| @@ -11,40 +11,11 @@ | ||
| 11 | 11 | private $productCardHtml; |
| 12 | 12 | private $word_handler; |
| 13 | 13 | private $last_similarity_analysis = null; |
| 14 | 14 | private $current_valid_urls = []; |
| 15 | - private $last_vectorstore_error = null; | |
| 16 | 15 | private $is_streaming = false; // ADDED: Track if current request is streaming |
| 17 | - private $streaming_headers_sent = false; // Track if streaming headers have been sent | |
| 18 | 16 | |
| 19 | 17 | /** |
| 20 | - * Setup streaming headers - call this right before actually streaming | |
| 21 | - * This delays header setup to allow actions/forms to return JSON responses | |
| 22 | - */ | |
| 23 | -private function setup_streaming_headers() { | |
| 24 | - if ($this->streaming_headers_sent || headers_sent()) { | |
| 25 | - return false; | |
| 26 | - } | |
| 27 | - | |
| 28 | - // Disable output buffering | |
| 29 | - while (ob_get_level()) { | |
| 30 | - ob_end_flush(); | |
| 31 | - } | |
| 32 | - | |
| 33 | - // Set headers for SSE | |
| 34 | - header('Content-Type: text/event-stream'); | |
| 35 | - header('Cache-Control: no-cache'); | |
| 36 | - header('Connection: keep-alive'); | |
| 37 | - header('X-Accel-Buffering: no'); | |
| 38 | - | |
| 39 | - ob_implicit_flush(true); | |
| 40 | - flush(); | |
| 41 | - | |
| 42 | - $this->streaming_headers_sent = true; | |
| 43 | - return true; | |
| 44 | -} | |
| 45 | - | |
| 46 | -/** | |
| 47 | 18 | * Class constructor |
| 48 | 19 | */ |
| 49 | 20 | public function __construct() { |
| 50 | 21 | $this->options = get_option('mxchat_options'); |
| @@ -111,28 +82,16 @@ | ||
| 111 | 82 | // Add chat mode checking actions |
| 112 | 83 | add_action('wp_ajax_mxchat_get_current_chat_mode', array($this, 'mxchat_get_current_chat_mode')); |
| 113 | 84 | add_action('wp_ajax_nopriv_mxchat_get_current_chat_mode', array($this, 'mxchat_get_current_chat_mode')); |
| 114 | 85 | |
| 115 | - // Nonce refresh for page-cache compatibility (WP Rocket, LiteSpeed, etc.) | |
| 116 | - add_action('wp_ajax_mxchat_refresh_nonce', array($this, 'mxchat_refresh_nonce')); | |
| 117 | - add_action('wp_ajax_nopriv_mxchat_refresh_nonce', array($this, 'mxchat_refresh_nonce')); | |
| 118 | - | |
| 119 | 86 | // Auto-email transcript action |
| 120 | 87 | add_action('mxchat_send_delayed_transcript', array($this, 'mxchat_send_delayed_transcript'), 10, 1); |
| 121 | - | |
| 88 | + | |
| 122 | 89 | add_filter('mxchat_check_actions_only', array($this, 'check_actions_for_addons'), 10, 4); |
| 123 | 90 | |
| 124 | 91 | |
| 125 | 92 | } |
| 126 | 93 | |
| 127 | -/** | |
| 128 | - * Return a fresh nonce so cached pages can replace the stale one. | |
| 129 | - */ | |
| 130 | -public function mxchat_refresh_nonce() { | |
| 131 | - nocache_headers(); | |
| 132 | - wp_send_json_success(array('nonce' => wp_create_nonce('mxchat_chat_nonce'))); | |
| 133 | -} | |
| 134 | - | |
| 135 | 94 | // In your core plugin's check_actions_for_addons method: |
| 136 | 95 | public function check_actions_for_addons($default, $message, $user_id, $session_id) { |
| 137 | 96 | //error_log('MxChat Core: check_actions_for_addons called with message: ' . $message); |
| 138 | 97 | |
| @@ -157,17 +116,21 @@ | ||
| 157 | 116 | |
| 158 | 117 | $session_id = sanitize_text_field($_POST['session_id']); |
| 159 | 118 | |
| 160 | 119 | // SECURITY FIX: Verify session ownership before retrieving data |
| 161 | - // If IP/user changed, signal frontend to reset session instead of blocking | |
| 162 | 120 | $current_user_identifier = MxChat_User::mxchat_get_user_identifier(); |
| 163 | - | |
| 121 | + | |
| 164 | 122 | // Check if this session has an owner recorded |
| 165 | 123 | $session_owner = get_option("mxchat_session_owner_{$session_id}"); |
| 166 | - | |
| 167 | - // Update session owner if it changed (e.g. IP changed due to network switch) | |
| 168 | - // The session ID itself is the authentication — if the client has it, they own it | |
| 169 | - if (!$session_owner || $session_owner !== $current_user_identifier) { | |
| 124 | + | |
| 125 | + // If session has an owner and it doesn't match current user, deny access | |
| 126 | + if ($session_owner && $session_owner !== $current_user_identifier) { | |
| 127 | + wp_send_json_error(['message' => esc_html__('Unauthorized access.', 'mxchat')]); | |
| 128 | + wp_die(); | |
| 129 | + } | |
| 130 | + | |
| 131 | + // If no owner is set yet, claim ownership (for legacy sessions) | |
| 132 | + if (!$session_owner) { | |
| 170 | 133 | update_option("mxchat_session_owner_{$session_id}", $current_user_identifier, 'no'); |
| 171 | 134 | } |
| 172 | 135 | |
| 173 | 136 | $history = get_option("mxchat_history_{$session_id}", []); // Retrieve stored history |
| @@ -187,25 +150,10 @@ | ||
| 187 | 150 | 'chat_mode' => $chat_mode |
| 188 | 151 | ]); |
| 189 | 152 | wp_die(); |
| 190 | 153 | } |
| 191 | -private function mxchat_fetch_conversation_history_for_ai($session_id, $session_start_timestamp = 0) { | |
| 154 | +private function mxchat_fetch_conversation_history_for_ai($session_id) { | |
| 192 | 155 | $history = get_option("mxchat_history_{$session_id}", []); |
| 193 | - | |
| 194 | - // Check persistence setting - when OFF, only include messages from current page load | |
| 195 | - $options = get_option('mxchat_options', []); | |
| 196 | - $persistence_enabled = isset($options['chat_persistence_toggle']) && $options['chat_persistence_toggle'] === 'on'; | |
| 197 | - | |
| 198 | - // Filter history when persistence is OFF to match what the user sees | |
| 199 | - if (!$persistence_enabled && $session_start_timestamp > 0) { | |
| 200 | - $history = array_filter($history, function($entry) use ($session_start_timestamp) { | |
| 201 | - // Include messages from this page load onwards | |
| 202 | - return isset($entry['timestamp']) && $entry['timestamp'] >= $session_start_timestamp; | |
| 203 | - }); | |
| 204 | - // Re-index array after filtering | |
| 205 | - $history = array_values($history); | |
| 206 | - } | |
| 207 | - | |
| 208 | 156 | $formatted_history = []; |
| 209 | 157 | |
| 210 | 158 | // Adjusted for code-heavy conversations |
| 211 | 159 | $max_tokens = 120000; // Context window size |
| @@ -303,15 +251,8 @@ | ||
| 303 | 251 | 'callback' => [$this, 'handle_slack_messages'], |
| 304 | 252 | 'permission_callback' => [$this, 'verify_slack_request'], |
| 305 | 253 | ]); |
| 306 | 254 | |
| 307 | - // Telegram webhook endpoint | |
| 308 | - register_rest_route('mxchat/v1', '/telegram-webhook', [ | |
| 309 | - 'methods' => 'POST', | |
| 310 | - 'callback' => [$this, 'handle_telegram_webhook'], | |
| 311 | - 'permission_callback' => [$this, 'verify_telegram_request'], | |
| 312 | - ]); | |
| 313 | - | |
| 314 | 255 | //error_log(esc_html__('MxChat REST routes registered', 'mxchat')); |
| 315 | 256 | } |
| 316 | 257 | |
| 317 | 258 | /** |
| @@ -351,11 +292,10 @@ | ||
| 351 | 292 | //error_log(esc_html__('Slack request timestamp too old', 'mxchat')); |
| 352 | 293 | return false; |
| 353 | 294 | } |
| 354 | 295 | |
| 355 | - // Get raw request body from the WP_REST_Request object | |
| 356 | - // (php://input may already be consumed by WordPress at this point) | |
| 357 | - $request_body = $request->get_body(); | |
| 296 | + // Get raw request body | |
| 297 | + $request_body = file_get_contents('php://input'); | |
| 358 | 298 | |
| 359 | 299 | // Create the signature base string |
| 360 | 300 | $sig_basestring = "v0:{$timestamp}:{$request_body}"; |
| 361 | 301 | |
| @@ -364,43 +304,8 @@ | ||
| 364 | 304 | |
| 365 | 305 | // Compare signatures |
| 366 | 306 | return hash_equals($my_signature, $slack_signature); |
| 367 | 307 | } |
| 368 | - | |
| 369 | -/** | |
| 370 | - * Verify request is coming from Telegram. | |
| 371 | - * | |
| 372 | - * @param WP_REST_Request $request | |
| 373 | - * @return bool True if valid, false otherwise. | |
| 374 | - */ | |
| 375 | -public function verify_telegram_request($request) { | |
| 376 | - $secret_token = $this->options['telegram_webhook_secret'] ?? ''; | |
| 377 | - | |
| 378 | - //error_log('[MxChat Telegram DEBUG] verify_telegram_request called'); | |
| 379 | - //error_log('[MxChat Telegram DEBUG] Stored secret: ' . (empty($secret_token) ? 'EMPTY' : substr($secret_token, 0, 10) . '...')); | |
| 380 | - | |
| 381 | - if (empty($secret_token)) { | |
| 382 | - // If no secret is configured, allow the request (for initial setup) | |
| 383 | - //error_log('[MxChat Telegram DEBUG] No secret configured, allowing request'); | |
| 384 | - return true; | |
| 385 | - } | |
| 386 | - | |
| 387 | - // Telegram sends the secret token in the X-Telegram-Bot-Api-Secret-Token header | |
| 388 | - $request_token = $request->get_header('X-Telegram-Bot-Api-Secret-Token'); | |
| 389 | - | |
| 390 | - //error_log('[MxChat Telegram DEBUG] Request token: ' . (empty($request_token) ? 'EMPTY' : substr($request_token, 0, 10) . '...')); | |
| 391 | - | |
| 392 | - if (empty($request_token)) { | |
| 393 | - //error_log('[MxChat Telegram DEBUG] Request rejected: No token in header'); | |
| 394 | - return false; | |
| 395 | - } | |
| 396 | - | |
| 397 | - // Timing-safe comparison | |
| 398 | - $result = hash_equals($secret_token, $request_token); | |
| 399 | - //error_log('[MxChat Telegram DEBUG] Token comparison result: ' . ($result ? 'MATCH' : 'MISMATCH')); | |
| 400 | - return $result; | |
| 401 | -} | |
| 402 | - | |
| 403 | 308 | public function mxchat_stream_events(WP_REST_Request $request) { |
| 404 | 309 | header('Content-Type: text/event-stream'); |
| 405 | 310 | header('Cache-Control: no-cache'); |
| 406 | 311 | header('Connection: keep-alive'); |
| @@ -434,9 +339,9 @@ | ||
| 434 | 339 | |
| 435 | 340 | |
| 436 | 341 | |
| 437 | 342 | |
| 438 | -private function mxchat_save_chat_message($session_id, $role, $message, $originating_page = null, $rag_context = null) { | |
| 343 | +private function mxchat_save_chat_message($session_id, $role, $message, $originating_page = null) { | |
| 439 | 344 | global $wpdb; |
| 440 | 345 | $table_name = $wpdb->prefix . 'mxchat_chat_transcripts'; |
| 441 | 346 | //error_log("[DEBUG] mxchat_save_chat_message -> START for session_id: {$session_id}, role: {$role}"); |
| 442 | 347 | |
| @@ -610,17 +515,9 @@ | ||
| 610 | 515 | $insert_data['originating_page_title'] = $stored_originating['title'] ?? ''; |
| 611 | 516 | } |
| 612 | 517 | } |
| 613 | 518 | } |
| 614 | - | |
| 615 | - // Add RAG context if provided (for bot messages) | |
| 616 | - if ($rag_context !== null && $role === 'bot') { | |
| 617 | - $rag_context_column_exists = $wpdb->get_var("SHOW COLUMNS FROM $table_name LIKE 'rag_context'"); | |
| 618 | - if ($rag_context_column_exists) { | |
| 619 | - $insert_data['rag_context'] = is_array($rag_context) ? wp_json_encode($rag_context) : $rag_context; | |
| 620 | - } | |
| 621 | - } | |
| 622 | - | |
| 519 | + | |
| 623 | 520 | $wpdb->insert($table_name, $insert_data); |
| 624 | 521 | //error_log("[DEBUG] mxchat_save_chat_message -> Inserted message into DB. row_id: {$wpdb->insert_id}, data: " . print_r($insert_data, true)); |
| 625 | 522 | |
| 626 | 523 | // 9) Send notification email if this is the first user message in a new session |
| @@ -833,17 +730,12 @@ | ||
| 833 | 730 | $transcript_content .= "[{$msg->timestamp}] {$role_label}:\n"; |
| 834 | 731 | $transcript_content .= $msg->message . "\n\n"; |
| 835 | 732 | } |
| 836 | 733 | |
| 837 | - // Create temporary file for attachment using WP_Filesystem | |
| 734 | + // Create temporary file for attachment | |
| 838 | 735 | $upload_dir = wp_upload_dir(); |
| 839 | 736 | $temp_file = $upload_dir['basedir'] . '/mxchat-transcript-' . $session_id . '.txt'; |
| 840 | - global $wp_filesystem; | |
| 841 | - if (empty($wp_filesystem)) { | |
| 842 | - require_once ABSPATH . 'wp-admin/includes/file.php'; | |
| 843 | - WP_Filesystem(); | |
| 844 | - } | |
| 845 | - $wp_filesystem->put_contents($temp_file, $transcript_content, FS_CHMOD_FILE); | |
| 737 | + file_put_contents($temp_file, $transcript_content); | |
| 846 | 738 | |
| 847 | 739 | // Prepare email |
| 848 | 740 | $subject = sprintf('[%s] Chat Transcript - Session %s', get_bloginfo('name'), substr($session_id, 0, 8)); |
| 849 | 741 | |
| @@ -874,10 +766,8 @@ | ||
| 874 | 766 | public function mxchat_handle_save_email_and_response() { |
| 875 | 767 | //error_log('[DEBUG] ---------- mxchat_handle_save_email_and_response START ----------'); |
| 876 | 768 | //error_log('DEBUG: POST data: ' . print_r($_POST, true)); |
| 877 | 769 | |
| 878 | - nocache_headers(); | |
| 879 | - | |
| 880 | 770 | // Validate nonce |
| 881 | 771 | if (!isset($_POST['nonce']) || !wp_verify_nonce($_POST['nonce'], 'mxchat_chat_nonce')) { |
| 882 | 772 | //error_log(esc_html__('[ERROR] Invalid nonce in mxchat_handle_save_email_and_response', 'mxchat')); |
| 883 | 773 | wp_send_json_error(['message' => esc_html__('Invalid nonce.', 'mxchat')]); |
| @@ -889,9 +779,9 @@ | ||
| 889 | 779 | $name = isset($_POST['name']) ? sanitize_text_field($_POST['name']) : ''; |
| 890 | 780 | |
| 891 | 781 | //error_log("[DEBUG] handle_save_email_and_response -> session_id: {$session_id}, email: {$email}, name: {$name}"); |
| 892 | 782 | |
| 893 | - if (empty($session_id) || $session_id === 'null' || empty($email)) { | |
| 783 | + if (empty($session_id) || empty($email)) { | |
| 894 | 784 | //error_log("[ERROR] Missing session_id or email: session_id={$session_id}, email={$email}"); |
| 895 | 785 | wp_send_json_error(['message' => esc_html__('Session ID or email is missing.', 'mxchat')]); |
| 896 | 786 | wp_die(); |
| 897 | 787 | } |
| @@ -908,15 +798,15 @@ | ||
| 908 | 798 | } |
| 909 | 799 | |
| 910 | 800 | // 1) Always store email in wp_options |
| 911 | 801 | $email_option_key = "mxchat_email_{$session_id}"; |
| 912 | - update_option($email_option_key, $email, 'no'); | |
| 802 | + update_option($email_option_key, $email); | |
| 913 | 803 | //error_log("[DEBUG] handle_save_email_and_response -> updated option: {$email_option_key} => {$email}"); |
| 914 | 804 | |
| 915 | 805 | // Store name in wp_options if provided |
| 916 | 806 | if (!empty($name)) { |
| 917 | 807 | $name_option_key = "mxchat_name_{$session_id}"; |
| 918 | - update_option($name_option_key, $name, 'no'); | |
| 808 | + update_option($name_option_key, $name); | |
| 919 | 809 | //error_log("[DEBUG] handle_save_email_and_response -> updated option: {$name_option_key} => {$name}"); |
| 920 | 810 | } |
| 921 | 811 | |
| 922 | 812 | // 2) (Optional) Also store in DB if a row already exists |
| @@ -960,10 +850,8 @@ | ||
| 960 | 850 | |
| 961 | 851 | public function mxchat_check_email_provided() { |
| 962 | 852 | //error_log('[DEBUG] ---------- mxchat_check_email_provided START ----------'); |
| 963 | 853 | |
| 964 | - nocache_headers(); | |
| 965 | - | |
| 966 | 854 | if (!isset($_POST['nonce']) || !wp_verify_nonce($_POST['nonce'], 'mxchat_chat_nonce')) { |
| 967 | 855 | //error_log('[ERROR] Invalid nonce in mxchat_check_email_provided'); |
| 968 | 856 | wp_send_json_error(['message' => esc_html__('Invalid nonce', 'mxchat')]); |
| 969 | 857 | } |
| @@ -968,9 +856,9 @@ | ||
| 968 | 856 | wp_send_json_error(['message' => esc_html__('Invalid nonce', 'mxchat')]); |
| 969 | 857 | } |
| 970 | 858 | |
| 971 | 859 | $session_id = isset($_POST['session_id']) ? sanitize_text_field($_POST['session_id']) : ''; |
| 972 | - if (empty($session_id) || $session_id === 'null') { | |
| 860 | + if (empty($session_id)) { | |
| 973 | 861 | //error_log('[ERROR] No session ID provided in mxchat_check_email_provided'); |
| 974 | 862 | wp_send_json_error(['message' => esc_html__('No session ID provided', 'mxchat')]); |
| 975 | 863 | } |
| 976 | 864 | |
| @@ -1059,10 +947,10 @@ | ||
| 1059 | 947 | global $wpdb; |
| 1060 | 948 | |
| 1061 | 949 | // Debug: Log incoming bot_id |
| 1062 | 950 | $bot_id = isset($_POST['bot_id']) ? sanitize_key($_POST['bot_id']) : 'default'; |
| 1063 | - //error_log("=== MXCHAT DEBUG: Starting chat request ==="); | |
| 1064 | - //error_log("MXCHAT DEBUG: Bot ID received: " . $bot_id); | |
| 951 | + error_log("=== MXCHAT DEBUG: Starting chat request ==="); | |
| 952 | + error_log("MXCHAT DEBUG: Bot ID received: " . $bot_id); | |
| 1065 | 953 | |
| 1066 | 954 | // Get bot-specific options |
| 1067 | 955 | $bot_options = $this->get_bot_options($bot_id); |
| 1068 | 956 | $current_options = !empty($bot_options) ? $bot_options : $this->options; |
| @@ -1067,18 +955,31 @@ | ||
| 1067 | 955 | $bot_options = $this->get_bot_options($bot_id); |
| 1068 | 956 | $current_options = !empty($bot_options) ? $bot_options : $this->options; |
| 1069 | 957 | |
| 1070 | 958 | // Check if this is a streaming request |
| 1071 | - // Allow force_streaming_test parameter to bypass the setting check (for admin compatibility testing) | |
| 1072 | - $force_streaming_test = isset($_POST['force_streaming_test']) && $_POST['force_streaming_test'] === '1' && current_user_can('administrator'); | |
| 1073 | 959 | $is_streaming = isset($_POST['action']) && $_POST['action'] === 'mxchat_stream_chat' && |
| 1074 | - ($force_streaming_test || (isset($current_options['enable_streaming_toggle']) && $current_options['enable_streaming_toggle'] === 'on')); | |
| 960 | + isset($current_options['enable_streaming_toggle']) && $current_options['enable_streaming_toggle'] === 'on'; | |
| 1075 | 961 | |
| 1076 | 962 | // ADDED: Store streaming state in class property for use in private methods |
| 1077 | 963 | $this->is_streaming = $is_streaming; |
| 1078 | 964 | |
| 1079 | - // NOTE: Streaming headers are now set later via setup_streaming_headers() | |
| 1080 | - // This allows actions/forms to return JSON responses without header conflicts | |
| 965 | + // Set streaming headers if needed | |
| 966 | + if ($is_streaming) { | |
| 967 | + // Disable output buffering | |
| 968 | + while (ob_get_level()) { | |
| 969 | + ob_end_flush(); // Changed from ob_end_clean() | |
| 970 | + } | |
| 971 | + | |
| 972 | + // Set headers for SSE | |
| 973 | + header('Content-Type: text/event-stream'); | |
| 974 | + header('Cache-Control: no-cache'); | |
| 975 | + header('Connection: keep-alive'); | |
| 976 | + header('X-Accel-Buffering: no'); | |
| 977 | + | |
| 978 | + // Add these new lines: | |
| 979 | + ob_implicit_flush(true); | |
| 980 | + flush(); | |
| 981 | + } | |
| 1081 | 982 | |
| 1082 | 983 | // Check if MX Chat Moderation is active |
| 1083 | 984 | if (class_exists('MX_Chat_Moderation')) { |
| 1084 | 985 | // Get user email and IP |
| @@ -1144,29 +1045,20 @@ | ||
| 1144 | 1045 | |
| 1145 | 1046 | // Rest of your existing code... |
| 1146 | 1047 | $session_id = isset($_POST['session_id']) ? sanitize_text_field($_POST['session_id']) : ''; |
| 1147 | 1048 | |
| 1148 | - // Treat the literal strings 'null' / 'undefined' as missing too. Browser edge cases | |
| 1149 | - // (Safari ITP, private mode, cross-origin iframes with partitioned storage) can cause | |
| 1150 | - // the frontend FormData.append() to stringify a null session_id into the literal | |
| 1151 | - // "null", which would otherwise pass empty() and pollute the transcripts table with | |
| 1152 | - // ghost sessions that group every visitor's first message under one row. | |
| 1153 | - if ($session_id === 'null' || $session_id === 'undefined') { | |
| 1154 | - $session_id = ''; | |
| 1155 | - } | |
| 1156 | - | |
| 1157 | 1049 | if (empty($session_id)) { |
| 1158 | 1050 | wp_send_json_error(esc_html__('Session ID is missing.', 'mxchat')); |
| 1159 | 1051 | wp_die(); |
| 1160 | 1052 | } |
| 1161 | 1053 | |
| 1162 | - // Update session owner if it changed (e.g. IP changed due to network switch) | |
| 1163 | - // The session ID itself is the authentication — if the client has it, they own it | |
| 1054 | + // SECURITY FIX: Verify session ownership before processing chat request | |
| 1164 | 1055 | $current_user_identifier = MxChat_User::mxchat_get_user_identifier(); |
| 1165 | 1056 | $session_owner = get_option("mxchat_session_owner_{$session_id}"); |
| 1166 | 1057 | |
| 1167 | - if (!$session_owner || $session_owner !== $current_user_identifier) { | |
| 1168 | - update_option("mxchat_session_owner_{$session_id}", $current_user_identifier, 'no'); | |
| 1058 | + if ($session_owner && $session_owner !== $current_user_identifier) { | |
| 1059 | + wp_send_json_error(esc_html__('Unauthorized access.', 'mxchat')); | |
| 1060 | + wp_die(); | |
| 1169 | 1061 | } |
| 1170 | 1062 | |
| 1171 | 1063 | // Validate and sanitize the incoming message |
| 1172 | 1064 | if (empty($_POST['message'])) { |
| @@ -1573,37 +1465,54 @@ | ||
| 1573 | 1465 | |
| 1574 | 1466 | if ($testing_data !== null) { |
| 1575 | 1467 | $response_data['testing_data'] = $testing_data; |
| 1576 | 1468 | } |
| 1577 | - | |
| 1469 | + | |
| 1470 | + // Clear streaming headers if they were set | |
| 1471 | + if ($is_streaming) { | |
| 1472 | + header_remove('Content-Type'); | |
| 1473 | + header_remove('Cache-Control'); | |
| 1474 | + header_remove('Connection'); | |
| 1475 | + header_remove('X-Accel-Buffering'); | |
| 1476 | + header('Content-Type: application/json'); | |
| 1477 | + } | |
| 1478 | + | |
| 1578 | 1479 | wp_send_json($response_data); |
| 1579 | 1480 | wp_die(); |
| 1580 | 1481 | } else if ($intent_result === true && (!empty($this->fallbackResponse['text']) || !empty($this->fallbackResponse['html']))) { |
| 1581 | 1482 | // Intent returned true and set fallbackResponse |
| 1582 | - | |
| 1583 | - // SAVE TO TRANSCRIPT | |
| 1483 | + | |
| 1484 | + // SAVE TO TRANSCRIPT FIRST | |
| 1584 | 1485 | if (!empty($this->fallbackResponse['text'])) { |
| 1585 | 1486 | $this->mxchat_save_chat_message($session_id, 'bot', $this->fallbackResponse['text']); |
| 1586 | 1487 | } |
| 1587 | - // Save action HTML (product cards, featured products, etc.) so it renders in transcripts | |
| 1588 | 1488 | if (!empty($this->fallbackResponse['html'])) { |
| 1589 | 1489 | $this->mxchat_save_chat_message($session_id, 'bot', $this->fallbackResponse['html']); |
| 1590 | 1490 | } |
| 1591 | - | |
| 1491 | + | |
| 1592 | 1492 | $response_data = [ |
| 1593 | 1493 | 'text' => $this->fallbackResponse['text'] ?? '', |
| 1594 | 1494 | 'html' => $this->fallbackResponse['html'] ?? '', |
| 1595 | 1495 | 'session_id' => $session_id |
| 1596 | 1496 | ]; |
| 1597 | - | |
| 1497 | + | |
| 1598 | 1498 | if (isset($this->fallbackResponse['chat_mode'])) { |
| 1599 | 1499 | $response_data['chat_mode'] = $this->fallbackResponse['chat_mode']; |
| 1600 | 1500 | } |
| 1601 | - | |
| 1501 | + | |
| 1602 | 1502 | if ($testing_data !== null) { |
| 1603 | 1503 | $response_data['testing_data'] = $testing_data; |
| 1604 | 1504 | } |
| 1605 | - | |
| 1505 | + | |
| 1506 | + // Clear streaming headers if they were set | |
| 1507 | + if ($is_streaming) { | |
| 1508 | + header_remove('Content-Type'); | |
| 1509 | + header_remove('Cache-Control'); | |
| 1510 | + header_remove('Connection'); | |
| 1511 | + header_remove('X-Accel-Buffering'); | |
| 1512 | + header('Content-Type: application/json'); | |
| 1513 | + } | |
| 1514 | + | |
| 1606 | 1515 | wp_send_json($response_data); |
| 1607 | 1516 | wp_die(); |
| 1608 | 1517 | } |
| 1609 | 1518 | } |
| @@ -1608,13 +1517,11 @@ | ||
| 1608 | 1517 | } |
| 1609 | 1518 | } |
| 1610 | 1519 | |
| 1611 | 1520 | // If we get here, no intent matched OR the intent didn't provide a usable response |
| 1612 | - | |
| 1521 | + | |
| 1613 | 1522 | // Step 4: Generate AI response |
| 1614 | - // Get session start timestamp - when persistence is OFF, only include messages from this page load | |
| 1615 | - $session_start_timestamp = isset($_POST['session_start_timestamp']) ? intval($_POST['session_start_timestamp']) : 0; | |
| 1616 | - $conversation_history = $this->mxchat_fetch_conversation_history_for_ai($session_id, $session_start_timestamp); | |
| 1523 | + $conversation_history = $this->mxchat_fetch_conversation_history_for_ai($session_id); | |
| 1617 | 1524 | $this->mxchat_increment_chat_count(); |
| 1618 | 1525 | |
| 1619 | 1526 | // Generate embedding for the user's query - USE BOT-SPECIFIC API KEY |
| 1620 | 1527 | $api_key = $current_options['api_key'] ?? $this->options['api_key']; |
| @@ -1692,34 +1599,30 @@ | ||
| 1692 | 1599 | $context_content .= "Page Content: " . $page_context['content'] . "\n"; |
| 1693 | 1600 | $context_content .= "===== END CURRENT PAGE CONTEXT =====\n\n"; |
| 1694 | 1601 | } |
| 1695 | 1602 | |
| 1696 | - // Get relevant content from knowledge base - PASS BOT_ID and MESSAGE for Vector Store | |
| 1697 | - $relevant_content = $this->mxchat_find_relevant_content($user_message_embedding, $bot_id, $message); | |
| 1603 | + // Get relevant content from knowledge base - PASS BOT_ID | |
| 1604 | + $relevant_content = $this->mxchat_find_relevant_content($user_message_embedding, $bot_id); | |
| 1698 | 1605 | |
| 1699 | - // NEW: Also extract URLs from system instructions (only if citation links enabled) | |
| 1700 | - // Use fresh options to ensure we get the latest setting value | |
| 1701 | - $fresh_options = get_option('mxchat_options', []); | |
| 1702 | - $citation_links_enabled = isset($fresh_options['citation_links_toggle']) ? ($fresh_options['citation_links_toggle'] === 'on') : true; | |
| 1703 | - | |
| 1704 | - $system_instructions = $this->get_system_instructions($bot_id, $session_id); | |
| 1705 | - if ($citation_links_enabled && !empty($system_instructions)) { | |
| 1606 | + // NEW: Also extract URLs from system instructions | |
| 1607 | + $system_instructions = $this->get_system_instructions($bot_id); | |
| 1608 | + if (!empty($system_instructions)) { | |
| 1706 | 1609 | preg_match_all( |
| 1707 | - '#\bhttps?://[^\s<>"\']+#i', | |
| 1708 | - $system_instructions, | |
| 1610 | + '#\bhttps?://[^\s<>"\']+#i', | |
| 1611 | + $system_instructions, | |
| 1709 | 1612 | $system_instruction_urls |
| 1710 | 1613 | ); |
| 1711 | - | |
| 1614 | + | |
| 1712 | 1615 | if (!empty($system_instruction_urls[0])) { |
| 1713 | 1616 | // Merge with existing valid URLs |
| 1714 | 1617 | $this->current_valid_urls = array_merge( |
| 1715 | - $this->current_valid_urls, | |
| 1618 | + $this->current_valid_urls, | |
| 1716 | 1619 | $system_instruction_urls[0] |
| 1717 | 1620 | ); |
| 1718 | 1621 | // Remove duplicates |
| 1719 | 1622 | $this->current_valid_urls = array_unique($this->current_valid_urls); |
| 1720 | - | |
| 1721 | - //error_log("Added " . count($system_instruction_urls[0]) . " URLs from system instructions"); | |
| 1623 | + | |
| 1624 | + error_log("Added " . count($system_instruction_urls[0]) . " URLs from system instructions"); | |
| 1722 | 1625 | } |
| 1723 | 1626 | } |
| 1724 | 1627 | |
| 1725 | 1628 | // ===== CAPTURE REAL SIMILARITY DATA FOR ADMINS ===== |
| @@ -1727,10 +1630,8 @@ | ||
| 1727 | 1630 | // Update testing data with the REAL similarity analysis |
| 1728 | 1631 | $testing_data['top_matches'] = $this->last_similarity_analysis['top_matches']; |
| 1729 | 1632 | $testing_data['total_documents_checked'] = $this->last_similarity_analysis['total_checked'] ?? 0; |
| 1730 | 1633 | $testing_data['knowledge_base_type'] = $this->last_similarity_analysis['knowledge_base_type']; |
| 1731 | - $testing_data['sources_used'] = $this->last_similarity_analysis['sources_used'] ?? 0; | |
| 1732 | - $testing_data['total_chunks_used'] = $this->last_similarity_analysis['total_chunks_used'] ?? 0; | |
| 1733 | 1634 | } |
| 1734 | 1635 | // ===== END SIMILARITY DATA CAPTURE ===== |
| 1735 | 1636 | |
| 1736 | 1637 | // NEW: Add valid URLs to testing data for admin panel display (AFTER similarity data) |
| @@ -1735,9 +1636,9 @@ | ||
| 1735 | 1636 | |
| 1736 | 1637 | // NEW: Add valid URLs to testing data for admin panel display (AFTER similarity data) |
| 1737 | 1638 | if ($testing_data !== null && !empty($this->current_valid_urls)) { |
| 1738 | 1639 | $testing_data['approved_urls'] = array_values($this->current_valid_urls); |
| 1739 | - //error_log("Added " . count($this->current_valid_urls) . " approved URLs to testing data"); | |
| 1640 | + error_log("Added " . count($this->current_valid_urls) . " approved URLs to testing data"); | |
| 1740 | 1641 | } |
| 1741 | 1642 | |
| 1742 | 1643 | if (!empty($relevant_content)) { |
| 1743 | 1644 | $context_content .= "===== OFFICIAL KNOWLEDGE DATABASE CONTENT =====\n" . $relevant_content . "\n===== END OF OFFICIAL KNOWLEDGE DATABASE CONTENT =====\n\n"; |
| @@ -1744,10 +1645,10 @@ | ||
| 1744 | 1645 | } else { |
| 1745 | 1646 | $context_content .= "===== NO RELEVANT CONTENT FOUND IN KNOWLEDGE DATABASE =====\n"; |
| 1746 | 1647 | } |
| 1747 | 1648 | |
| 1748 | - // NEW: Add approved URLs list to context for AI (only if citation links enabled) | |
| 1749 | - if ($citation_links_enabled && !empty($this->current_valid_urls)) { | |
| 1649 | + // NEW: Add approved URLs list to context for AI | |
| 1650 | + if (!empty($this->current_valid_urls)) { | |
| 1750 | 1651 | $context_content .= "===== APPROVED URLS FOR CITATIONS =====\n"; |
| 1751 | 1652 | $context_content .= "You may ONLY use these exact URLs in your response:\n"; |
| 1752 | 1653 | foreach ($this->current_valid_urls as $url) { |
| 1753 | 1654 | $context_content .= "- " . $url . "\n"; |
| @@ -1788,9 +1689,9 @@ | ||
| 1788 | 1689 | |
| 1789 | 1690 | $context_content = apply_filters('mxchat_prepare_context', $context_content, $session_id); |
| 1790 | 1691 | |
| 1791 | 1692 | // Extract model from current options for bot-specific model support |
| 1792 | - $selected_model = isset($current_options['model']) ? $current_options['model'] : 'gpt-5.1-chat-latest'; | |
| 1693 | + $selected_model = isset($current_options['model']) ? $current_options['model'] : 'gpt-4o'; | |
| 1793 | 1694 | |
| 1794 | 1695 | $response = $this->mxchat_generate_response( |
| 1795 | 1696 | $context_content, |
| 1796 | 1697 | $current_options['api_key'] ?? $this->options['api_key'], |
| @@ -1840,50 +1741,25 @@ | ||
| 1840 | 1741 | wp_die(); |
| 1841 | 1742 | } |
| 1842 | 1743 | |
| 1843 | 1744 | // DEBUG: Check what we have |
| 1844 | - //error_log("=== BEFORE URL VALIDATION ==="); | |
| 1845 | - //error_log("current_valid_urls is empty? " . (empty($this->current_valid_urls) ? 'YES' : 'NO')); | |
| 1846 | - //error_log("current_valid_urls count: " . count($this->current_valid_urls)); | |
| 1847 | - //error_log("current_valid_urls content: " . print_r($this->current_valid_urls, true)); | |
| 1745 | + error_log("=== BEFORE URL VALIDATION ==="); | |
| 1746 | + error_log("current_valid_urls is empty? " . (empty($this->current_valid_urls) ? 'YES' : 'NO')); | |
| 1747 | + error_log("current_valid_urls count: " . count($this->current_valid_urls)); | |
| 1748 | + error_log("current_valid_urls content: " . print_r($this->current_valid_urls, true)); | |
| 1848 | 1749 | |
| 1849 | 1750 | // If we get here, the response is valid text - now validate URLs |
| 1850 | 1751 | if (!empty($this->current_valid_urls)) { |
| 1851 | - //error_log("CALLING validate_and_clean_urls"); | |
| 1752 | + error_log("CALLING validate_and_clean_urls"); | |
| 1852 | 1753 | $response = $this->validate_and_clean_urls($response, $this->current_valid_urls); |
| 1853 | 1754 | } else { |
| 1854 | - //error_log("SKIPPING validation - current_valid_urls is empty"); | |
| 1755 | + error_log("SKIPPING validation - current_valid_urls is empty"); | |
| 1855 | 1756 | } |
| 1856 | 1757 | // ===== END URL VALIDATION ===== |
| 1758 | + | |
| 1759 | + // Save the cleaned response | |
| 1760 | + $this->mxchat_save_chat_message($session_id, 'bot', $response); | |
| 1857 | 1761 | |
| 1858 | - // Prepare RAG context data for storage (only include documents used for context) | |
| 1859 | - $rag_context_for_storage = null; | |
| 1860 | - $has_rag_data = $this->last_similarity_analysis !== null && !empty($this->last_similarity_analysis['top_matches']); | |
