| 1 |
<?php |
| 2 |
/** |
| 3 |
* File: admin/class-knowledge-manager.php |
| 4 |
* |
| 5 |
* Handles all knowledge base content processing for MxChat |
| 6 |
* Including PDF, sitemap, content processing, and WordPress post management |
| 7 |
*/ |
| 8 |
if (!defined('ABSPATH')) { |
| 9 |
exit; // Exit if accessed directly |
| 10 |
} |
| 11 |
|
| 12 |
class MxChat_Knowledge_Manager { |
| 13 |
|
| 14 |
private $options; |
| 15 |
|
| 16 |
// Post IDs whose vectors were already deleted by mxchat_handle_status_transition this |
| 17 |
// request, so the transient-based branch in mxchat_handle_post_update can skip the |
| 18 |
// redundant (idempotent but network-visible) second deletion. |
| 19 |
private $transition_deleted_posts = array(); |
| 20 |
|
| 21 |
// Post IDs already INDEXED by mxchat_handle_status_transition's arrival edge this |
| 22 |
// request. Normal editor publishes fire transition_post_status first, then |
| 23 |
// post_updated — without this guard every editor publish would embed twice. |
| 24 |
private $transition_indexed_posts = array(); |
| 25 |
|
| 26 |
// Post IDs core has announced an in-flight UPDATE for. pre_post_update fires only |
| 27 |
// inside wp_insert_post's update branch and always before wp_transition_post_status, |
| 28 |
// so this is an exact "a post_updated is coming later this request" signal — which is |
| 29 |
// what makes it safe to arm transition_indexed_posts (plan a664f3). |
| 30 |
private $pending_post_update = array(); |
| 31 |
|
| 32 |
/** |
| 33 |
* Constructor - Register hooks for content processing |
| 34 |
*/ |
| 35 |
public function __construct() { |
| 36 |
$this->options = get_option('mxchat_options', array()); |
| 37 |
$this->mxchat_init_hooks(); |
| 38 |
|
| 39 |
$this->mxchat_init_role_hooks(); |
| 40 |
} |
| 41 |
|
| 42 |
/** |
| 43 |
* Initialize WordPress hooks for content processing |
| 44 |
* |
| 45 |
*/ |
| 46 |
private function mxchat_init_hooks() { |
| 47 |
// Admin post handlers for form submissions |
| 48 |
add_action('admin_post_mxchat_submit_content', array($this, 'mxchat_handle_content_submission')); |
| 49 |
add_action('admin_post_mxchat_submit_sitemap', array($this, 'mxchat_handle_sitemap_submission')); |
| 50 |
add_action('admin_post_mxchat_submit_pdf_file', array($this, 'mxchat_handle_pdf_file_submission')); |
| 51 |
add_action('admin_post_mxchat_submit_document_file', array($this, 'mxchat_handle_document_file_submission')); |
| 52 |
add_action('admin_post_mxchat_submit_youtube', array($this, 'mxchat_handle_youtube_submission')); |
| 53 |
add_action('admin_post_mxchat_stop_processing', array($this, 'mxchat_stop_processing')); |
| 54 |
|
| 55 |
// AJAX handlers for real-time processing and status updates |
| 56 |
add_action('wp_ajax_mxchat_get_status_updates', array($this, 'mxchat_ajax_get_status_updates')); |
| 57 |
add_action('wp_ajax_mxchat_dismiss_completed_status', array($this, 'mxchat_ajax_dismiss_completed_status')); |
| 58 |
add_action('wp_ajax_mxchat_get_content_list', array($this, 'ajax_mxchat_get_content_list')); |
| 59 |
add_action('wp_ajax_mxchat_process_selected_content', array($this, 'ajax_mxchat_process_selected_content')); |
| 60 |
add_action('wp_ajax_mxchat_save_inline_prompt', array($this, 'mxchat_save_inline_prompt')); |
| 61 |
add_action('admin_post_mxchat_delete_pinecone_prompt', array($this, 'mxchat_handle_pinecone_prompt_delete')); |
| 62 |
add_action('wp_ajax_mxchat_delete_pinecone_prompt', array($this, 'ajax_mxchat_delete_pinecone_prompt')); |
| 63 |
add_action('wp_ajax_mxchat_delete_chunks_by_url', array($this, 'ajax_mxchat_delete_chunks_by_url')); |
| 64 |
add_action('wp_ajax_mxchat_delete_wordpress_prompt', array($this, 'ajax_mxchat_delete_wordpress_prompt')); |
| 65 |
add_action('wp_ajax_mxchat_bulk_delete_knowledge', array($this, 'ajax_mxchat_bulk_delete_knowledge')); |
| 66 |
add_action('wp_ajax_mxchat_update_role_restriction', array($this, 'ajax_mxchat_update_role_restriction')); |
| 67 |
|
| 68 |
// Queue-based processing AJAX handlers |
| 69 |
add_action('wp_ajax_mxchat_get_next_queue_item', array($this, 'ajax_mxchat_get_next_queue_item')); |
| 70 |
add_action('wp_ajax_mxchat_process_queue_item', array($this, 'ajax_mxchat_process_queue_item')); |
| 71 |
add_action('wp_ajax_mxchat_get_queue_status', array($this, 'ajax_mxchat_get_queue_status')); |
| 72 |
add_action('wp_ajax_mxchat_clear_queue', array($this, 'ajax_mxchat_clear_queue')); |
| 73 |
add_action('wp_ajax_mxchat_retry_failed', array($this, 'ajax_mxchat_retry_failed')); |
| 74 |
add_action('wp_ajax_mxchat_get_recent_entries', array($this, 'ajax_mxchat_get_recent_entries')); |
| 75 |
add_action('wp_ajax_mxchat_detect_sitemaps', array($this, 'ajax_mxchat_detect_sitemaps')); |
| 76 |
add_action('wp_ajax_mxchat_refresh_pinecone_entries', array($this, 'ajax_mxchat_refresh_pinecone_entries')); |
| 77 |
add_action('wp_ajax_mxchat_paginate_entries', array($this, 'ajax_mxchat_paginate_entries')); |
| 78 |
add_action('wp_ajax_mxchat_get_entry_content', array($this, 'ajax_mxchat_get_entry_content')); |
| 79 |
add_action('wp_ajax_mxchat_save_entry_content', array($this, 'ajax_mxchat_save_entry_content')); |
| 80 |
add_action('wp_ajax_mxchat_inspect_entry', array($this, 'ajax_mxchat_inspect_entry')); |
| 81 |
|
| 82 |
// WordPress post management hooks |
| 83 |
add_action('pre_post_update', array($this, 'mxchat_store_pre_update_status'), 10, 2); |
| 84 |
add_action('post_updated', array($this, 'mxchat_handle_post_update'), 10, 3); |
| 85 |
add_action('before_delete_post', array($this, 'mxchat_handle_post_delete')); |
| 86 |
add_action('wp_trash_post', array($this, 'mxchat_handle_post_delete')); |
| 87 |
// Authoritative unpublish detection: core hands this hook the REAL previous status, so |
| 88 |
// removal no longer depends on the mxchat_prev_status_* transients (evictable by persistent |
| 89 |
// object caches, never written by paths that bypass wp_update_post, e.g. plugins flipping |
| 90 |
// post_status directly and calling wp_transition_post_status themselves). |
| 91 |
add_action('transition_post_status', array($this, 'mxchat_handle_status_transition'), 10, 3); |
| 92 |
|
| 93 |
// ACF hook - fires AFTER ACF fields are saved, ensuring ACF data is available |
| 94 |
// Priority 20 to run after ACF's own save (which runs at priority 10) |
| 95 |
add_action('acf/save_post', array($this, 'mxchat_handle_acf_save'), 20); |
| 96 |
|
| 97 |
// One-time cleanup for vectors orphaned by unpublishes that predate the |
| 98 |
// transition_post_status handler (plan 816fb1): wp mxchat prune-unpublished |
| 99 |
if (defined('WP_CLI') && WP_CLI) { |
| 100 |
WP_CLI::add_command('mxchat prune-unpublished', array($this, 'cli_prune_unpublished')); |
| 101 |
// In-place repair for RTL KB rows imported in visual order before the |
| 102 |
// 32bf9e normalizer existed: wp mxchat rtl-repair (plan d1e6f7) |
| 103 |
WP_CLI::add_command('mxchat rtl-repair', array($this, 'cli_rtl_repair')); |
| 104 |
} |
| 105 |
|
| 106 |
add_action('wp_ajax_mxchat_mark_queue_complete', array($this, 'ajax_mxchat_mark_queue_complete')); |
| 107 |
|
| 108 |
// WooCommerce product hooks (if WooCommerce is active) |
| 109 |
if (class_exists('WooCommerce')) { |
| 110 |
add_action('pre_post_update', array($this, 'mxchat_store_pre_update_status'), 10, 2); |
| 111 |
add_action('save_post_product', array($this, 'mxchat_handle_product_change'), 10, 3); |
| 112 |
add_action('wp_trash_post', array($this, 'mxchat_handle_product_delete')); |
| 113 |
add_action('before_delete_post', array($this, 'mxchat_handle_product_delete')); |
| 114 |
} |
| 115 |
} |
| 116 |
|
| 117 |
/** |
| 118 |
* Get current options (refreshed) |
| 119 |
*/ |
| 120 |
private function mxchat_get_options() { |
| 121 |
if (empty($this->options)) { |
| 122 |
$this->options = get_option('mxchat_options', array()); |
| 123 |
} |
| 124 |
return $this->options; |
| 125 |
} |
| 126 |
|
| 127 |
|
| 128 |
// ======================================== |
| 129 |
// MAIN CONTENT SUBMISSION HANDLERS |
| 130 |
// ======================================== |
| 131 |
|
| 132 |
public function mxchat_handle_content_submission() { |
| 133 |
// Check if the form was submitted and the user has permission. |
| 134 |
if (!isset($_POST['submit_content']) || !current_user_can('manage_options')) { |
| 135 |
return; |
| 136 |
} |
| 137 |
|
| 138 |
// Verify the nonce. |
| 139 |
$nonce = isset($_POST['mxchat_submit_content_nonce']) ? sanitize_text_field(wp_unslash($_POST['mxchat_submit_content_nonce'])) : ''; |
| 140 |
if (!wp_verify_nonce($nonce, 'mxchat_submit_content_action')) { |
| 141 |
wp_die(esc_html__('Nonce verification failed.', 'mxchat')); |
| 142 |
} |
| 143 |
|
| 144 |
// Sanitize the inputs. |
| 145 |
// Use wp_kses_post to allow safe HTML and wp_unslash to remove WordPress added slashes |
| 146 |
$article_content = wp_kses_post(wp_unslash($_POST['article_content'])); |
| 147 |
$article_url = isset($_POST['article_url']) ? esc_url_raw($_POST['article_url']) : ''; |
| 148 |
|
| 149 |
// Get bot_id from form submission |
| 150 |
$bot_id = isset($_POST['bot_id']) ? sanitize_key($_POST['bot_id']) : 'default'; |
| 151 |
|
| 152 |
// Get bot-specific options and API key |
| 153 |
$bot_options = $this->get_bot_options($bot_id); |
| 154 |
$options = !empty($bot_options) ? $bot_options : get_option('mxchat_options'); |
| 155 |
|
| 156 |
// Custom-provider-aware decision; keyless custom sites must pass (plan cbd5fd). |
| 157 |
$preflight = MxChat_Utils::embedding_preflight($options); |
| 158 |
if (!$preflight['ok']) { |
| 159 |
set_transient('mxchat_admin_notice_error', esc_html($preflight['reason']), 30); |
| 160 |
wp_safe_redirect(esc_url(admin_url('admin.php?page=mxchat-prompts'))); |
| 161 |
exit; |
| 162 |
} |
| 163 |
$api_key = $preflight['api_key']; |
| 164 |
|
| 165 |
// Use centralized utility function with bot_id |
| 166 |
$result = MxChat_Utils::submit_content_to_db($article_content, $article_url, $api_key, null, $bot_id); |
| 167 |
|
| 168 |
if (is_wp_error($result)) { |
| 169 |
set_transient('mxchat_admin_notice_error', |
| 170 |
esc_html__('Error storing content: ', 'mxchat') . $result->get_error_message(), |
| 171 |
30 |
| 172 |
); |
| 173 |
} else { |
| 174 |
set_transient('mxchat_admin_notice_success', |
| 175 |
esc_html__('Content successfully submitted!', 'mxchat'), |
| 176 |
30 |
| 177 |
); |
| 178 |
} |
| 179 |
|
| 180 |
wp_safe_redirect(esc_url(admin_url('admin.php?page=mxchat-prompts'))); |
| 181 |
exit; |
| 182 |
} |
| 183 |
|
| 184 |
/** |
| 185 |
* Handle the "YouTube" KB import source (admin-post form submission). |
| 186 |
* |
| 187 |
* Per-video description mode: |
| 188 |
* - auto: fetch oEmbed metadata (reliable) + best-effort captions transcript. |
| 189 |
* If no usable transcript, index the metadata anyway, tell the admin, |
| 190 |
* and bounce back with the manual box pre-filled (never fail silently). |
| 191 |
* - manual: the admin's own description is what gets indexed; metadata rides along. |
| 192 |
* |
| 193 |
* The row is stored with content_type 'youtube' and source_url = the canonical |
| 194 |
* watch URL, so re-importing the same video UPDATES the entry (source_url |
| 195 |
* duplicate handling in MxChat_Utils::store_in_wordpress_db) — that is also the |
| 196 |
* "augment a metadata-only entry" path. |
| 197 |
*/ |
| 198 |
public function mxchat_handle_youtube_submission() { |
| 199 |
if (!isset($_POST['submit_youtube']) || !current_user_can('manage_options')) { |
| 200 |
wp_die(esc_html__('Unauthorized access', 'mxchat')); |
| 201 |
} |
| 202 |
|
| 203 |
check_admin_referer('mxchat_submit_youtube_action', 'mxchat_submit_youtube_nonce'); |
| 204 |
|
| 205 |
$redirect_url = admin_url('admin.php?page=mxchat-prompts'); |
| 206 |
|
| 207 |
$youtube_url = isset($_POST['youtube_url']) ? esc_url_raw(wp_unslash($_POST['youtube_url'])) : ''; |
| 208 |
$video_id = MxChat_Utils::parse_youtube_id($youtube_url); |
| 209 |
|
| 210 |
if (empty($video_id)) { |
| 211 |
set_transient('mxchat_admin_notice_error', |
| 212 |
esc_html__('That does not look like a link to a single YouTube video. Please paste a watch, youtu.be, or Shorts URL.', 'mxchat'), |
| 213 |
30 |
| 214 |
); |
| 215 |
wp_safe_redirect(esc_url($redirect_url)); |
| 216 |
exit; |
| 217 |
} |
| 218 |
|
| 219 |
$canonical_url = 'https://www.youtube.com/watch?v=' . $video_id; |
| 220 |
|
| 221 |
$description_mode = (isset($_POST['youtube_description_mode']) && $_POST['youtube_description_mode'] === 'manual') ? 'manual' : 'auto'; |
| 222 |
$manual_description = isset($_POST['youtube_description']) ? trim(wp_kses_post(wp_unslash($_POST['youtube_description']))) : ''; |
| 223 |
|
| 224 |
$bot_id = isset($_POST['bot_id']) ? sanitize_key($_POST['bot_id']) : 'default'; |
| 225 |
|
| 226 |
// Resolve the embedding decision exactly like the sibling handlers — |
| 227 |
// custom-provider-aware (plan cbd5fd). |
| 228 |
$bot_options = $this->get_bot_options($bot_id); |
| 229 |
$options = !empty($bot_options) ? $bot_options : get_option('mxchat_options'); |
| 230 |
|
| 231 |
$preflight = MxChat_Utils::embedding_preflight($options); |
| 232 |
if (!$preflight['ok']) { |
| 233 |
set_transient('mxchat_admin_notice_error', esc_html($preflight['reason']), 30); |
| 234 |
wp_safe_redirect(esc_url($redirect_url)); |
| 235 |
exit; |
| 236 |
} |
| 237 |
$api_key = $preflight['api_key']; |
| 238 |
|
| 239 |
// Metadata is fetched in BOTH modes — it is the reliable half of auto, and in |
| 240 |
// manual mode it enriches the indexed text with the real title/channel. |
| 241 |
$meta = $this->mxchat_fetch_youtube_oembed($video_id); |
| 242 |
$video_title = isset($meta['title']) ? sanitize_text_field($meta['title']) : ''; |
| 243 |
$video_channel = isset($meta['author_name']) ? sanitize_text_field($meta['author_name']) : ''; |
| 244 |
|
| 245 |
$header_lines = 'YouTube Video: ' . ($video_title !== '' ? $video_title : $canonical_url) . "\n"; |
| 246 |
if ($video_channel !== '') { |
| 247 |
$header_lines .= 'Channel: ' . $video_channel . "\n"; |
| 248 |
} |
| 249 |
$header_lines .= 'URL: ' . $canonical_url . "\n\n"; |
| 250 |
|
| 251 |
$transcript_missing = false; |
| 252 |
|
| 253 |
if ($description_mode === 'manual') { |
| 254 |
if ($manual_description === '') { |
| 255 |
set_transient('mxchat_admin_notice_error', |
| 256 |
esc_html__('Please write a description for the video, or switch to auto-fetch.', 'mxchat'), |
| 257 |
30 |
| 258 |
); |
| 259 |
wp_safe_redirect(esc_url($redirect_url)); |
| 260 |
exit; |
| 261 |
} |
| 262 |
$indexed_text = $header_lines . $manual_description; |
| 263 |
} else { |
| 264 |
$transcript = $this->mxchat_fetch_youtube_transcript($video_id); |
| 265 |
|
| 266 |
if (strlen($transcript) >= 200) { |
| 267 |
$indexed_text = $header_lines . $transcript; |
| 268 |
} else { |
| 269 |
// Graceful fallback: captions disabled / blocked / no speech. Auto |
| 270 |
// reliably gets metadata; it does NOT guarantee a transcript. |
| 271 |
$transcript_missing = true; |
| 272 |
|
| 273 |
if ($video_title === '' && $video_channel === '') { |
| 274 |
// Both halves failed — nothing meaningful to index. |
| 275 |
set_transient('mxchat_admin_notice_error', |
| 276 |
esc_html__('Could not retrieve any information for that video (no metadata and no captions). Please check the URL, or use the manual description option.', 'mxchat'), |
| 277 |
30 |
| 278 |
); |
| 279 |
wp_safe_redirect(esc_url($redirect_url)); |
| 280 |
exit; |
| 281 |
} |
| 282 |
|
| 283 |
$indexed_text = $header_lines . sprintf( |
| 284 |
/* translators: 1: video title, 2: channel name */ |
| 285 |
__('A YouTube video titled "%1$s" from the channel %2$s.', 'mxchat'), |
| 286 |
$video_title !== '' ? $video_title : $canonical_url, |
| 287 |
$video_channel !== '' ? $video_channel : 'YouTube' |
| 288 |
); |
| 289 |
} |
| 290 |
} |
| 291 |
|
| 292 |
$result = MxChat_Utils::submit_content_to_db($indexed_text, $canonical_url, $api_key, null, $bot_id, 'youtube'); |
| 293 |
|
| 294 |
if (is_wp_error($result)) { |
| 295 |
set_transient('mxchat_admin_notice_error', |
| 296 |
esc_html__('Error storing video in the knowledge base: ', 'mxchat') . $result->get_error_message(), |
| 297 |
30 |
| 298 |
); |
| 299 |
wp_safe_redirect(esc_url($redirect_url)); |
| 300 |
exit; |
| 301 |
} |
| 302 |
|
| 303 |
if ($transcript_missing) { |
| 304 |
set_transient('mxchat_admin_notice_success', |
| 305 |
esc_html__('Video indexed from its title and channel — no captions were available for a transcript. The form below is pre-filled: write your own description and import again to improve matching (it updates the same entry).', 'mxchat'), |
| 306 |
30 |
| 307 |
); |
| 308 |
// Bounce back with prefill args so the page reopens the YouTube form in |
| 309 |
// manual mode with the URL + fetched title ready to augment. |
| 310 |
$redirect_url = add_query_arg(array( |
| 311 |
'mxchat_yt_prefill' => '1', |
| 312 |
'yt_url' => rawurlencode($canonical_url), |
| 313 |
'yt_title' => rawurlencode($video_title), |
| 314 |
), $redirect_url); |
| 315 |
} else { |
| 316 |
set_transient('mxchat_admin_notice_success', |
| 317 |
esc_html__('YouTube video successfully added to the knowledge base!', 'mxchat'), |
| 318 |
30 |
| 319 |
); |
| 320 |
} |
| 321 |
|
| 322 |
wp_safe_redirect(esc_url_raw($redirect_url)); |
| 323 |
exit; |
| 324 |
} |
| 325 |
|
| 326 |
/** |
| 327 |
* Fetch YouTube oEmbed metadata for a video (no API key required). |
| 328 |
* Returns the decoded array (title, author_name, thumbnail_url, ...) or array(). |
| 329 |
*/ |
| 330 |
private function mxchat_fetch_youtube_oembed($video_id) { |
| 331 |
$oembed_url = 'https://www.youtube.com/oembed?url=' . rawurlencode('https://www.youtube.com/watch?v=' . $video_id) . '&format=json'; |
| 332 |
$response = wp_remote_get($oembed_url, array('timeout' => 15)); |
| 333 |
if (is_wp_error($response) || wp_remote_retrieve_response_code($response) !== 200) { |
| 334 |
return array(); |
| 335 |
} |
| 336 |
$data = json_decode(wp_remote_retrieve_body($response), true); |
| 337 |
return is_array($data) ? $data : array(); |
| 338 |
} |
| 339 |
|
| 340 |
/** |
| 341 |
* Best-effort captions transcript for a video. Deliberately ISOLATED: this uses |
| 342 |
* YouTube's unofficial timedtext route (the caption track list embedded in the |
| 343 |
* watch page), which YouTube has broken before and will break again. Every |
| 344 |
* failure mode returns '' so a break degrades to the metadata-only import path |
| 345 |
* instead of erroring the whole submission. Do not let anything in here throw. |
| 346 |
*/ |
| 347 |
private function mxchat_fetch_youtube_transcript($video_id) { |
| 348 |
$watch_url = 'https://www.youtube.com/watch?v=' . $video_id . '&hl=en'; |
| 349 |
|
| 350 |
// First try the honest ingest UA; some responses omit the player config for |
| 351 |
// bot UAs, so retry once with a browser UA before giving up. |
| 352 |
$user_agents = array( |
| 353 |
mxchat_ingest_user_agent(), |
| 354 |
'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/126.0.0.0 Safari/537.36', |
| 355 |
); |
| 356 |
|
| 357 |
$tracks = array(); |
| 358 |
foreach ($user_agents as $ua) { |
| 359 |
$response = wp_remote_get($watch_url, array( |
| 360 |
'timeout' => 20, |
| 361 |
'user-agent' => $ua, |
| 362 |
)); |
| 363 |
if (is_wp_error($response) || wp_remote_retrieve_response_code($response) !== 200) { |
| 364 |
continue; |
| 365 |
} |
| 366 |
$body = wp_remote_retrieve_body($response); |
| 367 |
if (!is_string($body) || $body === '' || !preg_match('/"captionTracks":(\[.*?\])(?=,")/s', $body, $m)) { |
| 368 |
continue; |
| 369 |
} |
| 370 |
$decoded = json_decode($m[1], true); |
| 371 |
if (is_array($decoded) && !empty($decoded)) { |
| 372 |
$tracks = $decoded; |
| 373 |
break; |
| 374 |
} |
| 375 |
} |
| 376 |
|
| 377 |
if (empty($tracks)) { |
| 378 |
return ''; |
| 379 |
} |
| 380 |
|
| 381 |
// Prefer an English track, else take the first offered. |
| 382 |
$chosen = null; |
| 383 |
foreach ($tracks as $track) { |
| 384 |
if (isset($track['languageCode']) && strpos($track['languageCode'], 'en') === 0) { |
| 385 |
$chosen = $track; |
| 386 |
break; |
| 387 |
} |
| 388 |
} |
| 389 |
if ($chosen === null) { |
| 390 |
$chosen = $tracks[0]; |
| 391 |
} |
| 392 |
if (empty($chosen['baseUrl']) || !is_string($chosen['baseUrl'])) { |
| 393 |
return ''; |
| 394 |
} |
| 395 |
|
| 396 |
$timedtext = wp_remote_get($chosen['baseUrl'], array('timeout' => 20)); |
| 397 |
if (is_wp_error($timedtext) || wp_remote_retrieve_response_code($timedtext) !== 200) { |
| 398 |
return ''; |
| 399 |
} |
| 400 |
$xml = wp_remote_retrieve_body($timedtext); |
| 401 |
if (!is_string($xml) || strpos($xml, '<text') === false) { |
| 402 |
return ''; |
| 403 |
} |
| 404 |
|
| 405 |
// <text start=".." dur="..">caption</text> — strip tags, decode the |
| 406 |
// double-encoded entities timedtext ships, collapse whitespace. |
| 407 |
$text = preg_replace('/<[^>]+>/', ' ', $xml); |
| 408 |
$text = html_entity_decode(html_entity_decode($text, ENT_QUOTES | ENT_HTML5, 'UTF-8'), ENT_QUOTES | ENT_HTML5, 'UTF-8'); |
| 409 |
$text = trim(preg_replace('/\s+/u', ' ', $text)); |
| 410 |
|
| 411 |
return $text; |
| 412 |
} |
| 413 |
|
| 414 |
public function mxchat_is_pdf_url($url, $response) { |
| 415 |
$content_type = wp_remote_retrieve_header($response, 'content-type'); |
| 416 |
$file_extension = strtolower(pathinfo($url, PATHINFO_EXTENSION)); |
| 417 |
|
| 418 |
// Check Content-Disposition header for .pdf filename (Google Drive sends this) |
| 419 |
$disposition = wp_remote_retrieve_header($response, 'content-disposition'); |
| 420 |
$has_pdf_disposition = ! empty($disposition) && stripos($disposition, '.pdf') !== false; |
| 421 |
|
| 422 |
return strpos($content_type, 'pdf') !== false || $file_extension === 'pdf' || $has_pdf_disposition; |
| 423 |
} |
| 424 |
|
| 425 |
|
| 426 |
public function mxchat_handle_pdf_for_knowledge_base($pdf_url, $response, $bot_id = 'default') { |
| 427 |
if (!current_user_can('manage_options')) { |
| 428 |
return false; |
| 429 |
} |
| 430 |
|
| 431 |
$pdf_url = esc_url_raw($pdf_url); |
| 432 |
$upload_dir = wp_upload_dir(); |
| 433 |
|
| 434 |
if (isset($upload_dir['error']) && $upload_dir['error'] !== false) { |
| 435 |
return false; |
| 436 |
} |
| 437 |
|
| 438 |
$pdf_filename = sanitize_file_name('mxchat_kb_' . time() . '.pdf'); |
| 439 |
$pdf_path = trailingslashit($upload_dir['path']) . $pdf_filename; |
| 440 |
|
| 441 |
$response_body = wp_remote_retrieve_body($response); |
| 442 |
if (empty($response_body)) { |
| 443 |
return false; |
| 444 |
} |
| 445 |
|
| 446 |
if (!wp_mkdir_p(dirname($pdf_path))) { |
| 447 |
return false; |
| 448 |
} |
| 449 |
|
| 450 |
try { |
| 451 |
file_put_contents($pdf_path, $response_body); |
| 452 |
|
| 453 |
if (!file_exists($pdf_path)) { |
| 454 |
throw new Exception(__('Failed to save PDF file', 'mxchat')); |
| 455 |
} |
| 456 |
|
| 457 |
$total_pages = $this->mxchat_validate_and_count_pdf_pages($pdf_path); |
| 458 |
|
| 459 |
if ($total_pages === false || $total_pages < 1) { |
| 460 |
throw new Exception(__('Invalid PDF: Unable to parse or no pages found', 'mxchat')); |
| 461 |
} |
| 462 |
|
| 463 |
// Create unique queue ID |
| 464 |
$queue_id = 'pdf_' . md5($pdf_url . time()); |
| 465 |
|
| 466 |
// Create array of pages to process |
| 467 |
$pages = array(); |
| 468 |
for ($i = 1; $i <= $total_pages; $i++) { |
| 469 |
$pages[] = array( |
| 470 |
'pdf_path' => $pdf_path, |
| 471 |
'pdf_url' => $pdf_url, |
| 472 |
'page_number' => $i, |
| 473 |
'total_pages' => $total_pages |
| 474 |
); |
| 475 |
} |
| 476 |
|
| 477 |
// Add pages to queue |
| 478 |
$queued_count = $this->mxchat_add_to_queue($queue_id, 'pdf_page', $pages, $bot_id); |
| 479 |
|
| 480 |
if ($queued_count === 0) { |
| 481 |
wp_delete_file($pdf_path); |
| 482 |
throw new Exception(__('Failed to add PDF pages to processing queue', 'mxchat')); |
| 483 |
} |
| 484 |
|
| 485 |
// Store queue metadata |
| 486 |
$this->mxchat_set_queue_meta($queue_id, 'source_url', $pdf_url); |
| 487 |
$this->mxchat_set_queue_meta($queue_id, 'queue_type', 'pdf'); |
| 488 |
$this->mxchat_set_queue_meta($queue_id, 'total_items', $total_pages); |
| 489 |
$this->mxchat_set_queue_meta($queue_id, 'bot_id', $bot_id); |
| 490 |
$this->mxchat_set_queue_meta($queue_id, 'pdf_path', $pdf_path); |
| 491 |
$this->mxchat_set_queue_meta($queue_id, 'created_at', current_time('mysql')); |
| 492 |
|
| 493 |
// Store queue ID in transient for status tracking |
| 494 |
set_transient('mxchat_active_queue_pdf', $queue_id, DAY_IN_SECONDS); |
| 495 |
set_transient('mxchat_last_pdf_url', $pdf_url, DAY_IN_SECONDS); |
| 496 |
|
| 497 |
return 'queued'; |
| 498 |
|
| 499 |
} catch (Exception $e) { |
| 500 |
if (file_exists($pdf_path)) { |
| 501 |
wp_delete_file($pdf_path); |
| 502 |
} |
| 503 |
return $e->getMessage(); |
| 504 |
} |
| 505 |
} |
| 506 |
|
| 507 |
/** |
| 508 |
* Handle direct PDF file upload from the knowledge base page |
| 509 |
*/ |
| 510 |
public function mxchat_handle_pdf_file_submission() { |
| 511 |
if (!isset($_POST['submit_pdf_file']) || !current_user_can('manage_options')) { |
| 512 |
wp_die(esc_html__('Unauthorized access', 'mxchat')); |
| 513 |
} |
| 514 |
|
| 515 |
check_admin_referer('mxchat_submit_pdf_file_action', 'mxchat_submit_pdf_file_nonce'); |
| 516 |
|
| 517 |
$redirect_url = admin_url('admin.php?page=mxchat-prompts'); |
| 518 |
|
| 519 |
// Validate file upload |
| 520 |
if (empty($_FILES['pdf_file']) || $_FILES['pdf_file']['error'] !== UPLOAD_ERR_OK) { |
| 521 |
$error_code = isset($_FILES['pdf_file']['error']) ? $_FILES['pdf_file']['error'] : UPLOAD_ERR_NO_FILE; |
| 522 |
$error_messages = array( |
| 523 |
UPLOAD_ERR_INI_SIZE => __('The uploaded file exceeds the server upload_max_filesize limit.', 'mxchat'), |
| 524 |
UPLOAD_ERR_FORM_SIZE => __('The uploaded file exceeds the form MAX_FILE_SIZE limit.', 'mxchat'), |
| 525 |
UPLOAD_ERR_PARTIAL => __('The file was only partially uploaded.', 'mxchat'), |
| 526 |
UPLOAD_ERR_NO_FILE => __('No file was uploaded. Please select a PDF file.', 'mxchat'), |
| 527 |
UPLOAD_ERR_NO_TMP_DIR => __('Server missing temporary folder.', 'mxchat'), |
| 528 |
UPLOAD_ERR_CANT_WRITE => __('Server failed to write file to disk.', 'mxchat'), |
| 529 |
); |
| 530 |
$error_msg = isset($error_messages[$error_code]) ? $error_messages[$error_code] : __('Unknown upload error.', 'mxchat'); |
| 531 |
set_transient('mxchat_admin_notice_error', $error_msg, 30); |
| 532 |
wp_safe_redirect(esc_url($redirect_url)); |
| 533 |
exit; |
| 534 |
} |
| 535 |
|
| 536 |
$file = $_FILES['pdf_file']; |
| 537 |
|
| 538 |
// Validate MIME type |
| 539 |
$finfo = finfo_open(FILEINFO_MIME_TYPE); |
| 540 |
$mime_type = finfo_file($finfo, $file['tmp_name']); |
| 541 |
finfo_close($finfo); |
| 542 |
|
| 543 |
if ($mime_type !== 'application/pdf') { |
| 544 |
set_transient('mxchat_admin_notice_error', |
| 545 |
esc_html__('Invalid file type. Only PDF files are accepted.', 'mxchat'), |
| 546 |
30 |
| 547 |
); |
| 548 |
wp_safe_redirect(esc_url($redirect_url)); |
| 549 |
exit; |
| 550 |
} |
| 551 |
|
| 552 |
// Validate extension |
| 553 |
$ext = strtolower(pathinfo($file['name'], PATHINFO_EXTENSION)); |
| 554 |
if ($ext !== 'pdf') { |
| 555 |
set_transient('mxchat_admin_notice_error', |
| 556 |
esc_html__('Invalid file extension. Only .pdf files are accepted.', 'mxchat'), |
| 557 |
30 |
| 558 |
); |
| 559 |
wp_safe_redirect(esc_url($redirect_url)); |
| 560 |
exit; |
| 561 |
} |
| 562 |
|
| 563 |
$bot_id = isset($_POST['bot_id']) ? sanitize_key($_POST['bot_id']) : 'default'; |
| 564 |
$original_filename = sanitize_file_name($file['name']); |
| 565 |
|
| 566 |
$upload_dir = wp_upload_dir(); |
| 567 |
if (isset($upload_dir['error']) && $upload_dir['error'] !== false) { |
| 568 |
set_transient('mxchat_admin_notice_error', |
| 569 |
esc_html__('WordPress upload directory is not writable.', 'mxchat'), |
| 570 |
30 |
| 571 |
); |
| 572 |
wp_safe_redirect(esc_url($redirect_url)); |
| 573 |
exit; |
| 574 |
} |
| 575 |
|
| 576 |
$pdf_filename = sanitize_file_name('mxchat_kb_' . time() . '.pdf'); |
| 577 |
$pdf_path = trailingslashit($upload_dir['path']) . $pdf_filename; |
| 578 |
|
| 579 |
if (!wp_mkdir_p(dirname($pdf_path))) { |
| 580 |
set_transient('mxchat_admin_notice_error', |
| 581 |
esc_html__('Failed to create upload directory.', 'mxchat'), |
| 582 |
30 |
| 583 |
); |
| 584 |
wp_safe_redirect(esc_url($redirect_url)); |
| 585 |
exit; |
| 586 |
} |
| 587 |
|
| 588 |
// Move uploaded file |
| 589 |
if (!move_uploaded_file($file['tmp_name'], $pdf_path)) { |
| 590 |
set_transient('mxchat_admin_notice_error', |
| 591 |
esc_html__('Failed to save uploaded PDF file.', 'mxchat'), |
| 592 |
30 |
| 593 |
); |
| 594 |
wp_safe_redirect(esc_url($redirect_url)); |
| 595 |
exit; |
| 596 |
} |
| 597 |
|
| 598 |
try { |
| 599 |
$total_pages = $this->mxchat_validate_and_count_pdf_pages($pdf_path); |
| 600 |
|
| 601 |
if ($total_pages === false || $total_pages < 1) { |
| 602 |
throw new Exception(__('Invalid PDF: Unable to parse or no pages found', 'mxchat')); |
| 603 |
} |
| 604 |
|
| 605 |
// Use original filename as the source identifier |
| 606 |
$source_label = 'upload://' . $original_filename; |
| 607 |
|
| 608 |
$queue_id = 'pdf_' . md5($source_label . time()); |
| 609 |
|
| 610 |
$pages = array(); |
| 611 |
for ($i = 1; $i <= $total_pages; $i++) { |
| 612 |
$pages[] = array( |
| 613 |
'pdf_path' => $pdf_path, |
| 614 |
'pdf_url' => $source_label, |
| 615 |
'page_number' => $i, |
| 616 |
'total_pages' => $total_pages, |
| 617 |
); |
| 618 |
} |
| 619 |
|
| 620 |
$queued_count = $this->mxchat_add_to_queue($queue_id, 'pdf_page', $pages, $bot_id); |
| 621 |
|
| 622 |
if ($queued_count === 0) { |
| 623 |
wp_delete_file($pdf_path); |
| 624 |
throw new Exception(__('Failed to add PDF pages to processing queue', 'mxchat')); |
| 625 |
} |
| 626 |
|
| 627 |
$this->mxchat_set_queue_meta($queue_id, 'source_url', $source_label); |
| 628 |
$this->mxchat_set_queue_meta($queue_id, 'queue_type', 'pdf'); |
| 629 |
$this->mxchat_set_queue_meta($queue_id, 'total_items', $total_pages); |
| 630 |
$this->mxchat_set_queue_meta($queue_id, 'bot_id', $bot_id); |
| 631 |
$this->mxchat_set_queue_meta($queue_id, 'pdf_path', $pdf_path); |
| 632 |
$this->mxchat_set_queue_meta($queue_id, 'created_at', current_time('mysql')); |
| 633 |
|
| 634 |
set_transient('mxchat_active_queue_pdf', $queue_id, DAY_IN_SECONDS); |
| 635 |
set_transient('mxchat_last_pdf_url', $source_label, DAY_IN_SECONDS); |
| 636 |
|
| 637 |
set_transient('mxchat_admin_notice_success', |
| 638 |
sprintf( |
| 639 |
esc_html__('PDF "%s" (%d pages) queued for processing. Processing will start automatically.', 'mxchat'), |
| 640 |
esc_html($original_filename), |
| 641 |
$total_pages |
| 642 |
), |
| 643 |
30 |
| 644 |
); |
| 645 |
|
| 646 |
} catch (Exception $e) { |
| 647 |
if (file_exists($pdf_path)) { |
| 648 |
wp_delete_file($pdf_path); |
| 649 |
} |
| 650 |
set_transient('mxchat_admin_notice_error', |
| 651 |
esc_html__('Failed to process uploaded PDF: ', 'mxchat') . esc_html($e->getMessage()), |
| 652 |
30 |
| 653 |
); |
| 654 |
} |
| 655 |
|
| 656 |
wp_safe_redirect(esc_url($redirect_url)); |
| 657 |
exit; |
| 658 |
} |
| 659 |
|
| 660 |
/** |
| 661 |
* Handle direct document upload (.docx / .txt / .md) from the knowledge base |
| 662 |
* page (plan 0485e5). Unlike PDF Upload there is no per-page queue: the text |
| 663 |
* extracts in one pass and routes through submit_content_to_db, whose chunker |
| 664 |
* takes over for long content. The uploaded file is read from the PHP temp |
| 665 |
* file and never persisted — only its extracted text enters the KB. |
| 666 |
* |
| 667 |
* Source identity matches PDF Upload's scheme: upload://<filename>, stable |
| 668 |
* across re-uploads so a re-import REPLACES (delete_chunks_for_url + upsert |
| 669 |
* per identity) instead of duplicating. |
| 670 |
*/ |
| 671 |
public function mxchat_handle_document_file_submission() { |
| 672 |
if (!isset($_POST['submit_document_file']) || !current_user_can('manage_options')) { |
| 673 |
wp_die(esc_html__('Unauthorized access', 'mxchat')); |
| 674 |
} |
| 675 |
|
| 676 |
check_admin_referer('mxchat_submit_document_file_action', 'mxchat_submit_document_file_nonce'); |
| 677 |
|
| 678 |
$redirect_url = admin_url('admin.php?page=mxchat-prompts'); |
| 679 |
|
| 680 |
if (empty($_FILES['document_file']) || $_FILES['document_file']['error'] !== UPLOAD_ERR_OK) { |
| 681 |
$error_code = isset($_FILES['document_file']['error']) ? $_FILES['document_file']['error'] : UPLOAD_ERR_NO_FILE; |
| 682 |
$error_messages = array( |
| 683 |
UPLOAD_ERR_INI_SIZE => __('The uploaded file exceeds the server upload_max_filesize limit.', 'mxchat'), |
| 684 |
UPLOAD_ERR_FORM_SIZE => __('The uploaded file exceeds the form MAX_FILE_SIZE limit.', 'mxchat'), |
| 685 |
UPLOAD_ERR_PARTIAL => __('The file was only partially uploaded.', 'mxchat'), |
| 686 |
UPLOAD_ERR_NO_FILE => __('No file was uploaded. Please select a document.', 'mxchat'), |
| 687 |
UPLOAD_ERR_NO_TMP_DIR => __('Server missing temporary folder.', 'mxchat'), |
| 688 |
UPLOAD_ERR_CANT_WRITE => __('Server failed to write file to disk.', 'mxchat'), |
| 689 |
); |
| 690 |
$error_msg = isset($error_messages[$error_code]) ? $error_messages[$error_code] : __('Unknown upload error.', 'mxchat'); |
| 691 |
set_transient('mxchat_admin_notice_error', $error_msg, 30); |
| 692 |
wp_safe_redirect(esc_url($redirect_url)); |
| 693 |
exit; |
| 694 |
} |
| 695 |
|
| 696 |
$file = $_FILES['document_file']; |
| 697 |
$ext = strtolower(pathinfo($file['name'], PATHINFO_EXTENSION)); |
| 698 |
|
| 699 |
$finfo = finfo_open(FILEINFO_MIME_TYPE); |
| 700 |
$mime_type = finfo_file($finfo, $file['tmp_name']); |
| 701 |
finfo_close($finfo); |
| 702 |
|
| 703 |
// Per-extension MIME expectations. finfo commonly reports .docx as |
| 704 |
// application/zip (it IS a Zip container) and .md as plain text. |
| 705 |
$mime_ok = false; |
| 706 |
if ($ext === 'docx') { |
| 707 |
$mime_ok = in_array($mime_type, array( |
| 708 |
'application/vnd.openxmlformats-officedocument.wordprocessingml.document', |
| 709 |
'application/zip', |
| 710 |
), true); |
| 711 |
} elseif ($ext === 'txt' || $ext === 'md') { |
| 712 |
$mime_ok = (strpos((string) $mime_type, 'text/') === 0); |
| 713 |
} |
| 714 |
|
| 715 |
if (!$mime_ok) { |
| 716 |
set_transient('mxchat_admin_notice_error', |
| 717 |
esc_html__('Invalid or unreadable document. Accepted types: .docx, .txt, .md.', 'mxchat'), |
| 718 |
30 |
| 719 |
); |
| 720 |
wp_safe_redirect(esc_url($redirect_url)); |
| 721 |
exit; |
| 722 |
} |
| 723 |
|
| 724 |
$bot_id = isset($_POST['bot_id']) ? sanitize_key($_POST['bot_id']) : 'default'; |
| 725 |
$original_filename = sanitize_file_name($file['name']); |
| 726 |
|
| 727 |
// ---- Extract text (ONE extractor for .docx — the word handler's) ---- |
| 728 |
if ($ext === 'docx') { |
| 729 |
$text = MXChat_Word_Handler::extract_docx_text($file['tmp_name']); |
| 730 |
if ($text === false) { |
| 731 |
set_transient('mxchat_admin_notice_error', |
| 732 |
esc_html__('The .docx file could not be read. It may be corrupt, empty, or not a real Word document.', 'mxchat'), |
| 733 |
30 |
| 734 |
); |
| 735 |
wp_safe_redirect(esc_url($redirect_url)); |
| 736 |
exit; |
| 737 |
} |
| 738 |
} else { |
| 739 |
// .txt / .md read as-is. Markdown keeps its syntax on purpose — |
| 740 |
// headings are useful retrieval signal. |
| 741 |
$text = (string) file_get_contents($file['tmp_name']); |
| 742 |
$text = wp_check_invalid_utf8($text); |
| 743 |
$text = trim($text); |
| 744 |
} |
| 745 |
|
| 746 |
if ($text === '') { |
| 747 |
set_transient('mxchat_admin_notice_error', |
| 748 |
esc_html__('The uploaded document contains no readable text.', 'mxchat'), |
| 749 |
30 |
| 750 |
); |
| 751 |
wp_safe_redirect(esc_url($redirect_url)); |
| 752 |
exit; |
| 753 |
} |
| 754 |
|
| 755 |
// Size cap — same pdf_max_pages setting the PDF/toolbar paths use, but |
| 756 |
// estimated by CHARACTERS (~2500/page): the .docx cleaner collapses all |
| 757 |
// newlines to spaces, so a paragraph count reads 1 for any Word file. |
| 758 |
// Processing is synchronous — an unbounded document risks a timeout. |
| 759 |
$options = get_option('mxchat_options', array()); |
| 760 |
$max_pages = isset($options['pdf_max_pages']) ? intval($options['pdf_max_pages']) : 69; |
| 761 |
$estimated_pages = (int) ceil(strlen($text) / 2500); |
| 762 |
if ($estimated_pages > $max_pages) { |
| 763 |
set_transient('mxchat_admin_notice_error', |
| 764 |
sprintf( |
| 765 |
esc_html__('The document is too large (about %1$d pages; the limit is %2$d). Split it into smaller files, or raise the PDF max pages setting.', 'mxchat'), |
| 766 |
$estimated_pages, |
| 767 |
$max_pages |
| 768 |
), |
| 769 |
30 |
| 770 |
); |
| 771 |
wp_safe_redirect(esc_url($redirect_url)); |
| 772 |
exit; |
| 773 |
} |
| 774 |
|
| 775 |
// Embedding API key — bot-aware, same shape as the direct-content handler. |
| 776 |
$api_key = ''; |
| 777 |
if ($bot_id !== 'default' && class_exists('MxChat_Multi_Bot_Manager')) { |
| 778 |
$bot_options = apply_filters('mxchat_get_bot_options', array(), $bot_id); |
| 779 |
$api_key = $bot_options['api_key'] ?? ''; |
| 780 |
} |
| 781 |
if (empty($api_key)) { |
| 782 |
$api_key = $options['api_key'] ?? ''; |
| 783 |
} |
| 784 |
|
| 785 |
// Stable identity — PDF Upload's scheme. A re-upload of the same filename |
| 786 |
// replaces: clear old chunks first (covers a doc shrinking below the chunk |
| 787 |
// threshold, where the single-vector path would not clean them), then |
| 788 |
// submit — the chunked path re-deletes harmlessly. |
| 789 |
$source_label = 'upload://' . $original_filename; |
| 790 |
MxChat_Utils::delete_chunks_for_url($source_label, $bot_id); |
| 791 |
$result = MxChat_Utils::submit_content_to_db($text, $source_label, $api_key, null, $bot_id, 'document'); |
| 792 |
|
| 793 |
if (is_wp_error($result)) { |
| 794 |
set_transient('mxchat_admin_notice_error', |
| 795 |
esc_html__('Failed to import the document: ', 'mxchat') . esc_html($result->get_error_message()), |
| 796 |
30 |
| 797 |
); |
| 798 |
} else { |
| 799 |
set_transient('mxchat_admin_notice_success', |
| 800 |
sprintf( |
| 801 |
esc_html__('Document "%s" imported into the knowledge base.', 'mxchat'), |
| 802 |
esc_html($original_filename) |
| 803 |
), |
| 804 |
30 |
| 805 |
); |
| 806 |
} |
| 807 |
|
| 808 |
wp_safe_redirect(esc_url($redirect_url)); |
| 809 |
exit; |
| 810 |
} |
| 811 |
|
| 812 |
/** |
| 813 |
* Validate PDF and count pages with multiple parser attempts |
| 814 |
*/ |
| 815 |
private function mxchat_validate_and_count_pdf_pages($pdf_path) { |
| 816 |
// Method 1: Try with Smalot PDF Parser (your current method) |
| 817 |
try { |
| 818 |
mxchat_load_pdf_parser(); |
| 819 |
$parser = new \Smalot\PdfParser\Parser(); |
| 820 |
$pdf = $parser->parseFile($pdf_path); |
| 821 |
$pages = $pdf->getPages(); |
| 822 |
$page_count = count($pages); |
| 823 |
|
| 824 |
if ($page_count > 0) { |
| 825 |
//error_log('PDF parsed successfully with Smalot parser: ' . $page_count . ' pages'); |
| 826 |
return $page_count; |
| 827 |
} |
| 828 |
} catch (Exception $e) { |
| 829 |
//error_log('Smalot PDF parser failed: ' . $e->getMessage()); |
| 830 |
} |
| 831 |
|
| 832 |
// Method 2: Try with pdfinfo command (if available) |
| 833 |
if (function_exists('shell_exec') && !$this->mxchat_is_shell_disabled()) { |
| 834 |
try { |
| 835 |
$command = 'pdfinfo ' . escapeshellarg($pdf_path) . ' 2>&1'; |
| 836 |
$output = shell_exec($command); |
| 837 |
|
| 838 |
if ($output && preg_match('/Pages:\s*(\d+)/', $output, $matches)) { |
| 839 |
$page_count = intval($matches[1]); |
| 840 |
if ($page_count > 0) { |
| 841 |
//error_log('PDF parsed successfully with pdfinfo: ' . $page_count . ' pages'); |
| 842 |
return $page_count; |
| 843 |
} |
| 844 |
} |
| 845 |
} catch (Exception $e) { |
| 846 |
//error_log('pdfinfo command failed: ' . $e->getMessage()); |
| 847 |
} |
| 848 |
} |
| 849 |
|
| 850 |
// Method 3: Try to repair PDF and parse again |
| 851 |
try { |
| 852 |
$repaired_path = $this->mxchat_attempt_pdf_repair($pdf_path); |
| 853 |
if ($repaired_path && $repaired_path !== $pdf_path) { |
| 854 |
mxchat_load_pdf_parser(); |
| 855 |
$parser = new \Smalot\PdfParser\Parser(); |
| 856 |
$pdf = $parser->parseFile($repaired_path); |
| 857 |
$pages = $pdf->getPages(); |
| 858 |
$page_count = count($pages); |
| 859 |
|
| 860 |
if ($page_count > 0) { |
| 861 |
// Replace original with repaired version |
| 862 |
copy($repaired_path, $pdf_path); |
| 863 |
unlink($repaired_path); |
| 864 |
//error_log('PDF repaired and parsed successfully: ' . $page_count . ' pages'); |
| 865 |
return $page_count; |
| 866 |
} |
| 867 |
|
| 868 |
// Clean up repaired file if it didn't work |
| 869 |
unlink($repaired_path); |
| 870 |
} |
| 871 |
} catch (Exception $e) { |
| 872 |
//error_log('PDF repair attempt failed: ' . $e->getMessage()); |
| 873 |
} |
| 874 |
|
| 875 |
// Method 4: Manual PDF structure analysis (basic page count) |
| 876 |
try { |
| 877 |
$page_count = $this->mxchat_manual_pdf_page_count($pdf_path); |
| 878 |
if ($page_count > 0) { |
| 879 |
//error_log('PDF page count determined manually: ' . $page_count . ' pages'); |
| 880 |
return $page_count; |
| 881 |
} |
| 882 |
} catch (Exception $e) { |
| 883 |
//error_log('Manual PDF analysis failed: ' . $e->getMessage()); |
| 884 |
} |
| 885 |
|
| 886 |
//error_log('All PDF parsing methods failed for: ' . $pdf_path); |
| 887 |
return false; |
| 888 |
} |
| 889 |
|
| 890 |
/** |
| 891 |
* Check if shell_exec is disabled |
| 892 |
*/ |
| 893 |
private function mxchat_is_shell_disabled() { |
| 894 |
$disabled = explode(',', ini_get('disable_functions')); |
| 895 |
return in_array('shell_exec', $disabled); |
| 896 |
} |
| 897 |
|
| 898 |
/** |
| 899 |
* Attempt to repair PDF using basic methods |
| 900 |
*/ |
| 901 |
private function mxchat_attempt_pdf_repair($pdf_path) { |
| 902 |
try { |
| 903 |
$content = file_get_contents($pdf_path); |
| 904 |
if (!$content) { |
| 905 |
return false; |
| 906 |
} |
| 907 |
|
| 908 |
// Check if PDF starts with proper header |
| 909 |
if (substr($content, 0, 4) !== '%PDF') { |
| 910 |
// Try to find PDF header in the content |
| 911 |
$header_pos = strpos($content, '%PDF'); |
| 912 |
if ($header_pos !== false && $header_pos < 1024) { |
| 913 |
// Remove junk before PDF header |
| 914 |
$content = substr($content, $header_pos); |
| 915 |
$repaired_path = $pdf_path . '.repaired'; |
| 916 |
file_put_contents($repaired_path, $content); |
| 917 |
return $repaired_path; |
| 918 |
} |
| 919 |
} |
| 920 |
|
| 921 |
// Check for EOF marker |
| 922 |
$content = rtrim($content); |
| 923 |
if (!preg_match('/%%EOF\s*$/', $content)) { |
| 924 |
// Add EOF marker if missing |
| 925 |
$content .= "\n%%EOF"; |
| 926 |
$repaired_path = $pdf_path . '.repaired'; |
| 927 |
file_put_contents($repaired_path, $content); |
| 928 |
return $repaired_path; |
| 929 |
} |
| 930 |
|
| 931 |
} catch (Exception $e) { |
| 932 |
//error_log('PDF repair error: ' . $e->getMessage()); |
| 933 |
} |
| 934 |
|
| 935 |
return false; |
| 936 |
} |
| 937 |
|
| 938 |
/** |
| 939 |
* Manual PDF page counting by analyzing PDF structure |
| 940 |
*/ |
| 941 |
private function mxchat_manual_pdf_page_count($pdf_path) { |
| 942 |
try { |
| 943 |
$content = file_get_contents($pdf_path); |
| 944 |
if (!$content) { |
| 945 |
return 0; |
| 946 |
} |
| 947 |
|
| 948 |
// Method 1: Count /Type /Page objects |
| 949 |
$page_count = preg_match_all('/\/Type\s*\/Page[^s]/', $content); |
| 950 |
if ($page_count > 0) { |
| 951 |
return $page_count; |
| 952 |
} |
| 953 |
|
| 954 |
// Method 2: Look for /Count in pages object |
| 955 |
if (preg_match('/\/Type\s*\/Pages.*?\/Count\s+(\d+)/', $content, $matches)) { |
| 956 |
return intval($matches[1]); |
| 957 |
} |
| 958 |
|
| 959 |
// Method 3: Count page references |
| 960 |
$page_count = preg_match_all('/\d+\s+0\s+obj\s*<<[^>]*\/Type\s*\/Page/', $content); |
| 961 |
if ($page_count > 0) { |
| 962 |
return $page_count; |
| 963 |
} |
| 964 |
|
| 965 |
} catch (Exception $e) { |
| 966 |
//error_log('Manual PDF analysis error: ' . $e->getMessage()); |
| 967 |
} |
| 968 |
|
| 969 |
return 0; |
| 970 |
} |
| 971 |
|
| 972 |
|
| 973 |
public function mxchat_save_inline_prompt() { |
| 974 |
// DEBUG: Log what we're receiving |
| 975 |
//error_log('=== MXCHAT DEBUG ==='); |
| 976 |
//error_log('POST data: ' . print_r($_POST, true)); |
| 977 |
//error_log('Nonce from POST: ' . ($_POST['_ajax_nonce'] ?? 'NOT FOUND')); |
| 978 |
|
| 979 |
// Check for nonce security |
| 980 |
check_ajax_referer('mxchat_save_inline_nonce', '_ajax_nonce'); |
| 981 |
|
| 982 |
// If we get here, nonce passed |
| 983 |
//error_log('Nonce verification PASSED'); |
| 984 |
|
| 985 |
// Verify permissions |
| 986 |
if (!current_user_can('manage_options')) { |
| 987 |
wp_send_json_error(esc_html__('Permission denied.', 'mxchat')); |
| 988 |
return; |
| 989 |
} |
| 990 |
|
| 991 |
global $wpdb; |
| 992 |
$table_name = $wpdb->prefix . 'mxchat_system_prompt_content'; |
| 993 |
|
| 994 |
// Validate and sanitize input data |
| 995 |
$prompt_id = isset($_POST['id']) ? absint($_POST['id']) : 0; |
| 996 |
$article_content = isset($_POST['article_content']) ? wp_kses_post(wp_unslash($_POST['article_content'])) : ''; |
| 997 |
$article_url = isset($_POST['article_url']) ? esc_url_raw($_POST['article_url']) : ''; |
| 998 |
|
| 999 |
if ($prompt_id > 0 && !empty($article_content)) { |
| 1000 |
// Re-generate the embedding vector for the updated content |
| 1001 |
$embedding_vector = $this->mxchat_generate_embedding($article_content); |
| 1002 |
if (is_array($embedding_vector)) { |
| 1003 |
// Serialize the embedding vector before storing it |
| 1004 |
$embedding_vector_serialized = serialize($embedding_vector); |
| 1005 |
// Update the prompt in the database |
| 1006 |
$updated = $wpdb->update( |
| 1007 |
$table_name, |
| 1008 |
array( |
| 1009 |
'article_content' => $article_content, |
| 1010 |
'embedding_vector' => $embedding_vector_serialized, |
| 1011 |
'source_url' => $article_url, |
| 1012 |
), |
| 1013 |
array('id' => $prompt_id), |
| 1014 |
array('%s', '%s', '%s'), |
| 1015 |
array('%d') |
| 1016 |
); |
| 1017 |
if ($updated !== false) { |
| 1018 |
wp_send_json_success(); |
| 1019 |
} else { |
| 1020 |
MxChat_Admin::mxchat_log_debug('knowledge_error', 'Database update failed when saving knowledge entry'); |
| 1021 |
wp_send_json_error(esc_html__('Database update failed.', 'mxchat')); |
| 1022 |
} |
| 1023 |
} else { |
| 1024 |
MxChat_Admin::mxchat_log_debug('embedding_error', 'Embedding generation failed for knowledge entry'); |
| 1025 |
wp_send_json_error(esc_html__('Embedding generation failed.', 'mxchat')); |
| 1026 |
} |
| 1027 |
} else { |
| 1028 |
wp_send_json_error(esc_html__('Invalid data.', 'mxchat')); |
| 1029 |
} |
| 1030 |
} |
| 1031 |
|
| 1032 |
|
| 1033 |
/** |
| 1034 |
* AJAX: Get full content for editing — reassembles chunks if needed. |
| 1035 |
* Works for both WordPress DB and Pinecone entries. |
| 1036 |
*/ |
| 1037 |
/** |
| 1038 |
* Sanitize a knowledge entry's source_url from an AJAX request WITHOUT destroying |
| 1039 |
* its identity. sanitize_text_field() strips percent-encoded octets (%20, %D7%A9…), |
| 1040 |
* so a percent-encoded URL — every non-ASCII permalink — would md5 to a DIFFERENT |
| 1041 |
* id than the one it was stored under: reads miss the entry and saves write an |
| 1042 |
* orphan copy while the original keeps its stale text. URLs get esc_url_raw |
| 1043 |
* (identity-preserving, matches what import stored); non-URL keys (mxchat://, |
| 1044 |
* _ungrouped_) keep the old sanitizer. |
| 1045 |
*/ |
| 1046 |
private function sanitize_entry_source_url( $raw ) { |
| 1047 |
$raw = trim( (string) $raw ); |
| 1048 |
if ( preg_match( '#^https?://#i', $raw ) ) { |
| 1049 |
return esc_url_raw( $raw ); |
| 1050 |
} |
| 1051 |
return sanitize_text_field( $raw ); |
| 1052 |
} |
| 1053 |
|
| 1054 |
public function ajax_mxchat_get_entry_content() { |
| 1055 |
check_ajax_referer('mxchat_edit_entry_nonce', 'nonce'); |
| 1056 |
|
| 1057 |
if ( ! current_user_can('manage_options') ) { |
| 1058 |
wp_send_json_error( array( 'message' => 'Permission denied.' ) ); |
| 1059 |
} |
| 1060 |
|
| 1061 |
$source_url = $this->sanitize_entry_source_url( isset($_POST['source_url']) ? wp_unslash($_POST['source_url']) : '' ); |
| 1062 |
$entry_id = isset($_POST['entry_id']) ? absint($_POST['entry_id']) : 0; |
| 1063 |
$data_source = isset($_POST['data_source']) ? sanitize_key($_POST['data_source']) : 'wordpress'; |
| 1064 |
$bot_id = isset($_POST['bot_id']) ? sanitize_key($_POST['bot_id']) : 'default'; |
| 1065 |
|
| 1066 |
if ( $data_source === 'pinecone' ) { |
| 1067 |
// Pinecone ids are strings (md5 hashes, manual_* ids) — absint() would |
| 1068 |
// destroy them, so re-read the raw value for this branch only. |
| 1069 |
$vector_id = isset($_POST['entry_id']) ? sanitize_text_field( wp_unslash($_POST['entry_id']) ) : ''; |
| 1070 |
// Pinecone: fetch vectors by source_url, reassemble chunks |
| 1071 |
$content = $this->get_pinecone_entry_content( $source_url, $vector_id, $bot_id ); |
| 1072 |
} else { |
| 1073 |
// WordPress DB |
| 1074 |
$content = $this->get_wordpress_entry_content( $source_url, $entry_id ); |
| 1075 |
} |
| 1076 |
|
| 1077 |
if ( is_wp_error( $content ) ) { |
| 1078 |
wp_send_json_error( array( 'message' => $content->get_error_message() ) ); |
| 1079 |
} |
| 1080 |
|
| 1081 |
wp_send_json_success( $content ); |
| 1082 |
} |
| 1083 |
|
| 1084 |
/** |
| 1085 |
* Get content from WordPress DB — reassembles chunks by source_url. |
| 1086 |
*/ |
| 1087 |
private function get_wordpress_entry_content( $source_url, $entry_id ) { |
| 1088 |
global $wpdb; |
| 1089 |
$table = $wpdb->prefix . 'mxchat_system_prompt_content'; |
| 1090 |
|
| 1091 |
// If we have a source_url, check for chunks |
| 1092 |
if ( ! empty( $source_url ) && strpos( $source_url, 'mxchat://' ) !== 0 ) { |
| 1093 |
$rows = $wpdb->get_results( $wpdb->prepare( |
| 1094 |
"SELECT id, article_content, source_url, content_type FROM {$table} WHERE source_url = %s ORDER BY id ASC", |
| 1095 |
$source_url |
| 1096 |
) ); |
| 1097 |
|
| 1098 |
if ( $rows && count( $rows ) > 1 ) { |
| 1099 |
// Multiple rows = chunked. Reassemble. |
| 1100 |
$chunks = array(); |
| 1101 |
foreach ( $rows as $row ) { |
| 1102 |
$parsed = MxChat_Chunker::parse_stored_chunk( $row->article_content ); |
| 1103 |
$index = isset( $parsed['metadata']['chunk_index'] ) ? intval( $parsed['metadata']['chunk_index'] ) : count( $chunks ); |
| 1104 |
$chunks[ $index ] = $parsed['text']; |
| 1105 |
} |
| 1106 |
ksort( $chunks ); |
| 1107 |
return array( |
| 1108 |
'content' => implode( "\n\n", $chunks ), |
| 1109 |
'source_url' => $source_url, |
| 1110 |
'is_chunked' => true, |
| 1111 |
'chunk_count' => count( $chunks ), |
| 1112 |
'content_type' => $rows[0]->content_type, |
| 1113 |
); |
| 1114 |
} elseif ( $rows && count( $rows ) === 1 ) { |
| 1115 |
$parsed = MxChat_Chunker::parse_stored_chunk( $rows[0]->article_content ); |
| 1116 |
return array( |
| 1117 |
'content' => $parsed['text'], |
| 1118 |
'source_url' => $source_url, |
| 1119 |
'entry_id' => $rows[0]->id, |
| 1120 |
'is_chunked' => false, |
| 1121 |
'content_type' => $rows[0]->content_type, |
| 1122 |
); |
| 1123 |
} |
| 1124 |
} |
| 1125 |
|
| 1126 |
// Fallback: fetch by ID |
