| 1 |
<?php |
| 2 |
|
| 3 |
namespace wpforo\classes; |
| 4 |
|
| 5 |
// Exit if accessed directly |
| 6 |
if ( ! defined( 'ABSPATH' ) ) exit; |
| 7 |
|
| 8 |
/** |
| 9 |
* Vector Storage Manager - Unified Abstraction Layer |
| 10 |
* |
| 11 |
* Central manager that routes all vector storage operations to the appropriate |
| 12 |
* backend (Local WordPress DB or Cloud Storage) based on the current storage mode. |
| 13 |
* |
| 14 |
* All AI features that use indexed content should go through this abstraction: |
| 15 |
* - Content indexing |
| 16 |
* - Semantic search |
| 17 |
* - Statistics and status |
| 18 |
* - Similar content recommendations |
| 19 |
* |
| 20 |
* @since 3.0.0 |
| 21 |
*/ |
| 22 |
class VectorStorageManager { |
| 23 |
|
| 24 |
/** |
| 25 |
* Storage mode constants |
| 26 |
*/ |
| 27 |
const MODE_LOCAL = 'local'; |
| 28 |
const MODE_CLOUD = 'cloud'; |
| 29 |
|
| 30 |
/** |
| 31 |
* Local storage instance |
| 32 |
* |
| 33 |
* @var VectorStorageLocal|null |
| 34 |
*/ |
| 35 |
private $local_storage = null; |
| 36 |
|
| 37 |
/** |
| 38 |
* AI Client instance (for cloud operations) |
| 39 |
* |
| 40 |
* @var AIClient|null |
| 41 |
*/ |
| 42 |
private $ai_client = null; |
| 43 |
|
| 44 |
/** |
| 45 |
* Current board ID |
| 46 |
* |
| 47 |
* @var int |
| 48 |
*/ |
| 49 |
private $board_id = 0; |
| 50 |
|
| 51 |
/** |
| 52 |
* Cached storage mode |
| 53 |
* |
| 54 |
* @var string|null |
| 55 |
*/ |
| 56 |
private $storage_mode = null; |
| 57 |
|
| 58 |
/** |
| 59 |
* Singleton instance |
| 60 |
* |
| 61 |
* @var VectorStorageManager|null |
| 62 |
*/ |
| 63 |
private static $instance = null; |
| 64 |
|
| 65 |
/** |
| 66 |
* Get singleton instance |
| 67 |
* |
| 68 |
* @return VectorStorageManager |
| 69 |
*/ |
| 70 |
public static function instance() { |
| 71 |
if ( self::$instance === null ) { |
| 72 |
self::$instance = new self(); |
| 73 |
} |
| 74 |
return self::$instance; |
| 75 |
} |
| 76 |
|
| 77 |
/** |
| 78 |
* Constructor |
| 79 |
*/ |
| 80 |
public function __construct() { |
| 81 |
$this->board_id = WPF()->board->get_current( 'boardid' ); |
| 82 |
$this->register_cron_hooks(); |
| 83 |
|
| 84 |
// Update board_id when WPF()->change_board() is called |
| 85 |
add_action( 'wpforo_after_change_board', [ $this, 'on_board_change' ] ); |
| 86 |
} |
| 87 |
|
| 88 |
/** |
| 89 |
* Handle board change event |
| 90 |
* |
| 91 |
* Updates internal board_id to match the new board context. |
| 92 |
* Called by wpforo_after_change_board hook. |
| 93 |
* |
| 94 |
* @return void |
| 95 |
*/ |
| 96 |
public function on_board_change() { |
| 97 |
$this->board_id = WPF()->board->get_current( 'boardid' ); |
| 98 |
$this->storage_mode = null; // Reset cached storage mode |
| 99 |
} |
| 100 |
|
| 101 |
/** |
| 102 |
* Register cron hooks for local storage maintenance |
| 103 |
*/ |
| 104 |
private function register_cron_hooks() { |
| 105 |
// Register cleanup action - must be done here so callback exists when cron fires |
| 106 |
add_action( 'wpforo_ai_cleanup_expired_cache', [ $this, 'cleanup_expired_cache' ] ); |
| 107 |
|
| 108 |
// Schedule if not already scheduled |
| 109 |
if ( ! wp_next_scheduled( 'wpforo_ai_cleanup_expired_cache' ) ) { |
| 110 |
wp_schedule_event( time(), 'hourly', 'wpforo_ai_cleanup_expired_cache' ); |
| 111 |
} |
| 112 |
} |
| 113 |
|
| 114 |
/** |
| 115 |
* Cleanup expired cache entries (cron callback) |
| 116 |
*/ |
| 117 |
public function cleanup_expired_cache() { |
| 118 |
if ( $this->is_local_mode() ) { |
| 119 |
$local = $this->get_local_storage(); |
| 120 |
$local->cleanup_expired_cache(); |
| 121 |
} |
| 122 |
} |
| 123 |
|
| 124 |
/** |
| 125 |
* Get the current storage mode for the board |
| 126 |
* |
| 127 |
* @param int|null $board_id Optional board ID (uses current if not specified) |
| 128 |
* @return string 'local' or 'cloud' |
| 129 |
*/ |
| 130 |
public function get_storage_mode( $board_id = null ) { |
| 131 |
if ( $board_id === null ) { |
| 132 |
$board_id = $this->board_id; |
| 133 |
} |
| 134 |
|
| 135 |
// Return cached value if same board |
| 136 |
if ( $this->storage_mode !== null && $board_id === $this->board_id ) { |
| 137 |
return $this->storage_mode; |
| 138 |
} |
| 139 |
|
| 140 |
$this->storage_mode = get_option( 'wpforo_ai_storage_mode_' . $board_id, self::MODE_LOCAL ); |
| 141 |
return $this->storage_mode; |
| 142 |
} |
| 143 |
|
| 144 |
/** |
| 145 |
* Check if using local storage mode |
| 146 |
* |
| 147 |
* @param int|null $board_id Optional board ID |
| 148 |
* @return bool |
| 149 |
*/ |
| 150 |
public function is_local_mode( $board_id = null ) { |
| 151 |
return $this->get_storage_mode( $board_id ) === self::MODE_LOCAL; |
| 152 |
} |
| 153 |
|
| 154 |
/** |
| 155 |
* Check if using cloud storage mode |
| 156 |
* |
| 157 |
* @param int|null $board_id Optional board ID |
| 158 |
* @return bool |
| 159 |
*/ |
| 160 |
public function is_cloud_mode( $board_id = null ) { |
| 161 |
return $this->get_storage_mode( $board_id ) === self::MODE_CLOUD; |
| 162 |
} |
| 163 |
|
| 164 |
/** |
| 165 |
* Check if any content has been indexed in the current storage mode |
| 166 |
* |
| 167 |
* Lightweight check using SELECT EXISTS — no API calls. |
| 168 |
* |
| 169 |
* @return bool |
| 170 |
*/ |
| 171 |
public function has_indexed_content() { |
| 172 |
global $wpdb; |
| 173 |
|
| 174 |
$column = $this->is_local_mode() ? 'local' : 'cloud'; |
| 175 |
|
| 176 |
return (bool) $wpdb->get_var( |
| 177 |
"SELECT EXISTS( SELECT 1 FROM `" . WPF()->tables->topics . "` WHERE `{$column}` = 1 LIMIT 1 )" |
| 178 |
); |
| 179 |
} |
| 180 |
|
| 181 |
/** |
| 182 |
* Get local storage instance |
| 183 |
* |
| 184 |
* @return VectorStorageLocal |
| 185 |
*/ |
| 186 |
public function get_local_storage() { |
| 187 |
if ( $this->local_storage === null ) { |
| 188 |
$this->local_storage = new VectorStorageLocal(); |
| 189 |
} |
| 190 |
return $this->local_storage; |
| 191 |
} |
| 192 |
|
| 193 |
/** |
| 194 |
* Get AI client instance (for cloud operations) |
| 195 |
* |
| 196 |
* @return AIClient |
| 197 |
*/ |
| 198 |
public function get_ai_client() { |
| 199 |
if ( $this->ai_client === null ) { |
| 200 |
$this->ai_client = WPF()->ai_client; |
| 201 |
} |
| 202 |
return $this->ai_client; |
| 203 |
} |
| 204 |
|
| 205 |
/** |
| 206 |
* Set the board context |
| 207 |
* |
| 208 |
* @param int $board_id Board ID |
| 209 |
* @return $this |
| 210 |
*/ |
| 211 |
public function for_board( $board_id ) { |
| 212 |
$this->board_id = (int) $board_id; |
| 213 |
$this->storage_mode = null; // Reset cache |
| 214 |
return $this; |
| 215 |
} |
| 216 |
|
| 217 |
/** |
| 218 |
* Reset cached storage mode |
| 219 |
* |
| 220 |
* Call this after changing the storage mode option to ensure |
| 221 |
* subsequent calls use the new mode. |
| 222 |
* |
| 223 |
* @return void |
| 224 |
*/ |
| 225 |
public function reset_storage_mode_cache() { |
| 226 |
$this->storage_mode = null; |
| 227 |
} |
| 228 |
|
| 229 |
/** |
| 230 |
* Clear indexing statistics caches |
| 231 |
* |
| 232 |
* Call this after batch indexing completes, after clearing embeddings, |
| 233 |
* or after any bulk operation that changes indexed topic counts. |
| 234 |
* Individual topic indexing does NOT need to call this - the 5-minute |
| 235 |
* TTL handles gradual updates during indexing. |
| 236 |
* |
| 237 |
* @param int|null $board_id Specific board ID, or null for current board |
| 238 |
* @return void |
| 239 |
*/ |
| 240 |
public function clear_indexing_stats_cache( $board_id = null ) { |
| 241 |
$board_id = $board_id ?? $this->board_id; |
| 242 |
|
| 243 |
// Clear cloud indexed count cache |
| 244 |
delete_transient( 'wpforo_ai_cloud_indexed_count_' . $board_id ); |
| 245 |
|
| 246 |
// Clear storage recommendation cache |
| 247 |
delete_transient( 'wpforo_ai_srec_' . $board_id . '_local' ); |
| 248 |
delete_transient( 'wpforo_ai_srec_' . $board_id . '_cloud' ); |
| 249 |
|
| 250 |
// Clear total topics count cache (from admin page, includes storage mode suffix) |
| 251 |
delete_transient( 'wpforo_ai_ttc_' . $board_id . '_local' ); |
| 252 |
delete_transient( 'wpforo_ai_ttc_' . $board_id . '_cloud' ); |
| 253 |
|
| 254 |
// Clear forum topic counts cache (from admin page) |
| 255 |
delete_transient( 'wpforo_ai_ftc_' . $board_id ); |
| 256 |
} |
| 257 |
|
| 258 |
// ========================================================================= |
| 259 |
// STATISTICS & STATUS |
| 260 |
// ========================================================================= |
| 261 |
|
| 262 |
/** |
| 263 |
* Get indexing statistics |
| 264 |
* |
| 265 |
* Returns unified statistics regardless of storage mode. |
| 266 |
* |
| 267 |
* @return array { |
| 268 |
* @type int $total_indexed Total number of indexed items |
| 269 |
* @type int $total_topics Total topics with embeddings |
| 270 |
* @type int $indexing_progress Progress percentage (0-100) |
| 271 |
* @type bool $is_indexing Whether indexing is in progress |
| 272 |
* @type string $last_indexed_at ISO 8601 timestamp of last index |
| 273 |
* @type string $storage_mode Current storage mode |
| 274 |
* @type string $storage_size Storage size (local only) |
| 275 |
* } |
| 276 |
*/ |
| 277 |
public function get_indexing_stats() { |
| 278 |
if ( $this->is_local_mode() ) { |
| 279 |
return $this->get_local_stats(); |
| 280 |
} else { |
| 281 |
return $this->get_cloud_stats(); |
| 282 |
} |
| 283 |
} |
| 284 |
|
| 285 |
/** |
| 286 |
* Get local storage statistics |
| 287 |
* |
| 288 |
* @return array |
| 289 |
*/ |
| 290 |
private function get_local_stats() { |
| 291 |
$local = $this->get_local_storage(); |
| 292 |
$stats = $local->get_stats(); |
| 293 |
|
| 294 |
// Only report is_indexing when a batch is actually being processed |
| 295 |
// (lock transient is held), not when topics are merely queued. |
| 296 |
$board_id = $this->board_id; |
| 297 |
$is_actively_indexing = (bool) get_transient( 'wpforo_ai_indexing_lock_local_' . $board_id ) |
| 298 |
|| (bool) get_transient( 'wpforo_ai_indexing_lock_' . $board_id ); |
| 299 |
|
| 300 |
return [ |
| 301 |
'total_indexed' => (int) ( $stats['total_embeddings'] ?? 0 ), |
| 302 |
'total_topics' => (int) ( $stats['total_topics'] ?? 0 ), |
| 303 |
'indexing_progress' => $this->calculate_local_indexing_progress( $board_id ), |
| 304 |
'is_indexing' => $is_actively_indexing, |
| 305 |
'last_indexed_at' => $stats['last_indexed_at'] ?? null, |
| 306 |
'storage_mode' => self::MODE_LOCAL, |
| 307 |
'storage_size' => $stats['storage_size_mb'] ?? '0', |
| 308 |
'storage_size_mb' => $stats['storage_size_mb'] ?? '0', |
| 309 |
]; |
| 310 |
} |
| 311 |
|
| 312 |
/** |
| 313 |
* Calculate local indexing progress percentage. |
| 314 |
* |
| 315 |
* Reads the queue and settings options to calculate how many topics |
| 316 |
* have been processed out of the total queued for indexing. |
| 317 |
* |
| 318 |
* @param int $board_id Board ID |
| 319 |
* @return int Progress percentage (0-100), or 0 if no indexing in progress |
| 320 |
*/ |
| 321 |
private function calculate_local_indexing_progress( $board_id ) { |
| 322 |
$settings_key = 'wpforo_ai_indexing_settings_' . $board_id; |
| 323 |
$settings = get_option( $settings_key, [] ); |
| 324 |
|
| 325 |
// No settings = no indexing session started |
| 326 |
$total = (int) ( $settings['total_topics'] ?? 0 ); |
| 327 |
if ( $total <= 0 ) { |
| 328 |
return 0; |
| 329 |
} |
| 330 |
|
| 331 |
// Count remaining topics in queue |
| 332 |
$queue_key = 'wpforo_ai_indexing_queue_' . $board_id; |
| 333 |
$pending = get_option( $queue_key, [] ); |
| 334 |
$remaining = is_array( $pending ) ? count( $pending ) : 0; |
| 335 |
|
| 336 |
// Calculate progress |
| 337 |
$processed = $total - $remaining; |
| 338 |
if ( $processed < 0 ) { |
| 339 |
$processed = 0; // Safety: queue grew larger than original total |
| 340 |
} |
| 341 |
|
| 342 |
return (int) round( ( $processed / $total ) * 100 ); |
| 343 |
} |
| 344 |
|
| 345 |
/** |
| 346 |
* Get cloud storage statistics |
| 347 |
* |
| 348 |
* @return array |
| 349 |
*/ |
| 350 |
private function get_cloud_stats() { |
| 351 |
global $wpdb; |
| 352 |
|
| 353 |
$ai_client = $this->get_ai_client(); |
| 354 |
$rag_status = $ai_client->get_rag_status( $this->board_id ); |
| 355 |
|
| 356 |
if ( is_wp_error( $rag_status ) ) { |
| 357 |
$rag_status = []; |
| 358 |
} |
| 359 |
|
| 360 |
// Only report is_indexing when the backend is actively processing. |
| 361 |
// Queued WP Cron topics are reported separately via pending_cron_jobs. |
| 362 |
$is_indexing = (bool) ( $rag_status['is_indexing'] ?? false ); |
| 363 |
|
| 364 |
// Use local cloud column for accurate count (reflects manual changes) |
| 365 |
// Cache for 5 minutes to avoid heavy COUNT on large forums |
| 366 |
$cache_key = 'wpforo_ai_cloud_indexed_count_' . $this->board_id; |
| 367 |
$total_indexed = get_transient( $cache_key ); |
| 368 |
if ( false === $total_indexed ) { |
| 369 |
$total_indexed = (int) $wpdb->get_var( |
| 370 |
"SELECT COUNT(*) FROM `" . WPF()->tables->topics . "` WHERE `cloud` = 1" |
| 371 |
); |
| 372 |
set_transient( $cache_key, $total_indexed, 5 * MINUTE_IN_SECONDS ); |
| 373 |
} |
| 374 |
$total_indexed = (int) $total_indexed; |
| 375 |
|
| 376 |
return [ |
| 377 |
'total_indexed' => $total_indexed, |
| 378 |
'total_topics' => $total_indexed, |
| 379 |
'indexing_progress' => (int) ( $rag_status['indexing_progress'] ?? 0 ), |
| 380 |
'is_indexing' => $is_indexing, |
| 381 |
'last_indexed_at' => $rag_status['last_indexed_at'] ?? null, |
| 382 |
'storage_mode' => self::MODE_CLOUD, |
| 383 |
'storage_size' => null, // Cloud doesn't expose size |
| 384 |
'storage_size_mb' => null, |
| 385 |
// Async media (image/document) sub-progress from the backend. |
| 386 |
// Present when the image_worker has queued work for this tenant. |
| 387 |
// See GET /v1/rag/status and the admin UI sub-progress bar. |
| 388 |
'media_progress' => $rag_status['media_progress'] ?? null, |
| 389 |
]; |
| 390 |
} |
| 391 |
|
| 392 |
/** |
| 393 |
* Get pending WP Cron jobs info |
| 394 |
* |
| 395 |
* @return array |
| 396 |
*/ |
| 397 |
public function get_pending_cron_jobs() { |
| 398 |
$ai_client = $this->get_ai_client(); |
| 399 |
return $ai_client->get_pending_cron_jobs(); |
| 400 |
} |
| 401 |
|
| 402 |
/** |
| 403 |
* Get storage recommendation for current forum size |
| 404 |
* |
| 405 |
* @return array { |
| 406 |
* @type string $status 'good', 'warning', or 'critical' |
| 407 |
* @type string $message Human-readable recommendation |
| 408 |
* @type string $icon Dashicon name |
| 409 |
* } |
| 410 |
*/ |
| 411 |
public function get_storage_recommendation() { |
| 412 |
// Cache the recommendation to avoid heavy COUNT(*) on posts table |
| 413 |
// Board and storage mode specific to prevent stale data across switches |
| 414 |
$cache_key = 'wpforo_ai_srec_' . $this->board_id . '_' . $this->get_storage_mode(); |
| 415 |
$cached = get_transient( $cache_key ); |
| 416 |
if ( false !== $cached && is_array( $cached ) ) { |
| 417 |
return $cached; |
| 418 |
} |
| 419 |
|
| 420 |
$local = $this->get_local_storage(); |
| 421 |
$result = $local->get_storage_recommendation(); |
| 422 |
set_transient( $cache_key, $result, 10 * MINUTE_IN_SECONDS ); |
| 423 |
return $result; |
| 424 |
} |
| 425 |
|
| 426 |
/** |
| 427 |
* Get indexed counts grouped by forum |
| 428 |
* |
| 429 |
* Cached for 5 minutes to avoid heavy GROUP BY queries on large forums. |
| 430 |
* |
| 431 |
* @return array Forum ID => count mapping |
| 432 |
*/ |
| 433 |
public function get_indexed_counts_by_forum() { |
| 434 |
if ( $this->is_local_mode() ) { |
| 435 |
return $this->get_local_indexed_counts_by_forum(); |
| 436 |
} else { |
| 437 |
return $this->get_cloud_indexed_counts_by_forum(); |
| 438 |
} |
| 439 |
} |
| 440 |
|
| 441 |
/** |
| 442 |
* Get local indexed counts by forum |
| 443 |
* |
| 444 |
* @return array |
| 445 |
*/ |
| 446 |
private function get_local_indexed_counts_by_forum() { |
| 447 |
global $wpdb; |
| 448 |
|
| 449 |
$results = $wpdb->get_results( |
| 450 |
"SELECT forumid, COUNT(DISTINCT topicid) as count |
| 451 |
FROM " . WPF()->tables->ai_embeddings . " |
| 452 |
GROUP BY forumid", |
| 453 |
ARRAY_A |
| 454 |
); |
| 455 |
|
| 456 |
$counts = []; |
| 457 |
foreach ( $results as $row ) { |
| 458 |
$counts[ (int) $row['forumid'] ] = (int) $row['count']; |
| 459 |
} |
| 460 |
|
| 461 |
return $counts; |
| 462 |
} |
| 463 |
|
| 464 |
/** |
| 465 |
* Get cloud indexed counts by forum |
| 466 |
* |
| 467 |
* Uses the local `cloud` column from wpforo_topics table which is |
| 468 |
* synced from the cloud API when switching storage modes. |
| 469 |
* |
| 470 |
* @return array |
| 471 |
*/ |
| 472 |
private function get_cloud_indexed_counts_by_forum() { |
| 473 |
global $wpdb; |
| 474 |
|
| 475 |
$results = $wpdb->get_results( |
| 476 |
"SELECT forumid, COUNT(*) as count |
| 477 |
FROM " . WPF()->tables->topics . " |