| 1861 | - $has_action_data = isset($this->last_action_analysis) && !empty($this->last_action_analysis); | |
| 1862 | - | |
| 1863 | - if ($has_rag_data || $has_action_data) { | |
| 1864 | - $rag_context_for_storage = []; | |
| 1865 | - | |
| 1866 | - // Add RAG/source data if available | |
| 1867 | - if ($has_rag_data) { | |
| 1868 | - $rag_context_for_storage['top_matches'] = $this->last_similarity_analysis['top_matches']; | |
| 1869 | - $rag_context_for_storage['approved_urls'] = $this->current_valid_urls ?? []; | |
| 1870 | - $rag_context_for_storage['similarity_threshold'] = $this->last_similarity_analysis['threshold_used'] ?? 0.35; | |
| 1871 | - $rag_context_for_storage['knowledge_base_type'] = $this->last_similarity_analysis['knowledge_base_type'] ?? 'WordPress Database'; | |
| 1872 | - $rag_context_for_storage['total_documents_checked'] = $this->last_similarity_analysis['total_checked'] ?? 0; | |
| 1873 | - $rag_context_for_storage['sources_used'] = $this->last_similarity_analysis['sources_used'] ?? 0; | |
| 1874 | - $rag_context_for_storage['total_chunks_used'] = $this->last_similarity_analysis['total_chunks_used'] ?? 0; | |
| 1875 | - } | |
| 1876 | - | |
| 1877 | - // Add action analysis data if available | |
| 1878 | - if ($has_action_data) { | |
| 1879 | - $rag_context_for_storage['action_analysis'] = $this->last_action_analysis; | |
| 1880 | - } | |
| 1881 | - } | |
| 1882 | - | |
| 1883 | - // Save the cleaned response with RAG context | |
| 1884 | - $this->mxchat_save_chat_message($session_id, 'bot', $response, null, $rag_context_for_storage); | |
| 1885 | - | |
| 1886 | 1762 | // Step 5: Save additional content if available |
| 1887 | 1763 | if (!empty($this->productCardHtml)) { |
| 1888 | 1764 | $this->mxchat_save_chat_message($session_id, 'bot', $this->productCardHtml); |
| 1889 | 1765 | } |
| @@ -1892,13 +1768,8 @@ | ||
| 1892 | 1768 | $this->mxchat_save_chat_message($session_id, 'bot', $this->fallbackResponse['html']); |
| 1893 | 1769 | } |
| 1894 | 1770 | |
| 1895 | 1771 | // Step 6: Return the response |
| 1896 | - // DEBUG: Check if newlines exist in the response | |
| 1897 | - //error_log("=== MXCHAT NON-STREAMING RESPONSE DEBUG ==="); | |
| 1898 | - //error_log("Response has newlines: " . (strpos($response, "\n") !== false ? 'YES' : 'NO')); | |
| 1899 | - //error_log("Response first 500 chars: " . substr($response, 0, 500)); | |
| 1900 | - | |
| 1901 | 1772 | $response_data = [ |
| 1902 | 1773 | 'text' => $response, |
| 1903 | 1774 | 'html' => !empty($this->productCardHtml) ? $this->productCardHtml : ($this->fallbackResponse['html'] ?? ''), |
| 1904 | 1775 | 'session_id' => $session_id |
| @@ -1903,18 +1774,8 @@ | ||
| 1903 | 1774 | 'html' => !empty($this->productCardHtml) ? $this->productCardHtml : ($this->fallbackResponse['html'] ?? ''), |
| 1904 | 1775 | 'session_id' => $session_id |
| 1905 | 1776 | ]; |
| 1906 | 1777 | |
| 1907 | - // Include vectorstore error info for admin debugging (only visible to admins via testing_data) | |
| 1908 | - if (!empty($this->last_vectorstore_error) && $testing_data !== null) { | |
| 1909 | - $testing_data['vectorstore_error'] = $this->last_vectorstore_error; | |
| 1910 | - } | |
| 1911 | - | |
| 1912 | - // Also pass it as a top-level field so JS can show a better error message to admins | |
| 1913 | - if (!empty($this->last_vectorstore_error) && current_user_can('manage_options')) { | |
| 1914 | - $response_data['vectorstore_error'] = $this->last_vectorstore_error; | |
| 1915 | - } | |
| 1916 | - | |
| 1917 | 1778 | // Always add testing data for admins (no toggle needed) |
| 1918 | 1779 | if ($testing_data !== null) { |
| 1919 | 1780 | $response_data['testing_data'] = $testing_data; |
| 1920 | 1781 | } |
| @@ -1928,12 +1789,12 @@ | ||
| 1928 | 1789 | * Falls back to default options if bot_id is 'default' or multi-bot add-on is not active |
| 1929 | 1790 | */ |
| 1930 | 1791 | // Also debug the bot options retrieval |
| 1931 | 1792 | private function get_bot_options($bot_id = 'default') { |
| 1932 | - //error_log("MXCHAT DEBUG: get_bot_options called for bot: " . $bot_id); | |
| 1793 | + error_log("MXCHAT DEBUG: get_bot_options called for bot: " . $bot_id); | |
| 1933 | 1794 | |
| 1934 | 1795 | if ($bot_id === 'default' || !class_exists('MxChat_Multi_Bot_Manager')) { |
| 1935 | - //error_log("MXCHAT DEBUG: Using default options (no multi-bot or bot is 'default')"); | |
| 1796 | + error_log("MXCHAT DEBUG: Using default options (no multi-bot or bot is 'default')"); | |
| 1936 | 1797 | return array(); |
| 1937 | 1798 | } |
| 1938 | 1799 | |
| 1939 | 1800 | $bot_options = apply_filters('mxchat_get_bot_options', array(), $bot_id); |
| @@ -1938,11 +1799,11 @@ | ||
| 1938 | 1799 | |
| 1939 | 1800 | $bot_options = apply_filters('mxchat_get_bot_options', array(), $bot_id); |
| 1940 | 1801 | |
| 1941 | 1802 | if (!empty($bot_options)) { |
| 1942 | - //error_log("MXCHAT DEBUG: Got bot-specific options from filter"); | |
| 1803 | + error_log("MXCHAT DEBUG: Got bot-specific options from filter"); | |
| 1943 | 1804 | if (isset($bot_options['similarity_threshold'])) { |
| 1944 | - //error_log(" - similarity_threshold: " . $bot_options['similarity_threshold']); | |
| 1805 | + error_log(" - similarity_threshold: " . $bot_options['similarity_threshold']); | |
| 1945 | 1806 | } |
| 1946 | 1807 | } |
| 1947 | 1808 | |
| 1948 | 1809 | return is_array($bot_options) ? $bot_options : array(); |
| @@ -1953,13 +1814,13 @@ | ||
| 1953 | 1814 | * Used in the knowledge retrieval functions |
| 1954 | 1815 | */ |
| 1955 | 1816 | // Also add debugging to your get_bot_pinecone_config function |
| 1956 | 1817 | private function get_bot_pinecone_config($bot_id = 'default') { |
| 1957 | - //error_log("MXCHAT DEBUG: get_bot_pinecone_config called for bot: " . $bot_id); | |
| 1818 | + error_log("MXCHAT DEBUG: get_bot_pinecone_config called for bot: " . $bot_id); | |
| 1958 | 1819 | |
| 1959 | 1820 | // If default bot or multi-bot add-on not active, use default Pinecone config |
| 1960 | 1821 | if ($bot_id === 'default' || !class_exists('MxChat_Multi_Bot_Manager')) { |
| 1961 | - //error_log("MXCHAT DEBUG: Using default Pinecone config (no multi-bot or bot is 'default')"); | |
| 1822 | + error_log("MXCHAT DEBUG: Using default Pinecone config (no multi-bot or bot is 'default')"); | |
| 1962 | 1823 | $addon_options = get_option('mxchat_pinecone_addon_options', array()); |
| 1963 | 1824 | $config = array( |
| 1964 | 1825 | 'use_pinecone' => (isset($addon_options['mxchat_use_pinecone']) && $addon_options['mxchat_use_pinecone'] === '1'), |
| 1965 | 1826 | 'api_key' => $addon_options['mxchat_pinecone_api_key'] ?? '', |
| @@ -1965,24 +1826,24 @@ | ||
| 1965 | 1826 | 'api_key' => $addon_options['mxchat_pinecone_api_key'] ?? '', |
| 1966 | 1827 | 'host' => $addon_options['mxchat_pinecone_host'] ?? '', |
| 1967 | 1828 | 'namespace' => $addon_options['mxchat_pinecone_namespace'] ?? '' |
| 1968 | 1829 | ); |
| 1969 | - //error_log("MXCHAT DEBUG: Default config - use_pinecone: " . ($config['use_pinecone'] ? 'true' : 'false')); | |
| 1830 | + error_log("MXCHAT DEBUG: Default config - use_pinecone: " . ($config['use_pinecone'] ? 'true' : 'false')); | |
| 1970 | 1831 | return $config; |
| 1971 | 1832 | } |
| 1972 | 1833 | |
| 1973 | - //error_log("MXCHAT DEBUG: Calling filter 'mxchat_get_bot_pinecone_config' for bot: " . $bot_id); | |
| 1834 | + error_log("MXCHAT DEBUG: Calling filter 'mxchat_get_bot_pinecone_config' for bot: " . $bot_id); | |
| 1974 | 1835 | |
| 1975 | 1836 | // Hook for multi-bot add-on to provide bot-specific Pinecone config |
| 1976 | 1837 | $bot_pinecone_config = apply_filters('mxchat_get_bot_pinecone_config', array(), $bot_id); |
| 1977 | 1838 | |
| 1978 | 1839 | if (!empty($bot_pinecone_config)) { |
| 1979 | - //error_log("MXCHAT DEBUG: Got bot-specific config from filter"); | |
| 1980 | - //error_log(" - use_pinecone: " . (isset($bot_pinecone_config['use_pinecone']) ? ($bot_pinecone_config['use_pinecone'] ? 'true' : 'false') : 'not set')); | |
| 1981 | - //error_log(" - host: " . ($bot_pinecone_config['host'] ?? 'not set')); | |
| 1982 | - //error_log(" - namespace: " . ($bot_pinecone_config['namespace'] ?? 'not set')); | |
| 1840 | + error_log("MXCHAT DEBUG: Got bot-specific config from filter"); | |
| 1841 | + error_log(" - use_pinecone: " . (isset($bot_pinecone_config['use_pinecone']) ? ($bot_pinecone_config['use_pinecone'] ? 'true' : 'false') : 'not set')); | |
| 1842 | + error_log(" - host: " . ($bot_pinecone_config['host'] ?? 'not set')); | |
| 1843 | + error_log(" - namespace: " . ($bot_pinecone_config['namespace'] ?? 'not set')); | |
| 1983 | 1844 | } else { |
| 1984 | - //error_log("MXCHAT DEBUG: Filter returned empty config!"); | |
| 1845 | + error_log("MXCHAT DEBUG: Filter returned empty config!"); | |
| 1985 | 1846 | } |
| 1986 | 1847 | |
| 1987 | 1848 | return is_array($bot_pinecone_config) ? $bot_pinecone_config : array(); |
| 1988 | 1849 | } |
| @@ -2058,29 +1919,19 @@ | ||
| 2058 | 1919 | $intents = $wpdb->get_results($query); |
| 2059 | 1920 | } else { |
| 2060 | 1921 | $intents = $wpdb->get_results("SELECT * FROM $table_name WHERE enabled = 1 OR enabled IS NULL"); |
| 2061 | 1922 | } |
| 2062 | - | |
| 1923 | + | |
| 2063 | 1924 | if (empty($intents)) { |
| 2064 | 1925 | return false; |
| 2065 | 1926 | } |
| 2066 | - | |
| 2067 | - // Prefetch individual phrase vectors from wp_mxchat_intent_phrases (grouped by intent_id) | |
| 2068 | - $phrases_table = $wpdb->prefix . 'mxchat_intent_phrases'; | |
| 2069 | - $phrases_by_intent = []; | |
| 2070 | - if ($wpdb->get_var("SHOW TABLES LIKE '$phrases_table'") === $phrases_table) { | |
| 2071 | - $all_phrases = $wpdb->get_results("SELECT intent_id, phrase, embedding_vector FROM $phrases_table"); | |
| 2072 | - foreach ($all_phrases as $p) { | |
| 2073 | - $phrases_by_intent[$p->intent_id][] = $p; | |
| 2074 | - } | |
| 2075 | - } | |
| 2076 | - | |
| 1927 | + | |
| 2077 | 1928 | $highest_similarity = -INF; |
| 2078 | 1929 | $matched_intent = null; |
| 2079 | - | |
| 1930 | + | |
| 2080 | 1931 | // Array to store action analysis for testing panel |
| 2081 | 1932 | $action_analysis = []; |
| 2082 | - | |
| 1933 | + | |
| 2083 | 1934 | foreach ($intents as $intent) { |
| 2084 | 1935 | // Additional check for enabled state |
| 2085 | 1936 | $is_enabled = isset($intent->enabled) ? (bool)$intent->enabled : true; |
| 2086 | 1937 | if (!$is_enabled) { |
| @@ -2085,56 +1936,26 @@ | ||
| 2085 | 1936 | $is_enabled = isset($intent->enabled) ? (bool)$intent->enabled : true; |
| 2086 | 1937 | if (!$is_enabled) { |
| 2087 | 1938 | continue; |
| 2088 | 1939 | } |
| 2089 | - | |
| 1940 | + | |
| 2090 | 1941 | // Check if this action is enabled for the current bot |
| 2091 | 1942 | if (!$this->is_action_enabled_for_bot($intent, $current_bot_id)) { |
| 2092 | 1943 | continue; |
| 2093 | 1944 | } |
| 2094 | - | |
| 2095 | - $best_similarity = -INF; | |
| 2096 | - $matched_phrase_text = ''; | |
| 2097 | - | |
| 2098 | - // Check legacy embedding vector (existing behavior) | |
| 1945 | + | |
| 2099 | 1946 | $intent_embedding_serialized = $intent->embedding_vector; |
| 2100 | 1947 | $intent_embedding = $intent_embedding_serialized |
| 2101 | 1948 | ? unserialize($intent_embedding_serialized, ['allowed_classes' => false]) |
| 2102 | 1949 | : null; |
| 2103 | - | |
| 2104 | - if (is_array($intent_embedding) && !empty($intent_embedding)) { | |
| 2105 | - $legacy_similarity = $this->mxchat_calculate_cosine_similarity($user_embedding, $intent_embedding); | |
| 2106 | - if ($legacy_similarity > $best_similarity) { | |
| 2107 | - $best_similarity = $legacy_similarity; | |
| 2108 | - $matched_phrase_text = 'legacy'; | |
| 2109 | - } | |
| 2110 | - } | |
| 2111 | - | |
| 2112 | - // Check individual phrase vectors | |
| 2113 | - if (isset($phrases_by_intent[$intent->id])) { | |
| 2114 | - foreach ($phrases_by_intent[$intent->id] as $phrase_row) { | |
| 2115 | - $phrase_embedding = $phrase_row->embedding_vector | |
| 2116 | - ? unserialize($phrase_row->embedding_vector, ['allowed_classes' => false]) | |
| 2117 | - : null; | |
| 2118 | - if (!is_array($phrase_embedding)) { | |
| 2119 | - continue; | |
| 2120 | - } | |
| 2121 | - $phrase_similarity = $this->mxchat_calculate_cosine_similarity($user_embedding, $phrase_embedding); | |
| 2122 | - if ($phrase_similarity > $best_similarity) { | |
| 2123 | - $best_similarity = $phrase_similarity; | |
| 2124 | - $matched_phrase_text = $phrase_row->phrase; | |
| 2125 | - } | |
| 2126 | - } | |
| 2127 | - } | |
| 2128 | - | |
| 2129 | - // Skip if no valid embedding was found at all | |
| 2130 | - if ($best_similarity === -INF) { | |
| 1950 | + | |
| 1951 | + if (!is_array($intent_embedding)) { | |
| 2131 | 1952 | continue; |
| 2132 | 1953 | } |
| 2133 | - | |
| 2134 | - $similarity = $best_similarity; | |
| 1954 | + | |
| 1955 | + $similarity = $this->mxchat_calculate_cosine_similarity($user_embedding, $intent_embedding); | |
| 2135 | 1956 | $intent_threshold = isset($intent->similarity_threshold) ? $intent->similarity_threshold : 0.85; |
| 2136 | - | |
| 1957 | + | |
| 2137 | 1958 | // Store action analysis data for testing panel |
| 2138 | 1959 | $action_analysis[] = [ |
| 2139 | 1960 | 'intent_label' => $intent->intent_label, |
| 2140 | 1961 | 'callback_function' => $intent->callback_function, |
| @@ -2142,12 +1963,11 @@ | ||
| 2142 | 1963 | 'similarity_percentage' => round($similarity * 100, 2), |
| 2143 | 1964 | 'threshold' => $intent_threshold, |
| 2144 | 1965 | 'threshold_percentage' => round($intent_threshold * 100, 2), |
| 2145 | 1966 | 'above_threshold' => $similarity >= $intent_threshold, |
| 2146 | - 'matched_phrase' => $matched_phrase_text, | |
| 2147 | 1967 | 'triggered' => false // Will be updated below if this intent is triggered |
| 2148 | 1968 | ]; |
| 2149 | - | |
| 1969 | + | |
| 2150 | 1970 | if ($similarity >= $intent_threshold && $similarity > $highest_similarity) { |
| 2151 | 1971 | $highest_similarity = $similarity; |
| 2152 | 1972 | $matched_intent = $intent; |
| 2153 | 1973 | } |
| @@ -2218,22 +2038,16 @@ | ||
| 2218 | 2038 | // If enabled_bots column doesn't exist or is null, assume it's enabled for all bots (backward compatibility) |
| 2219 | 2039 | if (!isset($intent->enabled_bots) || empty($intent->enabled_bots)) { |
| 2220 | 2040 | return true; |
| 2221 | 2041 | } |
| 2222 | - | |
| 2042 | + | |
| 2223 | 2043 | $enabled_bots = json_decode($intent->enabled_bots, true); |
| 2224 | - | |
| 2044 | + | |
| 2225 | 2045 | // If JSON decode fails or returns empty array, assume enabled for all (backward compatibility) |
| 2226 | 2046 | if (!is_array($enabled_bots) || empty($enabled_bots)) { |
| 2227 | 2047 | return true; |
| 2228 | 2048 | } |
| 2229 | - | |
| 2230 | - // Admin testing tab uses bot_id "testing" — treat it as "default" so all | |
| 2231 | - // default-bot actions are testable from the admin panel | |
| 2232 | - if ($bot_id === 'testing') { | |
| 2233 | - $bot_id = 'default'; | |
| 2234 | - } | |
| 2235 | - | |
| 2049 | + | |
| 2236 | 2050 | // Check if the current bot is in the enabled bots list |
| 2237 | 2051 | return in_array($bot_id, $enabled_bots); |
| 2238 | 2052 | } |
| 2239 | 2053 | |
| @@ -2272,17 +2086,17 @@ | ||
| 2272 | 2086 | |
| 2273 | 2087 | public function mxchat_generate_image($message, $user_id, $session_id) { |
| 2274 | 2088 | //error_log("Starting image generation for message: " . $message); |
| 2275 | 2089 | |
| 2276 | - // Prepare a prompt for OpenAI image generation | |
| 2090 | + // Prepare a prompt for DALL-E | |
| 2277 | 2091 | $prompt = esc_html__('Create an image of ', 'mxchat') . sanitize_text_field($message); |
| 2278 | - | |
| 2092 | + | |
| 2279 | 2093 | // Use the existing OpenAI API key |
| 2280 | 2094 | $openai_api_key = sanitize_text_field($this->options['api_key']); |
| 2281 | - | |
| 2282 | - // Call OpenAI GPT Image to generate an image | |
| 2283 | - $image_response = $this->mxchat_generate_openai_image($prompt, $openai_api_key); | |
| 2284 | 2095 | |
| 2096 | + // Call DALL-E to generate an image | |
| 2097 | + $image_response = $this->mxchat_generate_dalle_image($prompt, $openai_api_key); | |
| 2098 | + | |
| 2285 | 2099 | // Check if the response contains an image URL |
| 2286 | 2100 | if (isset($image_response['imageUrl'])) { |
| 2287 | 2101 | $image_url = esc_url_raw($image_response['imageUrl']); |
| 2288 | 2102 | |
| @@ -2325,103 +2139,24 @@ | ||
| 2325 | 2139 | // Return the response directly instead of relying on the property |
| 2326 | 2140 | return $this->fallbackResponse; |
| 2327 | 2141 | } |
| 2328 | 2142 | } |
| 2329 | - | |
| 2330 | -public function mxchat_generate_gemini_image($message, $user_id, $session_id) { | |
| 2331 | - $prompt = esc_html__('Create an image of ', 'mxchat') . sanitize_text_field($message); | |
| 2332 | - | |
| 2333 | - $gemini_api_key = sanitize_text_field($this->options['gemini_api_key'] ?? ''); | |
| 2334 | - if (empty($gemini_api_key)) { | |
| 2335 | - $response_text = esc_html__("Gemini API key is not configured.", 'mxchat'); | |
| 2336 | - $this->mxchat_save_chat_message($session_id, 'bot', $response_text); | |
| 2337 | - return ['text' => $response_text, 'html' => '', 'images' => []]; | |
| 2338 | - } | |
| 2339 | - | |
| 2340 | - $image_response = $this->mxchat_generate_imagen_image($prompt, $gemini_api_key); | |
| 2341 | - | |
| 2342 | - if (isset($image_response['imageUrl'])) { | |
| 2343 | - $image_url = esc_url_raw($image_response['imageUrl']); | |
| 2344 | - | |
| 2345 | - $response_html = '<img src="' . esc_url($image_url) . '" alt="' . esc_attr__('Generated Image', 'mxchat') . '" class="mxchat-generated-image" />'; | |
| 2346 | - $response_text = esc_html__('Here is the image I generated:', 'mxchat'); | |
| 2347 | - | |
| 2348 | - $this->mxchat_save_chat_message($session_id, 'bot', $response_text); | |
| 2349 | - $this->mxchat_save_chat_message($session_id, 'bot', $response_html); | |
| 2350 | - | |
| 2351 | - $this->fallbackResponse = [ | |
| 2352 | - 'text' => $response_text, | |
| 2353 | - 'html' => $response_html, | |
| 2354 | - 'images' => [$image_url] | |
| 2355 | - ]; | |
| 2356 | - | |
| 2357 | - return $this->fallbackResponse; | |
| 2358 | - } else { | |
| 2359 | - $response_text = esc_html__("I'm sorry, but I couldn't generate an image based on your request.", 'mxchat'); | |
| 2360 | - | |
| 2361 | - $this->mxchat_save_chat_message($session_id, 'bot', $response_text); | |
| 2362 | - | |
| 2363 | - $this->fallbackResponse = [ | |
| 2364 | - 'text' => $response_text, | |
| 2365 | - 'html' => '', | |
| 2366 | - 'images' => [] | |
| 2367 | - ]; | |
| 2368 | - | |
| 2369 | - return $this->fallbackResponse; | |
| 2370 | - } | |
| 2371 | -} | |
| 2372 | - | |
| 2373 | -private function mxchat_save_generated_image($base64_data, $mime_type = 'image/png', $prefix = 'mxchat-generated') { | |
| 2374 | - $extension = ($mime_type === 'image/jpeg') ? 'jpg' : 'png'; | |
| 2375 | - $filename = sanitize_file_name($prefix . '-' . wp_generate_uuid4() . '.' . $extension); | |
| 2376 | - $decoded = base64_decode($base64_data); | |
| 2377 | - | |
| 2378 | - if ($decoded === false) { | |
| 2379 | - return new \WP_Error('decode_failed', esc_html__('Failed to decode image data.', 'mxchat')); | |
| 2380 | - } | |
| 2381 | - | |
| 2382 | - $upload = wp_upload_bits($filename, null, $decoded); | |
| 2383 | - | |
| 2384 | - if (!empty($upload['error'])) { | |
| 2385 | - return new \WP_Error('upload_failed', $upload['error']); | |
| 2386 | - } | |
| 2387 | - | |
| 2388 | - $attach_id = wp_insert_attachment([ | |
| 2389 | - 'post_mime_type' => $mime_type, | |
| 2390 | - 'post_title' => $prefix, | |
| 2391 | - 'post_content' => '', | |
| 2392 | - 'post_status' => 'inherit', | |
| 2393 | - ], $upload['file']); | |
| 2394 | - | |
| 2395 | - if (is_wp_error($attach_id)) { | |
| 2396 | - return $attach_id; | |
| 2397 | - } | |
| 2398 | - | |
| 2399 | - require_once ABSPATH . 'wp-admin/includes/image.php'; | |
| 2400 | - $metadata = wp_generate_attachment_metadata($attach_id, $upload['file']); | |
| 2401 | - wp_update_attachment_metadata($attach_id, $metadata); | |
| 2402 | - | |
| 2403 | - return esc_url_raw(wp_get_attachment_url($attach_id)); | |
| 2404 | -} | |
| 2405 | - | |
| 2406 | -private function mxchat_generate_openai_image($prompt, $api_key, $model = 'gpt-image-1', $timeout = 60) { | |
| 2143 | +private function mxchat_generate_dalle_image($prompt, $api_key, $model = 'dall-e-3', $timeout = 60) { | |
| 2407 | 2144 | $api_url = 'https://api.openai.com/v1/images/generations'; |
| 2408 | 2145 | $body = json_encode([ |
| 2409 | - 'prompt' => sanitize_text_field($prompt), | |
| 2410 | - 'n' => 1, | |
| 2411 | - 'size' => '1024x1024', | |
| 2412 | - 'quality' => 'medium', | |
| 2413 | - 'output_format' => 'png', | |
| 2414 | - 'model' => sanitize_text_field($model), | |
| 2146 | + 'prompt' => sanitize_text_field($prompt), | |
| 2147 | + 'n' => 1, | |
| 2148 | + 'size' => '1024x1024', | |
| 2149 | + 'model' => sanitize_text_field($model), | |
| 2415 | 2150 | ]); |
| 2416 | 2151 | |
| 2417 | 2152 | $args = [ |
| 2418 | - 'body' => $body, | |
| 2153 | + 'body' => $body, | |
| 2419 | 2154 | 'headers' => [ |
| 2420 | - 'Content-Type' => 'application/json', | |
| 2155 | + 'Content-Type' => 'application/json', | |
| 2421 | 2156 | 'Authorization' => 'Bearer ' . sanitize_text_field($api_key), |
| 2422 | 2157 | ], |
| 2423 | - 'method' => 'POST', | |
| 2158 | + 'method' => 'POST', | |
| 2424 | 2159 | 'timeout' => absint($timeout), |
| 2425 | 2160 | ]; |
| 2426 | 2161 | |
| 2427 | 2162 | $response = wp_remote_post($api_url, $args); |
| @@ -2426,67 +2161,22 @@ | ||
| 2426 | 2161 | |
| 2427 | 2162 | $response = wp_remote_post($api_url, $args); |
| 2428 | 2163 | |
| 2429 | 2164 | if (is_wp_error($response)) { |
| 2165 | + //error_log("DALL-E request failed: " . $response->get_error_message()); | |
| 2430 | 2166 | return ['error' => esc_html__('Error generating image: ', 'mxchat') . $response->get_error_message()]; |
| 2431 | 2167 | } |
| 2432 | 2168 | |
| 2433 | 2169 | $response_body = json_decode(wp_remote_retrieve_body($response), true); |
| 2434 | 2170 | |
| 2435 | - $b64 = $response_body['data'][0]['b64_json'] ?? $response_body['data'][0]['b64'] ?? null; | |
| 2436 | - if ($b64) { | |
| 2437 | - $saved_url = $this->mxchat_save_generated_image($b64, 'image/png', 'mxchat-openai'); | |
| 2438 | - if (is_wp_error($saved_url)) { | |
| 2439 | - return ['error' => $saved_url->get_error_message()]; | |
| 2440 | - } | |
| 2441 | - return ['imageUrl' => $saved_url]; | |
| 2171 | + if (isset($response_body['data'][0]['url'])) { | |
| 2172 | + return ['imageUrl' => esc_url_raw($response_body['data'][0]['url'])]; | |
| 2442 | 2173 | } else { |
| 2174 | + //error_log("DALL-E response error: " . wp_remote_retrieve_body($response)); | |
| 2443 | 2175 | return ['error' => esc_html__('Failed to generate image.', 'mxchat')]; |
| 2444 | 2176 | } |
| 2445 | 2177 | } |
| 2446 | 2178 | |
| 2447 | -private function mxchat_generate_imagen_image($prompt, $api_key, $timeout = 60) { | |
| 2448 | - $api_url = 'https://generativelanguage.googleapis.com/v1beta/models/imagen-4.0-generate-001:predict'; | |
| 2449 | - | |
| 2450 | - $body = json_encode([ | |
| 2451 | - 'instances' => [['prompt' => sanitize_text_field($prompt)]], | |
| 2452 | - 'parameters' => [ | |
| 2453 | - 'sampleCount' => 1, | |
| 2454 | - 'aspectRatio' => '1:1', | |
| 2455 | - ], | |
| 2456 | - ]); | |
| 2457 | - | |
| 2458 | - $args = [ | |
| 2459 | - 'body' => $body, | |
| 2460 | - 'headers' => [ | |
| 2461 | - 'Content-Type' => 'application/json', | |
| 2462 | - 'x-goog-api-key' => sanitize_text_field($api_key), | |
| 2463 | - ], | |
| 2464 | - 'method' => 'POST', | |
| 2465 | - 'timeout' => absint($timeout), | |
| 2466 | - ]; | |
| 2467 | - | |
| 2468 | - $response = wp_remote_post($api_url, $args); | |
| 2469 | - | |
| 2470 | - if (is_wp_error($response)) { | |
| 2471 | - return ['error' => esc_html__('Error generating image: ', 'mxchat') . $response->get_error_message()]; | |
| 2472 | - } | |
| 2473 | - | |
| 2474 | - $response_body = json_decode(wp_remote_retrieve_body($response), true); | |
| 2475 | - | |
| 2476 | - $b64 = $response_body['predictions'][0]['bytesBase64Encoded'] ?? $response_body['predictions'][0]['imageBytes'] ?? null; | |
| 2477 | - if ($b64) { | |
| 2478 | - $mime = $response_body['predictions'][0]['mimeType'] ?? 'image/png'; | |
| 2479 | - $saved_url = $this->mxchat_save_generated_image($b64, $mime, 'mxchat-gemini'); | |
| 2480 | - if (is_wp_error($saved_url)) { | |
| 2481 | - return ['error' => $saved_url->get_error_message()]; | |
| 2482 | - } | |
| 2483 | - return ['imageUrl' => $saved_url]; | |
| 2484 | - } else { | |
| 2485 | - return ['error' => esc_html__('Failed to generate image.', 'mxchat')]; | |
| 2486 | - } | |
| 2487 | -} | |
| 2488 | - | |
| 2489 | 2179 | /** |
| 2490 | 2180 | * Handle web search requests. |
| 2491 | 2181 | * |
| 2492 | 2182 | * Sends the refined search query to the Brave Search API and uses the |
| @@ -2753,9 +2443,9 @@ | ||
| 2753 | 2443 | $system_prompt = esc_html__("Interpret the user's request to provide only the essential keywords or phrases for image searching. Remove conversational language, politeness, or extra context. Return a concise search query that doesn't lose any of the original meaning.", 'mxchat'); |
| 2754 | 2444 | |
| 2755 | 2445 | // Get options and determine the selected model |
| 2756 | 2446 | $options = $this->options ?? get_option('mxchat_options'); |
| 2757 | - $selected_model = isset($options['model']) ? $options['model'] : 'gpt-5.1-chat-latest'; | |
| 2447 | + $selected_model = isset($options['model']) ? $options['model'] : 'gpt-4o'; | |
| 2758 | 2448 | |
| 2759 | 2449 | // Extract model prefix to determine the provider |
| 2760 | 2450 | $model_parts = explode('-', $selected_model); |
| 2761 | 2451 | $provider = strtolower($model_parts[0]); |
| @@ -2803,9 +2493,9 @@ | ||
| 2803 | 2493 | |
| 2804 | 2494 | /** |
| 2805 | 2495 | * Interpret query using OpenAI models |
| 2806 | 2496 | */ |
| 2807 | -private function interpret_query_with_openai($user_query, $system_prompt, $api_key, $model = 'gpt-5.1-chat-latest') { | |
| 2497 | +private function interpret_query_with_openai($user_query, $system_prompt, $api_key, $model = 'gpt-4o') { | |
| 2808 | 2498 | $url = 'https://api.openai.com/v1/chat/completions'; |
| 2809 | 2499 | $args = [ |
| 2810 | 2500 | 'headers' => [ |
| 2811 | 2501 | 'Authorization' => 'Bearer ' . $api_key, |
| @@ -2876,13 +2566,13 @@ | ||
| 2876 | 2566 | /** |
| 2877 | 2567 | * Interpret query using Gemini models |
| 2878 | 2568 | */ |
| 2879 | 2569 | private function interpret_query_with_gemini($user_query, $system_prompt, $api_key, $model) { |
| 2880 | - // Use v1beta for preview models, v1 for stable models | |
| 2881 | - $api_version = (strpos($model, 'preview') !== false || strpos($model, 'exp') !== false) ? 'v1beta' : 'v1'; | |
| 2882 | - | |
| 2883 | - $url = "https://generativelanguage.googleapis.com/{$api_version}/models/{$model}:generateContent?key=" . urlencode($api_key); | |
| 2570 | + // Strip "gemini-" prefix for the API | |
| 2571 | + $model_version = str_replace('gemini-', '', $model); | |
| 2884 | 2572 | |
| 2573 | + $url = "https://generativelanguage.googleapis.com/v1/models/$model_version:generateContent?key=" . urlencode($api_key); | |
| 2574 | + | |
| 2885 | 2575 | $args = [ |
| 2886 | 2576 | 'headers' => [ |
| 2887 | 2577 | 'Content-Type' => 'application/json', |
| 2888 | 2578 | ], |
| @@ -3156,14 +2846,9 @@ | ||
| 3156 | 2846 | //error_log("❌ BASIC PROCESSING: Failed to download PDF: " . $error_message); |
| 3157 | 2847 | return false; |
| 3158 | 2848 | } |
| 3159 | 2849 | |
| 3160 | - global $wp_filesystem; | |
| 3161 | - if (empty($wp_filesystem)) { | |
| 3162 | - require_once ABSPATH . 'wp-admin/includes/file.php'; | |
| 3163 | - WP_Filesystem(); | |
| 3164 | - } | |
| 3165 | - $wp_filesystem->put_contents($temp_file, wp_remote_retrieve_body($response), FS_CHMOD_FILE); | |
| 2850 | + file_put_contents($temp_file, wp_remote_retrieve_body($response)); | |
| 3166 | 2851 | //error_log("✅ PDF downloaded successfully"); |
| 3167 | 2852 | } else { |
| 3168 | 2853 | $temp_file = $pdf_source; |
| 3169 | 2854 | //error_log("Using local PDF file: " . $temp_file); |
| @@ -3170,9 +2855,8 @@ | ||
| 3170 | 2855 | } |
| 3171 | 2856 | |
| 3172 | 2857 | // Parse PDF |
| 3173 | 2858 | //error_log("Parsing PDF with basic parser..."); |
| 3174 | - mxchat_load_pdf_parser(); | |
| 3175 | 2859 | $parser = new \Smalot\PdfParser\Parser(); |
| 3176 | 2860 | $pdf = $parser->parseFile($temp_file); |
| 3177 | 2861 | $pages = $pdf->getPages(); |
| 3178 | 2862 | |
| @@ -3323,14 +3007,15 @@ | ||
| 3323 | 3007 | $file = $_FILES['pdf_file']; |
| 3324 | 3008 | $session_id = sanitize_text_field($_POST['session_id']); |
| 3325 | 3009 | $original_filename = sanitize_text_field($file['name']); |
| 3326 | 3010 | |
| 3327 | - // Update session owner if it changed (e.g. IP changed due to network switch) | |
| 3011 | + // SECURITY FIX: Verify session ownership before allowing upload | |
| 3328 | 3012 | $current_user_identifier = MxChat_User::mxchat_get_user_identifier(); |
| 3329 | 3013 | $session_owner = get_option("mxchat_session_owner_{$session_id}"); |
| 3330 | - | |
| 3331 | - if (!$session_owner || $session_owner !== $current_user_identifier) { | |
| 3332 | - update_option("mxchat_session_owner_{$session_id}", $current_user_identifier, 'no'); | |
| 3014 | + | |
| 3015 | + if ($session_owner && $session_owner !== $current_user_identifier) { | |
| 3016 | + wp_send_json_error(esc_html__('Unauthorized access.', 'mxchat')); | |
| 3017 | + return; | |
| 3333 | 3018 | } |
| 3334 | 3019 | |
| 3335 | 3020 | $file_type = wp_check_filetype($file['name'], ['pdf' => 'application/pdf']); |
| 3336 | 3021 | if ($file_type['type'] !== 'application/pdf') { |
| @@ -3434,15 +3119,15 @@ | ||
| 3434 | 3119 | } |
| 3435 | 3120 | |
| 3436 | 3121 | $history = get_option("mxchat_history_{$session_id}", []); |
| 3437 | 3122 | |
| 3438 | - //error_log("MxChat WhatsApp DEBUG: Fetch new messages for session {$session_id}"); | |
| 3439 | - //error_log("MxChat WhatsApp DEBUG: last_seen_id = " . var_export($last_seen_id, true)); | |
| 3440 | - //error_log("MxChat WhatsApp DEBUG: History count = " . count($history)); | |
| 3441 | - //error_log("MxChat WhatsApp DEBUG: Full history = " . print_r($history, true)); | |
| 3123 | + error_log("MxChat WhatsApp DEBUG: Fetch new messages for session {$session_id}"); | |
| 3124 | + error_log("MxChat WhatsApp DEBUG: last_seen_id = " . var_export($last_seen_id, true)); | |
| 3125 | + error_log("MxChat WhatsApp DEBUG: History count = " . count($history)); | |
| 3126 | + error_log("MxChat WhatsApp DEBUG: Full history = " . print_r($history, true)); | |
| 3442 | 3127 | |
| 3443 | 3128 | $new_messages = array_filter($history, function ($message) use ($last_seen_id, $persistence_enabled, $initial_timestamp) { |
| 3444 | - //error_log("MxChat WhatsApp DEBUG: Checking message - ID: " . ($message['id'] ?? 'NO_ID') . ", Role: " . ($message['role'] ?? 'NO_ROLE')); | |
| 3129 | + error_log("MxChat WhatsApp DEBUG: Checking message - ID: " . ($message['id'] ?? 'NO_ID') . ", Role: " . ($message['role'] ?? 'NO_ROLE')); | |
| 3445 | 3130 | |
| 3446 | 3131 | // If persistence is enabled, show all new messages |
| 3447 | 3132 | if ($persistence_enabled) { |
| 3448 | 3133 | $has_id = !empty($message['id']); |
| @@ -3454,9 +3139,9 @@ | ||
| 3454 | 3139 | } else { |
| 3455 | 3140 | $is_newer = strcmp($message['id'] ?? '', $last_seen_id) > 0; |
| 3456 | 3141 | } |
| 3457 | 3142 | |
| 3458 | - //error_log("MxChat WhatsApp DEBUG: has_id={$has_id}, is_newer={$is_newer}, is_agent={$is_agent}"); | |
