| @@ -1,7894 +1,11413 @@ | ||
| 1 | -<?php | |
| 2 | -/** | |
| 3 | - * File: admin/class-knowledge-manager.php | |
| 4 | - * | |
| 5 | - * Handles all knowledge base content processing for MxChat | |
| 6 | - * Including PDF, sitemap, content processing, and WordPress post management | |
| 7 | - */ | |
| 8 | -if (!defined('ABSPATH')) { | |
| 9 | - exit; // Exit if accessed directly | |
| 10 | -} | |
| 11 | - | |
| 12 | -class MxChat_Knowledge_Manager { | |
| 13 | - | |
| 14 | - private $options; | |
| 15 | - | |
| 16 | - /** | |
| 17 | - * Constructor - Register hooks for content processing | |
| 18 | - */ | |
| 19 | -public function __construct() { | |
| 20 | - $this->options = get_option('mxchat_options', array()); | |
| 21 | - $this->mxchat_init_hooks(); | |
| 22 | - | |
| 23 | - $this->mxchat_init_role_hooks(); | |
| 24 | -} | |
| 25 | - | |
| 26 | -/** | |
| 27 | - * Initialize WordPress hooks for content processing | |
| 28 | - * | |
| 29 | - */ | |
| 30 | -private function mxchat_init_hooks() { | |
| 31 | - // Admin post handlers for form submissions | |
| 32 | - add_action('admin_post_mxchat_submit_content', array($this, 'mxchat_handle_content_submission')); | |
| 33 | - add_action('admin_post_mxchat_submit_sitemap', array($this, 'mxchat_handle_sitemap_submission')); | |
| 34 | - add_action('admin_post_mxchat_submit_pdf_file', array($this, 'mxchat_handle_pdf_file_submission')); | |
| 35 | - add_action('admin_post_mxchat_stop_processing', array($this, 'mxchat_stop_processing')); | |
| 36 | - | |
| 37 | - // AJAX handlers for real-time processing and status updates | |
| 38 | - add_action('wp_ajax_mxchat_get_status_updates', array($this, 'mxchat_ajax_get_status_updates')); | |
| 39 | - add_action('wp_ajax_mxchat_dismiss_completed_status', array($this, 'mxchat_ajax_dismiss_completed_status')); | |
| 40 | - add_action('wp_ajax_mxchat_get_content_list', array($this, 'ajax_mxchat_get_content_list')); | |
| 41 | - add_action('wp_ajax_mxchat_process_selected_content', array($this, 'ajax_mxchat_process_selected_content')); | |
| 42 | - add_action('wp_ajax_mxchat_save_inline_prompt', array($this, 'mxchat_save_inline_prompt')); | |
| 43 | - add_action('admin_post_mxchat_delete_pinecone_prompt', array($this, 'mxchat_handle_pinecone_prompt_delete')); | |
| 44 | - add_action('wp_ajax_mxchat_delete_pinecone_prompt', array($this, 'ajax_mxchat_delete_pinecone_prompt')); | |
| 45 | - add_action('wp_ajax_mxchat_delete_chunks_by_url', array($this, 'ajax_mxchat_delete_chunks_by_url')); | |
| 46 | - add_action('wp_ajax_mxchat_delete_wordpress_prompt', array($this, 'ajax_mxchat_delete_wordpress_prompt')); | |
| 47 | - add_action('wp_ajax_mxchat_bulk_delete_knowledge', array($this, 'ajax_mxchat_bulk_delete_knowledge')); | |
| 48 | - add_action('wp_ajax_mxchat_update_role_restriction', array($this, 'ajax_mxchat_update_role_restriction')); | |
| 49 | - | |
| 50 | - // Queue-based processing AJAX handlers | |
| 51 | - add_action('wp_ajax_mxchat_get_next_queue_item', array($this, 'ajax_mxchat_get_next_queue_item')); | |
| 52 | - add_action('wp_ajax_mxchat_process_queue_item', array($this, 'ajax_mxchat_process_queue_item')); | |
| 53 | - add_action('wp_ajax_mxchat_get_queue_status', array($this, 'ajax_mxchat_get_queue_status')); | |
| 54 | - add_action('wp_ajax_mxchat_clear_queue', array($this, 'ajax_mxchat_clear_queue')); | |
| 55 | - add_action('wp_ajax_mxchat_retry_failed', array($this, 'ajax_mxchat_retry_failed')); | |
| 56 | - add_action('wp_ajax_mxchat_get_recent_entries', array($this, 'ajax_mxchat_get_recent_entries')); | |
| 57 | - add_action('wp_ajax_mxchat_detect_sitemaps', array($this, 'ajax_mxchat_detect_sitemaps')); | |
| 58 | - add_action('wp_ajax_mxchat_refresh_pinecone_entries', array($this, 'ajax_mxchat_refresh_pinecone_entries')); | |
| 59 | - add_action('wp_ajax_mxchat_paginate_entries', array($this, 'ajax_mxchat_paginate_entries')); | |
| 60 | - add_action('wp_ajax_mxchat_get_entry_content', array($this, 'ajax_mxchat_get_entry_content')); | |
| 61 | - add_action('wp_ajax_mxchat_save_entry_content', array($this, 'ajax_mxchat_save_entry_content')); | |
| 62 | - | |
| 63 | - // WordPress post management hooks | |
| 64 | - add_action('pre_post_update', array($this, 'mxchat_store_pre_update_status'), 10, 2); | |
| 65 | - add_action('post_updated', array($this, 'mxchat_handle_post_update'), 10, 3); | |
| 66 | - add_action('before_delete_post', array($this, 'mxchat_handle_post_delete')); | |
| 67 | - add_action('wp_trash_post', array($this, 'mxchat_handle_post_delete')); | |
| 68 | - | |
| 69 | - // ACF hook - fires AFTER ACF fields are saved, ensuring ACF data is available | |
| 70 | - // Priority 20 to run after ACF's own save (which runs at priority 10) | |
| 71 | - add_action('acf/save_post', array($this, 'mxchat_handle_acf_save'), 20); | |
| 72 | - | |
| 73 | - add_action('wp_ajax_mxchat_mark_queue_complete', array($this, 'ajax_mxchat_mark_queue_complete')); | |
| 74 | - | |
| 75 | - // WooCommerce product hooks (if WooCommerce is active) | |
| 76 | - if (class_exists('WooCommerce')) { | |
| 77 | - add_action('pre_post_update', array($this, 'mxchat_store_pre_update_status'), 10, 2); | |
| 78 | - add_action('save_post_product', array($this, 'mxchat_handle_product_change'), 10, 3); | |
| 79 | - add_action('wp_trash_post', array($this, 'mxchat_handle_product_delete')); | |
| 80 | - add_action('before_delete_post', array($this, 'mxchat_handle_product_delete')); | |
| 81 | - } | |
| 82 | -} | |
| 83 | - | |
| 84 | - /** | |
| 85 | - * Get current options (refreshed) | |
| 86 | - */ | |
| 87 | - private function mxchat_get_options() { | |
| 88 | - if (empty($this->options)) { | |
| 89 | - $this->options = get_option('mxchat_options', array()); | |
| 90 | - } | |
| 91 | - return $this->options; | |
| 92 | - } | |
| 93 | - | |
| 94 | - | |
| 95 | - // ======================================== | |
| 96 | - // MAIN CONTENT SUBMISSION HANDLERS | |
| 97 | - // ======================================== | |
| 98 | - | |
| 99 | -public function mxchat_handle_content_submission() { | |
| 100 | - // Check if the form was submitted and the user has permission. | |
| 101 | - if (!isset($_POST['submit_content']) || !current_user_can('manage_options')) { | |
| 102 | - return; | |
| 103 | - } | |
| 104 | - | |
| 105 | - // Verify the nonce. | |
| 106 | - $nonce = isset($_POST['mxchat_submit_content_nonce']) ? sanitize_text_field(wp_unslash($_POST['mxchat_submit_content_nonce'])) : ''; | |
| 107 | - if (!wp_verify_nonce($nonce, 'mxchat_submit_content_action')) { | |
| 108 | - wp_die(esc_html__('Nonce verification failed.', 'mxchat')); | |
| 109 | - } | |
| 110 | - | |
| 111 | - // Sanitize the inputs. | |
| 112 | - // Use wp_kses_post to allow safe HTML and wp_unslash to remove WordPress added slashes | |
| 113 | - $article_content = wp_kses_post(wp_unslash($_POST['article_content'])); | |
| 114 | - $article_url = isset($_POST['article_url']) ? esc_url_raw($_POST['article_url']) : ''; | |
| 115 | - | |
| 116 | - // Get bot_id from form submission | |
| 117 | - $bot_id = isset($_POST['bot_id']) ? sanitize_key($_POST['bot_id']) : 'default'; | |
| 118 | - | |
| 119 | - // Get bot-specific options and API key | |
| 120 | - $bot_options = $this->get_bot_options($bot_id); | |
| 121 | - $options = !empty($bot_options) ? $bot_options : get_option('mxchat_options'); | |
| 122 | - $selected_model = $options['embedding_model'] ?? 'text-embedding-ada-002'; | |
| 123 | - | |
| 124 | - if (strpos($selected_model, 'voyage') === 0) { | |
| 125 | - $api_key = $options['voyage_api_key'] ?? ''; | |
| 126 | - } elseif (strpos($selected_model, 'gemini-embedding') === 0) { | |
| 127 | - $api_key = $options['gemini_api_key'] ?? ''; | |
| 128 | - } else { | |
| 129 | - $api_key = $options['api_key'] ?? ''; | |
| 130 | - } | |
| 131 | - | |
| 132 | - if (empty($api_key)) { | |
| 133 | - set_transient('mxchat_admin_notice_error', | |
| 134 | - esc_html__('API key is not configured. Please add your API key in the settings before submitting content.', 'mxchat'), | |
| 135 | - 30 | |
| 136 | - ); | |
| 137 | - wp_safe_redirect(esc_url(admin_url('admin.php?page=mxchat-prompts'))); | |
| 138 | - exit; | |
| 139 | - } | |
| 140 | - | |
| 141 | - // Use centralized utility function with bot_id | |
| 142 | - $result = MxChat_Utils::submit_content_to_db($article_content, $article_url, $api_key, null, $bot_id); | |
| 143 | - | |
| 144 | - if (is_wp_error($result)) { | |
| 145 | - set_transient('mxchat_admin_notice_error', | |
| 146 | - esc_html__('Error storing content: ', 'mxchat') . $result->get_error_message(), | |
| 147 | - 30 | |
| 148 | - ); | |
| 149 | - } else { | |
| 150 | - set_transient('mxchat_admin_notice_success', | |
| 151 | - esc_html__('Content successfully submitted!', 'mxchat'), | |
| 152 | - 30 | |
| 153 | - ); | |
| 154 | - } | |
| 155 | - | |
| 156 | - wp_safe_redirect(esc_url(admin_url('admin.php?page=mxchat-prompts'))); | |
| 157 | - exit; | |
| 158 | -} | |
| 159 | - | |
| 160 | -public function mxchat_is_pdf_url($url, $response) { | |
| 161 | - $content_type = wp_remote_retrieve_header($response, 'content-type'); | |
| 162 | - $file_extension = strtolower(pathinfo($url, PATHINFO_EXTENSION)); | |
| 163 | - | |
| 164 | - // Check Content-Disposition header for .pdf filename (Google Drive sends this) | |
| 165 | - $disposition = wp_remote_retrieve_header($response, 'content-disposition'); | |
| 166 | - $has_pdf_disposition = ! empty($disposition) && stripos($disposition, '.pdf') !== false; | |
| 167 | - | |
| 168 | - return strpos($content_type, 'pdf') !== false || $file_extension === 'pdf' || $has_pdf_disposition; | |
| 169 | -} | |
| 170 | - | |
| 171 | - | |
| 172 | -public function mxchat_handle_pdf_for_knowledge_base($pdf_url, $response, $bot_id = 'default') { | |
| 173 | - if (!current_user_can('manage_options')) { | |
| 174 | - return false; | |
| 175 | - } | |
| 176 | - | |
| 177 | - $pdf_url = esc_url_raw($pdf_url); | |
| 178 | - $upload_dir = wp_upload_dir(); | |
| 179 | - | |
| 180 | - if (isset($upload_dir['error']) && $upload_dir['error'] !== false) { | |
| 181 | - return false; | |
| 182 | - } | |
| 183 | - | |
| 184 | - $pdf_filename = sanitize_file_name('mxchat_kb_' . time() . '.pdf'); | |
| 185 | - $pdf_path = trailingslashit($upload_dir['path']) . $pdf_filename; | |
| 186 | - | |
| 187 | - $response_body = wp_remote_retrieve_body($response); | |
| 188 | - if (empty($response_body)) { | |
| 189 | - return false; | |
| 190 | - } | |
| 191 | - | |
| 192 | - if (!wp_mkdir_p(dirname($pdf_path))) { | |
| 193 | - return false; | |
| 194 | - } | |
| 195 | - | |
| 196 | - try { | |
| 197 | - file_put_contents($pdf_path, $response_body); | |
| 198 | - | |
| 199 | - if (!file_exists($pdf_path)) { | |
| 200 | - throw new Exception(__('Failed to save PDF file', 'mxchat')); | |
| 201 | - } | |
| 202 | - | |
| 203 | - $total_pages = $this->mxchat_validate_and_count_pdf_pages($pdf_path); | |
| 204 | - | |
| 205 | - if ($total_pages === false || $total_pages < 1) { | |
| 206 | - throw new Exception(__('Invalid PDF: Unable to parse or no pages found', 'mxchat')); | |
| 207 | - } | |
| 208 | - | |
| 209 | - // Create unique queue ID | |
| 210 | - $queue_id = 'pdf_' . md5($pdf_url . time()); | |
| 211 | - | |
| 212 | - // Create array of pages to process | |
| 213 | - $pages = array(); | |
| 214 | - for ($i = 1; $i <= $total_pages; $i++) { | |
| 215 | - $pages[] = array( | |
| 216 | - 'pdf_path' => $pdf_path, | |
| 217 | - 'pdf_url' => $pdf_url, | |
| 218 | - 'page_number' => $i, | |
| 219 | - 'total_pages' => $total_pages | |
| 220 | - ); | |
| 221 | - } | |
| 222 | - | |
| 223 | - // Add pages to queue | |
| 224 | - $queued_count = $this->mxchat_add_to_queue($queue_id, 'pdf_page', $pages, $bot_id); | |
| 225 | - | |
| 226 | - if ($queued_count === 0) { | |
| 227 | - wp_delete_file($pdf_path); | |
| 228 | - throw new Exception(__('Failed to add PDF pages to processing queue', 'mxchat')); | |
| 229 | - } | |
| 230 | - | |
| 231 | - // Store queue metadata | |
| 232 | - $this->mxchat_set_queue_meta($queue_id, 'source_url', $pdf_url); | |
| 233 | - $this->mxchat_set_queue_meta($queue_id, 'queue_type', 'pdf'); | |
| 234 | - $this->mxchat_set_queue_meta($queue_id, 'total_items', $total_pages); | |
| 235 | - $this->mxchat_set_queue_meta($queue_id, 'bot_id', $bot_id); | |
| 236 | - $this->mxchat_set_queue_meta($queue_id, 'pdf_path', $pdf_path); | |
| 237 | - $this->mxchat_set_queue_meta($queue_id, 'created_at', current_time('mysql')); | |
| 238 | - | |
| 239 | - // Store queue ID in transient for status tracking | |
| 240 | - set_transient('mxchat_active_queue_pdf', $queue_id, DAY_IN_SECONDS); | |
| 241 | - set_transient('mxchat_last_pdf_url', $pdf_url, DAY_IN_SECONDS); | |
| 242 | - | |
| 243 | - return 'queued'; | |
| 244 | - | |
| 245 | - } catch (Exception $e) { | |
| 246 | - if (file_exists($pdf_path)) { | |
| 247 | - wp_delete_file($pdf_path); | |
| 248 | - } | |
| 249 | - return $e->getMessage(); | |
| 250 | - } | |
| 251 | -} | |
| 252 | - | |
| 253 | -/** | |
| 254 | - * Handle direct PDF file upload from the knowledge base page | |
| 255 | - */ | |
| 256 | -public function mxchat_handle_pdf_file_submission() { | |
| 257 | - if (!isset($_POST['submit_pdf_file']) || !current_user_can('manage_options')) { | |
| 258 | - wp_die(esc_html__('Unauthorized access', 'mxchat')); | |
| 259 | - } | |
| 260 | - | |
| 261 | - check_admin_referer('mxchat_submit_pdf_file_action', 'mxchat_submit_pdf_file_nonce'); | |
| 262 | - | |
| 263 | - $redirect_url = admin_url('admin.php?page=mxchat-prompts'); | |
| 264 | - | |
| 265 | - // Validate file upload | |
| 266 | - if (empty($_FILES['pdf_file']) || $_FILES['pdf_file']['error'] !== UPLOAD_ERR_OK) { | |
| 267 | - $error_code = isset($_FILES['pdf_file']['error']) ? $_FILES['pdf_file']['error'] : UPLOAD_ERR_NO_FILE; | |
| 268 | - $error_messages = array( | |
| 269 | - UPLOAD_ERR_INI_SIZE => __('The uploaded file exceeds the server upload_max_filesize limit.', 'mxchat'), | |
| 270 | - UPLOAD_ERR_FORM_SIZE => __('The uploaded file exceeds the form MAX_FILE_SIZE limit.', 'mxchat'), | |
| 271 | - UPLOAD_ERR_PARTIAL => __('The file was only partially uploaded.', 'mxchat'), | |
| 272 | - UPLOAD_ERR_NO_FILE => __('No file was uploaded. Please select a PDF file.', 'mxchat'), | |
| 273 | - UPLOAD_ERR_NO_TMP_DIR => __('Server missing temporary folder.', 'mxchat'), | |
| 274 | - UPLOAD_ERR_CANT_WRITE => __('Server failed to write file to disk.', 'mxchat'), | |
| 275 | - ); | |
| 276 | - $error_msg = isset($error_messages[$error_code]) ? $error_messages[$error_code] : __('Unknown upload error.', 'mxchat'); | |
| 277 | - set_transient('mxchat_admin_notice_error', $error_msg, 30); | |
| 278 | - wp_safe_redirect(esc_url($redirect_url)); | |
| 279 | - exit; | |
| 280 | - } | |
| 281 | - | |
| 282 | - $file = $_FILES['pdf_file']; | |
| 283 | - | |
| 284 | - // Validate MIME type | |
| 285 | - $finfo = finfo_open(FILEINFO_MIME_TYPE); | |
| 286 | - $mime_type = finfo_file($finfo, $file['tmp_name']); | |
| 287 | - finfo_close($finfo); | |
| 288 | - | |
| 289 | - if ($mime_type !== 'application/pdf') { | |
| 290 | - set_transient('mxchat_admin_notice_error', | |
| 291 | - esc_html__('Invalid file type. Only PDF files are accepted.', 'mxchat'), | |
| 292 | - 30 | |
| 293 | - ); | |
| 294 | - wp_safe_redirect(esc_url($redirect_url)); | |
| 295 | - exit; | |
| 296 | - } | |
| 297 | - | |
| 298 | - // Validate extension | |
| 299 | - $ext = strtolower(pathinfo($file['name'], PATHINFO_EXTENSION)); | |
| 300 | - if ($ext !== 'pdf') { | |
| 301 | - set_transient('mxchat_admin_notice_error', | |
| 302 | - esc_html__('Invalid file extension. Only .pdf files are accepted.', 'mxchat'), | |
| 303 | - 30 | |
| 304 | - ); | |
| 305 | - wp_safe_redirect(esc_url($redirect_url)); | |
| 306 | - exit; | |
| 307 | - } | |
| 308 | - | |
| 309 | - $bot_id = isset($_POST['bot_id']) ? sanitize_key($_POST['bot_id']) : 'default'; | |
| 310 | - $original_filename = sanitize_file_name($file['name']); | |
| 311 | - | |
| 312 | - $upload_dir = wp_upload_dir(); | |
| 313 | - if (isset($upload_dir['error']) && $upload_dir['error'] !== false) { | |
| 314 | - set_transient('mxchat_admin_notice_error', | |
| 315 | - esc_html__('WordPress upload directory is not writable.', 'mxchat'), | |
| 316 | - 30 | |
| 317 | - ); | |
| 318 | - wp_safe_redirect(esc_url($redirect_url)); | |
| 319 | - exit; | |
| 320 | - } | |
| 321 | - | |
| 322 | - $pdf_filename = sanitize_file_name('mxchat_kb_' . time() . '.pdf'); | |
| 323 | - $pdf_path = trailingslashit($upload_dir['path']) . $pdf_filename; | |
| 324 | - | |
| 325 | - if (!wp_mkdir_p(dirname($pdf_path))) { | |
| 326 | - set_transient('mxchat_admin_notice_error', | |
| 327 | - esc_html__('Failed to create upload directory.', 'mxchat'), | |
| 328 | - 30 | |
| 329 | - ); | |
| 330 | - wp_safe_redirect(esc_url($redirect_url)); | |
| 331 | - exit; | |
| 332 | - } | |
| 333 | - | |
| 334 | - // Move uploaded file | |
| 335 | - if (!move_uploaded_file($file['tmp_name'], $pdf_path)) { | |
| 336 | - set_transient('mxchat_admin_notice_error', | |
| 337 | - esc_html__('Failed to save uploaded PDF file.', 'mxchat'), | |
| 338 | - 30 | |
| 339 | - ); | |
| 340 | - wp_safe_redirect(esc_url($redirect_url)); | |
| 341 | - exit; | |
| 342 | - } | |
| 343 | - | |
| 344 | - try { | |
| 345 | - $total_pages = $this->mxchat_validate_and_count_pdf_pages($pdf_path); | |
| 346 | - | |
| 347 | - if ($total_pages === false || $total_pages < 1) { | |
| 348 | - throw new Exception(__('Invalid PDF: Unable to parse or no pages found', 'mxchat')); | |
| 349 | - } | |
| 350 | - | |
| 351 | - // Use original filename as the source identifier | |
| 352 | - $source_label = 'upload://' . $original_filename; | |
| 353 | - | |
| 354 | - $queue_id = 'pdf_' . md5($source_label . time()); | |
| 355 | - | |
| 356 | - $pages = array(); | |
| 357 | - for ($i = 1; $i <= $total_pages; $i++) { | |
| 358 | - $pages[] = array( | |
| 359 | - 'pdf_path' => $pdf_path, | |
| 360 | - 'pdf_url' => $source_label, | |
| 361 | - 'page_number' => $i, | |
| 362 | - 'total_pages' => $total_pages, | |
| 363 | - ); | |
| 364 | - } | |
| 365 | - | |
| 366 | - $queued_count = $this->mxchat_add_to_queue($queue_id, 'pdf_page', $pages, $bot_id); | |
| 367 | - | |
| 368 | - if ($queued_count === 0) { | |
| 369 | - wp_delete_file($pdf_path); | |
| 370 | - throw new Exception(__('Failed to add PDF pages to processing queue', 'mxchat')); | |
| 371 | - } | |
| 372 | - | |
| 373 | - $this->mxchat_set_queue_meta($queue_id, 'source_url', $source_label); | |
| 374 | - $this->mxchat_set_queue_meta($queue_id, 'queue_type', 'pdf'); | |
| 375 | - $this->mxchat_set_queue_meta($queue_id, 'total_items', $total_pages); | |
| 376 | - $this->mxchat_set_queue_meta($queue_id, 'bot_id', $bot_id); | |
| 377 | - $this->mxchat_set_queue_meta($queue_id, 'pdf_path', $pdf_path); | |
| 378 | - $this->mxchat_set_queue_meta($queue_id, 'created_at', current_time('mysql')); | |
| 379 | - | |
| 380 | - set_transient('mxchat_active_queue_pdf', $queue_id, DAY_IN_SECONDS); | |
| 381 | - set_transient('mxchat_last_pdf_url', $source_label, DAY_IN_SECONDS); | |
| 382 | - | |
| 383 | - set_transient('mxchat_admin_notice_success', | |
| 384 | - sprintf( | |
| 385 | - esc_html__('PDF "%s" (%d pages) queued for processing. Processing will start automatically.', 'mxchat'), | |
| 386 | - esc_html($original_filename), | |
| 387 | - $total_pages | |
| 388 | - ), | |
| 389 | - 30 | |
| 390 | - ); | |
| 391 | - | |
| 392 | - } catch (Exception $e) { | |
| 393 | - if (file_exists($pdf_path)) { | |
| 394 | - wp_delete_file($pdf_path); | |
| 395 | - } | |
| 396 | - set_transient('mxchat_admin_notice_error', | |
| 397 | - esc_html__('Failed to process uploaded PDF: ', 'mxchat') . esc_html($e->getMessage()), | |
| 398 | - 30 | |
| 399 | - ); | |
| 400 | - } | |
| 401 | - | |
| 402 | - wp_safe_redirect(esc_url($redirect_url)); | |
| 403 | - exit; | |
| 404 | -} | |
| 405 | - | |
| 406 | -/** | |
| 407 | - * Validate PDF and count pages with multiple parser attempts | |
| 408 | - */ | |
| 409 | -private function mxchat_validate_and_count_pdf_pages($pdf_path) { | |
| 410 | - // Method 1: Try with Smalot PDF Parser (your current method) | |
| 411 | - try { | |
| 412 | - mxchat_load_pdf_parser(); | |
| 413 | - $parser = new \Smalot\PdfParser\Parser(); | |
| 414 | - $pdf = $parser->parseFile($pdf_path); | |
| 415 | - $pages = $pdf->getPages(); | |
| 416 | - $page_count = count($pages); | |
| 417 | - | |
| 418 | - if ($page_count > 0) { | |
| 419 | - //error_log('PDF parsed successfully with Smalot parser: ' . $page_count . ' pages'); | |
| 420 | - return $page_count; | |
| 421 | - } | |
| 422 | - } catch (Exception $e) { | |
| 423 | - //error_log('Smalot PDF parser failed: ' . $e->getMessage()); | |
| 424 | - } | |
| 425 | - | |
| 426 | - // Method 2: Try with pdfinfo command (if available) | |
| 427 | - if (function_exists('shell_exec') && !$this->mxchat_is_shell_disabled()) { | |
| 428 | - try { | |
| 429 | - $command = 'pdfinfo ' . escapeshellarg($pdf_path) . ' 2>&1'; | |
| 430 | - $output = shell_exec($command); | |
| 431 | - | |
| 432 | - if ($output && preg_match('/Pages:\s*(\d+)/', $output, $matches)) { | |
| 433 | - $page_count = intval($matches[1]); | |
| 434 | - if ($page_count > 0) { | |
| 435 | - //error_log('PDF parsed successfully with pdfinfo: ' . $page_count . ' pages'); | |
| 436 | - return $page_count; | |
| 437 | - } | |
| 438 | - } | |
| 439 | - } catch (Exception $e) { | |
| 440 | - //error_log('pdfinfo command failed: ' . $e->getMessage()); | |
| 441 | - } | |
| 442 | - } | |
| 443 | - | |
| 444 | - // Method 3: Try to repair PDF and parse again | |
| 445 | - try { | |
| 446 | - $repaired_path = $this->mxchat_attempt_pdf_repair($pdf_path); | |
| 447 | - if ($repaired_path && $repaired_path !== $pdf_path) { | |
| 448 | - mxchat_load_pdf_parser(); | |
| 449 | - $parser = new \Smalot\PdfParser\Parser(); | |
| 450 | - $pdf = $parser->parseFile($repaired_path); | |
| 451 | - $pages = $pdf->getPages(); | |
| 452 | - $page_count = count($pages); | |
| 453 | - | |
| 454 | - if ($page_count > 0) { | |
| 455 | - // Replace original with repaired version | |
| 456 | - copy($repaired_path, $pdf_path); | |
| 457 | - unlink($repaired_path); | |
| 458 | - //error_log('PDF repaired and parsed successfully: ' . $page_count . ' pages'); | |
| 459 | - return $page_count; | |
| 460 | - } | |
| 461 | - | |
| 462 | - // Clean up repaired file if it didn't work | |
| 463 | - unlink($repaired_path); | |
| 464 | - } | |
| 465 | - } catch (Exception $e) { | |
| 466 | - //error_log('PDF repair attempt failed: ' . $e->getMessage()); | |
| 467 | - } | |
| 468 | - | |
| 469 | - // Method 4: Manual PDF structure analysis (basic page count) | |
| 470 | - try { | |
| 471 | - $page_count = $this->mxchat_manual_pdf_page_count($pdf_path); | |
| 472 | - if ($page_count > 0) { | |
| 473 | - //error_log('PDF page count determined manually: ' . $page_count . ' pages'); | |
| 474 | - return $page_count; | |
| 475 | - } | |
| 476 | - } catch (Exception $e) { | |
| 477 | - //error_log('Manual PDF analysis failed: ' . $e->getMessage()); | |
| 478 | - } | |
| 479 | - | |
| 480 | - //error_log('All PDF parsing methods failed for: ' . $pdf_path); | |
| 481 | - return false; | |
| 482 | -} | |
| 483 | - | |
| 484 | -/** | |
| 485 | - * Check if shell_exec is disabled | |
| 486 | - */ | |
| 487 | -private function mxchat_is_shell_disabled() { | |
| 488 | - $disabled = explode(',', ini_get('disable_functions')); | |
| 489 | - return in_array('shell_exec', $disabled); | |
| 490 | -} | |
| 491 | - | |
| 492 | -/** | |
| 493 | - * Attempt to repair PDF using basic methods | |
| 494 | - */ | |
| 495 | -private function mxchat_attempt_pdf_repair($pdf_path) { | |
| 496 | - try { | |
| 497 | - $content = file_get_contents($pdf_path); | |
| 498 | - if (!$content) { | |
| 499 | - return false; | |
| 500 | - } | |
| 501 | - | |
| 502 | - // Check if PDF starts with proper header | |
| 503 | - if (substr($content, 0, 4) !== '%PDF') { | |
| 504 | - // Try to find PDF header in the content | |
| 505 | - $header_pos = strpos($content, '%PDF'); | |
| 506 | - if ($header_pos !== false && $header_pos < 1024) { | |
| 507 | - // Remove junk before PDF header | |
| 508 | - $content = substr($content, $header_pos); | |
| 509 | - $repaired_path = $pdf_path . '.repaired'; | |
| 510 | - file_put_contents($repaired_path, $content); | |
| 511 | - return $repaired_path; | |
| 512 | - } | |
| 513 | - } | |
| 514 | - | |
| 515 | - // Check for EOF marker | |
| 516 | - $content = rtrim($content); | |
| 517 | - if (!preg_match('/%%EOF\s*$/', $content)) { | |
| 518 | - // Add EOF marker if missing | |
| 519 | - $content .= "\n%%EOF"; | |
| 520 | - $repaired_path = $pdf_path . '.repaired'; | |
| 521 | - file_put_contents($repaired_path, $content); | |
| 522 | - return $repaired_path; | |
| 523 | - } | |
| 524 | - | |
| 525 | - } catch (Exception $e) { | |
| 526 | - //error_log('PDF repair error: ' . $e->getMessage()); | |
| 527 | - } | |
| 528 | - | |
| 529 | - return false; | |
| 530 | -} | |
| 531 | - | |
| 532 | -/** | |
| 533 | - * Manual PDF page counting by analyzing PDF structure | |
| 534 | - */ | |
| 535 | -private function mxchat_manual_pdf_page_count($pdf_path) { | |
| 536 | - try { | |
| 537 | - $content = file_get_contents($pdf_path); | |
| 538 | - if (!$content) { | |
| 539 | - return 0; | |
| 540 | - } | |
| 541 | - | |
| 542 | - // Method 1: Count /Type /Page objects | |
| 543 | - $page_count = preg_match_all('/\/Type\s*\/Page[^s]/', $content); | |
| 544 | - if ($page_count > 0) { | |
| 545 | - return $page_count; | |
| 546 | - } | |
| 547 | - | |
| 548 | - // Method 2: Look for /Count in pages object | |
| 549 | - if (preg_match('/\/Type\s*\/Pages.*?\/Count\s+(\d+)/', $content, $matches)) { | |
| 550 | - return intval($matches[1]); | |
| 551 | - } | |
| 552 | - | |
| 553 | - // Method 3: Count page references | |
| 554 | - $page_count = preg_match_all('/\d+\s+0\s+obj\s*<<[^>]*\/Type\s*\/Page/', $content); | |
| 555 | - if ($page_count > 0) { | |
| 556 | - return $page_count; | |
| 557 | - } | |
| 558 | - | |
| 559 | - } catch (Exception $e) { | |
| 560 | - //error_log('Manual PDF analysis error: ' . $e->getMessage()); | |
| 561 | - } | |
| 562 | - | |
| 563 | - return 0; | |
| 564 | -} | |
| 565 | - | |
| 566 | - | |
| 567 | -public function mxchat_save_inline_prompt() { | |
| 568 | - // DEBUG: Log what we're receiving | |
| 569 | - //error_log('=== MXCHAT DEBUG ==='); | |
| 570 | - //error_log('POST data: ' . print_r($_POST, true)); | |
| 571 | - //error_log('Nonce from POST: ' . ($_POST['_ajax_nonce'] ?? 'NOT FOUND')); | |
| 572 | - | |
| 573 | - // Check for nonce security | |
| 574 | - check_ajax_referer('mxchat_save_inline_nonce', '_ajax_nonce'); | |
| 575 | - | |
| 576 | - // If we get here, nonce passed | |
| 577 | - //error_log('Nonce verification PASSED'); | |
| 578 | - | |
| 579 | - // Verify permissions | |
| 580 | - if (!current_user_can('manage_options')) { | |
| 581 | - wp_send_json_error(esc_html__('Permission denied.', 'mxchat')); | |
| 582 | - return; | |
| 583 | - } | |
| 584 | - | |
| 585 | - global $wpdb; | |
| 586 | - $table_name = $wpdb->prefix . 'mxchat_system_prompt_content'; | |
| 587 | - | |
| 588 | - // Validate and sanitize input data | |
| 589 | - $prompt_id = isset($_POST['id']) ? absint($_POST['id']) : 0; | |
| 590 | - $article_content = isset($_POST['article_content']) ? wp_kses_post(wp_unslash($_POST['article_content'])) : ''; | |
| 591 | - $article_url = isset($_POST['article_url']) ? esc_url_raw($_POST['article_url']) : ''; | |
| 592 | - | |
| 593 | - if ($prompt_id > 0 && !empty($article_content)) { | |
| 594 | - // Re-generate the embedding vector for the updated content | |
| 595 | - $embedding_vector = $this->mxchat_generate_embedding($article_content); | |
| 596 | - if (is_array($embedding_vector)) { | |
| 597 | - // Serialize the embedding vector before storing it | |
| 598 | - $embedding_vector_serialized = serialize($embedding_vector); | |
| 599 | - // Update the prompt in the database | |
| 600 | - $updated = $wpdb->update( | |
| 601 | - $table_name, | |
| 602 | - array( | |
| 603 | - 'article_content' => $article_content, | |
| 604 | - 'embedding_vector' => $embedding_vector_serialized, | |
| 605 | - 'source_url' => $article_url, | |
| 606 | - 'embedding_model' => MxChat_Utils::get_active_embedding_model(), | |
| 607 | - ), | |
| 608 | - array('id' => $prompt_id), | |
| 609 | - array('%s', '%s', '%s', '%s'), | |
| 610 | - array('%d') | |
| 611 | - ); | |
| 612 | - if ($updated !== false) { | |
| 613 | - wp_send_json_success(); | |
| 614 | - } else { | |
| 615 | - MxChat_Admin::mxchat_log_debug('knowledge_error', 'Database update failed when saving knowledge entry'); | |
| 616 | - wp_send_json_error(esc_html__('Database update failed.', 'mxchat')); | |
| 617 | - } | |
| 618 | - } else { | |
| 619 | - MxChat_Admin::mxchat_log_debug('embedding_error', 'Embedding generation failed for knowledge entry'); | |
| 620 | - wp_send_json_error(esc_html__('Embedding generation failed.', 'mxchat')); | |
| 621 | - } | |
| 622 | - } else { | |
| 623 | - wp_send_json_error(esc_html__('Invalid data.', 'mxchat')); | |
| 624 | - } | |
| 625 | -} | |
| 626 | - | |
| 627 | - | |
| 628 | -/** | |
| 629 | - * AJAX: Get full content for editing — reassembles chunks if needed. | |
| 630 | - * Works for both WordPress DB and Pinecone entries. | |
| 631 | - */ | |
| 632 | -public function ajax_mxchat_get_entry_content() { | |
| 633 | - check_ajax_referer('mxchat_edit_entry_nonce', 'nonce'); | |
| 634 | - | |
| 635 | - if ( ! current_user_can('manage_options') ) { | |
| 636 | - wp_send_json_error( array( 'message' => 'Permission denied.' ) ); | |
| 637 | - } | |
| 638 | - | |
| 639 | - $source_url = isset($_POST['source_url']) ? sanitize_text_field( wp_unslash($_POST['source_url']) ) : ''; | |
| 640 | - $entry_id = isset($_POST['entry_id']) ? absint($_POST['entry_id']) : 0; | |
| 641 | - $data_source = isset($_POST['data_source']) ? sanitize_key($_POST['data_source']) : 'wordpress'; | |
| 642 | - $bot_id = isset($_POST['bot_id']) ? sanitize_key($_POST['bot_id']) : 'default'; | |
| 643 | - | |
| 644 | - if ( $data_source === 'pinecone' ) { | |
| 645 | - // Pinecone: fetch vectors by source_url, reassemble chunks | |
| 646 | - $content = $this->get_pinecone_entry_content( $source_url, $entry_id, $bot_id ); | |
| 647 | - } else { | |
| 648 | - // WordPress DB | |
| 649 | - $content = $this->get_wordpress_entry_content( $source_url, $entry_id ); | |
| 650 | - } | |
| 651 | - | |
| 652 | - if ( is_wp_error( $content ) ) { | |
| 653 | - wp_send_json_error( array( 'message' => $content->get_error_message() ) ); | |
| 654 | - } | |
| 655 | - | |
| 656 | - wp_send_json_success( $content ); | |
| 657 | -} | |
| 658 | - | |
| 659 | -/** | |
| 660 | - * Get content from WordPress DB — reassembles chunks by source_url. | |
| 661 | - */ | |
| 662 | -private function get_wordpress_entry_content( $source_url, $entry_id ) { | |
| 663 | - global $wpdb; | |
| 664 | - $table = $wpdb->prefix . 'mxchat_system_prompt_content'; | |
| 665 | - | |
| 666 | - // If we have a source_url, check for chunks | |
| 667 | - if ( ! empty( $source_url ) && strpos( $source_url, 'mxchat://' ) !== 0 ) { | |
| 668 | - $rows = $wpdb->get_results( $wpdb->prepare( | |
| 669 | - "SELECT id, article_content, source_url, content_type FROM {$table} WHERE source_url = %s ORDER BY id ASC", | |
| 670 | - $source_url | |
| 671 | - ) ); | |
| 672 | - | |
| 673 | - if ( $rows && count( $rows ) > 1 ) { | |
| 674 | - // Multiple rows = chunked. Reassemble. | |
| 675 | - $chunks = array(); | |
| 676 | - foreach ( $rows as $row ) { | |
| 677 | - $parsed = MxChat_Chunker::parse_stored_chunk( $row->article_content ); | |
| 678 | - $index = isset( $parsed['metadata']['chunk_index'] ) ? intval( $parsed['metadata']['chunk_index'] ) : count( $chunks ); | |
| 679 | - $chunks[ $index ] = $parsed['text']; | |
| 680 | - } | |
| 681 | - ksort( $chunks ); | |
| 682 | - return array( | |
| 683 | - 'content' => implode( "\n\n", $chunks ), | |
| 684 | - 'source_url' => $source_url, | |
| 685 | - 'is_chunked' => true, | |
| 686 | - 'chunk_count' => count( $chunks ), | |
| 687 | - 'content_type' => $rows[0]->content_type, | |
| 688 | - ); | |
| 689 | - } elseif ( $rows && count( $rows ) === 1 ) { | |
| 690 | - $parsed = MxChat_Chunker::parse_stored_chunk( $rows[0]->article_content ); | |
| 691 | - return array( | |
| 692 | - 'content' => $parsed['text'], | |
| 693 | - 'source_url' => $source_url, | |
| 694 | - 'entry_id' => $rows[0]->id, | |
| 695 | - 'is_chunked' => false, | |
| 696 | - 'content_type' => $rows[0]->content_type, | |
| 697 | - ); | |
| 698 | - } | |
| 699 | - } | |
| 700 | - | |
| 701 | - // Fallback: fetch by ID | |
| 702 | - if ( $entry_id > 0 ) { | |
| 703 | - $row = $wpdb->get_row( $wpdb->prepare( | |
| 704 | - "SELECT id, article_content, source_url, content_type FROM {$table} WHERE id = %d", | |
| 705 | - $entry_id | |
| 706 | - ) ); | |
| 707 | - if ( $row ) { | |
| 708 | - $parsed = MxChat_Chunker::parse_stored_chunk( $row->article_content ); | |
| 709 | - return array( | |
| 710 | - 'content' => $parsed['text'], | |
| 711 | - 'source_url' => $row->source_url, | |
| 712 | - 'entry_id' => $row->id, | |
| 713 | - 'is_chunked' => false, | |
| 714 | - 'content_type' => $row->content_type, | |
| 715 | - ); | |
| 716 | - } | |
| 717 | - } | |
| 718 | - | |
| 719 | - return new WP_Error( 'not_found', 'Entry not found.' ); | |
| 720 | -} | |
| 721 | - | |
| 722 | -/** | |
| 723 | - * Get content from Pinecone — fetches vectors by source_url, reassembles chunks. | |
| 724 | - */ | |
| 725 | -private function get_pinecone_entry_content( $source_url, $entry_id, $bot_id ) { | |
| 726 | - if ( ! class_exists('MxChat_Pinecone_Manager') ) { | |
| 727 | - return new WP_Error( 'pinecone_unavailable', 'Pinecone manager not available.' ); | |
| 728 | - } | |
| 729 | - | |
| 730 | - // Get Pinecone config | |
| 731 | - if ( $bot_id === 'default' || ! class_exists('MxChat_Multi_Bot_Manager') ) { | |
| 732 | - $pinecone_options = get_option('mxchat_pinecone_addon_options'); | |
| 733 | - $api_key = $pinecone_options['mxchat_pinecone_api_key'] ?? ''; | |
| 734 | - $host = $pinecone_options['mxchat_pinecone_host'] ?? ''; | |
| 735 | - $namespace = $pinecone_options['mxchat_pinecone_namespace'] ?? ''; | |
| 736 | - } else { | |
| 737 | - $bot_config = apply_filters('mxchat_get_bot_pinecone_config', array(), $bot_id); | |
| 738 | - $api_key = $bot_config['api_key'] ?? ''; | |
| 739 | - $host = $bot_config['host'] ?? ''; | |
| 740 | - $namespace = $bot_config['namespace'] ?? ''; | |
| 741 | - } | |
| 742 | - | |
| 743 | - if ( empty($host) || empty($api_key) ) { | |
| 744 | - return new WP_Error( 'pinecone_config', 'Pinecone not configured.' ); | |
| 745 | - } | |
| 746 | - | |
| 747 | - // List vectors with the source_url prefix | |
| 748 | - $base_id = md5( $source_url ); | |
| 749 | - $vector_ids = array( $base_id ); | |
| 750 | - | |
| 751 | - // Find chunk vectors | |
| 752 | - $list_url = "https://{$host}/vectors/list"; | |
| 753 | - $list_body = array( 'prefix' => $base_id . '_chunk_', 'limit' => 100 ); | |
| 754 | - if ( ! empty($namespace) ) { | |
| 755 | - $list_body['namespace'] = $namespace; | |
| 756 | - } | |
| 757 | - | |
| 758 | - $list_resp = wp_remote_post( $list_url, array( | |
| 759 | - 'headers' => array( 'Api-Key' => $api_key, 'Content-Type' => 'application/json' ), | |
| 760 | - 'body' => wp_json_encode( $list_body ), | |
| 761 | - 'timeout' => 15, | |
| 762 | - ) ); | |
| 763 | - | |
| 764 | - if ( ! is_wp_error($list_resp) ) { | |
| 765 | - $list_data = json_decode( wp_remote_retrieve_body($list_resp), true ); | |
| 766 | - if ( ! empty($list_data['vectors']) ) { | |
| 767 | - foreach ( $list_data['vectors'] as $v ) { | |
| 768 | - $vector_ids[] = $v['id']; | |
| 769 | - } | |
| 770 | - } | |
| 771 | - } | |
| 772 | - | |
| 773 | - // Fetch vectors with metadata | |
| 774 | - $fetch_url = "https://{$host}/vectors/fetch"; | |
| 775 | - $fetch_body = array( 'ids' => $vector_ids ); | |
| 776 | - if ( ! empty($namespace) ) { | |
| 777 | - $fetch_body['namespace'] = $namespace; | |
| 778 | - } | |
| 779 | - | |
| 780 | - $fetch_resp = wp_remote_post( $fetch_url, array( | |
| 781 | - 'headers' => array( 'Api-Key' => $api_key, 'Content-Type' => 'application/json' ), | |
| 782 | - 'body' => wp_json_encode( $fetch_body ), | |
| 783 | - 'timeout' => 15, | |
| 784 | - ) ); | |
| 785 | - | |
| 786 | - if ( is_wp_error($fetch_resp) ) { | |
| 787 | - return new WP_Error( 'pinecone_fetch', 'Failed to fetch from Pinecone.' ); | |
| 788 | - } | |
| 789 | - | |
| 790 | - $fetch_data = json_decode( wp_remote_retrieve_body($fetch_resp), true ); | |
| 791 | - $vectors = $fetch_data['vectors'] ?? array(); | |
| 792 | - | |
| 793 | - if ( empty($vectors) ) { | |
| 794 | - return new WP_Error( 'not_found', 'Entry not found in Pinecone.' ); | |
| 795 | - } | |
| 796 | - | |
| 797 | - // Reassemble chunks | |
| 798 | - $chunks = array(); | |
| 799 | - $content_type = 'content'; | |
| 800 | - foreach ( $vectors as $vid => $vector ) { | |
| 801 | - $meta = $vector['metadata'] ?? array(); | |
| 802 | - $text = $meta['text'] ?? ''; | |
| 803 | - $index = $meta['chunk_index'] ?? 0; | |
| 804 | - $content_type = $meta['type'] ?? 'content'; | |
| 805 | - $chunks[ intval($index) ] = $text; | |
| 806 | - } | |
| 807 | - ksort( $chunks ); | |
| 808 | - | |
| 809 | - return array( | |
| 810 | - 'content' => implode( "\n\n", $chunks ), | |
| 811 | - 'source_url' => $source_url, | |
| 812 | - 'is_chunked' => count($chunks) > 1, | |
| 813 | - 'chunk_count' => count($chunks), | |
| 814 | - 'content_type' => $content_type, | |
| 815 | - ); | |
| 816 | -} | |
| 817 | - | |
| 818 | -/** | |
| 819 | - * AJAX: Save edited content — re-chunks and re-embeds as needed. | |
| 820 | - * Works for both WordPress DB and Pinecone entries. | |
| 821 | - */ | |
| 822 | -public function ajax_mxchat_save_entry_content() { | |
| 823 | - check_ajax_referer('mxchat_edit_entry_nonce', 'nonce'); | |
| 824 | - | |
| 825 | - if ( ! current_user_can('manage_options') ) { | |
| 826 | - wp_send_json_error( array( 'message' => 'Permission denied.' ) ); | |
| 827 | - } | |
| 828 | - | |
| 829 | - $source_url = isset($_POST['source_url']) ? sanitize_text_field( wp_unslash($_POST['source_url']) ) : ''; | |
| 830 | - $entry_id = isset($_POST['entry_id']) ? absint($_POST['entry_id']) : 0; | |
| 831 | - $content = isset($_POST['content']) ? wp_kses_post( wp_unslash($_POST['content']) ) : ''; | |
| 832 | - $data_source = isset($_POST['data_source']) ? sanitize_key($_POST['data_source']) : 'wordpress'; | |
| 833 | - $bot_id = isset($_POST['bot_id']) ? sanitize_key($_POST['bot_id']) : 'default'; | |
| 834 | - $content_type = isset($_POST['content_type']) ? sanitize_key($_POST['content_type']) : 'content'; | |
| 835 | - | |
| 836 | - if ( empty($content) ) { | |
| 837 | - wp_send_json_error( array( 'message' => 'Content cannot be empty.' ) ); | |
| 838 | - } | |
| 839 | - | |
| 840 | - // Get the embedding API key | |
| 841 | - $options = get_option('mxchat_options', array()); | |
| 842 | - $api_key = ''; | |
| 843 | - | |
| 844 | - if ( $bot_id !== 'default' && class_exists('MxChat_Multi_Bot_Manager') ) { | |
| 845 | - $bot_options = apply_filters('mxchat_get_bot_options', array(), $bot_id); | |
| 846 | - $api_key = $bot_options['api_key'] ?? ''; | |
| 847 | - } | |
| 848 | - if ( empty($api_key) ) { | |
| 849 | - $api_key = $options['api_key'] ?? ''; | |
| 850 | - } | |
| 851 | - | |
| 852 | - global $wpdb; | |
| 853 | - $table = $wpdb->prefix . 'mxchat_system_prompt_content'; | |
| 854 | - | |
| 855 | - // If source_url is empty but we have an entry_id, look it up | |
| 856 | - if ( empty($source_url) && $entry_id > 0 && $data_source === 'wordpress' ) { | |
| 857 | - $row = $wpdb->get_row( $wpdb->prepare( "SELECT source_url FROM {$table} WHERE id = %d", $entry_id ) ); | |
| 858 | - if ( $row && ! empty($row->source_url) ) { | |
| 859 | - $source_url = $row->source_url; | |
| 860 | - } | |
| 861 | - } | |
| 862 | - | |
| 863 | - // For manual entries (no source_url or mxchat:// prefix), delete the old entry by ID first | |
| 864 | - // so submit_content_to_db creates a replacement instead of a duplicate | |
| 865 | - // Also treat legacy mxchat.ai source URLs as manual — old bug assigned the site URL to manual entries | |
| 866 | - $is_legacy_manual = !empty($source_url) && strpos($source_url, 'mxchat.ai') !== false && strpos($source_url, 'mxchat://') !== 0; | |
| 867 | - if ( $entry_id > 0 && (empty($source_url) || strpos($source_url, 'mxchat://') === 0 || $is_legacy_manual) ) { | |
| 868 | - $wpdb->delete( $table, array( 'id' => $entry_id ), array( '%d' ) ); | |
| 869 | - // Clear legacy URL so submit_content_to_db generates a unique mxchat:// identifier | |
| 870 | - // instead of reusing the shared URL (which would mass-delete other entries with the same URL) | |
| 871 | - if ( $is_legacy_manual ) { | |
| 872 | - $source_url = ''; | |
| 873 | - } | |
| 874 | - } | |
| 875 | - | |
| 876 | - // Use the existing submit_content_to_db which handles chunking, Pinecone, and WP DB | |
| 877 | - $vector_id = ! empty($source_url) ? md5($source_url) : md5('mxchat_manual_' . $entry_id); | |
| 878 | - | |
| 879 | - // submit_content_to_db already handles: delete old chunks → re-chunk → re-embed → store | |
| 880 | - $result = MxChat_Utils::submit_content_to_db( $content, $source_url, $api_key, $vector_id, $bot_id, $content_type ); | |
| 881 | - | |
| 882 | - if ( is_wp_error($result) ) { | |
| 883 | - wp_send_json_error( array( 'message' => $result->get_error_message() ) ); | |
| 884 | - } | |
| 885 | - | |
| 886 | - wp_send_json_success( array( 'message' => 'Content saved and re-embedded successfully.' ) ); | |
| 887 | -} | |
| 888 | - | |
| 889 | -public function mxchat_get_pdf_processing_status($pdf_url) { | |
| 890 | - $pdf_url = esc_url_raw($pdf_url); | |
| 891 | - $status = get_transient(sanitize_key('mxchat_pdf_status_' . md5($pdf_url))); | |
| 892 | - | |
| 893 | - if (!$status || !is_array($status)) { | |
| 894 | - return false; | |
| 895 | - } | |
| 896 | - | |
| 897 | - // Check for stalled processing (no updates for 5 minutes) | |
| 898 | - if ($status['status'] === 'processing' && (time() - absint($status['last_update'])) > 300) { | |
| 899 | - $status['status'] = 'error'; | |
| 900 | - $status['error'] = __('PDF processing appears to be stalled. No updates for over 5 minutes.', 'mxchat'); | |
| 901 | - | |
| 902 | - // Save the updated status | |
| 903 | - set_transient( | |
| 904 | - sanitize_key('mxchat_pdf_status_' . md5($pdf_url)), | |
| 905 | - array_map('sanitize_text_field', $status), | |
| 906 | - DAY_IN_SECONDS | |
| 907 | - ); | |
| 908 | - } | |
| 909 | - | |
| 910 | - $result = array( | |
| 911 | - 'total_pages' => absint($status['total_pages']), | |
| 912 | - 'processed_pages' => absint($status['processed_pages']), | |
| 913 | - 'failed_pages' => absint($status['failed_pages'] ?? 0), | |
| 914 | - 'percentage' => ($status['total_pages'] > 0) | |
| 915 | - ? round((absint($status['processed_pages']) / absint($status['total_pages'])) * 100) | |
| 916 | - : 0, | |
| 917 | - 'status' => sanitize_text_field($status['status']), | |
| 918 | - 'last_update' => human_time_diff(absint($status['last_update']), time()) . ' ' . esc_html__('ago', 'mxchat'), | |
| 919 | - 'failed_pages_list' => isset($status['failed_pages_list']) ? $status['failed_pages_list'] : array(), | |
| 920 | - 'completion_summary' => isset($status['completion_summary']) ? $status['completion_summary'] : null | |
| 921 | - ); | |
| 922 | - | |
| 923 | - // Add error message if present | |
| 924 | - if (isset($status['error']) && !empty($status['error'])) { | |
| 925 | - $result['error'] = sanitize_text_field($status['error']); | |
| 926 | - } | |
| 927 | - | |
| 928 | - return $result; | |
| 929 | -} | |
| 930 | - | |
| 931 | - | |
| 932 | -public function mxchat_handle_sitemap_submission() { | |
| 933 | - // Check if the form was submitted and verify permissions | |
| 934 | - if (!isset($_POST['submit_sitemap']) || !current_user_can('manage_options')) { | |
| 935 | - wp_die(esc_html__('Unauthorized access', 'mxchat')); | |
| 936 | - } | |
| 937 | - | |
| 938 | - // Verify nonce | |
| 939 | - check_admin_referer('mxchat_submit_sitemap_action', 'mxchat_submit_sitemap_nonce'); | |
| 940 | - | |
| 941 | - // Validate URL | |
| 942 | - if (!isset($_POST['sitemap_url']) || empty($_POST['sitemap_url'])) { | |
| 943 | - set_transient('mxchat_admin_notice_error', | |
| 944 | - esc_html__('Please provide a valid URL.', 'mxchat'), | |
| 945 | - 30 | |
| 946 | - ); | |
| 947 | - wp_safe_redirect(esc_url(admin_url('admin.php?page=mxchat-prompts'))); | |
| 948 | - exit; | |
| 949 | - } | |
| 950 | - | |
| 951 | - $submitted_url = esc_url_raw($_POST['sitemap_url']); | |
| 952 | - | |
| 953 | - // Convert Google Drive sharing URLs to direct download URLs | |
| 954 | - if ( strpos($submitted_url, 'drive.google.com') !== false ) { | |
| 955 | - $file_id = ''; | |
| 956 | - if ( preg_match('/[?&]id=([a-zA-Z0-9_-]+)/', $submitted_url, $m) ) { | |
| 957 | - $file_id = $m[1]; | |
| 958 | - } elseif ( preg_match('#/file/d/([a-zA-Z0-9_-]+)#', $submitted_url, $m) ) { | |
| 959 | - $file_id = $m[1]; | |
| 960 | - } | |
| 961 | - if ( ! empty($file_id) ) { | |
| 962 | - $submitted_url = 'https://drive.google.com/uc?export=download&id=' . $file_id; | |
| 963 | - } | |
| 964 | - } | |
| 965 | - | |
| 966 | - // Get bot_id from form submission | |
| 967 | - $bot_id = isset($_POST['bot_id']) ? sanitize_key($_POST['bot_id']) : 'default'; | |
| 968 | - | |
| 969 | - // Get bot-specific options and validate API key | |
| 970 | - $bot_options = $this->get_bot_options($bot_id); | |
| 971 | - $options = !empty($bot_options) ? $bot_options : get_option('mxchat_options'); | |
| 972 | - $selected_model = $options['embedding_model'] ?? 'text-embedding-ada-002'; | |
| 973 | - | |
| 974 | - if (strpos($selected_model, 'voyage') === 0) { | |
| 975 | - $api_key = $options['voyage_api_key'] ?? ''; | |
| 976 | - $provider_name = 'Voyage AI'; | |
| 977 | - } elseif (strpos($selected_model, 'gemini-embedding') === 0) { | |
| 978 | - $api_key = $options['gemini_api_key'] ?? ''; | |
| 979 | - $provider_name = 'Google Gemini'; | |
| 980 | - } else { | |
| 981 | - $api_key = $options['api_key'] ?? ''; | |
| 982 | - $provider_name = 'OpenAI'; | |
| 983 | - } | |
| 984 | - | |
| 985 | - if (empty($api_key)) { | |
| 986 | - $error_message = sprintf( | |
| 987 | - esc_html__('%s API key is not configured. Please add your API key in the settings before submitting content.', 'mxchat'), | |
| 988 | - $provider_name | |
| 989 | - ); | |
| 990 | - set_transient('mxchat_admin_notice_error', $error_message, 30); | |
| 991 | - wp_safe_redirect(esc_url(admin_url('admin.php?page=mxchat-prompts'))); | |
| 992 | - exit; | |
| 993 | - } | |
| 994 | - | |
| 995 | - // Fetch URL — use browser-like headers so servers with bot protection don't block us | |
| 996 | - $response = wp_remote_get($submitted_url, array( | |
| 997 | - 'timeout' => 30, | |
| 998 | - 'sslverify' => false, | |
| 999 | - 'user-agent' => 'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/131.0.0.0 Safari/537.36', | |
| 1000 | - 'headers' => array( | |
| 1001 | - 'Accept' => 'text/html,application/xhtml+xml,application/xml;q=0.9,*/*;q=0.8', | |
| 1002 | - 'Accept-Language' => 'en-US,en;q=0.9', | |
| 1003 | - ), | |
| 1004 | - )); | |
| 1005 | - | |
| 1006 | - if (is_wp_error($response) || wp_remote_retrieve_response_code($response) !== 200) { | |
| 1007 | - $error_message = is_wp_error($response) ? $response->get_error_message() : 'HTTP Status: ' . wp_remote_retrieve_response_code($response); | |
| 1008 | - set_transient('mxchat_admin_notice_error', | |
| 1009 | - sprintf( | |
| 1010 | - esc_html__('Failed to fetch the URL: %s', 'mxchat'), | |
| 1011 | - esc_html($error_message) | |
| 1012 | - ), | |
| 1013 | - 30 | |
| 1014 | - ); | |
| 1015 | - wp_safe_redirect(esc_url(admin_url('admin.php?page=mxchat-prompts'))); | |
| 1016 | - exit; | |
| 1017 | - } | |
| 1018 | - | |
| 1019 | - $content_type = wp_remote_retrieve_header($response, 'content-type'); | |
| 1020 | - $body_content = wp_remote_retrieve_body($response); | |
| 1021 | - | |
| 1022 | - if (empty($body_content)) { | |
| 1023 | - set_transient('mxchat_admin_notice_error', | |
| 1024 | - esc_html__('Empty response received from URL.', 'mxchat'), | |
| 1025 | - 30 | |
| 1026 | - ); | |
| 1027 | - wp_safe_redirect(esc_url(admin_url('admin.php?page=mxchat-prompts'))); | |
| 1028 | - exit; | |
| 1029 | - } | |
| 1030 | - | |
| 1031 | - // Handle PDF URL | |
| 1032 | - if ($this->mxchat_is_pdf_url($submitted_url, $response)) { | |
| 1033 | - $result = $this->mxchat_handle_pdf_for_knowledge_base($submitted_url, $response, $bot_id); | |
| 1034 | - | |
| 1035 | - if ($result === 'queued') { | |
| 1036 | - set_transient('mxchat_admin_notice_success', | |
| 1037 | - esc_html__('PDF queued for processing. Processing will start automatically.', 'mxchat'), | |
| 1038 | - 30 | |
| 1039 | - ); | |
| 1040 | - } else { | |
| 1041 | - set_transient('mxchat_admin_notice_error', | |
| 1042 | - esc_html__('Failed to queue PDF processing: ', 'mxchat') . esc_html($result), | |
| 1043 | - 30 | |
| 1044 | - ); | |
| 1045 | - } | |
| 1046 | - | |
| 1047 | - wp_safe_redirect(esc_url(admin_url('admin.php?page=mxchat-prompts'))); | |
| 1048 | - exit; | |
| 1049 | - } | |
| 1050 | - | |
| 1051 | - // Handle Sitemap XML | |
| 1052 | - if (strpos($content_type, 'xml') !== false || strpos($body_content, '<urlset') !== false) { | |
| 1053 | - libxml_use_internal_errors(true); | |
| 1054 | - $xml = simplexml_load_string($body_content); | |
| 1055 | - $xml_errors = libxml_get_errors(); | |
| 1056 | - libxml_clear_errors(); | |
| 1057 | - | |
| 1058 | - if ($xml === false || !empty($xml_errors)) { | |
| 1059 | - set_transient('mxchat_admin_notice_error', | |
| 1060 | - esc_html__('Invalid sitemap XML. Please provide a valid sitemap.', 'mxchat'), | |
| 1061 | - 30 | |
| 1062 | - ); | |
| 1063 | - wp_safe_redirect(esc_url(admin_url('admin.php?page=mxchat-prompts'))); | |
| 1064 | - exit; | |
| 1065 | - } | |
| 1066 | - | |
| 1067 | - $result = $this->mxchat_handle_sitemap_for_knowledge_base($xml, $submitted_url, $bot_id); | |
| 1068 | - | |
| 1069 | - if ($result === 'queued') { | |
| 1070 | - set_transient('mxchat_admin_notice_success', | |
| 1071 | - esc_html__('Sitemap queued for processing. Processing will start automatically.', 'mxchat'), | |
| 1072 | - 30 | |
| 1073 | - ); | |
| 1074 | - } else { | |
| 1075 | - set_transient('mxchat_admin_notice_error', | |
| 1076 | - esc_html__('Failed to queue sitemap processing. Please check the status below for details.', 'mxchat'), | |
| 1077 | - 30 | |
| 1078 | - ); | |
| 1079 | - } | |
| 1080 | - | |
| 1081 | - wp_safe_redirect(esc_url(admin_url('admin.php?page=mxchat-prompts'))); | |
| 1082 | - exit; | |
| 1083 | - } | |
| 1084 | - | |
| 1085 | - // Handle Regular URL (single page) | |
| 1086 | - $page_content = $this->mxchat_extract_main_content($body_content); | |
| 1087 | - $sanitized_content = $this->mxchat_sanitize_content_for_api($page_content); | |
| 1088 | - | |
| 1089 | - //error_log('[MXCHAT-URL-DEBUG] URL: ' . $submitted_url); | |
| 1090 | - //error_log('[MXCHAT-URL-DEBUG] Extracted page_content length: ' . strlen($page_content)); | |
| 1091 | - //error_log('[MXCHAT-URL-DEBUG] Sanitized content length: ' . strlen($sanitized_content)); | |
| 1092 | - //error_log('[MXCHAT-URL-DEBUG] Content preview: ' . substr($sanitized_content, 0, 500)); | |
| 1093 | - | |
| 1094 | - if (empty($sanitized_content)) { | |
| 1095 | - set_transient('mxchat_admin_notice_error', | |
| 1096 | - esc_html__('No valid content found on the provided URL.', 'mxchat'), | |
| 1097 | - 30 | |
| 1098 | - ); | |
| 1099 | - wp_safe_redirect(esc_url(admin_url('admin.php?page=mxchat-prompts'))); | |
| 1100 | - exit; | |
| 1101 | - } | |
| 1102 | - | |
| 1103 | - // For single URLs, process immediately using submit_content_to_db | |
| 1104 | - // This handles chunking automatically for large content | |
| 1105 | - $db_result = MxChat_Utils::submit_content_to_db( | |
| 1106 | - $sanitized_content, | |
| 1107 | - $submitted_url, | |
| 1108 | - $api_key, | |
| 1109 | - null, | |
| 1110 | - $bot_id, | |
| 1111 | - 'url' // content_type | |
| 1112 | - ); | |
| 1113 | - | |
| 1114 | - if (is_wp_error($db_result)) { | |
| 1115 | - $error_message = esc_html__('Failed to store content in database: ', 'mxchat') . esc_html($db_result->get_error_message()); | |
| 1116 | - set_transient('mxchat_admin_notice_error', $error_message, 30); | |
| 1117 | - } else { | |
| 1118 | - $success_message = esc_html__('URL content successfully submitted!', 'mxchat'); | |
| 1119 | - set_transient('mxchat_admin_notice_success', $success_message, 30); | |
| 1120 | - } | |
| 1121 | - | |
| 1122 | - wp_safe_redirect(esc_url(admin_url('admin.php?page=mxchat-prompts'))); | |
| 1123 | - exit; | |
| 1124 | -} | |
| 1125 | - | |
| 1126 | - | |
| 1127 | -public function mxchat_get_single_url_status() { | |
| 1128 | - $status = get_transient('mxchat_single_url_status'); | |
| 1129 | - if (!$status) { | |
| 1130 | - return null; | |
| 1131 | - } | |
| 1132 | - | |
| 1133 | - // Add human-readable time | |
| 1134 | - if (isset($status['timestamp'])) { | |
| 1135 | - $status['human_time'] = human_time_diff(strtotime($status['timestamp']), current_time('timestamp')) . ' ' . __('ago', 'mxchat'); | |
| 1136 | - } | |
| 1137 | - | |
| 1138 | - return $status; | |
| 1139 | -} | |
| 1140 | - | |
| 1141 | -public function mxchat_handle_sitemap_for_knowledge_base($xml, $sitemap_url, $bot_id = 'default') { | |
| 1142 | - if (!current_user_can('manage_options')) { | |
| 1143 | - return false; | |
| 1144 | - } | |
| 1145 | - | |
| 1146 | - try { | |
| 1147 | - $sitemap_url = esc_url_raw($sitemap_url); | |
| 1148 | - | |
| 1149 | - if (!$xml || !is_object($xml)) { | |
| 1150 | - throw new Exception(__('Invalid XML object provided', 'mxchat')); | |
| 1151 | - } | |
| 1152 | - | |
| 1153 | - // Get bot-specific embedding API for validation | |
| 1154 | - $bot_options = $this->get_bot_options($bot_id); | |
| 1155 | - $options = !empty($bot_options) ? $bot_options : get_option('mxchat_options'); | |
| 1156 | - | |
| 1157 | - // Test the embedding API before processing | |
| 1158 | - $test_phrase = "Test embedding generation for MxChat"; | |
| 1159 | - $test_result = $this->mxchat_generate_embedding($test_phrase, $bot_id); | |
| 1160 | - | |
| 1161 | - if (is_string($test_result)) { | |
| 1162 | - throw new Exception(__('Embedding API validation failed: ', 'mxchat') . $test_result); | |
| 1163 | - } | |
| 1164 | - | |
| 1165 | - if (!is_array($test_result)) { | |
| 1166 | - throw new Exception(__('Embedding API returned unexpected result type. Please check your configuration.', 'mxchat')); | |
| 1167 | - } | |
| 1168 | - | |
| 1169 | - // Extract URLs from sitemap | |
| 1170 | - $urls = array(); | |
| 1171 | - foreach ($xml->url as $url_element) { | |
| 1172 | - $url = esc_url_raw((string)$url_element->loc); | |
| 1173 | - if ($url) { | |
| 1174 | - $urls[] = array('url' => $url); | |
| 1175 | - } | |
| 1176 | - } | |
| 1177 | - | |
| 1178 | - $total_urls = count($urls); | |
| 1179 | - | |
| 1180 | - if ($total_urls < 1) { | |
| 1181 | - throw new Exception(__('No valid URLs found in sitemap', 'mxchat')); | |
| 1182 | - } | |
| 1183 | - | |
| 1184 | - // Create unique queue ID | |
| 1185 | - $queue_id = 'sitemap_' . md5($sitemap_url . time()); | |
| 1186 | - | |
| 1187 | - // Add URLs to queue | |
| 1188 | - $queued_count = $this->mxchat_add_to_queue($queue_id, 'url', $urls, $bot_id); | |
| 1189 | - | |
| 1190 | - if ($queued_count === 0) { | |
| 1191 | - throw new Exception(__('Failed to add URLs to processing queue', 'mxchat')); | |
| 1192 | - } | |
| 1193 | - | |
| 1194 | - // Store queue metadata | |
| 1195 | - $this->mxchat_set_queue_meta($queue_id, 'source_url', $sitemap_url); | |
| 1196 | - $this->mxchat_set_queue_meta($queue_id, 'queue_type', 'sitemap'); | |
| 1197 | - $this->mxchat_set_queue_meta($queue_id, 'total_items', $total_urls); | |
| 1198 | - $this->mxchat_set_queue_meta($queue_id, 'bot_id', $bot_id); | |
| 1199 | - $this->mxchat_set_queue_meta($queue_id, 'created_at', current_time('mysql')); | |
| 1200 | - | |
| 1201 | - // Store queue ID in transient for status tracking | |
| 1202 | - set_transient('mxchat_active_queue_sitemap', $queue_id, DAY_IN_SECONDS); | |
| 1203 | - set_transient('mxchat_last_sitemap_url', $sitemap_url, DAY_IN_SECONDS); | |
| 1204 | - | |
| 1205 | - return 'queued'; | |
| 1206 | - | |
| 1207 | - } catch (Exception $e) { | |
| 1208 | - $error_message = $e->getMessage(); | |
| 1209 | - //error_log(sprintf(esc_html__('Error preparing sitemap for processing: %s', 'mxchat'), esc_html($error_message))); | |
| 1210 | - | |
| 1211 | - return $error_message; | |
| 1212 | - } | |
| 1213 | - | |
| 1214 | -} | |
| 1215 | - | |
| 1216 | -/** | |
| 1217 | - * Remove shortcode tags but preserve the content inside them | |
| 1218 | - * Example: [vc_column]Hello World[/vc_column] becomes "Hello World" | |
| 1219 | - * | |
| 1220 | - * @param string $content The content containing shortcodes | |
| 1221 | - * @return string Content with shortcode tags removed but inner content preserved | |
| 1222 | - */ | |
| 1223 | -private function strip_shortcode_tags_preserve_content($content) { | |
| 1224 | - // Single-pass regex removes all shortcode brackets: [tag], [tag attr="val"], [/tag], [tag /] | |
| 1225 | - // Content between tags is inherently preserved since only brackets are targeted | |
| 1226 | - $result = preg_replace('/\[\/?\w[\w-]*[^\]]*\]/', '', $content); | |
| 1227 | - return ($result !== null) ? $result : $content; | |
| 1228 | -} | |
| 1229 | - | |
| 1230 | -public function mxchat_sanitize_content_for_api($content) { | |
| 1231 | - //error_log('[MXCHAT-SANITIZE] Original content preview: ' . substr($content, 0, 500) . '...'); | |
| 1232 | - | |
| 1233 | - // Remove shortcode tags but PRESERVE content inside them | |
| 1234 | - $content = $this->strip_shortcode_tags_preserve_content($content); | |
| 1235 | - | |
| 1236 | - // Remove script, style tags, and HTML comments | |
| 1237 | - $content = preg_replace('/<script\b[^>]*>(.*?)<\/script>/is', '', $content); | |
| 1238 | - $content = preg_replace('/<style\b[^>]*>(.*?)<\/style>/is', '', $content); | |
| 1239 | - $content = preg_replace('/<!--(.|\s)*?-->/', '', $content); | |
| 1240 | - | |
| 1241 | - // Remove all HTML tags and decode HTML entities | |
| 1242 | - $content = wp_strip_all_tags($content); | |
| 1243 | - $content = html_entity_decode($content, ENT_QUOTES | ENT_HTML5); | |
| 1244 | - | |
| 1245 | - // Normalize whitespace but preserve paragraph breaks | |
| 1246 | - // First, normalize line endings to \n | |
| 1247 | - $content = str_replace(["\r\n", "\r"], "\n", $content); | |
| 1248 | - // Replace multiple spaces/tabs with single space, but preserve newlines | |
| 1249 | - $content = preg_replace('/[ \t]+/', ' ', $content); | |
| 1250 | - // Replace 3+ newlines with 2 newlines (max 2 blank lines) | |
| 1251 | - $content = preg_replace('/\n{3,}/', "\n\n", $content); | |
| 1252 | - // Trim each line | |
| 1253 | - $lines = explode("\n", $content); | |
| 1254 | - $lines = array_map('trim', $lines); | |
| 1255 | - $content = implode("\n", $lines); | |
| 1256 | - // Final trim | |
| 1257 | - $content = trim($content); | |
| 1258 | - | |
| 1259 | - // Remove control characters (which can cause database issues) but preserve \n (0x0A) and \r (0x0D) | |
| 1260 | - $content = preg_replace('/[\x00-\x08\x0B\x0C\x0E-\x1F\x7F]/u', '', $content); | |
| 1261 | - | |
| 1262 | - // Remove NULL bytes which can cause database errors | |
| 1263 | - $content = str_replace("\0", "", $content); | |
| 1264 | - | |
| 1265 | - // Ensure valid UTF-8 encoding | |
| 1266 | - $content = wp_check_invalid_utf8($content); | |
| 1267 | - | |
| 1268 | - // Remove any extremely long strings without spaces (often garbage) | |
| 1269 | - $content = preg_replace('/\S{300,}/', ' ', $content); | |
| 1270 | - | |
| 1271 | - // Replace problematic characters that often cause database issues | |
| 1272 | - $content = preg_replace('/[\x{10000}-\x{10FFFF}]/u', '', $content); // Remove emoji and other high Unicode characters | |
| 1273 | - | |
| 1274 | - // Replace any remaining potentially problematic characters with spaces | |
| 1275 | - // BUT preserve newlines by temporarily replacing them | |
| 1276 | - $content = str_replace("\n", "NEWLINE_PLACEHOLDER", $content); | |
| 1277 | - $content = preg_replace('/[^\p{L}\p{N}\p{P}\p{Z}\p{Sm}]/u', ' ', $content); | |
| 1278 | - $content = str_replace("NEWLINE_PLACEHOLDER", "\n", $content); | |
| 1279 | - | |
| 1280 | - // Limit to reasonable length if needed | |
| 1281 | - $max_length = 65000; // Just under MySQL TEXT field limit | |
| 1282 | - if (strlen($content) > $max_length) { | |
| 1283 | - $content = substr($content, 0, $max_length); | |
| 1284 | - } | |
| 1285 | - | |
| 1286 | - //error_log('[MXCHAT-SANITIZE] Sanitized content preview: ' . substr($content, 0, 500) . '...'); | |
| 1287 | - return $content; | |
| 1288 | -} | |
| 1289 | -public function mxchat_extract_main_content($html) { | |
| 1290 | - if (empty($html)) { | |
| 1291 | - return ''; | |
| 1292 | - } | |
| 1293 | - try { | |
| 1294 | - $dom = new DOMDocument; | |
| 1295 | - libxml_use_internal_errors(true); // Suppress HTML parsing errors | |
| 1296 | - @$dom->loadHTML(mb_convert_encoding($html, 'HTML-ENTITIES', 'UTF-8'), LIBXML_HTML_NOIMPLIED | LIBXML_HTML_NODEFDTD); | |
| 1297 | - $xpath = new DOMXPath($dom); | |
| 1298 | - | |
| 1299 | - // For debugging purposes | |
| 1300 | - $debugEnabled = true; // Set to true to enable debugging output | |
| 1301 | - $debug = function($message) use ($debugEnabled) { | |
| 1302 | - if ($debugEnabled) { | |
| 1303 | - //error_log('[MXCHAT-EXTRACT-DEBUG] ' . $message); | |
| 1304 | - } | |
| 1305 | - }; | |
| 1306 | - | |
| 1307 | - // Direct targeting for Gerow theme posts | |
| 1308 | - $post_text = $xpath->query('//div[contains(@class, "post-text")]'); | |
| 1309 | - if ($post_text && $post_text->length > 0) { | |
| 1310 | - $debug("Found post-text directly"); | |
| 1311 | - $content = ''; | |
| 1312 | - foreach ($post_text as $node) { | |
| 1313 | - $content .= $dom->saveHTML($node); | |
| 1314 | - } | |
| 1315 | - if (!empty($content)) { | |
| 1316 | - $debug("Returning post-text content"); | |
| 1317 | - return $content; | |
| 1318 | - } | |
| 1319 | - } | |
| 1320 | - | |
| 1321 | - // Try to get the blog details content which contains the post-text | |
| 1322 | - $blog_details = $xpath->query('//div[contains(@class, "blog-details-content")]'); | |
| 1323 | - if ($blog_details && $blog_details->length > 0) { | |
| 1324 | - $debug("Found blog-details-content"); | |
| 1325 | - $content = ''; | |
| 1326 | - foreach ($blog_details as $node) { | |
| 1327 | - $content .= $dom->saveHTML($node); | |
| 1328 | - } | |
| 1329 | - if (!empty($content)) { | |
| 1330 | - $debug("Returning blog-details-content"); | |
| 1331 | - return $content; | |
| 1332 | - } | |
| 1333 | - } | |
| 1334 | - | |
| 1335 | - // Try to get the article which contains the blog details | |
| 1336 | - $article = $xpath->query('//article[contains(@class, "blog-details-wrap")]'); | |
| 1337 | - if ($article && $article->length > 0) { | |
| 1338 | - $debug("Found article with blog-details-wrap"); | |
| 1339 | - $content = ''; | |
| 1340 | - foreach ($article as $node) { | |
| 1341 | - $content .= $dom->saveHTML($node); | |
| 1342 | - } | |
| 1343 | - if (!empty($content)) { | |
| 1344 | - $debug("Returning article content"); | |
| 1345 | - return $content; | |
| 1346 | - } | |
| 1347 | - } | |
| 1348 | - | |
| 1349 | - // Try even broader with the blog-item-wrap | |
| 1350 | - $blog_item = $xpath->query('//div[contains(@class, "blog-item-wrap")]'); | |
| 1351 | - if ($blog_item && $blog_item->length > 0) { | |
| 1352 | - $debug("Found blog-item-wrap"); | |
| 1353 | - $content = ''; | |
| 1354 | - foreach ($blog_item as $node) { | |
| 1355 | - $content .= $dom->saveHTML($node); | |
| 1356 | - } | |
| 1357 | - if (!empty($content)) { | |
| 1358 | - $debug("Returning blog-item-wrap content"); | |
| 1359 | - return $content; | |
| 1360 | - } | |
| 1361 | - } | |
| 1362 | - | |
| 1363 | - // Specific Gerow theme path | |
| 1364 | - $gerow_path = $xpath->query('//section[contains(@class, "blog-area")]//div[contains(@class, "post-text")]'); | |
| 1365 | - if ($gerow_path && $gerow_path->length > 0) { | |
| 1366 | - $debug("Found Gerow theme path to post-text"); | |
| 1367 | - $content = ''; | |
| 1368 | - foreach ($gerow_path as $node) { | |
| 1369 | - $content .= $dom->saveHTML($node); | |
| 1370 | - } | |
| 1371 | - if (!empty($content)) { | |
| 1372 | - $debug("Returning Gerow post-text content"); | |
| 1373 | - return $content; | |
| 1374 | - } | |
| 1375 | - } | |
| 1376 | - | |
| 1377 | - // Generic blog post selectors | |
| 1378 | - $selectors = [ | |
| 1379 | - // Blog post specific selectors | |
| 1380 | - '//div[contains(@class, "post-text")]', | |
| 1381 | - '//article[contains(@class, "blog-post-item")]//div[contains(@class, "post-text")]', | |
| 1382 | - '//div[contains(@class, "blog-details-content")]', | |
| 1383 | - '//article[contains(@class, "blog-details-wrap")]', | |
| 1384 | - '//div[contains(@class, "entry-content")]', | |
| 1385 | - '//div[contains(@class, "blog-content")]', | |
| 1386 | - '//div[contains(@class, "blog-item-wrap")]', | |
| 1387 | - | |
| 1388 | - // More general content selectors | |
| 1389 | - '//div[contains(@class, "page__content")]', | |
| 1390 | - '//div[contains(@class, "elementor-widget-container")]', | |
| 1391 | - '//div[contains(@class, "elementor-text-editor")]', | |
| 1392 | - '//div[contains(@class, "elementor-widget-text-editor")]', | |
| 1393 | - '//*[contains(@class, "entry-content")]', | |
| 1394 | - '//*[contains(@class, "post-content")]', | |
| 1395 | - '//*[contains(@class, "article-content")]', | |
| 1396 | - '//*[@id="content"]', | |
| 1397 | - '//*[@id="main-content"]', | |
| 1398 | - '//section[contains(@class, "blog-area")]', | |
| 1399 | - '//article', | |
| 1400 | - '//main', | |
| 1401 | - '//div[contains(@class, "content")]' | |
| 1402 | - ]; | |
| 1403 | - | |
| 1404 | - // First handle Elementor content - get only leaf widget containers to avoid duplicates | |
| 1405 | - $debug("Checking for Elementor content"); | |
| 1406 | - // Get widget containers that are direct children of widgets (not nested inside other widget containers) | |
| 1407 | - $elementor_widgets = $xpath->query('//div[contains(@class, "elementor-widget")]//div[contains(@class, "elementor-widget-container")]'); | |
| 1408 | - if ($elementor_widgets && $elementor_widgets->length > 0) { | |
| 1409 | - $debug("Found Elementor widgets"); | |
| 1410 | - $seen_content = array(); // Track seen content to avoid duplicates | |
| 1411 | - $combined_content = ''; | |
| 1412 | - foreach ($elementor_widgets as $widget) { | |
| 1413 | - $widget_content = $dom->saveHTML($widget); | |
| 1414 | - if (!empty($widget_content)) { | |
| 1415 | - // Create a hash of the content to detect duplicates | |
| 1416 | - $content_hash = md5($widget_content); | |
| 1417 | - if (!isset($seen_content[$content_hash])) { | |
| 1418 | - $seen_content[$content_hash] = true; | |
| 1419 | - $combined_content .= $widget_content; | |
| 1420 | - } | |
| 1421 | - } | |
| 1422 | - } | |
| 1423 | - if (!empty($combined_content)) { | |
| 1424 | - $debug("Returning Elementor content"); | |
| 1425 | - return $combined_content; | |
| 1426 | - } | |
| 1427 | - } | |
| 1428 | - | |
| 1429 | - // Try standard selectors one by one | |
| 1430 | - foreach ($selectors as $selector) { | |
| 1431 | - $debug("Trying selector: " . $selector); | |
| 1432 | - $nodes = $xpath->query($selector); | |
| 1433 | - if ($nodes && $nodes->length > 0) { | |
| 1434 | - $debug("Found " . $nodes->length . " matches for selector: " . $selector); | |
| 1435 | - // Only take the FIRST matching node to avoid duplicate content | |
| 1436 | - // (pages often have nested or multiple containers with same class) | |
| 1437 | - $content = $dom->saveHTML($nodes->item(0)); | |
| 1438 | - if (!empty($content)) { | |
| 1439 | - $debug("Returning content from selector: " . $selector . " (first match only)"); | |
| 1440 | - return $content; | |
| 1441 | - } | |
| 1442 | - } | |
| 1443 | - } | |
| 1444 | - | |
| 1445 | - // Manual regex fallback for post-text if DOM methods fail | |
| 1446 | - $debug("Trying regex fallback"); | |
| 1447 | - if (preg_match('/<div class="post-text">(.*?)<\/div>\s*<\/div>\s*<\/div>/s', $html, $matches)) { | |
| 1448 | - $debug("Found post-text via regex"); | |
| 1449 | - return '<div class="post-text">' . $matches[1] . '</div>'; | |
| 1450 | - } | |
| 1451 | - | |
| 1452 | - // Try to extract the blog section as a whole | |
| 1453 | - $blog_section = $xpath->query('//section[contains(@class, "blog-area")]'); | |
| 1454 | - if ($blog_section && $blog_section->length > 0) { | |
| 1455 | - $debug("Found blog-area section"); | |
| 1456 | - $content = ''; | |
| 1457 | - foreach ($blog_section as $node) { | |
| 1458 | - $content .= $dom->saveHTML($node); | |
| 1459 | - } | |
| 1460 | - if (!empty($content)) { | |
| 1461 | - $debug("Returning blog-area section content"); | |
| 1462 | - return $content; | |
| 1463 | - } | |
| 1464 | - } | |
| 1465 | - | |
| 1466 | - // Generic container selectors for non-CMS sites (like .asp pages) | |
| 1467 | - $debug("Trying generic container selectors"); | |
| 1468 | - $generic_selectors = [ | |
| 1469 | - '//div[@id="main"]', | |
| 1470 | - '//div[@id="wrapper"]', | |
| 1471 | - '//div[@id="page"]', | |
| 1472 | - '//div[@id="site-content"]', | |
| 1473 | - '//div[contains(@class, "main-content")]', | |
| 1474 | - '//div[contains(@class, "page-content")]', | |
| 1475 | - '//div[contains(@class, "site-content")]', | |
| 1476 | - ]; | |
| 1477 | - | |
| 1478 | - foreach ($generic_selectors as $selector) { | |
| 1479 | - $debug("Trying generic selector: " . $selector); | |
| 1480 | - $nodes = $xpath->query($selector); | |
| 1481 | - if ($nodes && $nodes->length > 0) { | |
| 1482 | - $content = $dom->saveHTML($nodes->item(0)); | |
| 1483 | - if (!empty($content)) { | |
| 1484 | - $debug("Returning content from generic selector: " . $selector); | |
| 1485 | - return $content; | |
| 1486 | - } | |
| 1487 | - } | |
| 1488 | - } | |
| 1489 | - | |
| 1490 | - // Paragraph-based content detection - find regions with substantial text | |
| 1491 | - $debug("Trying paragraph-based content detection"); | |
| 1492 | - $paragraphs = $xpath->query('//p[string-length(normalize-space()) > 50]'); | |
| 1493 | - if ($paragraphs && $paragraphs->length >= 3) { | |
| 1494 | - $debug("Found " . $paragraphs->length . " substantial paragraphs"); | |
| 1495 | - // Collect all substantial paragraphs and their content | |
| 1496 | - $paragraph_content = ''; | |
| 1497 | - foreach ($paragraphs as $p) { | |
| 1498 | - $paragraph_content .= $dom->saveHTML($p) . "\n"; | |
| 1499 | - } | |
| 1500 | - if (!empty($paragraph_content)) { | |
| 1501 | - $debug("Returning paragraph-based content"); | |
| 1502 | - return $paragraph_content; | |
| 1503 | - } | |
| 1504 | - } | |
| 1505 | - | |
| 1506 | - // Improved body fallback - strip nav/header/footer elements first | |
| 1507 | - $debug("Using improved body fallback"); | |
| 1508 | - $body = $dom->getElementsByTagName('body'); | |
| 1509 | - if ($body->length > 0) { | |
| 1510 | - // Clone the body to avoid modifying the original DOM | |
| 1511 | - $body_clone = $body->item(0)->cloneNode(true); | |
| 1512 | - | |
| 1513 | - // Remove common non-content elements by tag name | |
| 1514 | - $remove_tags = ['nav', 'header', 'footer', 'aside', 'script', 'style', 'noscript']; | |
| 1515 | - foreach ($remove_tags as $tag) { | |
| 1516 | - $elements = $body_clone->getElementsByTagName($tag); | |
| 1517 | - // Iterate backwards to safely remove elements | |
| 1518 | - for ($i = $elements->length - 1; $i >= 0; $i--) { | |
| 1519 | - $el = $elements->item($i); | |
| 1520 | - if ($el && $el->parentNode) { | |
| 1521 | - $el->parentNode->removeChild($el); | |
| 1522 | - } | |
| 1523 | - } | |
| 1524 | - } | |
| 1525 | - | |
| 1526 | - // Remove elements with common non-content class names using XPath on the cloned body | |
| 1527 | - $temp_dom = new DOMDocument(); | |
| 1528 | - @$temp_dom->appendChild($temp_dom->importNode($body_clone, true)); | |
| 1529 | - $temp_xpath = new DOMXPath($temp_dom); | |
| 1530 | - | |
| 1531 | - $remove_class_patterns = [ | |
| 1532 | - '//*[contains(@class, "nav")]', | |
| 1533 | - '//*[contains(@class, "menu")]', | |
| 1534 | - '//*[contains(@class, "sidebar")]', | |
| 1535 | - '//*[contains(@class, "footer")]', | |
| 1536 | - '//*[contains(@class, "header")]', | |
| 1537 | - '//*[contains(@id, "nav")]', | |
| 1538 | - '//*[contains(@id, "menu")]', | |
| 1539 | - '//*[contains(@id, "sidebar")]', | |
| 1540 | - '//*[contains(@id, "footer")]', | |
| 1541 | - '//*[contains(@id, "header")]', | |
| 1542 | - ]; | |
| 1543 | - | |
| 1544 | - foreach ($remove_class_patterns as $pattern) { | |
| 1545 | - $elements = $temp_xpath->query($pattern); | |
| 1546 | - if ($elements) { | |
| 1547 | - for ($i = $elements->length - 1; $i >= 0; $i--) { | |
| 1548 | - $el = $elements->item($i); | |
| 1549 | - if ($el && $el->parentNode) { | |
| 1550 | - $el->parentNode->removeChild($el); | |
| 1551 | - } | |
| 1552 | - } | |
| 1553 | - } | |
| 1554 | - } | |
| 1555 | - | |
| 1556 | - $cleaned_content = $temp_dom->saveHTML(); | |
| 1557 | - if (!empty($cleaned_content)) { | |
| 1558 | - $debug("Returning cleaned body content"); | |
| 1559 | - return $cleaned_content; | |
| 1560 | - } | |
| 1561 | - } | |
| 1562 | - | |
| 1563 | - // Last resort: return the original HTML | |
| 1564 | - $debug("Returning original HTML"); | |
| 1565 | - return $html; | |
| 1566 | - } catch (Exception $e) { | |
| 1567 | - //error_log('[MXCHAT-ERROR] Content extraction failed: ' . $e->getMessage()); | |
| 1568 | - return $html; // Return original HTML if parsing fails | |
| 1569 | - } finally { | |
| 1570 | - libxml_clear_errors(); | |
| 1571 | - } | |
| 1572 | -} | |
| 1573 | -public function mxchat_get_sitemap_processing_status($sitemap_url) { | |
| 1574 | - $sitemap_url = esc_url_raw($sitemap_url); | |
| 1575 | - $status_key = sanitize_key('mxchat_sitemap_status_' . md5($sitemap_url)); | |
| 1576 | - $status = get_transient($status_key); | |
| 1577 | - | |
| 1578 | - if (!$status || !is_array($status)) { | |
| 1579 | - return false; | |
| 1580 | - } | |
| 1581 | - | |
| 1582 | - // Auto-complete check: if all URLs are processed but status isn't complete | |
| 1583 | - if (isset($status['processed_urls']) && isset($status['total_urls']) && | |
| 1584 | - $status['processed_urls'] >= $status['total_urls'] && | |
| 1585 | - isset($status['status']) && $status['status'] !== 'complete' && | |
| 1586 | - $status['status'] !== 'error') { | |
| 1587 | - | |
| 1588 | - // Mark as complete | |
| 1589 | - $status['status'] = 'complete'; | |
| 1590 | - $status['processed_urls'] = $status['total_urls']; // Ensure exact match | |
| 1591 | - | |
| 1592 | - // Update the transient with the corrected status | |
| 1593 | - set_transient($status_key, $status, DAY_IN_SECONDS); | |
| 1594 | - } | |
| 1595 | - | |
| 1596 | - return array( | |
| 1597 | - 'total_urls' => absint($status['total_urls']), | |
| 1598 | - 'processed_urls' => absint($status['processed_urls']), | |
| 1599 | - 'failed_urls' => absint($status['failed_urls'] ?? 0), | |
| 1600 | - 'percentage' => ($status['total_urls'] > 0) | |
| 1601 | - ? round((absint($status['processed_urls']) / absint($status['total_urls'])) * 100) | |
| 1602 | - : 0, | |
| 1603 | - 'status' => sanitize_text_field($status['status']), | |
| 1604 | - 'last_update' => human_time_diff(absint($status['last_update']), time()) . ' ' . esc_html__('ago', 'mxchat'), | |
| 1605 | - 'error' => isset($status['error']) ? sanitize_text_field($status['error']) : '', | |
| 1606 | - 'last_error' => isset($status['last_error']) ? sanitize_text_field($status['last_error']) : '', | |
| 1607 | - 'failed_urls_list' => isset($status['failed_urls_list']) ? $status['failed_urls_list'] : array() | |
| 1608 | - ); | |
| 1609 | -} | |
| 1610 | - | |
| 1611 | -public function mxchat_ajax_get_status_updates() { | |
| 1612 | - try { | |
| 1613 | - // Verify the request | |
| 1614 | - check_ajax_referer('mxchat_status_nonce', 'nonce'); | |
| 1615 | - | |
| 1616 | - // Get active queue IDs | |
| 1617 | - $sitemap_queue_id = get_transient('mxchat_active_queue_sitemap'); | |
| 1618 | - $pdf_queue_id = get_transient('mxchat_active_queue_pdf'); | |
| 1619 | - | |
| 1620 | - $sitemap_status = false; | |
| 1621 | - $pdf_status = false; | |
| 1622 | - | |
| 1623 | - // Get sitemap queue status | |
| 1624 | - if ($sitemap_queue_id) { | |
| 1625 | - $sitemap_status = $this->mxchat_get_queue_status_data($sitemap_queue_id, 'sitemap'); | |
| 1626 | - } | |
| 1627 | - | |
| 1628 | - // Get PDF queue status | |
| 1629 | - if ($pdf_queue_id) { | |
| 1630 | - $pdf_status = $this->mxchat_get_queue_status_data($pdf_queue_id, 'pdf'); | |
| 1631 | - } | |
| 1632 | - | |
| 1633 | - $is_active_processing = | |
| 1634 | - ($sitemap_status && $sitemap_status['status'] === 'processing') || | |
| 1635 | - ($pdf_status && $pdf_status['status'] === 'processing'); | |
| 1636 | - | |
| 1637 | - // Return JSON response with the status data | |
| 1638 | - wp_send_json(array( | |
| 1639 | - 'pdf_status' => $pdf_status, | |
| 1640 | - 'sitemap_status' => $sitemap_status, | |
| 1641 | - 'is_processing' => $is_active_processing, | |
| 1642 | - 'sitemap_queue_id' => $sitemap_queue_id, | |
| 1643 | - 'pdf_queue_id' => $pdf_queue_id | |
| 1644 | - )); | |
| 1645 | - | |
| 1646 | - } catch (Exception $e) { | |
| 1647 | - //error_log('MxChat Status Update Error: ' . $e->getMessage()); | |
| 1648 | - | |
| 1649 | - wp_send_json_error(array( | |
| 1650 | - 'message' => 'Error getting status updates: ' . $e->getMessage(), | |
| 1651 | - 'status' => 'error' | |
| 1652 | - )); | |
| 1653 | - } | |
| 1654 | -} | |
| 1655 | - | |
| 1656 | -/** | |
| 1657 | - * Helper function to get queue status data | |
| 1658 | - */ | |
| 1659 | -private function mxchat_get_queue_status_data($queue_id, $type = 'sitemap') { | |
| 1660 | - global $wpdb; | |
| 1661 | - $table_name = $wpdb->prefix . 'mxchat_processing_queue'; | |
| 1662 | - | |
| 1663 | - // Get counts by status | |
| 1664 | - $counts = $wpdb->get_results($wpdb->prepare( | |
| 1665 | - "SELECT status, COUNT(*) as count | |
| 1666 | - FROM $table_name | |
| 1667 | - WHERE queue_id = %s | |
| 1668 | - GROUP BY status", | |
| 1669 | - $queue_id | |
| 1670 | - ), OBJECT_K); | |
| 1671 | - | |
| 1672 | - $total = 0; | |
| 1673 | - $completed = 0; | |
| 1674 | - $failed = 0; | |
| 1675 | - $processing = 0; | |
| 1676 | - $pending = 0; | |
| 1677 | - | |
| 1678 | - foreach ($counts as $status => $data) { | |
| 1679 | - $count = absint($data->count); | |
| 1680 | - $total += $count; | |
| 1681 | - | |
| 1682 | - switch ($status) { | |
| 1683 | - case 'completed': | |
| 1684 | - $completed = $count; | |
| 1685 | - break; | |
| 1686 | - case 'failed': | |
| 1687 | - $failed = $count; | |
| 1688 | - break; | |
| 1689 | - case 'processing': | |
| 1690 | - $processing = $count; | |
| 1691 | - break; | |
| 1692 | - case 'pending': | |
| 1693 | - $pending = $count; | |
| 1694 | - break; | |
| 1695 | - } | |
| 1696 | - } | |
| 1697 | - | |
| 1698 | - if ($total === 0) { | |
| 1699 | - return false; | |
| 1700 | - } | |
| 1701 | - | |
| 1702 | - // Calculate percentage | |
| 1703 | - $percentage = round((($completed + $failed) / $total) * 100); | |
| 1704 | - | |
| 1705 | - // Get failed items details (limit to 50) | |
| 1706 | - $failed_items = array(); | |
| 1707 | - if ($failed > 0) { | |
| 1708 | - $failed_results = $wpdb->get_results($wpdb->prepare( | |
| 1709 | - "SELECT item_type, item_data, error_message, attempts, completed_at | |
| 1710 | - FROM $table_name | |
| 1711 | - WHERE queue_id = %s | |
| 1712 | - AND status = 'failed' | |
| 1713 | - AND attempts >= max_attempts | |
| 1714 | - ORDER BY id DESC | |
| 1715 | - LIMIT 50", | |
| 1716 | - $queue_id | |
| 1717 | - )); | |
| 1718 | - | |
| 1719 | - foreach ($failed_results as $item) { | |
| 1720 | - $data = json_decode($item->item_data, true); | |
| 1721 | - $url = $type === 'pdf' ? ($data['pdf_url'] ?? '') : ($data['url'] ?? ''); | |
| 1722 | - | |
| 1723 | - $failed_items[] = array( | |
| 1724 | - 'url' => $url, | |
| 1725 | - 'page' => $type === 'pdf' ? ($data['page_number'] ?? 0) : 0, | |
| 1726 | - 'error' => $item->error_message, | |
| 1727 | - 'retries' => $item->attempts, | |
| 1728 | - 'time' => strtotime($item->completed_at) | |
| 1729 | - ); | |
| 1730 | - } | |
| 1731 | - } | |
| 1732 | - | |
| 1733 | - // Get queue metadata | |
| 1734 | - $source_url = $this->mxchat_get_queue_meta($queue_id, 'source_url'); | |
| 1735 | - | |
| 1736 | - // Determine if queue is complete | |
| 1737 | - $is_complete = ($pending === 0 && $processing === 0); | |
| 1738 | - | |
| 1739 | - // Get last update time | |
| 1740 | - $last_update = $wpdb->get_var($wpdb->prepare( | |
| 1741 | - "SELECT MAX(UNIX_TIMESTAMP(COALESCE(completed_at, started_at, created_at))) | |
| 1742 | - FROM $table_name | |
| 1743 | - WHERE queue_id = %s", | |
| 1744 | - $queue_id | |
| 1745 | - )); | |
| 1746 | - | |
| 1747 | - $last_update_text = $last_update ? human_time_diff($last_update, time()) . ' ' . __('ago', 'mxchat') : __('Just now', 'mxchat'); | |
| 1748 | - | |
| 1749 | - // Format based on type | |
| 1750 | - if ($type === 'pdf') { | |
| 1751 | - return array( | |
| 1752 | - 'total_pages' => $total, | |
| 1753 | - 'processed_pages' => $completed + $failed, | |
| 1754 | - 'failed_pages' => $failed, | |
| 1755 | - 'percentage' => $percentage, | |
| 1756 | - 'status' => $is_complete ? 'complete' : 'processing', | |
| 1757 | - 'last_update' => $last_update_text, | |
| 1758 | - 'failed_pages_list' => $failed_items, | |
| 1759 | - 'pdf_url' => $source_url, | |
| 1760 | - 'queue_id' => $queue_id | |
| 1761 | - ); | |
| 1762 | - } else { | |
| 1763 | - return array( | |
| 1764 | - 'total_urls' => $total, | |
| 1765 | - 'processed_urls' => $completed + $failed, | |
| 1766 | - 'failed_urls' => $failed, | |
| 1767 | - 'percentage' => $percentage, | |
| 1768 | - 'status' => $is_complete ? 'complete' : 'processing', | |
| 1769 | - 'last_update' => $last_update_text, | |
| 1770 | - 'failed_urls_list' => $failed_items, | |
| 1771 | - 'sitemap_url' => $source_url, | |
| 1772 | - 'queue_id' => $queue_id | |
| 1773 | - ); | |
| 1774 | - } | |
| 1775 | -} | |
| 1776 | - | |
| 1777 | -/** | |
| 1778 | - * Public method to get processing status for both sitemap and PDF queues | |
| 1779 | - * Used by admin pages to display processing status | |
| 1780 | - * | |
| 1781 | - * @return array Array with 'sitemap_status', 'pdf_status', and 'is_processing' keys | |
| 1782 | - */ | |
| 1783 | -public function mxchat_get_processing_statuses() { | |
| 1784 | - $sitemap_queue_id = get_transient('mxchat_active_queue_sitemap'); | |
| 1785 | - $pdf_queue_id = get_transient('mxchat_active_queue_pdf'); | |
| 1786 | - | |
| 1787 | - $sitemap_status = false; | |
| 1788 | - $pdf_status = false; | |
| 1789 | - | |
| 1790 | - if ($sitemap_queue_id) { | |
| 1791 | - $sitemap_status = $this->mxchat_get_queue_status_data($sitemap_queue_id, 'sitemap'); | |
| 1792 | - } | |
| 1793 | - | |
| 1794 | - if ($pdf_queue_id) { | |
| 1795 | - $pdf_status = $this->mxchat_get_queue_status_data($pdf_queue_id, 'pdf'); | |
| 1796 | - } | |
| 1797 | - | |
| 1798 | - $is_processing = | |
| 1799 | - ($sitemap_status && $sitemap_status['status'] === 'processing') || | |
| 1800 | - ($pdf_status && $pdf_status['status'] === 'processing'); | |
| 1801 | - | |
| 1802 | - return array( | |
| 1803 | - 'sitemap_status' => $sitemap_status, | |
| 1804 | - 'pdf_status' => $pdf_status, | |
| 1805 | - 'is_processing' => $is_processing | |
| 1806 | - ); | |
| 1807 | -} | |
| 1808 | - | |
| 1809 | -/** | |
| 1810 | - * AJAX handler to get recent knowledge entries for real-time table updates | |
| 1811 | - * UPDATED: Now supports both WordPress DB and Pinecone data sources | |
| 1812 | - */ | |
| 1813 | -public function ajax_mxchat_get_recent_entries() { | |
| 1814 | - check_ajax_referer('mxchat_entries_nonce', 'nonce'); | |
| 1815 | - | |
| 1816 | - if (!current_user_can('manage_options')) { | |
| 1817 | - wp_send_json_error(array('message' => 'Unauthorized')); | |
| 1818 | - return; | |
| 1819 | - } | |
| 1820 | - | |
| 1821 | - global $wpdb; | |
| 1822 | - $table_name = $wpdb->prefix . 'mxchat_system_prompt_content'; | |
| 1823 | - | |
| 1824 | - // Get parameters | |
| 1825 | - $last_id = isset($_POST['last_id']) ? absint($_POST['last_id']) : 0; | |
| 1826 | - $limit = isset($_POST['limit']) ? min(absint($_POST['limit']), 50) : 10; | |
| 1827 | - $bot_id = isset($_POST['bot_id']) ? sanitize_text_field($_POST['bot_id']) : 'default'; | |
| 1828 | - | |
| 1829 | - // Check if Pinecone is enabled for this bot | |
| 1830 | - $pinecone_manager = $this->mxchat_get_pinecone_manager(); | |
| 1831 | - $pinecone_options = $pinecone_manager ? $pinecone_manager->mxchat_get_bot_pinecone_options($bot_id) : array(); | |
| 1832 | - $use_pinecone = ($pinecone_options['mxchat_use_pinecone'] ?? '0') === '1'; | |
| 1833 | - $has_pinecone_api = !empty($pinecone_options['mxchat_pinecone_api_key']); | |
| 1834 | - | |
| 1835 | - if ($use_pinecone && $has_pinecone_api) { | |
| 1836 | - // PINECONE DATA SOURCE - Get grouped entry count (not raw vector count) | |
| 1837 | - // Use mxchat_fetch_pinecone_records which returns total_unique_entries | |
| 1838 | - $records = $pinecone_manager->mxchat_fetch_pinecone_records($pinecone_options, '', 1, 10, $bot_id, ''); | |
| 1839 | - $total_count = $records['total'] ?? 0; | |
| 1840 | - | |
| 1841 | - // For Pinecone, we don't return individual entries during polling | |
| 1842 | - // (entries are already displayed on page load via mxchat_fetch_pinecone_records) | |
| 1843 | - // We just return the updated count | |
| 1844 | - wp_send_json_success(array( | |
| 1845 | - 'entries' => array(), | |
| 1846 | - 'total_count' => absint($total_count), | |
| 1847 | - 'max_id' => $last_id, | |
| 1848 | - 'data_source' => 'pinecone' | |
| 1849 | - )); | |
| 1850 | - return; | |
| 1851 | - } | |
| 1852 | - | |
| 1853 | - // WORDPRESS DB DATA SOURCE | |
| 1854 | - // Build query to get entries newer than last_id | |
| 1855 | - $where_clauses = array('1=1'); | |
| 1856 | - $where_values = array(); | |
| 1857 | - | |
| 1858 | - if ($last_id > 0) { | |
| 1859 | - $where_clauses[] = 'id > %d'; | |
| 1860 | - $where_values[] = $last_id; | |
| 1861 | - } | |
| 1862 | - | |
| 1863 | - // Note: WordPress DB table doesn't have bot_id column | |
| 1864 | - // Multi-bot filtering is handled via Pinecone namespaces | |
| 1865 | - | |
| 1866 | - $where_sql = implode(' AND ', $where_clauses); | |
| 1867 | - | |
| 1868 | - // Get recent entries | |
| 1869 | - $query = "SELECT id, article_content, source_url, timestamp | |
| 1870 | - FROM $table_name | |
| 1871 | - WHERE $where_sql | |
| 1872 | - ORDER BY id DESC | |
| 1873 | - LIMIT %d"; | |
| 1874 | - | |
| 1875 | - $where_values[] = $limit; | |
| 1876 | - | |
| 1877 | - $entries = $wpdb->get_results($wpdb->prepare($query, $where_values)); | |
| 1878 | - | |
| 1879 | - // Get total count of GROUPED entries (by source_url) - matches pagination display | |
| 1880 | - // Count unique source_urls (excluding mxchat:// internal refs) + count of ungrouped rows | |
| 1881 | - $total_count = $wpdb->get_var( | |
| 1882 | - "SELECT (SELECT COUNT(DISTINCT source_url) FROM {$table_name} WHERE source_url != '' AND source_url NOT LIKE 'mxchat://%') + | |
| 1883 | - (SELECT COUNT(*) FROM {$table_name} WHERE source_url = '' OR source_url IS NULL OR source_url LIKE 'mxchat://%')" | |
| 1884 | - ); | |
| 1885 | - | |
| 1886 | - // Format entries for response | |
| 1887 | - $formatted_entries = array(); | |
| 1888 | - $preview_length = 150; | |
| 1889 | - foreach ($entries as $entry) { | |
| 1890 | - // Parse chunk metadata using the proper chunker method (same as initial page load) | |
| 1891 | - if (class_exists('MxChat_Chunker')) { | |
| 1892 | - $chunk_meta = MxChat_Chunker::parse_stored_chunk($entry->article_content); | |
| 1893 | - $display_content = $chunk_meta['text']; | |
| 1894 | - $chunk_metadata = $chunk_meta['metadata']; | |
| 1895 | - } else { | |
| 1896 | - $display_content = $entry->article_content; | |
| 1897 | - $chunk_metadata = array(); | |
| 1898 | - } | |
| 1899 | - | |
| 1900 | - $content_preview = mb_strlen($display_content) > $preview_length | |
| 1901 | - ? mb_substr($display_content, 0, $preview_length) . '...' | |
| 1902 | - : $display_content; | |
| 1903 | - | |
| 1904 | - $formatted_entries[] = array( | |
| 1905 | - 'id' => $entry->id, | |
| 1906 | - 'preview' => esc_html($content_preview), | |
| 1907 | - 'full_content' => wp_kses_post(wpautop($display_content)), | |
| 1908 | - 'content_length' => mb_strlen($display_content), | |
| 1909 | - 'preview_length' => $preview_length, | |
| 1910 | - 'source_url' => $entry->source_url, | |
| 1911 | - 'has_link' => !empty($entry->source_url) && strpos($entry->source_url, 'mxchat://') !== 0, | |
| 1912 | - 'chunk_metadata' => $chunk_metadata, | |
| 1913 | - 'bot_id' => $entry->bot_id ?? 'default', | |
| 1914 | - 'edit_nonce' => wp_create_nonce('mxchat_edit_entry_nonce'), | |
| 1915 | - 'delete_nonce' => wp_create_nonce('mxchat_delete_prompt_nonce') | |
| 1916 | - ); | |
| 1917 | - } | |
| 1918 | - | |
| 1919 | - wp_send_json_success(array( | |
| 1920 | - 'entries' => $formatted_entries, | |
| 1921 | - 'total_count' => absint($total_count), | |
| 1922 | - 'max_id' => !empty($entries) ? $entries[0]->id : $last_id, | |
| 1923 | - 'data_source' => 'wordpress' | |
| 1924 | - )); | |
| 1925 | -} | |
| 1926 | - | |
| 1927 | -/** | |
| 1928 | - * Get Pinecone total count from stats API | |
| 1929 | - * Helper function for ajax_mxchat_get_recent_entries | |
| 1930 | - */ | |
| 1931 | -private function mxchat_get_pinecone_count_from_stats($pinecone_options) { | |
| 1932 | - $api_key = $pinecone_options['mxchat_pinecone_api_key'] ?? ''; | |
| 1933 | - $host = $pinecone_options['mxchat_pinecone_host'] ?? ''; | |
| 1934 | - $namespace = $pinecone_options['mxchat_pinecone_namespace'] ?? ''; | |
| 1935 | - | |
| 1936 | - if (empty($api_key) || empty($host)) { | |
| 1937 | - return 0; | |
| 1938 | - } | |
| 1939 | - | |
| 1940 | - try { | |
| 1941 | - $stats_url = "https://{$host}/describe_index_stats"; | |
| 1942 | - | |
| 1943 | - $response = wp_remote_post($stats_url, array( | |
| 1944 | - 'headers' => array( | |
| 1945 | - 'Api-Key' => $api_key, | |
| 1946 | - 'Content-Type' => 'application/json' | |
| 1947 | - ), | |
| 1948 | - 'body' => '{}', | |
| 1949 | - 'timeout' => 10 | |
| 1950 | - )); | |
| 1951 | - | |
| 1952 | - if (!is_wp_error($response) && wp_remote_retrieve_response_code($response) === 200) { | |
| 1953 | - $body = wp_remote_retrieve_body($response); | |
| 1954 | - $stats_data = json_decode($body, true); | |
| 1955 | - | |
| 1956 | - // If namespace is specified, get count from that specific namespace | |
| 1957 | - if (!empty($namespace) && isset($stats_data['namespaces'][$namespace]['vectorCount'])) { | |
| 1958 | - return intval($stats_data['namespaces'][$namespace]['vectorCount']); | |
| 1959 | - } | |
| 1960 | - | |
| 1961 | - // If no namespace specified or namespace not found in response, use total | |
| 1962 | - return intval($stats_data['totalVectorCount'] ?? 0); | |
| 1963 | - } | |
| 1964 | - | |
| 1965 | - return 0; | |
| 1966 | - | |
| 1967 | - } catch (Exception $e) { | |
| 1968 | - return 0; | |
| 1969 | - } | |
| 1970 | -} | |
| 1971 | - | |
| 1972 | -/** | |
| 1973 | - * AJAX handler to refresh Pinecone entries table via AJAX | |
| 1974 | - * Returns the table HTML for updating the UI without a full page reload | |
| 1975 | - */ | |
| 1976 | -public function ajax_mxchat_refresh_pinecone_entries() { | |
| 1977 | - check_ajax_referer('mxchat_entries_nonce', 'nonce'); | |
| 1978 | - | |
| 1979 | - if (!current_user_can('manage_options')) { | |
| 1980 | - wp_send_json_error(array('message' => 'Unauthorized')); | |
| 1981 | - return; | |
| 1982 | - } | |
| 1983 | - | |
| 1984 | - $bot_id = isset($_POST['bot_id']) ? sanitize_text_field($_POST['bot_id']) : 'default'; | |
| 1985 | - $page = isset($_POST['page']) ? absint($_POST['page']) : 1; | |
| 1986 | - $per_page = 25; | |
| 1987 | - $search_query = isset($_POST['search']) ? sanitize_text_field($_POST['search']) : ''; | |
| 1988 | - $content_type_filter = isset($_POST['content_type']) ? sanitize_key($_POST['content_type']) : ''; | |
| 1989 | - | |
| 1990 | - // Get Pinecone manager and options | |
| 1991 | - $pinecone_manager = $this->mxchat_get_pinecone_manager(); | |
| 1992 | - if (!$pinecone_manager) { | |
| 1993 | - wp_send_json_error(array('message' => 'Pinecone manager not available')); | |
| 1994 | - return; | |
| 1995 | - } | |
| 1996 | - | |
| 1997 | - $pinecone_options = $pinecone_manager->mxchat_get_bot_pinecone_options($bot_id); | |
| 1998 | - $use_pinecone = ($pinecone_options['mxchat_use_pinecone'] ?? '0') === '1'; | |
| 1999 | - $pinecone_api_key = $pinecone_options['mxchat_pinecone_api_key'] ?? ''; | |
| 2000 | - | |
| 2001 | - if (!$use_pinecone || empty($pinecone_api_key)) { | |
| 2002 | - wp_send_json_error(array('message' => 'Pinecone not configured')); | |
| 2003 | - return; | |
| 2004 | - } | |
| 2005 | - | |
| 2006 | - // Fetch records from Pinecone | |
| 2007 | - $records = $pinecone_manager->mxchat_fetch_pinecone_records($pinecone_options, $search_query, $page, $per_page, $bot_id, $content_type_filter); | |
| 2008 | - $prompts = $records['data'] ?? array(); | |
| 2009 | - $total_records = $records['total'] ?? 0; | |
| 2010 | - | |
| 2011 | - // Preprocess Pinecone records — set chunk_metadata and display_content | |
| 2012 | - // (matches admin-knowledge-page.php preprocessing) | |
| 2013 | - foreach ($prompts as $prompt) { | |
| 2014 | - if (isset($prompt->chunk_index) && $prompt->chunk_index !== null) { | |
| 2015 | - $prompt->chunk_metadata = array( | |
| 2016 | - 'chunk_index' => intval($prompt->chunk_index), | |
| 2017 | - 'total_chunks' => isset($prompt->total_chunks) ? intval($prompt->total_chunks) : null, | |
| 2018 | - 'is_chunked' => isset($prompt->is_chunked) ? (bool) $prompt->is_chunked : true, | |
| 2019 | - 'source_url' => $prompt->source_url ?? '' | |
| 2020 | - ); | |
| 2021 | - $prompt->display_content = $prompt->article_content; | |
| 2022 | - } else { | |
| 2023 | - $prompt->chunk_metadata = array(); | |
| 2024 | - $prompt->display_content = $prompt->article_content ?? ''; | |
| 2025 | - } | |
| 2026 | - } | |
| 2027 | - | |
| 2028 | - // Group prompts by source_url | |
| 2029 | - $grouped_prompts = array(); | |
| 2030 | - foreach ($prompts as $prompt) { | |
| 2031 | - $source_url = ''; | |
| 2032 | - if (!empty($prompt->chunk_metadata['source_url'])) { | |
| 2033 | - $source_url = $prompt->chunk_metadata['source_url']; | |
| 2034 | - } elseif (!empty($prompt->source_url)) { | |
| 2035 | - $source_url = $prompt->source_url; | |
| 2036 | - } | |
| 2037 | - | |
| 2038 | - if (!empty($source_url)) { | |
| 2039 | - if (!isset($grouped_prompts[$source_url])) { | |
| 2040 | - $grouped_prompts[$source_url] = array(); | |
| 2041 | - } | |
| 2042 | - $grouped_prompts[$source_url][] = $prompt; | |
| 2043 | - } else { | |
| 2044 | - $grouped_prompts['_ungrouped_' . $prompt->id] = array($prompt); | |
| 2045 | - } | |
| 2046 | - } | |
| 2047 | - | |
| 2048 | - // Sort each group by chunk_index | |
| 2049 | - foreach ($grouped_prompts as $source_url => &$group) { | |
| 2050 | - usort($group, function($a, $b) { | |
| 2051 | - $index_a = isset($a->chunk_metadata['chunk_index']) ? intval($a->chunk_metadata['chunk_index']) : 0; | |
| 2052 | - $index_b = isset($b->chunk_metadata['chunk_index']) ? intval($b->chunk_metadata['chunk_index']) : 0; | |
| 2053 | - return $index_a - $index_b; | |
| 2054 | - }); | |
| 2055 | - } | |
| 2056 | - unset($group); | |
| 2057 | - | |
| 2058 | - // Build HTML for the table rows - must match admin-knowledge-page.php structure exactly | |
| 2059 | - ob_start(); | |
| 2060 | - $display_index = 0; | |
| 2061 | - $current_page = $page; | |
| 2062 | - $data_source = 'pinecone'; | |
| 2063 | - $current_bot_id = $bot_id; | |
| 2064 | - $preview_length = 150; | |
| 2065 | - | |
| 2066 | - if (empty($grouped_prompts)) { | |
| 2067 | - echo '<tr><td colspan="4" style="padding: 40px; text-align: center; color: var(--mxch-text-muted);">'; | |
| 2068 | - esc_html_e('No knowledge entries found in Pinecone.', 'mxchat'); | |
| 2069 | - echo '</td></tr>'; | |
| 2070 | - } else { | |
| 2071 | - foreach ($grouped_prompts as $source_url => $group) { | |
| 2072 | - $chunk_count = count($group); | |
| 2073 | - $first_prompt = $group[0]; | |
| 2074 | - $display_index++; | |
| 2075 | - | |
| 2076 | - if ($chunk_count > 1) { | |
| 2077 | - // Multiple chunks - show grouped row with expand button | |
| 2078 | - $group_id = 'group-' . md5($source_url); | |
| 2079 | - ?> | |
| 2080 | - <tr id="prompt-<?php echo esc_attr($first_prompt->id); ?>" | |
| 2081 | - class="mxchat-chunk-group-header" | |
| 2082 | - data-source="<?php echo esc_attr($data_source); ?>" | |
| 2083 | - data-group-id="<?php echo esc_attr($group_id); ?>" | |
| 2084 | - style="border-bottom: 1px solid var(--mxch-card-border); background: rgba(33, 150, 243, 0.02);"> | |
| 2085 | - <td style="padding: 12px 16px; text-align: center;"> | |
| 2086 | - <input type="checkbox" | |
| 2087 | - class="mxchat-entry-checkbox" | |
| 2088 | - data-entry-id="<?php echo esc_attr($first_prompt->id); ?>" | |
| 2089 | - data-source="<?php echo esc_attr($data_source); ?>" | |
| 2090 | - data-source-url="<?php echo esc_attr($source_url); ?>" | |
| 2091 | - data-is-group="true" | |
| 2092 | - data-chunk-count="<?php echo esc_attr($chunk_count); ?>"> | |
| 2093 | - </td> | |
| 2094 | - <td style="padding: 12px 16px; font-size: 13px;"> | |
| 2095 | - <?php echo esc_html($display_index + (($current_page - 1) * $per_page)); ?> | |
| 2096 | - </td> | |
| 2097 | - <td class="mxchat-content-cell" style="padding: 12px 16px; font-size: 13px;"> | |
| 2098 | - <div class="mxchat-chunk-group-info"> | |
| 2099 | - <button type="button" class="mxchat-chunk-toggle" data-group-id="<?php echo esc_attr($group_id); ?>"> | |
| 2100 | - <span class="dashicons dashicons-arrow-right-alt2"></span> | |
| 2101 | - </button> | |
| 2102 | - <span class="mxchat-chunk-badge"><?php echo esc_html($chunk_count); ?> <?php esc_html_e('chunks', 'mxchat'); ?></span> | |
| 2103 | - <span class="mxchat-chunk-preview"> | |
| 2104 | - <?php | |
| 2105 | - $parent_content = isset($first_prompt->display_content) ? $first_prompt->display_content : (isset($first_prompt->article_content) ? $first_prompt->article_content : ''); | |
| 2106 | - $content_preview = mb_substr($parent_content, 0, 100); | |
| 2107 | - echo esc_html($content_preview . '...'); | |
| 2108 | - ?> | |
| 2109 | - </span> | |
| 2110 | - </div> | |
| 2111 | - </td> | |
| 2112 | - <td class="mxchat-url-cell" style="padding: 12px 16px; font-size: 13px;"> | |
| 2113 | - <?php if (!empty($source_url) && strpos($source_url, 'mxchat://') !== 0 && strpos($source_url, '_ungrouped_') !== 0) : ?> | |
| 2114 | - <a href="<?php echo esc_url($source_url); ?>" target="_blank" style="color: var(--mxch-primary); text-decoration: none;"> | |
| 2115 | - <span class="dashicons dashicons-external" style="font-size: 14px;"></span> | |
| 2116 | - <?php esc_html_e('View Source', 'mxchat'); ?> | |
| 2117 | - </a> | |
| 2118 | - <?php else : ?> | |
| 2119 | - <span style="color: var(--mxch-text-muted);"><?php esc_html_e('Manual Content', 'mxchat'); ?></span> | |
| 2120 | - <?php endif; ?> | |
| 2121 | - </td> | |
| 2122 | - <td class="mxchat-actions-cell" style="padding: 12px 16px; white-space: nowrap;"> | |
| 2123 | - <?php if ($data_source !== 'pinecone') : ?> | |
| 2124 | - <button type="button" | |
| 2125 | - class="mxch-btn mxch-btn-ghost mxch-btn-sm mxchat-edit-entry-btn" | |
| 2126 | - data-source-url="<?php echo esc_attr($source_url); ?>" | |
| 2127 | - data-entry-id="<?php echo esc_attr($first_prompt->id); ?>" | |
| 2128 | - data-data-source="<?php echo esc_attr($data_source); ?>" | |
| 2129 | - data-bot-id="<?php echo esc_attr($current_bot_id); ?>" | |
| 2130 | - data-nonce="<?php echo wp_create_nonce('mxchat_edit_entry_nonce'); ?>" | |
| 2131 | - title="<?php esc_attr_e('Edit content', 'mxchat'); ?>"> | |
| 2132 | - <span class="dashicons dashicons-edit" style="font-size: 14px;"></span> | |
| 2133 | - </button> | |
| 2134 | - <?php endif; ?> | |
| 2135 | - <button type="button" | |
| 2136 | - class="mxch-btn mxch-btn-ghost mxch-btn-sm delete-button-group" | |
| 2137 | - data-source-url="<?php echo esc_attr($source_url); ?>" | |
| 2138 | - data-chunk-count="<?php echo esc_attr($chunk_count); ?>" | |
| 2139 | - data-data-source="<?php echo esc_attr($data_source); ?>" | |
| 2140 | - data-bot-id="<?php echo esc_attr($current_bot_id); ?>" | |
| 2141 | - data-nonce="<?php echo wp_create_nonce('mxchat_delete_chunks_nonce'); ?>" | |
| 2142 | - style="color: var(--mxch-error);" | |
| 2143 | - title="<?php esc_attr_e('Delete all chunks', 'mxchat'); ?>"> | |
| 2144 | - <span class="dashicons dashicons-trash" style="font-size: 14px;"></span> | |
| 2145 | - </button> | |
| 2146 | - </td> | |
| 2147 | - </tr> | |
| 2148 | - <?php | |
| 2149 | - // Render hidden chunk rows | |
| 2150 | - foreach ($group as $chunk_index => $chunk) { | |
| 2151 | - $meta_chunk_index = isset($chunk->chunk_metadata['chunk_index']) ? intval($chunk->chunk_metadata['chunk_index']) : $chunk_index; | |
| 2152 | - $meta_total_chunks = isset($chunk->chunk_metadata['total_chunks']) ? intval($chunk->chunk_metadata['total_chunks']) : $chunk_count; | |
| 2153 | - $content = isset($chunk->display_content) ? $chunk->display_content : (isset($chunk->article_content) ? $chunk->article_content : ''); | |
| 2154 | - $content_preview = mb_strlen($content) > $preview_length | |
| 2155 | - ? mb_substr($content, 0, $preview_length) . '...' | |
| 2156 | - : $content; | |
| 2157 | - ?> | |
| 2158 | - <tr id="prompt-<?php echo esc_attr($chunk->id); ?>" | |
| 2159 | - class="mxchat-chunk-row <?php echo esc_attr($group_id); ?>" | |
| 2160 | - data-source="<?php echo esc_attr($data_source); ?>" | |
| 2161 | - style="display: none; background: #f8f9fa; border-bottom: 1px solid var(--mxch-card-border);"> | |
| 2162 | - <td style="padding: 12px 16px; text-align: center;"> | |
| 2163 | - <!-- Checkbox column placeholder for chunks (managed by group) --> | |
| 2164 | - </td> | |
| 2165 | - <td style="padding: 12px 16px 12px 30px; font-size: 13px;"> | |
| 2166 | - <!-- Hidden ID column for chunks --> | |
| 2167 | - </td> | |
| 2168 | - <td class="mxchat-content-cell" style="padding: 12px 16px; font-size: 13px;"> | |
| 2169 | - <div class="mxchat-accordion-wrapper"> | |
| 2170 | - <div class="mxchat-content-preview"> | |
| 2171 | - <span class="mxchat-chunk-indicator" style="margin-right: 10px; color: var(--mxch-text-secondary); font-size: 12px;"> | |
| 2172 | - <?php printf(esc_html__('Chunk %d of %d', 'mxchat'), $meta_chunk_index + 1, $meta_total_chunks); ?> | |
| 2173 | - </span> | |
| 2174 | - <span class="preview-text"><?php echo esc_html($content_preview); ?></span> | |
| 2175 | - <?php if (mb_strlen($content) > $preview_length) : ?> | |
| 2176 | - <button class="mxchat-expand-toggle" type="button"> | |
| 2177 | - <span class="dashicons dashicons-arrow-down-alt2"></span> | |
| 2178 | - </button> | |
| 2179 | - <?php endif; ?> | |
| 2180 | - </div> | |
| 2181 | - <div class="mxchat-content-full" style="display: none;"> | |
| 2182 | - <div class="content-view"> | |
| 2183 | - <?php | |
| 2184 | - if (preg_match('/[\x{0590}-\x{05FF}]/u', $content)) { | |
| 2185 | - echo '<div dir="rtl" lang="he" class="rtl-content">'; | |
| 2186 | - echo wp_kses_post(wpautop($content)); | |
| 2187 | - echo '</div>'; | |
| 2188 | - } else { | |
| 2189 | - echo wp_kses_post(wpautop($content)); | |
| 2190 | - } | |
| 2191 | - ?> | |
| 2192 | - </div> | |
| 2193 | - </div> | |
| 2194 | - </div> | |
| 2195 | - </td> | |
| 2196 | - <td class="mxchat-url-cell" style="padding: 12px 16px; font-size: 13px;"> | |
| 2197 | - <span class="mxchat-chunk-label" style="color: var(--mxch-text-muted);"><?php esc_html_e('Same as parent', 'mxchat'); ?></span> | |
| 2198 | - </td> | |
| 2199 | - <td class="mxchat-actions-cell" style="padding: 12px 16px;"> | |
| 2200 | - <span class="mxchat-chunk-label" style="color: var(--mxch-text-muted);"><?php esc_html_e('Managed by group', 'mxchat'); ?></span> | |
| 2201 | - </td> | |
| 2202 | - </tr> | |
| 2203 | - <?php | |
| 2204 | - } | |
| 2205 | - } else { | |
| 2206 | - // Single entry - display normally with accordion | |
| 2207 | - $prompt = $first_prompt; | |
| 2208 | - $content = isset($prompt->display_content) ? $prompt->display_content : (isset($prompt->article_content) ? $prompt->article_content : ''); | |
| 2209 | - $content_preview = mb_strlen($content) > $preview_length | |
| 2210 | - ? mb_substr($content, 0, $preview_length) . '...' | |
| 2211 | - : $content; | |
| 2212 | - ?> | |
| 2213 | - <tr id="prompt-<?php echo esc_attr($prompt->id); ?>" | |
| 2214 | - data-source="<?php echo esc_attr($data_source); ?>" | |
| 2215 | - style="border-bottom: 1px solid var(--mxch-card-border); background: rgba(33, 150, 243, 0.02);"> | |
| 2216 | - <td style="padding: 12px 16px; text-align: center;"> | |
| 2217 | - <input type="checkbox" | |
| 2218 | - class="mxchat-entry-checkbox" | |
| 2219 | - data-entry-id="<?php echo esc_attr($prompt->id); ?>" | |
| 2220 | - data-source="<?php echo esc_attr($data_source); ?>" | |
| 2221 | - data-source-url="<?php echo esc_attr($prompt->source_url ?? ''); ?>" | |
| 2222 | - data-is-group="false" | |
| 2223 | - data-chunk-count="1"> | |
| 2224 | - </td> | |
| 2225 | - <td style="padding: 12px 16px; font-size: 13px;"> | |
| 2226 | - <?php echo esc_html($display_index + (($current_page - 1) * $per_page)); ?> | |
| 2227 | - </td> | |
| 2228 | - <td class="mxchat-content-cell" style="padding: 12px 16px; font-size: 13px;"> | |
| 2229 | - <div class="mxchat-accordion-wrapper"> | |
| 2230 | - <div class="mxchat-content-preview"> | |
| 2231 | - <span class="preview-text"><?php echo esc_html($content_preview); ?></span> | |
| 2232 | - <?php if (mb_strlen($content) > $preview_length) : ?> | |
| 2233 | - <button class="mxchat-expand-toggle" type="button"> | |
| 2234 | - <span class="dashicons dashicons-arrow-down-alt2"></span> | |
| 2235 | - </button> | |
| 2236 | - <?php endif; ?> | |
| 2237 | - </div> | |
| 2238 | - <div class="mxchat-content-full" style="display: none;"> | |
| 2239 | - <div class="content-view"> | |
| 2240 | - <?php | |
| 2241 | - if (preg_match('/[\x{0590}-\x{05FF}]/u', $content)) { | |
| 2242 | - echo '<div dir="rtl" lang="he" class="rtl-content">'; | |
| 2243 | - echo wp_kses_post(wpautop($content)); | |
| 2244 | - echo '</div>'; | |
| 2245 | - } else { | |
| 2246 | - echo wp_kses_post(wpautop($content)); | |
| 2247 | - } | |
| 2248 | - ?> | |
| 2249 | - </div> | |
| 2250 | - </div> | |
| 2251 | - </div> | |
| 2252 | - </td> | |
| 2253 | - <td class="mxchat-url-cell" style="padding: 12px 16px; font-size: 13px;"> | |
| 2254 | - <?php | |
| 2255 | - $actual_source = $source_url; | |
| 2256 | - if (strpos($source_url, '_ungrouped_') === 0) { | |
| 2257 | - $actual_source = $prompt->source_url ?? ''; | |
| 2258 | - } | |
| 2259 | - if (!empty($actual_source) && strpos($actual_source, 'mxchat://') !== 0) : ?> | |
| 2260 | - <a href="<?php echo esc_url($actual_source); ?>" target="_blank" style="color: var(--mxch-primary); text-decoration: none;"> | |
| 2261 | - <span class="dashicons dashicons-external" style="font-size: 14px;"></span> | |
| 2262 | - <?php esc_html_e('View', 'mxchat'); ?> | |
| 2263 | - </a> | |
| 2264 | - <?php else : ?> | |
| 2265 | - <span style="color: var(--mxch-text-muted);"><?php esc_html_e('Manual', 'mxchat'); ?></span> | |
| 2266 | - <?php endif; ?> | |
| 2267 | - </td> | |
| 2268 | - <td style="padding: 12px 16px;"> | |
| 2269 | - <button type="button" class="mxch-btn mxch-btn-ghost mxch-btn-sm delete-button-ajax" data-vector-id="<?php echo esc_attr($prompt->id); ?>" data-bot-id="<?php echo esc_attr($current_bot_id); ?>" data-nonce="<?php echo wp_create_nonce('mxchat_delete_pinecone_prompt_nonce'); ?>" style="color: var(--mxch-error);"> | |
| 2270 | - <span class="dashicons dashicons-trash" style="font-size: 14px;"></span> | |
| 2271 | - </button> | |
| 2272 | - </td> | |
| 2273 | - </tr> | |
| 2274 | - <?php | |
| 2275 | - } | |
| 2276 | - } | |
| 2277 | - } | |
| 2278 | - $html = ob_get_clean(); | |
| 2279 | - | |
| 2280 | - // Generate pagination HTML for Pinecone | |
| 2281 | - $total_pages = ceil($total_records / $per_page); | |
| 2282 | - $pagination_html = ''; | |
| 2283 | - if ($total_pages > 1) { | |
| 2284 | - $pagination_html = '<div class="mxchat-ajax-pagination" data-current-page="' . esc_attr($page) . '" data-total-pages="' . esc_attr($total_pages) . '" data-search="' . esc_attr($search_query) . '" data-content-type="' . esc_attr($content_type_filter) . '">'; | |
| 2285 | - | |
| 2286 | - // Previous button | |
| 2287 | - if ($page > 1) { | |
| 2288 | - $pagination_html .= '<a href="#" class="mxchat-page-link" data-page="' . ($page - 1) . '">' . esc_html__('« Previous', 'mxchat') . '</a> '; | |
| 2289 | - } | |
| 2290 | - | |
| 2291 | - // Page numbers | |
| 2292 | - $start_page = max(1, $page - 2); | |
| 2293 | - $end_page = min($total_pages, $page + 2); | |
| 2294 | - | |
| 2295 | - if ($start_page > 1) { | |
| 2296 | - $pagination_html .= '<a href="#" class="mxchat-page-link" data-page="1">1</a> '; | |
| 2297 | - if ($start_page > 2) { | |
| 2298 | - $pagination_html .= '<span class="mxchat-page-dots">...</span> '; | |
| 2299 | - } | |
| 2300 | - } | |
| 2301 | - | |
| 2302 | - for ($i = $start_page; $i <= $end_page; $i++) { | |
| 2303 | - if ($i == $page) { | |
| 2304 | - $pagination_html .= '<span class="mxchat-page-current">' . $i . '</span> '; | |
| 2305 | - } else { | |
| 2306 | - $pagination_html .= '<a href="#" class="mxchat-page-link" data-page="' . $i . '">' . $i . '</a> '; | |
| 2307 | - } | |
| 2308 | - } | |
| 2309 | - | |
| 2310 | - if ($end_page < $total_pages) { | |
| 2311 | - if ($end_page < $total_pages - 1) { | |
| 2312 | - $pagination_html .= '<span class="mxchat-page-dots">...</span> '; | |
| 2313 | - } | |
| 2314 | - $pagination_html .= '<a href="#" class="mxchat-page-link" data-page="' . $total_pages . '">' . $total_pages . '</a> '; | |
| 2315 | - } | |
| 2316 | - | |
| 2317 | - // Next button | |
| 2318 | - if ($page < $total_pages) { | |
| 2319 | - $pagination_html .= '<a href="#" class="mxchat-page-link" data-page="' . ($page + 1) . '">' . esc_html__('Next »', 'mxchat') . '</a>'; | |
| 2320 | - } | |
| 2321 | - | |
| 2322 | - $pagination_html .= '</div>'; | |
| 2323 | - } | |
| 2324 | - | |
| 2325 | - wp_send_json_success(array( | |
| 2326 | - 'html' => $html, | |
| 2327 | - 'pagination_html' => $pagination_html, | |
| 2328 | - 'total_count' => $total_records, | |
| 2329 | - 'total_pages' => $total_pages, | |
| 2330 | - 'page' => $page, | |
| 2331 | - 'per_page' => $per_page, | |
| 2332 | - 'data_source' => 'pinecone' | |
| 2333 | - )); | |
| 2334 | -} | |
| 2335 | - | |
| 2336 | -/** | |
| 2337 | - * AJAX handler for pagination - handles both WordPress DB and Pinecone data sources | |
| 2338 | - * Returns paginated entries without requiring a full page reload | |
| 2339 | - */ | |
| 2340 | -public function ajax_mxchat_paginate_entries() { | |
| 2341 | - check_ajax_referer('mxchat_entries_nonce', 'nonce'); | |
| 2342 | - | |
| 2343 | - if (!current_user_can('manage_options')) { | |
| 2344 | - wp_send_json_error(array('message' => 'Unauthorized')); | |
| 2345 | - return; | |
| 2346 | - } | |
| 2347 | - | |
| 2348 | - $bot_id = isset($_POST['bot_id']) ? sanitize_text_field($_POST['bot_id']) : 'default'; | |
| 2349 | - $page = isset($_POST['page']) ? absint($_POST['page']) : 1; | |
| 2350 | - $search_query = isset($_POST['search']) ? sanitize_text_field($_POST['search']) : ''; | |
| 2351 | - $content_type_filter = isset($_POST['content_type']) ? sanitize_key($_POST['content_type']) : ''; | |
| 2352 | - $per_page = 25; | |
| 2353 | - | |
| 2354 | - // Check if Pinecone is enabled for this bot | |
| 2355 | - $pinecone_manager = $this->mxchat_get_pinecone_manager(); | |
| 2356 | - $pinecone_options = $pinecone_manager ? $pinecone_manager->mxchat_get_bot_pinecone_options($bot_id) : array(); | |
| 2357 | - $use_pinecone = ($pinecone_options['mxchat_use_pinecone'] ?? '0') === '1'; | |
| 2358 | - $has_pinecone_api = !empty($pinecone_options['mxchat_pinecone_api_key']); | |
| 2359 | - | |
| 2360 | - if ($use_pinecone && $has_pinecone_api) { | |
| 2361 | - // Delegate to Pinecone pagination handler (pass search params) | |
| 2362 | - $_POST['page'] = $page; | |
| 2363 | - $_POST['search'] = $search_query; | |
| 2364 | - $_POST['content_type'] = $content_type_filter; | |
| 2365 | - $this->ajax_mxchat_refresh_pinecone_entries(); | |
| 2366 | - return; | |
| 2367 | - } | |
| 2368 | - | |
| 2369 | - // WordPress DB pagination - MUST match initial page load logic exactly | |
| 2370 | - global $wpdb; | |
| 2371 | - $table_name = $wpdb->prefix . 'mxchat_system_prompt_content'; | |
