| @@ -1,1862 +1,1998 @@ | ||
| 1 | -<?php | |
| 2 | -/** | |
| 3 | - * File: admin/class-pinecone-manager.php | |
| 4 | - * | |
| 5 | - * Handles all Pinecone vector database operations for MxChat | |
| 6 | - */ | |
| 7 | - | |
| 8 | -if (!defined('ABSPATH')) { | |
| 9 | - exit; // Exit if accessed directly | |
| 10 | -} | |
| 11 | - | |
| 12 | -class MxChat_Pinecone_Manager { | |
| 13 | - | |
| 14 | - /** | |
| 15 | - * Constructor | |
| 16 | - */ | |
| 17 | - public function __construct() { | |
| 18 | - } | |
| 19 | - | |
| 20 | - // ======================================== | |
| 21 | - // PINECONE FETCH OPERATIONS | |
| 22 | - // ======================================== | |
| 23 | - | |
| 24 | - | |
| 25 | -/** | |
| 26 | - * Fetches records from Pinecone with bot-specific filtering | |
| 27 | - * UPDATED 2.6.1: Optimized for large datasets - uses server-side pagination | |
| 28 | - */ | |
| 29 | -public function mxchat_fetch_pinecone_records($pinecone_options, $search_query = '', $page = 1, $per_page = 20, $bot_id = 'default', $content_type = '') { | |
| 30 | - $api_key = $pinecone_options['mxchat_pinecone_api_key'] ?? ''; | |
| 31 | - $host = $pinecone_options['mxchat_pinecone_host'] ?? ''; | |
| 32 | - $namespace = $pinecone_options['mxchat_pinecone_namespace'] ?? ''; | |
| 33 | - | |
| 34 | - if (empty($api_key) || empty($host)) { | |
| 35 | - return array('data' => array(), 'total' => 0, 'total_in_database' => 0, 'showing_recent_only' => false); | |
| 36 | - } | |
| 37 | - | |
| 38 | - try { | |
| 39 | - // Get total count for the banner message (bot-specific) - lightweight call | |
| 40 | - $total_in_database = $this->mxchat_get_pinecone_total_count($pinecone_options, $bot_id); | |
| 41 | - | |
| 42 | - // For large databases, use optimized paginated fetching | |
| 43 | - // Only fetch what we need for the current page, not all 1K records | |
| 44 | - if ($total_in_database > 500) { | |
| 45 | - $result = $this->mxchat_fetch_pinecone_page_optimized($pinecone_options, $search_query, $page, $per_page, $bot_id, $content_type); | |
| 46 | - $result['total_in_database'] = $total_in_database; | |
| 47 | - $result['showing_recent_only'] = true; // Always show banner when we're limiting results | |
| 48 | - return $result; | |
| 49 | - } | |
| 50 | - | |
| 51 | - // For smaller databases, use the existing approach but with safety limits | |
| 52 | - $all_records = $this->mxchat_get_recent_entries_safe($pinecone_options, $bot_id, 500); | |
| 53 | - | |
| 54 | - // Filter by content type if provided | |
| 55 | - if (!empty($content_type)) { | |
| 56 | - $all_records = array_filter($all_records, function($record) use ($content_type) { | |
| 57 | - $record_type = $record->type ?? 'content'; | |
| 58 | - return $record_type === $content_type; | |
| 59 | - }); | |
| 60 | - $all_records = array_values($all_records); // Re-index array | |
| 61 | - } | |
| 62 | - | |
| 63 | - // Filter by search query if provided | |
| 64 | - if (!empty($search_query)) { | |
| 65 | - $all_records = array_filter($all_records, function($record) use ($search_query) { | |
| 66 | - $content = $record->article_content ?? ''; | |
| 67 | - $source_url = $record->source_url ?? ''; | |
| 68 | - return stripos($content, $search_query) !== false || stripos($source_url, $search_query) !== false; | |
| 69 | - }); | |
| 70 | - $all_records = array_values($all_records); // Re-index array | |
| 71 | - } | |
| 72 | - | |
| 73 | - // UPDATED 2.6.3: Group records by source_url for chunk-aware pagination | |
| 74 | - // This ensures pagination shows X entries per page, not X chunks | |
| 75 | - $grouped_by_url = array(); | |
| 76 | - $empty_url_records = array(); | |
| 77 | - | |
| 78 | - foreach ($all_records as $record) { | |
| 79 | - $source_url = $record->source_url ?? ''; | |
| 80 | - if (empty($source_url)) { | |
| 81 | - $empty_url_records[] = $record; | |
| 82 | - } else { | |
| 83 | - if (!isset($grouped_by_url[$source_url])) { | |
| 84 | - $grouped_by_url[$source_url] = array(); | |
| 85 | - } | |
| 86 | - $grouped_by_url[$source_url][] = $record; | |
| 87 | - } | |
| 88 | - } | |
| 89 | - | |
| 90 | - // Count unique entries (unique URLs + individual empty-URL records) | |
| 91 | - $total_unique_entries = count($grouped_by_url) + count($empty_url_records); | |
| 92 | - | |
| 93 | - // Paginate by unique entries | |
| 94 | - $offset = ($page - 1) * $per_page; | |
| 95 | - | |
| 96 | - // Build ordered list of URL groups (newest first based on first record) | |
| 97 | - $url_groups_ordered = array_keys($grouped_by_url); | |
| 98 | - | |
| 99 | - // Get the URLs for this page | |
| 100 | - $page_urls = array_slice($url_groups_ordered, $offset, $per_page); | |
| 101 | - | |
| 102 | - // Collect all records for this page's URLs | |
| 103 | - $paged_records = array(); | |
| 104 | - foreach ($page_urls as $url) { | |
| 105 | - foreach ($grouped_by_url[$url] as $record) { | |
| 106 | - $paged_records[] = $record; | |
| 107 | - } | |
| 108 | - } | |
| 109 | - | |
| 110 | - // Add empty-URL records if they fall within this page's range | |
| 111 | - $remaining_slots = $per_page - count($page_urls); | |
| 112 | - $empty_offset = max(0, $offset - count($grouped_by_url)); | |
| 113 | - if ($remaining_slots > 0 && $empty_offset < count($empty_url_records)) { | |
| 114 | - $empty_page_records = array_slice($empty_url_records, $empty_offset, $remaining_slots); | |
| 115 | - $paged_records = array_merge($paged_records, $empty_page_records); | |
| 116 | - } | |
| 117 | - | |
| 118 | - return array( | |
| 119 | - 'data' => $paged_records, | |
| 120 | - 'total' => $total_unique_entries, | |
| 121 | - 'total_in_database' => $total_in_database, | |
| 122 | - 'showing_recent_only' => ($total_in_database > 500) | |
| 123 | - ); | |
| 124 | - | |
| 125 | - } catch (Exception $e) { | |
| 126 | - //error_log('MxChat Pinecone fetch error: ' . $e->getMessage()); | |
| 127 | - MxChat_Admin::mxchat_log_debug('pinecone_error', 'Pinecone fetch error: ' . $e->getMessage()); | |
| 128 | - return array('data' => array(), 'total' => 0, 'total_in_database' => 0, 'showing_recent_only' => false); | |
| 129 | - } | |
| 130 | -} | |
| 131 | - | |
| 132 | -/** | |
| 133 | - * Optimized fetch for large Pinecone databases | |
| 134 | - * ADDED 2.6.1: Prevents crashes with large datasets | |
| 135 | - * UPDATED 2.6.1: When searching, uses semantic search with embedded query for accurate results across all 13K+ records | |
| 136 | - */ | |
| 137 | -private function mxchat_fetch_pinecone_page_optimized($pinecone_options, $search_query = '', $page = 1, $per_page = 20, $bot_id = 'default', $content_type = '') { | |
| 138 | - $api_key = $pinecone_options['mxchat_pinecone_api_key'] ?? ''; | |
| 139 | - $host = $pinecone_options['mxchat_pinecone_host'] ?? ''; | |
| 140 | - $namespace = $pinecone_options['mxchat_pinecone_namespace'] ?? ''; | |
| 141 | - | |
| 142 | - try { | |
| 143 | - // If user is searching, use semantic search with embedded query | |
| 144 | - // This searches ALL records in Pinecone, not just fetched ones | |
| 145 | - if (!empty($search_query)) { | |
| 146 | - return $this->mxchat_semantic_search_pinecone($pinecone_options, $search_query, $page, $per_page, $bot_id, $content_type); | |
| 147 | - } | |
| 148 | - | |
| 149 | - // For browsing (no search), use Pinecone's list endpoint for true pagination | |
| 150 | - return $this->mxchat_list_pinecone_records($pinecone_options, $page, $per_page, $bot_id, $content_type); | |
| 151 | - | |
| 152 | - } catch (Exception $e) { | |
| 153 | - //error_log('MxChat optimized fetch exception: ' . $e->getMessage()); | |
| 154 | - MxChat_Admin::mxchat_log_debug('pinecone_error', 'Pinecone optimized fetch error: ' . $e->getMessage()); | |
| 155 | - return array('data' => array(), 'total' => 0); | |
| 156 | - } | |
| 157 | -} | |
| 158 | - | |
| 159 | -/** | |
| 160 | - * Semantic search across ALL Pinecone records using embedded search query | |
| 161 | - * This allows users to find any of their 13K+ products by searching | |
| 162 | - * ADDED 2.6.1 | |
| 163 | - */ | |
| 164 | -private function mxchat_semantic_search_pinecone($pinecone_options, $search_query, $page = 1, $per_page = 20, $bot_id = 'default', $content_type = '') { | |
| 165 | - $api_key = $pinecone_options['mxchat_pinecone_api_key'] ?? ''; | |
| 166 | - $host = $pinecone_options['mxchat_pinecone_host'] ?? ''; | |
| 167 | - $namespace = $pinecone_options['mxchat_pinecone_namespace'] ?? ''; | |
| 168 | - | |
| 169 | - // Get embedding for the search query | |
| 170 | - $query_embedding = $this->mxchat_get_search_embedding($search_query); | |
| 171 | - | |
| 172 | - if (empty($query_embedding)) { | |
| 173 | - // Fallback to text-based search if embedding fails | |
| 174 | - return $this->mxchat_text_search_fallback($pinecone_options, $search_query, $page, $per_page, $bot_id, $content_type); | |
| 175 | - } | |
| 176 | - | |
| 177 | - $query_url = "https://{$host}/query"; | |
| 178 | - | |
| 179 | - // Fetch more results to allow for filtering and pagination | |
| 180 | - $fetch_limit = min(($page * $per_page) + 100, 500); | |
| 181 | - | |
| 182 | - $query_data = array( | |
| 183 | - 'includeMetadata' => true, | |
| 184 | - 'includeValues' => false, | |
| 185 | - 'topK' => $fetch_limit, | |
| 186 | - 'vector' => $query_embedding | |
| 187 | - ); | |
| 188 | - | |
| 189 | - if (!empty($namespace)) { | |
| 190 | - $query_data['namespace'] = $namespace; | |
| 191 | - } | |
| 192 | - | |
| 193 | - // Add metadata filter for content type if specified | |
| 194 | - if (!empty($content_type)) { | |
| 195 | - $query_data['filter'] = array( | |
| 196 | - 'type' => array('$eq' => $content_type) | |
| 197 | - ); | |
| 198 | - } | |
| 199 | - | |
| 200 | - $response = wp_remote_post($query_url, array( | |
| 201 | - 'headers' => array( | |
| 202 | - 'Api-Key' => $api_key, | |
| 203 | - 'Content-Type' => 'application/json' | |
| 204 | - ), | |
| 205 | - 'body' => json_encode($query_data), | |
| 206 | - 'timeout' => 20 | |
| 207 | - )); | |
| 208 | - | |
| 209 | - if (is_wp_error($response) || wp_remote_retrieve_response_code($response) !== 200) { | |
| 210 | - return $this->mxchat_text_search_fallback($pinecone_options, $search_query, $page, $per_page, $bot_id, $content_type); | |
| 211 | - } | |
| 212 | - | |
| 213 | - $body = wp_remote_retrieve_body($response); | |
| 214 | - $data = json_decode($body, true); | |
| 215 | - | |
| 216 | - $records = array(); | |
| 217 | - if (isset($data['matches'])) { | |
| 218 | - foreach ($data['matches'] as $match) { | |
| 219 | - $metadata = $match['metadata'] ?? array(); | |
| 220 | - $created_at = $metadata['created_at'] ?? $metadata['last_updated'] ?? $metadata['timestamp'] ?? time(); | |
| 221 | - if (!is_numeric($created_at)) { | |
| 222 | - $created_at = strtotime($created_at) ?: time(); | |
| 223 | - } | |
| 224 | - | |
| 225 | - $records[] = (object) array( | |
| 226 | - 'id' => $match['id'] ?? '', | |
| 227 | - 'article_content' => $metadata['text'] ?? '', | |
| 228 | - 'source_url' => $metadata['source_url'] ?? '', | |
| 229 | - 'role_restriction' => $metadata['role_restriction'] ?? 'public', | |
| 230 | - 'type' => $metadata['type'] ?? 'content', | |
| 231 | - 'bot_id' => $bot_id, | |
| 232 | - 'created_at' => $created_at, | |
| 233 | - 'data_source' => 'pinecone', | |
| 234 | - 'relevance_score' => $match['score'] ?? 0, | |
| 235 | - 'chunk_index' => isset($metadata['chunk_index']) ? intval($metadata['chunk_index']) : null, | |
| 236 | - 'total_chunks' => isset($metadata['total_chunks']) ? intval($metadata['total_chunks']) : null, | |
| 237 | - 'is_chunked' => isset($metadata['is_chunked']) ? (bool) $metadata['is_chunked'] : false | |
| 238 | - ); | |
| 239 | - } | |
| 240 | - } | |
| 241 | - | |
| 242 | - // For semantic search, results are already sorted by relevance (score) | |
| 243 | - // No need to re-sort by date | |
| 244 | - | |
| 245 | - $total = count($records); | |
| 246 | - $offset = ($page - 1) * $per_page; | |
| 247 | - $paged_records = array_slice($records, $offset, $per_page); | |
| 248 | - | |
| 249 | - $this->mxchat_batch_fetch_role_restrictions($paged_records, $bot_id); | |
| 250 | - | |
| 251 | - return array( | |
| 252 | - 'data' => $paged_records, | |
| 253 | - 'total' => $total | |
| 254 | - ); | |
| 255 | -} | |
| 256 | - | |
| 257 | -/** | |
| 258 | - * Get embedding vector for a search query | |
| 259 | - * Uses the same embedding model configured for the knowledge base | |
| 260 | - * ADDED 2.6.1 | |
| 261 | - */ | |
| 262 | -private function mxchat_get_search_embedding($search_query) { | |
| 263 | - $options = get_option('mxchat_options', array()); | |
| 264 | - $embedding_model = $options['embedding_model'] ?? 'text-embedding-ada-002'; | |
| 265 | - | |
| 266 | - // Determine which API to use based on model | |
| 267 | - if (strpos($embedding_model, 'voyage-') === 0) { | |
| 268 | - return $this->mxchat_get_voyage_embedding($search_query, $options, $embedding_model); | |
| 269 | - } elseif (strpos($embedding_model, 'gemini-') === 0) { | |
| 270 | - return $this->mxchat_get_gemini_embedding($search_query, $options, $embedding_model); | |
| 271 | - } else { | |
| 272 | - return $this->mxchat_get_openai_embedding($search_query, $options, $embedding_model); | |
| 273 | - } | |
| 274 | -} | |
| 275 | - | |
| 276 | -/** | |
| 277 | - * Get OpenAI embedding for search query | |
| 278 | - */ | |
| 279 | -private function mxchat_get_openai_embedding($text, $options, $model) { | |
| 280 | - $api_key = $options['api_key'] ?? ''; | |
| 281 | - if (empty($api_key)) { | |
| 282 | - return null; | |
| 283 | - } | |
| 284 | - | |
| 285 | - $response = wp_remote_post('https://api.openai.com/v1/embeddings', array( | |
| 286 | - 'headers' => array( | |
| 287 | - 'Authorization' => 'Bearer ' . $api_key, | |
| 288 | - 'Content-Type' => 'application/json' | |
| 289 | - ), | |
| 290 | - 'body' => json_encode(array( | |
| 291 | - 'model' => $model, | |
| 292 | - 'input' => $text | |
| 293 | - )), | |
| 294 | - 'timeout' => 15 | |
| 295 | - )); | |
| 296 | - | |
| 297 | - if (is_wp_error($response) || wp_remote_retrieve_response_code($response) !== 200) { | |
| 298 | - return null; | |
| 299 | - } | |
| 300 | - | |
| 301 | - $body = json_decode(wp_remote_retrieve_body($response), true); | |
| 302 | - return $body['data'][0]['embedding'] ?? null; | |
| 303 | -} | |
| 304 | - | |
| 305 | -/** | |
| 306 | - * Get Voyage AI embedding for search query | |
| 307 | - */ | |
| 308 | -private function mxchat_get_voyage_embedding($text, $options, $model) { | |
| 309 | - $api_key = $options['voyage_api_key'] ?? ''; | |
| 310 | - if (empty($api_key)) { | |
| 311 | - return null; | |
| 312 | - } | |
| 313 | - | |
| 314 | - $request_body = array( | |
| 315 | - 'model' => $model, | |
| 316 | - 'input' => $text, | |
| 317 | - 'input_type' => 'query' | |
| 318 | - ); | |
| 319 | - | |
| 320 | - // Add output dimensions for voyage-3-large if configured | |
| 321 | - if (strpos($model, 'voyage-3-large') === 0 && !empty($options['voyage_output_dimension'])) { | |
| 322 | - $request_body['output_dimension'] = intval($options['voyage_output_dimension']); | |
| 323 | - } | |
| 324 | - | |
| 325 | - $response = wp_remote_post('https://api.voyageai.com/v1/embeddings', array( | |
| 326 | - 'headers' => array( | |
| 327 | - 'Authorization' => 'Bearer ' . $api_key, | |
| 328 | - 'Content-Type' => 'application/json' | |
| 329 | - ), | |
| 330 | - 'body' => json_encode($request_body), | |
| 331 | - 'timeout' => 15 | |
| 332 | - )); | |
| 333 | - | |
| 334 | - if (is_wp_error($response) || wp_remote_retrieve_response_code($response) !== 200) { | |
| 335 | - return null; | |
| 336 | - } | |
| 337 | - | |
| 338 | - $body = json_decode(wp_remote_retrieve_body($response), true); | |
| 339 | - return $body['data'][0]['embedding'] ?? null; | |
| 340 | -} | |
| 341 | - | |
| 342 | -/** | |
| 343 | - * Get Google Gemini embedding for search query | |
| 344 | - */ | |
| 345 | -private function mxchat_get_gemini_embedding($text, $options, $model) { | |
| 346 | - $api_key = $options['gemini_api_key'] ?? ''; | |
| 347 | - if (empty($api_key)) { | |
| 348 | - return null; | |
| 349 | - } | |
| 350 | - | |
| 351 | - $url = "https://generativelanguage.googleapis.com/v1beta/models/{$model}:embedContent?key={$api_key}"; | |
| 352 | - | |
| 353 | - $request_body = array( | |
| 354 | - 'model' => "models/{$model}", | |
| 355 | - 'content' => array( | |
| 356 | - 'parts' => array( | |
| 357 | - array('text' => $text) | |
| 358 | - ) | |
| 359 | - ), | |
| 360 | - 'taskType' => 'RETRIEVAL_QUERY' | |
| 361 | - ); | |
| 362 | - | |
| 363 | - // Add output dimensions if configured | |
| 364 | - if (!empty($options['gemini_output_dimension'])) { | |
| 365 | - $request_body['outputDimensionality'] = intval($options['gemini_output_dimension']); | |
| 366 | - } | |
| 367 | - | |
| 368 | - $response = wp_remote_post($url, array( | |
| 369 | - 'headers' => array( | |
| 370 | - 'Content-Type' => 'application/json' | |
| 371 | - ), | |
| 372 | - 'body' => json_encode($request_body), | |
| 373 | - 'timeout' => 15 | |
| 374 | - )); | |
| 375 | - | |
| 376 | - if (is_wp_error($response) || wp_remote_retrieve_response_code($response) !== 200) { | |
| 377 | - return null; | |
| 378 | - } | |
| 379 | - | |
| 380 | - $body = json_decode(wp_remote_retrieve_body($response), true); | |
| 381 | - return $body['embedding']['values'] ?? null; | |
| 382 | -} | |
| 383 | - | |
| 384 | -/** | |
| 385 | - * Fallback text search when embedding fails | |
| 386 | - * Fetches more records and filters by text match | |
| 387 | - * ADDED 2.6.1 | |
| 388 | - */ | |
| 389 | -private function mxchat_text_search_fallback($pinecone_options, $search_query, $page, $per_page, $bot_id, $content_type) { | |
| 390 | - $api_key = $pinecone_options['mxchat_pinecone_api_key'] ?? ''; | |
| 391 | - $host = $pinecone_options['mxchat_pinecone_host'] ?? ''; | |
| 392 | - $namespace = $pinecone_options['mxchat_pinecone_namespace'] ?? ''; | |
| 393 | - | |
| 394 | - $query_url = "https://{$host}/query"; | |
| 395 | - $query_vector = $this->mxchat_generate_optimized_query_vector(); | |
| 396 | - | |
| 397 | - // Fetch more records to search through | |
| 398 | - $query_data = array( | |
| 399 | - 'includeMetadata' => true, | |
| 400 | - 'includeValues' => false, | |
| 401 | - 'topK' => 1000, // Fetch more for text search | |
| 402 | - 'vector' => $query_vector | |
| 403 | - ); | |
| 404 | - | |
| 405 | - if (!empty($namespace)) { | |
| 406 | - $query_data['namespace'] = $namespace; | |
| 407 | - } | |
| 408 | - | |
| 409 | - $response = wp_remote_post($query_url, array( | |
| 410 | - 'headers' => array( | |
| 411 | - 'Api-Key' => $api_key, | |
| 412 | - 'Content-Type' => 'application/json' | |
| 413 | - ), | |
| 414 | - 'body' => json_encode($query_data), | |
| 415 | - 'timeout' => 20 | |
| 416 | - )); | |
| 417 | - | |
| 418 | - if (is_wp_error($response) || wp_remote_retrieve_response_code($response) !== 200) { | |
| 419 | - return array('data' => array(), 'total' => 0); | |
| 420 | - } | |
| 421 | - | |
| 422 | - $body = wp_remote_retrieve_body($response); | |
| 423 | - $data = json_decode($body, true); | |
| 424 | - | |
| 425 | - $records = array(); | |
| 426 | - $search_lower = strtolower($search_query); | |
| 427 | - | |
| 428 | - if (isset($data['matches'])) { | |
| 429 | - foreach ($data['matches'] as $match) { | |
| 430 | - $metadata = $match['metadata'] ?? array(); | |
| 431 | - | |
| 432 | - // Filter by content type | |
| 433 | - if (!empty($content_type)) { | |
| 434 | - $record_type = $metadata['type'] ?? 'content'; | |
| 435 | - if ($record_type !== $content_type) { | |
| 436 | - continue; | |
| 437 | - } | |
| 438 | - } | |
| 439 | - | |
| 440 | - // Text search filter | |
| 441 | - $content = strtolower($metadata['text'] ?? ''); | |
| 442 | - $source_url = strtolower($metadata['source_url'] ?? ''); | |
| 443 | - if (strpos($content, $search_lower) === false && strpos($source_url, $search_lower) === false) { | |
| 444 | - continue; | |
| 445 | - } | |
| 446 | - | |
| 447 | - $created_at = $metadata['created_at'] ?? $metadata['last_updated'] ?? $metadata['timestamp'] ?? time(); | |
| 448 | - if (!is_numeric($created_at)) { | |
| 449 | - $created_at = strtotime($created_at) ?: time(); | |
| 450 | - } | |
| 451 | - | |
| 452 | - $records[] = (object) array( | |
| 453 | - 'id' => $match['id'] ?? '', | |
| 454 | - 'article_content' => $metadata['text'] ?? '', | |
| 455 | - 'source_url' => $metadata['source_url'] ?? '', | |
| 456 | - 'role_restriction' => $metadata['role_restriction'] ?? 'public', | |
| 457 | - 'type' => $metadata['type'] ?? 'content', | |
| 458 | - 'bot_id' => $bot_id, | |
| 459 | - 'created_at' => $created_at, | |
| 460 | - 'data_source' => 'pinecone', | |
| 461 | - 'chunk_index' => isset($metadata['chunk_index']) ? intval($metadata['chunk_index']) : null, | |
| 462 | - 'total_chunks' => isset($metadata['total_chunks']) ? intval($metadata['total_chunks']) : null, | |
| 463 | - 'is_chunked' => isset($metadata['is_chunked']) ? (bool) $metadata['is_chunked'] : false | |
| 464 | - ); | |
| 465 | - } | |
| 466 | - } | |
| 467 | - | |
| 468 | - // Sort by date for text search results | |
| 469 | - usort($records, function($a, $b) { | |
| 470 | - return $b->created_at - $a->created_at; | |
| 471 | - }); | |
| 472 | - | |
| 473 | - $total = count($records); | |
| 474 | - $offset = ($page - 1) * $per_page; | |
| 475 | - $paged_records = array_slice($records, $offset, $per_page); | |
