| 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 |
// NOTE (plan 793b82): /vectors/list is a GET endpoint with query parameters. |
| 498 |
// A POST is answered 200-with-an-empty-body, which reads as "no vectors" — |
| 499 |
// this call used to POST and only ever "worked" because the empty result |
| 500 |
// tripped the query-based fallback below. |
| 501 |
$list_url = "https://{$host}/vectors/list"; |
| 502 |
|
| 503 |
// Calculate pagination token from page number |
| 504 |
// Pinecone uses cursor-based pagination, so we need to handle this differently |
| 505 |
$limit = $per_page * 2; // Fetch extra to account for filtering |
| 506 |
|
| 507 |
$list_params = array( |
| 508 |
'limit' => $limit |
| 509 |
); |
| 510 |
|
| 511 |
if (!empty($namespace)) { |
| 512 |
$list_params['namespace'] = $namespace; |
| 513 |
} |
| 514 |
|
| 515 |
// For pages beyond first, we need to use pagination_token |
| 516 |
// Store/retrieve pagination tokens in transients |
| 517 |
$pagination_key = 'mxchat_pinecone_page_' . md5($host . $namespace . $content_type); |
| 518 |
|
| 519 |
if ($page > 1) { |
| 520 |
$stored_tokens = get_transient($pagination_key); |
| 521 |
if ($stored_tokens && isset($stored_tokens[$page])) { |
| 522 |
$list_params['paginationToken'] = $stored_tokens[$page]; |
| 523 |
} |
| 524 |
} |
| 525 |
|
| 526 |
$response = wp_remote_get($list_url . '?' . http_build_query($list_params), array( |
| 527 |
'headers' => array( |
| 528 |
'Api-Key' => $api_key, |
| 529 |
'accept' => 'application/json' |
| 530 |
), |
| 531 |
'timeout' => 15 |
| 532 |
)); |
| 533 |
|
| 534 |
if (is_wp_error($response) || wp_remote_retrieve_response_code($response) !== 200) { |
| 535 |
// Genuine transport/API failure — fall back to the query-based approach, |
| 536 |
// but say so: a silent fallback is what masked the POST bug for a year. |
| 537 |
error_log('MxChat Pinecone: /vectors/list GET failed (' . (is_wp_error($response) ? $response->get_error_message() : wp_remote_retrieve_response_code($response)) . ') — falling back to query-based listing.'); |
| 538 |
return $this->mxchat_query_based_list($pinecone_options, $page, $per_page, $bot_id, $content_type); |
| 539 |
} |
| 540 |
|
| 541 |
$body = wp_remote_retrieve_body($response); |
| 542 |
$data = json_decode($body, true); |
| 543 |
|
| 544 |
// Store pagination token for next page |
| 545 |
if (!empty($data['pagination']['next'])) { |
| 546 |
$stored_tokens = get_transient($pagination_key) ?: array(); |
| 547 |
$stored_tokens[$page + 1] = $data['pagination']['next']; |
| 548 |
set_transient($pagination_key, $stored_tokens, 300); // 5 minute cache |
| 549 |
} |
| 550 |
|
| 551 |
$vector_ids = array(); |
| 552 |
if (isset($data['vectors'])) { |
| 553 |
foreach ($data['vectors'] as $vector) { |
| 554 |
$vector_ids[] = $vector['id']; |
| 555 |
} |
| 556 |
} |
| 557 |
|
| 558 |
// A 200 with no vector ids is a REAL answer now (the namespace is empty), |
| 559 |
// not the failure signature it was under POST — return the empty listing |
| 560 |
// rather than routing into the fallback and masking a broken primary. |
| 561 |
if (empty($vector_ids)) { |
| 562 |
return array('data' => array(), 'total' => 0); |
| 563 |
} |
| 564 |
|
| 565 |
// Fetch metadata for these vector IDs |
| 566 |
$fetched = $this->mxchat_fetch_vectors_by_ids_for_list($pinecone_options, $vector_ids, $page, $per_page, $bot_id, $content_type); |
| 567 |
|
| 568 |
// null = the fetch itself failed in transport (distinct from "ids had no |
| 569 |
// metadata matches") — log and fall back so the screen still lists. |
| 570 |
if ($fetched === null) { |
| 571 |
error_log('MxChat Pinecone: /vectors/fetch GET failed for the listing — falling back to query-based listing.'); |
| 572 |
return $this->mxchat_query_based_list($pinecone_options, $page, $per_page, $bot_id, $content_type); |
| 573 |
} |
| 574 |
|
| 575 |
return $fetched; |
| 576 |
} |
| 577 |
|
| 578 |
/** |
| 579 |
* Query-based listing fallback when list endpoint fails |
| 580 |
* ADDED 2.6.1 |
| 581 |
*/ |
| 582 |
private function mxchat_query_based_list($pinecone_options, $page, $per_page, $bot_id, $content_type) { |
| 583 |
$api_key = $pinecone_options['mxchat_pinecone_api_key'] ?? ''; |
| 584 |
$host = $pinecone_options['mxchat_pinecone_host'] ?? ''; |
| 585 |
$namespace = $pinecone_options['mxchat_pinecone_namespace'] ?? ''; |
| 586 |
|
| 587 |
$query_url = "https://{$host}/query"; |
| 588 |
$query_vector = $this->mxchat_generate_optimized_query_vector(); |
| 589 |
|
| 590 |
// Fetch records - use higher limit to cover large databases |
| 591 |
// Pinecone query API supports up to 10,000 topK |
| 592 |
// We fetch more than needed to get accurate total count and enable pagination |
| 593 |
$fetch_limit = 5000; |
| 594 |
|
| 595 |
$query_data = array( |
| 596 |
'includeMetadata' => true, |
| 597 |
'includeValues' => false, |
| 598 |
'topK' => $fetch_limit, |
| 599 |
'vector' => $query_vector |
| 600 |
); |
| 601 |
|
| 602 |
if (!empty($namespace)) { |
| 603 |
$query_data['namespace'] = $namespace; |
| 604 |
} |
| 605 |
|
| 606 |
// Add content type filter if specified |
| 607 |
if (!empty($content_type)) { |
| 608 |
$query_data['filter'] = array( |
| 609 |
'type' => array('$eq' => $content_type) |
| 610 |
); |
| 611 |
} |
| 612 |
|
| 613 |
$response = wp_remote_post($query_url, array( |
| 614 |
'headers' => array( |
| 615 |
'Api-Key' => $api_key, |
| 616 |
'Content-Type' => 'application/json' |
| 617 |
), |
| 618 |
'body' => json_encode($query_data), |
| 619 |
'timeout' => 30 |
| 620 |
)); |
| 621 |
|
| 622 |
if (is_wp_error($response) || wp_remote_retrieve_response_code($response) !== 200) { |
| 623 |
return array('data' => array(), 'total' => 0); |
| 624 |
} |
| 625 |
|
| 626 |
$body = wp_remote_retrieve_body($response); |
| 627 |
$data = json_decode($body, true); |
| 628 |
|
| 629 |
$records = array(); |
| 630 |
if (isset($data['matches'])) { |
| 631 |
foreach ($data['matches'] as $match) { |
| 632 |
$metadata = $match['metadata'] ?? array(); |
| 633 |
|
| 634 |
$created_at = $metadata['created_at'] ?? $metadata['last_updated'] ?? $metadata['timestamp'] ?? time(); |
| 635 |
if (!is_numeric($created_at)) { |
| 636 |
$created_at = strtotime($created_at) ?: time(); |
| 637 |
} |
| 638 |
|
| 639 |
$records[] = (object) array( |
| 640 |
'id' => $match['id'] ?? '', |
| 641 |
'article_content' => $metadata['text'] ?? '', |
| 642 |
'source_url' => $metadata['source_url'] ?? '', |
| 643 |
'role_restriction' => $metadata['role_restriction'] ?? 'public', |
| 644 |
'type' => $metadata['type'] ?? 'content', |
| 645 |
'bot_id' => $bot_id, |
| 646 |
'created_at' => $created_at, |
| 647 |
'data_source' => 'pinecone', |
| 648 |
'chunk_index' => isset($metadata['chunk_index']) ? intval($metadata['chunk_index']) : null, |
| 649 |
'total_chunks' => isset($metadata['total_chunks']) ? intval($metadata['total_chunks']) : null, |
| 650 |
'is_chunked' => isset($metadata['is_chunked']) ? (bool) $metadata['is_chunked'] : false |
| 651 |
); |
| 652 |
} |
| 653 |
} |
| 654 |
|
| 655 |