| 478 |
WHERE `cloud` = 1 |
| 479 |
GROUP BY forumid", |
| 480 |
ARRAY_A |
| 481 |
); |
| 482 |
|
| 483 |
$counts = []; |
| 484 |
foreach ( $results as $row ) { |
| 485 |
$counts[ (int) $row['forumid'] ] = (int) $row['count']; |
| 486 |
} |
| 487 |
|
| 488 |
return $counts; |
| 489 |
} |
| 490 |
|
| 491 |
// ========================================================================= |
| 492 |
// INDEXING OPERATIONS |
| 493 |
// ========================================================================= |
| 494 |
|
| 495 |
/** |
| 496 |
* Index a single topic with all its posts |
| 497 |
* |
| 498 |
* @param int $topicid Topic ID |
| 499 |
* @param array $options Optional indexing options |
| 500 |
* @return array|WP_Error Result or error |
| 501 |
*/ |
| 502 |
public function index_topic( $topicid, $options = [] ) { |
| 503 |
if ( $this->is_local_mode() ) { |
| 504 |
return $this->index_topic_local( $topicid, $options ); |
| 505 |
} else { |
| 506 |
return $this->index_topic_cloud( $topicid, $options ); |
| 507 |
} |
| 508 |
} |
| 509 |
|
| 510 |
/** |
| 511 |
* Index topic to local storage |
| 512 |
* |
| 513 |
* @param int $topicid Topic ID |
| 514 |
* @param array $options Options |
| 515 |
* @return array|WP_Error |
| 516 |
*/ |
| 517 |
private function index_topic_local( $topicid, $options = [] ) { |
| 518 |
$topic = WPF()->topic->get_topic( $topicid ); |
| 519 |
if ( ! $topic ) { |
| 520 |
return new \WP_Error( 'topic_not_found', wpforo_phrase( 'Topic not found', false ) ); |
| 521 |
} |
| 522 |
|
| 523 |
// Skip private topics - they should never be indexed |
| 524 |
if ( ! empty( $topic['private'] ) ) { |
| 525 |
return new \WP_Error( 'private_topic', wpforo_phrase( 'Private topics cannot be indexed', false ) ); |
| 526 |
} |
| 527 |
|
| 528 |
// Skip unapproved topics |
| 529 |
if ( isset( $topic['status'] ) && (int) $topic['status'] !== 0 ) { |
| 530 |
return new \WP_Error( 'unapproved_topic', wpforo_phrase( 'Unapproved topics cannot be indexed', false ) ); |
| 531 |
} |
| 532 |
|
| 533 |
// Get all posts for this topic ordered by creation date |
| 534 |
$posts = WPF()->post->get_posts( [ |
| 535 |
'topicid' => $topicid, |
| 536 |
'orderby' => 'created', |
| 537 |
'order' => 'ASC', |
| 538 |
] ); |
| 539 |
if ( empty( $posts ) ) { |
| 540 |
return new \WP_Error( 'no_posts', wpforo_phrase( 'No posts found for topic', false ) ); |
| 541 |
} |
| 542 |
|
| 543 |
// Pre-fetch forum name for enrichment tags (Phase 2.1) |
| 544 |
// Added to $topic array so prepare_content_for_embedding() doesn't need wpforo_forum() |
| 545 |
if ( ! empty( $topic['forumid'] ) && empty( $topic['forum_name'] ) ) { |
| 546 |
$forum = wpforo_forum( $topic['forumid'] ); |
| 547 |
$topic['forum_name'] = is_array( $forum ) ? ( $forum['title'] ?? '' ) : ''; |
| 548 |
} |
| 549 |
|
| 550 |
// Check if image/document indexing is enabled (Professional+ plans) |
| 551 |
$ai_client = $this->get_ai_client(); |
| 552 |
$include_images = $ai_client->is_image_indexing_enabled(); |
| 553 |
$include_documents = $ai_client->is_document_indexing_enabled(); |
| 554 |
$images_processed = 0; |
| 555 |
$documents_processed = 0; |
| 556 |
|
| 557 |
$local = $this->get_local_storage(); |
| 558 |
$indexed_count = 0; |
| 559 |
$noise_filtered_count = 0; |
| 560 |
$errors = []; |
| 561 |
$is_first = true; |
| 562 |
|
| 563 |
foreach ( $posts as $post ) { |
| 564 |
// Mark first post for special handling |
| 565 |
$post['is_first_post'] = $is_first; |
| 566 |
$current_is_first = $is_first; |
| 567 |
$is_first = false; |
| 568 |
|
| 569 |
// Extract images if image indexing is enabled |
| 570 |
$images = []; |
| 571 |
if ( $include_images && ! empty( $post['body'] ) ) { |
| 572 |
$images = $ai_client->extract_post_images( $post['body'] ); |
| 573 |
} |
| 574 |
|
| 575 |
// Extract documents if document indexing is enabled |
| 576 |
$documents = []; |
| 577 |
if ( $include_documents && ! empty( $post['body'] ) ) { |
| 578 |
$documents = $ai_client->extract_post_documents( $post['body'] ); |
| 579 |
} |
| 580 |
|
| 581 |
// Skip noise posts, but NEVER skip: |
| 582 |
// - First post (topic body is always indexed) |
| 583 |
// - Posts with images/documents that WILL be indexed (attachments have value even with short text) |
| 584 |
// Note: $images/$documents are only populated if indexing is enabled, so this check is correct |
| 585 |
$has_indexable_attachments = ! empty( $images ) || ! empty( $documents ); |
| 586 |
if ( ! $current_is_first && ! $has_indexable_attachments && $this->is_noise_post( $post['body'] ?? '' ) ) { |
| 587 |
$noise_filtered_count++; |
| 588 |
continue; |
| 589 |
} |
| 590 |
|
| 591 |
// Generate content for embedding |
| 592 |
$content = $this->prepare_content_for_embedding( $post, $topic ); |
| 593 |
|
| 594 |
// Use content hash for deduplication |
| 595 |
// Include image and document counts so re-index happens when attachments change |
| 596 |
$hash_input = $content . '|images:' . count( $images ) . '|docs:' . count( $documents ); |
| 597 |
$content_hash = md5( $hash_input ); |
| 598 |
|
| 599 |
// Check if already indexed with same content |
| 600 |
$existing = $local->get_embedding( $post['postid'] ); |
| 601 |
if ( $existing && $existing['content_hash'] === $content_hash ) { |
| 602 |
continue; // Skip, already indexed |
| 603 |
} |
| 604 |
|
| 605 |
// Generate embedding via cloud API |
| 606 |
// If images or documents provided, API will process them and return processed_content |
| 607 |
if ( ! empty( $images ) || ! empty( $documents ) ) { |
| 608 |
$embedding_result = $this->generate_embedding( |
| 609 |
$content, |
| 610 |
$images, |
| 611 |
$topic['title'], |
| 612 |
true, // Return full response to get processed_content |
| 613 |
$documents |
| 614 |
); |
| 615 |
|
| 616 |
if ( is_wp_error( $embedding_result ) ) { |
| 617 |
$errors[] = $embedding_result->get_error_message(); |
| 618 |
continue; |
| 619 |
} |
| 620 |
|
| 621 |
// Track image processing stats |
| 622 |
if ( ! empty( $embedding_result['image_processing']['images_processed'] ) ) { |
| 623 |
$images_processed += $embedding_result['image_processing']['images_processed']; |
| 624 |
} |
| 625 |
|
| 626 |
// Track document processing stats |
| 627 |
if ( ! empty( $embedding_result['document_processing']['documents_processed'] ) ) { |
| 628 |
$documents_processed += $embedding_result['document_processing']['documents_processed']; |
| 629 |
} |
| 630 |
|
| 631 |
$embedding = $embedding_result['embedding']; |
| 632 |
// Build clean preview with document/image summary |
| 633 |
$preview = $this->build_content_preview( $content, $embedding_result ); |
| 634 |
} else { |
| 635 |
// No images or documents, simple embedding generation |
| 636 |
$embedding = $this->generate_embedding( $content ); |
| 637 |
if ( is_wp_error( $embedding ) ) { |
| 638 |
$errors[] = $embedding->get_error_message(); |
| 639 |
continue; |
| 640 |
} |
| 641 |
$preview = $this->build_content_preview( $content ); |
| 642 |
} |
| 643 |
|
| 644 |
// Store locally |
| 645 |
$result = $local->store_embedding( |
| 646 |
$topicid, |
| 647 |
$post['postid'], |
| 648 |
$topic['forumid'], |
| 649 |
$post['userid'], |
| 650 |
$embedding, |
| 651 |
$content_hash, |
| 652 |
$preview |
| 653 |
); |
| 654 |
|
| 655 |
if ( $result ) { |
| 656 |
$indexed_count++; |
| 657 |
} |
| 658 |
} |
| 659 |
|
| 660 |
$response = [ |
| 661 |
'success' => true, |
| 662 |
'indexed_count' => $indexed_count, |
| 663 |
'noise_filtered_count' => $noise_filtered_count, |
| 664 |
'total_posts' => count( $posts ), |
| 665 |
'errors' => $errors, |
| 666 |
]; |
| 667 |
|
| 668 |
// Add image processing stats if images were processed |
| 669 |
if ( $images_processed > 0 ) { |
| 670 |
$response['images_processed'] = $images_processed; |
| 671 |
} |
| 672 |
|
| 673 |
// Add document processing stats if documents were processed |
| 674 |
if ( $documents_processed > 0 ) { |
| 675 |
$response['documents_processed'] = $documents_processed; |
| 676 |
} |
| 677 |
|
| 678 |
return $response; |
| 679 |
} |
| 680 |
|
| 681 |
/** |
| 682 |
* Index topic to cloud storage |
| 683 |
* |
| 684 |
* @param int $topicid Topic ID |
| 685 |
* @param array $options Options |
| 686 |
* @return array|WP_Error |
| 687 |
*/ |
| 688 |
private function index_topic_cloud( $topicid, $options = [] ) { |
| 689 |
$ai_client = $this->get_ai_client(); |
| 690 |
|
| 691 |
// Use existing cloud indexing flow via cron |
| 692 |
$topic = WPF()->topic->get_topic( $topicid ); |
| 693 |
if ( ! $topic ) { |
| 694 |
return new \WP_Error( 'topic_not_found', wpforo_phrase( 'Topic not found', false ) ); |
| 695 |
} |
| 696 |
|
| 697 |
// Skip private topics - they should never be indexed |
| 698 |
if ( ! empty( $topic['private'] ) ) { |
| 699 |
return new \WP_Error( 'private_topic', wpforo_phrase( 'Private topics cannot be indexed', false ) ); |
| 700 |
} |
| 701 |
|
| 702 |
// Skip unapproved topics |
| 703 |
if ( isset( $topic['status'] ) && (int) $topic['status'] !== 0 ) { |
| 704 |
return new \WP_Error( 'unapproved_topic', wpforo_phrase( 'Unapproved topics cannot be indexed', false ) ); |
| 705 |
} |
| 706 |
|
| 707 |
// Queue for background processing (existing behavior) |
| 708 |
$result = $ai_client->queue_topic_for_indexing( $topicid, $this->board_id ); |
| 709 |
|
| 710 |
return $result; |
| 711 |
} |
| 712 |
|
| 713 |
/** |
| 714 |
* Index multiple topics using batch embedding API |
| 715 |
* |
| 716 |
* Efficient method that collects all posts from multiple topics, |
| 717 |
* generates embeddings in a single API call, and stores locally. |
| 718 |
* This matches the cloud indexing pattern for efficiency. |
| 719 |
* |
| 720 |
* @param array $topic_ids Array of topic IDs to index |
| 721 |
* @param array $options Optional indexing options |
| 722 |
* @return array Result with counts |
| 723 |
*/ |
| 724 |
public function index_topics_batch_local( $topic_ids, $options = [] ) { |
| 725 |
if ( empty( $topic_ids ) ) { |
| 726 |
return [ |
| 727 |
'success' => true, |
| 728 |
'indexed_count' => 0, |
| 729 |
'skipped_count' => 0, |
| 730 |
'noise_filtered_count' => 0, |
| 731 |
'total_posts' => 0, |
| 732 |
'errors' => [], |
| 733 |
]; |
| 734 |
} |
| 735 |
|
| 736 |
$local = $this->get_local_storage(); |
| 737 |
$ai_client = $this->get_ai_client(); |
| 738 |
|
| 739 |
// One-time hash migration (v1 → v2): updates existing hashes to new format |
| 740 |
// that is independent of enrichment tag formatting. No API calls, no credit cost. |
| 741 |
$this->maybe_migrate_content_hashes(); |
| 742 |
|
| 743 |
// Check if image/document indexing is enabled (Professional+ plans) |
| 744 |
$include_images = $ai_client->is_image_indexing_enabled(); |
| 745 |
$include_documents = $ai_client->is_document_indexing_enabled(); |
| 746 |
|
| 747 |
// Collect all posts from all topics |
| 748 |
$items_to_embed = []; // Items without images/documents (batch endpoint) |
| 749 |
$items_with_images = []; // Items with images or documents (single endpoint) |
| 750 |
$post_metadata = []; // Metadata for storing after embedding |
| 751 |
$skipped_count = 0; |
| 752 |
$noise_filtered_count = 0; // Posts filtered as noise (short, gratitude, etc.) |
| 753 |
$topics_with_embeddings = []; // Topics that have at least one indexed post |
| 754 |
|
| 755 |
// Batch-fetch all topics in one query to avoid N+1 |
| 756 |
$_items_count = 0; |
| 757 |
$all_topics = WPF()->topic->get_topics( [ 'include' => $topic_ids, 'row_count' => count( $topic_ids ), 'access_filter' => false ], $_items_count, false ); |
| 758 |
$topics_map = []; |
| 759 |
foreach ( $all_topics as $t ) { |
| 760 |
$topics_map[ (int) $t['topicid'] ] = $t; |
| 761 |
} |
| 762 |
|
| 763 |
// Batch-fetch forum names for all unique forum IDs |
| 764 |
$forum_ids = array_unique( array_filter( array_column( $all_topics, 'forumid' ) ) ); |
| 765 |
$forums_map = []; |
| 766 |
foreach ( $forum_ids as $fid ) { |
| 767 |
$forums_map[ (int) $fid ] = wpforo_forum( $fid ); |
| 768 |
} |
| 769 |
|
| 770 |
foreach ( $topic_ids as $topicid ) { |
| 771 |
$topic = isset( $topics_map[ (int) $topicid ] ) ? $topics_map[ (int) $topicid ] : null; |
| 772 |
if ( ! $topic ) { |
| 773 |
continue; |
| 774 |
} |
| 775 |
|
| 776 |
// Use pre-fetched forum name for enrichment tags |
| 777 |
if ( ! empty( $topic['forumid'] ) && empty( $topic['forum_name'] ) ) { |
| 778 |
$forum = isset( $forums_map[ (int) $topic['forumid'] ] ) ? $forums_map[ (int) $topic['forumid'] ] : null; |
| 779 |
$topic['forum_name'] = is_array( $forum ) ? ( isset( $forum['title'] ) ? $forum['title'] : '' ) : ''; |
| 780 |
} |
| 781 |
|
| 782 |
// Skip private topics - they should never be indexed |
| 783 |
if ( ! empty( $topic['private'] ) ) { |
| 784 |
$skipped_count++; |
| 785 |
continue; |
| 786 |
} |
| 787 |
|
| 788 |
// Skip unapproved topics |
| 789 |
if ( isset( $topic['status'] ) && (int) $topic['status'] !== 0 ) { |
| 790 |
$skipped_count++; |
| 791 |
continue; |
| 792 |
} |
| 793 |
|
| 794 |
// Bypass user permission check - this is admin-initiated backend indexing. |
| 795 |
// Private/unapproved topics are already filtered above. |
| 796 |
$posts = WPF()->post->get_posts( [ |
| 797 |
'topicid' => $topicid, |
| 798 |
'orderby' => 'created', |
| 799 |
'order' => 'ASC', |
| 800 |
'check_private' => false, |
| 801 |
] ); |
| 802 |
|
| 803 |
if ( empty( $posts ) ) { |
| 804 |
continue; |
| 805 |
} |
| 806 |
|
| 807 |
$is_first = true; |
| 808 |
foreach ( $posts as $post ) { |
| 809 |
$post['is_first_post'] = $is_first; |
| 810 |
$current_is_first = $is_first; |
| 811 |
$is_first = false; |
| 812 |
|
| 813 |
// Extract images if image indexing is enabled |
| 814 |
$images = []; |
| 815 |
if ( $include_images && ! empty( $post['body'] ) ) { |
| 816 |
$images = $ai_client->extract_post_images( $post['body'] ); |
| 817 |
} |
| 818 |
|
| 819 |
// Extract documents if document indexing is enabled |
| 820 |
$documents = []; |
| 821 |
if ( $include_documents && ! empty( $post['body'] ) ) { |
| 822 |
$documents = $ai_client->extract_post_documents( $post['body'] ); |
| 823 |
} |
| 824 |
|
| 825 |
// Skip noise posts, but NEVER skip: |
| 826 |
// - First post (topic body is always indexed) |
| 827 |
// - Posts with images/documents that WILL be indexed (attachments have value even with short text) |
| 828 |
// Note: Only exempt if indexing is enabled AND post actually has attachments |
| 829 |
$has_indexable_attachments = ! empty( $images ) || ! empty( $documents ); |
| 830 |
if ( ! $current_is_first && ! $has_indexable_attachments && $this->is_noise_post( $post['body'] ?? '' ) ) { |
| 831 |
$noise_filtered_count++; |
| 832 |
continue; |
| 833 |
} |
| 834 |
|
| 835 |
// Prepare content for embedding |
| 836 |
$content = $this->prepare_content_for_embedding( $post, $topic ); |
| 837 |
$post_id = $post['postid']; |
| 838 |
|
| 839 |
// Stable hash from raw content — not from prepare_content_for_embedding() output. |
| 840 |
// This ensures enrichment formatting changes don't invalidate all existing hashes. |
| 841 |
$content_hash = $this->compute_content_hash( $post, $topic, count( $images ), count( $documents ) ); |
| 842 |
|
| 843 |
// Check if already indexed with same content (deduplication) |
| 844 |
$existing = $local->get_embedding( $post_id ); |
| 845 |
if ( $existing && $existing['content_hash'] === $content_hash ) { |
| 846 |
$skipped_count++; |
| 847 |
// Track that this topic has at least one indexed post |
| 848 |
$topics_with_embeddings[ $topicid ] = true; |
| 849 |
continue; |
| 850 |
} |
| 851 |
|
| 852 |
// Store metadata for later |
| 853 |
$item_id = 'post_' . $post_id; |
| 854 |
$post_metadata[ $item_id ] = [ |
| 855 |
'topicid' => $topicid, |
| 856 |
'postid' => $post_id, |
| 857 |
'forumid' => $topic['forumid'], |
| 858 |
'userid' => $post['userid'], |
| 859 |
'content_hash' => $content_hash, |
| 860 |
'preview' => $this->build_content_preview( $content ), |
| 861 |
'topic_title' => $topic['title'], |
| 862 |
]; |
| 863 |
|
| 864 |
// Separate posts with images/documents from text-only posts |
| 865 |
// Single endpoint handles image analysis and document processing |
| 866 |
if ( ! empty( $images ) || ! empty( $documents ) ) { |
| 867 |
$items_with_images[] = [ |
| 868 |
'id' => $item_id, |
| 869 |
'content' => $content, |
| 870 |
'images' => $images, |
| 871 |
'documents' => $documents, |
| 872 |
]; |
| 873 |
} else { |
| 874 |
$items_to_embed[] = [ |
| 875 |
'id' => $item_id, |
| 876 |
'content' => $content, |
| 877 |
]; |
| 878 |
} |
| 879 |
} |
| 880 |
} |
| 881 |
|
| 882 |
// If nothing to embed, update indexed hashes for topics with existing embeddings and return |
| 883 |
if ( empty( $items_to_embed ) && empty( $items_with_images ) ) { |
| 884 |
// Still update indexed hashes for topics that have existing embeddings |
| 885 |
// This ensures stats show correct count even if re-indexing finds no changes |
| 886 |
if ( ! empty( $topics_with_embeddings ) ) { |