| 1127 |
if ( $entry_id > 0 ) { |
| 1128 |
$row = $wpdb->get_row( $wpdb->prepare( |
| 1129 |
"SELECT id, article_content, source_url, content_type FROM {$table} WHERE id = %d", |
| 1130 |
$entry_id |
| 1131 |
) ); |
| 1132 |
if ( $row ) { |
| 1133 |
$parsed = MxChat_Chunker::parse_stored_chunk( $row->article_content ); |
| 1134 |
return array( |
| 1135 |
'content' => $parsed['text'], |
| 1136 |
'source_url' => $row->source_url, |
| 1137 |
'entry_id' => $row->id, |
| 1138 |
'is_chunked' => false, |
| 1139 |
'content_type' => $row->content_type, |
| 1140 |
); |
| 1141 |
} |
| 1142 |
} |
| 1143 |
|
| 1144 |
return new WP_Error( 'not_found', 'Entry not found.' ); |
| 1145 |
} |
| 1146 |
|
| 1147 |
/** |
| 1148 |
* Get content from Pinecone — fetches vectors by source_url, reassembles chunks. |
| 1149 |
*/ |
| 1150 |
private function get_pinecone_entry_content( $source_url, $entry_id, $bot_id ) { |
| 1151 |
if ( ! class_exists('MxChat_Pinecone_Manager') ) { |
| 1152 |
return new WP_Error( 'pinecone_unavailable', 'Pinecone manager not available.' ); |
| 1153 |
} |
| 1154 |
|
| 1155 |
// Get Pinecone config |
| 1156 |
if ( $bot_id === 'default' || ! class_exists('MxChat_Multi_Bot_Manager') ) { |
| 1157 |
$pinecone_options = get_option('mxchat_pinecone_addon_options'); |
| 1158 |
$api_key = $pinecone_options['mxchat_pinecone_api_key'] ?? ''; |
| 1159 |
$host = $pinecone_options['mxchat_pinecone_host'] ?? ''; |
| 1160 |
$namespace = $pinecone_options['mxchat_pinecone_namespace'] ?? ''; |
| 1161 |
} else { |
| 1162 |
$bot_config = apply_filters('mxchat_get_bot_pinecone_config', array(), $bot_id); |
| 1163 |
$api_key = $bot_config['api_key'] ?? ''; |
| 1164 |
$host = $bot_config['host'] ?? ''; |
| 1165 |
$namespace = $bot_config['namespace'] ?? ''; |
| 1166 |
} |
| 1167 |
|
| 1168 |
if ( empty($host) || empty($api_key) ) { |
| 1169 |
return new WP_Error( 'pinecone_config', 'Pinecone not configured.' ); |
| 1170 |
} |
| 1171 |
|
| 1172 |
// Manual entries carry no source_url (their vector id is a minted manual_* string, |
| 1173 |
// not md5 of anything the row can hand us) — fetch the exact vector instead. |
| 1174 |
// '_ungrouped_' is the table view's synthetic display key for such rows. |
| 1175 |
if ( ( empty($source_url) || strpos($source_url, '_ungrouped_') === 0 ) && ! empty($entry_id) && is_string($entry_id) ) { |
| 1176 |
$vector_ids = array( $entry_id ); |
| 1177 |
} else { |
| 1178 |
// List vectors with the source_url prefix |
| 1179 |
$base_id = md5( $source_url ); |
| 1180 |
$vector_ids = array( $base_id ); |
| 1181 |
|
| 1182 |
// Find chunk vectors |
| 1183 |
// NOTE: Pinecone's /vectors/list is a GET endpoint with query parameters. |
| 1184 |
// A POST is answered 200-with-an-empty-body, which reads as "no vectors". |
| 1185 |
$list_url = "https://{$host}/vectors/list"; |
| 1186 |
$list_params = array( 'prefix' => $base_id . '_chunk_', 'limit' => 100 ); |
| 1187 |
if ( ! empty($namespace) ) { |
| 1188 |
$list_params['namespace'] = $namespace; |
| 1189 |
} |
| 1190 |
|
| 1191 |
$list_resp = wp_remote_get( $list_url . '?' . http_build_query( $list_params ), array( |
| 1192 |
'headers' => array( 'Api-Key' => $api_key, 'accept' => 'application/json' ), |
| 1193 |
'timeout' => 15, |
| 1194 |
) ); |
| 1195 |
|
| 1196 |
if ( ! is_wp_error($list_resp) ) { |
| 1197 |
$list_data = json_decode( wp_remote_retrieve_body($list_resp), true ); |
| 1198 |
if ( ! empty($list_data['vectors']) ) { |
| 1199 |
foreach ( $list_data['vectors'] as $v ) { |
| 1200 |
$vector_ids[] = $v['id']; |
| 1201 |
} |
| 1202 |
} |
| 1203 |
} |
| 1204 |
} |
| 1205 |
|
| 1206 |
// Fetch vectors with metadata |
| 1207 |
// NOTE: /vectors/fetch is a GET endpoint too, and Pinecone expects the ids |
| 1208 |
// repeated (ids=a&ids=b) — http_build_query would emit ids[0]=a, so build |
| 1209 |
// the query string explicitly. |
| 1210 |
$fetch_query = array(); |
| 1211 |
foreach ( $vector_ids as $fetch_vid ) { |
| 1212 |
$fetch_query[] = 'ids=' . rawurlencode( $fetch_vid ); |
| 1213 |
} |
| 1214 |
if ( ! empty($namespace) ) { |
| 1215 |
$fetch_query[] = 'namespace=' . rawurlencode( $namespace ); |
| 1216 |
} |
| 1217 |
|
| 1218 |
$fetch_resp = wp_remote_get( "https://{$host}/vectors/fetch?" . implode( '&', $fetch_query ), array( |
| 1219 |
'headers' => array( 'Api-Key' => $api_key, 'accept' => 'application/json' ), |
| 1220 |
'timeout' => 15, |
| 1221 |
) ); |
| 1222 |
|
| 1223 |
if ( is_wp_error($fetch_resp) ) { |
| 1224 |
return new WP_Error( 'pinecone_fetch', 'Failed to fetch from Pinecone.' ); |
| 1225 |
} |
| 1226 |
|
| 1227 |
$fetch_data = json_decode( wp_remote_retrieve_body($fetch_resp), true ); |
| 1228 |
$vectors = $fetch_data['vectors'] ?? array(); |
| 1229 |
|
| 1230 |
if ( empty($vectors) ) { |
| 1231 |
return new WP_Error( 'not_found', 'Entry not found in Pinecone.' ); |
| 1232 |
} |
| 1233 |
|
| 1234 |
// Reassemble chunks |
| 1235 |
$chunks = array(); |
| 1236 |
$content_type = 'content'; |
| 1237 |
foreach ( $vectors as $vid => $vector ) { |
| 1238 |
$meta = $vector['metadata'] ?? array(); |
| 1239 |
$text = $meta['text'] ?? ''; |
| 1240 |
$index = $meta['chunk_index'] ?? 0; |
| 1241 |
$content_type = $meta['type'] ?? 'content'; |
| 1242 |
$chunks[ intval($index) ] = $text; |
| 1243 |
} |
| 1244 |
ksort( $chunks ); |
| 1245 |
|
| 1246 |
return array( |
| 1247 |
'content' => implode( "\n\n", $chunks ), |
| 1248 |
'source_url' => $source_url, |
| 1249 |
'is_chunked' => count($chunks) > 1, |
| 1250 |
'chunk_count' => count($chunks), |
| 1251 |
'content_type' => $content_type, |
| 1252 |
); |
| 1253 |
} |
| 1254 |
|
| 1255 |
/** |
| 1256 |
* AJAX: Inspect a knowledge entry — returns the per-chunk stored text + metadata |
| 1257 |
* WITHOUT collapsing it, so a site owner can see exactly what was indexed for an |
| 1258 |
* entry (plan-mxchat-20260628-d8cb4b). READ-ONLY: never re-embeds or mutates. |
| 1259 |
*/ |
| 1260 |
public function ajax_mxchat_inspect_entry() { |
| 1261 |
check_ajax_referer('mxchat_inspect_entry_nonce', 'nonce'); |
| 1262 |
|
| 1263 |
if ( ! current_user_can('manage_options') ) { |
| 1264 |
wp_send_json_error( array( 'message' => esc_html__('Permission denied.', 'mxchat') ) ); |
| 1265 |
} |
| 1266 |
|
| 1267 |
$source_url = isset($_POST['source_url']) ? sanitize_text_field( wp_unslash($_POST['source_url']) ) : ''; |
| 1268 |
$entry_id = isset($_POST['entry_id']) ? absint($_POST['entry_id']) : 0; |
| 1269 |
$data_source = isset($_POST['data_source']) ? sanitize_key($_POST['data_source']) : 'wordpress'; |
| 1270 |
$bot_id = isset($_POST['bot_id']) ? sanitize_key($_POST['bot_id']) : 'default'; |
| 1271 |
|
| 1272 |
if ( $data_source === 'pinecone' ) { |
| 1273 |
$result = $this->inspect_pinecone_entry( $source_url, $entry_id, $bot_id ); |
| 1274 |
} else { |
| 1275 |
$result = $this->inspect_wordpress_entry( $source_url, $entry_id ); |
| 1276 |
} |
| 1277 |
|
| 1278 |
if ( is_wp_error( $result ) ) { |
| 1279 |
wp_send_json_error( array( 'message' => $result->get_error_message() ) ); |
| 1280 |
} |
| 1281 |
|
| 1282 |
wp_send_json_success( $result ); |
| 1283 |
} |
| 1284 |
|
| 1285 |
/** |
| 1286 |
* Read-only inspector for WordPress-DB entries. Mirrors get_wordpress_entry_content() |
| 1287 |
* but returns each STORED chunk's exact text + length (no implode), plus the assembled |
| 1288 |
* embedded text. This shows what is actually in the index, not a re-derivation from the post. |
| 1289 |
*/ |
| 1290 |
private function inspect_wordpress_entry( $source_url, $entry_id ) { |
| 1291 |
global $wpdb; |
| 1292 |
$table = $wpdb->prefix . 'mxchat_system_prompt_content'; |
| 1293 |
|
| 1294 |
$rows = array(); |
| 1295 |
|
| 1296 |
// Group by the real stored source_url — this INCLUDES "mxchat://" manual |
| 1297 |
// Direct Content entries (the spec's manual-entry case), which share one |
| 1298 |
// source_url across their chunk rows. Only the synthetic "_ungrouped_<id>" |
| 1299 |
// display key (invented by the table view for rows with no source_url) is |
| 1300 |
// excluded; those fall through to the entry_id lookup below. |
| 1301 |
if ( ! empty( $source_url ) && strpos( $source_url, '_ungrouped_' ) !== 0 ) { |
| 1302 |
$rows = $wpdb->get_results( $wpdb->prepare( |
| 1303 |
"SELECT id, article_content, source_url, content_type FROM {$table} WHERE source_url = %s ORDER BY id ASC", |
| 1304 |
$source_url |
| 1305 |
) ); |
| 1306 |
} |
| 1307 |
|
| 1308 |
// Fallback / manual "Direct Content" entries: fetch the single row by id. |
| 1309 |
if ( empty( $rows ) && $entry_id > 0 ) { |
| 1310 |
$row = $wpdb->get_row( $wpdb->prepare( |
| 1311 |
"SELECT id, article_content, source_url, content_type FROM {$table} WHERE id = %d", |
| 1312 |
$entry_id |
| 1313 |
) ); |
| 1314 |
if ( $row ) { |
| 1315 |
$rows = array( $row ); |
| 1316 |
} |
| 1317 |
} |
| 1318 |
|
| 1319 |
if ( empty( $rows ) ) { |
| 1320 |
return new WP_Error( 'not_found', esc_html__('Entry not found in the local knowledge database.', 'mxchat') ); |
| 1321 |
} |
| 1322 |
|
| 1323 |
$chunks = array(); |
| 1324 |
$content_type = ''; |
| 1325 |
foreach ( $rows as $row ) { |
| 1326 |
$parsed = MxChat_Chunker::parse_stored_chunk( $row->article_content ); |
| 1327 |
$text = isset( $parsed['text'] ) ? $parsed['text'] : ''; |
| 1328 |
$index = isset( $parsed['metadata']['chunk_index'] ) ? intval( $parsed['metadata']['chunk_index'] ) : count( $chunks ); |
| 1329 |
$content_type = $row->content_type; |
| 1330 |
$chunks[] = array( |
| 1331 |
'index' => $index, |
| 1332 |
'text' => $text, |
| 1333 |
'length' => function_exists('mb_strlen') ? mb_strlen( $text ) : strlen( $text ), |
| 1334 |
'row_id' => intval( $row->id ), |
| 1335 |
); |
| 1336 |
} |
| 1337 |
|
| 1338 |
usort( $chunks, function( $a, $b ) { return $a['index'] - $b['index']; } ); |
| 1339 |
|
| 1340 |
$assembled = implode( "\n\n", wp_list_pluck( $chunks, 'text' ) ); |
| 1341 |
|
| 1342 |
return array( |
| 1343 |
'store' => 'wordpress', |
| 1344 |
'source_url' => $source_url, |
| 1345 |
'content_type' => $content_type, |
| 1346 |
'is_chunked' => count( $chunks ) > 1, |
| 1347 |
'chunk_count' => count( $chunks ), |
| 1348 |
'assembled' => $assembled, |
| 1349 |
'assembled_length' => function_exists('mb_strlen') ? mb_strlen( $assembled ) : strlen( $assembled ), |
| 1350 |
'chunks' => array_values( $chunks ), |
| 1351 |
// WP-DB storage carries no separate vector metadata; surface that fact |
| 1352 |
// rather than letting the owner guess (the spec's taxonomy question). |
| 1353 |
'metadata' => array(), |
| 1354 |
'metadata_note' => esc_html__('Stored in the local WordPress database. Only the assembled text shown here is embedded — there are no separate vector metadata fields (e.g. taxonomy terms are not stored unless they were injected into the text itself).', 'mxchat'), |
| 1355 |
); |
| 1356 |
} |
| 1357 |
|
| 1358 |
/** |
| 1359 |
* Read-only inspector for Pinecone entries. Mirrors get_pinecone_entry_content() |
| 1360 |
* but keeps each vector's text + metadata instead of imploding, so the owner can |
| 1361 |
* confirm exactly which metadata fields (text/source_url/type/last_updated/created_at/bot_id) |
| 1362 |
* are present per chunk. READ-ONLY. |
| 1363 |
*/ |
| 1364 |
private function inspect_pinecone_entry( $source_url, $entry_id, $bot_id ) { |
| 1365 |
if ( ! class_exists('MxChat_Pinecone_Manager') ) { |
| 1366 |
return new WP_Error( 'pinecone_unavailable', esc_html__('Pinecone manager not available.', 'mxchat') ); |
| 1367 |
} |
| 1368 |
|
| 1369 |
if ( $bot_id === 'default' || ! class_exists('MxChat_Multi_Bot_Manager') ) { |
| 1370 |
$pinecone_options = get_option('mxchat_pinecone_addon_options'); |
| 1371 |
$api_key = $pinecone_options['mxchat_pinecone_api_key'] ?? ''; |
| 1372 |
$host = $pinecone_options['mxchat_pinecone_host'] ?? ''; |
| 1373 |
$namespace = $pinecone_options['mxchat_pinecone_namespace'] ?? ''; |
| 1374 |
} else { |
| 1375 |
$bot_config = apply_filters('mxchat_get_bot_pinecone_config', array(), $bot_id); |
| 1376 |
$api_key = $bot_config['api_key'] ?? ''; |
| 1377 |
$host = $bot_config['host'] ?? ''; |
| 1378 |
$namespace = $bot_config['namespace'] ?? ''; |
| 1379 |
} |
| 1380 |
|
| 1381 |
if ( empty($host) || empty($api_key) ) { |
| 1382 |
return new WP_Error( 'pinecone_config', esc_html__('Pinecone not configured.', 'mxchat') ); |
| 1383 |
} |
| 1384 |
|
| 1385 |
$base_id = md5( $source_url ); |
| 1386 |
$vector_ids = array( $base_id ); |
| 1387 |
|
| 1388 |
// NOTE: Pinecone's /vectors/list is a GET endpoint with query parameters. |
| 1389 |
// A POST is answered 200-with-an-empty-body, which reads as "no vectors". |
| 1390 |
$list_url = "https://{$host}/vectors/list"; |
| 1391 |
$list_params = array( 'prefix' => $base_id . '_chunk_', 'limit' => 100 ); |
| 1392 |
if ( ! empty($namespace) ) { |
| 1393 |
$list_params['namespace'] = $namespace; |
| 1394 |
} |
| 1395 |
|
| 1396 |
$list_resp = wp_remote_get( $list_url . '?' . http_build_query( $list_params ), array( |
| 1397 |
'headers' => array( 'Api-Key' => $api_key, 'accept' => 'application/json' ), |
| 1398 |
'timeout' => 15, |
| 1399 |
) ); |
| 1400 |
|
| 1401 |
if ( ! is_wp_error($list_resp) ) { |
| 1402 |
$list_data = json_decode( wp_remote_retrieve_body($list_resp), true ); |
| 1403 |
if ( ! empty($list_data['vectors']) ) { |
| 1404 |
foreach ( $list_data['vectors'] as $v ) { |
| 1405 |
$vector_ids[] = $v['id']; |
| 1406 |
} |
| 1407 |
} |
| 1408 |
} |
| 1409 |
|
| 1410 |
// NOTE: /vectors/fetch is a GET endpoint too, and Pinecone expects the ids |
| 1411 |
// repeated (ids=a&ids=b) — http_build_query would emit ids[0]=a, so build |
| 1412 |
// the query string explicitly. |
| 1413 |
$fetch_query = array(); |
| 1414 |
foreach ( $vector_ids as $fetch_vid ) { |
| 1415 |
$fetch_query[] = 'ids=' . rawurlencode( $fetch_vid ); |
| 1416 |
} |
| 1417 |
if ( ! empty($namespace) ) { |
| 1418 |
$fetch_query[] = 'namespace=' . rawurlencode( $namespace ); |
| 1419 |
} |
| 1420 |
|
| 1421 |
$fetch_resp = wp_remote_get( "https://{$host}/vectors/fetch?" . implode( '&', $fetch_query ), array( |
| 1422 |
'headers' => array( 'Api-Key' => $api_key, 'accept' => 'application/json' ), |
| 1423 |
'timeout' => 15, |
| 1424 |
) ); |
| 1425 |
|
| 1426 |
if ( is_wp_error($fetch_resp) ) { |
| 1427 |
return new WP_Error( 'pinecone_fetch', esc_html__('Failed to fetch from Pinecone.', 'mxchat') ); |
| 1428 |
} |
| 1429 |
|
| 1430 |
$fetch_data = json_decode( wp_remote_retrieve_body($fetch_resp), true ); |
| 1431 |
$vectors = $fetch_data['vectors'] ?? array(); |
| 1432 |
|
| 1433 |
if ( empty($vectors) ) { |
| 1434 |
return new WP_Error( 'not_found', esc_html__('Entry not found in Pinecone.', 'mxchat') ); |
| 1435 |
} |
| 1436 |
|
| 1437 |
// Whitelisted metadata fields the spec calls out — shown so devs can confirm |
| 1438 |
// what is (and is NOT) stored per vector. |
| 1439 |
$meta_fields = array( 'text', 'source_url', 'type', 'last_updated', 'created_at', 'bot_id', 'chunk_index', 'total_chunks' ); |
| 1440 |
$chunks = array(); |
| 1441 |
$content_type = ''; |
| 1442 |
foreach ( $vectors as $vid => $vector ) { |
| 1443 |
$meta = isset($vector['metadata']) && is_array($vector['metadata']) ? $vector['metadata'] : array(); |
| 1444 |
$text = $meta['text'] ?? ''; |
| 1445 |
$index = isset($meta['chunk_index']) ? intval($meta['chunk_index']) : count($chunks); |
| 1446 |
$content_type = $meta['type'] ?? $content_type; |
| 1447 |
|
| 1448 |
$clean_meta = array(); |
| 1449 |
foreach ( $meta_fields as $field ) { |
| 1450 |
if ( array_key_exists( $field, $meta ) && $field !== 'text' ) { |
| 1451 |
$clean_meta[ $field ] = is_scalar( $meta[ $field ] ) ? (string) $meta[ $field ] : wp_json_encode( $meta[ $field ] ); |
| 1452 |
} |
| 1453 |
} |
| 1454 |
|
| 1455 |
$chunks[] = array( |
| 1456 |
'index' => $index, |
| 1457 |
'text' => $text, |
| 1458 |
'length' => function_exists('mb_strlen') ? mb_strlen( $text ) : strlen( $text ), |
| 1459 |
'vector_id' => (string) $vid, |
| 1460 |
'metadata' => $clean_meta, |
| 1461 |
); |
| 1462 |
} |
| 1463 |
|
| 1464 |
usort( $chunks, function( $a, $b ) { return $a['index'] - $b['index']; } ); |
| 1465 |
|
| 1466 |
$assembled = implode( "\n\n", wp_list_pluck( $chunks, 'text' ) ); |
| 1467 |
|
| 1468 |
return array( |
| 1469 |
'store' => 'pinecone', |
| 1470 |
'source_url' => $source_url, |
| 1471 |
'content_type' => $content_type, |
| 1472 |
'is_chunked' => count( $chunks ) > 1, |
| 1473 |
'chunk_count' => count( $chunks ), |
| 1474 |
'assembled' => $assembled, |
| 1475 |
'assembled_length' => function_exists('mb_strlen') ? mb_strlen( $assembled ) : strlen( $assembled ), |
| 1476 |
'chunks' => array_values( $chunks ), |
| 1477 |
'metadata' => array(), |
| 1478 |
'metadata_note' => esc_html__('Stored in Pinecone. Each chunk above lists the vector metadata fields actually present — if a field you expect (such as taxonomy terms) is missing here, it was not stored as metadata and is only searchable if it appears in the embedded text.', 'mxchat'), |
| 1479 |
); |
| 1480 |
} |
| 1481 |
|
| 1482 |
/** |
| 1483 |
* AJAX: Save edited content — re-chunks and re-embeds as needed. |
| 1484 |
* Works for both WordPress DB and Pinecone entries. |
| 1485 |
*/ |
| 1486 |
public function ajax_mxchat_save_entry_content() { |
| 1487 |
check_ajax_referer('mxchat_edit_entry_nonce', 'nonce'); |
| 1488 |
|
| 1489 |
if ( ! current_user_can('manage_options') ) { |
| 1490 |
wp_send_json_error( array( 'message' => 'Permission denied.' ) ); |
| 1491 |
} |
| 1492 |
|
| 1493 |
$source_url = $this->sanitize_entry_source_url( isset($_POST['source_url']) ? wp_unslash($_POST['source_url']) : '' ); |
| 1494 |
$entry_id = isset($_POST['entry_id']) ? absint($_POST['entry_id']) : 0; |
| 1495 |
$content = isset($_POST['content']) ? wp_kses_post( wp_unslash($_POST['content']) ) : ''; |
| 1496 |
$data_source = isset($_POST['data_source']) ? sanitize_key($_POST['data_source']) : 'wordpress'; |
| 1497 |
$bot_id = isset($_POST['bot_id']) ? sanitize_key($_POST['bot_id']) : 'default'; |
| 1498 |
$content_type = isset($_POST['content_type']) ? sanitize_key($_POST['content_type']) : 'content'; |
| 1499 |
|
| 1500 |
if ( empty($content) ) { |
| 1501 |
wp_send_json_error( array( 'message' => 'Content cannot be empty.' ) ); |
| 1502 |
} |
| 1503 |
|
| 1504 |
// Get the embedding API key |
| 1505 |
$options = get_option('mxchat_options', array()); |
| 1506 |
$api_key = ''; |
| 1507 |
|
| 1508 |
if ( $bot_id !== 'default' && class_exists('MxChat_Multi_Bot_Manager') ) { |
| 1509 |
$bot_options = apply_filters('mxchat_get_bot_options', array(), $bot_id); |
| 1510 |
$api_key = $bot_options['api_key'] ?? ''; |
| 1511 |
} |
| 1512 |
if ( empty($api_key) ) { |
| 1513 |
$api_key = $options['api_key'] ?? ''; |
| 1514 |
} |
| 1515 |
|
| 1516 |
if ( $data_source === 'pinecone' ) { |
| 1517 |
// Pinecone branch. The WP-DB manual-entry delete below must never run here: |
| 1518 |
// Pinecone ids are strings, and absint() on a digit-leading md5 hash would |
| 1519 |
// yield a real (unrelated) WP row id. |
| 1520 |
$raw_vector_id = isset($_POST['entry_id']) ? sanitize_text_field( wp_unslash($_POST['entry_id']) ) : ''; |
| 1521 |
$is_manual_single = empty($source_url) || strpos($source_url, '_ungrouped_') === 0; |
| 1522 |
$is_manual_chunked = strpos($source_url, 'mxchat://') === 0; |
| 1523 |
|
| 1524 |
if ( $is_manual_single || $is_manual_chunked ) { |
| 1525 |
// Manual content: remove the old vectors first, then store as fresh manual |
| 1526 |
// content — submit_content_to_db mints a new unique identity (manual_* id |
| 1527 |
// for a single vector, an mxchat:// chunk prefix if it now chunks). |
| 1528 |
if ( $is_manual_chunked ) { |
| 1529 |
// Minted identity: base + chunk vectors share the md5(mxchat://...) prefix. |
| 1530 |
MxChat_Utils::delete_chunks_for_url( $source_url, $bot_id ); |
| 1531 |
} elseif ( ! empty($raw_vector_id) && class_exists('MxChat_Pinecone_Manager') ) { |
| 1532 |
$pinecone_manager = MxChat_Pinecone_Manager::get_instance(); |
| 1533 |
$pinecone_options = $pinecone_manager->mxchat_get_bot_pinecone_options( $bot_id ); |
| 1534 |
if ( ! empty($pinecone_options['mxchat_pinecone_api_key']) && ! empty($pinecone_options['mxchat_pinecone_host']) ) { |
| 1535 |
$pinecone_manager->mxchat_delete_from_pinecone_by_vector_id( |
| 1536 |
$raw_vector_id, |
| 1537 |
$pinecone_options['mxchat_pinecone_api_key'], |
| 1538 |
$pinecone_options['mxchat_pinecone_host'], |
| 1539 |
$pinecone_options['mxchat_pinecone_namespace'] ?? '' |
| 1540 |
); |
| 1541 |
} |
| 1542 |
} |
| 1543 |
$result = MxChat_Utils::submit_content_to_db( $content, '', $api_key, null, $bot_id, $content_type ); |
| 1544 |
} else { |
| 1545 |
// URL-sourced entry: identity is md5(source_url). submit_content_to_db |
| 1546 |
// handles delete-old-chunks → re-chunk → re-embed → store, and sweeps |
| 1547 |
// stale chunk vectors when the content now fits in a single vector. |
| 1548 |
$result = MxChat_Utils::submit_content_to_db( $content, $source_url, $api_key, md5($source_url), $bot_id, $content_type ); |
| 1549 |
} |
| 1550 |
} else { |
| 1551 |
global $wpdb; |
| 1552 |
$table = $wpdb->prefix . 'mxchat_system_prompt_content'; |
| 1553 |
|
| 1554 |
// If source_url is empty but we have an entry_id, look it up |
| 1555 |
if ( empty($source_url) && $entry_id > 0 ) { |
| 1556 |
$row = $wpdb->get_row( $wpdb->prepare( "SELECT source_url FROM {$table} WHERE id = %d", $entry_id ) ); |
| 1557 |
if ( $row && ! empty($row->source_url) ) { |
| 1558 |
$source_url = $row->source_url; |
| 1559 |
} |
| 1560 |
} |
| 1561 |
|
| 1562 |
// For manual entries (no source_url or mxchat:// prefix), delete the old entry by ID first |
| 1563 |
// so submit_content_to_db creates a replacement instead of a duplicate |
| 1564 |
// Also treat legacy mxchat.ai source URLs as manual — old bug assigned the site URL to manual entries |
| 1565 |
$is_legacy_manual = !empty($source_url) && strpos($source_url, 'mxchat.ai') !== false && strpos($source_url, 'mxchat://') !== 0; |
| 1566 |
if ( $entry_id > 0 && (empty($source_url) || strpos($source_url, 'mxchat://') === 0 || $is_legacy_manual) ) { |
| 1567 |
$wpdb->delete( $table, array( 'id' => $entry_id ), array( '%d' ) ); |
| 1568 |
// Clear legacy URL so submit_content_to_db generates a unique mxchat:// identifier |
| 1569 |
// instead of reusing the shared URL (which would mass-delete other entries with the same URL) |
| 1570 |
if ( $is_legacy_manual ) { |
| 1571 |
$source_url = ''; |
| 1572 |
} |
| 1573 |
} |
| 1574 |
|
| 1575 |
// Use the existing submit_content_to_db which handles chunking, Pinecone, and WP DB |
| 1576 |
$vector_id = ! empty($source_url) ? md5($source_url) : md5('mxchat_manual_' . $entry_id); |
| 1577 |
|
| 1578 |
// submit_content_to_db already handles: delete old chunks → re-chunk → re-embed → store |
| 1579 |
$result = MxChat_Utils::submit_content_to_db( $content, $source_url, $api_key, $vector_id, $bot_id, $content_type ); |
| 1580 |
} |
| 1581 |
|
| 1582 |
if ( is_wp_error($result) ) { |
| 1583 |
wp_send_json_error( array( 'message' => $result->get_error_message() ) ); |
| 1584 |
} |
| 1585 |
|
| 1586 |
wp_send_json_success( array( 'message' => 'Content saved and re-embedded successfully.' ) ); |
| 1587 |
} |
| 1588 |
|
| 1589 |
public function mxchat_get_pdf_processing_status($pdf_url) { |
| 1590 |
$pdf_url = esc_url_raw($pdf_url); |
| 1591 |
$status = get_transient(sanitize_key('mxchat_pdf_status_' . md5($pdf_url))); |
| 1592 |
|
| 1593 |
if (!$status || !is_array($status)) { |
| 1594 |
return false; |
| 1595 |
} |
| 1596 |
|
| 1597 |
// Check for stalled processing (no updates for 5 minutes) |
| 1598 |
if ($status['status'] === 'processing' && (time() - absint($status['last_update'])) > 300) { |
| 1599 |
$status['status'] = 'error'; |
| 1600 |
$status['error'] = __('PDF processing appears to be stalled. No updates for over 5 minutes.', 'mxchat'); |
| 1601 |
|
| 1602 |
// Save the updated status |
| 1603 |
set_transient( |
| 1604 |
sanitize_key('mxchat_pdf_status_' . md5($pdf_url)), |
| 1605 |
array_map('sanitize_text_field', $status), |
| 1606 |
DAY_IN_SECONDS |
| 1607 |
); |
| 1608 |
} |
| 1609 |
|
| 1610 |
$result = array( |
| 1611 |
'total_pages' => absint($status['total_pages']), |
| 1612 |
'processed_pages' => absint($status['processed_pages']), |
| 1613 |
'failed_pages' => absint($status['failed_pages'] ?? 0), |
| 1614 |
'percentage' => ($status['total_pages'] > 0) |
| 1615 |
? round((absint($status['processed_pages']) / absint($status['total_pages'])) * 100) |
| 1616 |
: 0, |
| 1617 |
'status' => sanitize_text_field($status['status']), |
| 1618 |
'last_update' => human_time_diff(absint($status['last_update']), time()) . ' ' . esc_html__('ago', 'mxchat'), |
| 1619 |
'failed_pages_list' => isset($status['failed_pages_list']) ? $status['failed_pages_list'] : array(), |
| 1620 |
'completion_summary' => isset($status['completion_summary']) ? $status['completion_summary'] : null |
| 1621 |
); |
| 1622 |
|
| 1623 |
// Add error message if present |
| 1624 |
if (isset($status['error']) && !empty($status['error'])) { |
| 1625 |
$result['error'] = sanitize_text_field($status['error']); |
| 1626 |
} |
| 1627 |
|
| 1628 |
return $result; |
| 1629 |
} |
| 1630 |
|
| 1631 |
|
| 1632 |
public function mxchat_handle_sitemap_submission() { |
| 1633 |
// Check if the form was submitted and verify permissions |
| 1634 |
if (!isset($_POST['submit_sitemap']) || !current_user_can('manage_options')) { |
| 1635 |
wp_die(esc_html__('Unauthorized access', 'mxchat')); |
| 1636 |
} |
| 1637 |
|
| 1638 |
// Verify nonce |
| 1639 |
check_admin_referer('mxchat_submit_sitemap_action', 'mxchat_submit_sitemap_nonce'); |
| 1640 |
|
| 1641 |
// Validate URL |
| 1642 |
if (!isset($_POST['sitemap_url']) || empty($_POST['sitemap_url'])) { |
| 1643 |
set_transient('mxchat_admin_notice_error', |
| 1644 |
esc_html__('Please provide a valid URL.', 'mxchat'), |
| 1645 |
30 |
| 1646 |
); |
| 1647 |
wp_safe_redirect(esc_url(admin_url('admin.php?page=mxchat-prompts'))); |
| 1648 |
exit; |
| 1649 |
} |
| 1650 |
|
| 1651 |
$submitted_url = esc_url_raw($_POST['sitemap_url']); |
| 1652 |
|
| 1653 |
// Convert Google Drive sharing URLs to direct download URLs |
| 1654 |
if ( strpos($submitted_url, 'drive.google.com') !== false ) { |
| 1655 |
$file_id = ''; |
| 1656 |
if ( preg_match('/[?&]id=([a-zA-Z0-9_-]+)/', $submitted_url, $m) ) { |
| 1657 |
$file_id = $m[1]; |
| 1658 |
} elseif ( preg_match('#/file/d/([a-zA-Z0-9_-]+)#', $submitted_url, $m) ) { |
| 1659 |
$file_id = $m[1]; |
| 1660 |
} |
| 1661 |
if ( ! empty($file_id) ) { |
| 1662 |
$submitted_url = 'https://drive.google.com/uc?export=download&id=' . $file_id; |
| 1663 |
} |
| 1664 |
} |
| 1665 |
|
| 1666 |
// Get bot_id from form submission |
| 1667 |
$bot_id = isset($_POST['bot_id']) ? sanitize_key($_POST['bot_id']) : 'default'; |
| 1668 |
|
| 1669 |
// Get bot-specific options and validate the embedding decision — |
| 1670 |
// custom-provider-aware (plan cbd5fd). |
| 1671 |
$bot_options = $this->get_bot_options($bot_id); |
| 1672 |
$options = !empty($bot_options) ? $bot_options : get_option('mxchat_options'); |
| 1673 |
|
| 1674 |
$preflight = MxChat_Utils::embedding_preflight($options); |
| 1675 |
if (!$preflight['ok']) { |
| 1676 |
set_transient('mxchat_admin_notice_error', esc_html($preflight['reason']), 30); |
| 1677 |
wp_safe_redirect(esc_url(admin_url('admin.php?page=mxchat-prompts'))); |
| 1678 |
exit; |
| 1679 |
} |
| 1680 |
$api_key = $preflight['api_key']; |
| 1681 |
|
| 1682 |
// Fetch URL — send an honest, versioned MXChat crawler UA (not a spoofed |
| 1683 |
// browser). Stale browser UAs are exactly what WAFs like SiteGround's |
| 1684 |
// ModSecurity flag as scrapers, 403-ing the fetch (including PDFs served |
| 1685 |
// from the site's own media library, which route through this same call). |
| 1686 |
// See mxchat_ingest_user_agent(). Accept is kept for content negotiation; |
| 1687 |
// the browser-only Accept-Language fingerprint is dropped so it stays |
| 1688 |
// coherent with a bot identity. |
| 1689 |
$response = wp_remote_get($submitted_url, array( |
| 1690 |
'timeout' => 30, |
| 1691 |
'sslverify' => false, |
| 1692 |
'user-agent' => mxchat_ingest_user_agent(), |
| 1693 |
'headers' => array( |
| 1694 |
'Accept' => 'text/html,application/xhtml+xml,application/xml;q=0.9,*/*;q=0.8', |
| 1695 |
), |
| 1696 |
)); |
| 1697 |
|
| 1698 |
if (is_wp_error($response) || wp_remote_retrieve_response_code($response) !== 200) { |
| 1699 |
$error_message = is_wp_error($response) ? $response->get_error_message() : 'HTTP Status: ' . wp_remote_retrieve_response_code($response); |
| 1700 |
set_transient('mxchat_admin_notice_error', |
| 1701 |
sprintf( |
| 1702 |
esc_html__('Failed to fetch the URL: %s', 'mxchat'), |
| 1703 |
esc_html($error_message) |
| 1704 |
), |
| 1705 |
30 |
| 1706 |
); |
| 1707 |
wp_safe_redirect(esc_url(admin_url('admin.php?page=mxchat-prompts'))); |
| 1708 |
exit; |
| 1709 |
} |
| 1710 |
|
| 1711 |
$content_type = wp_remote_retrieve_header($response, 'content-type'); |
| 1712 |
$body_content = wp_remote_retrieve_body($response); |
| 1713 |
|
| 1714 |
if (empty($body_content)) { |
| 1715 |
set_transient('mxchat_admin_notice_error', |
| 1716 |
esc_html__('Empty response received from URL.', 'mxchat'), |
| 1717 |
30 |
| 1718 |
); |
| 1719 |
wp_safe_redirect(esc_url(admin_url('admin.php?page=mxchat-prompts'))); |
| 1720 |
exit; |
| 1721 |
} |
| 1722 |
|
| 1723 |
// Handle PDF URL |
| 1724 |
if ($this->mxchat_is_pdf_url($submitted_url, $response)) { |
| 1725 |
$result = $this->mxchat_handle_pdf_for_knowledge_base($submitted_url, $response, $bot_id); |
| 1726 |
|
| 1727 |
if ($result === 'queued') { |
| 1728 |
set_transient('mxchat_admin_notice_success', |
| 1729 |
esc_html__('PDF queued for processing. Processing will start automatically.', 'mxchat'), |
| 1730 |
30 |
| 1731 |
); |
| 1732 |
} else { |
| 1733 |
set_transient('mxchat_admin_notice_error', |
| 1734 |
esc_html__('Failed to queue PDF processing: ', 'mxchat') . esc_html($result), |
| 1735 |
30 |
| 1736 |
); |
| 1737 |
} |
| 1738 |
|
| 1739 |
wp_safe_redirect(esc_url(admin_url('admin.php?page=mxchat-prompts'))); |
| 1740 |
exit; |
| 1741 |
} |
| 1742 |
|
| 1743 |
// Handle Sitemap XML |
| 1744 |
if (strpos($content_type, 'xml') !== false || strpos($body_content, '<urlset') !== false) { |
| 1745 |
libxml_use_internal_errors(true); |
| 1746 |
$xml = simplexml_load_string($body_content); |
| 1747 |
$xml_errors = libxml_get_errors(); |
| 1748 |
libxml_clear_errors(); |
| 1749 |
|
| 1750 |
if ($xml === false || !empty($xml_errors)) { |
| 1751 |
set_transient('mxchat_admin_notice_error', |
| 1752 |
esc_html__('Invalid sitemap XML. Please provide a valid sitemap.', 'mxchat'), |
| 1753 |
30 |
| 1754 |
); |
| 1755 |
wp_safe_redirect(esc_url(admin_url('admin.php?page=mxchat-prompts'))); |
| 1756 |
exit; |
| 1757 |
} |
| 1758 |
|
| 1759 |
$result = $this->mxchat_handle_sitemap_for_knowledge_base($xml, $submitted_url, $bot_id); |
| 1760 |
|
| 1761 |
if ($result === 'queued') { |
| 1762 |
set_transient('mxchat_admin_notice_success', |
| 1763 |
esc_html__('Sitemap queued for processing. Processing will start automatically.', 'mxchat'), |
| 1764 |
30 |
| 1765 |
); |
| 1766 |
} else { |
| 1767 |
// Surface the reason the handler already computed (embedding pre-flight, |
| 1768 |
// empty sitemap, queue failure). The old message pointed at the status |
| 1769 |
// area, which is empty on this path — nothing was ever queued. |
| 1770 |
if (is_string($result) && $result !== '') { |
| 1771 |
set_transient('mxchat_admin_notice_error', |
| 1772 |
esc_html__('Failed to queue sitemap processing: ', 'mxchat') . esc_html($result), |
| 1773 |
30 |
| 1774 |
); |
| 1775 |
} else { |
| 1776 |
set_transient('mxchat_admin_notice_error', |
| 1777 |
esc_html__('Failed to queue sitemap processing.', 'mxchat'), |
| 1778 |
30 |
| 1779 |
); |
| 1780 |
} |
| 1781 |
} |
| 1782 |
|
| 1783 |
wp_safe_redirect(esc_url(admin_url('admin.php?page=mxchat-prompts'))); |
| 1784 |
exit; |
| 1785 |
} |
| 1786 |
|
| 1787 |
// Handle Regular URL (single page) |
| 1788 |
$page_content = $this->mxchat_extract_main_content($body_content); |
| 1789 |
$sanitized_content = $this->mxchat_sanitize_content_for_api($page_content); |
| 1790 |
|
| 1791 |
//error_log('[MXCHAT-URL-DEBUG] URL: ' . $submitted_url); |
| 1792 |
//error_log('[MXCHAT-URL-DEBUG] Extracted page_content length: ' . strlen($page_content)); |
| 1793 |
//error_log('[MXCHAT-URL-DEBUG] Sanitized content length: ' . strlen($sanitized_content)); |
| 1794 |
//error_log('[MXCHAT-URL-DEBUG] Content preview: ' . substr($sanitized_content, 0, 500)); |
| 1795 |
|
| 1796 |
if (empty($sanitized_content)) { |
| 1797 |
set_transient('mxchat_admin_notice_error', |
| 1798 |
esc_html__('No valid content found on the provided URL.', 'mxchat'), |
| 1799 |
30 |
| 1800 |
); |
| 1801 |
wp_safe_redirect(esc_url(admin_url('admin.php?page=mxchat-prompts'))); |
| 1802 |
exit; |
| 1803 |
} |
| 1804 |
|
| 1805 |
// For single URLs, process immediately using submit_content_to_db |
| 1806 |
// This handles chunking automatically for large content |
| 1807 |
$db_result = MxChat_Utils::submit_content_to_db( |
| 1808 |
$sanitized_content, |
| 1809 |
$submitted_url, |
| 1810 |
$api_key, |
| 1811 |
null, |
| 1812 |
$bot_id, |
| 1813 |
'url' // content_type |
| 1814 |
); |
| 1815 |
|
| 1816 |
if (is_wp_error($db_result)) { |
| 1817 |
$error_message = esc_html__('Failed to store content in database: ', 'mxchat') . esc_html($db_result->get_error_message()); |
| 1818 |
set_transient('mxchat_admin_notice_error', $error_message, 30); |
| 1819 |
} else { |
| 1820 |
$success_message = esc_html__('URL content successfully submitted!', 'mxchat'); |
| 1821 |
set_transient('mxchat_admin_notice_success', $success_message, 30); |
| 1822 |
} |
| 1823 |
|
| 1824 |
wp_safe_redirect(esc_url(admin_url('admin.php?page=mxchat-prompts'))); |
| 1825 |
exit; |
| 1826 |
} |
| 1827 |
|
| 1828 |
|
| 1829 |
public function mxchat_get_single_url_status() { |
| 1830 |
$status = get_transient('mxchat_single_url_status'); |
| 1831 |
if (!$status) { |
| 1832 |
return null; |
| 1833 |
} |
| 1834 |
|
| 1835 |
// Add human-readable time |
| 1836 |
if (isset($status['timestamp'])) { |
| 1837 |
$status['human_time'] = human_time_diff(strtotime($status['timestamp']), current_time('timestamp')) . ' ' . __('ago', 'mxchat'); |
| 1838 |
} |
| 1839 |
|
| 1840 |
return $status; |
| 1841 |
} |
| 1842 |
|
| 1843 |
public function mxchat_handle_sitemap_for_knowledge_base($xml, $sitemap_url, $bot_id = 'default') { |
| 1844 |
if (!current_user_can('manage_options')) { |
| 1845 |
return false; |
| 1846 |
} |
| 1847 |
|
| 1848 |
try { |
| 1849 |
$sitemap_url = esc_url_raw($sitemap_url); |
| 1850 |
|
| 1851 |
if (!$xml || !is_object($xml)) { |
| 1852 |
throw new Exception(__('Invalid XML object provided', 'mxchat')); |
| 1853 |
} |
| 1854 |
|
| 1855 |
// Get bot-specific embedding API for validation |
| 1856 |
$bot_options = $this->get_bot_options($bot_id); |
| 1857 |
$options = !empty($bot_options) ? $bot_options : get_option('mxchat_options'); |
| 1858 |
|
| 1859 |
// Test the embedding API before processing |
| 1860 |
$test_phrase = "Test embedding generation for MxChat"; |
| 1861 |
$test_result = $this->mxchat_generate_embedding($test_phrase, $bot_id); |
| 1862 |
|
| 1863 |
if (is_string($test_result)) { |
| 1864 |
throw new Exception(__('Embedding API validation failed: ', 'mxchat') . $test_result); |
| 1865 |
} |
| 1866 |
|
| 1867 |
if (!is_array($test_result)) { |
| 1868 |
throw new Exception(__('Embedding API returned unexpected result type. Please check your configuration.', 'mxchat')); |
| 1869 |
} |
| 1870 |
|
| 1871 |
// Extract URLs from sitemap |
| 1872 |
$urls = array(); |
| 1873 |
foreach ($xml->url as $url_element) { |
| 1874 |
$url = esc_url_raw((string)$url_element->loc); |
| 1875 |
if ($url) { |
| 1876 |
$urls[] = array('url' => $url); |
| 1877 |
} |
| 1878 |
} |
| 1879 |
|
| 1880 |
$total_urls = count($urls); |
| 1881 |
|
| 1882 |
if ($total_urls < 1) { |
| 1883 |
throw new Exception(__('No valid URLs found in sitemap', 'mxchat')); |
| 1884 |
} |
| 1885 |
|
| 1886 |
// Create unique queue ID |
| 1887 |
$queue_id = 'sitemap_' . md5($sitemap_url . time()); |
| 1888 |
|
| 1889 |
// Add URLs to queue |
| 1890 |
$queued_count = $this->mxchat_add_to_queue($queue_id, 'url', $urls, $bot_id); |
| 1891 |
|
| 1892 |
if ($queued_count === 0) { |
| 1893 |
throw new Exception(__('Failed to add URLs to processing queue', 'mxchat')); |
| 1894 |
} |
| 1895 |
|
| 1896 |
// Store queue metadata |
| 1897 |
$this->mxchat_set_queue_meta($queue_id, 'source_url', $sitemap_url); |
| 1898 |
$this->mxchat_set_queue_meta($queue_id, 'queue_type', 'sitemap'); |
| 1899 |
$this->mxchat_set_queue_meta($queue_id, 'total_items', $total_urls); |
| 1900 |
$this->mxchat_set_queue_meta($queue_id, 'bot_id', $bot_id); |
| 1901 |
$this->mxchat_set_queue_meta($queue_id, 'created_at', current_time('mysql')); |
| 1902 |
|
| 1903 |
// Store queue ID in transient for status tracking |
| 1904 |
set_transient('mxchat_active_queue_sitemap', $queue_id, DAY_IN_SECONDS); |
| 1905 |
set_transient('mxchat_last_sitemap_url', $sitemap_url, DAY_IN_SECONDS); |
| 1906 |
|
| 1907 |
return 'queued'; |
| 1908 |
|
| 1909 |
} catch (Exception $e) { |
| 1910 |
$error_message = $e->getMessage(); |
| 1911 |
//error_log(sprintf(esc_html__('Error preparing sitemap for processing: %s', 'mxchat'), esc_html($error_message))); |
| 1912 |
|
| 1913 |
return $error_message; |
| 1914 |
} |
| 1915 |
|
| 1916 |
} |
| 1917 |
|
| 1918 |
/** |
| 1919 |
* Remove shortcode tags but preserve the content inside them |
| 1920 |
* Example: [vc_column]Hello World[/vc_column] becomes "Hello World" |
| 1921 |
* |
| 1922 |
* @param string $content The content containing shortcodes |
| 1923 |
* @return string Content with shortcode tags removed but inner content preserved |
| 1924 |
*/ |
| 1925 |
/** |
| 1926 |
* Single-pass HTML entity decode for text entering the knowledge base. |
| 1927 |
* The corpus should hold what a human reads: a stored `&` consumes |
| 1928 |
* extra tokens, distorts the vector away from the form a visitor's |
| 1929 |
* question uses, and can be quoted back verbatim in an answer. |
| 1930 |
* Deliberately NOT looped to a fixed point — a stored `&amp;` is a |
| 1931 |
* legitimate literal `&` and must not collapse further (data loss). |
| 1932 |
* UTF-8 charset keeps multibyte (CJK/RTL) text untouched. Both assembly |
| 1933 |
* paths call this at their output points so the treatment cannot drift. |
| 1934 |
* (Plan d2c92e.) |
| 1935 |
*/ |
| 1936 |
private function mxchat_decode_entities_for_indexing($text) { |
| 1937 |
return html_entity_decode((string) $text, ENT_QUOTES | ENT_HTML5, 'UTF-8'); |
| 1938 |
} |
| 1939 |
|
| 1940 |
/** |
| 1941 |
* Price lines for a product's indexed text, pinned to the store's BASE currency. |
| 1942 |
* |
| 1943 |
* The four product assembly paths each used to call get_woocommerce_currency_symbol() |
| 1944 |
* with no argument, which resolves the currency active on the CURRENT request. |
| 1945 |
* Multi-currency plugins (CURCY, WOOCS, Aelia, WPML Multicurrency) filter that per |
| 1946 |
* request, so whichever currency the store happened to be serving when an import ran |
| 1947 |
* was frozen into every product it indexed. The amounts have the mirror problem: the |
| 1948 |
* woocommerce_product_get_* filters convert prices in the 'view' context but not in |
| 1949 |
* 'edit', so a converted amount could be paired with an unconverted symbol and produce |
| 1950 |
* a price that is not merely wrong but incoherent. |
| 1951 |
* |
| 1952 |
* Base currency option + 'edit' context makes both halves agree and makes the output |
| 1953 |
* independent of when the import ran. The currency CODE is emitted alongside the symbol |
| 1954 |
* so the model cannot read a bare "$" as USD on a store whose plugin swapped symbols. |
| 1955 |
* (Plan 7403ec.) |
| 1956 |
*/ |
| 1957 |
private function mxchat_product_price_lines($product) { |
| 1958 |
if (!is_object($product) || !method_exists($product, 'get_regular_price')) { |
| 1959 |
return ''; |
| 1960 |
} |
| 1961 |
|
| 1962 |
$currency = get_option('woocommerce_currency'); |
| 1963 |
$currency = is_string($currency) ? trim($currency) : ''; |
| 1964 |
$symbol = ($currency !== '') |
| 1965 |
? get_woocommerce_currency_symbol($currency) |
| 1966 |
: get_woocommerce_currency_symbol(); |
| 1967 |
$symbol = $this->mxchat_decode_entities_for_indexing($symbol); |
| 1968 |
|
| 1969 |
$regular_price = $product->get_regular_price('edit'); |
| 1970 |
$sale_price = $product->get_sale_price('edit'); |
| 1971 |
$price = $product->get_price('edit'); |
| 1972 |
|
| 1973 |
$lines = ''; |
| 1974 |
|
| 1975 |
if (!empty($regular_price)) { |
| 1976 |
$lines .= "Price: " . $this->mxchat_format_indexed_price($regular_price, $currency, $symbol) . "\n"; |
| 1977 |
} elseif (!empty($price)) { |
| 1978 |
$lines .= "Price: " . $this->mxchat_format_indexed_price($price, $currency, $symbol) . "\n"; |
| 1979 |
} |
| 1980 |
|
| 1981 |
if (!empty($sale_price) && $sale_price !== $regular_price) { |
| 1982 |
$lines .= "Sale Price: " . $this->mxchat_format_indexed_price($sale_price, $currency, $symbol) . "\n"; |
| 1983 |
} |
| 1984 |
|
| 1985 |
if ($product->is_type('variable')) { |
| 1986 |
list($min_price, $max_price) = $this->mxchat_variation_price_range($product); |
| 1987 |
if ($min_price !== null && $max_price !== null && (float) $min_price !== (float) $max_price) { |
| 1988 |
$lines .= "Price Range: " . $this->mxchat_format_indexed_price($min_price, $currency, $symbol) |
| 1989 |
. " - " . $this->mxchat_format_indexed_price($max_price, $currency, $symbol) . "\n"; |
| 1990 |
} |
| 1991 |
} |
| 1992 |
|
| 1993 |
return $lines; |
| 1994 |
} |
| 1995 |
|
| 1996 |
/** |
| 1997 |
* One indexed price amount, labelled with its currency code. |
| 1998 |
* |
| 1999 |
* "INR 1299.00 (Rs.1299.00)" — the code is what the model should reason from; the symbol |
| 2000 |
* is kept so a quoted price still reads naturally. Falls back to the old symbol-only |
| 2001 |
* shape when WooCommerce has no base currency configured, and drops the parenthetical |
| 2002 |
* when the symbol is absent or IS the code (several currencies have no distinct glyph). |
| 2003 |
*/ |
| 2004 |
private function mxchat_format_indexed_price($amount, $currency, $symbol) { |
| 2005 |
$amount = (string) $amount; |
| 2006 |
|
| 2007 |
if ($currency === '') { |
| 2008 |
return $symbol . $amount; |
| 2009 |
} |
| 2010 |
|
| 2011 |
if ($symbol === '' || $symbol === $currency) { |
| 2012 |
return $currency . ' ' . $amount; |
| 2013 |
} |
| 2014 |
|
| 2015 |
return $currency . ' ' . $amount . ' (' . $symbol . $amount . ')'; |
| 2016 |
} |
| 2017 |
|
| 2018 |
/** |
| 2019 |
* Min/max variation price read from the variations themselves in 'edit' context. |
| 2020 |
* |
| 2021 |
* get_variation_price() reads WooCommerce's display price cache, which multi-currency |
| 2022 |
* plugins populate with converted values — the same defect the rest of this helper |
| 2023 |
* exists to remove. Returns raw stored strings (not floats) so the indexed text keeps |
| 2024 |
* the store's own price formatting, and (null, null) when no variation carries a price. |
| 2025 |
*/ |
| 2026 |
private function mxchat_variation_price_range($product) { |
| 2027 |
$min_raw = null; |
| 2028 |
$max_raw = null; |
| 2029 |
$min_val = null; |
| 2030 |
$max_val = null; |
| 2031 |
|
| 2032 |
$children = method_exists($product, 'get_children') ? $product->get_children() : array(); |
| 2033 |
|
| 2034 |
foreach ($children as $child_id) { |
| 2035 |
$variation = wc_get_product($child_id); |
| 2036 |
if (!$variation) { |
| 2037 |
continue; |
| 2038 |
} |
| 2039 |
$raw = $variation->get_price('edit'); |
| 2040 |
if ($raw === '' || $raw === null) { |
| 2041 |
continue; |
| 2042 |
} |
| 2043 |
$val = (float) $raw; |
| 2044 |
if ($min_val === null || $val < $min_val) { |
| 2045 |
$min_val = $val; |
| 2046 |
$min_raw = $raw; |
| 2047 |
} |
| 2048 |
if ($max_val === null || $val > $max_val) { |
| 2049 |
$max_val = $val; |
| 2050 |
$max_raw = $raw; |
| 2051 |
} |
| 2052 |
} |
| 2053 |
|
| 2054 |
return array($min_raw, $max_raw); |
| 2055 |
} |
| 2056 |
|
| 2057 |
private function strip_shortcode_tags_preserve_content($content) { |
| 2058 |
// Single-pass regex removes all shortcode brackets: [tag], [tag attr="val"], [/tag], [tag /] |
| 2059 |
// Content between tags is inherently preserved since only brackets are targeted |
| 2060 |
$result = preg_replace('/\[\/?\w[\w-]*[^\]]*\]/', '', $content); |
| 2061 |
return ($result !== null) ? $result : $content; |
| 2062 |
} |
| 2063 |
|
| 2064 |
public function mxchat_sanitize_content_for_api($content) { |
| 2065 |
//error_log('[MXCHAT-SANITIZE] Original content preview: ' . substr($content, 0, 500) . '...'); |
| 2066 |
|
| 2067 |
// Remove shortcode tags but PRESERVE content inside them |
| 2068 |
$content = $this->strip_shortcode_tags_preserve_content($content); |
| 2069 |
|
| 2070 |
// Remove script, style tags, and HTML comments |
| 2071 |
$content = preg_replace('/<script\b[^>]*>(.*?)<\/script>/is', '', $content); |
| 2072 |
$content = preg_replace('/<style\b[^>]*>(.*?)<\/style>/is', '', $content); |
| 2073 |
$content = preg_replace('/<!--(.|\s)*?-->/', '', $content); |
| 2074 |
|
| 2075 |
// Remove all HTML tags and decode HTML entities |
| 2076 |
$content = wp_strip_all_tags($content); |
| 2077 |
$content = html_entity_decode($content, ENT_QUOTES | ENT_HTML5); |
| 2078 |
|
| 2079 |
// Normalize whitespace but preserve paragraph breaks |
| 2080 |
// First, normalize line endings to \n |
| 2081 |
$content = str_replace(["\r\n", "\r"], "\n", $content); |
| 2082 |
// Replace multiple spaces/tabs with single space, but preserve newlines |
| 2083 |
$content = preg_replace('/[ \t]+/', ' ', $content); |
| 2084 |
// Replace 3+ newlines with 2 newlines (max 2 blank lines) |
| 2085 |
$content = preg_replace('/\n{3,}/', "\n\n", $content); |
| 2086 |
// Trim each line |
| 2087 |
$lines = explode("\n", $content); |
| 2088 |
$lines = array_map('trim', $lines); |
| 2089 |
$content = implode("\n", $lines); |
| 2090 |
// Final trim |
| 2091 |
$content = trim($content); |
| 2092 |
|
| 2093 |
// Remove control characters (which can cause database issues) but preserve \n (0x0A) and \r (0x0D) |
| 2094 |
$content = preg_replace('/[\x00-\x08\x0B\x0C\x0E-\x1F\x7F]/u', '', $content); |
| 2095 |
|
| 2096 |
// Remove NULL bytes which can cause database errors |
| 2097 |
$content = str_replace("\0", "", $content); |
| 2098 |
|
| 2099 |
// Ensure valid UTF-8 encoding |
| 2100 |
$content = wp_check_invalid_utf8($content); |
| 2101 |
|
| 2102 |
// Remove extremely long runs with no whitespace (base64 blobs, minified JS). |
| 2103 |
// Counts CHARACTERS (/u), and never strips a run containing characters from a |
| 2104 |
// script written without spaces — Japanese, Chinese, Thai, Khmer, Lao, Myanmar — |
| 2105 |
// where a normal paragraph is legitimately one unbroken run. |
| 2106 |
$content = preg_replace_callback('/\S{300,}/u', function ($m) { |
| 2107 |
return preg_match('/[\p{Han}\p{Hiragana}\p{Katakana}\p{Thai}\p{Khmer}\p{Lao}\p{Myanmar}]/u', $m[0]) ? $m[0] : ' '; |
| 2108 |
}, $content); |
| 2109 |
|
| 2110 |
// Remove emoji/symbol blocks only — not the whole supplementary plane, which |
| 2111 |
// also holds CJK Extension B ideographs used in real Chinese/Japanese names. |
| 2112 |
// A ZWJ (U+200D) BETWEEN stripped pictographs is consumed with them, so a |
| 2113 |
// family sequence like 👨👩👧 leaves no invisible zero-width residue behind |
| 2114 |
// (the joiner between NON-emoji characters — Hindi conjuncts — is untouched). |
| 2115 |
$content = preg_replace('/[\x{1F000}-\x{1F0FF}\x{1F300}-\x{1FAFF}](?:\x{200D}[\x{1F000}-\x{1F0FF}\x{1F300}-\x{1FAFF}])*/u', '', $content); |
| 2116 |
|
| 2117 |
// There is deliberately NO catch-all character allowlist here (plan 209e57; |
| 2118 |
// one existed until 3.2.20). Every genuinely dangerous byte is already gone: |
| 2119 |
// control characters, null bytes, invalid UTF-8 and the emoji blocks are all |
| 2120 |
// stripped above. The allowlist's only remaining effect was to damage scripts |
| 2121 |
// nobody thought to enumerate — Unicode Cf (Format) was missing, so it |
| 2122 |
// replaced the zero-width joiner/non-joiner with spaces and silently split |
| 2123 |
// Persian words (� |
| 2124 |
یرو� |
| 2125 |
→ � |
| 2126 |
ی رو� |
| 2127 |
) and broke Hindi conjuncts (क्ष → क् ष). |
| 2128 |
// Do not add one back; the failure mode of an allowlist is exactly this. |
| 2129 |
|
| 2130 |
// Limit to reasonable length if needed (byte limit — MySQL TEXT is byte-sized, |
| 2131 |
// but cut on a character boundary so a multibyte char is never split mid-sequence) |
| 2132 |
$max_length = 65000; // Just under MySQL TEXT field limit |
| 2133 |
if (strlen($content) > $max_length) { |
| 2134 |
$content = mb_strcut($content, 0, $max_length, 'UTF-8'); |
| 2135 |
} |
| 2136 |
|
| 2137 |
//error_log('[MXCHAT-SANITIZE] Sanitized content preview: ' . substr($content, 0, 500) . '...'); |
| 2138 |
return $content; |
| 2139 |
} |
| 2140 |
public function mxchat_extract_main_content($html) { |
| 2141 |
if (empty($html)) { |
| 2142 |
return ''; |
| 2143 |
} |
| 2144 |
try { |
| 2145 |
$dom = new DOMDocument; |
| 2146 |
libxml_use_internal_errors(true); // Suppress HTML parsing errors |
| 2147 |
@$dom->loadHTML(mb_convert_encoding($html, 'HTML-ENTITIES', 'UTF-8'), LIBXML_HTML_NOIMPLIED | LIBXML_HTML_NODEFDTD); |
| 2148 |
$xpath = new DOMXPath($dom); |
| 2149 |
|
| 2150 |
// For debugging purposes |
| 2151 |
$debugEnabled = true; // Set to true to enable debugging output |
| 2152 |
$debug = function($message) use ($debugEnabled) { |
| 2153 |
if ($debugEnabled) { |
| 2154 |
//error_log('[MXCHAT-EXTRACT-DEBUG] ' . $message); |
| 2155 |
} |
| 2156 |
}; |
| 2157 |
|
| 2158 |
// Direct targeting for Gerow theme posts |
| 2159 |
$post_text = $xpath->query('//div[contains(@class, "post-text")]'); |
| 2160 |
if ($post_text && $post_text->length > 0) { |
| 2161 |
$debug("Found post-text directly"); |
| 2162 |
$content = ''; |
| 2163 |
foreach ($post_text as $node) { |
| 2164 |
$content .= $dom->saveHTML($node); |
| 2165 |
} |
| 2166 |
if (!empty($content)) { |
| 2167 |
$debug("Returning post-text content"); |
| 2168 |
return $content; |
| 2169 |
} |
| 2170 |
} |
| 2171 |
|
| 2172 |
// Try to get the blog details content which contains the post-text |
| 2173 |
$blog_details = $xpath->query('//div[contains(@class, "blog-details-content")]'); |
| 2174 |
if ($blog_details && $blog_details->length > 0) { |
| 2175 |
$debug("Found blog-details-content"); |
| 2176 |
$content = ''; |
| 2177 |
foreach ($blog_details as $node) { |
| 2178 |
$content .= $dom->saveHTML($node); |
| 2179 |
} |
| 2180 |
if (!empty($content)) { |
| 2181 |
$debug("Returning blog-details-content"); |
| 2182 |
return $content; |
| 2183 |
} |
| 2184 |
} |
| 2185 |
|
| 2186 |
// Try to get the article which contains the blog details |
| 2187 |
$article = $xpath->query('//article[contains(@class, "blog-details-wrap")]'); |
| 2188 |
if ($article && $article->length > 0) { |
| 2189 |
$debug("Found article with blog-details-wrap"); |
| 2190 |
$content = ''; |
| 2191 |
foreach ($article as $node) { |
| 2192 |
$content .= $dom->saveHTML($node); |
| 2193 |
} |
| 2194 |
if (!empty($content)) { |
