| 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 |
// Hook into WordPress actions if needed |
| 19 |
add_action('mxchat_delete_content', array($this, 'mxchat_delete_from_pinecone_by_url'), 10, 1); |
| 20 |
} |
| 21 |
|
| 22 |
// ======================================== |
| 23 |
// PINECONE FETCH OPERATIONS |
| 24 |
// ======================================== |
| 25 |
|
| 26 |
|
| 27 |
/** |
| 28 |
* Fetches 1K most recent records from Pinecone (SIMPLIFIED VERSION) |
| 29 |
*/ |
| 30 |
public function mxchat_fetch_pinecone_records($pinecone_options, $search_query = '', $page = 1, $per_page = 20) { |
| 31 |
//error_log('=== DEBUG: mxchat_fetch_pinecone_records started (simplified 1K) ==='); |
| 32 |
|
| 33 |
$api_key = $pinecone_options['mxchat_pinecone_api_key'] ?? ''; |
| 34 |
$host = $pinecone_options['mxchat_pinecone_host'] ?? ''; |
| 35 |
|
| 36 |
if (empty($api_key) || empty($host)) { |
| 37 |
//error_log('DEBUG: Missing required Pinecone parameters'); |
| 38 |
return array('data' => array(), 'total' => 0, 'total_in_database' => 0, 'showing_recent_only' => false); |
| 39 |
} |
| 40 |
|
| 41 |
try { |
| 42 |
// Get total count for the banner message |
| 43 |
$total_in_database = $this->mxchat_get_pinecone_total_count($pinecone_options); |
| 44 |
|
| 45 |
// Get 1K most recent entries |
| 46 |
$all_records = $this->mxchat_get_recent_1k_entries($pinecone_options); |
| 47 |
|
| 48 |
// Filter by search query if provided |
| 49 |
if (!empty($search_query)) { |
| 50 |
$all_records = array_filter($all_records, function($record) use ($search_query) { |
| 51 |
$content = $record->article_content ?? ''; |
| 52 |
$source_url = $record->source_url ?? ''; |
| 53 |
return stripos($content, $search_query) !== false || stripos($source_url, $search_query) !== false; |
| 54 |
}); |
| 55 |
} |
| 56 |
|
| 57 |
// Handle pagination |
| 58 |
$total = count($all_records); |
| 59 |
$offset = ($page - 1) * $per_page; |
| 60 |
$paged_records = array_slice($all_records, $offset, $per_page); |
| 61 |
|
| 62 |
//error_log('DEBUG: Returning ' . count($paged_records) . ' records (page ' . $page . ' of ' . ceil($total / $per_page) . ')'); |
| 63 |
|
| 64 |
return array( |
| 65 |
'data' => $paged_records, |
| 66 |
'total' => $total, |
| 67 |
'total_in_database' => $total_in_database, |
| 68 |
'showing_recent_only' => ($total_in_database > 1000) |
| 69 |
); |
| 70 |
|
| 71 |
} catch (Exception $e) { |
| 72 |
//error_log('DEBUG: Exception: ' . $e->getMessage()); |
| 73 |
return array('data' => array(), 'total' => 0, 'total_in_database' => 0, 'showing_recent_only' => false); |
| 74 |
} |
| 75 |
} |
| 76 |
|
| 77 |
/** |
| 78 |
* Get embedding dimensions based on the selected model |
| 79 |
* ADD THIS NEW FUNCTION |
| 80 |
*/ |
| 81 |
private function mxchat_get_embedding_dimensions() { |
| 82 |
$options = get_option('mxchat_options', array()); |
| 83 |
$selected_model = $options['embedding_model'] ?? 'text-embedding-ada-002'; |
| 84 |
|
| 85 |
// Define dimensions for different models |
| 86 |
$model_dimensions = array( |
| 87 |
'text-embedding-ada-002' => 1536, |
| 88 |
'text-embedding-3-small' => 1536, |
| 89 |
'text-embedding-3-large' => 3072, |
| 90 |
'voyage-2' => 1024, |
| 91 |
'voyage-large-2' => 1536, |
| 92 |
'voyage-3-large' => 2048, |
| 93 |
'gemini-embedding-001' => 1536, |
| 94 |
); |
| 95 |
|
| 96 |
// Check if it's a voyage model with custom dimensions |
| 97 |
if (strpos($selected_model, 'voyage-3-large') === 0) { |
| 98 |
$custom_dimensions = $options['voyage_output_dimension'] ?? 2048; |
| 99 |
return intval($custom_dimensions); |
| 100 |
} |
| 101 |
|
| 102 |
// Check if it's a gemini model with custom dimensions |
| 103 |
if (strpos($selected_model, 'gemini-embedding') === 0) { |
| 104 |
$custom_dimensions = $options['gemini_output_dimension'] ?? 1536; |
| 105 |
return intval($custom_dimensions); |
| 106 |
} |
| 107 |
|
| 108 |
// Return known dimensions or default to 1536 |
| 109 |
return $model_dimensions[$selected_model] ?? 1536; |
| 110 |
} |
| 111 |
|
| 112 |
/** |
| 113 |
* Generate random unit vector with correct dimensions |
| 114 |
* ADD THIS NEW FUNCTION |
| 115 |
*/ |
| 116 |
private function mxchat_generate_random_vector() { |
| 117 |
$dimensions = $this->mxchat_get_embedding_dimensions(); |
| 118 |
|
| 119 |
$random_vector = array(); |
| 120 |
for ($i = 0; $i < $dimensions; $i++) { |
| 121 |
$random_vector[] = (rand(-1000, 1000) / 1000.0); |
| 122 |
} |
| 123 |
|
| 124 |
// Normalize the vector to unit length |
| 125 |
$magnitude = sqrt(array_sum(array_map(function($x) { return $x * $x; }, $random_vector))); |
| 126 |
if ($magnitude > 0) { |
| 127 |
$random_vector = array_map(function($x) use ($magnitude) { return $x / $magnitude; }, $random_vector); |
| 128 |