// Sort by created_at (newest first) |
| 656 |
usort($records, function($a, $b) { |
| 657 |
return $b->created_at - $a->created_at; |
| 658 |
}); |
| 659 |
|
| 660 |
$total = count($records); |
| 661 |
$offset = ($page - 1) * $per_page; |
| 662 |
$paged_records = array_slice($records, $offset, $per_page); |
| 663 |
|
| 664 |
$this->mxchat_batch_fetch_role_restrictions($paged_records, $bot_id); |
| 665 |
|
| 666 |
return array( |
| 667 |
'data' => $paged_records, |
| 668 |
'total' => $total |
| 669 |
); |
| 670 |
} |
| 671 |
|
| 672 |
/** |
| 673 |
* Fetch specific vectors by their IDs and format for display |
| 674 |
* ADDED 2.6.1 |
| 675 |
*/ |
| 676 |
private function mxchat_fetch_vectors_by_ids_for_list($pinecone_options, $vector_ids, $page, $per_page, $bot_id, $content_type) { |
| 677 |
$api_key = $pinecone_options['mxchat_pinecone_api_key'] ?? ''; |
| 678 |
$host = $pinecone_options['mxchat_pinecone_host'] ?? ''; |
| 679 |
$namespace = $pinecone_options['mxchat_pinecone_namespace'] ?? ''; |
| 680 |
|
| 681 |
// NOTE (plan 793b82): /vectors/fetch is a GET endpoint, and Pinecone expects |
| 682 |
// the ids repeated (ids=a&ids=b) — http_build_query would emit ids[0]=a. |
| 683 |
// A POST here is answered 200-with-an-empty-body, which reads as "no vectors". |
| 684 |
$fetch_query = array(); |
| 685 |
foreach ($vector_ids as $fetch_vid) { |
| 686 |
$fetch_query[] = 'ids=' . rawurlencode($fetch_vid); |
| 687 |
} |
| 688 |
if (!empty($namespace)) { |
| 689 |
$fetch_query[] = 'namespace=' . rawurlencode($namespace); |
| 690 |
} |
| 691 |
|
| 692 |
$response = wp_remote_get("https://{$host}/vectors/fetch?" . implode('&', $fetch_query), array( |
| 693 |
'headers' => array( |
| 694 |
'Api-Key' => $api_key, |
| 695 |
'accept' => 'application/json' |
| 696 |
), |
| 697 |
'timeout' => 15 |
| 698 |
)); |
| 699 |
|
| 700 |
// Transport/API failure is distinct from "nothing matched": return null so |
| 701 |
// the caller can tell the difference (plan 793b82's empty-vs-failed rule). |
| 702 |
if (is_wp_error($response) || wp_remote_retrieve_response_code($response) !== 200) { |
| 703 |
return null; |
| 704 |
} |
| 705 |
|
| 706 |
$body = wp_remote_retrieve_body($response); |
| 707 |
$data = json_decode($body, true); |
| 708 |
|
| 709 |
$records = array(); |
| 710 |
if (isset($data['vectors'])) { |
| 711 |
foreach ($data['vectors'] as $vector_id => $vector_data) { |
| 712 |
$metadata = $vector_data['metadata'] ?? array(); |
| 713 |
|
| 714 |
// Filter by content type if specified |
| 715 |
if (!empty($content_type)) { |
| 716 |
$record_type = $metadata['type'] ?? 'content'; |
| 717 |
if ($record_type !== $content_type) { |
| 718 |
continue; |
| 719 |
} |
| 720 |
} |
| 721 |
|
| 722 |
$created_at = $metadata['created_at'] ?? $metadata['last_updated'] ?? $metadata['timestamp'] ?? time(); |
| 723 |
if (!is_numeric($created_at)) { |
| 724 |
$created_at = strtotime($created_at) ?: time(); |
| 725 |
} |
| 726 |
|
| 727 |
$records[] = (object) array( |
| 728 |
'id' => $vector_id, |
| 729 |
'article_content' => $metadata['text'] ?? '', |
| 730 |
'source_url' => $metadata['source_url'] ?? '', |
| 731 |
'role_restriction' => $metadata['role_restriction'] ?? 'public', |
| 732 |
'type' => $metadata['type'] ?? 'content', |
| 733 |
'bot_id' => $bot_id, |
| 734 |
'created_at' => $created_at, |
| 735 |
'data_source' => 'pinecone', |
| 736 |
'chunk_index' => isset($metadata['chunk_index']) ? intval($metadata['chunk_index']) : null, |
| 737 |
'total_chunks' => isset($metadata['total_chunks']) ? intval($metadata['total_chunks']) : null, |
| 738 |
'is_chunked' => isset($metadata['is_chunked']) ? (bool) $metadata['is_chunked'] : false |
| 739 |
); |
| 740 |
} |
| 741 |
} |
| 742 |
|
| 743 |
// Sort by date |
| 744 |
usort($records, function($a, $b) { |
| 745 |
return $b->created_at - $a->created_at; |
| 746 |
}); |
| 747 |
|
| 748 |
$total = count($records); |
| 749 |
$paged_records = array_slice($records, 0, $per_page); |
| 750 |
|
| 751 |
$this->mxchat_batch_fetch_role_restrictions($paged_records, $bot_id); |
| 752 |
|
| 753 |
return array( |
| 754 |
'data' => $paged_records, |
| 755 |
'total' => $total |
| 756 |
); |
| 757 |
} |
| 758 |
|
| 759 |
/** |
| 760 |
* Generate a single optimized query vector for fetching records |
| 761 |
* Uses a center-weighted approach for best coverage |
| 762 |
*/ |
| 763 |
private function mxchat_generate_optimized_query_vector() { |
| 764 |
$dimensions = $this->mxchat_get_embedding_dimensions(); |
| 765 |
$vector = array(); |
| 766 |
|
| 767 |
// Create a normalized center-weighted vector |
| 768 |
$center = $dimensions / 2; |
| 769 |
for ($i = 0; $i < $dimensions; $i++) { |
| 770 |
$distance = abs($i - $center) / $center; |
| 771 |
$vector[] = (1 - $distance) * 0.5; |
| 772 |
} |
| 773 |
|
| 774 |
// Normalize to unit length |
| 775 |
$magnitude = sqrt(array_sum(array_map(function($x) { return $x * $x; }, $vector))); |
| 776 |
if ($magnitude > 0) { |
| 777 |
$vector = array_map(function($x) use ($magnitude) { return $x / $magnitude; }, $vector); |
| 778 |
} |
| 779 |
|
| 780 |
return $vector; |
| 781 |
} |
| 782 |
|
| 783 |
/** |
| 784 |
* Batch fetch role restrictions for a set of records |
| 785 |
* Uses a single query instead of N queries |
| 786 |
* ADDED 2.6.1: Prevents N+1 query problem |
| 787 |
*/ |
| 788 |
private function mxchat_batch_fetch_role_restrictions(&$records, $bot_id = 'default') { |
| 789 |
if (empty($records)) { |
| 790 |
return; |
| 791 |
} |
| 792 |
|
| 793 |
global $wpdb; |
| 794 |
$roles_table = $wpdb->prefix . 'mxchat_pinecone_roles'; |
| 795 |
|
| 796 |
// Check if table exists |
| 797 |
$table_exists = $wpdb->get_var("SHOW TABLES LIKE '{$roles_table}'"); |
| 798 |
if (!$table_exists) { |
| 799 |
return; |
| 800 |
} |
| 801 |
|
| 802 |
// Get all vector IDs that need role lookup |
| 803 |
$vector_ids = array(); |
| 804 |
foreach ($records as $record) { |
| 805 |
if (empty($record->role_restriction) || $record->role_restriction === 'public') { |
| 806 |
$vector_ids[] = $record->id; |
| 807 |
} |
| 808 |
} |
| 809 |
|
| 810 |
if (empty($vector_ids)) { |
| 811 |
return; |
| 812 |
} |
| 813 |
|
| 814 |
// Check if bot_id column exists |
| 815 |
$columns = $wpdb->get_col("SHOW COLUMNS FROM {$roles_table}"); |
| 816 |
$has_bot_id = in_array('bot_id', $columns); |
| 817 |
|
| 818 |
// Build single query with IN clause |
| 819 |
$placeholders = implode(',', array_fill(0, count($vector_ids), '%s')); |
| 820 |
|
| 821 |
if ($has_bot_id) { |
| 822 |
$query = $wpdb->prepare( |
| 823 |
"SELECT vector_id, role_restriction FROM {$roles_table} WHERE vector_id IN ({$placeholders}) AND bot_id = %s", |
| 824 |
array_merge($vector_ids, array($bot_id)) |
| 825 |
); |
| 826 |
} else { |
| 827 |
$query = $wpdb->prepare( |
| 828 |
"SELECT vector_id, role_restriction FROM {$roles_table} WHERE vector_id IN ({$placeholders})", |
| 829 |
$vector_ids |
| 830 |
); |
| 831 |
} |
| 832 |
|
| 833 |
$results = $wpdb->get_results($query, OBJECT_K); |
| 834 |
|
| 835 |
// Apply role restrictions to records |
| 836 |