| 476 | - | |
| 477 | - $this->mxchat_batch_fetch_role_restrictions($paged_records, $bot_id); | |
| 478 | - | |
| 479 | - return array( | |
| 480 | - 'data' => $paged_records, | |
| 481 | - 'total' => $total | |
| 482 | - ); | |
| 483 | -} | |
| 484 | - | |
| 485 | -/** | |
| 486 | - * List Pinecone records using the list endpoint for true pagination (no search) | |
| 487 | - * This allows browsing through all 13K+ records page by page | |
| 488 | - * ADDED 2.6.1 | |
| 489 | - */ | |
| 490 | -private function mxchat_list_pinecone_records($pinecone_options, $page = 1, $per_page = 20, $bot_id = 'default', $content_type = '') { | |
| 491 | - $api_key = $pinecone_options['mxchat_pinecone_api_key'] ?? ''; | |
| 492 | - $host = $pinecone_options['mxchat_pinecone_host'] ?? ''; | |
| 493 | - $namespace = $pinecone_options['mxchat_pinecone_namespace'] ?? ''; | |
| 494 | - | |
| 495 | - // Pinecone's list endpoint returns vector IDs with pagination | |
| 496 | - // We then fetch the metadata for those specific IDs | |
| 497 | - $list_url = "https://{$host}/vectors/list"; | |
| 498 | - | |
| 499 | - // Calculate pagination token from page number | |
| 500 | - // Pinecone uses cursor-based pagination, so we need to handle this differently | |
| 501 | - $limit = $per_page * 2; // Fetch extra to account for filtering | |
| 502 | - | |
| 503 | - $list_params = array( | |
| 504 | - 'limit' => $limit | |
| 505 | - ); | |
| 506 | - | |
| 507 | - if (!empty($namespace)) { | |
| 508 | - $list_params['namespace'] = $namespace; | |
| 509 | - } | |
| 510 | - | |
| 511 | - // For pages beyond first, we need to use pagination_token | |
| 512 | - // Store/retrieve pagination tokens in transients | |
| 513 | - $pagination_key = 'mxchat_pinecone_page_' . md5($host . $namespace . $content_type); | |
| 514 | - | |
| 515 | - if ($page > 1) { | |
| 516 | - $stored_tokens = get_transient($pagination_key); | |
| 517 | - if ($stored_tokens && isset($stored_tokens[$page])) { | |
| 518 | - $list_params['paginationToken'] = $stored_tokens[$page]; | |
| 519 | - } | |
| 520 | - } | |
| 521 | - | |
| 522 | - $response = wp_remote_post($list_url, array( | |
| 523 | - 'headers' => array( | |
| 524 | - 'Api-Key' => $api_key, | |
| 525 | - 'Content-Type' => 'application/json' | |
| 526 | - ), | |
| 527 | - 'body' => json_encode($list_params), | |
| 528 | - 'timeout' => 15 | |
| 529 | - )); | |
| 530 | - | |
| 531 | - if (is_wp_error($response) || wp_remote_retrieve_response_code($response) !== 200) { | |
| 532 | - // Fallback to query-based approach | |
| 533 | - return $this->mxchat_query_based_list($pinecone_options, $page, $per_page, $bot_id, $content_type); | |
| 534 | - } | |
| 535 | - | |
| 536 | - $body = wp_remote_retrieve_body($response); | |
| 537 | - $data = json_decode($body, true); | |
| 538 | - | |
| 539 | - // Store pagination token for next page | |
| 540 | - if (!empty($data['pagination']['next'])) { | |
| 541 | - $stored_tokens = get_transient($pagination_key) ?: array(); | |
| 542 | - $stored_tokens[$page + 1] = $data['pagination']['next']; | |
| 543 | - set_transient($pagination_key, $stored_tokens, 300); // 5 minute cache | |
| 544 | - } | |
| 545 | - | |
| 546 | - $vector_ids = array(); | |
| 547 | - if (isset($data['vectors'])) { | |
| 548 | - foreach ($data['vectors'] as $vector) { | |
| 549 | - $vector_ids[] = $vector['id']; | |
| 550 | - } | |
| 551 | - } | |
| 552 | - | |
| 553 | - // If list endpoint returned empty, fall back to query-based approach | |
| 554 | - if (empty($vector_ids)) { | |
| 555 | - return $this->mxchat_query_based_list($pinecone_options, $page, $per_page, $bot_id, $content_type); | |
| 556 | - } | |
| 557 | - | |
| 558 | - // Fetch metadata for these vector IDs | |
| 559 | - return $this->mxchat_fetch_vectors_by_ids_for_list($pinecone_options, $vector_ids, $page, $per_page, $bot_id, $content_type); | |
| 560 | -} | |
| 561 | - | |
| 562 | -/** | |
| 563 | - * Query-based listing fallback when list endpoint fails | |
| 564 | - * ADDED 2.6.1 | |
| 565 | - */ | |
| 566 | -private function mxchat_query_based_list($pinecone_options, $page, $per_page, $bot_id, $content_type) { | |
| 567 | - $api_key = $pinecone_options['mxchat_pinecone_api_key'] ?? ''; | |
| 568 | - $host = $pinecone_options['mxchat_pinecone_host'] ?? ''; | |
| 569 | - $namespace = $pinecone_options['mxchat_pinecone_namespace'] ?? ''; | |
| 570 | - | |
| 571 | - $query_url = "https://{$host}/query"; | |
| 572 | - $query_vector = $this->mxchat_generate_optimized_query_vector(); | |
| 573 | - | |
| 574 | - // Fetch records - use higher limit to cover large databases | |
| 575 | - // Pinecone query API supports up to 10,000 topK | |
| 576 | - // We fetch more than needed to get accurate total count and enable pagination | |
| 577 | - $fetch_limit = 5000; | |
| 578 | - | |
| 579 | - $query_data = array( | |
| 580 | - 'includeMetadata' => true, | |
| 581 | - 'includeValues' => false, | |
| 582 | - 'topK' => $fetch_limit, | |
| 583 | - 'vector' => $query_vector | |
| 584 | - ); | |
| 585 | - | |
| 586 | - if (!empty($namespace)) { | |
| 587 | - $query_data['namespace'] = $namespace; | |
| 588 | - } | |
| 589 | - | |
| 590 | - // Add content type filter if specified | |
| 591 | - if (!empty($content_type)) { | |
| 592 | - $query_data['filter'] = array( | |
| 593 | - 'type' => array('$eq' => $content_type) | |
| 594 | - ); | |
| 595 | - } | |
| 596 | - | |
| 597 | - $response = wp_remote_post($query_url, array( | |
| 598 | - 'headers' => array( | |
| 599 | - 'Api-Key' => $api_key, | |
| 600 | - 'Content-Type' => 'application/json' | |
| 601 | - ), | |
| 602 | - 'body' => json_encode($query_data), | |
| 603 | - 'timeout' => 30 | |
| 604 | - )); | |
| 605 | - | |
| 606 | - if (is_wp_error($response) || wp_remote_retrieve_response_code($response) !== 200) { | |
| 607 | - return array('data' => array(), 'total' => 0); | |
| 608 | - } | |
| 609 | - | |
| 610 | - $body = wp_remote_retrieve_body($response); | |
| 611 | - $data = json_decode($body, true); | |
| 612 | - | |
| 613 | - $records = array(); | |
| 614 | - if (isset($data['matches'])) { | |
| 615 | - foreach ($data['matches'] as $match) { | |
| 616 | - $metadata = $match['metadata'] ?? array(); | |
| 617 | - | |
| 618 | - $created_at = $metadata['created_at'] ?? $metadata['last_updated'] ?? $metadata['timestamp'] ?? time(); | |
| 619 | - if (!is_numeric($created_at)) { | |
| 620 | - $created_at = strtotime($created_at) ?: time(); | |
| 621 | - } | |
| 622 | - | |
| 623 | - $records[] = (object) array( | |
| 624 | - 'id' => $match['id'] ?? '', | |
| 625 | - 'article_content' => $metadata['text'] ?? '', | |
| 626 | - 'source_url' => $metadata['source_url'] ?? '', | |
| 627 | - 'role_restriction' => $metadata['role_restriction'] ?? 'public', | |
| 628 | - 'type' => $metadata['type'] ?? 'content', | |
| 629 | - 'bot_id' => $bot_id, | |
| 630 | - 'created_at' => $created_at, | |
| 631 | - 'data_source' => 'pinecone', | |
| 632 | - 'chunk_index' => isset($metadata['chunk_index']) ? intval($metadata['chunk_index']) : null, | |
| 633 | - 'total_chunks' => isset($metadata['total_chunks']) ? intval($metadata['total_chunks']) : null, | |
| 634 | - 'is_chunked' => isset($metadata['is_chunked']) ? (bool) $metadata['is_chunked'] : false | |
| 635 | - ); | |
| 636 | - } | |
| 637 | - } | |
| 638 | - | |
| 639 | - // Sort by created_at (newest first) | |
| 640 | - usort($records, function($a, $b) { | |
| 641 | - return $b->created_at - $a->created_at; | |
| 642 | - }); | |
| 643 | - | |
| 644 | - $total = count($records); | |
| 645 | - $offset = ($page - 1) * $per_page; | |
| 646 | - $paged_records = array_slice($records, $offset, $per_page); | |
| 647 | - | |
| 648 | - $this->mxchat_batch_fetch_role_restrictions($paged_records, $bot_id); | |
| 649 | - | |
| 650 | - return array( | |
| 651 | - 'data' => $paged_records, | |
| 652 | - 'total' => $total | |
| 653 | - ); | |
| 654 | -} | |
| 655 | - | |
| 656 | -/** | |
| 657 | - * Fetch specific vectors by their IDs and format for display | |
| 658 | - * ADDED 2.6.1 | |
| 659 | - */ | |
| 660 | -private function mxchat_fetch_vectors_by_ids_for_list($pinecone_options, $vector_ids, $page, $per_page, $bot_id, $content_type) { | |
| 661 | - $api_key = $pinecone_options['mxchat_pinecone_api_key'] ?? ''; | |
| 662 | - $host = $pinecone_options['mxchat_pinecone_host'] ?? ''; | |
| 663 | - $namespace = $pinecone_options['mxchat_pinecone_namespace'] ?? ''; | |
| 664 | - | |
| 665 | - $fetch_url = "https://{$host}/vectors/fetch"; | |
| 666 | - | |
| 667 | - $fetch_data = array( | |
| 668 | - 'ids' => $vector_ids | |
| 669 | - ); | |
| 670 | - | |
| 671 | - if (!empty($namespace)) { | |
| 672 | - $fetch_data['namespace'] = $namespace; | |
| 673 | - } | |
| 674 | - | |
| 675 | - $response = wp_remote_post($fetch_url, array( | |
| 676 | - 'headers' => array( | |
| 677 | - 'Api-Key' => $api_key, | |
| 678 | - 'Content-Type' => 'application/json' | |
| 679 | - ), | |
| 680 | - 'body' => json_encode($fetch_data), | |
| 681 | - 'timeout' => 15 | |
| 682 | - )); | |
| 683 | - | |
| 684 | - if (is_wp_error($response) || wp_remote_retrieve_response_code($response) !== 200) { | |
| 685 | - return array('data' => array(), 'total' => 0); | |
| 686 | - } | |
| 687 | - | |
| 688 | - $body = wp_remote_retrieve_body($response); | |
| 689 | - $data = json_decode($body, true); | |
| 690 | - | |
| 691 | - $records = array(); | |
| 692 | - if (isset($data['vectors'])) { | |
| 693 | - foreach ($data['vectors'] as $vector_id => $vector_data) { | |
| 694 | - $metadata = $vector_data['metadata'] ?? array(); | |
| 695 | - | |
| 696 | - // Filter by content type if specified | |
| 697 | - if (!empty($content_type)) { | |
| 698 | - $record_type = $metadata['type'] ?? 'content'; | |
| 699 | - if ($record_type !== $content_type) { | |
| 700 | - continue; | |
| 701 | - } | |
| 702 | - } | |
| 703 | - | |
| 704 | - $created_at = $metadata['created_at'] ?? $metadata['last_updated'] ?? $metadata['timestamp'] ?? time(); | |
| 705 | - if (!is_numeric($created_at)) { | |
| 706 | - $created_at = strtotime($created_at) ?: time(); | |
| 707 | - } | |
| 708 | - | |
| 709 | - $records[] = (object) array( | |
| 710 | - 'id' => $vector_id, | |
| 711 | - 'article_content' => $metadata['text'] ?? '', | |
| 712 | - 'source_url' => $metadata['source_url'] ?? '', | |
| 713 | - 'role_restriction' => $metadata['role_restriction'] ?? 'public', | |
| 714 | - 'type' => $metadata['type'] ?? 'content', | |
| 715 | - 'bot_id' => $bot_id, | |
| 716 | - 'created_at' => $created_at, | |
| 717 | - 'data_source' => 'pinecone', | |
| 718 | - 'chunk_index' => isset($metadata['chunk_index']) ? intval($metadata['chunk_index']) : null, | |
| 719 | - 'total_chunks' => isset($metadata['total_chunks']) ? intval($metadata['total_chunks']) : null, | |
| 720 | - 'is_chunked' => isset($metadata['is_chunked']) ? (bool) $metadata['is_chunked'] : false | |
| 721 | - ); | |
| 722 | - } | |
| 723 | - } | |
| 724 | - | |
| 725 | - // Sort by date | |
| 726 | - usort($records, function($a, $b) { | |
| 727 | - return $b->created_at - $a->created_at; | |
| 728 | - }); | |
| 729 | - | |
| 730 | - $total = count($records); | |
| 731 | - $paged_records = array_slice($records, 0, $per_page); | |
| 732 | - | |
| 733 | - $this->mxchat_batch_fetch_role_restrictions($paged_records, $bot_id); | |
| 734 | - | |
| 735 | - return array( | |
| 736 | - 'data' => $paged_records, | |
| 737 | - 'total' => $total | |
| 738 | - ); | |
| 739 | -} | |
| 740 | - | |
| 741 | -/** | |
| 742 | - * Generate a single optimized query vector for fetching records | |
| 743 | - * Uses a center-weighted approach for best coverage | |
| 744 | - */ | |
| 745 | -private function mxchat_generate_optimized_query_vector() { | |
| 746 | - $dimensions = $this->mxchat_get_embedding_dimensions(); | |
| 747 | - $vector = array(); | |
| 748 | - | |
| 749 | - // Create a normalized center-weighted vector | |
| 750 | - $center = $dimensions / 2; | |
| 751 | - for ($i = 0; $i < $dimensions; $i++) { | |
| 752 | - $distance = abs($i - $center) / $center; | |
| 753 | - $vector[] = (1 - $distance) * 0.5; | |
| 754 | - } | |
| 755 | - | |
| 756 | - // Normalize to unit length | |
| 757 | - $magnitude = sqrt(array_sum(array_map(function($x) { return $x * $x; }, $vector))); | |
| 758 | - if ($magnitude > 0) { | |
| 759 | - $vector = array_map(function($x) use ($magnitude) { return $x / $magnitude; }, $vector); | |
| 760 | - } | |
| 761 | - | |
| 762 | - return $vector; | |
| 763 | -} | |
| 764 | - | |
| 765 | -/** | |
| 766 | - * Batch fetch role restrictions for a set of records | |
| 767 | - * Uses a single query instead of N queries | |
| 768 | - * ADDED 2.6.1: Prevents N+1 query problem | |
| 769 | - */ | |
| 770 | -private function mxchat_batch_fetch_role_restrictions(&$records, $bot_id = 'default') { | |
| 771 | - if (empty($records)) { | |
| 772 | - return; | |
| 773 | - } | |
| 774 | - | |
| 775 | - global $wpdb; | |
| 776 | - $roles_table = $wpdb->prefix . 'mxchat_pinecone_roles'; | |
| 777 | - | |
| 778 | - // Check if table exists | |
| 779 | - $table_exists = $wpdb->get_var("SHOW TABLES LIKE '{$roles_table}'"); | |
| 780 | - if (!$table_exists) { | |
| 781 | - return; | |
| 782 | - } | |
| 783 | - | |
| 784 | - // Get all vector IDs that need role lookup | |
| 785 | - $vector_ids = array(); | |
| 786 | - foreach ($records as $record) { | |
| 787 | - if (empty($record->role_restriction) || $record->role_restriction === 'public') { | |
| 788 | - $vector_ids[] = $record->id; | |
| 789 | - } | |
| 790 | - } | |
| 791 | - | |
| 792 | - if (empty($vector_ids)) { | |
| 793 | - return; | |
| 794 | - } | |
| 795 | - | |
| 796 | - // Check if bot_id column exists | |
| 797 | - $columns = $wpdb->get_col("SHOW COLUMNS FROM {$roles_table}"); | |
| 798 | - $has_bot_id = in_array('bot_id', $columns); | |
| 799 | - | |
| 800 | - // Build single query with IN clause | |
| 801 | - $placeholders = implode(',', array_fill(0, count($vector_ids), '%s')); | |
| 802 | - | |
| 803 | - if ($has_bot_id) { | |
| 804 | - $query = $wpdb->prepare( | |
| 805 | - "SELECT vector_id, role_restriction FROM {$roles_table} WHERE vector_id IN ({$placeholders}) AND bot_id = %s", | |
| 806 | - array_merge($vector_ids, array($bot_id)) | |
| 807 | - ); | |
| 808 | - } else { | |
| 809 | - $query = $wpdb->prepare( | |
| 810 | - "SELECT vector_id, role_restriction FROM {$roles_table} WHERE vector_id IN ({$placeholders})", | |
| 811 | - $vector_ids | |
| 812 | - ); | |
| 813 | - } | |
| 814 | - | |
| 815 | - $results = $wpdb->get_results($query, OBJECT_K); | |
| 816 | - | |
| 817 | - // Apply role restrictions to records | |
| 818 | - foreach ($records as &$record) { | |
| 819 | - if (isset($results[$record->id])) { | |
| 820 | - $record->role_restriction = $results[$record->id]->role_restriction; | |
| 821 | - } | |
| 822 | - } | |
| 823 | -} | |
| 824 | - | |
| 825 | -/** | |
| 826 | - * Safe version of get_recent_entries with memory limits | |
| 827 | - * ADDED 2.6.1: Prevents memory exhaustion | |
| 828 | - */ | |
| 829 | -private function mxchat_get_recent_entries_safe($pinecone_options, $bot_id = 'default', $limit = 500) { | |
| 830 | - global $wpdb; | |
| 831 | - | |
| 832 | - $api_key = $pinecone_options['mxchat_pinecone_api_key'] ?? ''; | |
| 833 | - $host = $pinecone_options['mxchat_pinecone_host'] ?? ''; | |
| 834 | - $namespace = $pinecone_options['mxchat_pinecone_namespace'] ?? ''; | |
| 835 | - | |
| 836 | - if (empty($api_key) || empty($host)) { | |
| 837 | - return array(); | |
| 838 | - } | |
| 839 | - | |
| 840 | - try { | |
| 841 | - $all_records = array(); | |
| 842 | - $seen_ids = array(); | |
| 843 | - $query_url = "https://{$host}/query"; | |
| 844 | - | |
| 845 | - // Use only 2 query vectors instead of 5 for better performance | |
| 846 | - $fixed_vectors = array_slice($this->mxchat_generate_fixed_query_vectors(), 0, 2); | |
| 847 | - | |
| 848 | - foreach ($fixed_vectors as $query_vector) { | |
| 849 | - // Limit topK to prevent memory issues | |
| 850 | - $topK = min(500, $limit); | |
| 851 | - | |
| 852 | - $query_data = array( | |
| 853 | - 'includeMetadata' => true, | |
| 854 | - 'includeValues' => false, | |
| 855 | - 'topK' => $topK, | |
| 856 | - 'vector' => $query_vector | |
| 857 | - ); | |
| 858 | - | |
| 859 | - if (!empty($namespace)) { | |
| 860 | - $query_data['namespace'] = $namespace; | |
| 861 | - } | |
| 862 | - | |
| 863 | - $response = wp_remote_post($query_url, array( | |
| 864 | - 'headers' => array( | |
| 865 | - 'Api-Key' => $api_key, | |
| 866 | - 'Content-Type' => 'application/json' | |
| 867 | - ), | |
| 868 | - 'body' => json_encode($query_data), | |
| 869 | - 'timeout' => 15 | |
| 870 | - )); | |
| 871 | - | |
| 872 | - if (!is_wp_error($response) && wp_remote_retrieve_response_code($response) === 200) { | |
| 873 | - $body = wp_remote_retrieve_body($response); | |
| 874 | - $data = json_decode($body, true); | |
| 875 | - | |
| 876 | - if (isset($data['matches'])) { | |
| 877 | - foreach ($data['matches'] as $match) { | |
| 878 | - $match_id = $match['id'] ?? ''; | |
| 879 | - if (!empty($match_id) && !isset($seen_ids[$match_id])) { | |
| 880 | - $metadata = $match['metadata'] ?? array(); | |
| 881 | - | |
| 882 | - $created_at = $metadata['created_at'] ?? $metadata['last_updated'] ?? $metadata['timestamp'] ?? time(); | |
| 883 | - if (!is_numeric($created_at)) { | |
| 884 | - $created_at = strtotime($created_at) ?: time(); | |
| 885 | - } | |
| 886 | - | |
| 887 | - $all_records[] = (object) array( | |
| 888 | - 'id' => $match_id, | |
| 889 | - 'article_content' => $metadata['text'] ?? '', | |
| 890 | - 'source_url' => $metadata['source_url'] ?? '', | |
| 891 | - 'role_restriction' => $metadata['role_restriction'] ?? 'public', | |
| 892 | - 'type' => $metadata['type'] ?? 'content', | |
| 893 | - 'bot_id' => $bot_id, | |
| 894 | - 'created_at' => $created_at, | |
| 895 | - 'data_source' => 'pinecone', | |
| 896 | - 'chunk_index' => isset($metadata['chunk_index']) ? intval($metadata['chunk_index']) : null, | |
| 897 | - 'total_chunks' => isset($metadata['total_chunks']) ? intval($metadata['total_chunks']) : null, | |
| 898 | - 'is_chunked' => isset($metadata['is_chunked']) ? (bool) $metadata['is_chunked'] : false | |
| 899 | - ); | |
| 900 | - | |
| 901 | - $seen_ids[$match_id] = true; | |
| 902 | - | |
| 903 | - // Stop if we've reached our limit | |
| 904 | - if (count($all_records) >= $limit) { | |
| 905 | - break 2; | |
| 906 | - } | |
| 907 | - } | |
| 908 | - } | |
| 909 | - } | |
| 910 | - } | |
| 911 | - | |
| 912 | - // Minimal delay between requests | |
| 913 | - usleep(50000); // 0.05 second delay | |
| 914 | - } | |
| 915 | - | |
| 916 | - // Sort by created_at (newest first) and apply limit | |
| 917 | - usort($all_records, function($a, $b) { | |
| 918 | - return $b->created_at - $a->created_at; | |
| 919 | - }); | |
| 920 | - | |
| 921 | - $limited_records = array_slice($all_records, 0, $limit); | |
| 922 | - | |
| 923 | - // Batch fetch role restrictions | |
| 924 | - $this->mxchat_batch_fetch_role_restrictions($limited_records, $bot_id); | |
| 925 | - | |
| 926 | - return $limited_records; | |
| 927 | - | |
| 928 | - } catch (Exception $e) { | |
| 929 | - //error_log('MxChat safe fetch exception: ' . $e->getMessage()); | |
| 930 | - MxChat_Admin::mxchat_log_debug('pinecone_error', 'Pinecone safe fetch error: ' . $e->getMessage()); | |
| 931 | - return array(); | |
| 932 | - } | |
| 933 | -} | |
| 934 | -/** | |
| 935 | - * Get embedding dimensions based on the selected model | |
| 936 | - * ADD THIS NEW FUNCTION | |
| 937 | - */ | |
| 938 | - private function mxchat_get_embedding_dimensions() { | |
| 939 | - $options = get_option('mxchat_options', array()); | |
| 940 | - $selected_model = $options['embedding_model'] ?? 'text-embedding-ada-002'; | |
| 941 | - | |
| 942 | - // Define dimensions for different models | |
| 943 | - $model_dimensions = array( | |
| 944 | - 'text-embedding-ada-002' => 1536, | |
| 945 | - 'text-embedding-3-small' => 1536, | |
| 946 | - 'text-embedding-3-large' => 3072, | |
| 947 | - 'voyage-2' => 1024, | |
| 948 | - 'voyage-large-2' => 1536, | |
| 949 | - 'voyage-3-large' => 2048, | |
| 950 | - 'gemini-embedding-001' => 1536, | |
| 951 | - ); | |
| 952 | - | |
| 953 | - // Check if it's a voyage model with custom dimensions | |
| 954 | - if (strpos($selected_model, 'voyage-3-large') === 0) { | |
| 955 | - $custom_dimensions = $options['voyage_output_dimension'] ?? 2048; | |
| 956 | - return intval($custom_dimensions); | |
| 957 | - } | |
| 958 | - | |
| 959 | - // Check if it's a gemini model with custom dimensions | |
| 960 | - if (strpos($selected_model, 'gemini-embedding') === 0) { | |
| 961 | - $custom_dimensions = $options['gemini_output_dimension'] ?? 1536; | |
| 962 | - return intval($custom_dimensions); | |
| 963 | - } | |
| 964 | - | |
| 965 | - // Return known dimensions or default to 1536 | |
| 966 | - return $model_dimensions[$selected_model] ?? 1536; | |
| 967 | - } | |
| 968 | - | |
| 969 | - /** | |
| 970 | - * Generate random unit vector with correct dimensions | |
| 971 | - * ADD THIS NEW FUNCTION | |
| 972 | - */ | |
| 973 | - private function mxchat_generate_random_vector() { | |
| 974 | - $dimensions = $this->mxchat_get_embedding_dimensions(); | |
| 975 | - | |
| 976 | - $random_vector = array(); | |
| 977 | - for ($i = 0; $i < $dimensions; $i++) { | |
| 978 | - $random_vector[] = (rand(-1000, 1000) / 1000.0); | |
| 979 | - } | |
| 980 | - | |