| 2195 |
$debug("Returning article content"); |
| 2196 |
return $content; |
| 2197 |
} |
| 2198 |
} |
| 2199 |
|
| 2200 |
// Try even broader with the blog-item-wrap |
| 2201 |
$blog_item = $xpath->query('//div[contains(@class, "blog-item-wrap")]'); |
| 2202 |
if ($blog_item && $blog_item->length > 0) { |
| 2203 |
$debug("Found blog-item-wrap"); |
| 2204 |
$content = ''; |
| 2205 |
foreach ($blog_item as $node) { |
| 2206 |
$content .= $dom->saveHTML($node); |
| 2207 |
} |
| 2208 |
if (!empty($content)) { |
| 2209 |
$debug("Returning blog-item-wrap content"); |
| 2210 |
return $content; |
| 2211 |
} |
| 2212 |
} |
| 2213 |
|
| 2214 |
// Specific Gerow theme path |
| 2215 |
$gerow_path = $xpath->query('//section[contains(@class, "blog-area")]//div[contains(@class, "post-text")]'); |
| 2216 |
if ($gerow_path && $gerow_path->length > 0) { |
| 2217 |
$debug("Found Gerow theme path to post-text"); |
| 2218 |
$content = ''; |
| 2219 |
foreach ($gerow_path as $node) { |
| 2220 |
$content .= $dom->saveHTML($node); |
| 2221 |
} |
| 2222 |
if (!empty($content)) { |
| 2223 |
$debug("Returning Gerow post-text content"); |
| 2224 |
return $content; |
| 2225 |
} |
| 2226 |
} |
| 2227 |
|
| 2228 |
// Generic blog post selectors |
| 2229 |
$selectors = [ |
| 2230 |
// Blog post specific selectors |
| 2231 |
'//div[contains(@class, "post-text")]', |
| 2232 |
'//article[contains(@class, "blog-post-item")]//div[contains(@class, "post-text")]', |
| 2233 |
'//div[contains(@class, "blog-details-content")]', |
| 2234 |
'//article[contains(@class, "blog-details-wrap")]', |
| 2235 |
'//div[contains(@class, "entry-content")]', |
| 2236 |
'//div[contains(@class, "blog-content")]', |
| 2237 |
'//div[contains(@class, "blog-item-wrap")]', |
| 2238 |
|
| 2239 |
// More general content selectors |
| 2240 |
'//div[contains(@class, "page__content")]', |
| 2241 |
'//div[contains(@class, "elementor-widget-container")]', |
| 2242 |
'//div[contains(@class, "elementor-text-editor")]', |
| 2243 |
'//div[contains(@class, "elementor-widget-text-editor")]', |
| 2244 |
'//*[contains(@class, "entry-content")]', |
| 2245 |
'//*[contains(@class, "post-content")]', |
| 2246 |
'//*[contains(@class, "article-content")]', |
| 2247 |
'//*[@id="content"]', |
| 2248 |
'//*[@id="main-content"]', |
| 2249 |
'//section[contains(@class, "blog-area")]', |
| 2250 |
'//article', |
| 2251 |
'//main', |
| 2252 |
'//div[contains(@class, "content")]' |
| 2253 |
]; |
| 2254 |
|
| 2255 |
// First handle Elementor content - get only leaf widget containers to avoid duplicates |
| 2256 |
$debug("Checking for Elementor content"); |
| 2257 |
// Get widget containers that are direct children of widgets (not nested inside other widget containers) |
| 2258 |
$elementor_widgets = $xpath->query('//div[contains(@class, "elementor-widget")]//div[contains(@class, "elementor-widget-container")]'); |
| 2259 |
if ($elementor_widgets && $elementor_widgets->length > 0) { |
| 2260 |
$debug("Found Elementor widgets"); |
| 2261 |
$seen_content = array(); // Track seen content to avoid duplicates |
| 2262 |
$combined_content = ''; |
| 2263 |
foreach ($elementor_widgets as $widget) { |
| 2264 |
$widget_content = $dom->saveHTML($widget); |
| 2265 |
if (!empty($widget_content)) { |
| 2266 |
// Create a hash of the content to detect duplicates |
| 2267 |
$content_hash = md5($widget_content); |
| 2268 |
if (!isset($seen_content[$content_hash])) { |
| 2269 |
$seen_content[$content_hash] = true; |
| 2270 |
$combined_content .= $widget_content; |
| 2271 |
} |
| 2272 |
} |
| 2273 |
} |
| 2274 |
if (!empty($combined_content)) { |
| 2275 |
$debug("Returning Elementor content"); |
| 2276 |
return $combined_content; |
| 2277 |
} |
| 2278 |
} |
| 2279 |
|
| 2280 |
// Try standard selectors one by one |
| 2281 |
foreach ($selectors as $selector) { |
| 2282 |
$debug("Trying selector: " . $selector); |
| 2283 |
$nodes = $xpath->query($selector); |
| 2284 |
if ($nodes && $nodes->length > 0) { |
| 2285 |
$debug("Found " . $nodes->length . " matches for selector: " . $selector); |
| 2286 |
// Only take the FIRST matching node to avoid duplicate content |
| 2287 |
// (pages often have nested or multiple containers with same class) |
| 2288 |
$content = $dom->saveHTML($nodes->item(0)); |
| 2289 |
if (!empty($content)) { |
| 2290 |
$debug("Returning content from selector: " . $selector . " (first match only)"); |
| 2291 |
return $content; |
| 2292 |
} |
| 2293 |
} |
| 2294 |
} |
| 2295 |
|
| 2296 |
// Manual regex fallback for post-text if DOM methods fail |
| 2297 |
$debug("Trying regex fallback"); |
| 2298 |
if (preg_match('/<div class="post-text">(.*?)<\/div>\s*<\/div>\s*<\/div>/s', $html, $matches)) { |
| 2299 |
$debug("Found post-text via regex"); |
| 2300 |
return '<div class="post-text">' . $matches[1] . '</div>'; |
| 2301 |
} |
| 2302 |
|
| 2303 |
// Try to extract the blog section as a whole |
| 2304 |
$blog_section = $xpath->query('//section[contains(@class, "blog-area")]'); |
| 2305 |
if ($blog_section && $blog_section->length > 0) { |
| 2306 |
$debug("Found blog-area section"); |
| 2307 |
$content = ''; |
| 2308 |
foreach ($blog_section as $node) { |
| 2309 |
$content .= $dom->saveHTML($node); |
| 2310 |
} |
| 2311 |
if (!empty($content)) { |
| 2312 |
$debug("Returning blog-area section content"); |
| 2313 |
return $content; |
| 2314 |
} |
| 2315 |
} |
| 2316 |
|
| 2317 |
// Generic container selectors for non-CMS sites (like .asp pages) |
| 2318 |
$debug("Trying generic container selectors"); |
| 2319 |
$generic_selectors = [ |
| 2320 |
'//div[@id="main"]', |
| 2321 |
'//div[@id="wrapper"]', |
| 2322 |
'//div[@id="page"]', |
| 2323 |
'//div[@id="site-content"]', |
| 2324 |
'//div[contains(@class, "main-content")]', |
| 2325 |
'//div[contains(@class, "page-content")]', |
| 2326 |
'//div[contains(@class, "site-content")]', |
| 2327 |
]; |
| 2328 |
|
| 2329 |
foreach ($generic_selectors as $selector) { |
| 2330 |
$debug("Trying generic selector: " . $selector); |
| 2331 |
$nodes = $xpath->query($selector); |
| 2332 |
if ($nodes && $nodes->length > 0) { |
| 2333 |
$content = $dom->saveHTML($nodes->item(0)); |
| 2334 |
if (!empty($content)) { |
| 2335 |
$debug("Returning content from generic selector: " . $selector); |
| 2336 |
return $content; |
| 2337 |
} |
| 2338 |
} |
| 2339 |
} |
| 2340 |
|
| 2341 |
// Paragraph-based content detection - find regions with substantial text |
| 2342 |
$debug("Trying paragraph-based content detection"); |
| 2343 |
$paragraphs = $xpath->query('//p[string-length(normalize-space()) > 50]'); |
| 2344 |
if ($paragraphs && $paragraphs->length >= 3) { |
| 2345 |
$debug("Found " . $paragraphs->length . " substantial paragraphs"); |
| 2346 |
// Collect all substantial paragraphs and their content |
| 2347 |
$paragraph_content = ''; |
| 2348 |
foreach ($paragraphs as $p) { |
| 2349 |
$paragraph_content .= $dom->saveHTML($p) . "\n"; |
| 2350 |
} |
| 2351 |
if (!empty($paragraph_content)) { |
| 2352 |
$debug("Returning paragraph-based content"); |
| 2353 |
return $paragraph_content; |
| 2354 |
} |
| 2355 |
} |
| 2356 |
|
| 2357 |
// Improved body fallback - strip nav/header/footer elements first |
| 2358 |
$debug("Using improved body fallback"); |
| 2359 |
$body = $dom->getElementsByTagName('body'); |
| 2360 |
if ($body->length > 0) { |
| 2361 |
// Clone the body to avoid modifying the original DOM |
| 2362 |
$body_clone = $body->item(0)->cloneNode(true); |
| 2363 |
|
| 2364 |
// Remove common non-content elements by tag name |
| 2365 |
$remove_tags = ['nav', 'header', 'footer', 'aside', 'script', 'style', 'noscript']; |
| 2366 |
foreach ($remove_tags as $tag) { |
| 2367 |
$elements = $body_clone->getElementsByTagName($tag); |
| 2368 |
// Iterate backwards to safely remove elements |
| 2369 |
for ($i = $elements->length - 1; $i >= 0; $i--) { |
| 2370 |
$el = $elements->item($i); |
| 2371 |
if ($el && $el->parentNode) { |
| 2372 |
$el->parentNode->removeChild($el); |
| 2373 |
} |
| 2374 |
} |
| 2375 |
} |
| 2376 |
|
| 2377 |
// Remove elements with common non-content class names using XPath on the cloned body |
| 2378 |
$temp_dom = new DOMDocument(); |
| 2379 |
@$temp_dom->appendChild($temp_dom->importNode($body_clone, true)); |
| 2380 |
$temp_xpath = new DOMXPath($temp_dom); |
| 2381 |
|
| 2382 |
$remove_class_patterns = [ |
| 2383 |
'//*[contains(@class, "nav")]', |
| 2384 |
'//*[contains(@class, "menu")]', |
| 2385 |
'//*[contains(@class, "sidebar")]', |
| 2386 |
'//*[contains(@class, "footer")]', |
| 2387 |
'//*[contains(@class, "header")]', |
| 2388 |
'//*[contains(@id, "nav")]', |
| 2389 |
'//*[contains(@id, "menu")]', |
| 2390 |
'//*[contains(@id, "sidebar")]', |
| 2391 |
'//*[contains(@id, "footer")]', |
| 2392 |
'//*[contains(@id, "header")]', |
| 2393 |
]; |
| 2394 |
|
| 2395 |
foreach ($remove_class_patterns as $pattern) { |
| 2396 |
$elements = $temp_xpath->query($pattern); |
| 2397 |
if ($elements) { |
| 2398 |
for ($i = $elements->length - 1; $i >= 0; $i--) { |
| 2399 |
$el = $elements->item($i); |
| 2400 |
if ($el && $el->parentNode) { |
| 2401 |
$el->parentNode->removeChild($el); |
| 2402 |
} |
| 2403 |
} |
| 2404 |
} |
| 2405 |
} |
| 2406 |
|
| 2407 |
$cleaned_content = $temp_dom->saveHTML(); |
| 2408 |
if (!empty($cleaned_content)) { |
| 2409 |
$debug("Returning cleaned body content"); |
| 2410 |
return $cleaned_content; |
| 2411 |
} |
| 2412 |
} |
| 2413 |
|
| 2414 |
// Last resort: return the original HTML |
| 2415 |
$debug("Returning original HTML"); |
| 2416 |
return $html; |
| 2417 |
} catch (Exception $e) { |
| 2418 |
//error_log('[MXCHAT-ERROR] Content extraction failed: ' . $e->getMessage()); |
| 2419 |
return $html; // Return original HTML if parsing fails |
| 2420 |
} finally { |
| 2421 |
libxml_clear_errors(); |
| 2422 |
} |
| 2423 |
} |
| 2424 |
public function mxchat_get_sitemap_processing_status($sitemap_url) { |
| 2425 |
$sitemap_url = esc_url_raw($sitemap_url); |
| 2426 |
$status_key = sanitize_key('mxchat_sitemap_status_' . md5($sitemap_url)); |
| 2427 |
$status = get_transient($status_key); |
| 2428 |
|
| 2429 |
if (!$status || !is_array($status)) { |
| 2430 |
return false; |
| 2431 |
} |
| 2432 |
|
| 2433 |
// Auto-complete check: if all URLs are processed but status isn't complete |
| 2434 |
if (isset($status['processed_urls']) && isset($status['total_urls']) && |
| 2435 |
$status['processed_urls'] >= $status['total_urls'] && |
| 2436 |
isset($status['status']) && $status['status'] !== 'complete' && |
| 2437 |
$status['status'] !== 'error') { |
| 2438 |
|
| 2439 |
// Mark as complete |
| 2440 |
$status['status'] = 'complete'; |
| 2441 |
$status['processed_urls'] = $status['total_urls']; // Ensure exact match |
| 2442 |
|
| 2443 |
// Update the transient with the corrected status |
| 2444 |
set_transient($status_key, $status, DAY_IN_SECONDS); |
| 2445 |
} |
| 2446 |
|
| 2447 |
return array( |
| 2448 |
'total_urls' => absint($status['total_urls']), |
| 2449 |
'processed_urls' => absint($status['processed_urls']), |
| 2450 |
'failed_urls' => absint($status['failed_urls'] ?? 0), |
| 2451 |
'percentage' => ($status['total_urls'] > 0) |
| 2452 |
? round((absint($status['processed_urls']) / absint($status['total_urls'])) * 100) |
| 2453 |
: 0, |
| 2454 |
'status' => sanitize_text_field($status['status']), |
| 2455 |
'last_update' => human_time_diff(absint($status['last_update']), time()) . ' ' . esc_html__('ago', 'mxchat'), |
| 2456 |
'error' => isset($status['error']) ? sanitize_text_field($status['error']) : '', |
| 2457 |
'last_error' => isset($status['last_error']) ? sanitize_text_field($status['last_error']) : '', |
| 2458 |
'failed_urls_list' => isset($status['failed_urls_list']) ? $status['failed_urls_list'] : array() |
| 2459 |
); |
| 2460 |
} |
| 2461 |
|
| 2462 |
public function mxchat_ajax_get_status_updates() { |
| 2463 |
try { |
| 2464 |
// Verify the request |
| 2465 |
check_ajax_referer('mxchat_status_nonce', 'nonce'); |
| 2466 |
|
| 2467 |
// Get active queue IDs |
| 2468 |
$sitemap_queue_id = get_transient('mxchat_active_queue_sitemap'); |
| 2469 |
$pdf_queue_id = get_transient('mxchat_active_queue_pdf'); |
| 2470 |
|
| 2471 |
$sitemap_status = false; |
| 2472 |
$pdf_status = false; |
| 2473 |
|
| 2474 |
// Get sitemap queue status |
| 2475 |
if ($sitemap_queue_id) { |
| 2476 |
$sitemap_status = $this->mxchat_get_queue_status_data($sitemap_queue_id, 'sitemap'); |
| 2477 |
} |
| 2478 |
|
| 2479 |
// Get PDF queue status |
| 2480 |
if ($pdf_queue_id) { |
| 2481 |
$pdf_status = $this->mxchat_get_queue_status_data($pdf_queue_id, 'pdf'); |
| 2482 |
} |
| 2483 |
|
| 2484 |
$is_active_processing = |
| 2485 |
($sitemap_status && $sitemap_status['status'] === 'processing') || |
| 2486 |
($pdf_status && $pdf_status['status'] === 'processing'); |
| 2487 |
|
| 2488 |
// Return JSON response with the status data |
| 2489 |
wp_send_json(array( |
| 2490 |
'pdf_status' => $pdf_status, |
| 2491 |
'sitemap_status' => $sitemap_status, |
| 2492 |
'is_processing' => $is_active_processing, |
| 2493 |
'sitemap_queue_id' => $sitemap_queue_id, |
| 2494 |
'pdf_queue_id' => $pdf_queue_id |
| 2495 |
)); |
| 2496 |
|
| 2497 |
} catch (Exception $e) { |
| 2498 |
//error_log('MxChat Status Update Error: ' . $e->getMessage()); |
| 2499 |
|
| 2500 |
wp_send_json_error(array( |
| 2501 |
'message' => 'Error getting status updates: ' . $e->getMessage(), |
| 2502 |
'status' => 'error' |
| 2503 |
)); |
| 2504 |
} |
| 2505 |
} |
| 2506 |
|
| 2507 |
/** |
| 2508 |
* Helper function to get queue status data |
| 2509 |
*/ |
| 2510 |
private function mxchat_get_queue_status_data($queue_id, $type = 'sitemap') { |
| 2511 |
global $wpdb; |
| 2512 |
$table_name = $wpdb->prefix . 'mxchat_processing_queue'; |
| 2513 |
|
| 2514 |
// Get counts by status |
| 2515 |
$counts = $wpdb->get_results($wpdb->prepare( |
| 2516 |
"SELECT status, COUNT(*) as count |
| 2517 |
FROM $table_name |
| 2518 |
WHERE queue_id = %s |
| 2519 |
GROUP BY status", |
| 2520 |
$queue_id |
| 2521 |
), OBJECT_K); |
| 2522 |
|
| 2523 |
$total = 0; |
| 2524 |
$completed = 0; |
| 2525 |
$failed = 0; |
| 2526 |
$processing = 0; |
| 2527 |
$pending = 0; |
| 2528 |
|
| 2529 |
foreach ($counts as $status => $data) { |
| 2530 |
$count = absint($data->count); |
| 2531 |
$total += $count; |
| 2532 |
|
| 2533 |
switch ($status) { |
| 2534 |
case 'completed': |
| 2535 |
$completed = $count; |
| 2536 |
break; |
| 2537 |
case 'failed': |
| 2538 |
$failed = $count; |
| 2539 |
break; |
| 2540 |
case 'processing': |
| 2541 |
$processing = $count; |
| 2542 |
break; |
| 2543 |
case 'pending': |
| 2544 |
$pending = $count; |
| 2545 |
break; |
| 2546 |
} |
| 2547 |
} |
| 2548 |
|
| 2549 |
if ($total === 0) { |
| 2550 |
return false; |
| 2551 |
} |
| 2552 |
|
| 2553 |
// Calculate percentage |
| 2554 |
$percentage = round((($completed + $failed) / $total) * 100); |
| 2555 |
|
| 2556 |
// Get failed items details (limit to 50) |
| 2557 |
$failed_items = array(); |
| 2558 |
if ($failed > 0) { |
| 2559 |
$failed_results = $wpdb->get_results($wpdb->prepare( |
| 2560 |
"SELECT item_type, item_data, error_message, attempts, completed_at |
| 2561 |
FROM $table_name |
| 2562 |
WHERE queue_id = %s |
| 2563 |
AND status = 'failed' |
| 2564 |
AND attempts >= max_attempts |
| 2565 |
ORDER BY id DESC |
| 2566 |
LIMIT 50", |
| 2567 |
$queue_id |
| 2568 |
)); |
| 2569 |
|
| 2570 |
foreach ($failed_results as $item) { |
| 2571 |
$data = json_decode($item->item_data, true); |
| 2572 |
$url = $type === 'pdf' ? ($data['pdf_url'] ?? '') : ($data['url'] ?? ''); |
| 2573 |
|
| 2574 |
$failed_items[] = array( |
| 2575 |
'url' => $url, |
| 2576 |
'page' => $type === 'pdf' ? ($data['page_number'] ?? 0) : 0, |
| 2577 |
'error' => $item->error_message, |
| 2578 |
'retries' => $item->attempts, |
| 2579 |
'time' => strtotime($item->completed_at) |
| 2580 |
); |
| 2581 |
} |
| 2582 |
} |
| 2583 |
|
| 2584 |
// Get queue metadata |
| 2585 |
$source_url = $this->mxchat_get_queue_meta($queue_id, 'source_url'); |
| 2586 |
|
| 2587 |
// Determine if queue is complete |
| 2588 |
$is_complete = ($pending === 0 && $processing === 0); |
| 2589 |
|
| 2590 |
// Get last update time |
| 2591 |
$last_update = $wpdb->get_var($wpdb->prepare( |
| 2592 |
"SELECT MAX(UNIX_TIMESTAMP(COALESCE(completed_at, started_at, created_at))) |
| 2593 |
FROM $table_name |
| 2594 |
WHERE queue_id = %s", |
| 2595 |
$queue_id |
| 2596 |
)); |
| 2597 |
|
| 2598 |
$last_update_text = $last_update ? human_time_diff($last_update, time()) . ' ' . __('ago', 'mxchat') : __('Just now', 'mxchat'); |
| 2599 |
|
| 2600 |
// Format based on type |
| 2601 |
if ($type === 'pdf') { |
| 2602 |
return array( |
| 2603 |
'total_pages' => $total, |
| 2604 |
'processed_pages' => $completed + $failed, |
| 2605 |
'failed_pages' => $failed, |
| 2606 |
'percentage' => $percentage, |
| 2607 |
'status' => $is_complete ? 'complete' : 'processing', |
| 2608 |
'last_update' => $last_update_text, |
| 2609 |
'failed_pages_list' => $failed_items, |
| 2610 |
'pdf_url' => $source_url, |
| 2611 |
'queue_id' => $queue_id |
| 2612 |
); |
| 2613 |
} else { |
| 2614 |
return array( |
| 2615 |
'total_urls' => $total, |
| 2616 |
'processed_urls' => $completed + $failed, |
| 2617 |
'failed_urls' => $failed, |
| 2618 |
'percentage' => $percentage, |
| 2619 |
'status' => $is_complete ? 'complete' : 'processing', |
| 2620 |
'last_update' => $last_update_text, |
| 2621 |
'failed_urls_list' => $failed_items, |
| 2622 |
'sitemap_url' => $source_url, |
| 2623 |
'queue_id' => $queue_id |
| 2624 |
); |
| 2625 |
} |
| 2626 |
} |
| 2627 |
|
| 2628 |
/** |
| 2629 |
* Public method to get processing status for both sitemap and PDF queues |
| 2630 |
* Used by admin pages to display processing status |
| 2631 |
* |
| 2632 |
* @return array Array with 'sitemap_status', 'pdf_status', and 'is_processing' keys |
| 2633 |
*/ |
| 2634 |
public function mxchat_get_processing_statuses() { |
| 2635 |
$sitemap_queue_id = get_transient('mxchat_active_queue_sitemap'); |
| 2636 |
$pdf_queue_id = get_transient('mxchat_active_queue_pdf'); |
| 2637 |
|
| 2638 |
$sitemap_status = false; |
| 2639 |
$pdf_status = false; |
| 2640 |
|
| 2641 |
if ($sitemap_queue_id) { |
| 2642 |
$sitemap_status = $this->mxchat_get_queue_status_data($sitemap_queue_id, 'sitemap'); |
| 2643 |
} |
| 2644 |
|
| 2645 |
if ($pdf_queue_id) { |
| 2646 |
$pdf_status = $this->mxchat_get_queue_status_data($pdf_queue_id, 'pdf'); |
| 2647 |
} |
| 2648 |
|
| 2649 |
$is_processing = |
| 2650 |
($sitemap_status && $sitemap_status['status'] === 'processing') || |
| 2651 |
($pdf_status && $pdf_status['status'] === 'processing'); |
| 2652 |
|
| 2653 |
return array( |
| 2654 |
'sitemap_status' => $sitemap_status, |
| 2655 |
'pdf_status' => $pdf_status, |
| 2656 |
'is_processing' => $is_processing |
| 2657 |
); |
| 2658 |
} |
| 2659 |
|
| 2660 |
/** |
| 2661 |
* AJAX handler to get recent knowledge entries for real-time table updates |
| 2662 |
* UPDATED: Now supports both WordPress DB and Pinecone data sources |
| 2663 |
*/ |
| 2664 |
public function ajax_mxchat_get_recent_entries() { |
| 2665 |
check_ajax_referer('mxchat_entries_nonce', 'nonce'); |
| 2666 |
|
| 2667 |
if (!current_user_can('manage_options')) { |
| 2668 |
wp_send_json_error(array('message' => 'Unauthorized')); |
| 2669 |
return; |
| 2670 |
} |
| 2671 |
|
| 2672 |
global $wpdb; |
| 2673 |
$table_name = $wpdb->prefix . 'mxchat_system_prompt_content'; |
| 2674 |
|
| 2675 |
// Get parameters |
| 2676 |
$last_id = isset($_POST['last_id']) ? absint($_POST['last_id']) : 0; |
| 2677 |
$limit = isset($_POST['limit']) ? min(absint($_POST['limit']), 50) : 10; |
| 2678 |
$bot_id = isset($_POST['bot_id']) ? sanitize_text_field($_POST['bot_id']) : 'default'; |
| 2679 |
|
| 2680 |
// Check if Pinecone is enabled for this bot |
| 2681 |
$pinecone_manager = $this->mxchat_get_pinecone_manager(); |
| 2682 |
$pinecone_options = $pinecone_manager ? $pinecone_manager->mxchat_get_bot_pinecone_options($bot_id) : array(); |
| 2683 |
$use_pinecone = ($pinecone_options['mxchat_use_pinecone'] ?? '0') === '1'; |
| 2684 |
$has_pinecone_api = !empty($pinecone_options['mxchat_pinecone_api_key']); |
| 2685 |
|
| 2686 |
if ($use_pinecone && $has_pinecone_api) { |
| 2687 |
// PINECONE DATA SOURCE - Get grouped entry count (not raw vector count) |
| 2688 |
// Use mxchat_fetch_pinecone_records which returns total_unique_entries |
| 2689 |
$records = $pinecone_manager->mxchat_fetch_pinecone_records($pinecone_options, '', 1, 10, $bot_id, ''); |
| 2690 |
$total_count = $records['total'] ?? 0; |
| 2691 |
|
| 2692 |
// For Pinecone, we don't return individual entries during polling |
| 2693 |
// (entries are already displayed on page load via mxchat_fetch_pinecone_records) |
| 2694 |
// We just return the updated count |
| 2695 |
wp_send_json_success(array( |
| 2696 |
'entries' => array(), |
| 2697 |
'total_count' => absint($total_count), |
| 2698 |
'max_id' => $last_id, |
| 2699 |
'data_source' => 'pinecone' |
| 2700 |
)); |
| 2701 |
return; |
| 2702 |
} |
| 2703 |
|
| 2704 |
// WORDPRESS DB DATA SOURCE |
| 2705 |
// Build query to get entries newer than last_id |
| 2706 |
$where_clauses = array('1=1'); |
| 2707 |
$where_values = array(); |
| 2708 |
|
| 2709 |
if ($last_id > 0) { |
| 2710 |
$where_clauses[] = 'id > %d'; |
| 2711 |
$where_values[] = $last_id; |
| 2712 |
} |
| 2713 |
|
| 2714 |
// Note: WordPress DB table doesn't have bot_id column |
| 2715 |
// Multi-bot filtering is handled via Pinecone namespaces |
| 2716 |
|
| 2717 |
$where_sql = implode(' AND ', $where_clauses); |
| 2718 |
|
| 2719 |
// Get recent entries |
| 2720 |
$query = "SELECT id, article_content, source_url, timestamp |
| 2721 |
FROM $table_name |
| 2722 |
WHERE $where_sql |
| 2723 |
ORDER BY id DESC |
| 2724 |
LIMIT %d"; |
| 2725 |
|
| 2726 |
$where_values[] = $limit; |
| 2727 |
|
| 2728 |
$entries = $wpdb->get_results($wpdb->prepare($query, $where_values)); |
| 2729 |
|
| 2730 |
// Get total count of GROUPED entries (by source_url) - matches pagination display |
| 2731 |
// Count unique source_urls (excluding mxchat:// internal refs) + count of ungrouped rows |
| 2732 |
$total_count = $wpdb->get_var( |
| 2733 |
"SELECT (SELECT COUNT(DISTINCT source_url) FROM {$table_name} WHERE source_url != '' AND source_url NOT LIKE 'mxchat://%') + |
| 2734 |
(SELECT COUNT(*) FROM {$table_name} WHERE source_url = '' OR source_url IS NULL OR source_url LIKE 'mxchat://%')" |
| 2735 |
); |
| 2736 |
|
| 2737 |
// Format entries for response |
| 2738 |
$formatted_entries = array(); |
| 2739 |
$preview_length = 150; |
| 2740 |
foreach ($entries as $entry) { |
| 2741 |
// Parse chunk metadata using the proper chunker method (same as initial page load) |
| 2742 |
if (class_exists('MxChat_Chunker')) { |
| 2743 |
$chunk_meta = MxChat_Chunker::parse_stored_chunk($entry->article_content); |
| 2744 |
$display_content = $chunk_meta['text']; |
| 2745 |
$chunk_metadata = $chunk_meta['metadata']; |
| 2746 |
} else { |
| 2747 |
$display_content = $entry->article_content; |
| 2748 |
$chunk_metadata = array(); |
| 2749 |
} |
| 2750 |
|
| 2751 |
$content_preview = mb_strlen($display_content) > $preview_length |
| 2752 |
? mb_substr($display_content, 0, $preview_length) . '...' |
| 2753 |
: $display_content; |
| 2754 |
|
| 2755 |
$formatted_entries[] = array( |
| 2756 |
'id' => $entry->id, |
| 2757 |
'preview' => esc_html($content_preview), |
| 2758 |
'full_content' => wp_kses_post(wpautop($display_content)), |
| 2759 |
'content_length' => mb_strlen($display_content), |
| 2760 |
'preview_length' => $preview_length, |
| 2761 |
'source_url' => $entry->source_url, |
| 2762 |
'has_link' => !empty($entry->source_url) && strpos($entry->source_url, 'mxchat://') !== 0, |
| 2763 |
'chunk_metadata' => $chunk_metadata, |
| 2764 |
'bot_id' => $entry->bot_id ?? 'default', |
| 2765 |
'edit_nonce' => wp_create_nonce('mxchat_edit_entry_nonce'), |
| 2766 |
'delete_nonce' => wp_create_nonce('mxchat_delete_prompt_nonce') |
| 2767 |
); |
| 2768 |
} |
| 2769 |
|
| 2770 |
wp_send_json_success(array( |
| 2771 |
'entries' => $formatted_entries, |
| 2772 |
'total_count' => absint($total_count), |
| 2773 |
'max_id' => !empty($entries) ? $entries[0]->id : $last_id, |
| 2774 |
'data_source' => 'wordpress' |
| 2775 |
)); |
| 2776 |
} |
| 2777 |
|
| 2778 |
/** |
| 2779 |
* Get Pinecone total count from stats API |
| 2780 |
* Helper function for ajax_mxchat_get_recent_entries |
| 2781 |
*/ |
| 2782 |
private function mxchat_get_pinecone_count_from_stats($pinecone_options) { |
| 2783 |
$api_key = $pinecone_options['mxchat_pinecone_api_key'] ?? ''; |
| 2784 |
$host = $pinecone_options['mxchat_pinecone_host'] ?? ''; |
| 2785 |
$namespace = $pinecone_options['mxchat_pinecone_namespace'] ?? ''; |
| 2786 |
|
| 2787 |
if (empty($api_key) || empty($host)) { |
| 2788 |
return 0; |
| 2789 |
} |
| 2790 |
|
| 2791 |
try { |
| 2792 |
$stats_url = "https://{$host}/describe_index_stats"; |
| 2793 |
|
| 2794 |
$response = wp_remote_post($stats_url, array( |
| 2795 |
'headers' => array( |
| 2796 |
'Api-Key' => $api_key, |
| 2797 |
'Content-Type' => 'application/json' |
| 2798 |
), |
| 2799 |
'body' => '{}', |
| 2800 |
'timeout' => 10 |
| 2801 |
)); |
| 2802 |
|
| 2803 |
if (!is_wp_error($response) && wp_remote_retrieve_response_code($response) === 200) { |
| 2804 |
$body = wp_remote_retrieve_body($response); |
| 2805 |
$stats_data = json_decode($body, true); |
| 2806 |
|
| 2807 |
// If namespace is specified, get count from that specific namespace |
| 2808 |
if (!empty($namespace) && isset($stats_data['namespaces'][$namespace]['vectorCount'])) { |
| 2809 |
return intval($stats_data['namespaces'][$namespace]['vectorCount']); |
| 2810 |
} |
| 2811 |
|
| 2812 |
// If no namespace specified or namespace not found in response, use total |
| 2813 |
return intval($stats_data['totalVectorCount'] ?? 0); |
| 2814 |
} |
| 2815 |
|
| 2816 |
return 0; |
| 2817 |
|
| 2818 |
} catch (Exception $e) { |
| 2819 |
return 0; |
| 2820 |
} |
| 2821 |
} |
| 2822 |
|
| 2823 |
/** |
| 2824 |
* AJAX handler to refresh Pinecone entries table via AJAX |
| 2825 |
* Returns the table HTML for updating the UI without a full page reload |
| 2826 |
*/ |
| 2827 |
public function ajax_mxchat_refresh_pinecone_entries() { |
| 2828 |
check_ajax_referer('mxchat_entries_nonce', 'nonce'); |
| 2829 |
|
| 2830 |
if (!current_user_can('manage_options')) { |
| 2831 |
wp_send_json_error(array('message' => 'Unauthorized')); |
| 2832 |
return; |
| 2833 |
} |
| 2834 |
|
| 2835 |
$bot_id = isset($_POST['bot_id']) ? sanitize_text_field($_POST['bot_id']) : 'default'; |
| 2836 |
$page = isset($_POST['page']) ? absint($_POST['page']) : 1; |
| 2837 |
$per_page = 25; |
| 2838 |
$search_query = isset($_POST['search']) ? sanitize_text_field($_POST['search']) : ''; |
| 2839 |
$content_type_filter = isset($_POST['content_type']) ? sanitize_key($_POST['content_type']) : ''; |
| 2840 |
|
| 2841 |
// Get Pinecone manager and options |
| 2842 |
$pinecone_manager = $this->mxchat_get_pinecone_manager(); |
| 2843 |
if (!$pinecone_manager) { |
| 2844 |
wp_send_json_error(array('message' => 'Pinecone manager not available')); |
| 2845 |
return; |
| 2846 |
} |
| 2847 |
|
| 2848 |
$pinecone_options = $pinecone_manager->mxchat_get_bot_pinecone_options($bot_id); |
| 2849 |
$use_pinecone = ($pinecone_options['mxchat_use_pinecone'] ?? '0') === '1'; |
| 2850 |
$pinecone_api_key = $pinecone_options['mxchat_pinecone_api_key'] ?? ''; |
| 2851 |
|
| 2852 |
if (!$use_pinecone || empty($pinecone_api_key)) { |
| 2853 |
wp_send_json_error(array('message' => 'Pinecone not configured')); |
| 2854 |
return; |
| 2855 |
} |
| 2856 |
|
| 2857 |
// Fetch records from Pinecone |
| 2858 |
$records = $pinecone_manager->mxchat_fetch_pinecone_records($pinecone_options, $search_query, $page, $per_page, $bot_id, $content_type_filter); |
| 2859 |
$prompts = $records['data'] ?? array(); |
| 2860 |
$total_records = $records['total'] ?? 0; |
| 2861 |
|
| 2862 |
// Preprocess Pinecone records — set chunk_metadata and display_content |
| 2863 |
// (matches admin-knowledge-page.php preprocessing) |
| 2864 |
foreach ($prompts as $prompt) { |
| 2865 |
if (isset($prompt->chunk_index) && $prompt->chunk_index !== null) { |
| 2866 |
$prompt->chunk_metadata = array( |
| 2867 |
'chunk_index' => intval($prompt->chunk_index), |
| 2868 |
'total_chunks' => isset($prompt->total_chunks) ? intval($prompt->total_chunks) : null, |
| 2869 |
'is_chunked' => isset($prompt->is_chunked) ? (bool) $prompt->is_chunked : true, |
| 2870 |
'source_url' => $prompt->source_url ?? '' |
| 2871 |
); |
| 2872 |
$prompt->display_content = $prompt->article_content; |
| 2873 |
} else { |
| 2874 |
$prompt->chunk_metadata = array(); |
| 2875 |
$prompt->display_content = $prompt->article_content ?? ''; |
| 2876 |
} |
| 2877 |
} |
| 2878 |
|
| 2879 |
// Group prompts by source_url |
| 2880 |
$grouped_prompts = array(); |
| 2881 |
foreach ($prompts as $prompt) { |
| 2882 |
$source_url = ''; |
| 2883 |
if (!empty($prompt->chunk_metadata['source_url'])) { |
| 2884 |
$source_url = $prompt->chunk_metadata['source_url']; |
| 2885 |
} elseif (!empty($prompt->source_url)) { |
| 2886 |
$source_url = $prompt->source_url; |
| 2887 |
} |
| 2888 |
|
| 2889 |
if (!empty($source_url)) { |
| 2890 |
if (!isset($grouped_prompts[$source_url])) { |
| 2891 |
$grouped_prompts[$source_url] = array(); |
| 2892 |
} |
| 2893 |
$grouped_prompts[$source_url][] = $prompt; |
| 2894 |
} else { |
| 2895 |
$grouped_prompts['_ungrouped_' . $prompt->id] = array($prompt); |
| 2896 |
} |
| 2897 |
} |
| 2898 |
|
| 2899 |
// Sort each group by chunk_index |
| 2900 |
foreach ($grouped_prompts as $source_url => &$group) { |
| 2901 |
usort($group, function($a, $b) { |
| 2902 |
$index_a = isset($a->chunk_metadata['chunk_index']) ? intval($a->chunk_metadata['chunk_index']) : 0; |
| 2903 |
$index_b = isset($b->chunk_metadata['chunk_index']) ? intval($b->chunk_metadata['chunk_index']) : 0; |
| 2904 |
return $index_a - $index_b; |
| 2905 |
}); |
| 2906 |
} |
| 2907 |
unset($group); |
| 2908 |
|
| 2909 |
// Build HTML for the table rows - must match admin-knowledge-page.php structure exactly |
| 2910 |
ob_start(); |
| 2911 |
$display_index = 0; |
| 2912 |
$current_page = $page; |
| 2913 |
$data_source = 'pinecone'; |
| 2914 |
$current_bot_id = $bot_id; |
| 2915 |
$preview_length = 150; |
| 2916 |
|
| 2917 |
if (empty($grouped_prompts)) { |
| 2918 |
echo '<tr><td colspan="4" style="padding: 40px; text-align: center; color: var(--mxch-text-muted);">'; |
| 2919 |
esc_html_e('No knowledge entries found in Pinecone.', 'mxchat'); |
| 2920 |
echo '</td></tr>'; |
| 2921 |
} else { |
| 2922 |
foreach ($grouped_prompts as $source_url => $group) { |
| 2923 |
$chunk_count = count($group); |
| 2924 |
$first_prompt = $group[0]; |
| 2925 |
$display_index++; |
| 2926 |
|
| 2927 |
if ($chunk_count > 1) { |
| 2928 |
// Multiple chunks - show grouped row with expand button |
| 2929 |
$group_id = 'group-' . md5($source_url); |
| 2930 |
?> |
| 2931 |
<tr id="prompt-<?php echo esc_attr($first_prompt->id); ?>" |
| 2932 |
class="mxchat-chunk-group-header" |
| 2933 |
data-source="<?php echo esc_attr($data_source); ?>" |
| 2934 |
data-group-id="<?php echo esc_attr($group_id); ?>" |
| 2935 |
style="border-bottom: 1px solid var(--mxch-card-border); background: rgba(33, 150, 243, 0.02);"> |
| 2936 |
<td style="padding: 12px 16px; text-align: center;"> |
| 2937 |
<input type="checkbox" |
| 2938 |
class="mxchat-entry-checkbox" |
| 2939 |
data-entry-id="<?php echo esc_attr($first_prompt->id); ?>" |
| 2940 |
data-source="<?php echo esc_attr($data_source); ?>" |
| 2941 |
data-source-url="<?php echo esc_attr($source_url); ?>" |
| 2942 |
data-is-group="true" |
| 2943 |
data-chunk-count="<?php echo esc_attr($chunk_count); ?>"> |
| 2944 |
</td> |
| 2945 |
<td style="padding: 12px 16px; font-size: 13px;"> |
| 2946 |
<?php echo esc_html($display_index + (($current_page - 1) * $per_page)); ?> |
| 2947 |
</td> |
| 2948 |
<td class="mxchat-content-cell" style="padding: 12px 16px; font-size: 13px;"> |
| 2949 |
<div class="mxchat-chunk-group-info"> |
| 2950 |
<button type="button" class="mxchat-chunk-toggle" data-group-id="<?php echo esc_attr($group_id); ?>"> |
| 2951 |
<span class="dashicons dashicons-arrow-right-alt2"></span> |
| 2952 |
</button> |
| 2953 |
<span class="mxchat-chunk-badge"><?php echo esc_html($chunk_count); ?> <?php esc_html_e('chunks', 'mxchat'); ?></span> |
| 2954 |
<span class="mxchat-chunk-preview"> |
| 2955 |
<?php |
| 2956 |
$parent_content = isset($first_prompt->display_content) ? $first_prompt->display_content : (isset($first_prompt->article_content) ? $first_prompt->article_content : ''); |
| 2957 |
$content_preview = mb_substr($parent_content, 0, 100); |
| 2958 |
echo esc_html($content_preview . '...'); |
| 2959 |
?> |
| 2960 |
</span> |
| 2961 |
</div> |
| 2962 |
</td> |
| 2963 |
<td class="mxchat-url-cell" style="padding: 12px 16px; font-size: 13px;"> |
| 2964 |
<?php if (!empty($source_url) && strpos($source_url, 'mxchat://') !== 0 && strpos($source_url, '_ungrouped_') !== 0) : ?> |
| 2965 |
<a href="<?php echo esc_url($source_url); ?>" target="_blank" style="color: var(--mxch-primary); text-decoration: none;"> |
| 2966 |
<span class="dashicons dashicons-external" style="font-size: 14px;"></span> |
| 2967 |
<?php esc_html_e('View Source', 'mxchat'); ?> |
| 2968 |
</a> |
| 2969 |
<?php else : ?> |
| 2970 |
<span style="color: var(--mxch-text-muted);"><?php esc_html_e('Manual Content', 'mxchat'); ?></span> |
| 2971 |
<?php endif; ?> |
| 2972 |
</td> |
| 2973 |
<td class="mxchat-actions-cell" style="padding: 12px 16px; white-space: nowrap;"> |
| 2974 |
<button type="button" |
| 2975 |
class="mxch-btn mxch-btn-ghost mxch-btn-sm mxchat-inspect-entry-btn" |
| 2976 |
data-source-url="<?php echo esc_attr($source_url); ?>" |
| 2977 |
data-entry-id="<?php echo esc_attr($first_prompt->id); ?>" |
| 2978 |
data-data-source="<?php echo esc_attr($data_source); ?>" |
| 2979 |
data-bot-id="<?php echo esc_attr($current_bot_id); ?>" |
| 2980 |
data-nonce="<?php echo wp_create_nonce('mxchat_inspect_entry_nonce'); ?>" |
| 2981 |
title="<?php esc_attr_e('View indexed content', 'mxchat'); ?>"> |
| 2982 |
<span class="dashicons dashicons-visibility" style="font-size: 14px;"></span> |
| 2983 |
</button> |
| 2984 |
<button type="button" |
| 2985 |
class="mxch-btn mxch-btn-ghost mxch-btn-sm mxchat-edit-entry-btn" |
| 2986 |
data-source-url="<?php echo esc_attr($source_url); ?>" |
| 2987 |
data-entry-id="<?php echo esc_attr($first_prompt->id); ?>" |
| 2988 |
data-data-source="<?php echo esc_attr($data_source); ?>" |
| 2989 |
data-bot-id="<?php echo esc_attr($current_bot_id); ?>" |
| 2990 |
data-nonce="<?php echo wp_create_nonce('mxchat_edit_entry_nonce'); ?>" |
| 2991 |
title="<?php esc_attr_e('Edit content', 'mxchat'); ?>"> |
| 2992 |
<span class="dashicons dashicons-edit" style="font-size: 14px;"></span> |
| 2993 |
</button> |
| 2994 |
<button type="button" |
| 2995 |
class="mxch-btn mxch-btn-ghost mxch-btn-sm delete-button-group" |
| 2996 |
data-source-url="<?php echo esc_attr($source_url); ?>" |
| 2997 |
data-chunk-count="<?php echo esc_attr($chunk_count); ?>" |
| 2998 |
data-data-source="<?php echo esc_attr($data_source); ?>" |
| 2999 |
data-bot-id="<?php echo esc_attr($current_bot_id); ?>" |
| 3000 |
data-nonce="<?php echo wp_create_nonce('mxchat_delete_chunks_nonce'); ?>" |
| 3001 |
style="color: var(--mxch-error);" |
| 3002 |
title="<?php esc_attr_e('Delete all chunks', 'mxchat'); ?>"> |
| 3003 |
<span class="dashicons dashicons-trash" style="font-size: 14px;"></span> |
| 3004 |
</button> |
| 3005 |
</td> |
| 3006 |
</tr> |
| 3007 |
<?php |
| 3008 |
// Render hidden chunk rows |
| 3009 |
foreach ($group as $chunk_index => $chunk) { |
| 3010 |
$meta_chunk_index = isset($chunk->chunk_metadata['chunk_index']) ? intval($chunk->chunk_metadata['chunk_index']) : $chunk_index; |
| 3011 |
$meta_total_chunks = isset($chunk->chunk_metadata['total_chunks']) ? intval($chunk->chunk_metadata['total_chunks']) : $chunk_count; |
| 3012 |
$content = isset($chunk->display_content) ? $chunk->display_content : (isset($chunk->article_content) ? $chunk->article_content : ''); |
| 3013 |
$content_preview = mb_strlen($content) > $preview_length |
| 3014 |
? mb_substr($content, 0, $preview_length) . '...' |
| 3015 |
: $content; |
| 3016 |
?> |
| 3017 |
<tr id="prompt-<?php echo esc_attr($chunk->id); ?>" |
| 3018 |
class="mxchat-chunk-row <?php echo esc_attr($group_id); ?>" |
| 3019 |
data-source="<?php echo esc_attr($data_source); ?>" |
| 3020 |
style="display: none; background: #f8f9fa; border-bottom: 1px solid var(--mxch-card-border);"> |
| 3021 |
<td style="padding: 12px 16px; text-align: center;"> |
| 3022 |
<!-- Checkbox column placeholder for chunks (managed by group) --> |
| 3023 |
</td> |
| 3024 |
<td style="padding: 12px 16px 12px 30px; font-size: 13px;"> |
| 3025 |
<!-- Hidden ID column for chunks --> |
| 3026 |
</td> |
| 3027 |
<td class="mxchat-content-cell" style="padding: 12px 16px; font-size: 13px;"> |
| 3028 |
<div class="mxchat-accordion-wrapper"> |
| 3029 |
<div class="mxchat-content-preview"> |
| 3030 |
<span class="mxchat-chunk-indicator" style="margin-right: 10px; color: var(--mxch-text-secondary); font-size: 12px;"> |
| 3031 |
<?php printf(esc_html__('Chunk %d of %d', 'mxchat'), $meta_chunk_index + 1, $meta_total_chunks); ?> |
| 3032 |
</span> |
| 3033 |
<span class="preview-text"><?php echo esc_html($content_preview); ?></span> |
| 3034 |
<?php if (mb_strlen($content) > $preview_length) : ?> |
| 3035 |
<button class="mxchat-expand-toggle" type="button"> |
| 3036 |
<span class="dashicons dashicons-arrow-down-alt2"></span> |
| 3037 |
</button> |
| 3038 |
<?php endif; ?> |
| 3039 |
</div> |
| 3040 |
<div class="mxchat-content-full" style="display: none;"> |
| 3041 |
<div class="content-view"> |
| 3042 |
<?php |
| 3043 |
if (preg_match('/[\x{0590}-\x{05FF}]/u', $content)) { |
| 3044 |
echo '<div dir="rtl" lang="he" class="rtl-content">'; |
| 3045 |
echo wp_kses_post(wpautop($content)); |
| 3046 |
echo '</div>'; |
| 3047 |
} else { |
| 3048 |
echo wp_kses_post(wpautop($content)); |
| 3049 |
} |
| 3050 |
?> |
| 3051 |
</div> |
| 3052 |
</div> |
| 3053 |
</div> |
| 3054 |
</td> |
| 3055 |
<td class="mxchat-url-cell" style="padding: 12px 16px; font-size: 13px;"> |
| 3056 |
<span class="mxchat-chunk-label" style="color: var(--mxch-text-muted);"><?php esc_html_e('Same as parent', 'mxchat'); ?></span> |
| 3057 |
</td> |
| 3058 |
<td class="mxchat-actions-cell" style="padding: 12px 16px;"> |
| 3059 |
<span class="mxchat-chunk-label" style="color: var(--mxch-text-muted);"><?php esc_html_e('Managed by group', 'mxchat'); ?></span> |
| 3060 |
</td> |
| 3061 |
</tr> |
| 3062 |
<?php |
| 3063 |
} |
| 3064 |
} else { |
| 3065 |
// Single entry - display normally with accordion |
| 3066 |
$prompt = $first_prompt; |
| 3067 |
$content = isset($prompt->display_content) ? $prompt->display_content : (isset($prompt->article_content) ? $prompt->article_content : ''); |
| 3068 |
$content_preview = mb_strlen($content) > $preview_length |
| 3069 |
? mb_substr($content, 0, $preview_length) . '...' |
| 3070 |
: $content; |
| 3071 |
?> |
| 3072 |
<tr id="prompt-<?php echo esc_attr($prompt->id); ?>" |
| 3073 |
data-source="<?php echo esc_attr($data_source); ?>" |
| 3074 |
style="border-bottom: 1px solid var(--mxch-card-border); background: rgba(33, 150, 243, 0.02);"> |
| 3075 |
<td style="padding: 12px 16px; text-align: center;"> |
| 3076 |
<input type="checkbox" |
| 3077 |
class="mxchat-entry-checkbox" |
| 3078 |
data-entry-id="<?php echo esc_attr($prompt->id); ?>" |
| 3079 |
data-source="<?php echo esc_attr($data_source); ?>" |
| 3080 |
data-source-url="<?php echo esc_attr($prompt->source_url ?? ''); ?>" |
| 3081 |
data-is-group="false" |
| 3082 |
data-chunk-count="1"> |
| 3083 |
</td> |
| 3084 |
<td style="padding: 12px 16px; font-size: 13px;"> |
| 3085 |
<?php echo esc_html($display_index + (($current_page - 1) * $per_page)); ?> |
| 3086 |
</td> |
| 3087 |
<td class="mxchat-content-cell" style="padding: 12px 16px; font-size: 13px;"> |
| 3088 |
<div class="mxchat-accordion-wrapper"> |
| 3089 |
<div class="mxchat-content-preview"> |
| 3090 |
<span class="preview-text"><?php echo esc_html($content_preview); ?></span> |
| 3091 |
<?php if (mb_strlen($content) > $preview_length) : ?> |
| 3092 |
<button class="mxchat-expand-toggle" type="button"> |
| 3093 |
<span class="dashicons dashicons-arrow-down-alt2"></span> |
| 3094 |
</button> |
| 3095 |
<?php endif; ?> |
| 3096 |
</div> |
| 3097 |
<div class="mxchat-content-full" style="display: none;"> |
| 3098 |
<div class="content-view"> |
| 3099 |
<?php |
| 3100 |
if (preg_match('/[\x{0590}-\x{05FF}]/u', $content)) { |
| 3101 |
echo '<div dir="rtl" lang="he" class="rtl-content">'; |
| 3102 |
echo wp_kses_post(wpautop($content)); |
| 3103 |
echo '</div>'; |
| 3104 |
} else { |
| 3105 |
echo wp_kses_post(wpautop($content)); |
| 3106 |
} |
| 3107 |
?> |
| 3108 |
</div> |
| 3109 |
</div> |
| 3110 |
</div> |
| 3111 |
</td> |
| 3112 |
<td class="mxchat-url-cell" style="padding: 12px 16px; font-size: 13px;"> |
| 3113 |
<?php |
| 3114 |
$actual_source = $source_url; |
| 3115 |
if (strpos($source_url, '_ungrouped_') === 0) { |
| 3116 |
$actual_source = $prompt->source_url ?? ''; |
| 3117 |
} |
| 3118 |
if (!empty($actual_source) && strpos($actual_source, 'mxchat://') !== 0) : ?> |
| 3119 |
<a href="<?php echo esc_url($actual_source); ?>" target="_blank" style="color: var(--mxch-primary); text-decoration: none;"> |
| 3120 |
<span class="dashicons dashicons-external" style="font-size: 14px;"></span> |
| 3121 |
<?php esc_html_e('View', 'mxchat'); ?> |
| 3122 |
</a> |
| 3123 |
<?php else : ?> |
| 3124 |
<span style="color: var(--mxch-text-muted);"><?php esc_html_e('Manual', 'mxchat'); ?></span> |
| 3125 |
<?php endif; ?> |
| 3126 |
</td> |
| 3127 |
<td style="padding: 12px 16px; white-space: nowrap;"> |
| 3128 |
<button type="button" |
| 3129 |
class="mxch-btn mxch-btn-ghost mxch-btn-sm mxchat-inspect-entry-btn" |
| 3130 |
data-source-url="<?php echo esc_attr($prompt->source_url ?? ''); ?>" |
| 3131 |
data-entry-id="<?php echo esc_attr($prompt->id); ?>" |
| 3132 |
data-data-source="<?php echo esc_attr($data_source); ?>" |
| 3133 |
data-bot-id="<?php echo esc_attr($current_bot_id); ?>" |
| 3134 |
data-nonce="<?php echo wp_create_nonce('mxchat_inspect_entry_nonce'); ?>" |
| 3135 |
title="<?php esc_attr_e('View indexed content', 'mxchat'); ?>"> |
| 3136 |
<span class="dashicons dashicons-visibility" style="font-size: 14px;"></span> |
| 3137 |
</button> |
| 3138 |
<button type="button" |
| 3139 |
class="mxch-btn mxch-btn-ghost mxch-btn-sm mxchat-edit-entry-btn" |
| 3140 |
data-source-url="<?php echo esc_attr($prompt->source_url ?? ''); ?>" |
| 3141 |
data-entry-id="<?php echo esc_attr($prompt->id); ?>" |
| 3142 |
data-data-source="<?php echo esc_attr($data_source); ?>" |
| 3143 |
data-bot-id="<?php echo esc_attr($current_bot_id); ?>" |
| 3144 |
data-nonce="<?php echo wp_create_nonce('mxchat_edit_entry_nonce'); ?>" |
| 3145 |
title="<?php esc_attr_e('Edit content', 'mxchat'); ?>"> |
| 3146 |
<span class="dashicons dashicons-edit" style="font-size: 14px;"></span> |
| 3147 |
</button> |
| 3148 |
<button type="button" class="mxch-btn mxch-btn-ghost mxch-btn-sm delete-button-ajax" data-vector-id="<?php echo esc_attr($prompt->id); ?>" data-bot-id="<?php echo esc_attr($current_bot_id); ?>" data-nonce="<?php echo wp_create_nonce('mxchat_delete_pinecone_prompt_nonce'); ?>" style="color: var(--mxch-error);"> |
| 3149 |
<span class="dashicons dashicons-trash" style="font-size: 14px;"></span> |
| 3150 |
</button> |
| 3151 |
</td> |
| 3152 |
</tr> |
| 3153 |
<?php |
| 3154 |
} |
| 3155 |
} |
| 3156 |
} |
| 3157 |
$html = ob_get_clean(); |
| 3158 |
|
| 3159 |
// Generate pagination HTML for Pinecone |
| 3160 |
$total_pages = ceil($total_records / $per_page); |
| 3161 |
$pagination_html = ''; |
| 3162 |
if ($total_pages > 1) { |
| 3163 |
$pagination_html = '<div class="mxchat-ajax-pagination" data-current-page="' . esc_attr($page) . '" data-total-pages="' . esc_attr($total_pages) . '" data-search="' . esc_attr($search_query) . '" data-content-type="' . esc_attr($content_type_filter) . '">'; |
| 3164 |
|
| 3165 |
// Previous button |
| 3166 |
if ($page > 1) { |
| 3167 |
$pagination_html .= '<a href="#" class="mxchat-page-link" data-page="' . ($page - 1) . '">' . esc_html__('« Previous', 'mxchat') . '</a> '; |
| 3168 |
} |
| 3169 |
|
| 3170 |
// Page numbers |
| 3171 |
$start_page = max(1, $page - 2); |
| 3172 |
$end_page = min($total_pages, $page + 2); |
| 3173 |
|
| 3174 |
if ($start_page > 1) { |
| 3175 |
$pagination_html .= '<a href="#" class="mxchat-page-link" data-page="1">1</a> '; |
| 3176 |
if ($start_page > 2) { |
| 3177 |
$pagination_html .= '<span class="mxchat-page-dots">...</span> '; |
| 3178 |
} |
| 3179 |
} |
| 3180 |
|
| 3181 |
for ($i = $start_page; $i <= $end_page; $i++) { |
| 3182 |
if ($i == $page) { |
| 3183 |
$pagination_html .= '<span class="mxchat-page-current">' . $i . '</span> '; |
| 3184 |
} else { |
| 3185 |
$pagination_html .= '<a href="#" class="mxchat-page-link" data-page="' . $i . '">' . $i . '</a> '; |
| 3186 |
} |
| 3187 |
} |
| 3188 |
|
| 3189 |
if ($end_page < $total_pages) { |
| 3190 |
if ($end_page < $total_pages - 1) { |
| 3191 |
$pagination_html .= '<span class="mxchat-page-dots">...</span> '; |
| 3192 |
} |
| 3193 |
$pagination_html .= '<a href="#" class="mxchat-page-link" data-page="' . $total_pages . '">' . $total_pages . '</a> '; |
| 3194 |
} |
| 3195 |
|
| 3196 |
// Next button |
| 3197 |
if ($page < $total_pages) { |
| 3198 |
$pagination_html .= '<a href="#" class="mxchat-page-link" data-page="' . ($page + 1) . '">' . esc_html__('Next »', 'mxchat') . '</a>'; |
| 3199 |
} |
| 3200 |
|
| 3201 |
$pagination_html .= '</div>'; |
| 3202 |
} |
| 3203 |
|
| 3204 |
wp_send_json_success(array( |
| 3205 |
'html' => $html, |
| 3206 |
'pagination_html' => $pagination_html, |
| 3207 |
'total_count' => $total_records, |
| 3208 |
'total_pages' => $total_pages, |
| 3209 |
'page' => $page, |
| 3210 |
'per_page' => $per_page, |
| 3211 |
'data_source' => 'pinecone' |
| 3212 |
)); |
| 3213 |
} |
| 3214 |
|
| 3215 |
/** |
| 3216 |
* AJAX handler for pagination - handles both WordPress DB and Pinecone data sources |
| 3217 |
* Returns paginated entries without requiring a full page reload |
| 3218 |
*/ |
| 3219 |
public function ajax_mxchat_paginate_entries() { |
| 3220 |
check_ajax_referer('mxchat_entries_nonce', 'nonce'); |
| 3221 |
|
| 3222 |
if (!current_user_can('manage_options')) { |
| 3223 |
wp_send_json_error(array('message' => 'Unauthorized')); |
| 3224 |
return; |
| 3225 |
} |
| 3226 |
|
| 3227 |
$bot_id = isset($_POST['bot_id']) ? sanitize_text_field($_POST['bot_id']) : 'default'; |
| 3228 |
$page = isset($_POST['page']) ? absint($_POST['page']) : 1; |
| 3229 |
$search_query = isset($_POST['search']) ? sanitize_text_field($_POST['search']) : ''; |
| 3230 |
$content_type_filter = isset($_POST['content_type']) ? sanitize_key($_POST['content_type']) : ''; |
| 3231 |
$per_page = 25; |
| 3232 |
|
| 3233 |
// Check if Pinecone is enabled for this bot |
| 3234 |
$pinecone_manager = $this->mxchat_get_pinecone_manager(); |
| 3235 |
$pinecone_options = $pinecone_manager ? $pinecone_manager->mxchat_get_bot_pinecone_options($bot_id) : array(); |
| 3236 |
$use_pinecone = ($pinecone_options['mxchat_use_pinecone'] ?? '0') === '1'; |
| 3237 |
$has_pinecone_api = !empty($pinecone_options['mxchat_pinecone_api_key']); |
| 3238 |
|
| 3239 |
if ($use_pinecone && $has_pinecone_api) { |
| 3240 |
// Delegate to Pinecone pagination handler (pass search params) |
| 3241 |
$_POST['page'] = $page; |
| 3242 |
$_POST['search'] = $search_query; |
| 3243 |
$_POST['content_type'] = $content_type_filter; |
| 3244 |
$this->ajax_mxchat_refresh_pinecone_entries(); |
| 3245 |
return; |
| 3246 |
} |
| 3247 |
|
| 3248 |
// WordPress DB pagination - MUST match initial page load logic exactly |
| 3249 |
global $wpdb; |
| 3250 |
$table_name = $wpdb->prefix . 'mxchat_system_prompt_content'; |
| 3251 |
$offset = ($page - 1) * $per_page; |
| 3252 |
|
| 3253 |
// Build WHERE clause for search and content type filtering |
| 3254 |
$where_clauses = array(); |
| 3255 |
$where_values = array(); |
| 3256 |
|
| 3257 |
if ($search_query) { |
| 3258 |
$where_clauses[] = "article_content LIKE %s"; |
| 3259 |
$where_values[] = '%' . $wpdb->esc_like($search_query) . '%'; |
| 3260 |
} |
| 3261 |
|
| 3262 |
if ($content_type_filter) { |
| 3263 |
switch ($content_type_filter) { |
| 3264 |
case 'manual': |
| 3265 |
$where_clauses[] = "(source_url = '' OR source_url IS NULL OR source_url LIKE 'mxchat://%')"; |
| 3266 |
break; |
| 3267 |
case 'pdf': |
| 3268 |
$where_clauses[] = "source_url LIKE '%.pdf'"; |
| 3269 |
break; |
| 3270 |
case 'url': |
| 3271 |
$where_clauses[] = "source_url != '' AND source_url IS NOT NULL AND source_url NOT LIKE 'mxchat://%' AND source_url NOT LIKE '%.pdf'"; |
| 3272 |
break; |
| 3273 |
} |
| 3274 |
} |
| 3275 |
|
| 3276 |
$where_sql = !empty($where_clauses) ? 'WHERE ' . implode(' AND ', $where_clauses) : ''; |
| 3277 |
|
| 3278 |
// Count grouped entries with filters applied |
| 3279 |
if (!empty($where_values)) { |
| 3280 |
$count_args = array_merge($where_values, $where_values); |
| 3281 |
$count_query = $wpdb->prepare( |
| 3282 |
"SELECT (SELECT COUNT(DISTINCT source_url) FROM {$table_name} {$where_sql} AND source_url != '' AND source_url NOT LIKE 'mxchat://%') + |
| 3283 |
(SELECT COUNT(*) FROM {$table_name} {$where_sql} AND (source_url = '' OR source_url IS NULL OR source_url LIKE 'mxchat://%'))", |
| 3284 |
...$count_args |
| 3285 |
); |
| 3286 |
$total_records = $wpdb->get_var($count_query); |
| 3287 |
} else if (!empty($where_sql)) { |
| 3288 |
// Content type filter only (no search), no prepared values needed |
| 3289 |
$total_records = $wpdb->get_var( |
| 3290 |
"SELECT (SELECT COUNT(DISTINCT source_url) FROM {$table_name} {$where_sql} AND source_url != '' AND source_url NOT LIKE 'mxchat://%') + |
| 3291 |
(SELECT COUNT(*) FROM {$table_name} {$where_sql} AND (source_url = '' OR source_url IS NULL OR source_url LIKE 'mxchat://%'))" |
| 3292 |
); |
| 3293 |
} else { |
| 3294 |
// No filters |
| 3295 |
$total_records = $wpdb->get_var( |
| 3296 |
"SELECT (SELECT COUNT(DISTINCT source_url) FROM {$table_name} WHERE source_url != '' AND source_url NOT LIKE 'mxchat://%') + |
| 3297 |
(SELECT COUNT(*) FROM {$table_name} WHERE source_url = '' OR source_url IS NULL OR source_url LIKE 'mxchat://%')" |
| 3298 |
); |
| 3299 |
} |
| 3300 |
$total_pages = ceil($total_records / $per_page); |
| 3301 |
|
| 3302 |
// Step 1: Get unique source_urls for this page (ordered by latest timestamp) with filters |
| 3303 |
if (!empty($where_values)) { |
| 3304 |
$query_args = array_merge($where_values, array($per_page, $offset)); |
| 3305 |
$urls_query = $wpdb->prepare( |
| 3306 |
"SELECT source_url, MAX(timestamp) as latest_ts FROM {$table_name} |
| 3307 |
{$where_sql} |
| 3308 |
GROUP BY source_url ORDER BY latest_ts DESC LIMIT %d OFFSET %d", |
| 3309 |
...$query_args |
| 3310 |
); |
| 3311 |
} else if (!empty($where_sql)) { |
| 3312 |
$urls_query = $wpdb->prepare( |
| 3313 |
"SELECT source_url, MAX(timestamp) as latest_ts FROM {$table_name} |
| 3314 |
{$where_sql} |
| 3315 |
GROUP BY source_url ORDER BY latest_ts DESC LIMIT %d OFFSET %d", |
| 3316 |
$per_page, $offset |
| 3317 |
); |
| 3318 |