foreach ($records as &$record) { |
| 837 |
if (isset($results[$record->id])) { |
| 838 |
$record->role_restriction = $results[$record->id]->role_restriction; |
| 839 |
} |
| 840 |
} |
| 841 |
} |
| 842 |
|
| 843 |
/** |
| 844 |
* Safe version of get_recent_entries with memory limits |
| 845 |
* ADDED 2.6.1: Prevents memory exhaustion |
| 846 |
*/ |
| 847 |
private function mxchat_get_recent_entries_safe($pinecone_options, $bot_id = 'default', $limit = 500) { |
| 848 |
global $wpdb; |
| 849 |
|
| 850 |
$api_key = $pinecone_options['mxchat_pinecone_api_key'] ?? ''; |
| 851 |
$host = $pinecone_options['mxchat_pinecone_host'] ?? ''; |
| 852 |
$namespace = $pinecone_options['mxchat_pinecone_namespace'] ?? ''; |
| 853 |
|
| 854 |
if (empty($api_key) || empty($host)) { |
| 855 |
return array(); |
| 856 |
} |
| 857 |
|
| 858 |
try { |
| 859 |
$all_records = array(); |
| 860 |
$seen_ids = array(); |
| 861 |
$query_url = "https://{$host}/query"; |
| 862 |
|
| 863 |
// Use only 2 query vectors instead of 5 for better performance |
| 864 |
$fixed_vectors = array_slice($this->mxchat_generate_fixed_query_vectors(), 0, 2); |
| 865 |
|
| 866 |
foreach ($fixed_vectors as $query_vector) { |
| 867 |
// Limit topK to prevent memory issues |
| 868 |
$topK = min(500, $limit); |
| 869 |
|
| 870 |
$query_data = array( |
| 871 |
'includeMetadata' => true, |
| 872 |
'includeValues' => false, |
| 873 |
'topK' => $topK, |
| 874 |
'vector' => $query_vector |
| 875 |
); |
| 876 |
|
| 877 |
if (!empty($namespace)) { |
| 878 |
$query_data['namespace'] = $namespace; |
| 879 |
} |
| 880 |
|
| 881 |
$response = wp_remote_post($query_url, array( |
| 882 |
'headers' => array( |
| 883 |
'Api-Key' => $api_key, |
| 884 |
'Content-Type' => 'application/json' |
| 885 |
), |
| 886 |
'body' => json_encode($query_data), |
| 887 |
'timeout' => 15 |
| 888 |
)); |
| 889 |
|
| 890 |
if (!is_wp_error($response) && wp_remote_retrieve_response_code($response) === 200) { |
| 891 |
$body = wp_remote_retrieve_body($response); |
| 892 |
$data = json_decode($body, true); |
| 893 |
|
| 894 |
if (isset($data['matches'])) { |
| 895 |
foreach ($data['matches'] as $match) { |
| 896 |
$match_id = $match['id'] ?? ''; |
| 897 |
if (!empty($match_id) && !isset($seen_ids[$match_id])) { |
| 898 |
$metadata = $match['metadata'] ?? array(); |
| 899 |
|
| 900 |
$created_at = $metadata['created_at'] ?? $metadata['last_updated'] ?? $metadata['timestamp'] ?? time(); |
| 901 |
if (!is_numeric($created_at)) { |
| 902 |
$created_at = strtotime($created_at) ?: time(); |
| 903 |
} |
| 904 |
|
| 905 |
$all_records[] = (object) array( |
| 906 |
'id' => $match_id, |
| 907 |
'article_content' => $metadata['text'] ?? '', |
| 908 |
'source_url' => $metadata['source_url'] ?? '', |
| 909 |
'role_restriction' => $metadata['role_restriction'] ?? 'public', |
| 910 |
'type' => $metadata['type'] ?? 'content', |
| 911 |
'bot_id' => $bot_id, |
| 912 |
'created_at' => $created_at, |
| 913 |
'data_source' => 'pinecone', |
| 914 |
'chunk_index' => isset($metadata['chunk_index']) ? intval($metadata['chunk_index']) : null, |
| 915 |
'total_chunks' => isset($metadata['total_chunks']) ? intval($metadata['total_chunks']) : null, |
| 916 |
'is_chunked' => isset($metadata['is_chunked']) ? (bool) $metadata['is_chunked'] : false |
| 917 |
); |
| 918 |
|
| 919 |
$seen_ids[$match_id] = true; |
| 920 |
|
| 921 |
// Stop if we've reached our limit |
| 922 |
if (count($all_records) >= $limit) { |
| 923 |
break 2; |
| 924 |
} |
| 925 |
} |
| 926 |
} |
| 927 |
} |
| 928 |
} |
| 929 |
|
| 930 |
// Minimal delay between requests |
| 931 |
usleep(50000); // 0.05 second delay |
| 932 |
} |
| 933 |
|
| 934 |
// Sort by created_at (newest first) and apply limit |
| 935 |
usort($all_records, function($a, $b) { |
| 936 |
return $b->created_at - $a->created_at; |
| 937 |
}); |
| 938 |
|
| 939 |
$limited_records = array_slice($all_records, 0, $limit); |
| 940 |
|
| 941 |
// Batch fetch role restrictions |
| 942 |
$this->mxchat_batch_fetch_role_restrictions($limited_records, $bot_id); |
| 943 |
|
| 944 |
return $limited_records; |
| 945 |
|
| 946 |
} catch (Exception $e) { |
| 947 |
//error_log('MxChat safe fetch exception: ' . $e->getMessage()); |
| 948 |
MxChat_Admin::mxchat_log_debug('pinecone_error', 'Pinecone safe fetch error: ' . $e->getMessage()); |
| 949 |
return array(); |
| 950 |
} |
| 951 |
} |
| 952 |
/** |
| 953 |
* Get embedding dimensions based on the selected model |
| 954 |
* ADD THIS NEW FUNCTION |
| 955 |
*/ |
| 956 |
private function mxchat_get_embedding_dimensions() { |
| 957 |
$options = get_option('mxchat_options', array()); |
| 958 |
$selected_model = $options['embedding_model'] ?? 'text-embedding-ada-002'; |
| 959 |
|
| 960 |
// Define dimensions for different models |
| 961 |
$model_dimensions = array( |
| 962 |
'text-embedding-ada-002' => 1536, |
| 963 |
'text-embedding-3-small' => 1536, |
| 964 |
'text-embedding-3-large' => 3072, |
| 965 |
'voyage-2' => 1024, |
| 966 |
'voyage-large-2' => 1536, |
| 967 |
'voyage-3-large' => 2048, |
| 968 |
'gemini-embedding-001' => 1536, |
| 969 |
); |
| 970 |
|
| 971 |
// Check if it's a voyage model with custom dimensions |
| 972 |
if (strpos($selected_model, 'voyage-3-large') === 0) { |
| 973 |
$custom_dimensions = $options['voyage_output_dimension'] ?? 2048; |
| 974 |
return intval($custom_dimensions); |
| 975 |
} |
| 976 |
|
| 977 |
// Check if it's a gemini model with custom dimensions |
| 978 |
if (strpos($selected_model, 'gemini-embedding') === 0) { |
| 979 |
$custom_dimensions = $options['gemini_output_dimension'] ?? 1536; |
| 980 |
return intval($custom_dimensions); |
| 981 |
} |
| 982 |
|
| 983 |
// Return known dimensions or default to 1536 |
| 984 |
return $model_dimensions[$selected_model] ?? 1536; |
| 985 |
} |
| 986 |
|
| 987 |
/** |
| 988 |
* Generate random unit vector with correct dimensions |
| 989 |
* ADD THIS NEW FUNCTION |
| 990 |
*/ |
| 991 |
private function mxchat_generate_random_vector() { |
| 992 |
$dimensions = $this->mxchat_get_embedding_dimensions(); |
| 993 |
|
| 994 |
$random_vector = array(); |
| 995 |
for ($i = 0; $i < $dimensions; $i++) { |
| 996 |
$random_vector[] = (rand(-1000, 1000) / 1000.0); |
| 997 |
} |
| 998 |
|
| 999 |
// Normalize the vector to unit length |
| 1000 |
$magnitude = sqrt(array_sum(array_map(function($x) { return $x * $x; }, $random_vector))); |
| 1001 |
if ($magnitude > 0) { |
| 1002 |
$random_vector = array_map(function($x) use ($magnitude) { return $x / $magnitude; }, $random_vector); |
| 1003 |
} |
| 1004 |
|
| 1005 |
return $random_vector; |
| 1006 |
} |
| 1007 |
|
| 1008 |
|
| 1009 |
/** |
| 1010 |
* Get recent entries from Pinecone |
| 1011 |
* UPDATED 2.6.1: Now uses safe version with memory limits to prevent crashes |
| 1012 |
* @deprecated Use mxchat_get_recent_entries_safe() instead for new code |
| 1013 |
*/ |
| 1014 |
private function mxchat_get_recent_1k_entries($pinecone_options, $bot_id = 'default') { |
| 1015 |
// Delegate to the safe version with a reasonable limit |
| 1016 |
// This prevents crashes with large datasets (13K+ products) |
