| @@ -1,965 +1,488 @@ | ||
| 1 | -<?php | |
| 2 | -if (!defined('ABSPATH')) { | |
| 3 | - exit; // Exit if accessed directly | |
| 4 | -} | |
| 5 | - | |
| 6 | -class MxChat_Utils { | |
| 7 | - | |
| 8 | -/** | |
| 9 | - * Centralized embedding model registry. Single source of truth for dimensions | |
| 10 | - * and provider, so model-switch protection logic doesn't drift across files. | |
| 11 | - */ | |
| 12 | -public static function embedding_model_registry() { | |
| 13 | - return array( | |
| 14 | - 'text-embedding-ada-002' => array('dims' => 1536, 'provider' => 'openai', 'label' => 'Ada 2'), | |
| 15 | - 'text-embedding-3-small' => array('dims' => 1536, 'provider' => 'openai', 'label' => 'TE3 Small'), | |
| 16 | - 'text-embedding-3-large' => array('dims' => 3072, 'provider' => 'openai', 'label' => 'TE3 Large'), | |
| 17 | - 'voyage-3-large' => array('dims' => 2048, 'provider' => 'voyage', 'label' => 'Voyage-3 Large'), | |
| 18 | - 'gemini-embedding-001' => array('dims' => 1536, 'provider' => 'gemini', 'label' => 'Gemini Embedding'), | |
| 19 | - ); | |
| 20 | -} | |
| 21 | - | |
| 22 | -public static function embedding_model_dimensions($model) { | |
| 23 | - $registry = self::embedding_model_registry(); | |
| 24 | - return isset($registry[$model]) ? (int) $registry[$model]['dims'] : 0; | |
| 25 | -} | |
| 26 | - | |
| 27 | -public static function embedding_model_label($model) { | |
| 28 | - $registry = self::embedding_model_registry(); | |
| 29 | - return isset($registry[$model]) ? $registry[$model]['label'] : $model; | |
| 30 | -} | |
| 31 | - | |
| 32 | -/** | |
| 33 | - * Returns the model that was last used to actually write embeddings into the | |
| 34 | - * KB. Differs from the user-selected setting once a switch has happened but | |
| 35 | - * no re-embed has occurred yet — that's the mismatch state we warn about. | |
| 36 | - */ | |
| 37 | -public static function get_active_embedding_model() { | |
| 38 | - return get_option('mxchat_active_embedding_model', ''); | |
| 39 | -} | |
| 40 | - | |
| 41 | -/** | |
| 42 | - * Stamp the model that produced the most recent successful embedding. Called | |
| 43 | - * from generate_embedding() right after the API responds with a valid vector. | |
| 44 | - */ | |
| 45 | -public static function stamp_active_embedding_model($model) { | |
| 46 | - if (!empty($model) && $model !== self::get_active_embedding_model()) { | |
| 47 | - update_option('mxchat_active_embedding_model', $model, false); | |
| 48 | - } | |
| 49 | -} | |
| 50 | - | |
| 51 | -/** | |
| 52 | - * UPDATED: Submit or update content (and its embedding) in the database. | |
| 53 | - * Stores in Pinecone if enabled, otherwise stores in WordPress DB. | |
| 54 | - * | |
| 55 | - * @param string $content The content to be embedded. | |
| 56 | - * @param string $source_url The source URL of the content. | |
| 57 | - * @param string $api_key The API key used for generating embeddings. | |
| 58 | - * @param string $vector_id Optional vector ID for Pinecone (if not provided, will use md5 of URL) | |
| 59 | - * @param string $bot_id The bot ID for multi-bot support | |
| 60 | - * @param string $content_type The type of content (post, page, pdf, url, manual, product, etc.) | |
| 61 | - * @return bool|WP_Error True on success, WP_Error on failure | |
| 62 | - */ | |
| 63 | -public static function submit_content_to_db($content, $source_url, $api_key, $vector_id = null, $bot_id = 'default', $content_type = 'content') { | |
| 64 | - global $wpdb; | |
| 65 | - $table_name = $wpdb->prefix . 'mxchat_system_prompt_content'; | |
| 66 | - | |
| 67 | - //error_log('[MXCHAT-DB] Starting database submission for URL: ' . $source_url . ' (Bot: ' . $bot_id . ', Type: ' . $content_type . ')'); | |
| 68 | - //error_log('[MXCHAT-DB] Content length: ' . strlen($content) . ' bytes'); | |
| 69 | - | |
| 70 | - // Sanitize the source URL | |
| 71 | - $source_url = esc_url_raw($source_url); | |
| 72 | - | |
| 73 | - // Sanitize content_type | |
| 74 | - $content_type = sanitize_key($content_type); | |
| 75 | - if (empty($content_type)) { | |
| 76 | - $content_type = 'content'; // Fallback for backwards compatibility | |
| 77 | - } | |
| 78 | - | |
| 79 | - // Just ensure UTF-8 validity without aggressive escaping | |
| 80 | - $safe_content = wp_check_invalid_utf8($content); | |
| 81 | - // Remove only null bytes and other control characters, but preserve newlines (\n = \x0A) and carriage returns (\r = \x0D) | |
| 82 | - $safe_content = preg_replace('/[\x00-\x08\x0B\x0C\x0E-\x1F\x7F]/u', '', $safe_content); | |
| 83 | - | |
| 84 | - // Check if chunking should be applied | |
| 85 | - $chunker = MxChat_Chunker::from_settings(); | |
| 86 | - if ($chunker->should_chunk($safe_content)) { | |
| 87 | - //error_log('[MXCHAT-DB] Content exceeds chunk threshold, using chunked submission'); | |
| 88 | - return self::submit_chunked_content($safe_content, $source_url, $api_key, $bot_id, $content_type, $chunker); | |
| 89 | - } | |
| 90 | - | |
| 91 | - // UPDATED: Generate the embedding using bot-specific configuration | |
| 92 | - $embedding_vector = self::generate_embedding($content, $api_key, $bot_id); | |
| 93 | - | |
| 94 | - if (!is_array($embedding_vector)) { | |
| 95 | - //error_log('[MXCHAT-DB] Error: Embedding generation failed'); | |
| 96 | - return new WP_Error('embedding_failed', 'Failed to generate embedding for content'); | |
| 97 | - } | |
| 98 | - | |
| 99 | - //error_log('[MXCHAT-DB] Embedding generated successfully'); | |
| 100 | - | |
| 101 | - // UPDATED: Check if Pinecone is enabled for this specific bot | |
| 102 | - if (self::is_pinecone_enabled_for_bot($bot_id)) { | |
| 103 | - //error_log('[MXCHAT-DB] Pinecone is enabled for bot ' . $bot_id . ' - using Pinecone storage'); | |
| 104 | - // Store in Pinecone only | |
| 105 | - return self::store_in_pinecone_only($embedding_vector, $content, $source_url, $vector_id, $bot_id, $content_type); | |
| 106 | - } else { | |
| 107 | - //error_log('[MXCHAT-DB] Pinecone not enabled for bot ' . $bot_id . ' - using WordPress storage'); | |
| 108 | - // Store in WordPress database only | |
| 109 | - $embedding_vector_serialized = maybe_serialize($embedding_vector); | |
| 110 | - return self::store_in_wordpress_db($safe_content, $source_url, $embedding_vector_serialized, $table_name, $content_type); | |
| 111 | - } | |
| 112 | -} | |
| 113 | - | |
| 114 | -/** | |
| 115 | - * UPDATED: Check if Pinecone is enabled and properly configured for a specific bot | |
| 116 | - */ | |
| 117 | -private static function is_pinecone_enabled_for_bot($bot_id = 'default') { | |
| 118 | - // For default bot or when multi-bot is not active, use original method | |
| 119 | - if ($bot_id === 'default' || !class_exists('MxChat_Multi_Bot_Manager')) { | |
| 120 | - return self::is_pinecone_enabled(); | |
| 121 | - } | |
| 122 | - | |
| 123 | - // Get bot-specific Pinecone configuration | |
| 124 | - $bot_pinecone_config = apply_filters('mxchat_get_bot_pinecone_config', array(), $bot_id); | |
| 125 | - | |
| 126 | - if (empty($bot_pinecone_config)) { | |
| 127 | - // Fallback to default configuration | |
| 128 | - return self::is_pinecone_enabled(); | |
| 129 | - } | |
| 130 | - | |
| 131 | - $enabled_check = !empty($bot_pinecone_config['use_pinecone']) && $bot_pinecone_config['use_pinecone']; | |
| 132 | - $api_key_check = !empty($bot_pinecone_config['api_key']); | |
| 133 | - $host_check = !empty($bot_pinecone_config['host']); | |
| 134 | - | |
| 135 | - return $enabled_check && $api_key_check && $host_check; | |
| 136 | -} | |
| 137 | - | |
| 138 | -/** | |
| 139 | - * Check if Pinecone is enabled and properly configured (original method for default bot) | |
| 140 | - */ | |
| 141 | -private static function is_pinecone_enabled() { | |
| 142 | - $pinecone_options = get_option('mxchat_pinecone_addon_options'); | |
| 143 | - | |
| 144 | - if (empty($pinecone_options)) { | |
| 145 | - return false; | |
| 146 | - } | |
| 147 | - | |
| 148 | - $enabled_check = !empty($pinecone_options['mxchat_use_pinecone']) && $pinecone_options['mxchat_use_pinecone'] !== '0'; | |
| 149 | - $api_key_check = !empty($pinecone_options['mxchat_pinecone_api_key']); | |
| 150 | - $host_check = !empty($pinecone_options['mxchat_pinecone_host']); | |
| 151 | - | |
| 152 | - return $enabled_check && $api_key_check && $host_check; | |
| 153 | -} | |
| 154 | - | |
| 155 | -/** | |
| 156 | - * UPDATED: Store content in Pinecone only with bot support | |
| 157 | - */ | |
| 158 | -private static function store_in_pinecone_only($embedding_vector, $content, $source_url, $vector_id = null, $bot_id = 'default', $content_type = 'content') { | |
| 159 | - //error_log('[MXCHAT-PINECONE] ===== Using Pinecone-only storage for bot ' . $bot_id . ' ====='); | |
