| 1 |
<?php |
| 2 |
if (!defined('ABSPATH')) { |
| 3 |
exit; // Exit if accessed directly |
| 4 |
} |
| 5 |
|
| 6 |
class MxChat_Utils { |
| 7 |
|
| 8 |
/** |
| 9 |
* UPDATED: Submit or update content (and its embedding) in the database. |
| 10 |
* Stores in Pinecone if enabled, otherwise stores in WordPress DB. |
| 11 |
* |
| 12 |
* @param string $content The content to be embedded. |
| 13 |
* @param string $source_url The source URL of the content. |
| 14 |
* @param string $api_key The API key used for generating embeddings. |
| 15 |
* @param string $vector_id Optional vector ID for Pinecone (if not provided, will use md5 of URL) |
| 16 |
* @param string $bot_id The bot ID for multi-bot support |
| 17 |
* @param string $content_type The type of content (post, page, pdf, url, manual, product, etc.) |
| 18 |
* @return bool|WP_Error True on success, WP_Error on failure |
| 19 |
*/ |
| 20 |
public static function submit_content_to_db($content, $source_url, $api_key, $vector_id = null, $bot_id = 'default', $content_type = 'content') { |
| 21 |
global $wpdb; |
| 22 |
$table_name = $wpdb->prefix . 'mxchat_system_prompt_content'; |
| 23 |
|
| 24 |
//error_log('[MXCHAT-DB] Starting database submission for URL: ' . $source_url . ' (Bot: ' . $bot_id . ', Type: ' . $content_type . ')'); |
| 25 |
//error_log('[MXCHAT-DB] Content length: ' . strlen($content) . ' bytes'); |
| 26 |
|
| 27 |
// Sanitize the source URL |
| 28 |
$source_url = esc_url_raw($source_url); |
| 29 |
|
| 30 |
// Sanitize content_type |
| 31 |
$content_type = sanitize_key($content_type); |
| 32 |
if (empty($content_type)) { |
| 33 |
$content_type = 'content'; // Fallback for backwards compatibility |
| 34 |
} |
| 35 |
|
| 36 |
// Just ensure UTF-8 validity without aggressive escaping |
| 37 |
$safe_content = wp_check_invalid_utf8($content); |
| 38 |
// Remove only null bytes and other control characters, but preserve newlines (\n = \x0A) and carriage returns (\r = \x0D) |
| 39 |
$safe_content = preg_replace('/[\x00-\x08\x0B\x0C\x0E-\x1F\x7F]/u', '', $safe_content); |
| 40 |
|
| 41 |
// Check if chunking should be applied |
| 42 |
$chunker = MxChat_Chunker::from_settings(); |
| 43 |
if ($chunker->should_chunk($safe_content)) { |
| 44 |
//error_log('[MXCHAT-DB] Content exceeds chunk threshold, using chunked submission'); |
| 45 |
return self::submit_chunked_content($safe_content, $source_url, $api_key, $bot_id, $content_type, $chunker); |
| 46 |
} |
| 47 |
|
| 48 |
// UPDATED: Generate the embedding using bot-specific configuration |
| 49 |
$embedding_vector = self::generate_embedding($content, $api_key, $bot_id); |
| 50 |
|
| 51 |
if (!is_array($embedding_vector)) { |
| 52 |
//error_log('[MXCHAT-DB] Error: Embedding generation failed'); |
| 53 |
return new WP_Error('embedding_failed', 'Failed to generate embedding for content'); |
| 54 |
} |
| 55 |
|
| 56 |
//error_log('[MXCHAT-DB] Embedding generated successfully'); |
| 57 |
|
| 58 |
// UPDATED: Check if Pinecone is enabled for this specific bot |
| 59 |
if (self::is_pinecone_enabled_for_bot($bot_id)) { |
| 60 |
//error_log('[MXCHAT-DB] Pinecone is enabled for bot ' . $bot_id . ' - using Pinecone storage'); |
| 61 |
// Store in Pinecone only |
| 62 |
return self::store_in_pinecone_only($embedding_vector, $content, $source_url, $vector_id, $bot_id, $content_type); |
| 63 |
} else { |
| 64 |
//error_log('[MXCHAT-DB] Pinecone not enabled for bot ' . $bot_id . ' - using WordPress storage'); |
| 65 |
// Store in WordPress database only |
| 66 |
$embedding_vector_serialized = maybe_serialize($embedding_vector); |
| 67 |
return self::store_in_wordpress_db($safe_content, $source_url, $embedding_vector_serialized, $table_name, $content_type); |
| 68 |
} |
| 69 |
} |
| 70 |
|
| 71 |
/** |
| 72 |
* UPDATED: Check if Pinecone is enabled and properly configured for a specific bot |
| 73 |
*/ |
| 74 |
private static function is_pinecone_enabled_for_bot($bot_id = 'default') { |
| 75 |
// For default bot or when multi-bot is not active, use original method |
| 76 |
if ($bot_id === 'default' || !class_exists('MxChat_Multi_Bot_Manager')) { |
| 77 |
return self::is_pinecone_enabled(); |
| 78 |
} |
| 79 |
|
| 80 |
// Get bot-specific Pinecone configuration |
| 81 |
$bot_pinecone_config = apply_filters('mxchat_get_bot_pinecone_config', array(), $bot_id); |
| 82 |
|
| 83 |
if (empty($bot_pinecone_config)) { |
| 84 |
// Fallback to default configuration |
| 85 |
return self::is_pinecone_enabled(); |
| 86 |
} |
| 87 |
|
| 88 |
$enabled_check = !empty($bot_pinecone_config['use_pinecone']) && $bot_pinecone_config['use_pinecone']; |
| 89 |
$api_key_check = !empty($bot_pinecone_config['api_key']); |
| 90 |
$host_check = !empty($bot_pinecone_config['host']); |
| 91 |
|
| 92 |
return $enabled_check && $api_key_check && $host_check; |
| 93 |
} |
| 94 |
|
| 95 |
/** |
| 96 |
* Check if Pinecone is enabled and properly configured (original method for default bot) |
