PluginProbe
MxChat – AI Chatbot & Content Generation for WordPress / 2.4.0
MxChat – AI Chatbot & Content Generation for WordPress v2.4.0
3.2.21 3.2.20 3.2.19 3.2.18 3.2.17 3.2.16 3.2.15 3.2.14 3.2.12 3.2.13 3.2.11 3.2.10 3.2.9 3.2.8 3.2.7 3.2.6 3.2.5 3.2.4 3.2.3 3.2.2 3.2.1 2.0.3 2.0.4 2.0.5 2.0.6 All 152 releases
← All changes | includes/class-mxchat-integrator.php +1512 -6192 3.2.82.4.0 View file →
@@ -10,244 +10,11 @@
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 $streaming_headers_sent = false; // Track if streaming headers have been sent
18 14
19 -/**
20 - * Setup streaming headers - call this right before actually streaming
21 - * This delays header setup to allow actions/forms to return JSON responses
22 - */
23 -/**
24 - * Auto-retry wrapper around wp_remote_post for chat-send provider calls.
25 - *
26 - * Retries up to twice (750ms then 2000ms backoff) when the upstream provider
27 - * returns a TRANSIENT error: WP timeout, 429, 502, 503, 504, or a provider-
28 - * specific "overloaded" / "rate limit" body string. Returns immediately on
29 - * permanent errors (401/403/404/422) so misconfiguration surfaces fast.
30 - *
31 - * Drop-in replacement for wp_remote_post — returns the same shape
32 - * (WP_Error or response array) so the caller's existing error-handling
33 - * code path is unchanged.
34 - *
35 - * STREAMING PATH NOTE: this helper is ONLY for non-streaming chat-send
36 - * paths (the *_response_openai / *_response_claude / etc functions).
37 - * For the *_stream variants, the cURL initial-connect happens inside a
38 - * read-chunks loop — retrying there safely (without re-emitting partial
39 - * stream chunks to the client) is a separate problem. Streaming paths
40 - * are NOT wrapped in this build; tracked as a follow-on.
41 - *
42 - * Honors the `mxchat_options['auto_retry_on_transient_error']` toggle
43 - * (default true). When false, behavior is identical to plain wp_remote_post.
44 - */
45 -private function mxchat_provider_call_with_retry($url, $args, $provider_hint = '') {
46 - $opts = is_array($this->options ?? null) ? $this->options : array();
47 - $enabled = !isset($opts['auto_retry_on_transient_error']) ||
48 - (string) $opts['auto_retry_on_transient_error'] !== '0';
49 15
50 - if (!$enabled) {
51 - return wp_remote_post($url, $args);
52 - }
53 -
54 - $backoffs = array(0, 750, 2000); // ms — first attempt 0, then retry waits
55 - $last_response = null;
56 -
57 - foreach ($backoffs as $i => $delay_ms) {
58 - if ($delay_ms > 0) {
59 - usleep($delay_ms * 1000);
60 - }
61 - $response = wp_remote_post($url, $args);
62 - $last_response = $response;
63 -
64 - if (!$this->mxchat_is_transient_provider_error($response, $provider_hint)) {
65 - return $response;
66 - }
67 -
68 - if (defined('WP_DEBUG') && WP_DEBUG) {
69 - $code_for_log = is_wp_error($response) ? 'wp_error:' . $response->get_error_code()
70 - : (int) wp_remote_retrieve_response_code($response);
71 - error_log(sprintf(
72 - '[MxChat] Transient provider error (provider=%s, attempt=%d/3, status=%s). %s',
73 - $provider_hint ?: 'unknown',
74 - $i + 1,
75 - $code_for_log,
76 - ($i + 1) < count($backoffs) ? 'Retrying.' : 'Giving up.'
77 - ));
78 - }
79 - }
80 -
81 - return $last_response;
82 -}
83 -
84 16 /**
85 - * Returns true if a wp_remote_post response represents a TRANSIENT
86 - * provider error worth retrying. Conservative — only retries on signals
87 - * that are very likely to clear within a few seconds.
88 - *
89 - * Transient signals:
90 - * - WP_Error with timeout / connection / dns / ssl
91 - * - HTTP 429, 502, 503, 504
92 - * - Provider-specific overload bodies (gemini "overloaded", openai
93 - * "server_error", anthropic "overloaded_error", xai/grok "Rate limit")
94 - *
95 - * NOT transient (return false — fail-fast):
96 - * - 200/2xx (success)
97 - * - 401, 403, 404, 422 (auth / config errors — retrying wastes the
98 - * budget; the user needs to fix something)
99 - * - Any other 4xx (assume permanent unless explicitly listed above)
100 - * - 5xx other than the four listed above (e.g. 500 generic server error
101 - * is often a malformed request on our side, not a transient outage)
102 - */
103 -private function mxchat_is_transient_provider_error($response, $provider_hint = '') {
104 - if (is_wp_error($response)) {
105 - $code = $response->get_error_code();
106 - return in_array($code, array('http_request_failed', 'connection_failed', 'connection_timeout'), true)
107 - || stripos((string) $response->get_error_message(), 'timed out') !== false
108 - || stripos((string) $response->get_error_message(), 'timeout') !== false;
109 - }
110 -
111 - $status = (int) wp_remote_retrieve_response_code($response);
112 - if (in_array($status, array(429, 502, 503, 504), true)) {
113 - return true;
114 - }
115 - if ($status >= 200 && $status < 300) {
116 - return false;
117 - }
118 - // Permanent 4xx that should fail fast — even with no body.
119 - if (in_array($status, array(401, 403, 404, 405, 422), true)) {
120 - return false;
121 - }
122 -
123 - // Provider-specific body inspection for the cases where the upstream
124 - // returns 200 with an error envelope (gemini does this for overload).
125 - $body = (string) wp_remote_retrieve_body($response);
126 - if ($body === '') {
127 - return false;
128 - }
129 - $lower = strtolower($body);
130 - $hint = strtolower((string) $provider_hint);
131 -
132 - if ($hint === 'gemini' && (strpos($lower, 'overloaded') !== false
133 - || strpos($lower, 'high demand') !== false
134 - || strpos($lower, 'model is overloaded') !== false)) {
135 - return true;
136 - }
137 - if ($hint === 'openai' && (strpos($lower, 'rate limit reached') !== false
138 - || strpos($lower, '"type":"server_error"') !== false
139 - || strpos($lower, '"code":"server_error"') !== false)) {
140 - return true;
141 - }
142 - if ($hint === 'anthropic' && (strpos($lower, '"type":"overloaded_error"') !== false
143 - || strpos($lower, 'overloaded_error') !== false)) {
144 - return true;
145 - }
146 - if (($hint === 'xai' || $hint === 'grok') && strpos($lower, 'rate limit') !== false) {
147 - return true;
148 - }
149 -
150 - return false;
151 -}
152 -
153 -/**
154 - * Streaming-path classifier: same rules as mxchat_is_transient_provider_error
155 - * but takes a raw (http_code, body, provider_hint, curl_errno) tuple as
156 - * captured during a cURL streaming exec. cURL's WRITEFUNCTION/HEADERFUNCTION
157 - * collect status separately from a plain wp_remote_post array shape, so the
158 - * non-streaming helper above can't be called directly. This delegate keeps
159 - * the classification rules identical across both paths.
160 - */
161 -private function mxchat_is_transient_provider_error_raw($http_code, $body, $provider_hint = '', $curl_errno = 0) {
162 - if ($curl_errno) {
163 - // cURL transport-level error (timeout, connection failure, DNS, etc.)
164 - // Match the same WP_Error timeout/connection signals the array variant treats as transient.
165 - return in_array($curl_errno, array(
166 - CURLE_OPERATION_TIMEDOUT,
167 - CURLE_COULDNT_CONNECT,
168 - CURLE_COULDNT_RESOLVE_HOST,
169 - CURLE_SSL_CONNECT_ERROR,
170 - CURLE_GOT_NOTHING,
171 - CURLE_SEND_ERROR,
172 - CURLE_RECV_ERROR,
173 - ), true);
174 - }
175 -
176 - $status = (int) $http_code;
177 - if (in_array($status, array(429, 502, 503, 504), true)) {
178 - return true;
179 - }
180 - if ($status >= 200 && $status < 300) {
181 - return false;
182 - }
183 - if (in_array($status, array(401, 403, 404, 405, 422), true)) {
184 - return false;
185 - }
186 -
187 - $body = (string) $body;
188 - if ($body === '') {
189 - return false;
190 - }
191 - $lower = strtolower($body);
192 - $hint = strtolower((string) $provider_hint);
193 -
194 - if ($hint === 'gemini' && (strpos($lower, 'overloaded') !== false
195 - || strpos($lower, 'high demand') !== false
196 - || strpos($lower, 'model is overloaded') !== false)) {
197 - return true;
198 - }
199 - if ($hint === 'openai' && (strpos($lower, 'rate limit reached') !== false
200 - || strpos($lower, '"type":"server_error"') !== false
201 - || strpos($lower, '"code":"server_error"') !== false)) {
202 - return true;
203 - }
204 - if ($hint === 'anthropic' && (strpos($lower, '"type":"overloaded_error"') !== false
205 - || strpos($lower, 'overloaded_error') !== false)) {
206 - return true;
207 - }
208 - if (($hint === 'xai' || $hint === 'grok') && strpos($lower, 'rate limit') !== false) {
209 - return true;
210 - }
211 -
212 - return false;
213 -}
214 -
215 -/**
216 - * Whether transient-error auto-retry is enabled in admin settings.
217 - * Default true unless explicitly set to '0'. Used by both wp_remote_post
218 - * (mxchat_provider_call_with_retry) and cURL streaming paths.
219 - */
220 -private function mxchat_retry_enabled() {
221 - $opts = is_array($this->options ?? null) ? $this->options : array();
222 - return !isset($opts['auto_retry_on_transient_error']) ||
223 - (string) $opts['auto_retry_on_transient_error'] !== '0';
224 -}
225 -
226 -private function setup_streaming_headers() {
227 - if ($this->streaming_headers_sent || headers_sent()) {
228 - return false;
229 - }
230 -
231 - // Disable output buffering
232 - while (ob_get_level()) {
233 - ob_end_flush();
234 - }
235 -
236 - // Set headers for SSE
237 - header('Content-Type: text/event-stream');
238 - header('Cache-Control: no-cache');
239 - header('Connection: keep-alive');
240 - header('X-Accel-Buffering: no');
241 -
242 - ob_implicit_flush(true);
243 - flush();
244 -
245 - $this->streaming_headers_sent = true;
246 - return true;
247 -}
248 -
249 -/**
250 17 * Class constructor
251 18 */
252 19 public function __construct() {
253 20 $this->options = get_option('mxchat_options');
@@ -314,28 +81,13 @@
314 81 // Add chat mode checking actions
315 82 add_action('wp_ajax_mxchat_get_current_chat_mode', array($this, 'mxchat_get_current_chat_mode'));
316 83 add_action('wp_ajax_nopriv_mxchat_get_current_chat_mode', array($this, 'mxchat_get_current_chat_mode'));
317 84
318 - // Nonce refresh for page-cache compatibility (WP Rocket, LiteSpeed, etc.)
319 - add_action('wp_ajax_mxchat_refresh_nonce', array($this, 'mxchat_refresh_nonce'));
320 - add_action('wp_ajax_nopriv_mxchat_refresh_nonce', array($this, 'mxchat_refresh_nonce'));
321 -
322 - // Auto-email transcript action
323 - add_action('mxchat_send_delayed_transcript', array($this, 'mxchat_send_delayed_transcript'), 10, 1);
324 -
325 85 add_filter('mxchat_check_actions_only', array($this, 'check_actions_for_addons'), 10, 4);
326 86
327 87
328 88 }
329 89
330 -/**
331 - * Return a fresh nonce so cached pages can replace the stale one.
332 - */
333 -public function mxchat_refresh_nonce() {
334 - nocache_headers();
335 - wp_send_json_success(array('nonce' => wp_create_nonce('mxchat_chat_nonce')));
336 -}
337 -
338 90 // In your core plugin's check_actions_for_addons method:
339 91 public function check_actions_for_addons($default, $message, $user_id, $session_id) {
340 92 //error_log('MxChat Core: check_actions_for_addons called with message: ' . $message);
341 93
@@ -358,22 +110,8 @@
358 110 wp_die();
359 111 }
360 112
361 113 $session_id = sanitize_text_field($_POST['session_id']);
362 -
363 - // SECURITY FIX: Verify session ownership before retrieving data
364 - // If IP/user changed, signal frontend to reset session instead of blocking
365 - $current_user_identifier = MxChat_User::mxchat_get_user_identifier();
366 -
367 - // Check if this session has an owner recorded
368 - $session_owner = get_option("mxchat_session_owner_{$session_id}");
369 -
370 - // Update session owner if it changed (e.g. IP changed due to network switch)
371 - // The session ID itself is the authentication — if the client has it, they own it
372 - if (!$session_owner || $session_owner !== $current_user_identifier) {
373 - update_option("mxchat_session_owner_{$session_id}", $current_user_identifier, 'no');
374 - }
375 -
376 114 $history = get_option("mxchat_history_{$session_id}", []); // Retrieve stored history
377 115 $chat_mode = get_option("mxchat_mode_{$session_id}", 'ai'); // Get current chat mode
378 116
379 117 if (empty($history)) {
@@ -390,25 +128,11 @@
390 128 'chat_mode' => $chat_mode
391 129 ]);
392 130 wp_die();
393 131 }
394 -private function mxchat_fetch_conversation_history_for_ai($session_id, $session_start_timestamp = 0) {
132 +
133 +private function mxchat_fetch_conversation_history_for_ai($session_id) {
395 134 $history = get_option("mxchat_history_{$session_id}", []);
396 -
397 - // Check persistence setting - when OFF, only include messages from current page load
398 - $options = get_option('mxchat_options', []);
399 - $persistence_enabled = isset($options['chat_persistence_toggle']) && $options['chat_persistence_toggle'] === 'on';
400 -
401 - // Filter history when persistence is OFF to match what the user sees
402 - if (!$persistence_enabled && $session_start_timestamp > 0) {
403 - $history = array_filter($history, function($entry) use ($session_start_timestamp) {
404 - // Include messages from this page load onwards
405 - return isset($entry['timestamp']) && $entry['timestamp'] >= $session_start_timestamp;
406 - });
407 - // Re-index array after filtering
408 - $history = array_values($history);
409 - }
410 -
411 135 $formatted_history = [];
412 136
413 137 // Adjusted for code-heavy conversations
414 138 $max_tokens = 120000; // Context window size
@@ -482,17 +206,8 @@
482 206
483 207 public function register_routes() {
484 208 //error_log(esc_html__('Registering MxChat REST routes', 'mxchat'));
485 209
486 - // Per-request chat-send nonce endpoint — issues a fresh nonce on demand
487 - // so the chat widget never depends on a stale nonce embedded in cached HTML.
488 - // Public (no auth), rate-limited (1 call / IP / second via a transient).
489 - register_rest_route('mxchat/v1', '/nonce', [
490 - 'methods' => 'GET',
491 - 'callback' => [$this, 'mxchat_issue_chat_send_nonce'],
492 - 'permission_callback' => '__return_true',
493 - ]);
494 -
495 210 register_rest_route('mxchat/v1', '/stream', [
496 211 'methods' => 'GET',
497 212 'callback' => [$this, 'mxchat_stream_events'],
498 213 'permission_callback' => [$this, 'verify_chat_session'],
@@ -515,83 +230,12 @@
515 230 'callback' => [$this, 'handle_slack_messages'],
516 231 'permission_callback' => [$this, 'verify_slack_request'],
517 232 ]);
518 233
519 - // Telegram webhook endpoint
520 - register_rest_route('mxchat/v1', '/telegram-webhook', [
521 - 'methods' => 'POST',
522 - 'callback' => [$this, 'handle_telegram_webhook'],
523 - 'permission_callback' => [$this, 'verify_telegram_request'],
524 - ]);
525 -
526 234 //error_log(esc_html__('MxChat REST routes registered', 'mxchat'));
527 235 }
528 236
529 237 /**
530 - * Issue a fresh per-request nonce for chat-send. Returned to the widget which
531 - * caches it for the session and includes it on every chat-send / stream-send /
532 - * upload call. By moving the nonce out of inline `window.mxchatChat = {...}` HTML
533 - * we eliminate the entire class of "first-message Access denied" failures that
534 - * plague WP installs behind a full-page cache (WP Rocket, LiteSpeed, FlyingPress,
535 - * W3 Total Cache, Cloudflare APO) — the nonce is never cached because it never
536 - * lives in the HTML body.
537 - *
538 - * Public endpoint. Rate-limited to 1 call / IP / 1s via a transient so a single
539 - * client browser can't be used to flood the nonce-issuance path.
540 - *
541 - * Nonce action: `mxchat_chat_send` (new). The chat-send AJAX handlers accept
542 - * BOTH this action AND the legacy `mxchat_chat_nonce` action for a 30-day
543 - * backwards-compat window so cached pages still in users' browsers don't break
544 - * mid-session.
545 - *
546 - * @since 3.2.7
547 - */
548 -public function mxchat_issue_chat_send_nonce(WP_REST_Request $request) {
549 - $ip = '';
550 - if (!empty($_SERVER['REMOTE_ADDR'])) {
551 - $ip = preg_replace('#[^0-9a-fA-F:\.]#', '', wp_unslash((string) $_SERVER['REMOTE_ADDR']));
552 - }
553 - if ($ip !== '') {
554 - // Best-effort rate limit. WP transients with sub-second TTL are racy
555 - // (parallel bursts can squeak through before set_transient completes);
556 - // we use 2s to make the gate slightly more reliable. Real production
557 - // rate-limiting at sub-second granularity needs Redis or DB row locks
558 - // — out of scope for this endpoint, which is already cheap.
559 - $key = 'mxchat_nonce_rl_' . md5($ip);
560 - if (get_transient($key)) {
561 - return new WP_REST_Response(array(
562 - 'error' => 'rate_limited',
563 - 'message' => __('Too many nonce requests. Try again shortly.', 'mxchat'),
564 - ), 429);
565 - }
566 - set_transient($key, 1, 2);
567 - }
568 -
569 - return new WP_REST_Response(array(
570 - 'nonce' => wp_create_nonce('mxchat_chat_send'),
571 - 'expires_in' => 86400, // WP nonces live 24h; widget caches for 12h conservatively.
572 - ), 200);
573 -}
574 -
575 -/**
576 - * Verify a chat-send nonce. Accepts BOTH the new `mxchat_chat_send` action
577 - * (issued by /wp-json/mxchat/v1/nonce) AND the legacy `mxchat_chat_nonce`
578 - * action (inline-localized in older cached HTML). The legacy acceptance is
579 - * a 30-day backwards-compat window — to be removed in a follow-up release
580 - * after 2026-06-27.
581 - *
582 - * @param string $posted_nonce
583 - * @return bool
584 - */
585 -public static function mxchat_verify_chat_send_nonce($posted_nonce) {
586 - if (!is_string($posted_nonce) || $posted_nonce === '') {
587 - return false;
588 - }
589 - return (bool) wp_verify_nonce($posted_nonce, 'mxchat_chat_send')
590 - || (bool) wp_verify_nonce($posted_nonce, 'mxchat_chat_nonce');
591 -}
592 -
593 -/**
594 238 * Verify valid chat session
595 239 */
596 240 public function verify_chat_session($request) {
597 241 $session_id = $request->get_param('session_id');
@@ -627,11 +271,10 @@
627 271 //error_log(esc_html__('Slack request timestamp too old', 'mxchat'));
628 272 return false;
629 273 }
630 274
631 - // Get raw request body from the WP_REST_Request object
632 - // (php://input may already be consumed by WordPress at this point)
633 - $request_body = $request->get_body();
275 + // Get raw request body
276 + $request_body = file_get_contents('php://input');
634 277
635 278 // Create the signature base string
636 279 $sig_basestring = "v0:{$timestamp}:{$request_body}";
637 280
@@ -640,43 +283,8 @@
640 283
641 284 // Compare signatures
642 285 return hash_equals($my_signature, $slack_signature);
643 286 }
644 -
645 -/**
646 - * Verify request is coming from Telegram.
647 - *
648 - * @param WP_REST_Request $request
649 - * @return bool True if valid, false otherwise.
650 - */
651 -public function verify_telegram_request($request) {
652 - $secret_token = $this->options['telegram_webhook_secret'] ?? '';
653 -
654 - //error_log('[MxChat Telegram DEBUG] verify_telegram_request called');
655 - //error_log('[MxChat Telegram DEBUG] Stored secret: ' . (empty($secret_token) ? 'EMPTY' : substr($secret_token, 0, 10) . '...'));
656 -
657 - if (empty($secret_token)) {
658 - // If no secret is configured, allow the request (for initial setup)
659 - //error_log('[MxChat Telegram DEBUG] No secret configured, allowing request');
660 - return true;
661 - }
662 -
663 - // Telegram sends the secret token in the X-Telegram-Bot-Api-Secret-Token header
664 - $request_token = $request->get_header('X-Telegram-Bot-Api-Secret-Token');
665 -
666 - //error_log('[MxChat Telegram DEBUG] Request token: ' . (empty($request_token) ? 'EMPTY' : substr($request_token, 0, 10) . '...'));
667 -
668 - if (empty($request_token)) {
669 - //error_log('[MxChat Telegram DEBUG] Request rejected: No token in header');
670 - return false;
671 - }
672 -
673 - // Timing-safe comparison
674 - $result = hash_equals($secret_token, $request_token);
675 - //error_log('[MxChat Telegram DEBUG] Token comparison result: ' . ($result ? 'MATCH' : 'MISMATCH'));
676 - return $result;
677 -}
678 -
679 287 public function mxchat_stream_events(WP_REST_Request $request) {
680 288 header('Content-Type: text/event-stream');
681 289 header('Cache-Control: no-cache');
682 290 header('Connection: keep-alive');
@@ -710,9 +318,9 @@
710 318
711 319
712 320
713 321
714 -private function mxchat_save_chat_message($session_id, $role, $message, $originating_page = null, $rag_context = null) {
322 +private function mxchat_save_chat_message($session_id, $role, $message, $originating_page = null) {
715 323 global $wpdb;
716 324 $table_name = $wpdb->prefix . 'mxchat_chat_transcripts';
717 325 //error_log("[DEBUG] mxchat_save_chat_message -> START for session_id: {$session_id}, role: {$role}");
718 326
@@ -724,26 +332,14 @@
724 332 $session_id
725 333 ));
726 334 $is_new_session = ($existing_messages == 0);
727 335
728 - // Log for debugging
336 + // NEW: Log for debugging
729 337 if ($is_new_session) {
730 338 //error_log("[DEBUG] This is a NEW session - first message");
731 339 }
732 340 }
733 341
734 - // SECURITY FIX: Set session ownership for new sessions
735 - if ($is_new_session && $role === 'user') {
736 - $current_user_identifier = MxChat_User::mxchat_get_user_identifier();
737 - $session_owner_key = "mxchat_session_owner_{$session_id}";
738 -
739 - // Only set ownership if not already set
740 - if (!get_option($session_owner_key)) {
741 - update_option($session_owner_key, $current_user_identifier, 'no');
742 - //error_log("[DEBUG] Set session ownership for {$session_id} to {$current_user_identifier}");
743 - }
744 - }
745 -
746 342 // 1) Extract agent name if present
747 343 $agent_name = '';
748 344 if (preg_match('/^Agent: (.*?) - /', $message, $matches)) {
749 345 $agent_name = $matches[1];
@@ -775,9 +371,9 @@
775 371 $email_option_key = "mxchat_email_{$session_id}";
776 372 $saved_email = get_option($email_option_key);
777 373 //error_log("[DEBUG] mxchat_save_chat_message -> Checking wp_options for email_option_key: {$email_option_key}, found: {$saved_email}");
778 374
779 - // Check for a saved name in wp_options
375 + // NEW: Check for a saved name in wp_options
780 376 $name_option_key = "mxchat_name_{$session_id}";
781 377 $saved_name = get_option($name_option_key);
782 378 //error_log("[DEBUG] mxchat_save_chat_message -> Checking wp_options for name_option_key: {$name_option_key}, found: {$saved_name}");
783 379
@@ -820,9 +416,9 @@
820 416 $insert_data = [
821 417 'user_id' => $user_id,
822 418 'user_identifier'=> $user_identifier,
823 419 'user_email' => $saved_email ?: $user_email,
824 - 'user_name' => $saved_name ?: '', // Add name to insert data
420 + 'user_name' => $saved_name ?: '', // NEW: Add name to insert data
825 421 'session_id' => $session_id,
826 422 'role' => $role,
827 423 'message' => $message,
828 424 'timestamp' => current_time('mysql', 1),
@@ -886,17 +482,9 @@
886 482 $insert_data['originating_page_title'] = $stored_originating['title'] ?? '';
887 483 }
888 484 }
889 485 }
890 -
891 - // Add RAG context if provided (for bot messages)
892 - if ($rag_context !== null && $role === 'bot') {
893 - $rag_context_column_exists = $wpdb->get_var("SHOW COLUMNS FROM $table_name LIKE 'rag_context'");
894 - if ($rag_context_column_exists) {
895 - $insert_data['rag_context'] = is_array($rag_context) ? wp_json_encode($rag_context) : $rag_context;
896 - }
897 - }
898 -
486 +
899 487 $wpdb->insert($table_name, $insert_data);
900 488 //error_log("[DEBUG] mxchat_save_chat_message -> Inserted message into DB. row_id: {$wpdb->insert_id}, data: " . print_r($insert_data, true));
901 489
902 490 // 9) Send notification email if this is the first user message in a new session
@@ -907,17 +495,11 @@
907 495 'ip' => $_SERVER['REMOTE_ADDR']
908 496 ));
909 497 }
910 498
911 - // 10) Schedule delayed transcript email if enabled and message is from user
912 - if ($wpdb->insert_id && $role === 'user') {
913 - $this->schedule_delayed_transcript_email($session_id);
914 - }
915 -
916 499 //error_log("[DEBUG] mxchat_save_chat_message -> END for session_id: {$session_id}");
917 500 return $message_id;
918 501 }
919 -
920 502 private function send_new_chat_notification($session_id, $user_info = array()) {
921 503 $options = get_option('mxchat_transcripts_options');
922 504
923 505 // Check if notifications are enabled
@@ -960,202 +542,14 @@
960 542 // Send email
961 543 return wp_mail($to, $subject, $message);
962 544 }
963 545
964 -/**
965 - * Schedule delayed transcript email for a session
966 - * Reschedules if a new user message is received
967 - */
968 -private function schedule_delayed_transcript_email($session_id) {
969 - $options = get_option('mxchat_transcripts_options');
970 -
971 - // Check if auto-email is enabled
972 - if (empty($options['mxchat_auto_email_transcript_enabled'])) {
973 - return;
974 - }
975 -
976 - // Get notification email
977 - $email = !empty($options['mxchat_notification_email']) ?
978 - $options['mxchat_notification_email'] :
979 - get_option('admin_email');
980 -
981 - if (!is_email($email)) {
982 - return;
983 - }
984 -
985 - // Get delay in minutes (default 30)
986 - $delay_minutes = isset($options['mxchat_auto_email_transcript_delay']) ?
987 - intval($options['mxchat_auto_email_transcript_delay']) : 30;
988 -
989 - // Clear any existing scheduled event for this session
990 - $hook = 'mxchat_send_delayed_transcript';
991 - $args = array($session_id);
992 - $timestamp = wp_next_scheduled($hook, $args);
993 -
994 - if ($timestamp) {
995 - wp_unschedule_event($timestamp, $hook, $args);
996 - }
997 -
998 - // Schedule new event
999 - $schedule_time = time() + ($delay_minutes * 60);
1000 - wp_schedule_single_event($schedule_time, $hook, $args);
1001 -}
1002 -
1003 -/**
1004 - * Check if chat messages contain contact information (email or phone number)
1005 - *
1006 - * @param array $messages Array of message objects with 'message' property
1007 - * @param object|null $session_data Session data object with user_email property
1008 - * @return bool True if contact info found, false otherwise
1009 - */
1010 -private function chat_contains_contact_info($messages, $session_data = null) {
1011 - // Check if session already has a stored email
1012 - if ($session_data && !empty($session_data->user_email)) {
1013 - return true;
1014 - }
1015 -
1016 - // Email regex pattern
1017 - $email_pattern = '/\b[A-Za-z0-9._%+-]+@[A-Za-z0-9.-]+\.[A-Z|a-z]{2,}\b/';
1018 -
1019 - // Phone number patterns (covers various formats including international, WhatsApp style)
1020 - // Matches: +1234567890, (123) 456-7890, 123-456-7890, 123.456.7890, 1234567890, +1 234 567 8900, etc.
1021 - $phone_pattern = '/(?:\+?\d{1,3}[-.\s]?)?\(?\d{2,4}\)?[-.\s]?\d{2,4}[-.\s]?\d{2,4}(?:[-.\s]?\d{1,4})?/';
1022 -
1023 - // Only check user messages (not assistant responses)
1024 - foreach ($messages as $msg) {
1025 - if ($msg->role !== 'user') {
1026 - continue;
1027 - }
1028 -
1029 - $message_text = $msg->message;
1030 -
1031 - // Check for email
1032 - if (preg_match($email_pattern, $message_text)) {
1033 - return true;
1034 - }
1035 -
1036 - // Check for phone number (must be at least 7 digits total to avoid false positives)
1037 - if (preg_match($phone_pattern, $message_text, $matches)) {
1038 - // Count actual digits to avoid matching short numbers
1039 - $digits_only = preg_replace('/\D/', '', $matches[0]);
1040 - if (strlen($digits_only) >= 7) {
1041 - return true;
1042 - }
1043 - }
1044 - }
1045 -
1046 - return false;
1047 -}
1048 -
1049 -/**
1050 - * Send the delayed transcript email with .txt attachment
1051 - */
1052 -public function mxchat_send_delayed_transcript($session_id) {
1053 - global $wpdb;
1054 -
1055 - $options = get_option('mxchat_transcripts_options');
1056 -
1057 - // Get notification email
1058 - $to = !empty($options['mxchat_notification_email']) ?
1059 - $options['mxchat_notification_email'] :
1060 - get_option('admin_email');
1061 -
1062 - if (!is_email($to)) {
1063 - return false;
1064 - }
1065 -
1066 - // Get all messages for this session
1067 - $table_name = $wpdb->prefix . 'mxchat_chat_transcripts';
1068 - $messages = $wpdb->get_results($wpdb->prepare(
1069 - "SELECT role, message, timestamp FROM {$table_name}
1070 - WHERE session_id = %s
1071 - ORDER BY timestamp ASC",
1072 - $session_id
1073 - ));
1074 -
1075 - if (empty($messages)) {
1076 - return false;
1077 - }
1078 -
1079 - // Get session metadata
1080 - $sessions_table = $wpdb->prefix . 'mxchat_sessions';
1081 - $session_data = $wpdb->get_row($wpdb->prepare(
1082 - "SELECT * FROM {$sessions_table} WHERE session_id = %s",
1083 - $session_id
1084 - ));
1085 -
1086 - // Check if contact info is required and if it's present
1087 - $require_contact = !empty($options['mxchat_auto_email_transcript_require_contact']);
1088 - if ($require_contact && !$this->chat_contains_contact_info($messages, $session_data)) {
1089 - // Contact info required but not found - skip sending
1090 - return false;
1091 - }
1092 -
1093 - // Build transcript content
1094 - $transcript_content = "Chat Transcript\n";
1095 - $transcript_content .= "================\n\n";
1096 - $transcript_content .= "Session ID: " . $session_id . "\n";
1097 -
1098 - if ($session_data) {
1099 - $transcript_content .= "User: " . ($session_data->user_identifier ?: 'Guest') . "\n";
1100 - $transcript_content .= "Email: " . ($session_data->user_email ?: 'Not provided') . "\n";
1101 - $transcript_content .= "Started: " . $session_data->created_at . "\n";
1102 - }
1103 -
1104 - $transcript_content .= "\n" . str_repeat("=", 50) . "\n\n";
1105 -
1106 - // Add messages
1107 - foreach ($messages as $msg) {
1108 - $role_label = ($msg->role === 'user') ? 'User' : 'Assistant';
1109 - $transcript_content .= "[{$msg->timestamp}] {$role_label}:\n";
1110 - $transcript_content .= $msg->message . "\n\n";
1111 - }
1112 -
1113 - // Create temporary file for attachment using WP_Filesystem
1114 - $upload_dir = wp_upload_dir();
1115 - $temp_file = $upload_dir['basedir'] . '/mxchat-transcript-' . $session_id . '.txt';
1116 - global $wp_filesystem;
1117 - if (empty($wp_filesystem)) {
1118 - require_once ABSPATH . 'wp-admin/includes/file.php';
1119 - WP_Filesystem();
1120 - }
1121 - $wp_filesystem->put_contents($temp_file, $transcript_content, FS_CHMOD_FILE);
1122 -
1123 - // Prepare email
1124 - $subject = sprintf('[%s] Chat Transcript - Session %s', get_bloginfo('name'), substr($session_id, 0, 8));
1125 -
1126 - $message = "Please find attached the full chat transcript.\n\n";
1127 - $message .= "Session ID: {$session_id}\n";
1128 -
1129 - if ($session_data) {
1130 - $message .= "User: " . ($session_data->user_identifier ?: 'Guest') . "\n";
1131 - $message .= "Email: " . ($session_data->user_email ?: 'Not provided') . "\n";
1132 - }
1133 -
1134 - $message .= "\nView online: " . admin_url('admin.php?page=mxchat-transcripts');
1135 -
1136 - // Send email with attachment
1137 - $attachments = array($temp_file);
1138 - $result = wp_mail($to, $subject, $message, '', $attachments);
1139 -
1140 - // Clean up temporary file
1141 - if (file_exists($temp_file)) {
1142 - unlink($temp_file);
1143 - }
1144 -
1145 - return $result;
1146 -}
1147 -
1148 -
1149 -
1150 546 public function mxchat_handle_save_email_and_response() {
1151 547 //error_log('[DEBUG] ---------- mxchat_handle_save_email_and_response START ----------');
1152 548 //error_log('DEBUG: POST data: ' . print_r($_POST, true));
1153 549
1154 - nocache_headers();
1155 -
1156 550 // Validate nonce
1157 - if (!isset($_POST['nonce']) || !MxChat_Integrator::mxchat_verify_chat_send_nonce($_POST['nonce'])) {
551 + if (!isset($_POST['nonce']) || !wp_verify_nonce($_POST['nonce'], 'mxchat_chat_nonce')) {
1158 552 //error_log(esc_html__('[ERROR] Invalid nonce in mxchat_handle_save_email_and_response', 'mxchat'));
1159 553 wp_send_json_error(['message' => esc_html__('Invalid nonce.', 'mxchat')]);
1160 554 wp_die();
1161 555 }
@@ -1165,15 +559,15 @@
1165 559 $name = isset($_POST['name']) ? sanitize_text_field($_POST['name']) : '';
1166 560
1167 561 //error_log("[DEBUG] handle_save_email_and_response -> session_id: {$session_id}, email: {$email}, name: {$name}");
1168 562
1169 - if (empty($session_id) || $session_id === 'null' || empty($email)) {
563 + if (empty($session_id) || empty($email)) {
1170 564 //error_log("[ERROR] Missing session_id or email: session_id={$session_id}, email={$email}");
1171 565 wp_send_json_error(['message' => esc_html__('Session ID or email is missing.', 'mxchat')]);
1172 566 wp_die();
1173 567 }
1174 568
1175 - // Validate name if provided (check if name field is enabled and name is required)
569 + // NEW: Validate name if provided (check if name field is enabled and name is required)
1176 570 $options = get_option('mxchat_options', []);
1177 571 $name_field_enabled = isset($options['enable_name_field']) &&
1178 572 ($options['enable_name_field'] === '1' || $options['enable_name_field'] === 'on');
1179 573
@@ -1184,15 +578,15 @@
1184 578 }
1185 579
1186 580 // 1) Always store email in wp_options
1187 581 $email_option_key = "mxchat_email_{$session_id}";
1188 - update_option($email_option_key, $email, 'no');
582 + update_option($email_option_key, $email);
1189 583 //error_log("[DEBUG] handle_save_email_and_response -> updated option: {$email_option_key} => {$email}");
1190 584
1191 - // Store name in wp_options if provided
585 + // NEW: Store name in wp_options if provided
1192 586 if (!empty($name)) {
1193 587 $name_option_key = "mxchat_name_{$session_id}";
1194 - update_option($name_option_key, $name, 'no');
588 + update_option($name_option_key, $name);
1195 589 //error_log("[DEBUG] handle_save_email_and_response -> updated option: {$name_option_key} => {$name}");
1196 590 }
1197 591
1198 592 // 2) (Optional) Also store in DB if a row already exists
@@ -1205,9 +599,9 @@
1205 599
1206 600 //error_log("[DEBUG] handle_save_email_and_response -> session_count for {$session_id}: {$session_count} (SQL: {$sql})");
1207 601
1208 602 if ($session_count) {
1209 - // Update both user_email and user_name if row(s) exist
603 + // NEW: Update both user_email and user_name if row(s) exist
1210 604 if (!empty($name)) {
1211 605 $update_sql = $wpdb->prepare(
1212 606 "UPDATE {$table_name} SET user_email = %s, user_name = %s WHERE session_id = %s",
1213 607 $email,
@@ -1236,17 +630,15 @@
1236 630
1237 631 public function mxchat_check_email_provided() {
1238 632 //error_log('[DEBUG] ---------- mxchat_check_email_provided START ----------');
1239 633
1240 - nocache_headers();
1241 -
1242 - if (!isset($_POST['nonce']) || !MxChat_Integrator::mxchat_verify_chat_send_nonce($_POST['nonce'])) {
634 + if (!isset($_POST['nonce']) || !wp_verify_nonce($_POST['nonce'], 'mxchat_chat_nonce')) {
1243 635 //error_log('[ERROR] Invalid nonce in mxchat_check_email_provided');
1244 636 wp_send_json_error(['message' => esc_html__('Invalid nonce', 'mxchat')]);
1245 637 }
1246 638
1247 639 $session_id = isset($_POST['session_id']) ? sanitize_text_field($_POST['session_id']) : '';
1248 - if (empty($session_id) || $session_id === 'null') {
640 + if (empty($session_id)) {
1249 641 //error_log('[ERROR] No session ID provided in mxchat_check_email_provided');
1250 642 wp_send_json_error(['message' => esc_html__('No session ID provided', 'mxchat')]);
1251 643 }
1252 644
@@ -1254,9 +646,9 @@
1254 646 if (is_user_logged_in()) {
1255 647 $current_user = wp_get_current_user();
1256 648 //error_log("[DEBUG] User is logged in as {$current_user->user_email}");
1257 649
1258 - // Get user's display name for logged in users
650 + // NEW: Get user's display name for logged in users
1259 651 $user_name = !empty($current_user->display_name) ? $current_user->display_name :
1260 652 (!empty($current_user->first_name) ? $current_user->first_name : '');
1261 653
1262 654 $response_data = ['logged_in' => true, 'email' => $current_user->user_email];
@@ -1266,9 +658,9 @@
1266 658
1267 659 wp_send_json_success($response_data);
1268 660 }
1269 661
1270 - // Check if name field is required
662 + // NEW: Check if name field is required
1271 663 $options = get_option('mxchat_options', []);
1272 664 $name_field_enabled = isset($options['enable_name_field']) &&
1273 665 ($options['enable_name_field'] === '1' || $options['enable_name_field'] === 'on');
1274 666
@@ -1274,9 +666,9 @@
1274 666
1275 667 $email_option_key = "mxchat_email_{$session_id}";
1276 668 $stored_email = get_option($email_option_key, '');
1277 669
1278 - // Check for stored name
670 + // NEW: Check for stored name
1279 671 $name_option_key = "mxchat_name_{$session_id}";
1280 672 $stored_name = get_option($name_option_key, '');
1281 673
1282 674 //error_log("[DEBUG] mxchat_check_email_provided -> Checking email option: {$email_option_key}, found: {$stored_email}");
@@ -1281,9 +673,9 @@
1281 673
1282 674 //error_log("[DEBUG] mxchat_check_email_provided -> Checking email option: {$email_option_key}, found: {$stored_email}");
1283 675 //error_log("[DEBUG] mxchat_check_email_provided -> Checking name option: {$name_option_key}, found: {$stored_name}, required: " . ($name_field_enabled ? 'yes' : 'no'));
1284 676
1285 - // Check if we have email and name (if name is required)
677 + // NEW: Check if we have email and name (if name is required)
1286 678 $has_required_info = !empty($stored_email);
1287 679
1288 680 if ($name_field_enabled) {
1289 681 $has_required_info = $has_required_info && !empty($stored_name);
@@ -1303,59 +695,32 @@
1303 695 wp_send_json_error(['message' => esc_html__('No email found', 'mxchat')]);
1304 696 }
1305 697 }
1306 698
1307 -/**
1308 - * Send error response in appropriate format based on streaming mode
1309 - * ADDED: Helper method to consistently handle errors in both streaming and non-streaming modes
1310 - *
1311 - * @param string $error_message The error message to display
1312 - * @param string $error_code Optional error code for debugging
1313 - */
1314 -private function send_error_response($error_message, $error_code = 'api_error') {
1315 - if ($this->is_streaming) {
1316 - echo "data: " . json_encode([
1317 - 'error' => true,
1318 - 'error_message' => $error_message,
1319 - 'error_code' => $error_code,
1320 - 'text' => $error_message,
1321 - 'message' => $error_message
1322 - ]) . "\n\n";
1323 - echo "data: [DONE]\n\n";
1324 - flush();
1325 - } else {
1326 - wp_send_json_error([
1327 - 'error_message' => $error_message,
1328 - 'error_code' => $error_code
1329 - ]);
1330 - }
1331 - wp_die();
1332 -}
1333 -
1334 699 public function mxchat_handle_chat_request() {
1335 700 global $wpdb;
1336 701
1337 - // Debug: Log incoming bot_id
1338 - $bot_id = isset($_POST['bot_id']) ? sanitize_key($_POST['bot_id']) : 'default';
1339 - //error_log("=== MXCHAT DEBUG: Starting chat request ===");
1340 - //error_log("MXCHAT DEBUG: Bot ID received: " . $bot_id);
702 + // NEW: Check if this is a streaming request
703 + $is_streaming = isset($_POST['action']) && $_POST['action'] === 'mxchat_stream_chat';
1341 704
1342 - // Get bot-specific options
1343 - $bot_options = $this->get_bot_options($bot_id);
1344 - $current_options = !empty($bot_options) ? $bot_options : $this->options;
705 + // NEW: Set streaming headers if needed
706 + if ($is_streaming) {
707 + // Disable output buffering
708 + while (ob_get_level()) {
709 + ob_end_flush(); // Changed from ob_end_clean()
710 + }
711 +
712 + // Set headers for SSE
713 + header('Content-Type: text/event-stream');
714 + header('Cache-Control: no-cache');
715 + header('Connection: keep-alive');
716 + header('X-Accel-Buffering: no');
717 +
718 + // Add these new lines:
719 + ob_implicit_flush(true);
720 + flush();
721 + }
1345 722
1346 - // Check if this is a streaming request
1347 - // Allow force_streaming_test parameter to bypass the setting check (for admin compatibility testing)
1348 - $force_streaming_test = isset($_POST['force_streaming_test']) && $_POST['force_streaming_test'] === '1' && current_user_can('administrator');
1349 - $is_streaming = isset($_POST['action']) && $_POST['action'] === 'mxchat_stream_chat' &&
1350 - ($force_streaming_test || (isset($current_options['enable_streaming_toggle']) && $current_options['enable_streaming_toggle'] === 'on'));
1351 -
1352 - // ADDED: Store streaming state in class property for use in private methods
1353 - $this->is_streaming = $is_streaming;
1354 -
1355 - // NOTE: Streaming headers are now set later via setup_streaming_headers()
1356 - // This allows actions/forms to return JSON responses without header conflicts
1357 -
1358 723 // Check if MX Chat Moderation is active
1359 724 if (class_exists('MX_Chat_Moderation')) {
1360 725 // Get user email and IP
1361 726 $user_email = '';
@@ -1420,31 +785,13 @@
1420 785
1421 786 // Rest of your existing code...
1422 787 $session_id = isset($_POST['session_id']) ? sanitize_text_field($_POST['session_id']) : '';
1423 788
1424 - // Treat the literal strings 'null' / 'undefined' as missing too. Browser edge cases
1425 - // (Safari ITP, private mode, cross-origin iframes with partitioned storage) can cause
1426 - // the frontend FormData.append() to stringify a null session_id into the literal
1427 - // "null", which would otherwise pass empty() and pollute the transcripts table with
1428 - // ghost sessions that group every visitor's first message under one row.
