| 1 |
<?php |
| 2 |
if (!defined('ABSPATH')) { |
| 3 |
exit; // Exit if accessed directly |
| 4 |
} |
| 5 |
|
| 6 |
class MxChat_Utils { |
| 7 |
|
| 8 |
/** |
| 9 |
* Validate a client-supplied session id (plan-mxchat-20260731-d42bec). |
| 10 |
* |
| 11 |
* sanitize_text_field() — which every session_id read site used before this — |
| 12 |
* preserves '/' and '..'. Harmless where the value is only an option or |
| 13 |
* transient key suffix, but mxchat_send_delayed_transcript() interpolates it |
| 14 |
* into a filesystem path, so '../../../../path/x' wrote, emailed and deleted a |
| 15 |
* file outside the uploads dir. |
| 16 |
* |
| 17 |
* REJECTS rather than rewrites: a silently-stripped id would orphan the |
| 18 |
* conversation it belongs to, which is harder to diagnose than a clean refusal. |
| 19 |
* Returns '' for anything malformed, so call sites fall into the empty-session |
| 20 |
* error paths they already have. |
| 21 |
* |
| 22 |
* The generator only ever emits 'mxchat_chat_' + 32 hex chars |
| 23 |
* (class-mxchat-integrator.php, js/chat-script.js), so this is not restrictive |
| 24 |
* in practice. Length ceiling is deliberate — session ids are also used as |
| 25 |
* option-name suffixes, and WP option names cap at 191 chars. |
| 26 |
* |
| 27 |
* @param mixed $raw Raw request value. |
| 28 |
* @return string The id if well-formed, '' otherwise. |
| 29 |
*/ |
| 30 |
public static function sanitize_session_id($raw) { |
| 31 |
if (!is_scalar($raw)) { |
| 32 |
return ''; |
| 33 |
} |
| 34 |
$val = trim((string) $raw); |
| 35 |
if ($val === '') { |
| 36 |
return ''; |
| 37 |
} |
| 38 |
return preg_match('/\A[A-Za-z0-9_-]{1,128}\z/', $val) ? $val : ''; |
| 39 |
} |
| 40 |
|
| 41 |
/** |
| 42 |
* Centralized embedding model registry. Single source of truth for dimensions |
| 43 |
* and provider, so model-switch protection logic doesn't drift across files. |
| 44 |
*/ |
| 45 |
public static function embedding_model_registry() { |
| 46 |
return array( |
| 47 |
'text-embedding-ada-002' => array('dims' => 1536, 'provider' => 'openai', 'label' => 'Ada 2'), |
| 48 |
'text-embedding-3-small' => array('dims' => 1536, 'provider' => 'openai', 'label' => 'TE3 Small'), |
| 49 |
'text-embedding-3-large' => array('dims' => 3072, 'provider' => 'openai', 'label' => 'TE3 Large'), |
| 50 |
'voyage-3-large' => array('dims' => 2048, 'provider' => 'voyage', 'label' => 'Voyage-3 Large'), |
| 51 |
'gemini-embedding-001' => array('dims' => 1536, 'provider' => 'gemini', 'label' => 'Gemini Embedding'), |
| 52 |
); |
| 53 |
} |
| 54 |
|
| 55 |
public static function embedding_model_dimensions($model) { |
| 56 |
$registry = self::embedding_model_registry(); |
| 57 |
return isset($registry[$model]) ? (int) $registry[$model]['dims'] : 0; |
| 58 |
} |
| 59 |
|
| 60 |
public static function embedding_model_label($model) { |
| 61 |
$registry = self::embedding_model_registry(); |
| 62 |
return isset($registry[$model]) ? $registry[$model]['label'] : $model; |
| 63 |
} |
| 64 |
|
| 65 |
/** |
| 66 |
* Returns the model that was last used to actually write embeddings into the |
| 67 |
* KB. Differs from the user-selected setting once a switch has happened but |
| 68 |
* no re-embed has occurred yet — that's the mismatch state we warn about. |
| 69 |
*/ |
| 70 |
public static function get_active_embedding_model() { |
| 71 |
return get_option('mxchat_active_embedding_model', ''); |
| 72 |
} |
| 73 |
|
| 74 |
/** |
| 75 |
* Stamp the model that produced the most recent successful embedding. Called |
| 76 |
* from generate_embedding() right after the API responds with a valid vector. |
| 77 |
*/ |
| 78 |
public static function stamp_active_embedding_model($model) { |
| 79 |
if (!empty($model) && $model !== self::get_active_embedding_model()) { |
| 80 |
update_option('mxchat_active_embedding_model', $model, false); |
| 81 |
} |
| 82 |
} |
| 83 |
|
| 84 |
/** |
| 85 |
* Extract the 11-character YouTube video ID from a URL, or '' if the URL is |
| 86 |
* not a single-video YouTube link. Single source of truth for both the KB |
| 87 |
* ingestion side and the chat render side — do not duplicate this parsing. |
| 88 |
* Channel, playlist, and search URLs deliberately return '' (only a URL that |
| 89 |
* identifies one video can be embedded). |
| 90 |
*/ |
| 91 |
public static function parse_youtube_id($url) { |
| 92 |
if (!is_string($url) || $url === '' || !preg_match('#^https?://#i', $url)) { |
| 93 |
return ''; |
| 94 |
} |
| 95 |
$host = strtolower((string) wp_parse_url($url, PHP_URL_HOST)); |
| 96 |
$host = preg_replace('/^(www|m)\./', '', $host); |
| 97 |
$path = (string) wp_parse_url($url, PHP_URL_PATH); |
| 98 |
$id = ''; |
| 99 |
if ($host === 'youtu.be') { |
| 100 |
$segments = explode('/', ltrim($path, '/')); |
| 101 |
$id = $segments[0] ?? ''; |
| 102 |
} elseif (in_array($host, array('youtube.com', 'youtube-nocookie.com'), true)) { |
| 103 |
if (preg_match('#^/(?:shorts|embed|live|v)/([A-Za-z0-9_-]+)#', $path, $m)) { |
| 104 |
$id = $m[1]; |
| 105 |
} elseif ($path === '/watch') { |
| 106 |
parse_str((string) wp_parse_url($url, PHP_URL_QUERY), $query_vars); |
| 107 |
$id = isset($query_vars['v']) ? (string) $query_vars['v'] : ''; |
| 108 |
} |
| 109 |
} |
| 110 |
$id = preg_replace('/[^A-Za-z0-9_-]/', '', (string) $id); |
| 111 |
return (strlen($id) === 11) ? $id : ''; |
| 112 |
} |
| 113 |
|
| 114 |
/** |
| 115 |
* UPDATED: Submit or update content (and its embedding) in the database. |
| 116 |
* Stores in Pinecone if enabled, otherwise stores in WordPress DB. |
| 117 |
* |
| 118 |
* @param string $content The content to be embedded. |
| 119 |
* @param string $source_url The source URL of the content. |
| 120 |
* @param string $api_key The API key used for generating embeddings. |
| 121 |
* @param string $vector_id Optional vector ID for Pinecone (if not provided, will use md5 of URL) |
| 122 |
* @param string $bot_id The bot ID for multi-bot support |
| 123 |
* @param string $content_type The type of content (post, page, pdf, url, manual, product, etc.) |
| 124 |
* @return bool|WP_Error True on success, WP_Error on failure |
| 125 |
*/ |
| 126 |
public static function submit_content_to_db($content, $source_url, $api_key, $vector_id = null, $bot_id = 'default', $content_type = 'content') { |
| 127 |
global $wpdb; |
| 128 |
$table_name = $wpdb->prefix . 'mxchat_system_prompt_content'; |
| 129 |
|
| 130 |
//error_log('[MXCHAT-DB] Starting database submission for URL: ' . $source_url . ' (Bot: ' . $bot_id . ', Type: ' . $content_type . ')'); |
| 131 |
//error_log('[MXCHAT-DB] Content length: ' . strlen($content) . ' bytes'); |
| 132 |
|
| 133 |
// Sanitize the source URL |
| 134 |
$source_url = esc_url_raw($source_url); |
| 135 |
|
| 136 |
// Sanitize content_type |
| 137 |
$content_type = sanitize_key($content_type); |
| 138 |
if (empty($content_type)) { |
| 139 |
$content_type = 'content'; // Fallback for backwards compatibility |
| 140 |
} |
| 141 |
|
| 142 |
// Just ensure UTF-8 validity without aggressive escaping |
| 143 |
$safe_content = wp_check_invalid_utf8($content); |
| 144 |
// Remove only null bytes and other control characters, but preserve newlines (\n = \x0A) and carriage returns (\r = \x0D) |
| 145 |