| 97 |
*/ |
| 98 |
private static function is_pinecone_enabled() { |
| 99 |
$pinecone_options = get_option('mxchat_pinecone_addon_options'); |
| 100 |
|
| 101 |
if (empty($pinecone_options)) { |
| 102 |
return false; |
| 103 |
} |
| 104 |
|
| 105 |
$enabled_check = !empty($pinecone_options['mxchat_use_pinecone']) && $pinecone_options['mxchat_use_pinecone'] !== '0'; |
| 106 |
$api_key_check = !empty($pinecone_options['mxchat_pinecone_api_key']); |
| 107 |
$host_check = !empty($pinecone_options['mxchat_pinecone_host']); |
| 108 |
|
| 109 |
return $enabled_check && $api_key_check && $host_check; |
| 110 |
} |
| 111 |
|
| 112 |
/** |
| 113 |
* UPDATED: Store content in Pinecone only with bot support |
| 114 |
*/ |
| 115 |
private static function store_in_pinecone_only($embedding_vector, $content, $source_url, $vector_id = null, $bot_id = 'default', $content_type = 'content') { |
| 116 |
//error_log('[MXCHAT-PINECONE] ===== Using Pinecone-only storage for bot ' . $bot_id . ' ====='); |
| 117 |
|
| 118 |
// Get bot-specific Pinecone configuration |
| 119 |
if ($bot_id === 'default' || !class_exists('MxChat_Multi_Bot_Manager')) { |
| 120 |
$pinecone_options = get_option('mxchat_pinecone_addon_options'); |
| 121 |
$api_key = $pinecone_options['mxchat_pinecone_api_key']; |
| 122 |
$environment = $pinecone_options['mxchat_pinecone_environment'] ?? ''; |
| 123 |
$index_name = $pinecone_options['mxchat_pinecone_index'] ?? ''; |
| 124 |
$namespace = $pinecone_options['mxchat_pinecone_namespace'] ?? ''; |
| 125 |
} else { |
| 126 |
$bot_pinecone_config = apply_filters('mxchat_get_bot_pinecone_config', array(), $bot_id); |
| 127 |
if (empty($bot_pinecone_config)) { |
| 128 |
// Fallback to default configuration |
| 129 |
$pinecone_options = get_option('mxchat_pinecone_addon_options'); |
| 130 |
$api_key = $pinecone_options['mxchat_pinecone_api_key']; |
| 131 |
$environment = $pinecone_options['mxchat_pinecone_environment'] ?? ''; |
| 132 |
$index_name = $pinecone_options['mxchat_pinecone_index'] ?? ''; |
| 133 |
$namespace = $pinecone_options['mxchat_pinecone_namespace'] ?? ''; |
| 134 |
} else { |
| 135 |
$api_key = $bot_pinecone_config['api_key']; |
| 136 |
$environment = ''; // Not used in new Pinecone API |
| 137 |
$index_name = ''; // Not used in new Pinecone API |
| 138 |
$namespace = $bot_pinecone_config['namespace'] ?? ''; |
| 139 |
} |
| 140 |
} |
| 141 |
|
| 142 |
$result = self::store_in_pinecone_main( |
| 143 |
$embedding_vector, |
| 144 |
$content, |
| 145 |
$source_url, |
| 146 |
$api_key, |
| 147 |
$environment, |
| 148 |
$index_name, |
| 149 |
$vector_id, |
| 150 |
$bot_id, |
| 151 |
$namespace, |
| 152 |
$content_type |
| 153 |
); |
| 154 |
|
| 155 |
if (is_wp_error($result)) { |
| 156 |
//error_log('[MXCHAT-PINECONE] Pinecone storage failed for bot ' . $bot_id . ': ' . $result->get_error_message()); |
| 157 |
return $result; |
| 158 |
} |
| 159 |
|
| 160 |
//error_log('[MXCHAT-PINECONE] Pinecone storage completed successfully for bot ' . $bot_id); |
| 161 |
return true; |
| 162 |
} |
| 163 |
|
| 164 |
/** |
| 165 |
* Store content in WordPress database with progressive fallback |
| 166 |
* UPDATED 2.5.6: Now includes content_type parameter |
| 167 |
*/ |
| 168 |
private static function store_in_wordpress_db($safe_content, $source_url, $embedding_vector_serialized, $table_name, $content_type = 'content') { |
| 169 |
global $wpdb; |
| 170 |
|
| 171 |
//error_log('[MXCHAT-DB] ===== Using WordPress-only storage ====='); |
| 172 |
|
| 173 |
// Sanitize content_type |
| 174 |
$content_type = sanitize_key($content_type); |
| 175 |
if (empty($content_type)) { |
| 176 |
$content_type = 'content'; // Fallback for backwards compatibility |
| 177 |
} |
| 178 |
|
| 179 |
// ===== FIXED: Generate unique identifier for manual content ===== |
| 180 |
$original_source_url = $source_url; |
| 181 |
$is_manual_content = empty($source_url) || $source_url === '' || !filter_var($source_url, FILTER_VALIDATE_URL); |
| 182 |
|
| 183 |
if ($is_manual_content) { |
| 184 |
// Generate unique identifier for manual content to prevent overwrites |
| 185 |
$source_url = 'mxchat://manual-content/' . time() . '-' . wp_generate_password(8, false); |
| 186 |
//error_log('[MXCHAT-DB] Generated unique ID for manual content: ' . $source_url); |
| 187 |
} |
| 188 |
|
| 189 |
// Only check for duplicates if we have a valid source URL (not manual content) |
| 190 |
$existing_id = null; |
| 191 |
if (!$is_manual_content) { |
| 192 |
$existing_id = $wpdb->get_var( |
| 193 |
$wpdb->prepare( |
| 194 |
"SELECT id FROM {$table_name} WHERE source_url = %s LIMIT 1", |
| 195 |
$source_url |
| 196 |
) |
| 197 |
); |
| 198 |
//error_log('[MXCHAT-DB] Checked for existing URL, found ID: ' . ($existing_id ?: 'none')); |
| 199 |
} else { |
| 200 |
//error_log('[MXCHAT-DB] Manual content - will create new entry (no duplicate check)'); |
| 201 |
