| @@ -10,40 +10,10 @@ | ||
| 10 | 10 | private $fallbackResponse; |
| 11 | 11 | private $productCardHtml; |
| 12 | 12 | private $word_handler; |
| 13 | 13 | private $last_similarity_analysis = null; |
| 14 | - private $current_valid_urls = []; | |
| 15 | - private $last_vectorstore_error = null; | |
| 16 | - private $is_streaming = false; // ADDED: Track if current request is streaming | |
| 17 | - private $streaming_headers_sent = false; // Track if streaming headers have been sent | |
| 18 | 14 | |
| 19 | -/** | |
| 20 | - * Setup streaming headers - call this right before actually streaming | |
| 21 | - * This delays header setup to allow actions/forms to return JSON responses | |
| 22 | - */ | |
| 23 | -private function setup_streaming_headers() { | |
| 24 | - if ($this->streaming_headers_sent || headers_sent()) { | |
| 25 | - return false; | |
| 26 | - } | |
| 27 | 15 | |
| 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 | - $this->streaming_headers_sent = true; | |
| 43 | - return true; | |
| 44 | -} | |
| 45 | - | |
| 46 | 16 | /** |
| 47 | 17 | * Class constructor |
| 48 | 18 | */ |
| 49 | 19 | public function __construct() { |
| @@ -111,28 +81,13 @@ | ||
| 111 | 81 | // Add chat mode checking actions |
| 112 | 82 | add_action('wp_ajax_mxchat_get_current_chat_mode', array($this, 'mxchat_get_current_chat_mode')); |
| 113 | 83 | add_action('wp_ajax_nopriv_mxchat_get_current_chat_mode', array($this, 'mxchat_get_current_chat_mode')); |
| 114 | 84 | |
| 115 | - // Nonce refresh for page-cache compatibility (WP Rocket, LiteSpeed, etc.) | |
| 116 | - add_action('wp_ajax_mxchat_refresh_nonce', array($this, 'mxchat_refresh_nonce')); | |
| 117 | - add_action('wp_ajax_nopriv_mxchat_refresh_nonce', array($this, 'mxchat_refresh_nonce')); | |
| 118 | - | |
| 119 | - // Auto-email transcript action | |
| 120 | - add_action('mxchat_send_delayed_transcript', array($this, 'mxchat_send_delayed_transcript'), 10, 1); | |
| 121 | - | |
| 122 | 85 | add_filter('mxchat_check_actions_only', array($this, 'check_actions_for_addons'), 10, 4); |
| 123 | 86 | |
| 124 | 87 | |
| 125 | 88 | } |
| 126 | 89 | |
| 127 | -/** | |
| 128 | - * Return a fresh nonce so cached pages can replace the stale one. | |
| 129 | - */ | |
| 130 | -public function mxchat_refresh_nonce() { | |
| 131 | - nocache_headers(); | |
| 132 | - wp_send_json_success(array('nonce' => wp_create_nonce('mxchat_chat_nonce'))); | |
| 133 | -} | |
| 134 | - | |
| 135 | 90 | // In your core plugin's check_actions_for_addons method: |
| 136 | 91 | public function check_actions_for_addons($default, $message, $user_id, $session_id) { |
| 137 | 92 | //error_log('MxChat Core: check_actions_for_addons called with message: ' . $message); |
| 138 | 93 | |
| @@ -155,22 +110,8 @@ | ||
| 155 | 110 | wp_die(); |
| 156 | 111 | } |
| 157 | 112 | |
| 158 | 113 | $session_id = sanitize_text_field($_POST['session_id']); |
| 159 | - | |
| 160 | - // SECURITY FIX: Verify session ownership before retrieving data | |
| 161 | - // If IP/user changed, signal frontend to reset session instead of blocking | |
| 162 | - $current_user_identifier = MxChat_User::mxchat_get_user_identifier(); | |
| 163 | - | |
| 164 | - // Check if this session has an owner recorded | |
| 165 | - $session_owner = get_option("mxchat_session_owner_{$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 !== $current_user_identifier) { | |
| 170 | - update_option("mxchat_session_owner_{$session_id}", $current_user_identifier, 'no'); | |
| 171 | - } | |
| 172 | - | |
| 173 | 114 | $history = get_option("mxchat_history_{$session_id}", []); // Retrieve stored history |
| 174 | 115 | $chat_mode = get_option("mxchat_mode_{$session_id}", 'ai'); // Get current chat mode |
| 175 | 116 | |
| 176 | 117 | if (empty($history)) { |
| @@ -187,25 +128,11 @@ | ||
| 187 | 128 | 'chat_mode' => $chat_mode |
| 188 | 129 | ]); |
| 189 | 130 | wp_die(); |
| 190 | 131 | } |
| 191 | -private function mxchat_fetch_conversation_history_for_ai($session_id, $session_start_timestamp = 0) { | |
| 132 | + | |
| 133 | +private function mxchat_fetch_conversation_history_for_ai($session_id) { | |
| 192 | 134 | $history = get_option("mxchat_history_{$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 | 135 | $formatted_history = []; |
| 209 | 136 | |
| 210 | 137 | // Adjusted for code-heavy conversations |
| 211 | 138 | $max_tokens = 120000; // Context window size |
| @@ -303,15 +230,8 @@ | ||
| 303 | 230 | 'callback' => [$this, 'handle_slack_messages'], |
| 304 | 231 | 'permission_callback' => [$this, 'verify_slack_request'], |
| 305 | 232 | ]); |
| 306 | 233 | |
| 307 | - // Telegram webhook endpoint | |
| 308 | - register_rest_route('mxchat/v1', '/telegram-webhook', [ | |
| 309 | - 'methods' => 'POST', | |
| 310 | - 'callback' => [$this, 'handle_telegram_webhook'], | |
| 311 | - 'permission_callback' => [$this, 'verify_telegram_request'], | |
| 312 | - ]); | |
| 313 | - | |
| 314 | 234 | //error_log(esc_html__('MxChat REST routes registered', 'mxchat')); |
| 315 | 235 | } |
| 316 | 236 | |
| 317 | 237 | /** |
| @@ -351,11 +271,10 @@ | ||
| 351 | 271 | //error_log(esc_html__('Slack request timestamp too old', 'mxchat')); |
| 352 | 272 | return false; |
| 353 | 273 | } |
| 354 | 274 | |
| 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(); | |
| 275 | + // Get raw request body | |
| 276 | + $request_body = file_get_contents('php://input'); | |
| 358 | 277 | |
| 359 | 278 | // Create the signature base string |
| 360 | 279 | $sig_basestring = "v0:{$timestamp}:{$request_body}"; |
| 361 | 280 | |
| @@ -364,43 +283,8 @@ | ||
| 364 | 283 | |
| 365 | 284 | // Compare signatures |
| 366 | 285 | return hash_equals($my_signature, $slack_signature); |
| 367 | 286 | } |
| 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 | 287 | public function mxchat_stream_events(WP_REST_Request $request) { |
| 404 | 288 | header('Content-Type: text/event-stream'); |
| 405 | 289 | header('Cache-Control: no-cache'); |
| 406 | 290 | header('Connection: keep-alive'); |
| @@ -434,9 +318,9 @@ | ||
| 434 | 318 | |
| 435 | 319 | |
| 436 | 320 | |
| 437 | 321 | |
| 438 | -private function mxchat_save_chat_message($session_id, $role, $message, $originating_page = null, $rag_context = null) { | |
| 322 | +private function mxchat_save_chat_message($session_id, $role, $message, $originating_page = null) { | |
| 439 | 323 | global $wpdb; |
| 440 | 324 | $table_name = $wpdb->prefix . 'mxchat_chat_transcripts'; |
| 441 | 325 | //error_log("[DEBUG] mxchat_save_chat_message -> START for session_id: {$session_id}, role: {$role}"); |
| 442 | 326 | |
| @@ -454,20 +338,8 @@ | ||
| 454 | 338 | //error_log("[DEBUG] This is a NEW session - first message"); |
| 455 | 339 | } |
| 456 | 340 | } |
| 457 | 341 | |
| 458 | - // SECURITY FIX: Set session ownership for new sessions | |
| 459 | - if ($is_new_session && $role === 'user') { | |
| 460 | - $current_user_identifier = MxChat_User::mxchat_get_user_identifier(); | |
| 461 | - $session_owner_key = "mxchat_session_owner_{$session_id}"; | |
| 462 | - | |
| 463 | - // Only set ownership if not already set | |
| 464 | - if (!get_option($session_owner_key)) { | |
| 465 | - update_option($session_owner_key, $current_user_identifier, 'no'); | |
| 466 | - //error_log("[DEBUG] Set session ownership for {$session_id} to {$current_user_identifier}"); | |
| 467 | - } | |
| 468 | - } | |
| 469 | - | |
| 470 | 342 | // 1) Extract agent name if present |
| 471 | 343 | $agent_name = ''; |
| 472 | 344 | if (preg_match('/^Agent: (.*?) - /', $message, $matches)) { |
| 473 | 345 | $agent_name = $matches[1]; |
| @@ -610,17 +482,9 @@ | ||
| 610 | 482 | $insert_data['originating_page_title'] = $stored_originating['title'] ?? ''; |
| 611 | 483 | } |
| 612 | 484 | } |
| 613 | 485 | } |
| 614 | - | |
| 615 | - // Add RAG context if provided (for bot messages) | |
| 616 | - if ($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 | - | |
| 486 | + | |
| 623 | 487 | $wpdb->insert($table_name, $insert_data); |
| 624 | 488 | //error_log("[DEBUG] mxchat_save_chat_message -> Inserted message into DB. row_id: {$wpdb->insert_id}, data: " . print_r($insert_data, true)); |
| 625 | 489 | |
| 626 | 490 | // 9) Send notification email if this is the first user message in a new session |
| @@ -631,17 +495,11 @@ | ||
| 631 | 495 | 'ip' => $_SERVER['REMOTE_ADDR'] |
| 632 | 496 | )); |
| 633 | 497 | } |
| 634 | 498 | |
| 635 | - // 10) Schedule delayed transcript email if enabled and message is from user | |
| 636 | - if ($wpdb->insert_id && $role === 'user') { | |
| 637 | - $this->schedule_delayed_transcript_email($session_id); | |
| 638 | - } | |
| 639 | - | |
| 640 | 499 | //error_log("[DEBUG] mxchat_save_chat_message -> END for session_id: {$session_id}"); |
| 641 | 500 | return $message_id; |
| 642 | 501 | } |
| 643 | - | |
| 644 | 502 | private function send_new_chat_notification($session_id, $user_info = array()) { |
| 645 | 503 | $options = get_option('mxchat_transcripts_options'); |
| 646 | 504 | |
| 647 | 505 | // Check if notifications are enabled |
| @@ -684,200 +542,12 @@ | ||
| 684 | 542 | // Send email |
| 685 | 543 | return wp_mail($to, $subject, $message); |
| 686 | 544 | } |
| 687 | 545 | |
| 688 | -/** | |
| 689 | - * Schedule delayed transcript email for a session | |
| 690 | - * Reschedules if a new user message is received | |
| 691 | - */ | |
| 692 | -private function schedule_delayed_transcript_email($session_id) { | |
| 693 | - $options = get_option('mxchat_transcripts_options'); | |
| 694 | - | |
| 695 | - // Check if auto-email is enabled | |
| 696 | - if (empty($options['mxchat_auto_email_transcript_enabled'])) { | |
| 697 | - return; | |
| 698 | - } | |
| 699 | - | |
| 700 | - // Get notification email | |
| 701 | - $email = !empty($options['mxchat_notification_email']) ? | |
| 702 | - $options['mxchat_notification_email'] : | |
| 703 | - get_option('admin_email'); | |
| 704 | - | |
| 705 | - if (!is_email($email)) { | |
| 706 | - return; | |
| 707 | - } | |
| 708 | - | |
| 709 | - // Get delay in minutes (default 30) | |
| 710 | - $delay_minutes = isset($options['mxchat_auto_email_transcript_delay']) ? | |
| 711 | - intval($options['mxchat_auto_email_transcript_delay']) : 30; | |
| 712 | - | |
| 713 | - // Clear any existing scheduled event for this session | |
| 714 | - $hook = 'mxchat_send_delayed_transcript'; | |
| 715 | - $args = array($session_id); | |
| 716 | - $timestamp = wp_next_scheduled($hook, $args); | |
| 717 | - | |
| 718 | - if ($timestamp) { | |
| 719 | - wp_unschedule_event($timestamp, $hook, $args); | |
| 720 | - } | |
| 721 | - | |
| 722 | - // Schedule new event | |
| 723 | - $schedule_time = time() + ($delay_minutes * 60); | |
| 724 | - wp_schedule_single_event($schedule_time, $hook, $args); | |
| 725 | -} | |
| 726 | - | |
| 727 | -/** | |
| 728 | - * Check if chat messages contain contact information (email or phone number) | |
| 729 | - * | |
| 730 | - * @param array $messages Array of message objects with 'message' property | |
| 731 | - * @param object|null $session_data Session data object with user_email property | |
| 732 | - * @return bool True if contact info found, false otherwise | |
| 733 | - */ | |
| 734 | -private function chat_contains_contact_info($messages, $session_data = null) { | |
| 735 | - // Check if session already has a stored email | |
| 736 | - if ($session_data && !empty($session_data->user_email)) { | |
| 737 | - return true; | |
| 738 | - } | |
| 739 | - | |
| 740 | - // Email regex pattern | |
| 741 | - $email_pattern = '/\b[A-Za-z0-9._%+-]+@[A-Za-z0-9.-]+\.[A-Z|a-z]{2,}\b/'; | |
| 742 | - | |
| 743 | - // Phone number patterns (covers various formats including international, WhatsApp style) | |
| 744 | - // Matches: +1234567890, (123) 456-7890, 123-456-7890, 123.456.7890, 1234567890, +1 234 567 8900, etc. | |
| 745 | - $phone_pattern = '/(?:\+?\d{1,3}[-.\s]?)?\(?\d{2,4}\)?[-.\s]?\d{2,4}[-.\s]?\d{2,4}(?:[-.\s]?\d{1,4})?/'; | |
| 746 | - | |
| 747 | - // Only check user messages (not assistant responses) | |
| 748 | - foreach ($messages as $msg) { | |
| 749 | - if ($msg->role !== 'user') { | |
| 750 | - continue; | |
| 751 | - } | |
| 752 | - | |
| 753 | - $message_text = $msg->message; | |
| 754 | - | |
| 755 | - // Check for email | |
| 756 | - if (preg_match($email_pattern, $message_text)) { | |
| 757 | - return true; | |
| 758 | - } | |
| 759 | - | |
| 760 | - // Check for phone number (must be at least 7 digits total to avoid false positives) | |
| 761 | - if (preg_match($phone_pattern, $message_text, $matches)) { | |
| 762 | - // Count actual digits to avoid matching short numbers | |
| 763 | - $digits_only = preg_replace('/\D/', '', $matches[0]); | |
| 764 | - if (strlen($digits_only) >= 7) { | |
| 765 | - return true; | |
| 766 | - } | |
| 767 | - } | |
| 768 | - } | |
| 769 | - | |
| 770 | - return false; | |
| 771 | -} | |
| 772 | - | |
| 773 | -/** | |
| 774 | - * Send the delayed transcript email with .txt attachment | |
| 775 | - */ | |
| 776 | -public function mxchat_send_delayed_transcript($session_id) { | |
| 777 | - global $wpdb; | |
| 778 | - | |
| 779 | - $options = get_option('mxchat_transcripts_options'); | |
| 780 | - | |
| 781 | - // Get notification email | |
| 782 | - $to = !empty($options['mxchat_notification_email']) ? | |
| 783 | - $options['mxchat_notification_email'] : | |
| 784 | - get_option('admin_email'); | |
| 785 | - | |
| 786 | - if (!is_email($to)) { | |
| 787 | - return false; | |
| 788 | - } | |
| 789 | - | |
| 790 | - // Get all messages for this session | |
| 791 | - $table_name = $wpdb->prefix . 'mxchat_chat_transcripts'; | |
| 792 | - $messages = $wpdb->get_results($wpdb->prepare( | |
| 793 | - "SELECT role, message, timestamp FROM {$table_name} | |
| 794 | - WHERE session_id = %s | |
| 795 | - ORDER BY timestamp ASC", | |
| 796 | - $session_id | |
| 797 | - )); | |
| 798 | - | |
| 799 | - if (empty($messages)) { | |
| 800 | - return false; | |
| 801 | - } | |
| 802 | - | |
| 803 | - // Get session metadata | |
| 804 | - $sessions_table = $wpdb->prefix . 'mxchat_sessions'; | |
| 805 | - $session_data = $wpdb->get_row($wpdb->prepare( | |
| 806 | - "SELECT * FROM {$sessions_table} WHERE session_id = %s", | |
| 807 | - $session_id | |
| 808 | - )); | |
| 809 | - | |
| 810 | - // Check if contact info is required and if it's present | |
| 811 | - $require_contact = !empty($options['mxchat_auto_email_transcript_require_contact']); | |
| 812 | - if ($require_contact && !$this->chat_contains_contact_info($messages, $session_data)) { | |
| 813 | - // Contact info required but not found - skip sending | |
| 814 | - return false; | |
| 815 | - } | |
| 816 | - | |
| 817 | - // Build transcript content | |
| 818 | - $transcript_content = "Chat Transcript\n"; | |
| 819 | - $transcript_content .= "================\n\n"; | |
| 820 | - $transcript_content .= "Session ID: " . $session_id . "\n"; | |
| 821 | - | |
| 822 | - if ($session_data) { | |
| 823 | - $transcript_content .= "User: " . ($session_data->user_identifier ?: 'Guest') . "\n"; | |
| 824 | - $transcript_content .= "Email: " . ($session_data->user_email ?: 'Not provided') . "\n"; | |
| 825 | - $transcript_content .= "Started: " . $session_data->created_at . "\n"; | |
| 826 | - } | |
| 827 | - | |
| 828 | - $transcript_content .= "\n" . str_repeat("=", 50) . "\n\n"; | |
| 829 | - | |
| 830 | - // Add messages | |
| 831 | - foreach ($messages as $msg) { | |
| 832 | - $role_label = ($msg->role === 'user') ? 'User' : 'Assistant'; | |
| 833 | - $transcript_content .= "[{$msg->timestamp}] {$role_label}:\n"; | |
| 834 | - $transcript_content .= $msg->message . "\n\n"; | |
| 835 | - } | |
| 836 | - | |
| 837 | - // Create temporary file for attachment using WP_Filesystem | |
| 838 | - $upload_dir = wp_upload_dir(); | |
| 839 | - $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); | |
| 846 | - | |
| 847 | - // Prepare email | |
| 848 | - $subject = sprintf('[%s] Chat Transcript - Session %s', get_bloginfo('name'), substr($session_id, 0, 8)); | |
| 849 | - | |
| 850 | - $message = "Please find attached the full chat transcript.\n\n"; | |
| 851 | - $message .= "Session ID: {$session_id}\n"; | |
| 852 | - | |
| 853 | - if ($session_data) { | |
| 854 | - $message .= "User: " . ($session_data->user_identifier ?: 'Guest') . "\n"; | |
| 855 | - $message .= "Email: " . ($session_data->user_email ?: 'Not provided') . "\n"; | |
| 856 | - } | |
| 857 | - | |
| 858 | - $message .= "\nView online: " . admin_url('admin.php?page=mxchat-transcripts'); | |
| 859 | - | |
| 860 | - // Send email with attachment | |
| 861 | - $attachments = array($temp_file); | |
| 862 | - $result = wp_mail($to, $subject, $message, '', $attachments); | |
| 863 | - | |
| 864 | - // Clean up temporary file | |
| 865 | - if (file_exists($temp_file)) { | |
| 866 | - unlink($temp_file); | |
| 867 | - } | |
| 868 | - | |
| 869 | - return $result; | |
| 870 | -} | |
| 871 | - | |
| 872 | - | |
| 873 | - | |
| 874 | 546 | public function mxchat_handle_save_email_and_response() { |
| 875 | 547 | //error_log('[DEBUG] ---------- mxchat_handle_save_email_and_response START ----------'); |
| 876 | 548 | //error_log('DEBUG: POST data: ' . print_r($_POST, true)); |
| 877 | 549 | |
| 878 | - nocache_headers(); | |
| 879 | - | |
| 880 | 550 | // Validate nonce |
| 881 | 551 | if (!isset($_POST['nonce']) || !wp_verify_nonce($_POST['nonce'], 'mxchat_chat_nonce')) { |
| 882 | 552 | //error_log(esc_html__('[ERROR] Invalid nonce in mxchat_handle_save_email_and_response', 'mxchat')); |
| 883 | 553 | wp_send_json_error(['message' => esc_html__('Invalid nonce.', 'mxchat')]); |
| @@ -889,9 +559,9 @@ | ||
| 889 | 559 | $name = isset($_POST['name']) ? sanitize_text_field($_POST['name']) : ''; |
| 890 | 560 | |
| 891 | 561 | //error_log("[DEBUG] handle_save_email_and_response -> session_id: {$session_id}, email: {$email}, name: {$name}"); |
| 892 | 562 | |
| 893 | - if (empty($session_id) || $session_id === 'null' || empty($email)) { | |
| 563 | + if (empty($session_id) || empty($email)) { | |
| 894 | 564 | //error_log("[ERROR] Missing session_id or email: session_id={$session_id}, email={$email}"); |
| 895 | 565 | wp_send_json_error(['message' => esc_html__('Session ID or email is missing.', 'mxchat')]); |
| 896 | 566 | wp_die(); |
| 897 | 567 | } |
| @@ -908,15 +578,15 @@ | ||
| 908 | 578 | } |
| 909 | 579 | |
| 910 | 580 | // 1) Always store email in wp_options |
| 911 | 581 | $email_option_key = "mxchat_email_{$session_id}"; |
| 912 | - update_option($email_option_key, $email, 'no'); | |
| 582 | + update_option($email_option_key, $email); | |
| 913 | 583 | //error_log("[DEBUG] handle_save_email_and_response -> updated option: {$email_option_key} => {$email}"); |
| 914 | 584 | |
| 915 | 585 | // Store name in wp_options if provided |
| 916 | 586 | if (!empty($name)) { |
| 917 | 587 | $name_option_key = "mxchat_name_{$session_id}"; |
| 918 | - update_option($name_option_key, $name, 'no'); | |
| 588 | + update_option($name_option_key, $name); | |
| 919 | 589 | //error_log("[DEBUG] handle_save_email_and_response -> updated option: {$name_option_key} => {$name}"); |
| 920 | 590 | } |
| 921 | 591 | |
| 922 | 592 | // 2) (Optional) Also store in DB if a row already exists |
| @@ -960,10 +630,8 @@ | ||
| 960 | 630 | |
| 961 | 631 | public function mxchat_check_email_provided() { |
| 962 | 632 | //error_log('[DEBUG] ---------- mxchat_check_email_provided START ----------'); |
| 963 | 633 | |
| 964 | - nocache_headers(); | |
| 965 | - | |
| 966 | 634 | if (!isset($_POST['nonce']) || !wp_verify_nonce($_POST['nonce'], 'mxchat_chat_nonce')) { |
| 967 | 635 | //error_log('[ERROR] Invalid nonce in mxchat_check_email_provided'); |
| 968 | 636 | wp_send_json_error(['message' => esc_html__('Invalid nonce', 'mxchat')]); |
| 969 | 637 | } |
| @@ -968,9 +636,9 @@ | ||
| 968 | 636 | wp_send_json_error(['message' => esc_html__('Invalid nonce', 'mxchat')]); |
| 969 | 637 | } |
| 970 | 638 | |
| 971 | 639 | $session_id = isset($_POST['session_id']) ? sanitize_text_field($_POST['session_id']) : ''; |
| 972 | - if (empty($session_id) || $session_id === 'null') { | |
| 640 | + if (empty($session_id)) { | |
| 973 | 641 | //error_log('[ERROR] No session ID provided in mxchat_check_email_provided'); |
| 974 | 642 | wp_send_json_error(['message' => esc_html__('No session ID provided', 'mxchat')]); |
| 975 | 643 | } |
| 976 | 644 | |
| @@ -1027,42 +695,15 @@ | ||
| 1027 | 695 | wp_send_json_error(['message' => esc_html__('No email found', 'mxchat')]); |
| 1028 | 696 | } |
| 1029 | 697 | } |
| 1030 | 698 | |
| 1031 | -/** | |
| 1032 | - * Send error response in appropriate format based on streaming mode | |
| 1033 | - * ADDED: Helper method to consistently handle errors in both streaming and non-streaming modes | |
| 1034 | - * | |
| 1035 | - * @param string $error_message The error message to display | |
| 1036 | - * @param string $error_code Optional error code for debugging | |
| 1037 | - */ | |
| 1038 | -private function send_error_response($error_message, $error_code = 'api_error') { | |
| 1039 | - if ($this->is_streaming) { | |
| 1040 | - echo "data: " . json_encode([ | |
| 1041 | - 'error' => true, | |
| 1042 | - 'error_message' => $error_message, | |
| 1043 | - 'error_code' => $error_code, | |
| 1044 | - 'text' => $error_message, | |
| 1045 | - 'message' => $error_message | |
| 1046 | - ]) . "\n\n"; | |
| 1047 | - echo "data: [DONE]\n\n"; | |
| 1048 | - flush(); | |
| 1049 | - } else { | |
| 1050 | - wp_send_json_error([ | |
| 1051 | - 'error_message' => $error_message, | |
| 1052 | - 'error_code' => $error_code | |
| 1053 | - ]); | |
| 1054 | - } | |
| 1055 | - wp_die(); | |
| 1056 | -} | |
| 1057 | - | |
| 1058 | 699 | public function mxchat_handle_chat_request() { |
| 1059 | 700 | global $wpdb; |
| 1060 | 701 | |
| 1061 | 702 | // Debug: Log incoming bot_id |
| 1062 | 703 | $bot_id = isset($_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); | |
| 704 | + error_log("=== MXCHAT DEBUG: Starting chat request ==="); | |
| 705 | + error_log("MXCHAT DEBUG: Bot ID received: " . $bot_id); | |
| 1065 | 706 | |
| 1066 | 707 | // Get bot-specific options |
| 1067 | 708 | $bot_options = $this->get_bot_options($bot_id); |
| 1068 | 709 | $current_options = !empty($bot_options) ? $bot_options : $this->options; |
| @@ -1067,19 +708,29 @@ | ||
| 1067 | 708 | $bot_options = $this->get_bot_options($bot_id); |
| 1068 | 709 | $current_options = !empty($bot_options) ? $bot_options : $this->options; |
| 1069 | 710 | |
| 1070 | 711 | // 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 | - $is_streaming = isset($_POST['action']) && $_POST['action'] === 'mxchat_stream_chat' && | |
| 1074 | - ($force_streaming_test || (isset($current_options['enable_streaming_toggle']) && $current_options['enable_streaming_toggle'] === 'on')); | |
| 712 | + $is_streaming = isset($_POST['action']) && $_POST['action'] === 'mxchat_stream_chat' && | |
| 713 | + isset($current_options['enable_streaming_toggle']) && $current_options['enable_streaming_toggle'] === 'on'; | |
| 714 | + | |
| 715 | + // Set streaming headers if needed | |
| 716 | + if ($is_streaming) { | |
| 717 | + // Disable output buffering | |
| 718 | + while (ob_get_level()) { | |
| 719 | + ob_end_flush(); // Changed from ob_end_clean() | |
| 720 | + } | |
| 721 | + | |
| 722 | + // Set headers for SSE | |
| 723 | + header('Content-Type: text/event-stream'); | |
| 724 | + header('Cache-Control: no-cache'); | |
| 725 | + header('Connection: keep-alive'); | |
| 726 | + header('X-Accel-Buffering: no'); | |
| 727 | + | |
| 728 | + // Add these new lines: | |
| 729 | + ob_implicit_flush(true); | |
| 730 | + flush(); | |
| 731 | + } | |
| 1075 | 732 | |
| 1076 | - // ADDED: Store streaming state in class property for use in private methods | |
| 1077 | - $this->is_streaming = $is_streaming; | |
| 1078 | - | |
| 1079 | - // NOTE: Streaming headers are now set later via setup_streaming_headers() | |
| 1080 | - // This allows actions/forms to return JSON responses without header conflicts | |
| 1081 | - | |
| 1082 | 733 | // Check if MX Chat Moderation is active |
| 1083 | 734 | if (class_exists('MX_Chat_Moderation')) { |
| 1084 | 735 | // Get user email and IP |
| 1085 | 736 | $user_email = ''; |
| @@ -1149,17 +800,8 @@ | ||
| 1149 | 800 | wp_send_json_error(esc_html__('Session ID is missing.', 'mxchat')); |
| 1150 | 801 | wp_die(); |
| 1151 | 802 | } |
| 1152 | 803 | |
| 1153 | - // Update session owner if it changed (e.g. IP changed due to network switch) | |
| 1154 | - // The session ID itself is the authentication — if the client has it, they own it | |
| 1155 | - $current_user_identifier = MxChat_User::mxchat_get_user_identifier(); | |
| 1156 | - $session_owner = get_option("mxchat_session_owner_{$session_id}"); | |
| 1157 | - | |
| 1158 | - if (!$session_owner || $session_owner !== $current_user_identifier) { | |
| 1159 | - update_option("mxchat_session_owner_{$session_id}", $current_user_identifier, 'no'); | |
| 1160 | - } | |
| 1161 | - | |
| 1162 | 804 | // Validate and sanitize the incoming message |
| 1163 | 805 | if (empty($_POST['message'])) { |
| 1164 | 806 | wp_send_json_error(esc_html__('No message received.', 'mxchat')); |
| 1165 | 807 | wp_die(); |
| @@ -1535,8 +1177,38 @@ | ||
| 1535 | 1177 | } |
| 1536 | 1178 | } |
| 1537 | 1179 | } |
| 1538 | 1180 | |
| 1181 | + // Check if there's an active recommendation flow session | |
| 1182 | + $flow_state = get_option("mxchat_sr_flow_state_{$session_id}", array()); | |
| 1183 | + if (!empty($flow_state) && isset($flow_state['flow_id'])) { | |
| 1184 | + // Create a dummy intent object that matches the original intent | |
| 1185 | + $dummy_intent = new stdClass(); | |
| 1186 | + $dummy_intent->intent_label = 'Recommendation Flow ' . $flow_state['flow_id']; | |
| 1187 | + $dummy_intent->phrases = ''; // Empty phrases to avoid matching the original trigger | |
| 1188 | + | |
| 1189 | + // Call the recommendation flow handler directly | |
| 1190 | + $response_data = apply_filters('mxchat_sr_recommendation_flow', false, $message, $user_id, $session_id, $dummy_intent); | |
| 1191 | + | |
| 1192 | + // If the handler returned a response, send it | |
| 1193 | + if (is_array($response_data) && (isset($response_data['text']) || isset($response_data['html']))) { | |
| 1194 | + // Save the bot's response to the chat history | |
| 1195 | + if (!empty($response_data['text'])) { | |
| 1196 | + $this->mxchat_save_chat_message($session_id, 'bot', $response_data['text']); | |
| 1197 | + } | |
| 1198 | + if (!empty($response_data['html'])) { | |
| 1199 | + $this->mxchat_save_chat_message($session_id, 'bot', $response_data['html']); | |
| 1200 | + } | |
| 1201 | + | |
| 1202 | + if ($testing_data !== null) { | |
| 1203 | + $response_data['testing_data'] = $testing_data; | |
| 1204 | + } | |
| 1205 | + | |
| 1206 | + // Send the response | |
| 1207 | + wp_send_json($response_data); | |
| 1208 | + wp_die(); | |
| 1209 | + } | |
| 1210 | + } | |
| 1539 | 1211 | |
| 1540 | 1212 | // Step 2: Detect intent and handle intent-based responses |
| 1541 | 1213 | $intent_result = $this->mxchat_check_intent_and_invoke_callback($message, $user_id, $session_id); |
| 1542 | 1214 | |
| @@ -1555,46 +1227,58 @@ | ||
| 1555 | 1227 | 'text' => $intent_result['text'] ?? '', |
| 1556 | 1228 | 'html' => $intent_result['html'] ?? '', |
| 1557 | 1229 | 'session_id' => $session_id |
| 1558 | 1230 | ]; |
| 1559 | - | |
| 1560 | - // IMPORTANT: Include chat_mode if present (for WhatsApp, Slack, etc.) | |
| 1561 | - if (isset($intent_result['chat_mode'])) { | |
| 1562 | - $response_data['chat_mode'] = $intent_result['chat_mode']; | |
| 1563 | - } | |
| 1564 | - | |
| 1231 | + | |
| 1565 | 1232 | if ($testing_data !== null) { |
| 1566 | 1233 | $response_data['testing_data'] = $testing_data; |
| 1567 | 1234 | } |
| 1568 | - | |
| 1235 | + | |
| 1236 | + // Clear streaming headers if they were set | |
| 1237 | + if ($is_streaming) { | |
| 1238 | + header_remove('Content-Type'); | |
| 1239 | + header_remove('Cache-Control'); | |
| 1240 | + header_remove('Connection'); | |
| 1241 | + header_remove('X-Accel-Buffering'); | |
| 1242 | + header('Content-Type: application/json'); | |
| 1243 | + } | |
| 1244 | + | |
| 1569 | 1245 | wp_send_json($response_data); |
| 1570 | 1246 | wp_die(); |
| 1571 | 1247 | } else if ($intent_result === true && (!empty($this->fallbackResponse['text']) || !empty($this->fallbackResponse['html']))) { |
| 1572 | 1248 | // Intent returned true and set fallbackResponse |
| 1573 | - | |
| 1574 | - // SAVE TO TRANSCRIPT | |
| 1249 | + | |
| 1250 | + // SAVE TO TRANSCRIPT FIRST | |
| 1575 | 1251 | if (!empty($this->fallbackResponse['text'])) { |
| 1576 | 1252 | $this->mxchat_save_chat_message($session_id, 'bot', $this->fallbackResponse['text']); |
| 1577 | 1253 | } |
| 1578 | - // Save action HTML (product cards, featured products, etc.) so it renders in transcripts | |
| 1579 | 1254 | if (!empty($this->fallbackResponse['html'])) { |
| 1580 | 1255 | $this->mxchat_save_chat_message($session_id, 'bot', $this->fallbackResponse['html']); |
| 1581 | 1256 | } |
| 1582 | - | |
| 1257 | + | |
| 1583 | 1258 | $response_data = [ |
| 1584 | 1259 | 'text' => $this->fallbackResponse['text'] ?? '', |
| 1585 | 1260 | 'html' => $this->fallbackResponse['html'] ?? '', |
| 1586 | 1261 | 'session_id' => $session_id |
| 1587 | 1262 | ]; |
| 1588 | - | |
| 1263 | + | |
| 1589 | 1264 | if (isset($this->fallbackResponse['chat_mode'])) { |
| 1590 | 1265 | $response_data['chat_mode'] = $this->fallbackResponse['chat_mode']; |
| 1591 | 1266 | } |
| 1592 | - | |
| 1267 | + | |
| 1593 | 1268 | if ($testing_data !== null) { |
| 1594 | 1269 | $response_data['testing_data'] = $testing_data; |
| 1595 | 1270 | } |
| 1596 | - | |
| 1271 | + | |
| 1272 | + // Clear streaming headers if they were set | |
| 1273 | + if ($is_streaming) { | |
| 1274 | + header_remove('Content-Type'); | |
| 1275 | + header_remove('Cache-Control'); | |
| 1276 | + header_remove('Connection'); | |
| 1277 | + header_remove('X-Accel-Buffering'); | |
| 1278 | + header('Content-Type: application/json'); | |
| 1279 | + } | |
| 1280 | + | |
| 1597 | 1281 | wp_send_json($response_data); |
| 1598 | 1282 | wp_die(); |
| 1599 | 1283 | } |
| 1600 | 1284 | } |
| @@ -1599,13 +1283,11 @@ | ||
| 1599 | 1283 | } |
| 1600 | 1284 | } |
| 1601 | 1285 | |
| 1602 | 1286 | // If we get here, no intent matched OR the intent didn't provide a usable response |
| 1603 | - | |
| 1287 | + | |
| 1604 | 1288 | // Step 4: Generate AI response |
| 1605 | - // Get session start timestamp - when persistence is OFF, only include messages from this page load | |
| 1606 | - $session_start_timestamp = isset($_POST['session_start_timestamp']) ? intval($_POST['session_start_timestamp']) : 0; | |
| 1607 | - $conversation_history = $this->mxchat_fetch_conversation_history_for_ai($session_id, $session_start_timestamp); | |
| 1289 | + $conversation_history = $this->mxchat_fetch_conversation_history_for_ai($session_id); | |
| 1608 | 1290 | $this->mxchat_increment_chat_count(); |
| 1609 | 1291 | |
| 1610 | 1292 | // Generate embedding for the user's query - USE BOT-SPECIFIC API KEY |
| 1611 | 1293 | $api_key = $current_options['api_key'] ?? $this->options['api_key']; |
| @@ -1614,50 +1296,22 @@ | ||
| 1614 | 1296 | // Check if the embedding generation returned an error |
| 1615 | 1297 | if (is_array($user_message_embedding) && isset($user_message_embedding['error'])) { |
| 1616 | 1298 | $error_message = $user_message_embedding['error']; |
| 1617 | 1299 | $error_code = $user_message_embedding['error_code'] ?? 'embedding_error'; |
| 1618 | - | |
| 1619 | - // FIXED: Send error in appropriate format based on streaming mode | |
| 1620 | - if ($is_streaming) { | |
| 1621 | - echo "data: " . json_encode([ | |
| 1622 | - 'error' => true, | |
| 1623 | - 'error_message' => $error_message, | |
| 1624 | - 'error_code' => $error_code, | |
| 1625 | - 'text' => $error_message, | |
| 1626 | - 'message' => $error_message | |
| 1627 | - ]) . "\n\n"; | |
| 1628 | - echo "data: [DONE]\n\n"; | |
| 1629 | - flush(); | |
| 1630 | - } else { | |
| 1631 | - wp_send_json_error([ | |
| 1632 | - 'error_message' => $error_message, | |
| 1633 | - 'error_code' => $error_code | |
| 1634 | - ]); | |
| 1635 | - } | |
| 1300 | + | |
| 1301 | + wp_send_json_error([ | |
| 1302 | + 'error_message' => $error_message, | |
| 1303 | + 'error_code' => $error_code | |
| 1304 | + ]); | |
| 1636 | 1305 | wp_die(); |
| 1637 | 1306 | } |
| 1638 | - | |
| 1307 | + | |
| 1639 | 1308 | // Check if the embedding is valid |
| 1640 | 1309 | if (!is_array($user_message_embedding) || empty($user_message_embedding)) { |
| 1641 | - $error_message = esc_html__('Unable to process your message. The embedding service is not responding correctly.', 'mxchat'); | |
| 1642 | - | |
| 1643 | - // FIXED: Send error in appropriate format based on streaming mode | |
| 1644 | - if ($is_streaming) { | |
| 1645 | - echo "data: " . json_encode([ | |
| 1646 | - 'error' => true, | |
| 1647 | - 'error_message' => $error_message, | |
| 1648 | - 'error_code' => 'invalid_embedding', | |
| 1649 | - 'text' => $error_message, | |
| 1650 | - 'message' => $error_message | |
| 1651 | - ]) . "\n\n"; | |
| 1652 | - echo "data: [DONE]\n\n"; | |
| 1653 | - flush(); | |
| 1654 | - } else { | |
| 1655 | - wp_send_json_error([ | |
| 1656 | - 'error_message' => $error_message, | |
| 1657 | - 'error_code' => 'invalid_embedding' | |
| 1658 | - ]); | |
| 1659 | - } | |
| 1310 | + wp_send_json_error([ | |
| 1311 | + 'error_message' => esc_html__('Unable to process your message. The embedding service is not responding correctly.', 'mxchat'), | |
| 1312 | + 'error_code' => 'invalid_embedding' | |
| 1313 | + ]); | |
| 1660 | 1314 | wp_die(); |
| 1661 | 1315 | } |
| 1662 | 1316 | |
| 1663 | 1317 | // Build context with both knowledge base and PDF content if available |
| @@ -1683,71 +1337,26 @@ | ||
| 1683 | 1337 | $context_content .= "Page Content: " . $page_context['content'] . "\n"; |
| 1684 | 1338 | $context_content .= "===== END CURRENT PAGE CONTEXT =====\n\n"; |
| 1685 | 1339 | } |
| 1686 | 1340 | |
| 1687 | - // Get relevant content from knowledge base - PASS BOT_ID and MESSAGE for Vector Store | |
| 1688 | - $relevant_content = $this->mxchat_find_relevant_content($user_message_embedding, $bot_id, $message); | |
| 1689 | - | |
| 1690 | - // NEW: Also extract URLs from system instructions (only if citation links enabled) | |
| 1691 | - // Use fresh options to ensure we get the latest setting value | |
| 1692 | - $fresh_options = get_option('mxchat_options', []); | |
| 1693 | - $citation_links_enabled = isset($fresh_options['citation_links_toggle']) ? ($fresh_options['citation_links_toggle'] === 'on') : true; | |
| 1341 | + // Get relevant content from knowledge base - PASS BOT_ID | |
| 1342 | + $relevant_content = $this->mxchat_find_relevant_content($user_message_embedding, $bot_id); | |
| 1694 | 1343 | |
| 1695 | - $system_instructions = $this->get_system_instructions($bot_id, $session_id); | |
| 1696 | - if ($citation_links_enabled && !empty($system_instructions)) { | |
| 1697 | - preg_match_all( | |
| 1698 | - '#\bhttps?://[^\s<>"\']+#i', | |
| 1699 | - $system_instructions, | |
| 1700 | - $system_instruction_urls | |
| 1701 | - ); | |
| 1702 | - | |
| 1703 | - if (!empty($system_instruction_urls[0])) { | |
| 1704 | - // Merge with existing valid URLs | |
| 1705 | - $this->current_valid_urls = array_merge( | |
| 1706 | - $this->current_valid_urls, | |
| 1707 | - $system_instruction_urls[0] | |
| 1708 | - ); | |
| 1709 | - // Remove duplicates | |
| 1710 | - $this->current_valid_urls = array_unique($this->current_valid_urls); | |
| 1711 | - | |
| 1712 | - //error_log("Added " . count($system_instruction_urls[0]) . " URLs from system instructions"); | |
| 1713 | - } | |
| 1344 | + // ===== CAPTURE REAL SIMILARITY DATA FOR ADMINS ===== | |
| 1345 | + if ($testing_data !== null && $this->last_similarity_analysis !== null) { | |
| 1346 | + // Update testing data with the REAL similarity analysis | |
| 1347 | + $testing_data['top_matches'] = $this->last_similarity_analysis['top_matches']; | |
| 1348 | + $testing_data['total_documents_checked'] = $this->last_similarity_analysis['total_checked'] ?? 0; | |
| 1349 | + $testing_data['knowledge_base_type'] = $this->last_similarity_analysis['knowledge_base_type']; | |
| 1714 | 1350 | } |
| 1351 | + // ===== END SIMILARITY DATA CAPTURE ===== | |
| 1715 | 1352 | |
| 1716 | -// ===== CAPTURE REAL SIMILARITY DATA FOR ADMINS ===== | |
| 1717 | -if ($testing_data !== null && $this->last_similarity_analysis !== null) { | |
| 1718 | - // Update testing data with the REAL similarity analysis | |
| 1719 | - $testing_data['top_matches'] = $this->last_similarity_analysis['top_matches']; | |
| 1720 | - $testing_data['total_documents_checked'] = $this->last_similarity_analysis['total_checked'] ?? 0; | |
| 1721 | - $testing_data['knowledge_base_type'] = $this->last_similarity_analysis['knowledge_base_type']; | |
| 1722 | - $testing_data['sources_used'] = $this->last_similarity_analysis['sources_used'] ?? 0; | |
| 1723 | - $testing_data['total_chunks_used'] = $this->last_similarity_analysis['total_chunks_used'] ?? 0; | |
| 1724 | -} | |
| 1725 | -// ===== END SIMILARITY DATA CAPTURE ===== | |
| 1726 | - | |
| 1727 | -// NEW: Add valid URLs to testing data for admin panel display (AFTER similarity data) | |
| 1728 | -if ($testing_data !== null && !empty($this->current_valid_urls)) { | |
| 1729 | - $testing_data['approved_urls'] = array_values($this->current_valid_urls); | |
| 1730 | - //error_log("Added " . count($this->current_valid_urls) . " approved URLs to testing data"); | |
| 1731 | -} | |
| 1732 | - | |
| 1733 | 1353 | if (!empty($relevant_content)) { |
| 1734 | 1354 | $context_content .= "===== OFFICIAL KNOWLEDGE DATABASE CONTENT =====\n" . $relevant_content . "\n===== END OF OFFICIAL KNOWLEDGE DATABASE CONTENT =====\n\n"; |
