advisor.php
2 years ago
chatbot.php
1 year ago
discussions.php
1 year ago
files.php
1 year ago
gdpr.php
1 year ago
search.php
1 year ago
security.php
1 year ago
tasks.php
1 year ago
wand.php
1 year ago
search.php
923 lines
| 1 | <?php |
| 2 | |
| 3 | class Meow_MWAI_Modules_Search { |
| 4 | private $core = null; |
| 5 | private $namespace = 'mwai-ui/v1'; |
| 6 | |
| 7 | public function __construct( $core ) { |
| 8 | $this->core = $core; |
| 9 | add_action( 'rest_api_init', [ $this, 'rest_api_init' ] ); |
| 10 | add_action( 'pre_get_posts', [ $this, 'pre_get_posts' ] ); |
| 11 | add_filter( 'the_posts', [ $this, 'filter_embeddings_search_results' ], 10, 2 ); |
| 12 | add_filter( 'get_search_query', [ $this, 'customize_search_display' ] ); |
| 13 | |
| 14 | // Initialize search frontend settings if they don't exist |
| 15 | $this->init_frontend_settings(); |
| 16 | } |
| 17 | |
| 18 | private function init_frontend_settings() { |
| 19 | // Ensure frontend search settings exist with defaults |
| 20 | if ( $this->core->get_option( 'search_frontend_method' ) === null ) { |
| 21 | $this->core->update_option( 'search_frontend_method', 'wordpress' ); |
| 22 | } |
| 23 | if ( $this->core->get_option( 'search_frontend_env_id' ) === null ) { |
| 24 | $this->core->update_option( 'search_frontend_env_id', null ); |
| 25 | } |
| 26 | if ( $this->core->get_option( 'search_website_context' ) === null ) { |
| 27 | $this->core->update_option( 'search_website_context', 'This is a website with useful information and content.' ); |
| 28 | } |
| 29 | } |
| 30 | |
| 31 | public function rest_api_init() { |
| 32 | register_rest_route( $this->namespace, '/search', [ |
| 33 | 'methods' => 'POST', |
| 34 | 'callback' => [ $this, 'rest_search' ], |
| 35 | 'permission_callback' => '__return_true' |
| 36 | ] ); |
| 37 | } |
| 38 | |
| 39 | public function pre_get_posts( $query ) { |
| 40 | if ( is_admin() || !$query->is_main_query() || !$query->is_search() ) { |
| 41 | return; |
| 42 | } |
| 43 | $search = $query->get( 's' ); |
| 44 | if ( empty( $search ) ) { |
| 45 | return; |
| 46 | } |
| 47 | |
| 48 | // Get the frontend search method setting |
| 49 | $frontend_method = $this->core->get_option( 'search_frontend_method', 'wordpress' ); |
| 50 | |
| 51 | // If WordPress method is selected, do nothing (use default WordPress search) |
| 52 | if ( $frontend_method === 'wordpress' ) { |
| 53 | return; |
| 54 | } |
| 55 | |
| 56 | // For Keywords method, use the full progressive search like admin |
| 57 | if ( $frontend_method === 'keywords' ) { |
| 58 | // Store the original search for later use |
| 59 | $query->set( 'mwai_original_search', $search ); |
| 60 | // Set a unique search term that won't match anything - we'll handle this in the_posts filter |
| 61 | $query->set( 's', 'MWAI_KEYWORDS_SEARCH_' . md5( $search ) ); |
| 62 | return; |
| 63 | } |
| 64 | |
| 65 | // For Embeddings method, we need to handle this differently |
| 66 | // Since we can't easily replace WordPress search with embeddings in pre_get_posts, |
| 67 | // we'll modify the query to return no results and handle embeddings in template_redirect |
| 68 | if ( $frontend_method === 'embeddings' ) { |
| 69 | // Store the original search for later use |
| 70 | $query->set( 'mwai_original_search', $search ); |
| 71 | // Set a unique search term that won't match anything |
| 72 | $query->set( 's', 'MWAI_EMBEDDINGS_SEARCH_' . md5( $search ) ); |
| 73 | } |
| 74 | } |
| 75 | |
| 76 | public function filter_embeddings_search_results( $posts, $query ) { |
| 77 | // Only handle main search queries with our special markers |
| 78 | if ( !$query->is_main_query() || !$query->is_search() || is_admin() ) { |
| 79 | return $posts; |
| 80 | } |
| 81 | |
| 82 | $search_term = $query->get( 's' ); |
| 83 | $original_search = $query->get( 'mwai_original_search' ); |
| 84 | |
| 85 | // Check if this is our embeddings or keywords search |
| 86 | if ( empty( $original_search ) || |
| 87 | ( strpos( $search_term, 'MWAI_EMBEDDINGS_SEARCH_' ) !== 0 && |
| 88 | strpos( $search_term, 'MWAI_KEYWORDS_SEARCH_' ) !== 0 ) ) { |
| 89 | return $posts; |
| 90 | } |
| 91 | |
| 92 | // Handle Keywords search |
| 93 | if ( strpos( $search_term, 'MWAI_KEYWORDS_SEARCH_' ) === 0 ) { |
| 94 | return $this->handle_frontend_keywords_search( $original_search, $query ); |
| 95 | } |
| 96 | |
| 97 | // Handle Embeddings search (existing logic below) |