| 3143 | + error_log("MxChat WhatsApp DEBUG: has_id={$has_id}, is_newer={$is_newer}, is_agent={$is_agent}"); | |
| 3459 | 3144 | |
| 3460 | 3145 | return $has_id && $is_newer && $is_agent; |
| 3461 | 3146 | } |
| 3462 | 3147 | |
| @@ -3465,16 +3150,12 @@ | ||
| 3465 | 3150 | $message['role'] === 'agent' && |
| 3466 | 3151 | $message['timestamp'] > $initial_timestamp; |
| 3467 | 3152 | }); |
| 3468 | 3153 | |
| 3469 | - //error_log("MxChat WhatsApp DEBUG: Filtered messages count = " . count($new_messages)); | |
| 3154 | + error_log("MxChat WhatsApp DEBUG: Filtered messages count = " . count($new_messages)); | |
| 3470 | 3155 | |
| 3471 | - // Include current chat mode so frontend can detect agent→AI transitions | |
| 3472 | - $chat_mode = get_option("mxchat_mode_{$session_id}", 'ai'); | |
| 3473 | - | |
| 3474 | 3156 | wp_send_json_success([ |
| 3475 | - 'new_messages' => array_values($new_messages), | |
| 3476 | - 'chat_mode' => $chat_mode | |
| 3157 | + 'new_messages' => array_values($new_messages) | |
| 3477 | 3158 | ]); |
| 3478 | 3159 | wp_die(); |
| 3479 | 3160 | } |
| 3480 | 3161 | public function mxchat_live_agent_handover($message, $user_id, $session_id) { |
| @@ -3759,393 +3440,9 @@ | ||
| 3759 | 3440 | |
| 3760 | 3441 | //error_log("[DEBUG] Generated channel name: {$channel_name}"); |
| 3761 | 3442 | return $channel_name; |
| 3762 | 3443 | } |
| 3763 | - | |
| 3764 | -/** | |
| 3765 | - * Telegram Live Agent Handover | |
| 3766 | - * Creates a forum topic in the Telegram group and notifies agents | |
| 3767 | - */ | |
| 3768 | -public function mxchat_telegram_live_agent_handover($message, $user_id, $session_id) { | |
| 3769 | - // Check if Telegram agents are available | |
| 3770 | - $telegram_available = $this->options['telegram_status'] ?? 'off'; | |
| 3771 | - if ($telegram_available !== 'on') { | |
| 3772 | - $away_message = $this->options['telegram_away_message'] ?? 'Sorry, live agents are currently unavailable. I can continue helping you as an AI assistant.'; | |
| 3773 | - $this->fallbackResponse = [ | |
| 3774 | - 'text' => $away_message, | |
| 3775 | - 'html' => '', | |
| 3776 | - 'images' => [], | |
| 3777 | - 'chat_mode' => 'ai' | |
| 3778 | - ]; | |
| 3779 | - wp_send_json([ | |
| 3780 | - 'text' => $away_message, | |
| 3781 | - 'html' => '', | |
| 3782 | - 'chat_mode' => 'ai', | |
| 3783 | - 'session_id' => $session_id | |
| 3784 | - ]); | |
| 3785 | - wp_die(); | |
| 3786 | - } | |
| 3787 | - | |
| 3788 | - $telegram_bot_token = $this->options['telegram_bot_token'] ?? ''; | |
| 3789 | - $telegram_group_id = $this->options['telegram_group_id'] ?? ''; | |
| 3790 | - | |
| 3791 | - if (empty($telegram_bot_token) || empty($telegram_group_id)) { | |
| 3792 | - return false; | |
| 3793 | - } | |
| 3794 | - | |
| 3795 | - // Check if topic already exists for this session | |
| 3796 | - $topic_id = get_option("mxchat_telegram_topic_{$session_id}", ''); | |
| 3797 | - | |
| 3798 | - if (empty($topic_id)) { | |
| 3799 | - // Generate topic name | |
| 3800 | - $topic_name = $this->generate_telegram_topic_name($session_id); | |
| 3801 | - | |
| 3802 | - // Random icon color (Telegram forum topic colors) | |
| 3803 | - $icon_colors = [0x6FB9F0, 0xFFD67E, 0xCB86DB, 0x8EEE98, 0xFF93B2, 0xFB6F5F]; | |
| 3804 | - $icon_color = $icon_colors[array_rand($icon_colors)]; | |
| 3805 | - | |
| 3806 | - // Create forum topic | |
| 3807 | - $response = wp_remote_post("https://api.telegram.org/bot{$telegram_bot_token}/createForumTopic", [ | |
| 3808 | - 'headers' => ['Content-Type' => 'application/json'], | |
| 3809 | - 'body' => json_encode([ | |
| 3810 | - 'chat_id' => $telegram_group_id, | |
| 3811 | - 'name' => $topic_name, | |
| 3812 | - 'icon_color' => $icon_color | |
| 3813 | - ]) | |
| 3814 | - ]); | |
| 3815 | - | |
| 3816 | - if (!is_wp_error($response)) { | |
| 3817 | - $response_body = wp_remote_retrieve_body($response); | |
| 3818 | - $response_data = json_decode($response_body, true); | |
| 3819 | - | |
| 3820 | - if (isset($response_data['ok']) && $response_data['ok']) { | |
| 3821 | - $topic_id = $response_data['result']['message_thread_id']; | |
| 3822 | - update_option("mxchat_telegram_topic_{$session_id}", $topic_id); | |
| 3823 | - update_option("mxchat_telegram_group_{$session_id}", $telegram_group_id); | |
| 3824 | - } | |
| 3825 | - } | |
| 3826 | - | |
| 3827 | - if (empty($topic_id)) { | |
| 3828 | - return false; // Failed to create topic | |
| 3829 | - } | |
| 3830 | - } | |
| 3831 | - | |
| 3832 | - // Get recent chat history | |
| 3833 | - $history = get_option("mxchat_history_{$session_id}", []); | |
| 3834 | - $recent_history = array_slice($history, -5); | |
| 3835 | - | |
| 3836 | - // Format conversation context for Telegram (HTML format) | |
| 3837 | - $conversation_context = ""; | |
| 3838 | - if (!empty($recent_history)) { | |
| 3839 | - $conversation_context = "<b>Recent Conversation:</b>\n"; | |
| 3840 | - foreach ($recent_history as $hist_message) { | |
| 3841 | - $role_display = $hist_message['role'] === 'user' ? '👤 User' : '🤖 AI'; | |
| 3842 | - $escaped_content = htmlspecialchars($hist_message['content'], ENT_QUOTES, 'UTF-8'); | |
| 3843 | - $conversation_context .= "{$role_display}: {$escaped_content}\n"; | |
| 3844 | - } | |
| 3845 | - $conversation_context .= "\n"; | |
| 3846 | - } | |
| 3847 | - | |
| 3848 | - // Get user info | |
| 3849 | - $user_email = get_option("mxchat_email_{$session_id}", 'Not provided'); | |
| 3850 | - $user_name = get_option("mxchat_name_{$session_id}", 'Anonymous'); | |
| 3851 | - | |
| 3852 | - // Update session mode | |
| 3853 | - update_option("mxchat_mode_{$session_id}", 'agent'); | |
| 3854 | - | |
| 3855 | - // Send initial message to topic | |
| 3856 | - $escaped_message = htmlspecialchars($message, ENT_QUOTES, 'UTF-8'); | |
| 3857 | - $topic_message = "🔔 <b>New Live Agent Request</b>\n\n"; | |
| 3858 | - $topic_message .= "<b>Session ID:</b> <code>{$session_id}</code>\n"; | |
| 3859 | - $topic_message .= "<b>User:</b> {$user_name}\n"; | |
| 3860 | - $topic_message .= "<b>Email:</b> {$user_email}\n\n"; | |
| 3861 | - | |
| 3862 | - if (!empty($conversation_context)) { | |
| 3863 | - $topic_message .= $conversation_context; | |
| 3864 | - } | |
| 3865 | - | |
| 3866 | - $topic_message .= "<b>Current Message:</b>\n{$escaped_message}\n\n"; | |
| 3867 | - $topic_message .= "<i>Reply in this topic - messages will be sent to the user</i>\n"; | |
| 3868 | - $topic_message .= "<i>Type #close, #end, #disconnect, or #done to end the session</i>"; | |
| 3869 | - | |
| 3870 | - wp_remote_post("https://api.telegram.org/bot{$telegram_bot_token}/sendMessage", [ | |
| 3871 | - 'headers' => ['Content-Type' => 'application/json'], | |
| 3872 | - 'body' => json_encode([ | |
| 3873 | - 'chat_id' => $telegram_group_id, | |
| 3874 | - 'message_thread_id' => $topic_id, | |
| 3875 | - 'text' => $topic_message, | |
| 3876 | - 'parse_mode' => 'HTML' | |
| 3877 | - ]) | |
| 3878 | - ]); | |
| 3879 | - | |
| 3880 | - $success_message = $this->options['telegram_notification_message'] ?? "I've notified a support agent. Please allow a moment for them to respond."; | |
| 3881 | - $this->mxchat_save_chat_message($session_id, 'bot', $success_message); | |
| 3882 | - | |
| 3883 | - $this->fallbackResponse = [ | |
| 3884 | - 'text' => $success_message, | |
| 3885 | - 'html' => '', | |
| 3886 | - 'images' => [], | |
| 3887 | - 'chat_mode' => 'agent' | |
| 3888 | - ]; | |
| 3889 | - | |
| 3890 | - wp_send_json([ | |
| 3891 | - 'success' => true, | |
| 3892 | - 'text' => $success_message, | |
| 3893 | - 'html' => '', | |
| 3894 | - 'chat_mode' => 'agent', | |
| 3895 | - 'session_id' => $session_id, | |
| 3896 | - 'fallbackResponse' => $this->fallbackResponse | |
| 3897 | - ]); | |
| 3898 | - wp_die(); | |
| 3899 | -} | |
| 3900 | - | |
| 3901 | -/** | |
| 3902 | - * Generate topic name for Telegram forum | |
| 3903 | - */ | |
| 3904 | -private function generate_telegram_topic_name($session_id) { | |
| 3905 | - $name = null; | |
| 3906 | - $email = null; | |
| 3907 | - | |
| 3908 | - // Check logged in user | |
| 3909 | - if (is_user_logged_in()) { | |
| 3910 | - $current_user = wp_get_current_user(); | |
| 3911 | - if (!empty($current_user->display_name)) { | |
| 3912 | - $name = $current_user->display_name; | |
| 3913 | - } | |
| 3914 | - if (!empty($current_user->user_email)) { | |
| 3915 | - $email = $current_user->user_email; | |
| 3916 | - } | |
| 3917 | - } | |
| 3918 | - | |
| 3919 | - // Check session data | |
| 3920 | - if (empty($name)) { | |
| 3921 | - $name = get_option("mxchat_name_{$session_id}"); | |
| 3922 | - } | |
| 3923 | - if (empty($email)) { | |
| 3924 | - $email = get_option("mxchat_email_{$session_id}"); | |
| 3925 | - } | |
| 3926 | - | |
| 3927 | - // Generate topic name | |
| 3928 | - $session_suffix = substr($session_id, -6); | |
| 3929 | - | |
| 3930 | - if (!empty($name)) { | |
| 3931 | - // Clean name for topic (max 128 chars in Telegram) | |
| 3932 | - $clean_name = preg_replace('/[^\p{L}\p{N}\s\-]/u', '', $name); | |
| 3933 | - $clean_name = trim($clean_name); | |
| 3934 | - if (strlen($clean_name) > 50) { | |
| 3935 | - $clean_name = substr($clean_name, 0, 50); | |
| 3936 | - } | |
| 3937 | - return "Chat - {$clean_name} ({$session_suffix})"; | |
| 3938 | - } elseif (!empty($email)) { | |
| 3939 | - // Use email prefix | |
| 3940 | - $email_prefix = explode('@', $email)[0]; | |
| 3941 | - if (strlen($email_prefix) > 30) { | |
| 3942 | - $email_prefix = substr($email_prefix, 0, 30); | |
| 3943 | - } | |
| 3944 | - return "Chat - {$email_prefix} ({$session_suffix})"; | |
| 3945 | - } | |
| 3946 | - | |
| 3947 | - return "Chat - {$session_suffix}"; | |
| 3948 | -} | |
| 3949 | - | |
| 3950 | -/** | |
| 3951 | - * Send user message to Telegram agent | |
| 3952 | - */ | |
| 3953 | -public function mxchat_send_user_message_to_telegram_agent($message, $user_id, $session_id) { | |
| 3954 | - $telegram_bot_token = $this->options['telegram_bot_token'] ?? ''; | |
| 3955 | - $topic_id = get_option("mxchat_telegram_topic_{$session_id}", ''); | |
| 3956 | - $group_id = get_option("mxchat_telegram_group_{$session_id}", ''); | |
| 3957 | - | |
| 3958 | - if (empty($telegram_bot_token) || empty($topic_id) || empty($group_id)) { | |
| 3959 | - return false; | |
| 3960 | - } | |
| 3961 | - | |
| 3962 | - $escaped_message = htmlspecialchars($message, ENT_QUOTES, 'UTF-8'); | |
| 3963 | - $user_message = "👤 <b>User:</b> {$escaped_message}"; | |
| 3964 | - | |
| 3965 | - $response = wp_remote_post("https://api.telegram.org/bot{$telegram_bot_token}/sendMessage", [ | |
| 3966 | - 'headers' => ['Content-Type' => 'application/json'], | |
| 3967 | - 'body' => json_encode([ | |
| 3968 | - 'chat_id' => $group_id, | |
| 3969 | - 'message_thread_id' => $topic_id, | |
| 3970 | - 'text' => $user_message, | |
| 3971 | - 'parse_mode' => 'HTML' | |
| 3972 | - ]) | |
| 3973 | - ]); | |
| 3974 | - | |
| 3975 | - return !is_wp_error($response); | |
| 3976 | -} | |
| 3977 | - | |
| 3978 | -/** | |
| 3979 | - * Handle incoming Telegram webhook | |
| 3980 | - */ | |
| 3981 | -public function handle_telegram_webhook(WP_REST_Request $request) { | |
| 3982 | - $body = $request->get_body(); | |
| 3983 | - $data = json_decode($body, true); | |
| 3984 | - | |
| 3985 | - //error_log('[MxChat Telegram DEBUG] Webhook received: ' . $body); | |
| 3986 | - | |
| 3987 | - // Handle message events from forum topics | |
| 3988 | - if (isset($data['message'])) { | |
| 3989 | - $message_data = $data['message']; | |
| 3990 | - | |
| 3991 | - // Skip if not from a forum topic | |
| 3992 | - if (!isset($message_data['message_thread_id'])) { | |
| 3993 | - //error_log('[MxChat Telegram DEBUG] Skipped: No message_thread_id (not a forum topic message)'); | |
| 3994 | - return new WP_REST_Response(['ok' => true]); | |
| 3995 | - } | |
| 3996 | - | |
| 3997 | - // Skip bot messages | |
| 3998 | - if (isset($message_data['from']['is_bot']) && $message_data['from']['is_bot']) { | |
| 3999 | - //error_log('[MxChat Telegram DEBUG] Skipped: Message from bot'); | |
| 4000 | - return new WP_REST_Response(['ok' => true]); | |
| 4001 | - } | |
| 4002 | - | |
| 4003 | - $chat_id = $message_data['chat']['id'] ?? ''; | |
| 4004 | - $topic_id = $message_data['message_thread_id']; | |
| 4005 | - $message_text = $message_data['text'] ?? ''; | |
| 4006 | - $message_id = $message_data['message_id'] ?? ''; | |
| 4007 | - $from = $message_data['from'] ?? []; | |
| 4008 | - $agent_name = trim(($from['first_name'] ?? '') . ' ' . ($from['last_name'] ?? '')); | |
| 4009 | - if (empty($agent_name)) { | |
| 4010 | - $agent_name = $from['username'] ?? 'Agent'; | |
| 4011 | - } | |
| 4012 | - | |
| 4013 | - //error_log("[MxChat Telegram DEBUG] Parsed: chat_id={$chat_id}, topic_id={$topic_id}, agent={$agent_name}, text={$message_text}"); | |
| 4014 | - | |
| 4015 | - // Skip empty messages | |
| 4016 | - if (empty($message_text)) { | |
| 4017 | - //error_log('[MxChat Telegram DEBUG] Skipped: Empty message text'); | |
| 4018 | - return new WP_REST_Response(['ok' => true]); | |
| 4019 | - } | |
| 4020 | - | |
| 4021 | - // Find session ID by topic ID - cast to string for comparison | |
| 4022 | - global $wpdb; | |
| 4023 | - $topic_id_str = strval($topic_id); | |
| 4024 | - $session_option = $wpdb->get_var( | |
| 4025 | - $wpdb->prepare( | |
| 4026 | - "SELECT option_name FROM {$wpdb->options} | |
| 4027 | - WHERE option_name LIKE %s | |
| 4028 | - AND option_value = %s", | |
| 4029 | - 'mxchat_telegram_topic_%', | |
| 4030 | - $topic_id_str | |
| 4031 | - ) | |
| 4032 | - ); | |
| 4033 | - | |
| 4034 | - //error_log("[MxChat Telegram DEBUG] Looking for topic_id={$topic_id_str} in options, found: " . ($session_option ?: 'NULL')); | |
| 4035 | - | |
| 4036 | - if ($session_option) { | |
| 4037 | - $session_id = str_replace('mxchat_telegram_topic_', '', $session_option); | |
| 4038 | - //error_log("[MxChat Telegram DEBUG] Session ID: {$session_id}"); | |
| 4039 | - | |
| 4040 | - // Verify the group ID matches | |
| 4041 | - $stored_group_id = get_option("mxchat_telegram_group_{$session_id}", ''); | |
| 4042 | - //error_log("[MxChat Telegram DEBUG] Stored group_id={$stored_group_id}, received chat_id={$chat_id}"); | |
| 4043 | - | |
| 4044 | - if (strval($stored_group_id) != strval($chat_id)) { | |
| 4045 | - //error_log('[MxChat Telegram DEBUG] Skipped: Group ID mismatch'); | |
| 4046 | - return new WP_REST_Response(['ok' => true]); | |
| 4047 | - } | |
| 4048 | - | |
| 4049 | - // Check for closure commands | |
| 4050 | - $lower_text = strtolower(trim($message_text)); | |
| 4051 | - if (in_array($lower_text, ['#close', '#end', '#disconnect', '#done'])) { | |
| 4052 | - //error_log("[MxChat Telegram DEBUG] Closure command received: {$lower_text}"); | |
| 4053 | - // End the live agent session | |
| 4054 | - update_option("mxchat_mode_{$session_id}", 'ai'); | |
| 4055 | - | |
| 4056 | - // Save disconnect message | |
| 4057 | - $disconnect_message = "Live agent session ended. You're now chatting with the AI assistant."; | |
| 4058 | - $this->mxchat_save_chat_message($session_id, 'bot', $disconnect_message); | |
| 4059 | - | |
| 4060 | - // Notify in Telegram | |
| 4061 | - $telegram_bot_token = $this->options['telegram_bot_token'] ?? ''; | |
| 4062 | - if (!empty($telegram_bot_token)) { | |
| 4063 | - wp_remote_post("https://api.telegram.org/bot{$telegram_bot_token}/sendMessage", [ | |
| 4064 | - 'headers' => ['Content-Type' => 'application/json'], | |
| 4065 | - 'body' => json_encode([ | |
| 4066 | - 'chat_id' => $chat_id, | |
| 4067 | - 'message_thread_id' => $topic_id, | |
| 4068 | - 'text' => "✅ Session closed. User returned to AI chatbot.", | |
| 4069 | - 'parse_mode' => 'HTML' | |
| 4070 | - ]) | |
| 4071 | - ]); | |
| 4072 | - | |
| 4073 | - // Optionally close the topic | |
| 4074 | - wp_remote_post("https://api.telegram.org/bot{$telegram_bot_token}/closeForumTopic", [ | |
| 4075 | - 'headers' => ['Content-Type' => 'application/json'], | |
| 4076 | - 'body' => json_encode([ | |
| 4077 | - 'chat_id' => $chat_id, | |
| 4078 | - 'message_thread_id' => $topic_id | |
| 4079 | - ]) | |
| 4080 | - ]); | |
| 4081 | - } | |
| 4082 | - | |
| 4083 | - return new WP_REST_Response(['ok' => true]); | |
| 4084 | - } | |
| 4085 | - | |
| 4086 | - // Deduplicate messages | |
| 4087 | - $message_key = md5($session_id . $message_id . $message_text); | |
| 4088 | - $processed_messages = get_transient('mxchat_telegram_messages_' . $session_id) ?: []; | |
| 4089 | - | |
| 4090 | - if (in_array($message_key, $processed_messages)) { | |
| 4091 | - //error_log('[MxChat Telegram DEBUG] Skipped: Duplicate message'); | |
| 4092 | - return new WP_REST_Response(['ok' => true]); | |
| 4093 | - } | |
| 4094 | - | |
| 4095 | - $processed_messages[] = $message_key; | |
| 4096 | - if (count($processed_messages) > 50) { | |
| 4097 | - $processed_messages = array_slice($processed_messages, -50); | |
| 4098 | - } | |
| 4099 | - set_transient('mxchat_telegram_messages_' . $session_id, $processed_messages, HOUR_IN_SECONDS); | |
| 4100 | - | |
| 4101 | - // Save the agent message - format with agent name prefix for proper parsing | |
| 4102 | - $formatted_message = "Agent: {$agent_name} - {$message_text}"; | |
| 4103 | - //error_log("[MxChat Telegram DEBUG] Saving agent message: {$formatted_message}"); | |
| 4104 | - | |
| 4105 | - $this->mxchat_save_chat_message($session_id, 'agent', $formatted_message); | |
| 4106 | - | |
| 4107 | - // Verify the message was saved to history | |
| 4108 | - $history = get_option("mxchat_history_{$session_id}", []); | |
| 4109 | - $last_message = end($history); | |
| 4110 | - //error_log("[MxChat Telegram DEBUG] History after save - count: " . count($history) . ", last message role: " . ($last_message['role'] ?? 'none')); | |
| 4111 | - | |
| 4112 | - // Send confirmation back to Telegram | |
| 4113 | - $telegram_bot_token = $this->options['telegram_bot_token'] ?? ''; | |
| 4114 | - if (!empty($telegram_bot_token)) { | |
| 4115 | - $confirm_key = 'mxchat_telegram_confirm_' . $message_key; | |
| 4116 | - if (!get_transient($confirm_key)) { | |
| 4117 | - wp_remote_post("https://api.telegram.org/bot{$telegram_bot_token}/sendMessage", [ | |
| 4118 | - 'headers' => ['Content-Type' => 'application/json'], | |
| 4119 | - 'body' => json_encode([ | |
| 4120 | - 'chat_id' => $chat_id, | |
| 4121 | - 'message_thread_id' => $topic_id, | |
| 4122 | - 'text' => "✅ <i>Message sent to user</i>", | |
| 4123 | - 'parse_mode' => 'HTML', | |
| 4124 | - 'reply_to_message_id' => $message_id | |
| 4125 | - ]) | |
| 4126 | - ]); | |
| 4127 | - set_transient($confirm_key, true, 300); | |
| 4128 | - } | |
| 4129 | - } | |
| 4130 | - } else { | |
| 4131 | - //error_log("[MxChat Telegram DEBUG] No session found for topic_id={$topic_id}"); | |
| 4132 | - } | |
| 4133 | - } else { | |
| 4134 | - //error_log('[MxChat Telegram DEBUG] No message in webhook data'); | |
| 4135 | - } | |
| 4136 | - | |
| 4137 | - return new WP_REST_Response(['ok' => true]); | |
| 4138 | -} | |
| 4139 | - | |
| 4140 | 3444 | public function mxchat_send_user_message_to_agent($message, $user_id, $session_id) { |
| 4141 | - // Check if this is a Telegram agent session | |
| 4142 | - $telegram_topic_id = get_option("mxchat_telegram_topic_{$session_id}", ''); | |
| 4143 | - if (!empty($telegram_topic_id)) { | |
| 4144 | - return $this->mxchat_send_user_message_to_telegram_agent($message, $user_id, $session_id); | |
| 4145 | - } | |
| 4146 | - | |
| 4147 | - // Otherwise, try Slack | |
| 4148 | 3445 | $slack_bot_token = $this->options['live_agent_bot_token'] ?? ''; |
| 4149 | 3446 | $channel_id = get_option("mxchat_channel_{$session_id}", ''); |
| 4150 | 3447 | |
| 4151 | 3448 | if (empty($slack_bot_token) || empty($channel_id)) { |
| @@ -4371,33 +3668,33 @@ | ||
| 4371 | 3668 | |
| 4372 | 3669 | $channel_id = $event['channel']; |
| 4373 | 3670 | $message_text = $event['text'] ?? ''; |
| 4374 | 3671 | $message_ts = $event['ts'] ?? ''; |
| 4375 | - | |
| 3672 | + | |
| 4376 | 3673 | // Find session ID by looking for matching channel |
| 4377 | 3674 | global $wpdb; |
| 4378 | 3675 | $session_option = $wpdb->get_var( |
| 4379 | 3676 | $wpdb->prepare( |
| 4380 | - "SELECT option_name FROM {$wpdb->options} | |
| 4381 | - WHERE option_name LIKE 'mxchat_channel_%' | |
| 3677 | + "SELECT option_name FROM {$wpdb->options} | |
| 3678 | + WHERE option_name LIKE 'mxchat_channel_%' | |
| 4382 | 3679 | AND option_value = %s", |
| 4383 | 3680 | $channel_id |
| 4384 | 3681 | ) |
| 4385 | 3682 | ); |
| 4386 | - | |
| 3683 | + | |
| 4387 | 3684 | if ($session_option) { |
| 4388 | 3685 | $session_id = str_replace('mxchat_channel_', '', $session_option); |
| 4389 | - | |
| 3686 | + | |
| 4390 | 3687 | // Create a unique key for this specific message |
| 4391 | 3688 | $message_key = md5($session_id . $message_ts . $message_text); |
| 4392 | 3689 | $processed_messages = get_transient('mxchat_processed_messages_' . $session_id) ?: []; |
| 4393 | - | |
| 3690 | + | |
| 4394 | 3691 | // Check if we've already processed this exact message |
| 4395 | 3692 | if (in_array($message_key, $processed_messages)) { |
| 4396 | 3693 | //error_log("Duplicate message detected for session $session_id"); |
| 4397 | 3694 | return new WP_REST_Response(['ok' => true]); |
| 4398 | 3695 | } |
| 4399 | - | |
| 3696 | + | |
| 4400 | 3697 | // Add to processed messages |
| 4401 | 3698 | $processed_messages[] = $message_key; |
| 4402 | 3699 | // Keep only last 50 messages per session |
| 4403 | 3700 | if (count($processed_messages) > 50) { |
| @@ -4403,46 +3700,14 @@ | ||
| 4403 | 3700 | if (count($processed_messages) > 50) { |
| 4404 | 3701 | $processed_messages = array_slice($processed_messages, -50); |
| 4405 | 3702 | } |
| 4406 | 3703 | set_transient('mxchat_processed_messages_' . $session_id, $processed_messages, HOUR_IN_SECONDS); |
| 4407 | - | |
| 4408 | - $slack_bot_token = $this->options['live_agent_bot_token'] ?? ''; | |
| 4409 | - | |
| 4410 | - // Handle agent ending the chat — transfer back to AI | |
| 4411 | - // Format: "!endchat" or "!endchat <custom message to user>" | |
| 4412 | - if (preg_match('/^!endchat\b/i', trim($message_text))) { | |
| 4413 | - update_option("mxchat_mode_{$session_id}", 'ai'); | |
| 4414 | - | |
| 4415 | - // Extract custom message after !endchat, or use empty string | |
| 4416 | - $custom_message = trim(preg_replace('/^!endchat\s*/i', '', trim($message_text))); | |
| 4417 | - | |
| 4418 | - // Send the agent's custom farewell message if provided | |
| 4419 | - if (!empty($custom_message)) { | |
| 4420 | - $this->mxchat_save_chat_message($session_id, 'agent', $custom_message); | |
| 4421 | - } | |
| 4422 | - | |
| 4423 | - // Confirm in Slack channel | |
| 4424 | - if (!empty($slack_bot_token)) { | |
| 4425 | - wp_remote_post('https://slack.com/api/chat.postMessage', [ | |
| 4426 | - 'headers' => [ | |
| 4427 | - 'Content-Type' => 'application/json', | |
| 4428 | - 'Authorization' => 'Bearer ' . $slack_bot_token | |
| 4429 | - ], | |
| 4430 | - 'body' => json_encode([ | |
| 4431 | - 'channel' => $channel_id, | |
| 4432 | - 'text' => "✅ *Chat ended.* User has been transferred back to AI mode.", | |
| 4433 | - 'mrkdwn' => true | |
| 4434 | - ]) | |
| 4435 | - ]); | |
| 4436 | - } | |
| 4437 | - | |
| 4438 | - return new WP_REST_Response(['ok' => true]); | |
| 4439 | - } | |
| 4440 | - | |
| 3704 | + | |
| 4441 | 3705 | // Save the agent message |
| 4442 | 3706 | $this->mxchat_save_chat_message($session_id, 'agent', $message_text); |
| 4443 | - | |
| 3707 | + | |
| 4444 | 3708 | // Send confirmation back to Slack (only once) |
| 3709 | + $slack_bot_token = $this->options['live_agent_bot_token'] ?? ''; | |
| 4445 | 3710 | if (!empty($slack_bot_token)) { |
| 4446 | 3711 | // Use a transient to prevent duplicate confirmations |
| 4447 | 3712 | $confirm_key = 'mxchat_confirm_' . $message_key; |
| 4448 | 3713 | if (!get_transient($confirm_key)) { |
| @@ -4452,9 +3717,9 @@ | ||
| 4452 | 3717 | 'Authorization' => 'Bearer ' . $slack_bot_token |
| 4453 | 3718 | ], |
| 4454 | 3719 | 'body' => json_encode([ |
| 4455 | 3720 | 'channel' => $channel_id, |
| 4456 | - 'text' => "✅ _Message sent to user_", | |
| 3721 | + 'text' => "✅ _Message sent to user_", | |
| 4457 | 3722 | 'thread_ts' => $event['ts'] // Reply in thread |
| 4458 | 3723 | ]) |
| 4459 | 3724 | ]); |
| 4460 | 3725 | // Set transient to prevent duplicate confirmations |
| @@ -4693,44 +3958,26 @@ | ||
| 4693 | 3958 | } |
| 4694 | 3959 | } |
| 4695 | 3960 | |
| 4696 | 3961 | |
| 4697 | -private function mxchat_find_relevant_content($user_embedding, $bot_id = 'default', $user_query = '') { | |
| 4698 | - //error_log("MXCHAT DEBUG: find_relevant_content called with bot_id: " . $bot_id); | |
| 4699 | - | |
| 4700 | - // Check for OpenAI Vector Store first (takes priority when enabled) | |
| 4701 | - $bot_vectorstore_config = $this->get_bot_vectorstore_config($bot_id); | |
| 4702 | - | |
| 4703 | - if ($bot_vectorstore_config['use_vectorstore']) { | |
| 4704 | - // Get current model to verify it's an OpenAI model | |
| 4705 | - $bot_options = $this->get_bot_options($bot_id); | |
| 4706 | - $mxchat_options = get_option('mxchat_options', array()); | |
| 4707 | - $current_options = !empty($bot_options) ? $bot_options : $mxchat_options; | |
| 4708 | - $selected_model = $current_options['model'] ?? 'gpt-5.1-chat-latest'; | |
| 4709 | - | |
| 4710 | - if ($this->is_openai_chat_model($selected_model)) { | |
| 4711 | - //error_log("MXCHAT DEBUG: Using OpenAI Vector Store for knowledge retrieval"); | |
| 4712 | - return $this->find_relevant_content_openai_vectorstore($user_query, $bot_id, $bot_vectorstore_config); | |
| 4713 | - } else { | |
| 4714 | - //error_log("MXCHAT DEBUG: Vector Store enabled but model is not OpenAI (" . $selected_model . "), skipping Vector Store"); | |
| 4715 | - } | |
| 4716 | - } | |
| 4717 | - | |
| 3962 | +private function mxchat_find_relevant_content($user_embedding, $bot_id = 'default') { | |
| 3963 | + error_log("MXCHAT DEBUG: find_relevant_content called with bot_id: " . $bot_id); | |
| 3964 | + | |
| 4718 | 3965 | // Get bot-specific Pinecone configuration |
| 4719 | 3966 | $bot_pinecone_config = $this->get_bot_pinecone_config($bot_id); |
| 4720 | - | |
| 3967 | + | |
| 4721 | 3968 | // Debug: Log the Pinecone configuration |
| 4722 | - //error_log("MXCHAT DEBUG: Pinecone config for bot '$bot_id':"); | |
| 4723 | - //error_log(" - use_pinecone: " . ($bot_pinecone_config['use_pinecone'] ? 'true' : 'false')); | |
| 4724 | - //error_log(" - api_key: " . (empty($bot_pinecone_config['api_key']) ? 'EMPTY' : 'SET (hidden)')); | |
| 4725 | - //error_log(" - host: " . ($bot_pinecone_config['host'] ?? 'NOT SET')); | |
| 4726 | - //error_log(" - namespace: " . ($bot_pinecone_config['namespace'] ?? 'NOT SET')); | |
| 4727 | - | |
| 3969 | + error_log("MXCHAT DEBUG: Pinecone config for bot '$bot_id':"); | |
| 3970 | + error_log(" - use_pinecone: " . ($bot_pinecone_config['use_pinecone'] ? 'true' : 'false')); | |
| 3971 | + error_log(" - api_key: " . (empty($bot_pinecone_config['api_key']) ? 'EMPTY' : 'SET (hidden)')); | |
| 3972 | + error_log(" - host: " . ($bot_pinecone_config['host'] ?? 'NOT SET')); | |
| 3973 | + error_log(" - namespace: " . ($bot_pinecone_config['namespace'] ?? 'NOT SET')); | |
| 3974 | + | |
| 4728 | 3975 | // Determine whether to use Pinecone based on bot configuration |
| 4729 | 3976 | $use_pinecone = isset($bot_pinecone_config['use_pinecone']) ? $bot_pinecone_config['use_pinecone'] : false; |
| 3977 | + | |
| 3978 | + error_log("MXCHAT DEBUG: Using " . ($use_pinecone ? "Pinecone" : "WordPress Database") . " for knowledge retrieval"); | |
| 4730 | 3979 | |
| 4731 | - //error_log("MXCHAT DEBUG: Using " . ($use_pinecone ? "Pinecone" : "WordPress Database") . " for knowledge retrieval"); | |
| 4732 | - | |
| 4733 | 3980 | if ($use_pinecone) { |
| 4734 | 3981 | return $this->find_relevant_content_pinecone($user_embedding, $bot_id, $bot_pinecone_config); |
| 4735 | 3982 | } else { |
| 4736 | 3983 | return $this->find_relevant_content_wordpress($user_embedding, $bot_id); |
| @@ -4739,8 +3986,11 @@ | ||
| 4739 | 3986 | |
| 4740 | 3987 | private function find_relevant_content_wordpress($user_embedding, $bot_id = 'default') { |
| 4741 | 3988 | global $wpdb; |
| 4742 | 3989 | $system_prompt_table = $wpdb->prefix . 'mxchat_system_prompt_content'; |
| 3990 | + $cache_key = 'mxchat_system_prompt_embeddings_' . $bot_id; | |
| 3991 | + $batch_size = 500; | |
| 3992 | + | |
| 4743 | 3993 | // Initialize similarity analysis storage |
| 4744 | 3994 | $this->last_similarity_analysis = [ |
| 4745 | 3995 | 'knowledge_base_type' => 'WordPress Database', |
| 4746 | 3996 | 'bot_id' => $bot_id, |
| @@ -4755,467 +4005,263 @@ | ||
| 4755 | 4005 | // Get bot-specific options for similarity threshold |
| 4756 | 4006 | $bot_options = $this->get_bot_options($bot_id); |
| 4757 | 4007 | $current_options = !empty($bot_options) ? $bot_options : $this->options; |
| 4758 | 4008 | |
| 4759 | - // Get knowledge manager instance for role checking | |
| 4760 | - $knowledge_manager = MxChat_Knowledge_Manager::get_instance(); | |
| 4009 | + // Retrieve embeddings from cache or database | |
| 4010 | + $embeddings = wp_cache_get($cache_key, 'mxchat_system_prompts'); | |
| 4011 | + if ($embeddings === false) { | |
| 4012 | + // Cache miss - load embeddings from database WITH CONTENT and ROLE RESTRICTION for testing | |
| 4013 | + $embeddings = []; | |
| 4014 | + $offset = 0; | |
| 4761 | 4015 | |
| 4762 | - // Get base similarity threshold from bot options or default options | |
| 4763 | - $similarity_threshold = isset($current_options['similarity_threshold']) | |
| 4764 | - ? ((int) $current_options['similarity_threshold']) / 100 | |
| 4765 | - : 0.35; | |
| 4766 | - $this->last_similarity_analysis['threshold_used'] = $similarity_threshold; | |
| 4016 | + do { | |
| 4017 | + // Add bot_id filter if not default and if bot_metadata column exists | |
| 4018 | + $bot_filter = ''; | |
| 4019 | + if ($bot_id !== 'default') { | |
| 4020 | + // Check if bot_metadata column exists | |
| 4021 | + $column_exists = $wpdb->get_var("SHOW COLUMNS FROM {$system_prompt_table} LIKE 'bot_metadata'"); | |
| 4022 | + if ($column_exists) { | |
| 4023 | + $bot_filter = $wpdb->prepare(" AND (bot_metadata = %s OR bot_metadata IS NULL OR bot_metadata = '')", $bot_id); | |
| 4024 | + } | |
| 4025 | + } | |
| 4767 | 4026 | |
| 4768 | - // Precompute bot_filter once, outside the streaming loop | |
| 4769 | - $bot_filter = ''; | |
| 4770 | - if ($bot_id !== 'default') { | |
| 4771 | - $column_exists = $wpdb->get_var("SHOW COLUMNS FROM {$system_prompt_table} LIKE 'bot_metadata'"); | |
| 4772 | - if ($column_exists) { | |
| 4773 | - $bot_filter = $wpdb->prepare(" AND (bot_metadata = %s OR bot_metadata IS NULL OR bot_metadata = '')", $bot_id); | |
| 4774 | - } | |
| 4775 | - } | |
| 4027 | + $query = $wpdb->prepare( | |
| 4028 | + "SELECT id, embedding_vector, article_content, source_url, role_restriction | |
| 4029 | + FROM {$system_prompt_table} | |
| 4030 | + WHERE 1=1 {$bot_filter} | |
| 4031 | + LIMIT %d OFFSET %d", | |
| 4032 | + $batch_size, | |
| 4033 | + $offset | |
| 4034 | + ); | |
| 4776 | 4035 | |
| 4777 | - // ===== STREAMING TOP-K PASS ===== | |
| 4778 | - // Stream rows in small batches, compute cosine similarity per row, and keep only: | |
| 4779 | - // - top 10 by raw similarity (for the testing/debug display panel) | |
| 4780 | - // - candidates above threshold with access (capped) for context assembly | |
| 4781 | - // This bounds peak memory regardless of knowledge base size and avoids loading | |