| 1735 | 1355 | } else { |
| 1736 | 1356 | $context_content .= "===== NO RELEVANT CONTENT FOUND IN KNOWLEDGE DATABASE =====\n"; |
| 1737 | 1357 | } |
| 1738 | - | |
| 1739 | - // NEW: Add approved URLs list to context for AI (only if citation links enabled) | |
| 1740 | - if ($citation_links_enabled && !empty($this->current_valid_urls)) { | |
| 1741 | - $context_content .= "===== APPROVED URLS FOR CITATIONS =====\n"; | |
| 1742 | - $context_content .= "You may ONLY use these exact URLs in your response:\n"; | |
| 1743 | - foreach ($this->current_valid_urls as $url) { | |
| 1744 | - $context_content .= "- " . $url . "\n"; | |
| 1745 | - } | |
| 1746 | - $context_content .= "\nCRITICAL: Do NOT create, modify, extend, or invent any other URLs. "; | |
| 1747 | - $context_content .= "===== END APPROVED URLS =====\n\n"; | |
| 1748 | - } | |
| 1749 | - | |
| 1358 | + | |
| 1750 | 1359 | // Check for and include PDF content |
| 1751 | 1360 | $pdf_url = get_transient('mxchat_pdf_url_' . $session_id); |
| 1752 | 1361 | $pdf_embeddings = get_transient('mxchat_pdf_embeddings_' . $session_id); |
| 1753 | 1362 | $pdf_filename = get_transient('mxchat_pdf_filename_' . $session_id); |
| @@ -1779,9 +1388,9 @@ | ||
| 1779 | 1388 | |
| 1780 | 1389 | $context_content = apply_filters('mxchat_prepare_context', $context_content, $session_id); |
| 1781 | 1390 | |
| 1782 | 1391 | // Extract model from current options for bot-specific model support |
| 1783 | - $selected_model = isset($current_options['model']) ? $current_options['model'] : 'gpt-5.1-chat-latest'; | |
| 1392 | + $selected_model = isset($current_options['model']) ? $current_options['model'] : 'gpt-4o'; | |
| 1784 | 1393 | |
| 1785 | 1394 | $response = $this->mxchat_generate_response( |
| 1786 | 1395 | $context_content, |
| 1787 | 1396 | $current_options['api_key'] ?? $this->options['api_key'], |
| @@ -1788,14 +1397,13 @@ | ||
| 1788 | 1397 | $current_options['xai_api_key'] ?? $this->options['xai_api_key'], |
| 1789 | 1398 | $current_options['claude_api_key'] ?? $this->options['claude_api_key'], |
| 1790 | 1399 | $current_options['deepseek_api_key'] ?? $this->options['deepseek_api_key'], |
| 1791 | 1400 | $current_options['gemini_api_key'] ?? $this->options['gemini_api_key'], |
| 1792 | - $current_options['openrouter_api_key'] ?? $this->options['openrouter_api_key'], | |
| 1793 | 1401 | $conversation_history, |
| 1794 | 1402 | $is_streaming, |
| 1795 | 1403 | $session_id, |
| 1796 | 1404 | $testing_data, |
| 1797 | - $selected_model | |
| 1405 | + $selected_model // ADD THIS LINE | |
| 1798 | 1406 | ); |
| 1799 | 1407 | |
| 1800 | 1408 | // Handle streaming vs non-streaming responses |
| 1801 | 1409 | if ($is_streaming) { |
| @@ -1803,27 +1411,11 @@ | ||
| 1803 | 1411 | if ($response === true) { |
| 1804 | 1412 | wp_die(); |
| 1805 | 1413 | } |
| 1806 | 1414 | // If we get here, streaming fell back to regular response, continue |
| 1807 | - // But if there's an error, we need to send it as SSE format since headers are already set | |
| 1808 | - if (is_array($response) && isset($response['error'])) { | |
| 1809 | - $error_message = $response['error']; | |
| 1810 | - $error_code = $response['error_code'] ?? 'api_error'; | |
| 1811 | - // Send error in SSE format that the client JS can handle | |
| 1812 | - echo "data: " . json_encode([ | |
| 1813 | - 'error' => true, | |
| 1814 | - 'error_message' => $error_message, | |
| 1815 | - 'error_code' => $error_code, | |
| 1816 | - 'text' => $error_message, // Also include as text for fallback handling | |
| 1817 | - 'message' => $error_message | |
| 1818 | - ]) . "\n\n"; | |
| 1819 | - echo "data: [DONE]\n\n"; | |
| 1820 | - flush(); | |
| 1821 | - wp_die(); | |
| 1822 | - } | |
| 1823 | 1415 | } |
| 1824 | - | |
| 1825 | - // Check if the response is an error array (non-streaming mode) | |
| 1416 | + | |
| 1417 | + // Check if the response is an error array | |
| 1826 | 1418 | if (is_array($response) && isset($response['error'])) { |
| 1827 | 1419 | wp_send_json_error([ |
| 1828 | 1420 | 'error_message' => $response['error'], |
| 1829 | 1421 | 'error_code' => $response['error_code'] ?? 'api_error' |
| @@ -1830,51 +1422,11 @@ | ||
| 1830 | 1422 | ]); |
| 1831 | 1423 | wp_die(); |
| 1832 | 1424 | } |
| 1833 | 1425 | |
| 1834 | - // DEBUG: Check what we have | |
| 1835 | - //error_log("=== BEFORE URL VALIDATION ==="); | |
| 1836 | - //error_log("current_valid_urls is empty? " . (empty($this->current_valid_urls) ? 'YES' : 'NO')); | |
| 1837 | - //error_log("current_valid_urls count: " . count($this->current_valid_urls)); | |
| 1838 | - //error_log("current_valid_urls content: " . print_r($this->current_valid_urls, true)); | |
| 1839 | - | |
| 1840 | - // If we get here, the response is valid text - now validate URLs | |
| 1841 | - if (!empty($this->current_valid_urls)) { | |
| 1842 | - //error_log("CALLING validate_and_clean_urls"); | |
| 1843 | - $response = $this->validate_and_clean_urls($response, $this->current_valid_urls); | |
| 1844 | - } else { | |
| 1845 | - //error_log("SKIPPING validation - current_valid_urls is empty"); | |
| 1846 | - } | |
| 1847 | - // ===== END URL VALIDATION ===== | |
| 1426 | + // If we get here, the response is valid text | |
| 1427 | + $this->mxchat_save_chat_message($session_id, 'bot', $response); | |
| 1848 | 1428 | |
| 1849 | - // Prepare RAG context data for storage (only include documents used for context) | |
| 1850 | - $rag_context_for_storage = null; | |
| 1851 | - $has_rag_data = $this->last_similarity_analysis !== null && !empty($this->last_similarity_analysis['top_matches']); | |
| 1852 | - $has_action_data = isset($this->last_action_analysis) && !empty($this->last_action_analysis); | |
| 1853 | - | |
| 1854 | - if ($has_rag_data || $has_action_data) { | |
| 1855 | - $rag_context_for_storage = []; | |
| 1856 | - | |
| 1857 | - // Add RAG/source data if available | |
| 1858 | - if ($has_rag_data) { | |
| 1859 | - $rag_context_for_storage['top_matches'] = $this->last_similarity_analysis['top_matches']; | |
| 1860 | - $rag_context_for_storage['approved_urls'] = $this->current_valid_urls ?? []; | |
| 1861 | - $rag_context_for_storage['similarity_threshold'] = $this->last_similarity_analysis['threshold_used'] ?? 0.35; | |
| 1862 | - $rag_context_for_storage['knowledge_base_type'] = $this->last_similarity_analysis['knowledge_base_type'] ?? 'WordPress Database'; | |
| 1863 | - $rag_context_for_storage['total_documents_checked'] = $this->last_similarity_analysis['total_checked'] ?? 0; | |
| 1864 | - $rag_context_for_storage['sources_used'] = $this->last_similarity_analysis['sources_used'] ?? 0; | |
| 1865 | - $rag_context_for_storage['total_chunks_used'] = $this->last_similarity_analysis['total_chunks_used'] ?? 0; | |
| 1866 | - } | |
| 1867 | - | |
| 1868 | - // Add action analysis data if available | |
| 1869 | - if ($has_action_data) { | |
| 1870 | - $rag_context_for_storage['action_analysis'] = $this->last_action_analysis; | |
| 1871 | - } | |
| 1872 | - } | |
| 1873 | - | |
| 1874 | - // Save the cleaned response with RAG context | |
| 1875 | - $this->mxchat_save_chat_message($session_id, 'bot', $response, null, $rag_context_for_storage); | |
| 1876 | - | |
| 1877 | 1429 | // Step 5: Save additional content if available |
| 1878 | 1430 | if (!empty($this->productCardHtml)) { |
| 1879 | 1431 | $this->mxchat_save_chat_message($session_id, 'bot', $this->productCardHtml); |
| 1880 | 1432 | } |
| @@ -1883,13 +1435,8 @@ | ||
| 1883 | 1435 | $this->mxchat_save_chat_message($session_id, 'bot', $this->fallbackResponse['html']); |
| 1884 | 1436 | } |
| 1885 | 1437 | |
| 1886 | 1438 | // Step 6: Return the response |
| 1887 | - // DEBUG: Check if newlines exist in the response | |
| 1888 | - //error_log("=== MXCHAT NON-STREAMING RESPONSE DEBUG ==="); | |
| 1889 | - //error_log("Response has newlines: " . (strpos($response, "\n") !== false ? 'YES' : 'NO')); | |
| 1890 | - //error_log("Response first 500 chars: " . substr($response, 0, 500)); | |
| 1891 | - | |
| 1892 | 1439 | $response_data = [ |
| 1893 | 1440 | 'text' => $response, |
| 1894 | 1441 | 'html' => !empty($this->productCardHtml) ? $this->productCardHtml : ($this->fallbackResponse['html'] ?? ''), |
| 1895 | 1442 | 'session_id' => $session_id |
| @@ -1894,18 +1441,8 @@ | ||
| 1894 | 1441 | 'html' => !empty($this->productCardHtml) ? $this->productCardHtml : ($this->fallbackResponse['html'] ?? ''), |
| 1895 | 1442 | 'session_id' => $session_id |
| 1896 | 1443 | ]; |
| 1897 | 1444 | |
| 1898 | - // Include vectorstore error info for admin debugging (only visible to admins via testing_data) | |
| 1899 | - if (!empty($this->last_vectorstore_error) && $testing_data !== null) { | |
| 1900 | - $testing_data['vectorstore_error'] = $this->last_vectorstore_error; | |
| 1901 | - } | |
| 1902 | - | |
| 1903 | - // Also pass it as a top-level field so JS can show a better error message to admins | |
| 1904 | - if (!empty($this->last_vectorstore_error) && current_user_can('manage_options')) { | |
| 1905 | - $response_data['vectorstore_error'] = $this->last_vectorstore_error; | |
| 1906 | - } | |
| 1907 | - | |
| 1908 | 1445 | // Always add testing data for admins (no toggle needed) |
| 1909 | 1446 | if ($testing_data !== null) { |
| 1910 | 1447 | $response_data['testing_data'] = $testing_data; |
| 1911 | 1448 | } |
| @@ -1913,8 +1450,9 @@ | ||
| 1913 | 1450 | wp_send_json($response_data); |
| 1914 | 1451 | wp_die(); |
| 1915 | 1452 | } |
| 1916 | 1453 | |
| 1454 | + | |
| 1917 | 1455 | /** |
| 1918 | 1456 | * Get bot-specific options for multi-bot functionality |
| 1919 | 1457 | * Falls back to default options if bot_id is 'default' or multi-bot add-on is not active |
| 1920 | 1458 | */ |
| @@ -1919,12 +1457,12 @@ | ||
| 1919 | 1457 | * Falls back to default options if bot_id is 'default' or multi-bot add-on is not active |
| 1920 | 1458 | */ |
| 1921 | 1459 | // Also debug the bot options retrieval |
| 1922 | 1460 | private function get_bot_options($bot_id = 'default') { |
| 1923 | - //error_log("MXCHAT DEBUG: get_bot_options called for bot: " . $bot_id); | |
| 1461 | + error_log("MXCHAT DEBUG: get_bot_options called for bot: " . $bot_id); | |
| 1924 | 1462 | |
| 1925 | 1463 | if ($bot_id === 'default' || !class_exists('MxChat_Multi_Bot_Manager')) { |
| 1926 | - //error_log("MXCHAT DEBUG: Using default options (no multi-bot or bot is 'default')"); | |
| 1464 | + error_log("MXCHAT DEBUG: Using default options (no multi-bot or bot is 'default')"); | |
| 1927 | 1465 | return array(); |
| 1928 | 1466 | } |
| 1929 | 1467 | |
| 1930 | 1468 | $bot_options = apply_filters('mxchat_get_bot_options', array(), $bot_id); |
| @@ -1929,11 +1467,11 @@ | ||
| 1929 | 1467 | |
| 1930 | 1468 | $bot_options = apply_filters('mxchat_get_bot_options', array(), $bot_id); |
| 1931 | 1469 | |
| 1932 | 1470 | if (!empty($bot_options)) { |
| 1933 | - //error_log("MXCHAT DEBUG: Got bot-specific options from filter"); | |
| 1471 | + error_log("MXCHAT DEBUG: Got bot-specific options from filter"); | |
| 1934 | 1472 | if (isset($bot_options['similarity_threshold'])) { |
| 1935 | - //error_log(" - similarity_threshold: " . $bot_options['similarity_threshold']); | |
| 1473 | + error_log(" - similarity_threshold: " . $bot_options['similarity_threshold']); | |
| 1936 | 1474 | } |
| 1937 | 1475 | } |
| 1938 | 1476 | |
| 1939 | 1477 | return is_array($bot_options) ? $bot_options : array(); |
| @@ -1944,13 +1482,13 @@ | ||
| 1944 | 1482 | * Used in the knowledge retrieval functions |
| 1945 | 1483 | */ |
| 1946 | 1484 | // Also add debugging to your get_bot_pinecone_config function |
| 1947 | 1485 | private function get_bot_pinecone_config($bot_id = 'default') { |
| 1948 | - //error_log("MXCHAT DEBUG: get_bot_pinecone_config called for bot: " . $bot_id); | |
| 1486 | + error_log("MXCHAT DEBUG: get_bot_pinecone_config called for bot: " . $bot_id); | |
| 1949 | 1487 | |
| 1950 | 1488 | // If default bot or multi-bot add-on not active, use default Pinecone config |
| 1951 | 1489 | if ($bot_id === 'default' || !class_exists('MxChat_Multi_Bot_Manager')) { |
| 1952 | - //error_log("MXCHAT DEBUG: Using default Pinecone config (no multi-bot or bot is 'default')"); | |
| 1490 | + error_log("MXCHAT DEBUG: Using default Pinecone config (no multi-bot or bot is 'default')"); | |
| 1953 | 1491 | $addon_options = get_option('mxchat_pinecone_addon_options', array()); |
| 1954 | 1492 | $config = array( |
| 1955 | 1493 | 'use_pinecone' => (isset($addon_options['mxchat_use_pinecone']) && $addon_options['mxchat_use_pinecone'] === '1'), |
| 1956 | 1494 | 'api_key' => $addon_options['mxchat_pinecone_api_key'] ?? '', |
| @@ -1956,24 +1494,24 @@ | ||
| 1956 | 1494 | 'api_key' => $addon_options['mxchat_pinecone_api_key'] ?? '', |
| 1957 | 1495 | 'host' => $addon_options['mxchat_pinecone_host'] ?? '', |
| 1958 | 1496 | 'namespace' => $addon_options['mxchat_pinecone_namespace'] ?? '' |
| 1959 | 1497 | ); |
| 1960 | - //error_log("MXCHAT DEBUG: Default config - use_pinecone: " . ($config['use_pinecone'] ? 'true' : 'false')); | |
| 1498 | + error_log("MXCHAT DEBUG: Default config - use_pinecone: " . ($config['use_pinecone'] ? 'true' : 'false')); | |
| 1961 | 1499 | return $config; |
| 1962 | 1500 | } |
| 1963 | 1501 | |
| 1964 | - //error_log("MXCHAT DEBUG: Calling filter 'mxchat_get_bot_pinecone_config' for bot: " . $bot_id); | |
| 1502 | + error_log("MXCHAT DEBUG: Calling filter 'mxchat_get_bot_pinecone_config' for bot: " . $bot_id); | |
| 1965 | 1503 | |
| 1966 | 1504 | // Hook for multi-bot add-on to provide bot-specific Pinecone config |
| 1967 | 1505 | $bot_pinecone_config = apply_filters('mxchat_get_bot_pinecone_config', array(), $bot_id); |
| 1968 | 1506 | |
| 1969 | 1507 | if (!empty($bot_pinecone_config)) { |
| 1970 | - //error_log("MXCHAT DEBUG: Got bot-specific config from filter"); | |
| 1971 | - //error_log(" - use_pinecone: " . (isset($bot_pinecone_config['use_pinecone']) ? ($bot_pinecone_config['use_pinecone'] ? 'true' : 'false') : 'not set')); | |
| 1972 | - //error_log(" - host: " . ($bot_pinecone_config['host'] ?? 'not set')); | |
| 1973 | - //error_log(" - namespace: " . ($bot_pinecone_config['namespace'] ?? 'not set')); | |
| 1508 | + error_log("MXCHAT DEBUG: Got bot-specific config from filter"); | |
| 1509 | + error_log(" - use_pinecone: " . (isset($bot_pinecone_config['use_pinecone']) ? ($bot_pinecone_config['use_pinecone'] ? 'true' : 'false') : 'not set')); | |
| 1510 | + error_log(" - host: " . ($bot_pinecone_config['host'] ?? 'not set')); | |
| 1511 | + error_log(" - namespace: " . ($bot_pinecone_config['namespace'] ?? 'not set')); | |
| 1974 | 1512 | } else { |
| 1975 | - //error_log("MXCHAT DEBUG: Filter returned empty config!"); | |
| 1513 | + error_log("MXCHAT DEBUG: Filter returned empty config!"); | |
| 1976 | 1514 | } |
| 1977 | 1515 | |
| 1978 | 1516 | return is_array($bot_pinecone_config) ? $bot_pinecone_config : array(); |
| 1979 | 1517 | } |
| @@ -1983,63 +1521,35 @@ | ||
| 1983 | 1521 | private function mxchat_check_intent_and_invoke_callback($message, $user_id, $session_id) { |
| 1984 | 1522 | global $wpdb; |
| 1985 | 1523 | $chat_mode = get_option("mxchat_mode_{$session_id}", 'ai'); |
| 1986 | 1524 | |
| 1987 | - // Get the current bot_id | |
| 1525 | + // NEW: Get the current bot_id | |
| 1988 | 1526 | $current_bot_id = $this->get_current_bot_id($session_id); |
| 1989 | 1527 | |
| 1990 | 1528 | // Generate the user embedding |
| 1991 | 1529 | $user_embedding = $this->mxchat_generate_embedding($message, $this->options['api_key']); |
| 1992 | - | |
| 1530 | + | |
| 1993 | 1531 | // Check if embedding generation returned an error |
| 1994 | 1532 | if (is_array($user_embedding) && isset($user_embedding['error'])) { |
| 1995 | 1533 | $error_message = $user_embedding['error']; |
| 1996 | 1534 | $error_code = $user_embedding['error_code'] ?? 'embedding_error'; |
| 1997 | - | |
| 1998 | - // FIXED: Send error in appropriate format based on streaming mode | |
| 1999 | - if ($this->is_streaming) { | |
| 2000 | - echo "data: " . json_encode([ | |
| 2001 | - 'error' => true, | |
| 2002 | - 'error_message' => $error_message, | |
| 2003 | - 'error_code' => $error_code, | |
| 2004 | - 'text' => $error_message, | |
| 2005 | - 'message' => $error_message | |
| 2006 | - ]) . "\n\n"; | |
| 2007 | - echo "data: [DONE]\n\n"; | |
| 2008 | - flush(); | |
| 2009 | - } else { | |
| 2010 | - wp_send_json_error([ | |
| 2011 | - 'error_message' => $error_message, | |
| 2012 | - 'error_code' => $error_code | |
| 2013 | - ]); | |
| 2014 | - } | |
| 1535 | + | |
| 1536 | + wp_send_json_error([ | |
| 1537 | + 'error_message' => $error_message, | |
| 1538 | + 'error_code' => $error_code | |
| 1539 | + ]); | |
| 2015 | 1540 | wp_die(); |
| 2016 | 1541 | } |
| 2017 | - | |
| 1542 | + | |
| 2018 | 1543 | // Check if embedding is valid |
| 2019 | 1544 | if (!is_array($user_embedding) || empty($user_embedding)) { |
| 2020 | - $error_message = esc_html__('Unable to process your message. The embedding service is not responding correctly.', 'mxchat'); | |
| 2021 | - | |
| 2022 | - // FIXED: Send error in appropriate format based on streaming mode | |
| 2023 | - if ($this->is_streaming) { | |
| 2024 | - echo "data: " . json_encode([ | |
| 2025 | - 'error' => true, | |
| 2026 | - 'error_message' => $error_message, | |
| 2027 | - 'error_code' => 'invalid_embedding', | |
| 2028 | - 'text' => $error_message, | |
| 2029 | - 'message' => $error_message | |
| 2030 | - ]) . "\n\n"; | |
| 2031 | - echo "data: [DONE]\n\n"; | |
| 2032 | - flush(); | |
| 2033 | - } else { | |
| 2034 | - wp_send_json_error([ | |
| 2035 | - 'error_message' => $error_message, | |
| 2036 | - 'error_code' => 'invalid_embedding' | |
| 2037 | - ]); | |
| 2038 | - } | |
| 1545 | + wp_send_json_error([ | |
| 1546 | + 'error_message' => esc_html__('Unable to process your message. The embedding service is not responding correctly.', 'mxchat'), | |
| 1547 | + 'error_code' => 'invalid_embedding' | |
| 1548 | + ]); | |
| 2039 | 1549 | wp_die(); |
| 2040 | 1550 | } |
| 2041 | - | |
| 1551 | + | |
| 2042 | 1552 | // Fetch intents from the database |
| 2043 | 1553 | $table_name = $wpdb->prefix . 'mxchat_intents'; |
| 2044 | 1554 | if ($chat_mode === 'agent') { |
| 2045 | 1555 | $query = $wpdb->prepare( |
| @@ -2049,29 +1559,19 @@ | ||
| 2049 | 1559 | $intents = $wpdb->get_results($query); |
| 2050 | 1560 | } else { |
| 2051 | 1561 | $intents = $wpdb->get_results("SELECT * FROM $table_name WHERE enabled = 1 OR enabled IS NULL"); |
| 2052 | 1562 | } |
| 2053 | - | |
| 1563 | + | |
| 2054 | 1564 | if (empty($intents)) { |
| 2055 | 1565 | return false; |
| 2056 | 1566 | } |
| 2057 | - | |
| 2058 | - // Prefetch individual phrase vectors from wp_mxchat_intent_phrases (grouped by intent_id) | |
| 2059 | - $phrases_table = $wpdb->prefix . 'mxchat_intent_phrases'; | |
| 2060 | - $phrases_by_intent = []; | |
| 2061 | - if ($wpdb->get_var("SHOW TABLES LIKE '$phrases_table'") === $phrases_table) { | |
| 2062 | - $all_phrases = $wpdb->get_results("SELECT intent_id, phrase, embedding_vector FROM $phrases_table"); | |
| 2063 | - foreach ($all_phrases as $p) { | |
| 2064 | - $phrases_by_intent[$p->intent_id][] = $p; | |
| 2065 | - } | |
| 2066 | - } | |
| 2067 | - | |
| 1567 | + | |
| 2068 | 1568 | $highest_similarity = -INF; |
| 2069 | 1569 | $matched_intent = null; |
| 2070 | - | |
| 1570 | + | |
| 2071 | 1571 | // Array to store action analysis for testing panel |
| 2072 | 1572 | $action_analysis = []; |
| 2073 | - | |
| 1573 | + | |
| 2074 | 1574 | foreach ($intents as $intent) { |
| 2075 | 1575 | // Additional check for enabled state |
| 2076 | 1576 | $is_enabled = isset($intent->enabled) ? (bool)$intent->enabled : true; |
| 2077 | 1577 | if (!$is_enabled) { |
| @@ -2076,56 +1576,26 @@ | ||
| 2076 | 1576 | $is_enabled = isset($intent->enabled) ? (bool)$intent->enabled : true; |
| 2077 | 1577 | if (!$is_enabled) { |
| 2078 | 1578 | continue; |
| 2079 | 1579 | } |
| 2080 | - | |
| 2081 | - // Check if this action is enabled for the current bot | |
| 1580 | + | |
| 1581 | + // NEW: Check if this action is enabled for the current bot | |
| 2082 | 1582 | if (!$this->is_action_enabled_for_bot($intent, $current_bot_id)) { |
| 2083 | 1583 | continue; |
| 2084 | 1584 | } |
| 2085 | - | |
| 2086 | - $best_similarity = -INF; | |
| 2087 | - $matched_phrase_text = ''; | |
| 2088 | - | |
| 2089 | - // Check legacy embedding vector (existing behavior) | |
| 1585 | + | |
| 2090 | 1586 | $intent_embedding_serialized = $intent->embedding_vector; |
| 2091 | 1587 | $intent_embedding = $intent_embedding_serialized |
| 2092 | 1588 | ? unserialize($intent_embedding_serialized, ['allowed_classes' => false]) |
| 2093 | 1589 | : null; |
| 2094 | - | |
| 2095 | - if (is_array($intent_embedding) && !empty($intent_embedding)) { | |
| 2096 | - $legacy_similarity = $this->mxchat_calculate_cosine_similarity($user_embedding, $intent_embedding); | |
| 2097 | - if ($legacy_similarity > $best_similarity) { | |
| 2098 | - $best_similarity = $legacy_similarity; | |
| 2099 | - $matched_phrase_text = 'legacy'; | |
| 2100 | - } | |
| 2101 | - } | |
| 2102 | - | |
| 2103 | - // Check individual phrase vectors | |
| 2104 | - if (isset($phrases_by_intent[$intent->id])) { | |
| 2105 | - foreach ($phrases_by_intent[$intent->id] as $phrase_row) { | |
| 2106 | - $phrase_embedding = $phrase_row->embedding_vector | |
| 2107 | - ? unserialize($phrase_row->embedding_vector, ['allowed_classes' => false]) | |
| 2108 | - : null; | |
| 2109 | - if (!is_array($phrase_embedding)) { | |
| 2110 | - continue; | |
| 2111 | - } | |
| 2112 | - $phrase_similarity = $this->mxchat_calculate_cosine_similarity($user_embedding, $phrase_embedding); | |
| 2113 | - if ($phrase_similarity > $best_similarity) { | |
| 2114 | - $best_similarity = $phrase_similarity; | |
| 2115 | - $matched_phrase_text = $phrase_row->phrase; | |
| 2116 | - } | |
| 2117 | - } | |
| 2118 | - } | |
| 2119 | - | |
| 2120 | - // Skip if no valid embedding was found at all | |
| 2121 | - if ($best_similarity === -INF) { | |
| 1590 | + | |
| 1591 | + if (!is_array($intent_embedding)) { | |
| 2122 | 1592 | continue; |
| 2123 | 1593 | } |
| 2124 | - | |
| 2125 | - $similarity = $best_similarity; | |
| 1594 | + | |
| 1595 | + $similarity = $this->mxchat_calculate_cosine_similarity($user_embedding, $intent_embedding); | |
| 2126 | 1596 | $intent_threshold = isset($intent->similarity_threshold) ? $intent->similarity_threshold : 0.85; |
| 2127 | - | |
| 1597 | + | |
| 2128 | 1598 | // Store action analysis data for testing panel |
| 2129 | 1599 | $action_analysis[] = [ |
| 2130 | 1600 | 'intent_label' => $intent->intent_label, |
| 2131 | 1601 | 'callback_function' => $intent->callback_function, |
| @@ -2133,12 +1603,11 @@ | ||
| 2133 | 1603 | 'similarity_percentage' => round($similarity * 100, 2), |
| 2134 | 1604 | 'threshold' => $intent_threshold, |
| 2135 | 1605 | 'threshold_percentage' => round($intent_threshold * 100, 2), |
| 2136 | 1606 | 'above_threshold' => $similarity >= $intent_threshold, |
| 2137 | - 'matched_phrase' => $matched_phrase_text, | |
| 2138 | 1607 | 'triggered' => false // Will be updated below if this intent is triggered |
| 2139 | 1608 | ]; |
| 2140 | - | |
| 1609 | + | |
| 2141 | 1610 | if ($similarity >= $intent_threshold && $similarity > $highest_similarity) { |
| 2142 | 1611 | $highest_similarity = $similarity; |
| 2143 | 1612 | $matched_intent = $intent; |
| 2144 | 1613 | } |
| @@ -2209,22 +1678,16 @@ | ||
| 2209 | 1678 | // If enabled_bots column doesn't exist or is null, assume it's enabled for all bots (backward compatibility) |
| 2210 | 1679 | if (!isset($intent->enabled_bots) || empty($intent->enabled_bots)) { |
| 2211 | 1680 | return true; |
| 2212 | 1681 | } |
| 2213 | - | |
| 1682 | + | |
| 2214 | 1683 | $enabled_bots = json_decode($intent->enabled_bots, true); |
| 2215 | - | |
| 1684 | + | |
| 2216 | 1685 | // If JSON decode fails or returns empty array, assume enabled for all (backward compatibility) |
| 2217 | 1686 | if (!is_array($enabled_bots) || empty($enabled_bots)) { |
| 2218 | 1687 | return true; |
| 2219 | 1688 | } |
| 2220 | - | |
| 2221 | - // Admin testing tab uses bot_id "testing" — treat it as "default" so all | |
| 2222 | - // default-bot actions are testable from the admin panel | |
| 2223 | - if ($bot_id === 'testing') { | |
| 2224 | - $bot_id = 'default'; | |
| 2225 | - } | |
| 2226 | - | |
| 1689 | + | |
| 2227 | 1690 | // Check if the current bot is in the enabled bots list |
| 2228 | 1691 | return in_array($bot_id, $enabled_bots); |
| 2229 | 1692 | } |
| 2230 | 1693 | |
| @@ -2263,17 +1726,17 @@ | ||
| 2263 | 1726 | |
| 2264 | 1727 | public function mxchat_generate_image($message, $user_id, $session_id) { |
| 2265 | 1728 | //error_log("Starting image generation for message: " . $message); |
| 2266 | 1729 | |
| 2267 | - // Prepare a prompt for OpenAI image generation | |
| 1730 | + // Prepare a prompt for DALL-E | |
| 2268 | 1731 | $prompt = esc_html__('Create an image of ', 'mxchat') . sanitize_text_field($message); |
| 2269 | - | |
| 1732 | + | |
| 2270 | 1733 | // Use the existing OpenAI API key |
| 2271 | 1734 | $openai_api_key = sanitize_text_field($this->options['api_key']); |
| 2272 | - | |
| 2273 | - // Call OpenAI GPT Image to generate an image | |
| 2274 | - $image_response = $this->mxchat_generate_openai_image($prompt, $openai_api_key); | |
| 2275 | 1735 | |
| 1736 | + // Call DALL-E to generate an image | |
| 1737 | + $image_response = $this->mxchat_generate_dalle_image($prompt, $openai_api_key); | |
| 1738 | + | |
| 2276 | 1739 | // Check if the response contains an image URL |
| 2277 | 1740 | if (isset($image_response['imageUrl'])) { |
| 2278 | 1741 | $image_url = esc_url_raw($image_response['imageUrl']); |
| 2279 | 1742 | |
| @@ -2316,103 +1779,24 @@ | ||
| 2316 | 1779 | // Return the response directly instead of relying on the property |
| 2317 | 1780 | return $this->fallbackResponse; |
| 2318 | 1781 | } |
| 2319 | 1782 | } |
| 2320 | - | |
| 2321 | -public function mxchat_generate_gemini_image($message, $user_id, $session_id) { | |
| 2322 | - $prompt = esc_html__('Create an image of ', 'mxchat') . sanitize_text_field($message); | |
| 2323 | - | |
| 2324 | - $gemini_api_key = sanitize_text_field($this->options['gemini_api_key'] ?? ''); | |
| 2325 | - if (empty($gemini_api_key)) { | |
| 2326 | - $response_text = esc_html__("Gemini API key is not configured.", 'mxchat'); | |
| 2327 | - $this->mxchat_save_chat_message($session_id, 'bot', $response_text); | |
| 2328 | - return ['text' => $response_text, 'html' => '', 'images' => []]; | |
| 2329 | - } | |
| 2330 | - | |
| 2331 | - $image_response = $this->mxchat_generate_imagen_image($prompt, $gemini_api_key); | |
| 2332 | - | |
| 2333 | - if (isset($image_response['imageUrl'])) { | |
| 2334 | - $image_url = esc_url_raw($image_response['imageUrl']); | |
| 2335 | - | |
| 2336 | - $response_html = '<img src="' . esc_url($image_url) . '" alt="' . esc_attr__('Generated Image', 'mxchat') . '" class="mxchat-generated-image" />'; | |
| 2337 | - $response_text = esc_html__('Here is the image I generated:', 'mxchat'); | |
| 2338 | - | |
| 2339 | - $this->mxchat_save_chat_message($session_id, 'bot', $response_text); | |
| 2340 | - $this->mxchat_save_chat_message($session_id, 'bot', $response_html); | |
| 2341 | - | |
| 2342 | - $this->fallbackResponse = [ | |
| 2343 | - 'text' => $response_text, | |
| 2344 | - 'html' => $response_html, | |
| 2345 | - 'images' => [$image_url] | |
| 2346 | - ]; | |
| 2347 | - | |
| 2348 | - return $this->fallbackResponse; | |
| 2349 | - } else { | |
| 2350 | - $response_text = esc_html__("I'm sorry, but I couldn't generate an image based on your request.", 'mxchat'); | |
| 2351 | - | |
| 2352 | - $this->mxchat_save_chat_message($session_id, 'bot', $response_text); | |
| 2353 | - | |
| 2354 | - $this->fallbackResponse = [ | |
| 2355 | - 'text' => $response_text, | |
| 2356 | - 'html' => '', | |
| 2357 | - 'images' => [] | |
| 2358 | - ]; | |
| 2359 | - | |
| 2360 | - return $this->fallbackResponse; | |
| 2361 | - } | |
| 2362 | -} | |
| 2363 | - | |
| 2364 | -private function mxchat_save_generated_image($base64_data, $mime_type = 'image/png', $prefix = 'mxchat-generated') { | |
| 2365 | - $extension = ($mime_type === 'image/jpeg') ? 'jpg' : 'png'; | |
| 2366 | - $filename = sanitize_file_name($prefix . '-' . wp_generate_uuid4() . '.' . $extension); | |
| 2367 | - $decoded = base64_decode($base64_data); | |
| 2368 | - | |
| 2369 | - if ($decoded === false) { | |
| 2370 | - return new \WP_Error('decode_failed', esc_html__('Failed to decode image data.', 'mxchat')); | |
| 2371 | - } | |
| 2372 | - | |
| 2373 | - $upload = wp_upload_bits($filename, null, $decoded); | |
| 2374 | - | |
| 2375 | - if (!empty($upload['error'])) { | |
| 2376 | - return new \WP_Error('upload_failed', $upload['error']); | |
| 2377 | - } | |
| 2378 | - | |
| 2379 | - $attach_id = wp_insert_attachment([ | |
| 2380 | - 'post_mime_type' => $mime_type, | |
| 2381 | - 'post_title' => $prefix, | |
| 2382 | - 'post_content' => '', | |
| 2383 | - 'post_status' => 'inherit', | |
| 2384 | - ], $upload['file']); | |
| 2385 | - | |
| 2386 | - if (is_wp_error($attach_id)) { | |
| 2387 | - return $attach_id; | |
| 2388 | - } | |
| 2389 | - | |
| 2390 | - require_once ABSPATH . 'wp-admin/includes/image.php'; | |
| 2391 | - $metadata = wp_generate_attachment_metadata($attach_id, $upload['file']); | |
| 2392 | - wp_update_attachment_metadata($attach_id, $metadata); | |
| 2393 | - | |
| 2394 | - return esc_url_raw(wp_get_attachment_url($attach_id)); | |
| 2395 | -} | |
| 2396 | - | |
| 2397 | -private function mxchat_generate_openai_image($prompt, $api_key, $model = 'gpt-image-1', $timeout = 60) { | |
| 1783 | +private function mxchat_generate_dalle_image($prompt, $api_key, $model = 'dall-e-3', $timeout = 60) { | |
| 2398 | 1784 | $api_url = 'https://api.openai.com/v1/images/generations'; |
| 2399 | 1785 | $body = json_encode([ |
| 2400 | - 'prompt' => sanitize_text_field($prompt), | |
| 2401 | - 'n' => 1, | |
| 2402 | - 'size' => '1024x1024', | |
| 2403 | - 'quality' => 'medium', | |
| 2404 | - 'output_format' => 'png', | |
| 2405 | - 'model' => sanitize_text_field($model), | |
| 1786 | + 'prompt' => sanitize_text_field($prompt), | |
| 1787 | + 'n' => 1, | |
| 1788 | + 'size' => '1024x1024', | |
| 1789 | + 'model' => sanitize_text_field($model), | |
| 2406 | 1790 | ]); |
| 2407 | 1791 | |
| 2408 | 1792 | $args = [ |
| 2409 | - 'body' => $body, | |
| 1793 | + 'body' => $body, | |
| 2410 | 1794 | 'headers' => [ |
| 2411 | - 'Content-Type' => 'application/json', | |
| 1795 | + 'Content-Type' => 'application/json', | |
| 2412 | 1796 | 'Authorization' => 'Bearer ' . sanitize_text_field($api_key), |
| 2413 | 1797 | ], |
| 2414 | - 'method' => 'POST', | |
| 1798 | + 'method' => 'POST', | |
| 2415 | 1799 | 'timeout' => absint($timeout), |
| 2416 | 1800 | ]; |
| 2417 | 1801 | |
| 2418 | 1802 | $response = wp_remote_post($api_url, $args); |
| @@ -2417,67 +1801,22 @@ | ||
| 2417 | 1801 | |
| 2418 | 1802 | $response = wp_remote_post($api_url, $args); |
| 2419 | 1803 | |
| 2420 | 1804 | if (is_wp_error($response)) { |
| 1805 | + //error_log("DALL-E request failed: " . $response->get_error_message()); | |
| 2421 | 1806 | return ['error' => esc_html__('Error generating image: ', 'mxchat') . $response->get_error_message()]; |
| 2422 | 1807 | } |
| 2423 | 1808 | |
| 2424 | 1809 | $response_body = json_decode(wp_remote_retrieve_body($response), true); |
| 2425 | 1810 | |
| 2426 | - $b64 = $response_body['data'][0]['b64_json'] ?? $response_body['data'][0]['b64'] ?? null; | |
| 2427 | - if ($b64) { | |
| 2428 | - $saved_url = $this->mxchat_save_generated_image($b64, 'image/png', 'mxchat-openai'); | |
| 2429 | - if (is_wp_error($saved_url)) { | |
| 2430 | - return ['error' => $saved_url->get_error_message()]; | |
| 2431 | - } | |
| 2432 | - return ['imageUrl' => $saved_url]; | |
| 1811 | + if (isset($response_body['data'][0]['url'])) { | |
| 1812 | + return ['imageUrl' => esc_url_raw($response_body['data'][0]['url'])]; | |
| 2433 | 1813 | } else { |
| 1814 | + //error_log("DALL-E response error: " . wp_remote_retrieve_body($response)); | |
| 2434 | 1815 | return ['error' => esc_html__('Failed to generate image.', 'mxchat')]; |
| 2435 | 1816 | } |
| 2436 | 1817 | } |
| 2437 | 1818 | |
| 2438 | -private function mxchat_generate_imagen_image($prompt, $api_key, $timeout = 60) { | |
| 2439 | - $api_url = 'https://generativelanguage.googleapis.com/v1beta/models/imagen-4.0-generate-001:predict'; | |
| 2440 | - | |
| 2441 | - $body = json_encode([ | |
| 2442 | - 'instances' => [['prompt' => sanitize_text_field($prompt)]], | |
| 2443 | - 'parameters' => [ | |
| 2444 | - 'sampleCount' => 1, | |
| 2445 | - 'aspectRatio' => '1:1', | |
| 2446 | - ], | |
| 2447 | - ]); | |
| 2448 | - | |
| 2449 | - $args = [ | |
| 2450 | - 'body' => $body, | |
| 2451 | - 'headers' => [ | |
| 2452 | - 'Content-Type' => 'application/json', | |
| 2453 | - 'x-goog-api-key' => sanitize_text_field($api_key), | |
| 2454 | - ], | |
| 2455 | - 'method' => 'POST', | |
| 2456 | - 'timeout' => absint($timeout), | |
| 2457 | - ]; | |
| 2458 | - | |
| 2459 | - $response = wp_remote_post($api_url, $args); | |
| 2460 | - | |
| 2461 | - if (is_wp_error($response)) { | |
| 2462 | - return ['error' => esc_html__('Error generating image: ', 'mxchat') . $response->get_error_message()]; | |
| 2463 | - } | |
| 2464 | - | |
| 2465 | - $response_body = json_decode(wp_remote_retrieve_body($response), true); | |
| 2466 | - | |
| 2467 | - $b64 = $response_body['predictions'][0]['bytesBase64Encoded'] ?? $response_body['predictions'][0]['imageBytes'] ?? null; | |
| 2468 | - if ($b64) { | |
| 2469 | - $mime = $response_body['predictions'][0]['mimeType'] ?? 'image/png'; | |
| 2470 | - $saved_url = $this->mxchat_save_generated_image($b64, $mime, 'mxchat-gemini'); | |
| 2471 | - if (is_wp_error($saved_url)) { | |
| 2472 | - return ['error' => $saved_url->get_error_message()]; | |
| 2473 | - } | |
| 2474 | - return ['imageUrl' => $saved_url]; | |
| 2475 | - } else { | |
| 2476 | - return ['error' => esc_html__('Failed to generate image.', 'mxchat')]; | |
| 2477 | - } | |
| 2478 | -} | |
| 2479 | - | |
| 2480 | 1819 | /** |
| 2481 | 1820 | * Handle web search requests. |
| 2482 | 1821 | * |
| 2483 | 1822 | * Sends the refined search query to the Brave Search API and uses the |
| @@ -2526,10 +1865,10 @@ | ||
| 2526 | 1865 | $transient_key = 'mxchat_search_' . md5($refined_search_query); |
| 2527 | 1866 | $results = get_transient($transient_key); |
| 2528 | 1867 | |
| 2529 | 1868 | if (false === $results) { |
| 2530 | - // SECURITY FIX: Changed to wp_safe_remote_get | |
| 2531 | - $response = wp_safe_remote_get( | |
| 1869 | + // Fetch new results from the Brave Search API | |
| 1870 | + $response = wp_remote_get( | |
| 2532 | 1871 | $api_url, |
| 2533 | 1872 | array( |
| 2534 | 1873 | 'headers' => array( |
| 2535 | 1874 | 'Accept' => 'application/json', |
| @@ -2668,10 +2007,9 @@ | ||
| 2668 | 2007 | ], |
| 2669 | 2008 | 'timeout' => 10, |
| 2670 | 2009 | ]; |
| 2671 | 2010 | |
| 2672 | - // SECURITY FIX: Changed to wp_safe_remote_get | |
| 2673 | - $response = wp_safe_remote_get($api_url, $args); | |
| 2011 | + $response = wp_remote_get($api_url, $args); | |
| 2674 | 2012 | |
| 2675 | 2013 | if (is_wp_error($response)) { |
| 2676 | 2014 | return array( |
| 2677 | 2015 | 'text' => __("I'm sorry, I couldn't retrieve any images based on your request.", 'mxchat'), |
| @@ -2744,9 +2082,9 @@ | ||
| 2744 | 2082 | $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'); |
| 2745 | 2083 | |
| 2746 | 2084 | // Get options and determine the selected model |
| 2747 | 2085 | $options = $this->options ?? get_option('mxchat_options'); |
| 2748 | - $selected_model = isset($options['model']) ? $options['model'] : 'gpt-5.1-chat-latest'; | |
| 2086 | + $selected_model = isset($options['model']) ? $options['model'] : 'gpt-4o'; | |
| 2749 | 2087 | |
| 2750 | 2088 | // Extract model prefix to determine the provider |
| 2751 | 2089 | $model_parts = explode('-', $selected_model); |
| 2752 | 2090 | $provider = strtolower($model_parts[0]); |
| @@ -2794,9 +2132,9 @@ | ||
| 2794 | 2132 | |
| 2795 | 2133 | /** |
| 2796 | 2134 | * Interpret query using OpenAI models |
| 2797 | 2135 | */ |
| 2798 | -private function interpret_query_with_openai($user_query, $system_prompt, $api_key, $model = 'gpt-5.1-chat-latest') { | |
| 2136 | +private function interpret_query_with_openai($user_query, $system_prompt, $api_key, $model = 'gpt-4o') { | |
| 2799 | 2137 | $url = 'https://api.openai.com/v1/chat/completions'; |
| 2800 | 2138 | $args = [ |
| 2801 | 2139 | 'headers' => [ |
| 2802 | 2140 | 'Authorization' => 'Bearer ' . $api_key, |
| @@ -2867,13 +2205,13 @@ | ||
| 2867 | 2205 | /** |
| 2868 | 2206 | * Interpret query using Gemini models |
| 2869 | 2207 | */ |
| 2870 | 2208 | private function interpret_query_with_gemini($user_query, $system_prompt, $api_key, $model) { |
| 2871 | - // Use v1beta for preview models, v1 for stable models | |
| 2872 | - $api_version = (strpos($model, 'preview') !== false || strpos($model, 'exp') !== false) ? 'v1beta' : 'v1'; | |
| 2873 | - | |
| 2874 | - $url = "https://generativelanguage.googleapis.com/{$api_version}/models/{$model}:generateContent?key=" . urlencode($api_key); | |
| 2209 | + // Strip "gemini-" prefix for the API | |
| 2210 | + $model_version = str_replace('gemini-', '', $model); | |
| 2875 | 2211 | |
| 2212 | + $url = "https://generativelanguage.googleapis.com/v1/models/$model_version:generateContent?key=" . urlencode($api_key); | |
| 2213 | + | |
| 2876 | 2214 | $args = [ |
| 2877 | 2215 | 'headers' => [ |
| 2878 | 2216 | 'Content-Type' => 'application/json', |
| 2879 | 2217 | ], |
| @@ -3069,9 +2407,9 @@ | ||
| 3069 | 2407 | } |
| 3070 | 2408 | |
| 3071 | 2409 | |
| 3072 | 2410 | /** |