} else { |
| 3319 |
$urls_query = $wpdb->prepare( |
| 3320 |
"SELECT source_url, MAX(timestamp) as latest_ts FROM {$table_name} |
| 3321 |
GROUP BY source_url ORDER BY latest_ts DESC LIMIT %d OFFSET %d", |
| 3322 |
$per_page, $offset |
| 3323 |
); |
| 3324 |
} |
| 3325 |
$page_urls = $wpdb->get_results($urls_query); |
| 3326 |
|
| 3327 |
// Step 2: Build list of source_urls to fetch |
| 3328 |
$url_list = array(); |
| 3329 |
$url_order_map = array(); |
| 3330 |
$order_index = 0; |
| 3331 |
foreach ($page_urls as $url_row) { |
| 3332 |
$url = $url_row->source_url; |
| 3333 |
$url_list[] = $url; |
| 3334 |
$url_order_map[$url] = $order_index++; |
| 3335 |
} |
| 3336 |
|
| 3337 |
// Step 3: Fetch all rows for these source_urls (with search filter if applicable) |
| 3338 |
$prompts = array(); |
| 3339 |
if (!empty($url_list)) { |
| 3340 |
$placeholders = implode(',', array_fill(0, count($url_list), '%s')); |
| 3341 |
if ($search_query) { |
| 3342 |
// Include search filter in the final fetch |
| 3343 |
$prompts_query = $wpdb->prepare( |
| 3344 |
"SELECT id, article_content, source_url, timestamp, role_restriction |
| 3345 |
FROM {$table_name} |
| 3346 |
WHERE source_url IN ($placeholders) AND article_content LIKE %s |
| 3347 |
ORDER BY timestamp DESC", |
| 3348 |
...array_merge($url_list, ['%' . $wpdb->esc_like($search_query) . '%']) |
| 3349 |
); |
| 3350 |
} else { |
| 3351 |
$prompts_query = $wpdb->prepare( |
| 3352 |
"SELECT id, article_content, source_url, timestamp, role_restriction |
| 3353 |
FROM {$table_name} |
| 3354 |
WHERE source_url IN ($placeholders) |
| 3355 |
ORDER BY timestamp DESC", |
| 3356 |
$url_list |
| 3357 |
); |
| 3358 |
} |
| 3359 |
$prompts = $wpdb->get_results($prompts_query); |
| 3360 |
} |
| 3361 |
|
| 3362 |
// Group prompts by source_url for chunk display |
| 3363 |
$grouped_prompts = array(); |
| 3364 |
foreach ($prompts as $prompt) { |
| 3365 |
$source_url = $prompt->source_url ?? ''; |
| 3366 |
|
| 3367 |
// Parse chunk metadata using the proper chunker method (same as initial page load) |
| 3368 |
if (class_exists('MxChat_Chunker')) { |
| 3369 |
$chunk_meta = MxChat_Chunker::parse_stored_chunk($prompt->article_content); |
| 3370 |
$prompt->chunk_metadata = $chunk_meta['metadata']; |
| 3371 |
$prompt->display_content = $chunk_meta['text']; |
| 3372 |
} else { |
| 3373 |
$prompt->chunk_metadata = array(); |
| 3374 |
$prompt->display_content = $prompt->article_content; |
| 3375 |
} |
| 3376 |
|
| 3377 |
if (!empty($source_url) && strpos($source_url, 'mxchat://') !== 0) { |
| 3378 |
if (!isset($grouped_prompts[$source_url])) { |
| 3379 |
$grouped_prompts[$source_url] = array(); |
| 3380 |
} |
| 3381 |
$grouped_prompts[$source_url][] = $prompt; |
| 3382 |
} else { |
| 3383 |
// Ungrouped entries |
| 3384 |
$grouped_prompts['_ungrouped_' . $prompt->id] = array($prompt); |
| 3385 |
} |
| 3386 |
} |
| 3387 |
|
| 3388 |
// Sort groups by the original URL order (newest first) |
| 3389 |
uksort($grouped_prompts, function($a, $b) use ($url_order_map) { |
| 3390 |
$order_a = isset($url_order_map[$a]) ? $url_order_map[$a] : PHP_INT_MAX; |
| 3391 |
$order_b = isset($url_order_map[$b]) ? $url_order_map[$b] : PHP_INT_MAX; |
| 3392 |
return $order_a - $order_b; |
| 3393 |
}); |
| 3394 |
|
| 3395 |
// Sort each group internally by chunk_index |
| 3396 |
foreach ($grouped_prompts as $source_url => &$group) { |
| 3397 |
usort($group, function($a, $b) { |
| 3398 |
$index_a = isset($a->chunk_metadata['chunk_index']) ? intval($a->chunk_metadata['chunk_index']) : 0; |
| 3399 |
$index_b = isset($b->chunk_metadata['chunk_index']) ? intval($b->chunk_metadata['chunk_index']) : 0; |
| 3400 |
return $index_a - $index_b; |
| 3401 |
}); |
| 3402 |
} |
| 3403 |
unset($group); |
| 3404 |
|
| 3405 |
// Build HTML for the table rows |
| 3406 |
ob_start(); |
| 3407 |
$display_index = 0; |
| 3408 |
$current_page = $page; |
| 3409 |
$data_source = 'wordpress'; |
| 3410 |
$current_bot_id = $bot_id; |
| 3411 |
$preview_length = 150; |
| 3412 |
|
| 3413 |
if (empty($grouped_prompts)) { |
| 3414 |
echo '<tr><td colspan="5" style="padding: 40px; text-align: center; color: var(--mxch-text-muted);">'; |
| 3415 |
esc_html_e('No knowledge entries found. Use the Import Options to add content.', 'mxchat'); |
| 3416 |
echo '</td></tr>'; |
| 3417 |
} else { |
| 3418 |
foreach ($grouped_prompts as $source_url => $group) { |
| 3419 |
$chunk_count = count($group); |
| 3420 |
$first_prompt = $group[0]; |
| 3421 |
$display_index++; |
| 3422 |
|
| 3423 |
if ($chunk_count > 1) { |
| 3424 |
// Multiple chunks - show grouped row with expand button |
| 3425 |
$group_id = 'group-' . md5($source_url); |
| 3426 |
?> |
| 3427 |
<tr id="prompt-<?php echo esc_attr($first_prompt->id); ?>" |
| 3428 |
class="mxchat-chunk-group-header" |
| 3429 |
data-source="<?php echo esc_attr($data_source); ?>" |
| 3430 |
data-group-id="<?php echo esc_attr($group_id); ?>" |
| 3431 |
style="border-bottom: 1px solid var(--mxch-card-border);"> |
| 3432 |
<td style="padding: 12px 16px; text-align: center;"> |
| 3433 |
<input type="checkbox" |
| 3434 |
class="mxchat-entry-checkbox" |
| 3435 |
data-entry-id="<?php echo esc_attr($first_prompt->id); ?>" |
| 3436 |
data-source="<?php echo esc_attr($data_source); ?>" |
| 3437 |
data-source-url="<?php echo esc_attr($source_url); ?>" |
| 3438 |
data-is-group="true" |
| 3439 |
data-chunk-count="<?php echo esc_attr($chunk_count); ?>"> |
| 3440 |
</td> |
| 3441 |
<td style="padding: 12px 16px; font-size: 13px;"> |
| 3442 |
<?php echo esc_html($first_prompt->id); ?> |
| 3443 |
</td> |
| 3444 |
<td class="mxchat-content-cell" style="padding: 12px 16px; font-size: 13px;"> |
| 3445 |
<div class="mxchat-chunk-group-info"> |
| 3446 |
<button type="button" class="mxchat-chunk-toggle" data-group-id="<?php echo esc_attr($group_id); ?>"> |
| 3447 |
<span class="dashicons dashicons-arrow-right-alt2"></span> |
| 3448 |
</button> |
| 3449 |
<span class="mxchat-chunk-badge"><?php echo esc_html($chunk_count); ?> <?php esc_html_e('chunks', 'mxchat'); ?></span> |
| 3450 |
<span class="mxchat-chunk-preview"> |
| 3451 |
<?php |
| 3452 |
$parent_content = isset($first_prompt->display_content) ? $first_prompt->display_content : $first_prompt->article_content; |
| 3453 |
$content_preview = mb_substr($parent_content, 0, 100); |
| 3454 |
echo esc_html($content_preview . '...'); |
| 3455 |
?> |
| 3456 |
</span> |
| 3457 |
</div> |
| 3458 |
</td> |
| 3459 |
<td class="mxchat-url-cell" style="padding: 12px 16px; font-size: 13px;"> |
| 3460 |
<?php if (!empty($source_url) && strpos($source_url, 'mxchat://') !== 0 && strpos($source_url, '_ungrouped_') !== 0) : ?> |
| 3461 |
<a href="<?php echo esc_url($source_url); ?>" target="_blank" style="color: var(--mxch-primary); text-decoration: none;"> |
| 3462 |
<span class="dashicons dashicons-external" style="font-size: 14px;"></span> |
| 3463 |
<?php esc_html_e('View Source', 'mxchat'); ?> |
| 3464 |
</a> |
| 3465 |
<?php else : ?> |
| 3466 |
<span style="color: var(--mxch-text-muted);"><?php esc_html_e('Manual Content', 'mxchat'); ?></span> |
| 3467 |
<?php endif; ?> |
| 3468 |
</td> |
| 3469 |
<td class="mxchat-actions-cell" style="padding: 12px 16px; white-space: nowrap;"> |
| 3470 |
<?php if ($data_source !== 'pinecone') : ?> |
| 3471 |
<button type="button" |
| 3472 |
class="mxch-btn mxch-btn-ghost mxch-btn-sm mxchat-edit-entry-btn" |
| 3473 |
data-source-url="<?php echo esc_attr($source_url); ?>" |
| 3474 |
data-entry-id="<?php echo esc_attr($first_prompt->id); ?>" |
| 3475 |
data-data-source="<?php echo esc_attr($data_source); ?>" |
| 3476 |
data-bot-id="<?php echo esc_attr($current_bot_id); ?>" |
| 3477 |
data-nonce="<?php echo wp_create_nonce('mxchat_edit_entry_nonce'); ?>" |
| 3478 |
title="<?php esc_attr_e('Edit content', 'mxchat'); ?>"> |
| 3479 |
<span class="dashicons dashicons-edit" style="font-size: 14px;"></span> |
| 3480 |
</button> |
| 3481 |
<?php endif; ?> |
| 3482 |
<button type="button" |
| 3483 |
class="mxch-btn mxch-btn-ghost mxch-btn-sm delete-button-group" |
| 3484 |
data-source-url="<?php echo esc_attr($source_url); ?>" |
| 3485 |
data-chunk-count="<?php echo esc_attr($chunk_count); ?>" |
| 3486 |
data-data-source="<?php echo esc_attr($data_source); ?>" |
| 3487 |
data-bot-id="<?php echo esc_attr($current_bot_id); ?>" |
| 3488 |
data-nonce="<?php echo wp_create_nonce('mxchat_delete_chunks_nonce'); ?>" |
| 3489 |
style="color: var(--mxch-error);" |
| 3490 |
title="<?php esc_attr_e('Delete all chunks', 'mxchat'); ?>"> |
| 3491 |
<span class="dashicons dashicons-trash" style="font-size: 14px;"></span> |
| 3492 |
</button> |
| 3493 |
</td> |
| 3494 |
</tr> |
| 3495 |
<?php |
| 3496 |
// Render hidden chunk rows |
| 3497 |
foreach ($group as $chunk_index => $chunk) { |
| 3498 |
$meta_chunk_index = isset($chunk->chunk_metadata['chunk_index']) ? intval($chunk->chunk_metadata['chunk_index']) : $chunk_index; |
| 3499 |
$meta_total_chunks = isset($chunk->chunk_metadata['total_chunks']) ? intval($chunk->chunk_metadata['total_chunks']) : $chunk_count; |
| 3500 |
$content = isset($chunk->display_content) ? $chunk->display_content : $chunk->article_content; |
| 3501 |
$content_preview = mb_strlen($content) > $preview_length |
| 3502 |
? mb_substr($content, 0, $preview_length) . '...' |
| 3503 |
: $content; |
| 3504 |
?> |
| 3505 |
<tr id="prompt-<?php echo esc_attr($chunk->id); ?>" |
| 3506 |
class="mxchat-chunk-row <?php echo esc_attr($group_id); ?>" |
| 3507 |
data-source="<?php echo esc_attr($data_source); ?>" |
| 3508 |
style="display: none; background: #f8f9fa; border-bottom: 1px solid var(--mxch-card-border);"> |
| 3509 |
<td style="padding: 12px 16px; text-align: center;"> |
| 3510 |
<!-- Checkbox column placeholder for chunks (managed by group) --> |
| 3511 |
</td> |
| 3512 |
<td style="padding: 12px 16px 12px 30px; font-size: 13px;"> |
| 3513 |
<!-- Hidden ID column for chunks --> |
| 3514 |
</td> |
| 3515 |
<td class="mxchat-content-cell" style="padding: 12px 16px; font-size: 13px;"> |
| 3516 |
<div class="mxchat-accordion-wrapper"> |
| 3517 |
<div class="mxchat-content-preview"> |
| 3518 |
<span class="mxchat-chunk-indicator" style="margin-right: 10px; color: var(--mxch-text-secondary); font-size: 12px;"> |
| 3519 |
<?php printf(esc_html__('Chunk %d of %d', 'mxchat'), $meta_chunk_index + 1, $meta_total_chunks); ?> |
| 3520 |
</span> |
| 3521 |
<span class="preview-text"><?php echo esc_html($content_preview); ?></span> |
| 3522 |
<?php if (mb_strlen($content) > $preview_length) : ?> |
| 3523 |
<button class="mxchat-expand-toggle" type="button"> |
| 3524 |
<span class="dashicons dashicons-arrow-down-alt2"></span> |
| 3525 |
</button> |
| 3526 |
<?php endif; ?> |
| 3527 |
</div> |
| 3528 |
<div class="mxchat-content-full" style="display: none;"> |
| 3529 |
<div class="content-view"> |
| 3530 |
<?php |
| 3531 |
if (preg_match('/[\x{0590}-\x{05FF}]/u', $content)) { |
| 3532 |
echo '<div dir="rtl" lang="he" class="rtl-content">'; |
| 3533 |
echo wp_kses_post(wpautop($content)); |
| 3534 |
echo '</div>'; |
| 3535 |
} else { |
| 3536 |
echo wp_kses_post(wpautop($content)); |
| 3537 |
} |
| 3538 |
?> |
| 3539 |
</div> |
| 3540 |
</div> |
| 3541 |
</div> |
| 3542 |
</td> |
| 3543 |
<td class="mxchat-url-cell" style="padding: 12px 16px; font-size: 13px;"> |
| 3544 |
<span class="mxchat-chunk-label" style="color: var(--mxch-text-muted);"><?php esc_html_e('Same as parent', 'mxchat'); ?></span> |
| 3545 |
</td> |
| 3546 |
<td class="mxchat-actions-cell" style="padding: 12px 16px;"> |
| 3547 |
<span class="mxchat-chunk-label" style="color: var(--mxch-text-muted);"><?php esc_html_e('Managed by group', 'mxchat'); ?></span> |
| 3548 |
</td> |
| 3549 |
</tr> |
| 3550 |
<?php |
| 3551 |
} |
| 3552 |
} else { |
| 3553 |
// Single entry - display normally with accordion |
| 3554 |
$prompt = $first_prompt; |
| 3555 |
$content = isset($prompt->display_content) ? $prompt->display_content : $prompt->article_content; |
| 3556 |
$content_preview = mb_strlen($content) > $preview_length |
| 3557 |
? mb_substr($content, 0, $preview_length) . '...' |
| 3558 |
: $content; |
| 3559 |
?> |
| 3560 |
<tr id="prompt-<?php echo esc_attr($prompt->id); ?>" |
| 3561 |
data-source="<?php echo esc_attr($data_source); ?>" |
| 3562 |
style="border-bottom: 1px solid var(--mxch-card-border);"> |
| 3563 |
<td style="padding: 12px 16px; text-align: center;"> |
| 3564 |
<input type="checkbox" |
| 3565 |
class="mxchat-entry-checkbox" |
| 3566 |
data-entry-id="<?php echo esc_attr($prompt->id); ?>" |
| 3567 |
data-source="<?php echo esc_attr($data_source); ?>" |
| 3568 |
data-source-url="<?php echo esc_attr($source_url); ?>" |
| 3569 |
data-is-group="false"> |
| 3570 |
</td> |
| 3571 |
<td style="padding: 12px 16px; font-size: 13px;"> |
| 3572 |
<?php echo esc_html($prompt->id); ?> |
| 3573 |
</td> |
| 3574 |
<td class="mxchat-content-cell" style="padding: 12px 16px; font-size: 13px;"> |
| 3575 |
<div class="mxchat-accordion-wrapper"> |
| 3576 |
<div class="mxchat-content-preview"> |
| 3577 |
<span class="preview-text"><?php echo esc_html($content_preview); ?></span> |
| 3578 |
<?php if (mb_strlen($content) > $preview_length) : ?> |
| 3579 |
<button class="mxchat-expand-toggle" type="button"> |
| 3580 |
<span class="dashicons dashicons-arrow-down-alt2"></span> |
| 3581 |
</button> |
| 3582 |
<?php endif; ?> |
| 3583 |
</div> |
| 3584 |
<div class="mxchat-content-full" style="display: none;"> |
| 3585 |
<div class="content-view"> |
| 3586 |
<?php |
| 3587 |
if (preg_match('/[\x{0590}-\x{05FF}]/u', $content)) { |
| 3588 |
echo '<div dir="rtl" lang="he" class="rtl-content">'; |
| 3589 |
echo wp_kses_post(wpautop($content)); |
| 3590 |
echo '</div>'; |
| 3591 |
} else { |
| 3592 |
echo wp_kses_post(wpautop($content)); |
| 3593 |
} |
| 3594 |
?> |
| 3595 |
</div> |
| 3596 |
</div> |
| 3597 |
</div> |
| 3598 |
</td> |
| 3599 |
<td class="mxchat-url-cell" style="padding: 12px 16px; font-size: 13px;"> |
| 3600 |
<?php |
| 3601 |
$actual_source = $source_url; |
| 3602 |
if (strpos($source_url, '_ungrouped_') === 0) { |
| 3603 |
$actual_source = $prompt->source_url ?? ''; |
| 3604 |
} |
| 3605 |
if (!empty($actual_source) && strpos($actual_source, 'mxchat://') !== 0) : ?> |
| 3606 |
<a href="<?php echo esc_url($actual_source); ?>" target="_blank" style="color: var(--mxch-primary); text-decoration: none;"> |
| 3607 |
<span class="dashicons dashicons-external" style="font-size: 14px;"></span> |
| 3608 |
<?php esc_html_e('View', 'mxchat'); ?> |
| 3609 |
</a> |
| 3610 |
<?php else : ?> |
| 3611 |
<span style="color: var(--mxch-text-muted);"><?php esc_html_e('Manual', 'mxchat'); ?></span> |
| 3612 |
<?php endif; ?> |
| 3613 |
</td> |
| 3614 |
<td style="padding: 12px 16px; white-space: nowrap;"> |
| 3615 |
<button type="button" |
| 3616 |
class="mxch-btn mxch-btn-ghost mxch-btn-sm mxchat-edit-entry-btn" |
| 3617 |
data-source-url="<?php echo esc_attr($prompt->source_url ?? ''); ?>" |
| 3618 |
data-entry-id="<?php echo esc_attr($prompt->id); ?>" |
| 3619 |
data-data-source="<?php echo esc_attr($data_source); ?>" |
| 3620 |
data-bot-id="<?php echo esc_attr($current_bot_id); ?>" |
| 3621 |
data-nonce="<?php echo wp_create_nonce('mxchat_edit_entry_nonce'); ?>" |
| 3622 |
title="<?php esc_attr_e('Edit content', 'mxchat'); ?>"> |
| 3623 |
<span class="dashicons dashicons-edit" style="font-size: 14px;"></span> |
| 3624 |
</button> |
| 3625 |
<button type="button" class="mxch-btn mxch-btn-ghost mxch-btn-sm delete-button-wordpress" data-entry-id="<?php echo esc_attr($prompt->id); ?>" data-bot-id="<?php echo esc_attr($current_bot_id); ?>" data-nonce="<?php echo wp_create_nonce('mxchat_delete_wordpress_prompt_nonce'); ?>" style="color: var(--mxch-error);"> |
| 3626 |
<span class="dashicons dashicons-trash" style="font-size: 14px;"></span> |
| 3627 |
</button> |
| 3628 |
</td> |
| 3629 |
</tr> |
| 3630 |
<?php |
| 3631 |
} |
| 3632 |
} |
| 3633 |
} |
| 3634 |
$html = ob_get_clean(); |
| 3635 |
|
| 3636 |
// Generate pagination HTML (include search/filter data for subsequent pages) |
| 3637 |
$pagination_html = ''; |
| 3638 |
if ($total_pages > 1) { |
| 3639 |
$pagination_html = '<div class="mxchat-ajax-pagination" data-current-page="' . esc_attr($page) . '" data-total-pages="' . esc_attr($total_pages) . '" data-search="' . esc_attr($search_query) . '" data-content-type="' . esc_attr($content_type_filter) . '">'; |
| 3640 |
|
| 3641 |
// Previous button |
| 3642 |
if ($page > 1) { |
| 3643 |
$pagination_html .= '<a href="#" class="mxchat-page-link" data-page="' . ($page - 1) . '">' . esc_html__('« Previous', 'mxchat') . '</a> '; |
| 3644 |
} |
| 3645 |
|
| 3646 |
// Page numbers |
| 3647 |
$start_page = max(1, $page - 2); |
| 3648 |
$end_page = min($total_pages, $page + 2); |
| 3649 |
|
| 3650 |
if ($start_page > 1) { |
| 3651 |
$pagination_html .= '<a href="#" class="mxchat-page-link" data-page="1">1</a> '; |
| 3652 |
if ($start_page > 2) { |
| 3653 |
$pagination_html .= '<span class="mxchat-page-dots">...</span> '; |
| 3654 |
} |
| 3655 |
} |
| 3656 |
|
| 3657 |
for ($i = $start_page; $i <= $end_page; $i++) { |
| 3658 |
if ($i == $page) { |
| 3659 |
$pagination_html .= '<span class="mxchat-page-current">' . $i . '</span> '; |
| 3660 |
} else { |
| 3661 |
$pagination_html .= '<a href="#" class="mxchat-page-link" data-page="' . $i . '">' . $i . '</a> '; |
| 3662 |
} |
| 3663 |
} |
| 3664 |
|
| 3665 |
if ($end_page < $total_pages) { |
| 3666 |
if ($end_page < $total_pages - 1) { |
| 3667 |
$pagination_html .= '<span class="mxchat-page-dots">...</span> '; |
| 3668 |
} |
| 3669 |
$pagination_html .= '<a href="#" class="mxchat-page-link" data-page="' . $total_pages . '">' . $total_pages . '</a> '; |
| 3670 |
} |
| 3671 |
|
| 3672 |
// Next button |
| 3673 |
if ($page < $total_pages) { |
| 3674 |
$pagination_html .= '<a href="#" class="mxchat-page-link" data-page="' . ($page + 1) . '">' . esc_html__('Next »', 'mxchat') . '</a>'; |
| 3675 |
} |
| 3676 |
|
| 3677 |
$pagination_html .= '</div>'; |
| 3678 |
} |
| 3679 |
|
| 3680 |
wp_send_json_success(array( |
| 3681 |
'html' => $html, |
| 3682 |
'pagination_html' => $pagination_html, |
| 3683 |
'total_count' => $total_records, |
| 3684 |
'total_pages' => $total_pages, |
| 3685 |
'page' => $page, |
| 3686 |
'per_page' => $per_page, |
| 3687 |
'data_source' => 'wordpress' |
| 3688 |
)); |
| 3689 |
} |
| 3690 |
|
| 3691 |
/** |
| 3692 |
* AJAX handler to detect available sitemaps on the site |
| 3693 |
* Optimized for speed - only checks primary sitemap indexes first |
| 3694 |
*/ |
| 3695 |
public function ajax_mxchat_detect_sitemaps() { |
| 3696 |
check_ajax_referer('mxchat_detect_sitemaps_nonce', 'nonce'); |
| 3697 |
|
| 3698 |
if (!current_user_can('manage_options')) { |
| 3699 |
wp_send_json_error(array('message' => 'Unauthorized')); |
| 3700 |
return; |
| 3701 |
} |
| 3702 |
|
| 3703 |
$site_url = get_site_url(); |
| 3704 |
$sitemaps = array(); |
| 3705 |
$found_index = false; |
| 3706 |
|
| 3707 |
// Only check the main sitemap index files first (much faster) |
| 3708 |
// These are the primary entry points that contain sub-sitemaps |
| 3709 |
$primary_indexes = array( |
| 3710 |
'sitemap_index.xml' => 'Yoast SEO / Rank Math', // Yoast & Rank Math |
| 3711 |
'wp-sitemap.xml' => 'WordPress Core', // WordPress Core |
| 3712 |
'sitemap.xml' => 'Standard', // Generic/AIOSEO |
| 3713 |
); |
| 3714 |
|
| 3715 |
foreach ($primary_indexes as $path => $source) { |
| 3716 |
$url = trailingslashit($site_url) . $path; |
| 3717 |
|
| 3718 |
$response = wp_remote_head($url, array( |
| 3719 |
'timeout' => 10, |
| 3720 |
'sslverify' => false, |
| 3721 |
'redirection' => 1, |
| 3722 |
'user-agent' => mxchat_ingest_user_agent(), |
| 3723 |
)); |
| 3724 |
|
| 3725 |
if (!is_wp_error($response) && wp_remote_retrieve_response_code($response) === 200) { |
| 3726 |
// Found a sitemap index - parse it to get sub-sitemaps |
| 3727 |
$sub_sitemaps = $this->parse_sitemap_index($url); |
| 3728 |
if (!empty($sub_sitemaps)) { |
| 3729 |
$sitemaps[] = array( |
| 3730 |
'url' => $url, |
| 3731 |
'type' => 'index', |
| 3732 |
'source' => $source, |
| 3733 |
'sub_sitemaps' => $sub_sitemaps |
| 3734 |
); |
| 3735 |
$found_index = true; |
| 3736 |
// Found a valid index, no need to check others |
| 3737 |
break; |
| 3738 |
} |
| 3739 |
} |
| 3740 |
} |
| 3741 |
|
| 3742 |
// If no sitemap index found, check for standalone sitemaps |
| 3743 |
if (!$found_index) { |
| 3744 |
$standalone_sitemaps = array( |
| 3745 |
'post-sitemap.xml' => array('type' => 'content', 'source' => 'Yoast SEO'), |
| 3746 |
'page-sitemap.xml' => array('type' => 'content', 'source' => 'Yoast SEO'), |
| 3747 |
); |
| 3748 |
|
| 3749 |
foreach ($standalone_sitemaps as $path => $info) { |
| 3750 |
$url = trailingslashit($site_url) . $path; |
| 3751 |
|
| 3752 |
$response = wp_remote_head($url, array( |
| 3753 |
'timeout' => 2, |
| 3754 |
'sslverify' => false |
| 3755 |
)); |
| 3756 |
|
| 3757 |
if (!is_wp_error($response) && wp_remote_retrieve_response_code($response) === 200) { |
| 3758 |
$sitemaps[] = array( |
| 3759 |
'url' => $url, |
| 3760 |
'type' => $info['type'], |
| 3761 |
'source' => $info['source'], |
| 3762 |
'url_count' => 0 // Skip URL count for speed |
| 3763 |
); |
| 3764 |
} |
| 3765 |
} |
| 3766 |
} |
| 3767 |
|
| 3768 |
wp_send_json_success(array( |
| 3769 |
'sitemaps' => $sitemaps, |
| 3770 |
'site_url' => $site_url |
| 3771 |
)); |
| 3772 |
} |
| 3773 |
|
| 3774 |
/** |
| 3775 |
* Parse a sitemap index to get sub-sitemaps |
| 3776 |
* Optimized: doesn't fetch URL count for each sub-sitemap (too slow) |
| 3777 |
*/ |
| 3778 |
private function parse_sitemap_index($url) { |
| 3779 |
$sub_sitemaps = array(); |
| 3780 |
|
| 3781 |
$response = wp_remote_get($url, array( |
| 3782 |
'timeout' => 30, |
| 3783 |
'sslverify' => false, |
| 3784 |
'user-agent' => mxchat_ingest_user_agent(), |
| 3785 |
'headers' => array( |
| 3786 |
'Accept' => 'text/html,application/xhtml+xml,application/xml;q=0.9,*/*;q=0.8', |
| 3787 |
), |
| 3788 |
)); |
| 3789 |
|
| 3790 |
if (is_wp_error($response)) { |
| 3791 |
return $sub_sitemaps; |
| 3792 |
} |
| 3793 |
|
| 3794 |
$body = wp_remote_retrieve_body($response); |
| 3795 |
if (empty($body)) { |
| 3796 |
return $sub_sitemaps; |
| 3797 |
} |
| 3798 |
|
| 3799 |
// Suppress XML errors |
| 3800 |
libxml_use_internal_errors(true); |
| 3801 |
$xml = simplexml_load_string($body); |
| 3802 |
libxml_clear_errors(); |
| 3803 |
|
| 3804 |
if ($xml === false) { |
| 3805 |
return $sub_sitemaps; |
| 3806 |
} |
| 3807 |
|
| 3808 |
// Check if it's a sitemap index (contains <sitemap> elements) |
| 3809 |
if (isset($xml->sitemap)) { |
| 3810 |
foreach ($xml->sitemap as $sitemap) { |
| 3811 |
$loc = (string) $sitemap->loc; |
| 3812 |
if (!empty($loc)) { |
| 3813 |
// Try to determine the type from the URL |
| 3814 |
$type = 'content'; |
| 3815 |
if (strpos($loc, 'category') !== false || strpos($loc, 'tag') !== false || strpos($loc, 'taxonomy') !== false) { |
| 3816 |
$type = 'taxonomy'; |
| 3817 |
} elseif (strpos($loc, 'author') !== false || strpos($loc, 'user') !== false) { |
| 3818 |
$type = 'author'; |
| 3819 |
} |
| 3820 |
|
| 3821 |
// Skip URL count - too slow to fetch for each sitemap |
| 3822 |
$sub_sitemaps[] = array( |
| 3823 |
'url' => $loc, |
| 3824 |
'type' => $type, |
| 3825 |
'url_count' => 0, // Don't fetch - takes too long |
| 3826 |
'name' => basename(parse_url($loc, PHP_URL_PATH)) |
| 3827 |
); |
| 3828 |
} |
| 3829 |
} |
| 3830 |
} |
| 3831 |
|
| 3832 |
return $sub_sitemaps; |
| 3833 |
} |
| 3834 |
|
| 3835 |
/** |
| 3836 |
* Get URL count from a sitemap |
| 3837 |
*/ |
| 3838 |
private function get_sitemap_url_count($url) { |
| 3839 |
$response = wp_remote_get($url, array( |
| 3840 |
'timeout' => 30, |
| 3841 |
'sslverify' => false, |
| 3842 |
'user-agent' => mxchat_ingest_user_agent(), |
| 3843 |
'headers' => array( |
| 3844 |
'Accept' => 'text/html,application/xhtml+xml,application/xml;q=0.9,*/*;q=0.8', |
| 3845 |
), |
| 3846 |
)); |
| 3847 |
|
| 3848 |
if (is_wp_error($response)) { |
| 3849 |
return 0; |
| 3850 |
} |
| 3851 |
|
| 3852 |
$body = wp_remote_retrieve_body($response); |
| 3853 |
if (empty($body)) { |
| 3854 |
return 0; |
| 3855 |
} |
| 3856 |
|
| 3857 |
// Count <url> or <loc> elements |
| 3858 |
$count = preg_match_all('/<url>/i', $body, $matches); |
| 3859 |
return $count ?: 0; |
| 3860 |
} |
| 3861 |
|
| 3862 |
/** |
| 3863 |
* Get sitemaps declared in robots.txt |
| 3864 |
*/ |
| 3865 |
private function get_sitemaps_from_robots($site_url) { |
| 3866 |
$sitemaps = array(); |
| 3867 |
$robots_url = trailingslashit($site_url) . 'robots.txt'; |
| 3868 |
|
| 3869 |
$response = wp_remote_get($robots_url, array( |
| 3870 |
'timeout' => 15, |
| 3871 |
'sslverify' => false, |
| 3872 |
'user-agent' => mxchat_ingest_user_agent(), |
| 3873 |
)); |
| 3874 |
|
| 3875 |
if (is_wp_error($response)) { |
| 3876 |
return $sitemaps; |
| 3877 |
} |
| 3878 |
|
| 3879 |
$body = wp_remote_retrieve_body($response); |
| 3880 |
if (empty($body)) { |
| 3881 |
return $sitemaps; |
| 3882 |
} |
| 3883 |
|
| 3884 |
// Find Sitemap: declarations |
| 3885 |
if (preg_match_all('/^Sitemap:\s*(.+)$/mi', $body, $matches)) { |
| 3886 |
foreach ($matches[1] as $sitemap_url) { |
| 3887 |
$sitemap_url = trim($sitemap_url); |
| 3888 |
if (filter_var($sitemap_url, FILTER_VALIDATE_URL)) { |
| 3889 |
$sitemaps[] = $sitemap_url; |
| 3890 |
} |
| 3891 |
} |
| 3892 |
} |
| 3893 |
|
| 3894 |
return $sitemaps; |
| 3895 |
} |
| 3896 |
|
| 3897 |
public function mxchat_stop_processing() { |
| 3898 |
// Verify permissions |
| 3899 |
if (!current_user_can('manage_options')) { |
| 3900 |
wp_die(esc_html__('Unauthorized access', 'mxchat')); |
| 3901 |
} |
| 3902 |
|
| 3903 |
// Verify nonce |
| 3904 |
check_admin_referer('mxchat_stop_processing_action', 'mxchat_stop_processing_nonce'); |
| 3905 |
|
| 3906 |
global $wpdb; |
| 3907 |
$table_name = $wpdb->prefix . 'mxchat_processing_queue'; |
| 3908 |
|
| 3909 |
// Get active queue IDs |
| 3910 |
$sitemap_queue_id = get_transient('mxchat_active_queue_sitemap'); |
| 3911 |
$pdf_queue_id = get_transient('mxchat_active_queue_pdf'); |
| 3912 |
|
| 3913 |
// Delete all pending items from active queues |
| 3914 |
if ($sitemap_queue_id) { |
| 3915 |
$wpdb->delete( |
| 3916 |
$table_name, |
| 3917 |
array( |
| 3918 |
'queue_id' => $sitemap_queue_id, |
| 3919 |
'status' => 'pending' |
| 3920 |
), |
| 3921 |
array('%s', '%s') |
| 3922 |
); |
| 3923 |
|
| 3924 |
delete_transient('mxchat_active_queue_sitemap'); |
| 3925 |
delete_transient('mxchat_last_sitemap_url'); |
| 3926 |
} |
| 3927 |
|
| 3928 |
if ($pdf_queue_id) { |
| 3929 |
// Get PDF path before deleting |
| 3930 |
$pdf_path = $this->mxchat_get_queue_meta($pdf_queue_id, 'pdf_path'); |
| 3931 |
|
| 3932 |
$wpdb->delete( |
| 3933 |
$table_name, |
| 3934 |
array( |
| 3935 |
'queue_id' => $pdf_queue_id, |
| 3936 |
'status' => 'pending' |
| 3937 |
), |
| 3938 |
array('%s', '%s') |
| 3939 |
); |
| 3940 |
|
| 3941 |
// Delete PDF file |
| 3942 |
if ($pdf_path && file_exists($pdf_path)) { |
| 3943 |
wp_delete_file($pdf_path); |
| 3944 |
} |
| 3945 |
|
| 3946 |
delete_transient('mxchat_active_queue_pdf'); |
| 3947 |
delete_transient('mxchat_last_pdf_url'); |
| 3948 |
} |
| 3949 |
|
| 3950 |
// Redirect back with a success message |
| 3951 |
set_transient('mxchat_admin_notice_success', |
| 3952 |
esc_html__('Processing has been stopped successfully.', 'mxchat'), |
| 3953 |
30 |
| 3954 |
); |
| 3955 |
wp_safe_redirect(admin_url('admin.php?page=mxchat-prompts')); |
| 3956 |
exit; |
| 3957 |
} |
| 3958 |
|
| 3959 |
/** |
| 3960 |
* Get content list for processing |
| 3961 |
*/ |
| 3962 |
public function ajax_mxchat_get_content_list() { |
| 3963 |
// Verify the nonce |
| 3964 |
check_ajax_referer('mxchat_content_selector_nonce', 'nonce'); |
| 3965 |
|
| 3966 |
if (!current_user_can('manage_options')) { |
| 3967 |
wp_send_json_error(__('Unauthorized access', 'mxchat')); |
| 3968 |
} |
| 3969 |
|
| 3970 |
$page = isset($_GET['page']) ? absint($_GET['page']) : 1; |
| 3971 |
$per_page = isset($_GET['per_page']) ? absint($_GET['per_page']) : 100; |
| 3972 |
$search = isset($_GET['search']) ? sanitize_text_field($_GET['search']) : ''; |
| 3973 |
$post_type = isset($_GET['post_type']) ? sanitize_text_field($_GET['post_type']) : 'all'; |
| 3974 |
$post_status = isset($_GET['post_status']) ? sanitize_text_field($_GET['post_status']) : 'publish'; |
| 3975 |
$processed_filter = isset($_GET['processed_filter']) ? sanitize_text_field($_GET['processed_filter']) : 'all'; |
| 3976 |
|
| 3977 |
// Build query args |
| 3978 |
$args = array( |
| 3979 |
'posts_per_page' => $per_page, |
| 3980 |
'paged' => $page, |
| 3981 |
'post_status' => $post_status !== 'all' ? $post_status : array('publish', 'draft', 'pending'), |
| 3982 |
'orderby' => 'date', |
| 3983 |
'order' => 'DESC', |
| 3984 |
); |
| 3985 |
|
| 3986 |
// Handle post types - IMPROVED VERSION |
| 3987 |
if ($post_type !== 'all') { |
| 3988 |
$args['post_type'] = $post_type; |
| 3989 |
} else { |
| 3990 |
// Get all available post types that might contain content |
| 3991 |
$all_post_types = array(); |
| 3992 |
|
| 3993 |
// First get all public post types |
| 3994 |
$public_types = get_post_types(array('public' => true), 'names'); |
| 3995 |
$all_post_types = array_merge($all_post_types, $public_types); |
| 3996 |
|
| 3997 |
// Add common forum/community post types |
| 3998 |
$forum_types = array('topic', 'reply', 'forum', 'wpforo_topic', 'wpforo_post'); |
| 3999 |
foreach ($forum_types as $forum_type) { |
| 4000 |
if (post_type_exists($forum_type)) { |
| 4001 |
$all_post_types[] = $forum_type; |
| 4002 |
} |
| 4003 |
} |
| 4004 |
|
| 4005 |
// Add other commonly used post types |
| 4006 |
$common_types = array('product', 'job_listing', 'event', 'portfolio'); |
| 4007 |
foreach ($common_types as $common_type) { |
| 4008 |
if (post_type_exists($common_type)) { |
| 4009 |
$all_post_types[] = $common_type; |
| 4010 |
} |
| 4011 |
} |
| 4012 |
|
| 4013 |
// Remove duplicates and ensure we have at least some post types |
| 4014 |
$all_post_types = array_unique($all_post_types); |
| 4015 |
|
| 4016 |
if (empty($all_post_types)) { |
| 4017 |
// Fallback to basic post types |
| 4018 |
$all_post_types = array('post', 'page'); |
| 4019 |
} |
| 4020 |
|
| 4021 |
$args['post_type'] = $all_post_types; |
| 4022 |
|
| 4023 |
// Debug logging to see what post types are being queried |
| 4024 |
//error_log('MxChat Debug: Querying post types: ' . implode(', ', $all_post_types)); |
| 4025 |
} |
| 4026 |
|
| 4027 |
if (!empty($search)) { |
| 4028 |
$args['s'] = $search; |
| 4029 |
} |
| 4030 |
|
| 4031 |
// Get processed data from storage |
| 4032 |
$processed_data = array(); |
| 4033 |
|
| 4034 |
$pinecone_options = get_option('mxchat_pinecone_addon_options', array()); |
| 4035 |
$use_pinecone = ($pinecone_options['mxchat_use_pinecone'] ?? '0') === '1'; |
| 4036 |
|
| 4037 |
if ($use_pinecone && !empty($pinecone_options['mxchat_pinecone_api_key'])) { |
| 4038 |
// Get fresh data from Pinecone - no caching |
| 4039 |
$processed_data = $this->mxchat_get_pinecone_processed_content($pinecone_options); |
| 4040 |
} else { |
| 4041 |
// WordPress DB checking with better URL matching for all post types |
| 4042 |
global $wpdb; |
| 4043 |
$table_name = $wpdb->prefix . 'mxchat_system_prompt_content'; |
| 4044 |
$processed_items = $wpdb->get_results("SELECT id, source_url, article_content, timestamp FROM {$table_name}"); |
| 4045 |
|
| 4046 |
// Group items by source_url to count chunks |
| 4047 |
$url_chunk_counts = array(); |
| 4048 |
$url_latest_timestamp = array(); |
| 4049 |
$url_first_id = array(); |
| 4050 |
|
| 4051 |
if (!empty($processed_items)) { |
| 4052 |
foreach ($processed_items as $item) { |
| 4053 |
$url = $item->source_url; |
| 4054 |
if (empty($url)) continue; |
| 4055 |
|
| 4056 |
// Count chunks per URL |
| 4057 |
if (!isset($url_chunk_counts[$url])) { |
| 4058 |
$url_chunk_counts[$url] = 0; |
| 4059 |
$url_latest_timestamp[$url] = $item->timestamp; |
| 4060 |
$url_first_id[$url] = $item->id; |
| 4061 |
} |
| 4062 |
$url_chunk_counts[$url]++; |
| 4063 |
|
| 4064 |
// Track latest timestamp |
| 4065 |
if (strtotime($item->timestamp) > strtotime($url_latest_timestamp[$url])) { |
| 4066 |
$url_latest_timestamp[$url] = $item->timestamp; |
| 4067 |
} |
| 4068 |
} |
| 4069 |
|
| 4070 |
// Now build processed_data with chunk counts |
| 4071 |
foreach ($url_chunk_counts as $url => $chunk_count) { |
| 4072 |
$post_id = $this->mxchat_url_to_post_id_improved($url); |
| 4073 |
|
| 4074 |
if ($post_id) { |
| 4075 |
$processed_data[$post_id] = array( |
| 4076 |
'db_id' => $url_first_id[$url], |
| 4077 |
'timestamp' => $url_latest_timestamp[$url], |
| 4078 |
'url' => $url, |
| 4079 |
'source' => 'wordpress', |
| 4080 |
'chunk_count' => $chunk_count |
| 4081 |
); |
| 4082 |
} |
| 4083 |
} |
| 4084 |
} |
| 4085 |
} |
| 4086 |
|
| 4087 |
// Get processed IDs as a simple array for in_array checks |
| 4088 |
$processed_ids = array_keys($processed_data); |
| 4089 |
|
| 4090 |
// Handle processed/unprocessed filter |
| 4091 |
if ($processed_filter === 'processed' && !empty($processed_ids)) { |
| 4092 |
$args['post__in'] = $processed_ids; |
| 4093 |
} elseif ($processed_filter === 'unprocessed' && !empty($processed_ids)) { |
| 4094 |
$args['post__not_in'] = $processed_ids; |
| 4095 |
} |
| 4096 |
|
| 4097 |
// Run the query |
| 4098 |
$query = new WP_Query($args); |
| 4099 |
$content_items = array(); |
| 4100 |
|
| 4101 |
if ($query->have_posts()) { |
| 4102 |
while ($query->have_posts()) { |
| 4103 |
$query->the_post(); |
| 4104 |
$id = get_the_ID(); |
| 4105 |
$post_date = get_the_date(); |
| 4106 |
$excerpt = wp_trim_words(get_the_excerpt(), 20, '...'); |
| 4107 |
$word_count = str_word_count(strip_tags(get_the_content())); |
| 4108 |
|
| 4109 |
$is_processed = in_array($id, $processed_ids); |
| 4110 |
$processed_date = ''; |
| 4111 |
$db_record_id = 0; |
| 4112 |
$data_source = 'none'; |
| 4113 |
|
| 4114 |
if ($is_processed && isset($processed_data[$id])) { |
| 4115 |
$item_data = $processed_data[$id]; |
| 4116 |
$data_source = $item_data['source']; |
| 4117 |
|
| 4118 |
if ($data_source === 'wordpress' && isset($item_data['timestamp'])) { |
| 4119 |
// WordPress DB format |
| 4120 |
$timestamp = strtotime($item_data['timestamp']); |
| 4121 |
$processed_date = human_time_diff($timestamp, current_time('timestamp')) . ' ago'; |
| 4122 |
$db_record_id = $item_data['db_id']; |
| 4123 |
} elseif ($data_source === 'pinecone') { |
| 4124 |
// Pinecone format |
| 4125 |
$processed_date = $item_data['processed_date']; |
| 4126 |
$db_record_id = $item_data['db_id']; |
| 4127 |
} |
| 4128 |
} |
| 4129 |
|
| 4130 |
// Get chunk count for this item |
| 4131 |
$chunk_count = 0; |
| 4132 |
if ($is_processed && isset($processed_data[$id]['chunk_count'])) { |
| 4133 |
$chunk_count = intval($processed_data[$id]['chunk_count']); |
| 4134 |
} |
| 4135 |
|
| 4136 |
$content_items[] = array( |
| 4137 |
'id' => $id, |
| 4138 |
'title' => get_the_title(), |
| 4139 |
'permalink' => get_permalink(), |
| 4140 |
'date' => $post_date, |
| 4141 |
'type' => get_post_type(), |
| 4142 |
'status' => get_post_status(), |
| 4143 |
'excerpt' => $excerpt, |
| 4144 |
'word_count' => $word_count, |
| 4145 |
'already_processed' => $is_processed, |
| 4146 |
'processed_date' => $processed_date, |
| 4147 |
'db_record_id' => $db_record_id, |
| 4148 |
'data_source' => $data_source, |
| 4149 |
'chunk_count' => $chunk_count |
| 4150 |
); |
| 4151 |
} |
| 4152 |
wp_reset_postdata(); |
| 4153 |
} |
| 4154 |
|
| 4155 |
$response = array( |
| 4156 |
'items' => $content_items, |
| 4157 |
'total' => $query->found_posts, |
| 4158 |
'total_pages' => $query->max_num_pages, |
| 4159 |
'current_page' => $page, |
| 4160 |
'processed_count' => count($processed_ids) |
| 4161 |
); |
| 4162 |
|
| 4163 |
wp_send_json_success($response); |
| 4164 |
exit; |
| 4165 |
} |
| 4166 |
|
| 4167 |
|
| 4168 |
/** |
| 4169 |
* This function handles various WooCommerce URL formats and permalink structures |
| 4170 |
*/ |
| 4171 |
private function mxchat_url_to_post_id_improved($url) { |
| 4172 |
// First try the standard WordPress function |
| 4173 |
$post_id = url_to_postid($url); |
| 4174 |
|
| 4175 |
if ($post_id > 0) { |
| 4176 |
return $post_id; |
| 4177 |
} |
| 4178 |
|
| 4179 |
// If that fails, try more aggressive URL matching |
| 4180 |
// Remove trailing slashes and query parameters for better matching |
| 4181 |
$clean_url = rtrim($url, '/'); |
| 4182 |
$clean_url = strtok($clean_url, '?'); // Remove query parameters |
| 4183 |
|
| 4184 |
// Try again with cleaned URL |
| 4185 |
$post_id = url_to_postid($clean_url); |
| 4186 |
if ($post_id > 0) { |
| 4187 |
return $post_id; |
| 4188 |
} |
| 4189 |
|
| 4190 |
// For bbPress forum topics, try extracting slug from URL |
| 4191 |
if (strpos($url, '/topic/') !== false || strpos($url, '/forums/') !== false) { |
| 4192 |
// Handle bbPress URLs: /forums/topic/topic-name/ |
| 4193 |
if (preg_match('/\/forums\/topic\/([^\/\?]+)/', $url, $matches)) { |
| 4194 |
$topic_slug = $matches[1]; |
| 4195 |
|
| 4196 |
// Look up topic by slug |
| 4197 |
$topic = get_page_by_path($topic_slug, OBJECT, 'topic'); |
| 4198 |
if ($topic) { |
| 4199 |
return $topic->ID; |
| 4200 |
} |
| 4201 |
|
| 4202 |
// Alternative method: query by post_name |
| 4203 |
global $wpdb; |
| 4204 |
$post_id = $wpdb->get_var($wpdb->prepare( |
| 4205 |
"SELECT ID FROM {$wpdb->posts} WHERE post_name = %s AND post_type = 'topic' AND post_status IN ('publish', 'closed')", |
| 4206 |
$topic_slug |
| 4207 |
)); |
| 4208 |
|
| 4209 |
if ($post_id) { |
| 4210 |
return intval($post_id); |
| 4211 |
} |
| 4212 |
} |
| 4213 |
|
| 4214 |
// Handle simpler topic URLs: /topic/topic-name/ |
| 4215 |
if (preg_match('/\/topic\/([^\/\?]+)/', $url, $matches)) { |
| 4216 |
$topic_slug = $matches[1]; |
| 4217 |
|
| 4218 |
global $wpdb; |
| 4219 |
$post_id = $wpdb->get_var($wpdb->prepare( |
| 4220 |
"SELECT ID FROM {$wpdb->posts} WHERE post_name = %s AND post_type = 'topic' AND post_status IN ('publish', 'closed')", |
| 4221 |
$topic_slug |
| 4222 |
)); |
| 4223 |
|
| 4224 |
if ($post_id) { |
| 4225 |
return intval($post_id); |
| 4226 |
} |
| 4227 |
} |
| 4228 |
} |
| 4229 |
|
| 4230 |
// For WooCommerce products |
| 4231 |
if (strpos($url, '/product/') !== false || strpos($url, 'product=') !== false) { |
| 4232 |
// Extract product slug from various URL formats |
| 4233 |
$product_slug = ''; |
| 4234 |
|
| 4235 |
// Handle pretty permalinks: /product/product-name/ |
| 4236 |
if (preg_match('/\/product\/([^\/\?]+)/', $url, $matches)) { |
| 4237 |
$product_slug = $matches[1]; |
| 4238 |
} |
| 4239 |
// Handle query parameters: ?product=product-name |
| 4240 |
elseif (preg_match('/[\?&]product=([^&]+)/', $url, $matches)) { |
| 4241 |
$product_slug = $matches[1]; |
| 4242 |
} |
| 4243 |
|
| 4244 |
if (!empty($product_slug)) { |
| 4245 |
// Look up product by slug |
| 4246 |
$product = get_page_by_path($product_slug, OBJECT, 'product'); |
| 4247 |
if ($product) { |
| 4248 |
return $product->ID; |
| 4249 |
} |
| 4250 |
|
| 4251 |
// Alternative method: query by post_name |
| 4252 |
global $wpdb; |
| 4253 |
$post_id = $wpdb->get_var($wpdb->prepare( |
| 4254 |
"SELECT ID FROM {$wpdb->posts} WHERE post_name = %s AND post_type = 'product' AND post_status = 'publish'", |
| 4255 |
$product_slug |
| 4256 |
)); |
| 4257 |
|
| 4258 |
if ($post_id) { |
| 4259 |
return intval($post_id); |
| 4260 |
} |
| 4261 |
} |
| 4262 |
} |
| 4263 |
|
| 4264 |
// Generic approach: try to extract slug and match against all post types |
| 4265 |
$parsed_url = wp_parse_url($clean_url); |
| 4266 |
$path = $parsed_url['path'] ?? ''; |
| 4267 |
|
| 4268 |
if (!empty($path)) { |
| 4269 |
// Get the last part of the path as potential slug |
| 4270 |
$path_parts = array_filter(explode('/', trim($path, '/'))); |
| 4271 |
$potential_slug = end($path_parts); |
| 4272 |
|
| 4273 |
if (!empty($potential_slug)) { |
| 4274 |
global $wpdb; |
| 4275 |
|
| 4276 |
// Try to find any post with this slug |
| 4277 |
$post_id = $wpdb->get_var($wpdb->prepare( |
| 4278 |
"SELECT ID FROM {$wpdb->posts} |
| 4279 |
WHERE post_name = %s |
| 4280 |
AND post_status IN ('publish', 'closed', 'private') |
| 4281 |
AND post_type NOT IN ('revision', 'attachment', 'nav_menu_item') |
| 4282 |
ORDER BY CASE |
| 4283 |
WHEN post_type = 'post' THEN 1 |
| 4284 |
WHEN post_type = 'page' THEN 2 |
| 4285 |
WHEN post_type = 'topic' THEN 3 |
| 4286 |
WHEN post_type = 'product' THEN 4 |
| 4287 |
ELSE 5 |
| 4288 |
END |
| 4289 |
LIMIT 1", |
| 4290 |
$potential_slug |
| 4291 |
)); |
| 4292 |
|
| 4293 |
if ($post_id) { |
| 4294 |
return intval($post_id); |
| 4295 |
} |
| 4296 |
} |
| 4297 |
} |
| 4298 |
|
| 4299 |
// ADDITIONAL: Try direct database lookup by URL variations |
| 4300 |
global $wpdb; |
| 4301 |
$table_name = $wpdb->prefix . 'mxchat_system_prompt_content'; |
| 4302 |
|
| 4303 |
// Try variations of the URL (with/without trailing slash, http/https) |
| 4304 |
$url_variations = array( |
| 4305 |
$url, |
| 4306 |
rtrim($url, '/'), |
| 4307 |
$url . '/', |
| 4308 |
str_replace('http://', 'https://', $url), |
| 4309 |
str_replace('https://', 'http://', $url), |
| 4310 |
str_replace('http://', 'https://', rtrim($url, '/')), |
| 4311 |
str_replace('https://', 'http://', rtrim($url, '/')) |
| 4312 |
); |
| 4313 |
|
| 4314 |
// Remove duplicates |
| 4315 |
$url_variations = array_unique($url_variations); |
| 4316 |
|
| 4317 |
foreach ($url_variations as $variation) { |
| 4318 |
$existing_record = $wpdb->get_row($wpdb->prepare( |
| 4319 |
"SELECT id, source_url FROM $table_name WHERE source_url = %s", |
| 4320 |
$variation |
| 4321 |
)); |
| 4322 |
|
| 4323 |
if ($existing_record) { |
| 4324 |
// Try to get post ID from this stored URL |
| 4325 |
$stored_post_id = url_to_postid($existing_record->source_url); |
| 4326 |
if ($stored_post_id > 0) { |
| 4327 |
return $stored_post_id; |
| 4328 |
} |
| 4329 |
} |
| 4330 |
} |
| 4331 |
|
| 4332 |
return 0; // No match found |
| 4333 |
} |
| 4334 |
/** |
| 4335 |
* Process selected content via AJAX |
| 4336 |
*/ |
| 4337 |
public function ajax_mxchat_process_selected_content() { |
| 4338 |
// Basic request validation |
| 4339 |
if (!check_ajax_referer('mxchat_content_selector_nonce', 'nonce', false)) { |
| 4340 |
wp_send_json_error('Invalid nonce'); |
| 4341 |
exit; |
| 4342 |
} |
| 4343 |
|
| 4344 |
if (!current_user_can('manage_options')) { |
| 4345 |
wp_send_json_error('Unauthorized access'); |
| 4346 |
exit; |
| 4347 |
} |
| 4348 |
|
| 4349 |
// Get post IDs - safely parse the array |
| 4350 |
$post_ids = array(); |
| 4351 |
if (isset($_POST['post_ids']) && is_array($_POST['post_ids'])) { |
| 4352 |
foreach ($_POST['post_ids'] as $id) { |
| 4353 |
$post_ids[] = absint($id); |
| 4354 |
} |
| 4355 |
} |
| 4356 |
|
| 4357 |
if (empty($post_ids)) { |
| 4358 |
wp_send_json_error('No content selected'); |
| 4359 |
exit; |
| 4360 |
} |
| 4361 |
|
| 4362 |
// Get bot_id from request |
| 4363 |
$bot_id = isset($_POST['bot_id']) ? sanitize_key($_POST['bot_id']) : 'default'; |
| 4364 |
|
| 4365 |
// ACF→PDF extraction is an install-level setting (Knowledge → ACF Fields, |
| 4366 |
// plan 11720c). The import modal shows a passive status line pointing |
| 4367 |
// there; the old per-batch checkbox and its remembered default are gone. |
| 4368 |
$extract_acf_pdfs = get_option('mxchat_acf_pdf_extraction', '0') === '1'; |
| 4369 |
|
| 4370 |
// Process only ONE post at a time to avoid request size issues |
| 4371 |
$post_id = reset($post_ids); |
| 4372 |
$post = get_post($post_id); |
| 4373 |
|
| 4374 |
if (!$post) { |
| 4375 |
wp_send_json_error('Post not found'); |
| 4376 |
exit; |
| 4377 |
} |
| 4378 |
|
| 4379 |
/** |
| 4380 |
* Allow developers to modify post data before processing into the knowledge base. |
| 4381 |
* Applied on BOTH content-preparation paths (this manual bulk import and the |
| 4382 |
* auto-sync path in mxchat_handle_post_update) with the same signature, so a |
| 4383 |
* callback registered once covers every indexing route. Purely additive — |
| 4384 |
* zero behaviour change when unhooked. |
| 4385 |
* |
| 4386 |
* @param WP_Post $post The post about to be indexed. |
| 4387 |
* @param string $bot_id Bot context for this import. |
| 4388 |
*/ |
| 4389 |
$post = apply_filters('mxchat_before_process_post', $post, $bot_id); |
| 4390 |
if (!($post instanceof WP_Post)) { |
| 4391 |
$post = get_post($post_id); // defend against a bad callback return |
| 4392 |
} |
| 4393 |
|
| 4394 |
// Assemble the indexable text via the shared post-kind assembler (a3d60c). |
| 4395 |
// Bulk import reads raw post fields, has always included product custom tabs, |
| 4396 |
// and passes the install-level ACF→PDF option. |
| 4397 |
$prepared = $this->mxchat_prepare_post_content_for_indexing($post_id, $post, array( |
| 4398 |
'read_display' => false, |
| 4399 |
'extract_acf_pdfs' => $extract_acf_pdfs, |
| 4400 |
'include_product_tabs' => true, |
| 4401 |
)); |
| 4402 |
$content = $prepared['content']; |
| 4403 |
$pdf_extracted_count = $prepared['pdf_extracted_count']; |
| 4404 |
|
| 4405 |
// Debug logging for WordPress Import content |
| 4406 |
//error_log('[MXCHAT-WP-IMPORT-DEBUG] Post ID: ' . $post_id . ' Title: ' . $post->post_title); |
| 4407 |
//error_log('[MXCHAT-WP-IMPORT-DEBUG] Raw post_content length: ' . strlen($post->post_content)); |
| 4408 |
//error_log('[MXCHAT-WP-IMPORT-DEBUG] Final content length: ' . strlen($content)); |
| 4409 |
//error_log('[MXCHAT-WP-IMPORT-DEBUG] Content preview: ' . substr($content, 0, 300)); |
| 4410 |
|
| 4411 |
// Note: Removed 10,000 char limit - chunking now handles large content properly |
| 4412 |
|
| 4413 |
// Get bot-specific embedding decision — custom-provider-aware (plan cbd5fd) |
| 4414 |
$bot_options = $this->get_bot_options($bot_id); |
| 4415 |
$options = !empty($bot_options) ? $bot_options : get_option('mxchat_options'); |
| 4416 |
|
| 4417 |
$preflight = MxChat_Utils::embedding_preflight($options); |
| 4418 |
if (!$preflight['ok']) { |
| 4419 |
MxChat_Admin::mxchat_log_debug('api_error', $preflight['reason'] . ' (knowledge processing)'); |
| 4420 |
wp_send_json_error($preflight['reason']); |
| 4421 |
exit; |
| 4422 |
} |
| 4423 |
$api_key = $preflight['api_key']; |
| 4424 |
|
| 4425 |
$source_url = get_permalink($post_id); |
| 4426 |
$vector_id = md5($source_url); // Vector ID for Pinecone |
| 4427 |
|
| 4428 |
// Check for existing content in bot-specific storage |
| 4429 |
$is_update = false; |
| 4430 |
|
| 4431 |
// Get bot-specific Pinecone configuration |
| 4432 |
$bot_pinecone_config = $this->get_bot_pinecone_config($bot_id); |
| 4433 |
$use_pinecone = !empty($bot_pinecone_config) && ($bot_pinecone_config['use_pinecone'] ?? false); |
| 4434 |
|
| 4435 |
if ($use_pinecone && !empty($bot_pinecone_config['api_key'])) { |
| 4436 |
// Check Pinecone for this bot |
| 4437 |
$pinecone_data = $this->mxchat_get_pinecone_processed_content($bot_pinecone_config); |
| 4438 |
if (isset($pinecone_data[$post_id])) { |
| 4439 |
$is_update = true; |
| 4440 |
} |
| 4441 |
} else { |
| 4442 |
// Check WordPress DB (same as before since it's shared) |
| 4443 |
global $wpdb; |
| 4444 |
$table_name = $wpdb->prefix . 'mxchat_system_prompt_content'; |
| 4445 |
$existing_record = $wpdb->get_row($wpdb->prepare( |
| 4446 |
"SELECT id FROM $table_name WHERE source_url = %s", |
| 4447 |
$source_url |
| 4448 |
)); |
| 4449 |
|
| 4450 |
if ($existing_record) { |
| 4451 |
$is_update = true; |
| 4452 |
} |
| 4453 |
} |
| 4454 |
|
| 4455 |
// UPDATED 2.5.6: Determine content type based on post_type |
| 4456 |
$post_type = $post->post_type; |
| 4457 |
$content_type = 'content'; // Default fallback |
| 4458 |
|
| 4459 |
// Map WordPress post types to content types |
| 4460 |
switch ($post_type) { |
| 4461 |
case 'post': |
| 4462 |
$content_type = 'post'; |
| 4463 |
break; |
| 4464 |
case 'page': |
| 4465 |
$content_type = 'page'; |
| 4466 |
break; |
| 4467 |
case 'product': |
| 4468 |
$content_type = 'product'; |
| 4469 |
break; |
| 4470 |
default: |
| 4471 |
// For custom post types, use the post type name |
| 4472 |
$content_type = sanitize_key($post_type); |
| 4473 |
break; |
| 4474 |
} |
| 4475 |
|
| 4476 |
// Use the centralized utility function with bot_id and content_type |
| 4477 |
$result = MxChat_Utils::submit_content_to_db( |
| 4478 |
$content, |
| 4479 |
$source_url, |
| 4480 |
$api_key, |
| 4481 |
$vector_id, |
| 4482 |
$bot_id, |
| 4483 |
$content_type |
| 4484 |
); |
| 4485 |
|
| 4486 |
if (is_wp_error($result)) { |
| 4487 |
MxChat_Admin::mxchat_log_debug('storage_error', 'Knowledge storage failed: ' . $result->get_error_message(), array('source_url' => $source_url)); |
| 4488 |
wp_send_json_error('Storage failed: ' . $result->get_error_message()); |
| 4489 |
exit; |
| 4490 |
} |
| 4491 |
|
| 4492 |
// Automatically apply role restriction based on tags |
| 4493 |
$this->apply_role_restriction_to_post($post_id, $source_url); |
| 4494 |
|
| 4495 |
$operation_type = $is_update ? 'update' : 'new'; |
| 4496 |
|
| 4497 |
// Count ACF fields for debugging |
| 4498 |
$acf_field_count = $prepared['acf_fields_found']; |
| 4499 |
|
| 4500 |
// Success response with minimal data |
| 4501 |
wp_send_json_success(array( |
| 4502 |
'message' => $operation_type === 'update' ? 'Content updated successfully' : 'Content processed successfully', |
| 4503 |
'post_id' => $post_id, |
| 4504 |
'title' => $post->post_title, |
| 4505 |
'operation_type' => $operation_type, |
| 4506 |
'vector_id' => $vector_id, |
| 4507 |
'acf_fields_found' => $acf_field_count, |
| 4508 |
'pdf_extracted_count' => (int) $pdf_extracted_count, |
| 4509 |
'content_preview' => substr($content, 0, 100) . '...', |
| 4510 |
'bot_id' => $bot_id |
| 4511 |
)); |
| 4512 |
exit; |
| 4513 |
} |
| 4514 |
|
| 4515 |
private function apply_role_restriction_to_post($post_id, $source_url) { |
| 4516 |
// Get tag-role mappings |
| 4517 |
$mappings = get_option('mxchat_tag_role_mappings', array()); |
| 4518 |
|
| 4519 |
if (empty($mappings)) { |
| 4520 |
return; // No mappings, leave as public |
| 4521 |
} |
| 4522 |
|
| 4523 |
// Get all tags for the post |
| 4524 |
$post_tags = wp_get_post_tags($post_id, array('fields' => 'slugs')); |
| 4525 |
|
| 4526 |
if (empty($post_tags)) { |
| 4527 |
return; // No tags, leave as public |
| 4528 |
} |
| 4529 |
|
| 4530 |
// Determine the highest role restriction based on tags |
| 4531 |
$highest_role = 'public'; |
| 4532 |
$role_hierarchy = array( |
| 4533 |
'public' => 0, |
| 4534 |
'logged_in' => 1, |
| 4535 |
'subscriber' => 2, |
| 4536 |
'contributor' => 3, |
| 4537 |
'author' => 4, |
| 4538 |
'editor' => 5, |
| 4539 |
'administrator' => 6 |
| 4540 |
); |
| 4541 |
|
| 4542 |
foreach ($post_tags as $tag_slug) { |
| 4543 |
if (isset($mappings[$tag_slug])) { |
| 4544 |
$role = $mappings[$tag_slug]; |
| 4545 |
if (isset($role_hierarchy[$role]) && $role_hierarchy[$role] > $role_hierarchy[$highest_role]) { |
| 4546 |
$highest_role = $role; |
| 4547 |
} |
| 4548 |
} |
| 4549 |
} |
| 4550 |
|
| 4551 |
// If no restricted tags found, return (leave as public) |
| 4552 |
if ($highest_role === 'public') { |
| 4553 |
return; |
| 4554 |
} |
| 4555 |
|
| 4556 |
// Update the role restriction in the database |
| 4557 |
global $wpdb; |
| 4558 |
|
| 4559 |
// Check if using Pinecone |
| 4560 |
$pinecone_options = get_option('mxchat_pinecone_addon_options', array()); |