$safe_content = preg_replace('/[\x00-\x08\x0B\x0C\x0E-\x1F\x7F]/u', '', $safe_content); |
| 146 |
|
| 147 |
// Check if chunking should be applied |
| 148 |
$chunker = MxChat_Chunker::from_settings(); |
| 149 |
if ($chunker->should_chunk($safe_content)) { |
| 150 |
//error_log('[MXCHAT-DB] Content exceeds chunk threshold, using chunked submission'); |
| 151 |
return self::submit_chunked_content($safe_content, $source_url, $api_key, $bot_id, $content_type, $chunker); |
| 152 |
} |
| 153 |
|
| 154 |
// UPDATED: Generate the embedding using bot-specific configuration |
| 155 |
$embedding_vector = self::generate_embedding($content, $api_key, $bot_id); |
| 156 |
|
| 157 |
if (!is_array($embedding_vector)) { |
| 158 |
// Surface the provider's real reason instead of a fixed string (4a7c0a). |
| 159 |
$reason = is_wp_error($embedding_vector) |
| 160 |
? $embedding_vector->get_error_message() |
| 161 |
: 'Failed to generate embedding for content'; |
| 162 |
return new WP_Error('embedding_failed', $reason); |
| 163 |
} |
| 164 |
|
| 165 |
//error_log('[MXCHAT-DB] Embedding generated successfully'); |
| 166 |
|
| 167 |
// UPDATED: Check if Pinecone is enabled for this specific bot |
| 168 |
if (self::is_pinecone_enabled_for_bot($bot_id)) { |
| 169 |
//error_log('[MXCHAT-DB] Pinecone is enabled for bot ' . $bot_id . ' - using Pinecone storage'); |
| 170 |
// Store in Pinecone only |
| 171 |
return self::store_in_pinecone_only($embedding_vector, $content, $source_url, $vector_id, $bot_id, $content_type); |
| 172 |
} else { |
| 173 |
//error_log('[MXCHAT-DB] Pinecone not enabled for bot ' . $bot_id . ' - using WordPress storage'); |
| 174 |
// Store in WordPress database only |
| 175 |
$embedding_vector_serialized = maybe_serialize($embedding_vector); |
| 176 |
return self::store_in_wordpress_db($safe_content, $source_url, $embedding_vector_serialized, $table_name, $content_type); |
| 177 |
} |
| 178 |
} |
| 179 |
|
| 180 |
/** |
| 181 |
* UPDATED: Check if Pinecone is enabled and properly configured for a specific bot |
| 182 |
*/ |
| 183 |
private static function is_pinecone_enabled_for_bot($bot_id = 'default') { |
| 184 |
// For default bot or when multi-bot is not active, use original method |
| 185 |
if ($bot_id === 'default' || !class_exists('MxChat_Multi_Bot_Manager')) { |
| 186 |
return self::is_pinecone_enabled(); |
| 187 |
} |
| 188 |
|
| 189 |
// Get bot-specific Pinecone configuration |
| 190 |
$bot_pinecone_config = apply_filters('mxchat_get_bot_pinecone_config', array(), $bot_id); |
| 191 |
|
| 192 |
if (empty($bot_pinecone_config)) { |
| 193 |
// Fallback to default configuration |
| 194 |
return self::is_pinecone_enabled(); |
| 195 |
} |
| 196 |
|
| 197 |
$enabled_check = !empty($bot_pinecone_config['use_pinecone']) && $bot_pinecone_config['use_pinecone']; |
| 198 |
$api_key_check = !empty($bot_pinecone_config['api_key']); |
| 199 |
$host_check = !empty($bot_pinecone_config['host']); |
| 200 |
|
| 201 |
return $enabled_check && $api_key_check && $host_check; |
| 202 |
} |
| 203 |
|
| 204 |
/** |
| 205 |
* Check if Pinecone is enabled and properly configured (original method for default bot) |
| 206 |
*/ |
| 207 |
private static function is_pinecone_enabled() { |
| 208 |
$pinecone_options = get_option('mxchat_pinecone_addon_options'); |
| 209 |
|
| 210 |
if (empty($pinecone_options)) { |
| 211 |
return false; |
| 212 |
} |
| 213 |
|
| 214 |
$enabled_check = !empty($pinecone_options['mxchat_use_pinecone']) && $pinecone_options['mxchat_use_pinecone'] !== '0'; |
| 215 |
$api_key_check = !empty($pinecone_options['mxchat_pinecone_api_key']); |
| 216 |
$host_check = !empty($pinecone_options['mxchat_pinecone_host']); |
| 217 |
|
| 218 |
return $enabled_check && $api_key_check && $host_check; |
| 219 |
} |
| 220 |
|
| 221 |
/** |
| 222 |
* UPDATED: Store content in Pinecone only with bot support |
| 223 |
*/ |
| 224 |
private static function store_in_pinecone_only($embedding_vector, $content, $source_url, $vector_id = null, $bot_id = 'default', $content_type = 'content') { |
| 225 |
//error_log('[MXCHAT-PINECONE] ===== Using Pinecone-only storage for bot ' . $bot_id . ' ====='); |
| 226 |
|
| 227 |
// Get bot-specific Pinecone configuration |
| 228 |
if ($bot_id === 'default' || !class_exists('MxChat_Multi_Bot_Manager')) { |
| 229 |
$pinecone_options = get_option('mxchat_pinecone_addon_options'); |
| 230 |
$api_key = $pinecone_options['mxchat_pinecone_api_key']; |
| 231 |
$environment = $pinecone_options['mxchat_pinecone_environment'] ?? ''; |
| 232 |
$index_name = $pinecone_options['mxchat_pinecone_index'] ?? ''; |
| 233 |
$namespace = $pinecone_options['mxchat_pinecone_namespace'] ?? ''; |
| 234 |
} else { |
| 235 |
$bot_pinecone_config = apply_filters('mxchat_get_bot_pinecone_config', array(), $bot_id); |
| 236 |
if (empty($bot_pinecone_config)) { |
| 237 |
// Fallback to default configuration |
| 238 |
$pinecone_options = get_option('mxchat_pinecone_addon_options'); |
| 239 |
$api_key = $pinecone_options['mxchat_pinecone_api_key']; |
| 240 |
$environment = $pinecone_options['mxchat_pinecone_environment'] ?? ''; |
| 241 |
$index_name = $pinecone_options['mxchat_pinecone_index'] ?? ''; |
| 242 |
$namespace = $pinecone_options['mxchat_pinecone_namespace'] ?? ''; |
| 243 |
} else { |
| 244 |
$api_key = $bot_pinecone_config['api_key']; |
| 245 |
$environment = ''; // Not used in new Pinecone API |
| 246 |
$index_name = ''; // Not used in new Pinecone API |
| 247 |
$namespace = $bot_pinecone_config['namespace'] ?? ''; |
| 248 |
} |
| 249 |
} |
| 250 |
|
| 251 |
$result = self::store_in_pinecone_main( |
| 252 |
$embedding_vector, |
| 253 |
$content, |
| 254 |
$source_url, |
| 255 |
$api_key, |
| 256 |
$environment, |
| 257 |
$index_name, |
| 258 |
$vector_id, |
| 259 |
$bot_id, |
| 260 |
$namespace, |
| 261 |
$content_type |
| 262 |
); |
| 263 |
|
| 264 |
if (is_wp_error($result)) { |
| 265 |
//error_log('[MXCHAT-PINECONE] Pinecone storage failed for bot ' . $bot_id . ': ' . $result->get_error_message()); |
| 266 |
return $result; |
| 267 |
} |
| 268 |
|
| 269 |
//error_log('[MXCHAT-PINECONE] Pinecone storage completed successfully for bot ' . $bot_id); |
| 270 |
return true; |
| 271 |
} |
| 272 |
|
| 273 |
/** |
| 274 |
* Store content in WordPress database with progressive fallback |
| 275 |
* UPDATED 2.5.6: Now includes content_type parameter |
| 276 |
*/ |
| 277 |
private static function store_in_wordpress_db($safe_content, $source_url, $embedding_vector_serialized, $table_name, $content_type = 'content') { |
| 278 |
global $wpdb; |
| 279 |
|
| 280 |
//error_log('[MXCHAT-DB] ===== Using WordPress-only storage ====='); |
| 281 |
|
| 282 |
// Sanitize content_type |
| 283 |
$content_type = sanitize_key($content_type); |
| 284 |
if (empty($content_type)) { |
| 285 |
$content_type = 'content'; // Fallback for backwards compatibility |
| 286 |
} |
| 287 |
|
| 288 |
// ===== FIXED: Generate unique identifier for manual content ===== |
| 289 |
$original_source_url = $source_url; |
| 290 |
// Check if this is truly manual content (no URL at all) vs a real URL that filter_var rejects |
| 291 |
// filter_var(FILTER_VALIDATE_URL) rejects valid URLs with encoded chars, non-ASCII, fragments, etc. |
| 292 |
// Use a looser check: if it starts with http(s):// or has a scheme, it's a URL |
| 293 |