| 2372 | - $offset = ($page - 1) * $per_page; | |
| 2373 | - | |
| 2374 | - // Build WHERE clause for search and content type filtering | |
| 2375 | - $where_clauses = array(); | |
| 2376 | - $where_values = array(); | |
| 2377 | - | |
| 2378 | - if ($search_query) { | |
| 2379 | - $where_clauses[] = "article_content LIKE %s"; | |
| 2380 | - $where_values[] = '%' . $wpdb->esc_like($search_query) . '%'; | |
| 2381 | - } | |
| 2382 | - | |
| 2383 | - if ($content_type_filter) { | |
| 2384 | - switch ($content_type_filter) { | |
| 2385 | - case 'manual': | |
| 2386 | - $where_clauses[] = "(source_url = '' OR source_url IS NULL OR source_url LIKE 'mxchat://%')"; | |
| 2387 | - break; | |
| 2388 | - case 'pdf': | |
| 2389 | - $where_clauses[] = "source_url LIKE '%.pdf'"; | |
| 2390 | - break; | |
| 2391 | - case 'url': | |
| 2392 | - $where_clauses[] = "source_url != '' AND source_url IS NOT NULL AND source_url NOT LIKE 'mxchat://%' AND source_url NOT LIKE '%.pdf'"; | |
| 2393 | - break; | |
| 2394 | - } | |
| 2395 | - } | |
| 2396 | - | |
| 2397 | - $where_sql = !empty($where_clauses) ? 'WHERE ' . implode(' AND ', $where_clauses) : ''; | |
| 2398 | - | |
| 2399 | - // Count grouped entries with filters applied | |
| 2400 | - if (!empty($where_values)) { | |
| 2401 | - $count_args = array_merge($where_values, $where_values); | |
| 2402 | - $count_query = $wpdb->prepare( | |
| 2403 | - "SELECT (SELECT COUNT(DISTINCT source_url) FROM {$table_name} {$where_sql} AND source_url != '' AND source_url NOT LIKE 'mxchat://%') + | |
| 2404 | - (SELECT COUNT(*) FROM {$table_name} {$where_sql} AND (source_url = '' OR source_url IS NULL OR source_url LIKE 'mxchat://%'))", | |
| 2405 | - ...$count_args | |
| 2406 | - ); | |
| 2407 | - $total_records = $wpdb->get_var($count_query); | |
| 2408 | - } else if (!empty($where_sql)) { | |
| 2409 | - // Content type filter only (no search), no prepared values needed | |
| 2410 | - $total_records = $wpdb->get_var( | |
| 2411 | - "SELECT (SELECT COUNT(DISTINCT source_url) FROM {$table_name} {$where_sql} AND source_url != '' AND source_url NOT LIKE 'mxchat://%') + | |
| 2412 | - (SELECT COUNT(*) FROM {$table_name} {$where_sql} AND (source_url = '' OR source_url IS NULL OR source_url LIKE 'mxchat://%'))" | |
| 2413 | - ); | |
| 2414 | - } else { | |
| 2415 | - // No filters | |
| 2416 | - $total_records = $wpdb->get_var( | |
| 2417 | - "SELECT (SELECT COUNT(DISTINCT source_url) FROM {$table_name} WHERE source_url != '' AND source_url NOT LIKE 'mxchat://%') + | |
| 2418 | - (SELECT COUNT(*) FROM {$table_name} WHERE source_url = '' OR source_url IS NULL OR source_url LIKE 'mxchat://%')" | |
| 2419 | - ); | |
| 2420 | - } | |
| 2421 | - $total_pages = ceil($total_records / $per_page); | |
| 2422 | - | |
| 2423 | - // Step 1: Get unique source_urls for this page (ordered by latest timestamp) with filters | |
| 2424 | - if (!empty($where_values)) { | |
| 2425 | - $query_args = array_merge($where_values, array($per_page, $offset)); | |
| 2426 | - $urls_query = $wpdb->prepare( | |
| 2427 | - "SELECT source_url, MAX(timestamp) as latest_ts FROM {$table_name} | |
| 2428 | - {$where_sql} | |
| 2429 | - GROUP BY source_url ORDER BY latest_ts DESC LIMIT %d OFFSET %d", | |
| 2430 | - ...$query_args | |
| 2431 | - ); | |
| 2432 | - } else if (!empty($where_sql)) { | |
| 2433 | - $urls_query = $wpdb->prepare( | |
| 2434 | - "SELECT source_url, MAX(timestamp) as latest_ts FROM {$table_name} | |
| 2435 | - {$where_sql} | |
| 2436 | - GROUP BY source_url ORDER BY latest_ts DESC LIMIT %d OFFSET %d", | |
| 2437 | - $per_page, $offset | |
| 2438 | - ); | |
| 2439 | - } else { | |
| 2440 | - $urls_query = $wpdb->prepare( | |
| 2441 | - "SELECT source_url, MAX(timestamp) as latest_ts FROM {$table_name} | |
| 2442 | - GROUP BY source_url ORDER BY latest_ts DESC LIMIT %d OFFSET %d", | |
| 2443 | - $per_page, $offset | |
| 2444 | - ); | |
| 2445 | - } | |
| 2446 | - $page_urls = $wpdb->get_results($urls_query); | |
| 2447 | - | |
| 2448 | - // Step 2: Build list of source_urls to fetch | |
| 2449 | - $url_list = array(); | |
| 2450 | - $url_order_map = array(); | |
| 2451 | - $order_index = 0; | |
| 2452 | - foreach ($page_urls as $url_row) { | |
| 2453 | - $url = $url_row->source_url; | |
| 2454 | - $url_list[] = $url; | |
| 2455 | - $url_order_map[$url] = $order_index++; | |
| 2456 | - } | |
| 2457 | - | |
| 2458 | - // Step 3: Fetch all rows for these source_urls (with search filter if applicable) | |
| 2459 | - $prompts = array(); | |
| 2460 | - if (!empty($url_list)) { | |
| 2461 | - $placeholders = implode(',', array_fill(0, count($url_list), '%s')); | |
| 2462 | - if ($search_query) { | |
| 2463 | - // Include search filter in the final fetch | |
| 2464 | - $prompts_query = $wpdb->prepare( | |
| 2465 | - "SELECT id, article_content, source_url, timestamp, role_restriction | |
| 2466 | - FROM {$table_name} | |
| 2467 | - WHERE source_url IN ($placeholders) AND article_content LIKE %s | |
| 2468 | - ORDER BY timestamp DESC", | |
| 2469 | - ...array_merge($url_list, ['%' . $wpdb->esc_like($search_query) . '%']) | |
| 2470 | - ); | |
| 2471 | - } else { | |
| 2472 | - $prompts_query = $wpdb->prepare( | |
| 2473 | - "SELECT id, article_content, source_url, timestamp, role_restriction | |
| 2474 | - FROM {$table_name} | |
| 2475 | - WHERE source_url IN ($placeholders) | |
| 2476 | - ORDER BY timestamp DESC", | |
| 2477 | - $url_list | |
| 2478 | - ); | |
| 2479 | - } | |
| 2480 | - $prompts = $wpdb->get_results($prompts_query); | |
| 2481 | - } | |
| 2482 | - | |
| 2483 | - // Group prompts by source_url for chunk display | |
| 2484 | - $grouped_prompts = array(); | |
| 2485 | - foreach ($prompts as $prompt) { | |
| 2486 | - $source_url = $prompt->source_url ?? ''; | |
| 2487 | - | |
| 2488 | - // Parse chunk metadata using the proper chunker method (same as initial page load) | |
| 2489 | - if (class_exists('MxChat_Chunker')) { | |
| 2490 | - $chunk_meta = MxChat_Chunker::parse_stored_chunk($prompt->article_content); | |
| 2491 | - $prompt->chunk_metadata = $chunk_meta['metadata']; | |
| 2492 | - $prompt->display_content = $chunk_meta['text']; | |
| 2493 | - } else { | |
| 2494 | - $prompt->chunk_metadata = array(); | |
| 2495 | - $prompt->display_content = $prompt->article_content; | |
| 2496 | - } | |
| 2497 | - | |
| 2498 | - if (!empty($source_url) && strpos($source_url, 'mxchat://') !== 0) { | |
| 2499 | - if (!isset($grouped_prompts[$source_url])) { | |
| 2500 | - $grouped_prompts[$source_url] = array(); | |
| 2501 | - } | |
| 2502 | - $grouped_prompts[$source_url][] = $prompt; | |
| 2503 | - } else { | |
| 2504 | - // Ungrouped entries | |
| 2505 | - $grouped_prompts['_ungrouped_' . $prompt->id] = array($prompt); | |
| 2506 | - } | |
| 2507 | - } | |
| 2508 | - | |
| 2509 | - // Sort groups by the original URL order (newest first) | |
| 2510 | - uksort($grouped_prompts, function($a, $b) use ($url_order_map) { | |
| 2511 | - $order_a = isset($url_order_map[$a]) ? $url_order_map[$a] : PHP_INT_MAX; | |
| 2512 | - $order_b = isset($url_order_map[$b]) ? $url_order_map[$b] : PHP_INT_MAX; | |
| 2513 | - return $order_a - $order_b; | |
| 2514 | - }); | |
| 2515 | - | |
| 2516 | - // Sort each group internally by chunk_index | |
| 2517 | - foreach ($grouped_prompts as $source_url => &$group) { | |
| 2518 | - usort($group, function($a, $b) { | |
| 2519 | - $index_a = isset($a->chunk_metadata['chunk_index']) ? intval($a->chunk_metadata['chunk_index']) : 0; | |
| 2520 | - $index_b = isset($b->chunk_metadata['chunk_index']) ? intval($b->chunk_metadata['chunk_index']) : 0; | |
| 2521 | - return $index_a - $index_b; | |
| 2522 | - }); | |
| 2523 | - } | |
| 2524 | - unset($group); | |
| 2525 | - | |
| 2526 | - // Build HTML for the table rows | |
| 2527 | - ob_start(); | |
| 2528 | - $display_index = 0; | |
| 2529 | - $current_page = $page; | |
| 2530 | - $data_source = 'wordpress'; | |
| 2531 | - $current_bot_id = $bot_id; | |
| 2532 | - $preview_length = 150; | |
| 2533 | - | |
| 2534 | - if (empty($grouped_prompts)) { | |
| 2535 | - echo '<tr><td colspan="5" style="padding: 40px; text-align: center; color: var(--mxch-text-muted);">'; | |
| 2536 | - esc_html_e('No knowledge entries found. Use the Import Options to add content.', 'mxchat'); | |
| 2537 | - echo '</td></tr>'; | |
| 2538 | - } else { | |
| 2539 | - foreach ($grouped_prompts as $source_url => $group) { | |
| 2540 | - $chunk_count = count($group); | |
| 2541 | - $first_prompt = $group[0]; | |
| 2542 | - $display_index++; | |
| 2543 | - | |
| 2544 | - if ($chunk_count > 1) { | |
| 2545 | - // Multiple chunks - show grouped row with expand button | |
| 2546 | - $group_id = 'group-' . md5($source_url); | |
| 2547 | - ?> | |
| 2548 | - <tr id="prompt-<?php echo esc_attr($first_prompt->id); ?>" | |
| 2549 | - class="mxchat-chunk-group-header" | |
| 2550 | - data-source="<?php echo esc_attr($data_source); ?>" | |
| 2551 | - data-group-id="<?php echo esc_attr($group_id); ?>" | |
| 2552 | - style="border-bottom: 1px solid var(--mxch-card-border);"> | |
| 2553 | - <td style="padding: 12px 16px; text-align: center;"> | |
| 2554 | - <input type="checkbox" | |
| 2555 | - class="mxchat-entry-checkbox" | |
| 2556 | - data-entry-id="<?php echo esc_attr($first_prompt->id); ?>" | |
| 2557 | - data-source="<?php echo esc_attr($data_source); ?>" | |
| 2558 | - data-source-url="<?php echo esc_attr($source_url); ?>" | |
| 2559 | - data-is-group="true" | |
| 2560 | - data-chunk-count="<?php echo esc_attr($chunk_count); ?>"> | |
| 2561 | - </td> | |
| 2562 | - <td style="padding: 12px 16px; font-size: 13px;"> | |
| 2563 | - <?php echo esc_html($first_prompt->id); ?> | |
| 2564 | - </td> | |
| 2565 | - <td class="mxchat-content-cell" style="padding: 12px 16px; font-size: 13px;"> | |
| 2566 | - <div class="mxchat-chunk-group-info"> | |
| 2567 | - <button type="button" class="mxchat-chunk-toggle" data-group-id="<?php echo esc_attr($group_id); ?>"> | |
| 2568 | - <span class="dashicons dashicons-arrow-right-alt2"></span> | |
| 2569 | - </button> | |
| 2570 | - <span class="mxchat-chunk-badge"><?php echo esc_html($chunk_count); ?> <?php esc_html_e('chunks', 'mxchat'); ?></span> | |
| 2571 | - <span class="mxchat-chunk-preview"> | |
| 2572 | - <?php | |
| 2573 | - $parent_content = isset($first_prompt->display_content) ? $first_prompt->display_content : $first_prompt->article_content; | |
| 2574 | - $content_preview = mb_substr($parent_content, 0, 100); | |
| 2575 | - echo esc_html($content_preview . '...'); | |
| 2576 | - ?> | |
| 2577 | - </span> | |
| 2578 | - </div> | |
| 2579 | - </td> | |
| 2580 | - <td class="mxchat-url-cell" style="padding: 12px 16px; font-size: 13px;"> | |
| 2581 | - <?php if (!empty($source_url) && strpos($source_url, 'mxchat://') !== 0 && strpos($source_url, '_ungrouped_') !== 0) : ?> | |
| 2582 | - <a href="<?php echo esc_url($source_url); ?>" target="_blank" style="color: var(--mxch-primary); text-decoration: none;"> | |
| 2583 | - <span class="dashicons dashicons-external" style="font-size: 14px;"></span> | |
| 2584 | - <?php esc_html_e('View Source', 'mxchat'); ?> | |
| 2585 | - </a> | |
| 2586 | - <?php else : ?> | |
| 2587 | - <span style="color: var(--mxch-text-muted);"><?php esc_html_e('Manual Content', 'mxchat'); ?></span> | |
| 2588 | - <?php endif; ?> | |
| 2589 | - </td> | |
| 2590 | - <td class="mxchat-actions-cell" style="padding: 12px 16px; white-space: nowrap;"> | |
| 2591 | - <?php if ($data_source !== 'pinecone') : ?> | |
| 2592 | - <button type="button" | |
| 2593 | - class="mxch-btn mxch-btn-ghost mxch-btn-sm mxchat-edit-entry-btn" | |
| 2594 | - data-source-url="<?php echo esc_attr($source_url); ?>" | |
| 2595 | - data-entry-id="<?php echo esc_attr($first_prompt->id); ?>" | |
| 2596 | - data-data-source="<?php echo esc_attr($data_source); ?>" | |
| 2597 | - data-bot-id="<?php echo esc_attr($current_bot_id); ?>" | |
| 2598 | - data-nonce="<?php echo wp_create_nonce('mxchat_edit_entry_nonce'); ?>" | |
| 2599 | - title="<?php esc_attr_e('Edit content', 'mxchat'); ?>"> | |
| 2600 | - <span class="dashicons dashicons-edit" style="font-size: 14px;"></span> | |
| 2601 | - </button> | |
| 2602 | - <?php endif; ?> | |
| 2603 | - <button type="button" | |
| 2604 | - class="mxch-btn mxch-btn-ghost mxch-btn-sm delete-button-group" | |
| 2605 | - data-source-url="<?php echo esc_attr($source_url); ?>" | |
| 2606 | - data-chunk-count="<?php echo esc_attr($chunk_count); ?>" | |
| 2607 | - data-data-source="<?php echo esc_attr($data_source); ?>" | |
| 2608 | - data-bot-id="<?php echo esc_attr($current_bot_id); ?>" | |
| 2609 | - data-nonce="<?php echo wp_create_nonce('mxchat_delete_chunks_nonce'); ?>" | |
| 2610 | - style="color: var(--mxch-error);" | |
| 2611 | - title="<?php esc_attr_e('Delete all chunks', 'mxchat'); ?>"> | |
| 2612 | - <span class="dashicons dashicons-trash" style="font-size: 14px;"></span> | |
| 2613 | - </button> | |
| 2614 | - </td> | |
| 2615 | - </tr> | |
| 2616 | - <?php | |
| 2617 | - // Render hidden chunk rows | |
| 2618 | - foreach ($group as $chunk_index => $chunk) { | |
| 2619 | - $meta_chunk_index = isset($chunk->chunk_metadata['chunk_index']) ? intval($chunk->chunk_metadata['chunk_index']) : $chunk_index; | |
| 2620 | - $meta_total_chunks = isset($chunk->chunk_metadata['total_chunks']) ? intval($chunk->chunk_metadata['total_chunks']) : $chunk_count; | |
| 2621 | - $content = isset($chunk->display_content) ? $chunk->display_content : $chunk->article_content; | |
| 2622 | - $content_preview = mb_strlen($content) > $preview_length | |
| 2623 | - ? mb_substr($content, 0, $preview_length) . '...' | |
| 2624 | - : $content; | |
| 2625 | - ?> | |
| 2626 | - <tr id="prompt-<?php echo esc_attr($chunk->id); ?>" | |
| 2627 | - class="mxchat-chunk-row <?php echo esc_attr($group_id); ?>" | |
| 2628 | - data-source="<?php echo esc_attr($data_source); ?>" | |
| 2629 | - style="display: none; background: #f8f9fa; border-bottom: 1px solid var(--mxch-card-border);"> | |
| 2630 | - <td style="padding: 12px 16px; text-align: center;"> | |
| 2631 | - <!-- Checkbox column placeholder for chunks (managed by group) --> | |
| 2632 | - </td> | |
| 2633 | - <td style="padding: 12px 16px 12px 30px; font-size: 13px;"> | |
| 2634 | - <!-- Hidden ID column for chunks --> | |
| 2635 | - </td> | |
| 2636 | - <td class="mxchat-content-cell" style="padding: 12px 16px; font-size: 13px;"> | |
| 2637 | - <div class="mxchat-accordion-wrapper"> | |
| 2638 | - <div class="mxchat-content-preview"> | |
| 2639 | - <span class="mxchat-chunk-indicator" style="margin-right: 10px; color: var(--mxch-text-secondary); font-size: 12px;"> | |
| 2640 | - <?php printf(esc_html__('Chunk %d of %d', 'mxchat'), $meta_chunk_index + 1, $meta_total_chunks); ?> | |
| 2641 | - </span> | |
| 2642 | - <span class="preview-text"><?php echo esc_html($content_preview); ?></span> | |
| 2643 | - <?php if (mb_strlen($content) > $preview_length) : ?> | |
| 2644 | - <button class="mxchat-expand-toggle" type="button"> | |
| 2645 | - <span class="dashicons dashicons-arrow-down-alt2"></span> | |
| 2646 | - </button> | |
| 2647 | - <?php endif; ?> | |
| 2648 | - </div> | |
| 2649 | - <div class="mxchat-content-full" style="display: none;"> | |
| 2650 | - <div class="content-view"> | |
| 2651 | - <?php | |
| 2652 | - if (preg_match('/[\x{0590}-\x{05FF}]/u', $content)) { | |
| 2653 | - echo '<div dir="rtl" lang="he" class="rtl-content">'; | |
| 2654 | - echo wp_kses_post(wpautop($content)); | |
| 2655 | - echo '</div>'; | |
| 2656 | - } else { | |
| 2657 | - echo wp_kses_post(wpautop($content)); | |
| 2658 | - } | |
| 2659 | - ?> | |
| 2660 | - </div> | |
| 2661 | - </div> | |
| 2662 | - </div> | |
| 2663 | - </td> | |
| 2664 | - <td class="mxchat-url-cell" style="padding: 12px 16px; font-size: 13px;"> | |
| 2665 | - <span class="mxchat-chunk-label" style="color: var(--mxch-text-muted);"><?php esc_html_e('Same as parent', 'mxchat'); ?></span> | |
| 2666 | - </td> | |
| 2667 | - <td class="mxchat-actions-cell" style="padding: 12px 16px;"> | |
| 2668 | - <span class="mxchat-chunk-label" style="color: var(--mxch-text-muted);"><?php esc_html_e('Managed by group', 'mxchat'); ?></span> | |
| 2669 | - </td> | |
| 2670 | - </tr> | |
| 2671 | - <?php | |
| 2672 | - } | |
| 2673 | - } else { | |
| 2674 | - // Single entry - display normally with accordion | |
| 2675 | - $prompt = $first_prompt; | |
| 2676 | - $content = isset($prompt->display_content) ? $prompt->display_content : $prompt->article_content; | |
| 2677 | - $content_preview = mb_strlen($content) > $preview_length | |
| 2678 | - ? mb_substr($content, 0, $preview_length) . '...' | |
| 2679 | - : $content; | |
| 2680 | - ?> | |
| 2681 | - <tr id="prompt-<?php echo esc_attr($prompt->id); ?>" | |
| 2682 | - data-source="<?php echo esc_attr($data_source); ?>" | |
| 2683 | - style="border-bottom: 1px solid var(--mxch-card-border);"> | |
| 2684 | - <td style="padding: 12px 16px; text-align: center;"> | |
| 2685 | - <input type="checkbox" | |
| 2686 | - class="mxchat-entry-checkbox" | |
| 2687 | - data-entry-id="<?php echo esc_attr($prompt->id); ?>" | |
| 2688 | - data-source="<?php echo esc_attr($data_source); ?>" | |
| 2689 | - data-source-url="<?php echo esc_attr($source_url); ?>" | |
| 2690 | - data-is-group="false"> | |
| 2691 | - </td> | |
| 2692 | - <td style="padding: 12px 16px; font-size: 13px;"> | |
| 2693 | - <?php echo esc_html($prompt->id); ?> | |
| 2694 | - </td> | |
| 2695 | - <td class="mxchat-content-cell" style="padding: 12px 16px; font-size: 13px;"> | |
| 2696 | - <div class="mxchat-accordion-wrapper"> | |
| 2697 | - <div class="mxchat-content-preview"> | |
| 2698 | - <span class="preview-text"><?php echo esc_html($content_preview); ?></span> | |
| 2699 | - <?php if (mb_strlen($content) > $preview_length) : ?> | |
| 2700 | - <button class="mxchat-expand-toggle" type="button"> | |
| 2701 | - <span class="dashicons dashicons-arrow-down-alt2"></span> | |
| 2702 | - </button> | |
| 2703 | - <?php endif; ?> | |
| 2704 | - </div> | |
| 2705 | - <div class="mxchat-content-full" style="display: none;"> | |
| 2706 | - <div class="content-view"> | |
| 2707 | - <?php | |
| 2708 | - if (preg_match('/[\x{0590}-\x{05FF}]/u', $content)) { | |
| 2709 | - echo '<div dir="rtl" lang="he" class="rtl-content">'; | |
| 2710 | - echo wp_kses_post(wpautop($content)); | |
| 2711 | - echo '</div>'; | |
| 2712 | - } else { | |
| 2713 | - echo wp_kses_post(wpautop($content)); | |
| 2714 | - } | |
| 2715 | - ?> | |
| 2716 | - </div> | |
| 2717 | - </div> | |
| 2718 | - </div> | |
| 2719 | - </td> | |
| 2720 | - <td class="mxchat-url-cell" style="padding: 12px 16px; font-size: 13px;"> | |
| 2721 | - <?php | |
| 2722 | - $actual_source = $source_url; | |
| 2723 | - if (strpos($source_url, '_ungrouped_') === 0) { | |
| 2724 | - $actual_source = $prompt->source_url ?? ''; | |
| 2725 | - } | |
| 2726 | - if (!empty($actual_source) && strpos($actual_source, 'mxchat://') !== 0) : ?> | |
| 2727 | - <a href="<?php echo esc_url($actual_source); ?>" target="_blank" style="color: var(--mxch-primary); text-decoration: none;"> | |
| 2728 | - <span class="dashicons dashicons-external" style="font-size: 14px;"></span> | |
| 2729 | - <?php esc_html_e('View', 'mxchat'); ?> | |
| 2730 | - </a> | |
| 2731 | - <?php else : ?> | |
| 2732 | - <span style="color: var(--mxch-text-muted);"><?php esc_html_e('Manual', 'mxchat'); ?></span> | |
| 2733 | - <?php endif; ?> | |
| 2734 | - </td> | |
| 2735 | - <td style="padding: 12px 16px; white-space: nowrap;"> | |
| 2736 | - <button type="button" | |
| 2737 | - class="mxch-btn mxch-btn-ghost mxch-btn-sm mxchat-edit-entry-btn" | |
| 2738 | - data-source-url="<?php echo esc_attr($prompt->source_url ?? ''); ?>" | |
| 2739 | - data-entry-id="<?php echo esc_attr($prompt->id); ?>" | |
| 2740 | - data-data-source="<?php echo esc_attr($data_source); ?>" | |
| 2741 | - data-bot-id="<?php echo esc_attr($current_bot_id); ?>" | |
| 2742 | - data-nonce="<?php echo wp_create_nonce('mxchat_edit_entry_nonce'); ?>" | |
| 2743 | - title="<?php esc_attr_e('Edit content', 'mxchat'); ?>"> | |
| 2744 | - <span class="dashicons dashicons-edit" style="font-size: 14px;"></span> | |
| 2745 | - </button> | |
| 2746 | - <button type="button" class="mxch-btn mxch-btn-ghost mxch-btn-sm delete-button-wordpress" data-entry-id="<?php echo esc_attr($prompt->id); ?>" data-bot-id="<?php echo esc_attr($current_bot_id); ?>" data-nonce="<?php echo wp_create_nonce('mxchat_delete_wordpress_prompt_nonce'); ?>" style="color: var(--mxch-error);"> | |
| 2747 | - <span class="dashicons dashicons-trash" style="font-size: 14px;"></span> | |
| 2748 | - </button> | |
| 2749 | - </td> | |
| 2750 | - </tr> | |
| 2751 | - <?php | |
| 2752 | - } | |
| 2753 | - } | |
| 2754 | - } | |
| 2755 | - $html = ob_get_clean(); | |
| 2756 | - | |
| 2757 | - // Generate pagination HTML (include search/filter data for subsequent pages) | |
| 2758 | - $pagination_html = ''; | |
| 2759 | - if ($total_pages > 1) { | |
| 2760 | - $pagination_html = '<div class="mxchat-ajax-pagination" data-current-page="' . esc_attr($page) . '" data-total-pages="' . esc_attr($total_pages) . '" data-search="' . esc_attr($search_query) . '" data-content-type="' . esc_attr($content_type_filter) . '">'; | |
| 2761 | - | |
| 2762 | - // Previous button | |
| 2763 | - if ($page > 1) { | |
| 2764 | - $pagination_html .= '<a href="#" class="mxchat-page-link" data-page="' . ($page - 1) . '">' . esc_html__('« Previous', 'mxchat') . '</a> '; | |
| 2765 | - } | |
| 2766 | - | |
| 2767 | - // Page numbers | |
| 2768 | - $start_page = max(1, $page - 2); | |
| 2769 | - $end_page = min($total_pages, $page + 2); | |
| 2770 | - | |
| 2771 | - if ($start_page > 1) { | |
| 2772 | - $pagination_html .= '<a href="#" class="mxchat-page-link" data-page="1">1</a> '; | |
| 2773 | - if ($start_page > 2) { | |
| 2774 | - $pagination_html .= '<span class="mxchat-page-dots">...</span> '; | |
| 2775 | - } | |
| 2776 | - } | |
| 2777 | - | |
| 2778 | - for ($i = $start_page; $i <= $end_page; $i++) { | |
| 2779 | - if ($i == $page) { | |
| 2780 | - $pagination_html .= '<span class="mxchat-page-current">' . $i . '</span> '; | |
| 2781 | - } else { | |
| 2782 | - $pagination_html .= '<a href="#" class="mxchat-page-link" data-page="' . $i . '">' . $i . '</a> '; | |
| 2783 | - } | |
| 2784 | - } | |
| 2785 | - | |
| 2786 | - if ($end_page < $total_pages) { | |
| 2787 | - if ($end_page < $total_pages - 1) { | |
| 2788 | - $pagination_html .= '<span class="mxchat-page-dots">...</span> '; | |
| 2789 | - } | |
| 2790 | - $pagination_html .= '<a href="#" class="mxchat-page-link" data-page="' . $total_pages . '">' . $total_pages . '</a> '; | |
| 2791 | - } | |
| 2792 | - | |
| 2793 | - // Next button | |
| 2794 | - if ($page < $total_pages) { | |
| 2795 | - $pagination_html .= '<a href="#" class="mxchat-page-link" data-page="' . ($page + 1) . '">' . esc_html__('Next »', 'mxchat') . '</a>'; | |
| 2796 | - } | |
| 2797 | - | |
| 2798 | - $pagination_html .= '</div>'; | |
| 2799 | - } | |
| 2800 | - | |
| 2801 | - wp_send_json_success(array( | |
| 2802 | - 'html' => $html, | |
| 2803 | - 'pagination_html' => $pagination_html, | |
| 2804 | - 'total_count' => $total_records, | |
| 2805 | - 'total_pages' => $total_pages, | |
| 2806 | - 'page' => $page, | |
| 2807 | - 'per_page' => $per_page, | |
| 2808 | - 'data_source' => 'wordpress' | |
| 2809 | - )); | |
| 2810 | -} | |
| 2811 | - | |
| 2812 | -/** | |
| 2813 | - * AJAX handler to detect available sitemaps on the site | |
| 2814 | - * Optimized for speed - only checks primary sitemap indexes first | |
| 2815 | - */ | |
| 2816 | -public function ajax_mxchat_detect_sitemaps() { | |
| 2817 | - check_ajax_referer('mxchat_detect_sitemaps_nonce', 'nonce'); | |
| 2818 | - | |
| 2819 | - if (!current_user_can('manage_options')) { | |
| 2820 | - wp_send_json_error(array('message' => 'Unauthorized')); | |
| 2821 | - return; | |
| 2822 | - } | |
| 2823 | - | |
| 2824 | - $site_url = get_site_url(); | |
| 2825 | - $sitemaps = array(); | |
| 2826 | - $found_index = false; | |
| 2827 | - | |
| 2828 | - // Only check the main sitemap index files first (much faster) | |
| 2829 | - // These are the primary entry points that contain sub-sitemaps | |
| 2830 | - $primary_indexes = array( | |
| 2831 | - 'sitemap_index.xml' => 'Yoast SEO / Rank Math', // Yoast & Rank Math | |
| 2832 | - 'wp-sitemap.xml' => 'WordPress Core', // WordPress Core | |
| 2833 | - 'sitemap.xml' => 'Standard', // Generic/AIOSEO | |
| 2834 | - ); | |
| 2835 | - | |
| 2836 | - foreach ($primary_indexes as $path => $source) { | |
| 2837 | - $url = trailingslashit($site_url) . $path; | |
| 2838 | - | |
| 2839 | - $response = wp_remote_head($url, array( | |
| 2840 | - 'timeout' => 10, | |
| 2841 | - 'sslverify' => false, | |
| 2842 | - 'redirection' => 1, | |
| 2843 | - 'user-agent' => 'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/131.0.0.0 Safari/537.36', | |
| 2844 | - )); | |
| 2845 | - | |
| 2846 | - if (!is_wp_error($response) && wp_remote_retrieve_response_code($response) === 200) { | |
| 2847 | - // Found a sitemap index - parse it to get sub-sitemaps | |
| 2848 | - $sub_sitemaps = $this->parse_sitemap_index($url); | |
| 2849 | - if (!empty($sub_sitemaps)) { | |
| 2850 | - $sitemaps[] = array( | |
| 2851 | - 'url' => $url, | |
| 2852 | - 'type' => 'index', | |
| 2853 | - 'source' => $source, | |
| 2854 | - 'sub_sitemaps' => $sub_sitemaps | |
| 2855 | - ); | |
| 2856 | - $found_index = true; | |
| 2857 | - // Found a valid index, no need to check others | |
| 2858 | - break; | |
| 2859 | - } | |
| 2860 | - } | |
| 2861 | - } | |
| 2862 | - | |
| 2863 | - // If no sitemap index found, check for standalone sitemaps | |
| 2864 | - if (!$found_index) { | |
| 2865 | - $standalone_sitemaps = array( | |
| 2866 | - 'post-sitemap.xml' => array('type' => 'content', 'source' => 'Yoast SEO'), | |
| 2867 | - 'page-sitemap.xml' => array('type' => 'content', 'source' => 'Yoast SEO'), | |
| 2868 | - ); | |
| 2869 | - | |
| 2870 | - foreach ($standalone_sitemaps as $path => $info) { | |
| 2871 | - $url = trailingslashit($site_url) . $path; | |
| 2872 | - | |
| 2873 | - $response = wp_remote_head($url, array( | |
| 2874 | - 'timeout' => 2, | |
| 2875 | - 'sslverify' => false | |
| 2876 | - )); | |
| 2877 | - | |
| 2878 | - if (!is_wp_error($response) && wp_remote_retrieve_response_code($response) === 200) { | |
| 2879 | - $sitemaps[] = array( | |
| 2880 | - 'url' => $url, | |
| 2881 | - 'type' => $info['type'], | |
| 2882 | - 'source' => $info['source'], | |
| 2883 | - 'url_count' => 0 // Skip URL count for speed | |
| 2884 | - ); | |
| 2885 | - } | |
| 2886 | - } | |
| 2887 | - } | |
| 2888 | - | |
| 2889 | - wp_send_json_success(array( | |
| 2890 | - 'sitemaps' => $sitemaps, | |
| 2891 | - 'site_url' => $site_url | |
| 2892 | - )); | |
| 2893 | -} | |
| 2894 | - | |
| 2895 | -/** | |
| 2896 | - * Parse a sitemap index to get sub-sitemaps | |
| 2897 | - * Optimized: doesn't fetch URL count for each sub-sitemap (too slow) | |
| 2898 | - */ | |
| 2899 | -private function parse_sitemap_index($url) { | |
| 2900 | - $sub_sitemaps = array(); | |
| 2901 | - | |
| 2902 | - $response = wp_remote_get($url, array( | |
| 2903 | - 'timeout' => 30, | |
| 2904 | - 'sslverify' => false, | |
| 2905 | - 'user-agent' => 'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/131.0.0.0 Safari/537.36', | |
| 2906 | - 'headers' => array( | |
| 2907 | - 'Accept' => 'text/html,application/xhtml+xml,application/xml;q=0.9,*/*;q=0.8', | |
| 2908 | - 'Accept-Language' => 'en-US,en;q=0.9', | |
| 2909 | - ), | |
| 2910 | - )); | |
| 2911 | - | |
| 2912 | - if (is_wp_error($response)) { | |
| 2913 | - return $sub_sitemaps; | |
| 2914 | - } | |
| 2915 | - | |
| 2916 | - $body = wp_remote_retrieve_body($response); | |
| 2917 | - if (empty($body)) { | |
| 2918 | - return $sub_sitemaps; | |
| 2919 | - } | |
| 2920 | - | |
| 2921 | - // Suppress XML errors | |
| 2922 | - libxml_use_internal_errors(true); | |
| 2923 | - $xml = simplexml_load_string($body); | |
| 2924 | - libxml_clear_errors(); | |
| 2925 | - | |
| 2926 | - if ($xml === false) { | |
| 2927 | - return $sub_sitemaps; | |
| 2928 | - } | |
| 2929 | - | |
| 2930 | - // Check if it's a sitemap index (contains <sitemap> elements) | |
| 2931 | - if (isset($xml->sitemap)) { | |
| 2932 | - foreach ($xml->sitemap as $sitemap) { | |
| 2933 | - $loc = (string) $sitemap->loc; | |
| 2934 | - if (!empty($loc)) { | |
| 2935 | - // Try to determine the type from the URL | |
| 2936 | - $type = 'content'; | |
| 2937 | - if (strpos($loc, 'category') !== false || strpos($loc, 'tag') !== false || strpos($loc, 'taxonomy') !== false) { | |
| 2938 | - $type = 'taxonomy'; | |
| 2939 | - } elseif (strpos($loc, 'author') !== false || strpos($loc, 'user') !== false) { | |
| 2940 | - $type = 'author'; | |
| 2941 | - } | |
| 2942 | - | |
| 2943 | - // Skip URL count - too slow to fetch for each sitemap | |
| 2944 | - $sub_sitemaps[] = array( | |
| 2945 | - 'url' => $loc, | |
| 2946 | - 'type' => $type, | |
| 2947 | - 'url_count' => 0, // Don't fetch - takes too long | |
| 2948 | - 'name' => basename(parse_url($loc, PHP_URL_PATH)) | |
| 2949 | - ); | |
| 2950 | - } | |
| 2951 | - } | |
| 2952 | - } | |
| 2953 | - | |
| 2954 | - return $sub_sitemaps; | |
| 2955 | -} | |
| 2956 | - | |
| 2957 | -/** | |
| 2958 | - * Get URL count from a sitemap | |
| 2959 | - */ | |
| 2960 | -private function get_sitemap_url_count($url) { | |
| 2961 | - $response = wp_remote_get($url, array( | |
| 2962 | - 'timeout' => 30, | |
| 2963 | - 'sslverify' => false, | |
| 2964 | - 'user-agent' => 'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/131.0.0.0 Safari/537.36', | |
| 2965 | - 'headers' => array( | |
| 2966 | - 'Accept' => 'text/html,application/xhtml+xml,application/xml;q=0.9,*/*;q=0.8', | |
| 2967 | - 'Accept-Language' => 'en-US,en;q=0.9', | |
| 2968 | - ), | |
| 2969 | - )); | |
| 2970 | - | |
| 2971 | - if (is_wp_error($response)) { | |
| 2972 | - return 0; | |
| 2973 | - } | |
| 2974 | - | |
| 2975 | - $body = wp_remote_retrieve_body($response); | |
| 2976 | - if (empty($body)) { | |
| 2977 | - return 0; | |
| 2978 | - } | |
| 2979 | - | |
| 2980 | - // Count <url> or <loc> elements | |
| 2981 | - $count = preg_match_all('/<url>/i', $body, $matches); | |
| 2982 | - return $count ?: 0; | |
| 2983 | -} | |
| 2984 | - | |
| 2985 | -/** | |
| 2986 | - * Get sitemaps declared in robots.txt | |
| 2987 | - */ | |
| 2988 | -private function get_sitemaps_from_robots($site_url) { | |
| 2989 | - $sitemaps = array(); | |
| 2990 | - $robots_url = trailingslashit($site_url) . 'robots.txt'; | |
| 2991 | - | |
| 2992 | - $response = wp_remote_get($robots_url, array( | |
| 2993 | - 'timeout' => 15, | |
| 2994 | - 'sslverify' => false, | |
| 2995 | - 'user-agent' => 'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/131.0.0.0 Safari/537.36', | |
| 2996 | - )); | |
| 2997 | - | |
| 2998 | - if (is_wp_error($response)) { | |
| 2999 | - return $sitemaps; | |
| 3000 | - } | |
| 3001 | - | |
| 3002 | - $body = wp_remote_retrieve_body($response); | |
| 3003 | - if (empty($body)) { | |
| 3004 | - return $sitemaps; | |
| 3005 | - } | |
| 3006 | - | |
| 3007 | - // Find Sitemap: declarations | |
| 3008 | - if (preg_match_all('/^Sitemap:\s*(.+)$/mi', $body, $matches)) { | |
| 3009 | - foreach ($matches[1] as $sitemap_url) { | |
| 3010 | - $sitemap_url = trim($sitemap_url); | |
| 3011 | - if (filter_var($sitemap_url, FILTER_VALIDATE_URL)) { | |
| 3012 | - $sitemaps[] = $sitemap_url; | |
| 3013 | - } | |
| 3014 | - } | |
| 3015 | - } | |
| 3016 | - | |
| 3017 | - return $sitemaps; | |
| 3018 | -} | |
| 3019 | - | |
| 3020 | -public function mxchat_stop_processing() { | |
| 3021 | - // Verify permissions | |
| 3022 | - if (!current_user_can('manage_options')) { | |
| 3023 | - wp_die(esc_html__('Unauthorized access', 'mxchat')); | |
| 3024 | - } | |
| 3025 | - | |
| 3026 | - // Verify nonce | |
| 3027 | - check_admin_referer('mxchat_stop_processing_action', 'mxchat_stop_processing_nonce'); | |
| 3028 | - | |
| 3029 | - global $wpdb; | |
| 3030 | - $table_name = $wpdb->prefix . 'mxchat_processing_queue'; | |
| 3031 | - | |
| 3032 | - // Get active queue IDs | |
| 3033 | - $sitemap_queue_id = get_transient('mxchat_active_queue_sitemap'); | |
| 3034 | - $pdf_queue_id = get_transient('mxchat_active_queue_pdf'); | |
| 3035 | - | |
| 3036 | - // Delete all pending items from active queues | |
| 3037 | - if ($sitemap_queue_id) { | |
| 3038 | - $wpdb->delete( | |
| 3039 | - $table_name, | |
| 3040 | - array( | |
| 3041 | - 'queue_id' => $sitemap_queue_id, | |
| 3042 | - 'status' => 'pending' | |
| 3043 | - ), | |
| 3044 | - array('%s', '%s') | |
| 3045 | - ); | |
| 3046 | - | |
| 3047 | - delete_transient('mxchat_active_queue_sitemap'); | |
| 3048 | - delete_transient('mxchat_last_sitemap_url'); | |
| 3049 | - } | |
| 3050 | - | |
| 3051 | - if ($pdf_queue_id) { | |
| 3052 | - // Get PDF path before deleting | |
| 3053 | - $pdf_path = $this->mxchat_get_queue_meta($pdf_queue_id, 'pdf_path'); | |
| 3054 | - | |
| 3055 | - $wpdb->delete( | |
| 3056 | - $table_name, | |
| 3057 | - array( | |
| 3058 | - 'queue_id' => $pdf_queue_id, | |
| 3059 | - 'status' => 'pending' | |
| 3060 | - ), | |
| 3061 | - array('%s', '%s') | |
| 3062 | - ); | |
| 3063 | - | |
| 3064 | - // Delete PDF file | |
| 3065 | - if ($pdf_path && file_exists($pdf_path)) { | |
| 3066 | - wp_delete_file($pdf_path); | |
| 3067 | - } | |
| 3068 | - | |
| 3069 | - delete_transient('mxchat_active_queue_pdf'); | |
| 3070 | - delete_transient('mxchat_last_pdf_url'); | |
| 3071 | - } | |
| 3072 | - | |
| 3073 | - // Redirect back with a success message | |
| 3074 | - set_transient('mxchat_admin_notice_success', | |
| 3075 | - esc_html__('Processing has been stopped successfully.', 'mxchat'), | |
| 3076 | - 30 | |
| 3077 | - ); | |
| 3078 | - wp_safe_redirect(admin_url('admin.php?page=mxchat-prompts')); | |
| 3079 | - exit; | |
| 3080 | -} | |
| 3081 | - | |
| 3082 | -/** | |
| 3083 | - * Get content list for processing | |
| 3084 | - */ | |
| 3085 | -public function ajax_mxchat_get_content_list() { | |
| 3086 | - // Verify the nonce | |
| 3087 | - check_ajax_referer('mxchat_content_selector_nonce', 'nonce'); | |
| 3088 | - | |
| 3089 | - if (!current_user_can('manage_options')) { | |
| 3090 | - wp_send_json_error(__('Unauthorized access', 'mxchat')); | |
| 3091 | - } | |
| 3092 | - | |
| 3093 | - $page = isset($_GET['page']) ? absint($_GET['page']) : 1; | |
| 3094 | - $per_page = isset($_GET['per_page']) ? absint($_GET['per_page']) : 100; | |
| 3095 | - $search = isset($_GET['search']) ? sanitize_text_field($_GET['search']) : ''; | |
| 3096 | - $post_type = isset($_GET['post_type']) ? sanitize_text_field($_GET['post_type']) : 'all'; | |
| 3097 | - $post_status = isset($_GET['post_status']) ? sanitize_text_field($_GET['post_status']) : 'publish'; | |
| 3098 | - $processed_filter = isset($_GET['processed_filter']) ? sanitize_text_field($_GET['processed_filter']) : 'all'; | |
| 3099 | - | |
| 3100 | - // Build query args | |
| 3101 | - $args = array( | |
| 3102 | - 'posts_per_page' => $per_page, | |
| 3103 | - 'paged' => $page, | |
| 3104 | - 'post_status' => $post_status !== 'all' ? $post_status : array('publish', 'draft', 'pending'), | |
| 3105 | - 'orderby' => 'date', | |
| 3106 | - 'order' => 'DESC', | |
| 3107 | - ); | |
| 3108 | - | |
| 3109 | - // Handle post types - IMPROVED VERSION | |
| 3110 | - if ($post_type !== 'all') { | |
| 3111 | - $args['post_type'] = $post_type; | |
| 3112 | - } else { | |
| 3113 | - // Get all available post types that might contain content | |
| 3114 | - $all_post_types = array(); | |
| 3115 | - | |
| 3116 | - // First get all public post types | |
| 3117 | - $public_types = get_post_types(array('public' => true), 'names'); | |
| 3118 | - $all_post_types = array_merge($all_post_types, $public_types); | |
| 3119 | - | |
| 3120 | - // Add common forum/community post types | |
| 3121 | - $forum_types = array('topic', 'reply', 'forum', 'wpforo_topic', 'wpforo_post'); | |
| 3122 | - foreach ($forum_types as $forum_type) { | |
| 3123 | - if (post_type_exists($forum_type)) { | |
| 3124 | - $all_post_types[] = $forum_type; | |
| 3125 | - } | |
| 3126 | - } | |
| 3127 | - | |
| 3128 | - // Add other commonly used post types | |
| 3129 | - $common_types = array('product', 'job_listing', 'event', 'portfolio'); | |
| 3130 | - foreach ($common_types as $common_type) { | |
| 3131 | - if (post_type_exists($common_type)) { | |
| 3132 | - $all_post_types[] = $common_type; | |
| 3133 | - } | |
| 3134 | - } | |
| 3135 | - | |
| 3136 | - // Remove duplicates and ensure we have at least some post types | |
| 3137 | - $all_post_types = array_unique($all_post_types); | |
| 3138 | - | |
| 3139 | - if (empty($all_post_types)) { | |
| 3140 | - // Fallback to basic post types | |
| 3141 | - $all_post_types = array('post', 'page'); | |
| 3142 | - } | |
| 3143 | - | |
| 3144 | - $args['post_type'] = $all_post_types; | |
| 3145 | - | |
| 3146 | - // Debug logging to see what post types are being queried | |
| 3147 | - //error_log('MxChat Debug: Querying post types: ' . implode(', ', $all_post_types)); | |
| 3148 | - } | |
| 3149 | - | |
| 3150 | - if (!empty($search)) { | |
| 3151 | - $args['s'] = $search; | |
| 3152 | - } | |
| 3153 | - | |
| 3154 | - // Get processed data from storage | |
| 3155 | - $processed_data = array(); | |
| 3156 | - | |
| 3157 | - $pinecone_options = get_option('mxchat_pinecone_addon_options', array()); | |
| 3158 | - $use_pinecone = ($pinecone_options['mxchat_use_pinecone'] ?? '0') === '1'; | |
| 3159 | - | |
| 3160 | - if ($use_pinecone && !empty($pinecone_options['mxchat_pinecone_api_key'])) { | |
| 3161 | - // Get fresh data from Pinecone - no caching | |
| 3162 | - $processed_data = $this->mxchat_get_pinecone_processed_content($pinecone_options); | |
| 3163 | - } else { | |
| 3164 | - // WordPress DB checking with better URL matching for all post types | |
| 3165 | - global $wpdb; | |
| 3166 | - $table_name = $wpdb->prefix . 'mxchat_system_prompt_content'; | |
| 3167 | - $processed_items = $wpdb->get_results("SELECT id, source_url, article_content, timestamp FROM {$table_name}"); | |
| 3168 | - | |
| 3169 | - // Group items by source_url to count chunks | |
| 3170 | - $url_chunk_counts = array(); | |
| 3171 | - $url_latest_timestamp = array(); | |
| 3172 | - $url_first_id = array(); | |
| 3173 | - | |
| 3174 | - if (!empty($processed_items)) { | |
| 3175 | - foreach ($processed_items as $item) { | |
| 3176 | - $url = $item->source_url; | |
| 3177 | - if (empty($url)) continue; | |
| 3178 | - | |
| 3179 | - // Count chunks per URL | |
| 3180 | - if (!isset($url_chunk_counts[$url])) { | |
| 3181 | - $url_chunk_counts[$url] = 0; | |
| 3182 | - $url_latest_timestamp[$url] = $item->timestamp; | |
| 3183 | - $url_first_id[$url] = $item->id; | |
| 3184 | - } | |
| 3185 | - $url_chunk_counts[$url]++; | |
| 3186 | - | |
| 3187 | - // Track latest timestamp | |
| 3188 | - if (strtotime($item->timestamp) > strtotime($url_latest_timestamp[$url])) { | |
| 3189 | - $url_latest_timestamp[$url] = $item->timestamp; | |
| 3190 | - } | |
| 3191 | - } | |
| 3192 | - | |
| 3193 | - // Now build processed_data with chunk counts | |
| 3194 | - foreach ($url_chunk_counts as $url => $chunk_count) { | |
| 3195 | - $post_id = $this->mxchat_url_to_post_id_improved($url); | |
| 3196 | - | |
| 3197 | - if ($post_id) { | |
| 3198 | - $processed_data[$post_id] = array( | |
| 3199 | - 'db_id' => $url_first_id[$url], | |
| 3200 | - 'timestamp' => $url_latest_timestamp[$url], | |
| 3201 | - 'url' => $url, | |
| 3202 | - 'source' => 'wordpress', | |
| 3203 | - 'chunk_count' => $chunk_count | |
| 3204 | - ); | |
| 3205 | - } | |
| 3206 | - } | |
| 3207 | - } | |
| 3208 | - } | |
| 3209 | - | |
| 3210 | - // Get processed IDs as a simple array for in_array checks | |
| 3211 | - $processed_ids = array_keys($processed_data); | |
| 3212 | - | |
| 3213 | - // Handle processed/unprocessed filter | |
| 3214 | - if ($processed_filter === 'processed' && !empty($processed_ids)) { | |
| 3215 | - $args['post__in'] = $processed_ids; | |
| 3216 | - } elseif ($processed_filter === 'unprocessed' && !empty($processed_ids)) { | |
| 3217 | - $args['post__not_in'] = $processed_ids; | |
| 3218 | - } | |
| 3219 | - | |
| 3220 | - // Run the query | |
| 3221 | - $query = new WP_Query($args); | |
| 3222 | - $content_items = array(); | |
| 3223 | - | |
| 3224 | - if ($query->have_posts()) { | |
| 3225 | - while ($query->have_posts()) { | |
| 3226 | - $query->the_post(); | |
| 3227 | - $id = get_the_ID(); | |
| 3228 | - $post_date = get_the_date(); | |
| 3229 | - $excerpt = wp_trim_words(get_the_excerpt(), 20, '...'); | |
| 3230 | - $word_count = str_word_count(strip_tags(get_the_content())); | |
| 3231 | - | |
| 3232 | - $is_processed = in_array($id, $processed_ids); | |
| 3233 | - $processed_date = ''; | |
| 3234 | - $db_record_id = 0; | |
| 3235 | - $data_source = 'none'; | |
| 3236 | - | |
| 3237 | - if ($is_processed && isset($processed_data[$id])) { | |
| 3238 | - $item_data = $processed_data[$id]; | |
| 3239 | - $data_source = $item_data['source']; | |
| 3240 | - | |
| 3241 | - if ($data_source === 'wordpress' && isset($item_data['timestamp'])) { | |
| 3242 | - // WordPress DB format | |
| 3243 | - $timestamp = strtotime($item_data['timestamp']); | |
| 3244 | - $processed_date = human_time_diff($timestamp, current_time('timestamp')) . ' ago'; | |
| 3245 | - $db_record_id = $item_data['db_id']; | |
| 3246 | - } elseif ($data_source === 'pinecone') { | |
| 3247 | - // Pinecone format | |
| 3248 | - $processed_date = $item_data['processed_date']; | |
| 3249 | - $db_record_id = $item_data['db_id']; | |
| 3250 | - } | |
| 3251 | - } | |
| 3252 | - | |
| 3253 | - // Get chunk count for this item | |
| 3254 | - $chunk_count = 0; | |
| 3255 | - if ($is_processed && isset($processed_data[$id]['chunk_count'])) { | |
| 3256 | - $chunk_count = intval($processed_data[$id]['chunk_count']); | |
| 3257 | - } | |
| 3258 | - | |
| 3259 | - $content_items[] = array( | |
| 3260 | - 'id' => $id, | |
| 3261 | - 'title' => get_the_title(), | |
| 3262 | - 'permalink' => get_permalink(), | |
| 3263 | - 'date' => $post_date, | |
| 3264 | - 'type' => get_post_type(), | |
| 3265 | - 'status' => get_post_status(), | |
| 3266 | - 'excerpt' => $excerpt, | |
| 3267 | - 'word_count' => $word_count, | |
| 3268 | - 'already_processed' => $is_processed, | |
| 3269 | - 'processed_date' => $processed_date, | |
| 3270 | - 'db_record_id' => $db_record_id, | |
| 3271 | - 'data_source' => $data_source, | |
| 3272 | - 'chunk_count' => $chunk_count | |
| 3273 | - ); | |
| 3274 | - } | |
| 3275 | - wp_reset_postdata(); | |
| 3276 | - } | |
| 3277 | - | |
| 3278 | - $response = array( | |
| 3279 | - 'items' => $content_items, | |
| 3280 | - 'total' => $query->found_posts, | |
| 3281 | - 'total_pages' => $query->max_num_pages, | |
| 3282 | - 'current_page' => $page, | |
| 3283 | - 'processed_count' => count($processed_ids) | |
| 3284 | - ); | |
| 3285 | - | |
| 3286 | - wp_send_json_success($response); | |
| 3287 | - exit; | |
| 3288 | -} | |
| 3289 | - | |
| 3290 | - | |
| 3291 | -/** | |
| 3292 | - * This function handles various WooCommerce URL formats and permalink structures | |
| 3293 | - */ | |
| 3294 | -private function mxchat_url_to_post_id_improved($url) { | |
| 3295 | - // First try the standard WordPress function | |
| 3296 | - $post_id = url_to_postid($url); | |
| 3297 | - | |
| 3298 | - if ($post_id > 0) { | |
| 3299 | - return $post_id; | |
| 3300 | - } | |
| 3301 | - | |
| 3302 | - // If that fails, try more aggressive URL matching | |
| 3303 | - // Remove trailing slashes and query parameters for better matching | |
| 3304 | - $clean_url = rtrim($url, '/'); | |
| 3305 | - $clean_url = strtok($clean_url, '?'); // Remove query parameters | |
| 3306 | - | |
| 3307 | - // Try again with cleaned URL | |
| 3308 | - $post_id = url_to_postid($clean_url); | |
| 3309 | - if ($post_id > 0) { | |
| 3310 | - return $post_id; | |
| 3311 | - } | |
| 3312 | - | |
| 3313 | - // For bbPress forum topics, try extracting slug from URL | |
| 3314 | - if (strpos($url, '/topic/') !== false || strpos($url, '/forums/') !== false) { | |
| 3315 | - // Handle bbPress URLs: /forums/topic/topic-name/ | |
| 3316 | - if (preg_match('/\/forums\/topic\/([^\/\?]+)/', $url, $matches)) { | |
| 3317 | - $topic_slug = $matches[1]; | |
| 3318 | - | |
| 3319 | - // Look up topic by slug | |
| 3320 | - $topic = get_page_by_path($topic_slug, OBJECT, 'topic'); | |
| 3321 | - if ($topic) { | |
| 3322 | - return $topic->ID; | |
| 3323 | - } | |
| 3324 | - | |
| 3325 | - // Alternative method: query by post_name | |
| 3326 | - global $wpdb; | |
| 3327 | - $post_id = $wpdb->get_var($wpdb->prepare( | |
| 3328 | - "SELECT ID FROM {$wpdb->posts} WHERE post_name = %s AND post_type = 'topic' AND post_status IN ('publish', 'closed')", | |
| 3329 | - $topic_slug | |
| 3330 | - )); | |
| 3331 | - | |
| 3332 | - if ($post_id) { | |
| 3333 | - return intval($post_id); | |
| 3334 | - } | |
| 3335 | - } | |
| 3336 | - | |
| 3337 | - // Handle simpler topic URLs: /topic/topic-name/ | |
| 3338 | - if (preg_match('/\/topic\/([^\/\?]+)/', $url, $matches)) { | |
| 3339 | - $topic_slug = $matches[1]; | |
| 3340 | - | |
| 3341 | - global $wpdb; | |
| 3342 | - $post_id = $wpdb->get_var($wpdb->prepare( | |
| 3343 | - "SELECT ID FROM {$wpdb->posts} WHERE post_name = %s AND post_type = 'topic' AND post_status IN ('publish', 'closed')", | |
| 3344 | - $topic_slug | |
| 3345 | - )); | |
| 3346 | - | |
| 3347 | - if ($post_id) { | |
| 3348 | - return intval($post_id); | |
| 3349 | - } | |
| 3350 | - } | |
| 3351 | - } | |
| 3352 | - | |
| 3353 | - // For WooCommerce products | |
| 3354 | - if (strpos($url, '/product/') !== false || strpos($url, 'product=') !== false) { | |
| 3355 | - // Extract product slug from various URL formats | |
| 3356 | - $product_slug = ''; | |
| 3357 | - | |
| 3358 | - // Handle pretty permalinks: /product/product-name/ | |
| 3359 | - if (preg_match('/\/product\/([^\/\?]+)/', $url, $matches)) { | |
| 3360 | - $product_slug = $matches[1]; | |
| 3361 | - } | |
| 3362 | - // Handle query parameters: ?product=product-name | |
| 3363 | - elseif (preg_match('/[\?&]product=([^&]+)/', $url, $matches)) { | |
| 3364 | - $product_slug = $matches[1]; | |
| 3365 | - } | |
| 3366 | - | |
| 3367 | - if (!empty($product_slug)) { | |
| 3368 | - // Look up product by slug | |
| 3369 | - $product = get_page_by_path($product_slug, OBJECT, 'product'); | |
| 3370 | - if ($product) { | |
| 3371 | - return $product->ID; | |
| 3372 | - } | |
| 3373 | - | |
| 3374 | - // Alternative method: query by post_name | |
| 3375 | - global $wpdb; | |
| 3376 | - $post_id = $wpdb->get_var($wpdb->prepare( | |
| 3377 | - "SELECT ID FROM {$wpdb->posts} WHERE post_name = %s AND post_type = 'product' AND post_status = 'publish'", | |
| 3378 | - $product_slug | |
| 3379 | - )); | |
| 3380 | - | |
| 3381 | - if ($post_id) { | |
| 3382 | - return intval($post_id); | |
| 3383 | - } | |
| 3384 | - } | |
| 3385 | - } | |
| 3386 | - | |
| 3387 | - // Generic approach: try to extract slug and match against all post types | |
| 3388 | - $parsed_url = wp_parse_url($clean_url); | |
| 3389 | - $path = $parsed_url['path'] ?? ''; | |
| 3390 | - | |
| 3391 | - if (!empty($path)) { | |
| 3392 | - // Get the last part of the path as potential slug | |
| 3393 | - $path_parts = array_filter(explode('/', trim($path, '/'))); | |
| 3394 | - $potential_slug = end($path_parts); | |
| 3395 | - | |
| 3396 | - if (!empty($potential_slug)) { | |
| 3397 | - global $wpdb; | |
| 3398 | - | |
| 3399 | - // Try to find any post with this slug | |
| 3400 | - $post_id = $wpdb->get_var($wpdb->prepare( | |
| 3401 | - "SELECT ID FROM {$wpdb->posts} | |
| 3402 | - WHERE post_name = %s | |
| 3403 | - AND post_status IN ('publish', 'closed', 'private') | |
| 3404 | - AND post_type NOT IN ('revision', 'attachment', 'nav_menu_item') | |
| 3405 | - ORDER BY CASE | |
| 3406 | - WHEN post_type = 'post' THEN 1 | |
| 3407 | - WHEN post_type = 'page' THEN 2 | |
| 3408 | - WHEN post_type = 'topic' THEN 3 | |
| 3409 | - WHEN post_type = 'product' THEN 4 | |
| 3410 | - ELSE 5 | |
| 3411 | - END | |
| 3412 | - LIMIT 1", | |
| 3413 | - $potential_slug | |
| 3414 | - )); | |
| 3415 | - | |
| 3416 | - if ($post_id) { | |
| 3417 | - return intval($post_id); | |
| 3418 | - } | |
| 3419 | - } | |
| 3420 | - } | |
| 3421 | - | |
| 3422 | - // ADDITIONAL: Try direct database lookup by URL variations | |
| 3423 | - global $wpdb; | |
| 3424 | - $table_name = $wpdb->prefix . 'mxchat_system_prompt_content'; | |
| 3425 | - | |
| 3426 | - // Try variations of the URL (with/without trailing slash, http/https) | |
| 3427 | - $url_variations = array( | |
| 3428 | - $url, | |
| 3429 | - rtrim($url, '/'), | |
| 3430 | - $url . '/', | |
| 3431 | - str_replace('http://', 'https://', $url), | |
| 3432 | - str_replace('https://', 'http://', $url), | |
| 3433 | - str_replace('http://', 'https://', rtrim($url, '/')), | |
| 3434 | - str_replace('https://', 'http://', rtrim($url, '/')) | |
| 3435 | - ); | |
| 3436 | - | |
| 3437 | - // Remove duplicates | |
| 3438 | - $url_variations = array_unique($url_variations); | |
| 3439 | - | |
| 3440 | - foreach ($url_variations as $variation) { | |
| 3441 | - $existing_record = $wpdb->get_row($wpdb->prepare( | |
| 3442 | - "SELECT id, source_url FROM $table_name WHERE source_url = %s", | |
| 3443 | - $variation | |
| 3444 | - )); | |
| 3445 | - | |
| 3446 | - if ($existing_record) { | |
| 3447 | - // Try to get post ID from this stored URL | |
| 3448 | - $stored_post_id = url_to_postid($existing_record->source_url); | |
| 3449 | - if ($stored_post_id > 0) { | |
| 3450 | - return $stored_post_id; | |
| 3451 | - } | |
| 3452 | - } | |
| 3453 | - } | |
| 3454 | - | |
| 3455 | - return 0; // No match found | |
| 3456 | -} | |
| 3457 | -/** | |
| 3458 | - * Process selected content via AJAX | |
| 3459 | - */ | |
| 3460 | -public function ajax_mxchat_process_selected_content() { | |
| 3461 | - // Basic request validation | |
| 3462 | - if (!check_ajax_referer('mxchat_content_selector_nonce', 'nonce', false)) { | |
| 3463 | - wp_send_json_error('Invalid nonce'); | |
| 3464 | - exit; | |
| 3465 | - } | |
| 3466 | - | |
| 3467 | - if (!current_user_can('manage_options')) { | |
| 3468 | - wp_send_json_error('Unauthorized access'); | |
| 3469 | - exit; | |
| 3470 | - } | |
| 3471 | - | |
| 3472 | - // Get post IDs - safely parse the array | |
| 3473 | - $post_ids = array(); | |
| 3474 | - if (isset($_POST['post_ids']) && is_array($_POST['post_ids'])) { | |
| 3475 | - foreach ($_POST['post_ids'] as $id) { | |
| 3476 | - $post_ids[] = absint($id); | |
| 3477 | - } | |
| 3478 | - } | |
| 3479 | - | |
| 3480 | - if (empty($post_ids)) { | |
| 3481 | - wp_send_json_error('No content selected'); | |
| 3482 | - exit; | |
| 3483 | - } | |
| 3484 | - | |
| 3485 | - // Get bot_id from request | |
| 3486 | - $bot_id = isset($_POST['bot_id']) ? sanitize_key($_POST['bot_id']) : 'default'; | |
| 3487 | - | |
| 3488 | - // Process only ONE post at a time to avoid request size issues | |
| 3489 | - $post_id = reset($post_ids); | |
| 3490 | - $post = get_post($post_id); | |
| 3491 | - | |
| 3492 | - if (!$post) { | |
| 3493 | - wp_send_json_error('Post not found'); | |
| 3494 | - exit; | |
| 3495 | - } | |
| 3496 | - | |
| 3497 | - // Allow developers to modify post data before processing into knowledge base | |
| 3498 | - $post = apply_filters('mxchat_before_process_post', $post, $bot_id); | |
| 3499 | - | |
| 3500 | - // Get content including title, short description (for WooCommerce), and main content | |
| 3501 | - $content = $post->post_title . "\n\n"; | |
| 3502 | - | |
| 3503 | - // Add short description if it exists (WooCommerce products use post_excerpt for short description) | |
| 3504 | - if (!empty($post->post_excerpt)) { | |
| 3505 | - // Remove shortcode tags but preserve content inside them | |
| 3506 | - $clean_excerpt = $this->strip_shortcode_tags_preserve_content($post->post_excerpt); | |
| 3507 | - $content .= "Short Description: " . wp_strip_all_tags($clean_excerpt) . "\n\n"; | |
| 3508 | - } | |
| 3509 | - | |
| 3510 | - // Add main content - remove shortcode tags but preserve content inside them | |
| 3511 | - $clean_content = $this->strip_shortcode_tags_preserve_content($post->post_content); | |
| 3512 | - $content .= wp_strip_all_tags($clean_content); | |
| 3513 | - | |
| 3514 | - // ADD WOOCOMMERCE PRODUCT DATA (pricing, stock, categories, custom tabs) | |
| 3515 | - if (get_post_type($post_id) === 'product' && class_exists('WooCommerce')) { | |
| 3516 | - $product = wc_get_product($post_id); | |
| 3517 | - | |
| 3518 | - if ($product) { | |
| 3519 | - // Get pricing information | |
| 3520 | - $regular_price = $product->get_regular_price(); | |
| 3521 | - $sale_price = $product->get_sale_price(); | |
| 3522 | - $price = $product->get_price(); | |
| 3523 | - $sku = $product->get_sku(); | |
| 3524 | - | |
| 3525 | - // Get currency symbol | |
| 3526 | - $currency_symbol = get_woocommerce_currency_symbol(); | |
| 3527 | - | |
| 3528 | - // Add pricing information | |
| 3529 | - $content .= "\n"; | |
| 3530 | - if (!empty($regular_price)) { | |
| 3531 | - $content .= "Price: " . $currency_symbol . $regular_price . "\n"; | |
| 3532 | - } elseif (!empty($price)) { | |
| 3533 | - $content .= "Price: " . $currency_symbol . $price . "\n"; | |
| 3534 | - } | |
| 3535 | - | |
| 3536 | - if (!empty($sale_price) && $sale_price !== $regular_price) { | |
| 3537 | - $content .= "Sale Price: " . $currency_symbol . $sale_price . "\n"; | |
| 3538 | - } | |
| 3539 | - | |
| 3540 | - // Handle variable products - show price range | |
| 3541 | - if ($product->is_type('variable')) { | |
| 3542 | - $min_price = $product->get_variation_price('min'); | |
| 3543 | - $max_price = $product->get_variation_price('max'); | |
| 3544 | - if ($min_price !== $max_price) { | |
| 3545 | - $content .= "Price Range: " . $currency_symbol . $min_price . " - " . $currency_symbol . $max_price . "\n"; | |
| 3546 | - } | |
| 3547 | - } | |
| 3548 | - | |
| 3549 | - if (!empty($sku)) { | |
| 3550 | - $content .= "SKU: " . $sku . "\n"; | |
| 3551 | - } | |
| 3552 | - | |
| 3553 | - // Get product categories | |
| 3554 | - $categories = wp_get_post_terms($post_id, 'product_cat', array('fields' => 'names')); | |
| 3555 | - if (!empty($categories) && !is_wp_error($categories)) { | |
| 3556 | - $content .= "Categories: " . implode(', ', $categories) . "\n"; | |
| 3557 | - } | |
| 3558 | - } | |
| 3559 | - | |
| 3560 | - // Get Custom Product Tabs (supports "Custom Product Tabs for WooCommerce" by Code Parrots) | |
| 3561 | - $custom_tabs = get_post_meta($post_id, 'yikes_woo_products_tabs', true); | |
| 3562 | - if (!empty($custom_tabs) && is_array($custom_tabs)) { | |
| 3563 | - foreach ($custom_tabs as $tab) { | |
| 3564 | - $tab_title = isset($tab['title']) ? $tab['title'] : (isset($tab['tab_title']) ? $tab['tab_title'] : ''); | |
| 3565 | - $tab_content = isset($tab['content']) ? $tab['content'] : ''; | |
| 3566 | - | |
| 3567 | - if (!empty($tab_title) && !empty($tab_content)) { | |
| 3568 | - $content .= "\n" . $tab_title . ": " . wp_strip_all_tags($tab_content) . "\n"; | |
| 3569 | - } | |
| 3570 | - } | |
| 3571 | - } | |
| 3572 | - | |
| 3573 | - // Also check for reusable/saved tabs applied to this product | |
| 3574 | - $applied_saved_tabs = get_post_meta($post_id, 'yikes_woo_reusable_products_tabs_applied', true); | |
| 3575 | - if (!empty($applied_saved_tabs) && is_array($applied_saved_tabs)) { | |
| 3576 | - $saved_tabs = get_option('yikes_woo_reusable_products_tabs', array()); | |
| 3577 | - if (!empty($saved_tabs) && is_array($saved_tabs)) { | |
| 3578 | - foreach ($applied_saved_tabs as $saved_tab_id) { | |
| 3579 | - if (isset($saved_tabs[$saved_tab_id])) { | |
| 3580 | - $tab = $saved_tabs[$saved_tab_id]; | |
| 3581 | - $tab_title = isset($tab['title']) ? $tab['title'] : (isset($tab['tab_title']) ? $tab['tab_title'] : ''); | |
| 3582 | - $tab_content = isset($tab['content']) ? $tab['content'] : ''; | |
| 3583 | - | |
| 3584 | - if (!empty($tab_title) && !empty($tab_content)) { | |
| 3585 | - $content .= "\n" . $tab_title . ": " . wp_strip_all_tags($tab_content) . "\n"; | |
| 3586 | - } | |
| 3587 | - } | |
| 3588 | - } | |
| 3589 | - } | |
| 3590 | - } | |
| 3591 | - } | |
| 3592 | - | |
| 3593 | - // ADD ACF FIELDS SUPPORT | |
| 3594 | - $acf_fields = $this->mxchat_get_acf_fields_for_post($post_id); | |
| 3595 | - if (!empty($acf_fields)) { | |
| 3596 | - $acf_content_parts = array(); | |
| 3597 | - | |
| 3598 | - foreach ($acf_fields as $field_name => $field_value) { | |
| 3599 | - $formatted_value = $this->mxchat_format_acf_field_value($field_value, $field_name, $post_id); | |
| 3600 | - | |
| 3601 | - if (!empty($formatted_value)) { | |
| 3602 | - $field_label = ucwords(str_replace('_', ' ', $field_name)); | |
| 3603 | - $acf_content_parts[] = $field_label . ": " . $formatted_value; | |
| 3604 | - } | |
| 3605 | - } | |
| 3606 | - | |
| 3607 | - if (!empty($acf_content_parts)) { | |
| 3608 | - $content .= "\n\n" . implode("\n", $acf_content_parts); | |
| 3609 | - } | |
| 3610 | - } | |
| 3611 | - | |
| 3612 | - // ADD CUSTOM POST META SUPPORT (whitelisted non-ACF meta fields) | |
| 3613 | - $custom_meta = $this->mxchat_get_whitelisted_post_meta($post_id); | |
| 3614 | - if (!empty($custom_meta)) { | |
| 3615 | - $meta_content_parts = array(); | |
| 3616 | - | |
| 3617 | - foreach ($custom_meta as $meta_key => $meta_value) { | |
| 3618 | - // Convert meta key to readable label | |
| 3619 | - $meta_label = ucwords(str_replace(array('_', '-'), ' ', $meta_key)); | |
| 3620 | - $meta_content_parts[] = $meta_label . ": " . $meta_value; | |
| 3621 | - } | |
| 3622 | - | |
| 3623 | - if (!empty($meta_content_parts)) { | |
| 3624 | - $content .= "\n\n" . implode("\n", $meta_content_parts); | |
| 3625 | - } | |
| 3626 | - } | |
| 3627 | - | |
| 3628 | - // Debug logging for WordPress Import content | |
| 3629 | - //error_log('[MXCHAT-WP-IMPORT-DEBUG] Post ID: ' . $post_id . ' Title: ' . $post->post_title); | |
| 3630 | - //error_log('[MXCHAT-WP-IMPORT-DEBUG] Raw post_content length: ' . strlen($post->post_content)); | |
| 3631 | - //error_log('[MXCHAT-WP-IMPORT-DEBUG] Final content length: ' . strlen($content)); | |
| 3632 | - //error_log('[MXCHAT-WP-IMPORT-DEBUG] Content preview: ' . substr($content, 0, 300)); | |
| 3633 | - | |
| 3634 | - // Note: Removed 10,000 char limit - chunking now handles large content properly | |
| 3635 | - | |
| 3636 | - // Get bot-specific API key | |
| 3637 | - $bot_options = $this->get_bot_options($bot_id); | |
| 3638 | - $options = !empty($bot_options) ? $bot_options : get_option('mxchat_options'); | |
| 3639 | - $selected_model = $options['embedding_model'] ?? 'text-embedding-ada-002'; | |
| 3640 | - | |
| 3641 | - if (strpos($selected_model, 'voyage') === 0) { | |
| 3642 | - $api_key = $options['voyage_api_key'] ?? ''; | |
| 3643 | - $provider_name = 'Voyage AI'; | |
| 3644 | - } elseif (strpos($selected_model, 'gemini-embedding') === 0) { | |
| 3645 | - $api_key = $options['gemini_api_key'] ?? ''; | |
| 3646 | - $provider_name = 'Google Gemini'; | |
| 3647 | - } else { | |
| 3648 | - $api_key = $options['api_key'] ?? ''; | |
| 3649 | - $provider_name = 'OpenAI'; | |
| 3650 | - } | |
| 3651 | - | |
| 3652 | - if (empty($api_key)) { | |
| 3653 | - MxChat_Admin::mxchat_log_debug('api_error', $provider_name . ' API key not configured for knowledge processing'); | |
| 3654 | - wp_send_json_error($provider_name . ' API key not configured'); | |
| 3655 | - exit; | |
| 3656 | - } | |
| 3657 | - | |
| 3658 | - $source_url = get_permalink($post_id); | |
| 3659 | - $vector_id = md5($source_url); // Vector ID for Pinecone | |
| 3660 | - | |
| 3661 | - // Check for existing content in bot-specific storage | |
| 3662 | - $is_update = false; | |
| 3663 | - | |
| 3664 | - // Get bot-specific Pinecone configuration | |
| 3665 | - $bot_pinecone_config = $this->get_bot_pinecone_config($bot_id); | |
| 3666 | - $use_pinecone = !empty($bot_pinecone_config) && ($bot_pinecone_config['use_pinecone'] ?? false); | |
| 3667 | - | |
| 3668 | - if ($use_pinecone && !empty($bot_pinecone_config['api_key'])) { | |
| 3669 | - // Check Pinecone for this bot | |
| 3670 | - $pinecone_data = $this->mxchat_get_pinecone_processed_content($bot_pinecone_config); | |
| 3671 | - if (isset($pinecone_data[$post_id])) { | |
| 3672 | - $is_update = true; | |
| 3673 | - } | |
| 3674 | - } else { | |
| 3675 | - // Check WordPress DB (same as before since it's shared) | |
| 3676 | - global $wpdb; | |
| 3677 | - $table_name = $wpdb->prefix . 'mxchat_system_prompt_content'; | |
| 3678 | - $existing_record = $wpdb->get_row($wpdb->prepare( | |
| 3679 | - "SELECT id FROM $table_name WHERE source_url = %s", | |
| 3680 | - $source_url | |
| 3681 | - )); | |
| 3682 | - | |
| 3683 | - if ($existing_record) { | |
| 3684 | - $is_update = true; | |
| 3685 | - } | |
| 3686 | - } | |
| 3687 | - | |
| 3688 | - // UPDATED 2.5.6: Determine content type based on post_type | |
| 3689 | - $post_type = $post->post_type; | |
| 3690 | - $content_type = 'content'; // Default fallback | |
| 3691 | - | |
| 3692 | - // Map WordPress post types to content types | |
| 3693 | - switch ($post_type) { | |
| 3694 | - case 'post': | |
| 3695 | - $content_type = 'post'; | |
| 3696 | - break; | |
| 3697 | - case 'page': | |
| 3698 | - $content_type = 'page'; | |
| 3699 | - break; | |
| 3700 | - case 'product': | |
| 3701 | - $content_type = 'product'; | |
| 3702 | - break; | |
| 3703 | - default: | |
| 3704 | - // For custom post types, use the post type name | |
| 3705 | - $content_type = sanitize_key($post_type); | |
| 3706 | - break; | |
| 3707 | - } | |
| 3708 | - | |
| 3709 | - // Use the centralized utility function with bot_id and content_type | |
| 3710 | - $result = MxChat_Utils::submit_content_to_db( | |
| 3711 | - $content, | |
| 3712 | - $source_url, | |
| 3713 | - $api_key, | |
| 3714 | - $vector_id, | |
| 3715 | - $bot_id, | |
| 3716 | - $content_type | |
| 3717 | - ); | |
| 3718 | - | |
| 3719 | - if (is_wp_error($result)) { | |
| 3720 | - MxChat_Admin::mxchat_log_debug('storage_error', 'Knowledge storage failed: ' . $result->get_error_message(), array('source_url' => $source_url)); | |
| 3721 | - wp_send_json_error('Storage failed: ' . $result->get_error_message()); | |
| 3722 | - exit; | |
| 3723 | - } | |
| 3724 | - | |
| 3725 | - // Automatically apply role restriction based on tags | |
| 3726 | - $this->apply_role_restriction_to_post($post_id, $source_url); | |
| 3727 | - | |
| 3728 | - $operation_type = $is_update ? 'update' : 'new'; | |
| 3729 | - | |
| 3730 | - // Count ACF fields for debugging | |
| 3731 | - $acf_field_count = count($acf_fields); | |
| 3732 | - | |
| 3733 | - // Success response with minimal data | |
| 3734 | - wp_send_json_success(array( | |
| 3735 | - 'message' => $operation_type === 'update' ? 'Content updated successfully' : 'Content processed successfully', | |
| 3736 | - 'post_id' => $post_id, | |
| 3737 | - 'title' => $post->post_title, | |
| 3738 | - 'operation_type' => $operation_type, | |
| 3739 | - 'vector_id' => $vector_id, | |
| 3740 | - 'acf_fields_found' => $acf_field_count, | |
| 3741 | - 'content_preview' => substr($content, 0, 100) . '...', | |
| 3742 | - 'bot_id' => $bot_id | |
| 3743 | - )); | |
| 3744 | - exit; | |
| 3745 | -} | |
| 3746 | - | |
| 3747 | -private function apply_role_restriction_to_post($post_id, $source_url) { | |
| 3748 | - // Get tag-role mappings | |
| 3749 | - $mappings = get_option('mxchat_tag_role_mappings', array()); | |
| 3750 | - | |
| 3751 | - if (empty($mappings)) { | |
| 3752 | - return; // No mappings, leave as public | |
| 3753 | - } | |
| 3754 | - | |
| 3755 | - // Get all tags for the post | |
| 3756 | - $post_tags = wp_get_post_tags($post_id, array('fields' => 'slugs')); | |
| 3757 | - | |
| 3758 | - if (empty($post_tags)) { | |
| 3759 | - return; // No tags, leave as public | |
| 3760 | - } | |
| 3761 | - | |
| 3762 | - // Determine the highest role restriction based on tags | |
| 3763 | - $highest_role = 'public'; | |
| 3764 | - $role_hierarchy = array( | |
| 3765 | - 'public' => 0, | |
| 3766 | - 'logged_in' => 1, | |
| 3767 | - 'subscriber' => 2, | |
| 3768 | - 'contributor' => 3, | |
| 3769 | - 'author' => 4, | |
| 3770 | - 'editor' => 5, | |
| 3771 | - 'administrator' => 6 | |
| 3772 | - ); | |
| 3773 | - | |
| 3774 | - foreach ($post_tags as $tag_slug) { | |
| 3775 | - if (isset($mappings[$tag_slug])) { | |
| 3776 | - $role = $mappings[$tag_slug]; | |
| 3777 | - if (isset($role_hierarchy[$role]) && $role_hierarchy[$role] > $role_hierarchy[$highest_role]) { | |
| 3778 | - $highest_role = $role; | |
| 3779 | - } | |
| 3780 | - } | |
| 3781 | - } | |
| 3782 | - | |
| 3783 | - // If no restricted tags found, return (leave as public) | |
| 3784 | - if ($highest_role === 'public') { | |
| 3785 | - return; | |
| 3786 | - } | |
| 3787 | - | |
| 3788 | - // Update the role restriction in the database | |
| 3789 | - global $wpdb; | |
| 3790 | - | |
| 3791 | - // Check if using Pinecone | |
| 3792 | - $pinecone_options = get_option('mxchat_pinecone_addon_options', array()); | |
| 3793 | - $use_pinecone = ($pinecone_options['mxchat_use_pinecone'] ?? '0') === '1'; | |
| 3794 | - | |
| 3795 | - if ($use_pinecone && !empty($pinecone_options['mxchat_pinecone_api_key'])) { | |
| 3796 | - // Update Pinecone role restriction | |
| 3797 | - $roles_table = $wpdb->prefix . 'mxchat_pinecone_roles'; | |
| 3798 | - $vector_id = md5($source_url); | |
| 3799 | - | |
| 3800 | - $wpdb->replace( | |
| 3801 | - $roles_table, | |
| 3802 | - array( | |
| 3803 | - 'vector_id' => $vector_id, | |
| 3804 | - 'role_restriction' => $highest_role, | |
| 3805 | - 'updated_at' => current_time('mysql') | |
| 3806 | - ), | |
| 3807 | - array('%s', '%s', '%s') | |
| 3808 | - ); | |
| 3809 | - } else { | |
| 3810 | - // Update WordPress DB | |
| 3811 | - $table_name = $wpdb->prefix . 'mxchat_system_prompt_content'; | |
| 3812 | - | |
| 3813 | - $wpdb->update( | |
| 3814 | - $table_name, | |
| 3815 | - array('role_restriction' => $highest_role), | |
| 3816 | - array('source_url' => $source_url), | |
| 3817 | - array('%s'), | |
| 3818 | - array('%s') | |
| 3819 | - ); | |
| 3820 | - } | |
| 3821 | -} | |
| 3822 | - | |
| 3823 | -public function mxchat_get_public_post_types() { | |
| 3824 | - // Get all public post types | |
| 3825 | - $post_types = get_post_types(array('public' => true), 'objects'); | |
| 3826 | - $post_type_options = array(); | |
| 3827 | - | |
| 3828 | - foreach ($post_types as $post_type) { | |
| 3829 | - $post_type_options[$post_type->name] = $post_type->label; | |
| 3830 | - } | |
| 3831 | - | |
| 3832 | - // Also include common forum/community post types that might not be marked as public | |
| 3833 | - $additional_types = array( | |
| 3834 | - 'topic' => 'Forum Topics (bbPress)', | |
| 3835 | - 'reply' => 'Forum Replies (bbPress)', | |
| 3836 | - 'forum' => 'Forums (bbPress)', | |
| 3837 | - 'wpforo_topic' => 'wpForo Topics', | |
| 3838 | - 'wpforo_post' => 'wpForo Posts' | |
| 3839 | - ); | |
| 3840 | - | |
| 3841 | - foreach ($additional_types as $type_name => $type_label) { | |
| 3842 | - if (post_type_exists($type_name) && !isset($post_type_options[$type_name])) { | |
| 3843 | - $post_type_options[$type_name] = $type_label; | |
| 3844 | - } | |
| 3845 | - } | |
| 3846 | - | |
| 3847 | - return $post_type_options; | |
| 3848 | -} | |
| 3849 | - | |
| 3850 | -/** | |
| 3851 | - * Retrieves processed content from Pinecone API | |
| 3852 | - */ | |
| 3853 | -public function mxchat_get_pinecone_processed_content($pinecone_options) { | |
| 3854 | - $api_key = $pinecone_options['mxchat_pinecone_api_key'] ?? ''; | |
| 3855 | - $host = $pinecone_options['mxchat_pinecone_host'] ?? ''; | |
| 3856 | - | |
| 3857 | - if (empty($api_key) || empty($host)) { | |
| 3858 | - return array(); | |
| 3859 | - } | |
| 3860 | - | |
| 3861 | - $pinecone_data = array(); | |
| 3862 | - | |
| 3863 | - try { | |
| 3864 | - // Always get fresh data from Pinecone | |
| 3865 | - $pinecone_data = $this->mxchat_scan_pinecone_for_processed_content($pinecone_options); | |
| 3866 | - | |
| 3867 | - // Method 2: Final fallback - try stats endpoint (if available) | |
| 3868 | - if (empty($pinecone_data)) { | |
| 3869 | - $stats_url = "https://{$host}/describe_index_stats"; | |
| 3870 | - | |
| 3871 | - $response = wp_remote_post($stats_url, array( | |
| 3872 | - 'headers' => array( | |
| 3873 | - 'Api-Key' => $api_key, | |
| 3874 | - 'Content-Type' => 'application/json' | |
| 3875 | - ), | |
| 3876 | - 'body' => json_encode(array()), | |
| 3877 | - 'timeout' => 30 | |
| 3878 | - )); | |
| 3879 | - | |
| 3880 | - if (!is_wp_error($response) && wp_remote_retrieve_response_code($response) === 200) { | |
| 3881 | - $body = wp_remote_retrieve_body($response); | |
| 3882 | - $stats_data = json_decode($body, true); | |
| 3883 | - } | |
| 3884 | - } | |
| 3885 | - | |
| 3886 | - } catch (Exception $e) { | |
| 3887 | - // Log error but return fresh data only | |
| 3888 | - } | |
| 3889 | - | |
| 3890 | - return $pinecone_data; | |
| 3891 | -} | |
| 3892 | -public function mxchat_fetch_pinecone_vectors_by_ids($pinecone_options, $vector_ids) { | |
| 3893 | - //error_log('=== DEBUG: Starting mxchat_fetch_pinecone_vectors_by_ids ==='); | |
| 3894 | - | |
| 3895 | - $api_key = $pinecone_options['mxchat_pinecone_api_key'] ?? ''; | |
| 3896 | - $host = $pinecone_options['mxchat_pinecone_host'] ?? ''; | |
| 3897 | - | |
| 3898 | - if (empty($api_key) || empty($host) || empty($vector_ids)) { | |
| 3899 | - //error_log('DEBUG: Missing parameters for fetch by IDs'); | |
| 3900 | - return array(); | |
| 3901 | - } | |
| 3902 | - | |
| 3903 | - try { | |
| 3904 | - $fetch_url = "https://{$host}/vectors/fetch"; | |
| 3905 | - //error_log('DEBUG: Fetch URL: ' . $fetch_url); | |
| 3906 | - //error_log('DEBUG: Fetching ' . count($vector_ids) . ' vector IDs'); | |
| 3907 | - | |
| 3908 | - // Pinecone fetch API allows fetching specific vectors by ID | |
| 3909 | - $fetch_data = array( | |
| 3910 | - 'ids' => array_values($vector_ids) | |
| 3911 | - ); | |
| 3912 | - | |
| 3913 | - $response = wp_remote_post($fetch_url, array( | |
| 3914 | - 'headers' => array( | |
| 3915 | - 'Api-Key' => $api_key, | |
| 3916 | - 'Content-Type' => 'application/json' | |
| 3917 | - ), | |
| 3918 | - 'body' => json_encode($fetch_data), | |
| 3919 | - 'timeout' => 30 | |
| 3920 | - )); | |
| 3921 | - | |
| 3922 | - if (is_wp_error($response)) { | |
| 3923 | - //error_log('DEBUG: Fetch by IDs WP error: ' . $response->get_error_message()); | |
| 3924 | - return array(); | |
| 3925 | - } | |
| 3926 | - | |
| 3927 | - $response_code = wp_remote_retrieve_response_code($response); | |
| 3928 | - //error_log('DEBUG: Fetch response code: ' . $response_code); | |
| 3929 | - | |
| 3930 | - if ($response_code !== 200) { | |
| 3931 | - $error_body = wp_remote_retrieve_body($response); | |
| 3932 | - //error_log('DEBUG: Fetch failed with body: ' . $error_body); | |
| 3933 | - return array(); | |
| 3934 | - } | |
| 3935 | - | |
| 3936 | - $body = wp_remote_retrieve_body($response); | |
| 3937 | - $data = json_decode($body, true); | |
| 3938 | - | |
| 3939 | - //error_log('DEBUG: Fetch response structure: ' . print_r(array_keys($data), true)); | |
| 3940 | - | |
| 3941 | - if (!isset($data['vectors'])) { | |
| 3942 | - //error_log('DEBUG: No vectors key in response'); | |
| 3943 | - return array(); | |
| 3944 | - } | |
| 3945 | - | |
| 3946 | - $processed_data = array(); | |
| 3947 | - | |
| 3948 | - foreach ($data['vectors'] as $vector_id => $vector_data) { | |
| 3949 | - $metadata = $vector_data['metadata'] ?? array(); | |
| 3950 | - $source_url = $metadata['source_url'] ?? ''; | |
| 3951 | - | |
| 3952 | - if (!empty($source_url)) { | |
| 3953 | - $post_id = url_to_postid($source_url); | |
| 3954 | - if ($post_id) { | |
| 3955 | - $created_at = $metadata['created_at'] ?? ''; | |
| 3956 | - $processed_date = 'Recently'; | |
| 3957 | - | |
| 3958 | - if (!empty($created_at)) { | |
| 3959 | - $timestamp = is_numeric($created_at) ? $created_at : strtotime($created_at); | |
| 3960 | - if ($timestamp) { | |
| 3961 | - $processed_date = human_time_diff($timestamp, current_time('timestamp')) . ' ago'; | |
| 3962 | - } | |
| 3963 | - } | |
| 3964 | - | |
| 3965 | - $processed_data[$post_id] = array( | |
| 3966 | - 'db_id' => $vector_id, | |
| 3967 | - 'processed_date' => $processed_date, | |
| 3968 | - 'url' => $source_url, | |
| 3969 | - 'source' => 'pinecone', | |
| 3970 | - 'timestamp' => $timestamp ?? current_time('timestamp') | |
| 3971 | - ); | |
| 3972 | - } | |
| 3973 | - } | |
| 3974 | - } | |
| 3975 | - | |
| 3976 | - //error_log('DEBUG: Processed ' . count($processed_data) . ' vectors from fetch'); | |
| 3977 | - return $processed_data; | |
| 3978 | - | |
| 3979 | - } catch (Exception $e) { | |
| 3980 | - //error_log('DEBUG: Exception in fetch_pinecone_vectors_by_ids: ' . $e->getMessage()); | |
| 3981 | - return array(); | |
| 3982 | - } | |
| 3983 | -} | |
| 3984 | - | |
| 3985 | -/** | |
| 3986 | - * Get embedding dimensions based on the selected model. | |
| 3987 | - */ | |
| 3988 | -private function mxchat_get_embedding_dimensions() { | |
| 3989 | - $options = get_option('mxchat_options', array()); | |
| 3990 | - $selected_model = $options['embedding_model'] ?? 'text-embedding-ada-002'; | |
| 3991 | - | |
| 3992 | - $model_dimensions = array( | |
| 3993 | - 'text-embedding-ada-002' => 1536, | |
| 3994 | - 'text-embedding-3-small' => 1536, | |
| 3995 | - 'text-embedding-3-large' => 3072, | |
| 3996 | - 'voyage-2' => 1024, | |
| 3997 | - 'voyage-large-2' => 1536, | |
| 3998 | - 'voyage-3-large' => 2048, | |
| 3999 | - 'gemini-embedding-001' => 1536, | |
| 4000 | - ); | |
| 4001 | - | |
| 4002 | - if (strpos($selected_model, 'voyage-3-large') === 0) { | |
| 4003 | - $custom_dimensions = $options['voyage_output_dimension'] ?? 2048; | |
| 4004 | - return intval($custom_dimensions); | |
| 4005 | - } | |
| 4006 | - | |
| 4007 | - if (strpos($selected_model, 'gemini-embedding') === 0) { | |
| 4008 | - $custom_dimensions = $options['gemini_output_dimension'] ?? 1536; | |
| 4009 | - return intval($custom_dimensions); | |
| 4010 | - } | |
| 4011 | - | |
| 4012 | - return $model_dimensions[$selected_model] ?? 1536; | |
| 4013 | -} | |
| 4014 | - | |
| 4015 | -/** | |
| 4016 | - * Scan Pinecone for processed content | |
| 4017 | - */ | |
| 4018 | -public function mxchat_scan_pinecone_for_processed_content($pinecone_options) { | |
| 4019 | - $api_key = $pinecone_options['mxchat_pinecone_api_key'] ?? ''; | |
| 4020 | - $host = $pinecone_options['mxchat_pinecone_host'] ?? ''; | |
| 4021 | - | |
| 4022 | - if (empty($api_key) || empty($host)) { | |
| 4023 | - return array(); | |
| 4024 | - } | |
| 4025 | - | |
| 4026 | - try { | |
| 4027 | - // Use multiple random vectors to get better coverage | |
| 4028 | - $all_matches = array(); | |
| 4029 | - $seen_ids = array(); | |
| 4030 | - | |
| 4031 | - // Get correct dimensions for the configured embedding model | |
| 4032 | - $dimensions = $this->mxchat_get_embedding_dimensions(); | |
| 4033 | - | |
| 4034 | - // Try 3 different random vectors to get better coverage | |
| 4035 | - for ($i = 0; $i < 3; $i++) { | |
| 4036 | - $query_url = "https://{$host}/query"; | |
| 4037 | - | |
| 4038 | - // Generate a random unit vector instead of zeros | |
| 4039 | - $random_vector = array(); | |
| 4040 | - for ($j = 0; $j < $dimensions; $j++) { | |
| 4041 | - $random_vector[] = (rand(-1000, 1000) / 1000.0); | |
| 4042 | - } | |
| 4043 | - | |
| 4044 | - // Normalize the vector to unit length | |
| 4045 | - $magnitude = sqrt(array_sum(array_map(function($x) { return $x * $x; }, $random_vector))); | |
| 4046 | - if ($magnitude > 0) { | |
| 4047 | - $random_vector = array_map(function($x) use ($magnitude) { return $x / $magnitude; }, $random_vector); | |
| 4048 | - } | |
| 4049 | - | |
| 4050 | - $query_data = array( | |
| 4051 | - 'includeMetadata' => true, | |
| 4052 | - 'includeValues' => false, | |
| 4053 | - 'topK' => 10000, | |
| 4054 | - 'vector' => $random_vector | |
| 4055 | - ); | |
| 4056 | - | |
| 4057 | - $response = wp_remote_post($query_url, array( | |
| 4058 | - 'headers' => array( | |
| 4059 | - 'Api-Key' => $api_key, | |
| 4060 | - 'Content-Type' => 'application/json' | |
| 4061 | - ), | |
| 4062 | - 'body' => json_encode($query_data), | |
| 4063 | - 'timeout' => 30 | |
| 4064 | - )); | |
| 4065 | - | |
| 4066 | - if (is_wp_error($response)) { | |
| 4067 | - continue; | |
| 4068 | - } | |
| 4069 | - | |
| 4070 | - $response_code = wp_remote_retrieve_response_code($response); | |
| 4071 | - | |
| 4072 | - if ($response_code !== 200) { | |
| 4073 | - continue; | |
| 4074 | - } | |
| 4075 | - | |
| 4076 | - $body = wp_remote_retrieve_body($response); | |
| 4077 | - $data = json_decode($body, true); | |
| 4078 | - | |
| 4079 | - if (isset($data['matches'])) { | |
| 4080 | - foreach ($data['matches'] as $match) { | |
| 4081 | - $match_id = $match['id'] ?? ''; | |
| 4082 | - if (!empty($match_id) && !isset($seen_ids[$match_id])) { | |
| 4083 | - $all_matches[] = $match; | |
| 4084 | - $seen_ids[$match_id] = true; | |
| 4085 | - } | |
| 4086 | - } | |
| 4087 | - } | |
| 4088 | - } | |
| 4089 | - | |
| 4090 | - // Convert matches to processed data format, grouping by URL to count chunks | |
| 4091 | - $processed_data = array(); | |
| 4092 | - $url_chunk_counts = array(); | |
| 4093 | - | |
| 4094 | - foreach ($all_matches as $match) { | |
| 4095 | - $metadata = $match['metadata'] ?? array(); | |
| 4096 | - $source_url = $metadata['source_url'] ?? ''; | |
| 4097 | - $match_id = $match['id'] ?? ''; | |
| 4098 | - | |
| 4099 | - if (!empty($source_url) && !empty($match_id)) { | |
| 4100 | - $post_id = url_to_postid($source_url); | |
| 4101 | - if ($post_id) { | |
| 4102 | - // Count chunks per post_id | |
| 4103 | - if (!isset($url_chunk_counts[$post_id])) { | |
| 4104 | - $url_chunk_counts[$post_id] = 0; | |
| 4105 | - } | |
| 4106 | - $url_chunk_counts[$post_id]++; | |
| 4107 | - | |
| 4108 | - $created_at = $metadata['created_at'] ?? ''; | |
| 4109 | - $processed_date = 'Recently'; | |
| 4110 | - | |
| 4111 | - if (!empty($created_at)) { | |
| 4112 | - $timestamp = is_numeric($created_at) ? $created_at : strtotime($created_at); | |
| 4113 | - if ($timestamp) { | |
| 4114 | - $processed_date = human_time_diff($timestamp, current_time('timestamp')) . ' ago'; | |
| 4115 | - } | |
| 4116 | - } | |
| 4117 | - | |
| 4118 | - // Only store if not already set, or update with newer timestamp | |
| 4119 | - if (!isset($processed_data[$post_id]) || | |
| 4120 | - ($timestamp ?? 0) > ($processed_data[$post_id]['timestamp'] ?? 0)) { | |
| 4121 | - $processed_data[$post_id] = array( | |
| 4122 | - 'db_id' => $match_id, | |
| 4123 | - 'processed_date' => $processed_date, | |
| 4124 | - 'url' => $source_url, | |
| 4125 | - 'source' => 'pinecone', | |
| 4126 | - 'timestamp' => $timestamp ?? current_time('timestamp') | |
| 4127 | - ); | |
| 4128 | - } | |
| 4129 | - } | |
| 4130 | - } | |
| 4131 | - } | |
| 4132 | - | |
| 4133 | - // Add chunk counts to processed data | |
| 4134 | - foreach ($url_chunk_counts as $post_id => $chunk_count) { | |
| 4135 | - if (isset($processed_data[$post_id])) { | |
| 4136 | - $processed_data[$post_id]['chunk_count'] = $chunk_count; | |
| 4137 | - } | |
| 4138 | - } | |
| 4139 | - | |
| 4140 | - return $processed_data; | |
| 4141 | - | |
| 4142 | - } catch (Exception $e) { | |
| 4143 | - return array(); | |
| 4144 | - } | |
| 4145 | -} | |
| 4146 | -/** | |
| 4147 | - * Generate embeddings from input text for MXChat with bot support | |
| 4148 | - */ | |
| 4149 | -private function mxchat_generate_embedding($text, $bot_id = 'default') { | |
| 4150 | - // Enable detailed logging for debugging | |
| 4151 | - //error_log('[MXCHAT-EMBED] Starting embedding generation for bot: ' . $bot_id . '. Text length: ' . strlen($text) . ' bytes'); | |
| 4152 | - //error_log('[MXCHAT-EMBED] Text preview: ' . substr($text, 0, 100) . '...'); | |
| 4153 | - | |
| 4154 | - // Get bot-specific options | |
| 4155 | - $bot_options = $this->get_bot_options($bot_id); | |
| 4156 | - $options = !empty($bot_options) ? $bot_options : get_option('mxchat_options'); | |
| 4157 | - | |
| 4158 | - $selected_model = $options['embedding_model'] ?? 'text-embedding-ada-002'; | |
| 4159 | - //error_log('[MXCHAT-EMBED] Selected embedding model for bot ' . $bot_id . ': ' . $selected_model); | |
| 4160 | - | |
| 4161 | - // Determine provider and endpoint | |
| 4162 | - if (strpos($selected_model, 'voyage') === 0) { | |
| 4163 | - $api_key = $options['voyage_api_key'] ?? ''; | |
| 4164 | - $endpoint = 'https://api.voyageai.com/v1/embeddings'; | |
| 4165 | - $provider_name = 'Voyage AI'; | |
| 4166 | - //error_log('[MXCHAT-EMBED] Using Voyage AI API for bot ' . $bot_id); | |
| 4167 | - } elseif (strpos($selected_model, 'gemini-embedding') === 0) { | |
| 4168 | - $api_key = $options['gemini_api_key'] ?? ''; | |
| 4169 | - $endpoint = 'https://generativelanguage.googleapis.com/v1beta/models/' . $selected_model . ':embedContent'; | |
| 4170 | - $provider_name = 'Google Gemini'; | |
| 4171 | - //error_log('[MXCHAT-EMBED] Using Google Gemini API for bot ' . $bot_id); | |
| 4172 | - } else { | |
| 4173 | - $api_key = $options['api_key'] ?? ''; | |
| 4174 | - $endpoint = 'https://api.openai.com/v1/embeddings'; | |
| 4175 | - $provider_name = 'OpenAI'; | |
| 4176 | - //error_log('[MXCHAT-EMBED] Using OpenAI API for bot ' . $bot_id); | |
| 4177 | - } | |
| 4178 | - | |
| 4179 | - //error_log('[MXCHAT-EMBED] Using endpoint: ' . $endpoint); | |
| 4180 | - | |
| 4181 | - if (empty($api_key)) { | |
| 4182 | - $error_message = sprintf('Missing %s API key for bot %s. Please configure your API key in the bot settings.', $provider_name, $bot_id); | |
| 4183 | - //error_log('[MXCHAT-EMBED] Error: ' . $error_message); | |
| 4184 | - return $error_message; | |
| 4185 | - } | |
| 4186 | - | |
| 4187 | - // Check if text is too long (for OpenAI, roughly estimate tokens as words/0.75) | |
| 4188 | - $estimated_tokens = ceil(str_word_count($text) / 0.75); | |
| 4189 | - //error_log('[MXCHAT-EMBED] Estimated token count: ~' . $estimated_tokens); | |
| 4190 | - | |
| 4191 | - if ($estimated_tokens > 8000 && strpos($selected_model, 'voyage') === false && strpos($selected_model, 'gemini-embedding') === false) { | |
| 4192 | - //error_log('[MXCHAT-EMBED] Warning: Text may exceed OpenAI token limits (8K for most models)'); | |
| 4193 | - // Consider truncating text here | |
| 4194 | - } | |
| 4195 | - | |
| 4196 | - // Prepare request body based on provider | |
| 4197 | - if (strpos($selected_model, 'gemini-embedding') === 0) { | |
| 4198 | - // Gemini API format | |
| 4199 | - $request_body = array( | |
| 4200 | - 'model' => 'models/' . $selected_model, | |
| 4201 | - 'content' => array( | |
| 4202 | - 'parts' => array( | |
| 4203 | - array('text' => $text) | |
| 4204 | - ) | |
| 4205 | - ) | |
| 4206 | - ); | |
| 4207 | - | |
| 4208 | - // Set output dimensionality to 1536 for consistency with other models | |
| 4209 | - $request_body['outputDimensionality'] = 1536; | |
| 4210 | - } else { | |
| 4211 | - // OpenAI/Voyage API format | |
| 4212 | - $request_body = array( | |
| 4213 | - 'model' => $selected_model, | |
| 4214 | - 'input' => $text | |
| 4215 | - ); | |
| 4216 | - | |
| 4217 | - // Add output_dimension for voyage-3-large model | |
| 4218 | - if ($selected_model === 'voyage-3-large') { | |
| 4219 | - $request_body['output_dimension'] = 2048; | |
| 4220 | - } | |
| 4221 | - } | |
| 4222 | - | |
| 4223 | - //error_log('[MXCHAT-EMBED] Request prepared with model: ' . $selected_model); | |
| 4224 | - | |
| 4225 | - // Prepare headers based on provider | |
| 4226 | - if (strpos($selected_model, 'gemini-embedding') === 0) { | |
| 4227 | - // Gemini uses API key as query parameter | |
| 4228 | - $endpoint .= '?key=' . $api_key; | |
| 4229 | - $headers = array( | |
| 4230 | - 'Content-Type' => 'application/json' | |
| 4231 | - ); | |
| 4232 | - } else { | |
| 4233 | - // OpenAI/Voyage use Bearer token | |
| 4234 | - $headers = array( | |
| 4235 | - 'Authorization' => 'Bearer ' . $api_key, | |
| 4236 | - 'Content-Type' => 'application/json' | |
| 4237 | - ); | |
| 4238 | - } | |
| 4239 | - | |
| 4240 | - // Make API request | |
| 4241 | - //error_log('[MXCHAT-EMBED] Sending API request to: ' . $endpoint); | |
| 4242 | - $response = wp_remote_post($endpoint, array( | |
| 4243 | - 'body' => wp_json_encode($request_body), | |
| 4244 | - 'headers' => $headers, | |
| 4245 | - 'timeout' => 60 // Increased timeout for large inputs | |
| 4246 | - )); | |
| 4247 | - | |
| 4248 | - // Handle wp_remote_post errors | |
| 4249 | - if (is_wp_error($response)) { | |
| 4250 | - $error_message = $response->get_error_message(); | |
| 4251 | - //error_log('[MXCHAT-EMBED] WP Remote Post Error: ' . $error_message); | |
| 4252 | - return 'Connection error: ' . $error_message; | |
| 4253 | - } | |
| 4254 | - | |
| 4255 | - // Get and check HTTP response code | |
| 4256 | - $http_code = wp_remote_retrieve_response_code($response); | |
| 4257 | - //error_log('[MXCHAT-EMBED] API Response Code: ' . $http_code); | |
| 4258 | - | |
| 4259 | - if ($http_code !== 200) { | |
| 4260 | - $error_body = wp_remote_retrieve_body($response); | |
| 4261 | - //error_log('[MXCHAT-EMBED] API Error Response Body: ' . $error_body); | |
| 4262 | - | |
| 4263 | - // Try to parse error for more details | |
| 4264 | - $error_json = json_decode($error_body, true); | |
| 4265 | - if (json_last_error() === JSON_ERROR_NONE && isset($error_json['error'])) { | |
| 4266 | - $error_type = $error_json['error']['type'] ?? 'unknown'; | |
| 4267 | - $error_message = $error_json['error']['message'] ?? 'No message'; | |
| 4268 | - //error_log('[MXCHAT-EMBED] API Error Type: ' . $error_type); | |
| 4269 | - //error_log('[MXCHAT-EMBED] API Error Message: ' . $error_message); | |
| 4270 | - | |
| 4271 | - // Customize error message for common API errors | |
| 4272 | - if ($error_type === 'invalid_request_error' && strpos($error_message, 'API key') !== false) { | |
| 4273 | - $error_message = sprintf('Invalid %s API key for bot %s. Please check your API key in the bot settings.', $provider_name, $bot_id); | |
| 4274 | - } elseif ($error_type === 'authentication_error') { | |
| 4275 | - $error_message = sprintf('%s authentication failed for bot %s. Please verify your API key in the bot settings.', $provider_name, $bot_id); | |
| 4276 | - } | |
| 4277 | - | |
| 4278 | - //error_log('[MXCHAT-EMBED] Returning error: ' . $error_message); | |
| 4279 | - return $error_message; | |
| 4280 | - } | |
| 4281 | - | |
| 4282 | - $error_message = sprintf("API Error (HTTP %d): Unable to generate embedding for bot %s", $http_code, $bot_id); | |
| 4283 | - //error_log('[MXCHAT-EMBED] Returning error: ' . $error_message); | |
| 4284 | - return $error_message; | |
| 4285 | - } | |
| 4286 | - | |
| 4287 | - // Parse response body | |
| 4288 | - $response_body = wp_remote_retrieve_body($response); | |
| 4289 | - //error_log('[MXCHAT-EMBED] Received response length: ' . strlen($response_body) . ' bytes'); | |
| 4290 | - | |
| 4291 | - $response_data = json_decode($response_body, true); | |
| 4292 | - | |
| 4293 | - if (json_last_error() !== JSON_ERROR_NONE) { | |
| 4294 | - $error = json_last_error_msg(); | |
| 4295 | - //error_log('[MXCHAT-EMBED] JSON Parse Error: ' . $error); | |
| 4296 | - //error_log('[MXCHAT-EMBED] Response preview: ' . substr($response_body, 0, 200)); | |
| 4297 | - return "Failed to parse API response: $error"; | |
| 4298 | - } | |
| 4299 | - | |
| 4300 | - // Handle different response formats based on provider | |
| 4301 | - if (strpos($selected_model, 'gemini-embedding') === 0) { | |
| 4302 | - // Gemini API response format | |
| 4303 | - if (isset($response_data['embedding']['values'])) { | |
| 4304 | - $embedding_dimensions = count($response_data['embedding']['values']); | |
| 4305 | - //error_log('[MXCHAT-EMBED] Successfully extracted Gemini embedding with ' . $embedding_dimensions . ' dimensions'); | |
| 4306 | - | |
| 4307 | - // Check if embedding dimensions are as expected (should be 1536) | |
| 4308 | - if ($embedding_dimensions !== 1536) { | |
| 4309 | - //error_log('[MXCHAT-EMBED] Warning: Unexpected Gemini embedding dimensions: ' . $embedding_dimensions); | |
| 4310 | - } | |
| 4311 | - | |
| 4312 | - return $response_data['embedding']['values']; | |
| 4313 | - } else { | |
| 4314 | - //error_log('[MXCHAT-EMBED] Error: No embedding found in Gemini response'); | |
| 4315 | - //error_log('[MXCHAT-EMBED] Response structure: ' . wp_json_encode(array_keys($response_data))); | |
| 4316 | - | |
| 4317 | - if (isset($response_data['error'])) { | |
| 4318 | - $error_message = "Gemini API Error in response: " . wp_json_encode($response_data['error']); | |
| 4319 | - //error_log('[MXCHAT-EMBED] ' . $error_message); | |
| 4320 | - return $error_message; | |
| 4321 | - } | |
| 4322 | - | |
| 4323 | - $error_message = "Invalid Gemini API response format: No embedding found"; | |
| 4324 | - //error_log('[MXCHAT-EMBED] ' . $error_message); | |
| 4325 | - return $error_message; | |
| 4326 | - } | |
| 4327 | - } else { | |
| 4328 | - // OpenAI/Voyage API response format | |
| 4329 | - if (isset($response_data['data'][0]['embedding'])) { | |
| 4330 | - $embedding_dimensions = count($response_data['data'][0]['embedding']); | |
| 4331 | - //error_log('[MXCHAT-EMBED] Successfully extracted embedding with ' . $embedding_dimensions . ' dimensions'); | |
| 4332 | - | |
| 4333 | - // Check if embedding dimensions are as expected | |
| 4334 | - if (($selected_model === 'text-embedding-ada-002' && $embedding_dimensions !== 1536) || | |
| 4335 | - ($selected_model === 'voyage-3-large' && $embedding_dimensions !== 2048)) { | |
| 4336 | - //error_log('[MXCHAT-EMBED] Warning: Unexpected embedding dimensions'); | |
| 4337 | - } | |
| 4338 | - | |
| 4339 | - return $response_data['data'][0]['embedding']; | |
| 4340 | - } else { | |
| 4341 | - //error_log('[MXCHAT-EMBED] Error: No embedding found in response'); | |
| 4342 | - //error_log('[MXCHAT-EMBED] Response structure: ' . wp_json_encode(array_keys($response_data))); | |
| 4343 | - | |
| 4344 | - if (isset($response_data['error'])) { | |
| 4345 | - $error_message = "API Error in response: " . wp_json_encode($response_data['error']); | |
| 4346 | - //error_log('[MXCHAT-EMBED] ' . $error_message); | |
| 4347 | - return $error_message; | |
| 4348 | - } | |
| 4349 | - | |
| 4350 | - $error_message = "Invalid API response format: No embedding found"; | |
| 4351 | - //error_log('[MXCHAT-EMBED] ' . $error_message); | |
| 4352 | - return $error_message; | |
| 4353 | - } | |
| 4354 | - } | |
| 4355 | -} | |
| 4356 | - | |
| 4357 | -/** | |
| 4358 | - * Get bot-specific options for multi-bot functionality | |
| 4359 | - * Falls back to default options if bot_id is 'default' or multi-bot add-on is not active | |
| 4360 | - */ | |
| 4361 | -private function get_bot_options($bot_id = 'default') { | |
| 4362 | - //error_log("MXCHAT DEBUG: get_bot_options called for bot: " . $bot_id); | |
| 4363 | - | |
| 4364 | - if ($bot_id === 'default' || !class_exists('MxChat_Multi_Bot_Manager')) { | |
| 4365 | - //error_log("MXCHAT DEBUG: Using default options (no multi-bot or bot is 'default')"); | |
| 4366 | - return array(); | |
| 4367 | - } | |
| 4368 | - | |
| 4369 | - $bot_options = apply_filters('mxchat_get_bot_options', array(), $bot_id); | |
| 4370 | - | |
| 4371 | - if (!empty($bot_options)) { | |
| 4372 | - //error_log("MXCHAT DEBUG: Got bot-specific options from filter"); | |
| 4373 | - if (isset($bot_options['similarity_threshold'])) { | |
| 4374 | - //error_log(" - similarity_threshold: " . $bot_options['similarity_threshold']); | |
| 4375 | - } | |
| 4376 | - } | |
| 4377 | - | |
| 4378 | - return is_array($bot_options) ? $bot_options : array(); | |
| 4379 | -} | |
| 4380 | - | |
| 4381 | -/** | |
| 4382 | - * Get bot-specific Pinecone configuration | |
| 4383 | - * Used in the knowledge retrieval functions | |
| 4384 | - */ | |
| 4385 | -// Also add debugging to your get_bot_pinecone_config function | |
| 4386 | -private function get_bot_pinecone_config($bot_id = 'default') { | |
| 4387 | - //error_log("MXCHAT DEBUG: get_bot_pinecone_config called for bot: " . $bot_id); | |
| 4388 | - | |
| 4389 | - // If default bot or multi-bot add-on not active, use default Pinecone config | |
| 4390 | - if ($bot_id === 'default' || !class_exists('MxChat_Multi_Bot_Manager')) { | |
| 4391 | - //error_log("MXCHAT DEBUG: Using default Pinecone config (no multi-bot or bot is 'default')"); | |
| 4392 | - $addon_options = get_option('mxchat_pinecone_addon_options', array()); | |
| 4393 | - $config = array( | |
| 4394 | - 'use_pinecone' => (isset($addon_options['mxchat_use_pinecone']) && $addon_options['mxchat_use_pinecone'] === '1'), | |
| 4395 | - 'api_key' => $addon_options['mxchat_pinecone_api_key'] ?? '', | |
| 4396 | - 'host' => $addon_options['mxchat_pinecone_host'] ?? '', | |
| 4397 | - 'namespace' => $addon_options['mxchat_pinecone_namespace'] ?? '' | |
| 4398 | - ); | |
| 4399 | - //error_log("MXCHAT DEBUG: Default config - use_pinecone: " . ($config['use_pinecone'] ? 'true' : 'false')); | |
| 4400 | - return $config; | |
| 4401 | - } | |
| 4402 | - | |
| 4403 | - //error_log("MXCHAT DEBUG: Calling filter 'mxchat_get_bot_pinecone_config' for bot: " . $bot_id); | |
| 4404 | - | |
| 4405 | - // Hook for multi-bot add-on to provide bot-specific Pinecone config | |
| 4406 | - $bot_pinecone_config = apply_filters('mxchat_get_bot_pinecone_config', array(), $bot_id); | |
| 4407 | - | |
| 4408 | - if (!empty($bot_pinecone_config)) { | |
| 4409 | - //error_log("MXCHAT DEBUG: Got bot-specific config from filter"); | |
| 4410 | - //error_log(" - use_pinecone: " . (isset($bot_pinecone_config['use_pinecone']) ? ($bot_pinecone_config['use_pinecone'] ? 'true' : 'false') : 'not set')); | |
| 4411 | - //error_log(" - host: " . ($bot_pinecone_config['host'] ?? 'not set')); | |
| 4412 | - //error_log(" - namespace: " . ($bot_pinecone_config['namespace'] ?? 'not set')); | |
| 4413 | - } else { | |
| 4414 | - //error_log("MXCHAT DEBUG: Filter returned empty config!"); | |
| 4415 | - } | |
| 4416 | - | |
| 4417 | - return is_array($bot_pinecone_config) ? $bot_pinecone_config : array(); | |
| 4418 | -} | |
| 4419 | - | |
| 4420 | - | |
| 4421 | -public function mxchat_ajax_dismiss_completed_status() { | |
| 4422 | - try { | |
| 4423 | - // Verify the request | |
| 4424 | - check_ajax_referer('mxchat_status_nonce', 'nonce'); | |
| 4425 | - | |
| 4426 | - if (!current_user_can('manage_options')) { | |
| 4427 | - wp_send_json_error('Unauthorized access'); | |
| 4428 | - exit; | |
| 4429 | - } | |
| 4430 | - | |
| 4431 | - $card_type = isset($_POST['card_type']) ? sanitize_text_field($_POST['card_type']) : ''; | |
| 4432 | - | |
| 4433 | - if ($card_type === 'pdf') { | |
| 4434 | - // Clear PDF status | |
| 4435 | - $pdf_url = get_transient('mxchat_last_pdf_url'); | |
| 4436 | - if ($pdf_url) { | |
| 4437 | - delete_transient('mxchat_pdf_status_' . md5($pdf_url)); | |
| 4438 | - delete_transient('mxchat_last_pdf_url'); | |
| 4439 | - } | |
| 4440 | - } elseif ($card_type === 'sitemap') { | |
| 4441 | - // Clear sitemap status | |
| 4442 | - $sitemap_url = get_transient('mxchat_last_sitemap_url'); | |
| 4443 | - if ($sitemap_url) { | |
| 4444 | - delete_transient('mxchat_sitemap_status_' . md5($sitemap_url)); | |
| 4445 | - delete_transient('mxchat_last_sitemap_url'); | |
| 4446 | - } | |
| 4447 | - } | |
| 4448 | - | |
| 4449 | - wp_send_json_success(array('message' => 'Status dismissed successfully')); | |
| 4450 | - | |
| 4451 | - } catch (Exception $e) { | |
| 4452 | - wp_send_json_error(array('message' => 'Error dismissing status: ' . $e->getMessage())); | |
| 4453 | - } | |
| 4454 | -} | |
| 4455 | - | |
| 4456 | -/** | |
| 4457 | - * Render completed status cards on page load | |
| 4458 | - * This ensures completed processing status persists through page refreshes | |
| 4459 | - */ | |
| 4460 | -public function mxchat_render_completed_status_cards() { | |
| 4461 | - $output = ''; | |
| 4462 | - | |
| 4463 | - // Check for completed PDF status | |
| 4464 | - $pdf_url = get_transient('mxchat_last_pdf_url'); | |
| 4465 | - if ($pdf_url) { | |
| 4466 | - $pdf_status = $this->mxchat_get_pdf_processing_status($pdf_url); | |
| 4467 | - if ($pdf_status && ($pdf_status['status'] === 'complete' || $pdf_status['status'] === 'error')) { | |
| 4468 | - $output .= $this->mxchat_render_pdf_status_card($pdf_status, $pdf_url); | |
| 4469 | - } | |
| 4470 | - } | |
| 4471 | - | |
| 4472 | - // Check for completed sitemap status | |
| 4473 | - $sitemap_url = get_transient('mxchat_last_sitemap_url'); | |
| 4474 | - if ($sitemap_url) { | |
| 4475 | - $sitemap_status = $this->mxchat_get_sitemap_processing_status($sitemap_url); | |
| 4476 | - if ($sitemap_status && ($sitemap_status['status'] === 'complete' || $sitemap_status['status'] === 'error')) { | |
| 4477 | - $output .= $this->mxchat_render_sitemap_status_card($sitemap_status, $sitemap_url); | |
| 4478 | - } | |
| 4479 | - } | |
| 4480 | - | |
| 4481 | - return $output; | |
| 4482 | -} | |
| 4483 | - | |
| 4484 | -/** | |
| 4485 | - * Render PDF status card HTML | |
| 4486 | - */ | |
| 4487 | -private function mxchat_render_pdf_status_card($status, $pdf_url) { | |
| 4488 | - $html = '<div class="mxchat-status-card" data-card-type="pdf">'; | |
| 4489 | - $html .= '<div class="mxchat-status-header">'; | |
| 4490 | - $html .= '<h4>' . esc_html__('PDF Processing Status', 'mxchat') . '</h4>'; | |
| 4491 | - | |
| 4492 | - // Add dismiss button for completed status | |
| 4493 | - if ($status['status'] === 'complete' || $status['status'] === 'error') { | |
| 4494 | - $html .= '<button type="button" class="mxchat-dismiss-button">' . esc_html__('Dismiss', 'mxchat') . '</button>'; | |
| 4495 | - } | |
| 4496 | - | |
| 4497 | - // Process Batch button for processing status | |
| 4498 | - if ($status['status'] === 'processing') { | |
| 4499 | - $html .= '<button type="button" class="mxchat-manual-batch-btn" | |
| 4500 | - data-process-type="pdf" | |
| 4501 | - data-url="' . esc_attr($pdf_url) . '"> | |
| 4502 | - ' . esc_html__('Process Batch', 'mxchat') . '</button>'; | |
| 4503 | - } | |
| 4504 | - | |
| 4505 | - // Add status badges | |
| 4506 | - if ($status['status'] === 'error') { | |
| 4507 | - $html .= '<span class="mxchat-status-badge mxchat-status-failed">' . esc_html__('Error', 'mxchat') . '</span>'; | |
| 4508 | - } elseif ($status['status'] === 'complete') { | |
| 4509 | - if ($status['failed_pages'] > 0) { | |
| 4510 | - $html .= '<span class="mxchat-status-badge mxchat-status-warning">' . | |
| 4511 | - sprintf(esc_html__('Completed with %d failures', 'mxchat'), $status['failed_pages']) . '</span>'; | |
| 4512 | - } else { | |
| 4513 | - $html .= '<span class="mxchat-status-badge mxchat-status-success">' . esc_html__('Complete', 'mxchat') . '</span>'; | |
| 4514 | - } | |
| 4515 | - } | |
| 4516 | - | |
| 4517 | - $html .= '</div>'; // End header | |
| 4518 | - | |
| 4519 | - // Progress bar | |
| 4520 | - $html .= '<div class="mxchat-progress-bar">'; | |
| 4521 | - $html .= '<div class="mxchat-progress-fill" style="width: ' . esc_attr($status['percentage']) . '%"></div>'; | |
| 4522 | - $html .= '</div>'; | |
| 4523 | - | |
| 4524 | - // Status details | |
| 4525 | - $html .= '<div class="mxchat-status-details">'; | |
| 4526 | - $html .= '<p>' . sprintf( | |
| 4527 | - esc_html__('Progress: %d of %d pages (%d%%)', 'mxchat'), | |
| 4528 | - $status['processed_pages'], | |
| 4529 | - $status['total_pages'], | |
| 4530 | - $status['percentage'] | |
| 4531 | - ) . '</p>'; | |
| 4532 | - | |
| 4533 | - // Show failed pages count if any | |
| 4534 | - if ($status['failed_pages'] > 0) { | |
| 4535 | - $html .= '<p><strong>' . esc_html__('Failed pages:', 'mxchat') . '</strong> ' . esc_html($status['failed_pages']) . '</p>'; | |
| 4536 | - } | |
| 4537 | - | |
| 4538 | - $html .= '<p><strong>' . esc_html__('Status:', 'mxchat') . '</strong> ' . esc_html(ucfirst($status['status'])) . '</p>'; | |
| 4539 | - $html .= '<p><strong>' . esc_html__('Last update:', 'mxchat') . '</strong> ' . esc_html($status['last_update']) . '</p>'; | |
| 4540 | - | |
| 4541 | - // Add completion summary if available AND it's an array | |
| 4542 | - if (isset($status['completion_summary']) && is_array($status['completion_summary']) && !empty($status['completion_summary'])) { | |
| 4543 | - $summary = $status['completion_summary']; | |
| 4544 | - $html .= '<div class="mxchat-completion-summary">'; | |
| 4545 | - $html .= '<h5>' . esc_html__('Processing Summary', 'mxchat') . '</h5>'; | |
| 4546 | - $html .= '<p><strong>' . esc_html__('Total Pages:', 'mxchat') . '</strong> ' . esc_html($summary['total_pages']) . '</p>'; | |
| 4547 | - $html .= '<p><strong>' . esc_html__('Successful:', 'mxchat') . '</strong> ' . esc_html($summary['successful_pages']) . '</p>'; | |
| 4548 | - $html .= '<p><strong>' . esc_html__('Failed:', 'mxchat') . '</strong> ' . esc_html($summary['failed_pages']) . '</p>'; | |
| 4549 | - $html .= '<p><strong>' . esc_html__('Completed:', 'mxchat') . '</strong> ' . esc_html($summary['completion_time']) . '</p>'; | |
| 4550 | - $html .= '</div>'; | |
| 4551 | - } | |
| 4552 | - | |
| 4553 | - // Add failed pages list if any AND it's an array | |
| 4554 | - if (isset($status['failed_pages_list']) && is_array($status['failed_pages_list']) && !empty($status['failed_pages_list'])) { | |
| 4555 | - $html .= $this->mxchat_render_failed_pages_list($status['failed_pages_list']); | |
| 4556 | - } | |
| 4557 | - | |
| 4558 | - // Add error message if any | |
| 4559 | - if (isset($status['error']) && !empty($status['error'])) { | |
| 4560 | - $html .= '<div class="mxchat-error-notice">'; | |
| 4561 | - $html .= '<p class="error">' . esc_html($status['error']) . '</p>'; | |
| 4562 | - $html .= '</div>'; | |
| 4563 | - } | |
| 4564 | - | |
| 4565 | - $html .= '</div>'; // End details | |
| 4566 | - $html .= '</div>'; // End card | |
| 4567 | - | |
| 4568 | - return $html; | |
| 4569 | -} | |
| 4570 | -/** | |
| 4571 | - * Render sitemap status card HTML | |
| 4572 | - */ | |
| 4573 | -private function mxchat_render_sitemap_status_card($status, $sitemap_url) { | |
| 4574 | - $html = '<div class="mxchat-status-card" data-card-type="sitemap">'; | |
| 4575 | - $html .= '<div class="mxchat-status-header">'; | |
| 4576 | - $html .= '<h4>' . esc_html__('Sitemap Processing Status', 'mxchat') . '</h4>'; | |
| 4577 | - | |
| 4578 | - // Add dismiss button for completed status | |
| 4579 | - if ($status['status'] === 'complete' || $status['status'] === 'error') { | |
| 4580 | - $html .= '<button type="button" class="mxchat-dismiss-button">' . esc_html__('Dismiss', 'mxchat') . '</button>'; | |
| 4581 | - } | |
| 4582 | - | |
| 4583 | - // Process Batch button for processing status | |
| 4584 | - if ($status['status'] === 'processing') { | |
| 4585 | - $html .= '<button type="button" class="mxchat-manual-batch-btn" | |
| 4586 | - data-process-type="sitemap" | |
| 4587 | - data-url="' . esc_attr($sitemap_url) . '"> | |
| 4588 | - ' . esc_html__('Process Batch', 'mxchat') . '</button>'; | |
| 4589 | - } | |
| 4590 | - | |
| 4591 | - // Add status badges | |
| 4592 | - if ($status['status'] === 'error') { | |
| 4593 | - $html .= '<span class="mxchat-status-badge mxchat-status-failed">' . esc_html__('Error', 'mxchat') . '</span>'; | |
| 4594 | - } elseif ($status['status'] === 'complete') { | |
| 4595 | - if ($status['failed_urls'] > 0) { | |
| 4596 | - $html .= '<span class="mxchat-status-badge mxchat-status-warning">' . | |
| 4597 | - sprintf(esc_html__('Completed with %d failures', 'mxchat'), $status['failed_urls']) . '</span>'; | |
| 4598 | - } else { | |
| 4599 | - $html .= '<span class="mxchat-status-badge mxchat-status-success">' . esc_html__('Complete', 'mxchat') . '</span>'; | |
| 4600 | - } | |
| 4601 | - } | |
| 4602 | - | |
| 4603 | - $html .= '</div>'; // End header | |
| 4604 | - | |
| 4605 | - // Progress bar | |
| 4606 | - $html .= '<div class="mxchat-progress-bar">'; | |
| 4607 | - $html .= '<div class="mxchat-progress-fill" style="width: ' . esc_attr($status['percentage']) . '%"></div>'; | |
| 4608 | - $html .= '</div>'; | |
| 4609 | - | |
| 4610 | - // Status details | |
| 4611 | - $html .= '<div class="mxchat-status-details">'; | |
| 4612 | - $html .= '<p>' . sprintf( | |
| 4613 | - esc_html__('Progress: %d of %d URLs (%d%%)', 'mxchat'), | |
| 4614 | - $status['processed_urls'], | |
| 4615 | - $status['total_urls'], | |
| 4616 | - $status['percentage'] | |
| 4617 | - ) . '</p>'; | |
| 4618 | - | |
| 4619 | - // Show failed URLs count if any | |
| 4620 | - if ($status['failed_urls'] > 0) { | |
| 4621 | - $html .= '<p><strong>' . esc_html__('Failed URLs:', 'mxchat') . '</strong> ' . esc_html($status['failed_urls']) . '</p>'; | |
| 4622 | - } | |
| 4623 | - | |
| 4624 | - $html .= '<p><strong>' . esc_html__('Status:', 'mxchat') . '</strong> ' . esc_html(ucfirst($status['status'])) . '</p>'; | |
| 4625 | - $html .= '<p><strong>' . esc_html__('Last update:', 'mxchat') . '</strong> ' . esc_html($status['last_update']) . '</p>'; | |
| 4626 | - | |
| 4627 | - // Add completion summary if available AND it's an array | |
| 4628 | - if (isset($status['completion_summary']) && is_array($status['completion_summary']) && !empty($status['completion_summary'])) { | |
| 4629 | - $summary = $status['completion_summary']; | |
| 4630 | - $html .= '<div class="mxchat-completion-summary">'; | |
| 4631 | - $html .= '<h5>' . esc_html__('Processing Summary', 'mxchat') . '</h5>'; | |
| 4632 | - $html .= '<p><strong>' . esc_html__('Total URLs:', 'mxchat') . '</strong> ' . esc_html($summary['total_urls']) . '</p>'; | |
| 4633 | - $html .= '<p><strong>' . esc_html__('Successful:', 'mxchat') . '</strong> ' . esc_html($summary['successful_urls']) . '</p>'; | |
| 4634 | - $html .= '<p><strong>' . esc_html__('Failed:', 'mxchat') . '</strong> ' . esc_html($summary['failed_urls']) . '</p>'; | |
| 4635 | - $html .= '<p><strong>' . esc_html__('Completed:', 'mxchat') . '</strong> ' . esc_html($summary['completion_time']) . '</p>'; | |
| 4636 | - $html .= '</div>'; | |
| 4637 | - } | |
| 4638 | - | |
| 4639 | - // Add error messages if any (but not the failed URLs list) | |
| 4640 | - if (!empty($status['error']) || !empty($status['last_error'])) { | |
| 4641 | - $html .= '<div class="mxchat-error-notice">'; | |
| 4642 | - | |
| 4643 | - if (!empty($status['error'])) { | |
| 4644 | - $html .= '<p class="error">' . esc_html($status['error']) . '</p>'; | |
| 4645 | - } | |
| 4646 | - | |
| 4647 | - if (!empty($status['last_error'])) { | |
| 4648 | - $html .= '<p class="last-error">' . esc_html__('Last error:', 'mxchat') . ' ' . esc_html($status['last_error']) . '</p>'; | |
| 4649 | - } | |
| 4650 | - | |
| 4651 | - $html .= '</div>'; | |
| 4652 | - } | |
| 4653 | - | |
| 4654 | - $html .= '</div>'; // End details | |
| 4655 | - $html .= '</div>'; // End card | |
| 4656 | - | |
| 4657 | - return $html; | |
| 4658 | -} | |
| 4659 | - | |
| 4660 | - | |
| 4661 | -/** | |
| 4662 | - * Render failed pages list | |
| 4663 | - */ | |
| 4664 | -private function mxchat_render_failed_pages_list($failed_pages_list) { | |
| 4665 | - // Validate that $failed_pages_list is an array and not empty | |
| 4666 | - if (!is_array($failed_pages_list) || empty($failed_pages_list)) { | |
| 4667 | - return ''; | |
| 4668 | - } | |
| 4669 | - | |
| 4670 | - $html = '<div class="mxchat-error-notice">'; | |
| 4671 | - $html .= '<div class="mxchat-failed-pages-container">'; | |
| 4672 | - $html .= '<h5>' . sprintf(esc_html__('Failed Pages (%d)', 'mxchat'), count($failed_pages_list)) . '</h5>'; | |
| 4673 | - $html .= '<details>'; | |
| 4674 | - $html .= '<summary>' . esc_html__('Show Failed Pages', 'mxchat') . '</summary>'; | |
| 4675 | - $html .= '<div class="mxchat-failed-pages-list">'; | |
| 4676 | - | |
| 4677 | - // Create table for failed pages | |
| 4678 | - $html .= '<table class="widefat striped">'; | |
| 4679 | - $html .= '<thead><tr>'; | |
| 4680 | - $html .= '<th>' . esc_html__('Page', 'mxchat') . '</th>'; | |
| 4681 | - $html .= '<th>' . esc_html__('Error', 'mxchat') . '</th>'; | |
| 4682 | - $html .= '<th>' . esc_html__('Retries', 'mxchat') . '</th>'; | |
| 4683 | - $html .= '<th>' . esc_html__('Time', 'mxchat') . '</th>'; | |
| 4684 | - $html .= '</tr></thead><tbody>'; | |
| 4685 | - | |
| 4686 | - // Sort failed pages by most recent | |
| 4687 | - $sorted_failed_pages = $failed_pages_list; | |
| 4688 | - usort($sorted_failed_pages, function($a, $b) { | |
| 4689 | - return ($b['time'] ?? 0) - ($a['time'] ?? 0); | |
| 4690 | - }); | |
| 4691 | - | |
| 4692 | - foreach ($sorted_failed_pages as $item) { | |
| 4693 | - // Ensure $item is an array before accessing its elements | |
| 4694 | - if (!is_array($item)) { | |
| 4695 | - continue; | |
| 4696 | - } | |
| 4697 | - | |
| 4698 | - $time_ago = isset($item['time']) ? human_time_diff($item['time'], current_time('timestamp')) . ' ' . esc_html__('ago', 'mxchat') : 'Unknown'; | |
| 4699 | - $html .= '<tr>'; | |
| 4700 | - $html .= '<td>' . esc_html__('Page', 'mxchat') . ' ' . esc_html($item['page'] ?? 'Unknown') . '</td>'; | |
| 4701 | - $html .= '<td style="word-break: break-word;">' . esc_html($item['error'] ?? 'Unknown error') . '</td>'; | |
| 4702 | - $html .= '<td>' . esc_html($item['retries'] ?? 'N/A') . '</td>'; | |
| 4703 | - $html .= '<td>' . esc_html($time_ago) . '</td>'; | |
| 4704 | - $html .= '</tr>'; | |
| 4705 | - } | |
| 4706 | - | |
| 4707 | - $html .= '</tbody></table>'; | |
| 4708 | - $html .= '</div></details></div></div>'; | |
| 4709 | - | |
| 4710 | - return $html; | |
| 4711 | -} | |
| 4712 | - | |
| 4713 | -/** | |
| 4714 | - * Render failed URLs list | |
| 4715 | - */ | |
| 4716 | -private function mxchat_render_failed_urls_list($failed_urls_list) { | |
| 4717 | - // Validate that $failed_urls_list is an array and not empty | |
| 4718 | - if (!is_array($failed_urls_list) || empty($failed_urls_list)) { | |
| 4719 | - return ''; | |
| 4720 | - } | |
| 4721 | - | |
| 4722 | - $html = '<div class="mxchat-failed-urls-container">'; | |
| 4723 | - $html .= '<h5>' . sprintf(esc_html__('Failed URLs (%d)', 'mxchat'), count($failed_urls_list)) . '</h5>'; | |
| 4724 | - $html .= '<details>'; | |
| 4725 | - $html .= '<summary>' . esc_html__('Show Failed URLs', 'mxchat') . '</summary>'; | |
| 4726 | - $html .= '<div class="mxchat-failed-urls-list">'; | |
| 4727 | - | |
| 4728 | - // Create table for failed URLs | |
| 4729 | - $html .= '<table class="widefat striped">'; | |
| 4730 | - $html .= '<thead><tr>'; | |
| 4731 | - $html .= '<th>' . esc_html__('URL', 'mxchat') . '</th>'; | |
| 4732 | - $html .= '<th>' . esc_html__('Error', 'mxchat') . '</th>'; | |
| 4733 | - $html .= '<th>' . esc_html__('Retries', 'mxchat') . '</th>'; | |
| 4734 | - $html .= '<th>' . esc_html__('Time', 'mxchat') . '</th>'; | |
| 4735 | - $html .= '</tr></thead><tbody>'; | |
| 4736 | - | |
| 4737 | - // Sort failed URLs by most recent | |
| 4738 | - $sorted_failed_urls = $failed_urls_list; | |
| 4739 | - usort($sorted_failed_urls, function($a, $b) { | |
| 4740 | - return ($b['time'] ?? 0) - ($a['time'] ?? 0); | |
| 4741 | - }); | |
| 4742 | - | |
| 4743 | - // Show up to 50 failed URLs | |
| 4744 | - $display_urls = array_slice($sorted_failed_urls, 0, 50); | |
| 4745 | - | |
| 4746 | - foreach ($display_urls as $item) { | |
| 4747 | - // Ensure $item is an array before accessing its elements | |
| 4748 | - if (!is_array($item)) { | |
| 4749 | - continue; | |
| 4750 | - } | |
| 4751 | - | |
| 4752 | - $url = $item['url'] ?? ''; | |
| 4753 | - $time_ago = isset($item['time']) ? human_time_diff($item['time'], current_time('timestamp')) . ' ' . esc_html__('ago', 'mxchat') : 'Unknown'; | |
| 4754 | - | |
| 4755 | - // Truncate URL for display | |
| 4756 | - $display_url = strlen($url) > 50 ? substr($url, 0, 47) . '...' : $url; | |
| 4757 | - | |
| 4758 | - $html .= '<tr>'; | |
| 4759 | - $html .= '<td style="word-break: break-all;">'; | |
| 4760 | - if (!empty($url)) { | |
| 4761 | - $html .= '<a href="' . esc_url($url) . '" target="_blank" rel="noopener noreferrer">' . esc_html($display_url) . '</a>'; | |
| 4762 | - } else { | |
| 4763 | - $html .= esc_html__('Unknown URL', 'mxchat'); | |
| 4764 | - } | |
| 4765 | - $html .= '</td>'; | |
| 4766 | - $html .= '<td style="word-break: break-word;">' . esc_html($item['error'] ?? 'Unknown error') . '</td>'; | |
| 4767 | - $html .= '<td>' . esc_html($item['retries'] ?? 'N/A') . '</td>'; | |
| 4768 | - $html .= '<td>' . esc_html($time_ago) . '</td>'; | |
| 4769 | - $html .= '</tr>'; | |
| 4770 | - } | |
| 4771 | - | |
| 4772 | - $html .= '</tbody></table>'; | |
| 4773 | - | |
| 4774 | - if (count($failed_urls_list) > 50) { | |
| 4775 | - $html .= '<div class="mxchat-failed-urls-more">+ ' . | |
| 4776 | - (count($failed_urls_list) - 50) . | |
| 4777 | - ' ' . esc_html__('more failed URLs not shown', 'mxchat') . '</div>'; | |
| 4778 | - } | |
| 4779 | - | |
| 4780 | - $html .= '</div></details></div>'; | |
| 4781 | - | |
| 4782 | - return $html; | |
| 4783 | -} | |
| 4784 | - | |
| 4785 | -/** | |
| 4786 | - * Get all ACF fields for a specific post, excluding any fields the user has disabled | |
| 4787 | - */ | |
| 4788 | -public function mxchat_get_acf_fields_for_post($post_id) { | |
| 4789 | - if (!function_exists('get_fields')) { | |
| 4790 | - return array(); | |
| 4791 | - } | |
| 4792 | - | |
| 4793 | - $fields = get_fields($post_id); | |
| 4794 | - if (!$fields || !is_array($fields)) { | |
| 4795 | - return array(); | |
| 4796 | - } | |