} |
| 202 |
// ===== END FIX ===== |
| 203 |
|
| 204 |
// Progressive fallback mechanism for problematic content |
| 205 |
$attempt = 1; |
| 206 |
$max_attempts = 3; |
| 207 |
$current_content = $safe_content; |
| 208 |
$result = false; |
| 209 |
|
| 210 |
while ($attempt <= $max_attempts && $result === false) { |
| 211 |
try { |
| 212 |
if ($existing_id) { |
| 213 |
//error_log('[MXCHAT-DB] Found existing entry (ID: ' . $existing_id . '). Updating... (Attempt ' . $attempt . ')'); |
| 214 |
|
| 215 |
// Update the existing row - UPDATED 2.5.6: Added content_type |
| 216 |
$result = $wpdb->update( |
| 217 |
$table_name, |
| 218 |
array( |
| 219 |
'url' => $source_url, |
| 220 |
'article_content' => $current_content, |
| 221 |
'embedding_vector' => $embedding_vector_serialized, |
| 222 |
'source_url' => $source_url, |
| 223 |
'content_type' => $content_type, |
| 224 |
'timestamp' => current_time('mysql'), |
| 225 |
), |
| 226 |
array('id' => $existing_id), |
| 227 |
array('%s','%s','%s','%s','%s','%s'), |
| 228 |
array('%d') |
| 229 |
); |
| 230 |
} else { |
| 231 |
//error_log('[MXCHAT-DB] No existing entry found. Inserting new row... (Attempt ' . $attempt . ')'); |
| 232 |
//error_log('[MXCHAT-DB] Content sample: ' . substr($current_content, 0, 1000)); |
| 233 |
|
| 234 |
// Insert a new row - UPDATED 2.5.6: Added content_type |
| 235 |
$result = $wpdb->insert( |
| 236 |
$table_name, |
| 237 |
array( |
| 238 |
'url' => $source_url, // Now unique for manual content |
| 239 |
'article_content' => $current_content, |
| 240 |
'embedding_vector' => $embedding_vector_serialized, |
| 241 |
'source_url' => $source_url, // Now unique for manual content |
| 242 |
'content_type' => $content_type, |
| 243 |
'timestamp' => current_time('mysql'), |
| 244 |
), |
| 245 |
array('%s','%s','%s','%s','%s','%s') |
| 246 |
); |
| 247 |
} |
| 248 |
|
| 249 |
if ($result === false) { |
| 250 |
error_log('[MXCHAT-DB] Database operation failed (Attempt ' . $attempt . ')'); |
| 251 |
error_log('[MXCHAT-DB] MySQL Error: ' . $wpdb->last_error); |
| 252 |
error_log('[MXCHAT-DB] MySQL Error Number: ' . $wpdb->last_errno); |
| 253 |
error_log('[MXCHAT-DB] Last Query: ' . substr($wpdb->last_query, 0, 500)); |
| 254 |
error_log('[MXCHAT-DB] Content length: ' . strlen($current_content) . ' bytes'); |
| 255 |
error_log('[MXCHAT-DB] Embedding vector length: ' . strlen($embedding_vector_serialized) . ' bytes'); |
| 256 |
|
| 257 |
// Progressively apply more aggressive sanitization on failure |
| 258 |
if ($attempt === 1) { |
| 259 |
// First fallback: Use a more aggressive character filter and shorten |
| 260 |
$current_content = preg_replace('/[^\p{L}\p{N}\s.,;:!?()-]/u', '', $current_content); |
| 261 |
$current_content = substr($current_content, 0, 50000); |
| 262 |
} else if ($attempt === 2) { |
| 263 |
// Second fallback: Keep only alphanumeric and basic punctuation, shorten further |
| 264 |
$current_content = preg_replace('/[^a-zA-Z0-9\s.,;:!?()-]/u', '', $current_content); |
| 265 |
$current_content = substr($current_content, 0, 30000); |
| 266 |
} |
| 267 |
|
| 268 |
$attempt++; |
| 269 |
} |
| 270 |
} catch (Exception $e) { |
| 271 |
//error_log('[MXCHAT-DB] Exception during database operation: ' . $e->getMessage()); |
| 272 |
$attempt++; |
| 273 |
} |
| 274 |
} |
| 275 |
|
| 276 |
if ($result === false) { |
| 277 |
error_log('[MXCHAT-DB] All database operation attempts failed'); |
| 278 |
error_log('[MXCHAT-DB] Final MySQL Error: ' . $wpdb->last_error); |
| 279 |
|
| 280 |
$detailed_error = sprintf( |
| 281 |
'Failed to store content in WordPress database after %d attempts. MySQL Error: %s (Error #%d). Content size: %d bytes, Embedding size: %d bytes', |
| 282 |
$max_attempts, |
| 283 |
$wpdb->last_error, |
| 284 |
$wpdb->last_errno, |
| 285 |
strlen($current_content), |
| 286 |
strlen($embedding_vector_serialized) |
| 287 |
); |
| 288 |
|
| 289 |
return new WP_Error('database_failed', $detailed_error); |
| 290 |
} |
| 291 |
|
| 292 |
//error_log('[MXCHAT-DB] WordPress database operation completed successfully (Attempt ' . ($attempt - 1) . ')'); |
| 293 |
return true; |
| 294 |
} |
| 295 |
|
| 296 |
/** |
| 297 |
* UPDATED: Store content in Pinecone database with bot support |
| 298 |
* UPDATED 2.5.6: Now accepts content_type parameter |
| 299 |
*/ |
| 300 |
private static function store_in_pinecone_main($embedding_vector, $content, $url, $api_key, $environment, $index_name, $vector_id = null, $bot_id = 'default', $namespace = '', $content_type = 'content') { |
| 301 |
//error_log('[MXCHAT-PINECONE-MAIN] ===== Starting Pinecone storage for bot ' . $bot_id . ' ====='); |
| 302 |
|
| 303 |
// ===== UPDATED: Handle manual content with unique vector IDs ===== |
| 304 |
if ($vector_id) { |
| 305 |
// Use provided vector ID |
| 306 |
//error_log('[MXCHAT-PINECONE-MAIN] Using provided vector ID: ' . $vector_id); |
| 307 |