| 160 | - | |
| 161 | - // Get bot-specific Pinecone configuration | |
| 162 | - if ($bot_id === 'default' || !class_exists('MxChat_Multi_Bot_Manager')) { | |
| 163 | - $pinecone_options = get_option('mxchat_pinecone_addon_options'); | |
| 164 | - $api_key = $pinecone_options['mxchat_pinecone_api_key']; | |
| 165 | - $environment = $pinecone_options['mxchat_pinecone_environment'] ?? ''; | |
| 166 | - $index_name = $pinecone_options['mxchat_pinecone_index'] ?? ''; | |
| 167 | - $namespace = $pinecone_options['mxchat_pinecone_namespace'] ?? ''; | |
| 168 | - } else { | |
| 169 | - $bot_pinecone_config = apply_filters('mxchat_get_bot_pinecone_config', array(), $bot_id); | |
| 170 | - if (empty($bot_pinecone_config)) { | |
| 171 | - // Fallback to default configuration | |
| 172 | - $pinecone_options = get_option('mxchat_pinecone_addon_options'); | |
| 173 | - $api_key = $pinecone_options['mxchat_pinecone_api_key']; | |
| 174 | - $environment = $pinecone_options['mxchat_pinecone_environment'] ?? ''; | |
| 175 | - $index_name = $pinecone_options['mxchat_pinecone_index'] ?? ''; | |
| 176 | - $namespace = $pinecone_options['mxchat_pinecone_namespace'] ?? ''; | |
| 177 | - } else { | |
| 178 | - $api_key = $bot_pinecone_config['api_key']; | |
| 179 | - $environment = ''; // Not used in new Pinecone API | |
| 180 | - $index_name = ''; // Not used in new Pinecone API | |
| 181 | - $namespace = $bot_pinecone_config['namespace'] ?? ''; | |
| 182 | - } | |
| 183 | - } | |
| 184 | - | |
| 185 | - $result = self::store_in_pinecone_main( | |
| 186 | - $embedding_vector, | |
| 187 | - $content, | |
| 188 | - $source_url, | |
| 189 | - $api_key, | |
| 190 | - $environment, | |
| 191 | - $index_name, | |
| 192 | - $vector_id, | |
| 193 | - $bot_id, | |
| 194 | - $namespace, | |
| 195 | - $content_type | |
| 196 | - ); | |
| 197 | - | |
| 198 | - if (is_wp_error($result)) { | |
| 199 | - //error_log('[MXCHAT-PINECONE] Pinecone storage failed for bot ' . $bot_id . ': ' . $result->get_error_message()); | |
| 200 | - return $result; | |
| 201 | - } | |
| 202 | - | |
| 203 | - //error_log('[MXCHAT-PINECONE] Pinecone storage completed successfully for bot ' . $bot_id); | |
| 204 | - return true; | |
| 205 | -} | |
| 206 | - | |
| 207 | -/** | |
| 208 | - * Store content in WordPress database with progressive fallback | |
| 209 | - * UPDATED 2.5.6: Now includes content_type parameter | |
| 210 | - */ | |
| 211 | -private static function store_in_wordpress_db($safe_content, $source_url, $embedding_vector_serialized, $table_name, $content_type = 'content') { | |
| 212 | - global $wpdb; | |
| 213 | - | |
| 214 | - //error_log('[MXCHAT-DB] ===== Using WordPress-only storage ====='); | |
| 215 | - | |
| 216 | - // Sanitize content_type | |
| 217 | - $content_type = sanitize_key($content_type); | |
| 218 | - if (empty($content_type)) { | |
| 219 | - $content_type = 'content'; // Fallback for backwards compatibility | |
| 220 | - } | |
| 221 | - | |
| 222 | - // ===== FIXED: Generate unique identifier for manual content ===== | |
| 223 | - $original_source_url = $source_url; | |
| 224 | - // Check if this is truly manual content (no URL at all) vs a real URL that filter_var rejects | |
| 225 | - // filter_var(FILTER_VALIDATE_URL) rejects valid URLs with encoded chars, non-ASCII, fragments, etc. | |
| 226 | - // Use a looser check: if it starts with http(s):// or has a scheme, it's a URL | |
| 227 | - $has_url_scheme = !empty($source_url) && preg_match('#^https?://#i', $source_url); | |
| 228 | - // Treat legacy mxchat.ai source URLs as manual — old bug assigned the site URL to manual entries | |
| 229 | - $is_legacy_mxchat_url = $has_url_scheme && strpos($source_url, 'mxchat.ai') !== false; | |
| 230 | - $is_manual_content = empty($source_url) || $source_url === '' || !$has_url_scheme || $is_legacy_mxchat_url; | |
| 231 | - | |
| 232 | - if ($is_manual_content) { | |
| 233 | - // Generate unique identifier for manual content to prevent overwrites | |
| 234 | - $source_url = 'mxchat://manual-content/' . time() . '-' . wp_generate_password(8, false); | |
| 235 | - //error_log('[MXCHAT-DB] Generated unique ID for manual content: ' . $source_url); | |
| 236 | - } | |
| 237 | - | |
| 238 | - // Only check for duplicates if we have a valid source URL (not manual content) | |
| 239 | - $existing_id = null; | |
| 240 | - if (!$is_manual_content) { | |
| 241 | - $existing_id = $wpdb->get_var( | |
| 242 | - $wpdb->prepare( | |
| 243 | - "SELECT id FROM {$table_name} WHERE source_url = %s LIMIT 1", | |
| 244 | - $source_url | |
| 245 | - ) | |
| 246 | - ); | |
| 247 | - //error_log('[MXCHAT-DB] Checked for existing URL, found ID: ' . ($existing_id ?: 'none')); | |
| 248 | - } else { | |
| 249 | - //error_log('[MXCHAT-DB] Manual content - will create new entry (no duplicate check)'); | |
| 250 | - } | |
| 251 | - // ===== END FIX ===== | |
| 252 | - | |
| 253 | - // Progressive fallback mechanism for problematic content | |
| 254 | - $attempt = 1; | |
| 255 | - $max_attempts = 3; | |
| 256 | - $current_content = $safe_content; | |
| 257 | - $result = false; | |
| 258 | - | |
| 259 | - $active_model = self::get_active_embedding_model(); | |
| 260 | - | |
| 261 | - while ($attempt <= $max_attempts && $result === false) { | |
| 262 | - try { | |
| 263 | - if ($existing_id) { | |
| 264 | - //error_log('[MXCHAT-DB] Found existing entry (ID: ' . $existing_id . '). Updating... (Attempt ' . $attempt . ')'); | |
| 265 | - | |
| 266 | - // Update the existing row - UPDATED 2.5.6: Added content_type | |
| 267 | - $result = $wpdb->update( | |
| 268 | - $table_name, | |
| 269 | - array( | |
| 270 | - 'url' => $source_url, | |
| 271 | - 'article_content' => $current_content, | |
| 272 | - 'embedding_vector' => $embedding_vector_serialized, | |
| 273 | - 'source_url' => $source_url, | |
| 274 | - 'content_type' => $content_type, | |
| 275 | - 'embedding_model' => $active_model, | |
| 276 | - 'timestamp' => current_time('mysql'), | |
| 277 | - ), | |
| 278 | - array('id' => $existing_id), | |
| 279 | - array('%s','%s','%s','%s','%s','%s','%s'), | |
| 280 | - array('%d') | |
| 281 | - ); | |
| 282 | - } else { | |
| 283 | - //error_log('[MXCHAT-DB] No existing entry found. Inserting new row... (Attempt ' . $attempt . ')'); | |
| 284 | - //error_log('[MXCHAT-DB] Content sample: ' . substr($current_content, 0, 1000)); | |
| 285 | - | |
| 286 | - // Insert a new row - UPDATED 2.5.6: Added content_type | |
| 287 | - $result = $wpdb->insert( | |
| 288 | - $table_name, | |
| 289 | - array( | |
| 290 | - 'url' => $source_url, // Now unique for manual content | |
| 291 | - 'article_content' => $current_content, | |
| 292 | - 'embedding_vector' => $embedding_vector_serialized, | |
| 293 | - 'source_url' => $source_url, // Now unique for manual content | |
| 294 | - 'content_type' => $content_type, | |
| 295 | - 'embedding_model' => $active_model, | |
| 296 | - 'timestamp' => current_time('mysql'), | |
| 297 | - ), | |
| 298 | - array('%s','%s','%s','%s','%s','%s','%s') | |
| 299 | - ); | |
| 300 | - } | |
| 301 | - | |
| 302 | - if ($result === false) { | |
| 303 | - //error_log('[MXCHAT-DB] Database operation failed (Attempt ' . $attempt . ')'); | |
| 304 | - //error_log('[MXCHAT-DB] MySQL Error: ' . $wpdb->last_error); | |
| 305 | - //error_log('[MXCHAT-DB] MySQL Error Number: ' . $wpdb->last_errno); | |
| 306 | - //error_log('[MXCHAT-DB] Last Query: ' . substr($wpdb->last_query, 0, 500)); | |
| 307 | - //error_log('[MXCHAT-DB] Content length: ' . strlen($current_content) . ' bytes'); | |
| 308 | - //error_log('[MXCHAT-DB] Embedding vector length: ' . strlen($embedding_vector_serialized) . ' bytes'); | |
| 309 | - | |
| 310 | - // Progressively apply more aggressive sanitization on failure | |
| 311 | - if ($attempt === 1) { | |
| 312 | - // First fallback: Use a more aggressive character filter and shorten | |
| 313 | - $current_content = preg_replace('/[^\p{L}\p{N}\s.,;:!?()-]/u', '', $current_content); | |
| 314 | - $current_content = substr($current_content, 0, 50000); | |
| 315 | - } else if ($attempt === 2) { | |
| 316 | - // Second fallback: Keep only alphanumeric and basic punctuation, shorten further | |
| 317 | - $current_content = preg_replace('/[^a-zA-Z0-9\s.,;:!?()-]/u', '', $current_content); | |
| 318 | - $current_content = substr($current_content, 0, 30000); | |
| 319 | - } | |
| 320 | - | |
| 321 | - $attempt++; | |
| 322 | - } | |
| 323 | - } catch (Exception $e) { | |
| 324 | - //error_log('[MXCHAT-DB] Exception during database operation: ' . $e->getMessage()); | |
| 325 | - $attempt++; | |
| 326 | - } | |
| 327 | - } | |
| 328 | - | |
| 329 | -if ($result === false) { | |
| 330 | - //error_log('[MXCHAT-DB] All database operation attempts failed'); | |
