| @@ -9,27 +9,11 @@ | ||
| 9 | 9 | exit; // Exit if accessed directly |
| 10 | 10 | } |
| 11 | 11 | |
| 12 | 12 | class MxChat_Knowledge_Manager { |
| 13 | - | |
| 13 | + | |
| 14 | 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 $pending_post_update = array(); | |
| 31 | - | |
| 15 | + | |
| 32 | 16 | /** |
| 33 | 17 | * Constructor - Register hooks for content processing |
| 34 | 18 | */ |
| 35 | 19 | public function __construct() { |
| @@ -47,10 +31,8 @@ | ||
| 47 | 31 | // Admin post handlers for form submissions |
| 48 | 32 | add_action('admin_post_mxchat_submit_content', array($this, 'mxchat_handle_content_submission')); |
| 49 | 33 | add_action('admin_post_mxchat_submit_sitemap', array($this, 'mxchat_handle_sitemap_submission')); |
| 50 | 34 | add_action('admin_post_mxchat_submit_pdf_file', array($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 | 35 | add_action('admin_post_mxchat_stop_processing', array($this, 'mxchat_stop_processing')); |
| 54 | 36 | |
| 55 | 37 | // AJAX handlers for real-time processing and status updates |
| 56 | 38 | add_action('wp_ajax_mxchat_get_status_updates', array($this, 'mxchat_ajax_get_status_updates')); |
| @@ -76,9 +58,8 @@ | ||
| 76 | 58 | add_action('wp_ajax_mxchat_refresh_pinecone_entries', array($this, 'ajax_mxchat_refresh_pinecone_entries')); |
| 77 | 59 | add_action('wp_ajax_mxchat_paginate_entries', array($this, 'ajax_mxchat_paginate_entries')); |
| 78 | 60 | add_action('wp_ajax_mxchat_get_entry_content', array($this, 'ajax_mxchat_get_entry_content')); |
| 79 | 61 | add_action('wp_ajax_mxchat_save_entry_content', array($this, 'ajax_mxchat_save_entry_content')); |
| 80 | - add_action('wp_ajax_mxchat_inspect_entry', array($this, 'ajax_mxchat_inspect_entry')); | |
| 81 | 62 | |
| 82 | 63 | // WordPress post management hooks |
| 83 | 64 | add_action('pre_post_update', array($this, 'mxchat_store_pre_update_status'), 10, 2); |
| 84 | 65 | add_action('post_updated', array($this, 'mxchat_handle_post_update'), 10, 3); |
| @@ -83,27 +64,13 @@ | ||
| 83 | 64 | add_action('pre_post_update', array($this, 'mxchat_store_pre_update_status'), 10, 2); |
| 84 | 65 | add_action('post_updated', array($this, 'mxchat_handle_post_update'), 10, 3); |
| 85 | 66 | add_action('before_delete_post', array($this, 'mxchat_handle_post_delete')); |
| 86 | 67 | 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 | 68 | |
| 93 | 69 | // ACF hook - fires AFTER ACF fields are saved, ensuring ACF data is available |
| 94 | 70 | // Priority 20 to run after ACF's own save (which runs at priority 10) |
| 95 | 71 | add_action('acf/save_post', array($this, 'mxchat_handle_acf_save'), 20); |
| 96 | 72 | |
| 97 | - // One-time cleanup for vectors orphaned by unpublishes that predate the | |
| 98 | - // transition_post_status handler (plan 816fb1): wp mxchat prune-unpublished | |
| 99 | - if (defined('WP_CLI') && WP_CLI) { | |
| 100 | - WP_CLI::add_command('mxchat prune-unpublished', array($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($this, 'cli_rtl_repair')); | |
| 104 | - } | |
| 105 | - | |
| 106 | 73 | add_action('wp_ajax_mxchat_mark_queue_complete', array($this, 'ajax_mxchat_mark_queue_complete')); |
| 107 | 74 | |
| 108 | 75 | // WooCommerce product hooks (if WooCommerce is active) |
| 109 | 76 | if (class_exists('WooCommerce')) { |
| @@ -151,17 +118,26 @@ | ||
| 151 | 118 | |
| 152 | 119 | // Get bot-specific options and API key |
| 153 | 120 | $bot_options = $this->get_bot_options($bot_id); |
| 154 | 121 | $options = !empty($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); | |
| 122 | + $selected_model = $options['embedding_model'] ?? 'text-embedding-ada-002'; | |
| 123 | + | |
| 124 | + if (strpos($selected_model, 'voyage') === 0) { | |
| 125 | + $api_key = $options['voyage_api_key'] ?? ''; | |
| 126 | + } elseif (strpos($selected_model, 'gemini-embedding') === 0) { | |
| 127 | + $api_key = $options['gemini_api_key'] ?? ''; | |
| 128 | + } else { | |
| 129 | + $api_key = $options['api_key'] ?? ''; | |
| 130 | + } | |
| 131 | + | |
| 132 | + if (empty($api_key)) { | |
| 133 | + set_transient('mxchat_admin_notice_error', | |
| 134 | + esc_html__('API key is not configured. Please add your API key in the settings before submitting content.', 'mxchat'), | |
| 135 | + 30 | |
| 136 | + ); | |
| 160 | 137 | wp_safe_redirect(esc_url(admin_url('admin.php?page=mxchat-prompts'))); |
| 161 | 138 | exit; |
| 162 | 139 | } |
| 163 | - $api_key = $preflight['api_key']; | |
| 164 | 140 | |
| 165 | 141 | // Use centralized utility function with bot_id |
| 166 | 142 | $result = MxChat_Utils::submit_content_to_db($article_content, $article_url, $api_key, null, $bot_id); |
| 167 | 143 | |
| @@ -180,238 +156,8 @@ | ||
| 180 | 156 | wp_safe_redirect(esc_url(admin_url('admin.php?page=mxchat-prompts'))); |
| 181 | 157 | exit; |
| 182 | 158 | } |
| 183 | 159 | |
| 184 | -/** | |
| 185 | - * Handle the "YouTube" KB import source (admin-post form submission). | |
| 186 | - * | |
| 187 | - * Per-video description mode: | |
| 188 | - * - auto: fetch oEmbed metadata (reliable) + best-effort captions transcript. | |
| 189 | - * If no usable transcript, index the metadata anyway, tell the admin, | |
| 190 | - * and bounce back with the manual box pre-filled (never fail silently). | |
| 191 | - * - manual: the admin's own description is what gets indexed; metadata rides along. | |
| 192 | - * | |
| 193 | - * The row is stored with content_type 'youtube' and source_url = the canonical | |
| 194 | - * watch URL, so re-importing the same video UPDATES the entry (source_url | |
| 195 | - * duplicate handling in MxChat_Utils::store_in_wordpress_db) — that is also the | |
| 196 | - * "augment a metadata-only entry" path. | |
| 197 | - */ | |
| 198 | -public function mxchat_handle_youtube_submission() { | |
| 199 | - if (!isset($_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 $text; | |
| 412 | -} | |
| 413 | - | |
| 414 | 160 | public function mxchat_is_pdf_url($url, $response) { |
| 415 | 161 | $content_type = wp_remote_retrieve_header($response, 'content-type'); |
| 416 | 162 | $file_extension = strtolower(pathinfo($url, PATHINFO_EXTENSION)); |
| 417 | 163 | |
| @@ -657,160 +403,8 @@ | ||
| 657 | 403 | exit; |
| 658 | 404 | } |
| 659 | 405 | |
| 660 | 406 | /** |
| 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($redirect_url)); | |
| 809 | - exit; | |
| 810 | -} | |
| 811 | - | |
| 812 | -/** | |
| 813 | 407 | * Validate PDF and count pages with multiple parser attempts |
| 814 | 408 | */ |
| 815 | 409 | private function mxchat_validate_and_count_pdf_pages($pdf_path) { |
| 816 | 410 | // Method 1: Try with Smalot PDF Parser (your current method) |
| @@ -1033,25 +627,8 @@ | ||
| 1033 | 627 | /** |
| 1034 | 628 | * AJAX: Get full content for editing — reassembles chunks if needed. |
| 1035 | 629 | * Works for both WordPress DB and Pinecone entries. |
| 1036 | 630 | */ |
| 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 | 631 | public function ajax_mxchat_get_entry_content() { |
| 1055 | 632 | check_ajax_referer('mxchat_edit_entry_nonce', 'nonce'); |
| 1056 | 633 | |
| 1057 | 634 | if ( ! current_user_can('manage_options') ) { |
| @@ -1057,19 +634,16 @@ | ||
| 1057 | 634 | if ( ! current_user_can('manage_options') ) { |
| 1058 | 635 | wp_send_json_error( array( 'message' => 'Permission denied.' ) ); |
| 1059 | 636 | } |
| 1060 | 637 | |
| 1061 | - $source_url = $this->sanitize_entry_source_url( isset($_POST['source_url']) ? wp_unslash($_POST['source_url']) : '' ); | |
| 638 | + $source_url = isset($_POST['source_url']) ? sanitize_text_field( wp_unslash($_POST['source_url']) ) : ''; | |
| 1062 | 639 | $entry_id = isset($_POST['entry_id']) ? absint($_POST['entry_id']) : 0; |
| 1063 | 640 | $data_source = isset($_POST['data_source']) ? sanitize_key($_POST['data_source']) : 'wordpress'; |
| 1064 | 641 | $bot_id = isset($_POST['bot_id']) ? sanitize_key($_POST['bot_id']) : 'default'; |
| 1065 | 642 | |
| 1066 | 643 | 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 | 644 | // Pinecone: fetch vectors by source_url, reassemble chunks |
| 1071 | - $content = $this->get_pinecone_entry_content( $source_url, $vector_id, $bot_id ); | |
| 645 | + $content = $this->get_pinecone_entry_content( $source_url, $entry_id, $bot_id ); | |
| 1072 | 646 | } else { |
| 1073 | 647 | // WordPress DB |
| 1074 | 648 | $content = $this->get_wordpress_entry_content( $source_url, $entry_id ); |
| 1075 | 649 | } |
| @@ -1168,56 +742,44 @@ | ||
| 1168 | 742 | if ( empty($host) || empty($api_key) ) { |
| 1169 | 743 | return new WP_Error( 'pinecone_config', 'Pinecone not configured.' ); |
| 1170 | 744 | } |
| 1171 | 745 | |
| 1172 | - // Manual entries carry no source_url (their vector id is a minted manual_* string, | |
| 1173 | - // not md5 of anything the row can hand us) — fetch the exact vector instead. | |
| 1174 | - // '_ungrouped_' is the table view's synthetic display key for such rows. | |
| 1175 | - if ( ( empty($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 ); | |
| 746 | + // List vectors with the source_url prefix | |
| 747 | + $base_id = md5( $source_url ); | |
| 748 | + $vector_ids = array( $base_id ); | |
| 1181 | 749 | |
| 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 | - } | |
| 750 | + // Find chunk vectors | |
| 751 | + $list_url = "https://{$host}/vectors/list"; | |
| 752 | + $list_body = array( 'prefix' => $base_id . '_chunk_', 'limit' => 100 ); | |
| 753 | + if ( ! empty($namespace) ) { | |
| 754 | + $list_body['namespace'] = $namespace; | |
| 755 | + } | |
| 1190 | 756 | |
| 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 | - ) ); | |
| 757 | + $list_resp = wp_remote_post( $list_url, array( | |
| 758 | + 'headers' => array( 'Api-Key' => $api_key, 'Content-Type' => 'application/json' ), | |
| 759 | + 'body' => wp_json_encode( $list_body ), | |
| 760 | + 'timeout' => 15, | |
| 761 | + ) ); | |
| 1195 | 762 | |
| 1196 | - if ( ! is_wp_error($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[] = $v['id']; | |
| 1201 | - } | |
| 763 | + if ( ! is_wp_error($list_resp) ) { | |
| 764 | + $list_data = json_decode( wp_remote_retrieve_body($list_resp), true ); | |
| 765 | + if ( ! empty($list_data['vectors']) ) { | |
| 766 | + foreach ( $list_data['vectors'] as $v ) { | |
| 767 | + $vector_ids[] = $v['id']; | |
| 1202 | 768 | } |
| 1203 | 769 | } |
| 1204 | 770 | } |
| 1205 | 771 | |
| 1206 | 772 | // 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 | - } | |
| 773 | + $fetch_url = "https://{$host}/vectors/fetch"; | |
| 774 | + $fetch_body = array( 'ids' => $vector_ids ); | |
| 1214 | 775 | if ( ! empty($namespace) ) { |
| 1215 | - $fetch_query[] = 'namespace=' . rawurlencode( $namespace ); | |
| 776 | + $fetch_body['namespace'] = $namespace; | |
| 1216 | 777 | } |
| 1217 | 778 | |
| 1218 | - $fetch_resp = wp_remote_get( "https://{$host}/vectors/fetch?" . implode( '&', $fetch_query ), array( | |
| 1219 | - 'headers' => array( 'Api-Key' => $api_key, 'accept' => 'application/json' ), | |
| 779 | + $fetch_resp = wp_remote_post( $fetch_url, array( | |
| 780 | + 'headers' => array( 'Api-Key' => $api_key, 'Content-Type' => 'application/json' ), | |
| 781 | + 'body' => wp_json_encode( $fetch_body ), | |
| 1220 | 782 | 'timeout' => 15, |
| 1221 | 783 | ) ); |
| 1222 | 784 | |
| 1223 | 785 | if ( is_wp_error($fetch_resp) ) { |
| @@ -1252,235 +814,8 @@ | ||
| 1252 | 814 | ); |
| 1253 | 815 | } |
| 1254 | 816 | |
| 1255 | 817 | /** |
| 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 | 818 | * AJAX: Save edited content — re-chunks and re-embeds as needed. |
| 1484 | 819 | * Works for both WordPress DB and Pinecone entries. |
| 1485 | 820 | */ |
| 1486 | 821 | public function ajax_mxchat_save_entry_content() { |
| @@ -1489,9 +824,9 @@ | ||
| 1489 | 824 | if ( ! current_user_can('manage_options') ) { |
| 1490 | 825 | wp_send_json_error( array( 'message' => 'Permission denied.' ) ); |
| 1491 | 826 | } |
| 1492 | 827 | |
| 1493 | - $source_url = $this->sanitize_entry_source_url( isset($_POST['source_url']) ? wp_unslash($_POST['source_url']) : '' ); | |
| 828 | + $source_url = isset($_POST['source_url']) ? sanitize_text_field( wp_unslash($_POST['source_url']) ) : ''; | |
| 1494 | 829 | $entry_id = isset($_POST['entry_id']) ? absint($_POST['entry_id']) : 0; |
| 1495 | 830 | $content = isset($_POST['content']) ? wp_kses_post( wp_unslash($_POST['content']) ) : ''; |
| 1496 | 831 | $data_source = isset($_POST['data_source']) ? sanitize_key($_POST['data_source']) : 'wordpress'; |
| 1497 | 832 | $bot_id = isset($_POST['bot_id']) ? sanitize_key($_POST['bot_id']) : 'default'; |
| @@ -1512,74 +847,38 @@ | ||
| 1512 | 847 | if ( empty($api_key) ) { |
| 1513 | 848 | $api_key = $options['api_key'] ?? ''; |
| 1514 | 849 | } |
| 1515 | 850 | |
| 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; | |
| 851 | + global $wpdb; | |
| 852 | + $table = $wpdb->prefix . 'mxchat_system_prompt_content'; | |
| 1523 | 853 | |
| 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, $content_type ); | |
| 854 | + // If source_url is empty but we have an entry_id, look it up | |
| 855 | + if ( empty($source_url) && $entry_id > 0 && $data_source === 'wordpress' ) { | |
| 856 | + $row = $wpdb->get_row( $wpdb->prepare( "SELECT source_url FROM {$table} WHERE id = %d", $entry_id ) ); | |
| 857 | + if ( $row && ! empty($row->source_url) ) { | |
| 858 | + $source_url = $row->source_url; | |
| 1549 | 859 | } |
| 1550 | - } else { | |
| 1551 | - global $wpdb; | |
| 1552 | - $table = $wpdb->prefix . 'mxchat_system_prompt_content'; | |
| 860 | + } | |
| 1553 | 861 | |
| 1554 | - // If source_url is empty but we have an entry_id, look it up | |
| 1555 | - if ( empty($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 | - } | |
| 862 | + // For manual entries (no source_url or mxchat:// prefix), delete the old entry by ID first | |
| 863 | + // so submit_content_to_db creates a replacement instead of a duplicate | |
| 864 | + // Also treat legacy mxchat.ai source URLs as manual — old bug assigned the site URL to manual entries | |
| 865 | + $is_legacy_manual = !empty($source_url) && strpos($source_url, 'mxchat.ai') !== false && strpos($source_url, 'mxchat://') !== 0; | |
| 866 | + if ( $entry_id > 0 && (empty($source_url) || strpos($source_url, 'mxchat://') === 0 || $is_legacy_manual) ) { | |
| 867 | + $wpdb->delete( $table, array( 'id' => $entry_id ), array( '%d' ) ); | |
| 868 | + // Clear legacy URL so submit_content_to_db generates a unique mxchat:// identifier | |
| 869 | + // instead of reusing the shared URL (which would mass-delete other entries with the same URL) | |
| 870 | + if ( $is_legacy_manual ) { | |
| 871 | + $source_url = ''; | |
| 1560 | 872 | } |
| 873 | + } | |
| 1561 | 874 | |
| 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 | - } | |
| 875 | + // Use the existing submit_content_to_db which handles chunking, Pinecone, and WP DB | |
| 876 | + $vector_id = ! empty($source_url) ? md5($source_url) : md5('mxchat_manual_' . $entry_id); | |
| 1574 | 877 | |
| 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); | |
| 878 | + // submit_content_to_db already handles: delete old chunks → re-chunk → re-embed → store | |
| 879 | + $result = MxChat_Utils::submit_content_to_db( $content, $source_url, $api_key, $vector_id, $bot_id, $content_type ); | |
| 1577 | 880 | |
| 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, $content_type ); | |
| 1580 | - } | |
| 1581 | - | |
| 1582 | 881 | if ( is_wp_error($result) ) { |
| 1583 | 882 | wp_send_json_error( array( 'message' => $result->get_error_message() ) ); |
| 1584 | 883 | } |
| 1585 | 884 | |
| @@ -1665,34 +964,42 @@ | ||
| 1665 | 964 | |
| 1666 | 965 | // Get bot_id from form submission |
| 1667 | 966 | $bot_id = isset($_POST['bot_id']) ? sanitize_key($_POST['bot_id']) : 'default'; |
| 1668 | 967 | |
| 1669 | - // Get bot-specific options and validate the embedding decision — | |
| 1670 | - // custom-provider-aware (plan cbd5fd). | |
| 968 | + // Get bot-specific options and validate API key | |
| 1671 | 969 | $bot_options = $this->get_bot_options($bot_id); |
| 1672 | 970 | $options = !empty($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); | |
| 971 | + $selected_model = $options['embedding_model'] ?? 'text-embedding-ada-002'; | |
| 972 | + | |
| 973 | + if (strpos($selected_model, 'voyage') === 0) { | |
| 974 | + $api_key = $options['voyage_api_key'] ?? ''; | |
| 975 | + $provider_name = 'Voyage AI'; | |
| 976 | + } elseif (strpos($selected_model, 'gemini-embedding') === 0) { | |
| 977 | + $api_key = $options['gemini_api_key'] ?? ''; | |
| 978 | + $provider_name = 'Google Gemini'; | |
| 979 | + } else { | |
| 980 | + $api_key = $options['api_key'] ?? ''; | |
| 981 | + $provider_name = 'OpenAI'; | |
| 982 | + } | |
| 983 | + | |
| 984 | + if (empty($api_key)) { | |
| 985 | + $error_message = sprintf( | |
| 986 | + esc_html__('%s API key is not configured. Please add your API key in the settings before submitting content.', 'mxchat'), | |
| 987 | + $provider_name | |
| 988 | + ); | |
| 989 | + set_transient('mxchat_admin_notice_error', $error_message, 30); | |
| 1677 | 990 | wp_safe_redirect(esc_url(admin_url('admin.php?page=mxchat-prompts'))); |
| 1678 | 991 | exit; |
| 1679 | 992 | } |
| 1680 | - $api_key = $preflight['api_key']; | |
| 1681 | 993 | |
| 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. | |
| 994 | + // Fetch URL — use browser-like headers so servers with bot protection don't block us | |
| 1689 | 995 | $response = wp_remote_get($submitted_url, array( |
| 1690 | 996 | 'timeout' => 30, |
| 1691 | 997 | 'sslverify' => false, |
| 1692 | - 'user-agent' => mxchat_ingest_user_agent(), | |
| 998 | + 'user-agent' => 'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/131.0.0.0 Safari/537.36', | |
| 1693 | 999 | 'headers' => array( |
| 1694 | 1000 | 'Accept' => 'text/html,application/xhtml+xml,application/xml;q=0.9,*/*;q=0.8', |
| 1001 | + 'Accept-Language' => 'en-US,en;q=0.9', | |
| 1695 | 1002 | ), |
| 1696 | 1003 | )); |
| 1697 | 1004 | |
| 1698 | 1005 | if (is_wp_error($response) || wp_remote_retrieve_response_code($response) !== 200) { |
| @@ -1763,22 +1070,12 @@ | ||
| 1763 | 1070 | esc_html__('Sitemap queued for processing. Processing will start automatically.', 'mxchat'), |
| 1764 | 1071 | 30 |
| 1765 | 1072 | ); |