| 4561 |
$use_pinecone = ($pinecone_options['mxchat_use_pinecone'] ?? '0') === '1'; |
| 4562 |
|
| 4563 |
if ($use_pinecone && !empty($pinecone_options['mxchat_pinecone_api_key'])) { |
| 4564 |
// Update Pinecone role restriction |
| 4565 |
$roles_table = $wpdb->prefix . 'mxchat_pinecone_roles'; |
| 4566 |
$vector_id = md5($source_url); |
| 4567 |
|
| 4568 |
$wpdb->replace( |
| 4569 |
$roles_table, |
| 4570 |
array( |
| 4571 |
'vector_id' => $vector_id, |
| 4572 |
'role_restriction' => $highest_role, |
| 4573 |
'updated_at' => current_time('mysql') |
| 4574 |
), |
| 4575 |
array('%s', '%s', '%s') |
| 4576 |
); |
| 4577 |
} else { |
| 4578 |
// Update WordPress DB |
| 4579 |
$table_name = $wpdb->prefix . 'mxchat_system_prompt_content'; |
| 4580 |
|
| 4581 |
$wpdb->update( |
| 4582 |
$table_name, |
| 4583 |
array('role_restriction' => $highest_role), |
| 4584 |
array('source_url' => $source_url), |
| 4585 |
array('%s'), |
| 4586 |
array('%s') |
| 4587 |
); |
| 4588 |
} |
| 4589 |
|
| 4590 |
// The entry's restriction just changed — keep the OpenAI Vector Store |
| 4591 |
// mirror consistent: non-public pulls the file (file_search has no |
| 4592 |
// per-role filtering), public re-mirrors it (plan 15b5c6). |
| 4593 |
if (class_exists('MxChat_Vectorstore_Manager')) { |
| 4594 |
MxChat_Vectorstore_Manager::handle_role_change($source_url, 'default', $highest_role); |
| 4595 |
} |
| 4596 |
} |
| 4597 |
|
| 4598 |
public function mxchat_get_public_post_types() { |
| 4599 |
// Get all public post types |
| 4600 |
$post_types = get_post_types(array('public' => true), 'objects'); |
| 4601 |
$post_type_options = array(); |
| 4602 |
|
| 4603 |
foreach ($post_types as $post_type) { |
| 4604 |
$post_type_options[$post_type->name] = $post_type->label; |
| 4605 |
} |
| 4606 |
|
| 4607 |
// Also include common forum/community post types that might not be marked as public |
| 4608 |
$additional_types = array( |
| 4609 |
'topic' => 'Forum Topics (bbPress)', |
| 4610 |
'reply' => 'Forum Replies (bbPress)', |
| 4611 |
'forum' => 'Forums (bbPress)', |
| 4612 |
'wpforo_topic' => 'wpForo Topics', |
| 4613 |
'wpforo_post' => 'wpForo Posts' |
| 4614 |
); |
| 4615 |
|
| 4616 |
foreach ($additional_types as $type_name => $type_label) { |
| 4617 |
if (post_type_exists($type_name) && !isset($post_type_options[$type_name])) { |
| 4618 |
$post_type_options[$type_name] = $type_label; |
| 4619 |
} |
| 4620 |
} |
| 4621 |
|
| 4622 |
return $post_type_options; |
| 4623 |
} |
| 4624 |
|
| 4625 |
/** |
| 4626 |
* Retrieves processed content from Pinecone API |
| 4627 |
*/ |
| 4628 |
public function mxchat_get_pinecone_processed_content($pinecone_options) { |
| 4629 |
$api_key = $pinecone_options['mxchat_pinecone_api_key'] ?? ''; |
| 4630 |
$host = $pinecone_options['mxchat_pinecone_host'] ?? ''; |
| 4631 |
|
| 4632 |
if (empty($api_key) || empty($host)) { |
| 4633 |
return array(); |
| 4634 |
} |
| 4635 |
|
| 4636 |
$pinecone_data = array(); |
| 4637 |
|
| 4638 |
try { |
| 4639 |
// Always get fresh data from Pinecone |
| 4640 |
$pinecone_data = $this->mxchat_scan_pinecone_for_processed_content($pinecone_options); |
| 4641 |
|
| 4642 |
// Method 2: Final fallback - try stats endpoint (if available) |
| 4643 |
if (empty($pinecone_data)) { |
| 4644 |
$stats_url = "https://{$host}/describe_index_stats"; |
| 4645 |
|
| 4646 |
$response = wp_remote_post($stats_url, array( |
| 4647 |
'headers' => array( |
| 4648 |
'Api-Key' => $api_key, |
| 4649 |
'Content-Type' => 'application/json' |
| 4650 |
), |
| 4651 |
'body' => json_encode(array()), |
| 4652 |
'timeout' => 30 |
| 4653 |
)); |
| 4654 |
|
| 4655 |
if (!is_wp_error($response) && wp_remote_retrieve_response_code($response) === 200) { |
| 4656 |
$body = wp_remote_retrieve_body($response); |
| 4657 |
$stats_data = json_decode($body, true); |
| 4658 |
} |
| 4659 |
} |
| 4660 |
|
| 4661 |
} catch (Exception $e) { |
| 4662 |
// Log error but return fresh data only |
| 4663 |
} |
| 4664 |
|
| 4665 |
return $pinecone_data; |
| 4666 |
} |
| 4667 |
public function mxchat_fetch_pinecone_vectors_by_ids($pinecone_options, $vector_ids) { |
| 4668 |
$api_key = $pinecone_options['mxchat_pinecone_api_key'] ?? ''; |
| 4669 |
$host = $pinecone_options['mxchat_pinecone_host'] ?? ''; |
| 4670 |
$namespace = $pinecone_options['mxchat_pinecone_namespace'] ?? ''; |
| 4671 |
|
| 4672 |
if (empty($api_key) || empty($host) || empty($vector_ids)) { |
| 4673 |
return array(); |
| 4674 |
} |
| 4675 |
|
| 4676 |
try { |
| 4677 |
// NOTE (plan 793b82): /vectors/fetch is a GET endpoint with the ids |
| 4678 |
// repeated in the query string (ids=a&ids=b — http_build_query would |
| 4679 |
// emit ids[0]=a); the old POST here was answered 200-with-an-empty-body, |
| 4680 |
// which read as "nothing indexed". Chunked at 100 ids to stay well |
| 4681 |
// under the measured HTTP 414 URL-length boundary. |
| 4682 |
$vectors = array(); |
| 4683 |
foreach (array_chunk(array_values($vector_ids), 100) as $chunk) { |
| 4684 |
$fetch_query = array(); |
| 4685 |
foreach ($chunk as $fetch_vid) { |
| 4686 |
$fetch_query[] = 'ids=' . rawurlencode($fetch_vid); |
| 4687 |
} |
| 4688 |
if (!empty($namespace)) { |
| 4689 |
$fetch_query[] = 'namespace=' . rawurlencode($namespace); |
| 4690 |
} |
| 4691 |
|
| 4692 |
$response = wp_remote_get("https://{$host}/vectors/fetch?" . implode('&', $fetch_query), array( |
| 4693 |
'headers' => array( |
| 4694 |
'Api-Key' => $api_key, |
| 4695 |
'accept' => 'application/json' |
| 4696 |
), |
| 4697 |
'timeout' => 30 |
| 4698 |
)); |
| 4699 |
|
| 4700 |
if (is_wp_error($response)) { |
| 4701 |
error_log('MxChat Pinecone: mxchat_fetch_pinecone_vectors_by_ids GET failed: ' . $response->get_error_message()); |
| 4702 |
continue; |
| 4703 |
} |
| 4704 |
|
| 4705 |
if (wp_remote_retrieve_response_code($response) !== 200) { |
| 4706 |
error_log('MxChat Pinecone: mxchat_fetch_pinecone_vectors_by_ids GET returned HTTP ' . wp_remote_retrieve_response_code($response)); |
| 4707 |
continue; |
| 4708 |
} |
| 4709 |
|
| 4710 |
$data = json_decode(wp_remote_retrieve_body($response), true); |
| 4711 |
if (isset($data['vectors']) && is_array($data['vectors'])) { |
| 4712 |
$vectors += $data['vectors']; |
| 4713 |
} |
| 4714 |
} |
| 4715 |
|
| 4716 |
if (empty($vectors)) { |
| 4717 |
return array(); |
| 4718 |
} |
| 4719 |
|
| 4720 |
$processed_data = array(); |
| 4721 |
|
| 4722 |
foreach ($vectors as $vector_id => $vector_data) { |
| 4723 |
$metadata = $vector_data['metadata'] ?? array(); |
| 4724 |
$source_url = $metadata['source_url'] ?? ''; |
| 4725 |
|
| 4726 |
if (!empty($source_url)) { |
| 4727 |
$post_id = url_to_postid($source_url); |
| 4728 |
if ($post_id) { |
| 4729 |
$created_at = $metadata['created_at'] ?? ''; |
| 4730 |
$processed_date = 'Recently'; |
| 4731 |
|
| 4732 |
if (!empty($created_at)) { |
| 4733 |
$timestamp = is_numeric($created_at) ? $created_at : strtotime($created_at); |
| 4734 |
if ($timestamp) { |
| 4735 |
$processed_date = human_time_diff($timestamp, current_time('timestamp')) . ' ago'; |
| 4736 |
} |
| 4737 |
} |
| 4738 |
|
| 4739 |
$processed_data[$post_id] = array( |
| 4740 |
'db_id' => $vector_id, |
| 4741 |
'processed_date' => $processed_date, |
| 4742 |
'url' => $source_url, |
| 4743 |
'source' => 'pinecone', |
| 4744 |
'timestamp' => $timestamp ?? current_time('timestamp') |
| 4745 |
); |
| 4746 |
} |
| 4747 |
} |
| 4748 |
} |
| 4749 |
|
| 4750 |
//error_log('DEBUG: Processed ' . count($processed_data) . ' vectors from fetch'); |
| 4751 |
return $processed_data; |
| 4752 |
|
| 4753 |
} catch (Exception $e) { |
| 4754 |
//error_log('DEBUG: Exception in fetch_pinecone_vectors_by_ids: ' . $e->getMessage()); |
| 4755 |
return array(); |
| 4756 |
} |
| 4757 |
} |
| 4758 |
|
| 4759 |
/** |
| 4760 |
* Get embedding dimensions based on the selected model. |
| 4761 |
*/ |
| 4762 |
private function mxchat_get_embedding_dimensions() { |
| 4763 |
$options = get_option('mxchat_options', array()); |
| 4764 |
$selected_model = $options['embedding_model'] ?? 'text-embedding-ada-002'; |
| 4765 |
|
| 4766 |
$model_dimensions = array( |
| 4767 |
'text-embedding-ada-002' => 1536, |
| 4768 |
'text-embedding-3-small' => 1536, |
| 4769 |
'text-embedding-3-large' => 3072, |
| 4770 |
'voyage-2' => 1024, |
| 4771 |
'voyage-large-2' => 1536, |
| 4772 |
'voyage-3-large' => 2048, |
| 4773 |
'gemini-embedding-001' => 1536, |
| 4774 |
); |
| 4775 |
|
| 4776 |
if (strpos($selected_model, 'voyage-3-large') === 0) { |
| 4777 |
$custom_dimensions = $options['voyage_output_dimension'] ?? 2048; |
| 4778 |
return intval($custom_dimensions); |
| 4779 |
} |
| 4780 |
|
| 4781 |
if (strpos($selected_model, 'gemini-embedding') === 0) { |
| 4782 |
$custom_dimensions = $options['gemini_output_dimension'] ?? 1536; |
| 4783 |
return intval($custom_dimensions); |
| 4784 |
} |
| 4785 |
|
| 4786 |
return $model_dimensions[$selected_model] ?? 1536; |
| 4787 |
} |
| 4788 |
|
| 4789 |
/** |
| 4790 |
* Scan Pinecone for processed content |
| 4791 |
*/ |
| 4792 |
public function mxchat_scan_pinecone_for_processed_content($pinecone_options) { |
| 4793 |
$api_key = $pinecone_options['mxchat_pinecone_api_key'] ?? ''; |
| 4794 |
$host = $pinecone_options['mxchat_pinecone_host'] ?? ''; |
| 4795 |
// plan 793b82: this scan was namespace-blind — on a namespaced setup it |
| 4796 |
// surveyed the default namespace and reported the wrong content as indexed. |
| 4797 |
$namespace = $pinecone_options['mxchat_pinecone_namespace'] ?? ''; |
| 4798 |
|
| 4799 |
if (empty($api_key) || empty($host)) { |
| 4800 |
return array(); |
| 4801 |
} |
| 4802 |
|
| 4803 |
try { |
| 4804 |
// Use multiple random vectors to get better coverage |
| 4805 |
$all_matches = array(); |
| 4806 |
$seen_ids = array(); |
| 4807 |
|
| 4808 |
// Get correct dimensions for the configured embedding model |
| 4809 |
$dimensions = $this->mxchat_get_embedding_dimensions(); |
| 4810 |
|
| 4811 |
// Try 3 different random vectors to get better coverage |
| 4812 |
for ($i = 0; $i < 3; $i++) { |
| 4813 |
$query_url = "https://{$host}/query"; |
| 4814 |
|
| 4815 |
// Generate a random unit vector instead of zeros |
| 4816 |
$random_vector = array(); |
| 4817 |
for ($j = 0; $j < $dimensions; $j++) { |
| 4818 |
$random_vector[] = (rand(-1000, 1000) / 1000.0); |
| 4819 |
} |
| 4820 |
|
| 4821 |
// Normalize the vector to unit length |
| 4822 |
$magnitude = sqrt(array_sum(array_map(function($x) { return $x * $x; }, $random_vector))); |
| 4823 |
if ($magnitude > 0) { |
| 4824 |
$random_vector = array_map(function($x) use ($magnitude) { return $x / $magnitude; }, $random_vector); |
| 4825 |
} |
| 4826 |
|
| 4827 |
$query_data = array( |
| 4828 |
'includeMetadata' => true, |
| 4829 |
'includeValues' => false, |
| 4830 |
'topK' => 10000, |
| 4831 |
'vector' => $random_vector |
| 4832 |
); |
| 4833 |
|
| 4834 |
if (!empty($namespace)) { |
| 4835 |
$query_data['namespace'] = $namespace; |
| 4836 |
} |
| 4837 |
|
| 4838 |
$response = wp_remote_post($query_url, array( |
| 4839 |
'headers' => array( |
| 4840 |
'Api-Key' => $api_key, |
| 4841 |
'Content-Type' => 'application/json' |
| 4842 |
), |
| 4843 |
'body' => json_encode($query_data), |
| 4844 |
'timeout' => 30 |
| 4845 |
)); |
| 4846 |
|
| 4847 |
if (is_wp_error($response)) { |
| 4848 |
continue; |
| 4849 |
} |
| 4850 |
|
| 4851 |
$response_code = wp_remote_retrieve_response_code($response); |
| 4852 |
|
| 4853 |
if ($response_code !== 200) { |
| 4854 |
continue; |
| 4855 |
} |
| 4856 |
|
| 4857 |
$body = wp_remote_retrieve_body($response); |
| 4858 |
$data = json_decode($body, true); |
| 4859 |
|
| 4860 |
if (isset($data['matches'])) { |
| 4861 |
foreach ($data['matches'] as $match) { |
| 4862 |
$match_id = $match['id'] ?? ''; |
| 4863 |
if (!empty($match_id) && !isset($seen_ids[$match_id])) { |
| 4864 |
$all_matches[] = $match; |
| 4865 |
$seen_ids[$match_id] = true; |
| 4866 |
} |
| 4867 |
} |
| 4868 |
} |
| 4869 |
} |
| 4870 |
|
| 4871 |
// Convert matches to processed data format, grouping by URL to count chunks |
| 4872 |
$processed_data = array(); |
| 4873 |
$url_chunk_counts = array(); |
| 4874 |
|
| 4875 |
foreach ($all_matches as $match) { |
| 4876 |
$metadata = $match['metadata'] ?? array(); |
| 4877 |
$source_url = $metadata['source_url'] ?? ''; |
| 4878 |
$match_id = $match['id'] ?? ''; |
| 4879 |
|
| 4880 |
if (!empty($source_url) && !empty($match_id)) { |
| 4881 |
$post_id = url_to_postid($source_url); |
| 4882 |
if ($post_id) { |
| 4883 |
// Count chunks per post_id |
| 4884 |
if (!isset($url_chunk_counts[$post_id])) { |
| 4885 |
$url_chunk_counts[$post_id] = 0; |
| 4886 |
} |
| 4887 |
$url_chunk_counts[$post_id]++; |
| 4888 |
|
| 4889 |
$created_at = $metadata['created_at'] ?? ''; |
| 4890 |
$processed_date = 'Recently'; |
| 4891 |
|
| 4892 |
if (!empty($created_at)) { |
| 4893 |
$timestamp = is_numeric($created_at) ? $created_at : strtotime($created_at); |
| 4894 |
if ($timestamp) { |
| 4895 |
$processed_date = human_time_diff($timestamp, current_time('timestamp')) . ' ago'; |
| 4896 |
} |
| 4897 |
} |
| 4898 |
|
| 4899 |
// Only store if not already set, or update with newer timestamp |
| 4900 |
if (!isset($processed_data[$post_id]) || |
| 4901 |
($timestamp ?? 0) > ($processed_data[$post_id]['timestamp'] ?? 0)) { |
| 4902 |
$processed_data[$post_id] = array( |
| 4903 |
'db_id' => $match_id, |
| 4904 |
'processed_date' => $processed_date, |
| 4905 |
'url' => $source_url, |
| 4906 |
'source' => 'pinecone', |
| 4907 |
'timestamp' => $timestamp ?? current_time('timestamp') |
| 4908 |
); |
| 4909 |
} |
| 4910 |
} |
| 4911 |
} |
| 4912 |
} |
| 4913 |
|
| 4914 |
// Add chunk counts to processed data |
| 4915 |
foreach ($url_chunk_counts as $post_id => $chunk_count) { |
| 4916 |
if (isset($processed_data[$post_id])) { |
| 4917 |
$processed_data[$post_id]['chunk_count'] = $chunk_count; |
| 4918 |
} |
| 4919 |
} |
| 4920 |
|
| 4921 |
return $processed_data; |
| 4922 |
|
| 4923 |
} catch (Exception $e) { |
| 4924 |
return array(); |
| 4925 |
} |
| 4926 |
} |
| 4927 |
/** |
| 4928 |
* Generate embeddings from input text for MXChat with bot support |
| 4929 |
*/ |
| 4930 |
private function mxchat_generate_embedding($text, $bot_id = 'default') { |
| 4931 |
// Enable detailed logging for debugging |
| 4932 |
//error_log('[MXCHAT-EMBED] Starting embedding generation for bot: ' . $bot_id . '. Text length: ' . strlen($text) . ' bytes'); |
| 4933 |
//error_log('[MXCHAT-EMBED] Text preview: ' . substr($text, 0, 100) . '...'); |
| 4934 |
|
| 4935 |
// Get bot-specific options |
| 4936 |
$bot_options = $this->get_bot_options($bot_id); |
| 4937 |
$options = !empty($bot_options) ? $bot_options : get_option('mxchat_options'); |
| 4938 |
|
| 4939 |
// Opt-in: when the custom provider is selected for embeddings, index through |
| 4940 |
// the same custom endpoint the query path uses so stored vectors and query |
| 4941 |
// vectors share a model. Returns the vector array on success, or an error |
| 4942 |
// string on failure (this function's existing failure contract). |
| 4943 |
if (isset($options['custom_provider_for_embeddings']) && $options['custom_provider_for_embeddings'] === 'on') { |
| 4944 |
if (!class_exists('MxChat_Utils')) { |
| 4945 |
require_once dirname(__FILE__) . '/../includes/class-mxchat-utils.php'; |
| 4946 |
} |
| 4947 |
return MxChat_Utils::generate_embedding_custom($text, $options); |
| 4948 |
} |
| 4949 |
|
| 4950 |
$selected_model = $options['embedding_model'] ?? 'text-embedding-ada-002'; |
| 4951 |
//error_log('[MXCHAT-EMBED] Selected embedding model for bot ' . $bot_id . ': ' . $selected_model); |
| 4952 |
|
| 4953 |
// Determine provider and endpoint |
| 4954 |
if (strpos($selected_model, 'voyage') === 0) { |
| 4955 |
$api_key = $options['voyage_api_key'] ?? ''; |
| 4956 |
$endpoint = 'https://api.voyageai.com/v1/embeddings'; |
| 4957 |
$provider_name = 'Voyage AI'; |
| 4958 |
//error_log('[MXCHAT-EMBED] Using Voyage AI API for bot ' . $bot_id); |
| 4959 |
} elseif (strpos($selected_model, 'gemini-embedding') === 0) { |
| 4960 |
$api_key = $options['gemini_api_key'] ?? ''; |
| 4961 |
$endpoint = 'https://generativelanguage.googleapis.com/v1beta/models/' . $selected_model . ':embedContent'; |
| 4962 |
$provider_name = 'Google Gemini'; |
| 4963 |
//error_log('[MXCHAT-EMBED] Using Google Gemini API for bot ' . $bot_id); |
| 4964 |
} else { |
| 4965 |
$api_key = $options['api_key'] ?? ''; |
| 4966 |
$endpoint = 'https://api.openai.com/v1/embeddings'; |
| 4967 |
$provider_name = 'OpenAI'; |
| 4968 |
//error_log('[MXCHAT-EMBED] Using OpenAI API for bot ' . $bot_id); |
| 4969 |
} |
| 4970 |
|
| 4971 |
//error_log('[MXCHAT-EMBED] Using endpoint: ' . $endpoint); |
| 4972 |
|
| 4973 |
if (empty($api_key)) { |
| 4974 |
$error_message = sprintf('Missing %s API key for bot %s. Please configure your API key in the bot settings.', $provider_name, $bot_id); |
| 4975 |
//error_log('[MXCHAT-EMBED] Error: ' . $error_message); |
| 4976 |
return $error_message; |
| 4977 |
} |
| 4978 |
|
| 4979 |
// Check if text is too long (for OpenAI, roughly estimate tokens as words/0.75) |
| 4980 |
$estimated_tokens = ceil(str_word_count($text) / 0.75); |
| 4981 |
//error_log('[MXCHAT-EMBED] Estimated token count: ~' . $estimated_tokens); |
| 4982 |
|
| 4983 |
if ($estimated_tokens > 8000 && strpos($selected_model, 'voyage') === false && strpos($selected_model, 'gemini-embedding') === false) { |
| 4984 |
//error_log('[MXCHAT-EMBED] Warning: Text may exceed OpenAI token limits (8K for most models)'); |
| 4985 |
// Consider truncating text here |
| 4986 |
} |
| 4987 |
|
| 4988 |
// Prepare request body based on provider |
| 4989 |
if (strpos($selected_model, 'gemini-embedding') === 0) { |
| 4990 |
// Gemini API format |
| 4991 |
$request_body = array( |
| 4992 |
'model' => 'models/' . $selected_model, |
| 4993 |
'content' => array( |
| 4994 |
'parts' => array( |
| 4995 |
array('text' => $text) |
| 4996 |
) |
| 4997 |
) |
| 4998 |
); |
| 4999 |
|
| 5000 |
// Set output dimensionality to 1536 for consistency with other models |
| 5001 |
$request_body['outputDimensionality'] = 1536; |
| 5002 |
} else { |
| 5003 |
// OpenAI/Voyage API format |
| 5004 |
$request_body = array( |
| 5005 |
'model' => $selected_model, |
| 5006 |
'input' => $text |
| 5007 |
); |
| 5008 |
|
| 5009 |
// Add output_dimension for voyage-3-large model |
| 5010 |
if ($selected_model === 'voyage-3-large') { |
| 5011 |
$request_body['output_dimension'] = 2048; |
| 5012 |
} |
| 5013 |
} |
| 5014 |
|
| 5015 |
//error_log('[MXCHAT-EMBED] Request prepared with model: ' . $selected_model); |
| 5016 |
|
| 5017 |
// Prepare headers based on provider |
| 5018 |
if (strpos($selected_model, 'gemini-embedding') === 0) { |
| 5019 |
// Gemini uses API key as query parameter |
| 5020 |
$endpoint .= '?key=' . $api_key; |
| 5021 |
$headers = array( |
| 5022 |
'Content-Type' => 'application/json' |
| 5023 |
); |
| 5024 |
} else { |
| 5025 |
// OpenAI/Voyage use Bearer token |
| 5026 |
$headers = array( |
| 5027 |
'Authorization' => 'Bearer ' . $api_key, |
| 5028 |
'Content-Type' => 'application/json' |
| 5029 |
); |
| 5030 |
} |
| 5031 |
|
| 5032 |
// Make API request |
| 5033 |
//error_log('[MXCHAT-EMBED] Sending API request to: ' . $endpoint); |
| 5034 |
$response = wp_remote_post($endpoint, array( |
| 5035 |
'body' => wp_json_encode($request_body), |
| 5036 |
'headers' => $headers, |
| 5037 |
'timeout' => 60 // Increased timeout for large inputs |
| 5038 |
)); |
| 5039 |
|
| 5040 |
// Handle wp_remote_post errors |
| 5041 |
if (is_wp_error($response)) { |
| 5042 |
$error_message = $response->get_error_message(); |
| 5043 |
//error_log('[MXCHAT-EMBED] WP Remote Post Error: ' . $error_message); |
| 5044 |
return 'Connection error: ' . $error_message; |
| 5045 |
} |
| 5046 |
|
| 5047 |
// Get and check HTTP response code |
| 5048 |
$http_code = wp_remote_retrieve_response_code($response); |
| 5049 |
//error_log('[MXCHAT-EMBED] API Response Code: ' . $http_code); |
| 5050 |
|
| 5051 |
if ($http_code !== 200) { |
| 5052 |
$error_body = wp_remote_retrieve_body($response); |
| 5053 |
//error_log('[MXCHAT-EMBED] API Error Response Body: ' . $error_body); |
| 5054 |
|
| 5055 |
// Try to parse error for more details |
| 5056 |
$error_json = json_decode($error_body, true); |
| 5057 |
if (json_last_error() === JSON_ERROR_NONE && isset($error_json['error'])) { |
| 5058 |
$error_type = $error_json['error']['type'] ?? 'unknown'; |
| 5059 |
$error_message = $error_json['error']['message'] ?? 'No message'; |
| 5060 |
//error_log('[MXCHAT-EMBED] API Error Type: ' . $error_type); |
| 5061 |
//error_log('[MXCHAT-EMBED] API Error Message: ' . $error_message); |
| 5062 |
|
| 5063 |
// Keep the provider's own diagnostic — a restricted-key 401 names the |
| 5064 |
// exact missing scope, and replacing it with "check your API key" sent |
| 5065 |
// a customer to regenerate two keys (plan 46b596). Same shape as |
| 5066 |
// MxChat_Utils::embedding_failure_error() so both ingestion paths read |
| 5067 |
// identically. Key never appears in provider messages, but scrub anyway. |
| 5068 |
if ($error_type === 'invalid_request_error' || $error_type === 'authentication_error') { |
| 5069 |
if (is_string($api_key) && $api_key !== '') { |
| 5070 |
$error_message = str_replace($api_key, '[redacted]', $error_message); |
| 5071 |
} |
| 5072 |
$error_message = sprintf( |
| 5073 |
'Embedding failed (%s, HTTP %d): %s', |
| 5074 |
$selected_model, |
| 5075 |
$http_code, |
| 5076 |
substr($error_message, 0, 300) |
| 5077 |
); |
| 5078 |
} |
| 5079 |
|
| 5080 |
//error_log('[MXCHAT-EMBED] Returning error: ' . $error_message); |
| 5081 |
return $error_message; |
| 5082 |
} |
| 5083 |
|
| 5084 |
$error_message = sprintf("API Error (HTTP %d): Unable to generate embedding for bot %s", $http_code, $bot_id); |
| 5085 |
//error_log('[MXCHAT-EMBED] Returning error: ' . $error_message); |
| 5086 |
return $error_message; |
| 5087 |
} |
| 5088 |
|
| 5089 |
// Parse response body |
| 5090 |
$response_body = wp_remote_retrieve_body($response); |
| 5091 |
//error_log('[MXCHAT-EMBED] Received response length: ' . strlen($response_body) . ' bytes'); |
| 5092 |
|
| 5093 |
$response_data = json_decode($response_body, true); |
| 5094 |
|
| 5095 |
if (json_last_error() !== JSON_ERROR_NONE) { |
| 5096 |
$error = json_last_error_msg(); |
| 5097 |
//error_log('[MXCHAT-EMBED] JSON Parse Error: ' . $error); |
| 5098 |
//error_log('[MXCHAT-EMBED] Response preview: ' . substr($response_body, 0, 200)); |
| 5099 |
return "Failed to parse API response: $error"; |
| 5100 |
} |
| 5101 |
|
| 5102 |
// Handle different response formats based on provider |
| 5103 |
if (strpos($selected_model, 'gemini-embedding') === 0) { |
| 5104 |
// Gemini API response format |
| 5105 |
if (isset($response_data['embedding']['values'])) { |
| 5106 |
$embedding_dimensions = count($response_data['embedding']['values']); |
| 5107 |
//error_log('[MXCHAT-EMBED] Successfully extracted Gemini embedding with ' . $embedding_dimensions . ' dimensions'); |
| 5108 |
|
| 5109 |
// Check if embedding dimensions are as expected (should be 1536) |
| 5110 |
if ($embedding_dimensions !== 1536) { |
| 5111 |
//error_log('[MXCHAT-EMBED] Warning: Unexpected Gemini embedding dimensions: ' . $embedding_dimensions); |
| 5112 |
} |
| 5113 |
|
| 5114 |
return $response_data['embedding']['values']; |
| 5115 |
} else { |
| 5116 |
//error_log('[MXCHAT-EMBED] Error: No embedding found in Gemini response'); |
| 5117 |
//error_log('[MXCHAT-EMBED] Response structure: ' . wp_json_encode(array_keys($response_data))); |
| 5118 |
|
| 5119 |
if (isset($response_data['error'])) { |
| 5120 |
$error_message = "Gemini API Error in response: " . wp_json_encode($response_data['error']); |
| 5121 |
//error_log('[MXCHAT-EMBED] ' . $error_message); |
| 5122 |
return $error_message; |
| 5123 |
} |
| 5124 |
|
| 5125 |
$error_message = "Invalid Gemini API response format: No embedding found"; |
| 5126 |
//error_log('[MXCHAT-EMBED] ' . $error_message); |
| 5127 |
return $error_message; |
| 5128 |
} |
| 5129 |
} else { |
| 5130 |
// OpenAI/Voyage API response format |
| 5131 |
if (isset($response_data['data'][0]['embedding'])) { |
| 5132 |
$embedding_dimensions = count($response_data['data'][0]['embedding']); |
| 5133 |
//error_log('[MXCHAT-EMBED] Successfully extracted embedding with ' . $embedding_dimensions . ' dimensions'); |
| 5134 |
|
| 5135 |
// Check if embedding dimensions are as expected |
| 5136 |
if (($selected_model === 'text-embedding-ada-002' && $embedding_dimensions !== 1536) || |
| 5137 |
($selected_model === 'voyage-3-large' && $embedding_dimensions !== 2048)) { |
| 5138 |
//error_log('[MXCHAT-EMBED] Warning: Unexpected embedding dimensions'); |
| 5139 |
} |
| 5140 |
|
| 5141 |
return $response_data['data'][0]['embedding']; |
| 5142 |
} else { |
| 5143 |
//error_log('[MXCHAT-EMBED] Error: No embedding found in response'); |
| 5144 |
//error_log('[MXCHAT-EMBED] Response structure: ' . wp_json_encode(array_keys($response_data))); |
| 5145 |
|
| 5146 |
if (isset($response_data['error'])) { |
| 5147 |
$error_message = "API Error in response: " . wp_json_encode($response_data['error']); |
| 5148 |
//error_log('[MXCHAT-EMBED] ' . $error_message); |
| 5149 |
return $error_message; |
| 5150 |
} |
| 5151 |
|
| 5152 |
$error_message = "Invalid API response format: No embedding found"; |
| 5153 |
//error_log('[MXCHAT-EMBED] ' . $error_message); |
| 5154 |
return $error_message; |
| 5155 |
} |
| 5156 |
} |
| 5157 |
} |
| 5158 |
|
| 5159 |
/** |
| 5160 |
* Get bot-specific options for multi-bot functionality |
| 5161 |
* Falls back to default options if bot_id is 'default' or multi-bot add-on is not active |
| 5162 |
*/ |
| 5163 |
private function get_bot_options($bot_id = 'default') { |
| 5164 |
//error_log("MXCHAT DEBUG: get_bot_options called for bot: " . $bot_id); |
| 5165 |
|
| 5166 |
if ($bot_id === 'default' || !class_exists('MxChat_Multi_Bot_Manager')) { |
| 5167 |
//error_log("MXCHAT DEBUG: Using default options (no multi-bot or bot is 'default')"); |
| 5168 |
return array(); |
| 5169 |
} |
| 5170 |
|
| 5171 |
$bot_options = apply_filters('mxchat_get_bot_options', array(), $bot_id); |
| 5172 |
|
| 5173 |
if (!empty($bot_options)) { |
| 5174 |
//error_log("MXCHAT DEBUG: Got bot-specific options from filter"); |
| 5175 |
if (isset($bot_options['similarity_threshold'])) { |
| 5176 |
//error_log(" - similarity_threshold: " . $bot_options['similarity_threshold']); |
| 5177 |
} |
| 5178 |
} |
| 5179 |
|
| 5180 |
return is_array($bot_options) ? $bot_options : array(); |
| 5181 |
} |
| 5182 |
|
| 5183 |
/** |
| 5184 |
* Get bot-specific Pinecone configuration |
| 5185 |
* Used in the knowledge retrieval functions |
| 5186 |
*/ |
| 5187 |
// Also add debugging to your get_bot_pinecone_config function |
| 5188 |
private function get_bot_pinecone_config($bot_id = 'default') { |
| 5189 |
//error_log("MXCHAT DEBUG: get_bot_pinecone_config called for bot: " . $bot_id); |
| 5190 |
|
| 5191 |
// If default bot or multi-bot add-on not active, use default Pinecone config |
| 5192 |
if ($bot_id === 'default' || !class_exists('MxChat_Multi_Bot_Manager')) { |
| 5193 |
//error_log("MXCHAT DEBUG: Using default Pinecone config (no multi-bot or bot is 'default')"); |
| 5194 |
$addon_options = get_option('mxchat_pinecone_addon_options', array()); |
| 5195 |
$config = array( |
| 5196 |
'use_pinecone' => (isset($addon_options['mxchat_use_pinecone']) && $addon_options['mxchat_use_pinecone'] === '1'), |
| 5197 |
'api_key' => $addon_options['mxchat_pinecone_api_key'] ?? '', |
| 5198 |
'host' => $addon_options['mxchat_pinecone_host'] ?? '', |
| 5199 |
'namespace' => $addon_options['mxchat_pinecone_namespace'] ?? '' |
| 5200 |
); |
| 5201 |
//error_log("MXCHAT DEBUG: Default config - use_pinecone: " . ($config['use_pinecone'] ? 'true' : 'false')); |
| 5202 |
return $config; |
| 5203 |
} |
| 5204 |
|
| 5205 |
//error_log("MXCHAT DEBUG: Calling filter 'mxchat_get_bot_pinecone_config' for bot: " . $bot_id); |
| 5206 |
|
| 5207 |
// Hook for multi-bot add-on to provide bot-specific Pinecone config |
| 5208 |
$bot_pinecone_config = apply_filters('mxchat_get_bot_pinecone_config', array(), $bot_id); |
| 5209 |
|
| 5210 |
if (!empty($bot_pinecone_config)) { |
| 5211 |
//error_log("MXCHAT DEBUG: Got bot-specific config from filter"); |
| 5212 |
//error_log(" - use_pinecone: " . (isset($bot_pinecone_config['use_pinecone']) ? ($bot_pinecone_config['use_pinecone'] ? 'true' : 'false') : 'not set')); |
| 5213 |
//error_log(" - host: " . ($bot_pinecone_config['host'] ?? 'not set')); |
| 5214 |
//error_log(" - namespace: " . ($bot_pinecone_config['namespace'] ?? 'not set')); |
| 5215 |
} else { |
| 5216 |
//error_log("MXCHAT DEBUG: Filter returned empty config!"); |
| 5217 |
} |
| 5218 |
|
| 5219 |
return is_array($bot_pinecone_config) ? $bot_pinecone_config : array(); |
| 5220 |
} |
| 5221 |
|
| 5222 |
|
| 5223 |
public function mxchat_ajax_dismiss_completed_status() { |
| 5224 |
try { |
| 5225 |
// Verify the request |
| 5226 |
check_ajax_referer('mxchat_status_nonce', 'nonce'); |
| 5227 |
|
| 5228 |
if (!current_user_can('manage_options')) { |
| 5229 |
wp_send_json_error('Unauthorized access'); |
| 5230 |
exit; |
| 5231 |
} |
| 5232 |
|
| 5233 |
$card_type = isset($_POST['card_type']) ? sanitize_text_field($_POST['card_type']) : ''; |
| 5234 |
|
| 5235 |
if ($card_type === 'pdf') { |
| 5236 |
// Clear PDF status |
| 5237 |
$pdf_url = get_transient('mxchat_last_pdf_url'); |
| 5238 |
if ($pdf_url) { |
| 5239 |
delete_transient('mxchat_pdf_status_' . md5($pdf_url)); |
| 5240 |
delete_transient('mxchat_last_pdf_url'); |
| 5241 |
} |
| 5242 |
} elseif ($card_type === 'sitemap') { |
| 5243 |
// Clear sitemap status |
| 5244 |
$sitemap_url = get_transient('mxchat_last_sitemap_url'); |
| 5245 |
if ($sitemap_url) { |
| 5246 |
delete_transient('mxchat_sitemap_status_' . md5($sitemap_url)); |
| 5247 |
delete_transient('mxchat_last_sitemap_url'); |
| 5248 |
} |
| 5249 |
} |
| 5250 |
|
| 5251 |
wp_send_json_success(array('message' => 'Status dismissed successfully')); |
| 5252 |
|
| 5253 |
} catch (Exception $e) { |
| 5254 |
wp_send_json_error(array('message' => 'Error dismissing status: ' . $e->getMessage())); |
| 5255 |
} |
| 5256 |
} |
| 5257 |
|
| 5258 |
/** |
| 5259 |
* Render completed status cards on page load |
| 5260 |
* This ensures completed processing status persists through page refreshes |
| 5261 |
*/ |
| 5262 |
public function mxchat_render_completed_status_cards() { |
| 5263 |
$output = ''; |
| 5264 |
|
| 5265 |
// Check for completed PDF status |
| 5266 |
$pdf_url = get_transient('mxchat_last_pdf_url'); |
| 5267 |
if ($pdf_url) { |
| 5268 |
$pdf_status = $this->mxchat_get_pdf_processing_status($pdf_url); |
| 5269 |
if ($pdf_status && ($pdf_status['status'] === 'complete' || $pdf_status['status'] === 'error')) { |
| 5270 |
$output .= $this->mxchat_render_pdf_status_card($pdf_status, $pdf_url); |
| 5271 |
} |
| 5272 |
} |
| 5273 |
|
| 5274 |
// Check for completed sitemap status |
| 5275 |
$sitemap_url = get_transient('mxchat_last_sitemap_url'); |
| 5276 |
if ($sitemap_url) { |
| 5277 |
$sitemap_status = $this->mxchat_get_sitemap_processing_status($sitemap_url); |
| 5278 |
if ($sitemap_status && ($sitemap_status['status'] === 'complete' || $sitemap_status['status'] === 'error')) { |
| 5279 |
$output .= $this->mxchat_render_sitemap_status_card($sitemap_status, $sitemap_url); |
| 5280 |
} |
| 5281 |
} |
| 5282 |
|
| 5283 |
return $output; |
| 5284 |
} |
| 5285 |
|
| 5286 |
/** |
| 5287 |
* Render PDF status card HTML |
| 5288 |
*/ |
| 5289 |
private function mxchat_render_pdf_status_card($status, $pdf_url) { |
| 5290 |
$html = '<div class="mxchat-status-card" data-card-type="pdf">'; |
| 5291 |
$html .= '<div class="mxchat-status-header">'; |
| 5292 |
$html .= '<h4>' . esc_html__('PDF Processing Status', 'mxchat') . '</h4>'; |
| 5293 |
|
| 5294 |
// Add dismiss button for completed status |
| 5295 |
if ($status['status'] === 'complete' || $status['status'] === 'error') { |
| 5296 |
$html .= '<button type="button" class="mxchat-dismiss-button">' . esc_html__('Dismiss', 'mxchat') . '</button>'; |
| 5297 |
} |
| 5298 |
|
| 5299 |
// Process Batch button for processing status |
| 5300 |
if ($status['status'] === 'processing') { |
| 5301 |
$html .= '<button type="button" class="mxchat-manual-batch-btn" |
| 5302 |
data-process-type="pdf" |
| 5303 |
data-url="' . esc_attr($pdf_url) . '"> |
| 5304 |
' . esc_html__('Process Batch', 'mxchat') . '</button>'; |
| 5305 |
} |
| 5306 |
|
| 5307 |
// Add status badges |
| 5308 |
if ($status['status'] === 'error') { |
| 5309 |
$html .= '<span class="mxchat-status-badge mxchat-status-failed">' . esc_html__('Error', 'mxchat') . '</span>'; |
| 5310 |
} elseif ($status['status'] === 'complete') { |
| 5311 |
if ($status['failed_pages'] > 0) { |
| 5312 |
$html .= '<span class="mxchat-status-badge mxchat-status-warning">' . |
| 5313 |
sprintf(esc_html__('Completed with %d failures', 'mxchat'), $status['failed_pages']) . '</span>'; |
| 5314 |
} else { |
| 5315 |
$html .= '<span class="mxchat-status-badge mxchat-status-success">' . esc_html__('Complete', 'mxchat') . '</span>'; |
| 5316 |
} |
| 5317 |
} |
| 5318 |
|
| 5319 |
$html .= '</div>'; // End header |
| 5320 |
|
| 5321 |
// Progress bar |
| 5322 |
$html .= '<div class="mxchat-progress-bar">'; |
| 5323 |
$html .= '<div class="mxchat-progress-fill" style="width: ' . esc_attr($status['percentage']) . '%"></div>'; |
| 5324 |
$html .= '</div>'; |
| 5325 |
|
| 5326 |
// Status details |
| 5327 |
$html .= '<div class="mxchat-status-details">'; |
| 5328 |
$html .= '<p>' . sprintf( |
| 5329 |
esc_html__('Progress: %d of %d pages (%d%%)', 'mxchat'), |
| 5330 |
$status['processed_pages'], |
| 5331 |
$status['total_pages'], |
| 5332 |
$status['percentage'] |
| 5333 |
) . '</p>'; |
| 5334 |
|
| 5335 |
// Show failed pages count if any |
| 5336 |
if ($status['failed_pages'] > 0) { |
| 5337 |
$html .= '<p><strong>' . esc_html__('Failed pages:', 'mxchat') . '</strong> ' . esc_html($status['failed_pages']) . '</p>'; |
| 5338 |
} |
| 5339 |
|
| 5340 |
$html .= '<p><strong>' . esc_html__('Status:', 'mxchat') . '</strong> ' . esc_html(ucfirst($status['status'])) . '</p>'; |
| 5341 |
$html .= '<p><strong>' . esc_html__('Last update:', 'mxchat') . '</strong> ' . esc_html($status['last_update']) . '</p>'; |
| 5342 |
|
| 5343 |
// Add completion summary if available AND it's an array |
| 5344 |
if (isset($status['completion_summary']) && is_array($status['completion_summary']) && !empty($status['completion_summary'])) { |
| 5345 |
$summary = $status['completion_summary']; |
| 5346 |
$html .= '<div class="mxchat-completion-summary">'; |
| 5347 |
$html .= '<h5>' . esc_html__('Processing Summary', 'mxchat') . '</h5>'; |
| 5348 |
$html .= '<p><strong>' . esc_html__('Total Pages:', 'mxchat') . '</strong> ' . esc_html($summary['total_pages']) . '</p>'; |
| 5349 |
$html .= '<p><strong>' . esc_html__('Successful:', 'mxchat') . '</strong> ' . esc_html($summary['successful_pages']) . '</p>'; |
| 5350 |
$html .= '<p><strong>' . esc_html__('Failed:', 'mxchat') . '</strong> ' . esc_html($summary['failed_pages']) . '</p>'; |
| 5351 |
$html .= '<p><strong>' . esc_html__('Completed:', 'mxchat') . '</strong> ' . esc_html($summary['completion_time']) . '</p>'; |
| 5352 |
$html .= '</div>'; |
| 5353 |
} |
| 5354 |
|
| 5355 |
// Add failed pages list if any AND it's an array |
| 5356 |
if (isset($status['failed_pages_list']) && is_array($status['failed_pages_list']) && !empty($status['failed_pages_list'])) { |
| 5357 |
$html .= $this->mxchat_render_failed_pages_list($status['failed_pages_list']); |
| 5358 |
} |
| 5359 |
|
| 5360 |
// Add error message if any |
| 5361 |
if (isset($status['error']) && !empty($status['error'])) { |
| 5362 |
$html .= '<div class="mxchat-error-notice">'; |
| 5363 |
$html .= '<p class="error">' . esc_html($status['error']) . '</p>'; |
| 5364 |
$html .= '</div>'; |
| 5365 |
} |
| 5366 |
|
| 5367 |
$html .= '</div>'; // End details |
| 5368 |
$html .= '</div>'; // End card |
| 5369 |
|
| 5370 |
return $html; |
| 5371 |
} |
| 5372 |
/** |
| 5373 |
* Render sitemap status card HTML |
| 5374 |
*/ |
| 5375 |
private function mxchat_render_sitemap_status_card($status, $sitemap_url) { |
| 5376 |
$html = '<div class="mxchat-status-card" data-card-type="sitemap">'; |
| 5377 |
$html .= '<div class="mxchat-status-header">'; |
| 5378 |
$html .= '<h4>' . esc_html__('Sitemap Processing Status', 'mxchat') . '</h4>'; |
| 5379 |
|
| 5380 |
// Add dismiss button for completed status |
| 5381 |
if ($status['status'] === 'complete' || $status['status'] === 'error') { |
| 5382 |
$html .= '<button type="button" class="mxchat-dismiss-button">' . esc_html__('Dismiss', 'mxchat') . '</button>'; |
| 5383 |
} |
| 5384 |
|
| 5385 |
// Process Batch button for processing status |
| 5386 |
if ($status['status'] === 'processing') { |
| 5387 |
$html .= '<button type="button" class="mxchat-manual-batch-btn" |
| 5388 |
data-process-type="sitemap" |
| 5389 |
data-url="' . esc_attr($sitemap_url) . '"> |
| 5390 |
' . esc_html__('Process Batch', 'mxchat') . '</button>'; |
| 5391 |
} |
| 5392 |
|
| 5393 |
// Add status badges |
| 5394 |
if ($status['status'] === 'error') { |
| 5395 |
$html .= '<span class="mxchat-status-badge mxchat-status-failed">' . esc_html__('Error', 'mxchat') . '</span>'; |
| 5396 |
} elseif ($status['status'] === 'complete') { |
| 5397 |
if ($status['failed_urls'] > 0) { |
| 5398 |
$html .= '<span class="mxchat-status-badge mxchat-status-warning">' . |
| 5399 |
sprintf(esc_html__('Completed with %d failures', 'mxchat'), $status['failed_urls']) . '</span>'; |
| 5400 |
} else { |
| 5401 |
$html .= '<span class="mxchat-status-badge mxchat-status-success">' . esc_html__('Complete', 'mxchat') . '</span>'; |
| 5402 |
} |
| 5403 |
} |
| 5404 |
|
| 5405 |
$html .= '</div>'; // End header |
| 5406 |
|
| 5407 |
// Progress bar |
| 5408 |
$html .= '<div class="mxchat-progress-bar">'; |
| 5409 |
$html .= '<div class="mxchat-progress-fill" style="width: ' . esc_attr($status['percentage']) . '%"></div>'; |
| 5410 |
$html .= '</div>'; |
| 5411 |
|
| 5412 |
// Status details |
| 5413 |
$html .= '<div class="mxchat-status-details">'; |
| 5414 |
$html .= '<p>' . sprintf( |
| 5415 |
esc_html__('Progress: %d of %d URLs (%d%%)', 'mxchat'), |
| 5416 |
$status['processed_urls'], |
| 5417 |
$status['total_urls'], |
| 5418 |
$status['percentage'] |
| 5419 |
) . '</p>'; |
| 5420 |
|
| 5421 |
// Show failed URLs count if any |
| 5422 |
if ($status['failed_urls'] > 0) { |
| 5423 |
$html .= '<p><strong>' . esc_html__('Failed URLs:', 'mxchat') . '</strong> ' . esc_html($status['failed_urls']) . '</p>'; |
| 5424 |
} |
| 5425 |
|
| 5426 |
$html .= '<p><strong>' . esc_html__('Status:', 'mxchat') . '</strong> ' . esc_html(ucfirst($status['status'])) . '</p>'; |
| 5427 |
$html .= '<p><strong>' . esc_html__('Last update:', 'mxchat') . '</strong> ' . esc_html($status['last_update']) . '</p>'; |
| 5428 |
|
| 5429 |
// Add completion summary if available AND it's an array |
| 5430 |
if (isset($status['completion_summary']) && is_array($status['completion_summary']) && !empty($status['completion_summary'])) { |
| 5431 |
$summary = $status['completion_summary']; |
| 5432 |
$html .= '<div class="mxchat-completion-summary">'; |
| 5433 |
$html .= '<h5>' . esc_html__('Processing Summary', 'mxchat') . '</h5>'; |
| 5434 |
$html .= '<p><strong>' . esc_html__('Total URLs:', 'mxchat') . '</strong> ' . esc_html($summary['total_urls']) . '</p>'; |
| 5435 |
$html .= '<p><strong>' . esc_html__('Successful:', 'mxchat') . '</strong> ' . esc_html($summary['successful_urls']) . '</p>'; |
| 5436 |
$html .= '<p><strong>' . esc_html__('Failed:', 'mxchat') . '</strong> ' . esc_html($summary['failed_urls']) . '</p>'; |
| 5437 |
$html .= '<p><strong>' . esc_html__('Completed:', 'mxchat') . '</strong> ' . esc_html($summary['completion_time']) . '</p>'; |
| 5438 |
$html .= '</div>'; |
| 5439 |
} |
| 5440 |
|
| 5441 |
// Add error messages if any (but not the failed URLs list) |
| 5442 |
if (!empty($status['error']) || !empty($status['last_error'])) { |
| 5443 |
$html .= '<div class="mxchat-error-notice">'; |
| 5444 |
|
| 5445 |
if (!empty($status['error'])) { |
| 5446 |
$html .= '<p class="error">' . esc_html($status['error']) . '</p>'; |
| 5447 |
} |
| 5448 |
|
| 5449 |
if (!empty($status['last_error'])) { |
| 5450 |
$html .= '<p class="last-error">' . esc_html__('Last error:', 'mxchat') . ' ' . esc_html($status['last_error']) . '</p>'; |
| 5451 |
} |
| 5452 |
|
| 5453 |
$html .= '</div>'; |
| 5454 |
} |
| 5455 |
|
| 5456 |
$html .= '</div>'; // End details |
| 5457 |
$html .= '</div>'; // End card |
| 5458 |
|
| 5459 |
return $html; |
| 5460 |
} |
| 5461 |
|
| 5462 |
|
| 5463 |
/** |
| 5464 |
* Render failed pages list |
| 5465 |
*/ |
| 5466 |
private function mxchat_render_failed_pages_list($failed_pages_list) { |
| 5467 |
// Validate that $failed_pages_list is an array and not empty |
| 5468 |
if (!is_array($failed_pages_list) || empty($failed_pages_list)) { |
| 5469 |
return ''; |
| 5470 |
} |
| 5471 |
|
| 5472 |
$html = '<div class="mxchat-error-notice">'; |
| 5473 |
$html .= '<div class="mxchat-failed-pages-container">'; |
| 5474 |
$html .= '<h5>' . sprintf(esc_html__('Failed Pages (%d)', 'mxchat'), count($failed_pages_list)) . '</h5>'; |
| 5475 |
$html .= '<details>'; |
| 5476 |
$html .= '<summary>' . esc_html__('Show Failed Pages', 'mxchat') . '</summary>'; |
| 5477 |
$html .= '<div class="mxchat-failed-pages-list">'; |
| 5478 |
|
| 5479 |
// Create table for failed pages |
| 5480 |
$html .= '<table class="widefat striped">'; |
| 5481 |
$html .= '<thead><tr>'; |
| 5482 |
$html .= '<th>' . esc_html__('Page', 'mxchat') . '</th>'; |
| 5483 |
$html .= '<th>' . esc_html__('Error', 'mxchat') . '</th>'; |
| 5484 |
$html .= '<th>' . esc_html__('Retries', 'mxchat') . '</th>'; |
| 5485 |
$html .= '<th>' . esc_html__('Time', 'mxchat') . '</th>'; |
| 5486 |
$html .= '</tr></thead><tbody>'; |
| 5487 |
|
| 5488 |
// Sort failed pages by most recent |
| 5489 |
$sorted_failed_pages = $failed_pages_list; |
| 5490 |
usort($sorted_failed_pages, function($a, $b) { |
| 5491 |
return ($b['time'] ?? 0) - ($a['time'] ?? 0); |
| 5492 |
}); |
| 5493 |
|
| 5494 |
foreach ($sorted_failed_pages as $item) { |
| 5495 |
// Ensure $item is an array before accessing its elements |
| 5496 |
if (!is_array($item)) { |
| 5497 |
continue; |
| 5498 |
} |
| 5499 |
|
| 5500 |
$time_ago = isset($item['time']) ? human_time_diff($item['time'], current_time('timestamp')) . ' ' . esc_html__('ago', 'mxchat') : 'Unknown'; |
| 5501 |
$html .= '<tr>'; |
| 5502 |
$html .= '<td>' . esc_html__('Page', 'mxchat') . ' ' . esc_html($item['page'] ?? 'Unknown') . '</td>'; |
| 5503 |
$html .= '<td style="word-break: break-word;">' . esc_html($item['error'] ?? 'Unknown error') . '</td>'; |
| 5504 |
$html .= '<td>' . esc_html($item['retries'] ?? 'N/A') . '</td>'; |
| 5505 |
$html .= '<td>' . esc_html($time_ago) . '</td>'; |
| 5506 |
$html .= '</tr>'; |
| 5507 |
} |
| 5508 |
|
| 5509 |
$html .= '</tbody></table>'; |
| 5510 |
$html .= '</div></details></div></div>'; |
| 5511 |
|
| 5512 |
return $html; |
| 5513 |
} |
| 5514 |
|
| 5515 |
/** |
| 5516 |
* Render failed URLs list |
| 5517 |
*/ |
| 5518 |
private function mxchat_render_failed_urls_list($failed_urls_list) { |
| 5519 |
// Validate that $failed_urls_list is an array and not empty |
| 5520 |
if (!is_array($failed_urls_list) || empty($failed_urls_list)) { |
| 5521 |
return ''; |
| 5522 |
} |
| 5523 |
|
| 5524 |
$html = '<div class="mxchat-failed-urls-container">'; |
| 5525 |
$html .= '<h5>' . sprintf(esc_html__('Failed URLs (%d)', 'mxchat'), count($failed_urls_list)) . '</h5>'; |
| 5526 |
$html .= '<details>'; |
| 5527 |
$html .= '<summary>' . esc_html__('Show Failed URLs', 'mxchat') . '</summary>'; |
| 5528 |
$html .= '<div class="mxchat-failed-urls-list">'; |
| 5529 |
|
| 5530 |
// Create table for failed URLs |
| 5531 |
$html .= '<table class="widefat striped">'; |
| 5532 |
$html .= '<thead><tr>'; |
| 5533 |
$html .= '<th>' . esc_html__('URL', 'mxchat') . '</th>'; |
| 5534 |
$html .= '<th>' . esc_html__('Error', 'mxchat') . '</th>'; |
| 5535 |
$html .= '<th>' . esc_html__('Retries', 'mxchat') . '</th>'; |
| 5536 |
$html .= '<th>' . esc_html__('Time', 'mxchat') . '</th>'; |
| 5537 |
$html .= '</tr></thead><tbody>'; |
| 5538 |
|
| 5539 |
// Sort failed URLs by most recent |
| 5540 |
$sorted_failed_urls = $failed_urls_list; |
| 5541 |
usort($sorted_failed_urls, function($a, $b) { |
| 5542 |
return ($b['time'] ?? 0) - ($a['time'] ?? 0); |
| 5543 |
}); |
| 5544 |
|
| 5545 |
// Show up to 50 failed URLs |
| 5546 |
$display_urls = array_slice($sorted_failed_urls, 0, 50); |
| 5547 |
|
| 5548 |
foreach ($display_urls as $item) { |
| 5549 |
// Ensure $item is an array before accessing its elements |
| 5550 |
if (!is_array($item)) { |
| 5551 |
continue; |
| 5552 |
} |
| 5553 |
|
| 5554 |
$url = $item['url'] ?? ''; |
| 5555 |
$time_ago = isset($item['time']) ? human_time_diff($item['time'], current_time('timestamp')) . ' ' . esc_html__('ago', 'mxchat') : 'Unknown'; |
| 5556 |
|
| 5557 |
// Truncate URL for display |
| 5558 |
$display_url = strlen($url) > 50 ? substr($url, 0, 47) . '...' : $url; |
| 5559 |
|
| 5560 |
$html .= '<tr>'; |
| 5561 |
$html .= '<td style="word-break: break-all;">'; |
| 5562 |
if (!empty($url)) { |
| 5563 |
$html .= '<a href="' . esc_url($url) . '" target="_blank" rel="noopener noreferrer">' . esc_html($display_url) . '</a>'; |
| 5564 |
} else { |
| 5565 |
$html .= esc_html__('Unknown URL', 'mxchat'); |
| 5566 |
} |
| 5567 |
$html .= '</td>'; |
| 5568 |
$html .= '<td style="word-break: break-word;">' . esc_html($item['error'] ?? 'Unknown error') . '</td>'; |
| 5569 |
$html .= '<td>' . esc_html($item['retries'] ?? 'N/A') . '</td>'; |
| 5570 |
$html .= '<td>' . esc_html($time_ago) . '</td>'; |
| 5571 |
$html .= '</tr>'; |
| 5572 |
} |
| 5573 |
|
| 5574 |
$html .= '</tbody></table>'; |
| 5575 |
|
| 5576 |
if (count($failed_urls_list) > 50) { |
| 5577 |
$html .= '<div class="mxchat-failed-urls-more">+ ' . |
| 5578 |
(count($failed_urls_list) - 50) . |
| 5579 |
' ' . esc_html__('more failed URLs not shown', 'mxchat') . '</div>'; |
| 5580 |
} |
| 5581 |
|
| 5582 |
$html .= '</div></details></div>'; |
| 5583 |
|
| 5584 |
return $html; |
| 5585 |
} |
| 5586 |
|
| 5587 |
/** |
| 5588 |
* Get all ACF fields for a specific post, excluding any fields the user has disabled |
| 5589 |
*/ |
| 5590 |
public function mxchat_get_acf_fields_for_post($post_id) { |
| 5591 |
if (!function_exists('get_field_objects')) { |
| 5592 |
return array(); |
| 5593 |
} |
| 5594 |
|
| 5595 |
// Field OBJECTS, not get_fields(): exclusion matches on the field KEY |
| 5596 |
// (unique per field) rather than the name (shared across groups — plan |
| 5597 |
// 30e81f). ACF's own get_fields() is implemented as get_field_objects() |
| 5598 |
// reduced to name => value, so the un-excluded reduction below is the |
| 5599 |
// identical shape and order the previous get_fields() call produced. |
| 5600 |
$field_objects = get_field_objects($post_id); |
| 5601 |
if (!$field_objects || !is_array($field_objects)) { |
| 5602 |
return array(); |
| 5603 |
} |
| 5604 |
|
| 5605 |
$excluded_fields = get_option('mxchat_acf_excluded_fields', array()); |
| 5606 |
if (!is_array($excluded_fields)) { |
| 5607 |
$excluded_fields = array(); |
| 5608 |
} |
| 5609 |
|
| 5610 |
$fields = array(); |
| 5611 |
foreach ($field_objects as $field_name => $field_object) { |
| 5612 |
if (!empty($excluded_fields)) { |
| 5613 |
$field_key = isset($field_object['key']) ? $field_object['key'] : ''; |
| 5614 |
// Legacy name entries stay honored: a stored name whose group was |
| 5615 |
// inactive at migration time still excludes every field wearing it. |
| 5616 |
if (in_array($field_key, $excluded_fields, true) || in_array($field_name, $excluded_fields, true)) { |
| 5617 |
continue; |
| 5618 |
} |
| 5619 |
} |
| 5620 |
$fields[$field_name] = isset($field_object['value']) ? $field_object['value'] : null; |
| 5621 |
} |
| 5622 |
|
| 5623 |
return $fields; |
| 5624 |
} |
| 5625 |
|
| 5626 |
/** |
| 5627 |
* Get all registered ACF field groups and their fields for the settings UI |
| 5628 |
*/ |
| 5629 |
public function mxchat_get_all_acf_fields() { |
| 5630 |
if (!function_exists('acf_get_field_groups') || !function_exists('acf_get_fields')) { |
| 5631 |
return array(); |
| 5632 |
} |
| 5633 |
|
| 5634 |
// Keyed by GROUP KEY, not title (titles are not unique), and each field |
| 5635 |
// entry carries its ACF field key — the unique identifier every toggle, |
| 5636 |
// save, and index-time exclusion now runs on (plans 30e81f / bf57e0). |
| 5637 |
$all_fields = array(); |
| 5638 |
$field_groups = acf_get_field_groups(); |
| 5639 |
|
| 5640 |
if (!empty($field_groups)) { |
| 5641 |
foreach ($field_groups as $group) { |
| 5642 |
$group_fields = acf_get_fields($group['key']); |
| 5643 |
if (!empty($group_fields)) { |
| 5644 |
$entry = array( |
| 5645 |
'title' => $group['title'], |
| 5646 |
'fields' => array(), |
| 5647 |
); |
| 5648 |
foreach ($group_fields as $field) { |
| 5649 |
$entry['fields'][] = array( |
| 5650 |
'key' => $field['key'], |
| 5651 |
'name' => $field['name'], |
| 5652 |
'label' => $field['label'], |
| 5653 |
'type' => $field['type'] |
| 5654 |
); |
| 5655 |
} |
| 5656 |
$all_fields[$group['key']] = $entry; |
| 5657 |
} |
| 5658 |
} |
| 5659 |
} |
| 5660 |
|
| 5661 |
return $all_fields; |
| 5662 |
} |
| 5663 |
|
| 5664 |
/** |
| 5665 |
* Get whitelisted custom post meta for a given post |
| 5666 |
* This allows non-ACF meta fields (like OptionTree, theme meta boxes) to be included in embeddings |
| 5667 |
*/ |
| 5668 |
public function mxchat_get_whitelisted_post_meta($post_id) { |
| 5669 |
$whitelist = get_option('mxchat_custom_meta_whitelist', ''); |
| 5670 |
|
| 5671 |
if (empty($whitelist)) { |
| 5672 |
return array(); |
| 5673 |
} |
| 5674 |
|
| 5675 |
// Parse the whitelist - one meta key per line |
| 5676 |
$meta_keys = array_filter(array_map('trim', explode("\n", $whitelist))); |
| 5677 |
|
| 5678 |
if (empty($meta_keys)) { |
| 5679 |
return array(); |
| 5680 |
} |
| 5681 |
|
| 5682 |
$result = array(); |
| 5683 |
|
| 5684 |
foreach ($meta_keys as $key) { |
| 5685 |
// Skip empty keys |
| 5686 |
if (empty($key)) { |
| 5687 |
continue; |
| 5688 |
} |
| 5689 |
|
| 5690 |
$value = get_post_meta($post_id, $key, true); |
| 5691 |
|
| 5692 |
// Only include non-empty string values |
| 5693 |
if (!empty($value) && is_string($value)) { |
| 5694 |
$result[$key] = $value; |
| 5695 |
} elseif (!empty($value) && is_array($value)) { |
| 5696 |
// Handle array values by joining them |
| 5697 |
$flat_value = $this->mxchat_flatten_meta_array($value); |
| 5698 |
if (!empty($flat_value)) { |
| 5699 |
$result[$key] = $flat_value; |
| 5700 |
} |
| 5701 |
} |
| 5702 |
} |
| 5703 |
|
| 5704 |
return $result; |
| 5705 |
} |
| 5706 |
|
| 5707 |
/** |
| 5708 |
* Flatten array meta values into a readable string |
| 5709 |
*/ |
| 5710 |
private function mxchat_flatten_meta_array($array, $depth = 0) { |
| 5711 |
if ($depth > 3) { |
| 5712 |
return ''; // Prevent infinite recursion |
| 5713 |
} |
| 5714 |
|
| 5715 |
$parts = array(); |
| 5716 |
|
| 5717 |
foreach ($array as $key => $value) { |
| 5718 |
if (is_string($value) && !empty($value)) { |
| 5719 |
$parts[] = $value; |