| 4782 | - // article_content for every row. article_content is fetched in Phase 2 for winners only. | |
| 4783 | - $batch_size = 250; | |
| 4784 | - $max_candidates = 200; // safety cap, well above rag_sources_limit * max_chunks_per_source | |
| 4785 | - $top_display = []; | |
| 4786 | - $candidates = []; | |
| 4787 | - $total_checked = 0; | |
| 4788 | - $offset = 0; | |
| 4036 | + $batch = $wpdb->get_results($query); | |
| 4037 | + if (empty($batch)) { | |
| 4038 | + break; | |
| 4039 | + } | |
| 4789 | 4040 | |
| 4790 | - do { | |
| 4791 | - $batch = $wpdb->get_results($wpdb->prepare( | |
| 4792 | - "SELECT id, embedding_vector, source_url, role_restriction | |
| 4793 | - FROM {$system_prompt_table} | |
| 4794 | - WHERE 1=1 {$bot_filter} | |
| 4795 | - LIMIT %d OFFSET %d", | |
| 4796 | - $batch_size, | |
| 4797 | - $offset | |
| 4798 | - )); | |
| 4041 | + $embeddings = array_merge($embeddings, $batch); | |
| 4042 | + $offset += $batch_size; | |
| 4043 | + unset($batch); | |
| 4044 | + } while (true); | |
| 4799 | 4045 | |
| 4800 | - if (empty($batch)) { | |
| 4801 | - break; | |
| 4046 | + if (empty($embeddings)) { | |
| 4047 | + // Store empty array for valid URLs since no content found | |
| 4048 | + $this->current_valid_urls = []; | |
| 4049 | + return ''; | |
| 4802 | 4050 | } |
| 4051 | + | |
| 4052 | + // Cache embeddings for future use (but note: this now includes content and role restrictions) | |
| 4053 | + wp_cache_set($cache_key, $embeddings, 'mxchat_system_prompts', 3600); | |
| 4054 | + } | |
| 4803 | 4055 | |
| 4804 | - foreach ($batch as $row) { | |
| 4805 | - $database_embedding = $row->embedding_vector | |
| 4806 | - ? unserialize($row->embedding_vector, ['allowed_classes' => false]) | |
| 4807 | - : null; | |
| 4056 | + // Get knowledge manager instance for role checking | |
| 4057 | + $knowledge_manager = MxChat_Knowledge_Manager::get_instance(); | |
| 4808 | 4058 | |
| 4809 | - if (!is_array($database_embedding) || !is_array($user_embedding)) { | |
| 4810 | - unset($database_embedding); | |
| 4811 | - continue; | |
| 4812 | - } | |
| 4813 | - | |
| 4059 | + // Get base similarity threshold from bot options or default options | |
| 4060 | + $similarity_threshold = isset($current_options['similarity_threshold']) | |
| 4061 | + ? ((int) $current_options['similarity_threshold']) / 100 | |
| 4062 | + : 0.35; | |
| 4063 | + | |
| 4064 | + $this->last_similarity_analysis['threshold_used'] = $similarity_threshold; | |
| 4065 | + | |
| 4066 | + // Calculate similarities and build results array | |
| 4067 | + $all_similarities = []; | |
| 4068 | + $relevant_results = []; | |
| 4069 | + | |
| 4070 | + foreach ($embeddings as $embedding) { | |
| 4071 | + $database_embedding = $embedding->embedding_vector | |
| 4072 | + ? unserialize($embedding->embedding_vector, ['allowed_classes' => false]) | |
| 4073 | + : null; | |
| 4074 | + | |
| 4075 | + if (is_array($database_embedding) && is_array($user_embedding)) { | |
| 4814 | 4076 | $similarity = $this->mxchat_calculate_cosine_similarity($user_embedding, $database_embedding); |
| 4815 | - unset($database_embedding); | |
| 4816 | - | |
| 4817 | - $role_restriction = $row->role_restriction ?? 'public'; | |
| 4077 | + | |
| 4078 | + // Check role access | |
| 4079 | + $role_restriction = $embedding->role_restriction ?? 'public'; | |
| 4818 | 4080 | $has_access = $knowledge_manager->mxchat_user_has_content_access($role_restriction); |
| 4819 | - $source_url = $row->source_url ?? ''; | |
| 4820 | - | |
| 4821 | - // Maintain top 10 display buffer (insert-if-beats-worst) | |
| 4822 | - if (count($top_display) < 10) { | |
| 4823 | - $top_display[] = [ | |
| 4824 | - 'id' => $row->id, | |
| 4825 | - 'similarity' => $similarity, | |
| 4826 | - 'source_url' => $source_url, | |
| 4827 | - 'role_restriction' => $role_restriction, | |
| 4828 | - 'has_access' => $has_access, | |
| 4829 | - ]; | |
| 4830 | - usort($top_display, function ($a, $b) { | |
| 4831 | - return $b['similarity'] <=> $a['similarity']; | |
| 4832 | - }); | |
| 4833 | - } elseif ($similarity > $top_display[9]['similarity']) { | |
| 4834 | - $top_display[9] = [ | |
| 4835 | - 'id' => $row->id, | |
| 4836 | - 'similarity' => $similarity, | |
| 4837 | - 'source_url' => $source_url, | |
| 4838 | - 'role_restriction' => $role_restriction, | |
| 4839 | - 'has_access' => $has_access, | |
| 4840 | - ]; | |
| 4841 | - usort($top_display, function ($a, $b) { | |
| 4842 | - return $b['similarity'] <=> $a['similarity']; | |
| 4843 | - }); | |
| 4081 | + | |
| 4082 | + // Store ALL similarities for testing (top 10) | |
| 4083 | + $source_display = ''; | |
| 4084 | + if (!empty($embedding->source_url) && $embedding->source_url !== '#') { | |
| 4085 | + $source_display = $embedding->source_url; | |
| 4086 | + } else { | |
| 4087 | + $content_preview = strip_tags($embedding->article_content ?? ''); | |
| 4088 | + $content_preview = preg_replace('/\s+/', ' ', $content_preview); | |
| 4089 | + $source_display = substr(trim($content_preview), 0, 50) . '...'; | |
| 4844 | 4090 | } |
| 4845 | - | |
| 4846 | - // Track candidates for context assembly (above threshold + has access) | |
| 4091 | + | |
| 4092 | + $all_similarities[] = [ | |
| 4093 | + 'document_id' => $embedding->id, | |
| 4094 | + 'similarity' => $similarity, | |
| 4095 | + 'similarity_percentage' => round($similarity * 100, 2), | |
| 4096 | + 'above_threshold' => $similarity >= $similarity_threshold, | |
| 4097 | + 'source_display' => $source_display, | |
| 4098 | + 'content_preview' => substr(strip_tags($embedding->article_content ?? ''), 0, 100) . '...', | |
| 4099 | + 'used_for_context' => false, // Initialize as false, we'll update this later | |
| 4100 | + 'role_restriction' => $role_restriction, | |
| 4101 | + 'has_access' => $has_access, | |
| 4102 | + 'filtered_out' => !$has_access | |
| 4103 | + ]; | |
| 4104 | + | |
| 4105 | + // Only consider results above threshold AND with access for actual content retrieval | |
| 4847 | 4106 | if ($similarity >= $similarity_threshold && $has_access) { |
| 4848 | - $candidates[] = [ | |
| 4849 | - 'id' => $row->id, | |
| 4850 | - 'similarity' => $similarity, | |
| 4851 | - 'source_url' => $source_url, | |
| 4107 | + $relevant_results[] = [ | |
| 4108 | + 'id' => $embedding->id, | |
| 4109 | + 'similarity' => $similarity | |
| 4852 | 4110 | ]; |
| 4853 | 4111 | } |
| 4854 | - | |
| 4855 | - $total_checked++; | |
| 4856 | 4112 | } |
| 4857 | - | |
| 4858 | - unset($batch); | |
| 4859 | - | |
| 4860 | - // Trim candidates periodically to cap memory during long scans | |
| 4861 | - if (count($candidates) > $max_candidates) { | |
| 4862 | - usort($candidates, function ($a, $b) { | |
| 4863 | - return $b['similarity'] <=> $a['similarity']; | |
| 4864 | - }); | |
| 4865 | - $candidates = array_slice($candidates, 0, $max_candidates); | |
| 4866 | - } | |
| 4867 | - | |
| 4868 | - $offset += $batch_size; | |
| 4869 | - } while (true); | |
| 4870 | - | |
| 4871 | - if ($total_checked === 0) { | |
| 4872 | - $this->current_valid_urls = []; | |
| 4873 | - return ''; | |
| 4113 | + | |
| 4114 | + unset($database_embedding); | |
| 4874 | 4115 | } |
| 4875 | 4116 | |
| 4876 | - // Final candidates sort (best first) | |
| 4877 | - if (count($candidates) > 1) { | |
| 4878 | - usort($candidates, function ($a, $b) { | |
| 4879 | - return $b['similarity'] <=> $a['similarity']; | |
| 4880 | - }); | |
| 4881 | - } | |
| 4882 | - | |
| 4883 | - // ===== PHASE 2: FETCH ARTICLE CONTENT ONLY FOR WINNERS ===== | |
| 4884 | - // Gather unique IDs we actually need (top_display + candidates) and pull | |
| 4885 | - // article_content in bounded IN() batches. This avoids loading content for | |
| 4886 | - // every row during the similarity scan. | |
| 4887 | - $needed_ids = []; | |
| 4888 | - foreach ($top_display as $item) { | |
| 4889 | - $needed_ids[$item['id']] = true; | |
| 4890 | - } | |
| 4891 | - foreach ($candidates as $item) { | |
| 4892 | - $needed_ids[$item['id']] = true; | |
| 4893 | - } | |
| 4894 | - $needed_ids = array_keys($needed_ids); | |
| 4895 | - | |
| 4896 | - $content_map = []; | |
| 4897 | - if (!empty($needed_ids)) { | |
| 4898 | - foreach (array_chunk($needed_ids, 250) as $chunk_ids) { | |
| 4899 | - $placeholders = implode(',', array_fill(0, count($chunk_ids), '%d')); | |
| 4900 | - $rows = $wpdb->get_results($wpdb->prepare( | |
| 4901 | - "SELECT id, article_content FROM {$system_prompt_table} WHERE id IN ($placeholders)", | |
| 4902 | - ...$chunk_ids | |
| 4903 | - )); | |
| 4904 | - foreach ($rows as $r) { | |
| 4905 | - $content_map[$r->id] = $r->article_content; | |
| 4906 | - } | |
| 4907 | - unset($rows); | |
| 4908 | - } | |
| 4909 | - } | |
| 4910 | - | |
| 4911 | - // Build the all_similarities display array from the top 10 | |
| 4912 | - $all_similarities = []; | |
| 4913 | - foreach ($top_display as $item) { | |
| 4914 | - $article_content_for_parse = $content_map[$item['id']] ?? ''; | |
| 4915 | - $parsed_for_display = MxChat_Chunker::parse_stored_chunk($article_content_for_parse); | |
| 4916 | - $is_chunk = $parsed_for_display['is_chunked']; | |
| 4917 | - $chunk_meta = $parsed_for_display['metadata']; | |
| 4918 | - | |
| 4919 | - if (!empty($item['source_url']) && $item['source_url'] !== '#') { | |
| 4920 | - $source_display = $item['source_url']; | |
| 4921 | - } else { | |
| 4922 | - $content_preview = strip_tags($article_content_for_parse); | |
| 4923 | - $content_preview = preg_replace('/\s+/', ' ', $content_preview); | |
| 4924 | - $source_display = substr(trim($content_preview), 0, 50) . '...'; | |
| 4925 | - } | |
| 4926 | - | |
| 4927 | - $all_similarities[] = [ | |
| 4928 | - 'document_id' => $item['id'], | |
| 4929 | - 'similarity' => $item['similarity'], | |
| 4930 | - 'similarity_percentage' => round($item['similarity'] * 100, 2), | |
| 4931 | - 'above_threshold' => $item['similarity'] >= $similarity_threshold, | |
| 4932 | - 'source_display' => $source_display, | |
| 4933 | - 'content_preview' => substr(strip_tags($parsed_for_display['text'] ?? ''), 0, 100) . '...', | |
| 4934 | - 'used_for_context' => false, | |
| 4935 | - 'role_restriction' => $item['role_restriction'], | |
| 4936 | - 'has_access' => $item['has_access'], | |
| 4937 | - 'filtered_out' => !$item['has_access'], | |
| 4938 | - 'is_chunk' => $is_chunk, | |
| 4939 | - 'chunk_index' => $is_chunk ? ($chunk_meta['chunk_index'] ?? 0) : null, | |
| 4940 | - 'total_chunks' => $is_chunk ? ($chunk_meta['total_chunks'] ?? 1) : null | |
| 4941 | - ]; | |
| 4942 | - } | |
| 4943 | - | |
| 4944 | - // Build url_groups from candidates for chunk reassembly | |
| 4945 | - $url_groups = array(); | |
| 4946 | - foreach ($candidates as $cand) { | |
| 4947 | - $article_content = $content_map[$cand['id']] ?? ''; | |
| 4948 | - $parsed = MxChat_Chunker::parse_stored_chunk($article_content); | |
| 4949 | - $is_chunked = $parsed['is_chunked']; | |
| 4950 | - $chunk_index = $parsed['metadata']['chunk_index'] ?? 0; | |
| 4951 | - $text_content = $parsed['text']; | |
| 4952 | - | |
| 4953 | - $source_url = $cand['source_url']; | |
| 4954 | - $group_key = !empty($source_url) ? $source_url : '_manual_' . $cand['id']; | |
| 4955 | - | |
| 4956 | - if (!isset($url_groups[$group_key])) { | |
| 4957 | - $url_groups[$group_key] = array( | |
| 4958 | - 'source_url' => $source_url, | |
| 4959 | - 'best_score' => 0, | |
| 4960 | - 'is_chunked' => $is_chunked, | |
| 4961 | - 'chunks' => array(), | |
| 4962 | - 'single_text' => '', | |
| 4963 | - 'single_id' => null | |
| 4964 | - ); | |
| 4965 | - } | |
| 4966 | - | |
| 4967 | - if ($cand['similarity'] > $url_groups[$group_key]['best_score']) { | |
| 4968 | - $url_groups[$group_key]['best_score'] = $cand['similarity']; | |
| 4969 | - } | |
| 4970 | - | |
| 4971 | - if ($is_chunked) { | |
| 4972 | - $url_groups[$group_key]['is_chunked'] = true; | |
| 4973 | - $url_groups[$group_key]['chunks'][] = array( | |
| 4974 | - 'id' => $cand['id'], | |
| 4975 | - 'score' => $cand['similarity'], | |
| 4976 | - 'chunk_index' => $chunk_index, | |
| 4977 | - 'text' => $text_content | |
| 4978 | - ); | |
| 4979 | - } else { | |
| 4980 | - $url_groups[$group_key]['single_text'] = $text_content; | |
| 4981 | - $url_groups[$group_key]['single_id'] = $cand['id']; | |
| 4982 | - } | |
| 4983 | - } | |
| 4984 | - | |
| 4985 | 4117 | // Sort ALL similarities for testing display (highest first) |
| 4986 | 4118 | usort($all_similarities, function ($a, $b) { |
| 4987 | 4119 | return $b['similarity'] <=> $a['similarity']; |
| 4988 | 4120 | }); |
| 4989 | - | |
| 4990 | - // Sort URL groups by best score (highest first) | |
| 4991 | - uasort($url_groups, function($a, $b) { | |
| 4992 | - return $b['best_score'] <=> $a['best_score']; | |
| 4121 | + | |
| 4122 | + // Sort relevant results by similarity (highest first) | |
| 4123 | + usort($relevant_results, function ($a, $b) { | |
| 4124 | + return $b['similarity'] <=> $a['similarity']; | |
| 4993 | 4125 | }); |
| 4994 | - | |
| 4995 | - // Get RAG sources limit from options (default 6, min 3, max 10) | |
| 4996 | - $rag_sources_limit = isset($current_options['rag_sources_limit']) ? intval($current_options['rag_sources_limit']) : 3; | |
| 4997 | - if ($rag_sources_limit < 3) $rag_sources_limit = 3; | |
| 4998 | - if ($rag_sources_limit > 10) $rag_sources_limit = 10; | |
| 4999 | - | |
| 5000 | - // Take top N unique URLs based on user setting | |
| 5001 | - $top_urls = array_slice($url_groups, 0, $rag_sources_limit, true); | |
| 5002 | - | |
| 5003 | - // Track which document IDs are used for context | |
| 4126 | + | |
| 4127 | + // Get top 5 results for actual content (standard approach) | |
| 4128 | + $top_results = array_slice($relevant_results, 0, 3); | |
| 4129 | + | |
| 4130 | + // NOW mark which documents are actually used for context | |
| 5004 | 4131 | $used_document_ids = []; |
| 5005 | - foreach ($top_urls as $group) { | |
| 5006 | - if ($group['is_chunked']) { | |
| 5007 | - foreach ($group['chunks'] as $chunk) { | |
| 5008 | - $used_document_ids[] = $chunk['id']; | |
| 5009 | - } | |
| 5010 | - } elseif ($group['single_id']) { | |
| 5011 | - $used_document_ids[] = $group['single_id']; | |
| 5012 | - } | |
| 4132 | + foreach ($top_results as $result) { | |
| 4133 | + $used_document_ids[] = $result['id']; | |
| 5013 | 4134 | } |
| 5014 | - | |
| 4135 | + | |
| 5015 | 4136 | // Update the all_similarities array to mark which were actually used |
| 5016 | 4137 | foreach ($all_similarities as &$similarity_item) { |
| 5017 | 4138 | $similarity_item['used_for_context'] = in_array($similarity_item['document_id'], $used_document_ids); |
| 5018 | 4139 | } |
| 5019 | - | |
| 5020 | - // Store top 10 for testing panel | |
| 4140 | + | |
| 4141 | + // Store top 10 for testing panel (now with correct used_for_context flags and role info) | |
| 5021 | 4142 | $this->last_similarity_analysis['top_matches'] = array_slice($all_similarities, 0, 10); |
| 5022 | - $this->last_similarity_analysis['total_checked'] = $total_checked; | |
| 5023 | - | |
| 4143 | + $this->last_similarity_analysis['total_checked'] = count($embeddings); | |
| 4144 | + | |
| 5024 | 4145 | // Initialize final content |
| 5025 | 4146 | $content = ''; |
| 5026 | - $matches_used = 0; | |
| 5027 | - $total_chunks_used = 0; | |
| 5028 | - $max_total_chunks = isset($current_options['rag_chunks_limit']) ? intval($current_options['rag_chunks_limit']) : 15; | |
| 5029 | - if ($max_total_chunks < 8) $max_total_chunks = 8; | |
| 5030 | - if ($max_total_chunks > 20) $max_total_chunks = 20; | |
| 5031 | - $max_chunks_per_source = 5; // Cap per individual source to limit token usage | |
| 5032 | - | |
| 5033 | - // Check if citation links are enabled (default to 'on' for backwards compatibility) | |
| 5034 | - // Use fresh options to ensure we get the latest setting value | |
| 5035 | - $fresh_options = get_option('mxchat_options', []); | |
| 5036 | - $citation_links_enabled = isset($fresh_options['citation_links_toggle']) ? ($fresh_options['citation_links_toggle'] === 'on') : true; | |
| 5037 | - | |
| 5038 | - // Build content from top sources | |
| 5039 | - foreach ($top_urls as $group_key => $group) { | |
| 5040 | - $source_url = $group['source_url']; // Use actual source_url, not the group key | |
| 5041 | - | |
| 5042 | - // Stop if we've hit the total chunk limit | |
| 5043 | - if ($total_chunks_used >= $max_total_chunks) { | |
| 5044 | - break; | |
| 4147 | + | |
| 4148 | + // Track document IDs to avoid duplicates | |
| 4149 | + $added_document_ids = []; | |
| 4150 | + | |
| 4151 | + // Fetch and format content for each selected result | |
| 4152 | + foreach ($top_results as $index => $result) { | |
| 4153 | + if (in_array($result['id'], $added_document_ids)) { | |
| 4154 | + continue; | |
| 5045 | 4155 | } |
| 5046 | - | |
| 5047 | - $full_text = ''; | |
| 5048 | - $chunks_in_this_source = 1; // Default for non-chunked content | |
| 5049 | - | |
| 5050 | - if ($group['is_chunked']) { | |
| 5051 | - // Calculate how many chunks we can still use (respect both total and per-source caps) | |
| 5052 | - $chunks_remaining = min($max_chunks_per_source, $max_total_chunks - $total_chunks_used); | |
| 5053 | - | |
| 5054 | - // Fetch chunks for this URL with limit | |
| 5055 | - $full_text = $this->reassemble_chunks_from_wordpress($source_url, $chunks_remaining, $chunks_in_this_source); | |
| 5056 | - | |
| 5057 | - // If fetching all chunks fails, fall back to matched chunks | |
| 5058 | - if (empty($full_text)) { | |
| 5059 | - // Sort matched chunks by index and concatenate | |
| 5060 | - usort($group['chunks'], function($a, $b) { | |
| 5061 | - return $a['chunk_index'] <=> $b['chunk_index']; | |
| 5062 | - }); | |
| 5063 | - | |
| 5064 | - $chunk_texts = array(); | |
| 5065 | - $chunks_in_this_source = 0; | |
| 5066 | - foreach ($group['chunks'] as $chunk) { | |
| 5067 | - if ($total_chunks_used + $chunks_in_this_source >= $max_total_chunks) { | |
| 5068 | - break; | |
| 4156 | + | |
| 4157 | + $chunk_content = $this->fetch_content_with_product_links($result['id']); | |
| 4158 | + $added_document_ids[] = $result['id']; | |
| 4159 | + | |
| 4160 | + // NEW: Extract source_url from database for this result | |
| 4161 | + $source_url = $wpdb->get_var($wpdb->prepare( | |
| 4162 | + "SELECT source_url FROM {$system_prompt_table} WHERE id = %d", | |
| 4163 | + $result['id'] | |
| 4164 | + )); | |
| 4165 | + | |
| 4166 | + // NEW: Add source_url to valid URLs list if it exists and is not empty/placeholder | |
| 4167 | + if (!empty($source_url) && $source_url !== '#') { | |
| 4168 | + $valid_urls[] = $source_url; | |
| 4169 | + } | |
| 4170 | + | |
| 4171 | + // NEW: Extract any URLs from the article content itself | |
| 4172 | + preg_match_all( | |
| 4173 | + '#\bhttps?://[^\s<>"\']+#i', | |
| 4174 | + $chunk_content, | |
| 4175 | + $content_urls | |
| 4176 | + ); | |
| 4177 | + if (!empty($content_urls[0])) { | |
| 4178 | + $valid_urls = array_merge($valid_urls, $content_urls[0]); | |
| 4179 | + } | |
| 4180 | + | |
| 4181 | + $content .= "## Reference " . ($index + 1) . " ##\n"; | |
| 4182 | + $content .= $chunk_content . "\n\n"; | |
| 4183 | + | |
| 4184 | + // PDF surrounding pages logic (unchanged) | |
| 4185 | + if (strpos($chunk_content, '{"document_type":"pdf"') !== false) { | |
| 4186 | + $surrounding_content = $wpdb->get_results($wpdb->prepare( | |
| 4187 | + "SELECT id, article_content, role_restriction FROM {$system_prompt_table} | |
| 4188 | + WHERE id IN ( | |
| 4189 | + (SELECT id FROM {$system_prompt_table} WHERE id < %d ORDER BY id DESC LIMIT 1), | |
| 4190 | + (SELECT id FROM {$system_prompt_table} WHERE id > %d ORDER BY id ASC LIMIT 1) | |
| 4191 | + )", | |
| 4192 | + $result['id'], | |
| 4193 | + $result['id'] | |
| 4194 | + )); | |
| 4195 | + | |
| 4196 | + // Check role access for surrounding content too | |
| 4197 | + if (!empty($surrounding_content[0])) { | |
| 4198 | + $surrounding_role = $surrounding_content[0]->role_restriction ?? 'public'; | |
| 4199 | + if ($knowledge_manager->mxchat_user_has_content_access($surrounding_role)) { | |
| 4200 | + // NEW: Extract URLs from surrounding content too | |
| 4201 | + preg_match_all( | |
| 4202 | + '#\bhttps?://[^\s<>"\']+#i', | |
| 4203 | + $surrounding_content[0]->article_content, | |
| 4204 | + $surrounding_urls | |
| 4205 | + ); | |
| 4206 | + if (!empty($surrounding_urls[0])) { | |
| 4207 | + $valid_urls = array_merge($valid_urls, $surrounding_urls[0]); | |
| 5069 | 4208 | } |
| 5070 | - $chunk_texts[] = $chunk['text']; | |
| 5071 | - $chunks_in_this_source++; | |
| 4209 | + | |
| 4210 | + $content .= "## Related Content ##\n"; | |
| 4211 | + $content .= $surrounding_content[0]->article_content . "\n\n"; | |
| 4212 | + $added_document_ids[] = $surrounding_content[0]->id; | |
| 5072 | 4213 | } |
| 5073 | - $full_text = implode("\n\n", $chunk_texts); | |
| 5074 | 4214 | } |
| 5075 | - } else { | |
| 5076 | - $full_text = $group['single_text']; | |
| 5077 | - $chunks_in_this_source = 1; | |
| 5078 | - } | |
| 5079 | - | |
| 5080 | - if (!empty($full_text)) { | |
| 5081 | - // Strip URLs from content if citation links are disabled | |
| 5082 | - if (!$citation_links_enabled) { | |
| 5083 | - $full_text = preg_replace('#\bhttps?://[^\s<>"\']+#i', '', $full_text); | |
| 5084 | - $full_text = preg_replace('/\s+/', ' ', trim($full_text)); // Clean up extra spaces | |
| 5085 | - } | |
| 5086 | - | |
| 5087 | - // Use numbered reference for URL-based entries, plain info label for manual entries | |
| 5088 | - // Manual entries are stored with an internal mxchat:// placeholder URL — never expose them as citations | |
| 5089 | - if (!empty($source_url) && $source_url !== '#' && strpos($source_url, 'mxchat://') !== 0) { | |
| 5090 | - $matches_used++; | |
| 5091 | - $content .= "## Reference " . $matches_used . " ##\n"; | |
| 5092 | - $content .= $full_text . "\n\n"; | |
| 5093 | - | |
| 5094 | - // Only include citation URLs if citation links are enabled | |
| 5095 | - if ($citation_links_enabled) { | |
| 5096 | - $valid_urls[] = $source_url; | |
| 5097 | - $content .= "URL: " . $source_url . "\n\n"; | |
| 4215 | + | |
| 4216 | + if (!empty($surrounding_content[1])) { | |
| 4217 | + $surrounding_role = $surrounding_content[1]->role_restriction ?? 'public'; | |
| 4218 | + if ($knowledge_manager->mxchat_user_has_content_access($surrounding_role)) { | |
| 4219 | + // NEW: Extract URLs from surrounding content too | |
| 4220 | + preg_match_all( | |
| 4221 | + '#\bhttps?://[^\s<>"\']+#i', | |
| 4222 | + $surrounding_content[1]->article_content, | |
| 4223 | + $surrounding_urls | |
| 4224 | + ); | |
| 4225 | + if (!empty($surrounding_urls[0])) { | |
| 4226 | + $valid_urls = array_merge($valid_urls, $surrounding_urls[0]); | |
| 4227 | + } | |
| 4228 | + | |
| 4229 | + $content .= "## Related Content ##\n"; | |
| 4230 | + $content .= $surrounding_content[1]->article_content . "\n\n"; | |
| 4231 | + $added_document_ids[] = $surrounding_content[1]->id; | |
| 5098 | 4232 | } |
| 5099 | - } else { | |
| 5100 | - // Manual entry — no reference number, no citation | |
| 5101 | - $content .= "## Information ##\n"; | |
| 5102 | - $content .= $full_text . "\n\n"; | |
| 5103 | 4233 | } |
| 5104 | - | |
| 5105 | - // Extract any URLs from the text content itself (only if citation links enabled) | |
| 5106 | - if ($citation_links_enabled) { | |
| 5107 | - preg_match_all( | |
| 5108 | - '#\bhttps?://[^\s<>"\']+#i', | |
| 5109 | - $full_text, | |
| 5110 | - $content_urls | |
| 5111 | - ); | |
| 5112 | - if (!empty($content_urls[0])) { | |
| 5113 | - $valid_urls = array_merge($valid_urls, $content_urls[0]); | |
| 5114 | - } | |
| 5115 | - } | |
| 5116 | - | |
| 5117 | - $total_chunks_used += $chunks_in_this_source; | |
| 5118 | 4234 | } |
| 5119 | 4235 | } |
| 5120 | - | |
| 4236 | + | |
| 5121 | 4237 | // NEW: Store unique valid URLs for validation |
| 5122 | 4238 | $this->current_valid_urls = array_unique($valid_urls); |
| 5123 | - | |
| 5124 | - // Store sources and chunks counts for testing/transcript display | |
| 5125 | - $this->last_similarity_analysis['sources_used'] = $matches_used; | |
| 5126 | - $this->last_similarity_analysis['total_chunks_used'] = $total_chunks_used; | |
| 5127 | - | |
| 5128 | - // Allow add-ons to act on similarity results (e.g. WooCommerce product card display) | |
| 5129 | - do_action('mxchat_similarity_results', $this->last_similarity_analysis['top_matches'], $bot_id); | |
| 5130 | - | |
| 4239 | + | |
| 5131 | 4240 | // Add response guidelines |
| 5132 | - if (empty($top_urls)) { | |
| 4241 | + if (empty($top_results)) { | |
| 5133 | 4242 | $content = "No reference information was found for this query.\n\n"; |
| 5134 | 4243 | } else { |
| 5135 | - // Build response guidelines based on citation links setting | |
| 5136 | 4244 | $content .= "\n## Response Guidelines ##\n" . |
| 5137 | 4245 | "You are an AI Chatbot. Answer naturally and helpfully using only the information from the references above. " . |
| 5138 | 4246 | "Be conversational and friendly, but never mention your knowledge base or training data. " . |
| 5139 | 4247 | "If you don't have specific information or are uncertain about any details, it's always " . |
| 5140 | 4248 | "better to honestly say you don't know rather than making up or guessing at answers. " . |
| 5141 | - "When information is incomplete, let them know you are unsure.\n\n"; | |
| 5142 | - | |
| 5143 | - // Only add hyperlink instructions if citation links are enabled | |
| 5144 | - if ($citation_links_enabled) { | |
| 5145 | - $content .= "CRITICAL: When creating hyperlinks, always use proper markdown format with descriptive text: " . | |
| 5146 | - "[descriptive text](url). NEVER use empty brackets like [](url). The text in brackets must describe what the link is about. " . | |
| 5147 | - "Only cite references that have a URL. Do not cite or add source labels to Information sections that have no URL."; | |
| 5148 | - } else { | |
| 5149 | - $content .= "IMPORTANT: Do not include any citation links, source URLs, or hyperlinks in your responses. " . | |
| 5150 | - "Simply provide helpful answers based on the reference information without citing sources."; | |
| 5151 | - } | |
| 4249 | + "When information is incomplete, let them know you are unsure.\n\n" . | |
| 4250 | + "CRITICAL: When creating hyperlinks, always use proper markdown format with descriptive text: " . | |
| 4251 | + "[descriptive text](url). NEVER use empty brackets like [](url). The text in brackets must describe what the link is about."; | |
| 5152 | 4252 | } |
| 5153 | 4253 | |
| 5154 | 4254 | return trim($content); |
| 5155 | 4255 | } |
| 5156 | 4256 | |
| 5157 | -/** | |
| 5158 | - * Fetch and reassemble chunks for a URL from WordPress database | |
| 5159 | - * | |
| 5160 | - * @param string $source_url The source URL to fetch chunks for | |
| 5161 | - * @param int $max_chunks Maximum number of chunks to return (0 = unlimited) | |
| 5162 | - * @param int &$chunk_count Reference to store the actual number of chunks returned | |
| 5163 | - * @return string Reassembled content from chunks | |
| 5164 | - */ | |
| 5165 | -private function reassemble_chunks_from_wordpress($source_url, $max_chunks = 0, &$chunk_count = 0) { | |
| 5166 | - global $wpdb; | |
| 5167 | - $table = $wpdb->prefix . 'mxchat_system_prompt_content'; | |
| 5168 | - | |
| 5169 | - // Fetch all rows with this source_url | |
| 5170 | - $rows = $wpdb->get_results($wpdb->prepare( | |
| 5171 | - "SELECT article_content FROM {$table} | |
| 5172 | - WHERE source_url = %s | |
| 5173 | - ORDER BY id ASC", | |
| 5174 | - $source_url | |
| 5175 | - )); | |
| 5176 | - | |
| 5177 | - if (empty($rows)) { | |
| 5178 | - $chunk_count = 0; | |
| 5179 | - return ''; | |
| 5180 | - } | |
| 5181 | - | |
| 5182 | - // Parse and sort chunks by index | |
| 5183 | - $chunks = array(); | |
| 5184 | - foreach ($rows as $row) { | |
| 5185 | - $parsed = MxChat_Chunker::parse_stored_chunk($row->article_content); | |
| 5186 | - | |
| 5187 | - if ($parsed['is_chunked']) { | |
| 5188 | - $chunk_index = $parsed['metadata']['chunk_index'] ?? 0; | |
| 5189 | - $chunks[$chunk_index] = $parsed['text']; | |
| 5190 | - } else { | |
| 5191 | - // Non-chunked content - just return it | |
| 5192 | - $chunks[] = $parsed['text']; | |
| 5193 | - } | |
| 5194 | - } | |
| 5195 | - | |
| 5196 | - // Sort by chunk index | |
| 5197 | - ksort($chunks); | |
| 5198 | - | |
| 5199 | - // Apply chunk limit if specified | |
| 5200 | - if ($max_chunks > 0 && count($chunks) > $max_chunks) { | |
| 5201 | - $chunks = array_slice($chunks, 0, $max_chunks, true); | |
| 5202 | - } | |
| 5203 | - | |
| 5204 | - // Store actual chunk count | |
| 5205 | - $chunk_count = count($chunks); | |
| 5206 | - | |
| 5207 | - // Reassemble content | |
| 5208 | - return implode("\n\n", $chunks); | |
| 5209 | -} | |
| 5210 | - | |
| 5211 | 4257 | private function find_relevant_content_pinecone($user_embedding, $bot_id = 'default', $bot_config = null) { |
| 5212 | 4258 | global $wpdb; |
| 5213 | 4259 | |
| 5214 | - //error_log("MXCHAT DEBUG: find_relevant_content_pinecone called"); | |
| 5215 | - //error_log(" - bot_id: " . $bot_id); | |
| 5216 | - //error_log(" - user_embedding is array: " . (is_array($user_embedding) ? 'yes' : 'no')); | |
| 5217 | - //error_log(" - user_embedding count: " . (is_array($user_embedding) ? count($user_embedding) : 'N/A')); | |
| 4260 | + error_log("MXCHAT DEBUG: find_relevant_content_pinecone called"); | |
| 4261 | + error_log(" - bot_id: " . $bot_id); | |
| 4262 | + error_log(" - user_embedding is array: " . (is_array($user_embedding) ? 'yes' : 'no')); | |
| 4263 | + error_log(" - user_embedding count: " . (is_array($user_embedding) ? count($user_embedding) : 'N/A')); | |
| 5218 | 4264 | |
| 5219 | 4265 | // Use bot-specific config or fall back to default |
| 5220 | 4266 | if ($bot_config === null) { |
| 5221 | 4267 | $bot_config = $this->get_bot_pinecone_config($bot_id); |
| @@ -5224,12 +4270,12 @@ | ||
| 5224 | 4270 | $api_key = $bot_config['api_key'] ?? ''; |
| 5225 | 4271 | $host = $bot_config['host'] ?? ''; |
| 5226 | 4272 | $namespace = $bot_config['namespace'] ?? ''; |
| 5227 | 4273 | |
| 5228 | - //error_log("MXCHAT DEBUG: Pinecone query parameters:"); | |
| 5229 | - //error_log(" - API Key: " . (empty($api_key) ? 'EMPTY - ERROR!' : 'Present (length: ' . strlen($api_key) . ')')); | |
| 5230 | - //error_log(" - Host: " . (empty($host) ? 'EMPTY - ERROR!' : $host)); | |
| 5231 | - //error_log(" - Namespace: " . (empty($namespace) ? 'EMPTY (will use default)' : $namespace)); | |
| 4274 | + error_log("MXCHAT DEBUG: Pinecone query parameters:"); | |
| 4275 | + error_log(" - API Key: " . (empty($api_key) ? 'EMPTY - ERROR!' : 'Present (length: ' . strlen($api_key) . ')')); | |
| 4276 | + error_log(" - Host: " . (empty($host) ? 'EMPTY - ERROR!' : $host)); | |
| 4277 | + error_log(" - Namespace: " . (empty($namespace) ? 'EMPTY (will use default)' : $namespace)); | |
| 5232 | 4278 | |
| 5233 | 4279 | // Initialize similarity analysis storage |
| 5234 | 4280 | $this->last_similarity_analysis = [ |
| 5235 | 4281 | 'knowledge_base_type' => 'Pinecone', |
| @@ -5243,11 +4289,11 @@ | ||
| 5243 | 4289 | // NEW: Initialize valid URLs array |
| 5244 | 4290 | $valid_urls = []; |
| 5245 | 4291 | |
| 5246 | 4292 | if (empty($host) || empty($api_key)) { |
| 5247 | - //error_log("MXCHAT DEBUG ERROR: Missing Pinecone host or API key!"); | |
| 5248 | - //error_log(" - Host empty: " . (empty($host) ? 'YES' : 'NO')); | |
| 5249 | - //error_log(" - API key empty: " . (empty($api_key) ? 'YES' : 'NO')); | |
| 4293 | + error_log("MXCHAT DEBUG ERROR: Missing Pinecone host or API key!"); | |
| 4294 | + error_log(" - Host empty: " . (empty($host) ? 'YES' : 'NO')); | |