| 331 | - //error_log('[MXCHAT-DB] Final MySQL Error: ' . $wpdb->last_error); | |
| 332 | - | |
| 333 | - $detailed_error = sprintf( | |
| 334 | - 'Failed to store content in WordPress database after %d attempts. MySQL Error: %s (Error #%d). Content size: %d bytes, Embedding size: %d bytes', | |
| 335 | - $max_attempts, | |
| 336 | - $wpdb->last_error, | |
| 337 | - $wpdb->last_errno, | |
| 338 | - strlen($current_content), | |
| 339 | - strlen($embedding_vector_serialized) | |
| 340 | - ); | |
| 341 | - | |
| 342 | - return new WP_Error('database_failed', $detailed_error); | |
| 343 | -} | |
| 344 | - | |
| 345 | - //error_log('[MXCHAT-DB] WordPress database operation completed successfully (Attempt ' . ($attempt - 1) . ')'); | |
| 346 | - return true; | |
| 347 | -} | |
| 348 | - | |
| 349 | -/** | |
| 350 | - * UPDATED: Store content in Pinecone database with bot support | |
| 351 | - * UPDATED 2.5.6: Now accepts content_type parameter | |
| 352 | - */ | |
| 353 | -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') { | |
| 354 | - //error_log('[MXCHAT-PINECONE-MAIN] ===== Starting Pinecone storage for bot ' . $bot_id . ' ====='); | |
| 355 | - | |
| 356 | - // ===== UPDATED: Handle manual content with unique vector IDs ===== | |
| 357 | - if ($vector_id) { | |
| 358 | - // Use provided vector ID | |
| 359 | - //error_log('[MXCHAT-PINECONE-MAIN] Using provided vector ID: ' . $vector_id); | |
| 360 | - } elseif (!empty($url) && preg_match('#^https?://#i', $url)) { | |
| 361 | - // For URLs, use URL-based ID (existing behavior) | |
| 362 | - $vector_id = md5($url); | |
| 363 | - //error_log('[MXCHAT-PINECONE-MAIN] Generated vector ID from URL: ' . $vector_id); | |
| 364 | - } else { | |
| 365 | - // For manual content (empty/no URL scheme), generate unique ID | |
| 366 | - $vector_id = 'manual_' . time() . '_' . substr(md5($content . microtime(true)), 0, 8); | |
| 367 | - //error_log('[MXCHAT-PINECONE-MAIN] Generated unique vector ID for manual content: ' . $vector_id); | |
| 368 | - } | |
| 369 | - // ===== END UPDATE ===== | |
| 370 | - | |
| 371 | - // Get host from bot-specific config or fallback to default | |
| 372 | - if ($bot_id === 'default' || !class_exists('MxChat_Multi_Bot_Manager')) { | |
| 373 | - $options = get_option('mxchat_pinecone_addon_options'); | |
| 374 | - $host = $options['mxchat_pinecone_host'] ?? ''; | |
| 375 | - } else { | |
| 376 | - $bot_pinecone_config = apply_filters('mxchat_get_bot_pinecone_config', array(), $bot_id); | |
| 377 | - if (!empty($bot_pinecone_config)) { | |
| 378 | - $host = $bot_pinecone_config['host'] ?? ''; | |
| 379 | - } else { | |
| 380 | - $options = get_option('mxchat_pinecone_addon_options'); | |
| 381 | - $host = $options['mxchat_pinecone_host'] ?? ''; | |
| 382 | - } | |
| 383 | - } | |
| 384 | - | |
| 385 | - //error_log('[MXCHAT-PINECONE-MAIN] Host: ' . $host); | |
| 386 | - //error_log('[MXCHAT-PINECONE-MAIN] API key length: ' . strlen($api_key)); | |
| 387 | - //error_log('[MXCHAT-PINECONE-MAIN] Bot ID: ' . $bot_id); | |
| 388 | - //error_log('[MXCHAT-PINECONE-MAIN] Namespace: ' . $namespace); | |
| 389 | - | |
| 390 | - if (empty($host)) { | |
| 391 | - //error_log('[MXCHAT-PINECONE-MAIN] ERROR: Host is empty'); | |
| 392 | - return new WP_Error('pinecone_config', 'Pinecone host is not configured. Please set the host in your bot settings.'); | |
| 393 | - } | |
| 394 | - | |
| 395 | - // ===== UPDATED 2.5.6: Use passed content_type or determine from URL if not provided ===== | |
| 396 | - // Sanitize content_type | |
| 397 | - $content_type = sanitize_key($content_type); | |
| 398 | - if (empty($content_type)) { | |
| 399 | - // Fallback to old detection logic for backwards compatibility | |
| 400 | - $is_product = false; | |
| 401 | - $content_type = 'manual'; // Default for manual content | |
| 402 | - | |
| 403 | - if (!empty($url) && preg_match('#^https?://#i', $url)) { | |
| 404 | - $is_product = (strpos($url, '/product/') !== false || strpos($url, '/shop/') !== false); | |
| 405 | - $content_type = $is_product ? 'product' : 'content'; | |
| 406 | - } | |
| 407 | - } | |
| 408 | - | |
| 409 | - //error_log('[MXCHAT-PINECONE-MAIN] Content type: ' . $content_type); | |
| 410 | - // ===== END UPDATE ===== | |
| 411 | - | |
| 412 | - $api_endpoint = "https://{$host}/vectors/upsert"; | |
| 413 | - //error_log('[MXCHAT-PINECONE-MAIN] API endpoint: ' . $api_endpoint); | |
| 414 | - | |
| 415 | - // UPDATED 2.5.6: Use provided content_type in metadata | |
| 416 | - $metadata = array( | |
| 417 | - 'text' => $content, | |
| 418 | - 'source_url' => $url, // Can be empty for manual content | |
| 419 | - 'type' => $content_type, // Now supports: post, page, pdf, url, manual, product, etc. | |
| 420 | - 'last_updated' => time(), | |
| 421 | - 'created_at' => time(), // Add creation timestamp | |
| 422 | - 'bot_id' => $bot_id, // Add bot identification | |
| 423 | - 'embedding_model' => self::get_active_embedding_model() // 3.2.3: track which model produced this vector | |
| 424 | - ); | |
| 425 | - | |
| 426 | - $vector_data = array( | |
| 427 | - 'id' => $vector_id, | |
| 428 | - 'values' => $embedding_vector, | |
| 429 | - 'metadata' => $metadata | |
| 430 | - ); | |
| 431 | - | |
| 432 | - $request_body = array( | |
| 433 | - 'vectors' => array($vector_data) | |
| 434 | - ); | |
| 435 | - | |
| 436 | - // Add namespace if specified for multi-bot separation | |
| 437 | - if (!empty($namespace)) { | |
| 438 | - $request_body['namespace'] = $namespace; | |
| 439 | - //error_log('[MXCHAT-PINECONE-MAIN] Using namespace: ' . $namespace); | |
| 440 | - } | |
| 441 | - | |
| 442 | - //error_log('[MXCHAT-PINECONE-MAIN] Request body prepared (embedding dimensions: ' . count($embedding_vector) . ')'); | |
| 443 | - | |
| 444 | - $response = wp_remote_post($api_endpoint, array( | |
| 445 | - 'headers' => array( | |
| 446 | - 'Api-Key' => $api_key, | |
| 447 | - 'accept' => 'application/json', | |
| 448 | - 'content-type' => 'application/json' | |
| 449 | - ), | |
| 450 | - 'body' => wp_json_encode($request_body), | |
| 451 | - 'timeout' => 30, | |
| 452 | - 'data_format' => 'body' | |
| 453 | - )); | |
| 454 | - | |
| 455 | - if (is_wp_error($response)) { | |
| 456 | - //error_log('[MXCHAT-PINECONE-MAIN] WordPress request error: ' . $response->get_error_message()); | |
| 457 | - return new WP_Error('pinecone_request', $response->get_error_message()); | |
| 458 | - } | |
| 459 | - | |
| 460 | - $response_code = wp_remote_retrieve_response_code($response); | |
| 461 | - //error_log('[MXCHAT-PINECONE-MAIN] Response code: ' . $response_code); | |
| 462 | - | |
| 463 | - if ($response_code !== 200) { | |
| 464 | - $body = wp_remote_retrieve_body($response); | |
| 465 | - //error_log('[MXCHAT-PINECONE-MAIN] API error - Response body: ' . $body); | |
| 466 | - return new WP_Error('pinecone_api', sprintf( | |
| 467 | - 'Pinecone API error (HTTP %d): %s', | |
| 468 | - $response_code, | |
| 469 | - $body | |
| 470 | - )); | |
| 471 | - } | |
| 472 | - | |
| 473 | - $response_body = wp_remote_retrieve_body($response); | |
| 474 | - //error_log('[MXCHAT-PINECONE-MAIN] Success response: ' . $response_body); | |
| 475 | - //error_log('[MXCHAT-PINECONE-MAIN] Successfully stored in Pinecone for bot ' . $bot_id); | |
| 476 | - //error_log('[MXCHAT-PINECONE-MAIN] ===== Pinecone storage complete ====='); | |
| 477 | - | |
| 478 | - return true; | |
| 479 | -} | |
| 480 | - | |
| 481 | -/** | |
| 482 | - * UPDATED: Generate an embedding for the given text using bot-specific configuration. | |
| 483 | - * | |
| 484 | - * @param string $text The text to be embedded. | |
| 485 | - * @param string $api_key The API key used for generating embeddings. | |
| 486 | - * @param string $bot_id The bot ID for multi-bot support | |
| 487 | - * @return array|null The embedding vector or null on failure. | |
| 488 | - */ | |
| 489 | -private static function generate_embedding($text, $api_key, $bot_id = 'default') { | |
| 490 | - // Get bot-specific options | |
| 491 | - if ($bot_id === 'default' || !class_exists('MxChat_Multi_Bot_Manager')) { | |
| 492 | - $options = get_option('mxchat_options'); | |
| 493 | - } else { | |
| 494 | - $bot_options = apply_filters('mxchat_get_bot_options', array(), $bot_id); | |
| 495 | - $options = !empty($bot_options) ? $bot_options : get_option('mxchat_options'); | |
| 496 | - } | |
| 497 | - | |
| 498 | - $selected_model = $options['embedding_model'] ?? 'text-embedding-ada-002'; | |
| 499 | - | |
| 500 | - // Determine endpoint and API key based on model | |
| 501 | - if (strpos($selected_model, 'voyage') === 0) { | |
| 502 | - $endpoint = 'https://api.voyageai.com/v1/embeddings'; | |
| 503 | - $api_key = $options['voyage_api_key'] ?? ''; | |