} elseif (!empty($url) && filter_var($url, FILTER_VALIDATE_URL)) { |
| 308 |
// For valid URLs, use URL-based ID (existing behavior) |
| 309 |
$vector_id = md5($url); |
| 310 |
//error_log('[MXCHAT-PINECONE-MAIN] Generated vector ID from URL: ' . $vector_id); |
| 311 |
} else { |
| 312 |
// For manual content (empty/invalid URL), generate unique ID |
| 313 |
$vector_id = 'manual_' . time() . '_' . substr(md5($content . microtime(true)), 0, 8); |
| 314 |
//error_log('[MXCHAT-PINECONE-MAIN] Generated unique vector ID for manual content: ' . $vector_id); |
| 315 |
} |
| 316 |
// ===== END UPDATE ===== |
| 317 |
|
| 318 |
// Get host from bot-specific config or fallback to default |
| 319 |
if ($bot_id === 'default' || !class_exists('MxChat_Multi_Bot_Manager')) { |
| 320 |
$options = get_option('mxchat_pinecone_addon_options'); |
| 321 |
$host = $options['mxchat_pinecone_host'] ?? ''; |
| 322 |
} else { |
| 323 |
$bot_pinecone_config = apply_filters('mxchat_get_bot_pinecone_config', array(), $bot_id); |
| 324 |
if (!empty($bot_pinecone_config)) { |
| 325 |
$host = $bot_pinecone_config['host'] ?? ''; |
| 326 |
} else { |
| 327 |
$options = get_option('mxchat_pinecone_addon_options'); |
| 328 |
$host = $options['mxchat_pinecone_host'] ?? ''; |
| 329 |
} |
| 330 |
} |
| 331 |
|
| 332 |
//error_log('[MXCHAT-PINECONE-MAIN] Host: ' . $host); |
| 333 |
//error_log('[MXCHAT-PINECONE-MAIN] API key length: ' . strlen($api_key)); |
| 334 |
//error_log('[MXCHAT-PINECONE-MAIN] Bot ID: ' . $bot_id); |
| 335 |
//error_log('[MXCHAT-PINECONE-MAIN] Namespace: ' . $namespace); |
| 336 |
|
| 337 |
if (empty($host)) { |
| 338 |
//error_log('[MXCHAT-PINECONE-MAIN] ERROR: Host is empty'); |
| 339 |
return new WP_Error('pinecone_config', 'Pinecone host is not configured. Please set the host in your bot settings.'); |
| 340 |
} |
| 341 |
|
| 342 |
// ===== UPDATED 2.5.6: Use passed content_type or determine from URL if not provided ===== |
| 343 |
// Sanitize content_type |
| 344 |
$content_type = sanitize_key($content_type); |
| 345 |
if (empty($content_type)) { |
| 346 |
// Fallback to old detection logic for backwards compatibility |
| 347 |
$is_product = false; |
| 348 |
$content_type = 'manual'; // Default for manual content |
| 349 |
|
| 350 |
if (!empty($url) && filter_var($url, FILTER_VALIDATE_URL)) { |
| 351 |
$is_product = (strpos($url, '/product/') !== false || strpos($url, '/shop/') !== false); |
| 352 |
$content_type = $is_product ? 'product' : 'content'; |
| 353 |
} |
| 354 |
} |
| 355 |
|
| 356 |
//error_log('[MXCHAT-PINECONE-MAIN] Content type: ' . $content_type); |
| 357 |
// ===== END UPDATE ===== |
| 358 |
|
| 359 |
$api_endpoint = "https://{$host}/vectors/upsert"; |
| 360 |
//error_log('[MXCHAT-PINECONE-MAIN] API endpoint: ' . $api_endpoint); |
| 361 |
|
| 362 |
// UPDATED 2.5.6: Use provided content_type in metadata |
| 363 |
$metadata = array( |
| 364 |
'text' => $content, |
| 365 |
'source_url' => $url, // Can be empty for manual content |
| 366 |
'type' => $content_type, // Now supports: post, page, pdf, url, manual, product, etc. |
| 367 |
'last_updated' => time(), |
| 368 |
'created_at' => time(), // Add creation timestamp |
| 369 |
'bot_id' => $bot_id // Add bot identification |
| 370 |
); |
| 371 |
|
| 372 |
$vector_data = array( |
| 373 |
'id' => $vector_id, |
| 374 |
'values' => $embedding_vector, |
| 375 |
'metadata' => $metadata |
| 376 |
); |
| 377 |
|
| 378 |
$request_body = array( |
| 379 |
'vectors' => array($vector_data) |
| 380 |
); |
| 381 |
|
| 382 |
// Add namespace if specified for multi-bot separation |
| 383 |
if (!empty($namespace)) { |
| 384 |
$request_body['namespace'] = $namespace; |
| 385 |
//error_log('[MXCHAT-PINECONE-MAIN] Using namespace: ' . $namespace); |
| 386 |
} |
| 387 |
|
| 388 |
//error_log('[MXCHAT-PINECONE-MAIN] Request body prepared (embedding dimensions: ' . count($embedding_vector) . ')'); |
| 389 |
|
| 390 |
$response = wp_remote_post($api_endpoint, array( |
| 391 |
'headers' => array( |
| 392 |
'Api-Key' => $api_key, |
| 393 |
'accept' => 'application/json', |
| 394 |
'content-type' => 'application/json' |
| 395 |
), |
| 396 |
'body' => wp_json_encode($request_body), |
| 397 |
'timeout' => 30, |
| 398 |
'data_format' => 'body' |
| 399 |
)); |
| 400 |
|
| 401 |
if (is_wp_error($response)) { |
| 402 |
//error_log('[MXCHAT-PINECONE-MAIN] WordPress request error: ' . $response->get_error_message()); |
| 403 |
return new WP_Error('pinecone_request', $response->get_error_message()); |
| 404 |
} |
| 405 |
|
| 406 |
$response_code = wp_remote_retrieve_response_code($response); |
| 407 |
//error_log('[MXCHAT-PINECONE-MAIN] Response code: ' . $response_code); |
| 408 |
|
| 409 |
if ($response_code !== 200) { |
| 410 |
$body = wp_remote_retrieve_body($response); |
| 411 |
//error_log('[MXCHAT-PINECONE-MAIN] API error - Response body: ' . $body); |
| 412 |
return new WP_Error('pinecone_api', sprintf( |
| 413 |