| 1766 | 1073 | } 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 | - } | |
| 1074 | + set_transient('mxchat_admin_notice_error', | |
| 1075 | + esc_html__('Failed to queue sitemap processing. Please check the status below for details.', 'mxchat'), | |
| 1076 | + 30 | |
| 1077 | + ); | |
| 1781 | 1078 | } |
| 1782 | 1079 | |
| 1783 | 1080 | wp_safe_redirect(esc_url(admin_url('admin.php?page=mxchat-prompts'))); |
| 1784 | 1081 | exit; |
| @@ -1921,140 +1218,8 @@ | ||
| 1921 | 1218 | * |
| 1922 | 1219 | * @param string $content The content containing shortcodes |
| 1923 | 1220 | * @return string Content with shortcode tags removed but inner content preserved |
| 1924 | 1221 | */ |
| 1925 | -/** | |
| 1926 | - * Single-pass HTML entity decode for text entering the knowledge base. | |
| 1927 | - * The corpus should hold what a human reads: a stored `&` consumes | |
| 1928 | - * extra tokens, distorts the vector away from the form a visitor's | |
| 1929 | - * question uses, and can be quoted back verbatim in an answer. | |
| 1930 | - * Deliberately NOT looped to a fixed point — a stored `&amp;` is a | |
| 1931 | - * legitimate literal `&` and must not collapse further (data loss). | |
| 1932 | - * UTF-8 charset keeps multibyte (CJK/RTL) text untouched. Both assembly | |
| 1933 | - * paths call this at their output points so the treatment cannot drift. | |
| 1934 | - * (Plan d2c92e.) | |
| 1935 | - */ | |
| 1936 | -private function mxchat_decode_entities_for_indexing($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, $max_raw); | |
| 2055 | -} | |
| 2056 | - | |
| 2057 | 1222 | private function strip_shortcode_tags_preserve_content($content) { |
| 2058 | 1223 | // Single-pass regex removes all shortcode brackets: [tag], [tag attr="val"], [/tag], [tag /] |
| 2059 | 1224 | // Content between tags is inherently preserved since only brackets are targeted |
| 2060 | 1225 | $result = preg_replace('/\[\/?\w[\w-]*[^\]]*\]/', '', $content); |
| @@ -2098,37 +1263,24 @@ | ||
| 2098 | 1263 | |
| 2099 | 1264 | // Ensure valid UTF-8 encoding |
| 2100 | 1265 | $content = wp_check_invalid_utf8($content); |
| 2101 | 1266 | |
| 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 (میروم → می روم) and broke Hindi conjuncts (क्ष → क् ष). | |
| 2124 | - // Do not add one back; the failure mode of an allowlist is exactly this. | |
| 2125 | - | |
| 2126 | - // Limit to reasonable length if needed (byte limit — MySQL TEXT is byte-sized, | |
| 2127 | - // but cut on a character boundary so a multibyte char is never split mid-sequence) | |
| 1267 | + // Remove any extremely long strings without spaces (often garbage) | |
| 1268 | + $content = preg_replace('/\S{300,}/', ' ', $content); | |
| 1269 | + | |
| 1270 | + // Replace problematic characters that often cause database issues | |
| 1271 | + $content = preg_replace('/[\x{10000}-\x{10FFFF}]/u', '', $content); // Remove emoji and other high Unicode characters | |
| 1272 | + | |
| 1273 | + // Replace any remaining potentially problematic characters with spaces | |
| 1274 | + // BUT preserve newlines by temporarily replacing them | |
| 1275 | + $content = str_replace("\n", "NEWLINE_PLACEHOLDER", $content); | |
| 1276 | + $content = preg_replace('/[^\p{L}\p{N}\p{P}\p{Z}\p{Sm}]/u', ' ', $content); | |
| 1277 | + $content = str_replace("NEWLINE_PLACEHOLDER", "\n", $content); | |
| 1278 | + | |
| 1279 | + // Limit to reasonable length if needed | |
| 2128 | 1280 | $max_length = 65000; // Just under MySQL TEXT field limit |
| 2129 | 1281 | if (strlen($content) > $max_length) { |
| 2130 | - $content = mb_strcut($content, 0, $max_length, 'UTF-8'); | |
| 1282 | + $content = substr($content, 0, $max_length); | |
| 2131 | 1283 | } |
| 2132 | 1284 | |
| 2133 | 1285 | //error_log('[MXCHAT-SANITIZE] Sanitized content preview: ' . substr($content, 0, 500) . '...'); |
| 2134 | 1286 | return $content; |
| @@ -2966,19 +2118,10 @@ | ||
| 2966 | 2118 | <span style="color: var(--mxch-text-muted);"><?php esc_html_e('Manual Content', 'mxchat'); ?></span> |
| 2967 | 2119 | <?php endif; ?> |
| 2968 | 2120 | </td> |
| 2969 | 2121 | <td class="mxchat-actions-cell" style="padding: 12px 16px; white-space: nowrap;"> |
| 2122 | + <?php if ($data_source !== 'pinecone') : ?> | |
| 2970 | 2123 | <button type="button" |
| 2971 | - class="mxch-btn mxch-btn-ghost mxch-btn-sm mxchat-inspect-entry-btn" | |
| 2972 | - data-source-url="<?php echo esc_attr($source_url); ?>" | |
| 2973 | - data-entry-id="<?php echo esc_attr($first_prompt->id); ?>" | |
| 2974 | - data-data-source="<?php echo esc_attr($data_source); ?>" | |
| 2975 | - data-bot-id="<?php echo esc_attr($current_bot_id); ?>" | |
| 2976 | - data-nonce="<?php echo wp_create_nonce('mxchat_inspect_entry_nonce'); ?>" | |
| 2977 | - title="<?php esc_attr_e('View indexed content', 'mxchat'); ?>"> | |
| 2978 | - <span class="dashicons dashicons-visibility" style="font-size: 14px;"></span> | |
| 2979 | - </button> | |
| 2980 | - <button type="button" | |
| 2981 | 2124 | class="mxch-btn mxch-btn-ghost mxch-btn-sm mxchat-edit-entry-btn" |
| 2982 | 2125 | data-source-url="<?php echo esc_attr($source_url); ?>" |
| 2983 | 2126 | data-entry-id="<?php echo esc_attr($first_prompt->id); ?>" |
| 2984 | 2127 | data-data-source="<?php echo esc_attr($data_source); ?>" |
| @@ -2986,8 +2129,9 @@ | ||
| 2986 | 2129 | data-nonce="<?php echo wp_create_nonce('mxchat_edit_entry_nonce'); ?>" |
| 2987 | 2130 | title="<?php esc_attr_e('Edit content', 'mxchat'); ?>"> |
| 2988 | 2131 | <span class="dashicons dashicons-edit" style="font-size: 14px;"></span> |
| 2989 | 2132 | </button> |
| 2133 | + <?php endif; ?> | |
| 2990 | 2134 | <button type="button" |
| 2991 | 2135 | class="mxch-btn mxch-btn-ghost mxch-btn-sm delete-button-group" |
| 2992 | 2136 | data-source-url="<?php echo esc_attr($source_url); ?>" |
| 2993 | 2137 | data-chunk-count="<?php echo esc_attr($chunk_count); ?>" |
| @@ -3119,29 +2263,9 @@ | ||
| 3119 | 2263 | <?php else : ?> |
| 3120 | 2264 | <span style="color: var(--mxch-text-muted);"><?php esc_html_e('Manual', 'mxchat'); ?></span> |
| 3121 | 2265 | <?php endif; ?> |
| 3122 | 2266 | </td> |
| 3123 | - <td style="padding: 12px 16px; white-space: nowrap;"> | |
| 3124 | - <button type="button" | |
| 3125 | - class="mxch-btn mxch-btn-ghost mxch-btn-sm mxchat-inspect-entry-btn" | |
| 3126 | - data-source-url="<?php echo esc_attr($prompt->source_url ?? ''); ?>" | |
| 3127 | - data-entry-id="<?php echo esc_attr($prompt->id); ?>" | |
| 3128 | - data-data-source="<?php echo esc_attr($data_source); ?>" | |
| 3129 | - data-bot-id="<?php echo esc_attr($current_bot_id); ?>" | |
| 3130 | - data-nonce="<?php echo wp_create_nonce('mxchat_inspect_entry_nonce'); ?>" | |
| 3131 | - title="<?php esc_attr_e('View indexed content', 'mxchat'); ?>"> | |
| 3132 | - <span class="dashicons dashicons-visibility" style="font-size: 14px;"></span> | |
| 3133 | - </button> | |
| 3134 | - <button type="button" | |
| 3135 | - class="mxch-btn mxch-btn-ghost mxch-btn-sm mxchat-edit-entry-btn" | |
| 3136 | - data-source-url="<?php echo esc_attr($prompt->source_url ?? ''); ?>" | |
| 3137 | - data-entry-id="<?php echo esc_attr($prompt->id); ?>" | |
| 3138 | - data-data-source="<?php echo esc_attr($data_source); ?>" | |
| 3139 | - data-bot-id="<?php echo esc_attr($current_bot_id); ?>" | |
| 3140 | - data-nonce="<?php echo wp_create_nonce('mxchat_edit_entry_nonce'); ?>" | |
| 3141 | - title="<?php esc_attr_e('Edit content', 'mxchat'); ?>"> | |
| 3142 | - <span class="dashicons dashicons-edit" style="font-size: 14px;"></span> | |
| 3143 | - </button> | |
| 2267 | + <td style="padding: 12px 16px;"> | |
| 3144 | 2268 | <button type="button" class="mxch-btn mxch-btn-ghost mxch-btn-sm delete-button-ajax" data-vector-id="<?php echo esc_attr($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);"> |
| 3145 | 2269 | <span class="dashicons dashicons-trash" style="font-size: 14px;"></span> |
| 3146 | 2270 | </button> |
| 3147 | 2271 | </td> |
| @@ -3714,9 +2838,9 @@ | ||
| 3714 | 2838 | $response = wp_remote_head($url, array( |
| 3715 | 2839 | 'timeout' => 10, |
| 3716 | 2840 | 'sslverify' => false, |
| 3717 | 2841 | 'redirection' => 1, |
| 3718 | - 'user-agent' => mxchat_ingest_user_agent(), | |
| 2842 | + 'user-agent' => 'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/131.0.0.0 Safari/537.36', | |
| 3719 | 2843 | )); |
| 3720 | 2844 | |
| 3721 | 2845 | if (!is_wp_error($response) && wp_remote_retrieve_response_code($response) === 200) { |
| 3722 | 2846 | // Found a sitemap index - parse it to get sub-sitemaps |
| @@ -3776,11 +2900,12 @@ | ||
| 3776 | 2900 | |
| 3777 | 2901 | $response = wp_remote_get($url, array( |
| 3778 | 2902 | 'timeout' => 30, |
| 3779 | 2903 | 'sslverify' => false, |
| 3780 | - 'user-agent' => mxchat_ingest_user_agent(), | |
| 2904 | + 'user-agent' => 'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/131.0.0.0 Safari/537.36', | |
| 3781 | 2905 | 'headers' => array( |
| 3782 | 2906 | 'Accept' => 'text/html,application/xhtml+xml,application/xml;q=0.9,*/*;q=0.8', |
| 2907 | + 'Accept-Language' => 'en-US,en;q=0.9', | |
| 3783 | 2908 | ), |
| 3784 | 2909 | )); |
| 3785 | 2910 | |
| 3786 | 2911 | if (is_wp_error($response)) { |
| @@ -3834,11 +2959,12 @@ | ||
| 3834 | 2959 | private function get_sitemap_url_count($url) { |
| 3835 | 2960 | $response = wp_remote_get($url, array( |
| 3836 | 2961 | 'timeout' => 30, |
| 3837 | 2962 | 'sslverify' => false, |
| 3838 | - 'user-agent' => mxchat_ingest_user_agent(), | |
| 2963 | + 'user-agent' => 'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/131.0.0.0 Safari/537.36', | |
| 3839 | 2964 | 'headers' => array( |
| 3840 | 2965 | 'Accept' => 'text/html,application/xhtml+xml,application/xml;q=0.9,*/*;q=0.8', |
| 2966 | + 'Accept-Language' => 'en-US,en;q=0.9', | |
| 3841 | 2967 | ), |
| 3842 | 2968 | )); |
| 3843 | 2969 | |
| 3844 | 2970 | if (is_wp_error($response)) { |
| @@ -3864,9 +2990,9 @@ | ||
| 3864 | 2990 | |
| 3865 | 2991 | $response = wp_remote_get($robots_url, array( |
| 3866 | 2992 | 'timeout' => 15, |
| 3867 | 2993 | 'sslverify' => false, |
| 3868 | - 'user-agent' => mxchat_ingest_user_agent(), | |
| 2994 | + 'user-agent' => 'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/131.0.0.0 Safari/537.36', | |
| 3869 | 2995 | )); |
| 3870 | 2996 | |
| 3871 | 2997 | if (is_wp_error($response)) { |
| 3872 | 2998 | return $sitemaps; |
| @@ -4357,12 +3483,20 @@ | ||
| 4357 | 3483 | |
| 4358 | 3484 | // Get bot_id from request |
| 4359 | 3485 | $bot_id = isset($_POST['bot_id']) ? sanitize_key($_POST['bot_id']) : 'default'; |
| 4360 | 3486 | |
| 4361 | - // ACF→PDF extraction is an install-level setting (Knowledge → ACF Fields, | |
| 4362 | - // plan 11720c). The import modal shows a passive status line pointing | |
| 4363 | - // there; the old per-batch checkbox and its remembered default are gone. | |
| 4364 | - $extract_acf_pdfs = get_option('mxchat_acf_pdf_extraction', '0') === '1'; | |
| 3487 | + // ACF→PDF extraction is opt-in per import batch. Persist the last-used value so users | |
| 3488 | + // don't re-check on every batch; the default is OFF for installs that haven't set it. | |
| 3489 | + $extract_acf_pdfs = !empty($_POST['extract_acf_pdfs']) && $_POST['extract_acf_pdfs'] !== 'false'; | |
| 3490 | + $mxchat_options = get_option('mxchat_options', array()); | |
| 3491 | + if (!is_array($mxchat_options)) { | |
| 3492 | + $mxchat_options = array(); | |
| 3493 | + } | |
| 3494 | + $prior_default = !empty($mxchat_options['acf_pdf_extract_default']); | |
| 3495 | + if ($prior_default !== $extract_acf_pdfs) { | |
| 3496 | + $mxchat_options['acf_pdf_extract_default'] = $extract_acf_pdfs ? 1 : 0; | |
| 3497 | + update_option('mxchat_options', $mxchat_options); | |
| 3498 | + } | |
| 4365 | 3499 | |
| 4366 | 3500 | // Process only ONE post at a time to avoid request size issues |
| 4367 | 3501 | $post_id = reset($post_ids); |
| 4368 | 3502 | $post = get_post($post_id); |
| @@ -4371,34 +3505,173 @@ | ||
| 4371 | 3505 | wp_send_json_error('Post not found'); |
| 4372 | 3506 | exit; |
| 4373 | 3507 | } |
| 4374 | 3508 | |
| 4375 | - /** | |
| 4376 | - * Allow developers to modify post data before processing into the knowledge base. | |
| 4377 | - * Applied on BOTH content-preparation paths (this manual bulk import and the | |
| 4378 | - * auto-sync path in mxchat_handle_post_update) with the same signature, so a | |
| 4379 | - * callback registered once covers every indexing route. Purely additive — | |
| 4380 | - * zero behaviour change when unhooked. | |
| 4381 | - * | |
| 4382 | - * @param WP_Post $post The post about to be indexed. | |
| 4383 | - * @param string $bot_id Bot context for this import. | |
| 4384 | - */ | |
| 3509 | + // Allow developers to modify post data before processing into knowledge base | |
| 4385 | 3510 | $post = apply_filters('mxchat_before_process_post', $post, $bot_id); |
| 4386 | - if (!($post instanceof WP_Post)) { | |
| 4387 | - $post = get_post($post_id); // defend against a bad callback return | |
| 3511 | + | |
| 3512 | + // Get content including title, short description (for WooCommerce), and main content | |
| 3513 | + $content = $post->post_title . "\n\n"; | |
| 3514 | + | |
| 3515 | + // Add short description if it exists (WooCommerce products use post_excerpt for short description) | |
| 3516 | + if (!empty($post->post_excerpt)) { | |
| 3517 | + // Remove shortcode tags but preserve content inside them | |
| 3518 | + $clean_excerpt = $this->strip_shortcode_tags_preserve_content($post->post_excerpt); | |
| 3519 | + $content .= "Short Description: " . wp_strip_all_tags($clean_excerpt) . "\n\n"; | |
| 4388 | 3520 | } |
| 4389 | 3521 | |
| 4390 | - // Assemble the indexable text via the shared post-kind assembler (a3d60c). | |
| 4391 | - // Bulk import reads raw post fields, has always included product custom tabs, | |
| 4392 | - // and passes the install-level ACF→PDF option. | |
| 4393 | - $prepared = $this->mxchat_prepare_post_content_for_indexing($post_id, $post, array( | |
| 4394 | - 'read_display' => false, | |
| 4395 | - 'extract_acf_pdfs' => $extract_acf_pdfs, | |
| 4396 | - 'include_product_tabs' => true, | |
| 4397 | - )); | |
| 4398 | - $content = $prepared['content']; | |
| 4399 | - $pdf_extracted_count = $prepared['pdf_extracted_count']; | |
| 3522 | + // Add main content - remove shortcode tags but preserve content inside them | |
| 3523 | + $clean_content = $this->strip_shortcode_tags_preserve_content($post->post_content); | |
| 3524 | + $content .= wp_strip_all_tags($clean_content); | |
| 4400 | 3525 | |
| 3526 | + // ADD WOOCOMMERCE PRODUCT DATA (pricing, stock, categories, custom tabs) | |
| 3527 | + if (get_post_type($post_id) === 'product' && class_exists('WooCommerce')) { | |
| 3528 | + $product = wc_get_product($post_id); | |
| 3529 | + | |
| 3530 | + if ($product) { | |
| 3531 | + // Get pricing information | |
| 3532 | + $regular_price = $product->get_regular_price(); | |
| 3533 | + $sale_price = $product->get_sale_price(); | |
| 3534 | + $price = $product->get_price(); | |
| 3535 | + $sku = $product->get_sku(); | |
| 3536 | + | |
| 3537 | + // Get currency symbol | |
| 3538 | + $currency_symbol = get_woocommerce_currency_symbol(); | |
| 3539 | + | |
| 3540 | + // Add pricing information | |
| 3541 | + $content .= "\n"; | |
| 3542 | + if (!empty($regular_price)) { | |
| 3543 | + $content .= "Price: " . $currency_symbol . $regular_price . "\n"; | |
| 3544 | + } elseif (!empty($price)) { | |
| 3545 | + $content .= "Price: " . $currency_symbol . $price . "\n"; | |
| 3546 | + } | |
| 3547 | + | |
| 3548 | + if (!empty($sale_price) && $sale_price !== $regular_price) { | |
| 3549 | + $content .= "Sale Price: " . $currency_symbol . $sale_price . "\n"; | |
| 3550 | + } | |
| 3551 | + | |
| 3552 | + // Handle variable products - show price range | |
| 3553 | + if ($product->is_type('variable')) { | |
| 3554 | + $min_price = $product->get_variation_price('min'); | |
| 3555 | + $max_price = $product->get_variation_price('max'); | |
| 3556 | + if ($min_price !== $max_price) { | |
| 3557 | + $content .= "Price Range: " . $currency_symbol . $min_price . " - " . $currency_symbol . $max_price . "\n"; | |
| 3558 | + } | |
| 3559 | + } | |
| 3560 | + | |
| 3561 | + if (!empty($sku)) { | |
| 3562 | + $content .= "SKU: " . $sku . "\n"; | |
| 3563 | + } | |
| 3564 | + | |
| 3565 | + // Get product categories | |
| 3566 | + $categories = wp_get_post_terms($post_id, 'product_cat', array('fields' => 'names')); | |
| 3567 | + if (!empty($categories) && !is_wp_error($categories)) { | |
| 3568 | + $content .= "Categories: " . implode(', ', $categories) . "\n"; | |
| 3569 | + } | |
| 3570 | + } | |
| 3571 | + | |
| 3572 | + // Get Custom Product Tabs (supports "Custom Product Tabs for WooCommerce" by Code Parrots) | |
| 3573 | + $custom_tabs = get_post_meta($post_id, 'yikes_woo_products_tabs', true); | |
| 3574 | + if (!empty($custom_tabs) && is_array($custom_tabs)) { | |
| 3575 | + foreach ($custom_tabs as $tab) { | |
| 3576 | + $tab_title = isset($tab['title']) ? $tab['title'] : (isset($tab['tab_title']) ? $tab['tab_title'] : ''); | |
| 3577 | + $tab_content = isset($tab['content']) ? $tab['content'] : ''; | |
| 3578 | + | |
| 3579 | + if (!empty($tab_title) && !empty($tab_content)) { | |
| 3580 | + $content .= "\n" . $tab_title . ": " . wp_strip_all_tags($tab_content) . "\n"; | |
| 3581 | + } | |
| 3582 | + } | |
| 3583 | + } | |
| 3584 | + | |
| 3585 | + // Also check for reusable/saved tabs applied to this product | |
| 3586 | + $applied_saved_tabs = get_post_meta($post_id, 'yikes_woo_reusable_products_tabs_applied', true); | |
| 3587 | + if (!empty($applied_saved_tabs) && is_array($applied_saved_tabs)) { | |
| 3588 | + $saved_tabs = get_option('yikes_woo_reusable_products_tabs', array()); | |
| 3589 | + if (!empty($saved_tabs) && is_array($saved_tabs)) { | |
| 3590 | + foreach ($applied_saved_tabs as $saved_tab_id) { | |
| 3591 | + if (isset($saved_tabs[$saved_tab_id])) { | |
| 3592 | + $tab = $saved_tabs[$saved_tab_id]; | |
| 3593 | + $tab_title = isset($tab['title']) ? $tab['title'] : (isset($tab['tab_title']) ? $tab['tab_title'] : ''); | |
| 3594 | + $tab_content = isset($tab['content']) ? $tab['content'] : ''; | |
| 3595 | + | |
| 3596 | + if (!empty($tab_title) && !empty($tab_content)) { | |