} |
| 129 |
|
| 130 |
return $random_vector; |
| 131 |
} |
| 132 |
|
| 133 |
/** |
| 134 |
* Get 1,000 most recent entries from Pinecone (SIMPLE VERSION) |
| 135 |
*/ |
| 136 |
private function mxchat_get_recent_1k_entries($pinecone_options) { |
| 137 |
//error_log('=== DEBUG: mxchat_get_recent_1k_entries started ==='); |
| 138 |
|
| 139 |
$api_key = $pinecone_options['mxchat_pinecone_api_key'] ?? ''; |
| 140 |
$host = $pinecone_options['mxchat_pinecone_host'] ?? ''; |
| 141 |
|
| 142 |
if (empty($api_key) || empty($host)) { |
| 143 |
return array(); |
| 144 |
} |
| 145 |
|
| 146 |
try { |
| 147 |
// Fetch multiple batches to get a good sample, then sort by date |
| 148 |
$all_records = array(); |
| 149 |
$seen_ids = array(); |
| 150 |
$query_url = "https://{$host}/query"; |
| 151 |
|
| 152 |
// Do 5 passes to get a good sample of records |
| 153 |
for ($pass = 0; $pass < 5; $pass++) { |
| 154 |
//error_log('DEBUG: Fetching batch ' . ($pass + 1) . '/5'); |
| 155 |
|
| 156 |
// Generate random vector with CORRECT dimensions |
| 157 |
$random_vector = $this->mxchat_generate_random_vector(); |
| 158 |
|
| 159 |
$query_data = array( |
| 160 |
'includeMetadata' => true, |
| 161 |
'includeValues' => false, |
| 162 |
'topK' => 2000, // Get 2K per batch |
| 163 |
'vector' => $random_vector |
| 164 |
); |
| 165 |
|
| 166 |
$response = wp_remote_post($query_url, array( |
| 167 |
'headers' => array( |
| 168 |
'Api-Key' => $api_key, |
| 169 |
'Content-Type' => 'application/json' |
| 170 |
), |
| 171 |
'body' => json_encode($query_data), |
| 172 |
'timeout' => 30 |
| 173 |
)); |
| 174 |
|
| 175 |
if (!is_wp_error($response) && wp_remote_retrieve_response_code($response) === 200) { |
| 176 |
$body = wp_remote_retrieve_body($response); |
| 177 |
$data = json_decode($body, true); |
| 178 |
|
| 179 |
if (isset($data['matches'])) { |
| 180 |
foreach ($data['matches'] as $match) { |
| 181 |
$match_id = $match['id'] ?? ''; |
| 182 |
if (!empty($match_id) && !isset($seen_ids[$match_id])) { |
| 183 |
$metadata = $match['metadata'] ?? array(); |
| 184 |
$created_at = $metadata['created_at'] ?? $metadata['last_updated'] ?? time(); |
| 185 |
|
| 186 |
// Ensure valid timestamp |
| 187 |
if (!is_numeric($created_at)) { |
| 188 |
$created_at = strtotime($created_at) ?: time(); |
| 189 |
} |
| 190 |
|
| 191 |
$all_records[] = (object) array( |
| 192 |
'id' => $match_id, |
| 193 |
'article_content' => $metadata['text'] ?? '', |
| 194 |
'source_url' => $metadata['source_url'] ?? '', |
| 195 |
'created_at' => $created_at, |
| 196 |
'data_source' => 'pinecone' |
| 197 |
); |
| 198 |
|
| 199 |
$seen_ids[$match_id] = true; |
| 200 |
} |
| 201 |
} |
| 202 |
} |
| 203 |
} |
| 204 |
|
| 205 |
usleep(100000); // 0.1 second delay between batches |
| 206 |
} |
| 207 |
|
| 208 |
// Sort by created_at (newest first) and take top 1K |
| 209 |
usort($all_records, function($a, $b) { |
| 210 |
return $b->created_at - $a->created_at; |
| 211 |
}); |
| 212 |
|
| 213 |
$recent_1k = array_slice($all_records, 0, 1000); |
| 214 |
|
| 215 |
//error_log('DEBUG: Found ' . count($all_records) . ' total records, returning top ' . count($recent_1k)); |
| 216 |
|
| 217 |
return $recent_1k; |
| 218 |
|
| 219 |
} catch (Exception $e) { |
| 220 |
//error_log('DEBUG: Exception in get_recent_1k_entries: ' . $e->getMessage()); |
| 221 |
return array(); |
| 222 |
} |
| 223 |
} |
| 224 |
|
| 225 |
/** |
| 226 |
* Scan Pinecone for processed content (MISSING FUNCTION - ADD THIS) |
| 227 |
*/ |
| 228 |
public function mxchat_scan_pinecone_for_processed_content($pinecone_options) { |
| 229 |
//error_log('=== DEBUG: Starting mxchat_scan_pinecone_for_processed_content ==='); |
| 230 |
|
| 231 |
$api_key = $pinecone_options['mxchat_pinecone_api_key'] ?? ''; |
| 232 |
$host = $pinecone_options['mxchat_pinecone_host'] ?? ''; |
| 233 |
|
| 234 |
if (empty($api_key) || empty($host)) { |
| 235 |
//error_log('DEBUG: Missing API credentials for scanning'); |
| 236 |
return array(); |
| 237 |
} |
| 238 |
|
| 239 |
try { |
| 240 |
// Use multiple random vectors to get better coverage |
| 241 |
$all_matches = array(); |
| 242 |
$seen_ids = array(); |
| 243 |
|
| 244 |
// Try 3 different random vectors to get better coverage |
| 245 |
for ($i = 0; $i < 3; $i++) { |
| 246 |
//error_log('DEBUG: Scanning attempt ' . ($i + 1) . '/3'); |
| 247 |
|
| 248 |
$query_url = "https://{$host}/query"; |
| 249 |
|
| 250 |
// Generate random vector with CORRECT dimensions |
| 251 |
$random_vector = $this->mxchat_generate_random_vector(); |
| 252 |
|
| 253 |
$query_data = array( |
| 254 |
'includeMetadata' => true, |
| 255 |
'includeValues' => false, |
| 256 |
'topK' => 10000, |
| 257 |
'vector' => $random_vector |
| 258 |
); |
| 259 |
|
| 260 |