| 4797 | - | |
| 4798 | - // Get excluded fields from settings | |
| 4799 | - $excluded_fields = get_option('mxchat_acf_excluded_fields', array()); | |
| 4800 | - if (!empty($excluded_fields) && is_array($excluded_fields)) { | |
| 4801 | - foreach ($excluded_fields as $excluded_field) { | |
| 4802 | - if (isset($fields[$excluded_field])) { | |
| 4803 | - unset($fields[$excluded_field]); | |
| 4804 | - } | |
| 4805 | - } | |
| 4806 | - } | |
| 4807 | - | |
| 4808 | - return $fields; | |
| 4809 | -} | |
| 4810 | - | |
| 4811 | -/** | |
| 4812 | - * Get all registered ACF field groups and their fields for the settings UI | |
| 4813 | - */ | |
| 4814 | -public function mxchat_get_all_acf_fields() { | |
| 4815 | - if (!function_exists('acf_get_field_groups') || !function_exists('acf_get_fields')) { | |
| 4816 | - return array(); | |
| 4817 | - } | |
| 4818 | - | |
| 4819 | - $all_fields = array(); | |
| 4820 | - $field_groups = acf_get_field_groups(); | |
| 4821 | - | |
| 4822 | - if (!empty($field_groups)) { | |
| 4823 | - foreach ($field_groups as $group) { | |
| 4824 | - $group_fields = acf_get_fields($group['key']); | |
| 4825 | - if (!empty($group_fields)) { | |
| 4826 | - $all_fields[$group['title']] = array(); | |
| 4827 | - foreach ($group_fields as $field) { | |
| 4828 | - $all_fields[$group['title']][] = array( | |
| 4829 | - 'name' => $field['name'], | |
| 4830 | - 'label' => $field['label'], | |
| 4831 | - 'type' => $field['type'] | |
| 4832 | - ); | |
| 4833 | - } | |
| 4834 | - } | |
| 4835 | - } | |
| 4836 | - } | |
| 4837 | - | |
| 4838 | - return $all_fields; | |
| 4839 | -} | |
| 4840 | - | |
| 4841 | -/** | |
| 4842 | - * Get whitelisted custom post meta for a given post | |
| 4843 | - * This allows non-ACF meta fields (like OptionTree, theme meta boxes) to be included in embeddings | |
| 4844 | - */ | |
| 4845 | -public function mxchat_get_whitelisted_post_meta($post_id) { | |
| 4846 | - $whitelist = get_option('mxchat_custom_meta_whitelist', ''); | |
| 4847 | - | |
| 4848 | - if (empty($whitelist)) { | |
| 4849 | - return array(); | |
| 4850 | - } | |
| 4851 | - | |
| 4852 | - // Parse the whitelist - one meta key per line | |
| 4853 | - $meta_keys = array_filter(array_map('trim', explode("\n", $whitelist))); | |
| 4854 | - | |
| 4855 | - if (empty($meta_keys)) { | |
| 4856 | - return array(); | |
| 4857 | - } | |
| 4858 | - | |
| 4859 | - $result = array(); | |
| 4860 | - | |
| 4861 | - foreach ($meta_keys as $key) { | |
| 4862 | - // Skip empty keys | |
| 4863 | - if (empty($key)) { | |
| 4864 | - continue; | |
| 4865 | - } | |
| 4866 | - | |
| 4867 | - $value = get_post_meta($post_id, $key, true); | |
| 4868 | - | |
| 4869 | - // Only include non-empty string values | |
| 4870 | - if (!empty($value) && is_string($value)) { | |
| 4871 | - $result[$key] = $value; | |
| 4872 | - } elseif (!empty($value) && is_array($value)) { | |
| 4873 | - // Handle array values by joining them | |
| 4874 | - $flat_value = $this->mxchat_flatten_meta_array($value); | |
| 4875 | - if (!empty($flat_value)) { | |
| 4876 | - $result[$key] = $flat_value; | |
| 4877 | - } | |
| 4878 | - } | |
| 4879 | - } | |
| 4880 | - | |
| 4881 | - return $result; | |
| 4882 | -} | |
| 4883 | - | |
| 4884 | -/** | |
| 4885 | - * Flatten array meta values into a readable string | |
| 4886 | - */ | |
| 4887 | -private function mxchat_flatten_meta_array($array, $depth = 0) { | |
| 4888 | - if ($depth > 3) { | |
| 4889 | - return ''; // Prevent infinite recursion | |
| 4890 | - } | |
| 4891 | - | |
| 4892 | - $parts = array(); | |
| 4893 | - | |
| 4894 | - foreach ($array as $key => $value) { | |
| 4895 | - if (is_string($value) && !empty($value)) { | |
| 4896 | - $parts[] = $value; | |
| 4897 | - } elseif (is_array($value)) { | |
| 4898 | - $nested = $this->mxchat_flatten_meta_array($value, $depth + 1); | |
| 4899 | - if (!empty($nested)) { | |
| 4900 | - $parts[] = $nested; | |
| 4901 | - } | |
| 4902 | - } | |
| 4903 | - } | |
| 4904 | - | |
| 4905 | - return implode(', ', $parts); | |
| 4906 | -} | |
| 4907 | - | |
| 4908 | -/** | |
| 4909 | - * Format ACF field values for content extraction | |
| 4910 | - */ | |
| 4911 | -public function mxchat_format_acf_field_value($value, $field_name = '', $post_id = 0) { | |
| 4912 | - if (empty($value)) { | |
| 4913 | - return ''; | |
| 4914 | - } | |
| 4915 | - | |
| 4916 | - // Handle WP_Post objects first (THIS IS THE KEY FIX) | |
| 4917 | - if ($value instanceof WP_Post) { | |
| 4918 | - return $value->post_title ?: ''; | |
| 4919 | - } | |
| 4920 | - | |
| 4921 | - // Handle other WP objects | |
| 4922 | - if (is_object($value)) { | |
| 4923 | - if (isset($value->post_title)) { | |
| 4924 | - return $value->post_title; | |
| 4925 | - } elseif (isset($value->display_name)) { | |
| 4926 | - return $value->display_name; | |
| 4927 | - } elseif (isset($value->name)) { | |
| 4928 | - return $value->name; | |
| 4929 | - } elseif (method_exists($value, '__toString')) { | |
| 4930 | - try { | |
| 4931 | - return (string) $value; | |
| 4932 | - } catch (Exception $e) { | |
| 4933 | - return ''; | |
| 4934 | - } | |
| 4935 | - } | |
| 4936 | - // For any other objects, return empty string | |
| 4937 | - return ''; | |
| 4938 | - } | |
| 4939 | - | |
| 4940 | - // Handle different ACF field types | |
| 4941 | - if (is_array($value)) { | |
| 4942 | - // Check if it's an image/file field | |
| 4943 | - if (isset($value['url'])) { | |
| 4944 | - // Image field - return alt text, title, or caption | |
| 4945 | - if (!empty($value['alt'])) { | |
| 4946 | - return $value['alt']; | |
| 4947 | - } elseif (!empty($value['title'])) { | |
| 4948 | - return $value['title']; | |
| 4949 | - } elseif (!empty($value['caption'])) { | |
| 4950 | - return $value['caption']; | |
| 4951 | - } else { | |
| 4952 | - return ''; // Don't include just the URL | |
| 4953 | - } | |
| 4954 | - } | |
| 4955 | - | |
| 4956 | - // Check if it's a post object or relationship field | |
| 4957 | - if (isset($value['post_title'])) { | |
| 4958 | - return $value['post_title']; | |
| 4959 | - } | |
| 4960 | - | |
| 4961 | - // Check if it's a user field | |
| 4962 | - if (isset($value['display_name'])) { | |
| 4963 | - return $value['display_name']; | |
| 4964 | - } | |
| 4965 | - | |
| 4966 | - // Check if it's a taxonomy term | |
| 4967 | - if (isset($value['name']) && isset($value['taxonomy'])) { | |
| 4968 | - return $value['name']; | |
| 4969 | - } | |
| 4970 | - | |
| 4971 | - // Check if it's a select field with label | |
| 4972 | - if (isset($value['label'])) { | |
| 4973 | - return $value['label']; | |
| 4974 | - } | |
| 4975 | - | |
| 4976 | - // Check for repeater field or flexible content | |
| 4977 | - if (is_numeric(key($value))) { | |
| 4978 | - $sub_values = array(); | |
| 4979 | - foreach ($value as $sub_item) { | |
| 4980 | - if (is_array($sub_item)) { | |
| 4981 | - // For repeater/flexible content, extract text values | |
| 4982 | - $sub_text = $this->mxchat_extract_text_from_acf_array($sub_item); | |
| 4983 | - if (!empty($sub_text)) { | |
| 4984 | - $sub_values[] = $sub_text; | |
| 4985 | - } | |
| 4986 | - } elseif ($sub_item instanceof WP_Post) { | |
| 4987 | - // Handle WP_Post objects in arrays | |
| 4988 | - $sub_values[] = $sub_item->post_title ?: ''; | |
| 4989 | - } else { | |
| 4990 | - $sub_values[] = (string) $sub_item; | |
| 4991 | - } | |
| 4992 | - } | |
| 4993 | - return implode(', ', array_filter($sub_values)); | |
| 4994 | - } | |
| 4995 | - | |
| 4996 | - // For other arrays, try to extract meaningful text | |
| 4997 | - $text_values = array(); | |
| 4998 | - foreach ($value as $key => $val) { | |
| 4999 | - if (is_string($val) && !empty(trim($val))) { | |
| 5000 | - $text_values[] = trim($val); | |
| 5001 | - } elseif ($val instanceof WP_Post) { | |
| 5002 | - // Handle WP_Post objects in associative arrays | |
| 5003 | - $text_values[] = $val->post_title ?: ''; | |
| 5004 | - } elseif (is_array($val) && isset($val['post_title'])) { | |
| 5005 | - $text_values[] = $val['post_title']; | |
| 5006 | - } elseif (is_array($val) && isset($val['name'])) { | |
| 5007 | - $text_values[] = $val['name']; | |
| 5008 | - } | |
| 5009 | - } | |
| 5010 | - | |
| 5011 | - return implode(', ', array_filter($text_values)); | |
| 5012 | - } | |
| 5013 | - | |
| 5014 | - // Handle boolean values | |
| 5015 | - if (is_bool($value)) { | |
| 5016 | - return $value ? 'Yes' : 'No'; | |
| 5017 | - } | |
| 5018 | - | |
| 5019 | - // Handle numeric values | |
| 5020 | - if (is_numeric($value)) { | |
| 5021 | - return (string) $value; | |
| 5022 | - } | |
| 5023 | - | |
| 5024 | - // Handle string values | |
| 5025 | - if (is_string($value)) { | |
| 5026 | - return trim($value); | |
| 5027 | - } | |
| 5028 | - | |
| 5029 | - // For anything else that we can't handle, return empty string | |
| 5030 | - // This prevents the "Object could not be converted to string" error | |
| 5031 | - return ''; | |
| 5032 | -} | |
| 5033 | - | |
| 5034 | -/** | |
| 5035 | - * Extract text from complex ACF array structures | |
| 5036 | - */ | |
| 5037 | -private function mxchat_extract_text_from_acf_array($array) { | |
| 5038 | - if (!is_array($array)) { | |
| 5039 | - return ''; | |
| 5040 | - } | |
| 5041 | - | |
| 5042 | - $text_parts = array(); | |
| 5043 | - | |
| 5044 | - foreach ($array as $key => $value) { | |
| 5045 | - if (is_string($value) && !empty(trim($value))) { | |
| 5046 | - // Skip keys that are likely to be IDs or technical values | |
| 5047 | - if (!is_numeric($value) || strlen($value) > 10) { | |
| 5048 | - $text_parts[] = trim($value); | |
| 5049 | - } | |
| 5050 | - } elseif ($value instanceof WP_Post) { | |
| 5051 | - // Handle WP_Post objects | |
| 5052 | - $text_parts[] = $value->post_title ?: ''; | |
| 5053 | - } elseif (is_array($value)) { | |
| 5054 | - if (isset($value['post_title'])) { | |
| 5055 | - $text_parts[] = $value['post_title']; | |
| 5056 | - } elseif (isset($value['name'])) { | |
| 5057 | - $text_parts[] = $value['name']; | |
| 5058 | - } elseif (isset($value['label'])) { | |
| 5059 | - $text_parts[] = $value['label']; | |
| 5060 | - } | |
| 5061 | - } elseif (is_object($value)) { | |
| 5062 | - // Handle other objects safely | |
| 5063 | - if (isset($value->post_title)) { | |
| 5064 | - $text_parts[] = $value->post_title; | |
| 5065 | - } elseif (isset($value->name)) { | |
| 5066 | - $text_parts[] = $value->name; | |
| 5067 | - } elseif (isset($value->display_name)) { | |
| 5068 | - $text_parts[] = $value->display_name; | |
| 5069 | - } | |
| 5070 | - } | |
| 5071 | - } | |
| 5072 | - | |
| 5073 | - return implode(', ', array_filter($text_parts)); | |
| 5074 | -} | |
| 5075 | - | |
| 5076 | -/** | |
| 5077 | - * Handle ACF save - fires after ACF fields are saved | |
| 5078 | - * This ensures ACF field data is available when syncing to knowledge base | |
| 5079 | - */ | |
| 5080 | -public function mxchat_handle_acf_save($post_id) { | |
| 5081 | - // Skip if not a valid post | |
| 5082 | - if (!$post_id || $post_id === 'options') { | |
| 5083 | - return; | |
| 5084 | - } | |
| 5085 | - | |
| 5086 | - // Skip autosaves and revisions | |
| 5087 | - if (defined('DOING_AUTOSAVE') && DOING_AUTOSAVE || wp_is_post_revision($post_id)) { | |
| 5088 | - return; | |
| 5089 | - } | |
| 5090 | - | |
| 5091 | - $post = get_post($post_id); | |
| 5092 | - if (!$post) { | |
| 5093 | - return; | |
| 5094 | - } | |
| 5095 | - | |
| 5096 | - $post_type = $post->post_type; | |
| 5097 | - | |
| 5098 | - // Check if sync is enabled for this post type | |
| 5099 | - $should_sync = false; | |
| 5100 | - | |
| 5101 | - if ($post_type === 'post' && get_option('mxchat_auto_sync_posts') === '1') { | |
| 5102 | - $should_sync = true; | |
| 5103 | - } else if ($post_type === 'page' && get_option('mxchat_auto_sync_pages') === '1') { | |
| 5104 | - $should_sync = true; | |
| 5105 | - } else if ($post_type === 'product' && class_exists('WooCommerce')) { | |
| 5106 | - // WooCommerce products - check if WooCommerce integration is enabled | |
| 5107 | - $options = get_option('mxchat_options', array()); | |
| 5108 | - if (isset($options['enable_woocommerce_integration']) && | |
| 5109 | - ($options['enable_woocommerce_integration'] === '1' || $options['enable_woocommerce_integration'] === 'on')) { | |
| 5110 | - $should_sync = true; | |
| 5111 | - } | |
| 5112 | - } else { | |
| 5113 | - // Check custom post types | |
| 5114 | - $option_name = 'mxchat_auto_sync_' . $post_type; | |
| 5115 | - if (get_option($option_name) === '1') { | |
| 5116 | - $should_sync = true; | |
| 5117 | - } | |
| 5118 | - } | |
| 5119 | - | |
| 5120 | - if (!$should_sync) { | |
| 5121 | - return; | |
| 5122 | - } | |
| 5123 | - | |
| 5124 | - // Only process published posts | |
| 5125 | - if ($post->post_status !== 'publish') { | |
| 5126 | - return; | |
| 5127 | - } | |
| 5128 | - | |
| 5129 | - // Check if this post has any ACF fields - if not, no need to re-sync | |
| 5130 | - $acf_fields = $this->mxchat_get_acf_fields_for_post($post_id); | |
| 5131 | - if (empty($acf_fields)) { | |
| 5132 | - return; | |
| 5133 | - } | |
| 5134 | - | |
| 5135 | - // Use a transient to prevent duplicate processing (post_updated may have already run) | |
| 5136 | - $transient_key = 'mxchat_acf_synced_' . $post_id; | |
| 5137 | - if (get_transient($transient_key)) { | |
| 5138 | - return; | |
| 5139 | - } | |
| 5140 | - set_transient($transient_key, true, 60); // Prevent re-processing for 60 seconds | |
| 5141 | - | |
| 5142 | - // Re-run the sync with ACF data now available | |
| 5143 | - // We pass $update=true since this is effectively an update with ACF data | |
| 5144 | - $this->mxchat_handle_post_update($post_id, $post, true); | |
| 5145 | -} | |
| 5146 | - | |
| 5147 | -public function mxchat_handle_post_update($post_id, $post, $update) { | |
| 5148 | - // Basic validation checks | |
| 5149 | - if (defined('DOING_AUTOSAVE') && DOING_AUTOSAVE || wp_is_post_revision($post_id)) { | |
| 5150 | - return; | |
| 5151 | - } | |
| 5152 | - | |
| 5153 | - $post_type = $post->post_type; | |
| 5154 | - | |
| 5155 | - // Check if sync is enabled for this post type | |
| 5156 | - $should_sync = false; | |
| 5157 | - | |
| 5158 | - // Check built-in post types first | |
| 5159 | - if ($post_type === 'post' && get_option('mxchat_auto_sync_posts') === '1') { | |
| 5160 | - $should_sync = true; | |
| 5161 | - } else if ($post_type === 'page' && get_option('mxchat_auto_sync_pages') === '1') { | |
| 5162 | - $should_sync = true; | |
| 5163 | - } else { | |
| 5164 | - // Check custom post types | |
| 5165 | - $option_name = 'mxchat_auto_sync_' . $post_type; | |
| 5166 | - if (get_option($option_name) === '1') { | |
| 5167 | - $should_sync = true; | |
| 5168 | - } | |
| 5169 | - } | |
| 5170 | - | |
| 5171 | - if (!$should_sync) { | |
| 5172 | - return; | |
| 5173 | - } | |
| 5174 | - | |
| 5175 | - // Check if we have stored the previous status and URL in our transients | |
| 5176 | - $previous_status_key = 'mxchat_prev_status_' . $post_id; | |
| 5177 | - $previous_status = get_transient($previous_status_key); | |
| 5178 | - | |
| 5179 | - $previous_url_key = 'mxchat_prev_url_' . $post_id; | |
| 5180 | - $previous_url = get_transient($previous_url_key); | |
| 5181 | - | |
| 5182 | - // If the post was previously published but is now not published, remove from knowledge base | |
| 5183 | - if ($previous_status === 'publish' && $post->post_status !== 'publish') { | |
| 5184 | - // Use the stored URL from when it was published, or fall back to current permalink | |
| 5185 | - $source_url = $previous_url ?: get_permalink($post_id); | |
| 5186 | - | |
| 5187 | - if ($source_url) { | |
| 5188 | - // Chunk-aware deletion (routes to Pinecone or WP DB and removes base + all chunks) | |
| 5189 | - MxChat_Utils::delete_chunks_for_url($source_url, 'default'); | |
| 5190 | - } | |
| 5191 | - | |
| 5192 | - // Clean up the transients and exit early | |
| 5193 | - delete_transient($previous_status_key); | |
| 5194 | - delete_transient($previous_url_key); | |
| 5195 | - return; | |
| 5196 | - } | |
| 5197 | - | |
| 5198 | - // Slug/permalink rename while still published: delete the old vectors before upserting new ones. | |
| 5199 | - // Without this, md5(old_url) vectors (base + chunks) would be orphaned under the stale URL. | |
| 5200 | - if ($post->post_status === 'publish' && !empty($previous_url)) { | |
| 5201 | - $current_url = get_permalink($post_id); | |
| 5202 | - if ($current_url && $current_url !== $previous_url) { | |
| 5203 | - MxChat_Utils::delete_chunks_for_url($previous_url, 'default'); | |
| 5204 | - } | |
| 5205 | - } | |
| 5206 | - | |
| 5207 | - // Store the current status for next time (if this is an update) | |
| 5208 | - if ($update) { | |
| 5209 | - set_transient($previous_status_key, $post->post_status, DAY_IN_SECONDS); | |
| 5210 | - | |
| 5211 | - // If the post is currently published, also store its URL | |
| 5212 | - if ($post->post_status === 'publish') { | |
| 5213 | - $current_url = get_permalink($post_id); | |
| 5214 | - set_transient($previous_url_key, $current_url, DAY_IN_SECONDS); | |
| 5215 | - } | |
| 5216 | - } | |
| 5217 | - | |
| 5218 | - // Only process currently published content for adding/updating | |
| 5219 | - if ($post->post_status === 'publish') { | |
| 5220 | - // Get the source URL | |
| 5221 | - $source_url = get_permalink($post_id); | |
| 5222 | - | |
| 5223 | - // Get content with proper formatting (matching ajax_mxchat_process_selected_content) | |
| 5224 | - $title = get_the_title($post_id); | |
| 5225 | - $content = get_post_field('post_content', $post_id); | |
| 5226 | - $excerpt = get_post_field('post_excerpt', $post_id); | |
| 5227 | - | |
| 5228 | - // Remove shortcode tags but preserve content inside them | |
| 5229 | - $content = $this->strip_shortcode_tags_preserve_content($content); | |
| 5230 | - $excerpt = $this->strip_shortcode_tags_preserve_content($excerpt); | |
| 5231 | - | |
| 5232 | - // Strip tags but preserve structure (don't use 'the_content' filter as it may re-add shortcodes) | |
| 5233 | - $content = wp_strip_all_tags($content); | |
| 5234 | - | |
| 5235 | - // Combine title, short description (if exists), and content | |
| 5236 | - $final_content = $title . "\n\n"; | |
| 5237 | - | |
| 5238 | - // Add short description if it exists (WooCommerce products use post_excerpt for short description) | |
| 5239 | - if (!empty($excerpt)) { | |
| 5240 | - $final_content .= "Short Description: " . wp_strip_all_tags($excerpt) . "\n\n"; | |
| 5241 | - } | |
| 5242 | - | |
| 5243 | - $final_content .= $content; | |
| 5244 | - | |
| 5245 | - // For WooCommerce products, include pricing and product details | |
| 5246 | - if ($post_type === 'product' && class_exists('WooCommerce')) { | |
| 5247 | - $product = wc_get_product($post_id); | |
| 5248 | - | |
| 5249 | - if ($product) { | |
| 5250 | - // Get pricing information | |
| 5251 | - $regular_price = $product->get_regular_price(); | |
| 5252 | - $sale_price = $product->get_sale_price(); | |
| 5253 | - $price = $product->get_price(); | |
| 5254 | - $sku = $product->get_sku(); | |
| 5255 | - | |
| 5256 | - // Get currency symbol | |
| 5257 | - $currency_symbol = get_woocommerce_currency_symbol(); | |
| 5258 | - | |
| 5259 | - // Add pricing information | |
| 5260 | - $final_content .= "\n"; | |
| 5261 | - if (!empty($regular_price)) { | |
| 5262 | - $final_content .= "Price: " . $currency_symbol . $regular_price . "\n"; | |
| 5263 | - } elseif (!empty($price)) { | |
| 5264 | - $final_content .= "Price: " . $currency_symbol . $price . "\n"; | |
| 5265 | - } | |
| 5266 | - | |
| 5267 | - if (!empty($sale_price) && $sale_price !== $regular_price) { | |
| 5268 | - $final_content .= "Sale Price: " . $currency_symbol . $sale_price . "\n"; | |
| 5269 | - } | |
| 5270 | - | |
| 5271 | - // Handle variable products - show price range | |
| 5272 | - if ($product->is_type('variable')) { | |
| 5273 | - $min_price = $product->get_variation_price('min'); | |
| 5274 | - $max_price = $product->get_variation_price('max'); | |
| 5275 | - if ($min_price !== $max_price) { | |
| 5276 | - $final_content .= "Price Range: " . $currency_symbol . $min_price . " - " . $currency_symbol . $max_price . "\n"; | |
| 5277 | - } | |
| 5278 | - } | |
| 5279 | - | |
| 5280 | - if (!empty($sku)) { | |
| 5281 | - $final_content .= "SKU: " . $sku . "\n"; | |
| 5282 | - } | |
| 5283 | - | |
| 5284 | - // Get product categories | |
| 5285 | - $categories = wp_get_post_terms($post_id, 'product_cat', array('fields' => 'names')); | |
| 5286 | - if (!empty($categories) && !is_wp_error($categories)) { | |
| 5287 | - $final_content .= "Categories: " . implode(', ', $categories) . "\n"; | |
| 5288 | - } | |
| 5289 | - } | |
| 5290 | - } | |
| 5291 | - | |
| 5292 | - // For custom post types like job_listing, include additional fields | |
| 5293 | - if ($post_type === 'job_listing') { | |
| 5294 | - // Add job-specific meta if available | |
| 5295 | - $job_location = get_post_meta($post_id, '_job_location', true); | |
| 5296 | - if (!empty($job_location)) { | |
| 5297 | - $final_content .= "\n\nLocation: " . $job_location; | |
| 5298 | - } | |
| 5299 | - | |
| 5300 | - // Get job type terms | |
| 5301 | - $job_types = get_the_terms($post_id, 'job_listing_type'); | |
| 5302 | - if (!empty($job_types) && !is_wp_error($job_types)) { | |
| 5303 | - $types = array(); | |
| 5304 | - foreach ($job_types as $type) { | |
| 5305 | - $types[] = $type->name; | |
| 5306 | - } | |
| 5307 | - $final_content .= "\n\nJob Type: " . implode(', ', $types); | |
| 5308 | - } | |
| 5309 | - | |
| 5310 | - // Get company name if available | |
| 5311 | - $company_name = get_post_meta($post_id, '_company_name', true); | |
| 5312 | - if (!empty($company_name)) { | |
| 5313 | - $final_content .= "\n\nCompany: " . $company_name; | |
| 5314 | - } | |
| 5315 | - } | |
| 5316 | - | |
| 5317 | - // ADD ACF FIELDS SUPPORT (matches ajax_mxchat_process_selected_content behavior) | |
| 5318 | - $acf_fields = $this->mxchat_get_acf_fields_for_post($post_id); | |
| 5319 | - if (!empty($acf_fields)) { | |
| 5320 | - $acf_content_parts = array(); | |
| 5321 | - | |
| 5322 | - foreach ($acf_fields as $field_name => $field_value) { | |
| 5323 | - $formatted_value = $this->mxchat_format_acf_field_value($field_value, $field_name, $post_id); | |
| 5324 | - if (!empty($formatted_value)) { | |
| 5325 | - // Convert field name to readable label | |
| 5326 | - $field_label = ucwords(str_replace(['_', '-'], ' ', $field_name)); | |
| 5327 | - $acf_content_parts[] = $field_label . ": " . $formatted_value; | |
| 5328 | - } | |
| 5329 | - } | |
| 5330 | - | |
| 5331 | - if (!empty($acf_content_parts)) { | |
| 5332 | - $final_content .= "\n\n" . implode("\n", $acf_content_parts); | |
| 5333 | - } | |
| 5334 | - } | |
| 5335 | - | |
| 5336 | - // ADD CUSTOM POST META SUPPORT (whitelisted non-ACF meta fields) | |
| 5337 | - $custom_meta = $this->mxchat_get_whitelisted_post_meta($post_id); | |
| 5338 | - if (!empty($custom_meta)) { | |
| 5339 | - $meta_content_parts = array(); | |
| 5340 | - | |
| 5341 | - foreach ($custom_meta as $meta_key => $meta_value) { | |
| 5342 | - // Convert meta key to readable label | |
| 5343 | - $meta_label = ucwords(str_replace(array('_', '-'), ' ', $meta_key)); | |
| 5344 | - $meta_content_parts[] = $meta_label . ": " . $meta_value; | |
| 5345 | - } | |
| 5346 | - | |
| 5347 | - if (!empty($meta_content_parts)) { | |
| 5348 | - $final_content .= "\n\n" . implode("\n", $meta_content_parts); | |
| 5349 | - } | |
| 5350 | - } | |
| 5351 | - | |
| 5352 | - // Get API key with proper model detection | |
| 5353 | - $options = get_option('mxchat_options'); | |
| 5354 | - $selected_model = $options['embedding_model'] ?? 'text-embedding-ada-002'; | |
| 5355 | - | |
| 5356 | - if (strpos($selected_model, 'voyage') === 0) { | |
| 5357 | - $api_key = $options['voyage_api_key'] ?? ''; | |
| 5358 | - } elseif (strpos($selected_model, 'gemini-embedding') === 0) { | |
| 5359 | - $api_key = $options['gemini_api_key'] ?? ''; | |
| 5360 | - } else { | |
| 5361 | - $api_key = $options['api_key'] ?? ''; | |
| 5362 | - } | |
| 5363 | - | |
| 5364 | - if (empty($api_key)) { | |
| 5365 | - return; | |
| 5366 | - } | |
| 5367 | - | |
| 5368 | - // Use the centralized utility function for storage | |
| 5369 | - $result = MxChat_Utils::submit_content_to_db( | |
| 5370 | - $final_content, | |
| 5371 | - $source_url, | |
| 5372 | - $api_key, | |
| 5373 | - md5($source_url) // Vector ID for Pinecone | |
| 5374 | - ); | |
| 5375 | - | |
| 5376 | - // After successful storage, apply role restriction based on tags | |
| 5377 | - if (!is_wp_error($result)) { | |
| 5378 | - $this->apply_role_restriction_to_post($post_id, $source_url); | |
| 5379 | - } | |
| 5380 | - } | |
| 5381 | - | |
| 5382 | - // Clean up the stored previous status if not used above | |
| 5383 | - if ($previous_status !== 'publish' || $post->post_status === 'publish') { | |
| 5384 | - delete_transient($previous_status_key); | |
| 5385 | - delete_transient($previous_url_key); | |
| 5386 | - } | |
| 5387 | -} | |
| 5388 | - | |
| 5389 | -/** | |
| 5390 | - * Store the post status and URL before update to detect status transitions | |
| 5391 | - * This runs before the post is actually updated in the database | |
| 5392 | - */ | |
| 5393 | -public function mxchat_store_pre_update_status($post_id, $data) { | |
| 5394 | - // Get the current post from database (before update) | |
| 5395 | - $current_post = get_post($post_id); | |
| 5396 | - | |
| 5397 | - if ($current_post) { | |
| 5398 | - // Store the current status temporarily | |
| 5399 | - $status_key = 'mxchat_prev_status_' . $post_id; | |
| 5400 | - set_transient($status_key, $current_post->post_status, HOUR_IN_SECONDS); | |
| 5401 | - | |
| 5402 | - // If the post is currently published, also store its URL | |
| 5403 | - if ($current_post->post_status === 'publish') { | |
| 5404 | - $url_key = 'mxchat_prev_url_' . $post_id; | |
| 5405 | - $current_url = get_permalink($post_id); | |
| 5406 | - set_transient($url_key, $current_url, HOUR_IN_SECONDS); | |
| 5407 | - } | |
| 5408 | - } | |
| 5409 | -} | |
| 5410 | - | |
| 5411 | -public function mxchat_handle_post_delete($post_id) { | |
| 5412 | - // Get post data before it's deleted | |
| 5413 | - $post = get_post($post_id); | |
| 5414 | - | |
| 5415 | - // Basic validation | |
| 5416 | - if (!$post || wp_is_post_revision($post_id)) { | |
| 5417 | - return; | |
| 5418 | - } | |
| 5419 | - | |
| 5420 | - $post_type = $post->post_type; | |
| 5421 | - | |
| 5422 | - // Check if sync is enabled for this post type | |
| 5423 | - $should_sync = false; | |
| 5424 | - | |
| 5425 | - // Check built-in post types first | |
| 5426 | - if ($post_type === 'post' && get_option('mxchat_auto_sync_posts') === '1') { | |
| 5427 | - $should_sync = true; | |
| 5428 | - } else if ($post_type === 'page' && get_option('mxchat_auto_sync_pages') === '1') { | |
| 5429 | - $should_sync = true; | |
| 5430 | - } else { | |
| 5431 | - // Check custom post types | |
| 5432 | - $option_name = 'mxchat_auto_sync_' . $post_type; | |
| 5433 | - if (get_option($option_name) === '1') { | |
| 5434 | - $should_sync = true; | |
| 5435 | - } | |
| 5436 | - } | |
| 5437 | - | |
| 5438 | - if (!$should_sync) { | |
| 5439 | - return; | |
| 5440 | - } | |
| 5441 | - | |
| 5442 | - // Resolve the pre-trash URL. wp_trash_post renames the slug with "__trashed" before firing | |
| 5443 | - // this hook, so get_permalink() here would return the trashed URL and md5() would miss the | |
| 5444 | - // real vector IDs stored under the original URL. | |
| 5445 | - $source_url = $this->mxchat_resolve_pre_trash_url($post_id); | |
| 5446 | - if (!$source_url) { | |
| 5447 | - //error_log('MXChat: Failed to resolve source URL for post ' . $post_id); | |
| 5448 | - return; | |
| 5449 | - } | |
| 5450 | - | |
| 5451 | - // Use chunk-aware deletion (handles both chunked and non-chunked content) | |
| 5452 | - $delete_result = MxChat_Utils::delete_chunks_for_url($source_url, 'default'); | |
| 5453 | - | |
| 5454 | - if (is_wp_error($delete_result)) { | |
| 5455 | - //error_log('MXChat: Chunk-aware deletion failed for URL: ' . $source_url . ' - ' . $delete_result->get_error_message()); | |
| 5456 | - } | |
| 5457 | - | |
| 5458 | - delete_transient('mxchat_prev_url_' . $post_id); | |
| 5459 | - delete_transient('mxchat_prev_status_' . $post_id); | |
| 5460 | -} | |
| 5461 | - | |
| 5462 | -/** | |
| 5463 | - * Resolve the source URL for a post being trashed/deleted. | |
| 5464 | - * | |
| 5465 | - * Why: wp_trash_post appends "__trashed" to the slug before the wp_trash_post action fires, so | |
| 5466 | - * get_permalink() returns a URL whose md5() won't match the vector IDs stored in Pinecone or | |
| 5467 | - * the source_url rows in the WP DB. Prefer the URL captured by mxchat_store_pre_update_status | |
| 5468 | - * (runs on pre_post_update, before the rename); fall back to stripping the __trashed suffix. | |
| 5469 | - */ | |
| 5470 | -private function mxchat_resolve_pre_trash_url($post_id) { | |
| 5471 | - $previous_url = get_transient('mxchat_prev_url_' . $post_id); | |
| 5472 | - if (!empty($previous_url)) { | |
| 5473 | - return $previous_url; | |
| 5474 | - } | |
| 5475 | - | |
| 5476 | - $current = get_permalink($post_id); | |
| 5477 | - if (!$current) { | |
| 5478 | - return ''; | |
| 5479 | - } | |
| 5480 | - return preg_replace('#__trashed(/?)$#', '$1', $current); | |
| 5481 | -} | |
| 5482 | - | |
| 5483 | - | |
| 5484 | - | |
| 5485 | -public function mxchat_handle_product_change($post_id, $post, $update) { | |
| 5486 | - if ($post->post_type !== 'product') { | |
| 5487 | - return; | |
| 5488 | - } | |
| 5489 | - | |
| 5490 | - if ($post->post_status === 'publish') { | |
| 5491 | - add_action('shutdown', function() use ($post_id) { | |
| 5492 | - $product = wc_get_product($post_id); | |
| 5493 | - if ($product) { | |
| 5494 | - $this->mxchat_store_product_embedding($product); | |
| 5495 | - } | |
| 5496 | - }); | |
| 5497 | - } | |
| 5498 | -} | |
| 5499 | - | |
| 5500 | -/** | |
| 5501 | - * Store WooCommerce product embeddings | |
| 5502 | - */ | |
| 5503 | -private function mxchat_store_product_embedding($product) { | |
| 5504 | - if (!isset($this->options['enable_woocommerce_integration']) || | |
| 5505 | - !in_array($this->options['enable_woocommerce_integration'], ['1', 'on'])) { | |
| 5506 | - return; | |
| 5507 | - } | |
| 5508 | - | |
| 5509 | - $source_url = get_permalink($product->get_id()); | |
| 5510 | - $product_id = $product->get_id(); | |
| 5511 | - | |
| 5512 | - // Build product content | |
| 5513 | - $title = $product->get_name(); | |
| 5514 | - $description = $product->get_description(); | |
| 5515 | - $short_description = $product->get_short_description(); | |
| 5516 | - $regular_price = $product->get_regular_price(); | |
| 5517 | - $sale_price = $product->get_sale_price(); | |
| 5518 | - $price = $product->get_price(); | |
| 5519 | - $sku = $product->get_sku(); | |
| 5520 | - | |
| 5521 | - // Get currency symbol | |
| 5522 | - $currency_symbol = get_woocommerce_currency_symbol(); | |
| 5523 | - | |
| 5524 | - // Format content consistently | |
| 5525 | - $content = $title . "\n\n"; | |
| 5526 | - | |
| 5527 | - if (!empty($short_description)) { | |
| 5528 | - $content .= "Short Description: " . wp_strip_all_tags($short_description) . "\n\n"; | |
| 5529 | - } | |
| 5530 | - | |
| 5531 | - if (!empty($description)) { | |
| 5532 | - $content .= wp_strip_all_tags($description) . "\n\n"; | |
| 5533 | - } | |
| 5534 | - | |
| 5535 | - // Add pricing information | |
| 5536 | - if (!empty($regular_price)) { | |
| 5537 | - $content .= "Price: " . $currency_symbol . $regular_price . "\n"; | |
| 5538 | - } elseif (!empty($price)) { | |
| 5539 | - $content .= "Price: " . $currency_symbol . $price . "\n"; | |
| 5540 | - } | |
| 5541 | - | |
| 5542 | - if (!empty($sale_price) && $sale_price !== $regular_price) { | |
| 5543 | - $content .= "Sale Price: " . $currency_symbol . $sale_price . "\n"; | |
| 5544 | - } | |
| 5545 | - | |
| 5546 | - // Handle variable products - show price range | |
| 5547 | - if ($product->is_type('variable')) { | |
| 5548 | - $min_price = $product->get_variation_price('min'); | |
| 5549 | - $max_price = $product->get_variation_price('max'); | |
| 5550 | - if ($min_price !== $max_price) { | |
| 5551 | - $content .= "Price Range: " . $currency_symbol . $min_price . " - " . $currency_symbol . $max_price . "\n"; | |
| 5552 | - } | |
| 5553 | - } | |
| 5554 | - | |
| 5555 | - if (!empty($sku)) { | |
| 5556 | - $content .= "SKU: " . $sku . "\n"; | |
| 5557 | - } | |
| 5558 | - | |
| 5559 | - // Get product categories | |
| 5560 | - $categories = wp_get_post_terms($product_id, 'product_cat', array('fields' => 'names')); | |
| 5561 | - if (!empty($categories) && !is_wp_error($categories)) { | |
| 5562 | - $content .= "Categories: " . implode(', ', $categories) . "\n"; | |
| 5563 | - } | |
| 5564 | - | |
| 5565 | - // Get Custom Product Tabs (supports "Custom Product Tabs for WooCommerce" by Code Parrots) | |
| 5566 | - $custom_tabs = get_post_meta($product_id, 'yikes_woo_products_tabs', true); | |
| 5567 | - if (!empty($custom_tabs) && is_array($custom_tabs)) { | |
| 5568 | - foreach ($custom_tabs as $tab) { | |
| 5569 | - $tab_title = isset($tab['title']) ? $tab['title'] : (isset($tab['tab_title']) ? $tab['tab_title'] : ''); | |
| 5570 | - $tab_content = isset($tab['content']) ? $tab['content'] : ''; | |
| 5571 | - | |
| 5572 | - if (!empty($tab_title) && !empty($tab_content)) { | |
| 5573 | - $content .= "\n" . $tab_title . ": " . wp_strip_all_tags($tab_content) . "\n"; | |
| 5574 | - } | |
| 5575 | - } | |
| 5576 | - } | |
| 5577 | - | |
| 5578 | - // Also check for reusable/saved tabs applied to this product | |
| 5579 | - $applied_saved_tabs = get_post_meta($product_id, 'yikes_woo_reusable_products_tabs_applied', true); | |
| 5580 | - if (!empty($applied_saved_tabs) && is_array($applied_saved_tabs)) { | |
| 5581 | - // Get the saved tabs option | |
| 5582 | - $saved_tabs = get_option('yikes_woo_reusable_products_tabs', array()); | |
| 5583 | - if (!empty($saved_tabs) && is_array($saved_tabs)) { | |
| 5584 | - foreach ($applied_saved_tabs as $saved_tab_id) { | |
| 5585 | - if (isset($saved_tabs[$saved_tab_id])) { | |
| 5586 | - $tab = $saved_tabs[$saved_tab_id]; | |
| 5587 | - $tab_title = isset($tab['title']) ? $tab['title'] : (isset($tab['tab_title']) ? $tab['tab_title'] : ''); | |
| 5588 | - $tab_content = isset($tab['content']) ? $tab['content'] : ''; | |
| 5589 | - | |
| 5590 | - if (!empty($tab_title) && !empty($tab_content)) { | |
| 5591 | - $content .= "\n" . $tab_title . ": " . wp_strip_all_tags($tab_content) . "\n"; | |
| 5592 | - } | |
| 5593 | - } | |
| 5594 | - } | |
| 5595 | - } | |
| 5596 | - } | |
| 5597 | - | |
| 5598 | - // Get API key with proper model detection | |
| 5599 | - $options = get_option('mxchat_options'); | |
| 5600 | - $selected_model = $options['embedding_model'] ?? 'text-embedding-ada-002'; | |
| 5601 | - | |
| 5602 | - if (strpos($selected_model, 'voyage') === 0) { | |
| 5603 | - $api_key = $options['voyage_api_key'] ?? ''; | |
| 5604 | - } elseif (strpos($selected_model, 'gemini-embedding') === 0) { | |
| 5605 | - $api_key = $options['gemini_api_key'] ?? ''; | |
| 5606 | - } else { | |
| 5607 | - $api_key = $options['api_key'] ?? ''; | |
| 5608 | - } | |
| 5609 | - | |
| 5610 | - if (empty($api_key)) { | |
| 5611 | - //error_log('MxChat Auto-sync: No API key configured for embedding model'); | |
| 5612 | - return; | |
| 5613 | - } | |
| 5614 | - | |
| 5615 | - // Use the centralized utility function for storage | |
| 5616 | - $result = MxChat_Utils::submit_content_to_db( | |
| 5617 | - $content, | |
| 5618 | - $source_url, | |
| 5619 | - $api_key, | |
| 5620 | - md5($source_url) // Vector ID for Pinecone | |
| 5621 | - ); | |
| 5622 | - | |
| 5623 | - // After successful storage, apply role restriction based on tags | |
| 5624 | - if (!is_wp_error($result)) { | |
| 5625 | - $this->apply_role_restriction_to_post($product_id, $source_url); | |
| 5626 | - } | |
| 5627 | - | |
| 5628 | - if (is_wp_error($result)) { | |
| 5629 | - //error_log('MxChat WooCommerce sync failed for product ' . $product_id . ': ' . $result->get_error_message()); | |
| 5630 | - } | |
| 5631 | -} | |
| 5632 | - | |
| 5633 | -public function mxchat_handle_product_delete($post_id) { | |
| 5634 | - if (get_post_type($post_id) !== 'product') { | |
| 5635 | - return; | |
| 5636 | - } | |
| 5637 | - | |
| 5638 | - $source_url = $this->mxchat_resolve_pre_trash_url($post_id); | |
| 5639 | - if (!$source_url) { | |
| 5640 | - return; | |
| 5641 | - } | |
| 5642 | - | |
| 5643 | - // Chunk-aware deletion (routes to Pinecone or WP DB and removes base + all chunks) | |
| 5644 | - MxChat_Utils::delete_chunks_for_url($source_url, 'default'); | |
| 5645 | - | |
| 5646 | - delete_transient('mxchat_prev_url_' . $post_id); | |
| 5647 | - delete_transient('mxchat_prev_status_' . $post_id); | |
| 5648 | -} | |
| 5649 | - | |
| 5650 | -/** | |
| 5651 | - * Handle individual Pinecone content deletion | |
| 5652 | - */ | |
| 5653 | -public function mxchat_handle_pinecone_prompt_delete() { | |
| 5654 | - // Check permissions | |
| 5655 | - if (!current_user_can('manage_options')) { | |
| 5656 | - wp_die(esc_html__('You do not have sufficient permissions.', 'mxchat')); | |
| 5657 | - } | |
| 5658 | - | |
| 5659 | - // Verify nonce | |
| 5660 | - if (!isset($_GET['_wpnonce']) || !wp_verify_nonce($_GET['_wpnonce'], 'mxchat_delete_pinecone_prompt_nonce')) { | |
| 5661 | - wp_die(esc_html__('Security check failed.', 'mxchat')); | |
| 5662 | - } | |
| 5663 | - | |
| 5664 | - $vector_id = isset($_GET['vector_id']) ? sanitize_text_field($_GET['vector_id']) : ''; | |
| 5665 | - | |
| 5666 | - if (empty($vector_id)) { | |
| 5667 | - set_transient('mxchat_admin_notice_error', | |
| 5668 | - esc_html__('Invalid vector ID.', 'mxchat'), | |
| 5669 | - 30 | |
| 5670 | - ); | |
| 5671 | - wp_safe_redirect(admin_url('admin.php?page=mxchat-prompts')); | |
| 5672 | - exit; | |
| 5673 | - } | |
| 5674 | - | |
| 5675 | - // Get Pinecone settings | |
| 5676 | - $pinecone_options = get_option('mxchat_pinecone_addon_options', array()); | |
| 5677 | - $use_pinecone = ($pinecone_options['mxchat_use_pinecone'] ?? '0') === '1'; | |
| 5678 | - | |
| 5679 | - if (!$use_pinecone || empty($pinecone_options['mxchat_pinecone_api_key'])) { | |
| 5680 | - set_transient('mxchat_admin_notice_error', | |
| 5681 | - esc_html__('Pinecone is not properly configured.', 'mxchat'), | |
| 5682 | - 30 | |
| 5683 | - ); | |
| 5684 | - wp_safe_redirect(admin_url('admin.php?page=mxchat-prompts')); | |
| 5685 | - exit; | |
| 5686 | - } | |
| 5687 | - | |
| 5688 | - // Delete from Pinecone | |
| 5689 | - $pinecone_manager = MxChat_Pinecone_Manager::get_instance(); | |
| 5690 | - $result = $pinecone_manager->mxchat_delete_from_pinecone_by_vector_id( | |
| 5691 | - $vector_id, | |
| 5692 | - $pinecone_options['mxchat_pinecone_api_key'], | |
| 5693 | - $pinecone_options['mxchat_pinecone_host'] | |
| 5694 | - ); | |
| 5695 | - | |
| 5696 | - if ($result['success']) { | |
| 5697 | - // No cache clearing needed since we removed caching | |
| 5698 | - set_transient('mxchat_admin_notice_success', | |
| 5699 | - esc_html__('Entry deleted successfully from Pinecone.', 'mxchat'), | |
| 5700 | - 30 | |
| 5701 | - ); | |
| 5702 | - } else { | |
| 5703 | - set_transient('mxchat_admin_notice_error', | |
| 5704 | - esc_html__('Failed to delete entry: ', 'mxchat') . $result['message'], | |
| 5705 | - 30 | |
| 5706 | - ); | |
| 5707 | - } | |
| 5708 | - | |
| 5709 | - wp_safe_redirect(admin_url('admin.php?page=mxchat-prompts')); | |
| 5710 | - exit; | |
| 5711 | -} | |
| 5712 | -/** | |
| 5713 | - * Handle individual Pinecone content deletion via AJAX | |
| 5714 | - */ | |
| 5715 | -public function ajax_mxchat_delete_pinecone_prompt() { | |
| 5716 | - // Verify nonce and permissions | |
| 5717 | - if (!check_ajax_referer('mxchat_delete_pinecone_prompt_nonce', 'nonce', false)) { | |
| 5718 | - wp_send_json_error('Invalid nonce'); | |
| 5719 | - exit; | |
| 5720 | - } | |
| 5721 | - | |
| 5722 | - if (!current_user_can('manage_options')) { | |
| 5723 | - wp_send_json_error('Unauthorized access'); | |
| 5724 | - exit; | |
| 5725 | - } | |
| 5726 | - | |
| 5727 | - $vector_id = isset($_POST['vector_id']) ? sanitize_text_field($_POST['vector_id']) : ''; | |
| 5728 | - $bot_id = isset($_POST['bot_id']) ? sanitize_text_field($_POST['bot_id']) : 'default'; | |
| 5729 | - | |
| 5730 | - if (empty($vector_id)) { | |
| 5731 | - wp_send_json_error('Missing vector ID'); | |
| 5732 | - exit; | |
| 5733 | - } | |
| 5734 | - | |
| 5735 | - // Get bot-specific Pinecone settings | |
| 5736 | - $pinecone_manager = MxChat_Pinecone_Manager::get_instance(); | |
| 5737 | - $pinecone_options = $pinecone_manager->mxchat_get_bot_pinecone_options($bot_id); | |
| 5738 | - | |
| 5739 | - $use_pinecone = ($pinecone_options['mxchat_use_pinecone'] ?? '0') === '1'; | |
| 5740 | - | |
| 5741 | - if (!$use_pinecone || empty($pinecone_options['mxchat_pinecone_api_key'])) { | |
| 5742 | - wp_send_json_error('Pinecone is not properly configured for bot: ' . $bot_id); | |
| 5743 | - exit; | |
| 5744 | - } | |
| 5745 | - | |
| 5746 | - // Delete from the correct Pinecone index | |
| 5747 | - $result = $pinecone_manager->mxchat_delete_from_pinecone_by_vector_id( | |
| 5748 | - $vector_id, | |
| 5749 | - $pinecone_options['mxchat_pinecone_api_key'], | |
| 5750 | - $pinecone_options['mxchat_pinecone_host'] | |
| 5751 | - ); | |
| 5752 | - | |
| 5753 | - if ($result['success']) { | |
| 5754 | - // No cache clearing needed since we removed caching | |
| 5755 | - wp_send_json_success(array( | |
| 5756 | - 'message' => 'Entry deleted successfully from Pinecone', | |
| 5757 | - 'vector_id' => $vector_id, | |
| 5758 | - 'bot_id' => $bot_id | |
| 5759 | - )); | |
| 5760 | - } else { | |
| 5761 | - MxChat_Admin::mxchat_log_debug('pinecone_error', 'Failed to delete from Pinecone: ' . $result['message'], array('vector_id' => $vector_id, 'bot_id' => $bot_id)); | |
| 5762 | - wp_send_json_error('Failed to delete from Pinecone: ' . $result['message']); | |
| 5763 | - } | |
| 5764 | - | |
| 5765 | - exit; | |
| 5766 | -} | |
| 5767 | - | |
| 5768 | -/** | |
| 5769 | - * Handle deletion of all chunks for a given source URL via AJAX | |
| 5770 | - * Follows the same pattern as ajax_mxchat_delete_pinecone_prompt | |
| 5771 | - */ | |
| 5772 | -public function ajax_mxchat_delete_chunks_by_url() { | |
| 5773 | - // Verify nonce and permissions | |
| 5774 | - if (!check_ajax_referer('mxchat_delete_chunks_nonce', 'nonce', false)) { | |
| 5775 | - wp_send_json_error('Invalid nonce'); | |
| 5776 | - exit; | |
| 5777 | - } | |
| 5778 | - | |
| 5779 | - if (!current_user_can('manage_options')) { | |
| 5780 | - wp_send_json_error('Unauthorized access'); | |
| 5781 | - exit; | |
| 5782 | - } | |
| 5783 | - | |
| 5784 | - $source_url = isset($_POST['source_url']) ? esc_url_raw($_POST['source_url']) : ''; | |
| 5785 | - $data_source = isset($_POST['data_source']) ? sanitize_text_field($_POST['data_source']) : 'wordpress'; | |
| 5786 | - $bot_id = isset($_POST['bot_id']) ? sanitize_text_field($_POST['bot_id']) : 'default'; | |
| 5787 | - | |
| 5788 | - if (empty($source_url)) { | |
| 5789 | - wp_send_json_error('Missing source URL'); | |
| 5790 | - exit; | |
| 5791 | - } | |
| 5792 | - | |
| 5793 | - // Generate the base vector ID from the source URL (same as how chunks are created) | |
| 5794 | - $base_vector_id = md5($source_url); | |
| 5795 | - | |
| 5796 | - if ($data_source === 'pinecone') { | |
| 5797 | - // Get bot-specific Pinecone settings (same as working delete function) | |
| 5798 | - $pinecone_manager = MxChat_Pinecone_Manager::get_instance(); | |
| 5799 | - $pinecone_options = $pinecone_manager->mxchat_get_bot_pinecone_options($bot_id); | |
| 5800 | - | |
| 5801 | - $use_pinecone = ($pinecone_options['mxchat_use_pinecone'] ?? '0') === '1'; | |
| 5802 | - | |
| 5803 | - if (!$use_pinecone || empty($pinecone_options['mxchat_pinecone_api_key'])) { | |
| 5804 | - wp_send_json_error('Pinecone is not properly configured for bot: ' . $bot_id); | |
| 5805 | - exit; | |
| 5806 | - } | |
| 5807 | - | |
| 5808 | - $api_key = $pinecone_options['mxchat_pinecone_api_key']; | |
| 5809 | - $host = $pinecone_options['mxchat_pinecone_host']; | |
| 5810 | - $namespace = $pinecone_options['mxchat_pinecone_namespace'] ?? ''; | |
| 5811 | - | |
| 5812 | - // Collect all vector IDs to delete | |
| 5813 | - $vectors_to_delete = array(); | |
| 5814 | - | |
| 5815 | - // Add the original single-vector ID (for non-chunked content) | |
| 5816 | - $vectors_to_delete[] = $base_vector_id; | |
| 5817 | - | |
| 5818 | - // Use Pinecone list API to find all chunk vectors with this prefix | |
| 5819 | - // NOTE: Pinecone List API is a GET request with query parameters, not POST | |
| 5820 | - $prefix = $base_vector_id . '_chunk_'; | |
| 5821 | - | |
| 5822 | - $query_params = array( | |
| 5823 | - 'prefix' => $prefix, | |
| 5824 | - 'limit' => 100 | |
| 5825 | - ); | |
| 5826 | - | |
| 5827 | - if (!empty($namespace)) { | |
| 5828 | - $query_params['namespace'] = $namespace; | |
| 5829 | - } | |
| 5830 | - | |
| 5831 | - $list_url = "https://{$host}/vectors/list?" . http_build_query($query_params); | |
| 5832 | - | |
| 5833 | - $list_response = wp_remote_get($list_url, array( | |
| 5834 | - 'headers' => array( | |
| 5835 | - 'Api-Key' => $api_key, | |
| 5836 | - 'accept' => 'application/json' | |
| 5837 | - ), | |
| 5838 | - 'timeout' => 30 | |
| 5839 | - )); | |
| 5840 | - | |
| 5841 | - if (!is_wp_error($list_response)) { | |
| 5842 | - $list_body_response = wp_remote_retrieve_body($list_response); | |
| 5843 | - $list_data = json_decode($list_body_response, true); | |
| 5844 | - if (!empty($list_data['vectors'])) { | |
| 5845 | - foreach ($list_data['vectors'] as $vector) { | |
| 5846 | - if (isset($vector['id'])) { | |
| 5847 | - $vectors_to_delete[] = $vector['id']; | |
| 5848 | - } | |
| 5849 | - } | |
| 5850 | - } | |
| 5851 | - } | |
| 5852 | - | |
| 5853 | - if (empty($vectors_to_delete)) { | |
| 5854 | - wp_send_json_success(array( | |
| 5855 | - 'message' => 'No vectors found to delete', | |
| 5856 | - 'source_url' => $source_url | |
| 5857 | - )); | |
| 5858 | - exit; | |
| 5859 | - } | |
| 5860 | - | |
| 5861 | - // Delete all vectors using the same endpoint as the working function | |
| 5862 | - $delete_url = "https://{$host}/vectors/delete"; | |
| 5863 | - | |
| 5864 | - $delete_body = array( | |
| 5865 | - 'ids' => $vectors_to_delete | |
| 5866 | - ); | |
| 5867 | - | |
| 5868 | - if (!empty($namespace)) { | |
| 5869 | - $delete_body['namespace'] = $namespace; | |
| 5870 | - } | |
| 5871 | - | |
| 5872 | - $delete_response = wp_remote_post($delete_url, array( | |
| 5873 | - 'headers' => array( | |
| 5874 | - 'Api-Key' => $api_key, | |
| 5875 | - 'accept' => 'application/json', | |
| 5876 | - 'content-type' => 'application/json' | |
| 5877 | - ), | |
| 5878 | - 'body' => wp_json_encode($delete_body), | |
| 5879 | - 'timeout' => 30 | |
| 5880 | - )); | |
| 5881 | - | |
| 5882 | - if (is_wp_error($delete_response)) { | |
| 5883 | - MxChat_Admin::mxchat_log_debug('pinecone_error', 'Failed to delete chunks from Pinecone: ' . $delete_response->get_error_message(), array('source_url' => $source_url)); | |
| 5884 | - wp_send_json_error('Failed to delete from Pinecone: ' . $delete_response->get_error_message()); | |
| 5885 | - exit; | |
| 5886 | - } | |
| 5887 | - | |
| 5888 | - $response_code = wp_remote_retrieve_response_code($delete_response); | |
| 5889 | - | |
| 5890 | - if ($response_code !== 200) { | |
| 5891 | - MxChat_Admin::mxchat_log_debug('pinecone_error', 'Pinecone API error (HTTP ' . $response_code . ')', array('source_url' => $source_url)); | |
| 5892 | - wp_send_json_error('Pinecone API error (HTTP ' . $response_code . ')'); | |
| 5893 | - exit; | |
| 5894 | - } | |
| 5895 | - | |
| 5896 | - wp_send_json_success(array( | |
| 5897 | - 'message' => 'All chunks deleted successfully from Pinecone', | |
| 5898 | - 'source_url' => $source_url, | |
| 5899 | - 'deleted_count' => count($vectors_to_delete) | |
| 5900 | - )); | |
| 5901 | - | |
| 5902 | - } else { | |
| 5903 | - // WordPress database deletion | |
| 5904 | - global $wpdb; | |
| 5905 | - $table_name = $wpdb->prefix . 'mxchat_system_prompt_content'; | |
| 5906 | - | |
| 5907 | - $result = $wpdb->delete( | |
| 5908 | - $table_name, | |
| 5909 | - array('source_url' => $source_url), | |
| 5910 | - array('%s') | |
| 5911 | - ); | |
| 5912 | - | |
| 5913 | - if ($result === false) { | |
| 5914 | - MxChat_Admin::mxchat_log_debug('database_error', 'Failed to delete from database: ' . $wpdb->last_error, array('source_url' => $source_url)); | |
| 5915 | - wp_send_json_error('Failed to delete from database: ' . $wpdb->last_error); | |
| 5916 | - exit; | |
| 5917 | - } | |
| 5918 | - | |
| 5919 | - wp_send_json_success(array( | |
| 5920 | - 'message' => 'All chunks deleted successfully from database', | |
| 5921 | - 'source_url' => $source_url, | |
| 5922 | - 'deleted_count' => $result | |
| 5923 | - )); | |
| 5924 | - } | |
| 5925 | - | |
| 5926 | - exit; | |
| 5927 | -} | |
| 5928 | - | |
| 5929 | -/** | |
| 5930 | - * Handle individual WordPress database content deletion via AJAX | |
| 5931 | - * Mirrors the Pinecone delete handler but for WordPress database entries | |
| 5932 | - */ | |
| 5933 | -public function ajax_mxchat_delete_wordpress_prompt() { | |
| 5934 | - // Verify nonce and permissions | |
| 5935 | - if (!check_ajax_referer('mxchat_delete_wordpress_prompt_nonce', 'nonce', false)) { | |
| 5936 | - wp_send_json_error('Invalid nonce'); | |
| 5937 | - exit; | |
| 5938 | - } | |
| 5939 | - | |
| 5940 | - if (!current_user_can('manage_options')) { | |
| 5941 | - wp_send_json_error('Unauthorized access'); | |
| 5942 | - exit; | |
| 5943 | - } | |
| 5944 | - | |
| 5945 | - $entry_id = isset($_POST['entry_id']) ? intval($_POST['entry_id']) : 0; | |
| 5946 | - | |
| 5947 | - if (empty($entry_id)) { | |
| 5948 | - wp_send_json_error('Missing entry ID'); | |
| 5949 | - exit; | |
| 5950 | - } | |
| 5951 | - | |
| 5952 | - global $wpdb; | |
| 5953 | - $table_name = $wpdb->prefix . 'mxchat_system_prompt_content'; | |
| 5954 | - | |
| 5955 | - // Clear cache for this entry | |
| 5956 | - wp_cache_delete('prompt_' . $entry_id, 'mxchat_prompts'); | |
| 5957 | - | |
| 5958 | - // Delete from database | |
| 5959 | - $result = $wpdb->delete( | |
| 5960 | - $table_name, | |
| 5961 | - array('id' => $entry_id), | |
| 5962 | - array('%d') | |
| 5963 | - ); | |
| 5964 | - | |
| 5965 | - if ($result !== false) { | |
| 5966 | - wp_send_json_success(array( | |
| 5967 | - 'message' => 'Entry deleted successfully', | |
| 5968 | - 'entry_id' => $entry_id | |
| 5969 | - )); | |
| 5970 | - } else { | |
| 5971 | - wp_send_json_error('Failed to delete entry from database'); | |
| 5972 | - } | |
| 5973 | - | |
| 5974 | - exit; | |
| 5975 | -} | |
| 5976 | - | |
| 5977 | -/** | |
| 5978 | - * Handle bulk deletion of knowledge entries via AJAX | |
| 5979 | - * Supports both Pinecone and WordPress database entries | |
| 5980 | - */ | |
| 5981 | -public function ajax_mxchat_bulk_delete_knowledge() { | |
| 5982 | - // Verify nonce and permissions | |
| 5983 | - if (!check_ajax_referer('mxchat_bulk_delete_knowledge_nonce', 'nonce', false)) { | |
| 5984 | - wp_send_json_error('Invalid nonce'); | |
| 5985 | - exit; | |
| 5986 | - } | |
| 5987 | - | |
| 5988 | - if (!current_user_can('manage_options')) { | |
| 5989 | - wp_send_json_error('Unauthorized access'); | |
| 5990 | - exit; | |
| 5991 | - } | |
| 5992 | - | |
| 5993 | - $entries = isset($_POST['entries']) ? $_POST['entries'] : array(); | |
| 5994 | - $bot_id = isset($_POST['bot_id']) ? sanitize_text_field($_POST['bot_id']) : 'default'; | |
| 5995 | - | |
| 5996 | - if (empty($entries) || !is_array($entries)) { | |
| 5997 | - wp_send_json_error('No entries provided'); | |
| 5998 | - exit; | |
| 5999 | - } | |
| 6000 | - | |
| 6001 | - // Extend execution time — bulk Pinecone operations can take a while | |
| 6002 | - if (function_exists('set_time_limit')) { | |
| 6003 | - set_time_limit(120); | |
| 6004 | - } | |
| 6005 | - | |
| 6006 | - $success_ids = array(); | |
| 6007 | - $failed_ids = array(); | |
| 6008 | - $errors = array(); | |
| 6009 | - | |
| 6010 | - global $wpdb; | |
| 6011 | - $table_name = $wpdb->prefix . 'mxchat_system_prompt_content'; | |
| 6012 | - | |
| 6013 | - // Get Pinecone manager for Pinecone deletions | |
| 6014 | - $pinecone_manager = MxChat_Pinecone_Manager::get_instance(); | |
| 6015 | - $pinecone_options = $pinecone_manager->mxchat_get_bot_pinecone_options($bot_id); | |
| 6016 | - $use_pinecone = ($pinecone_options['mxchat_use_pinecone'] ?? '0') === '1'; | |
| 6017 | - | |
| 6018 | - $api_key = $pinecone_options['mxchat_pinecone_api_key'] ?? ''; | |
| 6019 | - $host = $pinecone_options['mxchat_pinecone_host'] ?? ''; | |
| 6020 | - | |
| 6021 | - // ============================================= | |
| 6022 | - // PHASE 1: Collect all Pinecone vector IDs | |
| 6023 | - // and separate WordPress entries | |
| 6024 | - // ============================================= | |
| 6025 | - $pinecone_entry_ids = array(); // entry IDs that are pinecone-sourced | |
| 6026 | - $wordpress_entries = array(); // entries for WordPress DB deletion | |
| 6027 | - $all_vector_ids = array(); // all pinecone vector IDs to delete in one batch | |
| 6028 | - | |
| 6029 | - foreach ($entries as $entry) { | |
| 6030 | - $entry_id = sanitize_text_field($entry['id'] ?? ''); | |
| 6031 | - $source = sanitize_text_field($entry['source'] ?? 'wordpress'); | |
| 6032 | - $source_url = isset($entry['sourceUrl']) ? esc_url_raw($entry['sourceUrl']) : ''; | |
| 6033 | - $is_group = isset($entry['isGroup']) && ($entry['isGroup'] === true || $entry['isGroup'] === 'true'); | |
| 6034 | - | |
| 6035 | - if (empty($entry_id)) { | |
| 6036 | - continue; | |
| 6037 | - } | |
| 6038 | - | |
| 6039 | - if ($source === 'pinecone') { | |
| 6040 | - if (!$use_pinecone || empty($api_key)) { | |
| 6041 | - $failed_ids[] = $entry_id; | |
| 6042 | - $errors[] = "Pinecone not configured for entry: $entry_id"; | |
| 6043 | - continue; | |
| 6044 | - } | |
| 6045 | - | |
| 6046 | - $pinecone_entry_ids[] = $entry_id; | |
| 6047 | - | |
| 6048 | - if ($is_group && !empty($source_url)) { | |
| 6049 | - // Grouped/chunked entry: collect base ID + chunk IDs via List API | |
| 6050 | - $base_vector_id = md5($source_url); | |
| 6051 | - $all_vector_ids[] = $base_vector_id; | |
| 6052 | - | |
| 6053 | - $list_url = 'https://' . $host . '/vectors/list?prefix=' . urlencode($base_vector_id . '_chunk_') . '&limit=100'; | |
| 6054 | - $list_response = wp_remote_get($list_url, array( | |
| 6055 | - 'headers' => array( | |
| 6056 | - 'Api-Key' => $api_key, | |
| 6057 | - 'accept' => 'application/json' | |
| 6058 | - ), | |
| 6059 | - 'timeout' => 30 | |
| 6060 | - )); | |
| 6061 | - | |
| 6062 | - if (!is_wp_error($list_response)) { | |
| 6063 | - $list_body = json_decode(wp_remote_retrieve_body($list_response), true); | |
| 6064 | - if (!empty($list_body['vectors']) && is_array($list_body['vectors'])) { | |
| 6065 | - foreach ($list_body['vectors'] as $vector) { | |
| 6066 | - if (isset($vector['id'])) { | |
| 6067 | - $all_vector_ids[] = $vector['id']; | |
| 6068 | - } | |
| 6069 | - } | |
| 6070 | - } | |
| 6071 | - } | |
| 6072 | - } else { | |
| 6073 | - // Single entry: the entry_id IS the vector ID | |
| 6074 | - $all_vector_ids[] = $entry_id; | |
| 6075 | - } | |
| 6076 | - } else { | |
| 6077 | - $wordpress_entries[] = $entry; | |
| 6078 | - } | |
| 6079 | - } | |
| 6080 | - | |
| 6081 | - // ============================================= | |
| 6082 | - // PHASE 2: Single batch delete to Pinecone | |
| 6083 | - // ============================================= | |
| 6084 | - if (!empty($all_vector_ids)) { | |
| 6085 | - $all_vector_ids = array_values(array_unique($all_vector_ids)); | |
| 6086 | - $pinecone_success = true; | |
| 6087 | - $batches = array_chunk($all_vector_ids, 100); | |
| 6088 | - | |
| 6089 | - foreach ($batches as $batch) { | |
| 6090 | - $delete_response = wp_remote_post("https://{$host}/vectors/delete", array( | |
| 6091 | - 'headers' => array( | |
| 6092 | - 'Api-Key' => $api_key, | |
| 6093 | - 'accept' => 'application/json', | |
| 6094 | - 'content-type' => 'application/json' | |
| 6095 | - ), | |
| 6096 | - 'body' => wp_json_encode(array('ids' => $batch)), | |
| 6097 | - 'timeout' => 60 | |
| 6098 | - )); | |
| 6099 | - | |
| 6100 | - if (is_wp_error($delete_response)) { | |
| 6101 | - $pinecone_success = false; | |
| 6102 | - $errors[] = 'Pinecone batch deletion failed: ' . $delete_response->get_error_message(); | |
| 6103 | - MxChat_Admin::mxchat_log_debug('pinecone_error', 'Bulk delete batch failed: ' . $delete_response->get_error_message()); | |
| 6104 | - } else { | |
| 6105 | - $response_code = wp_remote_retrieve_response_code($delete_response); | |
| 6106 | - if ($response_code !== 200) { | |
| 6107 | - $pinecone_success = false; | |
| 6108 | - $response_body = wp_remote_retrieve_body($delete_response); | |
| 6109 | - $errors[] = "Pinecone API error (HTTP $response_code)"; | |
| 6110 | - MxChat_Admin::mxchat_log_debug('pinecone_error', 'Bulk delete batch failed (HTTP ' . $response_code . ')', array('response' => substr($response_body, 0, 200))); | |
| 6111 | - } | |
| 6112 | - } | |
| 6113 | - } | |
| 6114 | - | |
| 6115 | - // Mark all pinecone entries based on batch result | |
| 6116 | - foreach ($pinecone_entry_ids as $eid) { | |
| 6117 | - if ($pinecone_success) { | |
| 6118 | - $success_ids[] = $eid; | |
| 6119 | - } else { | |
| 6120 | - $failed_ids[] = $eid; | |
| 6121 | - } | |
| 6122 | - } | |
| 6123 | - } | |
| 6124 | - | |
| 6125 | - // ============================================= | |
| 6126 | - // PHASE 3: WordPress database deletions | |
| 6127 | - // ============================================= | |
| 6128 | - foreach ($wordpress_entries as $entry) { | |
| 6129 | - $entry_id = sanitize_text_field($entry['id'] ?? ''); | |
| 6130 | - $source_url = isset($entry['sourceUrl']) ? esc_url_raw($entry['sourceUrl']) : ''; | |
| 6131 | - $is_group = isset($entry['isGroup']) && ($entry['isGroup'] === true || $entry['isGroup'] === 'true'); | |
| 6132 | - | |
| 6133 | - if (empty($entry_id)) { | |
| 6134 | - continue; | |
| 6135 | - } | |
| 6136 | - | |
| 6137 | - try { | |
| 6138 | - if ($is_group && !empty($source_url)) { | |
| 6139 | - $result = $wpdb->delete( | |
| 6140 | - $table_name, | |
| 6141 | - array('source_url' => $source_url), | |
| 6142 | - array('%s') | |
| 6143 | - ); | |
| 6144 | - } else { | |
| 6145 | - wp_cache_delete('prompt_' . $entry_id, 'mxchat_prompts'); | |
| 6146 | - $result = $wpdb->delete( | |
| 6147 | - $table_name, | |
| 6148 | - array('id' => intval($entry_id)), | |
| 6149 | - array('%d') | |
| 6150 | - ); | |
| 6151 | - } | |
| 6152 | - | |
| 6153 | - if ($result !== false) { | |
| 6154 | - $success_ids[] = $entry_id; | |
| 6155 | - } else { | |
| 6156 | - $failed_ids[] = $entry_id; | |
| 6157 | - $errors[] = "Database error for entry: $entry_id"; | |
| 6158 | - } | |
| 6159 | - } catch (Exception $e) { | |
| 6160 | - $failed_ids[] = $entry_id; | |
| 6161 | - $errors[] = $e->getMessage(); | |
| 6162 | - } | |
| 6163 | - } | |
| 6164 | - | |
| 6165 | - wp_send_json_success(array( | |
| 6166 | - 'success_ids' => $success_ids, | |
| 6167 | - 'failed_ids' => $failed_ids, | |
| 6168 | - 'errors' => $errors, | |
| 6169 | - 'total_processed' => count($success_ids) + count($failed_ids) | |
| 6170 | - )); | |
| 6171 | - | |
| 6172 | - exit; | |
| 6173 | -} | |
| 6174 | - | |
| 6175 | -/** | |
| 6176 | - * Get hierarchical roles for dropdown | |
| 6177 | - */ | |
| 6178 | -public function mxchat_get_role_options() { | |
| 6179 | - return array( | |
| 6180 | - 'public' => __('Public (Everyone)', 'mxchat'), | |
| 6181 | - 'logged_in' => __('Logged In Users', 'mxchat'), | |
| 6182 | - 'subscriber' => __('Subscribers & Above', 'mxchat'), | |
| 6183 | - 'contributor' => __('Contributors & Above', 'mxchat'), | |
| 6184 | - 'author' => __('Authors & Above', 'mxchat'), | |
| 6185 | - 'editor' => __('Editors & Above', 'mxchat'), | |
| 6186 | - 'administrator' => __('Administrators Only', 'mxchat') | |
| 6187 | - ); | |
| 6188 | -} | |
| 6189 | - | |
| 6190 | -/** | |
| 6191 | - * Check if user has access to content based on role restriction | |
| 6192 | - */ | |
| 6193 | -public function mxchat_user_has_content_access($role_restriction) { | |
| 6194 | - // Public content is always accessible | |
| 6195 | - if ($role_restriction === 'public' || empty($role_restriction)) { | |
| 6196 | - return true; | |
| 6197 | - } | |
| 6198 | - | |
| 6199 | - // Check if user is logged in for logged_in restriction | |
| 6200 | - if ($role_restriction === 'logged_in') { | |
| 6201 | - return is_user_logged_in(); | |
| 6202 | - } | |
| 6203 | - | |
| 6204 | - // If not logged in, no access to role-restricted content | |
| 6205 | - if (!is_user_logged_in()) { | |
| 6206 | - return false; | |
| 6207 | - } | |
| 6208 | - | |
| 6209 | - $user = wp_get_current_user(); | |
| 6210 | - $user_roles = $user->roles; | |
| 6211 | - | |
| 6212 | - if (empty($user_roles)) { | |
| 6213 | - return false; | |
| 6214 | - } | |
| 6215 | - | |
| 6216 | - // Define role hierarchy (higher number = higher access) | |
| 6217 | - $hierarchy = array( | |
| 6218 | - 'subscriber' => 1, | |
| 6219 | - 'contributor' => 2, | |
| 6220 | - 'author' => 3, | |
| 6221 | - 'editor' => 4, | |
| 6222 | - 'administrator' => 5 | |
| 6223 | - ); | |
| 6224 | - | |
| 6225 | - // Get required level | |
| 6226 | - $required_level = isset($hierarchy[$role_restriction]) ? $hierarchy[$role_restriction] : 0; | |
| 6227 | - | |
| 6228 | - // Check if user has required level or higher | |
| 6229 | - foreach ($user_roles as $user_role) { | |
| 6230 | - $user_level = isset($hierarchy[$user_role]) ? $hierarchy[$user_role] : 0; | |
| 6231 | - if ($user_level >= $required_level) { | |
| 6232 | - return true; | |
| 6233 | - } | |
| 6234 | - } | |
| 6235 | - | |
| 6236 | - return false; | |
| 6237 | -} | |
| 6238 | - | |
| 6239 | -/** | |
| 6240 | - * Handle role restriction updates via AJAX | |
| 6241 | - * Removed cache clearing call since we removed caching | |
| 6242 | - */ | |
| 6243 | -public function ajax_mxchat_update_role_restriction() { | |
| 6244 | - // Verify nonce and permissions | |
| 6245 | - if (!check_ajax_referer('mxchat_update_role_nonce', 'nonce', false)) { | |
| 6246 | - wp_send_json_error('Invalid nonce'); | |
| 6247 | - exit; | |
| 6248 | - } | |
| 6249 | - | |
| 6250 | - if (!current_user_can('manage_options')) { | |
| 6251 | - wp_send_json_error('Unauthorized access'); | |
| 6252 | - exit; | |
| 6253 | - } | |
| 6254 | - | |
| 6255 | - $entry_id = isset($_POST['entry_id']) ? sanitize_text_field($_POST['entry_id']) : ''; | |
| 6256 | - $role_restriction = isset($_POST['role_restriction']) ? sanitize_text_field($_POST['role_restriction']) : 'public'; | |
| 6257 | - $data_source = isset($_POST['data_source']) ? sanitize_text_field($_POST['data_source']) : 'wordpress'; | |
| 6258 | - | |
| 6259 | - if (empty($entry_id)) { | |
| 6260 | - wp_send_json_error('Invalid entry ID'); | |
| 6261 | - exit; | |
| 6262 | - } | |
| 6263 | - | |
| 6264 | - // Get knowledge manager instance to validate role restriction | |
| 6265 | - $knowledge_manager = MxChat_Knowledge_Manager::get_instance(); | |
| 6266 | - $valid_roles = array_keys($knowledge_manager->mxchat_get_role_options()); | |
| 6267 | - if (!in_array($role_restriction, $valid_roles)) { | |
| 6268 | - wp_send_json_error('Invalid role restriction'); | |
| 6269 | - exit; | |
| 6270 | - } | |
| 6271 | - | |
| 6272 | - global $wpdb; | |
| 6273 | - | |
| 6274 | - if ($data_source === 'pinecone') { | |
| 6275 | - // Handle Pinecone role restriction (stored separately in WordPress table) | |
| 6276 | - $roles_table = $wpdb->prefix . 'mxchat_pinecone_roles'; | |
| 6277 | - | |
| 6278 | - // Use REPLACE to insert or update the role restriction | |
| 6279 | - $result = $wpdb->replace( | |
| 6280 | - $roles_table, | |
| 6281 | - array( | |
| 6282 | - 'vector_id' => $entry_id, | |
| 6283 | - 'role_restriction' => $role_restriction, | |
| 6284 | - 'updated_at' => current_time('mysql') | |
| 6285 | - ), | |
| 6286 | - array('%s', '%s', '%s') | |
| 6287 | - ); | |
| 6288 | - | |
| 6289 | - // No cache clearing needed since we removed caching | |
| 6290 | - | |
| 6291 | - } else { | |
| 6292 | - // Handle WordPress database role restriction (existing functionality) | |
| 6293 | - $table_name = $wpdb->prefix . 'mxchat_system_prompt_content'; | |
| 6294 | - | |
| 6295 | - $result = $wpdb->update( | |
| 6296 | - $table_name, | |
| 6297 | - array('role_restriction' => $role_restriction), | |
| 6298 | - array('id' => absint($entry_id)), | |
| 6299 | - array('%s'), | |
| 6300 | - array('%d') | |
| 6301 | - ); | |
| 6302 | - } | |
| 6303 | - | |
| 6304 | - if ($result === false) { | |
| 6305 | - wp_send_json_error('Database update failed: ' . $wpdb->last_error); | |
| 6306 | - exit; | |
| 6307 | - } | |
| 6308 | - | |
| 6309 | - wp_send_json_success(array( | |
| 6310 | - 'message' => 'Role restriction updated successfully', | |
| 6311 | - 'role_restriction' => $role_restriction, | |
| 6312 | - 'data_source' => $data_source, | |
| 6313 | - 'entry_id' => $entry_id | |
| 6314 | - )); | |
| 6315 | - exit; | |
| 6316 | -} | |
| 6317 | - | |
| 6318 | -// ======================================== | |
| 6319 | -// ROLE-BASED CONTENT RESTRICTIONS - NEW FUNCTIONS | |
| 6320 | -// Add these to your MxChat_Knowledge_Manager class | |
| 6321 | -// ======================================== | |
| 6322 | - | |
| 6323 | -/** | |
| 6324 | - * Initialize role-based content hooks | |
| 6325 | - * Add this call to your __construct() or mxchat_init_hooks() method | |
| 6326 | - */ | |
| 6327 | -private function mxchat_init_role_hooks() { | |
| 6328 | - // AJAX handlers for tag-role mappings | |
| 6329 | - add_action('wp_ajax_mxchat_add_tag_role_mapping', array($this, 'ajax_add_tag_role_mapping')); | |
| 6330 | - add_action('wp_ajax_mxchat_delete_tag_role_mapping', array($this, 'ajax_delete_tag_role_mapping')); | |
| 6331 | - add_action('wp_ajax_mxchat_get_tag_role_mappings', array($this, 'ajax_get_tag_role_mappings')); | |
| 6332 | - add_action('wp_ajax_mxchat_bulk_update_tag_roles', array($this, 'ajax_bulk_update_tag_roles')); | |
| 6333 | - | |
| 6334 | - // Hook to automatically update role restrictions when tags are added/removed | |
| 6335 | - add_action('set_object_terms', array($this, 'handle_tag_change'), 10, 6); | |
| 6336 | - | |
| 6337 | - // Hook to apply role restrictions on auto-sync | |
| 6338 | - add_action('mxchat_content_stored', array($this, 'apply_role_restriction_after_storage'), 10, 2); | |
| 6339 | -} | |
| 6340 | - | |
| 6341 | -/** | |
| 6342 | - * Add tag-role mapping via AJAX | |
| 6343 | - */ | |
| 6344 | -public function ajax_add_tag_role_mapping() { | |
| 6345 | - // Verify nonce and permissions | |
| 6346 | - check_ajax_referer('mxchat_prompts_setting_nonce', 'nonce'); | |
| 6347 | - | |
| 6348 | - if (!current_user_can('manage_options')) { | |
| 6349 | - wp_send_json_error('Unauthorized access'); | |
| 6350 | - exit; | |
| 6351 | - } | |
| 6352 | - | |
| 6353 | - $tag_slug = isset($_POST['tag_slug']) ? sanitize_text_field($_POST['tag_slug']) : ''; | |
| 6354 | - $role_restriction = isset($_POST['role_restriction']) ? sanitize_text_field($_POST['role_restriction']) : 'public'; | |
| 6355 | - | |
| 6356 | - if (empty($tag_slug)) { | |
| 6357 | - wp_send_json_error('Tag slug is required'); | |
| 6358 | - exit; | |
| 6359 | - } | |
| 6360 | - | |
| 6361 | - // Validate role restriction | |
| 6362 | - $valid_roles = array_keys($this->mxchat_get_role_options()); | |
| 6363 | - if (!in_array($role_restriction, $valid_roles)) { | |
| 6364 | - wp_send_json_error('Invalid role restriction'); | |
| 6365 | - exit; | |
| 6366 | - } | |
| 6367 | - | |
| 6368 | - // Check if tag exists in WordPress | |
| 6369 | - $term = get_term_by('slug', $tag_slug, 'post_tag'); | |
| 6370 | - if (!$term) { | |
| 6371 | - wp_send_json_error('Tag does not exist in WordPress'); | |
| 6372 | - exit; | |
| 6373 | - } | |
| 6374 | - | |
| 6375 | - // Get existing mappings | |
| 6376 | - $mappings = get_option('mxchat_tag_role_mappings', array()); | |
| 6377 | - | |
| 6378 | - // Check if mapping already exists | |
| 6379 | - if (isset($mappings[$tag_slug])) { | |
| 6380 | - wp_send_json_error('Mapping for this tag already exists'); | |
| 6381 | - exit; | |
| 6382 | - } | |
| 6383 | - | |
| 6384 | - // Add new mapping | |
| 6385 | - $mappings[$tag_slug] = $role_restriction; | |
| 6386 | - update_option('mxchat_tag_role_mappings', $mappings); | |
| 6387 | - | |
| 6388 | - wp_send_json_success(array( | |
| 6389 | - 'message' => 'Tag-role mapping added successfully', | |
| 6390 | - 'tag_slug' => $tag_slug, | |
| 6391 | - 'role_restriction' => $role_restriction | |
| 6392 | - )); | |
| 6393 | - exit; | |
| 6394 | -} | |
| 6395 | - | |
| 6396 | -/** | |
| 6397 | - * Delete tag-role mapping via AJAX | |
| 6398 | - */ | |
| 6399 | -public function ajax_delete_tag_role_mapping() { | |
| 6400 | - // Verify nonce and permissions | |
| 6401 | - check_ajax_referer('mxchat_prompts_setting_nonce', 'nonce'); | |
| 6402 | - | |
| 6403 | - if (!current_user_can('manage_options')) { | |
| 6404 | - wp_send_json_error('Unauthorized access'); | |
| 6405 | - exit; | |
| 6406 | - } | |
| 6407 | - | |
| 6408 | - $tag_slug = isset($_POST['tag_slug']) ? sanitize_text_field($_POST['tag_slug']) : ''; | |
| 6409 | - | |
| 6410 | - if (empty($tag_slug)) { | |
| 6411 | - wp_send_json_error('Tag slug is required'); | |
| 6412 | - exit; | |
| 6413 | - } | |
| 6414 | - | |
| 6415 | - // Get existing mappings | |
| 6416 | - $mappings = get_option('mxchat_tag_role_mappings', array()); | |
| 6417 | - | |
| 6418 | - // Check if mapping exists | |
| 6419 | - if (!isset($mappings[$tag_slug])) { | |
| 6420 | - wp_send_json_error('Mapping does not exist'); | |
| 6421 | - exit; | |
| 6422 | - } | |
| 6423 | - | |
| 6424 | - // Remove mapping | |
| 6425 | - unset($mappings[$tag_slug]); | |
| 6426 | - update_option('mxchat_tag_role_mappings', $mappings); | |
| 6427 | - | |
| 6428 | - wp_send_json_success(array( | |
| 6429 | - 'message' => 'Tag-role mapping deleted successfully', | |
| 6430 | - 'tag_slug' => $tag_slug | |
| 6431 | - )); | |
| 6432 | - exit; | |
| 6433 | -} | |
| 6434 | - | |
| 6435 | -/** | |
| 6436 | - * Get all tag-role mappings via AJAX | |
| 6437 | - */ | |
| 6438 | -public function ajax_get_tag_role_mappings() { | |
| 6439 | - // Verify nonce and permissions | |
| 6440 | - check_ajax_referer('mxchat_prompts_setting_nonce', 'nonce'); | |
| 6441 | - | |
| 6442 | - if (!current_user_can('manage_options')) { | |
| 6443 | - wp_send_json_error('Unauthorized access'); | |
| 6444 | - exit; | |
| 6445 | - } | |
| 6446 | - | |
| 6447 | - // Get mappings | |
| 6448 | - $mappings = get_option('mxchat_tag_role_mappings', array()); | |
| 6449 | - $role_options = $this->mxchat_get_role_options(); | |
| 6450 | - | |
| 6451 | - $formatted_mappings = array(); | |
| 6452 | - | |
| 6453 | - foreach ($mappings as $tag_slug => $role_restriction) { | |
| 6454 | - // Get tag object | |
| 6455 | - $term = get_term_by('slug', $tag_slug, 'post_tag'); | |
| 6456 | - | |
| 6457 | - // Count posts with this tag | |
| 6458 | - $post_count = 0; | |
| 6459 | - if ($term) { | |
| 6460 | - $post_count = $term->count; | |
| 6461 | - } | |
| 6462 | - | |
| 6463 | - $formatted_mappings[] = array( | |
| 6464 | - 'tag_slug' => $tag_slug, | |
| 6465 | - 'role_restriction' => $role_restriction, | |
| 6466 | - 'role_label' => $role_options[$role_restriction] ?? $role_restriction, | |
| 6467 | - 'post_count' => $post_count | |
| 6468 | - ); | |
| 6469 | - } | |
| 6470 | - | |
| 6471 | - wp_send_json_success(array( | |
| 6472 | - 'mappings' => $formatted_mappings | |
| 6473 | - )); | |
| 6474 | - exit; | |
| 6475 | -} | |
| 6476 | - | |
| 6477 | -/** | |
| 6478 | - * Bulk update role restrictions for all existing content with mapped tags | |
| 6479 | - */ | |
| 6480 | -public function ajax_bulk_update_tag_roles() { | |
| 6481 | - // Verify nonce and permissions | |
| 6482 | - check_ajax_referer('mxchat_prompts_setting_nonce', 'nonce'); | |
| 6483 | - | |
| 6484 | - if (!current_user_can('manage_options')) { | |
| 6485 | - wp_send_json_error('Unauthorized access'); | |
| 6486 | - exit; | |
| 6487 | - } | |
| 6488 | - | |
| 6489 | - // Get mappings | |
| 6490 | - $mappings = get_option('mxchat_tag_role_mappings', array()); | |
| 6491 | - | |
| 6492 | - if (empty($mappings)) { | |
| 6493 | - wp_send_json_error('No tag-role mappings found'); | |
| 6494 | - exit; | |
| 6495 | - } | |
| 6496 | - | |
| 6497 | - global $wpdb; | |
| 6498 | - | |
| 6499 | - // Check if using Pinecone | |
| 6500 | - $pinecone_options = get_option('mxchat_pinecone_addon_options', array()); | |
| 6501 | - $use_pinecone = ($pinecone_options['mxchat_use_pinecone'] ?? '0') === '1'; | |
| 6502 | - | |
| 6503 | - $updated_count = 0; | |
| 6504 | - $details = array(); | |
| 6505 | - | |
| 6506 | - foreach ($mappings as $tag_slug => $role_restriction) { | |
| 6507 | - // Get all posts with this tag | |
| 6508 | - $posts = get_posts(array( | |
| 6509 | - 'tag' => $tag_slug, | |
| 6510 | - 'post_type' => 'any', | |
| 6511 | - 'posts_per_page' => -1, | |
| 6512 | - 'fields' => 'ids', | |
| 6513 | - 'post_status' => 'publish' | |
| 6514 | - )); | |
| 6515 | - | |
| 6516 | - if (empty($posts)) { | |
| 6517 | - continue; | |
| 6518 | - } | |
| 6519 | - | |
| 6520 | - $tag_updated = 0; | |
| 6521 | - | |
| 6522 | - foreach ($posts as $post_id) { | |
| 6523 | - $source_url = get_permalink($post_id); | |
| 6524 | - if (!$source_url) { | |
| 6525 | - continue; | |
| 6526 | - } | |
| 6527 | - | |
| 6528 | - if ($use_pinecone && !empty($pinecone_options['mxchat_pinecone_api_key'])) { | |
| 6529 | - // Update Pinecone role restriction | |
| 6530 | - $roles_table = $wpdb->prefix . 'mxchat_pinecone_roles'; | |
| 6531 | - $vector_id = md5($source_url); | |
| 6532 | - | |
| 6533 | - $result = $wpdb->replace( | |
| 6534 | - $roles_table, | |
| 6535 | - array( | |
| 6536 | - 'vector_id' => $vector_id, | |
| 6537 | - 'role_restriction' => $role_restriction, | |
| 6538 | - 'updated_at' => current_time('mysql') | |
| 6539 | - ), | |
| 6540 | - array('%s', '%s', '%s') | |
| 6541 | - ); | |
| 6542 | - } else { | |
| 6543 | - // Update WordPress DB | |
| 6544 | - $table_name = $wpdb->prefix . 'mxchat_system_prompt_content'; | |
| 6545 | - | |
| 6546 | - $result = $wpdb->update( | |
| 6547 | - $table_name, | |
| 6548 | - array('role_restriction' => $role_restriction), | |
| 6549 | - array('source_url' => $source_url), | |
| 6550 | - array('%s'), | |
| 6551 | - array('%s') | |
| 6552 | - ); | |
| 6553 | - } | |
| 6554 | - | |
| 6555 | - if ($result !== false) { | |
| 6556 | - $tag_updated++; | |
| 6557 | - $updated_count++; | |
| 6558 | - } | |
| 6559 | - } | |
| 6560 | - | |
| 6561 | - if ($tag_updated > 0) { | |
| 6562 | - $details[] = sprintf( | |
| 6563 | - 'Tag "%s" (%s): %d posts updated', | |
| 6564 | - $tag_slug, | |
| 6565 | - $role_restriction, | |
| 6566 | - $tag_updated | |
| 6567 | - ); | |
| 6568 | - } | |
| 6569 | - } | |
| 6570 | - | |
| 6571 | - wp_send_json_success(array( | |
| 6572 | - 'message' => 'Bulk update completed', | |
| 6573 | - 'updated_count' => $updated_count, | |
| 6574 | - 'tags_processed' => count($mappings), | |
| 6575 | - 'details' => $details | |
| 6576 | - )); | |
| 6577 | - exit; | |
| 6578 | -} | |
| 6579 | - | |
| 6580 | -/** | |
| 6581 | - * Handle tag changes on posts (when tags are added or removed) | |
| 6582 | - */ | |
| 6583 | -public function handle_tag_change($object_id, $terms, $tt_ids, $taxonomy, $append, $old_tt_ids) { | |
| 6584 | - // Only process post tags | |
| 6585 | - if ($taxonomy !== 'post_tag') { | |
| 6586 | - return; | |
| 6587 | - } | |
| 6588 | - | |
| 6589 | - // Get tag-role mappings | |
| 6590 | - $mappings = get_option('mxchat_tag_role_mappings', array()); | |
| 6591 | - | |
| 6592 | - if (empty($mappings)) { | |
| 6593 | - return; | |
| 6594 | - } | |
| 6595 | - | |
| 6596 | - // Get the post's URL | |
| 6597 | - $source_url = get_permalink($object_id); | |
| 6598 | - if (!$source_url) { | |
| 6599 | - return; | |
| 6600 | - } | |
| 6601 | - | |
| 6602 | - // Determine the highest role restriction based on tags | |
| 6603 | - $highest_role = 'public'; | |
| 6604 | - $role_hierarchy = array( | |
| 6605 | - 'public' => 0, | |
| 6606 | - 'logged_in' => 1, | |
| 6607 | - 'subscriber' => 2, | |
| 6608 | - 'contributor' => 3, | |
| 6609 | - 'author' => 4, | |
| 6610 | - 'editor' => 5, | |
| 6611 | - 'administrator' => 6 | |
| 6612 | - ); | |
| 6613 | - | |
| 6614 | - // Get all current tags for the post | |
| 6615 | - $current_tags = wp_get_post_tags($object_id, array('fields' => 'slugs')); | |
| 6616 | - | |
| 6617 | - // Find the highest role restriction among the tags | |
| 6618 | - foreach ($current_tags as $tag_slug) { | |
| 6619 | - if (isset($mappings[$tag_slug])) { | |
| 6620 | - $role = $mappings[$tag_slug]; | |
| 6621 | - if (isset($role_hierarchy[$role]) && $role_hierarchy[$role] > $role_hierarchy[$highest_role]) { | |
| 6622 | - $highest_role = $role; | |
| 6623 | - } | |
| 6624 | - } | |
| 6625 | - } | |
| 6626 | - | |
| 6627 | - // Update the role restriction in the database | |
| 6628 | - global $wpdb; | |
| 6629 | - | |
| 6630 | - // Check if using Pinecone | |
| 6631 | - $pinecone_options = get_option('mxchat_pinecone_addon_options', array()); | |
| 6632 | - $use_pinecone = ($pinecone_options['mxchat_use_pinecone'] ?? '0') === '1'; | |
| 6633 | - | |
| 6634 | - if ($use_pinecone && !empty($pinecone_options['mxchat_pinecone_api_key'])) { | |
| 6635 | - // Update Pinecone role restriction | |
| 6636 | - $roles_table = $wpdb->prefix . 'mxchat_pinecone_roles'; | |
| 6637 | - $vector_id = md5($source_url); | |
| 6638 | - | |
| 6639 | - $wpdb->replace( | |
| 6640 | - $roles_table, | |
| 6641 | - array( | |
| 6642 | - 'vector_id' => $vector_id, | |
| 6643 | - 'role_restriction' => $highest_role, | |
| 6644 | - 'updated_at' => current_time('mysql') | |
| 6645 | - ), | |
| 6646 | - array('%s', '%s', '%s') | |
| 6647 | - ); | |
| 6648 | - } else { | |
| 6649 | - // Update WordPress DB | |
| 6650 | - $table_name = $wpdb->prefix . 'mxchat_system_prompt_content'; | |
| 6651 | - | |
| 6652 | - $wpdb->update( | |
| 6653 | - $table_name, | |
| 6654 | - array('role_restriction' => $highest_role), | |
| 6655 | - array('source_url' => $source_url), | |
| 6656 | - array('%s'), | |
| 6657 | - array('%s') | |
| 6658 | - ); | |
| 6659 | - } | |
| 6660 | -} | |
| 6661 | - | |
| 6662 | -/** | |
| 6663 | - * Apply role restriction after content is stored (for auto-sync) | |
| 6664 | - */ | |
| 6665 | -public function apply_role_restriction_after_storage($post_id, $source_url) { | |
| 6666 | - // Get tag-role mappings | |
| 6667 | - $mappings = get_option('mxchat_tag_role_mappings', array()); | |
| 6668 | - | |
| 6669 | - if (empty($mappings)) { | |
| 6670 | - return; | |
| 6671 | - } | |
| 6672 | - | |
| 6673 | - // Get all tags for the post | |
| 6674 | - $post_tags = wp_get_post_tags($post_id, array('fields' => 'slugs')); | |
| 6675 | - | |
| 6676 | - if (empty($post_tags)) { | |
| 6677 | - return; | |
| 6678 | - } | |
| 6679 | - | |
| 6680 | - // Determine the highest role restriction based on tags | |
| 6681 | - $highest_role = 'public'; | |
| 6682 | - $role_hierarchy = array( | |
| 6683 | - 'public' => 0, | |
| 6684 | - 'logged_in' => 1, | |
| 6685 | - 'subscriber' => 2, | |
| 6686 | - 'contributor' => 3, | |
| 6687 | - 'author' => 4, | |
| 6688 | - 'editor' => 5, | |
| 6689 | - 'administrator' => 6 | |
| 6690 | - ); | |
| 6691 | - | |
| 6692 | - foreach ($post_tags as $tag_slug) { | |
| 6693 | - if (isset($mappings[$tag_slug])) { | |
| 6694 | - $role = $mappings[$tag_slug]; | |
| 6695 | - if (isset($role_hierarchy[$role]) && $role_hierarchy[$role] > $role_hierarchy[$highest_role]) { | |
| 6696 | - $highest_role = $role; | |
| 6697 | - } | |
| 6698 | - } | |
| 6699 | - } | |
| 6700 | - | |
| 6701 | - // If no restricted tags found, return (leave as public) | |
| 6702 | - if ($highest_role === 'public') { | |
| 6703 | - return; | |
| 6704 | - } | |
| 6705 | - | |
| 6706 | - // Update the role restriction | |
| 6707 | - global $wpdb; | |
| 6708 | - | |
| 6709 | - // Check if using Pinecone | |
| 6710 | - $pinecone_options = get_option('mxchat_pinecone_addon_options', array()); | |
| 6711 | - $use_pinecone = ($pinecone_options['mxchat_use_pinecone'] ?? '0') === '1'; | |
| 6712 | - | |
| 6713 | - if ($use_pinecone && !empty($pinecone_options['mxchat_pinecone_api_key'])) { | |
| 6714 | - // Update Pinecone role restriction | |
| 6715 | - $roles_table = $wpdb->prefix . 'mxchat_pinecone_roles'; | |
| 6716 | - $vector_id = md5($source_url); | |
| 6717 | - | |
| 6718 | - $wpdb->replace( | |
| 6719 | - $roles_table, | |
| 6720 | - array( | |
| 6721 | - 'vector_id' => $vector_id, | |
| 6722 | - 'role_restriction' => $highest_role, | |
| 6723 | - 'updated_at' => current_time('mysql') | |
| 6724 | - ), | |
| 6725 | - array('%s', '%s', '%s') | |
| 6726 | - ); | |
| 6727 | - } else { | |
| 6728 | - // Update WordPress DB | |
| 6729 | - $table_name = $wpdb->prefix . 'mxchat_system_prompt_content'; | |
| 6730 | - | |
| 6731 | - $wpdb->update( | |
| 6732 | - $table_name, | |
| 6733 | - array('role_restriction' => $highest_role), | |
| 6734 | - array('source_url' => $source_url), | |
| 6735 | - array('%s'), | |
| 6736 | - array('%s') | |
| 6737 | - ); | |
| 6738 | - } | |
| 6739 | -} | |
| 6740 | - | |
| 6741 | - | |
| 6742 | - // ======================================== | |
| 6743 | - // HELPER METHODS | |
| 6744 | - // ======================================== | |
| 6745 | - | |
| 6746 | - /** | |
| 6747 | - * Check if user has required permissions for content processing | |
| 6748 | - */ | |
| 6749 | - private function mxchat_check_user_permissions() { | |
| 6750 | - if (!current_user_can('manage_options')) { | |
| 6751 | - wp_die(esc_html__('You do not have sufficient permissions.', 'mxchat')); | |
| 6752 | - } | |
| 6753 | - } | |
| 6754 | - | |
| 6755 | - /** | |
| 6756 | - * Validate nonce for security | |
| 6757 | - */ | |
| 6758 | - private function mxchat_validate_nonce($nonce_name, $nonce_action) { | |
| 6759 | - if (!isset($_POST[$nonce_name]) || !wp_verify_nonce($_POST[$nonce_name], $nonce_action)) { | |
| 6760 | - wp_die(esc_html__('Security check failed.', 'mxchat')); | |
| 6761 | - } | |
| 6762 | - } | |
| 6763 | - | |
| 6764 | - /** | |
| 6765 | - * Get embedding API credentials | |
| 6766 | - */ | |
| 6767 | - private function mxchat_get_embedding_credentials() { | |
| 6768 | - $embedding_model = $this->options['embedding_model'] ?? 'text-embedding-ada-002'; | |
| 6769 | - | |
| 6770 | - if (strpos($embedding_model, 'text-embedding-') !== false) { | |
| 6771 | - return array( | |
| 6772 | - 'type' => 'openai', | |
| 6773 | - 'api_key' => $this->options['api_key'] ?? '' | |
| 6774 | - ); | |
| 6775 | - } elseif (strpos($embedding_model, 'voyage-') !== false) { | |
| 6776 | - return array( | |
| 6777 | - 'type' => 'voyage', | |
| 6778 | - 'api_key' => $this->options['voyage_api_key'] ?? '' | |
| 6779 | - ); | |
| 6780 | - } elseif (strpos($embedding_model, 'gemini-embedding-') !== false) { | |
| 6781 | - return array( | |
| 6782 | - 'type' => 'gemini', | |
| 6783 | - 'api_key' => $this->options['gemini_api_key'] ?? '' | |
| 6784 | - ); | |
| 6785 | - } | |
| 6786 | - | |
| 6787 | - return array('type' => 'unknown', 'api_key' => ''); | |
| 6788 | - } | |
| 6789 | - | |
| 6790 | - /** | |
| 6791 | - * Log processing errors | |
| 6792 | - */ | |
| 6793 | - private function mxchat_log_processing_error($operation, $error_message) { | |
| 6794 | - //error_log("MxChat Knowledge Processing {$operation} Error: " . $error_message); | |
| 6795 | - } | |
| 6796 | - | |
| 6797 | - /** | |
| 6798 | - * Set admin notice transient | |
| 6799 | - */ | |
| 6800 | - private function mxchat_set_admin_notice($type, $message) { | |
| 6801 | - set_transient("mxchat_admin_notice_{$type}", $message, 30); | |
| 6802 | - } | |
| 6803 | - | |
| 6804 | - /** | |
| 6805 | - * Get Pinecone manager instance for vector operations | |
| 6806 | - */ | |
| 6807 | - private function mxchat_get_pinecone_manager() { | |
| 6808 | - return MxChat_Pinecone_Manager::get_instance(); | |
| 6809 | - } | |
| 6810 | - | |
| 6811 | - | |
| 6812 | - // ======================================== | |
| 6813 | -// DATABASE QUEUE TABLE MANAGEMENT | |
| 6814 | -// ======================================== | |
| 6815 | - | |
| 6816 | -/** | |
| 6817 | - * Create queue table on plugin activation | |
| 6818 | - * Call this from your plugin activation hook | |
| 6819 | - */ | |
| 6820 | -public function mxchat_create_queue_table() { | |
| 6821 | - global $wpdb; | |
| 6822 | - | |
| 6823 | - $table_name = $wpdb->prefix . 'mxchat_processing_queue'; | |
| 6824 | - $charset_collate = $wpdb->get_charset_collate(); | |
| 6825 | - | |
| 6826 | - $sql = "CREATE TABLE IF NOT EXISTS $table_name ( | |
| 6827 | - id bigint(20) unsigned NOT NULL AUTO_INCREMENT, | |
| 6828 | - queue_id varchar(64) NOT NULL, | |
| 6829 | - item_type varchar(20) NOT NULL, | |
| 6830 | - item_data longtext NOT NULL, | |
| 6831 | - status varchar(20) NOT NULL DEFAULT 'pending', | |
| 6832 | - bot_id varchar(50) NOT NULL DEFAULT 'default', | |
| 6833 | - priority int(11) NOT NULL DEFAULT 0, | |
| 6834 | - attempts int(11) NOT NULL DEFAULT 0, | |
| 6835 | - max_attempts int(11) NOT NULL DEFAULT 3, | |
| 6836 | - error_message text DEFAULT NULL, | |
| 6837 | - created_at datetime NOT NULL, | |
| 6838 | - started_at datetime DEFAULT NULL, | |
| 6839 | - completed_at datetime DEFAULT NULL, | |
| 6840 | - PRIMARY KEY (id), | |
| 6841 | - KEY queue_id (queue_id), | |
| 6842 | - KEY status (status), | |
| 6843 | - KEY item_type (item_type), | |
| 6844 | - KEY priority (priority) | |
| 6845 | - ) $charset_collate;"; | |
| 6846 | - | |
| 6847 | - require_once(ABSPATH . 'wp-admin/includes/upgrade.php'); | |
| 6848 | - dbDelta($sql); | |
| 6849 | - | |
| 6850 | - // Also create a meta table for queue metadata | |
| 6851 | - $meta_table = $wpdb->prefix . 'mxchat_queue_meta'; | |
| 6852 | - | |
| 6853 | - $meta_sql = "CREATE TABLE IF NOT EXISTS $meta_table ( | |
| 6854 | - id bigint(20) unsigned NOT NULL AUTO_INCREMENT, | |
| 6855 | - queue_id varchar(64) NOT NULL, | |
| 6856 | - meta_key varchar(255) NOT NULL, | |
| 6857 | - meta_value longtext, | |
| 6858 | - PRIMARY KEY (id), | |
| 6859 | - KEY queue_id (queue_id), | |
| 6860 | - KEY meta_key (meta_key) | |
| 6861 | - ) $charset_collate;"; | |
| 6862 | - | |
| 6863 | - dbDelta($meta_sql); | |
| 6864 | -} | |
| 6865 | - | |
| 6866 | -/** | |
| 6867 | - * Add items to the processing queue | |
| 6868 | - * | |
| 6869 | - * @param string $queue_id Unique identifier for this queue batch | |
| 6870 | - * @param string $item_type Type of item (url, pdf_page) | |
| 6871 | - * @param array $items Array of items to queue | |
| 6872 | - * @param string $bot_id Bot ID for processing | |
| 6873 | - * @return int Number of items queued | |
| 6874 | - */ | |
| 6875 | -private function mxchat_add_to_queue($queue_id, $item_type, $items, $bot_id = 'default') { | |
| 6876 | - global $wpdb; | |
| 6877 | - $table_name = $wpdb->prefix . 'mxchat_processing_queue'; | |
| 6878 | - | |
| 6879 | - $queued_count = 0; | |
| 6880 | - $priority = 0; | |
| 6881 | - | |
| 6882 | - foreach ($items as $item) { | |
| 6883 | - $result = $wpdb->insert( | |
| 6884 | - $table_name, | |
| 6885 | - array( | |
| 6886 | - 'queue_id' => $queue_id, | |
| 6887 | - 'item_type' => $item_type, | |
| 6888 | - 'item_data' => wp_json_encode($item), | |
| 6889 | - 'status' => 'pending', | |
| 6890 | - 'bot_id' => $bot_id, | |
| 6891 | - 'priority' => $priority, | |
| 6892 | - 'attempts' => 0, | |
| 6893 | - 'max_attempts' => 3, | |
| 6894 | - 'created_at' => current_time('mysql') | |
| 6895 | - ), | |
| 6896 | - array('%s', '%s', '%s', '%s', '%s', '%d', '%d', '%d', '%s') | |
| 6897 | - ); | |
| 6898 | - | |
| 6899 | - if ($result) { | |
| 6900 | - $queued_count++; | |
| 6901 | - } | |
| 6902 | - | |
| 6903 | - $priority++; // Process in order | |
| 6904 | - } | |
| 6905 | - | |
| 6906 | - return $queued_count; | |
| 6907 | -} | |
| 6908 | - | |
| 6909 | -/** | |
| 6910 | - * Store queue metadata (total counts, source URL, etc.) | |
| 6911 | - */ | |
| 6912 | -private function mxchat_set_queue_meta($queue_id, $meta_key, $meta_value) { | |
| 6913 | - global $wpdb; | |
| 6914 | - $meta_table = $wpdb->prefix . 'mxchat_queue_meta'; | |
| 6915 | - | |
| 6916 | - // Check if meta exists | |
| 6917 | - $existing = $wpdb->get_var($wpdb->prepare( | |
| 6918 | - "SELECT id FROM $meta_table WHERE queue_id = %s AND meta_key = %s", | |
| 6919 | - $queue_id, | |
| 6920 | - $meta_key | |
| 6921 | - )); | |
| 6922 | - | |
| 6923 | - if ($existing) { | |
| 6924 | - // Update | |
| 6925 | - $wpdb->update( | |
| 6926 | - $meta_table, | |
| 6927 | - array('meta_value' => maybe_serialize($meta_value)), | |
| 6928 | - array('queue_id' => $queue_id, 'meta_key' => $meta_key), | |
| 6929 | - array('%s'), | |
| 6930 | - array('%s', '%s') | |
| 6931 | - ); | |
| 6932 | - } else { | |
| 6933 | - // Insert | |
| 6934 | - $wpdb->insert( | |
| 6935 | - $meta_table, | |
| 6936 | - array( | |
| 6937 | - 'queue_id' => $queue_id, | |
| 6938 | - 'meta_key' => $meta_key, | |
| 6939 | - 'meta_value' => maybe_serialize($meta_value) | |
| 6940 | - ), | |
| 6941 | - array('%s', '%s', '%s') | |
| 6942 | - ); | |
| 6943 | - } | |
| 6944 | -} | |
| 6945 | - | |
| 6946 | -/** | |
| 6947 | - * Get queue metadata | |
| 6948 | - */ | |
| 6949 | -private function mxchat_get_queue_meta($queue_id, $meta_key) { | |
| 6950 | - global $wpdb; | |
| 6951 | - $meta_table = $wpdb->prefix . 'mxchat_queue_meta'; | |
| 6952 | - | |
| 6953 | - $value = $wpdb->get_var($wpdb->prepare( | |
| 6954 | - "SELECT meta_value FROM $meta_table WHERE queue_id = %s AND meta_key = %s", | |
| 6955 | - $queue_id, | |
| 6956 | - $meta_key | |
| 6957 | - )); | |
| 6958 | - | |
| 6959 | - return maybe_unserialize($value); | |
| 6960 | -} | |
| 6961 | - | |
| 6962 | -// ======================================== | |
| 6963 | -// AJAX QUEUE PROCESSING HANDLERS | |
| 6964 | -// ======================================== | |
| 6965 | - | |
| 6966 | -/** | |
| 6967 | - * AJAX: Get next item from queue to process | |
| 6968 | - */ | |
| 6969 | -public function ajax_mxchat_get_next_queue_item() { | |
| 6970 | - // Verify nonce and permissions | |
| 6971 | - check_ajax_referer('mxchat_queue_nonce', 'nonce'); | |
| 6972 | - | |
| 6973 | - if (!current_user_can('manage_options')) { | |
| 6974 | - wp_send_json_error('Unauthorized access'); | |
| 6975 | - } | |
| 6976 | - | |
| 6977 | - $queue_id = isset($_POST['queue_id']) ? sanitize_text_field($_POST['queue_id']) : ''; | |
| 6978 | - | |
| 6979 | - if (empty($queue_id)) { | |
| 6980 | - wp_send_json_error('Missing queue ID'); | |
| 6981 | - } | |
| 6982 | - | |
| 6983 | - global $wpdb; | |
| 6984 | - $table_name = $wpdb->prefix . 'mxchat_processing_queue'; | |
| 6985 | - | |
| 6986 | - // Get next pending item with retry logic for failed items | |
| 6987 | - $next_item = $wpdb->get_row($wpdb->prepare( | |
| 6988 | - "SELECT * FROM $table_name | |
| 6989 | - WHERE queue_id = %s | |
| 6990 | - AND status IN ('pending', 'failed') | |
| 6991 | - AND attempts < max_attempts | |
| 6992 | - ORDER BY priority ASC, id ASC | |
| 6993 | - LIMIT 1", | |
| 6994 | - $queue_id | |
| 6995 | - )); | |
| 6996 | - | |
| 6997 | - if (!$next_item) { | |
| 6998 | - // No more items - queue complete | |
| 6999 | - wp_send_json_success(array( | |
| 7000 | - 'complete' => true, | |
| 7001 | - 'message' => 'Queue processing complete' | |
| 7002 | - )); | |
| 7003 | - } | |
| 7004 | - | |
| 7005 | - // Mark item as processing | |
| 7006 | - $wpdb->update( | |
| 7007 | - $table_name, | |
| 7008 | - array( | |
| 7009 | - 'status' => 'processing', | |
| 7010 | - 'started_at' => current_time('mysql'), | |
| 7011 | - 'attempts' => $next_item->attempts + 1 | |
| 7012 | - ), | |
| 7013 | - array('id' => $next_item->id), | |
| 7014 | - array('%s', '%s', '%d'), | |
| 7015 | - array('%d') | |
| 7016 | - ); | |
| 7017 | - | |
| 7018 | - wp_send_json_success(array( | |
| 7019 | - 'complete' => false, | |
| 7020 | - 'item' => array( | |
| 7021 | - 'id' => $next_item->id, | |
| 7022 | - 'type' => $next_item->item_type, | |
| 7023 | - 'data' => json_decode($next_item->item_data, true), | |
| 7024 | - 'bot_id' => $next_item->bot_id, | |
| 7025 | - 'attempt' => $next_item->attempts + 1 | |
| 7026 | - ) | |
| 7027 | - )); | |
| 7028 | -} | |
| 7029 | - | |
| 7030 | -/** | |
| 7031 | - * AJAX: Process a single queue item | |
| 7032 | - */ | |
| 7033 | -public function ajax_mxchat_process_queue_item() { | |
| 7034 | - // Verify nonce and permissions | |
| 7035 | - check_ajax_referer('mxchat_queue_nonce', 'nonce'); | |
| 7036 | - | |
| 7037 | - if (!current_user_can('manage_options')) { | |
| 7038 | - wp_send_json_error('Unauthorized access'); | |
| 7039 | - } | |
| 7040 | - | |
| 7041 | - $item_id = isset($_POST['item_id']) ? absint($_POST['item_id']) : 0; | |
| 7042 | - $item_type = isset($_POST['item_type']) ? sanitize_text_field($_POST['item_type']) : ''; | |
| 7043 | - $item_data = isset($_POST['item_data']) ? $_POST['item_data'] : array(); | |
| 7044 | - $bot_id = isset($_POST['bot_id']) ? sanitize_key($_POST['bot_id']) : 'default'; | |
| 7045 | - | |
| 7046 | - if (empty($item_id) || empty($item_type)) { | |
| 7047 | - wp_send_json_error('Missing item data'); | |
| 7048 | - } | |
| 7049 | - | |
| 7050 | - global $wpdb; | |
| 7051 | - $table_name = $wpdb->prefix . 'mxchat_processing_queue'; | |
| 7052 | - | |
| 7053 | - // Process based on item type | |
| 7054 | - try { | |
| 7055 | - set_time_limit(60); // Give processing 60 seconds | |
| 7056 | - | |
| 7057 | - $result = false; | |
| 7058 | - $error_message = ''; | |
| 7059 | - | |
| 7060 | - // Read item directly from DB to get queue_id and preserve special chars in item_data | |
| 7061 | - // (POST round-trip through JS mangles characters like apostrophes in URLs) | |
| 7062 | - $db_item = $wpdb->get_row($wpdb->prepare( | |
| 7063 | - "SELECT queue_id, item_data FROM $table_name WHERE id = %d", | |
| 7064 | - $item_id | |
| 7065 | - )); | |
| 7066 | - $item_queue_id = $db_item ? $db_item->queue_id : ''; | |
| 7067 | - if ($db_item && !empty($db_item->item_data)) { | |
| 7068 | - $db_data = json_decode($db_item->item_data, true); | |
| 7069 | - if (is_array($db_data)) { | |
| 7070 | - $item_data = $db_data; | |
| 7071 | - } | |
| 7072 | - } | |
| 7073 | - | |
| 7074 | - switch ($item_type) { | |
| 7075 | - case 'url': | |
| 7076 | - $result = $this->mxchat_process_queue_url($item_data, $bot_id, $item_queue_id); | |
| 7077 | - break; | |
| 7078 | - | |
| 7079 | - case 'pdf_page': | |
| 7080 | - $result = $this->mxchat_process_queue_pdf_page($item_data, $bot_id); | |
| 7081 | - break; | |
| 7082 | - | |
| 7083 | - default: | |
| 7084 | - throw new Exception('Unknown item type: ' . $item_type); | |
| 7085 | - } | |
| 7086 | - | |
| 7087 | - if (is_wp_error($result)) { | |
| 7088 | - $error_code = $result->get_error_code(); | |
| 7089 | - // Content errors (empty page, sanitization) are permanent — retrying won't help | |
| 7090 | - $permanent_codes = array('empty_page', 'empty_after_sanitization', 'no_api_key', 'page_not_found'); | |
| 7091 | - if (in_array($error_code, $permanent_codes)) { | |
| 7092 | - // Mark as permanently failed — set attempts = max_attempts so it won't be retried | |
| 7093 | - $current_item = $wpdb->get_row($wpdb->prepare( | |
| 7094 | - "SELECT max_attempts FROM $table_name WHERE id = %d", $item_id | |
| 7095 | - )); | |
| 7096 | - $wpdb->update( | |
| 7097 | - $table_name, | |
| 7098 | - array( | |
| 7099 | - 'status' => 'failed', | |
| 7100 | - 'error_message' => $result->get_error_message(), | |
| 7101 | - 'attempts' => $current_item ? $current_item->max_attempts : 3 | |
| 7102 | - ), | |
| 7103 | - array('id' => $item_id), | |
| 7104 | - array('%s', '%s', '%d'), | |
| 7105 | - array('%d') | |
| 7106 | - ); | |
| 7107 | - wp_send_json_error(array( | |
| 7108 | - 'message' => $result->get_error_message(), | |
| 7109 | - 'permanent_failure' => true, | |
| 7110 | - 'item_id' => $item_id | |
| 7111 | - )); | |
| 7112 | - return; | |
| 7113 | - } | |
| 7114 | - throw new Exception($result->get_error_message()); | |
| 7115 | - } | |
| 7116 | - | |
| 7117 | - if ($result === false) { | |
| 7118 | - throw new Exception('Processing returned false - item may be empty or invalid'); | |
| 7119 | - } | |
| 7120 | - | |
| 7121 | - // Mark as completed | |
| 7122 | - $wpdb->update( | |
| 7123 | - $table_name, | |
| 7124 | - array( | |
| 7125 | - 'status' => 'completed', | |
| 7126 | - 'completed_at' => current_time('mysql'), | |
| 7127 | - 'error_message' => null | |
| 7128 | - ), | |
| 7129 | - array('id' => $item_id), | |
| 7130 | - array('%s', '%s', '%s'), | |
| 7131 | - array('%d') | |
| 7132 | - ); | |
| 7133 | - | |
| 7134 | - wp_send_json_success(array( | |
| 7135 | - 'processed' => true, | |
| 7136 | - 'item_id' => $item_id, | |
| 7137 | - 'message' => 'Item processed successfully' | |
| 7138 | - )); | |
| 7139 | - | |
| 7140 | - } catch (Exception $e) { | |
| 7141 | - $error_message = $e->getMessage(); | |
| 7142 | - | |
| 7143 | - // Get current attempt count | |
| 7144 | - $item = $wpdb->get_row($wpdb->prepare( | |
| 7145 | - "SELECT attempts, max_attempts FROM $table_name WHERE id = %d", | |
| 7146 | - $item_id | |
| 7147 | - )); | |
| 7148 | - | |
| 7149 | - // Check if we've exhausted retries | |
| 7150 | - if ($item && $item->attempts >= $item->max_attempts) { | |
| 7151 | - // Permanently failed | |
| 7152 | - $wpdb->update( | |
| 7153 | - $table_name, | |
| 7154 | - array( | |
| 7155 | - 'status' => 'failed', | |
| 7156 | - 'error_message' => $error_message | |
| 7157 | - ), | |
| 7158 | - array('id' => $item_id), | |
| 7159 | - array('%s', '%s'), | |
| 7160 | - array('%d') | |
| 7161 | - ); | |
| 7162 | - | |
| 7163 | - wp_send_json_error(array( | |
| 7164 | - 'message' => 'Item failed after maximum attempts: ' . $error_message, | |
| 7165 | - 'permanent_failure' => true, | |
| 7166 | - 'item_id' => $item_id | |
| 7167 | - )); | |
| 7168 | - } else { | |
| 7169 | - // Mark for retry | |
| 7170 | - $wpdb->update( | |
| 7171 | - $table_name, | |
| 7172 | - array( | |
| 7173 | - 'status' => 'failed', | |
| 7174 | - 'error_message' => $error_message | |
| 7175 | - ), | |
| 7176 | - array('id' => $item_id), | |
| 7177 | - array('%s', '%s'), | |
| 7178 | - array('%d') | |
| 7179 | - ); | |
| 7180 | - | |
| 7181 | - wp_send_json_error(array( | |
| 7182 | - 'message' => 'Item processing failed, will retry: ' . $error_message, | |
| 7183 | - 'can_retry' => true, | |
| 7184 | - 'item_id' => $item_id, | |
| 7185 | - 'attempts' => $item ? $item->attempts : 0 | |
| 7186 | - )); | |
| 7187 | - } | |
| 7188 | - } | |
| 7189 | -} | |
| 7190 | - | |
| 7191 | -/** | |
| 7192 | - * Process a URL from the queue | |
| 7193 | - */ | |
| 7194 | -private function mxchat_process_queue_url($item_data, $bot_id = 'default', $queue_id = '') { | |
| 7195 | - $url = isset($item_data['url']) ? $item_data['url'] : ''; | |
| 7196 | - | |
| 7197 | - if (empty($url)) { | |
| 7198 | - return new WP_Error('invalid_url', 'URL is empty'); | |
| 7199 | - } | |
| 7200 | - | |
| 7201 | - // Get bot-specific API key early (needed for both paths) | |
| 7202 | - $bot_options = $this->get_bot_options($bot_id); | |
| 7203 | - $options = !empty($bot_options) ? $bot_options : get_option('mxchat_options'); | |
| 7204 | - $selected_model = $options['embedding_model'] ?? 'text-embedding-ada-002'; | |
| 7205 | - | |
| 7206 | - if (strpos($selected_model, 'voyage') === 0) { | |
| 7207 | - $api_key = $options['voyage_api_key'] ?? ''; | |
| 7208 | - } elseif (strpos($selected_model, 'gemini-embedding') === 0) { | |
| 7209 | - $api_key = $options['gemini_api_key'] ?? ''; | |
| 7210 | - } else { | |
| 7211 | - $api_key = $options['api_key'] ?? ''; | |
| 7212 | - } | |
| 7213 | - | |
| 7214 | - if (empty($api_key)) { | |
| 7215 | - return new WP_Error('no_api_key', 'API key not configured for bot: ' . $bot_id); | |
| 7216 | - } | |
| 7217 | - | |
| 7218 | - // Check if this is a WooCommerce product URL and WooCommerce is active | |
| 7219 | - $is_product_url = (strpos($url, '/product/') !== false || strpos($url, '/shop/') !== false); | |
| 7220 | - $content_type = $is_product_url ? 'product' : 'url'; | |
| 7221 | - | |
| 7222 | - // Try to get WooCommerce product data if it's a product URL | |
| 7223 | - if ($is_product_url && class_exists('WooCommerce')) { | |
| 7224 | - $product_content = $this->mxchat_extract_woocommerce_product_content($url); | |
| 7225 | - | |
| 7226 | - if (!empty($product_content)) { | |
| 7227 | - // Successfully extracted WooCommerce product data with pricing | |
| 7228 | - $result = MxChat_Utils::submit_content_to_db( | |
| 7229 | - $product_content, | |
| 7230 | - $url, | |
| 7231 | - $api_key, | |
| 7232 | - null, | |
| 7233 | - $bot_id, | |
| 7234 | - 'product' | |
| 7235 | - ); | |
| 7236 | - return $result; | |
| 7237 | - } | |
| 7238 | - // If WooCommerce extraction failed, fall through to HTML extraction | |
| 7239 | - } | |
| 7240 | - | |
| 7241 | - // Fetch URL content (fallback for non-products or when WooCommerce extraction fails) | |
| 7242 | - $is_likely_pdf = (strtolower(pathinfo(parse_url($url, PHP_URL_PATH) ?: '', PATHINFO_EXTENSION)) === 'pdf'); | |
| 7243 | - $response = wp_remote_get($url, array( | |
| 7244 | - 'timeout' => $is_likely_pdf ? 120 : 30, | |
| 7245 | - 'redirection' => 5, | |
| 7246 | - 'user-agent' => 'MxChat/1.0' | |
| 7247 | - )); | |
| 7248 | - | |
| 7249 | - if (is_wp_error($response)) { | |
| 7250 | - return $response; | |
| 7251 | - } | |
| 7252 | - | |
| 7253 | - $response_code = wp_remote_retrieve_response_code($response); | |
| 7254 | - if ($response_code !== 200) { | |
| 7255 | - return new WP_Error('http_error', 'HTTP ' . $response_code . ' error for: ' . $url); | |
| 7256 | - } | |
| 7257 | - | |
| 7258 | - // Check if URL is a PDF — expand into per-page queue items using the standard PDF pipeline | |
| 7259 | - if ($this->mxchat_is_pdf_url($url, $response)) { | |
| 7260 | - return $this->mxchat_expand_pdf_to_queue($url, $response, $bot_id, $queue_id); | |
| 7261 | - } | |
| 7262 | - | |
| 7263 | - $html = wp_remote_retrieve_body($response); | |
| 7264 | - | |
| 7265 | - if (empty($html)) { | |
| 7266 | - return new WP_Error('empty_response', 'Empty response body'); | |
| 7267 | - } | |
| 7268 | - | |
| 7269 | - // Extract and sanitize content | |
| 7270 | - $content = $this->mxchat_extract_main_content($html); | |
| 7271 | - $sanitized = $this->mxchat_sanitize_content_for_api($content); | |
| 7272 | - | |
| 7273 | - if (empty($sanitized)) { | |
| 7274 | - // Not an error - just no content found (maybe a redirect or empty page) | |
| 7275 | - return false; | |
| 7276 | - } | |
| 7277 | - | |
| 7278 | - // Submit to database with content_type | |
| 7279 | - $result = MxChat_Utils::submit_content_to_db( | |
| 7280 | - $sanitized, | |
| 7281 | - $url, | |
| 7282 | - $api_key, | |
| 7283 | - null, | |
| 7284 | - $bot_id, | |
| 7285 | - $content_type | |
| 7286 | - ); | |
| 7287 | - | |
| 7288 | - return $result; | |
| 7289 | -} | |
| 7290 | - | |
| 7291 | -/** | |
| 7292 | - * Expand a PDF URL into per-page queue items using the standard PDF pipeline. | |
| 7293 | - * Called when a sitemap URL turns out to be a PDF — downloads, parses page count, | |
| 7294 | - * and adds pdf_page items to the same queue so they process with full progress tracking. | |
| 7295 | - */ | |
| 7296 | -private function mxchat_expand_pdf_to_queue($pdf_url, $response, $bot_id = 'default', $queue_id = '') { | |
| 7297 | - set_time_limit(120); // PDFs need extra time for download + parsing | |
| 7298 | - | |
| 7299 | - $upload_dir = wp_upload_dir(); | |
| 7300 | - $pdf_filename = sanitize_file_name('mxchat_kb_' . md5($pdf_url) . '.pdf'); | |
| 7301 | - $pdf_path = trailingslashit($upload_dir['path']) . $pdf_filename; | |
| 7302 | - | |
| 7303 | - $response_body = wp_remote_retrieve_body($response); | |
| 7304 | - if (empty($response_body)) { | |
| 7305 | - return new WP_Error('empty_pdf', 'Empty PDF response for: ' . $pdf_url); | |
| 7306 | - } | |
| 7307 | - | |
| 7308 | - if (!wp_mkdir_p(dirname($pdf_path))) { | |
| 7309 | - return new WP_Error('dir_error', 'Failed to create upload directory'); | |
| 7310 | - } | |
| 7311 | - | |
| 7312 | - file_put_contents($pdf_path, $response_body); | |
| 7313 | - | |
| 7314 | - if (!file_exists($pdf_path)) { | |
| 7315 | - return new WP_Error('save_error', 'Failed to save PDF file'); | |
| 7316 | - } | |
| 7317 | - | |
| 7318 | - try { | |
| 7319 | - $total_pages = $this->mxchat_validate_and_count_pdf_pages($pdf_path); | |
| 7320 | - | |
| 7321 | - if ($total_pages === false || $total_pages < 1) { | |
| 7322 | - wp_delete_file($pdf_path); | |
| 7323 | - return new WP_Error('no_pages', 'PDF has no pages: ' . $pdf_url); | |
| 7324 | - } | |
| 7325 | - | |
| 7326 | - // Build per-page items identical to mxchat_handle_pdf_for_knowledge_base | |
| 7327 | - $pages = array(); | |
| 7328 | - for ($i = 1; $i <= $total_pages; $i++) { | |
| 7329 | - $pages[] = array( | |
| 7330 | - 'pdf_path' => $pdf_path, | |
| 7331 | - 'pdf_url' => $pdf_url, | |
| 7332 | - 'page_number' => $i, | |
| 7333 | - 'total_pages' => $total_pages | |
| 7334 | - ); | |
| 7335 | - } | |
| 7336 | - | |
| 7337 | - // Add pdf_page items to the SAME queue so the JS picks them up automatically | |
| 7338 | - if (!empty($queue_id)) { | |
| 7339 | - $queued_count = $this->mxchat_add_to_queue($queue_id, 'pdf_page', $pages, $bot_id); | |
| 7340 | - } else { | |
| 7341 | - // Fallback: create a new PDF queue (shouldn't happen in sitemap flow) | |
| 7342 | - $new_queue_id = 'pdf_' . md5($pdf_url . time()); | |
| 7343 | - $queued_count = $this->mxchat_add_to_queue($new_queue_id, 'pdf_page', $pages, $bot_id); | |
| 7344 | - $this->mxchat_set_queue_meta($new_queue_id, 'source_url', $pdf_url); | |
| 7345 | - $this->mxchat_set_queue_meta($new_queue_id, 'queue_type', 'pdf'); | |
| 7346 | - $this->mxchat_set_queue_meta($new_queue_id, 'total_items', $total_pages); | |
| 7347 | - $this->mxchat_set_queue_meta($new_queue_id, 'bot_id', $bot_id); | |
| 7348 | - $this->mxchat_set_queue_meta($new_queue_id, 'pdf_path', $pdf_path); | |
| 7349 | - $this->mxchat_set_queue_meta($new_queue_id, 'created_at', current_time('mysql')); | |
| 7350 | - } | |
| 7351 | - | |
| 7352 | - if ($queued_count === 0) { | |
| 7353 | - wp_delete_file($pdf_path); | |
| 7354 | - return new WP_Error('queue_error', 'Failed to add PDF pages to queue'); | |
| 7355 | - } | |
| 7356 | - | |
| 7357 | - // Return true so the original URL item is marked complete | |
| 7358 | - // The new pdf_page items will be processed in subsequent batches | |
| 7359 | - return true; | |
| 7360 | - | |
| 7361 | - } catch (Exception $e) { | |
| 7362 | - if (file_exists($pdf_path)) { | |
| 7363 | - wp_delete_file($pdf_path); | |
| 7364 | - } | |
| 7365 | - return new WP_Error('pdf_parse_error', 'Error parsing PDF: ' . $e->getMessage()); | |
| 7366 | - } | |
| 7367 | -} | |
| 7368 | - | |
| 7369 | -/** | |
| 7370 | - * Legacy: Process a PDF URL inline during sitemap queue processing. | |
| 7371 | - * @deprecated Use mxchat_expand_pdf_to_queue instead — kept for reference only. | |
| 7372 | - */ | |
| 7373 | -private function mxchat_process_pdf_url_inline($pdf_url, $response, $api_key, $bot_id = 'default') { | |
| 7374 | - set_time_limit(120); // PDFs need more time — downloading + parsing all pages | |
| 7375 | - | |
| 7376 | - $upload_dir = wp_upload_dir(); | |
| 7377 | - $pdf_filename = sanitize_file_name('mxchat_kb_' . md5($pdf_url) . '.pdf'); | |
| 7378 | - $pdf_path = trailingslashit($upload_dir['path']) . $pdf_filename; | |
| 7379 | - | |
| 7380 | - $response_body = wp_remote_retrieve_body($response); | |
| 7381 | - if (empty($response_body)) { | |
| 7382 | - return new WP_Error('empty_pdf', 'Empty PDF response'); | |
| 7383 | - } | |
| 7384 | - | |
| 7385 | - if (!wp_mkdir_p(dirname($pdf_path))) { | |
| 7386 | - return new WP_Error('dir_error', 'Failed to create upload directory'); | |
| 7387 | - } | |
| 7388 | - | |
| 7389 | - file_put_contents($pdf_path, $response_body); | |
| 7390 | - | |
| 7391 | - if (!file_exists($pdf_path)) { | |
| 7392 | - return new WP_Error('save_error', 'Failed to save PDF file'); | |
| 7393 | - } | |
| 7394 | - | |
| 7395 | - try { | |
| 7396 | - mxchat_load_pdf_parser(); | |
| 7397 | - $parser = new \Smalot\PdfParser\Parser(); | |
| 7398 | - $pdf = $parser->parseFile($pdf_path); | |
| 7399 | - $pages = $pdf->getPages(); | |
| 7400 | - $total_pages = count($pages); | |
| 7401 | - | |
| 7402 | - if ($total_pages < 1) { | |
| 7403 | - wp_delete_file($pdf_path); | |
| 7404 | - return new WP_Error('no_pages', 'PDF has no pages'); | |
| 7405 | - } | |
| 7406 | - | |
| 7407 | - $processed = 0; | |
| 7408 | - $skipped_pages = array(); | |
| 7409 | - | |
| 7410 | - for ($i = 0; $i < $total_pages; $i++) { | |
| 7411 | - $page_num = $i + 1; | |
| 7412 | - $text = $pages[$i]->getText(); | |
| 7413 | - if (empty($text)) { | |
| 7414 | - $skipped_pages[] = 'Page ' . $page_num . ': No text could be extracted — page may contain only images, links, or non-standard encoding'; | |
| 7415 | - continue; | |
| 7416 | - } | |
| 7417 | - | |
| 7418 | - $sanitized = $this->mxchat_sanitize_content_for_api($text); | |
| 7419 | - if (empty($sanitized)) { | |
| 7420 | - $skipped_pages[] = 'Page ' . $page_num . ': Text was extracted but contained only special characters, control codes, or unsupported content'; | |
| 7421 | - continue; | |
| 7422 | - } | |
| 7423 | - | |
| 7424 | - $metadata = array( | |
| 7425 | - 'document_type' => 'pdf', | |
| 7426 | - 'total_pages' => $total_pages, | |
| 7427 | - 'current_page' => $page_num, | |
| 7428 | - 'source_url' => $pdf_url, | |
| 7429 | - ); | |
| 7430 | - | |
| 7431 | - $content_with_metadata = wp_json_encode($metadata) . "\n---\n" . $sanitized; | |
| 7432 | - $page_url = esc_url($pdf_url . '#page=' . $page_num); | |
| 7433 | - | |
| 7434 | - MxChat_Utils::submit_content_to_db( | |
| 7435 | - $content_with_metadata, | |
| 7436 | - $page_url, | |
| 7437 | - $api_key, | |
| 7438 | - null, | |
| 7439 | - $bot_id, | |
| 7440 | - 'pdf' | |
| 7441 | - ); | |
| 7442 | - | |
| 7443 | - $processed++; | |
| 7444 | - } | |
| 7445 | - | |
| 7446 | - // Clean up the temp PDF file | |
| 7447 | - wp_delete_file($pdf_path); | |
| 7448 | - | |
| 7449 | - if (!empty($skipped_pages)) { | |
| 7450 | - error_log('MxChat PDF: Skipped ' . count($skipped_pages) . ' of ' . $total_pages . ' pages: ' . implode('; ', $skipped_pages)); | |
| 7451 | - } | |
| 7452 | - | |
| 7453 | - return $processed > 0 ? true : false; | |
| 7454 | - | |
| 7455 | - } catch (Exception $e) { | |
| 7456 | - if (file_exists($pdf_path)) { | |
| 7457 | - wp_delete_file($pdf_path); | |
| 7458 | - } | |
| 7459 | - return new WP_Error('pdf_parse_error', 'Error parsing PDF: ' . $e->getMessage()); | |
| 7460 | - } | |
| 7461 | -} | |
| 7462 | - | |
| 7463 | -/** | |
| 7464 | - * Extract WooCommerce product content including pricing | |
| 7465 | - * | |
| 7466 | - * @param string $url The product URL | |
| 7467 | - * @return string|false Product content with pricing, or false if not found | |
| 7468 | - */ | |
| 7469 | -private function mxchat_extract_woocommerce_product_content($url) { | |
| 7470 | - // Try to get product ID from URL | |
| 7471 | - $product_id = url_to_postid($url); | |
| 7472 | - | |
| 7473 | - // If url_to_postid fails, try to extract from URL pattern | |
| 7474 | - if (!$product_id) { | |
| 7475 | - $product_slug = ''; | |
| 7476 | - | |
| 7477 | - // Handle pretty permalinks: /product/product-name/ | |
| 7478 | - if (preg_match('/\/product\/([^\/\?]+)/', $url, $matches)) { | |
| 7479 | - $product_slug = $matches[1]; | |
| 7480 | - } | |
| 7481 | - | |
| 7482 | - if (!empty($product_slug)) { | |
| 7483 | - $product_post = get_page_by_path($product_slug, OBJECT, 'product'); | |
| 7484 | - if ($product_post) { | |
| 7485 | - $product_id = $product_post->ID; | |
| 7486 | - } | |
| 7487 | - } | |
| 7488 | - } | |
| 7489 | - | |
| 7490 | - if (!$product_id) { | |
| 7491 | - return false; | |
| 7492 | - } | |
| 7493 | - | |
| 7494 | - // Get WooCommerce product object | |
| 7495 | - $product = wc_get_product($product_id); | |
| 7496 | - | |
| 7497 | - if (!$product) { | |
| 7498 | - return false; | |
| 7499 | - } | |
| 7500 | - | |
| 7501 | - // Build product content with pricing (similar to mxchat_store_product_embedding) | |
| 7502 | - $title = $product->get_name(); | |
| 7503 | - $description = $product->get_description(); | |
| 7504 | - $short_description = $product->get_short_description(); | |
| 7505 | - $sku = $product->get_sku(); | |
| 7506 | - | |
| 7507 | - // Get pricing information | |
| 7508 | - $regular_price = $product->get_regular_price(); | |
| 7509 | - $sale_price = $product->get_sale_price(); | |
| 7510 | - $price = $product->get_price(); // Current active price | |
| 7511 | - | |
| 7512 | - // Get currency symbol | |
| 7513 | - $currency_symbol = get_woocommerce_currency_symbol(); | |
| 7514 | - | |
| 7515 | - // Format content | |
| 7516 | - $content = $title . "\n\n"; | |
| 7517 | - | |
| 7518 | - if (!empty($short_description)) { | |