| 98 | if ( strpos( $search_term, 'MWAI_EMBEDDINGS_SEARCH_' ) !== 0 ) { |
| 99 | return $posts; |
| 100 | } |
| 101 | |
| 102 | // Get the frontend search method to double-check |
| 103 | $frontend_method = $this->core->get_option( 'search_frontend_method', 'wordpress' ); |
| 104 | if ( $frontend_method !== 'embeddings' ) { |
| 105 | return $posts; |
| 106 | } |
| 107 | |
| 108 | // Get the embeddings environment ID |
| 109 | $env_id = $this->core->get_option( 'search_frontend_env_id', null ); |
| 110 | if ( empty( $env_id ) ) { |
| 111 | return $posts; // No environment selected, return empty |
| 112 | } |
| 113 | |
| 114 | try { |
| 115 | // Perform embeddings search |
| 116 | $embedding_result = $this->search_with_embeddings( $original_search, $env_id ); |
| 117 | |
| 118 | if ( isset( $embedding_result['error'] ) || empty( $embedding_result['post_ids'] ) ) { |
| 119 | return []; // Return empty array if search failed or no results |
| 120 | } |
| 121 | |
| 122 | // Get the post IDs from embeddings results |
| 123 | $post_ids = array_map( function( $result ) { |
| 124 | return $result['id']; |
| 125 | }, $embedding_result['post_ids'] ); |
| 126 | |
| 127 | if ( empty( $post_ids ) ) { |
| 128 | return []; |
| 129 | } |
| 130 | |
| 131 | // Get the actual post objects |
| 132 | $embeddings_posts = get_posts( [ |
| 133 | 'post__in' => $post_ids, |
| 134 | 'orderby' => 'post__in', |
| 135 | 'posts_per_page' => count( $post_ids ), |
| 136 | 'post_type' => 'post', |
| 137 | 'post_status' => 'publish' |
| 138 | ] ); |
| 139 | |
| 140 | // Update the query's found_posts count |
| 141 | $query->found_posts = count( $embeddings_posts ); |
| 142 | $query->max_num_pages = 1; |
| 143 | |
| 144 | return $embeddings_posts; |
| 145 | |
| 146 | } catch ( Exception $e ) { |
| 147 | error_log( 'AI Engine Search: Frontend embeddings search failed - ' . $e->getMessage() ); |
| 148 | return []; |
| 149 | } |
| 150 | } |
| 151 | |
| 152 | private function handle_frontend_keywords_search( $original_search, $query ) { |
| 153 | // Get website context for keywords search |
| 154 | $website_context = $this->core->get_option( 'search_website_context', '' ); |
| 155 | |
| 156 | try { |
| 157 | // Use the same search logic as the admin REST API |
| 158 | $search_queries = $this->generate_keyword_tiers( $original_search, $website_context ); |
| 159 | $keyword_result = $this->search_with_keywords( $search_queries ); |
| 160 | |
| 161 | if ( !empty( $keyword_result['results'] ) ) { |
| 162 | // Extract post IDs from results |
| 163 | $post_ids = array_map( function( $result ) { |
| 164 | return $result['id']; |
| 165 | }, $keyword_result['results'] ); |
| 166 | |
| 167 | // Get the actual post objects |
| 168 | $posts = get_posts( [ |
| 169 | 'post__in' => $post_ids, |
| 170 | 'orderby' => 'post__in', |
| 171 | 'posts_per_page' => count( $post_ids ), |
| 172 | 'post_type' => 'post', |
| 173 | 'post_status' => 'publish' |
| 174 | ] ); |
| 175 | |
| 176 | // Update the query's found_posts count |
| 177 | $query->found_posts = count( $posts ); |
| 178 | $query->max_num_pages = 1; |
| 179 | |
| 180 | return $posts; |
| 181 | } |
| 182 | |
| 183 | // If no results with keywords, fallback to original search |
| 184 | $fallback_posts = get_posts( [ |
| 185 | 's' => $original_search, |
| 186 | 'posts_per_page' => 20, |
| 187 | 'post_type' => 'post', |
| 188 | 'post_status' => 'publish' |
| 189 | ] ); |
| 190 | |
| 191 | $query->found_posts = count( $fallback_posts ); |
| 192 | $query->max_num_pages = 1; |
| 193 | |
| 194 | return $fallback_posts; |
| 195 | |
| 196 | } catch ( Exception $e ) { |
| 197 | error_log( 'AI Engine Search: Frontend keywords search failed - ' . $e->getMessage() ); |
| 198 | |
| 199 | // Fallback to original search |
| 200 | $fallback_posts = get_posts( [ |
| 201 | 's' => $original_search, |
| 202 | 'posts_per_page' => 20, |
| 203 | 'post_type' => 'post', |
| 204 | 'post_status' => 'publish' |
| 205 | ] ); |
| 206 | |
| 207 | return $fallback_posts; |
| 208 | } |
| 209 | } |
| 210 | |
| 211 | private function get_site_context() { |
| 212 | // Get all categories |
| 213 | $categories = get_categories( [ 'hide_empty' => false ] ); |
| 214 | $category_names = array_map( function( $cat ) { |
| 215 | return $cat->name; |
| 216 | }, $categories ); |
| 217 | |
| 218 | // Get all tags |
| 219 | $tags = get_tags( [ 'hide_empty' => false ] ); |