1429 - if ($session_id === 'null' || $session_id === 'undefined') {
1430 - $session_id = '';
1431 - }
1432 -
1433 789 if (empty($session_id)) {
1434 790 wp_send_json_error(esc_html__('Session ID is missing.', 'mxchat'));
1435 791 wp_die();
1436 792 }
1437 793
1438 - // Update session owner if it changed (e.g. IP changed due to network switch)
1439 - // The session ID itself is the authentication — if the client has it, they own it
1440 - $current_user_identifier = MxChat_User::mxchat_get_user_identifier();
1441 - $session_owner = get_option("mxchat_session_owner_{$session_id}");
1442 -
1443 - if (!$session_owner || $session_owner !== $current_user_identifier) {
1444 - update_option("mxchat_session_owner_{$session_id}", $current_user_identifier, 'no');
1445 - }
1446 -
1447 794 // Validate and sanitize the incoming message
1448 795 if (empty($_POST['message'])) {
1449 796 wp_send_json_error(esc_html__('No message received.', 'mxchat'));
1450 797 wp_die();
@@ -1450,191 +797,209 @@
1450 797 wp_die();
1451 798 }
1452 799
1453 800
1454 - // Track originating page for first message in session
1455 - $table_name = $wpdb->prefix . 'mxchat_chat_transcripts';
801 + // NEW: Track originating page for first message in session
802 +$table_name = $wpdb->prefix . 'mxchat_chat_transcripts';
1456 803
1457 - // Check if originating page columns exist
1458 - $columns_exist = $wpdb->get_var("SHOW COLUMNS FROM $table_name LIKE 'originating_page_url'");
804 +// Check if originating page columns exist
805 +$columns_exist = $wpdb->get_var("SHOW COLUMNS FROM $table_name LIKE 'originating_page_url'");
1459 806
1460 - if ($columns_exist) {
1461 - // Check if this session already has messages
1462 - $message_count = $wpdb->get_var($wpdb->prepare(
1463 - "SELECT COUNT(*) FROM $table_name WHERE session_id = %s",
1464 - $session_id
1465 - ));
807 +if ($columns_exist) {
808 + // Check if this session already has messages
809 + $message_count = $wpdb->get_var($wpdb->prepare(
810 + "SELECT COUNT(*) FROM $table_name WHERE session_id = %s",
811 + $session_id
812 + ));
813 +
814 + // If this is the first message in the session
815 + if ($message_count == 0) {
816 + // Get originating page from JavaScript (preferred) or HTTP_REFERER (fallback)
817 + $originating_url = '';
818 + $originating_title = '';
1466 819
1467 - // If this is the first message in the session
1468 - if ($message_count == 0) {
1469 - // Get originating page from JavaScript (preferred) or HTTP_REFERER (fallback)
1470 - $originating_url = '';
1471 - $originating_title = '';
820 + // Try to get from POST data first (sent by JavaScript)
821 + if (isset($_POST['current_page_url'])) {
822 + $originating_url = esc_url_raw($_POST['current_page_url']);
823 + $originating_title = isset($_POST['current_page_title'])
824 + ? sanitize_text_field($_POST['current_page_title'])
825 + : '';
826 + }
827 + // Fallback to HTTP_REFERER if not provided by JavaScript
828 + else if (isset($_SERVER['HTTP_REFERER'])) {
829 + $originating_url = esc_url_raw($_SERVER['HTTP_REFERER']);
830 + }
831 +
832 + // Generate title if we have URL but no title
833 + if ($originating_url && empty($originating_title)) {
834 + $parsed_url = parse_url($originating_url);
835 + $path = isset($parsed_url['path']) ? trim($parsed_url['path'], '/') : '';
1472 836
1473 - // Try to get from POST data first (sent by JavaScript)
1474 - if (isset($_POST['current_page_url'])) {
1475 - $originating_url = esc_url_raw($_POST['current_page_url']);
1476 - $originating_title = isset($_POST['current_page_title'])
1477 - ? sanitize_text_field($_POST['current_page_title'])
1478 - : '';
837 + if (empty($path) || $path === 'index.php' || $path === 'index.html') {
838 + $originating_title = 'Homepage';
839 + } else {
840 + // Clean up the path to make a readable title
841 + $originating_title = str_replace(['-', '_', '/', '.php', '.html'], ' ', $path);
842 + $originating_title = ucwords(trim($originating_title));
1479 843 }
1480 - // Fallback to HTTP_REFERER if not provided by JavaScript
1481 - else if (isset($_SERVER['HTTP_REFERER'])) {
1482 - $originating_url = esc_url_raw($_SERVER['HTTP_REFERER']);
1483 - }
1484 -
1485 - // Generate title if we have URL but no title
1486 - if ($originating_url && empty($originating_title)) {
1487 - $parsed_url = parse_url($originating_url);
1488 - $path = isset($parsed_url['path']) ? trim($parsed_url['path'], '/') : '';
1489 -
1490 - if (empty($path) || $path === 'index.php' || $path === 'index.html') {
1491 - $originating_title = 'Homepage';
1492 - } else {
1493 - // Clean up the path to make a readable title
1494 - $originating_title = str_replace(['-', '_', '/', '.php', '.html'], ' ', $path);
1495 - $originating_title = ucwords(trim($originating_title));
1496 - }
1497 - }
1498 -
1499 - // Store for later use when saving the message
1500 - $this->pending_originating_page = [
1501 - 'url' => $originating_url,
1502 - 'title' => $originating_title
1503 - ];
1504 844 }
845 +
846 + // Store for later use when saving the message
847 + $this->pending_originating_page = [
848 + 'url' => $originating_url,
849 + 'title' => $originating_title
850 + ];
1505 851 }
852 +}
853 +
854 +
855 +
856 + // NEW: Get page context if provided
857 + $page_context = null;
858 + if (isset($_POST['page_context']) && !empty($_POST['page_context'])) {
859 + $page_context_raw = stripslashes($_POST['page_context']);
860 + $page_context = json_decode($page_context_raw, true);
1506 861
1507 -
1508 -
1509 - // Get page context if provided
1510 - $page_context = null;
1511 - if (isset($_POST['page_context']) && !empty($_POST['page_context'])) {
1512 - $page_context_raw = stripslashes($_POST['page_context']);
1513 - $page_context = json_decode($page_context_raw, true);
862 + // Validate page context structure
863 + if (is_array($page_context) &&
864 + isset($page_context['url']) &&
865 + isset($page_context['title']) &&
866 + isset($page_context['content'])) {
1514 867
1515 - // Validate page context structure
1516 - if (is_array($page_context) &&
1517 - isset($page_context['url']) &&
1518 - isset($page_context['title']) &&
1519 - isset($page_context['content'])) {
1520 -
1521 - // Sanitize page context
1522 - $page_context['url'] = esc_url_raw($page_context['url']);
1523 - $page_context['title'] = sanitize_text_field($page_context['title']);
1524 - $page_context['content'] = wp_kses_post($page_context['content']);
1525 - } else {
1526 - $page_context = null;
1527 - }
868 + // Sanitize page context
869 + $page_context['url'] = esc_url_raw($page_context['url']);
870 + $page_context['title'] = sanitize_text_field($page_context['title']);
871 + $page_context['content'] = wp_kses_post($page_context['content']);
872 + } else {
873 + $page_context = null;
1528 874 }
875 + }
1529 876
1530 - // Modify the message sanitization to preserve PHP tags in code blocks
1531 - $allowed_tags = [
1532 - 'pre' => [],
1533 - 'code' => ['class' => true],
1534 - 'span' => ['class' => true],
1535 - 'div' => ['class' => true],
1536 - ];
877 + // Modify the message sanitization to preserve PHP tags in code blocks
878 + $allowed_tags = [
879 + 'pre' => [],
880 + 'code' => ['class' => true],
881 + 'span' => ['class' => true],
882 + 'div' => ['class' => true],
883 + ];
1537 884
1538 - // First preserve code blocks
1539 - $message = preg_replace_callback('/<pre><code.*?>.*?<\/code><\/pre>/s', function($matches) {
1540 - return htmlspecialchars_decode($matches[0]);
1541 - }, $_POST['message']);
885 + // First preserve code blocks
886 + $message = preg_replace_callback('/<pre><code.*?>.*?<\/code><\/pre>/s', function($matches) {
887 + return htmlspecialchars_decode($matches[0]);
888 + }, $_POST['message']);
1542 889
1543 - // Then apply sanitization
1544 - $message = wp_kses($message, $allowed_tags);
890 + // Then apply sanitization
891 + $message = wp_kses($message, $allowed_tags);
1545 892
1546 - // Preserve code blocks from markdown conversion
1547 - $message = preg_replace('/```(\w+)?\s*([\s\S]+?)```/s', '<pre><code class="$1">$2</code></pre>', $message);
1548 - $message = apply_filters('mxchat_filter_message', $message, 'prompt', $session_id);
893 + // Preserve code blocks from markdown conversion
894 + $message = preg_replace('/```(\w+)?\s*([\s\S]+?)```/s', '<pre><code class="$1">$2</code></pre>', $message);
895 + $message = apply_filters('mxchat_filter_message', $message, 'prompt', $session_id);
1549 896
1550 - // ===== SIMPLIFIED TESTING PANEL INITIALIZATION =====
1551 - // Always initialize testing data for admins (no toggle needed)
1552 - $testing_data = null;
1553 - if (current_user_can('administrator')) {
1554 - // For vision messages, use the original user message for the query display
1555 - $query_for_testing = $message;
1556 - if (isset($_POST['vision_processed']) && $_POST['vision_processed'] && isset($_POST['original_user_message'])) {
1557 - $query_for_testing = sanitize_textarea_field($_POST['original_user_message']);
1558 - }
1559 -
1560 - $testing_data = [
1561 - 'query' => $query_for_testing,
1562 - 'timestamp' => time(),
1563 - 'top_matches' => [],
1564 - 'action_matches' => [], // Initialize action matches array
1565 - 'page_context' => $page_context, // Include page context in testing data
1566 - 'is_vision' => isset($_POST['vision_processed']) && $_POST['vision_processed'],
1567 - 'bot_id' => $bot_id // Include bot ID in testing data
1568 - ];
1569 -
1570 - // Get similarity threshold from bot options or default options
1571 - $similarity_threshold = isset($current_options['similarity_threshold'])
1572 - ? ((int) $current_options['similarity_threshold']) / 100
1573 - : 0.35;
1574 -
1575 - $testing_data['similarity_threshold'] = $similarity_threshold;
1576 -
1577 - // Determine knowledge base type using bot-specific config
1578 - $bot_pinecone_config = $this->get_bot_pinecone_config($bot_id);
1579 - $use_pinecone = isset($bot_pinecone_config['use_pinecone']) ? $bot_pinecone_config['use_pinecone'] : false;
1580 - $testing_data['knowledge_base_type'] = $use_pinecone ? 'Pinecone' : 'WordPress Database';
897 +// ===== SIMPLIFIED TESTING PANEL INITIALIZATION =====
898 + // Always initialize testing data for admins (no toggle needed)
899 + $testing_data = null;
900 + if (current_user_can('administrator')) {
901 + // For vision messages, use the original user message for the query display
902 + $query_for_testing = $message;
903 + if (isset($_POST['vision_processed']) && $_POST['vision_processed'] && isset($_POST['original_user_message'])) {
904 + $query_for_testing = sanitize_textarea_field($_POST['original_user_message']);
1581 905 }
1582 - // ===== END SIMPLIFIED TESTING INITIALIZATION =====
906 +
907 + $testing_data = [
908 + 'query' => $query_for_testing,
909 + 'timestamp' => time(),
910 + 'top_matches' => [],
911 + 'action_matches' => [], // NEW: Initialize action matches array
912 + 'page_context' => $page_context, // NEW: Include page context in testing data
913 + 'is_vision' => isset($_POST['vision_processed']) && $_POST['vision_processed']
914 + ];
915 +
916 + // Get similarity threshold
917 + $similarity_threshold = isset($this->options['similarity_threshold'])
918 + ? ((int) $this->options['similarity_threshold']) / 100
919 + : 0.75;
920 +
921 + $testing_data['similarity_threshold'] = $similarity_threshold;
922 +
923 + // Determine knowledge base type
924 + $addon_options = get_option('mxchat_pinecone_addon_options', array());
925 + $use_pinecone = (isset($addon_options['mxchat_use_pinecone']) && $addon_options['mxchat_use_pinecone'] === '1');
926 + $testing_data['knowledge_base_type'] = $use_pinecone ? 'Pinecone' : 'WordPress Database';
927 + }
928 + // ===== END SIMPLIFIED TESTING INITIALIZATION =====
1583 929
1584 - // Add debug before and after:
1585 - //error_log('MxChat Core: About to call mxchat_pre_process_message filter with message: ' . $message);
1586 - $pre_processed_result = apply_filters('mxchat_pre_process_message', $message, $user_id, $session_id);
1587 - //error_log('MxChat Core: Filter returned: ' . (is_array($pre_processed_result) ? 'array' : $pre_processed_result));
930 +// Add debug before and after:
931 +//error_log('MxChat Core: About to call mxchat_pre_process_message filter with message: ' . $message);
932 +$pre_processed_result = apply_filters('mxchat_pre_process_message', $message, $user_id, $session_id);
933 +//error_log('MxChat Core: Filter returned: ' . (is_array($pre_processed_result) ? 'array' : $pre_processed_result));
1588 934
1589 935
1590 - // If the pre-processing returned a result (not the original message), use it directly
1591 - if (is_array($pre_processed_result) && isset($pre_processed_result['text'])) {
1592 - // Save the AI response
1593 - $this->mxchat_save_chat_message($session_id, 'bot', $pre_processed_result['text']);
1594 -
1595 - // Save HTML content if provided
1596 - if (!empty($pre_processed_result['html'])) {
1597 - $this->mxchat_save_chat_message($session_id, 'bot', $pre_processed_result['html']);
1598 - }
1599 -
1600 - // Add testing data if admin
1601 - $response_data = [
1602 - 'text' => $pre_processed_result['text'],
1603 - 'html' => $pre_processed_result['html'] ?? '',
1604 - 'session_id' => $session_id
1605 - ];
1606 -
1607 - if ($testing_data !== null) {
1608 - $response_data['testing_data'] = $testing_data;
1609 - }
1610 -
1611 - wp_send_json($response_data);
1612 - wp_die();
936 + // If the pre-processing returned a result (not the original message), use it directly
937 + if (is_array($pre_processed_result) && isset($pre_processed_result['text'])) {
938 + // Save the AI response
939 + $this->mxchat_save_chat_message($session_id, 'bot', $pre_processed_result['text']);
940 +
941 + // Save HTML content if provided
942 + if (!empty($pre_processed_result['html'])) {
943 + $this->mxchat_save_chat_message($session_id, 'bot', $pre_processed_result['html']);
1613 944 }
945 +
946 + // Add testing data if admin
947 + $response_data = [
948 + 'text' => $pre_processed_result['text'],
949 + 'html' => $pre_processed_result['html'] ?? '',
950 + 'session_id' => $session_id
951 + ];
952 +
953 + if ($testing_data !== null) {
954 + $response_data['testing_data'] = $testing_data;
955 + }
956 +
957 + wp_send_json($response_data);
958 + wp_die();
959 + }
1614 960
1615 - // Save the user's message - handle vision processed messages differently
1616 - if (isset($_POST['vision_processed']) && $_POST['vision_processed'] && isset($_POST['original_user_message'])) {
1617 - // For vision messages, save the original user message with image indicator
1618 - $original_message = sanitize_textarea_field($_POST['original_user_message']);
1619 - if (isset($_POST['vision_images_count']) && $_POST['vision_images_count'] > 0) {
1620 - $image_count = intval($_POST['vision_images_count']);
1621 - $original_message .= " [{$image_count} image(s)]";
1622 - }
1623 - $this->mxchat_save_chat_message($session_id, 'user', $original_message);
1624 - } else {
1625 - // Regular message - save as normal
1626 - $this->mxchat_save_chat_message($session_id, 'user', $message);
961 + // Save the user's message - handle vision processed messages differently
962 + if (isset($_POST['vision_processed']) && $_POST['vision_processed'] && isset($_POST['original_user_message'])) {
963 + // For vision messages, save the original user message with image indicator
964 + $original_message = sanitize_textarea_field($_POST['original_user_message']);
965 + if (isset($_POST['vision_images_count']) && $_POST['vision_images_count'] > 0) {
966 + $image_count = intval($_POST['vision_images_count']);
967 + $original_message .= " [{$image_count} image(s)]";
1627 968 }
969 + $this->mxchat_save_chat_message($session_id, 'user', $original_message);
970 + } else {
971 + // Regular message - save as normal
972 + $this->mxchat_save_chat_message($session_id, 'user', $message);
973 + }
1628 974
975 +
976 +if (is_email($message)) {
977 + // Add the email to Loops
978 + $this->add_email_to_loops($message);
1629 979
1630 - if (is_email($message)) {
1631 - // Add the email to Loops
1632 - $this->add_email_to_loops($message);
980 + // Get the user's success message instruction
981 + $user_success_message = $this->options['email_capture_response'] ?? __('Thank you for providing your email! You\'ve been added to our list.', 'mxchat');
982 +
983 + // Set instruction for AI using the user's success message
984 + $this->current_action_instruction = $user_success_message;
985 +
986 + // Clear the email capture transient since we got the email
987 + delete_transient('mxchat_email_capture_' . $user_id);
988 + }
989 +
990 + // NEW: Check if we're in an email capture flow but user hasn't provided email yet
991 + elseif (get_transient('mxchat_email_capture_' . $user_id)) {
992 + // Check if the message contains an email (not the whole message being an email)
993 + if (preg_match('/\b[A-Za-z0-9._%+-]+@[A-Za-z0-9.-]+\.[A-Z|a-z]{2,}\b/', $message, $matches)) {
994 + $extracted_email = $matches[0];
1633 995
1634 - // Get the user's success message instruction using current_options
1635 - $user_success_message = $current_options['email_capture_response'] ?? __('Thank you for providing your email! You\'ve been added to our list.', 'mxchat');
996 + // Add the extracted email to Loops
997 + $this->add_email_to_loops($extracted_email);
1636 998
999 + // Get the user's success message instruction
1000 + $user_success_message = $this->options['email_capture_response'] ?? __('Thank you for providing your email! You\'ve been added to our list.', 'mxchat');
1001 +
1637 1002 // Set instruction for AI using the user's success message
1638 1003 $this->current_action_instruction = $user_success_message;
1639 1004
1640 1005 // Clear the email capture transient since we got the email
@@ -1639,692 +1004,468 @@
1639 1004
1640 1005 // Clear the email capture transient since we got the email
1641 1006 delete_transient('mxchat_email_capture_' . $user_id);
1642 1007 }
1643 -
1644 - // Check if we're in an email capture flow but user hasn't provided email yet
1645 - elseif (get_transient('mxchat_email_capture_' . $user_id)) {
1646 - // Check if the message contains an email (not the whole message being an email)
1647 - if (preg_match('/\b[A-Za-z0-9._%+-]+@[A-Za-z0-9.-]+\.[A-Z|a-z]{2,}\b/', $message, $matches)) {
1648 - $extracted_email = $matches[0];
1649 -
1650 - // Add the extracted email to Loops
1651 - $this->add_email_to_loops($extracted_email);
1652 -
1653 - // Get the user's success message instruction using current_options
1654 - $user_success_message = $current_options['email_capture_response'] ?? __('Thank you for providing your email! You\'ve been added to our list.', 'mxchat');
1655 -
1656 - // Set instruction for AI using the user's success message
1657 - $this->current_action_instruction = $user_success_message;
1658 -
1659 - // Clear the email capture transient since we got the email
1660 - delete_transient('mxchat_email_capture_' . $user_id);
1661 - }
1662 - // If no email found but we're in capture mode, remind them
1663 - else {
1664 - // Get the original instruction to remind them using current_options
1665 - $original_instruction = $current_options['triggered_phrase_response'] ?? __("Please provide your email address.", 'mxchat');
1666 - $this->current_action_instruction = $original_instruction;
1667 - }
1008 + // If no email found but we're in capture mode, remind them
1009 + else {
1010 + // Get the original instruction to remind them
1011 + $original_instruction = $this->options['triggered_phrase_response'] ?? __("Please provide your email address.", 'mxchat');
1012 + $this->current_action_instruction = $original_instruction;
1668 1013 }
1014 + }
1669 1015
1670 - $intent_info = '';
1016 + $intent_info = '';
1671 1017
1672 - // Check chat mode
1673 - $chat_mode = get_option("mxchat_mode_{$session_id}", 'ai');
1018 + // Check chat mode
1019 + $chat_mode = get_option("mxchat_mode_{$session_id}", 'ai');
1674 1020
1675 - // Handle agent mode
1676 1021 // Handle agent mode
1677 - if ($chat_mode === 'agent') {
1678 - // First, check for switch intent before doing anything else
1679 - $intent_matched = $this->mxchat_check_intent_and_invoke_callback($message, $user_id, $session_id);
1022 +// Handle agent mode
1023 + if ($chat_mode === 'agent') {
1024 + // First, check for switch intent before doing anything else
1025 + $intent_matched = $this->mxchat_check_intent_and_invoke_callback($message, $user_id, $session_id);
1680 1026
1681 - // Capture action analysis for testing panel after intent check
1682 - if ($testing_data !== null && isset($this->last_action_analysis) && !empty($this->last_action_analysis)) {
1683 - $testing_data['action_matches'] = $this->last_action_analysis;
1027 + // NEW: Capture action analysis for testing panel after intent check
1028 + if ($testing_data !== null && isset($this->last_action_analysis) && !empty($this->last_action_analysis)) {
1029 + $testing_data['action_matches'] = $this->last_action_analysis;
1030 + }
1031 +
1032 + // Around line 506, in the agent mode handling section:
1033 + if ($intent_matched && !empty($this->fallbackResponse['text'])) {
1034 + // Update chat mode first
1035 + update_option("mxchat_mode_{$session_id}", 'ai');
1036 +
1037 + // Clear any existing PDF context to start fresh
1038 + $this->clear_pdf_transients($session_id);
1039 +
1040 + // Prepare clean switch response with explicit chat_mode
1041 + $response_data = [
1042 + 'text' => $this->fallbackResponse['text'],
1043 + 'html' => $this->fallbackResponse['html'] ?? '',
1044 + 'session_id' => $session_id,
1045 + 'chat_mode' => 'ai' // EXPLICITLY SET THIS
1046 + ];
1047 +
1048 + if ($testing_data !== null) {
1049 + $response_data['testing_data'] = $testing_data;
1684 1050 }
1685 -
1686 - // Around line 506, in the agent mode handling section:
1687 - if ($intent_matched && !empty($this->fallbackResponse['text'])) {
1688 - // Update chat mode first
1689 - update_option("mxchat_mode_{$session_id}", 'ai');
1690 -
1691 - // Clear any existing PDF context to start fresh
1692 - $this->clear_pdf_transients($session_id);
1693 -
1694 - // Prepare clean switch response with explicit chat_mode
1695 - $response_data = [
1696 - 'text' => $this->fallbackResponse['text'],
1697 - 'html' => $this->fallbackResponse['html'] ?? '',
1698 - 'session_id' => $session_id,
1699 - 'chat_mode' => 'ai' // EXPLICITLY SET THIS
1051 +
1052 + // Save the mode switch message
1053 + $this->mxchat_save_chat_message($session_id, 'system', esc_html__('Switched to AI chat mode', 'mxchat'));
1054 + $this->mxchat_save_chat_message($session_id, 'bot', $this->fallbackResponse['text']);
1055 +
1056 + // Send response and exit
1057 + wp_send_json($response_data);
1058 + wp_die();
1059 + } elseif (!$intent_matched) {
1060 + // No intent matched, handle live agent message
1061 + try {
1062 + $this->mxchat_send_user_message_to_agent($message, $user_id, $session_id);
1063 +
1064 + $agent_response = [
1065 + 'status' => 'waiting_for_agent',
1066 + 'message' => esc_html__('Message sent to live agent.', 'mxchat')
1700 1067 ];
1701 -
1068 +
1702 1069 if ($testing_data !== null) {
1703 - $response_data['testing_data'] = $testing_data;
1070 + $agent_response['testing_data'] = $testing_data;
1704 1071 }
1705 -
1706 - // Save the mode switch message
1707 - $this->mxchat_save_chat_message($session_id, 'system', esc_html__('Switched to AI chat mode', 'mxchat'));
1708 - $this->mxchat_save_chat_message($session_id, 'bot', $this->fallbackResponse['text']);
1709 -
1710 - // Send response and exit
1711 - wp_send_json($response_data);
1712 - wp_die();
1713 - } elseif (!$intent_matched) {
1714 - // No intent matched, handle live agent message
1715 - try {
1716 - $this->mxchat_send_user_message_to_agent($message, $user_id, $session_id);
1717 1072
1718 - $agent_response = [
1719 - 'status' => 'waiting_for_agent',
1720 - 'message' => esc_html__('Message sent to live agent.', 'mxchat')
1721 - ];
1722 -
1723 - if ($testing_data !== null) {
1724 - $agent_response['testing_data'] = $testing_data;
1725 - }
1726 -
1727 - wp_send_json_success($agent_response);
1728 - } catch (\Exception $e) {
1729 - wp_send_json_error(esc_html__('Failed to send message to agent', 'mxchat'));
1730 - }
1731 - wp_die();
1073 + wp_send_json_success($agent_response);
1074 + } catch (\Exception $e) {
1075 + wp_send_json_error(esc_html__('Failed to send message to agent', 'mxchat'));
1732 1076 }
1077 + wp_die();
1733 1078 }
1079 + }
1734 1080
1735 - // Step 1: Check for new PDF URL in the message
1736 - if (!isset($_POST['vision_processed']) && preg_match('/https?:\/\/[^\s"]+/i', $message, $matches)) {
1737 - $new_pdf_url = $matches[0];
1081 + // Step 1: Check for new PDF URL in the message
1082 + if (!isset($_POST['vision_processed']) && preg_match('/https?:\/\/[^\s"]+/i', $message, $matches)) {
1083 + $new_pdf_url = $matches[0];
1738 1084
1739 - // Check if this is likely a PDF-related request
1740 - $pdf_keywords = ['pdf', 'document', 'read', 'analyze'];
1741 - $is_pdf_request = false;
1085 + // Check if this is likely a PDF-related request
1086 + $pdf_keywords = ['pdf', 'document', 'read', 'analyze'];
1087 + $is_pdf_request = false;
1742 1088
1743 - foreach ($pdf_keywords as $keyword) {
1744 - if (stripos($message, $keyword) !== false) {
1745 - $is_pdf_request = true;
1746 - break;
1747 - }
1089 + foreach ($pdf_keywords as $keyword) {
1090 + if (stripos($message, $keyword) !== false) {
1091 + $is_pdf_request = true;
1092 + break;
1748 1093 }
1094 + }
1749 1095
1750 - // If it looks like a PDF request or we're waiting for a PDF URL
1751 - if ($is_pdf_request || get_transient('mxchat_waiting_for_pdf_url_' . $session_id)) {
1752 - // Validate HTTPS
1753 - if (wp_http_validate_url($new_pdf_url) && parse_url($new_pdf_url, PHP_URL_SCHEME) === 'https') {
1754 - // Extract filename from URL
1755 - $pdf_filename = basename(parse_url($new_pdf_url, PHP_URL_PATH));
1096 + // If it looks like a PDF request or we're waiting for a PDF URL
1097 + if ($is_pdf_request || get_transient('mxchat_waiting_for_pdf_url_' . $session_id)) {
1098 + // Validate HTTPS
1099 + if (wp_http_validate_url($new_pdf_url) && parse_url($new_pdf_url, PHP_URL_SCHEME) === 'https') {
1100 + // Extract filename from URL
1101 + $pdf_filename = basename(parse_url($new_pdf_url, PHP_URL_PATH));
1756 1102
1757 - // Clear previous PDF transients
1758 - $this->clear_pdf_transients($session_id);
1103 + // Clear previous PDF transients
1104 + $this->clear_pdf_transients($session_id);
1759 1105
1760 - // Process new PDF using current_options
1761 - $max_pages = $current_options['pdf_max_pages'] ?? 69;
1762 - $embeddings = $this->fetch_and_split_pdf_pages($new_pdf_url, $max_pages);
1106 + // Process new PDF
1107 + $max_pages = $this->options['pdf_max_pages'] ?? 69;
1108 + $embeddings = $this->fetch_and_split_pdf_pages($new_pdf_url, $max_pages);
1763 1109
1764 - if ($embeddings === 'too_many_pages') {
1765 - $error_text = sprintf(
1766 - $current_options['pdf_intent_error_text'] ??
1767 - esc_html__("The provided PDF exceeds the maximum allowed limit of %d pages. Please provide a smaller document.", 'mxchat'),
1768 - $max_pages
1769 - );
1770 - $this->fallbackResponse['text'] = $error_text;
1771 - } elseif ($embeddings) {
1772 - // Store new PDF information
1773 - $pdf_filename = basename(parse_url($new_pdf_url, PHP_URL_PATH));
1110 + if ($embeddings === 'too_many_pages') {
1111 + $error_text = sprintf(
1112 + $this->options['pdf_intent_error_text'] ??
1113 + esc_html__("The provided PDF exceeds the maximum allowed limit of %d pages. Please provide a smaller document.", 'mxchat'),
1114 + $max_pages
1115 + );
1116 + $this->fallbackResponse['text'] = $error_text;
1117 + } elseif ($embeddings) {
1118 + // Store new PDF information
1119 + $pdf_filename = basename(parse_url($new_pdf_url, PHP_URL_PATH));
1774 1120
1775 - // If the filename is generic, create a more descriptive one
1776 - if (in_array($pdf_filename, ['results_download.php', 'download.php', 'view.php', 'pdf.php']) ||
1777 - strpos($pdf_filename, '.php') !== false) {
1778 - $pdf_filename = 'Document_' . date('Y-m-d_H-i') . '.pdf';
1779 - }
1121 + // If the filename is generic, create a more descriptive one
1122 + if (in_array($pdf_filename, ['results_download.php', 'download.php', 'view.php', 'pdf.php']) ||
1123 + strpos($pdf_filename, '.php') !== false) {
1124 + $pdf_filename = 'Document_' . date('Y-m-d_H-i') . '.pdf';
1125 + }
1780 1126
1781 - set_transient('mxchat_pdf_url_' . $session_id, $new_pdf_url, HOUR_IN_SECONDS);
1782 - set_transient('mxchat_pdf_filename_' . $session_id, $pdf_filename, HOUR_IN_SECONDS);
1783 - set_transient('mxchat_pdf_embeddings_' . $session_id, $embeddings, HOUR_IN_SECONDS);
1784 - set_transient('mxchat_include_pdf_in_context_' . $session_id, true, HOUR_IN_SECONDS);
1127 + set_transient('mxchat_pdf_url_' . $session_id, $new_pdf_url, HOUR_IN_SECONDS);
1128 + set_transient('mxchat_pdf_filename_' . $session_id, $pdf_filename, HOUR_IN_SECONDS);
1129 + set_transient('mxchat_pdf_embeddings_' . $session_id, $embeddings, HOUR_IN_SECONDS);
1130 + set_transient('mxchat_include_pdf_in_context_' . $session_id, true, HOUR_IN_SECONDS);
1785 1131
1786 - $success_text = $current_options['pdf_intent_success_text'] ??
1787 - esc_html__("I've processed the new PDF '{$pdf_filename}'. What questions do you have about it?", 'mxchat');
1132 + $success_text = $this->options['pdf_intent_success_text'] ??
1133 + esc_html__("I've processed the new PDF '{$pdf_filename}'. What questions do you have about it?", 'mxchat');
1788 1134
1789 - $pdf_response = [
1790 - 'success' => true,
1791 - 'message' => $success_text,
1792 - 'data' => [
1793 - 'filename' => $pdf_filename
1794 - ]
1795 - ];
1796 -
1797 - if ($testing_data !== null) {
1798 - $pdf_response['testing_data'] = $testing_data;
1799 - }
1800 -
1801 - wp_send_json($pdf_response);
1802 - wp_die();
1803 - } else {
1804 - $error_text = $current_options['pdf_intent_error_text'] ??
1805 - esc_html__("Sorry, I couldn't process the PDF. Please ensure it's a valid file.", 'mxchat');
1806 - $this->fallbackResponse['text'] = $error_text;
1807 - }
1808 -
1809 - $pdf_error_response = [
1810 - 'success' => false,
1811 - 'message' => $this->fallbackResponse['text']
1135 + $pdf_response = [
1136 + 'success' => true,
1137 + 'message' => $success_text,
1138 + 'data' => [
1139 + 'filename' => $pdf_filename
1140 + ]
1812 1141 ];
1813 1142
1814 1143 if ($testing_data !== null) {
1815 - $pdf_error_response['testing_data'] = $testing_data;
1144 + $pdf_response['testing_data'] = $testing_data;
1816 1145 }
1817 1146
1818 - wp_send_json($pdf_error_response);
1147 + wp_send_json($pdf_response);
1819 1148 wp_die();
1149 + } else {
1150 + $error_text = $this->options['pdf_intent_error_text'] ??
1151 + esc_html__("Sorry, I couldn't process the PDF. Please ensure it's a valid file.", 'mxchat');
1152 + $this->fallbackResponse['text'] = $error_text;
1820 1153 }
1821 - }
1822 - }
1823 1154
1824 -
1825 - // Step 2: Detect intent and handle intent-based responses
1826 - $intent_result = $this->mxchat_check_intent_and_invoke_callback($message, $user_id, $session_id);
1827 -
1828 - // Capture action analysis for testing panel after intent check
1829 - if ($testing_data !== null && isset($this->last_action_analysis) && !empty($this->last_action_analysis)) {
1830 - $testing_data['action_matches'] = $this->last_action_analysis;
1831 - }
1832 -
1833 - // Step 3: Handle the intent result appropriately
1834 - if ($intent_result !== false) {
1835 - // Intent was matched - ALWAYS send as JSON response, never streaming
1836 -
1837 - if (is_array($intent_result) && (isset($intent_result['text']) || isset($intent_result['html']))) {
1838 - // Intent returned a direct response array
1839 - $response_data = [
1840 - 'text' => $intent_result['text'] ?? '',
1841 - 'html' => $intent_result['html'] ?? '',
1842 - 'session_id' => $session_id
1155 + $pdf_error_response = [
1156 + 'success' => false,
1157 + 'message' => $this->fallbackResponse['text']
1843 1158 ];
1844 -
1845 - // IMPORTANT: Include chat_mode if present (for WhatsApp, Slack, etc.)
1846 - if (isset($intent_result['chat_mode'])) {
1847 - $response_data['chat_mode'] = $intent_result['chat_mode'];
1848 - }
1849 -
1159 +
1850 1160 if ($testing_data !== null) {
1851 - $response_data['testing_data'] = $testing_data;
1161 + $pdf_error_response['testing_data'] = $testing_data;
1852 1162 }
1853 1163
1854 - wp_send_json($response_data);
1164 + wp_send_json($pdf_error_response);
1855 1165 wp_die();
1856 - } else if ($intent_result === true && (!empty($this->fallbackResponse['text']) || !empty($this->fallbackResponse['html']))) {
1857 - // Intent returned true and set fallbackResponse
1858 -
1859 - // SAVE TO TRANSCRIPT
1860 - if (!empty($this->fallbackResponse['text'])) {
1861 - $this->mxchat_save_chat_message($session_id, 'bot', $this->fallbackResponse['text']);
1862 - }
1863 - // Save action HTML (product cards, featured products, etc.) so it renders in transcripts
1864 - if (!empty($this->fallbackResponse['html'])) {
1865 - $this->mxchat_save_chat_message($session_id, 'bot', $this->fallbackResponse['html']);
1866 - }
1867 -
1868 - $response_data = [
1869 - 'text' => $this->fallbackResponse['text'] ?? '',
1870 - 'html' => $this->fallbackResponse['html'] ?? '',
1871 - 'session_id' => $session_id
1872 - ];
1873 -
1874 - if (isset($this->fallbackResponse['chat_mode'])) {
1875 - $response_data['chat_mode'] = $this->fallbackResponse['chat_mode'];
1876 - }
1877 -
1878 - if ($testing_data !== null) {
1879 - $response_data['testing_data'] = $testing_data;
1880 - }
1881 -
1882 - wp_send_json($response_data);
1883 - wp_die();
1884 1166 }
1885 1167 }
1168 + }
1886 1169
1887 - // If we get here, no intent matched OR the intent didn't provide a usable response
1888 -
1889 - // Step 4: Generate AI response
1890 - // Get session start timestamp - when persistence is OFF, only include messages from this page load
1891 - $session_start_timestamp = isset($_POST['session_start_timestamp']) ? intval($_POST['session_start_timestamp']) : 0;
1892 - $conversation_history = $this->mxchat_fetch_conversation_history_for_ai($session_id, $session_start_timestamp);
1893 - $this->mxchat_increment_chat_count();
1170 + // Check if there's an active recommendation flow session
1171 + $flow_state = get_option("mxchat_sr_flow_state_{$session_id}", array());
1172 + if (!empty($flow_state) && isset($flow_state['flow_id'])) {
1173 + // Create a dummy intent object that matches the original intent
1174 + $dummy_intent = new stdClass();
1175 + $dummy_intent->intent_label = 'Recommendation Flow ' . $flow_state['flow_id'];
1176 + $dummy_intent->phrases = ''; // Empty phrases to avoid matching the original trigger
1894 1177
1895 - // Generate embedding for the user's query - USE BOT-SPECIFIC API KEY
1896 - $api_key = $current_options['api_key'] ?? $this->options['api_key'];
1897 - $user_message_embedding = $this->mxchat_generate_embedding($message, $api_key);
1178 + // Call the recommendation flow handler directly
1179 + $response_data = apply_filters('mxchat_sr_recommendation_flow', false, $message, $user_id, $session_id, $dummy_intent);
1898 1180
1899 - // Check if the embedding generation returned an error
1900 - if (is_array($user_message_embedding) && isset($user_message_embedding['error'])) {
1901 - $error_message = $user_message_embedding['error'];
1902 - $error_code = $user_message_embedding['error_code'] ?? 'embedding_error';
1903 -
1904 - // FIXED: Send error in appropriate format based on streaming mode
1905 - if ($is_streaming) {
1906 - echo "data: " . json_encode([
1907 - 'error' => true,
1908 - 'error_message' => $error_message,
1909 - 'error_code' => $error_code,
1910 - 'text' => $error_message,
1911 - 'message' => $error_message
1912 - ]) . "\n\n";
1913 - echo "data: [DONE]\n\n";
1914 - flush();
1915 - } else {
1916 - wp_send_json_error([
1917 - 'error_message' => $error_message,
1918 - 'error_code' => $error_code
1919 - ]);
1181 + // If the handler returned a response, send it
1182 + if (is_array($response_data) && (isset($response_data['text']) || isset($response_data['html']))) {
1183 + // Save the bot's response to the chat history
1184 + if (!empty($response_data['text'])) {
1185 + $this->mxchat_save_chat_message($session_id, 'bot', $response_data['text']);
1920 1186 }
1187 + if (!empty($response_data['html'])) {
1188 + $this->mxchat_save_chat_message($session_id, 'bot', $response_data['html']);
1189 + }
1190 +
1191 + if ($testing_data !== null) {
1192 + $response_data['testing_data'] = $testing_data;
1193 + }
1194 +
1195 + // Send the response
1196 + wp_send_json($response_data);
1921 1197 wp_die();
1922 1198 }
1199 + }
1923 1200
1924 - // Check if the embedding is valid
1925 - if (!is_array($user_message_embedding) || empty($user_message_embedding)) {
1926 - $error_message = esc_html__('Unable to process your message. The embedding service is not responding correctly.', 'mxchat');
1201 + // Step 2: Detect intent and handle intent-based responses
1202 + $intent_result = $this->mxchat_check_intent_and_invoke_callback($message, $user_id, $session_id);
1927 1203
1928 - // FIXED: Send error in appropriate format based on streaming mode
1204 + // NEW: Capture action analysis for testing panel after intent check
1205 + if ($testing_data !== null && isset($this->last_action_analysis) && !empty($this->last_action_analysis)) {
1206 + $testing_data['action_matches'] = $this->last_action_analysis;
1207 + }
1208 +
1209 + // Step 3: Handle the intent result appropriately
1210 + if ($intent_result !== false) {
1211 + // Intent was matched - ALWAYS send as JSON response, never streaming
1212 +
1213 + if (is_array($intent_result) && (isset($intent_result['text']) || isset($intent_result['html']))) {
1214 + // Intent returned a direct response array
1215 + $response_data = [
1216 + 'text' => $intent_result['text'] ?? '',
1217 + 'html' => $intent_result['html'] ?? '',
1218 + 'session_id' => $session_id
1219 + ];
1220 +
1221 + if ($testing_data !== null) {
1222 + $response_data['testing_data'] = $testing_data;
1223 + }
1224 +
1225 + // Clear streaming headers if they were set
1929 1226 if ($is_streaming) {
1930 - echo "data: " . json_encode([
1931 - 'error' => true,
1932 - 'error_message' => $error_message,
1933 - 'error_code' => 'invalid_embedding',
1934 - 'text' => $error_message,
1935 - 'message' => $error_message
1936 - ]) . "\n\n";
1937 - echo "data: [DONE]\n\n";
1938 - flush();
1939 - } else {
1940 - wp_send_json_error([
1941 - 'error_message' => $error_message,
1942 - 'error_code' => 'invalid_embedding'
1943 - ]);
1227 + header_remove('Content-Type');
1228 + header_remove('Cache-Control');
1229 + header_remove('Connection');
1230 + header_remove('X-Accel-Buffering');
1231 + header('Content-Type: application/json');
1944 1232 }
1233 +
1234 + wp_send_json($response_data);
1945 1235 wp_die();
1946 - }
1947 -
1948 - // Build context with both knowledge base and PDF content if available
1949 - $context_content = "User asked: '{$message}'\n\n";
1950 -
1951 - // Add action instruction if present (add this right after the above line)
1952 - if (!empty($this->current_action_instruction)) {
1953 - $context_content .= "===== SPECIAL INSTRUCTION =====\n";
1954 - $context_content .= "IMPORTANT: " . $this->current_action_instruction . "\n";
1955 - $context_content .= "Respond naturally and conversationally while following this instruction.\n";
1956 - $context_content .= "===== END SPECIAL INSTRUCTION =====\n\n";
1236 + } else if ($intent_result === true && (!empty($this->fallbackResponse['text']) || !empty($this->fallbackResponse['html']))) {
1237 + // Intent returned true and set fallbackResponse
1957 1238
1958 - // Clear the instruction after using it
1959 - $this->current_action_instruction = null;
1960 - }
1961 -
1962 -
1963 - // Add page context if available and contextual awareness is enabled using current_options
1964 - if ($page_context && isset($current_options['contextual_awareness_toggle']) && $current_options['contextual_awareness_toggle'] === 'on') {
1965 - $context_content .= "===== CURRENT PAGE CONTEXT =====\n";
1966 - $context_content .= "Page URL: " . $page_context['url'] . "\n";
1967 - $context_content .= "Page Title: " . $page_context['title'] . "\n";
1968 - $context_content .= "Page Content: " . $page_context['content'] . "\n";
1969 - $context_content .= "===== END CURRENT PAGE CONTEXT =====\n\n";
1970 - }
1971 -
1972 - // Get relevant content from knowledge base - PASS BOT_ID and MESSAGE for Vector Store
1973 - $relevant_content = $this->mxchat_find_relevant_content($user_message_embedding, $bot_id, $message);
1974 -
1975 - // NEW: Also extract URLs from system instructions (only if citation links enabled)
1976 - // Use fresh options to ensure we get the latest setting value
1977 - $fresh_options = get_option('mxchat_options', []);
1978 - $citation_links_enabled = isset($fresh_options['citation_links_toggle']) ? ($fresh_options['citation_links_toggle'] === 'on') : true;
1979 -
1980 - $system_instructions = $this->get_system_instructions($bot_id, $session_id);
1981 - if ($citation_links_enabled && !empty($system_instructions)) {
1982 - preg_match_all(
1983 - '#\bhttps?://[^\s<>"\']+#i',
1984 - $system_instructions,
1985 - $system_instruction_urls
1986 - );
1987 -
1988 - if (!empty($system_instruction_urls[0])) {
1989 - // Merge with existing valid URLs
1990 - $this->current_valid_urls = array_merge(
1991 - $this->current_valid_urls,
1992 - $system_instruction_urls[0]
1993 - );
1994 - // Remove duplicates
1995 - $this->current_valid_urls = array_unique($this->current_valid_urls);
1996 -
1997 - //error_log("Added " . count($system_instruction_urls[0]) . " URLs from system instructions");
1239 + // SAVE TO TRANSCRIPT FIRST
1240 + if (!empty($this->fallbackResponse['text'])) {
1241 + $this->mxchat_save_chat_message($session_id, 'bot', $this->fallbackResponse['text']);
1998 1242 }
1999 - }
2000 -
2001 -// ===== CAPTURE REAL SIMILARITY DATA FOR ADMINS =====
2002 -if ($testing_data !== null && $this->last_similarity_analysis !== null) {
2003 - // Update testing data with the REAL similarity analysis
2004 - $testing_data['top_matches'] = $this->last_similarity_analysis['top_matches'];
2005 - $testing_data['total_documents_checked'] = $this->last_similarity_analysis['total_checked'] ?? 0;
2006 - $testing_data['knowledge_base_type'] = $this->last_similarity_analysis['knowledge_base_type'];
2007 - $testing_data['sources_used'] = $this->last_similarity_analysis['sources_used'] ?? 0;
2008 - $testing_data['total_chunks_used'] = $this->last_similarity_analysis['total_chunks_used'] ?? 0;
2009 -}
2010 -// ===== END SIMILARITY DATA CAPTURE =====
2011 -
2012 -// NEW: Add valid URLs to testing data for admin panel display (AFTER similarity data)
2013 -if ($testing_data !== null && !empty($this->current_valid_urls)) {
2014 - $testing_data['approved_urls'] = array_values($this->current_valid_urls);
2015 - //error_log("Added " . count($this->current_valid_urls) . " approved URLs to testing data");
2016 -}
2017 -
2018 - if (!empty($relevant_content)) {
2019 - $context_content .= "===== OFFICIAL KNOWLEDGE DATABASE CONTENT =====\n" . $relevant_content . "\n===== END OF OFFICIAL KNOWLEDGE DATABASE CONTENT =====\n\n";
2020 - } else {
2021 - $context_content .= "===== NO RELEVANT CONTENT FOUND IN KNOWLEDGE DATABASE =====\n";
2022 - }
2023 -
2024 - // NEW: Add approved URLs list to context for AI (only if citation links enabled)
2025 - if ($citation_links_enabled && !empty($this->current_valid_urls)) {
2026 - $context_content .= "===== APPROVED URLS FOR CITATIONS =====\n";
2027 - $context_content .= "You may ONLY use these exact URLs in your response:\n";
2028 - foreach ($this->current_valid_urls as $url) {
2029 - $context_content .= "- " . $url . "\n";
1243 + if (!empty($this->fallbackResponse['html'])) {
1244 + $this->mxchat_save_chat_message($session_id, 'bot', $this->fallbackResponse['html']);
2030 1245 }
2031 - $context_content .= "\nCRITICAL: Do NOT create, modify, extend, or invent any other URLs. ";
2032 - $context_content .= "===== END APPROVED URLS =====\n\n";
2033 - }
2034 -
2035 - // Check for and include PDF content
2036 - $pdf_url = get_transient('mxchat_pdf_url_' . $session_id);
2037 - $pdf_embeddings = get_transient('mxchat_pdf_embeddings_' . $session_id);
2038 - $pdf_filename = get_transient('mxchat_pdf_filename_' . $session_id);
2039 - if ($pdf_url && $pdf_embeddings && get_transient('mxchat_include_pdf_in_context_' . $session_id)) {
2040 - $relevant_pdf_pages = $this->find_relevant_pdf_pages($user_message_embedding, $pdf_embeddings);
2041 - if (!empty($relevant_pdf_pages)) {
2042 - $context_content .= "Relevant content from PDF document '{$pdf_filename}':\n";
2043 - foreach ($relevant_pdf_pages as $page_data) {
2044 - $context_content .= "Page {$page_data['page_number']} of '{$pdf_filename}': {$page_data['text']}\n";
2045 - }
2046 - $context_content .= "\n";
1246 +
1247 + $response_data = [
1248 + 'text' => $this->fallbackResponse['text'] ?? '',
1249 + 'html' => $this->fallbackResponse['html'] ?? '',
1250 + 'session_id' => $session_id
1251 + ];
1252 +
1253 + if (isset($this->fallbackResponse['chat_mode'])) {
1254 + $response_data['chat_mode'] = $this->fallbackResponse['chat_mode'];
2047 1255 }
2048 - }
2049 -
2050 - // Check for and include Word content
2051 - $word_url = get_transient('mxchat_word_url_' . $session_id);
2052 - $word_embeddings = get_transient('mxchat_word_embeddings_' . $session_id);
2053 - $word_filename = get_transient('mxchat_word_filename_' . $session_id);
2054 - if ($word_url && $word_embeddings && get_transient('mxchat_include_word_in_context_' . $session_id)) {
2055 - $relevant_word_chunks = $this->word_handler->mxchat_find_relevant_word_chunks($user_message_embedding, $word_embeddings);
2056 - if (!empty($relevant_word_chunks)) {
2057 - $context_content .= "Relevant content from Word document '{$word_filename}':\n";
2058 - foreach ($relevant_word_chunks as $chunk_data) {
2059 - $context_content .= "Section {$chunk_data['chunk_number']} of '{$word_filename}': {$chunk_data['text']}\n";
2060 - }
2061 - $context_content .= "\n";
1256 +
1257 + if ($testing_data !== null) {
1258 + $response_data['testing_data'] = $testing_data;
2062 1259 }
1260 +
1261 + // Clear streaming headers if they were set
1262 + if ($is_streaming) {
1263 + header_remove('Content-Type');
1264 + header_remove('Cache-Control');
1265 + header_remove('Connection');
1266 + header_remove('X-Accel-Buffering');
1267 + header('Content-Type: application/json');
1268 + }
1269 +
1270 + wp_send_json($response_data);
1271 + wp_die();
2063 1272 }
2064 -
2065 - $context_content = apply_filters('mxchat_prepare_context', $context_content, $session_id);
1273 + }
2066 1274
2067 - // Extract model from current options for bot-specific model support
2068 - $selected_model = isset($current_options['model']) ? $current_options['model'] : 'gpt-5.1-chat-latest';
1275 + // If we get here, no intent matched OR the intent didn't provide a usable response
1276 +
1277 + // Step 4: Generate AI response
1278 + $conversation_history = $this->mxchat_fetch_conversation_history_for_ai($session_id);
1279 + $this->mxchat_increment_chat_count();
1280 +
1281 + // Generate embedding for the user's query
1282 + $user_message_embedding = $this->mxchat_generate_embedding($message, $this->options['api_key']);
1283 +
1284 + // Check if the embedding generation returned an error
1285 + if (is_array($user_message_embedding) && isset($user_message_embedding['error'])) {
1286 + $error_message = $user_message_embedding['error'];
1287 + $error_code = $user_message_embedding['error_code'] ?? 'embedding_error';
2069 1288
2070 - $response = $this->mxchat_generate_response(
2071 - $context_content,
2072 - $current_options['api_key'] ?? $this->options['api_key'],
2073 - $current_options['xai_api_key'] ?? $this->options['xai_api_key'],
2074 - $current_options['claude_api_key'] ?? $this->options['claude_api_key'],
2075 - $current_options['deepseek_api_key'] ?? $this->options['deepseek_api_key'],
2076 - $current_options['gemini_api_key'] ?? $this->options['gemini_api_key'],
2077 - $current_options['openrouter_api_key'] ?? $this->options['openrouter_api_key'],
2078 - $conversation_history,
2079 - $is_streaming,
2080 - $session_id,
2081 - $testing_data,
2082 - $selected_model
2083 - );
2084 -
2085 - // Handle streaming vs non-streaming responses
2086 - if ($is_streaming) {
2087 - // Check if streaming actually happened or if it fell back to regular response
2088 - if ($response === true) {
2089 - wp_die();
2090 - }
2091 - // If we get here, streaming fell back to regular response, continue
2092 - // But if there's an error, we need to send it as SSE format since headers are already set
2093 - if (is_array($response) && isset($response['error'])) {
2094 - $error_message = $response['error'];
2095 - $error_code = $response['error_code'] ?? 'api_error';
2096 - // Send error in SSE format that the client JS can handle
2097 - echo "data: " . json_encode([
2098 - 'error' => true,
2099 - 'error_message' => $error_message,
2100 - 'error_code' => $error_code,
2101 - 'text' => $error_message, // Also include as text for fallback handling
2102 - 'message' => $error_message
2103 - ]) . "\n\n";
2104 - echo "data: [DONE]\n\n";
2105 - flush();
2106 - wp_die();
2107 - }
2108 - }
1289 + wp_send_json_error([
1290 + 'error_message' => $error_message,
1291 + 'error_code' => $error_code
1292 + ]);
1293 + wp_die();
1294 + }
1295 +
1296 + // Check if the embedding is valid
1297 + if (!is_array($user_message_embedding) || empty($user_message_embedding)) {
1298 + wp_send_json_error([
1299 + 'error_message' => esc_html__('Unable to process your message. The embedding service is not responding correctly.', 'mxchat'),
1300 + 'error_code' => 'invalid_embedding'
1301 + ]);
1302 + wp_die();
1303 + }
2109 1304
2110 - // Check if the response is an error array (non-streaming mode)
2111 - if (is_array($response) && isset($response['error'])) {
2112 - wp_send_json_error([
2113 - 'error_message' => $response['error'],
2114 - 'error_code' => $response['error_code'] ?? 'api_error'
2115 - ]);
2116 - wp_die();
2117 - }
1305 + // Build context with both knowledge base and PDF content if available
1306 + $context_content = "User asked: '{$message}'\n\n";
1307 +
1308 + // NEW: Add action instruction if present (add this right after the above line)
1309 + if (!empty($this->current_action_instruction)) {
1310 + $context_content .= "===== SPECIAL INSTRUCTION =====\n";
1311 + $context_content .= "IMPORTANT: " . $this->current_action_instruction . "\n";
1312 + $context_content .= "Respond naturally and conversationally while following this instruction.\n";
1313 + $context_content .= "===== END SPECIAL INSTRUCTION =====\n\n";
2118 1314
2119 - // DEBUG: Check what we have
2120 - //error_log("=== BEFORE URL VALIDATION ===");
2121 - //error_log("current_valid_urls is empty? " . (empty($this->current_valid_urls) ? 'YES' : 'NO'));
2122 - //error_log("current_valid_urls count: " . count($this->current_valid_urls));
2123 - //error_log("current_valid_urls content: " . print_r($this->current_valid_urls, true));
2124 -
2125 - // If we get here, the response is valid text - now validate URLs
2126 - if (!empty($this->current_valid_urls)) {
2127 - //error_log("CALLING validate_and_clean_urls");
2128 - $response = $this->validate_and_clean_urls($response, $this->current_valid_urls);
2129 - } else {
2130 - //error_log("SKIPPING validation - current_valid_urls is empty");
2131 - }
2132 - // ===== END URL VALIDATION =====
1315 + // Clear the instruction after using it
1316 + $this->current_action_instruction = null;
1317 + }
2133 1318
2134 - // Prepare RAG context data for storage (only include documents used for context)
2135 - $rag_context_for_storage = null;
2136 - $has_rag_data = $this->last_similarity_analysis !== null && !empty($this->last_similarity_analysis['top_matches']);
2137 - $has_action_data = isset($this->last_action_analysis) && !empty($this->last_action_analysis);
2138 1319
2139 - if ($has_rag_data || $has_action_data) {
2140 - $rag_context_for_storage = [];
1320 + // NEW: Add page context if available and contextual awareness is enabled
1321 + if ($page_context && isset($this->options['contextual_awareness_toggle']) && $this->options['contextual_awareness_toggle'] === 'on') {
1322 + $context_content .= "===== CURRENT PAGE CONTEXT =====\n";
1323 + $context_content .= "Page URL: " . $page_context['url'] . "\n";
1324 + $context_content .= "Page Title: " . $page_context['title'] . "\n";
1325 + $context_content .= "Page Content: " . $page_context['content'] . "\n";
1326 + $context_content .= "===== END CURRENT PAGE CONTEXT =====\n\n";
1327 + }
2141 1328
2142 - // Add RAG/source data if available
2143 - if ($has_rag_data) {
2144 - $rag_context_for_storage['top_matches'] = $this->last_similarity_analysis['top_matches'];
2145 - $rag_context_for_storage['approved_urls'] = $this->current_valid_urls ?? [];
2146 - $rag_context_for_storage['similarity_threshold'] = $this->last_similarity_analysis['threshold_used'] ?? 0.35;
2147 - $rag_context_for_storage['knowledge_base_type'] = $this->last_similarity_analysis['knowledge_base_type'] ?? 'WordPress Database';
2148 - $rag_context_for_storage['total_documents_checked'] = $this->last_similarity_analysis['total_checked'] ?? 0;
2149 - $rag_context_for_storage['sources_used'] = $this->last_similarity_analysis['sources_used'] ?? 0;
2150 - $rag_context_for_storage['total_chunks_used'] = $this->last_similarity_analysis['total_chunks_used'] ?? 0;
2151 - }
1329 + // Get relevant content from knowledge base - THIS IS WHERE THE SIMILARITY ANALYSIS HAPPENS
1330 + $relevant_content = $this->mxchat_find_relevant_content($user_message_embedding);
1331 +
1332 + // ===== CAPTURE REAL SIMILARITY DATA FOR ADMINS =====
1333 + if ($testing_data !== null && $this->last_similarity_analysis !== null) {
1334 + // Update testing data with the REAL similarity analysis
1335 + $testing_data['top_matches'] = $this->last_similarity_analysis['top_matches'];
1336 + $testing_data['total_documents_checked'] = $this->last_similarity_analysis['total_checked'] ?? 0;
1337 + $testing_data['knowledge_base_type'] = $this->last_similarity_analysis['knowledge_base_type'];
1338 + }
1339 + // ===== END SIMILARITY DATA CAPTURE =====
1340 +
1341 + if (!empty($relevant_content)) {
1342 + $context_content .= "===== OFFICIAL KNOWLEDGE DATABASE CONTENT =====\n" . $relevant_content . "\n===== END OF OFFICIAL KNOWLEDGE DATABASE CONTENT =====\n\n";
1343 + } else {
1344 + $context_content .= "===== NO RELEVANT CONTENT FOUND IN KNOWLEDGE DATABASE =====\n";
1345 + }
2152 1346
2153 - // Add action analysis data if available
2154 - if ($has_action_data) {
2155 - $rag_context_for_storage['action_analysis'] = $this->last_action_analysis;
1347 + // Check for and include PDF content
1348 + $pdf_url = get_transient('mxchat_pdf_url_' . $session_id);
1349 + $pdf_embeddings = get_transient('mxchat_pdf_embeddings_' . $session_id);
1350 + $pdf_filename = get_transient('mxchat_pdf_filename_' . $session_id);
1351 + if ($pdf_url && $pdf_embeddings && get_transient('mxchat_include_pdf_in_context_' . $session_id)) {
1352 + $relevant_pdf_pages = $this->find_relevant_pdf_pages($user_message_embedding, $pdf_embeddings);
1353 + if (!empty($relevant_pdf_pages)) {
1354 + $context_content .= "Relevant content from PDF document '{$pdf_filename}':\n";
1355 + foreach ($relevant_pdf_pages as $page_data) {
1356 + $context_content .= "Page {$page_data['page_number']} of '{$pdf_filename}': {$page_data['text']}\n";
2156 1357 }
1358 + $context_content .= "\n";
2157 1359 }
1360 + }
2158 1361
2159 - // Save the cleaned response with RAG context
2160 - $this->mxchat_save_chat_message($session_id, 'bot', $response, null, $rag_context_for_storage);
2161 -
2162 - // Step 5: Save additional content if available
2163 - if (!empty($this->productCardHtml)) {
2164 - $this->mxchat_save_chat_message($session_id, 'bot', $this->productCardHtml);
1362 + // Check for and include Word content
1363 + $word_url = get_transient('mxchat_word_url_' . $session_id);
1364 + $word_embeddings = get_transient('mxchat_word_embeddings_' . $session_id);
1365 + $word_filename = get_transient('mxchat_word_filename_' . $session_id);
1366 + if ($word_url && $word_embeddings && get_transient('mxchat_include_word_in_context_' . $session_id)) {
1367 + $relevant_word_chunks = $this->word_handler->mxchat_find_relevant_word_chunks($user_message_embedding, $word_embeddings);
1368 + if (!empty($relevant_word_chunks)) {
1369 + $context_content .= "Relevant content from Word document '{$word_filename}':\n";
1370 + foreach ($relevant_word_chunks as $chunk_data) {
1371 + $context_content .= "Section {$chunk_data['chunk_number']} of '{$word_filename}': {$chunk_data['text']}\n";
1372 + }
1373 + $context_content .= "\n";
2165 1374 }
1375 + }
1376 +
1377 + $context_content = apply_filters('mxchat_prepare_context', $context_content, $session_id);
2166 1378
2167 - if (!empty($this->fallbackResponse['html'])) {
2168 - $this->mxchat_save_chat_message($session_id, 'bot', $this->fallbackResponse['html']);
1379 + // Generate response
1380 + $response = $this->mxchat_generate_response(
1381 + $context_content,
1382 + $this->options['api_key'],
1383 + $this->options['xai_api_key'],
1384 + $this->options['claude_api_key'],
1385 + $this->options['deepseek_api_key'],
1386 + $this->options['gemini_api_key'],
1387 + $conversation_history,
1388 + $is_streaming,
1389 + $session_id,
1390 + $testing_data
1391 + );
1392 +
1393 + // Handle streaming vs non-streaming responses
1394 + if ($is_streaming) {
1395 + // Check if streaming actually happened or if it fell back to regular response
1396 + if ($response === true) {
1397 + wp_die();
2169 1398 }
2170 -
2171 - // Step 6: Return the response
2172 - // DEBUG: Check if newlines exist in the response
2173 - //error_log("=== MXCHAT NON-STREAMING RESPONSE DEBUG ===");
2174 - //error_log("Response has newlines: " . (strpos($response, "\n") !== false ? 'YES' : 'NO'));
2175 - //error_log("Response first 500 chars: " . substr($response, 0, 500));
2176 -
2177 - $response_data = [
2178 - 'text' => $response,
2179 - 'html' => !empty($this->productCardHtml) ? $this->productCardHtml : ($this->fallbackResponse['html'] ?? ''),
2180 - 'session_id' => $session_id
2181 - ];
2182 -
2183 - // Include vectorstore error info for admin debugging (only visible to admins via testing_data)
2184 - if (!empty($this->last_vectorstore_error) && $testing_data !== null) {
2185 - $testing_data['vectorstore_error'] = $this->last_vectorstore_error;
2186 - }
2187 -
2188 - // Also pass it as a top-level field so JS can show a better error message to admins
2189 - if (!empty($this->last_vectorstore_error) && current_user_can('manage_options')) {
2190 - $response_data['vectorstore_error'] = $this->last_vectorstore_error;
2191 - }
2192 -
2193 - // Always add testing data for admins (no toggle needed)
2194 - if ($testing_data !== null) {
2195 - $response_data['testing_data'] = $testing_data;
2196 - }
2197 -
2198 - wp_send_json($response_data);
1399 + // If we get here, streaming fell back to regular response, continue
1400 + }
1401 +
1402 + // Check if the response is an error array
1403 + if (is_array($response) && isset($response['error'])) {
1404 + wp_send_json_error([
1405 + 'error_message' => $response['error'],
1406 + 'error_code' => $response['error_code'] ?? 'api_error'
1407 + ]);
2199 1408 wp_die();
2200 -}
2201 -
2202 -/**
2203 - * Get bot-specific options for multi-bot functionality
2204 - * Falls back to default options if bot_id is 'default' or multi-bot add-on is not active
2205 - */
2206 -// Also debug the bot options retrieval
2207 -private function get_bot_options($bot_id = 'default') {
2208 - //error_log("MXCHAT DEBUG: get_bot_options called for bot: " . $bot_id);
2209 -
2210 - if ($bot_id === 'default' || !class_exists('MxChat_Multi_Bot_Manager')) {
2211 - //error_log("MXCHAT DEBUG: Using default options (no multi-bot or bot is 'default')");
2212 - return array();
2213 1409 }
2214 1410
2215 - $bot_options = apply_filters('mxchat_get_bot_options', array(), $bot_id);
2216 -
2217 - if (!empty($bot_options)) {
2218 - //error_log("MXCHAT DEBUG: Got bot-specific options from filter");
2219 - if (isset($bot_options['similarity_threshold'])) {
2220 - //error_log(" - similarity_threshold: " . $bot_options['similarity_threshold']);
2221 - }
1411 + // If we get here, the response is valid text
1412 + $this->mxchat_save_chat_message($session_id, 'bot', $response);
1413 +
1414 + // Step 5: Save additional content if available
1415 + if (!empty($this->productCardHtml)) {
1416 + $this->mxchat_save_chat_message($session_id, 'bot', $this->productCardHtml);
2222 1417 }
2223 -
2224 - return is_array($bot_options) ? $bot_options : array();
2225 -}
2226 1418
2227 -/**
2228 - * Get bot-specific Pinecone configuration
2229 - * Used in the knowledge retrieval functions
2230 - */
2231 -// Also add debugging to your get_bot_pinecone_config function
2232 -private function get_bot_pinecone_config($bot_id = 'default') {
2233 - //error_log("MXCHAT DEBUG: get_bot_pinecone_config called for bot: " . $bot_id);
2234 -
2235 - // If default bot or multi-bot add-on not active, use default Pinecone config
2236 - if ($bot_id === 'default' || !class_exists('MxChat_Multi_Bot_Manager')) {
2237 - //error_log("MXCHAT DEBUG: Using default Pinecone config (no multi-bot or bot is 'default')");
2238 - $addon_options = get_option('mxchat_pinecone_addon_options', array());
2239 - $config = array(
2240 - 'use_pinecone' => (isset($addon_options['mxchat_use_pinecone']) && $addon_options['mxchat_use_pinecone'] === '1'),
2241 - 'api_key' => $addon_options['mxchat_pinecone_api_key'] ?? '',
2242 - 'host' => $addon_options['mxchat_pinecone_host'] ?? '',
2243 - 'namespace' => $addon_options['mxchat_pinecone_namespace'] ?? ''
2244 - );
2245 - //error_log("MXCHAT DEBUG: Default config - use_pinecone: " . ($config['use_pinecone'] ? 'true' : 'false'));
2246 - return $config;
1419 + if (!empty($this->fallbackResponse['html'])) {
1420 + $this->mxchat_save_chat_message($session_id, 'bot', $this->fallbackResponse['html']);
2247 1421 }
2248 -
2249 - //error_log("MXCHAT DEBUG: Calling filter 'mxchat_get_bot_pinecone_config' for bot: " . $bot_id);
2250 -
2251 - // Hook for multi-bot add-on to provide bot-specific Pinecone config
2252 - $bot_pinecone_config = apply_filters('mxchat_get_bot_pinecone_config', array(), $bot_id);
2253 -
2254 - if (!empty($bot_pinecone_config)) {
2255 - //error_log("MXCHAT DEBUG: Got bot-specific config from filter");
2256 - //error_log(" - use_pinecone: " . (isset($bot_pinecone_config['use_pinecone']) ? ($bot_pinecone_config['use_pinecone'] ? 'true' : 'false') : 'not set'));
2257 - //error_log(" - host: " . ($bot_pinecone_config['host'] ?? 'not set'));
2258 - //error_log(" - namespace: " . ($bot_pinecone_config['namespace'] ?? 'not set'));
2259 - } else {
2260 - //error_log("MXCHAT DEBUG: Filter returned empty config!");
1422 +
1423 + // Step 6: Return the response
1424 + $response_data = [
1425 + 'text' => $response,
1426 + 'html' => !empty($this->productCardHtml) ? $this->productCardHtml : ($this->fallbackResponse['html'] ?? ''),
1427 + 'session_id' => $session_id
1428 + ];
1429 +
1430 + // Always add testing data for admins (no toggle needed)
1431 + if ($testing_data !== null) {
1432 + $response_data['testing_data'] = $testing_data;
2261 1433 }
2262 -
2263 - return is_array($bot_pinecone_config) ? $bot_pinecone_config : array();
1434 +
1435 + wp_send_json($response_data);
1436 + wp_die();
2264 1437 }
2265 1438
2266 -
2267 1439 // Updated function to check intents and invoke the callback function
2268 1440 private function mxchat_check_intent_and_invoke_callback($message, $user_id, $session_id) {
2269 1441 global $wpdb;
2270 1442 $chat_mode = get_option("mxchat_mode_{$session_id}", 'ai');
2271 1443
2272 - // Get the current bot_id
2273 - $current_bot_id = $this->get_current_bot_id($session_id);
2274 -
2275 1444 // Generate the user embedding
2276 1445 $user_embedding = $this->mxchat_generate_embedding($message, $this->options['api_key']);
2277 -
1446 +
2278 1447 // Check if embedding generation returned an error
2279 1448 if (is_array($user_embedding) && isset($user_embedding['error'])) {
2280 1449 $error_message = $user_embedding['error'];
2281 1450 $error_code = $user_embedding['error_code'] ?? 'embedding_error';
2282 -
2283 - // FIXED: Send error in appropriate format based on streaming mode
2284 - if ($this->is_streaming) {
2285 - echo "data: " . json_encode([
2286 - 'error' => true,
2287 - 'error_message' => $error_message,
2288 - 'error_code' => $error_code,
2289 - 'text' => $error_message,
2290 - 'message' => $error_message
2291 - ]) . "\n\n";
2292 - echo "data: [DONE]\n\n";
2293 - flush();
2294 - } else {
2295 - wp_send_json_error([
2296 - 'error_message' => $error_message,
2297 - 'error_code' => $error_code
2298 - ]);
2299 - }
1451 +
1452 + wp_send_json_error([
1453 + 'error_message' => $error_message,
1454 + 'error_code' => $error_code
1455 + ]);
2300 1456 wp_die();
2301 1457 }
2302 -
1458 +
2303 1459 // Check if embedding is valid
2304 1460 if (!is_array($user_embedding) || empty($user_embedding)) {
2305 - $error_message = esc_html__('Unable to process your message. The embedding service is not responding correctly.', 'mxchat');
2306 -
2307 - // FIXED: Send error in appropriate format based on streaming mode
2308 - if ($this->is_streaming) {
2309 - echo "data: " . json_encode([
2310 - 'error' => true,
2311 - 'error_message' => $error_message,
2312 - 'error_code' => 'invalid_embedding',
2313 - 'text' => $error_message,
2314 - 'message' => $error_message
2315 - ]) . "\n\n";
2316 - echo "data: [DONE]\n\n";
2317 - flush();
2318 - } else {
2319 - wp_send_json_error([
2320 - 'error_message' => $error_message,
2321 - 'error_code' => 'invalid_embedding'
2322 - ]);
2323 - }
1461 + wp_send_json_error([
1462 + 'error_message' => esc_html__('Unable to process your message. The embedding service is not responding correctly.', 'mxchat'),
1463 + 'error_code' => 'invalid_embedding'
1464 + ]);
2324 1465 wp_die();
2325 1466 }
2326 -
1467 +
2327 1468 // Fetch intents from the database
2328 1469 $table_name = $wpdb->prefix . 'mxchat_intents';
2329 1470 if ($chat_mode === 'agent') {
2330 1471 $query = $wpdb->prepare(
@@ -2334,29 +1475,19 @@
2334 1475 $intents = $wpdb->get_results($query);
2335 1476 } else {
2336 1477 $intents = $wpdb->get_results("SELECT * FROM $table_name WHERE enabled = 1 OR enabled IS NULL");
2337 1478 }
2338 -
1479 +
2339 1480 if (empty($intents)) {
2340 1481 return false;
2341 1482 }
2342 -
2343 - // Prefetch individual phrase vectors from wp_mxchat_intent_phrases (grouped by intent_id)
2344 - $phrases_table = $wpdb->prefix . 'mxchat_intent_phrases';
2345 - $phrases_by_intent = [];
2346 - if ($wpdb->get_var("SHOW TABLES LIKE '$phrases_table'") === $phrases_table) {
2347 - $all_phrases = $wpdb->get_results("SELECT intent_id, phrase, embedding_vector FROM $phrases_table");
2348 - foreach ($all_phrases as $p) {
2349 - $phrases_by_intent[$p->intent_id][] = $p;
2350 - }
2351 - }
2352 -
1483 +
2353 1484 $highest_similarity = -INF;
2354 1485 $matched_intent = null;
2355 -
2356 - // Array to store action analysis for testing panel
1486 +
1487 + // NEW: Array to store action analysis for testing panel
2357 1488 $action_analysis = [];
2358 -
1489 +
2359 1490 foreach ($intents as $intent) {
2360 1491 // Additional check for enabled state
2361 1492 $is_enabled = isset($intent->enabled) ? (bool)$intent->enabled : true;
2362 1493 if (!$is_enabled) {
@@ -2361,57 +1492,22 @@
2361 1492 $is_enabled = isset($intent->enabled) ? (bool)$intent->enabled : true;
2362 1493 if (!$is_enabled) {
2363 1494 continue;
2364 1495 }
2365 -
2366 - // Check if this action is enabled for the current bot
2367 - if (!$this->is_action_enabled_for_bot($intent, $current_bot_id)) {
2368 - continue;
2369 - }
2370 -
2371 - $best_similarity = -INF;
2372 - $matched_phrase_text = '';
2373 -
2374 - // Check legacy embedding vector (existing behavior)
1496 +
2375 1497 $intent_embedding_serialized = $intent->embedding_vector;
2376 1498 $intent_embedding = $intent_embedding_serialized
2377 1499 ? unserialize($intent_embedding_serialized, ['allowed_classes' => false])
2378 1500 : null;
2379 -
2380 - if (is_array($intent_embedding) && !empty($intent_embedding)) {
2381 - $legacy_similarity = $this->mxchat_calculate_cosine_similarity($user_embedding, $intent_embedding);
2382 - if ($legacy_similarity > $best_similarity) {
2383 - $best_similarity = $legacy_similarity;
2384 - $matched_phrase_text = 'legacy';
2385 - }
2386 - }
2387 -
2388 - // Check individual phrase vectors
2389 - if (isset($phrases_by_intent[$intent->id])) {
2390 - foreach ($phrases_by_intent[$intent->id] as $phrase_row) {
2391 - $phrase_embedding = $phrase_row->embedding_vector
2392 - ? unserialize($phrase_row->embedding_vector, ['allowed_classes' => false])
2393 - : null;
2394 - if (!is_array($phrase_embedding)) {
2395 - continue;
2396 - }
2397 - $phrase_similarity = $this->mxchat_calculate_cosine_similarity($user_embedding, $phrase_embedding);
2398 - if ($phrase_similarity > $best_similarity) {
2399 - $best_similarity = $phrase_similarity;
2400 - $matched_phrase_text = $phrase_row->phrase;
2401 - }
2402 - }
2403 - }
2404 -
2405 - // Skip if no valid embedding was found at all
2406 - if ($best_similarity === -INF) {
1501 +
1502 + if (!is_array($intent_embedding)) {
2407 1503 continue;
2408 1504 }
2409 -
2410 - $similarity = $best_similarity;
1505 +
1506 + $similarity = $this->mxchat_calculate_cosine_similarity($user_embedding, $intent_embedding);
2411 1507 $intent_threshold = isset($intent->similarity_threshold) ? $intent->similarity_threshold : 0.85;
2412 -
2413 - // Store action analysis data for testing panel
1508 +
1509 + // NEW: Store action analysis data for testing panel
2414 1510 $action_analysis[] = [
2415 1511 'intent_label' => $intent->intent_label,
2416 1512 'callback_function' => $intent->callback_function,
2417 1513 'similarity' => round($similarity, 4),
@@ -2418,12 +1514,11 @@
2418 1514 'similarity_percentage' => round($similarity * 100, 2),
2419 1515 'threshold' => $intent_threshold,
2420 1516 'threshold_percentage' => round($intent_threshold * 100, 2),
2421 1517 'above_threshold' => $similarity >= $intent_threshold,
2422 - 'matched_phrase' => $matched_phrase_text,
2423 1518 'triggered' => false // Will be updated below if this intent is triggered
2424 1519 ];
2425 -
1520 +
2426 1521 if ($similarity >= $intent_threshold && $similarity > $highest_similarity) {
2427 1522 $highest_similarity = $similarity;
2428 1523 $matched_intent = $intent;
2429 1524 }
@@ -2428,9 +1523,9 @@
2428 1523 $matched_intent = $intent;
2429 1524 }
2430 1525 }
2431 1526
2432 - // Mark the triggered action if any
1527 + // NEW: Mark the triggered action if any
2433 1528 if ($matched_intent) {
2434 1529 foreach ($action_analysis as &$action) {
2435 1530 if ($action['intent_label'] === $matched_intent->intent_label) {
2436 1531 $action['triggered'] = true;
@@ -2438,9 +1533,9 @@
2438 1533 }
2439 1534 }
2440 1535 }
2441 1536
2442 - // Sort actions by similarity (highest first) and store for testing panel
1537 + // NEW: Sort actions by similarity (highest first) and store for testing panel
2443 1538 usort($action_analysis, function($a, $b) {
2444 1539 return $b['similarity'] <=> $a['similarity'];
2445 1540 });
2446 1541
@@ -2447,73 +1542,47 @@
2447 1542 // Store action analysis for testing panel capture
2448 1543 $this->last_action_analysis = $action_analysis;
2449 1544
2450 1545 // Around line 715 in your mxchat_check_intent_and_invoke_callback function
2451 - if ($matched_intent) {
2452 - // If the callback is a method on this instance (core callback), call it directly
2453 - if (method_exists($this, $matched_intent->callback_function)) {
2454 - $callback_result = call_user_func(
2455 - [$this, $matched_intent->callback_function],
2456 - $message,
2457 - $user_id,
2458 - $session_id,
2459 - $matched_intent,
2460 - $user_context ?? null
2461 - );
1546 +if ($matched_intent) {
1547 + // If the callback is a method on this instance (core callback), call it directly
1548 + if (method_exists($this, $matched_intent->callback_function)) {
1549 + $callback_result = call_user_func(
1550 + [$this, $matched_intent->callback_function],
1551 + $message,
1552 + $user_id,
1553 + $session_id,
1554 + $matched_intent,
1555 + $user_context ?? null
1556 + );
1557 + } else {
1558 + // Otherwise, use apply_filters for add-on callbacks
1559 + $callback_result = apply_filters(
1560 + $matched_intent->callback_function,
1561 + false,
1562 + $message,
1563 + $user_id,
1564 + $session_id,
1565 + $matched_intent
1566 + );
1567 + }
1568 +
1569 + // Handle the callback result properly
1570 + if ($callback_result !== false) {
1571 + // If callback returned an array with chat_mode, use it directly
1572 + if (is_array($callback_result) && isset($callback_result['chat_mode'])) {
1573 + $this->fallbackResponse = $callback_result;
1574 + return $callback_result; // Return the full array
2462 1575 } else {
2463 - // Otherwise, use apply_filters for add-on callbacks
2464 - $callback_result = apply_filters(
2465 - $matched_intent->callback_function,
2466 - false,
2467 - $message,
2468 - $user_id,
2469 - $session_id,
2470 - $matched_intent
2471 - );
1576 + $this->fallbackResponse = $callback_result;
1577 + return true;
2472 1578 }
2473 -
2474 - // Handle the callback result properly
2475 - if ($callback_result !== false) {
2476 - // If callback returned an array with chat_mode, use it directly
2477 - if (is_array($callback_result) && isset($callback_result['chat_mode'])) {
2478 - $this->fallbackResponse = $callback_result;
2479 - return $callback_result; // Return the full array
2480 - } else {
2481 - $this->fallbackResponse = $callback_result;
2482 - return true;
2483 - }
2484 - }
2485 1579 }
1580 +}
2486 1581
2487 1582 return false;
2488 1583 }
2489 1584
2490 -/**
2491 - * Check if an action is enabled for a specific bot
2492 - */
2493 -private function is_action_enabled_for_bot($intent, $bot_id) {
2494 - // If enabled_bots column doesn't exist or is null, assume it's enabled for all bots (backward compatibility)
2495 - if (!isset($intent->enabled_bots) || empty($intent->enabled_bots)) {
2496 - return true;
2497 - }
2498 -
2499 - $enabled_bots = json_decode($intent->enabled_bots, true);
2500 -
2501 - // If JSON decode fails or returns empty array, assume enabled for all (backward compatibility)
2502 - if (!is_array($enabled_bots) || empty($enabled_bots)) {
2503 - return true;
2504 - }
2505 -
2506 - // Admin testing tab uses bot_id "testing" — treat it as "default" so all
2507 - // default-bot actions are testable from the admin panel
2508 - if ($bot_id === 'testing') {
2509 - $bot_id = 'default';
2510 - }
2511 -
2512 - // Check if the current bot is in the enabled bots list
2513 - return in_array($bot_id, $enabled_bots);
2514 -}
2515 -
2516 1585 // Helper function to clear PDF and Word document related transients
2517 1586 private function clear_pdf_transients($session_id) {
2518 1587 // PDF transients
2519 1588 delete_transient('mxchat_pdf_url_' . $session_id);
@@ -2547,23 +1616,18 @@
2547 1616 }
2548 1617
2549 1618 public function mxchat_generate_image($message, $user_id, $session_id) {
2550 1619 //error_log("Starting image generation for message: " . $message);
2551 -
2552 - // Prepare a prompt for OpenAI image generation
1620 +
1621 + // Prepare a prompt for DALL-E
2553 1622 $prompt = esc_html__('Create an image of ', 'mxchat') . sanitize_text_field($message);
2554 -
2555 - // Opt-in routing: when 'custom_provider_for_images' is on, route image gen
2556 - // through the configured Custom (OpenAI-compatible) /images/generations route.