| 981 | - // Normalize the vector to unit length | |
| 982 | - $magnitude = sqrt(array_sum(array_map(function($x) { return $x * $x; }, $random_vector))); | |
| 983 | - if ($magnitude > 0) { | |
| 984 | - $random_vector = array_map(function($x) use ($magnitude) { return $x / $magnitude; }, $random_vector); | |
| 985 | - } | |
| 986 | - | |
| 987 | - return $random_vector; | |
| 988 | - } | |
| 989 | - | |
| 990 | - | |
| 991 | -/** | |
| 992 | - * Get recent entries from Pinecone | |
| 993 | - * UPDATED 2.6.1: Now uses safe version with memory limits to prevent crashes | |
| 994 | - * @deprecated Use mxchat_get_recent_entries_safe() instead for new code | |
| 995 | - */ | |
| 996 | -private function mxchat_get_recent_1k_entries($pinecone_options, $bot_id = 'default') { | |
| 997 | - // Delegate to the safe version with a reasonable limit | |
| 998 | - // This prevents crashes with large datasets (13K+ products) | |
| 999 | - return $this->mxchat_get_recent_entries_safe($pinecone_options, $bot_id, 500); | |
| 1000 | -} | |
| 1001 | -/** | |
| 1002 | - * Generate fixed query vectors for consistent results | |
| 1003 | - */ | |
| 1004 | -private function mxchat_generate_fixed_query_vectors() { | |
| 1005 | - $dimensions = $this->mxchat_get_embedding_dimensions(); | |
| 1006 | - $vectors = array(); | |
| 1007 | - | |
| 1008 | - // Create 5 fixed vectors with different patterns for better coverage | |
| 1009 | - $patterns = array( | |
| 1010 | - 'zeros_with_ones' => 0.1, // Mostly zeros with some 1s | |
| 1011 | - 'ascending' => 0.2, // Ascending pattern | |
| 1012 | - 'descending' => 0.3, // Descending pattern | |
| 1013 | - 'alternating' => 0.4, // Alternating positive/negative | |
| 1014 | - 'center_weighted' => 0.5 // Higher values in center | |
| 1015 | - ); | |
| 1016 | - | |
| 1017 | - foreach ($patterns as $pattern_name => $seed) { | |
| 1018 | - $vector = array(); | |
| 1019 | - | |
| 1020 | - for ($i = 0; $i < $dimensions; $i++) { | |
| 1021 | - switch ($pattern_name) { | |
| 1022 | - case 'zeros_with_ones': | |
| 1023 | - $vector[] = ($i % 10 === 0) ? 1.0 : 0.0; | |
| 1024 | - break; | |
| 1025 | - case 'ascending': | |
| 1026 | - $vector[] = ($i / $dimensions) * 2 - 1; // Range -1 to 1 | |
| 1027 | - break; | |
| 1028 | - case 'descending': | |
| 1029 | - $vector[] = (($dimensions - $i) / $dimensions) * 2 - 1; | |
| 1030 | - break; | |
| 1031 | - case 'alternating': | |
| 1032 | - $vector[] = ($i % 2 === 0) ? $seed : -$seed; | |
| 1033 | - break; | |
| 1034 | - case 'center_weighted': | |
| 1035 | - $center = $dimensions / 2; | |
| 1036 | - $distance = abs($i - $center) / $center; | |
| 1037 | - $vector[] = (1 - $distance) * $seed; | |
| 1038 | - break; | |
| 1039 | - } | |
| 1040 | - } | |
| 1041 | - | |
| 1042 | - // Normalize the vector to unit length | |
| 1043 | - $magnitude = sqrt(array_sum(array_map(function($x) { return $x * $x; }, $vector))); | |
| 1044 | - if ($magnitude > 0) { | |
| 1045 | - $vector = array_map(function($x) use ($magnitude) { return $x / $magnitude; }, $vector); | |
| 1046 | - } | |
| 1047 | - | |
| 1048 | - $vectors[] = $vector; | |
| 1049 | - } | |
| 1050 | - | |
| 1051 | - return $vectors; | |
| 1052 | -} | |
| 1053 | - | |
| 1054 | - | |
| 1055 | -/** | |
| 1056 | - * Scan Pinecone for processed content | |
| 1057 | - * UPDATED 2.6.2: Uses direct ID lookup via fetch API instead of random vector scanning | |
| 1058 | - * This removes the 10K record limit and scales to any database size | |
| 1059 | - * | |
| 1060 | - * @param array $pinecone_options Pinecone configuration options | |
| 1061 | - * @param array $post_ids Optional array of specific post IDs to check (if empty, checks all published posts) | |
| 1062 | - * @return array Processed data keyed by post ID | |
| 1063 | - */ | |
| 1064 | -public function mxchat_scan_pinecone_for_processed_content($pinecone_options, $post_ids = array()) { | |
| 1065 | - $api_key = $pinecone_options['mxchat_pinecone_api_key'] ?? ''; | |
| 1066 | - $host = $pinecone_options['mxchat_pinecone_host'] ?? ''; | |
| 1067 | - | |
| 1068 | - if (empty($api_key) || empty($host)) { | |
| 1069 | - return array(); | |
| 1070 | - } | |
| 1071 | - | |
| 1072 | - try { | |
| 1073 | - // If no specific post IDs provided, get all published posts | |
| 1074 | - if (empty($post_ids)) { | |
| 1075 | - $posts = get_posts(array( | |
| 1076 | - 'post_type' => 'any', | |
| 1077 | - 'post_status' => 'publish', | |
| 1078 | - 'posts_per_page' => -1, | |
| 1079 | - 'fields' => 'ids', | |
| 1080 | - 'no_found_rows' => true, | |
| 1081 | - 'update_post_meta_cache' => false, | |
| 1082 | - 'update_post_term_cache' => false, | |
| 1083 | - )); | |
| 1084 | - $post_ids = $posts; | |
| 1085 | - } | |
| 1086 | - | |
| 1087 | - if (empty($post_ids)) { | |
| 1088 | - return array(); | |
| 1089 | - } | |
| 1090 | - | |
| 1091 | - // Build a map of vector_id => post data for lookup | |
| 1092 | - $vector_id_map = array(); | |
| 1093 | - foreach ($post_ids as $post_id) { | |
| 1094 | - $permalink = get_permalink($post_id); | |
| 1095 | - if ($permalink) { | |
| 1096 | - $vector_id = md5($permalink); | |
| 1097 | - $vector_id_map[$vector_id] = array( | |
| 1098 | - 'post_id' => $post_id, | |
| 1099 | - 'url' => $permalink | |
| 1100 | - ); | |
| 1101 | - } | |
| 1102 | - } | |
| 1103 | - | |
| 1104 | - if (empty($vector_id_map)) { | |
| 1105 | - return array(); | |
| 1106 | - } | |
| 1107 | - | |
| 1108 | - // Batch check Pinecone using fetch API (max 1000 IDs per request) | |
| 1109 | - $all_vector_ids = array_keys($vector_id_map); | |
| 1110 | - $chunks = array_chunk($all_vector_ids, 1000); | |
| 1111 | - $processed_data = array(); | |
| 1112 | - | |
| 1113 | - foreach ($chunks as $chunk) { | |
| 1114 | - $fetch_url = "https://{$host}/vectors/fetch"; | |
| 1115 | - | |
| 1116 | - $response = wp_remote_post($fetch_url, array( | |
| 1117 | - 'headers' => array( | |
| 1118 | - 'Api-Key' => $api_key, | |
| 1119 | - 'Content-Type' => 'application/json' | |
| 1120 | - ), | |
| 1121 | - 'body' => json_encode(array('ids' => $chunk)), | |
| 1122 | - 'timeout' => 30 | |
| 1123 | - )); | |
| 1124 | - | |
| 1125 | - if (is_wp_error($response)) { | |
| 1126 | - continue; | |
| 1127 | - } | |
| 1128 | - | |
| 1129 | - $response_code = wp_remote_retrieve_response_code($response); | |
| 1130 | - if ($response_code !== 200) { | |
| 1131 | - continue; | |
| 1132 | - } | |
| 1133 | - | |
| 1134 | - $body = wp_remote_retrieve_body($response); | |
| 1135 | - $data = json_decode($body, true); | |
| 1136 | - | |
| 1137 | - // Process returned vectors | |
| 1138 | - if (isset($data['vectors']) && is_array($data['vectors'])) { | |
| 1139 | - foreach ($data['vectors'] as $vector_id => $vector_data) { | |
| 1140 | - if (isset($vector_id_map[$vector_id])) { | |
| 1141 | - $post_info = $vector_id_map[$vector_id]; | |
| 1142 | - $post_id = $post_info['post_id']; | |
| 1143 | - $metadata = $vector_data['metadata'] ?? array(); | |
| 1144 | - | |
| 1145 | - $created_at = $metadata['created_at'] ?? ''; | |
| 1146 | - $processed_date = 'Recently'; | |
| 1147 | - $timestamp = current_time('timestamp'); | |
| 1148 | - | |
| 1149 | - if (!empty($created_at)) { | |
| 1150 | - $ts = is_numeric($created_at) ? $created_at : strtotime($created_at); | |
| 1151 | - if ($ts) { | |
| 1152 | - $timestamp = $ts; | |
| 1153 | - $processed_date = human_time_diff($ts, current_time('timestamp')) . ' ago'; | |
| 1154 | - } | |
| 1155 | - } | |
| 1156 | - | |
| 1157 | - $processed_data[$post_id] = array( | |
| 1158 | - 'db_id' => $vector_id, | |
| 1159 | - 'processed_date' => $processed_date, | |
| 1160 | - 'url' => $post_info['url'], | |
| 1161 | - 'source' => 'pinecone', | |
| 1162 | - 'timestamp' => $timestamp | |
| 1163 | - ); | |
| 1164 | - } | |
| 1165 | - } | |
| 1166 | - } | |
| 1167 | - } | |
| 1168 | - | |
| 1169 | - return $processed_data; | |
| 1170 | - | |
| 1171 | - } catch (Exception $e) { | |
| 1172 | - return array(); | |
| 1173 | - } | |
| 1174 | -} | |
| 1175 | - | |
| 1176 | -/** | |
| 1177 | - * Get total count from Pinecone stats API | |
| 1178 | - * UPDATED: Removed cache fallback reference | |
| 1179 | - */ | |
| 1180 | -private function mxchat_get_pinecone_total_count($pinecone_options, $bot_id = 'default') { | |
| 1181 | - $api_key = $pinecone_options['mxchat_pinecone_api_key'] ?? ''; | |
| 1182 | - $host = $pinecone_options['mxchat_pinecone_host'] ?? ''; | |
| 1183 | - $namespace = $pinecone_options['mxchat_pinecone_namespace'] ?? ''; | |
| 1184 | - | |
| 1185 | - if (empty($api_key) || empty($host)) { | |
| 1186 | - return 0; | |
| 1187 | - } | |
| 1188 | - | |
| 1189 | - try { | |
| 1190 | - $stats_url = "https://{$host}/describe_index_stats"; | |
| 1191 | - | |
| 1192 | - // describe_index_stats doesn't need a body, just the POST request | |
| 1193 | - $response = wp_remote_post($stats_url, array( | |
| 1194 | - 'headers' => array( | |
| 1195 | - 'Api-Key' => $api_key, | |
| 1196 | - 'Content-Type' => 'application/json' | |
| 1197 | - ), | |
| 1198 | - 'body' => '{}', | |
| 1199 | - 'timeout' => 15 | |
| 1200 | - )); | |
| 1201 | - | |
| 1202 | - if (!is_wp_error($response) && wp_remote_retrieve_response_code($response) === 200) { | |
| 1203 | - $body = wp_remote_retrieve_body($response); | |
| 1204 | - $stats_data = json_decode($body, true); | |
| 1205 | - | |
| 1206 | - // If namespace is specified, get count from that specific namespace | |
| 1207 | - // Pinecone stats response format: { namespaces: { "ns": { vectorCount: N } }, totalVectorCount: N } | |
| 1208 | - if (!empty($namespace) && isset($stats_data['namespaces'][$namespace]['vectorCount'])) { | |
| 1209 | - $namespace_count = intval($stats_data['namespaces'][$namespace]['vectorCount']); | |
| 1210 | - //error_log('DEBUG: Got namespace-specific count: ' . $namespace_count . ' for namespace: ' . $namespace); | |
| 1211 | - return $namespace_count; | |
| 1212 | - } | |
| 1213 | - | |
| 1214 | - // If no namespace specified or namespace not found in response, use total | |
| 1215 | - $total_count = $stats_data['totalVectorCount'] ?? 0; | |
| 1216 | - if ($total_count > 0) { | |
| 1217 | - //error_log('DEBUG: Got total count from stats API: ' . $total_count); | |
| 1218 | - return intval($total_count); | |
| 1219 | - } | |
| 1220 | - } | |
| 1221 | - | |
| 1222 | - // If stats API fails, return 0 instead of using cache | |
| 1223 | - return 0; | |
| 1224 | - | |
| 1225 | - } catch (Exception $e) { | |
| 1226 | - //error_log('DEBUG: Exception getting total count: ' . $e->getMessage()); | |
| 1227 | - return 0; | |
| 1228 | - } | |
| 1229 | -} | |
| 1230 | - | |
| 1231 | -/** | |
| 1232 | - * Get bot-specific Pinecone configuration for database operations | |
| 1233 | - */ | |
| 1234 | -public function mxchat_get_bot_pinecone_options($bot_id = 'default') { | |
| 1235 | - //error_log('DEBUG: Getting Pinecone options for bot: ' . $bot_id); | |
| 1236 | - | |
| 1237 | - // If default bot or multi-bot add-on not active, use default Pinecone config | |
| 1238 | - if ($bot_id === 'default' || !class_exists('MxChat_Multi_Bot_Manager')) { | |
| 1239 | - $addon_options = get_option('mxchat_pinecone_addon_options', array()); | |
| 1240 | - //error_log('DEBUG: Using default Pinecone options'); | |
| 1241 | - return $addon_options; | |
| 1242 | - } | |
| 1243 | - | |
| 1244 | - // Get bot-specific configuration using the filter | |
| 1245 | - $bot_config = apply_filters('mxchat_get_bot_pinecone_config', array(), $bot_id); | |
| 1246 | - | |
| 1247 | - //error_log('DEBUG: Bot config from filter: ' . print_r($bot_config, true)); | |
| 1248 | - | |
| 1249 | - // Check if we got valid bot-specific config | |
| 1250 | - if (!empty($bot_config) && isset($bot_config['use_pinecone']) && $bot_config['use_pinecone']) { | |
| 1251 | - // Convert bot config to the format expected by fetch functions | |
| 1252 | - $pinecone_options = array( | |
| 1253 | - 'mxchat_use_pinecone' => '1', | |
| 1254 | - 'mxchat_pinecone_api_key' => $bot_config['api_key'] ?? '', | |
| 1255 | - 'mxchat_pinecone_host' => $bot_config['host'] ?? '', | |
| 1256 | - 'mxchat_pinecone_namespace' => $bot_config['namespace'] ?? '', | |
| 1257 | - 'mxchat_pinecone_environment' => '', | |
| 1258 | - 'mxchat_pinecone_index' => '' | |
| 1259 | - ); | |
| 1260 | - | |
| 1261 | - //error_log('DEBUG: Returning bot-specific Pinecone options for bot: ' . $bot_id); | |
| 1262 | - return $pinecone_options; | |
| 1263 | - } | |
| 1264 | - | |
| 1265 | - // Fallback to default options if bot-specific config is invalid | |
| 1266 | - //error_log('DEBUG: Bot-specific config invalid, falling back to default'); | |
| 1267 | - return get_option('mxchat_pinecone_addon_options', array()); | |
| 1268 | -} | |
| 1269 | - | |
| 1270 | -/** | |
| 1271 | - * Get bot-specific Pinecone configuration | |
| 1272 | - * Used in the knowledge retrieval functions | |
| 1273 | - */ | |
| 1274 | -private function get_bot_pinecone_config($bot_id = 'default') { | |
| 1275 | - //error_log("MXCHAT DEBUG: get_bot_pinecone_config called for bot: " . $bot_id); | |
| 1276 | - | |
| 1277 | - // If default bot or multi-bot add-on not active, use default Pinecone config | |
| 1278 | - if ($bot_id === 'default' || !class_exists('MxChat_Multi_Bot_Manager')) { | |
| 1279 | - //error_log("MXCHAT DEBUG: Using default Pinecone config (no multi-bot or bot is 'default')"); | |
| 1280 | - $addon_options = get_option('mxchat_pinecone_addon_options', array()); | |
| 1281 | - $config = array( | |
| 1282 | - 'use_pinecone' => (isset($addon_options['mxchat_use_pinecone']) && $addon_options['mxchat_use_pinecone'] === '1'), | |
| 1283 | - 'api_key' => $addon_options['mxchat_pinecone_api_key'] ?? '', | |
| 1284 | - 'host' => $addon_options['mxchat_pinecone_host'] ?? '', | |
| 1285 | - 'namespace' => $addon_options['mxchat_pinecone_namespace'] ?? '' | |
| 1286 | - ); | |
| 1287 | - //error_log("MXCHAT DEBUG: Default config - use_pinecone: " . ($config['use_pinecone'] ? 'true' : 'false')); | |
| 1288 | - return $config; | |
| 1289 | - } | |
| 1290 | - | |
| 1291 | - //error_log("MXCHAT DEBUG: Calling filter 'mxchat_get_bot_pinecone_config' for bot: " . $bot_id); | |
| 1292 | - | |
| 1293 | - // Hook for multi-bot add-on to provide bot-specific Pinecone config | |
| 1294 | - $bot_pinecone_config = apply_filters('mxchat_get_bot_pinecone_config', array(), $bot_id); | |
| 1295 | - | |
| 1296 | - if (!empty($bot_pinecone_config)) { | |
| 1297 | - //error_log("MXCHAT DEBUG: Got bot-specific config from filter"); | |
| 1298 | - //error_log(" - use_pinecone: " . (isset($bot_pinecone_config['use_pinecone']) ? ($bot_pinecone_config['use_pinecone'] ? 'true' : 'false') : 'not set')); | |
| 1299 | - //error_log(" - host: " . ($bot_pinecone_config['host'] ?? 'not set')); | |
| 1300 | - //error_log(" - namespace: " . ($bot_pinecone_config['namespace'] ?? 'not set')); | |
| 1301 | - } else { | |
| 1302 | - //error_log("MXCHAT DEBUG: Filter returned empty config!"); | |
| 1303 | - } | |
| 1304 | - | |
| 1305 | - return is_array($bot_pinecone_config) ? $bot_pinecone_config : array(); | |
| 1306 | -} | |
| 1307 | - | |
| 1308 | - | |
| 1309 | -/** | |
| 1310 | - * Fetches vectors from Pinecone using provided IDs (for content selection feature) | |
| 1311 | - */ | |
| 1312 | -public function fetch_pinecone_vectors_by_ids($pinecone_options, $vector_ids) { | |
| 1313 | - $api_key = $pinecone_options['mxchat_pinecone_api_key'] ?? ''; | |
| 1314 | - $host = $pinecone_options['mxchat_pinecone_host'] ?? ''; | |
| 1315 | - | |
| 1316 | - if (empty($api_key) || empty($host) || empty($vector_ids)) { | |
| 1317 | - return array(); | |
| 1318 | - } | |
| 1319 | - | |
| 1320 | - try { | |
| 1321 | - $fetch_url = "https://{$host}/vectors/fetch"; | |
| 1322 | - | |
| 1323 | - // Pinecone fetch API allows fetching specific vectors by ID | |
| 1324 | - $fetch_data = array( | |
| 1325 | - 'ids' => array_values($vector_ids) | |
| 1326 | - ); | |
| 1327 | - | |
| 1328 | - $response = wp_remote_post($fetch_url, array( | |
| 1329 | - 'headers' => array( | |
| 1330 | - 'Api-Key' => $api_key, | |
| 1331 | - 'Content-Type' => 'application/json' | |
| 1332 | - ), | |
| 1333 | - 'body' => json_encode($fetch_data), | |
| 1334 | - 'timeout' => 30 | |
| 1335 | - )); | |
| 1336 | - | |
| 1337 | - if (is_wp_error($response)) { | |
| 1338 | - return array(); | |
| 1339 | - } | |
| 1340 | - | |
| 1341 | - $response_code = wp_remote_retrieve_response_code($response); | |
| 1342 | - | |
| 1343 | - if ($response_code !== 200) { | |
| 1344 | - return array(); | |
| 1345 | - } | |
| 1346 | - | |
| 1347 | - $body = wp_remote_retrieve_body($response); | |
| 1348 | - $data = json_decode($body, true); | |
| 1349 | - | |
| 1350 | - if (!isset($data['vectors'])) { | |
| 1351 | - return array(); | |
| 1352 | - } | |
| 1353 | - | |
| 1354 | - $processed_data = array(); | |
| 1355 | - | |
| 1356 | - foreach ($data['vectors'] as $vector_id => $vector_data) { | |
| 1357 | - $metadata = $vector_data['metadata'] ?? array(); | |
| 1358 | - $source_url = $metadata['source_url'] ?? ''; | |
| 1359 | - | |
| 1360 | - if (!empty($source_url)) { | |
| 1361 | - $post_id = url_to_postid($source_url); | |
| 1362 | - if ($post_id) { | |
| 1363 | - $created_at = $metadata['created_at'] ?? ''; | |
| 1364 | - $processed_date = 'Recently'; // Default | |
| 1365 | - | |
| 1366 | - if (!empty($created_at)) { | |
| 1367 | - $timestamp = is_numeric($created_at) ? $created_at : strtotime($created_at); | |
| 1368 | - if ($timestamp) { | |
| 1369 | - $processed_date = human_time_diff($timestamp, current_time('timestamp')) . ' ago'; | |
| 1370 | - } | |
| 1371 | - } | |
| 1372 | - | |
| 1373 | - $processed_data[$post_id] = array( | |
| 1374 | - 'db_id' => $vector_id, | |
| 1375 | - 'processed_date' => $processed_date, | |
| 1376 | - 'url' => $source_url, | |
| 1377 | - 'source' => 'pinecone', | |
| 1378 | - 'timestamp' => $timestamp ?? current_time('timestamp') | |
| 1379 | - ); | |
| 1380 | - } | |
| 1381 | - } | |
| 1382 | - } | |
| 1383 | - | |
| 1384 | - return $processed_data; | |
| 1385 | - | |
| 1386 | - } catch (Exception $e) { | |
| 1387 | - return array(); | |
| 1388 | - } | |
| 1389 | -} | |
| 1390 | - // ======================================== | |
| 1391 | - // PINECONE DELETE OPERATIONS | |
| 1392 | - // ======================================== | |
| 1393 | - | |
| 1394 | -/** | |
| 1395 | - * Delete all vectors from Pinecone | |
| 1396 | - * Loops until all vectors are deleted (handles large databases) | |
| 1397 | - */ | |
| 1398 | -public function mxchat_delete_all_from_pinecone($pinecone_options, $content_type_filter = '') { | |
| 1399 | - $api_key = $pinecone_options['mxchat_pinecone_api_key'] ?? ''; | |
| 1400 | - $host = $pinecone_options['mxchat_pinecone_host'] ?? ''; | |
| 1401 | - | |
| 1402 | - if (empty($api_key) || empty($host)) { | |
| 1403 | - return array( | |
| 1404 | - 'success' => false, | |
| 1405 | - 'message' => 'Missing Pinecone API credentials' | |
| 1406 | - ); | |
| 1407 | - } | |
| 1408 | - | |
| 1409 | - try { | |
| 1410 | - $total_deleted = 0; | |
| 1411 | - $failed_batches = 0; | |
| 1412 | - $max_iterations = 100; // Safety limit to prevent infinite loops | |
| 1413 | - $iteration = 0; | |
| 1414 | - | |
| 1415 | - // Loop until no more vectors are found | |
| 1416 | - do { | |
| 1417 | - $iteration++; | |
| 1418 | - | |
| 1419 | - // Get a batch of vector IDs from Pinecone | |
| 1420 | - $records = $this->mxchat_get_recent_1k_entries($pinecone_options); | |
| 1421 | - $vector_ids = array(); | |
| 1422 | - | |
| 1423 | - foreach ($records as $record) { | |
| 1424 | - if (!empty($record->id)) { | |
| 1425 | - // If content type filter is active, only include matching records | |
| 1426 | - if (!empty($content_type_filter)) { | |
| 1427 | - $record_type = isset($record->type) ? $record->type : ''; | |
| 1428 | - if ($record_type !== $content_type_filter) { | |
| 1429 | - continue; | |
| 1430 | - } | |
| 1431 | - } | |
| 1432 | - $vector_ids[] = $record->id; | |
| 1433 | - } | |
| 1434 | - } | |
| 1435 | - | |
| 1436 | - // If no matching vectors found, we're done | |
| 1437 | - // (either no records at all, or all remaining records are non-matching types) | |
| 1438 | - if (empty($vector_ids)) { | |
| 1439 | - break; | |
| 1440 | - } | |
| 1441 | - | |
| 1442 | - // Delete vectors in batches (Pinecone has limits on batch operations) | |
| 1443 | - $batch_size = 100; | |
| 1444 | - $batches = array_chunk($vector_ids, $batch_size); | |
| 1445 | - | |
| 1446 | - foreach ($batches as $batch) { | |
| 1447 | - $result = $this->mxchat_delete_pinecone_batch($batch, $api_key, $host); | |
| 1448 | - if ($result['success']) { | |
| 1449 | - $total_deleted += count($batch); | |
| 1450 | - } else { | |
| 1451 | - $failed_batches++; | |
| 1452 | - } | |
| 1453 | - } | |
| 1454 | - | |
| 1455 | - // Small delay to avoid rate limiting | |
| 1456 | - usleep(100000); // 100ms | |
| 1457 | - | |
| 1458 | - } while ($iteration < $max_iterations); | |
| 1459 | - | |
| 1460 | - if ($total_deleted === 0) { | |
| 1461 | - return array( | |
| 1462 | - 'success' => true, | |
| 1463 | - 'message' => 'No vectors found to delete' | |
| 1464 | - ); | |
| 1465 | - } | |
| 1466 | - | |
| 1467 | - if ($failed_batches > 0) { | |
| 1468 | - return array( | |
| 1469 | - 'success' => false, | |
| 1470 | - 'message' => sprintf('Deleted %d vectors, but %d batches failed', $total_deleted, $failed_batches) | |
| 1471 | - ); | |
| 1472 | - } | |
| 1473 | - | |
| 1474 | - return array( | |
| 1475 | - 'success' => true, | |