| 220 | $tag_names = array_map( function( $tag ) { |
| 221 | return $tag->name; |
| 222 | }, $tags ); |
| 223 | |
| 224 | return [ |
| 225 | 'categories' => $category_names, |
| 226 | 'tags' => $tag_names |
| 227 | ]; |
| 228 | } |
| 229 | |
| 230 | private function generate_keyword_tiers( $text, $website_context = '' ) { |
| 231 | $context = $this->get_site_context(); |
| 232 | |
| 233 | $message = "Generate 40 progressive search queries for: \"$text\"\n\n"; |
| 234 | |
| 235 | if ( !empty( $website_context ) ) { |
| 236 | $message .= "Website Context: " . $website_context . "\n\n"; |
| 237 | } |
| 238 | |
| 239 | if ( !empty( $context['categories'] ) ) { |
| 240 | $message .= "Site Categories: " . implode( ', ', array_slice( $context['categories'], 0, 20 ) ) . "\n"; |
| 241 | } |
| 242 | if ( !empty( $context['tags'] ) ) { |
| 243 | $message .= "Site Tags: " . implode( ', ', array_slice( $context['tags'], 0, 20 ) ) . "\n"; |
| 244 | } |
| 245 | |
| 246 | $message .= "\nCreate 40 search queries optimized for WordPress search:\n"; |
| 247 | $message .= "- WordPress searches for EXACT WORDS in post content, not concepts\n"; |
| 248 | $message .= "- Use SIMPLE, COMMON words that authors actually write in posts\n"; |
| 249 | $message .= "- Avoid complex phrases, technical terms, or descriptive adjectives\n"; |
| 250 | $message .= "- Focus on core nouns, verbs, and basic adjectives\n"; |
| 251 | $message .= "- Consider the website context and categories\n"; |
| 252 | $message .= "- HIGH SCORES (100-70): 3 keywords that are really good matches\n"; |
| 253 | $message .= "- MEDIUM SCORES (69-31): 2 keywords for broader matches\n"; |
| 254 | $message .= "- LOW SCORES (30 and below): 1 keyword for broadest possible matches\n"; |
| 255 | $message .= "- Each line must be unique\n"; |
| 256 | $message .= "- Format exactly as: SCORE: keyword1 keyword2 keyword3\n"; |
| 257 | $message .= "- Do NOT add any other text, bullets, or formatting\n\n"; |
| 258 | |
| 259 | $message .= "Example for 'funny adventure game' on a gaming website:\n"; |
| 260 | $message .= "100: funny adventure game\n"; |
| 261 | $message .= "97: adventure game comedy\n"; |
| 262 | $message .= "94: comedy adventure game\n"; |
| 263 | $message .= "91: funny game adventure\n"; |
| 264 | $message .= "88: adventure comedy game\n"; |
| 265 | $message .= "85: game adventure funny\n"; |
| 266 | $message .= "82: humor adventure game\n"; |
| 267 | $message .= "79: adventure game humor\n"; |
| 268 | $message .= "76: funny adventure quest\n"; |
| 269 | $message .= "73: comedy game adventure\n"; |
| 270 | $message .= "70: adventure game fun\n"; |
| 271 | $message .= "67: adventure game\n"; |
| 272 | $message .= "64: funny game\n"; |
| 273 | $message .= "61: comedy game\n"; |
| 274 | $message .= "58: adventure comedy\n"; |
| 275 | $message .= "55: game humor\n"; |
| 276 | $message .= "52: funny adventure\n"; |
| 277 | $message .= "49: adventure quest\n"; |
| 278 | $message .= "46: game comedy\n"; |
| 279 | $message .= "43: adventure story\n"; |
| 280 | $message .= "40: game fun\n"; |
| 281 | $message .= "37: funny story\n"; |
| 282 | $message .= "34: comedy quest\n"; |
| 283 | $message .= "31: adventure play\n"; |
| 284 | $message .= "30: adventure\n"; |
| 285 | $message .= "29: game\n"; |
| 286 | $message .= "28: funny\n"; |
| 287 | $message .= "27: comedy\n"; |
| 288 | $message .= "26: quest\n"; |
| 289 | $message .= "25: humor\n"; |
| 290 | $message .= "24: story\n"; |
| 291 | $message .= "23: play\n"; |
| 292 | $message .= "22: fun\n"; |
| 293 | $message .= "21: character\n"; |
| 294 | $message .= "20: world\n"; |
| 295 | $message .= "19: level\n"; |
| 296 | $message .= "18: puzzle\n"; |
| 297 | $message .= "17: island\n"; |
| 298 | $message .= "16: pirate\n"; |
| 299 | $message .= "15: monkey\n\n"; |
| 300 | |
| 301 | $message .= "IMPORTANT: Generate EXACTLY 40 queries following this format. Start at 100 and decrease by 2-3 points each time.\n"; |
| 302 | $message .= "Now generate 40 queries for \"$text\":\n"; |
| 303 | |
| 304 | $query = new Meow_MWAI_Query_Text( $message ); |
| 305 | $query->set_max_tokens( 2000 ); // Use max_tokens instead of max_results |
| 306 | |
| 307 | try { |
| 308 | $reply = $this->core->run_query( $query ); |
| 309 | if ( !empty( $reply->result ) ) { |
| 310 | $parsed = $this->parse_search_queries( $reply->result ); |
| 311 | if ( $parsed !== null ) { |