| 1017 |
return $this->mxchat_get_recent_entries_safe($pinecone_options, $bot_id, 500); |
| 1018 |
} |
| 1019 |
/** |
| 1020 |
* Generate fixed query vectors for consistent results |
| 1021 |
*/ |
| 1022 |
private function mxchat_generate_fixed_query_vectors() { |
| 1023 |
$dimensions = $this->mxchat_get_embedding_dimensions(); |
| 1024 |
$vectors = array(); |
| 1025 |
|
| 1026 |
// Create 5 fixed vectors with different patterns for better coverage |
| 1027 |
$patterns = array( |
| 1028 |
'zeros_with_ones' => 0.1, // Mostly zeros with some 1s |
| 1029 |
'ascending' => 0.2, // Ascending pattern |
| 1030 |
'descending' => 0.3, // Descending pattern |
| 1031 |
'alternating' => 0.4, // Alternating positive/negative |
| 1032 |
'center_weighted' => 0.5 // Higher values in center |
| 1033 |
); |
| 1034 |
|
| 1035 |
foreach ($patterns as $pattern_name => $seed) { |
| 1036 |
$vector = array(); |
| 1037 |
|
| 1038 |
for ($i = 0; $i < $dimensions; $i++) { |
| 1039 |
switch ($pattern_name) { |
| 1040 |
case 'zeros_with_ones': |
| 1041 |
$vector[] = ($i % 10 === 0) ? 1.0 : 0.0; |
| 1042 |
break; |
| 1043 |
case 'ascending': |
| 1044 |
$vector[] = ($i / $dimensions) * 2 - 1; // Range -1 to 1 |
| 1045 |
break; |
| 1046 |
case 'descending': |
| 1047 |
$vector[] = (($dimensions - $i) / $dimensions) * 2 - 1; |
| 1048 |
break; |
| 1049 |
case 'alternating': |
| 1050 |
$vector[] = ($i % 2 === 0) ? $seed : -$seed; |
| 1051 |
break; |
| 1052 |
case 'center_weighted': |
| 1053 |
$center = $dimensions / 2; |
| 1054 |
$distance = abs($i - $center) / $center; |
| 1055 |
$vector[] = (1 - $distance) * $seed; |
| 1056 |
break; |
| 1057 |
} |
| 1058 |
} |
| 1059 |
|
| 1060 |
// Normalize the vector to unit length |
| 1061 |
$magnitude = sqrt(array_sum(array_map(function($x) { return $x * $x; }, $vector))); |
| 1062 |
if ($magnitude > 0) { |
| 1063 |
$vector = array_map(function($x) use ($magnitude) { return $x / $magnitude; }, $vector); |
| 1064 |
} |
| 1065 |
|
| 1066 |
$vectors[] = $vector; |
| 1067 |
} |
| 1068 |
|
| 1069 |
return $vectors; |
| 1070 |
} |
| 1071 |
|
| 1072 |
|
| 1073 |
/** |
| 1074 |
* Scan Pinecone for processed content |
| 1075 |
* UPDATED 2.6.2: Uses direct ID lookup via fetch API instead of random vector scanning |
| 1076 |
* This removes the 10K record limit and scales to any database size |
| 1077 |
* |
| 1078 |
* @param array $pinecone_options Pinecone configuration options |
| 1079 |
* @param array $post_ids Optional array of specific post IDs to check (if empty, checks all published posts) |
| 1080 |
* @return array Processed data keyed by post ID |
| 1081 |
*/ |
| 1082 |
public function mxchat_scan_pinecone_for_processed_content($pinecone_options, $post_ids = array()) { |
| 1083 |
$api_key = $pinecone_options['mxchat_pinecone_api_key'] ?? ''; |
| 1084 |
$host = $pinecone_options['mxchat_pinecone_host'] ?? ''; |
| 1085 |
|
| 1086 |
if (empty($api_key) || empty($host)) { |
| 1087 |
return array(); |
| 1088 |
} |
| 1089 |
|
| 1090 |
try { |
| 1091 |
// If no specific post IDs provided, get all published posts |
| 1092 |
if (empty($post_ids)) { |
| 1093 |
$posts = get_posts(array( |
| 1094 |
'post_type' => 'any', |
| 1095 |
'post_status' => 'publish', |
| 1096 |
'posts_per_page' => -1, |
| 1097 |
'fields' => 'ids', |
| 1098 |
'no_found_rows' => true, |
| 1099 |
'update_post_meta_cache' => false, |
| 1100 |
'update_post_term_cache' => false, |
| 1101 |
)); |
| 1102 |
$post_ids = $posts; |
| 1103 |
} |
| 1104 |
|
| 1105 |
if (empty($post_ids)) { |
| 1106 |
return array(); |
| 1107 |
} |
| 1108 |
|
| 1109 |
// Build a map of vector_id => post data for lookup |
| 1110 |
$vector_id_map = array(); |
| 1111 |
foreach ($post_ids as $post_id) { |
| 1112 |
$permalink = get_permalink($post_id); |
| 1113 |
if ($permalink) { |
| 1114 |
$vector_id = md5($permalink); |
| 1115 |
$vector_id_map[$vector_id] = array( |
| 1116 |
'post_id' => $post_id, |
| 1117 |
'url' => $permalink |
| 1118 |
); |
| 1119 |
} |
| 1120 |
} |
| 1121 |
|
| 1122 |
if (empty($vector_id_map)) { |
| 1123 |
return array(); |
| 1124 |
} |
| 1125 |
|
| 1126 |
// Batch check Pinecone using fetch API. |
| 1127 |
// NOTE (plan 793b82): /vectors/fetch is a GET endpoint with the ids |
| 1128 |
// repeated in the query string (ids=a&ids=b); the old POST here was |
| 1129 |
// answered 200-with-an-empty-body, so this scan saw NOTHING as indexed |
| 1130 |
// and callers re-embedded content Pinecone already had. Chunk size 100: |
| 1131 |
// measured on a live serverless index, ~9KB of URL is accepted and |
| 1132 |
// ~18KB draws HTTP 414, so 100 32-char ids (~4KB) leaves real margin. |
| 1133 |
$namespace = $pinecone_options['mxchat_pinecone_namespace'] ?? ''; |
| 1134 |
$all_vector_ids = array_keys($vector_id_map); |
| 1135 |
$chunks = array_chunk($all_vector_ids, 100); |
| 1136 |
$processed_data = array(); |
| 1137 |
|
| 1138 |
foreach ($chunks as $chunk) { |
| 1139 |
$fetch_query = array(); |
| 1140 |
foreach ($chunk as $fetch_vid) { |
| 1141 |
$fetch_query[] = 'ids=' . rawurlencode($fetch_vid); |
| 1142 |
} |
| 1143 |
if (!empty($namespace)) { |
| 1144 |
$fetch_query[] = 'namespace=' . rawurlencode($namespace); |
| 1145 |
} |
| 1146 |
|
| 1147 |
$response = wp_remote_get("https://{$host}/vectors/fetch?" . implode('&', $fetch_query), array( |
| 1148 |
'headers' => array( |
| 1149 |
'Api-Key' => $api_key, |
| 1150 |
'accept' => 'application/json' |
| 1151 |
), |
| 1152 |
'timeout' => 30 |
| 1153 |
)); |
| 1154 |
|
| 1155 |
if (is_wp_error($response)) { |
| 1156 |
error_log('MxChat Pinecone: processed-content scan /vectors/fetch GET failed: ' . $response->get_error_message()); |
| 1157 |
continue; |
| 1158 |
} |
| 1159 |
|
| 1160 |
$response_code = wp_remote_retrieve_response_code($response); |
| 1161 |
if ($response_code !== 200) { |
| 1162 |
error_log('MxChat Pinecone: processed-content scan /vectors/fetch GET returned HTTP ' . $response_code); |
| 1163 |
continue; |
| 1164 |
} |
| 1165 |
|
| 1166 |
$body = wp_remote_retrieve_body($response); |
| 1167 |
$data = json_decode($body, true); |
| 1168 |
|
| 1169 |
// Process returned vectors |
| 1170 |
if (isset($data['vectors']) && is_array($data['vectors'])) { |
| 1171 |
foreach ($data['vectors'] as $vector_id => $vector_data) { |
| 1172 |
if (isset($vector_id_map[$vector_id])) { |
| 1173 |
$post_info = $vector_id_map[$vector_id]; |
| 1174 |
$post_id = $post_info['post_id']; |
| 1175 |
$metadata = $vector_data['metadata'] ?? array(); |
| 1176 |
|
| 1177 |
$created_at = $metadata['created_at'] ?? ''; |
| 1178 |
$processed_date = 'Recently'; |
| 1179 |
$timestamp = current_time('timestamp'); |
| 1180 |
|
| 1181 |
if (!empty($created_at)) { |
| 1182 |
$ts = is_numeric($created_at) ? $created_at : strtotime($created_at); |
| 1183 |
if ($ts) { |
| 1184 |
$timestamp = $ts; |
| 1185 |
$processed_date = human_time_diff($ts, current_time('timestamp')) . ' ago'; |
| 1186 |
} |
| 1187 |
} |
| 1188 |
|
| 1189 |
$processed_data[$post_id] = array( |
| 1190 |
'db_id' => $vector_id, |
| 1191 |
'processed_date' => $processed_date, |
