PluginProbe
MxChat – AI Chatbot & Content Generation for WordPress / 1.0.8
MxChat – AI Chatbot & Content Generation for WordPress v1.0.8
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 +269 -13465 3.2.161.0.8 View file →
@@ -4,5926 +4,261 @@
4 4 }
5 5
6 6 class MxChat_Integrator {
7 7 private $options;
8 - private $prompts_options;
9 8 private $chat_count;
10 - private $fallbackResponse;
11 - private $productCardHtml;
12 - // plan-mxchat-20260717-03ba33 — consent-safe YouTube embed queued during RAG
13 - // retrieval when a video-backed KB entry is used as context. Emitted on the
14 - // response 'html' channel alongside productCardHtml (non-streaming path,
15 - // same constraint as product cards).
16 - private $videoEmbedHtml = '';
17 - // plan-mxchat-20260617-48a57a — function-calling UI payload capture. When a
18 - // model-invoked tool yields a UI element (generated image, woo product card,
19 - // image-search gallery), the FC loop stashes its html here so the FC outcome
20 - // handler can SURFACE it to the frontend the same way the intent path does,
21 - // instead of stripping it to text for the model (the bug: UI-bearing actions
22 - // rendered nothing under function calling).
23 - private $fc_ui_html = '';
24 - private $fc_ui_images = array();
25 - private $fc_ui_captured = false;
26 - // plan-mxchat-20260722-59bc1b — {context} placeholder support. When the
27 - // owner's system instructions carry {context}, the assembled KB block is
28 - // stashed here (instead of being appended to $context_content) and
29 - // get_system_instructions() injects it at the token's position. Null until
30 - // the per-turn KB assembly has run — the early URL-extraction call to
31 - // get_system_instructions() must NOT consume the token.
32 - private $context_kb_block = null;
33 - private $word_handler;
34 - private $last_similarity_analysis = null;
35 - private $current_valid_urls = [];
36 - private $last_vectorstore_error = null;
37 - private $is_streaming = false; // ADDED: Track if current request is streaming
38 - private $streaming_headers_sent = false; // Track if streaming headers have been sent
39 - private $pending_originating_page = null; // Originating page captured at session start, consumed on row insert
40 - private $current_action_instruction = null; // Success-message instruction injected into the next system context
41 - private $last_action_analysis = null; // Last action-match analysis for testing_data payloads
42 9
43 -/**
44 - * Setup streaming headers - call this right before actually streaming
45 - * This delays header setup to allow actions/forms to return JSON responses
46 - */
47 -/**
48 - * Auto-retry wrapper around wp_remote_post for chat-send provider calls.
49 - *
50 - * Retries up to twice (750ms then 2000ms backoff) when the upstream provider
51 - * returns a TRANSIENT error: WP timeout, 429, 502, 503, 504, or a provider-
52 - * specific "overloaded" / "rate limit" body string. Returns immediately on
53 - * permanent errors (401/403/404/422) so misconfiguration surfaces fast.
54 - *
55 - * Drop-in replacement for wp_remote_post — returns the same shape
56 - * (WP_Error or response array) so the caller's existing error-handling
57 - * code path is unchanged.
58 - *
59 - * STREAMING PATH NOTE: this helper is ONLY for non-streaming chat-send
60 - * paths (the *_response_openai / *_response_claude / etc functions).
61 - * For the *_stream variants, the cURL initial-connect happens inside a
62 - * read-chunks loop — retrying there safely (without re-emitting partial
63 - * stream chunks to the client) is a separate problem. Streaming paths
64 - * are NOT wrapped in this build; tracked as a follow-on.
65 - *
66 - * Honors the `mxchat_options['auto_retry_on_transient_error']` toggle
67 - * (default true). When false, behavior is identical to plain wp_remote_post.
68 - */
69 -private function mxchat_provider_call_with_retry($url, $args, $provider_hint = '') {
70 - $opts = is_array($this->options ?? null) ? $this->options : array();
71 - $enabled = !isset($opts['auto_retry_on_transient_error']) ||
72 - (string) $opts['auto_retry_on_transient_error'] !== '0';
10 +public function __construct() {
11 + $this->options = get_option('mxchat_options');
12 + $this->chat_count = get_option('mxchat_chat_count', 0);
73 13
74 - if (!$enabled) {
75 - return wp_remote_post($url, $args);
76 - }
14 + // Add WooCommerce hooks
15 + add_action('wp_insert_post', array($this, 'mxchat_handle_product_change'), 10, 3);
77 16
78 - $backoffs = array(0, 750, 2000); // ms — first attempt 0, then retry waits
79 - $last_response = null;
17 + // Ensure embeddings are removed when a product is moved to trash or permanently deleted
18 + add_action('wp_trash_post', array($this, 'mxchat_handle_product_delete'));
19 + add_action('before_delete_post', array($this, 'mxchat_handle_product_delete'));
80 20
81 - foreach ($backoffs as $i => $delay_ms) {
82 - if ($delay_ms > 0) {
83 - usleep($delay_ms * 1000);
84 - }
85 - $response = wp_remote_post($url, $args);
86 - $last_response = $response;
87 -
88 - if (!$this->mxchat_is_transient_provider_error($response, $provider_hint)) {
89 - return $response;
90 - }
91 -
92 - if (defined('WP_DEBUG') && WP_DEBUG) {
93 - $code_for_log = is_wp_error($response) ? 'wp_error:' . $response->get_error_code()
94 - : (int) wp_remote_retrieve_response_code($response);
95 - error_log(sprintf(
96 - '[MxChat] Transient provider error (provider=%s, attempt=%d/3, status=%s). %s',
97 - $provider_hint ?: 'unknown',
98 - $i + 1,
99 - $code_for_log,
100 - ($i + 1) < count($backoffs) ? 'Retrying.' : 'Giving up.'
101 - ));
102 - }
103 - }
104 -
105 - return $last_response;
106 -}
107 -
108 -/**
109 - * Returns true if a wp_remote_post response represents a TRANSIENT
110 - * provider error worth retrying. Conservative — only retries on signals
111 - * that are very likely to clear within a few seconds.
112 - *
113 - * Transient signals:
114 - * - WP_Error with timeout / connection / dns / ssl
115 - * - HTTP 429, 502, 503, 504
116 - * - Provider-specific overload bodies (gemini "overloaded", openai
117 - * "server_error", anthropic "overloaded_error", xai/grok "Rate limit")
118 - *
119 - * NOT transient (return false — fail-fast):
120 - * - 200/2xx (success)
121 - * - 401, 403, 404, 422 (auth / config errors — retrying wastes the
122 - * budget; the user needs to fix something)
123 - * - Any other 4xx (assume permanent unless explicitly listed above)
124 - * - 5xx other than the four listed above (e.g. 500 generic server error
125 - * is often a malformed request on our side, not a transient outage)
126 - */
127 -private function mxchat_is_transient_provider_error($response, $provider_hint = '') {
128 - if (is_wp_error($response)) {
129 - $code = $response->get_error_code();
130 - return in_array($code, array('http_request_failed', 'connection_failed', 'connection_timeout'), true)
131 - || stripos((string) $response->get_error_message(), 'timed out') !== false
132 - || stripos((string) $response->get_error_message(), 'timeout') !== false;
133 - }
134 -
135 - $status = (int) wp_remote_retrieve_response_code($response);
136 - if (in_array($status, array(429, 502, 503, 504), true)) {
137 - return true;
138 - }
139 - if ($status >= 200 && $status < 300) {
140 - return false;
141 - }
142 - // Permanent 4xx that should fail fast — even with no body.
143 - if (in_array($status, array(401, 403, 404, 405, 422), true)) {
144 - return false;
145 - }
146 -
147 - // Provider-specific body inspection for the cases where the upstream
148 - // returns 200 with an error envelope (gemini does this for overload).
