| 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 |
$this->init_hooks(); |
| 22 |
} |
| 23 |
|
| 24 |
/** |
| 25 |
* Initialize WordPress hooks for content processing |
| 26 |
*/ |
| 27 |
private function init_hooks() { |
| 28 |
// Admin post handlers for form submissions |
| 29 |
add_action('admin_post_mxchat_submit_content', array($this, 'mxchat_handle_content_submission')); |
| 30 |
add_action('admin_post_mxchat_submit_sitemap', array($this, 'mxchat_handle_sitemap_submission')); |
| 31 |
add_action('admin_post_mxchat_stop_processing', array($this, 'mxchat_stop_processing')); |
| 32 |
|
| 33 |
// AJAX handlers for real-time processing and status updates |
| 34 |
add_action('wp_ajax_mxchat_get_status_updates', array($this, 'ajax_get_status_updates')); |
| 35 |
add_action('wp_ajax_mxchat_dismiss_completed_status', array($this, 'ajax_dismiss_completed_status')); // NEW |
| 36 |
add_action('wp_ajax_mxchat_get_content_list', array($this, 'ajax_mxchat_get_content_list')); |
| 37 |
add_action('wp_ajax_mxchat_process_selected_content', array($this, 'ajax_mxchat_process_selected_content')); |
| 38 |
add_action('wp_ajax_mxchat_manual_batch_process', array($this, 'ajax_manual_batch_process')); |
| 39 |
|
| 40 |
|
| 41 |
// Cron handlers for background processing |
| 42 |
add_action('mxchat_process_sitemap_urls', array($this, 'process_sitemap_urls_cron'), 10, 5); |
| 43 |
add_action('mxchat_process_pdf_pages', array($this, 'process_pdf_pages_cron'), 10, 5); |
| 44 |
|
| 45 |
// WordPress post management hooks |
| 46 |
add_action('save_post', array($this, 'handle_post_update'), 10, 2); |
| 47 |
add_action('post_updated', array($this, 'handle_post_update'), 10, 3); |
| 48 |
add_action('before_delete_post', array($this, 'mxchat_handle_post_delete')); |
| 49 |
add_action('wp_trash_post', array($this, 'mxchat_handle_post_delete')); |
| 50 |
|
| 51 |
// WooCommerce product hooks (if WooCommerce is active) |
| 52 |
if (class_exists('WooCommerce')) { |
| 53 |
add_action('save_post_product', array($this, 'mxchat_handle_product_change')); |
| 54 |
add_action('wp_trash_post', array($this, 'mxchat_handle_product_delete')); |
| 55 |
add_action('before_delete_post', array($this, 'mxchat_handle_product_delete')); |
| 56 |
} |
| 57 |
|
| 58 |
// Admin notices for processing status |
| 59 |
add_action('admin_notices', array($this, 'display_processing_status')); |
| 60 |
} |
| 61 |
|
| 62 |
/** |
| 63 |
* Get current options (refreshed) |
| 64 |
*/ |
| 65 |
private function get_options() { |
| 66 |
if (empty($this->options)) { |
| 67 |
$this->options = get_option('mxchat_options', array()); |
| 68 |
} |
| 69 |
return $this->options; |
| 70 |
} |
| 71 |
|
| 72 |
|
| 73 |
/** |
| 74 |
* Handle manual batch processing via AJAX |
| 75 |
*/ |
| 76 |
public function ajax_manual_batch_process() { |
| 77 |
try { |
| 78 |
// Verify nonce and permissions |
| 79 |
check_ajax_referer('mxchat_status_nonce', 'nonce'); |
| 80 |
|
| 81 |
if (!current_user_can('manage_options')) { |
| 82 |
wp_send_json_error('Unauthorized access'); |
| 83 |
} |
| 84 |
|
| 85 |
$process_type = sanitize_text_field($_POST['process_type'] ?? ''); |
| 86 |
$url = sanitize_text_field($_POST['url'] ?? ''); |
| 87 |
|
| 88 |
if (empty($process_type) || empty($url)) { |
| 89 |
wp_send_json_error('Missing required parameters'); |
| 90 |
} |
| 91 |
|
| 92 |
$processed = 0; |
| 93 |
|
| 94 |
if ($process_type === 'pdf') { |
| 95 |
$processed = $this->manual_process_pdf_batch($url); |
| 96 |
} elseif ($process_type === 'sitemap') { |
| 97 |
$processed = $this->manual_process_sitemap_batch($url); |
| 98 |
} |
| 99 |
|
| 100 |
if ($processed > 0) { |
| 101 |
wp_send_json_success(array( |
| 102 |
'message' => "Processed {$processed} items successfully", |
| 103 |
'processed' => $processed |
| 104 |
)); |
| 105 |
} else { |
| 106 |
wp_send_json_error('No items were processed'); |
| 107 |
} |
| 108 |
|
| 109 |
} catch (Exception $e) { |
| 110 |
//error_log('Manual batch process error: ' . $e->getMessage()); |
| 111 |
wp_send_json_error('Processing failed: ' . $e->getMessage()); |
| 112 |
} |
| 113 |
} |
| 114 |
|
| 115 |
/** |
| 116 |
* Process a small PDF batch manually - DIRECT PROCESSING |
| 117 |
*/ |
| 118 |
private function manual_process_pdf_batch($pdf_url) { |
| 119 |
try { |
| 120 |
$status_key = sanitize_key('mxchat_pdf_status_' . md5($pdf_url)); |
| 121 |
$status = get_transient($status_key); |
| 122 |
|
| 123 |
if (!$status || $status['status'] !== 'processing') { |
| 124 |
//error_log('Manual PDF: No processing status found'); |
| 125 |
return 0; |
| 126 |
} |
| 127 |
|
| 128 |
//error_log('Manual PDF: Starting direct processing for ' . $pdf_url); |
| 129 |
|
| 130 |
// Get current progress |
| 131 |
$current_page = $status['processed_pages'] ?? 0; |
| 132 |
$total_pages = $status['total_pages'] ?? 0; |
| 133 |
|
| 134 |
if ($current_page >= $total_pages) { |
| 135 |
//error_log('Manual PDF: Already completed'); |
| 136 |
return 0; |
| 137 |
} |
| 138 |
|
| 139 |
// Try to download the PDF again for processing |
| 140 |
$response = wp_remote_get($pdf_url, array('timeout' => 30)); |
| 141 |
|
| 142 |
if (is_wp_error($response)) { |
| 143 |
//error_log('Manual PDF: Failed to download PDF: ' . $response->get_error_message()); |
| 144 |
return 0; |
| 145 |
} |
| 146 |
|
| 147 |
$pdf_content = wp_remote_retrieve_body($response); |
| 148 |
if (empty($pdf_content)) { |
| 149 |
//error_log('Manual PDF: Empty PDF content'); |
| 150 |
return 0; |
| 151 |
} |
| 152 |
|
| 153 |
// Save PDF temporarily |
| 154 |
$upload_dir = wp_upload_dir(); |
| 155 |
$temp_pdf_path = trailingslashit($upload_dir['path']) . 'temp_manual_' . time() . '.pdf'; |
| 156 |
file_put_contents($temp_pdf_path, $pdf_content); |
| 157 |
|
| 158 |
// Process 2 pages directly |
| 159 |
$processed = $this->process_pdf_pages_direct($temp_pdf_path, $pdf_url, $current_page, 5); |
| 160 |
|
| 161 |
// Clean up temp file |
| 162 |
if (file_exists($temp_pdf_path)) { |
| 163 |
wp_delete_file($temp_pdf_path); |
| 164 |
} |
| 165 |
|
| 166 |
//error_log('Manual PDF: Processed ' . $processed . ' pages'); |
| 167 |
return $processed; |
| 168 |
|
| 169 |
} catch (Exception $e) { |
| 170 |
//error_log('Manual PDF batch error: ' . $e->getMessage()); |
| 171 |
return 0; |
| 172 |
} |
| 173 |
} |
| 174 |
|
| 175 |
/** |
| 176 |
* Process PDF pages directly without cron |
| 177 |
*/ |
| 178 |
private function process_pdf_pages_direct($pdf_path, $pdf_url, $start_page, $batch_size) { |
| 179 |
try { |
| 180 |
if (!file_exists($pdf_path)) { |
| 181 |
//error_log('Direct PDF: File not found at ' . $pdf_path); |
| 182 |
return 0; |
| 183 |
} |
| 184 |
|
| 185 |
$parser = new \Smalot\PdfParser\Parser(); |
| 186 |
$pdf = $parser->parseFile($pdf_path); |
| 187 |
$pages = $pdf->getPages(); |
| 188 |
|
| 189 |
$status_key = sanitize_key('mxchat_pdf_status_' . md5($pdf_url)); |
| 190 |
$status = get_transient($status_key); |
| 191 |
|
| 192 |
if (!$status) { |
| 193 |
return 0; |
| 194 |
} |
| 195 |
|
| 196 |
$options = get_option('mxchat_options'); |
| 197 |
$api_key = $options['api_key'] ?? ''; |
| 198 |
|
| 199 |
if (empty($api_key)) { |
| 200 |
//error_log('Direct PDF: No API key'); |
| 201 |
return 0; |
| 202 |
} |
| 203 |
|
| 204 |
$processed = 0; |
| 205 |
$end_page = min($start_page + $batch_size, count($pages)); |
| 206 |
|
| 207 |
for ($i = $start_page; $i < $end_page; $i++) { |
| 208 |
try { |
| 209 |
$page_number = $i + 1; |
| 210 |
$text = $pages[$i]->getText(); |
| 211 |
|
| 212 |
if (empty($text)) { |
| 213 |
//error_log('Direct PDF: Empty text on page ' . $page_number); |
| 214 |
continue; |
| 215 |
} |
| 216 |
|
| 217 |
$sanitized_content = $this->mxchat_sanitize_content_for_api($text); |
| 218 |
if (empty($sanitized_content)) { |
| 219 |
//error_log('Direct PDF: No content after sanitization on page ' . $page_number); |
| 220 |
continue; |
| 221 |
} |
| 222 |
|
| 223 |
// Generate embedding |
| 224 |
$embedding_vector = $this->mxchat_generate_embedding($sanitized_content); |
| 225 |
if (is_string($embedding_vector)) { |
| 226 |
//error_log('Direct PDF: Embedding failed on page ' . $page_number . ': ' . $embedding_vector); |
| 227 |
continue; |
| 228 |
} |
| 229 |
|
| 230 |
// Create metadata |
| 231 |
$metadata = array( |
| 232 |
'document_type' => 'pdf', |
| 233 |
'total_pages' => count($pages), |
| 234 |
'current_page' => $page_number, |
| 235 |
'source_url' => $pdf_url |
| 236 |
); |
| 237 |
|
| 238 |
$content_with_metadata = wp_json_encode($metadata) . "\n---\n" . $sanitized_content; |
| 239 |
$page_url = esc_url($pdf_url . "#page=" . $page_number); |
| 240 |
|
| 241 |
// Store in database |
| 242 |
$db_result = MxChat_Utils::submit_content_to_db($content_with_metadata, $page_url, $api_key); |
| 243 |
|
| 244 |
if (is_wp_error($db_result)) { |
| 245 |
//error_log('Direct PDF: DB error on page ' . $page_number . ': ' . $db_result->get_error_message()); |
| 246 |
continue; |
| 247 |
} |
| 248 |
|
| 249 |
$processed++; |
| 250 |
//error_log('Direct PDF: Successfully processed page ' . $page_number); |
| 251 |
|
| 252 |
// Update status |
| 253 |
$status['processed_pages'] = $i + 1; |
| 254 |
$status['last_update'] = time(); |
| 255 |
$status['percentage'] = round(($status['processed_pages'] / $status['total_pages']) * 100); |
| 256 |
set_transient($status_key, $status, DAY_IN_SECONDS); |
| 257 |
|
| 258 |
} catch (Exception $e) { |
| 259 |
//error_log('Direct PDF: Error processing page ' . ($i + 1) . ': ' . $e->getMessage()); |
| 260 |
continue; |
| 261 |
} |
| 262 |
} |
| 263 |
|
| 264 |
// Check if completed |
| 265 |
if ($status['processed_pages'] >= $status['total_pages']) { |
| 266 |
$status['status'] = 'complete'; |
| 267 |
set_transient($status_key, $status, DAY_IN_SECONDS); |
| 268 |
//error_log('Direct PDF: Processing completed'); |
| 269 |
} |
| 270 |
|
| 271 |
return $processed; |
| 272 |
|
| 273 |
} catch (Exception $e) { |
| 274 |
//error_log('Direct PDF processing error: ' . $e->getMessage()); |
| 275 |
return 0; |
| 276 |
} |
| 277 |
} |
| 278 |
|
| 279 |
/** |
| 280 |
* Process a small sitemap batch manually - DIRECT PROCESSING |
| 281 |
*/ |
| 282 |
private function manual_process_sitemap_batch($sitemap_url) { |
| 283 |
try { |
| 284 |
$status_key = sanitize_key('mxchat_sitemap_status_' . md5($sitemap_url)); |
| 285 |
$status = get_transient($status_key); |
| 286 |
|
| 287 |
if (!$status || $status['status'] !== 'processing') { |
| 288 |
return 0; |
| 289 |
} |
| 290 |
|
| 291 |
//error_log('Manual Sitemap: Starting direct processing for ' . $sitemap_url); |
| 292 |
|
| 293 |
// Re-fetch the sitemap to get URLs |
| 294 |
$response = wp_remote_get($sitemap_url, array('timeout' => 30)); |
| 295 |
if (is_wp_error($response)) { |
| 296 |
//error_log('Manual Sitemap: Failed to fetch sitemap'); |
| 297 |
return 0; |
| 298 |
} |
| 299 |
|
| 300 |
$sitemap_content = wp_remote_retrieve_body($response); |
| 301 |
$xml = simplexml_load_string($sitemap_content); |
| 302 |
|
| 303 |
if (!$xml) { |
| 304 |
//error_log('Manual Sitemap: Invalid XML'); |
| 305 |
return 0; |
| 306 |
} |
| 307 |
|
| 308 |
$urls = array(); |
| 309 |
foreach ($xml->url as $url_element) { |
| 310 |
$urls[] = (string)$url_element->loc; |
| 311 |
} |
| 312 |
|
| 313 |
$current_processed = $status['processed_urls'] ?? 0; |
| 314 |
$batch_size = 5; |
| 315 |
$processed = 0; |
| 316 |
|
| 317 |
// Process next 2 URLs |
| 318 |
for ($i = $current_processed; $i < min($current_processed + $batch_size, count($urls)); $i++) { |
| 319 |
$url = $urls[$i]; |
| 320 |
|
| 321 |
if ($this->process_single_url_direct($url)) { |
| 322 |
$processed++; |
| 323 |
} |
| 324 |
|
| 325 |
// Update status |
| 326 |
$status['processed_urls'] = $i + 1; |
| 327 |
$status['last_update'] = time(); |
| 328 |
$status['percentage'] = round(($status['processed_urls'] / $status['total_urls']) * 100); |
| 329 |
set_transient($status_key, $status, DAY_IN_SECONDS); |
| 330 |
} |
| 331 |
|
| 332 |
// Check if completed |
| 333 |
if ($status['processed_urls'] >= $status['total_urls']) { |
| 334 |
$status['status'] = 'complete'; |
| 335 |
set_transient($status_key, $status, DAY_IN_SECONDS); |
| 336 |
} |
| 337 |
|
| 338 |
//error_log('Manual Sitemap: Processed ' . $processed . ' URLs'); |
| 339 |
return $processed; |
| 340 |
|
| 341 |
} catch (Exception $e) { |
| 342 |
//error_log('Manual sitemap batch error: ' . $e->getMessage()); |
| 343 |
return 0; |
| 344 |
} |
| 345 |
} |
| 346 |
|
| 347 |
/** |
| 348 |
* Process a single URL directly |
| 349 |
*/ |
| 350 |
private function process_single_url_direct($url) { |
| 351 |
try { |
| 352 |
$response = wp_remote_get($url, array('timeout' => 30)); |
| 353 |
if (is_wp_error($response)) { |
| 354 |
return false; |
| 355 |
} |
| 356 |
|
| 357 |
$html = wp_remote_retrieve_body($response); |
| 358 |
$content = $this->mxchat_extract_main_content($html); |
| 359 |
$sanitized = $this->mxchat_sanitize_content_for_api($content); |
| 360 |
|
| 361 |
if (empty($sanitized)) { |
| 362 |
return false; |
| 363 |
} |
| 364 |
|
| 365 |
$options = get_option('mxchat_options'); |
| 366 |
$api_key = $options['api_key'] ?? ''; |
| 367 |
|
| 368 |
$result = MxChat_Utils::submit_content_to_db($sanitized, $url, $api_key); |
| 369 |
|
| 370 |
return !is_wp_error($result); |
| 371 |
|
| 372 |
} catch (Exception $e) { |
| 373 |
//error_log('Single URL processing error: ' . $e->getMessage()); |
| 374 |
return false; |
| 375 |
} |
| 376 |
} |
| 377 |
|
| 378 |
|
| 379 |
// ======================================== |
| 380 |
// MAIN CONTENT SUBMISSION HANDLERS |
| 381 |
// ======================================== |
| 382 |
|
| 383 |
public function mxchat_handle_content_submission() { |
| 384 |
// Check if the form was submitted and the user has permission. |
| 385 |
if (!isset($_POST['submit_content']) || !current_user_can('manage_options')) { |
| 386 |
return; |
| 387 |
} |
| 388 |
|
| 389 |
// Verify the nonce. |
| 390 |
$nonce = isset($_POST['mxchat_submit_content_nonce']) ? sanitize_text_field(wp_unslash($_POST['mxchat_submit_content_nonce'])) : ''; |
| 391 |
if (!wp_verify_nonce($nonce, 'mxchat_submit_content_action')) { |
| 392 |
wp_die(esc_html__('Nonce verification failed.', 'mxchat')); |
| 393 |
} |
| 394 |
|
| 395 |
// Sanitize the inputs. |
| 396 |
$article_content = sanitize_textarea_field($_POST['article_content']); |
| 397 |
$article_url = isset($_POST['article_url']) ? esc_url_raw($_POST['article_url']) : ''; |
| 398 |
|
| 399 |
// Get API key for submission |
| 400 |
$options = get_option('mxchat_options'); |
| 401 |
$selected_model = $options['embedding_model'] ?? 'text-embedding-ada-002'; |
| 402 |
|
| 403 |
if (strpos($selected_model, 'voyage') === 0) { |
| 404 |
$api_key = $options['voyage_api_key'] ?? ''; |
| 405 |
} elseif (strpos($selected_model, 'gemini-embedding') === 0) { |
| 406 |
$api_key = $options['gemini_api_key'] ?? ''; |
| 407 |
} else { |
| 408 |
$api_key = $options['api_key'] ?? ''; |
| 409 |
} |
| 410 |
|
| 411 |
if (empty($api_key)) { |
| 412 |
set_transient('mxchat_admin_notice_error', |
| 413 |
esc_html__('API key is not configured. Please add your API key in the settings before submitting content.', 'mxchat'), |
| 414 |
30 |
| 415 |
); |
| 416 |
wp_safe_redirect(esc_url(admin_url('admin.php?page=mxchat-prompts'))); |
| 417 |
exit; |
| 418 |
} |
| 419 |
|
| 420 |
// Use centralized utility function for storage |
| 421 |
$result = MxChat_Utils::submit_content_to_db($article_content, $article_url, $api_key); |
| 422 |
|
| 423 |
if (is_wp_error($result)) { |