| 1192 |
'url' => $post_info['url'], |
| 1193 |
'source' => 'pinecone', |
| 1194 |
'timestamp' => $timestamp |
| 1195 |
); |
| 1196 |
} |
| 1197 |
} |
| 1198 |
} |
| 1199 |
} |
| 1200 |
|
| 1201 |
return $processed_data; |
| 1202 |
|
| 1203 |
} catch (Exception $e) { |
| 1204 |
return array(); |
| 1205 |
} |
| 1206 |
} |
| 1207 |
|
| 1208 |
/** |
| 1209 |
* Get total count from Pinecone stats API |
| 1210 |
* UPDATED: Removed cache fallback reference |
| 1211 |
*/ |
| 1212 |
private function mxchat_get_pinecone_total_count($pinecone_options, $bot_id = 'default') { |
| 1213 |
$api_key = $pinecone_options['mxchat_pinecone_api_key'] ?? ''; |
| 1214 |
$host = $pinecone_options['mxchat_pinecone_host'] ?? ''; |
| 1215 |
$namespace = $pinecone_options['mxchat_pinecone_namespace'] ?? ''; |
| 1216 |
|
| 1217 |
if (empty($api_key) || empty($host)) { |
| 1218 |
return 0; |
| 1219 |
} |
| 1220 |
|
| 1221 |
try { |
| 1222 |
$stats_url = "https://{$host}/describe_index_stats"; |
| 1223 |
|
| 1224 |
// describe_index_stats doesn't need a body, just the POST request |
| 1225 |
$response = wp_remote_post($stats_url, array( |
| 1226 |
'headers' => array( |
| 1227 |
'Api-Key' => $api_key, |
| 1228 |
'Content-Type' => 'application/json' |
| 1229 |
), |
| 1230 |
'body' => '{}', |
| 1231 |
'timeout' => 15 |
| 1232 |
)); |
| 1233 |
|
| 1234 |
if (!is_wp_error($response) && wp_remote_retrieve_response_code($response) === 200) { |
| 1235 |
$body = wp_remote_retrieve_body($response); |
| 1236 |
$stats_data = json_decode($body, true); |
| 1237 |
|
| 1238 |
// If namespace is specified, get count from that specific namespace |
| 1239 |
// Pinecone stats response format: { namespaces: { "ns": { vectorCount: N } }, totalVectorCount: N } |
| 1240 |
if (!empty($namespace) && isset($stats_data['namespaces'][$namespace]['vectorCount'])) { |
| 1241 |
$namespace_count = intval($stats_data['namespaces'][$namespace]['vectorCount']); |
| 1242 |
//error_log('DEBUG: Got namespace-specific count: ' . $namespace_count . ' for namespace: ' . $namespace); |
| 1243 |
return $namespace_count; |
| 1244 |
} |
| 1245 |
|
| 1246 |
// If no namespace specified or namespace not found in response, use total |
| 1247 |
$total_count = $stats_data['totalVectorCount'] ?? 0; |
| 1248 |
if ($total_count > 0) { |
| 1249 |
//error_log('DEBUG: Got total count from stats API: ' . $total_count); |
| 1250 |
return intval($total_count); |
| 1251 |
} |
| 1252 |
} |
| 1253 |
|
| 1254 |
// If stats API fails, return 0 instead of using cache |
| 1255 |
return 0; |
| 1256 |
|
| 1257 |
} catch (Exception $e) { |
| 1258 |
//error_log('DEBUG: Exception getting total count: ' . $e->getMessage()); |
| 1259 |
return 0; |
| 1260 |
} |
| 1261 |
} |
| 1262 |
|
| 1263 |
/** |
| 1264 |
* Get bot-specific Pinecone configuration for database operations |
| 1265 |
*/ |
| 1266 |
public function mxchat_get_bot_pinecone_options($bot_id = 'default') { |
| 1267 |
//error_log('DEBUG: Getting Pinecone options for bot: ' . $bot_id); |
| 1268 |
|
| 1269 |
// If default bot or multi-bot add-on not active, use default Pinecone config |
| 1270 |
if ($bot_id === 'default' || !class_exists('MxChat_Multi_Bot_Manager')) { |
| 1271 |
$addon_options = get_option('mxchat_pinecone_addon_options', array()); |
| 1272 |
//error_log('DEBUG: Using default Pinecone options'); |
| 1273 |
return $addon_options; |
| 1274 |
} |
| 1275 |
|
| 1276 |
// Get bot-specific configuration using the filter |
| 1277 |
$bot_config = apply_filters('mxchat_get_bot_pinecone_config', array(), $bot_id); |
| 1278 |
|
| 1279 |
//error_log('DEBUG: Bot config from filter: ' . print_r($bot_config, true)); |
| 1280 |
|
| 1281 |
// Check if we got valid bot-specific config |
| 1282 |
if (!empty($bot_config) && isset($bot_config['use_pinecone']) && $bot_config['use_pinecone']) { |
| 1283 |
// Convert bot config to the format expected by fetch functions |
| 1284 |
$pinecone_options = array( |
| 1285 |
'mxchat_use_pinecone' => '1', |
| 1286 |
'mxchat_pinecone_api_key' => $bot_config['api_key'] ?? '', |
| 1287 |
'mxchat_pinecone_host' => $bot_config['host'] ?? '', |
| 1288 |
'mxchat_pinecone_namespace' => $bot_config['namespace'] ?? '', |
| 1289 |
'mxchat_pinecone_environment' => '', |
| 1290 |
'mxchat_pinecone_index' => '' |
| 1291 |
); |
| 1292 |
|
| 1293 |
//error_log('DEBUG: Returning bot-specific Pinecone options for bot: ' . $bot_id); |
| 1294 |
return $pinecone_options; |
| 1295 |
} |
| 1296 |
|
| 1297 |
// Fallback to default options if bot-specific config is invalid |
| 1298 |
//error_log('DEBUG: Bot-specific config invalid, falling back to default'); |
| 1299 |
return get_option('mxchat_pinecone_addon_options', array()); |
| 1300 |
} |
| 1301 |
|
| 1302 |
/** |
| 1303 |
* Get bot-specific Pinecone configuration |
| 1304 |
* Used in the knowledge retrieval functions |
| 1305 |
*/ |
| 1306 |
private function get_bot_pinecone_config($bot_id = 'default') { |
| 1307 |
//error_log("MXCHAT DEBUG: get_bot_pinecone_config called for bot: " . $bot_id); |
| 1308 |
|
| 1309 |
// If default bot or multi-bot add-on not active, use default Pinecone config |
| 1310 |
if ($bot_id === 'default' || !class_exists('MxChat_Multi_Bot_Manager')) { |
| 1311 |
//error_log("MXCHAT DEBUG: Using default Pinecone config (no multi-bot or bot is 'default')"); |
| 1312 |
$addon_options = get_option('mxchat_pinecone_addon_options', array()); |
| 1313 |
$config = array( |
| 1314 |
'use_pinecone' => (isset($addon_options['mxchat_use_pinecone']) && $addon_options['mxchat_use_pinecone'] === '1'), |
| 1315 |
'api_key' => $addon_options['mxchat_pinecone_api_key'] ?? '', |
| 1316 |
'host' => $addon_options['mxchat_pinecone_host'] ?? '', |
| 1317 |
'namespace' => $addon_options['mxchat_pinecone_namespace'] ?? '' |
| 1318 |
); |
| 1319 |
//error_log("MXCHAT DEBUG: Default config - use_pinecone: " . ($config['use_pinecone'] ? 'true' : 'false')); |
| 1320 |
return $config; |
| 1321 |
} |
| 1322 |
|
| 1323 |
//error_log("MXCHAT DEBUG: Calling filter 'mxchat_get_bot_pinecone_config' for bot: " . $bot_id); |
| 1324 |
|
| 1325 |
// Hook for multi-bot add-on to provide bot-specific Pinecone config |
| 1326 |
$bot_pinecone_config = apply_filters('mxchat_get_bot_pinecone_config', array(), $bot_id); |
| 1327 |
|
| 1328 |
if (!empty($bot_pinecone_config)) { |
| 1329 |
//error_log("MXCHAT DEBUG: Got bot-specific config from filter"); |
| 1330 |
//error_log(" - use_pinecone: " . (isset($bot_pinecone_config['use_pinecone']) ? ($bot_pinecone_config['use_pinecone'] ? 'true' : 'false') : 'not set')); |
| 1331 |
//error_log(" - host: " . ($bot_pinecone_config['host'] ?? 'not set')); |
| 1332 |
//error_log(" - namespace: " . ($bot_pinecone_config['namespace'] ?? 'not set')); |
| 1333 |
} else { |
| 1334 |
//error_log("MXCHAT DEBUG: Filter returned empty config!"); |
| 1335 |
} |
| 1336 |
|
| 1337 |
return is_array($bot_pinecone_config) ? $bot_pinecone_config : array(); |
| 1338 |
} |
| 1339 |
|
| 1340 |
|
| 1341 |
/** |
| 1342 |
* Fetches vectors from Pinecone using provided IDs (for content selection feature) |