| 3597 | + $content .= "\n" . $tab_title . ": " . wp_strip_all_tags($tab_content) . "\n"; | |
| 3598 | + } | |
| 3599 | + } | |
| 3600 | + } | |
| 3601 | + } | |
| 3602 | + } | |
| 3603 | + } | |
| 3604 | + | |
| 3605 | + // ADD ACF FIELDS SUPPORT | |
| 3606 | + $acf_fields = $this->mxchat_get_acf_fields_for_post($post_id); | |
| 3607 | + $pdf_extracted_count = 0; | |
| 3608 | + if (!empty($acf_fields)) { | |
| 3609 | + $acf_content_parts = array(); | |
| 3610 | + $pdf_attachment_ids = array(); | |
| 3611 | + | |
| 3612 | + foreach ($acf_fields as $field_name => $field_value) { | |
| 3613 | + $formatted_value = $this->mxchat_format_acf_field_value($field_value, $field_name, $post_id); | |
| 3614 | + | |
| 3615 | + if (!empty($formatted_value)) { | |
| 3616 | + $field_label = ucwords(str_replace('_', ' ', $field_name)); | |
| 3617 | + $acf_content_parts[] = $field_label . ": " . $formatted_value; | |
| 3618 | + } | |
| 3619 | + | |
| 3620 | + // Walk this field's value tree for any PDF attachment references and queue them for extraction. | |
| 3621 | + // Only when the user opted into ACF→PDF extraction for this batch; otherwise the ACF text | |
| 3622 | + // still lands in the KB but the heavier PDF parsing is skipped. | |
| 3623 | + if ($extract_acf_pdfs) { | |
| 3624 | + $this->mxchat_collect_pdf_attachment_ids_from_acf_value($field_value, $pdf_attachment_ids); | |
| 3625 | + } | |
| 3626 | + } | |
| 3627 | + | |
| 3628 | + if (!empty($acf_content_parts)) { | |
| 3629 | + $content .= "\n\n" . implode("\n", $acf_content_parts); | |
| 3630 | + } | |
| 3631 | + | |
| 3632 | + // Extract text from each unique PDF found in ACF fields and append as a labeled section | |
| 3633 | + if ($extract_acf_pdfs && !empty($pdf_attachment_ids)) { | |
| 3634 | + $pdf_attachment_ids = array_unique(array_filter(array_map('intval', $pdf_attachment_ids))); | |
| 3635 | + $pdf_sections = array(); | |
| 3636 | + foreach ($pdf_attachment_ids as $att_id) { | |
| 3637 | + $pdf_text = $this->mxchat_extract_pdf_text_by_attachment_id($att_id); | |
| 3638 | + if (!empty($pdf_text)) { | |
| 3639 | + $pdf_title = get_the_title($att_id); | |
| 3640 | + $pdf_url = wp_get_attachment_url($att_id); | |
| 3641 | + $header = 'PDF Attachment'; | |
| 3642 | + if (!empty($pdf_title)) { | |
| 3643 | + $header .= ': ' . $pdf_title; | |
| 3644 | + } | |
| 3645 | + if (!empty($pdf_url)) { | |
| 3646 | + $header .= ' (' . $pdf_url . ')'; | |
| 3647 | + } | |
| 3648 | + $pdf_sections[] = $header . "\n" . $pdf_text; | |
| 3649 | + $pdf_extracted_count++; | |
| 3650 | + } | |
| 3651 | + } | |
| 3652 | + if (!empty($pdf_sections)) { | |
| 3653 | + $content .= "\n\nAttached PDFs:\n" . implode("\n\n", $pdf_sections); | |
| 3654 | + } | |
| 3655 | + } | |
| 3656 | + } | |
| 3657 | + | |
| 3658 | + // ADD CUSTOM POST META SUPPORT (whitelisted non-ACF meta fields) | |
| 3659 | + $custom_meta = $this->mxchat_get_whitelisted_post_meta($post_id); | |
| 3660 | + if (!empty($custom_meta)) { | |
| 3661 | + $meta_content_parts = array(); | |
| 3662 | + | |
| 3663 | + foreach ($custom_meta as $meta_key => $meta_value) { | |
| 3664 | + // Convert meta key to readable label | |
| 3665 | + $meta_label = ucwords(str_replace(array('_', '-'), ' ', $meta_key)); | |
| 3666 | + $meta_content_parts[] = $meta_label . ": " . $meta_value; | |
| 3667 | + } | |
| 3668 | + | |
| 3669 | + if (!empty($meta_content_parts)) { | |
| 3670 | + $content .= "\n\n" . implode("\n", $meta_content_parts); | |
| 3671 | + } | |
| 3672 | + } | |
| 3673 | + | |
| 4401 | 3674 | // Debug logging for WordPress Import content |
| 4402 | 3675 | //error_log('[MXCHAT-WP-IMPORT-DEBUG] Post ID: ' . $post_id . ' Title: ' . $post->post_title); |
| 4403 | 3676 | //error_log('[MXCHAT-WP-IMPORT-DEBUG] Raw post_content length: ' . strlen($post->post_content)); |
| 4404 | 3677 | //error_log('[MXCHAT-WP-IMPORT-DEBUG] Final content length: ' . strlen($content)); |
| @@ -4405,19 +3678,29 @@ | ||
| 4405 | 3678 | //error_log('[MXCHAT-WP-IMPORT-DEBUG] Content preview: ' . substr($content, 0, 300)); |
| 4406 | 3679 | |
| 4407 | 3680 | // Note: Removed 10,000 char limit - chunking now handles large content properly |
| 4408 | 3681 | |
| 4409 | - // Get bot-specific embedding decision — custom-provider-aware (plan cbd5fd) | |
| 3682 | + // Get bot-specific API key | |
| 4410 | 3683 | $bot_options = $this->get_bot_options($bot_id); |
| 4411 | 3684 | $options = !empty($bot_options) ? $bot_options : get_option('mxchat_options'); |
| 4412 | - | |
| 4413 | - $preflight = MxChat_Utils::embedding_preflight($options); | |
| 4414 | - if (!$preflight['ok']) { | |
| 4415 | - MxChat_Admin::mxchat_log_debug('api_error', $preflight['reason'] . ' (knowledge processing)'); | |
| 4416 | - wp_send_json_error($preflight['reason']); | |
| 3685 | + $selected_model = $options['embedding_model'] ?? 'text-embedding-ada-002'; | |
| 3686 | + | |
| 3687 | + if (strpos($selected_model, 'voyage') === 0) { | |
| 3688 | + $api_key = $options['voyage_api_key'] ?? ''; | |
| 3689 | + $provider_name = 'Voyage AI'; | |
| 3690 | + } elseif (strpos($selected_model, 'gemini-embedding') === 0) { | |
| 3691 | + $api_key = $options['gemini_api_key'] ?? ''; | |
| 3692 | + $provider_name = 'Google Gemini'; | |
| 3693 | + } else { | |
| 3694 | + $api_key = $options['api_key'] ?? ''; | |
| 3695 | + $provider_name = 'OpenAI'; | |
| 3696 | + } | |
| 3697 | + | |
| 3698 | + if (empty($api_key)) { | |
| 3699 | + MxChat_Admin::mxchat_log_debug('api_error', $provider_name . ' API key not configured for knowledge processing'); | |
| 3700 | + wp_send_json_error($provider_name . ' API key not configured'); | |
| 4417 | 3701 | exit; |
| 4418 | 3702 | } |
| 4419 | - $api_key = $preflight['api_key']; | |
| 4420 | 3703 | |
| 4421 | 3704 | $source_url = get_permalink($post_id); |
| 4422 | 3705 | $vector_id = md5($source_url); // Vector ID for Pinecone |
| 4423 | 3706 | |
| @@ -4488,11 +3771,11 @@ | ||
| 4488 | 3771 | // Automatically apply role restriction based on tags |
| 4489 | 3772 | $this->apply_role_restriction_to_post($post_id, $source_url); |
| 4490 | 3773 | |
| 4491 | 3774 | $operation_type = $is_update ? 'update' : 'new'; |
| 4492 | - | |
| 3775 | + | |
| 4493 | 3776 | // Count ACF fields for debugging |
| 4494 | - $acf_field_count = $prepared['acf_fields_found']; | |
| 3777 | + $acf_field_count = count($acf_fields); | |
| 4495 | 3778 | |
| 4496 | 3779 | // Success response with minimal data |
| 4497 | 3780 | wp_send_json_success(array( |
| 4498 | 3781 | 'message' => $operation_type === 'update' ? 'Content updated successfully' : 'Content processed successfully', |
| @@ -4572,9 +3855,9 @@ | ||
| 4572 | 3855 | ); |
| 4573 | 3856 | } else { |
| 4574 | 3857 | // Update WordPress DB |
| 4575 | 3858 | $table_name = $wpdb->prefix . 'mxchat_system_prompt_content'; |
| 4576 | - | |
| 3859 | + | |
| 4577 | 3860 | $wpdb->update( |
| 4578 | 3861 | $table_name, |
| 4579 | 3862 | array('role_restriction' => $highest_role), |
| 4580 | 3863 | array('source_url' => $source_url), |
| @@ -4581,15 +3864,8 @@ | ||
| 4581 | 3864 | array('%s'), |
| 4582 | 3865 | array('%s') |
| 4583 | 3866 | ); |
| 4584 | 3867 | } |
| 4585 | - | |
| 4586 | - // The entry's restriction just changed — keep the OpenAI Vector Store | |
| 4587 | - // mirror consistent: non-public pulls the file (file_search has no | |
| 4588 | - // per-role filtering), public re-mirrors it (plan 15b5c6). | |
| 4589 | - if (class_exists('MxChat_Vectorstore_Manager')) { | |
| 4590 | - MxChat_Vectorstore_Manager::handle_role_change($source_url, 'default', $highest_role); | |
| 4591 | - } | |
| 4592 | 3868 | } |
| 4593 | 3869 | |
| 4594 | 3870 | public function mxchat_get_public_post_types() { |
| 4595 | 3871 | // Get all public post types |
| @@ -4660,63 +3936,64 @@ | ||
| 4660 | 3936 | |
| 4661 | 3937 | return $pinecone_data; |
| 4662 | 3938 | } |
| 4663 | 3939 | public function mxchat_fetch_pinecone_vectors_by_ids($pinecone_options, $vector_ids) { |
| 3940 | + //error_log('=== DEBUG: Starting mxchat_fetch_pinecone_vectors_by_ids ==='); | |
| 3941 | + | |
| 4664 | 3942 | $api_key = $pinecone_options['mxchat_pinecone_api_key'] ?? ''; |
| 4665 | 3943 | $host = $pinecone_options['mxchat_pinecone_host'] ?? ''; |
| 4666 | - $namespace = $pinecone_options['mxchat_pinecone_namespace'] ?? ''; | |
| 4667 | 3944 | |
| 4668 | 3945 | if (empty($api_key) || empty($host) || empty($vector_ids)) { |
| 3946 | + //error_log('DEBUG: Missing parameters for fetch by IDs'); | |
| 4669 | 3947 | return array(); |
| 4670 | 3948 | } |
| 4671 | 3949 | |
| 4672 | 3950 | try { |
| 4673 | - // NOTE (plan 793b82): /vectors/fetch is a GET endpoint with the ids | |
| 4674 | - // repeated in the query string (ids=a&ids=b — http_build_query would | |
| 4675 | - // emit ids[0]=a); the old POST here was answered 200-with-an-empty-body, | |
| 4676 | - // which read as "nothing indexed". Chunked at 100 ids to stay well | |
| 4677 | - // under the measured HTTP 414 URL-length boundary. | |
| 4678 | - $vectors = array(); | |
| 4679 | - foreach (array_chunk(array_values($vector_ids), 100) as $chunk) { | |
| 4680 | - $fetch_query = array(); | |
| 4681 | - foreach ($chunk as $fetch_vid) { | |
| 4682 | - $fetch_query[] = 'ids=' . rawurlencode($fetch_vid); | |
| 4683 | - } | |
| 4684 | - if (!empty($namespace)) { | |
| 4685 | - $fetch_query[] = 'namespace=' . rawurlencode($namespace); | |
| 4686 | - } | |
| 3951 | + $fetch_url = "https://{$host}/vectors/fetch"; | |
| 3952 | + //error_log('DEBUG: Fetch URL: ' . $fetch_url); | |
| 3953 | + //error_log('DEBUG: Fetching ' . count($vector_ids) . ' vector IDs'); | |
| 4687 | 3954 | |
| 4688 | - $response = wp_remote_get("https://{$host}/vectors/fetch?" . implode('&', $fetch_query), array( | |
| 4689 | - 'headers' => array( | |
| 4690 | - 'Api-Key' => $api_key, | |
| 4691 | - 'accept' => 'application/json' | |
| 4692 | - ), | |
| 4693 | - 'timeout' => 30 | |
| 4694 | - )); | |
| 3955 | + // Pinecone fetch API allows fetching specific vectors by ID | |
| 3956 | + $fetch_data = array( | |
| 3957 | + 'ids' => array_values($vector_ids) | |
| 3958 | + ); | |
| 4695 | 3959 | |
| 4696 | - if (is_wp_error($response)) { | |
| 4697 | - error_log('MxChat Pinecone: mxchat_fetch_pinecone_vectors_by_ids GET failed: ' . $response->get_error_message()); | |
| 4698 | - continue; | |
| 4699 | - } | |
| 3960 | + $response = wp_remote_post($fetch_url, array( | |
| 3961 | + 'headers' => array( | |
| 3962 | + 'Api-Key' => $api_key, | |
| 3963 | + 'Content-Type' => 'application/json' | |
| 3964 | + ), | |
| 3965 | + 'body' => json_encode($fetch_data), | |
| 3966 | + 'timeout' => 30 | |
| 3967 | + )); | |
| 4700 | 3968 | |
| 4701 | - if (wp_remote_retrieve_response_code($response) !== 200) { | |
| 4702 | - error_log('MxChat Pinecone: mxchat_fetch_pinecone_vectors_by_ids GET returned HTTP ' . wp_remote_retrieve_response_code($response)); | |
| 4703 | - continue; | |
| 4704 | - } | |
| 3969 | + if (is_wp_error($response)) { | |
| 3970 | + //error_log('DEBUG: Fetch by IDs WP error: ' . $response->get_error_message()); | |
| 3971 | + return array(); | |
| 3972 | + } | |
| 4705 | 3973 | |
| 4706 | - $data = json_decode(wp_remote_retrieve_body($response), true); | |
| 4707 | - if (isset($data['vectors']) && is_array($data['vectors'])) { | |
| 4708 | - $vectors += $data['vectors']; | |
| 4709 | - } | |
| 3974 | + $response_code = wp_remote_retrieve_response_code($response); | |
| 3975 | + //error_log('DEBUG: Fetch response code: ' . $response_code); | |
| 3976 | + | |
| 3977 | + if ($response_code !== 200) { | |
| 3978 | + $error_body = wp_remote_retrieve_body($response); | |
| 3979 | + //error_log('DEBUG: Fetch failed with body: ' . $error_body); | |
| 3980 | + return array(); | |
| 4710 | 3981 | } |
| 4711 | 3982 | |
| 4712 | - if (empty($vectors)) { | |
| 3983 | + $body = wp_remote_retrieve_body($response); | |
| 3984 | + $data = json_decode($body, true); | |
| 3985 | + | |
| 3986 | + //error_log('DEBUG: Fetch response structure: ' . print_r(array_keys($data), true)); | |
| 3987 | + | |
| 3988 | + if (!isset($data['vectors'])) { | |
| 3989 | + //error_log('DEBUG: No vectors key in response'); | |
| 4713 | 3990 | return array(); |
| 4714 | 3991 | } |
| 4715 | 3992 | |
| 4716 | 3993 | $processed_data = array(); |
| 4717 | 3994 | |
| 4718 | - foreach ($vectors as $vector_id => $vector_data) { | |
| 3995 | + foreach ($data['vectors'] as $vector_id => $vector_data) { | |
| 4719 | 3996 | $metadata = $vector_data['metadata'] ?? array(); |
| 4720 | 3997 | $source_url = $metadata['source_url'] ?? ''; |
| 4721 | 3998 | |
| 4722 | 3999 | if (!empty($source_url)) { |
| @@ -4787,11 +4064,8 @@ | ||
| 4787 | 4064 | */ |
| 4788 | 4065 | public function mxchat_scan_pinecone_for_processed_content($pinecone_options) { |
| 4789 | 4066 | $api_key = $pinecone_options['mxchat_pinecone_api_key'] ?? ''; |
| 4790 | 4067 | $host = $pinecone_options['mxchat_pinecone_host'] ?? ''; |
| 4791 | - // plan 793b82: this scan was namespace-blind — on a namespaced setup it | |
| 4792 | - // surveyed the default namespace and reported the wrong content as indexed. | |
| 4793 | - $namespace = $pinecone_options['mxchat_pinecone_namespace'] ?? ''; | |
| 4794 | 4068 | |
| 4795 | 4069 | if (empty($api_key) || empty($host)) { |
| 4796 | 4070 | return array(); |
| 4797 | 4071 | } |
| @@ -4826,12 +4100,8 @@ | ||
| 4826 | 4100 | 'topK' => 10000, |
| 4827 | 4101 | 'vector' => $random_vector |
| 4828 | 4102 | ); |
| 4829 | 4103 | |
| 4830 | - if (!empty($namespace)) { | |
| 4831 | - $query_data['namespace'] = $namespace; | |
| 4832 | - } | |
| 4833 | - | |
| 4834 | 4104 | $response = wp_remote_post($query_url, array( |
| 4835 | 4105 | 'headers' => array( |
| 4836 | 4106 | 'Api-Key' => $api_key, |
| 4837 | 4107 | 'Content-Type' => 'application/json' |
| @@ -4844,9 +4114,9 @@ | ||
| 4844 | 4114 | continue; |
| 4845 | 4115 | } |
| 4846 | 4116 | |
| 4847 | 4117 | $response_code = wp_remote_retrieve_response_code($response); |
| 4848 | - | |
| 4118 | + | |
| 4849 | 4119 | if ($response_code !== 200) { |
| 4850 | 4120 | continue; |
| 4851 | 4121 | } |
| 4852 | 4122 | |
| @@ -5055,23 +4325,13 @@ | ||
| 5055 | 4325 | $error_message = $error_json['error']['message'] ?? 'No message'; |
| 5056 | 4326 | //error_log('[MXCHAT-EMBED] API Error Type: ' . $error_type); |
| 5057 | 4327 | //error_log('[MXCHAT-EMBED] API Error Message: ' . $error_message); |
| 5058 | 4328 | |
| 5059 | - // Keep the provider's own diagnostic — a restricted-key 401 names the | |
| 5060 | - // exact missing scope, and replacing it with "check your API key" sent | |
| 5061 | - // a customer to regenerate two keys (plan 46b596). Same shape as | |
| 5062 | - // MxChat_Utils::embedding_failure_error() so both ingestion paths read | |
| 5063 | - // identically. Key never appears in provider messages, but scrub anyway. | |
| 5064 | - if ($error_type === 'invalid_request_error' || $error_type === 'authentication_error') { | |
| 5065 | - if (is_string($api_key) && $api_key !== '') { | |
| 5066 | - $error_message = str_replace($api_key, '[redacted]', $error_message); | |
| 5067 | - } | |
| 5068 | - $error_message = sprintf( | |
| 5069 | - 'Embedding failed (%s, HTTP %d): %s', | |
| 5070 | - $selected_model, | |
| 5071 | - $http_code, | |
| 5072 | - substr($error_message, 0, 300) | |
| 5073 | - ); | |
| 4329 | + // Customize error message for common API errors | |
| 4330 | + if ($error_type === 'invalid_request_error' && strpos($error_message, 'API key') !== false) { | |
| 4331 | + $error_message = sprintf('Invalid %s API key for bot %s. Please check your API key in the bot settings.', $provider_name, $bot_id); | |
| 4332 | + } elseif ($error_type === 'authentication_error') { | |
| 4333 | + $error_message = sprintf('%s authentication failed for bot %s. Please verify your API key in the bot settings.', $provider_name, $bot_id); | |
| 5074 | 4334 | } |
| 5075 | 4335 | |
| 5076 | 4336 | //error_log('[MXCHAT-EMBED] Returning error: ' . $error_message); |
| 5077 | 4337 | return $error_message; |
| @@ -5583,38 +4843,25 @@ | ||
| 5583 | 4843 | /** |
| 5584 | 4844 | * Get all ACF fields for a specific post, excluding any fields the user has disabled |
| 5585 | 4845 | */ |
| 5586 | 4846 | public function mxchat_get_acf_fields_for_post($post_id) { |
| 5587 | - if (!function_exists('get_field_objects')) { | |
| 4847 | + if (!function_exists('get_fields')) { | |
| 5588 | 4848 | return array(); |
| 5589 | 4849 | } |
| 5590 | 4850 | |
| 5591 | - // Field OBJECTS, not get_fields(): exclusion matches on the field KEY | |
| 5592 | - // (unique per field) rather than the name (shared across groups — plan | |
| 5593 | - // 30e81f). ACF's own get_fields() is implemented as get_field_objects() | |
| 5594 | - // reduced to name => value, so the un-excluded reduction below is the | |
| 5595 | - // identical shape and order the previous get_fields() call produced. | |
| 5596 | - $field_objects = get_field_objects($post_id); | |
| 5597 | - if (!$field_objects || !is_array($field_objects)) { | |
| 4851 | + $fields = get_fields($post_id); | |
| 4852 | + if (!$fields || !is_array($fields)) { | |
| 5598 | 4853 | return array(); |
| 5599 | 4854 | } |
| 5600 | 4855 | |
| 4856 | + // Get excluded fields from settings | |
| 5601 | 4857 | $excluded_fields = get_option('mxchat_acf_excluded_fields', array()); |
| 5602 | - if (!is_array($excluded_fields)) { | |
| 5603 | - $excluded_fields = array(); | |
| 5604 | - } | |
| 5605 | - | |
| 5606 | - $fields = array(); | |
| 5607 | - foreach ($field_objects as $field_name => $field_object) { | |
| 5608 | - if (!empty($excluded_fields)) { | |
| 5609 | - $field_key = isset($field_object['key']) ? $field_object['key'] : ''; | |
| 5610 | - // Legacy name entries stay honored: a stored name whose group was | |
| 5611 | - // inactive at migration time still excludes every field wearing it. | |
| 5612 | - if (in_array($field_key, $excluded_fields, true) || in_array($field_name, $excluded_fields, true)) { | |
| 5613 | - continue; | |
| 4858 | + if (!empty($excluded_fields) && is_array($excluded_fields)) { | |
| 4859 | + foreach ($excluded_fields as $excluded_field) { | |
| 4860 | + if (isset($fields[$excluded_field])) { | |
| 4861 | + unset($fields[$excluded_field]); | |
| 5614 | 4862 | } |
| 5615 | 4863 | } |
| 5616 | - $fields[$field_name] = isset($field_object['value']) ? $field_object['value'] : null; | |
