| 1 |
<?php |
| 2 |
|
| 3 |
namespace wpforo\classes; |
| 4 |
|
| 5 |
/** |
| 6 |
* WordPress Content Type Indexer |
| 7 |
* |
| 8 |
* Handles AI content indexing for WordPress posts, pages, and custom post types. |
| 9 |
* Uses the existing RAG infrastructure via /rag/wordpress/* API endpoints. |
| 10 |
* |
| 11 |
* Content Types: |
| 12 |
* - Posts, Pages, and any public custom post types |
| 13 |
* - Excludes internal types (revisions, nav_menu_item, etc.) |
| 14 |
* |
| 15 |
* @since 3.0.0 |
| 16 |
*/ |
| 17 |
class AIWordPressIndexer { |
| 18 |
|
| 19 |
/** |
| 20 |
* Internal/structural post types to skip |
| 21 |
* These should never be indexed as they're not user-facing content. |
| 22 |
* |
| 23 |
* @var array |
| 24 |
*/ |
| 25 |
private $skip_post_types = [ |
| 26 |
// WordPress core internal |
| 27 |
'attachment', |
| 28 |
'revision', |
| 29 |
'nav_menu_item', |
| 30 |
'wp_block', |
| 31 |
'wp_template', |
| 32 |
'wp_template_part', |
| 33 |
'wp_navigation', |
| 34 |
'wp_font_family', |
| 35 |
'wp_font_face', |
| 36 |
'oembed_cache', |
| 37 |
'user_request', |
| 38 |
'wp_global_styles', |
| 39 |
'custom_css', |
| 40 |
'customize_changeset', |
| 41 |
|
| 42 |
// ACF internal |
| 43 |
'acf-field-group', |
| 44 |
'acf-field', |
| 45 |
'acf-post-type', |
| 46 |
'acf-taxonomy', |
| 47 |
|
| 48 |
// wpForo posts (indexed separately via forum indexing) |
| 49 |
'wpforo_post', |
| 50 |
]; |
| 51 |
|
| 52 |
/** |
| 53 |
* WooCommerce post types to skip (temporary exclusion) |
| 54 |
* TODO: Remove this exclusion when WooCommerce support is implemented |
| 55 |
* |
| 56 |
* @var array |
| 57 |
*/ |
| 58 |
private $skip_post_types_woocommerce = [ |
| 59 |
'product', // Main products |
| 60 |
'product_variation', // Product variations |
| 61 |
'shop_order', // Orders (legacy, pre-HPOS) |
| 62 |
'shop_order_refund', // Order refunds |
| 63 |
'shop_coupon', // Coupons |
| 64 |
'shop_order_placehold', // Order placeholders |
| 65 |
'shop_webhook', // Webhooks |
| 66 |
]; |
| 67 |
|
| 68 |
/** |
| 69 |
* Internal/structural taxonomies to skip |
| 70 |
* |
| 71 |
* @var array |
| 72 |
*/ |
| 73 |
private $skip_taxonomies = [ |
| 74 |
// WordPress core internal |
| 75 |
'nav_menu', |
| 76 |
'link_category', |
| 77 |
'post_format', |
| 78 |
'wp_theme', |
| 79 |
'wp_template_part_area', |
| 80 |
'wp_pattern_category', |
| 81 |
]; |
| 82 |
|
| 83 |
/** |
| 84 |
* WooCommerce taxonomies to skip (temporary exclusion) |
| 85 |
* TODO: Remove this exclusion when WooCommerce support is implemented |
| 86 |
* |
| 87 |
* @var array |
| 88 |
*/ |
| 89 |
private $skip_taxonomies_woocommerce = [ |
| 90 |
'product_cat', // Product categories |
| 91 |
'product_tag', // Product tags |
| 92 |
'product_type', // Product types (simple, variable, etc.) |
| 93 |
'product_visibility', // Product visibility |
| 94 |
'product_shipping_class', // Shipping classes |
| 95 |
// Note: pa_* (product attributes) are handled dynamically in get_taxonomies_for_post_type() |
| 96 |
]; |
| 97 |
|
| 98 |
/** |
| 99 |
* Batch size for indexing |
| 100 |
* |
| 101 |
* @var int |
| 102 |
*/ |
| 103 |
const BATCH_SIZE = 50; |
| 104 |
|
| 105 |
/** |
| 106 |
* Minimum content length after stripping HTML and shortcodes |
| 107 |
* |
| 108 |
* @var int |
| 109 |
*/ |
| 110 |
const MIN_CONTENT_LENGTH = 10; |
| 111 |
|
| 112 |
/** |
| 113 |
* Constructor |
| 114 |
*/ |
| 115 |
public function __construct() { |
| 116 |
// Register admin-only AJAX handlers |
| 117 |
if ( is_admin() ) { |
| 118 |
add_action( 'wp_ajax_wpforo_ai_wp_get_post_types', [ $this, 'ajax_get_post_types' ] ); |
| 119 |
add_action( 'wp_ajax_wpforo_ai_wp_get_taxonomies', [ $this, 'ajax_get_taxonomies' ] ); |
| 120 |
add_action( 'wp_ajax_wpforo_ai_wp_get_taxonomy_terms', [ $this, 'ajax_get_taxonomy_terms' ] ); |
| 121 |
add_action( 'wp_ajax_wpforo_ai_wp_get_indexing_status', [ $this, 'ajax_get_indexing_status' ] ); |
| 122 |
add_action( 'wp_ajax_wpforo_ai_wp_index_by_taxonomy', [ $this, 'ajax_index_by_taxonomy' ] ); |
| 123 |
add_action( 'wp_ajax_wpforo_ai_wp_index_custom', [ $this, 'ajax_index_custom' ] ); |
| 124 |
add_action( 'wp_ajax_wpforo_ai_wp_delete_content', [ $this, 'ajax_delete_content' ] ); |
| 125 |
} |
| 126 |
|
| 127 |
// Register WP Cron handler for batch processing |
| 128 |
// IMPORTANT: Must be registered unconditionally (not only in admin context) |
| 129 |
// because WP Cron runs in a separate request where is_admin() returns FALSE |
| 130 |
add_action( 'wpforo_ai_process_wp_batch', [ $this, 'process_batch_with_lock' ] ); |
| 131 |
} |
| 132 |
|
| 133 |
/** |
| 134 |
* Wrapper for process_batch() with transient-based lock. |
| 135 |
* |
| 136 |
* Prevents concurrent batch processing when inline cron nudge fires |
| 137 |
* simultaneously with WP-Cron or multiple admin page refreshes. |
| 138 |
* Follows same pattern as VectorStorageManager::cron_process_queue_mode(). |
| 139 |
* |
| 140 |
* @return array|WP_Error Result from process_batch() or lock status |
| 141 |
*/ |
| 142 |
public function process_batch_with_lock() { |
| 143 |
$lock_key = 'wpforo_ai_wp_indexing_lock'; |
| 144 |
|
| 145 |
// Check if already processing |
| 146 |
if ( get_transient( $lock_key ) ) { |
| 147 |
// Reschedule as backup (same pattern as forum indexing) |
| 148 |
if ( ! wp_next_scheduled( 'wpforo_ai_process_wp_batch' ) ) { |
| 149 |
wp_schedule_single_event( time() + 60, 'wpforo_ai_process_wp_batch' ); |
| 150 |
} |
| 151 |
return [ 'status' => 'locked', 'message' => 'Another batch is being processed' ]; |
| 152 |
} |
| 153 |
|
| 154 |
// Acquire lock (300s TTL - matches forum indexing) |
| 155 |
set_transient( $lock_key, 'processing_' . time(), 300 ); |
| 156 |
|
| 157 |
// Process the batch |
| 158 |
$result = $this->process_batch(); |
| 159 |
|
| 160 |
// Release lock |
| 161 |
delete_transient( $lock_key ); |
| 162 |
|
| 163 |
return $result; |
| 164 |
} |
| 165 |
|
| 166 |
/** |
| 167 |
* Check if using local storage mode for WordPress content |
| 168 |
* |
| 169 |
* WordPress content indexing uses a global storage mode setting. |
| 170 |
* This is independent of forum boards - it's a site-wide setting |
| 171 |
* stored in wpforo_ai_storage_mode_0. |
| 172 |
* |
| 173 |
* @return bool True if local mode, false if cloud mode |
| 174 |
*/ |
| 175 |
private function is_local_storage_mode() { |
| 176 |
$storage_mode = get_option( 'wpforo_ai_storage_mode_0', 'local' ); |
| 177 |
return $storage_mode === 'local'; |
| 178 |
} |
| 179 |
|
| 180 |
/** |
| 181 |
* Get the local storage instance for WordPress content |
| 182 |
* |
| 183 |
* Uses the VectorStorageManager's local storage instance which has |
| 184 |
* proper table references initialized. WordPress content uses the |
| 185 |
* same embeddings table as forum content. |
| 186 |
* |
| 187 |
* @return \wpforo\classes\VectorStorageLocal|null Local storage instance or null |
| 188 |
*/ |
| 189 |
private function get_local_storage() { |
| 190 |
if ( ! isset( WPF()->vector_storage ) || ! WPF()->vector_storage ) { |