| 3073 | - * Enhanced fetch_and_split_pdf_pages with SSRF protection | |
| 2411 | + * Enhanced fetch_and_split_pdf_pages with detailed debugging | |
| 3074 | 2412 | */ |
| 3075 | 2413 | private function fetch_and_split_pdf_pages($pdf_source, $max_pages) { |
| 3076 | 2414 | // CLEAR DEBUG LOGGING |
| 3077 | 2415 | //error_log("=== MXCHAT PDF PROCESSING START ==="); |
| @@ -3126,19 +2464,10 @@ | ||
| 3126 | 2464 | // (I'll include the key parts with debug logging) |
| 3127 | 2465 | |
| 3128 | 2466 | if (filter_var($pdf_source, FILTER_VALIDATE_URL)) { |
| 3129 | 2467 | //error_log("Downloading PDF from URL..."); |
| 3130 | - | |
| 3131 | - // SECURITY FIX: Validate URL before processing | |
| 3132 | - if (!$this->mxchat_is_safe_pdf_url($pdf_source)) { | |
| 3133 | - //error_log("❌ SECURITY: Blocked unsafe PDF URL"); | |
| 3134 | - return false; | |
| 3135 | - } | |
| 3136 | - | |
| 3137 | 2468 | $temp_file = wp_tempnam($pdf_source); |
| 3138 | - | |
| 3139 | - // SECURITY FIX: Changed from wp_remote_get to wp_safe_remote_get | |
| 3140 | - $response = wp_safe_remote_get($pdf_source, [ | |
| 2469 | + $response = wp_remote_get($pdf_source, [ | |
| 3141 | 2470 | 'timeout' => 60, |
| 3142 | 2471 | 'headers' => ['User-Agent' => 'MxChat PDF Processor'] |
| 3143 | 2472 | ]); |
| 3144 | 2473 | |
| @@ -3147,14 +2476,9 @@ | ||
| 3147 | 2476 | //error_log("❌ BASIC PROCESSING: Failed to download PDF: " . $error_message); |
| 3148 | 2477 | return false; |
| 3149 | 2478 | } |
| 3150 | 2479 | |
| 3151 | - global $wp_filesystem; | |
| 3152 | - if (empty($wp_filesystem)) { | |
| 3153 | - require_once ABSPATH . 'wp-admin/includes/file.php'; | |
| 3154 | - WP_Filesystem(); | |
| 3155 | - } | |
| 3156 | - $wp_filesystem->put_contents($temp_file, wp_remote_retrieve_body($response), FS_CHMOD_FILE); | |
| 2480 | + file_put_contents($temp_file, wp_remote_retrieve_body($response)); | |
| 3157 | 2481 | //error_log("✅ PDF downloaded successfully"); |
| 3158 | 2482 | } else { |
| 3159 | 2483 | $temp_file = $pdf_source; |
| 3160 | 2484 | //error_log("Using local PDF file: " . $temp_file); |
| @@ -3161,9 +2485,8 @@ | ||
| 3161 | 2485 | } |
| 3162 | 2486 | |
| 3163 | 2487 | // Parse PDF |
| 3164 | 2488 | //error_log("Parsing PDF with basic parser..."); |
| 3165 | - mxchat_load_pdf_parser(); | |
| 3166 | 2489 | $parser = new \Smalot\PdfParser\Parser(); |
| 3167 | 2490 | $pdf = $parser->parseFile($temp_file); |
| 3168 | 2491 | $pages = $pdf->getPages(); |
| 3169 | 2492 | |
| @@ -3226,33 +2549,8 @@ | ||
| 3226 | 2549 | return false; |
| 3227 | 2550 | } |
| 3228 | 2551 | } |
| 3229 | 2552 | |
| 3230 | - | |
| 3231 | -/** | |
| 3232 | - * Validate PDF URL for security | |
| 3233 | - * Prevents SSRF attacks by blocking dangerous URLs | |
| 3234 | - */ | |
| 3235 | - | |
| 3236 | -private function mxchat_is_safe_pdf_url($url) { | |
| 3237 | - // Use WordPress core function for comprehensive validation | |
| 3238 | - // This blocks localhost, private IPs, and reserved IP ranges | |
| 3239 | - $validated_url = wp_http_validate_url($url); | |
| 3240 | - | |
| 3241 | - if ($validated_url === false) { | |
| 3242 | - return false; | |
| 3243 | - } | |
| 3244 | - | |
| 3245 | - // Additional check: only allow HTTP/HTTPS schemes | |
| 3246 | - $parsed = parse_url($url); | |
| 3247 | - if (!isset($parsed['scheme']) || !in_array($parsed['scheme'], ['http', 'https'], true)) { | |
| 3248 | - return false; | |
| 3249 | - } | |
| 3250 | - | |
| 3251 | - return true; | |
| 3252 | -} | |
| 3253 | - | |
| 3254 | - | |
| 3255 | 2553 | private function mxchat_clean_text($text) { |
| 3256 | 2554 | // Remove excessive whitespace |
| 3257 | 2555 | $text = preg_replace('/\s+/', ' ', $text); |
| 3258 | 2556 | |
| @@ -3291,10 +2589,9 @@ | ||
| 3291 | 2589 | } |
| 3292 | 2590 | |
| 3293 | 2591 | return []; |
| 3294 | 2592 | } |
| 3295 | - | |
| 3296 | - | |
| 2593 | +// Add this to your class | |
| 3297 | 2594 | public function handle_pdf_upload() { |
| 3298 | 2595 | check_ajax_referer('mxchat_chat_nonce', 'nonce'); |
| 3299 | 2596 | |
| 3300 | 2597 | if (!isset($_FILES['pdf_file']) || !isset($_POST['session_id'])) { |
| @@ -3301,29 +2598,12 @@ | ||
| 3301 | 2598 | wp_send_json_error(esc_html__('Missing required parameters.', 'mxchat')); |
| 3302 | 2599 | return; |
| 3303 | 2600 | } |
| 3304 | 2601 | |
| 3305 | - // SECURITY FIX: Check if PDF uploads are enabled in settings | |
| 3306 | - $options = get_option('mxchat_options', array()); | |
| 3307 | - $show_pdf_button = isset($options['show_pdf_upload_button']) ? $options['show_pdf_upload_button'] : 'on'; | |
| 3308 | - | |
| 3309 | - if ($show_pdf_button !== 'on') { | |
| 3310 | - wp_send_json_error(esc_html__('PDF uploads are currently disabled.', 'mxchat')); | |
| 3311 | - return; | |
| 3312 | - } | |
| 3313 | - | |
| 3314 | 2602 | $file = $_FILES['pdf_file']; |
| 3315 | 2603 | $session_id = sanitize_text_field($_POST['session_id']); |
| 3316 | 2604 | $original_filename = sanitize_text_field($file['name']); |
| 3317 | 2605 | |
| 3318 | - // Update session owner if it changed (e.g. IP changed due to network switch) | |
| 3319 | - $current_user_identifier = MxChat_User::mxchat_get_user_identifier(); | |
| 3320 | - $session_owner = get_option("mxchat_session_owner_{$session_id}"); | |
| 3321 | - | |
| 3322 | - if (!$session_owner || $session_owner !== $current_user_identifier) { | |
| 3323 | - update_option("mxchat_session_owner_{$session_id}", $current_user_identifier, 'no'); | |
| 3324 | - } | |
| 3325 | - | |
| 3326 | 2606 | $file_type = wp_check_filetype($file['name'], ['pdf' => 'application/pdf']); |
| 3327 | 2607 | if ($file_type['type'] !== 'application/pdf') { |
| 3328 | 2608 | wp_send_json_error(esc_html__('Invalid file type. Only PDF files are allowed.', 'mxchat')); |
| 3329 | 2609 | return; |
| @@ -3329,12 +2609,9 @@ | ||
| 3329 | 2609 | return; |
| 3330 | 2610 | } |
| 3331 | 2611 | |
| 3332 | 2612 | $upload_dir = wp_upload_dir(); |
| 3333 | - | |
| 3334 | - // SECURITY FIX: Generate random filename without exposing session_id | |
| 3335 | - $random_string = wp_generate_password(20, false, false); // 20 char alphanumeric string | |
| 3336 | - $pdf_filename = 'mxchat_' . $random_string . '_' . time() . '.pdf'; | |
| 2613 | + $pdf_filename = 'mxchat_' . $session_id . '_' . time() . '.pdf'; | |
| 3337 | 2614 | $pdf_path = $upload_dir['path'] . '/' . $pdf_filename; |
| 3338 | 2615 | |
| 3339 | 2616 | if (!move_uploaded_file($file['tmp_name'], $pdf_path)) { |
| 3340 | 2617 | wp_send_json_error(esc_html__('Failed to upload file.', 'mxchat')); |
| @@ -3365,9 +2642,8 @@ | ||
| 3365 | 2642 | return; |
| 3366 | 2643 | } |
| 3367 | 2644 | |
| 3368 | 2645 | if (!empty($embeddings)) { |
| 3369 | - // Store the mapping between session and the random filename | |
| 3370 | 2646 | set_transient('mxchat_pdf_url_' . $session_id, $pdf_path, HOUR_IN_SECONDS); |
| 3371 | 2647 | set_transient('mxchat_pdf_filename_' . $session_id, $original_filename, HOUR_IN_SECONDS); |
| 3372 | 2648 | set_transient('mxchat_pdf_embeddings_' . $session_id, $embeddings, HOUR_IN_SECONDS); |
| 3373 | 2649 | set_transient('mxchat_include_pdf_in_context_' . $session_id, true, HOUR_IN_SECONDS); |
| @@ -3411,8 +2687,10 @@ | ||
| 3411 | 2687 | wp_die(); |
| 3412 | 2688 | } |
| 3413 | 2689 | |
| 3414 | 2690 | |
| 2691 | + | |
| 2692 | + | |
| 3415 | 2693 | function mxchat_fetch_new_messages() { |
| 3416 | 2694 | $session_id = sanitize_text_field($_POST['session_id']); |
| 3417 | 2695 | $last_seen_id = sanitize_text_field($_POST['last_seen_id']); |
| 3418 | 2696 | $persistence_enabled = $_POST['persistence_enabled'] === 'true'; |
| @@ -3425,31 +2703,14 @@ | ||
| 3425 | 2703 | } |
| 3426 | 2704 | |
| 3427 | 2705 | $history = get_option("mxchat_history_{$session_id}", []); |
| 3428 | 2706 | |
| 3429 | - //error_log("MxChat WhatsApp DEBUG: Fetch new messages for session {$session_id}"); | |
| 3430 | - //error_log("MxChat WhatsApp DEBUG: last_seen_id = " . var_export($last_seen_id, true)); | |
| 3431 | - //error_log("MxChat WhatsApp DEBUG: History count = " . count($history)); | |
| 3432 | - //error_log("MxChat WhatsApp DEBUG: Full history = " . print_r($history, true)); | |
| 3433 | - | |
| 3434 | 2707 | $new_messages = array_filter($history, function ($message) use ($last_seen_id, $persistence_enabled, $initial_timestamp) { |
| 3435 | - //error_log("MxChat WhatsApp DEBUG: Checking message - ID: " . ($message['id'] ?? 'NO_ID') . ", Role: " . ($message['role'] ?? 'NO_ROLE')); | |
| 3436 | - | |
| 3437 | 2708 | // If persistence is enabled, show all new messages |
| 3438 | 2709 | if ($persistence_enabled) { |
| 3439 | - $has_id = !empty($message['id']); | |
| 3440 | - $is_agent = $message['role'] === 'agent'; | |
| 3441 | - | |
| 3442 | - // If last_seen_id is empty, 'NaN', or invalid, show all agent messages | |
| 3443 | - if (empty($last_seen_id) || $last_seen_id === 'NaN' || $last_seen_id === 'undefined') { | |
| 3444 | - $is_newer = true; | |
| 3445 | - } else { | |
| 3446 | - $is_newer = strcmp($message['id'] ?? '', $last_seen_id) > 0; | |
| 3447 | - } | |
| 3448 | - | |
| 3449 | - //error_log("MxChat WhatsApp DEBUG: has_id={$has_id}, is_newer={$is_newer}, is_agent={$is_agent}"); | |
| 3450 | - | |
| 3451 | - return $has_id && $is_newer && $is_agent; | |
| 2710 | + return !empty($message['id']) && | |
| 2711 | + strcmp($message['id'], $last_seen_id) > 0 && | |
| 2712 | + $message['role'] === 'agent'; | |
| 3452 | 2713 | } |
| 3453 | 2714 | |
| 3454 | 2715 | // If persistence is disabled, only show messages after initial timestamp |
| 3455 | 2716 | return !empty($message['id']) && |
| @@ -3456,16 +2717,12 @@ | ||
| 3456 | 2717 | $message['role'] === 'agent' && |
| 3457 | 2718 | $message['timestamp'] > $initial_timestamp; |
| 3458 | 2719 | }); |
| 3459 | 2720 | |
| 3460 | - //error_log("MxChat WhatsApp DEBUG: Filtered messages count = " . count($new_messages)); | |
| 2721 | + //error_log(esc_html__("New agent messages fetched for session $session_id. Last seen ID: $last_seen_id", 'mxchat')); | |
| 3461 | 2722 | |
| 3462 | - // Include current chat mode so frontend can detect agent→AI transitions | |
| 3463 | - $chat_mode = get_option("mxchat_mode_{$session_id}", 'ai'); | |
| 3464 | - | |
| 3465 | 2723 | wp_send_json_success([ |
| 3466 | - 'new_messages' => array_values($new_messages), | |
| 3467 | - 'chat_mode' => $chat_mode | |
| 2724 | + 'new_messages' => array_values($new_messages) | |
| 3468 | 2725 | ]); |
| 3469 | 2726 | wp_die(); |
| 3470 | 2727 | } |
| 3471 | 2728 | public function mxchat_live_agent_handover($message, $user_id, $session_id) { |
| @@ -3750,393 +3007,9 @@ | ||
| 3750 | 3007 | |
| 3751 | 3008 | //error_log("[DEBUG] Generated channel name: {$channel_name}"); |
| 3752 | 3009 | return $channel_name; |
| 3753 | 3010 | } |
| 3754 | - | |
| 3755 | -/** | |
| 3756 | - * Telegram Live Agent Handover | |
| 3757 | - * Creates a forum topic in the Telegram group and notifies agents | |
| 3758 | - */ | |
| 3759 | -public function mxchat_telegram_live_agent_handover($message, $user_id, $session_id) { | |
| 3760 | - // Check if Telegram agents are available | |
| 3761 | - $telegram_available = $this->options['telegram_status'] ?? 'off'; | |
| 3762 | - if ($telegram_available !== 'on') { | |
| 3763 | - $away_message = $this->options['telegram_away_message'] ?? 'Sorry, live agents are currently unavailable. I can continue helping you as an AI assistant.'; | |
| 3764 | - $this->fallbackResponse = [ | |
| 3765 | - 'text' => $away_message, | |
| 3766 | - 'html' => '', | |
| 3767 | - 'images' => [], | |
| 3768 | - 'chat_mode' => 'ai' | |
| 3769 | - ]; | |
| 3770 | - wp_send_json([ | |
| 3771 | - 'text' => $away_message, | |
| 3772 | - 'html' => '', | |
| 3773 | - 'chat_mode' => 'ai', | |
| 3774 | - 'session_id' => $session_id | |
| 3775 | - ]); | |
| 3776 | - wp_die(); | |
| 3777 | - } | |
| 3778 | - | |
| 3779 | - $telegram_bot_token = $this->options['telegram_bot_token'] ?? ''; | |
| 3780 | - $telegram_group_id = $this->options['telegram_group_id'] ?? ''; | |
| 3781 | - | |
| 3782 | - if (empty($telegram_bot_token) || empty($telegram_group_id)) { | |
| 3783 | - return false; | |
| 3784 | - } | |
| 3785 | - | |
| 3786 | - // Check if topic already exists for this session | |
| 3787 | - $topic_id = get_option("mxchat_telegram_topic_{$session_id}", ''); | |
| 3788 | - | |
| 3789 | - if (empty($topic_id)) { | |
| 3790 | - // Generate topic name | |
| 3791 | - $topic_name = $this->generate_telegram_topic_name($session_id); | |
| 3792 | - | |
| 3793 | - // Random icon color (Telegram forum topic colors) | |
| 3794 | - $icon_colors = [0x6FB9F0, 0xFFD67E, 0xCB86DB, 0x8EEE98, 0xFF93B2, 0xFB6F5F]; | |
| 3795 | - $icon_color = $icon_colors[array_rand($icon_colors)]; | |
| 3796 | - | |
| 3797 | - // Create forum topic | |
| 3798 | - $response = wp_remote_post("https://api.telegram.org/bot{$telegram_bot_token}/createForumTopic", [ | |
| 3799 | - 'headers' => ['Content-Type' => 'application/json'], | |
| 3800 | - 'body' => json_encode([ | |
| 3801 | - 'chat_id' => $telegram_group_id, | |
| 3802 | - 'name' => $topic_name, | |
| 3803 | - 'icon_color' => $icon_color | |
| 3804 | - ]) | |
| 3805 | - ]); | |
| 3806 | - | |
| 3807 | - if (!is_wp_error($response)) { | |
| 3808 | - $response_body = wp_remote_retrieve_body($response); | |
| 3809 | - $response_data = json_decode($response_body, true); | |
| 3810 | - | |
| 3811 | - if (isset($response_data['ok']) && $response_data['ok']) { | |
| 3812 | - $topic_id = $response_data['result']['message_thread_id']; | |
| 3813 | - update_option("mxchat_telegram_topic_{$session_id}", $topic_id); | |
| 3814 | - update_option("mxchat_telegram_group_{$session_id}", $telegram_group_id); | |
| 3815 | - } | |
| 3816 | - } | |
| 3817 | - | |
| 3818 | - if (empty($topic_id)) { | |
| 3819 | - return false; // Failed to create topic | |
| 3820 | - } | |
| 3821 | - } | |
| 3822 | - | |
| 3823 | - // Get recent chat history | |
| 3824 | - $history = get_option("mxchat_history_{$session_id}", []); | |
| 3825 | - $recent_history = array_slice($history, -5); | |
| 3826 | - | |
| 3827 | - // Format conversation context for Telegram (HTML format) | |
| 3828 | - $conversation_context = ""; | |
| 3829 | - if (!empty($recent_history)) { | |
| 3830 | - $conversation_context = "<b>Recent Conversation:</b>\n"; | |
| 3831 | - foreach ($recent_history as $hist_message) { | |
| 3832 | - $role_display = $hist_message['role'] === 'user' ? '👤 User' : '🤖 AI'; | |
| 3833 | - $escaped_content = htmlspecialchars($hist_message['content'], ENT_QUOTES, 'UTF-8'); | |
| 3834 | - $conversation_context .= "{$role_display}: {$escaped_content}\n"; | |
| 3835 | - } | |
| 3836 | - $conversation_context .= "\n"; | |
| 3837 | - } | |
| 3838 | - | |
| 3839 | - // Get user info | |
| 3840 | - $user_email = get_option("mxchat_email_{$session_id}", 'Not provided'); | |
| 3841 | - $user_name = get_option("mxchat_name_{$session_id}", 'Anonymous'); | |
| 3842 | - | |
| 3843 | - // Update session mode | |
| 3844 | - update_option("mxchat_mode_{$session_id}", 'agent'); | |
| 3845 | - | |
| 3846 | - // Send initial message to topic | |
| 3847 | - $escaped_message = htmlspecialchars($message, ENT_QUOTES, 'UTF-8'); | |
| 3848 | - $topic_message = "🔔 <b>New Live Agent Request</b>\n\n"; | |
| 3849 | - $topic_message .= "<b>Session ID:</b> <code>{$session_id}</code>\n"; | |
| 3850 | - $topic_message .= "<b>User:</b> {$user_name}\n"; | |
| 3851 | - $topic_message .= "<b>Email:</b> {$user_email}\n\n"; | |
| 3852 | - | |
| 3853 | - if (!empty($conversation_context)) { | |
| 3854 | - $topic_message .= $conversation_context; | |
| 3855 | - } | |
| 3856 | - | |
| 3857 | - $topic_message .= "<b>Current Message:</b>\n{$escaped_message}\n\n"; | |
| 3858 | - $topic_message .= "<i>Reply in this topic - messages will be sent to the user</i>\n"; | |
| 3859 | - $topic_message .= "<i>Type #close, #end, #disconnect, or #done to end the session</i>"; | |
| 3860 | - | |
| 3861 | - wp_remote_post("https://api.telegram.org/bot{$telegram_bot_token}/sendMessage", [ | |
| 3862 | - 'headers' => ['Content-Type' => 'application/json'], | |
| 3863 | - 'body' => json_encode([ | |
| 3864 | - 'chat_id' => $telegram_group_id, | |
| 3865 | - 'message_thread_id' => $topic_id, | |
| 3866 | - 'text' => $topic_message, | |
| 3867 | - 'parse_mode' => 'HTML' | |
| 3868 | - ]) | |
| 3869 | - ]); | |
| 3870 | - | |
| 3871 | - $success_message = $this->options['telegram_notification_message'] ?? "I've notified a support agent. Please allow a moment for them to respond."; | |
| 3872 | - $this->mxchat_save_chat_message($session_id, 'bot', $success_message); | |
| 3873 | - | |
| 3874 | - $this->fallbackResponse = [ | |
| 3875 | - 'text' => $success_message, | |
| 3876 | - 'html' => '', | |
| 3877 | - 'images' => [], | |
| 3878 | - 'chat_mode' => 'agent' | |
| 3879 | - ]; | |
| 3880 | - | |
| 3881 | - wp_send_json([ | |
| 3882 | - 'success' => true, | |
| 3883 | - 'text' => $success_message, | |
| 3884 | - 'html' => '', | |
| 3885 | - 'chat_mode' => 'agent', | |
| 3886 | - 'session_id' => $session_id, | |
| 3887 | - 'fallbackResponse' => $this->fallbackResponse | |
| 3888 | - ]); | |
| 3889 | - wp_die(); | |
| 3890 | -} | |
| 3891 | - | |
| 3892 | -/** | |
| 3893 | - * Generate topic name for Telegram forum | |
| 3894 | - */ | |
| 3895 | -private function generate_telegram_topic_name($session_id) { | |
| 3896 | - $name = null; | |
| 3897 | - $email = null; | |
| 3898 | - | |
| 3899 | - // Check logged in user | |
| 3900 | - if (is_user_logged_in()) { | |
| 3901 | - $current_user = wp_get_current_user(); | |
| 3902 | - if (!empty($current_user->display_name)) { | |
| 3903 | - $name = $current_user->display_name; | |
| 3904 | - } | |
| 3905 | - if (!empty($current_user->user_email)) { | |
| 3906 | - $email = $current_user->user_email; | |
| 3907 | - } | |
| 3908 | - } | |
| 3909 | - | |
| 3910 | - // Check session data | |
| 3911 | - if (empty($name)) { | |
| 3912 | - $name = get_option("mxchat_name_{$session_id}"); | |
| 3913 | - } | |
| 3914 | - if (empty($email)) { | |
| 3915 | - $email = get_option("mxchat_email_{$session_id}"); | |
| 3916 | - } | |
| 3917 | - | |
| 3918 | - // Generate topic name | |
| 3919 | - $session_suffix = substr($session_id, -6); | |
| 3920 | - | |
| 3921 | - if (!empty($name)) { | |
| 3922 | - // Clean name for topic (max 128 chars in Telegram) | |
| 3923 | - $clean_name = preg_replace('/[^\p{L}\p{N}\s\-]/u', '', $name); | |
| 3924 | - $clean_name = trim($clean_name); | |
| 3925 | - if (strlen($clean_name) > 50) { | |
| 3926 | - $clean_name = substr($clean_name, 0, 50); | |
| 3927 | - } | |
| 3928 | - return "Chat - {$clean_name} ({$session_suffix})"; | |
| 3929 | - } elseif (!empty($email)) { | |
| 3930 | - // Use email prefix | |
| 3931 | - $email_prefix = explode('@', $email)[0]; | |
| 3932 | - if (strlen($email_prefix) > 30) { | |
| 3933 | - $email_prefix = substr($email_prefix, 0, 30); | |
| 3934 | - } | |
| 3935 | - return "Chat - {$email_prefix} ({$session_suffix})"; | |
| 3936 | - } | |
| 3937 | - | |
| 3938 | - return "Chat - {$session_suffix}"; | |
| 3939 | -} | |
| 3940 | - | |
| 3941 | -/** | |
| 3942 | - * Send user message to Telegram agent | |
| 3943 | - */ | |
| 3944 | -public function mxchat_send_user_message_to_telegram_agent($message, $user_id, $session_id) { | |
| 3945 | - $telegram_bot_token = $this->options['telegram_bot_token'] ?? ''; | |
| 3946 | - $topic_id = get_option("mxchat_telegram_topic_{$session_id}", ''); | |
| 3947 | - $group_id = get_option("mxchat_telegram_group_{$session_id}", ''); | |
| 3948 | - | |
| 3949 | - if (empty($telegram_bot_token) || empty($topic_id) || empty($group_id)) { | |
| 3950 | - return false; | |
| 3951 | - } | |
| 3952 | - | |
| 3953 | - $escaped_message = htmlspecialchars($message, ENT_QUOTES, 'UTF-8'); | |
| 3954 | - $user_message = "👤 <b>User:</b> {$escaped_message}"; | |
| 3955 | - | |
| 3956 | - $response = wp_remote_post("https://api.telegram.org/bot{$telegram_bot_token}/sendMessage", [ | |
| 3957 | - 'headers' => ['Content-Type' => 'application/json'], | |
| 3958 | - 'body' => json_encode([ | |
| 3959 | - 'chat_id' => $group_id, | |
| 3960 | - 'message_thread_id' => $topic_id, | |
| 3961 | - 'text' => $user_message, | |
| 3962 | - 'parse_mode' => 'HTML' | |
| 3963 | - ]) | |
| 3964 | - ]); | |
| 3965 | - | |
| 3966 | - return !is_wp_error($response); | |
| 3967 | -} | |
| 3968 | - | |
| 3969 | -/** | |
| 3970 | - * Handle incoming Telegram webhook | |
| 3971 | - */ | |
| 3972 | -public function handle_telegram_webhook(WP_REST_Request $request) { | |
| 3973 | - $body = $request->get_body(); | |
| 3974 | - $data = json_decode($body, true); | |
| 3975 | - | |
| 3976 | - //error_log('[MxChat Telegram DEBUG] Webhook received: ' . $body); | |
| 3977 | - | |
| 3978 | - // Handle message events from forum topics | |
| 3979 | - if (isset($data['message'])) { | |
| 3980 | - $message_data = $data['message']; | |
| 3981 | - | |
| 3982 | - // Skip if not from a forum topic | |
| 3983 | - if (!isset($message_data['message_thread_id'])) { | |
| 3984 | - //error_log('[MxChat Telegram DEBUG] Skipped: No message_thread_id (not a forum topic message)'); | |
| 3985 | - return new WP_REST_Response(['ok' => true]); | |
| 3986 | - } | |
| 3987 | - | |
| 3988 | - // Skip bot messages | |
| 3989 | - if (isset($message_data['from']['is_bot']) && $message_data['from']['is_bot']) { | |
| 3990 | - //error_log('[MxChat Telegram DEBUG] Skipped: Message from bot'); | |
| 3991 | - return new WP_REST_Response(['ok' => true]); | |
| 3992 | - } | |
| 3993 | - | |
| 3994 | - $chat_id = $message_data['chat']['id'] ?? ''; | |
| 3995 | - $topic_id = $message_data['message_thread_id']; | |
| 3996 | - $message_text = $message_data['text'] ?? ''; | |
| 3997 | - $message_id = $message_data['message_id'] ?? ''; | |
| 3998 | - $from = $message_data['from'] ?? []; | |
| 3999 | - $agent_name = trim(($from['first_name'] ?? '') . ' ' . ($from['last_name'] ?? '')); | |
| 4000 | - if (empty($agent_name)) { | |
| 4001 | - $agent_name = $from['username'] ?? 'Agent'; | |
| 4002 | - } | |
| 4003 | - | |
| 4004 | - //error_log("[MxChat Telegram DEBUG] Parsed: chat_id={$chat_id}, topic_id={$topic_id}, agent={$agent_name}, text={$message_text}"); | |
| 4005 | - | |
| 4006 | - // Skip empty messages | |
| 4007 | - if (empty($message_text)) { | |
| 4008 | - //error_log('[MxChat Telegram DEBUG] Skipped: Empty message text'); | |
| 4009 | - return new WP_REST_Response(['ok' => true]); | |
| 4010 | - } | |
| 4011 | - | |
| 4012 | - // Find session ID by topic ID - cast to string for comparison | |
| 4013 | - global $wpdb; | |
| 4014 | - $topic_id_str = strval($topic_id); | |
| 4015 | - $session_option = $wpdb->get_var( | |
| 4016 | - $wpdb->prepare( | |
| 4017 | - "SELECT option_name FROM {$wpdb->options} | |
| 4018 | - WHERE option_name LIKE %s | |
| 4019 | - AND option_value = %s", | |
| 4020 | - 'mxchat_telegram_topic_%', | |
| 4021 | - $topic_id_str | |
| 4022 | - ) | |
| 4023 | - ); | |
| 4024 | - | |
| 4025 | - //error_log("[MxChat Telegram DEBUG] Looking for topic_id={$topic_id_str} in options, found: " . ($session_option ?: 'NULL')); | |
| 4026 | - | |
| 4027 | - if ($session_option) { | |
| 4028 | - $session_id = str_replace('mxchat_telegram_topic_', '', $session_option); | |
| 4029 | - //error_log("[MxChat Telegram DEBUG] Session ID: {$session_id}"); | |
| 4030 | - | |
| 4031 | - // Verify the group ID matches | |
| 4032 | - $stored_group_id = get_option("mxchat_telegram_group_{$session_id}", ''); | |
| 4033 | - //error_log("[MxChat Telegram DEBUG] Stored group_id={$stored_group_id}, received chat_id={$chat_id}"); | |
| 4034 | - | |
| 4035 | - if (strval($stored_group_id) != strval($chat_id)) { | |
| 4036 | - //error_log('[MxChat Telegram DEBUG] Skipped: Group ID mismatch'); | |
| 4037 | - return new WP_REST_Response(['ok' => true]); | |
| 4038 | - } | |
| 4039 | - | |
| 4040 | - // Check for closure commands | |
| 4041 | - $lower_text = strtolower(trim($message_text)); | |
| 4042 | - if (in_array($lower_text, ['#close', '#end', '#disconnect', '#done'])) { | |
| 4043 | - //error_log("[MxChat Telegram DEBUG] Closure command received: {$lower_text}"); | |
| 4044 | - // End the live agent session | |
| 4045 | - update_option("mxchat_mode_{$session_id}", 'ai'); | |
| 4046 | - | |
| 4047 | - // Save disconnect message | |
| 4048 | - $disconnect_message = "Live agent session ended. You're now chatting with the AI assistant."; | |
| 4049 | - $this->mxchat_save_chat_message($session_id, 'bot', $disconnect_message); | |
| 4050 | - | |
| 4051 | - // Notify in Telegram | |
| 4052 | - $telegram_bot_token = $this->options['telegram_bot_token'] ?? ''; | |
| 4053 | - if (!empty($telegram_bot_token)) { | |
| 4054 | - wp_remote_post("https://api.telegram.org/bot{$telegram_bot_token}/sendMessage", [ | |
| 4055 | - 'headers' => ['Content-Type' => 'application/json'], | |
| 4056 | - 'body' => json_encode([ | |
| 4057 | - 'chat_id' => $chat_id, | |
| 4058 | - 'message_thread_id' => $topic_id, | |
| 4059 | - 'text' => "✅ Session closed. User returned to AI chatbot.", | |
| 4060 | - 'parse_mode' => 'HTML' | |
| 4061 | - ]) | |
| 4062 | - ]); | |
| 4063 | - | |
| 4064 | - // Optionally close the topic | |
| 4065 | - wp_remote_post("https://api.telegram.org/bot{$telegram_bot_token}/closeForumTopic", [ | |
| 4066 | - 'headers' => ['Content-Type' => 'application/json'], | |
| 4067 | - 'body' => json_encode([ | |
| 4068 | - 'chat_id' => $chat_id, | |
| 4069 | - 'message_thread_id' => $topic_id | |
| 4070 | - ]) | |
| 4071 | - ]); | |
| 4072 | - } | |
| 4073 | - | |
| 4074 | - return new WP_REST_Response(['ok' => true]); | |
| 4075 | - } | |
| 4076 | - | |
| 4077 | - // Deduplicate messages | |
| 4078 | - $message_key = md5($session_id . $message_id . $message_text); | |
| 4079 | - $processed_messages = get_transient('mxchat_telegram_messages_' . $session_id) ?: []; | |
| 4080 | - | |
| 4081 | - if (in_array($message_key, $processed_messages)) { | |
| 4082 | - //error_log('[MxChat Telegram DEBUG] Skipped: Duplicate message'); | |
| 4083 | - return new WP_REST_Response(['ok' => true]); | |
| 4084 | - } | |
| 4085 | - | |
| 4086 | - $processed_messages[] = $message_key; | |
| 4087 | - if (count($processed_messages) > 50) { | |
| 4088 | - $processed_messages = array_slice($processed_messages, -50); | |
| 4089 | - } | |
| 4090 | - set_transient('mxchat_telegram_messages_' . $session_id, $processed_messages, HOUR_IN_SECONDS); | |
| 4091 | - | |
| 4092 | - // Save the agent message - format with agent name prefix for proper parsing | |
| 4093 | - $formatted_message = "Agent: {$agent_name} - {$message_text}"; | |
| 4094 | - //error_log("[MxChat Telegram DEBUG] Saving agent message: {$formatted_message}"); | |
| 4095 | - | |
| 4096 | - $this->mxchat_save_chat_message($session_id, 'agent', $formatted_message); | |
| 4097 | - | |
| 4098 | - // Verify the message was saved to history | |
| 4099 | - $history = get_option("mxchat_history_{$session_id}", []); | |
| 4100 | - $last_message = end($history); | |
| 4101 | - //error_log("[MxChat Telegram DEBUG] History after save - count: " . count($history) . ", last message role: " . ($last_message['role'] ?? 'none')); | |
| 4102 | - | |
| 4103 | - // Send confirmation back to Telegram | |
| 4104 | - $telegram_bot_token = $this->options['telegram_bot_token'] ?? ''; | |
| 4105 | - if (!empty($telegram_bot_token)) { | |
| 4106 | - $confirm_key = 'mxchat_telegram_confirm_' . $message_key; | |
| 4107 | - if (!get_transient($confirm_key)) { | |
| 4108 | - wp_remote_post("https://api.telegram.org/bot{$telegram_bot_token}/sendMessage", [ | |
| 4109 | - 'headers' => ['Content-Type' => 'application/json'], | |
| 4110 | - 'body' => json_encode([ | |
| 4111 | - 'chat_id' => $chat_id, | |
| 4112 | - 'message_thread_id' => $topic_id, | |
| 4113 | - 'text' => "✅ <i>Message sent to user</i>", | |
| 4114 | - 'parse_mode' => 'HTML', | |
| 4115 | - 'reply_to_message_id' => $message_id | |
| 4116 | - ]) | |
| 4117 | - ]); | |
| 4118 | - set_transient($confirm_key, true, 300); | |
| 4119 | - } | |
| 4120 | - } | |
| 4121 | - } else { | |
| 4122 | - //error_log("[MxChat Telegram DEBUG] No session found for topic_id={$topic_id}"); | |
| 4123 | - } | |
| 4124 | - } else { | |
| 4125 | - //error_log('[MxChat Telegram DEBUG] No message in webhook data'); | |
| 4126 | - } | |
| 4127 | - | |
| 4128 | - return new WP_REST_Response(['ok' => true]); | |
| 4129 | -} | |
| 4130 | - | |
| 4131 | 3011 | public function mxchat_send_user_message_to_agent($message, $user_id, $session_id) { |
| 4132 | - // Check if this is a Telegram agent session | |
| 4133 | - $telegram_topic_id = get_option("mxchat_telegram_topic_{$session_id}", ''); | |
| 4134 | - if (!empty($telegram_topic_id)) { | |
| 4135 | - return $this->mxchat_send_user_message_to_telegram_agent($message, $user_id, $session_id); | |
| 4136 | - } | |
| 4137 | - | |
| 4138 | - // Otherwise, try Slack | |
| 4139 | 3012 | $slack_bot_token = $this->options['live_agent_bot_token'] ?? ''; |
| 4140 | 3013 | $channel_id = get_option("mxchat_channel_{$session_id}", ''); |
| 4141 | 3014 | |
| 4142 | 3015 | if (empty($slack_bot_token) || empty($channel_id)) { |
| @@ -4362,33 +3235,33 @@ | ||
| 4362 | 3235 | |
| 4363 | 3236 | $channel_id = $event['channel']; |
| 4364 | 3237 | $message_text = $event['text'] ?? ''; |
| 4365 | 3238 | $message_ts = $event['ts'] ?? ''; |
| 4366 | - | |
| 3239 | + | |
| 4367 | 3240 | // Find session ID by looking for matching channel |
| 4368 | 3241 | global $wpdb; |
| 4369 | 3242 | $session_option = $wpdb->get_var( |
| 4370 | 3243 | $wpdb->prepare( |
| 4371 | - "SELECT option_name FROM {$wpdb->options} | |
| 4372 | - WHERE option_name LIKE 'mxchat_channel_%' | |
| 3244 | + "SELECT option_name FROM {$wpdb->options} | |
| 3245 | + WHERE option_name LIKE 'mxchat_channel_%' | |
| 4373 | 3246 | AND option_value = %s", |
| 4374 | 3247 | $channel_id |
| 4375 | 3248 | ) |
| 4376 | 3249 | ); |
| 4377 | - | |
| 3250 | + | |
| 4378 | 3251 | if ($session_option) { |
| 4379 | 3252 | $session_id = str_replace('mxchat_channel_', '', $session_option); |
| 4380 | - | |
| 3253 | + | |
| 4381 | 3254 | // Create a unique key for this specific message |
| 4382 | 3255 | $message_key = md5($session_id . $message_ts . $message_text); |
| 4383 | 3256 | $processed_messages = get_transient('mxchat_processed_messages_' . $session_id) ?: []; |
| 4384 | - | |
| 3257 | + | |
| 4385 | 3258 | // Check if we've already processed this exact message |
| 4386 | 3259 | if (in_array($message_key, $processed_messages)) { |
| 4387 | 3260 | //error_log("Duplicate message detected for session $session_id"); |
| 4388 | 3261 | return new WP_REST_Response(['ok' => true]); |
| 4389 | 3262 | } |
| 4390 | - | |
| 3263 | + | |
| 4391 | 3264 | // Add to processed messages |
| 4392 | 3265 | $processed_messages[] = $message_key; |
| 4393 | 3266 | // Keep only last 50 messages per session |
| 4394 | 3267 | if (count($processed_messages) > 50) { |
| @@ -4394,46 +3267,14 @@ | ||
| 4394 | 3267 | if (count($processed_messages) > 50) { |
| 4395 | 3268 | $processed_messages = array_slice($processed_messages, -50); |
| 4396 | 3269 | } |
| 4397 | 3270 | set_transient('mxchat_processed_messages_' . $session_id, $processed_messages, HOUR_IN_SECONDS); |
| 4398 | - | |
| 4399 | - $slack_bot_token = $this->options['live_agent_bot_token'] ?? ''; | |
| 4400 | - | |
| 4401 | - // Handle agent ending the chat — transfer back to AI | |
| 4402 | - // Format: "!endchat" or "!endchat <custom message to user>" | |
| 4403 | - if (preg_match('/^!endchat\b/i', trim($message_text))) { | |
| 4404 | - update_option("mxchat_mode_{$session_id}", 'ai'); | |
| 4405 | - | |
| 4406 | - // Extract custom message after !endchat, or use empty string | |
| 4407 | - $custom_message = trim(preg_replace('/^!endchat\s*/i', '', trim($message_text))); | |
| 4408 | - | |
| 4409 | - // Send the agent's custom farewell message if provided | |
| 4410 | - if (!empty($custom_message)) { | |
| 4411 | - $this->mxchat_save_chat_message($session_id, 'agent', $custom_message); | |
| 4412 | - } | |
| 4413 | - | |
| 4414 | - // Confirm in Slack channel | |
| 4415 | - if (!empty($slack_bot_token)) { | |
| 4416 | - wp_remote_post('https://slack.com/api/chat.postMessage', [ | |
| 4417 | - 'headers' => [ | |
| 4418 | - 'Content-Type' => 'application/json', | |
| 4419 | - 'Authorization' => 'Bearer ' . $slack_bot_token | |
| 4420 | - ], | |
| 4421 | - 'body' => json_encode([ | |
| 4422 | - 'channel' => $channel_id, | |
| 4423 | - 'text' => "✅ *Chat ended.* User has been transferred back to AI mode.", | |
| 4424 | - 'mrkdwn' => true | |
| 4425 | - ]) | |
| 4426 | - ]); | |
| 4427 | - } | |
| 4428 | - | |
| 4429 | - return new WP_REST_Response(['ok' => true]); | |
| 4430 | - } | |
| 4431 | - | |
| 3271 | + | |
| 4432 | 3272 | // Save the agent message |
| 4433 | 3273 | $this->mxchat_save_chat_message($session_id, 'agent', $message_text); |
| 4434 | - | |
| 3274 | + | |
| 4435 | 3275 | // Send confirmation back to Slack (only once) |
| 3276 | + $slack_bot_token = $this->options['live_agent_bot_token'] ?? ''; | |
| 4436 | 3277 | if (!empty($slack_bot_token)) { |
| 4437 | 3278 | // Use a transient to prevent duplicate confirmations |
| 4438 | 3279 | $confirm_key = 'mxchat_confirm_' . $message_key; |
| 4439 | 3280 | if (!get_transient($confirm_key)) { |
| @@ -4443,9 +3284,9 @@ | ||
| 4443 | 3284 | 'Authorization' => 'Bearer ' . $slack_bot_token |
| 4444 | 3285 | ], |
| 4445 | 3286 | 'body' => json_encode([ |
| 4446 | 3287 | 'channel' => $channel_id, |
| 4447 | - 'text' => "✅ _Message sent to user_", | |
| 3288 | + 'text' => "✅ _Message sent to user_", | |
| 4448 | 3289 | 'thread_ts' => $event['ts'] // Reply in thread |
| 4449 | 3290 | ]) |
| 4450 | 3291 | ]); |
| 4451 | 3292 | // Set transient to prevent duplicate confirmations |
| @@ -4684,44 +3525,26 @@ | ||
| 4684 | 3525 | } |
| 4685 | 3526 | } |
| 4686 | 3527 | |
| 4687 | 3528 | |
| 4688 | -private function mxchat_find_relevant_content($user_embedding, $bot_id = 'default', $user_query = '') { | |
| 4689 | - //error_log("MXCHAT DEBUG: find_relevant_content called with bot_id: " . $bot_id); | |
| 4690 | - | |
| 4691 | - // Check for OpenAI Vector Store first (takes priority when enabled) | |
| 4692 | - $bot_vectorstore_config = $this->get_bot_vectorstore_config($bot_id); | |
| 4693 | - | |
| 4694 | - if ($bot_vectorstore_config['use_vectorstore']) { | |
| 4695 | - // Get current model to verify it's an OpenAI model | |
| 4696 | - $bot_options = $this->get_bot_options($bot_id); | |
| 4697 | - $mxchat_options = get_option('mxchat_options', array()); | |
| 4698 | - $current_options = !empty($bot_options) ? $bot_options : $mxchat_options; | |
| 4699 | - $selected_model = $current_options['model'] ?? 'gpt-5.1-chat-latest'; | |
| 4700 | - | |
| 4701 | - if ($this->is_openai_chat_model($selected_model)) { | |
| 4702 | - //error_log("MXCHAT DEBUG: Using OpenAI Vector Store for knowledge retrieval"); | |
| 4703 | - return $this->find_relevant_content_openai_vectorstore($user_query, $bot_id, $bot_vectorstore_config); | |
| 4704 | - } else { | |
| 4705 | - //error_log("MXCHAT DEBUG: Vector Store enabled but model is not OpenAI (" . $selected_model . "), skipping Vector Store"); | |
| 4706 | - } | |
| 4707 | - } | |
| 4708 | - | |
| 3529 | +private function mxchat_find_relevant_content($user_embedding, $bot_id = 'default') { | |
| 3530 | + error_log("MXCHAT DEBUG: find_relevant_content called with bot_id: " . $bot_id); | |
| 3531 | + | |
| 4709 | 3532 | // Get bot-specific Pinecone configuration |
| 4710 | 3533 | $bot_pinecone_config = $this->get_bot_pinecone_config($bot_id); |
| 4711 | - | |
| 3534 | + | |
| 4712 | 3535 | // Debug: Log the Pinecone configuration |
| 4713 | - //error_log("MXCHAT DEBUG: Pinecone config for bot '$bot_id':"); | |
| 4714 | - //error_log(" - use_pinecone: " . ($bot_pinecone_config['use_pinecone'] ? 'true' : 'false')); | |
| 4715 | - //error_log(" - api_key: " . (empty($bot_pinecone_config['api_key']) ? 'EMPTY' : 'SET (hidden)')); | |
| 4716 | - //error_log(" - host: " . ($bot_pinecone_config['host'] ?? 'NOT SET')); | |
| 4717 | - //error_log(" - namespace: " . ($bot_pinecone_config['namespace'] ?? 'NOT SET')); | |
| 4718 | - | |
| 3536 | + error_log("MXCHAT DEBUG: Pinecone config for bot '$bot_id':"); | |
| 3537 | + error_log(" - use_pinecone: " . ($bot_pinecone_config['use_pinecone'] ? 'true' : 'false')); | |
| 3538 | + error_log(" - api_key: " . (empty($bot_pinecone_config['api_key']) ? 'EMPTY' : 'SET (hidden)')); | |
| 3539 | + error_log(" - host: " . ($bot_pinecone_config['host'] ?? 'NOT SET')); | |
| 3540 | + error_log(" - namespace: " . ($bot_pinecone_config['namespace'] ?? 'NOT SET')); | |
| 3541 | + | |
| 4719 | 3542 | // Determine whether to use Pinecone based on bot configuration |
| 4720 | 3543 | $use_pinecone = isset($bot_pinecone_config['use_pinecone']) ? $bot_pinecone_config['use_pinecone'] : false; |
| 3544 | + | |
| 3545 | + error_log("MXCHAT DEBUG: Using " . ($use_pinecone ? "Pinecone" : "WordPress Database") . " for knowledge retrieval"); | |
| 4721 | 3546 | |
| 4722 | - //error_log("MXCHAT DEBUG: Using " . ($use_pinecone ? "Pinecone" : "WordPress Database") . " for knowledge retrieval"); | |
| 4723 | - | |
| 4724 | 3547 | if ($use_pinecone) { |
| 4725 | 3548 | return $this->find_relevant_content_pinecone($user_embedding, $bot_id, $bot_pinecone_config); |
| 4726 | 3549 | } else { |
| 4727 | 3550 | return $this->find_relevant_content_wordpress($user_embedding, $bot_id); |
| @@ -4730,482 +3553,233 @@ | ||
| 4730 | 3553 | |
| 4731 | 3554 | private function find_relevant_content_wordpress($user_embedding, $bot_id = 'default') { |
| 4732 | 3555 | global $wpdb; |
| 4733 | 3556 | $system_prompt_table = $wpdb->prefix . 'mxchat_system_prompt_content'; |
| 3557 | + $cache_key = 'mxchat_system_prompt_embeddings_' . $bot_id; // Bot-specific cache key | |
| 3558 | + $batch_size = 500; | |
| 3559 | + | |