| 7519 | - $content .= "Short Description: " . wp_strip_all_tags($short_description) . "\n\n"; | |
| 7520 | - } | |
| 7521 | - | |
| 7522 | - if (!empty($description)) { | |
| 7523 | - $content .= wp_strip_all_tags($description) . "\n\n"; | |
| 7524 | - } | |
| 7525 | - | |
| 7526 | - // Add pricing information | |
| 7527 | - if (!empty($regular_price)) { | |
| 7528 | - $content .= "Price: " . $currency_symbol . $regular_price . "\n"; | |
| 7529 | - } elseif (!empty($price)) { | |
| 7530 | - $content .= "Price: " . $currency_symbol . $price . "\n"; | |
| 7531 | - } | |
| 7532 | - | |
| 7533 | - if (!empty($sale_price) && $sale_price !== $regular_price) { | |
| 7534 | - $content .= "Sale Price: " . $currency_symbol . $sale_price . "\n"; | |
| 7535 | - } | |
| 7536 | - | |
| 7537 | - // Handle variable products - show price range | |
| 7538 | - if ($product->is_type('variable')) { | |
| 7539 | - $min_price = $product->get_variation_price('min'); | |
| 7540 | - $max_price = $product->get_variation_price('max'); | |
| 7541 | - if ($min_price !== $max_price) { | |
| 7542 | - $content .= "Price Range: " . $currency_symbol . $min_price . " - " . $currency_symbol . $max_price . "\n"; | |
| 7543 | - } | |
| 7544 | - } | |
| 7545 | - | |
| 7546 | - if (!empty($sku)) { | |
| 7547 | - $content .= "SKU: " . $sku . "\n"; | |
| 7548 | - } | |
| 7549 | - | |
| 7550 | - // Get product categories | |
| 7551 | - $categories = wp_get_post_terms($product_id, 'product_cat', array('fields' => 'names')); | |
| 7552 | - if (!empty($categories) && !is_wp_error($categories)) { | |
| 7553 | - $content .= "Categories: " . implode(', ', $categories) . "\n"; | |
| 7554 | - } | |
| 7555 | - | |
| 7556 | - // Get Custom Product Tabs (supports "Custom Product Tabs for WooCommerce" by Code Parrots) | |
| 7557 | - $custom_tabs = get_post_meta($product_id, 'yikes_woo_products_tabs', true); | |
| 7558 | - if (!empty($custom_tabs) && is_array($custom_tabs)) { | |
| 7559 | - foreach ($custom_tabs as $tab) { | |
| 7560 | - $tab_title = isset($tab['title']) ? $tab['title'] : (isset($tab['tab_title']) ? $tab['tab_title'] : ''); | |
| 7561 | - $tab_content = isset($tab['content']) ? $tab['content'] : ''; | |
| 7562 | - | |
| 7563 | - if (!empty($tab_title) && !empty($tab_content)) { | |
| 7564 | - $content .= "\n" . $tab_title . ": " . wp_strip_all_tags($tab_content) . "\n"; | |
| 7565 | - } | |
| 7566 | - } | |
| 7567 | - } | |
| 7568 | - | |
| 7569 | - // Also check for reusable/saved tabs applied to this product | |
| 7570 | - $applied_saved_tabs = get_post_meta($product_id, 'yikes_woo_reusable_products_tabs_applied', true); | |
| 7571 | - if (!empty($applied_saved_tabs) && is_array($applied_saved_tabs)) { | |
| 7572 | - $saved_tabs = get_option('yikes_woo_reusable_products_tabs', array()); | |
| 7573 | - if (!empty($saved_tabs) && is_array($saved_tabs)) { | |
| 7574 | - foreach ($applied_saved_tabs as $saved_tab_id) { | |
| 7575 | - if (isset($saved_tabs[$saved_tab_id])) { | |
| 7576 | - $tab = $saved_tabs[$saved_tab_id]; | |
| 7577 | - $tab_title = isset($tab['title']) ? $tab['title'] : (isset($tab['tab_title']) ? $tab['tab_title'] : ''); | |
| 7578 | - $tab_content = isset($tab['content']) ? $tab['content'] : ''; | |
| 7579 | - | |
| 7580 | - if (!empty($tab_title) && !empty($tab_content)) { | |
| 7581 | - $content .= "\n" . $tab_title . ": " . wp_strip_all_tags($tab_content) . "\n"; | |
| 7582 | - } | |
| 7583 | - } | |
| 7584 | - } | |
| 7585 | - } | |
| 7586 | - } | |
| 7587 | - | |
| 7588 | - return $this->mxchat_sanitize_content_for_api($content); | |
| 7589 | -} | |
| 7590 | - | |
| 7591 | -/** | |
| 7592 | - * Process a PDF page from the queue | |
| 7593 | - */ | |
| 7594 | -private function mxchat_process_queue_pdf_page($item_data, $bot_id = 'default') { | |
| 7595 | - $pdf_path = isset($item_data['pdf_path']) ? $item_data['pdf_path'] : ''; | |
| 7596 | - $pdf_url = isset($item_data['pdf_url']) ? $item_data['pdf_url'] : ''; | |
| 7597 | - $page_number = isset($item_data['page_number']) ? absint($item_data['page_number']) : 0; | |
| 7598 | - $total_pages = isset($item_data['total_pages']) ? absint($item_data['total_pages']) : 0; | |
| 7599 | - | |
| 7600 | - if (empty($pdf_path) || !file_exists($pdf_path)) { | |
| 7601 | - return new WP_Error('pdf_not_found', 'PDF file not found: ' . $pdf_path); | |
| 7602 | - } | |
| 7603 | - | |
| 7604 | - if ($page_number < 1) { | |
| 7605 | - return new WP_Error('invalid_page', 'Invalid page number'); | |
| 7606 | - } | |
| 7607 | - | |
| 7608 | - try { | |
| 7609 | - mxchat_load_pdf_parser(); | |
| 7610 | - $parser = new \Smalot\PdfParser\Parser(); | |
| 7611 | - $pdf = $parser->parseFile($pdf_path); | |
| 7612 | - $pages = $pdf->getPages(); | |
| 7613 | - | |
| 7614 | - if (!isset($pages[$page_number - 1])) { | |
| 7615 | - return new WP_Error('page_not_found', 'Page ' . $page_number . ' not found in PDF'); | |
| 7616 | - } | |
| 7617 | - | |
| 7618 | - $text = $pages[$page_number - 1]->getText(); | |
| 7619 | - | |
| 7620 | - if (empty($text)) { | |
| 7621 | - return new WP_Error('empty_page', 'Page ' . $page_number . ': No text could be extracted — page may contain only images, links, or non-standard encoding'); | |
| 7622 | - } | |
| 7623 | - | |
| 7624 | - $sanitized = $this->mxchat_sanitize_content_for_api($text); | |
| 7625 | - | |
| 7626 | - if (empty($sanitized)) { | |
| 7627 | - return new WP_Error('empty_after_sanitization', 'Page ' . $page_number . ': Text was extracted but contained only special characters, control codes, or unsupported content that was removed during cleanup'); | |
| 7628 | - } | |
| 7629 | - | |
| 7630 | - // Create metadata | |
| 7631 | - $metadata = array( | |
| 7632 | - 'document_type' => 'pdf', | |
| 7633 | - 'total_pages' => $total_pages, | |
| 7634 | - 'current_page' => $page_number, | |
| 7635 | - 'source_url' => $pdf_url | |
| 7636 | - ); | |
| 7637 | - | |
| 7638 | - $content_with_metadata = wp_json_encode($metadata) . "\n---\n" . $sanitized; | |
| 7639 | - $page_url = esc_url($pdf_url . "#page=" . $page_number); | |
| 7640 | - | |
| 7641 | - // Get bot-specific API key | |
| 7642 | - $bot_options = $this->get_bot_options($bot_id); | |
| 7643 | - $options = !empty($bot_options) ? $bot_options : get_option('mxchat_options'); | |
| 7644 | - $selected_model = $options['embedding_model'] ?? 'text-embedding-ada-002'; | |
| 7645 | - | |
| 7646 | - if (strpos($selected_model, 'voyage') === 0) { | |
| 7647 | - $api_key = $options['voyage_api_key'] ?? ''; | |
| 7648 | - } elseif (strpos($selected_model, 'gemini-embedding') === 0) { | |
| 7649 | - $api_key = $options['gemini_api_key'] ?? ''; | |
| 7650 | - } else { | |
| 7651 | - $api_key = $options['api_key'] ?? ''; | |
| 7652 | - } | |
| 7653 | - | |
| 7654 | - if (empty($api_key)) { | |
| 7655 | - return new WP_Error('no_api_key', 'API key not configured for bot: ' . $bot_id); | |
| 7656 | - } | |
| 7657 | - | |
| 7658 | - // Submit to database - UPDATED 2.5.6: Added content_type 'pdf' | |
| 7659 | - $result = MxChat_Utils::submit_content_to_db( | |
| 7660 | - $content_with_metadata, | |
| 7661 | - $page_url, | |
| 7662 | - $api_key, | |
| 7663 | - null, | |
| 7664 | - $bot_id, | |
| 7665 | - 'pdf' | |
| 7666 | - ); | |
| 7667 | - | |
| 7668 | - return $result; | |
| 7669 | - | |
| 7670 | - } catch (Exception $e) { | |
| 7671 | - return new WP_Error('pdf_parse_error', 'Error parsing PDF: ' . $e->getMessage()); | |
| 7672 | - } | |
| 7673 | -} | |
| 7674 | - | |
| 7675 | -/** | |
| 7676 | - * AJAX: Get queue processing status | |
| 7677 | - */ | |
| 7678 | -public function ajax_mxchat_get_queue_status() { | |
| 7679 | - // Verify nonce and permissions | |
| 7680 | - check_ajax_referer('mxchat_queue_nonce', 'nonce'); | |
| 7681 | - | |
| 7682 | - if (!current_user_can('manage_options')) { | |
| 7683 | - wp_send_json_error('Unauthorized access'); | |
| 7684 | - } | |
| 7685 | - | |
| 7686 | - $queue_id = isset($_POST['queue_id']) ? sanitize_text_field($_POST['queue_id']) : ''; | |
| 7687 | - | |
| 7688 | - if (empty($queue_id)) { | |
| 7689 | - wp_send_json_error('Missing queue ID'); | |
| 7690 | - } | |
| 7691 | - | |
| 7692 | - global $wpdb; | |
| 7693 | - $table_name = $wpdb->prefix . 'mxchat_processing_queue'; | |
| 7694 | - | |
| 7695 | - // Get counts by status | |
| 7696 | - $counts = $wpdb->get_results($wpdb->prepare( | |
| 7697 | - "SELECT status, COUNT(*) as count | |
| 7698 | - FROM $table_name | |
| 7699 | - WHERE queue_id = %s | |
| 7700 | - GROUP BY status", | |
| 7701 | - $queue_id | |
| 7702 | - ), OBJECT_K); | |
| 7703 | - | |
| 7704 | - $total = 0; | |
| 7705 | - $completed = 0; | |
| 7706 | - $failed = 0; | |
| 7707 | - $processing = 0; | |
| 7708 | - $pending = 0; | |
| 7709 | - | |
| 7710 | - foreach ($counts as $status => $data) { | |
| 7711 | - $count = absint($data->count); | |
| 7712 | - $total += $count; | |
| 7713 | - | |
| 7714 | - switch ($status) { | |
| 7715 | - case 'completed': | |
| 7716 | - $completed = $count; | |
| 7717 | - break; | |
| 7718 | - case 'failed': | |
| 7719 | - $failed = $count; | |
| 7720 | - break; | |
| 7721 | - case 'processing': | |
| 7722 | - $processing = $count; | |
| 7723 | - break; | |
| 7724 | - case 'pending': | |
| 7725 | - $pending = $count; | |
| 7726 | - break; | |
| 7727 | - } | |
| 7728 | - } | |
| 7729 | - | |
| 7730 | - // Calculate percentage | |
| 7731 | - $percentage = $total > 0 ? round((($completed + $failed) / $total) * 100) : 0; | |
| 7732 | - | |
| 7733 | - // Get failed items details (include all failed items, not just those that exhausted retries) | |
| 7734 | - $failed_items = array(); | |
| 7735 | - if ($failed > 0) { | |
| 7736 | - $failed_items = $wpdb->get_results($wpdb->prepare( | |
| 7737 | - "SELECT item_type, item_data, error_message, attempts | |
| 7738 | - FROM $table_name | |
| 7739 | - WHERE queue_id = %s | |
| 7740 | - AND status = 'failed' | |
| 7741 | - ORDER BY id DESC | |
| 7742 | - LIMIT 50", | |
| 7743 | - $queue_id | |
| 7744 | - )); | |
| 7745 | - } | |
| 7746 | - | |
| 7747 | - // Get queue metadata | |
| 7748 | - $source_url = $this->mxchat_get_queue_meta($queue_id, 'source_url'); | |
| 7749 | - $queue_type = $this->mxchat_get_queue_meta($queue_id, 'queue_type'); | |
| 7750 | - | |
| 7751 | - // Determine if queue is complete | |
| 7752 | - $is_complete = ($pending === 0 && $processing === 0); | |
| 7753 | - | |
| 7754 | - wp_send_json_success(array( | |
| 7755 | - 'queue_id' => $queue_id, | |
| 7756 | - 'queue_type' => $queue_type, | |
| 7757 | - 'source_url' => $source_url, | |
| 7758 | - 'total' => $total, | |
| 7759 | - 'completed' => $completed, | |
| 7760 | - 'failed' => $failed, | |
| 7761 | - 'processing' => $processing, | |
| 7762 | - 'pending' => $pending, | |
| 7763 | - 'percentage' => $percentage, | |
| 7764 | - 'is_complete' => $is_complete, | |
| 7765 | - 'failed_items' => $failed_items, | |
| 7766 | - 'status' => $is_complete ? 'complete' : 'processing' | |
| 7767 | - )); | |
| 7768 | -} | |
| 7769 | - | |
| 7770 | -/** | |
| 7771 | - * AJAX: Clear completed queue | |
| 7772 | - */ | |
| 7773 | -public function ajax_mxchat_clear_queue() { | |
| 7774 | - // Verify nonce and permissions | |
| 7775 | - check_ajax_referer('mxchat_queue_nonce', 'nonce'); | |
| 7776 | - | |
| 7777 | - if (!current_user_can('manage_options')) { | |
| 7778 | - wp_send_json_error('Unauthorized access'); | |
| 7779 | - } | |
| 7780 | - | |
| 7781 | - $queue_id = isset($_POST['queue_id']) ? sanitize_text_field($_POST['queue_id']) : ''; | |
| 7782 | - | |
| 7783 | - if (empty($queue_id)) { | |
| 7784 | - wp_send_json_error('Missing queue ID'); | |
| 7785 | - } | |
| 7786 | - | |
| 7787 | - global $wpdb; | |
| 7788 | - $table_name = $wpdb->prefix . 'mxchat_processing_queue'; | |
| 7789 | - $meta_table = $wpdb->prefix . 'mxchat_queue_meta'; | |
| 7790 | - | |
| 7791 | - // Delete queue items | |
| 7792 | - $wpdb->delete( | |
| 7793 | - $table_name, | |
| 7794 | - array('queue_id' => $queue_id), | |
| 7795 | - array('%s') | |
| 7796 | - ); | |
| 7797 | - | |
| 7798 | - // Delete queue metadata | |
| 7799 | - $wpdb->delete( | |
| 7800 | - $meta_table, | |
| 7801 | - array('queue_id' => $queue_id), | |
| 7802 | - array('%s') | |
| 7803 | - ); | |
| 7804 | - | |
| 7805 | - wp_send_json_success(array( | |
| 7806 | - 'message' => 'Queue cleared successfully' | |
| 7807 | - )); | |
| 7808 | -} | |
| 7809 | - | |
| 7810 | -/** | |
| 7811 | - * AJAX: Retry failed items in queue | |
| 7812 | - */ | |
| 7813 | -public function ajax_mxchat_retry_failed() { | |
| 7814 | - // Verify nonce and permissions | |
| 7815 | - check_ajax_referer('mxchat_queue_nonce', 'nonce'); | |
| 7816 | - | |
| 7817 | - if (!current_user_can('manage_options')) { | |
| 7818 | - wp_send_json_error('Unauthorized access'); | |
| 7819 | - } | |
| 7820 | - | |
| 7821 | - $queue_id = isset($_POST['queue_id']) ? sanitize_text_field($_POST['queue_id']) : ''; | |
| 7822 | - | |
| 7823 | - if (empty($queue_id)) { | |
| 7824 | - wp_send_json_error('Missing queue ID'); | |
| 7825 | - } | |
| 7826 | - | |
| 7827 | - global $wpdb; | |
| 7828 | - $table_name = $wpdb->prefix . 'mxchat_processing_queue'; | |
| 7829 | - | |
| 7830 | - // Reset failed items to pending and reset attempt count | |
| 7831 | - $updated = $wpdb->update( | |
| 7832 | - $table_name, | |
| 7833 | - array( | |
| 7834 | - 'status' => 'pending', | |
| 7835 | - 'attempts' => 0, | |
| 7836 | - 'error_message' => null | |
| 7837 | - ), | |
| 7838 | - array( | |
| 7839 | - 'queue_id' => $queue_id, | |
| 7840 | - 'status' => 'failed' | |
| 7841 | - ), | |
| 7842 | - array('%s', '%d', '%s'), | |
| 7843 | - array('%s', '%s') | |
| 7844 | - ); | |
| 7845 | - | |
| 7846 | - wp_send_json_success(array( | |
| 7847 | - 'message' => 'Reset ' . $updated . ' failed items for retry', | |
| 7848 | - 'reset_count' => $updated | |
| 7849 | - )); | |
| 7850 | -} | |
| 7851 | - | |
| 7852 | - | |
| 7853 | -public function ajax_mxchat_mark_queue_complete() { | |
| 7854 | - check_ajax_referer('mxchat_queue_nonce', 'nonce'); | |
| 7855 | - | |
| 7856 | - if (!current_user_can('manage_options')) { | |
| 7857 | - wp_send_json_error('Unauthorized access'); | |
| 7858 | - } | |
| 7859 | - | |
| 7860 | - $queue_id = isset($_POST['queue_id']) ? sanitize_text_field($_POST['queue_id']) : ''; | |
| 7861 | - | |
| 7862 | - if (empty($queue_id)) { | |
| 7863 | - wp_send_json_error('Missing queue ID'); | |
| 7864 | - } | |
| 7865 | - | |
| 7866 | - // Clear active queue transients | |
| 7867 | - if (strpos($queue_id, 'sitemap_') === 0) { | |
| 7868 | - delete_transient('mxchat_active_queue_sitemap'); | |
| 7869 | - } else if (strpos($queue_id, 'pdf_') === 0) { | |
| 7870 | - delete_transient('mxchat_active_queue_pdf'); | |
| 7871 | - } | |
| 7872 | - | |
| 7873 | - wp_send_json_success(array('message' => 'Queue marked as complete')); | |
| 7874 | -} | |
| 7875 | - | |
| 7876 | - | |
| 7877 | - // ======================================== | |
| 7878 | - // STATIC ACCESS METHODS | |
| 7879 | - // ======================================== | |
| 7880 | - | |
| 7881 | - /** | |
| 7882 | - * Get singleton instance | |
| 7883 | - */ | |
| 7884 | - public static function get_instance() { | |
| 7885 | - static $instance = null; | |
| 7886 | - if ($instance === null) { | |
| 7887 | - $instance = new self(); | |
| 7888 | - } | |
| 7889 | - return $instance; | |
| 7890 | - } | |
| 7891 | -} | |
| 7892 | - | |
| 7893 | -// Initialize the Knowledge manager | |
| 1 | +<?php | |
| 2 | +/** | |
| 3 | + * File: admin/class-knowledge-manager.php | |
| 4 | + * | |
| 5 | + * Handles all knowledge base content processing for MxChat | |
| 6 | + * Including PDF, sitemap, content processing, and WordPress post management | |
| 7 | + */ | |
| 8 | +if (!defined('ABSPATH')) { | |
| 9 | + exit; // Exit if accessed directly | |
| 10 | +} | |
| 11 | + | |
| 12 | +class MxChat_Knowledge_Manager { | |
| 13 | + | |
| 14 | + private $options; | |
| 15 | + | |
| 16 | + // Post IDs whose vectors were already deleted by mxchat_handle_status_transition this | |
| 17 | + // request, so the transient-based branch in mxchat_handle_post_update can skip the | |
| 18 | + // redundant (idempotent but network-visible) second deletion. | |
| 19 | + private $transition_deleted_posts = array(); | |
| 20 | + | |
| 21 | + // Post IDs already INDEXED by mxchat_handle_status_transition's arrival edge this | |
| 22 | + // request. Normal editor publishes fire transition_post_status first, then | |
| 23 | + // post_updated — without this guard every editor publish would embed twice. | |
| 24 | + private $transition_indexed_posts = array(); | |
| 25 | + | |
| 26 | + // Post IDs core has announced an in-flight UPDATE for. pre_post_update fires only | |
| 27 | + // inside wp_insert_post's update branch and always before wp_transition_post_status, | |
| 28 | + // so this is an exact "a post_updated is coming later this request" signal — which is | |
| 29 | + // what makes it safe to arm transition_indexed_posts (plan a664f3). | |
| 30 | + private $pending_post_update = array(); | |
| 31 | + | |
| 32 | + /** | |
| 33 | + * Constructor - Register hooks for content processing | |
| 34 | + */ | |
| 35 | +public function __construct() { | |
| 36 | + $this->options = get_option('mxchat_options', array()); | |
| 37 | + $this->mxchat_init_hooks(); | |
| 38 | + | |
| 39 | + $this->mxchat_init_role_hooks(); | |
| 40 | +} | |
| 41 | + | |
| 42 | +/** | |
| 43 | + * Initialize WordPress hooks for content processing | |
| 44 | + * | |
| 45 | + */ | |
| 46 | +private function mxchat_init_hooks() { | |
| 47 | + // Admin post handlers for form submissions | |
| 48 | + add_action('admin_post_mxchat_submit_content', array($this, 'mxchat_handle_content_submission')); | |
| 49 | + add_action('admin_post_mxchat_submit_sitemap', array($this, 'mxchat_handle_sitemap_submission')); | |
| 50 | + add_action('admin_post_mxchat_submit_pdf_file', array($this, 'mxchat_handle_pdf_file_submission')); | |
| 51 | + add_action('admin_post_mxchat_submit_document_file', array($this, 'mxchat_handle_document_file_submission')); | |
| 52 | + add_action('admin_post_mxchat_submit_youtube', array($this, 'mxchat_handle_youtube_submission')); | |
| 53 | + add_action('admin_post_mxchat_stop_processing', array($this, 'mxchat_stop_processing')); | |
| 54 | + | |
| 55 | + // AJAX handlers for real-time processing and status updates | |
| 56 | + add_action('wp_ajax_mxchat_get_status_updates', array($this, 'mxchat_ajax_get_status_updates')); | |
| 57 | + add_action('wp_ajax_mxchat_dismiss_completed_status', array($this, 'mxchat_ajax_dismiss_completed_status')); | |
| 58 | + add_action('wp_ajax_mxchat_get_content_list', array($this, 'ajax_mxchat_get_content_list')); | |
| 59 | + add_action('wp_ajax_mxchat_process_selected_content', array($this, 'ajax_mxchat_process_selected_content')); | |
| 60 | + add_action('wp_ajax_mxchat_save_inline_prompt', array($this, 'mxchat_save_inline_prompt')); | |
| 61 | + add_action('admin_post_mxchat_delete_pinecone_prompt', array($this, 'mxchat_handle_pinecone_prompt_delete')); | |
| 62 | + add_action('wp_ajax_mxchat_delete_pinecone_prompt', array($this, 'ajax_mxchat_delete_pinecone_prompt')); | |
| 63 | + add_action('wp_ajax_mxchat_delete_chunks_by_url', array($this, 'ajax_mxchat_delete_chunks_by_url')); | |
| 64 | + add_action('wp_ajax_mxchat_delete_wordpress_prompt', array($this, 'ajax_mxchat_delete_wordpress_prompt')); | |
| 65 | + add_action('wp_ajax_mxchat_bulk_delete_knowledge', array($this, 'ajax_mxchat_bulk_delete_knowledge')); | |
| 66 | + add_action('wp_ajax_mxchat_update_role_restriction', array($this, 'ajax_mxchat_update_role_restriction')); | |
| 67 | + | |
| 68 | + // Queue-based processing AJAX handlers | |
| 69 | + add_action('wp_ajax_mxchat_get_next_queue_item', array($this, 'ajax_mxchat_get_next_queue_item')); | |
| 70 | + add_action('wp_ajax_mxchat_process_queue_item', array($this, 'ajax_mxchat_process_queue_item')); | |
| 71 | + add_action('wp_ajax_mxchat_get_queue_status', array($this, 'ajax_mxchat_get_queue_status')); | |
| 72 | + add_action('wp_ajax_mxchat_clear_queue', array($this, 'ajax_mxchat_clear_queue')); | |
| 73 | + add_action('wp_ajax_mxchat_retry_failed', array($this, 'ajax_mxchat_retry_failed')); | |
| 74 | + add_action('wp_ajax_mxchat_get_recent_entries', array($this, 'ajax_mxchat_get_recent_entries')); | |
| 75 | + add_action('wp_ajax_mxchat_detect_sitemaps', array($this, 'ajax_mxchat_detect_sitemaps')); | |
| 76 | + add_action('wp_ajax_mxchat_refresh_pinecone_entries', array($this, 'ajax_mxchat_refresh_pinecone_entries')); | |
| 77 | + add_action('wp_ajax_mxchat_paginate_entries', array($this, 'ajax_mxchat_paginate_entries')); | |
| 78 | + add_action('wp_ajax_mxchat_get_entry_content', array($this, 'ajax_mxchat_get_entry_content')); | |
| 79 | + add_action('wp_ajax_mxchat_save_entry_content', array($this, 'ajax_mxchat_save_entry_content')); | |
| 80 | + add_action('wp_ajax_mxchat_inspect_entry', array($this, 'ajax_mxchat_inspect_entry')); | |
| 81 | + | |
| 82 | + // WordPress post management hooks | |
| 83 | + add_action('pre_post_update', array($this, 'mxchat_store_pre_update_status'), 10, 2); | |
| 84 | + add_action('post_updated', array($this, 'mxchat_handle_post_update'), 10, 3); | |
| 85 | + add_action('before_delete_post', array($this, 'mxchat_handle_post_delete')); | |
| 86 | + add_action('wp_trash_post', array($this, 'mxchat_handle_post_delete')); | |
| 87 | + // Authoritative unpublish detection: core hands this hook the REAL previous status, so | |
| 88 | + // removal no longer depends on the mxchat_prev_status_* transients (evictable by persistent | |
| 89 | + // object caches, never written by paths that bypass wp_update_post, e.g. plugins flipping | |
| 90 | + // post_status directly and calling wp_transition_post_status themselves). | |
| 91 | + add_action('transition_post_status', array($this, 'mxchat_handle_status_transition'), 10, 3); | |
| 92 | + | |
| 93 | + // Media-library attachments (plan 01010a). Attachments never reach the post | |
| 94 | + // hooks above in a usable state — post_updated fires but post_status is | |
| 95 | + // 'inherit', not 'publish', so the indexing branch always returns early — and | |
| 96 | + // wp_delete_attachment() fires ONLY 'delete_attachment' (wp_delete_post() | |
| 97 | + // delegates to it before its own before_delete_post), so the post-delete | |
| 98 | + // handler never sees one. Both paths therefore need their own hook, and both | |
| 99 | + // are gated on the SAME 'mxchat_auto_sync_attachment' toggle the Advanced | |
| 100 | + // Custom Post Sync Settings list already renders for Media. | |
| 101 | + add_action('add_attachment', array($this, 'mxchat_handle_attachment_change')); | |
| 102 | + add_action('edit_attachment', array($this, 'mxchat_handle_attachment_change')); | |
| 103 | + add_action('delete_attachment', array($this, 'mxchat_handle_attachment_delete')); | |
| 104 | + | |
| 105 | + // ACF hook - fires AFTER ACF fields are saved, ensuring ACF data is available | |
| 106 | + // Priority 20 to run after ACF's own save (which runs at priority 10) | |
| 107 | + add_action('acf/save_post', array($this, 'mxchat_handle_acf_save'), 20); | |
| 108 | + | |
| 109 | + // One-time cleanup for vectors orphaned by unpublishes that predate the | |
| 110 | + // transition_post_status handler (plan 816fb1): wp mxchat prune-unpublished | |
| 111 | + if (defined('WP_CLI') && WP_CLI) { | |
| 112 | + WP_CLI::add_command('mxchat prune-unpublished', array($this, 'cli_prune_unpublished')); | |
| 113 | + // In-place repair for RTL KB rows imported in visual order before the | |
| 114 | + // 32bf9e normalizer existed: wp mxchat rtl-repair (plan d1e6f7) | |
| 115 | + WP_CLI::add_command('mxchat rtl-repair', array($this, 'cli_rtl_repair')); | |
| 116 | + // One command for the entries that five import fixes left behind: | |
| 117 | + // wp mxchat kb-reimport (plan 89739b) | |
| 118 | + WP_CLI::add_command('mxchat kb-reimport', array($this, 'cli_kb_reimport')); | |
| 119 | + } | |
| 120 | + add_action('wp_ajax_mxchat_kb_reimport', array($this, 'ajax_mxchat_kb_reimport')); | |
| 121 | + | |
| 122 | + add_action('wp_ajax_mxchat_mark_queue_complete', array($this, 'ajax_mxchat_mark_queue_complete')); | |
| 123 | + | |
| 124 | + // WooCommerce product hooks (if WooCommerce is active) | |
| 125 | + if (class_exists('WooCommerce')) { | |
| 126 | + add_action('pre_post_update', array($this, 'mxchat_store_pre_update_status'), 10, 2); | |
| 127 | + add_action('save_post_product', array($this, 'mxchat_handle_product_change'), 10, 3); | |
| 128 | + add_action('wp_trash_post', array($this, 'mxchat_handle_product_delete')); | |
| 129 | + add_action('before_delete_post', array($this, 'mxchat_handle_product_delete')); | |
| 130 | + } | |
| 131 | +} | |
| 132 | + | |
| 133 | + /** | |
| 134 | + * Get current options (refreshed) | |
| 135 | + */ | |
| 136 | + private function mxchat_get_options() { | |
| 137 | + if (empty($this->options)) { | |
| 138 | + $this->options = get_option('mxchat_options', array()); | |
| 139 | + } | |
| 140 | + return $this->options; | |
| 141 | + } | |
| 142 | + | |
| 143 | + | |
| 144 | + // ======================================== | |
| 145 | + // MAIN CONTENT SUBMISSION HANDLERS | |
| 146 | + // ======================================== | |
| 147 | + | |
| 148 | +public function mxchat_handle_content_submission() { | |
| 149 | + // Check if the form was submitted and the user has permission. | |
| 150 | + if (!isset($_POST['submit_content']) || !current_user_can('manage_options')) { | |
| 151 | + return; | |
| 152 | + } | |
| 153 | + | |
| 154 | + // Verify the nonce. | |
| 155 | + $nonce = isset($_POST['mxchat_submit_content_nonce']) ? sanitize_text_field(wp_unslash($_POST['mxchat_submit_content_nonce'])) : ''; | |
| 156 | + if (!wp_verify_nonce($nonce, 'mxchat_submit_content_action')) { | |
| 157 | + wp_die(esc_html__('Nonce verification failed.', 'mxchat')); | |
| 158 | + } | |
| 159 | + | |
| 160 | + // Sanitize the inputs. | |
| 161 | + // Use wp_kses_post to allow safe HTML and wp_unslash to remove WordPress added slashes | |
| 162 | + $article_content = wp_kses_post(wp_unslash($_POST['article_content'])); | |
| 163 | + $article_url = isset($_POST['article_url']) ? esc_url_raw($_POST['article_url']) : ''; | |
| 164 | + | |
| 165 | + // Get bot_id from form submission | |
| 166 | + $bot_id = isset($_POST['bot_id']) ? sanitize_key($_POST['bot_id']) : 'default'; | |
| 167 | + | |
| 168 | + // Get bot-specific options and API key | |
| 169 | + $bot_options = $this->get_bot_options($bot_id); | |
| 170 | + $options = !empty($bot_options) ? $bot_options : get_option('mxchat_options'); | |
| 171 | + | |
| 172 | + // Custom-provider-aware decision; keyless custom sites must pass (plan cbd5fd). | |
| 173 | + $preflight = MxChat_Utils::embedding_preflight($options); | |
| 174 | + if (!$preflight['ok']) { | |
| 175 | + set_transient('mxchat_admin_notice_error', esc_html($preflight['reason']), 30); | |
| 176 | + wp_safe_redirect(esc_url(admin_url('admin.php?page=mxchat-prompts'))); | |
| 177 | + exit; | |
| 178 | + } | |
| 179 | + $api_key = $preflight['api_key']; | |
| 180 | + | |
| 181 | + // Use centralized utility function with bot_id | |
| 182 | + $result = MxChat_Utils::submit_content_to_db($article_content, $article_url, $api_key, null, $bot_id); | |
| 183 | + | |
| 184 | + if (is_wp_error($result)) { | |
| 185 | + set_transient('mxchat_admin_notice_error', | |
| 186 | + esc_html__('Error storing content: ', 'mxchat') . $result->get_error_message(), | |
| 187 | + 30 | |
| 188 | + ); | |
| 189 | + } else { | |
| 190 | + set_transient('mxchat_admin_notice_success', | |
| 191 | + esc_html__('Content successfully submitted!', 'mxchat'), | |
| 192 | + 30 | |
| 193 | + ); | |
| 194 | + } | |
| 195 | + | |
| 196 | + wp_safe_redirect(esc_url(admin_url('admin.php?page=mxchat-prompts'))); | |
| 197 | + exit; | |
| 198 | +} | |
| 199 | + | |
| 200 | +/** | |
| 201 | + * Handle the "YouTube" KB import source (admin-post form submission). | |
| 202 | + * | |
| 203 | + * Per-video description mode: | |
| 204 | + * - auto: fetch oEmbed metadata (reliable) + best-effort captions transcript. | |
| 205 | + * If no usable transcript, index the metadata anyway, tell the admin, | |
| 206 | + * and bounce back with the manual box pre-filled (never fail silently). | |
| 207 | + * - manual: the admin's own description is what gets indexed; metadata rides along. | |
| 208 | + * | |
| 209 | + * The row is stored with content_type 'youtube' and source_url = the canonical | |
| 210 | + * watch URL, so re-importing the same video UPDATES the entry (source_url | |
| 211 | + * duplicate handling in MxChat_Utils::store_in_wordpress_db) — that is also the | |
| 212 | + * "augment a metadata-only entry" path. | |
| 213 | + */ | |
| 214 | +public function mxchat_handle_youtube_submission() { | |
| 215 | + if (!isset($_POST['submit_youtube']) || !current_user_can('manage_options')) { | |
| 216 | + wp_die(esc_html__('Unauthorized access', 'mxchat')); | |
| 217 | + } | |
| 218 | + | |
| 219 | + check_admin_referer('mxchat_submit_youtube_action', 'mxchat_submit_youtube_nonce'); | |
| 220 | + | |
| 221 | + $redirect_url = admin_url('admin.php?page=mxchat-prompts'); | |
| 222 | + | |
| 223 | + $youtube_url = isset($_POST['youtube_url']) ? esc_url_raw(wp_unslash($_POST['youtube_url'])) : ''; | |
| 224 | + $video_id = MxChat_Utils::parse_youtube_id($youtube_url); | |
| 225 | + | |
| 226 | + if (empty($video_id)) { | |
| 227 | + set_transient('mxchat_admin_notice_error', | |
| 228 | + esc_html__('That does not look like a link to a single YouTube video. Please paste a watch, youtu.be, or Shorts URL.', 'mxchat'), | |
| 229 | + 30 | |
| 230 | + ); | |
| 231 | + wp_safe_redirect(esc_url($redirect_url)); | |
| 232 | + exit; | |
| 233 | + } | |
| 234 | + | |
| 235 | + $canonical_url = 'https://www.youtube.com/watch?v=' . $video_id; | |
| 236 | + | |
| 237 | + $description_mode = (isset($_POST['youtube_description_mode']) && $_POST['youtube_description_mode'] === 'manual') ? 'manual' : 'auto'; | |
| 238 | + $manual_description = isset($_POST['youtube_description']) ? trim(wp_kses_post(wp_unslash($_POST['youtube_description']))) : ''; | |
| 239 | + | |
| 240 | + $bot_id = isset($_POST['bot_id']) ? sanitize_key($_POST['bot_id']) : 'default'; | |
| 241 | + | |
| 242 | + // Resolve the embedding decision exactly like the sibling handlers — | |
| 243 | + // custom-provider-aware (plan cbd5fd). | |
| 244 | + $bot_options = $this->get_bot_options($bot_id); | |
| 245 | + $options = !empty($bot_options) ? $bot_options : get_option('mxchat_options'); | |
| 246 | + | |
| 247 | + $preflight = MxChat_Utils::embedding_preflight($options); | |
| 248 | + if (!$preflight['ok']) { | |
| 249 | + set_transient('mxchat_admin_notice_error', esc_html($preflight['reason']), 30); | |
| 250 | + wp_safe_redirect(esc_url($redirect_url)); | |
| 251 | + exit; | |
| 252 | + } | |
| 253 | + $api_key = $preflight['api_key']; | |
| 254 | + | |
| 255 | + // Metadata is fetched in BOTH modes — it is the reliable half of auto, and in | |
| 256 | + // manual mode it enriches the indexed text with the real title/channel. | |
| 257 | + $meta = $this->mxchat_fetch_youtube_oembed($video_id); | |
| 258 | + $video_title = isset($meta['title']) ? sanitize_text_field($meta['title']) : ''; | |
| 259 | + $video_channel = isset($meta['author_name']) ? sanitize_text_field($meta['author_name']) : ''; | |
| 260 | + | |
| 261 | + $header_lines = 'YouTube Video: ' . ($video_title !== '' ? $video_title : $canonical_url) . "\n"; | |
| 262 | + if ($video_channel !== '') { | |
| 263 | + $header_lines .= 'Channel: ' . $video_channel . "\n"; | |
| 264 | + } | |
| 265 | + $header_lines .= 'URL: ' . $canonical_url . "\n\n"; | |
| 266 | + | |
| 267 | + $transcript_missing = false; | |
| 268 | + | |
| 269 | + if ($description_mode === 'manual') { | |
| 270 | + if ($manual_description === '') { | |
| 271 | + set_transient('mxchat_admin_notice_error', | |
| 272 | + esc_html__('Please write a description for the video, or switch to auto-fetch.', 'mxchat'), | |
| 273 | + 30 | |
| 274 | + ); | |
| 275 | + wp_safe_redirect(esc_url($redirect_url)); | |
| 276 | + exit; | |
| 277 | + } | |
| 278 | + $indexed_text = $header_lines . $manual_description; | |
| 279 | + } else { | |
| 280 | + $transcript = $this->mxchat_fetch_youtube_transcript($video_id); | |
| 281 | + | |
| 282 | + if (strlen($transcript) >= 200) { | |
| 283 | + $indexed_text = $header_lines . $transcript; | |
| 284 | + } else { | |
| 285 | + // Graceful fallback: captions disabled / blocked / no speech. Auto | |
| 286 | + // reliably gets metadata; it does NOT guarantee a transcript. | |
| 287 | + $transcript_missing = true; | |
| 288 | + | |
| 289 | + if ($video_title === '' && $video_channel === '') { | |
| 290 | + // Both halves failed — nothing meaningful to index. | |
| 291 | + set_transient('mxchat_admin_notice_error', | |
| 292 | + esc_html__('Could not retrieve any information for that video (no metadata and no captions). Please check the URL, or use the manual description option.', 'mxchat'), | |
| 293 | + 30 | |
| 294 | + ); | |
| 295 | + wp_safe_redirect(esc_url($redirect_url)); | |
| 296 | + exit; | |
| 297 | + } | |
| 298 | + | |
| 299 | + $indexed_text = $header_lines . sprintf( | |
| 300 | + /* translators: 1: video title, 2: channel name */ | |
| 301 | + __('A YouTube video titled "%1$s" from the channel %2$s.', 'mxchat'), | |
| 302 | + $video_title !== '' ? $video_title : $canonical_url, | |
| 303 | + $video_channel !== '' ? $video_channel : 'YouTube' | |
| 304 | + ); | |
| 305 | + } | |
| 306 | + } | |
| 307 | + | |
| 308 | + $result = MxChat_Utils::submit_content_to_db($indexed_text, $canonical_url, $api_key, null, $bot_id, 'youtube'); | |
| 309 | + | |
| 310 | + if (is_wp_error($result)) { | |
| 311 | + set_transient('mxchat_admin_notice_error', | |
| 312 | + esc_html__('Error storing video in the knowledge base: ', 'mxchat') . $result->get_error_message(), | |
| 313 | + 30 | |
| 314 | + ); | |
| 315 | + wp_safe_redirect(esc_url($redirect_url)); | |
| 316 | + exit; | |
| 317 | + } | |
| 318 | + | |
| 319 | + if ($transcript_missing) { | |
| 320 | + set_transient('mxchat_admin_notice_success', | |
| 321 | + esc_html__('Video indexed from its title and channel — no captions were available for a transcript. The form below is pre-filled: write your own description and import again to improve matching (it updates the same entry).', 'mxchat'), | |
| 322 | + 30 | |
| 323 | + ); | |
| 324 | + // Bounce back with prefill args so the page reopens the YouTube form in | |
| 325 | + // manual mode with the URL + fetched title ready to augment. | |
| 326 | + $redirect_url = add_query_arg(array( | |
| 327 | + 'mxchat_yt_prefill' => '1', | |
| 328 | + 'yt_url' => rawurlencode($canonical_url), | |
| 329 | + 'yt_title' => rawurlencode($video_title), | |
| 330 | + ), $redirect_url); | |
| 331 | + } else { | |
| 332 | + set_transient('mxchat_admin_notice_success', | |
| 333 | + esc_html__('YouTube video successfully added to the knowledge base!', 'mxchat'), | |
| 334 | + 30 | |
| 335 | + ); | |
| 336 | + } | |
| 337 | + | |
| 338 | + wp_safe_redirect(esc_url_raw($redirect_url)); | |
| 339 | + exit; | |
| 340 | +} | |
| 341 | + | |
| 342 | +/** | |
| 343 | + * Fetch YouTube oEmbed metadata for a video (no API key required). | |
| 344 | + * Returns the decoded array (title, author_name, thumbnail_url, ...) or array(). | |
| 345 | + */ | |
| 346 | +private function mxchat_fetch_youtube_oembed($video_id) { | |
| 347 | + $oembed_url = 'https://www.youtube.com/oembed?url=' . rawurlencode('https://www.youtube.com/watch?v=' . $video_id) . '&format=json'; | |
| 348 | + $response = wp_remote_get($oembed_url, array('timeout' => 15)); | |
| 349 | + if (is_wp_error($response) || wp_remote_retrieve_response_code($response) !== 200) { | |
| 350 | + return array(); | |
| 351 | + } | |
| 352 | + $data = json_decode(wp_remote_retrieve_body($response), true); | |
| 353 | + return is_array($data) ? $data : array(); | |
| 354 | +} | |
| 355 | + | |
| 356 | +/** | |
| 357 | + * Best-effort captions transcript for a video. Deliberately ISOLATED: this uses | |
| 358 | + * YouTube's unofficial timedtext route (the caption track list embedded in the | |
| 359 | + * watch page), which YouTube has broken before and will break again. Every | |
| 360 | + * failure mode returns '' so a break degrades to the metadata-only import path | |
| 361 | + * instead of erroring the whole submission. Do not let anything in here throw. | |
| 362 | + */ | |
| 363 | +private function mxchat_fetch_youtube_transcript($video_id) { | |
| 364 | + $watch_url = 'https://www.youtube.com/watch?v=' . $video_id . '&hl=en'; | |
| 365 | + | |
| 366 | + // First try the honest ingest UA; some responses omit the player config for | |
| 367 | + // bot UAs, so retry once with a browser UA before giving up. | |
| 368 | + $user_agents = array( | |
| 369 | + mxchat_ingest_user_agent(), | |
| 370 | + 'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/126.0.0.0 Safari/537.36', | |
| 371 | + ); | |
| 372 | + | |
| 373 | + $tracks = array(); | |
| 374 | + foreach ($user_agents as $ua) { | |
| 375 | + $response = wp_remote_get($watch_url, array( | |
| 376 | + 'timeout' => 20, | |
| 377 | + 'user-agent' => $ua, | |
| 378 | + )); | |
| 379 | + if (is_wp_error($response) || wp_remote_retrieve_response_code($response) !== 200) { | |
| 380 | + continue; | |
| 381 | + } | |
| 382 | + $body = wp_remote_retrieve_body($response); | |
| 383 | + if (!is_string($body) || $body === '' || !preg_match('/"captionTracks":(\[.*?\])(?=,")/s', $body, $m)) { | |
| 384 | + continue; | |
| 385 | + } | |
| 386 | + $decoded = json_decode($m[1], true); | |
| 387 | + if (is_array($decoded) && !empty($decoded)) { | |
| 388 | + $tracks = $decoded; | |
| 389 | + break; | |
| 390 | + } | |
| 391 | + } | |
| 392 | + | |
| 393 | + if (empty($tracks)) { | |
| 394 | + return ''; | |
| 395 | + } | |
| 396 | + | |
| 397 | + // Prefer an English track, else take the first offered. | |
| 398 | + $chosen = null; | |
| 399 | + foreach ($tracks as $track) { | |
| 400 | + if (isset($track['languageCode']) && strpos($track['languageCode'], 'en') === 0) { | |
| 401 | + $chosen = $track; | |
| 402 | + break; | |
| 403 | + } | |
| 404 | + } | |
| 405 | + if ($chosen === null) { | |
| 406 | + $chosen = $tracks[0]; | |
| 407 | + } | |
| 408 | + if (empty($chosen['baseUrl']) || !is_string($chosen['baseUrl'])) { | |
| 409 | + return ''; | |
| 410 | + } | |
| 411 | + | |
| 412 | + $timedtext = wp_remote_get($chosen['baseUrl'], array('timeout' => 20)); | |
| 413 | + if (is_wp_error($timedtext) || wp_remote_retrieve_response_code($timedtext) !== 200) { | |
| 414 | + return ''; | |
| 415 | + } | |
| 416 | + $xml = wp_remote_retrieve_body($timedtext); | |
| 417 | + if (!is_string($xml) || strpos($xml, '<text') === false) { | |
| 418 | + return ''; | |
| 419 | + } | |
| 420 | + | |
| 421 | + // <text start=".." dur="..">caption</text> — strip tags, decode the | |
| 422 | + // double-encoded entities timedtext ships, collapse whitespace. | |
| 423 | + $text = preg_replace('/<[^>]+>/', ' ', $xml); | |
| 424 | + $text = html_entity_decode(html_entity_decode($text, ENT_QUOTES | ENT_HTML5, 'UTF-8'), ENT_QUOTES | ENT_HTML5, 'UTF-8'); | |
| 425 | + $text = trim(preg_replace('/\s+/u', ' ', $text)); | |
| 426 | + | |
| 427 | + return $text; | |
| 428 | +} | |
| 429 | + | |
| 430 | +public function mxchat_is_pdf_url($url, $response) { | |
| 431 | + $content_type = wp_remote_retrieve_header($response, 'content-type'); | |
| 432 | + $file_extension = strtolower(pathinfo($url, PATHINFO_EXTENSION)); | |
| 433 | + | |
| 434 | + // Check Content-Disposition header for .pdf filename (Google Drive sends this) | |
| 435 | + $disposition = wp_remote_retrieve_header($response, 'content-disposition'); | |
| 436 | + $has_pdf_disposition = ! empty($disposition) && stripos($disposition, '.pdf') !== false; | |
| 437 | + | |
| 438 | + return strpos($content_type, 'pdf') !== false || $file_extension === 'pdf' || $has_pdf_disposition; | |
| 439 | +} | |
| 440 | + | |
| 441 | + | |
| 442 | +public function mxchat_handle_pdf_for_knowledge_base($pdf_url, $response, $bot_id = 'default') { | |
| 443 | + if (!current_user_can('manage_options')) { | |
| 444 | + return false; | |
| 445 | + } | |
| 446 | + | |
| 447 | + $pdf_url = esc_url_raw($pdf_url); | |
| 448 | + $upload_dir = wp_upload_dir(); | |
| 449 | + | |
| 450 | + if (isset($upload_dir['error']) && $upload_dir['error'] !== false) { | |
| 451 | + return false; | |
| 452 | + } | |
| 453 | + | |
| 454 | + $pdf_filename = sanitize_file_name('mxchat_kb_' . time() . '.pdf'); | |
| 455 | + $pdf_path = trailingslashit($upload_dir['path']) . $pdf_filename; | |
| 456 | + | |
| 457 | + $response_body = wp_remote_retrieve_body($response); | |
| 458 | + if (empty($response_body)) { | |
| 459 | + return false; | |
| 460 | + } | |
| 461 | + | |
| 462 | + if (!wp_mkdir_p(dirname($pdf_path))) { | |
| 463 | + return false; | |
| 464 | + } | |
| 465 | + | |
| 466 | + try { | |
| 467 | + file_put_contents($pdf_path, $response_body); | |
| 468 | + | |
| 469 | + if (!file_exists($pdf_path)) { | |
| 470 | + throw new Exception(__('Failed to save PDF file', 'mxchat')); | |
| 471 | + } | |
| 472 | + | |
| 473 | + $total_pages = $this->mxchat_validate_and_count_pdf_pages($pdf_path); | |
| 474 | + | |
| 475 | + if ($total_pages === false || $total_pages < 1) { | |
| 476 | + throw new Exception(__('Invalid PDF: Unable to parse or no pages found', 'mxchat')); | |
| 477 | + } | |
| 478 | + | |
| 479 | + // Create unique queue ID | |
| 480 | + $queue_id = 'pdf_' . md5($pdf_url . time()); | |
| 481 | + | |
| 482 | + // Create array of pages to process | |
| 483 | + $pages = array(); | |
| 484 | + for ($i = 1; $i <= $total_pages; $i++) { | |
| 485 | + $pages[] = array( | |
| 486 | + 'pdf_path' => $pdf_path, | |
| 487 | + 'pdf_url' => $pdf_url, | |
| 488 | + 'page_number' => $i, | |
| 489 | + 'total_pages' => $total_pages | |
| 490 | + ); | |
| 491 | + } | |
| 492 | + | |
| 493 | + // Add pages to queue | |
| 494 | + $queued_count = $this->mxchat_add_to_queue($queue_id, 'pdf_page', $pages, $bot_id); | |
| 495 | + | |
| 496 | + if ($queued_count === 0) { | |
| 497 | + wp_delete_file($pdf_path); | |
| 498 | + throw new Exception(__('Failed to add PDF pages to processing queue', 'mxchat')); | |
| 499 | + } | |
| 500 | + | |
| 501 | + // Store queue metadata | |
| 502 | + $this->mxchat_set_queue_meta($queue_id, 'source_url', $pdf_url); | |
| 503 | + $this->mxchat_set_queue_meta($queue_id, 'queue_type', 'pdf'); | |
| 504 | + $this->mxchat_set_queue_meta($queue_id, 'total_items', $total_pages); | |
| 505 | + $this->mxchat_set_queue_meta($queue_id, 'bot_id', $bot_id); | |
| 506 | + $this->mxchat_set_queue_meta($queue_id, 'pdf_path', $pdf_path); | |
| 507 | + $this->mxchat_set_queue_meta($queue_id, 'created_at', current_time('mysql')); | |
| 508 | + | |
| 509 | + // Store queue ID in transient for status tracking | |
| 510 | + set_transient('mxchat_active_queue_pdf', $queue_id, DAY_IN_SECONDS); | |
| 511 | + set_transient('mxchat_last_pdf_url', $pdf_url, DAY_IN_SECONDS); | |
| 512 | + | |
| 513 | + return 'queued'; | |
| 514 | + | |
| 515 | + } catch (Exception $e) { | |
| 516 | + if (file_exists($pdf_path)) { | |
| 517 | + wp_delete_file($pdf_path); | |
| 518 | + } | |
| 519 | + return $e->getMessage(); | |
| 520 | + } | |
| 521 | +} | |
| 522 | + | |
| 523 | +/** | |
| 524 | + * Handle direct PDF file upload from the knowledge base page | |
| 525 | + */ | |
| 526 | +public function mxchat_handle_pdf_file_submission() { | |
| 527 | + if (!isset($_POST['submit_pdf_file']) || !current_user_can('manage_options')) { | |
| 528 | + wp_die(esc_html__('Unauthorized access', 'mxchat')); | |
| 529 | + } | |
| 530 | + | |
| 531 | + check_admin_referer('mxchat_submit_pdf_file_action', 'mxchat_submit_pdf_file_nonce'); | |
| 532 | + | |
| 533 | + $redirect_url = admin_url('admin.php?page=mxchat-prompts'); | |
| 534 | + | |
| 535 | + // Validate file upload | |
| 536 | + if (empty($_FILES['pdf_file']) || $_FILES['pdf_file']['error'] !== UPLOAD_ERR_OK) { | |
| 537 | + $error_code = isset($_FILES['pdf_file']['error']) ? $_FILES['pdf_file']['error'] : UPLOAD_ERR_NO_FILE; | |
| 538 | + $error_messages = array( | |
| 539 | + UPLOAD_ERR_INI_SIZE => __('The uploaded file exceeds the server upload_max_filesize limit.', 'mxchat'), | |
| 540 | + UPLOAD_ERR_FORM_SIZE => __('The uploaded file exceeds the form MAX_FILE_SIZE limit.', 'mxchat'), | |
| 541 | + UPLOAD_ERR_PARTIAL => __('The file was only partially uploaded.', 'mxchat'), | |
| 542 | + UPLOAD_ERR_NO_FILE => __('No file was uploaded. Please select a PDF file.', 'mxchat'), | |
| 543 | + UPLOAD_ERR_NO_TMP_DIR => __('Server missing temporary folder.', 'mxchat'), | |
| 544 | + UPLOAD_ERR_CANT_WRITE => __('Server failed to write file to disk.', 'mxchat'), | |
| 545 | + ); | |
| 546 | + $error_msg = isset($error_messages[$error_code]) ? $error_messages[$error_code] : __('Unknown upload error.', 'mxchat'); | |
| 547 | + set_transient('mxchat_admin_notice_error', $error_msg, 30); | |
| 548 | + wp_safe_redirect(esc_url($redirect_url)); | |
| 549 | + exit; | |
| 550 | + } | |
| 551 | + | |
| 552 | + $file = $_FILES['pdf_file']; | |
| 553 | + | |
| 554 | + // Validate MIME type | |
| 555 | + $finfo = finfo_open(FILEINFO_MIME_TYPE); | |
| 556 | + $mime_type = finfo_file($finfo, $file['tmp_name']); | |
| 557 | + finfo_close($finfo); | |
| 558 | + | |
| 559 | + if ($mime_type !== 'application/pdf') { | |
| 560 | + set_transient('mxchat_admin_notice_error', | |
| 561 | + esc_html__('Invalid file type. Only PDF files are accepted.', 'mxchat'), | |
| 562 | + 30 | |
| 563 | + ); | |
| 564 | + wp_safe_redirect(esc_url($redirect_url)); | |
| 565 | + exit; | |
| 566 | + } | |
| 567 | + | |
| 568 | + // Validate extension | |
| 569 | + $ext = strtolower(pathinfo($file['name'], PATHINFO_EXTENSION)); | |
| 570 | + if ($ext !== 'pdf') { | |
| 571 | + set_transient('mxchat_admin_notice_error', | |
| 572 | + esc_html__('Invalid file extension. Only .pdf files are accepted.', 'mxchat'), | |
| 573 | + 30 | |
| 574 | + ); | |
| 575 | + wp_safe_redirect(esc_url($redirect_url)); | |
| 576 | + exit; | |
| 577 | + } | |
| 578 | + | |
| 579 | + $bot_id = isset($_POST['bot_id']) ? sanitize_key($_POST['bot_id']) : 'default'; | |
| 580 | + $original_filename = sanitize_file_name($file['name']); | |
| 581 | + | |
| 582 | + $upload_dir = wp_upload_dir(); | |
| 583 | + if (isset($upload_dir['error']) && $upload_dir['error'] !== false) { | |
| 584 | + set_transient('mxchat_admin_notice_error', | |
| 585 | + esc_html__('WordPress upload directory is not writable.', 'mxchat'), | |
| 586 | + 30 | |
| 587 | + ); | |
| 588 | + wp_safe_redirect(esc_url($redirect_url)); | |
| 589 | + exit; | |
| 590 | + } | |
| 591 | + | |
| 592 | + $pdf_filename = sanitize_file_name('mxchat_kb_' . time() . '.pdf'); | |
| 593 | + $pdf_path = trailingslashit($upload_dir['path']) . $pdf_filename; | |
| 594 | + | |
| 595 | + if (!wp_mkdir_p(dirname($pdf_path))) { | |
| 596 | + set_transient('mxchat_admin_notice_error', | |
| 597 | + esc_html__('Failed to create upload directory.', 'mxchat'), | |
| 598 | + 30 | |
| 599 | + ); | |
| 600 | + wp_safe_redirect(esc_url($redirect_url)); | |
| 601 | + exit; | |
| 602 | + } | |
| 603 | + | |
| 604 | + // Move uploaded file | |
| 605 | + if (!move_uploaded_file($file['tmp_name'], $pdf_path)) { | |
| 606 | + set_transient('mxchat_admin_notice_error', | |
| 607 | + esc_html__('Failed to save uploaded PDF file.', 'mxchat'), | |
| 608 | + 30 | |
| 609 | + ); | |
| 610 | + wp_safe_redirect(esc_url($redirect_url)); | |
| 611 | + exit; | |
| 612 | + } | |
| 613 | + | |
| 614 | + try { | |
| 615 | + $total_pages = $this->mxchat_validate_and_count_pdf_pages($pdf_path); | |
| 616 | + | |
| 617 | + if ($total_pages === false || $total_pages < 1) { | |
| 618 | + throw new Exception(__('Invalid PDF: Unable to parse or no pages found', 'mxchat')); | |
| 619 | + } | |
| 620 | + | |
| 621 | + // Use original filename as the source identifier | |
| 622 | + $source_label = 'upload://' . $original_filename; | |
| 623 | + | |
| 624 | + // Optional page the PDF is cited as (plans 025ea2 / dcc19d). Kept | |
| 625 | + // beside the file's upload:// identity, which every page carries, so a | |
| 626 | + // re-upload still replaces the pages; leaving the field empty on a | |
| 627 | + // re-upload clears an earlier page. | |
| 628 | + MxChat_Utils::set_upload_source_url( | |
| 629 | + $source_label, | |
| 630 | + $bot_id, | |
| 631 | + isset($_POST['source_page_url']) ? wp_unslash($_POST['source_page_url']) : '' | |
| 632 | + ); | |
| 633 | + | |
| 634 | + $queue_id = 'pdf_' . md5($source_label . time()); | |
| 635 | + | |
| 636 | + $pages = array(); | |
| 637 | + for ($i = 1; $i <= $total_pages; $i++) { | |
| 638 | + $pages[] = array( | |
| 639 | + 'pdf_path' => $pdf_path, | |
| 640 | + 'pdf_url' => $source_label, | |
| 641 | + 'page_number' => $i, | |
| 642 | + 'total_pages' => $total_pages, | |
| 643 | + ); | |
| 644 | + } | |
| 645 | + | |
| 646 | + $queued_count = $this->mxchat_add_to_queue($queue_id, 'pdf_page', $pages, $bot_id); | |
| 647 | + | |
| 648 | + if ($queued_count === 0) { | |
| 649 | + wp_delete_file($pdf_path); | |
| 650 | + throw new Exception(__('Failed to add PDF pages to processing queue', 'mxchat')); | |
| 651 | + } | |
| 652 | + | |
| 653 | + $this->mxchat_set_queue_meta($queue_id, 'source_url', $source_label); | |
| 654 | + $this->mxchat_set_queue_meta($queue_id, 'queue_type', 'pdf'); | |
| 655 | + $this->mxchat_set_queue_meta($queue_id, 'total_items', $total_pages); | |
| 656 | + $this->mxchat_set_queue_meta($queue_id, 'bot_id', $bot_id); | |
| 657 | + $this->mxchat_set_queue_meta($queue_id, 'pdf_path', $pdf_path); | |
| 658 | + $this->mxchat_set_queue_meta($queue_id, 'created_at', current_time('mysql')); | |
| 659 | + | |
| 660 | + set_transient('mxchat_active_queue_pdf', $queue_id, DAY_IN_SECONDS); | |
| 661 | + set_transient('mxchat_last_pdf_url', $source_label, DAY_IN_SECONDS); | |
| 662 | + | |
| 663 | + set_transient('mxchat_admin_notice_success', | |
| 664 | + sprintf( | |
| 665 | + esc_html__('PDF "%s" (%d pages) queued for processing. Processing will start automatically.', 'mxchat'), | |
| 666 | + esc_html($original_filename), | |
| 667 | + $total_pages | |
| 668 | + ), | |
| 669 | + 30 | |
| 670 | + ); | |
| 671 | + | |
| 672 | + } catch (Exception $e) { | |
| 673 | + if (file_exists($pdf_path)) { | |
| 674 | + wp_delete_file($pdf_path); | |
| 675 | + } | |
| 676 | + set_transient('mxchat_admin_notice_error', | |
| 677 | + esc_html__('Failed to process uploaded PDF: ', 'mxchat') . esc_html($e->getMessage()), | |
| 678 | + 30 | |
| 679 | + ); | |
| 680 | + } | |
| 681 | + | |
| 682 | + wp_safe_redirect(esc_url($redirect_url)); | |
| 683 | + exit; | |
| 684 | +} | |
| 685 | + | |
| 686 | +/** | |
| 687 | + * Handle direct document upload (.docx / .txt / .md) from the knowledge base | |
| 688 | + * page (plan 0485e5). Unlike PDF Upload there is no per-page queue: the text | |
| 689 | + * extracts in one pass and routes through submit_content_to_db, whose chunker | |
| 690 | + * takes over for long content. The uploaded file is read from the PHP temp | |
| 691 | + * file and never persisted — only its extracted text enters the KB. | |
| 692 | + * | |
| 693 | + * Source identity matches PDF Upload's scheme: upload://<filename>, stable | |
| 694 | + * across re-uploads so a re-import REPLACES (delete_chunks_for_url + upsert | |
| 695 | + * per identity) instead of duplicating. | |
| 696 | + */ | |
| 697 | +public function mxchat_handle_document_file_submission() { | |
| 698 | + if (!isset($_POST['submit_document_file']) || !current_user_can('manage_options')) { | |
| 699 | + wp_die(esc_html__('Unauthorized access', 'mxchat')); | |
| 700 | + } | |
| 701 | + | |
| 702 | + check_admin_referer('mxchat_submit_document_file_action', 'mxchat_submit_document_file_nonce'); | |
| 703 | + | |
| 704 | + $redirect_url = admin_url('admin.php?page=mxchat-prompts'); | |
| 705 | + | |
| 706 | + if (empty($_FILES['document_file']) || $_FILES['document_file']['error'] !== UPLOAD_ERR_OK) { | |
| 707 | + $error_code = isset($_FILES['document_file']['error']) ? $_FILES['document_file']['error'] : UPLOAD_ERR_NO_FILE; | |
| 708 | + $error_messages = array( | |
| 709 | + UPLOAD_ERR_INI_SIZE => __('The uploaded file exceeds the server upload_max_filesize limit.', 'mxchat'), | |
| 710 | + UPLOAD_ERR_FORM_SIZE => __('The uploaded file exceeds the form MAX_FILE_SIZE limit.', 'mxchat'), | |
| 711 | + UPLOAD_ERR_PARTIAL => __('The file was only partially uploaded.', 'mxchat'), | |
| 712 | + UPLOAD_ERR_NO_FILE => __('No file was uploaded. Please select a document.', 'mxchat'), | |
| 713 | + UPLOAD_ERR_NO_TMP_DIR => __('Server missing temporary folder.', 'mxchat'), | |
| 714 | + UPLOAD_ERR_CANT_WRITE => __('Server failed to write file to disk.', 'mxchat'), | |
| 715 | + ); | |
| 716 | + $error_msg = isset($error_messages[$error_code]) ? $error_messages[$error_code] : __('Unknown upload error.', 'mxchat'); | |
| 717 | + set_transient('mxchat_admin_notice_error', $error_msg, 30); | |
| 718 | + wp_safe_redirect(esc_url($redirect_url)); | |
| 719 | + exit; | |
| 720 | + } | |
| 721 | + | |
| 722 | + $file = $_FILES['document_file']; | |
| 723 | + $ext = strtolower(pathinfo($file['name'], PATHINFO_EXTENSION)); | |
| 724 | + | |
| 725 | + $finfo = finfo_open(FILEINFO_MIME_TYPE); | |
| 726 | + $mime_type = finfo_file($finfo, $file['tmp_name']); | |
| 727 | + finfo_close($finfo); | |
| 728 | + | |
| 729 | + // Per-extension MIME expectations. finfo commonly reports .docx as | |
| 730 | + // application/zip (it IS a Zip container) and .md as plain text. | |
| 731 | + $mime_ok = false; | |
| 732 | + if ($ext === 'docx') { | |
| 733 | + $mime_ok = in_array($mime_type, array( | |
| 734 | + 'application/vnd.openxmlformats-officedocument.wordprocessingml.document', | |
| 735 | + 'application/zip', | |
| 736 | + ), true); | |
| 737 | + } elseif ($ext === 'txt' || $ext === 'md') { | |
| 738 | + $mime_ok = (strpos((string) $mime_type, 'text/') === 0); | |
| 739 | + } | |
| 740 | + | |
| 741 | + if (!$mime_ok) { | |
| 742 | + set_transient('mxchat_admin_notice_error', | |
| 743 | + esc_html__('Invalid or unreadable document. Accepted types: .docx, .txt, .md.', 'mxchat'), | |
| 744 | + 30 | |
| 745 | + ); | |
| 746 | + wp_safe_redirect(esc_url($redirect_url)); | |
| 747 | + exit; | |
| 748 | + } | |
| 749 | + | |
| 750 | + $bot_id = isset($_POST['bot_id']) ? sanitize_key($_POST['bot_id']) : 'default'; | |
| 751 | + $original_filename = sanitize_file_name($file['name']); | |
| 752 | + | |
| 753 | + // ---- Extract text (ONE extractor for .docx — the word handler's) ---- | |
| 754 | + if ($ext === 'docx') { | |
| 755 | + $text = MXChat_Word_Handler::extract_docx_text($file['tmp_name']); | |
| 756 | + if ($text === false) { | |
| 757 | + set_transient('mxchat_admin_notice_error', | |
| 758 | + esc_html__('The .docx file could not be read. It may be corrupt, empty, or not a real Word document.', 'mxchat'), | |
| 759 | + 30 | |
| 760 | + ); | |
| 761 | + wp_safe_redirect(esc_url($redirect_url)); | |
| 762 | + exit; | |
| 763 | + } | |
| 764 | + } else { | |
| 765 | + // .txt / .md read as-is. Markdown keeps its syntax on purpose — | |
| 766 | + // headings are useful retrieval signal. | |
| 767 | + $text = (string) file_get_contents($file['tmp_name']); | |
| 768 | + $text = wp_check_invalid_utf8($text); | |
| 769 | + $text = trim($text); | |
| 770 | + } | |
| 771 | + | |
| 772 | + if ($text === '') { | |
| 773 | + set_transient('mxchat_admin_notice_error', | |
| 774 | + esc_html__('The uploaded document contains no readable text.', 'mxchat'), | |
| 775 | + 30 | |
| 776 | + ); | |
| 777 | + wp_safe_redirect(esc_url($redirect_url)); | |
| 778 | + exit; | |
| 779 | + } | |
| 780 | + | |
| 781 | + // Size cap — same pdf_max_pages setting the PDF/toolbar paths use, but | |
| 782 | + // estimated by CHARACTERS (~2500/page): the .docx cleaner collapses all | |
| 783 | + // newlines to spaces, so a paragraph count reads 1 for any Word file. | |
| 784 | + // Processing is synchronous — an unbounded document risks a timeout. | |
| 785 | + $options = get_option('mxchat_options', array()); | |
| 786 | + $max_pages = isset($options['pdf_max_pages']) ? intval($options['pdf_max_pages']) : 69; | |
| 787 | + $estimated_pages = (int) ceil(strlen($text) / 2500); | |
| 788 | + if ($estimated_pages > $max_pages) { | |
| 789 | + set_transient('mxchat_admin_notice_error', | |
| 790 | + sprintf( | |
| 791 | + esc_html__('The document is too large (about %1$d pages; the limit is %2$d). Split it into smaller files, or raise the PDF max pages setting.', 'mxchat'), | |
| 792 | + $estimated_pages, | |
| 793 | + $max_pages | |
| 794 | + ), | |
| 795 | + 30 | |
| 796 | + ); | |
| 797 | + wp_safe_redirect(esc_url($redirect_url)); | |
| 798 | + exit; | |
| 799 | + } | |
| 800 | + | |
| 801 | + // Embedding API key — bot-aware, same shape as the direct-content handler. | |
| 802 | + $api_key = ''; | |
| 803 | + if ($bot_id !== 'default' && class_exists('MxChat_Multi_Bot_Manager')) { | |
| 804 | + $bot_options = apply_filters('mxchat_get_bot_options', array(), $bot_id); | |
| 805 | + $api_key = $bot_options['api_key'] ?? ''; | |
| 806 | + } | |
| 807 | + if (empty($api_key)) { | |
| 808 | + $api_key = $options['api_key'] ?? ''; | |
| 809 | + } | |
| 810 | + | |
| 811 | + // Stable identity — PDF Upload's scheme. A re-upload of the same filename | |
| 812 | + // replaces: clear old chunks first (covers a doc shrinking below the chunk | |
| 813 | + // threshold, where the single-vector path would not clean them), then | |
| 814 | + // submit — the chunked path re-deletes harmlessly. | |
| 815 | + $source_label = 'upload://' . $original_filename; | |
| 816 | + MxChat_Utils::delete_chunks_for_url($source_label, $bot_id); | |
| 817 | + $result = MxChat_Utils::submit_content_to_db($text, $source_label, $api_key, null, $bot_id, 'document'); | |
| 818 | + | |
| 819 | + // Optional page the document is cited as (plan 025ea2). Stored beside the | |
| 820 | + // entry, not as its identity; an empty field on a re-upload clears it. | |
| 821 | + if (!is_wp_error($result)) { | |
| 822 | + MxChat_Utils::set_upload_source_url( | |
| 823 | + $source_label, | |
| 824 | + $bot_id, | |
| 825 | + isset($_POST['source_page_url']) ? wp_unslash($_POST['source_page_url']) : '' | |
| 826 | + ); | |
| 827 | + } | |
| 828 | + | |
| 829 | + if (is_wp_error($result)) { | |
| 830 | + set_transient('mxchat_admin_notice_error', | |
| 831 | + esc_html__('Failed to import the document: ', 'mxchat') . esc_html($result->get_error_message()), | |
| 832 | + 30 | |
| 833 | + ); | |
| 834 | + } else { | |
| 835 | + set_transient('mxchat_admin_notice_success', | |
| 836 | + sprintf( | |
| 837 | + esc_html__('Document "%s" imported into the knowledge base.', 'mxchat'), | |
| 838 | + esc_html($original_filename) | |
| 839 | + ), | |
| 840 | + 30 | |
| 841 | + ); | |
| 842 | + } | |
| 843 | + | |
| 844 | + wp_safe_redirect(esc_url($redirect_url)); | |
| 845 | + exit; | |
| 846 | +} | |
| 847 | + | |
| 848 | +/** | |
| 849 | + * Validate PDF and count pages with multiple parser attempts | |
| 850 | + */ | |
| 851 | +private function mxchat_validate_and_count_pdf_pages($pdf_path) { | |
| 852 | + // Method 1: Try with Smalot PDF Parser (your current method) | |
| 853 | + try { | |
| 854 | + mxchat_load_pdf_parser(); | |
| 855 | + $parser = new \Smalot\PdfParser\Parser(); | |
| 856 | + $pdf = $parser->parseFile($pdf_path); | |
| 857 | + $pages = $pdf->getPages(); | |
| 858 | + $page_count = count($pages); | |
| 859 | + | |
| 860 | + if ($page_count > 0) { | |
| 861 | + //error_log('PDF parsed successfully with Smalot parser: ' . $page_count . ' pages'); | |
| 862 | + return $page_count; | |
| 863 | + } | |
| 864 | + } catch (Exception $e) { | |
| 865 | + //error_log('Smalot PDF parser failed: ' . $e->getMessage()); | |
| 866 | + } | |
| 867 | + | |
| 868 | + // Method 2: Try with pdfinfo command (if available) | |
| 869 | + if (function_exists('shell_exec') && !$this->mxchat_is_shell_disabled()) { | |
| 870 | + try { | |
| 871 | + $command = 'pdfinfo ' . escapeshellarg($pdf_path) . ' 2>&1'; | |
| 872 | + $output = shell_exec($command); | |
| 873 | + | |
| 874 | + if ($output && preg_match('/Pages:\s*(\d+)/', $output, $matches)) { | |
| 875 | + $page_count = intval($matches[1]); | |
| 876 | + if ($page_count > 0) { | |
| 877 | + //error_log('PDF parsed successfully with pdfinfo: ' . $page_count . ' pages'); | |
| 878 | + return $page_count; | |
| 879 | + } | |
| 880 | + } | |
| 881 | + } catch (Exception $e) { | |
| 882 | + //error_log('pdfinfo command failed: ' . $e->getMessage()); | |
| 883 | + } | |
| 884 | + } | |
| 885 | + | |
| 886 | + // Method 3: Try to repair PDF and parse again | |
| 887 | + try { | |
| 888 | + $repaired_path = $this->mxchat_attempt_pdf_repair($pdf_path); | |
| 889 | + if ($repaired_path && $repaired_path !== $pdf_path) { | |
| 890 | + mxchat_load_pdf_parser(); | |
| 891 | + $parser = new \Smalot\PdfParser\Parser(); | |
| 892 | + $pdf = $parser->parseFile($repaired_path); | |
| 893 | + $pages = $pdf->getPages(); | |
| 894 | + $page_count = count($pages); | |
| 895 | + | |
| 896 | + if ($page_count > 0) { | |
| 897 | + // Replace original with repaired version | |
| 898 | + copy($repaired_path, $pdf_path); | |
| 899 | + unlink($repaired_path); | |
| 900 | + //error_log('PDF repaired and parsed successfully: ' . $page_count . ' pages'); | |
| 901 | + return $page_count; | |
| 902 | + } | |
| 903 | + | |
| 904 | + // Clean up repaired file if it didn't work | |
| 905 | + unlink($repaired_path); | |
| 906 | + } | |
| 907 | + } catch (Exception $e) { | |
| 908 | + //error_log('PDF repair attempt failed: ' . $e->getMessage()); | |
| 909 | + } | |
| 910 | + | |
| 911 | + // Method 4: Manual PDF structure analysis (basic page count) | |
| 912 | + try { | |
| 913 | + $page_count = $this->mxchat_manual_pdf_page_count($pdf_path); | |
| 914 | + if ($page_count > 0) { | |
| 915 | + //error_log('PDF page count determined manually: ' . $page_count . ' pages'); | |
| 916 | + return $page_count; | |
| 917 | + } | |
| 918 | + } catch (Exception $e) { | |
| 919 | + //error_log('Manual PDF analysis failed: ' . $e->getMessage()); | |
| 920 | + } | |
| 921 | + | |
| 922 | + //error_log('All PDF parsing methods failed for: ' . $pdf_path); | |
| 923 | + return false; | |
| 924 | +} | |
| 925 | + | |
| 926 | +/** | |
| 927 | + * Check if shell_exec is disabled | |
| 928 | + */ | |
| 929 | +private function mxchat_is_shell_disabled() { | |
| 930 | + $disabled = explode(',', ini_get('disable_functions')); | |
| 931 | + return in_array('shell_exec', $disabled); | |
| 932 | +} | |
| 933 | + | |
| 934 | +/** | |
| 935 | + * Attempt to repair PDF using basic methods | |
| 936 | + */ | |
| 937 | +private function mxchat_attempt_pdf_repair($pdf_path) { | |
| 938 | + try { | |
| 939 | + $content = file_get_contents($pdf_path); | |
| 940 | + if (!$content) { | |
| 941 | + return false; | |
| 942 | + } | |
| 943 | + | |
| 944 | + // Check if PDF starts with proper header | |
| 945 | + if (substr($content, 0, 4) !== '%PDF') { | |
| 946 | + // Try to find PDF header in the content | |
| 947 | + $header_pos = strpos($content, '%PDF'); | |
| 948 | + if ($header_pos !== false && $header_pos < 1024) { | |
| 949 | + // Remove junk before PDF header | |
| 950 | + $content = substr($content, $header_pos); | |
| 951 | + $repaired_path = $pdf_path . '.repaired'; | |
| 952 | + file_put_contents($repaired_path, $content); | |
| 953 | + return $repaired_path; | |
| 954 | + } | |
| 955 | + } | |
| 956 | + | |
| 957 | + // Check for EOF marker | |
| 958 | + $content = rtrim($content); | |
| 959 | + if (!preg_match('/%%EOF\s*$/', $content)) { | |
| 960 | + // Add EOF marker if missing | |
| 961 | + $content .= "\n%%EOF"; | |
| 962 | + $repaired_path = $pdf_path . '.repaired'; | |
| 963 | + file_put_contents($repaired_path, $content); | |
| 964 | + return $repaired_path; | |
| 965 | + } | |
| 966 | + | |
| 967 | + } catch (Exception $e) { | |
| 968 | + //error_log('PDF repair error: ' . $e->getMessage()); | |
| 969 | + } | |
| 970 | + | |
| 971 | + return false; | |
| 972 | +} | |
| 973 | + | |
| 974 | +/** | |
| 975 | + * Manual PDF page counting by analyzing PDF structure | |
| 976 | + */ | |
| 977 | +private function mxchat_manual_pdf_page_count($pdf_path) { | |
| 978 | + try { | |
| 979 | + $content = file_get_contents($pdf_path); | |
| 980 | + if (!$content) { | |
| 981 | + return 0; | |
| 982 | + } | |
| 983 | + | |
| 984 | + // Method 1: Count /Type /Page objects | |
| 985 | + $page_count = preg_match_all('/\/Type\s*\/Page[^s]/', $content); | |
| 986 | + if ($page_count > 0) { | |
| 987 | + return $page_count; | |
| 988 | + } | |
| 989 | + | |
| 990 | + // Method 2: Look for /Count in pages object | |
| 991 | + if (preg_match('/\/Type\s*\/Pages.*?\/Count\s+(\d+)/', $content, $matches)) { | |
| 992 | + return intval($matches[1]); | |
| 993 | + } | |
| 994 | + | |
| 995 | + // Method 3: Count page references | |
| 996 | + $page_count = preg_match_all('/\d+\s+0\s+obj\s*<<[^>]*\/Type\s*\/Page/', $content); | |
| 997 | + if ($page_count > 0) { | |
| 998 | + return $page_count; | |
| 999 | + } | |
| 1000 | + | |
| 1001 | + } catch (Exception $e) { | |
| 1002 | + //error_log('Manual PDF analysis error: ' . $e->getMessage()); | |
| 1003 | + } | |
| 1004 | + | |
| 1005 | + return 0; | |
| 1006 | +} | |
| 1007 | + | |
| 1008 | + | |
| 1009 | +public function mxchat_save_inline_prompt() { | |
| 1010 | + // DEBUG: Log what we're receiving | |
| 1011 | + //error_log('=== MXCHAT DEBUG ==='); | |
| 1012 | + //error_log('POST data: ' . print_r($_POST, true)); | |
| 1013 | + //error_log('Nonce from POST: ' . ($_POST['_ajax_nonce'] ?? 'NOT FOUND')); | |
| 1014 | + | |
| 1015 | + // Check for nonce security | |
| 1016 | + check_ajax_referer('mxchat_save_inline_nonce', '_ajax_nonce'); | |
| 1017 | + | |
| 1018 | + // If we get here, nonce passed | |
| 1019 | + //error_log('Nonce verification PASSED'); | |
| 1020 | + | |
| 1021 | + // Verify permissions | |
| 1022 | + if (!current_user_can('manage_options')) { | |
| 1023 | + wp_send_json_error(esc_html__('Permission denied.', 'mxchat')); | |
| 1024 | + return; | |
| 1025 | + } | |
| 1026 | + | |
| 1027 | + global $wpdb; | |
| 1028 | + $table_name = $wpdb->prefix . 'mxchat_system_prompt_content'; | |
| 1029 | + | |
| 1030 | + // Validate and sanitize input data | |
| 1031 | + $prompt_id = isset($_POST['id']) ? absint($_POST['id']) : 0; | |
| 1032 | + $article_content = isset($_POST['article_content']) ? wp_kses_post(wp_unslash($_POST['article_content'])) : ''; | |
| 1033 | + $article_url = isset($_POST['article_url']) ? esc_url_raw($_POST['article_url']) : ''; | |
| 1034 | + | |
| 1035 | + if ($prompt_id > 0 && !empty($article_content)) { | |
| 1036 | + // Re-generate the embedding vector for the updated content | |
| 1037 | + $embedding_vector = $this->mxchat_generate_embedding($article_content); | |
| 1038 | + if (is_array($embedding_vector)) { | |
| 1039 | + // Serialize the embedding vector before storing it | |
| 1040 | + $embedding_vector_serialized = serialize($embedding_vector); | |
| 1041 | + // Update the prompt in the database | |
| 1042 | + $updated = $wpdb->update( | |
| 1043 | + $table_name, | |
| 1044 | + array( | |
| 1045 | + 'article_content' => $article_content, | |
| 1046 | + 'embedding_vector' => $embedding_vector_serialized, | |
| 1047 | + 'source_url' => $article_url, | |
| 1048 | + ), | |
| 1049 | + array('id' => $prompt_id), | |
| 1050 | + array('%s', '%s', '%s'), | |
| 1051 | + array('%d') | |
| 1052 | + ); | |
| 1053 | + if ($updated !== false) { | |
| 1054 | + wp_send_json_success(); | |
| 1055 | + } else { | |
| 1056 | + MxChat_Admin::mxchat_log_debug('knowledge_error', 'Database update failed when saving knowledge entry'); | |
| 1057 | + wp_send_json_error(esc_html__('Database update failed.', 'mxchat')); | |
| 1058 | + } | |
| 1059 | + } else { | |
| 1060 | + MxChat_Admin::mxchat_log_debug('embedding_error', 'Embedding generation failed for knowledge entry'); | |
| 1061 | + wp_send_json_error(esc_html__('Embedding generation failed.', 'mxchat')); | |
| 1062 | + } | |
| 1063 | + } else { | |
| 1064 | + wp_send_json_error(esc_html__('Invalid data.', 'mxchat')); | |
| 1065 | + } | |
| 1066 | +} | |
| 1067 | + | |
| 1068 | + | |
| 1069 | +/** | |
| 1070 | + * AJAX: Get full content for editing — reassembles chunks if needed. | |
| 1071 | + * Works for both WordPress DB and Pinecone entries. | |
| 1072 | + */ | |
| 1073 | +/** | |
| 1074 | + * Sanitize a knowledge entry's source_url from an AJAX request WITHOUT destroying | |
| 1075 | + * its identity. sanitize_text_field() strips percent-encoded octets (%20, %D7%A9…), | |
| 1076 | + * so a percent-encoded URL — every non-ASCII permalink — would md5 to a DIFFERENT | |
| 1077 | + * id than the one it was stored under: reads miss the entry and saves write an | |
| 1078 | + * orphan copy while the original keeps its stale text. URLs get esc_url_raw | |
| 1079 | + * (identity-preserving, matches what import stored); non-URL keys (mxchat://, | |
| 1080 | + * _ungrouped_) keep the old sanitizer. | |
| 1081 | + */ | |
| 1082 | +private function sanitize_entry_source_url( $raw ) { | |
| 1083 | + $raw = trim( (string) $raw ); | |
| 1084 | + if ( preg_match( '#^https?://#i', $raw ) ) { | |
| 1085 | + return esc_url_raw( $raw ); | |
| 1086 | + } | |
| 1087 | + return sanitize_text_field( $raw ); | |
| 1088 | +} | |
| 1089 | + | |
| 1090 | +public function ajax_mxchat_get_entry_content() { | |
| 1091 | + check_ajax_referer('mxchat_edit_entry_nonce', 'nonce'); | |
| 1092 | + | |
| 1093 | + if ( ! current_user_can('manage_options') ) { | |
| 1094 | + wp_send_json_error( array( 'message' => 'Permission denied.' ) ); | |
| 1095 | + } | |
| 1096 | + | |
| 1097 | + $source_url = $this->sanitize_entry_source_url( isset($_POST['source_url']) ? wp_unslash($_POST['source_url']) : '' ); | |
| 1098 | + $entry_id = isset($_POST['entry_id']) ? absint($_POST['entry_id']) : 0; | |
| 1099 | + $data_source = isset($_POST['data_source']) ? sanitize_key($_POST['data_source']) : 'wordpress'; | |
| 1100 | + $bot_id = isset($_POST['bot_id']) ? sanitize_key($_POST['bot_id']) : 'default'; | |
| 1101 | + | |
| 1102 | + if ( $data_source === 'pinecone' ) { | |
| 1103 | + // Pinecone ids are strings (md5 hashes, manual_* ids) — absint() would | |
| 1104 | + // destroy them, so re-read the raw value for this branch only. | |
| 1105 | + $vector_id = isset($_POST['entry_id']) ? sanitize_text_field( wp_unslash($_POST['entry_id']) ) : ''; | |
| 1106 | + // Pinecone: fetch vectors by source_url, reassemble chunks | |
| 1107 | + $content = $this->get_pinecone_entry_content( $source_url, $vector_id, $bot_id ); | |
| 1108 | + } else { | |
| 1109 | + // WordPress DB | |
| 1110 | + $content = $this->get_wordpress_entry_content( $source_url, $entry_id ); | |
| 1111 | + } | |
| 1112 | + | |
| 1113 | + if ( is_wp_error( $content ) ) { | |
| 1114 | + wp_send_json_error( array( 'message' => $content->get_error_message() ) ); | |
| 1115 | + } | |
| 1116 | + | |
| 1117 | + wp_send_json_success( $content ); | |
| 1118 | +} | |
| 1119 | + | |
| 1120 | +/** | |
| 1121 | + * Get content from WordPress DB — reassembles chunks by source_url. | |
| 1122 | + */ | |
| 1123 | +private function get_wordpress_entry_content( $source_url, $entry_id ) { | |
| 1124 | + global $wpdb; | |
| 1125 | + $table = $wpdb->prefix . 'mxchat_system_prompt_content'; | |
| 1126 | + | |
| 1127 | + // If we have a source_url, check for chunks | |
| 1128 | + if ( ! empty( $source_url ) && strpos( $source_url, 'mxchat://' ) !== 0 ) { | |
| 1129 | + $rows = $wpdb->get_results( $wpdb->prepare( | |
| 1130 | + "SELECT id, article_content, source_url, content_type FROM {$table} WHERE source_url = %s ORDER BY id ASC", | |
| 1131 | + $source_url | |
| 1132 | + ) ); | |
| 1133 | + | |
| 1134 | + if ( $rows && count( $rows ) > 1 ) { | |
| 1135 | + // Multiple rows = chunked. Reassemble. | |
| 1136 | + $chunks = array(); | |
| 1137 | + foreach ( $rows as $row ) { | |
| 1138 | + $parsed = MxChat_Chunker::parse_stored_chunk( $row->article_content ); | |
| 1139 | + $index = isset( $parsed['metadata']['chunk_index'] ) ? intval( $parsed['metadata']['chunk_index'] ) : count( $chunks ); | |
| 1140 | + $chunks[ $index ] = $parsed['text']; | |
| 1141 | + } | |
| 1142 | + ksort( $chunks ); | |
| 1143 | + return array( | |
| 1144 | + 'content' => implode( "\n\n", $chunks ), | |
| 1145 | + 'source_url' => $source_url, | |
| 1146 | + 'is_chunked' => true, | |
| 1147 | + 'chunk_count' => count( $chunks ), | |
| 1148 | + 'content_type' => $rows[0]->content_type, | |
| 1149 | + ); | |
| 1150 | + } elseif ( $rows && count( $rows ) === 1 ) { | |
| 1151 | + $parsed = MxChat_Chunker::parse_stored_chunk( $rows[0]->article_content ); | |
| 1152 | + return array( | |
| 1153 | + 'content' => $parsed['text'], | |
| 1154 | + 'source_url' => $source_url, | |
| 1155 | + 'entry_id' => $rows[0]->id, | |
| 1156 | + 'is_chunked' => false, | |
| 1157 | + 'content_type' => $rows[0]->content_type, | |
| 1158 | + ); | |
| 1159 | + } | |
| 1160 | + } | |
| 1161 | + | |
| 1162 | + // Fallback: fetch by ID | |
| 1163 | + if ( $entry_id > 0 ) { | |
| 1164 | + $row = $wpdb->get_row( $wpdb->prepare( | |
| 1165 | + "SELECT id, article_content, source_url, content_type FROM {$table} WHERE id = %d", | |
| 1166 | + $entry_id | |
| 1167 | + ) ); | |
| 1168 | + if ( $row ) { | |
| 1169 | + $parsed = MxChat_Chunker::parse_stored_chunk( $row->article_content ); | |
| 1170 | + return array( | |
| 1171 | + 'content' => $parsed['text'], | |
| 1172 | + 'source_url' => $row->source_url, | |
| 1173 | + 'entry_id' => $row->id, | |
| 1174 | + 'is_chunked' => false, | |
| 1175 | + 'content_type' => $row->content_type, | |
| 1176 | + ); | |
| 1177 | + } | |
| 1178 | + } | |
| 1179 | + | |
| 1180 | + return new WP_Error( 'not_found', 'Entry not found.' ); | |
| 1181 | +} | |
| 1182 | + | |
| 1183 | +/** | |
| 1184 | + * Get content from Pinecone — fetches vectors by source_url, reassembles chunks. | |
| 1185 | + */ | |
| 1186 | +private function get_pinecone_entry_content( $source_url, $entry_id, $bot_id ) { | |
| 1187 | + if ( ! class_exists('MxChat_Pinecone_Manager') ) { | |
| 1188 | + return new WP_Error( 'pinecone_unavailable', 'Pinecone manager not available.' ); | |
| 1189 | + } | |
| 1190 | + | |
| 1191 | + // Get Pinecone config | |
| 1192 | + if ( $bot_id === 'default' || ! class_exists('MxChat_Multi_Bot_Manager') ) { | |
| 1193 | + $pinecone_options = get_option('mxchat_pinecone_addon_options'); | |
| 1194 | + $api_key = $pinecone_options['mxchat_pinecone_api_key'] ?? ''; | |
| 1195 | + $host = $pinecone_options['mxchat_pinecone_host'] ?? ''; | |
| 1196 | + $namespace = $pinecone_options['mxchat_pinecone_namespace'] ?? ''; | |
| 1197 | + } else { | |
| 1198 | + $bot_config = apply_filters('mxchat_get_bot_pinecone_config', array(), $bot_id); | |
| 1199 | + $api_key = $bot_config['api_key'] ?? ''; | |
| 1200 | + $host = $bot_config['host'] ?? ''; | |
| 1201 | + $namespace = $bot_config['namespace'] ?? ''; | |
| 1202 | + } | |
| 1203 | + | |
| 1204 | + if ( empty($host) || empty($api_key) ) { | |
| 1205 | + return new WP_Error( 'pinecone_config', 'Pinecone not configured.' ); | |
| 1206 | + } | |
| 1207 | + | |
| 1208 | + // Document index (plan 362c31): same ids, read through the documents API. | |
| 1209 | + if ( class_exists('MxChat_Pinecone_Documents') && MxChat_Pinecone_Documents::is_document_index( $bot_id ) ) { | |
| 1210 | + $docs_cfg = array( 'api_key' => $api_key, 'host' => $host, 'namespace' => $namespace ); | |
| 1211 | + if ( ( empty($source_url) || strpos($source_url, '_ungrouped_') === 0 ) && ! empty($entry_id) && is_string($entry_id) ) { | |
| 1212 | + $doc_ids = array( $entry_id ); | |
| 1213 | + } else { | |
| 1214 | + $doc_ids = array_merge( array( md5( $source_url ) ), MxChat_Pinecone_Documents::list_ids( md5( $source_url ) . '_chunk_', $docs_cfg ) ); | |
| 1215 | + } | |
| 1216 | + $docs = MxChat_Pinecone_Documents::fetch( $doc_ids, $docs_cfg, array( 'text', 'chunk_index', 'type' ) ); | |
| 1217 | + if ( empty($docs) ) { | |
| 1218 | + return new WP_Error( 'not_found', 'Entry not found in Pinecone.' ); | |
| 1219 | + } | |
| 1220 | + $doc_chunks = array(); | |
| 1221 | + $doc_type = 'content'; | |
| 1222 | + foreach ( $docs as $doc ) { | |
| 1223 | + $doc_meta = $doc['metadata'] ?? array(); | |
| 1224 | + $doc_chunks[ intval( $doc_meta['chunk_index'] ?? 0 ) ] = $doc_meta['text'] ?? ''; | |
| 1225 | + $doc_type = $doc_meta['type'] ?? 'content'; | |
| 1226 | + } | |
| 1227 | + ksort( $doc_chunks ); | |
| 1228 | + return array( | |
| 1229 | + 'content' => implode( "\n\n", $doc_chunks ), | |
| 1230 | + 'source_url' => $source_url, | |
| 1231 | + 'is_chunked' => count($doc_chunks) > 1, | |
| 1232 | + 'chunk_count' => count($doc_chunks), | |
| 1233 | + 'content_type' => $doc_type, | |
| 1234 | + ); | |
| 1235 | + } | |
| 1236 | + | |
| 1237 | + // Manual entries carry no source_url (their vector id is a minted manual_* string, | |
| 1238 | + // not md5 of anything the row can hand us) — fetch the exact vector instead. | |
| 1239 | + // '_ungrouped_' is the table view's synthetic display key for such rows. | |
| 1240 | + if ( ( empty($source_url) || strpos($source_url, '_ungrouped_') === 0 ) && ! empty($entry_id) && is_string($entry_id) ) { | |
| 1241 | + $vector_ids = array( $entry_id ); | |
| 1242 | + } else { | |
| 1243 | + // List vectors with the source_url prefix | |
| 1244 | + $base_id = md5( $source_url ); | |
| 1245 | + $vector_ids = array( $base_id ); | |
| 1246 | + | |
| 1247 | + // Find chunk vectors | |
| 1248 | + // NOTE: Pinecone's /vectors/list is a GET endpoint with query parameters. | |
| 1249 | + // A POST is answered 200-with-an-empty-body, which reads as "no vectors". | |
| 1250 | + $list_url = "https://{$host}/vectors/list"; | |
| 1251 | + $list_params = array( 'prefix' => $base_id . '_chunk_', 'limit' => 100 ); | |
| 1252 | + if ( ! empty($namespace) ) { | |
| 1253 | + $list_params['namespace'] = $namespace; | |
| 1254 | + } | |
| 1255 | + | |
| 1256 | + $list_resp = wp_remote_get( $list_url . '?' . http_build_query( $list_params ), array( | |
| 1257 | + 'headers' => array( 'Api-Key' => $api_key, 'accept' => 'application/json' ), | |
| 1258 | + 'timeout' => 15, | |
| 1259 | + ) ); | |
| 1260 | + | |
| 1261 | + if ( ! is_wp_error($list_resp) ) { | |
| 1262 | + $list_data = json_decode( wp_remote_retrieve_body($list_resp), true ); | |
| 1263 | + if ( ! empty($list_data['vectors']) ) { | |
| 1264 | + foreach ( $list_data['vectors'] as $v ) { | |
| 1265 | + $vector_ids[] = $v['id']; | |
| 1266 | + } | |
| 1267 | + } | |
| 1268 | + } | |
| 1269 | + } | |
| 1270 | + | |
| 1271 | + // Fetch vectors with metadata | |
| 1272 | + // NOTE: /vectors/fetch is a GET endpoint too, and Pinecone expects the ids | |
| 1273 | + // repeated (ids=a&ids=b) — http_build_query would emit ids[0]=a, so build | |
| 1274 | + // the query string explicitly. | |
| 1275 | + $fetch_query = array(); | |
| 1276 | + foreach ( $vector_ids as $fetch_vid ) { | |
| 1277 | + $fetch_query[] = 'ids=' . rawurlencode( $fetch_vid ); | |
| 1278 | + } | |
| 1279 | + if ( ! empty($namespace) ) { | |
| 1280 | + $fetch_query[] = 'namespace=' . rawurlencode( $namespace ); | |
| 1281 | + } | |
| 1282 | + | |
| 1283 | + $fetch_resp = wp_remote_get( "https://{$host}/vectors/fetch?" . implode( '&', $fetch_query ), array( | |
| 1284 | + 'headers' => array( 'Api-Key' => $api_key, 'accept' => 'application/json' ), | |
| 1285 | + 'timeout' => 15, | |
| 1286 | + ) ); | |
| 1287 | + | |
| 1288 | + if ( is_wp_error($fetch_resp) ) { | |
| 1289 | + return new WP_Error( 'pinecone_fetch', 'Failed to fetch from Pinecone.' ); | |
| 1290 | + } | |
| 1291 | + | |
| 1292 | + $fetch_data = json_decode( wp_remote_retrieve_body($fetch_resp), true ); | |
| 1293 | + $vectors = $fetch_data['vectors'] ?? array(); | |
| 1294 | + | |
| 1295 | + if ( empty($vectors) ) { | |
| 1296 | + return new WP_Error( 'not_found', 'Entry not found in Pinecone.' ); | |
| 1297 | + } | |
| 1298 | + | |
| 1299 | + // Reassemble chunks | |
| 1300 | + $chunks = array(); | |
| 1301 | + $content_type = 'content'; | |
| 1302 | + foreach ( $vectors as $vid => $vector ) { | |
| 1303 | + $meta = $vector['metadata'] ?? array(); | |
| 1304 | + $text = $meta['text'] ?? ''; | |
| 1305 | + $index = $meta['chunk_index'] ?? 0; | |
| 1306 | + $content_type = $meta['type'] ?? 'content'; | |
| 1307 | + $chunks[ intval($index) ] = $text; | |
| 1308 | + } | |
| 1309 | + ksort( $chunks ); | |
| 1310 | + | |
| 1311 | + return array( | |
| 1312 | + 'content' => implode( "\n\n", $chunks ), | |
| 1313 | + 'source_url' => $source_url, | |
| 1314 | + 'is_chunked' => count($chunks) > 1, | |
| 1315 | + 'chunk_count' => count($chunks), | |
| 1316 | + 'content_type' => $content_type, | |
| 1317 | + ); | |
| 1318 | +} | |
| 1319 | + | |
| 1320 | +/** | |
| 1321 | + * AJAX: Inspect a knowledge entry — returns the per-chunk stored text + metadata | |
| 1322 | + * WITHOUT collapsing it, so a site owner can see exactly what was indexed for an | |
| 1323 | + * entry (plan-mxchat-20260628-d8cb4b). READ-ONLY: never re-embeds or mutates. | |
| 1324 | + */ | |
| 1325 | +public function ajax_mxchat_inspect_entry() { | |
| 1326 | + check_ajax_referer('mxchat_inspect_entry_nonce', 'nonce'); | |
| 1327 | + | |
| 1328 | + if ( ! current_user_can('manage_options') ) { | |
| 1329 | + wp_send_json_error( array( 'message' => esc_html__('Permission denied.', 'mxchat') ) ); | |
| 1330 | + } | |
| 1331 | + | |
| 1332 | + // Same identity-preserving read the Edit path uses: sanitize_text_field() | |
| 1333 | + // strips percent-encoded octets, so a non-ASCII permalink would look up a | |
| 1334 | + // different string (WordPress DB) and md5 to a different id (Pinecone). | |
| 1335 | + $source_url = $this->sanitize_entry_source_url( isset($_POST['source_url']) ? wp_unslash($_POST['source_url']) : '' ); | |
| 1336 | + $entry_id = isset($_POST['entry_id']) ? absint($_POST['entry_id']) : 0; | |
| 1337 | + $data_source = isset($_POST['data_source']) ? sanitize_key($_POST['data_source']) : 'wordpress'; | |
| 1338 | + $bot_id = isset($_POST['bot_id']) ? sanitize_key($_POST['bot_id']) : 'default'; | |
| 1339 | + | |
| 1340 | + if ( $data_source === 'pinecone' ) { | |
| 1341 | + // Pinecone ids are strings (md5 hashes, manual_* ids) — absint() would | |
| 1342 | + // destroy them, so re-read the raw value for this branch only. | |
| 1343 | + $vector_id = isset($_POST['entry_id']) ? sanitize_text_field( wp_unslash($_POST['entry_id']) ) : ''; | |
| 1344 | + $result = $this->inspect_pinecone_entry( $source_url, $vector_id, $bot_id ); | |
| 1345 | + } else { | |
| 1346 | + $result = $this->inspect_wordpress_entry( $source_url, $entry_id ); | |
| 1347 | + } | |
| 1348 | + | |
| 1349 | + if ( is_wp_error( $result ) ) { | |
| 1350 | + wp_send_json_error( array( 'message' => $result->get_error_message() ) ); | |
| 1351 | + } | |
| 1352 | + | |
| 1353 | + wp_send_json_success( $result ); | |
| 1354 | +} | |
| 1355 | + | |
| 1356 | +/** | |
| 1357 | + * Read-only inspector for WordPress-DB entries. Mirrors get_wordpress_entry_content() | |
| 1358 | + * but returns each STORED chunk's exact text + length (no implode), plus the assembled | |
| 1359 | + * embedded text. This shows what is actually in the index, not a re-derivation from the post. | |
| 1360 | + */ | |
| 1361 | +private function inspect_wordpress_entry( $source_url, $entry_id ) { | |
| 1362 | + global $wpdb; | |
| 1363 | + $table = $wpdb->prefix . 'mxchat_system_prompt_content'; | |
| 1364 | + | |
| 1365 | + $rows = array(); | |
| 1366 | + | |
| 1367 | + // Group by the real stored source_url — this INCLUDES "mxchat://" manual | |
| 1368 | + // Direct Content entries (the spec's manual-entry case), which share one | |
| 1369 | + // source_url across their chunk rows. Only the synthetic "_ungrouped_<id>" | |
| 1370 | + // display key (invented by the table view for rows with no source_url) is | |
| 1371 | + // excluded; those fall through to the entry_id lookup below. | |
| 1372 | + if ( ! empty( $source_url ) && strpos( $source_url, '_ungrouped_' ) !== 0 ) { | |
| 1373 | + $rows = $wpdb->get_results( $wpdb->prepare( | |
| 1374 | + "SELECT id, article_content, source_url, content_type FROM {$table} WHERE source_url = %s ORDER BY id ASC", | |
| 1375 | + $source_url | |
| 1376 | + ) ); | |
| 1377 | + } | |
| 1378 | + | |
| 1379 | + // Fallback / manual "Direct Content" entries: fetch the single row by id. | |
| 1380 | + if ( empty( $rows ) && $entry_id > 0 ) { | |
| 1381 | + $row = $wpdb->get_row( $wpdb->prepare( | |
| 1382 | + "SELECT id, article_content, source_url, content_type FROM {$table} WHERE id = %d", | |
| 1383 | + $entry_id | |
| 1384 | + ) ); | |
| 1385 | + if ( $row ) { | |
| 1386 | + $rows = array( $row ); | |
| 1387 | + } | |
| 1388 | + } | |
| 1389 | + | |
| 1390 | + if ( empty( $rows ) ) { | |
| 1391 | + return new WP_Error( 'not_found', esc_html__('Entry not found in the local knowledge database.', 'mxchat') ); | |
| 1392 | + } | |
| 1393 | + | |
| 1394 | + $chunks = array(); | |
| 1395 | + $content_type = ''; | |
| 1396 | + foreach ( $rows as $row ) { | |
| 1397 | + $parsed = MxChat_Chunker::parse_stored_chunk( $row->article_content ); | |
| 1398 | + $text = isset( $parsed['text'] ) ? $parsed['text'] : ''; | |
| 1399 | + $index = isset( $parsed['metadata']['chunk_index'] ) ? intval( $parsed['metadata']['chunk_index'] ) : count( $chunks ); | |
| 1400 | + $content_type = $row->content_type; | |
| 1401 | + $chunks[] = array( | |
| 1402 | + 'index' => $index, | |
| 1403 | + 'text' => $text, | |
| 1404 | + 'length' => function_exists('mb_strlen') ? mb_strlen( $text ) : strlen( $text ), | |
| 1405 | + 'row_id' => intval( $row->id ), | |
| 1406 | + ); | |
| 1407 | + } | |
| 1408 | + | |
| 1409 | + usort( $chunks, function( $a, $b ) { return $a['index'] - $b['index']; } ); | |
| 1410 | + | |
| 1411 | + $assembled = implode( "\n\n", wp_list_pluck( $chunks, 'text' ) ); | |
| 1412 | + | |
| 1413 | + return array( | |
| 1414 | + 'store' => 'wordpress', | |
| 1415 | + 'source_url' => $source_url, | |
| 1416 | + 'content_type' => $content_type, | |
| 1417 | + 'is_chunked' => count( $chunks ) > 1, | |
| 1418 | + 'chunk_count' => count( $chunks ), | |
| 1419 | + 'assembled' => $assembled, | |
| 1420 | + 'assembled_length' => function_exists('mb_strlen') ? mb_strlen( $assembled ) : strlen( $assembled ), | |
| 1421 | + 'chunks' => array_values( $chunks ), | |
| 1422 | + // WP-DB storage carries no separate vector metadata; surface that fact | |
| 1423 | + // rather than letting the owner guess (the spec's taxonomy question). | |
| 1424 | + 'metadata' => array(), | |
| 1425 | + 'metadata_note' => esc_html__('Stored in the local WordPress database. Only the assembled text shown here is embedded — there are no separate vector metadata fields (e.g. taxonomy terms are not stored unless they were injected into the text itself).', 'mxchat'), | |
| 1426 | + ); | |
| 1427 | +} | |
| 1428 | + | |
| 1429 | +/** | |
| 1430 | + * Read-only inspector for Pinecone entries. Mirrors get_pinecone_entry_content() | |
| 1431 | + * but keeps each vector's text + metadata instead of imploding, so the owner can | |
| 1432 | + * confirm exactly which metadata fields (text/source_url/type/last_updated/created_at/bot_id) | |
| 1433 | + * are present per chunk. READ-ONLY. | |
| 1434 | + */ | |
| 1435 | +private function inspect_pinecone_entry( $source_url, $entry_id, $bot_id ) { | |
| 1436 | + if ( ! class_exists('MxChat_Pinecone_Manager') ) { | |
| 1437 | + return new WP_Error( 'pinecone_unavailable', esc_html__('Pinecone manager not available.', 'mxchat') ); | |
| 1438 | + } | |
| 1439 | + | |
| 1440 | + if ( $bot_id === 'default' || ! class_exists('MxChat_Multi_Bot_Manager') ) { | |
| 1441 | + $pinecone_options = get_option('mxchat_pinecone_addon_options'); | |
| 1442 | + $api_key = $pinecone_options['mxchat_pinecone_api_key'] ?? ''; | |
| 1443 | + $host = $pinecone_options['mxchat_pinecone_host'] ?? ''; | |
| 1444 | + $namespace = $pinecone_options['mxchat_pinecone_namespace'] ?? ''; | |
| 1445 | + } else { | |
| 1446 | + $bot_config = apply_filters('mxchat_get_bot_pinecone_config', array(), $bot_id); | |
| 1447 | + $api_key = $bot_config['api_key'] ?? ''; | |
| 1448 | + $host = $bot_config['host'] ?? ''; | |
| 1449 | + $namespace = $bot_config['namespace'] ?? ''; | |
| 1450 | + } | |
| 1451 | + | |
| 1452 | + if ( empty($host) || empty($api_key) ) { | |
| 1453 | + return new WP_Error( 'pinecone_config', esc_html__('Pinecone not configured.', 'mxchat') ); | |
| 1454 | + } | |
| 1455 | + | |
| 1456 | + // Document index (plan 362c31): the documents API hands back the same | |
| 1457 | + // id => {metadata} map the classic fetch does, so the inspection below is shared. | |
| 1458 | + // Manual entries carry no source_url (their vector id is a minted manual_* string, | |
| 1459 | + // not md5 of anything the row can hand us) — fetch the exact vector instead, the | |
| 1460 | + // way get_pinecone_entry_content() does. '_ungrouped_' is the table view's | |
| 1461 | + // synthetic display key for such rows. | |
| 1462 | + $exact_id = ( ( empty($source_url) || strpos($source_url, '_ungrouped_') === 0 ) && ! empty($entry_id) && is_string($entry_id) ); | |
| 1463 | + | |
| 1464 | + if ( class_exists('MxChat_Pinecone_Documents') && MxChat_Pinecone_Documents::is_document_index( $bot_id ) ) { | |
| 1465 | + $docs_cfg = array( 'api_key' => $api_key, 'host' => $host, 'namespace' => $namespace ); | |
| 1466 | + if ( $exact_id ) { | |
| 1467 | + $vector_ids = array( $entry_id ); | |
| 1468 | + } else { | |
| 1469 | + $vector_ids = array_merge( array( md5( $source_url ) ), MxChat_Pinecone_Documents::list_ids( md5( $source_url ) . '_chunk_', $docs_cfg ) ); | |
| 1470 | + } | |
| 1471 | + $vectors = MxChat_Pinecone_Documents::fetch( $vector_ids, $docs_cfg ); | |
| 1472 | + if ( empty($vectors) ) { | |
| 1473 | + return new WP_Error( 'not_found', esc_html__('Entry not found in Pinecone.', 'mxchat') ); | |
| 1474 | + } | |
| 1475 | + } else { | |
| 1476 | + if ( $exact_id ) { | |
| 1477 | + $vector_ids = array( $entry_id ); | |
| 1478 | + } else { | |
| 1479 | + $base_id = md5( $source_url ); | |
| 1480 | + $vector_ids = array( $base_id ); | |
| 1481 | + | |
| 1482 | + // NOTE: Pinecone's /vectors/list is a GET endpoint with query parameters. | |
| 1483 | + // A POST is answered 200-with-an-empty-body, which reads as "no vectors". | |
| 1484 | + $list_url = "https://{$host}/vectors/list"; | |
| 1485 | + $list_params = array( 'prefix' => $base_id . '_chunk_', 'limit' => 100 ); | |
| 1486 | + if ( ! empty($namespace) ) { | |
| 1487 | + $list_params['namespace'] = $namespace; | |
| 1488 | + } | |
| 1489 | + | |
| 1490 | + $list_resp = wp_remote_get( $list_url . '?' . http_build_query( $list_params ), array( | |
| 1491 | + 'headers' => array( 'Api-Key' => $api_key, 'accept' => 'application/json' ), | |
| 1492 | + 'timeout' => 15, | |
| 1493 | + ) ); | |
| 1494 | + | |
| 1495 | + if ( ! is_wp_error($list_resp) ) { | |
| 1496 | + $list_data = json_decode( wp_remote_retrieve_body($list_resp), true ); | |
| 1497 | + if ( ! empty($list_data['vectors']) ) { | |
| 1498 | + foreach ( $list_data['vectors'] as $v ) { | |
| 1499 | + $vector_ids[] = $v['id']; | |
| 1500 | + } | |
| 1501 | + } | |
| 1502 | + } | |
| 1503 | + } // end md5(source_url) + chunk-prefix ids | |
| 1504 | + | |
| 1505 | + // NOTE: /vectors/fetch is a GET endpoint too, and Pinecone expects the ids | |
| 1506 | + // repeated (ids=a&ids=b) — http_build_query would emit ids[0]=a, so build | |
| 1507 | + // the query string explicitly. | |
| 1508 | + $fetch_query = array(); | |
| 1509 | + foreach ( $vector_ids as $fetch_vid ) { | |
| 1510 | + $fetch_query[] = 'ids=' . rawurlencode( $fetch_vid ); | |
| 1511 | + } | |
| 1512 | + if ( ! empty($namespace) ) { | |
| 1513 | + $fetch_query[] = 'namespace=' . rawurlencode( $namespace ); | |
| 1514 | + } | |
| 1515 | + | |
| 1516 | + $fetch_resp = wp_remote_get( "https://{$host}/vectors/fetch?" . implode( '&', $fetch_query ), array( | |
| 1517 | + 'headers' => array( 'Api-Key' => $api_key, 'accept' => 'application/json' ), | |
| 1518 | + 'timeout' => 15, | |
| 1519 | + ) ); | |
| 1520 | + | |
| 1521 | + if ( is_wp_error($fetch_resp) ) { | |
| 1522 | + return new WP_Error( 'pinecone_fetch', esc_html__('Failed to fetch from Pinecone.', 'mxchat') ); | |
| 1523 | + } | |
| 1524 | + | |
| 1525 | + $fetch_data = json_decode( wp_remote_retrieve_body($fetch_resp), true ); | |
| 1526 | + $vectors = $fetch_data['vectors'] ?? array(); | |
| 1527 | + | |
| 1528 | + if ( empty($vectors) ) { | |
| 1529 | + return new WP_Error( 'not_found', esc_html__('Entry not found in Pinecone.', 'mxchat') ); | |
| 1530 | + } | |
| 1531 | + } // end classic (vector index) read | |
| 1532 | + | |
| 1533 | + // Whitelisted metadata fields the spec calls out — shown so devs can confirm | |
| 1534 | + // what is (and is NOT) stored per vector. | |
| 1535 | + $meta_fields = array( 'text', 'source_url', 'type', 'last_updated', 'created_at', 'bot_id', 'chunk_index', 'total_chunks' ); | |
| 1536 | + $chunks = array(); | |
| 1537 | + $content_type = ''; | |
| 1538 | + foreach ( $vectors as $vid => $vector ) { | |
| 1539 | + $meta = isset($vector['metadata']) && is_array($vector['metadata']) ? $vector['metadata'] : array(); | |
| 1540 | + $text = $meta['text'] ?? ''; | |
| 1541 | + $index = isset($meta['chunk_index']) ? intval($meta['chunk_index']) : count($chunks); | |
| 1542 | + $content_type = $meta['type'] ?? $content_type; | |
| 1543 | + | |
| 1544 | + $clean_meta = array(); | |
| 1545 | + foreach ( $meta_fields as $field ) { | |
| 1546 | + if ( array_key_exists( $field, $meta ) && $field !== 'text' ) { | |
| 1547 | + $clean_meta[ $field ] = is_scalar( $meta[ $field ] ) ? (string) $meta[ $field ] : wp_json_encode( $meta[ $field ] ); | |
| 1548 | + } | |
| 1549 | + } | |
| 1550 | + | |
| 1551 | + $chunks[] = array( | |
| 1552 | + 'index' => $index, | |
| 1553 | + 'text' => $text, | |
| 1554 | + 'length' => function_exists('mb_strlen') ? mb_strlen( $text ) : strlen( $text ), | |
| 1555 | + 'vector_id' => (string) $vid, | |
| 1556 | + 'metadata' => $clean_meta, | |
| 1557 | + ); | |
| 1558 | + } | |
| 1559 | + | |
| 1560 | + usort( $chunks, function( $a, $b ) { return $a['index'] - $b['index']; } ); | |
| 1561 | + | |
| 1562 | + $assembled = implode( "\n\n", wp_list_pluck( $chunks, 'text' ) ); | |
| 1563 | + | |
| 1564 | + return array( | |
| 1565 | + 'store' => 'pinecone', | |
| 1566 | + 'source_url' => $source_url, | |
| 1567 | + 'content_type' => $content_type, | |
| 1568 | + 'is_chunked' => count( $chunks ) > 1, | |
| 1569 | + 'chunk_count' => count( $chunks ), | |
| 1570 | + 'assembled' => $assembled, | |
| 1571 | + 'assembled_length' => function_exists('mb_strlen') ? mb_strlen( $assembled ) : strlen( $assembled ), | |
| 1572 | + 'chunks' => array_values( $chunks ), | |
| 1573 | + 'metadata' => array(), | |
| 1574 | + 'metadata_note' => esc_html__('Stored in Pinecone. Each chunk above lists the vector metadata fields actually present — if a field you expect (such as taxonomy terms) is missing here, it was not stored as metadata and is only searchable if it appears in the embedded text.', 'mxchat'), | |
| 1575 | + ); | |
| 1576 | +} | |
| 1577 | + | |
| 1578 | +/** | |
| 1579 | + * AJAX: Save edited content — re-chunks and re-embeds as needed. | |
| 1580 | + * Works for both WordPress DB and Pinecone entries. | |
| 1581 | + */ | |
| 1582 | +public function ajax_mxchat_save_entry_content() { | |
| 1583 | + check_ajax_referer('mxchat_edit_entry_nonce', 'nonce'); | |
| 1584 | + | |
| 1585 | + if ( ! current_user_can('manage_options') ) { | |
| 1586 | + wp_send_json_error( array( 'message' => 'Permission denied.' ) ); | |
| 1587 | + } | |
| 1588 | + | |
| 1589 | + $source_url = $this->sanitize_entry_source_url( isset($_POST['source_url']) ? wp_unslash($_POST['source_url']) : '' ); | |
| 1590 | + $entry_id = isset($_POST['entry_id']) ? absint($_POST['entry_id']) : 0; | |
| 1591 | + $content = isset($_POST['content']) ? wp_kses_post( wp_unslash($_POST['content']) ) : ''; | |
| 1592 | + $data_source = isset($_POST['data_source']) ? sanitize_key($_POST['data_source']) : 'wordpress'; | |
| 1593 | + $bot_id = isset($_POST['bot_id']) ? sanitize_key($_POST['bot_id']) : 'default'; | |
| 1594 | + $content_type = isset($_POST['content_type']) ? sanitize_key($_POST['content_type']) : 'content'; | |
| 1595 | + | |
| 1596 | + if ( empty($content) ) { | |
| 1597 | + wp_send_json_error( array( 'message' => 'Content cannot be empty.' ) ); | |
| 1598 | + } | |
| 1599 | + | |
| 1600 | + // Get the embedding API key | |
| 1601 | + $options = get_option('mxchat_options', array()); | |
| 1602 | + $api_key = ''; | |
| 1603 | + | |
| 1604 | + if ( $bot_id !== 'default' && class_exists('MxChat_Multi_Bot_Manager') ) { | |
| 1605 | + $bot_options = apply_filters('mxchat_get_bot_options', array(), $bot_id); | |
| 1606 | + $api_key = $bot_options['api_key'] ?? ''; | |
| 1607 | + } | |
| 1608 | + if ( empty($api_key) ) { | |
| 1609 | + $api_key = $options['api_key'] ?? ''; | |
| 1610 | + } | |
| 1611 | + | |
| 1612 | + if ( $data_source === 'pinecone' ) { | |
| 1613 | + // Pinecone branch. The WP-DB manual-entry delete below must never run here: | |
| 1614 | + // Pinecone ids are strings, and absint() on a digit-leading md5 hash would | |
| 1615 | + // yield a real (unrelated) WP row id. | |
| 1616 | + $raw_vector_id = isset($_POST['entry_id']) ? sanitize_text_field( wp_unslash($_POST['entry_id']) ) : ''; | |
| 1617 | + $is_manual_single = empty($source_url) || strpos($source_url, '_ungrouped_') === 0; | |
| 1618 | + $is_manual_chunked = strpos($source_url, 'mxchat://') === 0; | |
| 1619 | + | |
| 1620 | + if ( $is_manual_single || $is_manual_chunked ) { | |
| 1621 | + // Manual content: remove the old vectors first, then store as fresh manual | |
| 1622 | + // content — submit_content_to_db mints a new unique identity (manual_* id | |
| 1623 | + // for a single vector, an mxchat:// chunk prefix if it now chunks). | |
| 1624 | + if ( $is_manual_chunked ) { | |
| 1625 | + // Minted identity: base + chunk vectors share the md5(mxchat://...) prefix. | |
| 1626 | + MxChat_Utils::delete_chunks_for_url( $source_url, $bot_id ); | |
| 1627 | + } elseif ( ! empty($raw_vector_id) && class_exists('MxChat_Pinecone_Manager') ) { | |
| 1628 | + $pinecone_manager = MxChat_Pinecone_Manager::get_instance(); | |
| 1629 | + $pinecone_options = $pinecone_manager->mxchat_get_bot_pinecone_options( $bot_id ); | |
| 1630 | + if ( ! empty($pinecone_options['mxchat_pinecone_api_key']) && ! empty($pinecone_options['mxchat_pinecone_host']) ) { | |
| 1631 | + $pinecone_manager->mxchat_delete_from_pinecone_by_vector_id( | |
| 1632 | + $raw_vector_id, | |
| 1633 | + $pinecone_options['mxchat_pinecone_api_key'], | |
| 1634 | + $pinecone_options['mxchat_pinecone_host'], | |
| 1635 | + $pinecone_options['mxchat_pinecone_namespace'] ?? '' | |
| 1636 | + ); | |
| 1637 | + } | |
| 1638 | + } | |
| 1639 | + $result = MxChat_Utils::submit_content_to_db( $content, '', $api_key, null, $bot_id, $content_type ); | |
| 1640 | + } else { | |
| 1641 | + // URL-sourced entry: identity is md5(source_url). submit_content_to_db | |
| 1642 | + // handles delete-old-chunks → re-chunk → re-embed → store, and sweeps | |
| 1643 | + // stale chunk vectors when the content now fits in a single vector. | |
| 1644 | + $result = MxChat_Utils::submit_content_to_db( $content, $source_url, $api_key, md5($source_url), $bot_id, $content_type ); | |
| 1645 | + } | |
| 1646 | + } else { | |
| 1647 | + global $wpdb; | |
| 1648 | + $table = $wpdb->prefix . 'mxchat_system_prompt_content'; | |
| 1649 | + | |
| 1650 | + // If source_url is empty but we have an entry_id, look it up | |
| 1651 | + if ( empty($source_url) && $entry_id > 0 ) { | |
| 1652 | + $row = $wpdb->get_row( $wpdb->prepare( "SELECT source_url FROM {$table} WHERE id = %d", $entry_id ) ); | |
| 1653 | + if ( $row && ! empty($row->source_url) ) { | |
| 1654 | + $source_url = $row->source_url; | |
| 1655 | + } | |
| 1656 | + } | |
| 1657 | + | |
| 1658 | + // For manual entries (no source_url or mxchat:// prefix), delete the old entry by ID first | |
| 1659 | + // so submit_content_to_db creates a replacement instead of a duplicate | |
| 1660 | + // Also treat legacy mxchat.ai source URLs as manual — old bug assigned the site URL to manual entries | |
| 1661 | + $is_legacy_manual = !empty($source_url) && strpos($source_url, 'mxchat.ai') !== false && strpos($source_url, 'mxchat://') !== 0; | |
| 1662 | + if ( $entry_id > 0 && (empty($source_url) || strpos($source_url, 'mxchat://') === 0 || $is_legacy_manual) ) { | |
| 1663 | + $wpdb->delete( $table, array( 'id' => $entry_id ), array( '%d' ) ); | |
| 1664 | + // Clear legacy URL so submit_content_to_db generates a unique mxchat:// identifier | |
| 1665 | + // instead of reusing the shared URL (which would mass-delete other entries with the same URL) | |
| 1666 | + if ( $is_legacy_manual ) { | |
| 1667 | + $source_url = ''; | |
| 1668 | + } | |
| 1669 | + } | |
| 1670 | + | |
| 1671 | + // Use the existing submit_content_to_db which handles chunking, Pinecone, and WP DB | |
| 1672 | + $vector_id = ! empty($source_url) ? md5($source_url) : md5('mxchat_manual_' . $entry_id); | |
| 1673 | + | |
| 1674 | + // submit_content_to_db already handles: delete old chunks → re-chunk → re-embed → store | |
| 1675 | + $result = MxChat_Utils::submit_content_to_db( $content, $source_url, $api_key, $vector_id, $bot_id, $content_type ); | |
| 1676 | + } | |
| 1677 | + | |
| 1678 | + if ( is_wp_error($result) ) { | |
| 1679 | + wp_send_json_error( array( 'message' => $result->get_error_message() ) ); | |
| 1680 | + } | |
| 1681 | + | |
| 1682 | + wp_send_json_success( array( 'message' => 'Content saved and re-embedded successfully.' ) ); | |
| 1683 | +} | |
| 1684 | + | |
| 1685 | +public function mxchat_get_pdf_processing_status($pdf_url) { | |
| 1686 | + $pdf_url = esc_url_raw($pdf_url); | |
| 1687 | + $status = get_transient(sanitize_key('mxchat_pdf_status_' . md5($pdf_url))); | |
| 1688 | + | |
| 1689 | + if (!$status || !is_array($status)) { | |
| 1690 | + return false; | |
| 1691 | + } | |
| 1692 | + | |
| 1693 | + // Check for stalled processing (no updates for 5 minutes) | |
| 1694 | + if ($status['status'] === 'processing' && (time() - absint($status['last_update'])) > 300) { | |
| 1695 | + $status['status'] = 'error'; | |
| 1696 | + $status['error'] = __('PDF processing appears to be stalled. No updates for over 5 minutes.', 'mxchat'); | |
| 1697 | + | |
| 1698 | + // Save the updated status | |
| 1699 | + set_transient( | |
| 1700 | + sanitize_key('mxchat_pdf_status_' . md5($pdf_url)), | |
| 1701 | + array_map('sanitize_text_field', $status), | |
| 1702 | + DAY_IN_SECONDS | |
| 1703 | + ); | |
| 1704 | + } | |
| 1705 | + | |
| 1706 | + $result = array( | |
| 1707 | + 'total_pages' => absint($status['total_pages']), | |
| 1708 | + 'processed_pages' => absint($status['processed_pages']), | |
| 1709 | + 'failed_pages' => absint($status['failed_pages'] ?? 0), | |
| 1710 | + 'percentage' => ($status['total_pages'] > 0) | |
| 1711 | + ? round((absint($status['processed_pages']) / absint($status['total_pages'])) * 100) | |
| 1712 | + : 0, | |
| 1713 | + 'status' => sanitize_text_field($status['status']), | |
| 1714 | + 'last_update' => human_time_diff(absint($status['last_update']), time()) . ' ' . esc_html__('ago', 'mxchat'), | |
| 1715 | + 'failed_pages_list' => isset($status['failed_pages_list']) ? $status['failed_pages_list'] : array(), | |
| 1716 | + 'completion_summary' => isset($status['completion_summary']) ? $status['completion_summary'] : null | |
| 1717 | + ); | |
| 1718 | + | |
| 1719 | + // Add error message if present | |
| 1720 | + if (isset($status['error']) && !empty($status['error'])) { | |
| 1721 | + $result['error'] = sanitize_text_field($status['error']); | |
| 1722 | + } | |
| 1723 | + | |
| 1724 | + return $result; | |
| 1725 | +} | |
| 1726 | + | |
| 1727 | + | |
| 1728 | +public function mxchat_handle_sitemap_submission() { | |
| 1729 | + // Check if the form was submitted and verify permissions | |
| 1730 | + if (!isset($_POST['submit_sitemap']) || !current_user_can('manage_options')) { | |
| 1731 | + wp_die(esc_html__('Unauthorized access', 'mxchat')); | |
| 1732 | + } | |
| 1733 | + | |
| 1734 | + // Verify nonce | |
| 1735 | + check_admin_referer('mxchat_submit_sitemap_action', 'mxchat_submit_sitemap_nonce'); | |
| 1736 | + | |
| 1737 | + // Validate URL | |
| 1738 | + if (!isset($_POST['sitemap_url']) || empty($_POST['sitemap_url'])) { | |
| 1739 | + set_transient('mxchat_admin_notice_error', | |
| 1740 | + esc_html__('Please provide a valid URL.', 'mxchat'), | |
| 1741 | + 30 | |
| 1742 | + ); | |
| 1743 | + wp_safe_redirect(esc_url(admin_url('admin.php?page=mxchat-prompts'))); | |
| 1744 | + exit; | |
| 1745 | + } | |
| 1746 | + | |
| 1747 | + $submitted_url = esc_url_raw($_POST['sitemap_url']); | |
| 1748 | + | |
| 1749 | + // Convert Google Drive sharing URLs to direct download URLs | |
| 1750 | + if ( strpos($submitted_url, 'drive.google.com') !== false ) { | |
| 1751 | + $file_id = ''; | |
| 1752 | + if ( preg_match('/[?&]id=([a-zA-Z0-9_-]+)/', $submitted_url, $m) ) { | |
| 1753 | + $file_id = $m[1]; | |
| 1754 | + } elseif ( preg_match('#/file/d/([a-zA-Z0-9_-]+)#', $submitted_url, $m) ) { | |
| 1755 | + $file_id = $m[1]; | |
| 1756 | + } | |
| 1757 | + if ( ! empty($file_id) ) { | |
| 1758 | + $submitted_url = 'https://drive.google.com/uc?export=download&id=' . $file_id; | |
| 1759 | + } | |
| 1760 | + } | |
| 1761 | + | |
| 1762 | + // Get bot_id from form submission | |
| 1763 | + $bot_id = isset($_POST['bot_id']) ? sanitize_key($_POST['bot_id']) : 'default'; | |
| 1764 | + | |
| 1765 | + // Get bot-specific options and validate the embedding decision — | |
| 1766 | + // custom-provider-aware (plan cbd5fd). | |
| 1767 | + $bot_options = $this->get_bot_options($bot_id); | |
| 1768 | + $options = !empty($bot_options) ? $bot_options : get_option('mxchat_options'); | |
| 1769 | + | |
| 1770 | + $preflight = MxChat_Utils::embedding_preflight($options); | |
| 1771 | + if (!$preflight['ok']) { | |
| 1772 | + set_transient('mxchat_admin_notice_error', esc_html($preflight['reason']), 30); | |
| 1773 | + wp_safe_redirect(esc_url(admin_url('admin.php?page=mxchat-prompts'))); | |
| 1774 | + exit; | |
| 1775 | + } | |
| 1776 | + $api_key = $preflight['api_key']; | |
| 1777 | + | |
| 1778 | + // Fetch URL — send an honest, versioned MXChat crawler UA (not a spoofed | |
| 1779 | + // browser). Stale browser UAs are exactly what WAFs like SiteGround's | |
| 1780 | + // ModSecurity flag as scrapers, 403-ing the fetch (including PDFs served | |
| 1781 | + // from the site's own media library, which route through this same call). | |
| 1782 | + // See mxchat_ingest_user_agent(). Accept is kept for content negotiation; | |
| 1783 | + // the browser-only Accept-Language fingerprint is dropped so it stays | |
| 1784 | + // coherent with a bot identity. | |
| 1785 | + $response = wp_remote_get($submitted_url, array( | |
| 1786 | + 'timeout' => 30, | |
| 1787 | + 'sslverify' => false, | |
| 1788 | + 'user-agent' => mxchat_ingest_user_agent(), | |
| 1789 | + 'headers' => array( | |
| 1790 | + 'Accept' => 'text/html,application/xhtml+xml,application/xml;q=0.9,*/*;q=0.8', | |
| 1791 | + ), | |
| 1792 | + )); | |
| 1793 | + | |
| 1794 | + if (is_wp_error($response) || wp_remote_retrieve_response_code($response) !== 200) { | |
| 1795 | + $error_message = is_wp_error($response) ? $response->get_error_message() : 'HTTP Status: ' . wp_remote_retrieve_response_code($response); | |
| 1796 | + set_transient('mxchat_admin_notice_error', | |
| 1797 | + sprintf( | |
| 1798 | + esc_html__('Failed to fetch the URL: %s', 'mxchat'), | |
| 1799 | + esc_html($error_message) | |
| 1800 | + ), | |
| 1801 | + 30 | |
| 1802 | + ); | |
| 1803 | + wp_safe_redirect(esc_url(admin_url('admin.php?page=mxchat-prompts'))); | |
| 1804 | + exit; | |
| 1805 | + } | |
| 1806 | + | |
| 1807 | + $content_type = wp_remote_retrieve_header($response, 'content-type'); | |
| 1808 | + $body_content = wp_remote_retrieve_body($response); | |
| 1809 | + | |
| 1810 | + if (empty($body_content)) { | |
| 1811 | + set_transient('mxchat_admin_notice_error', | |
| 1812 | + esc_html__('Empty response received from URL.', 'mxchat'), | |
| 1813 | + 30 | |
| 1814 | + ); | |
| 1815 | + wp_safe_redirect(esc_url(admin_url('admin.php?page=mxchat-prompts'))); | |
| 1816 | + exit; | |
| 1817 | + } | |
| 1818 | + | |
| 1819 | + // Handle PDF URL | |
| 1820 | + if ($this->mxchat_is_pdf_url($submitted_url, $response)) { | |
| 1821 | + $result = $this->mxchat_handle_pdf_for_knowledge_base($submitted_url, $response, $bot_id); | |
| 1822 | + | |
| 1823 | + if ($result === 'queued') { | |
| 1824 | + set_transient('mxchat_admin_notice_success', | |
| 1825 | + esc_html__('PDF queued for processing. Processing will start automatically.', 'mxchat'), | |
| 1826 | + 30 | |
| 1827 | + ); | |
| 1828 | + } else { | |
| 1829 | + set_transient('mxchat_admin_notice_error', | |
| 1830 | + esc_html__('Failed to queue PDF processing: ', 'mxchat') . esc_html($result), | |
| 1831 | + 30 | |
| 1832 | + ); | |
| 1833 | + } | |
| 1834 | + | |
| 1835 | + wp_safe_redirect(esc_url(admin_url('admin.php?page=mxchat-prompts'))); | |
| 1836 | + exit; | |
| 1837 | + } | |
| 1838 | + | |
| 1839 | + // Handle Sitemap XML | |
| 1840 | + if (strpos($content_type, 'xml') !== false || strpos($body_content, '<urlset') !== false) { | |
| 1841 | + libxml_use_internal_errors(true); | |
| 1842 | + $xml = simplexml_load_string($body_content); | |
| 1843 | + $xml_errors = libxml_get_errors(); | |
| 1844 | + libxml_clear_errors(); | |
| 1845 | + | |
| 1846 | + if ($xml === false || !empty($xml_errors)) { | |
| 1847 | + set_transient('mxchat_admin_notice_error', | |
| 1848 | + esc_html__('Invalid sitemap XML. Please provide a valid sitemap.', 'mxchat'), | |
| 1849 | + 30 | |
| 1850 | + ); | |
| 1851 | + wp_safe_redirect(esc_url(admin_url('admin.php?page=mxchat-prompts'))); | |
| 1852 | + exit; | |
| 1853 | + } | |
| 1854 | + | |
| 1855 | + $result = $this->mxchat_handle_sitemap_for_knowledge_base($xml, $submitted_url, $bot_id); | |
| 1856 | + | |
| 1857 | + if ($result === 'queued') { | |
| 1858 | + set_transient('mxchat_admin_notice_success', | |
| 1859 | + esc_html__('Sitemap queued for processing. Processing will start automatically.', 'mxchat'), | |
| 1860 | + 30 | |
| 1861 | + ); | |
| 1862 | + } else { | |
| 1863 | + // Surface the reason the handler already computed (embedding pre-flight, | |
| 1864 | + // empty sitemap, queue failure). The old message pointed at the status | |
| 1865 | + // area, which is empty on this path — nothing was ever queued. | |
| 1866 | + if (is_string($result) && $result !== '') { | |
| 1867 | + set_transient('mxchat_admin_notice_error', | |
| 1868 | + esc_html__('Failed to queue sitemap processing: ', 'mxchat') . esc_html($result), | |
| 1869 | + 30 | |
| 1870 | + ); | |
| 1871 | + } else { | |
| 1872 | + set_transient('mxchat_admin_notice_error', | |
| 1873 | + esc_html__('Failed to queue sitemap processing.', 'mxchat'), | |
| 1874 | + 30 | |
| 1875 | + ); | |
| 1876 | + } | |
| 1877 | + } | |
| 1878 | + | |
| 1879 | + wp_safe_redirect(esc_url(admin_url('admin.php?page=mxchat-prompts'))); | |
| 1880 | + exit; | |
| 1881 | + } | |
| 1882 | + | |
| 1883 | + // Handle Regular URL (single page) | |
| 1884 | + $page_content = $this->mxchat_extract_main_content($body_content); | |
| 1885 | + $sanitized_content = $this->mxchat_sanitize_content_for_api($page_content); | |
| 1886 | + | |
| 1887 | + //error_log('[MXCHAT-URL-DEBUG] URL: ' . $submitted_url); | |
| 1888 | + //error_log('[MXCHAT-URL-DEBUG] Extracted page_content length: ' . strlen($page_content)); | |
| 1889 | + //error_log('[MXCHAT-URL-DEBUG] Sanitized content length: ' . strlen($sanitized_content)); | |
| 1890 | + //error_log('[MXCHAT-URL-DEBUG] Content preview: ' . substr($sanitized_content, 0, 500)); | |
| 1891 | + | |
| 1892 | + if (empty($sanitized_content)) { | |
| 1893 | + set_transient('mxchat_admin_notice_error', | |
| 1894 | + esc_html__('No valid content found on the provided URL.', 'mxchat'), | |
| 1895 | + 30 | |
| 1896 | + ); | |
| 1897 | + wp_safe_redirect(esc_url(admin_url('admin.php?page=mxchat-prompts'))); | |
| 1898 | + exit; | |
| 1899 | + } | |
| 1900 | + | |
| 1901 | + // For single URLs, process immediately using submit_content_to_db | |
| 1902 | + // This handles chunking automatically for large content | |
| 1903 | + $db_result = MxChat_Utils::submit_content_to_db( | |
| 1904 | + $sanitized_content, | |
| 1905 | + $submitted_url, | |
| 1906 | + $api_key, | |
| 1907 | + null, | |
| 1908 | + $bot_id, | |
| 1909 | + 'url', // content_type | |
| 1910 | + MxChat_Utils::extract_page_title($body_content) // the page's own title (plan d9ee66) | |
| 1911 | + ); | |
| 1912 | + | |
| 1913 | + if (is_wp_error($db_result)) { | |
| 1914 | + $error_message = esc_html__('Failed to store content in database: ', 'mxchat') . esc_html($db_result->get_error_message()); | |
| 1915 | + set_transient('mxchat_admin_notice_error', $error_message, 30); | |
| 1916 | + } else { | |
| 1917 | + $success_message = esc_html__('URL content successfully submitted!', 'mxchat'); | |
| 1918 | + set_transient('mxchat_admin_notice_success', $success_message, 30); | |
| 1919 | + } | |
| 1920 | + | |