| 191 |
return null; |
| 192 |
} |
| 193 |
return WPF()->vector_storage->get_local_storage(); |
| 194 |
} |
| 195 |
|
| 196 |
/** |
| 197 |
* Get public post types available for indexing |
| 198 |
* |
| 199 |
* @return array Array of post type objects with name, label, and count |
| 200 |
*/ |
| 201 |
public function get_public_post_types() { |
| 202 |
// 'publicly_queryable' => true, 'public' => true |
| 203 |
$post_type_arguments = apply_filters('wpforo_ai_wp_indexing_post_types', [ 'public' => true ]); |
| 204 |
$post_types = get_post_types($post_type_arguments, 'objects'); |
| 205 |
|
| 206 |
// Also include 'page' which has publicly_queryable = false by default |
| 207 |
$page_type = get_post_type_object( 'page' ); |
| 208 |
if ( $page_type ) { |
| 209 |
$post_types['page'] = $page_type; |
| 210 |
} |
| 211 |
|
| 212 |
// Also include 'post' which may have different settings |
| 213 |
$post_type = get_post_type_object( 'post' ); |
| 214 |
if ( $post_type ) { |
| 215 |
$post_types['post'] = $post_type; |
| 216 |
} |
| 217 |
|
| 218 |
// Combine skip lists |
| 219 |
$all_skip_types = array_merge( $this->skip_post_types, $this->skip_post_types_woocommerce ); |
| 220 |
|
| 221 |
$result = []; |
| 222 |
foreach ( $post_types as $name => $type ) { |
| 223 |
// Skip internal/structural types and WooCommerce types |
| 224 |
if ( in_array( $name, $all_skip_types, true ) ) { |
| 225 |
continue; |
| 226 |
} |
| 227 |
|
| 228 |
// Count indexable published posts (for display accuracy) |
| 229 |
$indexable_count = $this->count_indexable_posts( $name ); |
| 230 |
|
| 231 |
// Include ALL public post types, regardless of current indexable count |
| 232 |
// Content quality filtering happens at indexing time, not at type listing |
| 233 |
$result[] = [ |
| 234 |
'name' => $name, |
| 235 |
'label' => $type->labels->name, |
| 236 |
'count' => $indexable_count, // Shows indexable posts (may be 0) |
| 237 |
]; |
| 238 |
} |
| 239 |
|
| 240 |
return $result; |
| 241 |
} |
| 242 |
|
| 243 |
/** |
| 244 |
* Count indexable published posts for a post type |
| 245 |
* |
| 246 |
* Only counts posts with meaningful textual content. |
| 247 |
* Results are cached for 5 minutes to improve performance. |
| 248 |
* |
| 249 |
* @param string $post_type Post type name |
| 250 |
* @return int Number of indexable posts |
| 251 |
*/ |
| 252 |
private function count_indexable_posts( $post_type ) { |
| 253 |
// Check transient cache first |
| 254 |
$cache_key = 'wpforo_ai_indexable_count_' . sanitize_key( $post_type ); |
| 255 |
$cached = get_transient( $cache_key ); |
| 256 |
if ( false !== $cached ) { |
| 257 |
return (int) $cached; |
| 258 |
} |
| 259 |
|
| 260 |
$query = new \WP_Query( [ |
| 261 |
'post_type' => $post_type, |
| 262 |
'post_status' => 'publish', |
| 263 |
'posts_per_page' => -1, |
| 264 |
'no_found_rows' => true, |
| 265 |
'update_post_meta_cache' => false, |
| 266 |
'update_post_term_cache' => false, |
| 267 |
'fields' => 'ids', // Only fetch IDs first for efficiency |
| 268 |
] ); |
| 269 |
|
| 270 |
// If no posts, return 0 |
| 271 |
if ( empty( $query->posts ) ) { |
| 272 |
set_transient( $cache_key, 0, 5 * MINUTE_IN_SECONDS ); |
| 273 |
return 0; |
| 274 |
} |
| 275 |
|
| 276 |
// For small result sets, check all posts |
| 277 |
// For large sets, sample to estimate (check first 500) |
| 278 |
$post_ids = $query->posts; |
| 279 |
$total_posts = count( $post_ids ); |
| 280 |
|
| 281 |
if( apply_filters('wpforo_ai_filter_indexable_post_types', false, $post_ids, $post_type, $total_posts) ){ |
| 282 |
$sample_size = min( 100, $total_posts ); |
| 283 |
$indexable_count = 0; |
| 284 |
|
| 285 |
// Check sample of posts |
| 286 |
for ( $i = 0; $i < $sample_size; $i++ ) { |
| 287 |
$post = get_post( $post_ids[ $i ] ); |
| 288 |
if ( $post && $this->is_content_indexable( $post ) ) { |
| 289 |
$indexable_count++; |
| 290 |
} |
| 291 |
} |
| 292 |
|
| 293 |
// If we sampled, extrapolate the count |
| 294 |
if ( $sample_size < $total_posts ) { |
| 295 |
$ratio = $indexable_count / $sample_size; |
| 296 |
$indexable_count = (int) round( $total_posts * $ratio ); |
| 297 |
} |
| 298 |
} else { |
| 299 |
|
| 300 |
// Keep the original number of post type items |
| 301 |
$indexable_count = $total_posts; |
| 302 |
} |
| 303 |
|
| 304 |
|
| 305 |
|
| 306 |
set_transient( $cache_key, $indexable_count, 5 * MINUTE_IN_SECONDS ); |
| 307 |
|
| 308 |
return $indexable_count; |
| 309 |
} |
| 310 |
|
| 311 |
/** |
| 312 |
* Get public taxonomies for a post type |
| 313 |
* |
| 314 |
* @param string $post_type Post type name |
| 315 |
* @return array Array of taxonomy objects |
| 316 |
*/ |
| 317 |
public function get_taxonomies_for_post_type( $post_type ) { |
| 318 |
$taxonomies = get_object_taxonomies( $post_type, 'objects' ); |
| 319 |
|
| 320 |
// Combine skip lists |
| 321 |
$all_skip_taxonomies = array_merge( $this->skip_taxonomies, $this->skip_taxonomies_woocommerce ); |
| 322 |
|
| 323 |
$result = []; |
| 324 |
foreach ( $taxonomies as $name => $taxonomy ) { |
| 325 |
// Only include public taxonomies |
| 326 |
if ( ! $taxonomy->public ) { |
| 327 |
continue; |
| 328 |
} |
| 329 |
|
| 330 |
// Skip internal/structural taxonomies and WooCommerce taxonomies |
| 331 |
if ( in_array( $name, $all_skip_taxonomies, true ) ) { |
| 332 |
continue; |
| 333 |
} |
| 334 |
|
| 335 |
// Skip WooCommerce product attributes (pa_* prefix) |
| 336 |
// TODO: Remove this when WooCommerce support is implemented |
| 337 |
if ( strpos( $name, 'pa_' ) === 0 ) { |
| 338 |
continue; |
| 339 |
} |
| 340 |
|
| 341 |
// Get term count (show all terms, not just those with posts) |
| 342 |
$count = wp_count_terms( [ 'taxonomy' => $name, 'hide_empty' => false ] ); |
| 343 |
|
| 344 |
$result[] = [ |
| 345 |
'name' => $name, |
| 346 |
'label' => $taxonomy->labels->name, |
| 347 |
'term_count' => is_wp_error( $count ) ? 0 : (int) $count, |
| 348 |
]; |
| 349 |
} |
| 350 |
|
| 351 |
return $result; |
| 352 |
} |
| 353 |
|
| 354 |
/** |
| 355 |
* Get terms for a taxonomy |
| 356 |
* |
| 357 |
* @param string $taxonomy Taxonomy name |
| 358 |
* @param bool $hierarchical Whether to return hierarchical structure |
| 359 |
* @param array $post_types Post types to count (defaults to all public types using this taxonomy) |
| 360 |
* @return array Array of terms |
| 361 |
*/ |
| 362 |
public function get_taxonomy_terms( $taxonomy, $hierarchical = true, $post_types = [] ) { |
| 363 |
$args = [ |
| 364 |
'taxonomy' => $taxonomy, |
| 365 |
'hide_empty' => false, // Show all terms, not just those with posts |
| 366 |
'orderby' => 'name', |
| 367 |
'order' => 'ASC', |
| 368 |
]; |
| 369 |
|
| 370 |
if ( $hierarchical && is_taxonomy_hierarchical( $taxonomy ) ) { |
| 371 |
$args['parent'] = 0; // Get top-level terms first |
| 372 |
} |
| 373 |
|
| 374 |
$terms = get_terms( $args ); |
| 375 |
|
| 376 |
if ( is_wp_error( $terms ) ) { |
| 377 |
return []; |
| 378 |
} |
| 379 |
|
| 380 |
// Get post types that use this taxonomy if not specified |