$has_url_scheme = !empty($source_url) && preg_match('#^https?://#i', $source_url); |
| 294 |
// Treat legacy mxchat.ai source URLs as manual — old bug assigned the site URL to manual entries |
| 295 |
$is_legacy_mxchat_url = $has_url_scheme && strpos($source_url, 'mxchat.ai') !== false; |
| 296 |
$is_manual_content = empty($source_url) || $source_url === '' || !$has_url_scheme || $is_legacy_mxchat_url; |
| 297 |
|
| 298 |
if ($is_manual_content) { |
| 299 |
// Generate unique identifier for manual content to prevent overwrites |
| 300 |
$source_url = 'mxchat://manual-content/' . time() . '-' . wp_generate_password(8, false); |
| 301 |
//error_log('[MXCHAT-DB] Generated unique ID for manual content: ' . $source_url); |
| 302 |
} |
| 303 |
|
| 304 |
// Only check for duplicates if we have a valid source URL (not manual content) |
| 305 |
$existing_id = null; |
| 306 |
if (!$is_manual_content) { |
| 307 |
$existing_id = $wpdb->get_var( |
| 308 |
$wpdb->prepare( |
| 309 |
"SELECT id FROM {$table_name} WHERE source_url = %s LIMIT 1", |
| 310 |
$source_url |
| 311 |
) |
| 312 |
); |
| 313 |
//error_log('[MXCHAT-DB] Checked for existing URL, found ID: ' . ($existing_id ?: 'none')); |
| 314 |
} else { |
| 315 |
//error_log('[MXCHAT-DB] Manual content - will create new entry (no duplicate check)'); |
| 316 |
} |
| 317 |
// ===== END FIX ===== |
| 318 |
|
| 319 |
// Progressive fallback mechanism for problematic content |
| 320 |
$attempt = 1; |
| 321 |
$max_attempts = 3; |
| 322 |
$current_content = $safe_content; |
| 323 |
$result = false; |
| 324 |
|
| 325 |
while ($attempt <= $max_attempts && $result === false) { |
| 326 |
try { |
| 327 |
if ($existing_id) { |
| 328 |
//error_log('[MXCHAT-DB] Found existing entry (ID: ' . $existing_id . '). Updating... (Attempt ' . $attempt . ')'); |
| 329 |
|
| 330 |
// Update the existing row - UPDATED 2.5.6: Added content_type |
| 331 |
$result = $wpdb->update( |
| 332 |
$table_name, |
| 333 |
array( |
| 334 |
'url' => $source_url, |
| 335 |
'article_content' => $current_content, |
| 336 |
'embedding_vector' => $embedding_vector_serialized, |
| 337 |
'source_url' => $source_url, |
| 338 |
'content_type' => $content_type, |
| 339 |
'timestamp' => current_time('mysql'), |
| 340 |
), |
| 341 |
array('id' => $existing_id), |
| 342 |
array('%s','%s','%s','%s','%s','%s'), |
| 343 |
array('%d') |
| 344 |
); |
| 345 |
} else { |
| 346 |
//error_log('[MXCHAT-DB] No existing entry found. Inserting new row... (Attempt ' . $attempt . ')'); |
| 347 |
//error_log('[MXCHAT-DB] Content sample: ' . substr($current_content, 0, 1000)); |
| 348 |
|
| 349 |
// Insert a new row - UPDATED 2.5.6: Added content_type |
| 350 |
$result = $wpdb->insert( |
| 351 |
$table_name, |
| 352 |
array( |
| 353 |
'url' => $source_url, // Now unique for manual content |
| 354 |
'article_content' => $current_content, |
| 355 |
'embedding_vector' => $embedding_vector_serialized, |
| 356 |
'source_url' => $source_url, // Now unique for manual content |
| 357 |
'content_type' => $content_type, |
| 358 |
'timestamp' => current_time('mysql'), |
| 359 |
), |
| 360 |
array('%s','%s','%s','%s','%s','%s') |
| 361 |
); |
| 362 |
} |
| 363 |
|
| 364 |
if ($result === false) { |
| 365 |
//error_log('[MXCHAT-DB] Database operation failed (Attempt ' . $attempt . ')'); |
| 366 |
//error_log('[MXCHAT-DB] MySQL Error: ' . $wpdb->last_error); |
| 367 |
//error_log('[MXCHAT-DB] MySQL Error Number: ' . $wpdb->last_errno); |
| 368 |
//error_log('[MXCHAT-DB] Last Query: ' . substr($wpdb->last_query, 0, 500)); |
| 369 |
//error_log('[MXCHAT-DB] Content length: ' . strlen($current_content) . ' bytes'); |
| 370 |
//error_log('[MXCHAT-DB] Embedding vector length: ' . strlen($embedding_vector_serialized) . ' bytes'); |
| 371 |
|
| 372 |
// Progressively apply more aggressive sanitization on failure |
| 373 |
if ($attempt === 1) { |
| 374 |
// First fallback: Use a more aggressive character filter and shorten |
| 375 |
$current_content = preg_replace('/[^\p{L}\p{N}\s.,;:!?()-]/u', '', $current_content); |
| 376 |
$current_content = substr($current_content, 0, 50000); |
| 377 |
} else if ($attempt === 2) { |
| 378 |
// Second fallback: Keep only alphanumeric and basic punctuation, shorten further |
| 379 |
$current_content = preg_replace('/[^a-zA-Z0-9\s.,;:!?()-]/u', '', $current_content); |
| 380 |
$current_content = substr($current_content, 0, 30000); |
| 381 |
} |
| 382 |
|
| 383 |
$attempt++; |
| 384 |
} |
| 385 |
} catch (Exception $e) { |
| 386 |
//error_log('[MXCHAT-DB] Exception during database operation: ' . $e->getMessage()); |
| 387 |
$attempt++; |
| 388 |
} |
| 389 |
} |
| 390 |
|
| 391 |
if ($result === false) { |
| 392 |
//error_log('[MXCHAT-DB] All database operation attempts failed'); |
| 393 |
//error_log('[MXCHAT-DB] Final MySQL Error: ' . $wpdb->last_error); |
| 394 |
|
| 395 |
$detailed_error = sprintf( |
| 396 |
'Failed to store content in WordPress database after %d attempts. MySQL Error: %s (Error #%d). Content size: %d bytes, Embedding size: %d bytes', |
| 397 |
$max_attempts, |
| 398 |
$wpdb->last_error, |
| 399 |
$wpdb->last_errno, |
| 400 |
strlen($current_content), |
| 401 |
strlen($embedding_vector_serialized) |
| 402 |
); |
| 403 |
|
| 404 |
return new WP_Error('database_failed', $detailed_error); |
| 405 |
} |
| 406 |
|
| 407 |
//error_log('[MXCHAT-DB] WordPress database operation completed successfully (Attempt ' . ($attempt - 1) . ')'); |
| 408 |
return true; |
| 409 |
} |
| 410 |
|
| 411 |
/** |
| 412 |
* UPDATED: Store content in Pinecone database with bot support |
| 413 |
* UPDATED 2.5.6: Now accepts content_type parameter |
| 414 |
*/ |
| 415 |
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') { |
| 416 |
//error_log('[MXCHAT-PINECONE-MAIN] ===== Starting Pinecone storage for bot ' . $bot_id . ' ====='); |
| 417 |
|
| 418 |
// ===== UPDATED: Handle manual content with unique vector IDs ===== |
| 419 |
if ($vector_id) { |
| 420 |
// Use provided vector ID |
| 421 |
//error_log('[MXCHAT-PINECONE-MAIN] Using provided vector ID: ' . $vector_id); |
| 422 |
} elseif (!empty($url) && preg_match('#^https?://#i', $url)) { |
| 423 |
// For URLs, use URL-based ID (existing behavior) |
| 424 |
$vector_id = md5($url); |
| 425 |
//error_log('[MXCHAT-PINECONE-MAIN] Generated vector ID from URL: ' . $vector_id); |
| 426 |
} else { |
| 427 |
// For manual content (empty/no URL scheme), generate unique ID |
| 428 |
$vector_id = 'manual_' . time() . '_' . substr(md5($content . microtime(true)), 0, 8); |
| 429 |
//error_log('[MXCHAT-PINECONE-MAIN] Generated unique vector ID for manual content: ' . $vector_id); |
| 430 |
} |
| 431 |
// ===== END UPDATE ===== |
| 432 |
|
| 433 |
// Get host from bot-specific config or fallback to default |
| 434 |
if ($bot_id === 'default' || !class_exists('MxChat_Multi_Bot_Manager')) { |
| 435 |
$options = get_option('mxchat_pinecone_addon_options'); |
| 436 |
$host = $options['mxchat_pinecone_host'] ?? ''; |
| 437 |
} else { |
| 438 |
$bot_pinecone_config = apply_filters('mxchat_get_bot_pinecone_config', array(), $bot_id); |
| 439 |
if (!empty($bot_pinecone_config)) { |
| 440 |
$host = $bot_pinecone_config['host'] ?? ''; |
| 441 |
} else { |
| 442 |
$options = get_option('mxchat_pinecone_addon_options'); |
| 443 |