149 - $body = (string) wp_remote_retrieve_body($response);
150 - if ($body === '') {
151 - return false;
152 - }
153 - $lower = strtolower($body);
154 - $hint = strtolower((string) $provider_hint);
155 -
156 - if ($hint === 'gemini' && (strpos($lower, 'overloaded') !== false
157 - || strpos($lower, 'high demand') !== false
158 - || strpos($lower, 'model is overloaded') !== false)) {
159 - return true;
160 - }
161 - if ($hint === 'openai' && (strpos($lower, 'rate limit reached') !== false
162 - || strpos($lower, '"type":"server_error"') !== false
163 - || strpos($lower, '"code":"server_error"') !== false)) {
164 - return true;
165 - }
166 - if ($hint === 'anthropic' && (strpos($lower, '"type":"overloaded_error"') !== false
167 - || strpos($lower, 'overloaded_error') !== false)) {
168 - return true;
169 - }
170 - if (($hint === 'xai' || $hint === 'grok') && strpos($lower, 'rate limit') !== false) {
171 - return true;
172 - }
173 -
174 - return false;
175 -}
176 -
177 -/**
178 - * Streaming-path classifier: same rules as mxchat_is_transient_provider_error
179 - * but takes a raw (http_code, body, provider_hint, curl_errno) tuple as
180 - * captured during a cURL streaming exec. cURL's WRITEFUNCTION/HEADERFUNCTION
181 - * collect status separately from a plain wp_remote_post array shape, so the
182 - * non-streaming helper above can't be called directly. This delegate keeps
183 - * the classification rules identical across both paths.
184 - */
185 -private function mxchat_is_transient_provider_error_raw($http_code, $body, $provider_hint = '', $curl_errno = 0) {
186 - if ($curl_errno) {
187 - // cURL transport-level error (timeout, connection failure, DNS, etc.)
188 - // Match the same WP_Error timeout/connection signals the array variant treats as transient.
189 - return in_array($curl_errno, array(
190 - CURLE_OPERATION_TIMEDOUT,
191 - CURLE_COULDNT_CONNECT,
192 - CURLE_COULDNT_RESOLVE_HOST,
193 - CURLE_SSL_CONNECT_ERROR,
194 - CURLE_GOT_NOTHING,
195 - CURLE_SEND_ERROR,
196 - CURLE_RECV_ERROR,
197 - ), true);
198 - }
199 -
200 - $status = (int) $http_code;
201 - if (in_array($status, array(429, 502, 503, 504), true)) {
202 - return true;
203 - }
204 - if ($status >= 200 && $status < 300) {
205 - return false;
206 - }
207 - if (in_array($status, array(401, 403, 404, 405, 422), true)) {
208 - return false;
209 - }
210 -
211 - $body = (string) $body;
212 - if ($body === '') {
213 - return false;
214 - }
215 - $lower = strtolower($body);
216 - $hint = strtolower((string) $provider_hint);
217 -
218 - if ($hint === 'gemini' && (strpos($lower, 'overloaded') !== false
219 - || strpos($lower, 'high demand') !== false
220 - || strpos($lower, 'model is overloaded') !== false)) {
221 - return true;
222 - }
223 - if ($hint === 'openai' && (strpos($lower, 'rate limit reached') !== false
224 - || strpos($lower, '"type":"server_error"') !== false
225 - || strpos($lower, '"code":"server_error"') !== false)) {
226 - return true;
227 - }
228 - if ($hint === 'anthropic' && (strpos($lower, '"type":"overloaded_error"') !== false
229 - || strpos($lower, 'overloaded_error') !== false)) {
230 - return true;
231 - }
232 - if (($hint === 'xai' || $hint === 'grok') && strpos($lower, 'rate limit') !== false) {
233 - return true;
234 - }
235 -
236 - return false;
237 -}
238 -
239 -/**
240 - * Whether transient-error auto-retry is enabled in admin settings.
241 - * Default true unless explicitly set to '0'. Used by both wp_remote_post
242 - * (mxchat_provider_call_with_retry) and cURL streaming paths.
243 - */
244 -private function mxchat_retry_enabled() {
245 - $opts = is_array($this->options ?? null) ? $this->options : array();
246 - return !isset($opts['auto_retry_on_transient_error']) ||
247 - (string) $opts['auto_retry_on_transient_error'] !== '0';
248 -}
249 -
250 -private function setup_streaming_headers() {
251 - if ($this->streaming_headers_sent || headers_sent()) {
252 - return false;
253 - }
254 -
255 - // Headers MUST be set BEFORE the buffers are torn down: flushing a
256 - // buffer that holds any stray output commits the response and turns
257 - // every later header() into a logged no-op — dropping all four SSE
258 - // headers, including the X-Accel-Buffering that stops nginx-fronted
259 - // hosts from de-streaming the reply (plan fe130d).
260 - header('Content-Type: text/event-stream');
261 - header('Cache-Control: no-cache');
262 - header('Connection: keep-alive');
263 - header('X-Accel-Buffering: no');
264 -
265 - // Dev-mode diagnostic: with the reorder, stray buffered bytes become
266 - // the first bytes of the SSE stream — record what they are so a future
267 - // switch to ob_end_clean() can be decided on evidence (fe130d follow-up).
268 - if (defined('MXCHAT_DEV_MODE') && MXCHAT_DEV_MODE && ob_get_level() > 0) {
269 - $buffered = ob_get_contents();
270 - if (is_string($buffered) && $buffered !== '') {
271 - error_log('MxChat SSE teardown: output buffer held ' . strlen($buffered) . ' byte(s): ' . substr($buffered, 0, 200));
272 - }
273 - }
274 -
275 - // Disable output buffering
276 - while (ob_get_level()) {
277 - ob_end_flush();
278 - }
279 -
280 - ob_implicit_flush(true);
281 - flush();
282 -
283 - $this->streaming_headers_sent = true;
284 - return true;
285 -}
286 -
287 -/**
288 - * Class constructor
289 - */
290 -public function __construct() {
291 - $this->options = get_option('mxchat_options');
292 - $this->prompts_options = get_option('mxchat_prompts_options', array());
293 - $this->chat_count = get_option('mxchat_chat_count', 0);
294 - $this->word_handler = new MXChat_Word_Handler($this->options);
295 -
296 - // Add all action hooks
297 21 add_action('wp_enqueue_scripts', array($this, 'mxchat_enqueue_scripts_styles'));
298 22 add_action('wp_ajax_mxchat_handle_chat_request', array($this, 'mxchat_handle_chat_request'));
299 23 add_action('wp_ajax_nopriv_mxchat_handle_chat_request', array($this, 'mxchat_handle_chat_request'));
300 24 add_action('wp_ajax_mxchat_dismiss_pre_chat_message', array($this, 'mxchat_dismiss_pre_chat_message'));
301 25 add_action('wp_ajax_nopriv_mxchat_dismiss_pre_chat_message', array($this, 'mxchat_dismiss_pre_chat_message'));
302 -
303 - // Add the AJAX actions for checking if the pre-chat message was dismissed
304 - add_action('wp_ajax_mxchat_check_pre_chat_message_status', array($this, 'mxchat_check_pre_chat_message_status'));
305 - add_action('wp_ajax_nopriv_mxchat_check_pre_chat_message_status', array($this, 'mxchat_check_pre_chat_message_status'));
306 - add_action('wp_ajax_mxchat_fetch_conversation_history', [$this, 'mxchat_fetch_conversation_history']);
307 - add_action('wp_ajax_nopriv_mxchat_fetch_conversation_history', [$this, 'mxchat_fetch_conversation_history']);
308 - add_action('wp_ajax_mxchat_add_to_cart', [$this, 'mxchat_add_to_cart']);
309 - add_action('wp_ajax_nopriv_mxchat_add_to_cart', [$this, 'mxchat_add_to_cart']);
310 -
311 - // Add REST API routes registration
312 - add_action('rest_api_init', array($this, 'register_routes'));
313 - add_action('wp_ajax_mxchat_fetch_new_messages', array($this, 'mxchat_fetch_new_messages'));
314 - add_action('wp_ajax_nopriv_mxchat_fetch_new_messages', array($this, 'mxchat_fetch_new_messages'));
315 -
316 - // Rate limit action - notice we removed the old schedule setup
317 - add_action('mxchat_reset_rate_limits', array($this, 'mxchat_reset_rate_limits'));
318 26
319 - // Self-heal: if the reset event is ever lost (cron row cleared, botched
320 - // migration, deactivate/reactivate race), an admin-context request brings it
321 - // back. Cheap by construction: 60s transient guard + early return when the
322 - // event is already scheduled. Without this, a lost event with the fallback
323 - // flag unset leaves visitors rate-limited forever.