| 1921 | + wp_safe_redirect(esc_url(admin_url('admin.php?page=mxchat-prompts'))); | |
| 1922 | + exit; | |
| 1923 | +} | |
| 1924 | + | |
| 1925 | + | |
| 1926 | +public function mxchat_get_single_url_status() { | |
| 1927 | + $status = get_transient('mxchat_single_url_status'); | |
| 1928 | + if (!$status) { | |
| 1929 | + return null; | |
| 1930 | + } | |
| 1931 | + | |
| 1932 | + // Add human-readable time | |
| 1933 | + if (isset($status['timestamp'])) { | |
| 1934 | + $status['human_time'] = human_time_diff(strtotime($status['timestamp']), current_time('timestamp')) . ' ' . __('ago', 'mxchat'); | |
| 1935 | + } | |
| 1936 | + | |
| 1937 | + return $status; | |
| 1938 | +} | |
| 1939 | + | |
| 1940 | +public function mxchat_handle_sitemap_for_knowledge_base($xml, $sitemap_url, $bot_id = 'default') { | |
| 1941 | + if (!current_user_can('manage_options')) { | |
| 1942 | + return false; | |
| 1943 | + } | |
| 1944 | + | |
| 1945 | + try { | |
| 1946 | + $sitemap_url = esc_url_raw($sitemap_url); | |
| 1947 | + | |
| 1948 | + if (!$xml || !is_object($xml)) { | |
| 1949 | + throw new Exception(__('Invalid XML object provided', 'mxchat')); | |
| 1950 | + } | |
| 1951 | + | |
| 1952 | + // Get bot-specific embedding API for validation | |
| 1953 | + $bot_options = $this->get_bot_options($bot_id); | |
| 1954 | + $options = !empty($bot_options) ? $bot_options : get_option('mxchat_options'); | |
| 1955 | + | |
| 1956 | + // Test the embedding API before processing | |
| 1957 | + $test_phrase = "Test embedding generation for MxChat"; | |
| 1958 | + $test_result = $this->mxchat_generate_embedding($test_phrase, $bot_id); | |
| 1959 | + | |
| 1960 | + if (is_string($test_result)) { | |
| 1961 | + throw new Exception(__('Embedding API validation failed: ', 'mxchat') . $test_result); | |
| 1962 | + } | |
| 1963 | + | |
| 1964 | + if (!is_array($test_result)) { | |
| 1965 | + throw new Exception(__('Embedding API returned unexpected result type. Please check your configuration.', 'mxchat')); | |
| 1966 | + } | |
| 1967 | + | |
| 1968 | + // Extract URLs from sitemap | |
| 1969 | + $urls = array(); | |
| 1970 | + foreach ($xml->url as $url_element) { | |
| 1971 | + $url = esc_url_raw((string)$url_element->loc); | |
| 1972 | + if ($url) { | |
| 1973 | + $urls[] = array('url' => $url); | |
| 1974 | + } | |
| 1975 | + } | |
| 1976 | + | |
| 1977 | + $total_urls = count($urls); | |
| 1978 | + | |
| 1979 | + if ($total_urls < 1) { | |
| 1980 | + throw new Exception(__('No valid URLs found in sitemap', 'mxchat')); | |
| 1981 | + } | |
| 1982 | + | |
| 1983 | + // Create unique queue ID | |
| 1984 | + $queue_id = 'sitemap_' . md5($sitemap_url . time()); | |
| 1985 | + | |
| 1986 | + // Add URLs to queue | |
| 1987 | + $queued_count = $this->mxchat_add_to_queue($queue_id, 'url', $urls, $bot_id); | |
| 1988 | + | |
| 1989 | + if ($queued_count === 0) { | |
| 1990 | + throw new Exception(__('Failed to add URLs to processing queue', 'mxchat')); | |
| 1991 | + } | |
| 1992 | + | |
| 1993 | + // Store queue metadata | |
| 1994 | + $this->mxchat_set_queue_meta($queue_id, 'source_url', $sitemap_url); | |
| 1995 | + $this->mxchat_set_queue_meta($queue_id, 'queue_type', 'sitemap'); | |
| 1996 | + $this->mxchat_set_queue_meta($queue_id, 'total_items', $total_urls); | |
| 1997 | + $this->mxchat_set_queue_meta($queue_id, 'bot_id', $bot_id); | |
| 1998 | + $this->mxchat_set_queue_meta($queue_id, 'created_at', current_time('mysql')); | |
| 1999 | + | |
| 2000 | + // Store queue ID in transient for status tracking | |
| 2001 | + set_transient('mxchat_active_queue_sitemap', $queue_id, DAY_IN_SECONDS); | |
| 2002 | + set_transient('mxchat_last_sitemap_url', $sitemap_url, DAY_IN_SECONDS); | |
| 2003 | + | |
| 2004 | + return 'queued'; | |
| 2005 | + | |
| 2006 | + } catch (Exception $e) { | |
| 2007 | + $error_message = $e->getMessage(); | |
| 2008 | + //error_log(sprintf(esc_html__('Error preparing sitemap for processing: %s', 'mxchat'), esc_html($error_message))); | |
| 2009 | + | |
| 2010 | + return $error_message; | |
| 2011 | + } | |
| 2012 | + | |
| 2013 | +} | |
| 2014 | + | |
| 2015 | +/** | |
| 2016 | + * Remove shortcode tags but preserve the content inside them | |
| 2017 | + * Example: [vc_column]Hello World[/vc_column] becomes "Hello World" | |
| 2018 | + * | |
| 2019 | + * @param string $content The content containing shortcodes | |
| 2020 | + * @return string Content with shortcode tags removed but inner content preserved | |
| 2021 | + */ | |
| 2022 | +/** | |
| 2023 | + * Single-pass HTML entity decode for text entering the knowledge base. | |
| 2024 | + * The corpus should hold what a human reads: a stored `&` consumes | |
| 2025 | + * extra tokens, distorts the vector away from the form a visitor's | |
| 2026 | + * question uses, and can be quoted back verbatim in an answer. | |
| 2027 | + * Deliberately NOT looped to a fixed point — a stored `&amp;` is a | |
| 2028 | + * legitimate literal `&` and must not collapse further (data loss). | |
| 2029 | + * UTF-8 charset keeps multibyte (CJK/RTL) text untouched. Both assembly | |
| 2030 | + * paths call this at their output points so the treatment cannot drift. | |
| 2031 | + * (Plan d2c92e.) | |
| 2032 | + */ | |
| 2033 | +private function mxchat_decode_entities_for_indexing($text) { | |
| 2034 | + return html_entity_decode((string) $text, ENT_QUOTES | ENT_HTML5, 'UTF-8'); | |
| 2035 | +} | |
| 2036 | + | |
| 2037 | +/** | |
| 2038 | + * Price lines for a product's indexed text, pinned to the store's BASE currency. | |
| 2039 | + * | |
| 2040 | + * The four product assembly paths each used to call get_woocommerce_currency_symbol() | |
| 2041 | + * with no argument, which resolves the currency active on the CURRENT request. | |
| 2042 | + * Multi-currency plugins (CURCY, WOOCS, Aelia, WPML Multicurrency) filter that per | |
| 2043 | + * request, so whichever currency the store happened to be serving when an import ran | |
| 2044 | + * was frozen into every product it indexed. The amounts have the mirror problem: the | |
| 2045 | + * woocommerce_product_get_* filters convert prices in the 'view' context but not in | |
| 2046 | + * 'edit', so a converted amount could be paired with an unconverted symbol and produce | |
| 2047 | + * a price that is not merely wrong but incoherent. | |
| 2048 | + * | |
| 2049 | + * Base currency option + 'edit' context makes both halves agree and makes the output | |
| 2050 | + * independent of when the import ran. The currency CODE is emitted alongside the symbol | |
| 2051 | + * so the model cannot read a bare "$" as USD on a store whose plugin swapped symbols. | |
| 2052 | + * (Plan 7403ec.) | |
| 2053 | + */ | |
| 2054 | +private function mxchat_product_price_lines($product) { | |
| 2055 | + if (!is_object($product) || !method_exists($product, 'get_regular_price')) { | |
| 2056 | + return ''; | |
| 2057 | + } | |
| 2058 | + | |
| 2059 | + $currency = get_option('woocommerce_currency'); | |
| 2060 | + $currency = is_string($currency) ? trim($currency) : ''; | |
| 2061 | + $symbol = ($currency !== '') | |
| 2062 | + ? get_woocommerce_currency_symbol($currency) | |
| 2063 | + : get_woocommerce_currency_symbol(); | |
| 2064 | + $symbol = $this->mxchat_decode_entities_for_indexing($symbol); | |
| 2065 | + | |
| 2066 | + $regular_price = $product->get_regular_price('edit'); | |
| 2067 | + $sale_price = $product->get_sale_price('edit'); | |
| 2068 | + $price = $product->get_price('edit'); | |
| 2069 | + | |
| 2070 | + $lines = ''; | |
| 2071 | + | |
| 2072 | + if (!empty($regular_price)) { | |
| 2073 | + $lines .= "Price: " . $this->mxchat_format_indexed_price($regular_price, $currency, $symbol) . "\n"; | |
| 2074 | + } elseif (!empty($price)) { | |
| 2075 | + $lines .= "Price: " . $this->mxchat_format_indexed_price($price, $currency, $symbol) . "\n"; | |
| 2076 | + } | |
| 2077 | + | |
| 2078 | + if (!empty($sale_price) && $sale_price !== $regular_price) { | |
| 2079 | + $lines .= "Sale Price: " . $this->mxchat_format_indexed_price($sale_price, $currency, $symbol) . "\n"; | |
| 2080 | + } | |
| 2081 | + | |
| 2082 | + if ($product->is_type('variable')) { | |
| 2083 | + list($min_price, $max_price) = $this->mxchat_variation_price_range($product); | |
| 2084 | + if ($min_price !== null && $max_price !== null && (float) $min_price !== (float) $max_price) { | |
| 2085 | + $lines .= "Price Range: " . $this->mxchat_format_indexed_price($min_price, $currency, $symbol) | |
| 2086 | + . " - " . $this->mxchat_format_indexed_price($max_price, $currency, $symbol) . "\n"; | |
| 2087 | + } | |
| 2088 | + } | |
| 2089 | + | |
| 2090 | + return $lines; | |
| 2091 | +} | |
| 2092 | + | |
| 2093 | +/** | |
| 2094 | + * One indexed price amount, labelled with its currency code. | |
| 2095 | + * | |
| 2096 | + * "INR 1299.00 (Rs.1299.00)" — the code is what the model should reason from; the symbol | |
| 2097 | + * is kept so a quoted price still reads naturally. Falls back to the old symbol-only | |
| 2098 | + * shape when WooCommerce has no base currency configured, and drops the parenthetical | |
| 2099 | + * when the symbol is absent or IS the code (several currencies have no distinct glyph). | |
| 2100 | + */ | |
| 2101 | +private function mxchat_format_indexed_price($amount, $currency, $symbol) { | |
| 2102 | + $amount = (string) $amount; | |
| 2103 | + | |
| 2104 | + if ($currency === '') { | |
| 2105 | + return $symbol . $amount; | |
| 2106 | + } | |
| 2107 | + | |
| 2108 | + if ($symbol === '' || $symbol === $currency) { | |
| 2109 | + return $currency . ' ' . $amount; | |
| 2110 | + } | |
| 2111 | + | |
| 2112 | + return $currency . ' ' . $amount . ' (' . $symbol . $amount . ')'; | |
| 2113 | +} | |
| 2114 | + | |
| 2115 | +/** | |
| 2116 | + * Min/max variation price read from the variations themselves in 'edit' context. | |
| 2117 | + * | |
| 2118 | + * get_variation_price() reads WooCommerce's display price cache, which multi-currency | |
| 2119 | + * plugins populate with converted values — the same defect the rest of this helper | |
| 2120 | + * exists to remove. Returns raw stored strings (not floats) so the indexed text keeps | |
| 2121 | + * the store's own price formatting, and (null, null) when no variation carries a price. | |
| 2122 | + */ | |
| 2123 | +private function mxchat_variation_price_range($product) { | |
| 2124 | + $min_raw = null; | |
| 2125 | + $max_raw = null; | |
| 2126 | + $min_val = null; | |
| 2127 | + $max_val = null; | |
| 2128 | + | |
| 2129 | + $children = method_exists($product, 'get_children') ? $product->get_children() : array(); | |
| 2130 | + | |
| 2131 | + foreach ($children as $child_id) { | |
| 2132 | + $variation = wc_get_product($child_id); | |
| 2133 | + if (!$variation) { | |
| 2134 | + continue; | |
| 2135 | + } | |
| 2136 | + $raw = $variation->get_price('edit'); | |
| 2137 | + if ($raw === '' || $raw === null) { | |
| 2138 | + continue; | |
| 2139 | + } | |
| 2140 | + $val = (float) $raw; | |
| 2141 | + if ($min_val === null || $val < $min_val) { | |
| 2142 | + $min_val = $val; | |
| 2143 | + $min_raw = $raw; | |
| 2144 | + } | |
| 2145 | + if ($max_val === null || $val > $max_val) { | |
| 2146 | + $max_val = $val; | |
| 2147 | + $max_raw = $raw; | |
| 2148 | + } | |
| 2149 | + } | |
| 2150 | + | |
| 2151 | + return array($min_raw, $max_raw); | |
| 2152 | +} | |
| 2153 | + | |
| 2154 | +private function strip_shortcode_tags_preserve_content($content) { | |
| 2155 | + // Single-pass regex removes all shortcode brackets: [tag], [tag attr="val"], [/tag], [tag /] | |
| 2156 | + // Content between tags is inherently preserved since only brackets are targeted | |
| 2157 | + $result = preg_replace('/\[\/?\w[\w-]*[^\]]*\]/', '', $content); | |
| 2158 | + return ($result !== null) ? $result : $content; | |
| 2159 | +} | |
| 2160 | + | |
| 2161 | +public function mxchat_sanitize_content_for_api($content) { | |
| 2162 | + //error_log('[MXCHAT-SANITIZE] Original content preview: ' . substr($content, 0, 500) . '...'); | |
| 2163 | + | |
| 2164 | + // Remove shortcode tags but PRESERVE content inside them | |
| 2165 | + $content = $this->strip_shortcode_tags_preserve_content($content); | |
| 2166 | + | |
| 2167 | + // Remove script, style tags, and HTML comments | |
| 2168 | + $content = preg_replace('/<script\b[^>]*>(.*?)<\/script>/is', '', $content); | |
| 2169 | + $content = preg_replace('/<style\b[^>]*>(.*?)<\/style>/is', '', $content); | |
| 2170 | + $content = preg_replace('/<!--(.|\s)*?-->/', '', $content); | |
| 2171 | + | |
| 2172 | + // Remove all HTML tags and decode HTML entities | |
| 2173 | + $content = wp_strip_all_tags($content); | |
| 2174 | + $content = html_entity_decode($content, ENT_QUOTES | ENT_HTML5); | |
| 2175 | + | |
| 2176 | + // Normalize whitespace but preserve paragraph breaks | |
| 2177 | + // First, normalize line endings to \n | |
| 2178 | + $content = str_replace(["\r\n", "\r"], "\n", $content); | |
| 2179 | + // Replace multiple spaces/tabs with single space, but preserve newlines | |
| 2180 | + $content = preg_replace('/[ \t]+/', ' ', $content); | |
| 2181 | + // Replace 3+ newlines with 2 newlines (max 2 blank lines) | |
| 2182 | + $content = preg_replace('/\n{3,}/', "\n\n", $content); | |
| 2183 | + // Trim each line | |
| 2184 | + $lines = explode("\n", $content); | |
| 2185 | + $lines = array_map('trim', $lines); | |
| 2186 | + $content = implode("\n", $lines); | |
| 2187 | + // Final trim | |
| 2188 | + $content = trim($content); | |
| 2189 | + | |
| 2190 | + // Remove control characters (which can cause database issues) but preserve \n (0x0A) and \r (0x0D) | |
| 2191 | + $content = preg_replace('/[\x00-\x08\x0B\x0C\x0E-\x1F\x7F]/u', '', $content); | |
| 2192 | + | |
| 2193 | + // Remove NULL bytes which can cause database errors | |
| 2194 | + $content = str_replace("\0", "", $content); | |
| 2195 | + | |
| 2196 | + // Ensure valid UTF-8 encoding | |
| 2197 | + $content = wp_check_invalid_utf8($content); | |
| 2198 | + | |
| 2199 | + // Remove extremely long runs with no whitespace (base64 blobs, minified JS). | |
| 2200 | + // Counts CHARACTERS (/u), and never strips a run containing characters from a | |
| 2201 | + // script written without spaces — Japanese, Chinese, Thai, Khmer, Lao, Myanmar — | |
| 2202 | + // where a normal paragraph is legitimately one unbroken run. | |
| 2203 | + $content = preg_replace_callback('/\S{300,}/u', function ($m) { | |
| 2204 | + return preg_match('/[\p{Han}\p{Hiragana}\p{Katakana}\p{Thai}\p{Khmer}\p{Lao}\p{Myanmar}]/u', $m[0]) ? $m[0] : ' '; | |
| 2205 | + }, $content); | |
| 2206 | + | |
| 2207 | + // Remove emoji/symbol blocks only — not the whole supplementary plane, which | |
| 2208 | + // also holds CJK Extension B ideographs used in real Chinese/Japanese names. | |
| 2209 | + // A ZWJ (U+200D) BETWEEN stripped pictographs is consumed with them, so a | |
| 2210 | + // family sequence like 👨👩👧 leaves no invisible zero-width residue behind | |
| 2211 | + // (the joiner between NON-emoji characters — Hindi conjuncts — is untouched). | |
| 2212 | + $content = preg_replace('/[\x{1F000}-\x{1F0FF}\x{1F300}-\x{1FAFF}](?:\x{200D}[\x{1F000}-\x{1F0FF}\x{1F300}-\x{1FAFF}])*/u', '', $content); | |
| 2213 | + | |
| 2214 | + // There is deliberately NO catch-all character allowlist here (plan 209e57; | |
| 2215 | + // one existed until 3.2.20). Every genuinely dangerous byte is already gone: | |
| 2216 | + // control characters, null bytes, invalid UTF-8 and the emoji blocks are all | |
| 2217 | + // stripped above. The allowlist's only remaining effect was to damage scripts | |
| 2218 | + // nobody thought to enumerate — Unicode Cf (Format) was missing, so it | |
| 2219 | + // replaced the zero-width joiner/non-joiner with spaces and silently split | |
| 2220 | + // Persian words (میروم → می روم) and broke Hindi conjuncts (क्ष → क् ष). | |
| 2221 | + // Do not add one back; the failure mode of an allowlist is exactly this. | |
| 2222 | + | |
| 2223 | + // Limit to reasonable length if needed (byte limit — MySQL TEXT is byte-sized, | |
| 2224 | + // but cut on a character boundary so a multibyte char is never split mid-sequence) | |
| 2225 | + $max_length = 65000; // Just under MySQL TEXT field limit | |
| 2226 | + if (strlen($content) > $max_length) { | |
| 2227 | + $content = mb_strcut($content, 0, $max_length, 'UTF-8'); | |
| 2228 | + } | |
| 2229 | + | |
| 2230 | + //error_log('[MXCHAT-SANITIZE] Sanitized content preview: ' . substr($content, 0, 500) . '...'); | |
| 2231 | + return $content; | |
| 2232 | +} | |
| 2233 | +public function mxchat_extract_main_content($html) { | |
| 2234 | + if (empty($html)) { | |
| 2235 | + return ''; | |
| 2236 | + } | |
| 2237 | + try { | |
| 2238 | + $dom = new DOMDocument; | |
| 2239 | + libxml_use_internal_errors(true); // Suppress HTML parsing errors | |
| 2240 | + @$dom->loadHTML(mb_convert_encoding($html, 'HTML-ENTITIES', 'UTF-8'), LIBXML_HTML_NOIMPLIED | LIBXML_HTML_NODEFDTD); | |
| 2241 | + $xpath = new DOMXPath($dom); | |
| 2242 | + | |
| 2243 | + // For debugging purposes | |
| 2244 | + $debugEnabled = true; // Set to true to enable debugging output | |
| 2245 | + $debug = function($message) use ($debugEnabled) { | |
| 2246 | + if ($debugEnabled) { | |
| 2247 | + //error_log('[MXCHAT-EXTRACT-DEBUG] ' . $message); | |
| 2248 | + } | |
| 2249 | + }; | |
| 2250 | + | |
| 2251 | + // Direct targeting for Gerow theme posts | |
| 2252 | + $post_text = $xpath->query('//div[contains(@class, "post-text")]'); | |
| 2253 | + if ($post_text && $post_text->length > 0) { | |
| 2254 | + $debug("Found post-text directly"); | |
| 2255 | + $content = ''; | |
| 2256 | + foreach ($post_text as $node) { | |
| 2257 | + $content .= $dom->saveHTML($node); | |
| 2258 | + } | |
| 2259 | + if (!empty($content)) { | |
| 2260 | + $debug("Returning post-text content"); | |
| 2261 | + return $content; | |
| 2262 | + } | |
| 2263 | + } | |
| 2264 | + | |
| 2265 | + // Try to get the blog details content which contains the post-text | |
| 2266 | + $blog_details = $xpath->query('//div[contains(@class, "blog-details-content")]'); | |
| 2267 | + if ($blog_details && $blog_details->length > 0) { | |
| 2268 | + $debug("Found blog-details-content"); | |
| 2269 | + $content = ''; | |
| 2270 | + foreach ($blog_details as $node) { | |
| 2271 | + $content .= $dom->saveHTML($node); | |
| 2272 | + } | |
| 2273 | + if (!empty($content)) { | |
| 2274 | + $debug("Returning blog-details-content"); | |
| 2275 | + return $content; | |
| 2276 | + } | |
| 2277 | + } | |
| 2278 | + | |
| 2279 | + // Try to get the article which contains the blog details | |
| 2280 | + $article = $xpath->query('//article[contains(@class, "blog-details-wrap")]'); | |
| 2281 | + if ($article && $article->length > 0) { | |
| 2282 | + $debug("Found article with blog-details-wrap"); | |
| 2283 | + $content = ''; | |
| 2284 | + foreach ($article as $node) { | |
| 2285 | + $content .= $dom->saveHTML($node); | |
| 2286 | + } | |
| 2287 | + if (!empty($content)) { | |
| 2288 | + $debug("Returning article content"); | |
| 2289 | + return $content; | |
| 2290 | + } | |
| 2291 | + } | |
| 2292 | + | |
| 2293 | + // Try even broader with the blog-item-wrap | |
| 2294 | + $blog_item = $xpath->query('//div[contains(@class, "blog-item-wrap")]'); | |
| 2295 | + if ($blog_item && $blog_item->length > 0) { | |
| 2296 | + $debug("Found blog-item-wrap"); | |
| 2297 | + $content = ''; | |
| 2298 | + foreach ($blog_item as $node) { | |
| 2299 | + $content .= $dom->saveHTML($node); | |
| 2300 | + } | |
| 2301 | + if (!empty($content)) { | |
| 2302 | + $debug("Returning blog-item-wrap content"); | |
| 2303 | + return $content; | |
| 2304 | + } | |
| 2305 | + } | |
| 2306 | + | |
| 2307 | + // Specific Gerow theme path | |
| 2308 | + $gerow_path = $xpath->query('//section[contains(@class, "blog-area")]//div[contains(@class, "post-text")]'); | |
| 2309 | + if ($gerow_path && $gerow_path->length > 0) { | |
| 2310 | + $debug("Found Gerow theme path to post-text"); | |
| 2311 | + $content = ''; | |
| 2312 | + foreach ($gerow_path as $node) { | |
| 2313 | + $content .= $dom->saveHTML($node); | |
| 2314 | + } | |
| 2315 | + if (!empty($content)) { | |
| 2316 | + $debug("Returning Gerow post-text content"); | |
| 2317 | + return $content; | |
| 2318 | + } | |
| 2319 | + } | |
| 2320 | + | |
| 2321 | + // Generic blog post selectors | |
| 2322 | + $selectors = [ | |
| 2323 | + // Blog post specific selectors | |
| 2324 | + '//div[contains(@class, "post-text")]', | |
| 2325 | + '//article[contains(@class, "blog-post-item")]//div[contains(@class, "post-text")]', | |
| 2326 | + '//div[contains(@class, "blog-details-content")]', | |
| 2327 | + '//article[contains(@class, "blog-details-wrap")]', | |
| 2328 | + '//div[contains(@class, "entry-content")]', | |
| 2329 | + '//div[contains(@class, "blog-content")]', | |
| 2330 | + '//div[contains(@class, "blog-item-wrap")]', | |
| 2331 | + | |
| 2332 | + // More general content selectors | |
| 2333 | + '//div[contains(@class, "page__content")]', | |
| 2334 | + '//div[contains(@class, "elementor-widget-container")]', | |
| 2335 | + '//div[contains(@class, "elementor-text-editor")]', | |
| 2336 | + '//div[contains(@class, "elementor-widget-text-editor")]', | |
| 2337 | + '//*[contains(@class, "entry-content")]', | |
| 2338 | + '//*[contains(@class, "post-content")]', | |
| 2339 | + '//*[contains(@class, "article-content")]', | |
| 2340 | + '//*[@id="content"]', | |
| 2341 | + '//*[@id="main-content"]', | |
| 2342 | + '//section[contains(@class, "blog-area")]', | |
| 2343 | + '//article', | |
| 2344 | + '//main', | |
| 2345 | + '//div[contains(@class, "content")]' | |
| 2346 | + ]; | |
| 2347 | + | |
| 2348 | + // First handle Elementor content - get only leaf widget containers to avoid duplicates | |
| 2349 | + $debug("Checking for Elementor content"); | |
| 2350 | + // Get widget containers that are direct children of widgets (not nested inside other widget containers) | |
| 2351 | + $elementor_widgets = $xpath->query('//div[contains(@class, "elementor-widget")]//div[contains(@class, "elementor-widget-container")]'); | |
| 2352 | + if ($elementor_widgets && $elementor_widgets->length > 0) { | |
| 2353 | + $debug("Found Elementor widgets"); | |
| 2354 | + $seen_content = array(); // Track seen content to avoid duplicates | |
| 2355 | + $combined_content = ''; | |
| 2356 | + foreach ($elementor_widgets as $widget) { | |
| 2357 | + $widget_content = $dom->saveHTML($widget); | |
| 2358 | + if (!empty($widget_content)) { | |
| 2359 | + // Create a hash of the content to detect duplicates | |
| 2360 | + $content_hash = md5($widget_content); | |
| 2361 | + if (!isset($seen_content[$content_hash])) { | |
| 2362 | + $seen_content[$content_hash] = true; | |
| 2363 | + $combined_content .= $widget_content; | |
| 2364 | + } | |
| 2365 | + } | |
| 2366 | + } | |
| 2367 | + if (!empty($combined_content)) { | |
| 2368 | + $debug("Returning Elementor content"); | |
| 2369 | + return $combined_content; | |
| 2370 | + } | |
| 2371 | + } | |
| 2372 | + | |
| 2373 | + // Try standard selectors one by one | |
| 2374 | + foreach ($selectors as $selector) { | |
| 2375 | + $debug("Trying selector: " . $selector); | |
| 2376 | + $nodes = $xpath->query($selector); | |
| 2377 | + if ($nodes && $nodes->length > 0) { | |
| 2378 | + $debug("Found " . $nodes->length . " matches for selector: " . $selector); | |
| 2379 | + // Only take the FIRST matching node to avoid duplicate content | |
| 2380 | + // (pages often have nested or multiple containers with same class) | |
| 2381 | + $content = $dom->saveHTML($nodes->item(0)); | |
| 2382 | + if (!empty($content)) { | |
| 2383 | + $debug("Returning content from selector: " . $selector . " (first match only)"); | |
| 2384 | + return $content; | |
| 2385 | + } | |
| 2386 | + } | |
| 2387 | + } | |
| 2388 | + | |
| 2389 | + // Manual regex fallback for post-text if DOM methods fail | |
| 2390 | + $debug("Trying regex fallback"); | |
| 2391 | + if (preg_match('/<div class="post-text">(.*?)<\/div>\s*<\/div>\s*<\/div>/s', $html, $matches)) { | |
| 2392 | + $debug("Found post-text via regex"); | |
| 2393 | + return '<div class="post-text">' . $matches[1] . '</div>'; | |
| 2394 | + } | |
| 2395 | + | |
| 2396 | + // Try to extract the blog section as a whole | |
| 2397 | + $blog_section = $xpath->query('//section[contains(@class, "blog-area")]'); | |
| 2398 | + if ($blog_section && $blog_section->length > 0) { | |
| 2399 | + $debug("Found blog-area section"); | |
| 2400 | + $content = ''; | |
| 2401 | + foreach ($blog_section as $node) { | |
| 2402 | + $content .= $dom->saveHTML($node); | |
| 2403 | + } | |
| 2404 | + if (!empty($content)) { | |
| 2405 | + $debug("Returning blog-area section content"); | |
| 2406 | + return $content; | |
| 2407 | + } | |
| 2408 | + } | |
| 2409 | + | |
| 2410 | + // Generic container selectors for non-CMS sites (like .asp pages) | |
| 2411 | + $debug("Trying generic container selectors"); | |
| 2412 | + $generic_selectors = [ | |
| 2413 | + '//div[@id="main"]', | |
| 2414 | + '//div[@id="wrapper"]', | |
| 2415 | + '//div[@id="page"]', | |
| 2416 | + '//div[@id="site-content"]', | |
| 2417 | + '//div[contains(@class, "main-content")]', | |
| 2418 | + '//div[contains(@class, "page-content")]', | |
| 2419 | + '//div[contains(@class, "site-content")]', | |
| 2420 | + ]; | |
| 2421 | + | |
| 2422 | + foreach ($generic_selectors as $selector) { | |
| 2423 | + $debug("Trying generic selector: " . $selector); | |
| 2424 | + $nodes = $xpath->query($selector); | |
| 2425 | + if ($nodes && $nodes->length > 0) { | |
| 2426 | + $content = $dom->saveHTML($nodes->item(0)); | |
| 2427 | + if (!empty($content)) { | |
| 2428 | + $debug("Returning content from generic selector: " . $selector); | |
| 2429 | + return $content; | |
| 2430 | + } | |
| 2431 | + } | |
| 2432 | + } | |
| 2433 | + | |
| 2434 | + // Paragraph-based content detection - find regions with substantial text | |
| 2435 | + $debug("Trying paragraph-based content detection"); | |
| 2436 | + $paragraphs = $xpath->query('//p[string-length(normalize-space()) > 50]'); | |
| 2437 | + if ($paragraphs && $paragraphs->length >= 3) { | |
| 2438 | + $debug("Found " . $paragraphs->length . " substantial paragraphs"); | |
| 2439 | + // Collect all substantial paragraphs and their content | |
| 2440 | + $paragraph_content = ''; | |
| 2441 | + foreach ($paragraphs as $p) { | |
| 2442 | + $paragraph_content .= $dom->saveHTML($p) . "\n"; | |
| 2443 | + } | |
| 2444 | + if (!empty($paragraph_content)) { | |
| 2445 | + $debug("Returning paragraph-based content"); | |
| 2446 | + return $paragraph_content; | |
| 2447 | + } | |
| 2448 | + } | |
| 2449 | + | |
| 2450 | + // Improved body fallback - strip nav/header/footer elements first | |
| 2451 | + $debug("Using improved body fallback"); | |
| 2452 | + $body = $dom->getElementsByTagName('body'); | |
| 2453 | + if ($body->length > 0) { | |
| 2454 | + // Clone the body to avoid modifying the original DOM | |
| 2455 | + $body_clone = $body->item(0)->cloneNode(true); | |
| 2456 | + | |
| 2457 | + // Remove common non-content elements by tag name | |
| 2458 | + $remove_tags = ['nav', 'header', 'footer', 'aside', 'script', 'style', 'noscript']; | |
| 2459 | + foreach ($remove_tags as $tag) { | |
| 2460 | + $elements = $body_clone->getElementsByTagName($tag); | |
| 2461 | + // Iterate backwards to safely remove elements | |
| 2462 | + for ($i = $elements->length - 1; $i >= 0; $i--) { | |
| 2463 | + $el = $elements->item($i); | |
| 2464 | + if ($el && $el->parentNode) { | |
| 2465 | + $el->parentNode->removeChild($el); | |
| 2466 | + } | |
| 2467 | + } | |
| 2468 | + } | |
| 2469 | + | |
| 2470 | + // Remove elements with common non-content class names using XPath on the cloned body | |
| 2471 | + $temp_dom = new DOMDocument(); | |
| 2472 | + @$temp_dom->appendChild($temp_dom->importNode($body_clone, true)); | |
| 2473 | + $temp_xpath = new DOMXPath($temp_dom); | |
| 2474 | + | |
| 2475 | + $remove_class_patterns = [ | |
| 2476 | + '//*[contains(@class, "nav")]', | |
| 2477 | + '//*[contains(@class, "menu")]', | |
| 2478 | + '//*[contains(@class, "sidebar")]', | |
| 2479 | + '//*[contains(@class, "footer")]', | |
| 2480 | + '//*[contains(@class, "header")]', | |
| 2481 | + '//*[contains(@id, "nav")]', | |
| 2482 | + '//*[contains(@id, "menu")]', | |
| 2483 | + '//*[contains(@id, "sidebar")]', | |
| 2484 | + '//*[contains(@id, "footer")]', | |
| 2485 | + '//*[contains(@id, "header")]', | |
| 2486 | + ]; | |
| 2487 | + | |
| 2488 | + foreach ($remove_class_patterns as $pattern) { | |
| 2489 | + $elements = $temp_xpath->query($pattern); | |
| 2490 | + if ($elements) { | |
| 2491 | + for ($i = $elements->length - 1; $i >= 0; $i--) { | |
| 2492 | + $el = $elements->item($i); | |
| 2493 | + if ($el && $el->parentNode) { | |
| 2494 | + $el->parentNode->removeChild($el); | |
| 2495 | + } | |
| 2496 | + } | |
| 2497 | + } | |
| 2498 | + } | |
| 2499 | + | |
| 2500 | + $cleaned_content = $temp_dom->saveHTML(); | |
| 2501 | + if (!empty($cleaned_content)) { | |
| 2502 | + $debug("Returning cleaned body content"); | |
| 2503 | + return $cleaned_content; | |
| 2504 | + } | |
| 2505 | + } | |
| 2506 | + | |
| 2507 | + // Last resort: return the original HTML | |
| 2508 | + $debug("Returning original HTML"); | |
| 2509 | + return $html; | |
| 2510 | + } catch (Exception $e) { | |
| 2511 | + //error_log('[MXCHAT-ERROR] Content extraction failed: ' . $e->getMessage()); | |
| 2512 | + return $html; // Return original HTML if parsing fails | |
| 2513 | + } finally { | |
| 2514 | + libxml_clear_errors(); | |
| 2515 | + } | |
| 2516 | +} | |
| 2517 | +public function mxchat_get_sitemap_processing_status($sitemap_url) { | |
| 2518 | + $sitemap_url = esc_url_raw($sitemap_url); | |
| 2519 | + $status_key = sanitize_key('mxchat_sitemap_status_' . md5($sitemap_url)); | |
| 2520 | + $status = get_transient($status_key); | |
| 2521 | + | |
| 2522 | + if (!$status || !is_array($status)) { | |
| 2523 | + return false; | |
| 2524 | + } | |
| 2525 | + | |
| 2526 | + // Auto-complete check: if all URLs are processed but status isn't complete | |
| 2527 | + if (isset($status['processed_urls']) && isset($status['total_urls']) && | |
| 2528 | + $status['processed_urls'] >= $status['total_urls'] && | |
| 2529 | + isset($status['status']) && $status['status'] !== 'complete' && | |
| 2530 | + $status['status'] !== 'error') { | |
| 2531 | + | |
| 2532 | + // Mark as complete | |
| 2533 | + $status['status'] = 'complete'; | |
| 2534 | + $status['processed_urls'] = $status['total_urls']; // Ensure exact match | |
| 2535 | + | |
| 2536 | + // Update the transient with the corrected status | |
| 2537 | + set_transient($status_key, $status, DAY_IN_SECONDS); | |
| 2538 | + } | |
| 2539 | + | |
| 2540 | + return array( | |
| 2541 | + 'total_urls' => absint($status['total_urls']), | |
| 2542 | + 'processed_urls' => absint($status['processed_urls']), | |
| 2543 | + 'failed_urls' => absint($status['failed_urls'] ?? 0), | |
| 2544 | + 'percentage' => ($status['total_urls'] > 0) | |
| 2545 | + ? round((absint($status['processed_urls']) / absint($status['total_urls'])) * 100) | |
| 2546 | + : 0, | |
| 2547 | + 'status' => sanitize_text_field($status['status']), | |
| 2548 | + 'last_update' => human_time_diff(absint($status['last_update']), time()) . ' ' . esc_html__('ago', 'mxchat'), | |
| 2549 | + 'error' => isset($status['error']) ? sanitize_text_field($status['error']) : '', | |
| 2550 | + 'last_error' => isset($status['last_error']) ? sanitize_text_field($status['last_error']) : '', | |
| 2551 | + 'failed_urls_list' => isset($status['failed_urls_list']) ? $status['failed_urls_list'] : array() | |
| 2552 | + ); | |
| 2553 | +} | |
| 2554 | + | |
| 2555 | +public function mxchat_ajax_get_status_updates() { | |
| 2556 | + try { | |
| 2557 | + // Verify the request | |
| 2558 | + check_ajax_referer('mxchat_status_nonce', 'nonce'); | |
| 2559 | + | |
| 2560 | + // Get active queue IDs | |
| 2561 | + $sitemap_queue_id = get_transient('mxchat_active_queue_sitemap'); | |
| 2562 | + $pdf_queue_id = get_transient('mxchat_active_queue_pdf'); | |
| 2563 | + | |
| 2564 | + $sitemap_status = false; | |
| 2565 | + $pdf_status = false; | |
| 2566 | + | |
| 2567 | + // Get sitemap queue status | |
| 2568 | + if ($sitemap_queue_id) { | |
| 2569 | + $sitemap_status = $this->mxchat_get_queue_status_data($sitemap_queue_id, 'sitemap'); | |
| 2570 | + } | |
| 2571 | + | |
| 2572 | + // Get PDF queue status | |
| 2573 | + if ($pdf_queue_id) { | |
| 2574 | + $pdf_status = $this->mxchat_get_queue_status_data($pdf_queue_id, 'pdf'); | |
| 2575 | + } | |
| 2576 | + | |
| 2577 | + $is_active_processing = | |
| 2578 | + ($sitemap_status && $sitemap_status['status'] === 'processing') || | |
| 2579 | + ($pdf_status && $pdf_status['status'] === 'processing'); | |
| 2580 | + | |
| 2581 | + // Return JSON response with the status data | |
| 2582 | + wp_send_json(array( | |
| 2583 | + 'pdf_status' => $pdf_status, | |
| 2584 | + 'sitemap_status' => $sitemap_status, | |
| 2585 | + 'is_processing' => $is_active_processing, | |
| 2586 | + 'sitemap_queue_id' => $sitemap_queue_id, | |
| 2587 | + 'pdf_queue_id' => $pdf_queue_id | |
| 2588 | + )); | |
| 2589 | + | |
| 2590 | + } catch (Exception $e) { | |
| 2591 | + //error_log('MxChat Status Update Error: ' . $e->getMessage()); | |
| 2592 | + | |
| 2593 | + wp_send_json_error(array( | |
| 2594 | + 'message' => 'Error getting status updates: ' . $e->getMessage(), | |
| 2595 | + 'status' => 'error' | |
| 2596 | + )); | |
| 2597 | + } | |
| 2598 | +} | |
| 2599 | + | |
| 2600 | +/** | |
| 2601 | + * Helper function to get queue status data | |
| 2602 | + */ | |
| 2603 | +private function mxchat_get_queue_status_data($queue_id, $type = 'sitemap') { | |
| 2604 | + global $wpdb; | |
| 2605 | + $table_name = $wpdb->prefix . 'mxchat_processing_queue'; | |
| 2606 | + | |
| 2607 | + // Get counts by status | |
| 2608 | + $counts = $wpdb->get_results($wpdb->prepare( | |
| 2609 | + "SELECT status, COUNT(*) as count | |
| 2610 | + FROM $table_name | |
| 2611 | + WHERE queue_id = %s | |
| 2612 | + GROUP BY status", | |
| 2613 | + $queue_id | |
| 2614 | + ), OBJECT_K); | |
| 2615 | + | |
| 2616 | + $total = 0; | |
| 2617 | + $completed = 0; | |
| 2618 | + $failed = 0; | |
| 2619 | + $processing = 0; | |
| 2620 | + $pending = 0; | |
| 2621 | + | |
| 2622 | + foreach ($counts as $status => $data) { | |
| 2623 | + $count = absint($data->count); | |
| 2624 | + $total += $count; | |
| 2625 | + | |
| 2626 | + switch ($status) { | |
| 2627 | + case 'completed': | |
| 2628 | + $completed = $count; | |
| 2629 | + break; | |
| 2630 | + case 'failed': | |
| 2631 | + $failed = $count; | |
| 2632 | + break; | |
| 2633 | + case 'processing': | |
| 2634 | + $processing = $count; | |
| 2635 | + break; | |
| 2636 | + case 'pending': | |
| 2637 | + $pending = $count; | |
| 2638 | + break; | |
| 2639 | + } | |
| 2640 | + } | |
| 2641 | + | |
| 2642 | + if ($total === 0) { | |
| 2643 | + return false; | |
| 2644 | + } | |
| 2645 | + | |
| 2646 | + // Calculate percentage | |
| 2647 | + $percentage = round((($completed + $failed) / $total) * 100); | |
| 2648 | + | |
| 2649 | + // Get failed items details (limit to 50) | |
| 2650 | + $failed_items = array(); | |
| 2651 | + if ($failed > 0) { | |
| 2652 | + $failed_results = $wpdb->get_results($wpdb->prepare( | |
| 2653 | + "SELECT item_type, item_data, error_message, attempts, completed_at | |
| 2654 | + FROM $table_name | |
| 2655 | + WHERE queue_id = %s | |
| 2656 | + AND status = 'failed' | |
| 2657 | + AND attempts >= max_attempts | |
| 2658 | + ORDER BY id DESC | |
| 2659 | + LIMIT 50", | |
| 2660 | + $queue_id | |
| 2661 | + )); | |
| 2662 | + | |
| 2663 | + foreach ($failed_results as $item) { | |
| 2664 | + $data = json_decode($item->item_data, true); | |
| 2665 | + $url = $type === 'pdf' ? ($data['pdf_url'] ?? '') : ($data['url'] ?? ''); | |
| 2666 | + | |
| 2667 | + $failed_items[] = array( | |
| 2668 | + 'url' => $url, | |
| 2669 | + 'page' => $type === 'pdf' ? ($data['page_number'] ?? 0) : 0, | |
| 2670 | + 'error' => $item->error_message, | |
| 2671 | + 'retries' => $item->attempts, | |
| 2672 | + 'time' => strtotime($item->completed_at) | |
| 2673 | + ); | |
| 2674 | + } | |
| 2675 | + } | |
| 2676 | + | |
| 2677 | + // Get queue metadata | |
| 2678 | + $source_url = $this->mxchat_get_queue_meta($queue_id, 'source_url'); | |
| 2679 | + | |
| 2680 | + // Determine if queue is complete | |
| 2681 | + $is_complete = ($pending === 0 && $processing === 0); | |
| 2682 | + | |
| 2683 | + // Get last update time | |
| 2684 | + $last_update = $wpdb->get_var($wpdb->prepare( | |
| 2685 | + "SELECT MAX(UNIX_TIMESTAMP(COALESCE(completed_at, started_at, created_at))) | |
| 2686 | + FROM $table_name | |
| 2687 | + WHERE queue_id = %s", | |
| 2688 | + $queue_id | |
| 2689 | + )); | |
| 2690 | + | |
| 2691 | + $last_update_text = $last_update ? human_time_diff($last_update, time()) . ' ' . __('ago', 'mxchat') : __('Just now', 'mxchat'); | |
| 2692 | + | |
| 2693 | + // Format based on type | |
| 2694 | + if ($type === 'pdf') { | |
| 2695 | + return array( | |
| 2696 | + 'total_pages' => $total, | |
| 2697 | + 'processed_pages' => $completed + $failed, | |
| 2698 | + 'failed_pages' => $failed, | |
| 2699 | + 'percentage' => $percentage, | |
| 2700 | + 'status' => $is_complete ? 'complete' : 'processing', | |
| 2701 | + 'last_update' => $last_update_text, | |
| 2702 | + 'failed_pages_list' => $failed_items, | |
| 2703 | + 'pdf_url' => $source_url, | |
| 2704 | + 'queue_id' => $queue_id | |
| 2705 | + ); | |
| 2706 | + } else { | |
| 2707 | + return array( | |
| 2708 | + 'total_urls' => $total, | |
| 2709 | + 'processed_urls' => $completed + $failed, | |
| 2710 | + 'failed_urls' => $failed, | |
| 2711 | + 'percentage' => $percentage, | |
| 2712 | + 'status' => $is_complete ? 'complete' : 'processing', | |
| 2713 | + 'last_update' => $last_update_text, | |
| 2714 | + 'failed_urls_list' => $failed_items, | |
| 2715 | + 'sitemap_url' => $source_url, | |
| 2716 | + 'queue_id' => $queue_id | |
| 2717 | + ); | |
| 2718 | + } | |
| 2719 | +} | |
| 2720 | + | |
| 2721 | +/** | |
| 2722 | + * Public method to get processing status for both sitemap and PDF queues | |
| 2723 | + * Used by admin pages to display processing status | |
| 2724 | + * | |
| 2725 | + * @return array Array with 'sitemap_status', 'pdf_status', and 'is_processing' keys | |
| 2726 | + */ | |
| 2727 | +public function mxchat_get_processing_statuses() { | |
| 2728 | + $sitemap_queue_id = get_transient('mxchat_active_queue_sitemap'); | |
| 2729 | + $pdf_queue_id = get_transient('mxchat_active_queue_pdf'); | |
| 2730 | + | |
| 2731 | + $sitemap_status = false; | |
| 2732 | + $pdf_status = false; | |
| 2733 | + | |
| 2734 | + if ($sitemap_queue_id) { | |
| 2735 | + $sitemap_status = $this->mxchat_get_queue_status_data($sitemap_queue_id, 'sitemap'); | |
| 2736 | + } | |
| 2737 | + | |
| 2738 | + if ($pdf_queue_id) { | |
| 2739 | + $pdf_status = $this->mxchat_get_queue_status_data($pdf_queue_id, 'pdf'); | |
| 2740 | + } | |
| 2741 | + | |
| 2742 | + $is_processing = | |
| 2743 | + ($sitemap_status && $sitemap_status['status'] === 'processing') || | |
| 2744 | + ($pdf_status && $pdf_status['status'] === 'processing'); | |
| 2745 | + | |
| 2746 | + return array( | |
| 2747 | + 'sitemap_status' => $sitemap_status, | |
| 2748 | + 'pdf_status' => $pdf_status, | |
| 2749 | + 'is_processing' => $is_processing | |
| 2750 | + ); | |
| 2751 | +} | |
| 2752 | + | |
| 2753 | +/** | |
| 2754 | + * AJAX handler to get recent knowledge entries for real-time table updates | |
| 2755 | + * UPDATED: Now supports both WordPress DB and Pinecone data sources | |
| 2756 | + */ | |
| 2757 | +public function ajax_mxchat_get_recent_entries() { | |
| 2758 | + check_ajax_referer('mxchat_entries_nonce', 'nonce'); | |
| 2759 | + | |
| 2760 | + if (!current_user_can('manage_options')) { | |
| 2761 | + wp_send_json_error(array('message' => 'Unauthorized')); | |
| 2762 | + return; | |
| 2763 | + } | |
| 2764 | + | |
| 2765 | + global $wpdb; | |
| 2766 | + $table_name = $wpdb->prefix . 'mxchat_system_prompt_content'; | |
| 2767 | + | |
| 2768 | + // Get parameters | |
| 2769 | + $last_id = isset($_POST['last_id']) ? absint($_POST['last_id']) : 0; | |
| 2770 | + $limit = isset($_POST['limit']) ? min(absint($_POST['limit']), 50) : 10; | |
| 2771 | + $bot_id = isset($_POST['bot_id']) ? sanitize_text_field($_POST['bot_id']) : 'default'; | |
| 2772 | + | |
| 2773 | + // Check if Pinecone is enabled for this bot | |
| 2774 | + $pinecone_manager = $this->mxchat_get_pinecone_manager(); | |
| 2775 | + $pinecone_options = $pinecone_manager ? $pinecone_manager->mxchat_get_bot_pinecone_options($bot_id) : array(); | |
| 2776 | + $use_pinecone = ($pinecone_options['mxchat_use_pinecone'] ?? '0') === '1'; | |
| 2777 | + $has_pinecone_api = !empty($pinecone_options['mxchat_pinecone_api_key']); | |
| 2778 | + | |
| 2779 | + if ($use_pinecone && $has_pinecone_api) { | |
| 2780 | + // PINECONE DATA SOURCE - Get grouped entry count (not raw vector count) | |
| 2781 | + // Use mxchat_fetch_pinecone_records which returns total_unique_entries | |
| 2782 | + $records = $pinecone_manager->mxchat_fetch_pinecone_records($pinecone_options, '', 1, 10, $bot_id, ''); | |
| 2783 | + $total_count = $records['total'] ?? 0; | |
| 2784 | + | |
| 2785 | + // For Pinecone, we don't return individual entries during polling | |
| 2786 | + // (entries are already displayed on page load via mxchat_fetch_pinecone_records) | |
| 2787 | + // We just return the updated count. Over 500 vectors that is the live | |
| 2788 | + // index / namespace vector count climbing during an import (plan dd6e10). | |
| 2789 | + wp_send_json_success(array( | |
| 2790 | + 'entries' => array(), | |
| 2791 | + 'total_count' => absint($total_count), | |
| 2792 | + 'count_unit' => $records['count_unit'] ?? '', | |
| 2793 | + 'max_id' => $last_id, | |
| 2794 | + 'data_source' => 'pinecone' | |
| 2795 | + )); | |
| 2796 | + return; | |
| 2797 | + } | |
| 2798 | + | |
| 2799 | + // WORDPRESS DB DATA SOURCE | |
| 2800 | + // Build query to get entries newer than last_id | |
| 2801 | + $where_clauses = array('1=1'); | |
| 2802 | + $where_values = array(); | |
| 2803 | + | |
| 2804 | + if ($last_id > 0) { | |
| 2805 | + $where_clauses[] = 'id > %d'; | |
| 2806 | + $where_values[] = $last_id; | |
| 2807 | + } | |
| 2808 | + | |
| 2809 | + // Note: WordPress DB table doesn't have bot_id column | |
| 2810 | + // Multi-bot filtering is handled via Pinecone namespaces | |
| 2811 | + | |
| 2812 | + $where_sql = implode(' AND ', $where_clauses); | |
| 2813 | + | |
| 2814 | + // Get recent entries | |
| 2815 | + $query = "SELECT id, article_content, source_url, timestamp | |
| 2816 | + FROM $table_name | |
| 2817 | + WHERE $where_sql | |
| 2818 | + ORDER BY id DESC | |
| 2819 | + LIMIT %d"; | |
| 2820 | + | |
| 2821 | + $where_values[] = $limit; | |
| 2822 | + | |
| 2823 | + $entries = $wpdb->get_results($wpdb->prepare($query, $where_values)); | |
| 2824 | + | |
| 2825 | + // Get total count of GROUPED entries (by source_url) - matches pagination display | |
| 2826 | + // Count unique source_urls (excluding mxchat:// internal refs) + count of ungrouped rows | |
| 2827 | + $total_count = $wpdb->get_var( | |
| 2828 | + "SELECT (SELECT COUNT(DISTINCT source_url) FROM {$table_name} WHERE source_url != '' AND source_url NOT LIKE 'mxchat://%') + | |
| 2829 | + (SELECT COUNT(*) FROM {$table_name} WHERE source_url = '' OR source_url IS NULL OR source_url LIKE 'mxchat://%')" | |
| 2830 | + ); | |
| 2831 | + | |
| 2832 | + // Format entries for response | |
| 2833 | + $formatted_entries = array(); | |
| 2834 | + $preview_length = 150; | |
| 2835 | + foreach ($entries as $entry) { | |
| 2836 | + // Parse chunk metadata using the proper chunker method (same as initial page load) | |
| 2837 | + if (class_exists('MxChat_Chunker')) { | |
| 2838 | + $chunk_meta = MxChat_Chunker::parse_stored_chunk($entry->article_content); | |
| 2839 | + $display_content = $chunk_meta['text']; | |
| 2840 | + $chunk_metadata = $chunk_meta['metadata']; | |
| 2841 | + } else { | |
| 2842 | + $display_content = $entry->article_content; | |
| 2843 | + $chunk_metadata = array(); | |
| 2844 | + } | |
| 2845 | + | |
| 2846 | + $content_preview = mb_strlen($display_content) > $preview_length | |
| 2847 | + ? mb_substr($display_content, 0, $preview_length) . '...' | |
| 2848 | + : $display_content; | |
| 2849 | + | |
| 2850 | + $formatted_entries[] = array( | |
| 2851 | + 'id' => $entry->id, | |
| 2852 | + 'preview' => esc_html($content_preview), | |
| 2853 | + 'full_content' => wp_kses_post(wpautop($display_content)), | |
| 2854 | + 'content_length' => mb_strlen($display_content), | |
| 2855 | + 'preview_length' => $preview_length, | |
| 2856 | + 'source_url' => $entry->source_url, | |
| 2857 | + 'has_link' => !empty($entry->source_url) && strpos($entry->source_url, 'mxchat://') !== 0, | |
| 2858 | + 'chunk_metadata' => $chunk_metadata, | |
| 2859 | + 'bot_id' => $entry->bot_id ?? 'default', | |
| 2860 | + 'edit_nonce' => wp_create_nonce('mxchat_edit_entry_nonce'), | |
| 2861 | + 'delete_nonce' => wp_create_nonce('mxchat_delete_prompt_nonce') | |
| 2862 | + ); | |
| 2863 | + } | |
| 2864 | + | |
| 2865 | + wp_send_json_success(array( | |
| 2866 | + 'entries' => $formatted_entries, | |
| 2867 | + 'total_count' => absint($total_count), | |
| 2868 | + 'max_id' => !empty($entries) ? $entries[0]->id : $last_id, | |
| 2869 | + 'data_source' => 'wordpress' | |
| 2870 | + )); | |
| 2871 | +} | |
| 2872 | + | |
| 2873 | +/** | |
| 2874 | + * Get Pinecone total count from stats API | |
| 2875 | + * Helper function for ajax_mxchat_get_recent_entries | |
| 2876 | + */ | |
| 2877 | +private function mxchat_get_pinecone_count_from_stats($pinecone_options) { | |
| 2878 | + $api_key = $pinecone_options['mxchat_pinecone_api_key'] ?? ''; | |
| 2879 | + $host = $pinecone_options['mxchat_pinecone_host'] ?? ''; | |
| 2880 | + $namespace = $pinecone_options['mxchat_pinecone_namespace'] ?? ''; | |
| 2881 | + | |
| 2882 | + if (empty($api_key) || empty($host)) { | |
| 2883 | + return 0; | |
| 2884 | + } | |
| 2885 | + | |
| 2886 | + // Document index (plan 362c31). | |
| 2887 | + if (class_exists('MxChat_Pinecone_Documents') && MxChat_Pinecone_Documents::is_document_options($pinecone_options)) { | |
| 2888 | + return MxChat_Pinecone_Documents::count(MxChat_Pinecone_Documents::cfg_from_options($pinecone_options)); | |
| 2889 | + } | |
| 2890 | + | |
| 2891 | + try { | |
| 2892 | + $stats_url = "https://{$host}/describe_index_stats"; | |
| 2893 | + | |
| 2894 | + $response = wp_remote_post($stats_url, array( | |
| 2895 | + 'headers' => array( | |
| 2896 | + 'Api-Key' => $api_key, | |
| 2897 | + 'Content-Type' => 'application/json' | |
| 2898 | + ), | |
| 2899 | + 'body' => '{}', | |
| 2900 | + 'timeout' => 10 | |
| 2901 | + )); | |
| 2902 | + | |
| 2903 | + if (!is_wp_error($response) && wp_remote_retrieve_response_code($response) === 200) { | |
| 2904 | + $body = wp_remote_retrieve_body($response); | |
| 2905 | + $stats_data = json_decode($body, true); | |
| 2906 | + | |
| 2907 | + // If namespace is specified, get count from that specific namespace | |
| 2908 | + if (!empty($namespace) && isset($stats_data['namespaces'][$namespace]['vectorCount'])) { | |
| 2909 | + return intval($stats_data['namespaces'][$namespace]['vectorCount']); | |
| 2910 | + } | |
| 2911 | + | |
| 2912 | + // If no namespace specified or namespace not found in response, use total | |
| 2913 | + return intval($stats_data['totalVectorCount'] ?? 0); | |
| 2914 | + } | |
| 2915 | + | |
| 2916 | + return 0; | |
| 2917 | + | |
| 2918 | + } catch (Exception $e) { | |
| 2919 | + return 0; | |
| 2920 | + } | |
| 2921 | +} | |
| 2922 | + | |
| 2923 | +/** | |
| 2924 | + * AJAX handler to refresh Pinecone entries table via AJAX | |
| 2925 | + * Returns the table HTML for updating the UI without a full page reload | |
| 2926 | + */ | |
| 2927 | +public function ajax_mxchat_refresh_pinecone_entries() { | |
| 2928 | + check_ajax_referer('mxchat_entries_nonce', 'nonce'); | |
| 2929 | + | |
| 2930 | + if (!current_user_can('manage_options')) { | |
| 2931 | + wp_send_json_error(array('message' => 'Unauthorized')); | |
| 2932 | + return; | |
| 2933 | + } | |
| 2934 | + | |
| 2935 | + $bot_id = isset($_POST['bot_id']) ? sanitize_text_field($_POST['bot_id']) : 'default'; | |
| 2936 | + $page = isset($_POST['page']) ? absint($_POST['page']) : 1; | |
| 2937 | + $per_page = 25; | |
| 2938 | + $search_query = isset($_POST['search']) ? sanitize_text_field($_POST['search']) : ''; | |
| 2939 | + $content_type_filter = isset($_POST['content_type']) ? sanitize_key($_POST['content_type']) : ''; | |
| 2940 | + | |
| 2941 | + // Get Pinecone manager and options | |
| 2942 | + $pinecone_manager = $this->mxchat_get_pinecone_manager(); | |
| 2943 | + if (!$pinecone_manager) { | |
| 2944 | + wp_send_json_error(array('message' => 'Pinecone manager not available')); | |
| 2945 | + return; | |
| 2946 | + } | |
| 2947 | + | |
| 2948 | + $pinecone_options = $pinecone_manager->mxchat_get_bot_pinecone_options($bot_id); | |
| 2949 | + $use_pinecone = ($pinecone_options['mxchat_use_pinecone'] ?? '0') === '1'; | |
| 2950 | + $pinecone_api_key = $pinecone_options['mxchat_pinecone_api_key'] ?? ''; | |
| 2951 | + | |
| 2952 | + if (!$use_pinecone || empty($pinecone_api_key)) { | |
| 2953 | + wp_send_json_error(array('message' => 'Pinecone not configured')); | |
| 2954 | + return; | |
| 2955 | + } | |
| 2956 | + | |
| 2957 | + // Fetch records from Pinecone | |
| 2958 | + $records = $pinecone_manager->mxchat_fetch_pinecone_records($pinecone_options, $search_query, $page, $per_page, $bot_id, $content_type_filter); | |
| 2959 | + $prompts = $records['data'] ?? array(); | |
| 2960 | + $total_records = $records['total'] ?? 0; | |
| 2961 | + | |
| 2962 | + // Preprocess Pinecone records — set chunk_metadata and display_content | |
| 2963 | + // (matches admin-knowledge-page.php preprocessing) | |
| 2964 | + foreach ($prompts as $prompt) { | |
| 2965 | + if (isset($prompt->chunk_index) && $prompt->chunk_index !== null) { | |
| 2966 | + $prompt->chunk_metadata = array( | |
| 2967 | + 'chunk_index' => intval($prompt->chunk_index), | |
| 2968 | + 'total_chunks' => isset($prompt->total_chunks) ? intval($prompt->total_chunks) : null, | |
| 2969 | + 'is_chunked' => isset($prompt->is_chunked) ? (bool) $prompt->is_chunked : true, | |
| 2970 | + 'source_url' => $prompt->source_url ?? '' | |
| 2971 | + ); | |
| 2972 | + $prompt->display_content = $prompt->article_content; | |
| 2973 | + } else { | |
| 2974 | + $prompt->chunk_metadata = array(); | |
| 2975 | + $prompt->display_content = $prompt->article_content ?? ''; | |
| 2976 | + } | |
| 2977 | + } | |
| 2978 | + | |
| 2979 | + // Group prompts by source_url | |
| 2980 | + $grouped_prompts = array(); | |
| 2981 | + foreach ($prompts as $prompt) { | |
| 2982 | + $source_url = ''; | |
| 2983 | + if (!empty($prompt->chunk_metadata['source_url'])) { | |
| 2984 | + $source_url = $prompt->chunk_metadata['source_url']; | |
| 2985 | + } elseif (!empty($prompt->source_url)) { | |
| 2986 | + $source_url = $prompt->source_url; | |
| 2987 | + } | |
| 2988 | + | |
| 2989 | + if (!empty($source_url)) { | |
| 2990 | + if (!isset($grouped_prompts[$source_url])) { | |
| 2991 | + $grouped_prompts[$source_url] = array(); | |
| 2992 | + } | |
| 2993 | + $grouped_prompts[$source_url][] = $prompt; | |
| 2994 | + } else { | |
| 2995 | + $grouped_prompts['_ungrouped_' . $prompt->id] = array($prompt); | |
| 2996 | + } | |
| 2997 | + } | |
| 2998 | + | |
| 2999 | + // Sort each group by chunk_index | |
| 3000 | + foreach ($grouped_prompts as $source_url => &$group) { | |
| 3001 | + usort($group, function($a, $b) { | |
| 3002 | + $index_a = isset($a->chunk_metadata['chunk_index']) ? intval($a->chunk_metadata['chunk_index']) : 0; | |
| 3003 | + $index_b = isset($b->chunk_metadata['chunk_index']) ? intval($b->chunk_metadata['chunk_index']) : 0; | |
| 3004 | + return $index_a - $index_b; | |
| 3005 | + }); | |
| 3006 | + } | |
| 3007 | + unset($group); | |
| 3008 | + | |
| 3009 | + // Build HTML for the table rows - must match admin-knowledge-page.php structure exactly | |
| 3010 | + ob_start(); | |
| 3011 | + $display_index = 0; | |
| 3012 | + $current_page = $page; | |
| 3013 | + $data_source = 'pinecone'; | |
| 3014 | + $current_bot_id = $bot_id; | |
| 3015 | + $preview_length = 150; | |
| 3016 | + | |
| 3017 | + if (empty($grouped_prompts)) { | |
| 3018 | + echo '<tr><td colspan="4" style="padding: 40px; text-align: center; color: var(--mxch-text-muted);">'; | |
| 3019 | + esc_html_e('No knowledge entries found in Pinecone.', 'mxchat'); | |
| 3020 | + echo '</td></tr>'; | |
| 3021 | + } else { | |
| 3022 | + foreach ($grouped_prompts as $source_url => $group) { | |
| 3023 | + $chunk_count = count($group); | |
| 3024 | + $first_prompt = $group[0]; | |
| 3025 | + $display_index++; | |
| 3026 | + | |
| 3027 | + if ($chunk_count > 1) { | |
| 3028 | + // Multiple chunks - show grouped row with expand button | |
| 3029 | + $group_id = 'group-' . md5($source_url); | |
| 3030 | + ?> | |
| 3031 | + <tr id="prompt-<?php echo esc_attr($first_prompt->id); ?>" | |
| 3032 | + class="mxchat-chunk-group-header" | |
| 3033 | + data-source="<?php echo esc_attr($data_source); ?>" | |
| 3034 | + data-group-id="<?php echo esc_attr($group_id); ?>" | |
| 3035 | + style="border-bottom: 1px solid var(--mxch-card-border); background: rgba(33, 150, 243, 0.02);"> | |
| 3036 | + <td style="padding: 12px 16px; text-align: center;"> | |
| 3037 | + <input type="checkbox" | |
| 3038 | + class="mxchat-entry-checkbox" | |
| 3039 | + data-entry-id="<?php echo esc_attr($first_prompt->id); ?>" | |
| 3040 | + data-source="<?php echo esc_attr($data_source); ?>" | |
| 3041 | + data-source-url="<?php echo esc_attr($source_url); ?>" | |
| 3042 | + data-is-group="true" | |
| 3043 | + data-chunk-count="<?php echo esc_attr($chunk_count); ?>"> | |
| 3044 | + </td> | |
| 3045 | + <td style="padding: 12px 16px; font-size: 13px;"> | |
| 3046 | + <?php echo esc_html($display_index + (($current_page - 1) * $per_page)); ?> | |
| 3047 | + </td> | |
| 3048 | + <td class="mxchat-content-cell" style="padding: 12px 16px; font-size: 13px;"> | |
| 3049 | + <div class="mxchat-chunk-group-info"> | |
| 3050 | + <button type="button" class="mxchat-chunk-toggle" data-group-id="<?php echo esc_attr($group_id); ?>"> | |
| 3051 | + <span class="dashicons dashicons-arrow-right-alt2"></span> | |
| 3052 | + </button> | |
| 3053 | + <span class="mxchat-chunk-badge"><?php echo esc_html($chunk_count); ?> <?php esc_html_e('chunks', 'mxchat'); ?></span> | |
| 3054 | + <span class="mxchat-chunk-preview"> | |
| 3055 | + <?php | |
| 3056 | + $parent_content = isset($first_prompt->display_content) ? $first_prompt->display_content : (isset($first_prompt->article_content) ? $first_prompt->article_content : ''); | |
| 3057 | + $content_preview = mb_substr($parent_content, 0, 100); | |
| 3058 | + echo esc_html($content_preview . '...'); | |
| 3059 | + ?> | |
| 3060 | + </span> | |
| 3061 | + </div> | |
| 3062 | + </td> | |
| 3063 | + <td class="mxchat-url-cell" style="padding: 12px 16px; font-size: 13px;"> | |
| 3064 | + <?php if (!empty($source_url) && strpos($source_url, 'mxchat://') !== 0 && strpos($source_url, '_ungrouped_') !== 0) : ?> | |
| 3065 | + <a href="<?php echo esc_url($source_url); ?>" target="_blank" style="color: var(--mxch-primary); text-decoration: none;"> | |
| 3066 | + <span class="dashicons dashicons-external" style="font-size: 14px;"></span> | |
| 3067 | + <?php esc_html_e('View Source', 'mxchat'); ?> | |
| 3068 | + </a> | |
| 3069 | + <?php else : ?> | |
| 3070 | + <span style="color: var(--mxch-text-muted);"><?php esc_html_e('Manual Content', 'mxchat'); ?></span> | |
| 3071 | + <?php endif; ?> | |
| 3072 | + </td> | |
| 3073 | + <td class="mxchat-actions-cell" style="padding: 12px 16px; white-space: nowrap;"> | |
| 3074 | + <button type="button" | |
| 3075 | + class="mxch-btn mxch-btn-ghost mxch-btn-sm mxchat-inspect-entry-btn" | |
| 3076 | + data-source-url="<?php echo esc_attr($source_url); ?>" | |
| 3077 | + data-entry-id="<?php echo esc_attr($first_prompt->id); ?>" | |
| 3078 | + data-data-source="<?php echo esc_attr($data_source); ?>" | |
| 3079 | + data-bot-id="<?php echo esc_attr($current_bot_id); ?>" | |
| 3080 | + data-nonce="<?php echo wp_create_nonce('mxchat_inspect_entry_nonce'); ?>" | |
| 3081 | + title="<?php esc_attr_e('View indexed content', 'mxchat'); ?>"> | |
| 3082 | + <span class="dashicons dashicons-visibility" style="font-size: 14px;"></span> | |
| 3083 | + </button> | |
| 3084 | + <button type="button" | |
| 3085 | + class="mxch-btn mxch-btn-ghost mxch-btn-sm mxchat-edit-entry-btn" | |
| 3086 | + data-source-url="<?php echo esc_attr($source_url); ?>" | |
| 3087 | + data-entry-id="<?php echo esc_attr($first_prompt->id); ?>" | |
| 3088 | + data-data-source="<?php echo esc_attr($data_source); ?>" | |
| 3089 | + data-bot-id="<?php echo esc_attr($current_bot_id); ?>" | |
| 3090 | + data-nonce="<?php echo wp_create_nonce('mxchat_edit_entry_nonce'); ?>" | |
| 3091 | + title="<?php esc_attr_e('Edit content', 'mxchat'); ?>"> | |
| 3092 | + <span class="dashicons dashicons-edit" style="font-size: 14px;"></span> | |
| 3093 | + </button> | |
| 3094 | + <button type="button" | |
| 3095 | + class="mxch-btn mxch-btn-ghost mxch-btn-sm delete-button-group" | |
| 3096 | + data-source-url="<?php echo esc_attr($source_url); ?>" | |
| 3097 | + data-chunk-count="<?php echo esc_attr($chunk_count); ?>" | |
| 3098 | + data-data-source="<?php echo esc_attr($data_source); ?>" | |
| 3099 | + data-bot-id="<?php echo esc_attr($current_bot_id); ?>" | |
| 3100 | + data-nonce="<?php echo wp_create_nonce('mxchat_delete_chunks_nonce'); ?>" | |
| 3101 | + style="color: var(--mxch-error);" | |
| 3102 | + title="<?php esc_attr_e('Delete all chunks', 'mxchat'); ?>"> | |
| 3103 | + <span class="dashicons dashicons-trash" style="font-size: 14px;"></span> | |
| 3104 | + </button> | |
| 3105 | + </td> | |
| 3106 | + </tr> | |
| 3107 | + <?php | |
| 3108 | + // Render hidden chunk rows | |
| 3109 | + foreach ($group as $chunk_index => $chunk) { | |
| 3110 | + $meta_chunk_index = isset($chunk->chunk_metadata['chunk_index']) ? intval($chunk->chunk_metadata['chunk_index']) : $chunk_index; | |
| 3111 | + $meta_total_chunks = isset($chunk->chunk_metadata['total_chunks']) ? intval($chunk->chunk_metadata['total_chunks']) : $chunk_count; | |
| 3112 | + $content = isset($chunk->display_content) ? $chunk->display_content : (isset($chunk->article_content) ? $chunk->article_content : ''); | |
| 3113 | + $content_preview = mb_strlen($content) > $preview_length | |
| 3114 | + ? mb_substr($content, 0, $preview_length) . '...' | |
| 3115 | + : $content; | |
| 3116 | + ?> | |
| 3117 | + <tr id="prompt-<?php echo esc_attr($chunk->id); ?>" | |
| 3118 | + class="mxchat-chunk-row <?php echo esc_attr($group_id); ?>" | |
| 3119 | + data-source="<?php echo esc_attr($data_source); ?>" | |
| 3120 | + style="display: none; background: #f8f9fa; border-bottom: 1px solid var(--mxch-card-border);"> | |
| 3121 | + <td style="padding: 12px 16px; text-align: center;"> | |
| 3122 | + <!-- Checkbox column placeholder for chunks (managed by group) --> | |
| 3123 | + </td> | |
| 3124 | + <td style="padding: 12px 16px 12px 30px; font-size: 13px;"> | |
| 3125 | + <!-- Hidden ID column for chunks --> | |
| 3126 | + </td> | |
| 3127 | + <td class="mxchat-content-cell" style="padding: 12px 16px; font-size: 13px;"> | |
| 3128 | + <div class="mxchat-accordion-wrapper"> | |
| 3129 | + <div class="mxchat-content-preview"> | |
| 3130 | + <span class="mxchat-chunk-indicator" style="margin-right: 10px; color: var(--mxch-text-secondary); font-size: 12px;"> | |
| 3131 | + <?php printf(esc_html__('Chunk %d of %d', 'mxchat'), $meta_chunk_index + 1, $meta_total_chunks); ?> | |
| 3132 | + </span> | |
| 3133 | + <span class="preview-text"><?php echo esc_html($content_preview); ?></span> | |
| 3134 | + <?php if (mb_strlen($content) > $preview_length) : ?> | |
| 3135 | + <button class="mxchat-expand-toggle" type="button"> | |
| 3136 | + <span class="dashicons dashicons-arrow-down-alt2"></span> | |
| 3137 | + </button> | |
| 3138 | + <?php endif; ?> | |
| 3139 | + </div> | |
| 3140 | + <div class="mxchat-content-full" style="display: none;"> | |
| 3141 | + <div class="content-view"> | |
| 3142 | + <?php | |
| 3143 | + if (preg_match('/[\x{0590}-\x{05FF}]/u', $content)) { | |
| 3144 | + echo '<div dir="rtl" lang="he" class="rtl-content">'; | |
| 3145 | + echo wp_kses_post(wpautop($content)); | |
| 3146 | + echo '</div>'; | |
| 3147 | + } else { | |
| 3148 | + echo wp_kses_post(wpautop($content)); | |
| 3149 | + } | |
| 3150 | + ?> | |
| 3151 | + </div> | |
| 3152 | + </div> | |
| 3153 | + </div> | |
| 3154 | + </td> | |
| 3155 | + <td class="mxchat-url-cell" style="padding: 12px 16px; font-size: 13px;"> | |
| 3156 | + <span class="mxchat-chunk-label" style="color: var(--mxch-text-muted);"><?php esc_html_e('Same as parent', 'mxchat'); ?></span> | |
| 3157 | + </td> | |
| 3158 | + <td class="mxchat-actions-cell" style="padding: 12px 16px;"> | |
| 3159 | + <span class="mxchat-chunk-label" style="color: var(--mxch-text-muted);"><?php esc_html_e('Managed by group', 'mxchat'); ?></span> | |
| 3160 | + </td> | |
| 3161 | + </tr> | |
| 3162 | + <?php | |
| 3163 | + } | |
| 3164 | + } else { | |
| 3165 | + // Single entry - display normally with accordion | |
| 3166 | + $prompt = $first_prompt; | |
| 3167 | + $content = isset($prompt->display_content) ? $prompt->display_content : (isset($prompt->article_content) ? $prompt->article_content : ''); | |
| 3168 | + $content_preview = mb_strlen($content) > $preview_length | |
| 3169 | + ? mb_substr($content, 0, $preview_length) . '...' | |
| 3170 | + : $content; | |
| 3171 | + ?> | |
| 3172 | + <tr id="prompt-<?php echo esc_attr($prompt->id); ?>" | |
| 3173 | + data-source="<?php echo esc_attr($data_source); ?>" | |
| 3174 | + style="border-bottom: 1px solid var(--mxch-card-border); background: rgba(33, 150, 243, 0.02);"> | |
| 3175 | + <td style="padding: 12px 16px; text-align: center;"> | |
| 3176 | + <input type="checkbox" | |
| 3177 | + class="mxchat-entry-checkbox" | |
| 3178 | + data-entry-id="<?php echo esc_attr($prompt->id); ?>" | |
| 3179 | + data-source="<?php echo esc_attr($data_source); ?>" | |
| 3180 | + data-source-url="<?php echo esc_attr($prompt->source_url ?? ''); ?>" | |
| 3181 | + data-is-group="false" | |
| 3182 | + data-chunk-count="1"> | |
| 3183 | + </td> | |
| 3184 | + <td style="padding: 12px 16px; font-size: 13px;"> | |
| 3185 | + <?php echo esc_html($display_index + (($current_page - 1) * $per_page)); ?> | |
| 3186 | + </td> | |
| 3187 | + <td class="mxchat-content-cell" style="padding: 12px 16px; font-size: 13px;"> | |
| 3188 | + <div class="mxchat-accordion-wrapper"> | |
| 3189 | + <div class="mxchat-content-preview"> | |
| 3190 | + <span class="preview-text"><?php echo esc_html($content_preview); ?></span> | |
| 3191 | + <?php if (mb_strlen($content) > $preview_length) : ?> | |
| 3192 | + <button class="mxchat-expand-toggle" type="button"> | |
| 3193 | + <span class="dashicons dashicons-arrow-down-alt2"></span> | |
| 3194 | + </button> | |
| 3195 | + <?php endif; ?> | |
| 3196 | + </div> | |
| 3197 | + <div class="mxchat-content-full" style="display: none;"> | |
| 3198 | + <div class="content-view"> | |
| 3199 | + <?php | |
| 3200 | + if (preg_match('/[\x{0590}-\x{05FF}]/u', $content)) { | |
| 3201 | + echo '<div dir="rtl" lang="he" class="rtl-content">'; | |
| 3202 | + echo wp_kses_post(wpautop($content)); | |
| 3203 | + echo '</div>'; | |
| 3204 | + } else { | |
| 3205 | + echo wp_kses_post(wpautop($content)); | |
| 3206 | + } | |
| 3207 | + ?> | |
| 3208 | + </div> | |
| 3209 | + </div> | |
| 3210 | + </div> | |
| 3211 | + </td> | |
| 3212 | + <td class="mxchat-url-cell" style="padding: 12px 16px; font-size: 13px;"> | |
| 3213 | + <?php | |
| 3214 | + $actual_source = $source_url; | |
| 3215 | + if (strpos($source_url, '_ungrouped_') === 0) { | |
| 3216 | + $actual_source = $prompt->source_url ?? ''; | |
| 3217 | + } | |
| 3218 | + if (!empty($actual_source) && strpos($actual_source, 'mxchat://') !== 0) : ?> | |
| 3219 | + <a href="<?php echo esc_url($actual_source); ?>" target="_blank" style="color: var(--mxch-primary); text-decoration: none;"> | |
| 3220 | + <span class="dashicons dashicons-external" style="font-size: 14px;"></span> | |
| 3221 | + <?php esc_html_e('View', 'mxchat'); ?> | |
| 3222 | + </a> | |
| 3223 | + <?php else : ?> | |
| 3224 | + <span style="color: var(--mxch-text-muted);"><?php esc_html_e('Manual', 'mxchat'); ?></span> | |
| 3225 | + <?php endif; ?> | |
| 3226 | + </td> | |
| 3227 | + <td style="padding: 12px 16px; white-space: nowrap;"> | |
| 3228 | + <button type="button" | |
| 3229 | + class="mxch-btn mxch-btn-ghost mxch-btn-sm mxchat-inspect-entry-btn" | |
| 3230 | + data-source-url="<?php echo esc_attr($prompt->source_url ?? ''); ?>" | |
| 3231 | + data-entry-id="<?php echo esc_attr($prompt->id); ?>" | |
| 3232 | + data-data-source="<?php echo esc_attr($data_source); ?>" | |
| 3233 | + data-bot-id="<?php echo esc_attr($current_bot_id); ?>" | |
| 3234 | + data-nonce="<?php echo wp_create_nonce('mxchat_inspect_entry_nonce'); ?>" | |
| 3235 | + title="<?php esc_attr_e('View indexed content', 'mxchat'); ?>"> | |
| 3236 | + <span class="dashicons dashicons-visibility" style="font-size: 14px;"></span> | |
| 3237 | + </button> | |
| 3238 | + <button type="button" | |
| 3239 | + class="mxch-btn mxch-btn-ghost mxch-btn-sm mxchat-edit-entry-btn" | |
| 3240 | + data-source-url="<?php echo esc_attr($prompt->source_url ?? ''); ?>" | |
| 3241 | + data-entry-id="<?php echo esc_attr($prompt->id); ?>" | |
| 3242 | + data-data-source="<?php echo esc_attr($data_source); ?>" | |
| 3243 | + data-bot-id="<?php echo esc_attr($current_bot_id); ?>" | |
| 3244 | + data-nonce="<?php echo wp_create_nonce('mxchat_edit_entry_nonce'); ?>" | |
| 3245 | + title="<?php esc_attr_e('Edit content', 'mxchat'); ?>"> | |
| 3246 | + <span class="dashicons dashicons-edit" style="font-size: 14px;"></span> | |
| 3247 | + </button> | |
| 3248 | + <button type="button" class="mxch-btn mxch-btn-ghost mxch-btn-sm delete-button-ajax" data-vector-id="<?php echo esc_attr($prompt->id); ?>" data-bot-id="<?php echo esc_attr($current_bot_id); ?>" data-nonce="<?php echo wp_create_nonce('mxchat_delete_pinecone_prompt_nonce'); ?>" style="color: var(--mxch-error);"> | |
| 3249 | + <span class="dashicons dashicons-trash" style="font-size: 14px;"></span> | |
| 3250 | + </button> | |
| 3251 | + </td> | |
| 3252 | + </tr> | |
| 3253 | + <?php | |
| 3254 | + } | |
| 3255 | + } | |
| 3256 | + } | |
| 3257 | + $html = ob_get_clean(); | |
| 3258 | + | |
| 3259 | + // Generate pagination HTML for Pinecone | |
| 3260 | + $total_pages = ceil($total_records / $per_page); | |
| 3261 | + $count_unit = $records['count_unit'] ?? ''; | |
| 3262 | + $cursor_pagination = !empty($records['cursor_pagination']); | |
| 3263 | + $has_next = !empty($records['has_next']); | |
| 3264 | + $pagination_html = ''; | |
| 3265 | + if ($cursor_pagination) { | |
| 3266 | + // Plan dd6e10: cursor path — Previous / Next only, no numbered pages. | |
| 3267 | + $pagination_html = self::mxchat_cursor_pagination_html($page, $has_next, $search_query, $content_type_filter); | |
| 3268 | + } elseif ($total_pages > 1) { | |
| 3269 | + $pagination_html = '<div class="mxchat-ajax-pagination" data-current-page="' . esc_attr($page) . '" data-total-pages="' . esc_attr($total_pages) . '" data-search="' . esc_attr($search_query) . '" data-content-type="' . esc_attr($content_type_filter) . '">'; | |
| 3270 | + | |
| 3271 | + // Previous button | |
| 3272 | + if ($page > 1) { | |
| 3273 | + $pagination_html .= '<a href="#" class="mxchat-page-link" data-page="' . ($page - 1) . '">' . esc_html__('« Previous', 'mxchat') . '</a> '; | |
| 3274 | + } | |
| 3275 | + | |
| 3276 | + // Page numbers | |
| 3277 | + $start_page = max(1, $page - 2); | |
| 3278 | + $end_page = min($total_pages, $page + 2); | |
| 3279 | + | |
| 3280 | + if ($start_page > 1) { | |
| 3281 | + $pagination_html .= '<a href="#" class="mxchat-page-link" data-page="1">1</a> '; | |
| 3282 | + if ($start_page > 2) { | |
| 3283 | + $pagination_html .= '<span class="mxchat-page-dots">...</span> '; | |
| 3284 | + } | |
| 3285 | + } | |
| 3286 | + | |
| 3287 | + for ($i = $start_page; $i <= $end_page; $i++) { | |
| 3288 | + if ($i == $page) { | |
| 3289 | + $pagination_html .= '<span class="mxchat-page-current">' . $i . '</span> '; | |
| 3290 | + } else { | |
| 3291 | + $pagination_html .= '<a href="#" class="mxchat-page-link" data-page="' . $i . '">' . $i . '</a> '; | |
| 3292 | + } | |
| 3293 | + } | |
| 3294 | + | |
| 3295 | + if ($end_page < $total_pages) { | |
| 3296 | + if ($end_page < $total_pages - 1) { | |
| 3297 | + $pagination_html .= '<span class="mxchat-page-dots">...</span> '; | |
| 3298 | + } | |
| 3299 | + $pagination_html .= '<a href="#" class="mxchat-page-link" data-page="' . $total_pages . '">' . $total_pages . '</a> '; | |
| 3300 | + } | |
| 3301 | + | |
| 3302 | + // Next button | |
| 3303 | + if ($page < $total_pages) { | |
| 3304 | + $pagination_html .= '<a href="#" class="mxchat-page-link" data-page="' . ($page + 1) . '">' . esc_html__('Next »', 'mxchat') . '</a>'; | |
| 3305 | + } | |
| 3306 | + | |
| 3307 | + $pagination_html .= '</div>'; | |
| 3308 | + } | |
| 3309 | + | |
| 3310 | + wp_send_json_success(array( | |
| 3311 | + 'html' => $html, | |
| 3312 | + 'pagination_html' => $pagination_html, | |
| 3313 | + 'total_count' => $total_records, | |
| 3314 | + 'count_unit' => $count_unit, | |
| 3315 | + 'total_pages' => $total_pages, | |
| 3316 | + 'page' => $page, | |
| 3317 | + 'per_page' => $per_page, | |
| 3318 | + 'data_source' => 'pinecone' | |
| 3319 | + )); | |
| 3320 | +} | |
| 3321 | + | |
| 3322 | +/** | |
| 3323 | + * Previous / Next markup for the Pinecone cursor path (plan dd6e10). | |
| 3324 | + * | |
| 3325 | + * Pinecone's /vectors/list has no random access, so numbered pages would be a | |
| 3326 | + * fiction; the page number here is only the cursor step, which the manager | |
| 3327 | + * maps to a stored paginationToken. Used by the page template on first render | |
| 3328 | + * and by the AJAX paginator, so the two can never drift. | |
| 3329 | + */ | |
| 3330 | +public static function mxchat_cursor_pagination_html($page, $has_next, $search_query = '', $content_type_filter = '') { | |
| 3331 | + $page = max(1, absint($page)); | |
| 3332 | + if ($page <= 1 && !$has_next) { | |
| 3333 | + return ''; | |
| 3334 | + } | |
| 3335 | + $html = '<div class="mxchat-ajax-pagination mxchat-cursor-pagination" data-current-page="' . esc_attr($page) . '" data-total-pages="" data-search="' . esc_attr($search_query) . '" data-content-type="' . esc_attr($content_type_filter) . '">'; | |
| 3336 | + if ($page > 1) { | |
| 3337 | + $html .= '<a href="#" class="mxchat-page-link" data-page="' . ($page - 1) . '">' . esc_html__('« Previous', 'mxchat') . '</a> '; | |
| 3338 | + } | |
| 3339 | + $html .= '<span class="mxchat-page-current">' . esc_html($page) . '</span> '; | |
| 3340 | + if ($has_next) { | |
| 3341 | + $html .= '<a href="#" class="mxchat-page-link" data-page="' . ($page + 1) . '">' . esc_html__('Next »', 'mxchat') . '</a>'; | |
| 3342 | + } | |
| 3343 | + $html .= '</div>'; | |
| 3344 | + return $html; | |
| 3345 | +} | |
| 3346 | + | |
| 3347 | +/** | |
| 3348 | + * AJAX handler for pagination - handles both WordPress DB and Pinecone data sources | |
| 3349 | + * Returns paginated entries without requiring a full page reload | |
| 3350 | + */ | |
| 3351 | +public function ajax_mxchat_paginate_entries() { | |
| 3352 | + check_ajax_referer('mxchat_entries_nonce', 'nonce'); | |
| 3353 | + | |
| 3354 | + if (!current_user_can('manage_options')) { | |
| 3355 | + wp_send_json_error(array('message' => 'Unauthorized')); | |
| 3356 | + return; | |
| 3357 | + } | |
| 3358 | + | |
| 3359 | + $bot_id = isset($_POST['bot_id']) ? sanitize_text_field($_POST['bot_id']) : 'default'; | |
| 3360 | + $page = isset($_POST['page']) ? absint($_POST['page']) : 1; | |
| 3361 | + $search_query = isset($_POST['search']) ? sanitize_text_field($_POST['search']) : ''; | |
| 3362 | + $content_type_filter = isset($_POST['content_type']) ? sanitize_key($_POST['content_type']) : ''; | |
| 3363 | + $per_page = 25; | |
| 3364 | + | |
| 3365 | + // Check if Pinecone is enabled for this bot | |
| 3366 | + $pinecone_manager = $this->mxchat_get_pinecone_manager(); | |
| 3367 | + $pinecone_options = $pinecone_manager ? $pinecone_manager->mxchat_get_bot_pinecone_options($bot_id) : array(); | |
| 3368 | + $use_pinecone = ($pinecone_options['mxchat_use_pinecone'] ?? '0') === '1'; | |
| 3369 | + $has_pinecone_api = !empty($pinecone_options['mxchat_pinecone_api_key']); | |
| 3370 | + | |
| 3371 | + if ($use_pinecone && $has_pinecone_api) { | |
| 3372 | + // Delegate to Pinecone pagination handler (pass search params) | |
| 3373 | + $_POST['page'] = $page; | |
| 3374 | + $_POST['search'] = $search_query; | |
| 3375 | + $_POST['content_type'] = $content_type_filter; | |
| 3376 | + $this->ajax_mxchat_refresh_pinecone_entries(); | |
| 3377 | + return; | |
| 3378 | + } | |
| 3379 | + | |
| 3380 | + // WordPress DB pagination - MUST match initial page load logic exactly | |
| 3381 | + global $wpdb; | |
| 3382 | + $table_name = $wpdb->prefix . 'mxchat_system_prompt_content'; | |
| 3383 | + $offset = ($page - 1) * $per_page; | |
| 3384 | + | |
| 3385 | + // Build WHERE clause for search and content type filtering | |
| 3386 | + $where_clauses = array(); | |
| 3387 | + $where_values = array(); | |
| 3388 | + | |
| 3389 | + if ($search_query) { | |
| 3390 | + $where_clauses[] = "article_content LIKE %s"; | |
| 3391 | + $where_values[] = '%' . $wpdb->esc_like($search_query) . '%'; | |
| 3392 | + } | |
| 3393 | + | |
| 3394 | + if ($content_type_filter) { | |
| 3395 | + switch ($content_type_filter) { | |
| 3396 | + case 'manual': | |
| 3397 | + $where_clauses[] = "(source_url = '' OR source_url IS NULL OR source_url LIKE 'mxchat://%')"; | |
| 3398 | + break; | |
| 3399 | + case 'pdf': | |
| 3400 | + $where_clauses[] = "source_url LIKE '%.pdf'"; | |
| 3401 | + break; | |
| 3402 | + case 'url': | |
| 3403 | + $where_clauses[] = "source_url != '' AND source_url IS NOT NULL AND source_url NOT LIKE 'mxchat://%' AND source_url NOT LIKE '%.pdf'"; | |
| 3404 | + break; | |
| 3405 | + } | |
| 3406 | + } | |
| 3407 | + | |
| 3408 | + $where_sql = !empty($where_clauses) ? 'WHERE ' . implode(' AND ', $where_clauses) : ''; | |
| 3409 | + | |
| 3410 | + // Count grouped entries with filters applied | |
| 3411 | + if (!empty($where_values)) { | |
| 3412 | + $count_args = array_merge($where_values, $where_values); | |
| 3413 | + $count_query = $wpdb->prepare( | |
| 3414 | + "SELECT (SELECT COUNT(DISTINCT source_url) FROM {$table_name} {$where_sql} AND source_url != '' AND source_url NOT LIKE 'mxchat://%') + | |
| 3415 | + (SELECT COUNT(*) FROM {$table_name} {$where_sql} AND (source_url = '' OR source_url IS NULL OR source_url LIKE 'mxchat://%'))", | |
| 3416 | + ...$count_args | |
| 3417 | + ); | |
| 3418 | + $total_records = $wpdb->get_var($count_query); | |
| 3419 | + } else if (!empty($where_sql)) { | |
| 3420 | + // Content type filter only (no search), no prepared values needed | |
| 3421 | + $total_records = $wpdb->get_var( | |
| 3422 | + "SELECT (SELECT COUNT(DISTINCT source_url) FROM {$table_name} {$where_sql} AND source_url != '' AND source_url NOT LIKE 'mxchat://%') + | |
| 3423 | + (SELECT COUNT(*) FROM {$table_name} {$where_sql} AND (source_url = '' OR source_url IS NULL OR source_url LIKE 'mxchat://%'))" | |
| 3424 | + ); | |
| 3425 | + } else { | |
| 3426 | + // No filters | |
| 3427 | + $total_records = $wpdb->get_var( | |
| 3428 | + "SELECT (SELECT COUNT(DISTINCT source_url) FROM {$table_name} WHERE source_url != '' AND source_url NOT LIKE 'mxchat://%') + | |
| 3429 | + (SELECT COUNT(*) FROM {$table_name} WHERE source_url = '' OR source_url IS NULL OR source_url LIKE 'mxchat://%')" | |
| 3430 | + ); | |
| 3431 | + } | |
| 3432 | + $total_pages = ceil($total_records / $per_page); | |
| 3433 | + | |
| 3434 | + // Step 1: Get unique source_urls for this page (ordered by latest timestamp) with filters | |
| 3435 | + if (!empty($where_values)) { | |
| 3436 | + $query_args = array_merge($where_values, array($per_page, $offset)); | |
| 3437 | + $urls_query = $wpdb->prepare( | |
| 3438 | + "SELECT source_url, MAX(timestamp) as latest_ts FROM {$table_name} | |
| 3439 | + {$where_sql} | |
| 3440 | + GROUP BY source_url ORDER BY latest_ts DESC LIMIT %d OFFSET %d", | |
| 3441 | + ...$query_args | |
| 3442 | + ); | |
| 3443 | + } else if (!empty($where_sql)) { | |
| 3444 | + $urls_query = $wpdb->prepare( | |
| 3445 | + "SELECT source_url, MAX(timestamp) as latest_ts FROM {$table_name} | |
| 3446 | + {$where_sql} | |
| 3447 | + GROUP BY source_url ORDER BY latest_ts DESC LIMIT %d OFFSET %d", | |
| 3448 | + $per_page, $offset | |
| 3449 | + ); | |
| 3450 | + } else { | |
| 3451 | + $urls_query = $wpdb->prepare( | |
| 3452 | + "SELECT source_url, MAX(timestamp) as latest_ts FROM {$table_name} | |
| 3453 | + GROUP BY source_url ORDER BY latest_ts DESC LIMIT %d OFFSET %d", | |
| 3454 | + $per_page, $offset | |
| 3455 | + ); | |
| 3456 | + } | |
| 3457 | + $page_urls = $wpdb->get_results($urls_query); | |
| 3458 | + | |
| 3459 | + // Step 2: Build list of source_urls to fetch | |
| 3460 | + $url_list = array(); | |
| 3461 | + $url_order_map = array(); | |
| 3462 | + $order_index = 0; | |
| 3463 | + foreach ($page_urls as $url_row) { | |
| 3464 | + $url = $url_row->source_url; | |
| 3465 | + $url_list[] = $url; | |
| 3466 | + $url_order_map[$url] = $order_index++; | |
| 3467 | + } | |
| 3468 | + | |
| 3469 | + // Step 3: Fetch all rows for these source_urls (with search filter if applicable) | |
| 3470 | + $prompts = array(); | |
| 3471 | + if (!empty($url_list)) { | |
| 3472 | + $placeholders = implode(',', array_fill(0, count($url_list), '%s')); | |
| 3473 | + if ($search_query) { | |
| 3474 | + // Include search filter in the final fetch | |
| 3475 | + $prompts_query = $wpdb->prepare( | |
| 3476 | + "SELECT id, article_content, source_url, timestamp, role_restriction | |
| 3477 | + FROM {$table_name} | |
| 3478 | + WHERE source_url IN ($placeholders) AND article_content LIKE %s | |
| 3479 | + ORDER BY timestamp DESC", | |
| 3480 | + ...array_merge($url_list, ['%' . $wpdb->esc_like($search_query) . '%']) | |
| 3481 | + ); | |
| 3482 | + } else { | |
| 3483 | + $prompts_query = $wpdb->prepare( | |
| 3484 | + "SELECT id, article_content, source_url, timestamp, role_restriction | |
| 3485 | + FROM {$table_name} | |
| 3486 | + WHERE source_url IN ($placeholders) | |
| 3487 | + ORDER BY timestamp DESC", | |
| 3488 | + $url_list | |
| 3489 | + ); | |
| 3490 | + } | |
| 3491 | + $prompts = $wpdb->get_results($prompts_query); | |
| 3492 | + } | |
| 3493 | + | |
| 3494 | + // Group prompts by source_url for chunk display | |
| 3495 | + $grouped_prompts = array(); | |
| 3496 | + foreach ($prompts as $prompt) { | |
| 3497 | + $source_url = $prompt->source_url ?? ''; | |
| 3498 | + | |
| 3499 | + // Parse chunk metadata using the proper chunker method (same as initial page load) | |
| 3500 | + if (class_exists('MxChat_Chunker')) { | |
| 3501 | + $chunk_meta = MxChat_Chunker::parse_stored_chunk($prompt->article_content); | |
| 3502 | + $prompt->chunk_metadata = $chunk_meta['metadata']; | |
| 3503 | + $prompt->display_content = $chunk_meta['text']; | |
| 3504 | + } else { | |
| 3505 | + $prompt->chunk_metadata = array(); | |
| 3506 | + $prompt->display_content = $prompt->article_content; | |
| 3507 | + } | |
| 3508 | + | |
| 3509 | + if (!empty($source_url) && strpos($source_url, 'mxchat://') !== 0) { | |
| 3510 | + if (!isset($grouped_prompts[$source_url])) { | |
| 3511 | + $grouped_prompts[$source_url] = array(); | |
| 3512 | + } | |
| 3513 | + $grouped_prompts[$source_url][] = $prompt; | |
| 3514 | + } else { | |
| 3515 | + // Ungrouped entries | |
| 3516 | + $grouped_prompts['_ungrouped_' . $prompt->id] = array($prompt); | |
| 3517 | + } | |
| 3518 | + } | |
| 3519 | + | |
| 3520 | + // Sort groups by the original URL order (newest first) | |
| 3521 | + uksort($grouped_prompts, function($a, $b) use ($url_order_map) { | |
| 3522 | + $order_a = isset($url_order_map[$a]) ? $url_order_map[$a] : PHP_INT_MAX; | |
| 3523 | + $order_b = isset($url_order_map[$b]) ? $url_order_map[$b] : PHP_INT_MAX; | |
| 3524 | + return $order_a - $order_b; | |
| 3525 | + }); | |
| 3526 | + | |
| 3527 | + // Sort each group internally by chunk_index | |
| 3528 | + foreach ($grouped_prompts as $source_url => &$group) { | |
| 3529 | + usort($group, function($a, $b) { | |
| 3530 | + $index_a = isset($a->chunk_metadata['chunk_index']) ? intval($a->chunk_metadata['chunk_index']) : 0; | |
| 3531 | + $index_b = isset($b->chunk_metadata['chunk_index']) ? intval($b->chunk_metadata['chunk_index']) : 0; | |
| 3532 | + return $index_a - $index_b; | |
| 3533 | + }); | |
| 3534 | + } | |
| 3535 | + unset($group); | |
| 3536 | + | |
| 3537 | + // Build HTML for the table rows | |
| 3538 | + ob_start(); | |
| 3539 | + $display_index = 0; | |
| 3540 | + $current_page = $page; | |
| 3541 | + $data_source = 'wordpress'; | |
| 3542 | + $current_bot_id = $bot_id; | |
| 3543 | + $preview_length = 150; | |
| 3544 | + | |
| 3545 | + if (empty($grouped_prompts)) { | |
| 3546 | + echo '<tr><td colspan="5" style="padding: 40px; text-align: center; color: var(--mxch-text-muted);">'; | |
| 3547 | + esc_html_e('No knowledge entries found. Use the Import Options to add content.', 'mxchat'); | |
| 3548 | + echo '</td></tr>'; | |
| 3549 | + } else { | |
| 3550 | + foreach ($grouped_prompts as $source_url => $group) { | |
| 3551 | + $chunk_count = count($group); | |
| 3552 | + $first_prompt = $group[0]; | |
| 3553 | + $display_index++; | |
| 3554 | + | |
| 3555 | + if ($chunk_count > 1) { | |
| 3556 | + // Multiple chunks - show grouped row with expand button | |
| 3557 | + $group_id = 'group-' . md5($source_url); | |
| 3558 | + ?> | |
| 3559 | + <tr id="prompt-<?php echo esc_attr($first_prompt->id); ?>" | |
| 3560 | + class="mxchat-chunk-group-header" | |
| 3561 | + data-source="<?php echo esc_attr($data_source); ?>" | |
| 3562 | + data-group-id="<?php echo esc_attr($group_id); ?>" | |
| 3563 | + style="border-bottom: 1px solid var(--mxch-card-border);"> | |
| 3564 | + <td style="padding: 12px 16px; text-align: center;"> | |
| 3565 | + <input type="checkbox" | |
| 3566 | + class="mxchat-entry-checkbox" | |
| 3567 | + data-entry-id="<?php echo esc_attr($first_prompt->id); ?>" | |
| 3568 | + data-source="<?php echo esc_attr($data_source); ?>" | |
| 3569 | + data-source-url="<?php echo esc_attr($source_url); ?>" | |
| 3570 | + data-is-group="true" | |
| 3571 | + data-chunk-count="<?php echo esc_attr($chunk_count); ?>"> | |
| 3572 | + </td> | |
| 3573 | + <td style="padding: 12px 16px; font-size: 13px;"> | |
| 3574 | + <?php echo esc_html($first_prompt->id); ?> | |
| 3575 | + </td> | |
| 3576 | + <td class="mxchat-content-cell" style="padding: 12px 16px; font-size: 13px;"> | |
| 3577 | + <div class="mxchat-chunk-group-info"> | |
| 3578 | + <button type="button" class="mxchat-chunk-toggle" data-group-id="<?php echo esc_attr($group_id); ?>"> | |
| 3579 | + <span class="dashicons dashicons-arrow-right-alt2"></span> | |
| 3580 | + </button> | |
| 3581 | + <span class="mxchat-chunk-badge"><?php echo esc_html($chunk_count); ?> <?php esc_html_e('chunks', 'mxchat'); ?></span> | |
| 3582 | + <span class="mxchat-chunk-preview"> | |
| 3583 | + <?php | |
| 3584 | + $parent_content = isset($first_prompt->display_content) ? $first_prompt->display_content : $first_prompt->article_content; | |
| 3585 | + $content_preview = mb_substr($parent_content, 0, 100); | |
| 3586 | + echo esc_html($content_preview . '...'); | |
| 3587 | + ?> | |
| 3588 | + </span> | |
| 3589 | + </div> | |
| 3590 | + </td> | |
| 3591 | + <td class="mxchat-url-cell" style="padding: 12px 16px; font-size: 13px;"> | |
| 3592 | + <?php if (!empty($source_url) && strpos($source_url, 'mxchat://') !== 0 && strpos($source_url, '_ungrouped_') !== 0) : ?> | |
| 3593 | + <a href="<?php echo esc_url($source_url); ?>" target="_blank" style="color: var(--mxch-primary); text-decoration: none;"> | |
| 3594 | + <span class="dashicons dashicons-external" style="font-size: 14px;"></span> | |
| 3595 | + <?php esc_html_e('View Source', 'mxchat'); ?> | |
| 3596 | + </a> | |
| 3597 | + <?php else : ?> | |
| 3598 | + <span style="color: var(--mxch-text-muted);"><?php esc_html_e('Manual Content', 'mxchat'); ?></span> | |
| 3599 | + <?php endif; ?> | |
| 3600 | + </td> | |
| 3601 | + <td class="mxchat-actions-cell" style="padding: 12px 16px; white-space: nowrap;"> | |
| 3602 | + <?php if ($data_source !== 'pinecone') : ?> | |
| 3603 | + <button type="button" | |
| 3604 | + class="mxch-btn mxch-btn-ghost mxch-btn-sm mxchat-edit-entry-btn" | |
| 3605 | + data-source-url="<?php echo esc_attr($source_url); ?>" | |
| 3606 | + data-entry-id="<?php echo esc_attr($first_prompt->id); ?>" | |
| 3607 | + data-data-source="<?php echo esc_attr($data_source); ?>" | |
| 3608 | + data-bot-id="<?php echo esc_attr($current_bot_id); ?>" | |
| 3609 | + data-nonce="<?php echo wp_create_nonce('mxchat_edit_entry_nonce'); ?>" | |
| 3610 | + title="<?php esc_attr_e('Edit content', 'mxchat'); ?>"> | |
| 3611 | + <span class="dashicons dashicons-edit" style="font-size: 14px;"></span> | |
| 3612 | + </button> | |
| 3613 | + <?php endif; ?> | |
| 3614 | + <button type="button" | |
| 3615 | + class="mxch-btn mxch-btn-ghost mxch-btn-sm delete-button-group" | |
| 3616 | + data-source-url="<?php echo esc_attr($source_url); ?>" | |
| 3617 | + data-chunk-count="<?php echo esc_attr($chunk_count); ?>" | |
| 3618 | + data-data-source="<?php echo esc_attr($data_source); ?>" | |
| 3619 | + data-bot-id="<?php echo esc_attr($current_bot_id); ?>" | |
| 3620 | + data-nonce="<?php echo wp_create_nonce('mxchat_delete_chunks_nonce'); ?>" | |
| 3621 | + style="color: var(--mxch-error);" | |
| 3622 | + title="<?php esc_attr_e('Delete all chunks', 'mxchat'); ?>"> | |
| 3623 | + <span class="dashicons dashicons-trash" style="font-size: 14px;"></span> | |
| 3624 | + </button> | |
| 3625 | + </td> | |
| 3626 | + </tr> | |
| 3627 | + <?php | |
| 3628 | + // Render hidden chunk rows | |
| 3629 | + foreach ($group as $chunk_index => $chunk) { | |
| 3630 | + $meta_chunk_index = isset($chunk->chunk_metadata['chunk_index']) ? intval($chunk->chunk_metadata['chunk_index']) : $chunk_index; | |
| 3631 | + $meta_total_chunks = isset($chunk->chunk_metadata['total_chunks']) ? intval($chunk->chunk_metadata['total_chunks']) : $chunk_count; | |
| 3632 | + $content = isset($chunk->display_content) ? $chunk->display_content : $chunk->article_content; | |
| 3633 | + $content_preview = mb_strlen($content) > $preview_length | |
| 3634 | + ? mb_substr($content, 0, $preview_length) . '...' | |
| 3635 | + : $content; | |
| 3636 | + ?> | |
| 3637 | + <tr id="prompt-<?php echo esc_attr($chunk->id); ?>" | |
| 3638 | + class="mxchat-chunk-row <?php echo esc_attr($group_id); ?>" | |
| 3639 | + data-source="<?php echo esc_attr($data_source); ?>" | |
| 3640 | + style="display: none; background: #f8f9fa; border-bottom: 1px solid var(--mxch-card-border);"> | |
| 3641 | + <td style="padding: 12px 16px; text-align: center;"> | |
| 3642 | + <!-- Checkbox column placeholder for chunks (managed by group) --> | |
| 3643 | + </td> | |
| 3644 | + <td style="padding: 12px 16px 12px 30px; font-size: 13px;"> | |
| 3645 | + <!-- Hidden ID column for chunks --> | |
| 3646 | + </td> | |
| 3647 | + <td class="mxchat-content-cell" style="padding: 12px 16px; font-size: 13px;"> | |
| 3648 | + <div class="mxchat-accordion-wrapper"> | |
| 3649 | + <div class="mxchat-content-preview"> | |
| 3650 | + <span class="mxchat-chunk-indicator" style="margin-right: 10px; color: var(--mxch-text-secondary); font-size: 12px;"> | |
| 3651 | + <?php printf(esc_html__('Chunk %d of %d', 'mxchat'), $meta_chunk_index + 1, $meta_total_chunks); ?> | |
| 3652 | + </span> | |
| 3653 | + <span class="preview-text"><?php echo esc_html($content_preview); ?></span> | |
| 3654 | + <?php if (mb_strlen($content) > $preview_length) : ?> | |
| 3655 | + <button class="mxchat-expand-toggle" type="button"> | |
| 3656 | + <span class="dashicons dashicons-arrow-down-alt2"></span> | |
| 3657 | + </button> | |
| 3658 | + <?php endif; ?> | |
| 3659 | + </div> | |
| 3660 | + <div class="mxchat-content-full" style="display: none;"> | |
| 3661 | + <div class="content-view"> | |
| 3662 | + <?php | |
| 3663 | + if (preg_match('/[\x{0590}-\x{05FF}]/u', $content)) { | |
| 3664 | + echo '<div dir="rtl" lang="he" class="rtl-content">'; | |
| 3665 | + echo wp_kses_post(wpautop($content)); | |
| 3666 | + echo '</div>'; | |
| 3667 | + } else { | |
| 3668 | + echo wp_kses_post(wpautop($content)); | |
| 3669 | + } | |
| 3670 | + ?> | |
| 3671 | + </div> | |
| 3672 | + </div> | |
| 3673 | + </div> | |
| 3674 | + </td> | |
| 3675 | + <td class="mxchat-url-cell" style="padding: 12px 16px; font-size: 13px;"> | |
| 3676 | + <span class="mxchat-chunk-label" style="color: var(--mxch-text-muted);"><?php esc_html_e('Same as parent', 'mxchat'); ?></span> | |
| 3677 | + </td> | |
| 3678 | + <td class="mxchat-actions-cell" style="padding: 12px 16px;"> | |
| 3679 | + <span class="mxchat-chunk-label" style="color: var(--mxch-text-muted);"><?php esc_html_e('Managed by group', 'mxchat'); ?></span> | |
| 3680 | + </td> | |
| 3681 | + </tr> | |
| 3682 | + <?php | |
| 3683 | + } | |
| 3684 | + } else { | |
| 3685 | + // Single entry - display normally with accordion | |
| 3686 | + $prompt = $first_prompt; | |
| 3687 | + $content = isset($prompt->display_content) ? $prompt->display_content : $prompt->article_content; | |
| 3688 | + $content_preview = mb_strlen($content) > $preview_length | |
| 3689 | + ? mb_substr($content, 0, $preview_length) . '...' | |
| 3690 | + : $content; | |
| 3691 | + ?> | |
| 3692 | + <tr id="prompt-<?php echo esc_attr($prompt->id); ?>" | |
| 3693 | + data-source="<?php echo esc_attr($data_source); ?>" | |
| 3694 | + style="border-bottom: 1px solid var(--mxch-card-border);"> | |
| 3695 | + <td style="padding: 12px 16px; text-align: center;"> | |
| 3696 | + <input type="checkbox" | |
| 3697 | + class="mxchat-entry-checkbox" | |
| 3698 | + data-entry-id="<?php echo esc_attr($prompt->id); ?>" | |
| 3699 | + data-source="<?php echo esc_attr($data_source); ?>" | |
| 3700 | + data-source-url="<?php echo esc_attr($source_url); ?>" | |
| 3701 | + data-is-group="false"> | |
| 3702 | + </td> | |
| 3703 | + <td style="padding: 12px 16px; font-size: 13px;"> | |
| 3704 | + <?php echo esc_html($prompt->id); ?> | |
| 3705 | + </td> | |
| 3706 | + <td class="mxchat-content-cell" style="padding: 12px 16px; font-size: 13px;"> | |
| 3707 | + <div class="mxchat-accordion-wrapper"> | |
| 3708 | + <div class="mxchat-content-preview"> | |
| 3709 | + <span class="preview-text"><?php echo esc_html($content_preview); ?></span> | |
| 3710 | + <?php if (mb_strlen($content) > $preview_length) : ?> | |
| 3711 | + <button class="mxchat-expand-toggle" type="button"> | |
| 3712 | + <span class="dashicons dashicons-arrow-down-alt2"></span> | |
| 3713 | + </button> | |
| 3714 | + <?php endif; ?> | |
| 3715 | + </div> | |
| 3716 | + <div class="mxchat-content-full" style="display: none;"> | |
| 3717 | + <div class="content-view"> | |
| 3718 | + <?php | |
| 3719 | + if (preg_match('/[\x{0590}-\x{05FF}]/u', $content)) { | |
| 3720 | + echo '<div dir="rtl" lang="he" class="rtl-content">'; | |
| 3721 | + echo wp_kses_post(wpautop($content)); | |
| 3722 | + echo '</div>'; | |
| 3723 | + } else { | |
| 3724 | + echo wp_kses_post(wpautop($content)); | |
| 3725 | + } | |
| 3726 | + ?> | |
| 3727 | + </div> | |
| 3728 | + </div> | |
| 3729 | + </div> | |
| 3730 | + </td> | |
| 3731 | + <td class="mxchat-url-cell" style="padding: 12px 16px; font-size: 13px;"> | |
| 3732 | + <?php | |
| 3733 | + $actual_source = $source_url; | |
| 3734 | + if (strpos($source_url, '_ungrouped_') === 0) { | |
| 3735 | + $actual_source = $prompt->source_url ?? ''; | |
| 3736 | + } | |
| 3737 | + if (!empty($actual_source) && strpos($actual_source, 'mxchat://') !== 0) : ?> | |
| 3738 | + <a href="<?php echo esc_url($actual_source); ?>" target="_blank" style="color: var(--mxch-primary); text-decoration: none;"> | |
| 3739 | + <span class="dashicons dashicons-external" style="font-size: 14px;"></span> | |
| 3740 | + <?php esc_html_e('View', 'mxchat'); ?> | |
| 3741 | + </a> | |
| 3742 | + <?php else : ?> | |
| 3743 | + <span style="color: var(--mxch-text-muted);"><?php esc_html_e('Manual', 'mxchat'); ?></span> | |
| 3744 | + <?php endif; ?> | |
| 3745 | + </td> | |
| 3746 | + <td style="padding: 12px 16px; white-space: nowrap;"> | |
| 3747 | + <button type="button" | |
| 3748 | + class="mxch-btn mxch-btn-ghost mxch-btn-sm mxchat-edit-entry-btn" | |
| 3749 | + data-source-url="<?php echo esc_attr($prompt->source_url ?? ''); ?>" | |
| 3750 | + data-entry-id="<?php echo esc_attr($prompt->id); ?>" | |
| 3751 | + data-data-source="<?php echo esc_attr($data_source); ?>" | |
| 3752 | + data-bot-id="<?php echo esc_attr($current_bot_id); ?>" | |
| 3753 | + data-nonce="<?php echo wp_create_nonce('mxchat_edit_entry_nonce'); ?>" | |
| 3754 | + title="<?php esc_attr_e('Edit content', 'mxchat'); ?>"> | |
| 3755 | + <span class="dashicons dashicons-edit" style="font-size: 14px;"></span> | |
| 3756 | + </button> | |
| 3757 | + <button type="button" class="mxch-btn mxch-btn-ghost mxch-btn-sm delete-button-wordpress" data-entry-id="<?php echo esc_attr($prompt->id); ?>" data-bot-id="<?php echo esc_attr($current_bot_id); ?>" data-nonce="<?php echo wp_create_nonce('mxchat_delete_wordpress_prompt_nonce'); ?>" style="color: var(--mxch-error);"> | |
| 3758 | + <span class="dashicons dashicons-trash" style="font-size: 14px;"></span> | |
| 3759 | + </button> | |
| 3760 | + </td> | |
| 3761 | + </tr> | |
| 3762 | + <?php | |
| 3763 | + } | |
| 3764 | + } | |
| 3765 | + } | |
| 3766 | + $html = ob_get_clean(); | |
| 3767 | + | |
| 3768 | + // Generate pagination HTML (include search/filter data for subsequent pages) | |
| 3769 | + $pagination_html = ''; | |
| 3770 | + if ($total_pages > 1) { | |
| 3771 | + $pagination_html = '<div class="mxchat-ajax-pagination" data-current-page="' . esc_attr($page) . '" data-total-pages="' . esc_attr($total_pages) . '" data-search="' . esc_attr($search_query) . '" data-content-type="' . esc_attr($content_type_filter) . '">'; | |
| 3772 | + | |
| 3773 | + // Previous button | |
| 3774 | + if ($page > 1) { | |
| 3775 | + $pagination_html .= '<a href="#" class="mxchat-page-link" data-page="' . ($page - 1) . '">' . esc_html__('« Previous', 'mxchat') . '</a> '; | |
| 3776 | + } | |
| 3777 | + | |
| 3778 | + // Page numbers | |
| 3779 | + $start_page = max(1, $page - 2); | |
| 3780 | + $end_page = min($total_pages, $page + 2); | |
| 3781 | + | |
| 3782 | + if ($start_page > 1) { | |
| 3783 | + $pagination_html .= '<a href="#" class="mxchat-page-link" data-page="1">1</a> '; | |
| 3784 | + if ($start_page > 2) { | |
| 3785 | + $pagination_html .= '<span class="mxchat-page-dots">...</span> '; | |
| 3786 | + } | |
| 3787 | + } | |
| 3788 | + | |
| 3789 | + for ($i = $start_page; $i <= $end_page; $i++) { | |
| 3790 | + if ($i == $page) { | |
| 3791 | + $pagination_html .= '<span class="mxchat-page-current">' . $i . '</span> '; | |
| 3792 | + } else { | |
| 3793 | + $pagination_html .= '<a href="#" class="mxchat-page-link" data-page="' . $i . '">' . $i . '</a> '; | |
| 3794 | + } | |
| 3795 | + } | |
| 3796 | + | |
| 3797 | + if ($end_page < $total_pages) { | |
| 3798 | + if ($end_page < $total_pages - 1) { | |
| 3799 | + $pagination_html .= '<span class="mxchat-page-dots">...</span> '; | |
| 3800 | + } | |
| 3801 | + $pagination_html .= '<a href="#" class="mxchat-page-link" data-page="' . $total_pages . '">' . $total_pages . '</a> '; | |
| 3802 | + } | |
| 3803 | + | |
| 3804 | + // Next button | |
| 3805 | + if ($page < $total_pages) { | |
| 3806 | + $pagination_html .= '<a href="#" class="mxchat-page-link" data-page="' . ($page + 1) . '">' . esc_html__('Next »', 'mxchat') . '</a>'; | |
| 3807 | + } | |
| 3808 | + | |
| 3809 | + $pagination_html .= '</div>'; | |
| 3810 | + } | |
| 3811 | + | |
| 3812 | + wp_send_json_success(array( | |
| 3813 | + 'html' => $html, | |
| 3814 | + 'pagination_html' => $pagination_html, | |
| 3815 | + 'total_count' => $total_records, | |
| 3816 | + 'total_pages' => $total_pages, | |
| 3817 | + 'page' => $page, | |
| 3818 | + 'per_page' => $per_page, | |
| 3819 | + 'data_source' => 'wordpress' | |
| 3820 | + )); | |
| 3821 | +} | |
| 3822 | + | |
| 3823 | +/** | |
| 3824 | + * AJAX handler to detect available sitemaps on the site | |
| 3825 | + * Optimized for speed - only checks primary sitemap indexes first | |
| 3826 | + */ | |
| 3827 | +public function ajax_mxchat_detect_sitemaps() { | |
| 3828 | + check_ajax_referer('mxchat_detect_sitemaps_nonce', 'nonce'); | |
| 3829 | + | |
| 3830 | + if (!current_user_can('manage_options')) { | |
| 3831 | + wp_send_json_error(array('message' => 'Unauthorized')); | |
| 3832 | + return; | |
| 3833 | + } | |
| 3834 | + | |
| 3835 | + $site_url = get_site_url(); | |
| 3836 | + $sitemaps = array(); | |
| 3837 | + $found_index = false; | |
| 3838 | + | |
| 3839 | + // Only check the main sitemap index files first (much faster) | |
| 3840 | + // These are the primary entry points that contain sub-sitemaps | |
| 3841 | + $primary_indexes = array( | |
| 3842 | + 'sitemap_index.xml' => 'Yoast SEO / Rank Math', // Yoast & Rank Math | |
| 3843 | + 'wp-sitemap.xml' => 'WordPress Core', // WordPress Core | |
| 3844 | + 'sitemap.xml' => 'Standard', // Generic/AIOSEO | |
| 3845 | + ); | |
| 3846 | + | |
| 3847 | + foreach ($primary_indexes as $path => $source) { | |
| 3848 | + $url = trailingslashit($site_url) . $path; | |
| 3849 | + | |
| 3850 | + $response = wp_remote_head($url, array( | |
| 3851 | + 'timeout' => 10, | |
| 3852 | + 'sslverify' => false, | |
| 3853 | + 'redirection' => 1, | |
| 3854 | + 'user-agent' => mxchat_ingest_user_agent(), | |
| 3855 | + )); | |
| 3856 | + | |
| 3857 | + if (!is_wp_error($response) && wp_remote_retrieve_response_code($response) === 200) { | |
| 3858 | + // Found a sitemap index - parse it to get sub-sitemaps | |
| 3859 | + $sub_sitemaps = $this->parse_sitemap_index($url); | |
| 3860 | + if (!empty($sub_sitemaps)) { | |
| 3861 | + $sitemaps[] = array( | |
| 3862 | + 'url' => $url, | |
| 3863 | + 'type' => 'index', | |
| 3864 | + 'source' => $source, | |
| 3865 | + 'sub_sitemaps' => $sub_sitemaps | |
| 3866 | + ); | |
| 3867 | + $found_index = true; | |
| 3868 | + // Found a valid index, no need to check others | |
| 3869 | + break; | |
| 3870 | + } | |
| 3871 | + } | |
| 3872 | + } | |
| 3873 | + | |
| 3874 | + // If no sitemap index found, check for standalone sitemaps | |
| 3875 | + if (!$found_index) { | |
| 3876 | + $standalone_sitemaps = array( | |
| 3877 | + 'post-sitemap.xml' => array('type' => 'content', 'source' => 'Yoast SEO'), | |
| 3878 | + 'page-sitemap.xml' => array('type' => 'content', 'source' => 'Yoast SEO'), | |
| 3879 | + ); | |
| 3880 | + | |
| 3881 | + foreach ($standalone_sitemaps as $path => $info) { | |
| 3882 | + $url = trailingslashit($site_url) . $path; | |
| 3883 | + | |
| 3884 | + $response = wp_remote_head($url, array( | |
| 3885 | + 'timeout' => 2, | |
| 3886 | + 'sslverify' => false | |
| 3887 | + )); | |
| 3888 | + | |
| 3889 | + if (!is_wp_error($response) && wp_remote_retrieve_response_code($response) === 200) { | |
| 3890 | + $sitemaps[] = array( | |
| 3891 | + 'url' => $url, | |
| 3892 | + 'type' => $info['type'], | |
| 3893 | + 'source' => $info['source'], | |
| 3894 | + 'url_count' => 0 // Skip URL count for speed | |
| 3895 | + ); | |
| 3896 | + } | |
| 3897 | + } | |
| 3898 | + } | |
| 3899 | + | |
| 3900 | + wp_send_json_success(array( | |
| 3901 | + 'sitemaps' => $sitemaps, | |
| 3902 | + 'site_url' => $site_url | |
| 3903 | + )); | |
| 3904 | +} | |
| 3905 | + | |
| 3906 | +/** | |
| 3907 | + * Parse a sitemap index to get sub-sitemaps | |
| 3908 | + * Optimized: doesn't fetch URL count for each sub-sitemap (too slow) | |
| 3909 | + */ | |
| 3910 | +private function parse_sitemap_index($url) { | |
| 3911 | + $sub_sitemaps = array(); | |
| 3912 | + | |
| 3913 | + $response = wp_remote_get($url, array( | |
| 3914 | + 'timeout' => 30, | |
| 3915 | + 'sslverify' => false, | |
| 3916 | + 'user-agent' => mxchat_ingest_user_agent(), | |
| 3917 | + 'headers' => array( | |
| 3918 | + 'Accept' => 'text/html,application/xhtml+xml,application/xml;q=0.9,*/*;q=0.8', | |
| 3919 | + ), | |
| 3920 | + )); | |
| 3921 | + | |
| 3922 | + if (is_wp_error($response)) { | |
| 3923 | + return $sub_sitemaps; | |
| 3924 | + } | |
| 3925 | + | |
| 3926 | + $body = wp_remote_retrieve_body($response); | |
| 3927 | + if (empty($body)) { | |
| 3928 | + return $sub_sitemaps; | |
| 3929 | + } | |
| 3930 | + | |
| 3931 | + // Suppress XML errors | |
| 3932 | + libxml_use_internal_errors(true); | |
| 3933 | + $xml = simplexml_load_string($body); | |
| 3934 | + libxml_clear_errors(); | |
| 3935 | + | |
| 3936 | + if ($xml === false) { | |
| 3937 | + return $sub_sitemaps; | |
| 3938 | + } | |
| 3939 | + | |
| 3940 | + // Check if it's a sitemap index (contains <sitemap> elements) | |
| 3941 | + if (isset($xml->sitemap)) { | |
| 3942 | + foreach ($xml->sitemap as $sitemap) { | |
| 3943 | + $loc = (string) $sitemap->loc; | |
| 3944 | + if (!empty($loc)) { | |
| 3945 | + // Try to determine the type from the URL | |
| 3946 | + $type = 'content'; | |
| 3947 | + if (strpos($loc, 'category') !== false || strpos($loc, 'tag') !== false || strpos($loc, 'taxonomy') !== false) { | |
| 3948 | + $type = 'taxonomy'; | |
| 3949 | + } elseif (strpos($loc, 'author') !== false || strpos($loc, 'user') !== false) { | |
| 3950 | + $type = 'author'; | |
| 3951 | + } | |
| 3952 | + | |
| 3953 | + // Skip URL count - too slow to fetch for each sitemap | |
| 3954 | + $sub_sitemaps[] = array( | |
| 3955 | + 'url' => $loc, | |
| 3956 | + 'type' => $type, | |
| 3957 | + 'url_count' => 0, // Don't fetch - takes too long | |
| 3958 | + 'name' => basename(parse_url($loc, PHP_URL_PATH)) | |
| 3959 | + ); | |
| 3960 | + } | |
| 3961 | + } | |
| 3962 | + } | |
| 3963 | + | |
| 3964 | + return $sub_sitemaps; | |
| 3965 | +} | |
| 3966 | + | |
| 3967 | +/** | |
| 3968 | + * Get URL count from a sitemap | |
| 3969 | + */ | |
| 3970 | +private function get_sitemap_url_count($url) { | |
| 3971 | + $response = wp_remote_get($url, array( | |
| 3972 | + 'timeout' => 30, | |
| 3973 | + 'sslverify' => false, | |
| 3974 | + 'user-agent' => mxchat_ingest_user_agent(), | |
| 3975 | + 'headers' => array( | |
| 3976 | + 'Accept' => 'text/html,application/xhtml+xml,application/xml;q=0.9,*/*;q=0.8', | |
| 3977 | + ), | |
| 3978 | + )); | |
| 3979 | + | |
| 3980 | + if (is_wp_error($response)) { | |
| 3981 | + return 0; | |
| 3982 | + } | |
| 3983 | + | |
| 3984 | + $body = wp_remote_retrieve_body($response); | |
| 3985 | + if (empty($body)) { | |
| 3986 | + return 0; | |
| 3987 | + } | |
| 3988 | + | |
| 3989 | + // Count <url> or <loc> elements | |
| 3990 | + $count = preg_match_all('/<url>/i', $body, $matches); | |
| 3991 | + return $count ?: 0; | |
| 3992 | +} | |
| 3993 | + | |
| 3994 | +/** | |
| 3995 | + * Get sitemaps declared in robots.txt | |
| 3996 | + */ | |
| 3997 | +private function get_sitemaps_from_robots($site_url) { | |
| 3998 | + $sitemaps = array(); | |
| 3999 | + $robots_url = trailingslashit($site_url) . 'robots.txt'; | |
| 4000 | + | |
| 4001 | + $response = wp_remote_get($robots_url, array( | |
| 4002 | + 'timeout' => 15, | |
| 4003 | + 'sslverify' => false, | |
| 4004 | + 'user-agent' => mxchat_ingest_user_agent(), | |
| 4005 | + )); | |
| 4006 | + | |
| 4007 | + if (is_wp_error($response)) { | |
| 4008 | + return $sitemaps; | |
| 4009 | + } | |
| 4010 | + | |
| 4011 | + $body = wp_remote_retrieve_body($response); | |
| 4012 | + if (empty($body)) { | |
| 4013 | + return $sitemaps; | |
| 4014 | + } | |
| 4015 | + | |
| 4016 | + // Find Sitemap: declarations | |
| 4017 | + if (preg_match_all('/^Sitemap:\s*(.+)$/mi', $body, $matches)) { | |
| 4018 | + foreach ($matches[1] as $sitemap_url) { | |
| 4019 | + $sitemap_url = trim($sitemap_url); | |
| 4020 | + if (filter_var($sitemap_url, FILTER_VALIDATE_URL)) { | |
| 4021 | + $sitemaps[] = $sitemap_url; | |
| 4022 | + } | |
| 4023 | + } | |
| 4024 | + } | |
| 4025 | + | |
| 4026 | + return $sitemaps; | |
| 4027 | +} | |
| 4028 | + | |
| 4029 | +public function mxchat_stop_processing() { | |
| 4030 | + // Verify permissions | |
| 4031 | + if (!current_user_can('manage_options')) { | |
| 4032 | + wp_die(esc_html__('Unauthorized access', 'mxchat')); | |
| 4033 | + } | |
| 4034 | + | |
| 4035 | + // Verify nonce | |
| 4036 | + check_admin_referer('mxchat_stop_processing_action', 'mxchat_stop_processing_nonce'); | |
| 4037 | + | |
| 4038 | + global $wpdb; | |
| 4039 | + $table_name = $wpdb->prefix . 'mxchat_processing_queue'; | |
| 4040 | + | |
| 4041 | + // Get active queue IDs | |
| 4042 | + $sitemap_queue_id = get_transient('mxchat_active_queue_sitemap'); | |
| 4043 | + $pdf_queue_id = get_transient('mxchat_active_queue_pdf'); | |
| 4044 | + | |
| 4045 | + // Delete all pending items from active queues | |
| 4046 | + if ($sitemap_queue_id) { | |
| 4047 | + $wpdb->delete( | |
| 4048 | + $table_name, | |
| 4049 | + array( | |
| 4050 | + 'queue_id' => $sitemap_queue_id, | |
| 4051 | + 'status' => 'pending' | |
| 4052 | + ), | |
| 4053 | + array('%s', '%s') | |
| 4054 | + ); | |
| 4055 | + | |
| 4056 | + delete_transient('mxchat_active_queue_sitemap'); | |
| 4057 | + delete_transient('mxchat_last_sitemap_url'); | |
| 4058 | + } | |
| 4059 | + | |
| 4060 | + if ($pdf_queue_id) { | |
| 4061 | + // Get PDF path before deleting | |
| 4062 | + $pdf_path = $this->mxchat_get_queue_meta($pdf_queue_id, 'pdf_path'); | |
| 4063 | + | |
| 4064 | + $wpdb->delete( | |
| 4065 | + $table_name, | |
| 4066 | + array( | |
| 4067 | + 'queue_id' => $pdf_queue_id, | |
| 4068 | + 'status' => 'pending' | |
| 4069 | + ), | |
| 4070 | + array('%s', '%s') | |
| 4071 | + ); | |
| 4072 | + | |
| 4073 | + // Delete PDF file | |
| 4074 | + if ($pdf_path && file_exists($pdf_path)) { | |
| 4075 | + wp_delete_file($pdf_path); | |
| 4076 | + } | |
| 4077 | + | |
| 4078 | + delete_transient('mxchat_active_queue_pdf'); | |
| 4079 | + delete_transient('mxchat_last_pdf_url'); | |
| 4080 | + } | |
| 4081 | + | |
| 4082 | + // Redirect back with a success message | |
| 4083 | + set_transient('mxchat_admin_notice_success', | |
| 4084 | + esc_html__('Processing has been stopped successfully.', 'mxchat'), | |
| 4085 | + 30 | |
| 4086 | + ); | |
| 4087 | + wp_safe_redirect(admin_url('admin.php?page=mxchat-prompts')); | |
| 4088 | + exit; | |
| 4089 | +} | |
| 4090 | + | |
| 4091 | +/** | |
| 4092 | + * Get content list for processing | |
| 4093 | + */ | |
| 4094 | +public function ajax_mxchat_get_content_list() { | |
| 4095 | + // Verify the nonce | |
| 4096 | + check_ajax_referer('mxchat_content_selector_nonce', 'nonce'); | |
| 4097 | + | |
| 4098 | + if (!current_user_can('manage_options')) { | |
| 4099 | + wp_send_json_error(__('Unauthorized access', 'mxchat')); | |
| 4100 | + } | |
| 4101 | + | |
| 4102 | + $page = isset($_GET['page']) ? absint($_GET['page']) : 1; | |
| 4103 | + $per_page = isset($_GET['per_page']) ? absint($_GET['per_page']) : 100; | |
| 4104 | + $search = isset($_GET['search']) ? sanitize_text_field($_GET['search']) : ''; | |
| 4105 | + $post_type = isset($_GET['post_type']) ? sanitize_text_field($_GET['post_type']) : 'all'; | |
| 4106 | + $post_status = isset($_GET['post_status']) ? sanitize_text_field($_GET['post_status']) : 'publish'; | |
| 4107 | + $processed_filter = isset($_GET['processed_filter']) ? sanitize_text_field($_GET['processed_filter']) : 'all'; | |
| 4108 | + | |
| 4109 | + // Build query args | |
| 4110 | + $args = array( | |
| 4111 | + 'posts_per_page' => $per_page, | |
| 4112 | + 'paged' => $page, | |
| 4113 | + 'post_status' => $post_status !== 'all' ? $post_status : array('publish', 'draft', 'pending'), | |
| 4114 | + 'orderby' => 'date', | |
| 4115 | + 'order' => 'DESC', | |
| 4116 | + ); | |
| 4117 | + | |
| 4118 | + // Supported media types, and the one-shot WHERE filter that scopes the MIME gate | |
| 4119 | + // to attachment rows when this query spans several post types (plan 01010a). | |
| 4120 | + $attachment_mimes = self::mxchat_supported_attachment_mime_types(); | |
| 4121 | + $attachment_where = null; | |
| 4122 | + | |
| 4123 | + // Handle post types - IMPROVED VERSION | |
| 4124 | + if ($post_type !== 'all') { | |
| 4125 | + $args['post_type'] = $post_type; | |
| 4126 | + | |
| 4127 | + // Media is offered by the dropdown above (attachment is a public post type) | |
| 4128 | + // but could never return a row: attachments are stored with post_status | |
| 4129 | + // 'inherit' and this query asks for publish/draft/pending, so the list was | |
| 4130 | + // empty by construction. Force the status attachments actually use, and | |
| 4131 | + // offer only the file types we can extract text from (plan 01010a). | |
| 4132 | + if ($post_type === 'attachment') { | |
| 4133 | + $args['post_status'] = 'inherit'; | |
| 4134 | + $args['post_mime_type'] = $attachment_mimes; | |
| 4135 | + } | |
| 4136 | + } else { | |
| 4137 | + // Get all available post types that might contain content | |
| 4138 | + $all_post_types = array(); | |
| 4139 | + | |
| 4140 | + // First get all public post types | |
| 4141 | + $public_types = get_post_types(array('public' => true), 'names'); | |
| 4142 | + $all_post_types = array_merge($all_post_types, $public_types); | |
| 4143 | + | |
| 4144 | + // Add common forum/community post types | |
| 4145 | + $forum_types = array('topic', 'reply', 'forum', 'wpforo_topic', 'wpforo_post'); | |
| 4146 | + foreach ($forum_types as $forum_type) { | |
| 4147 | + if (post_type_exists($forum_type)) { | |
| 4148 | + $all_post_types[] = $forum_type; | |
| 4149 | + } | |
| 4150 | + } | |
| 4151 | + | |
| 4152 | + // Add other commonly used post types | |
| 4153 | + $common_types = array('product', 'job_listing', 'event', 'portfolio'); | |
| 4154 | + foreach ($common_types as $common_type) { | |
| 4155 | + if (post_type_exists($common_type)) { | |
| 4156 | + $all_post_types[] = $common_type; | |
| 4157 | + } | |
| 4158 | + } | |
| 4159 | + | |
| 4160 | + // Remove duplicates and ensure we have at least some post types | |
| 4161 | + $all_post_types = array_unique($all_post_types); | |
| 4162 | + | |
| 4163 | + if (empty($all_post_types)) { | |
| 4164 | + // Fallback to basic post types | |
| 4165 | + $all_post_types = array('post', 'page'); | |
| 4166 | + } | |
| 4167 | + | |
| 4168 | + // Attachments in the "All Content Types" view (plan 01010a). Two things have | |
| 4169 | + // to happen at once and neither can ride $args on its own: | |
| 4170 | + // | |
| 4171 | + // - 'inherit' has to join the status list, because WP_Query takes ONE status | |
| 4172 | + // set for the whole query. Only attachments and revisions use it, and | |
| 4173 | + // revisions are not in $all_post_types, so nothing else widens. | |
| 4174 | + // - the MIME gate CANNOT be $args['post_mime_type'] here — that clause is | |
| 4175 | + // applied to every row, and an ordinary post has an empty post_mime_type, | |
| 4176 | + // so it would empty the entire listing rather than filter the media half. | |
| 4177 | + // A scoped WHERE is the only shape that filters attachments alone. | |
| 4178 | + // | |
| 4179 | + // Media joins this view under the default Published filter and under All | |
| 4180 | + // Statuses; picking Drafts deliberately drops it, because a file has no draft | |
| 4181 | + // state and answering a draft filter with published media is a wrong answer. | |
| 4182 | + $wants_attachments = in_array('attachment', $all_post_types, true) | |
| 4183 | + && ($post_status === 'publish' || $post_status === 'all'); | |
| 4184 | + | |
| 4185 | + if ($wants_attachments) { | |
| 4186 | + $statuses = array_values(array_unique(array_merge((array) $args['post_status'], array('inherit')))); | |
| 4187 | + $args['post_status'] = $statuses; | |
| 4188 | + | |
| 4189 | + global $wpdb; | |
| 4190 | + $mime_placeholders = implode(', ', array_fill(0, count($attachment_mimes), '%s')); | |
| 4191 | + $mime_clause = $wpdb->prepare( | |
| 4192 | + " AND ({$wpdb->posts}.post_type <> 'attachment' OR {$wpdb->posts}.post_mime_type IN ($mime_placeholders))", | |
| 4193 | + $attachment_mimes | |
| 4194 | + ); | |
| 4195 | + $attachment_where = function ($where) use ($mime_clause) { | |
| 4196 | + return $where . $mime_clause; | |
| 4197 | + }; | |
| 4198 | + add_filter('posts_where', $attachment_where); | |
| 4199 | + } else { | |
| 4200 | + $all_post_types = array_values(array_diff($all_post_types, array('attachment'))); | |
| 4201 | + } | |
| 4202 | + | |
| 4203 | + $args['post_type'] = $all_post_types; | |
| 4204 | + | |
| 4205 | + // Debug logging to see what post types are being queried | |
| 4206 | + //error_log('MxChat Debug: Querying post types: ' . implode(', ', $all_post_types)); | |
| 4207 | + } | |
| 4208 | + | |
| 4209 | + if (!empty($search)) { | |
| 4210 | + $args['s'] = $search; | |
| 4211 | + } | |
| 4212 | + | |
| 4213 | + // Get processed data from storage | |
| 4214 | + $processed_data = array(); | |
| 4215 | + | |
| 4216 | + $pinecone_options = get_option('mxchat_pinecone_addon_options', array()); | |