| 1476 | - 'message' => "Successfully deleted {$total_deleted} vectors from Pinecone" | |
| 1477 | - ); | |
| 1478 | - | |
| 1479 | - } catch (Exception $e) { | |
| 1480 | - return array( | |
| 1481 | - 'success' => false, | |
| 1482 | - 'message' => $e->getMessage() | |
| 1483 | - ); | |
| 1484 | - } | |
| 1485 | -} | |
| 1486 | - | |
| 1487 | - /** | |
| 1488 | - * Deletes batch of vectors from Pinecone database | |
| 1489 | - */ | |
| 1490 | - public function mxchat_delete_pinecone_batch($vector_ids, $api_key, $host) { | |
| 1491 | - // Build the API endpoint | |
| 1492 | - $api_endpoint = "https://{$host}/vectors/delete"; | |
| 1493 | - | |
| 1494 | - // Prepare the request body with the IDs | |
| 1495 | - $request_body = array( | |
| 1496 | - 'ids' => $vector_ids | |
| 1497 | - ); | |
| 1498 | - | |
| 1499 | - // Make the deletion request | |
| 1500 | - $response = wp_remote_post($api_endpoint, array( | |
| 1501 | - 'headers' => array( | |
| 1502 | - 'Api-Key' => $api_key, | |
| 1503 | - 'accept' => 'application/json', | |
| 1504 | - 'content-type' => 'application/json' | |
| 1505 | - ), | |
| 1506 | - 'body' => wp_json_encode($request_body), | |
| 1507 | - 'timeout' => 60, // Increased timeout for batch operations | |
| 1508 | - 'method' => 'POST' | |
| 1509 | - )); | |
| 1510 | - | |
| 1511 | - // Handle WordPress HTTP API errors | |
| 1512 | - if (is_wp_error($response)) { | |
| 1513 | - MxChat_Admin::mxchat_log_debug('pinecone_error', 'Pinecone batch deletion failed: ' . $response->get_error_message()); | |
| 1514 | - return array( | |
| 1515 | - 'success' => false, | |
| 1516 | - 'message' => $response->get_error_message() | |
| 1517 | - ); | |
| 1518 | - } | |
| 1519 | - | |
| 1520 | - // Check response status | |
| 1521 | - $response_code = wp_remote_retrieve_response_code($response); | |
| 1522 | - $response_body = wp_remote_retrieve_body($response); | |
| 1523 | - | |
| 1524 | - // Pinecone returns 200 for successful deletion | |
| 1525 | - if ($response_code !== 200) { | |
| 1526 | - MxChat_Admin::mxchat_log_debug('pinecone_error', 'Pinecone batch deletion failed (HTTP ' . $response_code . ')', array('response' => substr($response_body, 0, 200))); | |
| 1527 | - return array( | |
| 1528 | - 'success' => false, | |
| 1529 | - 'message' => sprintf( | |
| 1530 | - 'Pinecone API error (HTTP %d): %s', | |
| 1531 | - $response_code, | |
| 1532 | - $response_body | |
| 1533 | - ) | |
| 1534 | - ); | |
| 1535 | - } | |
| 1536 | - | |
| 1537 | - return array( | |
| 1538 | - 'success' => true, | |
| 1539 | - 'message' => 'Batch deleted successfully from Pinecone' | |
| 1540 | - ); | |
| 1541 | - } | |
| 1542 | - | |
| 1543 | - | |
| 1544 | -/** | |
| 1545 | - * Deletes vector from Pinecone using API request | |
| 1546 | - */ | |
| 1547 | -public function mxchat_delete_from_pinecone_by_vector_id($vector_id, $api_key, $host, $namespace = '') { | |
| 1548 | - //error_log('=== PINECONE DELETE OPERATION ==='); | |
| 1549 | - //error_log('Vector ID: ' . $vector_id); | |
| 1550 | - //error_log('Host: ' . $host); | |
| 1551 | - //error_log('API Key: ' . (empty($api_key) ? 'EMPTY' : 'SET')); | |
| 1552 | - | |
| 1553 | - // First, let's verify the vector exists before trying to delete | |
| 1554 | - $fetch_url = "https://{$host}/vectors/fetch"; | |
| 1555 | - | |
| 1556 | - $fetch_params = array( | |
| 1557 | - 'ids' => array($vector_id) | |
| 1558 | - ); | |
| 1559 | - | |
| 1560 | - // Add namespace if provided (though you said you're not using namespaces) | |
| 1561 | - if (!empty($namespace)) { | |
| 1562 | - $fetch_params['namespace'] = $namespace; | |
| 1563 | - } | |
| 1564 | - | |
| 1565 | - // Construct URL with query parameters for GET request | |
| 1566 | - $fetch_url_with_params = $fetch_url . '?' . http_build_query($fetch_params); | |
| 1567 | - | |
| 1568 | - $fetch_response = wp_remote_get($fetch_url_with_params, array( | |
| 1569 | - 'headers' => array( | |
| 1570 | - 'Api-Key' => $api_key, | |
| 1571 | - 'accept' => 'application/json' | |
| 1572 | - ), | |
| 1573 | - 'timeout' => 15 | |
| 1574 | - )); | |
| 1575 | - | |
| 1576 | - if (!is_wp_error($fetch_response) && wp_remote_retrieve_response_code($fetch_response) === 200) { | |
| 1577 | - $fetch_body = wp_remote_retrieve_body($fetch_response); | |
| 1578 | - $fetch_data = json_decode($fetch_body, true); | |
| 1579 | - | |
| 1580 | - //error_log('DEBUG: Fetch response: ' . print_r($fetch_data, true)); | |
| 1581 | - | |
| 1582 | - if (isset($fetch_data['vectors']) && isset($fetch_data['vectors'][$vector_id])) { | |
| 1583 | - //error_log('DEBUG: Vector EXISTS in this index before deletion'); | |
| 1584 | - } else { | |
| 1585 | - //error_log('WARNING: Vector NOT FOUND in this index! It may be in a different bot\'s index'); | |
| 1586 | - // You might want to return an error here | |
| 1587 | - } | |
| 1588 | - } else { | |
| 1589 | - //error_log('DEBUG: Could not fetch vector to verify existence'); | |
| 1590 | - } | |
| 1591 | - | |
| 1592 | - // Now proceed with deletion | |
| 1593 | - $api_endpoint = "https://{$host}/vectors/delete"; | |
| 1594 | - | |
| 1595 | - // Prepare the request body with the ID | |
| 1596 | - $request_body = array( | |
| 1597 | - 'ids' => array($vector_id) | |
| 1598 | - ); | |
| 1599 | - | |
| 1600 | - // Add namespace if provided | |
| 1601 | - if (!empty($namespace)) { | |
| 1602 | - $request_body['namespace'] = $namespace; | |
| 1603 | - } | |
| 1604 | - | |
| 1605 | - //error_log('DEBUG: Delete request body: ' . json_encode($request_body)); | |
| 1606 | - //error_log('DEBUG: Delete endpoint: ' . $api_endpoint); | |
| 1607 | - | |
| 1608 | - // Make the deletion request | |
| 1609 | - $response = wp_remote_post($api_endpoint, array( | |
| 1610 | - 'headers' => array( | |
| 1611 | - 'Api-Key' => $api_key, | |
| 1612 | - 'accept' => 'application/json', | |
| 1613 | - 'content-type' => 'application/json' | |
| 1614 | - ), | |
| 1615 | - 'body' => wp_json_encode($request_body), | |
| 1616 | - 'timeout' => 30 | |
| 1617 | - )); | |
| 1618 | - | |
| 1619 | - // Handle WordPress HTTP API errors | |
| 1620 | - if (is_wp_error($response)) { | |
| 1621 | - //error_log('DEBUG: WP Error: ' . $response->get_error_message()); | |
| 1622 | - return array( | |
| 1623 | - 'success' => false, | |
| 1624 | - 'message' => $response->get_error_message() | |
| 1625 | - ); | |
| 1626 | - } | |
| 1627 | - | |
| 1628 | - // Check response status | |
| 1629 | - $response_code = wp_remote_retrieve_response_code($response); | |
| 1630 | - $response_body = wp_remote_retrieve_body($response); | |
| 1631 | - | |
| 1632 | - //error_log('DEBUG: Delete response code: ' . $response_code); | |
| 1633 | - //error_log('DEBUG: Delete response body: ' . $response_body); | |
| 1634 | - | |
| 1635 | - // Pinecone returns 200 for successful deletion (even if vector didn't exist) | |
| 1636 | - if ($response_code !== 200) { | |
| 1637 | - //error_log('DEBUG: Non-200 response from Pinecone'); | |
| 1638 | - return array( | |
| 1639 | - 'success' => false, | |
| 1640 | - 'message' => sprintf( | |
| 1641 | - 'Pinecone API error (HTTP %d): %s', | |
| 1642 | - $response_code, | |
| 1643 | - $response_body | |
| 1644 | - ) | |
| 1645 | - ); | |
| 1646 | - } | |
| 1647 | - | |
| 1648 | - // After deletion, verify it's actually gone | |
| 1649 | - sleep(1); // Give Pinecone a moment to process | |
| 1650 | - | |
| 1651 | - $verify_response = wp_remote_get($fetch_url_with_params, array( | |
| 1652 | - 'headers' => array( | |
| 1653 | - 'Api-Key' => $api_key, | |
| 1654 | - 'accept' => 'application/json' | |
| 1655 | - ), | |
| 1656 | - 'timeout' => 15 | |
| 1657 | - )); | |
| 1658 | - | |
| 1659 | - if (!is_wp_error($verify_response) && wp_remote_retrieve_response_code($verify_response) === 200) { | |
| 1660 | - $verify_body = wp_remote_retrieve_body($verify_response); | |
| 1661 | - $verify_data = json_decode($verify_body, true); | |
| 1662 | - | |
| 1663 | - if (isset($verify_data['vectors']) && isset($verify_data['vectors'][$vector_id])) { | |
| 1664 | - //error_log('ERROR: Vector STILL EXISTS after deletion attempt!'); | |
| 1665 | - return array( | |
| 1666 | - 'success' => false, | |
| 1667 | - 'message' => 'Vector still exists after deletion attempt' | |
| 1668 | - ); | |
| 1669 | - } else { | |
| 1670 | - //error_log('SUCCESS: Vector confirmed deleted (or never existed)'); | |
| 1671 | - } | |
| 1672 | - } | |
| 1673 | - | |
| 1674 | - //error_log('=== END PINECONE DELETE OPERATION ==='); | |
| 1675 | - | |
| 1676 | - return array( | |
| 1677 | - 'success' => true, | |
| 1678 | - 'message' => 'Vector deleted successfully from Pinecone' | |
| 1679 | - ); | |
| 1680 | -} | |
| 1681 | - | |
| 1682 | - /** | |
| 1683 | - * Deletes data from Pinecone index using API key | |
| 1684 | - */ | |
| 1685 | - private function mxchat_delete_from_pinecone($urls, $api_key, $environment, $index_name) { | |
| 1686 | - // Get the Pinecone host from options (matching your store_in_pinecone_main pattern) | |
| 1687 | - $options = get_option('mxchat_pinecone_addon_options'); | |
| 1688 | - $host = $options['mxchat_pinecone_host'] ?? ''; | |
| 1689 | - | |
| 1690 | - if (empty($host)) { | |
| 1691 | - return array( | |
| 1692 | - 'success' => false, | |
| 1693 | - 'message' => 'Pinecone host is not configured. Please set the host in your settings.' | |
| 1694 | - ); | |
| 1695 | - } | |
| 1696 | - | |
| 1697 | - // Build API endpoint using the configured host | |
| 1698 | - $api_endpoint = "https://{$host}/vectors/delete"; | |
| 1699 | - | |
| 1700 | - // Create vector IDs from URLs (matching your store method's ID generation) | |
| 1701 | - $vector_ids = array_map('md5', $urls); | |
| 1702 | - | |
| 1703 | - // Prepare the delete request body | |
| 1704 | - $request_body = array( | |
| 1705 | - 'ids' => $vector_ids, | |
| 1706 | - 'filter' => array( | |
| 1707 | - 'source_url' => array( | |
| 1708 | - '$in' => $urls | |
| 1709 | - ) | |
| 1710 | - ) | |
| 1711 | - ); | |
| 1712 | - | |
| 1713 | - // Make the deletion request | |
| 1714 | - $response = wp_remote_post($api_endpoint, array( | |
| 1715 | - 'headers' => array( | |
| 1716 | - 'Api-Key' => $api_key, | |
| 1717 | - 'accept' => 'application/json', | |
| 1718 | - 'content-type' => 'application/json' | |
| 1719 | - ), | |
| 1720 | - 'body' => wp_json_encode($request_body), | |
| 1721 | - 'timeout' => 30, | |
| 1722 | - 'data_format' => 'body' | |
| 1723 | - )); | |
| 1724 | - | |
| 1725 | - // Handle WordPress HTTP API errors | |
| 1726 | - if (is_wp_error($response)) { | |
| 1727 | - return array( | |
| 1728 | - 'success' => false, | |
| 1729 | - 'message' => $response->get_error_message() | |
| 1730 | - ); | |
| 1731 | - } | |
| 1732 | - | |
| 1733 | - // Check response status | |
| 1734 | - $response_code = wp_remote_retrieve_response_code($response); | |
| 1735 | - if ($response_code !== 200) { | |
| 1736 | - $body = wp_remote_retrieve_body($response); | |
| 1737 | - return array( | |
| 1738 | - 'success' => false, | |
| 1739 | - 'message' => sprintf( | |
| 1740 | - 'Pinecone API error (HTTP %d): %s', | |
| 1741 | - $response_code, | |
| 1742 | - $body | |
| 1743 | - ) | |
| 1744 | - ); | |
| 1745 | - } | |
| 1746 | - | |
| 1747 | - // Parse response body | |
| 1748 | - $body = wp_remote_retrieve_body($response); | |
| 1749 | - $response_data = json_decode($body, true); | |
| 1750 | - | |
| 1751 | - // Final validation of the response | |
| 1752 | - if (json_last_error() !== JSON_ERROR_NONE) { | |
| 1753 | - return array( | |
| 1754 | - 'success' => false, | |
| 1755 | - 'message' => 'Failed to parse Pinecone response: ' . json_last_error_msg() | |
| 1756 | - ); | |
| 1757 | - } | |
| 1758 | - | |
| 1759 | - return array( | |
| 1760 | - 'success' => true, | |
| 1761 | - 'message' => sprintf('Successfully deleted %d vectors from Pinecone', count($vector_ids)) | |
| 1762 | - ); | |
| 1763 | - } | |
| 1764 | - | |
| 1765 | - | |
| 1766 | - | |
| 1767 | -/** | |
| 1768 | - * Retrieves processed content from Pinecone API | |
| 1769 | - * | |
| 1770 | - * @param array $pinecone_options Pinecone configuration options | |
| 1771 | - * @param array $post_ids Optional array of specific post IDs to check (if empty, checks all) | |
| 1772 | - * @return array Processed data keyed by post ID | |
| 1773 | - */ | |
| 1774 | -public function mxchat_get_pinecone_processed_content($pinecone_options, $post_ids = array()) { | |
| 1775 | - $api_key = $pinecone_options['mxchat_pinecone_api_key'] ?? ''; | |
| 1776 | - $host = $pinecone_options['mxchat_pinecone_host'] ?? ''; | |
| 1777 | - | |
| 1778 | - if (empty($api_key) || empty($host)) { | |
| 1779 | - return array(); | |
| 1780 | - } | |
| 1781 | - | |
| 1782 | - $pinecone_data = array(); | |
| 1783 | - | |
| 1784 | - try { | |
| 1785 | - // Always get fresh data from Pinecone | |
| 1786 | - $pinecone_data = $this->mxchat_scan_pinecone_for_processed_content($pinecone_options, $post_ids); | |
| 1787 | - | |
| 1788 | - // Method 2: Final fallback - try stats endpoint (if available) | |
| 1789 | - if (empty($pinecone_data)) { | |
| 1790 | - $stats_url = "https://{$host}/describe_index_stats"; | |
| 1791 | - | |
| 1792 | - $response = wp_remote_post($stats_url, array( | |
| 1793 | - 'headers' => array( | |
| 1794 | - 'Api-Key' => $api_key, | |
| 1795 | - 'Content-Type' => 'application/json' | |
| 1796 | - ), | |
| 1797 | - 'body' => json_encode(array()), | |
| 1798 | - 'timeout' => 30 | |
| 1799 | - )); | |
| 1800 | - | |
| 1801 | - if (!is_wp_error($response) && wp_remote_retrieve_response_code($response) === 200) { | |
| 1802 | - $body = wp_remote_retrieve_body($response); | |
| 1803 | - $stats_data = json_decode($body, true); | |
| 1804 | - } | |
| 1805 | - } | |
| 1806 | - | |
| 1807 | - } catch (Exception $e) { | |
| 1808 | - // Log error but don't return cached data | |
| 1809 | - } | |
| 1810 | - | |
| 1811 | - return $pinecone_data; | |
| 1812 | -} | |
| 1813 | - // ======================================== | |
| 1814 | - // HELPER METHODS | |
| 1815 | - // ======================================== | |
| 1816 | - | |
| 1817 | - /** | |
| 1818 | - * Validates Pinecone API credentials | |
| 1819 | - */ | |
| 1820 | - private function mxchat_validate_pinecone_credentials($api_key, $host) { | |
| 1821 | - if (empty($api_key) || empty($host)) { | |
| 1822 | - return false; | |
| 1823 | - } | |
| 1824 | - return true; | |
| 1825 | - } | |
| 1826 | - | |
| 1827 | - /** | |
| 1828 | - * Get Pinecone API credentials from options | |
| 1829 | - */ | |
| 1830 | - private function mxchat_get_pinecone_credentials() { | |
| 1831 | - $options = get_option('mxchat_options', array()); | |
| 1832 | - return array( | |
| 1833 | - 'api_key' => isset($options['pinecone_api_key']) ? $options['pinecone_api_key'] : '', | |
| 1834 | - 'host' => isset($options['pinecone_host']) ? $options['pinecone_host'] : '' | |
| 1835 | - ); | |
| 1836 | - } | |
| 1837 | - | |
| 1838 | - /** | |
| 1839 | - * Log Pinecone operation errors | |
| 1840 | - */ | |
| 1841 | - private function log_pinecone_error($operation, $error_message) { | |
| 1842 | - //error_log("MxChat Pinecone {$operation} Error: " . $error_message); | |
| 1843 | - } | |
| 1844 | - | |
| 1845 | - // ======================================== | |
| 1846 | - // STATIC ACCESS METHODS (for backward compatibility) | |
| 1847 | - // ======================================== | |
| 1848 | - | |
| 1849 | - /** | |
| 1850 | - * Get singleton instance | |
| 1851 | - */ | |
| 1852 | - public static function get_instance() { | |
| 1853 | - static $instance = null; | |
| 1854 | - if ($instance === null) { | |
| 1855 | - $instance = new self(); | |
| 1856 | - } | |
| 1857 | - return $instance; | |
| 1858 | - } | |
| 1859 | -} | |
| 1860 | - | |
| 1861 | -// Initialize the Pinecone manager | |
| 1862 | -$mxchat_pinecone_manager = MxChat_Pinecone_Manager::get_instance(); | |
| 1 | +<?php | |
| 2 | +/** | |
| 3 | + * File: admin/class-pinecone-manager.php | |
| 4 | + * | |
| 5 | + * Handles all Pinecone vector database operations for MxChat | |
| 6 | + */ | |
| 7 | + | |
| 8 | +if (!defined('ABSPATH')) { | |
| 9 | + exit; // Exit if accessed directly | |
| 10 | +} | |
| 11 | + | |
| 12 | +class MxChat_Pinecone_Manager { | |
| 13 | + | |
| 14 | + /** | |
| 15 | + * Constructor | |
| 16 | + */ | |
| 17 | + public function __construct() { | |
| 18 | + } | |
| 19 | + | |
| 20 | + // ======================================== | |
| 21 | + // PINECONE FETCH OPERATIONS | |
| 22 | + // ======================================== | |
| 23 | + | |
| 24 | + | |
| 25 | +/** | |
| 26 | + * Fetches records from Pinecone with bot-specific filtering | |
| 27 | + * UPDATED 2.6.1: Optimized for large datasets - uses server-side pagination | |
| 28 | + */ | |
| 29 | +public function mxchat_fetch_pinecone_records($pinecone_options, $search_query = '', $page = 1, $per_page = 20, $bot_id = 'default', $content_type = '') { | |
| 30 | + $api_key = $pinecone_options['mxchat_pinecone_api_key'] ?? ''; | |
| 31 | + $host = $pinecone_options['mxchat_pinecone_host'] ?? ''; | |
| 32 | + $namespace = $pinecone_options['mxchat_pinecone_namespace'] ?? ''; | |
| 33 | + | |
| 34 | + if (empty($api_key) || empty($host)) { | |
| 35 | + return array('data' => array(), 'total' => 0, 'total_in_database' => 0, 'showing_recent_only' => false); | |
| 36 | + } | |
| 37 | + | |
| 38 | + // Document index (plan 362c31): cursor-paged listing + BM25 search through the documents API. | |
| 39 | + if (class_exists('MxChat_Pinecone_Documents') && MxChat_Pinecone_Documents::is_document_options($pinecone_options)) { | |
| 40 | + $doc_result = MxChat_Pinecone_Documents::fetch_records($pinecone_options, $search_query, $page, $per_page, $bot_id, $content_type); | |
| 41 | + if (!empty($doc_result['data'])) { | |
| 42 | + $this->mxchat_batch_fetch_role_restrictions($doc_result['data'], $bot_id); | |
| 43 | + } | |
| 44 | + return $doc_result; | |
| 45 | + } | |
| 46 | + | |
| 47 | + try { | |
| 48 | + // Get total count for the banner message (bot-specific) - lightweight call | |
| 49 | + $total_in_database = $this->mxchat_get_pinecone_total_count($pinecone_options, $bot_id); | |
| 50 | + | |
| 51 | + // For large databases, use optimized paginated fetching | |
| 52 | + // Only fetch what we need for the current page, not all 1K records | |
| 53 | + if ($total_in_database > 500) { | |
| 54 | + $result = $this->mxchat_fetch_pinecone_page_optimized($pinecone_options, $search_query, $page, $per_page, $bot_id, $content_type); | |
| 55 | + $result['total_in_database'] = $total_in_database; | |
| 56 | + $result['showing_recent_only'] = true; // Always show banner when we're limiting results | |
| 57 | + | |
| 58 | + // Plan dd6e10: the list path used to report the size of its one page | |
| 59 | + // fetch (per_page * 2 = 50) as 'total', so every index over 500 | |
| 60 | + // vectors read "Knowledge Entries (50)" and paginated to two pages. | |
| 61 | + // The only honest number on this path is the one we already fetched | |
| 62 | + // above — the index / namespace vector count — so return that, | |
| 63 | + // say it is vectors (a long entry is several of them), and flag | |
| 64 | + // that pages come from a cursor, not from a total. Semantic search | |
| 65 | + // keeps its own total: that is a real count of what matched. | |
| 66 | + if (empty($search_query)) { | |
| 67 | + $result['total'] = $total_in_database; | |
| 68 | + $result['count_unit'] = 'vectors'; | |
| 69 | + $result['cursor_pagination'] = true; | |
| 70 | + $result['has_next'] = !empty($result['has_next']); | |
| 71 | + } | |
| 72 | + return $result; | |
| 73 | + } | |
| 74 | + | |
| 75 | + // For smaller databases, use the existing approach but with safety limits | |
| 76 | + $all_records = $this->mxchat_get_recent_entries_safe($pinecone_options, $bot_id, 500); | |
| 77 | + | |
| 78 | + // Filter by content type if provided | |
| 79 | + if (!empty($content_type)) { | |
| 80 | + $all_records = array_filter($all_records, function($record) use ($content_type) { | |
| 81 | + $record_type = $record->type ?? 'content'; | |
| 82 | + return $record_type === $content_type; | |
| 83 | + }); | |
| 84 | + $all_records = array_values($all_records); // Re-index array | |
| 85 | + } | |
| 86 | + | |
| 87 | + // Filter by search query if provided | |
| 88 | + if (!empty($search_query)) { | |
| 89 | + $all_records = array_filter($all_records, function($record) use ($search_query) { | |
| 90 | + $content = $record->article_content ?? ''; | |
| 91 | + $source_url = $record->source_url ?? ''; | |
| 92 | + return stripos($content, $search_query) !== false || stripos($source_url, $search_query) !== false; | |
| 93 | + }); | |
| 94 | + $all_records = array_values($all_records); // Re-index array | |
| 95 | + } | |
| 96 | + | |
| 97 | + // UPDATED 2.6.3: Group records by source_url for chunk-aware pagination | |
| 98 | + // This ensures pagination shows X entries per page, not X chunks | |
| 99 | + $grouped_by_url = array(); | |
| 100 | + $empty_url_records = array(); | |
| 101 | + | |
| 102 | + foreach ($all_records as $record) { | |
| 103 | + $source_url = $record->source_url ?? ''; | |
| 104 | + if (empty($source_url)) { | |
| 105 | + $empty_url_records[] = $record; | |
| 106 | + } else { | |
| 107 | + if (!isset($grouped_by_url[$source_url])) { | |
| 108 | + $grouped_by_url[$source_url] = array(); | |
| 109 | + } | |
| 110 | + $grouped_by_url[$source_url][] = $record; | |
| 111 | + } | |
| 112 | + } | |
| 113 | + | |