2557 - if (!empty($this->options['custom_provider_for_images']) && $this->options['custom_provider_for_images'] === 'on') {
2558 - $image_response = $this->mxchat_generate_custom_image($prompt);
2559 - } else {
2560 - // Use the existing OpenAI API key
2561 - $openai_api_key = sanitize_text_field($this->options['api_key']);
2562 - // Call OpenAI GPT Image to generate an image
2563 - $image_response = $this->mxchat_generate_openai_image($prompt, $openai_api_key);
2564 - }
2565 1623
1624 + // Use the existing OpenAI API key
1625 + $openai_api_key = sanitize_text_field($this->options['api_key']);
1626 +
1627 + // Call DALL-E to generate an image
1628 + $image_response = $this->mxchat_generate_dalle_image($prompt, $openai_api_key);
1629 +
2566 1630 // Check if the response contains an image URL
2567 1631 if (isset($image_response['imageUrl'])) {
2568 1632 $image_url = esc_url_raw($image_response['imageUrl']);
2569 1633
@@ -2606,103 +1670,24 @@
2606 1670 // Return the response directly instead of relying on the property
2607 1671 return $this->fallbackResponse;
2608 1672 }
2609 1673 }
2610 -
2611 -public function mxchat_generate_gemini_image($message, $user_id, $session_id) {
2612 - $prompt = esc_html__('Create an image of ', 'mxchat') . sanitize_text_field($message);
2613 -
2614 - $gemini_api_key = sanitize_text_field($this->options['gemini_api_key'] ?? '');
2615 - if (empty($gemini_api_key)) {
2616 - $response_text = esc_html__("Gemini API key is not configured.", 'mxchat');
2617 - $this->mxchat_save_chat_message($session_id, 'bot', $response_text);
2618 - return ['text' => $response_text, 'html' => '', 'images' => []];
2619 - }
2620 -
2621 - $image_response = $this->mxchat_generate_imagen_image($prompt, $gemini_api_key);
2622 -
2623 - if (isset($image_response['imageUrl'])) {
2624 - $image_url = esc_url_raw($image_response['imageUrl']);
2625 -
2626 - $response_html = '<img src="' . esc_url($image_url) . '" alt="' . esc_attr__('Generated Image', 'mxchat') . '" class="mxchat-generated-image" />';
2627 - $response_text = esc_html__('Here is the image I generated:', 'mxchat');
2628 -
2629 - $this->mxchat_save_chat_message($session_id, 'bot', $response_text);
2630 - $this->mxchat_save_chat_message($session_id, 'bot', $response_html);
2631 -
2632 - $this->fallbackResponse = [
2633 - 'text' => $response_text,
2634 - 'html' => $response_html,
2635 - 'images' => [$image_url]
2636 - ];
2637 -
2638 - return $this->fallbackResponse;
2639 - } else {
2640 - $response_text = esc_html__("I'm sorry, but I couldn't generate an image based on your request.", 'mxchat');
2641 -
2642 - $this->mxchat_save_chat_message($session_id, 'bot', $response_text);
2643 -
2644 - $this->fallbackResponse = [
2645 - 'text' => $response_text,
2646 - 'html' => '',
2647 - 'images' => []
2648 - ];
2649 -
2650 - return $this->fallbackResponse;
2651 - }
2652 -}
2653 -
2654 -private function mxchat_save_generated_image($base64_data, $mime_type = 'image/png', $prefix = 'mxchat-generated') {
2655 - $extension = ($mime_type === 'image/jpeg') ? 'jpg' : 'png';
2656 - $filename = sanitize_file_name($prefix . '-' . wp_generate_uuid4() . '.' . $extension);
2657 - $decoded = base64_decode($base64_data);
2658 -
2659 - if ($decoded === false) {
2660 - return new \WP_Error('decode_failed', esc_html__('Failed to decode image data.', 'mxchat'));
2661 - }
2662 -
2663 - $upload = wp_upload_bits($filename, null, $decoded);
2664 -
2665 - if (!empty($upload['error'])) {
2666 - return new \WP_Error('upload_failed', $upload['error']);
2667 - }
2668 -
2669 - $attach_id = wp_insert_attachment([
2670 - 'post_mime_type' => $mime_type,
2671 - 'post_title' => $prefix,
2672 - 'post_content' => '',
2673 - 'post_status' => 'inherit',
2674 - ], $upload['file']);
2675 -
2676 - if (is_wp_error($attach_id)) {
2677 - return $attach_id;
2678 - }
2679 -
2680 - require_once ABSPATH . 'wp-admin/includes/image.php';
2681 - $metadata = wp_generate_attachment_metadata($attach_id, $upload['file']);
2682 - wp_update_attachment_metadata($attach_id, $metadata);
2683 -
2684 - return esc_url_raw(wp_get_attachment_url($attach_id));
2685 -}
2686 -
2687 -private function mxchat_generate_openai_image($prompt, $api_key, $model = 'gpt-image-1', $timeout = 60) {
1674 +private function mxchat_generate_dalle_image($prompt, $api_key, $model = 'dall-e-3', $timeout = 60) {
2688 1675 $api_url = 'https://api.openai.com/v1/images/generations';
2689 1676 $body = json_encode([
2690 - 'prompt' => sanitize_text_field($prompt),
2691 - 'n' => 1,
2692 - 'size' => '1024x1024',
2693 - 'quality' => 'medium',
2694 - 'output_format' => 'png',
2695 - 'model' => sanitize_text_field($model),
1677 + 'prompt' => sanitize_text_field($prompt),
1678 + 'n' => 1,
1679 + 'size' => '1024x1024',
1680 + 'model' => sanitize_text_field($model),
2696 1681 ]);
2697 1682
2698 1683 $args = [
2699 - 'body' => $body,
1684 + 'body' => $body,
2700 1685 'headers' => [
2701 - 'Content-Type' => 'application/json',
1686 + 'Content-Type' => 'application/json',
2702 1687 'Authorization' => 'Bearer ' . sanitize_text_field($api_key),
2703 1688 ],
2704 - 'method' => 'POST',
1689 + 'method' => 'POST',
2705 1690 'timeout' => absint($timeout),
2706 1691 ];
2707 1692
2708 1693 $response = wp_remote_post($api_url, $args);
@@ -2707,114 +1692,23 @@
2707 1692
2708 1693 $response = wp_remote_post($api_url, $args);
2709 1694
2710 1695 if (is_wp_error($response)) {
1696 + //error_log("DALL-E request failed: " . $response->get_error_message());
2711 1697 return ['error' => esc_html__('Error generating image: ', 'mxchat') . $response->get_error_message()];
2712 1698 }
2713 1699
2714 1700 $response_body = json_decode(wp_remote_retrieve_body($response), true);
2715 1701
2716 - $b64 = $response_body['data'][0]['b64_json'] ?? $response_body['data'][0]['b64'] ?? null;
2717 - if ($b64) {
2718 - $saved_url = $this->mxchat_save_generated_image($b64, 'image/png', 'mxchat-openai');
2719 - if (is_wp_error($saved_url)) {
2720 - return ['error' => $saved_url->get_error_message()];
2721 - }
2722 - return ['imageUrl' => $saved_url];
1702 + if (isset($response_body['data'][0]['url'])) {
1703 + return ['imageUrl' => esc_url_raw($response_body['data'][0]['url'])];
2723 1704 } else {
1705 + //error_log("DALL-E response error: " . wp_remote_retrieve_body($response));
2724 1706 return ['error' => esc_html__('Failed to generate image.', 'mxchat')];
2725 1707 }
2726 1708 }
2727 1709
2728 1710 /**
2729 - * Generate an image via a Custom (OpenAI-compatible) provider's /images/generations route.
2730 - * Only called when the opt-in 'custom_provider_for_images' setting is on.
2731 - */
2732 -private function mxchat_generate_custom_image($prompt, $timeout = 90) {
2733 - $cfg = $this->mxchat_resolve_custom_provider();
2734 - if (empty($cfg['base_url'])) {
2735 - return ['error' => esc_html__('Custom provider Base URL is not configured.', 'mxchat')];
2736 - }
2737 - $url = $cfg['base_url'] . '/images/generations';
2738 - if (!empty($cfg['api_version'])) {
2739 - $url .= (strpos($url, '?') === false ? '?' : '&') . 'api-version=' . rawurlencode($cfg['api_version']);
2740 - }
2741 - $body = wp_json_encode([
2742 - 'prompt' => sanitize_text_field($prompt),
2743 - 'n' => 1,
2744 - 'size' => '1024x1024',
2745 - 'model' => $cfg['model'],
2746 - ]);
2747 - $response = wp_remote_post($url, [
2748 - 'headers' => $this->mxchat_custom_provider_assoc_headers($cfg),
2749 - 'body' => $body,
2750 - 'method' => 'POST',
2751 - 'timeout' => absint($timeout),
2752 - ]);
2753 - if (is_wp_error($response)) {
2754 - return ['error' => esc_html__('Error generating image (custom provider): ', 'mxchat') . $response->get_error_message()];
2755 - }
2756 - $resp = json_decode(wp_remote_retrieve_body($response), true);
2757 - // Try b64 first (matches OpenAI shape), then url-based fallback.
2758 - $b64 = $resp['data'][0]['b64_json'] ?? $resp['data'][0]['b64'] ?? null;
2759 - if ($b64) {
2760 - $saved = $this->mxchat_save_generated_image($b64, 'image/png', 'mxchat-custom');
2761 - if (is_wp_error($saved)) {
2762 - return ['error' => $saved->get_error_message()];
2763 - }
2764 - return ['imageUrl' => $saved];
2765 - }
2766 - $remote_url = $resp['data'][0]['url'] ?? null;
2767 - if ($remote_url) {
2768 - return ['imageUrl' => esc_url_raw($remote_url)];
2769 - }
2770 - $err_msg = $resp['error']['message'] ?? esc_html__('Custom provider did not return an image.', 'mxchat');
2771 - return ['error' => esc_html($err_msg)];
2772 -}
2773 -
2774 -private function mxchat_generate_imagen_image($prompt, $api_key, $timeout = 60) {
2775 - $api_url = 'https://generativelanguage.googleapis.com/v1beta/models/imagen-4.0-generate-001:predict';
2776 -
2777 - $body = json_encode([
2778 - 'instances' => [['prompt' => sanitize_text_field($prompt)]],
2779 - 'parameters' => [
2780 - 'sampleCount' => 1,
2781 - 'aspectRatio' => '1:1',
2782 - ],
2783 - ]);
2784 -
2785 - $args = [
2786 - 'body' => $body,
2787 - 'headers' => [
2788 - 'Content-Type' => 'application/json',
2789 - 'x-goog-api-key' => sanitize_text_field($api_key),
2790 - ],
2791 - 'method' => 'POST',
2792 - 'timeout' => absint($timeout),
2793 - ];
2794 -
2795 - $response = wp_remote_post($api_url, $args);
2796 -
2797 - if (is_wp_error($response)) {
2798 - return ['error' => esc_html__('Error generating image: ', 'mxchat') . $response->get_error_message()];
2799 - }
2800 -
2801 - $response_body = json_decode(wp_remote_retrieve_body($response), true);
2802 -
2803 - $b64 = $response_body['predictions'][0]['bytesBase64Encoded'] ?? $response_body['predictions'][0]['imageBytes'] ?? null;
2804 - if ($b64) {
2805 - $mime = $response_body['predictions'][0]['mimeType'] ?? 'image/png';
2806 - $saved_url = $this->mxchat_save_generated_image($b64, $mime, 'mxchat-gemini');
2807 - if (is_wp_error($saved_url)) {
2808 - return ['error' => $saved_url->get_error_message()];
2809 - }
2810 - return ['imageUrl' => $saved_url];
2811 - } else {
2812 - return ['error' => esc_html__('Failed to generate image.', 'mxchat')];
2813 - }
2814 -}
2815 -
2816 -/**
2817 1711 * Handle web search requests.
2818 1712 *
2819 1713 * Sends the refined search query to the Brave Search API and uses the
2820 1714 * results to generate a conversational response with the AI model.
@@ -2862,10 +1756,10 @@
2862 1756 $transient_key = 'mxchat_search_' . md5($refined_search_query);
2863 1757 $results = get_transient($transient_key);
2864 1758
2865 1759 if (false === $results) {
2866 - // SECURITY FIX: Changed to wp_safe_remote_get
2867 - $response = wp_safe_remote_get(
1760 + // Fetch new results from the Brave Search API
1761 + $response = wp_remote_get(
2868 1762 $api_url,
2869 1763 array(
2870 1764 'headers' => array(
2871 1765 'Accept' => 'application/json',
@@ -3004,10 +1898,9 @@
3004 1898 ],
3005 1899 'timeout' => 10,
3006 1900 ];
3007 1901
3008 - // SECURITY FIX: Changed to wp_safe_remote_get
3009 - $response = wp_safe_remote_get($api_url, $args);
1902 + $response = wp_remote_get($api_url, $args);
3010 1903
3011 1904 if (is_wp_error($response)) {
3012 1905 return array(
3013 1906 'text' => __("I'm sorry, I couldn't retrieve any images based on your request.", 'mxchat'),
@@ -3077,22 +1970,17 @@
3077 1970 * @return string The refined search query
3078 1971 */
3079 1972 public function mxchat_interpret_search_query($user_query) {
3080 1973 $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');
3081 -
1974 +
3082 1975 // Get options and determine the selected model
3083 1976 $options = $this->options ?? get_option('mxchat_options');
3084 - $selected_model = isset($options['model']) ? $options['model'] : 'gpt-5.1-chat-latest';
3085 -
3086 - // Custom (OpenAI-compatible) provider routes by model id, not prefix.
3087 - if ($selected_model === 'custom-provider') {
3088 - return $this->interpret_query_with_custom($user_query, $system_prompt);
3089 - }
3090 -
1977 + $selected_model = isset($options['model']) ? $options['model'] : 'gpt-4o';
1978 +
3091 1979 // Extract model prefix to determine the provider
3092 1980 $model_parts = explode('-', $selected_model);
3093 1981 $provider = strtolower($model_parts[0]);
3094 -
1982 +
3095 1983 // Determine which API key to use based on the provider
3096 1984 switch ($provider) {
3097 1985 case 'gemini':
3098 1986 $api_key = isset($options['gemini_api_key']) ? sanitize_text_field($options['gemini_api_key']) : '';
@@ -3133,60 +2021,11 @@
3133 2021 }
3134 2022 }
3135 2023
3136 2024 /**
3137 - * Interpret query against the configured Custom (OpenAI-compatible) provider.
3138 - * Uses the same base URL + auth scheme as the chat dispatcher.
3139 - */
3140 -private function interpret_query_with_custom($user_query, $system_prompt) {
3141 - $cfg = $this->mxchat_resolve_custom_provider();
3142 - if (empty($cfg['base_url'])) {
3143 - return sanitize_text_field($user_query);
3144 - }
3145 - $args = [
3146 - 'headers' => $this->mxchat_custom_provider_assoc_headers($cfg),
3147 - 'body' => wp_json_encode([
3148 - 'model' => $cfg['model'],
3149 - 'messages' => [
3150 - ['role' => 'system', 'content' => $system_prompt],
3151 - ['role' => 'user', 'content' => sanitize_text_field($user_query)],
3152 - ],
3153 - 'temperature' => 0.2,
3154 - 'max_tokens' => 20,
3155 - ]),
3156 - 'method' => 'POST',
3157 - 'timeout' => 15,
3158 - ];
3159 - $response = wp_remote_post($cfg['chat_url'], $args);
3160 - if (is_wp_error($response)) {
3161 - return sanitize_text_field($user_query);
3162 - }
3163 - $body = json_decode(wp_remote_retrieve_body($response), true);
3164 - return isset($body['choices'][0]['message']['content'])
3165 - ? sanitize_text_field(trim($body['choices'][0]['message']['content']))
3166 - : sanitize_text_field($user_query);
3167 -}
3168 -
3169 -/**
3170 - * Convert the colon-style header list returned by mxchat_resolve_custom_provider
3171 - * into the assoc-array form wp_remote_post expects.
3172 - */
3173 -private function mxchat_custom_provider_assoc_headers($cfg) {
3174 - $headers = ['Content-Type' => 'application/json'];
3175 - if (!empty($cfg['api_key'])) {
3176 - if (($cfg['auth_scheme'] ?? 'bearer') === 'api-key') {
3177 - $headers['api-key'] = $cfg['api_key'];
3178 - } else {
3179 - $headers['Authorization'] = 'Bearer ' . $cfg['api_key'];
3180 - }
3181 - }
3182 - return $headers;
3183 -}
3184 -
3185 -/**
3186 2025 * Interpret query using OpenAI models
3187 2026 */
3188 -private function interpret_query_with_openai($user_query, $system_prompt, $api_key, $model = 'gpt-5.1-chat-latest') {
2027 +private function interpret_query_with_openai($user_query, $system_prompt, $api_key, $model = 'gpt-4o') {
3189 2028 $url = 'https://api.openai.com/v1/chat/completions';
3190 2029 $args = [
3191 2030 'headers' => [
3192 2031 'Authorization' => 'Bearer ' . $api_key,
@@ -3216,34 +2055,13 @@
3216 2055 : sanitize_text_field($user_query);
3217 2056 }
3218 2057
3219 2058 /**
3220 - * Anthropic deprecated temperature/top_p/top_k starting with Opus 4.7 —
3221 - * add new reasoning-Opus model ids here. (We don't send top_p/top_k in any
3222 - * Claude body, so the list only needs to gate temperature stripping.)
3223 - */
3224 -private function mxchat_claude_omits_temperature($model) {
3225 - $no_temp = array('claude-opus-4-7', 'claude-opus-4-8');
3226 - return in_array($model, $no_temp, true);
3227 -}
3228 -
3229 -/**
3230 2059 * Interpret query using Claude models
3231 2060 */
3232 2061 private function interpret_query_with_claude($user_query, $system_prompt, $api_key, $model) {
3233 2062 $url = 'https://api.anthropic.com/v1/messages';
3234 -
3235 - $payload = [
3236 - 'model' => $model,
3237 - 'system' => $system_prompt,
3238 - 'messages' => [
3239 - ['role' => 'user', 'content' => sanitize_text_field($user_query)]
3240 - ],
3241 - 'max_tokens' => 20,
3242 - 'temperature' => 0.2,
3243 - ];
3244 - if ($this->mxchat_claude_omits_temperature($model)) { unset($payload['temperature']); }
3245 -
2063 +
3246 2064 $args = [
3247 2065 'headers' => [
3248 2066 'Content-Type' => 'application/json',
3249 2067 'x-api-key' => $api_key,
@@ -3248,9 +2066,17 @@
3248 2066 'Content-Type' => 'application/json',
3249 2067 'x-api-key' => $api_key,
3250 2068 'anthropic-version' => '2023-06-01',
3251 2069 ],
3252 - 'body' => wp_json_encode($payload),
2070 + 'body' => wp_json_encode([
2071 + 'model' => $model,
2072 + 'system' => $system_prompt,
2073 + 'messages' => [
2074 + ['role' => 'user', 'content' => sanitize_text_field($user_query)]
2075 + ],
2076 + 'max_tokens' => 20,
2077 + 'temperature' => 0.2,
2078 + ]),
3253 2079 'method' => 'POST',
3254 2080 'timeout' => 15,
3255 2081 ];
3256 2082
@@ -3270,16 +2096,13 @@
3270 2096 /**
3271 2097 * Interpret query using Gemini models
3272 2098 */
3273 2099 private function interpret_query_with_gemini($user_query, $system_prompt, $api_key, $model) {
3274 - if ($model === 'gemini-3-pro-preview') {
3275 - $model = 'gemini-3.1-pro-preview';
3276 - }
3277 - // Use v1beta for preview models, v1 for stable models
3278 - $api_version = (strpos($model, 'preview') !== false || strpos($model, 'exp') !== false) ? 'v1beta' : 'v1';
3279 -
3280 - $url = "https://generativelanguage.googleapis.com/{$api_version}/models/{$model}:generateContent?key=" . urlencode($api_key);
2100 + // Strip "gemini-" prefix for the API
2101 + $model_version = str_replace('gemini-', '', $model);
3281 2102
2103 + $url = "https://generativelanguage.googleapis.com/v1/models/$model_version:generateContent?key=" . urlencode($api_key);
2104 +
3282 2105 $args = [
3283 2106 'headers' => [
3284 2107 'Content-Type' => 'application/json',
3285 2108 ],
@@ -3475,9 +2298,9 @@
3475 2298 }
3476 2299
3477 2300
3478 2301 /**
3479 - * Enhanced fetch_and_split_pdf_pages with SSRF protection
2302 + * Enhanced fetch_and_split_pdf_pages with detailed debugging
3480 2303 */
3481 2304 private function fetch_and_split_pdf_pages($pdf_source, $max_pages) {
3482 2305 // CLEAR DEBUG LOGGING
3483 2306 //error_log("=== MXCHAT PDF PROCESSING START ===");
@@ -3532,19 +2355,10 @@
3532 2355 // (I'll include the key parts with debug logging)
3533 2356
3534 2357 if (filter_var($pdf_source, FILTER_VALIDATE_URL)) {
3535 2358 //error_log("Downloading PDF from URL...");
3536 -
3537 - // SECURITY FIX: Validate URL before processing
3538 - if (!$this->mxchat_is_safe_pdf_url($pdf_source)) {
3539 - //error_log("❌ SECURITY: Blocked unsafe PDF URL");
3540 - return false;
3541 - }
3542 -
3543 2359 $temp_file = wp_tempnam($pdf_source);
3544 -
3545 - // SECURITY FIX: Changed from wp_remote_get to wp_safe_remote_get
3546 - $response = wp_safe_remote_get($pdf_source, [
2360 + $response = wp_remote_get($pdf_source, [
3547 2361 'timeout' => 60,
3548 2362 'headers' => ['User-Agent' => 'MxChat PDF Processor']
3549 2363 ]);
3550 2364
@@ -3553,14 +2367,9 @@
3553 2367 //error_log("❌ BASIC PROCESSING: Failed to download PDF: " . $error_message);
3554 2368 return false;
3555 2369 }
3556 2370
3557 - global $wp_filesystem;
3558 - if (empty($wp_filesystem)) {
3559 - require_once ABSPATH . 'wp-admin/includes/file.php';
3560 - WP_Filesystem();
3561 - }
3562 - $wp_filesystem->put_contents($temp_file, wp_remote_retrieve_body($response), FS_CHMOD_FILE);
2371 + file_put_contents($temp_file, wp_remote_retrieve_body($response));
3563 2372 //error_log("✅ PDF downloaded successfully");
3564 2373 } else {
3565 2374 $temp_file = $pdf_source;
3566 2375 //error_log("Using local PDF file: " . $temp_file);
@@ -3567,9 +2376,8 @@
3567 2376 }
3568 2377
3569 2378 // Parse PDF
3570 2379 //error_log("Parsing PDF with basic parser...");
3571 - mxchat_load_pdf_parser();
3572 2380 $parser = new \Smalot\PdfParser\Parser();
3573 2381 $pdf = $parser->parseFile($temp_file);
3574 2382 $pages = $pdf->getPages();
3575 2383
@@ -3632,33 +2440,8 @@
3632 2440 return false;
3633 2441 }
3634 2442 }
3635 2443
3636 -
3637 -/**
3638 - * Validate PDF URL for security
3639 - * Prevents SSRF attacks by blocking dangerous URLs
3640 - */
3641 -
3642 -private function mxchat_is_safe_pdf_url($url) {
3643 - // Use WordPress core function for comprehensive validation
3644 - // This blocks localhost, private IPs, and reserved IP ranges
3645 - $validated_url = wp_http_validate_url($url);
3646 -
3647 - if ($validated_url === false) {
3648 - return false;
3649 - }
3650 -
3651 - // Additional check: only allow HTTP/HTTPS schemes
3652 - $parsed = parse_url($url);
3653 - if (!isset($parsed['scheme']) || !in_array($parsed['scheme'], ['http', 'https'], true)) {
3654 - return false;
3655 - }
3656 -
3657 - return true;
3658 -}
3659 -
3660 -
3661 2444 private function mxchat_clean_text($text) {
3662 2445 // Remove excessive whitespace
3663 2446 $text = preg_replace('/\s+/', ' ', $text);
3664 2447
@@ -3697,14 +2480,11 @@
3697 2480 }
3698 2481
3699 2482 return [];
3700 2483 }
3701 -
3702 -
2484 +// Add this to your class
3703 2485 public function handle_pdf_upload() {
3704 - if (!isset($_POST['nonce']) || !MxChat_Integrator::mxchat_verify_chat_send_nonce(wp_unslash((string) $_POST['nonce']))) {
3705 - wp_send_json_error(array('message' => esc_html__('Invalid nonce.', 'mxchat')), 403);
3706 - }
2486 + check_ajax_referer('mxchat_chat_nonce', 'nonce');
3707 2487
3708 2488 if (!isset($_FILES['pdf_file']) || !isset($_POST['session_id'])) {
3709 2489 wp_send_json_error(esc_html__('Missing required parameters.', 'mxchat'));
3710 2490 return;
@@ -3709,29 +2489,12 @@
3709 2489 wp_send_json_error(esc_html__('Missing required parameters.', 'mxchat'));
3710 2490 return;
3711 2491 }
3712 2492
3713 - // SECURITY FIX: Check if PDF uploads are enabled in settings
3714 - $options = get_option('mxchat_options', array());
3715 - $show_pdf_button = isset($options['show_pdf_upload_button']) ? $options['show_pdf_upload_button'] : 'on';
3716 -
3717 - if ($show_pdf_button !== 'on') {
3718 - wp_send_json_error(esc_html__('PDF uploads are currently disabled.', 'mxchat'));
3719 - return;
3720 - }
3721 -
3722 2493 $file = $_FILES['pdf_file'];
3723 2494 $session_id = sanitize_text_field($_POST['session_id']);
3724 2495 $original_filename = sanitize_text_field($file['name']);
3725 2496
3726 - // Update session owner if it changed (e.g. IP changed due to network switch)
3727 - $current_user_identifier = MxChat_User::mxchat_get_user_identifier();
3728 - $session_owner = get_option("mxchat_session_owner_{$session_id}");
3729 -
3730 - if (!$session_owner || $session_owner !== $current_user_identifier) {
3731 - update_option("mxchat_session_owner_{$session_id}", $current_user_identifier, 'no');
3732 - }
3733 -
3734 2497 $file_type = wp_check_filetype($file['name'], ['pdf' => 'application/pdf']);
3735 2498 if ($file_type['type'] !== 'application/pdf') {
3736 2499 wp_send_json_error(esc_html__('Invalid file type. Only PDF files are allowed.', 'mxchat'));
3737 2500 return;
@@ -3737,12 +2500,9 @@
3737 2500 return;
3738 2501 }
3739 2502
3740 2503 $upload_dir = wp_upload_dir();
3741 -
3742 - // SECURITY FIX: Generate random filename without exposing session_id
3743 - $random_string = wp_generate_password(20, false, false); // 20 char alphanumeric string
3744 - $pdf_filename = 'mxchat_' . $random_string . '_' . time() . '.pdf';
2504 + $pdf_filename = 'mxchat_' . $session_id . '_' . time() . '.pdf';
3745 2505 $pdf_path = $upload_dir['path'] . '/' . $pdf_filename;
3746 2506
3747 2507 if (!move_uploaded_file($file['tmp_name'], $pdf_path)) {
3748 2508 wp_send_json_error(esc_html__('Failed to upload file.', 'mxchat'));
@@ -3773,9 +2533,8 @@
3773 2533 return;
3774 2534 }
3775 2535
3776 2536 if (!empty($embeddings)) {
3777 - // Store the mapping between session and the random filename
3778 2537 set_transient('mxchat_pdf_url_' . $session_id, $pdf_path, HOUR_IN_SECONDS);
3779 2538 set_transient('mxchat_pdf_filename_' . $session_id, $original_filename, HOUR_IN_SECONDS);
3780 2539 set_transient('mxchat_pdf_embeddings_' . $session_id, $embeddings, HOUR_IN_SECONDS);
3781 2540 set_transient('mxchat_include_pdf_in_context_' . $session_id, true, HOUR_IN_SECONDS);
@@ -3796,11 +2555,9 @@
3796 2555 wp_send_json_error($error_message);
3797 2556 return;
3798 2557 }
3799 2558 public function handle_pdf_remove() {
3800 - if (!isset($_POST['nonce']) || !MxChat_Integrator::mxchat_verify_chat_send_nonce(wp_unslash((string) $_POST['nonce']))) {
3801 - wp_send_json_error(array('message' => esc_html__('Invalid nonce.', 'mxchat')), 403);
3802 - }
2559 + check_ajax_referer('mxchat_chat_nonce', 'nonce');
3803 2560
3804 2561 if (empty($_POST['session_id'])) {
3805 2562 wp_send_json_error(esc_html__('Session ID missing.', 'mxchat'));
3806 2563 wp_die();
@@ -3821,8 +2578,10 @@
3821 2578 wp_die();
3822 2579 }
3823 2580
3824 2581
2582 +
2583 +
3825 2584 function mxchat_fetch_new_messages() {
3826 2585 $session_id = sanitize_text_field($_POST['session_id']);
3827 2586 $last_seen_id = sanitize_text_field($_POST['last_seen_id']);
3828 2587 $persistence_enabled = $_POST['persistence_enabled'] === 'true';
@@ -3835,31 +2594,14 @@
3835 2594 }
3836 2595
3837 2596 $history = get_option("mxchat_history_{$session_id}", []);
3838 2597
3839 - //error_log("MxChat WhatsApp DEBUG: Fetch new messages for session {$session_id}");
3840 - //error_log("MxChat WhatsApp DEBUG: last_seen_id = " . var_export($last_seen_id, true));
3841 - //error_log("MxChat WhatsApp DEBUG: History count = " . count($history));
3842 - //error_log("MxChat WhatsApp DEBUG: Full history = " . print_r($history, true));
3843 -
3844 2598 $new_messages = array_filter($history, function ($message) use ($last_seen_id, $persistence_enabled, $initial_timestamp) {
3845 - //error_log("MxChat WhatsApp DEBUG: Checking message - ID: " . ($message['id'] ?? 'NO_ID') . ", Role: " . ($message['role'] ?? 'NO_ROLE'));
3846 -
3847 2599 // If persistence is enabled, show all new messages
3848 2600 if ($persistence_enabled) {
3849 - $has_id = !empty($message['id']);
3850 - $is_agent = $message['role'] === 'agent';
3851 -
3852 - // If last_seen_id is empty, 'NaN', or invalid, show all agent messages
3853 - if (empty($last_seen_id) || $last_seen_id === 'NaN' || $last_seen_id === 'undefined') {
3854 - $is_newer = true;
3855 - } else {
3856 - $is_newer = strcmp($message['id'] ?? '', $last_seen_id) > 0;
3857 - }
3858 -
3859 - //error_log("MxChat WhatsApp DEBUG: has_id={$has_id}, is_newer={$is_newer}, is_agent={$is_agent}");
3860 -
3861 - return $has_id && $is_newer && $is_agent;
2601 + return !empty($message['id']) &&
2602 + strcmp($message['id'], $last_seen_id) > 0 &&
2603 + $message['role'] === 'agent';
3862 2604 }
3863 2605
3864 2606 // If persistence is disabled, only show messages after initial timestamp
3865 2607 return !empty($message['id']) &&
@@ -3866,16 +2608,12 @@
3866 2608 $message['role'] === 'agent' &&
3867 2609 $message['timestamp'] > $initial_timestamp;
3868 2610 });
3869 2611
3870 - //error_log("MxChat WhatsApp DEBUG: Filtered messages count = " . count($new_messages));
2612 + //error_log(esc_html__("New agent messages fetched for session $session_id. Last seen ID: $last_seen_id", 'mxchat'));
3871 2613
3872 - // Include current chat mode so frontend can detect agent→AI transitions
3873 - $chat_mode = get_option("mxchat_mode_{$session_id}", 'ai');
3874 -
3875 2614 wp_send_json_success([
3876 - 'new_messages' => array_values($new_messages),
3877 - 'chat_mode' => $chat_mode
2615 + 'new_messages' => array_values($new_messages)
3878 2616 ]);
3879 2617 wp_die();
3880 2618 }
3881 2619 public function mxchat_live_agent_handover($message, $user_id, $session_id) {
@@ -4160,393 +2898,9 @@
4160 2898
4161 2899 //error_log("[DEBUG] Generated channel name: {$channel_name}");
4162 2900 return $channel_name;
4163 2901 }
4164 -
4165 -/**
4166 - * Telegram Live Agent Handover
4167 - * Creates a forum topic in the Telegram group and notifies agents
4168 - */
4169 -public function mxchat_telegram_live_agent_handover($message, $user_id, $session_id) {
4170 - // Check if Telegram agents are available
4171 - $telegram_available = $this->options['telegram_status'] ?? 'off';
4172 - if ($telegram_available !== 'on') {
4173 - $away_message = $this->options['telegram_away_message'] ?? 'Sorry, live agents are currently unavailable. I can continue helping you as an AI assistant.';
4174 - $this->fallbackResponse = [
4175 - 'text' => $away_message,
4176 - 'html' => '',
4177 - 'images' => [],
4178 - 'chat_mode' => 'ai'
4179 - ];
4180 - wp_send_json([
4181 - 'text' => $away_message,
4182 - 'html' => '',
4183 - 'chat_mode' => 'ai',
4184 - 'session_id' => $session_id
4185 - ]);
4186 - wp_die();
4187 - }
4188 -
4189 - $telegram_bot_token = $this->options['telegram_bot_token'] ?? '';
4190 - $telegram_group_id = $this->options['telegram_group_id'] ?? '';
4191 -
4192 - if (empty($telegram_bot_token) || empty($telegram_group_id)) {
4193 - return false;
4194 - }
4195 -
4196 - // Check if topic already exists for this session
4197 - $topic_id = get_option("mxchat_telegram_topic_{$session_id}", '');
4198 -
4199 - if (empty($topic_id)) {
4200 - // Generate topic name
4201 - $topic_name = $this->generate_telegram_topic_name($session_id);
4202 -
4203 - // Random icon color (Telegram forum topic colors)
4204 - $icon_colors = [0x6FB9F0, 0xFFD67E, 0xCB86DB, 0x8EEE98, 0xFF93B2, 0xFB6F5F];
4205 - $icon_color = $icon_colors[array_rand($icon_colors)];
4206 -
4207 - // Create forum topic
4208 - $response = wp_remote_post("https://api.telegram.org/bot{$telegram_bot_token}/createForumTopic", [
4209 - 'headers' => ['Content-Type' => 'application/json'],
4210 - 'body' => json_encode([
4211 - 'chat_id' => $telegram_group_id,
4212 - 'name' => $topic_name,
4213 - 'icon_color' => $icon_color
4214 - ])
4215 - ]);
4216 -
4217 - if (!is_wp_error($response)) {
4218 - $response_body = wp_remote_retrieve_body($response);
4219 - $response_data = json_decode($response_body, true);
4220 -
4221 - if (isset($response_data['ok']) && $response_data['ok']) {
4222 - $topic_id = $response_data['result']['message_thread_id'];
4223 - update_option("mxchat_telegram_topic_{$session_id}", $topic_id);
4224 - update_option("mxchat_telegram_group_{$session_id}", $telegram_group_id);
4225 - }
4226 - }
4227 -
4228 - if (empty($topic_id)) {
4229 - return false; // Failed to create topic
4230 - }
4231 - }
4232 -
4233 - // Get recent chat history
4234 - $history = get_option("mxchat_history_{$session_id}", []);
4235 - $recent_history = array_slice($history, -5);
4236 -
4237 - // Format conversation context for Telegram (HTML format)
4238 - $conversation_context = "";
4239 - if (!empty($recent_history)) {
4240 - $conversation_context = "<b>Recent Conversation:</b>\n";
4241 - foreach ($recent_history as $hist_message) {
4242 - $role_display = $hist_message['role'] === 'user' ? '👤 User' : '🤖 AI';
4243 - $escaped_content = htmlspecialchars($hist_message['content'], ENT_QUOTES, 'UTF-8');
4244 - $conversation_context .= "{$role_display}: {$escaped_content}\n";
4245 - }
4246 - $conversation_context .= "\n";
4247 - }
4248 -
4249 - // Get user info
4250 - $user_email = get_option("mxchat_email_{$session_id}", 'Not provided');
4251 - $user_name = get_option("mxchat_name_{$session_id}", 'Anonymous');
4252 -
4253 - // Update session mode
4254 - update_option("mxchat_mode_{$session_id}", 'agent');
4255 -
4256 - // Send initial message to topic
4257 - $escaped_message = htmlspecialchars($message, ENT_QUOTES, 'UTF-8');
4258 - $topic_message = "🔔 <b>New Live Agent Request</b>\n\n";
4259 - $topic_message .= "<b>Session ID:</b> <code>{$session_id}</code>\n";
4260 - $topic_message .= "<b>User:</b> {$user_name}\n";
4261 - $topic_message .= "<b>Email:</b> {$user_email}\n\n";
4262 -
4263 - if (!empty($conversation_context)) {
4264 - $topic_message .= $conversation_context;
4265 - }
4266 -
4267 - $topic_message .= "<b>Current Message:</b>\n{$escaped_message}\n\n";
4268 - $topic_message .= "<i>Reply in this topic - messages will be sent to the user</i>\n";
4269 - $topic_message .= "<i>Type #close, #end, #disconnect, or #done to end the session</i>";
4270 -
4271 - wp_remote_post("https://api.telegram.org/bot{$telegram_bot_token}/sendMessage", [
4272 - 'headers' => ['Content-Type' => 'application/json'],
4273 - 'body' => json_encode([
4274 - 'chat_id' => $telegram_group_id,
4275 - 'message_thread_id' => $topic_id,
4276 - 'text' => $topic_message,
4277 - 'parse_mode' => 'HTML'
4278 - ])
4279 - ]);
4280 -
4281 - $success_message = $this->options['telegram_notification_message'] ?? "I've notified a support agent. Please allow a moment for them to respond.";
4282 - $this->mxchat_save_chat_message($session_id, 'bot', $success_message);
4283 -
4284 - $this->fallbackResponse = [
4285 - 'text' => $success_message,
4286 - 'html' => '',
4287 - 'images' => [],
4288 - 'chat_mode' => 'agent'
4289 - ];
4290 -
4291 - wp_send_json([
4292 - 'success' => true,
4293 - 'text' => $success_message,
4294 - 'html' => '',
4295 - 'chat_mode' => 'agent',
4296 - 'session_id' => $session_id,
4297 - 'fallbackResponse' => $this->fallbackResponse
4298 - ]);
4299 - wp_die();
4300 -}
4301 -
4302 -/**
4303 - * Generate topic name for Telegram forum
4304 - */
4305 -private function generate_telegram_topic_name($session_id) {
4306 - $name = null;
4307 - $email = null;
4308 -
4309 - // Check logged in user
4310 - if (is_user_logged_in()) {
4311 - $current_user = wp_get_current_user();
4312 - if (!empty($current_user->display_name)) {
4313 - $name = $current_user->display_name;
4314 - }
4315 - if (!empty($current_user->user_email)) {
4316 - $email = $current_user->user_email;
4317 - }
4318 - }
4319 -
4320 - // Check session data
4321 - if (empty($name)) {
4322 - $name = get_option("mxchat_name_{$session_id}");
4323 - }
4324 - if (empty($email)) {
4325 - $email = get_option("mxchat_email_{$session_id}");
4326 - }
4327 -
4328 - // Generate topic name
4329 - $session_suffix = substr($session_id, -6);
4330 -
4331 - if (!empty($name)) {
4332 - // Clean name for topic (max 128 chars in Telegram)
4333 - $clean_name = preg_replace('/[^\p{L}\p{N}\s\-]/u', '', $name);
4334 - $clean_name = trim($clean_name);
4335 - if (strlen($clean_name) > 50) {
4336 - $clean_name = substr($clean_name, 0, 50);
4337 - }
4338 - return "Chat - {$clean_name} ({$session_suffix})";
4339 - } elseif (!empty($email)) {
4340 - // Use email prefix
4341 - $email_prefix = explode('@', $email)[0];
4342 - if (strlen($email_prefix) > 30) {
4343 - $email_prefix = substr($email_prefix, 0, 30);
4344 - }
4345 - return "Chat - {$email_prefix} ({$session_suffix})";
4346 - }
4347 -
4348 - return "Chat - {$session_suffix}";
4349 -}
4350 -
4351 -/**
4352 - * Send user message to Telegram agent
4353 - */
4354 -public function mxchat_send_user_message_to_telegram_agent($message, $user_id, $session_id) {
4355 - $telegram_bot_token = $this->options['telegram_bot_token'] ?? '';
4356 - $topic_id = get_option("mxchat_telegram_topic_{$session_id}", '');
4357 - $group_id = get_option("mxchat_telegram_group_{$session_id}", '');
4358 -
4359 - if (empty($telegram_bot_token) || empty($topic_id) || empty($group_id)) {
4360 - return false;
4361 - }
4362 -
4363 - $escaped_message = htmlspecialchars($message, ENT_QUOTES, 'UTF-8');
4364 - $user_message = "👤 <b>User:</b> {$escaped_message}";
4365 -
4366 - $response = wp_remote_post("https://api.telegram.org/bot{$telegram_bot_token}/sendMessage", [
4367 - 'headers' => ['Content-Type' => 'application/json'],
4368 - 'body' => json_encode([
4369 - 'chat_id' => $group_id,
4370 - 'message_thread_id' => $topic_id,
4371 - 'text' => $user_message,
4372 - 'parse_mode' => 'HTML'
4373 - ])
4374 - ]);
4375 -
4376 - return !is_wp_error($response);
4377 -}
4378 -
4379 -/**
4380 - * Handle incoming Telegram webhook
4381 - */
4382 -public function handle_telegram_webhook(WP_REST_Request $request) {
4383 - $body = $request->get_body();
4384 - $data = json_decode($body, true);
4385 -
4386 - //error_log('[MxChat Telegram DEBUG] Webhook received: ' . $body);
4387 -
4388 - // Handle message events from forum topics
4389 - if (isset($data['message'])) {
4390 - $message_data = $data['message'];
4391 -
4392 - // Skip if not from a forum topic
4393 - if (!isset($message_data['message_thread_id'])) {
4394 - //error_log('[MxChat Telegram DEBUG] Skipped: No message_thread_id (not a forum topic message)');
4395 - return new WP_REST_Response(['ok' => true]);
4396 - }
4397 -
4398 - // Skip bot messages
4399 - if (isset($message_data['from']['is_bot']) && $message_data['from']['is_bot']) {
4400 - //error_log('[MxChat Telegram DEBUG] Skipped: Message from bot');
4401 - return new WP_REST_Response(['ok' => true]);
4402 - }
4403 -
4404 - $chat_id = $message_data['chat']['id'] ?? '';
4405 - $topic_id = $message_data['message_thread_id'];
4406 - $message_text = $message_data['text'] ?? '';
4407 - $message_id = $message_data['message_id'] ?? '';
4408 - $from = $message_data['from'] ?? [];
4409 - $agent_name = trim(($from['first_name'] ?? '') . ' ' . ($from['last_name'] ?? ''));
4410 - if (empty($agent_name)) {
4411 - $agent_name = $from['username'] ?? 'Agent';
4412 - }
4413 -
4414 - //error_log("[MxChat Telegram DEBUG] Parsed: chat_id={$chat_id}, topic_id={$topic_id}, agent={$agent_name}, text={$message_text}");
4415 -
4416 - // Skip empty messages
4417 - if (empty($message_text)) {
4418 - //error_log('[MxChat Telegram DEBUG] Skipped: Empty message text');
4419 - return new WP_REST_Response(['ok' => true]);
4420 - }
4421 -
4422 - // Find session ID by topic ID - cast to string for comparison
4423 - global $wpdb;
4424 - $topic_id_str = strval($topic_id);
4425 - $session_option = $wpdb->get_var(
4426 - $wpdb->prepare(
4427 - "SELECT option_name FROM {$wpdb->options}
4428 - WHERE option_name LIKE %s
4429 - AND option_value = %s",
4430 - 'mxchat_telegram_topic_%',
4431 - $topic_id_str
4432 - )
4433 - );
4434 -
4435 - //error_log("[MxChat Telegram DEBUG] Looking for topic_id={$topic_id_str} in options, found: " . ($session_option ?: 'NULL'));
4436 -
4437 - if ($session_option) {
4438 - $session_id = str_replace('mxchat_telegram_topic_', '', $session_option);
4439 - //error_log("[MxChat Telegram DEBUG] Session ID: {$session_id}");
4440 -
4441 - // Verify the group ID matches
4442 - $stored_group_id = get_option("mxchat_telegram_group_{$session_id}", '');
4443 - //error_log("[MxChat Telegram DEBUG] Stored group_id={$stored_group_id}, received chat_id={$chat_id}");
4444 -
4445 - if (strval($stored_group_id) != strval($chat_id)) {
4446 - //error_log('[MxChat Telegram DEBUG] Skipped: Group ID mismatch');
4447 - return new WP_REST_Response(['ok' => true]);
4448 - }
4449 -
4450 - // Check for closure commands
4451 - $lower_text = strtolower(trim($message_text));
4452 - if (in_array($lower_text, ['#close', '#end', '#disconnect', '#done'])) {
4453 - //error_log("[MxChat Telegram DEBUG] Closure command received: {$lower_text}");
4454 - // End the live agent session
4455 - update_option("mxchat_mode_{$session_id}", 'ai');
4456 -
4457 - // Save disconnect message
4458 - $disconnect_message = "Live agent session ended. You're now chatting with the AI assistant.";
4459 - $this->mxchat_save_chat_message($session_id, 'bot', $disconnect_message);
4460 -
4461 - // Notify in Telegram
4462 - $telegram_bot_token = $this->options['telegram_bot_token'] ?? '';
4463 - if (!empty($telegram_bot_token)) {
4464 - wp_remote_post("https://api.telegram.org/bot{$telegram_bot_token}/sendMessage", [
4465 - 'headers' => ['Content-Type' => 'application/json'],
4466 - 'body' => json_encode([
4467 - 'chat_id' => $chat_id,
4468 - 'message_thread_id' => $topic_id,
4469 - 'text' => "✅ Session closed. User returned to AI chatbot.",
4470 - 'parse_mode' => 'HTML'
4471 - ])
4472 - ]);
4473 -
4474 - // Optionally close the topic
4475 - wp_remote_post("https://api.telegram.org/bot{$telegram_bot_token}/closeForumTopic", [
4476 - 'headers' => ['Content-Type' => 'application/json'],
4477 - 'body' => json_encode([
4478 - 'chat_id' => $chat_id,
4479 - 'message_thread_id' => $topic_id
4480 - ])
4481 - ]);
4482 - }
4483 -
4484 - return new WP_REST_Response(['ok' => true]);
4485 - }
4486 -
4487 - // Deduplicate messages
4488 - $message_key = md5($session_id . $message_id . $message_text);
4489 - $processed_messages = get_transient('mxchat_telegram_messages_' . $session_id) ?: [];
4490 -
4491 - if (in_array($message_key, $processed_messages)) {
4492 - //error_log('[MxChat Telegram DEBUG] Skipped: Duplicate message');
4493 - return new WP_REST_Response(['ok' => true]);
4494 - }
4495 -
4496 - $processed_messages[] = $message_key;
4497 - if (count($processed_messages) > 50) {
4498 - $processed_messages = array_slice($processed_messages, -50);
4499 - }
4500 - set_transient('mxchat_telegram_messages_' . $session_id, $processed_messages, HOUR_IN_SECONDS);
4501 -
4502 - // Save the agent message - format with agent name prefix for proper parsing
4503 - $formatted_message = "Agent: {$agent_name} - {$message_text}";
4504 - //error_log("[MxChat Telegram DEBUG] Saving agent message: {$formatted_message}");
4505 -
4506 - $this->mxchat_save_chat_message($session_id, 'agent', $formatted_message);
4507 -
4508 - // Verify the message was saved to history
4509 - $history = get_option("mxchat_history_{$session_id}", []);
4510 - $last_message = end($history);
4511 - //error_log("[MxChat Telegram DEBUG] History after save - count: " . count($history) . ", last message role: " . ($last_message['role'] ?? 'none'));
4512 -
4513 - // Send confirmation back to Telegram
4514 - $telegram_bot_token = $this->options['telegram_bot_token'] ?? '';
4515 - if (!empty($telegram_bot_token)) {
4516 - $confirm_key = 'mxchat_telegram_confirm_' . $message_key;
4517 - if (!get_transient($confirm_key)) {
4518 - wp_remote_post("https://api.telegram.org/bot{$telegram_bot_token}/sendMessage", [
4519 - 'headers' => ['Content-Type' => 'application/json'],
4520 - 'body' => json_encode([
4521 - 'chat_id' => $chat_id,
4522 - 'message_thread_id' => $topic_id,
4523 - 'text' => "✅ <i>Message sent to user</i>",
4524 - 'parse_mode' => 'HTML',
4525 - 'reply_to_message_id' => $message_id
4526 - ])
4527 - ]);
4528 - set_transient($confirm_key, true, 300);
4529 - }
4530 - }
4531 - } else {
4532 - //error_log("[MxChat Telegram DEBUG] No session found for topic_id={$topic_id}");
4533 - }
4534 - } else {
4535 - //error_log('[MxChat Telegram DEBUG] No message in webhook data');
4536 - }
4537 -
4538 - return new WP_REST_Response(['ok' => true]);
4539 -}
4540 -
4541 2902 public function mxchat_send_user_message_to_agent($message, $user_id, $session_id) {
4542 - // Check if this is a Telegram agent session
4543 - $telegram_topic_id = get_option("mxchat_telegram_topic_{$session_id}", '');
4544 - if (!empty($telegram_topic_id)) {
4545 - return $this->mxchat_send_user_message_to_telegram_agent($message, $user_id, $session_id);
4546 - }
4547 -
4548 - // Otherwise, try Slack
4549 2903 $slack_bot_token = $this->options['live_agent_bot_token'] ?? '';
4550 2904 $channel_id = get_option("mxchat_channel_{$session_id}", '');
4551 2905
4552 2906 if (empty($slack_bot_token) || empty($channel_id)) {
@@ -4772,33 +3126,33 @@
4772 3126
4773 3127 $channel_id = $event['channel'];
4774 3128 $message_text = $event['text'] ?? '';
4775 3129 $message_ts = $event['ts'] ?? '';
4776 -
3130 +
4777 3131 // Find session ID by looking for matching channel
4778 3132 global $wpdb;
4779 3133 $session_option = $wpdb->get_var(
4780 3134 $wpdb->prepare(
4781 - "SELECT option_name FROM {$wpdb->options}
4782 - WHERE option_name LIKE 'mxchat_channel_%'
3135 + "SELECT option_name FROM {$wpdb->options}
3136 + WHERE option_name LIKE 'mxchat_channel_%'
4783 3137 AND option_value = %s",
4784 3138 $channel_id
4785 3139 )
4786 3140 );
4787 -
3141 +
4788 3142 if ($session_option) {
4789 3143 $session_id = str_replace('mxchat_channel_', '', $session_option);
4790 -
3144 +
4791 3145 // Create a unique key for this specific message
4792 3146 $message_key = md5($session_id . $message_ts . $message_text);
4793 3147 $processed_messages = get_transient('mxchat_processed_messages_' . $session_id) ?: [];
4794 -
3148 +
4795 3149 // Check if we've already processed this exact message
4796 3150 if (in_array($message_key, $processed_messages)) {
4797 3151 //error_log("Duplicate message detected for session $session_id");
4798 3152 return new WP_REST_Response(['ok' => true]);
4799 3153 }
4800 -
3154 +
4801 3155 // Add to processed messages
4802 3156 $processed_messages[] = $message_key;
4803 3157 // Keep only last 50 messages per session
4804 3158 if (count($processed_messages) > 50) {
@@ -4804,46 +3158,14 @@
4804 3158 if (count($processed_messages) > 50) {
4805 3159 $processed_messages = array_slice($processed_messages, -50);
4806 3160 }
4807 3161 set_transient('mxchat_processed_messages_' . $session_id, $processed_messages, HOUR_IN_SECONDS);
4808 -
4809 - $slack_bot_token = $this->options['live_agent_bot_token'] ?? '';
4810 -
4811 - // Handle agent ending the chat — transfer back to AI
4812 - // Format: "!endchat" or "!endchat <custom message to user>"
4813 - if (preg_match('/^!endchat\b/i', trim($message_text))) {
4814 - update_option("mxchat_mode_{$session_id}", 'ai');
4815 -
4816 - // Extract custom message after !endchat, or use empty string
4817 - $custom_message = trim(preg_replace('/^!endchat\s*/i', '', trim($message_text)));
4818 -
4819 - // Send the agent's custom farewell message if provided
4820 - if (!empty($custom_message)) {
4821 - $this->mxchat_save_chat_message($session_id, 'agent', $custom_message);
4822 - }
4823 -
4824 - // Confirm in Slack channel
4825 - if (!empty($slack_bot_token)) {
4826 - wp_remote_post('https://slack.com/api/chat.postMessage', [
4827 - 'headers' => [
4828 - 'Content-Type' => 'application/json',
4829 - 'Authorization' => 'Bearer ' . $slack_bot_token
4830 - ],
4831 - 'body' => json_encode([
4832 - 'channel' => $channel_id,
4833 - 'text' => "✅ *Chat ended.* User has been transferred back to AI mode.",
4834 - 'mrkdwn' => true
4835 - ])
4836 - ]);
4837 - }
4838 -
4839 - return new WP_REST_Response(['ok' => true]);
4840 - }
4841 -
3162 +
4842 3163 // Save the agent message
4843 3164 $this->mxchat_save_chat_message($session_id, 'agent', $message_text);
4844 -
3165 +
4845 3166 // Send confirmation back to Slack (only once)
3167 + $slack_bot_token = $this->options['live_agent_bot_token'] ?? '';
4846 3168 if (!empty($slack_bot_token)) {
4847 3169 // Use a transient to prevent duplicate confirmations
4848 3170 $confirm_key = 'mxchat_confirm_' . $message_key;
4849 3171 if (!get_transient($confirm_key)) {
@@ -4853,9 +3175,9 @@
4853 3175 'Authorization' => 'Bearer ' . $slack_bot_token
4854 3176 ],
4855 3177 'body' => json_encode([
4856 3178 'channel' => $channel_id,
4857 - 'text' => "✅ _Message sent to user_",
3179 + 'text' => "✅ _Message sent to user_",
4858 3180 'thread_ts' => $event['ts'] // Reply in thread
4859 3181 ])
4860 3182 ]);
4861 3183 // Set transient to prevent duplicate confirmations
@@ -4895,15 +3217,9 @@
4895 3217 try {
4896 3218 // Get options and selected model
4897 3219 $options = get_option('mxchat_options');
4898 3220 $selected_model = $options['embedding_model'] ?? 'text-embedding-ada-002';
4899 -
4900 - // Opt-in: route embeddings through the Custom (OpenAI-compatible) provider.