$host = $options['mxchat_pinecone_host'] ?? ''; |
| 444 |
} |
| 445 |
} |
| 446 |
|
| 447 |
//error_log('[MXCHAT-PINECONE-MAIN] Host: ' . $host); |
| 448 |
//error_log('[MXCHAT-PINECONE-MAIN] API key length: ' . strlen($api_key)); |
| 449 |
//error_log('[MXCHAT-PINECONE-MAIN] Bot ID: ' . $bot_id); |
| 450 |
//error_log('[MXCHAT-PINECONE-MAIN] Namespace: ' . $namespace); |
| 451 |
|
| 452 |
if (empty($host)) { |
| 453 |
//error_log('[MXCHAT-PINECONE-MAIN] ERROR: Host is empty'); |
| 454 |
return new WP_Error('pinecone_config', 'Pinecone host is not configured. Please set the host in your bot settings.'); |
| 455 |
} |
| 456 |
|
| 457 |
// ===== UPDATED 2.5.6: Use passed content_type or determine from URL if not provided ===== |
| 458 |
// Sanitize content_type |
| 459 |
$content_type = sanitize_key($content_type); |
| 460 |
if (empty($content_type)) { |
| 461 |
// Fallback to old detection logic for backwards compatibility |
| 462 |
$is_product = false; |
| 463 |
$content_type = 'manual'; // Default for manual content |
| 464 |
|
| 465 |
if (!empty($url) && preg_match('#^https?://#i', $url)) { |
| 466 |
$is_product = (strpos($url, '/product/') !== false || strpos($url, '/shop/') !== false); |
| 467 |
$content_type = $is_product ? 'product' : 'content'; |
| 468 |
} |
| 469 |
} |
| 470 |
|
| 471 |
//error_log('[MXCHAT-PINECONE-MAIN] Content type: ' . $content_type); |
| 472 |
// ===== END UPDATE ===== |
| 473 |
|
| 474 |
$api_endpoint = "https://{$host}/vectors/upsert"; |
| 475 |
//error_log('[MXCHAT-PINECONE-MAIN] API endpoint: ' . $api_endpoint); |
| 476 |
|
| 477 |
// UPDATED 2.5.6: Use provided content_type in metadata |
| 478 |
$metadata = array( |
| 479 |
'text' => $content, |
| 480 |
'source_url' => $url, // Can be empty for manual content |
| 481 |
'type' => $content_type, // Now supports: post, page, pdf, url, manual, product, etc. |
| 482 |
'last_updated' => time(), |
| 483 |
'created_at' => time(), // Add creation timestamp |
| 484 |
'bot_id' => $bot_id, // Add bot identification |
| 485 |
); |
| 486 |
|
| 487 |
$vector_data = array( |
| 488 |
'id' => $vector_id, |
| 489 |
'values' => $embedding_vector, |
| 490 |
'metadata' => $metadata |
| 491 |
); |
| 492 |
|
| 493 |
$request_body = array( |
| 494 |
'vectors' => array($vector_data) |
| 495 |
); |
| 496 |
|
| 497 |
// Add namespace if specified for multi-bot separation |
| 498 |
if (!empty($namespace)) { |
| 499 |
$request_body['namespace'] = $namespace; |
| 500 |
//error_log('[MXCHAT-PINECONE-MAIN] Using namespace: ' . $namespace); |
| 501 |
} |
| 502 |
|
| 503 |
//error_log('[MXCHAT-PINECONE-MAIN] Request body prepared (embedding dimensions: ' . count($embedding_vector) . ')'); |
| 504 |
|
| 505 |
$response = wp_remote_post($api_endpoint, array( |
| 506 |
'headers' => array( |
| 507 |
'Api-Key' => $api_key, |
| 508 |
'accept' => 'application/json', |
| 509 |
'content-type' => 'application/json' |
| 510 |
), |
| 511 |
'body' => wp_json_encode($request_body), |
| 512 |
'timeout' => 30, |
| 513 |
'data_format' => 'body' |
| 514 |
)); |
| 515 |
|
| 516 |
if (is_wp_error($response)) { |
| 517 |
//error_log('[MXCHAT-PINECONE-MAIN] WordPress request error: ' . $response->get_error_message()); |
| 518 |
return new WP_Error('pinecone_request', $response->get_error_message()); |
| 519 |
} |
| 520 |
|
| 521 |
$response_code = wp_remote_retrieve_response_code($response); |
| 522 |
//error_log('[MXCHAT-PINECONE-MAIN] Response code: ' . $response_code); |
| 523 |
|
| 524 |
if ($response_code !== 200) { |
| 525 |
$body = wp_remote_retrieve_body($response); |
| 526 |
//error_log('[MXCHAT-PINECONE-MAIN] API error - Response body: ' . $body); |
| 527 |
return new WP_Error('pinecone_api', sprintf( |
| 528 |
'Pinecone API error (HTTP %d): %s', |
| 529 |
$response_code, |
| 530 |
$body |
| 531 |
)); |
| 532 |
} |
| 533 |
|
| 534 |
$response_body = wp_remote_retrieve_body($response); |
| 535 |
//error_log('[MXCHAT-PINECONE-MAIN] Success response: ' . $response_body); |
| 536 |
//error_log('[MXCHAT-PINECONE-MAIN] Successfully stored in Pinecone for bot ' . $bot_id); |
| 537 |
//error_log('[MXCHAT-PINECONE-MAIN] ===== Pinecone storage complete ====='); |
| 538 |
|
| 539 |
return true; |
| 540 |
} |
| 541 |
|
| 542 |
/** |
| 543 |
* UPDATED: Generate an embedding for the given text using bot-specific configuration. |
| 544 |
* |
| 545 |
* @param string $text The text to be embedded. |
| 546 |
* @param string $api_key The API key used for generating embeddings. |
| 547 |
* @param string $bot_id The bot ID for multi-bot support |
| 548 |
* @return array|null The embedding vector or null on failure. |
| 549 |
*/ |
| 550 |
private static function generate_embedding($text, $api_key, $bot_id = 'default') { |
| 551 |
// Get bot-specific options |
| 552 |
if ($bot_id === 'default' || !class_exists('MxChat_Multi_Bot_Manager')) { |
| 553 |
$options = get_option('mxchat_options'); |
| 554 |
} else { |
| 555 |
$bot_options = apply_filters('mxchat_get_bot_options', array(), $bot_id); |
| 556 |
$options = !empty($bot_options) ? $bot_options : get_option('mxchat_options'); |
| 557 |
} |
| 558 |
|
| 559 |
// Opt-in: when the custom provider is selected for embeddings, route the KB |
| 560 |
// INDEX side through the same custom endpoint the query side uses, so stored |
| 561 |
// vectors and query vectors come from the same model. Default-off behavior |
| 562 |
// below is untouched. |
| 563 |
if (isset($options['custom_provider_for_embeddings']) && $options['custom_provider_for_embeddings'] === 'on') { |
| 564 |
$custom = self::generate_embedding_custom($text, $options); |
| 565 |
// The custom path already returns a human-readable error string — |
| 566 |
// carry it instead of collapsing to null (plan 4a7c0a). |
| 567 |
return is_array($custom) ? $custom : new WP_Error('embedding_failed', (string) $custom); |
| 568 |
} |
| 569 |
|
| 570 |
$selected_model = $options['embedding_model'] ?? 'text-embedding-ada-002'; |
| 571 |
|
| 572 |
// Determine endpoint and API key based on model |
| 573 |
if (strpos($selected_model, 'voyage') === 0) { |
| 574 |
$endpoint = 'https://api.voyageai.com/v1/embeddings'; |
| 575 |
$api_key = $options['voyage_api_key'] ?? ''; |
| 576 |
} elseif (strpos($selected_model, 'gemini-embedding') === 0) { |
| 577 |
$endpoint = 'https://generativelanguage.googleapis.com/v1beta/models/' . $selected_model . ':embedContent'; |
| 578 |
$api_key = $options['gemini_api_key'] ?? ''; |
| 579 |
} else { |
| 580 |
$endpoint = 'https://api.openai.com/v1/embeddings'; |
| 581 |
// Use the bot-specific API key or fallback to passed API key |
| 582 |
$api_key = $options['api_key'] ?? $api_key; |
| 583 |
} |
| 584 |
|
| 585 |
// Prepare request body based on provider |
| 586 |
if (strpos($selected_model, 'gemini-embedding') === 0) { |
| 587 |
// Gemini API format |
| 588 |
$request_body = [ |
| 589 |
'model' => 'models/' . $selected_model, |
| 590 |
'content' => [ |
| 591 |
'parts' => [ |
| 592 |
['text' => $text] |
| 593 |
] |
| 594 |
], |
| 595 |
'outputDimensionality' => 1536 |
| 596 |
]; |
| 597 |
|
| 598 |
// Prepare headers for Gemini (API key as query parameter) |
| 599 |
$endpoint .= '?key=' . $api_key; |
| 600 |
$headers = [ |
| 601 |
'Content-Type' => 'application/json' |
| 602 |
]; |
| 603 |
} else { |
| 604 |