| 114 | + // Count unique entries (unique URLs + individual empty-URL records) | |
| 115 | + $total_unique_entries = count($grouped_by_url) + count($empty_url_records); | |
| 116 | + | |
| 117 | + // Paginate by unique entries | |
| 118 | + $offset = ($page - 1) * $per_page; | |
| 119 | + | |
| 120 | + // Build ordered list of URL groups (newest first based on first record) | |
| 121 | + $url_groups_ordered = array_keys($grouped_by_url); | |
| 122 | + | |
| 123 | + // Get the URLs for this page | |
| 124 | + $page_urls = array_slice($url_groups_ordered, $offset, $per_page); | |
| 125 | + | |
| 126 | + // Collect all records for this page's URLs | |
| 127 | + $paged_records = array(); | |
| 128 | + foreach ($page_urls as $url) { | |
| 129 | + foreach ($grouped_by_url[$url] as $record) { | |
| 130 | + $paged_records[] = $record; | |
| 131 | + } | |
| 132 | + } | |
| 133 | + | |
| 134 | + // Add empty-URL records if they fall within this page's range | |
| 135 | + $remaining_slots = $per_page - count($page_urls); | |
| 136 | + $empty_offset = max(0, $offset - count($grouped_by_url)); | |
| 137 | + if ($remaining_slots > 0 && $empty_offset < count($empty_url_records)) { | |
| 138 | + $empty_page_records = array_slice($empty_url_records, $empty_offset, $remaining_slots); | |
| 139 | + $paged_records = array_merge($paged_records, $empty_page_records); | |
| 140 | + } | |
| 141 | + | |
| 142 | + return array( | |
| 143 | + 'data' => $paged_records, | |
| 144 | + 'total' => $total_unique_entries, | |
| 145 | + 'total_in_database' => $total_in_database, | |
| 146 | + 'showing_recent_only' => ($total_in_database > 500) | |
| 147 | + ); | |
| 148 | + | |
| 149 | + } catch (Exception $e) { | |
| 150 | + //error_log('MxChat Pinecone fetch error: ' . $e->getMessage()); | |
| 151 | + MxChat_Admin::mxchat_log_debug('pinecone_error', 'Pinecone fetch error: ' . $e->getMessage()); | |
| 152 | + return array('data' => array(), 'total' => 0, 'total_in_database' => 0, 'showing_recent_only' => false); | |
| 153 | + } | |
| 154 | +} | |
| 155 | + | |
| 156 | +/** | |
| 157 | + * Optimized fetch for large Pinecone databases | |
| 158 | + * ADDED 2.6.1: Prevents crashes with large datasets | |
| 159 | + * UPDATED 2.6.1: When searching, uses semantic search with embedded query for accurate results across all 13K+ records | |
| 160 | + */ | |
| 161 | +private function mxchat_fetch_pinecone_page_optimized($pinecone_options, $search_query = '', $page = 1, $per_page = 20, $bot_id = 'default', $content_type = '') { | |
| 162 | + $api_key = $pinecone_options['mxchat_pinecone_api_key'] ?? ''; | |
| 163 | + $host = $pinecone_options['mxchat_pinecone_host'] ?? ''; | |
| 164 | + $namespace = $pinecone_options['mxchat_pinecone_namespace'] ?? ''; | |
| 165 | + | |
| 166 | + try { | |
| 167 | + // If user is searching, use semantic search with embedded query | |
| 168 | + // This searches ALL records in Pinecone, not just fetched ones | |
| 169 | + if (!empty($search_query)) { | |
| 170 | + return $this->mxchat_semantic_search_pinecone($pinecone_options, $search_query, $page, $per_page, $bot_id, $content_type); | |
| 171 | + } | |
| 172 | + | |
| 173 | + // For browsing (no search), use Pinecone's list endpoint for true pagination | |
| 174 | + return $this->mxchat_list_pinecone_records($pinecone_options, $page, $per_page, $bot_id, $content_type); | |
| 175 | + | |
| 176 | + } catch (Exception $e) { | |
| 177 | + //error_log('MxChat optimized fetch exception: ' . $e->getMessage()); | |
| 178 | + MxChat_Admin::mxchat_log_debug('pinecone_error', 'Pinecone optimized fetch error: ' . $e->getMessage()); | |
| 179 | + return array('data' => array(), 'total' => 0); | |
| 180 | + } | |
| 181 | +} | |
| 182 | + | |
| 183 | +/** | |
| 184 | + * Semantic search across ALL Pinecone records using embedded search query | |
| 185 | + * This allows users to find any of their 13K+ products by searching | |
| 186 | + * ADDED 2.6.1 | |
| 187 | + */ | |
| 188 | +private function mxchat_semantic_search_pinecone($pinecone_options, $search_query, $page = 1, $per_page = 20, $bot_id = 'default', $content_type = '') { | |
| 189 | + $api_key = $pinecone_options['mxchat_pinecone_api_key'] ?? ''; | |
| 190 | + $host = $pinecone_options['mxchat_pinecone_host'] ?? ''; | |
| 191 | + $namespace = $pinecone_options['mxchat_pinecone_namespace'] ?? ''; | |
| 192 | + | |
| 193 | + // Get embedding for the search query | |
| 194 | + $query_embedding = $this->mxchat_get_search_embedding($search_query); | |
| 195 | + | |
| 196 | + if (empty($query_embedding)) { | |
| 197 | + // Fallback to text-based search if embedding fails | |
| 198 | + return $this->mxchat_text_search_fallback($pinecone_options, $search_query, $page, $per_page, $bot_id, $content_type); | |
| 199 | + } | |
| 200 | + | |
| 201 | + $query_url = "https://{$host}/query"; | |
| 202 | + | |
| 203 | + // Fetch more results to allow for filtering and pagination | |
| 204 | + $fetch_limit = min(($page * $per_page) + 100, 500); | |
| 205 | + | |
| 206 | + $query_data = array( | |
| 207 | + 'includeMetadata' => true, | |
| 208 | + 'includeValues' => false, | |
| 209 | + 'topK' => $fetch_limit, | |
| 210 | + 'vector' => $query_embedding | |
| 211 | + ); | |
| 212 | + | |
| 213 | + if (!empty($namespace)) { | |
| 214 | + $query_data['namespace'] = $namespace; | |
| 215 | + } | |
| 216 | + | |
| 217 | + // Add metadata filter for content type if specified | |
| 218 | + if (!empty($content_type)) { | |
| 219 | + $query_data['filter'] = array( | |
| 220 | + 'type' => array('$eq' => $content_type) | |
| 221 | + ); | |
| 222 | + } | |
| 223 | + | |
| 224 | + $response = wp_remote_post($query_url, array( | |
| 225 | + 'headers' => array( | |
| 226 | + 'Api-Key' => $api_key, | |
| 227 | + 'Content-Type' => 'application/json' | |
| 228 | + ), | |
| 229 | + 'body' => json_encode($query_data), | |
| 230 | + 'timeout' => 20 | |
| 231 | + )); | |
| 232 | + | |
| 233 | + if (is_wp_error($response) || wp_remote_retrieve_response_code($response) !== 200) { | |
| 234 | + return $this->mxchat_text_search_fallback($pinecone_options, $search_query, $page, $per_page, $bot_id, $content_type); | |
| 235 | + } | |
| 236 | + | |
| 237 | + $body = wp_remote_retrieve_body($response); | |
| 238 | + $data = json_decode($body, true); | |
| 239 | + | |
| 240 | + $records = array(); | |
| 241 | + if (isset($data['matches'])) { | |
| 242 | + foreach ($data['matches'] as $match) { | |
| 243 | + $metadata = $match['metadata'] ?? array(); | |
| 244 | + $created_at = $metadata['created_at'] ?? $metadata['last_updated'] ?? $metadata['timestamp'] ?? time(); | |
| 245 | + if (!is_numeric($created_at)) { | |
| 246 | + $created_at = strtotime($created_at) ?: time(); | |
| 247 | + } | |
| 248 | + | |
| 249 | + $records[] = (object) array( | |
| 250 | + 'id' => $match['id'] ?? '', | |
| 251 | + 'article_content' => $metadata['text'] ?? '', | |
| 252 | + 'source_url' => $metadata['source_url'] ?? '', | |
| 253 | + 'role_restriction' => $metadata['role_restriction'] ?? 'public', | |
| 254 | + 'type' => $metadata['type'] ?? 'content', | |
| 255 | + 'bot_id' => $bot_id, | |
| 256 | + 'created_at' => $created_at, | |
| 257 | + 'data_source' => 'pinecone', | |
| 258 | + 'relevance_score' => $match['score'] ?? 0, | |
| 259 | + 'chunk_index' => isset($metadata['chunk_index']) ? intval($metadata['chunk_index']) : null, | |
| 260 | + 'total_chunks' => isset($metadata['total_chunks']) ? intval($metadata['total_chunks']) : null, | |
| 261 | + 'is_chunked' => isset($metadata['is_chunked']) ? (bool) $metadata['is_chunked'] : false | |
| 262 | + ); | |
| 263 | + } | |
| 264 | + } | |
| 265 | + | |
| 266 | + // For semantic search, results are already sorted by relevance (score) | |
| 267 | + // No need to re-sort by date | |
| 268 | + | |
| 269 | + $total = count($records); | |
| 270 | + $offset = ($page - 1) * $per_page; | |
| 271 | + $paged_records = array_slice($records, $offset, $per_page); | |
| 272 | + | |
| 273 | + $this->mxchat_batch_fetch_role_restrictions($paged_records, $bot_id); | |
| 274 | + | |
| 275 | + return array( | |
| 276 | + 'data' => $paged_records, | |
| 277 | + 'total' => $total | |
| 278 | + ); | |
| 279 | +} | |
| 280 | + | |
| 281 | +/** | |
| 282 | + * Get embedding vector for a search query | |
| 283 | + * Uses the same embedding model configured for the knowledge base | |
| 284 | + * ADDED 2.6.1 | |
| 285 | + */ | |
| 286 | +private function mxchat_get_search_embedding($search_query) { | |
| 287 | + $options = get_option('mxchat_options', array()); | |
| 288 | + | |
| 289 | + // Custom Provider embeddings first, as everywhere else (plan 418afd): the | |
| 290 | + // records were embedded with the custom model, so the query must be too. | |
| 291 | + // A failure returns null, which sends the caller to its text search. | |
| 292 | + if (isset($options['custom_provider_for_embeddings']) && $options['custom_provider_for_embeddings'] === 'on') { | |
| 293 | + $custom = MxChat_Utils::generate_embedding_custom($search_query, $options); | |
| 294 | + return is_array($custom) ? $custom : null; | |
| 295 | + } | |
| 296 | + | |
| 297 | + $embedding_model = $options['embedding_model'] ?? 'text-embedding-ada-002'; | |
| 298 | + | |
| 299 | + // Determine which API to use based on model | |
| 300 | + if (strpos($embedding_model, 'voyage-') === 0) { | |
| 301 | + return $this->mxchat_get_voyage_embedding($search_query, $options, $embedding_model); | |
| 302 | + } elseif (strpos($embedding_model, 'gemini-') === 0) { | |
| 303 | + return $this->mxchat_get_gemini_embedding($search_query, $options, $embedding_model); | |
| 304 | + } else { | |
| 305 | + return $this->mxchat_get_openai_embedding($search_query, $options, $embedding_model); | |
| 306 | + } | |
| 307 | +} | |
| 308 | + | |
| 309 | +/** | |
| 310 | + * Get OpenAI embedding for search query | |
| 311 | + */ | |
| 312 | +private function mxchat_get_openai_embedding($text, $options, $model) { | |
| 313 | + $api_key = $options['api_key'] ?? ''; | |
| 314 | + if (empty($api_key)) { | |
| 315 | + return null; | |
| 316 | + } | |
| 317 | + | |
| 318 | + $response = wp_remote_post('https://api.openai.com/v1/embeddings', array( | |
| 319 | + 'headers' => array( | |
| 320 | + 'Authorization' => 'Bearer ' . $api_key, | |
| 321 | + 'Content-Type' => 'application/json' | |
| 322 | + ), | |
| 323 | + 'body' => json_encode(array( | |
| 324 | + 'model' => $model, | |
| 325 | + 'input' => $text | |
| 326 | + )), | |
| 327 | + 'timeout' => 15 | |
| 328 | + )); | |
| 329 | + | |
| 330 | + if (is_wp_error($response) || wp_remote_retrieve_response_code($response) !== 200) { | |
| 331 | + return null; | |
| 332 | + } | |
| 333 | + | |
| 334 | + $body = json_decode(wp_remote_retrieve_body($response), true); | |
| 335 | + return $body['data'][0]['embedding'] ?? null; | |
| 336 | +} | |
| 337 | + | |
| 338 | +/** | |
| 339 | + * Get Voyage AI embedding for search query | |
| 340 | + */ | |
| 341 | +private function mxchat_get_voyage_embedding($text, $options, $model) { | |
| 342 | + $api_key = $options['voyage_api_key'] ?? ''; | |
| 343 | + if (empty($api_key)) { | |
| 344 | + return null; | |
| 345 | + } | |
| 346 | + | |
| 347 | + $request_body = array( | |
| 348 | + 'model' => $model, | |
| 349 | + 'input' => $text, | |
| 350 | + 'input_type' => 'query' | |
| 351 | + ); | |
| 352 | + | |
| 353 | + // Add output dimensions for voyage-3-large if configured | |
| 354 | + if (strpos($model, 'voyage-3-large') === 0 && !empty($options['voyage_output_dimension'])) { | |
| 355 | + $request_body['output_dimension'] = intval($options['voyage_output_dimension']); | |
| 356 | + } | |
| 357 | + | |
| 358 | + $response = wp_remote_post('https://api.voyageai.com/v1/embeddings', array( | |
| 359 | + 'headers' => array( | |
| 360 | + 'Authorization' => 'Bearer ' . $api_key, | |
| 361 | + 'Content-Type' => 'application/json' | |
| 362 | + ), | |
| 363 | + 'body' => json_encode($request_body), | |
| 364 | + 'timeout' => 15 | |
| 365 | + )); | |
| 366 | + | |
| 367 | + if (is_wp_error($response) || wp_remote_retrieve_response_code($response) !== 200) { | |
| 368 | + return null; | |
| 369 | + } | |
| 370 | + | |
| 371 | + $body = json_decode(wp_remote_retrieve_body($response), true); | |
| 372 | + return $body['data'][0]['embedding'] ?? null; | |
| 373 | +} | |
| 374 | + | |
| 375 | +/** | |
| 376 | + * Get Google Gemini embedding for search query | |
| 377 | + */ | |
| 378 | +private function mxchat_get_gemini_embedding($text, $options, $model) { | |
| 379 | + $api_key = $options['gemini_api_key'] ?? ''; | |
| 380 | + if (empty($api_key)) { | |
| 381 | + return null; | |
| 382 | + } | |
| 383 | + | |
| 384 | + $url = "https://generativelanguage.googleapis.com/v1beta/models/{$model}:embedContent?key={$api_key}"; | |
| 385 | + | |
| 386 | + $request_body = array( | |
| 387 | + 'model' => "models/{$model}", | |
| 388 | + 'content' => array( | |
| 389 | + 'parts' => array( | |
| 390 | + array('text' => $text) | |
| 391 | + ) | |
| 392 | + ), | |
| 393 | + 'taskType' => 'RETRIEVAL_QUERY' | |
| 394 | + ); | |
| 395 | + | |
| 396 | + // Add output dimensions if configured | |
| 397 | + if (!empty($options['gemini_output_dimension'])) { | |
| 398 | + $request_body['outputDimensionality'] = intval($options['gemini_output_dimension']); | |
| 399 | + } | |
| 400 | + | |
| 401 | + $response = wp_remote_post($url, array( | |
| 402 | + 'headers' => array( | |
| 403 | + 'Content-Type' => 'application/json' | |
| 404 | + ), | |
| 405 | + 'body' => json_encode($request_body), | |
| 406 | + 'timeout' => 15 | |
| 407 | + )); | |
| 408 | + | |
| 409 | + if (is_wp_error($response) || wp_remote_retrieve_response_code($response) !== 200) { | |
| 410 | + return null; | |
| 411 | + } | |
| 412 | + | |
| 413 | + $body = json_decode(wp_remote_retrieve_body($response), true); | |
| 414 | + return $body['embedding']['values'] ?? null; | |
| 415 | +} | |
| 416 | + | |
| 417 | +/** | |
| 418 | + * Fallback text search when embedding fails | |
| 419 | + * Fetches more records and filters by text match | |
| 420 | + * ADDED 2.6.1 | |
| 421 | + */ | |
| 422 | +private function mxchat_text_search_fallback($pinecone_options, $search_query, $page, $per_page, $bot_id, $content_type) { | |
| 423 | + $api_key = $pinecone_options['mxchat_pinecone_api_key'] ?? ''; | |
| 424 | + $host = $pinecone_options['mxchat_pinecone_host'] ?? ''; | |
| 425 | + $namespace = $pinecone_options['mxchat_pinecone_namespace'] ?? ''; | |
| 426 | + | |
| 427 | + $query_url = "https://{$host}/query"; | |
| 428 | + $query_vector = $this->mxchat_generate_optimized_query_vector(); | |
| 429 | + | |
| 430 | + // Fetch more records to search through | |
| 431 | + $query_data = array( | |
| 432 | + 'includeMetadata' => true, | |
| 433 | + 'includeValues' => false, | |
| 434 | + 'topK' => 1000, // Fetch more for text search | |
| 435 | + 'vector' => $query_vector | |
| 436 | + ); | |
| 437 | + | |
| 438 | + if (!empty($namespace)) { | |
| 439 | + $query_data['namespace'] = $namespace; | |
| 440 | + } | |
| 441 | + | |
| 442 | + $response = wp_remote_post($query_url, array( | |
| 443 | + 'headers' => array( | |
| 444 | + 'Api-Key' => $api_key, | |
| 445 | + 'Content-Type' => 'application/json' | |
| 446 | + ), | |
| 447 | + 'body' => json_encode($query_data), | |
| 448 | + 'timeout' => 20 | |
| 449 | + )); | |
| 450 | + | |
| 451 | + if (is_wp_error($response) || wp_remote_retrieve_response_code($response) !== 200) { | |
| 452 | + return array('data' => array(), 'total' => 0); | |
| 453 | + } | |
| 454 | + | |
| 455 | + $body = wp_remote_retrieve_body($response); | |
| 456 | + $data = json_decode($body, true); | |
| 457 | + | |
| 458 | + $records = array(); | |
| 459 | + $search_lower = strtolower($search_query); | |
| 460 | + | |
| 461 | + if (isset($data['matches'])) { | |
| 462 | + foreach ($data['matches'] as $match) { | |
| 463 | + $metadata = $match['metadata'] ?? array(); | |
| 464 | + | |
| 465 | + // Filter by content type | |
| 466 | + if (!empty($content_type)) { | |
| 467 | + $record_type = $metadata['type'] ?? 'content'; | |
| 468 | + if ($record_type !== $content_type) { | |
| 469 | + continue; | |
| 470 | + } | |
| 471 | + } | |
| 472 | + | |
| 473 | + // Text search filter | |
| 474 | + $content = strtolower($metadata['text'] ?? ''); | |
| 475 | + $source_url = strtolower($metadata['source_url'] ?? ''); | |
| 476 | + if (strpos($content, $search_lower) === false && strpos($source_url, $search_lower) === false) { | |
| 477 | + continue; | |
| 478 | + } | |
| 479 | + | |
| 480 | + $created_at = $metadata['created_at'] ?? $metadata['last_updated'] ?? $metadata['timestamp'] ?? time(); | |
| 481 | + if (!is_numeric($created_at)) { | |
| 482 | + $created_at = strtotime($created_at) ?: time(); | |
| 483 | + } | |
| 484 | + | |
| 485 | + $records[] = (object) array( | |
| 486 | + 'id' => $match['id'] ?? '', | |
| 487 | + 'article_content' => $metadata['text'] ?? '', | |
| 488 | + 'source_url' => $metadata['source_url'] ?? '', | |
| 489 | + 'role_restriction' => $metadata['role_restriction'] ?? 'public', | |
| 490 | + 'type' => $metadata['type'] ?? 'content', | |
| 491 | + 'bot_id' => $bot_id, | |
| 492 | + 'created_at' => $created_at, | |
| 493 | + 'data_source' => 'pinecone', | |
| 494 | + 'chunk_index' => isset($metadata['chunk_index']) ? intval($metadata['chunk_index']) : null, | |
| 495 | + 'total_chunks' => isset($metadata['total_chunks']) ? intval($metadata['total_chunks']) : null, | |
| 496 | + 'is_chunked' => isset($metadata['is_chunked']) ? (bool) $metadata['is_chunked'] : false | |
| 497 | + ); | |
| 498 | + } | |
| 499 | + } | |
| 500 | + | |
| 501 | + // Sort by date for text search results | |
| 502 | + usort($records, function($a, $b) { | |
| 503 | + return $b->created_at - $a->created_at; | |
| 504 | + }); | |
| 505 | + | |
| 506 | + $total = count($records); | |
| 507 | + $offset = ($page - 1) * $per_page; | |
| 508 | + $paged_records = array_slice($records, $offset, $per_page); | |
| 509 | + | |
| 510 | + $this->mxchat_batch_fetch_role_restrictions($paged_records, $bot_id); | |
| 511 | + | |
| 512 | + return array( | |
| 513 | + 'data' => $paged_records, | |
| 514 | + 'total' => $total | |
| 515 | + ); | |
| 516 | +} | |
| 517 | + | |
| 518 | +/** | |
| 519 | + * List Pinecone records using the list endpoint for true pagination (no search) | |
| 520 | + * This allows browsing through all 13K+ records page by page | |
| 521 | + * ADDED 2.6.1 | |
| 522 | + */ | |
| 523 | +private function mxchat_list_pinecone_records($pinecone_options, $page = 1, $per_page = 20, $bot_id = 'default', $content_type = '') { | |
| 524 | + $api_key = $pinecone_options['mxchat_pinecone_api_key'] ?? ''; | |
| 525 | + $host = $pinecone_options['mxchat_pinecone_host'] ?? ''; | |
| 526 | + $namespace = $pinecone_options['mxchat_pinecone_namespace'] ?? ''; | |
| 527 | + | |
| 528 | + // Pinecone's list endpoint returns vector IDs with pagination | |
| 529 | + // We then fetch the metadata for those specific IDs | |
| 530 | + // NOTE (plan 793b82): /vectors/list is a GET endpoint with query parameters. | |
| 531 | + // A POST is answered 200-with-an-empty-body, which reads as "no vectors" — | |
| 532 | + // this call used to POST and only ever "worked" because the empty result | |
| 533 | + // tripped the query-based fallback below. | |
| 534 | + $list_url = "https://{$host}/vectors/list"; | |
| 535 | + | |
| 536 | + // Calculate pagination token from page number | |
| 537 | + // Pinecone uses cursor-based pagination, so we need to handle this differently. | |
| 538 | + // Plan dd6e10: when browsing unfiltered, one cursor step must be exactly one | |
| 539 | + // page — the old per_page * 2 over-fetch was sliced back to per_page, so the | |
| 540 | + // next cursor started 25 ids past what the visitor had seen. The over-fetch | |
| 541 | + // only earns its keep when a content-type filter will discard some of the ids. | |
| 542 | + $limit = !empty($content_type) ? $per_page * 2 : $per_page; | |
| 543 | + | |
| 544 | + $list_params = array( | |
| 545 | + 'limit' => $limit | |
| 546 | + ); | |
| 547 | + | |
| 548 | + if (!empty($namespace)) { | |
| 549 | + $list_params['namespace'] = $namespace; | |
| 550 | + } | |
| 551 | + | |
| 552 | + // For pages beyond first, we need to use pagination_token | |
| 553 | + // Store/retrieve pagination tokens in transients | |
| 554 | + $pagination_key = 'mxchat_pinecone_page_' . md5($host . $namespace . $content_type); | |
| 555 | + | |
| 556 | + if ($page > 1) { | |
| 557 | + $stored_tokens = get_transient($pagination_key); | |
| 558 | + if ($stored_tokens && isset($stored_tokens[$page])) { | |
| 559 | + $list_params['paginationToken'] = $stored_tokens[$page]; | |
| 560 | + } | |
| 561 | + } | |
| 562 | + | |
| 563 | + $response = wp_remote_get($list_url . '?' . http_build_query($list_params), array( | |
| 564 | + 'headers' => array( | |
| 565 | + 'Api-Key' => $api_key, | |
| 566 | + 'accept' => 'application/json' | |
| 567 | + ), | |
| 568 | + 'timeout' => 15 | |
| 569 | + )); | |
| 570 | + | |
| 571 | + if (is_wp_error($response) || wp_remote_retrieve_response_code($response) !== 200) { | |
| 572 | + // Genuine transport/API failure — fall back to the query-based approach, | |
| 573 | + // but say so: a silent fallback is what masked the POST bug for a year. | |
| 574 | + error_log('MxChat Pinecone: /vectors/list GET failed (' . (is_wp_error($response) ? $response->get_error_message() : wp_remote_retrieve_response_code($response)) . ') — falling back to query-based listing.'); | |