| 381 |
if ( empty( $post_types ) ) { |
| 382 |
$tax_obj = get_taxonomy( $taxonomy ); |
| 383 |
$post_types = $tax_obj ? $tax_obj->object_type : [ 'post' ]; |
| 384 |
} |
| 385 |
|
| 386 |
// Get indexed post IDs for counting |
| 387 |
$indexed_post_ids = $this->get_indexed_post_ids(); |
| 388 |
|
| 389 |
$result = []; |
| 390 |
foreach ( $terms as $term ) { |
| 391 |
// Count only published posts for this term |
| 392 |
$published_count = $this->count_published_posts_in_term( $term->term_id, $taxonomy, $post_types ); |
| 393 |
// Count indexed posts in this term |
| 394 |
$indexed_count = $this->count_indexed_posts_in_term( $term->term_id, $taxonomy, $post_types, $indexed_post_ids ); |
| 395 |
|
| 396 |
$term_data = [ |
| 397 |
'term_id' => $term->term_id, |
| 398 |
'name' => $term->name, |
| 399 |
'slug' => $term->slug, |
| 400 |
'count' => $published_count, |
| 401 |
'indexed' => $indexed_count, |
| 402 |
]; |
| 403 |
|
| 404 |
// Get children for hierarchical taxonomies |
| 405 |
if ( $hierarchical && is_taxonomy_hierarchical( $taxonomy ) ) { |
| 406 |
$children = get_terms( [ |
| 407 |
'taxonomy' => $taxonomy, |
| 408 |
'hide_empty' => false, |
| 409 |
'parent' => $term->term_id, |
| 410 |
'orderby' => 'name', |
| 411 |
'order' => 'ASC', |
| 412 |
] ); |
| 413 |
|
| 414 |
if ( ! is_wp_error( $children ) && ! empty( $children ) ) { |
| 415 |
$term_data['children'] = []; |
| 416 |
foreach ( $children as $child ) { |
| 417 |
$child_published_count = $this->count_published_posts_in_term( $child->term_id, $taxonomy, $post_types ); |
| 418 |
$child_indexed_count = $this->count_indexed_posts_in_term( $child->term_id, $taxonomy, $post_types, $indexed_post_ids ); |
| 419 |
|
| 420 |
$term_data['children'][] = [ |
| 421 |
'term_id' => $child->term_id, |
| 422 |
'name' => $child->name, |
| 423 |
'slug' => $child->slug, |
| 424 |
'count' => $child_published_count, |
| 425 |
'indexed' => $child_indexed_count, |
| 426 |
]; |
| 427 |
} |
| 428 |
} |
| 429 |
} |
| 430 |
|
| 431 |
$result[] = $term_data; |
| 432 |
} |
| 433 |
|
| 434 |
return $result; |
| 435 |
} |
| 436 |
|
| 437 |
/** |
| 438 |
* Get all indexed WordPress post IDs |
| 439 |
* |
| 440 |
* @return array Array of indexed post IDs |
| 441 |
*/ |
| 442 |
private function get_indexed_post_ids() { |
| 443 |
if ( $this->is_local_storage_mode() ) { |
| 444 |
$local = $this->get_local_storage(); |
| 445 |
if ( ! $local ) { |
| 446 |
return []; |
| 447 |
} |
| 448 |
return $local->get_wp_indexed_post_ids(); |
| 449 |
} |
| 450 |
|
| 451 |
// Cloud mode: fetch indexed post IDs from backend API |
| 452 |
if ( ! WPF()->ai_client || ! WPF()->ai_client->is_service_available() ) { |
| 453 |
return []; |
| 454 |
} |
| 455 |
|
| 456 |
$response = WPF()->ai_client->api_get( '/rag/wordpress/indexed-posts' ); |
| 457 |
if ( is_wp_error( $response ) ) { |
| 458 |
return []; |
| 459 |
} |
| 460 |
|
| 461 |
return isset( $response['post_ids'] ) ? array_map( 'intval', $response['post_ids'] ) : []; |
| 462 |
} |
| 463 |
|
| 464 |
/** |
| 465 |
* Count indexed posts in a specific term |
| 466 |
* |
| 467 |
* @param int $term_id Term ID |
| 468 |
* @param string $taxonomy Taxonomy name |
| 469 |
* @param array $post_types Post types to check |
| 470 |
* @param array $indexed_post_ids Pre-fetched array of indexed post IDs |
| 471 |
* @return int Number of indexed posts in the term |
| 472 |
*/ |
| 473 |
private function count_indexed_posts_in_term( $term_id, $taxonomy, $post_types, $indexed_post_ids ) { |
| 474 |
if ( empty( $indexed_post_ids ) ) { |
| 475 |
return 0; |
| 476 |
} |
| 477 |
|
| 478 |
$query = new \WP_Query( [ |
| 479 |
'post_type' => $post_types, |
| 480 |
'post_status' => 'publish', |
| 481 |
'posts_per_page' => -1, |
| 482 |
'fields' => 'ids', |
| 483 |
'tax_query' => [ |
| 484 |
[ |
| 485 |
'taxonomy' => $taxonomy, |
| 486 |
'field' => 'term_id', |
| 487 |
'terms' => $term_id, |
| 488 |
], |
| 489 |
], |
| 490 |
'post__in' => $indexed_post_ids, |
| 491 |
] ); |
| 492 |
|
| 493 |
return $query->found_posts; |
| 494 |
} |
| 495 |
|
| 496 |
/** |
| 497 |
* Count published and indexable posts in a specific term |
| 498 |
* |
| 499 |
* Only counts posts that have meaningful textual content (not just shortcodes, |
| 500 |
* not binary data, and at least MIN_CONTENT_LENGTH characters). |
| 501 |
* Results are cached for 5 minutes to improve performance. |
| 502 |
* |
| 503 |
* @param int $term_id Term ID |
| 504 |
* @param string $taxonomy Taxonomy name |
| 505 |
* @param array $post_types Post types to count |
| 506 |
* @return int Number of indexable published posts |
| 507 |
*/ |
| 508 |
private function count_published_posts_in_term( $term_id, $taxonomy, $post_types = [ 'post' ] ) { |
| 509 |
// Check transient cache first |
| 510 |
$cache_key = 'wpforo_ai_term_count_' . $term_id . '_' . sanitize_key( $taxonomy ); |
| 511 |
$cached = get_transient( $cache_key ); |
| 512 |
if ( false !== $cached ) { |
| 513 |
return (int) $cached; |
| 514 |
} |
| 515 |
|
| 516 |
$query = new \WP_Query( [ |
| 517 |
'post_type' => $post_types, |
| 518 |
'post_status' => 'publish', |
| 519 |
'posts_per_page' => -1, |
| 520 |
'tax_query' => [ |
| 521 |
[ |
| 522 |
'taxonomy' => $taxonomy, |
| 523 |
'field' => 'term_id', |
| 524 |
'terms' => $term_id, |
| 525 |
], |
| 526 |
], |
| 527 |
'no_found_rows' => true, |
| 528 |
'update_post_meta_cache' => false, |
| 529 |
'update_post_term_cache' => false, |
| 530 |
'fields' => 'ids', // Only fetch IDs first |
| 531 |
] ); |
| 532 |
|
| 533 |
// If no posts, return 0 |
| 534 |
if ( empty( $query->posts ) ) { |
| 535 |
set_transient( $cache_key, 0, 5 * MINUTE_IN_SECONDS ); |
| 536 |
return 0; |
| 537 |
} |
| 538 |
|
| 539 |
$post_ids = $query->posts; |
| 540 |
$total_posts = count( $post_ids ); |
| 541 |
$sample_size = min( 200, $total_posts ); // Smaller sample for term counts |
| 542 |
$indexable_count = 0; |
| 543 |
|
| 544 |
for ( $i = 0; $i < $sample_size; $i++ ) { |
| 545 |
$post = get_post( $post_ids[ $i ] ); |
| 546 |
if ( $post && $this->is_content_indexable( $post ) ) { |
| 547 |
$indexable_count++; |
| 548 |
} |
| 549 |
} |
| 550 |
|
| 551 |
// Extrapolate if sampled |
| 552 |
if ( $sample_size < $total_posts ) { |
| 553 |
$ratio = $indexable_count / $sample_size; |
| 554 |
$indexable_count = (int) round( $total_posts * $ratio ); |
| 555 |
} |
| 556 |
|
| 557 |
set_transient( $cache_key, $indexable_count, 5 * MINUTE_IN_SECONDS ); |
| 558 |
|
| 559 |
return $indexable_count; |
| 560 |
} |
| 561 |
|
| 562 |
/** |
| 563 |
* Get WordPress posts for indexing |
| 564 |
* |
| 565 |
* @param array $args Query arguments |
| 566 |
* @return array Array of formatted post data |
| 567 |
*/ |
| 568 |
public function get_posts_for_indexing( $args = [] ) { |
| 569 |
$defaults = [ |
| 570 |
'post_type' => 'post', |
| 571 |
'post_status' => 'publish', |
| 572 |
'posts_per_page' => self::BATCH_SIZE, |
| 573 |
'paged' => 1, |
| 574 |
'orderby' => 'ID', |
| 575 |
'order' => 'ASC', |
| 576 |
]; |
| 577 |
|
| 578 |