| 312 | return $parsed; |
| 313 | } |
| 314 | } |
| 315 | } |
| 316 | catch ( Exception $e ) { |
| 317 | error_log( 'AI Engine Search: Failed to generate search queries - ' . $e->getMessage() ); |
| 318 | } |
| 319 | |
| 320 | // Fallback |
| 321 | return $this->fallback_keyword_tiers( $text ); |
| 322 | } |
| 323 | |
| 324 | private function parse_search_queries( $ai_response ) { |
| 325 | $searches = []; |
| 326 | |
| 327 | |
| 328 | $lines = explode( "\n", $ai_response ); |
| 329 | foreach ( $lines as $line ) { |
| 330 | $line = trim( $line ); |
| 331 | if ( empty( $line ) ) continue; |
| 332 | |
| 333 | // Parse lines like "100: keyword1 keyword2 keyword3" |
| 334 | // Also handle lines that might start with a dash or bullet |
| 335 | $line = preg_replace( '/^[-•*]\s*/', '', $line ); |
| 336 | |
| 337 | if ( preg_match( '/^(\d+)\s*:\s*(.+)$/', $line, $matches ) ) { |
| 338 | $score = intval( $matches[1] ); |
| 339 | $keywords = trim( $matches[2] ); |
| 340 | |
| 341 | if ( $score >= 0 && $score <= 100 && !empty( $keywords ) ) { |
| 342 | $searches[] = [ |
| 343 | 'score' => $score, |
| 344 | 'keywords' => $keywords |
| 345 | ]; |
| 346 | } |
| 347 | } |
| 348 | } |
| 349 | |
| 350 | // If we got good results, return them |
| 351 | if ( count( $searches ) >= 5 ) { |
| 352 | // Sort by score descending |
| 353 | usort( $searches, function( $a, $b ) { |
| 354 | return $b['score'] <=> $a['score']; |
| 355 | } ); |
| 356 | return $searches; |
| 357 | } |
| 358 | |
| 359 | // Otherwise use fallback |
| 360 | return null; |
| 361 | } |
| 362 | |
| 363 | private function fallback_keyword_tiers( $text ) { |
| 364 | // Extract meaningful words |
| 365 | $words = str_word_count( strtolower( $text ), 1 ); |
| 366 | |
| 367 | // Remove stop words |
| 368 | $stop_words = [ 'i', 'me', 'my', 'we', 'our', 'you', 'your', 'he', 'she', 'it', 'they', |
| 369 | 'want', 'need', 'like', 'love', 'to', 'a', 'an', 'the', 'with', 'is', |
| 370 | 'for', 'of', 'and', 'or', 'but', 'in', 'on', 'at', 'which', 'that' ]; |
| 371 | |
| 372 | $meaningful = array_diff( $words, $stop_words ); |
| 373 | $meaningful = array_values( $meaningful ); |
| 374 | |
| 375 | // Synonyms for common terms - focused on actual search intent |
| 376 | $synonyms = [ |
| 377 | 'game' => ['games', 'gaming', 'play', 'gameplay'], |
| 378 | 'space' => ['galaxy', 'universe', 'cosmic', 'stellar', 'sci-fi'], |
| 379 | 'funny' => ['humor', 'comedy', 'fun', 'humorous'], |
| 380 | 'adventure' => ['quest', 'journey', 'exploration', 'story'], |
| 381 | 'huge' => ['large', 'big', 'massive', 'vast'], |
| 382 | 'world' => ['universe', 'realm', 'map', 'environment'] |
| 383 | ]; |
| 384 | |
| 385 | // Create 40 search queries directly for better fallback |
| 386 | $search_queries = []; |
| 387 | $score = 100; |
| 388 | |
| 389 | // If we have no meaningful words, use the original text |
| 390 | if ( empty( $meaningful ) ) { |
| 391 | $meaningful = $words; |
| 392 | if ( empty( $meaningful ) ) { |
| 393 | // Last resort: split the original text |
| 394 | $meaningful = explode( ' ', $text ); |
| 395 | } |
| 396 | } |
| 397 | |
| 398 | // First batch: exact words from query (100-85) |
| 399 | if ( count( $meaningful ) >= 1 ) { |
| 400 | // 4-5 keywords |
| 401 | for ( $i = 0; $i < 5 && $score >= 85; $i++ ) { |
| 402 | $keywords = []; |
| 403 | shuffle( $meaningful ); |
| 404 | $num_keywords = min( 4 + rand(0, 1), count( $meaningful ) ); |
| 405 | $keywords = array_slice( $meaningful, 0, max( 1, $num_keywords ) ); |
| 406 | if ( count( $keywords ) >= 1 ) { |
| 407 | $search_queries[] = [ |
| 408 | 'score' => $score, |
| 409 | 'keywords' => implode( ' ', $keywords ) |
| 410 | ]; |
| 411 | $score -= 3; |
| 412 | } |
| 413 | } |
| 414 | } |
| 415 | |
| 416 | // Second batch: mix exact with synonyms (84-60) |
| 417 | $all_related = $meaningful; |
| 418 | foreach ( $meaningful as $word ) { |
| 419 | if ( isset( $synonyms[$word] ) ) { |
| 420 | $all_related = array_merge( $all_related, array_slice( $synonyms[$word], 0, 2 ) ); |
| 421 | } |
| 422 | } |
| 423 | $all_related = array_unique( $all_related ); |
| 424 | |
| 425 | for ( $i = 0; $i < 15 && $score >= 60; $i++ ) { |
| 426 | shuffle( $all_related ); |
| 427 | $num_keywords = 3 + rand( 0, 1 ); |
| 428 | $keywords = array_slice( $all_related, 0, min( $num_keywords, count( $all_related ) ) ); |
| 429 | if ( count( $keywords ) >= 2 ) { |