| 1343 |
*/ |
| 1344 |
public function fetch_pinecone_vectors_by_ids($pinecone_options, $vector_ids) { |
| 1345 |
$api_key = $pinecone_options['mxchat_pinecone_api_key'] ?? ''; |
| 1346 |
$host = $pinecone_options['mxchat_pinecone_host'] ?? ''; |
| 1347 |
$namespace = $pinecone_options['mxchat_pinecone_namespace'] ?? ''; |
| 1348 |
|
| 1349 |
if (empty($api_key) || empty($host) || empty($vector_ids)) { |
| 1350 |
return array(); |
| 1351 |
} |
| 1352 |
|
| 1353 |
try { |
| 1354 |
// NOTE (plan 793b82): /vectors/fetch is a GET endpoint with the ids |
| 1355 |
// repeated in the query string; the old POST here was answered |
| 1356 |
// 200-with-an-empty-body. Chunked at 100 ids to stay well under the |
| 1357 |
// measured HTTP 414 URL-length boundary. |
| 1358 |
$vectors = array(); |
| 1359 |
foreach (array_chunk(array_values($vector_ids), 100) as $chunk) { |
| 1360 |
$fetch_query = array(); |
| 1361 |
foreach ($chunk as $fetch_vid) { |
| 1362 |
$fetch_query[] = 'ids=' . rawurlencode($fetch_vid); |
| 1363 |
} |
| 1364 |
if (!empty($namespace)) { |
| 1365 |
$fetch_query[] = 'namespace=' . rawurlencode($namespace); |
| 1366 |
} |
| 1367 |
|
| 1368 |
$response = wp_remote_get("https://{$host}/vectors/fetch?" . implode('&', $fetch_query), array( |
| 1369 |
'headers' => array( |
| 1370 |
'Api-Key' => $api_key, |
| 1371 |
'accept' => 'application/json' |
| 1372 |
), |
| 1373 |
'timeout' => 30 |
| 1374 |
)); |
| 1375 |
|
| 1376 |
if (is_wp_error($response)) { |
| 1377 |
error_log('MxChat Pinecone: fetch_pinecone_vectors_by_ids GET failed: ' . $response->get_error_message()); |
| 1378 |
continue; |
| 1379 |
} |
| 1380 |
|
| 1381 |
if (wp_remote_retrieve_response_code($response) !== 200) { |
| 1382 |
error_log('MxChat Pinecone: fetch_pinecone_vectors_by_ids GET returned HTTP ' . wp_remote_retrieve_response_code($response)); |
| 1383 |
continue; |
| 1384 |
} |
| 1385 |
|
| 1386 |
$data = json_decode(wp_remote_retrieve_body($response), true); |
| 1387 |
if (isset($data['vectors']) && is_array($data['vectors'])) { |
| 1388 |
$vectors += $data['vectors']; |
| 1389 |
} |
| 1390 |
} |
| 1391 |
|
| 1392 |
if (empty($vectors)) { |
| 1393 |
return array(); |
| 1394 |
} |
| 1395 |
|
| 1396 |
$processed_data = array(); |
| 1397 |
|
| 1398 |
foreach ($vectors as $vector_id => $vector_data) { |
| 1399 |
$metadata = $vector_data['metadata'] ?? array(); |
| 1400 |
$source_url = $metadata['source_url'] ?? ''; |
| 1401 |
|
| 1402 |
if (!empty($source_url)) { |
| 1403 |
$post_id = url_to_postid($source_url); |
| 1404 |
if ($post_id) { |
| 1405 |
$created_at = $metadata['created_at'] ?? ''; |
| 1406 |
$processed_date = 'Recently'; // Default |
| 1407 |
|
| 1408 |
if (!empty($created_at)) { |
| 1409 |
$timestamp = is_numeric($created_at) ? $created_at : strtotime($created_at); |
| 1410 |
if ($timestamp) { |
| 1411 |
$processed_date = human_time_diff($timestamp, current_time('timestamp')) . ' ago'; |
| 1412 |
} |
| 1413 |
} |
| 1414 |
|
| 1415 |
$processed_data[$post_id] = array( |
| 1416 |
'db_id' => $vector_id, |
| 1417 |
'processed_date' => $processed_date, |
| 1418 |
'url' => $source_url, |
| 1419 |
'source' => 'pinecone', |
| 1420 |
'timestamp' => $timestamp ?? current_time('timestamp') |
| 1421 |
); |
| 1422 |
} |
| 1423 |
} |
| 1424 |
} |
| 1425 |
|
| 1426 |
return $processed_data; |
| 1427 |
|
| 1428 |
} catch (Exception $e) { |
| 1429 |
return array(); |
| 1430 |
} |
| 1431 |
} |
| 1432 |
// ======================================== |
| 1433 |
// PINECONE DELETE OPERATIONS |
| 1434 |
// ======================================== |
| 1435 |
|
| 1436 |
/** |
| 1437 |
* Delete all vectors from Pinecone |
| 1438 |
* Loops until all vectors are deleted (handles large databases) |
| 1439 |
*/ |
| 1440 |
public function mxchat_delete_all_from_pinecone($pinecone_options, $content_type_filter = '') { |
| 1441 |
$api_key = $pinecone_options['mxchat_pinecone_api_key'] ?? ''; |
| 1442 |
$host = $pinecone_options['mxchat_pinecone_host'] ?? ''; |
| 1443 |
$namespace = $pinecone_options['mxchat_pinecone_namespace'] ?? ''; |
| 1444 |
|
| 1445 |
if (empty($api_key) || empty($host)) { |
| 1446 |
return array( |
| 1447 |
'success' => false, |
| 1448 |
'message' => 'Missing Pinecone API credentials' |
| 1449 |
); |
| 1450 |
} |
| 1451 |
|
| 1452 |
try { |
| 1453 |
$total_deleted = 0; |
| 1454 |
$failed_batches = 0; |
| 1455 |
$max_iterations = 100; // Safety limit to prevent infinite loops |
| 1456 |
$iteration = 0; |
| 1457 |
|
| 1458 |
// Loop until no more vectors are found |
| 1459 |
do { |
| 1460 |
$iteration++; |
| 1461 |
|
| 1462 |
// Get a batch of vector IDs from Pinecone |
| 1463 |
$records = $this->mxchat_get_recent_1k_entries($pinecone_options); |
| 1464 |
$vector_ids = array(); |
| 1465 |
|
| 1466 |
foreach ($records as $record) { |
| 1467 |
if (!empty($record->id)) { |
| 1468 |
// If content type filter is active, only include matching records |
| 1469 |
if (!empty($content_type_filter)) { |
| 1470 |
$record_type = isset($record->type) ? $record->type : ''; |
| 1471 |
if ($record_type !== $content_type_filter) { |
| 1472 |
continue; |
| 1473 |
} |
| 1474 |
} |
| 1475 |
$vector_ids[] = $record->id; |
| 1476 |
} |
| 1477 |
} |
| 1478 |
|
| 1479 |
// If no matching vectors found, we're done |
| 1480 |
// (either no records at all, or all remaining records are non-matching types) |
| 1481 |
if (empty($vector_ids)) { |
| 1482 |
break; |
| 1483 |
} |
| 1484 |
|
| 1485 |
// Delete vectors in batches (Pinecone has limits on batch operations) |
| 1486 |
$batch_size = 100; |
| 1487 |
$batches = array_chunk($vector_ids, $batch_size); |
| 1488 |
|
| 1489 |
foreach ($batches as $batch) { |
| 1490 |
$result = $this->mxchat_delete_pinecone_batch($batch, $api_key, $host, $namespace); |
| 1491 |
if ($result['success']) { |
| 1492 |
$total_deleted += count($batch); |
| 1493 |
} else { |
| 1494 |
$failed_batches++; |
| 1495 |
} |
| 1496 |
} |
| 1497 |
|
| 1498 |
// Small delay to avoid rate limiting |
| 1499 |
usleep(100000); // 100ms |
| 1500 |
|
| 1501 |
} while ($iteration < $max_iterations); |
| 1502 |
|
| 1503 |
if ($total_deleted === 0) { |
| 1504 |
return array( |
| 1505 |
'success' => true, |
| 1506 |
'message' => 'No vectors found to delete' |
| 1507 |
); |
| 1508 |
} |
| 1509 |
|
| 1510 |
if ($failed_batches > 0) { |
| 1511 |
return array( |
| 1512 |
'success' => false, |
| 1513 |
'message' => sprintf('Deleted %d vectors, but %d batches failed', $total_deleted, $failed_batches) |
| 1514 |
); |
| 1515 |
} |
| 1516 |
|
| 1517 |
return array( |
| 1518 |
'success' => true, |
| 1519 |
'message' => "Successfully deleted {$total_deleted} vectors from Pinecone" |
| 1520 |
); |
| 1521 |
|
| 1522 |
} catch (Exception $e) { |
| 1523 |
return array( |
| 1524 |
'success' => false, |
| 1525 |
'message' => $e->getMessage() |
| 1526 |
); |
| 1527 |
} |
| 1528 |
} |
| 1529 |
|
| 1530 |
/** |
| 1531 |
* Deletes batch of vectors from Pinecone database |
| 1532 |
*/ |
| 1533 |
public function mxchat_delete_pinecone_batch($vector_ids, $api_key, $host, $namespace = '') { |