| 5720 |
} elseif (is_array($value)) { |
| 5721 |
$nested = $this->mxchat_flatten_meta_array($value, $depth + 1); |
| 5722 |
if (!empty($nested)) { |
| 5723 |
$parts[] = $nested; |
| 5724 |
} |
| 5725 |
} |
| 5726 |
} |
| 5727 |
|
| 5728 |
return implode(', ', $parts); |
| 5729 |
} |
| 5730 |
|
| 5731 |
/** |
| 5732 |
* Format ACF field values for content extraction |
| 5733 |
*/ |
| 5734 |
public function mxchat_format_acf_field_value($value, $field_name = '', $post_id = 0) { |
| 5735 |
if (empty($value)) { |
| 5736 |
return ''; |
| 5737 |
} |
| 5738 |
|
| 5739 |
// Handle WP_Post objects first (THIS IS THE KEY FIX) |
| 5740 |
if ($value instanceof WP_Post) { |
| 5741 |
return $value->post_title ?: ''; |
| 5742 |
} |
| 5743 |
|
| 5744 |
// Handle other WP objects |
| 5745 |
if (is_object($value)) { |
| 5746 |
if (isset($value->post_title)) { |
| 5747 |
return $value->post_title; |
| 5748 |
} elseif (isset($value->display_name)) { |
| 5749 |
return $value->display_name; |
| 5750 |
} elseif (isset($value->name)) { |
| 5751 |
return $value->name; |
| 5752 |
} elseif (method_exists($value, '__toString')) { |
| 5753 |
try { |
| 5754 |
return (string) $value; |
| 5755 |
} catch (Exception $e) { |
| 5756 |
return ''; |
| 5757 |
} |
| 5758 |
} |
| 5759 |
// For any other objects, return empty string |
| 5760 |
return ''; |
| 5761 |
} |
| 5762 |
|
| 5763 |
// Handle different ACF field types |
| 5764 |
if (is_array($value)) { |
| 5765 |
// Check if it's an image/file field |
| 5766 |
if (isset($value['url'])) { |
| 5767 |
// Image field - return alt text, title, or caption |
| 5768 |
if (!empty($value['alt'])) { |
| 5769 |
return $value['alt']; |
| 5770 |
} elseif (!empty($value['title'])) { |
| 5771 |
return $value['title']; |
| 5772 |
} elseif (!empty($value['caption'])) { |
| 5773 |
return $value['caption']; |
| 5774 |
} else { |
| 5775 |
return ''; // Don't include just the URL |
| 5776 |
} |
| 5777 |
} |
| 5778 |
|
| 5779 |
// Check if it's a post object or relationship field |
| 5780 |
if (isset($value['post_title'])) { |
| 5781 |
return $value['post_title']; |
| 5782 |
} |
| 5783 |
|
| 5784 |
// Check if it's a user field |
| 5785 |
if (isset($value['display_name'])) { |
| 5786 |
return $value['display_name']; |
| 5787 |
} |
| 5788 |
|
| 5789 |
// Check if it's a taxonomy term |
| 5790 |
if (isset($value['name']) && isset($value['taxonomy'])) { |
| 5791 |
return $value['name']; |
| 5792 |
} |
| 5793 |
|
| 5794 |
// Check if it's a select field with label |
| 5795 |
if (isset($value['label'])) { |
| 5796 |
return $value['label']; |
| 5797 |
} |
| 5798 |
|
| 5799 |
// Check for repeater field or flexible content |
| 5800 |
if (is_numeric(key($value))) { |
| 5801 |
$sub_values = array(); |
| 5802 |
foreach ($value as $sub_item) { |
| 5803 |
if (is_array($sub_item)) { |
| 5804 |
// For repeater/flexible content, extract text values |
| 5805 |
$sub_text = $this->mxchat_extract_text_from_acf_array($sub_item); |
| 5806 |
if (!empty($sub_text)) { |
| 5807 |
$sub_values[] = $sub_text; |
| 5808 |
} |
| 5809 |
} elseif ($sub_item instanceof WP_Post) { |
| 5810 |
// Handle WP_Post objects in arrays |
| 5811 |
$sub_values[] = $sub_item->post_title ?: ''; |
| 5812 |
} else { |
| 5813 |
$sub_values[] = (string) $sub_item; |
| 5814 |
} |
| 5815 |
} |
| 5816 |
return implode(', ', array_filter($sub_values)); |
| 5817 |
} |
| 5818 |
|
| 5819 |
// For other arrays, try to extract meaningful text |
| 5820 |
$text_values = array(); |
| 5821 |
foreach ($value as $key => $val) { |
| 5822 |
if (is_string($val) && !empty(trim($val))) { |
| 5823 |
$text_values[] = trim($val); |
| 5824 |
} elseif ($val instanceof WP_Post) { |
| 5825 |
// Handle WP_Post objects in associative arrays |
| 5826 |
$text_values[] = $val->post_title ?: ''; |
| 5827 |
} elseif (is_array($val) && isset($val['post_title'])) { |
| 5828 |
$text_values[] = $val['post_title']; |
| 5829 |
} elseif (is_array($val) && isset($val['name'])) { |
| 5830 |
$text_values[] = $val['name']; |
| 5831 |
} |
| 5832 |
} |
| 5833 |
|
| 5834 |
return implode(', ', array_filter($text_values)); |
| 5835 |
} |
| 5836 |
|
| 5837 |
// Handle boolean values |
| 5838 |
if (is_bool($value)) { |
| 5839 |
return $value ? 'Yes' : 'No'; |
| 5840 |
} |
| 5841 |
|
| 5842 |
// Handle numeric values |
| 5843 |
if (is_numeric($value)) { |
| 5844 |
return (string) $value; |
| 5845 |
} |
| 5846 |
|
| 5847 |
// Handle string values |
| 5848 |
if (is_string($value)) { |
| 5849 |
return trim($value); |
| 5850 |
} |
| 5851 |
|
| 5852 |
// For anything else that we can't handle, return empty string |
| 5853 |
// This prevents the "Object could not be converted to string" error |
| 5854 |
return ''; |
| 5855 |
} |
| 5856 |
|
| 5857 |
/** |
| 5858 |
* Extract text from complex ACF array structures |
| 5859 |
*/ |
| 5860 |
private function mxchat_extract_text_from_acf_array($array) { |
| 5861 |
if (!is_array($array)) { |
| 5862 |
return ''; |
| 5863 |
} |
| 5864 |
|
| 5865 |
$text_parts = array(); |
| 5866 |
|
| 5867 |
foreach ($array as $key => $value) { |
| 5868 |
if (is_string($value) && !empty(trim($value))) { |
| 5869 |
// Skip keys that are likely to be IDs or technical values |
| 5870 |
if (!is_numeric($value) || strlen($value) > 10) { |
| 5871 |
$text_parts[] = trim($value); |
| 5872 |
} |
| 5873 |
} elseif ($value instanceof WP_Post) { |
| 5874 |
// Handle WP_Post objects |
| 5875 |
$text_parts[] = $value->post_title ?: ''; |
| 5876 |
} elseif (is_array($value)) { |
| 5877 |
if (isset($value['post_title'])) { |
| 5878 |
$text_parts[] = $value['post_title']; |
| 5879 |
} elseif (isset($value['name'])) { |
| 5880 |
$text_parts[] = $value['name']; |
| 5881 |
} elseif (isset($value['label'])) { |
| 5882 |
$text_parts[] = $value['label']; |
| 5883 |
} |
| 5884 |
} elseif (is_object($value)) { |
| 5885 |
// Handle other objects safely |
| 5886 |
if (isset($value->post_title)) { |
| 5887 |
$text_parts[] = $value->post_title; |
| 5888 |
} elseif (isset($value->name)) { |
| 5889 |
$text_parts[] = $value->name; |
| 5890 |
} elseif (isset($value->display_name)) { |
| 5891 |
$text_parts[] = $value->display_name; |
| 5892 |
} |
| 5893 |
} |
| 5894 |
} |
| 5895 |
|
| 5896 |
return implode(', ', array_filter($text_parts)); |
| 5897 |
} |
| 5898 |
|
| 5899 |
/** |
| 5900 |
* Walk an ACF field value tree and collect attachment IDs for any value that |
| 5901 |
* resolves to a PDF in the WordPress media library. Handles the three shapes |
| 5902 |
* ACF returns for File/Image/URL fields (array with ID+url, integer attachment ID, |
| 5903 |
* plain URL string), and recurses through repeater/group/flexible content. |
| 5904 |
* |
| 5905 |
* @param mixed $value The ACF field value (any depth) |
| 5906 |
* @param array $out Accumulator (passed by reference) for attachment IDs |
| 5907 |
* @param int $depth Recursion guard |
| 5908 |
*/ |
| 5909 |
private function mxchat_collect_pdf_attachment_ids_from_acf_value($value, &$out, $depth = 0) { |
| 5910 |
if ($depth > 6) { |
| 5911 |
return; // prevent runaway recursion on circular/very-deep structures |
| 5912 |
} |
| 5913 |
|
| 5914 |
if (empty($value)) { |
| 5915 |
return; |
| 5916 |
} |
| 5917 |
|
| 5918 |
// Array shapes: ACF File/Image return value=array; repeaters/groups are arrays of arrays |
| 5919 |
if (is_array($value)) { |
| 5920 |
// Direct File/Image-style array (has 'url' and usually 'ID' + 'mime_type') |
| 5921 |
$looks_like_attachment = isset($value['url']) || isset($value['ID']) || isset($value['id']); |
| 5922 |
if ($looks_like_attachment) { |
| 5923 |
$att_id = 0; |
| 5924 |
if (!empty($value['ID']) && is_numeric($value['ID'])) { |
| 5925 |
$att_id = (int) $value['ID']; |
| 5926 |
} elseif (!empty($value['id']) && is_numeric($value['id'])) { |
| 5927 |
$att_id = (int) $value['id']; |
| 5928 |
} elseif (!empty($value['url']) && is_string($value['url'])) { |
| 5929 |
$att_id = (int) attachment_url_to_postid($value['url']); |
| 5930 |
} |
| 5931 |
|
| 5932 |
$is_pdf = false; |
| 5933 |
if (!empty($value['mime_type']) && $value['mime_type'] === 'application/pdf') { |
| 5934 |
$is_pdf = true; |
| 5935 |
} elseif (!empty($value['subtype']) && strtolower((string) $value['subtype']) === 'pdf') { |
| 5936 |
$is_pdf = true; |
| 5937 |
} elseif (!empty($value['url']) && is_string($value['url']) && $this->mxchat_url_looks_like_pdf($value['url'])) { |
| 5938 |
$is_pdf = true; |
| 5939 |
} elseif ($att_id && get_post_mime_type($att_id) === 'application/pdf') { |
| 5940 |
$is_pdf = true; |
| 5941 |
} |
| 5942 |
|
| 5943 |
if ($is_pdf && $att_id && get_post_mime_type($att_id) === 'application/pdf') { |
| 5944 |
$out[] = $att_id; |
| 5945 |
} |
| 5946 |
// An array node that represents one attachment doesn't contain other |
| 5947 |
// attachments inside it — done with this branch. |
| 5948 |
return; |
| 5949 |
} |
| 5950 |
|
| 5951 |
// Recurse: repeater rows, flexible-content layouts, groups, etc. |
| 5952 |
foreach ($value as $sub) { |
| 5953 |
$this->mxchat_collect_pdf_attachment_ids_from_acf_value($sub, $out, $depth + 1); |
| 5954 |
} |
| 5955 |
return; |
| 5956 |
} |
| 5957 |
|
| 5958 |
// Plain numeric attachment ID (ACF File field set to "Return: ID") |
| 5959 |
if (is_numeric($value)) { |
| 5960 |
$att_id = (int) $value; |
| 5961 |
if ($att_id > 0 && get_post_mime_type($att_id) === 'application/pdf') { |
| 5962 |
$out[] = $att_id; |
| 5963 |
} |
| 5964 |
return; |
| 5965 |
} |
| 5966 |
|
| 5967 |
// Plain string — URL pointing at a PDF (ACF File field set to "Return: URL", or a custom URL/text field) |
| 5968 |
if (is_string($value)) { |
| 5969 |
$trimmed = trim($value); |
| 5970 |
if ($trimmed !== '' && $this->mxchat_url_looks_like_pdf($trimmed)) { |
| 5971 |
$att_id = (int) attachment_url_to_postid($trimmed); |
| 5972 |
if ($att_id > 0 && get_post_mime_type($att_id) === 'application/pdf') { |
| 5973 |
$out[] = $att_id; |
| 5974 |
} |
| 5975 |
} |
| 5976 |
return; |
| 5977 |
} |
| 5978 |
} |
| 5979 |
|
| 5980 |
/** |
| 5981 |
* Heuristic: does this URL/string look like a PDF reference? |
| 5982 |
* Tolerates query strings and fragments (#page=2). |
| 5983 |
*/ |
| 5984 |
private function mxchat_url_looks_like_pdf($url) { |
| 5985 |
if (!is_string($url) || $url === '') { |
| 5986 |
return false; |
| 5987 |
} |
| 5988 |
// Strip query + fragment before checking extension |
| 5989 |
$path = preg_replace('/[?#].*$/', '', $url); |
| 5990 |
return (bool) preg_match('/\.pdf$/i', $path); |
| 5991 |
} |
| 5992 |
|
| 5993 |
/** |
| 5994 |
* Extract text from a PDF attachment by ID using the bundled Smalot parser. |
| 5995 |
* Reads the file directly from disk via get_attached_file (no HTTP fetch). |
| 5996 |
* Result is cached on the attachment as post_meta keyed by file mtime so we |
| 5997 |
* only parse the same PDF once unless the file changes on disk. |
| 5998 |
* |
| 5999 |
* @param int $attachment_id |
| 6000 |
* @return string Extracted plain text, or '' on failure. |
| 6001 |
*/ |
| 6002 |
private function mxchat_extract_pdf_text_by_attachment_id($attachment_id) { |
| 6003 |
$attachment_id = (int) $attachment_id; |
| 6004 |
if ($attachment_id <= 0) { |
| 6005 |
return ''; |
| 6006 |
} |
| 6007 |
if (get_post_mime_type($attachment_id) !== 'application/pdf') { |
| 6008 |
return ''; |
| 6009 |
} |
| 6010 |
|
| 6011 |
$pdf_path = get_attached_file($attachment_id); |
| 6012 |
if (empty($pdf_path) || !file_exists($pdf_path) || !is_readable($pdf_path)) { |
| 6013 |
return ''; |
| 6014 |
} |
| 6015 |
|
| 6016 |
// Raw-file size cap. Parsing very large PDFs can OOM the request; skip with a log entry |
| 6017 |
// and let the rest of the ACF content land in the KB. Filterable for users who need it bigger. |
| 6018 |
$default_max_bytes = 25 * 1024 * 1024; |
| 6019 |
$max_bytes = (int) apply_filters('mxchat_acf_pdf_max_bytes', $default_max_bytes, $attachment_id, $pdf_path); |
| 6020 |
if ($max_bytes > 0) { |
| 6021 |
$file_size = @filesize($pdf_path); |
| 6022 |
if ($file_size !== false && $file_size > $max_bytes) { |
| 6023 |
error_log(sprintf( |
| 6024 |
'[mxchat] ACF PDF skipped (over size cap): attachment %d "%s" %d bytes > cap %d', |
| 6025 |
$attachment_id, |
| 6026 |
basename($pdf_path), |
| 6027 |
$file_size, |
| 6028 |
$max_bytes |
| 6029 |
)); |
| 6030 |
return ''; |
| 6031 |
} |
| 6032 |
} |
| 6033 |
|
| 6034 |
$mtime = @filemtime($pdf_path); |
| 6035 |
$cache_meta_key = '_mxchat_acf_pdf_text_v1'; |
| 6036 |
$cached = get_post_meta($attachment_id, $cache_meta_key, true); |
| 6037 |
if (is_array($cached) && isset($cached['mtime'], $cached['text']) && (int) $cached['mtime'] === (int) $mtime) { |
| 6038 |
return (string) $cached['text']; |
| 6039 |
} |
| 6040 |
|
| 6041 |
$text = ''; |
| 6042 |
try { |
| 6043 |
if (function_exists('mxchat_load_pdf_parser')) { |
| 6044 |
mxchat_load_pdf_parser(); |
| 6045 |
} |
| 6046 |
if (!class_exists('\\Smalot\\PdfParser\\Parser')) { |
| 6047 |
return ''; |
| 6048 |
} |
| 6049 |
$parser = new \Smalot\PdfParser\Parser(); |
| 6050 |
$pdf = $parser->parseFile($pdf_path); |
| 6051 |
$pages = $pdf->getPages(); |
| 6052 |
$page_texts = array(); |
| 6053 |
$acf_page_num = 0; |
| 6054 |
foreach ($pages as $page) { |
| 6055 |
$acf_page_num++; |
| 6056 |
$page_text = ''; |
| 6057 |
try { |
| 6058 |
$page_text = $page->getText(); |
| 6059 |
} catch (\Exception $e) { |
| 6060 |
$page_text = ''; |
| 6061 |
} |
| 6062 |
if (!empty($page_text)) { |
| 6063 |
$page_text = MxChat_Utils::normalize_pdf_rtl($page_text, 'acf_pdf attachment ' . $attachment_id . ' page ' . $acf_page_num); |
| 6064 |
$page_texts[] = $page_text; |
| 6065 |
} |
| 6066 |
} |
| 6067 |
$text = trim(implode("\n\n", $page_texts)); |
| 6068 |
} catch (\Exception $e) { |
| 6069 |
error_log('[mxchat] ACF PDF extraction failed for attachment ' . $attachment_id . ': ' . $e->getMessage()); |
| 6070 |
return ''; |
| 6071 |
} catch (\Throwable $e) { |
| 6072 |
error_log('[mxchat] ACF PDF extraction error for attachment ' . $attachment_id . ': ' . $e->getMessage()); |
| 6073 |
return ''; |
| 6074 |
} |
| 6075 |
|
| 6076 |
// Cap per-PDF text to avoid blowing up the embedding payload on enormous PDFs. |
| 6077 |
// The chunker downstream will still split this into multiple vectors. |
| 6078 |
$max_len = (int) apply_filters('mxchat_acf_pdf_text_max_length', 50000); |
| 6079 |
if ($max_len > 0 && strlen($text) > $max_len) { |
| 6080 |
$text = substr($text, 0, $max_len); |
| 6081 |
} |
| 6082 |
|
| 6083 |
update_post_meta($attachment_id, $cache_meta_key, array( |
| 6084 |
'mtime' => (int) $mtime, |
| 6085 |
'text' => $text, |
| 6086 |
)); |
| 6087 |
|
| 6088 |
return $text; |
| 6089 |
} |
| 6090 |
|
| 6091 |
/** |
| 6092 |
* Handle ACF save - fires after ACF fields are saved |
| 6093 |
* This ensures ACF field data is available when syncing to knowledge base |
| 6094 |
*/ |
| 6095 |
public function mxchat_handle_acf_save($post_id) { |
| 6096 |
// Skip if not a valid post |
| 6097 |
if (!$post_id || $post_id === 'options') { |
| 6098 |
return; |
| 6099 |
} |
| 6100 |
|
| 6101 |
// Skip autosaves and revisions |
| 6102 |
if (defined('DOING_AUTOSAVE') && DOING_AUTOSAVE || wp_is_post_revision($post_id)) { |
| 6103 |
return; |
| 6104 |
} |
| 6105 |
|
| 6106 |
$post = get_post($post_id); |
| 6107 |
if (!$post) { |
| 6108 |
return; |
| 6109 |
} |
| 6110 |
|
| 6111 |
$post_type = $post->post_type; |
| 6112 |
|
| 6113 |
// Check if sync is enabled for this post type |
| 6114 |
$should_sync = false; |
| 6115 |
|
| 6116 |
if ($post_type === 'post' && get_option('mxchat_auto_sync_posts') === '1') { |
| 6117 |
$should_sync = true; |
| 6118 |
} else if ($post_type === 'page' && get_option('mxchat_auto_sync_pages') === '1') { |
| 6119 |
$should_sync = true; |
| 6120 |
} else if ($post_type === 'product' && class_exists('WooCommerce')) { |
| 6121 |
// WooCommerce products - check if WooCommerce integration is enabled |
| 6122 |
$options = get_option('mxchat_options', array()); |
| 6123 |
if (isset($options['enable_woocommerce_integration']) && |
| 6124 |
($options['enable_woocommerce_integration'] === '1' || $options['enable_woocommerce_integration'] === 'on')) { |
| 6125 |
$should_sync = true; |
| 6126 |
} |
| 6127 |
} else { |
| 6128 |
// Check custom post types |
| 6129 |
$option_name = 'mxchat_auto_sync_' . $post_type; |
| 6130 |
if (get_option($option_name) === '1') { |
| 6131 |
$should_sync = true; |
| 6132 |
} |
| 6133 |
} |
| 6134 |
|
| 6135 |
if (!$should_sync) { |
| 6136 |
return; |
| 6137 |
} |
| 6138 |
|
| 6139 |
// Only process published posts |
| 6140 |
if ($post->post_status !== 'publish') { |
| 6141 |
return; |
| 6142 |
} |
| 6143 |
|
| 6144 |
// Check if this post has any ACF fields - if not, no need to re-sync |
| 6145 |
$acf_fields = $this->mxchat_get_acf_fields_for_post($post_id); |
| 6146 |
if (empty($acf_fields)) { |
| 6147 |
return; |
| 6148 |
} |
| 6149 |
|
| 6150 |
// Use a transient to prevent duplicate processing (post_updated may have already run) |
| 6151 |
$transient_key = 'mxchat_acf_synced_' . $post_id; |
| 6152 |
if (get_transient($transient_key)) { |
| 6153 |
return; |
| 6154 |
} |
| 6155 |
set_transient($transient_key, true, 60); // Prevent re-processing for 60 seconds |
| 6156 |
|
| 6157 |
// Re-run the sync with ACF data now available |
| 6158 |
// We pass $update=true since this is effectively an update with ACF data |
| 6159 |
$this->mxchat_handle_post_update($post_id, $post, true); |
| 6160 |
} |
| 6161 |
|
| 6162 |
public function mxchat_handle_post_update($post_id, $post, $update) { |
| 6163 |
// The in-flight-update marker has done its job the moment post_updated runs; drop it |
| 6164 |
// before any early return so it can never outlive its own save (a failed $wpdb->update |
| 6165 |
// inside wp_insert_post returns after pre_post_update but before the transition). |
| 6166 |
unset($this->pending_post_update[$post_id]); |
| 6167 |
|
| 6168 |
// Basic validation checks |
| 6169 |
if (defined('DOING_AUTOSAVE') && DOING_AUTOSAVE || wp_is_post_revision($post_id)) { |
| 6170 |
return; |
| 6171 |
} |
| 6172 |
|
| 6173 |
$post_type = $post->post_type; |
| 6174 |
|
| 6175 |
// Check if sync is enabled for this post type |
| 6176 |
$should_sync = false; |
| 6177 |
|
| 6178 |
// Check built-in post types first |
| 6179 |
if ($post_type === 'post' && get_option('mxchat_auto_sync_posts') === '1') { |
| 6180 |
$should_sync = true; |
| 6181 |
} else if ($post_type === 'page' && get_option('mxchat_auto_sync_pages') === '1') { |
| 6182 |
$should_sync = true; |
| 6183 |
} else { |
| 6184 |
// Check custom post types |
| 6185 |
$option_name = 'mxchat_auto_sync_' . $post_type; |
| 6186 |
if (get_option($option_name) === '1') { |
| 6187 |
$should_sync = true; |
| 6188 |
} |
| 6189 |
} |
| 6190 |
|
| 6191 |
if (!$should_sync) { |
| 6192 |
return; |
| 6193 |
} |
| 6194 |
|
| 6195 |
// Check if we have stored the previous status and URL in our transients |
| 6196 |
$previous_status_key = 'mxchat_prev_status_' . $post_id; |
| 6197 |
$previous_status = get_transient($previous_status_key); |
| 6198 |
|
| 6199 |
$previous_url_key = 'mxchat_prev_url_' . $post_id; |
| 6200 |
$previous_url = get_transient($previous_url_key); |
| 6201 |
|
| 6202 |
// If the post was previously published but is now not published, remove from knowledge base |
| 6203 |
if ($previous_status === 'publish' && $post->post_status !== 'publish') { |
| 6204 |
// Use the stored URL from when it was published, or fall back to current permalink |
| 6205 |
$source_url = $previous_url ?: get_permalink($post_id); |
| 6206 |
|
| 6207 |
// mxchat_handle_status_transition already deleted for this post earlier in this |
| 6208 |
// request (it fires first inside wp_insert_post); skip the redundant round-trip. |
| 6209 |
if ($source_url && empty($this->transition_deleted_posts[$post_id])) { |
| 6210 |
// Chunk-aware deletion (routes to Pinecone or WP DB and removes base + all chunks) |
| 6211 |
MxChat_Utils::delete_chunks_for_url($source_url, 'default'); |
| 6212 |
} |
| 6213 |
|
| 6214 |
// Clean up the transients and exit early |
| 6215 |
delete_transient($previous_status_key); |
| 6216 |
delete_transient($previous_url_key); |
| 6217 |
return; |
| 6218 |
} |
| 6219 |
|
| 6220 |
// Slug/permalink rename while still published: delete the old vectors before upserting new ones. |
| 6221 |
// Without this, md5(old_url) vectors (base + chunks) would be orphaned under the stale URL. |
| 6222 |
if ($post->post_status === 'publish' && !empty($previous_url)) { |
| 6223 |
$current_url = get_permalink($post_id); |
| 6224 |
if ($current_url && $current_url !== $previous_url) { |
| 6225 |
MxChat_Utils::delete_chunks_for_url($previous_url, 'default'); |
| 6226 |
} |
| 6227 |
} |
| 6228 |
|
| 6229 |
// Store the current status for next time (if this is an update) |
| 6230 |
if ($update) { |
| 6231 |
set_transient($previous_status_key, $post->post_status, DAY_IN_SECONDS); |
| 6232 |
|
| 6233 |
// If the post is currently published, also store its URL |
| 6234 |
if ($post->post_status === 'publish') { |
| 6235 |
$current_url = get_permalink($post_id); |
| 6236 |
set_transient($previous_url_key, $current_url, DAY_IN_SECONDS); |
| 6237 |
} |
| 6238 |
} |
| 6239 |
|
| 6240 |
// Only process currently published content for adding/updating. |
| 6241 |
// transition_indexed_posts: mxchat_handle_status_transition's arrival edge may have |
| 6242 |
// already indexed this post earlier in this request (editor publishes fire |
| 6243 |
// transition_post_status first, then post_updated) — skip the duplicate embed. |
| 6244 |
// Consume-once: the flag is cleared when honoured, so a LATER save of the same |
| 6245 |
// post in one long-running process (WP-CLI scripts, importers) re-indexes normally. |
| 6246 |
if ($post->post_status === 'publish') { |
| 6247 |
if (!empty($this->transition_indexed_posts[$post_id])) { |
| 6248 |
unset($this->transition_indexed_posts[$post_id]); |
| 6249 |
} else { |
| 6250 |
$this->mxchat_index_published_post($post_id, $post); |
| 6251 |
} |
| 6252 |
} |
| 6253 |
|
| 6254 |
// Clean up the stored previous status if not used above |
| 6255 |
if ($previous_status !== 'publish' || $post->post_status === 'publish') { |
| 6256 |
delete_transient($previous_status_key); |
| 6257 |
delete_transient($previous_url_key); |
| 6258 |
} |
| 6259 |
} |
| 6260 |
|
| 6261 |
/** |
| 6262 |
* Index a published post into the knowledge base: preprocessing filter, content |
| 6263 |
* assembly (title/excerpt/body), WooCommerce product enrichment, job_listing meta, |
| 6264 |
* ACF fields (+ optional PDF extraction), whitelisted custom meta, embedding and |
| 6265 |
* upsert, then tag-based role restriction. |
| 6266 |
* |
| 6267 |
* Shared by the post_updated auto-sync path (mxchat_handle_post_update) and the |
| 6268 |
* transition_post_status arrival edge (mxchat_handle_status_transition), so |
| 6269 |
* scheduled publishes (wp_publish_post) and direct status=publish inserts index |
| 6270 |
* identically to editor saves (plan 3055e1). Pure extraction of the former |
| 6271 |
* publish branch — body indentation retained to keep the diff reviewable. |
| 6272 |
*/ |
| 6273 |
private function mxchat_index_published_post($post_id, $post) { |
| 6274 |
$post_type = $post->post_type; |
| 6275 |
|
| 6276 |
// WooCommerce products are owned by the WC-object assembler (plan a3d60c): |
| 6277 |
// whenever WooCommerce is active AND the integration is enabled, every |
| 6278 |
// product save also fires save_post_product, which queues |
| 6279 |
// mxchat_store_product_embedding on shutdown — and that writer runs LAST, |
| 6280 |
// overwriting the same md5(permalink) row this path would write. Assembling |
| 6281 |
// and embedding the product here was pure duplicate spend (measured: two |
| 6282 |
// embedding calls per product save, second one wins). Skip ONLY under the |
| 6283 |
// exact conditions the shutdown writer runs — same option read as its own |
| 6284 |
// gate — because with the integration off (or WooCommerce inactive) this |
| 6285 |
// path is the sole product indexer and must keep working. |
| 6286 |
if ($post_type === 'product' |
| 6287 |
&& class_exists('WooCommerce') |
| 6288 |
&& isset($this->options['enable_woocommerce_integration']) |
| 6289 |
&& in_array($this->options['enable_woocommerce_integration'], ['1', 'on'])) { |
| 6290 |
return; |
| 6291 |
} |
| 6292 |
|
| 6293 |
// Get the source URL |
| 6294 |
$source_url = get_permalink($post_id); |
| 6295 |
|
| 6296 |
// A draft published programmatically (wp_publish_post) can reach this |
| 6297 |
// point with an EMPTY post_name — wp_insert_post skips slug generation |
| 6298 |
// for draft/pending — and get_permalink() then resolves to the bare |
| 6299 |
// site root. A knowledge row keyed to the homepage cites the wrong URL |
| 6300 |
// and answers homepage questions with this post's body, so refuse to |
| 6301 |
// write it; the post indexes correctly on its next save, once the slug |
| 6302 |
// exists. The empty-post_name test is what keeps a legitimate static |
| 6303 |
// front page (which has a slug but a root permalink) indexable. |
| 6304 |
// (Plan d138c4.) |
| 6305 |
if ('' === $post->post_name |
| 6306 |
&& untrailingslashit($source_url) === untrailingslashit(home_url())) { |
| 6307 |
return; |
| 6308 |
} |
| 6309 |
|
| 6310 |
/** |
| 6311 |
* Allow developers to modify post data before processing into the knowledge base. |
| 6312 |
* Same filter and signature as the manual bulk-import path |
| 6313 |
* (ajax_mxchat_process_selected_content), so a callback registered once covers |
| 6314 |
* every indexing route. Purely additive — zero behaviour change when unhooked. |
| 6315 |
* Auto-sync runs under the 'default' bot context, matching the rest of this |
| 6316 |
* function. |
| 6317 |
* |
| 6318 |
* @param WP_Post $post The post about to be indexed. |
| 6319 |
* @param string $bot_id Bot context ('default' on auto-sync). |
| 6320 |
*/ |
| 6321 |
$post = apply_filters('mxchat_before_process_post', $post, 'default'); |
| 6322 |
if (!($post instanceof WP_Post)) { |
| 6323 |
$post = get_post($post_id); // defend against a bad callback return |
| 6324 |
} |
| 6325 |
|
| 6326 |
// Assemble the indexable text via the shared post-kind assembler (a3d60c), |
| 6327 |
// reading from the FILTERED post object — not re-fetched by ID, which would |
| 6328 |
// discard it. Auto-sync reads content/excerpt in its historical |
| 6329 |
// get_post_field() display context, never appended product custom tabs |
| 6330 |
// (its product branch is reachable only with the WooCommerce integration |
| 6331 |
// off), and gates ACF→PDF extraction behind its own opt-in option — |
| 6332 |
// default OFF, because re-parsing every ACF PDF on every editor save is |
| 6333 |
// expensive and most sites don't want it (the 25 MB size cap lives in the |
| 6334 |
// shared extractor either way). |
| 6335 |
$prepared = $this->mxchat_prepare_post_content_for_indexing($post_id, $post, array( |
| 6336 |
'read_display' => true, |
| 6337 |
'extract_acf_pdfs' => get_option('mxchat_auto_sync_acf_pdfs', '0') === '1', |
| 6338 |
'include_product_tabs' => false, |
| 6339 |
)); |
| 6340 |
$final_content = $prepared['content']; |
| 6341 |
|
| 6342 |
// Embedding decision — custom-provider-aware. Gating on a cloud API key |
| 6343 |
// here silently killed auto-sync on keyless custom-embeddings sites, |
| 6344 |
// because generate_embedding() routes custom FIRST and never needs the |
| 6345 |
// key (plan cbd5fd). Silent-return shape preserved. |
| 6346 |
$preflight = MxChat_Utils::embedding_preflight(get_option('mxchat_options')); |
| 6347 |
if (!$preflight['ok']) { |
| 6348 |
return; |
| 6349 |
} |
| 6350 |
$api_key = $preflight['api_key']; |
| 6351 |
|
| 6352 |
// Use the centralized utility function for storage |
| 6353 |
$result = MxChat_Utils::submit_content_to_db( |
| 6354 |
$final_content, |
| 6355 |
$source_url, |
| 6356 |
$api_key, |
| 6357 |
md5($source_url) // Vector ID for Pinecone |
| 6358 |
); |
| 6359 |
|
| 6360 |
// After successful storage, apply role restriction based on tags |
| 6361 |
if (!is_wp_error($result)) { |
| 6362 |
$this->apply_role_restriction_to_post($post_id, $source_url); |
| 6363 |
} |
| 6364 |
} |
| 6365 |
|
| 6366 |
/** |
| 6367 |
* Shared post-fields content assembler (plan a3d60c) — the ONE body behind both |
| 6368 |
* post-kind ingestion paths: manual bulk import (ajax_mxchat_process_selected_content) |
| 6369 |
* and auto-sync (mxchat_index_published_post). Behavior-preserving extraction; the |
| 6370 |
* measured per-caller differences ride $args instead of living as drifting copies: |
| 6371 |
* |
| 6372 |
* 'read_display' bool Auto-sync historically reads content/excerpt via |
| 6373 |
* get_post_field() in its default 'display' context |
| 6374 |
* (the post_content / post_excerpt display filters |
| 6375 |
* fire); bulk import reads the raw properties. Inert |
| 6376 |
* on a stock install — preserved per-path, not converged. |
| 6377 |
* 'extract_acf_pdfs' bool Each caller passes its OWN option (bulk: |
| 6378 |
* mxchat_acf_pdf_extraction; auto-sync: |
| 6379 |
* mxchat_auto_sync_acf_pdfs) — the two-option design |
| 6380 |
* is deliberate (plan 11720c). Gates BOTH the PDF-id |
| 6381 |
* collection walk and the extraction loop; the ids are |
| 6382 |
* only ever read inside the extraction branch, so |
| 6383 |
* gating collection is output-identical on every install. |
| 6384 |
* 'include_product_tabs' bool The bulk path has always appended yikes_woo custom |
| 6385 |
* tabs to product content; the auto-sync product branch |
| 6386 |
* (reachable only with the WooCommerce integration off) |
| 6387 |
* never did. Preserved per-path — converging it would be |
| 6388 |
* a behavior change, recorded on the plan instead. |
| 6389 |
* |
| 6390 |
* Returns array: 'content' (the assembled indexable text), 'acf_fields_found' and |
| 6391 |
* 'pdf_extracted_count' (the bulk path reports both in its AJAX response). |
| 6392 |
*/ |
| 6393 |
private function mxchat_prepare_post_content_for_indexing($post_id, $post, $args) { |
| 6394 |
$read_display = !empty($args['read_display']); |
| 6395 |
$extract_acf_pdfs = !empty($args['extract_acf_pdfs']); |
| 6396 |
$include_product_tabs = !empty($args['include_product_tabs']); |
| 6397 |
|
| 6398 |
// Raw post_title, NOT get_the_title(): the_title applies wptexturize + |
| 6399 |
// convert_chars and prepends the "Protected:" / "Private:" display chrome. |
| 6400 |
// The knowledge base stores facts, not display strings. Entity decode at |
| 6401 |
// output time (single-pass, shared helper) — a stored `&` embeds worse |
| 6402 |
// than `&` and gets quoted back to visitors (d2c92e). |
| 6403 |
$content = $this->mxchat_decode_entities_for_indexing($post->post_title) . "\n\n"; |
| 6404 |
|
| 6405 |
$raw_excerpt = $read_display ? get_post_field('post_excerpt', $post) : $post->post_excerpt; |
| 6406 |
$raw_content = $read_display ? get_post_field('post_content', $post) : $post->post_content; |
| 6407 |
|
| 6408 |
// Add short description if it exists (WooCommerce products use post_excerpt for short description) |
| 6409 |
// Strip FIRST, then test: an excerpt that is nothing but shortcodes strips to |
| 6410 |
// empty, and testing the raw value emitted a bare "Short Description: " label |
| 6411 |
// with no value after it. trim() only in the TEST — the emitted value is |
| 6412 |
// untouched, so a populated excerpt is byte-identical to before. A |
| 6413 |
// whitespace-only excerpt is an empty excerpt and must not produce a labelled |
| 6414 |
// line with nothing after it. |
| 6415 |
$clean_excerpt = $this->strip_shortcode_tags_preserve_content($raw_excerpt); |
| 6416 |
if (trim($clean_excerpt) !== '') { |
| 6417 |
$content .= "Short Description: " . $this->mxchat_decode_entities_for_indexing(wp_strip_all_tags($clean_excerpt)) . "\n\n"; |
| 6418 |
} |
| 6419 |
|
| 6420 |
// Main content — remove shortcode tags but preserve content inside them, then |
| 6421 |
// strip tags (don't use 'the_content' filter as it may re-add shortcodes). |
| 6422 |
$clean_content = $this->strip_shortcode_tags_preserve_content($raw_content); |
| 6423 |
$content .= $this->mxchat_decode_entities_for_indexing(wp_strip_all_tags($clean_content)); |
| 6424 |
|
| 6425 |
// WooCommerce product enrichment (post-fields kind). The WC-object assembler |
| 6426 |
// (mxchat_prepare_product_content_for_indexing) owns product rows whenever the |
| 6427 |
// integration is on; this branch serves the bulk import (all configurations) |
| 6428 |
// and auto-sync with the integration off. |
| 6429 |
if (get_post_type($post_id) === 'product' && class_exists('WooCommerce')) { |
| 6430 |
$product = wc_get_product($post_id); |
| 6431 |
|
| 6432 |
if ($product) { |
| 6433 |
$content .= "\n"; |
| 6434 |
$content .= $this->mxchat_woo_product_summary_lines($product); |
| 6435 |
} |
| 6436 |
|
| 6437 |
if ($include_product_tabs) { |
| 6438 |
$content .= $this->mxchat_woo_custom_tabs_text($post_id); |
| 6439 |
} |
| 6440 |
} |
| 6441 |
|
| 6442 |
// For custom post types like job_listing, include additional fields |
| 6443 |
if (get_post_type($post_id) === 'job_listing') { |
| 6444 |
// Add job-specific meta if available |
| 6445 |
$job_location = get_post_meta($post_id, '_job_location', true); |
| 6446 |
if (!empty($job_location)) { |
| 6447 |
$content .= "\n\nLocation: " . $job_location; |
| 6448 |
} |
| 6449 |
|
| 6450 |
// Get job type terms |
| 6451 |
$job_types = get_the_terms($post_id, 'job_listing_type'); |
| 6452 |
if (!empty($job_types) && !is_wp_error($job_types)) { |
| 6453 |
$types = array(); |
| 6454 |
foreach ($job_types as $type) { |
| 6455 |
$types[] = $type->name; |
| 6456 |
} |
| 6457 |
$content .= "\n\nJob Type: " . implode(', ', $types); |
| 6458 |
} |
| 6459 |
|
| 6460 |
// Get company name if available |
| 6461 |
$company_name = get_post_meta($post_id, '_company_name', true); |
| 6462 |
if (!empty($company_name)) { |
| 6463 |
$content .= "\n\nCompany: " . $company_name; |
| 6464 |
} |
| 6465 |
} |
| 6466 |
|
| 6467 |
// ADD ACF FIELDS SUPPORT |
| 6468 |
$acf_fields = $this->mxchat_get_acf_fields_for_post($post_id); |
| 6469 |
$pdf_extracted_count = 0; |
| 6470 |
if (!empty($acf_fields)) { |
| 6471 |
$acf_content_parts = array(); |
| 6472 |
$pdf_attachment_ids = array(); |
| 6473 |
|
| 6474 |
foreach ($acf_fields as $field_name => $field_value) { |
| 6475 |
$formatted_value = $this->mxchat_format_acf_field_value($field_value, $field_name, $post_id); |
| 6476 |
|
| 6477 |
if (!empty($formatted_value)) { |
| 6478 |
// Both separators: a hyphenated ACF name should read as words. |
| 6479 |
$field_label = ucwords(str_replace(['_', '-'], ' ', $field_name)); |
| 6480 |
$acf_content_parts[] = $field_label . ": " . $formatted_value; |
| 6481 |
} |
| 6482 |
|
| 6483 |
// Walk this field's value tree for any PDF attachment references and |
| 6484 |
// queue them for extraction — only when this caller's PDF option is on. |
| 6485 |
if ($extract_acf_pdfs) { |
| 6486 |
$this->mxchat_collect_pdf_attachment_ids_from_acf_value($field_value, $pdf_attachment_ids); |
| 6487 |
} |
| 6488 |
} |
| 6489 |
|
| 6490 |
if (!empty($acf_content_parts)) { |
| 6491 |
$content .= "\n\n" . implode("\n", $acf_content_parts); |
| 6492 |
} |
| 6493 |
|
| 6494 |
// Extract text from each unique PDF found in ACF fields and append as a labeled section |
| 6495 |
if ($extract_acf_pdfs && !empty($pdf_attachment_ids)) { |
| 6496 |
$pdf_attachment_ids = array_unique(array_filter(array_map('intval', $pdf_attachment_ids))); |
| 6497 |
$pdf_sections = array(); |
| 6498 |
foreach ($pdf_attachment_ids as $att_id) { |
| 6499 |
$pdf_text = $this->mxchat_extract_pdf_text_by_attachment_id($att_id); |
| 6500 |
if (!empty($pdf_text)) { |
| 6501 |
$pdf_title = get_the_title($att_id); |
| 6502 |
$pdf_url = wp_get_attachment_url($att_id); |
| 6503 |
$header = 'PDF Attachment'; |
| 6504 |
if (!empty($pdf_title)) { |
| 6505 |
$header .= ': ' . $pdf_title; |
| 6506 |
} |
| 6507 |
if (!empty($pdf_url)) { |
| 6508 |
$header .= ' (' . $pdf_url . ')'; |
| 6509 |
} |
| 6510 |
$pdf_sections[] = $header . "\n" . $pdf_text; |
| 6511 |
$pdf_extracted_count++; |
| 6512 |
} |
| 6513 |
} |
| 6514 |
if (!empty($pdf_sections)) { |
| 6515 |
$content .= "\n\nAttached PDFs:\n" . implode("\n\n", $pdf_sections); |
| 6516 |
} |
| 6517 |
} |
| 6518 |
} |
| 6519 |
|
| 6520 |
// ADD CUSTOM POST META SUPPORT (whitelisted non-ACF meta fields) |
| 6521 |
$custom_meta = $this->mxchat_get_whitelisted_post_meta($post_id); |
| 6522 |
if (!empty($custom_meta)) { |
| 6523 |
$meta_content_parts = array(); |
| 6524 |
|
| 6525 |
foreach ($custom_meta as $meta_key => $meta_value) { |
| 6526 |
// Convert meta key to readable label |
| 6527 |
$meta_label = ucwords(str_replace(array('_', '-'), ' ', $meta_key)); |
| 6528 |
$meta_content_parts[] = $meta_label . ": " . $meta_value; |
| 6529 |
} |
| 6530 |
|
| 6531 |
if (!empty($meta_content_parts)) { |
| 6532 |
$content .= "\n\n" . implode("\n", $meta_content_parts); |
| 6533 |
} |
| 6534 |
} |
| 6535 |
|
| 6536 |
return array( |
| 6537 |
'content' => $content, |
| 6538 |
'acf_fields_found' => count($acf_fields), |
| 6539 |
'pdf_extracted_count' => $pdf_extracted_count, |
| 6540 |
); |
| 6541 |
} |
| 6542 |
|
| 6543 |
/** |
| 6544 |
* Shared WC-object product assembler (plan a3d60c) — the ONE body behind the two |
| 6545 |
* WooCommerce-object ingestion paths: the auto-sync product writer |
| 6546 |
* (mxchat_store_product_embedding) and the URL/sitemap product import |
| 6547 |
* (mxchat_extract_woocommerce_product_content). Assembles from the WC_Product, |
| 6548 |
* the authoritative source for product rows (scope decision on the plan). |
| 6549 |
*/ |
| 6550 |
private function mxchat_prepare_product_content_for_indexing($product) { |
| 6551 |
$title = $product->get_name(); |
| 6552 |
$description = $product->get_description(); |
| 6553 |
$short_description = $product->get_short_description(); |
| 6554 |
|
| 6555 |
// Format content consistently |
| 6556 |
$content = $title . "\n\n"; |
| 6557 |
|
| 6558 |
if (!empty($short_description)) { |
| 6559 |
$content .= "Short Description: " . wp_strip_all_tags($short_description) . "\n\n"; |
| 6560 |
} |
| 6561 |
|
| 6562 |
if (!empty($description)) { |
| 6563 |
$content .= wp_strip_all_tags($description) . "\n\n"; |
| 6564 |
} |
| 6565 |
|
| 6566 |
$content .= $this->mxchat_woo_product_summary_lines($product); |
| 6567 |
$content .= $this->mxchat_woo_custom_tabs_text($product->get_id()); |
| 6568 |
|
| 6569 |
return $content; |
| 6570 |
} |
| 6571 |
|
| 6572 |
/** |
| 6573 |
* Pricing + SKU + categories lines for a product — shared by both assembler kinds |
| 6574 |
* (the post-fields product enrichment and the WC-object assembler). |
| 6575 |
*/ |
| 6576 |
private function mxchat_woo_product_summary_lines($product) { |
| 6577 |
// Add pricing information (base-currency pinned — see mxchat_product_price_lines) |
| 6578 |
$lines = $this->mxchat_product_price_lines($product); |
| 6579 |
|
| 6580 |
$sku = $product->get_sku(); |
| 6581 |
if (!empty($sku)) { |
| 6582 |
$lines .= "SKU: " . $sku . "\n"; |
| 6583 |
} |
| 6584 |
|
| 6585 |
// Get product categories |
| 6586 |
$categories = wp_get_post_terms($product->get_id(), 'product_cat', array('fields' => 'names')); |
| 6587 |
if (!empty($categories) && !is_wp_error($categories)) { |
| 6588 |
$lines .= "Categories: " . implode(', ', $categories) . "\n"; |
| 6589 |
} |
| 6590 |
|
| 6591 |
return $lines; |
| 6592 |
} |
| 6593 |
|
| 6594 |
/** |
| 6595 |
* Custom Product Tabs text (supports "Custom Product Tabs for WooCommerce" by |
| 6596 |
* Code Parrots) — direct tabs plus applied reusable/saved tabs. The ONE copy of |
| 6597 |
* the yikes_woo logic; three sites carried byte-identical clones before a3d60c. |
| 6598 |
*/ |
| 6599 |
private function mxchat_woo_custom_tabs_text($product_id) { |
| 6600 |
$text = ''; |
| 6601 |
|
| 6602 |
$custom_tabs = get_post_meta($product_id, 'yikes_woo_products_tabs', true); |
| 6603 |
if (!empty($custom_tabs) && is_array($custom_tabs)) { |
| 6604 |
foreach ($custom_tabs as $tab) { |
| 6605 |
$tab_title = isset($tab['title']) ? $tab['title'] : (isset($tab['tab_title']) ? $tab['tab_title'] : ''); |
| 6606 |
$tab_content = isset($tab['content']) ? $tab['content'] : ''; |
| 6607 |
|
| 6608 |
if (!empty($tab_title) && !empty($tab_content)) { |
| 6609 |
$text .= "\n" . $tab_title . ": " . wp_strip_all_tags($tab_content) . "\n"; |
| 6610 |
} |
| 6611 |
} |
| 6612 |
} |
| 6613 |
|
| 6614 |
// Also check for reusable/saved tabs applied to this product |
| 6615 |
$applied_saved_tabs = get_post_meta($product_id, 'yikes_woo_reusable_products_tabs_applied', true); |
| 6616 |
if (!empty($applied_saved_tabs) && is_array($applied_saved_tabs)) { |
| 6617 |
$saved_tabs = get_option('yikes_woo_reusable_products_tabs', array()); |
| 6618 |
if (!empty($saved_tabs) && is_array($saved_tabs)) { |
| 6619 |
foreach ($applied_saved_tabs as $saved_tab_id) { |
| 6620 |
if (isset($saved_tabs[$saved_tab_id])) { |
| 6621 |
$tab = $saved_tabs[$saved_tab_id]; |
| 6622 |
$tab_title = isset($tab['title']) ? $tab['title'] : (isset($tab['tab_title']) ? $tab['tab_title'] : ''); |
| 6623 |
$tab_content = isset($tab['content']) ? $tab['content'] : ''; |
| 6624 |
|
| 6625 |
if (!empty($tab_title) && !empty($tab_content)) { |
| 6626 |
$text .= "\n" . $tab_title . ": " . wp_strip_all_tags($tab_content) . "\n"; |
| 6627 |
} |
| 6628 |
} |
| 6629 |
} |
| 6630 |
} |
| 6631 |
} |
| 6632 |
|
| 6633 |
return $text; |
| 6634 |
} |
| 6635 |
|
| 6636 |
/** |
| 6637 |
* Store the post status and URL before update to detect status transitions |
| 6638 |
* This runs before the post is actually updated in the database |
| 6639 |
*/ |
| 6640 |
public function mxchat_store_pre_update_status($post_id, $data) { |
| 6641 |
// Core is inside wp_insert_post's update branch, so a post_updated WILL fire later |
| 6642 |
// this request and can consume the arrival-edge guard (plan a664f3). |
| 6643 |
$this->pending_post_update[$post_id] = true; |
| 6644 |
|
| 6645 |
// Get the current post from database (before update) |
| 6646 |
$current_post = get_post($post_id); |
| 6647 |
|
| 6648 |
if ($current_post) { |
| 6649 |
// Store the current status temporarily |
| 6650 |
$status_key = 'mxchat_prev_status_' . $post_id; |
| 6651 |
set_transient($status_key, $current_post->post_status, HOUR_IN_SECONDS); |
| 6652 |
|
| 6653 |
// If the post is currently published, also store its URL |
| 6654 |
if ($current_post->post_status === 'publish') { |
| 6655 |
$url_key = 'mxchat_prev_url_' . $post_id; |
| 6656 |
$current_url = get_permalink($post_id); |
| 6657 |
set_transient($url_key, $current_url, HOUR_IN_SECONDS); |
| 6658 |
} |
| 6659 |
} |
| 6660 |
} |
| 6661 |
|
| 6662 |
/** |
| 6663 |
* Whether auto-sync is enabled for a post type (mirrors the checks used by the |
| 6664 |
* update/delete handlers; kept as one helper so new call sites cannot drift). |
| 6665 |
*/ |
| 6666 |
private function mxchat_is_auto_sync_enabled($post_type) { |
| 6667 |
if ($post_type === 'post') { |
| 6668 |
return get_option('mxchat_auto_sync_posts') === '1'; |
| 6669 |
} |
| 6670 |
if ($post_type === 'page') { |
| 6671 |
return get_option('mxchat_auto_sync_pages') === '1'; |
| 6672 |
} |
| 6673 |
return get_option('mxchat_auto_sync_' . $post_type) === '1'; |
| 6674 |
} |
| 6675 |
|
| 6676 |
/** |
| 6677 |
* Remove a post's vectors the moment it leaves 'publish', using the authoritative |
| 6678 |
* old status core passes to transition_post_status — no transient involved (plan 816fb1). |
| 6679 |
* |
| 6680 |
* Covers status changes that never route through wp_update_post (scheduled-expiry |
| 6681 |
* plugins and others that flip post_status directly and call wp_transition_post_status), |
| 6682 |
* where neither pre_post_update nor post_updated fires and the old detection missed. |
| 6683 |
*/ |
| 6684 |
public function mxchat_handle_status_transition($new_status, $old_status, $post) { |
| 6685 |
if (!($post instanceof WP_Post) || wp_is_post_revision($post->ID)) { |
| 6686 |
return; |
| 6687 |
} |
| 6688 |
|
| 6689 |
// Arrival edge (plan 3055e1): a post BECOMING published is indexed here, because |
| 6690 |
// wp_publish_post() — the path scheduled posts take via check_and_publish_future_post — |
| 6691 |
// and direct wp_insert_post(status=publish) creates never fire post_updated, so the |
| 6692 |
// auto-sync ADD path alone misses them. Editor publishes also pass through here; |
| 6693 |
// the transition_indexed_posts guard keeps mxchat_handle_post_update from embedding |
| 6694 |
// a second time in the same request. |
| 6695 |
if ($new_status === 'publish' && $old_status !== 'publish') { |
| 6696 |
if ($this->mxchat_is_auto_sync_enabled($post->post_type)) { |
| 6697 |
$this->mxchat_index_published_post($post->ID, $post); |
| 6698 |
|
| 6699 |
// Arm the double-fire guard ONLY when a post_updated is actually coming to |
| 6700 |
// consume it (plan a664f3). Two publish paths never fire post_updated at all: |
| 6701 |
// a direct wp_insert_post(status=publish) create, and wp_publish_post() — the |
| 6702 |
// call check_and_publish_future_post() makes for scheduled posts. Arming the |
| 6703 |
// guard unconditionally left it set with nothing to consume it, so the NEXT |
| 6704 |
// update of that post was swallowed entirely: zero embed calls, no knowledge |
| 6705 |
// -base row, silently. Consume-once on this side too, so a guard can never |
| 6706 |
// outlive the single save it was armed for. |
| 6707 |
if (!empty($this->pending_post_update[$post->ID])) { |
| 6708 |
unset($this->pending_post_update[$post->ID]); |
| 6709 |
$this->transition_indexed_posts[$post->ID] = true; |
| 6710 |
} |
| 6711 |
} |
| 6712 |
return; |
| 6713 |
} |
| 6714 |
|
| 6715 |
// Only the publish -> not-publish edge matters here. |
| 6716 |
if ($old_status !== 'publish' || $new_status === 'publish') { |
| 6717 |
return; |
| 6718 |
} |
| 6719 |
// Trash is handled by mxchat_handle_post_delete (wp_trash_post) with pre-trash URL |
| 6720 |
// resolution; skip to avoid a second network round-trip per trash. |
| 6721 |
if ($new_status === 'trash') { |
| 6722 |
return; |
| 6723 |
} |
| 6724 |
if (!$this->mxchat_is_auto_sync_enabled($post->post_type)) { |
| 6725 |
return; |
| 6726 |
} |
| 6727 |
|
| 6728 |
$urls = array(); |
| 6729 |
|
| 6730 |
// The DB may already hold the new status when this fires, so get_permalink() on the |
| 6731 |
// live post could build a draft-style URL whose md5 misses the stored vector IDs. |
| 6732 |
// Reconstruct the published permalink from a clone instead. |
| 6733 |
$published_clone = clone $post; |
| 6734 |
$published_clone->post_status = 'publish'; |
| 6735 |
$published_url = get_permalink($published_clone); |
| 6736 |
if ($published_url) { |
| 6737 |
$urls[] = $published_url; |
| 6738 |
} |
| 6739 |
|
| 6740 |
// Honour the pre-update capture when present (covers a slug change in the same save). |
| 6741 |
$previous_url = get_transient('mxchat_prev_url_' . $post->ID); |
| 6742 |
if (!empty($previous_url)) { |
| 6743 |
$urls[] = $previous_url; |
| 6744 |
} |
| 6745 |
|
| 6746 |
foreach (array_unique($urls) as $url) { |
| 6747 |
MxChat_Utils::delete_chunks_for_url($url, 'default'); |
| 6748 |
} |
| 6749 |
|
| 6750 |
if (!empty($urls)) { |
| 6751 |
$this->transition_deleted_posts[$post->ID] = true; |
| 6752 |
} |
| 6753 |
} |
| 6754 |
|
| 6755 |
/** |
| 6756 |
* WP-CLI: remove knowledge-base entries left behind by posts that were unpublished, |
| 6757 |
* trashed, or made private before the transition_post_status handler existed. |
| 6758 |
* |
| 6759 |
* Walks every auto-synced post type's non-published posts, reconstructs each one's |
| 6760 |
* published-era permalink, and deletes its vectors (routes to Pinecone or the WP table). |
| 6761 |
* Deletion is idempotent, so never-indexed posts are a cheap no-op. |
| 6762 |
* |
| 6763 |
* ## OPTIONS |
| 6764 |
* |
| 6765 |
* [--dry-run] |
| 6766 |
* : Report what would be removed without deleting anything. |
| 6767 |
* |
| 6768 |
* ## EXAMPLES |
| 6769 |
* |
| 6770 |
* wp mxchat prune-unpublished --dry-run |
| 6771 |
* wp mxchat prune-unpublished |
| 6772 |
*/ |
| 6773 |
public function cli_prune_unpublished($args, $assoc_args) { |
| 6774 |
global $wpdb; |
| 6775 |
$dry_run = !empty($assoc_args['dry-run']); |
| 6776 |
$table = $wpdb->prefix . 'mxchat_system_prompt_content'; |
| 6777 |
|
| 6778 |
$candidate_types = array_merge(array('post', 'page'), array_values(get_post_types(array('_builtin' => false), 'names'))); |
| 6779 |
$synced_types = array(); |
| 6780 |
foreach ($candidate_types as $type) { |
| 6781 |
if ($this->mxchat_is_auto_sync_enabled($type)) { |
| 6782 |
$synced_types[] = $type; |
| 6783 |
} |
| 6784 |
} |
| 6785 |
if (empty($synced_types)) { |
| 6786 |
WP_CLI::success('No post types have auto-sync enabled; nothing to prune.'); |
| 6787 |
return; |
| 6788 |
} |
| 6789 |
|
| 6790 |
$scanned = 0; |
| 6791 |
$pruned = 0; |
| 6792 |
$paged = 1; |
| 6793 |
do { |
| 6794 |
$query = new WP_Query(array( |
| 6795 |
'post_type' => $synced_types, |
| 6796 |
'post_status' => array('draft', 'pending', 'private', 'future', 'trash'), |
| 6797 |
'posts_per_page' => 100, |
| 6798 |
'paged' => $paged, |
| 6799 |
'fields' => 'ids', |
| 6800 |
)); |
| 6801 |
foreach ($query->posts as $post_id) { |
| 6802 |
$post = get_post($post_id); |
| 6803 |
if (!$post) { |
| 6804 |
continue; |
| 6805 |
} |
| 6806 |
$scanned++; |
| 6807 |
|
| 6808 |
// Rebuild the permalink the post had while published: publish-status clone, |
| 6809 |
// with wp_trash_post's __trashed slug suffix stripped for trashed posts. |
| 6810 |
$clone = clone $post; |
| 6811 |
$clone->post_status = 'publish'; |
| 6812 |
if (substr($clone->post_name, -9) === '__trashed') { |
| 6813 |
$clone->post_name = substr($clone->post_name, 0, -9); |
| 6814 |
} |
| 6815 |
$url = get_permalink($clone); |
| 6816 |
if (!$url) { |
| 6817 |
continue; |
| 6818 |
} |
| 6819 |
|
| 6820 |
// Local-table row count is exact in WordPress-DB mode; in Pinecone mode it |
| 6821 |
// reads 0 but the delete below still routes to Pinecone and is idempotent. |
| 6822 |
$local_rows = (int) $wpdb->get_var($wpdb->prepare( |
| 6823 |
"SELECT COUNT(*) FROM {$table} WHERE source_url = %s", $url |
| 6824 |
)); |
| 6825 |
|
| 6826 |
if ($dry_run) { |
| 6827 |
if ($local_rows > 0) { |
| 6828 |
WP_CLI::log(sprintf('Would remove %d row(s): %s (post %d, %s)', $local_rows, $url, $post_id, $post->post_status)); |
| 6829 |
$pruned += $local_rows; |
| 6830 |
} |
| 6831 |
continue; |
| 6832 |
} |
| 6833 |
|
| 6834 |
MxChat_Utils::delete_chunks_for_url($url, 'default'); |
| 6835 |
if ($local_rows > 0) { |
| 6836 |
WP_CLI::log(sprintf('Removed %d row(s): %s (post %d, %s)', $local_rows, $url, $post_id, $post->post_status)); |
| 6837 |
$pruned += $local_rows; |
| 6838 |
} |
| 6839 |
} |
| 6840 |
$more = $paged < $query->max_num_pages; |
| 6841 |
$paged++; |
| 6842 |
} while ($more); |
| 6843 |
|
| 6844 |
WP_CLI::success(sprintf( |
| 6845 |
'%s %d local knowledge row(s) across %d non-published post(s) scanned.%s', |
| 6846 |
$dry_run ? 'Would remove' : 'Removed', |
| 6847 |
$pruned, |
| 6848 |
$scanned, |
| 6849 |
' (Pinecone-mode deletions are not counted locally.)' |
| 6850 |
)); |
| 6851 |
} |
| 6852 |
|
| 6853 |
/** |
| 6854 |
* WP-CLI: repair knowledge-base rows whose PDF text was imported in visual |
| 6855 |
* (reversed) order before the RTL normalizer existed. 32bf9e fixed new |