| 430 | $search_queries[] = [ |
| 431 | 'score' => $score, |
| 432 | 'keywords' => implode( ' ', $keywords ) |
| 433 | ]; |
| 434 | $score -= 2; |
| 435 | } |
| 436 | } |
| 437 | |
| 438 | // Third batch: fewer keywords (59-30) |
| 439 | for ( $i = 0; $i < 20 && $score >= 30; $i++ ) { |
| 440 | shuffle( $all_related ); |
| 441 | $keywords = array_slice( $all_related, 0, 2 ); |
| 442 | if ( count( $keywords ) >= 2 ) { |
| 443 | $search_queries[] = [ |
| 444 | 'score' => $score, |
| 445 | 'keywords' => implode( ' ', $keywords ) |
| 446 | ]; |
| 447 | $score -= 2; |
| 448 | } |
| 449 | } |
| 450 | |
| 451 | // If we couldn't generate any searches, create at least one with the original text |
| 452 | if ( empty( $search_queries ) ) { |
| 453 | $search_queries[] = [ |
| 454 | 'score' => 100, |
| 455 | 'keywords' => $text |
| 456 | ]; |
| 457 | } |
| 458 | |
| 459 | // Return search queries in same format as AI would generate |
| 460 | return array_slice( $search_queries, 0, 40 ); |
| 461 | } |
| 462 | |
| 463 | private function create_search_combinations( $searches, $max_searches = 40 ) { |
| 464 | // If we have AI-generated searches, use them directly |
| 465 | if ( is_array( $searches ) && !empty( $searches ) && isset( $searches[0] ) && isset( $searches[0]['keywords'] ) ) { |
| 466 | $combinations = []; |
| 467 | foreach ( $searches as $search ) { |
| 468 | // Skip empty keywords |
| 469 | if ( !empty( trim( $search['keywords'] ) ) ) { |
| 470 | $combinations[] = [ |
| 471 | 'keywords' => $search['keywords'], |
| 472 | 'score' => $search['score'], |
| 473 | 'strategy' => 'ai_generated' |
| 474 | ]; |
| 475 | } |
| 476 | if ( count( $combinations ) >= $max_searches ) { |
| 477 | break; |
| 478 | } |
| 479 | } |
| 480 | // If we have some combinations, return them |
| 481 | if ( !empty( $combinations ) ) { |
| 482 | return $combinations; |
| 483 | } |
| 484 | } |
| 485 | |
| 486 | // Otherwise, this is the fallback format, generate combinations |
| 487 | $tiers = $searches; |
| 488 | $combinations = []; |
| 489 | $exact = $tiers['exact'] ?? []; |
| 490 | $contextual = $tiers['contextual'] ?? []; |
| 491 | $general = $tiers['general'] ?? []; |
| 492 | |
| 493 | // Simple fallback algorithm |
| 494 | $search_count = 0; |
| 495 | |
| 496 | // Mix of different combinations - start at 100 |
| 497 | $strategies = [ |
| 498 | [ 'exact' => 4, 'score' => 100 ], |
| 499 | [ 'exact' => 3, 'score' => 90 ], |
| 500 | [ 'exact' => 2, 'contextual' => 2, 'score' => 80 ], |
| 501 | [ 'exact' => 2, 'score' => 70 ], |
| 502 | [ 'exact' => 1, 'contextual' => 2, 'score' => 60 ], |
| 503 | [ 'contextual' => 2, 'score' => 50 ], |
| 504 | [ 'exact' => 1, 'general' => 1, 'score' => 40 ], |
| 505 | [ 'general' => 2, 'score' => 35 ] |
| 506 | ]; |
| 507 | |
| 508 | foreach ( $strategies as $strategy ) { |
| 509 | for ( $i = 0; $i < 5 && $search_count < $max_searches; $i++ ) { |
| 510 | $keywords = []; |
| 511 | |
| 512 | if ( isset( $strategy['exact'] ) && count( $exact ) >= $strategy['exact'] ) { |
| 513 | shuffle( $exact ); |
| 514 | $keywords = array_merge( $keywords, array_slice( $exact, 0, $strategy['exact'] ) ); |
| 515 | } |
| 516 | |
| 517 | if ( isset( $strategy['contextual'] ) && count( $contextual ) >= $strategy['contextual'] ) { |
| 518 | shuffle( $contextual ); |
| 519 | $keywords = array_merge( $keywords, array_slice( $contextual, 0, $strategy['contextual'] ) ); |
| 520 | } |
| 521 | |
| 522 | if ( isset( $strategy['general'] ) && count( $general ) >= $strategy['general'] ) { |
| 523 | shuffle( $general ); |
| 524 | $keywords = array_merge( $keywords, array_slice( $general, 0, $strategy['general'] ) ); |
| 525 | } |
| 526 | |
| 527 | if ( count( $keywords ) >= 2 ) { |
| 528 | $keywords_str = implode( ' ', $keywords ); |
| 529 | if ( !empty( trim( $keywords_str ) ) ) { |
| 530 | $combinations[] = [ |
| 531 | 'keywords' => $keywords_str, |
| 532 | 'score' => $strategy['score'] + rand( -5, 5 ), |
| 533 | 'strategy' => 'fallback' |
| 534 | ]; |
| 535 | $search_count++; |
| 536 | } |
| 537 | } |
| 538 | } |
| 539 | } |
| 540 | |
| 541 | // If we have no combinations, create at least one from whatever we have |
| 542 | if ( empty( $combinations ) ) { |
| 543 | // Try to create a basic search from exact keywords |
| 544 | if ( !empty( $exact ) ) { |
| 545 | $combinations[] = [ |
| 546 | 'keywords' => implode( ' ', array_slice( $exact, 0, 3 ) ), |
| 547 | 'score' => 50, |