| 424 |
set_transient('mxchat_admin_notice_error', |
| 425 |
esc_html__('Error storing content: ', 'mxchat') . $result->get_error_message(), |
| 426 |
30 |
| 427 |
); |
| 428 |
} else { |
| 429 |
set_transient('mxchat_admin_notice_success', |
| 430 |
esc_html__('Content successfully submitted!', 'mxchat'), |
| 431 |
30 |
| 432 |
); |
| 433 |
} |
| 434 |
|
| 435 |
wp_safe_redirect(esc_url(admin_url('admin.php?page=mxchat-prompts'))); |
| 436 |
exit; |
| 437 |
} |
| 438 |
public function is_pdf_url($url, $response) { |
| 439 |
$content_type = wp_remote_retrieve_header($response, 'content-type'); |
| 440 |
$file_extension = strtolower(pathinfo($url, PATHINFO_EXTENSION)); |
| 441 |
|
| 442 |
return strpos($content_type, 'pdf') !== false || $file_extension === 'pdf'; |
| 443 |
} |
| 444 |
public function handle_pdf_for_knowledge_base($pdf_url, $response) { |
| 445 |
if (!current_user_can('manage_options')) { |
| 446 |
//error_log(esc_html__('Unauthorized PDF processing attempt', 'mxchat')); |
| 447 |
return false; |
| 448 |
} |
| 449 |
|
| 450 |
$pdf_url = esc_url_raw($pdf_url); |
| 451 |
$upload_dir = wp_upload_dir(); |
| 452 |
|
| 453 |
if (isset($upload_dir['error']) && $upload_dir['error'] !== false) { |
| 454 |
//error_log(sprintf(esc_html__('Upload directory error: %s', 'mxchat'), esc_html($upload_dir['error']))); |
| 455 |
return false; |
| 456 |
} |
| 457 |
|
| 458 |
$pdf_filename = sanitize_file_name('mxchat_kb_' . time() . '.pdf'); |
| 459 |
$pdf_path = trailingslashit($upload_dir['path']) . $pdf_filename; |
| 460 |
|
| 461 |
$response_body = wp_remote_retrieve_body($response); |
| 462 |
if (empty($response_body)) { |
| 463 |
//error_log(esc_html__('Empty PDF response body', 'mxchat')); |
| 464 |
return false; |
| 465 |
} |
| 466 |
|
| 467 |
if (!wp_mkdir_p(dirname($pdf_path))) { |
| 468 |
//error_log(sprintf(esc_html__('Failed to create directory for PDF: %s', 'mxchat'), esc_html($pdf_path))); |
| 469 |
return false; |
| 470 |
} |
| 471 |
|
| 472 |
try { |
| 473 |
file_put_contents($pdf_path, $response_body); |
| 474 |
|
| 475 |
if (!file_exists($pdf_path)) { |
| 476 |
throw new Exception(__('Failed to save PDF file', 'mxchat')); |
| 477 |
} |
| 478 |
|
| 479 |
$parser = new \Smalot\PdfParser\Parser(); |
| 480 |
$pdf = $parser->parseFile($pdf_path); |
| 481 |
$total_pages = absint(count($pdf->getPages())); |
| 482 |
|
| 483 |
if ($total_pages < 1) { |
| 484 |
throw new Exception(__('Invalid PDF: no pages found', 'mxchat')); |
| 485 |
} |
| 486 |
|
| 487 |
wp_schedule_single_event(time(), 'mxchat_process_pdf_pages', array( |
| 488 |
'pdf_path' => $pdf_path, |
| 489 |
'pdf_url' => $pdf_url, |
| 490 |
'total_pages' => $total_pages, |
| 491 |
'batch_size' => absint(15), |
| 492 |
'batch_pause' => absint(10) |
| 493 |
)); |
| 494 |
|
| 495 |
$status_data = array( |
| 496 |
'total_pages' => $total_pages, |
| 497 |
'processed_pages' => 0, |
| 498 |
'status' => 'processing', |
| 499 |
'last_update' => time() |
| 500 |
); |
| 501 |
|
| 502 |
set_transient( |
| 503 |
sanitize_key('mxchat_pdf_status_' . md5($pdf_url)), |
| 504 |
array_map('sanitize_text_field', $status_data), |
| 505 |
DAY_IN_SECONDS |
| 506 |
); |
| 507 |
|
| 508 |
return __('scheduled', 'mxchat'); |
| 509 |
|
| 510 |
} catch (Exception $e) { |
| 511 |
//error_log(sprintf(esc_html__('Error preparing PDF for processing: %s', 'mxchat'), esc_html($e->getMessage()))); |
| 512 |
if (file_exists($pdf_path)) { |
| 513 |
wp_delete_file($pdf_path); |
| 514 |
} |
| 515 |
return false; |
| 516 |
} |
| 517 |
} |
| 518 |
|
| 519 |
public function process_pdf_pages_cron($pdf_path, $pdf_url, $total_pages, $batch_size, $batch_pause) { |
| 520 |
// Validate inputs |
| 521 |
$pdf_path = sanitize_text_field($pdf_path); |
| 522 |
$pdf_url = esc_url_raw($pdf_url); |
| 523 |
$total_pages = absint($total_pages); |
| 524 |
$batch_size = absint($batch_size); |
| 525 |
$batch_pause = absint($batch_pause); |
| 526 |
|
| 527 |
try { |
| 528 |
if (!file_exists($pdf_path)) { |
| 529 |
throw new Exception(sprintf('PDF file not found at path: %s', esc_html($pdf_path))); |
| 530 |
} |
| 531 |
|
| 532 |
$parser = new \Smalot\PdfParser\Parser(); |
| 533 |
$pdf = $parser->parseFile($pdf_path); |
| 534 |
$pages = $pdf->getPages(); |
| 535 |
|
| 536 |
// Get current progress |
| 537 |
$status_key = sanitize_key('mxchat_pdf_status_' . md5($pdf_url)); |
| 538 |
$status = get_transient($status_key); |
| 539 |
|
| 540 |
if (!$status || !is_array($status)) { |
| 541 |
throw new Exception('Invalid status data retrieved from transient'); |
| 542 |
} |
| 543 |
|
| 544 |
// Initialize failed pages list if it doesn't exist |
| 545 |
if (!isset($status['failed_pages_list']) || !is_array($status['failed_pages_list'])) { |
| 546 |
$status['failed_pages_list'] = []; |
| 547 |
} |
| 548 |
|
| 549 |
$start_page = absint($status['processed_pages']); |
| 550 |
$end_page = min($start_page + $batch_size, $total_pages); |
| 551 |
$options = get_option('mxchat_options'); |
| 552 |
|
| 553 |
if (empty($options['api_key'])) { |
| 554 |
throw new Exception('API key is missing or invalid'); |
| 555 |
} |
| 556 |
|
| 557 |
$successful_pages = 0; |
| 558 |
$failed_pages = 0; |
| 559 |
|
| 560 |
for ($i = $start_page; $i < $end_page; $i++) { |
| 561 |
$page_number = $i + 1; |
| 562 |
$max_retries = 3; |
| 563 |
$retry_count = 0; |
| 564 |
$page_processed = false; |
| 565 |
$last_error = ''; |
| 566 |
|
| 567 |
while (!$page_processed && $retry_count < $max_retries) { |
| 568 |
try { |
| 569 |
$text = $pages[$i]->getText(); |
| 570 |
|
| 571 |
if (empty($text)) { |
| 572 |
throw new Exception("Empty text on page {$page_number}"); |
| 573 |
} |
| 574 |
|
| 575 |
$sanitized_content = $this->mxchat_sanitize_content_for_api($text); |
| 576 |
|
| 577 |
if (empty($sanitized_content)) { |
| 578 |
throw new Exception("No valid content after sanitization on page {$page_number}"); |
| 579 |
} |
| 580 |
|
| 581 |
$embedding_vector = $this->mxchat_generate_embedding($sanitized_content); |
| 582 |
|
| 583 |
if (is_string($embedding_vector)) { |
| 584 |
throw new Exception("Embedding generation failed: " . $embedding_vector); |
| 585 |
} |
| 586 |
|
| 587 |
if (!is_array($embedding_vector)) { |
| 588 |
throw new Exception("Embedding generation returned unexpected result type: " . gettype($embedding_vector)); |
| 589 |
} |
| 590 |
|
| 591 |
$metadata = array( |
| 592 |
'document_type' => 'pdf', |
| 593 |
'total_pages' => $total_pages, |
| 594 |
'current_page' => $page_number, |
| 595 |
'prev_page' => $i > 0 ? $i : null, |
| 596 |
'next_page' => $i < ($total_pages - 1) ? $i + 2 : null, |
| 597 |
'source_url' => $pdf_url |
| 598 |
); |
| 599 |
|
| 600 |
$content_with_metadata = wp_json_encode($metadata) . "\n---\n" . $sanitized_content; |
| 601 |
$page_url = esc_url($pdf_url . "#page=" . $page_number); |
| 602 |
|
| 603 |
$db_result = MxChat_Utils::submit_content_to_db($content_with_metadata, $page_url, $options['api_key']); |
| 604 |
|
| 605 |
if (is_wp_error($db_result)) { |
| 606 |
throw new Exception("Database submission failed: " . $db_result->get_error_message()); |
| 607 |
} |
| 608 |
|
| 609 |
// Success! |
| 610 |
$page_processed = true; |
| 611 |
$successful_pages++; |
| 612 |
|
| 613 |
} catch (Exception $e) { |
| 614 |
$retry_count++; |
| 615 |
$last_error = $e->getMessage(); |
| 616 |
|
| 617 |
//error_log("PDF page {$page_number} failed (attempt {$retry_count}/{$max_retries}): " . $last_error); |
| 618 |
|
| 619 |
if ($retry_count < $max_retries) { |
| 620 |
// Wait before retry (exponential backoff: 1s, 2s, 4s) |
| 621 |
sleep(pow(2, $retry_count - 1)); |
| 622 |
} |
| 623 |
} |
| 624 |
} |
| 625 |
|
| 626 |
// If page still not processed after all retries, mark as failed |
| 627 |
if (!$page_processed) { |
| 628 |
$failed_pages++; |
| 629 |
$status['failed_pages_list'][] = [ |
| 630 |
'page' => $page_number, |
| 631 |
'error' => $last_error, |
| 632 |
'time' => time(), |
| 633 |
'retries' => $max_retries |
| 634 |
]; |
| 635 |
|
| 636 |
// Limit failed pages list to prevent memory issues |
| 637 |
if (count($status['failed_pages_list']) > 50) { |
| 638 |
$status['failed_pages_list'] = array_slice($status['failed_pages_list'], -50); |
| 639 |
} |
| 640 |
|
| 641 |
//error_log("PDF page {$page_number} permanently failed after {$max_retries} attempts: " . $last_error); |
| 642 |
} |
| 643 |
|
| 644 |
// Update progress |
| 645 |
$status['processed_pages'] = absint($page_number); |
| 646 |
$status['last_update'] = time(); |
| 647 |
$status['failed_pages'] = absint($status['failed_pages'] ?? 0) + ($page_processed ? 0 : 1); |
| 648 |
|
| 649 |
set_transient($status_key, array_map('sanitize_text_field', $status), DAY_IN_SECONDS); |
| 650 |
} |
| 651 |
|
| 652 |
// Schedule next batch if needed |
| 653 |
if ($end_page < $total_pages) { |
| 654 |
wp_schedule_single_event(time() + $batch_pause, 'mxchat_process_pdf_pages', array( |
| 655 |
'pdf_path' => $pdf_path, |
| 656 |
'pdf_url' => $pdf_url, |
| 657 |
'total_pages' => $total_pages, |
| 658 |
'batch_size' => $batch_size, |
| 659 |
'batch_pause' => $batch_pause |
| 660 |
)); |
| 661 |
} else { |
| 662 |
// Processing complete |
| 663 |
$status['status'] = 'complete'; |
| 664 |
$status['processed_pages'] = $total_pages; |
| 665 |
|
| 666 |
// Add completion summary |
| 667 |
$status['completion_summary'] = [ |
| 668 |
'total_pages' => $total_pages, |
| 669 |
'successful_pages' => $total_pages - absint($status['failed_pages'] ?? 0), |
| 670 |
'failed_pages' => absint($status['failed_pages'] ?? 0), |
| 671 |
'completion_time' => current_time('mysql') |
| 672 |
]; |
| 673 |
|
| 674 |
// Save the completed status (don't delete it - let user dismiss manually) |
| 675 |
set_transient($status_key, array_map('sanitize_text_field', $status), DAY_IN_SECONDS); |
| 676 |
|
| 677 |
// Clean up the temporary PDF file |
| 678 |
if (file_exists($pdf_path)) { |
| 679 |
wp_delete_file($pdf_path); |
| 680 |
} |
| 681 |
|
| 682 |
// DON'T delete the status transients here - let user dismiss manually |
| 683 |
} |
| 684 |
|
| 685 |
} catch (\Exception $e) { |
| 686 |
//error_log(sprintf('[MXCHAT-PDF] Error processing PDF: %s', $e->getMessage())); |
| 687 |
|
| 688 |
$status_key = sanitize_key('mxchat_pdf_status_' . md5($pdf_url)); |
| 689 |
$status = get_transient($status_key); |
| 690 |
|
| 691 |
if (!$status || !is_array($status)) { |
| 692 |
$status = array( |
| 693 |
'total_pages' => $total_pages, |
| 694 |
'processed_pages' => 0, |
| 695 |
'status' => 'error', |
| 696 |
'error' => sanitize_text_field($e->getMessage()), |
| 697 |
'last_update' => time() |
| 698 |
); |
| 699 |
} else { |
| 700 |
$status['status'] = 'error'; |
| 701 |
$status['error'] = sanitize_text_field($e->getMessage()); |
| 702 |
$status['last_update'] = time(); |
| 703 |
} |
| 704 |
|
| 705 |
set_transient($status_key, array_map('sanitize_text_field', $status), DAY_IN_SECONDS); |
| 706 |
|
| 707 |
if (file_exists($pdf_path)) { |
| 708 |
wp_delete_file($pdf_path); |
| 709 |
} |
| 710 |
} |
| 711 |
} |
| 712 |
|
| 713 |
|
| 714 |
public function get_pdf_processing_status($pdf_url) { |
| 715 |
$pdf_url = esc_url_raw($pdf_url); |
| 716 |
$status = get_transient(sanitize_key('mxchat_pdf_status_' . md5($pdf_url))); |
| 717 |
|
| 718 |
if (!$status || !is_array($status)) { |
| 719 |
return false; |
| 720 |
} |
| 721 |
|
| 722 |
// Check for stalled processing (no updates for 5 minutes) |
| 723 |
if ($status['status'] === 'processing' && (time() - absint($status['last_update'])) > 300) { |
| 724 |
$status['status'] = 'error'; |
| 725 |
$status['error'] = __('PDF processing appears to be stalled. No updates for over 5 minutes.', 'mxchat'); |
| 726 |
|
| 727 |
// Save the updated status |
| 728 |
set_transient( |
| 729 |
sanitize_key('mxchat_pdf_status_' . md5($pdf_url)), |
| 730 |
array_map('sanitize_text_field', $status), |
| 731 |
DAY_IN_SECONDS |
| 732 |
); |
| 733 |
} |
| 734 |
|
| 735 |
$result = array( |
| 736 |
'total_pages' => absint($status['total_pages']), |
| 737 |
'processed_pages' => absint($status['processed_pages']), |
| 738 |
'failed_pages' => absint($status['failed_pages'] ?? 0), |
| 739 |
'percentage' => ($status['total_pages'] > 0) |
| 740 |
? round((absint($status['processed_pages']) / absint($status['total_pages'])) * 100) |
| 741 |
: 0, |
| 742 |
'status' => sanitize_text_field($status['status']), |
| 743 |
'last_update' => human_time_diff(absint($status['last_update']), time()) . ' ' . esc_html__('ago', 'mxchat'), |
| 744 |
'failed_pages_list' => isset($status['failed_pages_list']) ? $status['failed_pages_list'] : array(), |
| 745 |
'completion_summary' => isset($status['completion_summary']) ? $status['completion_summary'] : null |
| 746 |
); |
| 747 |
|
| 748 |
// Add error message if present |
| 749 |
if (isset($status['error']) && !empty($status['error'])) { |
| 750 |
$result['error'] = sanitize_text_field($status['error']); |
| 751 |
} |
| 752 |
|
| 753 |
return $result; |
| 754 |
} |
| 755 |
|
| 756 |
|
| 757 |
public function mxchat_handle_sitemap_submission() { |
| 758 |
// Start logging the submission process |
| 759 |
//error_log('[MXCHAT-URL] ===== Starting URL submission process ====='); |
| 760 |
|
| 761 |
// Check if the form was submitted and verify permissions |
| 762 |
if (!isset($_POST['submit_sitemap']) || !current_user_can('manage_options')) { |
| 763 |
//error_log('[MXCHAT-URL] Error: Unauthorized access or form not submitted properly'); |
| 764 |
wp_die(esc_html__('Unauthorized access', 'mxchat')); |
| 765 |
} |
| 766 |
|
| 767 |
// Verify nonce |
| 768 |
//error_log('[MXCHAT-URL] Verifying nonce'); |
| 769 |
check_admin_referer('mxchat_submit_sitemap_action', 'mxchat_submit_sitemap_nonce'); |
| 770 |
|
| 771 |
// Validate URL |
| 772 |
if (!isset($_POST['sitemap_url']) || empty($_POST['sitemap_url'])) { |
| 773 |
//error_log('[MXCHAT-URL] Error: Empty or missing URL'); |
| 774 |
set_transient('mxchat_admin_notice_error', |
| 775 |
esc_html__('Please provide a valid URL.', 'mxchat'), |
| 776 |
30 |
| 777 |
); |
| 778 |
wp_safe_redirect(esc_url(admin_url('admin.php?page=mxchat-prompts'))); |
| 779 |
exit; |
| 780 |
} |
| 781 |
|
| 782 |
$submitted_url = esc_url_raw($_POST['sitemap_url']); |
| 783 |
//error_log('[MXCHAT-URL] Processing URL: ' . $submitted_url); |
| 784 |
|
| 785 |
// Validate API key first |
| 786 |
$options = get_option('mxchat_options'); |
| 787 |
$selected_model = $options['embedding_model'] ?? 'text-embedding-ada-002'; |
| 788 |
|
| 789 |
if (strpos($selected_model, 'voyage') === 0) { |
| 790 |
$api_key = $options['voyage_api_key'] ?? ''; |
| 791 |
$provider_name = 'Voyage AI'; |
| 792 |
} elseif (strpos($selected_model, 'gemini-embedding') === 0) { |
| 793 |
$api_key = $options['gemini_api_key'] ?? ''; |
| 794 |
$provider_name = 'Google Gemini'; |
| 795 |
} else { |
| 796 |
$api_key = $options['api_key'] ?? ''; |
| 797 |
$provider_name = 'OpenAI'; |
| 798 |
} |
| 799 |
|
| 800 |
if (empty($api_key)) { |
| 801 |
$error_message = sprintf( |
| 802 |
esc_html__('%s API key is not configured. Please add your API key in the settings before submitting content.', 'mxchat'), |
| 803 |
$provider_name |
| 804 |
); |
| 805 |
//error_log('[MXCHAT-URL] Error: ' . $error_message); |
| 806 |
set_transient('mxchat_admin_notice_error', $error_message, 30); |
| 807 |
//error_log('[MXCHAT-URL] Set error transient: ' . $error_message); |
| 808 |
wp_safe_redirect(esc_url(admin_url('admin.php?page=mxchat-prompts'))); |
| 809 |
exit; |
| 810 |
} |
| 811 |
|
| 812 |