| 504 | - } elseif (strpos($selected_model, 'gemini-embedding') === 0) { | |
| 505 | - $endpoint = 'https://generativelanguage.googleapis.com/v1beta/models/' . $selected_model . ':embedContent'; | |
| 506 | - $api_key = $options['gemini_api_key'] ?? ''; | |
| 507 | - } else { | |
| 508 | - $endpoint = 'https://api.openai.com/v1/embeddings'; | |
| 509 | - // Use the bot-specific API key or fallback to passed API key | |
| 510 | - $api_key = $options['api_key'] ?? $api_key; | |
| 511 | - } | |
| 512 | - | |
| 513 | - // Prepare request body based on provider | |
| 514 | - if (strpos($selected_model, 'gemini-embedding') === 0) { | |
| 515 | - // Gemini API format | |
| 516 | - $request_body = [ | |
| 517 | - 'model' => 'models/' . $selected_model, | |
| 518 | - 'content' => [ | |
| 519 | - 'parts' => [ | |
| 520 | - ['text' => $text] | |
| 521 | - ] | |
| 522 | - ], | |
| 523 | - 'outputDimensionality' => 1536 | |
| 524 | - ]; | |
| 525 | - | |
| 526 | - // Prepare headers for Gemini (API key as query parameter) | |
| 527 | - $endpoint .= '?key=' . $api_key; | |
| 528 | - $headers = [ | |
| 529 | - 'Content-Type' => 'application/json' | |
| 530 | - ]; | |
| 531 | - } else { | |
| 532 | - // OpenAI/Voyage API format | |
| 533 | - $request_body = [ | |
| 534 | - 'input' => $text, | |
| 535 | - 'model' => $selected_model | |
| 536 | - ]; | |
| 537 | - | |
| 538 | - // Add output_dimension for voyage-3-large | |
| 539 | - if ($selected_model === 'voyage-3-large') { | |
| 540 | - $request_body['output_dimension'] = 2048; | |
| 541 | - } | |
| 542 | - | |
| 543 | - // Prepare headers for OpenAI/Voyage | |
| 544 | - $headers = [ | |
| 545 | - 'Content-Type' => 'application/json', | |
| 546 | - 'Authorization' => 'Bearer ' . $api_key | |
| 547 | - ]; | |
| 548 | - } | |
| 549 | - | |
| 550 | - $args = [ | |
| 551 | - 'body' => wp_json_encode($request_body), | |
| 552 | - 'headers' => $headers, | |
| 553 | - 'timeout' => 60, | |
| 554 | - 'redirection' => 5, | |
| 555 | - 'blocking' => true, | |
| 556 | - 'httpversion' => '1.0', | |
| 557 | - 'sslverify' => true, | |
| 558 | - ]; | |
| 559 | - | |
| 560 | - $response = wp_remote_post($endpoint, $args); | |
| 561 | - | |
| 562 | - if (is_wp_error($response)) { | |
| 563 | - //error_log('Error generating embedding for bot ' . $bot_id . ': ' . $response->get_error_message()); | |
| 564 | - return null; | |
| 565 | - } | |
| 566 | - | |
| 567 | - $response_body = json_decode(wp_remote_retrieve_body($response), true); | |
| 568 | - | |
| 569 | - // Handle different response formats based on provider | |
| 570 | - if (strpos($selected_model, 'gemini-embedding') === 0) { | |
| 571 | - // Gemini API response format | |
| 572 | - if (isset($response_body['embedding']['values']) && is_array($response_body['embedding']['values'])) { | |
| 573 | - self::stamp_active_embedding_model($selected_model); | |
| 574 | - return $response_body['embedding']['values']; | |
| 575 | - } else { | |
| 576 | - //error_log('Invalid response received from Gemini embedding API for bot ' . $bot_id . ': ' . wp_json_encode($response_body)); | |
| 577 | - return null; | |
| 578 | - } | |
| 579 | - } else { | |
| 580 | - // OpenAI/Voyage API response format | |
| 581 | - if (isset($response_body['data'][0]['embedding']) && is_array($response_body['data'][0]['embedding'])) { | |
| 582 | - self::stamp_active_embedding_model($selected_model); | |
| 583 | - return $response_body['data'][0]['embedding']; | |
| 584 | - } else { | |
| 585 | - //error_log('Invalid response received from embedding API for bot ' . $bot_id . ': ' . wp_json_encode($response_body)); | |
| 586 | - return null; | |
| 587 | - } | |
| 588 | - } | |
| 589 | -} | |
| 590 | - | |
| 591 | -/** | |
| 592 | - * Submit content as multiple chunks | |
| 593 | - * | |
| 594 | - * Splits large content into chunks, generates embeddings for each, | |
| 595 | - * and stores them with chunk metadata for later reassembly. | |
| 596 | - * | |
| 597 | - * @param string $content The content to chunk and store | |
| 598 | - * @param string $source_url The source URL | |
| 599 | - * @param string $api_key The API key for embeddings | |
| 600 | - * @param string $bot_id The bot ID | |
| 601 | - * @param string $content_type The content type | |
| 602 | - * @param MxChat_Chunker $chunker The chunker instance | |
| 603 | - * @return bool|WP_Error True on success, WP_Error on failure | |
| 604 | - */ | |
| 605 | -private static function submit_chunked_content($content, $source_url, $api_key, $bot_id, $content_type, $chunker) { | |
| 606 | - global $wpdb; | |
| 607 | - $table_name = $wpdb->prefix . 'mxchat_system_prompt_content'; | |
| 608 | - | |
| 609 | - //error_log('[MXCHAT-CHUNK-DEBUG] Starting chunked submission for: ' . $source_url); | |
| 610 | - //error_log('[MXCHAT-CHUNK-DEBUG] Content length: ' . strlen($content) . ' chars'); | |
| 611 | - | |
| 612 | - // First, delete any existing chunks for this URL (clean slate) | |
| 613 | - $delete_result = self::delete_chunks_for_url($source_url, $bot_id); | |
| 614 | - if (is_wp_error($delete_result)) { | |
| 615 | - //error_log('[MXCHAT-CHUNK-DEBUG] Warning: Failed to delete existing chunks: ' . $delete_result->get_error_message()); | |
| 616 | - // Continue anyway - we'll overwrite with upsert | |
| 617 | - } | |
| 618 | - | |
| 619 | - // Split content into chunks | |
| 620 | - $chunks = $chunker->chunk_text($content); | |
| 621 | - $total_chunks = count($chunks); | |
| 622 | - | |
| 623 | - //error_log('[MXCHAT-CHUNK-DEBUG] Created ' . $total_chunks . ' chunks'); | |
| 624 | - foreach ($chunks as $i => $chunk) { | |
| 625 | - //error_log('[MXCHAT-CHUNK-DEBUG] Chunk ' . $i . ' length: ' . strlen($chunk) . ' chars, preview: ' . substr($chunk, 0, 100)); | |
| 626 | - } | |
| 627 | - | |
| 628 | - //error_log('[MXCHAT-CHUNK] Split content into ' . $total_chunks . ' chunks'); | |
| 629 | - | |
| 630 | - if ($total_chunks === 0) { | |
| 631 | - return new WP_Error('chunking_failed', 'Content could not be split into chunks'); | |
| 632 | - } | |
| 633 | - | |
| 634 | - $errors = array(); | |
| 635 | - $is_pinecone = self::is_pinecone_enabled_for_bot($bot_id); | |
| 636 | - | |
| 637 | - foreach ($chunks as $index => $chunk_text) { | |
| 638 | - // Generate chunk metadata | |
| 639 | - $chunk_metadata = MxChat_Chunker::create_chunk_metadata($index, $total_chunks, $source_url); | |
| 640 | - $chunk_vector_id = MxChat_Chunker::generate_chunk_vector_id($source_url, $index); | |
| 641 | - | |
| 642 | - //error_log('[MXCHAT-CHUNK] Processing chunk ' . ($index + 1) . '/' . $total_chunks . ' (ID: ' . $chunk_vector_id . ')'); | |
| 643 | - | |
| 644 | - // Generate embedding for this chunk | |
| 645 | - $embedding_vector = self::generate_embedding($chunk_text, $api_key, $bot_id); | |
| 646 | - | |
| 647 | - if (!is_array($embedding_vector)) { | |
| 648 | - $errors[] = new WP_Error('embedding_failed', 'Failed to generate embedding for chunk ' . $index); | |
| 649 | - //error_log('[MXCHAT-CHUNK] Failed to generate embedding for chunk ' . $index); | |
| 650 | - continue; | |
| 651 | - } | |
| 652 | - | |
| 653 | - if ($is_pinecone) { | |
| 654 | - // Store in Pinecone with chunk metadata | |
| 655 | - $result = self::store_chunk_in_pinecone( | |
| 656 | - $embedding_vector, | |
| 657 | - $chunk_text, | |
| 658 | - $source_url, | |
| 659 | - $chunk_vector_id, | |
| 660 | - $bot_id, | |
| 661 | - $content_type, | |
| 662 | - $chunk_metadata | |
| 663 | - ); | |
| 664 | - } else { | |
| 665 | - // Store in WordPress DB with chunk metadata | |
| 666 | - $content_with_metadata = MxChat_Chunker::format_chunk_for_storage($chunk_text, $chunk_metadata); | |
| 667 | - $embedding_vector_serialized = maybe_serialize($embedding_vector); | |
| 668 | - | |
| 669 | - $result = self::store_chunk_in_wordpress_db( | |
| 670 | - $content_with_metadata, | |
| 671 | - $source_url, | |
| 672 | - $embedding_vector_serialized, | |
| 673 | - $table_name, | |
| 674 | - $content_type, | |
| 675 | - $chunk_metadata | |
| 676 | - ); | |
| 677 | - } | |
| 678 | - | |
| 679 | - if (is_wp_error($result)) { | |
| 680 | - $errors[] = $result; | |
| 681 | - //error_log('[MXCHAT-CHUNK] Failed to store chunk ' . $index . ': ' . $result->get_error_message()); | |
| 682 | - } | |
| 683 | - } | |
| 684 | - | |
| 685 | - if (count($errors) === $total_chunks) { | |
| 686 | - return new WP_Error('chunking_failed', 'Failed to store any chunks'); | |
| 687 | - } | |
| 688 | - | |
| 689 | - if (!empty($errors)) { | |
| 690 | - //error_log('[MXCHAT-CHUNK] Completed with ' . count($errors) . ' errors out of ' . $total_chunks . ' chunks'); | |
| 691 | - return new WP_Error('chunking_partial_failure', | |
| 692 | - sprintf('Failed to store %d of %d chunks', count($errors), $total_chunks)); | |
| 693 | - } | |