| 6856 |
* imports only; this fixes rows already in the table without the customer |
| 6857 |
* having to re-source and re-upload the original PDFs (plan d1e6f7). |
| 6858 |
* |
| 6859 |
* Detection reuses MxChat_Utils::normalize_pdf_rtl() on the stored text: a |
| 6860 |
* row is a candidate exactly when the normalizer would change it, so the |
| 6861 |
* import-time heuristic and the repair heuristic can never disagree. |
| 6862 |
* Repaired rows are RE-EMBEDDED — the stored vector was computed over |
| 6863 |
* reversed text and is as broken as the text — so a wet run calls the |
| 6864 |
* embedding provider once per repaired row on the site's API key. Runs |
| 6865 |
* beyond 25 rows therefore require --yes. |
| 6866 |
* |
| 6867 |
* Scope notes: |
| 6868 |
* - Scans the WordPress knowledge table. Pinecone-mode entries live in |
| 6869 |
* Pinecone, not this table, and are not scanned; if a scanned row's bot |
| 6870 |
* ALSO has Pinecone enabled (hybrid drift), the repaired entry is |
| 6871 |
* re-submitted through the normal import path so the md5-keyed Pinecone |
| 6872 |
* vector is replaced too. |
| 6873 |
* - Knowledge rows do not carry a bot id; --bot only selects whose |
| 6874 |
* embedding configuration (model + key) is used for re-embedding. |
| 6875 |
* - The mxchat_pdf_rtl_normalize filter is honoured: a site that disabled |
| 6876 |
* normalization gets detections of zero, not surprise rewrites. |
| 6877 |
* - The metadata header the PDF importer stores before the text separator |
| 6878 |
* is preserved byte-identical; only the text segment is repaired. |
| 6879 |
* |
| 6880 |
* ## OPTIONS |
| 6881 |
* |
| 6882 |
* [--dry-run] |
| 6883 |
* : List the rows that would be repaired without changing anything. |
| 6884 |
* |
| 6885 |
* [--bot=<id>] |
| 6886 |
* : Embedding configuration to use for re-embedding. Default: default. |
| 6887 |
* |
| 6888 |
* [--all-content] |
| 6889 |
* : Scan every row containing right-to-left text, not just rows with PDF |
| 6890 |
* provenance (a page anchor in the source URL, or pdf content type). |
| 6891 |
* |
| 6892 |
* [--yes] |
| 6893 |
* : Proceed even when more than 25 rows need re-embedding (API cost gate). |
| 6894 |
* |
| 6895 |
* ## EXAMPLES |
| 6896 |
* |
| 6897 |
* wp mxchat rtl-repair --dry-run |
| 6898 |
* wp mxchat rtl-repair |
| 6899 |
* wp mxchat rtl-repair --all-content --yes |
| 6900 |
*/ |
| 6901 |
public function cli_rtl_repair($args, $assoc_args) { |
| 6902 |
global $wpdb; |
| 6903 |
$dry_run = !empty($assoc_args['dry-run']); |
| 6904 |
$all = !empty($assoc_args['all-content']); |
| 6905 |
$yes = !empty($assoc_args['yes']); |
| 6906 |
$bot_id = isset($assoc_args['bot']) ? sanitize_key($assoc_args['bot']) : 'default'; |
| 6907 |
$table = $wpdb->prefix . 'mxchat_system_prompt_content'; |
| 6908 |
|
| 6909 |
// Detection pass — no API calls. Walk the table in id batches so a large |
| 6910 |
// knowledge base never loads at once. |
| 6911 |
$rtl_re = '/[\x{0590}-\x{05FF}\x{0600}-\x{06FF}\x{0750}-\x{077F}\x{FB50}-\x{FDFF}\x{FE70}-\x{FEFF}]/u'; |
| 6912 |
$candidates = array(); |
| 6913 |
$scanned = 0; |
| 6914 |
$last_id = 0; |
| 6915 |
do { |
| 6916 |
if ($all) { |
| 6917 |
$rows = $wpdb->get_results($wpdb->prepare( |
| 6918 |
"SELECT id, article_content, source_url, content_type FROM {$table} |
| 6919 |
WHERE id > %d ORDER BY id ASC LIMIT 200", |
| 6920 |
$last_id |
| 6921 |
)); |
| 6922 |
} else { |
| 6923 |
$rows = $wpdb->get_results($wpdb->prepare( |
| 6924 |
"SELECT id, article_content, source_url, content_type FROM {$table} |
| 6925 |
WHERE id > %d AND (source_url LIKE %s OR content_type = 'pdf') |
| 6926 |
ORDER BY id ASC LIMIT 200", |
| 6927 |
$last_id, |
| 6928 |
'%' . $wpdb->esc_like('#page=') . '%' |
| 6929 |
)); |
| 6930 |
} |
| 6931 |
foreach ($rows as $row) { |
| 6932 |
$last_id = (int) $row->id; |
| 6933 |
$scanned++; |
| 6934 |
$content = (string) $row->article_content; |
| 6935 |
if (!preg_match($rtl_re, $content)) { |
| 6936 |
continue; |
| 6937 |
} |
| 6938 |
list($header, $text) = $this->mxchat_rtl_repair_split($content); |
| 6939 |
$normalized = MxChat_Utils::normalize_pdf_rtl($text, 'rtl-repair row ' . $row->id); |
| 6940 |
if (is_string($normalized) && $normalized !== $text) { |
| 6941 |
$candidates[] = array( |
| 6942 |
'id' => (int) $row->id, |
| 6943 |
'source_url' => (string) $row->source_url, |
| 6944 |
'content_type' => (string) $row->content_type, |
| 6945 |
'new_content' => $header . $normalized, |
| 6946 |
); |
| 6947 |
} |
| 6948 |
} |
| 6949 |
} while (count($rows) === 200); |
| 6950 |
|
| 6951 |
WP_CLI::log(sprintf('Scanned %d row(s); %d stored in reversed (visual) order.', $scanned, count($candidates))); |
| 6952 |
if (empty($candidates)) { |
| 6953 |
WP_CLI::success('No reversed RTL rows found — nothing to repair.'); |
| 6954 |
return; |
| 6955 |
} |
| 6956 |
|
| 6957 |
foreach ($candidates as $c) { |
| 6958 |
WP_CLI::log(sprintf('%s row %d %s', $dry_run ? 'Would repair' : 'Will repair', $c['id'], $c['source_url'])); |
| 6959 |
} |
| 6960 |
if ($dry_run) { |
| 6961 |
WP_CLI::success(sprintf('Dry run: %d row(s) would be repaired and re-embedded. Run without --dry-run to apply.', count($candidates))); |
| 6962 |
return; |
| 6963 |
} |
| 6964 |
|
| 6965 |
// Cost gate: re-embedding spends the customer's API budget. |
| 6966 |
WP_CLI::log(sprintf('Re-embedding will call the embedding provider once per row — %d call(s) on this site\'s API key.', count($candidates))); |
| 6967 |
if (count($candidates) > 25 && !$yes) { |
| 6968 |
WP_CLI::error(sprintf('%d rows need re-embedding (more than 25). Re-run with --yes to confirm the API cost. No rows were changed.', count($candidates))); |
| 6969 |
} |
| 6970 |
|
| 6971 |
$bot_options = $this->get_bot_options($bot_id); |
| 6972 |
$options = !empty($bot_options) ? $bot_options : get_option('mxchat_options'); |
| 6973 |
$preflight = MxChat_Utils::embedding_preflight($options); |
| 6974 |
if (!$preflight['ok']) { |
| 6975 |
WP_CLI::error('Embedding configuration problem: ' . $preflight['reason']); |
| 6976 |
} |
| 6977 |
$api_key = $preflight['api_key']; |
| 6978 |
|
| 6979 |
$pinecone_hybrid = $this->mxchat_rtl_repair_pinecone_enabled($bot_id); |
| 6980 |
$repaired = 0; |
| 6981 |
$failed = 0; |
| 6982 |
foreach ($candidates as $c) { |
| 6983 |
$vector = MxChat_Utils::regenerate_embedding($c['new_content'], $api_key, $bot_id); |
| 6984 |
if (!is_array($vector)) { |
| 6985 |
$failed++; |
| 6986 |
$reason = is_wp_error($vector) ? $vector->get_error_message() : 'embedding request failed'; |
| 6987 |
// Text and vector must stay consistent: never write repaired text |
| 6988 |
// beside the stale reversed-text vector. |
| 6989 |
WP_CLI::warning(sprintf('Row %d NOT repaired — %s. Row left unchanged.', $c['id'], $reason)); |
| 6990 |
continue; |
| 6991 |
} |
| 6992 |
$wpdb->update( |
| 6993 |
$table, |
| 6994 |
array( |
| 6995 |
'article_content' => $c['new_content'], |
| 6996 |
'embedding_vector' => maybe_serialize($vector), |
| 6997 |
), |
| 6998 |
array('id' => $c['id']), |
| 6999 |
array('%s', '%s'), |
| 7000 |
array('%d') |
| 7001 |
); |
| 7002 |
$repaired++; |
| 7003 |
if (class_exists('MxChat_Admin')) { |
| 7004 |
MxChat_Admin::mxchat_log_debug('pdf_rtl_repaired', 'Stored KB row restored to logical order and re-embedded', array( |
| 7005 |
'row_id' => $c['id'], |
| 7006 |
'source_url' => $c['source_url'], |
| 7007 |
'bot' => $bot_id, |
| 7008 |
)); |
| 7009 |
} |
| 7010 |
// Hybrid drift: the bot indexes into Pinecone but this row sat in the |
| 7011 |
// WP table — push the repaired entry through the normal import path so |
| 7012 |
// the md5(source_url)-keyed Pinecone vector is replaced as well. |
| 7013 |
if ($pinecone_hybrid) { |
| 7014 |
MxChat_Utils::submit_content_to_db( |
| 7015 |
$c['new_content'], |
| 7016 |
$c['source_url'], |
| 7017 |
$api_key, |
| 7018 |
null, |
| 7019 |
$bot_id, |
| 7020 |
$c['content_type'] !== '' ? $c['content_type'] : 'pdf' |
| 7021 |
); |
| 7022 |
} |
| 7023 |
} |
| 7024 |
|
| 7025 |
WP_CLI::success(sprintf('Repaired + re-embedded %d row(s); %d failed; %d scanned.', $repaired, $failed, $scanned)); |
| 7026 |
} |
| 7027 |
|
| 7028 |
/** |
| 7029 |
* Split a stored KB row into (metadata header incl. separator, text segment). |
| 7030 |
* The PDF importer stores wp_json_encode($metadata) . "\n---\n" . $text — |
| 7031 |
* repair must touch only the text and keep the header byte-identical. |
| 7032 |
*/ |
| 7033 |
private function mxchat_rtl_repair_split($content) { |
| 7034 |
$sep = "\n---\n"; |
| 7035 |
$pos = strpos($content, $sep); |
| 7036 |
if ($pos !== false && $pos > 0 && $content[0] === '{') { |
| 7037 |
$maybe_json = substr($content, 0, $pos); |
| 7038 |
if (json_decode($maybe_json) !== null) { |
| 7039 |
return array(substr($content, 0, $pos + strlen($sep)), substr($content, $pos + strlen($sep))); |
| 7040 |
} |
| 7041 |
} |
| 7042 |
return array('', $content); |
| 7043 |
} |
| 7044 |
|
| 7045 |
/** |
| 7046 |
* Mirror of MxChat_Utils::is_pinecone_enabled_for_bot() (private there) for |
| 7047 |
* the repair CLI's hybrid-drift check. |
| 7048 |
*/ |
| 7049 |
private function mxchat_rtl_repair_pinecone_enabled($bot_id) { |
| 7050 |
if ($bot_id !== 'default' && class_exists('MxChat_Multi_Bot_Manager')) { |
| 7051 |
$cfg = apply_filters('mxchat_get_bot_pinecone_config', array(), $bot_id); |
| 7052 |
if (!empty($cfg)) { |
| 7053 |
return !empty($cfg['use_pinecone']) && !empty($cfg['api_key']) && !empty($cfg['host']); |
| 7054 |
} |
| 7055 |
} |
| 7056 |
$po = get_option('mxchat_pinecone_addon_options'); |
| 7057 |
return !empty($po['mxchat_use_pinecone']) && $po['mxchat_use_pinecone'] !== '0' |
| 7058 |
&& !empty($po['mxchat_pinecone_api_key']) && !empty($po['mxchat_pinecone_host']); |
| 7059 |
} |
| 7060 |
|
| 7061 |
public function mxchat_handle_post_delete($post_id) { |
| 7062 |
// Get post data before it's deleted |
| 7063 |
$post = get_post($post_id); |
| 7064 |
|
| 7065 |
// Basic validation |
| 7066 |
if (!$post || wp_is_post_revision($post_id)) { |
| 7067 |
return; |
| 7068 |
} |
| 7069 |
|
| 7070 |
$post_type = $post->post_type; |
| 7071 |
|
| 7072 |
// Check if sync is enabled for this post type |
| 7073 |
$should_sync = false; |
| 7074 |
|
| 7075 |
// Check built-in post types first |
| 7076 |
if ($post_type === 'post' && get_option('mxchat_auto_sync_posts') === '1') { |
| 7077 |
$should_sync = true; |
| 7078 |
} else if ($post_type === 'page' && get_option('mxchat_auto_sync_pages') === '1') { |
| 7079 |
$should_sync = true; |
| 7080 |
} else { |
| 7081 |
// Check custom post types |
| 7082 |
$option_name = 'mxchat_auto_sync_' . $post_type; |
| 7083 |
if (get_option($option_name) === '1') { |
| 7084 |
$should_sync = true; |
| 7085 |
} |
| 7086 |
} |
| 7087 |
|
| 7088 |
if (!$should_sync) { |
| 7089 |
return; |
| 7090 |
} |
| 7091 |
|
| 7092 |
// Resolve the pre-trash URL. wp_trash_post renames the slug with "__trashed" before firing |
| 7093 |
// this hook, so get_permalink() here would return the trashed URL and md5() would miss the |
| 7094 |
// real vector IDs stored under the original URL. |
| 7095 |
$source_url = $this->mxchat_resolve_pre_trash_url($post_id); |
| 7096 |
if (!$source_url) { |
| 7097 |
//error_log('MXChat: Failed to resolve source URL for post ' . $post_id); |
| 7098 |
return; |
| 7099 |
} |
| 7100 |
|
| 7101 |
// Use chunk-aware deletion (handles both chunked and non-chunked content) |
| 7102 |
$delete_result = MxChat_Utils::delete_chunks_for_url($source_url, 'default'); |
| 7103 |
|
| 7104 |
if (is_wp_error($delete_result)) { |
| 7105 |
//error_log('MXChat: Chunk-aware deletion failed for URL: ' . $source_url . ' - ' . $delete_result->get_error_message()); |
| 7106 |
} |
| 7107 |
|
| 7108 |
delete_transient('mxchat_prev_url_' . $post_id); |
| 7109 |
delete_transient('mxchat_prev_status_' . $post_id); |
| 7110 |
} |
| 7111 |
|
| 7112 |
/** |
| 7113 |
* Resolve the source URL for a post being trashed/deleted. |
| 7114 |
* |
| 7115 |
* Why: wp_trash_post appends "__trashed" to the slug before the wp_trash_post action fires, so |
| 7116 |
* get_permalink() returns a URL whose md5() won't match the vector IDs stored in Pinecone or |
| 7117 |
* the source_url rows in the WP DB. Prefer the URL captured by mxchat_store_pre_update_status |
| 7118 |
* (runs on pre_post_update, before the rename); fall back to stripping the __trashed suffix. |
| 7119 |
*/ |
| 7120 |
private function mxchat_resolve_pre_trash_url($post_id) { |
| 7121 |
$previous_url = get_transient('mxchat_prev_url_' . $post_id); |
| 7122 |
if (!empty($previous_url)) { |
| 7123 |
return $previous_url; |
| 7124 |
} |
| 7125 |
|
| 7126 |
$current = get_permalink($post_id); |
| 7127 |
if (!$current) { |
| 7128 |
return ''; |
| 7129 |
} |
| 7130 |
return preg_replace('#__trashed(/?)$#', '$1', $current); |
| 7131 |
} |
| 7132 |
|
| 7133 |
|
| 7134 |
|
| 7135 |
public function mxchat_handle_product_change($post_id, $post, $update) { |
| 7136 |
if ($post->post_type !== 'product') { |
| 7137 |
return; |
| 7138 |
} |
| 7139 |
|
| 7140 |
if ($post->post_status === 'publish') { |
| 7141 |
add_action('shutdown', function() use ($post_id) { |
| 7142 |
$product = wc_get_product($post_id); |
| 7143 |
if ($product) { |
| 7144 |
$this->mxchat_store_product_embedding($product); |
| 7145 |
} |
| 7146 |
}); |
| 7147 |
} |
| 7148 |
} |
| 7149 |
|
| 7150 |
/** |
| 7151 |
* Store WooCommerce product embeddings |
| 7152 |
*/ |
| 7153 |
private function mxchat_store_product_embedding($product) { |
| 7154 |
if (!isset($this->options['enable_woocommerce_integration']) || |
| 7155 |
!in_array($this->options['enable_woocommerce_integration'], ['1', 'on'])) { |
| 7156 |
return; |
| 7157 |
} |
| 7158 |
|
| 7159 |
$source_url = get_permalink($product->get_id()); |
| 7160 |
$product_id = $product->get_id(); |
| 7161 |
|
| 7162 |
// Build product content via the shared WC-object assembler (a3d60c) — this |
| 7163 |
// writer owns product rows whenever the integration is on. |
| 7164 |
$content = $this->mxchat_prepare_product_content_for_indexing($product); |
| 7165 |
|
| 7166 |
// Embedding decision — custom-provider-aware (plan cbd5fd); silent-return |
| 7167 |
// shape preserved. |
| 7168 |
$preflight = MxChat_Utils::embedding_preflight(get_option('mxchat_options')); |
| 7169 |
if (!$preflight['ok']) { |
| 7170 |
//error_log('MxChat Auto-sync: embedding pre-flight failed: ' . $preflight['reason']); |
| 7171 |
return; |
| 7172 |
} |
| 7173 |
$api_key = $preflight['api_key']; |
| 7174 |
|
| 7175 |
// Use the centralized utility function for storage |
| 7176 |
$result = MxChat_Utils::submit_content_to_db( |
| 7177 |
$content, |
| 7178 |
$source_url, |
| 7179 |
$api_key, |
| 7180 |
md5($source_url) // Vector ID for Pinecone |
| 7181 |
); |
| 7182 |
|
| 7183 |
// After successful storage, apply role restriction based on tags |
| 7184 |
if (!is_wp_error($result)) { |
| 7185 |
$this->apply_role_restriction_to_post($product_id, $source_url); |
| 7186 |
} |
| 7187 |
|
| 7188 |
if (is_wp_error($result)) { |
| 7189 |
//error_log('MxChat WooCommerce sync failed for product ' . $product_id . ': ' . $result->get_error_message()); |
| 7190 |
} |
| 7191 |
} |
| 7192 |
|
| 7193 |
public function mxchat_handle_product_delete($post_id) { |
| 7194 |
if (get_post_type($post_id) !== 'product') { |
| 7195 |
return; |
| 7196 |
} |
| 7197 |
|
| 7198 |
$source_url = $this->mxchat_resolve_pre_trash_url($post_id); |
| 7199 |
if (!$source_url) { |
| 7200 |
return; |
| 7201 |
} |
| 7202 |
|
| 7203 |
// Chunk-aware deletion (routes to Pinecone or WP DB and removes base + all chunks) |
| 7204 |
MxChat_Utils::delete_chunks_for_url($source_url, 'default'); |
| 7205 |
|
| 7206 |
delete_transient('mxchat_prev_url_' . $post_id); |
| 7207 |
delete_transient('mxchat_prev_status_' . $post_id); |
| 7208 |
} |
| 7209 |
|
| 7210 |
/** |
| 7211 |
* Handle individual Pinecone content deletion |
| 7212 |
*/ |
| 7213 |
public function mxchat_handle_pinecone_prompt_delete() { |
| 7214 |
// Check permissions |
| 7215 |
if (!current_user_can('manage_options')) { |
| 7216 |
wp_die(esc_html__('You do not have sufficient permissions.', 'mxchat')); |
| 7217 |
} |
| 7218 |
|
| 7219 |
// Verify nonce |
| 7220 |
if (!isset($_GET['_wpnonce']) || !wp_verify_nonce($_GET['_wpnonce'], 'mxchat_delete_pinecone_prompt_nonce')) { |
| 7221 |
wp_die(esc_html__('Security check failed.', 'mxchat')); |
| 7222 |
} |
| 7223 |
|
| 7224 |
$vector_id = isset($_GET['vector_id']) ? sanitize_text_field($_GET['vector_id']) : ''; |
| 7225 |
|
| 7226 |
if (empty($vector_id)) { |
| 7227 |
set_transient('mxchat_admin_notice_error', |
| 7228 |
esc_html__('Invalid vector ID.', 'mxchat'), |
| 7229 |
30 |
| 7230 |
); |
| 7231 |
wp_safe_redirect(admin_url('admin.php?page=mxchat-prompts')); |
| 7232 |
exit; |
| 7233 |
} |
| 7234 |
|
| 7235 |
// Get Pinecone settings |
| 7236 |
$pinecone_options = get_option('mxchat_pinecone_addon_options', array()); |
| 7237 |
$use_pinecone = ($pinecone_options['mxchat_use_pinecone'] ?? '0') === '1'; |
| 7238 |
|
| 7239 |
if (!$use_pinecone || empty($pinecone_options['mxchat_pinecone_api_key'])) { |
| 7240 |
set_transient('mxchat_admin_notice_error', |
| 7241 |
esc_html__('Pinecone is not properly configured.', 'mxchat'), |
| 7242 |
30 |
| 7243 |
); |
| 7244 |
wp_safe_redirect(admin_url('admin.php?page=mxchat-prompts')); |
| 7245 |
exit; |
| 7246 |
} |
| 7247 |
|
| 7248 |
// Delete from Pinecone |
| 7249 |
$pinecone_manager = MxChat_Pinecone_Manager::get_instance(); |
| 7250 |
$result = $pinecone_manager->mxchat_delete_from_pinecone_by_vector_id( |
| 7251 |
$vector_id, |
| 7252 |
$pinecone_options['mxchat_pinecone_api_key'], |
| 7253 |
$pinecone_options['mxchat_pinecone_host'], |
| 7254 |
$pinecone_options['mxchat_pinecone_namespace'] ?? '' |
| 7255 |
); |
| 7256 |
|
| 7257 |
if ($result['success']) { |
| 7258 |
// Mirror the removal to the OpenAI Vector Store mapping (plan 15b5c6); |
| 7259 |
// a chunk vector id reduces to its base entry there. |
| 7260 |
if (class_exists('MxChat_Vectorstore_Manager')) { |
| 7261 |
MxChat_Vectorstore_Manager::sync_delete_by_key($vector_id, 'default'); |
| 7262 |
} |
| 7263 |
// No cache clearing needed since we removed caching |
| 7264 |
set_transient('mxchat_admin_notice_success', |
| 7265 |
esc_html__('Entry deleted successfully from Pinecone.', 'mxchat'), |
| 7266 |
30 |
| 7267 |
); |
| 7268 |
} else { |
| 7269 |
set_transient('mxchat_admin_notice_error', |
| 7270 |
esc_html__('Failed to delete entry: ', 'mxchat') . $result['message'], |
| 7271 |
30 |
| 7272 |
); |
| 7273 |
} |
| 7274 |
|
| 7275 |
wp_safe_redirect(admin_url('admin.php?page=mxchat-prompts')); |
| 7276 |
exit; |
| 7277 |
} |
| 7278 |
/** |
| 7279 |
* Handle individual Pinecone content deletion via AJAX |
| 7280 |
*/ |
| 7281 |
public function ajax_mxchat_delete_pinecone_prompt() { |
| 7282 |
// Verify nonce and permissions |
| 7283 |
if (!check_ajax_referer('mxchat_delete_pinecone_prompt_nonce', 'nonce', false)) { |
| 7284 |
wp_send_json_error('Invalid nonce'); |
| 7285 |
exit; |
| 7286 |
} |
| 7287 |
|
| 7288 |
if (!current_user_can('manage_options')) { |
| 7289 |
wp_send_json_error('Unauthorized access'); |
| 7290 |
exit; |
| 7291 |
} |
| 7292 |
|
| 7293 |
$vector_id = isset($_POST['vector_id']) ? sanitize_text_field($_POST['vector_id']) : ''; |
| 7294 |
$bot_id = isset($_POST['bot_id']) ? sanitize_text_field($_POST['bot_id']) : 'default'; |
| 7295 |
|
| 7296 |
if (empty($vector_id)) { |
| 7297 |
wp_send_json_error('Missing vector ID'); |
| 7298 |
exit; |
| 7299 |
} |
| 7300 |
|
| 7301 |
// Get bot-specific Pinecone settings |
| 7302 |
$pinecone_manager = MxChat_Pinecone_Manager::get_instance(); |
| 7303 |
$pinecone_options = $pinecone_manager->mxchat_get_bot_pinecone_options($bot_id); |
| 7304 |
|
| 7305 |
$use_pinecone = ($pinecone_options['mxchat_use_pinecone'] ?? '0') === '1'; |
| 7306 |
|
| 7307 |
if (!$use_pinecone || empty($pinecone_options['mxchat_pinecone_api_key'])) { |
| 7308 |
wp_send_json_error('Pinecone is not properly configured for bot: ' . $bot_id); |
| 7309 |
exit; |
| 7310 |
} |
| 7311 |
|
| 7312 |
// Delete from the correct Pinecone index and namespace |
| 7313 |
$result = $pinecone_manager->mxchat_delete_from_pinecone_by_vector_id( |
| 7314 |
$vector_id, |
| 7315 |
$pinecone_options['mxchat_pinecone_api_key'], |
| 7316 |
$pinecone_options['mxchat_pinecone_host'], |
| 7317 |
$pinecone_options['mxchat_pinecone_namespace'] ?? '' |
| 7318 |
); |
| 7319 |
|
| 7320 |
if ($result['success']) { |
| 7321 |
// Mirror the removal to the OpenAI Vector Store mapping (plan 15b5c6) |
| 7322 |
if (class_exists('MxChat_Vectorstore_Manager')) { |
| 7323 |
MxChat_Vectorstore_Manager::sync_delete_by_key($vector_id, $bot_id); |
| 7324 |
} |
| 7325 |
// No cache clearing needed since we removed caching |
| 7326 |
wp_send_json_success(array( |
| 7327 |
'message' => 'Entry deleted successfully from Pinecone', |
| 7328 |
'vector_id' => $vector_id, |
| 7329 |
'bot_id' => $bot_id |
| 7330 |
)); |
| 7331 |
} else { |
| 7332 |
MxChat_Admin::mxchat_log_debug('pinecone_error', 'Failed to delete from Pinecone: ' . $result['message'], array('vector_id' => $vector_id, 'bot_id' => $bot_id)); |
| 7333 |
wp_send_json_error('Failed to delete from Pinecone: ' . $result['message']); |
| 7334 |
} |
| 7335 |
|
| 7336 |
exit; |
| 7337 |
} |
| 7338 |
|
| 7339 |
/** |
| 7340 |
* Handle deletion of all chunks for a given source URL via AJAX |
| 7341 |
* Follows the same pattern as ajax_mxchat_delete_pinecone_prompt |
| 7342 |
*/ |
| 7343 |
public function ajax_mxchat_delete_chunks_by_url() { |
| 7344 |
// Verify nonce and permissions |
| 7345 |
if (!check_ajax_referer('mxchat_delete_chunks_nonce', 'nonce', false)) { |
| 7346 |
wp_send_json_error('Invalid nonce'); |
| 7347 |
exit; |
| 7348 |
} |
| 7349 |
|
| 7350 |
if (!current_user_can('manage_options')) { |
| 7351 |
wp_send_json_error('Unauthorized access'); |
| 7352 |
exit; |
| 7353 |
} |
| 7354 |
|
| 7355 |
$source_url = isset($_POST['source_url']) ? esc_url_raw($_POST['source_url']) : ''; |
| 7356 |
$data_source = isset($_POST['data_source']) ? sanitize_text_field($_POST['data_source']) : 'wordpress'; |
| 7357 |
$bot_id = isset($_POST['bot_id']) ? sanitize_text_field($_POST['bot_id']) : 'default'; |
| 7358 |
|
| 7359 |
if (empty($source_url)) { |
| 7360 |
wp_send_json_error('Missing source URL'); |
| 7361 |
exit; |
| 7362 |
} |
| 7363 |
|
| 7364 |
// Generate the base vector ID from the source URL (same as how chunks are created) |
| 7365 |
$base_vector_id = md5($source_url); |
| 7366 |
|
| 7367 |
if ($data_source === 'pinecone') { |
| 7368 |
// Get bot-specific Pinecone settings (same as working delete function) |
| 7369 |
$pinecone_manager = MxChat_Pinecone_Manager::get_instance(); |
| 7370 |
$pinecone_options = $pinecone_manager->mxchat_get_bot_pinecone_options($bot_id); |
| 7371 |
|
| 7372 |
$use_pinecone = ($pinecone_options['mxchat_use_pinecone'] ?? '0') === '1'; |
| 7373 |
|
| 7374 |
if (!$use_pinecone || empty($pinecone_options['mxchat_pinecone_api_key'])) { |
| 7375 |
wp_send_json_error('Pinecone is not properly configured for bot: ' . $bot_id); |
| 7376 |
exit; |
| 7377 |
} |
| 7378 |
|
| 7379 |
$api_key = $pinecone_options['mxchat_pinecone_api_key']; |
| 7380 |
$host = $pinecone_options['mxchat_pinecone_host']; |
| 7381 |
$namespace = $pinecone_options['mxchat_pinecone_namespace'] ?? ''; |
| 7382 |
|
| 7383 |
// Collect all vector IDs to delete |
| 7384 |
$vectors_to_delete = array(); |
| 7385 |
|
| 7386 |
// Add the original single-vector ID (for non-chunked content) |
| 7387 |
$vectors_to_delete[] = $base_vector_id; |
| 7388 |
|
| 7389 |
// Use Pinecone list API to find all chunk vectors with this prefix |
| 7390 |
// NOTE: Pinecone List API is a GET request with query parameters, not POST |
| 7391 |
$prefix = $base_vector_id . '_chunk_'; |
| 7392 |
|
| 7393 |
$query_params = array( |
| 7394 |
'prefix' => $prefix, |
| 7395 |
'limit' => 100 |
| 7396 |
); |
| 7397 |
|
| 7398 |
if (!empty($namespace)) { |
| 7399 |
$query_params['namespace'] = $namespace; |
| 7400 |
} |
| 7401 |
|
| 7402 |
$list_url = "https://{$host}/vectors/list?" . http_build_query($query_params); |
| 7403 |
|
| 7404 |
$list_response = wp_remote_get($list_url, array( |
| 7405 |
'headers' => array( |
| 7406 |
'Api-Key' => $api_key, |
| 7407 |
'accept' => 'application/json' |
| 7408 |
), |
| 7409 |
'timeout' => 30 |
| 7410 |
)); |
| 7411 |
|
| 7412 |
if (!is_wp_error($list_response)) { |
| 7413 |
$list_body_response = wp_remote_retrieve_body($list_response); |
| 7414 |
$list_data = json_decode($list_body_response, true); |
| 7415 |
if (!empty($list_data['vectors'])) { |
| 7416 |
foreach ($list_data['vectors'] as $vector) { |
| 7417 |
if (isset($vector['id'])) { |
| 7418 |
$vectors_to_delete[] = $vector['id']; |
| 7419 |
} |
| 7420 |
} |
| 7421 |
} |
| 7422 |
} |
| 7423 |
|
| 7424 |
if (empty($vectors_to_delete)) { |
| 7425 |
// Entry already gone from Pinecone — still clear any mirrored |
| 7426 |
// Vector Store file so it can't outlive the entry (plan 15b5c6). |
| 7427 |
if (class_exists('MxChat_Vectorstore_Manager')) { |
| 7428 |
MxChat_Vectorstore_Manager::sync_delete_entry($source_url, $bot_id); |
| 7429 |
} |
| 7430 |
wp_send_json_success(array( |
| 7431 |
'message' => 'No vectors found to delete', |
| 7432 |
'source_url' => $source_url |
| 7433 |
)); |
| 7434 |
exit; |
| 7435 |
} |
| 7436 |
|
| 7437 |
// Delete all vectors using the same endpoint as the working function |
| 7438 |
$delete_url = "https://{$host}/vectors/delete"; |
| 7439 |
|
| 7440 |
$delete_body = array( |
| 7441 |
'ids' => $vectors_to_delete |
| 7442 |
); |
| 7443 |
|
| 7444 |
if (!empty($namespace)) { |
| 7445 |
$delete_body['namespace'] = $namespace; |
| 7446 |
} |
| 7447 |
|
| 7448 |
$delete_response = wp_remote_post($delete_url, array( |
| 7449 |
'headers' => array( |
| 7450 |
'Api-Key' => $api_key, |
| 7451 |
'accept' => 'application/json', |
| 7452 |
'content-type' => 'application/json' |
| 7453 |
), |
| 7454 |
'body' => wp_json_encode($delete_body), |
| 7455 |
'timeout' => 30 |
| 7456 |
)); |
| 7457 |
|
| 7458 |
if (is_wp_error($delete_response)) { |
| 7459 |
MxChat_Admin::mxchat_log_debug('pinecone_error', 'Failed to delete chunks from Pinecone: ' . $delete_response->get_error_message(), array('source_url' => $source_url)); |
| 7460 |
wp_send_json_error('Failed to delete from Pinecone: ' . $delete_response->get_error_message()); |
| 7461 |
exit; |
| 7462 |
} |
| 7463 |
|
| 7464 |
$response_code = wp_remote_retrieve_response_code($delete_response); |
| 7465 |
|
| 7466 |
if ($response_code !== 200) { |
| 7467 |
MxChat_Admin::mxchat_log_debug('pinecone_error', 'Pinecone API error (HTTP ' . $response_code . ')', array('source_url' => $source_url)); |
| 7468 |
wp_send_json_error('Pinecone API error (HTTP ' . $response_code . ')'); |
| 7469 |
exit; |
| 7470 |
} |
| 7471 |
|
| 7472 |
// Mirror the removal to the OpenAI Vector Store (plan 15b5c6) |
| 7473 |
if (class_exists('MxChat_Vectorstore_Manager')) { |
| 7474 |
MxChat_Vectorstore_Manager::sync_delete_entry($source_url, $bot_id); |
| 7475 |
} |
| 7476 |
|
| 7477 |
wp_send_json_success(array( |
| 7478 |
'message' => 'All chunks deleted successfully from Pinecone', |
| 7479 |
'source_url' => $source_url, |
| 7480 |
'deleted_count' => count($vectors_to_delete) |
| 7481 |
)); |
| 7482 |
|
| 7483 |
} else { |
| 7484 |
// WordPress database deletion |
| 7485 |
global $wpdb; |
| 7486 |
$table_name = $wpdb->prefix . 'mxchat_system_prompt_content'; |
| 7487 |
|
| 7488 |
$result = $wpdb->delete( |
| 7489 |
$table_name, |
| 7490 |
array('source_url' => $source_url), |
| 7491 |
array('%s') |
| 7492 |
); |
| 7493 |
|
| 7494 |
if ($result === false) { |
| 7495 |
MxChat_Admin::mxchat_log_debug('database_error', 'Failed to delete from database: ' . $wpdb->last_error, array('source_url' => $source_url)); |
| 7496 |
wp_send_json_error('Failed to delete from database: ' . $wpdb->last_error); |
| 7497 |
exit; |
| 7498 |
} |
| 7499 |
|
| 7500 |
// Mirror the removal to the OpenAI Vector Store (plan 15b5c6) |
| 7501 |
if (class_exists('MxChat_Vectorstore_Manager')) { |
| 7502 |
MxChat_Vectorstore_Manager::sync_delete_entry($source_url, $bot_id); |
| 7503 |
} |
| 7504 |
|
| 7505 |
wp_send_json_success(array( |
| 7506 |
'message' => 'All chunks deleted successfully from database', |
| 7507 |
'source_url' => $source_url, |
| 7508 |
'deleted_count' => $result |
| 7509 |
)); |
| 7510 |
} |
| 7511 |
|
| 7512 |
exit; |
| 7513 |
} |
| 7514 |
|
| 7515 |
/** |
| 7516 |
* Handle individual WordPress database content deletion via AJAX |
| 7517 |
* Mirrors the Pinecone delete handler but for WordPress database entries |
| 7518 |
*/ |
| 7519 |
public function ajax_mxchat_delete_wordpress_prompt() { |
| 7520 |
// Verify nonce and permissions |
| 7521 |
if (!check_ajax_referer('mxchat_delete_wordpress_prompt_nonce', 'nonce', false)) { |
| 7522 |
wp_send_json_error('Invalid nonce'); |
| 7523 |
exit; |
| 7524 |
} |
| 7525 |
|
| 7526 |
if (!current_user_can('manage_options')) { |
| 7527 |
wp_send_json_error('Unauthorized access'); |
| 7528 |
exit; |
| 7529 |
} |
| 7530 |
|
| 7531 |
$entry_id = isset($_POST['entry_id']) ? intval($_POST['entry_id']) : 0; |
| 7532 |
|
| 7533 |
if (empty($entry_id)) { |
| 7534 |
wp_send_json_error('Missing entry ID'); |
| 7535 |
exit; |
| 7536 |
} |
| 7537 |
|
| 7538 |
global $wpdb; |
| 7539 |
$table_name = $wpdb->prefix . 'mxchat_system_prompt_content'; |
| 7540 |
|
| 7541 |
// Capture the identity BEFORE the row disappears — needed to mirror the |
| 7542 |
// change into the Vector Store (plan 15b5c6). |
| 7543 |
$source_url = $wpdb->get_var($wpdb->prepare( |
| 7544 |
"SELECT source_url FROM {$table_name} WHERE id = %d", |
| 7545 |
$entry_id |
| 7546 |
)); |
| 7547 |
|
| 7548 |
// Clear cache for this entry |
| 7549 |
wp_cache_delete('prompt_' . $entry_id, 'mxchat_prompts'); |
| 7550 |
|
| 7551 |
// Delete from database |
| 7552 |
$result = $wpdb->delete( |
| 7553 |
$table_name, |
| 7554 |
array('id' => $entry_id), |
| 7555 |
array('%d') |
| 7556 |
); |
| 7557 |
|
| 7558 |
if ($result !== false) { |
| 7559 |
// Mirror to the Vector Store: if sibling rows remain (this was one |
| 7560 |
// chunk of a larger entry) the entry's file is REFRESHED from what's |
| 7561 |
// left; if none remain, the file is removed. |
| 7562 |
if (!empty($source_url) && class_exists('MxChat_Vectorstore_Manager')) { |
| 7563 |
$remaining = (int) $wpdb->get_var($wpdb->prepare( |
| 7564 |
"SELECT COUNT(*) FROM {$table_name} WHERE source_url = %s", |
| 7565 |
$source_url |
| 7566 |
)); |
| 7567 |
if ($remaining > 0) { |
| 7568 |
MxChat_Vectorstore_Manager::sync_upsert_entry($source_url, '', 'default'); |
| 7569 |
} else { |
| 7570 |
MxChat_Vectorstore_Manager::sync_delete_entry($source_url, 'default'); |
| 7571 |
} |
| 7572 |
} |
| 7573 |
|
| 7574 |
wp_send_json_success(array( |
| 7575 |
'message' => 'Entry deleted successfully', |
| 7576 |
'entry_id' => $entry_id |
| 7577 |
)); |
| 7578 |
} else { |
| 7579 |
wp_send_json_error('Failed to delete entry from database'); |
| 7580 |
} |
| 7581 |
|
| 7582 |
exit; |
| 7583 |
} |
| 7584 |
|
| 7585 |
/** |
| 7586 |
* Handle bulk deletion of knowledge entries via AJAX |
| 7587 |
* Supports both Pinecone and WordPress database entries |
| 7588 |
*/ |
| 7589 |
public function ajax_mxchat_bulk_delete_knowledge() { |
| 7590 |
// Verify nonce and permissions |
| 7591 |
if (!check_ajax_referer('mxchat_bulk_delete_knowledge_nonce', 'nonce', false)) { |
| 7592 |
wp_send_json_error('Invalid nonce'); |
| 7593 |
exit; |
| 7594 |
} |
| 7595 |
|
| 7596 |
if (!current_user_can('manage_options')) { |
| 7597 |
wp_send_json_error('Unauthorized access'); |
| 7598 |
exit; |
| 7599 |
} |
| 7600 |
|
| 7601 |
$entries = isset($_POST['entries']) ? $_POST['entries'] : array(); |
| 7602 |
$bot_id = isset($_POST['bot_id']) ? sanitize_text_field($_POST['bot_id']) : 'default'; |
| 7603 |
|
| 7604 |
if (empty($entries) || !is_array($entries)) { |
| 7605 |
wp_send_json_error('No entries provided'); |
| 7606 |
exit; |
| 7607 |
} |
| 7608 |
|
| 7609 |
// Extend execution time — bulk Pinecone operations can take a while |
| 7610 |
if (function_exists('set_time_limit')) { |
| 7611 |
set_time_limit(120); |
| 7612 |
} |
| 7613 |
|
| 7614 |
$success_ids = array(); |
| 7615 |
$failed_ids = array(); |
| 7616 |
$errors = array(); |
| 7617 |
|
| 7618 |
global $wpdb; |
| 7619 |
$table_name = $wpdb->prefix . 'mxchat_system_prompt_content'; |
| 7620 |
|
| 7621 |
// Get Pinecone manager for Pinecone deletions |
| 7622 |
$pinecone_manager = MxChat_Pinecone_Manager::get_instance(); |
| 7623 |
$pinecone_options = $pinecone_manager->mxchat_get_bot_pinecone_options($bot_id); |
| 7624 |
$use_pinecone = ($pinecone_options['mxchat_use_pinecone'] ?? '0') === '1'; |
| 7625 |
|
| 7626 |
$api_key = $pinecone_options['mxchat_pinecone_api_key'] ?? ''; |
| 7627 |
$host = $pinecone_options['mxchat_pinecone_host'] ?? ''; |
| 7628 |
$namespace = $pinecone_options['mxchat_pinecone_namespace'] ?? ''; |
| 7629 |
|
| 7630 |
// ============================================= |
| 7631 |
// PHASE 1: Collect all Pinecone vector IDs |
| 7632 |
// and separate WordPress entries |
| 7633 |
// ============================================= |
| 7634 |
$pinecone_entry_ids = array(); // entry IDs that are pinecone-sourced |
| 7635 |
$wordpress_entries = array(); // entries for WordPress DB deletion |
| 7636 |
$all_vector_ids = array(); // all pinecone vector IDs to delete in one batch |
| 7637 |
$vs_mirror_urls = array(); // Vector Store mirror: URLs to delete (plan 15b5c6) |
| 7638 |
$vs_mirror_keys = array(); // Vector Store mirror: bare vector ids to delete |
| 7639 |
|
| 7640 |
foreach ($entries as $entry) { |
| 7641 |
$entry_id = sanitize_text_field($entry['id'] ?? ''); |
| 7642 |
$source = sanitize_text_field($entry['source'] ?? 'wordpress'); |
| 7643 |
$source_url = isset($entry['sourceUrl']) ? esc_url_raw($entry['sourceUrl']) : ''; |
| 7644 |
$is_group = isset($entry['isGroup']) && ($entry['isGroup'] === true || $entry['isGroup'] === 'true'); |
| 7645 |
|
| 7646 |
if (empty($entry_id)) { |
| 7647 |
continue; |
| 7648 |
} |
| 7649 |
|
| 7650 |
if ($source === 'pinecone') { |
| 7651 |
if (!$use_pinecone || empty($api_key)) { |
| 7652 |
$failed_ids[] = $entry_id; |
| 7653 |
$errors[] = "Pinecone not configured for entry: $entry_id"; |
| 7654 |
continue; |
| 7655 |
} |
| 7656 |
|
| 7657 |
$pinecone_entry_ids[] = $entry_id; |
| 7658 |
|
| 7659 |
if ($is_group && !empty($source_url)) { |
| 7660 |
// Grouped/chunked entry: collect base ID + chunk IDs via List API |
| 7661 |
$base_vector_id = md5($source_url); |
| 7662 |
$all_vector_ids[] = $base_vector_id; |
| 7663 |
|
| 7664 |
$list_url = 'https://' . $host . '/vectors/list?prefix=' . urlencode($base_vector_id . '_chunk_') . '&limit=100'; |
| 7665 |
if (!empty($namespace)) { |
| 7666 |
$list_url .= '&namespace=' . rawurlencode($namespace); |
| 7667 |
} |
| 7668 |
$list_response = wp_remote_get($list_url, array( |
| 7669 |
'headers' => array( |
| 7670 |
'Api-Key' => $api_key, |
| 7671 |
'accept' => 'application/json' |
| 7672 |
), |
| 7673 |
'timeout' => 30 |
| 7674 |
)); |
| 7675 |
|
| 7676 |
if (!is_wp_error($list_response)) { |
| 7677 |
$list_body = json_decode(wp_remote_retrieve_body($list_response), true); |
| 7678 |
if (!empty($list_body['vectors']) && is_array($list_body['vectors'])) { |
| 7679 |
foreach ($list_body['vectors'] as $vector) { |
| 7680 |
if (isset($vector['id'])) { |
| 7681 |
$all_vector_ids[] = $vector['id']; |
| 7682 |
} |
| 7683 |
} |
| 7684 |
} |
| 7685 |
} |
| 7686 |
} else { |
| 7687 |
// Single entry: the entry_id IS the vector ID |
| 7688 |
$all_vector_ids[] = $entry_id; |
| 7689 |
} |
| 7690 |
|
| 7691 |
if (!empty($source_url)) { |
| 7692 |
$vs_mirror_urls[] = $source_url; |
| 7693 |
} else { |
| 7694 |
$vs_mirror_keys[] = $entry_id; |
| 7695 |
} |
| 7696 |
} else { |
| 7697 |
$wordpress_entries[] = $entry; |
| 7698 |
} |
| 7699 |
} |
| 7700 |
|
| 7701 |
// ============================================= |
| 7702 |
// PHASE 2: Single batch delete to Pinecone |
| 7703 |
// ============================================= |
| 7704 |
if (!empty($all_vector_ids)) { |
| 7705 |
$all_vector_ids = array_values(array_unique($all_vector_ids)); |
| 7706 |
$pinecone_success = true; |
| 7707 |
$batches = array_chunk($all_vector_ids, 100); |
| 7708 |
|
| 7709 |
foreach ($batches as $batch) { |
| 7710 |
$delete_body = array('ids' => $batch); |
| 7711 |
if (!empty($namespace)) { |
| 7712 |
$delete_body['namespace'] = $namespace; |
| 7713 |
} |
| 7714 |
$delete_response = wp_remote_post("https://{$host}/vectors/delete", array( |
| 7715 |
'headers' => array( |
| 7716 |
'Api-Key' => $api_key, |
| 7717 |
'accept' => 'application/json', |
| 7718 |
'content-type' => 'application/json' |
| 7719 |
), |
| 7720 |
'body' => wp_json_encode($delete_body), |
| 7721 |
'timeout' => 60 |
| 7722 |
)); |
| 7723 |
|
| 7724 |
if (is_wp_error($delete_response)) { |
| 7725 |
$pinecone_success = false; |
| 7726 |
$errors[] = 'Pinecone batch deletion failed: ' . $delete_response->get_error_message(); |
| 7727 |
MxChat_Admin::mxchat_log_debug('pinecone_error', 'Bulk delete batch failed: ' . $delete_response->get_error_message()); |
| 7728 |
} else { |
| 7729 |
$response_code = wp_remote_retrieve_response_code($delete_response); |
| 7730 |
if ($response_code !== 200) { |
| 7731 |
$pinecone_success = false; |
| 7732 |
$response_body = wp_remote_retrieve_body($delete_response); |
| 7733 |
$errors[] = "Pinecone API error (HTTP $response_code)"; |
| 7734 |
MxChat_Admin::mxchat_log_debug('pinecone_error', 'Bulk delete batch failed (HTTP ' . $response_code . ')', array('response' => substr($response_body, 0, 200))); |
| 7735 |
} |
| 7736 |
} |
| 7737 |
} |
| 7738 |
|
| 7739 |
// Mark all pinecone entries based on batch result |
| 7740 |
foreach ($pinecone_entry_ids as $eid) { |
| 7741 |
if ($pinecone_success) { |
| 7742 |
$success_ids[] = $eid; |
| 7743 |
} else { |
| 7744 |
$failed_ids[] = $eid; |
| 7745 |
} |
| 7746 |
} |
| 7747 |
|
| 7748 |
// Mirror the removals to the OpenAI Vector Store (plan 15b5c6) |
| 7749 |
if ($pinecone_success && class_exists('MxChat_Vectorstore_Manager')) { |
| 7750 |
foreach (array_unique($vs_mirror_urls) as $vs_url) { |
| 7751 |
MxChat_Vectorstore_Manager::sync_delete_entry($vs_url, $bot_id); |
| 7752 |
} |
| 7753 |
foreach (array_unique($vs_mirror_keys) as $vs_key) { |
| 7754 |
MxChat_Vectorstore_Manager::sync_delete_by_key($vs_key, $bot_id); |
| 7755 |
} |
| 7756 |
} |
| 7757 |
} |
| 7758 |
|
| 7759 |
// ============================================= |
| 7760 |
// PHASE 3: WordPress database deletions |
| 7761 |
// ============================================= |
| 7762 |
foreach ($wordpress_entries as $entry) { |
| 7763 |
$entry_id = sanitize_text_field($entry['id'] ?? ''); |
| 7764 |
$source_url = isset($entry['sourceUrl']) ? esc_url_raw($entry['sourceUrl']) : ''; |
| 7765 |
$is_group = isset($entry['isGroup']) && ($entry['isGroup'] === true || $entry['isGroup'] === 'true'); |
| 7766 |
|
| 7767 |
if (empty($entry_id)) { |
| 7768 |
continue; |
| 7769 |
} |
| 7770 |
|
| 7771 |
try { |
| 7772 |
if ($is_group && !empty($source_url)) { |
| 7773 |
$result = $wpdb->delete( |
| 7774 |
$table_name, |
| 7775 |
array('source_url' => $source_url), |
| 7776 |
array('%s') |
| 7777 |
); |
| 7778 |
$row_url = $source_url; |
| 7779 |
} else { |
| 7780 |
// Identity captured pre-delete for the Vector Store mirror |
| 7781 |
$row_url = $wpdb->get_var($wpdb->prepare( |
| 7782 |
"SELECT source_url FROM {$table_name} WHERE id = %d", |
| 7783 |
intval($entry_id) |
| 7784 |
)); |
| 7785 |
wp_cache_delete('prompt_' . $entry_id, 'mxchat_prompts'); |
| 7786 |
$result = $wpdb->delete( |
| 7787 |
$table_name, |
| 7788 |
array('id' => intval($entry_id)), |
| 7789 |
array('%d') |
| 7790 |
); |
| 7791 |
} |
| 7792 |
|
| 7793 |
if ($result !== false) { |
| 7794 |
$success_ids[] = $entry_id; |
| 7795 |
// Mirror to the Vector Store: refresh the entry's file when |
| 7796 |
// sibling chunk rows survive, remove it when none do. |
| 7797 |
if (!empty($row_url) && class_exists('MxChat_Vectorstore_Manager')) { |
| 7798 |
$remaining = (int) $wpdb->get_var($wpdb->prepare( |
| 7799 |
"SELECT COUNT(*) FROM {$table_name} WHERE source_url = %s", |
| 7800 |
$row_url |
| 7801 |
)); |
| 7802 |
if ($remaining > 0) { |
| 7803 |
MxChat_Vectorstore_Manager::sync_upsert_entry($row_url, '', $bot_id); |
| 7804 |
} else { |
| 7805 |
MxChat_Vectorstore_Manager::sync_delete_entry($row_url, $bot_id); |
| 7806 |
} |
| 7807 |
} |
| 7808 |
} else { |
| 7809 |
$failed_ids[] = $entry_id; |
| 7810 |
$errors[] = "Database error for entry: $entry_id"; |
| 7811 |
} |
| 7812 |
} catch (Exception $e) { |
| 7813 |
$failed_ids[] = $entry_id; |
| 7814 |
$errors[] = $e->getMessage(); |
| 7815 |
} |
| 7816 |
} |
| 7817 |
|
| 7818 |
wp_send_json_success(array( |
| 7819 |
'success_ids' => $success_ids, |
| 7820 |
'failed_ids' => $failed_ids, |
| 7821 |
'errors' => $errors, |
| 7822 |
'total_processed' => count($success_ids) + count($failed_ids) |
| 7823 |
)); |
| 7824 |
|
| 7825 |
exit; |
| 7826 |
} |
| 7827 |
|
| 7828 |
/** |
| 7829 |
* Get hierarchical roles for dropdown |
| 7830 |
*/ |
| 7831 |
public function mxchat_get_role_options() { |
| 7832 |
return array( |
| 7833 |
'public' => __('Public (Everyone)', 'mxchat'), |
| 7834 |
'logged_in' => __('Logged In Users', 'mxchat'), |
| 7835 |
'subscriber' => __('Subscribers & Above', 'mxchat'), |
| 7836 |
'contributor' => __('Contributors & Above', 'mxchat'), |
| 7837 |
'author' => __('Authors & Above', 'mxchat'), |
| 7838 |
'editor' => __('Editors & Above', 'mxchat'), |
| 7839 |
'administrator' => __('Administrators Only', 'mxchat') |
| 7840 |
); |
| 7841 |
} |
| 7842 |
|
| 7843 |
/** |
| 7844 |
* Check if user has access to content based on role restriction |
| 7845 |
*/ |
| 7846 |
public function mxchat_user_has_content_access($role_restriction) { |
| 7847 |
// Public content is always accessible |
| 7848 |
if ($role_restriction === 'public' || empty($role_restriction)) { |
| 7849 |
return true; |
| 7850 |
} |
| 7851 |
|
| 7852 |
// Check if user is logged in for logged_in restriction |
| 7853 |
if ($role_restriction === 'logged_in') { |
| 7854 |
return is_user_logged_in(); |
| 7855 |
} |
| 7856 |
|
| 7857 |
// If not logged in, no access to role-restricted content |
| 7858 |
if (!is_user_logged_in()) { |
| 7859 |
return false; |
| 7860 |
} |
| 7861 |
|
| 7862 |
$user = wp_get_current_user(); |
| 7863 |
$user_roles = $user->roles; |
| 7864 |
|
| 7865 |
if (empty($user_roles)) { |
| 7866 |
return false; |
| 7867 |
} |
| 7868 |
|
| 7869 |
// Define role hierarchy (higher number = higher access) |
| 7870 |
$hierarchy = array( |
| 7871 |
'subscriber' => 1, |
| 7872 |
'contributor' => 2, |
| 7873 |
'author' => 3, |
| 7874 |
'editor' => 4, |
| 7875 |
'administrator' => 5 |
| 7876 |
); |
| 7877 |
|
| 7878 |
// Get required level |
| 7879 |
$required_level = isset($hierarchy[$role_restriction]) ? $hierarchy[$role_restriction] : 0; |
| 7880 |
|
| 7881 |
// Check if user has required level or higher |
| 7882 |
foreach ($user_roles as $user_role) { |
| 7883 |
$user_level = isset($hierarchy[$user_role]) ? $hierarchy[$user_role] : 0; |
| 7884 |
if ($user_level >= $required_level) { |
| 7885 |
return true; |
| 7886 |
} |
| 7887 |
} |
| 7888 |
|
| 7889 |
return false; |
| 7890 |
} |
| 7891 |
|
| 7892 |
/** |
| 7893 |
* Handle role restriction updates via AJAX |
| 7894 |
* Removed cache clearing call since we removed caching |
| 7895 |
*/ |
| 7896 |
public function ajax_mxchat_update_role_restriction() { |
| 7897 |
// Verify nonce and permissions |
| 7898 |
if (!check_ajax_referer('mxchat_update_role_nonce', 'nonce', false)) { |
| 7899 |
wp_send_json_error('Invalid nonce'); |
| 7900 |
exit; |
| 7901 |
} |
| 7902 |
|
| 7903 |
if (!current_user_can('manage_options')) { |
| 7904 |
wp_send_json_error('Unauthorized access'); |
| 7905 |
exit; |
| 7906 |
} |
| 7907 |
|
| 7908 |
$entry_id = isset($_POST['entry_id']) ? sanitize_text_field($_POST['entry_id']) : ''; |
| 7909 |
$role_restriction = isset($_POST['role_restriction']) ? sanitize_text_field($_POST['role_restriction']) : 'public'; |
| 7910 |
$data_source = isset($_POST['data_source']) ? sanitize_text_field($_POST['data_source']) : 'wordpress'; |
| 7911 |
|
| 7912 |
if (empty($entry_id)) { |
| 7913 |
wp_send_json_error('Invalid entry ID'); |
| 7914 |
exit; |
| 7915 |
} |
| 7916 |
|
| 7917 |
// Get knowledge manager instance to validate role restriction |
| 7918 |
$knowledge_manager = MxChat_Knowledge_Manager::get_instance(); |
| 7919 |
$valid_roles = array_keys($knowledge_manager->mxchat_get_role_options()); |
| 7920 |
if (!in_array($role_restriction, $valid_roles)) { |
| 7921 |
wp_send_json_error('Invalid role restriction'); |
| 7922 |
exit; |
| 7923 |
} |
| 7924 |
|
| 7925 |
global $wpdb; |
| 7926 |
|
| 7927 |
if ($data_source === 'pinecone') { |
| 7928 |
// Handle Pinecone role restriction (stored separately in WordPress table) |
| 7929 |
$roles_table = $wpdb->prefix . 'mxchat_pinecone_roles'; |
| 7930 |
|
| 7931 |
// Use REPLACE to insert or update the role restriction |
| 7932 |
$result = $wpdb->replace( |
| 7933 |
$roles_table, |
| 7934 |
array( |
| 7935 |
'vector_id' => $entry_id, |
| 7936 |
'role_restriction' => $role_restriction, |
| 7937 |
'updated_at' => current_time('mysql') |
| 7938 |
), |
| 7939 |
array('%s', '%s', '%s') |
| 7940 |
); |
| 7941 |
|
| 7942 |
// No cache clearing needed since we removed caching |
| 7943 |
|
| 7944 |
} else { |
| 7945 |
// Handle WordPress database role restriction (existing functionality) |
| 7946 |
$table_name = $wpdb->prefix . 'mxchat_system_prompt_content'; |
| 7947 |
|
| 7948 |
$result = $wpdb->update( |
| 7949 |
$table_name, |
| 7950 |
array('role_restriction' => $role_restriction), |
| 7951 |
array('id' => absint($entry_id)), |
| 7952 |
array('%s'), |
| 7953 |
array('%d') |
| 7954 |
); |
| 7955 |
} |
| 7956 |
|
| 7957 |
if ($result === false) { |
| 7958 |
wp_send_json_error('Database update failed: ' . $wpdb->last_error); |
| 7959 |
exit; |
| 7960 |
} |
| 7961 |
|
| 7962 |
// Keep the OpenAI Vector Store mirror consistent with the new restriction |
| 7963 |
// (plan 15b5c6): non-public pulls the mirrored file, public re-mirrors. |
| 7964 |
if (class_exists('MxChat_Vectorstore_Manager')) { |
| 7965 |
if ($data_source === 'pinecone') { |
| 7966 |
if ($role_restriction !== 'public') { |
| 7967 |
MxChat_Vectorstore_Manager::sync_delete_by_key($entry_id, 'default'); |
| 7968 |
} |
| 7969 |
// Public again: Pinecone-mode content isn't held locally, so the |
| 7970 |
// entry re-mirrors on its next save/import rather than here. |
| 7971 |
} else { |
| 7972 |
$row_url = $wpdb->get_var($wpdb->prepare( |
| 7973 |
"SELECT source_url FROM {$wpdb->prefix}mxchat_system_prompt_content WHERE id = %d", |
| 7974 |
absint($entry_id) |
| 7975 |
)); |
| 7976 |
if (!empty($row_url)) { |
| 7977 |
MxChat_Vectorstore_Manager::handle_role_change($row_url, 'default', $role_restriction); |
| 7978 |
} |
| 7979 |
} |
| 7980 |
} |
| 7981 |
|
| 7982 |
wp_send_json_success(array( |
| 7983 |
'message' => 'Role restriction updated successfully', |
| 7984 |
'role_restriction' => $role_restriction, |
| 7985 |
'data_source' => $data_source, |
| 7986 |
'entry_id' => $entry_id |
| 7987 |
)); |
| 7988 |
exit; |
| 7989 |
} |
| 7990 |
|
| 7991 |
// ======================================== |
| 7992 |
// ROLE-BASED CONTENT RESTRICTIONS - NEW FUNCTIONS |
| 7993 |
// Add these to your MxChat_Knowledge_Manager class |
| 7994 |
// ======================================== |
| 7995 |
|
| 7996 |
/** |
| 7997 |
* Initialize role-based content hooks |
| 7998 |
* Add this call to your __construct() or mxchat_init_hooks() method |
| 7999 |
*/ |
| 8000 |
private function mxchat_init_role_hooks() { |
| 8001 |
// AJAX handlers for tag-role mappings |
| 8002 |
add_action('wp_ajax_mxchat_add_tag_role_mapping', array($this, 'ajax_add_tag_role_mapping')); |
| 8003 |
add_action('wp_ajax_mxchat_delete_tag_role_mapping', array($this, 'ajax_delete_tag_role_mapping')); |
| 8004 |
add_action('wp_ajax_mxchat_get_tag_role_mappings', array($this, 'ajax_get_tag_role_mappings')); |
| 8005 |
add_action('wp_ajax_mxchat_bulk_update_tag_roles', array($this, 'ajax_bulk_update_tag_roles')); |
| 8006 |
|
| 8007 |
// Hook to automatically update role restrictions when tags are added/removed |
| 8008 |
add_action('set_object_terms', array($this, 'handle_tag_change'), 10, 6); |
| 8009 |
|
| 8010 |
// Hook to apply role restrictions on auto-sync |
| 8011 |
add_action('mxchat_content_stored', array($this, 'apply_role_restriction_after_storage'), 10, 2); |
| 8012 |
} |
| 8013 |
|
| 8014 |
/** |
| 8015 |
* Add tag-role mapping via AJAX |
| 8016 |
*/ |
| 8017 |
public function ajax_add_tag_role_mapping() { |
| 8018 |
// Verify nonce and permissions |
| 8019 |
check_ajax_referer('mxchat_prompts_setting_nonce', 'nonce'); |
| 8020 |
|
| 8021 |
if (!current_user_can('manage_options')) { |
| 8022 |
wp_send_json_error('Unauthorized access'); |
| 8023 |
exit; |
| 8024 |
} |
| 8025 |
|
| 8026 |
$tag_input = isset($_POST['tag_slug']) ? sanitize_text_field($_POST['tag_slug']) : ''; |
| 8027 |
$role_restriction = isset($_POST['role_restriction']) ? sanitize_text_field($_POST['role_restriction']) : 'public'; |
| 8028 |
|
| 8029 |
if (empty($tag_input)) { |
| 8030 |
wp_send_json_error('Please enter a tag name or slug'); |
| 8031 |
exit; |
| 8032 |
} |
| 8033 |
|
| 8034 |
// Validate role restriction |
| 8035 |
$valid_roles = array_keys($this->mxchat_get_role_options()); |
| 8036 |
if (!in_array($role_restriction, $valid_roles)) { |
| 8037 |
wp_send_json_error('Invalid role restriction'); |
| 8038 |
exit; |
| 8039 |
} |
| 8040 |
|
| 8041 |
// Resolve the tag by slug first, then fall back to its display name, so users can |
| 8042 |
// enter either "premium-content" or "Premium Content". (plan b8bcf5 — the field is |
| 8043 |
// labeled by name but previously validated by slug only, producing the confusing |
| 8044 |
// "Tag does not exist in WordPress" error when a real tag's name was typed.) |
| 8045 |
$term = get_term_by('slug', $tag_input, 'post_tag'); |
| 8046 |
if (!$term) { |
| 8047 |
$term = get_term_by('name', $tag_input, 'post_tag'); |
| 8048 |
} |
| 8049 |
if (!$term) { |
| 8050 |
wp_send_json_error('No tag with that name or slug exists yet. Create it under Posts → Tags first, then enter its name or slug.'); |
| 8051 |
exit; |
| 8052 |
} |
| 8053 |
|
| 8054 |
// Always key the mapping by the RESOLVED slug — apply_role_restriction_to_post() |
| 8055 |
// compares against each post's tag slugs, so the stored key must be a slug, |
| 8056 |
// never the raw (possibly display-name) input. |
| 8057 |
$tag_slug = $term->slug; |
| 8058 |
|
| 8059 |
// Get existing mappings |
| 8060 |
$mappings = get_option('mxchat_tag_role_mappings', array()); |
| 8061 |
|
| 8062 |
// Check if mapping already exists |
| 8063 |
if (isset($mappings[$tag_slug])) { |
| 8064 |
wp_send_json_error('Mapping for this tag already exists'); |
| 8065 |
exit; |
| 8066 |
} |
| 8067 |
|
| 8068 |
// Add new mapping |
| 8069 |
$mappings[$tag_slug] = $role_restriction; |
| 8070 |
update_option('mxchat_tag_role_mappings', $mappings); |
| 8071 |
|
| 8072 |
wp_send_json_success(array( |
| 8073 |
'message' => 'Tag-role mapping added successfully', |
| 8074 |
'tag_slug' => $tag_slug, |
| 8075 |