$args = wp_parse_args( $args, $defaults ); |
| 579 |
|
| 580 |
// Ensure we only get published posts |
| 581 |
$args['post_status'] = 'publish'; |
| 582 |
|
| 583 |
$query = new \WP_Query( $args ); |
| 584 |
$posts = []; |
| 585 |
|
| 586 |
foreach ( $query->posts as $post ) { |
| 587 |
$posts[] = $this->format_post_for_indexing( $post ); |
| 588 |
} |
| 589 |
|
| 590 |
return [ |
| 591 |
'posts' => $posts, |
| 592 |
'total' => $query->found_posts, |
| 593 |
'total_pages' => $query->max_num_pages, |
| 594 |
'current' => $args['paged'], |
| 595 |
]; |
| 596 |
} |
| 597 |
|
| 598 |
/** |
| 599 |
* Format a WordPress post for indexing |
| 600 |
* |
| 601 |
* @param \WP_Post $post WordPress post object |
| 602 |
* @return array Formatted post data for API |
| 603 |
*/ |
| 604 |
public function format_post_for_indexing( $post ) { |
| 605 |
// Get taxonomy terms |
| 606 |
$taxonomies = get_object_taxonomies( $post->post_type, 'names' ); |
| 607 |
$taxonomy_terms = []; |
| 608 |
|
| 609 |
foreach ( $taxonomies as $taxonomy ) { |
| 610 |
$terms = wp_get_post_terms( $post->ID, $taxonomy, [ 'fields' => 'names' ] ); |
| 611 |
if ( ! is_wp_error( $terms ) && ! empty( $terms ) ) { |
| 612 |
$tax_obj = get_taxonomy( $taxonomy ); |
| 613 |
$tax_label = $tax_obj ? $tax_obj->labels->singular_name : $taxonomy; |
| 614 |
$taxonomy_terms[ $tax_label ] = $terms; |
| 615 |
} |
| 616 |
} |
| 617 |
|
| 618 |
// Get important post meta (configurable) |
| 619 |
$post_meta = $this->get_indexable_post_meta( $post ); |
| 620 |
|
| 621 |
return [ |
| 622 |
'post_id' => $post->ID, |
| 623 |
'post_type' => $post->post_type, |
| 624 |
'title' => $post->post_title, |
| 625 |
'content' => $post->post_content, |
| 626 |
'excerpt' => $post->post_excerpt, |
| 627 |
'author_id' => (int) $post->post_author, |
| 628 |
'post_status' => $post->post_status, |
| 629 |
'permalink' => get_permalink( $post->ID ), |
| 630 |
'created_at' => $post->post_date_gmt, |
| 631 |
'updated_at' => $post->post_modified_gmt, |
| 632 |
'taxonomy_terms' => $taxonomy_terms, |
| 633 |
'post_meta' => $post_meta, |
| 634 |
]; |
| 635 |
} |
| 636 |
|
| 637 |
/** |
| 638 |
* Get indexable post meta |
| 639 |
* |
| 640 |
* @param \WP_Post $post WordPress post object |
| 641 |
* @return array Filtered post meta |
| 642 |
*/ |
| 643 |
private function get_indexable_post_meta( $post ) { |
| 644 |
$meta = []; |
| 645 |
|
| 646 |
// WooCommerce product meta |
| 647 |
if ( $post->post_type === 'product' ) { |
| 648 |
$meta['_price'] = get_post_meta( $post->ID, '_price', true ); |
| 649 |
$meta['_sku'] = get_post_meta( $post->ID, '_sku', true ); |
| 650 |
$meta['_stock_status'] = get_post_meta( $post->ID, '_stock_status', true ); |
| 651 |
} |
| 652 |
|
| 653 |
// Allow plugins to add custom meta |
| 654 |
$meta = apply_filters( 'wpforo_ai_indexable_post_meta', $meta, $post ); |
| 655 |
|
| 656 |
// Remove empty values |
| 657 |
return array_filter( $meta, function( $v ) { |
| 658 |
return $v !== '' && $v !== null; |
| 659 |
} ); |
| 660 |
} |
| 661 |
|
| 662 |
/** |
| 663 |
* Check if post content is suitable for indexing |
| 664 |
* |
| 665 |
* Validates that content is: |
| 666 |
* - Not just shortcodes |
| 667 |
* - Has meaningful text (>= MIN_CONTENT_LENGTH chars after stripping) |
| 668 |
* - Is textual (not binary/garbage data) |
| 669 |
* |
| 670 |
* @param \WP_Post|int $post Post object or ID |
| 671 |
* @return bool True if content is indexable |
| 672 |
*/ |
| 673 |
public function is_content_indexable( $post ) { |
| 674 |
if ( is_numeric( $post ) ) { |
| 675 |
$post = get_post( $post ); |
| 676 |
} |
| 677 |
|
| 678 |
if ( ! $post || ! isset( $post->post_content ) ) { |
| 679 |
return false; |
| 680 |
} |
| 681 |
|
| 682 |
$content = $post->post_content; |
| 683 |
|
| 684 |
// Check for binary/non-textual content |
| 685 |
// Binary data often contains null bytes or high ratio of non-printable characters |
| 686 |
if ( $this->is_binary_content( $content ) ) { |
| 687 |
return false; |
| 688 |
} |
| 689 |
|
| 690 |
// Strip shortcodes first (e.g., [gallery], [contact-form-7 id="123"]) |
| 691 |
$content = strip_shortcodes( $content ); |
| 692 |
|
| 693 |
// Strip all HTML tags |
| 694 |
$content = wp_strip_all_tags( $content ); |
| 695 |
|
| 696 |
// Decode HTML entities |
| 697 |
$content = html_entity_decode( $content, ENT_QUOTES, 'UTF-8' ); |
| 698 |
|
| 699 |
// Normalize whitespace |
| 700 |
$content = preg_replace( '/\s+/', ' ', $content ); |
| 701 |
$content = trim( $content ); |
| 702 |
|
| 703 |
// Check minimum length |
| 704 |
$length = mb_strlen( $content, 'UTF-8' ); |
| 705 |
|
| 706 |
return $length >= self::MIN_CONTENT_LENGTH; |
| 707 |
} |
| 708 |
|
| 709 |
/** |
| 710 |
* Check if content appears to be binary/non-textual data |
| 711 |
* |
| 712 |
* @param string $content Content to check |
| 713 |
* @return bool True if content appears to be binary |
| 714 |
*/ |
| 715 |
private function is_binary_content( $content ) { |
| 716 |
if ( empty( $content ) ) { |
| 717 |
return false; |
| 718 |
} |
| 719 |
|
| 720 |
// Check for null bytes (common in binary data) |
| 721 |
if ( strpos( $content, "\0" ) !== false ) { |
| 722 |
return true; |
| 723 |
} |
| 724 |
|
| 725 |
// Sample the content (check first 1000 bytes for performance) |
| 726 |
$sample = substr( $content, 0, 1000 ); |
| 727 |
$sample_length = strlen( $sample ); |
| 728 |
|
| 729 |
if ( $sample_length === 0 ) { |
| 730 |
return false; |
| 731 |
} |
| 732 |
|
| 733 |
// Count non-printable characters (excluding common whitespace) |
| 734 |
$non_printable = 0; |
| 735 |
for ( $i = 0; $i < $sample_length; $i++ ) { |
| 736 |
$ord = ord( $sample[ $i ] ); |
| 737 |
// Allow: tab (9), newline (10), carriage return (13), space and above (32-126) |
| 738 |
// Allow extended ASCII/UTF-8 (128+) |
| 739 |
if ( $ord < 9 || ( $ord > 13 && $ord < 32 ) || ( $ord > 126 && $ord < 128 ) ) { |
| 740 |
$non_printable++; |
| 741 |
} |
| 742 |
} |
| 743 |
|
| 744 |
// If more than 10% is non-printable, likely binary |
| 745 |
$ratio = $non_printable / $sample_length; |
| 746 |
|
| 747 |
return $ratio > 0.1; |
| 748 |
} |
| 749 |
|
| 750 |
/** |
| 751 |
* Get the clean text content for a post (for display/counting purposes) |
| 752 |
* |
| 753 |
* @param \WP_Post|int $post Post object or ID |
| 754 |
* @return string Clean text content |
| 755 |
*/ |
| 756 |
public function get_clean_content( $post ) { |
| 757 |
if ( is_numeric( $post ) ) { |
| 758 |
$post = get_post( $post ); |
| 759 |
} |
| 760 |
|
| 761 |
if ( ! $post || ! isset( $post->post_content ) ) { |
| 762 |
return ''; |
| 763 |
} |
| 764 |
|
| 765 |
$content = $post->post_content; |
| 766 |
$content = strip_shortcodes( $content ); |
| 767 |
$content = wp_strip_all_tags( $content ); |
| 768 |
$content = html_entity_decode( $content, ENT_QUOTES, 'UTF-8' ); |
| 769 |
$content = preg_replace( '/\s+/', ' ', $content ); |
| 770 |
|
| 771 |
return trim( $content ); |
| 772 |
} |
| 773 |
|
| 774 |
/** |
| 775 |
* Index posts by taxonomy term(s) |