| 4295 | + error_log(" - API key empty: " . (empty($api_key) ? 'YES' : 'NO')); | |
| 5250 | 4296 | // Store empty array for valid URLs since we can't proceed |
| 5251 | 4297 | $this->current_valid_urls = []; |
| 5252 | 4298 | return ''; |
| 5253 | 4299 | } |
| @@ -5269,9 +4315,9 @@ | ||
| 5269 | 4315 | $api_endpoint = "https://{$host}/query"; |
| 5270 | 4316 | |
| 5271 | 4317 | $request_body = array( |
| 5272 | 4318 | 'vector' => $user_embedding, |
| 5273 | - 'topK' => 50, // Increased for chunked content grouping - need more candidates to find top N unique URLs | |
| 4319 | + 'topK' => 20, // Request more to get good testing data | |
| 5274 | 4320 | 'includeMetadata' => true, |
| 5275 | 4321 | 'includeValues' => true |
| 5276 | 4322 | ); |
| 5277 | 4323 | |
| @@ -5279,11 +4325,11 @@ | ||
| 5279 | 4325 | if (!empty($namespace)) { |
| 5280 | 4326 | $request_body['namespace'] = $namespace; |
| 5281 | 4327 | } |
| 5282 | 4328 | |
| 5283 | - //error_log("MXCHAT DEBUG: About to call Pinecone API"); | |
| 5284 | - //error_log(" - Endpoint: " . $api_endpoint); | |
| 5285 | - //error_log(" - Namespace in request: " . (!empty($namespace) ? $namespace : 'NOT SET')); | |
| 4329 | + error_log("MXCHAT DEBUG: About to call Pinecone API"); | |
| 4330 | + error_log(" - Endpoint: " . $api_endpoint); | |
| 4331 | + error_log(" - Namespace in request: " . (!empty($namespace) ? $namespace : 'NOT SET')); | |
| 5286 | 4332 | |
| 5287 | 4333 | $response = wp_remote_post($api_endpoint, array( |
| 5288 | 4334 | 'headers' => array( |
| 5289 | 4335 | 'Api-Key' => $api_key, |
| @@ -5294,9 +4340,9 @@ | ||
| 5294 | 4340 | 'timeout' => 30 |
| 5295 | 4341 | )); |
| 5296 | 4342 | |
| 5297 | 4343 | if (is_wp_error($response)) { |
| 5298 | - //error_log("MXCHAT DEBUG ERROR: WP Error in Pinecone request: " . $response->get_error_message()); | |
| 4344 | + error_log("MXCHAT DEBUG ERROR: WP Error in Pinecone request: " . $response->get_error_message()); | |
| 5299 | 4345 | // Store empty array for valid URLs |
| 5300 | 4346 | $this->current_valid_urls = []; |
| 5301 | 4347 | return ''; |
| 5302 | 4348 | } |
| @@ -5301,13 +4347,13 @@ | ||
| 5301 | 4347 | return ''; |
| 5302 | 4348 | } |
| 5303 | 4349 | |
| 5304 | 4350 | $response_code = wp_remote_retrieve_response_code($response); |
| 5305 | - //error_log("MXCHAT DEBUG: Pinecone response code: " . $response_code); | |
| 4351 | + error_log("MXCHAT DEBUG: Pinecone response code: " . $response_code); | |
| 5306 | 4352 | |
| 5307 | 4353 | if ($response_code !== 200) { |
| 5308 | 4354 | $response_body = wp_remote_retrieve_body($response); |
| 5309 | - //error_log("MXCHAT DEBUG ERROR: Pinecone API error response: " . substr($response_body, 0, 500)); | |
| 4355 | + error_log("MXCHAT DEBUG ERROR: Pinecone API error response: " . substr($response_body, 0, 500)); | |
| 5310 | 4356 | // Store empty array for valid URLs |
| 5311 | 4357 | $this->current_valid_urls = []; |
| 5312 | 4358 | return ''; |
| 5313 | 4359 | } |
| @@ -5313,42 +4359,42 @@ | ||
| 5313 | 4359 | } |
| 5314 | 4360 | |
| 5315 | 4361 | // ADD DETAILED DEBUG SECTION HERE |
| 5316 | 4362 | $response_body = wp_remote_retrieve_body($response); |
| 5317 | - //error_log("MXCHAT DEBUG: Raw Pinecone response length: " . strlen($response_body)); | |
| 4363 | + error_log("MXCHAT DEBUG: Raw Pinecone response length: " . strlen($response_body)); | |
| 5318 | 4364 | |
| 5319 | 4365 | $results = json_decode($response_body, true); |
| 5320 | 4366 | |
| 5321 | 4367 | if (json_last_error() !== JSON_ERROR_NONE) { |
| 5322 | - //error_log("MXCHAT DEBUG ERROR: JSON decode error: " . json_last_error_msg()); | |
| 5323 | - //error_log("MXCHAT DEBUG: First 500 chars of response: " . substr($response_body, 0, 500)); | |
| 4368 | + error_log("MXCHAT DEBUG ERROR: JSON decode error: " . json_last_error_msg()); | |
| 4369 | + error_log("MXCHAT DEBUG: First 500 chars of response: " . substr($response_body, 0, 500)); | |
| 5324 | 4370 | // Store empty array for valid URLs |
| 5325 | 4371 | $this->current_valid_urls = []; |
| 5326 | 4372 | return ''; |
| 5327 | 4373 | } |
| 5328 | 4374 | |
| 5329 | - //error_log("MXCHAT DEBUG: Pinecone response structure:"); | |
| 5330 | - //error_log(" - Has 'matches' key: " . (isset($results['matches']) ? 'yes' : 'no')); | |
| 5331 | - //error_log(" - Has 'namespace' key: " . (isset($results['namespace']) ? 'yes (' . $results['namespace'] . ')' : 'no')); | |
| 4375 | + error_log("MXCHAT DEBUG: Pinecone response structure:"); | |
| 4376 | + error_log(" - Has 'matches' key: " . (isset($results['matches']) ? 'yes' : 'no')); | |
| 4377 | + error_log(" - Has 'namespace' key: " . (isset($results['namespace']) ? 'yes (' . $results['namespace'] . ')' : 'no')); | |
| 5332 | 4378 | |
| 5333 | 4379 | if (empty($results['matches'])) { |
| 5334 | - //error_log("MXCHAT DEBUG: No matches found in Pinecone response"); | |
| 5335 | - //error_log("MXCHAT DEBUG: Response keys: " . implode(', ', array_keys($results))); | |
| 4380 | + error_log("MXCHAT DEBUG: No matches found in Pinecone response"); | |
| 4381 | + error_log("MXCHAT DEBUG: Response keys: " . implode(', ', array_keys($results))); | |
| 5336 | 4382 | // Store empty array for valid URLs |
| 5337 | 4383 | $this->current_valid_urls = []; |
| 5338 | 4384 | return ''; |
| 5339 | 4385 | } |
| 5340 | 4386 | |
| 5341 | - //error_log("MXCHAT DEBUG: Found " . count($results['matches']) . " matches in Pinecone"); | |
| 4387 | + error_log("MXCHAT DEBUG: Found " . count($results['matches']) . " matches in Pinecone"); | |
| 5342 | 4388 | |
| 5343 | 4389 | // Log first match details for debugging |
| 5344 | 4390 | if (!empty($results['matches'][0])) { |
| 5345 | 4391 | $first_match = $results['matches'][0]; |
| 5346 | - //error_log("MXCHAT DEBUG: First match details:"); | |
| 5347 | - //error_log(" - Score: " . ($first_match['score'] ?? 'no score')); | |
| 5348 | - //error_log(" - Has metadata: " . (isset($first_match['metadata']) ? 'yes' : 'no')); | |
| 4392 | + error_log("MXCHAT DEBUG: First match details:"); | |
| 4393 | + error_log(" - Score: " . ($first_match['score'] ?? 'no score')); | |
| 4394 | + error_log(" - Has metadata: " . (isset($first_match['metadata']) ? 'yes' : 'no')); | |
| 5349 | 4395 | if (isset($first_match['metadata'])) { |
| 5350 | - //error_log(" - Metadata keys: " . implode(', ', array_keys($first_match['metadata']))); | |
| 4396 | + error_log(" - Metadata keys: " . implode(', ', array_keys($first_match['metadata']))); | |
| 5351 | 4397 | } |
| 5352 | 4398 | } |
| 5353 | 4399 | |
| 5354 | 4400 | // Initialize the final content |
| @@ -5354,182 +4400,57 @@ | ||
| 5354 | 4400 | // Initialize the final content |
| 5355 | 4401 | $content = ''; |
| 5356 | 4402 | $matches_used = 0; |
| 5357 | 4403 | $matches_used_for_context = []; |
| 5358 | - $total_chunks_used = 0; | |
| 5359 | - $max_total_chunks = isset($current_options['rag_chunks_limit']) ? intval($current_options['rag_chunks_limit']) : 15; | |
| 5360 | - if ($max_total_chunks < 8) $max_total_chunks = 8; | |
| 5361 | - if ($max_total_chunks > 20) $max_total_chunks = 20; | |
| 5362 | - $max_chunks_per_source = 5; // Cap per individual source to limit token usage | |
| 5363 | - | |
| 5364 | - // Check if citation links are enabled (default to 'on' for backwards compatibility) | |
| 5365 | - // Use fresh options to ensure we get the latest setting value | |
| 5366 | - $fresh_options = get_option('mxchat_options', []); | |
| 5367 | - $citation_links_enabled = isset($fresh_options['citation_links_toggle']) ? ($fresh_options['citation_links_toggle'] === 'on') : true; | |
| 5368 | - | |
| 5369 | - // NEW CHUNKING LOGIC: Group results by source_url for chunk reassembly | |
| 5370 | - $url_groups = array(); | |
| 5371 | - | |
| 4404 | + | |
| 4405 | + // Process each match for actual content generation (lazy role checking) | |
| 5372 | 4406 | foreach ($results['matches'] as $index => $match) { |
| 5373 | 4407 | // Skip if similarity is below threshold |
| 5374 | 4408 | if ($match['score'] < $similarity_threshold) { |
| 5375 | 4409 | continue; |
| 5376 | 4410 | } |
| 5377 | - | |
| 5378 | - $metadata = $match['metadata'] ?? array(); | |
| 5379 | - $source_url = $metadata['source_url'] ?? ''; | |
| 5380 | - $match_id = $match['id'] ?? ''; | |
| 5381 | - | |
| 5382 | - // LAZY ROLE CHECK: Only check role for content we're actually considering | |
| 5383 | - $role_restriction = $this->get_single_vector_role($match_id, $metadata); | |
| 5384 | - $has_access = $knowledge_manager->mxchat_user_has_content_access($role_restriction); | |
| 5385 | - | |
| 5386 | - // Skip if user doesn't have access | |
| 5387 | - if (!$has_access) { | |
| 5388 | - continue; | |
| 5389 | - } | |
| 5390 | - | |
| 5391 | - // Use a unique key for manual entries without a source URL | |
| 5392 | - $group_key = !empty($source_url) ? $source_url : '_manual_' . $match_id; | |
| 5393 | - | |
| 5394 | - // Group by source URL (or unique key for manual entries) | |
| 5395 | - if (!isset($url_groups[$group_key])) { | |
| 5396 | - $url_groups[$group_key] = array( | |
| 5397 | - 'source_url' => $source_url, | |
| 5398 | - 'best_score' => 0, | |
| 5399 | - 'is_chunked' => isset($metadata['is_chunked']) && $metadata['is_chunked'], | |
| 5400 | - 'chunks' => array(), | |
| 5401 | - 'single_text' => '' | |
| 5402 | - ); | |
| 5403 | - } | |
| 5404 | - | |
| 5405 | - // Track best score for this group | |
| 5406 | - if ($match['score'] > $url_groups[$group_key]['best_score']) { | |
| 5407 | - $url_groups[$group_key]['best_score'] = $match['score']; | |
| 5408 | - } | |
| 5409 | - | |
| 5410 | - // Store chunk info or single text | |
| 5411 | - if ($url_groups[$group_key]['is_chunked']) { | |
| 5412 | - $url_groups[$group_key]['chunks'][] = array( | |
| 5413 | - 'id' => $match_id, | |
| 5414 | - 'score' => $match['score'], | |
| 5415 | - 'chunk_index' => $metadata['chunk_index'] ?? 0, | |
| 5416 | - 'text' => $metadata['text'] ?? '' | |
| 5417 | - ); | |
| 5418 | - } else { | |
| 5419 | - // Non-chunked content - just store the text | |
| 5420 | - $url_groups[$group_key]['single_text'] = $metadata['text'] ?? ''; | |
| 5421 | - $url_groups[$group_key]['single_id'] = $match_id; | |
| 5422 | - } | |
| 5423 | - } | |
| 5424 | - | |
| 5425 | - // Sort URL groups by best score (highest first) | |
| 5426 | - uasort($url_groups, function($a, $b) { | |
| 5427 | - return $b['best_score'] <=> $a['best_score']; | |
| 5428 | - }); | |
| 5429 | - | |
| 5430 | - // Get RAG sources limit from options (default 6, min 3, max 10) | |
| 5431 | - $rag_sources_limit = isset($current_options['rag_sources_limit']) ? intval($current_options['rag_sources_limit']) : 3; | |
| 5432 | - if ($rag_sources_limit < 3) $rag_sources_limit = 3; | |
| 5433 | - if ($rag_sources_limit > 10) $rag_sources_limit = 10; | |
| 5434 | - | |
| 5435 | - // Take top N unique URLs based on user setting | |
| 5436 | - $top_urls = array_slice($url_groups, 0, $rag_sources_limit, true); | |
| 5437 | - | |
| 5438 | - // Track which match IDs are actually used for context | |
| 5439 | - foreach ($top_urls as $group) { | |
| 5440 | - if ($group['is_chunked']) { | |
| 5441 | - foreach ($group['chunks'] as $chunk) { | |
| 5442 | - $matches_used_for_context[] = $chunk['id']; | |
| 5443 | - } | |
| 5444 | - } elseif (!empty($group['single_id'])) { | |
| 5445 | - $matches_used_for_context[] = $group['single_id']; | |
| 5446 | - } | |
| 5447 | - } | |
| 5448 | - | |
| 5449 | - // Build content from top sources | |
| 5450 | - foreach ($top_urls as $group_key => $group) { | |
| 5451 | - $source_url = $group['source_url']; // Use actual source_url, not the group key | |
| 5452 | - | |
| 5453 | - // Stop if we've hit the total chunk limit | |
| 5454 | - if ($total_chunks_used >= $max_total_chunks) { | |
| 4411 | + | |
| 4412 | + // Limit to top 3 matches above threshold | |
| 4413 | + if ($matches_used >= 3) { | |
| 5455 | 4414 | break; |
| 5456 | 4415 | } |
| 5457 | - | |
| 5458 | - $full_text = ''; | |
| 5459 | - $chunks_in_this_source = 1; // Default for non-chunked content | |
| 5460 | - | |
| 5461 | - if ($group['is_chunked']) { | |
| 5462 | - // Calculate how many chunks we can still use (respect both total and per-source caps) | |
| 5463 | - $chunks_remaining = min($max_chunks_per_source, $max_total_chunks - $total_chunks_used); | |
| 5464 | - | |
| 5465 | - // Fetch chunks for this URL with limit | |
| 5466 | - $full_text = $this->reassemble_chunks_from_pinecone($source_url, $bot_config, $chunks_remaining, $chunks_in_this_source); | |
| 5467 | - | |
| 5468 | - // If fetching all chunks fails, fall back to matched chunks | |
| 5469 | - if (empty($full_text)) { | |
| 5470 | - // Sort matched chunks by index and concatenate | |
| 5471 | - usort($group['chunks'], function($a, $b) { | |
| 5472 | - return $a['chunk_index'] <=> $b['chunk_index']; | |
| 5473 | - }); | |
| 5474 | - | |
| 5475 | - $chunk_texts = array(); | |
| 5476 | - $chunks_in_this_source = 0; | |
| 5477 | - foreach ($group['chunks'] as $chunk) { | |
| 5478 | - if ($total_chunks_used + $chunks_in_this_source >= $max_total_chunks) { | |
| 5479 | - break; | |
| 5480 | - } | |
| 5481 | - $chunk_texts[] = $chunk['text']; | |
| 5482 | - $chunks_in_this_source++; | |
| 5483 | - } | |
| 5484 | - $full_text = implode("\n\n", $chunk_texts); | |
| 4416 | + | |
| 4417 | + if (!empty($match['metadata']['text'])) { | |
| 4418 | + // LAZY ROLE CHECK: Only check role for content we're actually considering | |
| 4419 | + $match_id = $match['id'] ?? ''; | |
| 4420 | + $role_restriction = $this->get_single_vector_role($match_id, $match['metadata']); | |
| 4421 | + $has_access = $knowledge_manager->mxchat_user_has_content_access($role_restriction); | |
| 4422 | + | |
| 4423 | + // Skip if user doesn't have access | |
| 4424 | + if (!$has_access) { | |
| 4425 | + continue; | |
| 5485 | 4426 | } |
| 5486 | - } else { | |
| 5487 | - $full_text = $group['single_text']; | |
| 5488 | - $chunks_in_this_source = 1; | |
| 5489 | - } | |
| 5490 | - | |
| 5491 | - if (!empty($full_text)) { | |
| 5492 | - // Strip URLs from content if citation links are disabled | |
| 5493 | - if (!$citation_links_enabled) { | |
| 5494 | - $full_text = preg_replace('#\bhttps?://[^\s<>"\']+#i', '', $full_text); | |
| 5495 | - $full_text = preg_replace('/\s+/', ' ', trim($full_text)); // Clean up extra spaces | |
| 4427 | + | |
| 4428 | + // User has access - add to content | |
| 4429 | + $content .= "## Reference " . ($matches_used + 1) . " ##\n"; | |
| 4430 | + $content .= $match['metadata']['text'] . "\n\n"; | |
| 4431 | + | |
| 4432 | + // NEW: Extract source_url from metadata if it exists | |
| 4433 | + if (!empty($match['metadata']['source_url']) && $match['metadata']['source_url'] !== '#') { | |
| 4434 | + $valid_urls[] = $match['metadata']['source_url']; | |
| 4435 | + $content .= "URL: " . $match['metadata']['source_url'] . "\n\n"; | |
| 5496 | 4436 | } |
| 5497 | - | |
| 5498 | - // Use numbered reference for URL-based entries, plain info label for manual entries | |
| 5499 | - // Manual entries are stored with an internal mxchat:// placeholder URL — never expose them as citations | |
| 5500 | - if (!empty($source_url) && $source_url !== '#' && strpos($source_url, 'mxchat://') !== 0) { | |
| 5501 | - $matches_used++; | |
| 5502 | - $content .= "## Reference " . $matches_used . " ##\n"; | |
| 5503 | - $content .= $full_text . "\n\n"; | |
| 5504 | - | |
| 5505 | - // Only include citation URLs if citation links are enabled | |
| 5506 | - if ($citation_links_enabled) { | |
| 5507 | - $valid_urls[] = $source_url; | |
| 5508 | - $content .= "URL: " . $source_url . "\n\n"; | |
| 5509 | - } | |
| 5510 | - } else { | |
| 5511 | - // Manual entry — no reference number, no citation | |
| 5512 | - $content .= "## Information ##\n"; | |
| 5513 | - $content .= $full_text . "\n\n"; | |
| 4437 | + | |
| 4438 | + // NEW: Extract any URLs from the text content itself | |
| 4439 | + preg_match_all( | |
| 4440 | + '#\bhttps?://[^\s<>"\']+#i', | |
| 4441 | + $match['metadata']['text'], | |
| 4442 | + $content_urls | |
| 4443 | + ); | |
| 4444 | + if (!empty($content_urls[0])) { | |
| 4445 | + $valid_urls = array_merge($valid_urls, $content_urls[0]); | |
| 5514 | 4446 | } |
| 5515 | - | |
| 5516 | - // Extract any URLs from the text content itself (only if citation links enabled) | |
| 5517 | - if ($citation_links_enabled) { | |
| 5518 | - preg_match_all( | |
| 5519 | - '#\bhttps?://[^\s<>"\']+#i', | |
| 5520 | - $full_text, | |
| 5521 | - $content_urls | |
| 5522 | - ); | |
| 5523 | - if (!empty($content_urls[0])) { | |
| 5524 | - $valid_urls = array_merge($valid_urls, $content_urls[0]); | |
| 5525 | - } | |
| 5526 | - } | |
| 5527 | - | |
| 5528 | - $total_chunks_used += $chunks_in_this_source; | |
| 4447 | + | |
| 4448 | + $matches_used_for_context[] = $match['id'] ?? $index; | |
| 4449 | + $matches_used++; | |
| 5529 | 4450 | } |
| 5530 | 4451 | } |
| 5531 | - | |
| 4452 | + | |
| 5532 | 4453 | // Process ALL matches for testing data (top 10) - with role checking for testing display |
| 5533 | 4454 | $all_matches = []; |
| 5534 | 4455 | foreach ($results['matches'] as $index => $match) { |
| 5535 | 4456 | if ($index >= 10) break; // Limit to top 10 for testing |
| @@ -5549,19 +4470,9 @@ | ||
| 5549 | 4470 | $source_display = substr(trim($content_preview), 0, 50) . '...'; |
| 5550 | 4471 | } |
| 5551 | 4472 | |
| 5552 | 4473 | $match_id_for_display = $match['id'] ?? $index; |
| 5553 | - | |
| 5554 | - // Check for chunk metadata in Pinecone | |
| 5555 | - $is_chunk = isset($match['metadata']['is_chunked']) && $match['metadata']['is_chunked']; | |
| 5556 | - $chunk_index = isset($match['metadata']['chunk_index']) ? intval($match['metadata']['chunk_index']) : null; | |
| 5557 | - $total_chunks = isset($match['metadata']['total_chunks']) ? intval($match['metadata']['total_chunks']) : null; | |
| 5558 | - | |
| 5559 | - // Also detect chunk from vector ID pattern: {hash}_chunk_{index} | |
| 5560 | - if (!$is_chunk && MxChat_Chunker::is_chunk_vector_id($match_id_for_display)) { | |
| 5561 | - $is_chunk = true; | |
| 5562 | - } | |
| 5563 | - | |
| 4474 | + | |
| 5564 | 4475 | $all_matches[] = [ |
| 5565 | 4476 | 'document_id' => $match_id_for_display, |
| 5566 | 4477 | 'similarity' => $match['score'], |
| 5567 | 4478 | 'similarity_percentage' => round($match['score'] * 100, 2), |
| @@ -5570,12 +4481,9 @@ | ||
| 5570 | 4481 | 'content_preview' => substr(strip_tags($match['metadata']['text'] ?? ''), 0, 100) . '...', |
| 5571 | 4482 | 'used_for_context' => in_array($match_id_for_display, $matches_used_for_context), |
| 5572 | 4483 | 'role_restriction' => $role_restriction, |
| 5573 | 4484 | 'has_access' => $has_access, |
| 5574 | - 'filtered_out' => !$has_access, | |
| 5575 | - 'is_chunk' => $is_chunk, | |
| 5576 | - 'chunk_index' => $chunk_index, | |
| 5577 | - 'total_chunks' => $total_chunks | |
| 4485 | + 'filtered_out' => !$has_access | |
| 5578 | 4486 | ]; |
| 5579 | 4487 | } |
| 5580 | 4488 | |
| 5581 | 4489 | // Store for testing panel |
| @@ -5580,40 +4488,26 @@ | ||
| 5580 | 4488 | |
| 5581 | 4489 | // Store for testing panel |
| 5582 | 4490 | $this->last_similarity_analysis['top_matches'] = $all_matches; |
| 5583 | 4491 | $this->last_similarity_analysis['total_checked'] = count($results['matches']); |
| 5584 | - $this->last_similarity_analysis['sources_used'] = $matches_used; | |
| 5585 | - $this->last_similarity_analysis['total_chunks_used'] = $total_chunks_used; | |
| 5586 | - | |
| 4492 | + | |
| 5587 | 4493 | // NEW: Store unique valid URLs for validation |
| 5588 | 4494 | $this->current_valid_urls = array_unique($valid_urls); |
| 5589 | - | |
| 5590 | - // Allow add-ons to act on similarity results (e.g. WooCommerce product card display) | |
| 5591 | - do_action('mxchat_similarity_results', $this->last_similarity_analysis['top_matches'], $bot_id); | |
| 5592 | - | |
| 4495 | + | |
| 5593 | 4496 | // Add response guidelines |
| 5594 | 4497 | if ($matches_used === 0) { |
| 5595 | 4498 | $content = "No reference information was found for this query.\n\n"; |
| 5596 | 4499 | } else { |
| 5597 | - // Build response guidelines based on citation links setting | |
| 5598 | 4500 | $content .= "\n## Response Guidelines ##\n" . |
| 5599 | 4501 | "You are an AI Chatbot. Answer naturally and helpfully using only the information from the references above. " . |
| 5600 | 4502 | "Be conversational and friendly, but never mention your knowledge base or training data. " . |
| 5601 | 4503 | "If you don't have specific information or are uncertain about any details, it's always " . |
| 5602 | 4504 | "better to honestly say you don't know rather than making up or guessing at answers. " . |
| 5603 | - "When information is incomplete, let them know you are unsure.\n\n"; | |
| 5604 | - | |
| 5605 | - // Only add hyperlink instructions if citation links are enabled | |
| 5606 | - if ($citation_links_enabled) { | |
| 5607 | - $content .= "CRITICAL: When creating hyperlinks, always use proper markdown format with descriptive text: " . | |
| 5608 | - "[descriptive text](url). NEVER use empty brackets like [](url). The text in brackets must describe what the link is about. " . | |
| 5609 | - "Only cite references that have a URL. Do not cite or add source labels to Information sections that have no URL."; | |
| 5610 | - } else { | |
| 5611 | - $content .= "IMPORTANT: Do not include any citation links, source URLs, or hyperlinks in your responses. " . | |
| 5612 | - "Simply provide helpful answers based on the reference information without citing sources."; | |
| 5613 | - } | |
| 4505 | + "When information is incomplete, let them know you are unsure.\n\n" . | |
| 4506 | + "CRITICAL: When creating hyperlinks, always use proper markdown format with descriptive text: " . | |
| 4507 | + "[descriptive text](url). NEVER use empty brackets like [](url). The text in brackets must describe what the link is about."; | |
| 5614 | 4508 | } |
| 5615 | - | |
| 4509 | + | |
| 5616 | 4510 | return trim($content); |
| 5617 | 4511 | } |
| 5618 | 4512 | |
| 5619 | 4513 | /** |
| @@ -5653,518 +4547,12 @@ | ||
| 5653 | 4547 | } |
| 5654 | 4548 | |
| 5655 | 4549 | // Cache individual role for 1 hour |
| 5656 | 4550 | wp_cache_set($cache_key, $role_restriction, 'mxchat_vector_roles', 3600); |
| 5657 | - | |
| 4551 | + | |
| 5658 | 4552 | return $role_restriction; |
| 5659 | 4553 | } |
| 5660 | 4554 | |
| 5661 | -/** | |
| 5662 | - * Fetch and reassemble all chunks for a URL from Pinecone | |
| 5663 | - * | |
| 5664 | - * @param string $source_url The source URL to fetch chunks for | |
| 5665 | - * @param array $bot_config Bot-specific Pinecone configuration | |
| 5666 | - * @return string Reassembled content from all chunks | |
| 5667 | - */ | |
| 5668 | -private function reassemble_chunks_from_pinecone($source_url, $bot_config, $max_chunks = 0, &$chunk_count = 0) { | |
| 5669 | - $api_key = $bot_config['api_key'] ?? ''; | |
| 5670 | - $host = $bot_config['host'] ?? ''; | |
| 5671 | - $namespace = $bot_config['namespace'] ?? ''; | |
| 5672 | - | |
| 5673 | - if (empty($host) || empty($api_key)) { | |
| 5674 | - $chunk_count = 0; | |
| 5675 | - return ''; | |
| 5676 | - } | |
| 5677 | - | |
| 5678 | - $base_hash = md5($source_url); | |
| 5679 | - | |
| 5680 | - // Use Pinecone list API to find all chunk vectors with this prefix | |
| 5681 | - $list_url = "https://{$host}/vectors/list"; | |
| 5682 | - | |
| 5683 | - // Limit to max_chunks if specified, otherwise fetch up to 100 | |
| 5684 | - $fetch_limit = ($max_chunks > 0 && $max_chunks < 100) ? $max_chunks : 100; | |
| 5685 | - | |
| 5686 | - $list_body = array( | |
| 5687 | - 'prefix' => $base_hash . '_chunk_', | |
| 5688 | - 'limit' => $fetch_limit | |
| 5689 | - ); | |
| 5690 | - | |
| 5691 | - if (!empty($namespace)) { | |
| 5692 | - $list_body['namespace'] = $namespace; | |
| 5693 | - } | |
| 5694 | - | |
| 5695 | - $list_response = wp_remote_post($list_url, array( | |
| 5696 | - 'headers' => array( | |
| 5697 | - 'Api-Key' => $api_key, | |
| 5698 | - 'accept' => 'application/json', | |
| 5699 | - 'content-type' => 'application/json' | |
| 5700 | - ), | |
| 5701 | - 'body' => wp_json_encode($list_body), | |
| 5702 | - 'timeout' => 30 | |
| 5703 | - )); | |
| 5704 | - | |
| 5705 | - if (is_wp_error($list_response)) { | |
| 5706 | - //error_log('[MXCHAT-CHUNK] List API error: ' . $list_response->get_error_message()); | |
| 5707 | - return ''; | |
| 5708 | - } | |
| 5709 | - | |
| 5710 | - $list_data = json_decode(wp_remote_retrieve_body($list_response), true); | |
| 5711 | - | |
| 5712 | - if (empty($list_data['vectors'])) { | |
| 5713 | - //error_log('[MXCHAT-CHUNK] No chunk vectors found for URL: ' . $source_url); | |
| 5714 | - return ''; | |
| 5715 | - } | |
| 5716 | - | |
| 5717 | - // Extract vector IDs | |
| 5718 | - $vector_ids = array(); | |
| 5719 | - foreach ($list_data['vectors'] as $vector) { | |
| 5720 | - if (isset($vector['id'])) { | |
| 5721 | - $vector_ids[] = $vector['id']; | |
| 5722 | - } | |
| 5723 | - } | |
| 5724 | - | |
| 5725 | - if (empty($vector_ids)) { | |
| 5726 | - return ''; | |
| 5727 | - } | |
| 5728 | - | |
| 5729 | - // Fetch all chunk content | |
| 5730 | - $fetch_url = "https://{$host}/vectors/fetch"; | |
| 5731 | - | |
| 5732 | - $fetch_body = array( | |
| 5733 | - 'ids' => $vector_ids | |
| 5734 | - ); | |
| 5735 | - | |
| 5736 | - if (!empty($namespace)) { | |
| 5737 | - $fetch_body['namespace'] = $namespace; | |
| 5738 | - } | |
| 5739 | - | |
| 5740 | - $fetch_response = wp_remote_post($fetch_url, array( | |
| 5741 | - 'headers' => array( | |
| 5742 | - 'Api-Key' => $api_key, | |
| 5743 | - 'accept' => 'application/json', | |
| 5744 | - 'content-type' => 'application/json' | |
| 5745 | - ), | |
| 5746 | - 'body' => wp_json_encode($fetch_body), | |
| 5747 | - 'timeout' => 30 | |
| 5748 | - )); | |
| 5749 | - | |
| 5750 | - if (is_wp_error($fetch_response)) { | |
| 5751 | - //error_log('[MXCHAT-CHUNK] Fetch API error: ' . $fetch_response->get_error_message()); | |
| 5752 | - return ''; | |
| 5753 | - } | |
| 5754 | - | |
| 5755 | - $fetch_data = json_decode(wp_remote_retrieve_body($fetch_response), true); | |
| 5756 | - | |
| 5757 | - if (empty($fetch_data['vectors'])) { | |
| 5758 | - return ''; | |
| 5759 | - } | |
| 5760 | - | |
| 5761 | - // Sort chunks by index and reassemble | |
| 5762 | - $chunks = array(); | |
| 5763 | - foreach ($fetch_data['vectors'] as $id => $vector) { | |
| 5764 | - $metadata = $vector['metadata'] ?? array(); | |
| 5765 | - $chunk_index = $metadata['chunk_index'] ?? 0; | |
| 5766 | - $text = $metadata['text'] ?? ''; | |
| 5767 | - | |
| 5768 | - // Store chunk with its index | |
| 5769 | - $chunks[$chunk_index] = $text; | |
| 5770 | - } | |
| 5771 | - | |
| 5772 | - // Sort by chunk index | |
| 5773 | - ksort($chunks); | |
| 5774 | - | |
| 5775 | - // Apply chunk limit if specified | |
| 5776 | - if ($max_chunks > 0 && count($chunks) > $max_chunks) { | |
| 5777 | - $chunks = array_slice($chunks, 0, $max_chunks, true); | |
| 5778 | - } | |
| 5779 | - | |
| 5780 | - // Store actual chunk count | |
| 5781 | - $chunk_count = count($chunks); | |
| 5782 | - | |
| 5783 | - // Reassemble content | |
| 5784 | - return implode("\n\n", $chunks); | |
| 5785 | -} | |
| 5786 | - | |
| 5787 | -/** | |
| 5788 | - * Search for relevant content using OpenAI Vector Store (File Search) | |
| 5789 | - * | |
| 5790 | - * @param string $user_query The user's query text | |
| 5791 | - * @param string $bot_id The bot ID | |
| 5792 | - * @param array $vectorstore_config Vector Store configuration | |
| 5793 | - * @return string Formatted context string with references | |
| 5794 | - */ | |
| 5795 | -private function find_relevant_content_openai_vectorstore($user_query, $bot_id = 'default', $vectorstore_config = array()) { | |
| 5796 | - //error_log("MXCHAT DEBUG: find_relevant_content_openai_vectorstore called"); | |
| 5797 | - //error_log(" - bot_id: " . $bot_id); | |
| 5798 | - //error_log(" - user_query length: " . strlen($user_query)); | |
| 5799 | - | |
| 5800 | - // Get OpenAI API key | |
| 5801 | - $mxchat_options = get_option('mxchat_options', array()); | |
| 5802 | - $api_key = $mxchat_options['api_key'] ?? ''; | |
| 5803 | - | |
| 5804 | - // Reset vectorstore error tracking | |
| 5805 | - $this->last_vectorstore_error = null; | |
| 5806 | - | |
| 5807 | - if (empty($api_key)) { | |
| 5808 | - //error_log("MXCHAT DEBUG ERROR: OpenAI API key not configured"); | |
| 5809 | - $this->last_vectorstore_error = 'Vector Store search failed: OpenAI API key is not configured.'; | |
| 5810 | - $this->current_valid_urls = []; | |
| 5811 | - return ''; | |
| 5812 | - } | |
| 5813 | - | |
| 5814 | - // Get Vector Store configuration | |
| 5815 | - if (empty($vectorstore_config)) { | |
| 5816 | - $vectorstore_config = $this->get_bot_vectorstore_config($bot_id); | |
| 5817 | - } | |
| 5818 | - | |
| 5819 | - $vectorstore_ids_string = $vectorstore_config['vectorstore_ids'] ?? ''; | |
| 5820 | - $max_results = $vectorstore_config['max_results'] ?? 5; | |
| 5821 | - | |
| 5822 | - if (empty($vectorstore_ids_string)) { | |
| 5823 | - //error_log("MXCHAT DEBUG ERROR: No Vector Store IDs configured"); | |
| 5824 | - $this->last_vectorstore_error = 'Vector Store search failed: No Vector Store IDs are configured for this bot.'; | |
| 5825 | - $this->current_valid_urls = []; | |
| 5826 | - return ''; | |
| 5827 | - } | |
| 5828 | - | |
| 5829 | - // Parse Vector Store IDs | |
| 5830 | - $vectorstore_ids = array_map('trim', explode(',', $vectorstore_ids_string)); | |
| 5831 | - $vectorstore_ids = array_filter($vectorstore_ids); // Remove empty values | |
| 5832 | - | |
| 5833 | - //error_log("MXCHAT DEBUG: Vector Store IDs: " . implode(', ', $vectorstore_ids)); | |
| 5834 | - //error_log("MXCHAT DEBUG: Max results: " . $max_results); | |
| 5835 | - | |
| 5836 | - // Initialize similarity analysis storage | |
| 5837 | - $this->last_similarity_analysis = [ | |
| 5838 | - 'knowledge_base_type' => 'OpenAI Vector Store', | |
| 5839 | - 'bot_id' => $bot_id, | |
| 5840 | - 'vectorstore_ids' => $vectorstore_ids, | |
| 5841 | - 'top_matches' => [], | |
| 5842 | - 'threshold_used' => 0, | |
| 5843 | - 'total_checked' => 0 | |
| 5844 | - ]; | |
| 5845 | - | |
| 5846 | - $valid_urls = []; | |
| 5847 | - | |
| 5848 | - // Get the selected model | |
| 5849 | - $bot_options = $this->get_bot_options($bot_id); | |
| 5850 | - $current_options = !empty($bot_options) ? $bot_options : $mxchat_options; | |
| 5851 | - $selected_model = $current_options['model'] ?? 'gpt-5.1-chat-latest'; | |
| 5852 | - | |
| 5853 | - // Verify it's an OpenAI model | |
| 5854 | - if (!$this->is_openai_chat_model($selected_model)) { | |
| 5855 | - //error_log("MXCHAT DEBUG ERROR: Vector Store search requires OpenAI model. Current: " . $selected_model); | |
| 5856 | - $this->last_vectorstore_error = 'Vector Store search requires an OpenAI model. Current model: ' . $selected_model; | |
| 5857 | - $this->current_valid_urls = []; | |
| 5858 | - return ''; | |
| 5859 | - } | |
| 5860 | - | |
| 5861 | - // Use OpenAI Responses API with file_search tool | |
| 5862 | - $request_body = array( | |
| 5863 | - 'model' => $selected_model, | |
| 5864 | - 'input' => $user_query, | |
| 5865 | - 'tools' => array( | |
| 5866 | - array( | |
| 5867 | - 'type' => 'file_search', | |
| 5868 | - 'vector_store_ids' => $vectorstore_ids, | |
| 5869 | - 'max_num_results' => intval($max_results) | |
| 5870 | - ) | |
| 5871 | - ), | |
| 5872 | - 'include' => array('output[*].file_search_call.search_results') | |
| 5873 | - ); | |
| 5874 | - | |
| 5875 | - //error_log("MXCHAT VECTORSTORE: ========== REQUEST START =========="); | |