$response = wp_remote_post($query_url, array( |
| 261 |
'headers' => array( |
| 262 |
'Api-Key' => $api_key, |
| 263 |
'Content-Type' => 'application/json' |
| 264 |
), |
| 265 |
'body' => json_encode($query_data), |
| 266 |
'timeout' => 30 |
| 267 |
)); |
| 268 |
|
| 269 |
if (is_wp_error($response)) { |
| 270 |
//error_log('DEBUG: Query attempt ' . ($i + 1) . ' WP error: ' . $response->get_error_message()); |
| 271 |
continue; |
| 272 |
} |
| 273 |
|
| 274 |
$response_code = wp_remote_retrieve_response_code($response); |
| 275 |
//error_log('DEBUG: Query attempt ' . ($i + 1) . ' response code: ' . $response_code); |
| 276 |
|
| 277 |
if ($response_code !== 200) { |
| 278 |
$error_body = wp_remote_retrieve_body($response); |
| 279 |
//error_log('DEBUG: Query attempt ' . ($i + 1) . ' failed with body: ' . substr($error_body, 0, 500)); |
| 280 |
continue; |
| 281 |
} |
| 282 |
|
| 283 |
$body = wp_remote_retrieve_body($response); |
| 284 |
$data = json_decode($body, true); |
| 285 |
|
| 286 |
if (isset($data['matches'])) { |
| 287 |
//error_log('DEBUG: Query attempt ' . ($i + 1) . ' returned ' . count($data['matches']) . ' matches'); |
| 288 |
foreach ($data['matches'] as $match) { |
| 289 |
$match_id = $match['id'] ?? ''; |
| 290 |
if (!empty($match_id) && !isset($seen_ids[$match_id])) { |
| 291 |
$all_matches[] = $match; |
| 292 |
$seen_ids[$match_id] = true; |
| 293 |
} |
| 294 |
} |
| 295 |
} else { |
| 296 |
//error_log('DEBUG: Query attempt ' . ($i + 1) . ' - no matches key in response'); |
| 297 |
} |
| 298 |
} |
| 299 |
|
| 300 |
//error_log('DEBUG: Total unique matches found: ' . count($all_matches)); |
| 301 |
|
| 302 |
// Convert matches to processed data format |
| 303 |
$processed_data = array(); |
| 304 |
$vector_ids_for_cache = array(); |
| 305 |
|
| 306 |
foreach ($all_matches as $match) { |
| 307 |
$metadata = $match['metadata'] ?? array(); |
| 308 |
$source_url = $metadata['source_url'] ?? ''; |
| 309 |
$match_id = $match['id'] ?? ''; |
| 310 |
|
| 311 |
if (!empty($source_url) && !empty($match_id)) { |
| 312 |
$post_id = url_to_postid($source_url); |
| 313 |
if ($post_id) { |
| 314 |
$created_at = $metadata['created_at'] ?? ''; |
| 315 |
$processed_date = 'Recently'; |
| 316 |
|
| 317 |
if (!empty($created_at)) { |
| 318 |
$timestamp = is_numeric($created_at) ? $created_at : strtotime($created_at); |
| 319 |
if ($timestamp) { |
| 320 |
$processed_date = human_time_diff($timestamp, current_time('timestamp')) . ' ago'; |
| 321 |
} |
| 322 |
} |
| 323 |
|
| 324 |
$processed_data[$post_id] = array( |
| 325 |
'db_id' => $match_id, |
| 326 |
'processed_date' => $processed_date, |
| 327 |
'url' => $source_url, |
| 328 |
'source' => 'pinecone', |
| 329 |
'timestamp' => $timestamp ?? current_time('timestamp') |
| 330 |
); |
| 331 |
|
| 332 |
$vector_ids_for_cache[] = $match_id; |
| 333 |
} |
| 334 |
} |
| 335 |
} |
| 336 |
|
| 337 |
// Update the vector IDs cache for future use |
| 338 |
if (!empty($vector_ids_for_cache)) { |
| 339 |
update_option('mxchat_pinecone_vector_ids_cache', $vector_ids_for_cache); |
| 340 |
//error_log('DEBUG: Updated vector IDs cache with ' . count($vector_ids_for_cache) . ' IDs'); |
| 341 |
} |
| 342 |
|
| 343 |
//error_log('DEBUG: Returning ' . count($processed_data) . ' processed items from scanning'); |
| 344 |
return $processed_data; |
| 345 |
|
| 346 |
} catch (Exception $e) { |
| 347 |
//error_log('DEBUG: Exception in scan_pinecone_for_processed_content: ' . $e->getMessage()); |
| 348 |
return array(); |
| 349 |
} |
| 350 |
} |
| 351 |
|
| 352 |
/** |
| 353 |
* Get total count of vectors in Pinecone (SIMPLE VERSION) |
| 354 |
*/ |
| 355 |
private function mxchat_get_pinecone_total_count($pinecone_options) { |
| 356 |
// First try the stats API |
| 357 |
$api_key = $pinecone_options['mxchat_pinecone_api_key'] ?? ''; |
| 358 |
$host = $pinecone_options['mxchat_pinecone_host'] ?? ''; |
| 359 |
|
| 360 |
if (empty($api_key) || empty($host)) { |
| 361 |
return 0; |
| 362 |
} |
| 363 |
|
| 364 |
try { |
| 365 |
$stats_url = "https://{$host}/describe_index_stats"; |
| 366 |
|
| 367 |
// Try GET request |
| 368 |
$response = wp_remote_get($stats_url, array( |
| 369 |
'headers' => array( |
| 370 |
'Api-Key' => $api_key, |
| 371 |
'Accept' => 'application/json' |
| 372 |
), |
| 373 |
'timeout' => 15 |
| 374 |
)); |
| 375 |
|
| 376 |
if (!is_wp_error($response) && wp_remote_retrieve_response_code($response) === 200) { |
| 377 |
$body = wp_remote_retrieve_body($response); |
| 378 |
$stats_data = json_decode($body, true); |
| 379 |
|
| 380 |
$total_count = $stats_data['totalVectorCount'] ?? 0; |
| 381 |
if ($total_count > 0) { |
| 382 |
//error_log('DEBUG: Got total count from stats API: ' . $total_count); |
| 383 |
return intval($total_count); |