| 776 |
* |
| 777 |
* @param string $taxonomy Taxonomy name |
| 778 |
* @param int|array $term_ids Term ID or array of term IDs |
| 779 |
* @param array $post_types Post types to index |
| 780 |
* @param string $date_from Optional start date (Y-m-d format) |
| 781 |
* @param string $date_to Optional end date (Y-m-d format) |
| 782 |
* @return array|WP_Error Result or error |
| 783 |
*/ |
| 784 |
public function index_by_taxonomy( $taxonomy, $term_ids, $post_types = [ 'post' ], $date_from = '', $date_to = '' ) { |
| 785 |
if ( ! WPF()->ai_client || ! WPF()->ai_client->is_service_available() ) { |
| 786 |
return new \WP_Error( 'not_connected', __( 'AI service is not connected', 'wpforo' ) ); |
| 787 |
} |
| 788 |
|
| 789 |
// Ensure term_ids is an array |
| 790 |
$term_ids = (array) $term_ids; |
| 791 |
$term_ids = array_map( 'intval', $term_ids ); |
| 792 |
$term_ids = array_filter( $term_ids ); // Remove zeros |
| 793 |
|
| 794 |
if ( empty( $term_ids ) ) { |
| 795 |
return new \WP_Error( 'no_terms', __( 'No valid terms specified', 'wpforo' ) ); |
| 796 |
} |
| 797 |
|
| 798 |
// Get all posts in these terms |
| 799 |
$args = [ |
| 800 |
'post_type' => $post_types, |
| 801 |
'posts_per_page' => -1, // Get all |
| 802 |
'post_status' => 'publish', |
| 803 |
'tax_query' => [ |
| 804 |
[ |
| 805 |
'taxonomy' => $taxonomy, |
| 806 |
'field' => 'term_id', |
| 807 |
'terms' => $term_ids, |
| 808 |
], |
| 809 |
], |
| 810 |
'fields' => 'ids', |
| 811 |
]; |
| 812 |
|
| 813 |
// Add date range filter if specified |
| 814 |
if ( ! empty( $date_from ) ) { |
| 815 |
$args['date_query'][] = [ |
| 816 |
'after' => $date_from, |
| 817 |
'inclusive' => true, |
| 818 |
]; |
| 819 |
} |
| 820 |
|
| 821 |
if ( ! empty( $date_to ) ) { |
| 822 |
$args['date_query'][] = [ |
| 823 |
'before' => $date_to, |
| 824 |
'inclusive' => true, |
| 825 |
]; |
| 826 |
} |
| 827 |
|
| 828 |
$query = new \WP_Query( $args ); |
| 829 |
$post_ids = $query->posts; |
| 830 |
|
| 831 |
if ( empty( $post_ids ) ) { |
| 832 |
return new \WP_Error( 'no_posts', __( 'No posts found in this term', 'wpforo' ) ); |
| 833 |
} |
| 834 |
|
| 835 |
// Queue posts for batch indexing |
| 836 |
return $this->queue_posts_for_indexing( $post_ids ); |
| 837 |
} |
| 838 |
|
| 839 |
/** |
| 840 |
* Index posts with custom filters |
| 841 |
* |
| 842 |
* @param array $params Custom indexing parameters |
| 843 |
* @return array|WP_Error Result or error |
| 844 |
*/ |
| 845 |
public function index_custom( $params ) { |
| 846 |
if ( ! WPF()->ai_client || ! WPF()->ai_client->is_service_available() ) { |
| 847 |
return new \WP_Error( 'not_connected', __( 'AI service is not connected', 'wpforo' ) ); |
| 848 |
} |
| 849 |
|
| 850 |
$args = [ |
| 851 |
'post_type' => isset( $params['post_types'] ) ? $params['post_types'] : [ 'post' ], |
| 852 |
'posts_per_page' => -1, |
| 853 |
'post_status' => 'publish', |
| 854 |
'fields' => 'ids', |
| 855 |
]; |
| 856 |
|
| 857 |
// Date range filter |
| 858 |
if ( ! empty( $params['date_from'] ) ) { |
| 859 |
$args['date_query'][] = [ |
| 860 |
'after' => $params['date_from'], |
| 861 |
'inclusive' => true, |
| 862 |
]; |
| 863 |
} |
| 864 |
|
| 865 |
if ( ! empty( $params['date_to'] ) ) { |
| 866 |
$args['date_query'][] = [ |
| 867 |
'before' => $params['date_to'], |
| 868 |
'inclusive' => true, |
| 869 |
]; |
| 870 |
} |
| 871 |
|
| 872 |
// Specific post IDs |
| 873 |
if ( ! empty( $params['post_ids'] ) ) { |
| 874 |
$args['post__in'] = array_map( 'intval', (array) $params['post_ids'] ); |
| 875 |
} |
| 876 |
|
| 877 |
// Author filter |
| 878 |
if ( ! empty( $params['author'] ) ) { |
| 879 |
$args['author'] = intval( $params['author'] ); |
| 880 |
} |
| 881 |
|
| 882 |
$query = new \WP_Query( $args ); |
| 883 |
$post_ids = $query->posts; |
| 884 |
|
| 885 |
if ( empty( $post_ids ) ) { |
| 886 |
return new \WP_Error( 'no_posts', __( 'No posts found matching criteria', 'wpforo' ) ); |
| 887 |
} |
| 888 |
|
| 889 |
return $this->queue_posts_for_indexing( $post_ids ); |
| 890 |
} |
| 891 |
|
| 892 |
/** |
| 893 |
* Queue posts for batch indexing |
| 894 |
* |
| 895 |
* @param array $post_ids Array of post IDs to index |
| 896 |
* @return array Result with job info |
| 897 |
*/ |
| 898 |
public function queue_posts_for_indexing( $post_ids ) { |
| 899 |
$batches = array_chunk( $post_ids, self::BATCH_SIZE ); |
| 900 |
$job_id = 'wp_index_' . uniqid(); |
| 901 |
$total_posts = count( $post_ids ); |
| 902 |
|
| 903 |
// Clear the status cache so polling gets fresh data |
| 904 |
delete_transient( 'wpforo_ai_wp_indexing_status' ); |
| 905 |
|
| 906 |
// Store queue in options for processing |
| 907 |
update_option( 'wpforo_ai_wp_indexing_queue', [ |
| 908 |
'job_id' => $job_id, |
| 909 |
'batches' => $batches, |
| 910 |
'current' => 0, |
| 911 |
'total_posts' => $total_posts, |
| 912 |
'indexed' => 0, |
| 913 |
'failed' => 0, |
| 914 |
'skipped' => 0, |
| 915 |
'status' => 'processing', |
| 916 |
'started_at' => current_time( 'mysql', true ), |
| 917 |
] ); |
| 918 |
|
| 919 |
// Schedule first batch |
| 920 |
wp_schedule_single_event( time() + 1, 'wpforo_ai_process_wp_batch' ); |
| 921 |
|
| 922 |
return [ |
| 923 |
'job_id' => $job_id, |
| 924 |
'total_posts' => $total_posts, |
| 925 |
'batches' => count( $batches ), |
| 926 |
'status' => 'queued', |
| 927 |
]; |
| 928 |
} |
| 929 |
|
| 930 |
/** |
| 931 |
* Process a batch of posts for indexing |
| 932 |
* |
| 933 |
* @return array|WP_Error Result or error |
| 934 |
*/ |
| 935 |
public function process_batch() { |
| 936 |
$queue = get_option( 'wpforo_ai_wp_indexing_queue' ); |
| 937 |
|
| 938 |
if ( empty( $queue ) || empty( $queue['batches'] ) ) { |
| 939 |
return new \WP_Error( 'no_queue', 'No indexing queue found' ); |
| 940 |
} |
| 941 |
|
| 942 |
$current_batch_index = $queue['current']; |
| 943 |
|
| 944 |
if ( ! isset( $queue['batches'][ $current_batch_index ] ) ) { |
| 945 |
// All batches processed |
| 946 |
delete_option( 'wpforo_ai_wp_indexing_queue' ); |
| 947 |
return [ 'status' => 'completed' ]; |
| 948 |
} |
| 949 |
|
| 950 |
$post_ids = $queue['batches'][ $current_batch_index ]; |
| 951 |
$posts = []; |
| 952 |
$skipped = 0; |
| 953 |
|
| 954 |
foreach ( $post_ids as $post_id ) { |
| 955 |
$post = get_post( $post_id ); |
| 956 |
if ( $post && $post->post_status === 'publish' ) { |
| 957 |
// Skip posts with non-indexable content (shortcodes only, binary, too short) |
| 958 |
if ( ! $this->is_content_indexable( $post ) ) { |
| 959 |
$skipped++; |
| 960 |
continue; |
| 961 |
} |
| 962 |
$posts[] = $this->format_post_for_indexing( $post ); |
| 963 |
} |
| 964 |
} |
| 965 |
|
| 966 |
// Track skipped posts |
| 967 |
if ( ! isset( $queue['skipped'] ) ) { |
| 968 |
$queue['skipped'] = 0; |
| 969 |
} |
| 970 |
$queue['skipped'] += $skipped; |
| 971 |
|
| 972 |
if ( empty( $posts ) ) { |
| 973 |
// Move to next batch |
| 974 |
$queue['current']++; |
| 975 |
update_option( 'wpforo_ai_wp_indexing_queue', $queue ); |
| 976 |
|
| 977 |
// Schedule next batch |