324 - add_action('admin_init', array($this, 'setup_rate_limit_cron_jobs'));
325 -
326 - // File upload and handling actions
327 - add_action('wp_ajax_mxchat_upload_pdf', [$this, 'handle_pdf_upload']);
328 - add_action('wp_ajax_nopriv_mxchat_upload_pdf', [$this, 'handle_pdf_upload']);
329 - add_action('wp_ajax_mxchat_remove_pdf', [$this, 'handle_pdf_remove']);
330 - add_action('wp_ajax_nopriv_mxchat_remove_pdf', [$this, 'handle_pdf_remove']);
331 -
332 - // Word document handling actions
333 - add_action('wp_ajax_mxchat_upload_word', array($this, 'mxchat_handle_word_upload'));
334 - add_action('wp_ajax_nopriv_mxchat_upload_word', array($this, 'mxchat_handle_word_upload'));
335 - add_action('wp_ajax_mxchat_remove_word', array($this, 'mxchat_handle_word_remove'));
336 - add_action('wp_ajax_nopriv_mxchat_remove_word', array($this, 'mxchat_handle_word_remove'));
337 - add_action('wp_ajax_mxchat_check_word_status', array($this, 'mxchat_check_word_status'));
338 - add_action('wp_ajax_nopriv_mxchat_check_word_status', array($this, 'mxchat_check_word_status'));
339 -
340 - // Email handling actions
341 - add_action('wp_ajax_nopriv_mxchat_handle_save_email_and_response', [$this, 'mxchat_handle_save_email_and_response']);
342 - add_action('wp_ajax_mxchat_handle_save_email_and_response', [$this, 'mxchat_handle_save_email_and_response']);
343 - add_action('wp_ajax_nopriv_mxchat_check_email_provided', [$this, 'mxchat_check_email_provided']);
344 - add_action('wp_ajax_mxchat_check_email_provided', [$this, 'mxchat_check_email_provided']);
345 -
346 - add_action('wp_ajax_mxchat_stream_chat', array($this, 'mxchat_handle_chat_request'));
347 - add_action('wp_ajax_nopriv_mxchat_stream_chat', array($this, 'mxchat_handle_chat_request'));
348 -
349 - // Testing panel AJAX actions
350 - add_action('wp_ajax_mxchat_get_system_info', array($this, 'mxchat_get_system_info'));
351 - add_action('wp_ajax_mxchat_get_similarity_threshold', array($this, 'mxchat_get_similarity_threshold'));
352 - add_action('wp_ajax_mxchat_get_kb_status', array($this, 'mxchat_get_kb_status'));
353 - add_action('wp_ajax_mxchat_start_fresh_session', array($this, 'mxchat_start_fresh_session'));
354 - // Add to your existing constructor, in the section with other AJAX actions:
355 - add_action('wp_ajax_mxchat_track_url_click', array($this, 'mxchat_track_url_click'));
356 - add_action('wp_ajax_nopriv_mxchat_track_url_click', array($this, 'mxchat_track_url_click'));
357 - add_action('wp_ajax_mxchat_track_originating_page', array($this, 'mxchat_track_originating_page'));
358 - add_action('wp_ajax_nopriv_mxchat_track_originating_page', array($this, 'mxchat_track_originating_page'));
359 - // Add chat mode checking actions
360 - add_action('wp_ajax_mxchat_get_current_chat_mode', array($this, 'mxchat_get_current_chat_mode'));
361 - add_action('wp_ajax_nopriv_mxchat_get_current_chat_mode', array($this, 'mxchat_get_current_chat_mode'));
362 -
363 - // Nonce refresh for page-cache compatibility (WP Rocket, LiteSpeed, etc.)
364 - add_action('wp_ajax_mxchat_refresh_nonce', array($this, 'mxchat_refresh_nonce'));
365 - add_action('wp_ajax_nopriv_mxchat_refresh_nonce', array($this, 'mxchat_refresh_nonce'));
366 -
367 - // Auto-email transcript action
368 - add_action('mxchat_send_delayed_transcript', array($this, 'mxchat_send_delayed_transcript'), 10, 1);
369 -
370 - add_filter('mxchat_check_actions_only', array($this, 'check_actions_for_addons'), 10, 4);
371 -
372 -
373 -}
374 -
375 -/**
376 - * Return a fresh nonce so cached pages can replace the stale one.
377 - * With `with_settings`, also returns the current behavior-gate settings so
378 - * the widget can correct stale inline-localized values (plan-32db95).
379 - */
380 -public function mxchat_refresh_nonce() {
381 - nocache_headers();
382 - $payload = array('nonce' => wp_create_nonce('mxchat_chat_nonce'));
383 - if (!empty($_REQUEST['with_settings'])) {
384 - $payload['settings'] = $this->get_dynamic_widget_settings(true);
27 + if (!wp_next_scheduled('mxchat_reset_rate_limits')) {
28 + wp_schedule_event(time(), 'daily', 'mxchat_reset_rate_limits');
385 29 }
386 - wp_send_json_success($payload);
387 -}
388 30
389 -/**
390 - * Behavior-gate settings the widget may re-fetch at runtime (plan-32db95).
391 - *
392 - * Every widget setting ships inline in page HTML via wp_localize_script, so
393 - * full-page caches (host caches, WP Rocket, LiteSpeed, W3TC, FlyingPress,
394 - * WP Super Cache, Cloudflare APO, the browser itself) keep serving a stale
395 - * snapshot after an admin changes a setting. MxChat_Cache_Purge clears the
396 - * caches PHP can reach; this payload covers the rest — the widget requests
397 - * it on first open (via the nonce-refresh endpoints) and merges it over
398 - * `mxchatChat`, the same distrust-cached-HTML pattern the 3.2.7 per-request
399 - * nonce uses.
400 - *
401 - * Behavior gates + labels ONLY — colors stay inline because they're also
402 - * server-inline-styled, and a runtime swap would visibly flash.
403 - *
404 - * Both wp_localize_script blocks merge this exact array, so the inline and
405 - * refreshed payloads cannot drift.
406 - *
407 - * @param bool $fresh Re-read mxchat_options from the DB (endpoint paths)
408 - * instead of trusting the instance copy.
409 - * @return array
410 - */
411 -public function get_dynamic_widget_settings($fresh = false) {
412 - $options = $fresh ? get_option('mxchat_options', array()) : $this->options;
413 - if (!is_array($options)) {
414 - $options = array();
415 - }
416 - return array(
417 - 'model' => isset($options['model']) ? $options['model'] : 'gpt-5.6-sol',
418 - 'enable_streaming_toggle' => isset($options['enable_streaming_toggle']) ? $options['enable_streaming_toggle'] : 'on',
419 - 'rate_limit_message' => $options['rate_limit_message'] ?? 'Rate limit exceeded. Please try again later.',
420 - 'chat_toolbar_toggle' => $options['chat_toolbar_toggle'] ?? 'off',
421 - 'print_button_enabled' => $options['print_button_enabled'] ?? 'on',
422 - 'print_button_label' => esc_html__('Download Transcript', 'mxchat'),
423 - // "Start new chat" header-menu item (plan ac2e81). Default OFF.
424 - 'reset_chat_enabled' => $options['reset_chat_enabled'] ?? 'off',
425 - 'reset_chat_label' => !empty($options['reset_chat_label']) ? esc_html($options['reset_chat_label']) : esc_html__('Start new chat', 'mxchat'),
426 - 'reset_chat_confirm' => esc_html__('Start a new chat? This clears the current conversation.', 'mxchat'),
427 - 'stop_button_label' => esc_html__('Stop response', 'mxchat'),
428 - 'print_header_title' => esc_html(get_bloginfo('name')) . ' — ' . esc_html__('Chat transcript', 'mxchat'),
429 - // Emit 'on'/'off' STRINGS, never booleans: wp_localize_script casts
430 - // scalars to string, and (string) false === '' — which the widget's
431 - // old gate read as enabled (plan-4bba64). The filter keeps its
432 - // boolean contract; only the emitted value is stringified.