'Pinecone API error (HTTP %d): %s', |
| 414 |
$response_code, |
| 415 |
$body |
| 416 |
)); |
| 417 |
} |
| 418 |
|
| 419 |
$response_body = wp_remote_retrieve_body($response); |
| 420 |
//error_log('[MXCHAT-PINECONE-MAIN] Success response: ' . $response_body); |
| 421 |
//error_log('[MXCHAT-PINECONE-MAIN] Successfully stored in Pinecone for bot ' . $bot_id); |
| 422 |
//error_log('[MXCHAT-PINECONE-MAIN] ===== Pinecone storage complete ====='); |
| 423 |
|
| 424 |
return true; |
| 425 |
} |
| 426 |
|
| 427 |
/** |
| 428 |
* UPDATED: Generate an embedding for the given text using bot-specific configuration. |
| 429 |
* |
| 430 |
* @param string $text The text to be embedded. |
| 431 |
* @param string $api_key The API key used for generating embeddings. |
| 432 |
* @param string $bot_id The bot ID for multi-bot support |
| 433 |
* @return array|null The embedding vector or null on failure. |
| 434 |
*/ |
| 435 |
private static function generate_embedding($text, $api_key, $bot_id = 'default') { |
| 436 |
// Get bot-specific options |
| 437 |
if ($bot_id === 'default' || !class_exists('MxChat_Multi_Bot_Manager')) { |
| 438 |
$options = get_option('mxchat_options'); |
| 439 |
} else { |
| 440 |
$bot_options = apply_filters('mxchat_get_bot_options', array(), $bot_id); |
| 441 |
$options = !empty($bot_options) ? $bot_options : get_option('mxchat_options'); |
| 442 |
} |
| 443 |
|
| 444 |
$selected_model = $options['embedding_model'] ?? 'text-embedding-ada-002'; |
| 445 |
|
| 446 |
// Determine endpoint and API key based on model |
| 447 |
if (strpos($selected_model, 'voyage') === 0) { |
| 448 |
$endpoint = 'https://api.voyageai.com/v1/embeddings'; |
| 449 |
$api_key = $options['voyage_api_key'] ?? ''; |
| 450 |
} elseif (strpos($selected_model, 'gemini-embedding') === 0) { |
| 451 |
$endpoint = 'https://generativelanguage.googleapis.com/v1beta/models/' . $selected_model . ':embedContent'; |
| 452 |
$api_key = $options['gemini_api_key'] ?? ''; |
| 453 |
} else { |
| 454 |
$endpoint = 'https://api.openai.com/v1/embeddings'; |
| 455 |
// Use the bot-specific API key or fallback to passed API key |
| 456 |
$api_key = $options['api_key'] ?? $api_key; |
| 457 |
} |
| 458 |
|
| 459 |
// Prepare request body based on provider |
| 460 |
if (strpos($selected_model, 'gemini-embedding') === 0) { |
| 461 |
// Gemini API format |
| 462 |
$request_body = [ |
| 463 |
'model' => 'models/' . $selected_model, |
| 464 |
'content' => [ |
| 465 |
'parts' => [ |
| 466 |
['text' => $text] |
| 467 |
] |
| 468 |
], |
| 469 |
'outputDimensionality' => 1536 |
| 470 |
]; |
| 471 |
|
| 472 |
// Prepare headers for Gemini (API key as query parameter) |
| 473 |
$endpoint .= '?key=' . $api_key; |
| 474 |
$headers = [ |
| 475 |
'Content-Type' => 'application/json' |
| 476 |
]; |
| 477 |
} else { |
| 478 |
// OpenAI/Voyage API format |
| 479 |
$request_body = [ |
| 480 |
'input' => $text, |
| 481 |
'model' => $selected_model |
| 482 |
]; |
| 483 |
|
| 484 |
// Add output_dimension for voyage-3-large |
| 485 |
if ($selected_model === 'voyage-3-large') { |
| 486 |
$request_body['output_dimension'] = 2048; |
| 487 |
} |
| 488 |
|
| 489 |
// Prepare headers for OpenAI/Voyage |
| 490 |
$headers = [ |
| 491 |
'Content-Type' => 'application/json', |
| 492 |
'Authorization' => 'Bearer ' . $api_key |
| 493 |
]; |
| 494 |
} |
| 495 |
|
| 496 |
$args = [ |
| 497 |
'body' => wp_json_encode($request_body), |
| 498 |
'headers' => $headers, |
| 499 |
'timeout' => 60, |
| 500 |
'redirection' => 5, |
| 501 |
'blocking' => true, |
| 502 |
'httpversion' => '1.0', |
| 503 |
'sslverify' => true, |
| 504 |
]; |
| 505 |
|
| 506 |
$response = wp_remote_post($endpoint, $args); |
| 507 |
|
| 508 |
if (is_wp_error($response)) { |
| 509 |
//error_log('Error generating embedding for bot ' . $bot_id . ': ' . $response->get_error_message()); |
| 510 |
return null; |
| 511 |
} |
| 512 |
|
| 513 |
$response_body = json_decode(wp_remote_retrieve_body($response), true); |
| 514 |
|
| 515 |
// Handle different response formats based on provider |
| 516 |
if (strpos($selected_model, 'gemini-embedding') === 0) { |
| 517 |
// Gemini API response format |
| 518 |
if (isset($response_body['embedding']['values']) && is_array($response_body['embedding']['values'])) { |
| 519 |
return $response_body['embedding']['values']; |
| 520 |
} else { |
| 521 |
//error_log('Invalid response received from Gemini embedding API for bot ' . $bot_id . ': ' . wp_json_encode($response_body)); |
| 522 |
return null; |
| 523 |
} |
| 524 |
} else { |
| 525 |
// OpenAI/Voyage API response format |
| 526 |
if (isset($response_body['data'][0]['embedding']) && is_array($response_body['data'][0]['embedding'])) { |
| 527 |
return $response_body['data'][0]['embedding']; |
| 528 |
} else { |
| 529 |
//error_log('Invalid response received from embedding API for bot ' . $bot_id . ': ' . wp_json_encode($response_body)); |
| 530 |