//error_log('[MXCHAT-URL] Fetching URL content'); |
| 813 |
$response = wp_remote_get($submitted_url, array('timeout' => 30)); |
| 814 |
|
| 815 |
if (is_wp_error($response) || wp_remote_retrieve_response_code($response) !== 200) { |
| 816 |
$error_message = is_wp_error($response) ? $response->get_error_message() : 'HTTP Status: ' . wp_remote_retrieve_response_code($response); |
| 817 |
//error_log('[MXCHAT-URL] Error fetching URL: ' . $error_message); |
| 818 |
set_transient('mxchat_admin_notice_error', |
| 819 |
sprintf( |
| 820 |
esc_html__('Failed to fetch the URL: %s', 'mxchat'), |
| 821 |
esc_html($error_message) |
| 822 |
), |
| 823 |
30 |
| 824 |
); |
| 825 |
wp_safe_redirect(esc_url(admin_url('admin.php?page=mxchat-prompts'))); |
| 826 |
exit; |
| 827 |
} |
| 828 |
|
| 829 |
$content_type = wp_remote_retrieve_header($response, 'content-type'); |
| 830 |
//error_log('[MXCHAT-URL] Content type: ' . $content_type); |
| 831 |
$body_content = wp_remote_retrieve_body($response); |
| 832 |
|
| 833 |
if (empty($body_content)) { |
| 834 |
//error_log('[MXCHAT-URL] Error: Empty response body'); |
| 835 |
set_transient('mxchat_admin_notice_error', |
| 836 |
esc_html__('Empty response received from URL.', 'mxchat'), |
| 837 |
30 |
| 838 |
); |
| 839 |
wp_safe_redirect(esc_url(admin_url('admin.php?page=mxchat-prompts'))); |
| 840 |
exit; |
| 841 |
} |
| 842 |
//error_log('[MXCHAT-URL] Retrieved body content length: ' . strlen($body_content) . ' bytes'); |
| 843 |
|
| 844 |
// Handle PDF URL |
| 845 |
if ($this->is_pdf_url($submitted_url, $response)) { |
| 846 |
//error_log('[MXCHAT-URL] Detected PDF URL, handling PDF for knowledge base'); |
| 847 |
$result = $this->handle_pdf_for_knowledge_base($submitted_url, $response); |
| 848 |
//error_log('[MXCHAT-URL] PDF handling result: ' . $result); |
| 849 |
|
| 850 |
if ($result === 'scheduled') { |
| 851 |
set_transient( |
| 852 |
'mxchat_last_pdf_url', |
| 853 |
sanitize_text_field($submitted_url), |
| 854 |
DAY_IN_SECONDS |
| 855 |
); |
| 856 |
set_transient('mxchat_admin_notice_info', |
| 857 |
esc_html__('PDF processing has started in the background. You can check the progress in the Knowledge Base section.', 'mxchat'), |
| 858 |
30 |
| 859 |
); |
| 860 |
} else { |
| 861 |
set_transient('mxchat_admin_notice_error', |
| 862 |
esc_html__('Failed to start PDF processing: ', 'mxchat') . esc_html($result), |
| 863 |
30 |
| 864 |
); |
| 865 |
} |
| 866 |
|
| 867 |
wp_safe_redirect(esc_url(admin_url('admin.php?page=mxchat-prompts'))); |
| 868 |
exit; |
| 869 |
} |
| 870 |
|
| 871 |
// Handle Sitemap XML |
| 872 |
if (strpos($content_type, 'xml') !== false || strpos($body_content, '<urlset') !== false) { |
| 873 |
//error_log('[MXCHAT-URL] Detected XML content, processing as sitemap'); |
| 874 |
libxml_use_internal_errors(true); |
| 875 |
$xml = simplexml_load_string($body_content); |
| 876 |
$xml_errors = libxml_get_errors(); |
| 877 |
libxml_clear_errors(); |
| 878 |
|
| 879 |
if ($xml === false || !empty($xml_errors)) { |
| 880 |
//error_log('[MXCHAT-URL] Error: Invalid XML format'); |
| 881 |
if (!empty($xml_errors)) { |
| 882 |
foreach ($xml_errors as $error) { |
| 883 |
//error_log('[MXCHAT-URL] XML Error: ' . $error->message); |
| 884 |
} |
| 885 |
} |
| 886 |
|
| 887 |
set_transient('mxchat_admin_notice_error', |
| 888 |
esc_html__('Invalid sitemap XML. Please provide a valid sitemap.', 'mxchat'), |
| 889 |
30 |
| 890 |
); |
| 891 |
wp_safe_redirect(esc_url(admin_url('admin.php?page=mxchat-prompts'))); |
| 892 |
exit; |
| 893 |
} |
| 894 |
|
| 895 |
//error_log('[MXCHAT-URL] Valid XML found, handling sitemap for knowledge base'); |
| 896 |
$result = $this->handle_sitemap_for_knowledge_base($xml, $submitted_url); |
| 897 |
//error_log('[MXCHAT-URL] Sitemap handling result: ' . $result); |
| 898 |
|
| 899 |
if ($result === 'scheduled') { |
| 900 |
set_transient( |
| 901 |
'mxchat_last_sitemap_url', |
| 902 |
sanitize_text_field($submitted_url), |
| 903 |
DAY_IN_SECONDS |
| 904 |
); |
| 905 |
set_transient('mxchat_admin_notice_info', |
| 906 |
esc_html__('Sitemap processing has started in the background. You can check the progress in the Knowledge Base section.', 'mxchat'), |
| 907 |
30 |
| 908 |
); |
| 909 |
} else { |
| 910 |
// Return to the admin page without a redirect for better error display |
| 911 |
// The error is already stored in the sitemap status transient |
| 912 |
set_transient('mxchat_admin_notice_error', |
| 913 |
esc_html__('Failed to start sitemap processing. Please check the status below for details.', 'mxchat'), |
| 914 |
30 |
| 915 |
); |
| 916 |
} |
| 917 |
|
| 918 |
wp_safe_redirect(esc_url(admin_url('admin.php?page=mxchat-prompts'))); |
| 919 |
exit; |
| 920 |
} |
| 921 |
|
| 922 |
// Handle Regular URL |
| 923 |
//error_log('[MXCHAT-URL] Processing as regular webpage'); |
| 924 |
$page_content = $this->mxchat_extract_main_content($body_content); |
| 925 |
//error_log('[MXCHAT-URL] Extracted content length: ' . strlen($page_content) . ' bytes'); |
| 926 |
|
| 927 |
$sanitized_content = $this->mxchat_sanitize_content_for_api($page_content); |
| 928 |
//error_log('[MXCHAT-URL] Sanitized content length: ' . strlen($sanitized_content) . ' bytes'); |
| 929 |
|
| 930 |
if (empty($sanitized_content)) { |
| 931 |
//error_log('[MXCHAT-URL] Error: No valid content after sanitization'); |
| 932 |
|
| 933 |
// Set both transients - the error notice and the URL status |
| 934 |
set_transient('mxchat_admin_notice_error', |
| 935 |
esc_html__('No valid content found on the provided URL.', 'mxchat'), |
| 936 |
30 |
| 937 |
); |
| 938 |
|
| 939 |
// Set URL status transient |
| 940 |
set_transient('mxchat_single_url_status', [ |
| 941 |
'url' => $submitted_url, |
| 942 |
'timestamp' => current_time('mysql'), |
| 943 |
'status' => 'failed', |
| 944 |
'error' => esc_html__('No valid content found on the provided URL.', 'mxchat') |
| 945 |
], DAY_IN_SECONDS); |
| 946 |
|
| 947 |
wp_safe_redirect(esc_url(admin_url('admin.php?page=mxchat-prompts'))); |
| 948 |
exit; |
| 949 |
} |
| 950 |
|
| 951 |
//error_log('[MXCHAT-URL] Generating embedding for content'); |
| 952 |
$embedding_vector = $this->mxchat_generate_embedding($sanitized_content); |
| 953 |
|
| 954 |
// Check if embedding_vector is a string (error message) |
| 955 |
if (is_string($embedding_vector)) { |
| 956 |
//error_log('[MXCHAT-URL] Error generating embedding: ' . $embedding_vector); |
| 957 |
$error_message = esc_html__('Failed to generate embedding: ', 'mxchat') . esc_html($embedding_vector); |
| 958 |
//error_log('[MXCHAT-URL] Setting error transient: ' . $error_message); |
| 959 |
|
| 960 |
// Set both transients |
| 961 |
set_transient('mxchat_admin_notice_error', $error_message, 30); |
| 962 |
|
| 963 |
// Set URL status transient |
| 964 |
set_transient('mxchat_single_url_status', [ |
| 965 |
'url' => $submitted_url, |
| 966 |
'timestamp' => current_time('mysql'), |
| 967 |
'status' => 'failed', |
| 968 |
'error' => $error_message |
| 969 |
], DAY_IN_SECONDS); |
| 970 |
|
| 971 |
wp_safe_redirect(esc_url(admin_url('admin.php?page=mxchat-prompts'))); |
| 972 |
exit; |
| 973 |
} |
| 974 |
|
| 975 |
if (is_array($embedding_vector)) { |
| 976 |
//error_log('[MXCHAT-URL] Successfully generated embedding with ' . count($embedding_vector) . ' dimensions'); |
| 977 |
|
| 978 |
$db_result = MxChat_Utils::submit_content_to_db( |
| 979 |
$sanitized_content, |
| 980 |
$submitted_url, |
| 981 |
$api_key |
| 982 |
); |
| 983 |
|
| 984 |
if (is_wp_error($db_result)) { |
| 985 |
//error_log('[MXCHAT-URL] Error: Failed to store content in database: ' . $db_result->get_error_message()); |
| 986 |
$error_message = esc_html__('Failed to store content in database: ', 'mxchat') . esc_html($db_result->get_error_message()); |
| 987 |
//error_log('[MXCHAT-URL] Setting error transient: ' . $error_message); |
| 988 |
|
| 989 |
// Set both transients |
| 990 |
set_transient('mxchat_admin_notice_error', $error_message, 30); |
| 991 |
|
| 992 |
// Set URL status transient |
| 993 |
set_transient('mxchat_single_url_status', [ |
| 994 |
'url' => $submitted_url, |
| 995 |
'timestamp' => current_time('mysql'), |
| 996 |
'status' => 'failed', |
| 997 |
'error' => $error_message |
| 998 |
], DAY_IN_SECONDS); |
| 999 |
|
| 1000 |
wp_safe_redirect(esc_url(admin_url('admin.php?page=mxchat-prompts'))); |
| 1001 |
exit; |
| 1002 |
} |
| 1003 |
|
| 1004 |
//error_log('[MXCHAT-URL] Successfully stored content in database'); |
| 1005 |
$success_message = esc_html__('URL content successfully submitted!', 'mxchat'); |
| 1006 |
//error_log('[MXCHAT-URL] Setting success transient: ' . $success_message); |
| 1007 |
|
| 1008 |
// Set both transients |
| 1009 |
set_transient('mxchat_admin_notice_success', $success_message, 30); |
| 1010 |
|
| 1011 |
// Set URL status transient with success |
| 1012 |
set_transient('mxchat_single_url_status', [ |
| 1013 |
'url' => $submitted_url, |
| 1014 |
'timestamp' => current_time('mysql'), |
| 1015 |
'status' => 'complete', |
| 1016 |
'content_length' => strlen($sanitized_content), |
| 1017 |
'embedding_dimensions' => count($embedding_vector) |
| 1018 |
], DAY_IN_SECONDS); |
| 1019 |
|
| 1020 |
} else { |
| 1021 |
//error_log('[MXCHAT-URL] Error: Failed to generate embedding. Unexpected result type: ' . gettype($embedding_vector)); |
| 1022 |
$error_message = esc_html__('Failed to generate embedding: Unexpected result type. Please check your API key and try again.', 'mxchat'); |
| 1023 |
//error_log('[MXCHAT-URL] Setting error transient: ' . $error_message); |
| 1024 |
|
| 1025 |
// Set both transients |
| 1026 |
set_transient('mxchat_admin_notice_error', $error_message, 30); |
| 1027 |
|
| 1028 |
// Set URL status transient |
| 1029 |
set_transient('mxchat_single_url_status', [ |
| 1030 |
'url' => $submitted_url, |
| 1031 |
'timestamp' => current_time('mysql'), |
| 1032 |
'status' => 'failed', |
| 1033 |
'error' => $error_message |
| 1034 |
], DAY_IN_SECONDS); |
| 1035 |
} |
| 1036 |
|
| 1037 |
//error_log('[MXCHAT-URL] ===== Completed URL submission process ====='); |
| 1038 |
wp_safe_redirect(esc_url(admin_url('admin.php?page=mxchat-prompts'))); |
| 1039 |
exit; |
| 1040 |
} |
| 1041 |
public function get_single_url_status() { |
| 1042 |
$status = get_transient('mxchat_single_url_status'); |
| 1043 |
if (!$status) { |
| 1044 |
return null; |
| 1045 |
} |
| 1046 |
|
| 1047 |
// Add human-readable time |
| 1048 |
if (isset($status['timestamp'])) { |
| 1049 |
$status['human_time'] = human_time_diff(strtotime($status['timestamp']), current_time('timestamp')) . ' ' . __('ago', 'mxchat'); |
| 1050 |
} |
| 1051 |
|
| 1052 |
return $status; |
| 1053 |
} |
| 1054 |
public function handle_sitemap_for_knowledge_base($xml, $sitemap_url) { |
| 1055 |
// Clear any single URL status when starting sitemap processing |
| 1056 |
delete_transient('mxchat_single_url_status'); |
| 1057 |
if (!current_user_can('manage_options')) { |
| 1058 |
//error_log(esc_html__('Unauthorized sitemap processing attempt', 'mxchat')); |
| 1059 |
return false; |
| 1060 |
} |
| 1061 |
|
| 1062 |
try { |
| 1063 |
$sitemap_url = esc_url_raw($sitemap_url); |
| 1064 |
|
| 1065 |
if (!$xml || !is_object($xml)) { |
| 1066 |
throw new Exception(__('Invalid XML object provided', 'mxchat')); |
| 1067 |
} |
| 1068 |
|
| 1069 |
// Add embedding validation before processing |
| 1070 |
// Test embedding with a small sample text to verify API key is working |
| 1071 |
$test_result = $this->mxchat_generate_embedding("This is a test to verify the embedding API key is working."); |
| 1072 |
|
| 1073 |
// Check if test_result is a string (error message) rather than an array (valid embedding) |
| 1074 |
if (is_string($test_result)) { |
| 1075 |
//error_log('[MXCHAT-URL] Embedding API validation failed: ' . $test_result); |
| 1076 |
|
| 1077 |
// Store the error in the status transient so it can be displayed later |
| 1078 |
$status_data = array( |
| 1079 |
'total_urls' => 0, |
| 1080 |
'processed_urls' => 0, |
| 1081 |
'status' => 'error', |
| 1082 |
'error' => __('Embedding API validation failed: ', 'mxchat') . $test_result, |
| 1083 |
'last_update' => time() |
| 1084 |
); |
| 1085 |
|
| 1086 |
set_transient( |
| 1087 |
sanitize_key('mxchat_sitemap_status_' . md5($sitemap_url)), |
| 1088 |
array_map('sanitize_text_field', $status_data), |
| 1089 |
DAY_IN_SECONDS |
| 1090 |
); |
| 1091 |
|
| 1092 |
throw new Exception(__('Embedding API validation failed: ', 'mxchat') . $test_result); |
| 1093 |
} |
| 1094 |
|
| 1095 |
// Make sure it's an array (valid embedding) |
| 1096 |
if (!is_array($test_result)) { |
| 1097 |
//error_log('[MXCHAT-URL] Embedding API returned unexpected result type: ' . gettype($test_result)); |
| 1098 |
throw new Exception(__('Embedding API returned unexpected result type. Please check your configuration.', 'mxchat')); |
| 1099 |
} |
| 1100 |
|
| 1101 |
$urls = []; |
| 1102 |
foreach ($xml->url as $url_element) { |
| 1103 |
$url = esc_url_raw((string)$url_element->loc); |
| 1104 |
if ($url) { |
| 1105 |
$urls[] = $url; |
| 1106 |
} |
| 1107 |
} |
| 1108 |
|
| 1109 |
$total_urls = absint(count($urls)); |
| 1110 |
|
| 1111 |
if ($total_urls < 1) { |
| 1112 |
throw new Exception(__('No valid URLs found in sitemap', 'mxchat')); |
| 1113 |
} |
| 1114 |
|
| 1115 |
wp_schedule_single_event(time(), 'mxchat_process_sitemap_urls', array( |
| 1116 |
'urls' => $urls, |
| 1117 |
'sitemap_url' => $sitemap_url, |
| 1118 |
'total_urls' => $total_urls, |
| 1119 |
'batch_size' => absint(10), |
| 1120 |
'batch_pause' => absint(5) |
| 1121 |
)); |
| 1122 |
|
| 1123 |
$status_data = array( |
| 1124 |
'total_urls' => $total_urls, |
| 1125 |
'processed_urls' => 0, |
| 1126 |
'status' => 'processing', |
| 1127 |
'last_update' => time() |
| 1128 |
); |
| 1129 |
|
| 1130 |
set_transient( |
| 1131 |
sanitize_key('mxchat_sitemap_status_' . md5($sitemap_url)), |
| 1132 |
array_map('sanitize_text_field', $status_data), |
| 1133 |
DAY_IN_SECONDS |
| 1134 |
); |
| 1135 |
|
| 1136 |
return __('scheduled', 'mxchat'); |
| 1137 |
|
| 1138 |
} catch (\Exception $e) { |
| 1139 |
$error_message = $e->getMessage(); |
| 1140 |
//error_log(sprintf(esc_html__('Error preparing sitemap for processing: %s', 'mxchat'), esc_html($error_message))); |
| 1141 |
|
| 1142 |
// Store the sitemap URL and error in transients so they can be displayed |
| 1143 |
set_transient( |
| 1144 |
'mxchat_last_sitemap_url', |
| 1145 |
sanitize_text_field($sitemap_url), |
| 1146 |
DAY_IN_SECONDS |
| 1147 |
); |
| 1148 |
|
| 1149 |
$status_data = array( |
| 1150 |
'total_urls' => 0, |
| 1151 |
'processed_urls' => 0, |
| 1152 |
'status' => 'error', |
| 1153 |
'error' => $error_message, |
| 1154 |
'last_update' => time() |
| 1155 |
); |
| 1156 |
|
| 1157 |
set_transient( |
| 1158 |
sanitize_key('mxchat_sitemap_status_' . md5($sitemap_url)), |
| 1159 |
array_map('sanitize_text_field', $status_data), |
| 1160 |
DAY_IN_SECONDS |
| 1161 |
); |
| 1162 |
|
| 1163 |
return $error_message; |
| 1164 |
} |
| 1165 |
} |
| 1166 |
|
| 1167 |
public function process_sitemap_urls_cron($urls, $sitemap_url, $total_urls, $batch_size, $batch_pause) { |
| 1168 |
// Validate inputs |
| 1169 |
$sitemap_url = esc_url_raw($sitemap_url); |
| 1170 |
$total_urls = absint($total_urls); |
| 1171 |
$batch_size = absint($batch_size); |
| 1172 |
$batch_pause = absint($batch_pause); |
| 1173 |
|
| 1174 |
if (!is_array($urls) || empty($urls)) { |
| 1175 |
return; |
| 1176 |
} |
| 1177 |
|
| 1178 |
try { |
| 1179 |
$status_key = sanitize_key('mxchat_sitemap_status_' . md5($sitemap_url)); |