4901 - // Off by default so existing sites see byte-identical behavior.
4902 - if (!empty($options['custom_provider_for_embeddings']) && $options['custom_provider_for_embeddings'] === 'on') {
4903 - return $this->mxchat_generate_embedding_custom($text);
4904 - }
4905 -
3221 +
4906 3222 // Determine endpoint and API key based on model
4907 3223 if (strpos($selected_model, 'voyage') === 0) {
4908 3224 $endpoint = 'https://api.voyageai.com/v1/embeddings';
4909 3225 $api_key = $options['voyage_api_key'] ?? '';
@@ -5099,617 +3415,251 @@
5099 3415 ];
5100 3416 }
5101 3417 }
5102 3418
3419 +private function mxchat_find_relevant_content($user_embedding) {
3420 + //error_log('MXChat Vector Search: Starting content search...');
5103 3421
5104 -/**
5105 - * Generate embedding via a Custom (OpenAI-compatible) provider's /embeddings route.
5106 - * Only called when the opt-in 'custom_provider_for_embeddings' setting is on.
5107 - * Returns a numeric array (the embedding vector) on success, or ['error','error_code'] on failure.
5108 - */
5109 -private function mxchat_generate_embedding_custom($text) {
5110 - if (empty($text)) {
5111 - return ['error' => esc_html__('No text provided for embedding generation', 'mxchat'), 'error_code' => 'empty_embedding_text'];
5112 - }
5113 - $cfg = $this->mxchat_resolve_custom_provider();
5114 - if (empty($cfg['base_url'])) {
5115 - return ['error' => esc_html__('Custom provider Base URL is not configured.', 'mxchat'), 'error_code' => 'missing_custom_provider_base_url'];
5116 - }
3422 + // Retrieve the add-on settings from the database.
3423 + $addon_options = get_option('mxchat_pinecone_addon_options', array());
5117 3424
5118 - $options = get_option('mxchat_options');
5119 - $embed_url = $cfg['base_url'] . '/embeddings';
5120 - if (!empty($cfg['api_version'])) {
5121 - $embed_url .= (strpos($embed_url, '?') === false ? '?' : '&') . 'api-version=' . rawurlencode($cfg['api_version']);
5122 - }
5123 - $model = isset($options['custom_provider_embedding_model']) && trim((string) $options['custom_provider_embedding_model']) !== ''
5124 - ? trim((string) $options['custom_provider_embedding_model'])
5125 - : $cfg['model'];
3425 + // Determine whether Pinecone is enabled.
3426 + $use_pinecone = (isset($addon_options['mxchat_use_pinecone']) && $addon_options['mxchat_use_pinecone'] === '1') ? 1 : 0;
5126 3427
5127 - $response = wp_remote_post($embed_url, [
5128 - 'headers' => $this->mxchat_custom_provider_assoc_headers($cfg),
5129 - 'body' => wp_json_encode(['input' => $text, 'model' => $model]),
5130 - 'timeout' => 60,
5131 - ]);
5132 - if (is_wp_error($response)) {
5133 - return [
5134 - 'error' => esc_html__('Connection error when generating embeddings (custom provider): ', 'mxchat') . esc_html($response->get_error_message()),
5135 - 'error_code' => 'embedding_custom_connection_error',
5136 - ];
5137 - }
5138 - $status = wp_remote_retrieve_response_code($response);
5139 - $body = json_decode(wp_remote_retrieve_body($response), true);
5140 - if ($status !== 200) {
5141 - $msg = isset($body['error']['message']) ? $body['error']['message'] : 'HTTP ' . $status;
5142 - return [
5143 - 'error' => esc_html__('Custom embedding endpoint error: ', 'mxchat') . esc_html($msg),
5144 - 'error_code' => 'embedding_custom_api_error',
5145 - 'status_code' => $status,
5146 - ];
5147 - }
5148 - if (isset($body['data'][0]['embedding']) && is_array($body['data'][0]['embedding'])) {
5149 - return $body['data'][0]['embedding'];
5150 - }
5151 - return [
5152 - 'error' => esc_html__('Invalid embedding response from custom provider.', 'mxchat'),
5153 - 'error_code' => 'embedding_custom_invalid_response',
5154 - ];
5155 -}
3428 + //error_log('Pinecone enabled flag: ' . $use_pinecone);
5156 3429
5157 -private function mxchat_find_relevant_content($user_embedding, $bot_id = 'default', $user_query = '') {
5158 - //error_log("MXCHAT DEBUG: find_relevant_content called with bot_id: " . $bot_id);
5159 -
5160 - // Check for OpenAI Vector Store first (takes priority when enabled)
5161 - $bot_vectorstore_config = $this->get_bot_vectorstore_config($bot_id);
5162 -
5163 - if ($bot_vectorstore_config['use_vectorstore']) {
5164 - // Get current model to verify it's an OpenAI model
5165 - $bot_options = $this->get_bot_options($bot_id);
5166 - $mxchat_options = get_option('mxchat_options', array());
5167 - $current_options = !empty($bot_options) ? $bot_options : $mxchat_options;
5168 - $selected_model = $current_options['model'] ?? 'gpt-5.1-chat-latest';
5169 -
5170 - if ($this->is_openai_chat_model($selected_model)) {
5171 - //error_log("MXCHAT DEBUG: Using OpenAI Vector Store for knowledge retrieval");
5172 - return $this->find_relevant_content_openai_vectorstore($user_query, $bot_id, $bot_vectorstore_config);
5173 - } else {
5174 - //error_log("MXCHAT DEBUG: Vector Store enabled but model is not OpenAI (" . $selected_model . "), skipping Vector Store");
5175 - }
5176 - }
5177 -
5178 - // Get bot-specific Pinecone configuration
5179 - $bot_pinecone_config = $this->get_bot_pinecone_config($bot_id);
5180 -
5181 - // Debug: Log the Pinecone configuration
5182 - //error_log("MXCHAT DEBUG: Pinecone config for bot '$bot_id':");
5183 - //error_log(" - use_pinecone: " . ($bot_pinecone_config['use_pinecone'] ? 'true' : 'false'));
5184 - //error_log(" - api_key: " . (empty($bot_pinecone_config['api_key']) ? 'EMPTY' : 'SET (hidden)'));
5185 - //error_log(" - host: " . ($bot_pinecone_config['host'] ?? 'NOT SET'));
5186 - //error_log(" - namespace: " . ($bot_pinecone_config['namespace'] ?? 'NOT SET'));
5187 -
5188 - // Determine whether to use Pinecone based on bot configuration
5189 - $use_pinecone = isset($bot_pinecone_config['use_pinecone']) ? $bot_pinecone_config['use_pinecone'] : false;
5190 -
5191 - //error_log("MXCHAT DEBUG: Using " . ($use_pinecone ? "Pinecone" : "WordPress Database") . " for knowledge retrieval");
5192 -
5193 - if ($use_pinecone) {
5194 - return $this->find_relevant_content_pinecone($user_embedding, $bot_id, $bot_pinecone_config);
3430 + if ($use_pinecone === 1) {
3431 + //error_log('MXChat Vector Search: Using Pinecone database');
3432 + return $this->find_relevant_content_pinecone($user_embedding);
5195 3433 } else {
5196 - return $this->find_relevant_content_wordpress($user_embedding, $bot_id);
3434 + //error_log('MXChat Vector Search: Using WordPress database');
3435 + return $this->find_relevant_content_wordpress($user_embedding);
5197 3436 }
5198 3437 }
5199 3438
5200 -private function find_relevant_content_wordpress($user_embedding, $bot_id = 'default') {
3439 +private function find_relevant_content_wordpress($user_embedding) {
5201 3440 global $wpdb;
5202 3441 $system_prompt_table = $wpdb->prefix . 'mxchat_system_prompt_content';
3442 + $cache_key = 'mxchat_system_prompt_embeddings';
3443 + $batch_size = 500;
3444 +
5203 3445 // Initialize similarity analysis storage
5204 3446 $this->last_similarity_analysis = [
5205 3447 'knowledge_base_type' => 'WordPress Database',
5206 - 'bot_id' => $bot_id,
5207 3448 'top_matches' => [],
5208 3449 'threshold_used' => 0,
5209 3450 'total_checked' => 0
5210 3451 ];
5211 3452
5212 - // NEW: Initialize valid URLs array
5213 - $valid_urls = [];
3453 + // Retrieve embeddings from cache or database
3454 + $embeddings = wp_cache_get($cache_key, 'mxchat_system_prompts');
3455 + if ($embeddings === false) {
3456 + // Cache miss - load embeddings from database WITH CONTENT and ROLE RESTRICTION for testing
3457 + $embeddings = [];
3458 + $offset = 0;
5214 3459
5215 - // Get bot-specific options for similarity threshold
5216 - $bot_options = $this->get_bot_options($bot_id);
5217 - $current_options = !empty($bot_options) ? $bot_options : $this->options;
3460 + do {
3461 + $query = $wpdb->prepare(
3462 + "SELECT id, embedding_vector, article_content, source_url, role_restriction
3463 + FROM {$system_prompt_table}
3464 + LIMIT %d OFFSET %d",
3465 + $batch_size,
3466 + $offset
3467 + );
5218 3468
5219 - // Get knowledge manager instance for role checking
5220 - $knowledge_manager = MxChat_Knowledge_Manager::get_instance();
3469 + $batch = $wpdb->get_results($query);
3470 + if (empty($batch)) {
3471 + break;
3472 + }
5221 3473
5222 - // Get base similarity threshold from bot options or default options
5223 - $similarity_threshold = isset($current_options['similarity_threshold'])
5224 - ? ((int) $current_options['similarity_threshold']) / 100
5225 - : 0.35;
5226 - $this->last_similarity_analysis['threshold_used'] = $similarity_threshold;
3474 + $embeddings = array_merge($embeddings, $batch);
3475 + $offset += $batch_size;
3476 + unset($batch);
3477 + } while (true);
5227 3478
5228 - // Precompute bot_filter once, outside the streaming loop
5229 - $bot_filter = '';
5230 - if ($bot_id !== 'default') {
5231 - $column_exists = $wpdb->get_var("SHOW COLUMNS FROM {$system_prompt_table} LIKE 'bot_metadata'");
5232 - if ($column_exists) {
5233 - $bot_filter = $wpdb->prepare(" AND (bot_metadata = %s OR bot_metadata IS NULL OR bot_metadata = '')", $bot_id);
3479 + if (empty($embeddings)) {
3480 + return '';
5234 3481 }
3482 +
3483 + // Cache embeddings for future use (but note: this now includes content and role restrictions)
3484 + wp_cache_set($cache_key, $embeddings, 'mxchat_system_prompts', 3600);
5235 3485 }
5236 3486
5237 - // ===== STREAMING TOP-K PASS =====
5238 - // Stream rows in small batches, compute cosine similarity per row, and keep only:
5239 - // - top 10 by raw similarity (for the testing/debug display panel)
5240 - // - candidates above threshold with access (capped) for context assembly
5241 - // This bounds peak memory regardless of knowledge base size and avoids loading
5242 - // article_content for every row. article_content is fetched in Phase 2 for winners only.
5243 - $batch_size = 250;
5244 - $max_candidates = 200; // safety cap, well above rag_sources_limit * max_chunks_per_source
5245 - $top_display = [];
5246 - $candidates = [];
5247 - $total_checked = 0;
5248 - $offset = 0;
3487 + // NEW: Get knowledge manager instance for role checking
3488 + $knowledge_manager = MxChat_Knowledge_Manager::get_instance();
5249 3489
5250 - do {
5251 - $batch = $wpdb->get_results($wpdb->prepare(
5252 - "SELECT id, embedding_vector, source_url, role_restriction
5253 - FROM {$system_prompt_table}
5254 - WHERE 1=1 {$bot_filter}
5255 - LIMIT %d OFFSET %d",
5256 - $batch_size,
5257 - $offset
5258 - ));
5259 -
5260 - if (empty($batch)) {
5261 - break;
5262 - }
5263 -
5264 - foreach ($batch as $row) {
5265 - $database_embedding = $row->embedding_vector
5266 - ? unserialize($row->embedding_vector, ['allowed_classes' => false])
5267 - : null;
5268 -
5269 - if (!is_array($database_embedding) || !is_array($user_embedding)) {
5270 - unset($database_embedding);
5271 - continue;
5272 - }
5273 -
3490 + // Get configuration options
3491 + $main_options = get_option('mxchat_options', []);
3492 +
3493 + // Get base similarity threshold (default 75%)
3494 + $similarity_threshold = isset($main_options['similarity_threshold'])
3495 + ? ((int) $main_options['similarity_threshold']) / 100
3496 + : 0.75;
3497 +
3498 + $this->last_similarity_analysis['threshold_used'] = $similarity_threshold;
3499 +
3500 + // Calculate similarities and build results array
3501 + $all_similarities = [];
3502 + $relevant_results = [];
3503 +
3504 + foreach ($embeddings as $embedding) {
3505 + $database_embedding = $embedding->embedding_vector
3506 + ? unserialize($embedding->embedding_vector, ['allowed_classes' => false])
3507 + : null;
3508 +
3509 + if (is_array($database_embedding) && is_array($user_embedding)) {
5274 3510 $similarity = $this->mxchat_calculate_cosine_similarity($user_embedding, $database_embedding);
5275 - unset($database_embedding);
5276 -
5277 - $role_restriction = $row->role_restriction ?? 'public';
3511 +
3512 + // NEW: Check role access
3513 + $role_restriction = $embedding->role_restriction ?? 'public';
5278 3514 $has_access = $knowledge_manager->mxchat_user_has_content_access($role_restriction);
5279 - $source_url = $row->source_url ?? '';
5280 -
5281 - // Maintain top 10 display buffer (insert-if-beats-worst)
5282 - if (count($top_display) < 10) {
5283 - $top_display[] = [
5284 - 'id' => $row->id,
5285 - 'similarity' => $similarity,
5286 - 'source_url' => $source_url,
5287 - 'role_restriction' => $role_restriction,
5288 - 'has_access' => $has_access,
5289 - ];
5290 - usort($top_display, function ($a, $b) {
5291 - return $b['similarity'] <=> $a['similarity'];
5292 - });
5293 - } elseif ($similarity > $top_display[9]['similarity']) {
5294 - $top_display[9] = [
5295 - 'id' => $row->id,
5296 - 'similarity' => $similarity,
5297 - 'source_url' => $source_url,
5298 - 'role_restriction' => $role_restriction,
5299 - 'has_access' => $has_access,
5300 - ];
5301 - usort($top_display, function ($a, $b) {
5302 - return $b['similarity'] <=> $a['similarity'];
5303 - });
3515 +
3516 + // Store ALL similarities for testing (top 10)
3517 + $source_display = '';
3518 + if (!empty($embedding->source_url) && $embedding->source_url !== '#') {
3519 + $source_display = $embedding->source_url;
3520 + } else {
3521 + $content_preview = strip_tags($embedding->article_content ?? '');
3522 + $content_preview = preg_replace('/\s+/', ' ', $content_preview);
3523 + $source_display = substr(trim($content_preview), 0, 50) . '...';
5304 3524 }
5305 -
5306 - // Track candidates for context assembly (above threshold + has access)
3525 +
3526 + $all_similarities[] = [
3527 + 'document_id' => $embedding->id,
3528 + 'similarity' => $similarity,
3529 + 'similarity_percentage' => round($similarity * 100, 2),
3530 + 'above_threshold' => $similarity >= $similarity_threshold,
3531 + 'source_display' => $source_display,
3532 + 'content_preview' => substr(strip_tags($embedding->article_content ?? ''), 0, 100) . '...',
3533 + 'used_for_context' => false, // Initialize as false, we'll update this later
3534 + 'role_restriction' => $role_restriction, // NEW: Include role info for testing
3535 + 'has_access' => $has_access, // NEW: Include access info for testing
3536 + 'filtered_out' => !$has_access // NEW: Mark if filtered out by role
3537 + ];
3538 +
3539 + // Only consider results above threshold AND with access for actual content retrieval
5307 3540 if ($similarity >= $similarity_threshold && $has_access) {
5308 - $candidates[] = [
5309 - 'id' => $row->id,
5310 - 'similarity' => $similarity,
5311 - 'source_url' => $source_url,
3541 + $relevant_results[] = [
3542 + 'id' => $embedding->id,
3543 + 'similarity' => $similarity
5312 3544 ];
5313 3545 }
5314 -
5315 - $total_checked++;
5316 3546 }
5317 -
5318 - unset($batch);
5319 -
5320 - // Trim candidates periodically to cap memory during long scans
5321 - if (count($candidates) > $max_candidates) {
5322 - usort($candidates, function ($a, $b) {
5323 - return $b['similarity'] <=> $a['similarity'];
5324 - });
5325 - $candidates = array_slice($candidates, 0, $max_candidates);
5326 - }
5327 -
5328 - $offset += $batch_size;
5329 - } while (true);
5330 -
5331 - if ($total_checked === 0) {
5332 - $this->current_valid_urls = [];
5333 - return '';
3547 +
3548 + unset($database_embedding);
5334 3549 }
5335 3550
5336 - // Final candidates sort (best first)
5337 - if (count($candidates) > 1) {
5338 - usort($candidates, function ($a, $b) {
5339 - return $b['similarity'] <=> $a['similarity'];
5340 - });
5341 - }
5342 -
5343 - // ===== PHASE 2: FETCH ARTICLE CONTENT ONLY FOR WINNERS =====
5344 - // Gather unique IDs we actually need (top_display + candidates) and pull
5345 - // article_content in bounded IN() batches. This avoids loading content for
5346 - // every row during the similarity scan.
5347 - $needed_ids = [];
5348 - foreach ($top_display as $item) {
5349 - $needed_ids[$item['id']] = true;
5350 - }
5351 - foreach ($candidates as $item) {
5352 - $needed_ids[$item['id']] = true;
5353 - }
5354 - $needed_ids = array_keys($needed_ids);
5355 -
5356 - $content_map = [];
5357 - if (!empty($needed_ids)) {
5358 - foreach (array_chunk($needed_ids, 250) as $chunk_ids) {
5359 - $placeholders = implode(',', array_fill(0, count($chunk_ids), '%d'));
5360 - $rows = $wpdb->get_results($wpdb->prepare(
5361 - "SELECT id, article_content FROM {$system_prompt_table} WHERE id IN ($placeholders)",
5362 - ...$chunk_ids
5363 - ));
5364 - foreach ($rows as $r) {
5365 - $content_map[$r->id] = $r->article_content;
5366 - }
5367 - unset($rows);
5368 - }
5369 - }
5370 -
5371 - // Build the all_similarities display array from the top 10
5372 - $all_similarities = [];
5373 - foreach ($top_display as $item) {
5374 - $article_content_for_parse = $content_map[$item['id']] ?? '';
5375 - $parsed_for_display = MxChat_Chunker::parse_stored_chunk($article_content_for_parse);
5376 - $is_chunk = $parsed_for_display['is_chunked'];
5377 - $chunk_meta = $parsed_for_display['metadata'];
5378 -
5379 - if (!empty($item['source_url']) && $item['source_url'] !== '#') {
5380 - $source_display = $item['source_url'];
5381 - } else {
5382 - $content_preview = strip_tags($article_content_for_parse);
5383 - $content_preview = preg_replace('/\s+/', ' ', $content_preview);
5384 - $source_display = substr(trim($content_preview), 0, 50) . '...';
5385 - }
5386 -
5387 - $all_similarities[] = [
5388 - 'document_id' => $item['id'],
5389 - 'similarity' => $item['similarity'],
5390 - 'similarity_percentage' => round($item['similarity'] * 100, 2),
5391 - 'above_threshold' => $item['similarity'] >= $similarity_threshold,
5392 - 'source_display' => $source_display,
5393 - 'content_preview' => substr(strip_tags($parsed_for_display['text'] ?? ''), 0, 100) . '...',
5394 - 'used_for_context' => false,
5395 - 'role_restriction' => $item['role_restriction'],
5396 - 'has_access' => $item['has_access'],
5397 - 'filtered_out' => !$item['has_access'],
5398 - 'is_chunk' => $is_chunk,
5399 - 'chunk_index' => $is_chunk ? ($chunk_meta['chunk_index'] ?? 0) : null,
5400 - 'total_chunks' => $is_chunk ? ($chunk_meta['total_chunks'] ?? 1) : null
5401 - ];
5402 - }
5403 -
5404 - // Build url_groups from candidates for chunk reassembly
5405 - $url_groups = array();
5406 - foreach ($candidates as $cand) {
5407 - $article_content = $content_map[$cand['id']] ?? '';
5408 - $parsed = MxChat_Chunker::parse_stored_chunk($article_content);
5409 - $is_chunked = $parsed['is_chunked'];
5410 - $chunk_index = $parsed['metadata']['chunk_index'] ?? 0;
5411 - $text_content = $parsed['text'];
5412 -
5413 - $source_url = $cand['source_url'];
5414 - $group_key = !empty($source_url) ? $source_url : '_manual_' . $cand['id'];
5415 -
5416 - if (!isset($url_groups[$group_key])) {
5417 - $url_groups[$group_key] = array(
5418 - 'source_url' => $source_url,
5419 - 'best_score' => 0,
5420 - 'is_chunked' => $is_chunked,
5421 - 'chunks' => array(),
5422 - 'single_text' => '',
5423 - 'single_id' => null
5424 - );
5425 - }
5426 -
5427 - if ($cand['similarity'] > $url_groups[$group_key]['best_score']) {
5428 - $url_groups[$group_key]['best_score'] = $cand['similarity'];
5429 - }
5430 -
5431 - if ($is_chunked) {
5432 - $url_groups[$group_key]['is_chunked'] = true;
5433 - $url_groups[$group_key]['chunks'][] = array(
5434 - 'id' => $cand['id'],
5435 - 'score' => $cand['similarity'],
5436 - 'chunk_index' => $chunk_index,
5437 - 'text' => $text_content
5438 - );
5439 - } else {
5440 - $url_groups[$group_key]['single_text'] = $text_content;
5441 - $url_groups[$group_key]['single_id'] = $cand['id'];
5442 - }
5443 - }
5444 -
5445 3551 // Sort ALL similarities for testing display (highest first)
5446 3552 usort($all_similarities, function ($a, $b) {
5447 3553 return $b['similarity'] <=> $a['similarity'];
5448 3554 });
5449 -
5450 - // Sort URL groups by best score (highest first)
5451 - uasort($url_groups, function($a, $b) {
5452 - return $b['best_score'] <=> $a['best_score'];
3555 +
3556 + // Sort relevant results by similarity (highest first)
3557 + usort($relevant_results, function ($a, $b) {
3558 + return $b['similarity'] <=> $a['similarity'];
5453 3559 });
5454 -
5455 - // Get RAG sources limit from options (default 6, min 3, max 10)
5456 - $rag_sources_limit = isset($current_options['rag_sources_limit']) ? intval($current_options['rag_sources_limit']) : 3;
5457 - if ($rag_sources_limit < 3) $rag_sources_limit = 3;
5458 - if ($rag_sources_limit > 10) $rag_sources_limit = 10;
5459 -
5460 - // Take top N unique URLs based on user setting
5461 - $top_urls = array_slice($url_groups, 0, $rag_sources_limit, true);
5462 -
5463 - // Track which document IDs are used for context
3560 +
3561 + // Get top 5 results for actual content (standard approach)
3562 + $top_results = array_slice($relevant_results, 0, 5);
3563 +
3564 + // NOW mark which documents are actually used for context
5464 3565 $used_document_ids = [];
5465 - foreach ($top_urls as $group) {
5466 - if ($group['is_chunked']) {
5467 - foreach ($group['chunks'] as $chunk) {
5468 - $used_document_ids[] = $chunk['id'];
5469 - }
5470 - } elseif ($group['single_id']) {
5471 - $used_document_ids[] = $group['single_id'];
5472 - }
3566 + foreach ($top_results as $result) {
3567 + $used_document_ids[] = $result['id'];
5473 3568 }
5474 -
3569 +
5475 3570 // Update the all_similarities array to mark which were actually used
5476 3571 foreach ($all_similarities as &$similarity_item) {
5477 3572 $similarity_item['used_for_context'] = in_array($similarity_item['document_id'], $used_document_ids);
5478 3573 }
5479 -
5480 - // Store top 10 for testing panel
3574 +
3575 + // Store top 10 for testing panel (now with correct used_for_context flags and role info)
5481 3576 $this->last_similarity_analysis['top_matches'] = array_slice($all_similarities, 0, 10);
5482 - $this->last_similarity_analysis['total_checked'] = $total_checked;
5483 -
3577 + $this->last_similarity_analysis['total_checked'] = count($embeddings);
3578 +
3579 + //error_log("MxChat Testing: Stored " . count($this->last_similarity_analysis['top_matches']) . " top matches for testing");
3580 +
5484 3581 // Initialize final content
5485 3582 $content = '';
5486 - $matches_used = 0;
5487 - $total_chunks_used = 0;
5488 - $max_total_chunks = isset($current_options['rag_chunks_limit']) ? intval($current_options['rag_chunks_limit']) : 15;
5489 - if ($max_total_chunks < 8) $max_total_chunks = 8;
5490 - if ($max_total_chunks > 20) $max_total_chunks = 20;
5491 - $max_chunks_per_source = 5; // Cap per individual source to limit token usage
5492 -
5493 - // Check if citation links are enabled (default to 'on' for backwards compatibility)
5494 - // Use fresh options to ensure we get the latest setting value
5495 - $fresh_options = get_option('mxchat_options', []);
5496 - $citation_links_enabled = isset($fresh_options['citation_links_toggle']) ? ($fresh_options['citation_links_toggle'] === 'on') : true;
5497 -
5498 - // Build content from top sources
5499 - foreach ($top_urls as $group_key => $group) {
5500 - $source_url = $group['source_url']; // Use actual source_url, not the group key
5501 -
5502 - // Stop if we've hit the total chunk limit
5503 - if ($total_chunks_used >= $max_total_chunks) {
5504 - break;
3583 +
3584 + // Track document IDs to avoid duplicates
3585 + $added_document_ids = [];
3586 +
3587 + // Fetch and format content for each selected result
3588 + foreach ($top_results as $index => $result) {
3589 + if (in_array($result['id'], $added_document_ids)) {
3590 + continue;
5505 3591 }
5506 -
5507 - $full_text = '';
5508 - $chunks_in_this_source = 1; // Default for non-chunked content
5509 -
5510 - if ($group['is_chunked']) {
5511 - // Calculate how many chunks we can still use (respect both total and per-source caps)
5512 - $chunks_remaining = min($max_chunks_per_source, $max_total_chunks - $total_chunks_used);
5513 -
5514 - // Fetch chunks for this URL with limit
5515 - $full_text = $this->reassemble_chunks_from_wordpress($source_url, $chunks_remaining, $chunks_in_this_source);
5516 -
5517 - // If fetching all chunks fails, fall back to matched chunks
5518 - if (empty($full_text)) {
5519 - // Sort matched chunks by index and concatenate
5520 - usort($group['chunks'], function($a, $b) {
5521 - return $a['chunk_index'] <=> $b['chunk_index'];
5522 - });
5523 -
5524 - $chunk_texts = array();
5525 - $chunks_in_this_source = 0;
5526 - foreach ($group['chunks'] as $chunk) {
5527 - if ($total_chunks_used + $chunks_in_this_source >= $max_total_chunks) {
5528 - break;
5529 - }
5530 - $chunk_texts[] = $chunk['text'];
5531 - $chunks_in_this_source++;
3592 +
3593 + $chunk_content = $this->fetch_content_with_product_links($result['id']);
3594 + $added_document_ids[] = $result['id'];
3595 +
3596 + $content .= "## Reference " . ($index + 1) . " ##\n";
3597 + $content .= $chunk_content . "\n\n";
3598 +
3599 + // PDF surrounding pages logic (unchanged)
3600 + if (strpos($chunk_content, '{"document_type":"pdf"') !== false) {
3601 + $surrounding_content = $wpdb->get_results($wpdb->prepare(
3602 + "SELECT id, article_content, role_restriction FROM {$system_prompt_table}
3603 + WHERE id IN (
3604 + (SELECT id FROM {$system_prompt_table} WHERE id < %d ORDER BY id DESC LIMIT 1),
3605 + (SELECT id FROM {$system_prompt_table} WHERE id > %d ORDER BY id ASC LIMIT 1)
3606 + )",
3607 + $result['id'],
3608 + $result['id']
3609 + ));
3610 +
3611 + // NEW: Check role access for surrounding content too
3612 + if (!empty($surrounding_content[0])) {
3613 + $surrounding_role = $surrounding_content[0]->role_restriction ?? 'public';
3614 + if ($knowledge_manager->mxchat_user_has_content_access($surrounding_role)) {
3615 + $content .= "## Related Content ##\n";
3616 + $content .= $surrounding_content[0]->article_content . "\n\n";
3617 + $added_document_ids[] = $surrounding_content[0]->id;
5532 3618 }
5533 - $full_text = implode("\n\n", $chunk_texts);
5534 3619 }
5535 - } else {
5536 - $full_text = $group['single_text'];
5537 - $chunks_in_this_source = 1;
5538 - }
5539 -
5540 - if (!empty($full_text)) {
5541 - // Strip URLs from content if citation links are disabled
5542 - if (!$citation_links_enabled) {
5543 - $full_text = preg_replace('#\bhttps?://[^\s<>"\']+#i', '', $full_text);
5544 - $full_text = preg_replace('/\s+/', ' ', trim($full_text)); // Clean up extra spaces
5545 - }
5546 -
5547 - // Use numbered reference for URL-based entries, plain info label for manual entries
5548 - // Manual entries are stored with an internal mxchat:// placeholder URL — never expose them as citations
5549 - if (!empty($source_url) && $source_url !== '#' && strpos($source_url, 'mxchat://') !== 0) {
5550 - $matches_used++;
5551 - $content .= "## Reference " . $matches_used . " ##\n";
5552 - $content .= $full_text . "\n\n";
5553 -
5554 - // Only include citation URLs if citation links are enabled
5555 - if ($citation_links_enabled) {
5556 - $valid_urls[] = $source_url;
5557 - $content .= "URL: " . $source_url . "\n\n";
3620 +
3621 + if (!empty($surrounding_content[1])) {
3622 + $surrounding_role = $surrounding_content[1]->role_restriction ?? 'public';
3623 + if ($knowledge_manager->mxchat_user_has_content_access($surrounding_role)) {
3624 + $content .= "## Related Content ##\n";
3625 + $content .= $surrounding_content[1]->article_content . "\n\n";
3626 + $added_document_ids[] = $surrounding_content[1]->id;
5558 3627 }
5559 - } else {
5560 - // Manual entry — no reference number, no citation
5561 - $content .= "## Information ##\n";
5562 - $content .= $full_text . "\n\n";
5563 3628 }
5564 -
5565 - // Extract any URLs from the text content itself (only if citation links enabled)
5566 - if ($citation_links_enabled) {
5567 - preg_match_all(
5568 - '#\bhttps?://[^\s<>"\']+#i',
5569 - $full_text,
5570 - $content_urls
5571 - );
5572 - if (!empty($content_urls[0])) {
5573 - $valid_urls = array_merge($valid_urls, $content_urls[0]);
5574 - }
5575 - }
5576 -
5577 - $total_chunks_used += $chunks_in_this_source;
5578 3629 }
5579 3630 }
5580 -
5581 - // NEW: Store unique valid URLs for validation
5582 - $this->current_valid_urls = array_unique($valid_urls);
5583 -
5584 - // Store sources and chunks counts for testing/transcript display
5585 - $this->last_similarity_analysis['sources_used'] = $matches_used;
5586 - $this->last_similarity_analysis['total_chunks_used'] = $total_chunks_used;
5587 -
5588 - // Allow add-ons to act on similarity results (e.g. WooCommerce product card display)
5589 - do_action('mxchat_similarity_results', $this->last_similarity_analysis['top_matches'], $bot_id);
5590 -
3631 +
5591 3632 // Add response guidelines
5592 - if (empty($top_urls)) {
3633 + if (empty($top_results)) {
5593 3634 $content = "No reference information was found for this query.\n\n";
5594 3635 } else {
5595 - // Build response guidelines based on citation links setting
5596 3636 $content .= "\n## Response Guidelines ##\n" .
5597 3637 "You are an AI Chatbot. Answer naturally and helpfully using only the information from the references above. " .
5598 3638 "Be conversational and friendly, but never mention your knowledge base or training data. " .
5599 3639 "If you don't have specific information or are uncertain about any details, it's always " .
5600 3640 "better to honestly say you don't know rather than making up or guessing at answers. " .
5601 - "When information is incomplete, let them know you are unsure.\n\n";
5602 -
5603 - // Only add hyperlink instructions if citation links are enabled
5604 - if ($citation_links_enabled) {
5605 - $content .= "CRITICAL: When creating hyperlinks, always use proper markdown format with descriptive text: " .
5606 - "[descriptive text](url). NEVER use empty brackets like [](url). The text in brackets must describe what the link is about. " .
5607 - "Only cite references that have a URL. Do not cite or add source labels to Information sections that have no URL.";
5608 - } else {
5609 - $content .= "IMPORTANT: Do not include any citation links, source URLs, or hyperlinks in your responses. " .