// OpenAI/Voyage API format |
| 605 |
$request_body = [ |
| 606 |
'input' => $text, |
| 607 |
'model' => $selected_model |
| 608 |
]; |
| 609 |
|
| 610 |
// Add output_dimension for voyage-3-large |
| 611 |
if ($selected_model === 'voyage-3-large') { |
| 612 |
$request_body['output_dimension'] = 2048; |
| 613 |
} |
| 614 |
|
| 615 |
// Prepare headers for OpenAI/Voyage |
| 616 |
$headers = [ |
| 617 |
'Content-Type' => 'application/json', |
| 618 |
'Authorization' => 'Bearer ' . $api_key |
| 619 |
]; |
| 620 |
} |
| 621 |
|
| 622 |
$args = [ |
| 623 |
'body' => wp_json_encode($request_body), |
| 624 |
'headers' => $headers, |
| 625 |
'timeout' => 60, |
| 626 |
'redirection' => 5, |
| 627 |
'blocking' => true, |
| 628 |
'httpversion' => '1.0', |
| 629 |
'sslverify' => true, |
| 630 |
]; |
| 631 |
|
| 632 |
$response = wp_remote_post($endpoint, $args); |
| 633 |
|
| 634 |
if (is_wp_error($response)) { |
| 635 |
$message = 'Embedding request failed (connection): ' . $response->get_error_message(); |
| 636 |
if (class_exists('MxChat_Admin')) { |
| 637 |
MxChat_Admin::mxchat_log_debug('embedding_error', $message, array('model' => $selected_model, 'bot_id' => $bot_id)); |
| 638 |
} |
| 639 |
return new WP_Error('embedding_failed', $message); |
| 640 |
} |
| 641 |
|
| 642 |
$response_body = json_decode(wp_remote_retrieve_body($response), true); |
| 643 |
|
| 644 |
// Handle different response formats based on provider |
| 645 |
if (strpos($selected_model, 'gemini-embedding') === 0) { |
| 646 |
// Gemini API response format |
| 647 |
if (isset($response_body['embedding']['values']) && is_array($response_body['embedding']['values'])) { |
| 648 |
self::stamp_active_embedding_model($selected_model); |
| 649 |
return $response_body['embedding']['values']; |
| 650 |
} else { |
| 651 |
return self::embedding_failure_error($response, $selected_model, $api_key, $bot_id); |
| 652 |
} |
| 653 |
} else { |
| 654 |
// OpenAI/Voyage API response format |
| 655 |
if (isset($response_body['data'][0]['embedding']) && is_array($response_body['data'][0]['embedding'])) { |
| 656 |
self::stamp_active_embedding_model($selected_model); |
| 657 |
return $response_body['data'][0]['embedding']; |
| 658 |
} else { |
| 659 |
return self::embedding_failure_error($response, $selected_model, $api_key, $bot_id); |
| 660 |
} |
| 661 |
} |
| 662 |
} |
| 663 |
|
| 664 |
/** |
| 665 |
* Build a WP_Error carrying the embedding provider's REAL failure reason, |
| 666 |
* and record it in the Debug Mode log. Previously every failure path |
| 667 |
* returned bare null, so customers saw only "Failed to generate embedding |
| 668 |
* for content" / "Failed to store any chunks" with no cause (plan 4a7c0a). |
| 669 |
* |
| 670 |
* The API key never appears in provider response bodies (it travels in the |
| 671 |
* request headers), but the reason is scrubbed for it anyway before it can |
| 672 |
* reach a notice or the debug log. |
| 673 |
*/ |
| 674 |
private static function embedding_failure_error($response, $selected_model, $api_key, $bot_id) { |
| 675 |
$status = (int) wp_remote_retrieve_response_code($response); |
| 676 |
$raw = (string) wp_remote_retrieve_body($response); |
| 677 |
$decoded = json_decode($raw, true); |
| 678 |
|
| 679 |
// Provider error shapes: OpenAI + Gemini use {"error":{"message":…}}; |
| 680 |
// Voyage uses {"detail":…}. |
| 681 |
$reason = ''; |
| 682 |
if (is_array($decoded)) { |
| 683 |
if (isset($decoded['error']['message']) && is_string($decoded['error']['message'])) { |
| 684 |
$reason = $decoded['error']['message']; |
| 685 |
} elseif (isset($decoded['detail']) && is_string($decoded['detail'])) { |
| 686 |
$reason = $decoded['detail']; |
| 687 |
} |
| 688 |
} |
| 689 |
if ($reason === '') { |
| 690 |
$reason = ($raw !== '') ? substr($raw, 0, 200) : 'empty or malformed response'; |
| 691 |
} |
| 692 |
if (is_string($api_key) && $api_key !== '') { |
| 693 |
$reason = str_replace($api_key, '[redacted]', $reason); |
| 694 |
} |
| 695 |
$reason = substr($reason, 0, 300); |
| 696 |
$message = sprintf('Embedding failed (%s, HTTP %d): %s', $selected_model, $status, $reason); |
| 697 |
|
| 698 |
if (class_exists('MxChat_Admin')) { |
| 699 |
MxChat_Admin::mxchat_log_debug('embedding_error', $message, array( |
| 700 |
'model' => $selected_model, |
| 701 |
'status' => $status, |
| 702 |
'bot_id' => $bot_id, |
| 703 |
)); |
| 704 |
} |
| 705 |
|
| 706 |
return new WP_Error('embedding_failed', $message); |
| 707 |
} |
| 708 |
|
| 709 |
/** |
| 710 |
* Generate an embedding via a Custom (OpenAI-compatible) provider's /embeddings route. |
| 711 |
* Shared by every embedding entry point so the KNOWLEDGE-BASE INDEX side and the |
| 712 |
* QUERY side route through the same model when the opt-in |
| 713 |
* 'custom_provider_for_embeddings' setting is on. Mirrors the query-path logic in |
| 714 |
* MxChat_Integrator::mxchat_generate_embedding_custom() but takes an explicit |
| 715 |
* $options array so it is callable statically from utils + knowledge-manager. |
| 716 |
* |
| 717 |
* Returns a numeric array (the embedding vector) on success, or a human-readable |
| 718 |
* error string on failure (so callers expecting a string error, like the |
| 719 |
* knowledge-manager, can surface it directly; callers expecting array|null wrap it). |
| 720 |
* |
| 721 |
* @param string $text Text to embed. |
| 722 |
* @param array $options The resolved mxchat options (must contain the custom_provider_* keys). |
| 723 |
* @return array|string Embedding vector on success; error string on failure. |
| 724 |
*/ |
| 725 |
public static function generate_embedding_custom($text, $options) { |
| 726 |
if (empty($text)) { |
| 727 |
return 'No text provided for embedding generation'; |
| 728 |
} |
| 729 |
|
| 730 |
$base_url = isset($options['custom_provider_base_url']) ? rtrim(trim((string) $options['custom_provider_base_url']), '/') : ''; |
| 731 |
if (empty($base_url)) { |
| 732 |
return 'Custom provider Base URL is not configured.'; |
| 733 |
} |
| 734 |
|
| 735 |
$api_key = isset($options['custom_provider_api_key']) ? trim((string) $options['custom_provider_api_key']) : ''; |
| 736 |
$auth_scheme = isset($options['custom_provider_auth_scheme']) ? $options['custom_provider_auth_scheme'] : 'bearer'; |
| 737 |
$api_version = isset($options['custom_provider_api_version']) ? trim((string) $options['custom_provider_api_version']) : ''; |
| 738 |
|
| 739 |
// Embedding model: prefer the dedicated custom_provider_embedding_model, fall back to the chat model. |
| 740 |
$model = (isset($options['custom_provider_embedding_model']) && trim((string) $options['custom_provider_embedding_model']) !== '') |
| 741 |
? trim((string) $options['custom_provider_embedding_model']) |
| 742 |
: ((isset($options['custom_provider_model']) && trim((string) $options['custom_provider_model']) !== '') ? trim((string) $options['custom_provider_model']) : 'default'); |
| 743 |
|
| 744 |
$embed_url = $base_url . '/embeddings'; |
| 745 |
if (!empty($api_version)) { |
| 746 |
$embed_url .= (strpos($embed_url, '?') === false ? '?' : '&') . 'api-version=' . rawurlencode($api_version); |
| 747 |
} |
| 748 |
|
| 749 |
$headers = ['Content-Type' => 'application/json']; |
| 750 |
if (!empty($api_key)) { |
| 751 |
if ($auth_scheme === 'api-key') { |