| 887 |
$this->update_topics_indexed_hash( array_keys( $topics_with_embeddings ) ); |
| 888 |
} |
| 889 |
|
| 890 |
return [ |
| 891 |
'success' => true, |
| 892 |
'indexed_count' => 0, |
| 893 |
'skipped_count' => $skipped_count, |
| 894 |
'noise_filtered_count' => $noise_filtered_count, |
| 895 |
'total_posts' => $skipped_count + $noise_filtered_count, |
| 896 |
'errors' => [], |
| 897 |
'message' => sprintf( wpforo_phrase( 'All %d posts already indexed (unchanged).', false ), $skipped_count ), |
| 898 |
]; |
| 899 |
} |
| 900 |
|
| 901 |
$all_results = []; |
| 902 |
$errors = []; |
| 903 |
$total_credits_used = 0; |
| 904 |
$images_processed = 0; |
| 905 |
$documents_processed = 0; |
| 906 |
|
| 907 |
// ================================================================= |
| 908 |
// STEP 1: Process posts WITH images/documents via single endpoint |
| 909 |
// Single endpoint supports image analysis and document processing |
| 910 |
// ================================================================= |
| 911 |
if ( ! empty( $items_with_images ) ) { |
| 912 |
\wpforo_ai_log( 'debug', sprintf( |
| 913 |
'Processing %d posts with images/documents via single endpoint', |
| 914 |
count( $items_with_images ) |
| 915 |
), 'VectorStorage' ); |
| 916 |
|
| 917 |
foreach ( $items_with_images as $item ) { |
| 918 |
$item_id = $item['id']; |
| 919 |
$meta = $post_metadata[ $item_id ] ?? null; |
| 920 |
|
| 921 |
if ( ! $meta ) { |
| 922 |
continue; |
| 923 |
} |
| 924 |
|
| 925 |
// Generate embedding with images/documents via single endpoint |
| 926 |
$embedding_result = $this->generate_embedding( |
| 927 |
$item['content'], |
| 928 |
$item['images'] ?? [], |
| 929 |
$meta['topic_title'], |
| 930 |
true, // Return full response to get processed_content |
| 931 |
$item['documents'] ?? [] |
| 932 |
); |
| 933 |
|
| 934 |
if ( is_wp_error( $embedding_result ) ) { |
| 935 |
$errors[] = sprintf( 'Failed to embed %s: %s', $item_id, $embedding_result->get_error_message() ); |
| 936 |
continue; |
| 937 |
} |
| 938 |
|
| 939 |
// Track image processing stats |
| 940 |
if ( ! empty( $embedding_result['image_processing']['images_processed'] ) ) { |
| 941 |
$images_processed += $embedding_result['image_processing']['images_processed']; |
| 942 |
} |
| 943 |
|
| 944 |
// Track document processing stats |
| 945 |
if ( ! empty( $embedding_result['document_processing']['documents_processed'] ) ) { |
| 946 |
$documents_processed += $embedding_result['document_processing']['documents_processed']; |
| 947 |
} |
| 948 |
|
| 949 |
// Track credits from API response |
| 950 |
if ( isset( $embedding_result['credits_used'] ) ) { |
| 951 |
$total_credits_used += (int) $embedding_result['credits_used']; |
| 952 |
} |
| 953 |
|
| 954 |
// Build clean preview with document/image summary |
| 955 |
$clean_preview = $this->build_content_preview( $item['content'], $embedding_result ); |
| 956 |
|
| 957 |
// Add to results in same format as batch endpoint |
| 958 |
$all_results[] = [ |
| 959 |
'id' => $item_id, |
| 960 |
'success' => true, |
| 961 |
'embedding' => $embedding_result['embedding'], |
| 962 |
'preview' => $clean_preview, |
| 963 |
]; |
| 964 |
|
| 965 |
// Update metadata preview with clean content |
| 966 |
$post_metadata[ $item_id ]['preview'] = $clean_preview; |
| 967 |
} |
| 968 |
} |
| 969 |
|
| 970 |
// ================================================================= |
| 971 |
// STEP 2: Process posts WITHOUT images via batch endpoint |
| 972 |
// More efficient for text-only content |
| 973 |
// ================================================================= |
| 974 |
if ( ! empty( $items_to_embed ) ) { |
| 975 |
// Chunk items to max 100 per API call (API limit) |
| 976 |
$item_chunks = array_chunk( $items_to_embed, 100, true ); |
| 977 |
$api_call_index = 0; |
| 978 |
|
| 979 |
// Count unique topics that have text-only items (for credit calculation) |
| 980 |
$text_only_topic_ids = array_unique( array_map( function( $item ) use ( $post_metadata ) { |
| 981 |
return $post_metadata[ $item['id'] ]['topicid'] ?? 0; |
| 982 |
}, $items_to_embed ) ); |
| 983 |
|
| 984 |
// Exclude topics that already had image posts processed (they were already charged) |
| 985 |
$image_topic_ids = array_unique( array_map( function( $item ) use ( $post_metadata ) { |
| 986 |
return $post_metadata[ $item['id'] ]['topicid'] ?? 0; |
| 987 |
}, $items_with_images ) ); |
| 988 |
$text_only_topic_ids = array_diff( $text_only_topic_ids, $image_topic_ids ); |
| 989 |
$actual_new_topics_count = count( $text_only_topic_ids ); |
| 990 |
|
| 991 |
foreach ( $item_chunks as $chunk ) { |
| 992 |
// For credit charging: only charge topic_count on first chunk to avoid double-charging |
| 993 |
// Subsequent chunks are "free" since they're part of the same topic batch |
| 994 |
$chunk_topic_count = ( $api_call_index === 0 ) ? $actual_new_topics_count : 0; |
| 995 |
|
| 996 |
// DEBUG: Log what we're passing |
| 997 |
\wpforo_ai_log( 'debug', sprintf( |
| 998 |
'index_topics_batch_local chunk %d: batch_topics=%d, new_topics=%d, chunk_items=%d, chunk_topic_count=%d', |
| 999 |
$api_call_index, |
| 1000 |
count( $topic_ids ), |
| 1001 |
$actual_new_topics_count, |
| 1002 |
count( $chunk ), |
| 1003 |
$chunk_topic_count |
| 1004 |
), 'VectorStorage' ); |
| 1005 |
|
| 1006 |
$response = $ai_client->generate_embeddings_batch( array_values( $chunk ), $chunk_topic_count ); |
| 1007 |
|
| 1008 |
if ( is_wp_error( $response ) ) { |
| 1009 |
$errors[] = sprintf( 'Chunk %d failed: %s', $api_call_index + 1, $response->get_error_message() ); |
| 1010 |
} elseif ( ! empty( $response['results'] ) ) { |
| 1011 |
$all_results = array_merge( $all_results, $response['results'] ); |
| 1012 |
// Accumulate credits from each successful API call |
| 1013 |
$credits_from_api = (int) ( $response['credits_used'] ?? 0 ); |
| 1014 |
$total_credits_used += $credits_from_api; |
| 1015 |
|
| 1016 |
\wpforo_ai_log( 'debug', sprintf( |
| 1017 |
'API response chunk %d: credits_used=%d, successful=%d, failed=%d', |
| 1018 |
$api_call_index, |
| 1019 |
$credits_from_api, |
| 1020 |
$response['successful_items'] ?? 0, |
| 1021 |
$response['failed_items'] ?? 0 |
| 1022 |
), 'VectorStorage' ); |
| 1023 |
} |
| 1024 |
|
| 1025 |
$api_call_index++; |
| 1026 |
} |
| 1027 |
} |
| 1028 |
|
| 1029 |
// If all processing failed, return error |
| 1030 |
$total_items = count( $items_to_embed ) + count( $items_with_images ); |
| 1031 |
if ( empty( $all_results ) && ! empty( $errors ) ) { |
| 1032 |
// Log error to AILogs database |
| 1033 |
if ( isset( WPF()->ai_logs ) ) { |
| 1034 |
$user_type = AILogs::USER_TYPE_USER; |
| 1035 |
if ( defined( 'DOING_CRON' ) && DOING_CRON ) { |
| 1036 |
$user_type = AILogs::USER_TYPE_CRON; |
| 1037 |
} elseif ( ! get_current_user_id() ) { |
| 1038 |
$user_type = AILogs::USER_TYPE_SYSTEM; |
| 1039 |
} |
| 1040 |
|
| 1041 |
WPF()->ai_logs->log( [ |
| 1042 |
'action_type' => AILogs::ACTION_CONTENT_INDEXING, |
| 1043 |
'user_type' => $user_type, |
| 1044 |
'credits_used' => 0, |
| 1045 |
'status' => AILogs::STATUS_ERROR, |
| 1046 |
'content_type' => 'topic', |
| 1047 |
'request_summary' => sprintf( 'Local indexing: %d topics', count( $topic_ids ) ), |
| 1048 |
'error_message' => implode( '; ', array_slice( $errors, 0, 5 ) ), // First 5 errors |
| 1049 |
'extra_data' => wp_json_encode( [ |
| 1050 |
'storage_mode' => 'local', |
| 1051 |
'topic_ids' => array_slice( $topic_ids, 0, 20 ), |
| 1052 |
'errors_count' => count( $errors ), |
| 1053 |
] ), |
| 1054 |
] ); |
| 1055 |
} |
| 1056 |
|
| 1057 |
return [ |
| 1058 |
'success' => false, |
| 1059 |
'indexed_count' => 0, |
| 1060 |
'skipped_count' => $skipped_count, |
| 1061 |
'noise_filtered_count' => $noise_filtered_count, |
| 1062 |
'total_posts' => $total_items + $skipped_count + $noise_filtered_count, |
| 1063 |
'errors' => $errors, |
| 1064 |
]; |
| 1065 |
} |
| 1066 |
|
| 1067 |
// Store all embeddings locally |
| 1068 |
$indexed_count = 0; |
| 1069 |
|
| 1070 |
// Track which topics had posts successfully indexed |
| 1071 |
$topics_indexed = []; |
| 1072 |
|
| 1073 |
foreach ( $all_results as $result ) { |
| 1074 |
$item_id = $result['id']; |
| 1075 |
|
| 1076 |
if ( ! $result['success'] || empty( $result['embedding'] ) ) { |
| 1077 |
$errors[] = sprintf( 'Failed to embed %s: %s', $item_id, $result['error'] ?? 'Unknown error' ); |
| 1078 |
continue; |
| 1079 |
} |
| 1080 |
|
| 1081 |
$meta = $post_metadata[ $item_id ] ?? null; |
| 1082 |
if ( ! $meta ) { |
| 1083 |
continue; |
| 1084 |
} |
| 1085 |
|
| 1086 |
$stored = $local->store_embedding( |
| 1087 |
$meta['topicid'], |
| 1088 |
$meta['postid'], |
| 1089 |
$meta['forumid'], |
| 1090 |
$meta['userid'], |
| 1091 |
$result['embedding'], |
| 1092 |
$meta['content_hash'], |
| 1093 |
$meta['preview'] |
| 1094 |
); |
| 1095 |
|
| 1096 |
if ( $stored ) { |
| 1097 |
$indexed_count++; |
| 1098 |
// Track this topic as having indexed posts |
| 1099 |
$topics_indexed[ $meta['topicid'] ] = true; |
| 1100 |
} |
| 1101 |
} |
| 1102 |
|
| 1103 |
// Update indexed hash in wpforo_topics table for all topics with indexed content |
| 1104 |
// This includes both newly indexed topics AND topics with existing embeddings (skipped) |
| 1105 |
// Needed for: |
| 1106 |
// 1. Statistics to show correct indexed topic count |
| 1107 |
// 2. Summarization feature to know which topics have indexed content |
| 1108 |
// 3. Deduplication - to skip unchanged topics on re-indexing |
| 1109 |
$all_topics_with_content = array_keys( $topics_indexed + $topics_with_embeddings ); |
| 1110 |
if ( ! empty( $all_topics_with_content ) ) { |
| 1111 |
$this->update_topics_indexed_hash( $all_topics_with_content ); |
| 1112 |
} |
| 1113 |
|
| 1114 |
$response = [ |
| 1115 |
'success' => true, |
| 1116 |
'indexed_count' => $indexed_count, |
| 1117 |
'skipped_count' => $skipped_count, |
| 1118 |
'noise_filtered_count' => $noise_filtered_count, |
| 1119 |
'total_posts' => $total_items + $skipped_count + $noise_filtered_count, |
| 1120 |
'credits_used' => $total_credits_used > 0 ? $total_credits_used : count( $topic_ids ), |
| 1121 |
'errors' => $errors, |
| 1122 |
]; |
| 1123 |
|
| 1124 |
// Add image processing stats if images were processed |
| 1125 |
if ( $images_processed > 0 ) { |
| 1126 |
$response['images_processed'] = $images_processed; |
| 1127 |
} |
| 1128 |
|
| 1129 |
// Add document processing stats if documents were processed |
| 1130 |
if ( $documents_processed > 0 ) { |
| 1131 |
$response['documents_processed'] = $documents_processed; |
| 1132 |
} |
| 1133 |
|
| 1134 |
// Log to AILogs database for tracking |
| 1135 |
if ( $indexed_count > 0 && isset( WPF()->ai_logs ) ) { |
| 1136 |
$user_type = AILogs::USER_TYPE_USER; |
| 1137 |
if ( defined( 'DOING_CRON' ) && DOING_CRON ) { |
| 1138 |
$user_type = AILogs::USER_TYPE_CRON; |
| 1139 |
} elseif ( ! get_current_user_id() ) { |
| 1140 |
$user_type = AILogs::USER_TYPE_SYSTEM; |
| 1141 |
} |
| 1142 |
|
| 1143 |
WPF()->ai_logs->log( [ |
| 1144 |
'action_type' => AILogs::ACTION_CONTENT_INDEXING, |
| 1145 |
'user_type' => $user_type, |
| 1146 |
'credits_used' => $response['credits_used'] ?? 0, |
| 1147 |
'status' => AILogs::STATUS_SUCCESS, |
| 1148 |
'content_type' => 'topic', |
| 1149 |
'request_summary' => sprintf( 'Local indexing: %d topics', count( $topic_ids ) ), |
| 1150 |
'response_summary' => sprintf( |
| 1151 |
'Indexed %d posts from %d topics (%d skipped)', |
| 1152 |
$indexed_count, |
| 1153 |
count( $all_topics_with_content ), |
| 1154 |
$skipped_count |
| 1155 |
), |
| 1156 |
'extra_data' => wp_json_encode( [ |
| 1157 |
'storage_mode' => 'local', |
| 1158 |
'topic_ids' => array_slice( $topic_ids, 0, 20 ), // Limit to first 20 for log size |
| 1159 |
'topics_indexed' => count( $all_topics_with_content ), |
| 1160 |
'posts_indexed' => $indexed_count, |
| 1161 |
'skipped' => $skipped_count, |
| 1162 |
'errors_count' => count( $errors ), |
| 1163 |
] ), |
| 1164 |
] ); |
| 1165 |
} |
| 1166 |
|
| 1167 |
return $response; |
| 1168 |
} |
| 1169 |
|
| 1170 |
/** |
| 1171 |
* Prepare content for embedding |
| 1172 |
* |
| 1173 |
* Combines post content with relevant metadata for better semantic matching. |
| 1174 |
* |
| 1175 |
* @param array $post Post data |
| 1176 |
* @param array $topic Topic data |
| 1177 |
* @return string Prepared content |
| 1178 |
*/ |
| 1179 |
/** |
| 1180 |
* Build a clean content preview for search result display |
| 1181 |
* |
| 1182 |
* Strips Lambda processing markers (--- Document Content ---, [DOCUMENT: ...], [IMAGE: ...]) |
| 1183 |
* and appends a human-friendly document/image summary. |
| 1184 |
* |
| 1185 |
* @param string $content Original post content (before Lambda processing) |
| 1186 |
* @param array|null $embedding_result Full Lambda response (with document_processing/image_processing stats) |
| 1187 |
* @param int $max_length Maximum preview length |
| 1188 |
* @return string Clean preview text |
| 1189 |
*/ |
| 1190 |
private function build_content_preview( $content, $embedding_result = null, $max_length = 1000 ) { |
| 1191 |
// Start with clean post text (no Lambda markers or shortcodes) |
| 1192 |
$preview = strip_tags( $content ); |
| 1193 |
// Strip "Topic: Title" prefix — embedding content includes topic title for better |
| 1194 |
// vector similarity, but previews should show only the post body text. |
| 1195 |
$preview = preg_replace( '/^Topic:\s*[^\n]*\n*/i', '', $preview ); |
| 1196 |
// Strip trailing "Topic: Title" (first posts repeat title at end for embedding weight) |
| 1197 |
$preview = preg_replace( '/\n*\s*Topic:\s*[^\n]*$/i', '', $preview ); |
| 1198 |
// Strip enrichment tags added for embedding quality: [FORUM: name], [SOLVED], [BEST ANSWER] |
| 1199 |
$preview = preg_replace( '/\[(?:FORUM|SOLVED|BEST ANSWER)[^\]]*\]/', '', $preview ); |
| 1200 |
// Strip wpForo shortcodes: [attach]N[/attach], [attach]N,M[/attach] |
| 1201 |
$preview = preg_replace( '/\[attach\]\d+(?:,\d+)?\[\/attach\]/', '', $preview ); |
| 1202 |
// Strip any remaining shortcode-like patterns |
| 1203 |
$preview = preg_replace( '/\[(?:\/)?[a-zA-Z0-9_-]+(?:\s[^\]]*?)?\]/', '', $preview ); |
| 1204 |
$preview = preg_replace( '/\s+/', ' ', $preview ); |
| 1205 |
$preview = trim( $preview ); |
| 1206 |
|
| 1207 |
// Build document/image summary suffix |
| 1208 |
$attachments = []; |
| 1209 |
|
| 1210 |
if ( ! empty( $embedding_result['document_processing']['documents_processed'] ) ) { |
| 1211 |
$doc_count = (int) $embedding_result['document_processing']['documents_processed']; |
| 1212 |
$page_count = (int) ( $embedding_result['document_processing']['total_pages'] ?? 0 ); |
| 1213 |
$total_chunks = (int) ( $embedding_result['total_chunks'] ?? 1 ); |
| 1214 |
|
| 1215 |
if ( $page_count > 0 ) { |
| 1216 |
// If content was split into multiple chunks, we only store chunk 0 |
| 1217 |
// so not all pages are fully represented — show "X+ pages" |
| 1218 |
if ( $total_chunks > 1 ) { |
| 1219 |
$attachments[] = sprintf( '%d %s, %d+ %s', |
| 1220 |
$doc_count, |
| 1221 |
$doc_count === 1 ? 'document' : 'documents', |
| 1222 |
$page_count, |
| 1223 |
$page_count === 1 ? 'page' : 'pages' |
| 1224 |
); |
| 1225 |
} else { |
| 1226 |
$attachments[] = sprintf( '%d %s, %d %s', |
| 1227 |
$doc_count, |
| 1228 |
$doc_count === 1 ? 'document' : 'documents', |
| 1229 |
$page_count, |
| 1230 |
$page_count === 1 ? 'page' : 'pages' |
| 1231 |
); |
| 1232 |
} |
| 1233 |
} else { |
| 1234 |
$attachments[] = sprintf( '%d %s', |
| 1235 |
$doc_count, |
| 1236 |
$doc_count === 1 ? 'document' : 'documents' |
| 1237 |
); |
| 1238 |
} |
| 1239 |
} |
| 1240 |
|
| 1241 |
if ( ! empty( $embedding_result['image_processing']['images_processed'] ) ) { |
| 1242 |
$img_count = (int) $embedding_result['image_processing']['images_processed']; |
| 1243 |
$attachments[] = sprintf( '%d %s', |
| 1244 |
$img_count, |
| 1245 |
$img_count === 1 ? 'image' : 'images' |
| 1246 |
); |
| 1247 |
} |
| 1248 |
|
| 1249 |
// Append attachment summary |
| 1250 |
if ( ! empty( $attachments ) ) { |
| 1251 |
$suffix = ' [+ ' . implode( ', ', $attachments ) . ']'; |
| 1252 |
$preview = mb_substr( $preview, 0, $max_length - mb_strlen( $suffix ) ) . $suffix; |
| 1253 |
} else { |
| 1254 |
$preview = mb_substr( $preview, 0, $max_length ); |
| 1255 |
} |
| 1256 |
|
| 1257 |
return $preview; |
| 1258 |
} |
| 1259 |
|
| 1260 |
/** |
| 1261 |
* Noise filtering patterns - matches cloud mode (rag_ingestion/handler.py) |
| 1262 |
* These patterns identify non-informative content that shouldn't be indexed. |
| 1263 |
*/ |
| 1264 |
private static $gratitude_patterns = [ |
| 1265 |
'/^thanks?\.?!?$/i', |
| 1266 |
'/^thank you\.?!?$/i', |
| 1267 |
'/^thx\.?!?$/i', |
| 1268 |
'/^ty\.?!?$/i', |
| 1269 |
'/^thanks?,?\s*(a lot|so much|for .{0,30})\.?!?$/i', |
| 1270 |
'/^(great|awesome|perfect|helpful|nice|good),?\s*thanks?\.?!?$/i', |
| 1271 |