'role_restriction' => $role_restriction |
| 8076 |
)); |
| 8077 |
exit; |
| 8078 |
} |
| 8079 |
|
| 8080 |
/** |
| 8081 |
* Delete tag-role mapping via AJAX |
| 8082 |
*/ |
| 8083 |
public function ajax_delete_tag_role_mapping() { |
| 8084 |
// Verify nonce and permissions |
| 8085 |
check_ajax_referer('mxchat_prompts_setting_nonce', 'nonce'); |
| 8086 |
|
| 8087 |
if (!current_user_can('manage_options')) { |
| 8088 |
wp_send_json_error('Unauthorized access'); |
| 8089 |
exit; |
| 8090 |
} |
| 8091 |
|
| 8092 |
$tag_slug = isset($_POST['tag_slug']) ? sanitize_text_field($_POST['tag_slug']) : ''; |
| 8093 |
|
| 8094 |
if (empty($tag_slug)) { |
| 8095 |
wp_send_json_error('Tag slug is required'); |
| 8096 |
exit; |
| 8097 |
} |
| 8098 |
|
| 8099 |
// Get existing mappings |
| 8100 |
$mappings = get_option('mxchat_tag_role_mappings', array()); |
| 8101 |
|
| 8102 |
// Check if mapping exists |
| 8103 |
if (!isset($mappings[$tag_slug])) { |
| 8104 |
wp_send_json_error('Mapping does not exist'); |
| 8105 |
exit; |
| 8106 |
} |
| 8107 |
|
| 8108 |
// Remove mapping |
| 8109 |
unset($mappings[$tag_slug]); |
| 8110 |
update_option('mxchat_tag_role_mappings', $mappings); |
| 8111 |
|
| 8112 |
wp_send_json_success(array( |
| 8113 |
'message' => 'Tag-role mapping deleted successfully', |
| 8114 |
'tag_slug' => $tag_slug |
| 8115 |
)); |
| 8116 |
exit; |
| 8117 |
} |
| 8118 |
|
| 8119 |
/** |
| 8120 |
* Get all tag-role mappings via AJAX |
| 8121 |
*/ |
| 8122 |
public function ajax_get_tag_role_mappings() { |
| 8123 |
// Verify nonce and permissions |
| 8124 |
check_ajax_referer('mxchat_prompts_setting_nonce', 'nonce'); |
| 8125 |
|
| 8126 |
if (!current_user_can('manage_options')) { |
| 8127 |
wp_send_json_error('Unauthorized access'); |
| 8128 |
exit; |
| 8129 |
} |
| 8130 |
|
| 8131 |
// Get mappings |
| 8132 |
$mappings = get_option('mxchat_tag_role_mappings', array()); |
| 8133 |
$role_options = $this->mxchat_get_role_options(); |
| 8134 |
|
| 8135 |
$formatted_mappings = array(); |
| 8136 |
|
| 8137 |
foreach ($mappings as $tag_slug => $role_restriction) { |
| 8138 |
// Get tag object |
| 8139 |
$term = get_term_by('slug', $tag_slug, 'post_tag'); |
| 8140 |
|
| 8141 |
// Count posts with this tag |
| 8142 |
$post_count = 0; |
| 8143 |
if ($term) { |
| 8144 |
$post_count = $term->count; |
| 8145 |
} |
| 8146 |
|
| 8147 |
$formatted_mappings[] = array( |
| 8148 |
'tag_slug' => $tag_slug, |
| 8149 |
'role_restriction' => $role_restriction, |
| 8150 |
'role_label' => $role_options[$role_restriction] ?? $role_restriction, |
| 8151 |
'post_count' => $post_count |
| 8152 |
); |
| 8153 |
} |
| 8154 |
|
| 8155 |
wp_send_json_success(array( |
| 8156 |
'mappings' => $formatted_mappings |
| 8157 |
)); |
| 8158 |
exit; |
| 8159 |
} |
| 8160 |
|
| 8161 |
/** |
| 8162 |
* Bulk update role restrictions for all existing content with mapped tags |
| 8163 |
*/ |
| 8164 |
public function ajax_bulk_update_tag_roles() { |
| 8165 |
// Verify nonce and permissions |
| 8166 |
check_ajax_referer('mxchat_prompts_setting_nonce', 'nonce'); |
| 8167 |
|
| 8168 |
if (!current_user_can('manage_options')) { |
| 8169 |
wp_send_json_error('Unauthorized access'); |
| 8170 |
exit; |
| 8171 |
} |
| 8172 |
|
| 8173 |
// Get mappings |
| 8174 |
$mappings = get_option('mxchat_tag_role_mappings', array()); |
| 8175 |
|
| 8176 |
if (empty($mappings)) { |
| 8177 |
wp_send_json_error('No tag-role mappings found'); |
| 8178 |
exit; |
| 8179 |
} |
| 8180 |
|
| 8181 |
global $wpdb; |
| 8182 |
|
| 8183 |
// Check if using Pinecone |
| 8184 |
$pinecone_options = get_option('mxchat_pinecone_addon_options', array()); |
| 8185 |
$use_pinecone = ($pinecone_options['mxchat_use_pinecone'] ?? '0') === '1'; |
| 8186 |
|
| 8187 |
$updated_count = 0; |
| 8188 |
$details = array(); |
| 8189 |
|
| 8190 |
foreach ($mappings as $tag_slug => $role_restriction) { |
| 8191 |
// Get all posts with this tag |
| 8192 |
$posts = get_posts(array( |
| 8193 |
'tag' => $tag_slug, |
| 8194 |
'post_type' => 'any', |
| 8195 |
'posts_per_page' => -1, |
| 8196 |
'fields' => 'ids', |
| 8197 |
'post_status' => 'publish' |
| 8198 |
)); |
| 8199 |
|
| 8200 |
if (empty($posts)) { |
| 8201 |
continue; |
| 8202 |
} |
| 8203 |
|
| 8204 |
$tag_updated = 0; |
| 8205 |
|
| 8206 |
foreach ($posts as $post_id) { |
| 8207 |
$source_url = get_permalink($post_id); |
| 8208 |
if (!$source_url) { |
| 8209 |
continue; |
| 8210 |
} |
| 8211 |
|
| 8212 |
if ($use_pinecone && !empty($pinecone_options['mxchat_pinecone_api_key'])) { |
| 8213 |
// Update Pinecone role restriction |
| 8214 |
$roles_table = $wpdb->prefix . 'mxchat_pinecone_roles'; |
| 8215 |
$vector_id = md5($source_url); |
| 8216 |
|
| 8217 |
$result = $wpdb->replace( |
| 8218 |
$roles_table, |
| 8219 |
array( |
| 8220 |
'vector_id' => $vector_id, |
| 8221 |
'role_restriction' => $role_restriction, |
| 8222 |
'updated_at' => current_time('mysql') |
| 8223 |
), |
| 8224 |
array('%s', '%s', '%s') |
| 8225 |
); |
| 8226 |
} else { |
| 8227 |
// Update WordPress DB |
| 8228 |
$table_name = $wpdb->prefix . 'mxchat_system_prompt_content'; |
| 8229 |
|
| 8230 |
$result = $wpdb->update( |
| 8231 |
$table_name, |
| 8232 |
array('role_restriction' => $role_restriction), |
| 8233 |
array('source_url' => $source_url), |
| 8234 |
array('%s'), |
| 8235 |
array('%s') |
| 8236 |
); |
| 8237 |
} |
| 8238 |
|
| 8239 |
if ($result !== false) { |
| 8240 |
$tag_updated++; |
| 8241 |
$updated_count++; |
| 8242 |
} |
| 8243 |
} |
| 8244 |
|
| 8245 |
if ($tag_updated > 0) { |
| 8246 |
$details[] = sprintf( |
| 8247 |
'Tag "%s" (%s): %d posts updated', |
| 8248 |
$tag_slug, |
| 8249 |
$role_restriction, |
| 8250 |
$tag_updated |
| 8251 |
); |
| 8252 |
} |
| 8253 |
} |
| 8254 |
|
| 8255 |
wp_send_json_success(array( |
| 8256 |
'message' => 'Bulk update completed', |
| 8257 |
'updated_count' => $updated_count, |
| 8258 |
'tags_processed' => count($mappings), |
| 8259 |
'details' => $details |
| 8260 |
)); |
| 8261 |
exit; |
| 8262 |
} |
| 8263 |
|
| 8264 |
/** |
| 8265 |
* Handle tag changes on posts (when tags are added or removed) |
| 8266 |
*/ |
| 8267 |
public function handle_tag_change($object_id, $terms, $tt_ids, $taxonomy, $append, $old_tt_ids) { |
| 8268 |
// Only process post tags |
| 8269 |
if ($taxonomy !== 'post_tag') { |
| 8270 |
return; |
| 8271 |
} |
| 8272 |
|
| 8273 |
// Get tag-role mappings |
| 8274 |
$mappings = get_option('mxchat_tag_role_mappings', array()); |
| 8275 |
|
| 8276 |
if (empty($mappings)) { |
| 8277 |
return; |
| 8278 |
} |
| 8279 |
|
| 8280 |
// Get the post's URL |
| 8281 |
$source_url = get_permalink($object_id); |
| 8282 |
if (!$source_url) { |
| 8283 |
return; |
| 8284 |
} |
| 8285 |
|
| 8286 |
// Determine the highest role restriction based on tags |
| 8287 |
$highest_role = 'public'; |
| 8288 |
$role_hierarchy = array( |
| 8289 |
'public' => 0, |
| 8290 |
'logged_in' => 1, |
| 8291 |
'subscriber' => 2, |
| 8292 |
'contributor' => 3, |
| 8293 |
'author' => 4, |
| 8294 |
'editor' => 5, |
| 8295 |
'administrator' => 6 |
| 8296 |
); |
| 8297 |
|
| 8298 |
// Get all current tags for the post |
| 8299 |
$current_tags = wp_get_post_tags($object_id, array('fields' => 'slugs')); |
| 8300 |
|
| 8301 |
// Find the highest role restriction among the tags |
| 8302 |
foreach ($current_tags as $tag_slug) { |
| 8303 |
if (isset($mappings[$tag_slug])) { |
| 8304 |
$role = $mappings[$tag_slug]; |
| 8305 |
if (isset($role_hierarchy[$role]) && $role_hierarchy[$role] > $role_hierarchy[$highest_role]) { |
| 8306 |
$highest_role = $role; |
| 8307 |
} |
| 8308 |
} |
| 8309 |
} |
| 8310 |
|
| 8311 |
// Update the role restriction in the database |
| 8312 |
global $wpdb; |
| 8313 |
|
| 8314 |
// Check if using Pinecone |
| 8315 |
$pinecone_options = get_option('mxchat_pinecone_addon_options', array()); |
| 8316 |
$use_pinecone = ($pinecone_options['mxchat_use_pinecone'] ?? '0') === '1'; |
| 8317 |
|
| 8318 |
if ($use_pinecone && !empty($pinecone_options['mxchat_pinecone_api_key'])) { |
| 8319 |
// Update Pinecone role restriction |
| 8320 |
$roles_table = $wpdb->prefix . 'mxchat_pinecone_roles'; |
| 8321 |
$vector_id = md5($source_url); |
| 8322 |
|
| 8323 |
$wpdb->replace( |
| 8324 |
$roles_table, |
| 8325 |
array( |
| 8326 |
'vector_id' => $vector_id, |
| 8327 |
'role_restriction' => $highest_role, |
| 8328 |
'updated_at' => current_time('mysql') |
| 8329 |
), |
| 8330 |
array('%s', '%s', '%s') |
| 8331 |
); |
| 8332 |
} else { |
| 8333 |
// Update WordPress DB |
| 8334 |
$table_name = $wpdb->prefix . 'mxchat_system_prompt_content'; |
| 8335 |
|
| 8336 |
$wpdb->update( |
| 8337 |
$table_name, |
| 8338 |
array('role_restriction' => $highest_role), |
| 8339 |
array('source_url' => $source_url), |
| 8340 |
array('%s'), |
| 8341 |
array('%s') |
| 8342 |
); |
| 8343 |
} |
| 8344 |
|
| 8345 |
// The entry's restriction just changed — keep the OpenAI Vector Store |
| 8346 |
// mirror consistent: non-public pulls the file (file_search has no |
| 8347 |
// per-role filtering), public re-mirrors it (plan 15b5c6). |
| 8348 |
if (class_exists('MxChat_Vectorstore_Manager')) { |
| 8349 |
MxChat_Vectorstore_Manager::handle_role_change($source_url, 'default', $highest_role); |
| 8350 |
} |
| 8351 |
} |
| 8352 |
|
| 8353 |
/** |
| 8354 |
* Apply role restriction after content is stored (for auto-sync) |
| 8355 |
*/ |
| 8356 |
public function apply_role_restriction_after_storage($post_id, $source_url) { |
| 8357 |
// Get tag-role mappings |
| 8358 |
$mappings = get_option('mxchat_tag_role_mappings', array()); |
| 8359 |
|
| 8360 |
if (empty($mappings)) { |
| 8361 |
return; |
| 8362 |
} |
| 8363 |
|
| 8364 |
// Get all tags for the post |
| 8365 |
$post_tags = wp_get_post_tags($post_id, array('fields' => 'slugs')); |
| 8366 |
|
| 8367 |
if (empty($post_tags)) { |
| 8368 |
return; |
| 8369 |
} |
| 8370 |
|
| 8371 |
// Determine the highest role restriction based on tags |
| 8372 |
$highest_role = 'public'; |
| 8373 |
$role_hierarchy = array( |
| 8374 |
'public' => 0, |
| 8375 |
'logged_in' => 1, |
| 8376 |
'subscriber' => 2, |
| 8377 |
'contributor' => 3, |
| 8378 |
'author' => 4, |
| 8379 |
'editor' => 5, |
| 8380 |
'administrator' => 6 |
| 8381 |
); |
| 8382 |
|
| 8383 |
foreach ($post_tags as $tag_slug) { |
| 8384 |
if (isset($mappings[$tag_slug])) { |
| 8385 |
$role = $mappings[$tag_slug]; |
| 8386 |
if (isset($role_hierarchy[$role]) && $role_hierarchy[$role] > $role_hierarchy[$highest_role]) { |
| 8387 |
$highest_role = $role; |
| 8388 |
} |
| 8389 |
} |
| 8390 |
} |
| 8391 |
|
| 8392 |
// If no restricted tags found, return (leave as public) |
| 8393 |
if ($highest_role === 'public') { |
| 8394 |
return; |
| 8395 |
} |
| 8396 |
|
| 8397 |
// Update the role restriction |
| 8398 |
global $wpdb; |
| 8399 |
|
| 8400 |
// Check if using Pinecone |
| 8401 |
$pinecone_options = get_option('mxchat_pinecone_addon_options', array()); |
| 8402 |
$use_pinecone = ($pinecone_options['mxchat_use_pinecone'] ?? '0') === '1'; |
| 8403 |
|
| 8404 |
if ($use_pinecone && !empty($pinecone_options['mxchat_pinecone_api_key'])) { |
| 8405 |
// Update Pinecone role restriction |
| 8406 |
$roles_table = $wpdb->prefix . 'mxchat_pinecone_roles'; |
| 8407 |
$vector_id = md5($source_url); |
| 8408 |
|
| 8409 |
$wpdb->replace( |
| 8410 |
$roles_table, |
| 8411 |
array( |
| 8412 |
'vector_id' => $vector_id, |
| 8413 |
'role_restriction' => $highest_role, |
| 8414 |
'updated_at' => current_time('mysql') |
| 8415 |
), |
| 8416 |
array('%s', '%s', '%s') |
| 8417 |
); |
| 8418 |
} else { |
| 8419 |
// Update WordPress DB |
| 8420 |
$table_name = $wpdb->prefix . 'mxchat_system_prompt_content'; |
| 8421 |
|
| 8422 |
$wpdb->update( |
| 8423 |
$table_name, |
| 8424 |
array('role_restriction' => $highest_role), |
| 8425 |
array('source_url' => $source_url), |
| 8426 |
array('%s'), |
| 8427 |
array('%s') |
| 8428 |
); |
| 8429 |
} |
| 8430 |
|
| 8431 |
// The entry's restriction just changed — keep the OpenAI Vector Store |
| 8432 |
// mirror consistent: non-public pulls the file (file_search has no |
| 8433 |
// per-role filtering), public re-mirrors it (plan 15b5c6). |
| 8434 |
if (class_exists('MxChat_Vectorstore_Manager')) { |
| 8435 |
MxChat_Vectorstore_Manager::handle_role_change($source_url, 'default', $highest_role); |
| 8436 |
} |
| 8437 |
} |
| 8438 |
|
| 8439 |
|
| 8440 |
// ======================================== |
| 8441 |
// HELPER METHODS |
| 8442 |
// ======================================== |
| 8443 |
|
| 8444 |
/** |
| 8445 |
* Check if user has required permissions for content processing |
| 8446 |
*/ |
| 8447 |
private function mxchat_check_user_permissions() { |
| 8448 |
if (!current_user_can('manage_options')) { |
| 8449 |
wp_die(esc_html__('You do not have sufficient permissions.', 'mxchat')); |
| 8450 |
} |
| 8451 |
} |
| 8452 |
|
| 8453 |
/** |
| 8454 |
* Validate nonce for security |
| 8455 |
*/ |
| 8456 |
private function mxchat_validate_nonce($nonce_name, $nonce_action) { |
| 8457 |
if (!isset($_POST[$nonce_name]) || !wp_verify_nonce($_POST[$nonce_name], $nonce_action)) { |
| 8458 |
wp_die(esc_html__('Security check failed.', 'mxchat')); |
| 8459 |
} |
| 8460 |
} |
| 8461 |
|
| 8462 |
/** |
| 8463 |
* Get embedding API credentials |
| 8464 |
*/ |
| 8465 |
private function mxchat_get_embedding_credentials() { |
| 8466 |
$embedding_model = $this->options['embedding_model'] ?? 'text-embedding-ada-002'; |
| 8467 |
|
| 8468 |
if (strpos($embedding_model, 'text-embedding-') !== false) { |
| 8469 |
return array( |
| 8470 |
'type' => 'openai', |
| 8471 |
'api_key' => $this->options['api_key'] ?? '' |
| 8472 |
); |
| 8473 |
} elseif (strpos($embedding_model, 'voyage-') !== false) { |
| 8474 |
return array( |
| 8475 |
'type' => 'voyage', |
| 8476 |
'api_key' => $this->options['voyage_api_key'] ?? '' |
| 8477 |
); |
| 8478 |
} elseif (strpos($embedding_model, 'gemini-embedding-') !== false) { |
| 8479 |
return array( |
| 8480 |
'type' => 'gemini', |
| 8481 |
'api_key' => $this->options['gemini_api_key'] ?? '' |
| 8482 |
); |
| 8483 |
} |
| 8484 |
|
| 8485 |
return array('type' => 'unknown', 'api_key' => ''); |
| 8486 |
} |
| 8487 |
|
| 8488 |
/** |
| 8489 |
* Log processing errors |
| 8490 |
*/ |
| 8491 |
private function mxchat_log_processing_error($operation, $error_message) { |
| 8492 |
//error_log("MxChat Knowledge Processing {$operation} Error: " . $error_message); |
| 8493 |
} |
| 8494 |
|
| 8495 |
/** |
| 8496 |
* Set admin notice transient |
| 8497 |
*/ |
| 8498 |
private function mxchat_set_admin_notice($type, $message) { |
| 8499 |
set_transient("mxchat_admin_notice_{$type}", $message, 30); |
| 8500 |
} |
| 8501 |
|
| 8502 |
/** |
| 8503 |
* Get Pinecone manager instance for vector operations |
| 8504 |
*/ |
| 8505 |
private function mxchat_get_pinecone_manager() { |
| 8506 |
return MxChat_Pinecone_Manager::get_instance(); |
| 8507 |
} |
| 8508 |
|
| 8509 |
|
| 8510 |
// ======================================== |
| 8511 |
// DATABASE QUEUE TABLE MANAGEMENT |
| 8512 |
// ======================================== |
| 8513 |
|
| 8514 |
/** |
| 8515 |
* Create queue table on plugin activation |
| 8516 |
* Call this from your plugin activation hook |
| 8517 |
*/ |
| 8518 |
public function mxchat_create_queue_table() { |
| 8519 |
global $wpdb; |
| 8520 |
|
| 8521 |
$table_name = $wpdb->prefix . 'mxchat_processing_queue'; |
| 8522 |
$charset_collate = $wpdb->get_charset_collate(); |
| 8523 |
|
| 8524 |
$sql = "CREATE TABLE IF NOT EXISTS $table_name ( |
| 8525 |
id bigint(20) unsigned NOT NULL AUTO_INCREMENT, |
| 8526 |
queue_id varchar(64) NOT NULL, |
| 8527 |
item_type varchar(20) NOT NULL, |
| 8528 |
item_data longtext NOT NULL, |
| 8529 |
status varchar(20) NOT NULL DEFAULT 'pending', |
| 8530 |
bot_id varchar(50) NOT NULL DEFAULT 'default', |
| 8531 |
priority int(11) NOT NULL DEFAULT 0, |
| 8532 |
attempts int(11) NOT NULL DEFAULT 0, |
| 8533 |
max_attempts int(11) NOT NULL DEFAULT 3, |
| 8534 |
error_message text DEFAULT NULL, |
| 8535 |
created_at datetime NOT NULL, |
| 8536 |
started_at datetime DEFAULT NULL, |
| 8537 |
completed_at datetime DEFAULT NULL, |
| 8538 |
PRIMARY KEY (id), |
| 8539 |
KEY queue_id (queue_id), |
| 8540 |
KEY status (status), |
| 8541 |
KEY item_type (item_type), |
| 8542 |
KEY priority (priority) |
| 8543 |
) $charset_collate;"; |
| 8544 |
|
| 8545 |
require_once(ABSPATH . 'wp-admin/includes/upgrade.php'); |
| 8546 |
dbDelta($sql); |
| 8547 |
|
| 8548 |
// Also create a meta table for queue metadata |
| 8549 |
$meta_table = $wpdb->prefix . 'mxchat_queue_meta'; |
| 8550 |
|
| 8551 |
$meta_sql = "CREATE TABLE IF NOT EXISTS $meta_table ( |
| 8552 |
id bigint(20) unsigned NOT NULL AUTO_INCREMENT, |
| 8553 |
queue_id varchar(64) NOT NULL, |
| 8554 |
meta_key varchar(255) NOT NULL, |
| 8555 |
meta_value longtext, |
| 8556 |
PRIMARY KEY (id), |
| 8557 |
KEY queue_id (queue_id), |
| 8558 |
KEY meta_key (meta_key) |
| 8559 |
) $charset_collate;"; |
| 8560 |
|
| 8561 |
dbDelta($meta_sql); |
| 8562 |
} |
| 8563 |
|
| 8564 |
/** |
| 8565 |
* Add items to the processing queue |
| 8566 |
* |
| 8567 |
* @param string $queue_id Unique identifier for this queue batch |
| 8568 |
* @param string $item_type Type of item (url, pdf_page) |
| 8569 |
* @param array $items Array of items to queue |
| 8570 |
* @param string $bot_id Bot ID for processing |
| 8571 |
* @return int Number of items queued |
| 8572 |
*/ |
| 8573 |
private function mxchat_add_to_queue($queue_id, $item_type, $items, $bot_id = 'default') { |
| 8574 |
global $wpdb; |
| 8575 |
$table_name = $wpdb->prefix . 'mxchat_processing_queue'; |
| 8576 |
|
| 8577 |
$queued_count = 0; |
| 8578 |
$priority = 0; |
| 8579 |
|
| 8580 |
foreach ($items as $item) { |
| 8581 |
$result = $wpdb->insert( |
| 8582 |
$table_name, |
| 8583 |
array( |
| 8584 |
'queue_id' => $queue_id, |
| 8585 |
'item_type' => $item_type, |
| 8586 |
'item_data' => wp_json_encode($item), |
| 8587 |
'status' => 'pending', |
| 8588 |
'bot_id' => $bot_id, |
| 8589 |
'priority' => $priority, |
| 8590 |
'attempts' => 0, |
| 8591 |
'max_attempts' => 3, |
| 8592 |
'created_at' => current_time('mysql') |
| 8593 |
), |
| 8594 |
array('%s', '%s', '%s', '%s', '%s', '%d', '%d', '%d', '%s') |
| 8595 |
); |
| 8596 |
|
| 8597 |
if ($result) { |
| 8598 |
$queued_count++; |
| 8599 |
} |
| 8600 |
|
| 8601 |
$priority++; // Process in order |
| 8602 |
} |
| 8603 |
|
| 8604 |
return $queued_count; |
| 8605 |
} |
| 8606 |
|
| 8607 |
/** |
| 8608 |
* Store queue metadata (total counts, source URL, etc.) |
| 8609 |
*/ |
| 8610 |
private function mxchat_set_queue_meta($queue_id, $meta_key, $meta_value) { |
| 8611 |
global $wpdb; |
| 8612 |
$meta_table = $wpdb->prefix . 'mxchat_queue_meta'; |
| 8613 |
|
| 8614 |
// Check if meta exists |
| 8615 |
$existing = $wpdb->get_var($wpdb->prepare( |
| 8616 |
"SELECT id FROM $meta_table WHERE queue_id = %s AND meta_key = %s", |
| 8617 |
$queue_id, |
| 8618 |
$meta_key |
| 8619 |
)); |
| 8620 |
|
| 8621 |
if ($existing) { |
| 8622 |
// Update |
| 8623 |
$wpdb->update( |
| 8624 |
$meta_table, |
| 8625 |
array('meta_value' => maybe_serialize($meta_value)), |
| 8626 |
array('queue_id' => $queue_id, 'meta_key' => $meta_key), |
| 8627 |
array('%s'), |
| 8628 |
array('%s', '%s') |
| 8629 |
); |
| 8630 |
} else { |
| 8631 |
// Insert |
| 8632 |
$wpdb->insert( |
| 8633 |
$meta_table, |
| 8634 |
array( |
| 8635 |
'queue_id' => $queue_id, |
| 8636 |
'meta_key' => $meta_key, |
| 8637 |
'meta_value' => maybe_serialize($meta_value) |
| 8638 |
), |
| 8639 |
array('%s', '%s', '%s') |
| 8640 |
); |
| 8641 |
} |
| 8642 |
} |
| 8643 |
|
| 8644 |
/** |
| 8645 |
* Get queue metadata |
| 8646 |
*/ |
| 8647 |
private function mxchat_get_queue_meta($queue_id, $meta_key) { |
| 8648 |
global $wpdb; |
| 8649 |
$meta_table = $wpdb->prefix . 'mxchat_queue_meta'; |
| 8650 |
|
| 8651 |
$value = $wpdb->get_var($wpdb->prepare( |
| 8652 |
"SELECT meta_value FROM $meta_table WHERE queue_id = %s AND meta_key = %s", |
| 8653 |
$queue_id, |
| 8654 |
$meta_key |
| 8655 |
)); |
| 8656 |
|
| 8657 |
return maybe_unserialize($value); |
| 8658 |
} |
| 8659 |
|
| 8660 |
// ======================================== |
| 8661 |
// AJAX QUEUE PROCESSING HANDLERS |
| 8662 |
// ======================================== |
| 8663 |
|
| 8664 |
/** |
| 8665 |
* AJAX: Get next item from queue to process |
| 8666 |
*/ |
| 8667 |
public function ajax_mxchat_get_next_queue_item() { |
| 8668 |
// Verify nonce and permissions |
| 8669 |
check_ajax_referer('mxchat_queue_nonce', 'nonce'); |
| 8670 |
|
| 8671 |
if (!current_user_can('manage_options')) { |
| 8672 |
wp_send_json_error('Unauthorized access'); |
| 8673 |
} |
| 8674 |
|
| 8675 |
$queue_id = isset($_POST['queue_id']) ? sanitize_text_field($_POST['queue_id']) : ''; |
| 8676 |
|
| 8677 |
if (empty($queue_id)) { |
| 8678 |
wp_send_json_error('Missing queue ID'); |
| 8679 |
} |
| 8680 |
|
| 8681 |
global $wpdb; |
| 8682 |
$table_name = $wpdb->prefix . 'mxchat_processing_queue'; |
| 8683 |
|
| 8684 |
// Get next pending item with retry logic for failed items |
| 8685 |
$next_item = $wpdb->get_row($wpdb->prepare( |
| 8686 |
"SELECT * FROM $table_name |
| 8687 |
WHERE queue_id = %s |
| 8688 |
AND status IN ('pending', 'failed') |
| 8689 |
AND attempts < max_attempts |
| 8690 |
ORDER BY priority ASC, id ASC |
| 8691 |
LIMIT 1", |
| 8692 |
$queue_id |
| 8693 |
)); |
| 8694 |
|
| 8695 |
if (!$next_item) { |
| 8696 |
// No more items - queue complete |
| 8697 |
wp_send_json_success(array( |
| 8698 |
'complete' => true, |
| 8699 |
'message' => 'Queue processing complete' |
| 8700 |
)); |
| 8701 |
} |
| 8702 |
|
| 8703 |
// Mark item as processing |
| 8704 |
$wpdb->update( |
| 8705 |
$table_name, |
| 8706 |
array( |
| 8707 |
'status' => 'processing', |
| 8708 |
'started_at' => current_time('mysql'), |
| 8709 |
'attempts' => $next_item->attempts + 1 |
| 8710 |
), |
| 8711 |
array('id' => $next_item->id), |
| 8712 |
array('%s', '%s', '%d'), |
| 8713 |
array('%d') |
| 8714 |
); |
| 8715 |
|
| 8716 |
wp_send_json_success(array( |
| 8717 |
'complete' => false, |
| 8718 |
'item' => array( |
| 8719 |
'id' => $next_item->id, |
| 8720 |
'type' => $next_item->item_type, |
| 8721 |
'data' => json_decode($next_item->item_data, true), |
| 8722 |
'bot_id' => $next_item->bot_id, |
| 8723 |
'attempt' => $next_item->attempts + 1 |
| 8724 |
) |
| 8725 |
)); |
| 8726 |
} |
| 8727 |
|
| 8728 |
/** |
| 8729 |
* AJAX: Process a single queue item |
| 8730 |
*/ |
| 8731 |
public function ajax_mxchat_process_queue_item() { |
| 8732 |
// Verify nonce and permissions |
| 8733 |
check_ajax_referer('mxchat_queue_nonce', 'nonce'); |
| 8734 |
|
| 8735 |
if (!current_user_can('manage_options')) { |
| 8736 |
wp_send_json_error('Unauthorized access'); |
| 8737 |
} |
| 8738 |
|
| 8739 |
$item_id = isset($_POST['item_id']) ? absint($_POST['item_id']) : 0; |
| 8740 |
$item_type = isset($_POST['item_type']) ? sanitize_text_field($_POST['item_type']) : ''; |
| 8741 |
$item_data = isset($_POST['item_data']) ? $_POST['item_data'] : array(); |
| 8742 |
$bot_id = isset($_POST['bot_id']) ? sanitize_key($_POST['bot_id']) : 'default'; |
| 8743 |
|
| 8744 |
if (empty($item_id) || empty($item_type)) { |
| 8745 |
wp_send_json_error('Missing item data'); |
| 8746 |
} |
| 8747 |
|
| 8748 |
global $wpdb; |
| 8749 |
$table_name = $wpdb->prefix . 'mxchat_processing_queue'; |
| 8750 |
|
| 8751 |
// Process based on item type |
| 8752 |
try { |
| 8753 |
set_time_limit(60); // Give processing 60 seconds |
| 8754 |
|
| 8755 |
$result = false; |
| 8756 |
$error_message = ''; |
| 8757 |
|
| 8758 |
// Read item directly from DB to get queue_id and preserve special chars in item_data |
| 8759 |
// (POST round-trip through JS mangles characters like apostrophes in URLs) |
| 8760 |
$db_item = $wpdb->get_row($wpdb->prepare( |
| 8761 |
"SELECT queue_id, item_data FROM $table_name WHERE id = %d", |
| 8762 |
$item_id |
| 8763 |
)); |
| 8764 |
$item_queue_id = $db_item ? $db_item->queue_id : ''; |
| 8765 |
if ($db_item && !empty($db_item->item_data)) { |
| 8766 |
$db_data = json_decode($db_item->item_data, true); |
| 8767 |
if (is_array($db_data)) { |
| 8768 |
$item_data = $db_data; |
| 8769 |
} |
| 8770 |
} |
| 8771 |
|
| 8772 |
switch ($item_type) { |
| 8773 |
case 'url': |
| 8774 |
$result = $this->mxchat_process_queue_url($item_data, $bot_id, $item_queue_id); |
| 8775 |
break; |
| 8776 |
|
| 8777 |
case 'pdf_page': |
| 8778 |
$result = $this->mxchat_process_queue_pdf_page($item_data, $bot_id); |
| 8779 |
break; |
| 8780 |
|
| 8781 |
default: |
| 8782 |
throw new Exception('Unknown item type: ' . $item_type); |
| 8783 |
} |
| 8784 |
|
| 8785 |
if (is_wp_error($result)) { |
| 8786 |
$error_code = $result->get_error_code(); |
| 8787 |
// Content errors (empty page, sanitization) are permanent — retrying won't help |
| 8788 |
$permanent_codes = array('empty_page', 'empty_after_sanitization', 'no_api_key', 'page_not_found'); |
| 8789 |
if (in_array($error_code, $permanent_codes)) { |
| 8790 |
// Mark as permanently failed — set attempts = max_attempts so it won't be retried |
| 8791 |
$current_item = $wpdb->get_row($wpdb->prepare( |
| 8792 |
"SELECT max_attempts FROM $table_name WHERE id = %d", $item_id |
| 8793 |
)); |
| 8794 |
$wpdb->update( |
| 8795 |
$table_name, |
| 8796 |
array( |
| 8797 |
'status' => 'failed', |
| 8798 |
'error_message' => $result->get_error_message(), |
| 8799 |
'attempts' => $current_item ? $current_item->max_attempts : 3 |
| 8800 |
), |
| 8801 |
array('id' => $item_id), |
| 8802 |
array('%s', '%s', '%d'), |
| 8803 |
array('%d') |
| 8804 |
); |
| 8805 |
wp_send_json_error(array( |
| 8806 |
'message' => $result->get_error_message(), |
| 8807 |
'permanent_failure' => true, |
| 8808 |
'item_id' => $item_id |
| 8809 |
)); |
| 8810 |
return; |
| 8811 |
} |
| 8812 |
throw new Exception($result->get_error_message()); |
| 8813 |
} |
| 8814 |
|
| 8815 |
if ($result === false) { |
| 8816 |
throw new Exception('Processing returned false - item may be empty or invalid'); |
| 8817 |
} |
| 8818 |
|
| 8819 |
// Mark as completed |
| 8820 |
$wpdb->update( |
| 8821 |
$table_name, |
| 8822 |
array( |
| 8823 |
'status' => 'completed', |
| 8824 |
'completed_at' => current_time('mysql'), |
| 8825 |
'error_message' => null |
| 8826 |
), |
| 8827 |
array('id' => $item_id), |
| 8828 |
array('%s', '%s', '%s'), |
| 8829 |
array('%d') |
| 8830 |
); |
| 8831 |
|
| 8832 |
wp_send_json_success(array( |
| 8833 |
'processed' => true, |
| 8834 |
'item_id' => $item_id, |
| 8835 |
'message' => 'Item processed successfully' |
| 8836 |
)); |
| 8837 |
|
| 8838 |
} catch (Exception $e) { |
| 8839 |
$error_message = $e->getMessage(); |
| 8840 |
|
| 8841 |
// Get current attempt count |
| 8842 |
$item = $wpdb->get_row($wpdb->prepare( |
| 8843 |
"SELECT attempts, max_attempts FROM $table_name WHERE id = %d", |
| 8844 |
$item_id |
| 8845 |
)); |
| 8846 |
|
| 8847 |
// Check if we've exhausted retries |
| 8848 |
if ($item && $item->attempts >= $item->max_attempts) { |
| 8849 |
// Permanently failed |
| 8850 |
$wpdb->update( |
| 8851 |
$table_name, |
| 8852 |
array( |
| 8853 |
'status' => 'failed', |
| 8854 |
'error_message' => $error_message |
| 8855 |
), |
| 8856 |
array('id' => $item_id), |
| 8857 |
array('%s', '%s'), |
| 8858 |
array('%d') |
| 8859 |
); |
| 8860 |
|
| 8861 |
wp_send_json_error(array( |
| 8862 |
'message' => 'Item failed after maximum attempts: ' . $error_message, |
| 8863 |
'permanent_failure' => true, |
| 8864 |
'item_id' => $item_id |
| 8865 |
)); |
| 8866 |
} else { |
| 8867 |
// Mark for retry |
| 8868 |
$wpdb->update( |
| 8869 |
$table_name, |
| 8870 |
array( |
| 8871 |
'status' => 'failed', |
| 8872 |
'error_message' => $error_message |
| 8873 |
), |
| 8874 |
array('id' => $item_id), |
| 8875 |
array('%s', '%s'), |
| 8876 |
array('%d') |
| 8877 |
); |
| 8878 |
|
| 8879 |
wp_send_json_error(array( |
| 8880 |
'message' => 'Item processing failed, will retry: ' . $error_message, |
| 8881 |
'can_retry' => true, |
| 8882 |
'item_id' => $item_id, |
| 8883 |
'attempts' => $item ? $item->attempts : 0 |
| 8884 |
)); |
| 8885 |
} |
| 8886 |
} |
| 8887 |
} |
| 8888 |
|
| 8889 |
/** |
| 8890 |
* Process a URL from the queue |
| 8891 |
*/ |
| 8892 |
private function mxchat_process_queue_url($item_data, $bot_id = 'default', $queue_id = '') { |
| 8893 |
$url = isset($item_data['url']) ? $item_data['url'] : ''; |
| 8894 |
|
| 8895 |
if (empty($url)) { |
| 8896 |
return new WP_Error('invalid_url', 'URL is empty'); |
| 8897 |
} |
| 8898 |
|
| 8899 |
// Get bot-specific embedding decision early (needed for both paths) — |
| 8900 |
// custom-provider-aware (plan cbd5fd). Error code preserved. |
| 8901 |
$bot_options = $this->get_bot_options($bot_id); |
| 8902 |
$options = !empty($bot_options) ? $bot_options : get_option('mxchat_options'); |
| 8903 |
|
| 8904 |
$preflight = MxChat_Utils::embedding_preflight($options); |
| 8905 |
if (!$preflight['ok']) { |
| 8906 |
return new WP_Error('no_api_key', $preflight['reason'] . ' (bot: ' . $bot_id . ')'); |
| 8907 |
} |
| 8908 |
$api_key = $preflight['api_key']; |
| 8909 |
|
| 8910 |
// Check if this is a WooCommerce product URL and WooCommerce is active |
| 8911 |
$is_product_url = (strpos($url, '/product/') !== false || strpos($url, '/shop/') !== false); |
| 8912 |
$content_type = $is_product_url ? 'product' : 'url'; |
| 8913 |
|
| 8914 |
// Try to get WooCommerce product data if it's a product URL |
| 8915 |
if ($is_product_url && class_exists('WooCommerce')) { |
| 8916 |
$product_content = $this->mxchat_extract_woocommerce_product_content($url); |
| 8917 |
|
| 8918 |
if (!empty($product_content)) { |
| 8919 |
// Successfully extracted WooCommerce product data with pricing |
| 8920 |
$result = MxChat_Utils::submit_content_to_db( |
| 8921 |
$product_content, |
| 8922 |
$url, |
| 8923 |
$api_key, |
| 8924 |
null, |
| 8925 |
$bot_id, |
| 8926 |
'product' |
| 8927 |
); |
| 8928 |
return $result; |
| 8929 |
} |
| 8930 |
// If WooCommerce extraction failed, fall through to HTML extraction |
| 8931 |
} |
| 8932 |
|
| 8933 |
// Fetch URL content (fallback for non-products or when WooCommerce extraction fails) |
| 8934 |
$is_likely_pdf = (strtolower(pathinfo(parse_url($url, PHP_URL_PATH) ?: '', PATHINFO_EXTENSION)) === 'pdf'); |
| 8935 |
$response = wp_remote_get($url, array( |
| 8936 |
'timeout' => $is_likely_pdf ? 120 : 30, |
| 8937 |
'redirection' => 5, |
| 8938 |
'user-agent' => mxchat_ingest_user_agent(), |
| 8939 |
)); |
| 8940 |
|
| 8941 |
if (is_wp_error($response)) { |
| 8942 |
return $response; |
| 8943 |
} |
| 8944 |
|
| 8945 |
$response_code = wp_remote_retrieve_response_code($response); |
| 8946 |
if ($response_code !== 200) { |
| 8947 |
return new WP_Error('http_error', 'HTTP ' . $response_code . ' error for: ' . $url); |
| 8948 |
} |
| 8949 |
|
| 8950 |
// Check if URL is a PDF — expand into per-page queue items using the standard PDF pipeline |
| 8951 |
if ($this->mxchat_is_pdf_url($url, $response)) { |
| 8952 |
return $this->mxchat_expand_pdf_to_queue($url, $response, $bot_id, $queue_id); |
| 8953 |
} |
| 8954 |
|
| 8955 |
$html = wp_remote_retrieve_body($response); |
| 8956 |
|
| 8957 |
if (empty($html)) { |
| 8958 |
return new WP_Error('empty_response', 'Empty response body'); |
| 8959 |
} |
| 8960 |
|
| 8961 |
// Extract and sanitize content |
| 8962 |
$content = $this->mxchat_extract_main_content($html); |
| 8963 |
$sanitized = $this->mxchat_sanitize_content_for_api($content); |
| 8964 |
|
| 8965 |
if (empty($sanitized)) { |
| 8966 |
// Not an error - just no content found (maybe a redirect or empty page) |
| 8967 |
return false; |
| 8968 |
} |
| 8969 |
|
| 8970 |
// Submit to database with content_type |
| 8971 |
$result = MxChat_Utils::submit_content_to_db( |
| 8972 |
$sanitized, |
| 8973 |
$url, |
| 8974 |
$api_key, |
| 8975 |
null, |
| 8976 |
$bot_id, |
| 8977 |
$content_type |
| 8978 |
); |
| 8979 |
|
| 8980 |
return $result; |
| 8981 |
} |
| 8982 |
|
| 8983 |
/** |
| 8984 |
* Expand a PDF URL into per-page queue items using the standard PDF pipeline. |
| 8985 |
* Called when a sitemap URL turns out to be a PDF — downloads, parses page count, |
| 8986 |
* and adds pdf_page items to the same queue so they process with full progress tracking. |
| 8987 |
*/ |
| 8988 |
private function mxchat_expand_pdf_to_queue($pdf_url, $response, $bot_id = 'default', $queue_id = '') { |
| 8989 |
set_time_limit(120); // PDFs need extra time for download + parsing |
| 8990 |
|
| 8991 |
$upload_dir = wp_upload_dir(); |
| 8992 |
$pdf_filename = sanitize_file_name('mxchat_kb_' . md5($pdf_url) . '.pdf'); |
| 8993 |
$pdf_path = trailingslashit($upload_dir['path']) . $pdf_filename; |
| 8994 |
|
| 8995 |
$response_body = wp_remote_retrieve_body($response); |
| 8996 |
if (empty($response_body)) { |
| 8997 |
return new WP_Error('empty_pdf', 'Empty PDF response for: ' . $pdf_url); |
| 8998 |
} |
| 8999 |
|
| 9000 |
if (!wp_mkdir_p(dirname($pdf_path))) { |
| 9001 |
return new WP_Error('dir_error', 'Failed to create upload directory'); |
| 9002 |
} |
| 9003 |
|
| 9004 |
file_put_contents($pdf_path, $response_body); |
| 9005 |
|
| 9006 |
if (!file_exists($pdf_path)) { |
| 9007 |
return new WP_Error('save_error', 'Failed to save PDF file'); |
| 9008 |
} |
| 9009 |
|
| 9010 |
try { |
| 9011 |
$total_pages = $this->mxchat_validate_and_count_pdf_pages($pdf_path); |
| 9012 |
|
| 9013 |
if ($total_pages === false || $total_pages < 1) { |
| 9014 |
wp_delete_file($pdf_path); |
| 9015 |
return new WP_Error('no_pages', 'PDF has no pages: ' . $pdf_url); |
| 9016 |
} |
| 9017 |
|
| 9018 |
// Build per-page items identical to mxchat_handle_pdf_for_knowledge_base |
| 9019 |
$pages = array(); |
| 9020 |
for ($i = 1; $i <= $total_pages; $i++) { |
| 9021 |
$pages[] = array( |
| 9022 |
'pdf_path' => $pdf_path, |
| 9023 |
'pdf_url' => $pdf_url, |
| 9024 |
'page_number' => $i, |
| 9025 |
'total_pages' => $total_pages |
| 9026 |
); |
| 9027 |
} |
| 9028 |
|
| 9029 |
// Add pdf_page items to the SAME queue so the JS picks them up automatically |
| 9030 |
if (!empty($queue_id)) { |
| 9031 |
$queued_count = $this->mxchat_add_to_queue($queue_id, 'pdf_page', $pages, $bot_id); |
| 9032 |
} else { |
| 9033 |
// Fallback: create a new PDF queue (shouldn't happen in sitemap flow) |
| 9034 |
$new_queue_id = 'pdf_' . md5($pdf_url . time()); |
| 9035 |
$queued_count = $this->mxchat_add_to_queue($new_queue_id, 'pdf_page', $pages, $bot_id); |
| 9036 |
$this->mxchat_set_queue_meta($new_queue_id, 'source_url', $pdf_url); |
| 9037 |
$this->mxchat_set_queue_meta($new_queue_id, 'queue_type', 'pdf'); |
| 9038 |
$this->mxchat_set_queue_meta($new_queue_id, 'total_items', $total_pages); |
| 9039 |
$this->mxchat_set_queue_meta($new_queue_id, 'bot_id', $bot_id); |
| 9040 |
$this->mxchat_set_queue_meta($new_queue_id, 'pdf_path', $pdf_path); |
| 9041 |
$this->mxchat_set_queue_meta($new_queue_id, 'created_at', current_time('mysql')); |
| 9042 |
} |
| 9043 |
|
| 9044 |
if ($queued_count === 0) { |
| 9045 |
wp_delete_file($pdf_path); |
| 9046 |
return new WP_Error('queue_error', 'Failed to add PDF pages to queue'); |
| 9047 |
} |
| 9048 |
|
| 9049 |
// Return true so the original URL item is marked complete |
| 9050 |
// The new pdf_page items will be processed in subsequent batches |
| 9051 |
return true; |
| 9052 |
|
| 9053 |
} catch (Exception $e) { |
| 9054 |
if (file_exists($pdf_path)) { |
| 9055 |
wp_delete_file($pdf_path); |
| 9056 |
} |
| 9057 |
return new WP_Error('pdf_parse_error', 'Error parsing PDF: ' . $e->getMessage()); |
| 9058 |
} |
| 9059 |
} |
| 9060 |
|
| 9061 |
/** |
| 9062 |
* Legacy: Process a PDF URL inline during sitemap queue processing. |
| 9063 |
* @deprecated Use mxchat_expand_pdf_to_queue instead — kept for reference only. |
| 9064 |
*/ |
| 9065 |
private function mxchat_process_pdf_url_inline($pdf_url, $response, $api_key, $bot_id = 'default') { |
| 9066 |
set_time_limit(120); // PDFs need more time — downloading + parsing all pages |
| 9067 |
|
| 9068 |
$upload_dir = wp_upload_dir(); |
| 9069 |
$pdf_filename = sanitize_file_name('mxchat_kb_' . md5($pdf_url) . '.pdf'); |
| 9070 |
$pdf_path = trailingslashit($upload_dir['path']) . $pdf_filename; |
| 9071 |
|
| 9072 |
$response_body = wp_remote_retrieve_body($response); |
| 9073 |
if (empty($response_body)) { |
| 9074 |
return new WP_Error('empty_pdf', 'Empty PDF response'); |
| 9075 |
} |
| 9076 |
|
| 9077 |
if (!wp_mkdir_p(dirname($pdf_path))) { |
| 9078 |
return new WP_Error('dir_error', 'Failed to create upload directory'); |
| 9079 |
} |
| 9080 |
|
| 9081 |
file_put_contents($pdf_path, $response_body); |
| 9082 |
|
| 9083 |
if (!file_exists($pdf_path)) { |
| 9084 |
return new WP_Error('save_error', 'Failed to save PDF file'); |
| 9085 |
} |
| 9086 |
|
| 9087 |
try { |
| 9088 |
mxchat_load_pdf_parser(); |
| 9089 |
$parser = new \Smalot\PdfParser\Parser(); |
| 9090 |
$pdf = $parser->parseFile($pdf_path); |
| 9091 |
$pages = $pdf->getPages(); |
| 9092 |
$total_pages = count($pages); |
| 9093 |
|
| 9094 |
if ($total_pages < 1) { |
| 9095 |
wp_delete_file($pdf_path); |
| 9096 |
return new WP_Error('no_pages', 'PDF has no pages'); |
| 9097 |
} |
| 9098 |
|
| 9099 |
$processed = 0; |
| 9100 |
$skipped_pages = array(); |
| 9101 |
|
| 9102 |
for ($i = 0; $i < $total_pages; $i++) { |
| 9103 |
$page_num = $i + 1; |
| 9104 |
$text = $pages[$i]->getText(); |
| 9105 |
$text = MxChat_Utils::normalize_pdf_rtl($text, 'kb_pdf_import page ' . $page_num); |
| 9106 |
if (empty($text)) { |
| 9107 |
$skipped_pages[] = 'Page ' . $page_num . ': No text could be extracted — page may contain only images, links, or non-standard encoding'; |
| 9108 |
continue; |
| 9109 |
} |
| 9110 |
|
| 9111 |
$sanitized = $this->mxchat_sanitize_content_for_api($text); |
| 9112 |
if (empty($sanitized)) { |
| 9113 |
$skipped_pages[] = 'Page ' . $page_num . ': Text was extracted but contained only special characters, control codes, or unsupported content'; |
| 9114 |
continue; |
| 9115 |
} |
| 9116 |
|
| 9117 |
$metadata = array( |
| 9118 |
'document_type' => 'pdf', |
| 9119 |
'total_pages' => $total_pages, |
| 9120 |
'current_page' => $page_num, |
| 9121 |
'source_url' => $pdf_url, |
| 9122 |
); |
| 9123 |
|
| 9124 |
$content_with_metadata = wp_json_encode($metadata) . "\n---\n" . $sanitized; |
| 9125 |
$page_url = esc_url($pdf_url . '#page=' . $page_num); |
| 9126 |
|
| 9127 |
MxChat_Utils::submit_content_to_db( |
| 9128 |
$content_with_metadata, |
| 9129 |
$page_url, |
| 9130 |
$api_key, |
| 9131 |
null, |
| 9132 |
$bot_id, |
| 9133 |
'pdf' |
| 9134 |
); |
| 9135 |
|
| 9136 |
$processed++; |
| 9137 |
} |
| 9138 |
|
| 9139 |
// Clean up the temp PDF file |
| 9140 |
wp_delete_file($pdf_path); |
| 9141 |
|
| 9142 |
if (!empty($skipped_pages)) { |
| 9143 |
error_log('MxChat PDF: Skipped ' . count($skipped_pages) . ' of ' . $total_pages . ' pages: ' . implode('; ', $skipped_pages)); |
| 9144 |
} |
| 9145 |
|
| 9146 |
return $processed > 0 ? true : false; |
| 9147 |
|
| 9148 |
} catch (Exception $e) { |
| 9149 |
if (file_exists($pdf_path)) { |
| 9150 |
wp_delete_file($pdf_path); |
| 9151 |
} |
| 9152 |
return new WP_Error('pdf_parse_error', 'Error parsing PDF: ' . $e->getMessage()); |
| 9153 |
} |
| 9154 |
} |
| 9155 |
|
| 9156 |
/** |
| 9157 |
* Extract WooCommerce product content including pricing |
| 9158 |
* |
| 9159 |
* @param string $url The product URL |
| 9160 |
* @return string|false Product content with pricing, or false if not found |
| 9161 |
*/ |
| 9162 |
private function mxchat_extract_woocommerce_product_content($url) { |
| 9163 |
// Try to get product ID from URL |
| 9164 |
$product_id = url_to_postid($url); |
| 9165 |
|
| 9166 |
// If url_to_postid fails, try to extract from URL pattern |
| 9167 |
if (!$product_id) { |
| 9168 |
$product_slug = ''; |
| 9169 |
|
| 9170 |
// Handle pretty permalinks: /product/product-name/ |
| 9171 |
if (preg_match('/\/product\/([^\/\?]+)/', $url, $matches)) { |
| 9172 |
$product_slug = $matches[1]; |
| 9173 |
} |
| 9174 |
|
| 9175 |
if (!empty($product_slug)) { |
| 9176 |
$product_post = get_page_by_path($product_slug, OBJECT, 'product'); |
| 9177 |
if ($product_post) { |
| 9178 |
$product_id = $product_post->ID; |
| 9179 |
} |
| 9180 |
} |
| 9181 |
} |
| 9182 |
|
| 9183 |
if (!$product_id) { |
| 9184 |
return false; |
| 9185 |
} |
| 9186 |
|
| 9187 |
// Get WooCommerce product object |
| 9188 |
$product = wc_get_product($product_id); |
| 9189 |
|
| 9190 |
if (!$product) { |
| 9191 |
return false; |
| 9192 |
} |
| 9193 |
|
| 9194 |
// Build product content via the shared WC-object assembler (a3d60c) — same |
| 9195 |
// body as the auto-sync product writer, so the two paths can never drift. |
| 9196 |
$content = $this->mxchat_prepare_product_content_for_indexing($product); |
| 9197 |
|
| 9198 |
return $this->mxchat_sanitize_content_for_api($content); |
| 9199 |
} |
| 9200 |
|
| 9201 |
/** |
| 9202 |
* Process a PDF page from the queue |
| 9203 |
*/ |
| 9204 |
private function mxchat_process_queue_pdf_page($item_data, $bot_id = 'default') { |
| 9205 |
$pdf_path = isset($item_data['pdf_path']) ? $item_data['pdf_path'] : ''; |
| 9206 |
$pdf_url = isset($item_data['pdf_url']) ? $item_data['pdf_url'] : ''; |
| 9207 |
$page_number = isset($item_data['page_number']) ? absint($item_data['page_number']) : 0; |
| 9208 |
$total_pages = isset($item_data['total_pages']) ? absint($item_data['total_pages']) : 0; |
| 9209 |
|
| 9210 |
if (empty($pdf_path) || !file_exists($pdf_path)) { |
| 9211 |
return new WP_Error('pdf_not_found', 'PDF file not found: ' . $pdf_path); |
| 9212 |
} |
| 9213 |
|
| 9214 |
if ($page_number < 1) { |
| 9215 |
return new WP_Error('invalid_page', 'Invalid page number'); |
| 9216 |
} |
| 9217 |
|
| 9218 |
try { |
| 9219 |
mxchat_load_pdf_parser(); |
| 9220 |
$parser = new \Smalot\PdfParser\Parser(); |
| 9221 |
$pdf = $parser->parseFile($pdf_path); |
| 9222 |
$pages = $pdf->getPages(); |
| 9223 |
|
| 9224 |
if (!isset($pages[$page_number - 1])) { |
| 9225 |
return new WP_Error('page_not_found', 'Page ' . $page_number . ' not found in PDF'); |
| 9226 |
} |
| 9227 |
|
| 9228 |
$text = $pages[$page_number - 1]->getText(); |
| 9229 |
$text = MxChat_Utils::normalize_pdf_rtl($text, 'kb_pdf_page page ' . $page_number); |
| 9230 |
|
| 9231 |
if (empty($text)) { |
| 9232 |
return new WP_Error('empty_page', 'Page ' . $page_number . ': No text could be extracted — page may contain only images, links, or non-standard encoding'); |
| 9233 |
} |
| 9234 |
|
| 9235 |
$sanitized = $this->mxchat_sanitize_content_for_api($text); |
| 9236 |
|
| 9237 |
if (empty($sanitized)) { |
| 9238 |
return new WP_Error('empty_after_sanitization', 'Page ' . $page_number . ': Text was extracted but contained only special characters, control codes, or unsupported content that was removed during cleanup'); |
| 9239 |
} |
| 9240 |
|
| 9241 |
// Create metadata |
| 9242 |
$metadata = array( |
| 9243 |
'document_type' => 'pdf', |
| 9244 |
'total_pages' => $total_pages, |
| 9245 |
'current_page' => $page_number, |
| 9246 |
'source_url' => $pdf_url |
| 9247 |
); |
| 9248 |
|
| 9249 |
$content_with_metadata = wp_json_encode($metadata) . "\n---\n" . $sanitized; |
| 9250 |
$page_url = esc_url($pdf_url . "#page=" . $page_number); |
| 9251 |
|
| 9252 |
// Get bot-specific embedding decision — custom-provider-aware |
| 9253 |
// (plan cbd5fd). Error code preserved. |
| 9254 |
$bot_options = $this->get_bot_options($bot_id); |
| 9255 |
$options = !empty($bot_options) ? $bot_options : get_option('mxchat_options'); |
| 9256 |
|
| 9257 |
$preflight = MxChat_Utils::embedding_preflight($options); |
| 9258 |
if (!$preflight['ok']) { |
| 9259 |
return new WP_Error('no_api_key', $preflight['reason'] . ' (bot: ' . $bot_id . ')'); |
| 9260 |
} |
| 9261 |
$api_key = $preflight['api_key']; |
| 9262 |
|
| 9263 |
// Submit to database - UPDATED 2.5.6: Added content_type 'pdf' |
| 9264 |
$result = MxChat_Utils::submit_content_to_db( |
| 9265 |
$content_with_metadata, |
| 9266 |
$page_url, |
| 9267 |
$api_key, |
| 9268 |
null, |
| 9269 |
$bot_id, |
| 9270 |
'pdf' |
| 9271 |
); |
| 9272 |
|
| 9273 |
return $result; |
| 9274 |
|
| 9275 |
} catch (Exception $e) { |
| 9276 |
return new WP_Error('pdf_parse_error', 'Error parsing PDF: ' . $e->getMessage()); |
| 9277 |
} |
| 9278 |
} |
| 9279 |
|
| 9280 |
/** |
| 9281 |
* AJAX: Get queue processing status |
| 9282 |
*/ |
| 9283 |
public function ajax_mxchat_get_queue_status() { |
| 9284 |
// Verify nonce and permissions |
| 9285 |
check_ajax_referer('mxchat_queue_nonce', 'nonce'); |
| 9286 |
|
| 9287 |
if (!current_user_can('manage_options')) { |
| 9288 |
wp_send_json_error('Unauthorized access'); |
| 9289 |
} |
| 9290 |
|
| 9291 |
$queue_id = isset($_POST['queue_id']) ? sanitize_text_field($_POST['queue_id']) : ''; |
| 9292 |
|
| 9293 |
if (empty($queue_id)) { |
| 9294 |
wp_send_json_error('Missing queue ID'); |
| 9295 |
} |
| 9296 |
|
| 9297 |
global $wpdb; |
| 9298 |
$table_name = $wpdb->prefix . 'mxchat_processing_queue'; |
| 9299 |
|
| 9300 |
// Get counts by status |
| 9301 |
$counts = $wpdb->get_results($wpdb->prepare( |
| 9302 |
"SELECT status, COUNT(*) as count |
| 9303 |
FROM $table_name |
| 9304 |
WHERE queue_id = %s |
| 9305 |
GROUP BY status", |
| 9306 |
$queue_id |
| 9307 |
), OBJECT_K); |
| 9308 |
|
| 9309 |
$total = 0; |
| 9310 |
$completed = 0; |
| 9311 |
$failed = 0; |
| 9312 |
$processing = 0; |
| 9313 |
$pending = 0; |
| 9314 |
|
| 9315 |
foreach ($counts as $status => $data) { |
| 9316 |
$count = absint($data->count); |
| 9317 |
$total += $count; |
| 9318 |
|
| 9319 |
switch ($status) { |
| 9320 |
case 'completed': |
| 9321 |
$completed = $count; |
| 9322 |
break; |
| 9323 |
case 'failed': |
| 9324 |
$failed = $count; |
| 9325 |
break; |
| 9326 |
case 'processing': |
| 9327 |
$processing = $count; |
| 9328 |
break; |
| 9329 |
case 'pending': |
| 9330 |
$pending = $count; |
| 9331 |
break; |
| 9332 |
} |
| 9333 |
} |
| 9334 |
|
| 9335 |
// Calculate percentage |
| 9336 |
$percentage = $total > 0 ? round((($completed + $failed) / $total) * 100) : 0; |
| 9337 |
|
| 9338 |
// Get failed items details (include all failed items, not just those that exhausted retries) |
| 9339 |
$failed_items = array(); |
| 9340 |
if ($failed > 0) { |
| 9341 |
$failed_items = $wpdb->get_results($wpdb->prepare( |
| 9342 |
"SELECT item_type, item_data, error_message, attempts |
| 9343 |
FROM $table_name |
| 9344 |
WHERE queue_id = %s |
| 9345 |
AND status = 'failed' |
| 9346 |
ORDER BY id DESC |
| 9347 |
LIMIT 50", |
| 9348 |
$queue_id |
| 9349 |
)); |
| 9350 |
} |
| 9351 |
|
| 9352 |
// Get queue metadata |
| 9353 |
$source_url = $this->mxchat_get_queue_meta($queue_id, 'source_url'); |
| 9354 |
$queue_type = $this->mxchat_get_queue_meta($queue_id, 'queue_type'); |
| 9355 |
|
| 9356 |
// Determine if queue is complete |
| 9357 |
$is_complete = ($pending === 0 && $processing === 0); |
| 9358 |
|
| 9359 |
wp_send_json_success(array( |
| 9360 |
'queue_id' => $queue_id, |
| 9361 |
'queue_type' => $queue_type, |
| 9362 |
'source_url' => $source_url, |
| 9363 |
'total' => $total, |
| 9364 |
'completed' => $completed, |
| 9365 |
'failed' => $failed, |
| 9366 |
'processing' => $processing, |
| 9367 |
'pending' => $pending, |
| 9368 |
'percentage' => $percentage, |
| 9369 |
'is_complete' => $is_complete, |
| 9370 |
'failed_items' => $failed_items, |
| 9371 |
'status' => $is_complete ? 'complete' : 'processing' |
| 9372 |
)); |
| 9373 |
} |
| 9374 |
|
| 9375 |
/** |
| 9376 |
* AJAX: Clear completed queue |
| 9377 |
*/ |
| 9378 |
public function ajax_mxchat_clear_queue() { |
| 9379 |
// Verify nonce and permissions |
| 9380 |
check_ajax_referer('mxchat_queue_nonce', 'nonce'); |
| 9381 |
|
| 9382 |
if (!current_user_can('manage_options')) { |
| 9383 |
wp_send_json_error('Unauthorized access'); |
| 9384 |
} |
| 9385 |
|
| 9386 |
$queue_id = isset($_POST['queue_id']) ? sanitize_text_field($_POST['queue_id']) : ''; |
| 9387 |
|
| 9388 |
if (empty($queue_id)) { |
| 9389 |
wp_send_json_error('Missing queue ID'); |
| 9390 |
} |
| 9391 |
|
| 9392 |
global $wpdb; |
| 9393 |
$table_name = $wpdb->prefix . 'mxchat_processing_queue'; |
| 9394 |
$meta_table = $wpdb->prefix . 'mxchat_queue_meta'; |
| 9395 |
|
| 9396 |
// Delete queue items |
| 9397 |
$wpdb->delete( |
| 9398 |
$table_name, |
| 9399 |
array('queue_id' => $queue_id), |
| 9400 |
array('%s') |
| 9401 |
); |
| 9402 |
|
| 9403 |
// Delete queue metadata |
| 9404 |
$wpdb->delete( |
| 9405 |
$meta_table, |
| 9406 |
array('queue_id' => $queue_id), |
| 9407 |
array('%s') |
| 9408 |
); |
| 9409 |
|
| 9410 |
wp_send_json_success(array( |
| 9411 |
'message' => 'Queue cleared successfully' |
| 9412 |
)); |
| 9413 |
} |
| 9414 |
|
| 9415 |
/** |
| 9416 |
* AJAX: Retry failed items in queue |
| 9417 |
*/ |
| 9418 |
public function ajax_mxchat_retry_failed() { |
| 9419 |
// Verify nonce and permissions |
| 9420 |
check_ajax_referer('mxchat_queue_nonce', 'nonce'); |
| 9421 |
|
| 9422 |
if (!current_user_can('manage_options')) { |
| 9423 |
wp_send_json_error('Unauthorized access'); |
| 9424 |
} |
| 9425 |
|
| 9426 |
$queue_id = isset($_POST['queue_id']) ? sanitize_text_field($_POST['queue_id']) : ''; |
| 9427 |
|
| 9428 |
if (empty($queue_id)) { |
| 9429 |
wp_send_json_error('Missing queue ID'); |
| 9430 |
} |
| 9431 |
|
| 9432 |
global $wpdb; |
| 9433 |
$table_name = $wpdb->prefix . 'mxchat_processing_queue'; |
| 9434 |
|
| 9435 |
// Reset failed items to pending and reset attempt count |
| 9436 |
$updated = $wpdb->update( |
| 9437 |
$table_name, |
| 9438 |
array( |
| 9439 |
'status' => 'pending', |
| 9440 |
'attempts' => 0, |
| 9441 |
'error_message' => null |
| 9442 |
), |
| 9443 |
array( |
| 9444 |
'queue_id' => $queue_id, |
| 9445 |
'status' => 'failed' |
| 9446 |
), |
| 9447 |
array('%s', '%d', '%s'), |
| 9448 |
array('%s', '%s') |
| 9449 |
); |
| 9450 |
|
| 9451 |
wp_send_json_success(array( |
| 9452 |
'message' => 'Reset ' . $updated . ' failed items for retry', |
| 9453 |
'reset_count' => $updated |
| 9454 |
)); |
| 9455 |
} |
| 9456 |
|
| 9457 |
|
| 9458 |
public function ajax_mxchat_mark_queue_complete() { |
| 9459 |
check_ajax_referer('mxchat_queue_nonce', 'nonce'); |
| 9460 |
|
| 9461 |
if (!current_user_can('manage_options')) { |
| 9462 |
wp_send_json_error('Unauthorized access'); |
| 9463 |
} |
| 9464 |
|
| 9465 |
$queue_id = isset($_POST['queue_id']) ? sanitize_text_field($_POST['queue_id']) : ''; |
| 9466 |
|
| 9467 |
if (empty($queue_id)) { |
| 9468 |
wp_send_json_error('Missing queue ID'); |
| 9469 |
} |
| 9470 |
|
| 9471 |
// Clear active queue transients |
| 9472 |
if (strpos($queue_id, 'sitemap_') === 0) { |
| 9473 |
delete_transient('mxchat_active_queue_sitemap'); |
| 9474 |
} else if (strpos($queue_id, 'pdf_') === 0) { |
| 9475 |
delete_transient('mxchat_active_queue_pdf'); |
| 9476 |
} |
| 9477 |
|
| 9478 |
wp_send_json_success(array('message' => 'Queue marked as complete')); |
| 9479 |
} |
| 9480 |
|
| 9481 |
|
| 9482 |
// ======================================== |
| 9483 |
// STATIC ACCESS METHODS |
| 9484 |
// ======================================== |
| 9485 |
|
| 9486 |
/** |
| 9487 |
* Get singleton instance |
| 9488 |
*/ |
| 9489 |
public static function get_instance() { |
| 9490 |
static $instance = null; |
| 9491 |
if ($instance === null) { |
| 9492 |
$instance = new self(); |
| 9493 |
} |
| 9494 |
return $instance; |
| 9495 |
} |
| 9496 |
} |
| 9497 |
|
| 9498 |
// Initialize the Knowledge manager |
| 9499 |
$mxchat_knowledge_manager = MxChat_Knowledge_Manager::get_instance(); |