| 1180 |
$status = get_transient($status_key); |
| 1181 |
|
| 1182 |
if (!$status || !is_array($status)) { |
| 1183 |
throw new Exception('Invalid status data retrieved from transient'); |
| 1184 |
} |
| 1185 |
|
| 1186 |
// Initialize failed_urls array if it doesn't exist |
| 1187 |
if (!isset($status['failed_urls_list']) || !is_array($status['failed_urls_list'])) { |
| 1188 |
$status['failed_urls_list'] = []; |
| 1189 |
} |
| 1190 |
|
| 1191 |
$start_url = absint($status['processed_urls']); |
| 1192 |
$end_url = min($start_url + $batch_size, $total_urls); |
| 1193 |
|
| 1194 |
// Track batch statistics |
| 1195 |
$batch_stats = [ |
| 1196 |
'processed' => 0, |
| 1197 |
'failed' => 0, |
| 1198 |
'last_error' => '', |
| 1199 |
'embedding_errors' => 0 |
| 1200 |
]; |
| 1201 |
|
| 1202 |
// Check embedding configuration with first URL (only on first batch) |
| 1203 |
if ($start_url === 0) { |
| 1204 |
$test_url = esc_url_raw($urls[0]); |
| 1205 |
$test_response = wp_remote_get($test_url); |
| 1206 |
|
| 1207 |
if (!is_wp_error($test_response) && wp_remote_retrieve_response_code($test_response) === 200) { |
| 1208 |
$test_html = wp_remote_retrieve_body($test_response); |
| 1209 |
$test_content = $this->mxchat_extract_main_content($test_html); |
| 1210 |
$test_sanitized = $this->mxchat_sanitize_content_for_api($test_content); |
| 1211 |
|
| 1212 |
if (!empty($test_sanitized)) { |
| 1213 |
$test_embedding = $this->mxchat_generate_embedding($test_sanitized); |
| 1214 |
|
| 1215 |
if (is_string($test_embedding)) { |
| 1216 |
throw new Exception('Embedding generation failed: ' . $test_embedding); |
| 1217 |
} |
| 1218 |
|
| 1219 |
if (!is_array($test_embedding)) { |
| 1220 |
throw new Exception('Embedding generation returned unexpected result type: ' . gettype($test_embedding)); |
| 1221 |
} |
| 1222 |
} |
| 1223 |
} |
| 1224 |
} |
| 1225 |
|
| 1226 |
for ($i = $start_url; $i < $end_url; $i++) { |
| 1227 |
$page_url = esc_url_raw($urls[$i]); |
| 1228 |
$max_retries = 3; |
| 1229 |
$retry_count = 0; |
| 1230 |
$url_processed = false; |
| 1231 |
$last_error = ''; |
| 1232 |
|
| 1233 |
while (!$url_processed && $retry_count < $max_retries) { |
| 1234 |
try { |
| 1235 |
// Attempt to fetch the URL |
| 1236 |
$page_response = wp_remote_get($page_url, array('timeout' => 30)); |
| 1237 |
|
| 1238 |
if (is_wp_error($page_response)) { |
| 1239 |
throw new Exception('HTTP request failed: ' . $page_response->get_error_message()); |
| 1240 |
} |
| 1241 |
|
| 1242 |
$response_code = wp_remote_retrieve_response_code($page_response); |
| 1243 |
if ($response_code !== 200) { |
| 1244 |
throw new Exception('HTTP Status: ' . $response_code); |
| 1245 |
} |
| 1246 |
|
| 1247 |
$page_html = wp_remote_retrieve_body($page_response); |
| 1248 |
|
| 1249 |
if (empty($page_html)) { |
| 1250 |
throw new Exception('Empty response body'); |
| 1251 |
} |
| 1252 |
|
| 1253 |
$page_content = $this->mxchat_extract_main_content($page_html); |
| 1254 |
$sanitized_content = $this->mxchat_sanitize_content_for_api($page_content); |
| 1255 |
|
| 1256 |
if (empty($sanitized_content)) { |
| 1257 |
throw new Exception('No valid content found after processing'); |
| 1258 |
} |
| 1259 |
|
| 1260 |
$embedding_vector = $this->mxchat_generate_embedding($sanitized_content); |
| 1261 |
|
| 1262 |
if (is_string($embedding_vector)) { |
| 1263 |
throw new Exception('Embedding generation failed: ' . $embedding_vector); |
| 1264 |
} |
| 1265 |
|
| 1266 |
if (!is_array($embedding_vector)) { |
| 1267 |
throw new Exception('Embedding generation returned unexpected result type: ' . gettype($embedding_vector)); |
| 1268 |
} |
| 1269 |
|
| 1270 |
// Submit to database |
| 1271 |
$options = get_option('mxchat_options'); |
| 1272 |
$submission_result = MxChat_Utils::submit_content_to_db($sanitized_content, $page_url, $options['api_key']); |
| 1273 |
|
| 1274 |
if (is_wp_error($submission_result)) { |
| 1275 |
throw new Exception('Database submission failed: ' . $submission_result->get_error_message()); |
| 1276 |
} |
| 1277 |
|
| 1278 |
// Success! |
| 1279 |
$url_processed = true; |
| 1280 |
$batch_stats['processed']++; |
| 1281 |
|
| 1282 |
} catch (Exception $e) { |
| 1283 |
$retry_count++; |
| 1284 |
$last_error = $e->getMessage(); |
| 1285 |
|
| 1286 |
//error_log("URL {$page_url} failed (attempt {$retry_count}/{$max_retries}): " . $last_error); |
| 1287 |
|
| 1288 |
// Track embedding errors specifically |
| 1289 |
if (strpos($last_error, 'Embedding') !== false) { |
| 1290 |
$batch_stats['embedding_errors']++; |
| 1291 |
} |
| 1292 |
|
| 1293 |
if ($retry_count < $max_retries) { |
| 1294 |
// Wait before retry (exponential backoff: 1s, 2s, 4s) |
| 1295 |
sleep(pow(2, $retry_count - 1)); |
| 1296 |
} |
| 1297 |
} |
| 1298 |
} |
| 1299 |
|
| 1300 |
// If URL still not processed after all retries, mark as failed |
| 1301 |
if (!$url_processed) { |
| 1302 |
$batch_stats['failed']++; |
| 1303 |
$batch_stats['last_error'] = $last_error; |
| 1304 |
|
| 1305 |
// Add to failed URLs list |
| 1306 |
$status['failed_urls_list'][] = [ |
| 1307 |
'url' => $page_url, |
| 1308 |
'error' => $last_error, |
| 1309 |
'time' => time(), |
| 1310 |
'retries' => $max_retries |
| 1311 |
]; |
| 1312 |
|
| 1313 |
// Limit the number of failed URLs we store to prevent transient size issues |
| 1314 |
if (count($status['failed_urls_list']) > 100) { |
| 1315 |
$status['failed_urls_list'] = array_slice($status['failed_urls_list'], -100); |
| 1316 |
} |
| 1317 |
} |
| 1318 |
|
| 1319 |
// Update progress |
| 1320 |
$status['processed_urls'] = absint($i + 1); |
| 1321 |
$status['last_update'] = time(); |
| 1322 |
$status['failed_urls'] = absint($status['failed_urls'] ?? 0) + ($url_processed ? 0 : 1); |
| 1323 |
$status['last_error'] = $batch_stats['last_error']; |
| 1324 |
|
| 1325 |
set_transient($status_key, $status, DAY_IN_SECONDS); |
| 1326 |
|
| 1327 |
// If we have too many consecutive embedding errors, stop processing |
| 1328 |
if ($batch_stats['embedding_errors'] >= 10) { |
| 1329 |
throw new Exception('Too many consecutive embedding failures detected. Please check your API configuration.'); |
| 1330 |
} |
| 1331 |
} |
| 1332 |
|
| 1333 |
// If all URLs in this batch failed, stop processing |
| 1334 |
if ($batch_stats['processed'] === 0 && $batch_stats['failed'] > 0) { |
| 1335 |
$status['status'] = 'error'; |
| 1336 |
$status['error'] = sprintf( |
| 1337 |
'Processing stopped: %d consecutive failures in batch. Last error: %s', |
| 1338 |
$batch_stats['failed'], |
| 1339 |
$batch_stats['last_error'] |
| 1340 |
); |
| 1341 |
set_transient($status_key, $status, DAY_IN_SECONDS); |
| 1342 |
return; |
| 1343 |
} |
| 1344 |
|
| 1345 |
// Update final progress |
| 1346 |
$status['processed_urls'] = min($end_url, $total_urls); |
| 1347 |
$status['last_update'] = time(); |
| 1348 |
set_transient($status_key, $status, DAY_IN_SECONDS); |
| 1349 |
|
| 1350 |
// Check if we've processed all URLs |
| 1351 |
if ($end_url >= $total_urls) { |
| 1352 |
// All URLs have been processed - mark as complete |
| 1353 |
$status['status'] = 'complete'; |
| 1354 |
$status['processed_urls'] = $total_urls; |
| 1355 |
|
| 1356 |
// Add completion summary |
| 1357 |
$status['completion_summary'] = [ |
| 1358 |
'total_urls' => $total_urls, |
| 1359 |
'successful_urls' => $total_urls - absint($status['failed_urls'] ?? 0), |
| 1360 |
'failed_urls' => absint($status['failed_urls'] ?? 0), |
| 1361 |
'completion_time' => current_time('mysql') |
| 1362 |
]; |
| 1363 |
|
| 1364 |
// Save the completed status (don't delete it - let user dismiss manually) |
| 1365 |
set_transient($status_key, $status, DAY_IN_SECONDS); |
| 1366 |
|
| 1367 |
// DON'T delete the status transients here - let user dismiss manually |
| 1368 |
} else { |
| 1369 |
// Schedule next batch |
| 1370 |
wp_schedule_single_event(time() + $batch_pause, 'mxchat_process_sitemap_urls', array( |
| 1371 |
'urls' => $urls, |
| 1372 |
'sitemap_url' => $sitemap_url, |
| 1373 |
'total_urls' => $total_urls, |
| 1374 |
'batch_size' => $batch_size, |
| 1375 |
'batch_pause' => $batch_pause, |
| 1376 |
)); |
| 1377 |
} |
| 1378 |
} catch (\Exception $e) { |
| 1379 |
$status['status'] = 'error'; |
| 1380 |
$status['error'] = $e->getMessage(); |
| 1381 |
set_transient($status_key, $status, DAY_IN_SECONDS); |
| 1382 |
} |
| 1383 |
} |
| 1384 |
public function mxchat_sanitize_content_for_api($content) { |
| 1385 |
//error_log('[MXCHAT-SANITIZE] Original content preview: ' . substr($content, 0, 500) . '...'); |
| 1386 |
|
| 1387 |
// Remove script, style tags, and HTML comments |
| 1388 |
$content = preg_replace('/<script\b[^>]*>(.*?)<\/script>/is', '', $content); |
| 1389 |
$content = preg_replace('/<style\b[^>]*>(.*?)<\/style>/is', '', $content); |
| 1390 |
$content = preg_replace('/<!--(.|\s)*?-->/', '', $content); |
| 1391 |
|
| 1392 |
// Remove all HTML tags and decode HTML entities |
| 1393 |
$content = wp_strip_all_tags($content); |
| 1394 |
$content = html_entity_decode($content, ENT_QUOTES | ENT_HTML5); |
| 1395 |
|
| 1396 |
// Normalize whitespace but preserve paragraph breaks |
| 1397 |
// First, normalize line endings to \n |
| 1398 |
$content = str_replace(["\r\n", "\r"], "\n", $content); |
| 1399 |
// Replace multiple spaces/tabs with single space, but preserve newlines |
| 1400 |
$content = preg_replace('/[ \t]+/', ' ', $content); |
| 1401 |
// Replace 3+ newlines with 2 newlines (max 2 blank lines) |
| 1402 |
$content = preg_replace('/\n{3,}/', "\n\n", $content); |
| 1403 |
// Trim each line |
| 1404 |
$lines = explode("\n", $content); |
| 1405 |
$lines = array_map('trim', $lines); |
| 1406 |
$content = implode("\n", $lines); |
| 1407 |
// Final trim |
| 1408 |
$content = trim($content); |
| 1409 |
|
| 1410 |
// Remove control characters (which can cause database issues) but preserve \n (0x0A) and \r (0x0D) |
| 1411 |
$content = preg_replace('/[\x00-\x08\x0B\x0C\x0E-\x1F\x7F]/u', '', $content); |
| 1412 |
|
| 1413 |
// Remove NULL bytes which can cause database errors |
| 1414 |
$content = str_replace("\0", "", $content); |
| 1415 |
|
| 1416 |
// Ensure valid UTF-8 encoding |
| 1417 |
$content = wp_check_invalid_utf8($content); |
| 1418 |
|
| 1419 |
// Remove any extremely long strings without spaces (often garbage) |
| 1420 |
$content = preg_replace('/\S{300,}/', ' ', $content); |
| 1421 |
|
| 1422 |
// Replace problematic characters that often cause database issues |
| 1423 |
$content = preg_replace('/[\x{10000}-\x{10FFFF}]/u', '', $content); // Remove emoji and other high Unicode characters |
| 1424 |
|
| 1425 |
// Replace any remaining potentially problematic characters with spaces |
| 1426 |
// BUT preserve newlines by temporarily replacing them |
| 1427 |
$content = str_replace("\n", "NEWLINE_PLACEHOLDER", $content); |
| 1428 |
$content = preg_replace('/[^\p{L}\p{N}\p{P}\p{Z}\p{Sm}]/u', ' ', $content); |
| 1429 |
$content = str_replace("NEWLINE_PLACEHOLDER", "\n", $content); |
| 1430 |
|
| 1431 |
// Limit to reasonable length if needed |
| 1432 |
$max_length = 65000; // Just under MySQL TEXT field limit |
| 1433 |
if (strlen($content) > $max_length) { |
| 1434 |
$content = substr($content, 0, $max_length); |
| 1435 |
} |
| 1436 |
|
| 1437 |
//error_log('[MXCHAT-SANITIZE] Sanitized content preview: ' . substr($content, 0, 500) . '...'); |
| 1438 |
return $content; |
| 1439 |
} |
| 1440 |
public static function mxchat_extract_main_content($html) { |
| 1441 |
if (empty($html)) { |
| 1442 |
return ''; |
| 1443 |
} |
| 1444 |
try { |
| 1445 |
$dom = new DOMDocument; |
| 1446 |
libxml_use_internal_errors(true); // Suppress HTML parsing errors |
| 1447 |
@$dom->loadHTML(mb_convert_encoding($html, 'HTML-ENTITIES', 'UTF-8'), LIBXML_HTML_NOIMPLIED | LIBXML_HTML_NODEFDTD); |
| 1448 |
$xpath = new DOMXPath($dom); |
| 1449 |
|
| 1450 |
// For debugging purposes |
| 1451 |
$debugEnabled = false; // Set to true to enable debugging output |
| 1452 |
$debug = function($message) use ($debugEnabled) { |
| 1453 |
if ($debugEnabled) { |
| 1454 |
//error_log('[MXCHAT-DEBUG] ' . $message); |
| 1455 |
} |
| 1456 |
}; |
| 1457 |
|
| 1458 |
// Direct targeting for Gerow theme posts |
| 1459 |
$post_text = $xpath->query('//div[contains(@class, "post-text")]'); |
| 1460 |
if ($post_text && $post_text->length > 0) { |
| 1461 |
$debug("Found post-text directly"); |
| 1462 |
$content = ''; |
| 1463 |
foreach ($post_text as $node) { |
| 1464 |
$content .= $dom->saveHTML($node); |
| 1465 |
} |
| 1466 |
if (!empty($content)) { |
| 1467 |
$debug("Returning post-text content"); |
| 1468 |
return $content; |
| 1469 |
} |
| 1470 |
} |
| 1471 |
|
| 1472 |
// Try to get the blog details content which contains the post-text |
| 1473 |
$blog_details = $xpath->query('//div[contains(@class, "blog-details-content")]'); |
| 1474 |
if ($blog_details && $blog_details->length > 0) { |
| 1475 |
$debug("Found blog-details-content"); |
| 1476 |
$content = ''; |
| 1477 |
foreach ($blog_details as $node) { |
| 1478 |
$content .= $dom->saveHTML($node); |
| 1479 |
} |
| 1480 |
if (!empty($content)) { |
| 1481 |
$debug("Returning blog-details-content"); |
| 1482 |
return $content; |
| 1483 |
} |
| 1484 |
} |
| 1485 |
|
| 1486 |
// Try to get the article which contains the blog details |
| 1487 |
$article = $xpath->query('//article[contains(@class, "blog-details-wrap")]'); |
| 1488 |
if ($article && $article->length > 0) { |
| 1489 |
$debug("Found article with blog-details-wrap"); |
| 1490 |
$content = ''; |
| 1491 |
foreach ($article as $node) { |
| 1492 |
$content .= $dom->saveHTML($node); |
| 1493 |
} |
| 1494 |
if (!empty($content)) { |
| 1495 |
$debug("Returning article content"); |
| 1496 |
return $content; |
| 1497 |
} |
| 1498 |
} |
| 1499 |
|
| 1500 |
// Try even broader with the blog-item-wrap |
| 1501 |
$blog_item = $xpath->query('//div[contains(@class, "blog-item-wrap")]'); |
| 1502 |
if ($blog_item && $blog_item->length > 0) { |
| 1503 |
$debug("Found blog-item-wrap"); |
| 1504 |
$content = ''; |
| 1505 |
foreach ($blog_item as $node) { |
| 1506 |
$content .= $dom->saveHTML($node); |
| 1507 |
} |
| 1508 |
if (!empty($content)) { |
| 1509 |
$debug("Returning blog-item-wrap content"); |
| 1510 |
return $content; |
| 1511 |
} |
| 1512 |
} |
| 1513 |
|
| 1514 |
// Specific Gerow theme path |
| 1515 |
$gerow_path = $xpath->query('//section[contains(@class, "blog-area")]//div[contains(@class, "post-text")]'); |
| 1516 |
if ($gerow_path && $gerow_path->length > 0) { |
| 1517 |
$debug("Found Gerow theme path to post-text"); |
| 1518 |
$content = ''; |
| 1519 |
foreach ($gerow_path as $node) { |
| 1520 |
$content .= $dom->saveHTML($node); |
| 1521 |
} |
| 1522 |
if (!empty($content)) { |
| 1523 |
$debug("Returning Gerow post-text content"); |
| 1524 |
return $content; |
| 1525 |
} |
| 1526 |
} |
| 1527 |
|
| 1528 |
// Generic blog post selectors |
| 1529 |
$selectors = [ |
| 1530 |
// Blog post specific selectors |
| 1531 |
'//div[contains(@class, "post-text")]', |
| 1532 |
'//article[contains(@class, "blog-post-item")]//div[contains(@class, "post-text")]', |
| 1533 |
'//div[contains(@class, "blog-details-content")]', |
| 1534 |
'//article[contains(@class, "blog-details-wrap")]', |
| 1535 |
'//div[contains(@class, "entry-content")]', |
| 1536 |
'//div[contains(@class, "blog-content")]', |
| 1537 |
'//div[contains(@class, "blog-item-wrap")]', |
| 1538 |
|
| 1539 |
// More general content selectors |
| 1540 |
'//div[contains(@class, "page__content")]', |
| 1541 |
'//div[contains(@class, "elementor-widget-container")]', |