433 - 'satisfaction_rating_enabled' => apply_filters(
434 - 'mxchat_satisfaction_rating_enabled',
435 - ($options['satisfaction_rating_enabled'] ?? 'off') === 'on'
436 - ) ? 'on' : 'off',
437 - 'satisfaction_rating_idle_seconds' => max(5, min(600, intval($options['satisfaction_rating_idle_seconds'] ?? 60))),
438 - 'satisfaction_rating_copy' => array(
439 - 'question' => !empty($options['satisfaction_rating_question']) ? esc_html($options['satisfaction_rating_question']) : esc_html__('Was this helpful?', 'mxchat'),
440 - 'helpful' => esc_html__('Helpful', 'mxchat'),
441 - 'not_helpful' => esc_html__('Not helpful', 'mxchat'),
442 - 'dismiss' => esc_html__('Dismiss', 'mxchat'),
443 - 'thanks' => !empty($options['satisfaction_rating_thanks']) ? esc_html($options['satisfaction_rating_thanks']) : esc_html__('Thanks! Anything we should improve? (optional)', 'mxchat'),
444 - 'placeholder' => !empty($options['satisfaction_rating_placeholder']) ? esc_html($options['satisfaction_rating_placeholder']) : esc_html__('Tell us what could be better…', 'mxchat'),
445 - 'send' => esc_html__('Send', 'mxchat'),
446 - 'skip' => esc_html__('Skip', 'mxchat'),
447 - 'saved' => !empty($options['satisfaction_rating_saved']) ? esc_html($options['satisfaction_rating_saved']) : esc_html__('Thanks for the feedback.', 'mxchat'),
448 - ),
449 - );
31 + add_action('mxchat_reset_rate_limits', array($this, 'mxchat_reset_rate_limits'));
450 32 }
451 33
452 -// In your core plugin's check_actions_for_addons method:
453 -public function check_actions_for_addons($default, $message, $user_id, $session_id) {
454 - //error_log('MxChat Core: check_actions_for_addons called with message: ' . $message);
455 -
456 - $result = $this->mxchat_check_intent_and_invoke_callback($message, $user_id, $session_id);
457 -
458 - //error_log('MxChat Core: Intent check result = ' . ($result === false ? 'false' : 'true'));
459 -
460 - return $result;
461 -}
462 -
463 - private function mxchat_increment_chat_count() {
464 - $chat_count = get_option('mxchat_chat_count', 0);
465 - $chat_count++;
466 - update_option('mxchat_chat_count', $chat_count);
34 +public function mxchat_handle_product_change($post_id, $post, $update) {
35 + // Ensure this is a product post type
36 + if ($post->post_type !== 'product') {
37 + return;
467 38 }
468 39
469 -function mxchat_fetch_conversation_history() {
470 - if (empty($_POST['session_id'])) {
471 - wp_send_json_error(['message' => esc_html__('Session ID missing.', 'mxchat')]);
472 - wp_die();
473 - }
474 -
475 - $session_id = MxChat_Utils::sanitize_session_id(wp_unslash($_POST['session_id']));
476 -
477 - // SECURITY FIX: Verify session ownership before retrieving data
478 - // If IP/user changed, signal frontend to reset session instead of blocking
479 - $current_user_identifier = MxChat_User::mxchat_get_user_identifier();
480 -
481 - // Check if this session has an owner recorded
482 - $session_owner = get_option("mxchat_session_owner_{$session_id}");
483 -
484 - // Update session owner if it changed (e.g. IP changed due to network switch)
485 - // The session ID itself is the authentication — if the client has it, they own it
486 - if (!$session_owner || $session_owner !== $current_user_identifier) {
487 - update_option("mxchat_session_owner_{$session_id}", $current_user_identifier, 'no');
488 - }
489 -
490 - $history = get_option("mxchat_history_{$session_id}", []); // Retrieve stored history
491 - $chat_mode = get_option("mxchat_mode_{$session_id}", 'ai'); // Get current chat mode
492 -
493 - if (empty($history)) {
494 - // Even if history is empty, return the chat mode
495 - wp_send_json_success([
496 - 'conversation' => [],
497 - 'chat_mode' => $chat_mode
498 - ]);
499 - wp_die();
500 - }
501 -
502 - wp_send_json_success([
503 - 'conversation' => $history,
504 - 'chat_mode' => $chat_mode
505 - ]);
506 - wp_die();
507 -}
508 -private function mxchat_fetch_conversation_history_for_ai($session_id, $session_start_timestamp = 0) {
509 - $history = get_option("mxchat_history_{$session_id}", []);
510 -
511 - // Check persistence setting - when OFF, only include messages from current page load
512 - $options = get_option('mxchat_options', []);
513 - $persistence_enabled = isset($options['chat_persistence_toggle']) && $options['chat_persistence_toggle'] === 'on';
514 -
515 - // Filter history when persistence is OFF to match what the user sees
516 - if (!$persistence_enabled && $session_start_timestamp > 0) {
517 - $history = array_filter($history, function($entry) use ($session_start_timestamp) {
518 - // Include messages from this page load onwards
519 - return isset($entry['timestamp']) && $entry['timestamp'] >= $session_start_timestamp;
520 - });
521 - // Re-index array after filtering
522 - $history = array_values($history);
523 - }
524 -
525 - $formatted_history = [];
526 -
527 - // Adjusted for code-heavy conversations
528 - $max_tokens = 120000; // Context window size
529 - $reserved_tokens = 5000; // Space for system prompts + current query
530 - $current_token_count = 0;
531 -
532 - // Allowed HTML tags for content sanitization
533 - $allowed_tags = [
534 - 'pre' => ['class' => true],
535 - 'code' => ['class' => true],
536 - 'span' => ['class' => true],
537 - 'div' => ['class' => true],
538 - 'strong' => [],
539 - 'em' => []
540 - ];
541 -
542 - foreach (array_reverse($history) as $entry) {
543 - // Preserve code blocks while sanitizing other HTML
544 - $clean_content = wp_kses($entry['content'], $allowed_tags);
545 -
546 - // Detect code blocks in content
547 - $has_code = false;
548 -// Replace the HTML check with:
549 -// Allow messages that contain code blocks or are plain text
550 -if (strpos($clean_content, '<pre') === false &&
551 - strpos($clean_content, '<code') === false &&
552 - $clean_content !== strip_tags($entry['content'])) {
553 - continue;
554 -}
555 -
556 - // Skip entries that lost significant content during sanitization
557 - if (!$has_code && $clean_content !== strip_tags($entry['content'])) {
558 - continue;
559 - }
560 -
561 - // More accurate token estimation (1 token ≈ 4 characters)
562 - $token_estimate = ceil(mb_strlen($clean_content, 'UTF-8') / 4);
563 -
564 - // Check token budget with the new estimate
565 - if (($current_token_count + $token_estimate + $reserved_tokens) > $max_tokens) {
566 - // Try to fit partial content if it's the first entry
567 - if (empty($formatted_history)) {
568 - $clean_content = mb_substr($clean_content, 0, ($max_tokens - $reserved_tokens) * 4);
569 - $token_estimate = ceil(mb_strlen($clean_content, 'UTF-8') / 4);
40 + // Only generate embeddings if the product is published
41 + if ($post->post_status === 'publish') {
42 + // Delay the embedding slightly to ensure all product data is available
43 + add_action('shutdown', function() use ($post_id) {
44 + $product = wc_get_product($post_id);
45 + if ($product && $product->get_price() !== '') {
46 + $this->mxchat_store_product_embedding($product);
570 47 } else {
571 - break;
48 + // Optionally, log or handle the case where product data is incomplete
49 + error_log("Product {$post_id} does not have complete data. Embedding not generated.");
572 50 }
573 - }
574 -
575 - // Add to formatted history
576 - $formatted_history[] = [
577 - 'role' => $entry['role'],
578 - 'content' => $clean_content
579 - ];
580 -
581 - $current_token_count += $token_estimate;
51 + });
582 52 }
583 -
584 - // Reverse back to maintain chronological order
585 - $formatted_history = array_reverse($formatted_history);
586 -
587 - // Add system message about code context
588 - array_unshift($formatted_history, [
589 - 'role' => 'system',
590 - 'content' => 'Preserved code blocks are marked with [CODE BLOCK PRESERVED]. '
591 - . 'Maintain formatting and syntax highlighting when referencing code.'
592 - ]);
593 -
594 - return $formatted_history;
595 53 }
596 54
597 -public function register_routes() {
598 - //error_log(esc_html__('Registering MxChat REST routes', 'mxchat'));
599 -
600 - // Per-request chat-send nonce endpoint — issues a fresh nonce on demand
601 - // so the chat widget never depends on a stale nonce embedded in cached HTML.
602 - // Public (no auth), rate-limited (1 call / IP / second via a transient).