| 5617 | 4864 | } |
| 5618 | 4865 | |
| 5619 | 4866 | return $fields; |
| 5620 | 4867 | } |
| @@ -5626,11 +4873,8 @@ | ||
| 5626 | 4873 | if (!function_exists('acf_get_field_groups') || !function_exists('acf_get_fields')) { |
| 5627 | 4874 | return array(); |
| 5628 | 4875 | } |
| 5629 | 4876 | |
| 5630 | - // Keyed by GROUP KEY, not title (titles are not unique), and each field | |
| 5631 | - // entry carries its ACF field key — the unique identifier every toggle, | |
| 5632 | - // save, and index-time exclusion now runs on (plans 30e81f / bf57e0). | |
| 5633 | 4877 | $all_fields = array(); |
| 5634 | 4878 | $field_groups = acf_get_field_groups(); |
| 5635 | 4879 | |
| 5636 | 4880 | if (!empty($field_groups)) { |
| @@ -5636,21 +4880,16 @@ | ||
| 5636 | 4880 | if (!empty($field_groups)) { |
| 5637 | 4881 | foreach ($field_groups as $group) { |
| 5638 | 4882 | $group_fields = acf_get_fields($group['key']); |
| 5639 | 4883 | if (!empty($group_fields)) { |
| 5640 | - $entry = array( | |
| 5641 | - 'title' => $group['title'], | |
| 5642 | - 'fields' => array(), | |
| 5643 | - ); | |
| 4884 | + $all_fields[$group['title']] = array(); | |
| 5644 | 4885 | foreach ($group_fields as $field) { |
| 5645 | - $entry['fields'][] = array( | |
| 5646 | - 'key' => $field['key'], | |
| 5647 | - 'name' => $field['name'], | |
| 4886 | + $all_fields[$group['title']][] = array( | |
| 4887 | + 'name' => $field['name'], | |
| 5648 | 4888 | 'label' => $field['label'], |
| 5649 | - 'type' => $field['type'] | |
| 4889 | + 'type' => $field['type'] | |
| 5650 | 4890 | ); |
| 5651 | 4891 | } |
| 5652 | - $all_fields[$group['key']] = $entry; | |
| 5653 | 4892 | } |
| 5654 | 4893 | } |
| 5655 | 4894 | } |
| 5656 | 4895 | |
| @@ -6045,11 +5284,9 @@ | ||
| 6045 | 5284 | $parser = new \Smalot\PdfParser\Parser(); |
| 6046 | 5285 | $pdf = $parser->parseFile($pdf_path); |
| 6047 | 5286 | $pages = $pdf->getPages(); |
| 6048 | 5287 | $page_texts = array(); |
| 6049 | - $acf_page_num = 0; | |
| 6050 | 5288 | foreach ($pages as $page) { |
| 6051 | - $acf_page_num++; | |
| 6052 | 5289 | $page_text = ''; |
| 6053 | 5290 | try { |
| 6054 | 5291 | $page_text = $page->getText(); |
| 6055 | 5292 | } catch (\Exception $e) { |
| @@ -6055,9 +5292,8 @@ | ||
| 6055 | 5292 | } catch (\Exception $e) { |
| 6056 | 5293 | $page_text = ''; |
| 6057 | 5294 | } |
| 6058 | 5295 | if (!empty($page_text)) { |
| 6059 | - $page_text = MxChat_Utils::normalize_pdf_rtl($page_text, 'acf_pdf attachment ' . $attachment_id . ' page ' . $acf_page_num); | |
| 6060 | 5296 | $page_texts[] = $page_text; |
| 6061 | 5297 | } |
| 6062 | 5298 | } |
| 6063 | 5299 | $text = trim(implode("\n\n", $page_texts)); |
| @@ -6155,13 +5391,8 @@ | ||
| 6155 | 5391 | $this->mxchat_handle_post_update($post_id, $post, true); |
| 6156 | 5392 | } |
| 6157 | 5393 | |
| 6158 | 5394 | public function mxchat_handle_post_update($post_id, $post, $update) { |
| 6159 | - // The in-flight-update marker has done its job the moment post_updated runs; drop it | |
| 6160 | - // before any early return so it can never outlive its own save (a failed $wpdb->update | |
| 6161 | - // inside wp_insert_post returns after pre_post_update but before the transition). | |
| 6162 | - unset($this->pending_post_update[$post_id]); | |
| 6163 | - | |
| 6164 | 5395 | // Basic validation checks |
| 6165 | 5396 | if (defined('DOING_AUTOSAVE') && DOING_AUTOSAVE || wp_is_post_revision($post_id)) { |
| 6166 | 5397 | return; |
| 6167 | 5398 | } |
| @@ -6199,11 +5430,9 @@ | ||
| 6199 | 5430 | if ($previous_status === 'publish' && $post->post_status !== 'publish') { |
| 6200 | 5431 | // Use the stored URL from when it was published, or fall back to current permalink |
| 6201 | 5432 | $source_url = $previous_url ?: get_permalink($post_id); |
| 6202 | 5433 | |
| 6203 | - // mxchat_handle_status_transition already deleted for this post earlier in this | |
| 6204 | - // request (it fires first inside wp_insert_post); skip the redundant round-trip. | |
| 6205 | - if ($source_url && empty($this->transition_deleted_posts[$post_id])) { | |
| 5434 | + if ($source_url) { | |
| 6206 | 5435 | // Chunk-aware deletion (routes to Pinecone or WP DB and removes base + all chunks) |
| 6207 | 5436 | MxChat_Utils::delete_chunks_for_url($source_url, 'default'); |
| 6208 | 5437 | } |
| 6209 | 5438 | |
| @@ -6232,402 +5461,209 @@ | ||
| 6232 | 5461 | set_transient($previous_url_key, $current_url, DAY_IN_SECONDS); |
| 6233 | 5462 | } |
| 6234 | 5463 | } |
| 6235 | 5464 | |
| 6236 | - // Only process currently published content for adding/updating. | |
| 6237 | - // transition_indexed_posts: mxchat_handle_status_transition's arrival edge may have | |
| 6238 | - // already indexed this post earlier in this request (editor publishes fire | |
| 6239 | - // transition_post_status first, then post_updated) — skip the duplicate embed. | |
| 6240 | - // Consume-once: the flag is cleared when honoured, so a LATER save of the same | |
| 6241 | - // post in one long-running process (WP-CLI scripts, importers) re-indexes normally. | |
| 5465 | + // Only process currently published content for adding/updating | |
| 6242 | 5466 | if ($post->post_status === 'publish') { |
| 6243 | - if (!empty($this->transition_indexed_posts[$post_id])) { | |
| 6244 | - unset($this->transition_indexed_posts[$post_id]); | |
| 6245 | - } else { | |
| 6246 | - $this->mxchat_index_published_post($post_id, $post); | |
| 6247 | - } | |
| 6248 | - } | |
| 6249 | - | |
| 6250 | - // Clean up the stored previous status if not used above | |
| 6251 | - if ($previous_status !== 'publish' || $post->post_status === 'publish') { | |
| 6252 | - delete_transient($previous_status_key); | |
| 6253 | - delete_transient($previous_url_key); | |
| 6254 | - } | |
| 6255 | -} | |
| 6256 | - | |
| 6257 | -/** | |
| 6258 | - * Index a published post into the knowledge base: preprocessing filter, content | |
| 6259 | - * assembly (title/excerpt/body), WooCommerce product enrichment, job_listing meta, | |
| 6260 | - * ACF fields (+ optional PDF extraction), whitelisted custom meta, embedding and | |
| 6261 | - * upsert, then tag-based role restriction. | |
| 6262 | - * | |
| 6263 | - * Shared by the post_updated auto-sync path (mxchat_handle_post_update) and the | |
| 6264 | - * transition_post_status arrival edge (mxchat_handle_status_transition), so | |
| 6265 | - * scheduled publishes (wp_publish_post) and direct status=publish inserts index | |
| 6266 | - * identically to editor saves (plan 3055e1). Pure extraction of the former | |
| 6267 | - * publish branch — body indentation retained to keep the diff reviewable. | |
| 6268 | - */ | |
| 6269 | -private function mxchat_index_published_post($post_id, $post) { | |
| 6270 | - $post_type = $post->post_type; | |
| 6271 | - | |
| 6272 | - // WooCommerce products are owned by the WC-object assembler (plan a3d60c): | |
| 6273 | - // whenever WooCommerce is active AND the integration is enabled, every | |
| 6274 | - // product save also fires save_post_product, which queues | |
| 6275 | - // mxchat_store_product_embedding on shutdown — and that writer runs LAST, | |
| 6276 | - // overwriting the same md5(permalink) row this path would write. Assembling | |
| 6277 | - // and embedding the product here was pure duplicate spend (measured: two | |
| 6278 | - // embedding calls per product save, second one wins). Skip ONLY under the | |
| 6279 | - // exact conditions the shutdown writer runs — same option read as its own | |
| 6280 | - // gate — because with the integration off (or WooCommerce inactive) this | |
| 6281 | - // path is the sole product indexer and must keep working. | |
| 6282 | - if ($post_type === 'product' | |
| 6283 | - && class_exists('WooCommerce') | |
| 6284 | - && isset($this->options['enable_woocommerce_integration']) | |
| 6285 | - && in_array($this->options['enable_woocommerce_integration'], ['1', 'on'])) { | |
| 6286 | - return; | |
| 6287 | - } | |
| 6288 | - | |
| 6289 | 5467 | // Get the source URL |
| 6290 | 5468 | $source_url = get_permalink($post_id); |
| 5469 | + | |
| 5470 | + // Get content with proper formatting (matching ajax_mxchat_process_selected_content) | |
| 5471 | + $title = get_the_title($post_id); | |
| 5472 | + $content = get_post_field('post_content', $post_id); | |
| 5473 | + $excerpt = get_post_field('post_excerpt', $post_id); | |
| 6291 | 5474 | |
| 6292 | - // A draft published programmatically (wp_publish_post) can reach this | |
| 6293 | - // point with an EMPTY post_name — wp_insert_post skips slug generation | |
| 6294 | - // for draft/pending — and get_permalink() then resolves to the bare | |
| 6295 | - // site root. A knowledge row keyed to the homepage cites the wrong URL | |
| 6296 | - // and answers homepage questions with this post's body, so refuse to | |
| 6297 | - // write it; the post indexes correctly on its next save, once the slug | |
| 6298 | - // exists. The empty-post_name test is what keeps a legitimate static | |
| 6299 | - // front page (which has a slug but a root permalink) indexable. | |
| 6300 | - // (Plan d138c4.) | |
| 6301 | - if ('' === $post->post_name | |
| 6302 | - && untrailingslashit($source_url) === untrailingslashit(home_url())) { | |
| 6303 | - return; | |
| 6304 | - } | |
| 5475 | + // Remove shortcode tags but preserve content inside them | |
| 5476 | + $content = $this->strip_shortcode_tags_preserve_content($content); | |
| 5477 | + $excerpt = $this->strip_shortcode_tags_preserve_content($excerpt); | |
| 6305 | 5478 | |
| 6306 | - /** | |
| 6307 | - * Allow developers to modify post data before processing into the knowledge base. | |
| 6308 | - * Same filter and signature as the manual bulk-import path | |
| 6309 | - * (ajax_mxchat_process_selected_content), so a callback registered once covers | |
| 6310 | - * every indexing route. Purely additive — zero behaviour change when unhooked. | |
| 6311 | - * Auto-sync runs under the 'default' bot context, matching the rest of this | |
| 6312 | - * function. | |
| 6313 | - * | |
| 6314 | - * @param WP_Post $post The post about to be indexed. | |
| 6315 | - * @param string $bot_id Bot context ('default' on auto-sync). | |
| 6316 | - */ | |
| 6317 | - $post = apply_filters('mxchat_before_process_post', $post, 'default'); | |
| 6318 | - if (!($post instanceof WP_Post)) { | |
| 6319 | - $post = get_post($post_id); // defend against a bad callback return | |
| 6320 | - } | |
| 5479 | + // Strip tags but preserve structure (don't use 'the_content' filter as it may re-add shortcodes) | |
| 5480 | + $content = wp_strip_all_tags($content); | |
| 6321 | 5481 | |
| 6322 | - // Assemble the indexable text via the shared post-kind assembler (a3d60c), | |
| 6323 | - // reading from the FILTERED post object — not re-fetched by ID, which would | |
| 6324 | - // discard it. Auto-sync reads content/excerpt in its historical | |
| 6325 | - // get_post_field() display context, never appended product custom tabs | |
| 6326 | - // (its product branch is reachable only with the WooCommerce integration | |
| 6327 | - // off), and gates ACF→PDF extraction behind its own opt-in option — | |
| 6328 | - // default OFF, because re-parsing every ACF PDF on every editor save is | |
| 6329 | - // expensive and most sites don't want it (the 25 MB size cap lives in the | |
| 6330 | - // shared extractor either way). | |
| 6331 | - $prepared = $this->mxchat_prepare_post_content_for_indexing($post_id, $post, array( | |
| 6332 | - 'read_display' => true, | |
| 6333 | - 'extract_acf_pdfs' => get_option('mxchat_auto_sync_acf_pdfs', '0') === '1', | |
| 6334 | - 'include_product_tabs' => false, | |
| 6335 | - )); | |
| 6336 | - $final_content = $prepared['content']; | |
| 5482 | + // Combine title, short description (if exists), and content | |
| 5483 | + $final_content = $title . "\n\n"; | |
| 6337 | 5484 | |
| 6338 | - // Embedding decision — custom-provider-aware. Gating on a cloud API key | |
| 6339 | - // here silently killed auto-sync on keyless custom-embeddings sites, | |
| 6340 | - // because generate_embedding() routes custom FIRST and never needs the | |
| 6341 | - // key (plan cbd5fd). Silent-return shape preserved. | |
| 6342 | - $preflight = MxChat_Utils::embedding_preflight(get_option('mxchat_options')); | |
| 6343 | - if (!$preflight['ok']) { | |
| 6344 | - return; | |
| 5485 | + // Add short description if it exists (WooCommerce products use post_excerpt for short description) | |
| 5486 | + if (!empty($excerpt)) { | |
| 5487 | + $final_content .= "Short Description: " . wp_strip_all_tags($excerpt) . "\n\n"; | |
| 6345 | 5488 | } |
| 6346 | - $api_key = $preflight['api_key']; | |
| 6347 | 5489 | |
| 6348 | - // Use the centralized utility function for storage | |
| 6349 | - $result = MxChat_Utils::submit_content_to_db( | |
| 6350 | - $final_content, | |
| 6351 | - $source_url, | |
| 6352 | - $api_key, | |
| 6353 | - md5($source_url) // Vector ID for Pinecone | |
| 6354 | - ); | |
| 6355 | - | |
| 6356 | - // After successful storage, apply role restriction based on tags | |
| 6357 | - if (!is_wp_error($result)) { | |
| 6358 | - $this->apply_role_restriction_to_post($post_id, $source_url); | |
| 6359 | - } | |
| 6360 | -} | |
| 5490 | + $final_content .= $content; | |
| 6361 | 5491 | |
| 6362 | -/** | |
| 6363 | - * Shared post-fields content assembler (plan a3d60c) — the ONE body behind both | |
| 6364 | - * post-kind ingestion paths: manual bulk import (ajax_mxchat_process_selected_content) | |
| 6365 | - * and auto-sync (mxchat_index_published_post). Behavior-preserving extraction; the | |
| 6366 | - * measured per-caller differences ride $args instead of living as drifting copies: | |
| 6367 | - * | |
| 6368 | - * 'read_display' bool Auto-sync historically reads content/excerpt via | |
| 6369 | - * get_post_field() in its default 'display' context | |
| 6370 | - * (the post_content / post_excerpt display filters | |
| 6371 | - * fire); bulk import reads the raw properties. Inert | |
| 6372 | - * on a stock install — preserved per-path, not converged. | |
| 6373 | - * 'extract_acf_pdfs' bool Each caller passes its OWN option (bulk: | |
| 6374 | - * mxchat_acf_pdf_extraction; auto-sync: | |
| 6375 | - * mxchat_auto_sync_acf_pdfs) — the two-option design | |
| 6376 | - * is deliberate (plan 11720c). Gates BOTH the PDF-id | |
| 6377 | - * collection walk and the extraction loop; the ids are | |
| 6378 | - * only ever read inside the extraction branch, so | |
| 6379 | - * gating collection is output-identical on every install. | |
| 6380 | - * 'include_product_tabs' bool The bulk path has always appended yikes_woo custom | |
| 6381 | - * tabs to product content; the auto-sync product branch | |
| 6382 | - * (reachable only with the WooCommerce integration off) | |
| 6383 | - * never did. Preserved per-path — converging it would be | |
| 6384 | - * a behavior change, recorded on the plan instead. | |
| 6385 | - * | |
| 6386 | - * Returns array: 'content' (the assembled indexable text), 'acf_fields_found' and | |
| 6387 | - * 'pdf_extracted_count' (the bulk path reports both in its AJAX response). | |
| 6388 | - */ | |
| 6389 | -private function mxchat_prepare_post_content_for_indexing($post_id, $post, $args) { | |
| 6390 | - $read_display = !empty($args['read_display']); | |
| 6391 | - $extract_acf_pdfs = !empty($args['extract_acf_pdfs']); | |
| 6392 | - $include_product_tabs = !empty($args['include_product_tabs']); | |
| 5492 | + // For WooCommerce products, include pricing and product details | |
| 5493 | + if ($post_type === 'product' && class_exists('WooCommerce')) { | |
| 5494 | + $product = wc_get_product($post_id); | |
| 6393 | 5495 | |
| 6394 | - // Raw post_title, NOT get_the_title(): the_title applies wptexturize + | |
| 6395 | - // convert_chars and prepends the "Protected:" / "Private:" display chrome. | |
| 6396 | - // The knowledge base stores facts, not display strings. Entity decode at | |
| 6397 | - // output time (single-pass, shared helper) — a stored `&` embeds worse | |
| 6398 | - // than `&` and gets quoted back to visitors (d2c92e). | |
| 6399 | - $content = $this->mxchat_decode_entities_for_indexing($post->post_title) . "\n\n"; | |
| 5496 | + if ($product) { | |
| 5497 | + // Get pricing information | |
| 5498 | + $regular_price = $product->get_regular_price(); | |
| 5499 | + $sale_price = $product->get_sale_price(); | |
| 5500 | + $price = $product->get_price(); | |
| 5501 | + $sku = $product->get_sku(); | |
| 6400 | 5502 | |
| 6401 | - $raw_excerpt = $read_display ? get_post_field('post_excerpt', $post) : $post->post_excerpt; | |
| 6402 | - $raw_content = $read_display ? get_post_field('post_content', $post) : $post->post_content; | |
| 5503 | + // Get currency symbol | |
| 5504 | + $currency_symbol = get_woocommerce_currency_symbol(); | |
| 6403 | 5505 | |
| 6404 | - // Add short description if it exists (WooCommerce products use post_excerpt for short description) | |
| 6405 | - // Strip FIRST, then test: an excerpt that is nothing but shortcodes strips to | |
| 6406 | - // empty, and testing the raw value emitted a bare "Short Description: " label | |
| 6407 | - // with no value after it. trim() only in the TEST — the emitted value is | |
| 6408 | - // untouched, so a populated excerpt is byte-identical to before. A | |
| 6409 | - // whitespace-only excerpt is an empty excerpt and must not produce a labelled | |
| 6410 | - // line with nothing after it. | |
| 6411 | - $clean_excerpt = $this->strip_shortcode_tags_preserve_content($raw_excerpt); | |
| 6412 | - if (trim($clean_excerpt) !== '') { | |
| 6413 | - $content .= "Short Description: " . $this->mxchat_decode_entities_for_indexing(wp_strip_all_tags($clean_excerpt)) . "\n\n"; | |
| 6414 | - } | |
| 5506 | + // Add pricing information | |
| 5507 | + $final_content .= "\n"; | |
| 5508 | + if (!empty($regular_price)) { | |
| 5509 | + $final_content .= "Price: " . $currency_symbol . $regular_price . "\n"; | |
| 5510 | + } elseif (!empty($price)) { | |
| 5511 | + $final_content .= "Price: " . $currency_symbol . $price . "\n"; | |
| 5512 | + } | |
| 6415 | 5513 | |
| 6416 | - // Main content — remove shortcode tags but preserve content inside them, then | |
| 6417 | - // strip tags (don't use 'the_content' filter as it may re-add shortcodes). | |
| 6418 | - $clean_content = $this->strip_shortcode_tags_preserve_content($raw_content); | |