'/^thanks?,?\s*(it works|that worked|that fixed it|that helped)\.?!?$/i', |
| 1272 |
]; |
| 1273 |
|
| 1274 |
private static $agreement_patterns = [ |
| 1275 |
'/^i agree\.?!?$/i', |
| 1276 |
'/^same here\.?!?$/i', |
| 1277 |
'/^me too\.?!?$/i', |
| 1278 |
'/^same (problem|issue)\.?!?$/i', |
| 1279 |
'/^\+1\.?$/i', |
| 1280 |
'/^this\.?!?$/i', |
| 1281 |
'/^exactly\.?!?$/i', |
| 1282 |
'/^(yes|no|correct|right|true|indeed|absolutely|definitely)\.?!?$/i', |
| 1283 |
]; |
| 1284 |
|
| 1285 |
private static $bump_patterns = [ |
| 1286 |
'/^bump\.?!?$/i', |
| 1287 |
'/^following\.?$/i', |
| 1288 |
'/^subscribing\.?$/i', |
| 1289 |
'/^watching\.?$/i', |
| 1290 |
'/^any ?one\??$/i', |
| 1291 |
'/^any updates?\??$/i', |
| 1292 |
'/^hello\??$/i', |
| 1293 |
'/^still waiting\.?$/i', |
| 1294 |
'/^\?\?\?+$/i', |
| 1295 |
]; |
| 1296 |
|
| 1297 |
private static $noise_phrases = [ 'ok', 'okay', 'k', 'up' ]; |
| 1298 |
|
| 1299 |
/** Minimum character count for indexing */ |
| 1300 |
const MIN_CHAR_COUNT = 20; |
| 1301 |
|
| 1302 |
/** Minimum word count for indexing */ |
| 1303 |
const MIN_WORD_COUNT = 5; |
| 1304 |
|
| 1305 |
/** Cached locale check result for performance */ |
| 1306 |
private static $is_english_locale = null; |
| 1307 |
|
| 1308 |
/** |
| 1309 |
* Check if the current locale is English-like. |
| 1310 |
* Cached for performance since it doesn't change during a request. |
| 1311 |
* |
| 1312 |
* @return bool True if locale is English (en, en_US, en_GB, etc.) |
| 1313 |
*/ |
| 1314 |
private function is_english_locale() { |
| 1315 |
if ( self::$is_english_locale === null ) { |
| 1316 |
$locale = get_locale(); |
| 1317 |
// Match en, en_US, en_GB, en_AU, etc. |
| 1318 |
self::$is_english_locale = ( $locale === 'en' || strpos( $locale, 'en_' ) === 0 ); |
| 1319 |
} |
| 1320 |
return self::$is_english_locale; |
| 1321 |
} |
| 1322 |
|
| 1323 |
/** |
| 1324 |
* Check if a post is noise content that shouldn't be indexed. |
| 1325 |
* |
| 1326 |
* Filters out non-informative content to improve search quality: |
| 1327 |
* - Length < 20 characters (all languages) |
| 1328 |
* - Word count < 5 words (all languages) |
| 1329 |
* - Only emojis (no alphanumeric text) (all languages) |
| 1330 |
* - Gratitude-only posts ("thanks", etc.) (English only) |
| 1331 |
* - Agreement-only posts ("+1", "I agree", etc.) (English only) |
| 1332 |
* - Bump/follow posts ("bump", "following", etc.) (English only) |
| 1333 |
* - Common noise phrases ("ok", etc.) (English only) |
| 1334 |
* |
| 1335 |
* Pattern matching is only applied for English locales to avoid |
| 1336 |
* wasting resources on non-English forums where patterns won't match. |
| 1337 |
* |
| 1338 |
* Matches cloud mode behavior (rag_ingestion/handler.py:is_noise_reply). |
| 1339 |
* |
| 1340 |
* @param string $body Post body content (HTML allowed, will be stripped) |
| 1341 |
* @return bool True if post is noise and should be skipped, false if it should be indexed |
| 1342 |
*/ |
| 1343 |
public function is_noise_post( $body ) { |
| 1344 |
// Strip HTML and normalize whitespace |
| 1345 |
$cleaned = wp_strip_all_tags( $body ); |
| 1346 |
$cleaned = preg_replace( '/\s+/', ' ', $cleaned ); |
| 1347 |
$cleaned = trim( $cleaned ); |
| 1348 |
|
| 1349 |
// Rule 1: Minimum character count (language-agnostic) |
| 1350 |
if ( mb_strlen( $cleaned ) < self::MIN_CHAR_COUNT ) { |
| 1351 |
return true; |
| 1352 |
} |
| 1353 |
|
| 1354 |
// Rule 2: Minimum word count (language-agnostic) |
| 1355 |
$words = preg_split( '/\s+/', $cleaned, -1, PREG_SPLIT_NO_EMPTY ); |
| 1356 |
if ( count( $words ) < self::MIN_WORD_COUNT ) { |
| 1357 |
return true; |
| 1358 |
} |
| 1359 |
|
| 1360 |
// Rule 3: Check if only emojis/symbols (no alphanumeric chars) (language-agnostic) |
| 1361 |
if ( ! preg_match( '/[a-zA-Z0-9]/', $cleaned ) ) { |
| 1362 |
return true; |
| 1363 |
} |
| 1364 |
|
| 1365 |
// Pattern matching only for English locales - skip for other languages |
| 1366 |
// to avoid wasting CPU cycles on patterns that won't match |
| 1367 |
if ( ! $this->is_english_locale() ) { |
| 1368 |
return false; |
| 1369 |
} |
| 1370 |
|
| 1371 |
// Prepare for pattern matching (English only from here) |
| 1372 |
$lower_text = mb_strtolower( $cleaned ); |
| 1373 |
|
| 1374 |
// Rule 4: Gratitude-only detection |
| 1375 |
foreach ( self::$gratitude_patterns as $pattern ) { |
| 1376 |
if ( preg_match( $pattern, $lower_text ) ) { |
| 1377 |
return true; |
| 1378 |
} |
| 1379 |
} |
| 1380 |
|
| 1381 |
// Rule 5: Agreement-only detection |
| 1382 |
foreach ( self::$agreement_patterns as $pattern ) { |
| 1383 |
if ( preg_match( $pattern, $lower_text ) ) { |
| 1384 |
return true; |
| 1385 |
} |
| 1386 |
} |
| 1387 |
|
| 1388 |
// Rule 6: Bump/follow detection |
| 1389 |
foreach ( self::$bump_patterns as $pattern ) { |
| 1390 |
if ( preg_match( $pattern, $lower_text ) ) { |
| 1391 |
return true; |
| 1392 |
} |
| 1393 |
} |
| 1394 |
|
| 1395 |
// Rule 7: Legacy noise phrases (exact match) |
| 1396 |
if ( in_array( $lower_text, self::$noise_phrases, true ) ) { |
| 1397 |
return true; |
| 1398 |
} |
| 1399 |
|
| 1400 |
// Passed all filters - keep this post |
| 1401 |
return false; |
| 1402 |
} |
| 1403 |
|
| 1404 |
/** |
| 1405 |
* Compute a stable content hash for embedding deduplication. |
| 1406 |
* |
| 1407 |
* Uses raw content inputs (body text, topic title, attachment counts) rather than |
| 1408 |
* the formatted output of prepare_content_for_embedding(). This ensures that changes |
| 1409 |
* to embedding enrichment formatting (e.g., adding [FORUM:], [SOLVED], [BEST ANSWER] |
| 1410 |
* tags or topic title prefixes) don't invalidate all existing hashes and trigger |
| 1411 |
* unnecessary re-indexing that wastes credits. |
| 1412 |
* |
| 1413 |
* @param array $post Post data with 'body' key |
| 1414 |
* @param array $topic Topic data with 'title' key |
| 1415 |
* @param int $images_count Number of images in the post |
| 1416 |
* @param int $docs_count Number of documents in the post |
| 1417 |
* @return string MD5 hash |
| 1418 |
*/ |
| 1419 |
private function compute_content_hash( $post, $topic, $images_count = 0, $docs_count = 0 ) { |
| 1420 |
$body = wp_strip_all_tags( strip_shortcodes( $post['body'] ?? '' ) ); |
| 1421 |
$body = preg_replace( '/\[(?:\/)?[a-zA-Z0-9_-]+(?:\s[^\]]*?)?\]/', '', $body ); |
| 1422 |
$body = html_entity_decode( $body, ENT_QUOTES, 'UTF-8' ); |
| 1423 |
$body = preg_replace( '/\s+/', ' ', trim( $body ) ); |
| 1424 |
|
| 1425 |
$hash_input = ( $topic['title'] ?? '' ) . '|' . $body . '|images:' . $images_count . '|docs:' . $docs_count; |
| 1426 |
|
| 1427 |
return md5( $hash_input ); |
| 1428 |
} |
| 1429 |
|
| 1430 |
/** |
| 1431 |
* Migrate content hashes from v1 (full prepared content) to v2 (raw inputs only). |
| 1432 |
* |
| 1433 |
* Runs once automatically on the first indexing call after the update. |
| 1434 |
* Updates existing hashes in the DB without re-generating embeddings, |
| 1435 |
* so there is no API call cost or credit usage. |
| 1436 |
*/ |
| 1437 |
public function maybe_migrate_content_hashes() { |
| 1438 |
if ( (int) get_option( 'wpforo_ai_hash_version', 0 ) >= 2 ) { |
| 1439 |
return; |
| 1440 |
} |
| 1441 |
|
| 1442 |
global $wpdb; |
| 1443 |
|
| 1444 |
$embeddings_table = WPF()->tables->ai_embeddings; |
| 1445 |
$posts_table = WPF()->tables->posts; |
| 1446 |
$topics_table = WPF()->tables->topics; |
| 1447 |
|
| 1448 |
$batch_size = 500; |
| 1449 |
$offset = 0; |
| 1450 |
$updated = 0; |
| 1451 |
|
| 1452 |
do { |
| 1453 |
$rows = $wpdb->get_results( $wpdb->prepare( |
| 1454 |
"SELECT e.id, p.body, t.title |
| 1455 |
FROM {$embeddings_table} e |
| 1456 |
INNER JOIN {$posts_table} p ON e.postid = p.postid |
| 1457 |
INNER JOIN {$topics_table} t ON p.topicid = t.topicid |
| 1458 |
WHERE e.content_type = 'forum' |
| 1459 |
LIMIT %d OFFSET %d", |
| 1460 |
$batch_size, |
| 1461 |
$offset |
| 1462 |
), ARRAY_A ); |
| 1463 |
|
| 1464 |
if ( empty( $rows ) ) { |
| 1465 |
break; |
| 1466 |
} |
| 1467 |
|
| 1468 |
foreach ( $rows as $row ) { |
| 1469 |
$body = wp_strip_all_tags( strip_shortcodes( $row['body'] ?? '' ) ); |
| 1470 |
$body = preg_replace( '/\[(?:\/)?[a-zA-Z0-9_-]+(?:\s[^\]]*?)?\]/', '', $body ); |
| 1471 |
$body = html_entity_decode( $body, ENT_QUOTES, 'UTF-8' ); |
| 1472 |
$body = preg_replace( '/\s+/', ' ', trim( $body ) ); |
| 1473 |
|
| 1474 |
// Use 0 for image/doc counts — posts with attachments will be |
| 1475 |
// re-indexed once when their actual counts are computed. |
| 1476 |
$hash_input = ( $row['title'] ?? '' ) . '|' . $body . '|images:0|docs:0'; |
| 1477 |
$new_hash = md5( $hash_input ); |
| 1478 |
|
| 1479 |
$wpdb->update( |
| 1480 |
$embeddings_table, |
| 1481 |
[ 'content_hash' => $new_hash ], |
| 1482 |
[ 'id' => $row['id'] ], |
| 1483 |
[ '%s' ], |
| 1484 |
[ '%d' ] |
| 1485 |
); |
| 1486 |
$updated++; |
| 1487 |
} |
| 1488 |
|
| 1489 |
$offset += $batch_size; |
| 1490 |
} while ( count( $rows ) === $batch_size ); |
| 1491 |
|
| 1492 |
update_option( 'wpforo_ai_hash_version', 2, false ); |
| 1493 |
} |
| 1494 |
|
| 1495 |
public function prepare_content_for_embedding( $post, $topic ) { |
| 1496 |
// Max content length for embedding API (Titan Embed v2 handles ~32K chars / 8K tokens) |
| 1497 |
$max_content_length = 45000; |
| 1498 |
|
| 1499 |
$parts = []; |
| 1500 |
|
| 1501 |
// Add topic title for context on ALL posts (not just first post). |
| 1502 |
// This matches cloud mode behavior where every chunk gets [TOPIC: title] prefix. |
| 1503 |
// Without this, reply embeddings lack topic context and can't match topic-specific |
| 1504 |
// search terms (e.g., searching "postgres" won't find replies in a PostgreSQL topic). |
| 1505 |
// The title is stripped from content_preview by build_content_preview() so it |
| 1506 |
// won't appear in search result excerpts. |
| 1507 |
if ( ! empty( $topic['title'] ) ) { |
| 1508 |
$parts[] = 'Topic: ' . $topic['title']; |
| 1509 |
} |
| 1510 |
|
| 1511 |
// Contextual enrichment tags (Phase 2.1) — matches cloud mode chunk prefixes. |
| 1512 |
// These tags improve embedding quality by adding forum, solved, and best answer context. |
| 1513 |
// They are stripped from display by build_content_preview() via generic shortcode pattern. |
| 1514 |
if ( ! empty( $topic['forum_name'] ) ) { |
| 1515 |
$parts[] = '[FORUM: ' . $topic['forum_name'] . ']'; |
| 1516 |
} |
| 1517 |
if ( ! empty( $topic['solved'] ) ) { |
| 1518 |
$parts[] = '[SOLVED]'; |
| 1519 |
} |
| 1520 |
if ( ! empty( $post['is_answer'] ) ) { |
| 1521 |
$parts[] = '[BEST ANSWER]'; |
| 1522 |
} |
| 1523 |
|
| 1524 |
// Add post content (cleaned) |
| 1525 |
$content = $post['body'] ?? ''; |
| 1526 |
// Strip shortcodes (both WordPress and wpForo shortcodes like [attach]ID[/attach]) |
| 1527 |
$content = strip_shortcodes( $content ); |
| 1528 |
$content = preg_replace( '/\[(?:\/)?[a-zA-Z0-9_-]+(?:\s[^\]]*?)?\]/', '', $content ); |
| 1529 |
$content = wp_strip_all_tags( $content ); |
| 1530 |
$content = html_entity_decode( $content, ENT_QUOTES, 'UTF-8' ); |
| 1531 |
$content = preg_replace( '/\s+/', ' ', $content ); |
| 1532 |
$parts[] = trim( $content ); |
| 1533 |
|
| 1534 |
// Add tags if available (first post only) |
| 1535 |
if ( isset( $post['is_first_post'] ) && $post['is_first_post'] && ! empty( $topic['tags'] ) ) { |
| 1536 |
$tags = is_array( $topic['tags'] ) ? implode( ', ', $topic['tags'] ) : $topic['tags']; |
| 1537 |
$parts[] = 'Tags: ' . $tags; |
| 1538 |
} |
| 1539 |
|
| 1540 |
// Repeat title at the end to increase its weight in the embedding. |
| 1541 |
// This improves title-based search (topic suggestions) where users type |
| 1542 |
// a short title and need to match against long content embeddings. |
| 1543 |
if ( isset( $post['is_first_post'] ) && $post['is_first_post'] && ! empty( $topic['title'] ) ) { |
| 1544 |
$parts[] = 'Topic: ' . $topic['title']; |
| 1545 |
} |
| 1546 |
|
| 1547 |
$result = implode( "\n\n", array_filter( $parts ) ); |
| 1548 |
|
| 1549 |
// Truncate to max length if needed (embedding model and API limits) |
| 1550 |
if ( mb_strlen( $result ) > $max_content_length ) { |
| 1551 |
$result = mb_substr( $result, 0, $max_content_length ); |
| 1552 |
} |
| 1553 |
|
| 1554 |
return $result; |
| 1555 |
} |
| 1556 |
|
| 1557 |
/** |
| 1558 |
* Update indexed hash for topics after successful local indexing |
| 1559 |
* |
| 1560 |
* The indexed hash is an MD5 of "topicid_postcount" which changes when posts are added/removed. |
| 1561 |
* This enables: |
| 1562 |
* 1. Statistics to show correct indexed topic count |
| 1563 |
* 2. Summarization feature to use indexed content |
| 1564 |
* 3. Deduplication - skip unchanged topics on re-indexing |
| 1565 |
* |
| 1566 |
* @param array $topic_ids Array of topic IDs that were indexed |
| 1567 |
* @return int Number of topics updated |
| 1568 |
*/ |
| 1569 |
private function update_topics_indexed_hash( $topic_ids ) { |
| 1570 |
if ( empty( $topic_ids ) ) { |
| 1571 |
return 0; |
| 1572 |
} |
| 1573 |
|
| 1574 |
global $wpdb; |
| 1575 |
|
| 1576 |
// Build CASE statement for indexed hash updates |
| 1577 |
// Hash is MD5 of "topicid_postcount" |
| 1578 |
$topic_ids = array_map( 'intval', $topic_ids ); |
| 1579 |
$ids_list = implode( ',', $topic_ids ); |
| 1580 |
|
| 1581 |
// Update indexed hash and local column in a single query |
| 1582 |
// The hash is calculated as MD5(topicid + '_' + posts) |
| 1583 |
$updated = $wpdb->query( |
| 1584 |
"UPDATE `" . WPF()->tables->topics . "` |
| 1585 |
SET `indexed` = MD5(CONCAT(topicid, '_', posts)), |
| 1586 |
`local` = 1 |
| 1587 |
WHERE topicid IN ($ids_list)" |
| 1588 |
); |
| 1589 |
|
| 1590 |
if ( $updated > 0 ) { |
| 1591 |
// Clear topic cache to reflect indexed status |
| 1592 |
wpforo_clean_cache( 'topic' ); |
| 1593 |
} |
| 1594 |
|
| 1595 |
return (int) $updated; |
| 1596 |
} |
| 1597 |
|
| 1598 |
/** |
| 1599 |
* Generate embedding vector for content |
| 1600 |
* |
| 1601 |
* Uses the cloud API to generate embeddings. |
| 1602 |
* This is used for both local and cloud storage modes. |
| 1603 |
* |
| 1604 |
* Supports multimodal image indexing (Professional+ plans): |
| 1605 |
* - Pass images array with URLs from site domain |
| 1606 |
* - Images are processed by vision models on the backend |
| 1607 |
* - Returns full response including processed_content if images were processed |
| 1608 |
* |
| 1609 |
* Supports document indexing (Professional+ plans): |
| 1610 |
* - Pass documents array with URLs from site domain |
| 1611 |
* - Documents are processed (text extraction, OCR) on the backend |
| 1612 |
* |
| 1613 |
* @param string $content Content to embed |
| 1614 |
* @param array $images Optional. Array of image data for multimodal indexing |
| 1615 |
* @param string $topic_context Optional. Topic title for better image/document descriptions |
| 1616 |
* @param bool $full_response Optional. Return full response instead of just embedding |
| 1617 |
* @param array $documents Optional. Array of document data for document indexing |
| 1618 |
* @return array|WP_Error Vector array (default) or full response array if $full_response=true |
| 1619 |
*/ |
| 1620 |
public function generate_embedding( $content, $images = [], $topic_context = '', $full_response = false, $documents = [] ) { |
| 1621 |
$ai_client = $this->get_ai_client(); |
| 1622 |
|
| 1623 |
// Call cloud API to generate embedding (with optional images and documents) |
| 1624 |
$result = $ai_client->generate_embedding( $content, $images, $topic_context, $documents ); |
| 1625 |
|
| 1626 |
if ( is_wp_error( $result ) ) { |
| 1627 |
return $result; |
| 1628 |
} |
| 1629 |
|
| 1630 |
if ( ! isset( $result['embedding'] ) || ! is_array( $result['embedding'] ) ) { |
| 1631 |
return new \WP_Error( 'invalid_embedding', wpforo_phrase( 'Invalid embedding response from API', false ) ); |
| 1632 |
} |
| 1633 |
|
| 1634 |
// Return full response if requested (for image processing info) |
| 1635 |
if ( $full_response ) { |
| 1636 |
return $result; |
| 1637 |
} |
| 1638 |
|
| 1639 |
return $result['embedding']; |
| 1640 |
} |
| 1641 |
|
| 1642 |
/** |
| 1643 |
* Delete embeddings for a topic |
| 1644 |
* |
| 1645 |
* @param int $topicid Topic ID |
| 1646 |
* @return bool|WP_Error |
| 1647 |
*/ |
| 1648 |
public function delete_topic_embeddings( $topicid ) { |
| 1649 |
if ( $this->is_local_mode() ) { |
| 1650 |
$local = $this->get_local_storage(); |
| 1651 |
return $local->delete_topic_embeddings( $topicid ); |
| 1652 |
} else { |
| 1653 |
$ai_client = $this->get_ai_client(); |
| 1654 |
return $ai_client->delete_topic_from_index( $topicid, $this->board_id ); |
| 1655 |
} |
| 1656 |
} |
| 1657 |
|
| 1658 |
/** |
| 1659 |
* Delete embedding for a single post |
| 1660 |
* |
| 1661 |
* @param int $postid Post ID |
| 1662 |
* @return bool|WP_Error |
| 1663 |
*/ |
| 1664 |
public function delete_post_embedding( $postid ) { |
| 1665 |