| 694 | - | |
| 695 | - //error_log('[MXCHAT-CHUNK] Successfully stored all ' . $total_chunks . ' chunks'); | |
| 696 | - return true; | |
| 697 | -} | |
| 698 | - | |
| 699 | -/** | |
| 700 | - * Store a single chunk in Pinecone with chunk-specific metadata | |
| 701 | - */ | |
| 702 | -private static function store_chunk_in_pinecone($embedding_vector, $chunk_text, $source_url, $vector_id, $bot_id, $content_type, $chunk_metadata) { | |
| 703 | - // Get Pinecone configuration | |
| 704 | - if ($bot_id === 'default' || !class_exists('MxChat_Multi_Bot_Manager')) { | |
| 705 | - $pinecone_options = get_option('mxchat_pinecone_addon_options'); | |
| 706 | - $api_key = $pinecone_options['mxchat_pinecone_api_key'] ?? ''; | |
| 707 | - $host = $pinecone_options['mxchat_pinecone_host'] ?? ''; | |
| 708 | - $namespace = $pinecone_options['mxchat_pinecone_namespace'] ?? ''; | |
| 709 | - } else { | |
| 710 | - $bot_pinecone_config = apply_filters('mxchat_get_bot_pinecone_config', array(), $bot_id); | |
| 711 | - if (empty($bot_pinecone_config)) { | |
| 712 | - $pinecone_options = get_option('mxchat_pinecone_addon_options'); | |
| 713 | - $api_key = $pinecone_options['mxchat_pinecone_api_key'] ?? ''; | |
| 714 | - $host = $pinecone_options['mxchat_pinecone_host'] ?? ''; | |
| 715 | - $namespace = $pinecone_options['mxchat_pinecone_namespace'] ?? ''; | |
| 716 | - } else { | |
| 717 | - $api_key = $bot_pinecone_config['api_key'] ?? ''; | |
| 718 | - $host = $bot_pinecone_config['host'] ?? ''; | |
| 719 | - $namespace = $bot_pinecone_config['namespace'] ?? ''; | |
| 720 | - } | |
| 721 | - } | |
| 722 | - | |
| 723 | - if (empty($host) || empty($api_key)) { | |
| 724 | - return new WP_Error('pinecone_config', 'Pinecone is not properly configured'); | |
| 725 | - } | |
| 726 | - | |
| 727 | - $api_endpoint = "https://{$host}/vectors/upsert"; | |
| 728 | - | |
| 729 | - // Build metadata with chunk information | |
| 730 | - $metadata = array( | |
| 731 | - 'text' => $chunk_text, | |
| 732 | - 'source_url' => $source_url, | |
| 733 | - 'type' => $content_type, | |
| 734 | - 'is_chunked' => true, | |
| 735 | - 'chunk_index' => $chunk_metadata['chunk_index'], | |
| 736 | - 'total_chunks' => $chunk_metadata['total_chunks'], | |
| 737 | - 'parent_url_hash' => $chunk_metadata['parent_url_hash'], | |
| 738 | - 'last_updated' => time(), | |
| 739 | - 'created_at' => time(), | |
| 740 | - 'bot_id' => $bot_id, | |
| 741 | - 'embedding_model' => self::get_active_embedding_model() | |
| 742 | - ); | |
| 743 | - | |
| 744 | - $vector_data = array( | |
| 745 | - 'id' => $vector_id, | |
| 746 | - 'values' => $embedding_vector, | |
| 747 | - 'metadata' => $metadata | |
| 748 | - ); | |
| 749 | - | |
| 750 | - $request_body = array( | |
| 751 | - 'vectors' => array($vector_data) | |
| 752 | - ); | |
| 753 | - | |
| 754 | - if (!empty($namespace)) { | |
| 755 | - $request_body['namespace'] = $namespace; | |
| 756 | - } | |
| 757 | - | |
| 758 | - $response = wp_remote_post($api_endpoint, array( | |
| 759 | - 'headers' => array( | |
| 760 | - 'Api-Key' => $api_key, | |
| 761 | - 'accept' => 'application/json', | |
| 762 | - 'content-type' => 'application/json' | |
| 763 | - ), | |
| 764 | - 'body' => wp_json_encode($request_body), | |
| 765 | - 'timeout' => 30 | |
| 766 | - )); | |
| 767 | - | |
| 768 | - if (is_wp_error($response)) { | |
| 769 | - return $response; | |
| 770 | - } | |
| 771 | - | |
| 772 | - $response_code = wp_remote_retrieve_response_code($response); | |
| 773 | - if ($response_code !== 200) { | |
| 774 | - return new WP_Error('pinecone_api', 'Pinecone API error: HTTP ' . $response_code); | |
| 775 | - } | |
| 776 | - | |
| 777 | - return true; | |
| 778 | -} | |
| 779 | - | |
| 780 | -/** | |
| 781 | - * Store a single chunk in WordPress database | |
| 782 | - */ | |
| 783 | -private static function store_chunk_in_wordpress_db($content_with_metadata, $source_url, $embedding_vector_serialized, $table_name, $content_type, $chunk_metadata) { | |
| 784 | - global $wpdb; | |
| 785 | - | |
| 786 | - // For chunks, we always insert new rows (no duplicate checking) | |
| 787 | - // The URL includes chunk info in the metadata, but source_url stays the same for grouping | |
| 788 | - $result = $wpdb->insert( | |
| 789 | - $table_name, | |
| 790 | - array( | |
| 791 | - 'url' => $source_url, | |
| 792 | - 'article_content' => $content_with_metadata, | |
| 793 | - 'embedding_vector' => $embedding_vector_serialized, | |
| 794 | - 'source_url' => $source_url, | |
| 795 | - 'content_type' => $content_type, | |
| 796 | - 'embedding_model' => self::get_active_embedding_model(), | |
| 797 | - 'timestamp' => current_time('mysql') | |
| 798 | - ), | |
| 799 | - array('%s', '%s', '%s', '%s', '%s', '%s', '%s') | |
| 800 | - ); | |
| 801 | - | |
| 802 | - if ($result === false) { | |
| 803 | - return new WP_Error('database_failed', 'Failed to insert chunk: ' . $wpdb->last_error); | |
| 804 | - } | |
| 805 | - | |
| 806 | - return true; | |
| 807 | -} | |
| 808 | - | |
| 809 | -/** | |
| 810 | - * Delete all chunks for a given URL | |
| 811 | - * | |
| 812 | - * @param string $source_url The source URL | |
| 813 | - * @param string $bot_id The bot ID | |
| 814 | - * @return bool|WP_Error True on success, WP_Error on failure | |
| 815 | - */ | |
| 816 | -public static function delete_chunks_for_url($source_url, $bot_id = 'default') { | |
| 817 | - //error_log('[MXCHAT-CHUNK-DELETE] Deleting chunks for URL: ' . $source_url); | |
| 818 | - | |
| 819 | - if (self::is_pinecone_enabled_for_bot($bot_id)) { | |
| 820 | - return self::delete_pinecone_chunks_by_url($source_url, $bot_id); | |
| 821 | - } else { | |
| 822 | - return self::delete_wordpress_chunks_by_url($source_url); | |
| 823 | - } | |
| 824 | -} | |
| 825 | - | |
| 826 | -/** | |
| 827 | - * Delete all chunks for a URL from Pinecone | |
| 828 | - */ | |
| 829 | -private static function delete_pinecone_chunks_by_url($source_url, $bot_id) { | |
| 830 | - // Get Pinecone configuration | |
| 831 | - if ($bot_id === 'default' || !class_exists('MxChat_Multi_Bot_Manager')) { | |
| 832 | - $pinecone_options = get_option('mxchat_pinecone_addon_options'); | |
| 833 | - $api_key = $pinecone_options['mxchat_pinecone_api_key'] ?? ''; | |
| 834 | - $host = $pinecone_options['mxchat_pinecone_host'] ?? ''; | |
| 835 | - $namespace = $pinecone_options['mxchat_pinecone_namespace'] ?? ''; | |
| 836 | - } else { | |
| 837 | - $bot_pinecone_config = apply_filters('mxchat_get_bot_pinecone_config', array(), $bot_id); | |
| 838 | - if (empty($bot_pinecone_config)) { | |
| 839 | - $pinecone_options = get_option('mxchat_pinecone_addon_options'); | |
| 840 | - $api_key = $pinecone_options['mxchat_pinecone_api_key'] ?? ''; | |
| 841 | - $host = $pinecone_options['mxchat_pinecone_host'] ?? ''; | |
| 842 | - $namespace = $pinecone_options['mxchat_pinecone_namespace'] ?? ''; | |
| 843 | - } else { | |
| 844 | - $api_key = $bot_pinecone_config['api_key'] ?? ''; | |
| 845 | - $host = $bot_pinecone_config['host'] ?? ''; | |
| 846 | - $namespace = $bot_pinecone_config['namespace'] ?? ''; | |
| 847 | - } | |
| 848 | - } | |
| 849 | - | |
| 850 | - if (empty($host) || empty($api_key)) { | |
| 851 | - return new WP_Error('pinecone_config', 'Pinecone is not properly configured'); | |
| 852 | - } | |
| 853 | - | |
| 854 | - $base_vector_id = md5($source_url); | |
| 855 | - $vectors_to_delete = array(); | |
| 856 | - | |
| 857 | - // Add the original single-vector ID (for non-chunked content) | |
| 858 | - $vectors_to_delete[] = $base_vector_id; | |
| 859 | - | |
| 860 | - // Pinecone /vectors/list is a GET endpoint with query-string parameters; a POST here returns a | |
| 861 | - // non-200 silently and we end up only deleting the base vector, leaving chunks orphaned. | |
| 862 | - $query_params = array( | |
| 863 | - 'prefix' => $base_vector_id . '_chunk_', | |
| 864 | - 'limit' => 100, | |
| 865 | - ); | |
| 866 | - if (!empty($namespace)) { | |
| 867 | - $query_params['namespace'] = $namespace; | |
| 868 | - } | |
| 869 | - | |
| 870 | - $list_url = "https://{$host}/vectors/list?" . http_build_query($query_params); | |
| 871 | - | |
| 872 | - // Paginate in case a URL has more than 100 chunks. | |
| 873 | - do { | |
| 874 | - $list_response = wp_remote_get($list_url, array( | |
| 875 | - 'headers' => array( | |
| 876 | - 'Api-Key' => $api_key, | |
| 877 | - 'accept' => 'application/json', | |
| 878 | - ), | |
| 879 | - 'timeout' => 30, | |
| 880 | - )); | |
| 881 | - | |
| 882 | - if (is_wp_error($list_response) || wp_remote_retrieve_response_code($list_response) !== 200) { | |
| 883 | - break; | |
| 884 | - } | |
| 885 | - | |
| 886 | - $list_data = json_decode(wp_remote_retrieve_body($list_response), true); | |