return null; |
| 531 |
} |
| 532 |
} |
| 533 |
} |
| 534 |
|
| 535 |
/** |
| 536 |
* Submit content as multiple chunks |
| 537 |
* |
| 538 |
* Splits large content into chunks, generates embeddings for each, |
| 539 |
* and stores them with chunk metadata for later reassembly. |
| 540 |
* |
| 541 |
* @param string $content The content to chunk and store |
| 542 |
* @param string $source_url The source URL |
| 543 |
* @param string $api_key The API key for embeddings |
| 544 |
* @param string $bot_id The bot ID |
| 545 |
* @param string $content_type The content type |
| 546 |
* @param MxChat_Chunker $chunker The chunker instance |
| 547 |
* @return bool|WP_Error True on success, WP_Error on failure |
| 548 |
*/ |
| 549 |
private static function submit_chunked_content($content, $source_url, $api_key, $bot_id, $content_type, $chunker) { |
| 550 |
global $wpdb; |
| 551 |
$table_name = $wpdb->prefix . 'mxchat_system_prompt_content'; |
| 552 |
|
| 553 |
error_log('[MXCHAT-CHUNK-DEBUG] Starting chunked submission for: ' . $source_url); |
| 554 |
error_log('[MXCHAT-CHUNK-DEBUG] Content length: ' . strlen($content) . ' chars'); |
| 555 |
|
| 556 |
// First, delete any existing chunks for this URL (clean slate) |
| 557 |
$delete_result = self::delete_chunks_for_url($source_url, $bot_id); |
| 558 |
if (is_wp_error($delete_result)) { |
| 559 |
error_log('[MXCHAT-CHUNK-DEBUG] Warning: Failed to delete existing chunks: ' . $delete_result->get_error_message()); |
| 560 |
// Continue anyway - we'll overwrite with upsert |
| 561 |
} |
| 562 |
|
| 563 |
// Split content into chunks |
| 564 |
$chunks = $chunker->chunk_text($content); |
| 565 |
$total_chunks = count($chunks); |
| 566 |
|
| 567 |
error_log('[MXCHAT-CHUNK-DEBUG] Created ' . $total_chunks . ' chunks'); |
| 568 |
foreach ($chunks as $i => $chunk) { |
| 569 |
error_log('[MXCHAT-CHUNK-DEBUG] Chunk ' . $i . ' length: ' . strlen($chunk) . ' chars, preview: ' . substr($chunk, 0, 100)); |
| 570 |
} |
| 571 |
|
| 572 |
//error_log('[MXCHAT-CHUNK] Split content into ' . $total_chunks . ' chunks'); |
| 573 |
|
| 574 |
if ($total_chunks === 0) { |
| 575 |
return new WP_Error('chunking_failed', 'Content could not be split into chunks'); |
| 576 |
} |
| 577 |
|
| 578 |
$errors = array(); |
| 579 |
$is_pinecone = self::is_pinecone_enabled_for_bot($bot_id); |
| 580 |
|
| 581 |
foreach ($chunks as $index => $chunk_text) { |
| 582 |
// Generate chunk metadata |
| 583 |
$chunk_metadata = MxChat_Chunker::create_chunk_metadata($index, $total_chunks, $source_url); |
| 584 |
$chunk_vector_id = MxChat_Chunker::generate_chunk_vector_id($source_url, $index); |
| 585 |
|
| 586 |
//error_log('[MXCHAT-CHUNK] Processing chunk ' . ($index + 1) . '/' . $total_chunks . ' (ID: ' . $chunk_vector_id . ')'); |
| 587 |
|
| 588 |
// Generate embedding for this chunk |
| 589 |
$embedding_vector = self::generate_embedding($chunk_text, $api_key, $bot_id); |
| 590 |
|
| 591 |
if (!is_array($embedding_vector)) { |
| 592 |
$errors[] = new WP_Error('embedding_failed', 'Failed to generate embedding for chunk ' . $index); |
| 593 |
//error_log('[MXCHAT-CHUNK] Failed to generate embedding for chunk ' . $index); |
| 594 |
continue; |
| 595 |
} |
| 596 |
|
| 597 |
if ($is_pinecone) { |
| 598 |
// Store in Pinecone with chunk metadata |
| 599 |
$result = self::store_chunk_in_pinecone( |
| 600 |
$embedding_vector, |
| 601 |
$chunk_text, |
| 602 |
$source_url, |
| 603 |
$chunk_vector_id, |
| 604 |
$bot_id, |
| 605 |
$content_type, |
| 606 |
$chunk_metadata |
| 607 |
); |
| 608 |
} else { |
| 609 |
// Store in WordPress DB with chunk metadata |
| 610 |
$content_with_metadata = MxChat_Chunker::format_chunk_for_storage($chunk_text, $chunk_metadata); |
| 611 |
$embedding_vector_serialized = maybe_serialize($embedding_vector); |
| 612 |
|
| 613 |
$result = self::store_chunk_in_wordpress_db( |
| 614 |
$content_with_metadata, |
| 615 |
$source_url, |
| 616 |
$embedding_vector_serialized, |
| 617 |
$table_name, |
| 618 |
$content_type, |
| 619 |
$chunk_metadata |
| 620 |
); |
| 621 |
} |
| 622 |
|
| 623 |
if (is_wp_error($result)) { |
| 624 |
$errors[] = $result; |
| 625 |
//error_log('[MXCHAT-CHUNK] Failed to store chunk ' . $index . ': ' . $result->get_error_message()); |
| 626 |
} |
| 627 |
} |
| 628 |
|
| 629 |
if (count($errors) === $total_chunks) { |
| 630 |
return new WP_Error('chunking_failed', 'Failed to store any chunks'); |
| 631 |
} |
| 632 |
|
| 633 |
if (!empty($errors)) { |
| 634 |
//error_log('[MXCHAT-CHUNK] Completed with ' . count($errors) . ' errors out of ' . $total_chunks . ' chunks'); |
| 635 |
return new WP_Error('chunking_partial_failure', |
| 636 |
sprintf('Failed to store %d of %d chunks', count($errors), $total_chunks)); |
| 637 |
} |
| 638 |
|
| 639 |
//error_log('[MXCHAT-CHUNK] Successfully stored all ' . $total_chunks . ' chunks'); |
| 640 |
return true; |
| 641 |
} |
| 642 |
|
| 643 |
/** |
| 644 |