if ( $this->is_local_mode() ) { |
| 1666 |
$local = $this->get_local_storage(); |
| 1667 |
return $local->delete_embedding( $postid ); |
| 1668 |
} else { |
| 1669 |
$ai_client = $this->get_ai_client(); |
| 1670 |
return $ai_client->delete_post_from_index( $postid, $this->board_id ); |
| 1671 |
} |
| 1672 |
} |
| 1673 |
|
| 1674 |
/** |
| 1675 |
* Clear all embeddings for current board |
| 1676 |
* |
| 1677 |
* @return bool|WP_Error |
| 1678 |
*/ |
| 1679 |
public function clear_all_embeddings() { |
| 1680 |
if ( $this->is_local_mode() ) { |
| 1681 |
global $wpdb; |
| 1682 |
$wpdb->query( "TRUNCATE TABLE " . WPF()->tables->ai_embeddings ); |
| 1683 |
$wpdb->query( "TRUNCATE TABLE " . WPF()->tables->ai_embeddings_cache ); |
| 1684 |
// Also clear indexed status and local column in topics table so topics can be re-indexed |
| 1685 |
WPF()->db->query( |
| 1686 |
"UPDATE `" . WPF()->tables->topics . "` SET `indexed` = NULL, `local` = 0 WHERE `indexed` IS NOT NULL OR `local` = 1" |
| 1687 |
); |
| 1688 |
wpforo_clean_cache( 'topic' ); |
| 1689 |
$this->clear_indexing_stats_cache(); |
| 1690 |
return true; |
| 1691 |
} else { |
| 1692 |
$ai_client = $this->get_ai_client(); |
| 1693 |
$result = $ai_client->clear_rag_database( $this->board_id ); |
| 1694 |
$this->clear_indexing_stats_cache(); |
| 1695 |
return $result; |
| 1696 |
} |
| 1697 |
} |
| 1698 |
|
| 1699 |
/** |
| 1700 |
* Clear embeddings for specific forums in current storage mode. |
| 1701 |
* |
| 1702 |
* @param array $forum_ids Array of forum IDs to clear |
| 1703 |
* @return array|WP_Error Result with counts |
| 1704 |
*/ |
| 1705 |
public function clear_forum_embeddings( $forum_ids ) { |
| 1706 |
$forum_ids = array_values( array_filter( array_map( 'intval', (array) $forum_ids ), function( $id ) { return $id > 0; } ) ); |
| 1707 |
|
| 1708 |
if ( empty( $forum_ids ) ) { |
| 1709 |
return new \WP_Error( 'invalid_forums', __( 'No valid forum IDs provided.', 'wpforo' ) ); |
| 1710 |
} |
| 1711 |
|
| 1712 |
if ( $this->is_local_mode() ) { |
| 1713 |
return $this->clear_forum_embeddings_local( $forum_ids ); |
| 1714 |
} else { |
| 1715 |
return $this->clear_forum_embeddings_cloud( $forum_ids ); |
| 1716 |
} |
| 1717 |
} |
| 1718 |
|
| 1719 |
/** |
| 1720 |
* Clear forum embeddings in local storage mode. |
| 1721 |
* |
| 1722 |
* @param array $forum_ids Forum IDs to clear |
| 1723 |
* @return array Result with counts |
| 1724 |
*/ |
| 1725 |
private function clear_forum_embeddings_local( $forum_ids ) { |
| 1726 |
global $wpdb; |
| 1727 |
|
| 1728 |
$placeholders = implode( ',', array_fill( 0, count( $forum_ids ), '%d' ) ); |
| 1729 |
|
| 1730 |
$deleted = $wpdb->query( |
| 1731 |
$wpdb->prepare( |
| 1732 |
"DELETE FROM `" . WPF()->tables->ai_embeddings . "` WHERE `forumid` IN ({$placeholders})", |
| 1733 |
...$forum_ids |
| 1734 |
) |
| 1735 |
); |
| 1736 |
|
| 1737 |
$topics_cleared = $wpdb->query( |
| 1738 |
$wpdb->prepare( |
| 1739 |
"UPDATE `" . WPF()->tables->topics . "` SET `local` = 0 WHERE `forumid` IN ({$placeholders}) AND `local` = 1", |
| 1740 |
...$forum_ids |
| 1741 |
) |
| 1742 |
); |
| 1743 |
|
| 1744 |
wpforo_clean_cache( 'topic' ); |
| 1745 |
$this->clear_indexing_stats_cache(); |
| 1746 |
|
| 1747 |
return [ |
| 1748 |
'deleted_embeddings' => (int) $deleted, |
| 1749 |
'topics_cleared' => (int) $topics_cleared, |
| 1750 |
'forums_cleared' => count( $forum_ids ), |
| 1751 |
]; |
| 1752 |
} |
| 1753 |
|
| 1754 |
/** |
| 1755 |
* Clear forum embeddings in cloud storage mode. |
| 1756 |
* |
| 1757 |
* @param array $forum_ids Forum IDs to clear |
| 1758 |
* @return array|WP_Error Result with counts |
| 1759 |
*/ |
| 1760 |
private function clear_forum_embeddings_cloud( $forum_ids ) { |
| 1761 |
global $wpdb; |
| 1762 |
|
| 1763 |
$ai_client = $this->get_ai_client(); |
| 1764 |
$result = $ai_client->clear_forum_index( $forum_ids, $this->board_id ); |
| 1765 |
|
| 1766 |
if ( is_wp_error( $result ) ) { |
| 1767 |
return $result; |
| 1768 |
} |
| 1769 |
|
| 1770 |
$placeholders = implode( ',', array_fill( 0, count( $forum_ids ), '%d' ) ); |
| 1771 |
$topics_cleared = $wpdb->query( |
| 1772 |
$wpdb->prepare( |
| 1773 |
"UPDATE `" . WPF()->tables->topics . "` SET `cloud` = 0 WHERE `forumid` IN ({$placeholders}) AND `cloud` != 0", |
| 1774 |
...$forum_ids |
| 1775 |
) |
| 1776 |
); |
| 1777 |
|
| 1778 |
wpforo_clean_cache( 'topic' ); |
| 1779 |
$this->clear_indexing_stats_cache(); |
| 1780 |
|
| 1781 |
$result['topics_cleared'] = (int) $topics_cleared; |
| 1782 |
return $result; |
| 1783 |
} |
| 1784 |
|
| 1785 |
/** |
| 1786 |
* Ingest multiple topics |
| 1787 |
* |
| 1788 |
* For local mode, indexes topics directly. |
| 1789 |
* For cloud mode, sends to cloud API. |
| 1790 |
* |
| 1791 |
* @param array $topic_ids Array of topic IDs |
| 1792 |
* @param int $chunk_size Chunk size (used for cloud mode) |
| 1793 |
* @param int $overlap_percent Overlap percentage (used for cloud mode) |
| 1794 |
* @return array|WP_Error Result array or error |
| 1795 |
*/ |
| 1796 |
public function ingest_topics( $topic_ids, $chunk_size = 512, $overlap_percent = 20 ) { |
| 1797 |
if ( $this->is_local_mode() ) { |
| 1798 |
return $this->ingest_topics_local( $topic_ids ); |
| 1799 |
} else { |
| 1800 |
return $this->ingest_topics_cloud( $topic_ids, $chunk_size, $overlap_percent ); |
| 1801 |
} |
| 1802 |
} |
| 1803 |
|
| 1804 |
/** |
| 1805 |
* Ingest topics locally via WP Cron batches |
| 1806 |
* |
| 1807 |
* Uses batch embedding API for efficiency - multiple topics processed |
| 1808 |
* in a single API call, matching cloud indexing pattern. |
| 1809 |
* |
| 1810 |
* @param array $topic_ids Topic IDs to index |
| 1811 |
* @return array Result with success status and scheduled counts |
| 1812 |
*/ |
| 1813 |
private function ingest_topics_local( $topic_ids ) { |
| 1814 |
$total_topics = count( $topic_ids ); |
| 1815 |
|
| 1816 |
if ( empty( $topic_ids ) ) { |
| 1817 |
return [ |
| 1818 |
'success' => true, |
| 1819 |
'topics_queued' => 0, |
| 1820 |
'message' => wpforo_phrase( 'No topics to index.', false ), |
| 1821 |
]; |
| 1822 |
} |
| 1823 |
|
| 1824 |
// Check available credits before starting |
| 1825 |
$ai_client = $this->get_ai_client(); |
| 1826 |
$status = $ai_client->get_tenant_status( true ); // Force fresh status |
| 1827 |
if ( is_wp_error( $status ) ) { |
| 1828 |
return $status; |
| 1829 |
} |
| 1830 |
|
| 1831 |
$credits_available = isset( $status['subscription']['credits_remaining'] ) |
| 1832 |
? (int) $status['subscription']['credits_remaining'] |
| 1833 |
: 0; |
| 1834 |
|
| 1835 |
if ( $credits_available <= 0 ) { |
| 1836 |
return new \WP_Error( |
| 1837 |
'no_credits', |
| 1838 |
wpforo_phrase( 'No credits available for indexing. Please wait for your monthly credit reset or upgrade your plan.', false ) |
| 1839 |
); |
| 1840 |
} |
| 1841 |
|
| 1842 |
// Use self-rescheduling queue pattern: |
| 1843 |
// - Store all topic IDs in a queue (option) |
| 1844 |
// - Schedule ONE cron job |
| 1845 |
// - Job processes a batch, then reschedules itself if more remain |
| 1846 |
// This avoids overwhelming WP Cron with hundreds of jobs |
| 1847 |
$queue_key = 'wpforo_ai_indexing_queue_' . $this->board_id; |
| 1848 |
|
| 1849 |
// Get existing queue and merge (in case of concurrent requests) |
| 1850 |
$existing_queue = get_option( $queue_key, [] ); |
| 1851 |
$merged_queue = array_unique( array_merge( $existing_queue, $topic_ids ) ); |
| 1852 |
update_option( $queue_key, $merged_queue, false ); // No autoload |
| 1853 |
|
| 1854 |
// Schedule ONE job to start processing (if not already scheduled) |
| 1855 |
$cron_hook = 'wpforo_ai_process_queue'; |
| 1856 |
$cron_args = [ $this->board_id ]; |
| 1857 |
|
| 1858 |
if ( ! wp_next_scheduled( $cron_hook, $cron_args ) ) { |
| 1859 |
wp_schedule_single_event( time() + 5, $cron_hook, $cron_args ); |
| 1860 |
} |
| 1861 |
|
| 1862 |
return [ |
| 1863 |
'success' => true, |
| 1864 |
'threads_indexed' => 0, // Will be processed async |
| 1865 |
'posts_indexed' => 0, |
| 1866 |
'posts_unchanged' => 0, |
| 1867 |
'errors' => [], |
| 1868 |
'credits_consumed' => 0, // Credits consumed during cron processing |
| 1869 |
'topics_queued' => $total_topics, |
| 1870 |
'batches_queued' => 1, // Always just 1 job that self-reschedules |
| 1871 |
'message' => sprintf( |
| 1872 |
wpforo_phrase( 'Indexing queued! %d topics will be processed in batches. Processing starts in 5 seconds.', false ), |
| 1873 |
$total_topics |
| 1874 |
), |
| 1875 |
'stats' => [ |
| 1876 |
'deduplication' => [ |
| 1877 |
'posts_unchanged' => 0, |
| 1878 |
], |
| 1879 |
], |
| 1880 |
]; |
| 1881 |
} |
| 1882 |
|
| 1883 |
/** |
| 1884 |
* Ingest topics to cloud storage via option-backed self-rescheduling queue |
| 1885 |
* |
| 1886 |
* Writes topic IDs to a single DynamoDB-style queue option (autoload=false) |
| 1887 |
* and schedules ONE WP Cron event that drains the queue in batches, |
| 1888 |
* rescheduling itself until empty. |
| 1889 |
* |
| 1890 |
* This replaces the previous "sync first batch + N scheduled events" pattern, |
| 1891 |
* which caused O(n²) autoloaded `cron` option rewrites and OOM on large forums. |
| 1892 |
* It mirrors the pattern already used by ingest_topics_local() and |
| 1893 |
* queue_topic_for_auto_indexing(), so the existing UI (progress polling, |
| 1894 |
* stop button, cleanup session) works unchanged — the queue key and cron |
| 1895 |
* hook are already handled by get_pending_cron_jobs() and |
| 1896 |
* clear_pending_cron_jobs() in AIClient. |
| 1897 |
* |
| 1898 |
* Credits are enforced per-batch by the backend; the cron handler |
| 1899 |
* (VectorStorageManager::cron_process_queue_mode) detects 402 responses |
| 1900 |
* and stops rescheduling automatically. |
| 1901 |
* |
| 1902 |
* @param array $topic_ids Array of topic IDs |
| 1903 |
* @param int $chunk_size Chunk size for text splitting in tokens (unused here — read from options by cron handler) |
| 1904 |
* @param int $overlap_percent Overlap percentage for chunking (unused here — read from options by cron handler) |
| 1905 |
* @return array|WP_Error Result array or error |
| 1906 |
*/ |
| 1907 |
private function ingest_topics_cloud( $topic_ids, $chunk_size = 512, $overlap_percent = 20 ) { |
| 1908 |
$total_topics = count( $topic_ids ); |
| 1909 |
|
| 1910 |
if ( empty( $topic_ids ) ) { |
| 1911 |
return [ |
| 1912 |
'success' => true, |
| 1913 |
'topics_queued' => 0, |
| 1914 |
'message' => wpforo_phrase( 'No topics to index.', false ), |
| 1915 |
]; |
| 1916 |
} |
| 1917 |
|
| 1918 |
$ai_client = $this->get_ai_client(); |
| 1919 |
|
| 1920 |
// Check if database clearing is in progress |
| 1921 |
if ( $ai_client->is_clearing_in_progress() ) { |
| 1922 |
$remaining = $ai_client->get_clearing_time_remaining(); |
| 1923 |
$minutes = ceil( $remaining / 60 ); |
| 1924 |
return new \WP_Error( |
| 1925 |
'clearing_in_progress', |
| 1926 |
sprintf( |
| 1927 |
wpforo_phrase( 'Database clearing is in progress. Please wait approximately %d minute(s) before starting new indexing.', false ), |
| 1928 |
$minutes |
| 1929 |
) |
| 1930 |
); |
| 1931 |
} |
| 1932 |
|
| 1933 |
// Check available credits before starting |
| 1934 |
$status = $ai_client->get_tenant_status( true ); |
| 1935 |
if ( is_wp_error( $status ) ) { |
| 1936 |
return $status; |
| 1937 |
} |
| 1938 |
|
| 1939 |
$credits_available = isset( $status['subscription']['credits_remaining'] ) |
| 1940 |
? (int) $status['subscription']['credits_remaining'] |
| 1941 |
: 0; |
| 1942 |
|
| 1943 |
if ( $credits_available <= 0 ) { |
| 1944 |
return new \WP_Error( |
| 1945 |
'no_credits', |
| 1946 |
wpforo_phrase( 'No credits available for indexing. Please wait for your monthly credit reset or upgrade your plan.', false ) |
| 1947 |
); |
| 1948 |
} |
| 1949 |
|
| 1950 |
// Limit topics to available credits (worker will also stop on 402 as a safety net) |
| 1951 |
$topics_limited = false; |
| 1952 |
$skipped_topics = 0; |
| 1953 |
if ( $total_topics > $credits_available ) { |
| 1954 |
$topic_ids = array_slice( $topic_ids, 0, $credits_available ); |
| 1955 |
$topics_limited = true; |
| 1956 |
$skipped_topics = $total_topics - $credits_available; |
| 1957 |
$total_topics = count( $topic_ids ); |
| 1958 |
} |
| 1959 |
|
| 1960 |
// Normalize to integers to keep the queue array small and consistent with |
| 1961 |
// queue_topic_for_auto_indexing() which stores ints. |
| 1962 |
$topic_ids = array_values( array_unique( array_map( 'intval', $topic_ids ) ) ); |
| 1963 |
|
| 1964 |
// Write to option-backed queue (autoload=false). Merge with any existing |
| 1965 |
// queue so concurrent ingest calls don't clobber each other and auto- |
| 1966 |
// indexing entries are preserved. |
| 1967 |
$queue_key = 'wpforo_ai_indexing_queue_cloud_' . $this->board_id; |
| 1968 |
$existing_queue = get_option( $queue_key, [] ); |
| 1969 |
if ( ! is_array( $existing_queue ) ) { |
| 1970 |
$existing_queue = []; |
| 1971 |
} |
| 1972 |
$merged_queue = array_values( array_unique( array_merge( $existing_queue, $topic_ids ) ) ); |
| 1973 |
update_option( $queue_key, $merged_queue, false ); |
| 1974 |
|
| 1975 |
// Schedule ONE self-rescheduling cron event. The handler |
| 1976 |
// (cron_process_queue_mode in this class, wired via AIClient::cron_process_queue_cloud) |
| 1977 |
// will drain the queue in pagination_size batches and reschedule itself |
| 1978 |
// until empty. If an event is already scheduled, we don't schedule a second — |
| 1979 |
// the running worker will simply pick up the new entries on its next batch. |
| 1980 |
$cron_hook = 'wpforo_ai_process_queue_cloud'; |
| 1981 |
$cron_args = [ $this->board_id ]; |
| 1982 |
if ( ! wp_next_scheduled( $cron_hook, $cron_args ) ) { |
| 1983 |
wp_schedule_single_event( time() + 5, $cron_hook, $cron_args ); |
| 1984 |
} |
| 1985 |
|
| 1986 |
// Clear cached status so the UI picks up "indexing in progress" immediately |
| 1987 |
$ai_client->clear_rag_status_cache(); |
| 1988 |
|
| 1989 |
if ( $topics_limited ) { |
| 1990 |
$message = sprintf( |
| 1991 |
wpforo_phrase( 'Indexing queued: %1$d topics (limited by %2$d available credits). %3$d topics skipped. Processing starts in 5 seconds and continues in the background.', false ), |
| 1992 |
$total_topics, |
| 1993 |
$credits_available, |
| 1994 |
$skipped_topics |
| 1995 |
); |
| 1996 |
} else { |
| 1997 |
$message = sprintf( |
| 1998 |
wpforo_phrase( 'Indexing queued: %d topics. Processing starts in 5 seconds and continues in the background.', false ), |
| 1999 |
$total_topics |
| 2000 |
); |
| 2001 |
} |
| 2002 |
|
| 2003 |
return [ |
| 2004 |
'success' => true, |
| 2005 |
'message' => $message, |
| 2006 |
'topics_queued' => $total_topics, |
| 2007 |
'topics_limited' => $topics_limited, |
| 2008 |
'skipped_topics' => $skipped_topics, |
| 2009 |
'queue_size' => count( $merged_queue ), |
| 2010 |
]; |
| 2011 |
} |
| 2012 |
|
| 2013 |
/** |
| 2014 |
* Queue a single topic for auto-indexing |
| 2015 |
* |
| 2016 |
* Used for automatic indexing when: |
| 2017 |
* - A new approved topic is created |
| 2018 |
* - An unapproved topic is approved |
| 2019 |
* |
| 2020 |
* Adds the topic to the existing queue and schedules the cron processor |
| 2021 |
* if not already scheduled. This is a lightweight operation. |
| 2022 |
* |
| 2023 |
* @param int $topicid Topic ID to queue |
| 2024 |
* @return bool True if queued successfully |
| 2025 |
*/ |
| 2026 |
public function queue_topic_for_auto_indexing( $topicid ) { |
| 2027 |
$topicid = (int) $topicid; |
| 2028 |
|
| 2029 |
// Check if auto-indexing is enabled for this board |
| 2030 |
$auto_indexing_enabled = (bool) wpforo_get_option( 'ai_auto_indexing_enabled', 0 ); |
| 2031 |
if ( ! $auto_indexing_enabled ) { |
| 2032 |
return false; |
| 2033 |
} |
| 2034 |
|
| 2035 |
// Check if AI service is available |
| 2036 |
$ai_client = $this->get_ai_client(); |
| 2037 |
if ( ! $ai_client->is_service_available() ) { |
| 2038 |
return false; |
| 2039 |
} |
| 2040 |
|
| 2041 |
// Verify topic exists and is approved (status = 0) and not private |
| 2042 |
$topic = WPF()->topic->get_topic( $topicid ); |
| 2043 |
if ( ! $topic ) { |
| 2044 |
return false; |
| 2045 |
} |
| 2046 |
|
| 2047 |
// Only index approved (status=0), non-private topics |
| 2048 |
if ( intval( wpfval( $topic, 'status' ) ) !== 0 || intval( wpfval( $topic, 'private' ) ) === 1 ) { |
| 2049 |
return false; |
| 2050 |
} |
| 2051 |
|
| 2052 |
// Use mode-specific queue key to ensure topics are indexed in the correct mode |
| 2053 |
// This prevents cloud topics from being processed by local indexing and vice versa |
| 2054 |
$storage_mode = $this->get_storage_mode(); |
| 2055 |
$queue_key = 'wpforo_ai_indexing_queue_' . $storage_mode . '_' . $this->board_id; |
| 2056 |
|
| 2057 |
// Get existing queue and add this topic (avoid duplicates) |
| 2058 |
// Cast to int to prevent type mismatches — wpfval() returns strings, |