| 6419 | - $content .= $this->mxchat_decode_entities_for_indexing(wp_strip_all_tags($clean_content)); | |
| 5514 | + if (!empty($sale_price) && $sale_price !== $regular_price) { | |
| 5515 | + $final_content .= "Sale Price: " . $currency_symbol . $sale_price . "\n"; | |
| 5516 | + } | |
| 6420 | 5517 | |
| 6421 | - // WooCommerce product enrichment (post-fields kind). The WC-object assembler | |
| 6422 | - // (mxchat_prepare_product_content_for_indexing) owns product rows whenever the | |
| 6423 | - // integration is on; this branch serves the bulk import (all configurations) | |
| 6424 | - // and auto-sync with the integration off. | |
| 6425 | - if (get_post_type($post_id) === 'product' && class_exists('WooCommerce')) { | |
| 6426 | - $product = wc_get_product($post_id); | |
| 5518 | + // Handle variable products - show price range | |
| 5519 | + if ($product->is_type('variable')) { | |
| 5520 | + $min_price = $product->get_variation_price('min'); | |
| 5521 | + $max_price = $product->get_variation_price('max'); | |
| 5522 | + if ($min_price !== $max_price) { | |
| 5523 | + $final_content .= "Price Range: " . $currency_symbol . $min_price . " - " . $currency_symbol . $max_price . "\n"; | |
| 5524 | + } | |
| 5525 | + } | |
| 6427 | 5526 | |
| 6428 | - if ($product) { | |
| 6429 | - $content .= "\n"; | |
| 6430 | - $content .= $this->mxchat_woo_product_summary_lines($product); | |
| 6431 | - } | |
| 5527 | + if (!empty($sku)) { | |
| 5528 | + $final_content .= "SKU: " . $sku . "\n"; | |
| 5529 | + } | |
| 6432 | 5530 | |
| 6433 | - if ($include_product_tabs) { | |
| 6434 | - $content .= $this->mxchat_woo_custom_tabs_text($post_id); | |
| 5531 | + // Get product categories | |
| 5532 | + $categories = wp_get_post_terms($post_id, 'product_cat', array('fields' => 'names')); | |
| 5533 | + if (!empty($categories) && !is_wp_error($categories)) { | |
| 5534 | + $final_content .= "Categories: " . implode(', ', $categories) . "\n"; | |
| 5535 | + } | |
| 5536 | + } | |
| 6435 | 5537 | } |
| 6436 | - } | |
| 6437 | 5538 | |
| 6438 | - // For custom post types like job_listing, include additional fields | |
| 6439 | - if (get_post_type($post_id) === 'job_listing') { | |
| 6440 | - // Add job-specific meta if available | |
| 6441 | - $job_location = get_post_meta($post_id, '_job_location', true); | |
| 6442 | - if (!empty($job_location)) { | |
| 6443 | - $content .= "\n\nLocation: " . $job_location; | |
| 6444 | - } | |
| 5539 | + // For custom post types like job_listing, include additional fields | |
| 5540 | + if ($post_type === 'job_listing') { | |
| 5541 | + // Add job-specific meta if available | |
| 5542 | + $job_location = get_post_meta($post_id, '_job_location', true); | |
| 5543 | + if (!empty($job_location)) { | |
| 5544 | + $final_content .= "\n\nLocation: " . $job_location; | |
| 5545 | + } | |
| 6445 | 5546 | |
| 6446 | - // Get job type terms | |
| 6447 | - $job_types = get_the_terms($post_id, 'job_listing_type'); | |
| 6448 | - if (!empty($job_types) && !is_wp_error($job_types)) { | |
| 6449 | - $types = array(); | |
| 6450 | - foreach ($job_types as $type) { | |
| 6451 | - $types[] = $type->name; | |
| 5547 | + // Get job type terms | |
| 5548 | + $job_types = get_the_terms($post_id, 'job_listing_type'); | |
| 5549 | + if (!empty($job_types) && !is_wp_error($job_types)) { | |
| 5550 | + $types = array(); | |
| 5551 | + foreach ($job_types as $type) { | |
| 5552 | + $types[] = $type->name; | |
| 5553 | + } | |
| 5554 | + $final_content .= "\n\nJob Type: " . implode(', ', $types); | |
| 6452 | 5555 | } |
| 6453 | - $content .= "\n\nJob Type: " . implode(', ', $types); | |
| 6454 | - } | |
| 6455 | 5556 | |
| 6456 | - // Get company name if available | |
| 6457 | - $company_name = get_post_meta($post_id, '_company_name', true); | |
| 6458 | - if (!empty($company_name)) { | |
| 6459 | - $content .= "\n\nCompany: " . $company_name; | |
| 5557 | + // Get company name if available | |
| 5558 | + $company_name = get_post_meta($post_id, '_company_name', true); | |
| 5559 | + if (!empty($company_name)) { | |
| 5560 | + $final_content .= "\n\nCompany: " . $company_name; | |
| 5561 | + } | |
| 6460 | 5562 | } |
| 6461 | - } | |
| 6462 | 5563 | |
| 6463 | - // ADD ACF FIELDS SUPPORT | |
| 6464 | - $acf_fields = $this->mxchat_get_acf_fields_for_post($post_id); | |
| 6465 | - $pdf_extracted_count = 0; | |
| 6466 | - if (!empty($acf_fields)) { | |
| 6467 | - $acf_content_parts = array(); | |
| 6468 | - $pdf_attachment_ids = array(); | |
| 5564 | + // ADD ACF FIELDS SUPPORT (matches ajax_mxchat_process_selected_content behavior) | |
| 5565 | + $acf_fields = $this->mxchat_get_acf_fields_for_post($post_id); | |
| 5566 | + if (!empty($acf_fields)) { | |
| 5567 | + $acf_content_parts = array(); | |
| 5568 | + $pdf_attachment_ids = array(); | |
| 6469 | 5569 | |
| 6470 | - foreach ($acf_fields as $field_name => $field_value) { | |
| 6471 | - $formatted_value = $this->mxchat_format_acf_field_value($field_value, $field_name, $post_id); | |
| 5570 | + foreach ($acf_fields as $field_name => $field_value) { | |
| 5571 | + $formatted_value = $this->mxchat_format_acf_field_value($field_value, $field_name, $post_id); | |
| 5572 | + if (!empty($formatted_value)) { | |
| 5573 | + // Convert field name to readable label | |
| 5574 | + $field_label = ucwords(str_replace(['_', '-'], ' ', $field_name)); | |
| 5575 | + $acf_content_parts[] = $field_label . ": " . $formatted_value; | |
| 5576 | + } | |
| 6472 | 5577 | |
| 6473 | - if (!empty($formatted_value)) { | |
| 6474 | - // Both separators: a hyphenated ACF name should read as words. | |
| 6475 | - $field_label = ucwords(str_replace(['_', '-'], ' ', $field_name)); | |
| 6476 | - $acf_content_parts[] = $field_label . ": " . $formatted_value; | |
| 5578 | + $this->mxchat_collect_pdf_attachment_ids_from_acf_value($field_value, $pdf_attachment_ids); | |
| 6477 | 5579 | } |
| 6478 | 5580 | |
| 6479 | - // Walk this field's value tree for any PDF attachment references and | |
| 6480 | - // queue them for extraction — only when this caller's PDF option is on. | |
| 6481 | - if ($extract_acf_pdfs) { | |
| 6482 | - $this->mxchat_collect_pdf_attachment_ids_from_acf_value($field_value, $pdf_attachment_ids); | |
| 5581 | + if (!empty($acf_content_parts)) { | |
| 5582 | + $final_content .= "\n\n" . implode("\n", $acf_content_parts); | |
| 6483 | 5583 | } |
| 6484 | - } | |
| 6485 | 5584 | |
| 6486 | - if (!empty($acf_content_parts)) { | |
| 6487 | - $content .= "\n\n" . implode("\n", $acf_content_parts); | |
| 6488 | - } | |
| 6489 | - | |
| 6490 | - // Extract text from each unique PDF found in ACF fields and append as a labeled section | |
| 6491 | - if ($extract_acf_pdfs && !empty($pdf_attachment_ids)) { | |
| 6492 | - $pdf_attachment_ids = array_unique(array_filter(array_map('intval', $pdf_attachment_ids))); | |
| 6493 | - $pdf_sections = array(); | |
| 6494 | - foreach ($pdf_attachment_ids as $att_id) { | |
| 6495 | - $pdf_text = $this->mxchat_extract_pdf_text_by_attachment_id($att_id); | |
| 6496 | - if (!empty($pdf_text)) { | |
| 6497 | - $pdf_title = get_the_title($att_id); | |
| 6498 | - $pdf_url = wp_get_attachment_url($att_id); | |
| 6499 | - $header = 'PDF Attachment'; | |
| 6500 | - if (!empty($pdf_title)) { | |
| 6501 | - $header .= ': ' . $pdf_title; | |
| 5585 | + // Gate the auto-sync PDF-extraction loop behind an opt-in option. | |
| 5586 | + // Mirrors the per-batch checkbox the manual content selector has; the | |
| 5587 | + // 25 MB size cap lives in the shared extractor so it applies in both | |
| 5588 | + // paths regardless. Default OFF — re-parsing every ACF PDF on every | |
| 5589 | + // editor save is expensive and most sites don't want it. | |
| 5590 | + $autosync_extract_acf_pdfs = get_option('mxchat_auto_sync_acf_pdfs', '0') === '1'; | |
| 5591 | + if ($autosync_extract_acf_pdfs && !empty($pdf_attachment_ids)) { | |
| 5592 | + $pdf_attachment_ids = array_unique(array_filter(array_map('intval', $pdf_attachment_ids))); | |
| 5593 | + $pdf_sections = array(); | |
| 5594 | + foreach ($pdf_attachment_ids as $att_id) { | |
| 5595 | + $pdf_text = $this->mxchat_extract_pdf_text_by_attachment_id($att_id); | |
| 5596 | + if (!empty($pdf_text)) { | |
| 5597 | + $pdf_title = get_the_title($att_id); | |
| 5598 | + $pdf_url = wp_get_attachment_url($att_id); | |
| 5599 | + $header = 'PDF Attachment'; | |
| 5600 | + if (!empty($pdf_title)) { | |
| 5601 | + $header .= ': ' . $pdf_title; | |
| 5602 | + } | |
| 5603 | + if (!empty($pdf_url)) { | |
| 5604 | + $header .= ' (' . $pdf_url . ')'; | |
| 5605 | + } | |
| 5606 | + $pdf_sections[] = $header . "\n" . $pdf_text; | |
| 6502 | 5607 | } |
| 6503 | - if (!empty($pdf_url)) { | |
| 6504 | - $header .= ' (' . $pdf_url . ')'; | |
| 6505 | - } | |
| 6506 | - $pdf_sections[] = $header . "\n" . $pdf_text; | |
| 6507 | - $pdf_extracted_count++; | |
| 6508 | 5608 | } |
| 5609 | + if (!empty($pdf_sections)) { | |
| 5610 | + $final_content .= "\n\nAttached PDFs:\n" . implode("\n\n", $pdf_sections); | |
| 5611 | + } | |
| 6509 | 5612 | } |
| 6510 | - if (!empty($pdf_sections)) { | |
| 6511 | - $content .= "\n\nAttached PDFs:\n" . implode("\n\n", $pdf_sections); | |
| 6512 | - } | |
| 6513 | 5613 | } |
| 6514 | - } | |
| 6515 | 5614 | |
| 6516 | - // ADD CUSTOM POST META SUPPORT (whitelisted non-ACF meta fields) | |
| 6517 | - $custom_meta = $this->mxchat_get_whitelisted_post_meta($post_id); | |
| 6518 | - if (!empty($custom_meta)) { | |
| 6519 | - $meta_content_parts = array(); | |
| 5615 | + // ADD CUSTOM POST META SUPPORT (whitelisted non-ACF meta fields) | |
| 5616 | + $custom_meta = $this->mxchat_get_whitelisted_post_meta($post_id); | |
| 5617 | + if (!empty($custom_meta)) { | |
| 5618 | + $meta_content_parts = array(); | |
| 6520 | 5619 | |
| 6521 | - foreach ($custom_meta as $meta_key => $meta_value) { | |
| 6522 | - // Convert meta key to readable label | |
| 6523 | - $meta_label = ucwords(str_replace(array('_', '-'), ' ', $meta_key)); | |
| 6524 | - $meta_content_parts[] = $meta_label . ": " . $meta_value; | |
| 5620 | + foreach ($custom_meta as $meta_key => $meta_value) { | |
| 5621 | + // Convert meta key to readable label | |
| 5622 | + $meta_label = ucwords(str_replace(array('_', '-'), ' ', $meta_key)); | |
| 5623 | + $meta_content_parts[] = $meta_label . ": " . $meta_value; | |
| 5624 | + } | |
| 5625 | + | |
| 5626 | + if (!empty($meta_content_parts)) { | |
| 5627 | + $final_content .= "\n\n" . implode("\n", $meta_content_parts); | |
| 5628 | + } | |
| 6525 | 5629 | } |
| 6526 | 5630 | |
| 6527 | - if (!empty($meta_content_parts)) { | |
| 6528 | - $content .= "\n\n" . implode("\n", $meta_content_parts); | |
| 5631 | + // Get API key with proper model detection | |
| 5632 | + $options = get_option('mxchat_options'); | |
| 5633 | + $selected_model = $options['embedding_model'] ?? 'text-embedding-ada-002'; | |
| 5634 | + | |
| 5635 | + if (strpos($selected_model, 'voyage') === 0) { | |
| 5636 | + $api_key = $options['voyage_api_key'] ?? ''; | |
| 5637 | + } elseif (strpos($selected_model, 'gemini-embedding') === 0) { | |
| 5638 | + $api_key = $options['gemini_api_key'] ?? ''; | |
| 5639 | + } else { | |
| 5640 | + $api_key = $options['api_key'] ?? ''; | |
| 6529 | 5641 | } |
| 6530 | - } | |
| 6531 | - | |
| 6532 | - return array( | |
| 6533 | - 'content' => $content, | |
| 6534 | - 'acf_fields_found' => count($acf_fields), | |
| 6535 | - 'pdf_extracted_count' => $pdf_extracted_count, | |
| 6536 | - ); | |
| 6537 | -} | |
| 6538 | - | |
| 6539 | -/** | |
| 6540 | - * Shared WC-object product assembler (plan a3d60c) — the ONE body behind the two | |
| 6541 | - * WooCommerce-object ingestion paths: the auto-sync product writer | |
| 6542 | - * (mxchat_store_product_embedding) and the URL/sitemap product import | |
| 6543 | - * (mxchat_extract_woocommerce_product_content). Assembles from the WC_Product, | |
| 6544 | - * the authoritative source for product rows (scope decision on the plan). | |
| 6545 | - */ | |
| 6546 | -private function mxchat_prepare_product_content_for_indexing($product) { | |
| 6547 | - $title = $product->get_name(); | |
| 6548 | - $description = $product->get_description(); | |
| 6549 | - $short_description = $product->get_short_description(); | |
| 6550 | - | |
| 6551 | - // Format content consistently | |
| 6552 | - $content = $title . "\n\n"; | |
| 6553 | - | |
| 6554 | - if (!empty($short_description)) { | |
| 6555 | - $content .= "Short Description: " . wp_strip_all_tags($short_description) . "\n\n"; | |
| 6556 | - } | |
| 6557 | - | |
| 6558 | - if (!empty($description)) { | |
| 6559 | - $content .= wp_strip_all_tags($description) . "\n\n"; | |
| 6560 | - } | |
| 6561 | - | |
| 6562 | - $content .= $this->mxchat_woo_product_summary_lines($product); | |
| 6563 | - $content .= $this->mxchat_woo_custom_tabs_text($product->get_id()); | |
| 6564 | - | |
| 6565 | - return $content; | |
| 6566 | -} | |
| 6567 | - | |
| 6568 | -/** | |
| 6569 | - * Pricing + SKU + categories lines for a product — shared by both assembler kinds | |
| 6570 | - * (the post-fields product enrichment and the WC-object assembler). | |
| 6571 | - */ | |
| 6572 | -private function mxchat_woo_product_summary_lines($product) { | |
| 6573 | - // Add pricing information (base-currency pinned — see mxchat_product_price_lines) | |
| 6574 | - $lines = $this->mxchat_product_price_lines($product); | |
| 6575 | - | |
| 6576 | - $sku = $product->get_sku(); | |
| 6577 | - if (!empty($sku)) { | |
| 6578 | - $lines .= "SKU: " . $sku . "\n"; | |
| 6579 | - } | |
| 6580 | - | |
| 6581 | - // Get product categories | |
| 6582 | - $categories = wp_get_post_terms($product->get_id(), 'product_cat', array('fields' => 'names')); | |
| 6583 | - if (!empty($categories) && !is_wp_error($categories)) { | |
| 6584 | - $lines .= "Categories: " . implode(', ', $categories) . "\n"; | |
| 6585 | - } | |
| 6586 | - | |
| 6587 | - return $lines; | |
| 6588 | -} | |
| 6589 | - | |
| 6590 | -/** | |
| 6591 | - * Custom Product Tabs text (supports "Custom Product Tabs for WooCommerce" by | |
| 6592 | - * Code Parrots) — direct tabs plus applied reusable/saved tabs. The ONE copy of | |
| 6593 | - * the yikes_woo logic; three sites carried byte-identical clones before a3d60c. | |
| 6594 | - */ | |
| 6595 | -private function mxchat_woo_custom_tabs_text($product_id) { | |
| 6596 | - $text = ''; | |
| 6597 | - | |
| 6598 | - $custom_tabs = get_post_meta($product_id, 'yikes_woo_products_tabs', true); | |
| 6599 | - if (!empty($custom_tabs) && is_array($custom_tabs)) { | |
| 6600 | - foreach ($custom_tabs as $tab) { | |
| 6601 | - $tab_title = isset($tab['title']) ? $tab['title'] : (isset($tab['tab_title']) ? $tab['tab_title'] : ''); | |
| 6602 | - $tab_content = isset($tab['content']) ? $tab['content'] : ''; | |
| 6603 | - | |
| 6604 | - if (!empty($tab_title) && !empty($tab_content)) { | |
| 6605 | - $text .= "\n" . $tab_title . ": " . wp_strip_all_tags($tab_content) . "\n"; | |
| 6606 | - } | |
| 5642 | + | |
| 5643 | + if (empty($api_key)) { | |
| 5644 | + return; | |
| 6607 | 5645 | } |
| 5646 | + | |
| 5647 | + // Use the centralized utility function for storage | |
| 5648 | + $result = MxChat_Utils::submit_content_to_db( | |
| 5649 | + $final_content, | |
| 5650 | + $source_url, | |
| 5651 | + $api_key, | |
| 5652 | + md5($source_url) // Vector ID for Pinecone | |
| 5653 | + ); | |
| 5654 | + | |
| 5655 | + // After successful storage, apply role restriction based on tags | |
| 5656 | + if (!is_wp_error($result)) { | |
| 5657 | + $this->apply_role_restriction_to_post($post_id, $source_url); | |
| 5658 | + } | |
| 6608 | 5659 | } |
| 6609 | - | |
| 6610 | - // Also check for reusable/saved tabs applied to this product | |
| 6611 | - $applied_saved_tabs = get_post_meta($product_id, 'yikes_woo_reusable_products_tabs_applied', true); | |
| 6612 | - if (!empty($applied_saved_tabs) && is_array($applied_saved_tabs)) { | |
| 6613 | - $saved_tabs = get_option('yikes_woo_reusable_products_tabs', array()); | |
| 6614 | - if (!empty($saved_tabs) && is_array($saved_tabs)) { | |
| 6615 | - foreach ($applied_saved_tabs as $saved_tab_id) { | |
| 6616 | - if (isset($saved_tabs[$saved_tab_id])) { | |
| 6617 | - $tab = $saved_tabs[$saved_tab_id]; | |
| 6618 | - $tab_title = isset($tab['title']) ? $tab['title'] : (isset($tab['tab_title']) ? $tab['tab_title'] : ''); | |
| 6619 | - $tab_content = isset($tab['content']) ? $tab['content'] : ''; | |
| 6620 | - | |
| 6621 | - if (!empty($tab_title) && !empty($tab_content)) { | |
| 6622 | - $text .= "\n" . $tab_title . ": " . wp_strip_all_tags($tab_content) . "\n"; | |
| 6623 | - } | |
| 6624 | - } | |
| 6625 | - } | |
| 6626 | - } | |
| 5660 | + | |
| 5661 | + // Clean up the stored previous status if not used above | |
| 5662 | + if ($previous_status !== 'publish' || $post->post_status === 'publish') { | |
| 5663 | + delete_transient($previous_status_key); | |
| 5664 | + delete_transient($previous_url_key); | |
| 6627 | 5665 | } |
| 6628 | - | |
| 6629 | - return $text; | |
| 6630 | 5666 | } |
| 6631 | 5667 | |
| 6632 | 5668 | /** |
| 6633 | 5669 | * Store the post status and URL before update to detect status transitions |
| @@ -6633,12 +5669,8 @@ | ||
| 6633 | 5669 | * Store the post status and URL before update to detect status transitions |
| 6634 | 5670 | * This runs before the post is actually updated in the database |
| 6635 | 5671 | */ |
| 6636 | 5672 | public function mxchat_store_pre_update_status($post_id, $data) { |
| 6637 | - // Core is inside wp_insert_post's update branch, so a post_updated WILL fire later | |
| 6638 | - // this request and can consume the arrival-edge guard (plan a664f3). | |
| 6639 | - $this->pending_post_update[$post_id] = true; | |
| 6640 | - | |
| 6641 | 5673 | // Get the current post from database (before update) |
| 6642 | 5674 | $current_post = get_post($post_id); |
| 6643 | 5675 | |
| 6644 | 5676 | if ($current_post) { |
| @@ -6654,407 +5686,8 @@ | ||
| 6654 | 5686 | } |
| 6655 | 5687 | } |
| 6656 | 5688 | } |
| 6657 | 5689 | |
| 6658 | -/** | |
| 6659 | - * Whether auto-sync is enabled for a post type (mirrors the checks used by the | |
| 6660 | - * update/delete handlers; kept as one helper so new call sites cannot drift). | |
| 6661 | - */ | |