| 548 | 'strategy' => 'fallback_emergency' |
| 549 | ]; |
| 550 | } |
| 551 | // Or from the original text |
| 552 | else if ( is_string( $searches ) && !empty( trim( $searches ) ) ) { |
| 553 | $combinations[] = [ |
| 554 | 'keywords' => $searches, // This will be the original text in worst case |
| 555 | 'score' => 30, |
| 556 | 'strategy' => 'fallback_original' |
| 557 | ]; |
| 558 | } |
| 559 | } |
| 560 | |
| 561 | return array_slice( $combinations, 0, $max_searches ); |
| 562 | } |
| 563 | |
| 564 | private function get_combinations( $array, $length ) { |
| 565 | if ( $length == 1 ) { |
| 566 | return array_map( function( $el ) { return [ $el ]; }, $array ); |
| 567 | } |
| 568 | |
| 569 | $combinations = []; |
| 570 | $array_length = count( $array ); |
| 571 | |
| 572 | for ( $i = 0; $i < $array_length - $length + 1; $i++ ) { |
| 573 | $head = array_slice( $array, $i, 1 ); |
| 574 | $tail_combinations = $this->get_combinations( array_slice( $array, $i + 1 ), $length - 1 ); |
| 575 | foreach ( $tail_combinations as $tail ) { |
| 576 | $combinations[] = array_merge( $head, $tail ); |
| 577 | } |
| 578 | } |
| 579 | |
| 580 | return $combinations; |
| 581 | } |
| 582 | |
| 583 | private function search_with_keywords( $search_queries ) { |
| 584 | $all_results = []; |
| 585 | $searches_performed = 0; |
| 586 | $max_searches = 40; |
| 587 | $min_results_needed = 3; |
| 588 | |
| 589 | // Create search combinations with scores |
| 590 | $search_combinations = $this->create_search_combinations( $search_queries, $max_searches ); |
| 591 | |
| 592 | $debug_searches = []; |
| 593 | |
| 594 | foreach ( $search_combinations as $combination ) { |
| 595 | $searches_performed++; |
| 596 | $keywords = $combination['keywords']; |
| 597 | $score = $combination['score']; |
| 598 | $strategy = $combination['strategy']; |
| 599 | |
| 600 | // Record what we're searching |
| 601 | $debug_searches[] = [ |
| 602 | 'attempt' => $searches_performed, |
| 603 | 'keywords' => $keywords, |
| 604 | 'score' => $score, |
| 605 | 'strategy' => $strategy, |
| 606 | 'found' => 0 |
| 607 | ]; |
| 608 | |
| 609 | // Perform the search |
| 610 | $posts = get_posts( [ |
| 611 | 's' => $keywords, |
| 612 | 'posts_per_page' => 10, |
| 613 | 'post_type' => 'post', |
| 614 | 'post_status' => 'publish', |
| 615 | 'fields' => 'ids' |
| 616 | ] ); |
| 617 | |
| 618 | if ( !empty( $posts ) ) { |
| 619 | // Update found count |
| 620 | $debug_searches[count($debug_searches) - 1]['found'] = count( $posts ); |
| 621 | |
| 622 | // Add to results with score |
| 623 | foreach ( $posts as $post_id ) { |
| 624 | if ( !isset( $all_results[$post_id] ) ) { |
| 625 | $all_results[$post_id] = [ |
| 626 | 'id' => $post_id, |
| 627 | 'best_score' => $score, |
| 628 | 'found_with' => [] |
| 629 | ]; |
| 630 | } else { |
| 631 | // Keep the best (highest) score |
| 632 | if ( $score > $all_results[$post_id]['best_score'] ) { |
| 633 | $all_results[$post_id]['best_score'] = $score; |
| 634 | } |
| 635 | } |
| 636 | $all_results[$post_id]['found_with'][] = [ |
| 637 | 'keywords' => $keywords, |
| 638 | 'score' => $score |
| 639 | ]; |
| 640 | } |
| 641 | |
| 642 | // Stop if we have enough unique results |
| 643 | if ( count( $all_results ) >= $min_results_needed ) { |
| 644 | break; |
| 645 | } |
| 646 | } |
| 647 | |
| 648 | // Stop if we've done too many searches |
| 649 | if ( $searches_performed >= $max_searches ) { |
| 650 | break; |
| 651 | } |
| 652 | } |
| 653 | |
| 654 | // Sort results by score (highest first) |
| 655 | uasort( $all_results, function( $a, $b ) { |
| 656 | return $b['best_score'] <=> $a['best_score']; |
| 657 | } ); |
| 658 | |
| 659 | // Get full post data for results |
| 660 | $final_results = []; |
| 661 | $post_ids = array_keys( $all_results ); |
| 662 | |
| 663 | if ( !empty( $post_ids ) ) { |
| 664 | // Get posts but maintain our score order |
| 665 | $posts_data = []; |
| 666 | $posts = get_posts( [ |
| 667 | 'post__in' => $post_ids, |
| 668 | 'posts_per_page' => 20, |
| 669 | 'post_type' => 'post', |
| 670 | 'post_status' => 'publish' |
| 671 | ] ); |
| 672 | |
| 673 | // Create a map for easy access |
| 674 | foreach ( $posts as $post ) { |
| 675 | $posts_data[$post->ID] = $post; |
| 676 | } |
| 677 | |
| 678 | // Build results in score order |
| 679 | foreach ( $post_ids as $post_id ) { |
| 680 | if ( isset( $posts_data[$post_id] ) ) { |