| 2059 |
// but get_pending_topics_for_indexing() stores integers via array_map('intval') |
| 2060 |
$existing_queue = get_option( $queue_key, [] ); |
| 2061 |
if ( ! in_array( $topicid, $existing_queue ) ) { |
| 2062 |
$existing_queue[] = $topicid; |
| 2063 |
update_option( $queue_key, $existing_queue, false ); // No autoload |
| 2064 |
} |
| 2065 |
|
| 2066 |
// Schedule the mode-specific cron processor if not already scheduled |
| 2067 |
$cron_hook = 'wpforo_ai_process_queue_' . $storage_mode; |
| 2068 |
$cron_args = [ $this->board_id ]; |
| 2069 |
|
| 2070 |
if ( ! wp_next_scheduled( $cron_hook, $cron_args ) ) { |
| 2071 |
// Cloud: 24 hours, Local: 1 hour — batches replies to save credits |
| 2072 |
$default_delay = ( $storage_mode === 'cloud' ) ? 86400 : 3600; |
| 2073 |
|
| 2074 |
/** |
| 2075 |
* Filters the auto-indexing delay in seconds before the queued topics are processed. |
| 2076 |
* |
| 2077 |
* Cloud mode defaults to 86400 (24 hours), local mode defaults to 3600 (1 hour). |
| 2078 |
* |
| 2079 |
* @param int $delay Delay in seconds. |
| 2080 |
* @param string $storage_mode Storage mode ('local' or 'cloud'). |
| 2081 |
* @param int $board_id Board ID. |
| 2082 |
*/ |
| 2083 |
$delay = (int) apply_filters( 'wpforo_ai_auto_indexing_delay', $default_delay, $storage_mode, $this->board_id ); |
| 2084 |
|
| 2085 |
wp_schedule_single_event( time() + max( 1, $delay ), $cron_hook, $cron_args ); |
| 2086 |
} |
| 2087 |
|
| 2088 |
$this->log_info( 'topic_queued_for_auto_indexing', [ |
| 2089 |
'topicid' => $topicid, |
| 2090 |
'queue_size' => count( $existing_queue ), |
| 2091 |
'storage_mode' => $storage_mode, |
| 2092 |
'queue_key' => $queue_key, |
| 2093 |
] ); |
| 2094 |
|
| 2095 |
return true; |
| 2096 |
} |
| 2097 |
|
| 2098 |
/** |
| 2099 |
* Get new or recently modified topics that need indexing |
| 2100 |
* |
| 2101 |
* Only returns topics that are newer (by modified date) than the last |
| 2102 |
* indexed topic in the current storage mode. This ensures the daily cron |
| 2103 |
* only indexes genuinely new or updated topics, not old unindexed ones |
| 2104 |
* from forums the admin chose not to index. |
| 2105 |
* |
| 2106 |
* @param int $limit Maximum number of topics to return (default 100) |
| 2107 |
* @return array Array of topic IDs |
| 2108 |
*/ |
| 2109 |
public function get_pending_topics_for_indexing( $limit = 100 ) { |
| 2110 |
global $wpdb; |
| 2111 |
|
| 2112 |
$column = $this->is_local_mode() ? 'local' : 'cloud'; |
| 2113 |
$table = WPF()->tables->topics; |
| 2114 |
|
| 2115 |
// Find the modified date of the most recently indexed topic |
| 2116 |
$last_indexed_modified = $wpdb->get_var( |
| 2117 |
"SELECT `modified` FROM `{$table}` WHERE `{$column}` = 1 ORDER BY `modified` DESC LIMIT 1" |
| 2118 |
); |
| 2119 |
|
| 2120 |
if ( $last_indexed_modified ) { |
| 2121 |
// Select unindexed topics modified after the last indexed topic |
| 2122 |
$topic_ids = $wpdb->get_col( |
| 2123 |
$wpdb->prepare( |
| 2124 |
"SELECT topicid FROM `{$table}` |
| 2125 |
WHERE `{$column}` = 0 |
| 2126 |
AND `modified` > %s |
| 2127 |
AND `status` = 0 |
| 2128 |
AND `private` = 0 |
| 2129 |
ORDER BY `modified` ASC |
| 2130 |
LIMIT %d", |
| 2131 |
$last_indexed_modified, |
| 2132 |
$limit |
| 2133 |
) |
| 2134 |
); |
| 2135 |
} else { |
| 2136 |
// No indexed topics exist yet — only pick up topics from today forward |
| 2137 |
$topic_ids = $wpdb->get_col( |
| 2138 |
$wpdb->prepare( |
| 2139 |
"SELECT topicid FROM `{$table}` |
| 2140 |
WHERE `{$column}` = 0 |
| 2141 |
AND `modified` >= CURDATE() |
| 2142 |
AND `status` = 0 |
| 2143 |
AND `private` = 0 |
| 2144 |
ORDER BY `modified` ASC |
| 2145 |
LIMIT %d", |
| 2146 |
$limit |
| 2147 |
) |
| 2148 |
); |
| 2149 |
} |
| 2150 |
|
| 2151 |
return array_map( 'intval', $topic_ids ); |
| 2152 |
} |
| 2153 |
|
| 2154 |
/** |
| 2155 |
* Process pending topics (daily cron job handler) |
| 2156 |
* |
| 2157 |
* Finds topics with local=0 or cloud=0 (based on storage mode) |
| 2158 |
* and queues them for indexing using mode-specific queue. |
| 2159 |
* |
| 2160 |
* @return array Result with counts |
| 2161 |
*/ |
| 2162 |
public function cron_process_pending_topics() { |
| 2163 |
// Check if auto-indexing is enabled for this board |
| 2164 |
$auto_indexing_enabled = (bool) wpforo_get_option( 'ai_auto_indexing_enabled', 0 ); |
| 2165 |
if ( ! $auto_indexing_enabled ) { |
| 2166 |
return [ |
| 2167 |
'success' => false, |
| 2168 |
'message' => 'Auto-indexing is disabled', |
| 2169 |
'topics_queued' => 0, |
| 2170 |
]; |
| 2171 |
} |
| 2172 |
|
| 2173 |
// Check if AI service is available |
| 2174 |
$ai_client = $this->get_ai_client(); |
| 2175 |
if ( ! $ai_client->is_service_available() ) { |
| 2176 |
return [ |
| 2177 |
'success' => false, |
| 2178 |
'message' => 'AI service not available', |
| 2179 |
'topics_queued' => 0, |
| 2180 |
]; |
| 2181 |
} |
| 2182 |
|
| 2183 |
// Get current storage mode - this determines which column to check |
| 2184 |
$storage_mode = $this->get_storage_mode(); |
| 2185 |
|
| 2186 |
// Get pending topics (limit to 500 per day to avoid overloading) |
| 2187 |
// This checks local=0 for local mode, cloud=0 for cloud mode |
| 2188 |
$pending_topics = $this->get_pending_topics_for_indexing( 500 ); |
| 2189 |
|
| 2190 |
if ( empty( $pending_topics ) ) { |
| 2191 |
return [ |
| 2192 |
'success' => true, |
| 2193 |
'message' => 'No pending topics found', |
| 2194 |
'topics_queued' => 0, |
| 2195 |
'storage_mode' => $storage_mode, |
| 2196 |
]; |
| 2197 |
} |
| 2198 |
|
| 2199 |
// Use mode-specific queue key to ensure topics are indexed in the correct mode |
| 2200 |
$queue_key = 'wpforo_ai_indexing_queue_' . $storage_mode . '_' . $this->board_id; |
| 2201 |
|
| 2202 |
// Get existing queue and merge |
| 2203 |
$existing_queue = get_option( $queue_key, [] ); |
| 2204 |
$merged_queue = array_unique( array_merge( $existing_queue, $pending_topics ) ); |
| 2205 |
update_option( $queue_key, $merged_queue, false ); |
| 2206 |
|
| 2207 |
// Schedule the mode-specific cron processor if not already scheduled |
| 2208 |
$cron_hook = 'wpforo_ai_process_queue_' . $storage_mode; |
| 2209 |
$cron_args = [ $this->board_id ]; |
| 2210 |
|
| 2211 |
if ( ! wp_next_scheduled( $cron_hook, $cron_args ) ) { |
| 2212 |
wp_schedule_single_event( time() + 5, $cron_hook, $cron_args ); |
| 2213 |
} |
| 2214 |
|
| 2215 |
$this->log_info( 'daily_pending_topics_queued', [ |
| 2216 |
'topics_found' => count( $pending_topics ), |
| 2217 |
'queue_size' => count( $merged_queue ), |
| 2218 |
'storage_mode' => $storage_mode, |
| 2219 |
'queue_key' => $queue_key, |
| 2220 |
] ); |
| 2221 |
|
| 2222 |
return [ |
| 2223 |
'success' => true, |
| 2224 |
'message' => sprintf( 'Queued %d pending topics for %s indexing', count( $pending_topics ), $storage_mode ), |
| 2225 |
'topics_queued' => count( $pending_topics ), |
| 2226 |
'storage_mode' => $storage_mode, |
| 2227 |
]; |
| 2228 |
} |
| 2229 |
|
| 2230 |
/** |
| 2231 |
* Process the indexing queue (called by WP Cron) |
| 2232 |
* |
| 2233 |
* Self-rescheduling pattern: processes one batch at a time, |
| 2234 |
* then reschedules itself if more topics remain. |
| 2235 |
* |
| 2236 |
* @param int $board_id Board ID |
| 2237 |
* @return void |
| 2238 |
*/ |
| 2239 |
public function cron_process_queue( $board_id = 0 ) { |
| 2240 |
// Set the board context to ensure board-specific options are read correctly |
| 2241 |
WPF()->change_board( $board_id ); |
| 2242 |
|
| 2243 |
$lock_key = 'wpforo_ai_indexing_lock_' . $board_id; |
| 2244 |
$queue_key = 'wpforo_ai_indexing_queue_' . $board_id; |
| 2245 |
$batch_size = (int) wpforo_get_option( 'ai_pagination_size', 20 ); // Topics per batch from settings |
| 2246 |
|
| 2247 |
// Check if another process (AJAX or another cron) is running |
| 2248 |
$existing_lock = get_transient( $lock_key ); |
| 2249 |
if ( $existing_lock ) { |
| 2250 |
// Reschedule as backup |
| 2251 |
$cron_hook = 'wpforo_ai_process_queue'; |
| 2252 |
$cron_args = [ $board_id ]; |
| 2253 |
if ( ! wp_next_scheduled( $cron_hook, $cron_args ) ) { |
| 2254 |
wp_schedule_single_event( time() + 60, $cron_hook, $cron_args ); |
| 2255 |
} |
| 2256 |
return; |
| 2257 |
} |
| 2258 |
|
| 2259 |
// Lock per batch — expires after 2 minutes in case of crash |
| 2260 |
set_transient( $lock_key, 'cron_' . time(), 120 ); |
| 2261 |
|
| 2262 |
// Get pending topics from queue |
| 2263 |
$pending_topics = get_option( $queue_key, [] ); |
| 2264 |
|
| 2265 |
if ( empty( $pending_topics ) ) { |
| 2266 |
// Queue is empty, nothing to do |
| 2267 |
delete_option( $queue_key ); |
| 2268 |
delete_transient( $lock_key ); |
| 2269 |
return; |
| 2270 |
} |
| 2271 |
|
| 2272 |
// Take the next batch |
| 2273 |
$batch = array_slice( $pending_topics, 0, $batch_size ); |
| 2274 |
$remaining = array_slice( $pending_topics, $batch_size ); |
| 2275 |
|
| 2276 |
// Update queue with remaining topics BEFORE processing |
| 2277 |
// This prevents re-processing if cron runs twice |
| 2278 |
if ( ! empty( $remaining ) ) { |
| 2279 |
update_option( $queue_key, $remaining, false ); |
| 2280 |
} else { |
| 2281 |
delete_option( $queue_key ); |
| 2282 |
} |
| 2283 |
|
| 2284 |
// Use the batch embedding method with saved settings |
| 2285 |
$chunk_size = (int) wpforo_get_option( 'ai_chunk_size', 512 ); |
| 2286 |
$overlap_percent = (int) wpforo_get_option( 'ai_overlap_percent', 20 ); |
| 2287 |
$result = $this->index_topics_batch_local( $batch, [ |
| 2288 |
'chunk_size' => $chunk_size, |
| 2289 |
'overlap_percent' => $overlap_percent, |
| 2290 |
] ); |
| 2291 |
|
| 2292 |
// Check for credit exhaustion - stop rescheduling if no credits |
| 2293 |
$has_credit_error = false; |
| 2294 |
$errors = is_array( $result ) ? ( $result['errors'] ?? [] ) : []; |
| 2295 |
foreach ( $errors as $error ) { |
| 2296 |
if ( stripos( $error, 'insufficient credits' ) !== false || stripos( $error, '402' ) !== false ) { |
| 2297 |
$has_credit_error = true; |
| 2298 |
break; |
| 2299 |
} |
| 2300 |
} |
| 2301 |
|
| 2302 |
// Release lock after processing |
| 2303 |
delete_transient( $lock_key ); |
| 2304 |
|
| 2305 |
// Clear indexed counts cache so UI shows updated progress after each batch |
| 2306 |
$this->clear_indexing_stats_cache( $board_id ); |
| 2307 |
|
| 2308 |
if ( $has_credit_error ) { |
| 2309 |
// Clear queue so page reload doesn't auto-resume indexing |
| 2310 |
delete_option( $queue_key ); |
| 2311 |
delete_option( 'wpforo_ai_indexing_settings_' . $board_id ); |
| 2312 |
return; |
| 2313 |
} |
| 2314 |
|
| 2315 |
// If more topics remain, reschedule ourselves |
| 2316 |
if ( ! empty( $remaining ) ) { |
| 2317 |
$cron_hook = 'wpforo_ai_process_queue'; |
| 2318 |
$cron_args = [ $board_id ]; |
| 2319 |
|
| 2320 |
// Schedule next batch in 30 seconds |
| 2321 |
if ( ! wp_next_scheduled( $cron_hook, $cron_args ) ) { |
| 2322 |
wp_schedule_single_event( time() + 30, $cron_hook, $cron_args ); |
| 2323 |
} |
| 2324 |
} |
| 2325 |
} |
| 2326 |
|
| 2327 |
/** |
| 2328 |
* Process a mode-specific auto-indexing queue (called by WP Cron) |
| 2329 |
* |
| 2330 |
* This processes topics that were queued for a specific storage mode, |
| 2331 |
* ensuring they are indexed with the correct method regardless of |
| 2332 |
* what the current storage mode is set to. |
| 2333 |
* |
| 2334 |
* @param int $board_id Board ID |
| 2335 |
* @param string $mode Storage mode ('local' or 'cloud') |
| 2336 |
* @return void |
| 2337 |
*/ |
| 2338 |
public function cron_process_queue_mode( $board_id = 0, $mode = 'local' ) { |
| 2339 |
// Set the board context to ensure board-specific options are read correctly |
| 2340 |
WPF()->change_board( $board_id ); |
| 2341 |
|
| 2342 |
$lock_key = 'wpforo_ai_indexing_lock_' . $mode . '_' . $board_id; |
| 2343 |
$queue_key = 'wpforo_ai_indexing_queue_' . $mode . '_' . $board_id; |
| 2344 |
$batch_size = (int) wpforo_get_option( 'ai_pagination_size', 20 ); // Topics per batch from settings |
| 2345 |
|
| 2346 |
// Check if already processing |
| 2347 |
$existing_lock = get_transient( $lock_key ); |
| 2348 |
if ( $existing_lock ) { |
| 2349 |
// Already processing - reschedule as backup |
| 2350 |
$cron_hook = 'wpforo_ai_process_queue_' . $mode; |
| 2351 |
$cron_args = [ $board_id ]; |
| 2352 |
if ( ! wp_next_scheduled( $cron_hook, $cron_args ) ) { |
| 2353 |
wp_schedule_single_event( time() + 60, $cron_hook, $cron_args ); |
| 2354 |
} |
| 2355 |
return; |
| 2356 |
} |
| 2357 |
|
| 2358 |
// Set lock for processing |
| 2359 |
set_transient( $lock_key, 'cron', 300 ); |
| 2360 |
|
| 2361 |
// Get pending topics from mode-specific queue |
| 2362 |
$pending_topics = get_option( $queue_key, [] ); |
| 2363 |
|
| 2364 |
if ( empty( $pending_topics ) ) { |
| 2365 |
// Queue is empty, nothing to do |
| 2366 |
delete_option( $queue_key ); |
| 2367 |
delete_transient( $lock_key ); |
| 2368 |
return; |
| 2369 |
} |
| 2370 |
|
| 2371 |
// Take the next batch |
| 2372 |
$batch = array_slice( $pending_topics, 0, $batch_size ); |
| 2373 |
$remaining = array_slice( $pending_topics, $batch_size ); |
| 2374 |
|
| 2375 |
// Update queue with remaining topics BEFORE processing |
| 2376 |
if ( ! empty( $remaining ) ) { |
| 2377 |
update_option( $queue_key, $remaining, false ); |
| 2378 |
} else { |
| 2379 |
delete_option( $queue_key ); |
| 2380 |
} |
| 2381 |
|
| 2382 |
// Get chunk settings from options |
| 2383 |
$chunk_size = (int) wpforo_get_option( 'ai_chunk_size', 512 ); |
| 2384 |
$overlap_percent = (int) wpforo_get_option( 'ai_overlap_percent', 20 ); |
| 2385 |
|
| 2386 |
// Process using the specified mode |
| 2387 |
if ( $mode === 'local' ) { |
| 2388 |
// Use local batch embedding |
| 2389 |
$result = $this->index_topics_batch_local( $batch, [ |
| 2390 |
'chunk_size' => $chunk_size, |
| 2391 |
'overlap_percent' => $overlap_percent, |
| 2392 |
] ); |
| 2393 |
} else { |
| 2394 |
// Use cloud indexing via ingest_topics |
| 2395 |
$ai_client = $this->get_ai_client(); |
| 2396 |
$result = $ai_client->ingest_topics( $batch, $chunk_size, $overlap_percent ); |
| 2397 |
} |
| 2398 |
|
| 2399 |
// Check for credit exhaustion - stop rescheduling if no credits |
| 2400 |
$has_credit_error = false; |
| 2401 |
$errors = is_array( $result ) ? ( $result['errors'] ?? [] ) : []; |
| 2402 |
foreach ( $errors as $error ) { |
| 2403 |
if ( stripos( $error, 'insufficient credits' ) !== false || stripos( $error, '402' ) !== false ) { |
| 2404 |
$has_credit_error = true; |
| 2405 |
break; |
| 2406 |
} |
| 2407 |
} |
| 2408 |
|
| 2409 |
$this->log_info( 'cron_process_queue_mode_complete', [ |
| 2410 |
'mode' => $mode, |
| 2411 |
'batch_size' => count( $batch ), |
| 2412 |
'remaining' => count( $remaining ), |
| 2413 |
'credit_error' => $has_credit_error, |
| 2414 |
'result' => is_wp_error( $result ) ? $result->get_error_message() : 'success', |
| 2415 |
] ); |
| 2416 |
|
| 2417 |
// Release lock after processing |
| 2418 |
delete_transient( $lock_key ); |
| 2419 |
|
| 2420 |
// Clear indexed counts cache so UI shows updated progress after each batch |
| 2421 |
$this->clear_indexing_stats_cache( $board_id ); |
| 2422 |
|
| 2423 |
if ( $has_credit_error ) { |
| 2424 |
// Clear queue so page reload doesn't auto-resume indexing |
| 2425 |
delete_option( $queue_key ); |
| 2426 |
return; |
| 2427 |
} |
| 2428 |
|
| 2429 |
// If more topics remain, reschedule ourselves |
| 2430 |
if ( ! empty( $remaining ) ) { |
| 2431 |
$cron_hook = 'wpforo_ai_process_queue_' . $mode; |
| 2432 |
$cron_args = [ $board_id ]; |
| 2433 |
|
| 2434 |
// Schedule next batch in 30 seconds |
| 2435 |
if ( ! wp_next_scheduled( $cron_hook, $cron_args ) ) { |
| 2436 |
wp_schedule_single_event( time() + 30, $cron_hook, $cron_args ); |
| 2437 |
} |
| 2438 |
} |
| 2439 |
} |
| 2440 |
|
| 2441 |
/** |
| 2442 |
* Reindex all topics |
| 2443 |
* |
| 2444 |
* For local mode, schedules WP Cron jobs to index in batches. |
| 2445 |
* For cloud mode, sends to cloud API. |
| 2446 |
* |
| 2447 |
* @param int $chunk_size Chunk size |
| 2448 |
* @param int $overlap_percent Overlap percentage |
| 2449 |
* @return array|WP_Error Result or error |
| 2450 |
*/ |
| 2451 |
public function reindex_all_topics( $chunk_size = 512, $overlap_percent = 20 ) { |
| 2452 |
if ( $this->is_local_mode() ) { |
| 2453 |
return $this->reindex_all_topics_local(); |
| 2454 |
} else { |
| 2455 |
$ai_client = $this->get_ai_client(); |
| 2456 |
return $ai_client->reindex_all_topics( $chunk_size, $overlap_percent ); |
| 2457 |
} |
| 2458 |
} |
| 2459 |
|
| 2460 |
/** |
| 2461 |
* Clear local indexed status for all topics |
| 2462 |
* |
| 2463 |
* Sets `local` column to 0 for all topics without deleting actual embeddings. |
| 2464 |
* Used when user wants to force re-index all topics. |
| 2465 |
* |
| 2466 |
* @return int Number of topics updated |
| 2467 |
*/ |
| 2468 |
private function clear_topics_local_indexed_status() { |
| 2469 |