| 978 |
if ( isset( $queue['batches'][ $queue['current'] ] ) ) { |
| 979 |
wp_schedule_single_event( time() + 2, 'wpforo_ai_process_wp_batch' ); |
| 980 |
} |
| 981 |
|
| 982 |
return [ 'status' => 'batch_empty', 'current' => $queue['current'] ]; |
| 983 |
} |
| 984 |
|
| 985 |
// Check storage mode — local mode stores embeddings in WordPress DB |
| 986 |
// WordPress content uses the global storage mode setting (board 0) |
| 987 |
if ( $this->is_local_storage_mode() ) { |
| 988 |
$local_result = $this->process_batch_local( $posts, $queue ); |
| 989 |
|
| 990 |
return $local_result; |
| 991 |
} |
| 992 |
|
| 993 |
// CLOUD MODE: Send to cloud API |
| 994 |
$response = WPF()->ai_client->api_post( '/rag/wordpress/ingest', [ |
| 995 |
'posts' => $posts, |
| 996 |
'chunk_size' => 512, |
| 997 |
'overlap_percent' => 20, |
| 998 |
], 120 ); |
| 999 |
|
| 1000 |
if ( is_wp_error( $response ) ) { |
| 1001 |
$queue['failed'] += count( $post_ids ); |
| 1002 |
} else { |
| 1003 |
$queue['indexed'] += count( $posts ); |
| 1004 |
} |
| 1005 |
|
| 1006 |
// Move to next batch |
| 1007 |
$queue['current']++; |
| 1008 |
update_option( 'wpforo_ai_wp_indexing_queue', $queue ); |
| 1009 |
|
| 1010 |
// Schedule next batch |
| 1011 |
if ( isset( $queue['batches'][ $queue['current'] ] ) ) { |
| 1012 |
wp_schedule_single_event( time() + 2, 'wpforo_ai_process_wp_batch' ); |
| 1013 |
} else { |
| 1014 |
// All done |
| 1015 |
$queue['status'] = 'completed'; |
| 1016 |
$queue['completed_at'] = current_time( 'mysql', true ); |
| 1017 |
update_option( 'wpforo_ai_wp_indexing_queue', $queue ); |
| 1018 |
|
| 1019 |
// Clear cache |
| 1020 |
delete_transient( 'wpforo_ai_wp_indexing_status' ); |
| 1021 |
} |
| 1022 |
|
| 1023 |
return [ |
| 1024 |
'status' => 'processing', |
| 1025 |
'current' => $queue['current'], |
| 1026 |
'indexed' => $queue['indexed'], |
| 1027 |
'failed' => $queue['failed'], |
| 1028 |
'skipped' => $queue['skipped'], |
| 1029 |
]; |
| 1030 |
} |
| 1031 |
|
| 1032 |
/** |
| 1033 |
* Process a batch of posts for local storage mode |
| 1034 |
* |
| 1035 |
* Generates embeddings via cloud API but stores them in WordPress DB |
| 1036 |
* instead of cloud vector storage. Uses content_hash dedup to skip |
| 1037 |
* unchanged posts. |
| 1038 |
* |
| 1039 |
* @param array $posts Formatted post data from format_post_for_indexing() |
| 1040 |
* @param array $queue Current queue state (modified by reference via option update) |
| 1041 |
* @return array Result with status, indexed, failed, skipped counts |
| 1042 |
*/ |
| 1043 |
private function process_batch_local( $posts, $queue ) { |
| 1044 |
$local = $this->get_local_storage(); |
| 1045 |
if ( ! $local ) { |
| 1046 |
return [ |
| 1047 |
'status' => 'error', |
| 1048 |
'message' => 'Local storage is not available', |
| 1049 |
'indexed' => 0, |
| 1050 |
'failed' => count( $posts ), |
| 1051 |
'skipped' => 0, |
| 1052 |
]; |
| 1053 |
} |
| 1054 |
|
| 1055 |
$indexed = 0; |
| 1056 |
$failed = 0; |
| 1057 |
$skipped = 0; |
| 1058 |
|
| 1059 |
foreach ( $posts as $post_data ) { |
| 1060 |
$post_id = $post_data['post_id']; |
| 1061 |
$post_type = $post_data['post_type']; |
| 1062 |
|
| 1063 |
// Build text content for embedding |
| 1064 |
$content_parts = []; |
| 1065 |
if ( ! empty( $post_data['title'] ) ) { |
| 1066 |
$content_parts[] = $post_data['title']; |
| 1067 |
} |
| 1068 |
if ( ! empty( $post_data['excerpt'] ) ) { |
| 1069 |
$content_parts[] = wp_strip_all_tags( $post_data['excerpt'] ); |
| 1070 |
} |
| 1071 |
if ( ! empty( $post_data['content'] ) ) { |
| 1072 |
$clean_content = strip_shortcodes( $post_data['content'] ); |
| 1073 |
$clean_content = wp_strip_all_tags( $clean_content ); |
| 1074 |
$clean_content = html_entity_decode( $clean_content, ENT_QUOTES, 'UTF-8' ); |
| 1075 |
$clean_content = preg_replace( '/\s+/', ' ', trim( $clean_content ) ); |
| 1076 |
$content_parts[] = $clean_content; |
| 1077 |
} |
| 1078 |
|
| 1079 |
// Add taxonomy context |
| 1080 |
if ( ! empty( $post_data['taxonomy_terms'] ) ) { |
| 1081 |
foreach ( $post_data['taxonomy_terms'] as $tax_label => $terms ) { |
| 1082 |
$content_parts[] = $tax_label . ': ' . implode( ', ', $terms ); |
| 1083 |
} |
| 1084 |
} |
| 1085 |
|
| 1086 |
$content = implode( "\n\n", $content_parts ); |
| 1087 |
|
| 1088 |
// Truncate to fit within embedding model's token limit |
| 1089 |
// Titan Embed v2: 8192 tokens ≈ 28000 chars (safe margin) |
| 1090 |
$max_chars = 28000; |
| 1091 |
if ( mb_strlen( $content, 'UTF-8' ) > $max_chars ) { |
| 1092 |
$content = mb_substr( $content, 0, $max_chars, 'UTF-8' ); |
| 1093 |
} |
| 1094 |
|
| 1095 |
$content_hash = md5( $content ); |
| 1096 |
|
| 1097 |
// Check if already indexed with same content (dedup) |
| 1098 |
$existing = $local->get_embedding( $post_id ); |
| 1099 |
if ( $existing && $existing['content_hash'] === $content_hash ) { |
| 1100 |
$skipped++; |
| 1101 |
continue; |
| 1102 |
} |
| 1103 |
|
| 1104 |
// Generate embedding via cloud API |
| 1105 |
$result = WPF()->ai_client->generate_embedding( $content ); |
| 1106 |
$embedding = is_wp_error( $result ) ? $result : ( $result['embedding'] ?? null ); |
| 1107 |
|
| 1108 |
if ( is_wp_error( $embedding ) || ! is_array( $embedding ) ) { |
| 1109 |
$error_msg = is_wp_error( $embedding ) ? $embedding->get_error_message() : 'Invalid embedding response'; |
| 1110 |
\wpforo_ai_log( 'error', sprintf( |
| 1111 |
'Failed to generate embedding for WP post %d: %s', |
| 1112 |
$post_id, |
| 1113 |
$error_msg |
| 1114 |
), 'WPIndexer' ); |
| 1115 |
$failed++; |
| 1116 |
continue; |
| 1117 |
} |
| 1118 |
|
| 1119 |
// Build content preview |
| 1120 |
$preview = wp_trim_words( wp_strip_all_tags( strip_shortcodes( $post_data['content'] ?? '' ) ), 80, '...' ); |
| 1121 |
|
| 1122 |
// Store locally with content_type = post_type |
| 1123 |
$stored = $local->store_embedding( |
| 1124 |
0, // topicid (not a forum topic) |
| 1125 |
$post_id, // postid = WP post ID |
| 1126 |
0, // forumid (not a forum) |
| 1127 |
(int) $post_data['author_id'], |
| 1128 |
$embedding, |
| 1129 |
$content_hash, |
| 1130 |
$preview, |
| 1131 |
'amazon.titan-embed-text-v2', |
| 1132 |
$post_type // content_type = post type (page, post, product, etc.) |
| 1133 |
); |
| 1134 |
|
| 1135 |
if ( $stored ) { |
| 1136 |
$indexed++; |
| 1137 |
} else { |
| 1138 |
$failed++; |
| 1139 |
} |
| 1140 |
} |
| 1141 |
|
| 1142 |
// Update queue progress |
| 1143 |
$queue['indexed'] += $indexed; |
| 1144 |
$queue['failed'] += $failed; |
| 1145 |
$queue['skipped'] = ( $queue['skipped'] ?? 0 ) + $skipped; |
| 1146 |
|
| 1147 |
// Move to next batch |
| 1148 |
$queue['current']++; |
| 1149 |
update_option( 'wpforo_ai_wp_indexing_queue', $queue ); |
| 1150 |
|
| 1151 |
// Schedule next batch |
| 1152 |
if ( isset( $queue['batches'][ $queue['current'] ] ) ) { |
| 1153 |
wp_schedule_single_event( time() + 2, 'wpforo_ai_process_wp_batch' ); |
| 1154 |
} else { |
| 1155 |
// All done |
| 1156 |
$queue['status'] = 'completed'; |
| 1157 |