| 5876 | - //error_log("MXCHAT VECTORSTORE: Model: " . $selected_model); | |
| 5877 | - //error_log("MXCHAT VECTORSTORE: Query: " . substr($user_query, 0, 200)); | |
| 5878 | - //error_log("MXCHAT VECTORSTORE: Vector Store IDs: " . implode(', ', $vectorstore_ids)); | |
| 5879 | - //error_log("MXCHAT VECTORSTORE: Max Results: " . $max_results); | |
| 5880 | - //error_log("MXCHAT VECTORSTORE: Request body: " . wp_json_encode($request_body)); | |
| 5881 | - | |
| 5882 | - $response = wp_remote_post('https://api.openai.com/v1/responses', array( | |
| 5883 | - 'headers' => array( | |
| 5884 | - 'Authorization' => 'Bearer ' . $api_key, | |
| 5885 | - 'Content-Type' => 'application/json' | |
| 5886 | - ), | |
| 5887 | - 'body' => wp_json_encode($request_body), | |
| 5888 | - 'timeout' => 60 | |
| 5889 | - )); | |
| 5890 | - | |
| 5891 | - if (is_wp_error($response)) { | |
| 5892 | - //error_log("MXCHAT VECTORSTORE ERROR: WP Error: " . $response->get_error_message()); | |
| 5893 | - $this->last_vectorstore_error = 'Vector Store API request failed: ' . $response->get_error_message(); | |
| 5894 | - $this->current_valid_urls = []; | |
| 5895 | - return ''; | |
| 5896 | - } | |
| 5897 | - | |
| 5898 | - $response_code = wp_remote_retrieve_response_code($response); | |
| 5899 | - //error_log("MXCHAT VECTORSTORE: Response code: " . $response_code); | |
| 5900 | - | |
| 5901 | - $response_body = wp_remote_retrieve_body($response); | |
| 5902 | - //error_log("MXCHAT VECTORSTORE: Raw response (first 2000 chars): " . substr($response_body, 0, 2000)); | |
| 5903 | - | |
| 5904 | - if ($response_code !== 200) { | |
| 5905 | - //error_log("MXCHAT VECTORSTORE ERROR: API error response: " . $response_body); | |
| 5906 | - $api_error_detail = ''; | |
| 5907 | - $decoded_error = json_decode($response_body, true); | |
| 5908 | - if (isset($decoded_error['error']['message'])) { | |
| 5909 | - $api_error_detail = $decoded_error['error']['message']; | |
| 5910 | - } | |
| 5911 | - $this->last_vectorstore_error = 'Vector Store API returned HTTP ' . $response_code . ($api_error_detail ? ': ' . $api_error_detail : ''); | |
| 5912 | - $this->current_valid_urls = []; | |
| 5913 | - return ''; | |
| 5914 | - } | |
| 5915 | - $result = json_decode($response_body, true); | |
| 5916 | - | |
| 5917 | - if (json_last_error() !== JSON_ERROR_NONE) { | |
| 5918 | - //error_log("MXCHAT VECTORSTORE ERROR: JSON decode error: " . json_last_error_msg()); | |
| 5919 | - $this->last_vectorstore_error = 'Vector Store response could not be parsed: ' . json_last_error_msg(); | |
| 5920 | - $this->current_valid_urls = []; | |
| 5921 | - return ''; | |
| 5922 | - } | |
| 5923 | - | |
| 5924 | - // Debug: Log the structure of the result | |
| 5925 | - //error_log("MXCHAT VECTORSTORE: Result keys: " . implode(', ', array_keys($result))); | |
| 5926 | - if (isset($result['output'])) { | |
| 5927 | - //error_log("MXCHAT VECTORSTORE: Output count: " . count($result['output'])); | |
| 5928 | - foreach ($result['output'] as $idx => $out) { | |
| 5929 | - //error_log("MXCHAT VECTORSTORE: Output[$idx] type: " . ($out['type'] ?? 'unknown')); | |
| 5930 | - //error_log("MXCHAT VECTORSTORE: Output[$idx] keys: " . implode(', ', array_keys($out))); | |
| 5931 | - } | |
| 5932 | - } else { | |
| 5933 | - //error_log("MXCHAT VECTORSTORE: No 'output' key in result!"); | |
| 5934 | - } | |
| 5935 | - | |
| 5936 | - // Extract file search results from the response | |
| 5937 | - $content = ''; | |
| 5938 | - $matches_used = 0; | |
| 5939 | - $all_matches = []; | |
| 5940 | - | |
| 5941 | - // The Responses API returns output array with tool results | |
| 5942 | - if (isset($result['output']) && is_array($result['output'])) { | |
| 5943 | - foreach ($result['output'] as $output_item) { | |
| 5944 | - // Look for file_search_call results | |
| 5945 | - if (isset($output_item['type']) && $output_item['type'] === 'file_search_call') { | |
| 5946 | - //error_log("MXCHAT VECTORSTORE: Found file_search_call output item"); | |
| 5947 | - //error_log("MXCHAT VECTORSTORE: file_search_call keys: " . implode(', ', array_keys($output_item))); | |
| 5948 | - | |
| 5949 | - // Check for search_results in the output item directly | |
| 5950 | - $search_results = $output_item['search_results'] ?? $output_item['results'] ?? []; | |
| 5951 | - //error_log("MXCHAT VECTORSTORE: Search results count: " . count($search_results)); | |
| 5952 | - | |
| 5953 | - if (empty($search_results)) { | |
| 5954 | - //error_log("MXCHAT VECTORSTORE: No search results found in file_search_call"); | |
| 5955 | - //error_log("MXCHAT VECTORSTORE: file_search_call content: " . wp_json_encode($output_item)); | |
| 5956 | - } | |
| 5957 | - | |
| 5958 | - foreach ($search_results as $index => $search_result) { | |
| 5959 | - $filename = $search_result['filename'] ?? ''; | |
| 5960 | - $score = $search_result['score'] ?? 0; | |
| 5961 | - $text_content = ''; | |
| 5962 | - | |
| 5963 | - // Extract text content from the result | |
| 5964 | - // The text can be directly on the result OR nested under content array | |
| 5965 | - if (isset($search_result['text']) && !empty($search_result['text'])) { | |
| 5966 | - // Direct text field (OpenAI's actual format) | |
| 5967 | - $text_content = $search_result['text']; | |
| 5968 | - //error_log("MXCHAT VECTORSTORE: Found text directly on result[$index], length: " . strlen($text_content)); | |
| 5969 | - } elseif (isset($search_result['content']) && is_array($search_result['content'])) { | |
| 5970 | - // Nested content array format | |
| 5971 | - foreach ($search_result['content'] as $content_item) { | |
| 5972 | - if (isset($content_item['text'])) { | |
| 5973 | - $text_content .= $content_item['text'] . "\n"; | |
| 5974 | - } | |
| 5975 | - } | |
| 5976 | - //error_log("MXCHAT VECTORSTORE: Found text in content array for result[$index], length: " . strlen($text_content)); | |
| 5977 | - } else { | |
| 5978 | - //error_log("MXCHAT VECTORSTORE: No text found for result[$index]. Keys: " . implode(', ', array_keys($search_result))); | |
| 5979 | - } | |
| 5980 | - | |
| 5981 | - if (!empty($text_content)) { | |
| 5982 | - $content .= "## Reference " . ($matches_used + 1) . " ##\n"; | |
| 5983 | - $content .= trim($text_content) . "\n\n"; | |
| 5984 | - | |
| 5985 | - if (!empty($filename)) { | |
| 5986 | - $content .= "Source: " . $filename . "\n\n"; | |
| 5987 | - } | |
| 5988 | - | |
| 5989 | - // Extract URLs from content | |
| 5990 | - preg_match_all( | |
| 5991 | - '#\bhttps?://[^\s<>"\']+#i', | |
| 5992 | - $text_content, | |
| 5993 | - $content_urls | |
| 5994 | - ); | |
| 5995 | - if (!empty($content_urls[0])) { | |
| 5996 | - $valid_urls = array_merge($valid_urls, $content_urls[0]); | |
| 5997 | - } | |
| 5998 | - | |
| 5999 | - $matches_used++; | |
| 6000 | - } | |
| 6001 | - | |
| 6002 | - // Store for similarity analysis | |
| 6003 | - $all_matches[] = [ | |
| 6004 | - 'document_id' => $filename ?: ('result_' . $index), | |
| 6005 | - 'similarity' => $score, | |
| 6006 | - 'similarity_percentage' => round($score * 100, 2), | |
| 6007 | - 'above_threshold' => true, | |
| 6008 | - 'source_display' => $filename, | |
| 6009 | - 'content_preview' => substr(strip_tags($text_content), 0, 100) . '...', | |
| 6010 | - 'used_for_context' => true, | |
| 6011 | - 'role_restriction' => 'public', | |
| 6012 | - 'has_access' => true, | |
| 6013 | - 'filtered_out' => false | |
| 6014 | - ]; | |
| 6015 | - } | |
| 6016 | - } | |
| 6017 | - | |
| 6018 | - // Also check for message content with annotations (citations) | |
| 6019 | - if (isset($output_item['type']) && $output_item['type'] === 'message') { | |
| 6020 | - if (isset($output_item['content']) && is_array($output_item['content'])) { | |
| 6021 | - foreach ($output_item['content'] as $content_block) { | |
| 6022 | - if (isset($content_block['annotations']) && is_array($content_block['annotations'])) { | |
| 6023 | - foreach ($content_block['annotations'] as $annotation) { | |
| 6024 | - if (isset($annotation['filename'])) { | |
| 6025 | - $filename = $annotation['filename']; | |
| 6026 | - $score = $annotation['score'] ?? 0; | |
| 6027 | - $text_content = ''; | |
| 6028 | - | |
| 6029 | - if (isset($annotation['content']) && is_array($annotation['content'])) { | |
| 6030 | - foreach ($annotation['content'] as $ann_content) { | |
| 6031 | - if (isset($ann_content['text'])) { | |
| 6032 | - $text_content .= $ann_content['text'] . "\n"; | |
| 6033 | - } | |
| 6034 | - } | |
| 6035 | - } | |
| 6036 | - | |
| 6037 | - if (!empty($text_content) && $matches_used < $max_results) { | |
| 6038 | - $content .= "## Reference " . ($matches_used + 1) . " ##\n"; | |
| 6039 | - $content .= trim($text_content) . "\n\n"; | |
| 6040 | - $content .= "Source: " . $filename . "\n\n"; | |
| 6041 | - | |
| 6042 | - preg_match_all( | |
| 6043 | - '#\bhttps?://[^\s<>"\']+#i', | |
| 6044 | - $text_content, | |
| 6045 | - $content_urls | |
| 6046 | - ); | |
| 6047 | - if (!empty($content_urls[0])) { | |
| 6048 | - $valid_urls = array_merge($valid_urls, $content_urls[0]); | |
| 6049 | - } | |
| 6050 | - | |
| 6051 | - $matches_used++; | |
| 6052 | - | |
| 6053 | - $all_matches[] = [ | |
| 6054 | - 'document_id' => $filename, | |
| 6055 | - 'similarity' => $score, | |
| 6056 | - 'similarity_percentage' => round($score * 100, 2), | |
| 6057 | - 'above_threshold' => true, | |
| 6058 | - 'source_display' => $filename, | |
| 6059 | - 'content_preview' => substr(strip_tags($text_content), 0, 100) . '...', | |
| 6060 | - 'used_for_context' => true, | |
| 6061 | - 'role_restriction' => 'public', | |
| 6062 | - 'has_access' => true, | |
| 6063 | - 'filtered_out' => false | |
| 6064 | - ]; | |
| 6065 | - } | |
| 6066 | - } | |
| 6067 | - } | |
| 6068 | - } | |
| 6069 | - } | |
| 6070 | - } | |
| 6071 | - } | |
| 6072 | - } | |
| 6073 | - } | |
| 6074 | - | |
| 6075 | - // Store for testing panel | |
| 6076 | - $this->last_similarity_analysis['top_matches'] = $all_matches; | |
| 6077 | - $this->last_similarity_analysis['total_checked'] = count($all_matches); | |
| 6078 | - | |
| 6079 | - // Store unique valid URLs for validation | |
| 6080 | - $this->current_valid_urls = array_unique($valid_urls); | |
| 6081 | - | |
| 6082 | - // Allow add-ons to act on similarity results (e.g. WooCommerce product card display) | |
| 6083 | - do_action('mxchat_similarity_results', $this->last_similarity_analysis['top_matches'], $bot_id); | |
| 6084 | - | |
| 6085 | - //error_log("MXCHAT VECTORSTORE: ========== SEARCH COMPLETE =========="); | |
| 6086 | - //error_log("MXCHAT VECTORSTORE: Matches used: " . $matches_used); | |
| 6087 | - //error_log("MXCHAT VECTORSTORE: All matches count: " . count($all_matches)); | |
| 6088 | - //error_log("MXCHAT VECTORSTORE: Content length: " . strlen($content)); | |
| 6089 | - if ($matches_used > 0) { | |
| 6090 | - //error_log("MXCHAT VECTORSTORE: Content preview: " . substr($content, 0, 500)); | |
| 6091 | - } | |
| 6092 | - | |
| 6093 | - // Check if citation links are enabled | |
| 6094 | - $citation_links_enabled = ($mxchat_options['citation_links_toggle'] ?? 'on') === 'on'; | |
| 6095 | - | |
| 6096 | - // Add response guidelines | |
| 6097 | - if ($matches_used === 0) { | |
| 6098 | - //error_log("MXCHAT VECTORSTORE: No matches found - returning empty reference message"); | |
| 6099 | - $content = "No reference information was found for this query.\n\n"; | |
| 6100 | - } else { | |
| 6101 | - // Build response guidelines based on citation links setting | |
| 6102 | - $content .= "\n## Response Guidelines ##\n" . | |
| 6103 | - "You are an AI Chatbot. Answer naturally and helpfully using only the information from the references above. " . | |
| 6104 | - "Be conversational and friendly, but never mention your knowledge base or training data. " . | |
| 6105 | - "If you don't have specific information or are uncertain about any details, it's always " . | |
| 6106 | - "better to honestly say you don't know rather than making up or guessing at answers. " . | |
| 6107 | - "When information is incomplete, let them know you are unsure.\n\n"; | |
| 6108 | - | |
| 6109 | - // Only add hyperlink instructions if citation links are enabled | |
| 6110 | - if ($citation_links_enabled) { | |
| 6111 | - $content .= "CRITICAL: When creating hyperlinks, always use proper markdown format with descriptive text: " . | |
| 6112 | - "[descriptive text](url). NEVER use empty brackets like [](url). The text in brackets must describe what the link is about."; | |
| 6113 | - } else { | |
| 6114 | - $content .= "IMPORTANT: Do not include any citation links, source URLs, or hyperlinks in your responses. " . | |
| 6115 | - "Simply provide helpful answers based on the reference information without citing sources."; | |
| 6116 | - } | |
| 6117 | - } | |
| 6118 | - | |
| 6119 | - //error_log("MXCHAT DEBUG: Vector Store search complete. Matches used: " . $matches_used); | |
| 6120 | - | |
| 6121 | - return trim($content); | |
| 6122 | -} | |
| 6123 | - | |
| 6124 | -/** | |
| 6125 | - * Check if the given model is an OpenAI chat model | |
| 6126 | - * | |
| 6127 | - * @param string $model The model ID | |
| 6128 | - * @return bool True if it's an OpenAI model | |
| 6129 | - */ | |
| 6130 | -private function is_openai_chat_model($model) { | |
| 6131 | - $openai_prefixes = array('gpt-', 'o1-', 'o3-'); | |
| 6132 | - foreach ($openai_prefixes as $prefix) { | |
| 6133 | - if (strpos($model, $prefix) === 0) { | |
| 6134 | - return true; | |
| 6135 | - } | |
| 6136 | - } | |
| 6137 | - return false; | |
| 6138 | -} | |
| 6139 | - | |
| 6140 | -/** | |
| 6141 | - * Get bot-specific Vector Store configuration | |
| 6142 | - * | |
| 6143 | - * @param string $bot_id The bot ID | |
| 6144 | - * @return array Configuration array | |
| 6145 | - */ | |
| 6146 | -private function get_bot_vectorstore_config($bot_id = 'default') { | |
| 6147 | - $vectorstore_options = get_option('mxchat_openai_vectorstore_options', array()); | |
| 6148 | - | |
| 6149 | - // Default global settings | |
| 6150 | - $default_config = array( | |
| 6151 | - 'use_vectorstore' => ($vectorstore_options['mxchat_use_openai_vectorstore'] ?? '0') === '1', | |
| 6152 | - 'vectorstore_ids' => $vectorstore_options['mxchat_vectorstore_ids'] ?? '', | |
| 6153 | - 'max_results' => $vectorstore_options['mxchat_vectorstore_max_results'] ?? 5 | |
| 6154 | - ); | |
| 6155 | - | |
| 6156 | - // Allow multi-bot plugin to override with bot-specific settings | |
| 6157 | - $bot_config = apply_filters('mxchat_get_bot_vectorstore_config', $default_config, $bot_id); | |
| 6158 | - | |
| 6159 | - // Preserve max_results from global settings if not set in bot config | |
| 6160 | - if (!isset($bot_config['max_results'])) { | |
| 6161 | - $bot_config['max_results'] = $default_config['max_results']; | |
| 6162 | - } | |
| 6163 | - | |
| 6164 | - return $bot_config; | |
| 6165 | -} | |
| 6166 | - | |
| 6167 | 4555 | private function mxchat_find_relevant_products($user_embedding) { |
| 6168 | 4556 | //error_log('MXChat Vector Search: Starting product search...'); |
| 6169 | 4557 | |
| 6170 | 4558 | // Retrieve the add-on settings from the database |
| @@ -6185,75 +4573,73 @@ | ||
| 6185 | 4573 | } |
| 6186 | 4574 | private function find_relevant_products_wordpress($user_embedding) { |
| 6187 | 4575 | global $wpdb; |
| 6188 | 4576 | $system_prompt_table = $wpdb->prefix . 'mxchat_system_prompt_content'; |
| 4577 | + $cache_key = 'mxchat_system_prompt_embeddings'; | |
| 4578 | + $batch_size = 500; | |
| 6189 | 4579 | |
| 6190 | - if (!is_array($user_embedding)) { | |
| 6191 | - return ''; | |
| 6192 | - } | |
| 4580 | + // Original WordPress database search logic | |
| 4581 | + // [Previous implementation remains the same] | |
| 4582 | + $embeddings = wp_cache_get($cache_key, 'mxchat_system_prompts'); | |
| 4583 | + if ($embeddings === false) { | |
| 4584 | + $embeddings = []; | |
| 4585 | + $offset = 0; | |
| 6193 | 4586 | |
| 6194 | - // Streaming top-K pass: scan rows in small batches, keep only the top 3 | |
| 6195 | - // results above the similarity threshold. Peak memory is bounded by | |
| 6196 | - // $batch_size embedding rows plus a 3-element top list. | |
| 6197 | - $batch_size = 250; | |
| 6198 | - $similarity_threshold = 0.85; | |
| 6199 | - $top_k = 3; | |
| 6200 | - $top_results = []; | |
| 6201 | - $offset = 0; | |
| 4587 | + do { | |
| 4588 | + $query = $wpdb->prepare( | |
| 4589 | + "SELECT id, embedding_vector | |
| 4590 | + FROM {$system_prompt_table} | |
| 4591 | + LIMIT %d OFFSET %d", | |
| 4592 | + $batch_size, | |
| 4593 | + $offset | |
| 4594 | + ); | |
| 6202 | 4595 | |
| 6203 | - do { | |
| 6204 | - $batch = $wpdb->get_results($wpdb->prepare( | |
| 6205 | - "SELECT id, embedding_vector | |
| 6206 | - FROM {$system_prompt_table} | |
| 6207 | - LIMIT %d OFFSET %d", | |
| 6208 | - $batch_size, | |
| 6209 | - $offset | |
| 6210 | - )); | |
| 4596 | + $batch = $wpdb->get_results($query); | |
| 4597 | + if (empty($batch)) { | |
| 4598 | + break; | |
| 4599 | + } | |
| 6211 | 4600 | |
| 6212 | - if (empty($batch)) { | |
| 6213 | - break; | |
| 6214 | - } | |
| 4601 | + $embeddings = array_merge($embeddings, $batch); | |
| 4602 | + $offset += $batch_size; | |
| 6215 | 4603 | |
| 6216 | - foreach ($batch as $row) { | |
| 6217 | - $database_embedding = $row->embedding_vector | |
| 6218 | - ? unserialize($row->embedding_vector, ['allowed_classes' => false]) | |
| 6219 | - : null; | |
| 4604 | + unset($batch); | |
| 6220 | 4605 | |
| 6221 | - if (!is_array($database_embedding)) { | |
| 6222 | - unset($database_embedding); | |
| 6223 | - continue; | |
| 6224 | - } | |
| 4606 | + } while (true); | |
| 6225 | 4607 | |
| 4608 | + if (empty($embeddings)) { | |
| 4609 | + return ''; | |
| 4610 | + } | |
| 4611 | + wp_cache_set($cache_key, $embeddings, 'mxchat_system_prompts', 3600); | |
| 4612 | + } | |
| 4613 | + | |
| 4614 | + $relevant_results = []; | |
| 4615 | + foreach ($embeddings as $embedding) { | |
| 4616 | + $database_embedding = $embedding->embedding_vector | |
| 4617 | + ? unserialize($embedding->embedding_vector, ['allowed_classes' => false]) | |
| 4618 | + : null; | |
| 4619 | + if (is_array($database_embedding) && is_array($user_embedding)) { | |
| 6226 | 4620 | $similarity = $this->mxchat_calculate_cosine_similarity($user_embedding, $database_embedding); |
| 6227 | - unset($database_embedding); | |
| 6228 | - | |
| 6229 | - if ($similarity < $similarity_threshold) { | |
| 6230 | - continue; | |
| 6231 | - } | |
| 6232 | - | |
| 6233 | - // Insert into bounded top-K (kept sorted descending) | |
| 6234 | - if (count($top_results) < $top_k) { | |
| 6235 | - $top_results[] = ['id' => $row->id, 'similarity' => $similarity]; | |
| 6236 | - usort($top_results, function ($a, $b) { | |
| 6237 | - return $b['similarity'] <=> $a['similarity']; | |
| 6238 | - }); | |
| 6239 | - } elseif ($similarity > $top_results[$top_k - 1]['similarity']) { | |
| 6240 | - $top_results[$top_k - 1] = ['id' => $row->id, 'similarity' => $similarity]; | |
| 6241 | - usort($top_results, function ($a, $b) { | |
| 6242 | - return $b['similarity'] <=> $a['similarity']; | |
| 6243 | - }); | |
| 6244 | - } | |
| 4621 | + $relevant_results[] = [ | |
| 4622 | + 'id' => $embedding->id, | |
| 4623 | + 'similarity' => $similarity | |
| 4624 | + ]; | |
| 6245 | 4625 | } |
| 4626 | + unset($database_embedding); | |
| 4627 | + } | |
| 6246 | 4628 | |
| 6247 | - unset($batch); | |
| 6248 | - $offset += $batch_size; | |
| 6249 | - } while (true); | |
| 4629 | + // Use fixed threshold for products | |
| 4630 | + $similarity_threshold = 0.85; | |
| 6250 | 4631 | |
| 6251 | - if (empty($top_results)) { | |
| 6252 | - return ''; | |
| 6253 | - } | |
| 4632 | + $relevant_results = array_filter($relevant_results, function ($result) use ($similarity_threshold) { | |
| 4633 | + return $result['similarity'] >= $similarity_threshold; | |
| 4634 | + }); | |
| 4635 | + usort($relevant_results, function ($a, $b) { | |
| 4636 | + return $b['similarity'] <=> $a['similarity']; | |
| 4637 | + }); | |
| 6254 | 4638 | |
| 4639 | + $top_results = array_slice($relevant_results, 0, 3); | |
| 6255 | 4640 | $content = ''; |
| 4641 | + | |
| 6256 | 4642 | foreach ($top_results as $result) { |
| 6257 | 4643 | $chunk_content = $this->fetch_content_with_product_links($result['id']); |
| 6258 | 4644 | $content .= $chunk_content . "\n\n"; |
| 6259 | 4645 | } |
| @@ -6362,61 +4748,23 @@ | ||
| 6362 | 4748 | |
| 6363 | 4749 | /** |
| 6364 | 4750 | * Get system instructions for a specific bot or default |
| 6365 | 4751 | * Checks for multi-bot add-on and uses bot-specific instructions if available |
| 6366 | - * Automatically strips URLs if citation links are disabled | |
| 6367 | - * Replaces {visitor_name} placeholder with actual visitor name if available | |
| 6368 | - * | |
| 6369 | - * @param string $bot_id The bot ID to get instructions for | |
| 6370 | - * @param string $session_id Optional session ID to lookup visitor name | |
| 6371 | 4752 | */ |
| 6372 | -private function get_system_instructions($bot_id = 'default', $session_id = '') { | |
| 6373 | - $instructions = ''; | |
| 6374 | - | |
| 4753 | +private function get_system_instructions($bot_id = 'default') { | |
| 6375 | 4754 | // Check if multi-bot add-on is active |
| 6376 | 4755 | if (class_exists('MxChat_Multi_Bot_Core_Manager') && $bot_id !== 'default') { |
| 6377 | 4756 | // Get bot-specific options from multi-bot add-on |
| 6378 | 4757 | $bot_options = apply_filters('mxchat_get_bot_options', array(), $bot_id); |
| 6379 | - | |
| 4758 | + | |
| 6380 | 4759 | // If bot has custom system instructions, use those |
| 6381 | 4760 | if (!empty($bot_options['system_prompt_instructions'])) { |
| 6382 | - $instructions = $bot_options['system_prompt_instructions']; | |
| 4761 | + return $bot_options['system_prompt_instructions']; | |
| 6383 | 4762 | } |
| 6384 | 4763 | } |
| 6385 | - | |
| 4764 | + | |
| 6386 | 4765 | // Fall back to default system instructions |
| 6387 | - if (empty($instructions)) { | |
| 6388 | - $instructions = isset($this->options['system_prompt_instructions']) ? $this->options['system_prompt_instructions'] : ''; | |
| 6389 | - } | |
| 6390 | - | |
| 6391 | - // Check if citation links are disabled - if so, strip URLs from instructions | |
| 6392 | - $fresh_options = get_option('mxchat_options', []); | |
| 6393 | - $citation_links_enabled = isset($fresh_options['citation_links_toggle']) ? ($fresh_options['citation_links_toggle'] === 'on') : true; | |
| 6394 | - | |
| 6395 | - if (!$citation_links_enabled && !empty($instructions)) { | |
| 6396 | - $instructions = preg_replace('#\bhttps?://[^\s<>"\']+#i', '', $instructions); | |
| 6397 | - $instructions = preg_replace('/\s+/', ' ', trim($instructions)); // Clean up extra spaces | |
| 6398 | - } | |
| 6399 | - | |
| 6400 | - // Replace {visitor_name} placeholder with actual visitor name if available | |
| 6401 | - if (!empty($instructions) && !empty($session_id) && stripos($instructions, '{visitor_name}') !== false) { | |
| 6402 | - $name_option_key = "mxchat_name_{$session_id}"; | |
| 6403 | - $visitor_name = get_option($name_option_key, ''); | |
| 6404 | - | |
| 6405 | - if (!empty($visitor_name)) { | |
| 6406 | - $instructions = str_ireplace('{visitor_name}', sanitize_text_field($visitor_name), $instructions); | |
| 6407 | - } else { | |
| 6408 | - // Remove placeholder if no name is available | |
| 6409 | - $instructions = str_ireplace('{visitor_name}', '', $instructions); | |
| 6410 | - $instructions = preg_replace('/\s{2,}/', ' ', trim($instructions)); // Clean up extra spaces | |
| 6411 | - } | |
| 6412 | - } | |
| 6413 | - | |
| 6414 | - // Allow developers to filter system instructions and process shortcodes | |
| 6415 | - $instructions = apply_filters('mxchat_system_instructions', $instructions, $bot_id, $session_id); | |
| 6416 | - $instructions = do_shortcode($instructions); | |
| 6417 | - | |
| 6418 | - return $instructions; | |
| 4766 | + return isset($this->options['system_prompt_instructions']) ? $this->options['system_prompt_instructions'] : ''; | |
| 6419 | 4767 | } |
| 6420 | 4768 | /** |
| 6421 | 4769 | * Get the current bot ID from session or request context |
| 6422 | 4770 | */ |
| @@ -6436,9 +4784,9 @@ | ||
| 6436 | 4784 | |
| 6437 | 4785 | // Fall back to default |
| 6438 | 4786 | return 'default'; |
| 6439 | 4787 | } |
| 6440 | -private function mxchat_generate_response($relevant_content, $api_key, $xai_api_key, $claude_api_key, $deepseek_api_key, $gemini_api_key, $openrouter_api_key, $conversation_history, $streaming = false, $session_id = '', $testing_data = null, $selected_model = 'gpt-5.1-chat-latest') { | |
| 4788 | +private function mxchat_generate_response($relevant_content, $api_key, $xai_api_key, $claude_api_key, $deepseek_api_key, $gemini_api_key, $openrouter_api_key, $conversation_history, $streaming = false, $session_id = '', $testing_data = null, $selected_model = 'gpt-4o') { | |
| 6441 | 4789 | try { |
| 6442 | 4790 | if (!$relevant_content) { |
| 6443 | 4791 | $error_response = [ |
| 6444 | 4792 | 'error' => esc_html__("I couldn't find relevant information on that topic.", 'mxchat'), |
| @@ -6637,27 +4985,9 @@ | ||
| 6637 | 4985 | $error_response['testing_data'] = $testing_data; |
| 6638 | 4986 | } |
| 6639 | 4987 | return $error_response; |
| 6640 | 4988 | } |
| 6641 | - | |
| 6642 | - // Check if web search is enabled for this OpenAI model | |
| 6643 | - $web_search_enabled = isset($this->options['enable_web_search']) && $this->options['enable_web_search'] === 'on'; | |
| 6644 | - // Models that don't support web search | |
| 6645 | - $unsupported_web_search_models = array('gpt-4.1-nano'); | |
| 6646 | - $model_supports_web_search = !in_array($selected_model, $unsupported_web_search_models); | |
| 6647 | - | |
| 6648 | - if ($web_search_enabled && $model_supports_web_search) { | |
| 6649 | - // Use Responses API (required for some models, or when web search is enabled) | |
| 6650 | - return $this->mxchat_generate_response_openai_web_search( | |
| 6651 | - $selected_model, | |
| 6652 | - $api_key, | |
| 6653 | - $conversation_history, | |
| 6654 | - $relevant_content, | |
| 6655 | - $session_id, | |
| 6656 | - $testing_data, | |
| 6657 | - $streaming | |
| 6658 | - ); | |
| 6659 | - } elseif ($streaming) { | |
| 4989 | + if ($streaming) { | |
| 6660 | 4990 | return $this->mxchat_generate_response_openai_stream( |
| 6661 | 4991 | $selected_model, |
| 6662 | 4992 | $api_key, |
| 6663 | 4993 | $conversation_history, |
| @@ -6685,25 +5015,9 @@ | ||
| 6685 | 5015 | $error_response['testing_data'] = $testing_data; |
| 6686 | 5016 | } |
| 6687 | 5017 | return $error_response; |
| 6688 | 5018 | } |
| 6689 | - | |
| 6690 | - // Check if web search is enabled (default case also handles OpenAI models) | |
| 6691 | - $web_search_enabled = isset($this->options['enable_web_search']) && $this->options['enable_web_search'] === 'on'; | |
| 6692 | - $unsupported_web_search_models = array('gpt-4.1-nano'); | |
| 6693 | - $model_supports_web_search = !in_array($selected_model, $unsupported_web_search_models); | |
| 6694 | - | |
| 6695 | - if ($web_search_enabled && $model_supports_web_search) { | |
| 6696 | - return $this->mxchat_generate_response_openai_web_search( | |
| 6697 | - $selected_model, | |
| 6698 | - $api_key, | |
| 6699 | - $conversation_history, | |
| 6700 | - $relevant_content, | |
| 6701 | - $session_id, | |
| 6702 | - $testing_data, | |
| 6703 | - $streaming | |
| 6704 | - ); | |
| 6705 | - } elseif ($streaming) { | |
| 5019 | + if ($streaming) { | |
| 6706 | 5020 | return $this->mxchat_generate_response_openai_stream( |
| 6707 | 5021 | $selected_model, |
| 6708 | 5022 | $api_key, |
| 6709 | 5023 | $conversation_history, |
| @@ -6747,9 +5061,9 @@ | ||
| 6747 | 5061 | } |
| 6748 | 5062 | private function mxchat_generate_response_openrouter_stream($selected_model, $openrouter_api_key, $conversation_history, $relevant_content, $session_id, $testing_data = null) { |
| 6749 | 5063 | try { |
| 6750 | 5064 | $bot_id = $this->get_current_bot_id($session_id); |
| 6751 | - $system_prompt_instructions = $this->get_system_instructions($bot_id, $session_id); | |
| 5065 | + $system_prompt_instructions = $this->get_system_instructions($bot_id); | |
| 6752 | 5066 | |
| 6753 | 5067 | if (!is_array($conversation_history)) { |
| 6754 | 5068 | $conversation_history = array(); |
| 6755 | 5069 | } |
| @@ -6811,11 +5125,8 @@ | ||
| 6811 | 5125 | 'temperature' => 1, |
| 6812 | 5126 | 'stream' => true |
| 6813 | 5127 | ]); |
| 6814 | 5128 | |
| 6815 | - // Setup streaming headers now that we know we're actually streaming | |
| 6816 | - $this->setup_streaming_headers(); | |
| 6817 | - | |
| 6818 | 5129 | $ch = curl_init(); |
| 6819 | 5130 | curl_setopt($ch, CURLOPT_URL, 'https://openrouter.ai/api/v1/chat/completions'); |
| 6820 | 5131 | curl_setopt($ch, CURLOPT_RETURNTRANSFER, false); |
| 6821 | 5132 | curl_setopt($ch, CURLOPT_POST, true); |
| @@ -6904,33 +5215,13 @@ | ||
| 6904 | 5215 | |
| 6905 | 5216 | curl_close($ch); |
| 6906 | 5217 | |
| 6907 | 5218 | if (!empty($full_response) && !empty($session_id)) { |
| 6908 | - // Prepare RAG context for streaming response | |
| 6909 | - $rag_context_for_storage = null; | |
| 6910 | - $has_rag_data = $this->last_similarity_analysis !== null && !empty($this->last_similarity_analysis['top_matches']); | |
| 6911 | - $has_action_data = isset($this->last_action_analysis) && !empty($this->last_action_analysis); | |
| 6912 | - | |
| 6913 | - if ($has_rag_data || $has_action_data) { | |
| 6914 | - $rag_context_for_storage = []; | |
| 6915 | - | |
| 6916 | - if ($has_rag_data) { | |
| 6917 | - $rag_context_for_storage['top_matches'] = $this->last_similarity_analysis['top_matches']; | |
| 6918 | - $rag_context_for_storage['approved_urls'] = $this->current_valid_urls ?? []; | |
| 6919 | - $rag_context_for_storage['similarity_threshold'] = $this->last_similarity_analysis['threshold_used'] ?? 0.35; | |
| 6920 | - $rag_context_for_storage['knowledge_base_type'] = $this->last_similarity_analysis['knowledge_base_type'] ?? 'WordPress Database'; | |
| 6921 | - $rag_context_for_storage['total_documents_checked'] = $this->last_similarity_analysis['total_checked'] ?? 0; | |
| 6922 | - } | |
| 6923 | - | |
| 6924 | - if ($has_action_data) { | |
| 6925 | - $rag_context_for_storage['action_analysis'] = $this->last_action_analysis; | |
| 6926 | - } | |
| 6927 | - } | |
| 6928 | - $this->mxchat_save_chat_message($session_id, 'bot', $full_response, null, $rag_context_for_storage); | |
| 5219 | + $this->mxchat_save_chat_message($session_id, 'bot', $full_response); | |
| 6929 | 5220 | } |
| 6930 | - | |
| 5221 | + | |
| 6931 | 5222 | return true; |
| 6932 | - | |
| 5223 | + | |
| 6933 | 5224 | } catch (Exception $e) { |
| 6934 | 5225 | $regular_response = $this->mxchat_generate_response_openrouter( |
| 6935 | 5226 | $selected_model, |
| 6936 | 5227 | $openrouter_api_key, |
| @@ -6957,9 +5248,9 @@ | ||
| 6957 | 5248 | try { |
| 6958 | 5249 | $bot_id = $this->get_current_bot_id($session_id); |
| 6959 | 5250 | |
| 6960 | 5251 | // Get system prompt instructions using centralized function |
| 6961 | - $system_prompt_instructions = $this->get_system_instructions($bot_id, $session_id); | |
| 5252 | + $system_prompt_instructions = $this->get_system_instructions($bot_id); | |
| 6962 | 5253 | |
| 6963 | 5254 | // Ensure conversation_history is an array |
| 6964 | 5255 | if (!is_array($conversation_history)) { |
| 6965 | 5256 | $conversation_history = array(); |
| @@ -7021,9 +5312,8 @@ | ||
| 7021 | 5312 | |
| 7022 | 5313 | // Check if this is a GPT-5 model (supports reasoning_effort parameter) |
| 7023 | 5314 | $is_gpt5_model = ( |
| 7024 | 5315 | strpos($selected_model, 'gpt-5') === 0 || |