| 4734 | 3560 | // Initialize similarity analysis storage |
| 4735 | 3561 | $this->last_similarity_analysis = [ |
| 4736 | 3562 | 'knowledge_base_type' => 'WordPress Database', |
| 4737 | - 'bot_id' => $bot_id, | |
| 3563 | + 'bot_id' => $bot_id, // Track which bot is being used | |
| 4738 | 3564 | 'top_matches' => [], |
| 4739 | 3565 | 'threshold_used' => 0, |
| 4740 | 3566 | 'total_checked' => 0 |
| 4741 | 3567 | ]; |
| 4742 | 3568 | |
| 4743 | - // NEW: Initialize valid URLs array | |
| 4744 | - $valid_urls = []; | |
| 4745 | - | |
| 4746 | - // Get bot-specific options for similarity threshold | |
| 3569 | + // Get bot-specific options for similarity threshold | |
| 4747 | 3570 | $bot_options = $this->get_bot_options($bot_id); |
| 4748 | 3571 | $current_options = !empty($bot_options) ? $bot_options : $this->options; |
| 4749 | 3572 | |
| 4750 | - // Get knowledge manager instance for role checking | |
| 4751 | - $knowledge_manager = MxChat_Knowledge_Manager::get_instance(); | |
| 3573 | + // Retrieve embeddings from cache or database | |
| 3574 | + $embeddings = wp_cache_get($cache_key, 'mxchat_system_prompts'); | |
| 3575 | + if ($embeddings === false) { | |
| 3576 | + // Cache miss - load embeddings from database WITH CONTENT and ROLE RESTRICTION for testing | |
| 3577 | + $embeddings = []; | |
| 3578 | + $offset = 0; | |
| 4752 | 3579 | |
| 4753 | - // Get base similarity threshold from bot options or default options | |
| 4754 | - $similarity_threshold = isset($current_options['similarity_threshold']) | |
| 4755 | - ? ((int) $current_options['similarity_threshold']) / 100 | |
| 4756 | - : 0.35; | |
| 4757 | - $this->last_similarity_analysis['threshold_used'] = $similarity_threshold; | |
| 3580 | + do { | |
| 3581 | + // Add bot_id filter if not default and if bot_metadata column exists | |
| 3582 | + $bot_filter = ''; | |
| 3583 | + if ($bot_id !== 'default') { | |
| 3584 | + // Check if bot_metadata column exists | |
| 3585 | + $column_exists = $wpdb->get_var("SHOW COLUMNS FROM {$system_prompt_table} LIKE 'bot_metadata'"); | |
| 3586 | + if ($column_exists) { | |
| 3587 | + $bot_filter = $wpdb->prepare(" AND (bot_metadata = %s OR bot_metadata IS NULL OR bot_metadata = '')", $bot_id); | |
| 3588 | + } | |
| 3589 | + } | |
| 4758 | 3590 | |
| 4759 | - // Precompute bot_filter once, outside the streaming loop | |
| 4760 | - $bot_filter = ''; | |
| 4761 | - if ($bot_id !== 'default') { | |
| 4762 | - $column_exists = $wpdb->get_var("SHOW COLUMNS FROM {$system_prompt_table} LIKE 'bot_metadata'"); | |
| 4763 | - if ($column_exists) { | |
| 4764 | - $bot_filter = $wpdb->prepare(" AND (bot_metadata = %s OR bot_metadata IS NULL OR bot_metadata = '')", $bot_id); | |
| 4765 | - } | |
| 4766 | - } | |
| 3591 | + $query = $wpdb->prepare( | |
| 3592 | + "SELECT id, embedding_vector, article_content, source_url, role_restriction | |
| 3593 | + FROM {$system_prompt_table} | |
| 3594 | + WHERE 1=1 {$bot_filter} | |
| 3595 | + LIMIT %d OFFSET %d", | |
| 3596 | + $batch_size, | |
| 3597 | + $offset | |
| 3598 | + ); | |
| 4767 | 3599 | |
| 4768 | - // ===== STREAMING TOP-K PASS ===== | |
| 4769 | - // Stream rows in small batches, compute cosine similarity per row, and keep only: | |
| 4770 | - // - top 10 by raw similarity (for the testing/debug display panel) | |
| 4771 | - // - candidates above threshold with access (capped) for context assembly | |
| 4772 | - // This bounds peak memory regardless of knowledge base size and avoids loading | |
| 4773 | - // article_content for every row. article_content is fetched in Phase 2 for winners only. | |
| 4774 | - $batch_size = 250; | |
| 4775 | - $max_candidates = 200; // safety cap, well above rag_sources_limit * max_chunks_per_source | |
| 4776 | - $top_display = []; | |
| 4777 | - $candidates = []; | |
| 4778 | - $total_checked = 0; | |
| 4779 | - $offset = 0; | |
| 3600 | + $batch = $wpdb->get_results($query); | |
| 3601 | + if (empty($batch)) { | |
| 3602 | + break; | |
| 3603 | + } | |
| 4780 | 3604 | |
| 4781 | - do { | |
| 4782 | - $batch = $wpdb->get_results($wpdb->prepare( | |
| 4783 | - "SELECT id, embedding_vector, source_url, role_restriction | |
| 4784 | - FROM {$system_prompt_table} | |
| 4785 | - WHERE 1=1 {$bot_filter} | |
| 4786 | - LIMIT %d OFFSET %d", | |
| 4787 | - $batch_size, | |
| 4788 | - $offset | |
| 4789 | - )); | |
| 3605 | + $embeddings = array_merge($embeddings, $batch); | |
| 3606 | + $offset += $batch_size; | |
| 3607 | + unset($batch); | |
| 3608 | + } while (true); | |
| 4790 | 3609 | |
| 4791 | - if (empty($batch)) { | |
| 4792 | - break; | |
| 3610 | + if (empty($embeddings)) { | |
| 3611 | + return ''; | |
| 4793 | 3612 | } |
| 3613 | + | |
| 3614 | + // Cache embeddings for future use (but note: this now includes content and role restrictions) | |
| 3615 | + wp_cache_set($cache_key, $embeddings, 'mxchat_system_prompts', 3600); | |
| 3616 | + } | |
| 4794 | 3617 | |
| 4795 | - foreach ($batch as $row) { | |
| 4796 | - $database_embedding = $row->embedding_vector | |
| 4797 | - ? unserialize($row->embedding_vector, ['allowed_classes' => false]) | |
| 4798 | - : null; | |
| 3618 | + // Get knowledge manager instance for role checking | |
| 3619 | + $knowledge_manager = MxChat_Knowledge_Manager::get_instance(); | |
| 4799 | 3620 | |
| 4800 | - if (!is_array($database_embedding) || !is_array($user_embedding)) { | |
| 4801 | - unset($database_embedding); | |
| 4802 | - continue; | |
| 4803 | - } | |
| 4804 | - | |
| 3621 | + // Get base similarity threshold from bot options or default options | |
| 3622 | + $similarity_threshold = isset($current_options['similarity_threshold']) | |
| 3623 | + ? ((int) $current_options['similarity_threshold']) / 100 | |
| 3624 | + : 0.35; | |
| 3625 | + | |
| 3626 | + $this->last_similarity_analysis['threshold_used'] = $similarity_threshold; | |
| 3627 | + | |
| 3628 | + // Calculate similarities and build results array | |
| 3629 | + $all_similarities = []; | |
| 3630 | + $relevant_results = []; | |
| 3631 | + | |
| 3632 | + foreach ($embeddings as $embedding) { | |
| 3633 | + $database_embedding = $embedding->embedding_vector | |
| 3634 | + ? unserialize($embedding->embedding_vector, ['allowed_classes' => false]) | |
| 3635 | + : null; | |
| 3636 | + | |
| 3637 | + if (is_array($database_embedding) && is_array($user_embedding)) { | |
| 4805 | 3638 | $similarity = $this->mxchat_calculate_cosine_similarity($user_embedding, $database_embedding); |
| 4806 | - unset($database_embedding); | |
| 4807 | - | |
| 4808 | - $role_restriction = $row->role_restriction ?? 'public'; | |
| 3639 | + | |
| 3640 | + // Check role access | |
| 3641 | + $role_restriction = $embedding->role_restriction ?? 'public'; | |
| 4809 | 3642 | $has_access = $knowledge_manager->mxchat_user_has_content_access($role_restriction); |
| 4810 | - $source_url = $row->source_url ?? ''; | |
| 4811 | - | |
| 4812 | - // Maintain top 10 display buffer (insert-if-beats-worst) | |
| 4813 | - if (count($top_display) < 10) { | |
| 4814 | - $top_display[] = [ | |
| 4815 | - 'id' => $row->id, | |
| 4816 | - 'similarity' => $similarity, | |
| 4817 | - 'source_url' => $source_url, | |
| 4818 | - 'role_restriction' => $role_restriction, | |
| 4819 | - 'has_access' => $has_access, | |
| 4820 | - ]; | |
| 4821 | - usort($top_display, function ($a, $b) { | |
| 4822 | - return $b['similarity'] <=> $a['similarity']; | |
| 4823 | - }); | |
| 4824 | - } elseif ($similarity > $top_display[9]['similarity']) { | |
| 4825 | - $top_display[9] = [ | |
| 4826 | - 'id' => $row->id, | |
| 4827 | - 'similarity' => $similarity, | |
| 4828 | - 'source_url' => $source_url, | |
| 4829 | - 'role_restriction' => $role_restriction, | |
| 4830 | - 'has_access' => $has_access, | |
| 4831 | - ]; | |
| 4832 | - usort($top_display, function ($a, $b) { | |
| 4833 | - return $b['similarity'] <=> $a['similarity']; | |
| 4834 | - }); | |
| 3643 | + | |
| 3644 | + // Store ALL similarities for testing (top 10) | |
| 3645 | + $source_display = ''; | |
| 3646 | + if (!empty($embedding->source_url) && $embedding->source_url !== '#') { | |
| 3647 | + $source_display = $embedding->source_url; | |
| 3648 | + } else { | |
| 3649 | + $content_preview = strip_tags($embedding->article_content ?? ''); | |
| 3650 | + $content_preview = preg_replace('/\s+/', ' ', $content_preview); | |
| 3651 | + $source_display = substr(trim($content_preview), 0, 50) . '...'; | |
| 4835 | 3652 | } |
| 4836 | - | |
| 4837 | - // Track candidates for context assembly (above threshold + has access) | |
| 3653 | + | |
| 3654 | + $all_similarities[] = [ | |
| 3655 | + 'document_id' => $embedding->id, | |
| 3656 | + 'similarity' => $similarity, | |
| 3657 | + 'similarity_percentage' => round($similarity * 100, 2), | |
| 3658 | + 'above_threshold' => $similarity >= $similarity_threshold, | |
| 3659 | + 'source_display' => $source_display, | |
| 3660 | + 'content_preview' => substr(strip_tags($embedding->article_content ?? ''), 0, 100) . '...', | |
| 3661 | + 'used_for_context' => false, // Initialize as false, we'll update this later | |
| 3662 | + 'role_restriction' => $role_restriction, // Include role info for testing | |
| 3663 | + 'has_access' => $has_access, // Include access info for testing | |
| 3664 | + 'filtered_out' => !$has_access // Mark if filtered out by role | |
| 3665 | + ]; | |
| 3666 | + | |
| 3667 | + // Only consider results above threshold AND with access for actual content retrieval | |
| 4838 | 3668 | if ($similarity >= $similarity_threshold && $has_access) { |
| 4839 | - $candidates[] = [ | |
| 4840 | - 'id' => $row->id, | |
| 4841 | - 'similarity' => $similarity, | |
| 4842 | - 'source_url' => $source_url, | |
| 3669 | + $relevant_results[] = [ | |
| 3670 | + 'id' => $embedding->id, | |
| 3671 | + 'similarity' => $similarity | |
| 4843 | 3672 | ]; |
| 4844 | 3673 | } |
| 4845 | - | |
| 4846 | - $total_checked++; | |
| 4847 | 3674 | } |
| 4848 | - | |
| 4849 | - unset($batch); | |
| 4850 | - | |
| 4851 | - // Trim candidates periodically to cap memory during long scans | |
| 4852 | - if (count($candidates) > $max_candidates) { | |
| 4853 | - usort($candidates, function ($a, $b) { | |
| 4854 | - return $b['similarity'] <=> $a['similarity']; | |
| 4855 | - }); | |
| 4856 | - $candidates = array_slice($candidates, 0, $max_candidates); | |
| 4857 | - } | |
| 4858 | - | |
| 4859 | - $offset += $batch_size; | |
| 4860 | - } while (true); | |
| 4861 | - | |
| 4862 | - if ($total_checked === 0) { | |
| 4863 | - $this->current_valid_urls = []; | |
| 4864 | - return ''; | |
| 3675 | + | |
| 3676 | + unset($database_embedding); | |
| 4865 | 3677 | } |
| 4866 | 3678 | |
| 4867 | - // Final candidates sort (best first) | |
| 4868 | - if (count($candidates) > 1) { | |
| 4869 | - usort($candidates, function ($a, $b) { | |
| 4870 | - return $b['similarity'] <=> $a['similarity']; | |
| 4871 | - }); | |
| 4872 | - } | |
| 4873 | - | |
| 4874 | - // ===== PHASE 2: FETCH ARTICLE CONTENT ONLY FOR WINNERS ===== | |
| 4875 | - // Gather unique IDs we actually need (top_display + candidates) and pull | |
| 4876 | - // article_content in bounded IN() batches. This avoids loading content for | |
| 4877 | - // every row during the similarity scan. | |
| 4878 | - $needed_ids = []; | |
| 4879 | - foreach ($top_display as $item) { | |
| 4880 | - $needed_ids[$item['id']] = true; | |
| 4881 | - } | |
| 4882 | - foreach ($candidates as $item) { | |
| 4883 | - $needed_ids[$item['id']] = true; | |
| 4884 | - } | |
| 4885 | - $needed_ids = array_keys($needed_ids); | |
| 4886 | - | |
| 4887 | - $content_map = []; | |
| 4888 | - if (!empty($needed_ids)) { | |
| 4889 | - foreach (array_chunk($needed_ids, 250) as $chunk_ids) { | |
| 4890 | - $placeholders = implode(',', array_fill(0, count($chunk_ids), '%d')); | |
| 4891 | - $rows = $wpdb->get_results($wpdb->prepare( | |
| 4892 | - "SELECT id, article_content FROM {$system_prompt_table} WHERE id IN ($placeholders)", | |
| 4893 | - ...$chunk_ids | |
| 4894 | - )); | |
| 4895 | - foreach ($rows as $r) { | |
| 4896 | - $content_map[$r->id] = $r->article_content; | |
| 4897 | - } | |
| 4898 | - unset($rows); | |
| 4899 | - } | |
| 4900 | - } | |
| 4901 | - | |
| 4902 | - // Build the all_similarities display array from the top 10 | |
| 4903 | - $all_similarities = []; | |
| 4904 | - foreach ($top_display as $item) { | |
| 4905 | - $article_content_for_parse = $content_map[$item['id']] ?? ''; | |
| 4906 | - $parsed_for_display = MxChat_Chunker::parse_stored_chunk($article_content_for_parse); | |
| 4907 | - $is_chunk = $parsed_for_display['is_chunked']; | |
| 4908 | - $chunk_meta = $parsed_for_display['metadata']; | |
| 4909 | - | |
| 4910 | - if (!empty($item['source_url']) && $item['source_url'] !== '#') { | |
| 4911 | - $source_display = $item['source_url']; | |
| 4912 | - } else { | |
| 4913 | - $content_preview = strip_tags($article_content_for_parse); | |
| 4914 | - $content_preview = preg_replace('/\s+/', ' ', $content_preview); | |
| 4915 | - $source_display = substr(trim($content_preview), 0, 50) . '...'; | |
| 4916 | - } | |
| 4917 | - | |
| 4918 | - $all_similarities[] = [ | |
| 4919 | - 'document_id' => $item['id'], | |
| 4920 | - 'similarity' => $item['similarity'], | |
| 4921 | - 'similarity_percentage' => round($item['similarity'] * 100, 2), | |
| 4922 | - 'above_threshold' => $item['similarity'] >= $similarity_threshold, | |
| 4923 | - 'source_display' => $source_display, | |
| 4924 | - 'content_preview' => substr(strip_tags($parsed_for_display['text'] ?? ''), 0, 100) . '...', | |
| 4925 | - 'used_for_context' => false, | |
| 4926 | - 'role_restriction' => $item['role_restriction'], | |
| 4927 | - 'has_access' => $item['has_access'], | |
| 4928 | - 'filtered_out' => !$item['has_access'], | |
| 4929 | - 'is_chunk' => $is_chunk, | |
| 4930 | - 'chunk_index' => $is_chunk ? ($chunk_meta['chunk_index'] ?? 0) : null, | |
| 4931 | - 'total_chunks' => $is_chunk ? ($chunk_meta['total_chunks'] ?? 1) : null | |
| 4932 | - ]; | |
| 4933 | - } | |
| 4934 | - | |
| 4935 | - // Build url_groups from candidates for chunk reassembly | |
| 4936 | - $url_groups = array(); | |
| 4937 | - foreach ($candidates as $cand) { | |
| 4938 | - $article_content = $content_map[$cand['id']] ?? ''; | |
| 4939 | - $parsed = MxChat_Chunker::parse_stored_chunk($article_content); | |
| 4940 | - $is_chunked = $parsed['is_chunked']; | |
| 4941 | - $chunk_index = $parsed['metadata']['chunk_index'] ?? 0; | |
| 4942 | - $text_content = $parsed['text']; | |
| 4943 | - | |
| 4944 | - $source_url = $cand['source_url']; | |
| 4945 | - $group_key = !empty($source_url) ? $source_url : '_manual_' . $cand['id']; | |
| 4946 | - | |
| 4947 | - if (!isset($url_groups[$group_key])) { | |
| 4948 | - $url_groups[$group_key] = array( | |
| 4949 | - 'source_url' => $source_url, | |
| 4950 | - 'best_score' => 0, | |
| 4951 | - 'is_chunked' => $is_chunked, | |
| 4952 | - 'chunks' => array(), | |
| 4953 | - 'single_text' => '', | |
| 4954 | - 'single_id' => null | |
| 4955 | - ); | |
| 4956 | - } | |
| 4957 | - | |
| 4958 | - if ($cand['similarity'] > $url_groups[$group_key]['best_score']) { | |
| 4959 | - $url_groups[$group_key]['best_score'] = $cand['similarity']; | |
| 4960 | - } | |
| 4961 | - | |
| 4962 | - if ($is_chunked) { | |
| 4963 | - $url_groups[$group_key]['is_chunked'] = true; | |
| 4964 | - $url_groups[$group_key]['chunks'][] = array( | |
| 4965 | - 'id' => $cand['id'], | |
| 4966 | - 'score' => $cand['similarity'], | |
| 4967 | - 'chunk_index' => $chunk_index, | |
| 4968 | - 'text' => $text_content | |
| 4969 | - ); | |
| 4970 | - } else { | |
| 4971 | - $url_groups[$group_key]['single_text'] = $text_content; | |
| 4972 | - $url_groups[$group_key]['single_id'] = $cand['id']; | |
| 4973 | - } | |
| 4974 | - } | |
| 4975 | - | |
| 4976 | 3679 | // Sort ALL similarities for testing display (highest first) |
| 4977 | 3680 | usort($all_similarities, function ($a, $b) { |
| 4978 | 3681 | return $b['similarity'] <=> $a['similarity']; |
| 4979 | 3682 | }); |
| 4980 | - | |
| 4981 | - // Sort URL groups by best score (highest first) | |
| 4982 | - uasort($url_groups, function($a, $b) { | |
| 4983 | - return $b['best_score'] <=> $a['best_score']; | |
| 3683 | + | |
| 3684 | + // Sort relevant results by similarity (highest first) | |
| 3685 | + usort($relevant_results, function ($a, $b) { | |
| 3686 | + return $b['similarity'] <=> $a['similarity']; | |
| 4984 | 3687 | }); |
| 4985 | - | |
| 4986 | - // Get RAG sources limit from options (default 6, min 3, max 10) | |
| 4987 | - $rag_sources_limit = isset($options['rag_sources_limit']) ? intval($options['rag_sources_limit']) : 3; | |
| 4988 | - if ($rag_sources_limit < 3) $rag_sources_limit = 3; | |
| 4989 | - if ($rag_sources_limit > 10) $rag_sources_limit = 10; | |
| 4990 | - | |
| 4991 | - // Take top N unique URLs based on user setting | |
| 4992 | - $top_urls = array_slice($url_groups, 0, $rag_sources_limit, true); | |
| 4993 | - | |
| 4994 | - // Track which document IDs are used for context | |
| 3688 | + | |
| 3689 | + // Get top 5 results for actual content (standard approach) | |
| 3690 | + $top_results = array_slice($relevant_results, 0, 5); | |
| 3691 | + | |
| 3692 | + // NOW mark which documents are actually used for context | |
| 4995 | 3693 | $used_document_ids = []; |
| 4996 | - foreach ($top_urls as $group) { | |
| 4997 | - if ($group['is_chunked']) { | |
| 4998 | - foreach ($group['chunks'] as $chunk) { | |
| 4999 | - $used_document_ids[] = $chunk['id']; | |
| 5000 | - } | |
| 5001 | - } elseif ($group['single_id']) { | |
| 5002 | - $used_document_ids[] = $group['single_id']; | |
| 5003 | - } | |
| 3694 | + foreach ($top_results as $result) { | |
| 3695 | + $used_document_ids[] = $result['id']; | |
| 5004 | 3696 | } |
| 5005 | - | |
| 3697 | + | |
| 5006 | 3698 | // Update the all_similarities array to mark which were actually used |
| 5007 | 3699 | foreach ($all_similarities as &$similarity_item) { |
| 5008 | 3700 | $similarity_item['used_for_context'] = in_array($similarity_item['document_id'], $used_document_ids); |
| 5009 | 3701 | } |
| 5010 | - | |
| 5011 | - // Store top 10 for testing panel | |
| 3702 | + | |
| 3703 | + // Store top 10 for testing panel (now with correct used_for_context flags and role info) | |
| 5012 | 3704 | $this->last_similarity_analysis['top_matches'] = array_slice($all_similarities, 0, 10); |
| 5013 | - $this->last_similarity_analysis['total_checked'] = $total_checked; | |
| 5014 | - | |
| 3705 | + $this->last_similarity_analysis['total_checked'] = count($embeddings); | |
| 3706 | + | |
| 3707 | + //error_log("MxChat Testing: Stored " . count($this->last_similarity_analysis['top_matches']) . " top matches for testing"); | |
| 3708 | + | |
| 5015 | 3709 | // Initialize final content |
| 5016 | 3710 | $content = ''; |
| 5017 | - $matches_used = 0; | |
| 5018 | - $total_chunks_used = 0; | |
| 5019 | - $max_total_chunks = isset($options['rag_chunks_limit']) ? intval($options['rag_chunks_limit']) : 15; | |
| 5020 | - if ($max_total_chunks < 8) $max_total_chunks = 8; | |
| 5021 | - if ($max_total_chunks > 20) $max_total_chunks = 20; | |
| 5022 | - $max_chunks_per_source = 5; // Cap per individual source to limit token usage | |
| 5023 | - | |
| 5024 | - // Check if citation links are enabled (default to 'on' for backwards compatibility) | |
| 5025 | - // Use fresh options to ensure we get the latest setting value | |
| 5026 | - $fresh_options = get_option('mxchat_options', []); | |
| 5027 | - $citation_links_enabled = isset($fresh_options['citation_links_toggle']) ? ($fresh_options['citation_links_toggle'] === 'on') : true; | |
| 5028 | - | |
| 5029 | - // Build content from top sources | |
| 5030 | - foreach ($top_urls as $group_key => $group) { | |
| 5031 | - $source_url = $group['source_url']; // Use actual source_url, not the group key | |
| 5032 | - | |
| 5033 | - // Stop if we've hit the total chunk limit | |
| 5034 | - if ($total_chunks_used >= $max_total_chunks) { | |
| 5035 | - break; | |
| 3711 | + | |
| 3712 | + // Track document IDs to avoid duplicates | |
| 3713 | + $added_document_ids = []; | |
| 3714 | + | |
| 3715 | + // Fetch and format content for each selected result | |
| 3716 | + foreach ($top_results as $index => $result) { | |
| 3717 | + if (in_array($result['id'], $added_document_ids)) { | |
| 3718 | + continue; | |
| 5036 | 3719 | } |
| 5037 | - | |
| 5038 | - $full_text = ''; | |
| 5039 | - $chunks_in_this_source = 1; // Default for non-chunked content | |
| 5040 | - | |
| 5041 | - if ($group['is_chunked']) { | |
| 5042 | - // Calculate how many chunks we can still use (respect both total and per-source caps) | |
| 5043 | - $chunks_remaining = min($max_chunks_per_source, $max_total_chunks - $total_chunks_used); | |
| 5044 | - | |
| 5045 | - // Fetch chunks for this URL with limit | |
| 5046 | - $full_text = $this->reassemble_chunks_from_wordpress($source_url, $chunks_remaining, $chunks_in_this_source); | |
| 5047 | - | |
| 5048 | - // If fetching all chunks fails, fall back to matched chunks | |
| 5049 | - if (empty($full_text)) { | |
| 5050 | - // Sort matched chunks by index and concatenate | |
| 5051 | - usort($group['chunks'], function($a, $b) { | |
| 5052 | - return $a['chunk_index'] <=> $b['chunk_index']; | |
| 5053 | - }); | |
| 5054 | - | |
| 5055 | - $chunk_texts = array(); | |
| 5056 | - $chunks_in_this_source = 0; | |
| 5057 | - foreach ($group['chunks'] as $chunk) { | |
| 5058 | - if ($total_chunks_used + $chunks_in_this_source >= $max_total_chunks) { | |
| 5059 | - break; | |
| 5060 | - } | |
| 5061 | - $chunk_texts[] = $chunk['text']; | |
| 5062 | - $chunks_in_this_source++; | |
| 3720 | + | |
| 3721 | + $chunk_content = $this->fetch_content_with_product_links($result['id']); | |
| 3722 | + $added_document_ids[] = $result['id']; | |
| 3723 | + | |
| 3724 | + $content .= "## Reference " . ($index + 1) . " ##\n"; | |
| 3725 | + $content .= $chunk_content . "\n\n"; | |
| 3726 | + | |
| 3727 | + // PDF surrounding pages logic (unchanged) | |
| 3728 | + if (strpos($chunk_content, '{"document_type":"pdf"') !== false) { | |
| 3729 | + $surrounding_content = $wpdb->get_results($wpdb->prepare( | |
| 3730 | + "SELECT id, article_content, role_restriction FROM {$system_prompt_table} | |
| 3731 | + WHERE id IN ( | |
| 3732 | + (SELECT id FROM {$system_prompt_table} WHERE id < %d ORDER BY id DESC LIMIT 1), | |
| 3733 | + (SELECT id FROM {$system_prompt_table} WHERE id > %d ORDER BY id ASC LIMIT 1) | |
| 3734 | + )", | |
| 3735 | + $result['id'], | |
| 3736 | + $result['id'] | |
| 3737 | + )); | |
| 3738 | + | |
| 3739 | + // Check role access for surrounding content too | |
| 3740 | + if (!empty($surrounding_content[0])) { | |
| 3741 | + $surrounding_role = $surrounding_content[0]->role_restriction ?? 'public'; | |
| 3742 | + if ($knowledge_manager->mxchat_user_has_content_access($surrounding_role)) { | |
| 3743 | + $content .= "## Related Content ##\n"; | |
| 3744 | + $content .= $surrounding_content[0]->article_content . "\n\n"; | |
| 3745 | + $added_document_ids[] = $surrounding_content[0]->id; | |
| 5063 | 3746 | } |
| 5064 | - $full_text = implode("\n\n", $chunk_texts); | |
| 5065 | 3747 | } |
| 5066 | - } else { | |
| 5067 | - $full_text = $group['single_text']; | |
| 5068 | - $chunks_in_this_source = 1; | |
| 5069 | - } | |
| 5070 | - | |
| 5071 | - if (!empty($full_text)) { | |
| 5072 | - // Strip URLs from content if citation links are disabled | |
| 5073 | - if (!$citation_links_enabled) { | |
| 5074 | - $full_text = preg_replace('#\bhttps?://[^\s<>"\']+#i', '', $full_text); | |
| 5075 | - $full_text = preg_replace('/\s+/', ' ', trim($full_text)); // Clean up extra spaces | |
| 5076 | - } | |
| 5077 | - | |
| 5078 | - // Use numbered reference for URL-based entries, plain info label for manual entries | |
| 5079 | - if (!empty($source_url) && $source_url !== '#') { | |
| 5080 | - $matches_used++; | |
| 5081 | - $content .= "## Reference " . $matches_used . " ##\n"; | |
| 5082 | - $content .= $full_text . "\n\n"; | |
| 5083 | - | |
| 5084 | - // Only include citation URLs if citation links are enabled | |
| 5085 | - if ($citation_links_enabled) { | |
| 5086 | - $valid_urls[] = $source_url; | |
| 5087 | - $content .= "URL: " . $source_url . "\n\n"; | |
| 3748 | + | |
| 3749 | + if (!empty($surrounding_content[1])) { | |
| 3750 | + $surrounding_role = $surrounding_content[1]->role_restriction ?? 'public'; | |
| 3751 | + if ($knowledge_manager->mxchat_user_has_content_access($surrounding_role)) { | |
| 3752 | + $content .= "## Related Content ##\n"; | |
| 3753 | + $content .= $surrounding_content[1]->article_content . "\n\n"; | |
| 3754 | + $added_document_ids[] = $surrounding_content[1]->id; | |
| 5088 | 3755 | } |
| 5089 | - } else { | |
| 5090 | - // Manual entry — no reference number, no citation | |
| 5091 | - $content .= "## Information ##\n"; | |
| 5092 | - $content .= $full_text . "\n\n"; | |
| 5093 | 3756 | } |
| 5094 | - | |
| 5095 | - // Extract any URLs from the text content itself (only if citation links enabled) | |
| 5096 | - if ($citation_links_enabled) { | |
| 5097 | - preg_match_all( | |
| 5098 | - '#\bhttps?://[^\s<>"\']+#i', | |
| 5099 | - $full_text, | |
| 5100 | - $content_urls | |
| 5101 | - ); | |
| 5102 | - if (!empty($content_urls[0])) { | |
| 5103 | - $valid_urls = array_merge($valid_urls, $content_urls[0]); | |
| 5104 | - } | |
| 5105 | - } | |
| 5106 | - | |
| 5107 | - $total_chunks_used += $chunks_in_this_source; | |
| 5108 | 3757 | } |
| 5109 | 3758 | } |
| 5110 | - | |
| 5111 | - // NEW: Store unique valid URLs for validation | |
| 5112 | - $this->current_valid_urls = array_unique($valid_urls); | |
| 5113 | - | |
| 5114 | - // Store sources and chunks counts for testing/transcript display | |
| 5115 | - $this->last_similarity_analysis['sources_used'] = $matches_used; | |
| 5116 | - $this->last_similarity_analysis['total_chunks_used'] = $total_chunks_used; | |
| 5117 | - | |
| 5118 | - // Allow add-ons to act on similarity results (e.g. WooCommerce product card display) | |
| 5119 | - do_action('mxchat_similarity_results', $this->last_similarity_analysis['top_matches'], $bot_id); | |
| 5120 | - | |
| 3759 | + | |
| 5121 | 3760 | // Add response guidelines |
| 5122 | - if (empty($top_urls)) { | |
| 3761 | + if (empty($top_results)) { | |
| 5123 | 3762 | $content = "No reference information was found for this query.\n\n"; |
| 5124 | 3763 | } else { |
| 5125 | - // Build response guidelines based on citation links setting | |
| 5126 | 3764 | $content .= "\n## Response Guidelines ##\n" . |
| 5127 | 3765 | "You are an AI Chatbot. Answer naturally and helpfully using only the information from the references above. " . |
| 5128 | 3766 | "Be conversational and friendly, but never mention your knowledge base or training data. " . |
| 5129 | 3767 | "If you don't have specific information or are uncertain about any details, it's always " . |
| 5130 | 3768 | "better to honestly say you don't know rather than making up or guessing at answers. " . |
| 5131 | - "When information is incomplete, let them know you are unsure.\n\n"; | |
| 5132 | - | |
| 5133 | - // Only add hyperlink instructions if citation links are enabled | |
| 5134 | - if ($citation_links_enabled) { | |
| 5135 | - $content .= "CRITICAL: When creating hyperlinks, always use proper markdown format with descriptive text: " . | |
| 5136 | - "[descriptive text](url). NEVER use empty brackets like [](url). The text in brackets must describe what the link is about. " . | |
| 5137 | - "Only cite references that have a URL. Do not cite or add source labels to Information sections that have no URL."; | |
| 5138 | - } else { | |
| 5139 | - $content .= "IMPORTANT: Do not include any citation links, source URLs, or hyperlinks in your responses. " . | |
| 5140 | - "Simply provide helpful answers based on the reference information without citing sources."; | |
| 5141 | - } | |
| 3769 | + "When information is incomplete, let them know you are unsure."; | |
| 5142 | 3770 | } |
| 5143 | 3771 | |
| 5144 | 3772 | return trim($content); |
| 5145 | 3773 | } |
| 5146 | 3774 | |
| 5147 | -/** | |
| 5148 | - * Fetch and reassemble chunks for a URL from WordPress database | |
| 5149 | - * | |
| 5150 | - * @param string $source_url The source URL to fetch chunks for | |
| 5151 | - * @param int $max_chunks Maximum number of chunks to return (0 = unlimited) | |
| 5152 | - * @param int &$chunk_count Reference to store the actual number of chunks returned | |
| 5153 | - * @return string Reassembled content from chunks | |
| 5154 | - */ | |
| 5155 | -private function reassemble_chunks_from_wordpress($source_url, $max_chunks = 0, &$chunk_count = 0) { | |
| 5156 | - global $wpdb; | |
| 5157 | - $table = $wpdb->prefix . 'mxchat_system_prompt_content'; | |
| 5158 | - | |
| 5159 | - // Fetch all rows with this source_url | |
| 5160 | - $rows = $wpdb->get_results($wpdb->prepare( | |
| 5161 | - "SELECT article_content FROM {$table} | |
| 5162 | - WHERE source_url = %s | |
| 5163 | - ORDER BY id ASC", | |
| 5164 | - $source_url | |
| 5165 | - )); | |
| 5166 | - | |
| 5167 | - if (empty($rows)) { | |
| 5168 | - $chunk_count = 0; | |
| 5169 | - return ''; | |
| 5170 | - } | |
| 5171 | - | |
| 5172 | - // Parse and sort chunks by index | |
| 5173 | - $chunks = array(); | |
| 5174 | - foreach ($rows as $row) { | |
| 5175 | - $parsed = MxChat_Chunker::parse_stored_chunk($row->article_content); | |
| 5176 | - | |
| 5177 | - if ($parsed['is_chunked']) { | |
| 5178 | - $chunk_index = $parsed['metadata']['chunk_index'] ?? 0; | |
| 5179 | - $chunks[$chunk_index] = $parsed['text']; | |
| 5180 | - } else { | |
| 5181 | - // Non-chunked content - just return it | |
| 5182 | - $chunks[] = $parsed['text']; | |
| 5183 | - } | |
| 5184 | - } | |
| 5185 | - | |
| 5186 | - // Sort by chunk index | |
| 5187 | - ksort($chunks); | |
| 5188 | - | |
| 5189 | - // Apply chunk limit if specified | |
| 5190 | - if ($max_chunks > 0 && count($chunks) > $max_chunks) { | |
| 5191 | - $chunks = array_slice($chunks, 0, $max_chunks, true); | |
| 5192 | - } | |
| 5193 | - | |
| 5194 | - // Store actual chunk count | |
| 5195 | - $chunk_count = count($chunks); | |
| 5196 | - | |
| 5197 | - // Reassemble content | |
| 5198 | - return implode("\n\n", $chunks); | |
| 5199 | -} | |
| 5200 | - | |
| 5201 | 3775 | private function find_relevant_content_pinecone($user_embedding, $bot_id = 'default', $bot_config = null) { |
| 5202 | 3776 | global $wpdb; |
| 5203 | 3777 | |
| 5204 | - //error_log("MXCHAT DEBUG: find_relevant_content_pinecone called"); | |
| 5205 | - //error_log(" - bot_id: " . $bot_id); | |
| 5206 | - //error_log(" - user_embedding is array: " . (is_array($user_embedding) ? 'yes' : 'no')); | |
| 5207 | - //error_log(" - user_embedding count: " . (is_array($user_embedding) ? count($user_embedding) : 'N/A')); | |
| 3778 | + error_log("MXCHAT DEBUG: find_relevant_content_pinecone called"); | |
| 3779 | + error_log(" - bot_id: " . $bot_id); | |
| 3780 | + error_log(" - user_embedding is array: " . (is_array($user_embedding) ? 'yes' : 'no')); | |
| 3781 | + error_log(" - user_embedding count: " . (is_array($user_embedding) ? count($user_embedding) : 'N/A')); | |
| 5208 | 3782 | |
| 5209 | 3783 | // Use bot-specific config or fall back to default |
| 5210 | 3784 | if ($bot_config === null) { |
| 5211 | 3785 | $bot_config = $this->get_bot_pinecone_config($bot_id); |
| @@ -5214,12 +3788,12 @@ | ||
| 5214 | 3788 | $api_key = $bot_config['api_key'] ?? ''; |
| 5215 | 3789 | $host = $bot_config['host'] ?? ''; |
| 5216 | 3790 | $namespace = $bot_config['namespace'] ?? ''; |
| 5217 | 3791 | |
| 5218 | - //error_log("MXCHAT DEBUG: Pinecone query parameters:"); | |
| 5219 | - //error_log(" - API Key: " . (empty($api_key) ? 'EMPTY - ERROR!' : 'Present (length: ' . strlen($api_key) . ')')); | |
| 5220 | - //error_log(" - Host: " . (empty($host) ? 'EMPTY - ERROR!' : $host)); | |
| 5221 | - //error_log(" - Namespace: " . (empty($namespace) ? 'EMPTY (will use default)' : $namespace)); | |
| 3792 | + error_log("MXCHAT DEBUG: Pinecone query parameters:"); | |
| 3793 | + error_log(" - API Key: " . (empty($api_key) ? 'EMPTY - ERROR!' : 'Present (length: ' . strlen($api_key) . ')')); | |
| 3794 | + error_log(" - Host: " . (empty($host) ? 'EMPTY - ERROR!' : $host)); | |
| 3795 | + error_log(" - Namespace: " . (empty($namespace) ? 'EMPTY (will use default)' : $namespace)); | |
| 5222 | 3796 | |
| 5223 | 3797 | // Initialize similarity analysis storage |
| 5224 | 3798 | $this->last_similarity_analysis = [ |
| 5225 | 3799 | 'knowledge_base_type' => 'Pinecone', |
| @@ -5229,17 +3803,12 @@ | ||
| 5229 | 3803 | 'threshold_used' => 0, |
| 5230 | 3804 | 'total_checked' => 0 |
| 5231 | 3805 | ]; |
| 5232 | 3806 | |
| 5233 | - // NEW: Initialize valid URLs array | |
| 5234 | - $valid_urls = []; | |
| 5235 | - | |
| 5236 | 3807 | if (empty($host) || empty($api_key)) { |
| 5237 | - //error_log("MXCHAT DEBUG ERROR: Missing Pinecone host or API key!"); | |
| 5238 | - //error_log(" - Host empty: " . (empty($host) ? 'YES' : 'NO')); | |
| 5239 | - //error_log(" - API key empty: " . (empty($api_key) ? 'YES' : 'NO')); | |
| 5240 | - // Store empty array for valid URLs since we can't proceed | |
| 5241 | - $this->current_valid_urls = []; | |
| 3808 | + error_log("MXCHAT DEBUG ERROR: Missing Pinecone host or API key!"); | |
| 3809 | + error_log(" - Host empty: " . (empty($host) ? 'YES' : 'NO')); | |
| 3810 | + error_log(" - API key empty: " . (empty($api_key) ? 'YES' : 'NO')); | |
| 5242 | 3811 | return ''; |
| 5243 | 3812 | } |
| 5244 | 3813 | |
| 5245 | 3814 | // Get knowledge manager instance for role checking |
| @@ -5259,9 +3828,9 @@ | ||
| 5259 | 3828 | $api_endpoint = "https://{$host}/query"; |
| 5260 | 3829 | |
| 5261 | 3830 | $request_body = array( |
| 5262 | 3831 | 'vector' => $user_embedding, |
| 5263 | - 'topK' => 50, // Increased for chunked content grouping - need more candidates to find top N unique URLs | |
| 3832 | + 'topK' => 20, // Request more to get good testing data | |
| 5264 | 3833 | 'includeMetadata' => true, |
| 5265 | 3834 | 'includeValues' => true |
| 5266 | 3835 | ); |
| 5267 | 3836 | |
| @@ -5269,11 +3838,11 @@ | ||
| 5269 | 3838 | if (!empty($namespace)) { |
| 5270 | 3839 | $request_body['namespace'] = $namespace; |
| 5271 | 3840 | } |
| 5272 | 3841 | |
| 5273 | - //error_log("MXCHAT DEBUG: About to call Pinecone API"); | |
| 5274 | - //error_log(" - Endpoint: " . $api_endpoint); | |
| 5275 | - //error_log(" - Namespace in request: " . (!empty($namespace) ? $namespace : 'NOT SET')); | |
| 3842 | + error_log("MXCHAT DEBUG: About to call Pinecone API"); | |
| 3843 | + error_log(" - Endpoint: " . $api_endpoint); | |
| 3844 | + error_log(" - Namespace in request: " . (!empty($namespace) ? $namespace : 'NOT SET')); | |
| 5276 | 3845 | |
| 5277 | 3846 | $response = wp_remote_post($api_endpoint, array( |
| 5278 | 3847 | 'headers' => array( |
| 5279 | 3848 | 'Api-Key' => $api_key, |
| @@ -5284,61 +3853,53 @@ | ||
| 5284 | 3853 | 'timeout' => 30 |
| 5285 | 3854 | )); |
| 5286 | 3855 | |
| 5287 | 3856 | if (is_wp_error($response)) { |
| 5288 | - //error_log("MXCHAT DEBUG ERROR: WP Error in Pinecone request: " . $response->get_error_message()); | |
| 5289 | - // Store empty array for valid URLs | |
| 5290 | - $this->current_valid_urls = []; | |
| 3857 | + error_log("MXCHAT DEBUG ERROR: WP Error in Pinecone request: " . $response->get_error_message()); | |
| 5291 | 3858 | return ''; |
| 5292 | 3859 | } |
| 5293 | 3860 | |
| 5294 | 3861 | $response_code = wp_remote_retrieve_response_code($response); |
| 5295 | - //error_log("MXCHAT DEBUG: Pinecone response code: " . $response_code); | |
| 3862 | + error_log("MXCHAT DEBUG: Pinecone response code: " . $response_code); | |
| 5296 | 3863 | |
| 5297 | 3864 | if ($response_code !== 200) { |
| 5298 | 3865 | $response_body = wp_remote_retrieve_body($response); |
| 5299 | - //error_log("MXCHAT DEBUG ERROR: Pinecone API error response: " . substr($response_body, 0, 500)); | |
| 5300 | - // Store empty array for valid URLs | |
| 5301 | - $this->current_valid_urls = []; | |
| 3866 | + error_log("MXCHAT DEBUG ERROR: Pinecone API error response: " . substr($response_body, 0, 500)); | |
| 5302 | 3867 | return ''; |
| 5303 | 3868 | } |
| 5304 | 3869 | |
| 5305 | 3870 | // ADD DETAILED DEBUG SECTION HERE |
| 5306 | 3871 | $response_body = wp_remote_retrieve_body($response); |
| 5307 | - //error_log("MXCHAT DEBUG: Raw Pinecone response length: " . strlen($response_body)); | |
| 3872 | + error_log("MXCHAT DEBUG: Raw Pinecone response length: " . strlen($response_body)); | |
| 5308 | 3873 | |
| 5309 | 3874 | $results = json_decode($response_body, true); |
| 5310 | 3875 | |
| 5311 | 3876 | if (json_last_error() !== JSON_ERROR_NONE) { |
| 5312 | - //error_log("MXCHAT DEBUG ERROR: JSON decode error: " . json_last_error_msg()); | |
| 5313 | - //error_log("MXCHAT DEBUG: First 500 chars of response: " . substr($response_body, 0, 500)); | |
| 5314 | - // Store empty array for valid URLs | |
| 5315 | - $this->current_valid_urls = []; | |
| 3877 | + error_log("MXCHAT DEBUG ERROR: JSON decode error: " . json_last_error_msg()); | |
| 3878 | + error_log("MXCHAT DEBUG: First 500 chars of response: " . substr($response_body, 0, 500)); | |
| 5316 | 3879 | return ''; |
| 5317 | 3880 | } |
| 5318 | 3881 | |
| 5319 | - //error_log("MXCHAT DEBUG: Pinecone response structure:"); | |
| 5320 | - //error_log(" - Has 'matches' key: " . (isset($results['matches']) ? 'yes' : 'no')); | |
| 5321 | - //error_log(" - Has 'namespace' key: " . (isset($results['namespace']) ? 'yes (' . $results['namespace'] . ')' : 'no')); | |
| 3882 | + error_log("MXCHAT DEBUG: Pinecone response structure:"); | |
| 3883 | + error_log(" - Has 'matches' key: " . (isset($results['matches']) ? 'yes' : 'no')); | |
| 3884 | + error_log(" - Has 'namespace' key: " . (isset($results['namespace']) ? 'yes (' . $results['namespace'] . ')' : 'no')); | |