| 6662 | -private function mxchat_is_auto_sync_enabled($post_type) { | |
| 6663 | - if ($post_type === 'post') { | |
| 6664 | - return get_option('mxchat_auto_sync_posts') === '1'; | |
| 6665 | - } | |
| 6666 | - if ($post_type === 'page') { | |
| 6667 | - return get_option('mxchat_auto_sync_pages') === '1'; | |
| 6668 | - } | |
| 6669 | - return get_option('mxchat_auto_sync_' . $post_type) === '1'; | |
| 6670 | -} | |
| 6671 | - | |
| 6672 | -/** | |
| 6673 | - * Remove a post's vectors the moment it leaves 'publish', using the authoritative | |
| 6674 | - * old status core passes to transition_post_status — no transient involved (plan 816fb1). | |
| 6675 | - * | |
| 6676 | - * Covers status changes that never route through wp_update_post (scheduled-expiry | |
| 6677 | - * plugins and others that flip post_status directly and call wp_transition_post_status), | |
| 6678 | - * where neither pre_post_update nor post_updated fires and the old detection missed. | |
| 6679 | - */ | |
| 6680 | -public function mxchat_handle_status_transition($new_status, $old_status, $post) { | |
| 6681 | - if (!($post instanceof WP_Post) || wp_is_post_revision($post->ID)) { | |
| 6682 | - return; | |
| 6683 | - } | |
| 6684 | - | |
| 6685 | - // Arrival edge (plan 3055e1): a post BECOMING published is indexed here, because | |
| 6686 | - // wp_publish_post() — the path scheduled posts take via check_and_publish_future_post — | |
| 6687 | - // and direct wp_insert_post(status=publish) creates never fire post_updated, so the | |
| 6688 | - // auto-sync ADD path alone misses them. Editor publishes also pass through here; | |
| 6689 | - // the transition_indexed_posts guard keeps mxchat_handle_post_update from embedding | |
| 6690 | - // a second time in the same request. | |
| 6691 | - if ($new_status === 'publish' && $old_status !== 'publish') { | |
| 6692 | - if ($this->mxchat_is_auto_sync_enabled($post->post_type)) { | |
| 6693 | - $this->mxchat_index_published_post($post->ID, $post); | |
| 6694 | - | |
| 6695 | - // Arm the double-fire guard ONLY when a post_updated is actually coming to | |
| 6696 | - // consume it (plan a664f3). Two publish paths never fire post_updated at all: | |
| 6697 | - // a direct wp_insert_post(status=publish) create, and wp_publish_post() — the | |
| 6698 | - // call check_and_publish_future_post() makes for scheduled posts. Arming the | |
| 6699 | - // guard unconditionally left it set with nothing to consume it, so the NEXT | |
| 6700 | - // update of that post was swallowed entirely: zero embed calls, no knowledge | |
| 6701 | - // -base row, silently. Consume-once on this side too, so a guard can never | |
| 6702 | - // outlive the single save it was armed for. | |
| 6703 | - if (!empty($this->pending_post_update[$post->ID])) { | |
| 6704 | - unset($this->pending_post_update[$post->ID]); | |
| 6705 | - $this->transition_indexed_posts[$post->ID] = true; | |
| 6706 | - } | |
| 6707 | - } | |
| 6708 | - return; | |
| 6709 | - } | |
| 6710 | - | |
| 6711 | - // Only the publish -> not-publish edge matters here. | |
| 6712 | - if ($old_status !== 'publish' || $new_status === 'publish') { | |
| 6713 | - return; | |
| 6714 | - } | |
| 6715 | - // Trash is handled by mxchat_handle_post_delete (wp_trash_post) with pre-trash URL | |
| 6716 | - // resolution; skip to avoid a second network round-trip per trash. | |
| 6717 | - if ($new_status === 'trash') { | |
| 6718 | - return; | |
| 6719 | - } | |
| 6720 | - if (!$this->mxchat_is_auto_sync_enabled($post->post_type)) { | |
| 6721 | - return; | |
| 6722 | - } | |
| 6723 | - | |
| 6724 | - $urls = array(); | |
| 6725 | - | |
| 6726 | - // The DB may already hold the new status when this fires, so get_permalink() on the | |
| 6727 | - // live post could build a draft-style URL whose md5 misses the stored vector IDs. | |
| 6728 | - // Reconstruct the published permalink from a clone instead. | |
| 6729 | - $published_clone = clone $post; | |
| 6730 | - $published_clone->post_status = 'publish'; | |
| 6731 | - $published_url = get_permalink($published_clone); | |
| 6732 | - if ($published_url) { | |
| 6733 | - $urls[] = $published_url; | |
| 6734 | - } | |
| 6735 | - | |
| 6736 | - // Honour the pre-update capture when present (covers a slug change in the same save). | |
| 6737 | - $previous_url = get_transient('mxchat_prev_url_' . $post->ID); | |
| 6738 | - if (!empty($previous_url)) { | |
| 6739 | - $urls[] = $previous_url; | |
| 6740 | - } | |
| 6741 | - | |
| 6742 | - foreach (array_unique($urls) as $url) { | |
| 6743 | - MxChat_Utils::delete_chunks_for_url($url, 'default'); | |
| 6744 | - } | |
| 6745 | - | |
| 6746 | - if (!empty($urls)) { | |
| 6747 | - $this->transition_deleted_posts[$post->ID] = true; | |
| 6748 | - } | |
| 6749 | -} | |
| 6750 | - | |
| 6751 | -/** | |
| 6752 | - * WP-CLI: remove knowledge-base entries left behind by posts that were unpublished, | |
| 6753 | - * trashed, or made private before the transition_post_status handler existed. | |
| 6754 | - * | |
| 6755 | - * Walks every auto-synced post type's non-published posts, reconstructs each one's | |
| 6756 | - * published-era permalink, and deletes its vectors (routes to Pinecone or the WP table). | |
| 6757 | - * Deletion is idempotent, so never-indexed posts are a cheap no-op. | |
| 6758 | - * | |
| 6759 | - * ## OPTIONS | |
| 6760 | - * | |
| 6761 | - * [--dry-run] | |
| 6762 | - * : Report what would be removed without deleting anything. | |
| 6763 | - * | |
| 6764 | - * ## EXAMPLES | |
| 6765 | - * | |
| 6766 | - * wp mxchat prune-unpublished --dry-run | |
| 6767 | - * wp mxchat prune-unpublished | |
| 6768 | - */ | |
| 6769 | -public function cli_prune_unpublished($args, $assoc_args) { | |
| 6770 | - global $wpdb; | |
| 6771 | - $dry_run = !empty($assoc_args['dry-run']); | |
| 6772 | - $table = $wpdb->prefix . 'mxchat_system_prompt_content'; | |
| 6773 | - | |
| 6774 | - $candidate_types = array_merge(array('post', 'page'), array_values(get_post_types(array('_builtin' => false), 'names'))); | |
| 6775 | - $synced_types = array(); | |
| 6776 | - foreach ($candidate_types as $type) { | |
| 6777 | - if ($this->mxchat_is_auto_sync_enabled($type)) { | |
| 6778 | - $synced_types[] = $type; | |
| 6779 | - } | |
| 6780 | - } | |
| 6781 | - if (empty($synced_types)) { | |
| 6782 | - WP_CLI::success('No post types have auto-sync enabled; nothing to prune.'); | |
| 6783 | - return; | |
| 6784 | - } | |
| 6785 | - | |
| 6786 | - $scanned = 0; | |
| 6787 | - $pruned = 0; | |
| 6788 | - $paged = 1; | |
| 6789 | - do { | |
| 6790 | - $query = new WP_Query(array( | |
| 6791 | - 'post_type' => $synced_types, | |
| 6792 | - 'post_status' => array('draft', 'pending', 'private', 'future', 'trash'), | |
| 6793 | - 'posts_per_page' => 100, | |
| 6794 | - 'paged' => $paged, | |
| 6795 | - 'fields' => 'ids', | |
| 6796 | - )); | |
| 6797 | - foreach ($query->posts as $post_id) { | |
| 6798 | - $post = get_post($post_id); | |
| 6799 | - if (!$post) { | |
| 6800 | - continue; | |
| 6801 | - } | |
| 6802 | - $scanned++; | |
| 6803 | - | |
| 6804 | - // Rebuild the permalink the post had while published: publish-status clone, | |
| 6805 | - // with wp_trash_post's __trashed slug suffix stripped for trashed posts. | |
| 6806 | - $clone = clone $post; | |
| 6807 | - $clone->post_status = 'publish'; | |
| 6808 | - if (substr($clone->post_name, -9) === '__trashed') { | |
| 6809 | - $clone->post_name = substr($clone->post_name, 0, -9); | |
| 6810 | - } | |
| 6811 | - $url = get_permalink($clone); | |
| 6812 | - if (!$url) { | |
| 6813 | - continue; | |
| 6814 | - } | |
| 6815 | - | |
| 6816 | - // Local-table row count is exact in WordPress-DB mode; in Pinecone mode it | |
| 6817 | - // reads 0 but the delete below still routes to Pinecone and is idempotent. | |
| 6818 | - $local_rows = (int) $wpdb->get_var($wpdb->prepare( | |
| 6819 | - "SELECT COUNT(*) FROM {$table} WHERE source_url = %s", $url | |
| 6820 | - )); | |
| 6821 | - | |
| 6822 | - if ($dry_run) { | |
| 6823 | - if ($local_rows > 0) { | |
| 6824 | - WP_CLI::log(sprintf('Would remove %d row(s): %s (post %d, %s)', $local_rows, $url, $post_id, $post->post_status)); | |
| 6825 | - $pruned += $local_rows; | |
| 6826 | - } | |
| 6827 | - continue; | |
| 6828 | - } | |
| 6829 | - | |
| 6830 | - MxChat_Utils::delete_chunks_for_url($url, 'default'); | |
| 6831 | - if ($local_rows > 0) { | |
| 6832 | - WP_CLI::log(sprintf('Removed %d row(s): %s (post %d, %s)', $local_rows, $url, $post_id, $post->post_status)); | |
| 6833 | - $pruned += $local_rows; | |
| 6834 | - } | |
| 6835 | - } | |
| 6836 | - $more = $paged < $query->max_num_pages; | |
| 6837 | - $paged++; | |
| 6838 | - } while ($more); | |
| 6839 | - | |
| 6840 | - WP_CLI::success(sprintf( | |
| 6841 | - '%s %d local knowledge row(s) across %d non-published post(s) scanned.%s', | |
| 6842 | - $dry_run ? 'Would remove' : 'Removed', | |
| 6843 | - $pruned, | |
| 6844 | - $scanned, | |
| 6845 | - ' (Pinecone-mode deletions are not counted locally.)' | |
| 6846 | - )); | |
| 6847 | -} | |
| 6848 | - | |
| 6849 | -/** | |
| 6850 | - * WP-CLI: repair knowledge-base rows whose PDF text was imported in visual | |
| 6851 | - * (reversed) order before the RTL normalizer existed. 32bf9e fixed new | |
| 6852 | - * imports only; this fixes rows already in the table without the customer | |
| 6853 | - * having to re-source and re-upload the original PDFs (plan d1e6f7). | |
| 6854 | - * | |
| 6855 | - * Detection reuses MxChat_Utils::normalize_pdf_rtl() on the stored text: a | |
| 6856 | - * row is a candidate exactly when the normalizer would change it, so the | |
| 6857 | - * import-time heuristic and the repair heuristic can never disagree. | |
| 6858 | - * Repaired rows are RE-EMBEDDED — the stored vector was computed over | |
| 6859 | - * reversed text and is as broken as the text — so a wet run calls the | |
| 6860 | - * embedding provider once per repaired row on the site's API key. Runs | |
| 6861 | - * beyond 25 rows therefore require --yes. | |
| 6862 | - * | |
| 6863 | - * Scope notes: | |
| 6864 | - * - Scans the WordPress knowledge table. Pinecone-mode entries live in | |
| 6865 | - * Pinecone, not this table, and are not scanned; if a scanned row's bot | |
| 6866 | - * ALSO has Pinecone enabled (hybrid drift), the repaired entry is | |
| 6867 | - * re-submitted through the normal import path so the md5-keyed Pinecone | |
| 6868 | - * vector is replaced too. | |
| 6869 | - * - Knowledge rows do not carry a bot id; --bot only selects whose | |
| 6870 | - * embedding configuration (model + key) is used for re-embedding. | |
| 6871 | - * - The mxchat_pdf_rtl_normalize filter is honoured: a site that disabled | |
| 6872 | - * normalization gets detections of zero, not surprise rewrites. | |
| 6873 | - * - The metadata header the PDF importer stores before the text separator | |
| 6874 | - * is preserved byte-identical; only the text segment is repaired. | |
| 6875 | - * | |
| 6876 | - * ## OPTIONS | |
| 6877 | - * | |
| 6878 | - * [--dry-run] | |
| 6879 | - * : List the rows that would be repaired without changing anything. | |
| 6880 | - * | |
| 6881 | - * [--bot=<id>] | |
| 6882 | - * : Embedding configuration to use for re-embedding. Default: default. | |
| 6883 | - * | |
| 6884 | - * [--all-content] | |
| 6885 | - * : Scan every row containing right-to-left text, not just rows with PDF | |
| 6886 | - * provenance (a page anchor in the source URL, or pdf content type). | |
| 6887 | - * | |
| 6888 | - * [--yes] | |
| 6889 | - * : Proceed even when more than 25 rows need re-embedding (API cost gate). | |
| 6890 | - * | |
| 6891 | - * ## EXAMPLES | |
| 6892 | - * | |
| 6893 | - * wp mxchat rtl-repair --dry-run | |
| 6894 | - * wp mxchat rtl-repair | |
| 6895 | - * wp mxchat rtl-repair --all-content --yes | |
| 6896 | - */ | |
| 6897 | -public function cli_rtl_repair($args, $assoc_args) { | |
| 6898 | - global $wpdb; | |
| 6899 | - $dry_run = !empty($assoc_args['dry-run']); | |
| 6900 | - $all = !empty($assoc_args['all-content']); | |
| 6901 | - $yes = !empty($assoc_args['yes']); | |
| 6902 | - $bot_id = isset($assoc_args['bot']) ? sanitize_key($assoc_args['bot']) : 'default'; | |
| 6903 | - $table = $wpdb->prefix . 'mxchat_system_prompt_content'; | |
| 6904 | - | |
| 6905 | - // Detection pass — no API calls. Walk the table in id batches so a large | |
| 6906 | - // knowledge base never loads at once. | |
| 6907 | - $rtl_re = '/[\x{0590}-\x{05FF}\x{0600}-\x{06FF}\x{0750}-\x{077F}\x{FB50}-\x{FDFF}\x{FE70}-\x{FEFF}]/u'; | |
| 6908 | - $candidates = array(); | |
| 6909 | - $scanned = 0; | |
| 6910 | - $last_id = 0; | |
| 6911 | - do { | |
| 6912 | - if ($all) { | |
| 6913 | - $rows = $wpdb->get_results($wpdb->prepare( | |
| 6914 | - "SELECT id, article_content, source_url, content_type FROM {$table} | |
| 6915 | - WHERE id > %d ORDER BY id ASC LIMIT 200", | |
| 6916 | - $last_id | |
| 6917 | - )); | |
| 6918 | - } else { | |
| 6919 | - $rows = $wpdb->get_results($wpdb->prepare( | |
| 6920 | - "SELECT id, article_content, source_url, content_type FROM {$table} | |
| 6921 | - WHERE id > %d AND (source_url LIKE %s OR content_type = 'pdf') | |
| 6922 | - ORDER BY id ASC LIMIT 200", | |
| 6923 | - $last_id, | |
| 6924 | - '%' . $wpdb->esc_like('#page=') . '%' | |
| 6925 | - )); | |
| 6926 | - } | |
| 6927 | - foreach ($rows as $row) { | |
| 6928 | - $last_id = (int) $row->id; | |
| 6929 | - $scanned++; | |
| 6930 | - $content = (string) $row->article_content; | |
| 6931 | - if (!preg_match($rtl_re, $content)) { | |
| 6932 | - continue; | |
| 6933 | - } | |
| 6934 | - list($header, $text) = $this->mxchat_rtl_repair_split($content); | |
| 6935 | - $normalized = MxChat_Utils::normalize_pdf_rtl($text, 'rtl-repair row ' . $row->id); | |
| 6936 | - if (is_string($normalized) && $normalized !== $text) { | |
| 6937 | - $candidates[] = array( | |
| 6938 | - 'id' => (int) $row->id, | |
| 6939 | - 'source_url' => (string) $row->source_url, | |
| 6940 | - 'content_type' => (string) $row->content_type, | |
| 6941 | - 'new_content' => $header . $normalized, | |
| 6942 | - ); | |
| 6943 | - } | |
| 6944 | - } | |
| 6945 | - } while (count($rows) === 200); | |
| 6946 | - | |
| 6947 | - WP_CLI::log(sprintf('Scanned %d row(s); %d stored in reversed (visual) order.', $scanned, count($candidates))); | |
| 6948 | - if (empty($candidates)) { | |
| 6949 | - WP_CLI::success('No reversed RTL rows found — nothing to repair.'); | |
| 6950 | - return; | |
| 6951 | - } | |
| 6952 | - | |
| 6953 | - foreach ($candidates as $c) { | |
| 6954 | - WP_CLI::log(sprintf('%s row %d %s', $dry_run ? 'Would repair' : 'Will repair', $c['id'], $c['source_url'])); | |
| 6955 | - } | |
| 6956 | - if ($dry_run) { | |
| 6957 | - WP_CLI::success(sprintf('Dry run: %d row(s) would be repaired and re-embedded. Run without --dry-run to apply.', count($candidates))); | |
| 6958 | - return; | |
| 6959 | - } | |
| 6960 | - | |
| 6961 | - // Cost gate: re-embedding spends the customer's API budget. | |
| 6962 | - WP_CLI::log(sprintf('Re-embedding will call the embedding provider once per row — %d call(s) on this site\'s API key.', count($candidates))); | |
| 6963 | - if (count($candidates) > 25 && !$yes) { | |
| 6964 | - 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))); | |
| 6965 | - } | |
| 6966 | - | |
| 6967 | - $bot_options = $this->get_bot_options($bot_id); | |
| 6968 | - $options = !empty($bot_options) ? $bot_options : get_option('mxchat_options'); | |
| 6969 | - $preflight = MxChat_Utils::embedding_preflight($options); | |
| 6970 | - if (!$preflight['ok']) { | |
| 6971 | - WP_CLI::error('Embedding configuration problem: ' . $preflight['reason']); | |
| 6972 | - } | |
| 6973 | - $api_key = $preflight['api_key']; | |
| 6974 | - | |
| 6975 | - $pinecone_hybrid = $this->mxchat_rtl_repair_pinecone_enabled($bot_id); | |
| 6976 | - $repaired = 0; | |
| 6977 | - $failed = 0; | |
| 6978 | - foreach ($candidates as $c) { | |
| 6979 | - $vector = MxChat_Utils::regenerate_embedding($c['new_content'], $api_key, $bot_id); | |
| 6980 | - if (!is_array($vector)) { | |
| 6981 | - $failed++; | |
| 6982 | - $reason = is_wp_error($vector) ? $vector->get_error_message() : 'embedding request failed'; | |
| 6983 | - // Text and vector must stay consistent: never write repaired text | |
| 6984 | - // beside the stale reversed-text vector. | |
| 6985 | - WP_CLI::warning(sprintf('Row %d NOT repaired — %s. Row left unchanged.', $c['id'], $reason)); | |
| 6986 | - continue; | |
| 6987 | - } | |
| 6988 | - $wpdb->update( | |
| 6989 | - $table, | |
| 6990 | - array( | |
| 6991 | - 'article_content' => $c['new_content'], | |
| 6992 | - 'embedding_vector' => maybe_serialize($vector), | |
| 6993 | - ), | |
| 6994 | - array('id' => $c['id']), | |
| 6995 | - array('%s', '%s'), | |
| 6996 | - array('%d') | |
| 6997 | - ); | |
| 6998 | - $repaired++; | |
| 6999 | - if (class_exists('MxChat_Admin')) { | |
| 7000 | - MxChat_Admin::mxchat_log_debug('pdf_rtl_repaired', 'Stored KB row restored to logical order and re-embedded', array( | |
| 7001 | - 'row_id' => $c['id'], | |
| 7002 | - 'source_url' => $c['source_url'], | |
| 7003 | - 'bot' => $bot_id, | |
| 7004 | - )); | |
| 7005 | - } | |
| 7006 | - // Hybrid drift: the bot indexes into Pinecone but this row sat in the | |
| 7007 | - // WP table — push the repaired entry through the normal import path so | |
| 7008 | - // the md5(source_url)-keyed Pinecone vector is replaced as well. | |
| 7009 | - if ($pinecone_hybrid) { | |
| 7010 | - MxChat_Utils::submit_content_to_db( | |
| 7011 | - $c['new_content'], | |
| 7012 | - $c['source_url'], | |
| 7013 | - $api_key, | |
| 7014 | - null, | |
| 7015 | - $bot_id, | |
| 7016 | - $c['content_type'] !== '' ? $c['content_type'] : 'pdf' | |
| 7017 | - ); | |
| 7018 | - } | |
| 7019 | - } | |
| 7020 | - | |
| 7021 | - WP_CLI::success(sprintf('Repaired + re-embedded %d row(s); %d failed; %d scanned.', $repaired, $failed, $scanned)); | |
| 7022 | -} | |
| 7023 | - | |
| 7024 | -/** | |
| 7025 | - * Split a stored KB row into (metadata header incl. separator, text segment). | |
| 7026 | - * The PDF importer stores wp_json_encode($metadata) . "\n---\n" . $text — | |
| 7027 | - * repair must touch only the text and keep the header byte-identical. | |
| 7028 | - */ | |
| 7029 | -private function mxchat_rtl_repair_split($content) { | |