$queue['completed_at'] = current_time( 'mysql', true ); |
| 1158 |
update_option( 'wpforo_ai_wp_indexing_queue', $queue ); |
| 1159 |
delete_transient( 'wpforo_ai_wp_indexing_status' ); |
| 1160 |
} |
| 1161 |
|
| 1162 |
return [ |
| 1163 |
'status' => 'processing', |
| 1164 |
'current' => $queue['current'], |
| 1165 |
'indexed' => $queue['indexed'], |
| 1166 |
'failed' => $queue['failed'], |
| 1167 |
'skipped' => $queue['skipped'] ?? 0, |
| 1168 |
]; |
| 1169 |
} |
| 1170 |
|
| 1171 |
/** |
| 1172 |
* Get WordPress content indexing status |
| 1173 |
* |
| 1174 |
* @return array|WP_Error Status data or error |
| 1175 |
*/ |
| 1176 |
public function get_indexing_status( $skip_cache = false ) { |
| 1177 |
if ( ! WPF()->ai_client || ! WPF()->ai_client->is_service_available() ) { |
| 1178 |
return new \WP_Error( 'not_connected', __( 'AI service is not connected', 'wpforo' ) ); |
| 1179 |
} |
| 1180 |
|
| 1181 |
// Check cache (skip if explicitly requested or if indexing is in progress) |
| 1182 |
$queue = get_option( 'wpforo_ai_wp_indexing_queue' ); |
| 1183 |
$is_processing = ! empty( $queue ) && isset( $queue['status'] ) && $queue['status'] === 'processing'; |
| 1184 |
|
| 1185 |
if ( ! $skip_cache && ! $is_processing ) { |
| 1186 |
$cached = get_transient( 'wpforo_ai_wp_indexing_status' ); |
| 1187 |
if ( $cached !== false ) { |
| 1188 |
return $cached; |
| 1189 |
} |
| 1190 |
} |
| 1191 |
|
| 1192 |
// Get indexed counts — source depends on storage mode |
| 1193 |
if ( $this->is_local_storage_mode() ) { |
| 1194 |
// LOCAL mode: count from WordPress ai_embeddings table |
| 1195 |
$local = $this->get_local_storage(); |
| 1196 |
if ( ! $local ) { |
| 1197 |
return new \WP_Error( 'storage_unavailable', __( 'Local storage is not available', 'wpforo' ) ); |
| 1198 |
} |
| 1199 |
$indexed_counts = $local->get_wp_indexed_counts(); |
| 1200 |
$response = [ |
| 1201 |
'content_source' => 'wordpress', |
| 1202 |
'indexed_counts' => $indexed_counts, |
| 1203 |
'total_indexed' => array_sum( $indexed_counts ), |
| 1204 |
]; |
| 1205 |
} else { |
| 1206 |
// CLOUD mode: query backend API (has sync_state records) |
| 1207 |
$response = WPF()->ai_client->api_get( '/rag/wordpress/status' ); |
| 1208 |
if ( is_wp_error( $response ) ) { |
| 1209 |
return $response; |
| 1210 |
} |
| 1211 |
} |
| 1212 |
|
| 1213 |
// Build by_type structure that JavaScript expects |
| 1214 |
$post_types = $this->get_public_post_types(); |
| 1215 |
$indexed_counts = isset( $response['indexed_counts'] ) ? $response['indexed_counts'] : []; |
| 1216 |
$by_type = []; |
| 1217 |
|
| 1218 |
foreach ( $post_types as $type ) { |
| 1219 |
$type_key = 'wp_' . $type['name']; |
| 1220 |
$indexed = isset( $indexed_counts[ $type_key ] ) ? (int) $indexed_counts[ $type_key ] : 0; |
| 1221 |
$total = (int) $type['count']; |
| 1222 |
$by_type[ $type_key ] = [ |
| 1223 |
'indexed' => $indexed, |
| 1224 |
'total' => $total, |
| 1225 |
'percentage' => $total > 0 ? round( ( $indexed / $total ) * 100, 1 ) : 0, |
| 1226 |
]; |
| 1227 |
} |
| 1228 |
|
| 1229 |
$response['by_type'] = $by_type; |
| 1230 |
$response['total_indexed'] = isset( $response['total_indexed'] ) ? (int) $response['total_indexed'] : 0; |
| 1231 |
|
| 1232 |
// Check if there's an active queue |
| 1233 |
$queue = get_option( 'wpforo_ai_wp_indexing_queue' ); |
| 1234 |
$is_indexing = false; |
| 1235 |
if ( ! empty( $queue ) ) { |
| 1236 |
$queue_status = isset( $queue['status'] ) ? $queue['status'] : 'processing'; |
| 1237 |
$is_indexing = ( $queue_status === 'processing' ); |
| 1238 |
$response['queue'] = [ |
| 1239 |
'job_id' => $queue['job_id'], |
| 1240 |
'total_posts' => $queue['total_posts'], |
| 1241 |
'indexed' => $queue['indexed'], |
| 1242 |
'failed' => $queue['failed'], |
| 1243 |
'current' => $queue['current'], |
| 1244 |
'total' => count( $queue['batches'] ), |
| 1245 |
'status' => $queue_status, |
| 1246 |
]; |
| 1247 |
} |
| 1248 |
|
| 1249 |
// Add is_indexing flag for UI |
| 1250 |
$response['is_indexing'] = $is_indexing; |
| 1251 |
|
| 1252 |
// Get last activity timestamp from queue or sync_state |
| 1253 |
$last_indexed_at = null; |
| 1254 |
if ( ! empty( $queue['completed_at'] ) ) { |
| 1255 |
$last_indexed_at = $queue['completed_at']; |
| 1256 |
} elseif ( ! empty( $queue['started_at'] ) ) { |
| 1257 |
$last_indexed_at = $queue['started_at']; |
| 1258 |
} |
| 1259 |
$response['last_indexed_at'] = $last_indexed_at; |
| 1260 |
// Format date using WordPress function (always available in AJAX context) |
| 1261 |
$response['last_indexed_at_formatted'] = $last_indexed_at |
| 1262 |
? wp_date( get_option( 'date_format' ) . ' ' . get_option( 'time_format' ), strtotime( $last_indexed_at ) ) |
| 1263 |
: null; |
| 1264 |
|
| 1265 |
// Cache for 5 minutes |
| 1266 |
set_transient( 'wpforo_ai_wp_indexing_status', $response, 5 * MINUTE_IN_SECONDS ); |
| 1267 |
|
| 1268 |
return $response; |
| 1269 |
} |
| 1270 |
|
| 1271 |
/** |
| 1272 |
* Delete WordPress content from index |
| 1273 |
* |
| 1274 |
* @param array $params Delete parameters (post_types, post_ids, all) |
| 1275 |
* @return array|WP_Error Result or error |
| 1276 |
*/ |
| 1277 |
public function delete_content( $params ) { |
| 1278 |
if ( $this->is_local_storage_mode() ) { |
| 1279 |
// LOCAL mode: delete from WordPress ai_embeddings table |
| 1280 |
$local = $this->get_local_storage(); |
| 1281 |
if ( ! $local ) { |
| 1282 |
return new \WP_Error( 'storage_unavailable', __( 'Local storage is not available', 'wpforo' ) ); |
| 1283 |
} |
| 1284 |
|
| 1285 |
$post_types = isset( $params['post_types'] ) ? $params['post_types'] : null; |
| 1286 |
$post_ids = isset( $params['post_ids'] ) ? $params['post_ids'] : null; |
| 1287 |
|
| 1288 |
// 'all' flag means delete all non-forum CPT embeddings |
| 1289 |
if ( ! empty( $params['all'] ) ) { |
| 1290 |
$post_types = null; |
| 1291 |
$post_ids = null; |
| 1292 |
} |
| 1293 |
|
| 1294 |
$deleted = $local->delete_wp_embeddings( $post_types, $post_ids ); |
| 1295 |
|
| 1296 |
// Clear cache |
| 1297 |
delete_transient( 'wpforo_ai_wp_indexing_status' ); |
| 1298 |
|
| 1299 |
return [ |
| 1300 |
'deleted' => $deleted, |
| 1301 |
'message' => sprintf( 'Deleted %d embeddings from local storage.', $deleted ), |
| 1302 |
]; |
| 1303 |
} |
| 1304 |
|
| 1305 |
// CLOUD mode: delete via backend API |
| 1306 |
if ( ! WPF()->ai_client || ! WPF()->ai_client->is_service_available() ) { |
| 1307 |
return new \WP_Error( 'not_connected', __( 'AI service is not connected', 'wpforo' ) ); |
| 1308 |
} |
| 1309 |
|
| 1310 |
$response = WPF()->ai_client->api_post( '/rag/wordpress/delete', $params, 60 ); |
| 1311 |
|
| 1312 |
if ( ! is_wp_error( $response ) ) { |
| 1313 |
// Clear cache |
| 1314 |
delete_transient( 'wpforo_ai_wp_indexing_status' ); |
| 1315 |
} |
| 1316 |
|
| 1317 |
return $response; |
| 1318 |
} |
| 1319 |
|
| 1320 |
// =============================== |
| 1321 |
// AJAX Handlers |
| 1322 |
// =============================== |
| 1323 |
|
| 1324 |
/** |
| 1325 |
* AJAX: Get public post types |