| 1542 |
'//div[contains(@class, "elementor-text-editor")]', |
| 1543 |
'//div[contains(@class, "elementor-widget-text-editor")]', |
| 1544 |
'//*[contains(@class, "entry-content")]', |
| 1545 |
'//*[contains(@class, "post-content")]', |
| 1546 |
'//*[contains(@class, "article-content")]', |
| 1547 |
'//*[@id="content"]', |
| 1548 |
'//*[@id="main-content"]', |
| 1549 |
'//section[contains(@class, "blog-area")]', |
| 1550 |
'//article', |
| 1551 |
'//main', |
| 1552 |
'//div[contains(@class, "content")]' |
| 1553 |
]; |
| 1554 |
|
| 1555 |
// First handle Elementor content |
| 1556 |
$debug("Checking for Elementor content"); |
| 1557 |
$elementor_widgets = $xpath->query('//div[contains(@class, "elementor-element")]//div[contains(@class, "elementor-widget-container")]'); |
| 1558 |
if ($elementor_widgets && $elementor_widgets->length > 0) { |
| 1559 |
$debug("Found Elementor widgets"); |
| 1560 |
$combined_content = ''; |
| 1561 |
foreach ($elementor_widgets as $widget) { |
| 1562 |
$widget_content = $dom->saveHTML($widget); |
| 1563 |
if (!empty($widget_content)) { |
| 1564 |
$combined_content .= $widget_content; |
| 1565 |
} |
| 1566 |
} |
| 1567 |
if (!empty($combined_content)) { |
| 1568 |
$debug("Returning Elementor content"); |
| 1569 |
return $combined_content; |
| 1570 |
} |
| 1571 |
} |
| 1572 |
|
| 1573 |
// Try standard selectors one by one |
| 1574 |
foreach ($selectors as $selector) { |
| 1575 |
$debug("Trying selector: " . $selector); |
| 1576 |
$nodes = $xpath->query($selector); |
| 1577 |
if ($nodes && $nodes->length > 0) { |
| 1578 |
$debug("Found matches for selector: " . $selector); |
| 1579 |
$content = ''; |
| 1580 |
foreach ($nodes as $node) { |
| 1581 |
$content .= $dom->saveHTML($node); |
| 1582 |
} |
| 1583 |
if (!empty($content)) { |
| 1584 |
$debug("Returning content from selector: " . $selector); |
| 1585 |
return $content; |
| 1586 |
} |
| 1587 |
} |
| 1588 |
} |
| 1589 |
|
| 1590 |
// Manual regex fallback for post-text if DOM methods fail |
| 1591 |
$debug("Trying regex fallback"); |
| 1592 |
if (preg_match('/<div class="post-text">(.*?)<\/div>\s*<\/div>\s*<\/div>/s', $html, $matches)) { |
| 1593 |
$debug("Found post-text via regex"); |
| 1594 |
return '<div class="post-text">' . $matches[1] . '</div>'; |
| 1595 |
} |
| 1596 |
|
| 1597 |
// Try to extract the blog section as a whole |
| 1598 |
$blog_section = $xpath->query('//section[contains(@class, "blog-area")]'); |
| 1599 |
if ($blog_section && $blog_section->length > 0) { |
| 1600 |
$debug("Found blog-area section"); |
| 1601 |
$content = ''; |
| 1602 |
foreach ($blog_section as $node) { |
| 1603 |
$content .= $dom->saveHTML($node); |
| 1604 |
} |
| 1605 |
if (!empty($content)) { |
| 1606 |
$debug("Returning blog-area section content"); |
| 1607 |
return $content; |
| 1608 |
} |
| 1609 |
} |
| 1610 |
|
| 1611 |
// Fallback: Return the body content if no specific selector matches |
| 1612 |
$debug("Using body fallback"); |
| 1613 |
$body = $dom->getElementsByTagName('body'); |
| 1614 |
if ($body->length > 0) { |
| 1615 |
return $dom->saveHTML($body->item(0)); |
| 1616 |
} |
| 1617 |
|
| 1618 |
// Last resort: return the original HTML |
| 1619 |
$debug("Returning original HTML"); |
| 1620 |
return $html; |
| 1621 |
} catch (Exception $e) { |
| 1622 |
//error_log('[MXCHAT-ERROR] Content extraction failed: ' . $e->getMessage()); |
| 1623 |
return $html; // Return original HTML if parsing fails |
| 1624 |
} finally { |
| 1625 |
libxml_clear_errors(); |
| 1626 |
} |
| 1627 |
} |
| 1628 |
public function get_sitemap_processing_status($sitemap_url) { |
| 1629 |
$sitemap_url = esc_url_raw($sitemap_url); |
| 1630 |
$status_key = sanitize_key('mxchat_sitemap_status_' . md5($sitemap_url)); |
| 1631 |
$status = get_transient($status_key); |
| 1632 |
|
| 1633 |
if (!$status || !is_array($status)) { |
| 1634 |
return false; |
| 1635 |
} |
| 1636 |
|
| 1637 |
// Auto-complete check: if all URLs are processed but status isn't complete |
| 1638 |
if (isset($status['processed_urls']) && isset($status['total_urls']) && |
| 1639 |
$status['processed_urls'] >= $status['total_urls'] && |
| 1640 |
isset($status['status']) && $status['status'] !== 'complete' && |
| 1641 |
$status['status'] !== 'error') { |
| 1642 |
|
| 1643 |
// Mark as complete |
| 1644 |
$status['status'] = 'complete'; |
| 1645 |
$status['processed_urls'] = $status['total_urls']; // Ensure exact match |
| 1646 |
|
| 1647 |
// Update the transient with the corrected status |
| 1648 |
set_transient($status_key, $status, DAY_IN_SECONDS); |
| 1649 |
} |
| 1650 |
|
| 1651 |
return array( |
| 1652 |
'total_urls' => absint($status['total_urls']), |
| 1653 |
'processed_urls' => absint($status['processed_urls']), |
| 1654 |
'failed_urls' => absint($status['failed_urls'] ?? 0), |
| 1655 |
'percentage' => ($status['total_urls'] > 0) |
| 1656 |
? round((absint($status['processed_urls']) / absint($status['total_urls'])) * 100) |
| 1657 |
: 0, |
| 1658 |
'status' => sanitize_text_field($status['status']), |
| 1659 |
'last_update' => human_time_diff(absint($status['last_update']), time()) . ' ' . esc_html__('ago', 'mxchat'), |
| 1660 |
'error' => isset($status['error']) ? sanitize_text_field($status['error']) : '', |
| 1661 |
'last_error' => isset($status['last_error']) ? sanitize_text_field($status['last_error']) : '', |
| 1662 |
'failed_urls_list' => isset($status['failed_urls_list']) ? $status['failed_urls_list'] : array() |
| 1663 |
); |
| 1664 |
} |
| 1665 |
|
| 1666 |
public function ajax_get_status_updates() { |
| 1667 |
try { |
| 1668 |
// Verify the request |
| 1669 |
check_ajax_referer('mxchat_status_nonce', 'nonce'); |
| 1670 |
|
| 1671 |
// Get the status just like in your admin page |
| 1672 |
$pdf_url = get_transient('mxchat_last_pdf_url'); |
| 1673 |
$sitemap_url = get_transient('mxchat_last_sitemap_url'); |
| 1674 |
|
| 1675 |
$pdf_status = $pdf_url ? $this->get_pdf_processing_status($pdf_url) : false; |
| 1676 |
$sitemap_status = $sitemap_url ? $this->get_sitemap_processing_status($sitemap_url) : false; |
| 1677 |
|
| 1678 |
// Add the PDF URL to the status object |
| 1679 |
if ($pdf_status && $pdf_url) { |
| 1680 |
$pdf_status['pdf_url'] = $pdf_url; |
| 1681 |
} |
| 1682 |
|
| 1683 |
// Set the current PDF URL for the manual batch processing button |
| 1684 |
$current_pdf_url = $pdf_url; |
| 1685 |
|
| 1686 |
// Check for true processing status, not just presence of status |
| 1687 |
$is_active_processing = |
| 1688 |
($sitemap_status && isset($sitemap_status['status']) && $sitemap_status['status'] === 'processing') || |
| 1689 |
($pdf_status && isset($pdf_status['status']) && $pdf_status['status'] === 'processing'); |
| 1690 |
|
| 1691 |
// Get single URL status, but only if no processing is active |
| 1692 |
$single_url_status = !$is_active_processing ? $this->get_single_url_status() : false; |
| 1693 |
|
| 1694 |
// REMOVED: Auto-clearing of completed status - now only done via dismiss button |
| 1695 |
|
| 1696 |
// Return JSON response with the status data |
| 1697 |
wp_send_json(array( |
| 1698 |
'pdf_status' => $pdf_status, |
| 1699 |
'sitemap_status' => $sitemap_status, |
| 1700 |
'single_url_status' => $single_url_status, |
| 1701 |
'is_processing' => $is_active_processing, |
| 1702 |
'current_pdf_url' => $current_pdf_url |
| 1703 |
)); |
| 1704 |
|
| 1705 |
} catch (Exception $e) { |
| 1706 |
// Log the error |
| 1707 |
//error_log('MxChat Status Update Error: ' . $e->getMessage()); |
| 1708 |
|
| 1709 |
// Return a friendly error response |
| 1710 |
wp_send_json_error(array( |
| 1711 |
'message' => 'Error getting status updates: ' . $e->getMessage(), |
| 1712 |
'status' => 'error' |
| 1713 |
)); |
| 1714 |
} |
| 1715 |
} |
| 1716 |
public function mxchat_stop_processing() { |
| 1717 |
// Verify permissions |
| 1718 |
if (!current_user_can('manage_options')) { |
| 1719 |
wp_die(esc_html__('Unauthorized access', 'mxchat')); |
| 1720 |
} |
| 1721 |
|
| 1722 |
// Verify nonce |
| 1723 |
check_admin_referer('mxchat_stop_processing_action', 'mxchat_stop_processing_nonce'); |
| 1724 |
|
| 1725 |
// Get the last sitemap URL and clear its transient |
| 1726 |
$sitemap_url = get_transient('mxchat_last_sitemap_url'); |
| 1727 |
if ($sitemap_url) { |
| 1728 |
delete_transient('mxchat_sitemap_status_' . md5($sitemap_url)); |
| 1729 |
delete_transient('mxchat_last_sitemap_url'); |
| 1730 |
} |
| 1731 |
|
| 1732 |
// Get the last PDF URL and clear its transient |
| 1733 |
$pdf_url = get_transient('mxchat_last_pdf_url'); |
| 1734 |
if ($pdf_url) { |
| 1735 |
delete_transient('mxchat_pdf_status_' . md5($pdf_url)); |
| 1736 |
delete_transient('mxchat_last_pdf_url'); |
| 1737 |
} |
| 1738 |
|
| 1739 |
// Unschedule any pending sitemap events |
| 1740 |
$timestamp = wp_next_scheduled('mxchat_process_sitemap_urls'); |
| 1741 |
if ($timestamp) { |
| 1742 |
wp_unschedule_event($timestamp, 'mxchat_process_sitemap_urls'); |
| 1743 |
} |
| 1744 |
|
| 1745 |
// Redirect back with a success message |
| 1746 |
set_transient('mxchat_admin_notice_success', |
| 1747 |
esc_html__('Processing has been stopped successfully.', 'mxchat'), |
| 1748 |
30 |
| 1749 |
); |
| 1750 |
wp_safe_redirect(admin_url('admin.php?page=mxchat-prompts')); |
| 1751 |
exit; |
| 1752 |
} |
| 1753 |
public function ajax_mxchat_get_content_list() { |
| 1754 |
// Verify the nonce |
| 1755 |
check_ajax_referer('mxchat_content_selector_nonce', 'nonce'); |
| 1756 |
|
| 1757 |
if (!current_user_can('manage_options')) { |
| 1758 |
wp_send_json_error(__('Unauthorized access', 'mxchat')); |
| 1759 |
} |
| 1760 |
|
| 1761 |
$page = isset($_GET['page']) ? absint($_GET['page']) : 1; |
| 1762 |
$per_page = isset($_GET['per_page']) ? absint($_GET['per_page']) : 20; |
| 1763 |
$search = isset($_GET['search']) ? sanitize_text_field($_GET['search']) : ''; |
| 1764 |
$post_type = isset($_GET['post_type']) ? sanitize_text_field($_GET['post_type']) : 'all'; |
| 1765 |
$post_status = isset($_GET['post_status']) ? sanitize_text_field($_GET['post_status']) : 'publish'; |
| 1766 |
$processed_filter = isset($_GET['processed_filter']) ? sanitize_text_field($_GET['processed_filter']) : 'all'; |
| 1767 |
|
| 1768 |
// Build query args |
| 1769 |
$args = array( |
| 1770 |
'posts_per_page' => $per_page, |
| 1771 |
'paged' => $page, |
| 1772 |
'post_status' => $post_status !== 'all' ? $post_status : array('publish', 'draft', 'pending'), |
| 1773 |
'orderby' => 'date', |
| 1774 |
'order' => 'DESC', |
| 1775 |
); |
| 1776 |
|
| 1777 |
// Handle post types |
| 1778 |
if ($post_type !== 'all') { |
| 1779 |
$args['post_type'] = $post_type; |
| 1780 |
} else { |
| 1781 |
// Default to post and page if we can't get post types |
| 1782 |
$args['post_type'] = array('post', 'page'); |
| 1783 |
|
| 1784 |
// Try to get public post types |
| 1785 |
$public_types = $this->get_public_post_types(); |
| 1786 |
if (is_array($public_types) && !empty($public_types)) { |
| 1787 |
$args['post_type'] = array_keys($public_types); |
| 1788 |
} |
| 1789 |
} |
| 1790 |
|
| 1791 |
if (!empty($search)) { |
| 1792 |
$args['s'] = $search; |
| 1793 |
} |
| 1794 |
|
| 1795 |
// ================================ |
| 1796 |
// UPDATED: Get already vectorized content from BOTH WordPress DB AND Pinecone |
| 1797 |
// ================================ |
| 1798 |
|
| 1799 |
// FIRST: Get from WordPress DB (your original working code) |
| 1800 |
global $wpdb; |
| 1801 |
$table_name = $wpdb->prefix . 'mxchat_system_prompt_content'; |
| 1802 |
$processed_items = $wpdb->get_results("SELECT id, source_url, timestamp FROM {$table_name}"); |
| 1803 |
|
| 1804 |
// Create a lookup array of processed posts with timestamps |
| 1805 |
$processed_data = array(); |
| 1806 |
|
| 1807 |
if (!empty($processed_items)) { |
| 1808 |
foreach ($processed_items as $item) { |
| 1809 |
$post_id = url_to_postid($item->source_url); |
| 1810 |
if ($post_id) { |
| 1811 |
$processed_data[$post_id] = array( |
| 1812 |
'db_id' => $item->id, |
| 1813 |
'timestamp' => $item->timestamp, |
| 1814 |
'url' => $item->source_url, |
| 1815 |
'source' => 'wordpress' |
| 1816 |
); |
| 1817 |
} |
| 1818 |
} |
| 1819 |
} |
| 1820 |
|
| 1821 |
// SECOND: ALSO check Pinecone if it's enabled |
| 1822 |
$pinecone_options = get_option('mxchat_pinecone_addon_options', array()); |
| 1823 |
$use_pinecone = ($pinecone_options['mxchat_use_pinecone'] ?? '0') === '1'; |
| 1824 |
|
| 1825 |
if ($use_pinecone && !empty($pinecone_options['mxchat_pinecone_api_key'])) { |
| 1826 |
$pinecone_data = $this->get_pinecone_processed_content($pinecone_options); |
| 1827 |
|
| 1828 |
// Merge Pinecone data with WordPress data |
| 1829 |
foreach ($pinecone_data as $post_id => $pinecone_info) { |
| 1830 |
// If not already found in WordPress DB, add from Pinecone |
| 1831 |
if (!isset($processed_data[$post_id])) { |
| 1832 |
$processed_data[$post_id] = $pinecone_info; |
| 1833 |
} |
| 1834 |
} |
| 1835 |
} |
| 1836 |
|
| 1837 |
// ================================ |
| 1838 |
|
| 1839 |
// Get processed IDs as a simple array for in_array checks |
| 1840 |
$processed_ids = array_keys($processed_data); |
| 1841 |
|
| 1842 |
// Handle processed/unprocessed filter |
| 1843 |
if ($processed_filter === 'processed' && !empty($processed_ids)) { |
| 1844 |
$args['post__in'] = $processed_ids; |
| 1845 |
} elseif ($processed_filter === 'unprocessed' && !empty($processed_ids)) { |
| 1846 |
$args['post__not_in'] = $processed_ids; |
| 1847 |
} |
| 1848 |
|
| 1849 |
// Run the query |
| 1850 |
$query = new WP_Query($args); |
| 1851 |
$content_items = array(); |
| 1852 |
|
| 1853 |
if ($query->have_posts()) { |
| 1854 |
while ($query->have_posts()) { |
| 1855 |
$query->the_post(); |
| 1856 |
$id = get_the_ID(); |
| 1857 |
$post_date = get_the_date(); |
| 1858 |
$excerpt = wp_trim_words(get_the_excerpt(), 20, '...'); |
| 1859 |
$word_count = str_word_count(strip_tags(get_the_content())); |
| 1860 |
|
| 1861 |
$is_processed = in_array($id, $processed_ids); |
| 1862 |
$processed_date = ''; |
| 1863 |
$db_record_id = 0; |
| 1864 |
$data_source = 'none'; |
| 1865 |
|
| 1866 |
if ($is_processed && isset($processed_data[$id])) { |
| 1867 |
$item_data = $processed_data[$id]; |
| 1868 |
$data_source = $item_data['source']; |
| 1869 |
|
| 1870 |
if ($data_source === 'wordpress' && isset($item_data['timestamp'])) { |
| 1871 |
// WordPress DB format |
| 1872 |
$timestamp = strtotime($item_data['timestamp']); |
| 1873 |
$processed_date = human_time_diff($timestamp, current_time('timestamp')) . ' ago'; |
| 1874 |
$db_record_id = $item_data['db_id']; |
| 1875 |
} elseif ($data_source === 'pinecone') { |
| 1876 |
// Pinecone format |
| 1877 |
$processed_date = $item_data['processed_date']; |
| 1878 |
$db_record_id = $item_data['db_id']; |
| 1879 |
} |
| 1880 |
} |
| 1881 |
|
| 1882 |
$content_items[] = array( |
| 1883 |
'id' => $id, |
| 1884 |
'title' => get_the_title(), |
| 1885 |
'permalink' => get_permalink(), |
| 1886 |
'date' => $post_date, |
| 1887 |
'type' => get_post_type(), |
| 1888 |
'status' => get_post_status(), |
| 1889 |
'excerpt' => $excerpt, |
| 1890 |
'word_count' => $word_count, |
| 1891 |
'already_processed' => $is_processed, |
| 1892 |
'processed_date' => $processed_date, |
| 1893 |
'db_record_id' => $db_record_id, |
| 1894 |
'data_source' => $data_source |
| 1895 |
); |
| 1896 |
} |
| 1897 |
wp_reset_postdata(); |
| 1898 |
} |
| 1899 |
|
| 1900 |
$response = array( |
| 1901 |
'items' => $content_items, |
| 1902 |
'total' => $query->found_posts, |
| 1903 |
'total_pages' => $query->max_num_pages, |