| 4217 | + $use_pinecone = ($pinecone_options['mxchat_use_pinecone'] ?? '0') === '1'; | |
| 4218 | + | |
| 4219 | + if ($use_pinecone && !empty($pinecone_options['mxchat_pinecone_api_key'])) { | |
| 4220 | + // Get fresh data from Pinecone - no caching | |
| 4221 | + $processed_data = $this->mxchat_get_pinecone_processed_content($pinecone_options); | |
| 4222 | + } else { | |
| 4223 | + // WordPress DB checking with better URL matching for all post types | |
| 4224 | + global $wpdb; | |
| 4225 | + $table_name = $wpdb->prefix . 'mxchat_system_prompt_content'; | |
| 4226 | + $processed_items = $wpdb->get_results("SELECT id, source_url, article_content, timestamp FROM {$table_name}"); | |
| 4227 | + | |
| 4228 | + // Group items by source_url to count chunks | |
| 4229 | + $url_chunk_counts = array(); | |
| 4230 | + $url_latest_timestamp = array(); | |
| 4231 | + $url_first_id = array(); | |
| 4232 | + | |
| 4233 | + if (!empty($processed_items)) { | |
| 4234 | + foreach ($processed_items as $item) { | |
| 4235 | + $url = $item->source_url; | |
| 4236 | + if (empty($url)) continue; | |
| 4237 | + | |
| 4238 | + // Count chunks per URL | |
| 4239 | + if (!isset($url_chunk_counts[$url])) { | |
| 4240 | + $url_chunk_counts[$url] = 0; | |
| 4241 | + $url_latest_timestamp[$url] = $item->timestamp; | |
| 4242 | + $url_first_id[$url] = $item->id; | |
| 4243 | + } | |
| 4244 | + $url_chunk_counts[$url]++; | |
| 4245 | + | |
| 4246 | + // Track latest timestamp | |
| 4247 | + if (strtotime($item->timestamp) > strtotime($url_latest_timestamp[$url])) { | |
| 4248 | + $url_latest_timestamp[$url] = $item->timestamp; | |
| 4249 | + } | |
| 4250 | + } | |
| 4251 | + | |
| 4252 | + // Now build processed_data with chunk counts | |
| 4253 | + foreach ($url_chunk_counts as $url => $chunk_count) { | |
| 4254 | + $post_id = $this->mxchat_url_to_post_id_improved($url); | |
| 4255 | + | |
| 4256 | + if ($post_id) { | |
| 4257 | + $processed_data[$post_id] = array( | |
| 4258 | + 'db_id' => $url_first_id[$url], | |
| 4259 | + 'timestamp' => $url_latest_timestamp[$url], | |
| 4260 | + 'url' => $url, | |
| 4261 | + 'source' => 'wordpress', | |
| 4262 | + 'chunk_count' => $chunk_count | |
| 4263 | + ); | |
| 4264 | + } | |
| 4265 | + } | |
| 4266 | + } | |
| 4267 | + } | |
| 4268 | + | |
| 4269 | + // Get processed IDs as a simple array for in_array checks | |
| 4270 | + $processed_ids = array_keys($processed_data); | |
| 4271 | + | |
| 4272 | + // Handle processed/unprocessed filter | |
| 4273 | + if ($processed_filter === 'processed' && !empty($processed_ids)) { | |
| 4274 | + $args['post__in'] = $processed_ids; | |
| 4275 | + } elseif ($processed_filter === 'unprocessed' && !empty($processed_ids)) { | |
| 4276 | + $args['post__not_in'] = $processed_ids; | |
| 4277 | + } | |
| 4278 | + | |
| 4279 | + // Run the query | |
| 4280 | + $query = new WP_Query($args); | |
| 4281 | + | |
| 4282 | + // Scoped to this one query — leaving it attached would MIME-filter every later | |
| 4283 | + // WP_Query in the request. | |
| 4284 | + if ($attachment_where) { | |
| 4285 | + remove_filter('posts_where', $attachment_where); | |
| 4286 | + } | |
| 4287 | + | |
| 4288 | + $content_items = array(); | |
| 4289 | + | |
| 4290 | + if ($query->have_posts()) { | |
| 4291 | + while ($query->have_posts()) { | |
| 4292 | + $query->the_post(); | |
| 4293 | + $id = get_the_ID(); | |
| 4294 | + $post_date = get_the_date(); | |
| 4295 | + $excerpt = wp_trim_words(get_the_excerpt(), 20, '...'); | |
| 4296 | + $word_count = str_word_count(strip_tags(get_the_content())); | |
| 4297 | + | |
| 4298 | + $is_processed = in_array($id, $processed_ids); | |
| 4299 | + $processed_date = ''; | |
| 4300 | + $db_record_id = 0; | |
| 4301 | + $data_source = 'none'; | |
| 4302 | + | |
| 4303 | + if ($is_processed && isset($processed_data[$id])) { | |
| 4304 | + $item_data = $processed_data[$id]; | |
| 4305 | + $data_source = $item_data['source']; | |
| 4306 | + | |
| 4307 | + if ($data_source === 'wordpress' && isset($item_data['timestamp'])) { | |
| 4308 | + // WordPress DB format | |
| 4309 | + $timestamp = strtotime($item_data['timestamp']); | |
| 4310 | + $processed_date = human_time_diff($timestamp, current_time('timestamp')) . ' ago'; | |
| 4311 | + $db_record_id = $item_data['db_id']; | |
| 4312 | + } elseif ($data_source === 'pinecone') { | |
| 4313 | + // Pinecone format | |
| 4314 | + $processed_date = $item_data['processed_date']; | |
| 4315 | + $db_record_id = $item_data['db_id']; | |
| 4316 | + } | |
| 4317 | + } | |
| 4318 | + | |
| 4319 | + // Get chunk count for this item | |
| 4320 | + $chunk_count = 0; | |
| 4321 | + if ($is_processed && isset($processed_data[$id]['chunk_count'])) { | |
| 4322 | + $chunk_count = intval($processed_data[$id]['chunk_count']); | |
| 4323 | + } | |
| 4324 | + | |
| 4325 | + // Media rows describe themselves differently (plan 01010a): the link has | |
| 4326 | + // to be the FILE, not the attachment page nobody wants; "attachment" and | |
| 4327 | + // "0 words" tell the user nothing, so the row carries the file type and | |
| 4328 | + // its size instead. Word count is deliberately NOT computed — that would | |
| 4329 | + // mean parsing every PDF in the library just to render a list. | |
| 4330 | + $permalink = get_permalink(); | |
| 4331 | + $type_label = get_post_type(); | |
| 4332 | + $size_label = ''; | |
| 4333 | + | |
| 4334 | + if (get_post_type() === 'attachment') { | |
| 4335 | + $file_url = wp_get_attachment_url($id); | |
| 4336 | + $permalink = $file_url ? $file_url : $permalink; | |
| 4337 | + $type_label = self::mxchat_attachment_type_label(get_post_mime_type($id)); | |
| 4338 | + | |
| 4339 | + $file_path = get_attached_file($id); | |
| 4340 | + $file_size = ($file_path && file_exists($file_path)) ? @filesize($file_path) : false; | |
| 4341 | + if ($file_size !== false) { | |
| 4342 | + $size_label = size_format($file_size, 0); | |
| 4343 | + } | |
| 4344 | + | |
| 4345 | + if ($excerpt === '') { | |
| 4346 | + $excerpt = $file_url ? wp_basename($file_url) : ''; | |
| 4347 | + } | |
| 4348 | + } | |
| 4349 | + | |
| 4350 | + $content_items[] = array( | |
| 4351 | + 'id' => $id, | |
| 4352 | + 'title' => get_the_title(), | |
| 4353 | + 'permalink' => $permalink, | |
| 4354 | + 'date' => $post_date, | |
| 4355 | + 'type' => get_post_type(), | |
| 4356 | + 'type_label' => $type_label, | |
| 4357 | + 'size_label' => $size_label, | |
| 4358 | + 'status' => get_post_status(), | |
| 4359 | + 'excerpt' => $excerpt, | |
| 4360 | + 'word_count' => $word_count, | |
| 4361 | + 'already_processed' => $is_processed, | |
| 4362 | + 'processed_date' => $processed_date, | |
| 4363 | + 'db_record_id' => $db_record_id, | |
| 4364 | + 'data_source' => $data_source, | |
| 4365 | + 'chunk_count' => $chunk_count | |
| 4366 | + ); | |
| 4367 | + } | |
| 4368 | + wp_reset_postdata(); | |
| 4369 | + } | |
| 4370 | + | |
| 4371 | + $response = array( | |
| 4372 | + 'items' => $content_items, | |
| 4373 | + 'total' => $query->found_posts, | |
| 4374 | + 'total_pages' => $query->max_num_pages, | |
| 4375 | + 'current_page' => $page, | |
| 4376 | + 'processed_count' => count($processed_ids) | |
| 4377 | + ); | |
| 4378 | + | |
| 4379 | + wp_send_json_success($response); | |
| 4380 | + exit; | |
| 4381 | +} | |
| 4382 | + | |
| 4383 | + | |
| 4384 | +/** | |
| 4385 | + * This function handles various WooCommerce URL formats and permalink structures | |
| 4386 | + */ | |
| 4387 | +private function mxchat_url_to_post_id_improved($url) { | |
| 4388 | + // First try the standard WordPress function | |
| 4389 | + $post_id = url_to_postid($url); | |
| 4390 | + | |
| 4391 | + if ($post_id > 0) { | |
| 4392 | + return $post_id; | |
| 4393 | + } | |
| 4394 | + | |
| 4395 | + // Media entries are stored under the FILE url (…/wp-content/uploads/…), which is | |
| 4396 | + // not a permalink — url_to_postid() never resolves one, and every slug fallback | |
| 4397 | + // below excludes 'attachment' explicitly. Without this, an imported media file | |
| 4398 | + // renders as "Not In Knowledge Base" forever and re-imports as new every time | |
| 4399 | + // (plan 01010a). attachment_url_to_postid() matches on _wp_attached_file and | |
| 4400 | + // returns 0 for anything outside the uploads directory, so post permalinks are | |
| 4401 | + // unaffected. | |
| 4402 | + $attachment_id = (int) attachment_url_to_postid($url); | |
| 4403 | + if ($attachment_id > 0) { | |
| 4404 | + return $attachment_id; | |
| 4405 | + } | |
| 4406 | + | |
| 4407 | + // If that fails, try more aggressive URL matching | |
| 4408 | + // Remove trailing slashes and query parameters for better matching | |
| 4409 | + $clean_url = rtrim($url, '/'); | |
| 4410 | + $clean_url = strtok($clean_url, '?'); // Remove query parameters | |
| 4411 | + | |
| 4412 | + // Try again with cleaned URL | |
| 4413 | + $post_id = url_to_postid($clean_url); | |
| 4414 | + if ($post_id > 0) { | |
| 4415 | + return $post_id; | |
| 4416 | + } | |
| 4417 | + | |
| 4418 | + // For bbPress forum topics, try extracting slug from URL | |
| 4419 | + if (strpos($url, '/topic/') !== false || strpos($url, '/forums/') !== false) { | |
| 4420 | + // Handle bbPress URLs: /forums/topic/topic-name/ | |
| 4421 | + if (preg_match('/\/forums\/topic\/([^\/\?]+)/', $url, $matches)) { | |
| 4422 | + $topic_slug = $matches[1]; | |
| 4423 | + | |
| 4424 | + // Look up topic by slug | |
| 4425 | + $topic = get_page_by_path($topic_slug, OBJECT, 'topic'); | |
| 4426 | + if ($topic) { | |
| 4427 | + return $topic->ID; | |
| 4428 | + } | |
| 4429 | + | |
| 4430 | + // Alternative method: query by post_name | |
| 4431 | + global $wpdb; | |
| 4432 | + $post_id = $wpdb->get_var($wpdb->prepare( | |
| 4433 | + "SELECT ID FROM {$wpdb->posts} WHERE post_name = %s AND post_type = 'topic' AND post_status IN ('publish', 'closed')", | |
| 4434 | + $topic_slug | |
| 4435 | + )); | |
| 4436 | + | |
| 4437 | + if ($post_id) { | |
| 4438 | + return intval($post_id); | |
| 4439 | + } | |
| 4440 | + } | |
| 4441 | + | |
| 4442 | + // Handle simpler topic URLs: /topic/topic-name/ | |
| 4443 | + if (preg_match('/\/topic\/([^\/\?]+)/', $url, $matches)) { | |
| 4444 | + $topic_slug = $matches[1]; | |
| 4445 | + | |
| 4446 | + global $wpdb; | |
| 4447 | + $post_id = $wpdb->get_var($wpdb->prepare( | |
| 4448 | + "SELECT ID FROM {$wpdb->posts} WHERE post_name = %s AND post_type = 'topic' AND post_status IN ('publish', 'closed')", | |
| 4449 | + $topic_slug | |
| 4450 | + )); | |
| 4451 | + | |
| 4452 | + if ($post_id) { | |
| 4453 | + return intval($post_id); | |
| 4454 | + } | |
| 4455 | + } | |
| 4456 | + } | |
| 4457 | + | |
| 4458 | + // For WooCommerce products | |
| 4459 | + if (strpos($url, '/product/') !== false || strpos($url, 'product=') !== false) { | |
| 4460 | + // Extract product slug from various URL formats | |
| 4461 | + $product_slug = ''; | |
| 4462 | + | |
| 4463 | + // Handle pretty permalinks: /product/product-name/ | |
| 4464 | + if (preg_match('/\/product\/([^\/\?]+)/', $url, $matches)) { | |
| 4465 | + $product_slug = $matches[1]; | |
| 4466 | + } | |
| 4467 | + // Handle query parameters: ?product=product-name | |
| 4468 | + elseif (preg_match('/[\?&]product=([^&]+)/', $url, $matches)) { | |
| 4469 | + $product_slug = $matches[1]; | |
| 4470 | + } | |
| 4471 | + | |
| 4472 | + if (!empty($product_slug)) { | |
| 4473 | + // Look up product by slug | |
| 4474 | + $product = get_page_by_path($product_slug, OBJECT, 'product'); | |
| 4475 | + if ($product) { | |
| 4476 | + return $product->ID; | |
| 4477 | + } | |
| 4478 | + | |
| 4479 | + // Alternative method: query by post_name | |
| 4480 | + global $wpdb; | |
| 4481 | + $post_id = $wpdb->get_var($wpdb->prepare( | |
| 4482 | + "SELECT ID FROM {$wpdb->posts} WHERE post_name = %s AND post_type = 'product' AND post_status = 'publish'", | |
| 4483 | + $product_slug | |
| 4484 | + )); | |
| 4485 | + | |
| 4486 | + if ($post_id) { | |
| 4487 | + return intval($post_id); | |
| 4488 | + } | |
| 4489 | + } | |
| 4490 | + } | |
| 4491 | + | |
| 4492 | + // Generic approach: try to extract slug and match against all post types | |
| 4493 | + $parsed_url = wp_parse_url($clean_url); | |
| 4494 | + $path = $parsed_url['path'] ?? ''; | |
| 4495 | + | |
| 4496 | + if (!empty($path)) { | |
| 4497 | + // Get the last part of the path as potential slug | |
| 4498 | + $path_parts = array_filter(explode('/', trim($path, '/'))); | |
| 4499 | + $potential_slug = end($path_parts); | |
| 4500 | + | |
| 4501 | + if (!empty($potential_slug)) { | |
| 4502 | + global $wpdb; | |
| 4503 | + | |
| 4504 | + // Try to find any post with this slug | |
| 4505 | + $post_id = $wpdb->get_var($wpdb->prepare( | |
| 4506 | + "SELECT ID FROM {$wpdb->posts} | |
| 4507 | + WHERE post_name = %s | |
| 4508 | + AND post_status IN ('publish', 'closed', 'private') | |
| 4509 | + AND post_type NOT IN ('revision', 'attachment', 'nav_menu_item') | |
| 4510 | + ORDER BY CASE | |
| 4511 | + WHEN post_type = 'post' THEN 1 | |
| 4512 | + WHEN post_type = 'page' THEN 2 | |
| 4513 | + WHEN post_type = 'topic' THEN 3 | |
| 4514 | + WHEN post_type = 'product' THEN 4 | |
| 4515 | + ELSE 5 | |
| 4516 | + END | |
| 4517 | + LIMIT 1", | |
| 4518 | + $potential_slug | |
| 4519 | + )); | |
| 4520 | + | |
| 4521 | + if ($post_id) { | |
| 4522 | + return intval($post_id); | |
| 4523 | + } | |
| 4524 | + } | |
| 4525 | + } | |
| 4526 | + | |
| 4527 | + // ADDITIONAL: Try direct database lookup by URL variations | |
| 4528 | + global $wpdb; | |
| 4529 | + $table_name = $wpdb->prefix . 'mxchat_system_prompt_content'; | |
| 4530 | + | |
| 4531 | + // Try variations of the URL (with/without trailing slash, http/https) | |
| 4532 | + $url_variations = array( | |
| 4533 | + $url, | |
| 4534 | + rtrim($url, '/'), | |
| 4535 | + $url . '/', | |
| 4536 | + str_replace('http://', 'https://', $url), | |
| 4537 | + str_replace('https://', 'http://', $url), | |
| 4538 | + str_replace('http://', 'https://', rtrim($url, '/')), | |
| 4539 | + str_replace('https://', 'http://', rtrim($url, '/')) | |
| 4540 | + ); | |
| 4541 | + | |
| 4542 | + // Remove duplicates | |
| 4543 | + $url_variations = array_unique($url_variations); | |
| 4544 | + | |
| 4545 | + foreach ($url_variations as $variation) { | |
| 4546 | + $existing_record = $wpdb->get_row($wpdb->prepare( | |
| 4547 | + "SELECT id, source_url FROM $table_name WHERE source_url = %s", | |
| 4548 | + $variation | |
| 4549 | + )); | |
| 4550 | + | |
| 4551 | + if ($existing_record) { | |
| 4552 | + // Try to get post ID from this stored URL | |
| 4553 | + $stored_post_id = url_to_postid($existing_record->source_url); | |
| 4554 | + if ($stored_post_id > 0) { | |
| 4555 | + return $stored_post_id; | |
| 4556 | + } | |
| 4557 | + } | |
| 4558 | + } | |
| 4559 | + | |
| 4560 | + return 0; // No match found | |
| 4561 | +} | |
| 4562 | +/** | |
| 4563 | + * Process selected content via AJAX | |
| 4564 | + */ | |
| 4565 | +public function ajax_mxchat_process_selected_content() { | |
| 4566 | + // Basic request validation | |
| 4567 | + if (!check_ajax_referer('mxchat_content_selector_nonce', 'nonce', false)) { | |
| 4568 | + wp_send_json_error('Invalid nonce'); | |
| 4569 | + exit; | |
| 4570 | + } | |
| 4571 | + | |
| 4572 | + if (!current_user_can('manage_options')) { | |
| 4573 | + wp_send_json_error('Unauthorized access'); | |
| 4574 | + exit; | |
| 4575 | + } | |
| 4576 | + | |
| 4577 | + // Get post IDs - safely parse the array | |
| 4578 | + $post_ids = array(); | |
| 4579 | + if (isset($_POST['post_ids']) && is_array($_POST['post_ids'])) { | |
| 4580 | + foreach ($_POST['post_ids'] as $id) { | |
| 4581 | + $post_ids[] = absint($id); | |
| 4582 | + } | |
| 4583 | + } | |
| 4584 | + | |
| 4585 | + if (empty($post_ids)) { | |
| 4586 | + wp_send_json_error('No content selected'); | |
| 4587 | + exit; | |
| 4588 | + } | |
| 4589 | + | |
| 4590 | + // Get bot_id from request | |
| 4591 | + $bot_id = isset($_POST['bot_id']) ? sanitize_key($_POST['bot_id']) : 'default'; | |
| 4592 | + | |
| 4593 | + // ACF→PDF extraction is an install-level setting (Knowledge → ACF Fields, | |
| 4594 | + // plan 11720c). The import modal shows a passive status line pointing | |
| 4595 | + // there; the old per-batch checkbox and its remembered default are gone. | |
| 4596 | + $extract_acf_pdfs = get_option('mxchat_acf_pdf_extraction', '0') === '1'; | |
| 4597 | + | |
| 4598 | + // Process only ONE post at a time to avoid request size issues | |
| 4599 | + $post_id = reset($post_ids); | |
| 4600 | + $post = get_post($post_id); | |
| 4601 | + | |
| 4602 | + if (!$post) { | |
| 4603 | + wp_send_json_error('Post not found'); | |
| 4604 | + exit; | |
| 4605 | + } | |
| 4606 | + | |
| 4607 | + // Media-library attachments take their own route (plan 01010a) — different source | |
| 4608 | + // URL (the file, not the attachment page), different body (an extractor, not | |
| 4609 | + // post_content), and a file that yields no text is refused with a reason instead | |
| 4610 | + // of stored empty. Sends its own response and exits. | |
| 4611 | + if ($post->post_type === 'attachment') { | |
| 4612 | + $this->mxchat_import_attachment($post, $bot_id); | |
| 4613 | + } | |
| 4614 | + | |
| 4615 | + /** | |
| 4616 | + * Allow developers to modify post data before processing into the knowledge base. | |
| 4617 | + * Applied on BOTH content-preparation paths (this manual bulk import and the | |
| 4618 | + * auto-sync path in mxchat_handle_post_update) with the same signature, so a | |
| 4619 | + * callback registered once covers every indexing route. Purely additive — | |
| 4620 | + * zero behaviour change when unhooked. | |
| 4621 | + * | |
| 4622 | + * @param WP_Post $post The post about to be indexed. | |
| 4623 | + * @param string $bot_id Bot context for this import. | |
| 4624 | + */ | |
| 4625 | + $post = apply_filters('mxchat_before_process_post', $post, $bot_id); | |
| 4626 | + if (!($post instanceof WP_Post)) { | |
| 4627 | + $post = get_post($post_id); // defend against a bad callback return | |
| 4628 | + } | |
| 4629 | + | |
| 4630 | + // Assemble the indexable text via the shared post-kind assembler (a3d60c). | |
| 4631 | + // Bulk import reads raw post fields, has always included product custom tabs, | |
| 4632 | + // and passes the install-level ACF→PDF option. | |
| 4633 | + $prepared = $this->mxchat_prepare_post_content_for_indexing($post_id, $post, array( | |
| 4634 | + 'read_display' => false, | |
| 4635 | + 'extract_acf_pdfs' => $extract_acf_pdfs, | |
| 4636 | + 'include_product_tabs' => true, | |
| 4637 | + )); | |
| 4638 | + $content = $prepared['content']; | |
| 4639 | + $pdf_extracted_count = $prepared['pdf_extracted_count']; | |
| 4640 | + | |
| 4641 | + // Debug logging for WordPress Import content | |
| 4642 | + //error_log('[MXCHAT-WP-IMPORT-DEBUG] Post ID: ' . $post_id . ' Title: ' . $post->post_title); | |
| 4643 | + //error_log('[MXCHAT-WP-IMPORT-DEBUG] Raw post_content length: ' . strlen($post->post_content)); | |
| 4644 | + //error_log('[MXCHAT-WP-IMPORT-DEBUG] Final content length: ' . strlen($content)); | |
| 4645 | + //error_log('[MXCHAT-WP-IMPORT-DEBUG] Content preview: ' . substr($content, 0, 300)); | |
| 4646 | + | |
| 4647 | + // Note: Removed 10,000 char limit - chunking now handles large content properly | |
| 4648 | + | |
| 4649 | + // Get bot-specific embedding decision — custom-provider-aware (plan cbd5fd) | |
| 4650 | + $bot_options = $this->get_bot_options($bot_id); | |
| 4651 | + $options = !empty($bot_options) ? $bot_options : get_option('mxchat_options'); | |
| 4652 | + | |
| 4653 | + $preflight = MxChat_Utils::embedding_preflight($options); | |
| 4654 | + if (!$preflight['ok']) { | |
| 4655 | + MxChat_Admin::mxchat_log_debug('api_error', $preflight['reason'] . ' (knowledge processing)'); | |
| 4656 | + wp_send_json_error($preflight['reason']); | |
| 4657 | + exit; | |
| 4658 | + } | |
| 4659 | + $api_key = $preflight['api_key']; | |
| 4660 | + | |
| 4661 | + $source_url = get_permalink($post_id); | |
| 4662 | + $vector_id = md5($source_url); // Vector ID for Pinecone | |
| 4663 | + | |
| 4664 | + // Check for existing content in bot-specific storage | |
| 4665 | + $is_update = false; | |
| 4666 | + | |
| 4667 | + // Get bot-specific Pinecone configuration | |
| 4668 | + $bot_pinecone_config = $this->get_bot_pinecone_config($bot_id); | |
| 4669 | + $use_pinecone = !empty($bot_pinecone_config) && ($bot_pinecone_config['use_pinecone'] ?? false); | |
| 4670 | + | |
| 4671 | + if ($use_pinecone && !empty($bot_pinecone_config['api_key'])) { | |
| 4672 | + // Check Pinecone for this bot | |
| 4673 | + $pinecone_data = $this->mxchat_get_pinecone_processed_content($bot_pinecone_config); | |
| 4674 | + if (isset($pinecone_data[$post_id])) { | |
| 4675 | + $is_update = true; | |
| 4676 | + } | |
| 4677 | + } else { | |
| 4678 | + // Check WordPress DB (same as before since it's shared) | |
| 4679 | + global $wpdb; | |
| 4680 | + $table_name = $wpdb->prefix . 'mxchat_system_prompt_content'; | |
| 4681 | + $existing_record = $wpdb->get_row($wpdb->prepare( | |
| 4682 | + "SELECT id FROM $table_name WHERE source_url = %s", | |
| 4683 | + $source_url | |
| 4684 | + )); | |
| 4685 | + | |
| 4686 | + if ($existing_record) { | |
| 4687 | + $is_update = true; | |
| 4688 | + } | |
| 4689 | + } | |
| 4690 | + | |
| 4691 | + // UPDATED 2.5.6: Determine content type based on post_type | |
| 4692 | + $post_type = $post->post_type; | |
| 4693 | + $content_type = 'content'; // Default fallback | |
| 4694 | + | |
| 4695 | + // Map WordPress post types to content types | |
| 4696 | + switch ($post_type) { | |
| 4697 | + case 'post': | |
| 4698 | + $content_type = 'post'; | |
| 4699 | + break; | |
| 4700 | + case 'page': | |
| 4701 | + $content_type = 'page'; | |
| 4702 | + break; | |
| 4703 | + case 'product': | |
| 4704 | + $content_type = 'product'; | |
| 4705 | + break; | |
| 4706 | + default: | |
| 4707 | + // For custom post types, use the post type name | |
| 4708 | + $content_type = sanitize_key($post_type); | |
| 4709 | + break; | |
| 4710 | + } | |
| 4711 | + | |
| 4712 | + // Use the centralized utility function with bot_id and content_type | |
| 4713 | + $result = MxChat_Utils::submit_content_to_db( | |
| 4714 | + $content, | |
| 4715 | + $source_url, | |
| 4716 | + $api_key, | |
| 4717 | + $vector_id, | |
| 4718 | + $bot_id, | |
| 4719 | + $content_type | |
| 4720 | + ); | |
| 4721 | + | |
| 4722 | + if (is_wp_error($result)) { | |
| 4723 | + MxChat_Admin::mxchat_log_debug('storage_error', 'Knowledge storage failed: ' . $result->get_error_message(), array('source_url' => $source_url)); | |
| 4724 | + wp_send_json_error('Storage failed: ' . $result->get_error_message()); | |
| 4725 | + exit; | |
| 4726 | + } | |
| 4727 | + | |
| 4728 | + // Automatically apply role restriction based on tags | |
| 4729 | + $this->apply_role_restriction_to_post($post_id, $source_url); | |
| 4730 | + | |
| 4731 | + $operation_type = $is_update ? 'update' : 'new'; | |
| 4732 | + | |
| 4733 | + // Count ACF fields for debugging | |
| 4734 | + $acf_field_count = $prepared['acf_fields_found']; | |
| 4735 | + | |
| 4736 | + // Success response with minimal data | |
| 4737 | + wp_send_json_success(array( | |
| 4738 | + 'message' => $operation_type === 'update' ? 'Content updated successfully' : 'Content processed successfully', | |
| 4739 | + 'post_id' => $post_id, | |
| 4740 | + 'title' => $post->post_title, | |
| 4741 | + 'operation_type' => $operation_type, | |
| 4742 | + 'vector_id' => $vector_id, | |
| 4743 | + 'acf_fields_found' => $acf_field_count, | |
| 4744 | + 'pdf_extracted_count' => (int) $pdf_extracted_count, | |
| 4745 | + 'content_preview' => substr($content, 0, 100) . '...', | |
| 4746 | + 'bot_id' => $bot_id | |
| 4747 | + )); | |
| 4748 | + exit; | |
| 4749 | +} | |
| 4750 | + | |
| 4751 | +private function apply_role_restriction_to_post($post_id, $source_url) { | |
| 4752 | + $highest_role = 'public'; | |
| 4753 | + | |
| 4754 | + // Tag-derived restriction. Tags only exist on posts, which is why pages | |
| 4755 | + // could never be restricted from the Role Restrictions screen (plan 6ab002). | |
| 4756 | + $mappings = get_option('mxchat_tag_role_mappings', array()); | |
| 4757 | + if (!empty($mappings)) { | |
| 4758 | + $post_tags = wp_get_post_tags($post_id, array('fields' => 'slugs')); | |
| 4759 | + if (!empty($post_tags)) { | |
| 4760 | + foreach ($post_tags as $tag_slug) { | |
| 4761 | + if (isset($mappings[$tag_slug])) { | |
| 4762 | + $highest_role = self::mxchat_stricter_role($highest_role, $mappings[$tag_slug]); | |
| 4763 | + } | |
| 4764 | + } | |
| 4765 | + } | |
| 4766 | + } | |
| 4767 | + | |
| 4768 | + // Per-entry restriction set on the post itself, which works for pages and | |
| 4769 | + // every other public post type. The stricter of the two wins. | |
| 4770 | + $meta_role = get_post_meta($post_id, '_mxchat_kb_role_restriction', true); | |
| 4771 | + if (is_string($meta_role) && $meta_role !== '') { | |
| 4772 | + $highest_role = self::mxchat_stricter_role($highest_role, $meta_role); | |
| 4773 | + } | |
| 4774 | + | |
| 4775 | + // Nothing restricts this entry — leave it public. | |
| 4776 | + if ($highest_role === 'public' || $highest_role === '') { | |
| 4777 | + return; | |
| 4778 | + } | |
| 4779 | + | |
| 4780 | + // Update the role restriction in the database | |
| 4781 | + global $wpdb; | |
| 4782 | + | |
| 4783 | + // Check if using Pinecone | |
| 4784 | + $pinecone_options = get_option('mxchat_pinecone_addon_options', array()); | |
| 4785 | + $use_pinecone = ($pinecone_options['mxchat_use_pinecone'] ?? '0') === '1'; | |
| 4786 | + | |
| 4787 | + if ($use_pinecone && !empty($pinecone_options['mxchat_pinecone_api_key'])) { | |
| 4788 | + // Update Pinecone role restriction | |
| 4789 | + $roles_table = $wpdb->prefix . 'mxchat_pinecone_roles'; | |
| 4790 | + $vector_id = md5($source_url); | |
| 4791 | + | |
| 4792 | + $wpdb->replace( | |
| 4793 | + $roles_table, | |
| 4794 | + array( | |
| 4795 | + 'vector_id' => $vector_id, | |
| 4796 | + 'role_restriction' => $highest_role, | |
| 4797 | + 'updated_at' => current_time('mysql') | |
| 4798 | + ), | |
| 4799 | + array('%s', '%s', '%s') | |
| 4800 | + ); | |
| 4801 | + } else { | |
| 4802 | + // Update WordPress DB | |
| 4803 | + $table_name = $wpdb->prefix . 'mxchat_system_prompt_content'; | |
| 4804 | + | |
| 4805 | + $wpdb->update( | |
| 4806 | + $table_name, | |
| 4807 | + array('role_restriction' => $highest_role), | |
| 4808 | + array('source_url' => $source_url), | |
| 4809 | + array('%s'), | |
| 4810 | + array('%s') | |
| 4811 | + ); | |
| 4812 | + } | |
| 4813 | + | |
| 4814 | + // The entry's restriction just changed — keep the OpenAI Vector Store | |
| 4815 | + // mirror consistent: non-public pulls the file (file_search has no | |
| 4816 | + // per-role filtering), public re-mirrors it (plan 15b5c6). | |
| 4817 | + if (class_exists('MxChat_Vectorstore_Manager')) { | |
| 4818 | + MxChat_Vectorstore_Manager::handle_role_change($source_url, 'default', $highest_role); | |
| 4819 | + } | |
| 4820 | +} | |
| 4821 | + | |
| 4822 | +public function mxchat_get_public_post_types() { | |
| 4823 | + // Get all public post types | |
| 4824 | + $post_types = get_post_types(array('public' => true), 'objects'); | |
| 4825 | + $post_type_options = array(); | |
| 4826 | + | |
| 4827 | + foreach ($post_types as $post_type) { | |
| 4828 | + $post_type_options[$post_type->name] = $post_type->label; | |
| 4829 | + } | |
| 4830 | + | |
| 4831 | + // Also include common forum/community post types that might not be marked as public | |
| 4832 | + $additional_types = array( | |
| 4833 | + 'topic' => 'Forum Topics (bbPress)', | |
| 4834 | + 'reply' => 'Forum Replies (bbPress)', | |
| 4835 | + 'forum' => 'Forums (bbPress)', | |
| 4836 | + 'wpforo_topic' => 'wpForo Topics', | |
| 4837 | + 'wpforo_post' => 'wpForo Posts' | |
| 4838 | + ); | |
| 4839 | + | |
| 4840 | + foreach ($additional_types as $type_name => $type_label) { | |
| 4841 | + if (post_type_exists($type_name) && !isset($post_type_options[$type_name])) { | |
| 4842 | + $post_type_options[$type_name] = $type_label; | |
| 4843 | + } | |
| 4844 | + } | |
| 4845 | + | |
| 4846 | + return $post_type_options; | |
| 4847 | +} | |
| 4848 | + | |
| 4849 | +/** | |
| 4850 | + * Retrieves processed content from Pinecone API | |
| 4851 | + */ | |
| 4852 | +public function mxchat_get_pinecone_processed_content($pinecone_options) { | |
| 4853 | + $api_key = $pinecone_options['mxchat_pinecone_api_key'] ?? ''; | |
| 4854 | + $host = $pinecone_options['mxchat_pinecone_host'] ?? ''; | |
| 4855 | + | |
| 4856 | + if (empty($api_key) || empty($host)) { | |
| 4857 | + return array(); | |
| 4858 | + } | |
| 4859 | + | |
| 4860 | + $pinecone_data = array(); | |
| 4861 | + | |
| 4862 | + try { | |
| 4863 | + // Always get fresh data from Pinecone | |
| 4864 | + $pinecone_data = $this->mxchat_scan_pinecone_for_processed_content($pinecone_options); | |
| 4865 | + | |
| 4866 | + // Method 2: Final fallback - try stats endpoint (if available) | |
| 4867 | + if (empty($pinecone_data)) { | |
| 4868 | + $stats_url = "https://{$host}/describe_index_stats"; | |
| 4869 | + | |
| 4870 | + $response = wp_remote_post($stats_url, array( | |
| 4871 | + 'headers' => array( | |
| 4872 | + 'Api-Key' => $api_key, | |
| 4873 | + 'Content-Type' => 'application/json' | |
| 4874 | + ), | |
| 4875 | + 'body' => json_encode(array()), | |
| 4876 | + 'timeout' => 30 | |
| 4877 | + )); | |
| 4878 | + | |
| 4879 | + if (!is_wp_error($response) && wp_remote_retrieve_response_code($response) === 200) { | |
| 4880 | + $body = wp_remote_retrieve_body($response); | |
| 4881 | + $stats_data = json_decode($body, true); | |
| 4882 | + } | |
| 4883 | + } | |
| 4884 | + | |
| 4885 | + } catch (Exception $e) { | |
| 4886 | + // Log error but return fresh data only | |
| 4887 | + } | |
| 4888 | + | |
| 4889 | + return $pinecone_data; | |
| 4890 | +} | |
| 4891 | +public function mxchat_fetch_pinecone_vectors_by_ids($pinecone_options, $vector_ids) { | |
| 4892 | + $api_key = $pinecone_options['mxchat_pinecone_api_key'] ?? ''; | |
| 4893 | + $host = $pinecone_options['mxchat_pinecone_host'] ?? ''; | |
| 4894 | + $namespace = $pinecone_options['mxchat_pinecone_namespace'] ?? ''; | |
| 4895 | + | |
| 4896 | + if (empty($api_key) || empty($host) || empty($vector_ids)) { | |
| 4897 | + return array(); | |
| 4898 | + } | |
| 4899 | + | |
| 4900 | + try { | |
| 4901 | + // NOTE (plan 793b82): /vectors/fetch is a GET endpoint with the ids | |
| 4902 | + // repeated in the query string (ids=a&ids=b — http_build_query would | |
| 4903 | + // emit ids[0]=a); the old POST here was answered 200-with-an-empty-body, | |
| 4904 | + // which read as "nothing indexed". Chunked at 100 ids to stay well | |
| 4905 | + // under the measured HTTP 414 URL-length boundary. | |
| 4906 | + $vectors = array(); | |
| 4907 | + // Document index (plan 362c31): documents/fetch instead of the GET loop (same id => vector shape). | |
| 4908 | + if (class_exists('MxChat_Pinecone_Documents') && MxChat_Pinecone_Documents::is_document_options($pinecone_options)) { | |
| 4909 | + $vectors = MxChat_Pinecone_Documents::fetch(array_values($vector_ids), MxChat_Pinecone_Documents::cfg_from_options($pinecone_options), array('source_url', 'created_at')); | |
| 4910 | + } else | |
| 4911 | + foreach (array_chunk(array_values($vector_ids), 100) as $chunk) { | |
| 4912 | + $fetch_query = array(); | |
| 4913 | + foreach ($chunk as $fetch_vid) { | |
| 4914 | + $fetch_query[] = 'ids=' . rawurlencode($fetch_vid); | |
| 4915 | + } | |
| 4916 | + if (!empty($namespace)) { | |
| 4917 | + $fetch_query[] = 'namespace=' . rawurlencode($namespace); | |
| 4918 | + } | |
| 4919 | + | |
| 4920 | + $response = wp_remote_get("https://{$host}/vectors/fetch?" . implode('&', $fetch_query), array( | |
| 4921 | + 'headers' => array( | |
| 4922 | + 'Api-Key' => $api_key, | |
| 4923 | + 'accept' => 'application/json' | |
| 4924 | + ), | |
| 4925 | + 'timeout' => 30 | |
| 4926 | + )); | |
| 4927 | + | |
| 4928 | + if (is_wp_error($response)) { | |
| 4929 | + error_log('MxChat Pinecone: mxchat_fetch_pinecone_vectors_by_ids GET failed: ' . $response->get_error_message()); | |
| 4930 | + continue; | |
| 4931 | + } | |
| 4932 | + | |
| 4933 | + if (wp_remote_retrieve_response_code($response) !== 200) { | |
| 4934 | + error_log('MxChat Pinecone: mxchat_fetch_pinecone_vectors_by_ids GET returned HTTP ' . wp_remote_retrieve_response_code($response)); | |
| 4935 | + continue; | |
| 4936 | + } | |
| 4937 | + | |
| 4938 | + $data = json_decode(wp_remote_retrieve_body($response), true); | |
| 4939 | + if (isset($data['vectors']) && is_array($data['vectors'])) { | |
| 4940 | + $vectors += $data['vectors']; | |
| 4941 | + } | |
| 4942 | + } | |
| 4943 | + | |
| 4944 | + if (empty($vectors)) { | |
| 4945 | + return array(); | |
| 4946 | + } | |
| 4947 | + | |
| 4948 | + $processed_data = array(); | |
| 4949 | + | |
| 4950 | + foreach ($vectors as $vector_id => $vector_data) { | |
| 4951 | + $metadata = $vector_data['metadata'] ?? array(); | |
| 4952 | + $source_url = $metadata['source_url'] ?? ''; | |
| 4953 | + | |
| 4954 | + if (!empty($source_url)) { | |
| 4955 | + $post_id = url_to_postid($source_url); | |
| 4956 | + // A media entry is stored under its FILE url, which url_to_postid() cannot | |
| 4957 | + // resolve — without this the Pinecone read reports every imported | |
| 4958 | + // attachment as unprocessed while the WordPress-DB read (which goes | |
| 4959 | + // through mxchat_url_to_post_id_improved) reports it correctly (plan 01010a). | |
| 4960 | + if (!$post_id) { | |
| 4961 | + $post_id = (int) attachment_url_to_postid($source_url); | |
| 4962 | + } | |
| 4963 | + if ($post_id) { | |
| 4964 | + $created_at = $metadata['created_at'] ?? ''; | |
| 4965 | + $processed_date = 'Recently'; | |
| 4966 | + | |
| 4967 | + if (!empty($created_at)) { | |
| 4968 | + $timestamp = is_numeric($created_at) ? $created_at : strtotime($created_at); | |
| 4969 | + if ($timestamp) { | |
| 4970 | + $processed_date = human_time_diff($timestamp, current_time('timestamp')) . ' ago'; | |
| 4971 | + } | |
| 4972 | + } | |
| 4973 | + | |
| 4974 | + $processed_data[$post_id] = array( | |
| 4975 | + 'db_id' => $vector_id, | |
| 4976 | + 'processed_date' => $processed_date, | |
| 4977 | + 'url' => $source_url, | |
| 4978 | + 'source' => 'pinecone', | |
| 4979 | + 'timestamp' => $timestamp ?? current_time('timestamp') | |
| 4980 | + ); | |
| 4981 | + } | |
| 4982 | + } | |
| 4983 | + } | |
| 4984 | + | |
| 4985 | + //error_log('DEBUG: Processed ' . count($processed_data) . ' vectors from fetch'); | |
| 4986 | + return $processed_data; | |
| 4987 | + | |
| 4988 | + } catch (Exception $e) { | |
| 4989 | + //error_log('DEBUG: Exception in fetch_pinecone_vectors_by_ids: ' . $e->getMessage()); | |
| 4990 | + return array(); | |
| 4991 | + } | |
| 4992 | +} | |
| 4993 | + | |
| 4994 | +/** | |
| 4995 | + * Get embedding dimensions based on the selected model. | |
| 4996 | + */ | |
| 4997 | +private function mxchat_get_embedding_dimensions() { | |
| 4998 | + // See MxChat_Pinecone_Manager::mxchat_get_embedding_dimensions() — same answer (plan 418afd). | |
| 4999 | + $options = get_option('mxchat_options', array()); | |
| 5000 | + $dimension = MxChat_Utils::expected_embedding_dimension($options); | |
| 5001 | + if (!is_wp_error($dimension) && (int) $dimension > 0) { | |
| 5002 | + return (int) $dimension; | |
| 5003 | + } | |
| 5004 | + return MxChat_Utils::standard_embedding_dimension($options); | |
| 5005 | +} | |
| 5006 | + | |
| 5007 | +/** | |
| 5008 | + * Scan Pinecone for processed content | |
| 5009 | + */ | |
| 5010 | +public function mxchat_scan_pinecone_for_processed_content($pinecone_options) { | |
| 5011 | + $api_key = $pinecone_options['mxchat_pinecone_api_key'] ?? ''; | |
| 5012 | + $host = $pinecone_options['mxchat_pinecone_host'] ?? ''; | |
| 5013 | + // plan 793b82: this scan was namespace-blind — on a namespaced setup it | |
| 5014 | + // surveyed the default namespace and reported the wrong content as indexed. | |
| 5015 | + $namespace = $pinecone_options['mxchat_pinecone_namespace'] ?? ''; | |
| 5016 | + | |
| 5017 | + if (empty($api_key) || empty($host)) { | |
| 5018 | + return array(); | |
| 5019 | + } | |
| 5020 | + | |
| 5021 | + try { | |
| 5022 | + // Use multiple random vectors to get better coverage | |
| 5023 | + $all_matches = array(); | |
| 5024 | + $seen_ids = array(); | |
| 5025 | + | |
| 5026 | + // Get correct dimensions for the configured embedding model | |
| 5027 | + $dimensions = $this->mxchat_get_embedding_dimensions(); | |
| 5028 | + | |
| 5029 | + // Try 3 different random vectors to get better coverage | |
| 5030 | + for ($i = 0; $i < 3; $i++) { | |
| 5031 | + $query_url = "https://{$host}/query"; | |
| 5032 | + | |
| 5033 | + // Generate a random unit vector instead of zeros | |
| 5034 | + $random_vector = array(); | |
| 5035 | + for ($j = 0; $j < $dimensions; $j++) { | |
| 5036 | + $random_vector[] = (rand(-1000, 1000) / 1000.0); | |
| 5037 | + } | |
| 5038 | + | |
| 5039 | + // Normalize the vector to unit length | |
| 5040 | + $magnitude = sqrt(array_sum(array_map(function($x) { return $x * $x; }, $random_vector))); | |
| 5041 | + if ($magnitude > 0) { | |
| 5042 | + $random_vector = array_map(function($x) use ($magnitude) { return $x / $magnitude; }, $random_vector); | |
| 5043 | + } | |
| 5044 | + | |
| 5045 | + $query_data = array( | |
| 5046 | + 'includeMetadata' => true, | |
| 5047 | + 'includeValues' => false, | |
| 5048 | + 'topK' => 10000, | |
| 5049 | + 'vector' => $random_vector | |
| 5050 | + ); | |
| 5051 | + | |
| 5052 | + if (!empty($namespace)) { | |
| 5053 | + $query_data['namespace'] = $namespace; | |
| 5054 | + } | |
| 5055 | + | |
| 5056 | + // Document index (plan 362c31): same random-vector sweep through documents/search. | |
| 5057 | + if (class_exists('MxChat_Pinecone_Documents') && MxChat_Pinecone_Documents::is_document_options($pinecone_options)) { | |
| 5058 | + $doc_hits = MxChat_Pinecone_Documents::search_dense($random_vector, 10000, null, MxChat_Pinecone_Documents::cfg_from_options($pinecone_options), array('source_url', 'created_at')); | |
| 5059 | + foreach ((is_wp_error($doc_hits) ? array() : $doc_hits) as $match) { | |
| 5060 | + $match_id = $match['id'] ?? ''; | |
| 5061 | + if (!empty($match_id) && !isset($seen_ids[$match_id])) { | |
| 5062 | + $all_matches[] = $match; | |
| 5063 | + $seen_ids[$match_id] = true; | |
| 5064 | + } | |
| 5065 | + } | |
| 5066 | + continue; | |
| 5067 | + } | |
| 5068 | + | |
| 5069 | + $response = wp_remote_post($query_url, array( | |
| 5070 | + 'headers' => array( | |
| 5071 | + 'Api-Key' => $api_key, | |
| 5072 | + 'Content-Type' => 'application/json' | |
| 5073 | + ), | |
| 5074 | + 'body' => json_encode($query_data), | |
| 5075 | + 'timeout' => 30 | |
| 5076 | + )); | |
| 5077 | + | |
| 5078 | + if (is_wp_error($response)) { | |
| 5079 | + continue; | |
| 5080 | + } | |
| 5081 | + | |
| 5082 | + $response_code = wp_remote_retrieve_response_code($response); | |
| 5083 | + | |
| 5084 | + if ($response_code !== 200) { | |
| 5085 | + continue; | |
| 5086 | + } | |
| 5087 | + | |
| 5088 | + $body = wp_remote_retrieve_body($response); | |
| 5089 | + $data = json_decode($body, true); | |
| 5090 | + | |
| 5091 | + if (isset($data['matches'])) { | |
| 5092 | + foreach ($data['matches'] as $match) { | |
| 5093 | + $match_id = $match['id'] ?? ''; | |
| 5094 | + if (!empty($match_id) && !isset($seen_ids[$match_id])) { | |
| 5095 | + $all_matches[] = $match; | |
| 5096 | + $seen_ids[$match_id] = true; | |
| 5097 | + } | |
| 5098 | + } | |
| 5099 | + } | |
| 5100 | + } | |
| 5101 | + | |
| 5102 | + // Convert matches to processed data format, grouping by URL to count chunks | |
| 5103 | + $processed_data = array(); | |
| 5104 | + $url_chunk_counts = array(); | |
| 5105 | + | |
| 5106 | + foreach ($all_matches as $match) { | |
| 5107 | + $metadata = $match['metadata'] ?? array(); | |
| 5108 | + $source_url = $metadata['source_url'] ?? ''; | |
| 5109 | + $match_id = $match['id'] ?? ''; | |
| 5110 | + | |
| 5111 | + if (!empty($source_url) && !empty($match_id)) { | |
| 5112 | + $post_id = url_to_postid($source_url); | |
| 5113 | + // A media entry is stored under its FILE url, which url_to_postid() cannot | |
| 5114 | + // resolve — without this the Pinecone read reports every imported | |
| 5115 | + // attachment as unprocessed while the WordPress-DB read (which goes | |
| 5116 | + // through mxchat_url_to_post_id_improved) reports it correctly (plan 01010a). | |
| 5117 | + if (!$post_id) { | |
| 5118 | + $post_id = (int) attachment_url_to_postid($source_url); | |
| 5119 | + } | |
| 5120 | + if ($post_id) { | |
| 5121 | + // Count chunks per post_id | |
| 5122 | + if (!isset($url_chunk_counts[$post_id])) { | |
| 5123 | + $url_chunk_counts[$post_id] = 0; | |
| 5124 | + } | |
| 5125 | + $url_chunk_counts[$post_id]++; | |
| 5126 | + | |
| 5127 | + $created_at = $metadata['created_at'] ?? ''; | |
| 5128 | + $processed_date = 'Recently'; | |
| 5129 | + | |
| 5130 | + if (!empty($created_at)) { | |
| 5131 | + $timestamp = is_numeric($created_at) ? $created_at : strtotime($created_at); | |
| 5132 | + if ($timestamp) { | |
| 5133 | + $processed_date = human_time_diff($timestamp, current_time('timestamp')) . ' ago'; | |
| 5134 | + } | |
| 5135 | + } | |
| 5136 | + | |
| 5137 | + // Only store if not already set, or update with newer timestamp | |
| 5138 | + if (!isset($processed_data[$post_id]) || | |
| 5139 | + ($timestamp ?? 0) > ($processed_data[$post_id]['timestamp'] ?? 0)) { | |
| 5140 | + $processed_data[$post_id] = array( | |
| 5141 | + 'db_id' => $match_id, | |
| 5142 | + 'processed_date' => $processed_date, | |
| 5143 | + 'url' => $source_url, | |
| 5144 | + 'source' => 'pinecone', | |
| 5145 | + 'timestamp' => $timestamp ?? current_time('timestamp') | |
| 5146 | + ); | |
| 5147 | + } | |
| 5148 | + } | |
| 5149 | + } | |
| 5150 | + } | |
| 5151 | + | |
| 5152 | + // Add chunk counts to processed data | |
| 5153 | + foreach ($url_chunk_counts as $post_id => $chunk_count) { | |
| 5154 | + if (isset($processed_data[$post_id])) { | |
| 5155 | + $processed_data[$post_id]['chunk_count'] = $chunk_count; | |
| 5156 | + } | |
| 5157 | + } | |
| 5158 | + | |
| 5159 | + return $processed_data; | |
| 5160 | + | |
| 5161 | + } catch (Exception $e) { | |
| 5162 | + return array(); | |
| 5163 | + } | |
| 5164 | +} | |
| 5165 | +/** | |
| 5166 | + * Generate embeddings from input text for MXChat with bot support | |
| 5167 | + */ | |
| 5168 | +private function mxchat_generate_embedding($text, $bot_id = 'default') { | |
| 5169 | + // Enable detailed logging for debugging | |
| 5170 | + //error_log('[MXCHAT-EMBED] Starting embedding generation for bot: ' . $bot_id . '. Text length: ' . strlen($text) . ' bytes'); | |
| 5171 | + //error_log('[MXCHAT-EMBED] Text preview: ' . substr($text, 0, 100) . '...'); | |
| 5172 | + | |
| 5173 | + // Get bot-specific options | |
| 5174 | + $bot_options = $this->get_bot_options($bot_id); | |
| 5175 | + $options = !empty($bot_options) ? $bot_options : get_option('mxchat_options'); | |
| 5176 | + | |
| 5177 | + // Opt-in: when the custom provider is selected for embeddings, index through | |
| 5178 | + // the same custom endpoint the query path uses so stored vectors and query | |
| 5179 | + // vectors share a model. Returns the vector array on success, or an error | |
| 5180 | + // string on failure (this function's existing failure contract). | |
| 5181 | + if (isset($options['custom_provider_for_embeddings']) && $options['custom_provider_for_embeddings'] === 'on') { | |
| 5182 | + if (!class_exists('MxChat_Utils')) { | |
| 5183 | + require_once dirname(__FILE__) . '/../includes/class-mxchat-utils.php'; | |
| 5184 | + } | |
| 5185 | + return MxChat_Utils::generate_embedding_custom($text, $options); | |
| 5186 | + } | |
| 5187 | + | |
| 5188 | + $selected_model = $options['embedding_model'] ?? 'text-embedding-ada-002'; | |
| 5189 | + //error_log('[MXCHAT-EMBED] Selected embedding model for bot ' . $bot_id . ': ' . $selected_model); | |
| 5190 | + | |
| 5191 | + // Determine provider and endpoint | |
| 5192 | + if (strpos($selected_model, 'voyage') === 0) { | |
| 5193 | + $api_key = $options['voyage_api_key'] ?? ''; | |
| 5194 | + $endpoint = 'https://api.voyageai.com/v1/embeddings'; | |
| 5195 | + $provider_name = 'Voyage AI'; | |
| 5196 | + //error_log('[MXCHAT-EMBED] Using Voyage AI API for bot ' . $bot_id); | |
| 5197 | + } elseif (strpos($selected_model, 'gemini-embedding') === 0) { | |
| 5198 | + $api_key = $options['gemini_api_key'] ?? ''; | |
| 5199 | + $endpoint = 'https://generativelanguage.googleapis.com/v1beta/models/' . $selected_model . ':embedContent'; | |
| 5200 | + $provider_name = 'Google Gemini'; | |
| 5201 | + //error_log('[MXCHAT-EMBED] Using Google Gemini API for bot ' . $bot_id); | |
| 5202 | + } else { | |
| 5203 | + $api_key = $options['api_key'] ?? ''; | |
| 5204 | + $endpoint = 'https://api.openai.com/v1/embeddings'; | |
| 5205 | + $provider_name = 'OpenAI'; | |
| 5206 | + //error_log('[MXCHAT-EMBED] Using OpenAI API for bot ' . $bot_id); | |
| 5207 | + } | |
| 5208 | + | |
| 5209 | + //error_log('[MXCHAT-EMBED] Using endpoint: ' . $endpoint); | |
| 5210 | + | |
| 5211 | + if (empty($api_key)) { | |
| 5212 | + $error_message = sprintf('Missing %s API key for bot %s. Please configure your API key in the bot settings.', $provider_name, $bot_id); | |
| 5213 | + //error_log('[MXCHAT-EMBED] Error: ' . $error_message); | |
| 5214 | + return $error_message; | |
| 5215 | + } | |
| 5216 | + | |
| 5217 | + // Check if text is too long (for OpenAI, roughly estimate tokens as words/0.75) | |
| 5218 | + $estimated_tokens = ceil(str_word_count($text) / 0.75); | |
| 5219 | + //error_log('[MXCHAT-EMBED] Estimated token count: ~' . $estimated_tokens); | |
| 5220 | + | |
| 5221 | + if ($estimated_tokens > 8000 && strpos($selected_model, 'voyage') === false && strpos($selected_model, 'gemini-embedding') === false) { | |
| 5222 | + //error_log('[MXCHAT-EMBED] Warning: Text may exceed OpenAI token limits (8K for most models)'); | |
| 5223 | + // Consider truncating text here | |
| 5224 | + } | |
| 5225 | + | |
| 5226 | + // Prepare request body based on provider | |
| 5227 | + if (strpos($selected_model, 'gemini-embedding') === 0) { | |
| 5228 | + // Gemini API format | |
| 5229 | + $request_body = array( | |
| 5230 | + 'model' => 'models/' . $selected_model, | |
| 5231 | + 'content' => array( | |
| 5232 | + 'parts' => array( | |
| 5233 | + array('text' => $text) | |
| 5234 | + ) | |
| 5235 | + ) | |
| 5236 | + ); | |
| 5237 | + | |
| 5238 | + // Set output dimensionality to 1536 for consistency with other models | |
| 5239 | + $request_body['outputDimensionality'] = 1536; | |
| 5240 | + } else { | |
| 5241 | + // OpenAI/Voyage API format | |
| 5242 | + $request_body = array( | |
| 5243 | + 'model' => $selected_model, | |
| 5244 | + 'input' => $text | |
| 5245 | + ); | |
| 5246 | + | |
| 5247 | + // Add output_dimension for voyage-3-large model | |
| 5248 | + if ($selected_model === 'voyage-3-large') { | |
| 5249 | + $request_body['output_dimension'] = 2048; | |
| 5250 | + } | |
| 5251 | + } | |
| 5252 | + | |
| 5253 | + //error_log('[MXCHAT-EMBED] Request prepared with model: ' . $selected_model); | |
| 5254 | + | |
| 5255 | + // Prepare headers based on provider | |
| 5256 | + if (strpos($selected_model, 'gemini-embedding') === 0) { | |
| 5257 | + // Gemini uses API key as query parameter | |
| 5258 | + $endpoint .= '?key=' . $api_key; | |
| 5259 | + $headers = array( | |
| 5260 | + 'Content-Type' => 'application/json' | |
| 5261 | + ); | |
| 5262 | + } else { | |
| 5263 | + // OpenAI/Voyage use Bearer token | |
| 5264 | + $headers = array( | |
| 5265 | + 'Authorization' => 'Bearer ' . $api_key, | |
| 5266 | + 'Content-Type' => 'application/json' | |
| 5267 | + ); | |
| 5268 | + } | |
| 5269 | + | |
| 5270 | + // Make API request | |
| 5271 | + //error_log('[MXCHAT-EMBED] Sending API request to: ' . $endpoint); | |
| 5272 | + $response = wp_remote_post($endpoint, array( | |
| 5273 | + 'body' => wp_json_encode($request_body), | |
| 5274 | + 'headers' => $headers, | |
| 5275 | + 'timeout' => 60 // Increased timeout for large inputs | |
| 5276 | + )); | |
| 5277 | + | |
| 5278 | + // Handle wp_remote_post errors | |
| 5279 | + if (is_wp_error($response)) { | |
| 5280 | + $error_message = $response->get_error_message(); | |
| 5281 | + //error_log('[MXCHAT-EMBED] WP Remote Post Error: ' . $error_message); | |
| 5282 | + return 'Connection error: ' . $error_message; | |
| 5283 | + } | |
| 5284 | + | |
| 5285 | + // Get and check HTTP response code | |
| 5286 | + $http_code = wp_remote_retrieve_response_code($response); | |
| 5287 | + //error_log('[MXCHAT-EMBED] API Response Code: ' . $http_code); | |
| 5288 | + | |
| 5289 | + if ($http_code !== 200) { | |
| 5290 | + $error_body = wp_remote_retrieve_body($response); | |
| 5291 | + //error_log('[MXCHAT-EMBED] API Error Response Body: ' . $error_body); | |
| 5292 | + | |
| 5293 | + // Try to parse error for more details | |
| 5294 | + $error_json = json_decode($error_body, true); | |
| 5295 | + if (json_last_error() === JSON_ERROR_NONE && isset($error_json['error'])) { | |
| 5296 | + $error_type = $error_json['error']['type'] ?? 'unknown'; | |
| 5297 | + $error_message = $error_json['error']['message'] ?? 'No message'; | |
| 5298 | + //error_log('[MXCHAT-EMBED] API Error Type: ' . $error_type); | |
| 5299 | + //error_log('[MXCHAT-EMBED] API Error Message: ' . $error_message); | |
| 5300 | + | |
| 5301 | + // Keep the provider's own diagnostic — a restricted-key 401 names the | |
| 5302 | + // exact missing scope, and replacing it with "check your API key" sent | |
| 5303 | + // a customer to regenerate two keys (plan 46b596). Same shape as | |
| 5304 | + // MxChat_Utils::embedding_failure_error() so both ingestion paths read | |
| 5305 | + // identically. Key never appears in provider messages, but scrub anyway. | |
| 5306 | + if ($error_type === 'invalid_request_error' || $error_type === 'authentication_error') { | |
| 5307 | + if (is_string($api_key) && $api_key !== '') { | |
| 5308 | + $error_message = str_replace($api_key, '[redacted]', $error_message); | |
| 5309 | + } | |
| 5310 | + $error_message = sprintf( | |
| 5311 | + 'Embedding failed (%s, HTTP %d): %s', | |
| 5312 | + $selected_model, | |
| 5313 | + $http_code, | |
| 5314 | + substr($error_message, 0, 300) | |
| 5315 | + ); | |
| 5316 | + } | |
| 5317 | + | |
| 5318 | + //error_log('[MXCHAT-EMBED] Returning error: ' . $error_message); | |
| 5319 | + return $error_message; | |
| 5320 | + } | |
| 5321 | + | |
| 5322 | + $error_message = sprintf("API Error (HTTP %d): Unable to generate embedding for bot %s", $http_code, $bot_id); | |
| 5323 | + //error_log('[MXCHAT-EMBED] Returning error: ' . $error_message); | |
| 5324 | + return $error_message; | |
| 5325 | + } | |
| 5326 | + | |
| 5327 | + // Parse response body | |
| 5328 | + $response_body = wp_remote_retrieve_body($response); | |
| 5329 | + //error_log('[MXCHAT-EMBED] Received response length: ' . strlen($response_body) . ' bytes'); | |
| 5330 | + | |
| 5331 | + $response_data = json_decode($response_body, true); | |
| 5332 | + | |
| 5333 | + if (json_last_error() !== JSON_ERROR_NONE) { | |
| 5334 | + $error = json_last_error_msg(); | |
| 5335 | + //error_log('[MXCHAT-EMBED] JSON Parse Error: ' . $error); | |
| 5336 | + //error_log('[MXCHAT-EMBED] Response preview: ' . substr($response_body, 0, 200)); | |
| 5337 | + return "Failed to parse API response: $error"; | |
| 5338 | + } | |
| 5339 | + | |
| 5340 | + // Handle different response formats based on provider | |
| 5341 | + if (strpos($selected_model, 'gemini-embedding') === 0) { | |
| 5342 | + // Gemini API response format | |
| 5343 | + if (isset($response_data['embedding']['values'])) { | |
| 5344 | + $embedding_dimensions = count($response_data['embedding']['values']); | |
| 5345 | + //error_log('[MXCHAT-EMBED] Successfully extracted Gemini embedding with ' . $embedding_dimensions . ' dimensions'); | |
| 5346 | + | |
| 5347 | + // Check if embedding dimensions are as expected (should be 1536) | |
| 5348 | + if ($embedding_dimensions !== 1536) { | |
| 5349 | + //error_log('[MXCHAT-EMBED] Warning: Unexpected Gemini embedding dimensions: ' . $embedding_dimensions); | |
| 5350 | + } | |
| 5351 | + | |
| 5352 | + return $response_data['embedding']['values']; | |
| 5353 | + } else { | |
| 5354 | + //error_log('[MXCHAT-EMBED] Error: No embedding found in Gemini response'); | |
| 5355 | + //error_log('[MXCHAT-EMBED] Response structure: ' . wp_json_encode(array_keys($response_data))); | |
| 5356 | + | |
| 5357 | + if (isset($response_data['error'])) { | |
| 5358 | + $error_message = "Gemini API Error in response: " . wp_json_encode($response_data['error']); | |
| 5359 | + //error_log('[MXCHAT-EMBED] ' . $error_message); | |
| 5360 | + return $error_message; | |
| 5361 | + } | |
| 5362 | + | |
| 5363 | + $error_message = "Invalid Gemini API response format: No embedding found"; | |
| 5364 | + //error_log('[MXCHAT-EMBED] ' . $error_message); | |
| 5365 | + return $error_message; | |
| 5366 | + } | |
| 5367 | + } else { | |
| 5368 | + // OpenAI/Voyage API response format | |
| 5369 | + if (isset($response_data['data'][0]['embedding'])) { | |
| 5370 | + $embedding_dimensions = count($response_data['data'][0]['embedding']); | |
| 5371 | + //error_log('[MXCHAT-EMBED] Successfully extracted embedding with ' . $embedding_dimensions . ' dimensions'); | |
| 5372 | + | |
| 5373 | + // Check if embedding dimensions are as expected | |
| 5374 | + if (($selected_model === 'text-embedding-ada-002' && $embedding_dimensions !== 1536) || | |
| 5375 | + ($selected_model === 'voyage-3-large' && $embedding_dimensions !== 2048)) { | |
| 5376 | + //error_log('[MXCHAT-EMBED] Warning: Unexpected embedding dimensions'); | |
| 5377 | + } | |
| 5378 | + | |
| 5379 | + return $response_data['data'][0]['embedding']; | |
| 5380 | + } else { | |
| 5381 | + //error_log('[MXCHAT-EMBED] Error: No embedding found in response'); | |
| 5382 | + //error_log('[MXCHAT-EMBED] Response structure: ' . wp_json_encode(array_keys($response_data))); | |
| 5383 | + | |
| 5384 | + if (isset($response_data['error'])) { | |
| 5385 | + $error_message = "API Error in response: " . wp_json_encode($response_data['error']); | |
| 5386 | + //error_log('[MXCHAT-EMBED] ' . $error_message); | |
| 5387 | + return $error_message; | |
| 5388 | + } | |
| 5389 | + | |
| 5390 | + $error_message = "Invalid API response format: No embedding found"; | |
| 5391 | + //error_log('[MXCHAT-EMBED] ' . $error_message); | |
| 5392 | + return $error_message; | |
| 5393 | + } | |
| 5394 | + } | |
| 5395 | +} | |
| 5396 | + | |
| 5397 | +/** | |
| 5398 | + * Get bot-specific options for multi-bot functionality | |
| 5399 | + * Falls back to default options if bot_id is 'default' or multi-bot add-on is not active | |
| 5400 | + */ | |
| 5401 | +private function get_bot_options($bot_id = 'default') { | |
| 5402 | + //error_log("MXCHAT DEBUG: get_bot_options called for bot: " . $bot_id); | |
| 5403 | + | |
| 5404 | + if ($bot_id === 'default' || !class_exists('MxChat_Multi_Bot_Manager')) { | |
| 5405 | + //error_log("MXCHAT DEBUG: Using default options (no multi-bot or bot is 'default')"); | |
| 5406 | + return array(); | |
| 5407 | + } | |
| 5408 | + | |
| 5409 | + $bot_options = apply_filters('mxchat_get_bot_options', array(), $bot_id); | |
| 5410 | + | |
| 5411 | + if (!empty($bot_options)) { | |
| 5412 | + //error_log("MXCHAT DEBUG: Got bot-specific options from filter"); | |
| 5413 | + if (isset($bot_options['similarity_threshold'])) { | |
| 5414 | + //error_log(" - similarity_threshold: " . $bot_options['similarity_threshold']); | |
| 5415 | + } | |
| 5416 | + } | |
| 5417 | + | |
| 5418 | + return is_array($bot_options) ? $bot_options : array(); | |
| 5419 | +} | |
| 5420 | + | |
| 5421 | +/** | |
| 5422 | + * Get bot-specific Pinecone configuration | |
| 5423 | + * Used in the knowledge retrieval functions | |
| 5424 | + */ | |
| 5425 | +// Also add debugging to your get_bot_pinecone_config function | |
| 5426 | +private function get_bot_pinecone_config($bot_id = 'default') { | |
| 5427 | + //error_log("MXCHAT DEBUG: get_bot_pinecone_config called for bot: " . $bot_id); | |
| 5428 | + | |
| 5429 | + // If default bot or multi-bot add-on not active, use default Pinecone config | |
| 5430 | + if ($bot_id === 'default' || !class_exists('MxChat_Multi_Bot_Manager')) { | |
| 5431 | + //error_log("MXCHAT DEBUG: Using default Pinecone config (no multi-bot or bot is 'default')"); | |
| 5432 | + $addon_options = get_option('mxchat_pinecone_addon_options', array()); | |
| 5433 | + $config = array( | |
| 5434 | + 'use_pinecone' => (isset($addon_options['mxchat_use_pinecone']) && $addon_options['mxchat_use_pinecone'] === '1'), | |
| 5435 | + 'api_key' => $addon_options['mxchat_pinecone_api_key'] ?? '', | |
| 5436 | + 'host' => $addon_options['mxchat_pinecone_host'] ?? '', | |
| 5437 | + 'namespace' => $addon_options['mxchat_pinecone_namespace'] ?? '' | |
| 5438 | + ); | |
| 5439 | + //error_log("MXCHAT DEBUG: Default config - use_pinecone: " . ($config['use_pinecone'] ? 'true' : 'false')); | |
| 5440 | + return $config; | |
| 5441 | + } | |
| 5442 | + | |
| 5443 | + //error_log("MXCHAT DEBUG: Calling filter 'mxchat_get_bot_pinecone_config' for bot: " . $bot_id); | |
| 5444 | + | |
| 5445 | + // Hook for multi-bot add-on to provide bot-specific Pinecone config | |
| 5446 | + $bot_pinecone_config = apply_filters('mxchat_get_bot_pinecone_config', array(), $bot_id); | |
| 5447 | + | |
| 5448 | + if (!empty($bot_pinecone_config)) { | |
| 5449 | + //error_log("MXCHAT DEBUG: Got bot-specific config from filter"); | |
| 5450 | + //error_log(" - use_pinecone: " . (isset($bot_pinecone_config['use_pinecone']) ? ($bot_pinecone_config['use_pinecone'] ? 'true' : 'false') : 'not set')); | |
| 5451 | + //error_log(" - host: " . ($bot_pinecone_config['host'] ?? 'not set')); | |
| 5452 | + //error_log(" - namespace: " . ($bot_pinecone_config['namespace'] ?? 'not set')); | |
| 5453 | + } else { | |
| 5454 | + //error_log("MXCHAT DEBUG: Filter returned empty config!"); | |
| 5455 | + } | |
| 5456 | + | |
| 5457 | + return is_array($bot_pinecone_config) ? $bot_pinecone_config : array(); | |
| 5458 | +} | |
| 5459 | + | |
| 5460 | + | |
| 5461 | +public function mxchat_ajax_dismiss_completed_status() { | |
| 5462 | + try { | |
| 5463 | + // Verify the request | |
| 5464 | + check_ajax_referer('mxchat_status_nonce', 'nonce'); | |
| 5465 | + | |
| 5466 | + if (!current_user_can('manage_options')) { | |
| 5467 | + wp_send_json_error('Unauthorized access'); | |
| 5468 | + exit; | |
| 5469 | + } | |
| 5470 | + | |
| 5471 | + $card_type = isset($_POST['card_type']) ? sanitize_text_field($_POST['card_type']) : ''; | |
| 5472 | + | |
| 5473 | + if ($card_type === 'pdf') { | |
| 5474 | + // Clear PDF status | |
| 5475 | + $pdf_url = get_transient('mxchat_last_pdf_url'); | |
| 5476 | + if ($pdf_url) { | |
| 5477 | + delete_transient('mxchat_pdf_status_' . md5($pdf_url)); | |
| 5478 | + delete_transient('mxchat_last_pdf_url'); | |
| 5479 | + } | |
| 5480 | + } elseif ($card_type === 'sitemap') { | |
| 5481 | + // Clear sitemap status | |
| 5482 | + $sitemap_url = get_transient('mxchat_last_sitemap_url'); | |
| 5483 | + if ($sitemap_url) { | |
| 5484 | + delete_transient('mxchat_sitemap_status_' . md5($sitemap_url)); | |
| 5485 | + delete_transient('mxchat_last_sitemap_url'); | |
| 5486 | + } | |
| 5487 | + } | |
| 5488 | + | |
| 5489 | + wp_send_json_success(array('message' => 'Status dismissed successfully')); | |
| 5490 | + | |
| 5491 | + } catch (Exception $e) { | |
| 5492 | + wp_send_json_error(array('message' => 'Error dismissing status: ' . $e->getMessage())); | |
| 5493 | + } | |
| 5494 | +} | |
| 5495 | + | |
| 5496 | +/** | |
| 5497 | + * Render completed status cards on page load | |
| 5498 | + * This ensures completed processing status persists through page refreshes | |
| 5499 | + */ | |
| 5500 | +public function mxchat_render_completed_status_cards() { | |
| 5501 | + $output = ''; | |
| 5502 | + | |
| 5503 | + // Check for completed PDF status | |
| 5504 | + $pdf_url = get_transient('mxchat_last_pdf_url'); | |
| 5505 | + if ($pdf_url) { | |
| 5506 | + $pdf_status = $this->mxchat_get_pdf_processing_status($pdf_url); | |
| 5507 | + if ($pdf_status && ($pdf_status['status'] === 'complete' || $pdf_status['status'] === 'error')) { | |
| 5508 | + $output .= $this->mxchat_render_pdf_status_card($pdf_status, $pdf_url); | |
| 5509 | + } | |
| 5510 | + } | |
| 5511 | + | |
| 5512 | + // Check for completed sitemap status | |
| 5513 | + $sitemap_url = get_transient('mxchat_last_sitemap_url'); | |
| 5514 | + if ($sitemap_url) { | |
| 5515 | + $sitemap_status = $this->mxchat_get_sitemap_processing_status($sitemap_url); | |
| 5516 | + if ($sitemap_status && ($sitemap_status['status'] === 'complete' || $sitemap_status['status'] === 'error')) { | |
| 5517 | + $output .= $this->mxchat_render_sitemap_status_card($sitemap_status, $sitemap_url); | |
| 5518 | + } | |
| 5519 | + } | |
| 5520 | + | |
| 5521 | + return $output; | |
| 5522 | +} | |
| 5523 | + | |
| 5524 | +/** | |
| 5525 | + * Render PDF status card HTML | |
| 5526 | + */ | |
| 5527 | +private function mxchat_render_pdf_status_card($status, $pdf_url) { | |
| 5528 | + $html = '<div class="mxchat-status-card" data-card-type="pdf">'; | |
| 5529 | + $html .= '<div class="mxchat-status-header">'; | |
| 5530 | + $html .= '<h4>' . esc_html__('PDF Processing Status', 'mxchat') . '</h4>'; | |
| 5531 | + | |
| 5532 | + // Add dismiss button for completed status | |
| 5533 | + if ($status['status'] === 'complete' || $status['status'] === 'error') { | |
| 5534 | + $html .= '<button type="button" class="mxchat-dismiss-button">' . esc_html__('Dismiss', 'mxchat') . '</button>'; | |
| 5535 | + } | |
| 5536 | + | |
| 5537 | + // Process Batch button for processing status | |
| 5538 | + if ($status['status'] === 'processing') { | |
| 5539 | + $html .= '<button type="button" class="mxchat-manual-batch-btn" | |
| 5540 | + data-process-type="pdf" | |
| 5541 | + data-url="' . esc_attr($pdf_url) . '"> | |
| 5542 | + ' . esc_html__('Process Batch', 'mxchat') . '</button>'; | |
| 5543 | + } | |
| 5544 | + | |
| 5545 | + // Add status badges | |
| 5546 | + if ($status['status'] === 'error') { | |
| 5547 | + $html .= '<span class="mxchat-status-badge mxchat-status-failed">' . esc_html__('Error', 'mxchat') . '</span>'; | |
| 5548 | + } elseif ($status['status'] === 'complete') { | |
| 5549 | + if ($status['failed_pages'] > 0) { | |
| 5550 | + $html .= '<span class="mxchat-status-badge mxchat-status-warning">' . | |
| 5551 | + sprintf(esc_html__('Completed with %d failures', 'mxchat'), $status['failed_pages']) . '</span>'; | |
| 5552 | + } else { | |
| 5553 | + $html .= '<span class="mxchat-status-badge mxchat-status-success">' . esc_html__('Complete', 'mxchat') . '</span>'; | |
| 5554 | + } | |
| 5555 | + } | |
| 5556 | + | |
| 5557 | + $html .= '</div>'; // End header | |
| 5558 | + | |
| 5559 | + // Progress bar | |
| 5560 | + $html .= '<div class="mxchat-progress-bar">'; | |
| 5561 | + $html .= '<div class="mxchat-progress-fill" style="width: ' . esc_attr($status['percentage']) . '%"></div>'; | |
| 5562 | + $html .= '</div>'; | |
| 5563 | + | |
| 5564 | + // Status details | |
| 5565 | + $html .= '<div class="mxchat-status-details">'; | |
| 5566 | + $html .= '<p>' . sprintf( | |
| 5567 | + esc_html__('Progress: %d of %d pages (%d%%)', 'mxchat'), | |
| 5568 | + $status['processed_pages'], | |
| 5569 | + $status['total_pages'], | |
| 5570 | + $status['percentage'] | |
| 5571 | + ) . '</p>'; | |
| 5572 | + | |
| 5573 | + // Show failed pages count if any | |
| 5574 | + if ($status['failed_pages'] > 0) { | |
| 5575 | + $html .= '<p><strong>' . esc_html__('Failed pages:', 'mxchat') . '</strong> ' . esc_html($status['failed_pages']) . '</p>'; | |
| 5576 | + } | |
| 5577 | + | |
| 5578 | + $html .= '<p><strong>' . esc_html__('Status:', 'mxchat') . '</strong> ' . esc_html(ucfirst($status['status'])) . '</p>'; | |
| 5579 | + $html .= '<p><strong>' . esc_html__('Last update:', 'mxchat') . '</strong> ' . esc_html($status['last_update']) . '</p>'; | |
| 5580 | + | |
| 5581 | + // Add completion summary if available AND it's an array | |
| 5582 | + if (isset($status['completion_summary']) && is_array($status['completion_summary']) && !empty($status['completion_summary'])) { | |
| 5583 | + $summary = $status['completion_summary']; | |
| 5584 | + $html .= '<div class="mxchat-completion-summary">'; | |
| 5585 | + $html .= '<h5>' . esc_html__('Processing Summary', 'mxchat') . '</h5>'; | |
| 5586 | + $html .= '<p><strong>' . esc_html__('Total Pages:', 'mxchat') . '</strong> ' . esc_html($summary['total_pages']) . '</p>'; | |
| 5587 | + $html .= '<p><strong>' . esc_html__('Successful:', 'mxchat') . '</strong> ' . esc_html($summary['successful_pages']) . '</p>'; | |
| 5588 | + $html .= '<p><strong>' . esc_html__('Failed:', 'mxchat') . '</strong> ' . esc_html($summary['failed_pages']) . '</p>'; | |
| 5589 | + $html .= '<p><strong>' . esc_html__('Completed:', 'mxchat') . '</strong> ' . esc_html($summary['completion_time']) . '</p>'; | |
| 5590 | + $html .= '</div>'; | |
| 5591 | + } | |
| 5592 | + | |
| 5593 | + // Add failed pages list if any AND it's an array | |
| 5594 | + if (isset($status['failed_pages_list']) && is_array($status['failed_pages_list']) && !empty($status['failed_pages_list'])) { | |
| 5595 | + $html .= $this->mxchat_render_failed_pages_list($status['failed_pages_list']); | |
| 5596 | + } | |
| 5597 | + | |
| 5598 | + // Add error message if any | |
| 5599 | + if (isset($status['error']) && !empty($status['error'])) { | |
| 5600 | + $html .= '<div class="mxchat-error-notice">'; | |
| 5601 | + $html .= '<p class="error">' . esc_html($status['error']) . '</p>'; | |
| 5602 | + $html .= '</div>'; | |
| 5603 | + } | |
| 5604 | + | |
| 5605 | + $html .= '</div>'; // End details | |
| 5606 | + $html .= '</div>'; // End card | |
| 5607 | + | |
| 5608 | + return $html; | |
| 5609 | +} | |
| 5610 | +/** | |
| 5611 | + * Render sitemap status card HTML | |
| 5612 | + */ | |
| 5613 | +private function mxchat_render_sitemap_status_card($status, $sitemap_url) { | |
| 5614 | + $html = '<div class="mxchat-status-card" data-card-type="sitemap">'; | |
| 5615 | + $html .= '<div class="mxchat-status-header">'; | |
| 5616 | + $html .= '<h4>' . esc_html__('Sitemap Processing Status', 'mxchat') . '</h4>'; | |
| 5617 | + | |
| 5618 | + // Add dismiss button for completed status | |
| 5619 | + if ($status['status'] === 'complete' || $status['status'] === 'error') { | |
| 5620 | + $html .= '<button type="button" class="mxchat-dismiss-button">' . esc_html__('Dismiss', 'mxchat') . '</button>'; | |
| 5621 | + } | |
| 5622 | + | |
| 5623 | + // Process Batch button for processing status | |
| 5624 | + if ($status['status'] === 'processing') { | |
| 5625 | + $html .= '<button type="button" class="mxchat-manual-batch-btn" | |
| 5626 | + data-process-type="sitemap" | |
| 5627 | + data-url="' . esc_attr($sitemap_url) . '"> | |
| 5628 | + ' . esc_html__('Process Batch', 'mxchat') . '</button>'; | |
| 5629 | + } | |
| 5630 | + | |
| 5631 | + // Add status badges | |
| 5632 | + if ($status['status'] === 'error') { | |
| 5633 | + $html .= '<span class="mxchat-status-badge mxchat-status-failed">' . esc_html__('Error', 'mxchat') . '</span>'; | |
| 5634 | + } elseif ($status['status'] === 'complete') { | |
| 5635 | + if ($status['failed_urls'] > 0) { | |
| 5636 | + $html .= '<span class="mxchat-status-badge mxchat-status-warning">' . | |
| 5637 | + sprintf(esc_html__('Completed with %d failures', 'mxchat'), $status['failed_urls']) . '</span>'; | |
| 5638 | + } else { | |
| 5639 | + $html .= '<span class="mxchat-status-badge mxchat-status-success">' . esc_html__('Complete', 'mxchat') . '</span>'; | |
| 5640 | + } | |
| 5641 | + } | |
| 5642 | + | |
| 5643 | + $html .= '</div>'; // End header | |
| 5644 | + | |
| 5645 | + // Progress bar | |
| 5646 | + $html .= '<div class="mxchat-progress-bar">'; | |
| 5647 | + $html .= '<div class="mxchat-progress-fill" style="width: ' . esc_attr($status['percentage']) . '%"></div>'; | |
| 5648 | + $html .= '</div>'; | |
| 5649 | + | |
| 5650 | + // Status details | |
| 5651 | + $html .= '<div class="mxchat-status-details">'; | |
| 5652 | + $html .= '<p>' . sprintf( | |
| 5653 | + esc_html__('Progress: %d of %d URLs (%d%%)', 'mxchat'), | |
| 5654 | + $status['processed_urls'], | |
| 5655 | + $status['total_urls'], | |
| 5656 | + $status['percentage'] | |
| 5657 | + ) . '</p>'; | |
| 5658 | + | |
| 5659 | + // Show failed URLs count if any | |
| 5660 | + if ($status['failed_urls'] > 0) { | |
| 5661 | + $html .= '<p><strong>' . esc_html__('Failed URLs:', 'mxchat') . '</strong> ' . esc_html($status['failed_urls']) . '</p>'; | |
| 5662 | + } | |
| 5663 | + | |
| 5664 | + $html .= '<p><strong>' . esc_html__('Status:', 'mxchat') . '</strong> ' . esc_html(ucfirst($status['status'])) . '</p>'; | |
| 5665 | + $html .= '<p><strong>' . esc_html__('Last update:', 'mxchat') . '</strong> ' . esc_html($status['last_update']) . '</p>'; | |
| 5666 | + | |
| 5667 | + // Add completion summary if available AND it's an array | |
| 5668 | + if (isset($status['completion_summary']) && is_array($status['completion_summary']) && !empty($status['completion_summary'])) { | |
| 5669 | + $summary = $status['completion_summary']; | |
| 5670 | + $html .= '<div class="mxchat-completion-summary">'; | |
| 5671 | + $html .= '<h5>' . esc_html__('Processing Summary', 'mxchat') . '</h5>'; | |
| 5672 | + $html .= '<p><strong>' . esc_html__('Total URLs:', 'mxchat') . '</strong> ' . esc_html($summary['total_urls']) . '</p>'; | |
| 5673 | + $html .= '<p><strong>' . esc_html__('Successful:', 'mxchat') . '</strong> ' . esc_html($summary['successful_urls']) . '</p>'; | |
| 5674 | + $html .= '<p><strong>' . esc_html__('Failed:', 'mxchat') . '</strong> ' . esc_html($summary['failed_urls']) . '</p>'; | |
| 5675 | + $html .= '<p><strong>' . esc_html__('Completed:', 'mxchat') . '</strong> ' . esc_html($summary['completion_time']) . '</p>'; | |
| 5676 | + $html .= '</div>'; | |
| 5677 | + } | |
| 5678 | + | |
| 5679 | + // Add error messages if any (but not the failed URLs list) | |
| 5680 | + if (!empty($status['error']) || !empty($status['last_error'])) { | |
| 5681 | + $html .= '<div class="mxchat-error-notice">'; | |
| 5682 | + | |
| 5683 | + if (!empty($status['error'])) { | |
| 5684 | + $html .= '<p class="error">' . esc_html($status['error']) . '</p>'; | |
| 5685 | + } | |
| 5686 | + | |
| 5687 | + if (!empty($status['last_error'])) { | |
| 5688 | + $html .= '<p class="last-error">' . esc_html__('Last error:', 'mxchat') . ' ' . esc_html($status['last_error']) . '</p>'; | |
| 5689 | + } | |
| 5690 | + | |
| 5691 | + $html .= '</div>'; | |
| 5692 | + } | |
| 5693 | + | |
| 5694 | + $html .= '</div>'; // End details | |
| 5695 | + $html .= '</div>'; // End card | |
| 5696 | + | |
| 5697 | + return $html; | |
| 5698 | +} | |
| 5699 | + | |
| 5700 | + | |
| 5701 | +/** | |
| 5702 | + * Render failed pages list | |
| 5703 | + */ | |
| 5704 | +private function mxchat_render_failed_pages_list($failed_pages_list) { | |
| 5705 | + // Validate that $failed_pages_list is an array and not empty | |
| 5706 | + if (!is_array($failed_pages_list) || empty($failed_pages_list)) { | |
| 5707 | + return ''; | |
| 5708 | + } | |
| 5709 | + | |
| 5710 | + $html = '<div class="mxchat-error-notice">'; | |
| 5711 | + $html .= '<div class="mxchat-failed-pages-container">'; | |
| 5712 | + $html .= '<h5>' . sprintf(esc_html__('Failed Pages (%d)', 'mxchat'), count($failed_pages_list)) . '</h5>'; | |
| 5713 | + $html .= '<details>'; | |
| 5714 | + $html .= '<summary>' . esc_html__('Show Failed Pages', 'mxchat') . '</summary>'; | |
| 5715 | + $html .= '<div class="mxchat-failed-pages-list">'; | |
| 5716 | + | |
| 5717 | + // Create table for failed pages | |
| 5718 | + $html .= '<table class="widefat striped">'; | |
| 5719 | + $html .= '<thead><tr>'; | |
| 5720 | + $html .= '<th>' . esc_html__('Page', 'mxchat') . '</th>'; | |
| 5721 | + $html .= '<th>' . esc_html__('Error', 'mxchat') . '</th>'; | |
| 5722 | + $html .= '<th>' . esc_html__('Retries', 'mxchat') . '</th>'; | |
| 5723 | + $html .= '<th>' . esc_html__('Time', 'mxchat') . '</th>'; | |
| 5724 | + $html .= '</tr></thead><tbody>'; | |
| 5725 | + | |
| 5726 | + // Sort failed pages by most recent | |
| 5727 | + $sorted_failed_pages = $failed_pages_list; | |
| 5728 | + usort($sorted_failed_pages, function($a, $b) { | |
| 5729 | + return ($b['time'] ?? 0) - ($a['time'] ?? 0); | |
| 5730 | + }); | |
| 5731 | + | |
| 5732 | + foreach ($sorted_failed_pages as $item) { | |
| 5733 | + // Ensure $item is an array before accessing its elements | |
| 5734 | + if (!is_array($item)) { | |
| 5735 | + continue; | |
| 5736 | + } | |
| 5737 | + | |
| 5738 | + $time_ago = isset($item['time']) ? human_time_diff($item['time'], current_time('timestamp')) . ' ' . esc_html__('ago', 'mxchat') : 'Unknown'; | |
| 5739 | + $html .= '<tr>'; | |
| 5740 | + $html .= '<td>' . esc_html__('Page', 'mxchat') . ' ' . esc_html($item['page'] ?? 'Unknown') . '</td>'; | |
| 5741 | + $html .= '<td style="word-break: break-word;">' . esc_html($item['error'] ?? 'Unknown error') . '</td>'; | |
| 5742 | + $html .= '<td>' . esc_html($item['retries'] ?? 'N/A') . '</td>'; | |
| 5743 | + $html .= '<td>' . esc_html($time_ago) . '</td>'; | |
| 5744 | + $html .= '</tr>'; | |
| 5745 | + } | |
| 5746 | + | |
| 5747 | + $html .= '</tbody></table>'; | |
| 5748 | + $html .= '</div></details></div></div>'; | |
| 5749 | + | |
| 5750 | + return $html; | |
| 5751 | +} | |
| 5752 | + | |
| 5753 | +/** | |
| 5754 | + * Render failed URLs list | |
| 5755 | + */ | |
| 5756 | +private function mxchat_render_failed_urls_list($failed_urls_list) { | |
| 5757 | + // Validate that $failed_urls_list is an array and not empty | |
| 5758 | + if (!is_array($failed_urls_list) || empty($failed_urls_list)) { | |
| 5759 | + return ''; | |
| 5760 | + } | |
| 5761 | + | |
| 5762 | + $html = '<div class="mxchat-failed-urls-container">'; | |
| 5763 | + $html .= '<h5>' . sprintf(esc_html__('Failed URLs (%d)', 'mxchat'), count($failed_urls_list)) . '</h5>'; | |
| 5764 | + $html .= '<details>'; | |
| 5765 | + $html .= '<summary>' . esc_html__('Show Failed URLs', 'mxchat') . '</summary>'; | |
| 5766 | + $html .= '<div class="mxchat-failed-urls-list">'; | |
| 5767 | + | |
| 5768 | + // Create table for failed URLs | |
| 5769 | + $html .= '<table class="widefat striped">'; | |
| 5770 | + $html .= '<thead><tr>'; | |
| 5771 | + $html .= '<th>' . esc_html__('URL', 'mxchat') . '</th>'; | |
| 5772 | + $html .= '<th>' . esc_html__('Error', 'mxchat') . '</th>'; | |
| 5773 | + $html .= '<th>' . esc_html__('Retries', 'mxchat') . '</th>'; | |
| 5774 | + $html .= '<th>' . esc_html__('Time', 'mxchat') . '</th>'; | |
| 5775 | + $html .= '</tr></thead><tbody>'; | |
| 5776 | + | |
| 5777 | + // Sort failed URLs by most recent | |
| 5778 | + $sorted_failed_urls = $failed_urls_list; | |
| 5779 | + usort($sorted_failed_urls, function($a, $b) { | |
| 5780 | + return ($b['time'] ?? 0) - ($a['time'] ?? 0); | |
| 5781 | + }); | |
| 5782 | + | |
| 5783 | + // Show up to 50 failed URLs | |
| 5784 | + $display_urls = array_slice($sorted_failed_urls, 0, 50); | |
| 5785 | + | |
| 5786 | + foreach ($display_urls as $item) { | |
| 5787 | + // Ensure $item is an array before accessing its elements | |
| 5788 | + if (!is_array($item)) { | |
| 5789 | + continue; | |
| 5790 | + } | |
| 5791 | + | |
| 5792 | + $url = $item['url'] ?? ''; | |
| 5793 | + $time_ago = isset($item['time']) ? human_time_diff($item['time'], current_time('timestamp')) . ' ' . esc_html__('ago', 'mxchat') : 'Unknown'; | |
| 5794 | + | |
| 5795 | + // Truncate URL for display | |
| 5796 | + $display_url = strlen($url) > 50 ? substr($url, 0, 47) . '...' : $url; | |
| 5797 | + | |
| 5798 | + $html .= '<tr>'; | |
| 5799 | + $html .= '<td style="word-break: break-all;">'; | |
| 5800 | + if (!empty($url)) { | |
| 5801 | + $html .= '<a href="' . esc_url($url) . '" target="_blank" rel="noopener noreferrer">' . esc_html($display_url) . '</a>'; | |
| 5802 | + } else { | |
| 5803 | + $html .= esc_html__('Unknown URL', 'mxchat'); | |
| 5804 | + } | |
| 5805 | + $html .= '</td>'; | |
| 5806 | + $html .= '<td style="word-break: break-word;">' . esc_html($item['error'] ?? 'Unknown error') . '</td>'; | |
| 5807 | + $html .= '<td>' . esc_html($item['retries'] ?? 'N/A') . '</td>'; | |
| 5808 | + $html .= '<td>' . esc_html($time_ago) . '</td>'; | |
| 5809 | + $html .= '</tr>'; | |
| 5810 | + } | |
| 5811 | + | |
| 5812 | + $html .= '</tbody></table>'; | |
| 5813 | + | |
| 5814 | + if (count($failed_urls_list) > 50) { | |
| 5815 | + $html .= '<div class="mxchat-failed-urls-more">+ ' . | |
| 5816 | + (count($failed_urls_list) - 50) . | |
| 5817 | + ' ' . esc_html__('more failed URLs not shown', 'mxchat') . '</div>'; | |
| 5818 | + } | |
| 5819 | + | |
| 5820 | + $html .= '</div></details></div>'; | |
| 5821 | + | |
| 5822 | + return $html; | |
| 5823 | +} | |
| 5824 | + | |
| 5825 | +/** | |
| 5826 | + * Get all ACF fields for a specific post, excluding any fields the user has disabled | |
| 5827 | + */ | |
| 5828 | +public function mxchat_get_acf_fields_for_post($post_id) { | |
| 5829 | + if (!function_exists('get_field_objects')) { | |
| 5830 | + return array(); | |
| 5831 | + } | |
| 5832 | + | |
| 5833 | + // Field OBJECTS, not get_fields(): exclusion matches on the field KEY | |
| 5834 | + // (unique per field) rather than the name (shared across groups — plan | |
| 5835 | + // 30e81f). ACF's own get_fields() is implemented as get_field_objects() | |
| 5836 | + // reduced to name => value, so the un-excluded reduction below is the | |
| 5837 | + // identical shape and order the previous get_fields() call produced. | |
| 5838 | + $field_objects = get_field_objects($post_id); | |
| 5839 | + if (!$field_objects || !is_array($field_objects)) { | |
| 5840 | + return array(); | |
| 5841 | + } | |
| 5842 | + | |
| 5843 | + $excluded_fields = get_option('mxchat_acf_excluded_fields', array()); | |
| 5844 | + if (!is_array($excluded_fields)) { | |
| 5845 | + $excluded_fields = array(); | |
| 5846 | + } | |
| 5847 | + | |
| 5848 | + $fields = array(); | |
| 5849 | + foreach ($field_objects as $field_name => $field_object) { | |
| 5850 | + if (!empty($excluded_fields)) { | |
| 5851 | + $field_key = isset($field_object['key']) ? $field_object['key'] : ''; | |
| 5852 | + // Legacy name entries stay honored: a stored name whose group was | |
| 5853 | + // inactive at migration time still excludes every field wearing it. | |
| 5854 | + if (in_array($field_key, $excluded_fields, true) || in_array($field_name, $excluded_fields, true)) { | |
| 5855 | + continue; | |
| 5856 | + } | |
| 5857 | + } | |
| 5858 | + $fields[$field_name] = isset($field_object['value']) ? $field_object['value'] : null; | |
| 5859 | + } | |
| 5860 | + | |
| 5861 | + return $fields; | |
| 5862 | +} | |
| 5863 | + | |
| 5864 | +/** | |
| 5865 | + * Get all registered ACF field groups and their fields for the settings UI | |
| 5866 | + */ | |
| 5867 | +public function mxchat_get_all_acf_fields() { | |
| 5868 | + if (!function_exists('acf_get_field_groups') || !function_exists('acf_get_fields')) { | |
| 5869 | + return array(); | |
| 5870 | + } | |
| 5871 | + | |
| 5872 | + // Keyed by GROUP KEY, not title (titles are not unique), and each field | |
| 5873 | + // entry carries its ACF field key — the unique identifier every toggle, | |
| 5874 | + // save, and index-time exclusion now runs on (plans 30e81f / bf57e0). | |
| 5875 | + $all_fields = array(); | |
| 5876 | + $field_groups = acf_get_field_groups(); | |
| 5877 | + | |
| 5878 | + if (!empty($field_groups)) { | |
| 5879 | + foreach ($field_groups as $group) { | |
| 5880 | + $group_fields = acf_get_fields($group['key']); | |
| 5881 | + if (!empty($group_fields)) { | |
| 5882 | + $entry = array( | |
| 5883 | + 'title' => $group['title'], | |
| 5884 | + 'fields' => array(), | |
| 5885 | + ); | |
| 5886 | + foreach ($group_fields as $field) { | |
| 5887 | + $entry['fields'][] = array( | |
| 5888 | + 'key' => $field['key'], | |
| 5889 | + 'name' => $field['name'], | |
| 5890 | + 'label' => $field['label'], | |
| 5891 | + 'type' => $field['type'] | |
| 5892 | + ); | |
| 5893 | + } | |
| 5894 | + $all_fields[$group['key']] = $entry; | |
| 5895 | + } | |
| 5896 | + } | |
| 5897 | + } | |
| 5898 | + | |
| 5899 | + return $all_fields; | |
| 5900 | +} | |
| 5901 | + | |
| 5902 | +/** | |
| 5903 | + * Get whitelisted custom post meta for a given post | |
| 5904 | + * This allows non-ACF meta fields (like OptionTree, theme meta boxes) to be included in embeddings | |
| 5905 | + */ | |
| 5906 | +public function mxchat_get_whitelisted_post_meta($post_id) { | |
| 5907 | + $whitelist = get_option('mxchat_custom_meta_whitelist', ''); | |
| 5908 | + | |
| 5909 | + if (empty($whitelist)) { | |
| 5910 | + return array(); | |
| 5911 | + } | |
| 5912 | + | |
| 5913 | + // Parse the whitelist - one meta key per line | |
| 5914 | + $meta_keys = array_filter(array_map('trim', explode("\n", $whitelist))); | |
| 5915 | + | |
| 5916 | + if (empty($meta_keys)) { | |
| 5917 | + return array(); | |
| 5918 | + } | |
| 5919 | + | |
| 5920 | + $result = array(); | |
| 5921 | + | |
| 5922 | + foreach ($meta_keys as $key) { | |
| 5923 | + // Skip empty keys | |
| 5924 | + if (empty($key)) { | |
| 5925 | + continue; | |
| 5926 | + } | |
| 5927 | + | |
| 5928 | + $value = get_post_meta($post_id, $key, true); | |
| 5929 | + | |
| 5930 | + // Only include non-empty string values | |
| 5931 | + if (!empty($value) && is_string($value)) { | |
| 5932 | + $result[$key] = $value; | |
| 5933 | + } elseif (!empty($value) && is_array($value)) { | |
| 5934 | + // Handle array values by joining them | |
| 5935 | + $flat_value = $this->mxchat_flatten_meta_array($value); | |
| 5936 | + if (!empty($flat_value)) { | |
| 5937 | + $result[$key] = $flat_value; | |
| 5938 | + } | |
| 5939 | + } | |
| 5940 | + } | |
| 5941 | + | |
| 5942 | + return $result; | |
| 5943 | +} | |
| 5944 | + | |
| 5945 | +/** | |
| 5946 | + * Flatten array meta values into a readable string | |
| 5947 | + */ | |
| 5948 | +private function mxchat_flatten_meta_array($array, $depth = 0) { | |
| 5949 | + if ($depth > 3) { | |
| 5950 | + return ''; // Prevent infinite recursion | |
| 5951 | + } | |
| 5952 | + | |
| 5953 | + $parts = array(); | |
| 5954 | + | |
| 5955 | + foreach ($array as $key => $value) { | |
| 5956 | + if (is_string($value) && !empty($value)) { | |
| 5957 | + $parts[] = $value; | |
| 5958 | + } elseif (is_array($value)) { | |
| 5959 | + $nested = $this->mxchat_flatten_meta_array($value, $depth + 1); | |
| 5960 | + if (!empty($nested)) { | |
| 5961 | + $parts[] = $nested; | |
| 5962 | + } | |
| 5963 | + } | |
| 5964 | + } | |
| 5965 | + | |
| 5966 | + return implode(', ', $parts); | |
| 5967 | +} | |
| 5968 | + | |
| 5969 | +/** | |
| 5970 | + * Format ACF field values for content extraction | |
| 5971 | + */ | |
| 5972 | +public function mxchat_format_acf_field_value($value, $field_name = '', $post_id = 0) { | |
| 5973 | + if (empty($value)) { | |
| 5974 | + return ''; | |
| 5975 | + } | |
| 5976 | + | |
| 5977 | + // Handle WP_Post objects first (THIS IS THE KEY FIX) | |
| 5978 | + if ($value instanceof WP_Post) { | |
| 5979 | + return $value->post_title ?: ''; | |
| 5980 | + } | |
| 5981 | + | |
| 5982 | + // Handle other WP objects | |
| 5983 | + if (is_object($value)) { | |
| 5984 | + if (isset($value->post_title)) { | |
| 5985 | + return $value->post_title; | |
| 5986 | + } elseif (isset($value->display_name)) { | |
| 5987 | + return $value->display_name; | |
| 5988 | + } elseif (isset($value->name)) { | |
| 5989 | + return $value->name; | |
| 5990 | + } elseif (method_exists($value, '__toString')) { | |
| 5991 | + try { | |
| 5992 | + return (string) $value; | |
| 5993 | + } catch (Exception $e) { | |
| 5994 | + return ''; | |
| 5995 | + } | |
| 5996 | + } | |
| 5997 | + // For any other objects, return empty string | |
| 5998 | + return ''; | |
| 5999 | + } | |
| 6000 | + | |
| 6001 | + // Handle different ACF field types | |
| 6002 | + if (is_array($value)) { | |
| 6003 | + // Check if it's an image/file field | |
| 6004 | + if (isset($value['url'])) { | |
| 6005 | + // Image field - return alt text, title, or caption | |
| 6006 | + if (!empty($value['alt'])) { | |
| 6007 | + return $value['alt']; | |
| 6008 | + } elseif (!empty($value['title'])) { | |
| 6009 | + return $value['title']; | |
| 6010 | + } elseif (!empty($value['caption'])) { | |
| 6011 | + return $value['caption']; | |
| 6012 | + } else { | |
| 6013 | + return ''; // Don't include just the URL | |
| 6014 | + } | |
| 6015 | + } | |
| 6016 | + | |
| 6017 | + // Check if it's a post object or relationship field | |
| 6018 | + if (isset($value['post_title'])) { | |
| 6019 | + return $value['post_title']; | |
| 6020 | + } | |
| 6021 | + | |
| 6022 | + // Check if it's a user field | |
| 6023 | + if (isset($value['display_name'])) { | |
| 6024 | + return $value['display_name']; | |
| 6025 | + } | |
| 6026 | + | |
| 6027 | + // Check if it's a taxonomy term | |
| 6028 | + if (isset($value['name']) && isset($value['taxonomy'])) { | |
| 6029 | + return $value['name']; | |
| 6030 | + } | |
| 6031 | + | |
| 6032 | + // Check if it's a select field with label | |
| 6033 | + if (isset($value['label'])) { | |
| 6034 | + return $value['label']; | |
| 6035 | + } | |
| 6036 | + | |
| 6037 | + // Check for repeater field or flexible content | |
| 6038 | + if (is_numeric(key($value))) { | |
| 6039 | + $sub_values = array(); | |
| 6040 | + foreach ($value as $sub_item) { | |
| 6041 | + if (is_array($sub_item)) { | |
| 6042 | + // For repeater/flexible content, extract text values | |
| 6043 | + $sub_text = $this->mxchat_extract_text_from_acf_array($sub_item); | |
| 6044 | + if (!empty($sub_text)) { | |
| 6045 | + $sub_values[] = $sub_text; | |
| 6046 | + } | |
| 6047 | + } elseif ($sub_item instanceof WP_Post) { | |
| 6048 | + // Handle WP_Post objects in arrays | |
| 6049 | + $sub_values[] = $sub_item->post_title ?: ''; | |
| 6050 | + } else { | |
| 6051 | + $sub_values[] = (string) $sub_item; | |
| 6052 | + } | |
| 6053 | + } | |
| 6054 | + return implode(', ', array_filter($sub_values)); | |
| 6055 | + } | |
| 6056 | + | |
| 6057 | + // For other arrays, try to extract meaningful text | |
| 6058 | + $text_values = array(); | |
| 6059 | + foreach ($value as $key => $val) { | |
| 6060 | + if (is_string($val) && !empty(trim($val))) { | |
| 6061 | + $text_values[] = trim($val); | |
| 6062 | + } elseif ($val instanceof WP_Post) { | |
| 6063 | + // Handle WP_Post objects in associative arrays | |
| 6064 | + $text_values[] = $val->post_title ?: ''; | |
| 6065 | + } elseif (is_array($val) && isset($val['post_title'])) { | |
| 6066 | + $text_values[] = $val['post_title']; | |
| 6067 | + } elseif (is_array($val) && isset($val['name'])) { | |
| 6068 | + $text_values[] = $val['name']; | |
| 6069 | + } | |
| 6070 | + } | |
| 6071 | + | |
| 6072 | + return implode(', ', array_filter($text_values)); | |
| 6073 | + } | |
| 6074 | + | |
| 6075 | + // Handle boolean values | |
| 6076 | + if (is_bool($value)) { | |
| 6077 | + return $value ? 'Yes' : 'No'; | |
| 6078 | + } | |
| 6079 | + | |
| 6080 | + // Handle numeric values | |
| 6081 | + if (is_numeric($value)) { | |
| 6082 | + return (string) $value; | |
| 6083 | + } | |
| 6084 | + | |
| 6085 | + // Handle string values | |
| 6086 | + if (is_string($value)) { | |
| 6087 | + return trim($value); | |
| 6088 | + } | |
| 6089 | + | |
| 6090 | + // For anything else that we can't handle, return empty string | |
| 6091 | + // This prevents the "Object could not be converted to string" error | |
| 6092 | + return ''; | |
| 6093 | +} | |
| 6094 | + | |
| 6095 | +/** | |
| 6096 | + * Extract text from complex ACF array structures | |
| 6097 | + */ | |
| 6098 | +private function mxchat_extract_text_from_acf_array($array) { | |
| 6099 | + if (!is_array($array)) { | |
| 6100 | + return ''; | |
| 6101 | + } | |
| 6102 | + | |
| 6103 | + $text_parts = array(); | |
| 6104 | + | |
| 6105 | + foreach ($array as $key => $value) { | |
| 6106 | + if (is_string($value) && !empty(trim($value))) { | |
| 6107 | + // Skip keys that are likely to be IDs or technical values | |
| 6108 | + if (!is_numeric($value) || strlen($value) > 10) { | |
| 6109 | + $text_parts[] = trim($value); | |
| 6110 | + } | |
| 6111 | + } elseif ($value instanceof WP_Post) { | |
| 6112 | + // Handle WP_Post objects | |
| 6113 | + $text_parts[] = $value->post_title ?: ''; | |
| 6114 | + } elseif (is_array($value)) { | |
| 6115 | + if (isset($value['post_title'])) { | |
| 6116 | + $text_parts[] = $value['post_title']; | |
| 6117 | + } elseif (isset($value['name'])) { | |
| 6118 | + $text_parts[] = $value['name']; | |
| 6119 | + } elseif (isset($value['label'])) { | |
| 6120 | + $text_parts[] = $value['label']; | |
| 6121 | + } | |
| 6122 | + } elseif (is_object($value)) { | |
| 6123 | + // Handle other objects safely | |
| 6124 | + if (isset($value->post_title)) { | |
| 6125 | + $text_parts[] = $value->post_title; | |
| 6126 | + } elseif (isset($value->name)) { | |
| 6127 | + $text_parts[] = $value->name; | |
| 6128 | + } elseif (isset($value->display_name)) { | |
| 6129 | + $text_parts[] = $value->display_name; | |
| 6130 | + } | |
| 6131 | + } | |
| 6132 | + } | |
| 6133 | + | |
| 6134 | + return implode(', ', array_filter($text_parts)); | |
| 6135 | +} | |
| 6136 | + | |
| 6137 | +/** | |
| 6138 | + * Walk an ACF field value tree and collect attachment IDs for any value that | |
| 6139 | + * resolves to a PDF in the WordPress media library. Handles the three shapes | |
| 6140 | + * ACF returns for File/Image/URL fields (array with ID+url, integer attachment ID, | |
| 6141 | + * plain URL string), and recurses through repeater/group/flexible content. | |
| 6142 | + * | |
| 6143 | + * @param mixed $value The ACF field value (any depth) | |
| 6144 | + * @param array $out Accumulator (passed by reference) for attachment IDs | |
| 6145 | + * @param int $depth Recursion guard | |
| 6146 | + */ | |
| 6147 | +private function mxchat_collect_pdf_attachment_ids_from_acf_value($value, &$out, $depth = 0) { | |
| 6148 | + if ($depth > 6) { | |
| 6149 | + return; // prevent runaway recursion on circular/very-deep structures | |
| 6150 | + } | |
| 6151 | + | |
| 6152 | + if (empty($value)) { | |
| 6153 | + return; | |
| 6154 | + } | |
| 6155 | + | |
| 6156 | + // Array shapes: ACF File/Image return value=array; repeaters/groups are arrays of arrays | |
| 6157 | + if (is_array($value)) { | |
| 6158 | + // Direct File/Image-style array (has 'url' and usually 'ID' + 'mime_type') | |
| 6159 | + $looks_like_attachment = isset($value['url']) || isset($value['ID']) || isset($value['id']); | |
| 6160 | + if ($looks_like_attachment) { | |
| 6161 | + $att_id = 0; | |
| 6162 | + if (!empty($value['ID']) && is_numeric($value['ID'])) { | |
| 6163 | + $att_id = (int) $value['ID']; | |
| 6164 | + } elseif (!empty($value['id']) && is_numeric($value['id'])) { | |
| 6165 | + $att_id = (int) $value['id']; | |
| 6166 | + } elseif (!empty($value['url']) && is_string($value['url'])) { | |
| 6167 | + $att_id = (int) attachment_url_to_postid($value['url']); | |
| 6168 | + } | |
| 6169 | + | |
| 6170 | + $is_pdf = false; | |
| 6171 | + if (!empty($value['mime_type']) && $value['mime_type'] === 'application/pdf') { | |
| 6172 | + $is_pdf = true; | |
| 6173 | + } elseif (!empty($value['subtype']) && strtolower((string) $value['subtype']) === 'pdf') { | |
| 6174 | + $is_pdf = true; | |
| 6175 | + } elseif (!empty($value['url']) && is_string($value['url']) && $this->mxchat_url_looks_like_pdf($value['url'])) { | |
| 6176 | + $is_pdf = true; | |
| 6177 | + } elseif ($att_id && get_post_mime_type($att_id) === 'application/pdf') { | |
| 6178 | + $is_pdf = true; | |
| 6179 | + } | |
| 6180 | + | |
| 6181 | + if ($is_pdf && $att_id && get_post_mime_type($att_id) === 'application/pdf') { | |
| 6182 | + $out[] = $att_id; | |
| 6183 | + } | |
| 6184 | + // An array node that represents one attachment doesn't contain other | |
| 6185 | + // attachments inside it — done with this branch. | |
| 6186 | + return; | |
| 6187 | + } | |
| 6188 | + | |
| 6189 | + // Recurse: repeater rows, flexible-content layouts, groups, etc. | |
| 6190 | + foreach ($value as $sub) { | |
| 6191 | + $this->mxchat_collect_pdf_attachment_ids_from_acf_value($sub, $out, $depth + 1); | |
| 6192 | + } | |
| 6193 | + return; | |
| 6194 | + } | |
| 6195 | + | |
| 6196 | + // Plain numeric attachment ID (ACF File field set to "Return: ID") | |
| 6197 | + if (is_numeric($value)) { | |
| 6198 | + $att_id = (int) $value; | |
| 6199 | + if ($att_id > 0 && get_post_mime_type($att_id) === 'application/pdf') { | |
| 6200 | + $out[] = $att_id; | |
| 6201 | + } | |
| 6202 | + return; | |
| 6203 | + } | |
| 6204 | + | |
| 6205 | + // Plain string — URL pointing at a PDF (ACF File field set to "Return: URL", or a custom URL/text field) | |
| 6206 | + if (is_string($value)) { | |
| 6207 | + $trimmed = trim($value); | |
| 6208 | + if ($trimmed !== '' && $this->mxchat_url_looks_like_pdf($trimmed)) { | |
| 6209 | + $att_id = (int) attachment_url_to_postid($trimmed); | |
| 6210 | + if ($att_id > 0 && get_post_mime_type($att_id) === 'application/pdf') { | |
| 6211 | + $out[] = $att_id; | |
| 6212 | + } | |
| 6213 | + } | |
| 6214 | + return; | |
| 6215 | + } | |
| 6216 | +} | |
| 6217 | + | |
| 6218 | +/** | |
| 6219 | + * Heuristic: does this URL/string look like a PDF reference? | |
| 6220 | + * Tolerates query strings and fragments (#page=2). | |
| 6221 | + */ | |
| 6222 | +private function mxchat_url_looks_like_pdf($url) { | |
| 6223 | + if (!is_string($url) || $url === '') { | |
| 6224 | + return false; | |
| 6225 | + } | |
| 6226 | + // Strip query + fragment before checking extension | |
| 6227 | + $path = preg_replace('/[?#].*$/', '', $url); | |
| 6228 | + return (bool) preg_match('/\.pdf$/i', $path); | |
| 6229 | +} | |
| 6230 | + | |
| 6231 | +/** | |
| 6232 | + * Extract text from a PDF attachment by ID using the bundled Smalot parser. | |
| 6233 | + * Reads the file directly from disk via get_attached_file (no HTTP fetch). | |
| 6234 | + * Result is cached on the attachment as post_meta keyed by file mtime so we | |
| 6235 | + * only parse the same PDF once unless the file changes on disk. | |
| 6236 | + * | |
| 6237 | + * @param int $attachment_id | |
| 6238 | + * @return string Extracted plain text, or '' on failure. | |
| 6239 | + */ | |
| 6240 | +private function mxchat_extract_pdf_text_by_attachment_id($attachment_id) { | |
| 6241 | + $attachment_id = (int) $attachment_id; | |
| 6242 | + if ($attachment_id <= 0) { | |
| 6243 | + return ''; | |
| 6244 | + } | |
| 6245 | + if (get_post_mime_type($attachment_id) !== 'application/pdf') { | |
| 6246 | + return ''; | |
| 6247 | + } | |
| 6248 | + | |
| 6249 | + $pdf_path = get_attached_file($attachment_id); | |
| 6250 | + if (empty($pdf_path) || !file_exists($pdf_path) || !is_readable($pdf_path)) { | |
| 6251 | + return ''; | |
| 6252 | + } | |
| 6253 | + | |
| 6254 | + // Raw-file size cap. Parsing very large PDFs can OOM the request; skip with a log entry | |
| 6255 | + // and let the rest of the ACF content land in the KB. Filterable for users who need it bigger. | |
| 6256 | + $default_max_bytes = 25 * 1024 * 1024; | |
| 6257 | + $max_bytes = (int) apply_filters('mxchat_acf_pdf_max_bytes', $default_max_bytes, $attachment_id, $pdf_path); | |
| 6258 | + if ($max_bytes > 0) { | |
| 6259 | + $file_size = @filesize($pdf_path); | |
| 6260 | + if ($file_size !== false && $file_size > $max_bytes) { | |
| 6261 | + error_log(sprintf( | |
| 6262 | + '[mxchat] ACF PDF skipped (over size cap): attachment %d "%s" %d bytes > cap %d', | |
| 6263 | + $attachment_id, | |
| 6264 | + basename($pdf_path), | |
| 6265 | + $file_size, | |
| 6266 | + $max_bytes | |
| 6267 | + )); | |
| 6268 | + return ''; | |
| 6269 | + } | |
| 6270 | + } | |
| 6271 | + | |
| 6272 | + $mtime = @filemtime($pdf_path); | |
| 6273 | + $cache_meta_key = '_mxchat_acf_pdf_text_v1'; | |
| 6274 | + $cached = get_post_meta($attachment_id, $cache_meta_key, true); | |
| 6275 | + if (is_array($cached) && isset($cached['mtime'], $cached['text']) && (int) $cached['mtime'] === (int) $mtime) { | |
| 6276 | + return (string) $cached['text']; | |
| 6277 | + } | |
| 6278 | + | |
| 6279 | + $text = ''; | |
| 6280 | + try { | |
| 6281 | + if (function_exists('mxchat_load_pdf_parser')) { | |
| 6282 | + mxchat_load_pdf_parser(); | |
| 6283 | + } | |
| 6284 | + if (!class_exists('\\Smalot\\PdfParser\\Parser')) { | |
| 6285 | + return ''; | |
| 6286 | + } | |
| 6287 | + $parser = new \Smalot\PdfParser\Parser(); | |
| 6288 | + $pdf = $parser->parseFile($pdf_path); | |
| 6289 | + $pages = $pdf->getPages(); | |
| 6290 | + $page_texts = array(); | |
| 6291 | + $acf_page_num = 0; | |
| 6292 | + foreach ($pages as $page) { | |
| 6293 | + $acf_page_num++; | |
| 6294 | + $page_text = ''; | |
| 6295 | + try { | |
| 6296 | + $page_text = $page->getText(); | |
| 6297 | + } catch (\Exception $e) { | |
| 6298 | + $page_text = ''; | |
| 6299 | + } | |
| 6300 | + if (!empty($page_text)) { | |
| 6301 | + $page_text = MxChat_Utils::normalize_pdf_rtl($page_text, 'acf_pdf attachment ' . $attachment_id . ' page ' . $acf_page_num); | |
| 6302 | + $page_texts[] = $page_text; | |
| 6303 | + } | |
| 6304 | + } | |
| 6305 | + $text = trim(implode("\n\n", $page_texts)); | |
| 6306 | + } catch (\Exception $e) { | |
| 6307 | + error_log('[mxchat] ACF PDF extraction failed for attachment ' . $attachment_id . ': ' . $e->getMessage()); | |
| 6308 | + return ''; | |
| 6309 | + } catch (\Throwable $e) { | |
| 6310 | + error_log('[mxchat] ACF PDF extraction error for attachment ' . $attachment_id . ': ' . $e->getMessage()); | |
| 6311 | + return ''; | |
| 6312 | + } | |
| 6313 | + | |
| 6314 | + // Cap per-PDF text to avoid blowing up the embedding payload on enormous PDFs. | |
| 6315 | + // The chunker downstream will still split this into multiple vectors. | |
| 6316 | + $max_len = (int) apply_filters('mxchat_acf_pdf_text_max_length', 50000); | |
| 6317 | + if ($max_len > 0 && strlen($text) > $max_len) { | |
| 6318 | + $text = substr($text, 0, $max_len); | |
| 6319 | + } | |
| 6320 | + | |
| 6321 | + update_post_meta($attachment_id, $cache_meta_key, array( | |
| 6322 | + 'mtime' => (int) $mtime, | |
| 6323 | + 'text' => $text, | |
| 6324 | + )); | |
| 6325 | + | |
| 6326 | + return $text; | |
| 6327 | +} | |
| 6328 | + | |
| 6329 | +/** | |
| 6330 | + * The media-library MIME types the knowledge base can actually read (plan 01010a). | |
| 6331 | + * | |
| 6332 | + * The importer's post-type dropdown is built from get_post_types(['public' => true]), | |
| 6333 | + * and 'attachment' is public — so Media has always been OFFERED as a source. What it | |
| 6334 | + * could never do is return anything, and the fix is deliberately narrower than the | |
| 6335 | + * offer: only files one of the THREE extractors already shipping in this plugin can | |
| 6336 | + * turn into text belong on that list. Images, audio and video are not listed at all | |
| 6337 | + * rather than listed and greyed out, because "nothing to import here" is a clearer | |
| 6338 | + * answer than a row you cannot select. | |
| 6339 | + * | |
| 6340 | + * text/markdown has no entry in WordPress's default wp_get_mime_types(), so a .md | |
| 6341 | + * file is only ever in the library on a site that added the mapping itself; it stays | |
| 6342 | + * on the list for those sites and costs nothing on the rest. | |
| 6343 | + * | |
| 6344 | + * @return array MIME type strings. | |
| 6345 | + */ | |
| 6346 | +public static function mxchat_supported_attachment_mime_types() { | |
| 6347 | + /** | |
| 6348 | + * Filter the attachment MIME types offered by the WordPress Content importer. | |
| 6349 | + * | |
| 6350 | + * Adding a type here makes it listable and importable — mxchat_extract_attachment_text() | |
| 6351 | + * reads anything text/* as plain text, so a text-shaped MIME works out of the box. | |
| 6352 | + * A binary format needs its own extractor. | |
| 6353 | + * | |
| 6354 | + * @param array $mime_types Supported MIME types. | |
| 6355 | + */ | |
| 6356 | + return (array) apply_filters('mxchat_kb_attachment_mime_types', array( | |
| 6357 | + 'application/pdf', | |
| 6358 | + 'application/vnd.openxmlformats-officedocument.wordprocessingml.document', | |
| 6359 | + 'text/plain', | |
| 6360 | + 'text/markdown', | |
| 6361 | + )); | |
| 6362 | +} | |
| 6363 | + | |
| 6364 | +/** | |
| 6365 | + * Short human label for a supported attachment MIME type ("PDF", "Word", ...). | |
| 6366 | + * Used in the importer list, where "attachment" as a type and "0 words" as a size | |
| 6367 | + * would tell the user nothing about what they are about to import. | |
| 6368 | + */ | |
| 6369 | +public static function mxchat_attachment_type_label($mime_type) { | |
| 6370 | + $labels = array( | |
| 6371 | + 'application/pdf' => __('PDF', 'mxchat'), | |
| 6372 | + 'application/vnd.openxmlformats-officedocument.wordprocessingml.document' => __('Word', 'mxchat'), | |
| 6373 | + 'text/plain' => __('Text', 'mxchat'), | |
| 6374 | + 'text/markdown' => __('Markdown', 'mxchat'), | |
| 6375 | + ); | |
| 6376 | + | |
| 6377 | + return isset($labels[$mime_type]) ? $labels[$mime_type] : __('Document', 'mxchat'); | |
| 6378 | +} | |
| 6379 | + | |
| 6380 | +/** | |
| 6381 | + * Extract indexable text from a media-library attachment (plan 01010a). | |
| 6382 | + * | |
| 6383 | + * Wiring, not parsing: every extractor here already ships and is already used | |
| 6384 | + * elsewhere — the Smalot path via mxchat_extract_pdf_text_by_attachment_id() (ACF | |
| 6385 | + * PDFs), MXChat_Word_Handler::extract_docx_text() and the plain-text read (both from | |
| 6386 | + * Document Upload, which is also where the character-estimated page cap comes from). | |
| 6387 | + * No new parser was added for this. | |
| 6388 | + * | |
| 6389 | + * Returns a reason instead of an empty string on failure so the caller can say WHY a | |
| 6390 | + * file was skipped; an attachment that yields no text must never be stored as an | |
| 6391 | + * empty entry. | |
| 6392 | + * | |
| 6393 | + * @param int $attachment_id | |
| 6394 | + * @return array{text: string, reason: string} reason is '' on success. | |
| 6395 | + */ | |
| 6396 | +private function mxchat_extract_attachment_text($attachment_id) { | |
| 6397 | + $attachment_id = (int) $attachment_id; | |
| 6398 | + $mime_type = get_post_mime_type($attachment_id); | |
| 6399 | + | |
| 6400 | + if (!in_array($mime_type, self::mxchat_supported_attachment_mime_types(), true)) { | |
| 6401 | + return array( | |
| 6402 | + 'text' => '', | |
| 6403 | + 'reason' => __('Unsupported file type — only PDF, Word, text and Markdown files can be imported.', 'mxchat'), | |
| 6404 | + ); | |
| 6405 | + } | |
| 6406 | + | |
| 6407 | + // PDF: the shared extractor owns its own size cap (25 MB raw, filterable), its | |
| 6408 | + // 50,000-character output cap, the RTL normaliser and the mtime-keyed cache, so | |
| 6409 | + // re-importing the same PDF does not re-parse it. | |
| 6410 | + if ($mime_type === 'application/pdf') { | |
| 6411 | + $text = $this->mxchat_extract_pdf_text_by_attachment_id($attachment_id); | |
| 6412 | + $text = trim((string) $text); | |
| 6413 | + | |
| 6414 | + if ($text === '') { | |
| 6415 | + return array( | |
| 6416 | + 'text' => '', | |
| 6417 | + 'reason' => __('No extractable text — the PDF is empty, scanned images only, or larger than the size limit.', 'mxchat'), | |
| 6418 | + ); | |
| 6419 | + } | |
| 6420 | + | |
| 6421 | + return array('text' => $text, 'reason' => ''); | |
| 6422 | + } | |
| 6423 | + | |
| 6424 | + $file_path = get_attached_file($attachment_id); | |
| 6425 | + if (empty($file_path) || !file_exists($file_path) || !is_readable($file_path)) { | |
| 6426 | + return array( | |
| 6427 | + 'text' => '', | |
| 6428 | + 'reason' => __('The file could not be read from disk.', 'mxchat'), | |
| 6429 | + ); | |
| 6430 | + } | |
| 6431 | + | |
| 6432 | + if ($mime_type === 'application/vnd.openxmlformats-officedocument.wordprocessingml.document') { | |
| 6433 | + $text = MXChat_Word_Handler::extract_docx_text($file_path); | |
| 6434 | + if ($text === false) { | |
| 6435 | + return array( | |
| 6436 | + 'text' => '', | |
| 6437 | + 'reason' => __('The Word file could not be read. It may be corrupt, empty, or not a real Word document.', 'mxchat'), | |
| 6438 | + ); | |
| 6439 | + } | |
| 6440 | + $text = trim((string) $text); | |
| 6441 | + } else { | |
| 6442 | + // .txt / .md read as-is, the same way Document Upload reads them — | |
| 6443 | + // Markdown deliberately keeps its syntax, because headings are useful | |
| 6444 | + // retrieval signal. | |
| 6445 | + $text = (string) file_get_contents($file_path); | |
| 6446 | + $text = wp_check_invalid_utf8($text); | |
| 6447 | + $text = trim($text); | |
| 6448 | + } | |
| 6449 | + | |
| 6450 | + if ($text === '') { | |
| 6451 | + return array( | |
| 6452 | + 'text' => '', | |
| 6453 | + 'reason' => __('The file contains no readable text.', 'mxchat'), | |
| 6454 | + ); | |
| 6455 | + } | |
| 6456 | + | |
| 6457 | + // Same cap Document Upload applies to the same three formats: the pdf_max_pages | |
| 6458 | + // setting, estimated by CHARACTERS (~2500/page) because the .docx cleaner | |
| 6459 | + // collapses newlines and a paragraph count reads 1 for any Word file. Importing | |
| 6460 | + // is synchronous, so an unbounded document risks a timeout. | |
| 6461 | + $options = get_option('mxchat_options', array()); | |
| 6462 | + $max_pages = isset($options['pdf_max_pages']) ? intval($options['pdf_max_pages']) : 69; | |
| 6463 | + $estimated_pages = (int) ceil(strlen($text) / 2500); | |
| 6464 | + if ($max_pages > 0 && $estimated_pages > $max_pages) { | |
| 6465 | + return array( | |
| 6466 | + 'text' => '', | |
| 6467 | + 'reason' => sprintf( | |
| 6468 | + /* translators: 1: estimated page count, 2: configured page limit */ | |
| 6469 | + __('Too large (about %1$d pages; the limit is %2$d). Split the file, or raise the PDF max pages setting.', 'mxchat'), | |
| 6470 | + $estimated_pages, | |
| 6471 | + $max_pages | |
| 6472 | + ), | |
| 6473 | + ); | |
| 6474 | + } | |
| 6475 | + | |
| 6476 | + return array('text' => $text, 'reason' => ''); | |
| 6477 | +} | |
| 6478 | + | |
| 6479 | +/** | |
| 6480 | + * Assemble the stored text for an attachment: title first, then the extracted body. | |
| 6481 | + * Mirrors the post assembler's shape (title on its own line, blank line, content) so | |
| 6482 | + * a media entry retrieves on its filename/title the same way a post does. | |
| 6483 | + */ | |
| 6484 | +private function mxchat_build_attachment_content($post, $extracted_text) { | |
| 6485 | + $title = $this->mxchat_decode_entities_for_indexing($post->post_title); | |
| 6486 | + $parts = array($title); | |
| 6487 | + | |
| 6488 | + // The media modal's Caption and Description fields are real, editable metadata | |
| 6489 | + // and are frequently the only human-written context a file has. | |
| 6490 | + $caption = trim(wp_strip_all_tags((string) $post->post_excerpt)); | |
| 6491 | + if ($caption !== '') { | |
| 6492 | + $parts[] = $this->mxchat_decode_entities_for_indexing($caption); | |
| 6493 | + } | |
| 6494 | + | |
| 6495 | + $description = trim(wp_strip_all_tags((string) $post->post_content)); | |
| 6496 | + if ($description !== '') { | |
| 6497 | + $parts[] = $this->mxchat_decode_entities_for_indexing($description); | |
| 6498 | + } | |
| 6499 | + | |
| 6500 | + $parts[] = $extracted_text; | |
| 6501 | + | |
| 6502 | + return implode("\n\n", array_filter($parts, function ($part) { | |
| 6503 | + return $part !== ''; | |
| 6504 | + })); | |
| 6505 | +} | |
| 6506 | + | |
| 6507 | +/** | |
| 6508 | + * Import one media-library attachment into the knowledge base — the attachment | |
| 6509 | + * branch of ajax_mxchat_process_selected_content (plan 01010a). | |
| 6510 | + * | |
| 6511 | + * Separate from the post path rather than folded into it because almost every step | |
| 6512 | + * differs: the source URL is the FILE url (get_permalink() on an attachment returns | |
| 6513 | + * its attachment page, which is not what anyone means by "where this document | |
| 6514 | + * lives"), the body comes from an extractor instead of post_content, and a file that | |
| 6515 | + * yields no text has to be refused with a reason rather than stored empty. | |
| 6516 | + * | |
| 6517 | + * Sends its own JSON response and exits, exactly like its caller. | |
| 6518 | + */ | |
| 6519 | +private function mxchat_import_attachment($post, $bot_id) { | |
| 6520 | + $post_id = (int) $post->ID; | |
| 6521 | + | |
| 6522 | + $extracted = $this->mxchat_extract_attachment_text($post_id); | |
| 6523 | + if ($extracted['text'] === '') { | |
| 6524 | + wp_send_json_error(sprintf( | |
| 6525 | + /* translators: 1: attachment title, 2: reason the file was skipped */ | |
| 6526 | + __('Skipped "%1$s": %2$s', 'mxchat'), | |
| 6527 | + $post->post_title, | |
| 6528 | + $extracted['reason'] | |
| 6529 | + )); | |
| 6530 | + exit; | |
| 6531 | + } | |
| 6532 | + | |
| 6533 | + $bot_options = $this->get_bot_options($bot_id); | |
| 6534 | + $options = !empty($bot_options) ? $bot_options : get_option('mxchat_options'); | |
| 6535 | + | |
| 6536 | + $preflight = MxChat_Utils::embedding_preflight($options); | |
| 6537 | + if (!$preflight['ok']) { | |
| 6538 | + MxChat_Admin::mxchat_log_debug('api_error', $preflight['reason'] . ' (attachment import)'); | |
| 6539 | + wp_send_json_error($preflight['reason']); | |
| 6540 | + exit; | |
| 6541 | + } | |
| 6542 | + | |
| 6543 | + $source_url = wp_get_attachment_url($post_id); | |
| 6544 | + if (empty($source_url)) { | |
| 6545 | + wp_send_json_error(__('The file has no URL — it may be missing from the uploads folder.', 'mxchat')); | |
| 6546 | + exit; | |
| 6547 | + } | |
| 6548 | + | |
| 6549 | + $vector_id = md5($source_url); | |
| 6550 | + $content = $this->mxchat_build_attachment_content($post, $extracted['text']); | |
| 6551 | + | |
| 6552 | + // "Already imported" read, same two-backend split the post path uses. | |
| 6553 | + $is_update = false; | |
| 6554 | + $bot_pinecone_config = $this->get_bot_pinecone_config($bot_id); | |
| 6555 | + $use_pinecone = !empty($bot_pinecone_config) && ($bot_pinecone_config['use_pinecone'] ?? false); | |
| 6556 | + | |
| 6557 | + if ($use_pinecone && !empty($bot_pinecone_config['api_key'])) { | |
| 6558 | + $pinecone_data = $this->mxchat_get_pinecone_processed_content($bot_pinecone_config); | |
| 6559 | + if (isset($pinecone_data[$post_id])) { | |
| 6560 | + $is_update = true; | |
| 6561 | + } | |
| 6562 | + } else { | |
| 6563 | + global $wpdb; | |
| 6564 | + $table_name = $wpdb->prefix . 'mxchat_system_prompt_content'; | |
| 6565 | + $existing_record = $wpdb->get_row($wpdb->prepare( | |
| 6566 | + "SELECT id FROM $table_name WHERE source_url = %s", | |
| 6567 | + $source_url | |
| 6568 | + )); | |
| 6569 | + if ($existing_record) { | |
| 6570 | + $is_update = true; | |
| 6571 | + } | |
| 6572 | + } | |
| 6573 | + | |
| 6574 | + // content_type 'attachment' — NOT 'document'. The Knowledge screen's type filter | |
| 6575 | + // is built from public post types, so 'attachment' is exactly what its "Media" | |
| 6576 | + // option filters on; 'document' (Document Upload's label) would file these under | |
| 6577 | + // an option that dropdown does not offer. | |
| 6578 | + $result = MxChat_Utils::submit_content_to_db( | |
| 6579 | + $content, | |
| 6580 | + $source_url, | |
| 6581 | + $preflight['api_key'], | |
| 6582 | + $vector_id, | |
| 6583 | + $bot_id, | |
| 6584 | + 'attachment' | |
| 6585 | + ); | |
| 6586 | + | |
| 6587 | + if (is_wp_error($result)) { | |
| 6588 | + MxChat_Admin::mxchat_log_debug('storage_error', 'Attachment storage failed: ' . $result->get_error_message(), array('source_url' => $source_url)); | |
| 6589 | + wp_send_json_error('Storage failed: ' . $result->get_error_message()); | |
| 6590 | + exit; | |
| 6591 | + } | |
| 6592 | + | |
| 6593 | + // Attachments carry no tags, so the tag half of this is always a no-op — the | |
| 6594 | + // per-entry dropdown and the MxChat meta box on the attachment's own edit screen | |
| 6595 | + // are what restrict a media entry (plan 6ab002). | |
| 6596 | + $this->apply_role_restriction_to_post($post_id, $source_url); | |
| 6597 | + | |
| 6598 | + $this->mxchat_remember_attachment_signature($post_id, $source_url, $content); | |
| 6599 | + | |
| 6600 | + wp_send_json_success(array( | |
| 6601 | + 'message' => $is_update ? 'Content updated successfully' : 'Content processed successfully', | |
| 6602 | + 'post_id' => $post_id, | |
| 6603 | + 'title' => $post->post_title, | |
| 6604 | + 'operation_type' => $is_update ? 'update' : 'new', | |
| 6605 | + 'vector_id' => $vector_id, | |
| 6606 | + 'acf_fields_found' => 0, | |
| 6607 | + 'pdf_extracted_count' => (get_post_mime_type($post_id) === 'application/pdf') ? 1 : 0, | |
| 6608 | + 'content_preview' => substr($content, 0, 100) . '...', | |
| 6609 | + 'bot_id' => $bot_id | |
| 6610 | + )); | |
| 6611 | + exit; | |
| 6612 | +} | |
| 6613 | + | |
| 6614 | +/** | |
| 6615 | + * Is media auto-sync switched on? The toggle is the generic per-post-type one the | |
| 6616 | + * Advanced Custom Post Sync Settings list already renders for Media (attachment is a | |
| 6617 | + * public post type), so there is no new setting and no new UI for it. | |