| 7030 | - $sep = "\n---\n"; | |
| 7031 | - $pos = strpos($content, $sep); | |
| 7032 | - if ($pos !== false && $pos > 0 && $content[0] === '{') { | |
| 7033 | - $maybe_json = substr($content, 0, $pos); | |
| 7034 | - if (json_decode($maybe_json) !== null) { | |
| 7035 | - return array(substr($content, 0, $pos + strlen($sep)), substr($content, $pos + strlen($sep))); | |
| 7036 | - } | |
| 7037 | - } | |
| 7038 | - return array('', $content); | |
| 7039 | -} | |
| 7040 | - | |
| 7041 | -/** | |
| 7042 | - * Mirror of MxChat_Utils::is_pinecone_enabled_for_bot() (private there) for | |
| 7043 | - * the repair CLI's hybrid-drift check. | |
| 7044 | - */ | |
| 7045 | -private function mxchat_rtl_repair_pinecone_enabled($bot_id) { | |
| 7046 | - if ($bot_id !== 'default' && class_exists('MxChat_Multi_Bot_Manager')) { | |
| 7047 | - $cfg = apply_filters('mxchat_get_bot_pinecone_config', array(), $bot_id); | |
| 7048 | - if (!empty($cfg)) { | |
| 7049 | - return !empty($cfg['use_pinecone']) && !empty($cfg['api_key']) && !empty($cfg['host']); | |
| 7050 | - } | |
| 7051 | - } | |
| 7052 | - $po = get_option('mxchat_pinecone_addon_options'); | |
| 7053 | - return !empty($po['mxchat_use_pinecone']) && $po['mxchat_use_pinecone'] !== '0' | |
| 7054 | - && !empty($po['mxchat_pinecone_api_key']) && !empty($po['mxchat_pinecone_host']); | |
| 7055 | -} | |
| 7056 | - | |
| 7057 | 5690 | public function mxchat_handle_post_delete($post_id) { |
| 7058 | 5691 | // Get post data before it's deleted |
| 7059 | 5692 | $post = get_post($post_id); |
| 7060 | 5693 | |
| @@ -7154,20 +5787,110 @@ | ||
| 7154 | 5787 | |
| 7155 | 5788 | $source_url = get_permalink($product->get_id()); |
| 7156 | 5789 | $product_id = $product->get_id(); |
| 7157 | 5790 | |
| 7158 | - // Build product content via the shared WC-object assembler (a3d60c) — this | |
| 7159 | - // writer owns product rows whenever the integration is on. | |
| 7160 | - $content = $this->mxchat_prepare_product_content_for_indexing($product); | |
| 5791 | + // Build product content | |
| 5792 | + $title = $product->get_name(); | |
| 5793 | + $description = $product->get_description(); | |
| 5794 | + $short_description = $product->get_short_description(); | |
| 5795 | + $regular_price = $product->get_regular_price(); | |
| 5796 | + $sale_price = $product->get_sale_price(); | |
| 5797 | + $price = $product->get_price(); | |
| 5798 | + $sku = $product->get_sku(); | |
| 7161 | 5799 | |
| 7162 | - // Embedding decision — custom-provider-aware (plan cbd5fd); silent-return | |
| 7163 | - // shape preserved. | |
| 7164 | - $preflight = MxChat_Utils::embedding_preflight(get_option('mxchat_options')); | |
| 7165 | - if (!$preflight['ok']) { | |
| 7166 | - //error_log('MxChat Auto-sync: embedding pre-flight failed: ' . $preflight['reason']); | |
| 5800 | + // Get currency symbol | |
| 5801 | + $currency_symbol = get_woocommerce_currency_symbol(); | |
| 5802 | + | |
| 5803 | + // Format content consistently | |
| 5804 | + $content = $title . "\n\n"; | |
| 5805 | + | |
| 5806 | + if (!empty($short_description)) { | |
| 5807 | + $content .= "Short Description: " . wp_strip_all_tags($short_description) . "\n\n"; | |
| 5808 | + } | |
| 5809 | + | |
| 5810 | + if (!empty($description)) { | |
| 5811 | + $content .= wp_strip_all_tags($description) . "\n\n"; | |
| 5812 | + } | |
| 5813 | + | |
| 5814 | + // Add pricing information | |
| 5815 | + if (!empty($regular_price)) { | |
| 5816 | + $content .= "Price: " . $currency_symbol . $regular_price . "\n"; | |
| 5817 | + } elseif (!empty($price)) { | |
| 5818 | + $content .= "Price: " . $currency_symbol . $price . "\n"; | |
| 5819 | + } | |
| 5820 | + | |
| 5821 | + if (!empty($sale_price) && $sale_price !== $regular_price) { | |
| 5822 | + $content .= "Sale Price: " . $currency_symbol . $sale_price . "\n"; | |
| 5823 | + } | |
| 5824 | + | |
| 5825 | + // Handle variable products - show price range | |
| 5826 | + if ($product->is_type('variable')) { | |
| 5827 | + $min_price = $product->get_variation_price('min'); | |
| 5828 | + $max_price = $product->get_variation_price('max'); | |
| 5829 | + if ($min_price !== $max_price) { | |
| 5830 | + $content .= "Price Range: " . $currency_symbol . $min_price . " - " . $currency_symbol . $max_price . "\n"; | |
| 5831 | + } | |
| 5832 | + } | |
| 5833 | + | |
| 5834 | + if (!empty($sku)) { | |
| 5835 | + $content .= "SKU: " . $sku . "\n"; | |
| 5836 | + } | |
| 5837 | + | |
| 5838 | + // Get product categories | |
| 5839 | + $categories = wp_get_post_terms($product_id, 'product_cat', array('fields' => 'names')); | |
| 5840 | + if (!empty($categories) && !is_wp_error($categories)) { | |
| 5841 | + $content .= "Categories: " . implode(', ', $categories) . "\n"; | |
| 5842 | + } | |
| 5843 | + | |
| 5844 | + // Get Custom Product Tabs (supports "Custom Product Tabs for WooCommerce" by Code Parrots) | |
| 5845 | + $custom_tabs = get_post_meta($product_id, 'yikes_woo_products_tabs', true); | |
| 5846 | + if (!empty($custom_tabs) && is_array($custom_tabs)) { | |
| 5847 | + foreach ($custom_tabs as $tab) { | |
| 5848 | + $tab_title = isset($tab['title']) ? $tab['title'] : (isset($tab['tab_title']) ? $tab['tab_title'] : ''); | |
| 5849 | + $tab_content = isset($tab['content']) ? $tab['content'] : ''; | |
| 5850 | + | |
| 5851 | + if (!empty($tab_title) && !empty($tab_content)) { | |
| 5852 | + $content .= "\n" . $tab_title . ": " . wp_strip_all_tags($tab_content) . "\n"; | |
| 5853 | + } | |
| 5854 | + } | |
| 5855 | + } | |
| 5856 | + | |
| 5857 | + // Also check for reusable/saved tabs applied to this product | |
| 5858 | + $applied_saved_tabs = get_post_meta($product_id, 'yikes_woo_reusable_products_tabs_applied', true); | |
| 5859 | + if (!empty($applied_saved_tabs) && is_array($applied_saved_tabs)) { | |
| 5860 | + // Get the saved tabs option | |
| 5861 | + $saved_tabs = get_option('yikes_woo_reusable_products_tabs', array()); | |
| 5862 | + if (!empty($saved_tabs) && is_array($saved_tabs)) { | |
| 5863 | + foreach ($applied_saved_tabs as $saved_tab_id) { | |
| 5864 | + if (isset($saved_tabs[$saved_tab_id])) { | |
| 5865 | + $tab = $saved_tabs[$saved_tab_id]; | |
| 5866 | + $tab_title = isset($tab['title']) ? $tab['title'] : (isset($tab['tab_title']) ? $tab['tab_title'] : ''); | |
| 5867 | + $tab_content = isset($tab['content']) ? $tab['content'] : ''; | |
| 5868 | + | |
| 5869 | + if (!empty($tab_title) && !empty($tab_content)) { | |
| 5870 | + $content .= "\n" . $tab_title . ": " . wp_strip_all_tags($tab_content) . "\n"; | |
| 5871 | + } | |
| 5872 | + } | |
| 5873 | + } | |
| 5874 | + } | |
| 5875 | + } | |
| 5876 | + | |
| 5877 | + // Get API key with proper model detection | |
| 5878 | + $options = get_option('mxchat_options'); | |
| 5879 | + $selected_model = $options['embedding_model'] ?? 'text-embedding-ada-002'; | |
| 5880 | + | |
| 5881 | + if (strpos($selected_model, 'voyage') === 0) { | |
| 5882 | + $api_key = $options['voyage_api_key'] ?? ''; | |
| 5883 | + } elseif (strpos($selected_model, 'gemini-embedding') === 0) { | |
| 5884 | + $api_key = $options['gemini_api_key'] ?? ''; | |
| 5885 | + } else { | |
| 5886 | + $api_key = $options['api_key'] ?? ''; | |
| 5887 | + } | |
| 5888 | + | |
| 5889 | + if (empty($api_key)) { | |
| 5890 | + //error_log('MxChat Auto-sync: No API key configured for embedding model'); | |
| 7167 | 5891 | return; |
| 7168 | 5892 | } |
| 7169 | - $api_key = $preflight['api_key']; | |
| 7170 | 5893 | |
| 7171 | 5894 | // Use the centralized utility function for storage |
| 7172 | 5895 | $result = MxChat_Utils::submit_content_to_db( |
| 7173 | 5896 | $content, |
| @@ -7243,23 +5966,17 @@ | ||
| 7243 | 5966 | |
| 7244 | 5967 | // Delete from Pinecone |
| 7245 | 5968 | $pinecone_manager = MxChat_Pinecone_Manager::get_instance(); |
| 7246 | 5969 | $result = $pinecone_manager->mxchat_delete_from_pinecone_by_vector_id( |
| 7247 | - $vector_id, | |
| 7248 | - $pinecone_options['mxchat_pinecone_api_key'], | |
| 7249 | - $pinecone_options['mxchat_pinecone_host'], | |
| 7250 | - $pinecone_options['mxchat_pinecone_namespace'] ?? '' | |
| 5970 | + $vector_id, | |
| 5971 | + $pinecone_options['mxchat_pinecone_api_key'], | |
| 5972 | + $pinecone_options['mxchat_pinecone_host'] | |
| 7251 | 5973 | ); |
| 7252 | 5974 | |
| 7253 | 5975 | if ($result['success']) { |
| 7254 | - // Mirror the removal to the OpenAI Vector Store mapping (plan 15b5c6); | |
| 7255 | - // a chunk vector id reduces to its base entry there. | |
| 7256 | - if (class_exists('MxChat_Vectorstore_Manager')) { | |
| 7257 | - MxChat_Vectorstore_Manager::sync_delete_by_key($vector_id, 'default'); | |
| 7258 | - } | |
| 7259 | 5976 | // No cache clearing needed since we removed caching |
| 7260 | - set_transient('mxchat_admin_notice_success', | |
| 7261 | - esc_html__('Entry deleted successfully from Pinecone.', 'mxchat'), | |
| 5977 | + set_transient('mxchat_admin_notice_success', | |
| 5978 | + esc_html__('Entry deleted successfully from Pinecone.', 'mxchat'), | |
| 7262 | 5979 | 30 |
| 7263 | 5980 | ); |
| 7264 | 5981 | } else { |
| 7265 | 5982 | set_transient('mxchat_admin_notice_error', |
| @@ -7304,21 +6021,16 @@ | ||
| 7304 | 6021 | wp_send_json_error('Pinecone is not properly configured for bot: ' . $bot_id); |
| 7305 | 6022 | exit; |
| 7306 | 6023 | } |
| 7307 | 6024 | |
| 7308 | - // Delete from the correct Pinecone index and namespace | |
| 6025 | + // Delete from the correct Pinecone index | |
| 7309 | 6026 | $result = $pinecone_manager->mxchat_delete_from_pinecone_by_vector_id( |
| 7310 | - $vector_id, | |
| 7311 | - $pinecone_options['mxchat_pinecone_api_key'], | |
| 7312 | - $pinecone_options['mxchat_pinecone_host'], | |
| 7313 | - $pinecone_options['mxchat_pinecone_namespace'] ?? '' | |
| 6027 | + $vector_id, | |
| 6028 | + $pinecone_options['mxchat_pinecone_api_key'], | |
| 6029 | + $pinecone_options['mxchat_pinecone_host'] | |
| 7314 | 6030 | ); |
| 7315 | 6031 | |
| 7316 | 6032 | if ($result['success']) { |
| 7317 | - // Mirror the removal to the OpenAI Vector Store mapping (plan 15b5c6) | |
| 7318 | - if (class_exists('MxChat_Vectorstore_Manager')) { | |
| 7319 | - MxChat_Vectorstore_Manager::sync_delete_by_key($vector_id, $bot_id); | |
| 7320 | - } | |
| 7321 | 6033 | // No cache clearing needed since we removed caching |
| 7322 | 6034 | wp_send_json_success(array( |
| 7323 | 6035 | 'message' => 'Entry deleted successfully from Pinecone', |
| 7324 | 6036 | 'vector_id' => $vector_id, |
| @@ -7417,13 +6129,8 @@ | ||
| 7417 | 6129 | } |
| 7418 | 6130 | } |
| 7419 | 6131 | |
| 7420 | 6132 | if (empty($vectors_to_delete)) { |
| 7421 | - // Entry already gone from Pinecone — still clear any mirrored | |
| 7422 | - // Vector Store file so it can't outlive the entry (plan 15b5c6). | |
| 7423 | - if (class_exists('MxChat_Vectorstore_Manager')) { | |
| 7424 | - MxChat_Vectorstore_Manager::sync_delete_entry($source_url, $bot_id); | |
| 7425 | - } | |
| 7426 | 6133 | wp_send_json_success(array( |
| 7427 | 6134 | 'message' => 'No vectors found to delete', |
| 7428 | 6135 | 'source_url' => $source_url |
| 7429 | 6136 | )); |
| @@ -7464,13 +6171,8 @@ | ||
| 7464 | 6171 | wp_send_json_error('Pinecone API error (HTTP ' . $response_code . ')'); |
| 7465 | 6172 | exit; |
| 7466 | 6173 | } |
| 7467 | 6174 | |
| 7468 | - // Mirror the removal to the OpenAI Vector Store (plan 15b5c6) | |
| 7469 | - if (class_exists('MxChat_Vectorstore_Manager')) { | |
| 7470 | - MxChat_Vectorstore_Manager::sync_delete_entry($source_url, $bot_id); | |
| 7471 | - } | |
| 7472 | - | |
| 7473 | 6175 | wp_send_json_success(array( |
| 7474 | 6176 | 'message' => 'All chunks deleted successfully from Pinecone', |
| 7475 | 6177 | 'source_url' => $source_url, |
| 7476 | 6178 | 'deleted_count' => count($vectors_to_delete) |
| @@ -7492,13 +6194,8 @@ | ||
| 7492 | 6194 | wp_send_json_error('Failed to delete from database: ' . $wpdb->last_error); |
| 7493 | 6195 | exit; |
| 7494 | 6196 | } |
| 7495 | 6197 | |
| 7496 | - // Mirror the removal to the OpenAI Vector Store (plan 15b5c6) | |
| 7497 | - if (class_exists('MxChat_Vectorstore_Manager')) { | |
| 7498 | - MxChat_Vectorstore_Manager::sync_delete_entry($source_url, $bot_id); | |
| 7499 | - } | |
| 7500 | - | |
| 7501 | 6198 | wp_send_json_success(array( |
| 7502 | 6199 | 'message' => 'All chunks deleted successfully from database', |
| 7503 | 6200 | 'source_url' => $source_url, |
| 7504 | 6201 | 'deleted_count' => $result |
| @@ -7533,15 +6230,8 @@ | ||
| 7533 | 6230 | |
| 7534 | 6231 | global $wpdb; |
| 7535 | 6232 | $table_name = $wpdb->prefix . 'mxchat_system_prompt_content'; |
| 7536 | 6233 | |
| 7537 | - // Capture the identity BEFORE the row disappears — needed to mirror the | |
| 7538 | - // change into the Vector Store (plan 15b5c6). | |
| 7539 | - $source_url = $wpdb->get_var($wpdb->prepare( | |
| 7540 | - "SELECT source_url FROM {$table_name} WHERE id = %d", | |
| 7541 | - $entry_id | |
| 7542 | - )); | |
| 7543 | - | |
| 7544 | 6234 | // Clear cache for this entry |
| 7545 | 6235 | wp_cache_delete('prompt_' . $entry_id, 'mxchat_prompts'); |
| 7546 | 6236 | |
| 7547 | 6237 | // Delete from database |
| @@ -7551,23 +6241,8 @@ | ||
| 7551 | 6241 | array('%d') |
| 7552 | 6242 | ); |
| 7553 | 6243 | |
| 7554 | 6244 | if ($result !== false) { |
| 7555 | - // Mirror to the Vector Store: if sibling rows remain (this was one | |
| 7556 | - // chunk of a larger entry) the entry's file is REFRESHED from what's | |
| 7557 | - // left; if none remain, the file is removed. | |
| 7558 | - if (!empty($source_url) && class_exists('MxChat_Vectorstore_Manager')) { | |
| 7559 | - $remaining = (int) $wpdb->get_var($wpdb->prepare( | |
| 7560 | - "SELECT COUNT(*) FROM {$table_name} WHERE source_url = %s", | |
| 7561 | - $source_url | |
| 7562 | - )); | |
| 7563 | - if ($remaining > 0) { | |
| 7564 | - MxChat_Vectorstore_Manager::sync_upsert_entry($source_url, '', 'default'); | |
| 7565 | - } else { | |
| 7566 | - MxChat_Vectorstore_Manager::sync_delete_entry($source_url, 'default'); | |
| 7567 | - } | |
| 7568 | - } | |
| 7569 | - | |
| 7570 | 6245 | wp_send_json_success(array( |
| 7571 | 6246 | 'message' => 'Entry deleted successfully', |
| 7572 | 6247 | 'entry_id' => $entry_id |
| 7573 | 6248 | )); |
| @@ -7620,9 +6295,8 @@ | ||
| 7620 | 6295 | $use_pinecone = ($pinecone_options['mxchat_use_pinecone'] ?? '0') === '1'; |
| 7621 | 6296 | |
| 7622 | 6297 | $api_key = $pinecone_options['mxchat_pinecone_api_key'] ?? ''; |
| 7623 | 6298 | $host = $pinecone_options['mxchat_pinecone_host'] ?? ''; |
| 7624 | - $namespace = $pinecone_options['mxchat_pinecone_namespace'] ?? ''; | |
| 7625 | 6299 | |
| 7626 | 6300 | // ============================================= |
| 7627 | 6301 | // PHASE 1: Collect all Pinecone vector IDs |
| 7628 | 6302 | // and separate WordPress entries |
| @@ -7629,10 +6303,8 @@ | ||
| 7629 | 6303 | // ============================================= |
| 7630 | 6304 | $pinecone_entry_ids = array(); // entry IDs that are pinecone-sourced |
| 7631 | 6305 | $wordpress_entries = array(); // entries for WordPress DB deletion |
| 7632 | 6306 | $all_vector_ids = array(); // all pinecone vector IDs to delete in one batch |
| 7633 | - $vs_mirror_urls = array(); // Vector Store mirror: URLs to delete (plan 15b5c6) | |
| 7634 | - $vs_mirror_keys = array(); // Vector Store mirror: bare vector ids to delete | |
| 7635 | 6307 | |
| 7636 | 6308 | foreach ($entries as $entry) { |
| 7637 | 6309 | $entry_id = sanitize_text_field($entry['id'] ?? ''); |
| 7638 | 6310 | $source = sanitize_text_field($entry['source'] ?? 'wordpress'); |
| @@ -7657,11 +6329,8 @@ | ||
| 7657 | 6329 | $base_vector_id = md5($source_url); |
| 7658 | 6330 | $all_vector_ids[] = $base_vector_id; |
| 7659 | 6331 | |
| 7660 | 6332 | $list_url = 'https://' . $host . '/vectors/list?prefix=' . urlencode($base_vector_id . '_chunk_') . '&limit=100'; |
| 7661 | - if (!empty($namespace)) { | |
| 7662 | - $list_url .= '&namespace=' . rawurlencode($namespace); | |
| 7663 | - } | |
| 7664 | 6333 | $list_response = wp_remote_get($list_url, array( |
| 7665 | 6334 | 'headers' => array( |
| 7666 | 6335 | 'Api-Key' => $api_key, |
| 7667 | 6336 | 'accept' => 'application/json' |
| @@ -7682,14 +6351,8 @@ | ||
| 7682 | 6351 | } else { |
| 7683 | 6352 | // Single entry: the entry_id IS the vector ID |
| 7684 | 6353 | $all_vector_ids[] = $entry_id; |
| 7685 | 6354 | } |
| 7686 | - | |
| 7687 | - if (!empty($source_url)) { | |
| 7688 | - $vs_mirror_urls[] = $source_url; | |
| 7689 | - } else { | |
| 7690 | - $vs_mirror_keys[] = $entry_id; | |
| 7691 | - } | |
| 7692 | 6355 | } else { |
| 7693 | 6356 | $wordpress_entries[] = $entry; |
| 7694 | 6357 | } |
| 7695 | 6358 | } |
| @@ -7702,12 +6365,8 @@ | ||
| 7702 | 6365 | $pinecone_success = true; |
| 7703 | 6366 | $batches = array_chunk($all_vector_ids, 100); |
| 7704 | 6367 | |
| 7705 | 6368 | foreach ($batches as $batch) { |
| 7706 | - $delete_body = array('ids' => $batch); | |
| 7707 | - if (!empty($namespace)) { | |
| 7708 | - $delete_body['namespace'] = $namespace; | |
| 7709 | - } | |
| 7710 | 6369 | $delete_response = wp_remote_post("https://{$host}/vectors/delete", array( |
| 7711 | 6370 | 'headers' => array( |
| 7712 | 6371 | 'Api-Key' => $api_key, |
| 7713 | 6372 | 'accept' => 'application/json', |
| @@ -7712,9 +6371,9 @@ | ||
| 7712 | 6371 | 'Api-Key' => $api_key, |
| 7713 | 6372 | 'accept' => 'application/json', |
| 7714 | 6373 | 'content-type' => 'application/json' |
| 7715 | 6374 | ), |
| 7716 | - 'body' => wp_json_encode($delete_body), | |
| 6375 | + 'body' => wp_json_encode(array('ids' => $batch)), | |
| 7717 | 6376 | 'timeout' => 60 |
| 7718 | 6377 | )); |
| 7719 | 6378 | |
| 7720 | 6379 | if (is_wp_error($delete_response)) { |
| @@ -7739,18 +6398,8 @@ | ||
| 7739 | 6398 | } else { |
| 7740 | 6399 | $failed_ids[] = $eid; |
| 7741 | 6400 | } |
| 7742 | 6401 | } |
| 7743 | - | |
| 7744 | - // Mirror the removals to the OpenAI Vector Store (plan 15b5c6) | |
| 7745 | - if ($pinecone_success && class_exists('MxChat_Vectorstore_Manager')) { | |
| 7746 | - foreach (array_unique($vs_mirror_urls) as $vs_url) { | |
| 7747 | - MxChat_Vectorstore_Manager::sync_delete_entry($vs_url, $bot_id); | |
| 7748 | - } | |
| 7749 | - foreach (array_unique($vs_mirror_keys) as $vs_key) { | |
| 7750 | - MxChat_Vectorstore_Manager::sync_delete_by_key($vs_key, $bot_id); | |