603 - register_rest_route('mxchat/v1', '/nonce', [
604 - 'methods' => 'GET',
605 - 'callback' => [$this, 'mxchat_issue_chat_send_nonce'],
606 - 'permission_callback' => '__return_true',
607 - ]);
608 -
609 - register_rest_route('mxchat/v1', '/stream', [
610 - 'methods' => 'GET',
611 - 'callback' => [$this, 'mxchat_stream_events'],
612 - 'permission_callback' => [$this, 'verify_chat_session'],
613 - ]);
614 -
615 - register_rest_route('mxchat/v1', '/agent-response', [
616 - 'methods' => 'POST',
617 - 'callback' => [$this, 'mxchat_handle_agent_response'],
618 - 'permission_callback' => [$this, 'verify_slack_request'],
619 - ]);
620 -
621 - register_rest_route('mxchat/v1', '/slack-interaction', [
622 - 'methods' => 'POST',
623 - 'callback' => [$this, 'handle_slack_interaction'],
624 - 'permission_callback' => [$this, 'verify_slack_request'],
625 - ]);
626 -
627 - register_rest_route('mxchat/v1', '/slack-messages', [
628 - 'methods' => 'POST',
629 - 'callback' => [$this, 'handle_slack_messages'],
630 - 'permission_callback' => [$this, 'verify_slack_request'],
631 - ]);
632 -
633 - // Telegram webhook endpoint
634 - register_rest_route('mxchat/v1', '/telegram-webhook', [
635 - 'methods' => 'POST',
636 - 'callback' => [$this, 'handle_telegram_webhook'],
637 - 'permission_callback' => [$this, 'verify_telegram_request'],
638 - ]);
639 -
640 - //error_log(esc_html__('MxChat REST routes registered', 'mxchat'));
641 -}
642 -
643 -/**
644 - * Issue a fresh per-request nonce for chat-send. Returned to the widget which
645 - * caches it for the session and includes it on every chat-send / stream-send /
646 - * upload call. By moving the nonce out of inline `window.mxchatChat = {...}` HTML
647 - * we eliminate the entire class of "first-message Access denied" failures that
648 - * plague WP installs behind a full-page cache (WP Rocket, LiteSpeed, FlyingPress,
649 - * W3 Total Cache, Cloudflare APO) — the nonce is never cached because it never
650 - * lives in the HTML body.
651 - *
652 - * Public endpoint. Rate-limited to 1 call / IP / 1s via a transient so a single
653 - * client browser can't be used to flood the nonce-issuance path.
654 - *
655 - * Nonce action: `mxchat_chat_send` (new). The chat-send AJAX handlers accept
656 - * BOTH this action AND the legacy `mxchat_chat_nonce` action for a 30-day
657 - * backwards-compat window so cached pages still in users' browsers don't break
658 - * mid-session.
659 - *
660 - * @since 3.2.7
661 - */
662 -public function mxchat_issue_chat_send_nonce(WP_REST_Request $request) {
663 - $ip = '';
664 - if (!empty($_SERVER['REMOTE_ADDR'])) {
665 - $ip = preg_replace('#[^0-9a-fA-F:\.]#', '', wp_unslash((string) $_SERVER['REMOTE_ADDR']));
55 +public function mxchat_handle_product_delete($post_id) {
56 + if (get_post_type($post_id) !== 'product') {
57 + return;
666 58 }
667 - if ($ip !== '') {
668 - // Best-effort rate limit. WP transients with sub-second TTL are racy
669 - // (parallel bursts can squeak through before set_transient completes);
670 - // we use 2s to make the gate slightly more reliable. Real production
671 - // rate-limiting at sub-second granularity needs Redis or DB row locks
672 - // — out of scope for this endpoint, which is already cheap.
673 - $key = 'mxchat_nonce_rl_' . md5($ip);
674 - if (get_transient($key)) {
675 - return new WP_REST_Response(array(
676 - 'error' => 'rate_limited',
677 - 'message' => __('Too many nonce requests. Try again shortly.', 'mxchat'),
678 - ), 429);
679 - }
680 - set_transient($key, 1, 2);
681 - }
682 59
683 - // The widget calls this endpoint without an X-WP-Nonce header, so WordPress does not
684 - // honor the auth cookie and the request runs as uid=0 even for logged-in users. That
685 - // makes wp_create_nonce() bind the nonce to uid=0, which then fails wp_verify_nonce()
686 - // at admin-ajax (which runs as the real uid) -> logged-in users get a 403 on upload.
687 - // Resolve the real user from the logged_in cookie so the nonce binds to the correct uid.
688 - if ( ! is_user_logged_in() ) {
689 - $maybe_uid = wp_validate_auth_cookie( '', 'logged_in' );
690 - if ( $maybe_uid ) {
691 - wp_set_current_user( $maybe_uid );
692 - }
693 - }
60 + global $wpdb;
61 + $table_name = $wpdb->prefix . 'mxchat_system_prompt_content';
694 62
695 - $payload = array(
696 - 'nonce' => wp_create_nonce('mxchat_chat_send'),
697 - 'expires_in' => 86400, // WP nonces live 24h; widget caches for 12h conservatively.
698 - );
699 -
700 - // plan-32db95: the widget's first-open refresh asks for current behavior
701 - // settings in the same round-trip, so stale inline-localized values on
702 - // cached pages get corrected without a second request. All values in
703 - // this payload already ship in public page HTML — nothing sensitive.
704 - if ($request->get_param('with_settings')) {
705 - $payload['settings'] = $this->get_dynamic_widget_settings(true);
706 - }
707 -
708 - return new WP_REST_Response($payload, 200);
63 + // Delete the embedding associated with this product
64 + $wpdb->delete($table_name, array('source_url' => get_permalink($post_id)), array('%s'));
709 65 }
710 66
711 -/**
712 - * Verify a chat-send nonce. Accepts BOTH the new `mxchat_chat_send` action
713 - * (issued by /wp-json/mxchat/v1/nonce) AND the legacy `mxchat_chat_nonce`
714 - * action (inline-localized in older cached HTML). The legacy acceptance is
715 - * a 30-day backwards-compat window — to be removed in a follow-up release
716 - * after 2026-06-27.
717 - *
718 - * @param string $posted_nonce
719 - * @return bool
720 - */
721 -public static function mxchat_verify_chat_send_nonce($posted_nonce) {
722 - if (!is_string($posted_nonce) || $posted_nonce === '') {
723 - return false;
724 - }
725 - return (bool) wp_verify_nonce($posted_nonce, 'mxchat_chat_send')
726 - || (bool) wp_verify_nonce($posted_nonce, 'mxchat_chat_nonce');
727 -}
67 +private function mxchat_store_product_embedding($product) {
68 + if (isset($this->options['enable_woocommerce_integration']) && $this->options['enable_woocommerce_integration'] === '1') {
728 69
729 -/**
730 - * Verify valid chat session
731 - */
732 -public function verify_chat_session($request) {
733 - $session_id = $request->get_param('session_id');
734 - if (empty($session_id)) {
735 - //error_log(esc_html__('Empty session ID in chat request', 'mxchat'));
736 - return false;
737 - }
70 + $source_url = get_permalink($product->get_id());
71 + $regular_price = $product->get_regular_price();
72 + $sale_price = $product->get_sale_price();
73 + $price = $sale_price ?: $regular_price;
738 74
739 - $chat_mode = get_option("mxchat_mode_{$session_id}", 'ai');
740 - return $chat_mode === 'agent';
741 -}
75 + $description = $product->get_description() . "\n\n" .
76 + "Short Description: " . $product->get_short_description() . "\n" .
77 + "Price: " . $regular_price . "\n" .
78 + "Sale Price: " . ($sale_price ?: 'N/A') . "\n" .
79 + "SKU: " . $product->get_sku();
742 80
743 -/**
744 - * Verify request is coming from Slack.
745 - *
746 - * @param WP_REST_Request $request
747 - * @return bool True if valid, false otherwise.