| 384 |
} |
| 385 |
} |
| 386 |
|
| 387 |
// Fallback: estimate from previous scans |
| 388 |
$cached_vector_ids = get_option('mxchat_pinecone_vector_ids_cache', array()); |
| 389 |
if (!empty($cached_vector_ids)) { |
| 390 |
$estimated_count = count($cached_vector_ids); |
| 391 |
//error_log('DEBUG: Using estimated count from cache: ' . $estimated_count); |
| 392 |
return intval($estimated_count); |
| 393 |
} |
| 394 |
|
| 395 |
} catch (Exception $e) { |
| 396 |
//error_log('DEBUG: Exception getting total count: ' . $e->getMessage()); |
| 397 |
} |
| 398 |
|
| 399 |
// If all else fails, return 0 |
| 400 |
return 0; |
| 401 |
} |
| 402 |
|
| 403 |
/** |
| 404 |
* Call this after adding new content to refresh the view |
| 405 |
*/ |
| 406 |
public function mxchat_refresh_after_new_content($pinecone_options) { |
| 407 |
//error_log('DEBUG: Refreshing after new content added'); |
| 408 |
|
| 409 |
// Clear any old caches |
| 410 |
delete_transient('mxchat_pinecone_recent_1k'); |
| 411 |
delete_transient('mxchat_pinecone_total_count'); |
| 412 |
|
| 413 |
// Force fresh fetch on next page load |
| 414 |
// The next call to mxchat_fetch_pinecone_records will get fresh data |
| 415 |
|
| 416 |
return true; |
| 417 |
} |
| 418 |
|
| 419 |
|
| 420 |
/** |
| 421 |
* Fetches vectors from Pinecone using provided IDs (for content selection feature) |
| 422 |
*/ |
| 423 |
public function fetch_pinecone_vectors_by_ids($pinecone_options, $vector_ids) { |
| 424 |
//error_log('=== DEBUG: fetch_pinecone_vectors_by_ids started (content selection method) ==='); |
| 425 |
|
| 426 |
$api_key = $pinecone_options['mxchat_pinecone_api_key'] ?? ''; |
| 427 |
$host = $pinecone_options['mxchat_pinecone_host'] ?? ''; |
| 428 |
|
| 429 |
//error_log('DEBUG: API key present: ' . (!empty($api_key) ? 'YES' : 'NO')); |
| 430 |
//error_log('DEBUG: Host: ' . $host); |
| 431 |
//error_log('DEBUG: Vector IDs count: ' . count($vector_ids)); |
| 432 |
|
| 433 |
if (empty($api_key) || empty($host) || empty($vector_ids)) { |
| 434 |
//error_log('DEBUG: Missing parameters for fetch by IDs (content selection)'); |
| 435 |
return array(); |
| 436 |
} |
| 437 |
|
| 438 |
try { |
| 439 |
$fetch_url = "https://{$host}/vectors/fetch"; |
| 440 |
//error_log('DEBUG: Fetch URL: ' . $fetch_url); |
| 441 |
|
| 442 |
// Pinecone fetch API allows fetching specific vectors by ID |
| 443 |
$fetch_data = array( |
| 444 |
'ids' => array_values($vector_ids) |
| 445 |
); |
| 446 |
|
| 447 |
$response = wp_remote_post($fetch_url, array( |
| 448 |
'headers' => array( |
| 449 |
'Api-Key' => $api_key, |
| 450 |
'Content-Type' => 'application/json' |
| 451 |
), |
| 452 |
'body' => json_encode($fetch_data), |
| 453 |
'timeout' => 30 |
| 454 |
)); |
| 455 |
|
| 456 |
if (is_wp_error($response)) { |
| 457 |
//error_log('DEBUG: Fetch by IDs WP error (content selection): ' . $response->get_error_message()); |
| 458 |
return array(); |
| 459 |
} |
| 460 |
|
| 461 |
$response_code = wp_remote_retrieve_response_code($response); |
| 462 |
//error_log('DEBUG: Fetch response code (content selection): ' . $response_code); |
| 463 |
|
| 464 |
if ($response_code !== 200) { |
| 465 |
$error_body = wp_remote_retrieve_body($response); |
| 466 |
//error_log('DEBUG: Fetch failed with body (content selection): ' . $error_body); |
| 467 |
return array(); |
| 468 |
} |
| 469 |
|
| 470 |
$body = wp_remote_retrieve_body($response); |
| 471 |
$data = json_decode($body, true); |
| 472 |
|
| 473 |
if (!isset($data['vectors'])) { |
| 474 |
//error_log('DEBUG: No vectors key in response (content selection)'); |
| 475 |
return array(); |
| 476 |
} |
| 477 |
|
| 478 |
$processed_data = array(); |
| 479 |
|
| 480 |
foreach ($data['vectors'] as $vector_id => $vector_data) { |
| 481 |
$metadata = $vector_data['metadata'] ?? array(); |
| 482 |
$source_url = $metadata['source_url'] ?? ''; |
| 483 |
|
| 484 |
if (!empty($source_url)) { |
| 485 |
$post_id = url_to_postid($source_url); |
| 486 |
if ($post_id) { |
| 487 |
$created_at = $metadata['created_at'] ?? ''; |
| 488 |
$processed_date = 'Recently'; // Default |
| 489 |
|
| 490 |
if (!empty($created_at)) { |
| 491 |
$timestamp = is_numeric($created_at) ? $created_at : strtotime($created_at); |
| 492 |
if ($timestamp) { |
| 493 |
$processed_date = human_time_diff($timestamp, current_time('timestamp')) . ' ago'; |
| 494 |
} |
| 495 |
} |
| 496 |
|
| 497 |
$processed_data[$post_id] = array( |
| 498 |
'db_id' => $vector_id, |
| 499 |
'processed_date' => $processed_date, |
| 500 |
'url' => $source_url, |
| 501 |
'source' => 'pinecone', |
| 502 |
'timestamp' => $timestamp ?? current_time('timestamp') |
| 503 |
); |
| 504 |
} |
| 505 |
} |
| 506 |
} |
| 507 |
|
| 508 |
//error_log('DEBUG: Processed ' . count($processed_data) . ' records (content selection method)'); |