| 5322 | 3885 | |
| 5323 | 3886 | if (empty($results['matches'])) { |
| 5324 | - //error_log("MXCHAT DEBUG: No matches found in Pinecone response"); | |
| 5325 | - //error_log("MXCHAT DEBUG: Response keys: " . implode(', ', array_keys($results))); | |
| 5326 | - // Store empty array for valid URLs | |
| 5327 | - $this->current_valid_urls = []; | |
| 3887 | + error_log("MXCHAT DEBUG: No matches found in Pinecone response"); | |
| 3888 | + error_log("MXCHAT DEBUG: Response keys: " . implode(', ', array_keys($results))); | |
| 5328 | 3889 | return ''; |
| 5329 | 3890 | } |
| 5330 | 3891 | |
| 5331 | - //error_log("MXCHAT DEBUG: Found " . count($results['matches']) . " matches in Pinecone"); | |
| 3892 | + error_log("MXCHAT DEBUG: Found " . count($results['matches']) . " matches in Pinecone"); | |
| 5332 | 3893 | |
| 5333 | 3894 | // Log first match details for debugging |
| 5334 | 3895 | if (!empty($results['matches'][0])) { |
| 5335 | 3896 | $first_match = $results['matches'][0]; |
| 5336 | - //error_log("MXCHAT DEBUG: First match details:"); | |
| 5337 | - //error_log(" - Score: " . ($first_match['score'] ?? 'no score')); | |
| 5338 | - //error_log(" - Has metadata: " . (isset($first_match['metadata']) ? 'yes' : 'no')); | |
| 3897 | + error_log("MXCHAT DEBUG: First match details:"); | |
| 3898 | + error_log(" - Score: " . ($first_match['score'] ?? 'no score')); | |
| 3899 | + error_log(" - Has metadata: " . (isset($first_match['metadata']) ? 'yes' : 'no')); | |
| 5339 | 3900 | if (isset($first_match['metadata'])) { |
| 5340 | - //error_log(" - Metadata keys: " . implode(', ', array_keys($first_match['metadata']))); | |
| 3901 | + error_log(" - Metadata keys: " . implode(', ', array_keys($first_match['metadata']))); | |
| 5341 | 3902 | } |
| 5342 | 3903 | } |
| 5343 | 3904 | |
| 5344 | 3905 | // Initialize the final content |
| @@ -5344,181 +3905,45 @@ | ||
| 5344 | 3905 | // Initialize the final content |
| 5345 | 3906 | $content = ''; |
| 5346 | 3907 | $matches_used = 0; |
| 5347 | 3908 | $matches_used_for_context = []; |
| 5348 | - $total_chunks_used = 0; | |
| 5349 | - $max_total_chunks = isset($current_options['rag_chunks_limit']) ? intval($current_options['rag_chunks_limit']) : 15; | |
| 5350 | - if ($max_total_chunks < 8) $max_total_chunks = 8; | |
| 5351 | - if ($max_total_chunks > 20) $max_total_chunks = 20; | |
| 5352 | - $max_chunks_per_source = 5; // Cap per individual source to limit token usage | |
| 5353 | - | |
| 5354 | - // Check if citation links are enabled (default to 'on' for backwards compatibility) | |
| 5355 | - // Use fresh options to ensure we get the latest setting value | |
| 5356 | - $fresh_options = get_option('mxchat_options', []); | |
| 5357 | - $citation_links_enabled = isset($fresh_options['citation_links_toggle']) ? ($fresh_options['citation_links_toggle'] === 'on') : true; | |
| 5358 | - | |
| 5359 | - // NEW CHUNKING LOGIC: Group results by source_url for chunk reassembly | |
| 5360 | - $url_groups = array(); | |
| 5361 | - | |
| 3909 | + | |
| 3910 | + // Process each match for actual content generation (lazy role checking) | |
| 5362 | 3911 | foreach ($results['matches'] as $index => $match) { |
| 5363 | 3912 | // Skip if similarity is below threshold |
| 5364 | 3913 | if ($match['score'] < $similarity_threshold) { |
| 5365 | 3914 | continue; |
| 5366 | 3915 | } |
| 5367 | - | |
| 5368 | - $metadata = $match['metadata'] ?? array(); | |
| 5369 | - $source_url = $metadata['source_url'] ?? ''; | |
| 5370 | - $match_id = $match['id'] ?? ''; | |
| 5371 | - | |
| 5372 | - // LAZY ROLE CHECK: Only check role for content we're actually considering | |
| 5373 | - $role_restriction = $this->get_single_vector_role($match_id, $metadata); | |
| 5374 | - $has_access = $knowledge_manager->mxchat_user_has_content_access($role_restriction); | |
| 5375 | - | |
| 5376 | - // Skip if user doesn't have access | |
| 5377 | - if (!$has_access) { | |
| 5378 | - continue; | |
| 5379 | - } | |
| 5380 | - | |
| 5381 | - // Use a unique key for manual entries without a source URL | |
| 5382 | - $group_key = !empty($source_url) ? $source_url : '_manual_' . $match_id; | |
| 5383 | - | |
| 5384 | - // Group by source URL (or unique key for manual entries) | |
| 5385 | - if (!isset($url_groups[$group_key])) { | |
| 5386 | - $url_groups[$group_key] = array( | |
| 5387 | - 'source_url' => $source_url, | |
| 5388 | - 'best_score' => 0, | |
| 5389 | - 'is_chunked' => isset($metadata['is_chunked']) && $metadata['is_chunked'], | |
| 5390 | - 'chunks' => array(), | |
| 5391 | - 'single_text' => '' | |
| 5392 | - ); | |
| 5393 | - } | |
| 5394 | - | |
| 5395 | - // Track best score for this group | |
| 5396 | - if ($match['score'] > $url_groups[$group_key]['best_score']) { | |
| 5397 | - $url_groups[$group_key]['best_score'] = $match['score']; | |
| 5398 | - } | |
| 5399 | - | |
| 5400 | - // Store chunk info or single text | |
| 5401 | - if ($url_groups[$group_key]['is_chunked']) { | |
| 5402 | - $url_groups[$group_key]['chunks'][] = array( | |
| 5403 | - 'id' => $match_id, | |
| 5404 | - 'score' => $match['score'], | |
| 5405 | - 'chunk_index' => $metadata['chunk_index'] ?? 0, | |
| 5406 | - 'text' => $metadata['text'] ?? '' | |
| 5407 | - ); | |
| 5408 | - } else { | |
| 5409 | - // Non-chunked content - just store the text | |
| 5410 | - $url_groups[$group_key]['single_text'] = $metadata['text'] ?? ''; | |
| 5411 | - $url_groups[$group_key]['single_id'] = $match_id; | |
| 5412 | - } | |
| 5413 | - } | |
| 5414 | - | |
| 5415 | - // Sort URL groups by best score (highest first) | |
| 5416 | - uasort($url_groups, function($a, $b) { | |
| 5417 | - return $b['best_score'] <=> $a['best_score']; | |
| 5418 | - }); | |
| 5419 | - | |
| 5420 | - // Get RAG sources limit from options (default 6, min 3, max 10) | |
| 5421 | - $rag_sources_limit = isset($current_options['rag_sources_limit']) ? intval($current_options['rag_sources_limit']) : 3; | |
| 5422 | - if ($rag_sources_limit < 3) $rag_sources_limit = 3; | |
| 5423 | - if ($rag_sources_limit > 10) $rag_sources_limit = 10; | |
| 5424 | - | |
| 5425 | - // Take top N unique URLs based on user setting | |
| 5426 | - $top_urls = array_slice($url_groups, 0, $rag_sources_limit, true); | |
| 5427 | - | |
| 5428 | - // Track which match IDs are actually used for context | |
| 5429 | - foreach ($top_urls as $group) { | |
| 5430 | - if ($group['is_chunked']) { | |
| 5431 | - foreach ($group['chunks'] as $chunk) { | |
| 5432 | - $matches_used_for_context[] = $chunk['id']; | |
| 5433 | - } | |
| 5434 | - } elseif (!empty($group['single_id'])) { | |
| 5435 | - $matches_used_for_context[] = $group['single_id']; | |
| 5436 | - } | |
| 5437 | - } | |
| 5438 | - | |
| 5439 | - // Build content from top sources | |
| 5440 | - foreach ($top_urls as $group_key => $group) { | |
| 5441 | - $source_url = $group['source_url']; // Use actual source_url, not the group key | |
| 5442 | - | |
| 5443 | - // Stop if we've hit the total chunk limit | |
| 5444 | - if ($total_chunks_used >= $max_total_chunks) { | |
| 3916 | + | |
| 3917 | + // Limit to top 5 matches above threshold | |
| 3918 | + if ($matches_used >= 5) { | |
| 5445 | 3919 | break; |
| 5446 | 3920 | } |
| 5447 | - | |
| 5448 | - $full_text = ''; | |
| 5449 | - $chunks_in_this_source = 1; // Default for non-chunked content | |
| 5450 | - | |
| 5451 | - if ($group['is_chunked']) { | |
| 5452 | - // Calculate how many chunks we can still use (respect both total and per-source caps) | |
| 5453 | - $chunks_remaining = min($max_chunks_per_source, $max_total_chunks - $total_chunks_used); | |
| 5454 | - | |
| 5455 | - // Fetch chunks for this URL with limit | |
| 5456 | - $full_text = $this->reassemble_chunks_from_pinecone($source_url, $bot_config, $chunks_remaining, $chunks_in_this_source); | |
| 5457 | - | |
| 5458 | - // If fetching all chunks fails, fall back to matched chunks | |
| 5459 | - if (empty($full_text)) { | |
| 5460 | - // Sort matched chunks by index and concatenate | |
| 5461 | - usort($group['chunks'], function($a, $b) { | |
| 5462 | - return $a['chunk_index'] <=> $b['chunk_index']; | |
| 5463 | - }); | |
| 5464 | - | |
| 5465 | - $chunk_texts = array(); | |
| 5466 | - $chunks_in_this_source = 0; | |
| 5467 | - foreach ($group['chunks'] as $chunk) { | |
| 5468 | - if ($total_chunks_used + $chunks_in_this_source >= $max_total_chunks) { | |
| 5469 | - break; | |
| 5470 | - } | |
| 5471 | - $chunk_texts[] = $chunk['text']; | |
| 5472 | - $chunks_in_this_source++; | |
| 5473 | - } | |
| 5474 | - $full_text = implode("\n\n", $chunk_texts); | |
| 3921 | + | |
| 3922 | + if (!empty($match['metadata']['text'])) { | |
| 3923 | + // LAZY ROLE CHECK: Only check role for content we're actually considering | |
| 3924 | + $match_id = $match['id'] ?? ''; | |
| 3925 | + $role_restriction = $this->get_single_vector_role($match_id, $match['metadata']); | |
| 3926 | + $has_access = $knowledge_manager->mxchat_user_has_content_access($role_restriction); | |
| 3927 | + | |
| 3928 | + // Skip if user doesn't have access | |
| 3929 | + if (!$has_access) { | |
| 3930 | + continue; | |
| 5475 | 3931 | } |
| 5476 | - } else { | |
| 5477 | - $full_text = $group['single_text']; | |
| 5478 | - $chunks_in_this_source = 1; | |
| 5479 | - } | |
| 5480 | - | |
| 5481 | - if (!empty($full_text)) { | |
| 5482 | - // Strip URLs from content if citation links are disabled | |
| 5483 | - if (!$citation_links_enabled) { | |
| 5484 | - $full_text = preg_replace('#\bhttps?://[^\s<>"\']+#i', '', $full_text); | |
| 5485 | - $full_text = preg_replace('/\s+/', ' ', trim($full_text)); // Clean up extra spaces | |
| 3932 | + | |
| 3933 | + // User has access - add to content | |
| 3934 | + $content .= "## Reference " . ($matches_used + 1) . " ##\n"; | |
| 3935 | + $content .= $match['metadata']['text'] . "\n\n"; | |
| 3936 | + | |
| 3937 | + if (!empty($match['metadata']['source_url'])) { | |
| 3938 | + $content .= "URL: " . $match['metadata']['source_url'] . "\n\n"; | |
| 5486 | 3939 | } |
| 5487 | - | |
| 5488 | - // Use numbered reference for URL-based entries, plain info label for manual entries | |
| 5489 | - if (!empty($source_url) && $source_url !== '#') { | |
| 5490 | - $matches_used++; | |
| 5491 | - $content .= "## Reference " . $matches_used . " ##\n"; | |
| 5492 | - $content .= $full_text . "\n\n"; | |
| 5493 | - | |
| 5494 | - // Only include citation URLs if citation links are enabled | |
| 5495 | - if ($citation_links_enabled) { | |
| 5496 | - $valid_urls[] = $source_url; | |
| 5497 | - $content .= "URL: " . $source_url . "\n\n"; | |
| 5498 | - } | |
| 5499 | - } else { | |
| 5500 | - // Manual entry — no reference number, no citation | |
| 5501 | - $content .= "## Information ##\n"; | |
| 5502 | - $content .= $full_text . "\n\n"; | |
| 5503 | - } | |
| 5504 | - | |
| 5505 | - // Extract any URLs from the text content itself (only if citation links enabled) | |
| 5506 | - if ($citation_links_enabled) { | |
| 5507 | - preg_match_all( | |
| 5508 | - '#\bhttps?://[^\s<>"\']+#i', | |
| 5509 | - $full_text, | |
| 5510 | - $content_urls | |
| 5511 | - ); | |
| 5512 | - if (!empty($content_urls[0])) { | |
| 5513 | - $valid_urls = array_merge($valid_urls, $content_urls[0]); | |
| 5514 | - } | |
| 5515 | - } | |
| 5516 | - | |
| 5517 | - $total_chunks_used += $chunks_in_this_source; | |
| 3940 | + | |
| 3941 | + $matches_used_for_context[] = $match['id'] ?? $index; | |
| 3942 | + $matches_used++; | |
| 5518 | 3943 | } |
| 5519 | 3944 | } |
| 5520 | - | |
| 3945 | + | |
| 5521 | 3946 | // Process ALL matches for testing data (top 10) - with role checking for testing display |
| 5522 | 3947 | $all_matches = []; |
| 5523 | 3948 | foreach ($results['matches'] as $index => $match) { |
| 5524 | 3949 | if ($index >= 10) break; // Limit to top 10 for testing |
| @@ -5538,19 +3963,9 @@ | ||
| 5538 | 3963 | $source_display = substr(trim($content_preview), 0, 50) . '...'; |
| 5539 | 3964 | } |
| 5540 | 3965 | |
| 5541 | 3966 | $match_id_for_display = $match['id'] ?? $index; |
| 5542 | - | |
| 5543 | - // Check for chunk metadata in Pinecone | |
| 5544 | - $is_chunk = isset($match['metadata']['is_chunked']) && $match['metadata']['is_chunked']; | |
| 5545 | - $chunk_index = isset($match['metadata']['chunk_index']) ? intval($match['metadata']['chunk_index']) : null; | |
| 5546 | - $total_chunks = isset($match['metadata']['total_chunks']) ? intval($match['metadata']['total_chunks']) : null; | |
| 5547 | - | |
| 5548 | - // Also detect chunk from vector ID pattern: {hash}_chunk_{index} | |
| 5549 | - if (!$is_chunk && MxChat_Chunker::is_chunk_vector_id($match_id_for_display)) { | |
| 5550 | - $is_chunk = true; | |
| 5551 | - } | |
| 5552 | - | |
| 3967 | + | |
| 5553 | 3968 | $all_matches[] = [ |
| 5554 | 3969 | 'document_id' => $match_id_for_display, |
| 5555 | 3970 | 'similarity' => $match['score'], |
| 5556 | 3971 | 'similarity_percentage' => round($match['score'] * 100, 2), |
| @@ -5559,12 +3974,9 @@ | ||
| 5559 | 3974 | 'content_preview' => substr(strip_tags($match['metadata']['text'] ?? ''), 0, 100) . '...', |
| 5560 | 3975 | 'used_for_context' => in_array($match_id_for_display, $matches_used_for_context), |
| 5561 | 3976 | 'role_restriction' => $role_restriction, |
| 5562 | 3977 | 'has_access' => $has_access, |
| 5563 | - 'filtered_out' => !$has_access, | |
| 5564 | - 'is_chunk' => $is_chunk, | |
| 5565 | - 'chunk_index' => $chunk_index, | |
| 5566 | - 'total_chunks' => $total_chunks | |
| 3978 | + 'filtered_out' => !$has_access | |
| 5567 | 3979 | ]; |
| 5568 | 3980 | } |
| 5569 | 3981 | |
| 5570 | 3982 | // Store for testing panel |
| @@ -5569,43 +3981,25 @@ | ||
| 5569 | 3981 | |
| 5570 | 3982 | // Store for testing panel |
| 5571 | 3983 | $this->last_similarity_analysis['top_matches'] = $all_matches; |
| 5572 | 3984 | $this->last_similarity_analysis['total_checked'] = count($results['matches']); |
| 5573 | - $this->last_similarity_analysis['sources_used'] = $matches_used; | |
| 5574 | - $this->last_similarity_analysis['total_chunks_used'] = $total_chunks_used; | |
| 5575 | - | |
| 5576 | - // NEW: Store unique valid URLs for validation | |
| 5577 | - $this->current_valid_urls = array_unique($valid_urls); | |
| 5578 | - | |
| 5579 | - // Allow add-ons to act on similarity results (e.g. WooCommerce product card display) | |
| 5580 | - do_action('mxchat_similarity_results', $this->last_similarity_analysis['top_matches'], $bot_id); | |
| 5581 | - | |
| 3985 | + | |
| 5582 | 3986 | // Add response guidelines |
| 5583 | 3987 | if ($matches_used === 0) { |
| 5584 | 3988 | $content = "No reference information was found for this query.\n\n"; |
| 5585 | 3989 | } else { |
| 5586 | - // Build response guidelines based on citation links setting | |
| 5587 | 3990 | $content .= "\n## Response Guidelines ##\n" . |
| 5588 | 3991 | "You are an AI Chatbot. Answer naturally and helpfully using only the information from the references above. " . |
| 5589 | 3992 | "Be conversational and friendly, but never mention your knowledge base or training data. " . |
| 5590 | 3993 | "If you don't have specific information or are uncertain about any details, it's always " . |
| 5591 | 3994 | "better to honestly say you don't know rather than making up or guessing at answers. " . |
| 5592 | - "When information is incomplete, let them know you are unsure.\n\n"; | |
| 5593 | - | |
| 5594 | - // Only add hyperlink instructions if citation links are enabled | |
| 5595 | - if ($citation_links_enabled) { | |
| 5596 | - $content .= "CRITICAL: When creating hyperlinks, always use proper markdown format with descriptive text: " . | |
| 5597 | - "[descriptive text](url). NEVER use empty brackets like [](url). The text in brackets must describe what the link is about. " . | |
| 5598 | - "Only cite references that have a URL. Do not cite or add source labels to Information sections that have no URL."; | |
| 5599 | - } else { | |
| 5600 | - $content .= "IMPORTANT: Do not include any citation links, source URLs, or hyperlinks in your responses. " . | |
| 5601 | - "Simply provide helpful answers based on the reference information without citing sources."; | |
| 5602 | - } | |
| 3995 | + "When information is incomplete, let them know you are unsure."; | |
| 5603 | 3996 | } |
| 5604 | - | |
| 3997 | + | |
| 5605 | 3998 | return trim($content); |
| 5606 | 3999 | } |
| 5607 | 4000 | |
| 4001 | + | |
| 5608 | 4002 | /** |
| 5609 | 4003 | * Get role restriction for a single vector (with caching) |
| 5610 | 4004 | */ |
| 5611 | 4005 | private function get_single_vector_role($vector_id, $metadata = array()) { |
| @@ -5642,518 +4036,12 @@ | ||
| 5642 | 4036 | } |
| 5643 | 4037 | |
| 5644 | 4038 | // Cache individual role for 1 hour |
| 5645 | 4039 | wp_cache_set($cache_key, $role_restriction, 'mxchat_vector_roles', 3600); |
| 5646 | - | |
| 4040 | + | |
| 5647 | 4041 | return $role_restriction; |
| 5648 | 4042 | } |
| 5649 | 4043 | |
| 5650 | -/** | |
| 5651 | - * Fetch and reassemble all chunks for a URL from Pinecone | |
| 5652 | - * | |
| 5653 | - * @param string $source_url The source URL to fetch chunks for | |
| 5654 | - * @param array $bot_config Bot-specific Pinecone configuration | |
| 5655 | - * @return string Reassembled content from all chunks | |
| 5656 | - */ | |
| 5657 | -private function reassemble_chunks_from_pinecone($source_url, $bot_config, $max_chunks = 0, &$chunk_count = 0) { | |
| 5658 | - $api_key = $bot_config['api_key'] ?? ''; | |
| 5659 | - $host = $bot_config['host'] ?? ''; | |
| 5660 | - $namespace = $bot_config['namespace'] ?? ''; | |
| 5661 | - | |
| 5662 | - if (empty($host) || empty($api_key)) { | |
| 5663 | - $chunk_count = 0; | |
| 5664 | - return ''; | |
| 5665 | - } | |
| 5666 | - | |
| 5667 | - $base_hash = md5($source_url); | |
| 5668 | - | |
| 5669 | - // Use Pinecone list API to find all chunk vectors with this prefix | |
| 5670 | - $list_url = "https://{$host}/vectors/list"; | |
| 5671 | - | |
| 5672 | - // Limit to max_chunks if specified, otherwise fetch up to 100 | |
| 5673 | - $fetch_limit = ($max_chunks > 0 && $max_chunks < 100) ? $max_chunks : 100; | |
| 5674 | - | |
| 5675 | - $list_body = array( | |
| 5676 | - 'prefix' => $base_hash . '_chunk_', | |
| 5677 | - 'limit' => $fetch_limit | |
| 5678 | - ); | |
| 5679 | - | |
| 5680 | - if (!empty($namespace)) { | |
| 5681 | - $list_body['namespace'] = $namespace; | |
| 5682 | - } | |
| 5683 | - | |
| 5684 | - $list_response = wp_remote_post($list_url, array( | |
| 5685 | - 'headers' => array( | |
| 5686 | - 'Api-Key' => $api_key, | |
| 5687 | - 'accept' => 'application/json', | |
| 5688 | - 'content-type' => 'application/json' | |
| 5689 | - ), | |
| 5690 | - 'body' => wp_json_encode($list_body), | |
| 5691 | - 'timeout' => 30 | |
| 5692 | - )); | |
| 5693 | - | |
| 5694 | - if (is_wp_error($list_response)) { | |
| 5695 | - //error_log('[MXCHAT-CHUNK] List API error: ' . $list_response->get_error_message()); | |
| 5696 | - return ''; | |
| 5697 | - } | |
| 5698 | - | |
| 5699 | - $list_data = json_decode(wp_remote_retrieve_body($list_response), true); | |
| 5700 | - | |
| 5701 | - if (empty($list_data['vectors'])) { | |
| 5702 | - //error_log('[MXCHAT-CHUNK] No chunk vectors found for URL: ' . $source_url); | |
| 5703 | - return ''; | |
| 5704 | - } | |
| 5705 | - | |
| 5706 | - // Extract vector IDs | |
| 5707 | - $vector_ids = array(); | |
| 5708 | - foreach ($list_data['vectors'] as $vector) { | |
| 5709 | - if (isset($vector['id'])) { | |
| 5710 | - $vector_ids[] = $vector['id']; | |
| 5711 | - } | |
| 5712 | - } | |
| 5713 | - | |
| 5714 | - if (empty($vector_ids)) { | |
| 5715 | - return ''; | |
| 5716 | - } | |
| 5717 | - | |
| 5718 | - // Fetch all chunk content | |
| 5719 | - $fetch_url = "https://{$host}/vectors/fetch"; | |
| 5720 | - | |
| 5721 | - $fetch_body = array( | |
| 5722 | - 'ids' => $vector_ids | |
| 5723 | - ); | |
| 5724 | - | |
| 5725 | - if (!empty($namespace)) { | |
| 5726 | - $fetch_body['namespace'] = $namespace; | |
| 5727 | - } | |
| 5728 | - | |
| 5729 | - $fetch_response = wp_remote_post($fetch_url, array( | |
| 5730 | - 'headers' => array( | |
| 5731 | - 'Api-Key' => $api_key, | |
| 5732 | - 'accept' => 'application/json', | |
| 5733 | - 'content-type' => 'application/json' | |
| 5734 | - ), | |
| 5735 | - 'body' => wp_json_encode($fetch_body), | |
| 5736 | - 'timeout' => 30 | |
| 5737 | - )); | |
| 5738 | - | |
| 5739 | - if (is_wp_error($fetch_response)) { | |
| 5740 | - //error_log('[MXCHAT-CHUNK] Fetch API error: ' . $fetch_response->get_error_message()); | |
| 5741 | - return ''; | |
| 5742 | - } | |
| 5743 | - | |
| 5744 | - $fetch_data = json_decode(wp_remote_retrieve_body($fetch_response), true); | |
| 5745 | - | |
| 5746 | - if (empty($fetch_data['vectors'])) { | |
| 5747 | - return ''; | |
| 5748 | - } | |
| 5749 | - | |
| 5750 | - // Sort chunks by index and reassemble | |
| 5751 | - $chunks = array(); | |
| 5752 | - foreach ($fetch_data['vectors'] as $id => $vector) { | |
| 5753 | - $metadata = $vector['metadata'] ?? array(); | |
| 5754 | - $chunk_index = $metadata['chunk_index'] ?? 0; | |
| 5755 | - $text = $metadata['text'] ?? ''; | |
| 5756 | - | |
| 5757 | - // Store chunk with its index | |
| 5758 | - $chunks[$chunk_index] = $text; | |
| 5759 | - } | |
| 5760 | - | |
| 5761 | - // Sort by chunk index | |
| 5762 | - ksort($chunks); | |
| 5763 | - | |
| 5764 | - // Apply chunk limit if specified | |
| 5765 | - if ($max_chunks > 0 && count($chunks) > $max_chunks) { | |
| 5766 | - $chunks = array_slice($chunks, 0, $max_chunks, true); | |
| 5767 | - } | |
| 5768 | - | |
| 5769 | - // Store actual chunk count | |
| 5770 | - $chunk_count = count($chunks); | |
| 5771 | - | |
| 5772 | - // Reassemble content | |
| 5773 | - return implode("\n\n", $chunks); | |
| 5774 | -} | |
| 5775 | - | |
| 5776 | -/** | |
| 5777 | - * Search for relevant content using OpenAI Vector Store (File Search) | |
| 5778 | - * | |
| 5779 | - * @param string $user_query The user's query text | |
| 5780 | - * @param string $bot_id The bot ID | |
| 5781 | - * @param array $vectorstore_config Vector Store configuration | |
| 5782 | - * @return string Formatted context string with references | |
| 5783 | - */ | |
| 5784 | -private function find_relevant_content_openai_vectorstore($user_query, $bot_id = 'default', $vectorstore_config = array()) { | |
| 5785 | - //error_log("MXCHAT DEBUG: find_relevant_content_openai_vectorstore called"); | |
| 5786 | - //error_log(" - bot_id: " . $bot_id); | |
| 5787 | - //error_log(" - user_query length: " . strlen($user_query)); | |
| 5788 | - | |
| 5789 | - // Get OpenAI API key | |
| 5790 | - $mxchat_options = get_option('mxchat_options', array()); | |
| 5791 | - $api_key = $mxchat_options['api_key'] ?? ''; | |
| 5792 | - | |
| 5793 | - // Reset vectorstore error tracking | |
| 5794 | - $this->last_vectorstore_error = null; | |
| 5795 | - | |
| 5796 | - if (empty($api_key)) { | |
| 5797 | - //error_log("MXCHAT DEBUG ERROR: OpenAI API key not configured"); | |
| 5798 | - $this->last_vectorstore_error = 'Vector Store search failed: OpenAI API key is not configured.'; | |
| 5799 | - $this->current_valid_urls = []; | |
| 5800 | - return ''; | |
| 5801 | - } | |
| 5802 | - | |
| 5803 | - // Get Vector Store configuration | |
| 5804 | - if (empty($vectorstore_config)) { | |
| 5805 | - $vectorstore_config = $this->get_bot_vectorstore_config($bot_id); | |
| 5806 | - } | |
| 5807 | - | |
| 5808 | - $vectorstore_ids_string = $vectorstore_config['vectorstore_ids'] ?? ''; | |
| 5809 | - $max_results = $vectorstore_config['max_results'] ?? 5; | |
| 5810 | - | |
| 5811 | - if (empty($vectorstore_ids_string)) { | |
| 5812 | - //error_log("MXCHAT DEBUG ERROR: No Vector Store IDs configured"); | |
| 5813 | - $this->last_vectorstore_error = 'Vector Store search failed: No Vector Store IDs are configured for this bot.'; | |
| 5814 | - $this->current_valid_urls = []; | |
| 5815 | - return ''; | |
| 5816 | - } | |
| 5817 | - | |
| 5818 | - // Parse Vector Store IDs | |
| 5819 | - $vectorstore_ids = array_map('trim', explode(',', $vectorstore_ids_string)); | |
| 5820 | - $vectorstore_ids = array_filter($vectorstore_ids); // Remove empty values | |
| 5821 | - | |
| 5822 | - //error_log("MXCHAT DEBUG: Vector Store IDs: " . implode(', ', $vectorstore_ids)); | |
| 5823 | - //error_log("MXCHAT DEBUG: Max results: " . $max_results); | |
| 5824 | - | |
| 5825 | - // Initialize similarity analysis storage | |
| 5826 | - $this->last_similarity_analysis = [ | |
| 5827 | - 'knowledge_base_type' => 'OpenAI Vector Store', | |
| 5828 | - 'bot_id' => $bot_id, | |
| 5829 | - 'vectorstore_ids' => $vectorstore_ids, | |
| 5830 | - 'top_matches' => [], | |
| 5831 | - 'threshold_used' => 0, | |
| 5832 | - 'total_checked' => 0 | |
| 5833 | - ]; | |
| 5834 | - | |
| 5835 | - $valid_urls = []; | |
| 5836 | - | |
| 5837 | - // Get the selected model | |
| 5838 | - $bot_options = $this->get_bot_options($bot_id); | |
| 5839 | - $current_options = !empty($bot_options) ? $bot_options : $mxchat_options; | |
| 5840 | - $selected_model = $current_options['model'] ?? 'gpt-5.1-chat-latest'; | |
| 5841 | - | |
| 5842 | - // Verify it's an OpenAI model | |
| 5843 | - if (!$this->is_openai_chat_model($selected_model)) { | |
| 5844 | - //error_log("MXCHAT DEBUG ERROR: Vector Store search requires OpenAI model. Current: " . $selected_model); | |
| 5845 | - $this->last_vectorstore_error = 'Vector Store search requires an OpenAI model. Current model: ' . $selected_model; | |
| 5846 | - $this->current_valid_urls = []; | |
| 5847 | - return ''; | |
| 5848 | - } | |
| 5849 | - | |
| 5850 | - // Use OpenAI Responses API with file_search tool | |
| 5851 | - $request_body = array( | |
| 5852 | - 'model' => $selected_model, | |
| 5853 | - 'input' => $user_query, | |
| 5854 | - 'tools' => array( | |
| 5855 | - array( | |
| 5856 | - 'type' => 'file_search', | |
| 5857 | - 'vector_store_ids' => $vectorstore_ids, | |
| 5858 | - 'max_num_results' => intval($max_results) | |
| 5859 | - ) | |
| 5860 | - ), | |
| 5861 | - 'include' => array('output[*].file_search_call.search_results') | |
| 5862 | - ); | |
| 5863 | - | |
| 5864 | - //error_log("MXCHAT VECTORSTORE: ========== REQUEST START =========="); | |
| 5865 | - //error_log("MXCHAT VECTORSTORE: Model: " . $selected_model); | |
| 5866 | - //error_log("MXCHAT VECTORSTORE: Query: " . substr($user_query, 0, 200)); | |
| 5867 | - //error_log("MXCHAT VECTORSTORE: Vector Store IDs: " . implode(', ', $vectorstore_ids)); | |
| 5868 | - //error_log("MXCHAT VECTORSTORE: Max Results: " . $max_results); | |
| 5869 | - //error_log("MXCHAT VECTORSTORE: Request body: " . wp_json_encode($request_body)); | |
| 5870 | - | |
| 5871 | - $response = wp_remote_post('https://api.openai.com/v1/responses', array( | |
| 5872 | - 'headers' => array( | |
| 5873 | - 'Authorization' => 'Bearer ' . $api_key, | |
| 5874 | - 'Content-Type' => 'application/json' | |
| 5875 | - ), | |
| 5876 | - 'body' => wp_json_encode($request_body), | |
| 5877 | - 'timeout' => 60 | |
| 5878 | - )); | |
| 5879 | - | |
| 5880 | - if (is_wp_error($response)) { | |
| 5881 | - //error_log("MXCHAT VECTORSTORE ERROR: WP Error: " . $response->get_error_message()); | |
| 5882 | - $this->last_vectorstore_error = 'Vector Store API request failed: ' . $response->get_error_message(); | |
| 5883 | - $this->current_valid_urls = []; | |
| 5884 | - return ''; | |
| 5885 | - } | |
| 5886 | - | |
| 5887 | - $response_code = wp_remote_retrieve_response_code($response); | |
| 5888 | - //error_log("MXCHAT VECTORSTORE: Response code: " . $response_code); | |
| 5889 | - | |
| 5890 | - $response_body = wp_remote_retrieve_body($response); | |
| 5891 | - //error_log("MXCHAT VECTORSTORE: Raw response (first 2000 chars): " . substr($response_body, 0, 2000)); | |
| 5892 | - | |
| 5893 | - if ($response_code !== 200) { | |
| 5894 | - //error_log("MXCHAT VECTORSTORE ERROR: API error response: " . $response_body); | |
| 5895 | - $api_error_detail = ''; | |
| 5896 | - $decoded_error = json_decode($response_body, true); | |
| 5897 | - if (isset($decoded_error['error']['message'])) { | |
| 5898 | - $api_error_detail = $decoded_error['error']['message']; | |
| 5899 | - } | |
| 5900 | - $this->last_vectorstore_error = 'Vector Store API returned HTTP ' . $response_code . ($api_error_detail ? ': ' . $api_error_detail : ''); | |
| 5901 | - $this->current_valid_urls = []; | |
| 5902 | - return ''; | |
| 5903 | - } | |
| 5904 | - $result = json_decode($response_body, true); | |
| 5905 | - | |
| 5906 | - if (json_last_error() !== JSON_ERROR_NONE) { | |
| 5907 | - //error_log("MXCHAT VECTORSTORE ERROR: JSON decode error: " . json_last_error_msg()); | |
| 5908 | - $this->last_vectorstore_error = 'Vector Store response could not be parsed: ' . json_last_error_msg(); | |
| 5909 | - $this->current_valid_urls = []; | |
| 5910 | - return ''; | |
| 5911 | - } | |
| 5912 | - | |
| 5913 | - // Debug: Log the structure of the result | |
| 5914 | - //error_log("MXCHAT VECTORSTORE: Result keys: " . implode(', ', array_keys($result))); | |
| 5915 | - if (isset($result['output'])) { | |
| 5916 | - //error_log("MXCHAT VECTORSTORE: Output count: " . count($result['output'])); | |
| 5917 | - foreach ($result['output'] as $idx => $out) { | |
| 5918 | - //error_log("MXCHAT VECTORSTORE: Output[$idx] type: " . ($out['type'] ?? 'unknown')); | |
| 5919 | - //error_log("MXCHAT VECTORSTORE: Output[$idx] keys: " . implode(', ', array_keys($out))); | |
| 5920 | - } | |
| 5921 | - } else { | |
| 5922 | - //error_log("MXCHAT VECTORSTORE: No 'output' key in result!"); | |
| 5923 | - } | |
| 5924 | - | |
| 5925 | - // Extract file search results from the response | |
| 5926 | - $content = ''; | |
| 5927 | - $matches_used = 0; | |
| 5928 | - $all_matches = []; | |
| 5929 | - | |
| 5930 | - // The Responses API returns output array with tool results | |
| 5931 | - if (isset($result['output']) && is_array($result['output'])) { | |
| 5932 | - foreach ($result['output'] as $output_item) { | |
| 5933 | - // Look for file_search_call results | |
| 5934 | - if (isset($output_item['type']) && $output_item['type'] === 'file_search_call') { | |
| 5935 | - //error_log("MXCHAT VECTORSTORE: Found file_search_call output item"); | |
| 5936 | - //error_log("MXCHAT VECTORSTORE: file_search_call keys: " . implode(', ', array_keys($output_item))); | |
| 5937 | - | |
| 5938 | - // Check for search_results in the output item directly | |
| 5939 | - $search_results = $output_item['search_results'] ?? $output_item['results'] ?? []; | |
| 5940 | - //error_log("MXCHAT VECTORSTORE: Search results count: " . count($search_results)); | |
| 5941 | - | |
| 5942 | - if (empty($search_results)) { | |
| 5943 | - //error_log("MXCHAT VECTORSTORE: No search results found in file_search_call"); | |
| 5944 | - //error_log("MXCHAT VECTORSTORE: file_search_call content: " . wp_json_encode($output_item)); | |
| 5945 | - } | |
| 5946 | - | |
| 5947 | - foreach ($search_results as $index => $search_result) { | |
| 5948 | - $filename = $search_result['filename'] ?? ''; | |
| 5949 | - $score = $search_result['score'] ?? 0; | |
| 5950 | - $text_content = ''; | |
| 5951 | - | |
| 5952 | - // Extract text content from the result | |
| 5953 | - // The text can be directly on the result OR nested under content array | |
| 5954 | - if (isset($search_result['text']) && !empty($search_result['text'])) { | |
| 5955 | - // Direct text field (OpenAI's actual format) | |
| 5956 | - $text_content = $search_result['text']; | |
| 5957 | - //error_log("MXCHAT VECTORSTORE: Found text directly on result[$index], length: " . strlen($text_content)); | |
| 5958 | - } elseif (isset($search_result['content']) && is_array($search_result['content'])) { | |
| 5959 | - // Nested content array format | |
| 5960 | - foreach ($search_result['content'] as $content_item) { | |
| 5961 | - if (isset($content_item['text'])) { | |
| 5962 | - $text_content .= $content_item['text'] . "\n"; | |
| 5963 | - } | |
| 5964 | - } | |
| 5965 | - //error_log("MXCHAT VECTORSTORE: Found text in content array for result[$index], length: " . strlen($text_content)); | |
| 5966 | - } else { | |
| 5967 | - //error_log("MXCHAT VECTORSTORE: No text found for result[$index]. Keys: " . implode(', ', array_keys($search_result))); | |
| 5968 | - } | |
| 5969 | - | |
| 5970 | - if (!empty($text_content)) { | |
| 5971 | - $content .= "## Reference " . ($matches_used + 1) . " ##\n"; | |
| 5972 | - $content .= trim($text_content) . "\n\n"; | |
| 5973 | - | |
| 5974 | - if (!empty($filename)) { | |
| 5975 | - $content .= "Source: " . $filename . "\n\n"; | |
| 5976 | - } | |
| 5977 | - | |
| 5978 | - // Extract URLs from content | |
| 5979 | - preg_match_all( | |
| 5980 | - '#\bhttps?://[^\s<>"\']+#i', | |
| 5981 | - $text_content, | |
| 5982 | - $content_urls | |
| 5983 | - ); | |
| 5984 | - if (!empty($content_urls[0])) { | |
| 5985 | - $valid_urls = array_merge($valid_urls, $content_urls[0]); | |
| 5986 | - } | |
| 5987 | - | |
| 5988 | - $matches_used++; | |
| 5989 | - } | |
| 5990 | - | |
| 5991 | - // Store for similarity analysis | |
| 5992 | - $all_matches[] = [ | |
| 5993 | - 'document_id' => $filename ?: ('result_' . $index), | |
| 5994 | - 'similarity' => $score, | |
| 5995 | - 'similarity_percentage' => round($score * 100, 2), | |
| 5996 | - 'above_threshold' => true, | |
| 5997 | - 'source_display' => $filename, | |
| 5998 | - 'content_preview' => substr(strip_tags($text_content), 0, 100) . '...', | |
| 5999 | - 'used_for_context' => true, | |
| 6000 | - 'role_restriction' => 'public', | |
| 6001 | - 'has_access' => true, | |
| 6002 | - 'filtered_out' => false | |
| 6003 | - ]; | |
| 6004 | - } | |
| 6005 | - } | |
| 6006 | - | |
| 6007 | - // Also check for message content with annotations (citations) | |
| 6008 | - if (isset($output_item['type']) && $output_item['type'] === 'message') { | |
| 6009 | - if (isset($output_item['content']) && is_array($output_item['content'])) { | |
| 6010 | - foreach ($output_item['content'] as $content_block) { | |
| 6011 | - if (isset($content_block['annotations']) && is_array($content_block['annotations'])) { | |
| 6012 | - foreach ($content_block['annotations'] as $annotation) { | |
| 6013 | - if (isset($annotation['filename'])) { | |
| 6014 | - $filename = $annotation['filename']; | |
| 6015 | - $score = $annotation['score'] ?? 0; | |
| 6016 | - $text_content = ''; | |
| 6017 | - | |
| 6018 | - if (isset($annotation['content']) && is_array($annotation['content'])) { | |
| 6019 | - foreach ($annotation['content'] as $ann_content) { | |
| 6020 | - if (isset($ann_content['text'])) { | |
| 6021 | - $text_content .= $ann_content['text'] . "\n"; | |
| 6022 | - } | |
| 6023 | - } | |
| 6024 | - } | |
| 6025 | - | |
| 6026 | - if (!empty($text_content) && $matches_used < $max_results) { | |
| 6027 | - $content .= "## Reference " . ($matches_used + 1) . " ##\n"; | |
| 6028 | - $content .= trim($text_content) . "\n\n"; | |
| 6029 | - $content .= "Source: " . $filename . "\n\n"; | |
| 6030 | - | |
| 6031 | - preg_match_all( | |
| 6032 | - '#\bhttps?://[^\s<>"\']+#i', | |
| 6033 | - $text_content, | |
| 6034 | - $content_urls | |
| 6035 | - ); | |
| 6036 | - if (!empty($content_urls[0])) { | |
| 6037 | - $valid_urls = array_merge($valid_urls, $content_urls[0]); | |
| 6038 | - } | |
| 6039 | - | |
| 6040 | - $matches_used++; | |
| 6041 | - | |
| 6042 | - $all_matches[] = [ | |
| 6043 | - 'document_id' => $filename, | |
| 6044 | - 'similarity' => $score, | |
| 6045 | - 'similarity_percentage' => round($score * 100, 2), | |
| 6046 | - 'above_threshold' => true, | |
| 6047 | - 'source_display' => $filename, | |
| 6048 | - 'content_preview' => substr(strip_tags($text_content), 0, 100) . '...', | |
| 6049 | - 'used_for_context' => true, | |
| 6050 | - 'role_restriction' => 'public', | |
| 6051 | - 'has_access' => true, | |
| 6052 | - 'filtered_out' => false | |
| 6053 | - ]; | |
| 6054 | - } | |
| 6055 | - } | |
| 6056 | - } | |
| 6057 | - } | |
| 6058 | - } | |
| 6059 | - } | |
| 6060 | - } | |
| 6061 | - } | |
| 6062 | - } | |
| 6063 | - | |
| 6064 | - // Store for testing panel | |
| 6065 | - $this->last_similarity_analysis['top_matches'] = $all_matches; | |
| 6066 | - $this->last_similarity_analysis['total_checked'] = count($all_matches); | |
| 6067 | - | |
| 6068 | - // Store unique valid URLs for validation | |
| 6069 | - $this->current_valid_urls = array_unique($valid_urls); | |
| 6070 | - | |
| 6071 | - // Allow add-ons to act on similarity results (e.g. WooCommerce product card display) | |
| 6072 | - do_action('mxchat_similarity_results', $this->last_similarity_analysis['top_matches'], $bot_id); | |
| 6073 | - | |
| 6074 | - //error_log("MXCHAT VECTORSTORE: ========== SEARCH COMPLETE =========="); | |
| 6075 | - //error_log("MXCHAT VECTORSTORE: Matches used: " . $matches_used); | |
| 6076 | - //error_log("MXCHAT VECTORSTORE: All matches count: " . count($all_matches)); | |
| 6077 | - //error_log("MXCHAT VECTORSTORE: Content length: " . strlen($content)); | |
| 6078 | - if ($matches_used > 0) { | |
| 6079 | - //error_log("MXCHAT VECTORSTORE: Content preview: " . substr($content, 0, 500)); | |
| 6080 | - } | |
| 6081 | - | |
| 6082 | - // Check if citation links are enabled | |
| 6083 | - $citation_links_enabled = ($mxchat_options['citation_links_toggle'] ?? 'on') === 'on'; | |
| 6084 | - | |
| 6085 | - // Add response guidelines | |
| 6086 | - if ($matches_used === 0) { | |
| 6087 | - //error_log("MXCHAT VECTORSTORE: No matches found - returning empty reference message"); | |
| 6088 | - $content = "No reference information was found for this query.\n\n"; | |
| 6089 | - } else { | |
| 6090 | - // Build response guidelines based on citation links setting | |
| 6091 | - $content .= "\n## Response Guidelines ##\n" . | |
| 6092 | - "You are an AI Chatbot. Answer naturally and helpfully using only the information from the references above. " . | |