| 575 | + return $this->mxchat_query_based_list($pinecone_options, $page, $per_page, $bot_id, $content_type); | |
| 576 | + } | |
| 577 | + | |
| 578 | + $body = wp_remote_retrieve_body($response); | |
| 579 | + $data = json_decode($body, true); | |
| 580 | + | |
| 581 | + // Store pagination token for next page | |
| 582 | + $has_next = !empty($data['pagination']['next']); | |
| 583 | + if ($has_next) { | |
| 584 | + $stored_tokens = get_transient($pagination_key) ?: array(); | |
| 585 | + $stored_tokens[$page + 1] = $data['pagination']['next']; | |
| 586 | + set_transient($pagination_key, $stored_tokens, 300); // 5 minute cache | |
| 587 | + } | |
| 588 | + | |
| 589 | + $vector_ids = array(); | |
| 590 | + if (isset($data['vectors'])) { | |
| 591 | + foreach ($data['vectors'] as $vector) { | |
| 592 | + $vector_ids[] = $vector['id']; | |
| 593 | + } | |
| 594 | + } | |
| 595 | + | |
| 596 | + // A 200 with no vector ids is a REAL answer now (the namespace is empty), | |
| 597 | + // not the failure signature it was under POST — return the empty listing | |
| 598 | + // rather than routing into the fallback and masking a broken primary. | |
| 599 | + if (empty($vector_ids)) { | |
| 600 | + return array('data' => array(), 'total' => 0, 'has_next' => false); | |
| 601 | + } | |
| 602 | + | |
| 603 | + // Fetch metadata for these vector IDs | |
| 604 | + $fetched = $this->mxchat_fetch_vectors_by_ids_for_list($pinecone_options, $vector_ids, $page, $per_page, $bot_id, $content_type); | |
| 605 | + | |
| 606 | + // null = the fetch itself failed in transport (distinct from "ids had no | |
| 607 | + // metadata matches") — log and fall back so the screen still lists. | |
| 608 | + if ($fetched === null) { | |
| 609 | + error_log('MxChat Pinecone: /vectors/fetch GET failed for the listing — falling back to query-based listing.'); | |
| 610 | + return $this->mxchat_query_based_list($pinecone_options, $page, $per_page, $bot_id, $content_type); | |
| 611 | + } | |
| 612 | + | |
| 613 | + $fetched['has_next'] = $has_next; | |
| 614 | + return $fetched; | |
| 615 | +} | |
| 616 | + | |
| 617 | +/** | |
| 618 | + * Query-based listing fallback when list endpoint fails | |
| 619 | + * ADDED 2.6.1 | |
| 620 | + */ | |
| 621 | +private function mxchat_query_based_list($pinecone_options, $page, $per_page, $bot_id, $content_type) { | |
| 622 | + $api_key = $pinecone_options['mxchat_pinecone_api_key'] ?? ''; | |
| 623 | + $host = $pinecone_options['mxchat_pinecone_host'] ?? ''; | |
| 624 | + $namespace = $pinecone_options['mxchat_pinecone_namespace'] ?? ''; | |
| 625 | + | |
| 626 | + $query_url = "https://{$host}/query"; | |
| 627 | + $query_vector = $this->mxchat_generate_optimized_query_vector(); | |
| 628 | + | |
| 629 | + // Fetch records - use higher limit to cover large databases | |
| 630 | + // Pinecone query API supports up to 10,000 topK | |
| 631 | + // We fetch more than needed to get accurate total count and enable pagination | |
| 632 | + $fetch_limit = 5000; | |
| 633 | + | |
| 634 | + $query_data = array( | |
| 635 | + 'includeMetadata' => true, | |
| 636 | + 'includeValues' => false, | |
| 637 | + 'topK' => $fetch_limit, | |
| 638 | + 'vector' => $query_vector | |
| 639 | + ); | |
| 640 | + | |
| 641 | + if (!empty($namespace)) { | |
| 642 | + $query_data['namespace'] = $namespace; | |
| 643 | + } | |
| 644 | + | |
| 645 | + // Add content type filter if specified | |
| 646 | + if (!empty($content_type)) { | |
| 647 | + $query_data['filter'] = array( | |
| 648 | + 'type' => array('$eq' => $content_type) | |
| 649 | + ); | |
| 650 | + } | |
| 651 | + | |
| 652 | + $response = wp_remote_post($query_url, array( | |
| 653 | + 'headers' => array( | |
| 654 | + 'Api-Key' => $api_key, | |
| 655 | + 'Content-Type' => 'application/json' | |
| 656 | + ), | |
| 657 | + 'body' => json_encode($query_data), | |
| 658 | + 'timeout' => 30 | |
| 659 | + )); | |
| 660 | + | |
| 661 | + if (is_wp_error($response) || wp_remote_retrieve_response_code($response) !== 200) { | |
| 662 | + return array('data' => array(), 'total' => 0); | |
| 663 | + } | |
| 664 | + | |
| 665 | + $body = wp_remote_retrieve_body($response); | |
| 666 | + $data = json_decode($body, true); | |
| 667 | + | |
| 668 | + $records = array(); | |
| 669 | + if (isset($data['matches'])) { | |
| 670 | + foreach ($data['matches'] as $match) { | |
| 671 | + $metadata = $match['metadata'] ?? array(); | |
| 672 | + | |
| 673 | + $created_at = $metadata['created_at'] ?? $metadata['last_updated'] ?? $metadata['timestamp'] ?? time(); | |
| 674 | + if (!is_numeric($created_at)) { | |
| 675 | + $created_at = strtotime($created_at) ?: time(); | |
| 676 | + } | |
| 677 | + | |
| 678 | + $records[] = (object) array( | |
| 679 | + 'id' => $match['id'] ?? '', | |
| 680 | + 'article_content' => $metadata['text'] ?? '', | |
| 681 | + 'source_url' => $metadata['source_url'] ?? '', | |
| 682 | + 'role_restriction' => $metadata['role_restriction'] ?? 'public', | |
| 683 | + 'type' => $metadata['type'] ?? 'content', | |
| 684 | + 'bot_id' => $bot_id, | |
| 685 | + 'created_at' => $created_at, | |
| 686 | + 'data_source' => 'pinecone', | |
| 687 | + 'chunk_index' => isset($metadata['chunk_index']) ? intval($metadata['chunk_index']) : null, | |
| 688 | + 'total_chunks' => isset($metadata['total_chunks']) ? intval($metadata['total_chunks']) : null, | |
| 689 | + 'is_chunked' => isset($metadata['is_chunked']) ? (bool) $metadata['is_chunked'] : false | |
| 690 | + ); | |
| 691 | + } | |
| 692 | + } | |
| 693 | + | |
| 694 | + // Sort by created_at (newest first) | |
| 695 | + usort($records, function($a, $b) { | |
| 696 | + return $b->created_at - $a->created_at; | |
| 697 | + }); | |
| 698 | + | |
| 699 | + $total = count($records); | |
| 700 | + $offset = ($page - 1) * $per_page; | |
| 701 | + $paged_records = array_slice($records, $offset, $per_page); | |
| 702 | + | |
| 703 | + $this->mxchat_batch_fetch_role_restrictions($paged_records, $bot_id); | |
| 704 | + | |
| 705 | + return array( | |
| 706 | + 'data' => $paged_records, | |
| 707 | + 'total' => $total, | |
| 708 | + 'has_next' => ($offset + $per_page) < $total // offset-paged, but the cursor UI still asks (plan dd6e10) | |
| 709 | + ); | |
| 710 | +} | |
| 711 | + | |
| 712 | +/** | |
| 713 | + * Fetch specific vectors by their IDs and format for display | |
| 714 | + * ADDED 2.6.1 | |
| 715 | + */ | |
| 716 | +private function mxchat_fetch_vectors_by_ids_for_list($pinecone_options, $vector_ids, $page, $per_page, $bot_id, $content_type) { | |
| 717 | + $api_key = $pinecone_options['mxchat_pinecone_api_key'] ?? ''; | |
| 718 | + $host = $pinecone_options['mxchat_pinecone_host'] ?? ''; | |
| 719 | + $namespace = $pinecone_options['mxchat_pinecone_namespace'] ?? ''; | |
| 720 | + | |
| 721 | + // NOTE (plan 793b82): /vectors/fetch is a GET endpoint, and Pinecone expects | |
| 722 | + // the ids repeated (ids=a&ids=b) — http_build_query would emit ids[0]=a. | |
| 723 | + // A POST here is answered 200-with-an-empty-body, which reads as "no vectors". | |
| 724 | + $fetch_query = array(); | |
| 725 | + foreach ($vector_ids as $fetch_vid) { | |
| 726 | + $fetch_query[] = 'ids=' . rawurlencode($fetch_vid); | |
| 727 | + } | |
| 728 | + if (!empty($namespace)) { | |
| 729 | + $fetch_query[] = 'namespace=' . rawurlencode($namespace); | |
| 730 | + } | |
| 731 | + | |
| 732 | + $response = wp_remote_get("https://{$host}/vectors/fetch?" . implode('&', $fetch_query), array( | |
| 733 | + 'headers' => array( | |
| 734 | + 'Api-Key' => $api_key, | |
| 735 | + 'accept' => 'application/json' | |
| 736 | + ), | |
| 737 | + 'timeout' => 15 | |
| 738 | + )); | |
| 739 | + | |
| 740 | + // Transport/API failure is distinct from "nothing matched": return null so | |
| 741 | + // the caller can tell the difference (plan 793b82's empty-vs-failed rule). | |
| 742 | + if (is_wp_error($response) || wp_remote_retrieve_response_code($response) !== 200) { | |
| 743 | + return null; | |
| 744 | + } | |
| 745 | + | |
| 746 | + $body = wp_remote_retrieve_body($response); | |
| 747 | + $data = json_decode($body, true); | |
| 748 | + | |
| 749 | + $records = array(); | |
| 750 | + if (isset($data['vectors'])) { | |
| 751 | + foreach ($data['vectors'] as $vector_id => $vector_data) { | |
| 752 | + $metadata = $vector_data['metadata'] ?? array(); | |
| 753 | + | |
| 754 | + // Filter by content type if specified | |
| 755 | + if (!empty($content_type)) { | |
| 756 | + $record_type = $metadata['type'] ?? 'content'; | |
| 757 | + if ($record_type !== $content_type) { | |
| 758 | + continue; | |
| 759 | + } | |
| 760 | + } | |
| 761 | + | |
| 762 | + $created_at = $metadata['created_at'] ?? $metadata['last_updated'] ?? $metadata['timestamp'] ?? time(); | |
| 763 | + if (!is_numeric($created_at)) { | |
| 764 | + $created_at = strtotime($created_at) ?: time(); | |
| 765 | + } | |
| 766 | + | |
| 767 | + $records[] = (object) array( | |
| 768 | + 'id' => $vector_id, | |
| 769 | + 'article_content' => $metadata['text'] ?? '', | |
| 770 | + 'source_url' => $metadata['source_url'] ?? '', | |
| 771 | + 'role_restriction' => $metadata['role_restriction'] ?? 'public', | |
| 772 | + 'type' => $metadata['type'] ?? 'content', | |
| 773 | + 'bot_id' => $bot_id, | |
| 774 | + 'created_at' => $created_at, | |
| 775 | + 'data_source' => 'pinecone', | |
| 776 | + 'chunk_index' => isset($metadata['chunk_index']) ? intval($metadata['chunk_index']) : null, | |
| 777 | + 'total_chunks' => isset($metadata['total_chunks']) ? intval($metadata['total_chunks']) : null, | |
| 778 | + 'is_chunked' => isset($metadata['is_chunked']) ? (bool) $metadata['is_chunked'] : false | |
| 779 | + ); | |
| 780 | + } | |
| 781 | + } | |
| 782 | + | |
| 783 | + // Sort by date | |
| 784 | + usort($records, function($a, $b) { | |
| 785 | + return $b->created_at - $a->created_at; | |
| 786 | + }); | |
| 787 | + | |
| 788 | + $total = count($records); | |
| 789 | + $paged_records = array_slice($records, 0, $per_page); | |
| 790 | + | |
| 791 | + $this->mxchat_batch_fetch_role_restrictions($paged_records, $bot_id); | |
| 792 | + | |
| 793 | + return array( | |
| 794 | + 'data' => $paged_records, | |
| 795 | + 'total' => $total | |
| 796 | + ); | |
| 797 | +} | |
| 798 | + | |
| 799 | +/** | |
| 800 | + * Generate a single optimized query vector for fetching records | |
| 801 | + * Uses a center-weighted approach for best coverage | |
| 802 | + */ | |
| 803 | +private function mxchat_generate_optimized_query_vector() { | |
| 804 | + $dimensions = $this->mxchat_get_embedding_dimensions(); | |
| 805 | + $vector = array(); | |
| 806 | + | |
| 807 | + // Create a normalized center-weighted vector | |
| 808 | + $center = $dimensions / 2; | |
| 809 | + for ($i = 0; $i < $dimensions; $i++) { | |
| 810 | + $distance = abs($i - $center) / $center; | |
| 811 | + $vector[] = (1 - $distance) * 0.5; | |
| 812 | + } | |
| 813 | + | |
| 814 | + // Normalize to unit length | |
| 815 | + $magnitude = sqrt(array_sum(array_map(function($x) { return $x * $x; }, $vector))); | |
| 816 | + if ($magnitude > 0) { | |
| 817 | + $vector = array_map(function($x) use ($magnitude) { return $x / $magnitude; }, $vector); | |
| 818 | + } | |
| 819 | + | |
| 820 | + return $vector; | |
| 821 | +} | |
| 822 | + | |
| 823 | +/** | |
| 824 | + * Batch fetch role restrictions for a set of records | |
| 825 | + * Uses a single query instead of N queries | |
| 826 | + * ADDED 2.6.1: Prevents N+1 query problem | |
| 827 | + */ | |
| 828 | +private function mxchat_batch_fetch_role_restrictions(&$records, $bot_id = 'default') { | |
| 829 | + if (empty($records)) { | |
| 830 | + return; | |
| 831 | + } | |
| 832 | + | |
| 833 | + global $wpdb; | |
| 834 | + $roles_table = $wpdb->prefix . 'mxchat_pinecone_roles'; | |
| 835 | + | |
| 836 | + // Check if table exists | |
| 837 | + $table_exists = $wpdb->get_var("SHOW TABLES LIKE '{$roles_table}'"); | |
| 838 | + if (!$table_exists) { | |
| 839 | + return; | |
| 840 | + } | |
| 841 | + | |
| 842 | + // Get all vector IDs that need role lookup | |
| 843 | + $vector_ids = array(); | |
| 844 | + foreach ($records as $record) { | |
| 845 | + if (empty($record->role_restriction) || $record->role_restriction === 'public') { | |
| 846 | + $vector_ids[] = $record->id; | |
| 847 | + } | |
| 848 | + } | |
| 849 | + | |
| 850 | + if (empty($vector_ids)) { | |
| 851 | + return; | |
| 852 | + } | |
| 853 | + | |
| 854 | + // Check if bot_id column exists | |
| 855 | + $columns = $wpdb->get_col("SHOW COLUMNS FROM {$roles_table}"); | |
| 856 | + $has_bot_id = in_array('bot_id', $columns); | |
| 857 | + | |
| 858 | + // Build single query with IN clause | |
| 859 | + $placeholders = implode(',', array_fill(0, count($vector_ids), '%s')); | |
| 860 | + | |
| 861 | + if ($has_bot_id) { | |
| 862 | + $query = $wpdb->prepare( | |
| 863 | + "SELECT vector_id, role_restriction FROM {$roles_table} WHERE vector_id IN ({$placeholders}) AND bot_id = %s", | |
| 864 | + array_merge($vector_ids, array($bot_id)) | |
| 865 | + ); | |
| 866 | + } else { | |
| 867 | + $query = $wpdb->prepare( | |
| 868 | + "SELECT vector_id, role_restriction FROM {$roles_table} WHERE vector_id IN ({$placeholders})", | |
| 869 | + $vector_ids | |
| 870 | + ); | |
| 871 | + } | |
| 872 | + | |
| 873 | + $results = $wpdb->get_results($query, OBJECT_K); | |
| 874 | + | |
| 875 | + // Apply role restrictions to records | |
| 876 | + foreach ($records as &$record) { | |
| 877 | + if (isset($results[$record->id])) { | |
| 878 | + $record->role_restriction = $results[$record->id]->role_restriction; | |
| 879 | + } | |
| 880 | + } | |
| 881 | +} | |
| 882 | + | |
| 883 | +/** | |
| 884 | + * Safe version of get_recent_entries with memory limits | |
| 885 | + * ADDED 2.6.1: Prevents memory exhaustion | |
| 886 | + */ | |
| 887 | +private function mxchat_get_recent_entries_safe($pinecone_options, $bot_id = 'default', $limit = 500) { | |
| 888 | + global $wpdb; | |
| 889 | + | |
| 890 | + $api_key = $pinecone_options['mxchat_pinecone_api_key'] ?? ''; | |
| 891 | + $host = $pinecone_options['mxchat_pinecone_host'] ?? ''; | |
| 892 | + $namespace = $pinecone_options['mxchat_pinecone_namespace'] ?? ''; | |
| 893 | + | |
| 894 | + if (empty($api_key) || empty($host)) { | |
| 895 | + return array(); | |
| 896 | + } | |
| 897 | + | |
| 898 | + try { | |
| 899 | + $all_records = array(); | |
| 900 | + $seen_ids = array(); | |
| 901 | + $query_url = "https://{$host}/query"; | |
| 902 | + | |
| 903 | + // Use only 2 query vectors instead of 5 for better performance | |
| 904 | + $fixed_vectors = array_slice($this->mxchat_generate_fixed_query_vectors(), 0, 2); | |
| 905 | + | |
| 906 | + foreach ($fixed_vectors as $query_vector) { | |
| 907 | + // Limit topK to prevent memory issues | |
| 908 | + $topK = min(500, $limit); | |
| 909 | + | |
| 910 | + $query_data = array( | |
| 911 | + 'includeMetadata' => true, | |
| 912 | + 'includeValues' => false, | |
| 913 | + 'topK' => $topK, | |
| 914 | + 'vector' => $query_vector | |
| 915 | + ); | |
| 916 | + | |
| 917 | + if (!empty($namespace)) { | |
| 918 | + $query_data['namespace'] = $namespace; | |
| 919 | + } | |
| 920 | + | |
| 921 | + $response = wp_remote_post($query_url, array( | |
| 922 | + 'headers' => array( | |
| 923 | + 'Api-Key' => $api_key, | |
| 924 | + 'Content-Type' => 'application/json' | |
| 925 | + ), | |
| 926 | + 'body' => json_encode($query_data), | |
| 927 | + 'timeout' => 15 | |
| 928 | + )); | |
| 929 | + | |
| 930 | + if (!is_wp_error($response) && wp_remote_retrieve_response_code($response) === 200) { | |
| 931 | + $body = wp_remote_retrieve_body($response); | |
| 932 | + $data = json_decode($body, true); | |
| 933 | + | |
| 934 | + if (isset($data['matches'])) { | |
| 935 | + foreach ($data['matches'] as $match) { | |
| 936 | + $match_id = $match['id'] ?? ''; | |
| 937 | + if (!empty($match_id) && !isset($seen_ids[$match_id])) { | |
| 938 | + $metadata = $match['metadata'] ?? array(); | |
| 939 | + | |
| 940 | + $created_at = $metadata['created_at'] ?? $metadata['last_updated'] ?? $metadata['timestamp'] ?? time(); | |
| 941 | + if (!is_numeric($created_at)) { | |
| 942 | + $created_at = strtotime($created_at) ?: time(); | |
| 943 | + } | |
| 944 | + | |
| 945 | + $all_records[] = (object) array( | |
| 946 | + 'id' => $match_id, | |
| 947 | + 'article_content' => $metadata['text'] ?? '', | |
| 948 | + 'source_url' => $metadata['source_url'] ?? '', | |
| 949 | + 'role_restriction' => $metadata['role_restriction'] ?? 'public', | |
| 950 | + 'type' => $metadata['type'] ?? 'content', | |
| 951 | + 'bot_id' => $bot_id, | |
| 952 | + 'created_at' => $created_at, | |
| 953 | + 'data_source' => 'pinecone', | |
| 954 | + 'chunk_index' => isset($metadata['chunk_index']) ? intval($metadata['chunk_index']) : null, | |
| 955 | + 'total_chunks' => isset($metadata['total_chunks']) ? intval($metadata['total_chunks']) : null, | |
| 956 | + 'is_chunked' => isset($metadata['is_chunked']) ? (bool) $metadata['is_chunked'] : false | |
| 957 | + ); | |
| 958 | + | |
| 959 | + $seen_ids[$match_id] = true; | |
| 960 | + | |
| 961 | + // Stop if we've reached our limit | |
| 962 | + if (count($all_records) >= $limit) { | |
| 963 | + break 2; | |
| 964 | + } | |
| 965 | + } | |
| 966 | + } | |
| 967 | + } | |
| 968 | + } | |
| 969 | + | |
| 970 | + // Minimal delay between requests | |
| 971 | + usleep(50000); // 0.05 second delay | |
| 972 | + } | |
| 973 | + | |
| 974 | + // Sort by created_at (newest first) and apply limit | |
| 975 | + usort($all_records, function($a, $b) { | |
| 976 | + return $b->created_at - $a->created_at; | |
| 977 | + }); | |
| 978 | + | |
| 979 | + $limited_records = array_slice($all_records, 0, $limit); | |
| 980 | + | |
| 981 | + // Batch fetch role restrictions | |
| 982 | + $this->mxchat_batch_fetch_role_restrictions($limited_records, $bot_id); | |
| 983 | + | |
| 984 | + return $limited_records; | |
| 985 | + | |
| 986 | + } catch (Exception $e) { | |
| 987 | + //error_log('MxChat safe fetch exception: ' . $e->getMessage()); | |
| 988 | + MxChat_Admin::mxchat_log_debug('pinecone_error', 'Pinecone safe fetch error: ' . $e->getMessage()); | |
| 989 | + return array(); | |
| 990 | + } | |
| 991 | +} | |
| 992 | +/** | |
| 993 | + * Get embedding dimensions based on the selected model | |
| 994 | + * ADD THIS NEW FUNCTION | |
| 995 | + */ | |
| 996 | + private function mxchat_get_embedding_dimensions() { | |
| 997 | + // One answer for every caller (plan 418afd): the size of the model the | |
| 998 | + // site really embeds with, Custom Provider included. If a custom | |
| 999 | + // model cannot be asked, the dropdown's number is the old behaviour | |
| 1000 | + // and still better than no listing at all. | |
| 1001 | + $options = get_option('mxchat_options', array()); | |
| 1002 | + $dimension = MxChat_Utils::expected_embedding_dimension($options); | |
| 1003 | + if (!is_wp_error($dimension) && (int) $dimension > 0) { | |
| 1004 | + return (int) $dimension; | |
| 1005 | + } | |
| 1006 | + return MxChat_Utils::standard_embedding_dimension($options); | |
| 1007 | + } | |
| 1008 | + | |
| 1009 | + /** | |
| 1010 | + * Generate random unit vector with correct dimensions | |
| 1011 | + * ADD THIS NEW FUNCTION | |
| 1012 | + */ | |
| 1013 | + private function mxchat_generate_random_vector() { | |
| 1014 | + $dimensions = $this->mxchat_get_embedding_dimensions(); | |
| 1015 | + | |
| 1016 | + $random_vector = array(); | |
| 1017 | + for ($i = 0; $i < $dimensions; $i++) { | |
| 1018 | + $random_vector[] = (rand(-1000, 1000) / 1000.0); | |
| 1019 | + } | |
| 1020 | + | |
| 1021 | + // Normalize the vector to unit length | |
| 1022 | + $magnitude = sqrt(array_sum(array_map(function($x) { return $x * $x; }, $random_vector))); | |
| 1023 | + if ($magnitude > 0) { | |
| 1024 | + $random_vector = array_map(function($x) use ($magnitude) { return $x / $magnitude; }, $random_vector); | |
| 1025 | + } | |
| 1026 | + | |
| 1027 | + return $random_vector; | |
| 1028 | + } | |
| 1029 | + | |
| 1030 | + | |
| 1031 | +/** | |
| 1032 | + * Get recent entries from Pinecone | |
| 1033 | + * UPDATED 2.6.1: Now uses safe version with memory limits to prevent crashes | |
| 1034 | + * @deprecated Use mxchat_get_recent_entries_safe() instead for new code | |
| 1035 | + */ | |
| 1036 | +private function mxchat_get_recent_1k_entries($pinecone_options, $bot_id = 'default') { | |
| 1037 | + // Delegate to the safe version with a reasonable limit | |
| 1038 | + // This prevents crashes with large datasets (13K+ products) | |
| 1039 | + return $this->mxchat_get_recent_entries_safe($pinecone_options, $bot_id, 500); | |
| 1040 | +} | |
| 1041 | +/** | |
| 1042 | + * Generate fixed query vectors for consistent results | |
| 1043 | + */ | |
| 1044 | +private function mxchat_generate_fixed_query_vectors() { | |
| 1045 | + $dimensions = $this->mxchat_get_embedding_dimensions(); | |
| 1046 | + $vectors = array(); | |
| 1047 | + | |
| 1048 | + // Create 5 fixed vectors with different patterns for better coverage | |
| 1049 | + $patterns = array( | |
| 1050 | + 'zeros_with_ones' => 0.1, // Mostly zeros with some 1s | |
| 1051 | + 'ascending' => 0.2, // Ascending pattern | |
| 1052 | + 'descending' => 0.3, // Descending pattern | |
| 1053 | + 'alternating' => 0.4, // Alternating positive/negative | |
| 1054 | + 'center_weighted' => 0.5 // Higher values in center | |
| 1055 | + ); | |
| 1056 | + | |
| 1057 | + foreach ($patterns as $pattern_name => $seed) { | |