| 887 | - if (!empty($list_data['vectors'])) { | |
| 888 | - foreach ($list_data['vectors'] as $vector) { | |
| 889 | - if (isset($vector['id'])) { | |
| 890 | - $vectors_to_delete[] = $vector['id']; | |
| 891 | - } | |
| 892 | - } | |
| 893 | - } | |
| 894 | - | |
| 895 | - $next_token = $list_data['pagination']['next'] ?? ''; | |
| 896 | - if (empty($next_token)) { | |
| 897 | - break; | |
| 898 | - } | |
| 899 | - | |
| 900 | - $query_params['paginationToken'] = $next_token; | |
| 901 | - $list_url = "https://{$host}/vectors/list?" . http_build_query($query_params); | |
| 902 | - } while (true); | |
| 903 | - | |
| 904 | - if (empty($vectors_to_delete)) { | |
| 905 | - //error_log('[MXCHAT-CHUNK-DELETE] No vectors found to delete'); | |
| 906 | - return true; | |
| 907 | - } | |
| 908 | - | |
| 909 | - //error_log('[MXCHAT-CHUNK-DELETE] Deleting ' . count($vectors_to_delete) . ' vectors from Pinecone'); | |
| 910 | - | |
| 911 | - // Delete vectors | |
| 912 | - $delete_url = "https://{$host}/vectors/delete"; | |
| 913 | - | |
| 914 | - $delete_body = array( | |
| 915 | - 'ids' => $vectors_to_delete | |
| 916 | - ); | |
| 917 | - | |
| 918 | - if (!empty($namespace)) { | |
| 919 | - $delete_body['namespace'] = $namespace; | |
| 920 | - } | |
| 921 | - | |
| 922 | - $delete_response = wp_remote_post($delete_url, array( | |
| 923 | - 'headers' => array( | |
| 924 | - 'Api-Key' => $api_key, | |
| 925 | - 'accept' => 'application/json', | |
| 926 | - 'content-type' => 'application/json' | |
| 927 | - ), | |
| 928 | - 'body' => wp_json_encode($delete_body), | |
| 929 | - 'timeout' => 30 | |
| 930 | - )); | |
| 931 | - | |
| 932 | - if (is_wp_error($delete_response)) { | |
| 933 | - return $delete_response; | |
| 934 | - } | |
| 935 | - | |
| 936 | - $response_code = wp_remote_retrieve_response_code($delete_response); | |
| 937 | - if ($response_code !== 200) { | |
| 938 | - return new WP_Error('pinecone_delete', 'Failed to delete vectors: HTTP ' . $response_code); | |
| 939 | - } | |
| 940 | - | |
| 941 | - return true; | |
| 942 | -} | |
| 943 | - | |
| 944 | -/** | |
| 945 | - * Delete all chunks for a URL from WordPress database | |
| 946 | - */ | |
| 947 | -private static function delete_wordpress_chunks_by_url($source_url) { | |
| 948 | - global $wpdb; | |
| 949 | - $table_name = $wpdb->prefix . 'mxchat_system_prompt_content'; | |
| 950 | - | |
| 951 | - // Delete all rows with this source_url (handles both chunked and non-chunked) | |
| 952 | - $result = $wpdb->delete( | |
| 953 | - $table_name, | |
| 954 | - array('source_url' => $source_url), | |
| 955 | - array('%s') | |
| 956 | - ); | |
| 957 | - | |
| 958 | - if ($result === false) { | |
| 959 | - return new WP_Error('database_delete', 'Failed to delete chunks: ' . $wpdb->last_error); | |
| 960 | - } | |
| 961 | - | |
| 962 | - //error_log('[MXCHAT-CHUNK-DELETE] Deleted ' . $result . ' rows from WordPress DB'); | |
| 963 | - return true; | |
| 964 | -} | |
| 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 $bot_id The bot ID for multi-bot support | |
| 17 | + * @return bool|WP_Error True on success, WP_Error on failure | |
| 18 | + */ | |
| 19 | +public static function submit_content_to_db($content, $source_url, $api_key, $vector_id = null, $bot_id = 'default') { | |
| 20 | + global $wpdb; | |
| 21 | + $table_name = $wpdb->prefix . 'mxchat_system_prompt_content'; | |
| 22 | + | |
| 23 | + //error_log('[MXCHAT-DB] Starting database submission for URL: ' . $source_url . ' (Bot: ' . $bot_id . ')'); | |
| 24 | + //error_log('[MXCHAT-DB] Content length: ' . strlen($content) . ' bytes'); | |
| 25 | + | |
| 26 | + // Sanitize the source URL | |
| 27 | + $source_url = esc_url_raw($source_url); | |
| 28 | + | |
| 29 | + // Just ensure UTF-8 validity without aggressive escaping | |
| 30 | + $safe_content = wp_check_invalid_utf8($content); | |
| 31 | + // Remove only null bytes and other control characters, but preserve newlines (\n = \x0A) and carriage returns (\r = \x0D) | |
| 32 | + $safe_content = preg_replace('/[\x00-\x08\x0B\x0C\x0E-\x1F\x7F]/u', '', $safe_content); | |
| 33 | + | |
| 34 | + // UPDATED: Generate the embedding using bot-specific configuration | |
| 35 | + $embedding_vector = self::generate_embedding($content, $api_key, $bot_id); | |
| 36 | + | |
| 37 | + if (!is_array($embedding_vector)) { | |
| 38 | + //error_log('[MXCHAT-DB] Error: Embedding generation failed'); | |
| 39 | + return new WP_Error('embedding_failed', 'Failed to generate embedding for content'); | |
| 40 | + } | |
| 41 | + | |
| 42 | + //error_log('[MXCHAT-DB] Embedding generated successfully'); | |
| 43 | + | |
| 44 | + // UPDATED: Check if Pinecone is enabled for this specific bot | |
| 45 | + if (self::is_pinecone_enabled_for_bot($bot_id)) { | |
| 46 | + //error_log('[MXCHAT-DB] Pinecone is enabled for bot ' . $bot_id . ' - using Pinecone storage'); | |
| 47 | + // Store in Pinecone only | |
| 48 | + return self::store_in_pinecone_only($embedding_vector, $content, $source_url, $vector_id, $bot_id); | |
| 49 | + } else { | |
| 50 | + //error_log('[MXCHAT-DB] Pinecone not enabled for bot ' . $bot_id . ' - using WordPress storage'); | |
| 51 | + // Store in WordPress database only | |
| 52 | + $embedding_vector_serialized = maybe_serialize($embedding_vector); | |
| 53 | + return self::store_in_wordpress_db($safe_content, $source_url, $embedding_vector_serialized, $table_name); | |
| 54 | + } | |
| 55 | +} | |
| 56 | + | |
| 57 | +/** | |
| 58 | + * UPDATED: Check if Pinecone is enabled and properly configured for a specific bot | |
| 59 | + */ | |
| 60 | +private static function is_pinecone_enabled_for_bot($bot_id = 'default') { | |
| 61 | + // For default bot or when multi-bot is not active, use original method | |
| 62 | + if ($bot_id === 'default' || !class_exists('MxChat_Multi_Bot_Manager')) { | |
| 63 | + return self::is_pinecone_enabled(); | |
| 64 | + } | |
| 65 | + | |
| 66 | + // Get bot-specific Pinecone configuration | |
| 67 | + $bot_pinecone_config = apply_filters('mxchat_get_bot_pinecone_config', array(), $bot_id); | |
| 68 | + | |
| 69 | + if (empty($bot_pinecone_config)) { | |
| 70 | + // Fallback to default configuration | |
| 71 | + return self::is_pinecone_enabled(); | |
| 72 | + } | |
| 73 | + | |
| 74 | + $enabled_check = !empty($bot_pinecone_config['use_pinecone']) && $bot_pinecone_config['use_pinecone']; | |
| 75 | + $api_key_check = !empty($bot_pinecone_config['api_key']); | |
| 76 | + $host_check = !empty($bot_pinecone_config['host']); | |
| 77 | + | |
| 78 | + return $enabled_check && $api_key_check && $host_check; | |
| 79 | +} | |
| 80 | + | |
| 81 | +/** | |
| 82 | + * Check if Pinecone is enabled and properly configured (original method for default bot) | |
| 83 | + */ | |
| 84 | +private static function is_pinecone_enabled() { | |
| 85 | + $pinecone_options = get_option('mxchat_pinecone_addon_options'); | |
| 86 | + | |
| 87 | + if (empty($pinecone_options)) { | |
| 88 | + return false; | |
| 89 | + } | |
| 90 | + | |
| 91 | + $enabled_check = !empty($pinecone_options['mxchat_use_pinecone']) && $pinecone_options['mxchat_use_pinecone'] !== '0'; | |
| 92 | + $api_key_check = !empty($pinecone_options['mxchat_pinecone_api_key']); | |
| 93 | + $host_check = !empty($pinecone_options['mxchat_pinecone_host']); | |
| 94 | + | |
| 95 | + return $enabled_check && $api_key_check && $host_check; | |
| 96 | +} | |
| 97 | + | |
| 98 | +/** | |
| 99 | + * UPDATED: Store content in Pinecone only with bot support | |
| 100 | + */ | |
| 101 | +private static function store_in_pinecone_only($embedding_vector, $content, $source_url, $vector_id = null, $bot_id = 'default') { | |
| 102 | + //error_log('[MXCHAT-PINECONE] ===== Using Pinecone-only storage for bot ' . $bot_id . ' ====='); | |
| 103 | + | |
| 104 | + // Get bot-specific Pinecone configuration | |
| 105 | + if ($bot_id === 'default' || !class_exists('MxChat_Multi_Bot_Manager')) { | |
| 106 | + $pinecone_options = get_option('mxchat_pinecone_addon_options'); | |
| 107 | + $api_key = $pinecone_options['mxchat_pinecone_api_key']; | |
| 108 | + $environment = $pinecone_options['mxchat_pinecone_environment'] ?? ''; | |
| 109 | + $index_name = $pinecone_options['mxchat_pinecone_index'] ?? ''; | |
| 110 | + $namespace = $pinecone_options['mxchat_pinecone_namespace'] ?? ''; | |
| 111 | + } else { | |
| 112 | + $bot_pinecone_config = apply_filters('mxchat_get_bot_pinecone_config', array(), $bot_id); | |
| 113 | + if (empty($bot_pinecone_config)) { | |
| 114 | + // Fallback to default configuration | |
| 115 | + $pinecone_options = get_option('mxchat_pinecone_addon_options'); | |
| 116 | + $api_key = $pinecone_options['mxchat_pinecone_api_key']; | |
| 117 | + $environment = $pinecone_options['mxchat_pinecone_environment'] ?? ''; | |