| 7025 | - $selected_model === 'gpt-5.2' || | |
| 7026 | 5316 | $selected_model === 'gpt-5.1-2025-11-13' || |
| 7027 | 5317 | $selected_model === 'gpt-5' || |
| 7028 | 5318 | $selected_model === 'gpt-5-mini' || |
| 7029 | 5319 | $selected_model === 'gpt-5-nano' |
| @@ -7036,27 +5326,20 @@ | ||
| 7036 | 5326 | 'temperature' => 1, |
| 7037 | 5327 | 'stream' => true |
| 7038 | 5328 | ]; |
| 7039 | 5329 | |
| 7040 | - // Add reasoning_effort only for GPT-5 models that support it | |
| 7041 | - // These chat models don't support reasoning_effort parameter | |
| 7042 | - $no_reasoning_models = array('gpt-5.2', 'gpt-5.1-chat-latest', 'gpt-5.3-chat-latest', 'gpt-5.4-mini', 'gpt-5.4-nano'); | |
| 7043 | - if ($is_gpt5_model && !in_array($selected_model, $no_reasoning_models, true)) { | |
| 5330 | + // Add reasoning_effort only for GPT-5 models | |
| 5331 | + if ($is_gpt5_model) { | |
| 7044 | 5332 | // GPT-5.1 uses 'low' instead of 'minimal' |
| 7045 | 5333 | if ($selected_model === 'gpt-5.1-2025-11-13') { |
| 7046 | 5334 | $request_body['reasoning_effort'] = 'low'; |
| 7047 | - } elseif ($selected_model === 'gpt-5.4') { | |
| 7048 | - $request_body['reasoning_effort'] = 'none'; | |
| 7049 | 5335 | } else { |
| 7050 | - $request_body['reasoning_effort'] = 'minimal'; | |
| 5336 | + $request_body['reasoning_effort'] = 'minimal'; // For other GPT-5 models | |
| 7051 | 5337 | } |
| 7052 | 5338 | } |
| 7053 | 5339 | |
| 7054 | 5340 | $body = json_encode($request_body); |
| 7055 | 5341 | |
| 7056 | - // Setup streaming headers now that we know we're actually streaming | |
| 7057 | - $this->setup_streaming_headers(); | |
| 7058 | - | |
| 7059 | 5342 | // Use cURL for streaming support |
| 7060 | 5343 | $ch = curl_init(); |
| 7061 | 5344 | curl_setopt($ch, CURLOPT_URL, 'https://api.openai.com/v1/chat/completions'); |
| 7062 | 5345 | curl_setopt($ch, CURLOPT_RETURNTRANSFER, false); |
| @@ -7171,36 +5454,16 @@ | ||
| 7171 | 5454 | return true; |
| 7172 | 5455 | } |
| 7173 | 5456 | |
| 7174 | 5457 | curl_close($ch); |
| 7175 | - | |
| 5458 | + | |
| 7176 | 5459 | // Save the complete response to maintain chat persistence |
| 7177 | 5460 | if (!empty($full_response) && !empty($session_id)) { |
| 7178 | - // Prepare RAG context for streaming response | |
| 7179 | - $rag_context_for_storage = null; | |
| 7180 | - $has_rag_data = $this->last_similarity_analysis !== null && !empty($this->last_similarity_analysis['top_matches']); | |
| 7181 | - $has_action_data = isset($this->last_action_analysis) && !empty($this->last_action_analysis); | |
| 7182 | - | |
| 7183 | - if ($has_rag_data || $has_action_data) { | |
| 7184 | - $rag_context_for_storage = []; | |
| 7185 | - | |
| 7186 | - if ($has_rag_data) { | |
| 7187 | - $rag_context_for_storage['top_matches'] = $this->last_similarity_analysis['top_matches']; | |
| 7188 | - $rag_context_for_storage['approved_urls'] = $this->current_valid_urls ?? []; | |
| 7189 | - $rag_context_for_storage['similarity_threshold'] = $this->last_similarity_analysis['threshold_used'] ?? 0.35; | |
| 7190 | - $rag_context_for_storage['knowledge_base_type'] = $this->last_similarity_analysis['knowledge_base_type'] ?? 'WordPress Database'; | |
| 7191 | - $rag_context_for_storage['total_documents_checked'] = $this->last_similarity_analysis['total_checked'] ?? 0; | |
| 7192 | - } | |
| 7193 | - | |
| 7194 | - if ($has_action_data) { | |
| 7195 | - $rag_context_for_storage['action_analysis'] = $this->last_action_analysis; | |
| 7196 | - } | |
| 7197 | - } | |
| 7198 | - $this->mxchat_save_chat_message($session_id, 'bot', $full_response, null, $rag_context_for_storage); | |
| 5461 | + $this->mxchat_save_chat_message($session_id, 'bot', $full_response); | |
| 7199 | 5462 | } |
| 7200 | - | |
| 5463 | + | |
| 7201 | 5464 | return true; // Indicate streaming completed successfully |
| 7202 | - | |
| 5465 | + | |
| 7203 | 5466 | } catch (Exception $e) { |
| 7204 | 5467 | // Fallback to regular response |
| 7205 | 5468 | $regular_response = $this->mxchat_generate_response_openai( |
| 7206 | 5469 | $selected_model, |
| @@ -7238,363 +5501,8 @@ | ||
| 7238 | 5501 | echo json_encode($response_data); |
| 7239 | 5502 | return true; |
| 7240 | 5503 | } |
| 7241 | 5504 | } |
| 7242 | - | |
| 7243 | -/** | |
| 7244 | - * Generate response using OpenAI Responses API with web search tool | |
| 7245 | - * This uses the newer Responses API which supports web search functionality | |
| 7246 | - */ | |
| 7247 | -private function mxchat_generate_response_openai_web_search($selected_model, $api_key, $conversation_history, $relevant_content, $session_id, $testing_data = null, $streaming = false) { | |
| 7248 | - try { | |
| 7249 | - $bot_id = $this->get_current_bot_id($session_id); | |
| 7250 | - $system_prompt_instructions = $this->get_system_instructions($bot_id, $session_id); | |
| 7251 | - | |
| 7252 | - if (!is_array($conversation_history)) { | |
| 7253 | - $conversation_history = array(); | |
| 7254 | - } | |
| 7255 | - | |
| 7256 | - // Build the input for Responses API | |
| 7257 | - // The Responses API uses a different format - we need to construct the input properly | |
| 7258 | - $input_parts = []; | |
| 7259 | - | |
| 7260 | - // Add system instructions as context | |
| 7261 | - $system_context = $system_prompt_instructions . "\n\n" . $relevant_content; | |
| 7262 | - | |
| 7263 | - // Build conversation as input items for Responses API | |
| 7264 | - foreach ($conversation_history as $message) { | |
| 7265 | - if (is_array($message) && isset($message['role']) && isset($message['content'])) { | |
| 7266 | - $role = $message['role']; | |
| 7267 | - if ($role === 'bot' || $role === 'agent') { | |
| 7268 | - $role = 'assistant'; | |
| 7269 | - } | |
| 7270 | - if (!in_array($role, ['assistant', 'user'])) { | |
| 7271 | - $role = 'user'; | |
| 7272 | - } | |
| 7273 | - $input_parts[] = [ | |
| 7274 | - 'type' => 'message', | |
| 7275 | - 'role' => $role, | |
| 7276 | - 'content' => $message['content'] | |
| 7277 | - ]; | |
| 7278 | - } | |
| 7279 | - } | |
| 7280 | - | |
| 7281 | - // Build request body for Responses API | |
| 7282 | - $request_body = [ | |
| 7283 | - 'model' => $selected_model, | |
| 7284 | - 'input' => $input_parts, | |
| 7285 | - 'instructions' => $system_context, | |
| 7286 | - 'stream' => $streaming | |
| 7287 | - ]; | |
| 7288 | - | |
| 7289 | - // Only add web search tool if web search is enabled in settings | |
| 7290 | - $web_search_enabled = isset($this->options['enable_web_search']) && $this->options['enable_web_search'] === 'on'; | |
| 7291 | - if ($web_search_enabled) { | |
| 7292 | - $request_body['tools'] = [ | |
| 7293 | - ['type' => 'web_search'] | |
| 7294 | - ]; | |
| 7295 | - } | |
| 7296 | - | |
| 7297 | - // Add reasoning effort for supported models | |
| 7298 | - $is_gpt5_model = strpos($selected_model, 'gpt-5') === 0; | |
| 7299 | - $no_reasoning_web = array('gpt-5.2', 'gpt-5.3-chat-latest', 'gpt-5.4-mini', 'gpt-5.4-nano'); | |
| 7300 | - if ($is_gpt5_model && !in_array($selected_model, $no_reasoning_web, true)) { | |
| 7301 | - if ($selected_model === 'gpt-5.1-2025-11-13') { | |
| 7302 | - $request_body['reasoning'] = ['effort' => 'low']; | |
| 7303 | - } elseif ($selected_model === 'gpt-5.4') { | |
| 7304 | - $request_body['reasoning'] = ['effort' => 'low']; | |
| 7305 | - } | |
| 7306 | - } | |
| 7307 | - | |
| 7308 | - //error_log("MXCHAT WEB SEARCH: Request body: " . json_encode($request_body)); | |
| 7309 | - | |
| 7310 | - if ($streaming) { | |
| 7311 | - return $this->mxchat_web_search_streaming_response($request_body, $api_key, $session_id, $testing_data); | |
| 7312 | - } else { | |
| 7313 | - return $this->mxchat_web_search_non_streaming_response($request_body, $api_key, $session_id, $testing_data); | |
| 7314 | - } | |
| 7315 | - | |
| 7316 | - } catch (Exception $e) { | |
| 7317 | - //error_log("MXCHAT WEB SEARCH ERROR: " . $e->getMessage()); | |
| 7318 | - return [ | |
| 7319 | - 'error' => sprintf(esc_html__('Web search error: %s', 'mxchat'), esc_html($e->getMessage())), | |
| 7320 | - 'error_code' => 'web_search_exception' | |
| 7321 | - ]; | |
| 7322 | - } | |
| 7323 | -} | |
| 7324 | - | |
| 7325 | -/** | |
| 7326 | - * Handle non-streaming web search response | |
| 7327 | - */ | |
| 7328 | -private function mxchat_web_search_non_streaming_response($request_body, $api_key, $session_id, $testing_data) { | |
| 7329 | - $request_body['stream'] = false; | |
| 7330 | - | |
| 7331 | - $response = wp_remote_post('https://api.openai.com/v1/responses', array( | |
| 7332 | - 'headers' => array( | |
| 7333 | - 'Authorization' => 'Bearer ' . $api_key, | |
| 7334 | - 'Content-Type' => 'application/json' | |
| 7335 | - ), | |
| 7336 | - 'body' => json_encode($request_body), | |
| 7337 | - 'timeout' => 90 | |
| 7338 | - )); | |
| 7339 | - | |
| 7340 | - if (is_wp_error($response)) { | |
| 7341 | - //error_log("MXCHAT WEB SEARCH ERROR: WP Error: " . $response->get_error_message()); | |
| 7342 | - return [ | |
| 7343 | - 'error' => esc_html__('Failed to connect to OpenAI web search API', 'mxchat'), | |
| 7344 | - 'error_code' => 'web_search_connection_error' | |
| 7345 | - ]; | |
| 7346 | - } | |
| 7347 | - | |
| 7348 | - $response_code = wp_remote_retrieve_response_code($response); | |
| 7349 | - $response_body = wp_remote_retrieve_body($response); | |
| 7350 | - | |
| 7351 | - //error_log("MXCHAT WEB SEARCH: Response code: " . $response_code); | |
| 7352 | - //error_log("MXCHAT WEB SEARCH: Response body (first 2000): " . substr($response_body, 0, 2000)); | |
| 7353 | - | |
| 7354 | - if ($response_code !== 200) { | |
| 7355 | - $error_data = json_decode($response_body, true); | |
| 7356 | - $error_message = $error_data['error']['message'] ?? 'Unknown API error'; | |
| 7357 | - return [ | |
| 7358 | - 'error' => sprintf(esc_html__('OpenAI API error: %s', 'mxchat'), esc_html($error_message)), | |
| 7359 | - 'error_code' => 'web_search_api_error' | |
| 7360 | - ]; | |
| 7361 | - } | |
| 7362 | - | |
| 7363 | - $result = json_decode($response_body, true); | |
| 7364 | - | |
| 7365 | - if (json_last_error() !== JSON_ERROR_NONE) { | |
| 7366 | - return [ | |
| 7367 | - 'error' => esc_html__('Invalid response from OpenAI', 'mxchat'), | |
| 7368 | - 'error_code' => 'web_search_json_error' | |
| 7369 | - ]; | |
| 7370 | - } | |
| 7371 | - | |
| 7372 | - // Extract the response text and citations from Responses API format | |
| 7373 | - $output_text = ''; | |
| 7374 | - $citations = []; | |
| 7375 | - | |
| 7376 | - if (isset($result['output'])) { | |
| 7377 | - foreach ($result['output'] as $output_item) { | |
| 7378 | - if ($output_item['type'] === 'message' && isset($output_item['content'])) { | |
| 7379 | - foreach ($output_item['content'] as $content_item) { | |
| 7380 | - if ($content_item['type'] === 'output_text') { | |
| 7381 | - $output_text .= $content_item['text']; | |
| 7382 | - | |
| 7383 | - // Extract citations/annotations | |
| 7384 | - if (isset($content_item['annotations'])) { | |
| 7385 | - foreach ($content_item['annotations'] as $annotation) { | |
| 7386 | - if ($annotation['type'] === 'url_citation') { | |
| 7387 | - $citations[] = [ | |
| 7388 | - 'url' => $annotation['url'], | |
| 7389 | - 'title' => $annotation['title'] ?? '' | |
| 7390 | - ]; | |
| 7391 | - } | |
| 7392 | - } | |
| 7393 | - } | |
| 7394 | - } | |
| 7395 | - } | |
| 7396 | - } | |
| 7397 | - } | |
| 7398 | - } | |
| 7399 | - | |
| 7400 | - // If we have citations, append them to the response | |
| 7401 | - if (!empty($citations)) { | |
| 7402 | - $output_text .= "\n\n**Sources:**\n"; | |
| 7403 | - $seen_urls = []; | |
| 7404 | - foreach ($citations as $citation) { | |
| 7405 | - if (!in_array($citation['url'], $seen_urls)) { | |
| 7406 | - $seen_urls[] = $citation['url']; | |
| 7407 | - $title = !empty($citation['title']) ? $citation['title'] : $citation['url']; | |
| 7408 | - $output_text .= "- [" . $title . "](" . $citation['url'] . ")\n"; | |
| 7409 | - } | |
| 7410 | - } | |
| 7411 | - } | |
| 7412 | - | |
| 7413 | - // Transcript save is handled by the main handler (mxchat_handle_chat_request) | |
| 7414 | - // which includes rag_context for the "sources" link in transcripts. | |
| 7415 | - | |
| 7416 | - return $output_text; | |
| 7417 | -} | |
| 7418 | - | |
| 7419 | -/** | |
| 7420 | - * Handle streaming web search response using Responses API | |
| 7421 | - */ | |
| 7422 | -private function mxchat_web_search_streaming_response($request_body, $api_key, $session_id, $testing_data) { | |
| 7423 | - $request_body['stream'] = true; | |
| 7424 | - | |
| 7425 | - // Check if we can stream | |
| 7426 | - if (headers_sent() || !function_exists('curl_init')) { | |
| 7427 | - // Fallback to non-streaming | |
| 7428 | - return $this->mxchat_web_search_non_streaming_response($request_body, $api_key, $session_id, $testing_data); | |
| 7429 | - } | |
| 7430 | - | |
| 7431 | - // Setup streaming headers | |
| 7432 | - $this->setup_streaming_headers(); | |
| 7433 | - | |
| 7434 | - $ch = curl_init(); | |
| 7435 | - curl_setopt($ch, CURLOPT_URL, 'https://api.openai.com/v1/responses'); | |
| 7436 | - curl_setopt($ch, CURLOPT_RETURNTRANSFER, false); | |
| 7437 | - curl_setopt($ch, CURLOPT_POST, true); | |
| 7438 | - curl_setopt($ch, CURLOPT_POSTFIELDS, json_encode($request_body)); | |
| 7439 | - curl_setopt($ch, CURLOPT_HTTPHEADER, array( | |
| 7440 | - 'Content-Type: application/json', | |
| 7441 | - 'Authorization: Bearer ' . $api_key | |
| 7442 | - )); | |
| 7443 | - curl_setopt($ch, CURLOPT_SSL_VERIFYPEER, true); | |
| 7444 | - curl_setopt($ch, CURLOPT_TIMEOUT, 120); | |
| 7445 | - | |
| 7446 | - $full_response = ''; | |
| 7447 | - $stream_started = false; | |
| 7448 | - $buffer = ''; | |
| 7449 | - $citations = []; | |
| 7450 | - | |
| 7451 | - curl_setopt($ch, CURLOPT_WRITEFUNCTION, function($ch, $data) use (&$full_response, &$stream_started, &$buffer, &$citations, $testing_data) { | |
| 7452 | - // Send testing data as first event if available | |
| 7453 | - if (!$stream_started && $testing_data !== null) { | |
| 7454 | - echo "data: " . json_encode(['testing_data' => $testing_data]) . "\n\n"; | |
| 7455 | - flush(); | |
| 7456 | - $stream_started = true; | |
| 7457 | - } | |
| 7458 | - | |
| 7459 | - $buffer .= $data; | |
| 7460 | - $lines = explode("\n", $buffer); | |
| 7461 | - $buffer = array_pop($lines); | |
| 7462 | - | |
| 7463 | - foreach ($lines as $line) { | |
| 7464 | - if (trim($line) === '') continue; | |
| 7465 | - if (strpos($line, 'data: ') !== 0) continue; | |
| 7466 | - | |
| 7467 | - $json_str = substr($line, 6); | |
| 7468 | - | |
| 7469 | - if (trim($json_str) === '[DONE]') { | |
| 7470 | - // Append citations if we have any | |
| 7471 | - if (!empty($citations)) { | |
| 7472 | - $citation_text = "\n\n**Sources:**\n"; | |
| 7473 | - $seen_urls = []; | |
| 7474 | - foreach ($citations as $citation) { | |
| 7475 | - if (!in_array($citation['url'], $seen_urls)) { | |
| 7476 | - $seen_urls[] = $citation['url']; | |
| 7477 | - $title = !empty($citation['title']) ? $citation['title'] : $citation['url']; | |
| 7478 | - $citation_text .= "- [" . $title . "](" . $citation['url'] . ")\n"; | |
| 7479 | - } | |
| 7480 | - } | |
| 7481 | - echo "data: " . json_encode(['content' => $citation_text]) . "\n\n"; | |
| 7482 | - $full_response .= $citation_text; | |
| 7483 | - flush(); | |
| 7484 | - } | |
| 7485 | - echo "data: [DONE]\n\n"; | |
| 7486 | - flush(); | |
| 7487 | - continue; | |
| 7488 | - } | |
| 7489 | - | |
| 7490 | - $json = json_decode(trim($json_str), true); | |
| 7491 | - if (!$json) continue; | |
| 7492 | - | |
| 7493 | - // Handle Responses API streaming events | |
| 7494 | - // The format is different from Chat Completions | |
| 7495 | - if (isset($json['type'])) { | |
| 7496 | - switch ($json['type']) { | |
| 7497 | - case 'response.output_text.delta': | |
| 7498 | - // Text content delta | |
| 7499 | - if (isset($json['delta'])) { | |
| 7500 | - $content = $json['delta']; | |
| 7501 | - $full_response .= $content; | |
| 7502 | - echo "data: " . json_encode(['content' => $content]) . "\n\n"; | |
| 7503 | - flush(); | |
| 7504 | - } | |
| 7505 | - break; | |
| 7506 | - | |
| 7507 | - case 'response.output_item.done': | |
| 7508 | - // Check for citations in completed items | |
| 7509 | - if (isset($json['item']['content'])) { | |
| 7510 | - foreach ($json['item']['content'] as $content_item) { | |
| 7511 | - if (isset($content_item['annotations'])) { | |
| 7512 | - foreach ($content_item['annotations'] as $annotation) { | |
| 7513 | - if ($annotation['type'] === 'url_citation') { | |
| 7514 | - $citations[] = [ | |
| 7515 | - 'url' => $annotation['url'], | |
| 7516 | - 'title' => $annotation['title'] ?? '' | |
| 7517 | - ]; | |
| 7518 | - } | |
| 7519 | - } | |
| 7520 | - } | |
| 7521 | - } | |
| 7522 | - } | |
| 7523 | - break; | |
| 7524 | - } | |
| 7525 | - } | |
| 7526 | - } | |
| 7527 | - | |
| 7528 | - return strlen($data); | |
| 7529 | - }); | |
| 7530 | - | |
| 7531 | - $response = curl_exec($ch); | |
| 7532 | - $http_code = curl_getinfo($ch, CURLINFO_HTTP_CODE); | |
| 7533 | - | |
| 7534 | - if (curl_errno($ch) || $http_code !== 200) { | |
| 7535 | - $curl_error = curl_error($ch); | |
| 7536 | - curl_close($ch); | |
| 7537 | - | |
| 7538 | - //error_log("MXCHAT WEB SEARCH STREAM ERROR: HTTP $http_code, cURL error: $curl_error"); | |
| 7539 | - | |
| 7540 | - // Fallback to non-streaming | |
| 7541 | - $fallback_response = $this->mxchat_web_search_non_streaming_response($request_body, $api_key, $session_id, $testing_data); | |
| 7542 | - | |
| 7543 | - if (is_array($fallback_response) && isset($fallback_response['error'])) { | |
| 7544 | - echo "data: " . json_encode([ | |
| 7545 | - 'error' => true, | |
| 7546 | - 'error_message' => $fallback_response['error'], | |
| 7547 | - 'error_code' => $fallback_response['error_code'] ?? 'web_search_error' | |
| 7548 | - ]) . "\n\n"; | |
| 7549 | - echo "data: [DONE]\n\n"; | |
| 7550 | - flush(); | |
| 7551 | - return true; | |
| 7552 | - } | |
| 7553 | - | |
| 7554 | - $response_data = [ | |
| 7555 | - 'text' => $fallback_response, | |
| 7556 | - 'html' => '', | |
| 7557 | - 'session_id' => $session_id | |
| 7558 | - ]; | |
| 7559 | - if ($testing_data !== null) { | |
| 7560 | - $response_data['testing_data'] = $testing_data; | |
| 7561 | - } | |
| 7562 | - header('Content-Type: application/json'); | |
| 7563 | - echo json_encode($response_data); | |
| 7564 | - return true; | |
| 7565 | - } | |
| 7566 | - | |
| 7567 | - curl_close($ch); | |
| 7568 | - | |
| 7569 | - // Save the complete response with RAG context so the "sources" link | |
| 7570 | - // appears in transcripts — mirrors the pattern used by Claude/OpenAI streaming. | |
| 7571 | - if (!empty($full_response) && !empty($session_id)) { | |
| 7572 | - $rag_context_for_storage = null; | |
| 7573 | - $has_rag_data = $this->last_similarity_analysis !== null && !empty($this->last_similarity_analysis['top_matches']); | |
| 7574 | - $has_action_data = isset($this->last_action_analysis) && !empty($this->last_action_analysis); | |
| 7575 | - | |
| 7576 | - if ($has_rag_data || $has_action_data) { | |
| 7577 | - $rag_context_for_storage = []; | |
| 7578 | - | |
| 7579 | - if ($has_rag_data) { | |
| 7580 | - $rag_context_for_storage['top_matches'] = $this->last_similarity_analysis['top_matches']; | |
| 7581 | - $rag_context_for_storage['approved_urls'] = $this->current_valid_urls ?? []; | |
| 7582 | - $rag_context_for_storage['similarity_threshold'] = $this->last_similarity_analysis['threshold_used'] ?? 0.35; | |
| 7583 | - $rag_context_for_storage['knowledge_base_type'] = $this->last_similarity_analysis['knowledge_base_type'] ?? 'WordPress Database'; | |
| 7584 | - $rag_context_for_storage['total_documents_checked'] = $this->last_similarity_analysis['total_checked'] ?? 0; | |
| 7585 | - } | |
| 7586 | - | |
| 7587 | - if ($has_action_data) { | |
| 7588 | - $rag_context_for_storage['action_analysis'] = $this->last_action_analysis; | |
| 7589 | - } | |
| 7590 | - } | |
| 7591 | - $this->mxchat_save_chat_message($session_id, 'bot', $full_response, null, $rag_context_for_storage); | |
| 7592 | - } | |
| 7593 | - | |
| 7594 | - return true; | |
| 7595 | -} | |
| 7596 | - | |
| 7597 | 5505 | private function mxchat_generate_response_claude_stream($selected_model, $claude_api_key, $conversation_history, $relevant_content, $session_id, $testing_data = null) { |
| 7598 | 5506 | try { |
| 7599 | 5507 | // Get bot ID from session or request |
| 7600 | 5508 | $bot_id = $this->get_current_bot_id($session_id); |
| @@ -7599,9 +5507,9 @@ | ||
| 7599 | 5507 | // Get bot ID from session or request |
| 7600 | 5508 | $bot_id = $this->get_current_bot_id($session_id); |
| 7601 | 5509 | |
| 7602 | 5510 | // Get system prompt instructions using centralized function |
| 7603 | - $system_prompt_instructions = $this->get_system_instructions($bot_id, $session_id); | |
| 5511 | + $system_prompt_instructions = $this->get_system_instructions($bot_id); | |
| 7604 | 5512 | // Ensure conversation_history is an array |
| 7605 | 5513 | if (!is_array($conversation_history)) { |
| 7606 | 5514 | $conversation_history = array(); |
| 7607 | 5515 | } |
| @@ -7678,11 +5586,8 @@ | ||
| 7678 | 5586 | echo json_encode($response_data); |
| 7679 | 5587 | return true; // Indicate we handled the response |
| 7680 | 5588 | } |
| 7681 | 5589 | |
| 7682 | - // Setup streaming headers now that we know we're actually streaming | |
| 7683 | - $this->setup_streaming_headers(); | |
| 7684 | - | |
| 7685 | 5590 | // Use cURL for streaming support |
| 7686 | 5591 | $ch = curl_init(); |
| 7687 | 5592 | curl_setopt($ch, CURLOPT_URL, 'https://api.anthropic.com/v1/messages'); |
| 7688 | 5593 | curl_setopt($ch, CURLOPT_RETURNTRANSFER, false); |
| @@ -7821,29 +5726,9 @@ | ||
| 7821 | 5726 | } |
| 7822 | 5727 | |
| 7823 | 5728 | // Save the complete response to maintain chat persistence |
| 7824 | 5729 | if (!empty($full_response) && !empty($session_id)) { |
| 7825 | - // Prepare RAG context for streaming response | |
| 7826 | - $rag_context_for_storage = null; | |
| 7827 | - $has_rag_data = $this->last_similarity_analysis !== null && !empty($this->last_similarity_analysis['top_matches']); | |
| 7828 | - $has_action_data = isset($this->last_action_analysis) && !empty($this->last_action_analysis); | |
| 7829 | - | |
| 7830 | - if ($has_rag_data || $has_action_data) { | |
| 7831 | - $rag_context_for_storage = []; | |
| 7832 | - | |
| 7833 | - if ($has_rag_data) { | |
| 7834 | - $rag_context_for_storage['top_matches'] = $this->last_similarity_analysis['top_matches']; | |
| 7835 | - $rag_context_for_storage['approved_urls'] = $this->current_valid_urls ?? []; | |
| 7836 | - $rag_context_for_storage['similarity_threshold'] = $this->last_similarity_analysis['threshold_used'] ?? 0.35; | |
| 7837 | - $rag_context_for_storage['knowledge_base_type'] = $this->last_similarity_analysis['knowledge_base_type'] ?? 'WordPress Database'; | |
| 7838 | - $rag_context_for_storage['total_documents_checked'] = $this->last_similarity_analysis['total_checked'] ?? 0; | |
| 7839 | - } | |
| 7840 | - | |
| 7841 | - if ($has_action_data) { | |
| 7842 | - $rag_context_for_storage['action_analysis'] = $this->last_action_analysis; | |
| 7843 | - } | |
| 7844 | - } | |
| 7845 | - $this->mxchat_save_chat_message($session_id, 'bot', $full_response, null, $rag_context_for_storage); | |
| 5730 | + $this->mxchat_save_chat_message($session_id, 'bot', $full_response); | |
| 7846 | 5731 | } |
| 7847 | 5732 | |
| 7848 | 5733 | return true; // Indicate streaming completed successfully |
| 7849 | 5734 | |
| @@ -7894,9 +5779,9 @@ | ||
| 7894 | 5779 | // Get bot ID from session or request |
| 7895 | 5780 | $bot_id = $this->get_current_bot_id($session_id); |
| 7896 | 5781 | |
| 7897 | 5782 | // Get system prompt instructions using centralized function |
| 7898 | - $system_prompt_instructions = $this->get_system_instructions($bot_id, $session_id); | |
| 5783 | + $system_prompt_instructions = $this->get_system_instructions($bot_id); | |
| 7899 | 5784 | |
| 7900 | 5785 | // Ensure conversation_history is an array |
| 7901 | 5786 | if (!is_array($conversation_history)) { |
| 7902 | 5787 | $conversation_history = array(); |
| @@ -7965,11 +5850,8 @@ | ||
| 7965 | 5850 | 'temperature' => 0.8, |
| 7966 | 5851 | 'stream' => true |
| 7967 | 5852 | ]); |
| 7968 | 5853 | |
| 7969 | - // Setup streaming headers now that we know we're actually streaming | |
| 7970 | - $this->setup_streaming_headers(); | |
| 7971 | - | |
| 7972 | 5854 | // Use cURL for streaming support |
| 7973 | 5855 | $ch = curl_init(); |
| 7974 | 5856 | curl_setopt($ch, CURLOPT_URL, 'https://api.x.ai/v1/chat/completions'); |
| 7975 | 5857 | curl_setopt($ch, CURLOPT_RETURNTRANSFER, false); |
| @@ -8070,39 +5952,19 @@ | ||
| 8070 | 5952 | return true; |
| 8071 | 5953 | } |
| 8072 | 5954 | |
| 8073 | 5955 | curl_close($ch); |
| 8074 | - | |
| 5956 | + | |
| 8075 | 5957 | // Save the complete response to maintain chat persistence |
| 8076 | 5958 | if (!empty($full_response) && !empty($session_id)) { |
| 8077 | - // Prepare RAG context for streaming response | |
| 8078 | - $rag_context_for_storage = null; | |
| 8079 | - $has_rag_data = $this->last_similarity_analysis !== null && !empty($this->last_similarity_analysis['top_matches']); | |
| 8080 | - $has_action_data = isset($this->last_action_analysis) && !empty($this->last_action_analysis); | |
| 8081 | - | |
| 8082 | - if ($has_rag_data || $has_action_data) { | |
| 8083 | - $rag_context_for_storage = []; | |
| 8084 | - | |
| 8085 | - if ($has_rag_data) { | |
| 8086 | - $rag_context_for_storage['top_matches'] = $this->last_similarity_analysis['top_matches']; | |
| 8087 | - $rag_context_for_storage['approved_urls'] = $this->current_valid_urls ?? []; | |
| 8088 | - $rag_context_for_storage['similarity_threshold'] = $this->last_similarity_analysis['threshold_used'] ?? 0.35; | |
| 8089 | - $rag_context_for_storage['knowledge_base_type'] = $this->last_similarity_analysis['knowledge_base_type'] ?? 'WordPress Database'; | |
| 8090 | - $rag_context_for_storage['total_documents_checked'] = $this->last_similarity_analysis['total_checked'] ?? 0; | |
| 8091 | - } | |
| 8092 | - | |
| 8093 | - if ($has_action_data) { | |
| 8094 | - $rag_context_for_storage['action_analysis'] = $this->last_action_analysis; | |
| 8095 | - } | |
| 8096 | - } | |
| 8097 | - $this->mxchat_save_chat_message($session_id, 'bot', $full_response, null, $rag_context_for_storage); | |
| 5959 | + $this->mxchat_save_chat_message($session_id, 'bot', $full_response); | |
| 8098 | 5960 | } |
| 8099 | - | |
| 5961 | + | |
| 8100 | 5962 | return true; // Indicate streaming completed successfully |
| 8101 | - | |
| 5963 | + | |
| 8102 | 5964 | } catch (Exception $e) { |
| 8103 | 5965 | //error_log("MxChat X.AI streaming exception: " . $e->getMessage()); |
| 8104 | - | |
| 5966 | + | |
| 8105 | 5967 | // Fallback to regular response |
| 8106 | 5968 | $regular_response = $this->mxchat_generate_response_xai( |
| 8107 | 5969 | $selected_model, |
| 8108 | 5970 | $xai_api_key, |
| @@ -8131,9 +5993,9 @@ | ||
| 8131 | 5993 | // Get bot ID from session or request |
| 8132 | 5994 | $bot_id = $this->get_current_bot_id($session_id); |
| 8133 | 5995 | |
| 8134 | 5996 | // Get system prompt instructions using centralized function |
| 8135 | - $system_prompt_instructions = $this->get_system_instructions($bot_id, $session_id); | |
| 5997 | + $system_prompt_instructions = $this->get_system_instructions($bot_id); | |
| 8136 | 5998 | |
| 8137 | 5999 | // Ensure conversation_history is an array |
| 8138 | 6000 | if (!is_array($conversation_history)) { |
| 8139 | 6001 | $conversation_history = array(); |
| @@ -8202,11 +6064,8 @@ | ||
| 8202 | 6064 | 'temperature' => 0.8, |
| 8203 | 6065 | 'stream' => true |
| 8204 | 6066 | ]); |
| 8205 | 6067 | |
| 8206 | - // Setup streaming headers now that we know we're actually streaming | |
| 8207 | - $this->setup_streaming_headers(); | |
| 8208 | - | |
| 8209 | 6068 | // Use cURL for streaming support |
| 8210 | 6069 | $ch = curl_init(); |
| 8211 | 6070 | curl_setopt($ch, CURLOPT_URL, 'https://api.deepseek.com/v1/chat/completions'); |
| 8212 | 6071 | curl_setopt($ch, CURLOPT_RETURNTRANSFER, false); |
| @@ -8321,39 +6180,19 @@ | ||
| 8321 | 6180 | return true; |
| 8322 | 6181 | } |
| 8323 | 6182 | |
| 8324 | 6183 | curl_close($ch); |
| 8325 | - | |
| 6184 | + | |
| 8326 | 6185 | // Save the complete response to maintain chat persistence |
| 8327 | 6186 | if (!empty($full_response) && !empty($session_id)) { |
| 8328 | - // Prepare RAG context for streaming response | |
| 8329 | - $rag_context_for_storage = null; | |
| 8330 | - $has_rag_data = $this->last_similarity_analysis !== null && !empty($this->last_similarity_analysis['top_matches']); | |
| 8331 | - $has_action_data = isset($this->last_action_analysis) && !empty($this->last_action_analysis); | |
| 8332 | - | |
| 8333 | - if ($has_rag_data || $has_action_data) { | |
| 8334 | - $rag_context_for_storage = []; | |
| 8335 | - | |
| 8336 | - if ($has_rag_data) { | |
| 8337 | - $rag_context_for_storage['top_matches'] = $this->last_similarity_analysis['top_matches']; | |
| 8338 | - $rag_context_for_storage['approved_urls'] = $this->current_valid_urls ?? []; | |
| 8339 | - $rag_context_for_storage['similarity_threshold'] = $this->last_similarity_analysis['threshold_used'] ?? 0.35; | |
| 8340 | - $rag_context_for_storage['knowledge_base_type'] = $this->last_similarity_analysis['knowledge_base_type'] ?? 'WordPress Database'; | |
| 8341 | - $rag_context_for_storage['total_documents_checked'] = $this->last_similarity_analysis['total_checked'] ?? 0; | |
| 8342 | - } | |
| 8343 | - | |
| 8344 | - if ($has_action_data) { | |
| 8345 | - $rag_context_for_storage['action_analysis'] = $this->last_action_analysis; | |
| 8346 | - } | |
| 8347 | - } | |
| 8348 | - $this->mxchat_save_chat_message($session_id, 'bot', $full_response, null, $rag_context_for_storage); | |
| 6187 | + $this->mxchat_save_chat_message($session_id, 'bot', $full_response); | |
| 8349 | 6188 | } |
| 8350 | - | |
| 6189 | + | |
| 8351 | 6190 | return true; // Indicate streaming completed successfully |
| 8352 | - | |
| 6191 | + | |
| 8353 | 6192 | } catch (Exception $e) { |
| 8354 | 6193 | //error_log("MxChat DeepSeek streaming exception: " . $e->getMessage()); |
| 8355 | - | |
| 6194 | + | |
| 8356 | 6195 | // Fallback to regular response |
| 8357 | 6196 | $regular_response = $this->mxchat_generate_response_deepseek( |
| 8358 | 6197 | $selected_model, |
| 8359 | 6198 | $deepseek_api_key, |
| @@ -8395,9 +6234,9 @@ | ||
| 8395 | 6234 | $conversation_history = array(); |
| 8396 | 6235 | } |
| 8397 | 6236 | |
| 8398 | 6237 | $bot_id = $this->get_current_bot_id(''); |
| 8399 | - $system_prompt_instructions = $this->get_system_instructions($bot_id, $session_id); | |
| 6238 | + $system_prompt_instructions = $this->get_system_instructions($bot_id); | |
| 8400 | 6239 | |
| 8401 | 6240 | $formatted_conversation = array(); |
| 8402 | 6241 | |
| 8403 | 6242 | $formatted_conversation[] = array( |
| @@ -8497,9 +6336,9 @@ | ||
| 8497 | 6336 | // Get bot ID from session or request |
| 8498 | 6337 | $bot_id = $this->get_current_bot_id($session_id); |
| 8499 | 6338 | |
| 8500 | 6339 | // Get system prompt instructions using centralized function |
| 8501 | - $system_prompt_instructions = $this->get_system_instructions($bot_id, $session_id); | |
| 6340 | + $system_prompt_instructions = $this->get_system_instructions($bot_id); | |
| 8502 | 6341 | |
| 8503 | 6342 | // Clean and validate conversation history |
| 8504 | 6343 | foreach ($conversation_history as &$message) { |