* Store a single chunk in Pinecone with chunk-specific metadata |
| 645 |
*/ |
| 646 |
private static function store_chunk_in_pinecone($embedding_vector, $chunk_text, $source_url, $vector_id, $bot_id, $content_type, $chunk_metadata) { |
| 647 |
// Get Pinecone configuration |
| 648 |
if ($bot_id === 'default' || !class_exists('MxChat_Multi_Bot_Manager')) { |
| 649 |
$pinecone_options = get_option('mxchat_pinecone_addon_options'); |
| 650 |
$api_key = $pinecone_options['mxchat_pinecone_api_key'] ?? ''; |
| 651 |
$host = $pinecone_options['mxchat_pinecone_host'] ?? ''; |
| 652 |
$namespace = $pinecone_options['mxchat_pinecone_namespace'] ?? ''; |
| 653 |
} else { |
| 654 |
$bot_pinecone_config = apply_filters('mxchat_get_bot_pinecone_config', array(), $bot_id); |
| 655 |
if (empty($bot_pinecone_config)) { |
| 656 |
$pinecone_options = get_option('mxchat_pinecone_addon_options'); |
| 657 |
$api_key = $pinecone_options['mxchat_pinecone_api_key'] ?? ''; |
| 658 |
$host = $pinecone_options['mxchat_pinecone_host'] ?? ''; |
| 659 |
$namespace = $pinecone_options['mxchat_pinecone_namespace'] ?? ''; |
| 660 |
} else { |
| 661 |
$api_key = $bot_pinecone_config['api_key'] ?? ''; |
| 662 |
$host = $bot_pinecone_config['host'] ?? ''; |
| 663 |
$namespace = $bot_pinecone_config['namespace'] ?? ''; |
| 664 |
} |
| 665 |
} |
| 666 |
|
| 667 |
if (empty($host) || empty($api_key)) { |
| 668 |
return new WP_Error('pinecone_config', 'Pinecone is not properly configured'); |
| 669 |
} |
| 670 |
|
| 671 |
$api_endpoint = "https://{$host}/vectors/upsert"; |
| 672 |
|
| 673 |
// Build metadata with chunk information |
| 674 |
$metadata = array( |
| 675 |
'text' => $chunk_text, |
| 676 |
'source_url' => $source_url, |
| 677 |
'type' => $content_type, |
| 678 |
'is_chunked' => true, |
| 679 |
'chunk_index' => $chunk_metadata['chunk_index'], |
| 680 |
'total_chunks' => $chunk_metadata['total_chunks'], |
| 681 |
'parent_url_hash' => $chunk_metadata['parent_url_hash'], |
| 682 |
'last_updated' => time(), |
| 683 |
'created_at' => time(), |
| 684 |
'bot_id' => $bot_id |
| 685 |
); |
| 686 |
|
| 687 |
$vector_data = array( |
| 688 |
'id' => $vector_id, |
| 689 |
'values' => $embedding_vector, |
| 690 |
'metadata' => $metadata |
| 691 |
); |
| 692 |
|
| 693 |
$request_body = array( |
| 694 |
'vectors' => array($vector_data) |
| 695 |
); |
| 696 |
|
| 697 |
if (!empty($namespace)) { |
| 698 |
$request_body['namespace'] = $namespace; |
| 699 |
} |
| 700 |
|
| 701 |
$response = wp_remote_post($api_endpoint, array( |
| 702 |
'headers' => array( |
| 703 |
'Api-Key' => $api_key, |
| 704 |
'accept' => 'application/json', |
| 705 |
'content-type' => 'application/json' |
| 706 |
), |
| 707 |
'body' => wp_json_encode($request_body), |
| 708 |
'timeout' => 30 |
| 709 |
)); |
| 710 |
|
| 711 |
if (is_wp_error($response)) { |
| 712 |
return $response; |
| 713 |
} |
| 714 |
|
| 715 |
$response_code = wp_remote_retrieve_response_code($response); |
| 716 |
if ($response_code !== 200) { |
| 717 |
return new WP_Error('pinecone_api', 'Pinecone API error: HTTP ' . $response_code); |
| 718 |
} |
| 719 |
|
| 720 |
return true; |
| 721 |
} |
| 722 |
|
| 723 |
/** |
| 724 |
* Store a single chunk in WordPress database |
| 725 |
*/ |
| 726 |
private static function store_chunk_in_wordpress_db($content_with_metadata, $source_url, $embedding_vector_serialized, $table_name, $content_type, $chunk_metadata) { |
| 727 |
global $wpdb; |
| 728 |
|
| 729 |
// For chunks, we always insert new rows (no duplicate checking) |
| 730 |
// The URL includes chunk info in the metadata, but source_url stays the same for grouping |
| 731 |
$result = $wpdb->insert( |
| 732 |
$table_name, |
| 733 |
array( |
| 734 |
'url' => $source_url, |
| 735 |
'article_content' => $content_with_metadata, |
| 736 |
'embedding_vector' => $embedding_vector_serialized, |
| 737 |
'source_url' => $source_url, |
| 738 |
'content_type' => $content_type, |
| 739 |
'timestamp' => current_time('mysql') |
| 740 |
), |
| 741 |
array('%s', '%s', '%s', '%s', '%s', '%s') |
| 742 |
); |
| 743 |
|
| 744 |
if ($result === false) { |
| 745 |
return new WP_Error('database_failed', 'Failed to insert chunk: ' . $wpdb->last_error); |
| 746 |
} |
| 747 |
|
| 748 |
return true; |
| 749 |
} |
| 750 |
|
| 751 |
/** |
| 752 |
* Delete all chunks for a given URL |
| 753 |
* |
| 754 |
* @param string $source_url The source URL |
| 755 |
* @param string $bot_id The bot ID |
| 756 |
* @return bool|WP_Error True on success, WP_Error on failure |
| 757 |
*/ |
| 758 |
public static function delete_chunks_for_url($source_url, $bot_id = 'default') { |
| 759 |
//error_log('[MXCHAT-CHUNK-DELETE] Deleting chunks for URL: ' . $source_url); |
| 760 |
|
| 761 |
if (self::is_pinecone_enabled_for_bot($bot_id)) { |
| 762 |
return self::delete_pinecone_chunks_by_url($source_url, $bot_id); |
| 763 |
} else { |
| 764 |
return self::delete_wordpress_chunks_by_url($source_url); |