| 752 |
$headers['api-key'] = $api_key; |
| 753 |
} else { |
| 754 |
$headers['Authorization'] = 'Bearer ' . $api_key; |
| 755 |
} |
| 756 |
} |
| 757 |
|
| 758 |
$response = wp_remote_post($embed_url, [ |
| 759 |
'headers' => $headers, |
| 760 |
'body' => wp_json_encode(['input' => $text, 'model' => $model]), |
| 761 |
'timeout' => 60, |
| 762 |
]); |
| 763 |
if (is_wp_error($response)) { |
| 764 |
return 'Connection error when generating embeddings (custom provider): ' . $response->get_error_message(); |
| 765 |
} |
| 766 |
|
| 767 |
$status = wp_remote_retrieve_response_code($response); |
| 768 |
$body = json_decode(wp_remote_retrieve_body($response), true); |
| 769 |
if ($status !== 200) { |
| 770 |
$msg = isset($body['error']['message']) ? $body['error']['message'] : 'HTTP ' . $status; |
| 771 |
return 'Custom embedding endpoint error: ' . $msg; |
| 772 |
} |
| 773 |
if (isset($body['data'][0]['embedding']) && is_array($body['data'][0]['embedding'])) { |
| 774 |
// Stamp the custom model identity so the active-embedding-model mismatch |
| 775 |
// warning reflects the real (custom) model rather than the built-in setting. |
| 776 |
self::stamp_active_embedding_model('custom:' . $model); |
| 777 |
return $body['data'][0]['embedding']; |
| 778 |
} |
| 779 |
return 'Invalid embedding response from custom provider.'; |
| 780 |
} |
| 781 |
|
| 782 |
/** |
| 783 |
* Submit content as multiple chunks |
| 784 |
* |
| 785 |
* Splits large content into chunks, generates embeddings for each, |
| 786 |
* and stores them with chunk metadata for later reassembly. |
| 787 |
* |
| 788 |
* @param string $content The content to chunk and store |
| 789 |
* @param string $source_url The source URL |
| 790 |
* @param string $api_key The API key for embeddings |
| 791 |
* @param string $bot_id The bot ID |
| 792 |
* @param string $content_type The content type |
| 793 |
* @param MxChat_Chunker $chunker The chunker instance |
| 794 |
* @return bool|WP_Error True on success, WP_Error on failure |
| 795 |
*/ |
| 796 |
private static function submit_chunked_content($content, $source_url, $api_key, $bot_id, $content_type, $chunker) { |
| 797 |
global $wpdb; |
| 798 |
$table_name = $wpdb->prefix . 'mxchat_system_prompt_content'; |
| 799 |
|
| 800 |
//error_log('[MXCHAT-CHUNK-DEBUG] Starting chunked submission for: ' . $source_url); |
| 801 |
//error_log('[MXCHAT-CHUNK-DEBUG] Content length: ' . strlen($content) . ' chars'); |
| 802 |
|
| 803 |
// First, delete any existing chunks for this URL (clean slate) |
| 804 |
$delete_result = self::delete_chunks_for_url($source_url, $bot_id); |
| 805 |
if (is_wp_error($delete_result)) { |
| 806 |
//error_log('[MXCHAT-CHUNK-DEBUG] Warning: Failed to delete existing chunks: ' . $delete_result->get_error_message()); |
| 807 |
// Continue anyway - we'll overwrite with upsert |
| 808 |
} |
| 809 |
|
| 810 |
// Split content into chunks |
| 811 |
$chunks = $chunker->chunk_text($content); |
| 812 |
$total_chunks = count($chunks); |
| 813 |
|
| 814 |
//error_log('[MXCHAT-CHUNK-DEBUG] Created ' . $total_chunks . ' chunks'); |
| 815 |
foreach ($chunks as $i => $chunk) { |
| 816 |
//error_log('[MXCHAT-CHUNK-DEBUG] Chunk ' . $i . ' length: ' . strlen($chunk) . ' chars, preview: ' . substr($chunk, 0, 100)); |
| 817 |
} |
| 818 |
|
| 819 |
//error_log('[MXCHAT-CHUNK] Split content into ' . $total_chunks . ' chunks'); |
| 820 |
|
| 821 |
if ($total_chunks === 0) { |
| 822 |
return new WP_Error('chunking_failed', 'Content could not be split into chunks'); |
| 823 |
} |
| 824 |
|
| 825 |
$errors = array(); |
| 826 |
$embed_failures = 0; |
| 827 |
$first_embed_reason = ''; |
| 828 |
$first_store_reason = ''; |
| 829 |
$is_pinecone = self::is_pinecone_enabled_for_bot($bot_id); |
| 830 |
|
| 831 |
foreach ($chunks as $index => $chunk_text) { |
| 832 |
// Generate chunk metadata |
| 833 |
$chunk_metadata = MxChat_Chunker::create_chunk_metadata($index, $total_chunks, $source_url); |
| 834 |
|
| 835 |
// AI-Engine-style aliases so external consumers (Pinecone/Qdrant/Chroma) can rely on |
| 836 |
// a stable shorthand ('source'/'part_index'/'part_total') without parsing our internal names. |
| 837 |
$chunk_metadata['source'] = $source_url; |
| 838 |
$chunk_metadata['part_index'] = (int) $index; |
| 839 |
$chunk_metadata['part_total'] = (int) $total_chunks; |
| 840 |
|
| 841 |
/** |
| 842 |
* Filter the per-chunk metadata blob before it's written to the KB store. |
| 843 |
* |
| 844 |
* @param array $chunk_metadata Metadata array (source, part_index, part_total, chunk_index, total_chunks, source_url, parent_url_hash, document_type, ...). |
| 845 |
* @param string $chunk_text The chunk text being stored. |
| 846 |
* @param array $context ['bot_id' => string, 'content_type' => string, 'source_url' => string, 'part_index' => int, 'part_total' => int] |
| 847 |
* @return array Updated metadata array. |
| 848 |
*/ |
| 849 |
$chunk_metadata = apply_filters( |
| 850 |
'mxchat_embedding_chunk_metadata', |
| 851 |
$chunk_metadata, |
| 852 |
$chunk_text, |
| 853 |
array( |
| 854 |
'bot_id' => $bot_id, |
| 855 |
'content_type' => $content_type, |
| 856 |
'source_url' => $source_url, |
| 857 |
'part_index' => (int) $index, |
| 858 |
'part_total' => (int) $total_chunks, |
| 859 |
) |
| 860 |
); |
| 861 |
|
| 862 |
$chunk_vector_id = MxChat_Chunker::generate_chunk_vector_id($source_url, $index); |
| 863 |
|
| 864 |
//error_log('[MXCHAT-CHUNK] Processing chunk ' . ($index + 1) . '/' . $total_chunks . ' (ID: ' . $chunk_vector_id . ')'); |
| 865 |
|
| 866 |
// Generate embedding for this chunk |
| 867 |
$embedding_vector = self::generate_embedding($chunk_text, $api_key, $bot_id); |
| 868 |
|
| 869 |
if (!is_array($embedding_vector)) { |
| 870 |
// Track embedding failures separately from storage failures, and |
| 871 |
// keep the first provider reason seen — the two failure classes |
| 872 |
// have opposite remedies (API key vs Pinecone/DB) (plan 4a7c0a). |
| 873 |
$embed_failures++; |
| 874 |
$reason = is_wp_error($embedding_vector) ? $embedding_vector->get_error_message() : ''; |
| 875 |
if ($reason !== '' && $first_embed_reason === '') { |
| 876 |
$first_embed_reason = $reason; |
| 877 |
} |
| 878 |
$errors[] = new WP_Error('embedding_failed', 'Failed to generate embedding for chunk ' . $index . ($reason !== '' ? ' — ' . $reason : '')); |
| 879 |
continue; |
| 880 |
} |
| 881 |
|
| 882 |
if ($is_pinecone) { |
| 883 |
// Store in Pinecone with chunk metadata |
| 884 |
$result = self::store_chunk_in_pinecone( |
| 885 |
$embedding_vector, |
| 886 |
$chunk_text, |
| 887 |
$source_url, |
| 888 |
$chunk_vector_id, |
| 889 |
$bot_id, |
| 890 |
$content_type, |
| 891 |
$chunk_metadata |
| 892 |
); |
| 893 |
} else { |
| 894 |
// Store in WordPress DB with chunk metadata |
| 895 |
$content_with_metadata = MxChat_Chunker::format_chunk_for_storage($chunk_text, $chunk_metadata); |
| 896 |
$embedding_vector_serialized = maybe_serialize($embedding_vector); |
| 897 |
|
| 898 |
$result = self::store_chunk_in_wordpress_db( |
| 899 |
$content_with_metadata, |
| 900 |
$source_url, |
| 901 |
$embedding_vector_serialized, |
| 902 |
$table_name, |
| 903 |
$content_type, |
| 904 |
$chunk_metadata |
| 905 |
); |
| 906 |
} |
| 907 |
|
| 908 |
if (is_wp_error($result)) { |