| 7751 | - } | |
| 7752 | - } | |
| 7753 | 6402 | } |
| 7754 | 6403 | |
| 7755 | 6404 | // ============================================= |
| 7756 | 6405 | // PHASE 3: WordPress database deletions |
| @@ -7770,15 +6419,9 @@ | ||
| 7770 | 6419 | $table_name, |
| 7771 | 6420 | array('source_url' => $source_url), |
| 7772 | 6421 | array('%s') |
| 7773 | 6422 | ); |
| 7774 | - $row_url = $source_url; | |
| 7775 | 6423 | } else { |
| 7776 | - // Identity captured pre-delete for the Vector Store mirror | |
| 7777 | - $row_url = $wpdb->get_var($wpdb->prepare( | |
| 7778 | - "SELECT source_url FROM {$table_name} WHERE id = %d", | |
| 7779 | - intval($entry_id) | |
| 7780 | - )); | |
| 7781 | 6424 | wp_cache_delete('prompt_' . $entry_id, 'mxchat_prompts'); |
| 7782 | 6425 | $result = $wpdb->delete( |
| 7783 | 6426 | $table_name, |
| 7784 | 6427 | array('id' => intval($entry_id)), |
| @@ -7787,21 +6430,8 @@ | ||
| 7787 | 6430 | } |
| 7788 | 6431 | |
| 7789 | 6432 | if ($result !== false) { |
| 7790 | 6433 | $success_ids[] = $entry_id; |
| 7791 | - // Mirror to the Vector Store: refresh the entry's file when | |
| 7792 | - // sibling chunk rows survive, remove it when none do. | |
| 7793 | - if (!empty($row_url) && class_exists('MxChat_Vectorstore_Manager')) { | |
| 7794 | - $remaining = (int) $wpdb->get_var($wpdb->prepare( | |
| 7795 | - "SELECT COUNT(*) FROM {$table_name} WHERE source_url = %s", | |
| 7796 | - $row_url | |
| 7797 | - )); | |
| 7798 | - if ($remaining > 0) { | |
| 7799 | - MxChat_Vectorstore_Manager::sync_upsert_entry($row_url, '', $bot_id); | |
| 7800 | - } else { | |
| 7801 | - MxChat_Vectorstore_Manager::sync_delete_entry($row_url, $bot_id); | |
| 7802 | - } | |
| 7803 | - } | |
| 7804 | 6434 | } else { |
| 7805 | 6435 | $failed_ids[] = $entry_id; |
| 7806 | 6436 | $errors[] = "Database error for entry: $entry_id"; |
| 7807 | 6437 | } |
| @@ -7953,29 +6583,9 @@ | ||
| 7953 | 6583 | if ($result === false) { |
| 7954 | 6584 | wp_send_json_error('Database update failed: ' . $wpdb->last_error); |
| 7955 | 6585 | exit; |
| 7956 | 6586 | } |
| 7957 | - | |
| 7958 | - // Keep the OpenAI Vector Store mirror consistent with the new restriction | |
| 7959 | - // (plan 15b5c6): non-public pulls the mirrored file, public re-mirrors. | |
| 7960 | - if (class_exists('MxChat_Vectorstore_Manager')) { | |
| 7961 | - if ($data_source === 'pinecone') { | |
| 7962 | - if ($role_restriction !== 'public') { | |
| 7963 | - MxChat_Vectorstore_Manager::sync_delete_by_key($entry_id, 'default'); | |
| 7964 | - } | |
| 7965 | - // Public again: Pinecone-mode content isn't held locally, so the | |
| 7966 | - // entry re-mirrors on its next save/import rather than here. | |
| 7967 | - } else { | |
| 7968 | - $row_url = $wpdb->get_var($wpdb->prepare( | |
| 7969 | - "SELECT source_url FROM {$wpdb->prefix}mxchat_system_prompt_content WHERE id = %d", | |
| 7970 | - absint($entry_id) | |
| 7971 | - )); | |
| 7972 | - if (!empty($row_url)) { | |
| 7973 | - MxChat_Vectorstore_Manager::handle_role_change($row_url, 'default', $role_restriction); | |
| 7974 | - } | |
| 7975 | - } | |
| 7976 | - } | |
| 7977 | - | |
| 6587 | + | |
| 7978 | 6588 | wp_send_json_success(array( |
| 7979 | 6589 | 'message' => 'Role restriction updated successfully', |
| 7980 | 6590 | 'role_restriction' => $role_restriction, |
| 7981 | 6591 | 'data_source' => $data_source, |
| @@ -8018,16 +6628,16 @@ | ||
| 8018 | 6628 | wp_send_json_error('Unauthorized access'); |
| 8019 | 6629 | exit; |
| 8020 | 6630 | } |
| 8021 | 6631 | |
| 8022 | - $tag_input = isset($_POST['tag_slug']) ? sanitize_text_field($_POST['tag_slug']) : ''; | |
| 6632 | + $tag_slug = isset($_POST['tag_slug']) ? sanitize_text_field($_POST['tag_slug']) : ''; | |
| 8023 | 6633 | $role_restriction = isset($_POST['role_restriction']) ? sanitize_text_field($_POST['role_restriction']) : 'public'; |
| 8024 | - | |
| 8025 | - if (empty($tag_input)) { | |
| 8026 | - wp_send_json_error('Please enter a tag name or slug'); | |
| 6634 | + | |
| 6635 | + if (empty($tag_slug)) { | |
| 6636 | + wp_send_json_error('Tag slug is required'); | |
| 8027 | 6637 | exit; |
| 8028 | 6638 | } |
| 8029 | - | |
| 6639 | + | |
| 8030 | 6640 | // Validate role restriction |
| 8031 | 6641 | $valid_roles = array_keys($this->mxchat_get_role_options()); |
| 8032 | 6642 | if (!in_array($role_restriction, $valid_roles)) { |
| 8033 | 6643 | wp_send_json_error('Invalid role restriction'); |
| @@ -8032,27 +6642,16 @@ | ||
| 8032 | 6642 | if (!in_array($role_restriction, $valid_roles)) { |
| 8033 | 6643 | wp_send_json_error('Invalid role restriction'); |
| 8034 | 6644 | exit; |
| 8035 | 6645 | } |
| 8036 | - | |
| 8037 | - // Resolve the tag by slug first, then fall back to its display name, so users can | |
| 8038 | - // enter either "premium-content" or "Premium Content". (plan b8bcf5 — the field is | |
| 8039 | - // labeled by name but previously validated by slug only, producing the confusing | |
| 8040 | - // "Tag does not exist in WordPress" error when a real tag's name was typed.) | |
| 8041 | - $term = get_term_by('slug', $tag_input, 'post_tag'); | |
| 6646 | + | |
| 6647 | + // Check if tag exists in WordPress | |
| 6648 | + $term = get_term_by('slug', $tag_slug, 'post_tag'); | |
| 8042 | 6649 | if (!$term) { |
| 8043 | - $term = get_term_by('name', $tag_input, 'post_tag'); | |
| 8044 | - } | |
| 8045 | - if (!$term) { | |
| 8046 | - 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.'); | |
| 6650 | + wp_send_json_error('Tag does not exist in WordPress'); | |
| 8047 | 6651 | exit; |
| 8048 | 6652 | } |
| 8049 | - | |
| 8050 | - // Always key the mapping by the RESOLVED slug — apply_role_restriction_to_post() | |
| 8051 | - // compares against each post's tag slugs, so the stored key must be a slug, | |
| 8052 | - // never the raw (possibly display-name) input. | |
| 8053 | - $tag_slug = $term->slug; | |
| 8054 | - | |
| 6653 | + | |
| 8055 | 6654 | // Get existing mappings |
| 8056 | 6655 | $mappings = get_option('mxchat_tag_role_mappings', array()); |
| 8057 | 6656 | |
| 8058 | 6657 | // Check if mapping already exists |
| @@ -8327,9 +6926,9 @@ | ||
| 8327 | 6926 | ); |
| 8328 | 6927 | } else { |
| 8329 | 6928 | // Update WordPress DB |
| 8330 | 6929 | $table_name = $wpdb->prefix . 'mxchat_system_prompt_content'; |
| 8331 | - | |
| 6930 | + | |
| 8332 | 6931 | $wpdb->update( |
| 8333 | 6932 | $table_name, |
| 8334 | 6933 | array('role_restriction' => $highest_role), |
| 8335 | 6934 | array('source_url' => $source_url), |
| @@ -8336,15 +6935,8 @@ | ||
| 8336 | 6935 | array('%s'), |
| 8337 | 6936 | array('%s') |
| 8338 | 6937 | ); |
| 8339 | 6938 | } |
| 8340 | - | |
| 8341 | - // The entry's restriction just changed — keep the OpenAI Vector Store | |
| 8342 | - // mirror consistent: non-public pulls the file (file_search has no | |
| 8343 | - // per-role filtering), public re-mirrors it (plan 15b5c6). | |
| 8344 | - if (class_exists('MxChat_Vectorstore_Manager')) { | |
| 8345 | - MxChat_Vectorstore_Manager::handle_role_change($source_url, 'default', $highest_role); | |
| 8346 | - } | |
| 8347 | 6939 | } |
| 8348 | 6940 | |
| 8349 | 6941 | /** |
| 8350 | 6942 | * Apply role restriction after content is stored (for auto-sync) |
| @@ -8413,9 +7005,9 @@ | ||
| 8413 | 7005 | ); |
| 8414 | 7006 | } else { |
| 8415 | 7007 | // Update WordPress DB |
| 8416 | 7008 | $table_name = $wpdb->prefix . 'mxchat_system_prompt_content'; |
| 8417 | - | |
| 7009 | + | |
| 8418 | 7010 | $wpdb->update( |
| 8419 | 7011 | $table_name, |
| 8420 | 7012 | array('role_restriction' => $highest_role), |
| 8421 | 7013 | array('source_url' => $source_url), |
| @@ -8422,15 +7014,8 @@ | ||
| 8422 | 7014 | array('%s'), |
| 8423 | 7015 | array('%s') |
| 8424 | 7016 | ); |
| 8425 | 7017 | } |
| 8426 | - | |
| 8427 | - // The entry's restriction just changed — keep the OpenAI Vector Store | |
| 8428 | - // mirror consistent: non-public pulls the file (file_search has no | |
| 8429 | - // per-role filtering), public re-mirrors it (plan 15b5c6). | |
| 8430 | - if (class_exists('MxChat_Vectorstore_Manager')) { | |
| 8431 | - MxChat_Vectorstore_Manager::handle_role_change($source_url, 'default', $highest_role); | |
| 8432 | - } | |
| 8433 | 7018 | } |
| 8434 | 7019 | |
| 8435 | 7020 | |
| 8436 | 7021 | // ======================================== |
| @@ -8891,19 +7476,25 @@ | ||
| 8891 | 7476 | if (empty($url)) { |
| 8892 | 7477 | return new WP_Error('invalid_url', 'URL is empty'); |
| 8893 | 7478 | } |
| 8894 | 7479 | |
| 8895 | - // Get bot-specific embedding decision early (needed for both paths) — | |
| 8896 | - // custom-provider-aware (plan cbd5fd). Error code preserved. | |
| 7480 | + // Get bot-specific API key early (needed for both paths) | |
| 8897 | 7481 | $bot_options = $this->get_bot_options($bot_id); |
| 8898 | 7482 | $options = !empty($bot_options) ? $bot_options : get_option('mxchat_options'); |
| 7483 | + $selected_model = $options['embedding_model'] ?? 'text-embedding-ada-002'; | |
| 8899 | 7484 | |
| 8900 | - $preflight = MxChat_Utils::embedding_preflight($options); | |
| 8901 | - if (!$preflight['ok']) { | |
| 8902 | - return new WP_Error('no_api_key', $preflight['reason'] . ' (bot: ' . $bot_id . ')'); | |
| 7485 | + if (strpos($selected_model, 'voyage') === 0) { | |
| 7486 | + $api_key = $options['voyage_api_key'] ?? ''; | |
| 7487 | + } elseif (strpos($selected_model, 'gemini-embedding') === 0) { | |
| 7488 | + $api_key = $options['gemini_api_key'] ?? ''; | |
| 7489 | + } else { | |
| 7490 | + $api_key = $options['api_key'] ?? ''; | |
| 8903 | 7491 | } |
| 8904 | - $api_key = $preflight['api_key']; | |
| 8905 | 7492 | |
| 7493 | + if (empty($api_key)) { | |
| 7494 | + return new WP_Error('no_api_key', 'API key not configured for bot: ' . $bot_id); | |
| 7495 | + } | |
| 7496 | + | |
| 8906 | 7497 | // Check if this is a WooCommerce product URL and WooCommerce is active |
| 8907 | 7498 | $is_product_url = (strpos($url, '/product/') !== false || strpos($url, '/shop/') !== false); |
| 8908 | 7499 | $content_type = $is_product_url ? 'product' : 'url'; |
| 8909 | 7500 | |
| @@ -8930,9 +7521,9 @@ | ||
| 8930 | 7521 | $is_likely_pdf = (strtolower(pathinfo(parse_url($url, PHP_URL_PATH) ?: '', PATHINFO_EXTENSION)) === 'pdf'); |
| 8931 | 7522 | $response = wp_remote_get($url, array( |
| 8932 | 7523 | 'timeout' => $is_likely_pdf ? 120 : 30, |
| 8933 | 7524 | 'redirection' => 5, |
| 8934 | - 'user-agent' => mxchat_ingest_user_agent(), | |
| 7525 | + 'user-agent' => 'MxChat/1.0' | |
| 8935 | 7526 | )); |
| 8936 | 7527 | |
| 8937 | 7528 | if (is_wp_error($response)) { |
| 8938 | 7529 | return $response; |
| @@ -9097,9 +7688,8 @@ | ||
| 9097 | 7688 | |
| 9098 | 7689 | for ($i = 0; $i < $total_pages; $i++) { |
| 9099 | 7690 | $page_num = $i + 1; |
| 9100 | 7691 | $text = $pages[$i]->getText(); |
| 9101 | - $text = MxChat_Utils::normalize_pdf_rtl($text, 'kb_pdf_import page ' . $page_num); | |
| 9102 | 7692 | if (empty($text)) { |
| 9103 | 7693 | $skipped_pages[] = 'Page ' . $page_num . ': No text could be extracted — page may contain only images, links, or non-standard encoding'; |
| 9104 | 7694 | continue; |
| 9105 | 7695 | } |
| @@ -9186,12 +7776,95 @@ | ||
| 9186 | 7776 | if (!$product) { |
| 9187 | 7777 | return false; |
| 9188 | 7778 | } |
| 9189 | 7779 | |
| 9190 | - // Build product content via the shared WC-object assembler (a3d60c) — same | |
| 9191 | - // body as the auto-sync product writer, so the two paths can never drift. | |
| 9192 | - $content = $this->mxchat_prepare_product_content_for_indexing($product); | |
| 7780 | + // Build product content with pricing (similar to mxchat_store_product_embedding) | |
| 7781 | + $title = $product->get_name(); | |
| 7782 | + $description = $product->get_description(); | |
| 7783 | + $short_description = $product->get_short_description(); | |
| 7784 | + $sku = $product->get_sku(); | |
| 9193 | 7785 | |
| 7786 | + // Get pricing information | |
| 7787 | + $regular_price = $product->get_regular_price(); | |
| 7788 | + $sale_price = $product->get_sale_price(); | |
| 7789 | + $price = $product->get_price(); // Current active price | |
| 7790 | + | |
| 7791 | + // Get currency symbol | |
| 7792 | + $currency_symbol = get_woocommerce_currency_symbol(); | |
| 7793 | + | |
| 7794 | + // Format content | |
| 7795 | + $content = $title . "\n\n"; | |
| 7796 | + | |
| 7797 | + if (!empty($short_description)) { | |
| 7798 | + $content .= "Short Description: " . wp_strip_all_tags($short_description) . "\n\n"; | |
| 7799 | + } | |
| 7800 | + | |
| 7801 | + if (!empty($description)) { | |
| 7802 | + $content .= wp_strip_all_tags($description) . "\n\n"; | |
| 7803 | + } | |
| 7804 | + | |
| 7805 | + // Add pricing information | |
| 7806 | + if (!empty($regular_price)) { | |
| 7807 | + $content .= "Price: " . $currency_symbol . $regular_price . "\n"; | |
| 7808 | + } elseif (!empty($price)) { | |
| 7809 | + $content .= "Price: " . $currency_symbol . $price . "\n"; | |
| 7810 | + } | |
| 7811 | + | |
| 7812 | + if (!empty($sale_price) && $sale_price !== $regular_price) { | |
| 7813 | + $content .= "Sale Price: " . $currency_symbol . $sale_price . "\n"; | |
| 7814 | + } | |
| 7815 | + | |
| 7816 | + // Handle variable products - show price range | |
| 7817 | + if ($product->is_type('variable')) { | |
| 7818 | + $min_price = $product->get_variation_price('min'); | |
| 7819 | + $max_price = $product->get_variation_price('max'); | |
| 7820 | + if ($min_price !== $max_price) { | |
| 7821 | + $content .= "Price Range: " . $currency_symbol . $min_price . " - " . $currency_symbol . $max_price . "\n"; | |
| 7822 | + } | |
| 7823 | + } | |
| 7824 | + | |
| 7825 | + if (!empty($sku)) { | |
| 7826 | + $content .= "SKU: " . $sku . "\n"; | |
| 7827 | + } | |
| 7828 | + | |
| 7829 | + // Get product categories | |
| 7830 | + $categories = wp_get_post_terms($product_id, 'product_cat', array('fields' => 'names')); | |
| 7831 | + if (!empty($categories) && !is_wp_error($categories)) { | |
| 7832 | + $content .= "Categories: " . implode(', ', $categories) . "\n"; | |
| 7833 | + } | |
| 7834 | + | |
| 7835 | + // Get Custom Product Tabs (supports "Custom Product Tabs for WooCommerce" by Code Parrots) | |
| 7836 | + $custom_tabs = get_post_meta($product_id, 'yikes_woo_products_tabs', true); | |
| 7837 | + if (!empty($custom_tabs) && is_array($custom_tabs)) { | |
| 7838 | + foreach ($custom_tabs as $tab) { | |
| 7839 | + $tab_title = isset($tab['title']) ? $tab['title'] : (isset($tab['tab_title']) ? $tab['tab_title'] : ''); | |
| 7840 | + $tab_content = isset($tab['content']) ? $tab['content'] : ''; | |
| 7841 | + | |
| 7842 | + if (!empty($tab_title) && !empty($tab_content)) { | |
| 7843 | + $content .= "\n" . $tab_title . ": " . wp_strip_all_tags($tab_content) . "\n"; | |
| 7844 | + } | |
| 7845 | + } | |
| 7846 | + } | |
| 7847 | + | |
| 7848 | + // Also check for reusable/saved tabs applied to this product | |
| 7849 | + $applied_saved_tabs = get_post_meta($product_id, 'yikes_woo_reusable_products_tabs_applied', true); | |
| 7850 | + if (!empty($applied_saved_tabs) && is_array($applied_saved_tabs)) { | |
| 7851 | + $saved_tabs = get_option('yikes_woo_reusable_products_tabs', array()); | |
| 7852 | + if (!empty($saved_tabs) && is_array($saved_tabs)) { | |
| 7853 | + foreach ($applied_saved_tabs as $saved_tab_id) { | |
| 7854 | + if (isset($saved_tabs[$saved_tab_id])) { | |
| 7855 | + $tab = $saved_tabs[$saved_tab_id]; | |
| 7856 | + $tab_title = isset($tab['title']) ? $tab['title'] : (isset($tab['tab_title']) ? $tab['tab_title'] : ''); | |
| 7857 | + $tab_content = isset($tab['content']) ? $tab['content'] : ''; | |
| 7858 | + | |
| 7859 | + if (!empty($tab_title) && !empty($tab_content)) { | |
| 7860 | + $content .= "\n" . $tab_title . ": " . wp_strip_all_tags($tab_content) . "\n"; | |
| 7861 | + } | |
| 7862 | + } | |
| 7863 | + } | |
| 7864 | + } | |
| 7865 | + } | |
| 7866 | + | |
| 9194 | 7867 | return $this->mxchat_sanitize_content_for_api($content); |
| 9195 | 7868 | } |
| 9196 | 7869 | |
| 9197 | 7870 | /** |
| @@ -9221,9 +7894,8 @@ | ||
| 9221 | 7894 | return new WP_Error('page_not_found', 'Page ' . $page_number . ' not found in PDF'); |
| 9222 | 7895 | } |
| 9223 | 7896 | |
| 9224 | 7897 | $text = $pages[$page_number - 1]->getText(); |
| 9225 | - $text = MxChat_Utils::normalize_pdf_rtl($text, 'kb_pdf_page page ' . $page_number); | |
| 9226 | 7898 | |
| 9227 | 7899 | if (empty($text)) { |
| 9228 | 7900 | return new WP_Error('empty_page', 'Page ' . $page_number . ': No text could be extracted — page may contain only images, links, or non-standard encoding'); |
| 9229 | 7901 | } |
| @@ -9244,18 +7916,24 @@ | ||
| 9244 | 7916 | |
| 9245 | 7917 | $content_with_metadata = wp_json_encode($metadata) . "\n---\n" . $sanitized; |
| 9246 | 7918 | $page_url = esc_url($pdf_url . "#page=" . $page_number); |
| 9247 | 7919 | |
| 9248 | - // Get bot-specific embedding decision — custom-provider-aware | |
| 9249 | - // (plan cbd5fd). Error code preserved. | |
| 7920 | + // Get bot-specific API key | |
| 9250 | 7921 | $bot_options = $this->get_bot_options($bot_id); |
| 9251 | 7922 | $options = !empty($bot_options) ? $bot_options : get_option('mxchat_options'); |
| 9252 | - | |
| 9253 | - $preflight = MxChat_Utils::embedding_preflight($options); | |
| 9254 | - if (!$preflight['ok']) { | |
| 9255 | - return new WP_Error('no_api_key', $preflight['reason'] . ' (bot: ' . $bot_id . ')'); | |
| 7923 | + $selected_model = $options['embedding_model'] ?? 'text-embedding-ada-002'; | |
| 7924 | + | |
| 7925 | + if (strpos($selected_model, 'voyage') === 0) { | |
| 7926 | + $api_key = $options['voyage_api_key'] ?? ''; | |
| 7927 | + } elseif (strpos($selected_model, 'gemini-embedding') === 0) { | |
| 7928 | + $api_key = $options['gemini_api_key'] ?? ''; | |
| 7929 | + } else { | |
| 7930 | + $api_key = $options['api_key'] ?? ''; | |
| 9256 | 7931 | } |
| 9257 | - $api_key = $preflight['api_key']; | |
| 7932 | + | |
| 7933 | + if (empty($api_key)) { | |
| 7934 | + return new WP_Error('no_api_key', 'API key not configured for bot: ' . $bot_id); | |
| 7935 | + } | |
| 9258 | 7936 | |
| 9259 | 7937 | // Submit to database - UPDATED 2.5.6: Added content_type 'pdf' |
| 9260 | 7938 | $result = MxChat_Utils::submit_content_to_db( |
| 9261 | 7939 | $content_with_metadata, |