| 6093 | - "Be conversational and friendly, but never mention your knowledge base or training data. " . | |
| 6094 | - "If you don't have specific information or are uncertain about any details, it's always " . | |
| 6095 | - "better to honestly say you don't know rather than making up or guessing at answers. " . | |
| 6096 | - "When information is incomplete, let them know you are unsure.\n\n"; | |
| 6097 | - | |
| 6098 | - // Only add hyperlink instructions if citation links are enabled | |
| 6099 | - if ($citation_links_enabled) { | |
| 6100 | - $content .= "CRITICAL: When creating hyperlinks, always use proper markdown format with descriptive text: " . | |
| 6101 | - "[descriptive text](url). NEVER use empty brackets like [](url). The text in brackets must describe what the link is about."; | |
| 6102 | - } else { | |
| 6103 | - $content .= "IMPORTANT: Do not include any citation links, source URLs, or hyperlinks in your responses. " . | |
| 6104 | - "Simply provide helpful answers based on the reference information without citing sources."; | |
| 6105 | - } | |
| 6106 | - } | |
| 6107 | - | |
| 6108 | - //error_log("MXCHAT DEBUG: Vector Store search complete. Matches used: " . $matches_used); | |
| 6109 | - | |
| 6110 | - return trim($content); | |
| 6111 | -} | |
| 6112 | - | |
| 6113 | -/** | |
| 6114 | - * Check if the given model is an OpenAI chat model | |
| 6115 | - * | |
| 6116 | - * @param string $model The model ID | |
| 6117 | - * @return bool True if it's an OpenAI model | |
| 6118 | - */ | |
| 6119 | -private function is_openai_chat_model($model) { | |
| 6120 | - $openai_prefixes = array('gpt-', 'o1-', 'o3-'); | |
| 6121 | - foreach ($openai_prefixes as $prefix) { | |
| 6122 | - if (strpos($model, $prefix) === 0) { | |
| 6123 | - return true; | |
| 6124 | - } | |
| 6125 | - } | |
| 6126 | - return false; | |
| 6127 | -} | |
| 6128 | - | |
| 6129 | -/** | |
| 6130 | - * Get bot-specific Vector Store configuration | |
| 6131 | - * | |
| 6132 | - * @param string $bot_id The bot ID | |
| 6133 | - * @return array Configuration array | |
| 6134 | - */ | |
| 6135 | -private function get_bot_vectorstore_config($bot_id = 'default') { | |
| 6136 | - $vectorstore_options = get_option('mxchat_openai_vectorstore_options', array()); | |
| 6137 | - | |
| 6138 | - // Default global settings | |
| 6139 | - $default_config = array( | |
| 6140 | - 'use_vectorstore' => ($vectorstore_options['mxchat_use_openai_vectorstore'] ?? '0') === '1', | |
| 6141 | - 'vectorstore_ids' => $vectorstore_options['mxchat_vectorstore_ids'] ?? '', | |
| 6142 | - 'max_results' => $vectorstore_options['mxchat_vectorstore_max_results'] ?? 5 | |
| 6143 | - ); | |
| 6144 | - | |
| 6145 | - // Allow multi-bot plugin to override with bot-specific settings | |
| 6146 | - $bot_config = apply_filters('mxchat_get_bot_vectorstore_config', $default_config, $bot_id); | |
| 6147 | - | |
| 6148 | - // Preserve max_results from global settings if not set in bot config | |
| 6149 | - if (!isset($bot_config['max_results'])) { | |
| 6150 | - $bot_config['max_results'] = $default_config['max_results']; | |
| 6151 | - } | |
| 6152 | - | |
| 6153 | - return $bot_config; | |
| 6154 | -} | |
| 6155 | - | |
| 6156 | 4044 | private function mxchat_find_relevant_products($user_embedding) { |
| 6157 | 4045 | //error_log('MXChat Vector Search: Starting product search...'); |
| 6158 | 4046 | |
| 6159 | 4047 | // Retrieve the add-on settings from the database |
| @@ -6174,75 +4062,73 @@ | ||
| 6174 | 4062 | } |
| 6175 | 4063 | private function find_relevant_products_wordpress($user_embedding) { |
| 6176 | 4064 | global $wpdb; |
| 6177 | 4065 | $system_prompt_table = $wpdb->prefix . 'mxchat_system_prompt_content'; |
| 4066 | + $cache_key = 'mxchat_system_prompt_embeddings'; | |
| 4067 | + $batch_size = 500; | |
| 6178 | 4068 | |
| 6179 | - if (!is_array($user_embedding)) { | |
| 6180 | - return ''; | |
| 6181 | - } | |
| 4069 | + // Original WordPress database search logic | |
| 4070 | + // [Previous implementation remains the same] | |
| 4071 | + $embeddings = wp_cache_get($cache_key, 'mxchat_system_prompts'); | |
| 4072 | + if ($embeddings === false) { | |
| 4073 | + $embeddings = []; | |
| 4074 | + $offset = 0; | |
| 6182 | 4075 | |
| 6183 | - // Streaming top-K pass: scan rows in small batches, keep only the top 3 | |
| 6184 | - // results above the similarity threshold. Peak memory is bounded by | |
| 6185 | - // $batch_size embedding rows plus a 3-element top list. | |
| 6186 | - $batch_size = 250; | |
| 6187 | - $similarity_threshold = 0.85; | |
| 6188 | - $top_k = 3; | |
| 6189 | - $top_results = []; | |
| 6190 | - $offset = 0; | |
| 4076 | + do { | |
| 4077 | + $query = $wpdb->prepare( | |
| 4078 | + "SELECT id, embedding_vector | |
| 4079 | + FROM {$system_prompt_table} | |
| 4080 | + LIMIT %d OFFSET %d", | |
| 4081 | + $batch_size, | |
| 4082 | + $offset | |
| 4083 | + ); | |
| 6191 | 4084 | |
| 6192 | - do { | |
| 6193 | - $batch = $wpdb->get_results($wpdb->prepare( | |
| 6194 | - "SELECT id, embedding_vector | |
| 6195 | - FROM {$system_prompt_table} | |
| 6196 | - LIMIT %d OFFSET %d", | |
| 6197 | - $batch_size, | |
| 6198 | - $offset | |
| 6199 | - )); | |
| 4085 | + $batch = $wpdb->get_results($query); | |
| 4086 | + if (empty($batch)) { | |
| 4087 | + break; | |
| 4088 | + } | |
| 6200 | 4089 | |
| 6201 | - if (empty($batch)) { | |
| 6202 | - break; | |
| 6203 | - } | |
| 4090 | + $embeddings = array_merge($embeddings, $batch); | |
| 4091 | + $offset += $batch_size; | |
| 6204 | 4092 | |
| 6205 | - foreach ($batch as $row) { | |
| 6206 | - $database_embedding = $row->embedding_vector | |
| 6207 | - ? unserialize($row->embedding_vector, ['allowed_classes' => false]) | |
| 6208 | - : null; | |
| 4093 | + unset($batch); | |
| 6209 | 4094 | |
| 6210 | - if (!is_array($database_embedding)) { | |
| 6211 | - unset($database_embedding); | |
| 6212 | - continue; | |
| 6213 | - } | |
| 4095 | + } while (true); | |
| 6214 | 4096 | |
| 4097 | + if (empty($embeddings)) { | |
| 4098 | + return ''; | |
| 4099 | + } | |
| 4100 | + wp_cache_set($cache_key, $embeddings, 'mxchat_system_prompts', 3600); | |
| 4101 | + } | |
| 4102 | + | |
| 4103 | + $relevant_results = []; | |
| 4104 | + foreach ($embeddings as $embedding) { | |
| 4105 | + $database_embedding = $embedding->embedding_vector | |
| 4106 | + ? unserialize($embedding->embedding_vector, ['allowed_classes' => false]) | |
| 4107 | + : null; | |
| 4108 | + if (is_array($database_embedding) && is_array($user_embedding)) { | |
| 6215 | 4109 | $similarity = $this->mxchat_calculate_cosine_similarity($user_embedding, $database_embedding); |
| 6216 | - unset($database_embedding); | |
| 6217 | - | |
| 6218 | - if ($similarity < $similarity_threshold) { | |
| 6219 | - continue; | |
| 6220 | - } | |
| 6221 | - | |
| 6222 | - // Insert into bounded top-K (kept sorted descending) | |
| 6223 | - if (count($top_results) < $top_k) { | |
| 6224 | - $top_results[] = ['id' => $row->id, 'similarity' => $similarity]; | |
| 6225 | - usort($top_results, function ($a, $b) { | |
| 6226 | - return $b['similarity'] <=> $a['similarity']; | |
| 6227 | - }); | |
| 6228 | - } elseif ($similarity > $top_results[$top_k - 1]['similarity']) { | |
| 6229 | - $top_results[$top_k - 1] = ['id' => $row->id, 'similarity' => $similarity]; | |
| 6230 | - usort($top_results, function ($a, $b) { | |
| 6231 | - return $b['similarity'] <=> $a['similarity']; | |
| 6232 | - }); | |
| 6233 | - } | |
| 4110 | + $relevant_results[] = [ | |
| 4111 | + 'id' => $embedding->id, | |
| 4112 | + 'similarity' => $similarity | |
| 4113 | + ]; | |
| 6234 | 4114 | } |
| 4115 | + unset($database_embedding); | |
| 4116 | + } | |
| 6235 | 4117 | |
| 6236 | - unset($batch); | |
| 6237 | - $offset += $batch_size; | |
| 6238 | - } while (true); | |
| 4118 | + // Use fixed threshold for products | |
| 4119 | + $similarity_threshold = 0.85; | |
| 6239 | 4120 | |
| 6240 | - if (empty($top_results)) { | |
| 6241 | - return ''; | |
| 6242 | - } | |
| 4121 | + $relevant_results = array_filter($relevant_results, function ($result) use ($similarity_threshold) { | |
| 4122 | + return $result['similarity'] >= $similarity_threshold; | |
| 4123 | + }); | |
| 4124 | + usort($relevant_results, function ($a, $b) { | |
| 4125 | + return $b['similarity'] <=> $a['similarity']; | |
| 4126 | + }); | |
| 6243 | 4127 | |
| 4128 | + $top_results = array_slice($relevant_results, 0, 5); | |
| 6244 | 4129 | $content = ''; |
| 4130 | + | |
| 6245 | 4131 | foreach ($top_results as $result) { |
| 6246 | 4132 | $chunk_content = $this->fetch_content_with_product_links($result['id']); |
| 6247 | 4133 | $content .= $chunk_content . "\n\n"; |
| 6248 | 4134 | } |
| @@ -6351,61 +4237,23 @@ | ||
| 6351 | 4237 | |
| 6352 | 4238 | /** |
| 6353 | 4239 | * Get system instructions for a specific bot or default |
| 6354 | 4240 | * Checks for multi-bot add-on and uses bot-specific instructions if available |
| 6355 | - * Automatically strips URLs if citation links are disabled | |
| 6356 | - * Replaces {visitor_name} placeholder with actual visitor name if available | |
| 6357 | - * | |
| 6358 | - * @param string $bot_id The bot ID to get instructions for | |
| 6359 | - * @param string $session_id Optional session ID to lookup visitor name | |
| 6360 | 4241 | */ |
| 6361 | -private function get_system_instructions($bot_id = 'default', $session_id = '') { | |
| 6362 | - $instructions = ''; | |
| 6363 | - | |
| 4242 | +private function get_system_instructions($bot_id = 'default') { | |
| 6364 | 4243 | // Check if multi-bot add-on is active |
| 6365 | 4244 | if (class_exists('MxChat_Multi_Bot_Core_Manager') && $bot_id !== 'default') { |
| 6366 | 4245 | // Get bot-specific options from multi-bot add-on |
| 6367 | 4246 | $bot_options = apply_filters('mxchat_get_bot_options', array(), $bot_id); |
| 6368 | - | |
| 4247 | + | |
| 6369 | 4248 | // If bot has custom system instructions, use those |
| 6370 | 4249 | if (!empty($bot_options['system_prompt_instructions'])) { |
| 6371 | - $instructions = $bot_options['system_prompt_instructions']; | |
| 4250 | + return $bot_options['system_prompt_instructions']; | |
| 6372 | 4251 | } |
| 6373 | 4252 | } |
| 6374 | - | |
| 4253 | + | |
| 6375 | 4254 | // Fall back to default system instructions |
| 6376 | - if (empty($instructions)) { | |
| 6377 | - $instructions = isset($this->options['system_prompt_instructions']) ? $this->options['system_prompt_instructions'] : ''; | |
| 6378 | - } | |
| 6379 | - | |
| 6380 | - // Check if citation links are disabled - if so, strip URLs from instructions | |
| 6381 | - $fresh_options = get_option('mxchat_options', []); | |
| 6382 | - $citation_links_enabled = isset($fresh_options['citation_links_toggle']) ? ($fresh_options['citation_links_toggle'] === 'on') : true; | |
| 6383 | - | |
| 6384 | - if (!$citation_links_enabled && !empty($instructions)) { | |
| 6385 | - $instructions = preg_replace('#\bhttps?://[^\s<>"\']+#i', '', $instructions); | |
| 6386 | - $instructions = preg_replace('/\s+/', ' ', trim($instructions)); // Clean up extra spaces | |
| 6387 | - } | |
| 6388 | - | |
| 6389 | - // Replace {visitor_name} placeholder with actual visitor name if available | |
| 6390 | - if (!empty($instructions) && !empty($session_id) && stripos($instructions, '{visitor_name}') !== false) { | |
| 6391 | - $name_option_key = "mxchat_name_{$session_id}"; | |
| 6392 | - $visitor_name = get_option($name_option_key, ''); | |
| 6393 | - | |
| 6394 | - if (!empty($visitor_name)) { | |
| 6395 | - $instructions = str_ireplace('{visitor_name}', sanitize_text_field($visitor_name), $instructions); | |
| 6396 | - } else { | |
| 6397 | - // Remove placeholder if no name is available | |
| 6398 | - $instructions = str_ireplace('{visitor_name}', '', $instructions); | |
| 6399 | - $instructions = preg_replace('/\s{2,}/', ' ', trim($instructions)); // Clean up extra spaces | |
| 6400 | - } | |
| 6401 | - } | |
| 6402 | - | |
| 6403 | - // Allow developers to filter system instructions and process shortcodes | |
| 6404 | - $instructions = apply_filters('mxchat_system_instructions', $instructions, $bot_id, $session_id); | |
| 6405 | - $instructions = do_shortcode($instructions); | |
| 6406 | - | |
| 6407 | - return $instructions; | |
| 4255 | + return isset($this->options['system_prompt_instructions']) ? $this->options['system_prompt_instructions'] : ''; | |
| 6408 | 4256 | } |
| 6409 | 4257 | /** |
| 6410 | 4258 | * Get the current bot ID from session or request context |
| 6411 | 4259 | */ |
| @@ -6425,9 +4273,9 @@ | ||
| 6425 | 4273 | |
| 6426 | 4274 | // Fall back to default |
| 6427 | 4275 | return 'default'; |
| 6428 | 4276 | } |
| 6429 | -private function mxchat_generate_response($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, $selected_model = 'gpt-5.1-chat-latest') { | |
| 4277 | +private function mxchat_generate_response($relevant_content, $api_key, $xai_api_key, $claude_api_key, $deepseek_api_key, $gemini_api_key, $conversation_history, $streaming = false, $session_id = '', $testing_data = null, $selected_model = 'gpt-4o') { | |
| 6430 | 4278 | try { |
| 6431 | 4279 | if (!$relevant_content) { |
| 6432 | 4280 | $error_response = [ |
| 6433 | 4281 | 'error' => esc_html__("I couldn't find relevant information on that topic.", 'mxchat'), |
| @@ -6433,73 +4281,22 @@ | ||
| 6433 | 4281 | 'error' => esc_html__("I couldn't find relevant information on that topic.", 'mxchat'), |
| 6434 | 4282 | 'error_code' => 'no_relevant_content' |
| 6435 | 4283 | ]; |
| 6436 | 4284 | |
| 4285 | + // Add testing data to error response if available | |
| 6437 | 4286 | if ($testing_data !== null) { |
| 6438 | 4287 | $error_response['testing_data'] = $testing_data; |
| 4288 | + //error_log("MxChat Testing: Added testing data to no_relevant_content error"); | |
| 6439 | 4289 | } |
| 6440 | 4290 | |
| 6441 | 4291 | return $error_response; |
| 6442 | 4292 | } |
| 6443 | 4293 | |
| 4294 | + // Ensure conversation_history is an array | |
| 6444 | 4295 | if (!is_array($conversation_history)) { |
| 6445 | 4296 | $conversation_history = array(); |
| 6446 | 4297 | } |
| 6447 | 4298 | |
| 6448 | - // Check if this is an OpenRouter model | |
| 6449 | - if ($selected_model === 'openrouter') { | |
| 6450 | - // Get the actual OpenRouter model from options | |
| 6451 | - $openrouter_selected_model = $this->options['openrouter_selected_model'] ?? ''; | |
| 6452 | - | |
| 6453 | - if (empty($openrouter_selected_model)) { | |
| 6454 | - $error_response = [ | |
| 6455 | - 'error' => esc_html__('No OpenRouter model selected. Please select a model in settings.', 'mxchat'), | |
| 6456 | - 'error_code' => 'no_openrouter_model_selected' | |
| 6457 | - ]; | |
| 6458 | - if ($testing_data !== null) { | |
| 6459 | - $error_response['testing_data'] = $testing_data; | |
| 6460 | - } | |
| 6461 | - return $error_response; | |
| 6462 | - } | |
| 6463 | - | |
| 6464 | - if (empty($openrouter_api_key)) { | |
| 6465 | - $error_response = [ | |
| 6466 | - 'error' => esc_html__('OpenRouter API key is not configured', 'mxchat'), | |
| 6467 | - 'error_code' => 'missing_openrouter_api_key' | |
| 6468 | - ]; | |
| 6469 | - if ($testing_data !== null) { | |
| 6470 | - $error_response['testing_data'] = $testing_data; | |
| 6471 | - } | |
| 6472 | - return $error_response; | |
| 6473 | - } | |
| 6474 | - | |
| 6475 | - if ($streaming) { | |
| 6476 | - return $this->mxchat_generate_response_openrouter_stream( | |
| 6477 | - $openrouter_selected_model, | |
| 6478 | - $openrouter_api_key, | |
| 6479 | - $conversation_history, | |
| 6480 | - $relevant_content, | |
| 6481 | - $session_id, | |
| 6482 | - $testing_data | |
| 6483 | - ); | |
| 6484 | - } else { | |
| 6485 | - $response = $this->mxchat_generate_response_openrouter( | |
| 6486 | - $openrouter_selected_model, | |
| 6487 | - $openrouter_api_key, | |
| 6488 | - $conversation_history, | |
| 6489 | - $relevant_content | |
| 6490 | - ); | |
| 6491 | - } | |
| 6492 | - | |
| 6493 | - if (is_array($response) && isset($response['error'])) { | |
| 6494 | - if ($testing_data !== null) { | |
| 6495 | - $response['testing_data'] = $testing_data; | |
| 6496 | - } | |
| 6497 | - return $response; | |
| 6498 | - } | |
| 6499 | - | |
| 6500 | - return $response; | |
| 6501 | - } | |
| 6502 | 4299 | |
| 6503 | 4300 | // Extract model prefix to determine the provider |
| 6504 | 4301 | $model_parts = explode('-', $selected_model); |
| 6505 | 4302 | $provider = strtolower($model_parts[0]); |
| @@ -6542,9 +4339,9 @@ | ||
| 6542 | 4339 | $claude_api_key, |
| 6543 | 4340 | $conversation_history, |
| 6544 | 4341 | $relevant_content, |
| 6545 | 4342 | $session_id, |
| 6546 | - $testing_data | |
| 4343 | + $testing_data // Pass testing data | |
| 6547 | 4344 | ); |
| 6548 | 4345 | } else { |
| 6549 | 4346 | $response = $this->mxchat_generate_response_claude( |
| 6550 | 4347 | $selected_model, |
| @@ -6572,9 +4369,9 @@ | ||
| 6572 | 4369 | $xai_api_key, |
| 6573 | 4370 | $conversation_history, |
| 6574 | 4371 | $relevant_content, |
| 6575 | 4372 | $session_id, |
| 6576 | - $testing_data | |
| 4373 | + $testing_data // Pass testing data | |
| 6577 | 4374 | ); |
| 6578 | 4375 | } else { |
| 6579 | 4376 | $response = $this->mxchat_generate_response_xai( |
| 6580 | 4377 | $selected_model, |
| @@ -6602,9 +4399,9 @@ | ||
| 6602 | 4399 | $deepseek_api_key, |
| 6603 | 4400 | $conversation_history, |
| 6604 | 4401 | $relevant_content, |
| 6605 | 4402 | $session_id, |
| 6606 | - $testing_data | |
| 4403 | + $testing_data // Pass testing data | |
| 6607 | 4404 | ); |
| 6608 | 4405 | } else { |
| 6609 | 4406 | $response = $this->mxchat_generate_response_deepseek( |
| 6610 | 4407 | $selected_model, |
| @@ -6626,27 +4423,9 @@ | ||
| 6626 | 4423 | $error_response['testing_data'] = $testing_data; |
| 6627 | 4424 | } |
| 6628 | 4425 | return $error_response; |
| 6629 | 4426 | } |
| 6630 | - | |
| 6631 | - // Check if web search is enabled for this OpenAI model | |
| 6632 | - $web_search_enabled = isset($this->options['enable_web_search']) && $this->options['enable_web_search'] === 'on'; | |
| 6633 | - // Models that don't support web search | |
| 6634 | - $unsupported_web_search_models = array('gpt-4.1-nano'); | |
| 6635 | - $model_supports_web_search = !in_array($selected_model, $unsupported_web_search_models); | |
| 6636 | - | |
| 6637 | - if ($web_search_enabled && $model_supports_web_search) { | |
| 6638 | - // Use Responses API (required for some models, or when web search is enabled) | |
| 6639 | - return $this->mxchat_generate_response_openai_web_search( | |
| 6640 | - $selected_model, | |
| 6641 | - $api_key, | |
| 6642 | - $conversation_history, | |
| 6643 | - $relevant_content, | |
| 6644 | - $session_id, | |
| 6645 | - $testing_data, | |
| 6646 | - $streaming | |
| 6647 | - ); | |
| 6648 | - } elseif ($streaming) { | |
| 4427 | + if ($streaming) { | |
| 6649 | 4428 | return $this->mxchat_generate_response_openai_stream( |
| 6650 | 4429 | $selected_model, |
| 6651 | 4430 | $api_key, |
| 6652 | 4431 | $conversation_history, |
| @@ -6651,9 +4430,9 @@ | ||
| 6651 | 4430 | $api_key, |
| 6652 | 4431 | $conversation_history, |
| 6653 | 4432 | $relevant_content, |
| 6654 | 4433 | $session_id, |
| 6655 | - $testing_data | |
| 4434 | + $testing_data // Pass testing data | |
| 6656 | 4435 | ); |
| 6657 | 4436 | } else { |
| 6658 | 4437 | $response = $this->mxchat_generate_response_openai( |
| 6659 | 4438 | $selected_model, |
| @@ -6664,8 +4443,9 @@ | ||
| 6664 | 4443 | } |
| 6665 | 4444 | break; |
| 6666 | 4445 | |
| 6667 | 4446 | default: |
| 4447 | + // Default to OpenAI for custom models or unrecognized prefixes | |
| 6668 | 4448 | if (empty($api_key)) { |
| 6669 | 4449 | $error_response = [ |
| 6670 | 4450 | 'error' => esc_html__('OpenAI API key is not configured', 'mxchat'), |
| 6671 | 4451 | 'error_code' => 'missing_openai_api_key' |
| @@ -6674,25 +4454,9 @@ | ||
| 6674 | 4454 | $error_response['testing_data'] = $testing_data; |
| 6675 | 4455 | } |
| 6676 | 4456 | return $error_response; |
| 6677 | 4457 | } |
| 6678 | - | |
| 6679 | - // Check if web search is enabled (default case also handles OpenAI models) | |
| 6680 | - $web_search_enabled = isset($this->options['enable_web_search']) && $this->options['enable_web_search'] === 'on'; | |
| 6681 | - $unsupported_web_search_models = array('gpt-4.1-nano'); | |
| 6682 | - $model_supports_web_search = !in_array($selected_model, $unsupported_web_search_models); | |
| 6683 | - | |
| 6684 | - if ($web_search_enabled && $model_supports_web_search) { | |
| 6685 | - return $this->mxchat_generate_response_openai_web_search( | |
| 6686 | - $selected_model, | |
| 6687 | - $api_key, | |
| 6688 | - $conversation_history, | |
| 6689 | - $relevant_content, | |
| 6690 | - $session_id, | |
| 6691 | - $testing_data, | |
| 6692 | - $streaming | |
| 6693 | - ); | |
| 6694 | - } elseif ($streaming) { | |
| 4458 | + if ($streaming) { | |
| 6695 | 4459 | return $this->mxchat_generate_response_openai_stream( |
| 6696 | 4460 | $selected_model, |
| 6697 | 4461 | $api_key, |
| 6698 | 4462 | $conversation_history, |
| @@ -6697,9 +4461,9 @@ | ||
| 6697 | 4461 | $api_key, |
| 6698 | 4462 | $conversation_history, |
| 6699 | 4463 | $relevant_content, |
| 6700 | 4464 | $session_id, |
| 6701 | - $testing_data | |
| 4465 | + $testing_data // Pass testing data | |
| 6702 | 4466 | ); |
| 6703 | 4467 | } else { |
| 6704 | 4468 | $response = $this->mxchat_generate_response_openai( |
| 6705 | 4469 | $selected_model, |
| @@ -6710,18 +4474,24 @@ | ||
| 6710 | 4474 | } |
| 6711 | 4475 | break; |
| 6712 | 4476 | } |
| 6713 | 4477 | |
| 4478 | + // Check if the response is an error array from the provider-specific function | |
| 6714 | 4479 | if (is_array($response) && isset($response['error'])) { |
| 4480 | + // Add testing data to error response if available | |
| 6715 | 4481 | if ($testing_data !== null) { |
| 6716 | 4482 | $response['testing_data'] = $testing_data; |
| 4483 | + //error_log("MxChat Testing: Added testing data to provider error response"); | |
| 6717 | 4484 | } |
| 6718 | - return $response; | |
| 4485 | + return $response; // Pass through the error with testing data | |
| 6719 | 4486 | } |
| 6720 | 4487 | |
| 4488 | + // For successful non-streaming responses, we don't add testing data here | |
| 4489 | + // because it will be added in the main handler | |
| 6721 | 4490 | return $response; |
| 6722 | 4491 | |
| 6723 | 4492 | } catch (Exception $e) { |
| 4493 | + //error_log('MXChat Error: ' . $e->getMessage()); | |
| 6724 | 4494 | $error_response = [ |
| 6725 | 4495 | 'error' => sprintf(esc_html__('An error occurred: %s', 'mxchat'), esc_html($e->getMessage())), |
| 6726 | 4496 | 'error_code' => 'system_exception', |
| 6727 | 4497 | 'exception_details' => $e->getMessage() |
| @@ -6726,229 +4496,24 @@ | ||
| 6726 | 4496 | 'error_code' => 'system_exception', |
| 6727 | 4497 | 'exception_details' => $e->getMessage() |
| 6728 | 4498 | ]; |
| 6729 | 4499 | |
| 4500 | + // Add testing data to exception response if available | |
| 6730 | 4501 | if ($testing_data !== null) { |
| 6731 | 4502 | $error_response['testing_data'] = $testing_data; |
| 4503 | + //error_log("MxChat Testing: Added testing data to exception response"); | |
| 6732 | 4504 | } |
| 6733 | 4505 | |
| 6734 | 4506 | return $error_response; |
| 6735 | 4507 | } |
| 6736 | 4508 | } |
| 6737 | -private function mxchat_generate_response_openrouter_stream($selected_model, $openrouter_api_key, $conversation_history, $relevant_content, $session_id, $testing_data = null) { | |
| 6738 | - try { | |
| 6739 | - $bot_id = $this->get_current_bot_id($session_id); | |
| 6740 | - $system_prompt_instructions = $this->get_system_instructions($bot_id, $session_id); | |
| 6741 | - | |
| 6742 | - if (!is_array($conversation_history)) { | |
| 6743 | - $conversation_history = array(); | |
| 6744 | - } | |
| 6745 | 4509 | |
| 6746 | - $formatted_conversation = array(); | |
| 6747 | - | |
| 6748 | - $formatted_conversation[] = array( | |
| 6749 | - 'role' => 'system', | |
| 6750 | - 'content' => $system_prompt_instructions . " " . $relevant_content | |
| 6751 | - ); | |
| 6752 | - | |
| 6753 | - foreach ($conversation_history as $message) { | |
| 6754 | - if (is_array($message) && isset($message['role']) && isset($message['content'])) { | |
| 6755 | - $role = $message['role']; | |
| 6756 | - if ($role === 'bot' || $role === 'agent') { | |
| 6757 | - $role = 'assistant'; | |
| 6758 | - } | |
| 6759 | - if (!in_array($role, ['system', 'assistant', 'user'])) { | |
| 6760 | - $role = 'user'; | |
| 6761 | - } | |
| 6762 | - $formatted_conversation[] = array( | |
| 6763 | - 'role' => $role, | |
| 6764 | - 'content' => $message['content'] | |
| 6765 | - ); | |
| 6766 | - } | |
| 6767 | - } | |
| 6768 | - | |
| 6769 | - if (headers_sent() || !function_exists('curl_init')) { | |
| 6770 | - $regular_response = $this->mxchat_generate_response_openrouter( | |
| 6771 | - $selected_model, | |
| 6772 | - $openrouter_api_key, | |
| 6773 | - $conversation_history, | |
| 6774 | - $relevant_content | |
| 6775 | - ); | |
| 6776 | - | |
| 6777 | - // Save bot response to transcript | |
| 6778 | - if (!empty($regular_response) && !empty($session_id)) { | |
| 6779 | - $this->mxchat_save_chat_message($session_id, 'bot', $regular_response); | |
| 6780 | - } | |
| 6781 | - | |
| 6782 | - $response_data = [ | |
| 6783 | - 'text' => $regular_response, | |
| 6784 | - 'html' => '', | |
| 6785 | - 'session_id' => $session_id | |
| 6786 | - ]; | |
| 6787 | - | |
| 6788 | - if ($testing_data !== null) { | |
| 6789 | - $response_data['testing_data'] = $testing_data; | |
| 6790 | - } | |
| 6791 | - | |
| 6792 | - header('Content-Type: application/json'); | |
| 6793 | - echo json_encode($response_data); | |
| 6794 | - return true; | |
| 6795 | - } | |
| 6796 | - | |
| 6797 | - $body = json_encode([ | |
| 6798 | - 'model' => $selected_model, | |
| 6799 | - 'messages' => $formatted_conversation, | |
| 6800 | - 'temperature' => 1, | |
| 6801 | - 'stream' => true | |
| 6802 | - ]); | |
| 6803 | - | |
| 6804 | - // Setup streaming headers now that we know we're actually streaming | |
| 6805 | - $this->setup_streaming_headers(); | |
| 6806 | - | |
| 6807 | - $ch = curl_init(); | |
| 6808 | - curl_setopt($ch, CURLOPT_URL, 'https://openrouter.ai/api/v1/chat/completions'); | |
| 6809 | - curl_setopt($ch, CURLOPT_RETURNTRANSFER, false); | |
| 6810 | - curl_setopt($ch, CURLOPT_POST, true); | |
| 6811 | - curl_setopt($ch, CURLOPT_POSTFIELDS, $body); | |
| 6812 | - curl_setopt($ch, CURLOPT_HTTPHEADER, array( | |
| 6813 | - 'Content-Type: application/json', | |
| 6814 | - 'Authorization: Bearer ' . $openrouter_api_key, | |
| 6815 | - 'HTTP-Referer: ' . home_url(), | |
| 6816 | - 'X-Title: ' . get_bloginfo('name') | |
| 6817 | - )); | |
| 6818 | - curl_setopt($ch, CURLOPT_SSL_VERIFYPEER, true); | |
| 6819 | - curl_setopt($ch, CURLOPT_TIMEOUT, 60); | |
| 6820 | - | |
| 6821 | - $full_response = ''; | |
| 6822 | - $stream_started = false; | |
| 6823 | - $buffer = ''; | |
| 6824 | - | |
| 6825 | - curl_setopt($ch, CURLOPT_WRITEFUNCTION, function($ch, $data) use (&$full_response, &$stream_started, &$buffer, $testing_data) { | |
| 6826 | - if (!$stream_started && $testing_data !== null) { | |
| 6827 | - echo "data: " . json_encode(['testing_data' => $testing_data]) . "\n\n"; | |
| 6828 | - flush(); | |
| 6829 | - $stream_started = true; | |
| 6830 | - } | |
| 6831 | - | |
| 6832 | - $buffer .= $data; | |
| 6833 | - $lines = explode("\n", $buffer); | |
| 6834 | - $buffer = array_pop($lines); | |
| 6835 | - | |
| 6836 | - foreach ($lines as $line) { | |
| 6837 | - if (trim($line) === '') { | |
| 6838 | - continue; | |
| 6839 | - } | |
| 6840 | - | |
| 6841 | - if (strpos($line, 'data: ') !== 0) { | |
| 6842 | - continue; | |
| 6843 | - } | |
| 6844 | - | |
| 6845 | - $json_str = substr($line, 6); | |
| 6846 | - | |
| 6847 | - if (trim($json_str) === '[DONE]') { | |
| 6848 | - echo "data: [DONE]\n\n"; | |
| 6849 | - flush(); | |
| 6850 | - continue; | |
| 6851 | - } | |
| 6852 | - | |
| 6853 | - $json = json_decode(trim($json_str), true); | |
| 6854 | - if ($json && isset($json['choices'][0]['delta']['content'])) { | |
| 6855 | - $content = $json['choices'][0]['delta']['content']; | |
| 6856 | - $full_response .= $content; | |
| 6857 | - | |
| 6858 | - echo "data: " . json_encode(['content' => $content]) . "\n\n"; | |
| 6859 | - flush(); | |
| 6860 | - } | |
| 6861 | - } | |
| 6862 | - | |
| 6863 | - return strlen($data); | |
| 6864 | - }); | |
| 6865 | - | |
| 6866 | - $response = curl_exec($ch); | |
| 6867 | - $http_code = curl_getinfo($ch, CURLINFO_HTTP_CODE); | |
| 6868 | - | |
| 6869 | - if (curl_errno($ch) || $http_code !== 200) { | |
| 6870 | - curl_close($ch); | |
| 6871 | - | |
| 6872 | - $regular_response = $this->mxchat_generate_response_openrouter( | |
| 6873 | - $selected_model, | |
| 6874 | - $openrouter_api_key, | |
| 6875 | - $conversation_history, | |
| 6876 | - $relevant_content | |
| 6877 | - ); | |
| 6878 | - | |
| 6879 | - $response_data = [ | |
| 6880 | - 'text' => $regular_response, | |
| 6881 | - 'html' => '', | |
| 6882 | - 'session_id' => $session_id | |
| 6883 | - ]; | |
| 6884 | - | |
| 6885 | - if ($testing_data !== null) { | |
| 6886 | - $response_data['testing_data'] = $testing_data; | |
| 6887 | - } | |
| 6888 | - | |
| 6889 | - header('Content-Type: application/json'); | |
| 6890 | - echo json_encode($response_data); | |
| 6891 | - return true; | |
| 6892 | - } | |
| 6893 | - | |
| 6894 | - curl_close($ch); | |
| 6895 | - | |
| 6896 | - if (!empty($full_response) && !empty($session_id)) { | |
| 6897 | - // Prepare RAG context for streaming response | |
| 6898 | - $rag_context_for_storage = null; | |
| 6899 | - $has_rag_data = $this->last_similarity_analysis !== null && !empty($this->last_similarity_analysis['top_matches']); | |
| 6900 | - $has_action_data = isset($this->last_action_analysis) && !empty($this->last_action_analysis); | |
| 6901 | - | |
| 6902 | - if ($has_rag_data || $has_action_data) { | |
| 6903 | - $rag_context_for_storage = []; | |
| 6904 | - | |
| 6905 | - if ($has_rag_data) { | |
| 6906 | - $rag_context_for_storage['top_matches'] = $this->last_similarity_analysis['top_matches']; | |
| 6907 | - $rag_context_for_storage['approved_urls'] = $this->current_valid_urls ?? []; | |
| 6908 | - $rag_context_for_storage['similarity_threshold'] = $this->last_similarity_analysis['threshold_used'] ?? 0.35; | |
| 6909 | - $rag_context_for_storage['knowledge_base_type'] = $this->last_similarity_analysis['knowledge_base_type'] ?? 'WordPress Database'; | |
| 6910 | - $rag_context_for_storage['total_documents_checked'] = $this->last_similarity_analysis['total_checked'] ?? 0; | |
| 6911 | - } | |
| 6912 | - | |
| 6913 | - if ($has_action_data) { | |
| 6914 | - $rag_context_for_storage['action_analysis'] = $this->last_action_analysis; | |
| 6915 | - } | |
| 6916 | - } | |
| 6917 | - $this->mxchat_save_chat_message($session_id, 'bot', $full_response, null, $rag_context_for_storage); | |
| 6918 | - } | |
| 6919 | - | |
| 6920 | - return true; | |
| 6921 | - | |
| 6922 | - } catch (Exception $e) { | |
| 6923 | - $regular_response = $this->mxchat_generate_response_openrouter( | |
| 6924 | - $selected_model, | |
| 6925 | - $openrouter_api_key, | |
| 6926 | - $conversation_history, | |
| 6927 | - $relevant_content | |
| 6928 | - ); | |
| 6929 | - | |
| 6930 | - $response_data = [ | |
| 6931 | - 'text' => $regular_response, | |
| 6932 | - 'html' => '', | |
| 6933 | - 'session_id' => $session_id | |
| 6934 | - ]; | |
| 6935 | - | |
| 6936 | - if ($testing_data !== null) { | |
| 6937 | - $response_data['testing_data'] = $testing_data; | |
| 6938 | - } | |
| 6939 | - | |
| 6940 | - header('Content-Type: application/json'); | |
| 6941 | - echo json_encode($response_data); | |
| 6942 | - return true; | |
| 6943 | - } | |
| 6944 | -} | |
| 6945 | 4510 | private function mxchat_generate_response_openai_stream($selected_model, $api_key, $conversation_history, $relevant_content, $session_id, $testing_data = null) { |
| 6946 | 4511 | try { |
| 6947 | 4512 | $bot_id = $this->get_current_bot_id($session_id); |
| 6948 | 4513 | |
| 6949 | 4514 | // Get system prompt instructions using centralized function |
| 6950 | - $system_prompt_instructions = $this->get_system_instructions($bot_id, $session_id); | |
| 4515 | + $system_prompt_instructions = $this->get_system_instructions($bot_id); | |
| 6951 | 4516 | |
| 6952 | 4517 | // Ensure conversation_history is an array |
| 6953 | 4518 | if (!is_array($conversation_history)) { |
| 6954 | 4519 | $conversation_history = array(); |
| @@ -6987,13 +4552,8 @@ | ||
| 6987 | 4552 | $conversation_history, |
| 6988 | 4553 | $relevant_content |
| 6989 | 4554 | ); |
| 6990 | 4555 | |
| 6991 | - // Save bot response to transcript | |
| 6992 | - if (!empty($regular_response) && !empty($session_id)) { | |
| 6993 | - $this->mxchat_save_chat_message($session_id, 'bot', $regular_response); | |
| 6994 | - } | |
| 6995 | - | |
| 6996 | 4556 | $response_data = [ |
| 6997 | 4557 | 'text' => $regular_response, |
| 6998 | 4558 | 'html' => '', |
| 6999 | 4559 | 'session_id' => $session_id |
| @@ -7007,45 +4567,16 @@ | ||
| 7007 | 4567 | echo json_encode($response_data); |
| 7008 | 4568 | return true; |
| 7009 | 4569 | } |
| 7010 | 4570 | |
| 7011 | - // Check if this is a GPT-5 model (supports reasoning_effort parameter) | |
| 7012 | - $is_gpt5_model = ( | |
| 7013 | - strpos($selected_model, 'gpt-5') === 0 || | |
| 7014 | - $selected_model === 'gpt-5.2' || | |
| 7015 | - $selected_model === 'gpt-5.1-2025-11-13' || | |
| 7016 | - $selected_model === 'gpt-5' || | |
| 7017 | - $selected_model === 'gpt-5-mini' || | |
| 7018 | - $selected_model === 'gpt-5-nano' | |
| 7019 | - ); | |
| 7020 | - | |
| 7021 | - // Build request body with optimal settings for fast streaming | |
| 7022 | - $request_body = [ | |
| 4571 | + // Prepare the request body with stream: true | |
| 4572 | + $body = json_encode([ | |
| 7023 | 4573 | 'model' => $selected_model, |
| 7024 | 4574 | 'messages' => $formatted_conversation, |
| 7025 | 4575 | 'temperature' => 1, |
| 7026 | 4576 | 'stream' => true |
| 7027 | - ]; | |
| 4577 | + ]); | |
| 7028 | 4578 | |
| 7029 | - // Add reasoning_effort only for GPT-5 models that support it | |
| 7030 | - // These chat models don't support reasoning_effort parameter | |
| 7031 | - $no_reasoning_models = array('gpt-5.2', 'gpt-5.1-chat-latest', 'gpt-5.3-chat-latest', 'gpt-5.4-mini', 'gpt-5.4-nano'); | |
| 7032 | - if ($is_gpt5_model && !in_array($selected_model, $no_reasoning_models, true)) { | |
| 7033 | - // GPT-5.1 uses 'low' instead of 'minimal' | |
| 7034 | - if ($selected_model === 'gpt-5.1-2025-11-13') { | |
| 7035 | - $request_body['reasoning_effort'] = 'low'; | |
| 7036 | - } elseif ($selected_model === 'gpt-5.4') { | |
| 7037 | - $request_body['reasoning_effort'] = 'none'; | |
| 7038 | - } else { | |
| 7039 | - $request_body['reasoning_effort'] = 'minimal'; | |
| 7040 | - } | |
| 7041 | - } | |
| 7042 | - | |
| 7043 | - $body = json_encode($request_body); | |
| 7044 | - | |
| 7045 | - // Setup streaming headers now that we know we're actually streaming | |
| 7046 | - $this->setup_streaming_headers(); | |
| 7047 | - | |
| 7048 | 4579 | // Use cURL for streaming support |
| 7049 | 4580 | $ch = curl_init(); |
| 7050 | 4581 | curl_setopt($ch, CURLOPT_URL, 'https://api.openai.com/v1/chat/completions'); |
| 7051 | 4582 | curl_setopt($ch, CURLOPT_RETURNTRANSFER, false); |
| @@ -7118,11 +4649,10 @@ | ||
| 7118 | 4649 | $response = curl_exec($ch); |
| 7119 | 4650 | $http_code = curl_getinfo($ch, CURLINFO_HTTP_CODE); |
| 7120 | 4651 | |
| 7121 | 4652 | if (curl_errno($ch) || $http_code !== 200) { |
| 7122 | - $curl_error = curl_error($ch); | |
| 7123 | 4653 | curl_close($ch); |
| 7124 | - | |
| 4654 | + | |
| 7125 | 4655 | // Fallback to regular response |
| 7126 | 4656 | $regular_response = $this->mxchat_generate_response_openai( |
| 7127 | 4657 | $selected_model, |
| 7128 | 4658 | $api_key, |
| @@ -7128,34 +4658,19 @@ | ||
| 7128 | 4658 | $api_key, |
| 7129 | 4659 | $conversation_history, |
| 7130 | 4660 | $relevant_content |
| 7131 | 4661 | ); |
| 7132 | - | |
| 7133 | - // FIXED: Check if regular response returned an error | |
| 7134 | - if (is_array($regular_response) && isset($regular_response['error'])) { | |
| 7135 | - // Send error in SSE format since we're in streaming mode | |
| 7136 | - echo "data: " . json_encode([ | |
| 7137 | - 'error' => true, | |
| 7138 | - 'error_message' => $regular_response['error'], | |
| 7139 | - 'error_code' => $regular_response['error_code'] ?? 'api_error', | |
| 7140 | - 'text' => $regular_response['error'], | |
| 7141 | - 'message' => $regular_response['error'] | |
| 7142 | - ]) . "\n\n"; | |
| 7143 | - echo "data: [DONE]\n\n"; | |
| 7144 | - flush(); | |
| 7145 | - return true; | |
| 7146 | - } | |
| 7147 | - | |
| 4662 | + | |
| 7148 | 4663 | $response_data = [ |
| 7149 | 4664 | 'text' => $regular_response, |
| 7150 | 4665 | 'html' => '', |
| 7151 | 4666 | 'session_id' => $session_id |
| 7152 | 4667 | ]; |
| 7153 | - | |
| 4668 | + | |
| 7154 | 4669 | if ($testing_data !== null) { |
| 7155 | 4670 | $response_data['testing_data'] = $testing_data; |
| 7156 | 4671 | } |
| 7157 | - | |
| 4672 | + | |
| 7158 | 4673 | header('Content-Type: application/json'); |
| 7159 | 4674 | echo json_encode($response_data); |
| 7160 | 4675 | return true; |
| 7161 | 4676 | } |
| @@ -7160,36 +4675,16 @@ | ||
| 7160 | 4675 | return true; |
| 7161 | 4676 | } |
| 7162 | 4677 | |
| 7163 | 4678 | curl_close($ch); |
| 7164 | - | |
| 4679 | + | |
| 7165 | 4680 | // Save the complete response to maintain chat persistence |
| 7166 | 4681 | if (!empty($full_response) && !empty($session_id)) { |
| 7167 | - // Prepare RAG context for streaming response | |
| 7168 | - $rag_context_for_storage = null; | |