| 765 |
} |
| 766 |
} |
| 767 |
|
| 768 |
/** |
| 769 |
* Delete all chunks for a URL from Pinecone |
| 770 |
*/ |
| 771 |
private static function delete_pinecone_chunks_by_url($source_url, $bot_id) { |
| 772 |
// Get Pinecone configuration |
| 773 |
if ($bot_id === 'default' || !class_exists('MxChat_Multi_Bot_Manager')) { |
| 774 |
$pinecone_options = get_option('mxchat_pinecone_addon_options'); |
| 775 |
$api_key = $pinecone_options['mxchat_pinecone_api_key'] ?? ''; |
| 776 |
$host = $pinecone_options['mxchat_pinecone_host'] ?? ''; |
| 777 |
$namespace = $pinecone_options['mxchat_pinecone_namespace'] ?? ''; |
| 778 |
} else { |
| 779 |
$bot_pinecone_config = apply_filters('mxchat_get_bot_pinecone_config', array(), $bot_id); |
| 780 |
if (empty($bot_pinecone_config)) { |
| 781 |
$pinecone_options = get_option('mxchat_pinecone_addon_options'); |
| 782 |
$api_key = $pinecone_options['mxchat_pinecone_api_key'] ?? ''; |
| 783 |
$host = $pinecone_options['mxchat_pinecone_host'] ?? ''; |
| 784 |
$namespace = $pinecone_options['mxchat_pinecone_namespace'] ?? ''; |
| 785 |
} else { |
| 786 |
$api_key = $bot_pinecone_config['api_key'] ?? ''; |
| 787 |
$host = $bot_pinecone_config['host'] ?? ''; |
| 788 |
$namespace = $bot_pinecone_config['namespace'] ?? ''; |
| 789 |
} |
| 790 |
} |
| 791 |
|
| 792 |
if (empty($host) || empty($api_key)) { |
| 793 |
return new WP_Error('pinecone_config', 'Pinecone is not properly configured'); |
| 794 |
} |
| 795 |
|
| 796 |
$base_vector_id = md5($source_url); |
| 797 |
$vectors_to_delete = array(); |
| 798 |
|
| 799 |
// Add the original single-vector ID (for non-chunked content) |
| 800 |
$vectors_to_delete[] = $base_vector_id; |
| 801 |
|
| 802 |
// Use Pinecone list API to find all chunk vectors with this prefix |
| 803 |
$list_url = "https://{$host}/vectors/list"; |
| 804 |
|
| 805 |
$list_body = array( |
| 806 |
'prefix' => $base_vector_id . '_chunk_', |
| 807 |
'limit' => 100 |
| 808 |
); |
| 809 |
|
| 810 |
if (!empty($namespace)) { |
| 811 |
$list_body['namespace'] = $namespace; |
| 812 |
} |
| 813 |
|
| 814 |
$list_response = wp_remote_post($list_url, array( |
| 815 |
'headers' => array( |
| 816 |
'Api-Key' => $api_key, |
| 817 |
'accept' => 'application/json', |
| 818 |
'content-type' => 'application/json' |
| 819 |
), |
| 820 |
'body' => wp_json_encode($list_body), |
| 821 |
'timeout' => 30 |
| 822 |
)); |
| 823 |
|
| 824 |
if (!is_wp_error($list_response)) { |
| 825 |
$list_data = json_decode(wp_remote_retrieve_body($list_response), true); |
| 826 |
if (!empty($list_data['vectors'])) { |
| 827 |
foreach ($list_data['vectors'] as $vector) { |
| 828 |
if (isset($vector['id'])) { |
| 829 |
$vectors_to_delete[] = $vector['id']; |
| 830 |
} |
| 831 |
} |
| 832 |
} |
| 833 |
} |
| 834 |
|
| 835 |
if (empty($vectors_to_delete)) { |
| 836 |
//error_log('[MXCHAT-CHUNK-DELETE] No vectors found to delete'); |
| 837 |
return true; |
| 838 |
} |
| 839 |
|
| 840 |
//error_log('[MXCHAT-CHUNK-DELETE] Deleting ' . count($vectors_to_delete) . ' vectors from Pinecone'); |
| 841 |
|
| 842 |
// Delete vectors |
| 843 |
$delete_url = "https://{$host}/vectors/delete"; |
| 844 |
|
| 845 |
$delete_body = array( |
| 846 |
'ids' => $vectors_to_delete |
| 847 |
); |
| 848 |
|
| 849 |
if (!empty($namespace)) { |
| 850 |
$delete_body['namespace'] = $namespace; |
| 851 |
} |
| 852 |
|
| 853 |
$delete_response = wp_remote_post($delete_url, array( |
| 854 |
'headers' => array( |
| 855 |
'Api-Key' => $api_key, |
| 856 |
'accept' => 'application/json', |
| 857 |
'content-type' => 'application/json' |
| 858 |
), |
| 859 |
'body' => wp_json_encode($delete_body), |
| 860 |
'timeout' => 30 |
| 861 |
)); |
| 862 |
|
| 863 |
if (is_wp_error($delete_response)) { |
| 864 |
return $delete_response; |
| 865 |
} |
| 866 |
|
| 867 |
$response_code = wp_remote_retrieve_response_code($delete_response); |
| 868 |
if ($response_code !== 200) { |
| 869 |
return new WP_Error('pinecone_delete', 'Failed to delete vectors: HTTP ' . $response_code); |
| 870 |
} |
| 871 |
|
| 872 |
return true; |
| 873 |
} |
| 874 |
|
| 875 |
/** |
| 876 |
* Delete all chunks for a URL from WordPress database |
| 877 |
*/ |
| 878 |
private static function delete_wordpress_chunks_by_url($source_url) { |
| 879 |
global $wpdb; |
| 880 |
$table_name = $wpdb->prefix . 'mxchat_system_prompt_content'; |
| 881 |
|
| 882 |
// Delete all rows with this source_url (handles both chunked and non-chunked) |
| 883 |
$result = $wpdb->delete( |
| 884 |
$table_name, |
| 885 |
array('source_url' => $source_url), |
| 886 |
array('%s') |
| 887 |
); |
| 888 |
|
| 889 |
if ($result === false) { |
| 890 |
return new WP_Error('database_delete', 'Failed to delete chunks: ' . $wpdb->last_error); |
| 891 |
} |
| 892 |
|
| 893 |
//error_log('[MXCHAT-CHUNK-DELETE] Deleted ' . $result . ' rows from WordPress DB'); |
| 894 |
return true; |
| 895 |
} |
| 896 |
} |