| 509 |
//error_log('=== DEBUG: fetch_pinecone_vectors_by_ids completed (content selection) ==='); |
| 510 |
|
| 511 |
return $processed_data; |
| 512 |
|
| 513 |
} catch (Exception $e) { |
| 514 |
//error_log('DEBUG: Exception in fetch_pinecone_vectors_by_ids (content selection): ' . $e->getMessage()); |
| 515 |
return array(); |
| 516 |
} |
| 517 |
} |
| 518 |
|
| 519 |
|
| 520 |
// ======================================== |
| 521 |
// PINECONE DELETE OPERATIONS |
| 522 |
// ======================================== |
| 523 |
|
| 524 |
/** |
| 525 |
* Deletes data from Pinecone using provided API credentials |
| 526 |
*/ |
| 527 |
public function mxchat_delete_all_from_pinecone($pinecone_options) { |
| 528 |
$api_key = $pinecone_options['mxchat_pinecone_api_key'] ?? ''; |
| 529 |
$host = $pinecone_options['mxchat_pinecone_host'] ?? ''; |
| 530 |
|
| 531 |
if (empty($api_key) || empty($host)) { |
| 532 |
return array( |
| 533 |
'success' => false, |
| 534 |
'message' => 'Missing Pinecone API credentials' |
| 535 |
); |
| 536 |
} |
| 537 |
|
| 538 |
try { |
| 539 |
// First, get all vector IDs |
| 540 |
$all_vector_ids = array(); |
| 541 |
|
| 542 |
// Try to get from cache first |
| 543 |
$cached_vector_ids = get_option('mxchat_pinecone_vector_ids_cache', array()); |
| 544 |
if (!empty($cached_vector_ids)) { |
| 545 |
$all_vector_ids = $cached_vector_ids; |
| 546 |
} else { |
| 547 |
// Fallback: scan to get vector IDs |
| 548 |
$records = $this->mxchat_scan_pinecone_vectors($pinecone_options); |
| 549 |
foreach ($records as $record) { |
| 550 |
if (!empty($record->id)) { |
| 551 |
$all_vector_ids[] = $record->id; |
| 552 |
} |
| 553 |
} |
| 554 |
} |
| 555 |
|
| 556 |
if (empty($all_vector_ids)) { |
| 557 |
return array( |
| 558 |
'success' => true, |
| 559 |
'message' => 'No vectors found to delete' |
| 560 |
); |
| 561 |
} |
| 562 |
|
| 563 |
// Delete vectors in batches (Pinecone has limits on batch operations) |
| 564 |
$batch_size = 100; |
| 565 |
$batches = array_chunk($all_vector_ids, $batch_size); |
| 566 |
$deleted_count = 0; |
| 567 |
$failed_batches = 0; |
| 568 |
|
| 569 |
foreach ($batches as $batch) { |
| 570 |
$result = $this->mxchat_delete_pinecone_batch($batch, $api_key, $host); |
| 571 |
if ($result['success']) { |
| 572 |
$deleted_count += count($batch); |
| 573 |
} else { |
| 574 |
$failed_batches++; |
| 575 |
//error_log('Failed to delete Pinecone batch: ' . $result['message']); |
| 576 |
} |
| 577 |
} |
| 578 |
|
| 579 |
if ($failed_batches > 0) { |
| 580 |
return array( |
| 581 |
'success' => false, |
| 582 |
'message' => sprintf('Deleted %d vectors, but %d batches failed', $deleted_count, $failed_batches) |
| 583 |
); |
| 584 |
} |
| 585 |
|
| 586 |
return array( |
| 587 |
'success' => true, |
| 588 |
'message' => "Successfully deleted {$deleted_count} vectors from Pinecone" |
| 589 |
); |
| 590 |
|
| 591 |
} catch (Exception $e) { |
| 592 |
//error_log('Pinecone delete all exception: ' . $e->getMessage()); |
| 593 |
return array( |
| 594 |
'success' => false, |
| 595 |
'message' => $e->getMessage() |
| 596 |
); |
| 597 |
} |
| 598 |
} |
| 599 |
|
| 600 |
|
| 601 |
/** |
| 602 |
* Deletes batch of vectors from Pinecone database |
| 603 |
*/ |
| 604 |
private function mxchat_delete_pinecone_batch($vector_ids, $api_key, $host) { |
| 605 |
// Build the API endpoint |
| 606 |
$api_endpoint = "https://{$host}/vectors/delete"; |
| 607 |
|
| 608 |
// Prepare the request body with the IDs |
| 609 |
$request_body = array( |
| 610 |
'ids' => $vector_ids |
| 611 |
); |
| 612 |
|
| 613 |
// Make the deletion request |
| 614 |
$response = wp_remote_post($api_endpoint, array( |
| 615 |
'headers' => array( |
| 616 |
'Api-Key' => $api_key, |
| 617 |
'accept' => 'application/json', |
| 618 |
'content-type' => 'application/json' |
| 619 |
), |
| 620 |
'body' => wp_json_encode($request_body), |
| 621 |
'timeout' => 60, // Increased timeout for batch operations |
| 622 |
'method' => 'POST' |
| 623 |
)); |
| 624 |
|
| 625 |
// Handle WordPress HTTP API errors |
| 626 |
if (is_wp_error($response)) { |
| 627 |
return array( |
| 628 |
'success' => false, |
| 629 |
'message' => $response->get_error_message() |
| 630 |
); |
| 631 |
} |
| 632 |
|
| 633 |
// Check response status |
| 634 |
$response_code = wp_remote_retrieve_response_code($response); |
| 635 |
$response_body = wp_remote_retrieve_body($response); |
| 636 |
|
| 637 |
// Pinecone returns 200 for successful deletion |
| 638 |
if ($response_code !== 200) { |
| 639 |
//error_log('Pinecone batch deletion failed: HTTP ' . $response_code . ' - ' . $response_body); |
| 640 |
return array( |
| 641 |
'success' => false, |
| 642 |
'message' => sprintf( |
| 643 |