5610 - "Simply provide helpful answers based on the reference information without citing sources.";
5611 - }
3641 + "When information is incomplete, let them know you are unsure.";
5612 3642 }
5613 3643
5614 3644 return trim($content);
5615 3645 }
5616 3646
5617 -/**
5618 - * Fetch and reassemble chunks for a URL from WordPress database
5619 - *
5620 - * @param string $source_url The source URL to fetch chunks for
5621 - * @param int $max_chunks Maximum number of chunks to return (0 = unlimited)
5622 - * @param int &$chunk_count Reference to store the actual number of chunks returned
5623 - * @return string Reassembled content from chunks
5624 - */
5625 -private function reassemble_chunks_from_wordpress($source_url, $max_chunks = 0, &$chunk_count = 0) {
5626 - global $wpdb;
5627 - $table = $wpdb->prefix . 'mxchat_system_prompt_content';
5628 -
5629 - // Fetch all rows with this source_url
5630 - $rows = $wpdb->get_results($wpdb->prepare(
5631 - "SELECT article_content FROM {$table}
5632 - WHERE source_url = %s
5633 - ORDER BY id ASC",
5634 - $source_url
5635 - ));
5636 -
5637 - if (empty($rows)) {
5638 - $chunk_count = 0;
5639 - return '';
5640 - }
5641 -
5642 - // Parse and sort chunks by index
5643 - $chunks = array();
5644 - foreach ($rows as $row) {
5645 - $parsed = MxChat_Chunker::parse_stored_chunk($row->article_content);
5646 -
5647 - if ($parsed['is_chunked']) {
5648 - $chunk_index = $parsed['metadata']['chunk_index'] ?? 0;
5649 - $chunks[$chunk_index] = $parsed['text'];
5650 - } else {
5651 - // Non-chunked content - just return it
5652 - $chunks[] = $parsed['text'];
5653 - }
5654 - }
5655 -
5656 - // Sort by chunk index
5657 - ksort($chunks);
5658 -
5659 - // Apply chunk limit if specified
5660 - if ($max_chunks > 0 && count($chunks) > $max_chunks) {
5661 - $chunks = array_slice($chunks, 0, $max_chunks, true);
5662 - }
5663 -
5664 - // Store actual chunk count
5665 - $chunk_count = count($chunks);
5666 -
5667 - // Reassemble content
5668 - return implode("\n\n", $chunks);
5669 -}
5670 -
5671 -private function find_relevant_content_pinecone($user_embedding, $bot_id = 'default', $bot_config = null) {
5672 - global $wpdb;
3647 +private function find_relevant_content_pinecone($user_embedding) {
3648 + global $wpdb; // For single role lookups
3649 + $options = get_option('mxchat_pinecone_addon_options', array());
3650 + $api_key = $options['mxchat_pinecone_api_key'] ?? '';
3651 + $host = $options['mxchat_pinecone_host'] ?? '';
5673 3652
5674 - //error_log("MXCHAT DEBUG: find_relevant_content_pinecone called");
5675 - //error_log(" - bot_id: " . $bot_id);
5676 - //error_log(" - user_embedding is array: " . (is_array($user_embedding) ? 'yes' : 'no'));
5677 - //error_log(" - user_embedding count: " . (is_array($user_embedding) ? count($user_embedding) : 'N/A'));
5678 -
5679 - // Use bot-specific config or fall back to default
5680 - if ($bot_config === null) {
5681 - $bot_config = $this->get_bot_pinecone_config($bot_id);
5682 - }
5683 -
5684 - $api_key = $bot_config['api_key'] ?? '';
5685 - $host = $bot_config['host'] ?? '';
5686 - $namespace = $bot_config['namespace'] ?? '';
5687 -
5688 - //error_log("MXCHAT DEBUG: Pinecone query parameters:");
5689 - //error_log(" - API Key: " . (empty($api_key) ? 'EMPTY - ERROR!' : 'Present (length: ' . strlen($api_key) . ')'));
5690 - //error_log(" - Host: " . (empty($host) ? 'EMPTY - ERROR!' : $host));
5691 - //error_log(" - Namespace: " . (empty($namespace) ? 'EMPTY (will use default)' : $namespace));
5692 -
5693 3653 // Initialize similarity analysis storage
5694 3654 $this->last_similarity_analysis = [
5695 3655 'knowledge_base_type' => 'Pinecone',
5696 - 'bot_id' => $bot_id,
5697 - 'namespace' => $namespace,
5698 3656 'top_matches' => [],
5699 3657 'threshold_used' => 0,
5700 3658 'total_checked' => 0
5701 3659 ];
5702 3660
5703 - // NEW: Initialize valid URLs array
5704 - $valid_urls = [];
5705 -
5706 3661 if (empty($host) || empty($api_key)) {
5707 - //error_log("MXCHAT DEBUG ERROR: Missing Pinecone host or API key!");
5708 - //error_log(" - Host empty: " . (empty($host) ? 'YES' : 'NO'));
5709 - //error_log(" - API key empty: " . (empty($api_key) ? 'YES' : 'NO'));
5710 - // Store empty array for valid URLs since we can't proceed
5711 - $this->current_valid_urls = [];
5712 3662 return '';
5713 3663 }
5714 3664
5715 3665 // Get knowledge manager instance for role checking
@@ -5714,37 +3664,26 @@
5714 3664
5715 3665 // Get knowledge manager instance for role checking
5716 3666 $knowledge_manager = MxChat_Knowledge_Manager::get_instance();
5717 3667
5718 - // Get the similarity threshold from the bot options or main options
5719 - $bot_options = $this->get_bot_options($bot_id);
5720 - $current_options = !empty($bot_options) ? $bot_options : get_option('mxchat_options', []);
3668 + // Get the similarity threshold from the main options
3669 + $main_options = get_option('mxchat_options', []);
3670 + $similarity_threshold = isset($main_options['similarity_threshold'])
3671 + ? ((int) $main_options['similarity_threshold']) / 100
3672 + : 0.75;
5721 3673
5722 - $similarity_threshold = isset($current_options['similarity_threshold'])
5723 - ? ((int) $current_options['similarity_threshold']) / 100
5724 - : 0.35;
5725 -
5726 3674 $this->last_similarity_analysis['threshold_used'] = $similarity_threshold;
5727 3675
5728 - // Prepare the query request for Pinecone
3676 + // Prepare the query request for Pinecone (request more for testing)
5729 3677 $api_endpoint = "https://{$host}/query";
5730 3678
5731 3679 $request_body = array(
5732 3680 'vector' => $user_embedding,
5733 - 'topK' => 50, // Increased for chunked content grouping - need more candidates to find top N unique URLs
3681 + 'topK' => 20, // Request more to get good testing data
5734 3682 'includeMetadata' => true,
5735 3683 'includeValues' => true
5736 3684 );
5737 3685
5738 - // Add namespace if specified for this bot
5739 - if (!empty($namespace)) {
5740 - $request_body['namespace'] = $namespace;
5741 - }
5742 -
5743 - //error_log("MXCHAT DEBUG: About to call Pinecone API");
5744 - //error_log(" - Endpoint: " . $api_endpoint);
5745 - //error_log(" - Namespace in request: " . (!empty($namespace) ? $namespace : 'NOT SET'));
5746 -
5747 3686 $response = wp_remote_post($api_endpoint, array(
5748 3687 'headers' => array(
5749 3688 'Api-Key' => $api_key,
5750 3689 'accept' => 'application/json',
@@ -5754,242 +3693,62 @@
5754 3693 'timeout' => 30
5755 3694 ));
5756 3695
5757 3696 if (is_wp_error($response)) {
5758 - //error_log("MXCHAT DEBUG ERROR: WP Error in Pinecone request: " . $response->get_error_message());
5759 - // Store empty array for valid URLs
5760 - $this->current_valid_urls = [];
5761 3697 return '';
5762 3698 }
5763 3699
5764 3700 $response_code = wp_remote_retrieve_response_code($response);
5765 - //error_log("MXCHAT DEBUG: Pinecone response code: " . $response_code);
5766 -
5767 3701 if ($response_code !== 200) {
5768 - $response_body = wp_remote_retrieve_body($response);
5769 - //error_log("MXCHAT DEBUG ERROR: Pinecone API error response: " . substr($response_body, 0, 500));
5770 - // Store empty array for valid URLs
5771 - $this->current_valid_urls = [];
5772 3702 return '';
5773 3703 }
5774 3704
5775 - // ADD DETAILED DEBUG SECTION HERE
5776 - $response_body = wp_remote_retrieve_body($response);
5777 - //error_log("MXCHAT DEBUG: Raw Pinecone response length: " . strlen($response_body));
5778 -
5779 - $results = json_decode($response_body, true);
5780 -
5781 - if (json_last_error() !== JSON_ERROR_NONE) {
5782 - //error_log("MXCHAT DEBUG ERROR: JSON decode error: " . json_last_error_msg());
5783 - //error_log("MXCHAT DEBUG: First 500 chars of response: " . substr($response_body, 0, 500));
5784 - // Store empty array for valid URLs
5785 - $this->current_valid_urls = [];
5786 - return '';
5787 - }
5788 -
5789 - //error_log("MXCHAT DEBUG: Pinecone response structure:");
5790 - //error_log(" - Has 'matches' key: " . (isset($results['matches']) ? 'yes' : 'no'));
5791 - //error_log(" - Has 'namespace' key: " . (isset($results['namespace']) ? 'yes (' . $results['namespace'] . ')' : 'no'));
5792 -
3705 + $results = json_decode(wp_remote_retrieve_body($response), true);
5793 3706 if (empty($results['matches'])) {
5794 - //error_log("MXCHAT DEBUG: No matches found in Pinecone response");
5795 - //error_log("MXCHAT DEBUG: Response keys: " . implode(', ', array_keys($results)));
5796 - // Store empty array for valid URLs
5797 - $this->current_valid_urls = [];
5798 3707 return '';
5799 3708 }
5800 3709
5801 - //error_log("MXCHAT DEBUG: Found " . count($results['matches']) . " matches in Pinecone");
5802 -
5803 - // Log first match details for debugging
5804 - if (!empty($results['matches'][0])) {
5805 - $first_match = $results['matches'][0];
5806 - //error_log("MXCHAT DEBUG: First match details:");
5807 - //error_log(" - Score: " . ($first_match['score'] ?? 'no score'));
5808 - //error_log(" - Has metadata: " . (isset($first_match['metadata']) ? 'yes' : 'no'));
5809 - if (isset($first_match['metadata'])) {
5810 - //error_log(" - Metadata keys: " . implode(', ', array_keys($first_match['metadata'])));
5811 - }
5812 - }
5813 -
5814 3710 // Initialize the final content
5815 3711 $content = '';
5816 3712 $matches_used = 0;
5817 3713 $matches_used_for_context = [];
5818 - $total_chunks_used = 0;
5819 - $max_total_chunks = isset($current_options['rag_chunks_limit']) ? intval($current_options['rag_chunks_limit']) : 15;
5820 - if ($max_total_chunks < 8) $max_total_chunks = 8;
5821 - if ($max_total_chunks > 20) $max_total_chunks = 20;
5822 - $max_chunks_per_source = 5; // Cap per individual source to limit token usage
5823 -
5824 - // Check if citation links are enabled (default to 'on' for backwards compatibility)
5825 - // Use fresh options to ensure we get the latest setting value
5826 - $fresh_options = get_option('mxchat_options', []);
5827 - $citation_links_enabled = isset($fresh_options['citation_links_toggle']) ? ($fresh_options['citation_links_toggle'] === 'on') : true;
5828 -
5829 - // NEW CHUNKING LOGIC: Group results by source_url for chunk reassembly
5830 - $url_groups = array();
5831 -
3714 +
3715 + // Process each match for actual content generation (lazy role checking)
5832 3716 foreach ($results['matches'] as $index => $match) {
5833 3717 // Skip if similarity is below threshold
5834 3718 if ($match['score'] < $similarity_threshold) {
5835 3719 continue;
5836 3720 }
5837 -
5838 - $metadata = $match['metadata'] ?? array();
5839 - $source_url = $metadata['source_url'] ?? '';
5840 - $match_id = $match['id'] ?? '';
5841 -
5842 - // LAZY ROLE CHECK: Only check role for content we're actually considering
5843 - $role_restriction = $this->get_single_vector_role($match_id, $metadata);
5844 - $has_access = $knowledge_manager->mxchat_user_has_content_access($role_restriction);
5845 -
5846 - // Skip if user doesn't have access
5847 - if (!$has_access) {
5848 - continue;
5849 - }
5850 -
5851 - // Use a unique key for manual entries without a source URL
5852 - $group_key = !empty($source_url) ? $source_url : '_manual_' . $match_id;
5853 -
5854 - // Group by source URL (or unique key for manual entries)
5855 - if (!isset($url_groups[$group_key])) {
5856 - $url_groups[$group_key] = array(
5857 - 'source_url' => $source_url,
5858 - 'best_score' => 0,
5859 - 'is_chunked' => isset($metadata['is_chunked']) && $metadata['is_chunked'],
5860 - 'chunks' => array(),
5861 - 'single_text' => ''
5862 - );
5863 - }
5864 -
5865 - // Track best score for this group
5866 - if ($match['score'] > $url_groups[$group_key]['best_score']) {
5867 - $url_groups[$group_key]['best_score'] = $match['score'];
5868 - }
5869 -
5870 - // Store chunk info or single text
5871 - if ($url_groups[$group_key]['is_chunked']) {
5872 - $url_groups[$group_key]['chunks'][] = array(
5873 - 'id' => $match_id,
5874 - 'score' => $match['score'],
5875 - 'chunk_index' => $metadata['chunk_index'] ?? 0,
5876 - 'text' => $metadata['text'] ?? ''
5877 - );
5878 - } else {
5879 - // Non-chunked content - just store the text
5880 - $url_groups[$group_key]['single_text'] = $metadata['text'] ?? '';
5881 - $url_groups[$group_key]['single_id'] = $match_id;
5882 - }
5883 - }
5884 -
5885 - // Sort URL groups by best score (highest first)
5886 - uasort($url_groups, function($a, $b) {
5887 - return $b['best_score'] <=> $a['best_score'];
5888 - });
5889 -
5890 - // Get RAG sources limit from options (default 6, min 3, max 10)
5891 - $rag_sources_limit = isset($current_options['rag_sources_limit']) ? intval($current_options['rag_sources_limit']) : 3;
5892 - if ($rag_sources_limit < 3) $rag_sources_limit = 3;
5893 - if ($rag_sources_limit > 10) $rag_sources_limit = 10;
5894 -
5895 - // Take top N unique URLs based on user setting
5896 - $top_urls = array_slice($url_groups, 0, $rag_sources_limit, true);
5897 -
5898 - // Track which match IDs are actually used for context
5899 - foreach ($top_urls as $group) {
5900 - if ($group['is_chunked']) {
5901 - foreach ($group['chunks'] as $chunk) {
5902 - $matches_used_for_context[] = $chunk['id'];
5903 - }
5904 - } elseif (!empty($group['single_id'])) {
5905 - $matches_used_for_context[] = $group['single_id'];
5906 - }
5907 - }
5908 -
5909 - // Build content from top sources
5910 - foreach ($top_urls as $group_key => $group) {
5911 - $source_url = $group['source_url']; // Use actual source_url, not the group key
5912 -
5913 - // Stop if we've hit the total chunk limit
5914 - if ($total_chunks_used >= $max_total_chunks) {
3721 +
3722 + // Limit to top 5 matches above threshold
3723 + if ($matches_used >= 5) {
5915 3724 break;
5916 3725 }
5917 -
5918 - $full_text = '';
5919 - $chunks_in_this_source = 1; // Default for non-chunked content
5920 -
5921 - if ($group['is_chunked']) {
5922 - // Calculate how many chunks we can still use (respect both total and per-source caps)
5923 - $chunks_remaining = min($max_chunks_per_source, $max_total_chunks - $total_chunks_used);
5924 -
5925 - // Fetch chunks for this URL with limit
5926 - $full_text = $this->reassemble_chunks_from_pinecone($source_url, $bot_config, $chunks_remaining, $chunks_in_this_source);
5927 -
5928 - // If fetching all chunks fails, fall back to matched chunks
5929 - if (empty($full_text)) {
5930 - // Sort matched chunks by index and concatenate
5931 - usort($group['chunks'], function($a, $b) {
5932 - return $a['chunk_index'] <=> $b['chunk_index'];
5933 - });
5934 -
5935 - $chunk_texts = array();
5936 - $chunks_in_this_source = 0;
5937 - foreach ($group['chunks'] as $chunk) {
5938 - if ($total_chunks_used + $chunks_in_this_source >= $max_total_chunks) {
5939 - break;
5940 - }
5941 - $chunk_texts[] = $chunk['text'];
5942 - $chunks_in_this_source++;
5943 - }
5944 - $full_text = implode("\n\n", $chunk_texts);
3726 +
3727 + if (!empty($match['metadata']['text'])) {
3728 + // LAZY ROLE CHECK: Only check role for content we're actually considering
3729 + $match_id = $match['id'] ?? '';
3730 + $role_restriction = $this->get_single_vector_role($match_id, $match['metadata']);
3731 + $has_access = $knowledge_manager->mxchat_user_has_content_access($role_restriction);
3732 +
3733 + // Skip if user doesn't have access
3734 + if (!$has_access) {
3735 + continue;
5945 3736 }
5946 - } else {
5947 - $full_text = $group['single_text'];
5948 - $chunks_in_this_source = 1;
5949 - }
5950 -
5951 - if (!empty($full_text)) {
5952 - // Strip URLs from content if citation links are disabled
5953 - if (!$citation_links_enabled) {
5954 - $full_text = preg_replace('#\bhttps?://[^\s<>"\']+#i', '', $full_text);
5955 - $full_text = preg_replace('/\s+/', ' ', trim($full_text)); // Clean up extra spaces
3737 +
3738 + // User has access - add to content
3739 + $content .= "## Reference " . ($matches_used + 1) . " ##\n";
3740 + $content .= $match['metadata']['text'] . "\n\n";
3741 +
3742 + if (!empty($match['metadata']['source_url'])) {
3743 + $content .= "URL: " . $match['metadata']['source_url'] . "\n\n";
5956 3744 }
5957 -
5958 - // Use numbered reference for URL-based entries, plain info label for manual entries
5959 - // Manual entries are stored with an internal mxchat:// placeholder URL — never expose them as citations
5960 - if (!empty($source_url) && $source_url !== '#' && strpos($source_url, 'mxchat://') !== 0) {
5961 - $matches_used++;
5962 - $content .= "## Reference " . $matches_used . " ##\n";
5963 - $content .= $full_text . "\n\n";
5964 -
5965 - // Only include citation URLs if citation links are enabled
5966 - if ($citation_links_enabled) {
5967 - $valid_urls[] = $source_url;
5968 - $content .= "URL: " . $source_url . "\n\n";
5969 - }
5970 - } else {
5971 - // Manual entry — no reference number, no citation
5972 - $content .= "## Information ##\n";
5973 - $content .= $full_text . "\n\n";
5974 - }
5975 -
5976 - // Extract any URLs from the text content itself (only if citation links enabled)
5977 - if ($citation_links_enabled) {
5978 - preg_match_all(
5979 - '#\bhttps?://[^\s<>"\']+#i',
5980 - $full_text,
5981 - $content_urls
5982 - );
5983 - if (!empty($content_urls[0])) {
5984 - $valid_urls = array_merge($valid_urls, $content_urls[0]);
5985 - }
5986 - }
5987 -
5988 - $total_chunks_used += $chunks_in_this_source;
3745 +
3746 + $matches_used_for_context[] = $match['id'] ?? $index;
3747 + $matches_used++;
5989 3748 }
5990 3749 }
5991 -
3750 +
5992 3751 // Process ALL matches for testing data (top 10) - with role checking for testing display
5993 3752 $all_matches = [];
5994 3753 foreach ($results['matches'] as $index => $match) {
5995 3754 if ($index >= 10) break; // Limit to top 10 for testing
@@ -6009,19 +3768,9 @@
6009 3768 $source_display = substr(trim($content_preview), 0, 50) . '...';
6010 3769 }
6011 3770
6012 3771 $match_id_for_display = $match['id'] ?? $index;
6013 -
6014 - // Check for chunk metadata in Pinecone
6015 - $is_chunk = isset($match['metadata']['is_chunked']) && $match['metadata']['is_chunked'];
6016 - $chunk_index = isset($match['metadata']['chunk_index']) ? intval($match['metadata']['chunk_index']) : null;
6017 - $total_chunks = isset($match['metadata']['total_chunks']) ? intval($match['metadata']['total_chunks']) : null;
6018 -
6019 - // Also detect chunk from vector ID pattern: {hash}_chunk_{index}
6020 - if (!$is_chunk && MxChat_Chunker::is_chunk_vector_id($match_id_for_display)) {
6021 - $is_chunk = true;
6022 - }
6023 -
3772 +
6024 3773 $all_matches[] = [
6025 3774 'document_id' => $match_id_for_display,
6026 3775 'similarity' => $match['score'],
6027 3776 'similarity_percentage' => round($match['score'] * 100, 2),
@@ -6030,12 +3779,9 @@
6030 3779 'content_preview' => substr(strip_tags($match['metadata']['text'] ?? ''), 0, 100) . '...',
6031 3780 'used_for_context' => in_array($match_id_for_display, $matches_used_for_context),
6032 3781 'role_restriction' => $role_restriction,
6033 3782 'has_access' => $has_access,
6034 - 'filtered_out' => !$has_access,
6035 - 'is_chunk' => $is_chunk,
6036 - 'chunk_index' => $chunk_index,
6037 - 'total_chunks' => $total_chunks
3783 + 'filtered_out' => !$has_access
6038 3784 ];
6039 3785 }
6040 3786
6041 3787 // Store for testing panel
@@ -6040,40 +3786,23 @@
6040 3786
6041 3787 // Store for testing panel
6042 3788 $this->last_similarity_analysis['top_matches'] = $all_matches;
6043 3789 $this->last_similarity_analysis['total_checked'] = count($results['matches']);
6044 - $this->last_similarity_analysis['sources_used'] = $matches_used;
6045 - $this->last_similarity_analysis['total_chunks_used'] = $total_chunks_used;
6046 -
6047 - // NEW: Store unique valid URLs for validation
6048 - $this->current_valid_urls = array_unique($valid_urls);
6049 -
6050 - // Allow add-ons to act on similarity results (e.g. WooCommerce product card display)
6051 - do_action('mxchat_similarity_results', $this->last_similarity_analysis['top_matches'], $bot_id);
6052 -
3790 +
3791 + //error_log("MxChat Testing: Stored " . count($this->last_similarity_analysis['top_matches']) . " Pinecone matches for testing");
3792 +
6053 3793 // Add response guidelines
6054 3794 if ($matches_used === 0) {
6055 3795 $content = "No reference information was found for this query.\n\n";
6056 3796 } else {
6057 - // Build response guidelines based on citation links setting
6058 - $content .= "\n## Response Guidelines ##\n" .
6059 - "You are an AI Chatbot. Answer naturally and helpfully using only the information from the references above. " .
6060 - "Be conversational and friendly, but never mention your knowledge base or training data. " .
6061 - "If you don't have specific information or are uncertain about any details, it's always " .
6062 - "better to honestly say you don't know rather than making up or guessing at answers. " .
6063 - "When information is incomplete, let them know you are unsure.\n\n";
6064 -
6065 - // Only add hyperlink instructions if citation links are enabled
6066 - if ($citation_links_enabled) {
6067 - $content .= "CRITICAL: When creating hyperlinks, always use proper markdown format with descriptive text: " .
6068 - "[descriptive text](url). NEVER use empty brackets like [](url). The text in brackets must describe what the link is about. " .
6069 - "Only cite references that have a URL. Do not cite or add source labels to Information sections that have no URL.";
6070 - } else {
6071 - $content .= "IMPORTANT: Do not include any citation links, source URLs, or hyperlinks in your responses. " .
6072 - "Simply provide helpful answers based on the reference information without citing sources.";
6073 - }
3797 + $content .= "\n## Response Guidelines ##\n" .
3798 + "You are an AI Chatbot. Answer naturally and helpfully using only the information from the references above. " .
3799 + "Be conversational and friendly, but never mention your knowledge base or training data. " .
3800 + "If you don't have specific information or are uncertain about any details, it's always " .
3801 + "better to honestly say you don't know rather than making up or guessing at answers. " .
3802 + "When information is incomplete, let them know you are unsure.";
6074 3803 }
6075 -
3804 +
6076 3805 return trim($content);
6077 3806 }
6078 3807
6079 3808 /**
@@ -6113,518 +3842,12 @@
6113 3842 }
6114 3843
6115 3844 // Cache individual role for 1 hour
6116 3845 wp_cache_set($cache_key, $role_restriction, 'mxchat_vector_roles', 3600);
6117 -
3846 +
6118 3847 return $role_restriction;
6119 3848 }
6120 3849
6121 -/**
6122 - * Fetch and reassemble all chunks for a URL from Pinecone
6123 - *
6124 - * @param string $source_url The source URL to fetch chunks for
6125 - * @param array $bot_config Bot-specific Pinecone configuration
6126 - * @return string Reassembled content from all chunks
6127 - */
6128 -private function reassemble_chunks_from_pinecone($source_url, $bot_config, $max_chunks = 0, &$chunk_count = 0) {
6129 - $api_key = $bot_config['api_key'] ?? '';
6130 - $host = $bot_config['host'] ?? '';
6131 - $namespace = $bot_config['namespace'] ?? '';
6132 -
6133 - if (empty($host) || empty($api_key)) {
6134 - $chunk_count = 0;
6135 - return '';
6136 - }
6137 -
6138 - $base_hash = md5($source_url);
6139 -
6140 - // Use Pinecone list API to find all chunk vectors with this prefix
6141 - $list_url = "https://{$host}/vectors/list";
6142 -
6143 - // Limit to max_chunks if specified, otherwise fetch up to 100
6144 - $fetch_limit = ($max_chunks > 0 && $max_chunks < 100) ? $max_chunks : 100;
6145 -
6146 - $list_body = array(
6147 - 'prefix' => $base_hash . '_chunk_',
6148 - 'limit' => $fetch_limit
6149 - );
6150 -
6151 - if (!empty($namespace)) {
6152 - $list_body['namespace'] = $namespace;
6153 - }
6154 -
6155 - $list_response = wp_remote_post($list_url, array(
6156 - 'headers' => array(
6157 - 'Api-Key' => $api_key,
6158 - 'accept' => 'application/json',
6159 - 'content-type' => 'application/json'
6160 - ),
6161 - 'body' => wp_json_encode($list_body),
6162 - 'timeout' => 30
6163 - ));
6164 -
6165 - if (is_wp_error($list_response)) {
6166 - //error_log('[MXCHAT-CHUNK] List API error: ' . $list_response->get_error_message());
6167 - return '';
6168 - }
6169 -
6170 - $list_data = json_decode(wp_remote_retrieve_body($list_response), true);
6171 -
6172 - if (empty($list_data['vectors'])) {
6173 - //error_log('[MXCHAT-CHUNK] No chunk vectors found for URL: ' . $source_url);
6174 - return '';
6175 - }
6176 -
6177 - // Extract vector IDs
6178 - $vector_ids = array();
6179 - foreach ($list_data['vectors'] as $vector) {
6180 - if (isset($vector['id'])) {
6181 - $vector_ids[] = $vector['id'];
6182 - }
6183 - }
6184 -
6185 - if (empty($vector_ids)) {
6186 - return '';
6187 - }
6188 -
6189 - // Fetch all chunk content
6190 - $fetch_url = "https://{$host}/vectors/fetch";
6191 -
6192 - $fetch_body = array(
6193 - 'ids' => $vector_ids
6194 - );
6195 -
6196 - if (!empty($namespace)) {
6197 - $fetch_body['namespace'] = $namespace;
6198 - }
6199 -
6200 - $fetch_response = wp_remote_post($fetch_url, array(
6201 - 'headers' => array(
6202 - 'Api-Key' => $api_key,
6203 - 'accept' => 'application/json',
6204 - 'content-type' => 'application/json'
6205 - ),
6206 - 'body' => wp_json_encode($fetch_body),
6207 - 'timeout' => 30
6208 - ));
6209 -
6210 - if (is_wp_error($fetch_response)) {
6211 - //error_log('[MXCHAT-CHUNK] Fetch API error: ' . $fetch_response->get_error_message());
6212 - return '';
6213 - }
6214 -
6215 - $fetch_data = json_decode(wp_remote_retrieve_body($fetch_response), true);
6216 -
6217 - if (empty($fetch_data['vectors'])) {
6218 - return '';
6219 - }
6220 -
6221 - // Sort chunks by index and reassemble
6222 - $chunks = array();
6223 - foreach ($fetch_data['vectors'] as $id => $vector) {
6224 - $metadata = $vector['metadata'] ?? array();
6225 - $chunk_index = $metadata['chunk_index'] ?? 0;
6226 - $text = $metadata['text'] ?? '';
6227 -
6228 - // Store chunk with its index
6229 - $chunks[$chunk_index] = $text;
6230 - }
6231 -
6232 - // Sort by chunk index
6233 - ksort($chunks);
6234 -
6235 - // Apply chunk limit if specified
6236 - if ($max_chunks > 0 && count($chunks) > $max_chunks) {
6237 - $chunks = array_slice($chunks, 0, $max_chunks, true);
6238 - }
6239 -
6240 - // Store actual chunk count
6241 - $chunk_count = count($chunks);
6242 -
6243 - // Reassemble content
6244 - return implode("\n\n", $chunks);
6245 -}
6246 -
6247 -/**
6248 - * Search for relevant content using OpenAI Vector Store (File Search)
6249 - *
6250 - * @param string $user_query The user's query text
6251 - * @param string $bot_id The bot ID
6252 - * @param array $vectorstore_config Vector Store configuration
6253 - * @return string Formatted context string with references
6254 - */
6255 -private function find_relevant_content_openai_vectorstore($user_query, $bot_id = 'default', $vectorstore_config = array()) {
6256 - //error_log("MXCHAT DEBUG: find_relevant_content_openai_vectorstore called");
6257 - //error_log(" - bot_id: " . $bot_id);
6258 - //error_log(" - user_query length: " . strlen($user_query));
6259 -
6260 - // Get OpenAI API key
6261 - $mxchat_options = get_option('mxchat_options', array());
6262 - $api_key = $mxchat_options['api_key'] ?? '';
6263 -
6264 - // Reset vectorstore error tracking
6265 - $this->last_vectorstore_error = null;
6266 -
6267 - if (empty($api_key)) {
6268 - //error_log("MXCHAT DEBUG ERROR: OpenAI API key not configured");
6269 - $this->last_vectorstore_error = 'Vector Store search failed: OpenAI API key is not configured.';
6270 - $this->current_valid_urls = [];
6271 - return '';
6272 - }
6273 -
6274 - // Get Vector Store configuration
6275 - if (empty($vectorstore_config)) {
6276 - $vectorstore_config = $this->get_bot_vectorstore_config($bot_id);
6277 - }
6278 -
6279 - $vectorstore_ids_string = $vectorstore_config['vectorstore_ids'] ?? '';
6280 - $max_results = $vectorstore_config['max_results'] ?? 5;
6281 -
6282 - if (empty($vectorstore_ids_string)) {
6283 - //error_log("MXCHAT DEBUG ERROR: No Vector Store IDs configured");
6284 - $this->last_vectorstore_error = 'Vector Store search failed: No Vector Store IDs are configured for this bot.';
6285 - $this->current_valid_urls = [];
6286 - return '';
6287 - }
6288 -
6289 - // Parse Vector Store IDs
6290 - $vectorstore_ids = array_map('trim', explode(',', $vectorstore_ids_string));
6291 - $vectorstore_ids = array_filter($vectorstore_ids); // Remove empty values
6292 -
6293 - //error_log("MXCHAT DEBUG: Vector Store IDs: " . implode(', ', $vectorstore_ids));
6294 - //error_log("MXCHAT DEBUG: Max results: " . $max_results);
6295 -
6296 - // Initialize similarity analysis storage
6297 - $this->last_similarity_analysis = [
6298 - 'knowledge_base_type' => 'OpenAI Vector Store',
6299 - 'bot_id' => $bot_id,
6300 - 'vectorstore_ids' => $vectorstore_ids,
6301 - 'top_matches' => [],
6302 - 'threshold_used' => 0,
6303 - 'total_checked' => 0
6304 - ];
6305 -
6306 - $valid_urls = [];
6307 -
6308 - // Get the selected model
6309 - $bot_options = $this->get_bot_options($bot_id);
6310 - $current_options = !empty($bot_options) ? $bot_options : $mxchat_options;
6311 - $selected_model = $current_options['model'] ?? 'gpt-5.1-chat-latest';
6312 -
6313 - // Verify it's an OpenAI model
6314 - if (!$this->is_openai_chat_model($selected_model)) {
6315 - //error_log("MXCHAT DEBUG ERROR: Vector Store search requires OpenAI model. Current: " . $selected_model);
6316 - $this->last_vectorstore_error = 'Vector Store search requires an OpenAI model. Current model: ' . $selected_model;
6317 - $this->current_valid_urls = [];
6318 - return '';
6319 - }
6320 -
6321 - // Use OpenAI Responses API with file_search tool
6322 - $request_body = array(
6323 - 'model' => $selected_model,
6324 - 'input' => $user_query,
6325 - 'tools' => array(
6326 - array(
6327 - 'type' => 'file_search',
6328 - 'vector_store_ids' => $vectorstore_ids,
6329 - 'max_num_results' => intval($max_results)
6330 - )
6331 - ),
6332 - 'include' => array('output[*].file_search_call.search_results')
6333 - );
6334 -
6335 - //error_log("MXCHAT VECTORSTORE: ========== REQUEST START ==========");
6336 - //error_log("MXCHAT VECTORSTORE: Model: " . $selected_model);
6337 - //error_log("MXCHAT VECTORSTORE: Query: " . substr($user_query, 0, 200));
6338 - //error_log("MXCHAT VECTORSTORE: Vector Store IDs: " . implode(', ', $vectorstore_ids));
6339 - //error_log("MXCHAT VECTORSTORE: Max Results: " . $max_results);
6340 - //error_log("MXCHAT VECTORSTORE: Request body: " . wp_json_encode($request_body));
6341 -
6342 - $response = wp_remote_post('https://api.openai.com/v1/responses', array(
6343 - 'headers' => array(
6344 - 'Authorization' => 'Bearer ' . $api_key,
6345 - 'Content-Type' => 'application/json'
6346 - ),
6347 - 'body' => wp_json_encode($request_body),
6348 - 'timeout' => 60
6349 - ));
6350 -
6351 - if (is_wp_error($response)) {
6352 - //error_log("MXCHAT VECTORSTORE ERROR: WP Error: " . $response->get_error_message());
6353 - $this->last_vectorstore_error = 'Vector Store API request failed: ' . $response->get_error_message();
6354 - $this->current_valid_urls = [];
6355 - return '';
6356 - }
6357 -
6358 - $response_code = wp_remote_retrieve_response_code($response);
6359 - //error_log("MXCHAT VECTORSTORE: Response code: " . $response_code);
6360 -
6361 - $response_body = wp_remote_retrieve_body($response);
6362 - //error_log("MXCHAT VECTORSTORE: Raw response (first 2000 chars): " . substr($response_body, 0, 2000));
6363 -
6364 - if ($response_code !== 200) {
6365 - //error_log("MXCHAT VECTORSTORE ERROR: API error response: " . $response_body);
6366 - $api_error_detail = '';
6367 - $decoded_error = json_decode($response_body, true);
6368 - if (isset($decoded_error['error']['message'])) {
6369 - $api_error_detail = $decoded_error['error']['message'];
6370 - }
6371 - $this->last_vectorstore_error = 'Vector Store API returned HTTP ' . $response_code . ($api_error_detail ? ': ' . $api_error_detail : '');
6372 - $this->current_valid_urls = [];
6373 - return '';
6374 - }
6375 - $result = json_decode($response_body, true);
6376 -
6377 - if (json_last_error() !== JSON_ERROR_NONE) {
6378 - //error_log("MXCHAT VECTORSTORE ERROR: JSON decode error: " . json_last_error_msg());
6379 - $this->last_vectorstore_error = 'Vector Store response could not be parsed: ' . json_last_error_msg();
6380 - $this->current_valid_urls = [];
6381 - return '';
6382 - }
6383 -
6384 - // Debug: Log the structure of the result
6385 - //error_log("MXCHAT VECTORSTORE: Result keys: " . implode(', ', array_keys($result)));
6386 - if (isset($result['output'])) {
6387 - //error_log("MXCHAT VECTORSTORE: Output count: " . count($result['output']));
6388 - foreach ($result['output'] as $idx => $out) {
6389 - //error_log("MXCHAT VECTORSTORE: Output[$idx] type: " . ($out['type'] ?? 'unknown'));
6390 - //error_log("MXCHAT VECTORSTORE: Output[$idx] keys: " . implode(', ', array_keys($out)));
6391 - }
6392 - } else {
6393 - //error_log("MXCHAT VECTORSTORE: No 'output' key in result!");
6394 - }
6395 -
6396 - // Extract file search results from the response
6397 - $content = '';
6398 - $matches_used = 0;
6399 - $all_matches = [];
6400 -
6401 - // The Responses API returns output array with tool results
6402 - if (isset($result['output']) && is_array($result['output'])) {
6403 - foreach ($result['output'] as $output_item) {
6404 - // Look for file_search_call results
6405 - if (isset($output_item['type']) && $output_item['type'] === 'file_search_call') {
6406 - //error_log("MXCHAT VECTORSTORE: Found file_search_call output item");
6407 - //error_log("MXCHAT VECTORSTORE: file_search_call keys: " . implode(', ', array_keys($output_item)));
6408 -
6409 - // Check for search_results in the output item directly
6410 - $search_results = $output_item['search_results'] ?? $output_item['results'] ?? [];
6411 - //error_log("MXCHAT VECTORSTORE: Search results count: " . count($search_results));
6412 -
6413 - if (empty($search_results)) {
6414 - //error_log("MXCHAT VECTORSTORE: No search results found in file_search_call");
6415 - //error_log("MXCHAT VECTORSTORE: file_search_call content: " . wp_json_encode($output_item));
6416 - }
6417 -
6418 - foreach ($search_results as $index => $search_result) {
6419 - $filename = $search_result['filename'] ?? '';
6420 - $score = $search_result['score'] ?? 0;
6421 - $text_content = '';
6422 -
6423 - // Extract text content from the result
6424 - // The text can be directly on the result OR nested under content array
6425 - if (isset($search_result['text']) && !empty($search_result['text'])) {
6426 - // Direct text field (OpenAI's actual format)
6427 - $text_content = $search_result['text'];
6428 - //error_log("MXCHAT VECTORSTORE: Found text directly on result[$index], length: " . strlen($text_content));
6429 - } elseif (isset($search_result['content']) && is_array($search_result['content'])) {
6430 - // Nested content array format
6431 - foreach ($search_result['content'] as $content_item) {
6432 - if (isset($content_item['text'])) {
6433 - $text_content .= $content_item['text'] . "\n";
6434 - }
6435 - }
6436 - //error_log("MXCHAT VECTORSTORE: Found text in content array for result[$index], length: " . strlen($text_content));
6437 - } else {
6438 - //error_log("MXCHAT VECTORSTORE: No text found for result[$index]. Keys: " . implode(', ', array_keys($search_result)));
6439 - }
6440 -
6441 - if (!empty($text_content)) {
6442 - $content .= "## Reference " . ($matches_used + 1) . " ##\n";
6443 - $content .= trim($text_content) . "\n\n";
6444 -
6445 - if (!empty($filename)) {
6446 - $content .= "Source: " . $filename . "\n\n";
6447 - }
6448 -
6449 - // Extract URLs from content
6450 - preg_match_all(
6451 - '#\bhttps?://[^\s<>"\']+#i',
6452 - $text_content,
6453 - $content_urls
6454 - );
6455 - if (!empty($content_urls[0])) {
6456 - $valid_urls = array_merge($valid_urls, $content_urls[0]);
6457 - }
6458 -
6459 - $matches_used++;
6460 - }
6461 -
6462 - // Store for similarity analysis
6463 - $all_matches[] = [
6464 - 'document_id' => $filename ?: ('result_' . $index),
6465 - 'similarity' => $score,
6466 - 'similarity_percentage' => round($score * 100, 2),
6467 - 'above_threshold' => true,
6468 - 'source_display' => $filename,
6469 - 'content_preview' => substr(strip_tags($text_content), 0, 100) . '...',
6470 - 'used_for_context' => true,
6471 - 'role_restriction' => 'public',
6472 - 'has_access' => true,
6473 - 'filtered_out' => false
6474 - ];
6475 - }
6476 - }
6477 -
6478 - // Also check for message content with annotations (citations)
6479 - if (isset($output_item['type']) && $output_item['type'] === 'message') {
6480 - if (isset($output_item['content']) && is_array($output_item['content'])) {
6481 - foreach ($output_item['content'] as $content_block) {
6482 - if (isset($content_block['annotations']) && is_array($content_block['annotations'])) {
6483 - foreach ($content_block['annotations'] as $annotation) {
6484 - if (isset($annotation['filename'])) {
6485 - $filename = $annotation['filename'];
6486 - $score = $annotation['score'] ?? 0;
6487 - $text_content = '';
6488 -
6489 - if (isset($annotation['content']) && is_array($annotation['content'])) {
6490 - foreach ($annotation['content'] as $ann_content) {
6491 - if (isset($ann_content['text'])) {
6492 - $text_content .= $ann_content['text'] . "\n";
6493 - }
6494 - }
6495 - }
6496 -
6497 - if (!empty($text_content) && $matches_used < $max_results) {
6498 - $content .= "## Reference " . ($matches_used + 1) . " ##\n";
6499 - $content .= trim($text_content) . "\n\n";
6500 - $content .= "Source: " . $filename . "\n\n";
6501 -
6502 - preg_match_all(
6503 - '#\bhttps?://[^\s<>"\']+#i',
6504 - $text_content,
6505 - $content_urls
6506 - );
6507 - if (!empty($content_urls[0])) {
6508 - $valid_urls = array_merge($valid_urls, $content_urls[0]);
6509 - }
6510 -
6511 - $matches_used++;
6512 -
6513 - $all_matches[] = [
6514 - 'document_id' => $filename,
6515 - 'similarity' => $score,
6516 - 'similarity_percentage' => round($score * 100, 2),
6517 - 'above_threshold' => true,
6518 - 'source_display' => $filename,
6519 - 'content_preview' => substr(strip_tags($text_content), 0, 100) . '...',
6520 - 'used_for_context' => true,
6521 - 'role_restriction' => 'public',
6522 - 'has_access' => true,
6523 - 'filtered_out' => false
6524 - ];
6525 - }
6526 - }
6527 - }
6528 - }
6529 - }
6530 - }
6531 - }
6532 - }
6533 - }
6534 -
6535 - // Store for testing panel
6536 - $this->last_similarity_analysis['top_matches'] = $all_matches;
6537 - $this->last_similarity_analysis['total_checked'] = count($all_matches);
6538 -
6539 - // Store unique valid URLs for validation
6540 - $this->current_valid_urls = array_unique($valid_urls);
6541 -
6542 - // Allow add-ons to act on similarity results (e.g. WooCommerce product card display)
6543 - do_action('mxchat_similarity_results', $this->last_similarity_analysis['top_matches'], $bot_id);
6544 -
6545 - //error_log("MXCHAT VECTORSTORE: ========== SEARCH COMPLETE ==========");
6546 - //error_log("MXCHAT VECTORSTORE: Matches used: " . $matches_used);
6547 - //error_log("MXCHAT VECTORSTORE: All matches count: " . count($all_matches));
6548 - //error_log("MXCHAT VECTORSTORE: Content length: " . strlen($content));
6549 - if ($matches_used > 0) {
6550 - //error_log("MXCHAT VECTORSTORE: Content preview: " . substr($content, 0, 500));
6551 - }
6552 -
6553 - // Check if citation links are enabled
6554 - $citation_links_enabled = ($mxchat_options['citation_links_toggle'] ?? 'on') === 'on';
6555 -
6556 - // Add response guidelines
6557 - if ($matches_used === 0) {
6558 - //error_log("MXCHAT VECTORSTORE: No matches found - returning empty reference message");
6559 - $content = "No reference information was found for this query.\n\n";
6560 - } else {
6561 - // Build response guidelines based on citation links setting
6562 - $content .= "\n## Response Guidelines ##\n" .
6563 - "You are an AI Chatbot. Answer naturally and helpfully using only the information from the references above. " .
6564 - "Be conversational and friendly, but never mention your knowledge base or training data. " .
6565 - "If you don't have specific information or are uncertain about any details, it's always " .
6566 - "better to honestly say you don't know rather than making up or guessing at answers. " .
6567 - "When information is incomplete, let them know you are unsure.\n\n";
6568 -
6569 - // Only add hyperlink instructions if citation links are enabled
6570 - if ($citation_links_enabled) {
6571 - $content .= "CRITICAL: When creating hyperlinks, always use proper markdown format with descriptive text: " .
6572 - "[descriptive text](url). NEVER use empty brackets like [](url). The text in brackets must describe what the link is about.";
6573 - } else {
6574 - $content .= "IMPORTANT: Do not include any citation links, source URLs, or hyperlinks in your responses. " .
6575 - "Simply provide helpful answers based on the reference information without citing sources.";
6576 - }
6577 - }
6578 -
6579 - //error_log("MXCHAT DEBUG: Vector Store search complete. Matches used: " . $matches_used);
6580 -
6581 - return trim($content);
6582 -}
6583 -
6584 -/**
6585 - * Check if the given model is an OpenAI chat model
6586 - *
6587 - * @param string $model The model ID
6588 - * @return bool True if it's an OpenAI model
6589 - */
6590 -private function is_openai_chat_model($model) {
6591 - $openai_prefixes = array('gpt-', 'o1-', 'o3-');
6592 - foreach ($openai_prefixes as $prefix) {
6593 - if (strpos($model, $prefix) === 0) {
6594 - return true;
6595 - }
6596 - }
6597 - return false;
6598 -}
6599 -
6600 -/**
6601 - * Get bot-specific Vector Store configuration
6602 - *
6603 - * @param string $bot_id The bot ID
6604 - * @return array Configuration array
6605 - */
6606 -private function get_bot_vectorstore_config($bot_id = 'default') {
6607 - $vectorstore_options = get_option('mxchat_openai_vectorstore_options', array());
6608 -
6609 - // Default global settings
6610 - $default_config = array(
6611 - 'use_vectorstore' => ($vectorstore_options['mxchat_use_openai_vectorstore'] ?? '0') === '1',
6612 - 'vectorstore_ids' => $vectorstore_options['mxchat_vectorstore_ids'] ?? '',
6613 - 'max_results' => $vectorstore_options['mxchat_vectorstore_max_results'] ?? 5
6614 - );
6615 -
6616 - // Allow multi-bot plugin to override with bot-specific settings
6617 - $bot_config = apply_filters('mxchat_get_bot_vectorstore_config', $default_config, $bot_id);
6618 -
6619 - // Preserve max_results from global settings if not set in bot config
6620 - if (!isset($bot_config['max_results'])) {
6621 - $bot_config['max_results'] = $default_config['max_results'];
6622 - }
6623 -
6624 - return $bot_config;
6625 -}
6626 -
6627 3850 private function mxchat_find_relevant_products($user_embedding) {
6628 3851 //error_log('MXChat Vector Search: Starting product search...');
6629 3852
6630 3853 // Retrieve the add-on settings from the database
@@ -6645,75 +3868,73 @@
6645 3868 }
6646 3869 private function find_relevant_products_wordpress($user_embedding) {
6647 3870 global $wpdb;
6648 3871 $system_prompt_table = $wpdb->prefix . 'mxchat_system_prompt_content';
3872 + $cache_key = 'mxchat_system_prompt_embeddings';
3873 + $batch_size = 500;
6649 3874
6650 - if (!is_array($user_embedding)) {
6651 - return '';
6652 - }
3875 + // Original WordPress database search logic
3876 + // [Previous implementation remains the same]
3877 + $embeddings = wp_cache_get($cache_key, 'mxchat_system_prompts');
3878 + if ($embeddings === false) {
3879 + $embeddings = [];
3880 + $offset = 0;
6653 3881
6654 - // Streaming top-K pass: scan rows in small batches, keep only the top 3
6655 - // results above the similarity threshold. Peak memory is bounded by
6656 - // $batch_size embedding rows plus a 3-element top list.
6657 - $batch_size = 250;
6658 - $similarity_threshold = 0.85;
6659 - $top_k = 3;
6660 - $top_results = [];
6661 - $offset = 0;
3882 + do {
3883 + $query = $wpdb->prepare(
3884 + "SELECT id, embedding_vector
3885 + FROM {$system_prompt_table}
3886 + LIMIT %d OFFSET %d",
3887 + $batch_size,
3888 + $offset
3889 + );
6662 3890
6663 - do {
6664 - $batch = $wpdb->get_results($wpdb->prepare(
6665 - "SELECT id, embedding_vector
6666 - FROM {$system_prompt_table}
6667 - LIMIT %d OFFSET %d",
6668 - $batch_size,
6669 - $offset
6670 - ));
3891 + $batch = $wpdb->get_results($query);
3892 + if (empty($batch)) {
3893 + break;
3894 + }
6671 3895
6672 - if (empty($batch)) {
6673 - break;
6674 - }
3896 + $embeddings = array_merge($embeddings, $batch);
3897 + $offset += $batch_size;
6675 3898
6676 - foreach ($batch as $row) {
6677 - $database_embedding = $row->embedding_vector
6678 - ? unserialize($row->embedding_vector, ['allowed_classes' => false])
6679 - : null;
3899 + unset($batch);
6680 3900
6681 - if (!is_array($database_embedding)) {
6682 - unset($database_embedding);
6683 - continue;
6684 - }
3901 + } while (true);
6685 3902
3903 + if (empty($embeddings)) {
3904 + return '';
3905 + }
3906 + wp_cache_set($cache_key, $embeddings, 'mxchat_system_prompts', 3600);
3907 + }
3908 +
3909 + $relevant_results = [];
3910 + foreach ($embeddings as $embedding) {
3911 + $database_embedding = $embedding->embedding_vector
3912 + ? unserialize($embedding->embedding_vector, ['allowed_classes' => false])
3913 + : null;
3914 + if (is_array($database_embedding) && is_array($user_embedding)) {
6686 3915 $similarity = $this->mxchat_calculate_cosine_similarity($user_embedding, $database_embedding);
6687 - unset($database_embedding);
6688 -
6689 - if ($similarity < $similarity_threshold) {
6690 - continue;
6691 - }
6692 -
6693 - // Insert into bounded top-K (kept sorted descending)
6694 - if (count($top_results) < $top_k) {
6695 - $top_results[] = ['id' => $row->id, 'similarity' => $similarity];
6696 - usort($top_results, function ($a, $b) {
6697 - return $b['similarity'] <=> $a['similarity'];
6698 - });
6699 - } elseif ($similarity > $top_results[$top_k - 1]['similarity']) {
6700 - $top_results[$top_k - 1] = ['id' => $row->id, 'similarity' => $similarity];
6701 - usort($top_results, function ($a, $b) {
6702 - return $b['similarity'] <=> $a['similarity'];
6703 - });
6704 - }
3916 + $relevant_results[] = [
3917 + 'id' => $embedding->id,
3918 + 'similarity' => $similarity
3919 + ];
6705 3920 }
3921 + unset($database_embedding);
3922 + }
6706 3923
6707 - unset($batch);
6708 - $offset += $batch_size;
6709 - } while (true);
3924 + // Use fixed threshold for products
3925 + $similarity_threshold = 0.85;
6710 3926
6711 - if (empty($top_results)) {
6712 - return '';
6713 - }
3927 + $relevant_results = array_filter($relevant_results, function ($result) use ($similarity_threshold) {
3928 + return $result['similarity'] >= $similarity_threshold;
3929 + });
3930 + usort($relevant_results, function ($a, $b) {
3931 + return $b['similarity'] <=> $a['similarity'];
3932 + });
6714 3933
3934 + $top_results = array_slice($relevant_results, 0, 5);
6715 3935 $content = '';
3936 +
6716 3937 foreach ($top_results as $result) {
6717 3938 $chunk_content = $this->fetch_content_with_product_links($result['id']);
6718 3939 $content .= $chunk_content . "\n\n";
6719 3940 }
@@ -6719,10 +3940,8 @@
6719 3940 }
6720 3941
6721 3942 return trim($content);
6722 3943 }
6723 -
6724 -
6725 3944 private function find_relevant_products_pinecone($user_embedding) {
6726 3945 //error_log('Starting Pinecone product search...');
6727 3946
6728 3947 $options = get_option('mxchat_pinecone_addon_options', array());
@@ -6797,10 +4016,8 @@
6797 4016 }
6798 4017
6799 4018 return trim($content);
6800 4019 }
6801 -
6802 -
6803 4020 private function fetch_content_with_product_links($most_relevant_id) {
6804 4021 global $wpdb;
6805 4022 $system_prompt_table = $wpdb->prefix . 'mxchat_system_prompt_content';
6806 4023
@@ -6820,85 +4037,12 @@
6820 4037 return null;
6821 4038 }
6822 4039
6823 4040 /**
6824 - * Get system instructions for a specific bot or default
6825 - * Checks for multi-bot add-on and uses bot-specific instructions if available
6826 - * Automatically strips URLs if citation links are disabled
6827 - * Replaces {visitor_name} placeholder with actual visitor name if available
6828 - *
6829 - * @param string $bot_id The bot ID to get instructions for
6830 - * @param string $session_id Optional session ID to lookup visitor name
4041 + * Modified streaming functions to include testing data
6831 4042 */
6832 -private function get_system_instructions($bot_id = 'default', $session_id = '') {
6833 - $instructions = '';
6834 4043
6835 - // Check if multi-bot add-on is active
6836 - if (class_exists('MxChat_Multi_Bot_Core_Manager') && $bot_id !== 'default') {
6837 - // Get bot-specific options from multi-bot add-on
6838 - $bot_options = apply_filters('mxchat_get_bot_options', array(), $bot_id);
6839 -
6840 - // If bot has custom system instructions, use those
6841 - if (!empty($bot_options['system_prompt_instructions'])) {
6842 - $instructions = $bot_options['system_prompt_instructions'];
6843 - }
6844 - }
6845 -
6846 - // Fall back to default system instructions
6847 - if (empty($instructions)) {
6848 - $instructions = isset($this->options['system_prompt_instructions']) ? $this->options['system_prompt_instructions'] : '';
6849 - }
6850 -
6851 - // Check if citation links are disabled - if so, strip URLs from instructions
6852 - $fresh_options = get_option('mxchat_options', []);
6853 - $citation_links_enabled = isset($fresh_options['citation_links_toggle']) ? ($fresh_options['citation_links_toggle'] === 'on') : true;
6854 -
6855 - if (!$citation_links_enabled && !empty($instructions)) {
6856 - $instructions = preg_replace('#\bhttps?://[^\s<>"\']+#i', '', $instructions);
6857 - $instructions = preg_replace('/\s+/', ' ', trim($instructions)); // Clean up extra spaces
6858 - }
6859 -
6860 - // Replace {visitor_name} placeholder with actual visitor name if available
6861 - if (!empty($instructions) && !empty($session_id) && stripos($instructions, '{visitor_name}') !== false) {
6862 - $name_option_key = "mxchat_name_{$session_id}";
6863 - $visitor_name = get_option($name_option_key, '');
6864 -
6865 - if (!empty($visitor_name)) {
6866 - $instructions = str_ireplace('{visitor_name}', sanitize_text_field($visitor_name), $instructions);
6867 - } else {
6868 - // Remove placeholder if no name is available
6869 - $instructions = str_ireplace('{visitor_name}', '', $instructions);
6870 - $instructions = preg_replace('/\s{2,}/', ' ', trim($instructions)); // Clean up extra spaces
6871 - }
6872 - }
6873 -
6874 - // Allow developers to filter system instructions and process shortcodes
6875 - $instructions = apply_filters('mxchat_system_instructions', $instructions, $bot_id, $session_id);
6876 - $instructions = do_shortcode($instructions);
6877 -
6878 - return $instructions;
6879 -}
6880 -/**
6881 - * Get the current bot ID from session or request context
6882 - */
6883 -private function get_current_bot_id($session_id = '') {
6884 - // First, check if bot_id is passed in the current request
6885 - if (isset($_POST['bot_id']) && !empty($_POST['bot_id'])) {
6886 - return sanitize_key($_POST['bot_id']);
6887 - }
6888 -
6889 - // If not in POST, try to get it from session data
6890 - if (!empty($session_id)) {
6891 - $bot_id = get_option("mxchat_session_bot_{$session_id}", '');
6892 - if (!empty($bot_id)) {
6893 - return $bot_id;
6894 - }
6895 - }
6896 -
6897 - // Fall back to default
6898 - return 'default';
6899 -}
6900 -private function mxchat_generate_response($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, $selected_model = 'gpt-5.1-chat-latest') {
4044 +private function mxchat_generate_response($relevant_content, $api_key, $xai_api_key, $claude_api_key, $deepseek_api_key, $gemini_api_key, $conversation_history, $streaming = false, $session_id = '', $testing_data = null) {
6901 4045 try {
6902 4046 if (!$relevant_content) {
6903 4047 $error_response = [
6904 4048 'error' => esc_html__("I couldn't find relevant information on that topic.", 'mxchat'),
@@ -6904,74 +4048,25 @@
6904 4048 'error' => esc_html__("I couldn't find relevant information on that topic.", 'mxchat'),
6905 4049 'error_code' => 'no_relevant_content'
6906 4050 ];
6907 4051
4052 + // Add testing data to error response if available
6908 4053 if ($testing_data !== null) {
6909 4054 $error_response['testing_data'] = $testing_data;
4055 + //error_log("MxChat Testing: Added testing data to no_relevant_content error");
6910 4056 }
6911 4057
6912 4058 return $error_response;
6913 4059 }
6914 4060
4061 + // Ensure conversation_history is an array
6915 4062 if (!is_array($conversation_history)) {
6916 4063 $conversation_history = array();
6917 4064 }
6918 4065
6919 - // Check if this is an OpenRouter model
6920 - if ($selected_model === 'openrouter') {
6921 - // Get the actual OpenRouter model from options
6922 - $openrouter_selected_model = $this->options['openrouter_selected_model'] ?? '';
6923 -
6924 - if (empty($openrouter_selected_model)) {
6925 - $error_response = [
6926 - 'error' => esc_html__('No OpenRouter model selected. Please select a model in settings.', 'mxchat'),
6927 - 'error_code' => 'no_openrouter_model_selected'
6928 - ];
6929 - if ($testing_data !== null) {
6930 - $error_response['testing_data'] = $testing_data;
6931 - }
6932 - return $error_response;
6933 - }
6934 -
6935 - if (empty($openrouter_api_key)) {
6936 - $error_response = [
6937 - 'error' => esc_html__('OpenRouter API key is not configured', 'mxchat'),
6938 - 'error_code' => 'missing_openrouter_api_key'
6939 - ];
6940 - if ($testing_data !== null) {
6941 - $error_response['testing_data'] = $testing_data;
6942 - }
6943 - return $error_response;
6944 - }
6945 -
6946 - if ($streaming) {
6947 - return $this->mxchat_generate_response_openrouter_stream(
6948 - $openrouter_selected_model,
6949 - $openrouter_api_key,
6950 - $conversation_history,
6951 - $relevant_content,
6952 - $session_id,
6953 - $testing_data
6954 - );
6955 - } else {
6956 - $response = $this->mxchat_generate_response_openrouter(
6957 - $openrouter_selected_model,
6958 - $openrouter_api_key,
6959 - $conversation_history,
6960 - $relevant_content
6961 - );
6962 - }
6963 -
6964 - if (is_array($response) && isset($response['error'])) {
6965 - if ($testing_data !== null) {
6966 - $response['testing_data'] = $testing_data;
6967 - }
6968 - return $response;
6969 - }
6970 -
6971 - return $response;
6972 - }
6973 -
4066 + // Get selected model with default fallback
4067 + $selected_model = isset($this->options['model']) ? $this->options['model'] : 'gpt-4o';
4068 +
6974 4069 // Extract model prefix to determine the provider
6975 4070 $model_parts = explode('-', $selected_model);
6976 4071 $provider = strtolower($model_parts[0]);
6977 4072
@@ -7013,9 +4108,9 @@
7013 4108 $claude_api_key,
7014 4109 $conversation_history,
7015 4110 $relevant_content,
7016 4111 $session_id,
7017 - $testing_data
4112 + $testing_data // Pass testing data
7018 4113 );
7019 4114 } else {
7020 4115 $response = $this->mxchat_generate_response_claude(
7021 4116 $selected_model,
@@ -7043,9 +4138,9 @@
7043 4138 $xai_api_key,
7044 4139 $conversation_history,
7045 4140 $relevant_content,
7046 4141 $session_id,
7047 - $testing_data
4142 + $testing_data // Pass testing data
7048 4143 );
7049 4144 } else {
7050 4145 $response = $this->mxchat_generate_response_xai(
7051 4146 $selected_model,
@@ -7073,9 +4168,9 @@
7073 4168 $deepseek_api_key,
7074 4169 $conversation_history,
7075 4170 $relevant_content,
7076 4171 $session_id,
7077 - $testing_data
4172 + $testing_data // Pass testing data
7078 4173 );
7079 4174 } else {
7080 4175 $response = $this->mxchat_generate_response_deepseek(
7081 4176 $selected_model,
@@ -7085,38 +4180,8 @@
7085 4180 );
7086 4181 }
7087 4182 break;
7088 4183
7089 - case 'custom':
7090 - // Custom (OpenAI-compatible) provider — Ollama, LM Studio, vLLM, llama.cpp, Azure OpenAI
7091 - $cp_base_url = isset($this->options['custom_provider_base_url']) ? trim((string) $this->options['custom_provider_base_url']) : '';
7092 - if (empty($cp_base_url)) {
7093 - $error_response = [
7094 - 'error' => esc_html__('Custom provider is not configured. Set Base URL in MxChat → API Keys → Custom Provider.', 'mxchat'),
7095 - 'error_code' => 'missing_custom_provider_base_url'
7096 - ];
7097 - if ($testing_data !== null) {
7098 - $error_response['testing_data'] = $testing_data;
7099 - }
7100 - return $error_response;
7101 - }
7102 - if ($streaming) {
7103 - return $this->mxchat_generate_response_custom_stream(
7104 - $selected_model,
7105 - $conversation_history,
7106 - $relevant_content,
7107 - $session_id,
7108 - $testing_data
7109 - );
7110 - } else {
7111 - $response = $this->mxchat_generate_response_custom(
7112 - $selected_model,
7113 - $conversation_history,
7114 - $relevant_content
7115 - );
7116 - }
7117 - break;
7118 -
7119 4184 case 'gpt':
7120 4185 case 'o1':
7121 4186 if (empty($api_key)) {
7122 4187 $error_response = [
@@ -7127,27 +4192,9 @@
7127 4192 $error_response['testing_data'] = $testing_data;
7128 4193 }
7129 4194 return $error_response;
7130 4195 }
7131 -
7132 - // Check if web search is enabled for this OpenAI model
7133 - $web_search_enabled = isset($this->options['enable_web_search']) && $this->options['enable_web_search'] === 'on';
7134 - // Models that don't support web search
7135 - $unsupported_web_search_models = array('gpt-4.1-nano');
7136 - $model_supports_web_search = !in_array($selected_model, $unsupported_web_search_models);
7137 -
7138 - if ($web_search_enabled && $model_supports_web_search) {
7139 - // Use Responses API (required for some models, or when web search is enabled)
7140 - return $this->mxchat_generate_response_openai_web_search(
7141 - $selected_model,
7142 - $api_key,
7143 - $conversation_history,
7144 - $relevant_content,
7145 - $session_id,
7146 - $testing_data,
7147 - $streaming
7148 - );
7149 - } elseif ($streaming) {
4196 + if ($streaming) {
7150 4197 return $this->mxchat_generate_response_openai_stream(
7151 4198 $selected_model,
7152 4199 $api_key,
7153 4200 $conversation_history,
@@ -7152,9 +4199,9 @@
7152 4199 $api_key,
7153 4200 $conversation_history,
7154 4201 $relevant_content,
7155 4202 $session_id,
7156 - $testing_data
4203 + $testing_data // Pass testing data
7157 4204 );
7158 4205 } else {
7159 4206 $response = $this->mxchat_generate_response_openai(
7160 4207 $selected_model,
@@ -7165,8 +4212,9 @@
7165 4212 }
7166 4213 break;
7167 4214
7168 4215 default:
4216 + // Default to OpenAI for custom models or unrecognized prefixes
7169 4217 if (empty($api_key)) {
7170 4218 $error_response = [
7171 4219 'error' => esc_html__('OpenAI API key is not configured', 'mxchat'),
7172 4220 'error_code' => 'missing_openai_api_key'
@@ -7175,25 +4223,9 @@
7175 4223 $error_response['testing_data'] = $testing_data;
7176 4224 }
7177 4225 return $error_response;
7178 4226 }
7179 -
7180 - // Check if web search is enabled (default case also handles OpenAI models)
7181 - $web_search_enabled = isset($this->options['enable_web_search']) && $this->options['enable_web_search'] === 'on';
7182 - $unsupported_web_search_models = array('gpt-4.1-nano');
7183 - $model_supports_web_search = !in_array($selected_model, $unsupported_web_search_models);
7184 -
7185 - if ($web_search_enabled && $model_supports_web_search) {
7186 - return $this->mxchat_generate_response_openai_web_search(
7187 - $selected_model,
7188 - $api_key,
7189 - $conversation_history,
7190 - $relevant_content,
7191 - $session_id,
7192 - $testing_data,
7193 - $streaming
7194 - );
7195 - } elseif ($streaming) {
4227 + if ($streaming) {
7196 4228 return $this->mxchat_generate_response_openai_stream(
7197 4229 $selected_model,
7198 4230 $api_key,
7199 4231 $conversation_history,
@@ -7198,9 +4230,9 @@
7198 4230 $api_key,
7199 4231 $conversation_history,
7200 4232 $relevant_content,
7201 4233 $session_id,
7202 - $testing_data
4234 + $testing_data // Pass testing data
7203 4235 );
7204 4236 } else {
7205 4237 $response = $this->mxchat_generate_response_openai(
7206 4238 $selected_model,
@@ -7211,18 +4243,24 @@
7211 4243 }
7212 4244 break;
7213 4245 }
7214 4246
4247 + // Check if the response is an error array from the provider-specific function
7215 4248 if (is_array($response) && isset($response['error'])) {
4249 + // Add testing data to error response if available
7216 4250 if ($testing_data !== null) {
7217 4251 $response['testing_data'] = $testing_data;
4252 + //error_log("MxChat Testing: Added testing data to provider error response");
7218 4253 }
7219 - return $response;
4254 + return $response; // Pass through the error with testing data
7220 4255 }
7221 4256
4257 + // For successful non-streaming responses, we don't add testing data here
4258 + // because it will be added in the main handler
7222 4259 return $response;
7223 4260
7224 4261 } catch (Exception $e) {
4262 + //error_log('MXChat Error: ' . $e->getMessage());
7225 4263 $error_response = [
7226 4264 'error' => sprintf(esc_html__('An error occurred: %s', 'mxchat'), esc_html($e->getMessage())),
7227 4265 'error_code' => 'system_exception',
7228 4266 'exception_details' => $e->getMessage()
@@ -7227,24 +4265,29 @@
7227 4265 'error_code' => 'system_exception',
7228 4266 'exception_details' => $e->getMessage()
7229 4267 ];
7230 4268
4269 + // Add testing data to exception response if available
7231 4270 if ($testing_data !== null) {
7232 4271 $error_response['testing_data'] = $testing_data;
4272 + //error_log("MxChat Testing: Added testing data to exception response");
7233 4273 }
7234 4274
7235 4275 return $error_response;
7236 4276 }
7237 4277 }
7238 -private function mxchat_generate_response_openrouter_stream($selected_model, $openrouter_api_key, $conversation_history, $relevant_content, $session_id, $testing_data = null) {
4278 +
4279 +private function mxchat_generate_response_openai_stream($selected_model, $api_key, $conversation_history, $relevant_content, $session_id, $testing_data = null) {
7239 4280 try {
7240 - $bot_id = $this->get_current_bot_id($session_id);
7241 - $system_prompt_instructions = $this->get_system_instructions($bot_id, $session_id);
4281 + // Get system prompt instructions from options
4282 + $system_prompt_instructions = isset($this->options['system_prompt_instructions']) ? $this->options['system_prompt_instructions'] : '';
7242 4283
4284 + // Ensure conversation_history is an array
7243 4285 if (!is_array($conversation_history)) {
7244 4286 $conversation_history = array();
7245 4287 }
7246 4288
4289 + // Format conversation history for OpenAI
7247 4290 $formatted_conversation = array();
7248 4291
7249 4292 $formatted_conversation[] = array(
7250 4293 'role' => 'system',
@@ -7256,9 +4299,9 @@
7256 4299 $role = $message['role'];
7257 4300 if ($role === 'bot' || $role === 'agent') {
7258 4301 $role = 'assistant';
7259 4302 }
7260 - if (!in_array($role, ['system', 'assistant', 'user'])) {
4303 + if (!in_array($role, ['system', 'assistant', 'user', 'function', 'tool'])) {
7261 4304 $role = 'user';
7262 4305 }
7263 4306 $formatted_conversation[] = array(
7264 4307 'role' => $role,
@@ -7266,21 +4309,19 @@
7266 4309 );
7267 4310 }
7268 4311 }
7269 4312
4313 + // Check if we can actually stream
7270 4314 if (headers_sent() || !function_exists('curl_init')) {
7271 - $regular_response = $this->mxchat_generate_response_openrouter(
4315 + // Fallback to regular response with testing data
4316 + //error_log("MxChat: OpenAI streaming not possible, falling back to regular response");
4317 + $regular_response = $this->mxchat_generate_response_openai(
7272 4318 $selected_model,
7273 - $openrouter_api_key,
4319 + $api_key,
7274 4320 $conversation_history,
7275 4321 $relevant_content
7276 4322 );
7277 4323
7278 - // Save bot response to transcript
7279 - if (!empty($regular_response) && !empty($session_id)) {
7280 - $this->mxchat_save_chat_message($session_id, 'bot', $regular_response);
7281 - }
7282 -
7283 4324 $response_data = [
7284 4325 'text' => $regular_response,
7285 4326 'html' => '',
7286 4327 'session_id' => $session_id
@@ -7287,8 +4328,9 @@
7287 4328 ];
7288 4329
7289 4330 if ($testing_data !== null) {
7290 4331 $response_data['testing_data'] = $testing_data;
4332 + //error_log("MxChat Testing: Added testing data to OpenAI fallback response");
7291 4333 }
7292 4334
7293 4335 header('Content-Type: application/json');
7294 4336 echo json_encode($response_data);
@@ -7294,8 +4336,9 @@
7294 4336 echo json_encode($response_data);
7295 4337 return true;
7296 4338 }
7297 4339
4340 + // Prepare the request body with stream: true
7298 4341 $body = json_encode([
7299 4342 'model' => $selected_model,
7300 4343 'messages' => $formatted_conversation,
7301 4344 'temperature' => 1,
@@ -7301,213 +4344,71 @@
7301 4344 'temperature' => 1,
7302 4345 'stream' => true
7303 4346 ]);
7304 4347
7305 - // V2 retry-on-initial-connect: setup_streaming_headers is now lazy-fired
7306 - // inside WRITEFUNCTION on first byte of a successful upstream.