| 1904 |
'current_page' => $page, |
| 1905 |
'processed_count' => count($processed_ids) |
| 1906 |
); |
| 1907 |
|
| 1908 |
wp_send_json_success($response); |
| 1909 |
exit; |
| 1910 |
} |
| 1911 |
public function ajax_mxchat_process_selected_content() { |
| 1912 |
// Basic request validation |
| 1913 |
if (!check_ajax_referer('mxchat_content_selector_nonce', 'nonce', false)) { |
| 1914 |
wp_send_json_error('Invalid nonce'); |
| 1915 |
exit; |
| 1916 |
} |
| 1917 |
|
| 1918 |
if (!current_user_can('manage_options')) { |
| 1919 |
wp_send_json_error('Unauthorized access'); |
| 1920 |
exit; |
| 1921 |
} |
| 1922 |
|
| 1923 |
// Get post IDs - safely parse the array |
| 1924 |
$post_ids = array(); |
| 1925 |
if (isset($_POST['post_ids']) && is_array($_POST['post_ids'])) { |
| 1926 |
foreach ($_POST['post_ids'] as $id) { |
| 1927 |
$post_ids[] = absint($id); |
| 1928 |
} |
| 1929 |
} |
| 1930 |
|
| 1931 |
if (empty($post_ids)) { |
| 1932 |
wp_send_json_error('No content selected'); |
| 1933 |
exit; |
| 1934 |
} |
| 1935 |
|
| 1936 |
// Process only ONE post at a time to avoid request size issues |
| 1937 |
$post_id = reset($post_ids); |
| 1938 |
$post = get_post($post_id); |
| 1939 |
|
| 1940 |
if (!$post) { |
| 1941 |
wp_send_json_error('Post not found'); |
| 1942 |
exit; |
| 1943 |
} |
| 1944 |
|
| 1945 |
// Get minimal content |
| 1946 |
$content = $post->post_title . "\n\n" . wp_strip_all_tags($post->post_content); |
| 1947 |
$content = substr($content, 0, 10000); // Limit content size |
| 1948 |
|
| 1949 |
// Get API key with proper model detection |
| 1950 |
$options = get_option('mxchat_options'); |
| 1951 |
$selected_model = $options['embedding_model'] ?? 'text-embedding-ada-002'; |
| 1952 |
|
| 1953 |
if (strpos($selected_model, 'voyage') === 0) { |
| 1954 |
$api_key = $options['voyage_api_key'] ?? ''; |
| 1955 |
$provider_name = 'Voyage AI'; |
| 1956 |
} elseif (strpos($selected_model, 'gemini-embedding') === 0) { |
| 1957 |
$api_key = $options['gemini_api_key'] ?? ''; |
| 1958 |
$provider_name = 'Google Gemini'; |
| 1959 |
} else { |
| 1960 |
$api_key = $options['api_key'] ?? ''; |
| 1961 |
$provider_name = 'OpenAI'; |
| 1962 |
} |
| 1963 |
|
| 1964 |
if (empty($api_key)) { |
| 1965 |
wp_send_json_error($provider_name . ' API key not configured'); |
| 1966 |
exit; |
| 1967 |
} |
| 1968 |
|
| 1969 |
$source_url = get_permalink($post_id); |
| 1970 |
$vector_id = md5($source_url); // Vector ID for Pinecone |
| 1971 |
|
| 1972 |
// ================================ |
| 1973 |
// UPDATED: Check for existing content in BOTH sources for backwards compatibility |
| 1974 |
// ================================ |
| 1975 |
|
| 1976 |
$is_update = false; |
| 1977 |
|
| 1978 |
// Check WordPress DB first (backwards compatibility) |
| 1979 |
global $wpdb; |
| 1980 |
$table_name = $wpdb->prefix . 'mxchat_system_prompt_content'; |
| 1981 |
$existing_record = $wpdb->get_row($wpdb->prepare( |
| 1982 |
"SELECT id FROM $table_name WHERE source_url = %s", |
| 1983 |
$source_url |
| 1984 |
)); |
| 1985 |
|
| 1986 |
if ($existing_record) { |
| 1987 |
$is_update = true; |
| 1988 |
} else { |
| 1989 |
// Also check Pinecone if enabled |
| 1990 |
$pinecone_options = get_option('mxchat_pinecone_addon_options', array()); |
| 1991 |
$use_pinecone = ($pinecone_options['mxchat_use_pinecone'] ?? '0') === '1'; |
| 1992 |
|
| 1993 |
if ($use_pinecone && !empty($pinecone_options['mxchat_pinecone_api_key'])) { |
| 1994 |
$pinecone_data = $this->get_pinecone_processed_content($pinecone_options); |
| 1995 |
if (isset($pinecone_data[$post_id])) { |
| 1996 |
$is_update = true; |
| 1997 |
} |
| 1998 |
} |
| 1999 |
} |
| 2000 |
|
| 2001 |
// Use the centralized utility function for storage |
| 2002 |
$result = MxChat_Utils::submit_content_to_db( |
| 2003 |
$content, |
| 2004 |
$source_url, |
| 2005 |
$api_key, |
| 2006 |
$vector_id |
| 2007 |
); |
| 2008 |
|
| 2009 |
if (is_wp_error($result)) { |
| 2010 |
wp_send_json_error('Storage failed: ' . $result->get_error_message()); |
| 2011 |
exit; |
| 2012 |
} |
| 2013 |
|
| 2014 |
// ================================ |
| 2015 |
// NEW: Update caches immediately after successful storage |
| 2016 |
// ================================ |
| 2017 |
|
| 2018 |
// Check if Pinecone is enabled and update caches |
| 2019 |
$pinecone_options = get_option('mxchat_pinecone_addon_options', array()); |
| 2020 |
$use_pinecone = ($pinecone_options['mxchat_use_pinecone'] ?? '0') === '1'; |
| 2021 |
|
| 2022 |
if ($use_pinecone && !empty($pinecone_options['mxchat_pinecone_api_key'])) { |
| 2023 |
// Update vector ID cache for improved fetching |
| 2024 |
$this->update_pinecone_vector_cache($vector_id); |
| 2025 |
|
| 2026 |
// Update local processed content cache for immediate UI feedback |
| 2027 |
$pinecone_cache = get_option('mxchat_pinecone_processed_cache', array()); |
| 2028 |
$pinecone_cache[$post_id] = array( |
| 2029 |
'db_id' => $vector_id, |
| 2030 |
'processed_date' => 'Just now', |
| 2031 |
'url' => $source_url, |
| 2032 |
'source' => 'pinecone', |
| 2033 |
'timestamp' => current_time('timestamp') |
| 2034 |
); |
| 2035 |
update_option('mxchat_pinecone_processed_cache', $pinecone_cache); |
| 2036 |
|
| 2037 |
// Also update the general processed content cache |
| 2038 |
$processed_cache = get_option('mxchat_processed_content_cache', array()); |
| 2039 |
$processed_cache[$post_id] = array( |
| 2040 |
'db_id' => $vector_id, |
| 2041 |
'timestamp' => current_time('timestamp'), |
| 2042 |
'url' => $source_url, |
| 2043 |
'source' => 'pinecone' |
| 2044 |
); |
| 2045 |
update_option('mxchat_processed_content_cache', $processed_cache); |
| 2046 |
} |
| 2047 |
|
| 2048 |
$operation_type = $is_update ? 'update' : 'new'; |
| 2049 |
|
| 2050 |
// Success response with minimal data |
| 2051 |
wp_send_json_success(array( |
| 2052 |
'message' => $operation_type === 'update' ? 'Content updated successfully' : 'Content processed successfully', |
| 2053 |
'post_id' => $post_id, |
| 2054 |
'title' => $post->post_title, |
| 2055 |
'operation_type' => $operation_type, |
| 2056 |
'vector_id' => $vector_id, // Include vector ID for debugging |
| 2057 |
'cache_updated' => $use_pinecone // Indicate if cache was updated |
| 2058 |
)); |
| 2059 |
exit; |
| 2060 |
} |
| 2061 |
public function get_public_post_types() { |
| 2062 |
$post_types = get_post_types(array('public' => true), 'objects'); |
| 2063 |
$post_type_options = array(); |
| 2064 |
|
| 2065 |
foreach ($post_types as $post_type) { |
| 2066 |
$post_type_options[$post_type->name] = $post_type->label; |
| 2067 |
} |
| 2068 |
|
| 2069 |
return $post_type_options; |
| 2070 |
} |
| 2071 |
public function get_pinecone_processed_content($pinecone_options) { |
| 2072 |
// First check local cache for immediate updates |
| 2073 |
$cached_data = get_option('mxchat_pinecone_processed_cache', array()); |
| 2074 |
|
| 2075 |
$api_key = $pinecone_options['mxchat_pinecone_api_key'] ?? ''; |
| 2076 |
$host = $pinecone_options['mxchat_pinecone_host'] ?? ''; |
| 2077 |
|
| 2078 |
if (empty($api_key) || empty($host)) { |
| 2079 |
// Return only cached data if API credentials are missing |
| 2080 |
return $cached_data; |
| 2081 |
} |
| 2082 |
|
| 2083 |
$pinecone_data = array(); |
| 2084 |
|
| 2085 |
try { |
| 2086 |
// Method 1: Try to get vectors using cached vector IDs first |
| 2087 |
$cached_vector_ids = get_option('mxchat_pinecone_vector_ids_cache', array()); |
| 2088 |
|
| 2089 |
if (!empty($cached_vector_ids)) { |
| 2090 |
$pinecone_data = $this->fetch_pinecone_vectors_by_ids($pinecone_options, $cached_vector_ids); |
| 2091 |
} |
| 2092 |
|
| 2093 |
// Method 2: If no cached IDs or fetch failed, use scanning approach |
| 2094 |
if (empty($pinecone_data)) { |
| 2095 |
$pinecone_data = $this->scan_pinecone_for_processed_content($pinecone_options); |
| 2096 |
} |
| 2097 |
|
| 2098 |
// Method 3: Final fallback - try stats endpoint (if available) |
| 2099 |
if (empty($pinecone_data)) { |
| 2100 |
$stats_url = "https://{$host}/describe_index_stats"; |
| 2101 |
|
| 2102 |
$response = wp_remote_post($stats_url, array( |
| 2103 |
'headers' => array( |
| 2104 |
'Api-Key' => $api_key, |
| 2105 |
'Content-Type' => 'application/json' |
| 2106 |
), |
| 2107 |
'body' => json_encode(array()), |
| 2108 |
'timeout' => 30 |
| 2109 |
)); |
| 2110 |
|
| 2111 |
if (!is_wp_error($response) && wp_remote_retrieve_response_code($response) === 200) { |
| 2112 |
$body = wp_remote_retrieve_body($response); |
| 2113 |
$stats_data = json_decode($body, true); |
| 2114 |
|
| 2115 |
// Log stats for debugging but don't rely on them for vector listing |
| 2116 |
//error_log('Pinecone index stats: ' . print_r($stats_data, true)); |
| 2117 |
} |
| 2118 |
} |
| 2119 |
|
| 2120 |
} catch (Exception $e) { |
| 2121 |
//error_log('Pinecone processed content exception: ' . $e->getMessage()); |
| 2122 |
} |
| 2123 |
|
| 2124 |
// Merge cached data with Pinecone data |
| 2125 |
// Cache takes priority for recent updates (within last 5 minutes) |
| 2126 |
$merged_data = $pinecone_data; |
| 2127 |
|
| 2128 |
foreach ($cached_data as $post_id => $cache_item) { |
| 2129 |
$cache_timestamp = $cache_item['timestamp'] ?? 0; |
| 2130 |
$time_diff = current_time('timestamp') - $cache_timestamp; |
| 2131 |
|
| 2132 |
// If cache item is recent (less than 5 minutes), prioritize it |
| 2133 |
if ($time_diff < 300) { // 5 minutes = 300 seconds |
| 2134 |
$merged_data[$post_id] = $cache_item; |
| 2135 |
} else { |
| 2136 |
// If not in Pinecone data and cache is old, keep cache but mark as potentially stale |
| 2137 |
if (!isset($merged_data[$post_id])) { |
| 2138 |
$merged_data[$post_id] = $cache_item; |
| 2139 |
} |
| 2140 |
} |
| 2141 |
} |
| 2142 |
|
| 2143 |
return $merged_data; |
| 2144 |
} |
| 2145 |
public function fetch_pinecone_vectors_by_ids($pinecone_options, $vector_ids) { |
| 2146 |
$api_key = $pinecone_options['mxchat_pinecone_api_key'] ?? ''; |
| 2147 |
$host = $pinecone_options['mxchat_pinecone_host'] ?? ''; |
| 2148 |
|
| 2149 |
if (empty($api_key) || empty($host) || empty($vector_ids)) { |
| 2150 |
return array(); |
| 2151 |
} |
| 2152 |
|
| 2153 |
try { |
| 2154 |
$fetch_url = "https://{$host}/vectors/fetch"; |
| 2155 |
|
| 2156 |
// Pinecone fetch API allows fetching specific vectors by ID |
| 2157 |
$fetch_data = array( |
| 2158 |
'ids' => array_values($vector_ids) |
| 2159 |
); |
| 2160 |
|
| 2161 |
$response = wp_remote_post($fetch_url, array( |
| 2162 |
'headers' => array( |
| 2163 |
'Api-Key' => $api_key, |
| 2164 |
'Content-Type' => 'application/json' |
| 2165 |
), |
| 2166 |
'body' => json_encode($fetch_data), |
| 2167 |
'timeout' => 30 |
| 2168 |
)); |
| 2169 |
|
| 2170 |
if (is_wp_error($response)) { |
| 2171 |
//error_log('Pinecone fetch by IDs error: ' . $response->get_error_message()); |
| 2172 |
return array(); |
| 2173 |
} |
| 2174 |
|
| 2175 |
$response_code = wp_remote_retrieve_response_code($response); |
| 2176 |
if ($response_code !== 200) { |
| 2177 |
//error_log('Pinecone fetch by IDs failed with code: ' . $response_code); |
| 2178 |
return array(); |
| 2179 |
} |
| 2180 |
|
| 2181 |
$body = wp_remote_retrieve_body($response); |
| 2182 |
$data = json_decode($body, true); |
| 2183 |
|
| 2184 |
if (!isset($data['vectors'])) { |
| 2185 |
return array(); |
| 2186 |
} |
| 2187 |
|
| 2188 |
$processed_data = array(); |
| 2189 |
|
| 2190 |
foreach ($data['vectors'] as $vector_id => $vector_data) { |
| 2191 |
$metadata = $vector_data['metadata'] ?? array(); |
| 2192 |
$source_url = $metadata['source_url'] ?? ''; |
| 2193 |
|
| 2194 |
if (!empty($source_url)) { |
| 2195 |
$post_id = url_to_postid($source_url); |
| 2196 |
if ($post_id) { |
| 2197 |
$created_at = $metadata['created_at'] ?? ''; |
| 2198 |
$processed_date = 'Recently'; // Default |
| 2199 |
|
| 2200 |
if (!empty($created_at)) { |
| 2201 |
$timestamp = is_numeric($created_at) ? $created_at : strtotime($created_at); |
| 2202 |
if ($timestamp) { |
| 2203 |
$processed_date = human_time_diff($timestamp, current_time('timestamp')) . ' ago'; |
| 2204 |
} |
| 2205 |
} |
| 2206 |
|
| 2207 |
$processed_data[$post_id] = array( |
| 2208 |
'db_id' => $vector_id, |
| 2209 |
'processed_date' => $processed_date, |
| 2210 |
'url' => $source_url, |
| 2211 |
'source' => 'pinecone', |
| 2212 |
'timestamp' => $timestamp ?? current_time('timestamp') |
| 2213 |
); |
| 2214 |
} |
| 2215 |
} |
| 2216 |
} |
| 2217 |
|
| 2218 |
return $processed_data; |
| 2219 |
|
| 2220 |
} catch (Exception $e) { |
| 2221 |
//error_log('Pinecone fetch by IDs exception: ' . $e->getMessage()); |
| 2222 |
return array(); |
| 2223 |
} |
| 2224 |
} |
| 2225 |
public function scan_pinecone_for_processed_content($pinecone_options) { |
| 2226 |
$api_key = $pinecone_options['mxchat_pinecone_api_key'] ?? ''; |
| 2227 |
$host = $pinecone_options['mxchat_pinecone_host'] ?? ''; |
| 2228 |
|
| 2229 |
if (empty($api_key) || empty($host)) { |
| 2230 |
return array(); |
| 2231 |
} |
| 2232 |
|
| 2233 |
try { |
| 2234 |
// Use multiple random vectors to get better coverage |
| 2235 |
$all_matches = array(); |
| 2236 |
$seen_ids = array(); |
| 2237 |
|
| 2238 |
// Try 3 different random vectors to get better coverage |
| 2239 |
for ($i = 0; $i < 3; $i++) { |
| 2240 |
$query_url = "https://{$host}/query"; |
| 2241 |
|
| 2242 |
// Generate a random unit vector instead of zeros |
| 2243 |
$random_vector = array(); |
| 2244 |
for ($j = 0; $j < 1536; $j++) { |
| 2245 |
$random_vector[] = (rand(-1000, 1000) / 1000.0); // Random values between -1 and 1 |
| 2246 |
} |
| 2247 |
|
| 2248 |
// Normalize the vector to unit length |
| 2249 |
$magnitude = sqrt(array_sum(array_map(function($x) { return $x * $x; }, $random_vector))); |
| 2250 |
if ($magnitude > 0) { |
| 2251 |
$random_vector = array_map(function($x) use ($magnitude) { return $x / $magnitude; }, $random_vector); |
| 2252 |
} |
| 2253 |
|
| 2254 |
$query_data = array( |
| 2255 |
'includeMetadata' => true, |
| 2256 |
'includeValues' => false, |
| 2257 |
'topK' => 10000, // Get many results |
| 2258 |
'vector' => $random_vector |
| 2259 |
); |
| 2260 |
|
| 2261 |
$response = wp_remote_post($query_url, array( |
| 2262 |
'headers' => array( |
| 2263 |
'Api-Key' => $api_key, |
| 2264 |
'Content-Type' => 'application/json' |
| 2265 |
), |
| 2266 |
'body' => json_encode($query_data), |
| 2267 |
'timeout' => 30 |
| 2268 |
)); |
| 2269 |
|
| 2270 |
if (is_wp_error($response)) { |
| 2271 |
continue; |
| 2272 |
} |
| 2273 |
|
| 2274 |
$response_code = wp_remote_retrieve_response_code($response); |
| 2275 |
if ($response_code !== 200) { |
| 2276 |
continue; |
| 2277 |
} |
| 2278 |
|
| 2279 |
$body = wp_remote_retrieve_body($response); |
| 2280 |
$data = json_decode($body, true); |
| 2281 |
|
| 2282 |
if (isset($data['matches'])) { |
| 2283 |
foreach ($data['matches'] as $match) { |
| 2284 |
$match_id = $match['id'] ?? ''; |
| 2285 |
if (!empty($match_id) && !isset($seen_ids[$match_id])) { |
| 2286 |
$all_matches[] = $match; |
| 2287 |
$seen_ids[$match_id] = true; |
| 2288 |
} |
| 2289 |
} |
| 2290 |
} |
| 2291 |
} |
| 2292 |
|
| 2293 |
// Convert matches to processed data format |
| 2294 |
$processed_data = array(); |
| 2295 |
$vector_ids_for_cache = array(); |
| 2296 |
|
| 2297 |
foreach ($all_matches as $match) { |
| 2298 |
$metadata = $match['metadata'] ?? array(); |
| 2299 |
$source_url = $metadata['source_url'] ?? ''; |
| 2300 |
$match_id = $match['id'] ?? ''; |
| 2301 |
|
| 2302 |
if (!empty($source_url) && !empty($match_id)) { |
| 2303 |
$post_id = url_to_postid($source_url); |