'Pinecone API error (HTTP %d): %s', |
| 644 |
$response_code, |
| 645 |
$response_body |
| 646 |
) |
| 647 |
); |
| 648 |
} |
| 649 |
|
| 650 |
return array( |
| 651 |
'success' => true, |
| 652 |
'message' => 'Batch deleted successfully from Pinecone' |
| 653 |
); |
| 654 |
} |
| 655 |
|
| 656 |
|
| 657 |
/** |
| 658 |
* Deletes vector from Pinecone using API request |
| 659 |
*/ |
| 660 |
public function mxchat_delete_from_pinecone_by_vector_id($vector_id, $api_key, $host) { |
| 661 |
// Build the API endpoint |
| 662 |
$api_endpoint = "https://{$host}/vectors/delete"; |
| 663 |
|
| 664 |
// Prepare the request body with just the ID |
| 665 |
$request_body = array( |
| 666 |
'ids' => array($vector_id) |
| 667 |
); |
| 668 |
|
| 669 |
// Make the deletion request |
| 670 |
$response = wp_remote_post($api_endpoint, array( |
| 671 |
'headers' => array( |
| 672 |
'Api-Key' => $api_key, |
| 673 |
'accept' => 'application/json', |
| 674 |
'content-type' => 'application/json' |
| 675 |
), |
| 676 |
'body' => wp_json_encode($request_body), |
| 677 |
'timeout' => 30, |
| 678 |
'method' => 'POST' |
| 679 |
)); |
| 680 |
|
| 681 |
// Handle WordPress HTTP API errors |
| 682 |
if (is_wp_error($response)) { |
| 683 |
return array( |
| 684 |
'success' => false, |
| 685 |
'message' => $response->get_error_message() |
| 686 |
); |
| 687 |
} |
| 688 |
|
| 689 |
// Check response status |
| 690 |
$response_code = wp_remote_retrieve_response_code($response); |
| 691 |
$response_body = wp_remote_retrieve_body($response); |
| 692 |
|
| 693 |
// Pinecone returns 200 for successful deletion |
| 694 |
if ($response_code !== 200) { |
| 695 |
//error_log('Pinecone deletion failed: HTTP ' . $response_code . ' - ' . $response_body); |
| 696 |
return array( |
| 697 |
'success' => false, |
| 698 |
'message' => sprintf( |
| 699 |
'Pinecone API error (HTTP %d): %s', |
| 700 |
$response_code, |
| 701 |
$response_body |
| 702 |
) |
| 703 |
); |
| 704 |
} |
| 705 |
|
| 706 |
// Parse response to check if it was successful |
| 707 |
$response_data = json_decode($response_body, true); |
| 708 |
|
| 709 |
// Log successful deletion |
| 710 |
//error_log('Pinecone vector ' . $vector_id . ' deleted successfully'); |
| 711 |
|
| 712 |
return array( |
| 713 |
'success' => true, |
| 714 |
'message' => 'Vector deleted successfully from Pinecone' |
| 715 |
); |
| 716 |
} |
| 717 |
|
| 718 |
|
| 719 |
/** |
| 720 |
* Deletes data from Pinecone using a source URL |
| 721 |
*/ |
| 722 |
public function mxchat_delete_from_pinecone_by_url($source_url, $pinecone_options) { |
| 723 |
$host = $pinecone_options['mxchat_pinecone_host'] ?? ''; |
| 724 |
$api_key = $pinecone_options['mxchat_pinecone_api_key'] ?? ''; |
| 725 |
|
| 726 |
if (empty($host) || empty($api_key)) { |
| 727 |
//error_log('MXChat: Pinecone deletion failed - missing configuration'); |
| 728 |
return false; |
| 729 |
} |
| 730 |
|
| 731 |
$api_endpoint = "https://{$host}/vectors/delete"; |
| 732 |
$vector_id = md5($source_url); |
| 733 |
|
| 734 |
$request_body = array( |
| 735 |
'ids' => array($vector_id) |
| 736 |
); |
| 737 |
|
| 738 |
$response = wp_remote_post($api_endpoint, array( |
| 739 |
'headers' => array( |
| 740 |
'Api-Key' => $api_key, |
| 741 |
'accept' => 'application/json', |
| 742 |
'content-type' => 'application/json' |
| 743 |
), |
| 744 |
'body' => wp_json_encode($request_body), |
| 745 |
'timeout' => 30 |
| 746 |
)); |
| 747 |
|
| 748 |
if (is_wp_error($response)) { |
| 749 |
//error_log('MXChat: Pinecone deletion error - ' . $response->get_error_message()); |
| 750 |
return false; |
| 751 |
} |
| 752 |
|
| 753 |
$response_code = wp_remote_retrieve_response_code($response); |
| 754 |
if ($response_code !== 200) { |
| 755 |
//error_log('MXChat: Pinecone deletion failed with status ' . $response_code); |
| 756 |
return false; |
| 757 |
} |
| 758 |
|
| 759 |
return true; |
| 760 |
} |
| 761 |
|
| 762 |
|
| 763 |
/** |
| 764 |
* Deletes data from Pinecone index using API key |
| 765 |
*/ |
| 766 |
private function mxchat_delete_from_pinecone($urls, $api_key, $environment, $index_name) { |
| 767 |
// Get the Pinecone host from options (matching your store_in_pinecone_main pattern) |
| 768 |
$options = get_option('mxchat_pinecone_addon_options'); |
| 769 |
$host = $options['mxchat_pinecone_host'] ?? ''; |
| 770 |
|
| 771 |
if (empty($host)) { |
| 772 |
return array( |
| 773 |
'success' => false, |
| 774 |
'message' => 'Pinecone host is not configured. Please set the host in your settings.' |
| 775 |
); |
| 776 |
} |
| 777 |
|
| 778 |
// Build API endpoint using the configured host |
| 779 |
$api_endpoint = "https://{$host}/vectors/delete"; |
| 780 |
|
| 781 |
// Create vector IDs from URLs (matching your store method's ID generation) |