| 909 |
$errors[] = $result; |
| 910 |
if ($first_store_reason === '') { |
| 911 |
$first_store_reason = $result->get_error_message(); |
| 912 |
} |
| 913 |
} |
| 914 |
} |
| 915 |
|
| 916 |
if (count($errors) === $total_chunks) { |
| 917 |
// Say WHICH stage failed — "failed to store" used to cover pure |
| 918 |
// embedding failures too, sending customers to debug Pinecone when |
| 919 |
// the problem was their embedding API key (plan 4a7c0a). |
| 920 |
if ($embed_failures === $total_chunks) { |
| 921 |
return new WP_Error('chunking_failed', |
| 922 |
'Failed to store any chunks — every chunk failed to embed' |
| 923 |
. ($first_embed_reason !== '' ? ': ' . $first_embed_reason : '') |
| 924 |
. ' Check the embedding provider API key and model under MxChat Settings.'); |
| 925 |
} |
| 926 |
if ($embed_failures === 0) { |
| 927 |
return new WP_Error('chunking_failed', |
| 928 |
'Failed to store any chunks — embeddings generated but storage failed' |
| 929 |
. ($first_store_reason !== '' ? ': ' . $first_store_reason : '') |
| 930 |
. ' Check the knowledge base storage (Pinecone index or database).'); |
| 931 |
} |
| 932 |
return new WP_Error('chunking_failed', sprintf( |
| 933 |
'Failed to store any chunks — %d failed to embed%s and %d failed to store%s', |
| 934 |
$embed_failures, |
| 935 |
$first_embed_reason !== '' ? ' (' . $first_embed_reason . ')' : '', |
| 936 |
$total_chunks - $embed_failures, |
| 937 |
$first_store_reason !== '' ? ' (' . $first_store_reason . ')' : '' |
| 938 |
)); |
| 939 |
} |
| 940 |
|
| 941 |
if (!empty($errors)) { |
| 942 |
$detail = $first_embed_reason !== '' ? $first_embed_reason : $first_store_reason; |
| 943 |
return new WP_Error('chunking_partial_failure', |
| 944 |
sprintf('Failed to store %d of %d chunks', count($errors), $total_chunks) |
| 945 |
. ($detail !== '' ? ' — first error: ' . $detail : '')); |
| 946 |
} |
| 947 |
|
| 948 |
//error_log('[MXCHAT-CHUNK] Successfully stored all ' . $total_chunks . ' chunks'); |
| 949 |
return true; |
| 950 |
} |
| 951 |
|
| 952 |
/** |
| 953 |
* Store a single chunk in Pinecone with chunk-specific metadata |
| 954 |
*/ |
| 955 |
private static function store_chunk_in_pinecone($embedding_vector, $chunk_text, $source_url, $vector_id, $bot_id, $content_type, $chunk_metadata) { |
| 956 |
// Get Pinecone configuration |
| 957 |
if ($bot_id === 'default' || !class_exists('MxChat_Multi_Bot_Manager')) { |
| 958 |
$pinecone_options = get_option('mxchat_pinecone_addon_options'); |
| 959 |
$api_key = $pinecone_options['mxchat_pinecone_api_key'] ?? ''; |
| 960 |
$host = $pinecone_options['mxchat_pinecone_host'] ?? ''; |
| 961 |
$namespace = $pinecone_options['mxchat_pinecone_namespace'] ?? ''; |
| 962 |
} else { |
| 963 |
$bot_pinecone_config = apply_filters('mxchat_get_bot_pinecone_config', array(), $bot_id); |
| 964 |
if (empty($bot_pinecone_config)) { |
| 965 |
$pinecone_options = get_option('mxchat_pinecone_addon_options'); |
| 966 |
$api_key = $pinecone_options['mxchat_pinecone_api_key'] ?? ''; |
| 967 |
$host = $pinecone_options['mxchat_pinecone_host'] ?? ''; |
| 968 |
$namespace = $pinecone_options['mxchat_pinecone_namespace'] ?? ''; |
| 969 |
} else { |
| 970 |
$api_key = $bot_pinecone_config['api_key'] ?? ''; |
| 971 |
$host = $bot_pinecone_config['host'] ?? ''; |
| 972 |
$namespace = $bot_pinecone_config['namespace'] ?? ''; |
| 973 |
} |
| 974 |
} |
| 975 |
|
| 976 |
if (empty($host) || empty($api_key)) { |
| 977 |
return new WP_Error('pinecone_config', 'Pinecone is not properly configured'); |
| 978 |
} |
| 979 |
|
| 980 |
$api_endpoint = "https://{$host}/vectors/upsert"; |
| 981 |
|
| 982 |
// Build metadata with chunk information |
| 983 |
$metadata = array( |
| 984 |
'text' => $chunk_text, |
| 985 |
'source_url' => $source_url, |
| 986 |
'type' => $content_type, |
| 987 |
'is_chunked' => true, |
| 988 |
'chunk_index' => $chunk_metadata['chunk_index'], |
| 989 |
'total_chunks' => $chunk_metadata['total_chunks'], |
| 990 |
'parent_url_hash' => $chunk_metadata['parent_url_hash'], |
| 991 |
'last_updated' => time(), |
| 992 |
'created_at' => time(), |
| 993 |
'bot_id' => $bot_id, |
| 994 |
); |
| 995 |
|
| 996 |
$vector_data = array( |
| 997 |
'id' => $vector_id, |
| 998 |
'values' => $embedding_vector, |
| 999 |
'metadata' => $metadata |
| 1000 |
); |
| 1001 |
|
| 1002 |
$request_body = array( |
| 1003 |
'vectors' => array($vector_data) |
| 1004 |
); |
| 1005 |
|
| 1006 |
if (!empty($namespace)) { |
| 1007 |
$request_body['namespace'] = $namespace; |
| 1008 |
} |
| 1009 |
|
| 1010 |
$response = wp_remote_post($api_endpoint, array( |
| 1011 |
'headers' => array( |
| 1012 |
'Api-Key' => $api_key, |
| 1013 |
'accept' => 'application/json', |
| 1014 |
'content-type' => 'application/json' |
| 1015 |
), |
| 1016 |
'body' => wp_json_encode($request_body), |
| 1017 |
'timeout' => 30 |
| 1018 |
)); |
| 1019 |
|
| 1020 |
if (is_wp_error($response)) { |
| 1021 |
return $response; |
| 1022 |
} |
| 1023 |
|
| 1024 |
$response_code = wp_remote_retrieve_response_code($response); |
| 1025 |
if ($response_code !== 200) { |
| 1026 |
return new WP_Error('pinecone_api', 'Pinecone API error: HTTP ' . $response_code); |
| 1027 |
} |
| 1028 |
|
| 1029 |
return true; |
| 1030 |
} |
| 1031 |
|
| 1032 |
/** |
| 1033 |
* Store a single chunk in WordPress database |
| 1034 |
*/ |
| 1035 |
private static function store_chunk_in_wordpress_db($content_with_metadata, $source_url, $embedding_vector_serialized, $table_name, $content_type, $chunk_metadata) { |
| 1036 |
global $wpdb; |
| 1037 |
|
| 1038 |
// For chunks, we always insert new rows (no duplicate checking) |
| 1039 |
// The URL includes chunk info in the metadata, but source_url stays the same for grouping |
| 1040 |
$result = $wpdb->insert( |
| 1041 |
$table_name, |
| 1042 |
array( |
| 1043 |
'url' => $source_url, |
| 1044 |
'article_content' => $content_with_metadata, |
| 1045 |
'embedding_vector' => $embedding_vector_serialized, |
| 1046 |
'source_url' => $source_url, |
| 1047 |
'content_type' => $content_type, |
| 1048 |
'timestamp' => current_time('mysql') |
| 1049 |
), |
| 1050 |
array('%s', '%s', '%s', '%s', '%s', '%s') |
| 1051 |
); |
| 1052 |
|
| 1053 |
if ($result === false) { |
| 1054 |
return new WP_Error('database_failed', 'Failed to insert chunk: ' . $wpdb->last_error); |
| 1055 |
} |
| 1056 |
|
| 1057 |
return true; |
| 1058 |
} |
| 1059 |
|
| 1060 |
/** |
| 1061 |
* Delete all chunks for a given URL |
| 1062 |
* |
| 1063 |
* @param string $source_url The source URL |
| 1064 |
* @param string $bot_id The bot ID |
| 1065 |
* @return bool|WP_Error True on success, WP_Error on failure |
| 1066 |
*/ |
| 1067 |
public static function delete_chunks_for_url($source_url, $bot_id = 'default') { |
| 1068 |
//error_log('[MXCHAT-CHUNK-DELETE] Deleting chunks for URL: ' . $source_url); |
| 1069 |
|
| 1070 |
if (self::is_pinecone_enabled_for_bot($bot_id)) { |
| 1071 |
return self::delete_pinecone_chunks_by_url($source_url, $bot_id); |
| 1072 |
} else { |
| 1073 |
return self::delete_wordpress_chunks_by_url($source_url); |
| 1074 |
} |
| 1075 |
} |
| 1076 |
|
| 1077 |
/** |
| 1078 |
* Delete all chunks for a URL from Pinecone |
| 1079 |
*/ |
| 1080 |
private static function delete_pinecone_chunks_by_url($source_url, $bot_id) { |