global $wpdb; |
| 2470 |
|
| 2471 |
$updated = $wpdb->query( |
| 2472 |
"UPDATE `" . WPF()->tables->topics . "` SET `local` = 0 WHERE `local` = 1" |
| 2473 |
); |
| 2474 |
|
| 2475 |
if ( $updated > 0 ) { |
| 2476 |
wpforo_clean_cache( 'topic' ); |
| 2477 |
$this->clear_indexing_stats_cache(); |
| 2478 |
} |
| 2479 |
|
| 2480 |
return (int) $updated; |
| 2481 |
} |
| 2482 |
|
| 2483 |
/** |
| 2484 |
* Reindex all topics locally via WP Cron |
| 2485 |
* |
| 2486 |
* Uses batch embedding API for efficiency. |
| 2487 |
* Supports incremental indexing: only indexes topics with local=0. |
| 2488 |
* If all topics are already indexed, clears status and re-indexes all. |
| 2489 |
* |
| 2490 |
* @return array Result with queued count |
| 2491 |
*/ |
| 2492 |
private function reindex_all_topics_local() { |
| 2493 |
// Get unindexed topics (local = 0) |
| 2494 |
$unindexed_topic_ids = $this->get_unindexed_topic_ids(); |
| 2495 |
$unindexed_count = count( $unindexed_topic_ids ); |
| 2496 |
|
| 2497 |
// Determine if we're doing incremental indexing or full re-index |
| 2498 |
$is_reindex_all = ( $unindexed_count === 0 ); |
| 2499 |
|
| 2500 |
if ( $is_reindex_all ) { |
| 2501 |
// All topics are indexed - user wants to re-index everything |
| 2502 |
// Clear status first so all topics become "unindexed" |
| 2503 |
$this->clear_topics_local_indexed_status(); |
| 2504 |
|
| 2505 |
// Now get all topic IDs (they're all local=0 now) |
| 2506 |
$topic_ids = $this->get_unindexed_topic_ids(); |
| 2507 |
} else { |
| 2508 |
// Some topics need indexing - only index those (don't clear status) |
| 2509 |
$topic_ids = $unindexed_topic_ids; |
| 2510 |
} |
| 2511 |
|
| 2512 |
$total_topics = count( $topic_ids ); |
| 2513 |
|
| 2514 |
if ( $total_topics === 0 ) { |
| 2515 |
return [ |
| 2516 |
'success' => true, |
| 2517 |
'topics_queued' => 0, |
| 2518 |
'message' => wpforo_phrase( 'No topics found to index.', false ), |
| 2519 |
]; |
| 2520 |
} |
| 2521 |
|
| 2522 |
// Check available credits before starting (same as cloud mode) |
| 2523 |
$ai_client = $this->get_ai_client(); |
| 2524 |
$status = $ai_client->get_tenant_status( true ); // Force fresh status |
| 2525 |
if ( is_wp_error( $status ) ) { |
| 2526 |
return $status; |
| 2527 |
} |
| 2528 |
|
| 2529 |
$credits_available = isset( $status['subscription']['credits_remaining'] ) |
| 2530 |
? (int) $status['subscription']['credits_remaining'] |
| 2531 |
: 0; |
| 2532 |
|
| 2533 |
if ( $credits_available <= 0 ) { |
| 2534 |
return new \WP_Error( |
| 2535 |
'no_credits', |
| 2536 |
wpforo_phrase( 'No credits available for indexing. Please wait for your monthly credit reset or upgrade your plan.', false ) |
| 2537 |
); |
| 2538 |
} |
| 2539 |
|
| 2540 |
// Warn if credits are low but still proceed (deduplication may reduce actual usage) |
| 2541 |
$credits_warning = null; |
| 2542 |
if ( $credits_available < $total_topics ) { |
| 2543 |
$credits_warning = sprintf( |
| 2544 |
wpforo_phrase( 'Note: You have %d credits but %d topics to index. Unchanged topics will be skipped, but new topics may not all be indexed.', false ), |
| 2545 |
$credits_available, |
| 2546 |
$total_topics |
| 2547 |
); |
| 2548 |
} |
| 2549 |
|
| 2550 |
// Use self-rescheduling queue pattern: |
| 2551 |
// - Store all topic IDs in a queue (option) |
| 2552 |
// - Schedule ONE cron job |
| 2553 |
// - Job processes a batch (pagination_size topics), then reschedules itself if more remain |
| 2554 |
// This avoids overwhelming WP Cron with hundreds of jobs for large forums |
| 2555 |
$queue_key = 'wpforo_ai_indexing_queue_' . $this->board_id; |
| 2556 |
|
| 2557 |
// Clear any existing queue and set new one |
| 2558 |
update_option( $queue_key, $topic_ids, false ); // No autoload |
| 2559 |
|
| 2560 |
// Schedule ONE job to start processing (if not already scheduled) |
| 2561 |
$cron_hook = 'wpforo_ai_process_queue'; |
| 2562 |
$cron_args = [ $this->board_id ]; |
| 2563 |
|
| 2564 |
if ( ! wp_next_scheduled( $cron_hook, $cron_args ) ) { |
| 2565 |
wp_schedule_single_event( time() + 5, $cron_hook, $cron_args ); |
| 2566 |
} |
| 2567 |
|
| 2568 |
$batch_size = (int) wpforo_get_option( 'ai_pagination_size', 20 ); |
| 2569 |
$message = sprintf( |
| 2570 |
wpforo_phrase( 'Indexing queued! %d topics will be processed in batches of %d. Processing starts in 5 seconds.', false ), |
| 2571 |
$total_topics, |
| 2572 |
$batch_size |
| 2573 |
); |
| 2574 |
|
| 2575 |
// Append credits warning if applicable |
| 2576 |
if ( $credits_warning ) { |
| 2577 |
$message .= ' ' . $credits_warning; |
| 2578 |
} |
| 2579 |
|
| 2580 |
return [ |
| 2581 |
'success' => true, |
| 2582 |
'topics_queued' => $total_topics, |
| 2583 |
'total_topics' => $total_topics, |
| 2584 |
'batches_queued' => 1, // Just ONE self-rescheduling job |
| 2585 |
'credits_available' => $credits_available, |
| 2586 |
'message' => $message, |
| 2587 |
]; |
| 2588 |
} |
| 2589 |
|
| 2590 |
// ========================================================================= |
| 2591 |
// SEMANTIC SEARCH |
| 2592 |
// ========================================================================= |
| 2593 |
|
| 2594 |
/** |
| 2595 |
* Perform semantic search |
| 2596 |
* |
| 2597 |
* @param string $query Search query |
| 2598 |
* @param int $limit Maximum results |
| 2599 |
* @param array $filters Optional filters (forumid, userid, etc.) |
| 2600 |
* @return array|WP_Error Search results or error |
| 2601 |
*/ |
| 2602 |
public function semantic_search( $query, $limit = 10, $filters = [] ) { |
| 2603 |
// Add forum access filtering (restrict to forums user can view) |
| 2604 |
// This applies to BOTH local and cloud modes |
| 2605 |
// Skip if accessible_forumids already provided in filters (avoids duplicate computation) |
| 2606 |
if ( ! array_key_exists( 'accessible_forumids', $filters ) ) { |
| 2607 |
$ai_client = $this->get_ai_client(); |
| 2608 |
$accessible_forumids = $ai_client->get_accessible_forumids(); |
| 2609 |
|
| 2610 |
if ( $accessible_forumids !== null ) { |
| 2611 |
// User has restricted access - if empty array, they can't access any forums |
| 2612 |
if ( empty( $accessible_forumids ) ) { |
| 2613 |
return [ |
| 2614 |
'results' => [], |
| 2615 |
'total' => 0, |
| 2616 |
'message' => wpforo_phrase( 'No results found', false ), |
| 2617 |
]; |
| 2618 |
} |
| 2619 |
$filters['accessible_forumids'] = $accessible_forumids; |
| 2620 |
} |
| 2621 |
} elseif ( is_array( $filters['accessible_forumids'] ) && empty( $filters['accessible_forumids'] ) ) { |
| 2622 |
// Empty array passed = user can't access any forums |
| 2623 |
return [ |
| 2624 |
'results' => [], |
| 2625 |
'total' => 0, |
| 2626 |
'message' => wpforo_phrase( 'No results found', false ), |
| 2627 |
]; |
| 2628 |
} |
| 2629 |
|
| 2630 |
if ( $this->is_local_mode() ) { |
| 2631 |
return $this->semantic_search_local( $query, $limit, $filters ); |
| 2632 |
} else { |
| 2633 |
return $this->semantic_search_cloud( $query, $limit, $filters ); |
| 2634 |
} |
| 2635 |
} |
| 2636 |
|
| 2637 |
/** |
| 2638 |
* Perform local semantic search |
| 2639 |
* |
| 2640 |
* @param string $query Search query |
| 2641 |
* @param int $limit Maximum results |
| 2642 |
* @param array $filters Filters |
| 2643 |
* @return array|WP_Error |
| 2644 |
*/ |
| 2645 |
private function semantic_search_local( $query, $limit = 10, $filters = [] ) { |
| 2646 |
// Inject score threshold from settings so VectorStorageLocal filters at search time. |
| 2647 |
// Local cosine similarities are on a different scale (5-25%) than cloud scores (30-90%), |
| 2648 |
// so apply 1/3 of the configured threshold for local mode. |
| 2649 |
// Minimum absolute threshold of 15% prevents garbage results from nonsense queries |
| 2650 |
// (random vectors have ~5-15% cosine similarity with any query). |
| 2651 |
$min_score_setting = (int) wpfval( WPF()->settings->ai, 'search_min_score' ); |
| 2652 |
$local_threshold = $min_score_setting > 0 ? ( $min_score_setting / 100 ) / 3 : 0; |
| 2653 |
$absolute_min = 0.15; // 15% - below this, results are definitely garbage |
| 2654 |
if ( ! isset( $filters['min_score'] ) ) { |
| 2655 |
$filters['min_score'] = max( $local_threshold, $absolute_min ); |
| 2656 |
} |
| 2657 |
|
| 2658 |
// Convert accessible_forumids to forumids for local storage (include filter) |
| 2659 |
if ( ! empty( $filters['accessible_forumids'] ) && empty( $filters['forumids'] ) ) { |
| 2660 |
$filters['forumids'] = $filters['accessible_forumids']; |
| 2661 |
unset( $filters['accessible_forumids'] ); |
| 2662 |
} |
| 2663 |
|
| 2664 |
// Generate embedding for query |
| 2665 |
$query_embedding = $this->generate_embedding( $query ); |
| 2666 |
if ( is_wp_error( $query_embedding ) ) { |
| 2667 |
return $query_embedding; |
| 2668 |
} |
| 2669 |
|
| 2670 |
$local = $this->get_local_storage(); |
| 2671 |
// Fetch 3x more results to have headroom after deduplication |
| 2672 |
$results = $local->semantic_search( $query_embedding, $limit * 3, $filters ); |
| 2673 |
|
| 2674 |
if ( is_wp_error( $results ) ) { |
| 2675 |
return $results; |
| 2676 |
} |
| 2677 |
|
| 2678 |
// Deduplicate: group by topic/post, then by content fingerprint |
| 2679 |
$results = $this->deduplicate_results( $results, $limit ); |
| 2680 |
|
| 2681 |
// Format results to match cloud response format |
| 2682 |
return $this->format_search_results( $results ); |
| 2683 |
} |
| 2684 |
|
| 2685 |
/** |
| 2686 |
* Deduplicate local search results. |
| 2687 |
* |
| 2688 |
* Mirrors the cloud search's _deduplicate_by_topic() two-stage approach: |
| 2689 |
* 1. Group by topicid (forum) or postid (WordPress CPT) — keep best score per group |
| 2690 |
* 2. Content fingerprint dedup — catch identical text across different posts |
| 2691 |
* |
| 2692 |
* @param array $results Raw search results from VectorStorageLocal |
| 2693 |
* @param int $limit Maximum results to return |
| 2694 |
* @return array Deduplicated results |
| 2695 |
*/ |
| 2696 |
private function deduplicate_results( $results, $limit ) { |
| 2697 |
if ( empty( $results ) ) { |
| 2698 |
return $results; |
| 2699 |
} |
| 2700 |
|
| 2701 |
// Stage 1: Group by topic/post, keep best score per group |
| 2702 |
$best_by_group = []; |
| 2703 |
foreach ( $results as $result ) { |
| 2704 |
$content_type = $result['content_type'] ?? 'forum'; |
| 2705 |
|
| 2706 |
if ( $content_type !== 'forum' ) { |
| 2707 |
// WordPress CPT: group by postid (each post is unique content) |
| 2708 |
$group_key = 'wp_' . $result['postid']; |
| 2709 |
} else { |
| 2710 |
// Forum: group by topicid (multiple posts per topic) |
| 2711 |
$group_key = 'topic_' . $result['topicid']; |
| 2712 |
} |
| 2713 |
|
| 2714 |
if ( ! isset( $best_by_group[ $group_key ] ) |
| 2715 |
|| $result['similarity'] > $best_by_group[ $group_key ]['similarity'] ) { |
| 2716 |
$best_by_group[ $group_key ] = $result; |
| 2717 |
} |
| 2718 |
} |
| 2719 |
|
| 2720 |
// Sort by similarity descending |
| 2721 |
$deduplicated = array_values( $best_by_group ); |
| 2722 |
usort( $deduplicated, function ( $a, $b ) { |
| 2723 |
return $b['similarity'] <=> $a['similarity']; |
| 2724 |
} ); |
| 2725 |
|
| 2726 |
// Stage 2: Content fingerprint dedup (catches identical text across different posts) |
| 2727 |
$seen_content = []; |
| 2728 |
$final = []; |
| 2729 |
foreach ( $deduplicated as $result ) { |
| 2730 |
$preview = $result['content_preview'] ?? ''; |
| 2731 |
$fingerprint = substr( strtolower( trim( $preview ) ), 0, 200 ); |
| 2732 |
|
| 2733 |
if ( $fingerprint && isset( $seen_content[ $fingerprint ] ) ) { |
| 2734 |
continue; // Skip duplicate content |
| 2735 |
} |
| 2736 |
|
| 2737 |
if ( $fingerprint ) { |
| 2738 |
$seen_content[ $fingerprint ] = true; |
| 2739 |
} |
| 2740 |
$final[] = $result; |
| 2741 |
} |
| 2742 |
|
| 2743 |
return array_slice( $final, 0, $limit ); |
| 2744 |
} |
| 2745 |
|
| 2746 |
/** |
| 2747 |
* Perform cloud semantic search |
| 2748 |
* |
| 2749 |
* @param string $query Search query |
| 2750 |
* @param int $limit Maximum results |
| 2751 |
* @param array $filters Filters |
| 2752 |
* @return array|WP_Error |
| 2753 |
*/ |
| 2754 |
private function semantic_search_cloud( $query, $limit = 10, $filters = [] ) { |
| 2755 |
$ai_client = $this->get_ai_client(); |
| 2756 |
return $ai_client->semantic_search( $query, $limit, $filters ); |
| 2757 |
} |
| 2758 |
|
| 2759 |
/** |
| 2760 |
* Format local search results to match cloud response format |
| 2761 |
* |
| 2762 |
* @param array $results Raw local results |
| 2763 |
* @return array Formatted results |
| 2764 |
*/ |
| 2765 |
private function format_search_results( $results ) { |
| 2766 |
$formatted = [ |
| 2767 |
'results' => [], |
| 2768 |
'total' => count( $results ), |
| 2769 |
]; |
| 2770 |
|
| 2771 |
// Collect topic and post IDs for forum results to batch fetch (avoids N+1 queries) |
| 2772 |
$topic_ids = []; |
| 2773 |
$post_ids = []; |
| 2774 |
foreach ( $results as $result ) { |
| 2775 |
$content_type = $result['content_type'] ?? 'forum'; |
| 2776 |
if ( $content_type === 'forum' ) { |
| 2777 |
$topic_ids[] = (int) $result['topicid']; |
| 2778 |
$post_ids[] = (int) $result['postid']; |
| 2779 |
} |
| 2780 |
} |
| 2781 |
|
| 2782 |
// Batch fetch topics and posts in 2 queries instead of 2*N queries |
| 2783 |
$topics_map = []; |
| 2784 |
$posts_map = []; |
| 2785 |
|
| 2786 |
if ( ! empty( $topic_ids ) ) { |
| 2787 |
$topic_ids = array_unique( array_filter( $topic_ids ) ); |
| 2788 |
$_count = 0; |
| 2789 |
$all_topics = WPF()->topic->get_topics( [ |
| 2790 |
'include' => $topic_ids, |
| 2791 |
'row_count' => count( $topic_ids ), |
| 2792 |
'access_filter' => false, |
| 2793 |
], $_count, false ); |
| 2794 |
foreach ( $all_topics as $t ) { |
| 2795 |
$topics_map[ (int) $t['topicid'] ] = $t; |
| 2796 |
} |
| 2797 |
} |
| 2798 |
|
| 2799 |
if ( ! empty( $post_ids ) ) { |
| 2800 |
$post_ids = array_unique( array_filter( $post_ids ) ); |
| 2801 |
$_count = 0; |
| 2802 |
$all_posts = WPF()->post->get_posts( [ |
| 2803 |
'include' => $post_ids, |
| 2804 |
'row_count' => count( $post_ids ), |
| 2805 |
'check_private' => false, |
| 2806 |
], $_count, false ); |
| 2807 |
foreach ( $all_posts as $p ) { |
| 2808 |
$posts_map[ (int) $p['postid'] ] = $p; |
| 2809 |
} |
| 2810 |
} |
| 2811 |
|
| 2812 |
// Process results using pre-fetched maps |
| 2813 |
foreach ( $results as $result ) { |
| 2814 |
$content_type = $result['content_type'] ?? 'forum'; |
| 2815 |
|
| 2816 |
if ( $content_type !== 'forum' ) { |
| 2817 |
// WordPress CPT result from local DB |
| 2818 |
$wp_post = get_post( $result['postid'] ); |
| 2819 |
if ( ! $wp_post || $wp_post->post_status !== 'publish' ) { |
| 2820 |
continue; |
| 2821 |
} |
| 2822 |
|
| 2823 |
$post_type_obj = get_post_type_object( $wp_post->post_type ); |
| 2824 |
$post_type_label = $post_type_obj ? $post_type_obj->labels->singular_name : ucfirst( $wp_post->post_type ); |
| 2825 |
|
| 2826 |
$formatted['results'][] = [ |
| 2827 |
'title' => $wp_post->post_title, |
| 2828 |
'content' => $result['content_preview'] ?? wp_trim_words( wp_strip_all_tags( $wp_post->post_content ), 50 ), |
| 2829 |
'score' => (float) $result['similarity'], |
| 2830 |
'url' => get_permalink( $wp_post ), |
| 2831 |
'post_url' => get_permalink( $wp_post ), |
| 2832 |
'created' => $wp_post->post_date, |
| 2833 |
'user_id' => (int) $wp_post->post_author, |
| 2834 |
'post_id' => $wp_post->ID, |
| 2835 |
'content_type' => $content_type, |
| 2836 |
'content_source' => 'wordpress', |
| 2837 |
'post_type_label' => $post_type_label, |
| 2838 |
'metadata' => [ |
| 2839 |
'content_source' => 'wordpress', |
| 2840 |
'post_id' => $wp_post->ID, |
| 2841 |
'post_type' => $wp_post->post_type, |
| 2842 |
], |
| 2843 |
]; |
| 2844 |
continue; |
| 2845 |
} |
| 2846 |
|
| 2847 |
// Forum result - use pre-fetched maps |
| 2848 |
$topic = $topics_map[ (int) $result['topicid'] ] ?? null; |
| 2849 |
$post = $posts_map[ (int) $result['postid'] ] ?? null; |
| 2850 |
|
| 2851 |
if ( ! $topic || ! $post ) { |
| 2852 |
continue; |
| 2853 |
} |
| 2854 |
|
| 2855 |
// Calculate quality boost for ranking (uses live data, no schema changes) |
| 2856 |
// Boost is applied to ranking only; original score is preserved for display |
| 2857 |
$base_score = (float) $result['similarity']; |
| 2858 |
$boost = 1.0; |
| 2859 |
if ( ! empty( $topic['solved'] ) ) { |
| 2860 |
$boost += 0.05; // +5% for solved topics |
| 2861 |
} |
| 2862 |
if ( ! empty( $post['is_answer'] ) ) { |
| 2863 |
$boost += 0.10; // +10% for best answer posts |
| 2864 |
} |
| 2865 |
if ( ! empty( $post['votes'] ) && (int) $post['votes'] > 5 ) { |
| 2866 |
$boost += 0.03; // +3% for well-voted posts (>5 votes) |
| 2867 |
} |
| 2868 |
|
| 2869 |
$formatted['results'][] = [ |
| 2870 |
'topic_id' => (int) $result['topicid'], |
| 2871 |
'post_id' => (int) $result['postid'], |
| 2872 |
'forum_id' => (int) $result['forumid'], |
| 2873 |