| 1058 | + $vector = array(); | |
| 1059 | + | |
| 1060 | + for ($i = 0; $i < $dimensions; $i++) { | |
| 1061 | + switch ($pattern_name) { | |
| 1062 | + case 'zeros_with_ones': | |
| 1063 | + $vector[] = ($i % 10 === 0) ? 1.0 : 0.0; | |
| 1064 | + break; | |
| 1065 | + case 'ascending': | |
| 1066 | + $vector[] = ($i / $dimensions) * 2 - 1; // Range -1 to 1 | |
| 1067 | + break; | |
| 1068 | + case 'descending': | |
| 1069 | + $vector[] = (($dimensions - $i) / $dimensions) * 2 - 1; | |
| 1070 | + break; | |
| 1071 | + case 'alternating': | |
| 1072 | + $vector[] = ($i % 2 === 0) ? $seed : -$seed; | |
| 1073 | + break; | |
| 1074 | + case 'center_weighted': | |
| 1075 | + $center = $dimensions / 2; | |
| 1076 | + $distance = abs($i - $center) / $center; | |
| 1077 | + $vector[] = (1 - $distance) * $seed; | |
| 1078 | + break; | |
| 1079 | + } | |
| 1080 | + } | |
| 1081 | + | |
| 1082 | + // Normalize the vector to unit length | |
| 1083 | + $magnitude = sqrt(array_sum(array_map(function($x) { return $x * $x; }, $vector))); | |
| 1084 | + if ($magnitude > 0) { | |
| 1085 | + $vector = array_map(function($x) use ($magnitude) { return $x / $magnitude; }, $vector); | |
| 1086 | + } | |
| 1087 | + | |
| 1088 | + $vectors[] = $vector; | |
| 1089 | + } | |
| 1090 | + | |
| 1091 | + return $vectors; | |
| 1092 | +} | |
| 1093 | + | |
| 1094 | + | |
| 1095 | +/** | |
| 1096 | + * Scan Pinecone for processed content | |
| 1097 | + * UPDATED 2.6.2: Uses direct ID lookup via fetch API instead of random vector scanning | |
| 1098 | + * This removes the 10K record limit and scales to any database size | |
| 1099 | + * | |
| 1100 | + * @param array $pinecone_options Pinecone configuration options | |
| 1101 | + * @param array $post_ids Optional array of specific post IDs to check (if empty, checks all published posts) | |
| 1102 | + * @return array Processed data keyed by post ID | |
| 1103 | + */ | |
| 1104 | +public function mxchat_scan_pinecone_for_processed_content($pinecone_options, $post_ids = array()) { | |
| 1105 | + $api_key = $pinecone_options['mxchat_pinecone_api_key'] ?? ''; | |
| 1106 | + $host = $pinecone_options['mxchat_pinecone_host'] ?? ''; | |
| 1107 | + | |
| 1108 | + if (empty($api_key) || empty($host)) { | |
| 1109 | + return array(); | |
| 1110 | + } | |
| 1111 | + | |
| 1112 | + try { | |
| 1113 | + // If no specific post IDs provided, get all published posts | |
| 1114 | + if (empty($post_ids)) { | |
| 1115 | + $posts = get_posts(array( | |
| 1116 | + 'post_type' => 'any', | |
| 1117 | + 'post_status' => 'publish', | |
| 1118 | + 'posts_per_page' => -1, | |
| 1119 | + 'fields' => 'ids', | |
| 1120 | + 'no_found_rows' => true, | |
| 1121 | + 'update_post_meta_cache' => false, | |
| 1122 | + 'update_post_term_cache' => false, | |
| 1123 | + )); | |
| 1124 | + $post_ids = $posts; | |
| 1125 | + } | |
| 1126 | + | |
| 1127 | + if (empty($post_ids)) { | |
| 1128 | + return array(); | |
| 1129 | + } | |
| 1130 | + | |
| 1131 | + // Build a map of vector_id => post data for lookup | |
| 1132 | + $vector_id_map = array(); | |
| 1133 | + foreach ($post_ids as $post_id) { | |
| 1134 | + $permalink = get_permalink($post_id); | |
| 1135 | + if ($permalink) { | |
| 1136 | + $vector_id = md5($permalink); | |
| 1137 | + $vector_id_map[$vector_id] = array( | |
| 1138 | + 'post_id' => $post_id, | |
| 1139 | + 'url' => $permalink | |
| 1140 | + ); | |
| 1141 | + } | |
| 1142 | + } | |
| 1143 | + | |
| 1144 | + if (empty($vector_id_map)) { | |
| 1145 | + return array(); | |
| 1146 | + } | |
| 1147 | + | |
| 1148 | + // Batch check Pinecone using fetch API. | |
| 1149 | + // NOTE (plan 793b82): /vectors/fetch is a GET endpoint with the ids | |
| 1150 | + // repeated in the query string (ids=a&ids=b); the old POST here was | |
| 1151 | + // answered 200-with-an-empty-body, so this scan saw NOTHING as indexed | |
| 1152 | + // and callers re-embedded content Pinecone already had. Chunk size 100: | |
| 1153 | + // measured on a live serverless index, ~9KB of URL is accepted and | |
| 1154 | + // ~18KB draws HTTP 414, so 100 32-char ids (~4KB) leaves real margin. | |
| 1155 | + $namespace = $pinecone_options['mxchat_pinecone_namespace'] ?? ''; | |
| 1156 | + $all_vector_ids = array_keys($vector_id_map); | |
| 1157 | + $chunks = array_chunk($all_vector_ids, 100); | |
| 1158 | + $processed_data = array(); | |
| 1159 | + | |
| 1160 | + // Document index (plan 362c31): one documents/fetch instead of the GET loop. | |
| 1161 | + if (class_exists('MxChat_Pinecone_Documents') && MxChat_Pinecone_Documents::is_document_options($pinecone_options)) { | |
| 1162 | + $chunks = array(); | |
| 1163 | + $doc_vectors = MxChat_Pinecone_Documents::fetch($all_vector_ids, MxChat_Pinecone_Documents::cfg_from_options($pinecone_options), array('source_url', 'created_at')); | |
| 1164 | + foreach ($doc_vectors as $vector_id => $vector_data) { | |
| 1165 | + if (isset($vector_id_map[$vector_id])) { | |
| 1166 | + $processed_data[$vector_id_map[$vector_id]['post_id']] = MxChat_Pinecone_Documents::processed_row($vector_id, $vector_id_map[$vector_id]['url'], $vector_data['metadata'] ?? array()); | |
| 1167 | + } | |
| 1168 | + } | |
| 1169 | + } | |
| 1170 | + | |
| 1171 | + foreach ($chunks as $chunk) { | |
| 1172 | + $fetch_query = array(); | |
| 1173 | + foreach ($chunk as $fetch_vid) { | |
| 1174 | + $fetch_query[] = 'ids=' . rawurlencode($fetch_vid); | |
| 1175 | + } | |
| 1176 | + if (!empty($namespace)) { | |
| 1177 | + $fetch_query[] = 'namespace=' . rawurlencode($namespace); | |
| 1178 | + } | |
| 1179 | + | |
| 1180 | + $response = wp_remote_get("https://{$host}/vectors/fetch?" . implode('&', $fetch_query), array( | |
| 1181 | + 'headers' => array( | |
| 1182 | + 'Api-Key' => $api_key, | |
| 1183 | + 'accept' => 'application/json' | |
| 1184 | + ), | |
| 1185 | + 'timeout' => 30 | |
| 1186 | + )); | |
| 1187 | + | |
| 1188 | + if (is_wp_error($response)) { | |
| 1189 | + error_log('MxChat Pinecone: processed-content scan /vectors/fetch GET failed: ' . $response->get_error_message()); | |
| 1190 | + continue; | |
| 1191 | + } | |
| 1192 | + | |
| 1193 | + $response_code = wp_remote_retrieve_response_code($response); | |
| 1194 | + if ($response_code !== 200) { | |
| 1195 | + error_log('MxChat Pinecone: processed-content scan /vectors/fetch GET returned HTTP ' . $response_code); | |
| 1196 | + continue; | |
| 1197 | + } | |
| 1198 | + | |
| 1199 | + $body = wp_remote_retrieve_body($response); | |
| 1200 | + $data = json_decode($body, true); | |
| 1201 | + | |
| 1202 | + // Process returned vectors | |
| 1203 | + if (isset($data['vectors']) && is_array($data['vectors'])) { | |
| 1204 | + foreach ($data['vectors'] as $vector_id => $vector_data) { | |
| 1205 | + if (isset($vector_id_map[$vector_id])) { | |
| 1206 | + $post_info = $vector_id_map[$vector_id]; | |
| 1207 | + $post_id = $post_info['post_id']; | |
| 1208 | + $metadata = $vector_data['metadata'] ?? array(); | |
| 1209 | + | |
| 1210 | + $created_at = $metadata['created_at'] ?? ''; | |
| 1211 | + $processed_date = 'Recently'; | |
| 1212 | + $timestamp = current_time('timestamp'); | |
| 1213 | + | |
| 1214 | + if (!empty($created_at)) { | |
| 1215 | + $ts = is_numeric($created_at) ? $created_at : strtotime($created_at); | |
| 1216 | + if ($ts) { | |
| 1217 | + $timestamp = $ts; | |
| 1218 | + $processed_date = human_time_diff($ts, current_time('timestamp')) . ' ago'; | |
| 1219 | + } | |
| 1220 | + } | |
| 1221 | + | |
| 1222 | + $processed_data[$post_id] = array( | |
| 1223 | + 'db_id' => $vector_id, | |
| 1224 | + 'processed_date' => $processed_date, | |
| 1225 | + 'url' => $post_info['url'], | |
| 1226 | + 'source' => 'pinecone', | |
| 1227 | + 'timestamp' => $timestamp | |
| 1228 | + ); | |
| 1229 | + } | |
| 1230 | + } | |
| 1231 | + } | |
| 1232 | + } | |
| 1233 | + | |
| 1234 | + return $processed_data; | |
| 1235 | + | |
| 1236 | + } catch (Exception $e) { | |
| 1237 | + return array(); | |
| 1238 | + } | |
| 1239 | +} | |
| 1240 | + | |
| 1241 | +/** | |
| 1242 | + * Get total count from Pinecone stats API | |
| 1243 | + * UPDATED: Removed cache fallback reference | |
| 1244 | + */ | |
| 1245 | +private function mxchat_get_pinecone_total_count($pinecone_options, $bot_id = 'default') { | |
| 1246 | + $api_key = $pinecone_options['mxchat_pinecone_api_key'] ?? ''; | |
| 1247 | + $host = $pinecone_options['mxchat_pinecone_host'] ?? ''; | |
| 1248 | + $namespace = $pinecone_options['mxchat_pinecone_namespace'] ?? ''; | |
| 1249 | + | |
| 1250 | + if (empty($api_key) || empty($host)) { | |
| 1251 | + return 0; | |
| 1252 | + } | |
| 1253 | + | |
| 1254 | + // Document index (plan 362c31): same stats call, version-headed, namespace mapped. | |
| 1255 | + if (class_exists('MxChat_Pinecone_Documents') && MxChat_Pinecone_Documents::is_document_options($pinecone_options)) { | |
| 1256 | + return MxChat_Pinecone_Documents::count(MxChat_Pinecone_Documents::cfg_from_options($pinecone_options)); | |
| 1257 | + } | |
| 1258 | + | |
| 1259 | + try { | |
| 1260 | + $stats_url = "https://{$host}/describe_index_stats"; | |
| 1261 | + | |
| 1262 | + // describe_index_stats doesn't need a body, just the POST request | |
| 1263 | + $response = wp_remote_post($stats_url, array( | |
| 1264 | + 'headers' => array( | |
| 1265 | + 'Api-Key' => $api_key, | |
| 1266 | + 'Content-Type' => 'application/json' | |
| 1267 | + ), | |
| 1268 | + 'body' => '{}', | |
| 1269 | + 'timeout' => 15 | |
| 1270 | + )); | |
| 1271 | + | |
| 1272 | + if (!is_wp_error($response) && wp_remote_retrieve_response_code($response) === 200) { | |
| 1273 | + $body = wp_remote_retrieve_body($response); | |
| 1274 | + $stats_data = json_decode($body, true); | |
| 1275 | + | |
| 1276 | + // If namespace is specified, get count from that specific namespace | |
| 1277 | + // Pinecone stats response format: { namespaces: { "ns": { vectorCount: N } }, totalVectorCount: N } | |
| 1278 | + if (!empty($namespace)) { | |
| 1279 | + // Plan dd6e10: a configured namespace (including a Multi-Bot | |
| 1280 | + // per-bot namespace, which arrives through the same option key) | |
| 1281 | + // is counted on its own. Pinecone omits empty namespaces from | |
| 1282 | + // the stats, so "not listed" means 0 here — falling through to | |
| 1283 | + // totalVectorCount would show a bot the whole index's number. | |
| 1284 | + return isset($stats_data['namespaces'][$namespace]['vectorCount']) | |
| 1285 | + ? intval($stats_data['namespaces'][$namespace]['vectorCount']) | |
| 1286 | + : 0; | |
| 1287 | + } | |
| 1288 | + | |
| 1289 | + // If no namespace specified or namespace not found in response, use total | |
| 1290 | + $total_count = $stats_data['totalVectorCount'] ?? 0; | |
| 1291 | + if ($total_count > 0) { | |
| 1292 | + //error_log('DEBUG: Got total count from stats API: ' . $total_count); | |
| 1293 | + return intval($total_count); | |
| 1294 | + } | |
| 1295 | + } | |
| 1296 | + | |
| 1297 | + // If stats API fails, return 0 instead of using cache | |
| 1298 | + return 0; | |
| 1299 | + | |
| 1300 | + } catch (Exception $e) { | |
| 1301 | + //error_log('DEBUG: Exception getting total count: ' . $e->getMessage()); | |
| 1302 | + return 0; | |
| 1303 | + } | |
| 1304 | +} | |
| 1305 | + | |
| 1306 | +/** | |
| 1307 | + * Get bot-specific Pinecone configuration for database operations | |
| 1308 | + */ | |
| 1309 | +public function mxchat_get_bot_pinecone_options($bot_id = 'default') { | |
| 1310 | + //error_log('DEBUG: Getting Pinecone options for bot: ' . $bot_id); | |
| 1311 | + | |
| 1312 | + // If default bot or multi-bot add-on not active, use default Pinecone config | |
| 1313 | + if ($bot_id === 'default' || !class_exists('MxChat_Multi_Bot_Manager')) { | |
| 1314 | + $addon_options = get_option('mxchat_pinecone_addon_options', array()); | |
| 1315 | + //error_log('DEBUG: Using default Pinecone options'); | |
| 1316 | + return $addon_options; | |
| 1317 | + } | |
| 1318 | + | |
| 1319 | + // Get bot-specific configuration using the filter | |
| 1320 | + $bot_config = apply_filters('mxchat_get_bot_pinecone_config', array(), $bot_id); | |
| 1321 | + | |
| 1322 | + //error_log('DEBUG: Bot config from filter: ' . print_r($bot_config, true)); | |
| 1323 | + | |
| 1324 | + // Check if we got valid bot-specific config | |
| 1325 | + if (!empty($bot_config) && isset($bot_config['use_pinecone']) && $bot_config['use_pinecone']) { | |
| 1326 | + // Convert bot config to the format expected by fetch functions | |
| 1327 | + $pinecone_options = array( | |
| 1328 | + 'mxchat_use_pinecone' => '1', | |
| 1329 | + 'mxchat_pinecone_api_key' => $bot_config['api_key'] ?? '', | |
| 1330 | + 'mxchat_pinecone_host' => $bot_config['host'] ?? '', | |
| 1331 | + 'mxchat_pinecone_namespace' => $bot_config['namespace'] ?? '', | |
| 1332 | + 'mxchat_pinecone_environment' => '', | |
| 1333 | + 'mxchat_pinecone_index' => '', | |
| 1334 | + // Plan 362c31: a bot's own index may be a document index (the Multi-Bot | |
| 1335 | + // add-on supplies 'index_type' when it learns the choice; default vector). | |
| 1336 | + 'mxchat_pinecone_index_type' => (($bot_config['index_type'] ?? 'vector') === 'document') ? 'document' : 'vector' | |
| 1337 | + ); | |
| 1338 | + | |
| 1339 | + //error_log('DEBUG: Returning bot-specific Pinecone options for bot: ' . $bot_id); | |
| 1340 | + return $pinecone_options; | |
| 1341 | + } | |
| 1342 | + | |
| 1343 | + // Fallback to default options if bot-specific config is invalid | |
| 1344 | + //error_log('DEBUG: Bot-specific config invalid, falling back to default'); | |
| 1345 | + return get_option('mxchat_pinecone_addon_options', array()); | |
| 1346 | +} | |
| 1347 | + | |
| 1348 | +/** | |
| 1349 | + * Get bot-specific Pinecone configuration | |
| 1350 | + * Used in the knowledge retrieval functions | |
| 1351 | + */ | |
| 1352 | +private function get_bot_pinecone_config($bot_id = 'default') { | |
| 1353 | + //error_log("MXCHAT DEBUG: get_bot_pinecone_config called for bot: " . $bot_id); | |
| 1354 | + | |
| 1355 | + // If default bot or multi-bot add-on not active, use default Pinecone config | |
| 1356 | + if ($bot_id === 'default' || !class_exists('MxChat_Multi_Bot_Manager')) { | |
| 1357 | + //error_log("MXCHAT DEBUG: Using default Pinecone config (no multi-bot or bot is 'default')"); | |
| 1358 | + $addon_options = get_option('mxchat_pinecone_addon_options', array()); | |
| 1359 | + $config = array( | |
| 1360 | + 'use_pinecone' => (isset($addon_options['mxchat_use_pinecone']) && $addon_options['mxchat_use_pinecone'] === '1'), | |
| 1361 | + 'api_key' => $addon_options['mxchat_pinecone_api_key'] ?? '', | |
| 1362 | + 'host' => $addon_options['mxchat_pinecone_host'] ?? '', | |
| 1363 | + 'namespace' => $addon_options['mxchat_pinecone_namespace'] ?? '' | |
| 1364 | + ); | |
| 1365 | + //error_log("MXCHAT DEBUG: Default config - use_pinecone: " . ($config['use_pinecone'] ? 'true' : 'false')); | |
| 1366 | + return $config; | |
| 1367 | + } | |
| 1368 | + | |
| 1369 | + //error_log("MXCHAT DEBUG: Calling filter 'mxchat_get_bot_pinecone_config' for bot: " . $bot_id); | |
| 1370 | + | |
| 1371 | + // Hook for multi-bot add-on to provide bot-specific Pinecone config | |
| 1372 | + $bot_pinecone_config = apply_filters('mxchat_get_bot_pinecone_config', array(), $bot_id); | |
| 1373 | + | |
| 1374 | + if (!empty($bot_pinecone_config)) { | |
| 1375 | + //error_log("MXCHAT DEBUG: Got bot-specific config from filter"); | |
| 1376 | + //error_log(" - use_pinecone: " . (isset($bot_pinecone_config['use_pinecone']) ? ($bot_pinecone_config['use_pinecone'] ? 'true' : 'false') : 'not set')); | |
| 1377 | + //error_log(" - host: " . ($bot_pinecone_config['host'] ?? 'not set')); | |
| 1378 | + //error_log(" - namespace: " . ($bot_pinecone_config['namespace'] ?? 'not set')); | |
| 1379 | + } else { | |
| 1380 | + //error_log("MXCHAT DEBUG: Filter returned empty config!"); | |
| 1381 | + } | |
| 1382 | + | |
| 1383 | + return is_array($bot_pinecone_config) ? $bot_pinecone_config : array(); | |
| 1384 | +} | |
| 1385 | + | |
| 1386 | + | |
| 1387 | +/** | |
| 1388 | + * Fetches vectors from Pinecone using provided IDs (for content selection feature) | |
| 1389 | + */ | |
| 1390 | +public function fetch_pinecone_vectors_by_ids($pinecone_options, $vector_ids) { | |
| 1391 | + $api_key = $pinecone_options['mxchat_pinecone_api_key'] ?? ''; | |
| 1392 | + $host = $pinecone_options['mxchat_pinecone_host'] ?? ''; | |
| 1393 | + $namespace = $pinecone_options['mxchat_pinecone_namespace'] ?? ''; | |
| 1394 | + | |
| 1395 | + if (empty($api_key) || empty($host) || empty($vector_ids)) { | |
| 1396 | + return array(); | |
| 1397 | + } | |
| 1398 | + | |
| 1399 | + try { | |
| 1400 | + // NOTE (plan 793b82): /vectors/fetch is a GET endpoint with the ids | |
| 1401 | + // repeated in the query string; the old POST here was answered | |
| 1402 | + // 200-with-an-empty-body. Chunked at 100 ids to stay well under the | |
| 1403 | + // measured HTTP 414 URL-length boundary. | |
| 1404 | + $vectors = array(); | |
| 1405 | + // Document index (plan 362c31): documents/fetch instead of the GET loop (same id => vector shape). | |
| 1406 | + if (class_exists('MxChat_Pinecone_Documents') && MxChat_Pinecone_Documents::is_document_options($pinecone_options)) { | |
| 1407 | + $vectors = MxChat_Pinecone_Documents::fetch(array_values($vector_ids), MxChat_Pinecone_Documents::cfg_from_options($pinecone_options), array('source_url', 'created_at')); | |
| 1408 | + } else | |
| 1409 | + foreach (array_chunk(array_values($vector_ids), 100) as $chunk) { | |
| 1410 | + $fetch_query = array(); | |
| 1411 | + foreach ($chunk as $fetch_vid) { | |
| 1412 | + $fetch_query[] = 'ids=' . rawurlencode($fetch_vid); | |
| 1413 | + } | |
| 1414 | + if (!empty($namespace)) { | |
| 1415 | + $fetch_query[] = 'namespace=' . rawurlencode($namespace); | |
| 1416 | + } | |
| 1417 | + | |
| 1418 | + $response = wp_remote_get("https://{$host}/vectors/fetch?" . implode('&', $fetch_query), array( | |
| 1419 | + 'headers' => array( | |
| 1420 | + 'Api-Key' => $api_key, | |
| 1421 | + 'accept' => 'application/json' | |
| 1422 | + ), | |
| 1423 | + 'timeout' => 30 | |
| 1424 | + )); | |
| 1425 | + | |
| 1426 | + if (is_wp_error($response)) { | |
| 1427 | + error_log('MxChat Pinecone: fetch_pinecone_vectors_by_ids GET failed: ' . $response->get_error_message()); | |
| 1428 | + continue; | |
| 1429 | + } | |
| 1430 | + | |
| 1431 | + if (wp_remote_retrieve_response_code($response) !== 200) { | |
| 1432 | + error_log('MxChat Pinecone: fetch_pinecone_vectors_by_ids GET returned HTTP ' . wp_remote_retrieve_response_code($response)); | |
| 1433 | + continue; | |
| 1434 | + } | |
| 1435 | + | |
| 1436 | + $data = json_decode(wp_remote_retrieve_body($response), true); | |
| 1437 | + if (isset($data['vectors']) && is_array($data['vectors'])) { | |
| 1438 | + $vectors += $data['vectors']; | |
| 1439 | + } | |
| 1440 | + } | |
| 1441 | + | |
| 1442 | + if (empty($vectors)) { | |
| 1443 | + return array(); | |
| 1444 | + } | |
| 1445 | + | |
| 1446 | + $processed_data = array(); | |
| 1447 | + | |
| 1448 | + foreach ($vectors as $vector_id => $vector_data) { | |
| 1449 | + $metadata = $vector_data['metadata'] ?? array(); | |
| 1450 | + $source_url = $metadata['source_url'] ?? ''; | |
| 1451 | + | |
| 1452 | + if (!empty($source_url)) { | |
| 1453 | + $post_id = url_to_postid($source_url); | |
| 1454 | + if ($post_id) { | |
| 1455 | + $created_at = $metadata['created_at'] ?? ''; | |
| 1456 | + $processed_date = 'Recently'; // Default | |
| 1457 | + | |
| 1458 | + if (!empty($created_at)) { | |
| 1459 | + $timestamp = is_numeric($created_at) ? $created_at : strtotime($created_at); | |
| 1460 | + if ($timestamp) { | |
| 1461 | + $processed_date = human_time_diff($timestamp, current_time('timestamp')) . ' ago'; | |
| 1462 | + } | |
| 1463 | + } | |
| 1464 | + | |
| 1465 | + $processed_data[$post_id] = array( | |
| 1466 | + 'db_id' => $vector_id, | |
| 1467 | + 'processed_date' => $processed_date, | |
| 1468 | + 'url' => $source_url, | |
| 1469 | + 'source' => 'pinecone', | |
| 1470 | + 'timestamp' => $timestamp ?? current_time('timestamp') | |
| 1471 | + ); | |
| 1472 | + } | |
| 1473 | + } | |
| 1474 | + } | |
| 1475 | + | |
| 1476 | + return $processed_data; | |
| 1477 | + | |
| 1478 | + } catch (Exception $e) { | |
| 1479 | + return array(); | |
| 1480 | + } | |
| 1481 | +} | |
| 1482 | + // ======================================== | |
| 1483 | + // PINECONE DELETE OPERATIONS | |
| 1484 | + // ======================================== | |
| 1485 | + | |
| 1486 | +/** | |
| 1487 | + * Delete all vectors from Pinecone | |
| 1488 | + * Loops until all vectors are deleted (handles large databases) | |
| 1489 | + */ | |
| 1490 | +public function mxchat_delete_all_from_pinecone($pinecone_options, $content_type_filter = '') { | |
| 1491 | + $api_key = $pinecone_options['mxchat_pinecone_api_key'] ?? ''; | |
| 1492 | + $host = $pinecone_options['mxchat_pinecone_host'] ?? ''; | |
| 1493 | + $namespace = $pinecone_options['mxchat_pinecone_namespace'] ?? ''; | |
| 1494 | + | |
| 1495 | + if (empty($api_key) || empty($host)) { | |
| 1496 | + return array( | |
| 1497 | + 'success' => false, | |
| 1498 | + 'message' => 'Missing Pinecone API credentials' | |
| 1499 | + ); | |
| 1500 | + } | |
| 1501 | + | |
| 1502 | + // Document index (plan 362c31): one delete_all (or one filtered delete) call. | |
| 1503 | + if (class_exists('MxChat_Pinecone_Documents') && MxChat_Pinecone_Documents::is_document_options($pinecone_options)) { | |