7307 -
7308 - $captured_status_code = 0;
7309 - $captured_body_pre_stream = '';
7310 - $full_response = '';
4348 + // Use cURL for streaming support
4349 + $ch = curl_init();
4350 + curl_setopt($ch, CURLOPT_URL, 'https://api.openai.com/v1/chat/completions');
4351 + curl_setopt($ch, CURLOPT_RETURNTRANSFER, false);
4352 + curl_setopt($ch, CURLOPT_POST, true);
4353 + curl_setopt($ch, CURLOPT_POSTFIELDS, $body);
4354 + curl_setopt($ch, CURLOPT_HTTPHEADER, array(
4355 + 'Content-Type: application/json',
4356 + 'Authorization: Bearer ' . $api_key
4357 + ));
4358 + curl_setopt($ch, CURLOPT_SSL_VERIFYPEER, true);
4359 + curl_setopt($ch, CURLOPT_TIMEOUT, 60);
4360 +
4361 + $full_response = ''; // Accumulate full response for saving
7311 4362 $stream_started = false;
7312 - $buffer = '';
7313 - $errno = 0;
7314 - $last_curl_error = '';
7315 - $http_code = 0;
7316 - $max_attempts = $this->mxchat_retry_enabled() ? 3 : 1;
7317 - $backoff_ms = array(0, 750, 2000);
7318 -
7319 - for ($attempt = 0; $attempt < $max_attempts; $attempt++) {
7320 - if ($attempt > 0 && $backoff_ms[$attempt] > 0) {
7321 - usleep($backoff_ms[$attempt] * 1000);
4363 +
4364 + // Buffer control for real-time streaming
4365 + curl_setopt($ch, CURLOPT_WRITEFUNCTION, function($ch, $data) use (&$full_response, &$stream_started, $testing_data) {
4366 + // Send testing data as the first event if available
4367 + if (!$stream_started && $testing_data !== null) {
4368 + echo "data: " . json_encode(['testing_data' => $testing_data]) . "\n\n";
4369 + flush();
4370 + $stream_started = true;
4371 + //error_log("MxChat Testing: Sent testing data in OpenAI stream");
7322 4372 }
7323 -
7324 - $captured_status_code = 0;
7325 - $captured_body_pre_stream = '';
7326 - $full_response = '';
7327 - $stream_started = false;
7328 - $buffer = '';
7329 -
7330 - $ch = curl_init();
7331 - curl_setopt($ch, CURLOPT_URL, 'https://openrouter.ai/api/v1/chat/completions');
7332 - curl_setopt($ch, CURLOPT_RETURNTRANSFER, false);
7333 - curl_setopt($ch, CURLOPT_POST, true);
7334 - curl_setopt($ch, CURLOPT_POSTFIELDS, $body);
7335 - curl_setopt($ch, CURLOPT_HTTPHEADER, array(
7336 - 'Content-Type: application/json',
7337 - 'Authorization: Bearer ' . $openrouter_api_key,
7338 - 'HTTP-Referer: ' . home_url(),
7339 - 'X-Title: ' . get_bloginfo('name')
7340 - ));
7341 - curl_setopt($ch, CURLOPT_SSL_VERIFYPEER, true);
7342 - curl_setopt($ch, CURLOPT_TIMEOUT, 60);
7343 -
7344 - curl_setopt($ch, CURLOPT_HEADERFUNCTION, function($ch, $header) use (&$captured_status_code) {
7345 - if ($captured_status_code === 0 && preg_match('#^HTTP/\S+\s+(\d+)\b#', $header, $m)) {
7346 - $captured_status_code = (int) $m[1];
4373 +
4374 + // Process each chunk of data
4375 + $lines = explode("\n", $data);
4376 +
4377 + foreach ($lines as $line) {
4378 + if (trim($line) === '' || strpos($line, 'data: ') !== 0) {
4379 + continue;
7347 4380 }
7348 - return strlen($header);
7349 - });
7350 -
7351 - curl_setopt($ch, CURLOPT_WRITEFUNCTION, function($ch, $data) use (&$full_response, &$stream_started, &$buffer, &$captured_status_code, &$captured_body_pre_stream, $testing_data) {
7352 - if ($captured_status_code !== 0 && $captured_status_code !== 200) {
7353 - $captured_body_pre_stream .= $data;
7354 - return strlen($data);
4381 +
4382 + $json_str = substr($line, 6); // Remove 'data: ' prefix
4383 +
4384 + if ($json_str === '[DONE]') {
4385 + echo "data: [DONE]\n\n";
4386 + flush();
4387 + continue;
7355 4388 }
7356 -
7357 - if (!$this->streaming_headers_sent) {
7358 - $this->setup_streaming_headers();
7359 - }
7360 -
7361 - if (!$stream_started && $testing_data !== null) {
7362 - echo "data: " . json_encode(['testing_data' => $testing_data]) . "\n\n";
4389 +
4390 + $json = json_decode($json_str, true);
4391 + if (isset($json['choices'][0]['delta']['content'])) {
4392 + $content = $json['choices'][0]['delta']['content'];
4393 + $full_response .= $content; // Accumulate
4394 + // Send as SSE format
4395 + echo "data: " . json_encode(['content' => $content]) . "\n\n";
7363 4396 flush();
7364 - $stream_started = true;
7365 4397 }
7366 -
7367 - $buffer .= $data;
7368 - $lines = explode("\n", $buffer);
7369 - $buffer = array_pop($lines);
7370 -
7371 - foreach ($lines as $line) {
7372 - if (trim($line) === '') {
7373 - continue;
7374 - }
7375 - if (strpos($line, 'data: ') !== 0) {
7376 - continue;
7377 - }
7378 -
7379 - $json_str = substr($line, 6);
7380 -
7381 - if (trim($json_str) === '[DONE]') {
7382 - echo "data: [DONE]\n\n";
7383 - flush();
7384 - continue;
7385 - }
7386 -
7387 - $json = json_decode(trim($json_str), true);
7388 - if ($json && isset($json['choices'][0]['delta']['content'])) {
7389 - $content = $json['choices'][0]['delta']['content'];
7390 - $full_response .= $content;
7391 -
7392 - echo "data: " . json_encode(['content' => $content]) . "\n\n";
7393 - flush();
7394 - }
7395 - }
7396 -
7397 - return strlen($data);
7398 - });
7399 -
7400 - $response = curl_exec($ch);
7401 - $errno = curl_errno($ch);
7402 - $last_curl_error = curl_error($ch);
7403 - $http_code = $captured_status_code !== 0 ? $captured_status_code : (int) curl_getinfo($ch, CURLINFO_HTTP_CODE);
7404 - curl_close($ch);
7405 -
7406 - if (!$errno && $http_code === 200) {
7407 - break;
7408 4398 }
7409 -
7410 - $is_transient = $this->mxchat_is_transient_provider_error_raw($http_code, $captured_body_pre_stream, 'openai', $errno);
7411 - $can_retry = !$this->streaming_headers_sent
7412 - && ($attempt + 1) < $max_attempts
7413 - && $is_transient;
7414 -
7415 - if (defined('WP_DEBUG') && WP_DEBUG) {
7416 - error_log(sprintf(
7417 - '[MxChat] openrouter_stream initial-connect failure (attempt=%d/%d, status=%d, errno=%d, transient=%s, %s).',
7418 - $attempt + 1, $max_attempts, $http_code, $errno,
7419 - $is_transient ? 'yes' : 'no',
7420 - $can_retry ? 'Retrying.' : 'Giving up.'
7421 - ));
7422 - }
7423 -
7424 - if (!$can_retry) {
7425 - break;
7426 - }
7427 - }
7428 -
7429 - if (!$errno && $http_code === 200) {
7430 - if (!empty($full_response) && !empty($session_id)) {
7431 - $rag_context_for_storage = null;
7432 - $has_rag_data = $this->last_similarity_analysis !== null && !empty($this->last_similarity_analysis['top_matches']);
7433 - $has_action_data = isset($this->last_action_analysis) && !empty($this->last_action_analysis);
7434 -
7435 - if ($has_rag_data || $has_action_data) {
7436 - $rag_context_for_storage = [];
7437 -
7438 - if ($has_rag_data) {
7439 - $rag_context_for_storage['top_matches'] = $this->last_similarity_analysis['top_matches'];
7440 - $rag_context_for_storage['approved_urls'] = $this->current_valid_urls ?? [];
7441 - $rag_context_for_storage['similarity_threshold'] = $this->last_similarity_analysis['threshold_used'] ?? 0.35;
7442 - $rag_context_for_storage['knowledge_base_type'] = $this->last_similarity_analysis['knowledge_base_type'] ?? 'WordPress Database';
7443 - $rag_context_for_storage['total_documents_checked'] = $this->last_similarity_analysis['total_checked'] ?? 0;
7444 - }
7445 -
7446 - if ($has_action_data) {
7447 - $rag_context_for_storage['action_analysis'] = $this->last_action_analysis;
7448 - }
7449 - }
7450 - $this->mxchat_save_chat_message($session_id, 'bot', $full_response, null, $rag_context_for_storage);
7451 - }
7452 - return true;
7453 - }
7454 -
7455 - return $this->mxchat_stream_emit_fallback(
7456 - 'openai',
7457 - $this->mxchat_generate_response_openrouter($selected_model, $openrouter_api_key, $conversation_history, $relevant_content),
7458 - $session_id,
7459 - $testing_data
7460 - );
7461 -
7462 - } catch (Exception $e) {
7463 - return $this->mxchat_stream_emit_fallback(
7464 - 'openai',
7465 - $this->mxchat_generate_response_openrouter($selected_model, $openrouter_api_key, $conversation_history, $relevant_content),
7466 - $session_id,
7467 - $testing_data
7468 - );
7469 - }
7470 -}
7471 -private function mxchat_generate_response_openai_stream($selected_model, $api_key, $conversation_history, $relevant_content, $session_id, $testing_data = null) {
7472 - try {
7473 - $bot_id = $this->get_current_bot_id($session_id);
4399 +
4400 + return strlen($data);
4401 + });
7474 4402
7475 - // Get system prompt instructions using centralized function
7476 - $system_prompt_instructions = $this->get_system_instructions($bot_id, $session_id);
4403 + $response = curl_exec($ch);
4404 + $http_code = curl_getinfo($ch, CURLINFO_HTTP_CODE);
7477 4405
7478 - // Ensure conversation_history is an array
7479 - if (!is_array($conversation_history)) {
7480 - $conversation_history = array();
7481 - }
7482 -
7483 - // Format conversation history for OpenAI
7484 - $formatted_conversation = array();
7485 -
7486 - $formatted_conversation[] = array(
7487 - 'role' => 'system',
7488 - 'content' => $system_prompt_instructions . " " . $relevant_content
7489 - );
7490 -
7491 - foreach ($conversation_history as $message) {
7492 - if (is_array($message) && isset($message['role']) && isset($message['content'])) {
7493 - $role = $message['role'];
7494 - if ($role === 'bot' || $role === 'agent') {
7495 - $role = 'assistant';
7496 - }
7497 - if (!in_array($role, ['system', 'assistant', 'user', 'function', 'tool'])) {
7498 - $role = 'user';
7499 - }
7500 - $formatted_conversation[] = array(
7501 - 'role' => $role,
7502 - 'content' => $message['content']
7503 - );
7504 - }
7505 - }
7506 -
7507 - // Check if we can actually stream
7508 - if (headers_sent() || !function_exists('curl_init')) {
7509 - // Fallback to regular response with testing data
4406 + if (curl_errno($ch) || $http_code !== 200) {
4407 + curl_close($ch);
4408 +
4409 + // Fallback to regular response
4410 + //error_log("MxChat: OpenAI streaming failed, falling back");
7510 4411 $regular_response = $this->mxchat_generate_response_openai(
7511 4412 $selected_model,
7512 4413 $api_key,
7513 4414 $conversation_history,
@@ -7513,13 +4414,8 @@
7513 4414 $conversation_history,
7514 4415 $relevant_content
7515 4416 );
7516 4417
7517 - // Save bot response to transcript
7518 - if (!empty($regular_response) && !empty($session_id)) {
7519 - $this->mxchat_save_chat_message($session_id, 'bot', $regular_response);
7520 - }
7521 -
7522 4418 $response_data = [
7523 4419 'text' => $regular_response,
7524 4420 'html' => '',
7525 4421 'session_id' => $session_id
@@ -7526,8 +4422,9 @@
7526 4422 ];
7527 4423
7528 4424 if ($testing_data !== null) {
7529 4425 $response_data['testing_data'] = $testing_data;
4426 + //error_log("MxChat Testing: Added testing data to OpenAI error fallback");
7530 4427 }
7531 4428
7532 4429 header('Content-Type: application/json');
7533 4430 echo json_encode($response_data);
@@ -7532,912 +4429,50 @@
7532 4429 header('Content-Type: application/json');
7533 4430 echo json_encode($response_data);
7534 4431 return true;
7535 4432 }
7536 -
7537 - // Check if this is a GPT-5 model (supports reasoning_effort parameter)
7538 - $is_gpt5_model = (
7539 - strpos($selected_model, 'gpt-5') === 0 ||
7540 - $selected_model === 'gpt-5.2' ||
7541 - $selected_model === 'gpt-5.1-2025-11-13' ||
7542 - $selected_model === 'gpt-5' ||
7543 - $selected_model === 'gpt-5-mini' ||
7544 - $selected_model === 'gpt-5-nano'
7545 - );
7546 -
7547 - // Build request body with optimal settings for fast streaming
7548 - $request_body = [
7549 - 'model' => $selected_model,
7550 - 'messages' => $formatted_conversation,
7551 - 'temperature' => 1,
7552 - 'stream' => true
7553 - ];
7554 -
7555 - // Add reasoning_effort only for GPT-5 models that support it
7556 - // These chat models don't support reasoning_effort parameter
7557 - $no_reasoning_models = array('gpt-5.2', 'gpt-5.1-chat-latest', 'gpt-5.3-chat-latest', 'gpt-5.4-mini', 'gpt-5.4-nano');
7558 - if ($is_gpt5_model && !in_array($selected_model, $no_reasoning_models, true)) {
7559 - // GPT-5.1 uses 'low' instead of 'minimal'
7560 - if ($selected_model === 'gpt-5.1-2025-11-13') {
7561 - $request_body['reasoning_effort'] = 'low';
7562 - } elseif ($selected_model === 'gpt-5.5') {
7563 - $request_body['reasoning_effort'] = 'none';
7564 - } elseif ($selected_model === 'gpt-5.4') {
7565 - $request_body['reasoning_effort'] = 'none';
7566 - } else {
7567 - $request_body['reasoning_effort'] = 'minimal';
7568 - }
4433 +
4434 + curl_close($ch);
4435 +
4436 + // Save the complete response to maintain chat persistence
4437 + if (!empty($full_response) && !empty($session_id)) {
4438 + $this->mxchat_save_chat_message($session_id, 'bot', $full_response);
7569 4439 }
7570 -
7571 - $body = json_encode($request_body);
7572 -
7573 - // V2 retry-on-initial-connect: do NOT call setup_streaming_headers() here.
7574 - // It is now lazy-fired inside the WRITEFUNCTION on the first byte of a
7575 - // SUCCESSFUL upstream response, gated by the captured HTTP status.
7576 -
7577 - $captured_status_code = 0;
7578 - $captured_body_pre_stream = '';
7579 - $full_response = '';
7580 - $stream_started = false;
7581 - $buffer = '';
7582 - $errno = 0;
7583 - $last_curl_error = '';
7584 - $http_code = 0;
7585 - $max_attempts = $this->mxchat_retry_enabled() ? 3 : 1;
7586 - $backoff_ms = array(0, 750, 2000);
7587 -
7588 - for ($attempt = 0; $attempt < $max_attempts; $attempt++) {
7589 - if ($attempt > 0 && $backoff_ms[$attempt] > 0) {
7590 - usleep($backoff_ms[$attempt] * 1000);
7591 - }
7592 -
7593 - // Reset per-attempt capture state.
7594 - $captured_status_code = 0;
7595 - $captured_body_pre_stream = '';
7596 - $full_response = '';
7597 - $stream_started = false;
7598 - $buffer = '';
7599 -
7600 - $ch = curl_init();
7601 - curl_setopt($ch, CURLOPT_URL, 'https://api.openai.com/v1/chat/completions');
7602 - curl_setopt($ch, CURLOPT_RETURNTRANSFER, false);
7603 - curl_setopt($ch, CURLOPT_POST, true);
7604 - curl_setopt($ch, CURLOPT_POSTFIELDS, $body);
7605 - curl_setopt($ch, CURLOPT_HTTPHEADER, array(
7606 - 'Content-Type: application/json',
7607 - 'Authorization: Bearer ' . $api_key
7608 - ));
7609 - curl_setopt($ch, CURLOPT_SSL_VERIFYPEER, true);
7610 - curl_setopt($ch, CURLOPT_TIMEOUT, 60);
7611 -
7612 - // Capture HTTP status as soon as response headers arrive — fires before WRITEFUNCTION.
7613 - curl_setopt($ch, CURLOPT_HEADERFUNCTION, function($ch, $header) use (&$captured_status_code) {
7614 - if ($captured_status_code === 0 && preg_match('#^HTTP/\S+\s+(\d+)\b#', $header, $m)) {
7615 - $captured_status_code = (int) $m[1];
7616 - }
7617 - return strlen($header);
7618 - });
7619 -
7620 - // Buffer control for real-time streaming
7621 - curl_setopt($ch, CURLOPT_WRITEFUNCTION, function($ch, $data) use (&$full_response, &$stream_started, &$buffer, &$captured_status_code, &$captured_body_pre_stream, $testing_data) {
7622 - // V2 guard: if upstream returned non-200, buffer body for transient
7623 - // classification and DO NOT emit to client. Stream channel must NOT open.
7624 - if ($captured_status_code !== 0 && $captured_status_code !== 200) {
7625 - $captured_body_pre_stream .= $data;
7626 - return strlen($data);
7627 - }
7628 -
7629 - // Lazy-fire streaming headers on first byte of a SUCCESSFUL upstream.
7630 - // After this point streaming_headers_sent === true → retry is structurally blocked.
7631 - if (!$this->streaming_headers_sent) {
7632 - $this->setup_streaming_headers();
7633 - }
7634 -
7635 - // Send testing data as the first event if available
7636 - if (!$stream_started && $testing_data !== null) {
7637 - echo "data: " . json_encode(['testing_data' => $testing_data]) . "\n\n";
7638 - flush();
7639 - $stream_started = true;
7640 - }
7641 -
7642 - // CRITICAL FIX: Append new data to buffer
7643 - $buffer .= $data;
7644 -
7645 - // Process complete lines only
7646 - $lines = explode("\n", $buffer);
7647 -
7648 - // CRITICAL FIX: Keep the last incomplete line in the buffer
7649 - $buffer = array_pop($lines);
7650 -
7651 - foreach ($lines as $line) {
7652 - if (trim($line) === '') {
7653 - continue;
7654 - }
7655 - if (strpos($line, 'data: ') !== 0) {
7656 - continue;
7657 - }
7658 -
7659 - $json_str = substr($line, 6);
7660 -
7661 - if (trim($json_str) === '[DONE]') {
7662 - echo "data: [DONE]\n\n";
7663 - flush();
7664 - continue;
7665 - }
7666 -
7667 - $json = json_decode(trim($json_str), true);
7668 - if ($json && isset($json['choices'][0]['delta']['content'])) {
7669 - $content = $json['choices'][0]['delta']['content'];
7670 - $full_response .= $content;
7671 -
7672 - echo "data: " . json_encode(['content' => $content]) . "\n\n";
7673 - flush();
7674 - }
7675 - }
7676 -
7677 - return strlen($data);
7678 - });
7679 -
7680 - $response = curl_exec($ch);
7681 - $errno = curl_errno($ch);
7682 - $last_curl_error = curl_error($ch);
7683 - $http_code = $captured_status_code !== 0 ? $captured_status_code : (int) curl_getinfo($ch, CURLINFO_HTTP_CODE);
7684 - curl_close($ch);
7685 -
7686 - if (!$errno && $http_code === 200) {
7687 - break; // Happy path — WRITEFUNCTION already streamed everything.
7688 - }
7689 -
7690 - $is_transient = $this->mxchat_is_transient_provider_error_raw($http_code, $captured_body_pre_stream, 'openai', $errno);
7691 - $can_retry = !$this->streaming_headers_sent
7692 - && ($attempt + 1) < $max_attempts
7693 - && $is_transient;
7694 -
7695 - if (defined('WP_DEBUG') && WP_DEBUG) {
7696 - error_log(sprintf(
7697 - '[MxChat] openai_stream initial-connect failure (attempt=%d/%d, status=%d, errno=%d, transient=%s, %s).',
7698 - $attempt + 1, $max_attempts, $http_code, $errno,
7699 - $is_transient ? 'yes' : 'no',
7700 - $can_retry ? 'Retrying.' : 'Giving up.'
7701 - ));
7702 - }
7703 -
7704 - if (!$can_retry) {
7705 - break;
7706 - }
7707 - }
7708 -
7709 - // Post-loop branch.
7710 - if (!$errno && $http_code === 200) {
7711 - // Happy path — save the complete response to maintain chat persistence.
7712 - if (!empty($full_response) && !empty($session_id)) {
7713 - $rag_context_for_storage = null;
7714 - $has_rag_data = $this->last_similarity_analysis !== null && !empty($this->last_similarity_analysis['top_matches']);
7715 - $has_action_data = isset($this->last_action_analysis) && !empty($this->last_action_analysis);
7716 -
7717 - if ($has_rag_data || $has_action_data) {
7718 - $rag_context_for_storage = [];
7719 -
7720 - if ($has_rag_data) {
7721 - $rag_context_for_storage['top_matches'] = $this->last_similarity_analysis['top_matches'];
7722 - $rag_context_for_storage['approved_urls'] = $this->current_valid_urls ?? [];
7723 - $rag_context_for_storage['similarity_threshold'] = $this->last_similarity_analysis['threshold_used'] ?? 0.35;
7724 - $rag_context_for_storage['knowledge_base_type'] = $this->last_similarity_analysis['knowledge_base_type'] ?? 'WordPress Database';
7725 - $rag_context_for_storage['total_documents_checked'] = $this->last_similarity_analysis['total_checked'] ?? 0;
7726 - }
7727 -
7728 - if ($has_action_data) {
7729 - $rag_context_for_storage['action_analysis'] = $this->last_action_analysis;
7730 - }
7731 - }
7732 - $this->mxchat_save_chat_message($session_id, 'bot', $full_response, null, $rag_context_for_storage);
7733 - }
7734 -
7735 - return true;
7736 - }
7737 -
7738 - // Failure path — branch on whether SSE channel was opened.
7739 - return $this->mxchat_stream_emit_fallback(
7740 - 'openai',
7741 - $this->mxchat_generate_response_openai($selected_model, $api_key, $conversation_history, $relevant_content),
7742 - $session_id,
7743 - $testing_data
7744 - );
7745 -
4440 +
4441 + return true; // Indicate streaming completed successfully
4442 +
7746 4443 } catch (Exception $e) {
7747 - return $this->mxchat_stream_emit_fallback(
7748 - 'openai',
7749 - $this->mxchat_generate_response_openai($selected_model, $api_key, $conversation_history, $relevant_content),
7750 - $session_id,
7751 - $testing_data
4444 + //error_log("MxChat OpenAI streaming exception: " . $e->getMessage());
4445 +
4446 + // Fallback to regular response
4447 + $regular_response = $this->mxchat_generate_response_openai(
4448 + $selected_model,
4449 + $api_key,
4450 + $conversation_history,
4451 + $relevant_content
7752 4452 );
7753 - }
7754 -}
7755 -
7756 -/**
7757 - * Shared fallback emitter for streaming chat functions. Two outcomes:
7758 - * - streaming_headers_sent === true: SSE channel is open. Emit fallback content
7759 - * as `data: {...}\n\n` + `data: [DONE]\n\n` so the widget renders it as a
7760 - * normal bot bubble. Transcript row is persisted.
7761 - * - streaming_headers_sent === false: SSE channel never opened (retries
7762 - * exhausted on initial connect). Emit a clean JSON response — the path
7763 - * the widget would normally hit if streaming wasn't even attempted.
7764 - *
7765 - * Used by all six *_stream functions after their per-attempt retry loop.
7766 - */
7767 -private function mxchat_stream_emit_fallback($provider_hint, $regular_response, $session_id, $testing_data = null) {
7768 - $is_error_array = is_array($regular_response) && isset($regular_response['error']);
7769 -
7770 - if ($this->streaming_headers_sent) {
7771 - if ($is_error_array) {
7772 - echo "data: " . json_encode([
7773 - 'error' => true,
7774 - 'error_message' => $regular_response['error'],
7775 - 'error_code' => $regular_response['error_code'] ?? 'api_error',
7776 - 'text' => $regular_response['error'],
7777 - 'message' => $regular_response['error']
7778 - ]) . "\n\n";
7779 - echo "data: [DONE]\n\n";
7780 - flush();
7781 - return true;
4453 +
4454 + $response_data = [
4455 + 'text' => $regular_response,
4456 + 'html' => '',
4457 + 'session_id' => $session_id
4458 + ];
4459 +
4460 + if ($testing_data !== null) {
4461 + $response_data['testing_data'] = $testing_data;
4462 + //error_log("MxChat Testing: Added testing data to OpenAI exception fallback");
7782 4463 }
7783 - $fallback_message = (string) $regular_response;
7784 - if (!empty($fallback_message) && !empty($session_id)) {
7785 - $this->mxchat_save_chat_message($session_id, 'bot', $fallback_message);
7786 - }
7787 - echo "data: " . json_encode(['content' => $fallback_message]) . "\n\n";
7788 - echo "data: [DONE]\n\n";
7789 - flush();
7790 - return true;
7791 - }
7792 -
7793 - // SSE channel never opened — clean JSON fallback.
7794 - if ($is_error_array) {
4464 +
7795 4465 header('Content-Type: application/json');
7796 - echo json_encode(array(
7797 - 'error' => true,
7798 - 'error_message' => $regular_response['error'],
7799 - 'error_code' => $regular_response['error_code'] ?? 'api_error',
7800 - 'text' => $regular_response['error'],
7801 - 'message' => $regular_response['error'],
7802 - ));
4466 + echo json_encode($response_data);
7803 4467 return true;
7804 4468 }
7805 -
7806 - $fallback_message = (string) $regular_response;
7807 - if (!empty($fallback_message) && !empty($session_id)) {
7808 - $this->mxchat_save_chat_message($session_id, 'bot', $fallback_message);
7809 - }
7810 - $response_data = array(
7811 - 'text' => $fallback_message,
7812 - 'html' => '',
7813 - 'session_id' => $session_id,
7814 - );
7815 - if ($testing_data !== null) {
7816 - $response_data['testing_data'] = $testing_data;
7817 - }
7818 - header('Content-Type: application/json');
7819 - echo json_encode($response_data);
7820 - return true;
7821 4469 }
7822 -
7823 -/**
7824 - * Resolve custom (OpenAI-compatible) provider config from settings.
7825 - * Returns ['base_url','api_key','model','auth_scheme','api_version','chat_url','headers'].
7826 - */
7827 -private function mxchat_resolve_custom_provider() {
7828 - $base_url = isset($this->options['custom_provider_base_url']) ? rtrim(trim((string) $this->options['custom_provider_base_url']), '/') : '';
7829 - $api_key = isset($this->options['custom_provider_api_key']) ? trim((string) $this->options['custom_provider_api_key']) : '';
7830 - $model = isset($this->options['custom_provider_model']) ? trim((string) $this->options['custom_provider_model']) : '';
7831 - $auth_scheme = isset($this->options['custom_provider_auth_scheme']) ? $this->options['custom_provider_auth_scheme'] : 'bearer';
7832 - $api_version = isset($this->options['custom_provider_api_version']) ? trim((string) $this->options['custom_provider_api_version']) : '';
7833 -
7834 - $chat_url = $base_url . '/chat/completions';
7835 - if (!empty($api_version)) {
7836 - $chat_url .= (strpos($chat_url, '?') === false ? '?' : '&') . 'api-version=' . rawurlencode($api_version);
7837 - }
7838 -
7839 - $headers = array('Content-Type: application/json');
7840 - if (!empty($api_key)) {
7841 - if ($auth_scheme === 'api-key') {
7842 - $headers[] = 'api-key: ' . $api_key;
7843 - } else {
7844 - $headers[] = 'Authorization: Bearer ' . $api_key;
7845 - }
7846 - }
7847 -
7848 - return array(
7849 - 'base_url' => $base_url,
7850 - 'api_key' => $api_key,
7851 - 'model' => $model !== '' ? $model : 'default',
7852 - 'auth_scheme' => $auth_scheme,
7853 - 'api_version' => $api_version,
7854 - 'chat_url' => $chat_url,
7855 - 'headers' => $headers,
7856 - );
7857 -}
7858 -
7859 -/**
7860 - * Streaming chat completion against an OpenAI-compatible custom provider
7861 - * (Ollama, LM Studio, vLLM, llama.cpp, Azure OpenAI, etc.).
7862 - * Mirrors mxchat_generate_response_openai_stream but with parameterized URL/auth/model.
7863 - */
7864 -private function mxchat_generate_response_custom_stream($selected_model, $conversation_history, $relevant_content, $session_id, $testing_data = null) {
4470 +private function mxchat_generate_response_claude_stream($selected_model, $claude_api_key, $conversation_history, $relevant_content, $session_id, $testing_data = null) {
7865 4471 try {
7866 - $cfg = $this->mxchat_resolve_custom_provider();
7867 - if (empty($cfg['base_url'])) {
7868 - return array('error' => esc_html__('Custom provider Base URL is not configured.', 'mxchat'), 'error_code' => 'missing_custom_provider_base_url');
7869 - }
4472 + // Get system prompt instructions from options
4473 + $system_prompt_instructions = isset($this->options['system_prompt_instructions']) ? $this->options['system_prompt_instructions'] : '';
7870 4474
7871 - $bot_id = $this->get_current_bot_id($session_id);
7872 - $system_prompt_instructions = $this->get_system_instructions($bot_id, $session_id);
7873 - if (!is_array($conversation_history)) {
7874 - $conversation_history = array();
7875 - }
7876 -
7877 - $formatted_conversation = array();
7878 - $formatted_conversation[] = array(
7879 - 'role' => 'system',
7880 - 'content' => $system_prompt_instructions . ' ' . $relevant_content,
7881 - );
7882 - foreach ($conversation_history as $message) {
7883 - if (is_array($message) && isset($message['role']) && isset($message['content'])) {
7884 - $role = $message['role'];
7885 - if ($role === 'bot' || $role === 'agent') { $role = 'assistant'; }
7886 - if (!in_array($role, array('system', 'assistant', 'user', 'function', 'tool'))) { $role = 'user'; }
7887 - $formatted_conversation[] = array('role' => $role, 'content' => $message['content']);
7888 - }
7889 - }
7890 -
7891 - if (headers_sent() || !function_exists('curl_init')) {
7892 - // No streaming capability — fall through to non-stream wrapper
7893 - $regular = $this->mxchat_generate_response_custom($selected_model, $conversation_history, $relevant_content);
7894 - if (!empty($regular) && !empty($session_id) && is_string($regular)) {
7895 - $this->mxchat_save_chat_message($session_id, 'bot', $regular);
7896 - }
7897 - $response_data = array('text' => is_string($regular) ? $regular : '', 'html' => '', 'session_id' => $session_id);
7898 - if ($testing_data !== null) { $response_data['testing_data'] = $testing_data; }
7899 - header('Content-Type: application/json');
7900 - echo json_encode($response_data);
7901 - return true;
7902 - }
7903 -
7904 - $request_body = array(
7905 - 'model' => $cfg['model'],
7906 - 'messages' => $formatted_conversation,
7907 - 'stream' => true,
7908 - );
7909 - $body = json_encode($request_body);
7910 -
7911 - // V2 retry-on-initial-connect: setup_streaming_headers is lazy-fired in WRITEFUNCTION.
7912 -
7913 - $captured_status_code = 0;
7914 - $captured_body_pre_stream = '';
7915 - $full_response = '';
7916 - $stream_started = false;
7917 - $buffer = '';
7918 - $errno = 0;
7919 - $http_code = 0;
7920 - $max_attempts = $this->mxchat_retry_enabled() ? 3 : 1;
7921 - $backoff_ms = array(0, 750, 2000);
7922 -
7923 - for ($attempt = 0; $attempt < $max_attempts; $attempt++) {
7924 - if ($attempt > 0 && $backoff_ms[$attempt] > 0) {
7925 - usleep($backoff_ms[$attempt] * 1000);
7926 - }
7927 -
7928 - $captured_status_code = 0;
7929 - $captured_body_pre_stream = '';
7930 - $full_response = '';
7931 - $stream_started = false;
7932 - $buffer = '';
7933 -
7934 - $ch = curl_init();
7935 - curl_setopt($ch, CURLOPT_URL, $cfg['chat_url']);
7936 - curl_setopt($ch, CURLOPT_RETURNTRANSFER, false);
7937 - curl_setopt($ch, CURLOPT_POST, true);
7938 - curl_setopt($ch, CURLOPT_POSTFIELDS, $body);
7939 - curl_setopt($ch, CURLOPT_HTTPHEADER, $cfg['headers']);
7940 - curl_setopt($ch, CURLOPT_SSL_VERIFYPEER, true);
7941 - curl_setopt($ch, CURLOPT_TIMEOUT, 120);
7942 -
7943 - curl_setopt($ch, CURLOPT_HEADERFUNCTION, function($ch, $header) use (&$captured_status_code) {
7944 - if ($captured_status_code === 0 && preg_match('#^HTTP/\S+\s+(\d+)\b#', $header, $m)) {
7945 - $captured_status_code = (int) $m[1];
7946 - }
7947 - return strlen($header);
7948 - });
7949 -
7950 - curl_setopt($ch, CURLOPT_WRITEFUNCTION, function($ch, $data) use (&$full_response, &$stream_started, &$buffer, &$captured_status_code, &$captured_body_pre_stream, $testing_data) {
7951 - if ($captured_status_code !== 0 && $captured_status_code !== 200) {
7952 - $captured_body_pre_stream .= $data;
7953 - return strlen($data);
7954 - }
7955 -
7956 - if (!$this->streaming_headers_sent) {
7957 - $this->setup_streaming_headers();
7958 - }
7959 -
7960 - if (!$stream_started && $testing_data !== null) {
7961 - echo "data: " . json_encode(array('testing_data' => $testing_data)) . "\n\n";
7962 - flush();
7963 - $stream_started = true;
7964 - }
7965 - $buffer .= $data;
7966 - $lines = explode("\n", $buffer);
7967 - $buffer = array_pop($lines);
7968 - foreach ($lines as $line) {
7969 - if (trim($line) === '') { continue; }
7970 - if (strpos($line, 'data: ') !== 0) { continue; }
7971 - $json_str = substr($line, 6);
7972 - if (trim($json_str) === '[DONE]') {
7973 - echo "data: [DONE]\n\n";
7974 - flush();
7975 - continue;
7976 - }
7977 - $json = json_decode(trim($json_str), true);
7978 - if ($json && isset($json['choices'][0]['delta']['content'])) {
7979 - $content = $json['choices'][0]['delta']['content'];
7980 - $full_response .= $content;
7981 - echo "data: " . json_encode(array('content' => $content)) . "\n\n";
7982 - flush();
7983 - }
7984 - }
7985 - return strlen($data);
7986 - });
7987 -
7988 - $response = curl_exec($ch);
7989 - $errno = curl_errno($ch);
7990 - $http_code = $captured_status_code !== 0 ? $captured_status_code : (int) curl_getinfo($ch, CURLINFO_HTTP_CODE);
7991 - curl_close($ch);
7992 -
7993 - if (!$errno && $http_code === 200) {
7994 - break;
7995 - }
7996 -
7997 - $is_transient = $this->mxchat_is_transient_provider_error_raw($http_code, $captured_body_pre_stream, 'openai', $errno);
7998 - $can_retry = !$this->streaming_headers_sent
7999 - && ($attempt + 1) < $max_attempts
8000 - && $is_transient;
8001 -
8002 - if (defined('WP_DEBUG') && WP_DEBUG) {
8003 - error_log(sprintf(
8004 - '[MxChat] custom_stream initial-connect failure (attempt=%d/%d, status=%d, errno=%d, transient=%s, %s).',
8005 - $attempt + 1, $max_attempts, $http_code, $errno,
8006 - $is_transient ? 'yes' : 'no',
8007 - $can_retry ? 'Retrying.' : 'Giving up.'
8008 - ));
8009 - }
8010 -
8011 - if (!$can_retry) {
8012 - break;
8013 - }
8014 - }
8015 -
8016 - if (!$errno && $http_code === 200) {
8017 - if (!empty($full_response) && !empty($session_id)) {
8018 - $this->mxchat_save_chat_message($session_id, 'bot', $full_response);
8019 - }
8020 - return true;
8021 - }
8022 -
8023 - return $this->mxchat_stream_emit_fallback(
8024 - 'openai',
8025 - $this->mxchat_generate_response_custom($selected_model, $conversation_history, $relevant_content),
8026 - $session_id,
8027 - $testing_data
8028 - );
8029 -
8030 - } catch (Exception $e) {
8031 - return array('error' => sprintf(esc_html__('Custom provider error: %s', 'mxchat'), $e->getMessage()), 'error_code' => 'custom_provider_exception');
8032 - }
8033 -}
8034 -
8035 -/**
8036 - * Non-streaming chat completion against a custom OpenAI-compatible provider.
8037 - * Returns string content on success, array['error'=>...] on failure.
8038 - */
8039 -private function mxchat_generate_response_custom($selected_model, $conversation_history, $relevant_content) {
8040 - $cfg = $this->mxchat_resolve_custom_provider();
8041 - if (empty($cfg['base_url'])) {
8042 - return array('error' => esc_html__('Custom provider Base URL is not configured.', 'mxchat'), 'error_code' => 'missing_custom_provider_base_url');
8043 - }
8044 -
8045 - $bot_id = $this->get_current_bot_id(null);
8046 - $system_prompt_instructions = $this->get_system_instructions($bot_id, null);
8047 - if (!is_array($conversation_history)) {
8048 - $conversation_history = array();
8049 - }
8050 -
8051 - $messages = array(array(
8052 - 'role' => 'system',
8053 - 'content' => $system_prompt_instructions . ' ' . $relevant_content,
8054 - ));
8055 - foreach ($conversation_history as $message) {
8056 - if (is_array($message) && isset($message['role']) && isset($message['content'])) {
8057 - $role = $message['role'];
8058 - if ($role === 'bot' || $role === 'agent') { $role = 'assistant'; }
8059 - if (!in_array($role, array('system', 'assistant', 'user', 'function', 'tool'))) { $role = 'user'; }
8060 - $messages[] = array('role' => $role, 'content' => $message['content']);
8061 - }
8062 - }
8063 -
8064 - $headers_assoc = array('Content-Type' => 'application/json');
8065 - if (!empty($cfg['api_key'])) {
8066 - if ($cfg['auth_scheme'] === 'api-key') {
8067 - $headers_assoc['api-key'] = $cfg['api_key'];
8068 - } else {
8069 - $headers_assoc['Authorization'] = 'Bearer ' . $cfg['api_key'];
8070 - }
8071 - }
8072 -
8073 - $response = $this->mxchat_provider_call_with_retry($cfg['chat_url'], array(
8074 - 'headers' => $headers_assoc,
8075 - 'body' => wp_json_encode(array(
8076 - 'model' => $cfg['model'],
8077 - 'messages' => $messages,
8078 - )),
8079 - 'timeout' => 120,
8080 - ), 'openai');
8081 -
8082 - if (is_wp_error($response)) {
8083 - return array('error' => sprintf(esc_html__('Custom provider request failed: %s', 'mxchat'), $response->get_error_message()), 'error_code' => 'custom_provider_network_error');
8084 - }
8085 - $code = (int) wp_remote_retrieve_response_code($response);
8086 - if ($code < 200 || $code >= 300) {
8087 - return array('error' => sprintf(esc_html__('Custom provider returned HTTP %d.', 'mxchat'), $code), 'error_code' => 'custom_provider_http_error');
8088 - }
8089 - $body = json_decode(wp_remote_retrieve_body($response), true);
8090 - if (isset($body['choices'][0]['message']['content'])) {
8091 - return (string) $body['choices'][0]['message']['content'];
8092 - }
8093 - return array('error' => esc_html__('Custom provider returned an unexpected response shape.', 'mxchat'), 'error_code' => 'custom_provider_response_shape');
8094 -}
8095 -
8096 -/**
8097 - * Generate response using OpenAI Responses API with web search tool
8098 - * This uses the newer Responses API which supports web search functionality
8099 - */
8100 -private function mxchat_generate_response_openai_web_search($selected_model, $api_key, $conversation_history, $relevant_content, $session_id, $testing_data = null, $streaming = false) {
8101 - try {
8102 - $bot_id = $this->get_current_bot_id($session_id);
8103 - $system_prompt_instructions = $this->get_system_instructions($bot_id, $session_id);
8104 -
8105 - if (!is_array($conversation_history)) {
8106 - $conversation_history = array();
8107 - }
8108 -
8109 - // Build the input for Responses API
8110 - // The Responses API uses a different format - we need to construct the input properly
8111 - $input_parts = [];
8112 -
8113 - // Add system instructions as context
8114 - $system_context = $system_prompt_instructions . "\n\n" . $relevant_content;
8115 -
8116 - // Build conversation as input items for Responses API
8117 - foreach ($conversation_history as $message) {
8118 - if (is_array($message) && isset($message['role']) && isset($message['content'])) {
8119 - $role = $message['role'];
8120 - if ($role === 'bot' || $role === 'agent') {
8121 - $role = 'assistant';
8122 - }
8123 - if (!in_array($role, ['assistant', 'user'])) {
8124 - $role = 'user';
8125 - }
8126 - $input_parts[] = [
8127 - 'type' => 'message',
8128 - 'role' => $role,
8129 - 'content' => $message['content']
8130 - ];
8131 - }
8132 - }
8133 -
8134 - // Build request body for Responses API
8135 - $request_body = [
8136 - 'model' => $selected_model,
8137 - 'input' => $input_parts,
8138 - 'instructions' => $system_context,
8139 - 'stream' => $streaming
8140 - ];
8141 -
8142 - // Only add web search tool if web search is enabled in settings
8143 - $web_search_enabled = isset($this->options['enable_web_search']) && $this->options['enable_web_search'] === 'on';
8144 - if ($web_search_enabled) {
8145 - $request_body['tools'] = [
8146 - ['type' => 'web_search']
8147 - ];
8148 - }
8149 -
8150 - // Add reasoning effort for supported models
8151 - $is_gpt5_model = strpos($selected_model, 'gpt-5') === 0;
8152 - $no_reasoning_web = array('gpt-5.2', 'gpt-5.3-chat-latest', 'gpt-5.4-mini', 'gpt-5.4-nano');
8153 - if ($is_gpt5_model && !in_array($selected_model, $no_reasoning_web, true)) {
8154 - if ($selected_model === 'gpt-5.1-2025-11-13') {
8155 - $request_body['reasoning'] = ['effort' => 'low'];
8156 - } elseif ($selected_model === 'gpt-5.5') {
8157 - $request_body['reasoning'] = ['effort' => 'low'];
8158 - } elseif ($selected_model === 'gpt-5.4') {
8159 - $request_body['reasoning'] = ['effort' => 'low'];
8160 - }
8161 - }
8162 -
8163 - //error_log("MXCHAT WEB SEARCH: Request body: " . json_encode($request_body));
8164 -
8165 - if ($streaming) {
8166 - return $this->mxchat_web_search_streaming_response($request_body, $api_key, $session_id, $testing_data);
8167 - } else {
8168 - return $this->mxchat_web_search_non_streaming_response($request_body, $api_key, $session_id, $testing_data);
8169 - }
8170 -
8171 - } catch (Exception $e) {
8172 - //error_log("MXCHAT WEB SEARCH ERROR: " . $e->getMessage());
8173 - return [
8174 - 'error' => sprintf(esc_html__('Web search error: %s', 'mxchat'), esc_html($e->getMessage())),
8175 - 'error_code' => 'web_search_exception'
8176 - ];
8177 - }
8178 -}
8179 -
8180 -/**
8181 - * Handle non-streaming web search response
8182 - */
8183 -private function mxchat_web_search_non_streaming_response($request_body, $api_key, $session_id, $testing_data) {
8184 - $request_body['stream'] = false;
8185 -
8186 - $response = $this->mxchat_provider_call_with_retry('https://api.openai.com/v1/responses', array(
8187 - 'headers' => array(
8188 - 'Authorization' => 'Bearer ' . $api_key,
8189 - 'Content-Type' => 'application/json'
8190 - ),
8191 - 'body' => json_encode($request_body),
8192 - 'timeout' => 90
8193 - ), 'openai');
8194 -
8195 - if (is_wp_error($response)) {
8196 - //error_log("MXCHAT WEB SEARCH ERROR: WP Error: " . $response->get_error_message());
8197 - return [
8198 - 'error' => esc_html__('Failed to connect to OpenAI web search API', 'mxchat'),
8199 - 'error_code' => 'web_search_connection_error'
8200 - ];
8201 - }
8202 -
8203 - $response_code = wp_remote_retrieve_response_code($response);
8204 - $response_body = wp_remote_retrieve_body($response);
8205 -
8206 - //error_log("MXCHAT WEB SEARCH: Response code: " . $response_code);
8207 - //error_log("MXCHAT WEB SEARCH: Response body (first 2000): " . substr($response_body, 0, 2000));
8208 -
8209 - if ($response_code !== 200) {
8210 - $error_data = json_decode($response_body, true);
8211 - $error_message = $error_data['error']['message'] ?? 'Unknown API error';
8212 - return [
8213 - 'error' => sprintf(esc_html__('OpenAI API error: %s', 'mxchat'), esc_html($error_message)),
8214 - 'error_code' => 'web_search_api_error'
8215 - ];
8216 - }
8217 -
8218 - $result = json_decode($response_body, true);
8219 -
8220 - if (json_last_error() !== JSON_ERROR_NONE) {
8221 - return [
8222 - 'error' => esc_html__('Invalid response from OpenAI', 'mxchat'),
8223 - 'error_code' => 'web_search_json_error'
8224 - ];
8225 - }
8226 -
8227 - // Extract the response text and citations from Responses API format
8228 - $output_text = '';
8229 - $citations = [];
8230 -
8231 - if (isset($result['output'])) {
8232 - foreach ($result['output'] as $output_item) {
8233 - if ($output_item['type'] === 'message' && isset($output_item['content'])) {
8234 - foreach ($output_item['content'] as $content_item) {
8235 - if ($content_item['type'] === 'output_text') {
8236 - $output_text .= $content_item['text'];
8237 -
8238 - // Extract citations/annotations
8239 - if (isset($content_item['annotations'])) {
8240 - foreach ($content_item['annotations'] as $annotation) {
8241 - if ($annotation['type'] === 'url_citation') {
8242 - $citations[] = [
8243 - 'url' => $annotation['url'],
8244 - 'title' => $annotation['title'] ?? ''
8245 - ];
8246 - }
8247 - }
8248 - }
8249 - }
8250 - }
8251 - }
8252 - }
8253 - }
8254 -
8255 - // If we have citations, append them to the response
8256 - if (!empty($citations)) {
8257 - $output_text .= "\n\n**Sources:**\n";
8258 - $seen_urls = [];
8259 - foreach ($citations as $citation) {
8260 - if (!in_array($citation['url'], $seen_urls)) {
8261 - $seen_urls[] = $citation['url'];
8262 - $title = !empty($citation['title']) ? $citation['title'] : $citation['url'];
8263 - $output_text .= "- [" . $title . "](" . $citation['url'] . ")\n";
8264 - }
8265 - }
8266 - }
8267 -
8268 - // Transcript save is handled by the main handler (mxchat_handle_chat_request)
8269 - // which includes rag_context for the "sources" link in transcripts.