| 681 | $post = $posts_data[$post_id]; |
| 682 | $result_data = $all_results[$post_id]; |
| 683 | |
| 684 | // Get the keywords that found this post with the best score |
| 685 | $best_keywords = ''; |
| 686 | foreach ( $result_data['found_with'] as $found ) { |
| 687 | if ( $found['score'] == $result_data['best_score'] ) { |
| 688 | $best_keywords = $found['keywords']; |
| 689 | break; |
| 690 | } |
| 691 | } |
| 692 | |
| 693 | $final_results[] = [ |
| 694 | 'id' => $post->ID, |
| 695 | 'title' => get_the_title( $post ), |
| 696 | 'excerpt' => wp_trim_words( $post->post_content, 30 ), |
| 697 | 'score' => $result_data['best_score'] / 100, // Convert 0-100 to 0-1 for frontend consistency |
| 698 | 'found_with' => $best_keywords |
| 699 | ]; |
| 700 | } |
| 701 | } |
| 702 | } |
| 703 | |
| 704 | return [ |
| 705 | 'results' => $final_results, |
| 706 | 'debug' => [ |
| 707 | 'total_searches' => $searches_performed, |
| 708 | 'keyword_tiers' => is_array( $search_queries ) && isset( $search_queries['exact'] ) ? $search_queries : null, |
| 709 | 'searches' => $debug_searches |
| 710 | ] |
| 711 | ]; |
| 712 | } |
| 713 | |
| 714 | private function search_with_embeddings( $search_text, $env_id = null ) { |
| 715 | if ( !class_exists( 'MeowPro_MWAI_Embeddings' ) ) { |
| 716 | return [ 'error' => 'Embeddings module not available' ]; |
| 717 | } |
| 718 | |
| 719 | // Validate environment exists |
| 720 | if ( !$env_id ) { |
| 721 | return [ 'error' => 'No embeddings environment selected' ]; |
| 722 | } |
| 723 | |
| 724 | $env = $this->core->get_embeddings_env( $env_id ); |
| 725 | if ( !$env ) { |
| 726 | return [ 'error' => 'Invalid embeddings environment selected. Please select a valid environment.' ]; |
| 727 | } |
| 728 | |
| 729 | try { |
| 730 | // Get the embeddings instance |
| 731 | $embeddings = new MeowPro_MWAI_Embeddings( $this->core ); |
| 732 | |
| 733 | // Use the query_vectors method which handles everything internally |
| 734 | // Parameters: offset, limit, filters, sort |
| 735 | $filters = [ |
| 736 | 'envId' => $env_id, |
| 737 | 'search' => $search_text |
| 738 | ]; |
| 739 | $result = $embeddings->query_vectors( 0, 20, $filters ); |
| 740 | |
| 741 | $vectors = isset( $result['rows'] ) ? $result['rows'] : []; |
| 742 | |
| 743 | if ( empty( $vectors ) ) { |
| 744 | return [ |
| 745 | 'post_ids' => [], |
| 746 | 'debug' => [ |
| 747 | 'total_vectors' => 0, |
| 748 | 'message' => 'No matching vectors found' |
| 749 | ] |
| 750 | ]; |
| 751 | } |
| 752 | |
| 753 | // Extract post IDs from results |
| 754 | $post_ids = []; |
| 755 | $debug_info = []; |
| 756 | |
| 757 | foreach ( $vectors as $vector ) { |
| 758 | $debug_info[] = [ |
| 759 | 'refId' => $vector['refId'] ?? 'unknown', |
| 760 | 'score' => $vector['score'] ?? 0, |
| 761 | 'type' => $vector['type'] ?? 'unknown', |
| 762 | 'title' => $vector['title'] ?? 'unknown' |
| 763 | ]; |
| 764 | |
| 765 | // Check if this is a post embedding |
| 766 | if ( !empty( $vector['type'] ) && $vector['type'] === 'postId' && !empty( $vector['refId'] ) ) { |
| 767 | $score = isset( $vector['score'] ) ? (float)$vector['score'] : 0; |
| 768 | $post_ids[] = [ |
| 769 | 'id' => (int)$vector['refId'], |
| 770 | 'score' => $score |
| 771 | ]; |
| 772 | } |
| 773 | } |
| 774 | |
| 775 | // Sort by score descending |
| 776 | usort( $post_ids, function( $a, $b ) { |
| 777 | return $b['score'] <=> $a['score']; |
| 778 | } ); |
| 779 | |
| 780 | return [ |
| 781 | 'post_ids' => $post_ids, |
| 782 | 'debug' => [ |
| 783 | 'total_vectors' => count( $vectors ), |
| 784 | 'filtered_posts' => count( $post_ids ), |
| 785 | 'sample_vectors' => array_slice( $debug_info, 0, 5 ) |
| 786 | ] |
| 787 | ]; |
| 788 | } |
| 789 | catch ( Exception $e ) { |
| 790 | error_log( 'AI Engine Search: Embeddings search failed - ' . $e->getMessage() ); |
| 791 | return [ 'error' => 'Embeddings search failed: ' . $e->getMessage() ]; |
| 792 | } |
| 793 | } |
| 794 | |
| 795 | public function rest_search( $request ) { |
| 796 | $params = $request->get_json_params(); |
| 797 | $search = isset( $params['search'] ) ? sanitize_text_field( $params['search'] ) : ''; |
| 798 | $method = isset( $params['method'] ) ? $params['method'] : 'wordpress'; |
| 799 | $env_id = isset( $params['envId'] ) ? $params['envId'] : null; |
| 800 | $website_context = isset( $params['websiteContext'] ) ? sanitize_text_field( $params['websiteContext'] ) : ''; |