| 118 | + $index_name = $pinecone_options['mxchat_pinecone_index'] ?? ''; | |
| 119 | + $namespace = $pinecone_options['mxchat_pinecone_namespace'] ?? ''; | |
| 120 | + } else { | |
| 121 | + $api_key = $bot_pinecone_config['api_key']; | |
| 122 | + $environment = ''; // Not used in new Pinecone API | |
| 123 | + $index_name = ''; // Not used in new Pinecone API | |
| 124 | + $namespace = $bot_pinecone_config['namespace'] ?? ''; | |
| 125 | + } | |
| 126 | + } | |
| 127 | + | |
| 128 | + $result = self::store_in_pinecone_main( | |
| 129 | + $embedding_vector, | |
| 130 | + $content, | |
| 131 | + $source_url, | |
| 132 | + $api_key, | |
| 133 | + $environment, | |
| 134 | + $index_name, | |
| 135 | + $vector_id, | |
| 136 | + $bot_id, | |
| 137 | + $namespace | |
| 138 | + ); | |
| 139 | + | |
| 140 | + if (is_wp_error($result)) { | |
| 141 | + //error_log('[MXCHAT-PINECONE] Pinecone storage failed for bot ' . $bot_id . ': ' . $result->get_error_message()); | |
| 142 | + return $result; | |
| 143 | + } | |
| 144 | + | |
| 145 | + //error_log('[MXCHAT-PINECONE] Pinecone storage completed successfully for bot ' . $bot_id); | |
| 146 | + return true; | |
| 147 | +} | |
| 148 | + | |
| 149 | +/** | |
| 150 | + * Store content in WordPress database with progressive fallback (unchanged) | |
| 151 | + */ | |
| 152 | +private static function store_in_wordpress_db($safe_content, $source_url, $embedding_vector_serialized, $table_name) { | |
| 153 | + global $wpdb; | |
| 154 | + | |
| 155 | + //error_log('[MXCHAT-DB] ===== Using WordPress-only storage ====='); | |
| 156 | + | |
| 157 | + // ===== FIXED: Generate unique identifier for manual content ===== | |
| 158 | + $original_source_url = $source_url; | |
| 159 | + $is_manual_content = empty($source_url) || $source_url === '' || !filter_var($source_url, FILTER_VALIDATE_URL); | |
| 160 | + | |
| 161 | + if ($is_manual_content) { | |
| 162 | + // Generate unique identifier for manual content to prevent overwrites | |
| 163 | + $source_url = 'mxchat://manual-content/' . time() . '-' . wp_generate_password(8, false); | |
| 164 | + //error_log('[MXCHAT-DB] Generated unique ID for manual content: ' . $source_url); | |
| 165 | + } | |
| 166 | + | |
| 167 | + // Only check for duplicates if we have a valid source URL (not manual content) | |
| 168 | + $existing_id = null; | |
| 169 | + if (!$is_manual_content) { | |
| 170 | + $existing_id = $wpdb->get_var( | |
| 171 | + $wpdb->prepare( | |
| 172 | + "SELECT id FROM {$table_name} WHERE source_url = %s LIMIT 1", | |
| 173 | + $source_url | |
| 174 | + ) | |
| 175 | + ); | |
| 176 | + //error_log('[MXCHAT-DB] Checked for existing URL, found ID: ' . ($existing_id ?: 'none')); | |
| 177 | + } else { | |
| 178 | + //error_log('[MXCHAT-DB] Manual content - will create new entry (no duplicate check)'); | |
| 179 | + } | |
| 180 | + // ===== END FIX ===== | |
| 181 | + | |
| 182 | + // Progressive fallback mechanism for problematic content | |
| 183 | + $attempt = 1; | |
| 184 | + $max_attempts = 3; | |
| 185 | + $current_content = $safe_content; | |
| 186 | + $result = false; | |
| 187 | + | |
| 188 | + while ($attempt <= $max_attempts && $result === false) { | |
| 189 | + try { | |
| 190 | + if ($existing_id) { | |
| 191 | + //error_log('[MXCHAT-DB] Found existing entry (ID: ' . $existing_id . '). Updating... (Attempt ' . $attempt . ')'); | |
| 192 | + | |
| 193 | + // Update the existing row | |
| 194 | + $result = $wpdb->update( | |
| 195 | + $table_name, | |
| 196 | + array( | |
| 197 | + 'url' => $source_url, | |
| 198 | + 'article_content' => $current_content, | |
| 199 | + 'embedding_vector' => $embedding_vector_serialized, | |
| 200 | + 'source_url' => $source_url, | |
| 201 | + 'timestamp' => current_time('mysql'), | |
| 202 | + ), | |
| 203 | + array('id' => $existing_id), | |
| 204 | + array('%s','%s','%s','%s','%s'), | |
| 205 | + array('%d') | |
| 206 | + ); | |
| 207 | + } else { | |
| 208 | + //error_log('[MXCHAT-DB] No existing entry found. Inserting new row... (Attempt ' . $attempt . ')'); | |
| 209 | + //error_log('[MXCHAT-DB] Content sample: ' . substr($current_content, 0, 1000)); | |
| 210 | + | |
| 211 | + // Insert a new row (using generated unique ID for manual content) | |
| 212 | + $result = $wpdb->insert( | |
| 213 | + $table_name, | |
| 214 | + array( | |
| 215 | + 'url' => $source_url, // Now unique for manual content | |
| 216 | + 'article_content' => $current_content, | |
| 217 | + 'embedding_vector' => $embedding_vector_serialized, | |
| 218 | + 'source_url' => $source_url, // Now unique for manual content | |
| 219 | + 'timestamp' => current_time('mysql'), | |
| 220 | + ), | |
| 221 | + array('%s','%s','%s','%s','%s') | |
| 222 | + ); | |
| 223 | + } | |
| 224 | + | |
| 225 | + if ($result === false) { | |
| 226 | + //error_log('[MXCHAT-DB] Database operation failed (Attempt ' . $attempt . '): ' . $wpdb->last_error); | |
| 227 | + | |
| 228 | + // Progressively apply more aggressive sanitization on failure | |
| 229 | + if ($attempt === 1) { | |
| 230 | + // First fallback: Use a more aggressive character filter and shorten | |
| 231 | + $current_content = preg_replace('/[^\p{L}\p{N}\s.,;:!?()-]/u', '', $current_content); | |
| 232 | + $current_content = substr($current_content, 0, 50000); | |
| 233 | + } else if ($attempt === 2) { | |
| 234 | + // Second fallback: Keep only alphanumeric and basic punctuation, shorten further | |
| 235 | + $current_content = preg_replace('/[^a-zA-Z0-9\s.,;:!?()-]/u', '', $current_content); | |
| 236 | + $current_content = substr($current_content, 0, 30000); | |
| 237 | + } | |
| 238 | + | |
| 239 | + $attempt++; | |
| 240 | + } | |
| 241 | + } catch (Exception $e) { | |
| 242 | + //error_log('[MXCHAT-DB] Exception during database operation: ' . $e->getMessage()); | |
| 243 | + $attempt++; | |
| 244 | + } | |
| 245 | + } | |
| 246 | + | |
| 247 | + if ($result === false) { | |
| 248 | + //error_log('[MXCHAT-DB] All database operation attempts failed'); | |
| 249 | + return new WP_Error('database_failed', 'Failed to store content in WordPress database after ' . $max_attempts . ' attempts'); | |
| 250 | + } | |
| 251 | + | |
| 252 | + //error_log('[MXCHAT-DB] WordPress database operation completed successfully (Attempt ' . ($attempt - 1) . ')'); | |
| 253 | + return true; | |
| 254 | +} | |
| 255 | + | |
| 256 | +/** | |
| 257 | + * UPDATED: Store content in Pinecone database with bot support | |
| 258 | + */ | |
| 259 | +private static function store_in_pinecone_main($embedding_vector, $content, $url, $api_key, $environment, $index_name, $vector_id = null, $bot_id = 'default', $namespace = '') { | |
| 260 | + //error_log('[MXCHAT-PINECONE-MAIN] ===== Starting Pinecone storage for bot ' . $bot_id . ' ====='); | |
| 261 | + | |
| 262 | + // ===== UPDATED: Handle manual content with unique vector IDs ===== | |
| 263 | + if ($vector_id) { | |
| 264 | + // Use provided vector ID | |
| 265 | + //error_log('[MXCHAT-PINECONE-MAIN] Using provided vector ID: ' . $vector_id); | |
| 266 | + } elseif (!empty($url) && filter_var($url, FILTER_VALIDATE_URL)) { | |
| 267 | + // For valid URLs, use URL-based ID (existing behavior) | |
| 268 | + $vector_id = md5($url); | |
| 269 | + //error_log('[MXCHAT-PINECONE-MAIN] Generated vector ID from URL: ' . $vector_id); | |
| 270 | + } else { | |
| 271 | + // For manual content (empty/invalid URL), generate unique ID | |
| 272 | + $vector_id = 'manual_' . time() . '_' . substr(md5($content . microtime(true)), 0, 8); | |
| 273 | + //error_log('[MXCHAT-PINECONE-MAIN] Generated unique vector ID for manual content: ' . $vector_id); | |
| 274 | + } | |
| 275 | + // ===== END UPDATE ===== | |
| 276 | + | |
| 277 | + // Get host from bot-specific config or fallback to default | |
| 278 | + if ($bot_id === 'default' || !class_exists('MxChat_Multi_Bot_Manager')) { | |
| 279 | + $options = get_option('mxchat_pinecone_addon_options'); | |
| 280 | + $host = $options['mxchat_pinecone_host'] ?? ''; | |
| 281 | + } else { | |
| 282 | + $bot_pinecone_config = apply_filters('mxchat_get_bot_pinecone_config', array(), $bot_id); | |
| 283 | + if (!empty($bot_pinecone_config)) { | |
| 284 | + $host = $bot_pinecone_config['host'] ?? ''; | |
| 285 | + } else { | |
| 286 | + $options = get_option('mxchat_pinecone_addon_options'); | |
| 287 | + $host = $options['mxchat_pinecone_host'] ?? ''; | |
| 288 | + } | |
| 289 | + } | |
| 290 | + | |
| 291 | + //error_log('[MXCHAT-PINECONE-MAIN] Host: ' . $host); | |
| 292 | + //error_log('[MXCHAT-PINECONE-MAIN] API key length: ' . strlen($api_key)); | |
| 293 | + //error_log('[MXCHAT-PINECONE-MAIN] Bot ID: ' . $bot_id); | |
| 294 | + //error_log('[MXCHAT-PINECONE-MAIN] Namespace: ' . $namespace); | |
| 295 | + | |
| 296 | + if (empty($host)) { | |