| 7169 | - $has_rag_data = $this->last_similarity_analysis !== null && !empty($this->last_similarity_analysis['top_matches']); | |
| 7170 | - $has_action_data = isset($this->last_action_analysis) && !empty($this->last_action_analysis); | |
| 7171 | - | |
| 7172 | - if ($has_rag_data || $has_action_data) { | |
| 7173 | - $rag_context_for_storage = []; | |
| 7174 | - | |
| 7175 | - if ($has_rag_data) { | |
| 7176 | - $rag_context_for_storage['top_matches'] = $this->last_similarity_analysis['top_matches']; | |
| 7177 | - $rag_context_for_storage['approved_urls'] = $this->current_valid_urls ?? []; | |
| 7178 | - $rag_context_for_storage['similarity_threshold'] = $this->last_similarity_analysis['threshold_used'] ?? 0.35; | |
| 7179 | - $rag_context_for_storage['knowledge_base_type'] = $this->last_similarity_analysis['knowledge_base_type'] ?? 'WordPress Database'; | |
| 7180 | - $rag_context_for_storage['total_documents_checked'] = $this->last_similarity_analysis['total_checked'] ?? 0; | |
| 7181 | - } | |
| 7182 | - | |
| 7183 | - if ($has_action_data) { | |
| 7184 | - $rag_context_for_storage['action_analysis'] = $this->last_action_analysis; | |
| 7185 | - } | |
| 7186 | - } | |
| 7187 | - $this->mxchat_save_chat_message($session_id, 'bot', $full_response, null, $rag_context_for_storage); | |
| 4682 | + $this->mxchat_save_chat_message($session_id, 'bot', $full_response); | |
| 7188 | 4683 | } |
| 7189 | - | |
| 4684 | + | |
| 7190 | 4685 | return true; // Indicate streaming completed successfully |
| 7191 | - | |
| 4686 | + | |
| 7192 | 4687 | } catch (Exception $e) { |
| 7193 | 4688 | // Fallback to regular response |
| 7194 | 4689 | $regular_response = $this->mxchat_generate_response_openai( |
| 7195 | 4690 | $selected_model, |
| @@ -7196,394 +4691,24 @@ | ||
| 7196 | 4691 | $api_key, |
| 7197 | 4692 | $conversation_history, |
| 7198 | 4693 | $relevant_content |
| 7199 | 4694 | ); |
| 7200 | - | |
| 7201 | - // FIXED: Check if regular response returned an error | |
| 7202 | - if (is_array($regular_response) && isset($regular_response['error'])) { | |
| 7203 | - // Send error in SSE format since we're in streaming mode | |
| 7204 | - echo "data: " . json_encode([ | |
| 7205 | - 'error' => true, | |
| 7206 | - 'error_message' => $regular_response['error'], | |
| 7207 | - 'error_code' => $regular_response['error_code'] ?? 'api_error', | |
| 7208 | - 'text' => $regular_response['error'], | |
| 7209 | - 'message' => $regular_response['error'] | |
| 7210 | - ]) . "\n\n"; | |
| 7211 | - echo "data: [DONE]\n\n"; | |
| 7212 | - flush(); | |
| 7213 | - return true; | |
| 7214 | - } | |
| 7215 | - | |
| 4695 | + | |
| 7216 | 4696 | $response_data = [ |
| 7217 | 4697 | 'text' => $regular_response, |
| 7218 | 4698 | 'html' => '', |
| 7219 | 4699 | 'session_id' => $session_id |
| 7220 | 4700 | ]; |
| 7221 | - | |
| 4701 | + | |
| 7222 | 4702 | if ($testing_data !== null) { |
| 7223 | 4703 | $response_data['testing_data'] = $testing_data; |
| 7224 | 4704 | } |
| 7225 | - | |
| 4705 | + | |
| 7226 | 4706 | header('Content-Type: application/json'); |
| 7227 | 4707 | echo json_encode($response_data); |
| 7228 | 4708 | return true; |
| 7229 | 4709 | } |
| 7230 | 4710 | } |
| 7231 | - | |
| 7232 | -/** | |
| 7233 | - * Generate response using OpenAI Responses API with web search tool | |
| 7234 | - * This uses the newer Responses API which supports web search functionality | |
| 7235 | - */ | |
| 7236 | -private function mxchat_generate_response_openai_web_search($selected_model, $api_key, $conversation_history, $relevant_content, $session_id, $testing_data = null, $streaming = false) { | |
| 7237 | - try { | |
| 7238 | - $bot_id = $this->get_current_bot_id($session_id); | |
| 7239 | - $system_prompt_instructions = $this->get_system_instructions($bot_id, $session_id); | |
| 7240 | - | |
| 7241 | - if (!is_array($conversation_history)) { | |
| 7242 | - $conversation_history = array(); | |
| 7243 | - } | |
| 7244 | - | |
| 7245 | - // Build the input for Responses API | |
| 7246 | - // The Responses API uses a different format - we need to construct the input properly | |
| 7247 | - $input_parts = []; | |
| 7248 | - | |
| 7249 | - // Add system instructions as context | |
| 7250 | - $system_context = $system_prompt_instructions . "\n\n" . $relevant_content; | |
| 7251 | - | |
| 7252 | - // Build conversation as input items for Responses API | |
| 7253 | - foreach ($conversation_history as $message) { | |
| 7254 | - if (is_array($message) && isset($message['role']) && isset($message['content'])) { | |
| 7255 | - $role = $message['role']; | |
| 7256 | - if ($role === 'bot' || $role === 'agent') { | |
| 7257 | - $role = 'assistant'; | |
| 7258 | - } | |
| 7259 | - if (!in_array($role, ['assistant', 'user'])) { | |
| 7260 | - $role = 'user'; | |
| 7261 | - } | |
| 7262 | - $input_parts[] = [ | |
| 7263 | - 'type' => 'message', | |
| 7264 | - 'role' => $role, | |
| 7265 | - 'content' => $message['content'] | |
| 7266 | - ]; | |
| 7267 | - } | |
| 7268 | - } | |
| 7269 | - | |
| 7270 | - // Build request body for Responses API | |
| 7271 | - $request_body = [ | |
| 7272 | - 'model' => $selected_model, | |
| 7273 | - 'input' => $input_parts, | |
| 7274 | - 'instructions' => $system_context, | |
| 7275 | - 'stream' => $streaming | |
| 7276 | - ]; | |
| 7277 | - | |
| 7278 | - // Only add web search tool if web search is enabled in settings | |
| 7279 | - $web_search_enabled = isset($this->options['enable_web_search']) && $this->options['enable_web_search'] === 'on'; | |
| 7280 | - if ($web_search_enabled) { | |
| 7281 | - $request_body['tools'] = [ | |
| 7282 | - ['type' => 'web_search'] | |
| 7283 | - ]; | |
| 7284 | - } | |
| 7285 | - | |
| 7286 | - // Add reasoning effort for supported models | |
| 7287 | - $is_gpt5_model = strpos($selected_model, 'gpt-5') === 0; | |
| 7288 | - $no_reasoning_web = array('gpt-5.2', 'gpt-5.3-chat-latest', 'gpt-5.4-mini', 'gpt-5.4-nano'); | |
| 7289 | - if ($is_gpt5_model && !in_array($selected_model, $no_reasoning_web, true)) { | |
| 7290 | - if ($selected_model === 'gpt-5.1-2025-11-13') { | |
| 7291 | - $request_body['reasoning'] = ['effort' => 'low']; | |
| 7292 | - } elseif ($selected_model === 'gpt-5.4') { | |
| 7293 | - $request_body['reasoning'] = ['effort' => 'low']; | |
| 7294 | - } | |
| 7295 | - } | |
| 7296 | - | |
| 7297 | - //error_log("MXCHAT WEB SEARCH: Request body: " . json_encode($request_body)); | |
| 7298 | - | |
| 7299 | - if ($streaming) { | |
| 7300 | - return $this->mxchat_web_search_streaming_response($request_body, $api_key, $session_id, $testing_data); | |
| 7301 | - } else { | |
| 7302 | - return $this->mxchat_web_search_non_streaming_response($request_body, $api_key, $session_id, $testing_data); | |
| 7303 | - } | |
| 7304 | - | |
| 7305 | - } catch (Exception $e) { | |
| 7306 | - //error_log("MXCHAT WEB SEARCH ERROR: " . $e->getMessage()); | |
| 7307 | - return [ | |
| 7308 | - 'error' => sprintf(esc_html__('Web search error: %s', 'mxchat'), esc_html($e->getMessage())), | |
| 7309 | - 'error_code' => 'web_search_exception' | |
| 7310 | - ]; | |
| 7311 | - } | |
| 7312 | -} | |
| 7313 | - | |
| 7314 | -/** | |
| 7315 | - * Handle non-streaming web search response | |
| 7316 | - */ | |
| 7317 | -private function mxchat_web_search_non_streaming_response($request_body, $api_key, $session_id, $testing_data) { | |
| 7318 | - $request_body['stream'] = false; | |
| 7319 | - | |
| 7320 | - $response = wp_remote_post('https://api.openai.com/v1/responses', array( | |
| 7321 | - 'headers' => array( | |
| 7322 | - 'Authorization' => 'Bearer ' . $api_key, | |
| 7323 | - 'Content-Type' => 'application/json' | |
| 7324 | - ), | |
| 7325 | - 'body' => json_encode($request_body), | |
| 7326 | - 'timeout' => 90 | |
| 7327 | - )); | |
| 7328 | - | |
| 7329 | - if (is_wp_error($response)) { | |
| 7330 | - //error_log("MXCHAT WEB SEARCH ERROR: WP Error: " . $response->get_error_message()); | |
| 7331 | - return [ | |
| 7332 | - 'error' => esc_html__('Failed to connect to OpenAI web search API', 'mxchat'), | |
| 7333 | - 'error_code' => 'web_search_connection_error' | |
| 7334 | - ]; | |
| 7335 | - } | |
| 7336 | - | |
| 7337 | - $response_code = wp_remote_retrieve_response_code($response); | |
| 7338 | - $response_body = wp_remote_retrieve_body($response); | |
| 7339 | - | |
| 7340 | - //error_log("MXCHAT WEB SEARCH: Response code: " . $response_code); | |
| 7341 | - //error_log("MXCHAT WEB SEARCH: Response body (first 2000): " . substr($response_body, 0, 2000)); | |
| 7342 | - | |
| 7343 | - if ($response_code !== 200) { | |
| 7344 | - $error_data = json_decode($response_body, true); | |
| 7345 | - $error_message = $error_data['error']['message'] ?? 'Unknown API error'; | |
| 7346 | - return [ | |
| 7347 | - 'error' => sprintf(esc_html__('OpenAI API error: %s', 'mxchat'), esc_html($error_message)), | |
| 7348 | - 'error_code' => 'web_search_api_error' | |
| 7349 | - ]; | |
| 7350 | - } | |
| 7351 | - | |
| 7352 | - $result = json_decode($response_body, true); | |
| 7353 | - | |
| 7354 | - if (json_last_error() !== JSON_ERROR_NONE) { | |
| 7355 | - return [ | |
| 7356 | - 'error' => esc_html__('Invalid response from OpenAI', 'mxchat'), | |
| 7357 | - 'error_code' => 'web_search_json_error' | |
| 7358 | - ]; | |
| 7359 | - } | |
| 7360 | - | |
| 7361 | - // Extract the response text and citations from Responses API format | |
| 7362 | - $output_text = ''; | |
| 7363 | - $citations = []; | |
| 7364 | - | |
| 7365 | - if (isset($result['output'])) { | |
| 7366 | - foreach ($result['output'] as $output_item) { | |
| 7367 | - if ($output_item['type'] === 'message' && isset($output_item['content'])) { | |
| 7368 | - foreach ($output_item['content'] as $content_item) { | |
| 7369 | - if ($content_item['type'] === 'output_text') { | |
| 7370 | - $output_text .= $content_item['text']; | |
| 7371 | - | |
| 7372 | - // Extract citations/annotations | |
| 7373 | - if (isset($content_item['annotations'])) { | |
| 7374 | - foreach ($content_item['annotations'] as $annotation) { | |
| 7375 | - if ($annotation['type'] === 'url_citation') { | |
| 7376 | - $citations[] = [ | |
| 7377 | - 'url' => $annotation['url'], | |
| 7378 | - 'title' => $annotation['title'] ?? '' | |
| 7379 | - ]; | |
| 7380 | - } | |
| 7381 | - } | |
| 7382 | - } | |
| 7383 | - } | |
| 7384 | - } | |
| 7385 | - } | |
| 7386 | - } | |
| 7387 | - } | |
| 7388 | - | |
| 7389 | - // If we have citations, append them to the response | |
| 7390 | - if (!empty($citations)) { | |
| 7391 | - $output_text .= "\n\n**Sources:**\n"; | |
| 7392 | - $seen_urls = []; | |
| 7393 | - foreach ($citations as $citation) { | |
| 7394 | - if (!in_array($citation['url'], $seen_urls)) { | |
| 7395 | - $seen_urls[] = $citation['url']; | |
| 7396 | - $title = !empty($citation['title']) ? $citation['title'] : $citation['url']; | |
| 7397 | - $output_text .= "- [" . $title . "](" . $citation['url'] . ")\n"; | |
| 7398 | - } | |
| 7399 | - } | |
| 7400 | - } | |
| 7401 | - | |
| 7402 | - // Transcript save is handled by the main handler (mxchat_handle_chat_request) | |
| 7403 | - // which includes rag_context for the "sources" link in transcripts. | |
| 7404 | - | |
| 7405 | - return $output_text; | |
| 7406 | -} | |
| 7407 | - | |
| 7408 | -/** | |
| 7409 | - * Handle streaming web search response using Responses API | |
| 7410 | - */ | |
| 7411 | -private function mxchat_web_search_streaming_response($request_body, $api_key, $session_id, $testing_data) { | |
| 7412 | - $request_body['stream'] = true; | |
| 7413 | - | |
| 7414 | - // Check if we can stream | |
| 7415 | - if (headers_sent() || !function_exists('curl_init')) { | |
| 7416 | - // Fallback to non-streaming | |
| 7417 | - return $this->mxchat_web_search_non_streaming_response($request_body, $api_key, $session_id, $testing_data); | |
| 7418 | - } | |
| 7419 | - | |
| 7420 | - // Setup streaming headers | |
| 7421 | - $this->setup_streaming_headers(); | |
| 7422 | - | |
| 7423 | - $ch = curl_init(); | |
| 7424 | - curl_setopt($ch, CURLOPT_URL, 'https://api.openai.com/v1/responses'); | |
| 7425 | - curl_setopt($ch, CURLOPT_RETURNTRANSFER, false); | |
| 7426 | - curl_setopt($ch, CURLOPT_POST, true); | |
| 7427 | - curl_setopt($ch, CURLOPT_POSTFIELDS, json_encode($request_body)); | |
| 7428 | - curl_setopt($ch, CURLOPT_HTTPHEADER, array( | |
| 7429 | - 'Content-Type: application/json', | |
| 7430 | - 'Authorization: Bearer ' . $api_key | |
| 7431 | - )); | |
| 7432 | - curl_setopt($ch, CURLOPT_SSL_VERIFYPEER, true); | |
| 7433 | - curl_setopt($ch, CURLOPT_TIMEOUT, 120); | |
| 7434 | - | |
| 7435 | - $full_response = ''; | |
| 7436 | - $stream_started = false; | |
| 7437 | - $buffer = ''; | |
| 7438 | - $citations = []; | |
| 7439 | - | |
| 7440 | - curl_setopt($ch, CURLOPT_WRITEFUNCTION, function($ch, $data) use (&$full_response, &$stream_started, &$buffer, &$citations, $testing_data) { | |
| 7441 | - // Send testing data as first event if available | |
| 7442 | - if (!$stream_started && $testing_data !== null) { | |
| 7443 | - echo "data: " . json_encode(['testing_data' => $testing_data]) . "\n\n"; | |
| 7444 | - flush(); | |
| 7445 | - $stream_started = true; | |
| 7446 | - } | |
| 7447 | - | |
| 7448 | - $buffer .= $data; | |
| 7449 | - $lines = explode("\n", $buffer); | |
| 7450 | - $buffer = array_pop($lines); | |
| 7451 | - | |
| 7452 | - foreach ($lines as $line) { | |
| 7453 | - if (trim($line) === '') continue; | |
| 7454 | - if (strpos($line, 'data: ') !== 0) continue; | |
| 7455 | - | |
| 7456 | - $json_str = substr($line, 6); | |
| 7457 | - | |
| 7458 | - if (trim($json_str) === '[DONE]') { | |
| 7459 | - // Append citations if we have any | |
| 7460 | - if (!empty($citations)) { | |
| 7461 | - $citation_text = "\n\n**Sources:**\n"; | |
| 7462 | - $seen_urls = []; | |
| 7463 | - foreach ($citations as $citation) { | |
| 7464 | - if (!in_array($citation['url'], $seen_urls)) { | |
| 7465 | - $seen_urls[] = $citation['url']; | |
| 7466 | - $title = !empty($citation['title']) ? $citation['title'] : $citation['url']; | |
| 7467 | - $citation_text .= "- [" . $title . "](" . $citation['url'] . ")\n"; | |
| 7468 | - } | |
| 7469 | - } | |
| 7470 | - echo "data: " . json_encode(['content' => $citation_text]) . "\n\n"; | |
| 7471 | - $full_response .= $citation_text; | |
| 7472 | - flush(); | |
| 7473 | - } | |
| 7474 | - echo "data: [DONE]\n\n"; | |
| 7475 | - flush(); | |
| 7476 | - continue; | |
| 7477 | - } | |
| 7478 | - | |
| 7479 | - $json = json_decode(trim($json_str), true); | |
| 7480 | - if (!$json) continue; | |
| 7481 | - | |
| 7482 | - // Handle Responses API streaming events | |
| 7483 | - // The format is different from Chat Completions | |
| 7484 | - if (isset($json['type'])) { | |
| 7485 | - switch ($json['type']) { | |
| 7486 | - case 'response.output_text.delta': | |
| 7487 | - // Text content delta | |
| 7488 | - if (isset($json['delta'])) { | |
| 7489 | - $content = $json['delta']; | |
| 7490 | - $full_response .= $content; | |
| 7491 | - echo "data: " . json_encode(['content' => $content]) . "\n\n"; | |
| 7492 | - flush(); | |
| 7493 | - } | |
| 7494 | - break; | |
| 7495 | - | |
| 7496 | - case 'response.output_item.done': | |
| 7497 | - // Check for citations in completed items | |
| 7498 | - if (isset($json['item']['content'])) { | |
| 7499 | - foreach ($json['item']['content'] as $content_item) { | |
| 7500 | - if (isset($content_item['annotations'])) { | |
| 7501 | - foreach ($content_item['annotations'] as $annotation) { | |
| 7502 | - if ($annotation['type'] === 'url_citation') { | |
| 7503 | - $citations[] = [ | |
| 7504 | - 'url' => $annotation['url'], | |
| 7505 | - 'title' => $annotation['title'] ?? '' | |
| 7506 | - ]; | |
| 7507 | - } | |
| 7508 | - } | |
| 7509 | - } | |
| 7510 | - } | |
| 7511 | - } | |
| 7512 | - break; | |
| 7513 | - } | |
| 7514 | - } | |
| 7515 | - } | |
| 7516 | - | |
| 7517 | - return strlen($data); | |
| 7518 | - }); | |
| 7519 | - | |
| 7520 | - $response = curl_exec($ch); | |
| 7521 | - $http_code = curl_getinfo($ch, CURLINFO_HTTP_CODE); | |
| 7522 | - | |
| 7523 | - if (curl_errno($ch) || $http_code !== 200) { | |
| 7524 | - $curl_error = curl_error($ch); | |
| 7525 | - curl_close($ch); | |
| 7526 | - | |
| 7527 | - //error_log("MXCHAT WEB SEARCH STREAM ERROR: HTTP $http_code, cURL error: $curl_error"); | |
| 7528 | - | |
| 7529 | - // Fallback to non-streaming | |
| 7530 | - $fallback_response = $this->mxchat_web_search_non_streaming_response($request_body, $api_key, $session_id, $testing_data); | |
| 7531 | - | |
| 7532 | - if (is_array($fallback_response) && isset($fallback_response['error'])) { | |
| 7533 | - echo "data: " . json_encode([ | |
| 7534 | - 'error' => true, | |
| 7535 | - 'error_message' => $fallback_response['error'], | |
| 7536 | - 'error_code' => $fallback_response['error_code'] ?? 'web_search_error' | |
| 7537 | - ]) . "\n\n"; | |
| 7538 | - echo "data: [DONE]\n\n"; | |
| 7539 | - flush(); | |
| 7540 | - return true; | |
| 7541 | - } | |
| 7542 | - | |
| 7543 | - $response_data = [ | |
| 7544 | - 'text' => $fallback_response, | |
| 7545 | - 'html' => '', | |
| 7546 | - 'session_id' => $session_id | |
| 7547 | - ]; | |
| 7548 | - if ($testing_data !== null) { | |
| 7549 | - $response_data['testing_data'] = $testing_data; | |
| 7550 | - } | |
| 7551 | - header('Content-Type: application/json'); | |
| 7552 | - echo json_encode($response_data); | |
| 7553 | - return true; | |
| 7554 | - } | |
| 7555 | - | |
| 7556 | - curl_close($ch); | |
| 7557 | - | |
| 7558 | - // Save the complete response with RAG context so the "sources" link | |
| 7559 | - // appears in transcripts — mirrors the pattern used by Claude/OpenAI streaming. | |
| 7560 | - if (!empty($full_response) && !empty($session_id)) { | |
| 7561 | - $rag_context_for_storage = null; | |
| 7562 | - $has_rag_data = $this->last_similarity_analysis !== null && !empty($this->last_similarity_analysis['top_matches']); | |
| 7563 | - $has_action_data = isset($this->last_action_analysis) && !empty($this->last_action_analysis); | |
| 7564 | - | |
| 7565 | - if ($has_rag_data || $has_action_data) { | |
| 7566 | - $rag_context_for_storage = []; | |
| 7567 | - | |
| 7568 | - if ($has_rag_data) { | |
| 7569 | - $rag_context_for_storage['top_matches'] = $this->last_similarity_analysis['top_matches']; | |
| 7570 | - $rag_context_for_storage['approved_urls'] = $this->current_valid_urls ?? []; | |
| 7571 | - $rag_context_for_storage['similarity_threshold'] = $this->last_similarity_analysis['threshold_used'] ?? 0.35; | |
| 7572 | - $rag_context_for_storage['knowledge_base_type'] = $this->last_similarity_analysis['knowledge_base_type'] ?? 'WordPress Database'; | |
| 7573 | - $rag_context_for_storage['total_documents_checked'] = $this->last_similarity_analysis['total_checked'] ?? 0; | |
| 7574 | - } | |
| 7575 | - | |
| 7576 | - if ($has_action_data) { | |
| 7577 | - $rag_context_for_storage['action_analysis'] = $this->last_action_analysis; | |
| 7578 | - } | |
| 7579 | - } | |
| 7580 | - $this->mxchat_save_chat_message($session_id, 'bot', $full_response, null, $rag_context_for_storage); | |
| 7581 | - } | |
| 7582 | - | |
| 7583 | - return true; | |
| 7584 | -} | |
| 7585 | - | |
| 7586 | 4711 | private function mxchat_generate_response_claude_stream($selected_model, $claude_api_key, $conversation_history, $relevant_content, $session_id, $testing_data = null) { |
| 7587 | 4712 | try { |
| 7588 | 4713 | // Get bot ID from session or request |
| 7589 | 4714 | $bot_id = $this->get_current_bot_id($session_id); |
| @@ -7588,9 +4713,9 @@ | ||
| 7588 | 4713 | // Get bot ID from session or request |
| 7589 | 4714 | $bot_id = $this->get_current_bot_id($session_id); |
| 7590 | 4715 | |
| 7591 | 4716 | // Get system prompt instructions using centralized function |
| 7592 | - $system_prompt_instructions = $this->get_system_instructions($bot_id, $session_id); | |
| 4717 | + $system_prompt_instructions = $this->get_system_instructions($bot_id); | |
| 7593 | 4718 | // Ensure conversation_history is an array |
| 7594 | 4719 | if (!is_array($conversation_history)) { |
| 7595 | 4720 | $conversation_history = array(); |
| 7596 | 4721 | } |
| @@ -7642,13 +4767,8 @@ | ||
| 7642 | 4767 | array_slice($conversation_history, 0, -1), // Remove the added content |
| 7643 | 4768 | $relevant_content |
| 7644 | 4769 | ); |
| 7645 | 4770 | |
| 7646 | - // Save bot response to transcript | |
| 7647 | - if (!empty($regular_response) && !empty($session_id)) { | |
| 7648 | - $this->mxchat_save_chat_message($session_id, 'bot', $regular_response); | |
| 7649 | - } | |
| 7650 | - | |
| 7651 | 4771 | // Return as JSON with testing data |
| 7652 | 4772 | $response_data = [ |
| 7653 | 4773 | 'text' => $regular_response, |
| 7654 | 4774 | 'html' => '', |
| @@ -7667,11 +4787,8 @@ | ||
| 7667 | 4787 | echo json_encode($response_data); |
| 7668 | 4788 | return true; // Indicate we handled the response |
| 7669 | 4789 | } |
| 7670 | 4790 | |
| 7671 | - // Setup streaming headers now that we know we're actually streaming | |
| 7672 | - $this->setup_streaming_headers(); | |
| 7673 | - | |
| 7674 | 4791 | // Use cURL for streaming support |
| 7675 | 4792 | $ch = curl_init(); |
| 7676 | 4793 | curl_setopt($ch, CURLOPT_URL, 'https://api.anthropic.com/v1/messages'); |
| 7677 | 4794 | curl_setopt($ch, CURLOPT_RETURNTRANSFER, false); |
| @@ -7776,35 +4893,20 @@ | ||
| 7776 | 4893 | $claude_api_key, |
| 7777 | 4894 | array_slice($conversation_history, 0, -1), // Remove the added content |
| 7778 | 4895 | $relevant_content |
| 7779 | 4896 | ); |
| 7780 | - | |
| 7781 | - // FIXED: Check if regular response returned an error | |
| 7782 | - if (is_array($regular_response) && isset($regular_response['error'])) { | |
| 7783 | - // Send error in SSE format since we're in streaming mode | |
| 7784 | - echo "data: " . json_encode([ | |
| 7785 | - 'error' => true, | |
| 7786 | - 'error_message' => $regular_response['error'], | |
| 7787 | - 'error_code' => $regular_response['error_code'] ?? 'api_error', | |
| 7788 | - 'text' => $regular_response['error'], | |
| 7789 | - 'message' => $regular_response['error'] | |
| 7790 | - ]) . "\n\n"; | |
| 7791 | - echo "data: [DONE]\n\n"; | |
| 7792 | - flush(); | |
| 7793 | - return true; | |
| 7794 | - } | |
| 7795 | - | |
| 4897 | + | |
| 7796 | 4898 | $response_data = [ |
| 7797 | 4899 | 'text' => $regular_response, |
| 7798 | 4900 | 'html' => '', |
| 7799 | 4901 | 'session_id' => $session_id |
| 7800 | 4902 | ]; |
| 7801 | - | |
| 4903 | + | |
| 7802 | 4904 | if ($testing_data !== null) { |
| 7803 | 4905 | $response_data['testing_data'] = $testing_data; |
| 7804 | 4906 | //error_log("MxChat Testing: Added testing data to Claude error fallback"); |
| 7805 | 4907 | } |
| 7806 | - | |
| 4908 | + | |
| 7807 | 4909 | header('Content-Type: application/json'); |
| 7808 | 4910 | echo json_encode($response_data); |
| 7809 | 4911 | return true; |
| 7810 | 4912 | } |
| @@ -7810,29 +4912,9 @@ | ||
| 7810 | 4912 | } |
| 7811 | 4913 | |
| 7812 | 4914 | // Save the complete response to maintain chat persistence |
| 7813 | 4915 | if (!empty($full_response) && !empty($session_id)) { |
| 7814 | - // Prepare RAG context for streaming response | |
| 7815 | - $rag_context_for_storage = null; | |
| 7816 | - $has_rag_data = $this->last_similarity_analysis !== null && !empty($this->last_similarity_analysis['top_matches']); | |
| 7817 | - $has_action_data = isset($this->last_action_analysis) && !empty($this->last_action_analysis); | |
| 7818 | - | |
| 7819 | - if ($has_rag_data || $has_action_data) { | |
| 7820 | - $rag_context_for_storage = []; | |
| 7821 | - | |
| 7822 | - if ($has_rag_data) { | |
| 7823 | - $rag_context_for_storage['top_matches'] = $this->last_similarity_analysis['top_matches']; | |
| 7824 | - $rag_context_for_storage['approved_urls'] = $this->current_valid_urls ?? []; | |
| 7825 | - $rag_context_for_storage['similarity_threshold'] = $this->last_similarity_analysis['threshold_used'] ?? 0.35; | |
| 7826 | - $rag_context_for_storage['knowledge_base_type'] = $this->last_similarity_analysis['knowledge_base_type'] ?? 'WordPress Database'; | |
| 7827 | - $rag_context_for_storage['total_documents_checked'] = $this->last_similarity_analysis['total_checked'] ?? 0; | |
| 7828 | - } | |
| 7829 | - | |
| 7830 | - if ($has_action_data) { | |
| 7831 | - $rag_context_for_storage['action_analysis'] = $this->last_action_analysis; | |
| 7832 | - } | |
| 7833 | - } | |
| 7834 | - $this->mxchat_save_chat_message($session_id, 'bot', $full_response, null, $rag_context_for_storage); | |
| 4916 | + $this->mxchat_save_chat_message($session_id, 'bot', $full_response); | |
| 7835 | 4917 | } |
| 7836 | 4918 | |
| 7837 | 4919 | return true; // Indicate streaming completed successfully |
| 7838 | 4920 | |
| @@ -7837,9 +4919,9 @@ | ||
| 7837 | 4919 | return true; // Indicate streaming completed successfully |
| 7838 | 4920 | |
| 7839 | 4921 | } catch (Exception $e) { |
| 7840 | 4922 | //error_log("MxChat Claude streaming exception: " . $e->getMessage()); |
| 7841 | - | |
| 4923 | + | |
| 7842 | 4924 | // Fallback to regular response on exception |
| 7843 | 4925 | $regular_response = $this->mxchat_generate_response_claude( |
| 7844 | 4926 | $selected_model, |
| 7845 | 4927 | $claude_api_key, |
| @@ -7845,35 +4927,20 @@ | ||
| 7845 | 4927 | $claude_api_key, |
| 7846 | 4928 | $conversation_history, |
| 7847 | 4929 | $relevant_content |
| 7848 | 4930 | ); |
| 7849 | - | |
| 7850 | - // FIXED: Check if regular response returned an error | |
| 7851 | - if (is_array($regular_response) && isset($regular_response['error'])) { | |
| 7852 | - // Send error in SSE format since we're in streaming mode | |
| 7853 | - echo "data: " . json_encode([ | |
| 7854 | - 'error' => true, | |
| 7855 | - 'error_message' => $regular_response['error'], | |
| 7856 | - 'error_code' => $regular_response['error_code'] ?? 'api_error', | |
| 7857 | - 'text' => $regular_response['error'], | |
| 7858 | - 'message' => $regular_response['error'] | |
| 7859 | - ]) . "\n\n"; | |
| 7860 | - echo "data: [DONE]\n\n"; | |
| 7861 | - flush(); | |
| 7862 | - return true; | |
| 7863 | - } | |
| 7864 | - | |
| 4931 | + | |
| 7865 | 4932 | $response_data = [ |
| 7866 | 4933 | 'text' => $regular_response, |
| 7867 | 4934 | 'html' => '', |
| 7868 | 4935 | 'session_id' => $session_id |
| 7869 | 4936 | ]; |
| 7870 | - | |
| 4937 | + | |
| 7871 | 4938 | if ($testing_data !== null) { |
| 7872 | 4939 | $response_data['testing_data'] = $testing_data; |
| 7873 | 4940 | //error_log("MxChat Testing: Added testing data to Claude exception fallback"); |
| 7874 | 4941 | } |
| 7875 | - | |
| 4942 | + | |
| 7876 | 4943 | header('Content-Type: application/json'); |
| 7877 | 4944 | echo json_encode($response_data); |
| 7878 | 4945 | return true; |
| 7879 | 4946 | } |
| @@ -7883,9 +4950,9 @@ | ||
| 7883 | 4950 | // Get bot ID from session or request |
| 7884 | 4951 | $bot_id = $this->get_current_bot_id($session_id); |
| 7885 | 4952 | |
| 7886 | 4953 | // Get system prompt instructions using centralized function |
| 7887 | - $system_prompt_instructions = $this->get_system_instructions($bot_id, $session_id); | |
| 4954 | + $system_prompt_instructions = $this->get_system_instructions($bot_id); | |
| 7888 | 4955 | |
| 7889 | 4956 | // Ensure conversation_history is an array |
| 7890 | 4957 | if (!is_array($conversation_history)) { |
| 7891 | 4958 | $conversation_history = array(); |
| @@ -7925,13 +4992,8 @@ | ||
| 7925 | 4992 | $conversation_history, |
| 7926 | 4993 | $relevant_content |
| 7927 | 4994 | ); |
| 7928 | 4995 | |
| 7929 | - // Save bot response to transcript | |
| 7930 | - if (!empty($regular_response) && !empty($session_id)) { | |
| 7931 | - $this->mxchat_save_chat_message($session_id, 'bot', $regular_response); | |
| 7932 | - } | |
| 7933 | - | |
| 7934 | 4996 | $response_data = [ |
| 7935 | 4997 | 'text' => $regular_response, |
| 7936 | 4998 | 'html' => '', |
| 7937 | 4999 | 'session_id' => $session_id |
| @@ -7954,11 +5016,8 @@ | ||
| 7954 | 5016 | 'temperature' => 0.8, |
| 7955 | 5017 | 'stream' => true |
| 7956 | 5018 | ]); |
| 7957 | 5019 | |
| 7958 | - // Setup streaming headers now that we know we're actually streaming | |
| 7959 | - $this->setup_streaming_headers(); | |
| 7960 | - | |
| 7961 | 5020 | // Use cURL for streaming support |
| 7962 | 5021 | $ch = curl_init(); |
| 7963 | 5022 | curl_setopt($ch, CURLOPT_URL, 'https://api.x.ai/v1/chat/completions'); |
| 7964 | 5023 | curl_setopt($ch, CURLOPT_RETURNTRANSFER, false); |
| @@ -8059,39 +5118,19 @@ | ||
| 8059 | 5118 | return true; |
| 8060 | 5119 | } |
| 8061 | 5120 | |
| 8062 | 5121 | curl_close($ch); |
| 8063 | - | |
| 5122 | + | |
| 8064 | 5123 | // Save the complete response to maintain chat persistence |
| 8065 | 5124 | if (!empty($full_response) && !empty($session_id)) { |
| 8066 | - // Prepare RAG context for streaming response | |
| 8067 | - $rag_context_for_storage = null; | |
| 8068 | - $has_rag_data = $this->last_similarity_analysis !== null && !empty($this->last_similarity_analysis['top_matches']); | |
| 8069 | - $has_action_data = isset($this->last_action_analysis) && !empty($this->last_action_analysis); | |
| 8070 | - | |
| 8071 | - if ($has_rag_data || $has_action_data) { | |
| 8072 | - $rag_context_for_storage = []; | |
| 8073 | - | |
| 8074 | - if ($has_rag_data) { | |
| 8075 | - $rag_context_for_storage['top_matches'] = $this->last_similarity_analysis['top_matches']; | |
| 8076 | - $rag_context_for_storage['approved_urls'] = $this->current_valid_urls ?? []; | |
| 8077 | - $rag_context_for_storage['similarity_threshold'] = $this->last_similarity_analysis['threshold_used'] ?? 0.35; | |
| 8078 | - $rag_context_for_storage['knowledge_base_type'] = $this->last_similarity_analysis['knowledge_base_type'] ?? 'WordPress Database'; | |
| 8079 | - $rag_context_for_storage['total_documents_checked'] = $this->last_similarity_analysis['total_checked'] ?? 0; | |
| 8080 | - } | |
| 8081 | - | |
| 8082 | - if ($has_action_data) { | |
| 8083 | - $rag_context_for_storage['action_analysis'] = $this->last_action_analysis; | |
| 8084 | - } | |
| 8085 | - } | |
| 8086 | - $this->mxchat_save_chat_message($session_id, 'bot', $full_response, null, $rag_context_for_storage); | |
| 5125 | + $this->mxchat_save_chat_message($session_id, 'bot', $full_response); | |
| 8087 | 5126 | } |
| 8088 | - | |
| 5127 | + | |
| 8089 | 5128 | return true; // Indicate streaming completed successfully |
| 8090 | - | |
| 5129 | + | |
| 8091 | 5130 | } catch (Exception $e) { |
| 8092 | 5131 | //error_log("MxChat X.AI streaming exception: " . $e->getMessage()); |
| 8093 | - | |
| 5132 | + | |
| 8094 | 5133 | // Fallback to regular response |
| 8095 | 5134 | $regular_response = $this->mxchat_generate_response_xai( |
| 8096 | 5135 | $selected_model, |
| 8097 | 5136 | $xai_api_key, |
| @@ -8120,9 +5159,9 @@ | ||
| 8120 | 5159 | // Get bot ID from session or request |
| 8121 | 5160 | $bot_id = $this->get_current_bot_id($session_id); |
| 8122 | 5161 | |
| 8123 | 5162 | // Get system prompt instructions using centralized function |
| 8124 | - $system_prompt_instructions = $this->get_system_instructions($bot_id, $session_id); | |
| 5163 | + $system_prompt_instructions = $this->get_system_instructions($bot_id); | |
| 8125 | 5164 | |
| 8126 | 5165 | // Ensure conversation_history is an array |
| 8127 | 5166 | if (!is_array($conversation_history)) { |
| 8128 | 5167 | $conversation_history = array(); |
| @@ -8162,13 +5201,8 @@ | ||
| 8162 | 5201 | $conversation_history, |
| 8163 | 5202 | $relevant_content |
| 8164 | 5203 | ); |
| 8165 | 5204 | |
| 8166 | - // Save bot response to transcript | |
| 8167 | - if (!empty($regular_response) && !empty($session_id)) { | |
| 8168 | - $this->mxchat_save_chat_message($session_id, 'bot', $regular_response); | |
| 8169 | - } | |
| 8170 | - | |
| 8171 | 5205 | $response_data = [ |
| 8172 | 5206 | 'text' => $regular_response, |
| 8173 | 5207 | 'html' => '', |
| 8174 | 5208 | 'session_id' => $session_id |
| @@ -8191,11 +5225,8 @@ | ||
| 8191 | 5225 | 'temperature' => 0.8, |
| 8192 | 5226 | 'stream' => true |
| 8193 | 5227 | ]); |
| 8194 | 5228 | |
| 8195 | - // Setup streaming headers now that we know we're actually streaming | |
| 8196 | - $this->setup_streaming_headers(); | |
| 8197 | - | |
| 8198 | 5229 | // Use cURL for streaming support |
| 8199 | 5230 | $ch = curl_init(); |
| 8200 | 5231 | curl_setopt($ch, CURLOPT_URL, 'https://api.deepseek.com/v1/chat/completions'); |
| 8201 | 5232 | curl_setopt($ch, CURLOPT_RETURNTRANSFER, false); |
| @@ -8310,39 +5341,19 @@ | ||
| 8310 | 5341 | return true; |
| 8311 | 5342 | } |
| 8312 | 5343 | |
| 8313 | 5344 | curl_close($ch); |
| 8314 | - | |
| 5345 | + | |
| 8315 | 5346 | // Save the complete response to maintain chat persistence |
| 8316 | 5347 | if (!empty($full_response) && !empty($session_id)) { |
| 8317 | - // Prepare RAG context for streaming response | |
| 8318 | - $rag_context_for_storage = null; | |
| 8319 | - $has_rag_data = $this->last_similarity_analysis !== null && !empty($this->last_similarity_analysis['top_matches']); | |
| 8320 | - $has_action_data = isset($this->last_action_analysis) && !empty($this->last_action_analysis); | |
| 8321 | - | |
| 8322 | - if ($has_rag_data || $has_action_data) { | |
| 8323 | - $rag_context_for_storage = []; | |
| 8324 | - | |
| 8325 | - if ($has_rag_data) { | |
| 8326 | - $rag_context_for_storage['top_matches'] = $this->last_similarity_analysis['top_matches']; | |
| 8327 | - $rag_context_for_storage['approved_urls'] = $this->current_valid_urls ?? []; | |
| 8328 | - $rag_context_for_storage['similarity_threshold'] = $this->last_similarity_analysis['threshold_used'] ?? 0.35; | |
| 8329 | - $rag_context_for_storage['knowledge_base_type'] = $this->last_similarity_analysis['knowledge_base_type'] ?? 'WordPress Database'; | |
| 8330 | - $rag_context_for_storage['total_documents_checked'] = $this->last_similarity_analysis['total_checked'] ?? 0; | |
| 8331 | - } | |
| 8332 | - | |
| 8333 | - if ($has_action_data) { | |
| 8334 | - $rag_context_for_storage['action_analysis'] = $this->last_action_analysis; | |
| 8335 | - } | |
| 8336 | - } | |
| 8337 | - $this->mxchat_save_chat_message($session_id, 'bot', $full_response, null, $rag_context_for_storage); | |
| 5348 | + $this->mxchat_save_chat_message($session_id, 'bot', $full_response); | |
| 8338 | 5349 | } |
| 8339 | - | |
| 5350 | + | |
| 8340 | 5351 | return true; // Indicate streaming completed successfully |
| 8341 | - | |
| 5352 | + | |
| 8342 | 5353 | } catch (Exception $e) { |
| 8343 | 5354 | //error_log("MxChat DeepSeek streaming exception: " . $e->getMessage()); |
| 8344 | - | |
| 5355 | + | |
| 8345 | 5356 | // Fallback to regular response |
| 8346 | 5357 | $regular_response = $this->mxchat_generate_response_deepseek( |
| 8347 | 5358 | $selected_model, |
| 8348 | 5359 | $deepseek_api_key, |
| @@ -8376,112 +5387,8 @@ | ||
| 8376 | 5387 | return true; |
| 8377 | 5388 | } |
| 8378 | 5389 | } |
| 8379 | 5390 | |
| 8380 | - | |
| 8381 | -private function mxchat_generate_response_openrouter($selected_model, $openrouter_api_key, $conversation_history, $relevant_content) { | |
| 8382 | - try { | |
| 8383 | - if (!is_array($conversation_history)) { | |
| 8384 | - $conversation_history = array(); | |
| 8385 | - } | |
| 8386 | - | |
| 8387 | - $bot_id = $this->get_current_bot_id(''); | |
| 8388 | - $system_prompt_instructions = $this->get_system_instructions($bot_id, $session_id); | |
| 8389 | - | |
| 8390 | - $formatted_conversation = array(); | |
| 8391 | - | |
| 8392 | - $formatted_conversation[] = array( | |
| 8393 | - 'role' => 'system', | |
| 8394 | - 'content' => $system_prompt_instructions . " " . $relevant_content | |
| 8395 | - ); | |
| 8396 | - | |
| 8397 | - foreach ($conversation_history as $message) { | |
| 8398 | - if (is_array($message) && isset($message['role']) && isset($message['content'])) { | |
| 8399 | - $role = $message['role']; | |
| 8400 | - | |
| 8401 | - if ($role === 'bot' || $role === 'agent') { | |
| 8402 | - $role = 'assistant'; | |
| 8403 | - } | |
| 8404 | - if (!in_array($role, ['system', 'assistant', 'user'])) { | |
| 8405 | - $role = 'user'; | |
| 8406 | - } | |
| 8407 | - | |
| 8408 | - $formatted_conversation[] = array( | |
| 8409 | - 'role' => $role, | |
| 8410 | - 'content' => $message['content'] | |
| 8411 | - ); | |
| 8412 | - } | |
| 8413 | - } | |
| 8414 | - | |
| 8415 | - $body = json_encode([ | |
| 8416 | - 'model' => $selected_model, | |
| 8417 | - 'messages' => $formatted_conversation, | |
| 8418 | - 'temperature' => 1, | |
| 8419 | - ]); | |
| 8420 | - | |
| 8421 | - $args = [ | |
| 8422 | - 'body' => $body, | |
| 8423 | - 'headers' => [ | |
| 8424 | - 'Content-Type' => 'application/json', | |
| 8425 | - 'Authorization' => 'Bearer ' . $openrouter_api_key, | |
| 8426 | - 'HTTP-Referer' => home_url(), | |
| 8427 | - 'X-Title' => get_bloginfo('name'), | |
| 8428 | - ], | |
| 8429 | - 'timeout' => 60, | |
| 8430 | - 'redirection' => 5, | |
| 8431 | - 'blocking' => true, | |
| 8432 | - 'httpversion' => '1.0', | |
| 8433 | - 'sslverify' => true, | |
| 8434 | - ]; | |
| 8435 | - | |
| 8436 | - $response = wp_remote_post('https://openrouter.ai/api/v1/chat/completions', $args); | |
| 8437 | - | |
| 8438 | - if (is_wp_error($response)) { | |
| 8439 | - $error_message = $response->get_error_message(); | |
| 8440 | - return [ | |
| 8441 | - 'error' => esc_html__('Connection error when contacting OpenRouter: ', 'mxchat') . esc_html($error_message), | |
| 8442 | - 'error_code' => 'openrouter_connection_error', | |
| 8443 | - 'provider' => 'openrouter' | |
| 8444 | - ]; | |
| 8445 | - } | |
| 8446 | - | |
| 8447 | - $status_code = wp_remote_retrieve_response_code($response); | |
| 8448 | - if ($status_code !== 200) { | |
| 8449 | - $response_body = wp_remote_retrieve_body($response); | |
| 8450 | - $decoded_response = json_decode($response_body, true); | |
| 8451 | - | |