8270 -
8271 - return $output_text;
8272 -}
8273 -
8274 -/**
8275 - * Handle streaming web search response using Responses API
8276 - */
8277 -private function mxchat_web_search_streaming_response($request_body, $api_key, $session_id, $testing_data) {
8278 - $request_body['stream'] = true;
8279 -
8280 - // Check if we can stream
8281 - if (headers_sent() || !function_exists('curl_init')) {
8282 - // Fallback to non-streaming
8283 - return $this->mxchat_web_search_non_streaming_response($request_body, $api_key, $session_id, $testing_data);
8284 - }
8285 -
8286 - // Setup streaming headers
8287 - $this->setup_streaming_headers();
8288 -
8289 - $ch = curl_init();
8290 - curl_setopt($ch, CURLOPT_URL, 'https://api.openai.com/v1/responses');
8291 - curl_setopt($ch, CURLOPT_RETURNTRANSFER, false);
8292 - curl_setopt($ch, CURLOPT_POST, true);
8293 - curl_setopt($ch, CURLOPT_POSTFIELDS, json_encode($request_body));
8294 - curl_setopt($ch, CURLOPT_HTTPHEADER, array(
8295 - 'Content-Type: application/json',
8296 - 'Authorization: Bearer ' . $api_key
8297 - ));
8298 - curl_setopt($ch, CURLOPT_SSL_VERIFYPEER, true);
8299 - curl_setopt($ch, CURLOPT_TIMEOUT, 120);
8300 -
8301 - $full_response = '';
8302 - $stream_started = false;
8303 - $buffer = '';
8304 - $citations = [];
8305 -
8306 - curl_setopt($ch, CURLOPT_WRITEFUNCTION, function($ch, $data) use (&$full_response, &$stream_started, &$buffer, &$citations, $testing_data) {
8307 - // Send testing data as first event if available
8308 - if (!$stream_started && $testing_data !== null) {
8309 - echo "data: " . json_encode(['testing_data' => $testing_data]) . "\n\n";
8310 - flush();
8311 - $stream_started = true;
8312 - }
8313 -
8314 - $buffer .= $data;
8315 - $lines = explode("\n", $buffer);
8316 - $buffer = array_pop($lines);
8317 -
8318 - foreach ($lines as $line) {
8319 - if (trim($line) === '') continue;
8320 - if (strpos($line, 'data: ') !== 0) continue;
8321 -
8322 - $json_str = substr($line, 6);
8323 -
8324 - if (trim($json_str) === '[DONE]') {
8325 - // Append citations if we have any
8326 - if (!empty($citations)) {
8327 - $citation_text = "\n\n**Sources:**\n";
8328 - $seen_urls = [];
8329 - foreach ($citations as $citation) {
8330 - if (!in_array($citation['url'], $seen_urls)) {
8331 - $seen_urls[] = $citation['url'];
8332 - $title = !empty($citation['title']) ? $citation['title'] : $citation['url'];
8333 - $citation_text .= "- [" . $title . "](" . $citation['url'] . ")\n";
8334 - }
8335 - }
8336 - echo "data: " . json_encode(['content' => $citation_text]) . "\n\n";
8337 - $full_response .= $citation_text;
8338 - flush();
8339 - }
8340 - echo "data: [DONE]\n\n";
8341 - flush();
8342 - continue;
8343 - }
8344 -
8345 - $json = json_decode(trim($json_str), true);
8346 - if (!$json) continue;
8347 -
8348 - // Handle Responses API streaming events
8349 - // The format is different from Chat Completions
8350 - if (isset($json['type'])) {
8351 - switch ($json['type']) {
8352 - case 'response.output_text.delta':
8353 - // Text content delta
8354 - if (isset($json['delta'])) {
8355 - $content = $json['delta'];
8356 - $full_response .= $content;
8357 - echo "data: " . json_encode(['content' => $content]) . "\n\n";
8358 - flush();
8359 - }
8360 - break;
8361 -
8362 - case 'response.output_item.done':
8363 - // Check for citations in completed items
8364 - if (isset($json['item']['content'])) {
8365 - foreach ($json['item']['content'] as $content_item) {
8366 - if (isset($content_item['annotations'])) {
8367 - foreach ($content_item['annotations'] as $annotation) {
8368 - if ($annotation['type'] === 'url_citation') {
8369 - $citations[] = [
8370 - 'url' => $annotation['url'],
8371 - 'title' => $annotation['title'] ?? ''
8372 - ];
8373 - }
8374 - }
8375 - }
8376 - }
8377 - }
8378 - break;
8379 - }
8380 - }
8381 - }
8382 -
8383 - return strlen($data);
8384 - });
8385 -
8386 - $response = curl_exec($ch);
8387 - $http_code = curl_getinfo($ch, CURLINFO_HTTP_CODE);
8388 -
8389 - if (curl_errno($ch) || $http_code !== 200) {
8390 - $curl_error = curl_error($ch);
8391 - curl_close($ch);
8392 -
8393 - //error_log("MXCHAT WEB SEARCH STREAM ERROR: HTTP $http_code, cURL error: $curl_error");
8394 -
8395 - return $this->mxchat_stream_emit_fallback(
8396 - 'web_search',
8397 - $this->mxchat_web_search_non_streaming_response($request_body, $api_key, $session_id, $testing_data),
8398 - $session_id,
8399 - $testing_data
8400 - );
8401 - }
8402 -
8403 - curl_close($ch);
8404 -
8405 - // Save the complete response with RAG context so the "sources" link
8406 - // appears in transcripts — mirrors the pattern used by Claude/OpenAI streaming.
8407 - if (!empty($full_response) && !empty($session_id)) {
8408 - $rag_context_for_storage = null;
8409 - $has_rag_data = $this->last_similarity_analysis !== null && !empty($this->last_similarity_analysis['top_matches']);
8410 - $has_action_data = isset($this->last_action_analysis) && !empty($this->last_action_analysis);
8411 -
8412 - if ($has_rag_data || $has_action_data) {
8413 - $rag_context_for_storage = [];
8414 -
8415 - if ($has_rag_data) {
8416 - $rag_context_for_storage['top_matches'] = $this->last_similarity_analysis['top_matches'];
8417 - $rag_context_for_storage['approved_urls'] = $this->current_valid_urls ?? [];
8418 - $rag_context_for_storage['similarity_threshold'] = $this->last_similarity_analysis['threshold_used'] ?? 0.35;
8419 - $rag_context_for_storage['knowledge_base_type'] = $this->last_similarity_analysis['knowledge_base_type'] ?? 'WordPress Database';
8420 - $rag_context_for_storage['total_documents_checked'] = $this->last_similarity_analysis['total_checked'] ?? 0;
8421 - }
8422 -
8423 - if ($has_action_data) {
8424 - $rag_context_for_storage['action_analysis'] = $this->last_action_analysis;
8425 - }
8426 - }
8427 - $this->mxchat_save_chat_message($session_id, 'bot', $full_response, null, $rag_context_for_storage);
8428 - }
8429 -
8430 - return true;
8431 -}
8432 -
8433 -private function mxchat_generate_response_claude_stream($selected_model, $claude_api_key, $conversation_history, $relevant_content, $session_id, $testing_data = null) {
8434 - try {
8435 - // Get bot ID from session or request
8436 - $bot_id = $this->get_current_bot_id($session_id);
8437 -
8438 - // Get system prompt instructions using centralized function
8439 - $system_prompt_instructions = $this->get_system_instructions($bot_id, $session_id);
8440 4475 // Ensure conversation_history is an array
8441 4476 if (!is_array($conversation_history)) {
8442 4477 $conversation_history = array();
8443 4478 }
@@ -8469,9 +4504,9 @@
8469 4504 'content' => $relevant_content
8470 4505 ];
8471 4506
8472 4507 // Prepare the request body with stream: true
8473 - $payload = [
4508 + $body = json_encode([
8474 4509 'model' => $selected_model,
8475 4510 'messages' => $conversation_history,
8476 4511 'max_tokens' => 1000,
8477 4512 'temperature' => 0.8,
@@ -8476,11 +4511,9 @@
8476 4511 'max_tokens' => 1000,
8477 4512 'temperature' => 0.8,
8478 4513 'system' => $system_prompt_instructions,
8479 4514 'stream' => true
8480 - ];
8481 - if ($this->mxchat_claude_omits_temperature($selected_model)) { unset($payload['temperature']); }
8482 - $body = json_encode($payload);
4515 + ]);
8483 4516
8484 4517 // Check if we can actually stream (headers not sent, etc.)
8485 4518 if (headers_sent() || !function_exists('curl_init')) {
8486 4519 // Fallback to regular response with testing data
@@ -8491,13 +4524,8 @@
8491 4524 array_slice($conversation_history, 0, -1), // Remove the added content
8492 4525 $relevant_content
8493 4526 );
8494 4527
8495 - // Save bot response to transcript
8496 - if (!empty($regular_response) && !empty($session_id)) {
8497 - $this->mxchat_save_chat_message($session_id, 'bot', $regular_response);
8498 - }
8499 -
8500 4528 // Return as JSON with testing data
8501 4529 $response_data = [
8502 4530 'text' => $regular_response,
8503 4531 'html' => '',
@@ -8516,197 +4544,162 @@
8516 4544 echo json_encode($response_data);
8517 4545 return true; // Indicate we handled the response
8518 4546 }
8519 4547
8520 - // V2 retry-on-initial-connect: setup_streaming_headers is lazy-fired in WRITEFUNCTION.
4548 + // Use cURL for streaming support
4549 + $ch = curl_init();
4550 + curl_setopt($ch, CURLOPT_URL, 'https://api.anthropic.com/v1/messages');
4551 + curl_setopt($ch, CURLOPT_RETURNTRANSFER, false);
4552 + curl_setopt($ch, CURLOPT_POST, true);
4553 + curl_setopt($ch, CURLOPT_POSTFIELDS, $body);
4554 + curl_setopt($ch, CURLOPT_HTTPHEADER, array(
4555 + 'Content-Type: application/json',
4556 + 'x-api-key: ' . $claude_api_key,
4557 + 'anthropic-version: 2023-06-01'
4558 + ));
4559 + curl_setopt($ch, CURLOPT_SSL_VERIFYPEER, true);
4560 + curl_setopt($ch, CURLOPT_TIMEOUT, 60);
8521 4561
8522 - $captured_status_code = 0;
8523 - $captured_body_pre_stream = '';
8524 - $full_response = '';
4562 + $full_response = ''; // Accumulate full response for saving
8525 4563 $stream_started = false;
8526 - $buffer = '';
8527 - $errno = 0;
8528 - $http_code = 0;
8529 - $max_attempts = $this->mxchat_retry_enabled() ? 3 : 1;
8530 - $backoff_ms = array(0, 750, 2000);
8531 4564
8532 - for ($attempt = 0; $attempt < $max_attempts; $attempt++) {
8533 - if ($attempt > 0 && $backoff_ms[$attempt] > 0) {
8534 - usleep($backoff_ms[$attempt] * 1000);
4565 + // Buffer control for real-time streaming
4566 + curl_setopt($ch, CURLOPT_WRITEFUNCTION, function($ch, $data) use (&$full_response, &$stream_started, $testing_data) {
4567 + // Send testing data as the first event if available
4568 + if (!$stream_started && $testing_data !== null) {
4569 + echo "data: " . json_encode(['testing_data' => $testing_data]) . "\n\n";
4570 + flush();
4571 + $stream_started = true;
4572 + //error_log("MxChat Testing: Sent testing data in Claude stream");
8535 4573 }
4574 +
4575 + // Process each chunk of data
4576 + $lines = explode("\n", $data);
8536 4577
8537 - $captured_status_code = 0;
8538 - $captured_body_pre_stream = '';
8539 - $full_response = '';
8540 - $stream_started = false;
8541 - $buffer = '';
8542 -
8543 - $ch = curl_init();
8544 - curl_setopt($ch, CURLOPT_URL, 'https://api.anthropic.com/v1/messages');
8545 - curl_setopt($ch, CURLOPT_RETURNTRANSFER, false);
8546 - curl_setopt($ch, CURLOPT_POST, true);
8547 - curl_setopt($ch, CURLOPT_POSTFIELDS, $body);
8548 - curl_setopt($ch, CURLOPT_HTTPHEADER, array(
8549 - 'Content-Type: application/json',
8550 - 'x-api-key: ' . $claude_api_key,
8551 - 'anthropic-version: 2023-06-01'
8552 - ));
8553 - curl_setopt($ch, CURLOPT_SSL_VERIFYPEER, true);
8554 - curl_setopt($ch, CURLOPT_TIMEOUT, 60);
8555 -
8556 - curl_setopt($ch, CURLOPT_HEADERFUNCTION, function($ch, $header) use (&$captured_status_code) {
8557 - if ($captured_status_code === 0 && preg_match('#^HTTP/\S+\s+(\d+)\b#', $header, $m)) {
8558 - $captured_status_code = (int) $m[1];
4578 + foreach ($lines as $line) {
4579 + if (trim($line) === '') {
4580 + continue;
8559 4581 }
8560 - return strlen($header);
8561 - });
8562 4582
8563 - curl_setopt($ch, CURLOPT_WRITEFUNCTION, function($ch, $data) use (&$full_response, &$stream_started, &$buffer, &$captured_status_code, &$captured_body_pre_stream, $testing_data) {
8564 - if ($captured_status_code !== 0 && $captured_status_code !== 200) {
8565 - $captured_body_pre_stream .= $data;
8566 - return strlen($data);
4583 + // Claude uses event: and data: format
4584 + if (strpos($line, 'event: ') === 0) {
4585 + // Store the event type for the next data line
4586 + continue;
8567 4587 }
8568 4588
8569 - if (!$this->streaming_headers_sent) {
8570 - $this->setup_streaming_headers();
8571 - }
4589 + if (strpos($line, 'data: ') === 0) {
4590 + $json_str = substr($line, 6); // Remove 'data: ' prefix
8572 4591
8573 - if (!$stream_started && $testing_data !== null) {
8574 - echo "data: " . json_encode(['testing_data' => $testing_data]) . "\n\n";
8575 - flush();
8576 - $stream_started = true;
8577 - }
8578 -
8579 - $buffer .= $data;
8580 - $lines = explode("\n", $buffer);
8581 - $buffer = array_pop($lines);
8582 -
8583 - foreach ($lines as $line) {
8584 - if (trim($line) === '') {
4592 + $json = json_decode($json_str, true);
4593 + if (json_last_error() !== JSON_ERROR_NONE) {
8585 4594 continue;
8586 4595 }
8587 4596
8588 - if (strpos($line, 'event: ') === 0) {
8589 - continue;
8590 - }
4597 + // Handle different event types
4598 + if (isset($json['type'])) {
4599 + switch ($json['type']) {
4600 + case 'content_block_delta':
4601 + if (isset($json['delta']['text'])) {
4602 + $content = $json['delta']['text'];
4603 + $full_response .= $content; // Accumulate
4604 + // Send as SSE format compatible with your frontend
4605 + echo "data: " . json_encode(['content' => $content]) . "\n\n";
4606 + flush();
4607 + }
4608 + break;
8591 4609
8592 - if (strpos($line, 'data: ') === 0) {
8593 - $json_str = substr($line, 6);
4610 + case 'message_stop':
4611 + echo "data: [DONE]\n\n";
4612 + flush();
4613 + break;
8594 4614
8595 - $json = json_decode(trim($json_str), true);
8596 - if (json_last_error() !== JSON_ERROR_NONE) {
8597 - continue;
4615 + case 'error':
4616 + echo "data: " . json_encode(['error' => $json['error']['message'] ?? 'Unknown error']) . "\n\n";
4617 + flush();
4618 + break;
8598 4619 }
8599 -
8600 - if (isset($json['type'])) {
8601 - switch ($json['type']) {
8602 - case 'content_block_delta':
8603 - if (isset($json['delta']['text'])) {
8604 - $content = $json['delta']['text'];
8605 - $full_response .= $content;
8606 - echo "data: " . json_encode(['content' => $content]) . "\n\n";
8607 - flush();
8608 - }
8609 - break;
8610 -
8611 - case 'message_stop':
8612 - echo "data: [DONE]\n\n";
8613 - flush();
8614 - break;
8615 -
8616 - case 'error':
8617 - echo "data: " . json_encode(['error' => $json['error']['message'] ?? 'Unknown error']) . "\n\n";
8618 - flush();
8619 - break;
8620 - }
8621 - }
8622 4620 }
8623 4621 }
8624 -
8625 - return strlen($data);
8626 - });
8627 -
8628 - $response = curl_exec($ch);
8629 - $errno = curl_errno($ch);
8630 - $http_code = $captured_status_code !== 0 ? $captured_status_code : (int) curl_getinfo($ch, CURLINFO_HTTP_CODE);
8631 - curl_close($ch);
8632 -
8633 - if (!$errno && $http_code === 200) {
8634 - break;
8635 4622 }
8636 4623
8637 - $is_transient = $this->mxchat_is_transient_provider_error_raw($http_code, $captured_body_pre_stream, 'anthropic', $errno);
8638 - $can_retry = !$this->streaming_headers_sent
8639 - && ($attempt + 1) < $max_attempts
8640 - && $is_transient;
4624 + return strlen($data);
4625 + });
8641 4626
8642 - if (defined('WP_DEBUG') && WP_DEBUG) {
8643 - error_log(sprintf(
8644 - '[MxChat] claude_stream initial-connect failure (attempt=%d/%d, status=%d, errno=%d, transient=%s, %s).',
8645 - $attempt + 1, $max_attempts, $http_code, $errno,
8646 - $is_transient ? 'yes' : 'no',
8647 - $can_retry ? 'Retrying.' : 'Giving up.'
8648 - ));
8649 - }
4627 + $response = curl_exec($ch);
4628 + $http_code = curl_getinfo($ch, CURLINFO_HTTP_CODE);
8650 4629
8651 - if (!$can_retry) {
8652 - break;
8653 - }
4630 + if (curl_errno($ch)) {
4631 + curl_close($ch);
4632 + throw new Exception('cURL Error: ' . curl_error($ch));
8654 4633 }
8655 4634
8656 - if ($errno || $http_code !== 200) {
8657 - return $this->mxchat_stream_emit_fallback(
8658 - 'anthropic',
8659 - $this->mxchat_generate_response_claude($selected_model, $claude_api_key, array_slice($conversation_history, 0, -1), $relevant_content),
8660 - $session_id,
8661 - $testing_data
4635 + curl_close($ch);
4636 +
4637 + if ($http_code !== 200) {
4638 + // Fallback to regular response
4639 + //error_log("MxChat: Claude streaming failed with HTTP $http_code, falling back");
4640 + $regular_response = $this->mxchat_generate_response_claude(
4641 + $selected_model,
4642 + $claude_api_key,
4643 + array_slice($conversation_history, 0, -1), // Remove the added content
4644 + $relevant_content
8662 4645 );
4646 +
4647 + $response_data = [
4648 + 'text' => $regular_response,
4649 + 'html' => '',
4650 + 'session_id' => $session_id
4651 + ];
4652 +
4653 + if ($testing_data !== null) {
4654 + $response_data['testing_data'] = $testing_data;
4655 + //error_log("MxChat Testing: Added testing data to Claude error fallback");
4656 + }
4657 +
4658 + header('Content-Type: application/json');
4659 + echo json_encode($response_data);
4660 + return true;
8663 4661 }
8664 4662
8665 4663 // Save the complete response to maintain chat persistence
8666 4664 if (!empty($full_response) && !empty($session_id)) {
8667 - // Prepare RAG context for streaming response
8668 - $rag_context_for_storage = null;
8669 - $has_rag_data = $this->last_similarity_analysis !== null && !empty($this->last_similarity_analysis['top_matches']);
8670 - $has_action_data = isset($this->last_action_analysis) && !empty($this->last_action_analysis);
8671 -
8672 - if ($has_rag_data || $has_action_data) {
8673 - $rag_context_for_storage = [];
8674 -
8675 - if ($has_rag_data) {
8676 - $rag_context_for_storage['top_matches'] = $this->last_similarity_analysis['top_matches'];
8677 - $rag_context_for_storage['approved_urls'] = $this->current_valid_urls ?? [];
8678 - $rag_context_for_storage['similarity_threshold'] = $this->last_similarity_analysis['threshold_used'] ?? 0.35;
8679 - $rag_context_for_storage['knowledge_base_type'] = $this->last_similarity_analysis['knowledge_base_type'] ?? 'WordPress Database';
8680 - $rag_context_for_storage['total_documents_checked'] = $this->last_similarity_analysis['total_checked'] ?? 0;
8681 - }
8682 -
8683 - if ($has_action_data) {
8684 - $rag_context_for_storage['action_analysis'] = $this->last_action_analysis;
8685 - }
8686 - }
8687 - $this->mxchat_save_chat_message($session_id, 'bot', $full_response, null, $rag_context_for_storage);
4665 + $this->mxchat_save_chat_message($session_id, 'bot', $full_response);
8688 4666 }
8689 4667
8690 4668 return true; // Indicate streaming completed successfully
8691 4669
8692 4670 } catch (Exception $e) {
8693 - return $this->mxchat_stream_emit_fallback(
8694 - 'anthropic',
8695 - $this->mxchat_generate_response_claude($selected_model, $claude_api_key, $conversation_history, $relevant_content),
8696 - $session_id,
8697 - $testing_data
4671 + //error_log("MxChat Claude streaming exception: " . $e->getMessage());
4672 +
4673 + // Fallback to regular response on exception
4674 + $regular_response = $this->mxchat_generate_response_claude(
4675 + $selected_model,
4676 + $claude_api_key,
4677 + $conversation_history,
4678 + $relevant_content
8698 4679 );
4680 +
4681 + $response_data = [
4682 + 'text' => $regular_response,
4683 + 'html' => '',
4684 + 'session_id' => $session_id
4685 + ];
4686 +
4687 + if ($testing_data !== null) {
4688 + $response_data['testing_data'] = $testing_data;
4689 + //error_log("MxChat Testing: Added testing data to Claude exception fallback");
4690 + }
4691 +
4692 + header('Content-Type: application/json');
4693 + echo json_encode($response_data);
4694 + return true;
8699 4695 }
8700 4696 }
8701 4697 private function mxchat_generate_response_xai_stream($selected_model, $xai_api_key, $conversation_history, $relevant_content, $session_id, $testing_data = null) {
8702 4698 try {
8703 - // Get bot ID from session or request
8704 - $bot_id = $this->get_current_bot_id($session_id);
4699 + // Get system prompt instructions from options
4700 + $system_prompt_instructions = isset($this->options['system_prompt_instructions']) ? $this->options['system_prompt_instructions'] : '';
8705 4701
8706 - // Get system prompt instructions using centralized function
8707 - $system_prompt_instructions = $this->get_system_instructions($bot_id, $session_id);
8708 -
8709 4702 // Ensure conversation_history is an array
8710 4703 if (!is_array($conversation_history)) {
8711 4704 $conversation_history = array();
8712 4705 }
@@ -8745,13 +4738,8 @@
8745 4738 $conversation_history,
8746 4739 $relevant_content
8747 4740 );
8748 4741
8749 - // Save bot response to transcript
8750 - if (!empty($regular_response) && !empty($session_id)) {
8751 - $this->mxchat_save_chat_message($session_id, 'bot', $regular_response);
8752 - }
8753 -
8754 4742 $response_data = [
8755 4743 'text' => $regular_response,
8756 4744 'html' => '',
8757 4745 'session_id' => $session_id
@@ -8774,179 +4762,135 @@
8774 4762 'temperature' => 0.8,
8775 4763 'stream' => true
8776 4764 ]);
8777 4765
8778 - // V2 retry-on-initial-connect: setup_streaming_headers is lazy-fired in WRITEFUNCTION.
8779 -
8780 - $captured_status_code = 0;
8781 - $captured_body_pre_stream = '';
8782 - $full_response = '';
4766 + // Use cURL for streaming support
4767 + $ch = curl_init();
4768 + curl_setopt($ch, CURLOPT_URL, 'https://api.x.ai/v1/chat/completions');
4769 + curl_setopt($ch, CURLOPT_RETURNTRANSFER, false);
4770 + curl_setopt($ch, CURLOPT_POST, true);
4771 + curl_setopt($ch, CURLOPT_POSTFIELDS, $body);
4772 + curl_setopt($ch, CURLOPT_HTTPHEADER, array(
4773 + 'Content-Type: application/json',
4774 + 'Authorization: Bearer ' . $xai_api_key
4775 + ));
4776 + curl_setopt($ch, CURLOPT_SSL_VERIFYPEER, true);
4777 + curl_setopt($ch, CURLOPT_TIMEOUT, 60);
4778 +
4779 + $full_response = ''; // Accumulate full response for saving
8783 4780 $stream_started = false;
8784 - $buffer = '';
8785 - $errno = 0;
8786 - $http_code = 0;
8787 - $max_attempts = $this->mxchat_retry_enabled() ? 3 : 1;
8788 - $backoff_ms = array(0, 750, 2000);
8789 -
8790 - for ($attempt = 0; $attempt < $max_attempts; $attempt++) {
8791 - if ($attempt > 0 && $backoff_ms[$attempt] > 0) {
8792 - usleep($backoff_ms[$attempt] * 1000);
4781 +
4782 + // Buffer control for real-time streaming
4783 + curl_setopt($ch, CURLOPT_WRITEFUNCTION, function($ch, $data) use (&$full_response, &$stream_started, $testing_data) {
4784 + // Send testing data as the first event if available
4785 + if (!$stream_started && $testing_data !== null) {
4786 + echo "data: " . json_encode(['testing_data' => $testing_data]) . "\n\n";
4787 + flush();
4788 + $stream_started = true;
4789 + //error_log("MxChat Testing: Sent testing data in X.AI stream");
8793 4790 }
8794 -
8795 - $captured_status_code = 0;
8796 - $captured_body_pre_stream = '';
8797 - $full_response = '';
8798 - $stream_started = false;
8799 - $buffer = '';
8800 -
8801 - $ch = curl_init();
8802 - curl_setopt($ch, CURLOPT_URL, 'https://api.x.ai/v1/chat/completions');
8803 - curl_setopt($ch, CURLOPT_RETURNTRANSFER, false);
8804 - curl_setopt($ch, CURLOPT_POST, true);
8805 - curl_setopt($ch, CURLOPT_POSTFIELDS, $body);
8806 - curl_setopt($ch, CURLOPT_HTTPHEADER, array(
8807 - 'Content-Type: application/json',
8808 - 'Authorization: Bearer ' . $xai_api_key
8809 - ));
8810 - curl_setopt($ch, CURLOPT_SSL_VERIFYPEER, true);
8811 - curl_setopt($ch, CURLOPT_TIMEOUT, 60);
8812 -
8813 - curl_setopt($ch, CURLOPT_HEADERFUNCTION, function($ch, $header) use (&$captured_status_code) {
8814 - if ($captured_status_code === 0 && preg_match('#^HTTP/\S+\s+(\d+)\b#', $header, $m)) {
8815 - $captured_status_code = (int) $m[1];
4791 +
4792 + // Process each chunk of data
4793 + $lines = explode("\n", $data);
4794 +
4795 + foreach ($lines as $line) {
4796 + if (trim($line) === '' || strpos($line, 'data: ') !== 0) {
4797 + continue;
8816 4798 }
8817 - return strlen($header);
8818 - });
8819 -
8820 - curl_setopt($ch, CURLOPT_WRITEFUNCTION, function($ch, $data) use (&$full_response, &$stream_started, &$buffer, &$captured_status_code, &$captured_body_pre_stream, $testing_data) {
8821 - if ($captured_status_code !== 0 && $captured_status_code !== 200) {
8822 - $captured_body_pre_stream .= $data;
8823 - return strlen($data);
4799 +
4800 + $json_str = substr($line, 6); // Remove 'data: ' prefix
4801 +
4802 + if ($json_str === '[DONE]') {
4803 + echo "data: [DONE]\n\n";
4804 + flush();
4805 + continue;
8824 4806 }
8825 -
8826 - if (!$this->streaming_headers_sent) {
8827 - $this->setup_streaming_headers();
8828 - }
8829 -
8830 - if (!$stream_started && $testing_data !== null) {
8831 - echo "data: " . json_encode(['testing_data' => $testing_data]) . "\n\n";
4807 +
4808 + $json = json_decode($json_str, true);
4809 + if (isset($json['choices'][0]['delta']['content'])) {
4810 + $content = $json['choices'][0]['delta']['content'];
4811 + $full_response .= $content; // Accumulate
4812 + // Send as SSE format
4813 + echo "data: " . json_encode(['content' => $content]) . "\n\n";
8832 4814 flush();
8833 - $stream_started = true;
8834 4815 }
8835 -
8836 - $buffer .= $data;
8837 - $lines = explode("\n", $buffer);
8838 - $buffer = array_pop($lines);
8839 -
8840 - foreach ($lines as $line) {
8841 - if (trim($line) === '') {
8842 - continue;
8843 - }
8844 - if (strpos($line, 'data: ') !== 0) {
8845 - continue;
8846 - }
8847 -
8848 - $json_str = substr($line, 6);
8849 -
8850 - if (trim($json_str) === '[DONE]') {
8851 - echo "data: [DONE]\n\n";
8852 - flush();
8853 - continue;
8854 - }
8855 -
8856 - $json = json_decode(trim($json_str), true);
8857 - if ($json && isset($json['choices'][0]['delta']['content'])) {
8858 - $content = $json['choices'][0]['delta']['content'];
8859 - $full_response .= $content;
8860 - echo "data: " . json_encode(['content' => $content]) . "\n\n";
8861 - flush();
8862 - }
8863 - }
8864 -
8865 - return strlen($data);
8866 - });
8867 -
8868 - $response = curl_exec($ch);
8869 - $errno = curl_errno($ch);
8870 - $http_code = $captured_status_code !== 0 ? $captured_status_code : (int) curl_getinfo($ch, CURLINFO_HTTP_CODE);
4816 + }
4817 +
4818 + return strlen($data);
4819 + });
4820 +
4821 + $response = curl_exec($ch);
4822 + $http_code = curl_getinfo($ch, CURLINFO_HTTP_CODE);
4823 +
4824 + if (curl_errno($ch) || $http_code !== 200) {
8871 4825 curl_close($ch);
8872 -
8873 - if (!$errno && $http_code === 200) {
8874 - break;
4826 +
4827 + // Fallback to regular response
4828 + //error_log("MxChat: X.AI streaming failed, falling back");
4829 + $regular_response = $this->mxchat_generate_response_xai(
4830 + $selected_model,
4831 + $xai_api_key,
4832 + $conversation_history,
4833 + $relevant_content
4834 + );
4835 +
4836 + $response_data = [
4837 + 'text' => $regular_response,
4838 + 'html' => '',
4839 + 'session_id' => $session_id
4840 + ];
4841 +
4842 + if ($testing_data !== null) {
4843 + $response_data['testing_data'] = $testing_data;
4844 + //error_log("MxChat Testing: Added testing data to X.AI error fallback");
8875 4845 }
8876 -
8877 - $is_transient = $this->mxchat_is_transient_provider_error_raw($http_code, $captured_body_pre_stream, 'xai', $errno);
8878 - $can_retry = !$this->streaming_headers_sent
8879 - && ($attempt + 1) < $max_attempts
8880 - && $is_transient;
8881 -
8882 - if (defined('WP_DEBUG') && WP_DEBUG) {
8883 - error_log(sprintf(
8884 - '[MxChat] xai_stream initial-connect failure (attempt=%d/%d, status=%d, errno=%d, transient=%s, %s).',
8885 - $attempt + 1, $max_attempts, $http_code, $errno,
8886 - $is_transient ? 'yes' : 'no',
8887 - $can_retry ? 'Retrying.' : 'Giving up.'
8888 - ));
8889 - }
8890 -
8891 - if (!$can_retry) {
8892 - break;
8893 - }
4846 +
4847 + header('Content-Type: application/json');
4848 + echo json_encode($response_data);
4849 + return true;
8894 4850 }
8895 -
8896 - if ($errno || $http_code !== 200) {
8897 - return $this->mxchat_stream_emit_fallback(
8898 - 'xai',
8899 - $this->mxchat_generate_response_xai($selected_model, $xai_api_key, $conversation_history, $relevant_content),
8900 - $session_id,
8901 - $testing_data
8902 - );
8903 - }
8904 -
4851 +
4852 + curl_close($ch);
4853 +
8905 4854 // Save the complete response to maintain chat persistence
8906 4855 if (!empty($full_response) && !empty($session_id)) {
8907 - // Prepare RAG context for streaming response
8908 - $rag_context_for_storage = null;
8909 - $has_rag_data = $this->last_similarity_analysis !== null && !empty($this->last_similarity_analysis['top_matches']);
8910 - $has_action_data = isset($this->last_action_analysis) && !empty($this->last_action_analysis);
8911 -
8912 - if ($has_rag_data || $has_action_data) {
8913 - $rag_context_for_storage = [];
8914 -
8915 - if ($has_rag_data) {
8916 - $rag_context_for_storage['top_matches'] = $this->last_similarity_analysis['top_matches'];
8917 - $rag_context_for_storage['approved_urls'] = $this->current_valid_urls ?? [];
8918 - $rag_context_for_storage['similarity_threshold'] = $this->last_similarity_analysis['threshold_used'] ?? 0.35;
8919 - $rag_context_for_storage['knowledge_base_type'] = $this->last_similarity_analysis['knowledge_base_type'] ?? 'WordPress Database';
8920 - $rag_context_for_storage['total_documents_checked'] = $this->last_similarity_analysis['total_checked'] ?? 0;
8921 - }
8922 -
8923 - if ($has_action_data) {
8924 - $rag_context_for_storage['action_analysis'] = $this->last_action_analysis;
8925 - }
8926 - }
8927 - $this->mxchat_save_chat_message($session_id, 'bot', $full_response, null, $rag_context_for_storage);
4856 + $this->mxchat_save_chat_message($session_id, 'bot', $full_response);
8928 4857 }
8929 -
4858 +
8930 4859 return true; // Indicate streaming completed successfully
8931 -
4860 +
8932 4861 } catch (Exception $e) {
8933 - return $this->mxchat_stream_emit_fallback(
8934 - 'xai',
8935 - $this->mxchat_generate_response_xai($selected_model, $xai_api_key, $conversation_history, $relevant_content),
8936 - $session_id,
8937 - $testing_data
4862 + //error_log("MxChat X.AI streaming exception: " . $e->getMessage());
4863 +
4864 + // Fallback to regular response
4865 + $regular_response = $this->mxchat_generate_response_xai(
4866 + $selected_model,
4867 + $xai_api_key,
4868 + $conversation_history,
4869 + $relevant_content
8938 4870 );
4871 +
4872 + $response_data = [
4873 + 'text' => $regular_response,
4874 + 'html' => '',
4875 + 'session_id' => $session_id
4876 + ];
4877 +
4878 + if ($testing_data !== null) {
4879 + $response_data['testing_data'] = $testing_data;
4880 + //error_log("MxChat Testing: Added testing data to X.AI exception fallback");
4881 + }
4882 +
4883 + header('Content-Type: application/json');
4884 + echo json_encode($response_data);
4885 + return true;
8939 4886 }
8940 4887 }
8941 4888 private function mxchat_generate_response_deepseek_stream($selected_model, $deepseek_api_key, $conversation_history, $relevant_content, $session_id, $testing_data = null) {
8942 4889 try {
8943 - // Get bot ID from session or request
8944 - $bot_id = $this->get_current_bot_id($session_id);
4890 + // Get system prompt instructions from options
4891 + $system_prompt_instructions = isset($this->options['system_prompt_instructions']) ? $this->options['system_prompt_instructions'] : '';
8945 4892
8946 - // Get system prompt instructions using centralized function
8947 - $system_prompt_instructions = $this->get_system_instructions($bot_id, $session_id);
8948 -
8949 4893 // Ensure conversation_history is an array
8950 4894 if (!is_array($conversation_history)) {
8951 4895 $conversation_history = array();
8952 4896 }
@@ -8985,13 +4929,8 @@
8985 4929 $conversation_history,
8986 4930 $relevant_content
8987 4931 );
8988 4932
8989 - // Save bot response to transcript
8990 - if (!empty($regular_response) && !empty($session_id)) {
8991 - $this->mxchat_save_chat_message($session_id, 'bot', $regular_response);
8992 - }
8993 -
8994 4933 $response_data = [
8995 4934 'text' => $regular_response,
8996 4935 'html' => '',
8997 4936 'session_id' => $session_id
@@ -9014,284 +4953,158 @@
9014 4953 'temperature' => 0.8,
9015 4954 'stream' => true
9016 4955 ]);
9017 4956
9018 - // V2 retry-on-initial-connect: setup_streaming_headers is lazy-fired in WRITEFUNCTION.
9019 -
9020 - $captured_status_code = 0;
9021 - $captured_body_pre_stream = '';
9022 - $full_response = '';
4957 + // Use cURL for streaming support
4958 + $ch = curl_init();
4959 + curl_setopt($ch, CURLOPT_URL, 'https://api.deepseek.com/v1/chat/completions');
4960 + curl_setopt($ch, CURLOPT_RETURNTRANSFER, false);
4961 + curl_setopt($ch, CURLOPT_POST, true);
4962 + curl_setopt($ch, CURLOPT_POSTFIELDS, $body);
4963 + curl_setopt($ch, CURLOPT_HTTPHEADER, array(
4964 + 'Content-Type: application/json',
4965 + 'Authorization: Bearer ' . $deepseek_api_key
4966 + ));
4967 + curl_setopt($ch, CURLOPT_SSL_VERIFYPEER, true);
4968 + curl_setopt($ch, CURLOPT_TIMEOUT, 60);
4969 +
4970 + $full_response = ''; // Accumulate full response for saving
9023 4971 $stream_started = false;
9024 - $buffer = '';
9025 - $errno = 0;
9026 - $http_code = 0;
9027 - $max_attempts = $this->mxchat_retry_enabled() ? 3 : 1;
9028 - $backoff_ms = array(0, 750, 2000);
9029 -
9030 - for ($attempt = 0; $attempt < $max_attempts; $attempt++) {
9031 - if ($attempt > 0 && $backoff_ms[$attempt] > 0) {
9032 - usleep($backoff_ms[$attempt] * 1000);
4972 +
4973 + // Buffer control for real-time streaming
4974 + curl_setopt($ch, CURLOPT_WRITEFUNCTION, function($ch, $data) use (&$full_response, &$stream_started, $testing_data) {
4975 + // Send testing data as the first event if available
4976 + if (!$stream_started && $testing_data !== null) {
4977 + echo "data: " . json_encode(['testing_data' => $testing_data]) . "\n\n";
4978 + flush();
4979 + $stream_started = true;
4980 + //error_log("MxChat Testing: Sent testing data in DeepSeek stream");
9033 4981 }
9034 -
9035 - $captured_status_code = 0;
9036 - $captured_body_pre_stream = '';
9037 - $full_response = '';
9038 - $stream_started = false;
9039 - $buffer = '';
9040 -
9041 - $ch = curl_init();
9042 - curl_setopt($ch, CURLOPT_URL, 'https://api.deepseek.com/v1/chat/completions');
9043 - curl_setopt($ch, CURLOPT_RETURNTRANSFER, false);
9044 - curl_setopt($ch, CURLOPT_POST, true);
9045 - curl_setopt($ch, CURLOPT_POSTFIELDS, $body);
9046 - curl_setopt($ch, CURLOPT_HTTPHEADER, array(
9047 - 'Content-Type: application/json',
9048 - 'Authorization: Bearer ' . $deepseek_api_key
9049 - ));
9050 - curl_setopt($ch, CURLOPT_SSL_VERIFYPEER, true);
9051 - curl_setopt($ch, CURLOPT_TIMEOUT, 60);
9052 -
9053 - curl_setopt($ch, CURLOPT_HEADERFUNCTION, function($ch, $header) use (&$captured_status_code) {
9054 - if ($captured_status_code === 0 && preg_match('#^HTTP/\S+\s+(\d+)\b#', $header, $m)) {
9055 - $captured_status_code = (int) $m[1];
4982 +
4983 + // Process each chunk of data
4984 + $lines = explode("\n", $data);
4985 +
4986 + foreach ($lines as $line) {
4987 + if (trim($line) === '' || strpos($line, 'data: ') !== 0) {
4988 + continue;
9056 4989 }
9057 - return strlen($header);
9058 - });
9059 -
9060 - curl_setopt($ch, CURLOPT_WRITEFUNCTION, function($ch, $data) use (&$full_response, &$stream_started, &$buffer, &$captured_status_code, &$captured_body_pre_stream, $testing_data) {
9061 - if ($captured_status_code !== 0 && $captured_status_code !== 200) {
9062 - $captured_body_pre_stream .= $data;
9063 - return strlen($data);
4990 +
4991 + $json_str = substr($line, 6); // Remove 'data: ' prefix
4992 +
4993 + if ($json_str === '[DONE]') {
4994 + echo "data: [DONE]\n\n";
4995 + flush();
4996 + continue;
9064 4997 }
9065 -
9066 - if (!$this->streaming_headers_sent) {
9067 - $this->setup_streaming_headers();
9068 - }
9069 -
9070 - if (!$stream_started && $testing_data !== null) {
9071 - echo "data: " . json_encode(['testing_data' => $testing_data]) . "\n\n";
4998 +
4999 + $json = json_decode($json_str, true);
5000 + if (isset($json['choices'][0]['delta']['content'])) {
5001 + $content = $json['choices'][0]['delta']['content'];
5002 + $full_response .= $content; // Accumulate
5003 + // Send as SSE format
5004 + echo "data: " . json_encode(['content' => $content]) . "\n\n";
9072 5005 flush();
9073 - $stream_started = true;
9074 5006 }
9075 -
9076 - $buffer .= $data;
9077 - $lines = explode("\n", $buffer);
9078 - $buffer = array_pop($lines);
9079 -
9080 - foreach ($lines as $line) {
9081 - if (trim($line) === '') {
9082 - continue;
9083 - }
9084 - if (strpos($line, 'data: ') !== 0) {
9085 - continue;
9086 - }
9087 -
9088 - $json_str = substr($line, 6);
9089 -
9090 - if (trim($json_str) === '[DONE]') {
9091 - echo "data: [DONE]\n\n";
9092 - flush();
9093 - continue;
9094 - }
9095 -
9096 - $json = json_decode(trim($json_str), true);
9097 - if ($json && isset($json['choices'][0]['delta']['content'])) {
9098 - $content = $json['choices'][0]['delta']['content'];
9099 - $full_response .= $content;
9100 - echo "data: " . json_encode(['content' => $content]) . "\n\n";
9101 - flush();
9102 - }
5007 + }
5008 +
5009 + return strlen($data);
5010 + });
5011 +
5012 + $response = curl_exec($ch);
5013 + $http_code = curl_getinfo($ch, CURLINFO_HTTP_CODE);
5014 +
5015 + if (curl_errno($ch) || $http_code !== 200) {
5016 + $curl_error = curl_error($ch);
5017 + curl_close($ch);
5018 +
5019 + // Log the specific error for debugging
5020 + //error_log("MxChat: DeepSeek streaming failed - HTTP: $http_code, cURL: $curl_error");
5021 +
5022 + // Fallback to regular response
5023 + $regular_response = $this->mxchat_generate_response_deepseek(
5024 + $selected_model,
5025 + $deepseek_api_key,
5026 + $conversation_history,
5027 + $relevant_content
5028 + );
5029 +
5030 + // Handle error response from regular function
5031 + if (is_array($regular_response) && isset($regular_response['error'])) {
5032 + if ($testing_data !== null) {
5033 + $regular_response['testing_data'] = $testing_data;
9103 5034 }
9104 -
9105 - return strlen($data);
9106 - });
9107 -
9108 - $response = curl_exec($ch);
9109 - $errno = curl_errno($ch);
9110 - $http_code = $captured_status_code !== 0 ? $captured_status_code : (int) curl_getinfo($ch, CURLINFO_HTTP_CODE);
9111 - curl_close($ch);
9112 -
9113 - if (!$errno && $http_code === 200) {
9114 - break;
5035 + header('Content-Type: application/json');
5036 + echo json_encode($regular_response);
5037 + return true;
9115 5038 }
9116 -
9117 - $is_transient = $this->mxchat_is_transient_provider_error_raw($http_code, $captured_body_pre_stream, 'openai', $errno);
9118 - $can_retry = !$this->streaming_headers_sent
9119 - && ($attempt + 1) < $max_attempts
9120 - && $is_transient;
9121 -
9122 - if (defined('WP_DEBUG') && WP_DEBUG) {
9123 - error_log(sprintf(
9124 - '[MxChat] deepseek_stream initial-connect failure (attempt=%d/%d, status=%d, errno=%d, transient=%s, %s).',
9125 - $attempt + 1, $max_attempts, $http_code, $errno,
9126 - $is_transient ? 'yes' : 'no',
9127 - $can_retry ? 'Retrying.' : 'Giving up.'