748 - */
749 -public function verify_slack_request($request) {
750 - // Get the Slack signing secret from your plugin options
751 - $valid_key = $this->options['live_agent_secret_key'] ?? '';
81 + global $wpdb;
82 + $table_name = $wpdb->prefix . 'mxchat_system_prompt_content';
752 83
753 - if (empty($valid_key)) {
754 - //error_log(esc_html__('Slack signing secret not configured', 'mxchat'));
755 - return false;
756 - }
84 + // Delete any existing embedding for this product
85 + $wpdb->delete($table_name, array('source_url' => $source_url), array('%s'));
757 86
758 - $timestamp = $request->get_header('X-Slack-Request-Timestamp');
759 - $slack_signature = $request->get_header('X-Slack-Signature');
760 -
761 - // Verify timestamp to prevent replay attacks
762 - if (abs(time() - intval($timestamp)) > 300) {
763 - //error_log(esc_html__('Slack request timestamp too old', 'mxchat'));
764 - return false;
87 + // Submit the new content and embedding to the database
88 + MxChat_Utils::submit_content_to_db($description, $source_url, $this->options['api_key']);
765 89 }
766 -
767 - // Get raw request body from the WP_REST_Request object
768 - // (php://input may already be consumed by WordPress at this point)
769 - $request_body = $request->get_body();
770 -
771 - // Create the signature base string
772 - $sig_basestring = "v0:{$timestamp}:{$request_body}";
773 -
774 - // Calculate expected signature
775 - $my_signature = 'v0=' . hash_hmac('sha256', $sig_basestring, $valid_key);
776 -
777 - // Compare signatures
778 - return hash_equals($my_signature, $slack_signature);
779 90 }
780 91
781 -/**
782 - * Verify request is coming from Telegram.
783 - *
784 - * @param WP_REST_Request $request
785 - * @return bool True if valid, false otherwise.
786 - */
787 -public function verify_telegram_request($request) {
788 - $secret_token = $this->options['telegram_webhook_secret'] ?? '';
789 92
790 - //error_log('[MxChat Telegram DEBUG] verify_telegram_request called');
791 - //error_log('[MxChat Telegram DEBUG] Stored secret: ' . (empty($secret_token) ? 'EMPTY' : substr($secret_token, 0, 10) . '...'));
792 93
793 - if (empty($secret_token)) {
794 - // No secret configured (legacy setup). Do NOT fail open to the whole
795 - // internet — that lets an unauthenticated caller write agent-branded
796 - // messages. Fall back to verifying the request originates from
797 - // Telegram's published webhook IP ranges so existing no-secret installs
798 - // keep working while an arbitrary-internet caller is blocked. Setting a
799 - // real secret (see the admin notice) is the recommended path.
800 - // (plan-0c17b5)
801 - $peer = isset($_SERVER['REMOTE_ADDR']) ? (string) $_SERVER['REMOTE_ADDR'] : '';
802 - if ($this->mxchat_ip_in_telegram_ranges($peer)) {
803 - return true;
804 - }
805 - error_log('MxChat: Telegram webhook has no secret configured and the request '
806 - . 'is not from a Telegram IP range; rejected. Set a webhook secret to secure it.');
807 - return false;
808 - }
809 94
810 - // Telegram sends the secret token in the X-Telegram-Bot-Api-Secret-Token header
811 - $request_token = $request->get_header('X-Telegram-Bot-Api-Secret-Token');
812 95
813 - //error_log('[MxChat Telegram DEBUG] Request token: ' . (empty($request_token) ? 'EMPTY' : substr($request_token, 0, 10) . '...'));
814 -
815 - if (empty($request_token)) {
816 - //error_log('[MxChat Telegram DEBUG] Request rejected: No token in header');
817 - return false;
96 + private function mxchat_increment_chat_count() {
97 + $chat_count = get_option('mxchat_chat_count', 0);
98 + $chat_count++;
99 + update_option('mxchat_chat_count', $chat_count);
818 100 }
819 101
820 - // Timing-safe comparison
821 - $result = hash_equals($secret_token, $request_token);
822 - //error_log('[MxChat Telegram DEBUG] Token comparison result: ' . ($result ? 'MATCH' : 'MISMATCH'));
823 - return $result;
824 -}
825 -
826 -/**
827 - * Whether $ip falls within Telegram's published webhook IPv4 ranges
828 - * (149.154.160.0/20 and 91.108.4.0/22). Used as an authenticity fallback for
829 - * the Telegram webhook when no secret token is configured, so a legacy
830 - * no-secret install keeps working without failing open to the entire internet.
831 - *
832 - * Uses the real TCP peer (REMOTE_ADDR); a spoofable X-Forwarded-For is NOT
833 - * consulted. Behind a reverse proxy / CDN that rewrites REMOTE_ADDR this may
834 - * not match — which is exactly why configuring a real webhook secret is the
835 - * recommended path. (plan-0c17b5)
836 - *
837 - * @param string $ip Candidate IPv4 address.
838 - * @return bool
839 - */
840 -private function mxchat_ip_in_telegram_ranges($ip) {
841 - if (!is_string($ip) || $ip === '' || filter_var($ip, FILTER_VALIDATE_IP, FILTER_FLAG_IPV4) === false) {
842 - return false;
843 - }
844 - $ip_long = ip2long($ip);
845 - if ($ip_long === false) {
846 - return false;
847 - }
848 - $ranges = array(
849 - array('149.154.160.0', 20),
850 - array('91.108.4.0', 22),
851 - );
852 - foreach ($ranges as $range) {
853 - $subnet_long = ip2long($range[0]);
854 - if ($subnet_long === false) {
855 - continue;
856 - }
857 - $mask = (0xFFFFFFFF << (32 - $range[1])) & 0xFFFFFFFF;
858 - if (($ip_long & $mask) === ($subnet_long & $mask)) {
859 - return true;
860 - }
861 - }
862 - return false;
863 -}
864 -
865 -public function mxchat_stream_events(WP_REST_Request $request) {
866 - header('Content-Type: text/event-stream');
867 - header('Cache-Control: no-cache');
868 - header('Connection: keep-alive');
869 -
870 - $session_id = MxChat_Utils::sanitize_session_id($request->get_param('session_id'));
871 - $last_seen_id = sanitize_text_field($request->get_param('last_seen_id')) ?: '';
872 -
873 - if (empty($session_id)) {
874 - echo esc_html__("event: error\ndata: ", 'mxchat') . esc_html__('Missing session_id', 'mxchat') . "\n\n";
875 - flush();
876 - exit;
877 - }
878 -
879 - $history = get_option("mxchat_history_{$session_id}", []);
880 -
881 - // Filter only new messages
882 - $new_messages = array_filter($history, function ($message) use ($last_seen_id) {
883 - return !empty($message['id']) && $message['id'] > $last_seen_id;
884 - });
885 -
886 - // Send new messages if available
887 - if (!empty($new_messages)) {
888 - echo esc_html__("event: newMessages\ndata: ", 'mxchat') . json_encode(array_values($new_messages)) . "\n\n";
889 - } else {
890 - // Keep the connection alive
891 - echo esc_html__("event: keepAlive\ndata: ", 'mxchat') . "{}\n\n";
892 - }
893 - flush();
894 - exit;
895 -}
896 -
897 -
898 -
899 -
900 -private function mxchat_save_chat_message($session_id, $role, $message, $originating_page = null, $rag_context = null) {
102 +public function mxchat_fetch_conversation_history_for_ajax($session_id) {
901 103 global $wpdb;
902 104 $table_name = $wpdb->prefix . 'mxchat_chat_transcripts';
903 - //error_log("[DEBUG] mxchat_save_chat_message -> START for session_id: {$session_id}, role: {$role}");
904 -
905 - // Check if this is the first message in a new session (before any other database operations)
906 - $is_new_session = false;
907 - if ($role === 'user') { // Only check for user messages, not bot responses
908 - $existing_messages = $wpdb->get_var($wpdb->prepare(
909 - "SELECT COUNT(*) FROM $table_name WHERE session_id = %s",
910 - $session_id
911 - ));
912 - $is_new_session = ($existing_messages == 0);
913 -
914 - // Log for debugging
915 - if ($is_new_session) {
916 - //error_log("[DEBUG] This is a NEW session - first message");
917 - }
918 - }
919 -
920 - // SECURITY FIX: Set session ownership for new sessions
921 - if ($is_new_session && $role === 'user') {
922 - $current_user_identifier = MxChat_User::mxchat_get_user_identifier();
923 - $session_owner_key = "mxchat_session_owner_{$session_id}";
924 -
925 - // Only set ownership if not already set
926 - if (!get_option($session_owner_key)) {
927 - update_option($session_owner_key, $current_user_identifier, 'no');
928 - //error_log("[DEBUG] Set session ownership for {$session_id} to {$current_user_identifier}");
929 - }
930 - }
931 -
932 - // 1) Extract agent name if present