| 8452 | - $error_message = isset($decoded_response['error']['message']) | |
| 8453 | - ? $decoded_response['error']['message'] | |
| 8454 | - : 'HTTP Error ' . $status_code; | |
| 8455 | - | |
| 8456 | - return [ | |
| 8457 | - 'error' => esc_html__('OpenRouter API error: ', 'mxchat') . esc_html($error_message), | |
| 8458 | - 'error_code' => 'openrouter_api_error', | |
| 8459 | - 'provider' => 'openrouter', | |
| 8460 | - 'status_code' => $status_code | |
| 8461 | - ]; | |
| 8462 | - } | |
| 8463 | - | |
| 8464 | - $response_body = wp_remote_retrieve_body($response); | |
| 8465 | - $decoded_response = json_decode($response_body, true); | |
| 8466 | - | |
| 8467 | - if (isset($decoded_response['choices'][0]['message']['content'])) { | |
| 8468 | - return trim($decoded_response['choices'][0]['message']['content']); | |
| 8469 | - } else { | |
| 8470 | - return [ | |
| 8471 | - 'error' => esc_html__('Unexpected response format from OpenRouter.', 'mxchat'), | |
| 8472 | - 'error_code' => 'openrouter_response_format_error', | |
| 8473 | - 'provider' => 'openrouter' | |
| 8474 | - ]; | |
| 8475 | - } | |
| 8476 | - } catch (Exception $e) { | |
| 8477 | - return [ | |
| 8478 | - 'error' => esc_html__('System error when processing OpenRouter request: ', 'mxchat') . esc_html($e->getMessage()), | |
| 8479 | - 'error_code' => 'openrouter_exception', | |
| 8480 | - 'provider' => 'openrouter' | |
| 8481 | - ]; | |
| 8482 | - } | |
| 8483 | -} | |
| 8484 | 5391 | private function mxchat_generate_response_claude($selected_model, $claude_api_key, $conversation_history, $relevant_content) { |
| 8485 | 5392 | |
| 8486 | 5393 | // Get bot ID from session or request |
| 8487 | 5394 | $bot_id = $this->get_current_bot_id($session_id); |
| @@ -8486,9 +5393,9 @@ | ||
| 8486 | 5393 | // Get bot ID from session or request |
| 8487 | 5394 | $bot_id = $this->get_current_bot_id($session_id); |
| 8488 | 5395 | |
| 8489 | 5396 | // Get system prompt instructions using centralized function |
| 8490 | - $system_prompt_instructions = $this->get_system_instructions($bot_id, $session_id); | |
| 5397 | + $system_prompt_instructions = $this->get_system_instructions($bot_id); | |
| 8491 | 5398 | |
| 8492 | 5399 | // Clean and validate conversation history |
| 8493 | 5400 | foreach ($conversation_history as &$message) { |
| 8494 | 5401 | // Convert bot and agent roles to assistant |
| @@ -8592,12 +5499,12 @@ | ||
| 8592 | 5499 | $conversation_history = array(); |
| 8593 | 5500 | } |
| 8594 | 5501 | |
| 8595 | 5502 | // Get bot ID from session or request |
| 8596 | - $bot_id = $this->get_current_bot_id(''); | |
| 5503 | + $bot_id = $this->get_current_bot_id($session_id); | |
| 8597 | 5504 | |
| 8598 | 5505 | // Get system prompt instructions using centralized function |
| 8599 | - $system_prompt_instructions = $this->get_system_instructions($bot_id, $session_id); | |
| 5506 | + $system_prompt_instructions = $this->get_system_instructions($bot_id); | |
| 8600 | 5507 | |
| 8601 | 5508 | // Create a new array for the formatted conversation |
| 8602 | 5509 | $formatted_conversation = array(); |
| 8603 | 5510 | |
| @@ -8626,42 +5533,15 @@ | ||
| 8626 | 5533 | ); |
| 8627 | 5534 | } |
| 8628 | 5535 | } |
| 8629 | 5536 | |
| 8630 | - // Check if this is a GPT-5 model (supports reasoning_effort parameter) | |
| 8631 | - $is_gpt5_model = ( | |
| 8632 | - strpos($selected_model, 'gpt-5') === 0 || | |
| 8633 | - $selected_model === 'gpt-5.2' || | |
| 8634 | - $selected_model === 'gpt-5.1-2025-11-13' || | |
| 8635 | - $selected_model === 'gpt-5' || | |
| 8636 | - $selected_model === 'gpt-5-mini' || | |
| 8637 | - $selected_model === 'gpt-5-nano' | |
| 8638 | - ); | |
| 8639 | - | |
| 8640 | - // Build request body with optimal settings for fast responses | |
| 8641 | - $request_body = [ | |
| 5537 | + $body = json_encode([ | |
| 8642 | 5538 | 'model' => $selected_model, |
| 8643 | 5539 | 'messages' => $formatted_conversation, |
| 8644 | 5540 | 'temperature' => 1, |
| 8645 | 5541 | 'stream' => false |
| 8646 | - ]; | |
| 5542 | + ]); | |
| 8647 | 5543 | |
| 8648 | - // Add reasoning_effort only for GPT-5 models that support it | |
| 8649 | - // These chat models don't support reasoning_effort parameter | |
| 8650 | - $no_reasoning_models = array('gpt-5.2', 'gpt-5.1-chat-latest', 'gpt-5.3-chat-latest', 'gpt-5.4-mini', 'gpt-5.4-nano'); | |
| 8651 | - if ($is_gpt5_model && !in_array($selected_model, $no_reasoning_models, true)) { | |
| 8652 | - // GPT-5.1 uses 'low' instead of 'minimal' | |
| 8653 | - if ($selected_model === 'gpt-5.1-2025-11-13') { | |
| 8654 | - $request_body['reasoning_effort'] = 'low'; | |
| 8655 | - } elseif ($selected_model === 'gpt-5.4') { | |
| 8656 | - $request_body['reasoning_effort'] = 'none'; | |
| 8657 | - } else { | |
| 8658 | - $request_body['reasoning_effort'] = 'minimal'; | |
| 8659 | - } | |
| 8660 | - } | |
| 8661 | - | |
| 8662 | - $body = json_encode($request_body); | |
| 8663 | - | |
| 8664 | 5544 | $args = [ |
| 8665 | 5545 | 'body' => $body, |
| 8666 | 5546 | 'headers' => [ |
| 8667 | 5547 | 'Content-Type' => 'application/json', |
| @@ -8677,8 +5557,9 @@ | ||
| 8677 | 5557 | $response = wp_remote_post('https://api.openai.com/v1/chat/completions', $args); |
| 8678 | 5558 | |
| 8679 | 5559 | if (is_wp_error($response)) { |
| 8680 | 5560 | $error_message = $response->get_error_message(); |
| 5561 | + //error_log('OpenAI API Error: ' . $error_message); | |
| 8681 | 5562 | return [ |
| 8682 | 5563 | 'error' => esc_html__('Connection error when contacting OpenAI: ', 'mxchat') . esc_html($error_message), |
| 8683 | 5564 | 'error_code' => 'openai_connection_error', |
| 8684 | 5565 | 'provider' => 'openai' |
| @@ -8697,8 +5578,10 @@ | ||
| 8697 | 5578 | $error_type = isset($decoded_response['error']['type']) |
| 8698 | 5579 | ? $decoded_response['error']['type'] |
| 8699 | 5580 | : 'unknown'; |
| 8700 | 5581 | |
| 5582 | + //error_log('OpenAI API HTTP Error: ' . $status_code . ' - ' . $error_message); | |
| 5583 | + | |
| 8701 | 5584 | // Handle specific error types |
| 8702 | 5585 | switch ($error_type) { |
| 8703 | 5586 | case 'invalid_request_error': |
| 8704 | 5587 | if (strpos($error_message, 'API key') !== false) { |
| @@ -8746,8 +5629,9 @@ | ||
| 8746 | 5629 | |
| 8747 | 5630 | if (isset($decoded_response['choices'][0]['message']['content'])) { |
| 8748 | 5631 | return trim($decoded_response['choices'][0]['message']['content']); |
| 8749 | 5632 | } else { |
| 5633 | + //error_log('OpenAI API Response Format Error: ' . print_r($decoded_response, true)); | |
| 8750 | 5634 | return [ |
| 8751 | 5635 | 'error' => esc_html__('Unexpected response format from OpenAI.', 'mxchat'), |
| 8752 | 5636 | 'error_code' => 'openai_response_format_error', |
| 8753 | 5637 | 'provider' => 'openai' |
| @@ -8753,8 +5637,9 @@ | ||
| 8753 | 5637 | 'provider' => 'openai' |
| 8754 | 5638 | ]; |
| 8755 | 5639 | } |
| 8756 | 5640 | } catch (Exception $e) { |
| 5641 | + //error_log('OpenAI Exception: ' . $e->getMessage()); | |
| 8757 | 5642 | return [ |
| 8758 | 5643 | 'error' => esc_html__('System error when processing OpenAI request: ', 'mxchat') . esc_html($e->getMessage()), |
| 8759 | 5644 | 'error_code' => 'openai_exception', |
| 8760 | 5645 | 'provider' => 'openai' |
| @@ -8760,9 +5645,8 @@ | ||
| 8760 | 5645 | 'provider' => 'openai' |
| 8761 | 5646 | ]; |
| 8762 | 5647 | } |
| 8763 | 5648 | } |
| 8764 | - | |
| 8765 | 5649 | private function mxchat_generate_response_xai($selected_model, $xai_api_key, $conversation_history, $relevant_content) { |
| 8766 | 5650 | try { |
| 8767 | 5651 | // Get bot ID from session or request |
| 8768 | 5652 | $bot_id = $this->get_current_bot_id($session_id); |
| @@ -8767,9 +5651,9 @@ | ||
| 8767 | 5651 | // Get bot ID from session or request |
| 8768 | 5652 | $bot_id = $this->get_current_bot_id($session_id); |
| 8769 | 5653 | |
| 8770 | 5654 | // Get system prompt instructions using centralized function |
| 8771 | - $system_prompt_instructions = $this->get_system_instructions($bot_id, $session_id); | |
| 5655 | + $system_prompt_instructions = $this->get_system_instructions($bot_id); | |
| 8772 | 5656 | |
| 8773 | 5657 | // Add system prompt to relevant content |
| 8774 | 5658 | $content_with_instructions = $system_prompt_instructions . " " . $relevant_content; |
| 8775 | 5659 | |
| @@ -8966,9 +5850,9 @@ | ||
| 8966 | 5850 | // Get bot ID from session or request |
| 8967 | 5851 | $bot_id = $this->get_current_bot_id($session_id); |
| 8968 | 5852 | |
| 8969 | 5853 | // Get system prompt instructions using centralized function |
| 8970 | - $system_prompt_instructions = $this->get_system_instructions($bot_id, $session_id); | |
| 5854 | + $system_prompt_instructions = $this->get_system_instructions($bot_id); | |
| 8971 | 5855 | |
| 8972 | 5856 | // Create a new array for the formatted conversation |
| 8973 | 5857 | $formatted_conversation = array(); |
| 8974 | 5858 | |
| @@ -9125,9 +6009,9 @@ | ||
| 9125 | 6009 | // Get bot ID from session or request |
| 9126 | 6010 | $bot_id = $this->get_current_bot_id($session_id); |
| 9127 | 6011 | |
| 9128 | 6012 | // Get system prompt instructions using centralized function |
| 9129 | - $system_prompt_instructions = $this->get_system_instructions($bot_id, $session_id); | |
| 6013 | + $system_prompt_instructions = $this->get_system_instructions($bot_id); | |
| 9130 | 6014 | |
| 9131 | 6015 | // Add system prompt to relevant content |
| 9132 | 6016 | $content_with_instructions = $system_prompt_instructions . " " . $relevant_content; |
| 9133 | 6017 | |
| @@ -9223,11 +6107,9 @@ | ||
| 9223 | 6107 | ] |
| 9224 | 6108 | ]); |
| 9225 | 6109 | |
| 9226 | 6110 | // Prepare the API endpoint |
| 9227 | - // Use v1beta for preview models (Gemini 3, experimental), v1 for stable models | |
| 9228 | - $api_version = (strpos($selected_model, 'preview') !== false || strpos($selected_model, 'exp') !== false) ? 'v1beta' : 'v1'; | |
| 9229 | - $api_endpoint = 'https://generativelanguage.googleapis.com/' . $api_version . '/models/' . $selected_model . ':generateContent?key=' . $gemini_api_key; | |
| 6111 | + $api_endpoint = 'https://generativelanguage.googleapis.com/v1/models/' . $selected_model . ':generateContent?key=' . $gemini_api_key; | |
| 9230 | 6112 | |
| 9231 | 6113 | // Set up the API request |
| 9232 | 6114 | $args = [ |
| 9233 | 6115 | 'body' => $body, |
| @@ -9269,9 +6151,9 @@ | ||
| 9269 | 6151 | |
| 9270 | 6152 | |
| 9271 | 6153 | public function test_streaming_request() { |
| 9272 | 6154 | $options = get_option('mxchat_options', []); |
| 9273 | - $model = $options['model'] ?? 'gpt-5.1-chat-latest'; | |
| 6155 | + $model = $options['model'] ?? 'gpt-4o'; | |
| 9274 | 6156 | |
| 9275 | 6157 | // Detect provider from model prefix |
| 9276 | 6158 | $provider = strtolower(explode('-', $model)[0]); |
| 9277 | 6159 | |
| @@ -9454,54 +6336,35 @@ | ||
| 9454 | 6336 | } |
| 9455 | 6337 | |
| 9456 | 6338 | |
| 9457 | 6339 | public function mxchat_enqueue_scripts_styles() { |
| 9458 | - // Fetch options from the database first to check loading strategy | |
| 9459 | - $this->options = get_option('mxchat_options'); | |
| 9460 | - $loading_strategy = isset($this->options['script_loading_strategy']) ? $this->options['script_loading_strategy'] : 'default'; | |
| 9461 | - | |
| 9462 | - // Always enqueue CSS immediately | |
| 6340 | + // Define version numbers for the styles and scripts | |
| 6341 | + $chat_style_version = '2.4.6'; | |
| 6342 | + $chat_script_version = '2.4.6'; | |
| 6343 | + // Enqueue the script | |
| 6344 | + wp_enqueue_script( | |
| 6345 | + 'mxchat-chat-js', | |
| 6346 | + plugin_dir_url(__FILE__) . '../js/chat-script.js', | |
| 6347 | + array('jquery'), | |
| 6348 | + $chat_script_version, | |
| 6349 | + true | |
| 6350 | + ); | |
| 6351 | + // Enqueue the CSS | |
| 9463 | 6352 | wp_enqueue_style( |
| 9464 | 6353 | 'mxchat-chat-css', |
| 9465 | 6354 | plugin_dir_url(__FILE__) . '../css/chat-style.css', |
| 9466 | 6355 | array(), |
| 9467 | - MXCHAT_VERSION | |
| 6356 | + $chat_style_version | |
| 9468 | 6357 | ); |
| 9469 | - | |
| 9470 | - // Handle script loading based on strategy | |
| 9471 | - if ($loading_strategy === 'default' || $loading_strategy === 'defer') { | |
| 9472 | - // Enqueue the script normally | |
| 9473 | - wp_enqueue_script( | |
| 9474 | - 'mxchat-chat-js', | |
| 9475 | - plugin_dir_url(__FILE__) . '../js/chat-script.js', | |
| 9476 | - array('jquery'), | |
| 9477 | - MXCHAT_VERSION, | |
| 9478 | - true | |
| 9479 | - ); | |
| 9480 | - | |
| 9481 | - // Add defer attribute if strategy is 'defer' | |
| 9482 | - if ($loading_strategy === 'defer') { | |
| 9483 | - wp_script_add_data('mxchat-chat-js', 'strategy', 'defer'); | |
| 9484 | - } | |
| 9485 | - } else { | |
| 9486 | - // For delay or interaction-based loading, we'll use a custom loader | |
| 9487 | - // Don't enqueue the main script - we'll load it dynamically | |
| 9488 | - add_action('wp_footer', array($this, 'mxchat_output_delayed_script_loader'), 99); | |
| 9489 | - } | |
| 9490 | - | |
| 6358 | + // Fetch options from the database | |
| 6359 | + $this->options = get_option('mxchat_options'); | |
| 9491 | 6360 | $prompts_options = get_option('mxchat_prompts_options', array()); |
| 9492 | - | |
| 9493 | - // Check if AI theme is active - if so, skip inline colors in JavaScript | |
| 9494 | - $theme_options = get_option('mxchat_theme_options', array()); | |
| 9495 | - $ai_theme_active = !empty($theme_options['active_ai_theme_css']); | |
| 9496 | - $has_bot_theme_assignments = !empty($theme_options['bot_theme_assignments']); | |
| 9497 | - $skip_inline_colors = $ai_theme_active || $has_bot_theme_assignments; | |
| 9498 | - | |
| 6361 | + | |
| 9499 | 6362 | // Prepare settings for JavaScript |
| 9500 | 6363 | $style_settings = array( |
| 9501 | 6364 | 'ajax_url' => admin_url('admin-ajax.php'), |
| 9502 | 6365 | 'nonce' => wp_create_nonce('mxchat_chat_nonce'), |
| 9503 | - 'model' => isset($this->options['model']) ? $this->options['model'] : 'gpt-5.1-chat-latest', | |
| 6366 | + 'model' => isset($this->options['model']) ? $this->options['model'] : 'gpt-4o', | |
| 9504 | 6367 | 'enable_streaming_toggle' => isset($this->options['enable_streaming_toggle']) ? $this->options['enable_streaming_toggle'] : 'on', |
| 9505 | 6368 | 'contextual_awareness_toggle' => isset($this->options['contextual_awareness_toggle']) ? $this->options['contextual_awareness_toggle'] : 'off', |
| 9506 | 6369 | 'link_target_toggle' => $this->options['link_target_toggle'] ?? 'off', |
| 9507 | 6370 | 'rate_limit_message' => $this->options['rate_limit_message'] ?? 'Rate limit exceeded. Please try again later.', |
| @@ -9528,134 +6391,15 @@ | ||
| 9528 | 6391 | 'use_pinecone' => $prompts_options['mxchat_use_pinecone'] ?? '0', |
| 9529 | 6392 | 'email_collection_enabled' => $this->options['enable_email_block'] ?? 'off', // FIXED |
| 9530 | 6393 | 'initial_email_state' => null, // Also fixed this undefined variable |
| 9531 | 6394 | 'skip_email_check' => true, |
| 9532 | - 'pinecone_enabled' => isset($prompts_options['mxchat_use_pinecone']) && $prompts_options['mxchat_use_pinecone'] === '1', | |
| 9533 | - 'skip_inline_colors' => $skip_inline_colors, | |
| 9534 | - 'bot_theme_assignments' => $theme_options['bot_theme_assignments'] ?? array() | |
| 6395 | + 'pinecone_enabled' => isset($prompts_options['mxchat_use_pinecone']) && $prompts_options['mxchat_use_pinecone'] === '1' | |
| 9535 | 6396 | ); |
| 9536 | - | |
| 9537 | - // For normal/defer loading, use wp_localize_script | |
| 9538 | - // For delayed loading, we store settings in a transient to be output inline | |
| 9539 | - if ($loading_strategy === 'default' || $loading_strategy === 'defer') { | |
| 9540 | - wp_localize_script('mxchat-chat-js', 'mxchatChat', $style_settings); | |
| 9541 | - } else { | |
| 9542 | - // Store settings for the delayed loader to use | |
| 9543 | - set_transient('mxchat_delayed_settings_' . get_current_user_id(), $style_settings, 60); | |
| 9544 | - } | |
| 6397 | + // Pass the settings to the script | |
| 6398 | + wp_localize_script('mxchat-chat-js', 'mxchatChat', $style_settings); | |
| 9545 | 6399 | } |
| 9546 | 6400 | |
| 9547 | -/** | |
| 9548 | - * Output the delayed script loader for performance optimization | |
| 9549 | - */ | |
| 9550 | -public function mxchat_output_delayed_script_loader() { | |
| 9551 | - $this->options = get_option('mxchat_options'); | |
| 9552 | - $loading_strategy = isset($this->options['script_loading_strategy']) ? $this->options['script_loading_strategy'] : 'default'; | |
| 9553 | - $script_url = plugin_dir_url(__FILE__) . '../js/chat-script.js?ver=' . MXCHAT_VERSION; | |
| 9554 | 6401 | |
| 9555 | - // Get the stored settings | |
| 9556 | - $prompts_options = get_option('mxchat_prompts_options', array()); | |
| 9557 | - $theme_options = get_option('mxchat_theme_options', array()); | |
| 9558 | - $ai_theme_active = !empty($theme_options['active_ai_theme_css']); | |
| 9559 | - $has_bot_theme_assignments = !empty($theme_options['bot_theme_assignments']); | |
| 9560 | - $skip_inline_colors = $ai_theme_active || $has_bot_theme_assignments; | |
| 9561 | - | |
| 9562 | - $style_settings = array( | |
| 9563 | - 'ajax_url' => admin_url('admin-ajax.php'), | |
| 9564 | - 'nonce' => wp_create_nonce('mxchat_chat_nonce'), | |
| 9565 | - 'model' => isset($this->options['model']) ? $this->options['model'] : 'gpt-5.1-chat-latest', | |
| 9566 | - 'enable_streaming_toggle' => isset($this->options['enable_streaming_toggle']) ? $this->options['enable_streaming_toggle'] : 'on', | |
| 9567 | - 'contextual_awareness_toggle' => isset($this->options['contextual_awareness_toggle']) ? $this->options['contextual_awareness_toggle'] : 'off', | |
| 9568 | - 'link_target_toggle' => $this->options['link_target_toggle'] ?? 'off', | |
| 9569 | - 'rate_limit_message' => $this->options['rate_limit_message'] ?? 'Rate limit exceeded. Please try again later.', | |
| 9570 | - 'complianz_toggle' => isset($this->options['complianz_toggle']) && $this->options['complianz_toggle'] === 'on', | |
| 9571 | - 'user_message_bg_color' => $this->options['user_message_bg_color'] ?? '#fff', | |
| 9572 | - 'user_message_font_color' => $this->options['user_message_font_color'] ?? '#212121', | |
| 9573 | - 'bot_message_bg_color' => $this->options['bot_message_bg_color'] ?? '#212121', | |
| 9574 | - 'bot_message_font_color' => $this->options['bot_message_font_color'] ?? '#fff', | |
| 9575 | - 'top_bar_bg_color' => $this->options['top_bar_bg_color'] ?? '#212121', | |
| 9576 | - 'send_button_font_color' => $this->options['send_button_font_color'] ?? '#212121', | |
| 9577 | - 'close_button_color' => $this->options['close_button_color'] ?? '#fff', | |
| 9578 | - 'chatbot_background_color' => $this->options['chatbot_background_color'] ?? '#212121', | |
| 9579 | - 'chatbot_bg_color' => $this->options['chatbot_bg_color'] ?? '#fff', | |
| 9580 | - 'icon_color' => $this->options['icon_color'] ?? '#fff', | |
| 9581 | - 'chat_input_font_color' => $this->options['chat_input_font_color'] ?? '#212121', | |
| 9582 | - 'chat_persistence_toggle' => $this->options['chat_persistence_toggle'] ?? 'off', | |
| 9583 | - 'appendWidgetToBody' => $this->options['append_to_body'] ?? 'off', | |
| 9584 | - 'live_agent_message_bg_color' => $this->options['live_agent_message_bg_color'] ?? '#ffffff', | |
| 9585 | - 'live_agent_message_font_color' => $this->options['live_agent_message_font_color'] ?? '#333333', | |
| 9586 | - 'chat_toolbar_toggle' => $this->options['chat_toolbar_toggle'] ?? 'off', | |
| 9587 | - 'mode_indicator_bg_color' => $this->options['mode_indicator_bg_color'] ?? '#767676', | |
| 9588 | - 'mode_indicator_font_color' => $this->options['mode_indicator_font_color'] ?? '#ffffff', | |
| 9589 | - 'toolbar_icon_color' => $this->options['toolbar_icon_color'] ?? '#212121', | |
| 9590 | - 'use_pinecone' => $prompts_options['mxchat_use_pinecone'] ?? '0', | |
| 9591 | - 'email_collection_enabled' => $this->options['enable_email_block'] ?? 'off', | |
| 9592 | - 'initial_email_state' => null, | |
| 9593 | - 'skip_email_check' => true, | |
| 9594 | - 'pinecone_enabled' => isset($prompts_options['mxchat_use_pinecone']) && $prompts_options['mxchat_use_pinecone'] === '1', | |
| 9595 | - 'skip_inline_colors' => $skip_inline_colors, | |
| 9596 | - 'bot_theme_assignments' => $theme_options['bot_theme_assignments'] ?? array() | |
| 9597 | - ); | |
| 9598 | - | |
| 9599 | - // Determine delay time based on strategy | |
| 9600 | - $delay_ms = 0; | |
| 9601 | - switch ($loading_strategy) { | |
| 9602 | - case 'delay_1s': | |
| 9603 | - $delay_ms = 1000; | |
| 9604 | - break; | |
| 9605 | - case 'delay_3s': | |
| 9606 | - $delay_ms = 3000; | |
| 9607 | - break; | |
| 9608 | - case 'delay_5s': | |
| 9609 | - $delay_ms = 5000; | |
| 9610 | - break; | |
| 9611 | - } | |
| 9612 | - | |
| 9613 | - ?> | |
| 9614 | - <script type="text/javascript"> | |
| 9615 | - (function() { | |
| 9616 | - var mxchatLoaded = false; | |
| 9617 | - var mxchatChat = <?php echo wp_json_encode($style_settings); ?>; | |
| 9618 | - window.mxchatChat = mxchatChat; | |
| 9619 | - | |
| 9620 | - function loadMxChatScript() { | |
| 9621 | - if (mxchatLoaded) return; | |
| 9622 | - mxchatLoaded = true; | |
| 9623 | - | |
| 9624 | - function appendChatScript() { | |
| 9625 | - var script = document.createElement('script'); | |
| 9626 | - script.src = <?php echo wp_json_encode($script_url); ?>; | |
| 9627 | - script.type = 'text/javascript'; | |
| 9628 | - document.body.appendChild(script); | |
| 9629 | - } | |
| 9630 | - | |
| 9631 | - if (typeof jQuery !== 'undefined') { | |
| 9632 | - appendChatScript(); | |
| 9633 | - } else { | |
| 9634 | - var jq = document.createElement('script'); | |
| 9635 | - jq.src = <?php echo wp_json_encode(includes_url('js/jquery/jquery.min.js')); ?>; | |
| 9636 | - jq.onload = appendChatScript; | |
| 9637 | - document.body.appendChild(jq); | |
| 9638 | - } | |
| 9639 | - } | |
| 9640 | - | |
| 9641 | - <?php if ($loading_strategy === 'on_interaction'): ?> | |
| 9642 | - // Load on user interaction | |
| 9643 | - var events = ['scroll', 'mousemove', 'touchstart', 'keydown', 'click']; | |
| 9644 | - events.forEach(function(evt) { | |
| 9645 | - window.addEventListener(evt, loadMxChatScript, {once: true, passive: true}); | |
| 9646 | - }); | |
| 9647 | - // Fallback: load after 8 seconds if no interaction | |
| 9648 | - setTimeout(loadMxChatScript, 8000); | |
| 9649 | - <?php else: ?> | |
| 9650 | - // Load after specified delay | |
| 9651 | - setTimeout(loadMxChatScript, <?php echo intval($delay_ms); ?>); | |
| 9652 | - <?php endif; ?> | |
| 9653 | - })(); | |
| 9654 | - </script> | |
| 9655 | - <?php | |
| 9656 | -} | |
| 9657 | - | |
| 9658 | 6402 | /** |
| 9659 | 6403 | * Setup the cron jobs for rate limits with guard against multiple calls |
| 9660 | 6404 | */ |
| 9661 | 6405 | public function setup_rate_limit_cron_jobs() { |
| @@ -10190,11 +6934,8 @@ | ||
| 10190 | 6934 | |
| 10191 | 6935 | /** |
| 10192 | 6936 | * AJAX handler to get system information for testing panel |
| 10193 | 6937 | */ |
| 10194 | -/** | |
| 10195 | - * AJAX handler to get system information for testing panel | |
| 10196 | - */ | |
| 10197 | 6938 | public function mxchat_get_system_info() { |
| 10198 | 6939 | // Verify nonce for security |
| 10199 | 6940 | if (!wp_verify_nonce($_POST['nonce'], 'mxchat_test_nonce')) { |
| 10200 | 6941 | wp_send_json_error(['message' => 'Invalid nonce']); |
| @@ -10211,25 +6952,11 @@ | ||
| 10211 | 6952 | $system_prompt = isset($this->options['system_prompt_instructions']) |
| 10212 | 6953 | ? $this->options['system_prompt_instructions'] |
| 10213 | 6954 | : 'No system prompt configured'; |
| 10214 | 6955 | |
| 10215 | - // Get selected model | |
| 10216 | - $selected_model = isset($this->options['model']) ? $this->options['model'] : 'gpt-5.1-chat-latest'; | |
| 6956 | + // Get selected model - FIXED: Use $this->options instead of $current_options | |
| 6957 | + $selected_model = isset($this->options['model']) ? $this->options['model'] : 'gpt-4o'; | |
| 10217 | 6958 | |
| 10218 | - // Check if OpenRouter is being used | |
| 10219 | - $is_openrouter = ($selected_model === 'openrouter'); | |
| 10220 | - $openrouter_model = ''; | |
| 10221 | - | |
| 10222 | - if ($is_openrouter) { | |
| 10223 | - // Get the actual OpenRouter model that's selected | |
| 10224 | - $openrouter_model = isset($this->options['openrouter_selected_model']) | |
| 10225 | - ? $this->options['openrouter_selected_model'] | |
| 10226 | - : 'No OpenRouter model selected'; | |
| 10227 | - | |
| 10228 | - // Update selected_model display to show both | |
| 10229 | - $selected_model = 'OpenRouter: ' . $openrouter_model; | |
| 10230 | - } | |
| 10231 | - | |
| 10232 | 6959 | // Get API key status (just check if they exist, don't expose the keys) |
| 10233 | 6960 | $api_status = []; |
| 10234 | 6961 | $api_status['openai'] = !empty($this->options['api_key']); |
| 10235 | 6962 | $api_status['claude'] = !empty($this->options['claude_api_key']); |
| @@ -10235,15 +6962,12 @@ | ||
| 10235 | 6962 | $api_status['claude'] = !empty($this->options['claude_api_key']); |
| 10236 | 6963 | $api_status['gemini'] = !empty($this->options['gemini_api_key']); |
| 10237 | 6964 | $api_status['xai'] = !empty($this->options['xai_api_key']); |
| 10238 | 6965 | $api_status['deepseek'] = !empty($this->options['deepseek_api_key']); |
| 10239 | - $api_status['openrouter'] = !empty($this->options['openrouter_api_key']); | |
| 10240 | 6966 | |
| 10241 | 6967 | wp_send_json_success([ |
| 10242 | 6968 | 'system_prompt' => $system_prompt, |
| 10243 | 6969 | 'selected_model' => $selected_model, |
| 10244 | - 'is_openrouter' => $is_openrouter, | |
| 10245 | - 'openrouter_model' => $openrouter_model, | |
| 10246 | 6970 | 'api_status' => $api_status |
| 10247 | 6971 | ]); |
| 10248 | 6972 | } |
| 10249 | 6973 | |
| @@ -10282,42 +7006,24 @@ | ||
| 10282 | 7006 | if (!wp_verify_nonce($_POST['nonce'], 'mxchat_test_nonce')) { |
| 10283 | 7007 | wp_send_json_error(['message' => 'Invalid nonce']); |
| 10284 | 7008 | return; |
| 10285 | 7009 | } |
| 10286 | - | |
| 7010 | + | |
| 10287 | 7011 | // Only allow admin users |
| 10288 | 7012 | if (!current_user_can('administrator')) { |
| 10289 | 7013 | wp_send_json_error(['message' => 'Unauthorized']); |
| 10290 | 7014 | return; |
| 10291 | 7015 | } |
| 10292 | - | |
| 10293 | - // Check OpenAI Vector Store first (takes priority) | |
| 10294 | - $vectorstore_options = get_option('mxchat_openai_vectorstore_options', array()); | |
| 10295 | - $use_vectorstore = (isset($vectorstore_options['mxchat_use_openai_vectorstore']) && $vectorstore_options['mxchat_use_openai_vectorstore'] === '1'); | |
| 10296 | - | |
| 10297 | - if ($use_vectorstore) { | |
| 10298 | - $vectorstore_ids = $vectorstore_options['mxchat_openai_vectorstore_ids'] ?? ''; | |
| 10299 | - $id_count = !empty($vectorstore_ids) ? count(array_filter(array_map('trim', explode(',', $vectorstore_ids)))) : 0; | |
| 10300 | - | |
| 10301 | - $kb_info = [ | |
| 10302 | - 'type' => 'OpenAI Vector Store', | |
| 10303 | - 'status' => 'Active', | |
| 10304 | - 'documents' => $id_count > 0 ? $id_count . ' vector store' . ($id_count > 1 ? 's' : '') . ' configured' : 'No vector stores configured' | |
| 10305 | - ]; | |
| 10306 | - | |
| 10307 | - wp_send_json_success($kb_info); | |
| 10308 | - return; | |
| 10309 | - } | |
| 10310 | - | |
| 7016 | + | |
| 10311 | 7017 | // Check Pinecone vs WordPress |
| 10312 | 7018 | $addon_options = get_option('mxchat_pinecone_addon_options', array()); |
| 10313 | 7019 | $use_pinecone = (isset($addon_options['mxchat_use_pinecone']) && $addon_options['mxchat_use_pinecone'] === '1'); |
| 10314 | - | |
| 7020 | + | |
| 10315 | 7021 | $kb_info = [ |
| 10316 | 7022 | 'type' => $use_pinecone ? 'Pinecone' : 'WordPress Database', |
| 10317 | 7023 | 'status' => 'Active' |
| 10318 | 7024 | ]; |
| 10319 | - | |
| 7025 | + | |
| 10320 | 7026 | // Get document count |
| 10321 | 7027 | if ($use_pinecone) { |
| 10322 | 7028 | $kb_info['documents'] = 'Connected to Pinecone'; |
| 10323 | 7029 | $kb_info['api_configured'] = !empty($addon_options['mxchat_pinecone_api_key']); |
| @@ -10327,9 +7033,9 @@ | ||
| 10327 | 7033 | $table_name = $wpdb->prefix . 'mxchat_system_prompt_content'; |
| 10328 | 7034 | $count = $wpdb->get_var("SELECT COUNT(*) FROM {$table_name}"); |
| 10329 | 7035 | $kb_info['documents'] = $count ? $count . ' documents' : 'No documents'; |
| 10330 | 7036 | } |
| 10331 | - | |
| 7037 | + | |
| 10332 | 7038 | wp_send_json_success($kb_info); |
| 10333 | 7039 | } |
| 10334 | 7040 | |
| 10335 | 7041 | /** |
| @@ -10411,13 +7117,9 @@ | ||
| 10411 | 7117 | // Clear any other session-specific transients |
| 10412 | 7118 | delete_transient("mxchat_waiting_for_pdf_url_{$session_id}"); |
| 10413 | 7119 | delete_transient("mxchat_include_pdf_in_context_{$session_id}"); |
| 10414 | 7120 | delete_transient("mxchat_include_word_in_context_{$session_id}"); |
| 10415 | - | |
| 10416 | - // Clear form addon state (pending forms and submitted forms) | |
| 10417 | - delete_option("mxchat_pending_form_{$session_id}"); | |
| 10418 | - delete_option("mxchat_submitted_forms_{$session_id}"); | |
| 10419 | - | |
| 7121 | + | |
| 10420 | 7122 | //error_log("MxChat: Cleared all data for session: {$session_id}"); |
| 10421 | 7123 | } |
| 10422 | 7124 | |
| 10423 | 7125 | /** |
| @@ -10592,169 +7294,8 @@ | ||
| 10592 | 7294 | wp_send_json_success(['message' => 'Originating page tracked']); |
| 10593 | 7295 | wp_die(); |
| 10594 | 7296 | } |
| 10595 | 7297 | |
| 10596 | -/** | |
| 10597 | - * Validate and clean URLs from AI response | |
| 10598 | - * Removes any URLs that aren't in the knowledge base | |
| 10599 | - * | |
| 10600 | - * @param string $response_text The AI-generated response | |
| 10601 | - * @param array $valid_urls Array of URLs from the knowledge base | |
| 10602 | - * @return string Cleaned response with invalid URLs removed/flagged | |
| 10603 | - */ | |
| 10604 | -private function validate_and_clean_urls($response_text, $valid_urls) { | |
| 10605 | - // DEBUG: Log what we're working with | |
| 10606 | - //error_log("=== MxChat URL Validation Debug ==="); | |
| 10607 | - //error_log("Valid URLs count: " . count($valid_urls)); | |
| 10608 | - //error_log("Valid URLs: " . print_r($valid_urls, true)); | |
| 10609 | - //error_log("Response text length: " . strlen($response_text)); | |
| 10610 | - //error_log("Response text preview: " . substr($response_text, 0, 500)); | |
| 10611 | - | |
| 10612 | - // If no valid URLs provided or empty response, return as-is | |
| 10613 | - if (empty($valid_urls) || empty($response_text)) { | |
| 10614 | - //error_log("Validation skipped - empty valid_urls or response"); | |
| 10615 | - return $response_text; | |
| 10616 | - } | |
| 10617 | - | |
| 10618 | - // Extract all URLs from the AI response | |
| 10619 | - // This regex matches http:// and https:// URLs | |
| 10620 | - preg_match_all( | |
| 10621 | - '#\bhttps?://[^\s<>"\')\]]+#i', | |
| 10622 | - $response_text, | |
| 10623 | - $matches | |
| 10624 | - ); | |
| 10625 | - | |
| 10626 | - // If no URLs found in response, return as-is | |
| 10627 | - if (empty($matches[0])) { | |
| 10628 | - //error_log("No URLs found in response"); | |
| 10629 | - return $response_text; | |
| 10630 | - } | |
| 10631 | - | |
| 10632 | - $found_urls = $matches[0]; | |
| 10633 | - $cleaned_response = $response_text; | |
| 10634 | - $removed_count = 0; | |
| 10635 | - | |
| 10636 | - // Normalize valid URLs for comparison (remove trailing slashes, fragments, etc.) | |
| 10637 | - $normalized_valid_urls = array_map(function($url) { | |
| 10638 | - // Remove trailing slash | |
| 10639 | - $url = rtrim($url, '/'); | |
| 10640 | - // Remove URL fragments (#section) | |
| 10641 | - $url = preg_replace('/#.*$/', '', $url); | |
| 10642 | - // Remove trailing punctuation that might have been captured | |
| 10643 | - $url = rtrim($url, '.,;:!?'); | |
| 10644 | - return $url; | |
| 10645 | - }, $valid_urls); | |
| 10646 | - | |
| 10647 | - //error_log("Normalized valid URLs: " . print_r($normalized_valid_urls, true)); | |
| 10648 | - | |
| 10649 | - foreach ($found_urls as $found_url) { | |
| 10650 | - // Clean up the found URL (remove trailing punctuation that might have been captured) | |
| 10651 | - $clean_found_url = rtrim($found_url, '.,;:!?)'); | |
| 10652 | - | |
| 10653 | - // DEBUG: Log each URL being checked | |
| 10654 | - //error_log("Checking found URL: " . $found_url); | |
| 10655 | - | |
| 10656 | - // Normalize for comparison | |
| 10657 | - $normalized_found = rtrim($clean_found_url, '/'); | |
| 10658 | - $normalized_found = preg_replace('/#.*$/', '', $normalized_found); | |
| 10659 | - | |
| 10660 | - //error_log("Normalized found URL: " . $normalized_found); | |
| 10661 | - | |
| 10662 | - // Check if this URL exists in our valid URLs list | |
| 10663 | - $is_valid = false; | |
| 10664 | - | |
| 10665 | - //error_log("Starting validation checks for: " . $normalized_found); | |
| 10666 | - | |
| 10667 | - // First, try exact match | |
| 10668 | - if (in_array($normalized_found, $normalized_valid_urls)) { | |
| 10669 | - $is_valid = true; | |
| 10670 | - //error_log("EXACT MATCH FOUND"); | |
| 10671 | - } else { | |
| 10672 | - //error_log("No exact match, checking variations..."); | |
| 10673 | - // If no exact match, check if it's a variation (with query params, etc.) | |
| 10674 | - foreach ($normalized_valid_urls as $valid_url) { | |
| 10675 | - //error_log(" Comparing against valid URL: " . $valid_url); | |
| 10676 | - | |
| 10677 | - // Check if the found URL starts with a valid URL (handles query params) | |
| 10678 | - if (strpos($normalized_found, $valid_url) === 0) { | |
| 10679 | - // Check what comes after the valid URL | |
| 10680 | - $remainder = substr($normalized_found, strlen($valid_url)); | |
| 10681 | - | |
| 10682 | - // Only valid if: | |
| 10683 | - // 1. Exact match (remainder is empty) | |
| 10684 | - // 2. Query params (starts with ?) | |
| 10685 | - // 3. Fragment (starts with #) | |
| 10686 | - if (empty($remainder) || $remainder[0] === '?' || $remainder[0] === '#') { | |
| 10687 | - $is_valid = true; | |
| 10688 | - //error_log(" MATCH: Found URL is valid variation of base URL"); | |
| 10689 | - break; | |
| 10690 | - } else { | |
| 10691 | - //error_log(" NOT A MATCH: Found URL extends path beyond valid URL (remainder: " . $remainder . ")"); | |
| 10692 | - } | |
| 10693 | - } | |
| 10694 | - // Also check the reverse (in case valid URL has query params) | |
| 10695 | - if (strpos($valid_url, $normalized_found) === 0) { | |
| 10696 | - $is_valid = true; | |
| 10697 | - //error_log(" MATCH: Valid URL starts with found URL"); | |
| 10698 | - break; | |
| 10699 | - } | |
| 10700 | - } | |
| 10701 | - | |
| 10702 | - if (!$is_valid) { | |
| 10703 | - //error_log("NO MATCH FOUND - URL should be removed"); | |
| 10704 | - } | |
| 10705 | - } | |
| 10706 | - | |
| 10707 | - // If URL is not valid, remove it from the response | |
| 10708 | - if (!$is_valid) { | |
| 10709 | - // Log the removal for debugging | |
| 10710 | - //error_log("MxChat: Removed hallucinated URL: " . $found_url); | |
| 10711 | - //error_log("MxChat: Valid URLs were: " . implode(', ', array_slice($normalized_valid_urls, 0, 5))); | |
| 10712 | - | |
| 10713 | - $removed_count++; | |
| 10714 | - | |
| 10715 | - // Check if URL is part of a markdown link: [text](url) | |
| 10716 | - $markdown_pattern = '/\[([^\]]+)\]\(' . preg_quote($found_url, '/') . '\)/'; | |
| 10717 | - if (preg_match($markdown_pattern, $cleaned_response)) { | |
| 10718 | - //error_log("Found markdown link, removing but keeping text"); | |
| 10719 | - // Remove the markdown link but keep the text | |
| 10720 | - $cleaned_response = preg_replace($markdown_pattern, '$1', $cleaned_response); | |
| 10721 | - } | |
| 10722 | - // Check if URL is part of an HTML link: <a href="url">text</a> | |
| 10723 | - else if (preg_match('/<a[^>]*href=["\']' . preg_quote($found_url, '/') . '["\'][^>]*>(.*?)<\/a>/i', $cleaned_response, $link_match)) { | |
| 10724 | - //error_log("Found HTML link, removing but keeping text"); | |
| 10725 | - // Remove the HTML link but keep the text | |
| 10726 | - $link_text = $link_match[1]; | |
| 10727 | - $cleaned_response = preg_replace( | |
| 10728 | - '/<a[^>]*href=["\']' . preg_quote($found_url, '/') . '["\'][^>]*>.*?<\/a>/i', | |
| 10729 | - $link_text, | |
| 10730 | - $cleaned_response | |
| 10731 | - ); | |
| 10732 | - } | |
| 10733 | - // Otherwise just remove the bare URL | |
| 10734 | - else { | |
| 10735 | - //error_log("Removing bare URL"); | |
| 10736 | - $cleaned_response = str_replace($found_url, '', $cleaned_response); | |
| 10737 | - } | |
| 10738 | - } | |
| 10739 | - } | |
| 10740 | - | |
| 10741 | - // Log summary if any URLs were removed | |
| 10742 | - if ($removed_count > 0) { | |
| 10743 | - //error_log("MxChat: URL Validation Summary - Removed {$removed_count} hallucinated URL(s)"); | |
| 10744 | - } else { | |
| 10745 | - //error_log("MxChat: URL Validation Summary - No URLs removed, all were valid"); | |
| 10746 | - } | |
| 10747 | - | |
| 10748 | - // Clean up any double spaces or awkward punctuation left behind | |
| 10749 | - // IMPORTANT: Only collapse horizontal whitespace (spaces/tabs), preserve newlines for markdown formatting | |
| 10750 | - $cleaned_response = preg_replace('/[^\S\n]+/', ' ', $cleaned_response); // Collapse spaces/tabs but NOT newlines | |
| 10751 | - $cleaned_response = preg_replace('/[^\S\n]+([.,;:!?])/', '$1', $cleaned_response); // Same for punctuation cleanup | |
| 10752 | - | |
| 10753 | - //error_log("Final cleaned response: " . $cleaned_response); | |
| 10754 | - | |
| 10755 | - return trim($cleaned_response); | |
| 10756 | -} | |
| 10757 | 7298 | |
| 10758 | 7299 | /** |
| 10759 | 7300 | * AJAX handler to get current chat mode for a session |
| 10760 | 7301 | */ |