| 782 |
$vector_ids = array_map('md5', $urls); |
| 783 |
|
| 784 |
// Prepare the delete request body |
| 785 |
$request_body = array( |
| 786 |
'ids' => $vector_ids, |
| 787 |
'filter' => array( |
| 788 |
'source_url' => array( |
| 789 |
'$in' => $urls |
| 790 |
) |
| 791 |
) |
| 792 |
); |
| 793 |
|
| 794 |
// Make the deletion request |
| 795 |
$response = wp_remote_post($api_endpoint, array( |
| 796 |
'headers' => array( |
| 797 |
'Api-Key' => $api_key, |
| 798 |
'accept' => 'application/json', |
| 799 |
'content-type' => 'application/json' |
| 800 |
), |
| 801 |
'body' => wp_json_encode($request_body), |
| 802 |
'timeout' => 30, |
| 803 |
'data_format' => 'body' |
| 804 |
)); |
| 805 |
|
| 806 |
// Handle WordPress HTTP API errors |
| 807 |
if (is_wp_error($response)) { |
| 808 |
return array( |
| 809 |
'success' => false, |
| 810 |
'message' => $response->get_error_message() |
| 811 |
); |
| 812 |
} |
| 813 |
|
| 814 |
// Check response status |
| 815 |
$response_code = wp_remote_retrieve_response_code($response); |
| 816 |
if ($response_code !== 200) { |
| 817 |
$body = wp_remote_retrieve_body($response); |
| 818 |
return array( |
| 819 |
'success' => false, |
| 820 |
'message' => sprintf( |
| 821 |
'Pinecone API error (HTTP %d): %s', |
| 822 |
$response_code, |
| 823 |
$body |
| 824 |
) |
| 825 |
); |
| 826 |
} |
| 827 |
|
| 828 |
// Parse response body |
| 829 |
$body = wp_remote_retrieve_body($response); |
| 830 |
$response_data = json_decode($body, true); |
| 831 |
|
| 832 |
// Final validation of the response |
| 833 |
if (json_last_error() !== JSON_ERROR_NONE) { |
| 834 |
return array( |
| 835 |
'success' => false, |
| 836 |
'message' => 'Failed to parse Pinecone response: ' . json_last_error_msg() |
| 837 |
); |
| 838 |
} |
| 839 |
|
| 840 |
return array( |
| 841 |
'success' => true, |
| 842 |
'message' => sprintf('Successfully deleted %d vectors from Pinecone', count($vector_ids)) |
| 843 |
); |
| 844 |
} |
| 845 |
|
| 846 |
|
| 847 |
// ======================================== |
| 848 |
// VECTOR CACHE MANAGEMENT |
| 849 |
// ======================================== |
| 850 |
|
| 851 |
|
| 852 |
/** |
| 853 |
* Removes vector ID from cache array option |
| 854 |
*/ |
| 855 |
public function mxchat_remove_from_pinecone_vector_cache($vector_id) { |
| 856 |
$cached_ids = get_option('mxchat_pinecone_vector_ids_cache', array()); |
| 857 |
$key = array_search($vector_id, $cached_ids); |
| 858 |
if ($key !== false) { |
| 859 |
unset($cached_ids[$key]); |
| 860 |
update_option('mxchat_pinecone_vector_ids_cache', array_values($cached_ids)); |
| 861 |
} |
| 862 |
} |
| 863 |
|
| 864 |
/** |
| 865 |
* Removes vector ID from processed content caches |
| 866 |
*/ |
| 867 |
public function mxchat_remove_from_processed_content_caches($vector_id) { |
| 868 |
// Get all caches |
| 869 |
$pinecone_cache = get_option('mxchat_pinecone_processed_cache', array()); |
| 870 |
$processed_cache = get_option('mxchat_processed_content_cache', array()); |
| 871 |
|
| 872 |
// We need to find the post ID that corresponds to this vector ID |
| 873 |
// Vector ID is typically md5 of the source URL |
| 874 |
$post_id_to_remove = null; |
| 875 |
|
| 876 |
// Search through caches to find matching post |
| 877 |
foreach ($pinecone_cache as $post_id => $cache_data) { |
| 878 |
if (isset($cache_data['db_id']) && $cache_data['db_id'] === $vector_id) { |
| 879 |
$post_id_to_remove = $post_id; |
| 880 |
break; |
| 881 |
} |
| 882 |
} |
| 883 |
|
| 884 |
// Also check the processed cache |
| 885 |
if (!$post_id_to_remove) { |
| 886 |
foreach ($processed_cache as $post_id => $cache_data) { |
| 887 |
if (isset($cache_data['db_id']) && $cache_data['db_id'] === $vector_id) { |
| 888 |
$post_id_to_remove = $post_id; |
| 889 |
break; |
| 890 |
} |
| 891 |
} |
| 892 |
} |
| 893 |
|
| 894 |
// If we found the post ID, remove it from both caches |
| 895 |
if ($post_id_to_remove) { |
| 896 |
unset($pinecone_cache[$post_id_to_remove]); |
| 897 |
unset($processed_cache[$post_id_to_remove]); |
| 898 |
|
| 899 |
update_option('mxchat_pinecone_processed_cache', $pinecone_cache); |
| 900 |
update_option('mxchat_processed_content_cache', $processed_cache); |
| 901 |
|
| 902 |
//error_log('Removed post ID ' . $post_id_to_remove . ' from processed content caches'); |
| 903 |
} else { |
| 904 |
// If we can't find by vector ID, we might need to reconstruct the URL |
| 905 |
// and find the post ID that way |
| 906 |
//error_log('Could not find post ID for vector ID: ' . $vector_id); |
| 907 |
} |
| 908 |
} |
| 909 |
|
| 910 |
|
| 911 |
/** |
| 912 |
* Retrieves and caches Pinecone API processed content |
| 913 |
*/ |
| 914 |
public function mxchat_get_pinecone_processed_content($pinecone_options) { |
| 915 |