| 1534 |
// Build the API endpoint |
| 1535 |
$api_endpoint = "https://{$host}/vectors/delete"; |
| 1536 |
|
| 1537 |
// Prepare the request body with the IDs |
| 1538 |
$request_body = array( |
| 1539 |
'ids' => $vector_ids |
| 1540 |
); |
| 1541 |
|
| 1542 |
// Add namespace if provided — omitting it targets the default namespace |
| 1543 |
if (!empty($namespace)) { |
| 1544 |
$request_body['namespace'] = $namespace; |
| 1545 |
} |
| 1546 |
|
| 1547 |
// Make the deletion request |
| 1548 |
$response = wp_remote_post($api_endpoint, array( |
| 1549 |
'headers' => array( |
| 1550 |
'Api-Key' => $api_key, |
| 1551 |
'accept' => 'application/json', |
| 1552 |
'content-type' => 'application/json' |
| 1553 |
), |
| 1554 |
'body' => wp_json_encode($request_body), |
| 1555 |
'timeout' => 60, // Increased timeout for batch operations |
| 1556 |
'method' => 'POST' |
| 1557 |
)); |
| 1558 |
|
| 1559 |
// Handle WordPress HTTP API errors |
| 1560 |
if (is_wp_error($response)) { |
| 1561 |
MxChat_Admin::mxchat_log_debug('pinecone_error', 'Pinecone batch deletion failed: ' . $response->get_error_message()); |
| 1562 |
return array( |
| 1563 |
'success' => false, |
| 1564 |
'message' => $response->get_error_message() |
| 1565 |
); |
| 1566 |
} |
| 1567 |
|
| 1568 |
// Check response status |
| 1569 |
$response_code = wp_remote_retrieve_response_code($response); |
| 1570 |
$response_body = wp_remote_retrieve_body($response); |
| 1571 |
|
| 1572 |
// Pinecone returns 200 for successful deletion |
| 1573 |
if ($response_code !== 200) { |
| 1574 |
MxChat_Admin::mxchat_log_debug('pinecone_error', 'Pinecone batch deletion failed (HTTP ' . $response_code . ')', array('response' => substr($response_body, 0, 200))); |
| 1575 |
return array( |
| 1576 |
'success' => false, |
| 1577 |
'message' => sprintf( |
| 1578 |
'Pinecone API error (HTTP %d): %s', |
| 1579 |
$response_code, |
| 1580 |
$response_body |
| 1581 |
) |
| 1582 |
); |
| 1583 |
} |
| 1584 |
|
| 1585 |
return array( |
| 1586 |
'success' => true, |
| 1587 |
'message' => 'Batch deleted successfully from Pinecone' |
| 1588 |
); |
| 1589 |
} |
| 1590 |
|
| 1591 |
|
| 1592 |
/** |
| 1593 |
* Deletes vector from Pinecone using API request |
| 1594 |
*/ |
| 1595 |
public function mxchat_delete_from_pinecone_by_vector_id($vector_id, $api_key, $host, $namespace = '') { |
| 1596 |
//error_log('=== PINECONE DELETE OPERATION ==='); |
| 1597 |
//error_log('Vector ID: ' . $vector_id); |
| 1598 |
//error_log('Host: ' . $host); |
| 1599 |
//error_log('API Key: ' . (empty($api_key) ? 'EMPTY' : 'SET')); |
| 1600 |
|
| 1601 |
// First, let's verify the vector exists before trying to delete |
| 1602 |
$fetch_url = "https://{$host}/vectors/fetch"; |
| 1603 |
|
| 1604 |
$fetch_params = array( |
| 1605 |
'ids' => array($vector_id) |
| 1606 |
); |
| 1607 |
|
| 1608 |
// Add namespace if provided (though you said you're not using namespaces) |
| 1609 |
if (!empty($namespace)) { |
| 1610 |
$fetch_params['namespace'] = $namespace; |
| 1611 |
} |
| 1612 |
|
| 1613 |
// Construct URL with query parameters for GET request |
| 1614 |
$fetch_url_with_params = $fetch_url . '?' . http_build_query($fetch_params); |
| 1615 |
|
| 1616 |
$fetch_response = wp_remote_get($fetch_url_with_params, array( |
| 1617 |
'headers' => array( |
| 1618 |
'Api-Key' => $api_key, |
| 1619 |
'accept' => 'application/json' |
| 1620 |
), |
| 1621 |
'timeout' => 15 |
| 1622 |
)); |
| 1623 |
|
| 1624 |
if (!is_wp_error($fetch_response) && wp_remote_retrieve_response_code($fetch_response) === 200) { |
| 1625 |
$fetch_body = wp_remote_retrieve_body($fetch_response); |
| 1626 |
$fetch_data = json_decode($fetch_body, true); |
| 1627 |
|
| 1628 |
//error_log('DEBUG: Fetch response: ' . print_r($fetch_data, true)); |
| 1629 |
|
| 1630 |
if (isset($fetch_data['vectors']) && isset($fetch_data['vectors'][$vector_id])) { |
| 1631 |
//error_log('DEBUG: Vector EXISTS in this index before deletion'); |
| 1632 |
} else { |
| 1633 |
//error_log('WARNING: Vector NOT FOUND in this index! It may be in a different bot\'s index'); |
| 1634 |
// You might want to return an error here |
| 1635 |
} |
| 1636 |
} else { |
| 1637 |
//error_log('DEBUG: Could not fetch vector to verify existence'); |
| 1638 |
} |
| 1639 |
|
| 1640 |
// Now proceed with deletion |
| 1641 |
$api_endpoint = "https://{$host}/vectors/delete"; |
| 1642 |
|
| 1643 |
// Prepare the request body with the ID |
| 1644 |
$request_body = array( |
| 1645 |
'ids' => array($vector_id) |
| 1646 |
); |
| 1647 |
|
| 1648 |
// Add namespace if provided |
| 1649 |
if (!empty($namespace)) { |
| 1650 |
$request_body['namespace'] = $namespace; |
| 1651 |
} |
| 1652 |
|
| 1653 |
//error_log('DEBUG: Delete request body: ' . json_encode($request_body)); |
| 1654 |
//error_log('DEBUG: Delete endpoint: ' . $api_endpoint); |
| 1655 |
|
| 1656 |
// Make the deletion request |
| 1657 |
$response = wp_remote_post($api_endpoint, array( |
| 1658 |
'headers' => array( |
| 1659 |
'Api-Key' => $api_key, |
| 1660 |
'accept' => 'application/json', |
| 1661 |
'content-type' => 'application/json' |
| 1662 |
), |
| 1663 |
'body' => wp_json_encode($request_body), |
| 1664 |
'timeout' => 30 |
| 1665 |
)); |
| 1666 |
|
| 1667 |
// Handle WordPress HTTP API errors |
| 1668 |
if (is_wp_error($response)) { |
| 1669 |
//error_log('DEBUG: WP Error: ' . $response->get_error_message()); |
| 1670 |
return array( |
| 1671 |
'success' => false, |
| 1672 |
'message' => $response->get_error_message() |
| 1673 |
); |
| 1674 |
} |
| 1675 |
|
| 1676 |
// Check response status |
| 1677 |
$response_code = wp_remote_retrieve_response_code($response); |
| 1678 |
$response_body = wp_remote_retrieve_body($response); |
| 1679 |
|
| 1680 |
//error_log('DEBUG: Delete response code: ' . $response_code); |
| 1681 |
//error_log('DEBUG: Delete response body: ' . $response_body); |
| 1682 |
|
| 1683 |
// Pinecone returns 200 for successful deletion (even if vector didn't exist) |
| 1684 |
if ($response_code !== 200) { |
| 1685 |
//error_log('DEBUG: Non-200 response from Pinecone'); |
| 1686 |
return array( |
| 1687 |
'success' => false, |
| 1688 |
'message' => sprintf( |
| 1689 |
'Pinecone API error (HTTP %d): %s', |
| 1690 |
$response_code, |
| 1691 |
$response_body |
| 1692 |
) |
| 1693 |
); |
| 1694 |
} |
| 1695 |
|
| 1696 |
// After deletion, verify it's actually gone |
| 1697 |
sleep(1); // Give Pinecone a moment to process |
| 1698 |
|
| 1699 |
$verify_response = wp_remote_get($fetch_url_with_params, array( |
| 1700 |
'headers' => array( |
| 1701 |
'Api-Key' => $api_key, |
| 1702 |
'accept' => 'application/json' |
| 1703 |
), |
| 1704 |
'timeout' => 15 |
| 1705 |
)); |
| 1706 |
|
| 1707 |
if (!is_wp_error($verify_response) && wp_remote_retrieve_response_code($verify_response) === 200) { |
| 1708 |
$verify_body = wp_remote_retrieve_body($verify_response); |
| 1709 |
$verify_data = json_decode($verify_body, true); |
| 1710 |
|
| 1711 |
if (isset($verify_data['vectors']) && isset($verify_data['vectors'][$vector_id])) { |
| 1712 |