9128 - ));
5039 +
5040 + $response_data = [
5041 + 'text' => $regular_response,
5042 + 'html' => '',
5043 + 'session_id' => $session_id
5044 + ];
5045 +
5046 + if ($testing_data !== null) {
5047 + $response_data['testing_data'] = $testing_data;
5048 + //error_log("MxChat Testing: Added testing data to DeepSeek error fallback");
9129 5049 }
9130 -
9131 - if (!$can_retry) {
9132 - break;
9133 - }
5050 +
5051 + header('Content-Type: application/json');
5052 + echo json_encode($response_data);
5053 + return true;
9134 5054 }
9135 -
9136 - if ($errno || $http_code !== 200) {
9137 - return $this->mxchat_stream_emit_fallback(
9138 - 'openai',
9139 - $this->mxchat_generate_response_deepseek($selected_model, $deepseek_api_key, $conversation_history, $relevant_content),
9140 - $session_id,
9141 - $testing_data
9142 - );
9143 - }
9144 -
5055 +
5056 + curl_close($ch);
5057 +
9145 5058 // Save the complete response to maintain chat persistence
9146 5059 if (!empty($full_response) && !empty($session_id)) {
9147 - // Prepare RAG context for streaming response
9148 - $rag_context_for_storage = null;
9149 - $has_rag_data = $this->last_similarity_analysis !== null && !empty($this->last_similarity_analysis['top_matches']);
9150 - $has_action_data = isset($this->last_action_analysis) && !empty($this->last_action_analysis);
9151 -
9152 - if ($has_rag_data || $has_action_data) {
9153 - $rag_context_for_storage = [];
9154 -
9155 - if ($has_rag_data) {
9156 - $rag_context_for_storage['top_matches'] = $this->last_similarity_analysis['top_matches'];
9157 - $rag_context_for_storage['approved_urls'] = $this->current_valid_urls ?? [];
9158 - $rag_context_for_storage['similarity_threshold'] = $this->last_similarity_analysis['threshold_used'] ?? 0.35;
9159 - $rag_context_for_storage['knowledge_base_type'] = $this->last_similarity_analysis['knowledge_base_type'] ?? 'WordPress Database';
9160 - $rag_context_for_storage['total_documents_checked'] = $this->last_similarity_analysis['total_checked'] ?? 0;
9161 - }
9162 -
9163 - if ($has_action_data) {
9164 - $rag_context_for_storage['action_analysis'] = $this->last_action_analysis;
9165 - }
9166 - }
9167 - $this->mxchat_save_chat_message($session_id, 'bot', $full_response, null, $rag_context_for_storage);
5060 + $this->mxchat_save_chat_message($session_id, 'bot', $full_response);
9168 5061 }
9169 -
5062 +
9170 5063 return true; // Indicate streaming completed successfully
9171 -
5064 +
9172 5065 } catch (Exception $e) {
9173 - return $this->mxchat_stream_emit_fallback(
9174 - 'openai',
9175 - $this->mxchat_generate_response_deepseek($selected_model, $deepseek_api_key, $conversation_history, $relevant_content),
9176 - $session_id,
9177 - $testing_data
5066 + //error_log("MxChat DeepSeek streaming exception: " . $e->getMessage());
5067 +
5068 + // Fallback to regular response
5069 + $regular_response = $this->mxchat_generate_response_deepseek(
5070 + $selected_model,
5071 + $deepseek_api_key,
5072 + $conversation_history,
5073 + $relevant_content
9178 5074 );
9179 - }
9180 -}
9181 -
9182 -
9183 -private function mxchat_generate_response_openrouter($selected_model, $openrouter_api_key, $conversation_history, $relevant_content) {
9184 - try {
9185 - if (!is_array($conversation_history)) {
9186 - $conversation_history = array();
9187 - }
9188 -
9189 - $bot_id = $this->get_current_bot_id('');
9190 - $system_prompt_instructions = $this->get_system_instructions($bot_id, $session_id);
9191 5075
9192 - $formatted_conversation = array();
9193 -
9194 - $formatted_conversation[] = array(
9195 - 'role' => 'system',
9196 - 'content' => $system_prompt_instructions . " " . $relevant_content
9197 - );
9198 -
9199 - foreach ($conversation_history as $message) {
9200 - if (is_array($message) && isset($message['role']) && isset($message['content'])) {
9201 - $role = $message['role'];
9202 -
9203 - if ($role === 'bot' || $role === 'agent') {
9204 - $role = 'assistant';
9205 - }
9206 - if (!in_array($role, ['system', 'assistant', 'user'])) {
9207 - $role = 'user';
9208 - }
9209 -
9210 - $formatted_conversation[] = array(
9211 - 'role' => $role,
9212 - 'content' => $message['content']
9213 - );
5076 + // Handle error response from regular function
5077 + if (is_array($regular_response) && isset($regular_response['error'])) {
5078 + if ($testing_data !== null) {
5079 + $regular_response['testing_data'] = $testing_data;
9214 5080 }
5081 + header('Content-Type: application/json');
5082 + echo json_encode($regular_response);
5083 + return true;
9215 5084 }
9216 -
9217 - $body = json_encode([
9218 - 'model' => $selected_model,
9219 - 'messages' => $formatted_conversation,
9220 - 'temperature' => 1,
9221 - ]);
9222 -
9223 - $args = [
9224 - 'body' => $body,
9225 - 'headers' => [
9226 - 'Content-Type' => 'application/json',
9227 - 'Authorization' => 'Bearer ' . $openrouter_api_key,
9228 - 'HTTP-Referer' => home_url(),
9229 - 'X-Title' => get_bloginfo('name'),
9230 - ],
9231 - 'timeout' => 60,
9232 - 'redirection' => 5,
9233 - 'blocking' => true,
9234 - 'httpversion' => '1.0',
9235 - 'sslverify' => true,
5085 +
5086 + $response_data = [
5087 + 'text' => $regular_response,
5088 + 'html' => '',
5089 + 'session_id' => $session_id
9236 5090 ];
9237 -
9238 - $response = $this->mxchat_provider_call_with_retry('https://openrouter.ai/api/v1/chat/completions', $args, 'openai');
9239 -
9240 - if (is_wp_error($response)) {
9241 - $error_message = $response->get_error_message();
9242 - return [
9243 - 'error' => esc_html__('Connection error when contacting OpenRouter: ', 'mxchat') . esc_html($error_message),
9244 - 'error_code' => 'openrouter_connection_error',
9245 - 'provider' => 'openrouter'
9246 - ];
5091 +
5092 + if ($testing_data !== null) {
5093 + $response_data['testing_data'] = $testing_data;
5094 + //error_log("MxChat Testing: Added testing data to DeepSeek exception fallback");
9247 5095 }
9248 -
9249 - $status_code = wp_remote_retrieve_response_code($response);
9250 - if ($status_code !== 200) {
9251 - $response_body = wp_remote_retrieve_body($response);
9252 - $decoded_response = json_decode($response_body, true);
9253 -
9254 - $error_message = isset($decoded_response['error']['message'])
9255 - ? $decoded_response['error']['message']
9256 - : 'HTTP Error ' . $status_code;
9257 -
9258 - return [
9259 - 'error' => esc_html__('OpenRouter API error: ', 'mxchat') . esc_html($error_message),
9260 - 'error_code' => 'openrouter_api_error',
9261 - 'provider' => 'openrouter',
9262 - 'status_code' => $status_code
9263 - ];
9264 - }
9265 -
9266 - $response_body = wp_remote_retrieve_body($response);
9267 - $decoded_response = json_decode($response_body, true);
9268 -
9269 - if (isset($decoded_response['choices'][0]['message']['content'])) {
9270 - return trim($decoded_response['choices'][0]['message']['content']);
9271 - } else {
9272 - return [
9273 - 'error' => esc_html__('Unexpected response format from OpenRouter.', 'mxchat'),
9274 - 'error_code' => 'openrouter_response_format_error',
9275 - 'provider' => 'openrouter'
9276 - ];
9277 - }
9278 - } catch (Exception $e) {
9279 - return [
9280 - 'error' => esc_html__('System error when processing OpenRouter request: ', 'mxchat') . esc_html($e->getMessage()),
9281 - 'error_code' => 'openrouter_exception',
9282 - 'provider' => 'openrouter'
9283 - ];
5096 +
5097 + header('Content-Type: application/json');
5098 + echo json_encode($response_data);
5099 + return true;
9284 5100 }
9285 5101 }
5102 +
9286 5103 private function mxchat_generate_response_claude($selected_model, $claude_api_key, $conversation_history, $relevant_content) {
9287 -
9288 - // Get bot ID from session or request
9289 - $bot_id = $this->get_current_bot_id($session_id);
9290 -
9291 - // Get system prompt instructions using centralized function
9292 - $system_prompt_instructions = $this->get_system_instructions($bot_id, $session_id);
9293 -
5104 + // Get system prompt instructions from options
5105 + $system_prompt_instructions = isset($this->options['system_prompt_instructions']) ? $this->options['system_prompt_instructions'] : '';
5106 +
9294 5107 // Clean and validate conversation history
9295 5108 foreach ($conversation_history as &$message) {
9296 5109 // Convert bot and agent roles to assistant
9297 5110 if ($message['role'] === 'bot' || $message['role'] === 'agent') {
@@ -9318,17 +5131,15 @@
9318 5131 'content' => $relevant_content
9319 5132 ];
9320 5133
9321 5134 // Build request body
9322 - $payload = [
5135 + $body = json_encode([
9323 5136 'model' => $selected_model,
9324 5137 'max_tokens' => 1000,
9325 5138 'temperature' => 0.8,
9326 5139 'messages' => $conversation_history,
9327 5140 'system' => $system_prompt_instructions
9328 - ];
9329 - if ($this->mxchat_claude_omits_temperature($selected_model)) { unset($payload['temperature']); }
9330 - $body = json_encode($payload);
5141 + ]);
9331 5142
9332 5143 // Set up API request
9333 5144 $args = [
9334 5145 'body' => $body,
@@ -9344,9 +5155,9 @@
9344 5155 'sslverify' => true,
9345 5156 ];
9346 5157
9347 5158 // Make API request
9348 - $response = $this->mxchat_provider_call_with_retry('https://api.anthropic.com/v1/messages', $args, 'anthropic');
5159 + $response = wp_remote_post('https://api.anthropic.com/v1/messages', $args);
9349 5160
9350 5161 // Check for WordPress errors
9351 5162 if (is_wp_error($response)) {
9352 5163 //error_log("Claude API request error: " . $response->get_error_message());
@@ -9395,14 +5206,11 @@
9395 5206 if (!is_array($conversation_history)) {
9396 5207 $conversation_history = array();
9397 5208 }
9398 5209
9399 - // Get bot ID from session or request
9400 - $bot_id = $this->get_current_bot_id('');
9401 -
9402 - // Get system prompt instructions using centralized function
9403 - $system_prompt_instructions = $this->get_system_instructions($bot_id, $session_id);
9404 -
5210 + // Get system prompt instructions from options
5211 + $system_prompt_instructions = isset($this->options['system_prompt_instructions']) ? $this->options['system_prompt_instructions'] : '';
5212 +
9405 5213 // Create a new array for the formatted conversation
9406 5214 $formatted_conversation = array();
9407 5215
9408 5216 // Add system message first
@@ -9430,44 +5238,15 @@
9430 5238 );
9431 5239 }
9432 5240 }
9433 5241
9434 - // Check if this is a GPT-5 model (supports reasoning_effort parameter)
9435 - $is_gpt5_model = (
9436 - strpos($selected_model, 'gpt-5') === 0 ||
9437 - $selected_model === 'gpt-5.2' ||
9438 - $selected_model === 'gpt-5.1-2025-11-13' ||
9439 - $selected_model === 'gpt-5' ||
9440 - $selected_model === 'gpt-5-mini' ||
9441 - $selected_model === 'gpt-5-nano'
9442 - );
9443 -
9444 - // Build request body with optimal settings for fast responses
9445 - $request_body = [
5242 + $body = json_encode([
9446 5243 'model' => $selected_model,
9447 5244 'messages' => $formatted_conversation,
9448 5245 'temperature' => 1,
9449 5246 'stream' => false
9450 - ];
5247 + ]);
9451 5248
9452 - // Add reasoning_effort only for GPT-5 models that support it
9453 - // These chat models don't support reasoning_effort parameter
9454 - $no_reasoning_models = array('gpt-5.2', 'gpt-5.1-chat-latest', 'gpt-5.3-chat-latest', 'gpt-5.4-mini', 'gpt-5.4-nano');
9455 - if ($is_gpt5_model && !in_array($selected_model, $no_reasoning_models, true)) {
9456 - // GPT-5.1 uses 'low' instead of 'minimal'
9457 - if ($selected_model === 'gpt-5.1-2025-11-13') {
9458 - $request_body['reasoning_effort'] = 'low';
9459 - } elseif ($selected_model === 'gpt-5.5') {
9460 - $request_body['reasoning_effort'] = 'none';
9461 - } elseif ($selected_model === 'gpt-5.4') {
9462 - $request_body['reasoning_effort'] = 'none';
9463 - } else {
9464 - $request_body['reasoning_effort'] = 'minimal';
9465 - }
9466 - }
9467 -
9468 - $body = json_encode($request_body);
9469 -
9470 5249 $args = [
9471 5250 'body' => $body,
9472 5251 'headers' => [
9473 5252 'Content-Type' => 'application/json',
@@ -9479,12 +5258,13 @@
9479 5258 'httpversion' => '1.0',
9480 5259 'sslverify' => true,
9481 5260 ];
9482 5261
9483 - $response = $this->mxchat_provider_call_with_retry('https://api.openai.com/v1/chat/completions', $args, 'openai');
5262 + $response = wp_remote_post('https://api.openai.com/v1/chat/completions', $args);
9484 5263
9485 5264 if (is_wp_error($response)) {
9486 5265 $error_message = $response->get_error_message();
5266 + //error_log('OpenAI API Error: ' . $error_message);
9487 5267 return [
9488 5268 'error' => esc_html__('Connection error when contacting OpenAI: ', 'mxchat') . esc_html($error_message),
9489 5269 'error_code' => 'openai_connection_error',
9490 5270 'provider' => 'openai'
@@ -9503,8 +5283,10 @@
9503 5283 $error_type = isset($decoded_response['error']['type'])
9504 5284 ? $decoded_response['error']['type']
9505 5285 : 'unknown';
9506 5286
5287 + //error_log('OpenAI API HTTP Error: ' . $status_code . ' - ' . $error_message);
5288 +
9507 5289 // Handle specific error types
9508 5290 switch ($error_type) {
9509 5291 case 'invalid_request_error':
9510 5292 if (strpos($error_message, 'API key') !== false) {
@@ -9552,8 +5334,9 @@
9552 5334
9553 5335 if (isset($decoded_response['choices'][0]['message']['content'])) {
9554 5336 return trim($decoded_response['choices'][0]['message']['content']);
9555 5337 } else {
5338 + //error_log('OpenAI API Response Format Error: ' . print_r($decoded_response, true));
9556 5339 return [
9557 5340 'error' => esc_html__('Unexpected response format from OpenAI.', 'mxchat'),
9558 5341 'error_code' => 'openai_response_format_error',
9559 5342 'provider' => 'openai'
@@ -9559,8 +5342,9 @@
9559 5342 'provider' => 'openai'
9560 5343 ];
9561 5344 }
9562 5345 } catch (Exception $e) {
5346 + //error_log('OpenAI Exception: ' . $e->getMessage());
9563 5347 return [
9564 5348 'error' => esc_html__('System error when processing OpenAI request: ', 'mxchat') . esc_html($e->getMessage()),
9565 5349 'error_code' => 'openai_exception',
9566 5350 'provider' => 'openai'
@@ -9566,17 +5350,13 @@
9566 5350 'provider' => 'openai'
9567 5351 ];
9568 5352 }
9569 5353 }
9570 -
9571 5354 private function mxchat_generate_response_xai($selected_model, $xai_api_key, $conversation_history, $relevant_content) {
9572 5355 try {
9573 - // Get bot ID from session or request
9574 - $bot_id = $this->get_current_bot_id($session_id);
9575 -
9576 - // Get system prompt instructions using centralized function
9577 - $system_prompt_instructions = $this->get_system_instructions($bot_id, $session_id);
9578 -
5356 + // Get system prompt instructions from options
5357 + $system_prompt_instructions = isset($this->options['system_prompt_instructions']) ? $this->options['system_prompt_instructions'] : '';
5358 +
9579 5359 // Add system prompt to relevant content
9580 5360 $content_with_instructions = $system_prompt_instructions . " " . $relevant_content;
9581 5361
9582 5362 // Prepend system instructions to the conversation history
@@ -9625,9 +5405,9 @@
9625 5405 'sslverify' => true,
9626 5406 ];
9627 5407
9628 5408 // Make the API request
9629 - $response = $this->mxchat_provider_call_with_retry('https://api.x.ai/v1/chat/completions', $args, 'xai');
5409 + $response = wp_remote_post('https://api.x.ai/v1/chat/completions', $args);
9630 5410
9631 5411 // Process the response
9632 5412 if (is_wp_error($response)) {
9633 5413 $error_message = $response->get_error_message();
@@ -9768,14 +5548,11 @@
9768 5548 if (!is_array($conversation_history)) {
9769 5549 $conversation_history = array();
9770 5550 }
9771 5551
9772 - // Get bot ID from session or request
9773 - $bot_id = $this->get_current_bot_id($session_id);
9774 -
9775 - // Get system prompt instructions using centralized function
9776 - $system_prompt_instructions = $this->get_system_instructions($bot_id, $session_id);
9777 -
5552 + // Get system prompt instructions from options
5553 + $system_prompt_instructions = isset($this->options['system_prompt_instructions']) ? $this->options['system_prompt_instructions'] : '';
5554 +
9778 5555 // Create a new array for the formatted conversation
9779 5556 $formatted_conversation = array();
9780 5557
9781 5558 // Add system message first
@@ -9823,9 +5600,9 @@
9823 5600 'httpversion' => '1.0',
9824 5601 'sslverify' => true,
9825 5602 ];
9826 5603
9827 - $response = $this->mxchat_provider_call_with_retry('https://api.deepseek.com/v1/chat/completions', $args, 'openai');
5604 + $response = wp_remote_post('https://api.deepseek.com/v1/chat/completions', $args);
9828 5605
9829 5606 if (is_wp_error($response)) {
9830 5607 $error_message = $response->get_error_message();
9831 5608 //error_log('DeepSeek API Error: ' . $error_message);
@@ -9927,19 +5704,11 @@
9927 5704 ];
9928 5705 }
9929 5706 }
9930 5707 private function mxchat_generate_response_gemini($selected_model, $gemini_api_key, $conversation_history, $relevant_content) {
9931 - // Read-time remap: gemini-3-pro-preview was shut down March 9, 2026.
9932 - // Auto-rescue existing installs whose saved model is the dead ID.
9933 - if ($selected_model === 'gemini-3-pro-preview') {
9934 - $selected_model = 'gemini-3.1-pro-preview';
9935 - }
9936 - // Get bot ID from session or request
9937 - $bot_id = $this->get_current_bot_id($session_id);
9938 -
9939 - // Get system prompt instructions using centralized function
9940 - $system_prompt_instructions = $this->get_system_instructions($bot_id, $session_id);
9941 -
5708 + // Get system prompt instructions from options
5709 + $system_prompt_instructions = isset($this->options['system_prompt_instructions']) ? $this->options['system_prompt_instructions'] : '';
5710 +
9942 5711 // Add system prompt to relevant content
9943 5712 $content_with_instructions = $system_prompt_instructions . " " . $relevant_content;
9944 5713
9945 5714 // Format messages for Gemini API
@@ -10034,11 +5803,9 @@
10034 5803 ]
10035 5804 ]);
10036 5805
10037 5806 // Prepare the API endpoint
10038 - // Use v1beta for preview models (Gemini 3, experimental), v1 for stable models
10039 - $api_version = (strpos($selected_model, 'preview') !== false || strpos($selected_model, 'exp') !== false) ? 'v1beta' : 'v1';
10040 - $api_endpoint = 'https://generativelanguage.googleapis.com/' . $api_version . '/models/' . $selected_model . ':generateContent?key=' . $gemini_api_key;
5807 + $api_endpoint = 'https://generativelanguage.googleapis.com/v1/models/' . $selected_model . ':generateContent?key=' . $gemini_api_key;
10041 5808
10042 5809 // Set up the API request
10043 5810 $args = [
10044 5811 'body' => $body,
@@ -10052,10 +5819,10 @@
10052 5819 'sslverify' => true,
10053 5820 ];
10054 5821
10055 5822 // Make the API request
10056 - $response = $this->mxchat_provider_call_with_retry($api_endpoint, $args, 'gemini');
10057 -
5823 + $response = wp_remote_post($api_endpoint, $args);
5824 +
10058 5825 // Process the response
10059 5826 if (is_wp_error($response)) {
10060 5827 return "Sorry, there was an error processing your request: " . $response->get_error_message();
10061 5828 }
@@ -10077,12 +5844,11 @@
10077 5844 return "Sorry, I couldn't process that request. The response format was unexpected.";
10078 5845 }
10079 5846 }
10080 5847
10081 -
10082 5848 public function test_streaming_request() {
10083 5849 $options = get_option('mxchat_options', []);
10084 - $model = $options['model'] ?? 'gpt-5.1-chat-latest';
5850 + $model = $options['model'] ?? 'gpt-4o';
10085 5851
10086 5852 // Detect provider from model prefix
10087 5853 $provider = strtolower(explode('-', $model)[0]);
10088 5854
@@ -10208,8 +5974,9 @@
10208 5974
10209 5975 return true;
10210 5976 }
10211 5977
5978 +
10212 5979 public function mxchat_dismiss_pre_chat_message() {
10213 5980 // Get and sanitize the user identifier
10214 5981 $user_id = $this->mxchat_get_user_identifier();
10215 5982 $user_id = sanitize_key($user_id);
@@ -10263,62 +6030,36 @@
10263 6030
10264 6031 return $dotProduct / ($normA * $normB);
10265 6032 }
10266 6033
10267 -
10268 6034 public function mxchat_enqueue_scripts_styles() {
10269 - // Fetch options from the database first to check loading strategy
10270 - $this->options = get_option('mxchat_options');
10271 - $loading_strategy = isset($this->options['script_loading_strategy']) ? $this->options['script_loading_strategy'] : 'default';
10272 -
10273 - // Always enqueue CSS immediately
6035 + // Define version numbers for the styles and scripts
6036 + $chat_style_version = '2.4.0';
6037 + $chat_script_version = '2.4.0';
6038 + // Enqueue the script
6039 + wp_enqueue_script(
6040 + 'mxchat-chat-js',
6041 + plugin_dir_url(__FILE__) . '../js/chat-script.js',
6042 + array('jquery'),
6043 + $chat_script_version,
6044 + true
6045 + );
6046 + // Enqueue the CSS
10274 6047 wp_enqueue_style(
10275 6048 'mxchat-chat-css',
10276 6049 plugin_dir_url(__FILE__) . '../css/chat-style.css',
10277 6050 array(),
10278 - MXCHAT_VERSION
6051 + $chat_style_version
10279 6052 );
10280 -
10281 - // Handle script loading based on strategy
10282 - if ($loading_strategy === 'default' || $loading_strategy === 'defer') {
10283 - // Enqueue the script normally
10284 - wp_enqueue_script(
10285 - 'mxchat-chat-js',
10286 - plugin_dir_url(__FILE__) . '../js/chat-script.js',
10287 - array('jquery'),
10288 - MXCHAT_VERSION,
10289 - true
10290 - );
10291 -
10292 - // Add defer attribute if strategy is 'defer'
10293 - if ($loading_strategy === 'defer') {
10294 - wp_script_add_data('mxchat-chat-js', 'strategy', 'defer');
10295 - }
10296 - } else {
10297 - // For delay or interaction-based loading, we'll use a custom loader
10298 - // Don't enqueue the main script - we'll load it dynamically
10299 - add_action('wp_footer', array($this, 'mxchat_output_delayed_script_loader'), 99);
10300 - }
10301 -
6053 + // Fetch options from the database
6054 + $this->options = get_option('mxchat_options');
10302 6055 $prompts_options = get_option('mxchat_prompts_options', array());
10303 -
10304 - // Check if AI theme is active - if so, skip inline colors in JavaScript
10305 - $theme_options = get_option('mxchat_theme_options', array());
10306 - $ai_theme_active = !empty($theme_options['active_ai_theme_css']);
10307 - $has_bot_theme_assignments = !empty($theme_options['bot_theme_assignments']);
10308 - $skip_inline_colors = $ai_theme_active || $has_bot_theme_assignments;
10309 -
6056 +
10310 6057 // Prepare settings for JavaScript
10311 6058 $style_settings = array(
10312 6059 'ajax_url' => admin_url('admin-ajax.php'),
10313 - // The chat-send nonce is now fetched per-request from /wp-json/mxchat/v1/nonce
10314 - // (plan-6a68c9) so it never sits in cached HTML. We still emit a nonce here
10315 - // as a one-shot fallback for the first interaction on a fresh page load
10316 - // (so the very first chat-send doesn't need to wait for a REST round-trip),
10317 - // but the widget refetches before each subsequent send.
10318 - 'nonce' => wp_create_nonce('mxchat_chat_send'),
10319 - 'rest_url' => esc_url_raw(trailingslashit(rest_url('mxchat/v1'))),
10320 - 'model' => isset($this->options['model']) ? $this->options['model'] : 'gpt-5.1-chat-latest',
6060 + 'nonce' => wp_create_nonce('mxchat_chat_nonce'),
6061 + 'model' => isset($this->options['model']) ? $this->options['model'] : 'gpt-4o',
10321 6062 'enable_streaming_toggle' => isset($this->options['enable_streaming_toggle']) ? $this->options['enable_streaming_toggle'] : 'on',
10322 6063 'contextual_awareness_toggle' => isset($this->options['contextual_awareness_toggle']) ? $this->options['contextual_awareness_toggle'] : 'off',
10323 6064 'link_target_toggle' => $this->options['link_target_toggle'] ?? 'off',
10324 6065 'rate_limit_message' => $this->options['rate_limit_message'] ?? 'Rate limit exceeded. Please try again later.',
@@ -10345,175 +6086,15 @@
10345 6086 'use_pinecone' => $prompts_options['mxchat_use_pinecone'] ?? '0',
10346 6087 'email_collection_enabled' => $this->options['enable_email_block'] ?? 'off', // FIXED
10347 6088 'initial_email_state' => null, // Also fixed this undefined variable
10348 6089 'skip_email_check' => true,
10349 - 'pinecone_enabled' => isset($prompts_options['mxchat_use_pinecone']) && $prompts_options['mxchat_use_pinecone'] === '1',
10350 - 'skip_inline_colors' => $skip_inline_colors,
10351 - 'bot_theme_assignments' => $theme_options['bot_theme_assignments'] ?? array(),
10352 - 'print_button_enabled' => $this->options['print_button_enabled'] ?? 'on',
10353 - 'print_button_label' => esc_html__('Download Transcript', 'mxchat'),
10354 - 'print_header_title' => esc_html(get_bloginfo('name')) . ' — ' . esc_html__('Chat transcript', 'mxchat'),
10355 - 'satisfaction_rating_enabled' => apply_filters(
10356 - 'mxchat_satisfaction_rating_enabled',
10357 - ($this->options['satisfaction_rating_enabled'] ?? 'off') === 'on'
10358 - ),
10359 - 'satisfaction_rating_idle_seconds' => max(5, min(600, intval($this->options['satisfaction_rating_idle_seconds'] ?? 60))),
10360 - 'satisfaction_rating_copy' => array(
10361 - 'question' => !empty($this->options['satisfaction_rating_question']) ? esc_html($this->options['satisfaction_rating_question']) : esc_html__('Was this helpful?', 'mxchat'),
10362 - 'helpful' => esc_html__('Helpful', 'mxchat'),
10363 - 'not_helpful' => esc_html__('Not helpful', 'mxchat'),
10364 - 'dismiss' => esc_html__('Dismiss', 'mxchat'),
10365 - 'thanks' => !empty($this->options['satisfaction_rating_thanks']) ? esc_html($this->options['satisfaction_rating_thanks']) : esc_html__('Thanks! Anything we should improve? (optional)', 'mxchat'),
10366 - 'placeholder' => !empty($this->options['satisfaction_rating_placeholder']) ? esc_html($this->options['satisfaction_rating_placeholder']) : esc_html__('Tell us what could be better…', 'mxchat'),
10367 - 'send' => esc_html__('Send', 'mxchat'),
10368 - 'skip' => esc_html__('Skip', 'mxchat'),
10369 - 'saved' => !empty($this->options['satisfaction_rating_saved']) ? esc_html($this->options['satisfaction_rating_saved']) : esc_html__('Thanks for the feedback.', 'mxchat'),
10370 - ),
6090 + 'pinecone_enabled' => isset($prompts_options['mxchat_use_pinecone']) && $prompts_options['mxchat_use_pinecone'] === '1'
10371 6091 );
10372 -
10373 - // For normal/defer loading, use wp_localize_script
10374 - // For delayed loading, we store settings in a transient to be output inline
10375 - if ($loading_strategy === 'default' || $loading_strategy === 'defer') {
10376 - wp_localize_script('mxchat-chat-js', 'mxchatChat', $style_settings);
10377 - } else {
10378 - // Store settings for the delayed loader to use
10379 - set_transient('mxchat_delayed_settings_' . get_current_user_id(), $style_settings, 60);
10380 - }
6092 + // Pass the settings to the script
6093 + wp_localize_script('mxchat-chat-js', 'mxchatChat', $style_settings);
10381 6094 }
10382 6095
10383 -/**
10384 - * Output the delayed script loader for performance optimization
10385 - */
10386 -public function mxchat_output_delayed_script_loader() {
10387 - $this->options = get_option('mxchat_options');
10388 - $loading_strategy = isset($this->options['script_loading_strategy']) ? $this->options['script_loading_strategy'] : 'default';
10389 - $script_url = plugin_dir_url(__FILE__) . '../js/chat-script.js?ver=' . MXCHAT_VERSION;
10390 6096
10391 - // Get the stored settings
10392 - $prompts_options = get_option('mxchat_prompts_options', array());
10393 - $theme_options = get_option('mxchat_theme_options', array());
10394 - $ai_theme_active = !empty($theme_options['active_ai_theme_css']);
10395 - $has_bot_theme_assignments = !empty($theme_options['bot_theme_assignments']);
10396 - $skip_inline_colors = $ai_theme_active || $has_bot_theme_assignments;
10397 -
10398 - $style_settings = array(
10399 - 'ajax_url' => admin_url('admin-ajax.php'),
10400 - // Per-request nonce — see plan-6a68c9; widget fetches via /wp-json/mxchat/v1/nonce
10401 - // before each send. This inline value is a one-shot fallback for the first interaction.
10402 - 'nonce' => wp_create_nonce('mxchat_chat_send'),
10403 - 'rest_url' => esc_url_raw(trailingslashit(rest_url('mxchat/v1'))),
10404 - 'model' => isset($this->options['model']) ? $this->options['model'] : 'gpt-5.1-chat-latest',
10405 - 'enable_streaming_toggle' => isset($this->options['enable_streaming_toggle']) ? $this->options['enable_streaming_toggle'] : 'on',
10406 - 'contextual_awareness_toggle' => isset($this->options['contextual_awareness_toggle']) ? $this->options['contextual_awareness_toggle'] : 'off',
10407 - 'link_target_toggle' => $this->options['link_target_toggle'] ?? 'off',
10408 - 'rate_limit_message' => $this->options['rate_limit_message'] ?? 'Rate limit exceeded. Please try again later.',
10409 - 'complianz_toggle' => isset($this->options['complianz_toggle']) && $this->options['complianz_toggle'] === 'on',
10410 - 'user_message_bg_color' => $this->options['user_message_bg_color'] ?? '#fff',
10411 - 'user_message_font_color' => $this->options['user_message_font_color'] ?? '#212121',
10412 - 'bot_message_bg_color' => $this->options['bot_message_bg_color'] ?? '#212121',
10413 - 'bot_message_font_color' => $this->options['bot_message_font_color'] ?? '#fff',
10414 - 'top_bar_bg_color' => $this->options['top_bar_bg_color'] ?? '#212121',
10415 - 'send_button_font_color' => $this->options['send_button_font_color'] ?? '#212121',
10416 - 'close_button_color' => $this->options['close_button_color'] ?? '#fff',
10417 - 'chatbot_background_color' => $this->options['chatbot_background_color'] ?? '#212121',
10418 - 'chatbot_bg_color' => $this->options['chatbot_bg_color'] ?? '#fff',
10419 - 'icon_color' => $this->options['icon_color'] ?? '#fff',
10420 - 'chat_input_font_color' => $this->options['chat_input_font_color'] ?? '#212121',
10421 - 'chat_persistence_toggle' => $this->options['chat_persistence_toggle'] ?? 'off',
10422 - 'appendWidgetToBody' => $this->options['append_to_body'] ?? 'off',
10423 - 'live_agent_message_bg_color' => $this->options['live_agent_message_bg_color'] ?? '#ffffff',
10424 - 'live_agent_message_font_color' => $this->options['live_agent_message_font_color'] ?? '#333333',
10425 - 'chat_toolbar_toggle' => $this->options['chat_toolbar_toggle'] ?? 'off',
10426 - 'mode_indicator_bg_color' => $this->options['mode_indicator_bg_color'] ?? '#767676',
10427 - 'mode_indicator_font_color' => $this->options['mode_indicator_font_color'] ?? '#ffffff',
10428 - 'toolbar_icon_color' => $this->options['toolbar_icon_color'] ?? '#212121',
10429 - 'use_pinecone' => $prompts_options['mxchat_use_pinecone'] ?? '0',
10430 - 'email_collection_enabled' => $this->options['enable_email_block'] ?? 'off',
10431 - 'initial_email_state' => null,
10432 - 'skip_email_check' => true,
10433 - 'pinecone_enabled' => isset($prompts_options['mxchat_use_pinecone']) && $prompts_options['mxchat_use_pinecone'] === '1',
10434 - 'skip_inline_colors' => $skip_inline_colors,
10435 - 'bot_theme_assignments' => $theme_options['bot_theme_assignments'] ?? array(),
10436 - 'print_button_enabled' => $this->options['print_button_enabled'] ?? 'on',
10437 - 'print_button_label' => esc_html__('Download Transcript', 'mxchat'),
10438 - 'print_header_title' => esc_html(get_bloginfo('name')) . ' — ' . esc_html__('Chat transcript', 'mxchat'),
10439 - 'satisfaction_rating_enabled' => apply_filters(
10440 - 'mxchat_satisfaction_rating_enabled',
10441 - ($this->options['satisfaction_rating_enabled'] ?? 'off') === 'on'
10442 - ),
10443 - 'satisfaction_rating_idle_seconds' => max(5, min(600, intval($this->options['satisfaction_rating_idle_seconds'] ?? 60))),
10444 - 'satisfaction_rating_copy' => array(
10445 - 'question' => !empty($this->options['satisfaction_rating_question']) ? esc_html($this->options['satisfaction_rating_question']) : esc_html__('Was this helpful?', 'mxchat'),
10446 - 'helpful' => esc_html__('Helpful', 'mxchat'),
10447 - 'not_helpful' => esc_html__('Not helpful', 'mxchat'),
10448 - 'dismiss' => esc_html__('Dismiss', 'mxchat'),
10449 - 'thanks' => !empty($this->options['satisfaction_rating_thanks']) ? esc_html($this->options['satisfaction_rating_thanks']) : esc_html__('Thanks! Anything we should improve? (optional)', 'mxchat'),
10450 - 'placeholder' => !empty($this->options['satisfaction_rating_placeholder']) ? esc_html($this->options['satisfaction_rating_placeholder']) : esc_html__('Tell us what could be better…', 'mxchat'),
10451 - 'send' => esc_html__('Send', 'mxchat'),
10452 - 'skip' => esc_html__('Skip', 'mxchat'),
10453 - 'saved' => !empty($this->options['satisfaction_rating_saved']) ? esc_html($this->options['satisfaction_rating_saved']) : esc_html__('Thanks for the feedback.', 'mxchat'),
10454 - ),
10455 - );
10456 -
10457 - // Determine delay time based on strategy
10458 - $delay_ms = 0;
10459 - switch ($loading_strategy) {
10460 - case 'delay_1s':
10461 - $delay_ms = 1000;
10462 - break;
10463 - case 'delay_3s':
10464 - $delay_ms = 3000;
10465 - break;
10466 - case 'delay_5s':
10467 - $delay_ms = 5000;
10468 - break;
10469 - }
10470 -
10471 - ?>
10472 - <script type="text/javascript">
10473 - (function() {
10474 - var mxchatLoaded = false;
10475 - var mxchatChat = <?php echo wp_json_encode($style_settings); ?>;
10476 - window.mxchatChat = mxchatChat;
10477 -
10478 - function loadMxChatScript() {
10479 - if (mxchatLoaded) return;
10480 - mxchatLoaded = true;
10481 -
10482 - function appendChatScript() {
10483 - var script = document.createElement('script');
10484 - script.src = <?php echo wp_json_encode($script_url); ?>;
10485 - script.type = 'text/javascript';
10486 - document.body.appendChild(script);
10487 - }
10488 -
10489 - if (typeof jQuery !== 'undefined') {
10490 - appendChatScript();
10491 - } else {
10492 - var jq = document.createElement('script');
10493 - jq.src = <?php echo wp_json_encode(includes_url('js/jquery/jquery.min.js')); ?>;
10494 - jq.onload = appendChatScript;
10495 - document.body.appendChild(jq);
10496 - }
10497 - }
10498 -
10499 - <?php if ($loading_strategy === 'on_interaction'): ?>
10500 - // Load on user interaction
10501 - var events = ['scroll', 'mousemove', 'touchstart', 'keydown', 'click'];
10502 - events.forEach(function(evt) {
10503 - window.addEventListener(evt, loadMxChatScript, {once: true, passive: true});
10504 - });
10505 - // Fallback: load after 8 seconds if no interaction
10506 - setTimeout(loadMxChatScript, 8000);
10507 - <?php else: ?>
10508 - // Load after specified delay
10509 - setTimeout(loadMxChatScript, <?php echo intval($delay_ms); ?>);
10510 - <?php endif; ?>
10511 - })();
10512 - </script>
10513 - <?php
10514 -}
10515 -
10516 6097 /**
10517 6098 * Setup the cron jobs for rate limits with guard against multiple calls
10518 6099 */
10519 6100 public function setup_rate_limit_cron_jobs() {
@@ -10651,9 +6232,9 @@
10651 6232 update_option('mxchat_next_rate_limit_check', time() + $check_interval);
10652 6233 }
10653 6234 }
10654 6235 /**
10655 - * Enhanced rate limit check that includes fallback cleanup and bot-specific rate limits
6236 + * Enhanced rate limit check that includes fallback cleanup
10656 6237 */
10657 6238 public function check_rate_limit() {
10658 6239 // Check if we need to run fallback cleanup
10659 6240 $use_fallback = get_option('mxchat_use_fallback_rate_limits', false);
@@ -10663,66 +6244,11 @@
10663 6244 $this->mxchat_reset_rate_limits();
10664 6245 update_option('mxchat_next_rate_limit_check', time() + 3600); // Next hour
10665 6246 }
10666 6247
10667 - // Get bot ID from current request context
10668 - $bot_id = isset($_POST['bot_id']) ? sanitize_key($_POST['bot_id']) : 'default';
6248 + // Continue with your existing rate limit logic...
6249 + $all_options = get_option('mxchat_options', []);
10669 6250
10670 - // Get bot-specific options (includes rate limits if overridden)
10671 - $bot_options = $this->get_bot_options($bot_id);
10672 - $current_options = !empty($bot_options) ? $bot_options : $this->options;
10673 -
10674 - // Use bot-specific rate limits if available, otherwise fall back to default
10675 - $rate_limits_source = isset($current_options['rate_limits']) ? $current_options['rate_limits'] : get_option('mxchat_options', [])['rate_limits'] ?? [];
10676 -
10677 - // -------------------------------------------------------------------
10678 - // Whole-chatbot global cap (independent of role). Evaluated FIRST so
10679 - // it acts as a hard ceiling across all users + all roles. Default is
10680 - // 'unlimited' so existing installs are unchanged. Counter key drops
10681 - // both <role> and <user_id> segments — single pool per bot.
10682 - // -------------------------------------------------------------------
10683 - $global_cfg = isset($current_options['rate_limits_global']) && is_array($current_options['rate_limits_global'])
10684 - ? $current_options['rate_limits_global']
10685 - : (isset(get_option('mxchat_options', [])['rate_limits_global']) ? get_option('mxchat_options', [])['rate_limits_global'] : []);
10686 - $global_limit_raw = isset($global_cfg['limit']) ? (string) $global_cfg['limit'] : 'unlimited';
10687 - $global_timeframe = isset($global_cfg['timeframe']) ? (string) $global_cfg['timeframe'] : 'daily';
10688 - if ($global_limit_raw !== '' && $global_limit_raw !== 'unlimited' && (int) $global_limit_raw >= 1) {
10689 - $bot_id_for_global = isset($_POST['bot_id']) ? sanitize_key($_POST['bot_id']) : 'default';
10690 - $safe_bot_global = preg_replace('/[^a-zA-Z0-9_]/', '_', $bot_id_for_global);
10691 - $global_option = 'mxchat_chat_limit_' . $safe_bot_global . '_global';
10692 - $global_data = get_option($global_option, ['count' => 0, 'timestamp' => time()]);
10693 - if ((int) $global_data['count'] === 0) {
10694 - $global_data['timestamp'] = time();
10695 - update_option($global_option, $global_data);
10696 - }
10697 - $now = time();
10698 - $ts = (int) $global_data['timestamp'];
10699 - $reset = false;
10700 - switch ($global_timeframe) {
10701 - case 'hourly': $reset = ($now - $ts) >= 3600; break;
10702 - case 'daily': $reset = ($now - $ts) >= 86400; break;
10703 - case 'weekly': $reset = ($now - $ts) >= 604800; break;
10704 - case 'monthly': $reset = ($now - $ts) >= 2592000; break;
10705 - }
10706 - if ($reset) {
10707 - $global_data = ['count' => 0, 'timestamp' => $now];
10708 - update_option($global_option, $global_data);
10709 - }
10710 - if ((int) $global_data['count'] >= (int) $global_limit_raw) {
10711 - $global_msg = !empty($global_cfg['message'])
10712 - ? $global_cfg['message']
10713 - : __('This chatbot has reached its message limit. Please try again later.', 'mxchat');
10714 - return [
10715 - 'error' => true,
10716 - 'message' => $this->process_rate_limit_message_html($global_msg),
10717 - ];
10718 - }
10719 - // Reserve the slot for this request. Per-role check below also increments
10720 - // its own counter — that is intentional, both ceilings apply independently.
10721 - $global_data['count']++;
10722 - update_option($global_option, $global_data);
10723 - }
10724 -
10725 6251 // Determine user role or if logged out
10726 6252 if (is_user_logged_in()) {
10727 6253 $user = wp_get_current_user();
10728 6254 $user_id = $user->ID;
@@ -10742,13 +6268,13 @@
10742 6268 $user_id = $this->get_client_ip();
10743 6269 }
10744 6270
10745 6271 // Check if rate limits are configured for this role
10746 - if (!isset($rate_limits_source[$role])) {
6272 + if (!isset($all_options['rate_limits'][$role])) {
10747 6273 return true; // No limit set for this role
10748 6274 }
10749 6275
10750 - $limit = $rate_limits_source[$role]['limit'];
6276 + $limit = $all_options['rate_limits'][$role]['limit'];
10751 6277
10752 6278 // If unlimited, return true immediately
10753 6279 if ($limit === 'unlimited') {
10754 6280 return true;
@@ -10753,16 +6279,13 @@
10753 6279 if ($limit === 'unlimited') {
10754 6280 return true;
10755 6281 }
10756 6282
10757 - // Get the option name for this user/role with safer naming (include bot_id for bot-specific limits)
6283 + // Get the option name for this user/role with safer naming
10758 6284 $safe_role = preg_replace('/[^a-zA-Z0-9_]/', '_', $role);
10759 6285 $safe_user_id = preg_replace('/[^a-zA-Z0-9_]/', '_', $user_id);
10760 - $safe_bot_id = preg_replace('/[^a-zA-Z0-9_]/', '_', $bot_id);
6286 + $option_name = 'mxchat_chat_limit_' . $safe_role . '_' . $safe_user_id;
10761 6287
10762 - // Include bot_id in option name so each bot has separate rate limits
10763 - $option_name = 'mxchat_chat_limit_' . $safe_bot_id . '_' . $safe_role . '_' . $safe_user_id;
10764 -
10765 6288 // Get the counter data
10766 6289 $limit_data = get_option($option_name, ['count' => 0, 'timestamp' => time()]);
10767 6290
10768 6291 // If first request or counter reset needed, set the initial timestamp
@@ -10771,10 +6294,10 @@
10771 6294 update_option($option_name, $limit_data);
10772 6295 }
10773 6296
10774 6297 // Get the timeframe
10775 - $timeframe = isset($rate_limits_source[$role]['timeframe']) ?
10776 - $rate_limits_source[$role]['timeframe'] : 'daily';
6298 + $timeframe = isset($all_options['rate_limits'][$role]['timeframe']) ?
6299 + $all_options['rate_limits'][$role]['timeframe'] : 'daily';
10777 6300
10778 6301 // Check if the counter needs to be reset based on timeframe
10779 6302 $current_time = time();
10780 6303 $timestamp = $limit_data['timestamp'];
@@ -10803,10 +6326,10 @@
10803 6326
10804 6327 // Check if user has exceeded their limit
10805 6328 if ($limit_data['count'] >= intval($limit)) {
10806 6329 // Get the custom message for this role
10807 - $message = !empty($rate_limits_source[$role]['message'])
10808 - ? $rate_limits_source[$role]['message']
6330 + $message = !empty($all_options['rate_limits'][$role]['message'])
6331 + ? $all_options['rate_limits'][$role]['message']
10809 6332 : __('Rate limit exceeded. Please try again later.', 'mxchat');
10810 6333
10811 6334 // Add timeframe information to the message if placeholders exist
10812 6335 $timeframe_label = '';
@@ -11096,11 +6619,8 @@
11096 6619
11097 6620 /**
11098 6621 * AJAX handler to get system information for testing panel
11099 6622 */
11100 -/**
11101 - * AJAX handler to get system information for testing panel
11102 - */
11103 6623 public function mxchat_get_system_info() {
11104 6624 // Verify nonce for security
11105 6625 if (!wp_verify_nonce($_POST['nonce'], 'mxchat_test_nonce')) {
11106 6626 wp_send_json_error(['message' => 'Invalid nonce']);
@@ -11118,24 +6638,10 @@
11118 6638 ? $this->options['system_prompt_instructions']
11119 6639 : 'No system prompt configured';
11120 6640
11121 6641 // Get selected model
11122 - $selected_model = isset($this->options['model']) ? $this->options['model'] : 'gpt-5.1-chat-latest';
6642 + $selected_model = isset($this->options['model']) ? $this->options['model'] : 'gpt-4o';
11123 6643
11124 - // Check if OpenRouter is being used
11125 - $is_openrouter = ($selected_model === 'openrouter');
11126 - $openrouter_model = '';
11127 -
11128 - if ($is_openrouter) {
11129 - // Get the actual OpenRouter model that's selected
11130 - $openrouter_model = isset($this->options['openrouter_selected_model'])
11131 - ? $this->options['openrouter_selected_model']
11132 - : 'No OpenRouter model selected';
11133 -
11134 - // Update selected_model display to show both
11135 - $selected_model = 'OpenRouter: ' . $openrouter_model;
11136 - }
11137 -
11138 6644 // Get API key status (just check if they exist, don't expose the keys)
11139 6645 $api_status = [];
11140 6646 $api_status['openai'] = !empty($this->options['api_key']);
11141 6647 $api_status['claude'] = !empty($this->options['claude_api_key']);
@@ -11141,15 +6647,12 @@
11141 6647 $api_status['claude'] = !empty($this->options['claude_api_key']);
11142 6648 $api_status['gemini'] = !empty($this->options['gemini_api_key']);
11143 6649 $api_status['xai'] = !empty($this->options['xai_api_key']);
11144 6650 $api_status['deepseek'] = !empty($this->options['deepseek_api_key']);
11145 - $api_status['openrouter'] = !empty($this->options['openrouter_api_key']);
11146 6651
11147 6652 wp_send_json_success([
11148 6653 'system_prompt' => $system_prompt,
11149 6654 'selected_model' => $selected_model,
11150 - 'is_openrouter' => $is_openrouter,
11151 - 'openrouter_model' => $openrouter_model,
11152 6655 'api_status' => $api_status
11153 6656 ]);
11154 6657 }
11155 6658
@@ -11168,12 +6671,12 @@
11168 6671 wp_send_json_error(['message' => 'Unauthorized']);
11169 6672 return;
11170 6673 }
11171 6674
11172 - // Get similarity threshold from main options (default 35%)
6675 + // Get similarity threshold from main options (default 75%)
11173 6676 $similarity_threshold = isset($this->options['similarity_threshold'])
11174 6677 ? ((int) $this->options['similarity_threshold']) / 100
11175 - : 0.35;
6678 + : 0.75;
11176 6679
11177 6680 wp_send_json_success([
11178 6681 'threshold' => $similarity_threshold,
11179 6682 'threshold_percentage' => ($similarity_threshold * 100) . '%'
@@ -11188,42 +6691,24 @@
11188 6691 if (!wp_verify_nonce($_POST['nonce'], 'mxchat_test_nonce')) {
11189 6692 wp_send_json_error(['message' => 'Invalid nonce']);
11190 6693 return;
11191 6694 }
11192 -
6695 +
11193 6696 // Only allow admin users
11194 6697 if (!current_user_can('administrator')) {
11195 6698 wp_send_json_error(['message' => 'Unauthorized']);
11196 6699 return;
11197 6700 }
11198 -
11199 - // Check OpenAI Vector Store first (takes priority)
11200 - $vectorstore_options = get_option('mxchat_openai_vectorstore_options', array());
11201 - $use_vectorstore = (isset($vectorstore_options['mxchat_use_openai_vectorstore']) && $vectorstore_options['mxchat_use_openai_vectorstore'] === '1');
11202 -
11203 - if ($use_vectorstore) {
11204 - $vectorstore_ids = $vectorstore_options['mxchat_openai_vectorstore_ids'] ?? '';
11205 - $id_count = !empty($vectorstore_ids) ? count(array_filter(array_map('trim', explode(',', $vectorstore_ids)))) : 0;
11206 -
11207 - $kb_info = [
11208 - 'type' => 'OpenAI Vector Store',
11209 - 'status' => 'Active',
11210 - 'documents' => $id_count > 0 ? $id_count . ' vector store' . ($id_count > 1 ? 's' : '') . ' configured' : 'No vector stores configured'
11211 - ];
11212 -
11213 - wp_send_json_success($kb_info);
11214 - return;
11215 - }
11216 -
6701 +
11217 6702 // Check Pinecone vs WordPress
11218 6703 $addon_options = get_option('mxchat_pinecone_addon_options', array());
11219 6704 $use_pinecone = (isset($addon_options['mxchat_use_pinecone']) && $addon_options['mxchat_use_pinecone'] === '1');
11220 -
6705 +
11221 6706 $kb_info = [
11222 6707 'type' => $use_pinecone ? 'Pinecone' : 'WordPress Database',
11223 6708 'status' => 'Active'
11224 6709 ];
11225 -
6710 +
11226 6711 // Get document count
11227 6712 if ($use_pinecone) {
11228 6713 $kb_info['documents'] = 'Connected to Pinecone';
11229 6714 $kb_info['api_configured'] = !empty($addon_options['mxchat_pinecone_api_key']);
@@ -11233,9 +6718,9 @@
11233 6718 $table_name = $wpdb->prefix . 'mxchat_system_prompt_content';
11234 6719 $count = $wpdb->get_var("SELECT COUNT(*) FROM {$table_name}");
11235 6720 $kb_info['documents'] = $count ? $count . ' documents' : 'No documents';
11236 6721 }
11237 -
6722 +
11238 6723 wp_send_json_success($kb_info);
11239 6724 }
11240 6725
11241 6726 /**
@@ -11317,13 +6802,9 @@
11317 6802 // Clear any other session-specific transients
11318 6803 delete_transient("mxchat_waiting_for_pdf_url_{$session_id}");
11319 6804 delete_transient("mxchat_include_pdf_in_context_{$session_id}");
11320 6805 delete_transient("mxchat_include_word_in_context_{$session_id}");
11321 -
11322 - // Clear form addon state (pending forms and submitted forms)
11323 - delete_option("mxchat_pending_form_{$session_id}");
11324 - delete_option("mxchat_submitted_forms_{$session_id}");
11325 -
6806 +
11326 6807 //error_log("MxChat: Cleared all data for session: {$session_id}");
11327 6808 }
11328 6809
11329 6810 /**
@@ -11358,15 +6839,15 @@
11358 6839 $testing_data = [
11359 6840 'query' => $message,
11360 6841 'timestamp' => time(),
11361 6842 'top_matches' => [],
11362 - 'action_matches' => [] // Add action matches
6843 + 'action_matches' => [] // NEW: Add action matches
11363 6844 ];
11364 6845
11365 6846 // Get similarity threshold
11366 6847 $similarity_threshold = isset($this->options['similarity_threshold'])
11367 6848 ? ((int) $this->options['similarity_threshold']) / 100
11368 - : 0.35;
6849 + : 0.75;
11369 6850
11370 6851 $testing_data['similarity_threshold'] = $similarity_threshold;
11371 6852
11372 6853 // Use the real similarity analysis if available
@@ -11381,9 +6862,9 @@
11381 6862
11382 6863 $testing_data['knowledge_base_type'] = $use_pinecone ? 'Pinecone' : 'WordPress Database';
11383 6864 }
11384 6865
11385 - // Include action analysis if available
6866 + // NEW: Include action analysis if available
11386 6867 if (isset($this->last_action_analysis) && !empty($this->last_action_analysis)) {
11387 6868 $testing_data['action_matches'] = $this->last_action_analysis;
11388 6869
11389 6870 // Clear it after capturing to avoid stale data
@@ -11394,13 +6875,13 @@
11394 6875 }
11395 6876
11396 6877
11397 6878 /**
11398 - * Track URL clicks from chatbot responses
6879 + * NEW: Track URL clicks from chatbot responses
11399 6880 */
11400 6881 public function mxchat_track_url_click() {
11401 6882 // Verify nonce for security
11402 - if (!isset($_POST['nonce']) || !MxChat_Integrator::mxchat_verify_chat_send_nonce($_POST['nonce'])) {
6883 + if (!isset($_POST['nonce']) || !wp_verify_nonce($_POST['nonce'], 'mxchat_chat_nonce')) {
11403 6884 wp_send_json_error(['message' => 'Invalid nonce']);
11404 6885 wp_die();
11405 6886 }
11406 6887
@@ -11433,9 +6914,9 @@
11433 6914 wp_die();
11434 6915 }
11435 6916
11436 6917 /**
11437 - * Get URL click analytics for a session
6918 + * NEW: Get URL click analytics for a session
11438 6919 */
11439 6920 public function mxchat_get_url_clicks($session_id) {
11440 6921 global $wpdb;
11441 6922 $table_name = $wpdb->prefix . 'mxchat_url_clicks';
@@ -11447,13 +6928,13 @@
11447 6928
11448 6929 return $clicks;
11449 6930 }
11450 6931 /**
11451 - * Track the originating page where chat was started
6932 + * NEW: Track the originating page where chat was started
11452 6933 */
11453 6934 public function mxchat_track_originating_page() {
11454 6935 // Verify nonce
11455 - if (!isset($_POST['nonce']) || !MxChat_Integrator::mxchat_verify_chat_send_nonce($_POST['nonce'])) {
6936 + if (!isset($_POST['nonce']) || !wp_verify_nonce($_POST['nonce'], 'mxchat_chat_nonce')) {
11456 6937 wp_send_json_error(['message' => 'Invalid nonce']);
11457 6938 wp_die();
11458 6939 }
11459 6940
@@ -11498,169 +6979,8 @@
11498 6979 wp_send_json_success(['message' => 'Originating page tracked']);
11499 6980 wp_die();
11500 6981 }
11501 6982
11502 -/**
11503 - * Validate and clean URLs from AI response
11504 - * Removes any URLs that aren't in the knowledge base
11505 - *
11506 - * @param string $response_text The AI-generated response
11507 - * @param array $valid_urls Array of URLs from the knowledge base
11508 - * @return string Cleaned response with invalid URLs removed/flagged
11509 - */
11510 -private function validate_and_clean_urls($response_text, $valid_urls) {
11511 - // DEBUG: Log what we're working with
11512 - //error_log("=== MxChat URL Validation Debug ===");
11513 - //error_log("Valid URLs count: " . count($valid_urls));
11514 - //error_log("Valid URLs: " . print_r($valid_urls, true));
11515 - //error_log("Response text length: " . strlen($response_text));
11516 - //error_log("Response text preview: " . substr($response_text, 0, 500));
11517 -
11518 - // If no valid URLs provided or empty response, return as-is
11519 - if (empty($valid_urls) || empty($response_text)) {
11520 - //error_log("Validation skipped - empty valid_urls or response");
11521 - return $response_text;
11522 - }
11523 -
11524 - // Extract all URLs from the AI response
11525 - // This regex matches http:// and https:// URLs
11526 - preg_match_all(
11527 - '#\bhttps?://[^\s<>"\')\]]+#i',
11528 - $response_text,
11529 - $matches
11530 - );
11531 -
11532 - // If no URLs found in response, return as-is
11533 - if (empty($matches[0])) {
11534 - //error_log("No URLs found in response");
11535 - return $response_text;
11536 - }
11537 -
11538 - $found_urls = $matches[0];
11539 - $cleaned_response = $response_text;
11540 - $removed_count = 0;
11541 -
11542 - // Normalize valid URLs for comparison (remove trailing slashes, fragments, etc.)
11543 - $normalized_valid_urls = array_map(function($url) {
11544 - // Remove trailing slash
11545 - $url = rtrim($url, '/');
11546 - // Remove URL fragments (#section)
11547 - $url = preg_replace('/#.*$/', '', $url);
11548 - // Remove trailing punctuation that might have been captured
11549 - $url = rtrim($url, '.,;:!?');
11550 - return $url;
11551 - }, $valid_urls);
11552 -
11553 - //error_log("Normalized valid URLs: " . print_r($normalized_valid_urls, true));
11554 -
11555 - foreach ($found_urls as $found_url) {
11556 - // Clean up the found URL (remove trailing punctuation that might have been captured)
11557 - $clean_found_url = rtrim($found_url, '.,;:!?)');
11558 -
11559 - // DEBUG: Log each URL being checked
11560 - //error_log("Checking found URL: " . $found_url);
11561 -
11562 - // Normalize for comparison
11563 - $normalized_found = rtrim($clean_found_url, '/');
11564 - $normalized_found = preg_replace('/#.*$/', '', $normalized_found);
11565 -
11566 - //error_log("Normalized found URL: " . $normalized_found);
11567 -
11568 - // Check if this URL exists in our valid URLs list
11569 - $is_valid = false;
11570 -
11571 - //error_log("Starting validation checks for: " . $normalized_found);
11572 -
11573 - // First, try exact match
11574 - if (in_array($normalized_found, $normalized_valid_urls)) {
11575 - $is_valid = true;
11576 - //error_log("EXACT MATCH FOUND");
11577 - } else {
11578 - //error_log("No exact match, checking variations...");
11579 - // If no exact match, check if it's a variation (with query params, etc.)
11580 - foreach ($normalized_valid_urls as $valid_url) {
11581 - //error_log(" Comparing against valid URL: " . $valid_url);
11582 -
11583 - // Check if the found URL starts with a valid URL (handles query params)
11584 - if (strpos($normalized_found, $valid_url) === 0) {
11585 - // Check what comes after the valid URL
11586 - $remainder = substr($normalized_found, strlen($valid_url));
11587 -
11588 - // Only valid if:
11589 - // 1. Exact match (remainder is empty)
11590 - // 2. Query params (starts with ?)
11591 - // 3. Fragment (starts with #)
11592 - if (empty($remainder) || $remainder[0] === '?' || $remainder[0] === '#') {
11593 - $is_valid = true;
11594 - //error_log(" MATCH: Found URL is valid variation of base URL");
11595 - break;
11596 - } else {
11597 - //error_log(" NOT A MATCH: Found URL extends path beyond valid URL (remainder: " . $remainder . ")");
11598 - }
11599 - }
11600 - // Also check the reverse (in case valid URL has query params)
11601 - if (strpos($valid_url, $normalized_found) === 0) {
11602 - $is_valid = true;
11603 - //error_log(" MATCH: Valid URL starts with found URL");
11604 - break;
11605 - }
11606 - }
11607 -
11608 - if (!$is_valid) {
11609 - //error_log("NO MATCH FOUND - URL should be removed");
11610 - }
11611 - }
11612 -
11613 - // If URL is not valid, remove it from the response
11614 - if (!$is_valid) {
11615 - // Log the removal for debugging
11616 - //error_log("MxChat: Removed hallucinated URL: " . $found_url);
11617 - //error_log("MxChat: Valid URLs were: " . implode(', ', array_slice($normalized_valid_urls, 0, 5)));
11618 -
11619 - $removed_count++;
11620 -
11621 - // Check if URL is part of a markdown link: [text](url)
11622 - $markdown_pattern = '/\[([^\]]+)\]\(' . preg_quote($found_url, '/') . '\)/';
11623 - if (preg_match($markdown_pattern, $cleaned_response)) {
11624 - //error_log("Found markdown link, removing but keeping text");
11625 - // Remove the markdown link but keep the text
11626 - $cleaned_response = preg_replace($markdown_pattern, '$1', $cleaned_response);
11627 - }
11628 - // Check if URL is part of an HTML link: <a href="url">text</a>
11629 - else if (preg_match('/<a[^>]*href=["\']' . preg_quote($found_url, '/') . '["\'][^>]*>(.*?)<\/a>/i', $cleaned_response, $link_match)) {
11630 - //error_log("Found HTML link, removing but keeping text");
11631 - // Remove the HTML link but keep the text
11632 - $link_text = $link_match[1];
11633 - $cleaned_response = preg_replace(
11634 - '/<a[^>]*href=["\']' . preg_quote($found_url, '/') . '["\'][^>]*>.*?<\/a>/i',
11635 - $link_text,
11636 - $cleaned_response
11637 - );
11638 - }
11639 - // Otherwise just remove the bare URL
11640 - else {
11641 - //error_log("Removing bare URL");
11642 - $cleaned_response = str_replace($found_url, '', $cleaned_response);
11643 - }
11644 - }
11645 - }
11646 -
11647 - // Log summary if any URLs were removed
11648 - if ($removed_count > 0) {
11649 - //error_log("MxChat: URL Validation Summary - Removed {$removed_count} hallucinated URL(s)");
11650 - } else {
11651 - //error_log("MxChat: URL Validation Summary - No URLs removed, all were valid");
11652 - }
11653 -
11654 - // Clean up any double spaces or awkward punctuation left behind
11655 - // IMPORTANT: Only collapse horizontal whitespace (spaces/tabs), preserve newlines for markdown formatting
11656 - $cleaned_response = preg_replace('/[^\S\n]+/', ' ', $cleaned_response); // Collapse spaces/tabs but NOT newlines
11657 - $cleaned_response = preg_replace('/[^\S\n]+([.,;:!?])/', '$1', $cleaned_response); // Same for punctuation cleanup
11658 -
11659 - //error_log("Final cleaned response: " . $cleaned_response);
11660 -
11661 - return trim($cleaned_response);
11662 -}
11663 6983
11664 6984 /**
11665 6985 * AJAX handler to get current chat mode for a session
11666 6986 */
@@ -11665,9 +6985,9 @@
11665 6985 * AJAX handler to get current chat mode for a session
11666 6986 */
11667 6987 public function mxchat_get_current_chat_mode() {
11668 6988 // Verify nonce for security
11669 - if (!isset($_POST['nonce']) || !MxChat_Integrator::mxchat_verify_chat_send_nonce($_POST['nonce'])) {
6989 + if (!isset($_POST['nonce']) || !wp_verify_nonce($_POST['nonce'], 'mxchat_chat_nonce')) {
11670 6990 wp_send_json_error(['message' => 'Invalid nonce']);
11671 6991 wp_die();
11672 6992 }
11673 6993