// First check local cache for immediate updates |
| 916 |
$cached_data = get_option('mxchat_pinecone_processed_cache', array()); |
| 917 |
|
| 918 |
$api_key = $pinecone_options['mxchat_pinecone_api_key'] ?? ''; |
| 919 |
$host = $pinecone_options['mxchat_pinecone_host'] ?? ''; |
| 920 |
|
| 921 |
if (empty($api_key) || empty($host)) { |
| 922 |
// Return only cached data if API credentials are missing |
| 923 |
return $cached_data; |
| 924 |
} |
| 925 |
|
| 926 |
$pinecone_data = array(); |
| 927 |
|
| 928 |
try { |
| 929 |
// Method 1: Try to get vectors using cached vector IDs first |
| 930 |
$cached_vector_ids = get_option('mxchat_pinecone_vector_ids_cache', array()); |
| 931 |
|
| 932 |
if (!empty($cached_vector_ids)) { |
| 933 |
$pinecone_data = $this->fetch_pinecone_vectors_by_ids($pinecone_options, $cached_vector_ids); |
| 934 |
} |
| 935 |
|
| 936 |
// Method 2: If no cached IDs or fetch failed, use scanning approach |
| 937 |
if (empty($pinecone_data)) { |
| 938 |
$pinecone_data = $this->mxchat_scan_pinecone_for_processed_content($pinecone_options); |
| 939 |
} |
| 940 |
|
| 941 |
// Method 3: Final fallback - try stats endpoint (if available) |
| 942 |
if (empty($pinecone_data)) { |
| 943 |
$stats_url = "https://{$host}/describe_index_stats"; |
| 944 |
|
| 945 |
$response = wp_remote_post($stats_url, array( |
| 946 |
'headers' => array( |
| 947 |
'Api-Key' => $api_key, |
| 948 |
'Content-Type' => 'application/json' |
| 949 |
), |
| 950 |
'body' => json_encode(array()), |
| 951 |
'timeout' => 30 |
| 952 |
)); |
| 953 |
|
| 954 |
if (!is_wp_error($response) && wp_remote_retrieve_response_code($response) === 200) { |
| 955 |
$body = wp_remote_retrieve_body($response); |
| 956 |
$stats_data = json_decode($body, true); |
| 957 |
|
| 958 |
// Log stats for debugging but don't rely on them for vector listing |
| 959 |
//error_log('Pinecone index stats: ' . print_r($stats_data, true)); |
| 960 |
} |
| 961 |
} |
| 962 |
|
| 963 |
} catch (Exception $e) { |
| 964 |
//error_log('Pinecone processed content exception: ' . $e->getMessage()); |
| 965 |
} |
| 966 |
|
| 967 |
// Merge cached data with Pinecone data |
| 968 |
// Cache takes priority for recent updates (within last 5 minutes) |
| 969 |
$merged_data = $pinecone_data; |
| 970 |
|
| 971 |
foreach ($cached_data as $post_id => $cache_item) { |
| 972 |
$cache_timestamp = $cache_item['timestamp'] ?? 0; |
| 973 |
$time_diff = current_time('timestamp') - $cache_timestamp; |
| 974 |
|
| 975 |
// If cache item is recent (less than 5 minutes), prioritize it |
| 976 |
if ($time_diff < 300) { // 5 minutes = 300 seconds |
| 977 |
$merged_data[$post_id] = $cache_item; |
| 978 |
} else { |
| 979 |
// If not in Pinecone data and cache is old, keep cache but mark as potentially stale |
| 980 |
if (!isset($merged_data[$post_id])) { |
| 981 |
$merged_data[$post_id] = $cache_item; |
| 982 |
} |
| 983 |
} |
| 984 |
} |
| 985 |
|
| 986 |
return $merged_data; |
| 987 |
} |
| 988 |
|
| 989 |
|
| 990 |
// ======================================== |
| 991 |
// HELPER METHODS |
| 992 |
// ======================================== |
| 993 |
|
| 994 |
/** |
| 995 |
* Validates Pinecone API credentials |
| 996 |
*/ |
| 997 |
private function mxchat_validate_pinecone_credentials($api_key, $host) { |
| 998 |
if (empty($api_key) || empty($host)) { |
| 999 |
return false; |
| 1000 |
} |
| 1001 |
return true; |
| 1002 |
} |
| 1003 |
|
| 1004 |
/** |
| 1005 |
* Get Pinecone API credentials from options |
| 1006 |
*/ |
| 1007 |
private function mxchat_get_pinecone_credentials() { |
| 1008 |
$options = get_option('mxchat_options', array()); |
| 1009 |
return array( |
| 1010 |
'api_key' => isset($options['pinecone_api_key']) ? $options['pinecone_api_key'] : '', |
| 1011 |
'host' => isset($options['pinecone_host']) ? $options['pinecone_host'] : '' |
| 1012 |
); |
| 1013 |
} |
| 1014 |
|
| 1015 |
/** |
| 1016 |
* Log Pinecone operation errors |
| 1017 |
*/ |
| 1018 |
private function log_pinecone_error($operation, $error_message) { |
| 1019 |
//error_log("MxChat Pinecone {$operation} Error: " . $error_message); |
| 1020 |
} |
| 1021 |
|
| 1022 |
// ======================================== |
| 1023 |
// STATIC ACCESS METHODS (for backward compatibility) |
| 1024 |
// ======================================== |
| 1025 |
|
| 1026 |
/** |
| 1027 |
* Get singleton instance |
| 1028 |
*/ |
| 1029 |
public static function get_instance() { |
| 1030 |
static $instance = null; |
| 1031 |
if ($instance === null) { |
| 1032 |
$instance = new self(); |
| 1033 |
} |
| 1034 |
return $instance; |
| 1035 |
} |
| 1036 |
} |
| 1037 |
|
| 1038 |
// Initialize the Pinecone manager |
| 1039 |
$mxchat_pinecone_manager = MxChat_Pinecone_Manager::get_instance(); |
| 1040 |
|