//error_log('ERROR: Vector STILL EXISTS after deletion attempt!'); |
| 1713 |
return array( |
| 1714 |
'success' => false, |
| 1715 |
'message' => 'Vector still exists after deletion attempt' |
| 1716 |
); |
| 1717 |
} else { |
| 1718 |
//error_log('SUCCESS: Vector confirmed deleted (or never existed)'); |
| 1719 |
} |
| 1720 |
} |
| 1721 |
|
| 1722 |
//error_log('=== END PINECONE DELETE OPERATION ==='); |
| 1723 |
|
| 1724 |
return array( |
| 1725 |
'success' => true, |
| 1726 |
'message' => 'Vector deleted successfully from Pinecone' |
| 1727 |
); |
| 1728 |
} |
| 1729 |
|
| 1730 |
/** |
| 1731 |
* Deletes data from Pinecone index using API key |
| 1732 |
*/ |
| 1733 |
private function mxchat_delete_from_pinecone($urls, $api_key, $environment, $index_name) { |
| 1734 |
// Get the Pinecone host from options (matching your store_in_pinecone_main pattern) |
| 1735 |
$options = get_option('mxchat_pinecone_addon_options'); |
| 1736 |
$host = $options['mxchat_pinecone_host'] ?? ''; |
| 1737 |
|
| 1738 |
if (empty($host)) { |
| 1739 |
return array( |
| 1740 |
'success' => false, |
| 1741 |
'message' => 'Pinecone host is not configured. Please set the host in your settings.' |
| 1742 |
); |
| 1743 |
} |
| 1744 |
|
| 1745 |
// Build API endpoint using the configured host |
| 1746 |
$api_endpoint = "https://{$host}/vectors/delete"; |
| 1747 |
|
| 1748 |
// Create vector IDs from URLs (matching your store method's ID generation) |
| 1749 |
$vector_ids = array_map('md5', $urls); |
| 1750 |
|
| 1751 |
// Prepare the delete request body |
| 1752 |
$request_body = array( |
| 1753 |
'ids' => $vector_ids, |
| 1754 |
'filter' => array( |
| 1755 |
'source_url' => array( |
| 1756 |
'$in' => $urls |
| 1757 |
) |
| 1758 |
) |
| 1759 |
); |
| 1760 |
|
| 1761 |
// Make the deletion request |
| 1762 |
$response = wp_remote_post($api_endpoint, array( |
| 1763 |
'headers' => array( |
| 1764 |
'Api-Key' => $api_key, |
| 1765 |
'accept' => 'application/json', |
| 1766 |
'content-type' => 'application/json' |
| 1767 |
), |
| 1768 |
'body' => wp_json_encode($request_body), |
| 1769 |
'timeout' => 30, |
| 1770 |
'data_format' => 'body' |
| 1771 |
)); |
| 1772 |
|
| 1773 |
// Handle WordPress HTTP API errors |
| 1774 |
if (is_wp_error($response)) { |
| 1775 |
return array( |
| 1776 |
'success' => false, |
| 1777 |
'message' => $response->get_error_message() |
| 1778 |
); |
| 1779 |
} |
| 1780 |
|
| 1781 |
// Check response status |
| 1782 |
$response_code = wp_remote_retrieve_response_code($response); |
| 1783 |
if ($response_code !== 200) { |
| 1784 |
$body = wp_remote_retrieve_body($response); |
| 1785 |
return array( |
| 1786 |
'success' => false, |
| 1787 |
'message' => sprintf( |
| 1788 |
'Pinecone API error (HTTP %d): %s', |
| 1789 |
$response_code, |
| 1790 |
$body |
| 1791 |
) |
| 1792 |
); |
| 1793 |
} |
| 1794 |
|
| 1795 |
// Parse response body |
| 1796 |
$body = wp_remote_retrieve_body($response); |
| 1797 |
$response_data = json_decode($body, true); |
| 1798 |
|
| 1799 |
// Final validation of the response |
| 1800 |
if (json_last_error() !== JSON_ERROR_NONE) { |
| 1801 |
return array( |
| 1802 |
'success' => false, |
| 1803 |
'message' => 'Failed to parse Pinecone response: ' . json_last_error_msg() |
| 1804 |
); |
| 1805 |
} |
| 1806 |
|
| 1807 |
return array( |
| 1808 |
'success' => true, |
| 1809 |
'message' => sprintf('Successfully deleted %d vectors from Pinecone', count($vector_ids)) |
| 1810 |
); |
| 1811 |
} |
| 1812 |
|
| 1813 |
|
| 1814 |
|
| 1815 |
/** |
| 1816 |
* Retrieves processed content from Pinecone API |
| 1817 |
* |
| 1818 |
* @param array $pinecone_options Pinecone configuration options |
| 1819 |
* @param array $post_ids Optional array of specific post IDs to check (if empty, checks all) |
| 1820 |
* @return array Processed data keyed by post ID |
| 1821 |
*/ |
| 1822 |
public function mxchat_get_pinecone_processed_content($pinecone_options, $post_ids = array()) { |
| 1823 |
$api_key = $pinecone_options['mxchat_pinecone_api_key'] ?? ''; |
| 1824 |
$host = $pinecone_options['mxchat_pinecone_host'] ?? ''; |
| 1825 |
|
| 1826 |
if (empty($api_key) || empty($host)) { |
| 1827 |
return array(); |
| 1828 |
} |
| 1829 |
|
| 1830 |
$pinecone_data = array(); |
| 1831 |
|
| 1832 |
try { |
| 1833 |
// Always get fresh data from Pinecone |
| 1834 |
$pinecone_data = $this->mxchat_scan_pinecone_for_processed_content($pinecone_options, $post_ids); |
| 1835 |
|
| 1836 |
// Method 2: Final fallback - try stats endpoint (if available) |
| 1837 |
if (empty($pinecone_data)) { |
| 1838 |
$stats_url = "https://{$host}/describe_index_stats"; |
| 1839 |
|
| 1840 |
$response = wp_remote_post($stats_url, array( |
| 1841 |
'headers' => array( |
| 1842 |
'Api-Key' => $api_key, |
| 1843 |
'Content-Type' => 'application/json' |
| 1844 |
), |
| 1845 |
'body' => json_encode(array()), |
| 1846 |
'timeout' => 30 |
| 1847 |
)); |
| 1848 |
|
| 1849 |
if (!is_wp_error($response) && wp_remote_retrieve_response_code($response) === 200) { |
| 1850 |
$body = wp_remote_retrieve_body($response); |
| 1851 |
$stats_data = json_decode($body, true); |
| 1852 |
} |
| 1853 |
} |
| 1854 |
|
| 1855 |
} catch (Exception $e) { |
| 1856 |
// Log error but don't return cached data |
| 1857 |
} |
| 1858 |
|
| 1859 |
return $pinecone_data; |
| 1860 |
} |
| 1861 |
// ======================================== |
| 1862 |
// HELPER METHODS |
| 1863 |
// ======================================== |
| 1864 |
|
| 1865 |
/** |
| 1866 |
* Validates Pinecone API credentials |
| 1867 |
*/ |
| 1868 |
private function mxchat_validate_pinecone_credentials($api_key, $host) { |
| 1869 |
if (empty($api_key) || empty($host)) { |
| 1870 |
return false; |
| 1871 |
} |
| 1872 |
return true; |
| 1873 |
} |
| 1874 |
|
| 1875 |
/** |
| 1876 |
* Get Pinecone API credentials from options |
| 1877 |
*/ |
| 1878 |
private function mxchat_get_pinecone_credentials() { |
| 1879 |
$options = get_option('mxchat_options', array()); |
| 1880 |
return array( |
| 1881 |
'api_key' => isset($options['pinecone_api_key']) ? $options['pinecone_api_key'] : '', |
| 1882 |
'host' => isset($options['pinecone_host']) ? $options['pinecone_host'] : '' |
| 1883 |
); |
| 1884 |
} |
| 1885 |
|
| 1886 |
/** |
| 1887 |
* Log Pinecone operation errors |
| 1888 |
*/ |
| 1889 |
private function log_pinecone_error($operation, $error_message) { |
| 1890 |
//error_log("MxChat Pinecone {$operation} Error: " . $error_message); |
| 1891 |
} |
| 1892 |
|
| 1893 |
// ======================================== |
| 1894 |
// STATIC ACCESS METHODS (for backward compatibility) |
| 1895 |
// ======================================== |
| 1896 |
|
| 1897 |
/** |
| 1898 |
* Get singleton instance |
| 1899 |
*/ |
| 1900 |
public static function get_instance() { |
| 1901 |
static $instance = null; |
| 1902 |
if ($instance === null) { |
| 1903 |
$instance = new self(); |
| 1904 |
} |
| 1905 |
return $instance; |
| 1906 |
} |
| 1907 |
} |
| 1908 |
|
| 1909 |
// Initialize the Pinecone manager |
| 1910 |
$mxchat_pinecone_manager = MxChat_Pinecone_Manager::get_instance(); |
| 1911 |
|