933 - $agent_name = '';
934 - if (preg_match('/^Agent: (.*?) - /', $message, $matches)) {
935 - $agent_name = $matches[1];
936 - $message = str_replace("Agent: $agent_name - ", '', $message);
937 - $session_meta_key = "mxchat_agent_name_{$session_id}";
938 - if (empty(get_option($session_meta_key))) {
939 - update_option($session_meta_key, $agent_name);
940 - //error_log("[DEBUG] mxchat_save_chat_message -> Stored agent_name in option: {$session_meta_key} => {$agent_name}");
941 - }
942 - }
943 -
944 - // 2) Generate unique message_id
945 - $message_id = uniqid();
946 - //error_log("[DEBUG] mxchat_save_chat_message -> Generated message_id: {$message_id}");
947 -
948 - // 3) Determine user_id
949 - $user_id = is_user_logged_in() ? get_current_user_id() : 0;
950 -
951 - // 4) Determine user_identifier
952 - $user_identifier = $agent_name
953 - ? $agent_name
954 - : MxChat_User::mxchat_get_user_identifier();
955 -
956 - // 5) Determine displayed_name
957 - $user_email = MxChat_User::mxchat_get_user_email();
958 - $displayed_name = $agent_name ? $agent_name : ($user_email ?: $user_identifier);
959 -
960 - // 6) Check for a saved email in wp_options
961 - $email_option_key = "mxchat_email_{$session_id}";
962 - $saved_email = get_option($email_option_key);
963 - //error_log("[DEBUG] mxchat_save_chat_message -> Checking wp_options for email_option_key: {$email_option_key}, found: {$saved_email}");
964 -
965 - // Check for a saved name in wp_options
966 - $name_option_key = "mxchat_name_{$session_id}";
967 - $saved_name = get_option($name_option_key);
968 - //error_log("[DEBUG] mxchat_save_chat_message -> Checking wp_options for name_option_key: {$name_option_key}, found: {$saved_name}");
969 -
970 - // If found, update DB user_email and user_name
971 - if ($saved_email || $saved_name) {
972 - $update_data = [];
973 - if ($saved_email) {
974 - $update_data['user_email'] = $saved_email;
975 - }
976 - if ($saved_name) {
977 - $update_data['user_name'] = $saved_name;
978 - }
979 -
980 - if (!empty($update_data)) {
981 - $update_res = $wpdb->update(
982 - $table_name,
983 - $update_data,
984 - ['session_id' => $session_id],
985 - array_fill(0, count($update_data), '%s'),
986 - ['%s']
987 - );
988 - //error_log("[DEBUG] mxchat_save_chat_message -> Attempted DB user_email/user_name update for session_id {$session_id}. update_res: {$update_res}");
989 - }
990 - }
991 -
992 - // 7) Save to session history in wp_options
993 - $history_key = "mxchat_history_{$session_id}";
994 - $history = get_option($history_key, []);
995 - $history[] = [
996 - 'id' => $message_id,
997 - 'role' => $role,
998 - 'content' => $message,
999 - 'timestamp' => round(microtime(true) * 1000),
1000 - 'agent_name' => $displayed_name,
1001 - ];
1002 - update_option($history_key, $history, 'no');
1003 - //error_log("[DEBUG] mxchat_save_chat_message -> Updated session history in option: {$history_key}");
1004 -
1005 - // 8) Save the message to DB (INSERT)
1006 - $insert_data = [
1007 - 'user_id' => $user_id,
1008 - 'user_identifier'=> $user_identifier,
1009 - 'user_email' => $saved_email ?: $user_email,
1010 - 'user_name' => $saved_name ?: '', // Add name to insert data
1011 - 'session_id' => $session_id,
1012 - 'role' => $role,
1013 - 'message' => $message,
1014 - 'timestamp' => current_time('mysql', 1),
1015 - ];
1016 -
1017 - // IMPROVED: Handle originating page data
1018 - $columns_exist = $wpdb->get_var("SHOW COLUMNS FROM $table_name LIKE 'originating_page_url'");
1019 -
1020 - if ($columns_exist) {
1021 - if ($is_new_session && $role === 'user') {
1022 - // For the first user message, set originating page data
1023 -
1024 - // First check if we have it from the parameter
1025 - if ($originating_page && !empty($originating_page['url'])) {
1026 - $insert_data['originating_page_url'] = $originating_page['url'];
1027 - $insert_data['originating_page_title'] = $originating_page['title'] ?? '';
1028 -
1029 - //error_log("[DEBUG] Setting originating page from parameter: " . $originating_page['url']);
1030 - }
1031 - // Otherwise check if it's stored in the instance property
1032 - else if (isset($this->pending_originating_page) && !empty($this->pending_originating_page['url'])) {
1033 - $insert_data['originating_page_url'] = $this->pending_originating_page['url'];
1034 - $insert_data['originating_page_title'] = $this->pending_originating_page['title'] ?? '';
1035 -
1036 - //error_log("[DEBUG] Setting originating page from pending_originating_page: " . $this->pending_originating_page['url']);
1037 -
1038 - // Clear after using (= null, not unset(): unset() undeclares the property
1039 - // and the next assignment recreates it dynamic, re-triggering the PHP 8.2 deprecation)
1040 - $this->pending_originating_page = null;
1041 - }
1042 - // Fallback to HTTP_REFERER if nothing else is available
1043 - else if (isset($_SERVER['HTTP_REFERER'])) {
1044 - $referer_url = esc_url_raw($_SERVER['HTTP_REFERER']);
1045 - $insert_data['originating_page_url'] = $referer_url;
1046 -
1047 - // Generate title from URL
1048 - $parsed_url = parse_url($referer_url);
1049 - $path = isset($parsed_url['path']) ? trim($parsed_url['path'], '/') : '';
1050 -
1051 - if (empty($path) || $path === 'index.php' || $path === 'index.html') {
1052 - $insert_data['originating_page_title'] = 'Homepage';
1053 - } else {
1054 - $title = str_replace(['-', '_', '/', '.php', '.html'], ' ', $path);
1055 - $insert_data['originating_page_title'] = ucwords(trim($title));
1056 - }
1057 -
1058 - //error_log("[DEBUG] Setting originating page from HTTP_REFERER: " . $referer_url);
1059 - }
1060 -
1061 - // Store for this session so all messages have the same originating page
1062 - if (!empty($insert_data['originating_page_url'])) {
1063 - update_option("mxchat_originating_page_{$session_id}", [
1064 - 'url' => $insert_data['originating_page_url'],
1065 - 'title' => $insert_data['originating_page_title']
1066 - ], 'no');
1067 - }
1068 - } else {
1069 - // For subsequent messages in the session, use the stored originating page
1070 - $stored_originating = get_option("mxchat_originating_page_{$session_id}");
1071 - if ($stored_originating && !empty($stored_originating['url'])) {
1072 - $insert_data['originating_page_url'] = $stored_originating['url'];
1073 - $insert_data['originating_page_title'] = $stored_originating['title'] ?? '';
1074 - }
1075 - }
1076 - }
1077 105
1078 - // Add RAG context if provided (for bot messages)
1079 - if ($rag_context !== null && $role === 'bot') {
1080 - $rag_context_column_exists = $wpdb->get_var("SHOW COLUMNS FROM $table_name LIKE 'rag_context'");
1081 - if ($rag_context_column_exists) {
1082 - $insert_data['rag_context'] = is_array($rag_context) ? wp_json_encode($rag_context) : $rag_context;
1083 - }
1084 - }
1085 -
1086 - $wpdb->insert($table_name, $insert_data);
1087 - //error_log("[DEBUG] mxchat_save_chat_message -> Inserted message into DB. row_id: {$wpdb->insert_id}, data: " . print_r($insert_data, true));
1088 -
1089 - // 9) Send notification email if this is the first user message in a new session
1090 - if ($wpdb->insert_id && $is_new_session && $role === 'user') {
1091 - $this->send_new_chat_notification($session_id, array(
1092 - 'identifier' => $user_identifier,
1093 - 'email' => $saved_email ?: $user_email,
1094 - 'ip' => $_SERVER['REMOTE_ADDR']
1095 - ));
1096 - }
1097 -
1098 - // 10) Schedule delayed transcript email if enabled and message is from user
1099 - if ($wpdb->insert_id && $role === 'user') {
1100 - $this->schedule_delayed_transcript_email($session_id);
1101 - }
1102 -
1103 - //error_log("[DEBUG] mxchat_save_chat_message -> END for session_id: {$session_id}");
1104 - return $message_id;
1105 -}
1106 -
1107 -private function send_new_chat_notification($session_id, $user_info = array()) {
1108 - $options = get_option('mxchat_transcripts_options');
1109 -
1110 - // Check if notifications are enabled
1111 - if (empty($options['mxchat_enable_notifications'])) {
1112 - return false;
1113 - }
1114 -
1115 - // Get notification email
1116 - $to = !empty($options['mxchat_notification_email']) ?