| 1326 |
*/ |
| 1327 |
public function ajax_get_post_types() { |
| 1328 |
check_ajax_referer( 'wpforo_admin_ajax', 'security' ); |
| 1329 |
|
| 1330 |
if ( ! current_user_can( 'manage_options' ) ) { |
| 1331 |
wp_send_json_error( [ 'message' => __( 'Permission denied', 'wpforo' ) ] ); |
| 1332 |
} |
| 1333 |
|
| 1334 |
$post_types = $this->get_public_post_types(); |
| 1335 |
wp_send_json_success( [ 'post_types' => $post_types ] ); |
| 1336 |
} |
| 1337 |
|
| 1338 |
/** |
| 1339 |
* AJAX: Get taxonomies for post type |
| 1340 |
*/ |
| 1341 |
public function ajax_get_taxonomies() { |
| 1342 |
check_ajax_referer( 'wpforo_admin_ajax', 'security' ); |
| 1343 |
|
| 1344 |
if ( ! current_user_can( 'manage_options' ) ) { |
| 1345 |
wp_send_json_error( [ 'message' => __( 'Permission denied', 'wpforo' ) ] ); |
| 1346 |
} |
| 1347 |
|
| 1348 |
$post_type = isset( $_POST['post_type'] ) ? sanitize_key( $_POST['post_type'] ) : 'post'; |
| 1349 |
$taxonomies = $this->get_taxonomies_for_post_type( $post_type ); |
| 1350 |
|
| 1351 |
wp_send_json_success( [ 'taxonomies' => $taxonomies ] ); |
| 1352 |
} |
| 1353 |
|
| 1354 |
/** |
| 1355 |
* AJAX: Get terms for taxonomy |
| 1356 |
*/ |
| 1357 |
public function ajax_get_taxonomy_terms() { |
| 1358 |
check_ajax_referer( 'wpforo_admin_ajax', 'security' ); |
| 1359 |
|
| 1360 |
if ( ! current_user_can( 'manage_options' ) ) { |
| 1361 |
wp_send_json_error( [ 'message' => __( 'Permission denied', 'wpforo' ) ] ); |
| 1362 |
} |
| 1363 |
|
| 1364 |
$taxonomy = isset( $_POST['taxonomy'] ) ? sanitize_key( $_POST['taxonomy'] ) : 'category'; |
| 1365 |
|
| 1366 |
// Get post types if provided (to count only published posts for specific types) |
| 1367 |
$post_types = []; |
| 1368 |
if ( ! empty( $_POST['post_types'] ) ) { |
| 1369 |
$post_types = array_map( 'sanitize_key', (array) $_POST['post_types'] ); |
| 1370 |
} |
| 1371 |
|
| 1372 |
$terms = $this->get_taxonomy_terms( $taxonomy, true, $post_types ); |
| 1373 |
|
| 1374 |
wp_send_json_success( [ 'terms' => $terms ] ); |
| 1375 |
} |
| 1376 |
|
| 1377 |
/** |
| 1378 |
* AJAX: Get indexing status |
| 1379 |
*/ |
| 1380 |
public function ajax_get_indexing_status() { |
| 1381 |
check_ajax_referer( 'wpforo_admin_ajax', 'security' ); |
| 1382 |
|
| 1383 |
if ( ! current_user_can( 'manage_options' ) ) { |
| 1384 |
wp_send_json_error( [ 'message' => __( 'Permission denied', 'wpforo' ) ] ); |
| 1385 |
} |
| 1386 |
|
| 1387 |
$status = $this->get_indexing_status(); |
| 1388 |
|
| 1389 |
if ( is_wp_error( $status ) ) { |
| 1390 |
wp_send_json_error( [ 'message' => $status->get_error_message() ] ); |
| 1391 |
} |
| 1392 |
|
| 1393 |
wp_send_json_success( $status ); |
| 1394 |
} |
| 1395 |
|
| 1396 |
/** |
| 1397 |
* AJAX: Index by taxonomy |
| 1398 |
*/ |
| 1399 |
public function ajax_index_by_taxonomy() { |
| 1400 |
check_ajax_referer( 'wpforo_admin_ajax', 'security' ); |
| 1401 |
|
| 1402 |
if ( ! current_user_can( 'manage_options' ) ) { |
| 1403 |
wp_send_json_error( [ 'message' => __( 'Permission denied', 'wpforo' ) ] ); |
| 1404 |
} |
| 1405 |
|
| 1406 |
$taxonomy = isset( $_POST['taxonomy'] ) ? sanitize_key( $_POST['taxonomy'] ) : ''; |
| 1407 |
$post_types = isset( $_POST['post_types'] ) ? array_map( 'sanitize_key', (array) $_POST['post_types'] ) : [ 'post' ]; |
| 1408 |
$date_from = isset( $_POST['date_from'] ) ? sanitize_text_field( $_POST['date_from'] ) : ''; |
| 1409 |
$date_to = isset( $_POST['date_to'] ) ? sanitize_text_field( $_POST['date_to'] ) : ''; |
| 1410 |
|
| 1411 |
// Support both single term_id (legacy) and multiple term_ids |
| 1412 |
$term_ids = []; |
| 1413 |
if ( isset( $_POST['term_ids'] ) && is_array( $_POST['term_ids'] ) ) { |
| 1414 |
$term_ids = array_map( 'intval', $_POST['term_ids'] ); |
| 1415 |
} elseif ( isset( $_POST['term_id'] ) ) { |
| 1416 |
$term_ids = [ intval( $_POST['term_id'] ) ]; |
| 1417 |
} |
| 1418 |
|
| 1419 |
if ( empty( $taxonomy ) || empty( $term_ids ) ) { |
| 1420 |
wp_send_json_error( [ 'message' => __( 'Invalid taxonomy or term', 'wpforo' ) ] ); |
| 1421 |
} |
| 1422 |
|
| 1423 |
$result = $this->index_by_taxonomy( $taxonomy, $term_ids, $post_types, $date_from, $date_to ); |
| 1424 |
|
| 1425 |
if ( is_wp_error( $result ) ) { |
| 1426 |
wp_send_json_error( [ 'message' => $result->get_error_message() ] ); |
| 1427 |
} |
| 1428 |
|
| 1429 |
wp_send_json_success( $result ); |
| 1430 |
} |
| 1431 |
|
| 1432 |
/** |
| 1433 |
* AJAX: Custom indexing |
| 1434 |
*/ |
| 1435 |
public function ajax_index_custom() { |
| 1436 |
check_ajax_referer( 'wpforo_admin_ajax', 'security' ); |
| 1437 |
|
| 1438 |
if ( ! current_user_can( 'manage_options' ) ) { |
| 1439 |
wp_send_json_error( [ 'message' => __( 'Permission denied', 'wpforo' ) ] ); |
| 1440 |
} |
| 1441 |
|
| 1442 |
$params = []; |
| 1443 |
|
| 1444 |
if ( ! empty( $_POST['post_types'] ) ) { |
| 1445 |
$params['post_types'] = array_map( 'sanitize_key', (array) $_POST['post_types'] ); |
| 1446 |
} |
| 1447 |
|
| 1448 |
if ( ! empty( $_POST['date_from'] ) ) { |
| 1449 |
$params['date_from'] = sanitize_text_field( $_POST['date_from'] ); |
| 1450 |
} |
| 1451 |
|
| 1452 |
if ( ! empty( $_POST['date_to'] ) ) { |
| 1453 |
$params['date_to'] = sanitize_text_field( $_POST['date_to'] ); |
| 1454 |
} |
| 1455 |
|
| 1456 |
if ( ! empty( $_POST['post_ids'] ) ) { |
| 1457 |
$params['post_ids'] = array_map( 'intval', (array) $_POST['post_ids'] ); |
| 1458 |
} |
| 1459 |
|
| 1460 |
if ( ! empty( $_POST['author'] ) ) { |
| 1461 |
$params['author'] = intval( $_POST['author'] ); |
| 1462 |
} |
| 1463 |
|
| 1464 |
$result = $this->index_custom( $params ); |
| 1465 |
|
| 1466 |
if ( is_wp_error( $result ) ) { |
| 1467 |
wp_send_json_error( [ 'message' => $result->get_error_message() ] ); |
| 1468 |
} |
| 1469 |
|
| 1470 |
wp_send_json_success( $result ); |
| 1471 |
} |
| 1472 |
|
| 1473 |
/** |
| 1474 |
* AJAX: Delete content |
| 1475 |
*/ |
| 1476 |
public function ajax_delete_content() { |
| 1477 |
check_ajax_referer( 'wpforo_admin_ajax', 'security' ); |
| 1478 |
|
| 1479 |
if ( ! current_user_can( 'manage_options' ) ) { |
| 1480 |
wp_send_json_error( [ 'message' => __( 'Permission denied', 'wpforo' ) ] ); |
| 1481 |
} |
| 1482 |
|
| 1483 |
$params = []; |
| 1484 |
|
| 1485 |
if ( isset( $_POST['delete_all'] ) && $_POST['delete_all'] === 'true' ) { |
| 1486 |
$params['all'] = true; // API expects 'all', not 'delete_all' |
| 1487 |
} elseif ( ! empty( $_POST['post_types'] ) ) { |
| 1488 |
$params['post_types'] = array_map( 'sanitize_key', (array) $_POST['post_types'] ); |
| 1489 |
} elseif ( ! empty( $_POST['post_ids'] ) ) { |
| 1490 |
$params['post_ids'] = array_map( 'intval', (array) $_POST['post_ids'] ); |
| 1491 |
} |
| 1492 |
|
| 1493 |
$result = $this->delete_content( $params ); |
| 1494 |
|
| 1495 |
if ( is_wp_error( $result ) ) { |
| 1496 |
wp_send_json_error( [ 'message' => $result->get_error_message() ] ); |
| 1497 |
} |
| 1498 |
|
| 1499 |
wp_send_json_success( $result ); |
| 1500 |
} |
| 1501 |
} |
| 1502 |
|