| 297 | + //error_log('[MXCHAT-PINECONE-MAIN] ERROR: Host is empty'); | |
| 298 | + return new WP_Error('pinecone_config', 'Pinecone host is not configured. Please set the host in your bot settings.'); | |
| 299 | + } | |
| 300 | + | |
| 301 | + // ===== UPDATED: Determine content type more accurately ===== | |
| 302 | + $is_product = false; | |
| 303 | + $content_type = 'manual'; // Default for manual content | |
| 304 | + | |
| 305 | + if (!empty($url) && filter_var($url, FILTER_VALIDATE_URL)) { | |
| 306 | + $is_product = (strpos($url, '/product/') !== false || strpos($url, '/shop/') !== false); | |
| 307 | + $content_type = $is_product ? 'product' : 'content'; | |
| 308 | + } | |
| 309 | + | |
| 310 | + //error_log('[MXCHAT-PINECONE-MAIN] Content type: ' . $content_type); | |
| 311 | + // ===== END UPDATE ===== | |
| 312 | + | |
| 313 | + $api_endpoint = "https://{$host}/vectors/upsert"; | |
| 314 | + //error_log('[MXCHAT-PINECONE-MAIN] API endpoint: ' . $api_endpoint); | |
| 315 | + | |
| 316 | + // UPDATED: Add bot_id to metadata and handle namespace | |
| 317 | + $metadata = array( | |
| 318 | + 'text' => $content, | |
| 319 | + 'source_url' => $url, // Can be empty for manual content | |
| 320 | + 'type' => $content_type, // 'manual', 'content', or 'product' | |
| 321 | + 'last_updated' => time(), | |
| 322 | + 'created_at' => time(), // Add creation timestamp | |
| 323 | + 'bot_id' => $bot_id // Add bot identification | |
| 324 | + ); | |
| 325 | + | |
| 326 | + $vector_data = array( | |
| 327 | + 'id' => $vector_id, | |
| 328 | + 'values' => $embedding_vector, | |
| 329 | + 'metadata' => $metadata | |
| 330 | + ); | |
| 331 | + | |
| 332 | + $request_body = array( | |
| 333 | + 'vectors' => array($vector_data) | |
| 334 | + ); | |
| 335 | + | |
| 336 | + // Add namespace if specified for multi-bot separation | |
| 337 | + if (!empty($namespace)) { | |
| 338 | + $request_body['namespace'] = $namespace; | |
| 339 | + //error_log('[MXCHAT-PINECONE-MAIN] Using namespace: ' . $namespace); | |
| 340 | + } | |
| 341 | + | |
| 342 | + //error_log('[MXCHAT-PINECONE-MAIN] Request body prepared (embedding dimensions: ' . count($embedding_vector) . ')'); | |
| 343 | + | |
| 344 | + $response = wp_remote_post($api_endpoint, array( | |
| 345 | + 'headers' => array( | |
| 346 | + 'Api-Key' => $api_key, | |
| 347 | + 'accept' => 'application/json', | |
| 348 | + 'content-type' => 'application/json' | |
| 349 | + ), | |
| 350 | + 'body' => wp_json_encode($request_body), | |
| 351 | + 'timeout' => 30, | |
| 352 | + 'data_format' => 'body' | |
| 353 | + )); | |
| 354 | + | |
| 355 | + if (is_wp_error($response)) { | |
| 356 | + //error_log('[MXCHAT-PINECONE-MAIN] WordPress request error: ' . $response->get_error_message()); | |
| 357 | + return new WP_Error('pinecone_request', $response->get_error_message()); | |
| 358 | + } | |
| 359 | + | |
| 360 | + $response_code = wp_remote_retrieve_response_code($response); | |
| 361 | + //error_log('[MXCHAT-PINECONE-MAIN] Response code: ' . $response_code); | |
| 362 | + | |
| 363 | + if ($response_code !== 200) { | |
| 364 | + $body = wp_remote_retrieve_body($response); | |
| 365 | + //error_log('[MXCHAT-PINECONE-MAIN] API error - Response body: ' . $body); | |
| 366 | + return new WP_Error('pinecone_api', sprintf( | |
| 367 | + 'Pinecone API error (HTTP %d): %s', | |
| 368 | + $response_code, | |
| 369 | + $body | |
| 370 | + )); | |
| 371 | + } | |
| 372 | + | |
| 373 | + $response_body = wp_remote_retrieve_body($response); | |
| 374 | + //error_log('[MXCHAT-PINECONE-MAIN] Success response: ' . $response_body); | |
| 375 | + //error_log('[MXCHAT-PINECONE-MAIN] Successfully stored in Pinecone for bot ' . $bot_id); | |
| 376 | + //error_log('[MXCHAT-PINECONE-MAIN] ===== Pinecone storage complete ====='); | |
| 377 | + | |
| 378 | + return true; | |
| 379 | +} | |
| 380 | + | |
| 381 | +/** | |
| 382 | + * UPDATED: Generate an embedding for the given text using bot-specific configuration. | |
| 383 | + * | |
| 384 | + * @param string $text The text to be embedded. | |
| 385 | + * @param string $api_key The API key used for generating embeddings. | |
| 386 | + * @param string $bot_id The bot ID for multi-bot support | |
| 387 | + * @return array|null The embedding vector or null on failure. | |
| 388 | + */ | |
| 389 | +private static function generate_embedding($text, $api_key, $bot_id = 'default') { | |
| 390 | + // Get bot-specific options | |
| 391 | + if ($bot_id === 'default' || !class_exists('MxChat_Multi_Bot_Manager')) { | |
| 392 | + $options = get_option('mxchat_options'); | |
| 393 | + } else { | |
| 394 | + $bot_options = apply_filters('mxchat_get_bot_options', array(), $bot_id); | |
| 395 | + $options = !empty($bot_options) ? $bot_options : get_option('mxchat_options'); | |
| 396 | + } | |
| 397 | + | |
| 398 | + $selected_model = $options['embedding_model'] ?? 'text-embedding-ada-002'; | |
| 399 | + | |
| 400 | + // Determine endpoint and API key based on model | |
| 401 | + if (strpos($selected_model, 'voyage') === 0) { | |
| 402 | + $endpoint = 'https://api.voyageai.com/v1/embeddings'; | |
| 403 | + $api_key = $options['voyage_api_key'] ?? ''; | |
| 404 | + } elseif (strpos($selected_model, 'gemini-embedding') === 0) { | |
| 405 | + $endpoint = 'https://generativelanguage.googleapis.com/v1beta/models/' . $selected_model . ':embedContent'; | |
| 406 | + $api_key = $options['gemini_api_key'] ?? ''; | |
| 407 | + } else { | |
| 408 | + $endpoint = 'https://api.openai.com/v1/embeddings'; | |
| 409 | + // Use the bot-specific API key or fallback to passed API key | |
| 410 | + $api_key = $options['api_key'] ?? $api_key; | |
| 411 | + } | |
| 412 | + | |
| 413 | + // Prepare request body based on provider | |
| 414 | + if (strpos($selected_model, 'gemini-embedding') === 0) { | |
| 415 | + // Gemini API format | |
| 416 | + $request_body = [ | |
| 417 | + 'model' => 'models/' . $selected_model, | |
| 418 | + 'content' => [ | |
| 419 | + 'parts' => [ | |
| 420 | + ['text' => $text] | |
| 421 | + ] | |
| 422 | + ], | |
| 423 | + 'outputDimensionality' => 1536 | |
| 424 | + ]; | |
| 425 | + | |
| 426 | + // Prepare headers for Gemini (API key as query parameter) | |
| 427 | + $endpoint .= '?key=' . $api_key; | |
| 428 | + $headers = [ | |
| 429 | + 'Content-Type' => 'application/json' | |
| 430 | + ]; | |
| 431 | + } else { | |
| 432 | + // OpenAI/Voyage API format | |
| 433 | + $request_body = [ | |
| 434 | + 'input' => $text, | |
| 435 | + 'model' => $selected_model | |
| 436 | + ]; | |
| 437 | + | |
| 438 | + // Add output_dimension for voyage-3-large | |
| 439 | + if ($selected_model === 'voyage-3-large') { | |
| 440 | + $request_body['output_dimension'] = 2048; | |
| 441 | + } | |
| 442 | + | |
| 443 | + // Prepare headers for OpenAI/Voyage | |
| 444 | + $headers = [ | |
| 445 | + 'Content-Type' => 'application/json', | |
| 446 | + 'Authorization' => 'Bearer ' . $api_key | |
| 447 | + ]; | |
| 448 | + } | |
| 449 | + | |
| 450 | + $args = [ | |
| 451 | + 'body' => wp_json_encode($request_body), | |
| 452 | + 'headers' => $headers, | |
| 453 | + 'timeout' => 60, | |
| 454 | + 'redirection' => 5, | |
| 455 | + 'blocking' => true, | |
| 456 | + 'httpversion' => '1.0', | |
| 457 | + 'sslverify' => true, | |
| 458 | + ]; | |
| 459 | + | |
| 460 | + $response = wp_remote_post($endpoint, $args); | |
| 461 | + | |
| 462 | + if (is_wp_error($response)) { | |
| 463 | + //error_log('Error generating embedding for bot ' . $bot_id . ': ' . $response->get_error_message()); | |
| 464 | + return null; | |
| 465 | + } | |
| 466 | + | |
| 467 | + $response_body = json_decode(wp_remote_retrieve_body($response), true); | |
| 468 | + | |
| 469 | + // Handle different response formats based on provider | |
| 470 | + if (strpos($selected_model, 'gemini-embedding') === 0) { | |
| 471 | + // Gemini API response format | |
| 472 | + if (isset($response_body['embedding']['values']) && is_array($response_body['embedding']['values'])) { | |
| 473 | + return $response_body['embedding']['values']; | |
| 474 | + } else { | |
| 475 | + //error_log('Invalid response received from Gemini embedding API for bot ' . $bot_id . ': ' . wp_json_encode($response_body)); | |
| 476 | + return null; | |
| 477 | + } | |
| 478 | + } else { | |
| 479 | + // OpenAI/Voyage API response format | |
| 480 | + if (isset($response_body['data'][0]['embedding']) && is_array($response_body['data'][0]['embedding'])) { | |
| 481 | + return $response_body['data'][0]['embedding']; | |
| 482 | + } else { | |
| 483 | + //error_log('Invalid response received from embedding API for bot ' . $bot_id . ': ' . wp_json_encode($response_body)); | |
| 484 | + return null; | |
| 485 | + } | |
| 486 | + } | |
| 487 | +} | |
| 965 | 488 | } |