| 1081 |
// Get Pinecone configuration |
| 1082 |
if ($bot_id === 'default' || !class_exists('MxChat_Multi_Bot_Manager')) { |
| 1083 |
$pinecone_options = get_option('mxchat_pinecone_addon_options'); |
| 1084 |
$api_key = $pinecone_options['mxchat_pinecone_api_key'] ?? ''; |
| 1085 |
$host = $pinecone_options['mxchat_pinecone_host'] ?? ''; |
| 1086 |
$namespace = $pinecone_options['mxchat_pinecone_namespace'] ?? ''; |
| 1087 |
} else { |
| 1088 |
$bot_pinecone_config = apply_filters('mxchat_get_bot_pinecone_config', array(), $bot_id); |
| 1089 |
if (empty($bot_pinecone_config)) { |
| 1090 |
$pinecone_options = get_option('mxchat_pinecone_addon_options'); |
| 1091 |
$api_key = $pinecone_options['mxchat_pinecone_api_key'] ?? ''; |
| 1092 |
$host = $pinecone_options['mxchat_pinecone_host'] ?? ''; |
| 1093 |
$namespace = $pinecone_options['mxchat_pinecone_namespace'] ?? ''; |
| 1094 |
} else { |
| 1095 |
$api_key = $bot_pinecone_config['api_key'] ?? ''; |
| 1096 |
$host = $bot_pinecone_config['host'] ?? ''; |
| 1097 |
$namespace = $bot_pinecone_config['namespace'] ?? ''; |
| 1098 |
} |
| 1099 |
} |
| 1100 |
|
| 1101 |
if (empty($host) || empty($api_key)) { |
| 1102 |
return new WP_Error('pinecone_config', 'Pinecone is not properly configured'); |
| 1103 |
} |
| 1104 |
|
| 1105 |
$base_vector_id = md5($source_url); |
| 1106 |
$vectors_to_delete = array(); |
| 1107 |
|
| 1108 |
// Add the original single-vector ID (for non-chunked content) |
| 1109 |
$vectors_to_delete[] = $base_vector_id; |
| 1110 |
|
| 1111 |
// Pinecone /vectors/list is a GET endpoint with query-string parameters; a POST here returns a |
| 1112 |
// non-200 silently and we end up only deleting the base vector, leaving chunks orphaned. |
| 1113 |
$query_params = array( |
| 1114 |
'prefix' => $base_vector_id . '_chunk_', |
| 1115 |
'limit' => 100, |
| 1116 |
); |
| 1117 |
if (!empty($namespace)) { |
| 1118 |
$query_params['namespace'] = $namespace; |
| 1119 |
} |
| 1120 |
|
| 1121 |
$list_url = "https://{$host}/vectors/list?" . http_build_query($query_params); |
| 1122 |
|
| 1123 |
// Paginate in case a URL has more than 100 chunks. |
| 1124 |
do { |
| 1125 |
$list_response = wp_remote_get($list_url, array( |
| 1126 |
'headers' => array( |
| 1127 |
'Api-Key' => $api_key, |
| 1128 |
'accept' => 'application/json', |
| 1129 |
), |
| 1130 |
'timeout' => 30, |
| 1131 |
)); |
| 1132 |
|
| 1133 |
if (is_wp_error($list_response) || wp_remote_retrieve_response_code($list_response) !== 200) { |
| 1134 |
break; |
| 1135 |
} |
| 1136 |
|
| 1137 |
$list_data = json_decode(wp_remote_retrieve_body($list_response), true); |
| 1138 |
if (!empty($list_data['vectors'])) { |
| 1139 |
foreach ($list_data['vectors'] as $vector) { |
| 1140 |
if (isset($vector['id'])) { |
| 1141 |
$vectors_to_delete[] = $vector['id']; |
| 1142 |
} |
| 1143 |
} |
| 1144 |
} |
| 1145 |
|
| 1146 |
$next_token = $list_data['pagination']['next'] ?? ''; |
| 1147 |
if (empty($next_token)) { |
| 1148 |
break; |
| 1149 |
} |
| 1150 |
|
| 1151 |
$query_params['paginationToken'] = $next_token; |
| 1152 |
$list_url = "https://{$host}/vectors/list?" . http_build_query($query_params); |
| 1153 |
} while (true); |
| 1154 |
|
| 1155 |
if (empty($vectors_to_delete)) { |
| 1156 |
//error_log('[MXCHAT-CHUNK-DELETE] No vectors found to delete'); |
| 1157 |
return true; |
| 1158 |
} |
| 1159 |
|
| 1160 |
//error_log('[MXCHAT-CHUNK-DELETE] Deleting ' . count($vectors_to_delete) . ' vectors from Pinecone'); |
| 1161 |
|
| 1162 |
// Delete vectors |
| 1163 |
$delete_url = "https://{$host}/vectors/delete"; |
| 1164 |
|
| 1165 |
$delete_body = array( |
| 1166 |
'ids' => $vectors_to_delete |
| 1167 |
); |
| 1168 |
|
| 1169 |
if (!empty($namespace)) { |
| 1170 |
$delete_body['namespace'] = $namespace; |
| 1171 |
} |
| 1172 |
|
| 1173 |
$delete_response = wp_remote_post($delete_url, array( |
| 1174 |
'headers' => array( |
| 1175 |
'Api-Key' => $api_key, |
| 1176 |
'accept' => 'application/json', |
| 1177 |
'content-type' => 'application/json' |
| 1178 |
), |
| 1179 |
'body' => wp_json_encode($delete_body), |
| 1180 |
'timeout' => 30 |
| 1181 |
)); |
| 1182 |
|
| 1183 |
if (is_wp_error($delete_response)) { |
| 1184 |
return $delete_response; |
| 1185 |
} |
| 1186 |
|
| 1187 |
$response_code = wp_remote_retrieve_response_code($delete_response); |
| 1188 |
if ($response_code !== 200) { |
| 1189 |
return new WP_Error('pinecone_delete', 'Failed to delete vectors: HTTP ' . $response_code); |
| 1190 |
} |
| 1191 |
|
| 1192 |
return true; |
| 1193 |
} |
| 1194 |
|
| 1195 |
/** |
| 1196 |
* Delete all chunks for a URL from WordPress database |
| 1197 |
*/ |
| 1198 |
private static function delete_wordpress_chunks_by_url($source_url) { |
| 1199 |
global $wpdb; |
| 1200 |
$table_name = $wpdb->prefix . 'mxchat_system_prompt_content'; |
| 1201 |
|
| 1202 |
// Delete all rows with this source_url (handles both chunked and non-chunked) |
| 1203 |
$result = $wpdb->delete( |
| 1204 |
$table_name, |
| 1205 |
array('source_url' => $source_url), |
| 1206 |
array('%s') |
| 1207 |
); |
| 1208 |
|
| 1209 |
if ($result === false) { |
| 1210 |
return new WP_Error('database_delete', 'Failed to delete chunks: ' . $wpdb->last_error); |
| 1211 |
} |
| 1212 |
|
| 1213 |
//error_log('[MXCHAT-CHUNK-DELETE] Deleted ' . $result . ' rows from WordPress DB'); |
| 1214 |
return true; |
| 1215 |
} |
| 1216 |
|
| 1217 |
/** |
| 1218 |
* Hybrid keyword boost (plan-38ffa1): detect whether the WP-DB knowledge |
| 1219 |
* table can serve the keyword leg via a MySQL FULLTEXT index, creating the |
| 1220 |
* index if needed. Detection runs once and caches the answer in the |
| 1221 |
* mxchat_hybrid_keyword_capability option ('fulltext' | 'like'); pass |
| 1222 |
* $force to re-detect. LIKE is the graceful fallback for shared hosts |
| 1223 |
* whose ALTER fails — the feature works either way, FULLTEXT just ranks |
| 1224 |
* better and scales. |
| 1225 |
* |
| 1226 |
* @param bool $force Re-run detection even if a cached answer exists. |
| 1227 |
* @return string 'fulltext' or 'like' |
| 1228 |
*/ |
| 1229 |
public static function mxchat_hybrid_detect_capability($force = false) { |
| 1230 |
$cached = get_option('mxchat_hybrid_keyword_capability', ''); |
| 1231 |
if (!$force && in_array($cached, array('fulltext', 'like'), true)) { |
| 1232 |
return $cached; |
| 1233 |
} |
| 1234 |
|
| 1235 |
global $wpdb; |
| 1236 |
$table = $wpdb->prefix . 'mxchat_system_prompt_content'; |
| 1237 |
|
| 1238 |
$index_exists = $wpdb->get_var("SHOW INDEX FROM {$table} WHERE Key_name = 'mxchat_content_ft'"); |
| 1239 |
if (!$index_exists) { |
| 1240 |
// Suppress the visible error on hosts where this is not permitted — |
| 1241 |
// failure is an expected, handled outcome (LIKE fallback). |
| 1242 |
$suppress = $wpdb->suppress_errors(true); |
| 1243 |
$wpdb->query("ALTER TABLE {$table} ADD FULLTEXT INDEX mxchat_content_ft (article_content)"); |
| 1244 |
$wpdb->suppress_errors($suppress); |
| 1245 |
$index_exists = $wpdb->get_var("SHOW INDEX FROM {$table} WHERE Key_name = 'mxchat_content_ft'"); |
| 1246 |
} |
| 1247 |
|
| 1248 |
$capability = $index_exists ? 'fulltext' : 'like'; |
| 1249 |
update_option('mxchat_hybrid_keyword_capability', $capability); |
| 1250 |
return $capability; |
| 1251 |
} |
| 1252 |
} |