| 8505 | 6344 | // Convert bot and agent roles to assistant |
| @@ -8606,9 +6445,9 @@ | ||
| 8606 | 6445 | // Get bot ID from session or request |
| 8607 | 6446 | $bot_id = $this->get_current_bot_id(''); |
| 8608 | 6447 | |
| 8609 | 6448 | // Get system prompt instructions using centralized function |
| 8610 | - $system_prompt_instructions = $this->get_system_instructions($bot_id, $session_id); | |
| 6449 | + $system_prompt_instructions = $this->get_system_instructions($bot_id); | |
| 8611 | 6450 | |
| 8612 | 6451 | // Create a new array for the formatted conversation |
| 8613 | 6452 | $formatted_conversation = array(); |
| 8614 | 6453 | |
| @@ -8640,9 +6479,8 @@ | ||
| 8640 | 6479 | |
| 8641 | 6480 | // Check if this is a GPT-5 model (supports reasoning_effort parameter) |
| 8642 | 6481 | $is_gpt5_model = ( |
| 8643 | 6482 | strpos($selected_model, 'gpt-5') === 0 || |
| 8644 | - $selected_model === 'gpt-5.2' || | |
| 8645 | 6483 | $selected_model === 'gpt-5.1-2025-11-13' || |
| 8646 | 6484 | $selected_model === 'gpt-5' || |
| 8647 | 6485 | $selected_model === 'gpt-5-mini' || |
| 8648 | 6486 | $selected_model === 'gpt-5-nano' |
| @@ -8655,19 +6493,15 @@ | ||
| 8655 | 6493 | 'temperature' => 1, |
| 8656 | 6494 | 'stream' => false |
| 8657 | 6495 | ]; |
| 8658 | 6496 | |
| 8659 | - // Add reasoning_effort only for GPT-5 models that support it | |
| 8660 | - // These chat models don't support reasoning_effort parameter | |
| 8661 | - $no_reasoning_models = array('gpt-5.2', 'gpt-5.1-chat-latest', 'gpt-5.3-chat-latest', 'gpt-5.4-mini', 'gpt-5.4-nano'); | |
| 8662 | - if ($is_gpt5_model && !in_array($selected_model, $no_reasoning_models, true)) { | |
| 6497 | + // Add reasoning_effort only for GPT-5 models | |
| 6498 | + if ($is_gpt5_model) { | |
| 8663 | 6499 | // GPT-5.1 uses 'low' instead of 'minimal' |
| 8664 | 6500 | if ($selected_model === 'gpt-5.1-2025-11-13') { |
| 8665 | 6501 | $request_body['reasoning_effort'] = 'low'; |
| 8666 | - } elseif ($selected_model === 'gpt-5.4') { | |
| 8667 | - $request_body['reasoning_effort'] = 'none'; | |
| 8668 | 6502 | } else { |
| 8669 | - $request_body['reasoning_effort'] = 'minimal'; | |
| 6503 | + $request_body['reasoning_effort'] = 'minimal'; // For other GPT-5 models | |
| 8670 | 6504 | } |
| 8671 | 6505 | } |
| 8672 | 6506 | |
| 8673 | 6507 | $body = json_encode($request_body); |
| @@ -8778,9 +6612,9 @@ | ||
| 8778 | 6612 | // Get bot ID from session or request |
| 8779 | 6613 | $bot_id = $this->get_current_bot_id($session_id); |
| 8780 | 6614 | |
| 8781 | 6615 | // Get system prompt instructions using centralized function |
| 8782 | - $system_prompt_instructions = $this->get_system_instructions($bot_id, $session_id); | |
| 6616 | + $system_prompt_instructions = $this->get_system_instructions($bot_id); | |
| 8783 | 6617 | |
| 8784 | 6618 | // Add system prompt to relevant content |
| 8785 | 6619 | $content_with_instructions = $system_prompt_instructions . " " . $relevant_content; |
| 8786 | 6620 | |
| @@ -8977,9 +6811,9 @@ | ||
| 8977 | 6811 | // Get bot ID from session or request |
| 8978 | 6812 | $bot_id = $this->get_current_bot_id($session_id); |
| 8979 | 6813 | |
| 8980 | 6814 | // Get system prompt instructions using centralized function |
| 8981 | - $system_prompt_instructions = $this->get_system_instructions($bot_id, $session_id); | |
| 6815 | + $system_prompt_instructions = $this->get_system_instructions($bot_id); | |
| 8982 | 6816 | |
| 8983 | 6817 | // Create a new array for the formatted conversation |
| 8984 | 6818 | $formatted_conversation = array(); |
| 8985 | 6819 | |
| @@ -9136,9 +6970,9 @@ | ||
| 9136 | 6970 | // Get bot ID from session or request |
| 9137 | 6971 | $bot_id = $this->get_current_bot_id($session_id); |
| 9138 | 6972 | |
| 9139 | 6973 | // Get system prompt instructions using centralized function |
| 9140 | - $system_prompt_instructions = $this->get_system_instructions($bot_id, $session_id); | |
| 6974 | + $system_prompt_instructions = $this->get_system_instructions($bot_id); | |
| 9141 | 6975 | |
| 9142 | 6976 | // Add system prompt to relevant content |
| 9143 | 6977 | $content_with_instructions = $system_prompt_instructions . " " . $relevant_content; |
| 9144 | 6978 | |
| @@ -9234,11 +7068,9 @@ | ||
| 9234 | 7068 | ] |
| 9235 | 7069 | ]); |
| 9236 | 7070 | |
| 9237 | 7071 | // Prepare the API endpoint |
| 9238 | - // Use v1beta for preview models (Gemini 3, experimental), v1 for stable models | |
| 9239 | - $api_version = (strpos($selected_model, 'preview') !== false || strpos($selected_model, 'exp') !== false) ? 'v1beta' : 'v1'; | |
| 9240 | - $api_endpoint = 'https://generativelanguage.googleapis.com/' . $api_version . '/models/' . $selected_model . ':generateContent?key=' . $gemini_api_key; | |
| 7072 | + $api_endpoint = 'https://generativelanguage.googleapis.com/v1/models/' . $selected_model . ':generateContent?key=' . $gemini_api_key; | |
| 9241 | 7073 | |
| 9242 | 7074 | // Set up the API request |
| 9243 | 7075 | $args = [ |
| 9244 | 7076 | 'body' => $body, |
| @@ -9280,9 +7112,9 @@ | ||
| 9280 | 7112 | |
| 9281 | 7113 | |
| 9282 | 7114 | public function test_streaming_request() { |
| 9283 | 7115 | $options = get_option('mxchat_options', []); |
| 9284 | - $model = $options['model'] ?? 'gpt-5.1-chat-latest'; | |
| 7116 | + $model = $options['model'] ?? 'gpt-4o'; | |
| 9285 | 7117 | |
| 9286 | 7118 | // Detect provider from model prefix |
| 9287 | 7119 | $provider = strtolower(explode('-', $model)[0]); |
| 9288 | 7120 | |
| @@ -9465,13 +7297,17 @@ | ||
| 9465 | 7297 | } |
| 9466 | 7298 | |
| 9467 | 7299 | |
| 9468 | 7300 | public function mxchat_enqueue_scripts_styles() { |
| 9469 | - // Fetch options from the database first to check loading strategy | |
| 9470 | - $this->options = get_option('mxchat_options'); | |
| 9471 | - $loading_strategy = isset($this->options['script_loading_strategy']) ? $this->options['script_loading_strategy'] : 'default'; | |
| 9472 | - | |
| 9473 | - // Always enqueue CSS immediately | |
| 7301 | + // Enqueue the script | |
| 7302 | + wp_enqueue_script( | |
| 7303 | + 'mxchat-chat-js', | |
| 7304 | + plugin_dir_url(__FILE__) . '../js/chat-script.js', | |
| 7305 | + array('jquery'), | |
| 7306 | + MXCHAT_VERSION, | |
| 7307 | + true | |
| 7308 | + ); | |
| 7309 | + // Enqueue the CSS | |
| 9474 | 7310 | wp_enqueue_style( |
| 9475 | 7311 | 'mxchat-chat-css', |
| 9476 | 7312 | plugin_dir_url(__FILE__) . '../css/chat-style.css', |
| 9477 | 7313 | array(), |
| @@ -9476,43 +7312,17 @@ | ||
| 9476 | 7312 | plugin_dir_url(__FILE__) . '../css/chat-style.css', |
| 9477 | 7313 | array(), |
| 9478 | 7314 | MXCHAT_VERSION |
| 9479 | 7315 | ); |
| 9480 | - | |
| 9481 | - // Handle script loading based on strategy | |
| 9482 | - if ($loading_strategy === 'default' || $loading_strategy === 'defer') { | |
| 9483 | - // Enqueue the script normally | |
| 9484 | - wp_enqueue_script( | |
| 9485 | - 'mxchat-chat-js', | |
| 9486 | - plugin_dir_url(__FILE__) . '../js/chat-script.js', | |
| 9487 | - array('jquery'), | |
| 9488 | - MXCHAT_VERSION, | |
| 9489 | - true | |
| 9490 | - ); | |
| 9491 | - | |
| 9492 | - // Add defer attribute if strategy is 'defer' | |
| 9493 | - if ($loading_strategy === 'defer') { | |
| 9494 | - wp_script_add_data('mxchat-chat-js', 'strategy', 'defer'); | |
| 9495 | - } | |
| 9496 | - } else { | |
| 9497 | - // For delay or interaction-based loading, we'll use a custom loader | |
| 9498 | - // Don't enqueue the main script - we'll load it dynamically | |
| 9499 | - add_action('wp_footer', array($this, 'mxchat_output_delayed_script_loader'), 99); | |
| 9500 | - } | |
| 9501 | - | |
| 7316 | + // Fetch options from the database | |
| 7317 | + $this->options = get_option('mxchat_options'); | |
| 9502 | 7318 | $prompts_options = get_option('mxchat_prompts_options', array()); |
| 9503 | - | |
| 9504 | - // Check if AI theme is active - if so, skip inline colors in JavaScript | |
| 9505 | - $theme_options = get_option('mxchat_theme_options', array()); | |
| 9506 | - $ai_theme_active = !empty($theme_options['active_ai_theme_css']); | |
| 9507 | - $has_bot_theme_assignments = !empty($theme_options['bot_theme_assignments']); | |
| 9508 | - $skip_inline_colors = $ai_theme_active || $has_bot_theme_assignments; | |
| 9509 | - | |
| 7319 | + | |
| 9510 | 7320 | // Prepare settings for JavaScript |
| 9511 | 7321 | $style_settings = array( |
| 9512 | 7322 | 'ajax_url' => admin_url('admin-ajax.php'), |
| 9513 | 7323 | 'nonce' => wp_create_nonce('mxchat_chat_nonce'), |
| 9514 | - 'model' => isset($this->options['model']) ? $this->options['model'] : 'gpt-5.1-chat-latest', | |
| 7324 | + 'model' => isset($this->options['model']) ? $this->options['model'] : 'gpt-4o', | |
| 9515 | 7325 | 'enable_streaming_toggle' => isset($this->options['enable_streaming_toggle']) ? $this->options['enable_streaming_toggle'] : 'on', |
| 9516 | 7326 | 'contextual_awareness_toggle' => isset($this->options['contextual_awareness_toggle']) ? $this->options['contextual_awareness_toggle'] : 'off', |
| 9517 | 7327 | 'link_target_toggle' => $this->options['link_target_toggle'] ?? 'off', |
| 9518 | 7328 | 'rate_limit_message' => $this->options['rate_limit_message'] ?? 'Rate limit exceeded. Please try again later.', |
| @@ -9539,134 +7349,15 @@ | ||
| 9539 | 7349 | 'use_pinecone' => $prompts_options['mxchat_use_pinecone'] ?? '0', |
| 9540 | 7350 | 'email_collection_enabled' => $this->options['enable_email_block'] ?? 'off', // FIXED |
| 9541 | 7351 | 'initial_email_state' => null, // Also fixed this undefined variable |
| 9542 | 7352 | 'skip_email_check' => true, |
| 9543 | - 'pinecone_enabled' => isset($prompts_options['mxchat_use_pinecone']) && $prompts_options['mxchat_use_pinecone'] === '1', | |
| 9544 | - 'skip_inline_colors' => $skip_inline_colors, | |
| 9545 | - 'bot_theme_assignments' => $theme_options['bot_theme_assignments'] ?? array() | |
| 7353 | + 'pinecone_enabled' => isset($prompts_options['mxchat_use_pinecone']) && $prompts_options['mxchat_use_pinecone'] === '1' | |
| 9546 | 7354 | ); |
| 9547 | - | |
| 9548 | - // For normal/defer loading, use wp_localize_script | |
| 9549 | - // For delayed loading, we store settings in a transient to be output inline | |
| 9550 | - if ($loading_strategy === 'default' || $loading_strategy === 'defer') { | |
| 9551 | - wp_localize_script('mxchat-chat-js', 'mxchatChat', $style_settings); | |
| 9552 | - } else { | |
| 9553 | - // Store settings for the delayed loader to use | |
| 9554 | - set_transient('mxchat_delayed_settings_' . get_current_user_id(), $style_settings, 60); | |
| 9555 | - } | |
| 7355 | + // Pass the settings to the script | |
| 7356 | + wp_localize_script('mxchat-chat-js', 'mxchatChat', $style_settings); | |
| 9556 | 7357 | } |
| 9557 | 7358 | |
| 9558 | -/** | |
| 9559 | - * Output the delayed script loader for performance optimization | |
| 9560 | - */ | |
| 9561 | -public function mxchat_output_delayed_script_loader() { | |
| 9562 | - $this->options = get_option('mxchat_options'); | |
| 9563 | - $loading_strategy = isset($this->options['script_loading_strategy']) ? $this->options['script_loading_strategy'] : 'default'; | |
| 9564 | - $script_url = plugin_dir_url(__FILE__) . '../js/chat-script.js?ver=' . MXCHAT_VERSION; | |
| 9565 | 7359 | |
| 9566 | - // Get the stored settings | |
| 9567 | - $prompts_options = get_option('mxchat_prompts_options', array()); | |
| 9568 | - $theme_options = get_option('mxchat_theme_options', array()); | |
| 9569 | - $ai_theme_active = !empty($theme_options['active_ai_theme_css']); | |
| 9570 | - $has_bot_theme_assignments = !empty($theme_options['bot_theme_assignments']); | |
| 9571 | - $skip_inline_colors = $ai_theme_active || $has_bot_theme_assignments; | |
| 9572 | - | |
| 9573 | - $style_settings = array( | |
| 9574 | - 'ajax_url' => admin_url('admin-ajax.php'), | |
| 9575 | - 'nonce' => wp_create_nonce('mxchat_chat_nonce'), | |
| 9576 | - 'model' => isset($this->options['model']) ? $this->options['model'] : 'gpt-5.1-chat-latest', | |
| 9577 | - 'enable_streaming_toggle' => isset($this->options['enable_streaming_toggle']) ? $this->options['enable_streaming_toggle'] : 'on', | |
| 9578 | - 'contextual_awareness_toggle' => isset($this->options['contextual_awareness_toggle']) ? $this->options['contextual_awareness_toggle'] : 'off', | |
| 9579 | - 'link_target_toggle' => $this->options['link_target_toggle'] ?? 'off', | |
| 9580 | - 'rate_limit_message' => $this->options['rate_limit_message'] ?? 'Rate limit exceeded. Please try again later.', | |
| 9581 | - 'complianz_toggle' => isset($this->options['complianz_toggle']) && $this->options['complianz_toggle'] === 'on', | |
| 9582 | - 'user_message_bg_color' => $this->options['user_message_bg_color'] ?? '#fff', | |
| 9583 | - 'user_message_font_color' => $this->options['user_message_font_color'] ?? '#212121', | |
| 9584 | - 'bot_message_bg_color' => $this->options['bot_message_bg_color'] ?? '#212121', | |
| 9585 | - 'bot_message_font_color' => $this->options['bot_message_font_color'] ?? '#fff', | |
| 9586 | - 'top_bar_bg_color' => $this->options['top_bar_bg_color'] ?? '#212121', | |
| 9587 | - 'send_button_font_color' => $this->options['send_button_font_color'] ?? '#212121', | |
| 9588 | - 'close_button_color' => $this->options['close_button_color'] ?? '#fff', | |
| 9589 | - 'chatbot_background_color' => $this->options['chatbot_background_color'] ?? '#212121', | |
| 9590 | - 'chatbot_bg_color' => $this->options['chatbot_bg_color'] ?? '#fff', | |
| 9591 | - 'icon_color' => $this->options['icon_color'] ?? '#fff', | |
| 9592 | - 'chat_input_font_color' => $this->options['chat_input_font_color'] ?? '#212121', | |
| 9593 | - 'chat_persistence_toggle' => $this->options['chat_persistence_toggle'] ?? 'off', | |
| 9594 | - 'appendWidgetToBody' => $this->options['append_to_body'] ?? 'off', | |
| 9595 | - 'live_agent_message_bg_color' => $this->options['live_agent_message_bg_color'] ?? '#ffffff', | |
| 9596 | - 'live_agent_message_font_color' => $this->options['live_agent_message_font_color'] ?? '#333333', | |
| 9597 | - 'chat_toolbar_toggle' => $this->options['chat_toolbar_toggle'] ?? 'off', | |
| 9598 | - 'mode_indicator_bg_color' => $this->options['mode_indicator_bg_color'] ?? '#767676', | |
| 9599 | - 'mode_indicator_font_color' => $this->options['mode_indicator_font_color'] ?? '#ffffff', | |
| 9600 | - 'toolbar_icon_color' => $this->options['toolbar_icon_color'] ?? '#212121', | |
| 9601 | - 'use_pinecone' => $prompts_options['mxchat_use_pinecone'] ?? '0', | |
| 9602 | - 'email_collection_enabled' => $this->options['enable_email_block'] ?? 'off', | |
| 9603 | - 'initial_email_state' => null, | |
| 9604 | - 'skip_email_check' => true, | |
| 9605 | - 'pinecone_enabled' => isset($prompts_options['mxchat_use_pinecone']) && $prompts_options['mxchat_use_pinecone'] === '1', | |
| 9606 | - 'skip_inline_colors' => $skip_inline_colors, | |
| 9607 | - 'bot_theme_assignments' => $theme_options['bot_theme_assignments'] ?? array() | |
| 9608 | - ); | |
| 9609 | - | |
| 9610 | - // Determine delay time based on strategy | |
| 9611 | - $delay_ms = 0; | |
| 9612 | - switch ($loading_strategy) { | |
| 9613 | - case 'delay_1s': | |
| 9614 | - $delay_ms = 1000; | |
| 9615 | - break; | |
| 9616 | - case 'delay_3s': | |
| 9617 | - $delay_ms = 3000; | |
| 9618 | - break; | |
| 9619 | - case 'delay_5s': | |
| 9620 | - $delay_ms = 5000; | |
| 9621 | - break; | |
| 9622 | - } | |
| 9623 | - | |
| 9624 | - ?> | |
| 9625 | - <script type="text/javascript"> | |
| 9626 | - (function() { | |
| 9627 | - var mxchatLoaded = false; | |
| 9628 | - var mxchatChat = <?php echo wp_json_encode($style_settings); ?>; | |
| 9629 | - window.mxchatChat = mxchatChat; | |
| 9630 | - | |
| 9631 | - function loadMxChatScript() { | |
| 9632 | - if (mxchatLoaded) return; | |
| 9633 | - mxchatLoaded = true; | |
| 9634 | - | |
| 9635 | - function appendChatScript() { | |
| 9636 | - var script = document.createElement('script'); | |
| 9637 | - script.src = <?php echo wp_json_encode($script_url); ?>; | |
| 9638 | - script.type = 'text/javascript'; | |
| 9639 | - document.body.appendChild(script); | |
| 9640 | - } | |
| 9641 | - | |
| 9642 | - if (typeof jQuery !== 'undefined') { | |
| 9643 | - appendChatScript(); | |
| 9644 | - } else { | |
| 9645 | - var jq = document.createElement('script'); | |
| 9646 | - jq.src = <?php echo wp_json_encode(includes_url('js/jquery/jquery.min.js')); ?>; | |
| 9647 | - jq.onload = appendChatScript; | |
| 9648 | - document.body.appendChild(jq); | |
| 9649 | - } | |
| 9650 | - } | |
| 9651 | - | |
| 9652 | - <?php if ($loading_strategy === 'on_interaction'): ?> | |
| 9653 | - // Load on user interaction | |
| 9654 | - var events = ['scroll', 'mousemove', 'touchstart', 'keydown', 'click']; | |
| 9655 | - events.forEach(function(evt) { | |
| 9656 | - window.addEventListener(evt, loadMxChatScript, {once: true, passive: true}); | |
| 9657 | - }); | |
| 9658 | - // Fallback: load after 8 seconds if no interaction | |
| 9659 | - setTimeout(loadMxChatScript, 8000); | |
| 9660 | - <?php else: ?> | |
| 9661 | - // Load after specified delay | |
| 9662 | - setTimeout(loadMxChatScript, <?php echo intval($delay_ms); ?>); | |
| 9663 | - <?php endif; ?> | |
| 9664 | - })(); | |
| 9665 | - </script> | |
| 9666 | - <?php | |
| 9667 | -} | |
| 9668 | - | |
| 9669 | 7360 | /** |
| 9670 | 7361 | * Setup the cron jobs for rate limits with guard against multiple calls |
| 9671 | 7362 | */ |
| 9672 | 7363 | public function setup_rate_limit_cron_jobs() { |
| @@ -10223,9 +7914,9 @@ | ||
| 10223 | 7914 | ? $this->options['system_prompt_instructions'] |
| 10224 | 7915 | : 'No system prompt configured'; |
| 10225 | 7916 | |
| 10226 | 7917 | // Get selected model |
| 10227 | - $selected_model = isset($this->options['model']) ? $this->options['model'] : 'gpt-5.1-chat-latest'; | |
| 7918 | + $selected_model = isset($this->options['model']) ? $this->options['model'] : 'gpt-4o'; | |
| 10228 | 7919 | |
| 10229 | 7920 | // Check if OpenRouter is being used |
| 10230 | 7921 | $is_openrouter = ($selected_model === 'openrouter'); |
| 10231 | 7922 | $openrouter_model = ''; |
| @@ -10293,42 +7984,24 @@ | ||
| 10293 | 7984 | if (!wp_verify_nonce($_POST['nonce'], 'mxchat_test_nonce')) { |
| 10294 | 7985 | wp_send_json_error(['message' => 'Invalid nonce']); |
| 10295 | 7986 | return; |
| 10296 | 7987 | } |
| 10297 | - | |
| 7988 | + | |
| 10298 | 7989 | // Only allow admin users |
| 10299 | 7990 | if (!current_user_can('administrator')) { |
| 10300 | 7991 | wp_send_json_error(['message' => 'Unauthorized']); |
| 10301 | 7992 | return; |
| 10302 | 7993 | } |
| 10303 | - | |
| 10304 | - // Check OpenAI Vector Store first (takes priority) | |
| 10305 | - $vectorstore_options = get_option('mxchat_openai_vectorstore_options', array()); | |
| 10306 | - $use_vectorstore = (isset($vectorstore_options['mxchat_use_openai_vectorstore']) && $vectorstore_options['mxchat_use_openai_vectorstore'] === '1'); | |
| 10307 | - | |
| 10308 | - if ($use_vectorstore) { | |
| 10309 | - $vectorstore_ids = $vectorstore_options['mxchat_openai_vectorstore_ids'] ?? ''; | |
| 10310 | - $id_count = !empty($vectorstore_ids) ? count(array_filter(array_map('trim', explode(',', $vectorstore_ids)))) : 0; | |
| 10311 | - | |
| 10312 | - $kb_info = [ | |
| 10313 | - 'type' => 'OpenAI Vector Store', | |
| 10314 | - 'status' => 'Active', | |
| 10315 | - 'documents' => $id_count > 0 ? $id_count . ' vector store' . ($id_count > 1 ? 's' : '') . ' configured' : 'No vector stores configured' | |
| 10316 | - ]; | |
| 10317 | - | |
| 10318 | - wp_send_json_success($kb_info); | |
| 10319 | - return; | |
| 10320 | - } | |
| 10321 | - | |
| 7994 | + | |
| 10322 | 7995 | // Check Pinecone vs WordPress |
| 10323 | 7996 | $addon_options = get_option('mxchat_pinecone_addon_options', array()); |
| 10324 | 7997 | $use_pinecone = (isset($addon_options['mxchat_use_pinecone']) && $addon_options['mxchat_use_pinecone'] === '1'); |
| 10325 | - | |
| 7998 | + | |
| 10326 | 7999 | $kb_info = [ |
| 10327 | 8000 | 'type' => $use_pinecone ? 'Pinecone' : 'WordPress Database', |
| 10328 | 8001 | 'status' => 'Active' |
| 10329 | 8002 | ]; |
| 10330 | - | |
| 8003 | + | |
| 10331 | 8004 | // Get document count |
| 10332 | 8005 | if ($use_pinecone) { |
| 10333 | 8006 | $kb_info['documents'] = 'Connected to Pinecone'; |
| 10334 | 8007 | $kb_info['api_configured'] = !empty($addon_options['mxchat_pinecone_api_key']); |
| @@ -10338,9 +8011,9 @@ | ||
| 10338 | 8011 | $table_name = $wpdb->prefix . 'mxchat_system_prompt_content'; |
| 10339 | 8012 | $count = $wpdb->get_var("SELECT COUNT(*) FROM {$table_name}"); |
| 10340 | 8013 | $kb_info['documents'] = $count ? $count . ' documents' : 'No documents'; |
| 10341 | 8014 | } |
| 10342 | - | |
| 8015 | + | |
| 10343 | 8016 | wp_send_json_success($kb_info); |
| 10344 | 8017 | } |
| 10345 | 8018 | |
| 10346 | 8019 | /** |
| @@ -10422,13 +8095,9 @@ | ||
| 10422 | 8095 | // Clear any other session-specific transients |
| 10423 | 8096 | delete_transient("mxchat_waiting_for_pdf_url_{$session_id}"); |
| 10424 | 8097 | delete_transient("mxchat_include_pdf_in_context_{$session_id}"); |
| 10425 | 8098 | delete_transient("mxchat_include_word_in_context_{$session_id}"); |
| 10426 | - | |
| 10427 | - // Clear form addon state (pending forms and submitted forms) | |
| 10428 | - delete_option("mxchat_pending_form_{$session_id}"); | |
| 10429 | - delete_option("mxchat_submitted_forms_{$session_id}"); | |
| 10430 | - | |
| 8099 | + | |
| 10431 | 8100 | //error_log("MxChat: Cleared all data for session: {$session_id}"); |
| 10432 | 8101 | } |
| 10433 | 8102 | |
| 10434 | 8103 | /** |
| @@ -10613,17 +8282,17 @@ | ||
| 10613 | 8282 | * @return string Cleaned response with invalid URLs removed/flagged |
| 10614 | 8283 | */ |
| 10615 | 8284 | private function validate_and_clean_urls($response_text, $valid_urls) { |
| 10616 | 8285 | // DEBUG: Log what we're working with |
| 10617 | - //error_log("=== MxChat URL Validation Debug ==="); | |
| 10618 | - //error_log("Valid URLs count: " . count($valid_urls)); | |
| 10619 | - //error_log("Valid URLs: " . print_r($valid_urls, true)); | |
| 10620 | - //error_log("Response text length: " . strlen($response_text)); | |
| 10621 | - //error_log("Response text preview: " . substr($response_text, 0, 500)); | |
| 8286 | + error_log("=== MxChat URL Validation Debug ==="); | |
| 8287 | + error_log("Valid URLs count: " . count($valid_urls)); | |
| 8288 | + error_log("Valid URLs: " . print_r($valid_urls, true)); | |
| 8289 | + error_log("Response text length: " . strlen($response_text)); | |
| 8290 | + error_log("Response text preview: " . substr($response_text, 0, 500)); | |
| 10622 | 8291 | |
| 10623 | 8292 | // If no valid URLs provided or empty response, return as-is |
| 10624 | 8293 | if (empty($valid_urls) || empty($response_text)) { |
| 10625 | - //error_log("Validation skipped - empty valid_urls or response"); | |
| 8294 | + error_log("Validation skipped - empty valid_urls or response"); | |
| 10626 | 8295 | return $response_text; |
| 10627 | 8296 | } |
| 10628 | 8297 | |
| 10629 | 8298 | // Extract all URLs from the AI response |
| @@ -10635,9 +8304,9 @@ | ||
| 10635 | 8304 | ); |
| 10636 | 8305 | |
| 10637 | 8306 | // If no URLs found in response, return as-is |
| 10638 | 8307 | if (empty($matches[0])) { |
| 10639 | - //error_log("No URLs found in response"); | |
| 8308 | + error_log("No URLs found in response"); | |
| 10640 | 8309 | return $response_text; |
| 10641 | 8310 | } |
| 10642 | 8311 | |
| 10643 | 8312 | $found_urls = $matches[0]; |
| @@ -10654,9 +8323,9 @@ | ||
| 10654 | 8323 | $url = rtrim($url, '.,;:!?'); |
| 10655 | 8324 | return $url; |
| 10656 | 8325 | }, $valid_urls); |
| 10657 | 8326 | |
| 10658 | - //error_log("Normalized valid URLs: " . print_r($normalized_valid_urls, true)); | |
| 8327 | + error_log("Normalized valid URLs: " . print_r($normalized_valid_urls, true)); | |
| 10659 | 8328 | |
| 10660 | 8329 | foreach ($found_urls as $found_url) { |
| 10661 | 8330 | // Clean up the found URL (remove trailing punctuation that might have been captured) |
| 10662 | 8331 | $clean_found_url = rtrim($found_url, '.,;:!?)'); |
| @@ -10661,30 +8330,30 @@ | ||
| 10661 | 8330 | // Clean up the found URL (remove trailing punctuation that might have been captured) |
| 10662 | 8331 | $clean_found_url = rtrim($found_url, '.,;:!?)'); |
| 10663 | 8332 | |
| 10664 | 8333 | // DEBUG: Log each URL being checked |
| 10665 | - //error_log("Checking found URL: " . $found_url); | |
| 8334 | + error_log("Checking found URL: " . $found_url); | |
| 10666 | 8335 | |
| 10667 | 8336 | // Normalize for comparison |
| 10668 | 8337 | $normalized_found = rtrim($clean_found_url, '/'); |
| 10669 | 8338 | $normalized_found = preg_replace('/#.*$/', '', $normalized_found); |
| 10670 | 8339 | |
| 10671 | - //error_log("Normalized found URL: " . $normalized_found); | |
| 8340 | + error_log("Normalized found URL: " . $normalized_found); | |
| 10672 | 8341 | |
| 10673 | 8342 | // Check if this URL exists in our valid URLs list |
| 10674 | 8343 | $is_valid = false; |
| 10675 | 8344 | |
| 10676 | - //error_log("Starting validation checks for: " . $normalized_found); | |
| 8345 | + error_log("Starting validation checks for: " . $normalized_found); | |
| 10677 | 8346 | |
| 10678 | 8347 | // First, try exact match |
| 10679 | 8348 | if (in_array($normalized_found, $normalized_valid_urls)) { |
| 10680 | 8349 | $is_valid = true; |
| 10681 | - //error_log("EXACT MATCH FOUND"); | |
| 8350 | + error_log("EXACT MATCH FOUND"); | |
| 10682 | 8351 | } else { |
| 10683 | - //error_log("No exact match, checking variations..."); | |
| 8352 | + error_log("No exact match, checking variations..."); | |
| 10684 | 8353 | // If no exact match, check if it's a variation (with query params, etc.) |
| 10685 | 8354 | foreach ($normalized_valid_urls as $valid_url) { |
| 10686 | - //error_log(" Comparing against valid URL: " . $valid_url); | |
| 8355 | + error_log(" Comparing against valid URL: " . $valid_url); | |
| 10687 | 8356 | |
| 10688 | 8357 | // Check if the found URL starts with a valid URL (handles query params) |
| 10689 | 8358 | if (strpos($normalized_found, $valid_url) === 0) { |
| 10690 | 8359 | // Check what comes after the valid URL |
| @@ -10695,24 +8364,24 @@ | ||
| 10695 | 8364 | // 2. Query params (starts with ?) |
| 10696 | 8365 | // 3. Fragment (starts with #) |
| 10697 | 8366 | if (empty($remainder) || $remainder[0] === '?' || $remainder[0] === '#') { |
| 10698 | 8367 | $is_valid = true; |
| 10699 | - //error_log(" MATCH: Found URL is valid variation of base URL"); | |
| 8368 | + error_log(" MATCH: Found URL is valid variation of base URL"); | |
| 10700 | 8369 | break; |
| 10701 | 8370 | } else { |
| 10702 | - //error_log(" NOT A MATCH: Found URL extends path beyond valid URL (remainder: " . $remainder . ")"); | |
| 8371 | + error_log(" NOT A MATCH: Found URL extends path beyond valid URL (remainder: " . $remainder . ")"); | |
| 10703 | 8372 | } |
| 10704 | 8373 | } |
| 10705 | 8374 | // Also check the reverse (in case valid URL has query params) |
| 10706 | 8375 | if (strpos($valid_url, $normalized_found) === 0) { |
| 10707 | 8376 | $is_valid = true; |
| 10708 | - //error_log(" MATCH: Valid URL starts with found URL"); | |
| 8377 | + error_log(" MATCH: Valid URL starts with found URL"); | |
| 10709 | 8378 | break; |
| 10710 | 8379 | } |
| 10711 | 8380 | } |
| 10712 | 8381 | |
| 10713 | 8382 | if (!$is_valid) { |
| 10714 | - //error_log("NO MATCH FOUND - URL should be removed"); | |
| 8383 | + error_log("NO MATCH FOUND - URL should be removed"); | |
| 10715 | 8384 | } |
| 10716 | 8385 | } |
| 10717 | 8386 | |
| 10718 | 8387 | // If URL is not valid, remove it from the response |
| @@ -10717,10 +8386,10 @@ | ||
| 10717 | 8386 | |
| 10718 | 8387 | // If URL is not valid, remove it from the response |
| 10719 | 8388 | if (!$is_valid) { |
| 10720 | 8389 | // Log the removal for debugging |
| 10721 | - //error_log("MxChat: Removed hallucinated URL: " . $found_url); | |
| 10722 | - //error_log("MxChat: Valid URLs were: " . implode(', ', array_slice($normalized_valid_urls, 0, 5))); | |
| 8390 | + error_log("MxChat: Removed hallucinated URL: " . $found_url); | |
| 8391 | + error_log("MxChat: Valid URLs were: " . implode(', ', array_slice($normalized_valid_urls, 0, 5))); | |
| 10723 | 8392 | |
| 10724 | 8393 | $removed_count++; |
| 10725 | 8394 | |
| 10726 | 8395 | // Check if URL is part of a markdown link: [text](url) |
| @@ -10725,15 +8394,15 @@ | ||
| 10725 | 8394 | |
| 10726 | 8395 | // Check if URL is part of a markdown link: [text](url) |
| 10727 | 8396 | $markdown_pattern = '/\[([^\]]+)\]\(' . preg_quote($found_url, '/') . '\)/'; |
| 10728 | 8397 | if (preg_match($markdown_pattern, $cleaned_response)) { |
| 10729 | - //error_log("Found markdown link, removing but keeping text"); | |
| 8398 | + error_log("Found markdown link, removing but keeping text"); | |
| 10730 | 8399 | // Remove the markdown link but keep the text |
| 10731 | 8400 | $cleaned_response = preg_replace($markdown_pattern, '$1', $cleaned_response); |
| 10732 | 8401 | } |
| 10733 | 8402 | // Check if URL is part of an HTML link: <a href="url">text</a> |
| 10734 | 8403 | else if (preg_match('/<a[^>]*href=["\']' . preg_quote($found_url, '/') . '["\'][^>]*>(.*?)<\/a>/i', $cleaned_response, $link_match)) { |
| 10735 | - //error_log("Found HTML link, removing but keeping text"); | |
| 8404 | + error_log("Found HTML link, removing but keeping text"); | |
| 10736 | 8405 | // Remove the HTML link but keep the text |
| 10737 | 8406 | $link_text = $link_match[1]; |
| 10738 | 8407 | $cleaned_response = preg_replace( |
| 10739 | 8408 | '/<a[^>]*href=["\']' . preg_quote($found_url, '/') . '["\'][^>]*>.*?<\/a>/i', |
| @@ -10742,9 +8411,9 @@ | ||
| 10742 | 8411 | ); |
| 10743 | 8412 | } |
| 10744 | 8413 | // Otherwise just remove the bare URL |
| 10745 | 8414 | else { |
| 10746 | - //error_log("Removing bare URL"); | |
| 8415 | + error_log("Removing bare URL"); | |
| 10747 | 8416 | $cleaned_response = str_replace($found_url, '', $cleaned_response); |
| 10748 | 8417 | } |
| 10749 | 8418 | } |
| 10750 | 8419 | } |
| @@ -10750,19 +8419,18 @@ | ||
| 10750 | 8419 | } |
| 10751 | 8420 | |
| 10752 | 8421 | // Log summary if any URLs were removed |
| 10753 | 8422 | if ($removed_count > 0) { |
| 10754 | - //error_log("MxChat: URL Validation Summary - Removed {$removed_count} hallucinated URL(s)"); | |
| 8423 | + error_log("MxChat: URL Validation Summary - Removed {$removed_count} hallucinated URL(s)"); | |
| 10755 | 8424 | } else { |
| 10756 | - //error_log("MxChat: URL Validation Summary - No URLs removed, all were valid"); | |
| 8425 | + error_log("MxChat: URL Validation Summary - No URLs removed, all were valid"); | |
| 10757 | 8426 | } |
| 10758 | 8427 | |
| 10759 | 8428 | // Clean up any double spaces or awkward punctuation left behind |
| 10760 | - // IMPORTANT: Only collapse horizontal whitespace (spaces/tabs), preserve newlines for markdown formatting | |
| 10761 | - $cleaned_response = preg_replace('/[^\S\n]+/', ' ', $cleaned_response); // Collapse spaces/tabs but NOT newlines | |
| 10762 | - $cleaned_response = preg_replace('/[^\S\n]+([.,;:!?])/', '$1', $cleaned_response); // Same for punctuation cleanup | |
| 8429 | + $cleaned_response = preg_replace('/\s+/', ' ', $cleaned_response); | |
| 8430 | + $cleaned_response = preg_replace('/\s+([.,;:!?])/', '$1', $cleaned_response); | |
| 10763 | 8431 | |
| 10764 | - //error_log("Final cleaned response: " . $cleaned_response); | |
| 8432 | + error_log("Final cleaned response: " . $cleaned_response); | |
| 10765 | 8433 | |
| 10766 | 8434 | return trim($cleaned_response); |
| 10767 | 8435 | } |
| 10768 | 8436 | |