| 6618 | + */ | |
| 6619 | +private function mxchat_attachment_auto_sync_enabled() { | |
| 6620 | + return get_option('mxchat_auto_sync_attachment') === '1'; | |
| 6621 | +} | |
| 6622 | + | |
| 6623 | +/** | |
| 6624 | + * Record what we last indexed for an attachment: the URL it was stored under and a | |
| 6625 | + * hash of the stored text. | |
| 6626 | + * | |
| 6627 | + * This exists because 'edit_attachment' is noisy — it fires on every metadata touch, | |
| 6628 | + * including ones that change nothing we index (alt text, regenerating thumbnails). | |
| 6629 | + * Without a signature, each of those would pay for a fresh embedding call. The stored | |
| 6630 | + * URL is also what lets a file REPLACEMENT clean up after itself: the old URL's | |
| 6631 | + * vectors are deleted before the new ones are written, instead of being orphaned. | |
| 6632 | + */ | |
| 6633 | +private function mxchat_remember_attachment_signature($attachment_id, $source_url, $content) { | |
| 6634 | + update_post_meta((int) $attachment_id, '_mxchat_kb_attachment_sig', array( | |
| 6635 | + 'url' => (string) $source_url, | |
| 6636 | + 'hash' => md5((string) $content), | |
| 6637 | + )); | |
| 6638 | +} | |
| 6639 | + | |
| 6640 | +/** | |
| 6641 | + * Auto-sync an attachment on upload or edit (add_attachment / edit_attachment). | |
| 6642 | + * | |
| 6643 | + * Silent by design, like the post auto-sync path: a media upload is not a knowledge | |
| 6644 | + * base action from the user's point of view, so a file we cannot read is skipped | |
| 6645 | + * without an admin notice. The manual importer is where a skip gets a visible reason. | |
| 6646 | + */ | |
| 6647 | +public function mxchat_handle_attachment_change($attachment_id) { | |
| 6648 | + $attachment_id = (int) $attachment_id; | |
| 6649 | + | |
| 6650 | + if (!$this->mxchat_attachment_auto_sync_enabled()) { | |
| 6651 | + return; | |
| 6652 | + } | |
| 6653 | + | |
| 6654 | + $post = get_post($attachment_id); | |
| 6655 | + if (!$post || $post->post_type !== 'attachment') { | |
| 6656 | + return; | |
| 6657 | + } | |
| 6658 | + | |
| 6659 | + if (!in_array(get_post_mime_type($attachment_id), self::mxchat_supported_attachment_mime_types(), true)) { | |
| 6660 | + return; | |
| 6661 | + } | |
| 6662 | + | |
| 6663 | + $source_url = wp_get_attachment_url($attachment_id); | |
| 6664 | + if (empty($source_url)) { | |
| 6665 | + return; | |
| 6666 | + } | |
| 6667 | + | |
| 6668 | + $extracted = $this->mxchat_extract_attachment_text($attachment_id); | |
| 6669 | + if ($extracted['text'] === '') { | |
| 6670 | + return; | |
| 6671 | + } | |
| 6672 | + | |
| 6673 | + $content = $this->mxchat_build_attachment_content($post, $extracted['text']); | |
| 6674 | + $signature = get_post_meta($attachment_id, '_mxchat_kb_attachment_sig', true); | |
| 6675 | + | |
| 6676 | + if (is_array($signature) && isset($signature['url'], $signature['hash'])) { | |
| 6677 | + // Nothing we index changed — don't pay for another embedding. | |
| 6678 | + if ($signature['url'] === $source_url && $signature['hash'] === md5($content)) { | |
| 6679 | + return; | |
| 6680 | + } | |
| 6681 | + // The file was replaced under a new URL. md5(old_url) vectors would survive | |
| 6682 | + // as an entry citing a file that no longer exists there, so clear them first. | |
| 6683 | + if ($signature['url'] !== $source_url && $signature['url'] !== '') { | |
| 6684 | + MxChat_Utils::delete_chunks_for_url($signature['url'], 'default'); | |
| 6685 | + } | |
| 6686 | + } | |
| 6687 | + | |
| 6688 | + $preflight = MxChat_Utils::embedding_preflight(get_option('mxchat_options')); | |
| 6689 | + if (!$preflight['ok']) { | |
| 6690 | + return; | |
| 6691 | + } | |
| 6692 | + | |
| 6693 | + $result = MxChat_Utils::submit_content_to_db( | |
| 6694 | + $content, | |
| 6695 | + $source_url, | |
| 6696 | + $preflight['api_key'], | |
| 6697 | + md5($source_url), | |
| 6698 | + 'default', | |
| 6699 | + 'attachment' | |
| 6700 | + ); | |
| 6701 | + | |
| 6702 | + if (is_wp_error($result)) { | |
| 6703 | + return; | |
| 6704 | + } | |
| 6705 | + | |
| 6706 | + $this->apply_role_restriction_to_post($attachment_id, $source_url); | |
| 6707 | + $this->mxchat_remember_attachment_signature($attachment_id, $source_url, $content); | |
| 6708 | +} | |
| 6709 | + | |
| 6710 | +/** | |
| 6711 | + * Remove an attachment's knowledge entry when the file is deleted. | |
| 6712 | + * | |
| 6713 | + * Fires on 'delete_attachment', which is the ONLY delete hook an attachment gets — | |
| 6714 | + * wp_delete_post() hands attachments to wp_delete_attachment() before firing | |
| 6715 | + * before_delete_post, so mxchat_handle_post_delete never runs for one. | |
| 6716 | + * | |
| 6717 | + * Gated on the same auto-sync toggle as the post delete path, for the same reason it | |
| 6718 | + * is there: with auto-sync off, MxChat does not act on content changes by itself. | |
| 6719 | + */ | |
| 6720 | +public function mxchat_handle_attachment_delete($attachment_id) { | |
| 6721 | + $attachment_id = (int) $attachment_id; | |
| 6722 | + | |
| 6723 | + if (!$this->mxchat_attachment_auto_sync_enabled()) { | |
| 6724 | + return; | |
| 6725 | + } | |
| 6726 | + | |
| 6727 | + // Still resolvable — this hook fires before the row and the file are removed. | |
| 6728 | + $source_url = wp_get_attachment_url($attachment_id); | |
| 6729 | + $signature = get_post_meta($attachment_id, '_mxchat_kb_attachment_sig', true); | |
| 6730 | + | |
| 6731 | + $urls = array(); | |
| 6732 | + if (!empty($source_url)) { | |
| 6733 | + $urls[] = $source_url; | |
| 6734 | + } | |
| 6735 | + // A replaced-then-deleted file can still have vectors under the URL it was | |
| 6736 | + // last indexed at, if that differs from where it lives now. | |
| 6737 | + if (is_array($signature) && !empty($signature['url'])) { | |
| 6738 | + $urls[] = $signature['url']; | |
| 6739 | + } | |
| 6740 | + | |
| 6741 | + foreach (array_unique($urls) as $url) { | |
| 6742 | + MxChat_Utils::delete_chunks_for_url($url, 'default'); | |
| 6743 | + } | |
| 6744 | +} | |
| 6745 | + | |
| 6746 | +/** | |
| 6747 | + * Handle ACF save - fires after ACF fields are saved | |
| 6748 | + * This ensures ACF field data is available when syncing to knowledge base | |
| 6749 | + */ | |
| 6750 | +public function mxchat_handle_acf_save($post_id) { | |
| 6751 | + // Skip if not a valid post | |
| 6752 | + if (!$post_id || $post_id === 'options') { | |
| 6753 | + return; | |
| 6754 | + } | |
| 6755 | + | |
| 6756 | + // Skip autosaves and revisions | |
| 6757 | + if (defined('DOING_AUTOSAVE') && DOING_AUTOSAVE || wp_is_post_revision($post_id)) { | |
| 6758 | + return; | |
| 6759 | + } | |
| 6760 | + | |
| 6761 | + $post = get_post($post_id); | |
| 6762 | + if (!$post) { | |
| 6763 | + return; | |
| 6764 | + } | |
| 6765 | + | |
| 6766 | + $post_type = $post->post_type; | |
| 6767 | + | |
| 6768 | + // Check if sync is enabled for this post type | |
| 6769 | + $should_sync = false; | |
| 6770 | + | |
| 6771 | + if ($post_type === 'post' && get_option('mxchat_auto_sync_posts') === '1') { | |
| 6772 | + $should_sync = true; | |
| 6773 | + } else if ($post_type === 'page' && get_option('mxchat_auto_sync_pages') === '1') { | |
| 6774 | + $should_sync = true; | |
| 6775 | + } else if ($post_type === 'product' && class_exists('WooCommerce')) { | |
| 6776 | + // WooCommerce products - check if WooCommerce integration is enabled | |
| 6777 | + $options = get_option('mxchat_options', array()); | |
| 6778 | + if (isset($options['enable_woocommerce_integration']) && | |
| 6779 | + ($options['enable_woocommerce_integration'] === '1' || $options['enable_woocommerce_integration'] === 'on')) { | |
| 6780 | + $should_sync = true; | |
| 6781 | + } | |
| 6782 | + } else { | |
| 6783 | + // Check custom post types | |
| 6784 | + $option_name = 'mxchat_auto_sync_' . $post_type; | |
| 6785 | + if (get_option($option_name) === '1') { | |
| 6786 | + $should_sync = true; | |
| 6787 | + } | |
| 6788 | + } | |
| 6789 | + | |
| 6790 | + if (!$should_sync) { | |
| 6791 | + return; | |
| 6792 | + } | |
| 6793 | + | |
| 6794 | + // Only process published posts | |
| 6795 | + if ($post->post_status !== 'publish') { | |
| 6796 | + return; | |
| 6797 | + } | |
| 6798 | + | |
| 6799 | + // Check if this post has any ACF fields - if not, no need to re-sync | |
| 6800 | + $acf_fields = $this->mxchat_get_acf_fields_for_post($post_id); | |
| 6801 | + if (empty($acf_fields)) { | |
| 6802 | + return; | |
| 6803 | + } | |
| 6804 | + | |
| 6805 | + // Use a transient to prevent duplicate processing (post_updated may have already run) | |
| 6806 | + $transient_key = 'mxchat_acf_synced_' . $post_id; | |
| 6807 | + if (get_transient($transient_key)) { | |
| 6808 | + return; | |
| 6809 | + } | |
| 6810 | + set_transient($transient_key, true, 60); // Prevent re-processing for 60 seconds | |
| 6811 | + | |
| 6812 | + // Re-run the sync with ACF data now available | |
| 6813 | + // We pass $update=true since this is effectively an update with ACF data | |
| 6814 | + $this->mxchat_handle_post_update($post_id, $post, true); | |
| 6815 | +} | |
| 6816 | + | |
| 6817 | +public function mxchat_handle_post_update($post_id, $post, $update) { | |
| 6818 | + // The in-flight-update marker has done its job the moment post_updated runs; drop it | |
| 6819 | + // before any early return so it can never outlive its own save (a failed $wpdb->update | |
| 6820 | + // inside wp_insert_post returns after pre_post_update but before the transition). | |
| 6821 | + unset($this->pending_post_update[$post_id]); | |
| 6822 | + | |
| 6823 | + // Basic validation checks | |
| 6824 | + if (defined('DOING_AUTOSAVE') && DOING_AUTOSAVE || wp_is_post_revision($post_id)) { | |
| 6825 | + return; | |
| 6826 | + } | |
| 6827 | + | |
| 6828 | + $post_type = $post->post_type; | |
| 6829 | + | |
| 6830 | + // Check if sync is enabled for this post type | |
| 6831 | + $should_sync = false; | |
| 6832 | + | |
| 6833 | + // Check built-in post types first | |
| 6834 | + if ($post_type === 'post' && get_option('mxchat_auto_sync_posts') === '1') { | |
| 6835 | + $should_sync = true; | |
| 6836 | + } else if ($post_type === 'page' && get_option('mxchat_auto_sync_pages') === '1') { | |
| 6837 | + $should_sync = true; | |
| 6838 | + } else { | |
| 6839 | + // Check custom post types | |
| 6840 | + $option_name = 'mxchat_auto_sync_' . $post_type; | |
| 6841 | + if (get_option($option_name) === '1') { | |
| 6842 | + $should_sync = true; | |
| 6843 | + } | |
| 6844 | + } | |
| 6845 | + | |
| 6846 | + if (!$should_sync) { | |
| 6847 | + return; | |
| 6848 | + } | |
| 6849 | + | |
| 6850 | + // Check if we have stored the previous status and URL in our transients | |
| 6851 | + $previous_status_key = 'mxchat_prev_status_' . $post_id; | |
| 6852 | + $previous_status = get_transient($previous_status_key); | |
| 6853 | + | |
| 6854 | + $previous_url_key = 'mxchat_prev_url_' . $post_id; | |
| 6855 | + $previous_url = get_transient($previous_url_key); | |
| 6856 | + | |
| 6857 | + // If the post was previously published but is now not published, remove from knowledge base | |
| 6858 | + if ($previous_status === 'publish' && $post->post_status !== 'publish') { | |
| 6859 | + // Use the stored URL from when it was published, or fall back to current permalink | |
| 6860 | + $source_url = $previous_url ?: get_permalink($post_id); | |
| 6861 | + | |
| 6862 | + // mxchat_handle_status_transition already deleted for this post earlier in this | |
| 6863 | + // request (it fires first inside wp_insert_post); skip the redundant round-trip. | |
| 6864 | + if ($source_url && empty($this->transition_deleted_posts[$post_id])) { | |
| 6865 | + // Chunk-aware deletion (routes to Pinecone or WP DB and removes base + all chunks) | |
| 6866 | + MxChat_Utils::delete_chunks_for_url($source_url, 'default'); | |
| 6867 | + } | |
| 6868 | + | |
| 6869 | + // Clean up the transients and exit early | |
| 6870 | + delete_transient($previous_status_key); | |
| 6871 | + delete_transient($previous_url_key); | |
| 6872 | + return; | |
| 6873 | + } | |
| 6874 | + | |
| 6875 | + // Slug/permalink rename while still published: delete the old vectors before upserting new ones. | |
| 6876 | + // Without this, md5(old_url) vectors (base + chunks) would be orphaned under the stale URL. | |
| 6877 | + if ($post->post_status === 'publish' && !empty($previous_url)) { | |
| 6878 | + $current_url = get_permalink($post_id); | |
| 6879 | + if ($current_url && $current_url !== $previous_url) { | |
| 6880 | + MxChat_Utils::delete_chunks_for_url($previous_url, 'default'); | |
| 6881 | + } | |
| 6882 | + } | |
| 6883 | + | |
| 6884 | + // Store the current status for next time (if this is an update) | |
| 6885 | + if ($update) { | |
| 6886 | + set_transient($previous_status_key, $post->post_status, DAY_IN_SECONDS); | |
| 6887 | + | |
| 6888 | + // If the post is currently published, also store its URL | |
| 6889 | + if ($post->post_status === 'publish') { | |
| 6890 | + $current_url = get_permalink($post_id); | |
| 6891 | + set_transient($previous_url_key, $current_url, DAY_IN_SECONDS); | |
| 6892 | + } | |
| 6893 | + } | |
| 6894 | + | |
| 6895 | + // Only process currently published content for adding/updating. | |
| 6896 | + // transition_indexed_posts: mxchat_handle_status_transition's arrival edge may have | |
| 6897 | + // already indexed this post earlier in this request (editor publishes fire | |
| 6898 | + // transition_post_status first, then post_updated) — skip the duplicate embed. | |
| 6899 | + // Consume-once: the flag is cleared when honoured, so a LATER save of the same | |
| 6900 | + // post in one long-running process (WP-CLI scripts, importers) re-indexes normally. | |
| 6901 | + if ($post->post_status === 'publish') { | |
| 6902 | + if (!empty($this->transition_indexed_posts[$post_id])) { | |
| 6903 | + unset($this->transition_indexed_posts[$post_id]); | |
| 6904 | + } else { | |
| 6905 | + $this->mxchat_index_published_post($post_id, $post); | |
| 6906 | + } | |
| 6907 | + } | |
| 6908 | + | |
| 6909 | + // Clean up the stored previous status if not used above | |
| 6910 | + if ($previous_status !== 'publish' || $post->post_status === 'publish') { | |
| 6911 | + delete_transient($previous_status_key); | |
| 6912 | + delete_transient($previous_url_key); | |
| 6913 | + } | |
| 6914 | +} | |
| 6915 | + | |
| 6916 | +/** | |
| 6917 | + * Index a published post into the knowledge base: preprocessing filter, content | |
| 6918 | + * assembly (title/excerpt/body), WooCommerce product enrichment, job_listing meta, | |
| 6919 | + * ACF fields (+ optional PDF extraction), whitelisted custom meta, embedding and | |
| 6920 | + * upsert, then tag-based role restriction. | |
| 6921 | + * | |
| 6922 | + * Shared by the post_updated auto-sync path (mxchat_handle_post_update) and the | |
| 6923 | + * transition_post_status arrival edge (mxchat_handle_status_transition), so | |
| 6924 | + * scheduled publishes (wp_publish_post) and direct status=publish inserts index | |
| 6925 | + * identically to editor saves (plan 3055e1). Pure extraction of the former | |
| 6926 | + * publish branch — body indentation retained to keep the diff reviewable. | |
| 6927 | + */ | |
| 6928 | +private function mxchat_index_published_post($post_id, $post) { | |
| 6929 | + $post_type = $post->post_type; | |
| 6930 | + | |
| 6931 | + // WooCommerce products are owned by the WC-object assembler (plan a3d60c): | |
| 6932 | + // whenever WooCommerce is active AND the integration is enabled, every | |
| 6933 | + // product save also fires save_post_product, which queues | |
| 6934 | + // mxchat_store_product_embedding on shutdown — and that writer runs LAST, | |
| 6935 | + // overwriting the same md5(permalink) row this path would write. Assembling | |
| 6936 | + // and embedding the product here was pure duplicate spend (measured: two | |
| 6937 | + // embedding calls per product save, second one wins). Skip ONLY under the | |
| 6938 | + // exact conditions the shutdown writer runs — same option read as its own | |
| 6939 | + // gate — because with the integration off (or WooCommerce inactive) this | |
| 6940 | + // path is the sole product indexer and must keep working. | |
| 6941 | + if ($post_type === 'product' | |
| 6942 | + && class_exists('WooCommerce') | |
| 6943 | + && isset($this->options['enable_woocommerce_integration']) | |
| 6944 | + && in_array($this->options['enable_woocommerce_integration'], ['1', 'on'])) { | |
| 6945 | + return; | |
| 6946 | + } | |
| 6947 | + | |
| 6948 | + // Get the source URL | |
| 6949 | + $source_url = get_permalink($post_id); | |
| 6950 | + | |
| 6951 | + // A draft published programmatically (wp_publish_post) can reach this | |
| 6952 | + // point with an EMPTY post_name — wp_insert_post skips slug generation | |
| 6953 | + // for draft/pending — and get_permalink() then resolves to the bare | |
| 6954 | + // site root. A knowledge row keyed to the homepage cites the wrong URL | |
| 6955 | + // and answers homepage questions with this post's body, so refuse to | |
| 6956 | + // write it; the post indexes correctly on its next save, once the slug | |
| 6957 | + // exists. The empty-post_name test is what keeps a legitimate static | |
| 6958 | + // front page (which has a slug but a root permalink) indexable. | |
| 6959 | + // (Plan d138c4.) | |
| 6960 | + if ('' === $post->post_name | |
| 6961 | + && untrailingslashit($source_url) === untrailingslashit(home_url())) { | |
| 6962 | + return; | |
| 6963 | + } | |
| 6964 | + | |
| 6965 | + /** | |
| 6966 | + * Allow developers to modify post data before processing into the knowledge base. | |
| 6967 | + * Same filter and signature as the manual bulk-import path | |
| 6968 | + * (ajax_mxchat_process_selected_content), so a callback registered once covers | |
| 6969 | + * every indexing route. Purely additive — zero behaviour change when unhooked. | |
| 6970 | + * Auto-sync runs under the 'default' bot context, matching the rest of this | |
| 6971 | + * function. | |
| 6972 | + * | |
| 6973 | + * @param WP_Post $post The post about to be indexed. | |
| 6974 | + * @param string $bot_id Bot context ('default' on auto-sync). | |
| 6975 | + */ | |
| 6976 | + $post = apply_filters('mxchat_before_process_post', $post, 'default'); | |
| 6977 | + if (!($post instanceof WP_Post)) { | |
| 6978 | + $post = get_post($post_id); // defend against a bad callback return | |
| 6979 | + } | |
| 6980 | + | |
| 6981 | + // Assemble the indexable text via the shared post-kind assembler (a3d60c), | |
| 6982 | + // reading from the FILTERED post object — not re-fetched by ID, which would | |
| 6983 | + // discard it. Auto-sync reads content/excerpt in its historical | |
| 6984 | + // get_post_field() display context, never appended product custom tabs | |
| 6985 | + // (its product branch is reachable only with the WooCommerce integration | |
| 6986 | + // off), and gates ACF→PDF extraction behind its own opt-in option — | |
| 6987 | + // default OFF, because re-parsing every ACF PDF on every editor save is | |
| 6988 | + // expensive and most sites don't want it (the 25 MB size cap lives in the | |
| 6989 | + // shared extractor either way). | |
| 6990 | + $prepared = $this->mxchat_prepare_post_content_for_indexing($post_id, $post, array( | |
| 6991 | + 'read_display' => true, | |
| 6992 | + 'extract_acf_pdfs' => get_option('mxchat_auto_sync_acf_pdfs', '0') === '1', | |
| 6993 | + 'include_product_tabs' => false, | |
| 6994 | + )); | |
| 6995 | + $final_content = $prepared['content']; | |
| 6996 | + | |
| 6997 | + // Embedding decision — custom-provider-aware. Gating on a cloud API key | |
| 6998 | + // here silently killed auto-sync on keyless custom-embeddings sites, | |
| 6999 | + // because generate_embedding() routes custom FIRST and never needs the | |
| 7000 | + // key (plan cbd5fd). Silent-return shape preserved. | |
| 7001 | + $preflight = MxChat_Utils::embedding_preflight(get_option('mxchat_options')); | |
| 7002 | + if (!$preflight['ok']) { | |
| 7003 | + return; | |
| 7004 | + } | |
| 7005 | + $api_key = $preflight['api_key']; | |
| 7006 | + | |
| 7007 | + // Use the centralized utility function for storage | |
| 7008 | + $result = MxChat_Utils::submit_content_to_db( | |
| 7009 | + $final_content, | |
| 7010 | + $source_url, | |
| 7011 | + $api_key, | |
| 7012 | + md5($source_url) // Vector ID for Pinecone | |
| 7013 | + ); | |
| 7014 | + | |
| 7015 | + // After successful storage, apply role restriction based on tags | |
| 7016 | + if (!is_wp_error($result)) { | |
| 7017 | + $this->apply_role_restriction_to_post($post_id, $source_url); | |
| 7018 | + } | |
| 7019 | +} | |
| 7020 | + | |
| 7021 | +/** | |
| 7022 | + * Shared post-fields content assembler (plan a3d60c) — the ONE body behind both | |
| 7023 | + * post-kind ingestion paths: manual bulk import (ajax_mxchat_process_selected_content) | |
| 7024 | + * and auto-sync (mxchat_index_published_post). Behavior-preserving extraction; the | |
| 7025 | + * measured per-caller differences ride $args instead of living as drifting copies: | |
| 7026 | + * | |
| 7027 | + * 'read_display' bool Auto-sync historically reads content/excerpt via | |
| 7028 | + * get_post_field() in its default 'display' context | |
| 7029 | + * (the post_content / post_excerpt display filters | |
| 7030 | + * fire); bulk import reads the raw properties. Inert | |
| 7031 | + * on a stock install — preserved per-path, not converged. | |
| 7032 | + * 'extract_acf_pdfs' bool Each caller passes its OWN option (bulk: | |
| 7033 | + * mxchat_acf_pdf_extraction; auto-sync: | |
| 7034 | + * mxchat_auto_sync_acf_pdfs) — the two-option design | |
| 7035 | + * is deliberate (plan 11720c). Gates BOTH the PDF-id | |
| 7036 | + * collection walk and the extraction loop; the ids are | |
| 7037 | + * only ever read inside the extraction branch, so | |
| 7038 | + * gating collection is output-identical on every install. | |
| 7039 | + * 'include_product_tabs' bool The bulk path has always appended yikes_woo custom | |
| 7040 | + * tabs to product content; the auto-sync product branch | |
| 7041 | + * (reachable only with the WooCommerce integration off) | |
| 7042 | + * never did. Preserved per-path — converging it would be | |
| 7043 | + * a behavior change, recorded on the plan instead. | |
| 7044 | + * | |
| 7045 | + * Returns array: 'content' (the assembled indexable text), 'acf_fields_found' and | |
| 7046 | + * 'pdf_extracted_count' (the bulk path reports both in its AJAX response). | |
| 7047 | + */ | |
| 7048 | +private function mxchat_prepare_post_content_for_indexing($post_id, $post, $args) { | |
| 7049 | + $read_display = !empty($args['read_display']); | |
| 7050 | + $extract_acf_pdfs = !empty($args['extract_acf_pdfs']); | |
| 7051 | + $include_product_tabs = !empty($args['include_product_tabs']); | |
| 7052 | + | |
| 7053 | + // Raw post_title, NOT get_the_title(): the_title applies wptexturize + | |
| 7054 | + // convert_chars and prepends the "Protected:" / "Private:" display chrome. | |
| 7055 | + // The knowledge base stores facts, not display strings. Entity decode at | |
| 7056 | + // output time (single-pass, shared helper) — a stored `&` embeds worse | |
| 7057 | + // than `&` and gets quoted back to visitors (d2c92e). | |
| 7058 | + $content = $this->mxchat_decode_entities_for_indexing($post->post_title) . "\n\n"; | |
| 7059 | + | |
| 7060 | + $raw_excerpt = $read_display ? get_post_field('post_excerpt', $post) : $post->post_excerpt; | |
| 7061 | + $raw_content = $read_display ? get_post_field('post_content', $post) : $post->post_content; | |
| 7062 | + | |
| 7063 | + // Add short description if it exists (WooCommerce products use post_excerpt for short description) | |
| 7064 | + // Strip FIRST, then test: an excerpt that is nothing but shortcodes strips to | |
| 7065 | + // empty, and testing the raw value emitted a bare "Short Description: " label | |
| 7066 | + // with no value after it. trim() only in the TEST — the emitted value is | |
| 7067 | + // untouched, so a populated excerpt is byte-identical to before. A | |
| 7068 | + // whitespace-only excerpt is an empty excerpt and must not produce a labelled | |
| 7069 | + // line with nothing after it. | |
| 7070 | + $clean_excerpt = $this->strip_shortcode_tags_preserve_content($raw_excerpt); | |
| 7071 | + if (trim($clean_excerpt) !== '') { | |
| 7072 | + $content .= "Short Description: " . $this->mxchat_decode_entities_for_indexing(wp_strip_all_tags($clean_excerpt)) . "\n\n"; | |
| 7073 | + } | |
| 7074 | + | |
| 7075 | + // Main content — remove shortcode tags but preserve content inside them, then | |
| 7076 | + // strip tags (don't use 'the_content' filter as it may re-add shortcodes). | |
| 7077 | + $clean_content = $this->strip_shortcode_tags_preserve_content($raw_content); | |
| 7078 | + $content .= $this->mxchat_decode_entities_for_indexing(wp_strip_all_tags($clean_content)); | |
| 7079 | + | |
| 7080 | + // WooCommerce product enrichment (post-fields kind). The WC-object assembler | |
| 7081 | + // (mxchat_prepare_product_content_for_indexing) owns product rows whenever the | |
| 7082 | + // integration is on; this branch serves the bulk import (all configurations) | |
| 7083 | + // and auto-sync with the integration off. | |
| 7084 | + if (get_post_type($post_id) === 'product' && class_exists('WooCommerce')) { | |
| 7085 | + $product = wc_get_product($post_id); | |
| 7086 | + | |
| 7087 | + if ($product) { | |
| 7088 | + $content .= "\n"; | |
| 7089 | + $content .= $this->mxchat_woo_product_summary_lines($product); | |
| 7090 | + } | |
| 7091 | + | |
| 7092 | + if ($include_product_tabs) { | |
| 7093 | + $content .= $this->mxchat_woo_custom_tabs_text($post_id); | |
| 7094 | + } | |
| 7095 | + } | |
| 7096 | + | |
| 7097 | + // For custom post types like job_listing, include additional fields | |
| 7098 | + if (get_post_type($post_id) === 'job_listing') { | |
| 7099 | + // Add job-specific meta if available | |
| 7100 | + $job_location = get_post_meta($post_id, '_job_location', true); | |
| 7101 | + if (!empty($job_location)) { | |
| 7102 | + $content .= "\n\nLocation: " . $job_location; | |
| 7103 | + } | |
| 7104 | + | |
| 7105 | + // Get job type terms | |
| 7106 | + $job_types = get_the_terms($post_id, 'job_listing_type'); | |
| 7107 | + if (!empty($job_types) && !is_wp_error($job_types)) { | |
| 7108 | + $types = array(); | |
| 7109 | + foreach ($job_types as $type) { | |
| 7110 | + $types[] = $type->name; | |
| 7111 | + } | |
| 7112 | + $content .= "\n\nJob Type: " . implode(', ', $types); | |
| 7113 | + } | |
| 7114 | + | |
| 7115 | + // Get company name if available | |
| 7116 | + $company_name = get_post_meta($post_id, '_company_name', true); | |
| 7117 | + if (!empty($company_name)) { | |
| 7118 | + $content .= "\n\nCompany: " . $company_name; | |
| 7119 | + } | |
| 7120 | + } | |
| 7121 | + | |
| 7122 | + // ADD ACF FIELDS SUPPORT | |
| 7123 | + $acf_fields = $this->mxchat_get_acf_fields_for_post($post_id); | |
| 7124 | + $pdf_extracted_count = 0; | |
| 7125 | + if (!empty($acf_fields)) { | |
| 7126 | + $acf_content_parts = array(); | |
| 7127 | + $pdf_attachment_ids = array(); | |
| 7128 | + | |
| 7129 | + foreach ($acf_fields as $field_name => $field_value) { | |
| 7130 | + $formatted_value = $this->mxchat_format_acf_field_value($field_value, $field_name, $post_id); | |
| 7131 | + | |
| 7132 | + if (!empty($formatted_value)) { | |
| 7133 | + // Both separators: a hyphenated ACF name should read as words. | |
| 7134 | + $field_label = ucwords(str_replace(['_', '-'], ' ', $field_name)); | |
| 7135 | + $acf_content_parts[] = $field_label . ": " . $formatted_value; | |
| 7136 | + } | |
| 7137 | + | |
| 7138 | + // Walk this field's value tree for any PDF attachment references and | |
| 7139 | + // queue them for extraction — only when this caller's PDF option is on. | |
| 7140 | + if ($extract_acf_pdfs) { | |
| 7141 | + $this->mxchat_collect_pdf_attachment_ids_from_acf_value($field_value, $pdf_attachment_ids); | |
| 7142 | + } | |
| 7143 | + } | |
| 7144 | + | |
| 7145 | + if (!empty($acf_content_parts)) { | |
| 7146 | + $content .= "\n\n" . implode("\n", $acf_content_parts); | |
| 7147 | + } | |
| 7148 | + | |
| 7149 | + // Extract text from each unique PDF found in ACF fields and append as a labeled section | |
| 7150 | + if ($extract_acf_pdfs && !empty($pdf_attachment_ids)) { | |
| 7151 | + $pdf_attachment_ids = array_unique(array_filter(array_map('intval', $pdf_attachment_ids))); | |
| 7152 | + $pdf_sections = array(); | |
| 7153 | + foreach ($pdf_attachment_ids as $att_id) { | |
| 7154 | + $pdf_text = $this->mxchat_extract_pdf_text_by_attachment_id($att_id); | |
| 7155 | + if (!empty($pdf_text)) { | |
| 7156 | + $pdf_title = get_the_title($att_id); | |
| 7157 | + $pdf_url = wp_get_attachment_url($att_id); | |
| 7158 | + $header = 'PDF Attachment'; | |
| 7159 | + if (!empty($pdf_title)) { | |
| 7160 | + $header .= ': ' . $pdf_title; | |
| 7161 | + } | |
| 7162 | + if (!empty($pdf_url)) { | |
| 7163 | + $header .= ' (' . $pdf_url . ')'; | |
| 7164 | + } | |
| 7165 | + $pdf_sections[] = $header . "\n" . $pdf_text; | |
| 7166 | + $pdf_extracted_count++; | |
| 7167 | + } | |
| 7168 | + } | |
| 7169 | + if (!empty($pdf_sections)) { | |
| 7170 | + $content .= "\n\nAttached PDFs:\n" . implode("\n\n", $pdf_sections); | |
| 7171 | + } | |
| 7172 | + } | |
| 7173 | + } | |
| 7174 | + | |
| 7175 | + // ADD CUSTOM POST META SUPPORT (whitelisted non-ACF meta fields) | |
| 7176 | + $custom_meta = $this->mxchat_get_whitelisted_post_meta($post_id); | |
| 7177 | + if (!empty($custom_meta)) { | |
| 7178 | + $meta_content_parts = array(); | |
| 7179 | + | |
| 7180 | + foreach ($custom_meta as $meta_key => $meta_value) { | |
| 7181 | + // Convert meta key to readable label | |
| 7182 | + $meta_label = ucwords(str_replace(array('_', '-'), ' ', $meta_key)); | |
| 7183 | + $meta_content_parts[] = $meta_label . ": " . $meta_value; | |
| 7184 | + } | |
| 7185 | + | |
| 7186 | + if (!empty($meta_content_parts)) { | |
| 7187 | + $content .= "\n\n" . implode("\n", $meta_content_parts); | |
| 7188 | + } | |
| 7189 | + } | |
| 7190 | + | |
| 7191 | + return array( | |
| 7192 | + 'content' => $content, | |
| 7193 | + 'acf_fields_found' => count($acf_fields), | |
| 7194 | + 'pdf_extracted_count' => $pdf_extracted_count, | |
| 7195 | + ); | |
| 7196 | +} | |
| 7197 | + | |
| 7198 | +/** | |
| 7199 | + * Shared WC-object product assembler (plan a3d60c) — the ONE body behind the two | |
| 7200 | + * WooCommerce-object ingestion paths: the auto-sync product writer | |
| 7201 | + * (mxchat_store_product_embedding) and the URL/sitemap product import | |
| 7202 | + * (mxchat_extract_woocommerce_product_content). Assembles from the WC_Product, | |
| 7203 | + * the authoritative source for product rows (scope decision on the plan). | |
| 7204 | + */ | |
| 7205 | +private function mxchat_prepare_product_content_for_indexing($product) { | |
| 7206 | + $title = $product->get_name(); | |
| 7207 | + $description = $product->get_description(); | |
| 7208 | + $short_description = $product->get_short_description(); | |
| 7209 | + | |
| 7210 | + // Format content consistently | |
| 7211 | + $content = $title . "\n\n"; | |
| 7212 | + | |
| 7213 | + if (!empty($short_description)) { | |
| 7214 | + $content .= "Short Description: " . wp_strip_all_tags($short_description) . "\n\n"; | |
| 7215 | + } | |
| 7216 | + | |
| 7217 | + if (!empty($description)) { | |
| 7218 | + $content .= wp_strip_all_tags($description) . "\n\n"; | |
| 7219 | + } | |
| 7220 | + | |
| 7221 | + $content .= $this->mxchat_woo_product_summary_lines($product); | |
| 7222 | + $content .= $this->mxchat_woo_custom_tabs_text($product->get_id()); | |
| 7223 | + | |
| 7224 | + return $content; | |
| 7225 | +} | |
| 7226 | + | |
| 7227 | +/** | |
| 7228 | + * Pricing + SKU + categories + tags lines for a product — shared by both assembler kinds | |
| 7229 | + * (the post-fields product enrichment and the WC-object assembler). | |
| 7230 | + */ | |
| 7231 | +private function mxchat_woo_product_summary_lines($product) { | |
| 7232 | + // Add pricing information (base-currency pinned — see mxchat_product_price_lines) | |
| 7233 | + $lines = $this->mxchat_product_price_lines($product); | |
| 7234 | + | |
| 7235 | + $sku = $product->get_sku(); | |
| 7236 | + if (!empty($sku)) { | |
| 7237 | + $lines .= "SKU: " . $sku . "\n"; | |
| 7238 | + } | |
| 7239 | + | |
| 7240 | + // Get product categories | |
| 7241 | + $categories = wp_get_post_terms($product->get_id(), 'product_cat', array('fields' => 'names')); | |
| 7242 | + if (!empty($categories) && !is_wp_error($categories)) { | |
| 7243 | + $lines .= "Categories: " . implode(', ', $categories) . "\n"; | |
| 7244 | + } | |
| 7245 | + | |
| 7246 | + // Product tags (plan 799cc4). A store's tags often carry the words a | |
| 7247 | + // visitor asks with ("vegan", "gift") and none of them are in the | |
| 7248 | + // description. No line at all when the product has no tags. | |
| 7249 | + $tags = wp_get_post_terms($product->get_id(), 'product_tag', array('fields' => 'names')); | |
| 7250 | + if (!empty($tags) && !is_wp_error($tags)) { | |
| 7251 | + $lines .= "Tags: " . $this->mxchat_decode_entities_for_indexing(implode(', ', $tags)) . "\n"; | |
| 7252 | + } | |
| 7253 | + | |
| 7254 | + return $lines; | |
| 7255 | +} | |
| 7256 | + | |
| 7257 | +/** | |
| 7258 | + * Custom Product Tabs text (supports "Custom Product Tabs for WooCommerce" by | |
| 7259 | + * Code Parrots) — direct tabs plus applied reusable/saved tabs. The ONE copy of | |
| 7260 | + * the yikes_woo logic; three sites carried byte-identical clones before a3d60c. | |
| 7261 | + */ | |
| 7262 | +private function mxchat_woo_custom_tabs_text($product_id) { | |
| 7263 | + $text = ''; | |
| 7264 | + | |
| 7265 | + $custom_tabs = get_post_meta($product_id, 'yikes_woo_products_tabs', true); | |
| 7266 | + if (!empty($custom_tabs) && is_array($custom_tabs)) { | |
| 7267 | + foreach ($custom_tabs as $tab) { | |
| 7268 | + $tab_title = isset($tab['title']) ? $tab['title'] : (isset($tab['tab_title']) ? $tab['tab_title'] : ''); | |
| 7269 | + $tab_content = isset($tab['content']) ? $tab['content'] : ''; | |
| 7270 | + | |
| 7271 | + if (!empty($tab_title) && !empty($tab_content)) { | |
| 7272 | + $text .= "\n" . $tab_title . ": " . wp_strip_all_tags($tab_content) . "\n"; | |
| 7273 | + } | |
| 7274 | + } | |
| 7275 | + } | |
| 7276 | + | |
| 7277 | + // Also check for reusable/saved tabs applied to this product | |
| 7278 | + $applied_saved_tabs = get_post_meta($product_id, 'yikes_woo_reusable_products_tabs_applied', true); | |
| 7279 | + if (!empty($applied_saved_tabs) && is_array($applied_saved_tabs)) { | |
| 7280 | + $saved_tabs = get_option('yikes_woo_reusable_products_tabs', array()); | |
| 7281 | + if (!empty($saved_tabs) && is_array($saved_tabs)) { | |
| 7282 | + foreach ($applied_saved_tabs as $saved_tab_id) { | |
| 7283 | + if (isset($saved_tabs[$saved_tab_id])) { | |
| 7284 | + $tab = $saved_tabs[$saved_tab_id]; | |
| 7285 | + $tab_title = isset($tab['title']) ? $tab['title'] : (isset($tab['tab_title']) ? $tab['tab_title'] : ''); | |
| 7286 | + $tab_content = isset($tab['content']) ? $tab['content'] : ''; | |
| 7287 | + | |
| 7288 | + if (!empty($tab_title) && !empty($tab_content)) { | |
| 7289 | + $text .= "\n" . $tab_title . ": " . wp_strip_all_tags($tab_content) . "\n"; | |
| 7290 | + } | |
| 7291 | + } | |
| 7292 | + } | |
| 7293 | + } | |
| 7294 | + } | |
| 7295 | + | |
| 7296 | + return $text; | |
| 7297 | +} | |
| 7298 | + | |
| 7299 | +/** | |
| 7300 | + * Store the post status and URL before update to detect status transitions | |
| 7301 | + * This runs before the post is actually updated in the database | |
| 7302 | + */ | |
| 7303 | +public function mxchat_store_pre_update_status($post_id, $data) { | |
| 7304 | + // Core is inside wp_insert_post's update branch, so a post_updated WILL fire later | |
| 7305 | + // this request and can consume the arrival-edge guard (plan a664f3). | |
| 7306 | + $this->pending_post_update[$post_id] = true; | |
| 7307 | + | |
| 7308 | + // Get the current post from database (before update) | |
| 7309 | + $current_post = get_post($post_id); | |
| 7310 | + | |
| 7311 | + if ($current_post) { | |
| 7312 | + // Store the current status temporarily | |
| 7313 | + $status_key = 'mxchat_prev_status_' . $post_id; | |
| 7314 | + set_transient($status_key, $current_post->post_status, HOUR_IN_SECONDS); | |
| 7315 | + | |
| 7316 | + // If the post is currently published, also store its URL | |
| 7317 | + if ($current_post->post_status === 'publish') { | |
| 7318 | + $url_key = 'mxchat_prev_url_' . $post_id; | |
| 7319 | + $current_url = get_permalink($post_id); | |
| 7320 | + set_transient($url_key, $current_url, HOUR_IN_SECONDS); | |
| 7321 | + } | |
| 7322 | + } | |
| 7323 | +} | |
| 7324 | + | |
| 7325 | +/** | |
| 7326 | + * Whether auto-sync is enabled for a post type (mirrors the checks used by the | |
| 7327 | + * update/delete handlers; kept as one helper so new call sites cannot drift). | |
| 7328 | + */ | |
| 7329 | +private function mxchat_is_auto_sync_enabled($post_type) { | |
| 7330 | + if ($post_type === 'post') { | |
| 7331 | + return get_option('mxchat_auto_sync_posts') === '1'; | |
| 7332 | + } | |
| 7333 | + if ($post_type === 'page') { | |
| 7334 | + return get_option('mxchat_auto_sync_pages') === '1'; | |
| 7335 | + } | |
| 7336 | + return get_option('mxchat_auto_sync_' . $post_type) === '1'; | |
| 7337 | +} | |
| 7338 | + | |
| 7339 | +/** | |
| 7340 | + * Remove a post's vectors the moment it leaves 'publish', using the authoritative | |
| 7341 | + * old status core passes to transition_post_status — no transient involved (plan 816fb1). | |
| 7342 | + * | |
| 7343 | + * Covers status changes that never route through wp_update_post (scheduled-expiry | |
| 7344 | + * plugins and others that flip post_status directly and call wp_transition_post_status), | |
| 7345 | + * where neither pre_post_update nor post_updated fires and the old detection missed. | |
| 7346 | + */ | |
| 7347 | +public function mxchat_handle_status_transition($new_status, $old_status, $post) { | |
| 7348 | + if (!($post instanceof WP_Post) || wp_is_post_revision($post->ID)) { | |
| 7349 | + return; | |
| 7350 | + } | |
| 7351 | + | |
| 7352 | + // Arrival edge (plan 3055e1): a post BECOMING published is indexed here, because | |
| 7353 | + // wp_publish_post() — the path scheduled posts take via check_and_publish_future_post — | |
| 7354 | + // and direct wp_insert_post(status=publish) creates never fire post_updated, so the | |
| 7355 | + // auto-sync ADD path alone misses them. Editor publishes also pass through here; | |
| 7356 | + // the transition_indexed_posts guard keeps mxchat_handle_post_update from embedding | |
| 7357 | + // a second time in the same request. | |
| 7358 | + if ($new_status === 'publish' && $old_status !== 'publish') { | |
| 7359 | + if ($this->mxchat_is_auto_sync_enabled($post->post_type)) { | |
| 7360 | + $this->mxchat_index_published_post($post->ID, $post); | |
| 7361 | + | |
| 7362 | + // Arm the double-fire guard ONLY when a post_updated is actually coming to | |
| 7363 | + // consume it (plan a664f3). Two publish paths never fire post_updated at all: | |
| 7364 | + // a direct wp_insert_post(status=publish) create, and wp_publish_post() — the | |
| 7365 | + // call check_and_publish_future_post() makes for scheduled posts. Arming the | |
| 7366 | + // guard unconditionally left it set with nothing to consume it, so the NEXT | |
| 7367 | + // update of that post was swallowed entirely: zero embed calls, no knowledge | |
| 7368 | + // -base row, silently. Consume-once on this side too, so a guard can never | |
| 7369 | + // outlive the single save it was armed for. | |
| 7370 | + if (!empty($this->pending_post_update[$post->ID])) { | |
| 7371 | + unset($this->pending_post_update[$post->ID]); | |
| 7372 | + $this->transition_indexed_posts[$post->ID] = true; | |
| 7373 | + } | |
| 7374 | + } | |
| 7375 | + return; | |
| 7376 | + } | |
| 7377 | + | |
| 7378 | + // Only the publish -> not-publish edge matters here. | |
| 7379 | + if ($old_status !== 'publish' || $new_status === 'publish') { | |
| 7380 | + return; | |
| 7381 | + } | |
| 7382 | + // Trash is handled by mxchat_handle_post_delete (wp_trash_post) with pre-trash URL | |
| 7383 | + // resolution; skip to avoid a second network round-trip per trash. | |
| 7384 | + if ($new_status === 'trash') { | |
| 7385 | + return; | |
| 7386 | + } | |
| 7387 | + if (!$this->mxchat_is_auto_sync_enabled($post->post_type)) { | |
| 7388 | + return; | |
| 7389 | + } | |
| 7390 | + | |
| 7391 | + $urls = array(); | |
| 7392 | + | |
| 7393 | + // The DB may already hold the new status when this fires, so get_permalink() on the | |
| 7394 | + // live post could build a draft-style URL whose md5 misses the stored vector IDs. | |
| 7395 | + // Reconstruct the published permalink from a clone instead. | |
| 7396 | + $published_clone = clone $post; | |
| 7397 | + $published_clone->post_status = 'publish'; | |
| 7398 | + $published_url = get_permalink($published_clone); | |
| 7399 | + if ($published_url) { | |
| 7400 | + $urls[] = $published_url; | |
| 7401 | + } | |
| 7402 | + | |
| 7403 | + // Honour the pre-update capture when present (covers a slug change in the same save). | |
| 7404 | + $previous_url = get_transient('mxchat_prev_url_' . $post->ID); | |
| 7405 | + if (!empty($previous_url)) { | |
| 7406 | + $urls[] = $previous_url; | |
| 7407 | + } | |
| 7408 | + | |
| 7409 | + foreach (array_unique($urls) as $url) { | |
| 7410 | + MxChat_Utils::delete_chunks_for_url($url, 'default'); | |
| 7411 | + } | |
| 7412 | + | |
| 7413 | + if (!empty($urls)) { | |
| 7414 | + $this->transition_deleted_posts[$post->ID] = true; | |
| 7415 | + } | |
| 7416 | +} | |
| 7417 | + | |
| 7418 | +/** | |
| 7419 | + * WP-CLI: remove knowledge-base entries left behind by posts that were unpublished, | |
| 7420 | + * trashed, or made private before the transition_post_status handler existed. | |
| 7421 | + * | |
| 7422 | + * Walks every auto-synced post type's non-published posts, reconstructs each one's | |
| 7423 | + * published-era permalink, and deletes its vectors (routes to Pinecone or the WP table). | |
| 7424 | + * Deletion is idempotent, so never-indexed posts are a cheap no-op. | |
| 7425 | + * | |
| 7426 | + * ## OPTIONS | |
| 7427 | + * | |
| 7428 | + * [--dry-run] | |
| 7429 | + * : Report what would be removed without deleting anything. | |
| 7430 | + * | |
| 7431 | + * ## EXAMPLES | |
| 7432 | + * | |
| 7433 | + * wp mxchat prune-unpublished --dry-run | |
| 7434 | + * wp mxchat prune-unpublished | |
| 7435 | + */ | |
| 7436 | +public function cli_prune_unpublished($args, $assoc_args) { | |
| 7437 | + global $wpdb; | |
| 7438 | + $dry_run = !empty($assoc_args['dry-run']); | |
| 7439 | + $table = $wpdb->prefix . 'mxchat_system_prompt_content'; | |
| 7440 | + | |
| 7441 | + $candidate_types = array_merge(array('post', 'page'), array_values(get_post_types(array('_builtin' => false), 'names'))); | |
| 7442 | + $synced_types = array(); | |
| 7443 | + foreach ($candidate_types as $type) { | |
| 7444 | + if ($this->mxchat_is_auto_sync_enabled($type)) { | |
| 7445 | + $synced_types[] = $type; | |
| 7446 | + } | |
| 7447 | + } | |
| 7448 | + if (empty($synced_types)) { | |
| 7449 | + WP_CLI::success('No post types have auto-sync enabled; nothing to prune.'); | |
| 7450 | + return; | |
| 7451 | + } | |
| 7452 | + | |
| 7453 | + $scanned = 0; | |
| 7454 | + $pruned = 0; | |
| 7455 | + $paged = 1; | |
| 7456 | + do { | |
| 7457 | + $query = new WP_Query(array( | |
| 7458 | + 'post_type' => $synced_types, | |
| 7459 | + 'post_status' => array('draft', 'pending', 'private', 'future', 'trash'), | |
| 7460 | + 'posts_per_page' => 100, | |
| 7461 | + 'paged' => $paged, | |
| 7462 | + 'fields' => 'ids', | |
| 7463 | + )); | |
| 7464 | + foreach ($query->posts as $post_id) { | |
| 7465 | + $post = get_post($post_id); | |
| 7466 | + if (!$post) { | |
| 7467 | + continue; | |
| 7468 | + } | |
| 7469 | + $scanned++; | |
| 7470 | + | |
| 7471 | + // Rebuild the permalink the post had while published: publish-status clone, | |
| 7472 | + // with wp_trash_post's __trashed slug suffix stripped for trashed posts. | |
| 7473 | + $clone = clone $post; | |
| 7474 | + $clone->post_status = 'publish'; | |
| 7475 | + if (substr($clone->post_name, -9) === '__trashed') { | |
| 7476 | + $clone->post_name = substr($clone->post_name, 0, -9); | |
| 7477 | + } | |
| 7478 | + $url = get_permalink($clone); | |
| 7479 | + if (!$url) { | |
| 7480 | + continue; | |
| 7481 | + } | |
| 7482 | + | |
| 7483 | + // Local-table row count is exact in WordPress-DB mode; in Pinecone mode it | |
| 7484 | + // reads 0 but the delete below still routes to Pinecone and is idempotent. | |
| 7485 | + $local_rows = (int) $wpdb->get_var($wpdb->prepare( | |
| 7486 | + "SELECT COUNT(*) FROM {$table} WHERE source_url = %s", $url | |
| 7487 | + )); | |
| 7488 | + | |
| 7489 | + if ($dry_run) { | |
| 7490 | + if ($local_rows > 0) { | |
| 7491 | + WP_CLI::log(sprintf('Would remove %d row(s): %s (post %d, %s)', $local_rows, $url, $post_id, $post->post_status)); | |
| 7492 | + $pruned += $local_rows; | |
| 7493 | + } | |
| 7494 | + continue; | |
| 7495 | + } | |
| 7496 | + | |
| 7497 | + MxChat_Utils::delete_chunks_for_url($url, 'default'); | |
| 7498 | + if ($local_rows > 0) { | |
| 7499 | + WP_CLI::log(sprintf('Removed %d row(s): %s (post %d, %s)', $local_rows, $url, $post_id, $post->post_status)); | |
| 7500 | + $pruned += $local_rows; | |
| 7501 | + } | |
| 7502 | + } | |
| 7503 | + $more = $paged < $query->max_num_pages; | |
| 7504 | + $paged++; | |
| 7505 | + } while ($more); | |
| 7506 | + | |
| 7507 | + WP_CLI::success(sprintf( | |
| 7508 | + '%s %d local knowledge row(s) across %d non-published post(s) scanned.%s', | |
| 7509 | + $dry_run ? 'Would remove' : 'Removed', | |
| 7510 | + $pruned, | |
| 7511 | + $scanned, | |
| 7512 | + ' (Pinecone-mode deletions are not counted locally.)' | |
| 7513 | + )); | |
| 7514 | +} | |
| 7515 | + | |
| 7516 | +/** | |
| 7517 | + * WP-CLI: repair knowledge-base rows whose PDF text was imported in visual | |
| 7518 | + * (reversed) order before the RTL normalizer existed. 32bf9e fixed new | |
| 7519 | + * imports only; this fixes rows already in the table without the customer | |
| 7520 | + * having to re-source and re-upload the original PDFs (plan d1e6f7). | |
| 7521 | + * | |
| 7522 | + * Detection reuses MxChat_Utils::normalize_pdf_rtl() on the stored text: a | |
| 7523 | + * row is a candidate exactly when the normalizer would change it, so the | |
| 7524 | + * import-time heuristic and the repair heuristic can never disagree. | |
| 7525 | + * Repaired rows are RE-EMBEDDED — the stored vector was computed over | |
| 7526 | + * reversed text and is as broken as the text — so a wet run calls the | |
| 7527 | + * embedding provider once per repaired row on the site's API key. Runs | |
| 7528 | + * beyond 25 rows therefore require --yes. | |
| 7529 | + * | |
| 7530 | + * Scope notes: | |
| 7531 | + * - Scans the WordPress knowledge table. Pinecone-mode entries live in | |
| 7532 | + * Pinecone, not this table, and are not scanned; if a scanned row's bot | |
| 7533 | + * ALSO has Pinecone enabled (hybrid drift), the repaired entry is | |
| 7534 | + * re-submitted through the normal import path so the md5-keyed Pinecone | |
| 7535 | + * vector is replaced too. | |
| 7536 | + * - Knowledge rows do not carry a bot id; --bot only selects whose | |
| 7537 | + * embedding configuration (model + key) is used for re-embedding. | |
| 7538 | + * - The mxchat_pdf_rtl_normalize filter is honoured: a site that disabled | |
| 7539 | + * normalization gets detections of zero, not surprise rewrites. | |
| 7540 | + * - The metadata header the PDF importer stores before the text separator | |
| 7541 | + * is preserved byte-identical; only the text segment is repaired. | |
| 7542 | + * | |
| 7543 | + * ## OPTIONS | |
| 7544 | + * | |
| 7545 | + * [--dry-run] | |
| 7546 | + * : List the rows that would be repaired without changing anything. | |
| 7547 | + * | |
| 7548 | + * [--bot=<id>] | |
| 7549 | + * : Embedding configuration to use for re-embedding. Default: default. | |
| 7550 | + * | |
| 7551 | + * [--all-content] | |
| 7552 | + * : Scan every row containing right-to-left text, not just rows with PDF | |
| 7553 | + * provenance (a page anchor in the source URL, or pdf content type). | |
| 7554 | + * | |
| 7555 | + * [--yes] | |
| 7556 | + * : Proceed even when more than 25 rows need re-embedding (API cost gate). | |
| 7557 | + * | |
| 7558 | + * ## EXAMPLES | |
| 7559 | + * | |
| 7560 | + * wp mxchat rtl-repair --dry-run | |
| 7561 | + * wp mxchat rtl-repair | |
| 7562 | + * wp mxchat rtl-repair --all-content --yes | |
| 7563 | + */ | |
| 7564 | +public function cli_rtl_repair($args, $assoc_args) { | |
| 7565 | + global $wpdb; | |
| 7566 | + $dry_run = !empty($assoc_args['dry-run']); | |
| 7567 | + $all = !empty($assoc_args['all-content']); | |
| 7568 | + $yes = !empty($assoc_args['yes']); | |
| 7569 | + $bot_id = isset($assoc_args['bot']) ? sanitize_key($assoc_args['bot']) : 'default'; | |
| 7570 | + $table = $wpdb->prefix . 'mxchat_system_prompt_content'; | |
| 7571 | + | |
| 7572 | + // Detection pass — no API calls. Walk the table in id batches so a large | |
| 7573 | + // knowledge base never loads at once. | |
| 7574 | + $rtl_re = '/[\x{0590}-\x{05FF}\x{0600}-\x{06FF}\x{0750}-\x{077F}\x{FB50}-\x{FDFF}\x{FE70}-\x{FEFF}]/u'; | |
| 7575 | + $candidates = array(); | |
| 7576 | + $scanned = 0; | |
| 7577 | + $last_id = 0; | |
| 7578 | + do { | |
| 7579 | + if ($all) { | |
| 7580 | + $rows = $wpdb->get_results($wpdb->prepare( | |
| 7581 | + "SELECT id, article_content, source_url, content_type FROM {$table} | |
| 7582 | + WHERE id > %d ORDER BY id ASC LIMIT 200", | |
| 7583 | + $last_id | |
| 7584 | + )); | |
| 7585 | + } else { | |
| 7586 | + $rows = $wpdb->get_results($wpdb->prepare( | |
| 7587 | + "SELECT id, article_content, source_url, content_type FROM {$table} | |
| 7588 | + WHERE id > %d AND (source_url LIKE %s OR content_type = 'pdf') | |
| 7589 | + ORDER BY id ASC LIMIT 200", | |
| 7590 | + $last_id, | |
| 7591 | + '%' . $wpdb->esc_like('#page=') . '%' | |
| 7592 | + )); | |
| 7593 | + } | |
| 7594 | + foreach ($rows as $row) { | |
| 7595 | + $last_id = (int) $row->id; | |
| 7596 | + $scanned++; | |
| 7597 | + $content = (string) $row->article_content; | |
| 7598 | + if (!preg_match($rtl_re, $content)) { | |
| 7599 | + continue; | |
| 7600 | + } | |
| 7601 | + list($header, $text) = $this->mxchat_rtl_repair_split($content); | |
| 7602 | + $normalized = MxChat_Utils::normalize_pdf_rtl($text, 'rtl-repair row ' . $row->id); | |
| 7603 | + if (is_string($normalized) && $normalized !== $text) { | |
| 7604 | + $candidates[] = array( | |
| 7605 | + 'id' => (int) $row->id, | |
| 7606 | + 'source_url' => (string) $row->source_url, | |
| 7607 | + 'content_type' => (string) $row->content_type, | |
| 7608 | + 'new_content' => $header . $normalized, | |
| 7609 | + ); | |
| 7610 | + } | |
| 7611 | + } | |
| 7612 | + } while (count($rows) === 200); | |
| 7613 | + | |
| 7614 | + WP_CLI::log(sprintf('Scanned %d row(s); %d stored in reversed (visual) order.', $scanned, count($candidates))); | |
| 7615 | + if (empty($candidates)) { | |
| 7616 | + WP_CLI::success('No reversed RTL rows found — nothing to repair.'); | |
| 7617 | + return; | |
| 7618 | + } | |
| 7619 | + | |
| 7620 | + foreach ($candidates as $c) { | |
| 7621 | + WP_CLI::log(sprintf('%s row %d %s', $dry_run ? 'Would repair' : 'Will repair', $c['id'], $c['source_url'])); | |
| 7622 | + } | |
| 7623 | + if ($dry_run) { | |
| 7624 | + WP_CLI::success(sprintf('Dry run: %d row(s) would be repaired and re-embedded. Run without --dry-run to apply.', count($candidates))); | |
| 7625 | + return; | |
| 7626 | + } | |
| 7627 | + | |
| 7628 | + // Cost gate: re-embedding spends the customer's API budget. | |
| 7629 | + WP_CLI::log(sprintf('Re-embedding will call the embedding provider once per row — %d call(s) on this site\'s API key.', count($candidates))); | |
| 7630 | + if (count($candidates) > 25 && !$yes) { | |
| 7631 | + WP_CLI::error(sprintf('%d rows need re-embedding (more than 25). Re-run with --yes to confirm the API cost. No rows were changed.', count($candidates))); | |
| 7632 | + } | |
| 7633 | + | |
| 7634 | + $bot_options = $this->get_bot_options($bot_id); | |
| 7635 | + $options = !empty($bot_options) ? $bot_options : get_option('mxchat_options'); | |
| 7636 | + $preflight = MxChat_Utils::embedding_preflight($options); | |
| 7637 | + if (!$preflight['ok']) { | |
| 7638 | + WP_CLI::error('Embedding configuration problem: ' . $preflight['reason']); | |
| 7639 | + } | |
| 7640 | + $api_key = $preflight['api_key']; | |
| 7641 | + | |
| 7642 | + $pinecone_hybrid = $this->mxchat_rtl_repair_pinecone_enabled($bot_id); | |
| 7643 | + $repaired = 0; | |
| 7644 | + $failed = 0; | |
| 7645 | + foreach ($candidates as $c) { | |
| 7646 | + $vector = MxChat_Utils::regenerate_embedding($c['new_content'], $api_key, $bot_id); | |
| 7647 | + if (!is_array($vector)) { | |
| 7648 | + $failed++; | |
| 7649 | + $reason = is_wp_error($vector) ? $vector->get_error_message() : 'embedding request failed'; | |
| 7650 | + // Text and vector must stay consistent: never write repaired text | |
| 7651 | + // beside the stale reversed-text vector. | |
| 7652 | + WP_CLI::warning(sprintf('Row %d NOT repaired — %s. Row left unchanged.', $c['id'], $reason)); | |
| 7653 | + continue; | |
| 7654 | + } | |
| 7655 | + $wpdb->update( | |
| 7656 | + $table, | |
| 7657 | + array( | |
| 7658 | + 'article_content' => $c['new_content'], | |
| 7659 | + 'embedding_vector' => maybe_serialize($vector), | |
| 7660 | + ), | |
| 7661 | + array('id' => $c['id']), | |
| 7662 | + array('%s', '%s'), | |
| 7663 | + array('%d') | |
| 7664 | + ); | |
| 7665 | + $repaired++; | |
| 7666 | + if (class_exists('MxChat_Admin')) { | |
| 7667 | + MxChat_Admin::mxchat_log_debug('pdf_rtl_repaired', 'Stored KB row restored to logical order and re-embedded', array( | |
| 7668 | + 'row_id' => $c['id'], | |
| 7669 | + 'source_url' => $c['source_url'], | |
| 7670 | + 'bot' => $bot_id, | |
| 7671 | + )); | |
| 7672 | + } | |
| 7673 | + // Hybrid drift: the bot indexes into Pinecone but this row sat in the | |
| 7674 | + // WP table — push the repaired entry through the normal import path so | |
| 7675 | + // the md5(source_url)-keyed Pinecone vector is replaced as well. | |
| 7676 | + if ($pinecone_hybrid) { | |
| 7677 | + MxChat_Utils::submit_content_to_db( | |
| 7678 | + $c['new_content'], | |
| 7679 | + $c['source_url'], | |
| 7680 | + $api_key, | |
| 7681 | + null, | |
| 7682 | + $bot_id, | |
| 7683 | + $c['content_type'] !== '' ? $c['content_type'] : 'pdf' | |
| 7684 | + ); | |
| 7685 | + } | |
| 7686 | + } | |
| 7687 | + | |
| 7688 | + WP_CLI::success(sprintf('Repaired + re-embedded %d row(s); %d failed; %d scanned.', $repaired, $failed, $scanned)); | |
| 7689 | +} | |
| 7690 | + | |
| 7691 | +/** | |
| 7692 | + * Split a stored KB row into (metadata header incl. separator, text segment). | |
| 7693 | + * The PDF importer stores wp_json_encode($metadata) . "\n---\n" . $text — | |
| 7694 | + * repair must touch only the text and keep the header byte-identical. | |
| 7695 | + */ | |
| 7696 | +private function mxchat_rtl_repair_split($content) { | |
| 7697 | + $sep = "\n---\n"; | |
| 7698 | + $pos = strpos($content, $sep); | |
| 7699 | + if ($pos !== false && $pos > 0 && $content[0] === '{') { | |
| 7700 | + $maybe_json = substr($content, 0, $pos); | |
| 7701 | + if (json_decode($maybe_json) !== null) { | |
| 7702 | + return array(substr($content, 0, $pos + strlen($sep)), substr($content, $pos + strlen($sep))); | |
| 7703 | + } | |
| 7704 | + } | |
| 7705 | + return array('', $content); | |
| 7706 | +} | |
| 7707 | + | |
| 7708 | +/** | |
| 7709 | + * Mirror of MxChat_Utils::is_pinecone_enabled_for_bot() (private there) for | |
| 7710 | + * the repair CLI's hybrid-drift check. | |
| 7711 | + */ | |
| 7712 | +private function mxchat_rtl_repair_pinecone_enabled($bot_id) { | |
| 7713 | + if ($bot_id !== 'default' && class_exists('MxChat_Multi_Bot_Manager')) { | |
| 7714 | + $cfg = apply_filters('mxchat_get_bot_pinecone_config', array(), $bot_id); | |
| 7715 | + if (!empty($cfg)) { | |
| 7716 | + return !empty($cfg['use_pinecone']) && !empty($cfg['api_key']) && !empty($cfg['host']); | |
| 7717 | + } | |
| 7718 | + } | |
| 7719 | + $po = get_option('mxchat_pinecone_addon_options'); | |
| 7720 | + return !empty($po['mxchat_use_pinecone']) && $po['mxchat_use_pinecone'] !== '0' | |
| 7721 | + && !empty($po['mxchat_pinecone_api_key']) && !empty($po['mxchat_pinecone_host']); | |
| 7722 | +} | |
| 7723 | + | |
| 7724 | +/* ========================================================================= | |
| 7725 | + * Re-import affected entries (plan 89739b) | |
| 7726 | + * | |
| 7727 | + * Several releases corrected how knowledge text is imported, each only for | |
| 7728 | + * entries added afterwards: HTML entities (d2c92e), currency signs (7403ec), | |
| 7729 | + * combining marks and symbols (a19914), joiner characters (209e57) and | |
| 7730 | + * right-to-left PDF order (32bf9e). This walks the knowledge base, rebuilds | |
| 7731 | + * each entry's text from its source with the CURRENT import chain, and | |
| 7732 | + * rewrites an entry only when (a) the rebuilt text differs from the stored | |
| 7733 | + * text in one of those ways and (b) the two are otherwise the same document. | |
| 7734 | + * An entry whose source simply changed since it was imported is reported and | |
| 7735 | + * left alone: that is a different question from "was it damaged at import". | |
| 7736 | + * | |
| 7737 | + * One engine, two callers: wp mxchat kb-reimport and the Knowledge screen's | |
| 7738 | + * resumable action. A scan makes no embedding call and writes no entry. | |
| 7739 | + * ====================================================================== */ | |
| 7740 | + | |
| 7741 | +/** Release date of 3.2.20, the last release that changed how text is imported. */ | |
| 7742 | +const REIMPORT_DEFAULT_BEFORE = '2026-08-28'; | |
| 7743 | +const REIMPORT_STATE_OPTION = 'mxchat_kb_reimport_state'; | |
| 7744 | + | |
| 7745 | +/** Which store holds the entries for this bot: wordpress, pinecone or documents. */ | |
| 7746 | +private function mxchat_reimport_store($bot_id) { | |
| 7747 | + if (!$this->mxchat_rtl_repair_pinecone_enabled($bot_id)) { | |
| 7748 | + return 'wordpress'; | |
| 7749 | + } | |
| 7750 | + if (class_exists('MxChat_Pinecone_Documents') && MxChat_Pinecone_Documents::is_document_index($bot_id)) { | |
| 7751 | + return 'documents'; | |
| 7752 | + } | |
| 7753 | + return 'pinecone'; | |
| 7754 | +} | |
| 7755 | + | |
| 7756 | +private function mxchat_reimport_pinecone_cfg($bot_id) { | |
| 7757 | + $po = class_exists('MxChat_Pinecone_Manager') | |
| 7758 | + ? MxChat_Pinecone_Manager::get_instance()->mxchat_get_bot_pinecone_options($bot_id) | |
| 7759 | + : get_option('mxchat_pinecone_addon_options', array()); | |
| 7760 | + return array( | |
| 7761 | + 'api_key' => (string) ($po['mxchat_pinecone_api_key'] ?? ''), | |
| 7762 | + 'host' => (string) ($po['mxchat_pinecone_host'] ?? ''), | |
| 7763 | + 'namespace' => (string) ($po['mxchat_pinecone_namespace'] ?? ''), | |
| 7764 | + ); | |
| 7765 | +} | |
| 7766 | + | |
| 7767 | +private function mxchat_reimport_new_state($args) { | |
| 7768 | + $bot_id = isset($args['bot']) ? sanitize_key($args['bot']) : 'default'; | |
| 7769 | + $before = isset($args['before']) ? (string) $args['before'] : self::REIMPORT_DEFAULT_BEFORE; | |
| 7770 | + if (!preg_match('/^\d{4}-\d{2}-\d{2}$/', $before)) { | |
| 7771 | + $before = self::REIMPORT_DEFAULT_BEFORE; | |
| 7772 | + } | |
| 7773 | + global $wpdb; | |
| 7774 | + $store = $this->mxchat_reimport_store($bot_id); | |
| 7775 | + return array( | |
| 7776 | + 'phase' => 'scan', | |
| 7777 | + 'bot' => $bot_id, | |
| 7778 | + 'store' => $store, | |
| 7779 | + 'source' => isset($args['source']) && $args['source'] !== '' ? (string) $args['source'] : 'all', | |
| 7780 | + 'before' => $before, | |
| 7781 | + 'cap' => isset($args['cap']) ? (int) $args['cap'] : 100, | |
| 7782 | + 'cursor' => $store === 'wordpress' ? 0 : '', | |
| 7783 | + 'max_id' => $store === 'wordpress' ? (int) $wpdb->get_var("SELECT MAX(id) FROM {$wpdb->prefix}mxchat_system_prompt_content") : 0, | |
| 7784 | + 'scan_done' => false, | |
| 7785 | + 'scanned' => 0, | |
| 7786 | + 'unchanged' => 0, | |
| 7787 | + 'changed' => 0, | |
| 7788 | + 'skipped' => 0, | |
| 7789 | + 'candidates' => array(), | |
| 7790 | + 'notes' => array(), | |
| 7791 | + 'pos' => 0, | |
| 7792 | + 'repaired' => 0, | |
| 7793 | + 'failed' => 0, | |
| 7794 | + 'error' => '', | |
| 7795 | + 'updated' => time(), | |
| 7796 | + ); | |
| 7797 | +} | |
| 7798 | + | |
| 7799 | +/** True when the entry passes --source and --before. */ | |
| 7800 | +private function mxchat_reimport_selected(array $entry, array $state) { | |
| 7801 | + $url = (string) $entry['source_url']; | |
| 7802 | + $source = (string) $state['source']; | |
| 7803 | + if ($source === 'upload') { | |
| 7804 | + if (!MxChat_Utils::is_upload_source($url)) { | |
| 7805 | + return false; | |
| 7806 | + } | |
| 7807 | + } elseif ($source === 'manual') { | |
| 7808 | + if ($url !== '' && strpos($url, 'mxchat://') !== 0) { | |
| 7809 | + return false; | |
| 7810 | + } | |
| 7811 | + } elseif ($source !== 'all' && $source !== '') { | |
| 7812 | + if (strpos($url, $source) !== 0) { | |
| 7813 | + return false; | |
| 7814 | + } | |
| 7815 | + } | |
| 7816 | + $limit = strtotime($state['before'] . ' 00:00:00'); | |
| 7817 | + $when = $entry['updated']; | |
| 7818 | + if (is_string($when) && $when !== '' && !ctype_digit($when)) { | |
| 7819 | + $when = strtotime($when); | |
| 7820 | + } | |
| 7821 | + $when = (int) $when; | |
| 7822 | + // An entry with no date on record predates the date field: treat it as old. | |
| 7823 | + return ($when === 0) || ($limit === false) || ($when < $limit); | |
| 7824 | +} | |
| 7825 | + | |
| 7826 | +/** | |
| 7827 | + * One page of entries (not rows: a chunked entry is one entry). | |
| 7828 | + * @return array{entries:array,done:bool}|WP_Error | |
| 7829 | + */ | |
| 7830 | +private function mxchat_reimport_list_page(array &$state, $limit) { | |
| 7831 | + if ($state['store'] === 'wordpress') { | |
| 7832 | + global $wpdb; | |
| 7833 | + $table = $wpdb->prefix . 'mxchat_system_prompt_content'; | |
| 7834 | + $rows = $wpdb->get_results($wpdb->prepare( | |
| 7835 | + "SELECT MIN(id) AS first_id, MAX(source_url) AS source_url, MAX(content_type) AS content_type, | |
| 7836 | + MAX(`timestamp`) AS updated, COUNT(*) AS n | |
| 7837 | + FROM {$table} | |
| 7838 | + GROUP BY CASE WHEN source_url IS NULL OR source_url = '' THEN CONCAT('#', id) ELSE source_url END | |
| 7839 | + HAVING first_id > %d AND first_id <= %d | |
| 7840 | + ORDER BY first_id ASC LIMIT %d", | |
| 7841 | + (int) $state['cursor'], (int) $state['max_id'], (int) $limit | |
| 7842 | + )); | |
| 7843 | + $entries = array(); | |
| 7844 | + foreach ((array) $rows as $row) { | |
| 7845 | + $state['cursor'] = (int) $row->first_id; | |
| 7846 | + $entries[] = array( | |
| 7847 | + 'store' => 'wordpress', | |
| 7848 | + 'id' => (int) $row->first_id, | |
| 7849 | + 'source_url' => (string) $row->source_url, | |
| 7850 | + 'content_type' => (string) $row->content_type, | |
| 7851 | + 'updated' => (string) $row->updated, | |
| 7852 | + 'chunked' => ((int) $row->n) > 1, | |
| 7853 | + ); | |
| 7854 | + } | |
| 7855 | + return array('entries' => $entries, 'done' => count((array) $rows) < $limit); | |
| 7856 | + } | |
| 7857 | + | |
| 7858 | + $cfg = $this->mxchat_reimport_pinecone_cfg($state['bot']); | |
| 7859 | + if ($cfg['api_key'] === '' || $cfg['host'] === '') { | |
| 7860 | + return new WP_Error('pinecone_config', __('Pinecone is switched on but has no API key or host.', 'mxchat')); | |
| 7861 | + } | |
| 7862 | + $is_docs = ($state['store'] === 'documents'); | |
| 7863 | + $page = $is_docs | |
| 7864 | + ? MxChat_Pinecone_Documents::list_page('', $limit, (string) $state['cursor'], $cfg) | |
| 7865 | + : MxChat_Pinecone_Documents::classic_list_page($cfg['host'], $cfg['api_key'], $cfg['namespace'], $limit, (string) $state['cursor']); | |
| 7866 | + if (is_wp_error($page)) { | |
| 7867 | + return $page; | |
| 7868 | + } | |
| 7869 | + // A chunked entry is taken once, at its first chunk; the other chunk ids are passed over. | |
| 7870 | + $anchors = array(); | |
| 7871 | + foreach ($page['ids'] as $id) { | |
| 7872 | + if (preg_match('/_chunk_(\d+)$/', $id, $m) && (int) $m[1] !== 0) { | |
| 7873 | + continue; | |
| 7874 | + } | |
| 7875 | + $anchors[] = $id; | |
| 7876 | + } | |
| 7877 | + $records = array(); | |
| 7878 | + if (!empty($anchors)) { | |
| 7879 | + $records = $is_docs | |
| 7880 | + ? MxChat_Pinecone_Documents::fetch($anchors, $cfg) | |
| 7881 | + : MxChat_Pinecone_Documents::classic_fetch($cfg['host'], $cfg['api_key'], $cfg['namespace'], $anchors); | |
| 7882 | + if (is_wp_error($records)) { | |
| 7883 | + return $records; | |
| 7884 | + } | |
| 7885 | + } | |
| 7886 | + $entries = array(); | |
| 7887 | + foreach ($anchors as $id) { | |
| 7888 | + if (!isset($records[$id])) { | |
| 7889 | + continue; | |
| 7890 | + } | |
| 7891 | + $meta = isset($records[$id]['metadata']) && is_array($records[$id]['metadata']) ? $records[$id]['metadata'] : array(); | |
| 7892 | + $entries[] = array( | |
| 7893 | + 'store' => $state['store'], | |
| 7894 | + 'id' => (string) $id, | |
| 7895 | + 'source_url' => (string) ($meta['source_url'] ?? ''), | |
| 7896 | + 'content_type' => (string) ($meta['type'] ?? 'content'), | |
| 7897 | + 'updated' => (int) ($meta['last_updated'] ?? 0), | |
| 7898 | + 'chunked' => (bool) preg_match('/_chunk_0$/', $id), | |
| 7899 | + 'text' => (string) ($meta['text'] ?? ''), | |
| 7900 | + ); | |
| 7901 | + } | |
| 7902 | + $state['cursor'] = (string) $page['next']; | |
| 7903 | + return array('entries' => $entries, 'done' => ((string) $page['next'] === '')); | |
| 7904 | +} | |
| 7905 | + | |
| 7906 | +/** The entry's stored text, chunks joined. @return string|WP_Error */ | |
| 7907 | +private function mxchat_reimport_stored_text(array $entry, $bot_id) { | |
| 7908 | + if ($entry['store'] === 'wordpress') { | |
| 7909 | + $got = $this->get_wordpress_entry_content($entry['source_url'], (int) $entry['id']); | |
| 7910 | + } elseif (empty($entry['chunked']) && isset($entry['text'])) { | |
| 7911 | + return (string) $entry['text']; | |
| 7912 | + } else { | |
| 7913 | + // A chunked entry, or one read back from a saved scan (which keeps no text). | |
| 7914 | + $got = $this->get_pinecone_entry_content($entry['source_url'], (string) $entry['id'], $bot_id); | |
| 7915 | + } | |
| 7916 | + return is_wp_error($got) ? $got : (string) $got['content']; | |
| 7917 | +} | |
| 7918 | + | |
| 7919 | +private function mxchat_reimport_is_pdf(array $entry) { | |
| 7920 | + $url = (string) $entry['source_url']; | |
| 7921 | + return $entry['content_type'] === 'pdf' | |
| 7922 | + || strpos($url, '#page=') !== false | |
| 7923 | + || (MxChat_Utils::is_upload_source($url) && (bool) preg_match('/\.pdf$/i', (string) MxChat_Utils::upload_source_base($url))); | |
| 7924 | +} | |
| 7925 | + | |
| 7926 | +/** Text that came out of the PDF reader, page entry or whole file: reading order can apply. */ | |
| 7927 | +private function mxchat_reimport_from_pdf(array $entry) { | |
| 7928 | + return $this->mxchat_reimport_is_pdf($entry) || (bool) preg_match('/\.pdf(?:[?#]|$)/i', (string) $entry['source_url']); | |
| 7929 | +} | |
| 7930 | + | |
| 7931 | +private function mxchat_reimport_squash($text) { | |
| 7932 | + $out = preg_replace('/[\s\p{Z}]+/u', '', (string) $text); | |
| 7933 | + return is_string($out) ? $out : (string) $text; | |
| 7934 | +} | |
| 7935 | + | |
| 7936 | +/** | |
| 7937 | + * Comparison form only, never stored. Both texts are reduced to what the | |
| 7938 | + * oldest import chain kept (letters, numbers, punctuation, math symbols: | |
| 7939 | + * the allowlist that shipped until 3.2.18), with entities decoded and | |
| 7940 | + * right-to-left PDF text put in reading order. Two texts with the same form | |
| 7941 | + * are the same document, differing only in what the fixes restored. | |
| 7942 | + */ | |
| 7943 | +private function mxchat_reimport_norm($text, $is_pdf) { | |
| 7944 | + $text = (string) $text; | |
| 7945 | + for ($i = 0; $i < 3; $i++) { | |
| 7946 | + $decoded = html_entity_decode($text, ENT_QUOTES | ENT_HTML5, 'UTF-8'); | |
| 7947 | + if ($decoded === $text) { | |
| 7948 | + break; | |
| 7949 | + } | |
| 7950 | + $text = $decoded; | |
| 7951 | + } | |
| 7952 | + if ($is_pdf) { | |
| 7953 | + $logical = MxChat_Utils::normalize_pdf_rtl($text, 'kb-reimport compare'); | |
| 7954 | + if (is_string($logical)) { | |
| 7955 | + $text = $logical; | |
| 7956 | + } | |
| 7957 | + } | |
| 7958 | + $kept = preg_replace('/[^\p{L}\p{N}\p{P}\p{Sm}]+/u', '', $text); | |
| 7959 | + return is_string($kept) ? $kept : $this->mxchat_reimport_squash($text); | |
| 7960 | +} | |
| 7961 | + | |
| 7962 | +private function mxchat_reimport_count($pattern, $text) { | |
| 7963 | + $n = preg_match_all($pattern, (string) $text); | |
| 7964 | + return $n ? (int) $n : 0; | |
| 7965 | +} | |
| 7966 | + | |
| 7967 | +private function mxchat_reimport_entity_count($text) { | |
| 7968 | + if (!preg_match_all('/&(?:#[0-9]{1,7}|#x[0-9a-f]{1,6}|[a-z][a-z0-9]{1,31});/i', (string) $text, $m)) { | |
| 7969 | + return 0; | |
| 7970 | + } | |
| 7971 | + $n = 0; | |
| 7972 | + foreach ($m[0] as $entity) { | |
| 7973 | + if (html_entity_decode($entity, ENT_QUOTES | ENT_HTML5, 'UTF-8') !== $entity) { | |
| 7974 | + $n++; | |
| 7975 | + } | |
| 7976 | + } | |
| 7977 | + return $n; | |
| 7978 | +} | |
| 7979 | + | |
| 7980 | +/** Why the rebuilt text differs from the stored text, in the fixes' own terms. */ | |
| 7981 | +private function mxchat_reimport_reasons($stored, $fresh, $is_pdf) { | |
| 7982 | + $reasons = array(); | |
| 7983 | + if ($this->mxchat_reimport_entity_count($stored) > $this->mxchat_reimport_entity_count($fresh)) { | |
| 7984 | + $reasons[] = 'entities'; | |
| 7985 | + } | |
| 7986 | + if ($this->mxchat_reimport_count('/\p{Sc}/u', $fresh) > $this->mxchat_reimport_count('/\p{Sc}/u', $stored)) { | |
| 7987 | + $reasons[] = 'currency'; | |
| 7988 | + } | |
| 7989 | + if ($this->mxchat_reimport_count('/\p{M}/u', $fresh) > $this->mxchat_reimport_count('/\p{M}/u', $stored)) { | |
| 7990 | + $reasons[] = 'combining marks'; | |
| 7991 | + } | |
| 7992 | + if ($this->mxchat_reimport_count('/\p{So}/u', $fresh) > $this->mxchat_reimport_count('/\p{So}/u', $stored)) { | |
| 7993 | + $reasons[] = 'symbols'; | |
| 7994 | + } | |
| 7995 | + if ($this->mxchat_reimport_count('/[\x{200C}\x{200D}]/u', $fresh) > $this->mxchat_reimport_count('/[\x{200C}\x{200D}]/u', $stored)) { | |
| 7996 | + $reasons[] = 'joiners'; | |
| 7997 | + } | |
| 7998 | + if ($is_pdf) { | |
| 7999 | + $logical = MxChat_Utils::normalize_pdf_rtl($stored, 'kb-reimport detect'); | |
| 8000 | + if (is_string($logical) && $logical !== $stored) { | |
| 8001 | + $reasons[] = 'RTL'; | |
| 8002 | + } | |
| 8003 | + } | |
| 8004 | + return $reasons; | |
| 8005 | +} | |
| 8006 | + | |
| 8007 | +/** A product price line written before 7403ec has no currency code; one written since has. */ | |
| 8008 | +private function mxchat_reimport_price_code_missing($stored, $fresh) { | |
| 8009 | + $coded = '/^(?:Price|Sale Price|Price Range): [A-Z]{3} /m'; | |
| 8010 | + return (bool) preg_match('/^(?:Price|Sale Price|Price Range): /m', $stored) | |
| 8011 | + && !preg_match($coded, $stored) | |
| 8012 | + && (bool) preg_match($coded, $fresh); | |
| 8013 | +} | |
| 8014 | + | |
| 8015 | +/** What submit_content_to_db() does to text before it stores it. */ | |
| 8016 | +private function mxchat_reimport_storable($text) { | |
| 8017 | + $text = wp_check_invalid_utf8((string) $text); | |
| 8018 | + $out = preg_replace('/[\x00-\x08\x0B\x0C\x0E-\x1F\x7F]/u', '', $text); | |
| 8019 | + return is_string($out) ? $out : $text; | |
| 8020 | +} | |
| 8021 | + | |
| 8022 | +/** | |
| 8023 | + * Rebuild an entry's text from its source with the current import chain. | |
| 8024 | + * | |
| 8025 | + * @return array{mode:string,texts?:array,title?:string,post_id?:int,note?:string} | |
| 8026 | + * mode 'inplace' : no source to go back to (manual text, an uploaded file, a | |
| 8027 | + * PDF page) - only reading order can be repaired, in place. | |
| 8028 | + * mode 'refetch' : 'texts' holds one rebuilt text per import route that could | |
| 8029 | + * have written this entry (label => text). | |
| 8030 | + * mode 'skip' : 'note' says why. | |
| 8031 | + */ | |
| 8032 | +private function mxchat_reimport_fresh(array $entry, $bot_id) { | |
| 8033 | + $url = (string) $entry['source_url']; | |
| 8034 | + $type = (string) $entry['content_type']; | |
| 8035 | + | |
| 8036 | + if (!preg_match('#^https?://#i', $url) || $this->mxchat_reimport_is_pdf($entry)) { | |
| 8037 | + return array('mode' => 'inplace'); | |
| 8038 | + } | |
| 8039 | + // The WordPress store files any mxchat.ai address as manual text and would | |
| 8040 | + // add a second row instead of updating this one, so treat it as manual here too. | |
| 8041 | + if ($entry['store'] === 'wordpress' && strpos($url, 'mxchat.ai') !== false) { | |
| 8042 | + return array('mode' => 'inplace'); | |
| 8043 | + } | |
| 8044 | + if ($type === 'youtube') { | |
| 8045 | + return array('mode' => 'skip', 'note' => __('a video entry, not an imported page', 'mxchat'), 'quiet' => true); | |
| 8046 | + } | |
| 8047 | + | |
| 8048 | + // Imported by address (URL or sitemap import): fetch the page the way the importer does. | |
| 8049 | + if ($type === 'url' || $type === 'product') { | |
| 8050 | + $texts = array(); | |
| 8051 | + if (class_exists('WooCommerce') && (strpos($url, '/product/') !== false || strpos($url, '/shop/') !== false)) { | |
| 8052 | + $product_text = $this->mxchat_extract_woocommerce_product_content($url); | |
| 8053 | + if (!empty($product_text)) { | |
| 8054 | + $texts['product'] = $product_text; | |
| 8055 | + } | |
| 8056 | + } | |
| 8057 | + if ($type === 'url' || empty($texts)) { | |
| 8058 | + $response = wp_remote_get($url, array( | |
| 8059 | + 'timeout' => 30, | |
| 8060 | + 'redirection' => 5, | |
| 8061 | + 'user-agent' => function_exists('mxchat_ingest_user_agent') ? mxchat_ingest_user_agent() : 'MxChat', | |
| 8062 | + )); | |
| 8063 | + if (is_wp_error($response)) { | |
| 8064 | + if (empty($texts)) { | |
| 8065 | + return array('mode' => 'skip', 'note' => sprintf(__('the page could not be fetched (%s)', 'mxchat'), $response->get_error_message())); | |
| 8066 | + } | |
| 8067 | + } else { | |
| 8068 | + $code = (int) wp_remote_retrieve_response_code($response); | |
| 8069 | + if ($code !== 200) { | |
| 8070 | + if (empty($texts)) { | |
| 8071 | + return array('mode' => 'skip', 'note' => sprintf(__('the page is gone (HTTP %d)', 'mxchat'), $code)); | |
| 8072 | + } | |
| 8073 | + } elseif ($this->mxchat_is_pdf_url($url, $response)) { | |
| 8074 | + return array('mode' => 'inplace'); | |
| 8075 | + } else { | |
| 8076 | + $html = wp_remote_retrieve_body($response); | |
| 8077 | + $page = $this->mxchat_sanitize_content_for_api($this->mxchat_extract_main_content($html)); | |
| 8078 | + if ($page !== '') { | |
| 8079 | + $texts['page'] = $page; | |
| 8080 | + $title = MxChat_Utils::extract_page_title($html); | |
| 8081 | + } | |
| 8082 | + } | |
| 8083 | + } | |
| 8084 | + } | |
| 8085 | + if ($type === 'product' && !isset($texts['bulk'])) { | |
| 8086 | + $bulk = $this->mxchat_reimport_post_text($url, $bot_id, 'bulk'); | |
| 8087 | + if (is_array($bulk)) { | |
| 8088 | + $texts['bulk'] = $bulk['text']; | |
| 8089 | + } | |
| 8090 | + } | |
| 8091 | + if (empty($texts)) { | |
| 8092 | + return array('mode' => 'skip', 'note' => __('the page has no readable text now', 'mxchat')); | |
| 8093 | + } | |
| 8094 | + return array('mode' => 'refetch', 'texts' => $texts, 'title' => isset($title) ? (string) $title : ''); | |
| 8095 | + } | |
| 8096 | + | |
| 8097 | + // Everything else was built from a post, a product or a media file on this site. | |
| 8098 | + $post_id = (int) $this->mxchat_url_to_post_id_improved($url); | |
| 8099 | + $post = $post_id ? get_post($post_id) : null; | |
| 8100 | + if (!$post) { | |
| 8101 | + if ($type === 'content' || $type === '') { | |
| 8102 | + // Text an owner typed and gave an address to: there is no source to rebuild it from. | |
| 8103 | + return array('mode' => 'inplace'); | |
| 8104 | + } | |
| 8105 | + return array('mode' => 'skip', 'note' => __('its source is gone (no post or file on this site has this address)', 'mxchat')); | |
| 8106 | + } | |
| 8107 | + if ($post->post_type === 'attachment') { | |
| 8108 | + $extracted = $this->mxchat_extract_attachment_text($post_id); | |
| 8109 | + if ($extracted['text'] === '') { | |
| 8110 | + return array('mode' => 'skip', 'note' => sprintf(__('the file could not be read (%s)', 'mxchat'), $extracted['reason'])); | |
| 8111 | + } | |
| 8112 | + return array( | |
| 8113 | + 'mode' => 'refetch', | |
| 8114 | + 'texts' => array('file' => $this->mxchat_build_attachment_content($post, $extracted['text'])), | |
| 8115 | + 'post_id' => $post_id, | |
| 8116 | + 'is_file' => true, | |
| 8117 | + ); | |
| 8118 | + } | |
| 8119 | + if ($post->post_status !== 'publish') { | |
| 8120 | + return array('mode' => 'skip', 'note' => __('its source is no longer published', 'mxchat')); | |
| 8121 | + } | |
| 8122 | + | |
| 8123 | + $texts = array(); | |
| 8124 | + $routes = ($type === 'content' || $type === '') ? array('sync', 'bulk') : array('bulk', 'sync'); | |
| 8125 | + foreach ($routes as $route) { | |
| 8126 | + $built = $this->mxchat_reimport_post_text($url, $bot_id, $route, $post); | |
| 8127 | + if (is_array($built)) { | |
| 8128 | + $texts[$route] = $built['text']; | |
| 8129 | + } | |
| 8130 | + } | |
| 8131 | + if ($post->post_type === 'product' && class_exists('WooCommerce') && function_exists('wc_get_product')) { | |
| 8132 | + $product = wc_get_product($post_id); | |
| 8133 | + if ($product) { | |
| 8134 | + $from_product = array('product' => $this->mxchat_prepare_product_content_for_indexing($product)); | |
| 8135 | + // With the WooCommerce integration on, the product writer owns auto-synced product entries. | |
| 8136 | + $owns = isset($this->options['enable_woocommerce_integration']) && in_array($this->options['enable_woocommerce_integration'], array('1', 'on'), true); | |
| 8137 | + $texts = $owns ? $from_product + $texts : $texts + $from_product; | |
| 8138 | + } | |
| 8139 | + } | |
| 8140 | + if (empty($texts)) { | |
| 8141 | + return array('mode' => 'skip', 'note' => __('its source has no text now', 'mxchat')); | |
| 8142 | + } | |
| 8143 | + return array('mode' => 'refetch', 'texts' => $texts, 'post_id' => $post_id); | |
| 8144 | +} | |
| 8145 | + | |
| 8146 | +/** A post's text as the bulk import ('bulk') or auto-sync ('sync') builds it. @return array|null */ | |
| 8147 | +private function mxchat_reimport_post_text($url, $bot_id, $route, $post = null) { | |
| 8148 | + if (!$post) { | |
| 8149 | + $post_id = (int) $this->mxchat_url_to_post_id_improved($url); | |
| 8150 | + $post = $post_id ? get_post($post_id) : null; | |
| 8151 | + } | |
| 8152 | + if (!$post || $post->post_status !== 'publish' || $post->post_type === 'attachment') { | |
| 8153 | + return null; | |
| 8154 | + } | |
| 8155 | + $post_id = (int) $post->ID; | |
| 8156 | + $filtered = apply_filters('mxchat_before_process_post', $post, $route === 'sync' ? 'default' : $bot_id); | |
| 8157 | + if (!($filtered instanceof WP_Post)) { | |
| 8158 | + $filtered = $post; | |
| 8159 | + } | |
| 8160 | + $args = ($route === 'sync') | |
| 8161 | + ? array('read_display' => true, 'extract_acf_pdfs' => get_option('mxchat_auto_sync_acf_pdfs', '0') === '1', 'include_product_tabs' => false) | |
| 8162 | + : array('read_display' => false, 'extract_acf_pdfs' => get_option('mxchat_acf_pdf_extraction', '0') === '1', 'include_product_tabs' => true); | |
| 8163 | + $prepared = $this->mxchat_prepare_post_content_for_indexing($post_id, $filtered, $args); | |
| 8164 | + return array('text' => (string) $prepared['content']); | |
| 8165 | +} | |
| 8166 | + | |
| 8167 | +/** | |
| 8168 | + * Decide what to do with one entry. Makes no embedding call and writes nothing. | |
| 8169 | + * | |
| 8170 | + * @return array{outcome:string,reasons:array,note:string,mode:string,fresh:string,title:string,post_id:int,is_file:bool} | |
| 8171 | + * outcome: repair | unchanged | changed | skipped | |
| 8172 | + */ | |
| 8173 | +private function mxchat_reimport_evaluate(array $entry, $bot_id) { | |
| 8174 | + $result = array('outcome' => 'unchanged', 'reasons' => array(), 'note' => '', 'mode' => 'refetch', 'fresh' => '', 'title' => '', 'post_id' => 0, 'is_file' => false, 'quiet' => false); | |
| 8175 | + $is_pdf = $this->mxchat_reimport_from_pdf($entry); | |
| 8176 | + | |
| 8177 | + $fresh = $this->mxchat_reimport_fresh($entry, $bot_id); | |
| 8178 | + $result['mode'] = $fresh['mode']; | |
| 8179 | + if ($fresh['mode'] === 'skip') { | |
| 8180 | + $result['outcome'] = 'skipped'; | |
| 8181 | + $result['note'] = $fresh['note']; | |
| 8182 | + $result['quiet'] = !empty($fresh['quiet']); | |
| 8183 | + return $result; | |
| 8184 | + } | |
| 8185 | + | |
| 8186 | + if ($fresh['mode'] === 'inplace') { | |
| 8187 | + // Reading order is the one defect that can be repaired without the source. | |
| 8188 | + if (!$this->mxchat_reimport_is_pdf($entry) && !$this->mxchat_reimport_has_pdf_header($entry)) { | |
| 8189 | + return $result; | |
| 8190 | + } | |
| 8191 | + $rows = $this->mxchat_reimport_inplace_rows($entry, $bot_id); | |
| 8192 | + if (is_wp_error($rows)) { | |
| 8193 | + $result['outcome'] = 'skipped'; | |
| 8194 | + $result['note'] = $rows->get_error_message(); | |
| 8195 | + return $result; | |
| 8196 | + } | |
| 8197 | + foreach ($rows as $row) { | |
| 8198 | + if ($row['new'] !== $row['old']) { | |
| 8199 | + $result['outcome'] = 'repair'; | |
| 8200 | + $result['reasons'] = array('RTL'); | |
| 8201 | + return $result; | |
| 8202 | + } | |
| 8203 | + } | |
| 8204 | + return $result; | |
| 8205 | + } | |
| 8206 | + | |
| 8207 | + $stored = $this->mxchat_reimport_stored_text($entry, $bot_id); | |
| 8208 | + if (is_wp_error($stored)) { | |
| 8209 | + $result['outcome'] = 'skipped'; | |
| 8210 | + $result['note'] = $stored->get_error_message(); | |
| 8211 | + return $result; | |
| 8212 | + } | |
| 8213 | + $stored_squash = $this->mxchat_reimport_squash($stored); | |
| 8214 | + $stored_norm = $this->mxchat_reimport_norm($stored, $is_pdf); | |
| 8215 | + $stored_title = $this->mxchat_reimport_norm(strtok($stored, "\n"), false); | |
| 8216 | + | |
| 8217 | + // More than one import route can have written an entry (a product is built | |
| 8218 | + // one way by auto-sync and another by the bulk import). Take the route whose | |
| 8219 | + // text shares the most lines with what is stored. | |
| 8220 | + $stored_lines = array_flip(array_filter(array_map('trim', explode(" | |
| 8221 | +", $stored)), 'strlen')); | |
| 8222 | + $repair = null; | |
| 8223 | + $best = -1; | |
| 8224 | + foreach ($fresh['texts'] as $route => $text) { | |
| 8225 | + $text = $this->mxchat_reimport_storable($text); | |
| 8226 | + if ($this->mxchat_reimport_squash($text) === $stored_squash) { | |
| 8227 | + return $result; // identical to what one import route builds today | |
| 8228 | + } | |
| 8229 | + $reasons = $this->mxchat_reimport_reasons($stored, $text, $is_pdf); | |
| 8230 | + $same_doc = ($this->mxchat_reimport_norm($text, $is_pdf) === $stored_norm); | |
| 8231 | + if (!$same_doc && $route !== 'page' && $this->mxchat_reimport_price_code_missing($stored, $text) | |
| 8232 | + && $this->mxchat_reimport_norm(strtok($text, "\n"), false) === $stored_title) { | |
| 8233 | + // A product indexed before prices carried their currency code. | |
| 8234 | + $same_doc = true; | |
| 8235 | + if (!in_array('currency', $reasons, true)) { | |
| 8236 | + $reasons[] = 'currency'; | |
| 8237 | + } | |
| 8238 | + } | |
| 8239 | + if ($same_doc && !empty($reasons)) { | |
| 8240 | + $shared = count(array_intersect_key($stored_lines, array_flip(array_filter(array_map('trim', explode(" | |
| 8241 | +", $text)), 'strlen')))); | |
| 8242 | + if ($shared > $best) { | |
| 8243 | + $best = $shared; | |
| 8244 | + $repair = array('reasons' => $reasons, 'fresh' => $text); | |
| 8245 | + } | |
| 8246 | + } | |
| 8247 | + } | |
| 8248 | + | |
| 8249 | + if ($repair === null) { | |
| 8250 | + $result['outcome'] = 'changed'; | |
| 8251 | + $result['note'] = __('its source reads differently now, and not because of these fixes', 'mxchat'); | |
| 8252 | + return $result; | |
| 8253 | + } | |
| 8254 | + $result['outcome'] = 'repair'; | |
| 8255 | + $result['reasons'] = $repair['reasons']; | |
| 8256 | + $result['fresh'] = $repair['fresh']; | |
| 8257 | + $result['title'] = isset($fresh['title']) ? (string) $fresh['title'] : ''; | |
| 8258 | + $result['post_id'] = isset($fresh['post_id']) ? (int) $fresh['post_id'] : 0; | |
| 8259 | + $result['is_file'] = !empty($fresh['is_file']); | |
| 8260 | + return $result; | |
| 8261 | +} | |
| 8262 | + | |
| 8263 | +/** An uploaded PDF page stored before 3.2.24 has a manual identity; its header is what marks it as a PDF. */ | |
| 8264 | +private function mxchat_reimport_has_pdf_header(array $entry) { | |
| 8265 | + if ($entry['store'] !== 'wordpress') { | |
| 8266 | + return false; | |
| 8267 | + } | |
| 8268 | + global $wpdb; | |
| 8269 | + $head = (string) $wpdb->get_var($wpdb->prepare( | |
| 8270 | + "SELECT LEFT(article_content, 400) FROM {$wpdb->prefix}mxchat_system_prompt_content WHERE id = %d", | |
| 8271 | + (int) $entry['id'] | |
| 8272 | + )); | |
| 8273 | + return $head !== '' && $head[0] === '{' && stripos($head, 'pdf') !== false && strpos($head, "\n---\n") !== false; | |
| 8274 | +} | |
| 8275 | + | |
| 8276 | +/** | |
| 8277 | + * The stored pieces of an entry with reading order restored, piece by piece. | |
| 8278 | + * WordPress rows keep their metadata header byte for byte (as rtl-repair does). | |
| 8279 | + * @return array[]|WP_Error each: id, old, new | |
| 8280 | + */ | |
| 8281 | +private function mxchat_reimport_inplace_rows(array $entry, $bot_id) { | |
| 8282 | + $out = array(); | |
| 8283 | + if ($entry['store'] === 'wordpress') { | |
| 8284 | + global $wpdb; | |
| 8285 | + $table = $wpdb->prefix . 'mxchat_system_prompt_content'; | |
| 8286 | + $rows = ($entry['source_url'] !== '') | |
| 8287 | + ? $wpdb->get_results($wpdb->prepare("SELECT id, article_content FROM {$table} WHERE source_url = %s ORDER BY id ASC", $entry['source_url'])) | |
| 8288 | + : $wpdb->get_results($wpdb->prepare("SELECT id, article_content FROM {$table} WHERE id = %d", (int) $entry['id'])); | |
| 8289 | + foreach ((array) $rows as $row) { | |
| 8290 | + list($header, $text) = $this->mxchat_rtl_repair_split((string) $row->article_content); | |
| 8291 | + $logical = MxChat_Utils::normalize_pdf_rtl($text, 'kb-reimport row ' . $row->id); | |
| 8292 | + $out[] = array( | |
| 8293 | + 'id' => (int) $row->id, | |
| 8294 | + 'old' => (string) $row->article_content, | |
| 8295 | + 'new' => $header . (is_string($logical) ? $logical : $text), | |
| 8296 | + ); | |
| 8297 | + } | |
| 8298 | + return $out; | |
| 8299 | + } | |
| 8300 | + $stored = $this->mxchat_reimport_stored_text($entry, $bot_id); | |
| 8301 | + if (is_wp_error($stored)) { | |
| 8302 | + return $stored; | |
| 8303 | + } | |
| 8304 | + $logical = MxChat_Utils::normalize_pdf_rtl($stored, 'kb-reimport ' . $entry['id']); | |
| 8305 | + $out[] = array('id' => (string) $entry['id'], 'old' => $stored, 'new' => is_string($logical) ? $logical : $stored); | |
| 8306 | + return $out; | |
| 8307 | +} | |
| 8308 | + | |
| 8309 | +/** The role an entry is restricted to, read before a rewrite so it can be put back. */ | |
| 8310 | +private function mxchat_reimport_role(array $entry) { | |
| 8311 | + global $wpdb; | |
| 8312 | + if ($entry['store'] === 'wordpress') { | |
| 8313 | + $role = ($entry['source_url'] !== '') | |
| 8314 | + ? $wpdb->get_var($wpdb->prepare("SELECT role_restriction FROM {$wpdb->prefix}mxchat_system_prompt_content WHERE source_url = %s AND role_restriction IS NOT NULL AND role_restriction <> '' AND role_restriction <> 'public' LIMIT 1", $entry['source_url'])) | |
| 8315 | + : $wpdb->get_var($wpdb->prepare("SELECT role_restriction FROM {$wpdb->prefix}mxchat_system_prompt_content WHERE id = %d", (int) $entry['id'])); | |
| 8316 | + return ($role === null || $role === '') ? 'public' : (string) $role; | |
| 8317 | + } | |
| 8318 | + $base = preg_replace('/_chunk_\d+$/', '', (string) $entry['id']); | |
| 8319 | + $table = $wpdb->prefix . 'mxchat_pinecone_roles'; | |
| 8320 | + if ($wpdb->get_var($wpdb->prepare('SHOW TABLES LIKE %s', $table)) !== $table) { | |
| 8321 | + return 'public'; | |
| 8322 | + } | |
| 8323 | + $role = $wpdb->get_var($wpdb->prepare( | |
| 8324 | + "SELECT role_restriction FROM {$table} WHERE (vector_id = %s OR vector_id LIKE %s) AND role_restriction <> 'public' LIMIT 1", | |
| 8325 | + $base, $wpdb->esc_like($base . '_chunk_') . '%' | |
| 8326 | + )); | |
| 8327 | + return ($role === null || $role === '') ? 'public' : (string) $role; | |
| 8328 | +} | |
| 8329 | + | |
| 8330 | +/** Put a restriction back on every piece of a rewritten entry. */ | |
| 8331 | +private function mxchat_reimport_restore_role(array $entry, $role, $bot_id) { | |
| 8332 | + if ($role === 'public' || $entry['source_url'] === '') { | |
| 8333 | + return; | |
| 8334 | + } | |
| 8335 | + global $wpdb; | |
| 8336 | + if ($entry['store'] === 'wordpress') { | |
| 8337 | + $wpdb->update($wpdb->prefix . 'mxchat_system_prompt_content', array('role_restriction' => $role), array('source_url' => $entry['source_url']), array('%s'), array('%s')); | |
| 8338 | + if (class_exists('MxChat_Vectorstore_Manager')) { | |
| 8339 | + MxChat_Vectorstore_Manager::handle_role_change($entry['source_url'], $bot_id, $role); | |
| 8340 | + } | |
| 8341 | + return; | |
| 8342 | + } | |
| 8343 | + $base = md5($entry['source_url']); | |
| 8344 | + $ids = array($base); | |
| 8345 | + $n = MxChat_Utils::count_chunks_for_url($entry['source_url'], $bot_id); | |
| 8346 | + for ($i = 0; $i < (int) $n; $i++) { | |
| 8347 | + $ids[] = MxChat_Chunker::generate_chunk_vector_id($entry['source_url'], $i); | |
| 8348 | + } | |
| 8349 | + foreach (array_unique($ids) as $vector_id) { | |
| 8350 | + $wpdb->replace($wpdb->prefix . 'mxchat_pinecone_roles', array( | |
| 8351 | + 'vector_id' => $vector_id, | |
| 8352 | + 'role_restriction' => $role, | |
| 8353 | + 'updated_at' => current_time('mysql'), | |
| 8354 | + ), array('%s', '%s', '%s')); | |
| 8355 | + } | |
| 8356 | +} | |
| 8357 | + | |
| 8358 | +/** | |
| 8359 | + * Rewrite one entry. Re-evaluates first, so a second run (or a source that | |
| 8360 | + * changed between the scan and the confirm) cannot write something the scan | |
| 8361 | + * did not show. | |
| 8362 | + * @return array{outcome:string,reasons:array,note:string} | |
| 8363 | + */ | |
| 8364 | +private function mxchat_reimport_apply_entry(array $entry, $bot_id, $api_key) { | |
| 8365 | + $eval = $this->mxchat_reimport_evaluate($entry, $bot_id); | |
| 8366 | + if ($eval['outcome'] !== 'repair') { | |
| 8367 | + return array('outcome' => $eval['outcome'], 'reasons' => array(), 'note' => $eval['note']); | |
| 8368 | + } | |
| 8369 | + $role = $this->mxchat_reimport_role($entry); | |
| 8370 | + | |
| 8371 | + if ($eval['mode'] === 'inplace') { | |
| 8372 | + $rows = $this->mxchat_reimport_inplace_rows($entry, $bot_id); | |
| 8373 | + if (is_wp_error($rows)) { | |
| 8374 | + return array('outcome' => 'failed', 'reasons' => $eval['reasons'], 'note' => $rows->get_error_message()); | |
| 8375 | + } | |
| 8376 | + if ($entry['store'] === 'wordpress') { | |
| 8377 | + global $wpdb; | |
| 8378 | + foreach ($rows as $row) { | |
| 8379 | + if ($row['new'] === $row['old']) { | |
| 8380 | + continue; | |
| 8381 | + } | |
| 8382 | + $vector = $this->mxchat_reimport_with_backoff(function () use ($row, $api_key, $bot_id) { | |
| 8383 | + return MxChat_Utils::regenerate_embedding($row['new'], $api_key, $bot_id); | |
| 8384 | + }); | |
| 8385 | + if (!is_array($vector)) { | |
| 8386 | + // Text and vector stay consistent: no repaired text beside a stale vector. | |
| 8387 | + return array('outcome' => 'failed', 'reasons' => $eval['reasons'], 'note' => is_wp_error($vector) ? $vector->get_error_message() : __('the embedding request failed', 'mxchat')); | |
| 8388 | + } | |
| 8389 | + $wpdb->update( | |
| 8390 | + $wpdb->prefix . 'mxchat_system_prompt_content', | |
| 8391 | + array('article_content' => $row['new'], 'embedding_vector' => maybe_serialize($vector)), | |
| 8392 | + array('id' => $row['id']), | |
| 8393 | + array('%s', '%s'), | |
| 8394 | + array('%d') | |
| 8395 | + ); | |
| 8396 | + } | |
| 8397 | + return array('outcome' => 'repaired', 'reasons' => $eval['reasons'], 'note' => ''); | |
| 8398 | + } | |
| 8399 | + if ($entry['source_url'] === '') { | |
| 8400 | + // A Pinecone record with no source has no stable id to write back to. | |
| 8401 | + return array('outcome' => 'skipped', 'reasons' => array(), 'note' => __('a manual Pinecone entry: open it on the Knowledge screen and save it to repair it', 'mxchat')); | |
| 8402 | + } | |
| 8403 | + $eval['fresh'] = $rows[0]['new']; | |
| 8404 | + } | |
| 8405 | + | |
| 8406 | + $source_url = $entry['source_url']; | |
| 8407 | + $type = $entry['content_type'] !== '' ? $entry['content_type'] : 'content'; | |
| 8408 | + $fresh = $eval['fresh']; | |
| 8409 | + $title = $eval['title']; | |
| 8410 | + $result = $this->mxchat_reimport_with_backoff(function () use ($fresh, $source_url, $api_key, $bot_id, $type, $title) { | |
| 8411 | + return MxChat_Utils::submit_content_to_db($fresh, $source_url, $api_key, md5($source_url), $bot_id, $type, $title); | |
| 8412 | + }); | |
| 8413 | + if ($result !== true) { | |
| 8414 | + return array('outcome' => 'failed', 'reasons' => $eval['reasons'], 'note' => is_wp_error($result) ? $result->get_error_message() : __('the entry could not be stored', 'mxchat')); | |
| 8415 | + } | |
| 8416 | + $this->mxchat_reimport_restore_role($entry, $role, $bot_id); | |
| 8417 | + if ($eval['is_file'] && $eval['post_id']) { | |
| 8418 | + $this->mxchat_remember_attachment_signature($eval['post_id'], $source_url, $fresh); | |
| 8419 | + } | |
| 8420 | + return array('outcome' => 'repaired', 'reasons' => $eval['reasons'], 'note' => ''); | |
| 8421 | +} | |
| 8422 | + | |
| 8423 | +/** Run a write; on a rate-limit answer wait and try again, twice at most. */ | |
| 8424 | +private function mxchat_reimport_with_backoff(callable $call) { | |
| 8425 | + $result = null; | |
| 8426 | + for ($attempt = 0; $attempt < 3; $attempt++) { | |
| 8427 | + $result = $call(); | |
| 8428 | + if (!is_wp_error($result) || !preg_match('/\b429\b|rate.?limit|too many requests/i', $result->get_error_message())) { | |
| 8429 | + return $result; | |
| 8430 | + } | |
| 8431 | + if ($attempt < 2) { | |
| 8432 | + sleep($attempt === 0 ? 2 : 5); | |
| 8433 | + } | |
| 8434 | + } | |
| 8435 | + return $result; | |
| 8436 | +} | |
| 8437 | + | |
| 8438 | +private function mxchat_reimport_label(array $entry) { | |
| 8439 | + if ($entry['source_url'] !== '' && strpos($entry['source_url'], 'mxchat://') !== 0) { | |
| 8440 | + return $entry['source_url']; | |
| 8441 | + } | |
| 8442 | + return sprintf(__('Manual entry %s', 'mxchat'), $entry['id']); | |
| 8443 | +} | |
| 8444 | + | |
| 8445 | +private function mxchat_reimport_note(array &$state, $kind, array $entry, $note) { | |
| 8446 | + if ($state['cap'] > 0 && count($state['notes']) >= $state['cap']) { | |
| 8447 | + return; | |
| 8448 | + } | |
| 8449 | + $state['notes'][] = array('kind' => $kind, 'label' => $this->mxchat_reimport_label($entry), 'note' => $note); | |
| 8450 | +} | |
| 8451 | + | |
| 8452 | +/** Scan until the deadline: sort entries into repair / unchanged / changed / skipped. */ | |
| 8453 | +private function mxchat_reimport_scan_step(array &$state, $deadline) { | |
| 8454 | + while (!$state['scan_done'] && microtime(true) < $deadline) { | |
| 8455 | + $page = $this->mxchat_reimport_list_page($state, 10); | |
| 8456 | + if (is_wp_error($page)) { | |
| 8457 | + $state['error'] = $page->get_error_message(); | |
| 8458 | + $state['scan_done'] = true; | |
| 8459 | + break; | |
| 8460 | + } | |
| 8461 | + foreach ($page['entries'] as $entry) { | |
| 8462 | + if (!$this->mxchat_reimport_selected($entry, $state)) { | |
| 8463 | + continue; | |
| 8464 | + } | |
| 8465 | + $state['scanned']++; | |
| 8466 | + $eval = $this->mxchat_reimport_evaluate($entry, $state['bot']); | |
| 8467 | + if ($eval['outcome'] === 'repair') { | |
| 8468 | + unset($entry['text']); | |
| 8469 | + $entry['reasons'] = $eval['reasons']; | |
| 8470 | + $entry['label'] = $this->mxchat_reimport_label($entry); | |
| 8471 | + $state['candidates'][] = $entry; | |
| 8472 | + } elseif ($eval['outcome'] === 'unchanged') { | |
| 8473 | + $state['unchanged']++; | |
| 8474 | + } elseif ($eval['outcome'] === 'changed') { | |
| 8475 | + $state['changed']++; | |
| 8476 | + $this->mxchat_reimport_note($state, 'changed', $entry, $eval['note']); | |
| 8477 | + } elseif (!empty($eval['quiet'])) { | |
| 8478 | + $state['unchanged']++; | |
| 8479 | + } else { | |
| 8480 | + $state['skipped']++; | |
| 8481 | + $this->mxchat_reimport_note($state, 'skipped', $entry, $eval['note']); | |
| 8482 | + } | |
| 8483 | + } | |
| 8484 | + if ($page['done']) { | |
| 8485 | + $state['scan_done'] = true; | |
| 8486 | + } | |
| 8487 | + } | |
| 8488 | + if ($state['scan_done'] && $state['phase'] === 'scan') { | |
| 8489 | + $state['phase'] = 'scanned'; | |
| 8490 | + } | |
| 8491 | + $state['updated'] = time(); | |
| 8492 | +} | |
| 8493 | + | |
| 8494 | +/** Rewrite scanned candidates until the deadline (or $limit repairs). */ | |
| 8495 | +private function mxchat_reimport_apply_step(array &$state, $deadline, $api_key, $sleep_ms = 0, $limit = 0) { | |
| 8496 | + $total = count($state['candidates']); | |
| 8497 | + while ($state['pos'] < $total && microtime(true) < $deadline) { | |
| 8498 | + if ($limit > 0 && $state['repaired'] >= $limit) { | |
| 8499 | + break; | |
| 8500 | + } | |
| 8501 | + $i = $state['pos']; | |
| 8502 | + $entry = $state['candidates'][$i]; | |
| 8503 | + $done = $this->mxchat_reimport_apply_entry($entry, $state['bot'], $api_key); | |
| 8504 | + $state['candidates'][$i]['result'] = $done['outcome']; | |
| 8505 | + $state['candidates'][$i]['note'] = $done['note']; | |
| 8506 | + if ($done['outcome'] === 'repaired') { | |
| 8507 | + $state['repaired']++; | |
| 8508 | + if ($sleep_ms > 0) { | |
| 8509 | + usleep($sleep_ms * 1000); | |
| 8510 | + } | |
| 8511 | + } elseif ($done['outcome'] === 'failed') { | |
| 8512 | + $state['failed']++; | |
| 8513 | + } | |
| 8514 | + $state['pos']++; | |
| 8515 | + } | |
| 8516 | + if ($state['pos'] >= $total || ($limit > 0 && $state['repaired'] >= $limit)) { | |
| 8517 | + $state['phase'] = 'done'; | |
| 8518 | + } | |
| 8519 | + $state['updated'] = time(); | |
| 8520 | +} | |
| 8521 | + | |
| 8522 | +/** Embedding key for a bot, or WP_Error with the reason in plain words. */ | |
| 8523 | +private function mxchat_reimport_api_key($bot_id) { | |
| 8524 | + $bot_options = $this->get_bot_options($bot_id); | |
| 8525 | + $options = !empty($bot_options) ? $bot_options : get_option('mxchat_options'); | |
| 8526 | + $preflight = MxChat_Utils::embedding_preflight($options); | |
| 8527 | + if (!$preflight['ok']) { | |
| 8528 | + return new WP_Error('embedding_config', $preflight['reason']); | |
| 8529 | + } | |
| 8530 | + return $preflight['api_key']; | |
| 8531 | +} | |
| 8532 | + | |
| 8533 | +/** | |
| 8534 | + * WP-CLI: re-import knowledge entries that were damaged by an import defect | |
| 8535 | + * fixed in a later release (HTML entities left in the text, currency signs, | |
| 8536 | + * accents and other marks removed, joiner characters turned into spaces, | |
| 8537 | + * right-to-left PDF text stored backwards). | |
| 8538 | + * | |
| 8539 | + * Each entry's text is rebuilt from its source with the current importer and | |
| 8540 | + * compared with what is stored. An entry is rewritten and re-embedded only | |
| 8541 | + * when the difference is one of those defects. An entry whose source has | |
| 8542 | + * simply changed since it was imported is listed and left alone, and so is | |
| 8543 | + * an entry whose source is gone. Nothing is ever deleted. | |
| 8544 | + * | |
| 8545 | + * ## OPTIONS | |
| 8546 | + * | |
| 8547 | + * [--dry-run] | |
| 8548 | + * : List what would be rewritten, with the reason, without changing anything. | |
| 8549 | + * No embedding request is made. | |
| 8550 | + * | |
| 8551 | + * [--source=<source>] | |
| 8552 | + * : Which entries to check: all (default), upload, manual, or the start of an | |
| 8553 | + * address (for example https://example.com/docs/). | |
| 8554 | + * | |
| 8555 | + * [--before=<date>] | |
| 8556 | + * : Only entries last written before this date (YYYY-MM-DD). Default | |
| 8557 | + * 2026-08-28, the release date of the last of these fixes. | |
| 8558 | + * | |
| 8559 | + * [--bot=<id>] | |
| 8560 | + * : The bot whose knowledge base and embedding settings to use. Default: default. | |
| 8561 | + * | |
| 8562 | + * [--limit=<n>] | |
| 8563 | + * : Rewrite at most this many entries, then stop. | |
| 8564 | + * | |
| 8565 | + * [--sleep=<ms>] | |
| 8566 | + * : Pause this many milliseconds after each rewritten entry. | |
| 8567 | + * | |
| 8568 | + * [--yes] | |
| 8569 | + * : Proceed when more than 25 entries need re-embedding (API cost gate). | |
| 8570 | + * | |
| 8571 | + * ## EXAMPLES | |
| 8572 | + * | |
| 8573 | + * wp mxchat kb-reimport --dry-run | |
| 8574 | + * wp mxchat kb-reimport --source=https://example.com/docs/ --limit=50 | |
| 8575 | + * wp mxchat kb-reimport --before=2026-10-01 --yes | |
| 8576 | + */ | |
| 8577 | +public function cli_kb_reimport($args, $assoc_args) { | |
| 8578 | + $dry_run = !empty($assoc_args['dry-run']); | |
| 8579 | + $yes = !empty($assoc_args['yes']); | |
| 8580 | + $limit = isset($assoc_args['limit']) ? max(0, (int) $assoc_args['limit']) : 0; | |
| 8581 | + $sleep = isset($assoc_args['sleep']) ? max(0, (int) $assoc_args['sleep']) : 0; | |
| 8582 | + if (isset($assoc_args['before']) && !preg_match('/^\d{4}-\d{2}-\d{2}$/', (string) $assoc_args['before'])) { | |
| 8583 | + WP_CLI::error('--before takes a date as YYYY-MM-DD.'); | |
| 8584 | + } | |
| 8585 | + | |
| 8586 | + $state = $this->mxchat_reimport_new_state(array( | |
| 8587 | + 'bot' => isset($assoc_args['bot']) ? $assoc_args['bot'] : 'default', | |
| 8588 | + 'source' => isset($assoc_args['source']) ? $assoc_args['source'] : 'all', | |
| 8589 | + 'before' => isset($assoc_args['before']) ? $assoc_args['before'] : self::REIMPORT_DEFAULT_BEFORE, | |
| 8590 | + 'cap' => 0, | |
| 8591 | + )); | |
| 8592 | + $stores = array('wordpress' => 'the WordPress database', 'pinecone' => 'Pinecone', 'documents' => 'a Pinecone document index'); | |
| 8593 | + WP_CLI::log(sprintf('Checking entries in %s last written before %s (source: %s).', $stores[$state['store']], $state['before'], $state['source'])); | |
| 8594 | + | |
| 8595 | + while (!$state['scan_done']) { | |
| 8596 | + $this->mxchat_reimport_scan_step($state, microtime(true) + 20); | |
| 8597 | + } | |
| 8598 | + if ($state['error'] !== '') { | |
| 8599 | + WP_CLI::error('The scan stopped: ' . $state['error'] . ' Nothing was changed.'); | |
| 8600 | + } | |
| 8601 | + | |
| 8602 | + foreach ($state['notes'] as $note) { | |
| 8603 | + WP_CLI::log(sprintf('%s %s (%s)', $note['kind'] === 'changed' ? 'Left alone' : 'Skipped', $note['label'], $note['note'])); | |
| 8604 | + } | |
| 8605 | + foreach ($state['candidates'] as $c) { | |
| 8606 | + WP_CLI::log(sprintf('%s %s (%s)', $dry_run ? 'Would rewrite' : 'Will rewrite', $c['label'], implode(', ', $c['reasons']))); | |
| 8607 | + } | |
| 8608 | + $count = count($state['candidates']); | |
| 8609 | + $summary = sprintf('%d checked: %d to rewrite, %d already correct, %d left alone (source changed), %d skipped.', $state['scanned'], $count, $state['unchanged'], $state['changed'], $state['skipped']); | |
| 8610 | + if ($dry_run) { | |
| 8611 | + WP_CLI::success('Dry run. ' . $summary . ' Nothing was changed.'); | |
| 8612 | + return; | |
| 8613 | + } | |
| 8614 | + if ($count === 0) { | |
| 8615 | + WP_CLI::success($summary . ' Nothing to rewrite.'); | |
| 8616 | + return; | |
| 8617 | + } | |
| 8618 | + $planned = ($limit > 0) ? min($limit, $count) : $count; | |
| 8619 | + if ($planned > 25 && !$yes) { | |
| 8620 | + WP_CLI::error(sprintf('%d entries need re-embedding (more than 25). Re-run with --yes to confirm the API cost, or with --limit. Nothing was changed.', $planned)); | |
| 8621 | + } | |
| 8622 | + $api_key = $this->mxchat_reimport_api_key($state['bot']); | |
| 8623 | + if (is_wp_error($api_key)) { | |
| 8624 | + WP_CLI::error('Embedding configuration problem: ' . $api_key->get_error_message()); | |
| 8625 | + } | |
| 8626 | + | |
| 8627 | + $state['phase'] = 'apply'; | |
| 8628 | + $shown = 0; | |
| 8629 | + while ($state['phase'] === 'apply') { | |
| 8630 | + $this->mxchat_reimport_apply_step($state, microtime(true) + 20, $api_key, $sleep, $limit); | |
| 8631 | + for (; $shown < $state['pos']; $shown++) { | |
| 8632 | + $c = $state['candidates'][$shown]; | |
| 8633 | + if ($c['result'] === 'repaired') { | |
| 8634 | + WP_CLI::log(sprintf('Rewritten %s (%s)', $c['label'], implode(', ', $c['reasons']))); | |
| 8635 | + } elseif ($c['result'] === 'failed') { | |
| 8636 | + WP_CLI::warning(sprintf('Not rewritten %s: %s. Entry left unchanged.', $c['label'], $c['note'])); | |
| 8637 | + } else { | |
| 8638 | + WP_CLI::log(sprintf('No longer needed %s', $c['label'])); | |
| 8639 | + } | |
| 8640 | + } | |
| 8641 | + } | |
| 8642 | + $remaining = $count - $state['pos']; | |
| 8643 | + WP_CLI::success(sprintf('Rewrote and re-embedded %d entr%s; %d failed%s. %s', $state['repaired'], $state['repaired'] === 1 ? 'y' : 'ies', $state['failed'], $remaining > 0 ? sprintf('; %d left for the next run', $remaining) : '', $summary)); | |
| 8644 | +} | |
| 8645 | + | |
| 8646 | +/** What the Knowledge screen is shown of a state: counts, a capped list, no entry text. */ | |
| 8647 | +private function mxchat_reimport_public_state(array $state) { | |
| 8648 | + $list = array(); | |
| 8649 | + foreach (array_slice($state['candidates'], 0, 100) as $c) { | |
| 8650 | + $list[] = array( | |
| 8651 | + 'label' => $c['label'], | |
| 8652 | + 'reasons' => $c['reasons'], | |
| 8653 | + 'result' => isset($c['result']) ? $c['result'] : '', | |
| 8654 | + 'note' => isset($c['note']) ? $c['note'] : '', | |
| 8655 | + ); | |
| 8656 | + } | |
| 8657 | + return array( | |
| 8658 | + 'phase' => $state['phase'], | |
| 8659 | + 'before' => $state['before'], | |
| 8660 | + 'store' => $state['store'], | |
| 8661 | + 'scanned' => (int) $state['scanned'], | |
| 8662 | + 'to_repair' => count($state['candidates']), | |
| 8663 | + 'unchanged' => (int) $state['unchanged'], | |
| 8664 | + 'changed' => (int) $state['changed'], | |
| 8665 | + 'skipped' => (int) $state['skipped'], | |
| 8666 | + 'pos' => (int) $state['pos'], | |
| 8667 | + 'repaired' => (int) $state['repaired'], | |
| 8668 | + 'failed' => (int) $state['failed'], | |
| 8669 | + 'error' => (string) $state['error'], | |
| 8670 | + 'list' => $list, | |
| 8671 | + 'notes' => array_slice($state['notes'], 0, 100), | |
| 8672 | + ); | |
| 8673 | +} | |
| 8674 | + | |
| 8675 | +/** | |
| 8676 | + * AJAX: the Knowledge screen's "Re-import affected entries" action, in | |
| 8677 | + * resumable steps. op = scan (restart=1 begins a new scan), apply, cancel, | |
| 8678 | + * status. A scan step and an apply step each run for a few seconds and save | |
| 8679 | + * where they stopped, so a closed tab or a timeout loses nothing. | |
| 8680 | + */ | |
| 8681 | +public function ajax_mxchat_kb_reimport() { | |
| 8682 | + check_ajax_referer('mxchat_kb_reimport', 'nonce'); | |
| 8683 | + if (!current_user_can('manage_options')) { | |
| 8684 | + wp_send_json_error(array('message' => __('Permission denied.', 'mxchat'))); | |
| 8685 | + } | |
| 8686 | + $op = isset($_POST['op']) ? sanitize_key($_POST['op']) : 'status'; | |
| 8687 | + $state = get_option(self::REIMPORT_STATE_OPTION, array()); | |
| 8688 | + if (!is_array($state) || empty($state['phase'])) { | |
| 8689 | + $state = array(); | |
| 8690 | + } | |
| 8691 | + | |
| 8692 | + if ($op === 'cancel') { | |
| 8693 | + delete_option(self::REIMPORT_STATE_OPTION); | |
| 8694 | + wp_send_json_success(array('state' => array('phase' => 'idle'))); | |
| 8695 | + } | |
| 8696 | + if ($op === 'status') { | |
| 8697 | + wp_send_json_success(array('state' => empty($state) ? array('phase' => 'idle') : $this->mxchat_reimport_public_state($state))); | |
| 8698 | + } | |
| 8699 | + | |
| 8700 | + if ($op === 'scan') { | |
| 8701 | + if (!empty($_POST['restart']) || empty($state) || $state['phase'] !== 'scan') { | |
| 8702 | + $state = $this->mxchat_reimport_new_state(array( | |
| 8703 | + 'bot' => isset($_POST['bot_id']) ? sanitize_key($_POST['bot_id']) : 'default', | |
| 8704 | + 'before' => isset($_POST['before']) ? sanitize_text_field(wp_unslash($_POST['before'])) : self::REIMPORT_DEFAULT_BEFORE, | |
| 8705 | + )); | |
| 8706 | + } | |
| 8707 | + $this->mxchat_reimport_scan_step($state, microtime(true) + 8); | |
| 8708 | + update_option(self::REIMPORT_STATE_OPTION, $state, false); | |
| 8709 | + wp_send_json_success(array('state' => $this->mxchat_reimport_public_state($state))); | |
| 8710 | + } | |
| 8711 | + | |
| 8712 | + if ($op === 'apply') { | |
| 8713 | + if (empty($state) || !in_array($state['phase'], array('scanned', 'apply'), true)) { | |
| 8714 | + wp_send_json_error(array('message' => __('Check the knowledge base first.', 'mxchat'))); | |
| 8715 | + } | |
| 8716 | + $api_key = $this->mxchat_reimport_api_key($state['bot']); | |
| 8717 | + if (is_wp_error($api_key)) { | |
| 8718 | + wp_send_json_error(array('message' => $api_key->get_error_message())); | |
| 8719 | + } | |
| 8720 | + $state['phase'] = 'apply'; | |
| 8721 | + $this->mxchat_reimport_apply_step($state, microtime(true) + 8, $api_key); | |
| 8722 | + update_option(self::REIMPORT_STATE_OPTION, $state, false); | |
| 8723 | + wp_send_json_success(array('state' => $this->mxchat_reimport_public_state($state))); | |
| 8724 | + } | |
| 8725 | + | |
| 8726 | + wp_send_json_error(array('message' => __('Unknown request.', 'mxchat'))); | |
| 8727 | +} | |
| 8728 | + | |
| 8729 | +public function mxchat_handle_post_delete($post_id) { | |
| 8730 | + // Get post data before it's deleted | |
| 8731 | + $post = get_post($post_id); | |
| 8732 | + | |
| 8733 | + // Basic validation | |
| 8734 | + if (!$post || wp_is_post_revision($post_id)) { | |
| 8735 | + return; | |
| 8736 | + } | |
| 8737 | + | |
| 8738 | + $post_type = $post->post_type; | |
| 8739 | + | |
| 8740 | + // Check if sync is enabled for this post type | |
| 8741 | + $should_sync = false; | |
| 8742 | + | |
| 8743 | + // Check built-in post types first | |
| 8744 | + if ($post_type === 'post' && get_option('mxchat_auto_sync_posts') === '1') { | |
| 8745 | + $should_sync = true; | |
| 8746 | + } else if ($post_type === 'page' && get_option('mxchat_auto_sync_pages') === '1') { | |
| 8747 | + $should_sync = true; | |
| 8748 | + } else { | |
| 8749 | + // Check custom post types | |
| 8750 | + $option_name = 'mxchat_auto_sync_' . $post_type; | |
| 8751 | + if (get_option($option_name) === '1') { | |
| 8752 | + $should_sync = true; | |
| 8753 | + } | |
| 8754 | + } | |
| 8755 | + | |
| 8756 | + if (!$should_sync) { | |
| 8757 | + return; | |
| 8758 | + } | |
| 8759 | + | |
| 8760 | + // Resolve the pre-trash URL. wp_trash_post renames the slug with "__trashed" before firing | |
| 8761 | + // this hook, so get_permalink() here would return the trashed URL and md5() would miss the | |
| 8762 | + // real vector IDs stored under the original URL. | |
| 8763 | + $source_url = $this->mxchat_resolve_pre_trash_url($post_id); | |
| 8764 | + if (!$source_url) { | |
| 8765 | + //error_log('MXChat: Failed to resolve source URL for post ' . $post_id); | |
| 8766 | + return; | |
| 8767 | + } | |
| 8768 | + | |
| 8769 | + // Use chunk-aware deletion (handles both chunked and non-chunked content) | |
| 8770 | + $delete_result = MxChat_Utils::delete_chunks_for_url($source_url, 'default'); | |
| 8771 | + | |
| 8772 | + if (is_wp_error($delete_result)) { | |
| 8773 | + //error_log('MXChat: Chunk-aware deletion failed for URL: ' . $source_url . ' - ' . $delete_result->get_error_message()); | |
| 8774 | + } | |
| 8775 | + | |
| 8776 | + delete_transient('mxchat_prev_url_' . $post_id); | |
| 8777 | + delete_transient('mxchat_prev_status_' . $post_id); | |
| 8778 | +} | |
| 8779 | + | |
| 8780 | +/** | |
| 8781 | + * Resolve the source URL for a post being trashed/deleted. | |
| 8782 | + * | |
| 8783 | + * Why: wp_trash_post appends "__trashed" to the slug before the wp_trash_post action fires, so | |
| 8784 | + * get_permalink() returns a URL whose md5() won't match the vector IDs stored in Pinecone or | |
| 8785 | + * the source_url rows in the WP DB. Prefer the URL captured by mxchat_store_pre_update_status | |
| 8786 | + * (runs on pre_post_update, before the rename); fall back to stripping the __trashed suffix. | |