| 2304 |
if ($post_id) { |
| 2305 |
$created_at = $metadata['created_at'] ?? ''; |
| 2306 |
$processed_date = 'Recently'; // Default |
| 2307 |
|
| 2308 |
if (!empty($created_at)) { |
| 2309 |
$timestamp = is_numeric($created_at) ? $created_at : strtotime($created_at); |
| 2310 |
if ($timestamp) { |
| 2311 |
$processed_date = human_time_diff($timestamp, current_time('timestamp')) . ' ago'; |
| 2312 |
} |
| 2313 |
} |
| 2314 |
|
| 2315 |
$processed_data[$post_id] = array( |
| 2316 |
'db_id' => $match_id, |
| 2317 |
'processed_date' => $processed_date, |
| 2318 |
'url' => $source_url, |
| 2319 |
'source' => 'pinecone', |
| 2320 |
'timestamp' => $timestamp ?? current_time('timestamp') |
| 2321 |
); |
| 2322 |
|
| 2323 |
$vector_ids_for_cache[] = $match_id; |
| 2324 |
} |
| 2325 |
} |
| 2326 |
} |
| 2327 |
|
| 2328 |
// Update the vector IDs cache for future use |
| 2329 |
if (!empty($vector_ids_for_cache)) { |
| 2330 |
update_option('mxchat_pinecone_vector_ids_cache', $vector_ids_for_cache); |
| 2331 |
} |
| 2332 |
|
| 2333 |
return $processed_data; |
| 2334 |
|
| 2335 |
} catch (Exception $e) { |
| 2336 |
//error_log('Pinecone scan exception: ' . $e->getMessage()); |
| 2337 |
return array(); |
| 2338 |
} |
| 2339 |
} |
| 2340 |
|
| 2341 |
/** |
| 2342 |
* Generates embeddings from input text for MXChat |
| 2343 |
*/ |
| 2344 |
private function mxchat_generate_embedding($text) { |
| 2345 |
// Enable detailed logging for debugging |
| 2346 |
//error_log('[MXCHAT-EMBED] Starting embedding generation. Text length: ' . strlen($text) . ' bytes'); |
| 2347 |
//error_log('[MXCHAT-EMBED] Text preview: ' . substr($text, 0, 100) . '...'); |
| 2348 |
|
| 2349 |
$options = get_option('mxchat_options'); |
| 2350 |
$selected_model = $options['embedding_model'] ?? 'text-embedding-ada-002'; |
| 2351 |
//error_log('[MXCHAT-EMBED] Selected embedding model: ' . $selected_model); |
| 2352 |
|
| 2353 |
// Determine provider and endpoint |
| 2354 |
if (strpos($selected_model, 'voyage') === 0) { |
| 2355 |
$api_key = $options['voyage_api_key'] ?? ''; |
| 2356 |
$endpoint = 'https://api.voyageai.com/v1/embeddings'; |
| 2357 |
$provider_name = 'Voyage AI'; |
| 2358 |
//error_log('[MXCHAT-EMBED] Using Voyage AI API'); |
| 2359 |
} elseif (strpos($selected_model, 'gemini-embedding') === 0) { |
| 2360 |
$api_key = $options['gemini_api_key'] ?? ''; |
| 2361 |
$endpoint = 'https://generativelanguage.googleapis.com/v1beta/models/' . $selected_model . ':embedContent'; |
| 2362 |
$provider_name = 'Google Gemini'; |
| 2363 |
//error_log('[MXCHAT-EMBED] Using Google Gemini API'); |
| 2364 |
} else { |
| 2365 |
$api_key = $options['api_key'] ?? ''; |
| 2366 |
$endpoint = 'https://api.openai.com/v1/embeddings'; |
| 2367 |
$provider_name = 'OpenAI'; |
| 2368 |
//error_log('[MXCHAT-EMBED] Using OpenAI API'); |
| 2369 |
} |
| 2370 |
|
| 2371 |
//error_log('[MXCHAT-EMBED] Using endpoint: ' . $endpoint); |
| 2372 |
|
| 2373 |
if (empty($api_key)) { |
| 2374 |
$error_message = sprintf('Missing %s API key. Please configure your API key in the MxChat settings.', $provider_name); |
| 2375 |
//error_log('[MXCHAT-EMBED] Error: ' . $error_message); |
| 2376 |
return $error_message; |
| 2377 |
} |
| 2378 |
|
| 2379 |
// Check if text is too long (for OpenAI, roughly estimate tokens as words/0.75) |
| 2380 |
$estimated_tokens = ceil(str_word_count($text) / 0.75); |
| 2381 |
//error_log('[MXCHAT-EMBED] Estimated token count: ~' . $estimated_tokens); |
| 2382 |
|
| 2383 |
if ($estimated_tokens > 8000 && strpos($selected_model, 'voyage') === false && strpos($selected_model, 'gemini-embedding') === false) { |
| 2384 |
//error_log('[MXCHAT-EMBED] Warning: Text may exceed OpenAI token limits (8K for most models)'); |
| 2385 |
// Consider truncating text here |
| 2386 |
} |
| 2387 |
|
| 2388 |
// Prepare request body based on provider |
| 2389 |
if (strpos($selected_model, 'gemini-embedding') === 0) { |
| 2390 |
// Gemini API format |
| 2391 |
$request_body = array( |
| 2392 |
'model' => 'models/' . $selected_model, |
| 2393 |
'content' => array( |
| 2394 |
'parts' => array( |
| 2395 |
array('text' => $text) |
| 2396 |
) |
| 2397 |
) |
| 2398 |
); |
| 2399 |
|
| 2400 |
// Set output dimensionality to 1536 for consistency with other models |
| 2401 |
$request_body['outputDimensionality'] = 1536; |
| 2402 |
} else { |
| 2403 |
// OpenAI/Voyage API format |
| 2404 |
$request_body = array( |
| 2405 |
'model' => $selected_model, |
| 2406 |
'input' => $text |
| 2407 |
); |
| 2408 |
|
| 2409 |
// Add output_dimension for voyage-3-large model |
| 2410 |
if ($selected_model === 'voyage-3-large') { |
| 2411 |
$request_body['output_dimension'] = 2048; |
| 2412 |
} |
| 2413 |
} |
| 2414 |
|
| 2415 |
//error_log('[MXCHAT-EMBED] Request prepared with model: ' . $selected_model); |
| 2416 |
|
| 2417 |
// Prepare headers based on provider |
| 2418 |
if (strpos($selected_model, 'gemini-embedding') === 0) { |
| 2419 |
// Gemini uses API key as query parameter |
| 2420 |
$endpoint .= '?key=' . $api_key; |
| 2421 |
$headers = array( |
| 2422 |
'Content-Type' => 'application/json' |
| 2423 |
); |
| 2424 |
} else { |
| 2425 |
// OpenAI/Voyage use Bearer token |
| 2426 |
$headers = array( |
| 2427 |
'Authorization' => 'Bearer ' . $api_key, |
| 2428 |
'Content-Type' => 'application/json' |
| 2429 |
); |
| 2430 |
} |
| 2431 |
|
| 2432 |
// Make API request |
| 2433 |
//error_log('[MXCHAT-EMBED] Sending API request to: ' . $endpoint); |
| 2434 |
$response = wp_remote_post($endpoint, array( |
| 2435 |
'body' => wp_json_encode($request_body), |
| 2436 |
'headers' => $headers, |
| 2437 |
'timeout' => 60 // Increased timeout for large inputs |
| 2438 |
)); |
| 2439 |
|
| 2440 |
// Handle wp_remote_post errors |
| 2441 |
if (is_wp_error($response)) { |
| 2442 |
$error_message = $response->get_error_message(); |
| 2443 |
//error_log('[MXCHAT-EMBED] WP Remote Post Error: ' . $error_message); |
| 2444 |
return 'Connection error: ' . $error_message; |
| 2445 |
} |
| 2446 |
|
| 2447 |
// Get and check HTTP response code |
| 2448 |
$http_code = wp_remote_retrieve_response_code($response); |
| 2449 |
//error_log('[MXCHAT-EMBED] API Response Code: ' . $http_code); |
| 2450 |
|
| 2451 |
if ($http_code !== 200) { |
| 2452 |
$error_body = wp_remote_retrieve_body($response); |
| 2453 |
//error_log('[MXCHAT-EMBED] API Error Response Body: ' . $error_body); |
| 2454 |
|
| 2455 |
// Try to parse error for more details |
| 2456 |
$error_json = json_decode($error_body, true); |
| 2457 |
if (json_last_error() === JSON_ERROR_NONE && isset($error_json['error'])) { |
| 2458 |
$error_type = $error_json['error']['type'] ?? 'unknown'; |
| 2459 |
$error_message = $error_json['error']['message'] ?? 'No message'; |
| 2460 |
//error_log('[MXCHAT-EMBED] API Error Type: ' . $error_type); |
| 2461 |
//error_log('[MXCHAT-EMBED] API Error Message: ' . $error_message); |
| 2462 |
|
| 2463 |
// Customize error message for common API errors |
| 2464 |
if ($error_type === 'invalid_request_error' && strpos($error_message, 'API key') !== false) { |
| 2465 |
$error_message = sprintf('Invalid %s API key. Please check your API key in the MxChat settings.', $provider_name); |
| 2466 |
} elseif ($error_type === 'authentication_error') { |
| 2467 |
$error_message = sprintf('%s authentication failed. Please verify your API key in the MxChat settings.', $provider_name); |
| 2468 |
} |
| 2469 |
|
| 2470 |
//error_log('[MXCHAT-EMBED] Returning error: ' . $error_message); |
| 2471 |
return $error_message; |
| 2472 |
} |
| 2473 |
|
| 2474 |
$error_message = sprintf("API Error (HTTP %d): Unable to generate embedding", $http_code); |
| 2475 |
//error_log('[MXCHAT-EMBED] Returning error: ' . $error_message); |
| 2476 |
return $error_message; |
| 2477 |
} |
| 2478 |
|
| 2479 |
// Parse response body |
| 2480 |
$response_body = wp_remote_retrieve_body($response); |
| 2481 |
//error_log('[MXCHAT-EMBED] Received response length: ' . strlen($response_body) . ' bytes'); |
| 2482 |
|
| 2483 |
$response_data = json_decode($response_body, true); |
| 2484 |
|
| 2485 |
if (json_last_error() !== JSON_ERROR_NONE) { |
| 2486 |
$error = json_last_error_msg(); |
| 2487 |
//error_log('[MXCHAT-EMBED] JSON Parse Error: ' . $error); |
| 2488 |
//error_log('[MXCHAT-EMBED] Response preview: ' . substr($response_body, 0, 200)); |
| 2489 |
return "Failed to parse API response: $error"; |
| 2490 |
} |
| 2491 |
|
| 2492 |
// Handle different response formats based on provider |
| 2493 |
if (strpos($selected_model, 'gemini-embedding') === 0) { |
| 2494 |
// Gemini API response format |
| 2495 |
if (isset($response_data['embedding']['values'])) { |
| 2496 |
$embedding_dimensions = count($response_data['embedding']['values']); |
| 2497 |
//error_log('[MXCHAT-EMBED] Successfully extracted Gemini embedding with ' . $embedding_dimensions . ' dimensions'); |
| 2498 |
|
| 2499 |
// Check if embedding dimensions are as expected (should be 1536) |
| 2500 |
if ($embedding_dimensions !== 1536) { |
| 2501 |
//error_log('[MXCHAT-EMBED] Warning: Unexpected Gemini embedding dimensions: ' . $embedding_dimensions); |
| 2502 |
} |
| 2503 |
|
| 2504 |
return $response_data['embedding']['values']; |
| 2505 |
} else { |
| 2506 |
//error_log('[MXCHAT-EMBED] Error: No embedding found in Gemini response'); |
| 2507 |
//error_log('[MXCHAT-EMBED] Response structure: ' . wp_json_encode(array_keys($response_data))); |
| 2508 |
|
| 2509 |
if (isset($response_data['error'])) { |
| 2510 |
$error_message = "Gemini API Error in response: " . wp_json_encode($response_data['error']); |
| 2511 |
//error_log('[MXCHAT-EMBED] ' . $error_message); |
| 2512 |
return $error_message; |
| 2513 |
} |
| 2514 |
|
| 2515 |
$error_message = "Invalid Gemini API response format: No embedding found"; |
| 2516 |
//error_log('[MXCHAT-EMBED] ' . $error_message); |
| 2517 |
return $error_message; |
| 2518 |
} |
| 2519 |
} else { |
| 2520 |
// OpenAI/Voyage API response format |
| 2521 |
if (isset($response_data['data'][0]['embedding'])) { |
| 2522 |
$embedding_dimensions = count($response_data['data'][0]['embedding']); |
| 2523 |
//error_log('[MXCHAT-EMBED] Successfully extracted embedding with ' . $embedding_dimensions . ' dimensions'); |
| 2524 |
|
| 2525 |
// Check if embedding dimensions are as expected |
| 2526 |
if (($selected_model === 'text-embedding-ada-002' && $embedding_dimensions !== 1536) || |
| 2527 |
($selected_model === 'voyage-3-large' && $embedding_dimensions !== 2048)) { |
| 2528 |
//error_log('[MXCHAT-EMBED] Warning: Unexpected embedding dimensions'); |
| 2529 |
} |
| 2530 |
|
| 2531 |
return $response_data['data'][0]['embedding']; |
| 2532 |
} else { |
| 2533 |
//error_log('[MXCHAT-EMBED] Error: No embedding found in response'); |
| 2534 |
//error_log('[MXCHAT-EMBED] Response structure: ' . wp_json_encode(array_keys($response_data))); |
| 2535 |
|
| 2536 |
if (isset($response_data['error'])) { |
| 2537 |
$error_message = "API Error in response: " . wp_json_encode($response_data['error']); |
| 2538 |
//error_log('[MXCHAT-EMBED] ' . $error_message); |
| 2539 |
return $error_message; |
| 2540 |
} |
| 2541 |
|
| 2542 |
$error_message = "Invalid API response format: No embedding found"; |
| 2543 |
//error_log('[MXCHAT-EMBED] ' . $error_message); |
| 2544 |
return $error_message; |
| 2545 |
} |
| 2546 |
} |
| 2547 |
} |
| 2548 |
public function ajax_dismiss_completed_status() { |
| 2549 |
try { |
| 2550 |
// Verify the request |
| 2551 |
check_ajax_referer('mxchat_status_nonce', 'nonce'); |
| 2552 |
|
| 2553 |
if (!current_user_can('manage_options')) { |
| 2554 |
wp_send_json_error('Unauthorized access'); |
| 2555 |
exit; |
| 2556 |
} |
| 2557 |
|
| 2558 |
$card_type = isset($_POST['card_type']) ? sanitize_text_field($_POST['card_type']) : ''; |
| 2559 |
|
| 2560 |
if ($card_type === 'pdf') { |
| 2561 |
// Clear PDF status |
| 2562 |
$pdf_url = get_transient('mxchat_last_pdf_url'); |
| 2563 |
if ($pdf_url) { |
| 2564 |
delete_transient('mxchat_pdf_status_' . md5($pdf_url)); |
| 2565 |
delete_transient('mxchat_last_pdf_url'); |
| 2566 |
} |
| 2567 |
} elseif ($card_type === 'sitemap') { |
| 2568 |
// Clear sitemap status |
| 2569 |
$sitemap_url = get_transient('mxchat_last_sitemap_url'); |
| 2570 |
if ($sitemap_url) { |
| 2571 |
delete_transient('mxchat_sitemap_status_' . md5($sitemap_url)); |
| 2572 |
delete_transient('mxchat_last_sitemap_url'); |
| 2573 |
} |
| 2574 |
} |
| 2575 |
|
| 2576 |
wp_send_json_success(array('message' => 'Status dismissed successfully')); |
| 2577 |
|
| 2578 |
} catch (Exception $e) { |
| 2579 |
wp_send_json_error(array('message' => 'Error dismissing status: ' . $e->getMessage())); |
| 2580 |
} |
| 2581 |
} |
| 2582 |
|
| 2583 |
/** |
| 2584 |
* Render completed status cards on page load |
| 2585 |
* This ensures completed processing status persists through page refreshes |
| 2586 |
*/ |
| 2587 |
public function render_completed_status_cards() { |
| 2588 |
$output = ''; |
| 2589 |
|
| 2590 |
// Check for completed PDF status |
| 2591 |
$pdf_url = get_transient('mxchat_last_pdf_url'); |
| 2592 |
if ($pdf_url) { |
| 2593 |
$pdf_status = $this->get_pdf_processing_status($pdf_url); |
| 2594 |
if ($pdf_status && ($pdf_status['status'] === 'complete' || $pdf_status['status'] === 'error')) { |
| 2595 |
$output .= $this->render_pdf_status_card($pdf_status, $pdf_url); |
| 2596 |
} |
| 2597 |
} |
| 2598 |
|
| 2599 |
// Check for completed sitemap status |
| 2600 |
$sitemap_url = get_transient('mxchat_last_sitemap_url'); |
| 2601 |
if ($sitemap_url) { |
| 2602 |
$sitemap_status = $this->get_sitemap_processing_status($sitemap_url); |
| 2603 |
if ($sitemap_status && ($sitemap_status['status'] === 'complete' || $sitemap_status['status'] === 'error')) { |
| 2604 |
$output .= $this->render_sitemap_status_card($sitemap_status, $sitemap_url); |
| 2605 |
} |
| 2606 |
} |
| 2607 |
|
| 2608 |
return $output; |
| 2609 |
} |
| 2610 |
|
| 2611 |
/** |
| 2612 |
* Render PDF status card HTML |
| 2613 |
*/ |
| 2614 |
private function render_pdf_status_card($status, $pdf_url) { |
| 2615 |
$html = '<div class="mxchat-status-card" data-card-type="pdf">'; |
| 2616 |
$html .= '<div class="mxchat-status-header">'; |
| 2617 |
$html .= '<h4>' . esc_html__('PDF Processing Status', 'mxchat') . '</h4>'; |
| 2618 |
|
| 2619 |
// Add dismiss button for completed status |
| 2620 |
if ($status['status'] === 'complete' || $status['status'] === 'error') { |
| 2621 |
$html .= '<button type="button" class="mxchat-dismiss-button">' . esc_html__('Dismiss', 'mxchat') . '</button>'; |
| 2622 |
} |
| 2623 |
|
| 2624 |
// Process Batch button for processing status |
| 2625 |
if ($status['status'] === 'processing') { |
| 2626 |
$html .= '<button type="button" class="mxchat-manual-batch-btn" |
| 2627 |
data-process-type="pdf" |
| 2628 |
data-url="' . esc_attr($pdf_url) . '"> |
| 2629 |
' . esc_html__('Process Batch', 'mxchat') . '</button>'; |
| 2630 |
} |
| 2631 |
|
| 2632 |
// Add status badges |
| 2633 |
if ($status['status'] === 'error') { |
| 2634 |
$html .= '<span class="mxchat-status-badge mxchat-status-failed">' . esc_html__('Error', 'mxchat') . '</span>'; |
| 2635 |
} elseif ($status['status'] === 'complete') { |
| 2636 |
if ($status['failed_pages'] > 0) { |
| 2637 |
$html .= '<span class="mxchat-status-badge mxchat-status-warning">' . |
| 2638 |
sprintf(esc_html__('Completed with %d failures', 'mxchat'), $status['failed_pages']) . '</span>'; |
| 2639 |
} else { |
| 2640 |
$html .= '<span class="mxchat-status-badge mxchat-status-success">' . esc_html__('Complete', 'mxchat') . '</span>'; |