| 801 | |
| 802 | if ( empty( $search ) ) { |
| 803 | return new WP_REST_Response( [ 'success' => false, 'message' => 'Empty search' ], 400 ); |
| 804 | } |
| 805 | |
| 806 | $results = []; |
| 807 | $debug_info = []; |
| 808 | |
| 809 | if ( $method === 'wordpress' ) { |
| 810 | // Use standard WordPress search |
| 811 | $posts = get_posts( [ |
| 812 | 's' => $search, |
| 813 | 'posts_per_page' => 20, |
| 814 | 'post_type' => 'post', |
| 815 | 'post_status' => 'publish' |
| 816 | ] ); |
| 817 | |
| 818 | foreach ( $posts as $post ) { |
| 819 | $results[] = [ |
| 820 | 'id' => $post->ID, |
| 821 | 'title' => get_the_title( $post ), |
| 822 | 'excerpt' => wp_trim_words( $post->post_content, 30 ) |
| 823 | ]; |
| 824 | } |
| 825 | |
| 826 | $debug_info = [ |
| 827 | 'method' => 'Standard WordPress search', |
| 828 | 'query' => $search, |
| 829 | 'found' => count( $posts ) |
| 830 | ]; |
| 831 | } |
| 832 | elseif ( $method === 'embeddings' ) { |
| 833 | // Search using embeddings |
| 834 | $embedding_result = $this->search_with_embeddings( $search, $env_id ); |
| 835 | |
| 836 | if ( isset( $embedding_result['error'] ) ) { |
| 837 | return new WP_REST_Response( [ |
| 838 | 'success' => false, |
| 839 | 'message' => $embedding_result['error'], |
| 840 | 'debug' => $embedding_result['debug'] ?? null |
| 841 | ], 200 ); |
| 842 | } |
| 843 | |
| 844 | $debug_info = $embedding_result['debug'] ?? []; |
| 845 | |
| 846 | if ( !empty( $embedding_result['post_ids'] ) ) { |
| 847 | $post_ids = array_map( function( $result ) { |
| 848 | return $result['id']; |
| 849 | }, $embedding_result['post_ids'] ); |
| 850 | |
| 851 | $posts = get_posts( [ |
| 852 | 'post__in' => $post_ids, |
| 853 | 'orderby' => 'post__in', |
| 854 | 'posts_per_page' => count( $post_ids ), |
| 855 | 'post_type' => 'post', |
| 856 | 'post_status' => 'publish' |
| 857 | ] ); |
| 858 | |
| 859 | // Map posts with their scores |
| 860 | $score_map = []; |
| 861 | foreach ( $embedding_result['post_ids'] as $result ) { |
| 862 | $score_map[$result['id']] = $result['score']; |
| 863 | } |
| 864 | |
| 865 | foreach ( $posts as $post ) { |
| 866 | $results[] = [ |
| 867 | 'id' => $post->ID, |
| 868 | 'title' => get_the_title( $post ), |
| 869 | 'excerpt' => wp_trim_words( $post->post_content, 30 ), |
| 870 | 'score' => $score_map[$post->ID] ?? 0 |
| 871 | ]; |
| 872 | } |
| 873 | } |
| 874 | } else { |
| 875 | // Search using AI keywords with smart algorithm |
| 876 | $search_queries = $this->generate_keyword_tiers( $search, $website_context ); |
| 877 | $keyword_result = $this->search_with_keywords( $search_queries ); |
| 878 | |
| 879 | $results = $keyword_result['results']; |
| 880 | $debug_info = $keyword_result['debug']; |
| 881 | } |
| 882 | |
| 883 | $response = [ |
| 884 | 'success' => true, |
| 885 | 'results' => $results, |
| 886 | 'method' => $method, |
| 887 | 'debug' => $debug_info |
| 888 | ]; |
| 889 | |
| 890 | return new WP_REST_Response( $response, 200 ); |
| 891 | } |
| 892 | |
| 893 | public function customize_search_display( $search_query ) { |
| 894 | // Only customize on frontend search pages |
| 895 | if ( is_admin() || !is_search() ) { |
| 896 | return $search_query; |
| 897 | } |
| 898 | |
| 899 | // Get the frontend search method setting |
| 900 | $frontend_method = $this->core->get_option( 'search_frontend_method', 'wordpress' ); |
| 901 | |
| 902 | // If using standard WordPress search, no customization needed |
| 903 | if ( $frontend_method === 'wordpress' ) { |
| 904 | return $search_query; |
| 905 | } |
| 906 | |
| 907 | // Check if this was an AI-powered search by looking for our special markers |
| 908 | global $wp_query; |
| 909 | $current_search = $wp_query->get( 's' ); |
| 910 | $original_search = $wp_query->get( 'mwai_original_search' ); |
| 911 | |
| 912 | // Check if current search is one of our AI search markers |
| 913 | $is_keywords_search = strpos( $search_query, 'MWAI_KEYWORDS_SEARCH_' ) === 0; |
| 914 | $is_embeddings_search = strpos( $search_query, 'MWAI_EMBEDDINGS_SEARCH_' ) === 0; |
| 915 | |
| 916 | // If we have an original search stored and this is an AI search, just return the original search |
| 917 | if ( !empty( $original_search ) && ( $is_keywords_search || $is_embeddings_search ) ) { |
| 918 | return $original_search; |
| 919 | } |
| 920 | |
| 921 | return $search_query; |
| 922 | } |
| 923 | } |