| 1504 | + $doc_filter = !empty($content_type_filter) ? array('type' => array('$eq' => (string) $content_type_filter)) : null; | |
| 1505 | + $doc_result = MxChat_Pinecone_Documents::delete_all(MxChat_Pinecone_Documents::cfg_from_options($pinecone_options), $doc_filter); | |
| 1506 | + if (is_wp_error($doc_result)) { | |
| 1507 | + return array('success' => false, 'message' => $doc_result->get_error_message()); | |
| 1508 | + } | |
| 1509 | + return array('success' => true, 'message' => $doc_filter ? 'Deleted all matching documents from the Pinecone document index' : 'Deleted all documents from the Pinecone document index'); | |
| 1510 | + } | |
| 1511 | + | |
| 1512 | + try { | |
| 1513 | + $total_deleted = 0; | |
| 1514 | + $failed_batches = 0; | |
| 1515 | + $max_iterations = 100; // Safety limit to prevent infinite loops | |
| 1516 | + $iteration = 0; | |
| 1517 | + | |
| 1518 | + // Loop until no more vectors are found | |
| 1519 | + do { | |
| 1520 | + $iteration++; | |
| 1521 | + | |
| 1522 | + // Get a batch of vector IDs from Pinecone | |
| 1523 | + $records = $this->mxchat_get_recent_1k_entries($pinecone_options); | |
| 1524 | + $vector_ids = array(); | |
| 1525 | + | |
| 1526 | + foreach ($records as $record) { | |
| 1527 | + if (!empty($record->id)) { | |
| 1528 | + // If content type filter is active, only include matching records | |
| 1529 | + if (!empty($content_type_filter)) { | |
| 1530 | + $record_type = isset($record->type) ? $record->type : ''; | |
| 1531 | + if ($record_type !== $content_type_filter) { | |
| 1532 | + continue; | |
| 1533 | + } | |
| 1534 | + } | |
| 1535 | + $vector_ids[] = $record->id; | |
| 1536 | + } | |
| 1537 | + } | |
| 1538 | + | |
| 1539 | + // If no matching vectors found, we're done | |
| 1540 | + // (either no records at all, or all remaining records are non-matching types) | |
| 1541 | + if (empty($vector_ids)) { | |
| 1542 | + break; | |
| 1543 | + } | |
| 1544 | + | |
| 1545 | + // Delete vectors in batches (Pinecone has limits on batch operations) | |
| 1546 | + $batch_size = 100; | |
| 1547 | + $batches = array_chunk($vector_ids, $batch_size); | |
| 1548 | + | |
| 1549 | + foreach ($batches as $batch) { | |
| 1550 | + $result = $this->mxchat_delete_pinecone_batch($batch, $api_key, $host, $namespace); | |
| 1551 | + if ($result['success']) { | |
| 1552 | + $total_deleted += count($batch); | |
| 1553 | + } else { | |
| 1554 | + $failed_batches++; | |
| 1555 | + } | |
| 1556 | + } | |
| 1557 | + | |
| 1558 | + // Small delay to avoid rate limiting | |
| 1559 | + usleep(100000); // 100ms | |
| 1560 | + | |
| 1561 | + } while ($iteration < $max_iterations); | |
| 1562 | + | |
| 1563 | + if ($total_deleted === 0) { | |
| 1564 | + return array( | |
| 1565 | + 'success' => true, | |
| 1566 | + 'message' => 'No vectors found to delete' | |
| 1567 | + ); | |
| 1568 | + } | |
| 1569 | + | |
| 1570 | + if ($failed_batches > 0) { | |
| 1571 | + return array( | |
| 1572 | + 'success' => false, | |
| 1573 | + 'message' => sprintf('Deleted %d vectors, but %d batches failed', $total_deleted, $failed_batches) | |
| 1574 | + ); | |
| 1575 | + } | |
| 1576 | + | |
| 1577 | + return array( | |
| 1578 | + 'success' => true, | |
| 1579 | + 'message' => "Successfully deleted {$total_deleted} vectors from Pinecone" | |
| 1580 | + ); | |
| 1581 | + | |
| 1582 | + } catch (Exception $e) { | |
| 1583 | + return array( | |
| 1584 | + 'success' => false, | |
| 1585 | + 'message' => $e->getMessage() | |
| 1586 | + ); | |
| 1587 | + } | |
| 1588 | +} | |
| 1589 | + | |
| 1590 | + /** | |
| 1591 | + * Deletes batch of vectors from Pinecone database | |
| 1592 | + */ | |
| 1593 | + public function mxchat_delete_pinecone_batch($vector_ids, $api_key, $host, $namespace = '') { | |
| 1594 | + // Document index (plan 362c31): documents/delete by ids. | |
| 1595 | + if (class_exists('MxChat_Pinecone_Documents') && MxChat_Pinecone_Documents::is_document_host($host)) { | |
| 1596 | + $doc_result = MxChat_Pinecone_Documents::delete_ids((array) $vector_ids, array('api_key' => $api_key, 'host' => $host, 'namespace' => $namespace)); | |
| 1597 | + if (is_wp_error($doc_result)) { | |
| 1598 | + MxChat_Admin::mxchat_log_debug('pinecone_error', 'Pinecone document batch deletion failed: ' . $doc_result->get_error_message()); | |
| 1599 | + return array('success' => false, 'message' => $doc_result->get_error_message()); | |
| 1600 | + } | |
| 1601 | + return array('success' => true, 'message' => 'Batch deleted successfully from Pinecone'); | |
| 1602 | + } | |
| 1603 | + | |
| 1604 | + // Build the API endpoint | |
| 1605 | + $api_endpoint = "https://{$host}/vectors/delete"; | |
| 1606 | + | |
| 1607 | + // Prepare the request body with the IDs | |
| 1608 | + $request_body = array( | |
| 1609 | + 'ids' => $vector_ids | |
| 1610 | + ); | |
| 1611 | + | |
| 1612 | + // Add namespace if provided — omitting it targets the default namespace | |
| 1613 | + if (!empty($namespace)) { | |
| 1614 | + $request_body['namespace'] = $namespace; | |
| 1615 | + } | |
| 1616 | + | |
| 1617 | + // Make the deletion request | |
| 1618 | + $response = wp_remote_post($api_endpoint, array( | |
| 1619 | + 'headers' => array( | |
| 1620 | + 'Api-Key' => $api_key, | |
| 1621 | + 'accept' => 'application/json', | |
| 1622 | + 'content-type' => 'application/json' | |
| 1623 | + ), | |
| 1624 | + 'body' => wp_json_encode($request_body), | |
| 1625 | + 'timeout' => 60, // Increased timeout for batch operations | |
| 1626 | + 'method' => 'POST' | |
| 1627 | + )); | |
| 1628 | + | |
| 1629 | + // Handle WordPress HTTP API errors | |
| 1630 | + if (is_wp_error($response)) { | |
| 1631 | + MxChat_Admin::mxchat_log_debug('pinecone_error', 'Pinecone batch deletion failed: ' . $response->get_error_message()); | |
| 1632 | + return array( | |
| 1633 | + 'success' => false, | |
| 1634 | + 'message' => $response->get_error_message() | |
| 1635 | + ); | |
| 1636 | + } | |
| 1637 | + | |
| 1638 | + // Check response status | |
| 1639 | + $response_code = wp_remote_retrieve_response_code($response); | |
| 1640 | + $response_body = wp_remote_retrieve_body($response); | |
| 1641 | + | |
| 1642 | + // Pinecone returns 200 for successful deletion | |
| 1643 | + if ($response_code !== 200) { | |
| 1644 | + MxChat_Admin::mxchat_log_debug('pinecone_error', 'Pinecone batch deletion failed (HTTP ' . $response_code . ')', array('response' => substr($response_body, 0, 200))); | |
| 1645 | + return array( | |
| 1646 | + 'success' => false, | |
| 1647 | + 'message' => sprintf( | |
| 1648 | + 'Pinecone API error (HTTP %d): %s', | |
| 1649 | + $response_code, | |
| 1650 | + $response_body | |
| 1651 | + ) | |
| 1652 | + ); | |
| 1653 | + } | |
| 1654 | + | |
| 1655 | + return array( | |
| 1656 | + 'success' => true, | |
| 1657 | + 'message' => 'Batch deleted successfully from Pinecone' | |
| 1658 | + ); | |
| 1659 | + } | |
| 1660 | + | |
| 1661 | + | |
| 1662 | +/** | |
| 1663 | + * Deletes vector from Pinecone using API request | |
| 1664 | + */ | |
| 1665 | +public function mxchat_delete_from_pinecone_by_vector_id($vector_id, $api_key, $host, $namespace = '') { | |
| 1666 | + //error_log('=== PINECONE DELETE OPERATION ==='); | |
| 1667 | + //error_log('Vector ID: ' . $vector_id); | |
| 1668 | + //error_log('Host: ' . $host); | |
| 1669 | + //error_log('API Key: ' . (empty($api_key) ? 'EMPTY' : 'SET')); | |
| 1670 | + | |
| 1671 | + // Document index (plan 362c31): documents/delete by id. | |
| 1672 | + if (class_exists('MxChat_Pinecone_Documents') && MxChat_Pinecone_Documents::is_document_host($host)) { | |
| 1673 | + $doc_result = MxChat_Pinecone_Documents::delete_ids(array($vector_id), array('api_key' => $api_key, 'host' => $host, 'namespace' => $namespace)); | |
| 1674 | + if (is_wp_error($doc_result)) { | |
| 1675 | + return array('success' => false, 'message' => $doc_result->get_error_message()); | |
| 1676 | + } | |
| 1677 | + return array('success' => true, 'message' => 'Vector deleted successfully from Pinecone'); | |
| 1678 | + } | |
| 1679 | + | |
| 1680 | + // First, let's verify the vector exists before trying to delete | |
| 1681 | + $fetch_url = "https://{$host}/vectors/fetch"; | |
| 1682 | + | |
| 1683 | + $fetch_params = array( | |
| 1684 | + 'ids' => array($vector_id) | |
| 1685 | + ); | |
| 1686 | + | |
| 1687 | + // Add namespace if provided (though you said you're not using namespaces) | |
| 1688 | + if (!empty($namespace)) { | |
| 1689 | + $fetch_params['namespace'] = $namespace; | |
| 1690 | + } | |
| 1691 | + | |
| 1692 | + // Construct URL with query parameters for GET request | |
| 1693 | + $fetch_url_with_params = $fetch_url . '?' . http_build_query($fetch_params); | |
| 1694 | + | |
| 1695 | + $fetch_response = wp_remote_get($fetch_url_with_params, array( | |
| 1696 | + 'headers' => array( | |
| 1697 | + 'Api-Key' => $api_key, | |
| 1698 | + 'accept' => 'application/json' | |
| 1699 | + ), | |
| 1700 | + 'timeout' => 15 | |
| 1701 | + )); | |
| 1702 | + | |
| 1703 | + if (!is_wp_error($fetch_response) && wp_remote_retrieve_response_code($fetch_response) === 200) { | |
| 1704 | + $fetch_body = wp_remote_retrieve_body($fetch_response); | |
| 1705 | + $fetch_data = json_decode($fetch_body, true); | |
| 1706 | + | |
| 1707 | + //error_log('DEBUG: Fetch response: ' . print_r($fetch_data, true)); | |
| 1708 | + | |
| 1709 | + if (isset($fetch_data['vectors']) && isset($fetch_data['vectors'][$vector_id])) { | |
| 1710 | + //error_log('DEBUG: Vector EXISTS in this index before deletion'); | |
| 1711 | + } else { | |
| 1712 | + //error_log('WARNING: Vector NOT FOUND in this index! It may be in a different bot\'s index'); | |
| 1713 | + // You might want to return an error here | |
| 1714 | + } | |
| 1715 | + } else { | |
| 1716 | + //error_log('DEBUG: Could not fetch vector to verify existence'); | |
| 1717 | + } | |
| 1718 | + | |
| 1719 | + // Now proceed with deletion | |
| 1720 | + $api_endpoint = "https://{$host}/vectors/delete"; | |
| 1721 | + | |
| 1722 | + // Prepare the request body with the ID | |
| 1723 | + $request_body = array( | |
| 1724 | + 'ids' => array($vector_id) | |
| 1725 | + ); | |
| 1726 | + | |
| 1727 | + // Add namespace if provided | |
| 1728 | + if (!empty($namespace)) { | |
| 1729 | + $request_body['namespace'] = $namespace; | |
| 1730 | + } | |
| 1731 | + | |
| 1732 | + //error_log('DEBUG: Delete request body: ' . json_encode($request_body)); | |
| 1733 | + //error_log('DEBUG: Delete endpoint: ' . $api_endpoint); | |
| 1734 | + | |
| 1735 | + // Make the deletion request | |
| 1736 | + $response = wp_remote_post($api_endpoint, array( | |
| 1737 | + 'headers' => array( | |
| 1738 | + 'Api-Key' => $api_key, | |
| 1739 | + 'accept' => 'application/json', | |
| 1740 | + 'content-type' => 'application/json' | |
| 1741 | + ), | |
| 1742 | + 'body' => wp_json_encode($request_body), | |
| 1743 | + 'timeout' => 30 | |
| 1744 | + )); | |
| 1745 | + | |
| 1746 | + // Handle WordPress HTTP API errors | |
| 1747 | + if (is_wp_error($response)) { | |
| 1748 | + //error_log('DEBUG: WP Error: ' . $response->get_error_message()); | |
| 1749 | + return array( | |
| 1750 | + 'success' => false, | |
| 1751 | + 'message' => $response->get_error_message() | |
| 1752 | + ); | |
| 1753 | + } | |
| 1754 | + | |
| 1755 | + // Check response status | |
| 1756 | + $response_code = wp_remote_retrieve_response_code($response); | |
| 1757 | + $response_body = wp_remote_retrieve_body($response); | |
| 1758 | + | |
| 1759 | + //error_log('DEBUG: Delete response code: ' . $response_code); | |
| 1760 | + //error_log('DEBUG: Delete response body: ' . $response_body); | |
| 1761 | + | |
| 1762 | + // Pinecone returns 200 for successful deletion (even if vector didn't exist) | |
| 1763 | + if ($response_code !== 200) { | |
| 1764 | + //error_log('DEBUG: Non-200 response from Pinecone'); | |
| 1765 | + return array( | |
| 1766 | + 'success' => false, | |
| 1767 | + 'message' => sprintf( | |
| 1768 | + 'Pinecone API error (HTTP %d): %s', | |
| 1769 | + $response_code, | |
| 1770 | + $response_body | |
| 1771 | + ) | |
| 1772 | + ); | |
| 1773 | + } | |
| 1774 | + | |
| 1775 | + // After deletion, verify it's actually gone | |
| 1776 | + sleep(1); // Give Pinecone a moment to process | |
| 1777 | + | |
| 1778 | + $verify_response = wp_remote_get($fetch_url_with_params, array( | |
| 1779 | + 'headers' => array( | |
| 1780 | + 'Api-Key' => $api_key, | |
| 1781 | + 'accept' => 'application/json' | |
| 1782 | + ), | |
| 1783 | + 'timeout' => 15 | |
| 1784 | + )); | |
| 1785 | + | |
| 1786 | + if (!is_wp_error($verify_response) && wp_remote_retrieve_response_code($verify_response) === 200) { | |
| 1787 | + $verify_body = wp_remote_retrieve_body($verify_response); | |
| 1788 | + $verify_data = json_decode($verify_body, true); | |
| 1789 | + | |
| 1790 | + if (isset($verify_data['vectors']) && isset($verify_data['vectors'][$vector_id])) { | |
| 1791 | + //error_log('ERROR: Vector STILL EXISTS after deletion attempt!'); | |
| 1792 | + return array( | |
| 1793 | + 'success' => false, | |
| 1794 | + 'message' => 'Vector still exists after deletion attempt' | |
| 1795 | + ); | |
| 1796 | + } else { | |
| 1797 | + //error_log('SUCCESS: Vector confirmed deleted (or never existed)'); | |
| 1798 | + } | |
| 1799 | + } | |
| 1800 | + | |
| 1801 | + //error_log('=== END PINECONE DELETE OPERATION ==='); | |
| 1802 | + | |
| 1803 | + return array( | |
| 1804 | + 'success' => true, | |
| 1805 | + 'message' => 'Vector deleted successfully from Pinecone' | |
| 1806 | + ); | |
| 1807 | +} | |
| 1808 | + | |
| 1809 | + /** | |
| 1810 | + * Deletes data from Pinecone index using API key | |
| 1811 | + */ | |
| 1812 | + private function mxchat_delete_from_pinecone($urls, $api_key, $environment, $index_name) { | |
| 1813 | + // Get the Pinecone host from options (matching your store_in_pinecone_main pattern) | |
| 1814 | + $options = get_option('mxchat_pinecone_addon_options'); | |
| 1815 | + $host = $options['mxchat_pinecone_host'] ?? ''; | |
| 1816 | + | |
| 1817 | + if (empty($host)) { | |
| 1818 | + return array( | |
| 1819 | + 'success' => false, | |
| 1820 | + 'message' => 'Pinecone host is not configured. Please set the host in your settings.' | |
| 1821 | + ); | |
| 1822 | + } | |
| 1823 | + | |
| 1824 | + // Document index (plan 362c31): documents/delete by the same md5(url) ids. | |
| 1825 | + if (class_exists('MxChat_Pinecone_Documents') && MxChat_Pinecone_Documents::is_document_options($options)) { | |
| 1826 | + $doc_result = MxChat_Pinecone_Documents::delete_ids(array_map('md5', (array) $urls), MxChat_Pinecone_Documents::cfg_from_options($options)); | |
| 1827 | + if (is_wp_error($doc_result)) { | |
| 1828 | + return array('success' => false, 'message' => $doc_result->get_error_message()); | |
| 1829 | + } | |
| 1830 | + return array('success' => true, 'message' => sprintf('Successfully deleted %d vectors from Pinecone', count((array) $urls))); | |
| 1831 | + } | |
| 1832 | + | |
| 1833 | + // Build API endpoint using the configured host | |
| 1834 | + $api_endpoint = "https://{$host}/vectors/delete"; | |
| 1835 | + | |
| 1836 | + // Create vector IDs from URLs (matching your store method's ID generation) | |
| 1837 | + $vector_ids = array_map('md5', $urls); | |
| 1838 | + | |
| 1839 | + // Prepare the delete request body | |
| 1840 | + $request_body = array( | |
| 1841 | + 'ids' => $vector_ids, | |
| 1842 | + 'filter' => array( | |
| 1843 | + 'source_url' => array( | |
| 1844 | + '$in' => $urls | |
| 1845 | + ) | |
| 1846 | + ) | |
| 1847 | + ); | |
| 1848 | + | |
| 1849 | + // Make the deletion request | |
| 1850 | + $response = wp_remote_post($api_endpoint, array( | |
| 1851 | + 'headers' => array( | |
| 1852 | + 'Api-Key' => $api_key, | |
| 1853 | + 'accept' => 'application/json', | |
| 1854 | + 'content-type' => 'application/json' | |
| 1855 | + ), | |
| 1856 | + 'body' => wp_json_encode($request_body), | |
| 1857 | + 'timeout' => 30, | |
| 1858 | + 'data_format' => 'body' | |
| 1859 | + )); | |
| 1860 | + | |
| 1861 | + // Handle WordPress HTTP API errors | |
| 1862 | + if (is_wp_error($response)) { | |
| 1863 | + return array( | |
| 1864 | + 'success' => false, | |
| 1865 | + 'message' => $response->get_error_message() | |
| 1866 | + ); | |
| 1867 | + } | |
| 1868 | + | |
| 1869 | + // Check response status | |
| 1870 | + $response_code = wp_remote_retrieve_response_code($response); | |
| 1871 | + if ($response_code !== 200) { | |
| 1872 | + $body = wp_remote_retrieve_body($response); | |
| 1873 | + return array( | |
| 1874 | + 'success' => false, | |
| 1875 | + 'message' => sprintf( | |
| 1876 | + 'Pinecone API error (HTTP %d): %s', | |
| 1877 | + $response_code, | |
| 1878 | + $body | |
| 1879 | + ) | |
| 1880 | + ); | |
| 1881 | + } | |
| 1882 | + | |
| 1883 | + // Parse response body | |
| 1884 | + $body = wp_remote_retrieve_body($response); | |
| 1885 | + $response_data = json_decode($body, true); | |
| 1886 | + | |
| 1887 | + // Final validation of the response | |
| 1888 | + if (json_last_error() !== JSON_ERROR_NONE) { | |
| 1889 | + return array( | |
| 1890 | + 'success' => false, | |
| 1891 | + 'message' => 'Failed to parse Pinecone response: ' . json_last_error_msg() | |
| 1892 | + ); | |
| 1893 | + } | |
| 1894 | + | |
| 1895 | + return array( | |
| 1896 | + 'success' => true, | |
| 1897 | + 'message' => sprintf('Successfully deleted %d vectors from Pinecone', count($vector_ids)) | |
| 1898 | + ); | |
| 1899 | + } | |
| 1900 | + | |
| 1901 | + | |
| 1902 | + | |
| 1903 | +/** | |
| 1904 | + * Retrieves processed content from Pinecone API | |
| 1905 | + * | |
| 1906 | + * @param array $pinecone_options Pinecone configuration options | |
| 1907 | + * @param array $post_ids Optional array of specific post IDs to check (if empty, checks all) | |
| 1908 | + * @return array Processed data keyed by post ID | |
| 1909 | + */ | |
| 1910 | +public function mxchat_get_pinecone_processed_content($pinecone_options, $post_ids = array()) { | |
| 1911 | + $api_key = $pinecone_options['mxchat_pinecone_api_key'] ?? ''; | |
| 1912 | + $host = $pinecone_options['mxchat_pinecone_host'] ?? ''; | |
| 1913 | + | |
| 1914 | + if (empty($api_key) || empty($host)) { | |
| 1915 | + return array(); | |
| 1916 | + } | |
| 1917 | + | |
| 1918 | + $pinecone_data = array(); | |
| 1919 | + | |
| 1920 | + try { | |
| 1921 | + // Always get fresh data from Pinecone | |
| 1922 | + $pinecone_data = $this->mxchat_scan_pinecone_for_processed_content($pinecone_options, $post_ids); | |
| 1923 | + | |
| 1924 | + // Method 2: Final fallback - try stats endpoint (if available) | |
| 1925 | + if (empty($pinecone_data)) { | |
| 1926 | + $stats_url = "https://{$host}/describe_index_stats"; | |
| 1927 | + | |
| 1928 | + $response = wp_remote_post($stats_url, array( | |
| 1929 | + 'headers' => array( | |
| 1930 | + 'Api-Key' => $api_key, | |
| 1931 | + 'Content-Type' => 'application/json' | |
| 1932 | + ), | |
| 1933 | + 'body' => json_encode(array()), | |
| 1934 | + 'timeout' => 30 | |
| 1935 | + )); | |
| 1936 | + | |
| 1937 | + if (!is_wp_error($response) && wp_remote_retrieve_response_code($response) === 200) { | |
| 1938 | + $body = wp_remote_retrieve_body($response); | |
| 1939 | + $stats_data = json_decode($body, true); | |
| 1940 | + } | |
| 1941 | + } | |
| 1942 | + | |
| 1943 | + } catch (Exception $e) { | |
| 1944 | + // Log error but don't return cached data | |
| 1945 | + } | |
| 1946 | + | |
| 1947 | + return $pinecone_data; | |
| 1948 | +} | |
| 1949 | + // ======================================== | |
| 1950 | + // HELPER METHODS | |
| 1951 | + // ======================================== | |
| 1952 | + | |
| 1953 | + /** | |
| 1954 | + * Validates Pinecone API credentials | |
| 1955 | + */ | |
| 1956 | + private function mxchat_validate_pinecone_credentials($api_key, $host) { | |
| 1957 | + if (empty($api_key) || empty($host)) { | |
| 1958 | + return false; | |
| 1959 | + } | |
| 1960 | + return true; | |
| 1961 | + } | |
| 1962 | + | |
| 1963 | + /** | |
| 1964 | + * Get Pinecone API credentials from options | |
| 1965 | + */ | |
| 1966 | + private function mxchat_get_pinecone_credentials() { | |
| 1967 | + $options = get_option('mxchat_options', array()); | |
| 1968 | + return array( | |
| 1969 | + 'api_key' => isset($options['pinecone_api_key']) ? $options['pinecone_api_key'] : '', | |
| 1970 | + 'host' => isset($options['pinecone_host']) ? $options['pinecone_host'] : '' | |
| 1971 | + ); | |
| 1972 | + } | |
| 1973 | + | |
| 1974 | + /** | |
| 1975 | + * Log Pinecone operation errors | |
| 1976 | + */ | |
| 1977 | + private function log_pinecone_error($operation, $error_message) { | |
| 1978 | + //error_log("MxChat Pinecone {$operation} Error: " . $error_message); | |
| 1979 | + } | |
| 1980 | + | |
| 1981 | + // ======================================== | |
| 1982 | + // STATIC ACCESS METHODS (for backward compatibility) | |
| 1983 | + // ======================================== | |
| 1984 | + | |
| 1985 | + /** | |
| 1986 | + * Get singleton instance | |
| 1987 | + */ | |
| 1988 | + public static function get_instance() { | |
| 1989 | + static $instance = null; | |
| 1990 | + if ($instance === null) { | |
| 1991 | + $instance = new self(); | |
| 1992 | + } | |
| 1993 | + return $instance; | |
| 1994 | + } | |
| 1995 | +} | |
| 1996 | + | |
| 1997 | +// Initialize the Pinecone manager | |
| 1998 | +$mxchat_pinecone_manager = MxChat_Pinecone_Manager::get_instance(); | |