| 2641 |
} |
| 2642 |
} |
| 2643 |
|
| 2644 |
$html .= '</div>'; // End header |
| 2645 |
|
| 2646 |
// Progress bar |
| 2647 |
$html .= '<div class="mxchat-progress-bar">'; |
| 2648 |
$html .= '<div class="mxchat-progress-fill" style="width: ' . esc_attr($status['percentage']) . '%"></div>'; |
| 2649 |
$html .= '</div>'; |
| 2650 |
|
| 2651 |
// Status details |
| 2652 |
$html .= '<div class="mxchat-status-details">'; |
| 2653 |
$html .= '<p>' . sprintf( |
| 2654 |
esc_html__('Progress: %d of %d pages (%d%%)', 'mxchat'), |
| 2655 |
$status['processed_pages'], |
| 2656 |
$status['total_pages'], |
| 2657 |
$status['percentage'] |
| 2658 |
) . '</p>'; |
| 2659 |
|
| 2660 |
// Show failed pages count if any |
| 2661 |
if ($status['failed_pages'] > 0) { |
| 2662 |
$html .= '<p><strong>' . esc_html__('Failed pages:', 'mxchat') . '</strong> ' . esc_html($status['failed_pages']) . '</p>'; |
| 2663 |
} |
| 2664 |
|
| 2665 |
$html .= '<p><strong>' . esc_html__('Status:', 'mxchat') . '</strong> ' . esc_html(ucfirst($status['status'])) . '</p>'; |
| 2666 |
$html .= '<p><strong>' . esc_html__('Last update:', 'mxchat') . '</strong> ' . esc_html($status['last_update']) . '</p>'; |
| 2667 |
|
| 2668 |
// Add completion summary if available |
| 2669 |
if (isset($status['completion_summary'])) { |
| 2670 |
$summary = $status['completion_summary']; |
| 2671 |
$html .= '<div class="mxchat-completion-summary">'; |
| 2672 |
$html .= '<h5>' . esc_html__('Processing Summary', 'mxchat') . '</h5>'; |
| 2673 |
$html .= '<p><strong>' . esc_html__('Total Pages:', 'mxchat') . '</strong> ' . esc_html($summary['total_pages']) . '</p>'; |
| 2674 |
$html .= '<p><strong>' . esc_html__('Successful:', 'mxchat') . '</strong> ' . esc_html($summary['successful_pages']) . '</p>'; |
| 2675 |
$html .= '<p><strong>' . esc_html__('Failed:', 'mxchat') . '</strong> ' . esc_html($summary['failed_pages']) . '</p>'; |
| 2676 |
$html .= '<p><strong>' . esc_html__('Completed:', 'mxchat') . '</strong> ' . esc_html($summary['completion_time']) . '</p>'; |
| 2677 |
$html .= '</div>'; |
| 2678 |
} |
| 2679 |
|
| 2680 |
// Add failed pages list if any |
| 2681 |
if (!empty($status['failed_pages_list'])) { |
| 2682 |
$html .= $this->render_failed_pages_list($status['failed_pages_list']); |
| 2683 |
} |
| 2684 |
|
| 2685 |
// Add error message if any |
| 2686 |
if (isset($status['error']) && !empty($status['error'])) { |
| 2687 |
$html .= '<div class="mxchat-error-notice">'; |
| 2688 |
$html .= '<p class="error">' . esc_html($status['error']) . '</p>'; |
| 2689 |
$html .= '</div>'; |
| 2690 |
} |
| 2691 |
|
| 2692 |
$html .= '</div>'; // End details |
| 2693 |
$html .= '</div>'; // End card |
| 2694 |
|
| 2695 |
return $html; |
| 2696 |
} |
| 2697 |
|
| 2698 |
/** |
| 2699 |
* Render sitemap status card HTML |
| 2700 |
*/ |
| 2701 |
private function render_sitemap_status_card($status, $sitemap_url) { |
| 2702 |
$html = '<div class="mxchat-status-card" data-card-type="sitemap">'; |
| 2703 |
$html .= '<div class="mxchat-status-header">'; |
| 2704 |
$html .= '<h4>' . esc_html__('Sitemap Processing Status', 'mxchat') . '</h4>'; |
| 2705 |
|
| 2706 |
// Add dismiss button for completed status |
| 2707 |
if ($status['status'] === 'complete' || $status['status'] === 'error') { |
| 2708 |
$html .= '<button type="button" class="mxchat-dismiss-button">' . esc_html__('Dismiss', 'mxchat') . '</button>'; |
| 2709 |
} |
| 2710 |
|
| 2711 |
// Process Batch button for processing status |
| 2712 |
if ($status['status'] === 'processing') { |
| 2713 |
$html .= '<button type="button" class="mxchat-manual-batch-btn" |
| 2714 |
data-process-type="sitemap" |
| 2715 |
data-url="' . esc_attr($sitemap_url) . '"> |
| 2716 |
' . esc_html__('Process Batch', 'mxchat') . '</button>'; |
| 2717 |
} |
| 2718 |
|
| 2719 |
// Add status badges |
| 2720 |
if ($status['status'] === 'error') { |
| 2721 |
$html .= '<span class="mxchat-status-badge mxchat-status-failed">' . esc_html__('Error', 'mxchat') . '</span>'; |
| 2722 |
} elseif ($status['status'] === 'complete') { |
| 2723 |
if ($status['failed_urls'] > 0) { |
| 2724 |
$html .= '<span class="mxchat-status-badge mxchat-status-warning">' . |
| 2725 |
sprintf(esc_html__('Completed with %d failures', 'mxchat'), $status['failed_urls']) . '</span>'; |
| 2726 |
} else { |
| 2727 |
$html .= '<span class="mxchat-status-badge mxchat-status-success">' . esc_html__('Complete', 'mxchat') . '</span>'; |
| 2728 |
} |
| 2729 |
} |
| 2730 |
|
| 2731 |
$html .= '</div>'; // End header |
| 2732 |
|
| 2733 |
// Progress bar |
| 2734 |
$html .= '<div class="mxchat-progress-bar">'; |
| 2735 |
$html .= '<div class="mxchat-progress-fill" style="width: ' . esc_attr($status['percentage']) . '%"></div>'; |
| 2736 |
$html .= '</div>'; |
| 2737 |
|
| 2738 |
// Status details |
| 2739 |
$html .= '<div class="mxchat-status-details">'; |
| 2740 |
$html .= '<p>' . sprintf( |
| 2741 |
esc_html__('Progress: %d of %d URLs (%d%%)', 'mxchat'), |
| 2742 |
$status['processed_urls'], |
| 2743 |
$status['total_urls'], |
| 2744 |
$status['percentage'] |
| 2745 |
) . '</p>'; |
| 2746 |
|
| 2747 |
// Show failed URLs count if any |
| 2748 |
if ($status['failed_urls'] > 0) { |
| 2749 |
$html .= '<p><strong>' . esc_html__('Failed URLs:', 'mxchat') . '</strong> ' . esc_html($status['failed_urls']) . '</p>'; |
| 2750 |
} |
| 2751 |
|
| 2752 |
$html .= '<p><strong>' . esc_html__('Status:', 'mxchat') . '</strong> ' . esc_html(ucfirst($status['status'])) . '</p>'; |
| 2753 |
$html .= '<p><strong>' . esc_html__('Last update:', 'mxchat') . '</strong> ' . esc_html($status['last_update']) . '</p>'; |
| 2754 |
|
| 2755 |
// Add completion summary if available |
| 2756 |
if (isset($status['completion_summary'])) { |
| 2757 |
$summary = $status['completion_summary']; |
| 2758 |
$html .= '<div class="mxchat-completion-summary">'; |
| 2759 |
$html .= '<h5>' . esc_html__('Processing Summary', 'mxchat') . '</h5>'; |
| 2760 |
$html .= '<p><strong>' . esc_html__('Total URLs:', 'mxchat') . '</strong> ' . esc_html($summary['total_urls']) . '</p>'; |
| 2761 |
$html .= '<p><strong>' . esc_html__('Successful:', 'mxchat') . '</strong> ' . esc_html($summary['successful_urls']) . '</p>'; |
| 2762 |
$html .= '<p><strong>' . esc_html__('Failed:', 'mxchat') . '</strong> ' . esc_html($summary['failed_urls']) . '</p>'; |
| 2763 |
$html .= '<p><strong>' . esc_html__('Completed:', 'mxchat') . '</strong> ' . esc_html($summary['completion_time']) . '</p>'; |
| 2764 |
$html .= '</div>'; |
| 2765 |
} |
| 2766 |
|
| 2767 |
// Add error messages if any (but not the failed URLs list) |
| 2768 |
if (!empty($status['error']) || !empty($status['last_error'])) { |
| 2769 |
$html .= '<div class="mxchat-error-notice">'; |
| 2770 |
|
| 2771 |
if (!empty($status['error'])) { |
| 2772 |
$html .= '<p class="error">' . esc_html($status['error']) . '</p>'; |
| 2773 |
} |
| 2774 |
|
| 2775 |
if (!empty($status['last_error'])) { |
| 2776 |
$html .= '<p class="last-error">' . esc_html__('Last error:', 'mxchat') . ' ' . esc_html($status['last_error']) . '</p>'; |
| 2777 |
} |
| 2778 |
|
| 2779 |
$html .= '</div>'; |
| 2780 |
} |
| 2781 |
|
| 2782 |
$html .= '</div>'; // End details |
| 2783 |
$html .= '</div>'; // End card |
| 2784 |
|
| 2785 |
return $html; |
| 2786 |
} |
| 2787 |
/** |
| 2788 |
* Render failed pages list |
| 2789 |
*/ |
| 2790 |
private function render_failed_pages_list($failed_pages_list) { |
| 2791 |
if (empty($failed_pages_list)) { |
| 2792 |
return ''; |
| 2793 |
} |
| 2794 |
|
| 2795 |
$html = '<div class="mxchat-error-notice">'; |
| 2796 |
$html .= '<div class="mxchat-failed-pages-container">'; |
| 2797 |
$html .= '<h5>' . sprintf(esc_html__('Failed Pages (%d)', 'mxchat'), count($failed_pages_list)) . '</h5>'; |
| 2798 |
$html .= '<details>'; |
| 2799 |
$html .= '<summary>' . esc_html__('Show Failed Pages', 'mxchat') . '</summary>'; |
| 2800 |
$html .= '<div class="mxchat-failed-pages-list">'; |
| 2801 |
|
| 2802 |
// Create table for failed pages |
| 2803 |
$html .= '<table class="widefat striped">'; |
| 2804 |
$html .= '<thead><tr>'; |
| 2805 |
$html .= '<th>' . esc_html__('Page', 'mxchat') . '</th>'; |
| 2806 |
$html .= '<th>' . esc_html__('Error', 'mxchat') . '</th>'; |
| 2807 |
$html .= '<th>' . esc_html__('Retries', 'mxchat') . '</th>'; |
| 2808 |
$html .= '<th>' . esc_html__('Time', 'mxchat') . '</th>'; |
| 2809 |
$html .= '</tr></thead><tbody>'; |
| 2810 |
|
| 2811 |
// Sort failed pages by most recent |
| 2812 |
$sorted_failed_pages = $failed_pages_list; |
| 2813 |
usort($sorted_failed_pages, function($a, $b) { |
| 2814 |
return ($b['time'] ?? 0) - ($a['time'] ?? 0); |
| 2815 |
}); |
| 2816 |
|
| 2817 |
foreach ($sorted_failed_pages as $item) { |
| 2818 |
$time_ago = isset($item['time']) ? human_time_diff($item['time'], current_time('timestamp')) . ' ' . esc_html__('ago', 'mxchat') : 'Unknown'; |
| 2819 |
$html .= '<tr>'; |
| 2820 |
$html .= '<td>' . esc_html__('Page', 'mxchat') . ' ' . esc_html($item['page'] ?? 'Unknown') . '</td>'; |
| 2821 |
$html .= '<td style="word-break: break-word;">' . esc_html($item['error'] ?? 'Unknown error') . '</td>'; |
| 2822 |
$html .= '<td>' . esc_html($item['retries'] ?? 'N/A') . '</td>'; |
| 2823 |
$html .= '<td>' . esc_html($time_ago) . '</td>'; |
| 2824 |
$html .= '</tr>'; |
| 2825 |
} |
| 2826 |
|
| 2827 |
$html .= '</tbody></table>'; |
| 2828 |
$html .= '</div></details></div></div>'; |
| 2829 |
|
| 2830 |
return $html; |
| 2831 |
} |
| 2832 |
|
| 2833 |
/** |
| 2834 |
* Render failed URLs list |
| 2835 |
*/ |
| 2836 |
private function render_failed_urls_list($failed_urls_list) { |
| 2837 |
if (empty($failed_urls_list)) { |
| 2838 |
return ''; |
| 2839 |
} |
| 2840 |
|
| 2841 |
$html = '<div class="mxchat-failed-urls-container">'; |
| 2842 |
$html .= '<h5>' . sprintf(esc_html__('Failed URLs (%d)', 'mxchat'), count($failed_urls_list)) . '</h5>'; |
| 2843 |
$html .= '<details>'; |
| 2844 |
$html .= '<summary>' . esc_html__('Show Failed URLs', 'mxchat') . '</summary>'; |
| 2845 |
$html .= '<div class="mxchat-failed-urls-list">'; |
| 2846 |
|
| 2847 |
// Create table for failed URLs |
| 2848 |
$html .= '<table class="widefat striped">'; |
| 2849 |
$html .= '<thead><tr>'; |
| 2850 |
$html .= '<th>' . esc_html__('URL', 'mxchat') . '</th>'; |
| 2851 |
$html .= '<th>' . esc_html__('Error', 'mxchat') . '</th>'; |
| 2852 |
$html .= '<th>' . esc_html__('Retries', 'mxchat') . '</th>'; |
| 2853 |
$html .= '<th>' . esc_html__('Time', 'mxchat') . '</th>'; |
| 2854 |
$html .= '</tr></thead><tbody>'; |
| 2855 |
|
| 2856 |
// Sort failed URLs by most recent |
| 2857 |
$sorted_failed_urls = $failed_urls_list; |
| 2858 |
usort($sorted_failed_urls, function($a, $b) { |
| 2859 |
return ($b['time'] ?? 0) - ($a['time'] ?? 0); |
| 2860 |
}); |
| 2861 |
|
| 2862 |
// Show up to 50 failed URLs |
| 2863 |
$display_urls = array_slice($sorted_failed_urls, 0, 50); |
| 2864 |
|
| 2865 |
foreach ($display_urls as $item) { |
| 2866 |
$url = $item['url'] ?? ''; |
| 2867 |
$time_ago = isset($item['time']) ? human_time_diff($item['time'], current_time('timestamp')) . ' ' . esc_html__('ago', 'mxchat') : 'Unknown'; |
| 2868 |
|
| 2869 |
// Truncate URL for display |
| 2870 |
$display_url = strlen($url) > 50 ? substr($url, 0, 47) . '...' : $url; |
| 2871 |
|
| 2872 |
$html .= '<tr>'; |
| 2873 |
$html .= '<td style="word-break: break-all;">'; |
| 2874 |
if (!empty($url)) { |
| 2875 |
$html .= '<a href="' . esc_url($url) . '" target="_blank" rel="noopener noreferrer">' . esc_html($display_url) . '</a>'; |
| 2876 |
} else { |
| 2877 |
$html .= esc_html__('Unknown URL', 'mxchat'); |
| 2878 |
} |
| 2879 |
$html .= '</td>'; |
| 2880 |
$html .= '<td style="word-break: break-word;">' . esc_html($item['error'] ?? 'Unknown error') . '</td>'; |
| 2881 |
$html .= '<td>' . esc_html($item['retries'] ?? 'N/A') . '</td>'; |
| 2882 |
$html .= '<td>' . esc_html($time_ago) . '</td>'; |
| 2883 |
$html .= '</tr>'; |
| 2884 |
} |
| 2885 |
|
| 2886 |
$html .= '</tbody></table>'; |
| 2887 |
|
| 2888 |
if (count($failed_urls_list) > 50) { |
| 2889 |
$html .= '<div class="mxchat-failed-urls-more">+ ' . |
| 2890 |
(count($failed_urls_list) - 50) . |
| 2891 |
' ' . esc_html__('more failed URLs not shown', 'mxchat') . '</div>'; |
| 2892 |
} |
| 2893 |
|
| 2894 |
$html .= '</div></details></div>'; |
| 2895 |
|
| 2896 |
return $html; |
| 2897 |
} |
| 2898 |
|
| 2899 |
|
| 2900 |
// ======================================== |
| 2901 |
// HELPER METHODS |
| 2902 |
// ======================================== |
| 2903 |
|
| 2904 |
/** |
| 2905 |
* Check if user has required permissions for content processing |
| 2906 |
*/ |
| 2907 |
private function check_user_permissions() { |
| 2908 |
if (!current_user_can('manage_options')) { |
| 2909 |
wp_die(esc_html__('You do not have sufficient permissions.', 'mxchat')); |
| 2910 |
} |
| 2911 |
} |
| 2912 |
|
| 2913 |
/** |
| 2914 |
* Validate nonce for security |
| 2915 |
*/ |
| 2916 |
private function validate_nonce($nonce_name, $nonce_action) { |
| 2917 |
if (!isset($_POST[$nonce_name]) || !wp_verify_nonce($_POST[$nonce_name], $nonce_action)) { |
| 2918 |
wp_die(esc_html__('Security check failed.', 'mxchat')); |
| 2919 |
} |
| 2920 |
} |
| 2921 |
|
| 2922 |
/** |
| 2923 |
* Get embedding API credentials |
| 2924 |
*/ |
| 2925 |
private function get_embedding_credentials() { |
| 2926 |
$embedding_model = $this->options['embedding_model'] ?? 'text-embedding-ada-002'; |
| 2927 |
|
| 2928 |
if (strpos($embedding_model, 'text-embedding-') !== false) { |
| 2929 |
return array( |
| 2930 |
'type' => 'openai', |
| 2931 |
'api_key' => $this->options['api_key'] ?? '' |
| 2932 |
); |
| 2933 |
} elseif (strpos($embedding_model, 'voyage-') !== false) { |
| 2934 |
return array( |
| 2935 |
'type' => 'voyage', |
| 2936 |
'api_key' => $this->options['voyage_api_key'] ?? '' |
| 2937 |
); |
| 2938 |
} elseif (strpos($embedding_model, 'gemini-embedding-') !== false) { |
| 2939 |
return array( |
| 2940 |
'type' => 'gemini', |
| 2941 |
'api_key' => $this->options['gemini_api_key'] ?? '' |
| 2942 |
); |
| 2943 |
} |
| 2944 |
|
| 2945 |
return array('type' => 'unknown', 'api_key' => ''); |
| 2946 |
} |
| 2947 |
|
| 2948 |
/** |
| 2949 |
* Log processing errors |
| 2950 |
*/ |
| 2951 |
private function log_processing_error($operation, $error_message) { |
| 2952 |
//error_log("MxChat Knowledge Processing {$operation} Error: " . $error_message); |
| 2953 |
} |
| 2954 |
|
| 2955 |
/** |
| 2956 |
* Set admin notice transient |
| 2957 |
*/ |
| 2958 |
private function set_admin_notice($type, $message) { |
| 2959 |
set_transient("mxchat_admin_notice_{$type}", $message, 30); |
| 2960 |
} |
| 2961 |
|
| 2962 |
/** |
| 2963 |
* Get Pinecone manager instance for vector operations |
| 2964 |
*/ |
| 2965 |
private function get_pinecone_manager() { |
| 2966 |
return MxChat_Pinecone_Manager::get_instance(); |
| 2967 |
} |
| 2968 |
|
| 2969 |
// ======================================== |
| 2970 |
// STATIC ACCESS METHODS |
| 2971 |
// ======================================== |
| 2972 |
|
| 2973 |
/** |
| 2974 |
* Get singleton instance |
| 2975 |
*/ |
| 2976 |
public static function get_instance() { |
| 2977 |
static $instance = null; |
| 2978 |
if ($instance === null) { |
| 2979 |
$instance = new self(); |
| 2980 |
} |
| 2981 |
return $instance; |
| 2982 |
} |
| 2983 |
} |
| 2984 |
|
| 2985 |
// Initialize the Knowledge manager |
| 2986 |
$mxchat_knowledge_manager = MxChat_Knowledge_Manager::get_instance(); |