'title' => $topic['title'] ?? '', |
| 2874 |
'content' => preg_replace( '/\[(?:FORUM|SOLVED|BEST ANSWER)[^\]]*\]/', '', $result['content_preview'] ?? preg_replace( '/\[(?:\/)?[a-zA-Z0-9_-]+(?:\s[^\]]*?)?\]/', '', strip_tags( $post['body'] ) ) ), |
| 2875 |
'score' => $base_score, // Original similarity for display |
| 2876 |
'_boost_score' => $base_score * $boost, // Internal: for sorting only |
| 2877 |
'url' => WPF()->topic->get_url( $topic ), // Pass array to avoid extra query |
| 2878 |
'post_url' => WPF()->post->get_url( $post ), // Pass array to avoid extra query |
| 2879 |
'created' => $post['created'] ?? null, |
| 2880 |
'user_id' => (int) ( $result['userid'] ?? 0 ), |
| 2881 |
'content_type' => 'forum', |
| 2882 |
'content_source' => 'forum', |
| 2883 |
]; |
| 2884 |
} |
| 2885 |
|
| 2886 |
// Re-sort by boosted score (quality signals affect ranking) |
| 2887 |
if ( ! empty( $formatted['results'] ) ) { |
| 2888 |
usort( $formatted['results'], function( $a, $b ) { |
| 2889 |
return ( $b['_boost_score'] ?? $b['score'] ) <=> ( $a['_boost_score'] ?? $a['score'] ); |
| 2890 |
} ); |
| 2891 |
// Remove internal boost score from output |
| 2892 |
foreach ( $formatted['results'] as &$r ) { |
| 2893 |
unset( $r['_boost_score'] ); |
| 2894 |
} |
| 2895 |
unset( $r ); |
| 2896 |
} |
| 2897 |
|
| 2898 |
return $formatted; |
| 2899 |
} |
| 2900 |
|
| 2901 |
// ========================================================================= |
| 2902 |
// SIMILAR CONTENT |
| 2903 |
// ========================================================================= |
| 2904 |
|
| 2905 |
/** |
| 2906 |
* Find similar topics/posts |
| 2907 |
* |
| 2908 |
* @param string $type 'topic' or 'post' |
| 2909 |
* @param int $id Topic or post ID |
| 2910 |
* @param int $limit Maximum results |
| 2911 |
* @param bool $use_cache Whether to use cache |
| 2912 |
* @return array|WP_Error |
| 2913 |
*/ |
| 2914 |
public function find_similar( $type, $id, $limit = 5, $use_cache = true ) { |
| 2915 |
if ( $this->is_local_mode() ) { |
| 2916 |
$local = $this->get_local_storage(); |
| 2917 |
return $local->find_similar( $type, $id, $limit, ! $use_cache ); |
| 2918 |
} else { |
| 2919 |
$ai_client = $this->get_ai_client(); |
| 2920 |
return $ai_client->find_similar_topics( $id, $limit ); |
| 2921 |
} |
| 2922 |
} |
| 2923 |
|
| 2924 |
// ========================================================================= |
| 2925 |
// UTILITY METHODS |
| 2926 |
// ========================================================================= |
| 2927 |
|
| 2928 |
/** |
| 2929 |
* Check if content is indexed |
| 2930 |
* |
| 2931 |
* @param int $postid Post ID |
| 2932 |
* @param string $content_hash Optional content hash to check freshness |
| 2933 |
* @return bool |
| 2934 |
*/ |
| 2935 |
public function is_indexed( $postid, $content_hash = null ) { |
| 2936 |
if ( $this->is_local_mode() ) { |
| 2937 |
$local = $this->get_local_storage(); |
| 2938 |
$existing = $local->get_embedding( $postid ); |
| 2939 |
|
| 2940 |
if ( ! $existing ) { |
| 2941 |
return false; |
| 2942 |
} |
| 2943 |
|
| 2944 |
if ( $content_hash !== null ) { |
| 2945 |
return $existing['content_hash'] === $content_hash; |
| 2946 |
} |
| 2947 |
|
| 2948 |
return true; |
| 2949 |
} else { |
| 2950 |
// For cloud, we'd need to check via API |
| 2951 |
// For now, assume indexed if we've indexed before |
| 2952 |
return false; // Let cloud handle deduplication |
| 2953 |
} |
| 2954 |
} |
| 2955 |
|
| 2956 |
/** |
| 2957 |
* Get the current storage mode label |
| 2958 |
* |
| 2959 |
* @return string Human-readable label |
| 2960 |
*/ |
| 2961 |
public function get_storage_mode_label() { |
| 2962 |
if ( $this->is_local_mode() ) { |
| 2963 |
return wpforo_phrase( 'Local (WordPress)', false ); |
| 2964 |
} else { |
| 2965 |
return wpforo_phrase( 'Cloud (gVectors)', false ); |
| 2966 |
} |
| 2967 |
} |
| 2968 |
|
| 2969 |
/** |
| 2970 |
* Sync local indexed status in wpforo_topics table |
| 2971 |
* |
| 2972 |
* Queries wpforo_ai_embeddings table for all unique topic IDs and |
| 2973 |
* updates the `local` column in wpforo_topics accordingly. |
| 2974 |
* Called when switching to local storage mode. |
| 2975 |
* |
| 2976 |
* @return array { |
| 2977 |
* @type int $updated Number of topics marked as indexed |
| 2978 |
* @type int $cleared Number of topics marked as not indexed |
| 2979 |
* } |
| 2980 |
*/ |
| 2981 |
public function sync_local_indexed_status() { |
| 2982 |
global $wpdb; |
| 2983 |
|
| 2984 |
$embeddings_table = WPF()->tables->ai_embeddings; |
| 2985 |
$topics_table = WPF()->tables->topics; |
| 2986 |
|
| 2987 |
// Set local=1 for topics that have embeddings (single JOIN query) |
| 2988 |
$updated = (int) $wpdb->query( |
| 2989 |
"UPDATE `{$topics_table}` t |
| 2990 |
INNER JOIN (SELECT DISTINCT topicid FROM `{$embeddings_table}` WHERE topicid > 0) e |
| 2991 |
ON t.topicid = e.topicid |
| 2992 |
SET t.`local` = 1 |
| 2993 |
WHERE t.`local` != 1" |
| 2994 |
); |
| 2995 |
|
| 2996 |
// Set local=0 for topics that have no embeddings (single LEFT JOIN query) |
| 2997 |
$cleared = (int) $wpdb->query( |
| 2998 |
"UPDATE `{$topics_table}` t |
| 2999 |
LEFT JOIN (SELECT DISTINCT topicid FROM `{$embeddings_table}` WHERE topicid > 0) e |
| 3000 |
ON t.topicid = e.topicid |
| 3001 |
SET t.`local` = 0 |
| 3002 |
WHERE e.topicid IS NULL AND t.`local` != 0" |
| 3003 |
); |
| 3004 |
|
| 3005 |
// Count indexed topics |
| 3006 |
$indexed_count = (int) $wpdb->get_var( |
| 3007 |
"SELECT COUNT(DISTINCT topicid) FROM `{$embeddings_table}` WHERE topicid > 0" |
| 3008 |
); |
| 3009 |
|
| 3010 |
// Clear stats cache for fresh counts |
| 3011 |
if ( $updated > 0 || $cleared > 0 ) { |
| 3012 |
$this->clear_indexing_stats_cache(); |
| 3013 |
} |
| 3014 |
|
| 3015 |
return [ |
| 3016 |
'updated' => $updated, |
| 3017 |
'cleared' => $cleared, |
| 3018 |
'indexed_count' => $indexed_count, |
| 3019 |
]; |
| 3020 |
} |
| 3021 |
|
| 3022 |
/** |
| 3023 |
* Sync cloud indexed status in wpforo_topics table |
| 3024 |
* |
| 3025 |
* Calls the /v1/rag/indexed-topics API endpoint to get all indexed |
| 3026 |
* topic IDs and updates the `cloud` column in wpforo_topics accordingly. |
| 3027 |
* Called when switching to cloud storage mode. |
| 3028 |
* |
| 3029 |
* @return array|WP_Error { |
| 3030 |
* @type int $updated Number of topics marked as indexed |
| 3031 |
* @type int $cleared Number of topics marked as not indexed |
| 3032 |
* } |
| 3033 |
*/ |
| 3034 |
public function sync_cloud_indexed_status() { |
| 3035 |
global $wpdb; |
| 3036 |
|
| 3037 |
// Get indexed topic IDs from cloud API |
| 3038 |
$indexed_topic_ids = $this->get_cloud_indexed_topic_ids(); |
| 3039 |
|
| 3040 |
if ( is_wp_error( $indexed_topic_ids ) ) { |
| 3041 |
return $indexed_topic_ids; |
| 3042 |
} |
| 3043 |
|
| 3044 |
$updated = 0; |
| 3045 |
$cleared = 0; |
| 3046 |
|
| 3047 |
if ( ! empty( $indexed_topic_ids ) ) { |
| 3048 |
// Set cloud=1 for topics that are indexed |
| 3049 |
$placeholders = implode( ',', array_fill( 0, count( $indexed_topic_ids ), '%d' ) ); |
| 3050 |
$updated = $wpdb->query( |
| 3051 |
$wpdb->prepare( |
| 3052 |
"UPDATE `" . WPF()->tables->topics . "` SET `cloud` = 1 WHERE topicid IN ($placeholders)", |
| 3053 |
$indexed_topic_ids |
| 3054 |
) |
| 3055 |
); |
| 3056 |
|
| 3057 |
// Set cloud=0 for topics that are NOT indexed |
| 3058 |
$cleared = $wpdb->query( |
| 3059 |
$wpdb->prepare( |
| 3060 |
"UPDATE `" . WPF()->tables->topics . "` SET `cloud` = 0 WHERE topicid NOT IN ($placeholders)", |
| 3061 |
$indexed_topic_ids |
| 3062 |
) |
| 3063 |
); |
| 3064 |
} else { |
| 3065 |
// No indexed topics - set all to 0 |
| 3066 |
$cleared = $wpdb->query( |
| 3067 |
"UPDATE `" . WPF()->tables->topics . "` SET `cloud` = 0 WHERE `cloud` = 1" |
| 3068 |
); |
| 3069 |
} |
| 3070 |
|
| 3071 |
// Clear topic cache to ensure wpforo_topic() returns fresh cloud values |
| 3072 |
// This is critical for force_reindex to work correctly |
| 3073 |
if ( $updated > 0 || $cleared > 0 ) { |
| 3074 |
wpforo_clean_cache( 'topic' ); |
| 3075 |
// Clear indexing stats cache for fresh counts |
| 3076 |
$this->clear_indexing_stats_cache(); |
| 3077 |
} |
| 3078 |
|
| 3079 |
return [ |
| 3080 |
'updated' => (int) $updated, |
| 3081 |
'cleared' => (int) $cleared, |
| 3082 |
'indexed_count' => count( $indexed_topic_ids ) |
| 3083 |
]; |
| 3084 |
} |
| 3085 |
|
| 3086 |
/** |
| 3087 |
* Get all indexed topic IDs from cloud storage |
| 3088 |
* |
| 3089 |
* Calls the /v1/rag/indexed-topics API endpoint. |
| 3090 |
* |
| 3091 |
* @return array|WP_Error Array of topic IDs or error |
| 3092 |
*/ |
| 3093 |
public function get_cloud_indexed_topic_ids() { |
| 3094 |
$ai_client = $this->get_ai_client(); |
| 3095 |
if ( ! $ai_client ) { |
| 3096 |
return new \WP_Error( 'no_ai_client', wpforo_phrase( 'AI client not available', false ) ); |
| 3097 |
} |
| 3098 |
|
| 3099 |
$response = $ai_client->api_get( '/rag/indexed-topics' ); |
| 3100 |
|
| 3101 |
if ( is_wp_error( $response ) ) { |
| 3102 |
return $response; |
| 3103 |
} |
| 3104 |
|
| 3105 |
if ( ! isset( $response['topic_ids'] ) || ! is_array( $response['topic_ids'] ) ) { |
| 3106 |
return new \WP_Error( 'invalid_response', wpforo_phrase( 'Invalid response from API', false ) ); |
| 3107 |
} |
| 3108 |
|
| 3109 |
return $response['topic_ids']; |
| 3110 |
} |
| 3111 |
|
| 3112 |
/** |
| 3113 |
* Mark a topic as indexed in the current storage mode |
| 3114 |
* |
| 3115 |
* Updates the `local` or `cloud` column based on current mode. |
| 3116 |
* |
| 3117 |
* @param int $topicid Topic ID |
| 3118 |
* @return bool Success |
| 3119 |
*/ |
| 3120 |
public function mark_topic_indexed( $topicid ) { |
| 3121 |
global $wpdb; |
| 3122 |
|
| 3123 |
$column = $this->is_local_mode() ? 'local' : 'cloud'; |
| 3124 |
|
| 3125 |
return $wpdb->update( |
| 3126 |
WPF()->tables->topics, |
| 3127 |
[ $column => 1 ], |
| 3128 |
[ 'topicid' => $topicid ], |
| 3129 |
[ '%d' ], |
| 3130 |
[ '%d' ] |
| 3131 |
) !== false; |
| 3132 |
} |
| 3133 |
|
| 3134 |
/** |
| 3135 |
* Mark a topic as not indexed in the current storage mode |
| 3136 |
* |
| 3137 |
* Updates the `local` or `cloud` column based on current mode. |
| 3138 |
* |
| 3139 |
* @param int $topicid Topic ID |
| 3140 |
* @return bool Success |
| 3141 |
*/ |
| 3142 |
public function mark_topic_not_indexed( $topicid ) { |
| 3143 |
global $wpdb; |
| 3144 |
|
| 3145 |
$column = $this->is_local_mode() ? 'local' : 'cloud'; |
| 3146 |
|
| 3147 |
return $wpdb->update( |
| 3148 |
WPF()->tables->topics, |
| 3149 |
[ $column => 0 ], |
| 3150 |
[ 'topicid' => $topicid ], |
| 3151 |
[ '%d' ], |
| 3152 |
[ '%d' ] |
| 3153 |
) !== false; |
| 3154 |
} |
| 3155 |
|
| 3156 |
/** |
| 3157 |
* Get topics that are not indexed in the current storage mode |
| 3158 |
* |
| 3159 |
* Uses the `local` or `cloud` column based on current mode. |
| 3160 |
* |
| 3161 |
* @param int $limit Maximum number of topics to return (0 = no limit) |
| 3162 |
* @param int $offset Offset for pagination |
| 3163 |
* @return array Array of topic IDs |
| 3164 |
*/ |
| 3165 |
public function get_unindexed_topic_ids( $limit = 0, $offset = 0 ) { |
| 3166 |
global $wpdb; |
| 3167 |
|
| 3168 |
$column = $this->is_local_mode() ? 'local' : 'cloud'; |
| 3169 |
|
| 3170 |
// Exclude private topics (private = 1) and unapproved topics (status != 0) |
| 3171 |
$sql = "SELECT topicid FROM `" . WPF()->tables->topics . "` WHERE `{$column}` = 0 AND `status` = 0 AND `private` = 0 ORDER BY topicid ASC"; |
| 3172 |
|
| 3173 |
if ( $limit > 0 ) { |
| 3174 |
$sql .= $wpdb->prepare( " LIMIT %d OFFSET %d", $limit, $offset ); |
| 3175 |
} |
| 3176 |
|
| 3177 |
return array_map( 'intval', $wpdb->get_col( $sql ) ); |
| 3178 |
} |
| 3179 |
|
| 3180 |
/** |
| 3181 |
* Count topics that are not indexed in the current storage mode |
| 3182 |
* |
| 3183 |
* @return int Count of unindexed topics |
| 3184 |
*/ |
| 3185 |
public function count_unindexed_topics() { |
| 3186 |
global $wpdb; |
| 3187 |
|
| 3188 |
$column = $this->is_local_mode() ? 'local' : 'cloud'; |
| 3189 |
|
| 3190 |
// Exclude private topics (private = 1) and unapproved topics (status != 0) |
| 3191 |
return (int) $wpdb->get_var( |
| 3192 |
"SELECT COUNT(*) FROM `" . WPF()->tables->topics . "` WHERE `{$column}` = 0 AND `status` = 0 AND `private` = 0" |
| 3193 |
); |
| 3194 |
} |
| 3195 |
|
| 3196 |
// ========================================================================= |
| 3197 |
// INDEXING STATUS ANALYSIS |
| 3198 |
// ========================================================================= |
| 3199 |
|
| 3200 |
/** |
| 3201 |
* Get a detailed breakdown of topics by indexing status |
| 3202 |
* |
| 3203 |
* Analyzes all topics and categorizes them by why they are or aren't indexed. |
| 3204 |
* This helps users understand why some topics might not appear in AI search. |
| 3205 |
* |
| 3206 |
* Categories: |
| 3207 |
* - indexed: Successfully indexed in current storage mode |
| 3208 |
* - pending: Eligible for indexing but not yet indexed |
| 3209 |
* - private: Private topics (excluded from indexing) |
| 3210 |
* - unapproved: Unapproved topics (excluded from indexing) |
| 3211 |
* |
| 3212 |
* @return array Breakdown with counts for each category |
| 3213 |
*/ |
| 3214 |
public function get_indexing_status_breakdown() { |
| 3215 |
global $wpdb; |
| 3216 |
|
| 3217 |
$column = $this->is_local_mode() ? 'local' : 'cloud'; |
| 3218 |
$topics_table = WPF()->tables->topics; |
| 3219 |
|
| 3220 |
// Get counts for each category in a single query |
| 3221 |
$results = $wpdb->get_row( |
| 3222 |
"SELECT |
| 3223 |
COUNT(*) as total, |
| 3224 |
SUM(CASE WHEN `{$column}` = 1 AND `status` = 0 AND `private` = 0 THEN 1 ELSE 0 END) as indexed, |
| 3225 |
SUM(CASE WHEN `{$column}` = 0 AND `status` = 0 AND `private` = 0 THEN 1 ELSE 0 END) as pending, |
| 3226 |
SUM(CASE WHEN `private` = 1 THEN 1 ELSE 0 END) as private_topics, |
| 3227 |
SUM(CASE WHEN `status` != 0 AND `private` = 0 THEN 1 ELSE 0 END) as unapproved |
| 3228 |
FROM `{$topics_table}`", |
| 3229 |
ARRAY_A |
| 3230 |
); |
| 3231 |
|
| 3232 |
if ( ! $results ) { |
| 3233 |
return [ |
| 3234 |
'total' => 0, |
| 3235 |
'indexed' => 0, |
| 3236 |
'pending' => 0, |
| 3237 |
'private' => 0, |
| 3238 |
'unapproved' => 0, |
| 3239 |
'storage_mode' => $this->get_storage_mode(), |
| 3240 |
]; |
| 3241 |
} |
| 3242 |
|
| 3243 |
return [ |
| 3244 |
'total' => (int) $results['total'], |
| 3245 |
'indexed' => (int) $results['indexed'], |
| 3246 |
'pending' => (int) $results['pending'], |
| 3247 |
'private' => (int) $results['private_topics'], |
| 3248 |
'unapproved' => (int) $results['unapproved'], |
| 3249 |
'storage_mode' => $this->get_storage_mode(), |
| 3250 |
]; |
| 3251 |
} |
| 3252 |
|
| 3253 |
/** |
| 3254 |
* Get sample topics that are pending indexing |
| 3255 |
* |
| 3256 |
* Returns a limited sample of topics that should be indexed but aren't. |
| 3257 |
* Useful for debugging why topics aren't being picked up. |
| 3258 |
* |
| 3259 |
* @param int $limit Maximum number of topics to return |
| 3260 |
* @return array Array of topic data with basic info |
| 3261 |
*/ |
| 3262 |
public function get_pending_topics_sample( $limit = 10 ) { |
| 3263 |
global $wpdb; |
| 3264 |
|
| 3265 |
$column = $this->is_local_mode() ? 'local' : 'cloud'; |
| 3266 |
$topics_table = WPF()->tables->topics; |
| 3267 |
|
| 3268 |
$topics = $wpdb->get_results( |
| 3269 |
$wpdb->prepare( |
| 3270 |
"SELECT topicid, title, forumid, posts, created, modified |
| 3271 |
FROM `{$topics_table}` |
| 3272 |
WHERE `{$column}` = 0 |
| 3273 |
AND `status` = 0 |
| 3274 |
AND `private` = 0 |
| 3275 |
ORDER BY topicid DESC |
| 3276 |
LIMIT %d", |
| 3277 |
$limit |
| 3278 |
), |
| 3279 |
ARRAY_A |
| 3280 |
); |
| 3281 |
|
| 3282 |
return $topics ?: []; |
| 3283 |
} |
| 3284 |
|
| 3285 |
// ========================================================================= |
| 3286 |
// DEBUG LOGGING |
| 3287 |
// ========================================================================= |
| 3288 |
|
| 3289 |
/** |
| 3290 |
* Check if debug mode is enabled |
| 3291 |
* |
| 3292 |
* @return bool True if debug mode is enabled |
| 3293 |
*/ |
| 3294 |
private function is_debug_mode() { |
| 3295 |
return defined( 'WP_DEBUG' ) && WP_DEBUG && defined( 'WPFORO_AI_DEBUG' ) && WPFORO_AI_DEBUG; |
| 3296 |
} |
| 3297 |
|
| 3298 |
/** |
| 3299 |
* Log informational message (only in debug mode) |
| 3300 |
* |
| 3301 |
* @param string $context Log context identifier |
| 3302 |
* @param array $data Optional data to log |
| 3303 |
*/ |
| 3304 |
private function log_info( $context, $data = [] ) { |
| 3305 |
if ( ! $this->is_debug_mode() ) { |
| 3306 |
return; |
| 3307 |
} |
| 3308 |
|
| 3309 |
\wpforo_ai_log( 'info', sprintf( |
| 3310 |
'%s | Data: %s', |
| 3311 |
$context, |
| 3312 |
wp_json_encode( $data ) |
| 3313 |
), 'VectorStorage' ); |
| 3314 |
} |
| 3315 |
} |
| 3316 |
|