| 8787 | + */ | |
| 8788 | +private function mxchat_resolve_pre_trash_url($post_id) { | |
| 8789 | + $previous_url = get_transient('mxchat_prev_url_' . $post_id); | |
| 8790 | + if (!empty($previous_url)) { | |
| 8791 | + return $previous_url; | |
| 8792 | + } | |
| 8793 | + | |
| 8794 | + $current = get_permalink($post_id); | |
| 8795 | + if (!$current) { | |
| 8796 | + return ''; | |
| 8797 | + } | |
| 8798 | + return preg_replace('#__trashed(/?)$#', '$1', $current); | |
| 8799 | +} | |
| 8800 | + | |
| 8801 | + | |
| 8802 | + | |
| 8803 | +public function mxchat_handle_product_change($post_id, $post, $update) { | |
| 8804 | + if ($post->post_type !== 'product') { | |
| 8805 | + return; | |
| 8806 | + } | |
| 8807 | + | |
| 8808 | + if ($post->post_status === 'publish') { | |
| 8809 | + add_action('shutdown', function() use ($post_id) { | |
| 8810 | + $product = wc_get_product($post_id); | |
| 8811 | + if ($product) { | |
| 8812 | + $this->mxchat_store_product_embedding($product); | |
| 8813 | + } | |
| 8814 | + }); | |
| 8815 | + } | |
| 8816 | +} | |
| 8817 | + | |
| 8818 | +/** | |
| 8819 | + * Store WooCommerce product embeddings | |
| 8820 | + */ | |
| 8821 | +private function mxchat_store_product_embedding($product) { | |
| 8822 | + if (!isset($this->options['enable_woocommerce_integration']) || | |
| 8823 | + !in_array($this->options['enable_woocommerce_integration'], ['1', 'on'])) { | |
| 8824 | + return; | |
| 8825 | + } | |
| 8826 | + | |
| 8827 | + $source_url = get_permalink($product->get_id()); | |
| 8828 | + $product_id = $product->get_id(); | |
| 8829 | + | |
| 8830 | + // Build product content via the shared WC-object assembler (a3d60c) — this | |
| 8831 | + // writer owns product rows whenever the integration is on. | |
| 8832 | + $content = $this->mxchat_prepare_product_content_for_indexing($product); | |
| 8833 | + | |
| 8834 | + // Embedding decision — custom-provider-aware (plan cbd5fd); silent-return | |
| 8835 | + // shape preserved. | |
| 8836 | + $preflight = MxChat_Utils::embedding_preflight(get_option('mxchat_options')); | |
| 8837 | + if (!$preflight['ok']) { | |
| 8838 | + //error_log('MxChat Auto-sync: embedding pre-flight failed: ' . $preflight['reason']); | |
| 8839 | + return; | |
| 8840 | + } | |
| 8841 | + $api_key = $preflight['api_key']; | |
| 8842 | + | |
| 8843 | + // Use the centralized utility function for storage | |
| 8844 | + $result = MxChat_Utils::submit_content_to_db( | |
| 8845 | + $content, | |
| 8846 | + $source_url, | |
| 8847 | + $api_key, | |
| 8848 | + md5($source_url) // Vector ID for Pinecone | |
| 8849 | + ); | |
| 8850 | + | |
| 8851 | + // After successful storage, apply role restriction based on tags | |
| 8852 | + if (!is_wp_error($result)) { | |
| 8853 | + $this->apply_role_restriction_to_post($product_id, $source_url); | |
| 8854 | + } | |
| 8855 | + | |
| 8856 | + if (is_wp_error($result)) { | |
| 8857 | + //error_log('MxChat WooCommerce sync failed for product ' . $product_id . ': ' . $result->get_error_message()); | |
| 8858 | + } | |
| 8859 | +} | |
| 8860 | + | |
| 8861 | +public function mxchat_handle_product_delete($post_id) { | |
| 8862 | + if (get_post_type($post_id) !== 'product') { | |
| 8863 | + return; | |
| 8864 | + } | |
| 8865 | + | |
| 8866 | + $source_url = $this->mxchat_resolve_pre_trash_url($post_id); | |
| 8867 | + if (!$source_url) { | |
| 8868 | + return; | |
| 8869 | + } | |
| 8870 | + | |
| 8871 | + // Chunk-aware deletion (routes to Pinecone or WP DB and removes base + all chunks) | |
| 8872 | + MxChat_Utils::delete_chunks_for_url($source_url, 'default'); | |
| 8873 | + | |
| 8874 | + delete_transient('mxchat_prev_url_' . $post_id); | |
| 8875 | + delete_transient('mxchat_prev_status_' . $post_id); | |
| 8876 | +} | |
| 8877 | + | |
| 8878 | +/** | |
| 8879 | + * Handle individual Pinecone content deletion | |
| 8880 | + */ | |
| 8881 | +public function mxchat_handle_pinecone_prompt_delete() { | |
| 8882 | + // Check permissions | |
| 8883 | + if (!current_user_can('manage_options')) { | |
| 8884 | + wp_die(esc_html__('You do not have sufficient permissions.', 'mxchat')); | |
| 8885 | + } | |
| 8886 | + | |
| 8887 | + // Verify nonce | |
| 8888 | + if (!isset($_GET['_wpnonce']) || !wp_verify_nonce($_GET['_wpnonce'], 'mxchat_delete_pinecone_prompt_nonce')) { | |
| 8889 | + wp_die(esc_html__('Security check failed.', 'mxchat')); | |
| 8890 | + } | |
| 8891 | + | |
| 8892 | + $vector_id = isset($_GET['vector_id']) ? sanitize_text_field($_GET['vector_id']) : ''; | |
| 8893 | + | |
| 8894 | + if (empty($vector_id)) { | |
| 8895 | + set_transient('mxchat_admin_notice_error', | |
| 8896 | + esc_html__('Invalid vector ID.', 'mxchat'), | |
| 8897 | + 30 | |
| 8898 | + ); | |
| 8899 | + wp_safe_redirect(admin_url('admin.php?page=mxchat-prompts')); | |
| 8900 | + exit; | |
| 8901 | + } | |
| 8902 | + | |
| 8903 | + // Get Pinecone settings | |
| 8904 | + $pinecone_options = get_option('mxchat_pinecone_addon_options', array()); | |
| 8905 | + $use_pinecone = ($pinecone_options['mxchat_use_pinecone'] ?? '0') === '1'; | |
| 8906 | + | |
| 8907 | + if (!$use_pinecone || empty($pinecone_options['mxchat_pinecone_api_key'])) { | |
| 8908 | + set_transient('mxchat_admin_notice_error', | |
| 8909 | + esc_html__('Pinecone is not properly configured.', 'mxchat'), | |
| 8910 | + 30 | |
| 8911 | + ); | |
| 8912 | + wp_safe_redirect(admin_url('admin.php?page=mxchat-prompts')); | |
| 8913 | + exit; | |
| 8914 | + } | |
| 8915 | + | |
| 8916 | + // Delete from Pinecone | |
| 8917 | + $pinecone_manager = MxChat_Pinecone_Manager::get_instance(); | |
| 8918 | + $result = $pinecone_manager->mxchat_delete_from_pinecone_by_vector_id( | |
| 8919 | + $vector_id, | |
| 8920 | + $pinecone_options['mxchat_pinecone_api_key'], | |
| 8921 | + $pinecone_options['mxchat_pinecone_host'], | |
| 8922 | + $pinecone_options['mxchat_pinecone_namespace'] ?? '' | |
| 8923 | + ); | |
| 8924 | + | |
| 8925 | + if ($result['success']) { | |
| 8926 | + // Mirror the removal to the OpenAI Vector Store mapping (plan 15b5c6); | |
| 8927 | + // a chunk vector id reduces to its base entry there. | |
| 8928 | + if (class_exists('MxChat_Vectorstore_Manager')) { | |
| 8929 | + MxChat_Vectorstore_Manager::sync_delete_by_key($vector_id, 'default'); | |
| 8930 | + } | |
| 8931 | + MxChat_Utils::forget_attachment_signature_by_key($vector_id, 'default'); | |
| 8932 | + // No cache clearing needed since we removed caching | |
| 8933 | + set_transient('mxchat_admin_notice_success', | |
| 8934 | + esc_html__('Entry deleted successfully from Pinecone.', 'mxchat'), | |
| 8935 | + 30 | |
| 8936 | + ); | |
| 8937 | + } else { | |
| 8938 | + set_transient('mxchat_admin_notice_error', | |
| 8939 | + esc_html__('Failed to delete entry: ', 'mxchat') . $result['message'], | |
| 8940 | + 30 | |
| 8941 | + ); | |
| 8942 | + } | |
| 8943 | + | |
| 8944 | + wp_safe_redirect(admin_url('admin.php?page=mxchat-prompts')); | |
| 8945 | + exit; | |
| 8946 | +} | |
| 8947 | +/** | |
| 8948 | + * Handle individual Pinecone content deletion via AJAX | |
| 8949 | + */ | |
| 8950 | +public function ajax_mxchat_delete_pinecone_prompt() { | |
| 8951 | + // Verify nonce and permissions | |
| 8952 | + if (!check_ajax_referer('mxchat_delete_pinecone_prompt_nonce', 'nonce', false)) { | |
| 8953 | + wp_send_json_error('Invalid nonce'); | |
| 8954 | + exit; | |
| 8955 | + } | |
| 8956 | + | |
| 8957 | + if (!current_user_can('manage_options')) { | |
| 8958 | + wp_send_json_error('Unauthorized access'); | |
| 8959 | + exit; | |
| 8960 | + } | |
| 8961 | + | |
| 8962 | + $vector_id = isset($_POST['vector_id']) ? sanitize_text_field($_POST['vector_id']) : ''; | |
| 8963 | + $bot_id = isset($_POST['bot_id']) ? sanitize_text_field($_POST['bot_id']) : 'default'; | |
| 8964 | + | |
| 8965 | + if (empty($vector_id)) { | |
| 8966 | + wp_send_json_error('Missing vector ID'); | |
| 8967 | + exit; | |
| 8968 | + } | |
| 8969 | + | |
| 8970 | + // Get bot-specific Pinecone settings | |
| 8971 | + $pinecone_manager = MxChat_Pinecone_Manager::get_instance(); | |
| 8972 | + $pinecone_options = $pinecone_manager->mxchat_get_bot_pinecone_options($bot_id); | |
| 8973 | + | |
| 8974 | + $use_pinecone = ($pinecone_options['mxchat_use_pinecone'] ?? '0') === '1'; | |
| 8975 | + | |
| 8976 | + if (!$use_pinecone || empty($pinecone_options['mxchat_pinecone_api_key'])) { | |
| 8977 | + wp_send_json_error('Pinecone is not properly configured for bot: ' . $bot_id); | |
| 8978 | + exit; | |
| 8979 | + } | |
| 8980 | + | |
| 8981 | + // Delete from the correct Pinecone index and namespace | |
| 8982 | + $result = $pinecone_manager->mxchat_delete_from_pinecone_by_vector_id( | |
| 8983 | + $vector_id, | |
| 8984 | + $pinecone_options['mxchat_pinecone_api_key'], | |
| 8985 | + $pinecone_options['mxchat_pinecone_host'], | |
| 8986 | + $pinecone_options['mxchat_pinecone_namespace'] ?? '' | |
| 8987 | + ); | |
| 8988 | + | |
| 8989 | + if ($result['success']) { | |
| 8990 | + // Mirror the removal to the OpenAI Vector Store mapping (plan 15b5c6) | |
| 8991 | + if (class_exists('MxChat_Vectorstore_Manager')) { | |
| 8992 | + MxChat_Vectorstore_Manager::sync_delete_by_key($vector_id, $bot_id); | |
| 8993 | + } | |
| 8994 | + // A media entry removed here must lose its auto-sync signature too, | |
| 8995 | + // or the file is skipped as "unchanged" on every later save (b0ae54). | |
| 8996 | + MxChat_Utils::forget_attachment_signature_by_key($vector_id, $bot_id); | |
| 8997 | + // No cache clearing needed since we removed caching | |
| 8998 | + wp_send_json_success(array( | |
| 8999 | + 'message' => 'Entry deleted successfully from Pinecone', | |
| 9000 | + 'vector_id' => $vector_id, | |
| 9001 | + 'bot_id' => $bot_id | |
| 9002 | + )); | |
| 9003 | + } else { | |
| 9004 | + MxChat_Admin::mxchat_log_debug('pinecone_error', 'Failed to delete from Pinecone: ' . $result['message'], array('vector_id' => $vector_id, 'bot_id' => $bot_id)); | |
| 9005 | + wp_send_json_error('Failed to delete from Pinecone: ' . $result['message']); | |
| 9006 | + } | |
| 9007 | + | |
| 9008 | + exit; | |
| 9009 | +} | |
| 9010 | + | |
| 9011 | +/** | |
| 9012 | + * Handle deletion of all chunks for a given source URL via AJAX | |
| 9013 | + * Follows the same pattern as ajax_mxchat_delete_pinecone_prompt | |
| 9014 | + */ | |
| 9015 | +public function ajax_mxchat_delete_chunks_by_url() { | |
| 9016 | + // Verify nonce and permissions | |
| 9017 | + if (!check_ajax_referer('mxchat_delete_chunks_nonce', 'nonce', false)) { | |
| 9018 | + wp_send_json_error('Invalid nonce'); | |
| 9019 | + exit; | |
| 9020 | + } | |
| 9021 | + | |
| 9022 | + if (!current_user_can('manage_options')) { | |
| 9023 | + wp_send_json_error('Unauthorized access'); | |
| 9024 | + exit; | |
| 9025 | + } | |
| 9026 | + | |
| 9027 | + // Identity-preserving (plan dcc19d): esc_url_raw() empties upload:// and | |
| 9028 | + // mxchat:// identities, so a chunked uploaded file or manual entry could | |
| 9029 | + // not be deleted from its group row ("Missing source URL"). | |
| 9030 | + $source_url = $this->sanitize_entry_source_url(isset($_POST['source_url']) ? wp_unslash($_POST['source_url']) : ''); | |
| 9031 | + $data_source = isset($_POST['data_source']) ? sanitize_text_field($_POST['data_source']) : 'wordpress'; | |
| 9032 | + $bot_id = isset($_POST['bot_id']) ? sanitize_text_field($_POST['bot_id']) : 'default'; | |
| 9033 | + | |
| 9034 | + if (empty($source_url)) { | |
| 9035 | + wp_send_json_error('Missing source URL'); | |
| 9036 | + exit; | |
| 9037 | + } | |
| 9038 | + | |
| 9039 | + // Generate the base vector ID from the source URL (same as how chunks are created) | |
| 9040 | + $base_vector_id = md5($source_url); | |
| 9041 | + | |
| 9042 | + if ($data_source === 'pinecone') { | |
| 9043 | + // Get bot-specific Pinecone settings (same as working delete function) | |
| 9044 | + $pinecone_manager = MxChat_Pinecone_Manager::get_instance(); | |
| 9045 | + $pinecone_options = $pinecone_manager->mxchat_get_bot_pinecone_options($bot_id); | |
| 9046 | + | |
| 9047 | + $use_pinecone = ($pinecone_options['mxchat_use_pinecone'] ?? '0') === '1'; | |
| 9048 | + | |
| 9049 | + if (!$use_pinecone || empty($pinecone_options['mxchat_pinecone_api_key'])) { | |
| 9050 | + wp_send_json_error('Pinecone is not properly configured for bot: ' . $bot_id); | |
| 9051 | + exit; | |
| 9052 | + } | |
| 9053 | + | |
| 9054 | + $api_key = $pinecone_options['mxchat_pinecone_api_key']; | |
| 9055 | + $host = $pinecone_options['mxchat_pinecone_host']; | |
| 9056 | + $namespace = $pinecone_options['mxchat_pinecone_namespace'] ?? ''; | |
| 9057 | + | |
| 9058 | + // Document index (plan 362c31): base id + chunk ids through the documents API. | |
| 9059 | + if (class_exists('MxChat_Pinecone_Documents') && MxChat_Pinecone_Documents::is_document_options($pinecone_options)) { | |
| 9060 | + $docs_cfg = MxChat_Pinecone_Documents::cfg_from_options($pinecone_options); | |
| 9061 | + $doc_ids = array_merge(array($base_vector_id), MxChat_Pinecone_Documents::list_ids($base_vector_id . '_chunk_', $docs_cfg)); | |
| 9062 | + $doc_result = MxChat_Pinecone_Documents::delete_ids($doc_ids, $docs_cfg); | |
| 9063 | + if (is_wp_error($doc_result)) { | |
| 9064 | + MxChat_Admin::mxchat_log_debug('pinecone_error', 'Failed to delete documents from Pinecone: ' . $doc_result->get_error_message(), array('source_url' => $source_url)); | |
| 9065 | + wp_send_json_error('Failed to delete from Pinecone: ' . $doc_result->get_error_message()); | |
| 9066 | + exit; | |
| 9067 | + } | |
| 9068 | + if (class_exists('MxChat_Vectorstore_Manager')) { | |
| 9069 | + MxChat_Vectorstore_Manager::sync_delete_entry($source_url, $bot_id); | |
| 9070 | + } | |
| 9071 | + MxChat_Utils::forget_attachment_signature($source_url, $bot_id); | |
| 9072 | + wp_send_json_success(array( | |
| 9073 | + 'message' => 'All chunks deleted successfully from Pinecone', | |
| 9074 | + 'source_url' => $source_url, | |
| 9075 | + 'deleted_count' => count($doc_ids) | |
| 9076 | + )); | |
| 9077 | + exit; | |
| 9078 | + } | |
| 9079 | + | |
| 9080 | + // Collect all vector IDs to delete | |
| 9081 | + $vectors_to_delete = array(); | |
| 9082 | + | |
| 9083 | + // Add the original single-vector ID (for non-chunked content) | |
| 9084 | + $vectors_to_delete[] = $base_vector_id; | |
| 9085 | + | |
| 9086 | + // Use Pinecone list API to find all chunk vectors with this prefix | |
| 9087 | + // NOTE: Pinecone List API is a GET request with query parameters, not POST | |
| 9088 | + $prefix = $base_vector_id . '_chunk_'; | |
| 9089 | + | |
| 9090 | + $query_params = array( | |
| 9091 | + 'prefix' => $prefix, | |
| 9092 | + 'limit' => 100 | |
| 9093 | + ); | |
| 9094 | + | |
| 9095 | + if (!empty($namespace)) { | |
| 9096 | + $query_params['namespace'] = $namespace; | |
| 9097 | + } | |
| 9098 | + | |
| 9099 | + $list_url = "https://{$host}/vectors/list?" . http_build_query($query_params); | |
| 9100 | + | |
| 9101 | + $list_response = wp_remote_get($list_url, array( | |
| 9102 | + 'headers' => array( | |
| 9103 | + 'Api-Key' => $api_key, | |
| 9104 | + 'accept' => 'application/json' | |
| 9105 | + ), | |
| 9106 | + 'timeout' => 30 | |
| 9107 | + )); | |
| 9108 | + | |
| 9109 | + if (!is_wp_error($list_response)) { | |
| 9110 | + $list_body_response = wp_remote_retrieve_body($list_response); | |
| 9111 | + $list_data = json_decode($list_body_response, true); | |
| 9112 | + if (!empty($list_data['vectors'])) { | |
| 9113 | + foreach ($list_data['vectors'] as $vector) { | |
| 9114 | + if (isset($vector['id'])) { | |
| 9115 | + $vectors_to_delete[] = $vector['id']; | |
| 9116 | + } | |
| 9117 | + } | |
| 9118 | + } | |
| 9119 | + } | |
| 9120 | + | |
| 9121 | + if (empty($vectors_to_delete)) { | |
| 9122 | + // Entry already gone from Pinecone — still clear any mirrored | |
| 9123 | + // Vector Store file so it can't outlive the entry (plan 15b5c6). | |
| 9124 | + if (class_exists('MxChat_Vectorstore_Manager')) { | |
| 9125 | + MxChat_Vectorstore_Manager::sync_delete_entry($source_url, $bot_id); | |
| 9126 | + } | |
| 9127 | + MxChat_Utils::forget_attachment_signature($source_url, $bot_id); | |
| 9128 | + wp_send_json_success(array( | |
| 9129 | + 'message' => 'No vectors found to delete', | |
| 9130 | + 'source_url' => $source_url | |
| 9131 | + )); | |
| 9132 | + exit; | |
| 9133 | + } | |
| 9134 | + | |
| 9135 | + // Delete all vectors using the same endpoint as the working function | |
| 9136 | + $delete_url = "https://{$host}/vectors/delete"; | |
| 9137 | + | |
| 9138 | + $delete_body = array( | |
| 9139 | + 'ids' => $vectors_to_delete | |
| 9140 | + ); | |
| 9141 | + | |
| 9142 | + if (!empty($namespace)) { | |
| 9143 | + $delete_body['namespace'] = $namespace; | |
| 9144 | + } | |
| 9145 | + | |
| 9146 | + $delete_response = wp_remote_post($delete_url, array( | |
| 9147 | + 'headers' => array( | |
| 9148 | + 'Api-Key' => $api_key, | |
| 9149 | + 'accept' => 'application/json', | |
| 9150 | + 'content-type' => 'application/json' | |
| 9151 | + ), | |
| 9152 | + 'body' => wp_json_encode($delete_body), | |
| 9153 | + 'timeout' => 30 | |
| 9154 | + )); | |
| 9155 | + | |
| 9156 | + if (is_wp_error($delete_response)) { | |
| 9157 | + MxChat_Admin::mxchat_log_debug('pinecone_error', 'Failed to delete chunks from Pinecone: ' . $delete_response->get_error_message(), array('source_url' => $source_url)); | |
| 9158 | + wp_send_json_error('Failed to delete from Pinecone: ' . $delete_response->get_error_message()); | |
| 9159 | + exit; | |
| 9160 | + } | |
| 9161 | + | |
| 9162 | + $response_code = wp_remote_retrieve_response_code($delete_response); | |
| 9163 | + | |
| 9164 | + if ($response_code !== 200) { | |
| 9165 | + MxChat_Admin::mxchat_log_debug('pinecone_error', 'Pinecone API error (HTTP ' . $response_code . ')', array('source_url' => $source_url)); | |
| 9166 | + wp_send_json_error('Pinecone API error (HTTP ' . $response_code . ')'); | |
| 9167 | + exit; | |
| 9168 | + } | |
| 9169 | + | |
| 9170 | + // Mirror the removal to the OpenAI Vector Store (plan 15b5c6) | |
| 9171 | + if (class_exists('MxChat_Vectorstore_Manager')) { | |
| 9172 | + MxChat_Vectorstore_Manager::sync_delete_entry($source_url, $bot_id); | |
| 9173 | + } | |
| 9174 | + MxChat_Utils::forget_attachment_signature($source_url, $bot_id); | |
| 9175 | + | |
| 9176 | + wp_send_json_success(array( | |
| 9177 | + 'message' => 'All chunks deleted successfully from Pinecone', | |
| 9178 | + 'source_url' => $source_url, | |
| 9179 | + 'deleted_count' => count($vectors_to_delete) | |
| 9180 | + )); | |
| 9181 | + | |
| 9182 | + } else { | |
| 9183 | + // WordPress database deletion | |
| 9184 | + global $wpdb; | |
| 9185 | + $table_name = $wpdb->prefix . 'mxchat_system_prompt_content'; | |
| 9186 | + | |
| 9187 | + $result = $wpdb->delete( | |
| 9188 | + $table_name, | |
| 9189 | + array('source_url' => $source_url), | |
| 9190 | + array('%s') | |
| 9191 | + ); | |
| 9192 | + | |
| 9193 | + if ($result === false) { | |
| 9194 | + MxChat_Admin::mxchat_log_debug('database_error', 'Failed to delete from database: ' . $wpdb->last_error, array('source_url' => $source_url)); | |
| 9195 | + wp_send_json_error('Failed to delete from database: ' . $wpdb->last_error); | |
| 9196 | + exit; | |
| 9197 | + } | |
| 9198 | + | |
| 9199 | + // Mirror the removal to the OpenAI Vector Store (plan 15b5c6) | |
| 9200 | + if (class_exists('MxChat_Vectorstore_Manager')) { | |
| 9201 | + MxChat_Vectorstore_Manager::sync_delete_entry($source_url, $bot_id); | |
| 9202 | + } | |
| 9203 | + // The delete above removed every bot's rows for this URL. | |
| 9204 | + MxChat_Utils::forget_attachment_signature($source_url, 'default'); | |
| 9205 | + | |
| 9206 | + wp_send_json_success(array( | |
| 9207 | + 'message' => 'All chunks deleted successfully from database', | |
| 9208 | + 'source_url' => $source_url, | |
| 9209 | + 'deleted_count' => $result | |
| 9210 | + )); | |
| 9211 | + } | |
| 9212 | + | |
| 9213 | + exit; | |
| 9214 | +} | |
| 9215 | + | |
| 9216 | +/** | |
| 9217 | + * Handle individual WordPress database content deletion via AJAX | |
| 9218 | + * Mirrors the Pinecone delete handler but for WordPress database entries | |
| 9219 | + */ | |
| 9220 | +public function ajax_mxchat_delete_wordpress_prompt() { | |
| 9221 | + // Verify nonce and permissions | |
| 9222 | + if (!check_ajax_referer('mxchat_delete_wordpress_prompt_nonce', 'nonce', false)) { | |
| 9223 | + wp_send_json_error('Invalid nonce'); | |
| 9224 | + exit; | |
| 9225 | + } | |
| 9226 | + | |
| 9227 | + if (!current_user_can('manage_options')) { | |
| 9228 | + wp_send_json_error('Unauthorized access'); | |
| 9229 | + exit; | |
| 9230 | + } | |
| 9231 | + | |
| 9232 | + $entry_id = isset($_POST['entry_id']) ? intval($_POST['entry_id']) : 0; | |
| 9233 | + | |
| 9234 | + if (empty($entry_id)) { | |
| 9235 | + wp_send_json_error('Missing entry ID'); | |
| 9236 | + exit; | |
| 9237 | + } | |
| 9238 | + | |
| 9239 | + global $wpdb; | |
| 9240 | + $table_name = $wpdb->prefix . 'mxchat_system_prompt_content'; | |
| 9241 | + | |
| 9242 | + // Capture the identity BEFORE the row disappears — needed to mirror the | |
| 9243 | + // change into the Vector Store (plan 15b5c6). | |
| 9244 | + $source_url = $wpdb->get_var($wpdb->prepare( | |
| 9245 | + "SELECT source_url FROM {$table_name} WHERE id = %d", | |
| 9246 | + $entry_id | |
| 9247 | + )); | |
| 9248 | + | |
| 9249 | + // Clear cache for this entry | |
| 9250 | + wp_cache_delete('prompt_' . $entry_id, 'mxchat_prompts'); | |
| 9251 | + | |
| 9252 | + // Delete from database | |
| 9253 | + $result = $wpdb->delete( | |
| 9254 | + $table_name, | |
| 9255 | + array('id' => $entry_id), | |
| 9256 | + array('%d') | |
| 9257 | + ); | |
| 9258 | + | |
| 9259 | + if ($result !== false) { | |
| 9260 | + // Mirror to the Vector Store: if sibling rows remain (this was one | |
| 9261 | + // chunk of a larger entry) the entry's file is REFRESHED from what's | |
| 9262 | + // left; if none remain, the file is removed. | |
| 9263 | + if (!empty($source_url)) { | |
| 9264 | + $remaining = (int) $wpdb->get_var($wpdb->prepare( | |
| 9265 | + "SELECT COUNT(*) FROM {$table_name} WHERE source_url = %s", | |
| 9266 | + $source_url | |
| 9267 | + )); | |
| 9268 | + if (class_exists('MxChat_Vectorstore_Manager')) { | |
| 9269 | + if ($remaining > 0) { | |
| 9270 | + MxChat_Vectorstore_Manager::sync_upsert_entry($source_url, '', 'default'); | |
| 9271 | + } else { | |
| 9272 | + MxChat_Vectorstore_Manager::sync_delete_entry($source_url, 'default'); | |
| 9273 | + } | |
| 9274 | + } | |
| 9275 | + // Last row of a media entry gone: drop its auto-sync signature (b0ae54). | |
| 9276 | + if ($remaining === 0) { | |
| 9277 | + MxChat_Utils::forget_attachment_signature($source_url, 'default'); | |
| 9278 | + } | |
| 9279 | + } | |
| 9280 | + | |
| 9281 | + wp_send_json_success(array( | |
| 9282 | + 'message' => 'Entry deleted successfully', | |
| 9283 | + 'entry_id' => $entry_id | |
| 9284 | + )); | |
| 9285 | + } else { | |
| 9286 | + wp_send_json_error('Failed to delete entry from database'); | |
| 9287 | + } | |
| 9288 | + | |
| 9289 | + exit; | |
| 9290 | +} | |
| 9291 | + | |
| 9292 | +/** | |
| 9293 | + * Handle bulk deletion of knowledge entries via AJAX | |
| 9294 | + * Supports both Pinecone and WordPress database entries | |
| 9295 | + */ | |
| 9296 | +public function ajax_mxchat_bulk_delete_knowledge() { | |
| 9297 | + // Verify nonce and permissions | |
| 9298 | + if (!check_ajax_referer('mxchat_bulk_delete_knowledge_nonce', 'nonce', false)) { | |
| 9299 | + wp_send_json_error('Invalid nonce'); | |
| 9300 | + exit; | |
| 9301 | + } | |
| 9302 | + | |
| 9303 | + if (!current_user_can('manage_options')) { | |
| 9304 | + wp_send_json_error('Unauthorized access'); | |
| 9305 | + exit; | |
| 9306 | + } | |
| 9307 | + | |
| 9308 | + $entries = isset($_POST['entries']) ? $_POST['entries'] : array(); | |
| 9309 | + $bot_id = isset($_POST['bot_id']) ? sanitize_text_field($_POST['bot_id']) : 'default'; | |
| 9310 | + | |
| 9311 | + if (empty($entries) || !is_array($entries)) { | |
| 9312 | + wp_send_json_error('No entries provided'); | |
| 9313 | + exit; | |
| 9314 | + } | |
| 9315 | + | |
| 9316 | + // Extend execution time — bulk Pinecone operations can take a while | |
| 9317 | + if (function_exists('set_time_limit')) { | |
| 9318 | + set_time_limit(120); | |
| 9319 | + } | |
| 9320 | + | |
| 9321 | + $success_ids = array(); | |
| 9322 | + $failed_ids = array(); | |
| 9323 | + $errors = array(); | |
| 9324 | + | |
| 9325 | + global $wpdb; | |
| 9326 | + $table_name = $wpdb->prefix . 'mxchat_system_prompt_content'; | |
| 9327 | + | |
| 9328 | + // Get Pinecone manager for Pinecone deletions | |
| 9329 | + $pinecone_manager = MxChat_Pinecone_Manager::get_instance(); | |
| 9330 | + $pinecone_options = $pinecone_manager->mxchat_get_bot_pinecone_options($bot_id); | |
| 9331 | + $use_pinecone = ($pinecone_options['mxchat_use_pinecone'] ?? '0') === '1'; | |
| 9332 | + | |
| 9333 | + $api_key = $pinecone_options['mxchat_pinecone_api_key'] ?? ''; | |
| 9334 | + $host = $pinecone_options['mxchat_pinecone_host'] ?? ''; | |
| 9335 | + $namespace = $pinecone_options['mxchat_pinecone_namespace'] ?? ''; | |
| 9336 | + // Document index (plan 362c31): list + delete go through the documents API below. | |
| 9337 | + $bulk_docs = class_exists('MxChat_Pinecone_Documents') && MxChat_Pinecone_Documents::is_document_options($pinecone_options); | |
| 9338 | + $bulk_docs_cfg = $bulk_docs ? MxChat_Pinecone_Documents::cfg_from_options($pinecone_options) : array(); | |
| 9339 | + | |
| 9340 | + // ============================================= | |
| 9341 | + // PHASE 1: Collect all Pinecone vector IDs | |
| 9342 | + // and separate WordPress entries | |
| 9343 | + // ============================================= | |
| 9344 | + $pinecone_entry_ids = array(); // entry IDs that are pinecone-sourced | |
| 9345 | + $wordpress_entries = array(); // entries for WordPress DB deletion | |
| 9346 | + $all_vector_ids = array(); // all pinecone vector IDs to delete in one batch | |
| 9347 | + $vs_mirror_urls = array(); // Vector Store mirror: URLs to delete (plan 15b5c6) | |
| 9348 | + $vs_mirror_keys = array(); // Vector Store mirror: bare vector ids to delete | |
| 9349 | + | |
| 9350 | + foreach ($entries as $entry) { | |
| 9351 | + $entry_id = sanitize_text_field($entry['id'] ?? ''); | |
| 9352 | + $source = sanitize_text_field($entry['source'] ?? 'wordpress'); | |
| 9353 | + $source_url = isset($entry['sourceUrl']) ? esc_url_raw($entry['sourceUrl']) : ''; | |
| 9354 | + $is_group = isset($entry['isGroup']) && ($entry['isGroup'] === true || $entry['isGroup'] === 'true'); | |
| 9355 | + | |
| 9356 | + if (empty($entry_id)) { | |
| 9357 | + continue; | |
| 9358 | + } | |
| 9359 | + | |
| 9360 | + if ($source === 'pinecone') { | |
| 9361 | + if (!$use_pinecone || empty($api_key)) { | |
| 9362 | + $failed_ids[] = $entry_id; | |
| 9363 | + $errors[] = "Pinecone not configured for entry: $entry_id"; | |
| 9364 | + continue; | |
| 9365 | + } | |
| 9366 | + | |
| 9367 | + $pinecone_entry_ids[] = $entry_id; | |
| 9368 | + | |
| 9369 | + if ($is_group && !empty($source_url)) { | |
| 9370 | + // Grouped/chunked entry: collect base ID + chunk IDs via List API | |
| 9371 | + $base_vector_id = md5($source_url); | |
| 9372 | + $all_vector_ids[] = $base_vector_id; | |
| 9373 | + | |
| 9374 | + if ($bulk_docs) { | |
| 9375 | + $all_vector_ids = array_merge($all_vector_ids, MxChat_Pinecone_Documents::list_ids($base_vector_id . '_chunk_', $bulk_docs_cfg)); | |
| 9376 | + } else { | |
| 9377 | + $list_url = 'https://' . $host . '/vectors/list?prefix=' . urlencode($base_vector_id . '_chunk_') . '&limit=100'; | |
| 9378 | + if (!empty($namespace)) { | |
| 9379 | + $list_url .= '&namespace=' . rawurlencode($namespace); | |
| 9380 | + } | |
| 9381 | + $list_response = wp_remote_get($list_url, array( | |
| 9382 | + 'headers' => array( | |
| 9383 | + 'Api-Key' => $api_key, | |
| 9384 | + 'accept' => 'application/json' | |
| 9385 | + ), | |
| 9386 | + 'timeout' => 30 | |
| 9387 | + )); | |
| 9388 | + | |
| 9389 | + if (!is_wp_error($list_response)) { | |
| 9390 | + $list_body = json_decode(wp_remote_retrieve_body($list_response), true); | |
| 9391 | + if (!empty($list_body['vectors']) && is_array($list_body['vectors'])) { | |
| 9392 | + foreach ($list_body['vectors'] as $vector) { | |
| 9393 | + if (isset($vector['id'])) { | |
| 9394 | + $all_vector_ids[] = $vector['id']; | |
| 9395 | + } | |
| 9396 | + } | |
| 9397 | + } | |
| 9398 | + } | |
| 9399 | + } // end classic (vector index) list | |
| 9400 | + } else { | |
| 9401 | + // Single entry: the entry_id IS the vector ID | |
| 9402 | + $all_vector_ids[] = $entry_id; | |
| 9403 | + } | |
| 9404 | + | |
| 9405 | + if (!empty($source_url)) { | |
| 9406 | + $vs_mirror_urls[] = $source_url; | |
| 9407 | + } else { | |
| 9408 | + $vs_mirror_keys[] = $entry_id; | |
| 9409 | + } | |
| 9410 | + } else { | |
| 9411 | + $wordpress_entries[] = $entry; | |
| 9412 | + } | |
| 9413 | + } | |
| 9414 | + | |
| 9415 | + // ============================================= | |
| 9416 | + // PHASE 2: Single batch delete to Pinecone | |
| 9417 | + // ============================================= | |
| 9418 | + if (!empty($all_vector_ids)) { | |
| 9419 | + $all_vector_ids = array_values(array_unique($all_vector_ids)); | |
| 9420 | + $pinecone_success = true; | |
| 9421 | + $batches = array_chunk($all_vector_ids, 100); | |
| 9422 | + | |
| 9423 | + // Document index (plan 362c31): one documents/delete call for every id. | |
| 9424 | + if ($bulk_docs) { | |
| 9425 | + $doc_result = MxChat_Pinecone_Documents::delete_ids($all_vector_ids, $bulk_docs_cfg); | |
| 9426 | + if (is_wp_error($doc_result)) { | |
| 9427 | + $pinecone_success = false; | |
| 9428 | + $errors[] = 'Pinecone batch deletion failed: ' . $doc_result->get_error_message(); | |
| 9429 | + MxChat_Admin::mxchat_log_debug('pinecone_error', 'Bulk delete (document index) failed: ' . $doc_result->get_error_message()); | |
| 9430 | + } | |
| 9431 | + } else | |
| 9432 | + foreach ($batches as $batch) { | |
| 9433 | + $delete_body = array('ids' => $batch); | |
| 9434 | + if (!empty($namespace)) { | |
| 9435 | + $delete_body['namespace'] = $namespace; | |
| 9436 | + } | |
| 9437 | + $delete_response = wp_remote_post("https://{$host}/vectors/delete", array( | |
| 9438 | + 'headers' => array( | |
| 9439 | + 'Api-Key' => $api_key, | |
| 9440 | + 'accept' => 'application/json', | |
| 9441 | + 'content-type' => 'application/json' | |
| 9442 | + ), | |
| 9443 | + 'body' => wp_json_encode($delete_body), | |
| 9444 | + 'timeout' => 60 | |
| 9445 | + )); | |
| 9446 | + | |
| 9447 | + if (is_wp_error($delete_response)) { | |
| 9448 | + $pinecone_success = false; | |
| 9449 | + $errors[] = 'Pinecone batch deletion failed: ' . $delete_response->get_error_message(); | |
| 9450 | + MxChat_Admin::mxchat_log_debug('pinecone_error', 'Bulk delete batch failed: ' . $delete_response->get_error_message()); | |
| 9451 | + } else { | |
| 9452 | + $response_code = wp_remote_retrieve_response_code($delete_response); | |
| 9453 | + if ($response_code !== 200) { | |
| 9454 | + $pinecone_success = false; | |
| 9455 | + $response_body = wp_remote_retrieve_body($delete_response); | |
| 9456 | + $errors[] = "Pinecone API error (HTTP $response_code)"; | |
| 9457 | + MxChat_Admin::mxchat_log_debug('pinecone_error', 'Bulk delete batch failed (HTTP ' . $response_code . ')', array('response' => substr($response_body, 0, 200))); | |
| 9458 | + } | |
| 9459 | + } | |
| 9460 | + } | |
| 9461 | + | |
| 9462 | + // Mark all pinecone entries based on batch result | |
| 9463 | + foreach ($pinecone_entry_ids as $eid) { | |
| 9464 | + if ($pinecone_success) { | |
| 9465 | + $success_ids[] = $eid; | |
| 9466 | + } else { | |
| 9467 | + $failed_ids[] = $eid; | |
| 9468 | + } | |
| 9469 | + } | |
| 9470 | + | |
| 9471 | + // Mirror the removals to the OpenAI Vector Store (plan 15b5c6) | |
| 9472 | + if ($pinecone_success && class_exists('MxChat_Vectorstore_Manager')) { | |
| 9473 | + foreach (array_unique($vs_mirror_urls) as $vs_url) { | |
| 9474 | + MxChat_Vectorstore_Manager::sync_delete_entry($vs_url, $bot_id); | |
| 9475 | + } | |
| 9476 | + foreach (array_unique($vs_mirror_keys) as $vs_key) { | |
| 9477 | + MxChat_Vectorstore_Manager::sync_delete_by_key($vs_key, $bot_id); | |
| 9478 | + } | |
| 9479 | + } | |
| 9480 | + // Media entries removed here lose their auto-sync signature (b0ae54). | |
| 9481 | + if ($pinecone_success) { | |
| 9482 | + foreach (array_unique($vs_mirror_urls) as $vs_url) { | |
| 9483 | + MxChat_Utils::forget_attachment_signature($vs_url, $bot_id); | |
| 9484 | + } | |
| 9485 | + foreach (array_unique($vs_mirror_keys) as $vs_key) { | |
| 9486 | + MxChat_Utils::forget_attachment_signature_by_key($vs_key, $bot_id); | |
| 9487 | + } | |
| 9488 | + } | |
| 9489 | + } | |
| 9490 | + | |
| 9491 | + // ============================================= | |
| 9492 | + // PHASE 3: WordPress database deletions | |
| 9493 | + // ============================================= | |
| 9494 | + foreach ($wordpress_entries as $entry) { | |
| 9495 | + $entry_id = sanitize_text_field($entry['id'] ?? ''); | |
| 9496 | + $source_url = isset($entry['sourceUrl']) ? esc_url_raw($entry['sourceUrl']) : ''; | |
| 9497 | + $is_group = isset($entry['isGroup']) && ($entry['isGroup'] === true || $entry['isGroup'] === 'true'); | |
| 9498 | + | |
| 9499 | + if (empty($entry_id)) { | |
| 9500 | + continue; | |
| 9501 | + } | |
| 9502 | + | |
| 9503 | + try { | |
| 9504 | + if ($is_group && !empty($source_url)) { | |
| 9505 | + $result = $wpdb->delete( | |
| 9506 | + $table_name, | |
| 9507 | + array('source_url' => $source_url), | |
| 9508 | + array('%s') | |
| 9509 | + ); | |
| 9510 | + $row_url = $source_url; | |
| 9511 | + } else { | |
| 9512 | + // Identity captured pre-delete for the Vector Store mirror | |
| 9513 | + $row_url = $wpdb->get_var($wpdb->prepare( | |
| 9514 | + "SELECT source_url FROM {$table_name} WHERE id = %d", | |
| 9515 | + intval($entry_id) | |
| 9516 | + )); | |
| 9517 | + wp_cache_delete('prompt_' . $entry_id, 'mxchat_prompts'); | |
| 9518 | + $result = $wpdb->delete( | |
| 9519 | + $table_name, | |
| 9520 | + array('id' => intval($entry_id)), | |
| 9521 | + array('%d') | |
| 9522 | + ); | |
| 9523 | + } | |
| 9524 | + | |
| 9525 | + if ($result !== false) { | |
| 9526 | + $success_ids[] = $entry_id; | |
| 9527 | + // Mirror to the Vector Store: refresh the entry's file when | |
| 9528 | + // sibling chunk rows survive, remove it when none do. | |
| 9529 | + if (!empty($row_url)) { | |
| 9530 | + $remaining = (int) $wpdb->get_var($wpdb->prepare( | |
| 9531 | + "SELECT COUNT(*) FROM {$table_name} WHERE source_url = %s", | |
| 9532 | + $row_url | |
| 9533 | + )); | |
| 9534 | + if (class_exists('MxChat_Vectorstore_Manager')) { | |
| 9535 | + if ($remaining > 0) { | |
| 9536 | + MxChat_Vectorstore_Manager::sync_upsert_entry($row_url, '', $bot_id); | |
| 9537 | + } else { | |
| 9538 | + MxChat_Vectorstore_Manager::sync_delete_entry($row_url, $bot_id); | |
| 9539 | + } | |
| 9540 | + } | |
| 9541 | + if ($remaining === 0) { | |
| 9542 | + MxChat_Utils::forget_attachment_signature($row_url, 'default'); | |
| 9543 | + } | |
| 9544 | + } | |
| 9545 | + } else { | |
| 9546 | + $failed_ids[] = $entry_id; | |
| 9547 | + $errors[] = "Database error for entry: $entry_id"; | |
| 9548 | + } | |
| 9549 | + } catch (Exception $e) { | |
| 9550 | + $failed_ids[] = $entry_id; | |
| 9551 | + $errors[] = $e->getMessage(); | |
| 9552 | + } | |
| 9553 | + } | |
| 9554 | + | |
| 9555 | + wp_send_json_success(array( | |
| 9556 | + 'success_ids' => $success_ids, | |
| 9557 | + 'failed_ids' => $failed_ids, | |
| 9558 | + 'errors' => $errors, | |
| 9559 | + 'total_processed' => count($success_ids) + count($failed_ids) | |
| 9560 | + )); | |
| 9561 | + | |
| 9562 | + exit; | |
| 9563 | +} | |
| 9564 | + | |
| 9565 | +/** | |
| 9566 | + * Get the roles offered for a knowledge-entry restriction. | |
| 9567 | + * | |
| 9568 | + * Built from the site's ACTUAL roles (wp_roles()->get_names()) rather than a | |
| 9569 | + * hardcoded five, so roles added by User Role Editor / Members / a membership | |
| 9570 | + * plugin can be restricted to. The two virtual levels stay first, and the | |
| 9571 | + * built-in five keep their existing keys AND their existing labels, so | |
| 9572 | + * restrictions already stored on entries continue to mean what they meant. | |
| 9573 | + */ | |
| 9574 | +public function mxchat_get_role_options() { | |
| 9575 | + return self::mxchat_role_options(); | |
| 9576 | +} | |
| 9577 | + | |
| 9578 | +/** | |
| 9579 | + * The same list, callable without an instance. | |
| 9580 | + * | |
| 9581 | + * The constructor registers hooks, so anything outside this class that needs | |
| 9582 | + * the role list (the post meta box) must not instantiate one just to read it. | |
| 9583 | + */ | |
| 9584 | +public static function mxchat_role_options() { | |
| 9585 | + // Existing labels, in the existing order. A built-in that a site has | |
| 9586 | + // removed is simply not offered. | |
| 9587 | + $builtin_labels = array( | |
| 9588 | + 'subscriber' => __('Subscribers & Above', 'mxchat'), | |
| 9589 | + 'contributor' => __('Contributors & Above', 'mxchat'), | |
| 9590 | + 'author' => __('Authors & Above', 'mxchat'), | |
| 9591 | + 'editor' => __('Editors & Above', 'mxchat'), | |
| 9592 | + 'administrator' => __('Administrators Only', 'mxchat'), | |
| 9593 | + ); | |
| 9594 | + | |
| 9595 | + $options = array( | |
| 9596 | + 'public' => __('Public (Everyone)', 'mxchat'), | |
| 9597 | + 'logged_in' => __('Logged In Users', 'mxchat'), | |
| 9598 | + ); | |
| 9599 | + | |
| 9600 | + $site_roles = function_exists('wp_roles') ? wp_roles()->get_names() : array(); | |
| 9601 | + if (empty($site_roles) || !is_array($site_roles)) { | |
| 9602 | + // No roles readable (very early boot): fall back to the built-in set so | |
| 9603 | + // the dropdown is never empty. | |
| 9604 | + return array_merge($options, $builtin_labels); | |
| 9605 | + } | |
| 9606 | + | |
| 9607 | + foreach ($builtin_labels as $slug => $label) { | |
| 9608 | + if (isset($site_roles[$slug])) { | |
| 9609 | + $options[$slug] = $label; | |
| 9610 | + } | |
| 9611 | + } | |
| 9612 | + | |
| 9613 | + // Everything else the site defines, alphabetical by display name. | |
| 9614 | + $custom = array(); | |
| 9615 | + foreach ($site_roles as $slug => $name) { | |
| 9616 | + if (!isset($builtin_labels[$slug])) { | |
| 9617 | + $custom[$slug] = translate_user_role($name); | |
| 9618 | + } | |
| 9619 | + } | |
| 9620 | + asort($custom); | |
| 9621 | + | |
| 9622 | + return array_merge($options, $custom); | |
| 9623 | +} | |
| 9624 | + | |
| 9625 | +/** | |
| 9626 | + * Check if the current user may see content carrying a role restriction. | |
| 9627 | + * | |
| 9628 | + * Access is granted when the user holds the required role outright, or holds | |
| 9629 | + * a role whose capabilities are a superset of the required role's. That keeps | |
| 9630 | + * the built-in "& Above" ladder exactly as it was — administrator's caps | |
| 9631 | + * cover editor's, editor's cover author's, and so on down to subscriber — | |
| 9632 | + * while working for any custom role, which a hardcoded five-name hierarchy | |
| 9633 | + * could not. | |
| 9634 | + * | |
| 9635 | + * The old hierarchy scored an unknown role 0, so restricting an entry TO a | |
| 9636 | + * custom role silently granted it to every logged-in user. Fails closed now. | |
| 9637 | + */ | |
| 9638 | +public function mxchat_user_has_content_access($role_restriction) { | |
| 9639 | + // Public content is always accessible | |
| 9640 | + if ($role_restriction === 'public' || empty($role_restriction)) { | |
| 9641 | + return true; | |
| 9642 | + } | |
| 9643 | + | |
| 9644 | + // Check if user is logged in for logged_in restriction | |
| 9645 | + if ($role_restriction === 'logged_in') { | |
| 9646 | + return is_user_logged_in(); | |
| 9647 | + } | |
| 9648 | + | |
| 9649 | + // If not logged in, no access to role-restricted content | |
| 9650 | + if (!is_user_logged_in()) { | |
| 9651 | + return false; | |
| 9652 | + } | |
| 9653 | + | |
| 9654 | + $user = wp_get_current_user(); | |
| 9655 | + $user_roles = (array) $user->roles; | |
| 9656 | + | |
| 9657 | + if (empty($user_roles)) { | |
| 9658 | + return false; | |
| 9659 | + } | |
| 9660 | + | |
| 9661 | + // Holding the required role is always enough. | |
| 9662 | + if (in_array($role_restriction, $user_roles, true)) { | |
| 9663 | + return true; | |
| 9664 | + } | |
| 9665 | + | |
| 9666 | + // Site administrators keep the blanket access the old level-5 ladder gave | |
| 9667 | + // them. This also stops a restriction naming a since-deleted role from | |
| 9668 | + // locking the owner out of their own content. | |
| 9669 | + if (current_user_can('manage_options')) { | |
| 9670 | + return true; | |
| 9671 | + } | |
| 9672 | + | |
| 9673 | + foreach ($user_roles as $user_role) { | |
| 9674 | + if (self::mxchat_role_covers($user_role, $role_restriction)) { | |
| 9675 | + return true; | |
| 9676 | + } | |
| 9677 | + } | |
| 9678 | + | |
| 9679 | + return false; | |
| 9680 | +} | |
| 9681 | + | |
| 9682 | +/** | |
| 9683 | + * Does $candidate_role grant everything $required_role grants? | |
| 9684 | + * | |
| 9685 | + * Compared on capabilities, not names. Memoised per request because this runs | |
| 9686 | + * once per retrieved knowledge row. | |
| 9687 | + * | |
| 9688 | + * A required role that does not exist, or that grants nothing at all, is NOT | |
| 9689 | + * treated as covered by anything — an empty capability set is a superset of | |
| 9690 | + * itself for every role on the site, which would quietly make the restriction | |
| 9691 | + * meaningless. Those cases fall back to the exact-role check in the caller. | |
| 9692 | + * | |
| 9693 | + * @param string $candidate_role Role slug the user holds. | |
| 9694 | + * @param string $required_role Role slug the entry is restricted to. | |
| 9695 | + * @return bool | |
| 9696 | + */ | |
| 9697 | +private static function mxchat_role_covers($candidate_role, $required_role) { | |
| 9698 | + static $cache = array(); | |
| 9699 | + | |
| 9700 | + $key = $candidate_role . '|' . $required_role; | |
| 9701 | + if (isset($cache[$key])) { | |
| 9702 | + return $cache[$key]; | |
| 9703 | + } | |
| 9704 | + | |
| 9705 | + $cache[$key] = false; | |
| 9706 | + | |
| 9707 | + if (!function_exists('wp_roles')) { | |
| 9708 | + return false; | |
| 9709 | + } | |
| 9710 | + | |
| 9711 | + $roles = wp_roles(); | |
| 9712 | + $candidate = $roles->get_role($candidate_role); | |
| 9713 | + $required = $roles->get_role($required_role); | |
| 9714 | + | |
| 9715 | + if (!$candidate || !$required) { | |
| 9716 | + return false; | |
| 9717 | + } | |
| 9718 | + | |
| 9719 | + $required_caps = self::mxchat_real_caps($required); | |
| 9720 | + $candidate_caps = self::mxchat_real_caps($candidate); | |
| 9721 | + | |
| 9722 | + if (empty($required_caps)) { | |
| 9723 | + return false; | |
| 9724 | + } | |
| 9725 | + | |
| 9726 | + foreach (array_keys($required_caps) as $cap) { | |
| 9727 | + if (empty($candidate_caps[$cap])) { | |
| 9728 | + return false; | |
| 9729 | + } | |
| 9730 | + } | |
| 9731 | + | |
| 9732 | + $cache[$key] = true; | |
| 9733 | + return true; | |
| 9734 | +} | |
| 9735 | + | |
| 9736 | +/** | |
| 9737 | + * A role's granted capabilities, minus the vestigial user-level ones. | |
| 9738 | + * | |
| 9739 | + * WordPress still stamps level_0 ... level_10 onto its built-in roles for | |
| 9740 | + * back-compat with user levels, deprecated since 3.0. They carry no meaning | |
| 9741 | + * and a hand-built custom role usually has none of them — so comparing them | |
| 9742 | + * makes a role that genuinely does more than a subscriber fail to cover | |
| 9743 | + * subscriber, purely because it lacks level_0. Measured on this box: a | |
| 9744 | + * Teacher role with read/edit_posts/delete_posts/upload_files was denied | |
| 9745 | + * subscriber-restricted content for exactly that reason. | |
| 9746 | + * | |
| 9747 | + * @param WP_Role $role | |
| 9748 | + * @return array Capability name => true, for granted non-level capabilities. | |
| 9749 | + */ | |
| 9750 | +/** | |
| 9751 | + * Of two role restrictions, the one that admits FEWER people. | |
| 9752 | + * | |
| 9753 | + * 'public' loses to everything; 'logged_in' loses to any real role. Between | |
| 9754 | + * two real roles the comparison is the same capability-superset test the | |
| 9755 | + * access check uses: if A's capabilities cover B's, then everyone who | |
| 9756 | + * satisfies A also satisfies B, so A is the stricter requirement. | |
| 9757 | + * | |
| 9758 | + * Two unrelated custom roles are genuinely incomparable — neither covers the | |
| 9759 | + * other. In that case $b wins, because callers pass the per-entry choice as | |
| 9760 | + * $b: an owner who set the dropdown on the page itself meant it. | |
| 9761 | + * | |
| 9762 | + * @param string $a Current winner (may be 'public'). | |
| 9763 | + * @param string $b Challenger. | |
| 9764 | + * @return string | |
| 9765 | + */ | |
| 9766 | +private static function mxchat_stricter_role($a, $b) { | |
| 9767 | + $rank = array('public' => 0, 'logged_in' => 1); | |
| 9768 | + | |
| 9769 | + if ($b === '' || $b === null) { return $a; } | |
| 9770 | + if ($a === '' || $a === null) { return $b; } | |
| 9771 | + if ($a === $b) { return $a; } | |
| 9772 | + | |
| 9773 | + $a_virtual = isset($rank[$a]); | |
| 9774 | + $b_virtual = isset($rank[$b]); | |
| 9775 | + | |
| 9776 | + if ($a_virtual && $b_virtual) { | |
| 9777 | + return $rank[$b] > $rank[$a] ? $b : $a; | |
| 9778 | + } | |
| 9779 | + // A real role always outranks 'public' and 'logged_in'. | |
| 9780 | + if ($a_virtual) { return $b; } | |
| 9781 | + if ($b_virtual) { return $a; } | |
| 9782 | + | |
| 9783 | + // Both are real roles: whichever covers the other is the stricter one. | |
| 9784 | + if (self::mxchat_role_covers($a, $b)) { return $a; } | |
| 9785 | + if (self::mxchat_role_covers($b, $a)) { return $b; } | |
| 9786 | + | |
| 9787 | + // Incomparable — the explicit per-entry choice wins. | |
| 9788 | + return $b; | |
| 9789 | +} | |
| 9790 | + | |
| 9791 | +private static function mxchat_real_caps($role) { | |
| 9792 | + $caps = array(); | |
| 9793 | + foreach ((array) $role->capabilities as $cap => $granted) { | |
| 9794 | + if (!$granted) { | |
| 9795 | + continue; | |
| 9796 | + } | |
| 9797 | + if (preg_match('/^level_\d+$/', $cap)) { | |
| 9798 | + continue; | |
| 9799 | + } | |
| 9800 | + $caps[$cap] = true; | |
| 9801 | + } | |
| 9802 | + return $caps; | |
| 9803 | +} | |
| 9804 | + | |
| 9805 | +/** | |
| 9806 | + * Handle role restriction updates via AJAX | |
| 9807 | + * Removed cache clearing call since we removed caching | |
| 9808 | + */ | |
| 9809 | +public function ajax_mxchat_update_role_restriction() { | |
| 9810 | + // Verify nonce and permissions | |
| 9811 | + if (!check_ajax_referer('mxchat_update_role_nonce', 'nonce', false)) { | |
| 9812 | + wp_send_json_error('Invalid nonce'); | |
| 9813 | + exit; | |
| 9814 | + } | |
| 9815 | + | |
| 9816 | + if (!current_user_can('manage_options')) { | |
| 9817 | + wp_send_json_error('Unauthorized access'); | |
| 9818 | + exit; | |
| 9819 | + } | |
| 9820 | + | |
| 9821 | + $entry_id = isset($_POST['entry_id']) ? sanitize_text_field($_POST['entry_id']) : ''; | |
| 9822 | + $role_restriction = isset($_POST['role_restriction']) ? sanitize_text_field($_POST['role_restriction']) : 'public'; | |
| 9823 | + $data_source = isset($_POST['data_source']) ? sanitize_text_field($_POST['data_source']) : 'wordpress'; | |
| 9824 | + | |
| 9825 | + if (empty($entry_id)) { | |
| 9826 | + wp_send_json_error('Invalid entry ID'); | |
| 9827 | + exit; | |
| 9828 | + } | |
| 9829 | + | |
| 9830 | + // Get knowledge manager instance to validate role restriction | |
| 9831 | + $knowledge_manager = MxChat_Knowledge_Manager::get_instance(); | |
| 9832 | + $valid_roles = array_keys($knowledge_manager->mxchat_get_role_options()); | |
| 9833 | + if (!in_array($role_restriction, $valid_roles)) { | |
| 9834 | + wp_send_json_error('Invalid role restriction'); | |
| 9835 | + exit; | |
| 9836 | + } | |
| 9837 | + | |
| 9838 | + global $wpdb; | |
| 9839 | + | |
| 9840 | + if ($data_source === 'pinecone') { | |
| 9841 | + // Handle Pinecone role restriction (stored separately in WordPress table) | |
| 9842 | + $roles_table = $wpdb->prefix . 'mxchat_pinecone_roles'; | |
| 9843 | + | |
| 9844 | + // Use REPLACE to insert or update the role restriction | |
| 9845 | + $result = $wpdb->replace( | |
| 9846 | + $roles_table, | |
| 9847 | + array( | |
| 9848 | + 'vector_id' => $entry_id, | |
| 9849 | + 'role_restriction' => $role_restriction, | |
| 9850 | + 'updated_at' => current_time('mysql') | |
| 9851 | + ), | |
| 9852 | + array('%s', '%s', '%s') | |
| 9853 | + ); | |
| 9854 | + | |
| 9855 | + // No cache clearing needed since we removed caching | |
| 9856 | + | |
| 9857 | + } else { | |
| 9858 | + // Handle WordPress database role restriction (existing functionality) | |
| 9859 | + $table_name = $wpdb->prefix . 'mxchat_system_prompt_content'; | |
| 9860 | + | |
| 9861 | + $result = $wpdb->update( | |
| 9862 | + $table_name, | |
| 9863 | + array('role_restriction' => $role_restriction), | |
| 9864 | + array('id' => absint($entry_id)), | |
| 9865 | + array('%s'), | |
| 9866 | + array('%d') | |
| 9867 | + ); | |
| 9868 | + } | |
| 9869 | + | |
| 9870 | + if ($result === false) { | |
| 9871 | + wp_send_json_error('Database update failed: ' . $wpdb->last_error); | |
| 9872 | + exit; | |
| 9873 | + } | |
| 9874 | + | |
| 9875 | + // Keep the OpenAI Vector Store mirror consistent with the new restriction | |
| 9876 | + // (plan 15b5c6): non-public pulls the mirrored file, public re-mirrors. | |
| 9877 | + if (class_exists('MxChat_Vectorstore_Manager')) { | |
| 9878 | + if ($data_source === 'pinecone') { | |
| 9879 | + if ($role_restriction !== 'public') { | |
| 9880 | + MxChat_Vectorstore_Manager::sync_delete_by_key($entry_id, 'default'); | |
| 9881 | + } | |
| 9882 | + // Public again: Pinecone-mode content isn't held locally, so the | |
| 9883 | + // entry re-mirrors on its next save/import rather than here. | |
| 9884 | + } else { | |
| 9885 | + $row_url = $wpdb->get_var($wpdb->prepare( | |
| 9886 | + "SELECT source_url FROM {$wpdb->prefix}mxchat_system_prompt_content WHERE id = %d", | |
| 9887 | + absint($entry_id) | |
| 9888 | + )); | |
| 9889 | + if (!empty($row_url)) { | |
| 9890 | + MxChat_Vectorstore_Manager::handle_role_change($row_url, 'default', $role_restriction); | |
| 9891 | + } | |
| 9892 | + } | |
| 9893 | + } | |
| 9894 | + | |
| 9895 | + wp_send_json_success(array( | |
| 9896 | + 'message' => 'Role restriction updated successfully', | |
| 9897 | + 'role_restriction' => $role_restriction, | |
| 9898 | + 'data_source' => $data_source, | |
| 9899 | + 'entry_id' => $entry_id | |
| 9900 | + )); | |
| 9901 | + exit; | |
| 9902 | +} | |
| 9903 | + | |
| 9904 | +// ======================================== | |
| 9905 | +// ROLE-BASED CONTENT RESTRICTIONS - NEW FUNCTIONS | |
| 9906 | +// Add these to your MxChat_Knowledge_Manager class | |
| 9907 | +// ======================================== | |
| 9908 | + | |
| 9909 | +/** | |
| 9910 | + * Initialize role-based content hooks | |
| 9911 | + * Add this call to your __construct() or mxchat_init_hooks() method | |
| 9912 | + */ | |
| 9913 | +private function mxchat_init_role_hooks() { | |
| 9914 | + // AJAX handlers for tag-role mappings | |
| 9915 | + add_action('wp_ajax_mxchat_add_tag_role_mapping', array($this, 'ajax_add_tag_role_mapping')); | |
| 9916 | + add_action('wp_ajax_mxchat_delete_tag_role_mapping', array($this, 'ajax_delete_tag_role_mapping')); | |
| 9917 | + add_action('wp_ajax_mxchat_get_tag_role_mappings', array($this, 'ajax_get_tag_role_mappings')); | |
| 9918 | + add_action('wp_ajax_mxchat_bulk_update_tag_roles', array($this, 'ajax_bulk_update_tag_roles')); | |
| 9919 | + | |
| 9920 | + // Hook to automatically update role restrictions when tags are added/removed | |
| 9921 | + add_action('set_object_terms', array($this, 'handle_tag_change'), 10, 6); | |
| 9922 | + | |
| 9923 | + // Hook to apply role restrictions on auto-sync | |
| 9924 | + add_action('mxchat_content_stored', array($this, 'apply_role_restriction_after_storage'), 10, 2); | |
| 9925 | +} | |
| 9926 | + | |
| 9927 | +/** | |
| 9928 | + * Add tag-role mapping via AJAX | |
| 9929 | + */ | |
| 9930 | +public function ajax_add_tag_role_mapping() { | |
| 9931 | + // Verify nonce and permissions | |
| 9932 | + check_ajax_referer('mxchat_prompts_setting_nonce', 'nonce'); | |
| 9933 | + | |
| 9934 | + if (!current_user_can('manage_options')) { | |
| 9935 | + wp_send_json_error('Unauthorized access'); | |
| 9936 | + exit; | |
| 9937 | + } | |
| 9938 | + | |
| 9939 | + $tag_input = isset($_POST['tag_slug']) ? sanitize_text_field($_POST['tag_slug']) : ''; | |
| 9940 | + $role_restriction = isset($_POST['role_restriction']) ? sanitize_text_field($_POST['role_restriction']) : 'public'; | |
| 9941 | + | |
| 9942 | + if (empty($tag_input)) { | |
| 9943 | + wp_send_json_error('Please enter a tag name or slug'); | |
| 9944 | + exit; | |
| 9945 | + } | |
| 9946 | + | |
| 9947 | + // Validate role restriction | |
| 9948 | + $valid_roles = array_keys($this->mxchat_get_role_options()); | |
| 9949 | + if (!in_array($role_restriction, $valid_roles)) { | |
| 9950 | + wp_send_json_error('Invalid role restriction'); | |
| 9951 | + exit; | |
| 9952 | + } | |
| 9953 | + | |
| 9954 | + // Resolve the tag by slug first, then fall back to its display name, so users can | |
| 9955 | + // enter either "premium-content" or "Premium Content". (plan b8bcf5 — the field is | |
| 9956 | + // labeled by name but previously validated by slug only, producing the confusing | |
| 9957 | + // "Tag does not exist in WordPress" error when a real tag's name was typed.) | |
| 9958 | + $term = get_term_by('slug', $tag_input, 'post_tag'); | |
| 9959 | + if (!$term) { | |
| 9960 | + $term = get_term_by('name', $tag_input, 'post_tag'); | |
| 9961 | + } | |
| 9962 | + if (!$term) { | |
| 9963 | + wp_send_json_error('No tag with that name or slug exists yet. Create it under Posts → Tags first, then enter its name or slug.'); | |
| 9964 | + exit; | |
| 9965 | + } | |
| 9966 | + | |
| 9967 | + // Always key the mapping by the RESOLVED slug — apply_role_restriction_to_post() | |
| 9968 | + // compares against each post's tag slugs, so the stored key must be a slug, | |
| 9969 | + // never the raw (possibly display-name) input. | |
| 9970 | + $tag_slug = $term->slug; | |
| 9971 | + | |
| 9972 | + // Get existing mappings | |
| 9973 | + $mappings = get_option('mxchat_tag_role_mappings', array()); | |
| 9974 | + | |
| 9975 | + // Check if mapping already exists | |
| 9976 | + if (isset($mappings[$tag_slug])) { | |
| 9977 | + wp_send_json_error('Mapping for this tag already exists'); | |
| 9978 | + exit; | |
| 9979 | + } | |
| 9980 | + | |
| 9981 | + // Add new mapping | |
| 9982 | + $mappings[$tag_slug] = $role_restriction; | |
| 9983 | + update_option('mxchat_tag_role_mappings', $mappings); | |
| 9984 | + | |
| 9985 | + wp_send_json_success(array( | |
| 9986 | + 'message' => 'Tag-role mapping added successfully', | |
| 9987 | + 'tag_slug' => $tag_slug, | |
| 9988 | + 'role_restriction' => $role_restriction | |
| 9989 | + )); | |
| 9990 | + exit; | |
| 9991 | +} | |
| 9992 | + | |
| 9993 | +/** | |
| 9994 | + * Delete tag-role mapping via AJAX | |
| 9995 | + */ | |
| 9996 | +public function ajax_delete_tag_role_mapping() { | |
| 9997 | + // Verify nonce and permissions | |
| 9998 | + check_ajax_referer('mxchat_prompts_setting_nonce', 'nonce'); | |
| 9999 | + | |
| 10000 | + if (!current_user_can('manage_options')) { | |
| 10001 | + wp_send_json_error('Unauthorized access'); | |
| 10002 | + exit; | |
| 10003 | + } | |
| 10004 | + | |
| 10005 | + $tag_slug = isset($_POST['tag_slug']) ? sanitize_text_field($_POST['tag_slug']) : ''; | |
| 10006 | + | |
| 10007 | + if (empty($tag_slug)) { | |
| 10008 | + wp_send_json_error('Tag slug is required'); | |
| 10009 | + exit; | |
| 10010 | + } | |
| 10011 | + | |
| 10012 | + // Get existing mappings | |
| 10013 | + $mappings = get_option('mxchat_tag_role_mappings', array()); | |
| 10014 | + | |
| 10015 | + // Check if mapping exists | |
| 10016 | + if (!isset($mappings[$tag_slug])) { | |
| 10017 | + wp_send_json_error('Mapping does not exist'); | |
| 10018 | + exit; | |
| 10019 | + } | |
| 10020 | + | |
| 10021 | + // Remove mapping | |
| 10022 | + unset($mappings[$tag_slug]); | |
| 10023 | + update_option('mxchat_tag_role_mappings', $mappings); | |
| 10024 | + | |
| 10025 | + wp_send_json_success(array( | |
| 10026 | + 'message' => 'Tag-role mapping deleted successfully', | |
| 10027 | + 'tag_slug' => $tag_slug | |
| 10028 | + )); | |
| 10029 | + exit; | |
| 10030 | +} | |
| 10031 | + | |
| 10032 | +/** | |
| 10033 | + * Get all tag-role mappings via AJAX | |
| 10034 | + */ | |
| 10035 | +public function ajax_get_tag_role_mappings() { | |
| 10036 | + // Verify nonce and permissions | |
| 10037 | + check_ajax_referer('mxchat_prompts_setting_nonce', 'nonce'); | |
| 10038 | + | |
| 10039 | + if (!current_user_can('manage_options')) { | |
| 10040 | + wp_send_json_error('Unauthorized access'); | |
| 10041 | + exit; | |
| 10042 | + } | |
| 10043 | + | |
| 10044 | + // Get mappings | |
| 10045 | + $mappings = get_option('mxchat_tag_role_mappings', array()); | |
| 10046 | + $role_options = $this->mxchat_get_role_options(); | |
| 10047 | + | |
| 10048 | + $formatted_mappings = array(); | |
| 10049 | + | |
| 10050 | + foreach ($mappings as $tag_slug => $role_restriction) { | |
| 10051 | + // Get tag object | |
| 10052 | + $term = get_term_by('slug', $tag_slug, 'post_tag'); | |
| 10053 | + | |
| 10054 | + // Count posts with this tag | |
| 10055 | + $post_count = 0; | |
| 10056 | + if ($term) { | |
| 10057 | + $post_count = $term->count; | |
| 10058 | + } | |
| 10059 | + | |
| 10060 | + $formatted_mappings[] = array( | |
| 10061 | + 'tag_slug' => $tag_slug, | |
| 10062 | + 'role_restriction' => $role_restriction, | |
| 10063 | + 'role_label' => $role_options[$role_restriction] ?? $role_restriction, | |
| 10064 | + 'post_count' => $post_count | |
| 10065 | + ); | |
| 10066 | + } | |
| 10067 | + | |
| 10068 | + wp_send_json_success(array( | |
| 10069 | + 'mappings' => $formatted_mappings | |
| 10070 | + )); | |
| 10071 | + exit; | |
| 10072 | +} | |
| 10073 | + | |
| 10074 | +/** | |
| 10075 | + * Bulk update role restrictions for all existing content with mapped tags | |
| 10076 | + */ | |
| 10077 | +public function ajax_bulk_update_tag_roles() { | |
| 10078 | + // Verify nonce and permissions | |
| 10079 | + check_ajax_referer('mxchat_prompts_setting_nonce', 'nonce'); | |
| 10080 | + | |
| 10081 | + if (!current_user_can('manage_options')) { | |
| 10082 | + wp_send_json_error('Unauthorized access'); | |
| 10083 | + exit; | |
| 10084 | + } | |
| 10085 | + | |
| 10086 | + // Get mappings | |
| 10087 | + $mappings = get_option('mxchat_tag_role_mappings', array()); | |
| 10088 | + | |
| 10089 | + if (empty($mappings)) { | |
| 10090 | + wp_send_json_error('No tag-role mappings found'); | |
| 10091 | + exit; | |
| 10092 | + } | |
| 10093 | + | |
| 10094 | + global $wpdb; | |
| 10095 | + | |
| 10096 | + // Check if using Pinecone | |
| 10097 | + $pinecone_options = get_option('mxchat_pinecone_addon_options', array()); | |
| 10098 | + $use_pinecone = ($pinecone_options['mxchat_use_pinecone'] ?? '0') === '1'; | |
| 10099 | + | |
| 10100 | + $updated_count = 0; | |
| 10101 | + $details = array(); | |
| 10102 | + | |
| 10103 | + foreach ($mappings as $tag_slug => $role_restriction) { | |
| 10104 | + // Get all posts with this tag | |
| 10105 | + $posts = get_posts(array( | |
| 10106 | + 'tag' => $tag_slug, | |
| 10107 | + 'post_type' => 'any', | |
| 10108 | + 'posts_per_page' => -1, | |
| 10109 | + 'fields' => 'ids', | |
| 10110 | + 'post_status' => 'publish' | |
| 10111 | + )); | |
| 10112 | + | |
| 10113 | + if (empty($posts)) { | |
| 10114 | + continue; | |
| 10115 | + } | |
| 10116 | + | |
| 10117 | + $tag_updated = 0; | |
| 10118 | + | |
| 10119 | + foreach ($posts as $post_id) { | |
| 10120 | + $source_url = get_permalink($post_id); | |
| 10121 | + if (!$source_url) { | |
| 10122 | + continue; | |
| 10123 | + } | |
| 10124 | + | |
| 10125 | + if ($use_pinecone && !empty($pinecone_options['mxchat_pinecone_api_key'])) { | |
| 10126 | + // Update Pinecone role restriction | |
| 10127 | + $roles_table = $wpdb->prefix . 'mxchat_pinecone_roles'; | |
| 10128 | + $vector_id = md5($source_url); | |
| 10129 | + | |
| 10130 | + $result = $wpdb->replace( | |
| 10131 | + $roles_table, | |
| 10132 | + array( | |
| 10133 | + 'vector_id' => $vector_id, | |
| 10134 | + 'role_restriction' => $role_restriction, | |
| 10135 | + 'updated_at' => current_time('mysql') | |
| 10136 | + ), | |
| 10137 | + array('%s', '%s', '%s') | |
| 10138 | + ); | |
| 10139 | + } else { | |
| 10140 | + // Update WordPress DB | |
| 10141 | + $table_name = $wpdb->prefix . 'mxchat_system_prompt_content'; | |
| 10142 | + | |
| 10143 | + $result = $wpdb->update( | |
| 10144 | + $table_name, | |
| 10145 | + array('role_restriction' => $role_restriction), | |
| 10146 | + array('source_url' => $source_url), | |
| 10147 | + array('%s'), | |
| 10148 | + array('%s') | |
| 10149 | + ); | |
| 10150 | + } | |
| 10151 | + | |
| 10152 | + if ($result !== false) { | |
| 10153 | + $tag_updated++; | |
| 10154 | + $updated_count++; | |
| 10155 | + } | |
| 10156 | + } | |
| 10157 | + | |
| 10158 | + if ($tag_updated > 0) { | |
| 10159 | + $details[] = sprintf( | |
| 10160 | + 'Tag "%s" (%s): %d posts updated', | |
| 10161 | + $tag_slug, | |
| 10162 | + $role_restriction, | |
| 10163 | + $tag_updated | |
| 10164 | + ); | |
| 10165 | + } | |
| 10166 | + } | |
| 10167 | + | |
| 10168 | + wp_send_json_success(array( | |
| 10169 | + 'message' => 'Bulk update completed', | |
| 10170 | + 'updated_count' => $updated_count, | |
| 10171 | + 'tags_processed' => count($mappings), | |
| 10172 | + 'details' => $details | |
| 10173 | + )); | |
| 10174 | + exit; | |
| 10175 | +} | |
| 10176 | + | |
| 10177 | +/** | |
| 10178 | + * Handle tag changes on posts (when tags are added or removed) | |
| 10179 | + */ | |
| 10180 | +public function handle_tag_change($object_id, $terms, $tt_ids, $taxonomy, $append, $old_tt_ids) { | |
| 10181 | + // Only process post tags | |
| 10182 | + if ($taxonomy !== 'post_tag') { | |
| 10183 | + return; | |
| 10184 | + } | |
| 10185 | + | |
| 10186 | + // Get tag-role mappings | |
| 10187 | + $mappings = get_option('mxchat_tag_role_mappings', array()); | |
| 10188 | + | |
| 10189 | + if (empty($mappings)) { | |
| 10190 | + return; | |
| 10191 | + } | |
| 10192 | + | |
| 10193 | + // Get the post's URL | |
| 10194 | + $source_url = get_permalink($object_id); | |
| 10195 | + if (!$source_url) { | |
| 10196 | + return; | |
| 10197 | + } | |
| 10198 | + | |
| 10199 | + // Determine the highest role restriction based on tags | |
| 10200 | + $highest_role = 'public'; | |
| 10201 | + $role_hierarchy = array( | |
| 10202 | + 'public' => 0, | |
| 10203 | + 'logged_in' => 1, | |
| 10204 | + 'subscriber' => 2, | |
| 10205 | + 'contributor' => 3, | |
| 10206 | + 'author' => 4, | |
| 10207 | + 'editor' => 5, | |
| 10208 | + 'administrator' => 6 | |
| 10209 | + ); | |
| 10210 | + | |
| 10211 | + // Get all current tags for the post | |
| 10212 | + $current_tags = wp_get_post_tags($object_id, array('fields' => 'slugs')); | |
| 10213 | + | |
| 10214 | + // Find the highest role restriction among the tags | |
| 10215 | + foreach ($current_tags as $tag_slug) { | |
| 10216 | + if (isset($mappings[$tag_slug])) { | |
| 10217 | + $role = $mappings[$tag_slug]; | |
| 10218 | + if (isset($role_hierarchy[$role]) && $role_hierarchy[$role] > $role_hierarchy[$highest_role]) { | |
| 10219 | + $highest_role = $role; | |
| 10220 | + } | |
| 10221 | + } | |
| 10222 | + } | |
| 10223 | + | |
| 10224 | + // Update the role restriction in the database | |
| 10225 | + global $wpdb; | |
| 10226 | + | |
| 10227 | + // Check if using Pinecone | |
| 10228 | + $pinecone_options = get_option('mxchat_pinecone_addon_options', array()); | |
| 10229 | + $use_pinecone = ($pinecone_options['mxchat_use_pinecone'] ?? '0') === '1'; | |
| 10230 | + | |
| 10231 | + if ($use_pinecone && !empty($pinecone_options['mxchat_pinecone_api_key'])) { | |
| 10232 | + // Update Pinecone role restriction | |
| 10233 | + $roles_table = $wpdb->prefix . 'mxchat_pinecone_roles'; | |
| 10234 | + $vector_id = md5($source_url); | |
| 10235 | + | |
| 10236 | + $wpdb->replace( | |
| 10237 | + $roles_table, | |
| 10238 | + array( | |
| 10239 | + 'vector_id' => $vector_id, | |
| 10240 | + 'role_restriction' => $highest_role, | |
| 10241 | + 'updated_at' => current_time('mysql') | |
| 10242 | + ), | |
| 10243 | + array('%s', '%s', '%s') | |
| 10244 | + ); | |
| 10245 | + } else { | |
| 10246 | + // Update WordPress DB | |
| 10247 | + $table_name = $wpdb->prefix . 'mxchat_system_prompt_content'; | |
| 10248 | + | |
| 10249 | + $wpdb->update( | |
| 10250 | + $table_name, | |
| 10251 | + array('role_restriction' => $highest_role), | |
| 10252 | + array('source_url' => $source_url), | |
| 10253 | + array('%s'), | |
| 10254 | + array('%s') | |
| 10255 | + ); | |
| 10256 | + } | |
| 10257 | + | |
| 10258 | + // The entry's restriction just changed — keep the OpenAI Vector Store | |
| 10259 | + // mirror consistent: non-public pulls the file (file_search has no | |
| 10260 | + // per-role filtering), public re-mirrors it (plan 15b5c6). | |
| 10261 | + if (class_exists('MxChat_Vectorstore_Manager')) { | |
| 10262 | + MxChat_Vectorstore_Manager::handle_role_change($source_url, 'default', $highest_role); | |
| 10263 | + } | |
| 10264 | +} | |
| 10265 | + | |
| 10266 | +/** | |
| 10267 | + * Apply role restriction after content is stored (for auto-sync) | |
| 10268 | + */ | |
| 10269 | +public function apply_role_restriction_after_storage($post_id, $source_url) { | |
| 10270 | + // Get tag-role mappings | |
| 10271 | + $mappings = get_option('mxchat_tag_role_mappings', array()); | |
| 10272 | + | |
| 10273 | + if (empty($mappings)) { | |
| 10274 | + return; | |
| 10275 | + } | |
| 10276 | + | |
| 10277 | + // Get all tags for the post | |
| 10278 | + $post_tags = wp_get_post_tags($post_id, array('fields' => 'slugs')); | |
| 10279 | + | |
| 10280 | + if (empty($post_tags)) { | |
| 10281 | + return; | |
| 10282 | + } | |
| 10283 | + | |
| 10284 | + // Determine the highest role restriction based on tags | |
| 10285 | + $highest_role = 'public'; | |
| 10286 | + $role_hierarchy = array( | |
| 10287 | + 'public' => 0, | |
| 10288 | + 'logged_in' => 1, | |
| 10289 | + 'subscriber' => 2, | |
| 10290 | + 'contributor' => 3, | |
| 10291 | + 'author' => 4, | |
| 10292 | + 'editor' => 5, | |
| 10293 | + 'administrator' => 6 | |
| 10294 | + ); | |
| 10295 | + | |
| 10296 | + foreach ($post_tags as $tag_slug) { | |
| 10297 | + if (isset($mappings[$tag_slug])) { | |
| 10298 | + $role = $mappings[$tag_slug]; | |
| 10299 | + if (isset($role_hierarchy[$role]) && $role_hierarchy[$role] > $role_hierarchy[$highest_role]) { | |
| 10300 | + $highest_role = $role; | |
| 10301 | + } | |
| 10302 | + } | |
| 10303 | + } | |
| 10304 | + | |
| 10305 | + // If no restricted tags found, return (leave as public) | |
| 10306 | + if ($highest_role === 'public') { | |
| 10307 | + return; | |
| 10308 | + } | |
| 10309 | + | |
| 10310 | + // Update the role restriction | |
| 10311 | + global $wpdb; | |
| 10312 | + | |
| 10313 | + // Check if using Pinecone | |
| 10314 | + $pinecone_options = get_option('mxchat_pinecone_addon_options', array()); | |
| 10315 | + $use_pinecone = ($pinecone_options['mxchat_use_pinecone'] ?? '0') === '1'; | |
| 10316 | + | |
| 10317 | + if ($use_pinecone && !empty($pinecone_options['mxchat_pinecone_api_key'])) { | |
| 10318 | + // Update Pinecone role restriction | |
| 10319 | + $roles_table = $wpdb->prefix . 'mxchat_pinecone_roles'; | |
| 10320 | + $vector_id = md5($source_url); | |
| 10321 | + | |
| 10322 | + $wpdb->replace( | |
| 10323 | + $roles_table, | |
| 10324 | + array( | |
| 10325 | + 'vector_id' => $vector_id, | |
| 10326 | + 'role_restriction' => $highest_role, | |
| 10327 | + 'updated_at' => current_time('mysql') | |
| 10328 | + ), | |
| 10329 | + array('%s', '%s', '%s') | |
| 10330 | + ); | |
| 10331 | + } else { | |
| 10332 | + // Update WordPress DB | |
| 10333 | + $table_name = $wpdb->prefix . 'mxchat_system_prompt_content'; | |
| 10334 | + | |
| 10335 | + $wpdb->update( | |
| 10336 | + $table_name, | |
| 10337 | + array('role_restriction' => $highest_role), | |
| 10338 | + array('source_url' => $source_url), | |
| 10339 | + array('%s'), | |
| 10340 | + array('%s') | |
| 10341 | + ); | |
| 10342 | + } | |
| 10343 | + | |
| 10344 | + // The entry's restriction just changed — keep the OpenAI Vector Store | |
| 10345 | + // mirror consistent: non-public pulls the file (file_search has no | |
| 10346 | + // per-role filtering), public re-mirrors it (plan 15b5c6). | |
| 10347 | + if (class_exists('MxChat_Vectorstore_Manager')) { | |
| 10348 | + MxChat_Vectorstore_Manager::handle_role_change($source_url, 'default', $highest_role); | |
| 10349 | + } | |
| 10350 | +} | |
| 10351 | + | |
| 10352 | + | |
| 10353 | + // ======================================== | |
| 10354 | + // HELPER METHODS | |
| 10355 | + // ======================================== | |
| 10356 | + | |
| 10357 | + /** | |
| 10358 | + * Check if user has required permissions for content processing | |
| 10359 | + */ | |
| 10360 | + private function mxchat_check_user_permissions() { | |
| 10361 | + if (!current_user_can('manage_options')) { | |
| 10362 | + wp_die(esc_html__('You do not have sufficient permissions.', 'mxchat')); | |
| 10363 | + } | |
| 10364 | + } | |
| 10365 | + | |
| 10366 | + /** | |
| 10367 | + * Validate nonce for security | |
| 10368 | + */ | |
| 10369 | + private function mxchat_validate_nonce($nonce_name, $nonce_action) { | |
| 10370 | + if (!isset($_POST[$nonce_name]) || !wp_verify_nonce($_POST[$nonce_name], $nonce_action)) { | |
| 10371 | + wp_die(esc_html__('Security check failed.', 'mxchat')); | |
| 10372 | + } | |
| 10373 | + } | |
| 10374 | + | |
| 10375 | + /** | |
| 10376 | + * Get embedding API credentials | |
| 10377 | + */ | |
| 10378 | + private function mxchat_get_embedding_credentials() { | |
| 10379 | + $embedding_model = $this->options['embedding_model'] ?? 'text-embedding-ada-002'; | |
| 10380 | + | |
| 10381 | + if (strpos($embedding_model, 'text-embedding-') !== false) { | |
| 10382 | + return array( | |
| 10383 | + 'type' => 'openai', | |
| 10384 | + 'api_key' => $this->options['api_key'] ?? '' | |
| 10385 | + ); | |
| 10386 | + } elseif (strpos($embedding_model, 'voyage-') !== false) { | |
| 10387 | + return array( | |
| 10388 | + 'type' => 'voyage', | |
| 10389 | + 'api_key' => $this->options['voyage_api_key'] ?? '' | |
| 10390 | + ); | |
| 10391 | + } elseif (strpos($embedding_model, 'gemini-embedding-') !== false) { | |
| 10392 | + return array( | |
| 10393 | + 'type' => 'gemini', | |
| 10394 | + 'api_key' => $this->options['gemini_api_key'] ?? '' | |
| 10395 | + ); | |
| 10396 | + } | |
| 10397 | + | |
| 10398 | + return array('type' => 'unknown', 'api_key' => ''); | |
| 10399 | + } | |
| 10400 | + | |
| 10401 | + /** | |
| 10402 | + * Log processing errors | |
| 10403 | + */ | |
| 10404 | + private function mxchat_log_processing_error($operation, $error_message) { | |
| 10405 | + //error_log("MxChat Knowledge Processing {$operation} Error: " . $error_message); | |
| 10406 | + } | |
| 10407 | + | |
| 10408 | + /** | |
| 10409 | + * Set admin notice transient | |
| 10410 | + */ | |
| 10411 | + private function mxchat_set_admin_notice($type, $message) { | |
| 10412 | + set_transient("mxchat_admin_notice_{$type}", $message, 30); | |
| 10413 | + } | |
| 10414 | + | |
| 10415 | + /** | |
| 10416 | + * Get Pinecone manager instance for vector operations | |
| 10417 | + */ | |
| 10418 | + private function mxchat_get_pinecone_manager() { | |
| 10419 | + return MxChat_Pinecone_Manager::get_instance(); | |
| 10420 | + } | |
| 10421 | + | |
| 10422 | + | |
| 10423 | + // ======================================== | |
| 10424 | +// DATABASE QUEUE TABLE MANAGEMENT | |
| 10425 | +// ======================================== | |
| 10426 | + | |
| 10427 | +/** | |
| 10428 | + * Create queue table on plugin activation | |
| 10429 | + * Call this from your plugin activation hook | |
| 10430 | + */ | |
| 10431 | +public function mxchat_create_queue_table() { | |
| 10432 | + global $wpdb; | |
| 10433 | + | |
| 10434 | + $table_name = $wpdb->prefix . 'mxchat_processing_queue'; | |
| 10435 | + $charset_collate = $wpdb->get_charset_collate(); | |
| 10436 | + | |
| 10437 | + $sql = "CREATE TABLE IF NOT EXISTS $table_name ( | |
| 10438 | + id bigint(20) unsigned NOT NULL AUTO_INCREMENT, | |
| 10439 | + queue_id varchar(64) NOT NULL, | |
| 10440 | + item_type varchar(20) NOT NULL, | |
| 10441 | + item_data longtext NOT NULL, | |
| 10442 | + status varchar(20) NOT NULL DEFAULT 'pending', | |
| 10443 | + bot_id varchar(50) NOT NULL DEFAULT 'default', | |
| 10444 | + priority int(11) NOT NULL DEFAULT 0, | |
| 10445 | + attempts int(11) NOT NULL DEFAULT 0, | |
| 10446 | + max_attempts int(11) NOT NULL DEFAULT 3, | |
| 10447 | + error_message text DEFAULT NULL, | |
| 10448 | + created_at datetime NOT NULL, | |
| 10449 | + started_at datetime DEFAULT NULL, | |
| 10450 | + completed_at datetime DEFAULT NULL, | |
| 10451 | + PRIMARY KEY (id), | |
| 10452 | + KEY queue_id (queue_id), | |
| 10453 | + KEY status (status), | |
| 10454 | + KEY item_type (item_type), | |
| 10455 | + KEY priority (priority) | |
| 10456 | + ) $charset_collate;"; | |
| 10457 | + | |
| 10458 | + require_once(ABSPATH . 'wp-admin/includes/upgrade.php'); | |
| 10459 | + dbDelta($sql); | |
| 10460 | + | |
| 10461 | + // Also create a meta table for queue metadata | |
| 10462 | + $meta_table = $wpdb->prefix . 'mxchat_queue_meta'; | |
| 10463 | + | |
| 10464 | + $meta_sql = "CREATE TABLE IF NOT EXISTS $meta_table ( | |
| 10465 | + id bigint(20) unsigned NOT NULL AUTO_INCREMENT, | |
| 10466 | + queue_id varchar(64) NOT NULL, | |
| 10467 | + meta_key varchar(255) NOT NULL, | |
| 10468 | + meta_value longtext, | |
| 10469 | + PRIMARY KEY (id), | |
| 10470 | + KEY queue_id (queue_id), | |
| 10471 | + KEY meta_key (meta_key) | |
| 10472 | + ) $charset_collate;"; | |
| 10473 | + | |
| 10474 | + dbDelta($meta_sql); | |
| 10475 | +} | |
| 10476 | + | |
| 10477 | +/** | |
| 10478 | + * Add items to the processing queue | |
| 10479 | + * | |
| 10480 | + * @param string $queue_id Unique identifier for this queue batch | |
| 10481 | + * @param string $item_type Type of item (url, pdf_page) | |
| 10482 | + * @param array $items Array of items to queue | |
| 10483 | + * @param string $bot_id Bot ID for processing | |
| 10484 | + * @return int Number of items queued | |
| 10485 | + */ | |
| 10486 | +private function mxchat_add_to_queue($queue_id, $item_type, $items, $bot_id = 'default') { | |
| 10487 | + global $wpdb; | |
| 10488 | + $table_name = $wpdb->prefix . 'mxchat_processing_queue'; | |
| 10489 | + | |
| 10490 | + $queued_count = 0; | |
| 10491 | + $priority = 0; | |
| 10492 | + | |
| 10493 | + foreach ($items as $item) { | |
| 10494 | + $result = $wpdb->insert( | |
| 10495 | + $table_name, | |
| 10496 | + array( | |
| 10497 | + 'queue_id' => $queue_id, | |
| 10498 | + 'item_type' => $item_type, | |
| 10499 | + 'item_data' => wp_json_encode($item), | |
| 10500 | + 'status' => 'pending', | |
| 10501 | + 'bot_id' => $bot_id, | |
| 10502 | + 'priority' => $priority, | |
| 10503 | + 'attempts' => 0, | |
| 10504 | + 'max_attempts' => 3, | |
| 10505 | + 'created_at' => current_time('mysql') | |
| 10506 | + ), | |
| 10507 | + array('%s', '%s', '%s', '%s', '%s', '%d', '%d', '%d', '%s') | |
| 10508 | + ); | |
| 10509 | + | |
| 10510 | + if ($result) { | |
| 10511 | + $queued_count++; | |
| 10512 | + } | |
| 10513 | + | |
| 10514 | + $priority++; // Process in order | |
| 10515 | + } | |
| 10516 | + | |
| 10517 | + return $queued_count; | |
| 10518 | +} | |
| 10519 | + | |
| 10520 | +/** | |
| 10521 | + * Store queue metadata (total counts, source URL, etc.) | |
| 10522 | + */ | |
| 10523 | +private function mxchat_set_queue_meta($queue_id, $meta_key, $meta_value) { | |
| 10524 | + global $wpdb; | |
| 10525 | + $meta_table = $wpdb->prefix . 'mxchat_queue_meta'; | |
| 10526 | + | |
| 10527 | + // Check if meta exists | |
| 10528 | + $existing = $wpdb->get_var($wpdb->prepare( | |
| 10529 | + "SELECT id FROM $meta_table WHERE queue_id = %s AND meta_key = %s", | |
| 10530 | + $queue_id, | |
| 10531 | + $meta_key | |
| 10532 | + )); | |
| 10533 | + | |
| 10534 | + if ($existing) { | |
| 10535 | + // Update | |
| 10536 | + $wpdb->update( | |
| 10537 | + $meta_table, | |
| 10538 | + array('meta_value' => maybe_serialize($meta_value)), | |
| 10539 | + array('queue_id' => $queue_id, 'meta_key' => $meta_key), | |
| 10540 | + array('%s'), | |
| 10541 | + array('%s', '%s') | |
| 10542 | + ); | |
| 10543 | + } else { | |
| 10544 | + // Insert | |
| 10545 | + $wpdb->insert( | |
| 10546 | + $meta_table, | |
| 10547 | + array( | |
| 10548 | + 'queue_id' => $queue_id, | |
| 10549 | + 'meta_key' => $meta_key, | |
| 10550 | + 'meta_value' => maybe_serialize($meta_value) | |
| 10551 | + ), | |
| 10552 | + array('%s', '%s', '%s') | |
| 10553 | + ); | |
| 10554 | + } | |
| 10555 | +} | |
| 10556 | + | |
| 10557 | +/** | |
| 10558 | + * Get queue metadata | |
| 10559 | + */ | |
| 10560 | +private function mxchat_get_queue_meta($queue_id, $meta_key) { | |
| 10561 | + global $wpdb; | |
| 10562 | + $meta_table = $wpdb->prefix . 'mxchat_queue_meta'; | |
| 10563 | + | |
| 10564 | + $value = $wpdb->get_var($wpdb->prepare( | |
| 10565 | + "SELECT meta_value FROM $meta_table WHERE queue_id = %s AND meta_key = %s", | |
| 10566 | + $queue_id, | |
| 10567 | + $meta_key | |
| 10568 | + )); | |
| 10569 | + | |
| 10570 | + return maybe_unserialize($value); | |
| 10571 | +} | |
| 10572 | + | |
| 10573 | +// ======================================== | |
| 10574 | +// AJAX QUEUE PROCESSING HANDLERS | |
| 10575 | +// ======================================== | |
| 10576 | + | |
| 10577 | +/** | |
| 10578 | + * AJAX: Get next item from queue to process | |
| 10579 | + */ | |
| 10580 | +public function ajax_mxchat_get_next_queue_item() { | |
| 10581 | + // Verify nonce and permissions | |
| 10582 | + check_ajax_referer('mxchat_queue_nonce', 'nonce'); | |
| 10583 | + | |
| 10584 | + if (!current_user_can('manage_options')) { | |
| 10585 | + wp_send_json_error('Unauthorized access'); | |
| 10586 | + } | |
| 10587 | + | |
| 10588 | + $queue_id = isset($_POST['queue_id']) ? sanitize_text_field($_POST['queue_id']) : ''; | |
| 10589 | + | |
| 10590 | + if (empty($queue_id)) { | |
| 10591 | + wp_send_json_error('Missing queue ID'); | |
| 10592 | + } | |
| 10593 | + | |
| 10594 | + global $wpdb; | |
| 10595 | + $table_name = $wpdb->prefix . 'mxchat_processing_queue'; | |
| 10596 | + | |
| 10597 | + // Get next pending item with retry logic for failed items | |
| 10598 | + $next_item = $wpdb->get_row($wpdb->prepare( | |
| 10599 | + "SELECT * FROM $table_name | |
| 10600 | + WHERE queue_id = %s | |
| 10601 | + AND status IN ('pending', 'failed') | |
| 10602 | + AND attempts < max_attempts | |
| 10603 | + ORDER BY priority ASC, id ASC | |
| 10604 | + LIMIT 1", | |
| 10605 | + $queue_id | |
| 10606 | + )); | |
| 10607 | + | |
| 10608 | + if (!$next_item) { | |
| 10609 | + // No more items - queue complete | |
| 10610 | + wp_send_json_success(array( | |
| 10611 | + 'complete' => true, | |
| 10612 | + 'message' => 'Queue processing complete' | |
| 10613 | + )); | |
| 10614 | + } | |
| 10615 | + | |
| 10616 | + // Mark item as processing | |
| 10617 | + $wpdb->update( | |
| 10618 | + $table_name, | |
| 10619 | + array( | |
| 10620 | + 'status' => 'processing', | |
| 10621 | + 'started_at' => current_time('mysql'), | |
| 10622 | + 'attempts' => $next_item->attempts + 1 | |
| 10623 | + ), | |
| 10624 | + array('id' => $next_item->id), | |
| 10625 | + array('%s', '%s', '%d'), | |
| 10626 | + array('%d') | |
| 10627 | + ); | |
| 10628 | + | |
| 10629 | + wp_send_json_success(array( | |
| 10630 | + 'complete' => false, | |
| 10631 | + 'item' => array( | |
| 10632 | + 'id' => $next_item->id, | |
| 10633 | + 'type' => $next_item->item_type, | |
| 10634 | + 'data' => json_decode($next_item->item_data, true), | |
| 10635 | + 'bot_id' => $next_item->bot_id, | |
| 10636 | + 'attempt' => $next_item->attempts + 1 | |
| 10637 | + ) | |
| 10638 | + )); | |
| 10639 | +} | |
| 10640 | + | |
| 10641 | +/** | |
| 10642 | + * AJAX: Process a single queue item | |
| 10643 | + */ | |
| 10644 | +public function ajax_mxchat_process_queue_item() { | |
| 10645 | + // Verify nonce and permissions | |
| 10646 | + check_ajax_referer('mxchat_queue_nonce', 'nonce'); | |
| 10647 | + | |
| 10648 | + if (!current_user_can('manage_options')) { | |
| 10649 | + wp_send_json_error('Unauthorized access'); | |
| 10650 | + } | |
| 10651 | + | |
| 10652 | + $item_id = isset($_POST['item_id']) ? absint($_POST['item_id']) : 0; | |
| 10653 | + $item_type = isset($_POST['item_type']) ? sanitize_text_field($_POST['item_type']) : ''; | |
| 10654 | + $item_data = isset($_POST['item_data']) ? $_POST['item_data'] : array(); | |
| 10655 | + $bot_id = isset($_POST['bot_id']) ? sanitize_key($_POST['bot_id']) : 'default'; | |
| 10656 | + | |
| 10657 | + if (empty($item_id) || empty($item_type)) { | |
| 10658 | + wp_send_json_error('Missing item data'); | |
| 10659 | + } | |
| 10660 | + | |
| 10661 | + global $wpdb; | |
| 10662 | + $table_name = $wpdb->prefix . 'mxchat_processing_queue'; | |
| 10663 | + | |
| 10664 | + // Process based on item type | |
| 10665 | + try { | |
| 10666 | + set_time_limit(60); // Give processing 60 seconds | |
| 10667 | + | |
| 10668 | + $result = false; | |
| 10669 | + $error_message = ''; | |
| 10670 | + | |
| 10671 | + // Read item directly from DB to get queue_id and preserve special chars in item_data | |
| 10672 | + // (POST round-trip through JS mangles characters like apostrophes in URLs) | |
| 10673 | + $db_item = $wpdb->get_row($wpdb->prepare( | |
| 10674 | + "SELECT queue_id, item_data FROM $table_name WHERE id = %d", | |
| 10675 | + $item_id | |
| 10676 | + )); | |
| 10677 | + $item_queue_id = $db_item ? $db_item->queue_id : ''; | |
| 10678 | + if ($db_item && !empty($db_item->item_data)) { | |
| 10679 | + $db_data = json_decode($db_item->item_data, true); | |
| 10680 | + if (is_array($db_data)) { | |
| 10681 | + $item_data = $db_data; | |
| 10682 | + } | |
| 10683 | + } | |
| 10684 | + | |
| 10685 | + switch ($item_type) { | |
| 10686 | + case 'url': | |
| 10687 | + $result = $this->mxchat_process_queue_url($item_data, $bot_id, $item_queue_id); | |
| 10688 | + break; | |
| 10689 | + | |
| 10690 | + case 'pdf_page': | |
| 10691 | + $result = $this->mxchat_process_queue_pdf_page($item_data, $bot_id); | |
| 10692 | + break; | |
| 10693 | + | |
| 10694 | + default: | |
| 10695 | + throw new Exception('Unknown item type: ' . $item_type); | |
| 10696 | + } | |
| 10697 | + | |
| 10698 | + if (is_wp_error($result)) { | |
| 10699 | + $error_code = $result->get_error_code(); | |
| 10700 | + // Content errors (empty page, sanitization) are permanent — retrying won't help | |
| 10701 | + $permanent_codes = array('empty_page', 'empty_after_sanitization', 'no_api_key', 'page_not_found'); | |
| 10702 | + if (in_array($error_code, $permanent_codes)) { | |
| 10703 | + // Mark as permanently failed — set attempts = max_attempts so it won't be retried | |
| 10704 | + $current_item = $wpdb->get_row($wpdb->prepare( | |
| 10705 | + "SELECT max_attempts FROM $table_name WHERE id = %d", $item_id | |
| 10706 | + )); | |
| 10707 | + $wpdb->update( | |
| 10708 | + $table_name, | |
| 10709 | + array( | |
| 10710 | + 'status' => 'failed', | |
| 10711 | + 'error_message' => $result->get_error_message(), | |
| 10712 | + 'attempts' => $current_item ? $current_item->max_attempts : 3 | |
| 10713 | + ), | |
| 10714 | + array('id' => $item_id), | |
| 10715 | + array('%s', '%s', '%d'), | |
| 10716 | + array('%d') | |
| 10717 | + ); | |
| 10718 | + wp_send_json_error(array( | |
| 10719 | + 'message' => $result->get_error_message(), | |
| 10720 | + 'permanent_failure' => true, | |
| 10721 | + 'item_id' => $item_id | |
| 10722 | + )); | |
| 10723 | + return; | |
| 10724 | + } | |
| 10725 | + throw new Exception($result->get_error_message()); | |
| 10726 | + } | |
| 10727 | + | |
| 10728 | + if ($result === false) { | |
| 10729 | + throw new Exception('Processing returned false - item may be empty or invalid'); | |
| 10730 | + } | |
| 10731 | + | |
| 10732 | + // Mark as completed | |
| 10733 | + $wpdb->update( | |
| 10734 | + $table_name, | |
| 10735 | + array( | |
| 10736 | + 'status' => 'completed', | |
| 10737 | + 'completed_at' => current_time('mysql'), | |
| 10738 | + 'error_message' => null | |
| 10739 | + ), | |
| 10740 | + array('id' => $item_id), | |
| 10741 | + array('%s', '%s', '%s'), | |
| 10742 | + array('%d') | |
| 10743 | + ); | |
| 10744 | + | |
| 10745 | + wp_send_json_success(array( | |
| 10746 | + 'processed' => true, | |
| 10747 | + 'item_id' => $item_id, | |
| 10748 | + 'message' => 'Item processed successfully' | |
| 10749 | + )); | |
| 10750 | + | |
| 10751 | + } catch (Exception $e) { | |
| 10752 | + $error_message = $e->getMessage(); | |
| 10753 | + | |
| 10754 | + // Get current attempt count | |
| 10755 | + $item = $wpdb->get_row($wpdb->prepare( | |
| 10756 | + "SELECT attempts, max_attempts FROM $table_name WHERE id = %d", | |
| 10757 | + $item_id | |
| 10758 | + )); | |
| 10759 | + | |
| 10760 | + // Check if we've exhausted retries | |
| 10761 | + if ($item && $item->attempts >= $item->max_attempts) { | |
| 10762 | + // Permanently failed | |
| 10763 | + $wpdb->update( | |
| 10764 | + $table_name, | |
| 10765 | + array( | |
| 10766 | + 'status' => 'failed', | |
| 10767 | + 'error_message' => $error_message | |
| 10768 | + ), | |
| 10769 | + array('id' => $item_id), | |
| 10770 | + array('%s', '%s'), | |
| 10771 | + array('%d') | |
| 10772 | + ); | |
| 10773 | + | |
| 10774 | + wp_send_json_error(array( | |
| 10775 | + 'message' => 'Item failed after maximum attempts: ' . $error_message, | |
| 10776 | + 'permanent_failure' => true, | |
| 10777 | + 'item_id' => $item_id | |
| 10778 | + )); | |
| 10779 | + } else { | |
| 10780 | + // Mark for retry | |
| 10781 | + $wpdb->update( | |
| 10782 | + $table_name, | |
| 10783 | + array( | |
| 10784 | + 'status' => 'failed', | |
| 10785 | + 'error_message' => $error_message | |
| 10786 | + ), | |
| 10787 | + array('id' => $item_id), | |
| 10788 | + array('%s', '%s'), | |
| 10789 | + array('%d') | |
| 10790 | + ); | |
| 10791 | + | |
| 10792 | + wp_send_json_error(array( | |
| 10793 | + 'message' => 'Item processing failed, will retry: ' . $error_message, | |
| 10794 | + 'can_retry' => true, | |
| 10795 | + 'item_id' => $item_id, | |
| 10796 | + 'attempts' => $item ? $item->attempts : 0 | |
| 10797 | + )); | |
| 10798 | + } | |
| 10799 | + } | |
| 10800 | +} | |
| 10801 | + | |
| 10802 | +/** | |
| 10803 | + * Process a URL from the queue | |
| 10804 | + */ | |
| 10805 | +private function mxchat_process_queue_url($item_data, $bot_id = 'default', $queue_id = '') { | |
| 10806 | + $url = isset($item_data['url']) ? $item_data['url'] : ''; | |
| 10807 | + | |
| 10808 | + if (empty($url)) { | |
| 10809 | + return new WP_Error('invalid_url', 'URL is empty'); | |
| 10810 | + } | |
| 10811 | + | |
| 10812 | + // Get bot-specific embedding decision early (needed for both paths) — | |
| 10813 | + // custom-provider-aware (plan cbd5fd). Error code preserved. | |
| 10814 | + $bot_options = $this->get_bot_options($bot_id); | |
| 10815 | + $options = !empty($bot_options) ? $bot_options : get_option('mxchat_options'); | |
| 10816 | + | |
| 10817 | + $preflight = MxChat_Utils::embedding_preflight($options); | |
| 10818 | + if (!$preflight['ok']) { | |
| 10819 | + return new WP_Error('no_api_key', $preflight['reason'] . ' (bot: ' . $bot_id . ')'); | |
| 10820 | + } | |
| 10821 | + $api_key = $preflight['api_key']; | |
| 10822 | + | |
| 10823 | + // Check if this is a WooCommerce product URL and WooCommerce is active | |
| 10824 | + $is_product_url = (strpos($url, '/product/') !== false || strpos($url, '/shop/') !== false); | |
| 10825 | + $content_type = $is_product_url ? 'product' : 'url'; | |
| 10826 | + | |
| 10827 | + // Try to get WooCommerce product data if it's a product URL | |
| 10828 | + if ($is_product_url && class_exists('WooCommerce')) { | |
| 10829 | + $product_content = $this->mxchat_extract_woocommerce_product_content($url); | |
| 10830 | + | |
| 10831 | + if (!empty($product_content)) { | |
| 10832 | + // Successfully extracted WooCommerce product data with pricing | |
| 10833 | + $result = MxChat_Utils::submit_content_to_db( | |
| 10834 | + $product_content, | |
| 10835 | + $url, | |
| 10836 | + $api_key, | |
| 10837 | + null, | |
| 10838 | + $bot_id, | |
| 10839 | + 'product' | |
| 10840 | + ); | |
| 10841 | + return $result; | |
| 10842 | + } | |
| 10843 | + // If WooCommerce extraction failed, fall through to HTML extraction | |
| 10844 | + } | |
| 10845 | + | |
| 10846 | + // Fetch URL content (fallback for non-products or when WooCommerce extraction fails) | |
| 10847 | + $is_likely_pdf = (strtolower(pathinfo(parse_url($url, PHP_URL_PATH) ?: '', PATHINFO_EXTENSION)) === 'pdf'); | |
| 10848 | + $response = wp_remote_get($url, array( | |
| 10849 | + 'timeout' => $is_likely_pdf ? 120 : 30, | |
| 10850 | + 'redirection' => 5, | |
| 10851 | + 'user-agent' => mxchat_ingest_user_agent(), | |
| 10852 | + )); | |
| 10853 | + | |
| 10854 | + if (is_wp_error($response)) { | |
| 10855 | + return $response; | |
| 10856 | + } | |
| 10857 | + | |
| 10858 | + $response_code = wp_remote_retrieve_response_code($response); | |
| 10859 | + if ($response_code !== 200) { | |
| 10860 | + return new WP_Error('http_error', 'HTTP ' . $response_code . ' error for: ' . $url); | |
| 10861 | + } | |
| 10862 | + | |
| 10863 | + // Check if URL is a PDF — expand into per-page queue items using the standard PDF pipeline | |
| 10864 | + if ($this->mxchat_is_pdf_url($url, $response)) { | |
| 10865 | + return $this->mxchat_expand_pdf_to_queue($url, $response, $bot_id, $queue_id); | |
| 10866 | + } | |
| 10867 | + | |
| 10868 | + $html = wp_remote_retrieve_body($response); | |
| 10869 | + | |
| 10870 | + if (empty($html)) { | |
| 10871 | + return new WP_Error('empty_response', 'Empty response body'); | |
| 10872 | + } | |
| 10873 | + | |
| 10874 | + // Extract and sanitize content | |
| 10875 | + $content = $this->mxchat_extract_main_content($html); | |
| 10876 | + $sanitized = $this->mxchat_sanitize_content_for_api($content); | |
| 10877 | + | |
| 10878 | + if (empty($sanitized)) { | |
| 10879 | + // Not an error - just no content found (maybe a redirect or empty page) | |
| 10880 | + return false; | |
| 10881 | + } | |
| 10882 | + | |
| 10883 | + // Submit to database with content_type (+ the page's own title, plan d9ee66) | |
| 10884 | + $result = MxChat_Utils::submit_content_to_db( | |
| 10885 | + $sanitized, | |
| 10886 | + $url, | |
| 10887 | + $api_key, | |
| 10888 | + null, | |
| 10889 | + $bot_id, | |
| 10890 | + $content_type, | |
| 10891 | + MxChat_Utils::extract_page_title($html) | |
| 10892 | + ); | |
| 10893 | + | |
| 10894 | + return $result; | |
| 10895 | +} | |
| 10896 | + | |
| 10897 | +/** | |
| 10898 | + * Expand a PDF URL into per-page queue items using the standard PDF pipeline. | |
| 10899 | + * Called when a sitemap URL turns out to be a PDF — downloads, parses page count, | |
| 10900 | + * and adds pdf_page items to the same queue so they process with full progress tracking. | |
| 10901 | + */ | |
| 10902 | +private function mxchat_expand_pdf_to_queue($pdf_url, $response, $bot_id = 'default', $queue_id = '') { | |
| 10903 | + set_time_limit(120); // PDFs need extra time for download + parsing | |
| 10904 | + | |
| 10905 | + $upload_dir = wp_upload_dir(); | |
| 10906 | + $pdf_filename = sanitize_file_name('mxchat_kb_' . md5($pdf_url) . '.pdf'); | |
| 10907 | + $pdf_path = trailingslashit($upload_dir['path']) . $pdf_filename; | |
| 10908 | + | |
| 10909 | + $response_body = wp_remote_retrieve_body($response); | |
| 10910 | + if (empty($response_body)) { | |
| 10911 | + return new WP_Error('empty_pdf', 'Empty PDF response for: ' . $pdf_url); | |
| 10912 | + } | |
| 10913 | + | |
| 10914 | + if (!wp_mkdir_p(dirname($pdf_path))) { | |
| 10915 | + return new WP_Error('dir_error', 'Failed to create upload directory'); | |
| 10916 | + } | |
| 10917 | + | |
| 10918 | + file_put_contents($pdf_path, $response_body); | |
| 10919 | + | |
| 10920 | + if (!file_exists($pdf_path)) { | |
| 10921 | + return new WP_Error('save_error', 'Failed to save PDF file'); | |
| 10922 | + } | |
| 10923 | + | |
| 10924 | + try { | |
| 10925 | + $total_pages = $this->mxchat_validate_and_count_pdf_pages($pdf_path); | |
| 10926 | + | |
| 10927 | + if ($total_pages === false || $total_pages < 1) { | |
| 10928 | + wp_delete_file($pdf_path); | |
| 10929 | + return new WP_Error('no_pages', 'PDF has no pages: ' . $pdf_url); | |
| 10930 | + } | |
| 10931 | + | |
| 10932 | + // Build per-page items identical to mxchat_handle_pdf_for_knowledge_base | |
| 10933 | + $pages = array(); | |
| 10934 | + for ($i = 1; $i <= $total_pages; $i++) { | |
| 10935 | + $pages[] = array( | |
| 10936 | + 'pdf_path' => $pdf_path, | |
| 10937 | + 'pdf_url' => $pdf_url, | |
| 10938 | + 'page_number' => $i, | |
| 10939 | + 'total_pages' => $total_pages | |
| 10940 | + ); | |
| 10941 | + } | |
| 10942 | + | |
| 10943 | + // Add pdf_page items to the SAME queue so the JS picks them up automatically | |
| 10944 | + if (!empty($queue_id)) { | |
| 10945 | + $queued_count = $this->mxchat_add_to_queue($queue_id, 'pdf_page', $pages, $bot_id); | |
| 10946 | + } else { | |
| 10947 | + // Fallback: create a new PDF queue (shouldn't happen in sitemap flow) | |
| 10948 | + $new_queue_id = 'pdf_' . md5($pdf_url . time()); | |
| 10949 | + $queued_count = $this->mxchat_add_to_queue($new_queue_id, 'pdf_page', $pages, $bot_id); | |
| 10950 | + $this->mxchat_set_queue_meta($new_queue_id, 'source_url', $pdf_url); | |
| 10951 | + $this->mxchat_set_queue_meta($new_queue_id, 'queue_type', 'pdf'); | |
| 10952 | + $this->mxchat_set_queue_meta($new_queue_id, 'total_items', $total_pages); | |
| 10953 | + $this->mxchat_set_queue_meta($new_queue_id, 'bot_id', $bot_id); | |
| 10954 | + $this->mxchat_set_queue_meta($new_queue_id, 'pdf_path', $pdf_path); | |
| 10955 | + $this->mxchat_set_queue_meta($new_queue_id, 'created_at', current_time('mysql')); | |
| 10956 | + } | |
| 10957 | + | |
| 10958 | + if ($queued_count === 0) { | |
| 10959 | + wp_delete_file($pdf_path); | |
| 10960 | + return new WP_Error('queue_error', 'Failed to add PDF pages to queue'); | |
| 10961 | + } | |
| 10962 | + | |
| 10963 | + // Return true so the original URL item is marked complete | |
| 10964 | + // The new pdf_page items will be processed in subsequent batches | |
| 10965 | + return true; | |
| 10966 | + | |
| 10967 | + } catch (Exception $e) { | |
| 10968 | + if (file_exists($pdf_path)) { | |
| 10969 | + wp_delete_file($pdf_path); | |
| 10970 | + } | |
| 10971 | + return new WP_Error('pdf_parse_error', 'Error parsing PDF: ' . $e->getMessage()); | |
| 10972 | + } | |
| 10973 | +} | |
| 10974 | + | |
| 10975 | +/** | |
| 10976 | + * Legacy: Process a PDF URL inline during sitemap queue processing. | |
| 10977 | + * @deprecated Use mxchat_expand_pdf_to_queue instead — kept for reference only. | |
| 10978 | + */ | |
| 10979 | +private function mxchat_process_pdf_url_inline($pdf_url, $response, $api_key, $bot_id = 'default') { | |
| 10980 | + set_time_limit(120); // PDFs need more time — downloading + parsing all pages | |
| 10981 | + | |
| 10982 | + $upload_dir = wp_upload_dir(); | |
| 10983 | + $pdf_filename = sanitize_file_name('mxchat_kb_' . md5($pdf_url) . '.pdf'); | |
| 10984 | + $pdf_path = trailingslashit($upload_dir['path']) . $pdf_filename; | |
| 10985 | + | |
| 10986 | + $response_body = wp_remote_retrieve_body($response); | |
| 10987 | + if (empty($response_body)) { | |
| 10988 | + return new WP_Error('empty_pdf', 'Empty PDF response'); | |
| 10989 | + } | |
| 10990 | + | |
| 10991 | + if (!wp_mkdir_p(dirname($pdf_path))) { | |
| 10992 | + return new WP_Error('dir_error', 'Failed to create upload directory'); | |
| 10993 | + } | |
| 10994 | + | |
| 10995 | + file_put_contents($pdf_path, $response_body); | |
| 10996 | + | |
| 10997 | + if (!file_exists($pdf_path)) { | |
| 10998 | + return new WP_Error('save_error', 'Failed to save PDF file'); | |
| 10999 | + } | |
| 11000 | + | |
| 11001 | + try { | |
| 11002 | + mxchat_load_pdf_parser(); | |
| 11003 | + $parser = new \Smalot\PdfParser\Parser(); | |
| 11004 | + $pdf = $parser->parseFile($pdf_path); | |
| 11005 | + $pages = $pdf->getPages(); | |
| 11006 | + $total_pages = count($pages); | |
| 11007 | + | |
| 11008 | + if ($total_pages < 1) { | |
| 11009 | + wp_delete_file($pdf_path); | |
| 11010 | + return new WP_Error('no_pages', 'PDF has no pages'); | |
| 11011 | + } | |
| 11012 | + | |
| 11013 | + $processed = 0; | |
| 11014 | + $skipped_pages = array(); | |
| 11015 | + | |
| 11016 | + for ($i = 0; $i < $total_pages; $i++) { | |
| 11017 | + $page_num = $i + 1; | |
| 11018 | + $text = $pages[$i]->getText(); | |
| 11019 | + $text = MxChat_Utils::normalize_pdf_rtl($text, 'kb_pdf_import page ' . $page_num); | |
| 11020 | + if (empty($text)) { | |
| 11021 | + $skipped_pages[] = 'Page ' . $page_num . ': No text could be extracted — page may contain only images, links, or non-standard encoding'; | |
| 11022 | + continue; | |
| 11023 | + } | |
| 11024 | + | |
| 11025 | + $sanitized = $this->mxchat_sanitize_content_for_api($text); | |
| 11026 | + if (empty($sanitized)) { | |
| 11027 | + $skipped_pages[] = 'Page ' . $page_num . ': Text was extracted but contained only special characters, control codes, or unsupported content'; | |
| 11028 | + continue; | |
| 11029 | + } | |
| 11030 | + | |
| 11031 | + $metadata = array( | |
| 11032 | + 'document_type' => 'pdf', | |
| 11033 | + 'total_pages' => $total_pages, | |
| 11034 | + 'current_page' => $page_num, | |
| 11035 | + 'source_url' => $pdf_url, | |
| 11036 | + ); | |
| 11037 | + | |
| 11038 | + $content_with_metadata = wp_json_encode($metadata) . "\n---\n" . $sanitized; | |
| 11039 | + $page_url = MxChat_Utils::pdf_page_identity($pdf_url, $page_num); | |
| 11040 | + | |
| 11041 | + MxChat_Utils::submit_content_to_db( | |
| 11042 | + $content_with_metadata, | |
| 11043 | + $page_url, | |
| 11044 | + $api_key, | |
| 11045 | + null, | |
| 11046 | + $bot_id, | |
| 11047 | + 'pdf' | |
| 11048 | + ); | |
| 11049 | + | |
| 11050 | + $processed++; | |
| 11051 | + } | |
| 11052 | + | |
| 11053 | + // Clean up the temp PDF file | |
| 11054 | + wp_delete_file($pdf_path); | |
| 11055 | + | |
| 11056 | + if (!empty($skipped_pages)) { | |
| 11057 | + error_log('MxChat PDF: Skipped ' . count($skipped_pages) . ' of ' . $total_pages . ' pages: ' . implode('; ', $skipped_pages)); | |
| 11058 | + } | |
| 11059 | + | |
| 11060 | + return $processed > 0 ? true : false; | |
| 11061 | + | |
| 11062 | + } catch (Exception $e) { | |
| 11063 | + if (file_exists($pdf_path)) { | |
| 11064 | + wp_delete_file($pdf_path); | |
| 11065 | + } | |
| 11066 | + return new WP_Error('pdf_parse_error', 'Error parsing PDF: ' . $e->getMessage()); | |
| 11067 | + } | |
| 11068 | +} | |
| 11069 | + | |
| 11070 | +/** | |
| 11071 | + * Extract WooCommerce product content including pricing | |
| 11072 | + * | |
| 11073 | + * @param string $url The product URL | |
| 11074 | + * @return string|false Product content with pricing, or false if not found | |
| 11075 | + */ | |
| 11076 | +private function mxchat_extract_woocommerce_product_content($url) { | |
| 11077 | + // Try to get product ID from URL | |
| 11078 | + $product_id = url_to_postid($url); | |
| 11079 | + | |
| 11080 | + // If url_to_postid fails, try to extract from URL pattern | |
| 11081 | + if (!$product_id) { | |
| 11082 | + $product_slug = ''; | |
| 11083 | + | |
| 11084 | + // Handle pretty permalinks: /product/product-name/ | |
| 11085 | + if (preg_match('/\/product\/([^\/\?]+)/', $url, $matches)) { | |
| 11086 | + $product_slug = $matches[1]; | |
| 11087 | + } | |
| 11088 | + | |
| 11089 | + if (!empty($product_slug)) { | |
| 11090 | + $product_post = get_page_by_path($product_slug, OBJECT, 'product'); | |
| 11091 | + if ($product_post) { | |
| 11092 | + $product_id = $product_post->ID; | |
| 11093 | + } | |
| 11094 | + } | |
| 11095 | + } | |
| 11096 | + | |
| 11097 | + if (!$product_id) { | |
| 11098 | + return false; | |
| 11099 | + } | |
| 11100 | + | |
| 11101 | + // Get WooCommerce product object | |
| 11102 | + $product = wc_get_product($product_id); | |
| 11103 | + | |
| 11104 | + if (!$product) { | |
| 11105 | + return false; | |
| 11106 | + } | |
| 11107 | + | |
| 11108 | + // Build product content via the shared WC-object assembler (a3d60c) — same | |
| 11109 | + // body as the auto-sync product writer, so the two paths can never drift. | |
| 11110 | + $content = $this->mxchat_prepare_product_content_for_indexing($product); | |
| 11111 | + | |
| 11112 | + return $this->mxchat_sanitize_content_for_api($content); | |
| 11113 | +} | |
| 11114 | + | |
| 11115 | +/** | |
| 11116 | + * Process a PDF page from the queue | |
| 11117 | + */ | |
| 11118 | +private function mxchat_process_queue_pdf_page($item_data, $bot_id = 'default') { | |
| 11119 | + $pdf_path = isset($item_data['pdf_path']) ? $item_data['pdf_path'] : ''; | |
| 11120 | + $pdf_url = isset($item_data['pdf_url']) ? $item_data['pdf_url'] : ''; | |
| 11121 | + $page_number = isset($item_data['page_number']) ? absint($item_data['page_number']) : 0; | |
| 11122 | + $total_pages = isset($item_data['total_pages']) ? absint($item_data['total_pages']) : 0; | |
| 11123 | + | |
| 11124 | + if (empty($pdf_path) || !file_exists($pdf_path)) { | |
| 11125 | + return new WP_Error('pdf_not_found', 'PDF file not found: ' . $pdf_path); | |
| 11126 | + } | |
| 11127 | + | |
| 11128 | + if ($page_number < 1) { | |
| 11129 | + return new WP_Error('invalid_page', 'Invalid page number'); | |
| 11130 | + } | |
| 11131 | + | |
| 11132 | + try { | |
| 11133 | + mxchat_load_pdf_parser(); | |
| 11134 | + $parser = new \Smalot\PdfParser\Parser(); | |
| 11135 | + $pdf = $parser->parseFile($pdf_path); | |
| 11136 | + $pages = $pdf->getPages(); | |
| 11137 | + | |
| 11138 | + if (!isset($pages[$page_number - 1])) { | |
| 11139 | + return new WP_Error('page_not_found', 'Page ' . $page_number . ' not found in PDF'); | |
| 11140 | + } | |
| 11141 | + | |
| 11142 | + $text = $pages[$page_number - 1]->getText(); | |
| 11143 | + $text = MxChat_Utils::normalize_pdf_rtl($text, 'kb_pdf_page page ' . $page_number); | |
| 11144 | + | |
| 11145 | + if (empty($text)) { | |
| 11146 | + return new WP_Error('empty_page', 'Page ' . $page_number . ': No text could be extracted — page may contain only images, links, or non-standard encoding'); | |
| 11147 | + } | |
| 11148 | + | |
| 11149 | + $sanitized = $this->mxchat_sanitize_content_for_api($text); | |
| 11150 | + | |
| 11151 | + if (empty($sanitized)) { | |
| 11152 | + return new WP_Error('empty_after_sanitization', 'Page ' . $page_number . ': Text was extracted but contained only special characters, control codes, or unsupported content that was removed during cleanup'); | |
| 11153 | + } | |
| 11154 | + | |
| 11155 | + // Create metadata | |
| 11156 | + $metadata = array( | |
| 11157 | + 'document_type' => 'pdf', | |
| 11158 | + 'total_pages' => $total_pages, | |
| 11159 | + 'current_page' => $page_number, | |
| 11160 | + 'source_url' => $pdf_url | |
| 11161 | + ); | |
| 11162 | + | |
| 11163 | + $content_with_metadata = wp_json_encode($metadata) . "\n---\n" . $sanitized; | |
| 11164 | + $page_url = MxChat_Utils::pdf_page_identity($pdf_url, $page_number); | |
| 11165 | + | |
| 11166 | + // Get bot-specific embedding decision — custom-provider-aware | |
| 11167 | + // (plan cbd5fd). Error code preserved. | |
| 11168 | + $bot_options = $this->get_bot_options($bot_id); | |
| 11169 | + $options = !empty($bot_options) ? $bot_options : get_option('mxchat_options'); | |
| 11170 | + | |
| 11171 | + $preflight = MxChat_Utils::embedding_preflight($options); | |
| 11172 | + if (!$preflight['ok']) { | |
| 11173 | + return new WP_Error('no_api_key', $preflight['reason'] . ' (bot: ' . $bot_id . ')'); | |
| 11174 | + } | |
| 11175 | + $api_key = $preflight['api_key']; | |
| 11176 | + | |
| 11177 | + // Submit to database - UPDATED 2.5.6: Added content_type 'pdf' | |
| 11178 | + $result = MxChat_Utils::submit_content_to_db( | |
| 11179 | + $content_with_metadata, | |
| 11180 | + $page_url, | |
| 11181 | + $api_key, | |
| 11182 | + null, | |
| 11183 | + $bot_id, | |
| 11184 | + 'pdf' | |
| 11185 | + ); | |
| 11186 | + | |
| 11187 | + return $result; | |
| 11188 | + | |
| 11189 | + } catch (Exception $e) { | |
| 11190 | + return new WP_Error('pdf_parse_error', 'Error parsing PDF: ' . $e->getMessage()); | |
| 11191 | + } | |
| 11192 | +} | |
| 11193 | + | |
| 11194 | +/** | |
| 11195 | + * AJAX: Get queue processing status | |
| 11196 | + */ | |
| 11197 | +public function ajax_mxchat_get_queue_status() { | |
| 11198 | + // Verify nonce and permissions | |
| 11199 | + check_ajax_referer('mxchat_queue_nonce', 'nonce'); | |
| 11200 | + | |
| 11201 | + if (!current_user_can('manage_options')) { | |
| 11202 | + wp_send_json_error('Unauthorized access'); | |
| 11203 | + } | |
| 11204 | + | |
| 11205 | + $queue_id = isset($_POST['queue_id']) ? sanitize_text_field($_POST['queue_id']) : ''; | |
| 11206 | + | |
| 11207 | + if (empty($queue_id)) { | |
| 11208 | + wp_send_json_error('Missing queue ID'); | |
| 11209 | + } | |
| 11210 | + | |
| 11211 | + global $wpdb; | |
| 11212 | + $table_name = $wpdb->prefix . 'mxchat_processing_queue'; | |
| 11213 | + | |
| 11214 | + // Get counts by status | |
| 11215 | + $counts = $wpdb->get_results($wpdb->prepare( | |
| 11216 | + "SELECT status, COUNT(*) as count | |
| 11217 | + FROM $table_name | |
| 11218 | + WHERE queue_id = %s | |
| 11219 | + GROUP BY status", | |
| 11220 | + $queue_id | |
| 11221 | + ), OBJECT_K); | |
| 11222 | + | |
| 11223 | + $total = 0; | |
| 11224 | + $completed = 0; | |
| 11225 | + $failed = 0; | |
| 11226 | + $processing = 0; | |
| 11227 | + $pending = 0; | |
| 11228 | + | |
| 11229 | + foreach ($counts as $status => $data) { | |
| 11230 | + $count = absint($data->count); | |
| 11231 | + $total += $count; | |
| 11232 | + | |
| 11233 | + switch ($status) { | |
| 11234 | + case 'completed': | |
| 11235 | + $completed = $count; | |
| 11236 | + break; | |
| 11237 | + case 'failed': | |
| 11238 | + $failed = $count; | |
| 11239 | + break; | |
| 11240 | + case 'processing': | |
| 11241 | + $processing = $count; | |
| 11242 | + break; | |
| 11243 | + case 'pending': | |
| 11244 | + $pending = $count; | |
| 11245 | + break; | |
| 11246 | + } | |
| 11247 | + } | |
| 11248 | + | |
| 11249 | + // Calculate percentage | |
| 11250 | + $percentage = $total > 0 ? round((($completed + $failed) / $total) * 100) : 0; | |
| 11251 | + | |
| 11252 | + // Get failed items details (include all failed items, not just those that exhausted retries) | |
| 11253 | + $failed_items = array(); | |
| 11254 | + if ($failed > 0) { | |
| 11255 | + $failed_items = $wpdb->get_results($wpdb->prepare( | |
| 11256 | + "SELECT item_type, item_data, error_message, attempts | |
| 11257 | + FROM $table_name | |
| 11258 | + WHERE queue_id = %s | |
| 11259 | + AND status = 'failed' | |
| 11260 | + ORDER BY id DESC | |
| 11261 | + LIMIT 50", | |
| 11262 | + $queue_id | |
| 11263 | + )); | |
| 11264 | + } | |
| 11265 | + | |
| 11266 | + // Get queue metadata | |
| 11267 | + $source_url = $this->mxchat_get_queue_meta($queue_id, 'source_url'); | |
| 11268 | + $queue_type = $this->mxchat_get_queue_meta($queue_id, 'queue_type'); | |
| 11269 | + | |
| 11270 | + // Determine if queue is complete | |
| 11271 | + $is_complete = ($pending === 0 && $processing === 0); | |
| 11272 | + | |
| 11273 | + wp_send_json_success(array( | |
| 11274 | + 'queue_id' => $queue_id, | |
| 11275 | + 'queue_type' => $queue_type, | |
| 11276 | + 'source_url' => $source_url, | |
| 11277 | + 'total' => $total, | |
| 11278 | + 'completed' => $completed, | |
| 11279 | + 'failed' => $failed, | |
| 11280 | + 'processing' => $processing, | |
| 11281 | + 'pending' => $pending, | |
| 11282 | + 'percentage' => $percentage, | |
| 11283 | + 'is_complete' => $is_complete, | |
| 11284 | + 'failed_items' => $failed_items, | |
| 11285 | + 'status' => $is_complete ? 'complete' : 'processing' | |
| 11286 | + )); | |
| 11287 | +} | |
| 11288 | + | |
| 11289 | +/** | |
| 11290 | + * AJAX: Clear completed queue | |
| 11291 | + */ | |
| 11292 | +public function ajax_mxchat_clear_queue() { | |
| 11293 | + // Verify nonce and permissions | |
| 11294 | + check_ajax_referer('mxchat_queue_nonce', 'nonce'); | |
| 11295 | + | |
| 11296 | + if (!current_user_can('manage_options')) { | |
| 11297 | + wp_send_json_error('Unauthorized access'); | |
| 11298 | + } | |
| 11299 | + | |
| 11300 | + $queue_id = isset($_POST['queue_id']) ? sanitize_text_field($_POST['queue_id']) : ''; | |
| 11301 | + | |
| 11302 | + if (empty($queue_id)) { | |
| 11303 | + wp_send_json_error('Missing queue ID'); | |
| 11304 | + } | |
| 11305 | + | |
| 11306 | + global $wpdb; | |
| 11307 | + $table_name = $wpdb->prefix . 'mxchat_processing_queue'; | |
| 11308 | + $meta_table = $wpdb->prefix . 'mxchat_queue_meta'; | |
| 11309 | + | |
| 11310 | + // Delete queue items | |
| 11311 | + $wpdb->delete( | |
| 11312 | + $table_name, | |
| 11313 | + array('queue_id' => $queue_id), | |
| 11314 | + array('%s') | |
| 11315 | + ); | |
| 11316 | + | |
| 11317 | + // Delete queue metadata | |
| 11318 | + $wpdb->delete( | |
| 11319 | + $meta_table, | |
| 11320 | + array('queue_id' => $queue_id), | |
| 11321 | + array('%s') | |
| 11322 | + ); | |
| 11323 | + | |
| 11324 | + wp_send_json_success(array( | |
| 11325 | + 'message' => 'Queue cleared successfully' | |
| 11326 | + )); | |
| 11327 | +} | |
| 11328 | + | |
| 11329 | +/** | |
| 11330 | + * AJAX: Retry failed items in queue | |
| 11331 | + */ | |
| 11332 | +public function ajax_mxchat_retry_failed() { | |
| 11333 | + // Verify nonce and permissions | |
| 11334 | + check_ajax_referer('mxchat_queue_nonce', 'nonce'); | |
| 11335 | + | |
| 11336 | + if (!current_user_can('manage_options')) { | |
| 11337 | + wp_send_json_error('Unauthorized access'); | |
| 11338 | + } | |
| 11339 | + | |
| 11340 | + $queue_id = isset($_POST['queue_id']) ? sanitize_text_field($_POST['queue_id']) : ''; | |
| 11341 | + | |
| 11342 | + if (empty($queue_id)) { | |
| 11343 | + wp_send_json_error('Missing queue ID'); | |
| 11344 | + } | |
| 11345 | + | |
| 11346 | + global $wpdb; | |
| 11347 | + $table_name = $wpdb->prefix . 'mxchat_processing_queue'; | |
| 11348 | + | |
| 11349 | + // Reset failed items to pending and reset attempt count | |
| 11350 | + $updated = $wpdb->update( | |
| 11351 | + $table_name, | |
| 11352 | + array( | |
| 11353 | + 'status' => 'pending', | |
| 11354 | + 'attempts' => 0, | |
| 11355 | + 'error_message' => null | |
| 11356 | + ), | |
| 11357 | + array( | |
| 11358 | + 'queue_id' => $queue_id, | |
| 11359 | + 'status' => 'failed' | |
| 11360 | + ), | |
| 11361 | + array('%s', '%d', '%s'), | |
| 11362 | + array('%s', '%s') | |
| 11363 | + ); | |
| 11364 | + | |
| 11365 | + wp_send_json_success(array( | |
| 11366 | + 'message' => 'Reset ' . $updated . ' failed items for retry', | |
| 11367 | + 'reset_count' => $updated | |
| 11368 | + )); | |
| 11369 | +} | |
| 11370 | + | |
| 11371 | + | |
| 11372 | +public function ajax_mxchat_mark_queue_complete() { | |
| 11373 | + check_ajax_referer('mxchat_queue_nonce', 'nonce'); | |
| 11374 | + | |
| 11375 | + if (!current_user_can('manage_options')) { | |
| 11376 | + wp_send_json_error('Unauthorized access'); | |
| 11377 | + } | |
| 11378 | + | |
| 11379 | + $queue_id = isset($_POST['queue_id']) ? sanitize_text_field($_POST['queue_id']) : ''; | |
| 11380 | + | |
| 11381 | + if (empty($queue_id)) { | |
| 11382 | + wp_send_json_error('Missing queue ID'); | |
| 11383 | + } | |
| 11384 | + | |
| 11385 | + // Clear active queue transients | |
| 11386 | + if (strpos($queue_id, 'sitemap_') === 0) { | |
| 11387 | + delete_transient('mxchat_active_queue_sitemap'); | |
| 11388 | + } else if (strpos($queue_id, 'pdf_') === 0) { | |
| 11389 | + delete_transient('mxchat_active_queue_pdf'); | |
| 11390 | + } | |
| 11391 | + | |
| 11392 | + wp_send_json_success(array('message' => 'Queue marked as complete')); | |
| 11393 | +} | |
| 11394 | + | |
| 11395 | + | |
| 11396 | + // ======================================== | |
| 11397 | + // STATIC ACCESS METHODS | |
| 11398 | + // ======================================== | |
| 11399 | + | |
| 11400 | + /** | |
| 11401 | + * Get singleton instance | |
| 11402 | + */ | |
| 11403 | + public static function get_instance() { | |
| 11404 | + static $instance = null; | |
| 11405 | + if ($instance === null) { | |
| 11406 | + $instance = new self(); | |
| 11407 | + } | |
| 11408 | + return $instance; | |
| 11409 | + } | |
| 11410 | +} | |
| 11411 | + | |
| 11412 | +// Initialize the Knowledge manager | |
| 7894 | 11413 | $mxchat_knowledge_manager = MxChat_Knowledge_Manager::get_instance(); |