1117 - $options['mxchat_notification_email'] :
1118 - get_option('admin_email');
1119 -
1120 - if (!is_email($to)) {
1121 - return false;
1122 - }
1123 -
1124 - // Prepare email content
1125 - $subject = sprintf('[%s] New Chat Session Started', get_bloginfo('name'));
1126 -
1127 - $user_identifier = isset($user_info['identifier']) ? $user_info['identifier'] : 'Guest';
1128 - $user_email = isset($user_info['email']) ? $user_info['email'] : 'Not provided';
1129 - $user_ip = isset($user_info['ip']) ? $user_info['ip'] : $_SERVER['REMOTE_ADDR'];
1130 -
1131 - $message = sprintf(
1132 - "A new chat session has started on your website.\n\n" .
1133 - "Session ID: %s\n" .
1134 - "User: %s\n" .
1135 - "Email: %s\n" .
1136 - "IP Address: %s\n" .
1137 - "Time: %s\n\n" .
1138 - "View transcripts: %s",
1139 - $session_id,
1140 - $user_identifier,
1141 - $user_email,
1142 - $user_ip,
1143 - current_time('mysql'),
1144 - admin_url('admin.php?page=mxchat-transcripts')
106 + // Prepare and execute the query safely
107 + $chat_transcripts = $wpdb->get_results(
108 + $wpdb->prepare("SELECT * FROM $table_name WHERE session_id = %s ORDER BY timestamp ASC", sanitize_text_field($session_id))
1145 109 );
1146 -
1147 - // Send email
1148 - return wp_mail($to, $subject, $message);
1149 -}
1150 110
1151 -/**
1152 - * Schedule delayed transcript email for a session
1153 - * Reschedules if a new user message is received
1154 - */
1155 -private function schedule_delayed_transcript_email($session_id) {
1156 - $options = get_option('mxchat_transcripts_options');
1157 -
1158 - // Check if auto-email is enabled
1159 - if (empty($options['mxchat_auto_email_transcript_enabled'])) {
1160 - return;
111 + // Check if results are empty
112 + if (empty($chat_transcripts)) {
113 + return [];
1161 114 }
1162 -
1163 - // Get notification email
1164 - $email = !empty($options['mxchat_notification_email']) ?
1165 - $options['mxchat_notification_email'] :
1166 - get_option('admin_email');
1167 -
1168 - if (!is_email($email)) {
1169 - return;
1170 - }
1171 -
1172 - // Get delay in minutes (default 30)
1173 - $delay_minutes = isset($options['mxchat_auto_email_transcript_delay']) ?
1174 - intval($options['mxchat_auto_email_transcript_delay']) : 30;
1175 -
1176 - // Clear any existing scheduled event for this session
1177 - $hook = 'mxchat_send_delayed_transcript';
1178 - $args = array($session_id);
1179 - $timestamp = wp_next_scheduled($hook, $args);
1180 -
1181 - if ($timestamp) {
1182 - wp_unschedule_event($timestamp, $hook, $args);
1183 - }
1184 -
1185 - // Schedule new event
1186 - $schedule_time = time() + ($delay_minutes * 60);
1187 - wp_schedule_single_event($schedule_time, $hook, $args);
1188 -}
1189 115
1190 -/**
1191 - * Check if chat messages contain contact information (email or phone number)
1192 - *
1193 - * @param array $messages Array of message objects with 'message' property
1194 - * @param object|null $session_data Session data object with user_email property
1195 - * @return bool True if contact info found, false otherwise
1196 - */
1197 -private function chat_contains_contact_info($messages, $session_data = null) {
1198 - // Check if session already has a stored email
1199 - if ($session_data && !empty($session_data->user_email)) {
1200 - return true;
116 + // Build the conversation history
117 + $conversation_history = [];
118 + foreach ($chat_transcripts as $transcript) {
119 + $conversation_history[] = [
120 + 'role' => $transcript->role,
121 + 'content' => $transcript->message
122 + ];
1201 123 }
1202 124
1203 - // Email regex pattern
1204 - $email_pattern = '/\b[A-Za-z0-9._%+-]+@[A-Za-z0-9.-]+\.[A-Z|a-z]{2,}\b/';
1205 -
1206 - // Phone number patterns (covers various formats including international, WhatsApp style)
1207 - // Matches: +1234567890, (123) 456-7890, 123-456-7890, 123.456.7890, 1234567890, +1 234 567 8900, etc.
1208 - $phone_pattern = '/(?:\+?\d{1,3}[-.\s]?)?\(?\d{2,4}\)?[-.\s]?\d{2,4}[-.\s]?\d{2,4}(?:[-.\s]?\d{1,4})?/';
1209 -
1210 - // Only check user messages (not assistant responses)
1211 - foreach ($messages as $msg) {
1212 - if ($msg->role !== 'user') {
1213 - continue;
1214 - }
1215 -
1216 - $message_text = $msg->message;
1217 -
1218 - // Check for email
1219 - if (preg_match($email_pattern, $message_text)) {
1220 - return true;
1221 - }
1222 -
1223 - // Check for phone number (must be at least 7 digits total to avoid false positives)
1224 - if (preg_match($phone_pattern, $message_text, $matches)) {
1225 - // Count actual digits to avoid matching short numbers
1226 - $digits_only = preg_replace('/\D/', '', $matches[0]);
1227 - if (strlen($digits_only) >= 7) {
1228 - return true;
1229 - }
1230 - }
1231 - }
1232 -
1233 - return false;
125 + return $conversation_history;
1234 126 }
1235 127
1236 -/**
1237 - * Send the delayed transcript email with .txt attachment
1238 - */
1239 -public function mxchat_send_delayed_transcript($session_id) {
1240 - global $wpdb;
1241 128
1242 - // plan-mxchat-20260731-d42bec — this is the one place a session id becomes a
1243 - // filesystem path segment (see the $temp_file build below), so validate here
1244 - // too even though intake is now validated. This runs from a scheduled event,
1245 - // so its argument comes from whatever was stored at schedule time rather than
1246 - // straight from the current request.
1247 - $session_id = MxChat_Utils::sanitize_session_id($session_id);
1248 - if ($session_id === '') {
1249 - return false;
1250 - }
1251 -
1252 - $options = get_option('mxchat_transcripts_options');
1253 -
1254 - // Get notification email
1255 - $to = !empty($options['mxchat_notification_email']) ?
1256 - $options['mxchat_notification_email'] :
1257 - get_option('admin_email');
1258 -
1259 - if (!is_email($to)) {
1260 - return false;
1261 - }
1262 -
1263 - // Get all messages for this session
1264 - $table_name = $wpdb->prefix . 'mxchat_chat_transcripts';
1265 - $messages = $wpdb->get_results($wpdb->prepare(
1266 - "SELECT role, message, timestamp FROM {$table_name}
1267 - WHERE session_id = %s
1268 - ORDER BY timestamp ASC",
1269 - $session_id
1270 - ));
1271 -
1272 - if (empty($messages)) {
1273 - return false;
1274 - }
1275 -
1276 - // Get session metadata
1277 - $sessions_table = $wpdb->prefix . 'mxchat_sessions';
1278 - $session_data = $wpdb->get_row($wpdb->prepare(
1279 - "SELECT * FROM {$sessions_table} WHERE session_id = %s",
1280 - $session_id
1281 - ));
1282 -
1283 - // Check if contact info is required and if it's present
1284 - $require_contact = !empty($options['mxchat_auto_email_transcript_require_contact']);
1285 - if ($require_contact && !$this->chat_contains_contact_info($messages, $session_data)) {
1286 - // Contact info required but not found - skip sending
1287 - return false;
1288 - }
1289 -
1290 - // Build transcript content
1291 - $transcript_content = "Chat Transcript\n";
1292 - $transcript_content .= "================\n\n";
1293 - $transcript_content .= "Session ID: " . $session_id . "\n";
1294 -
1295 - if ($session_data) {
1296 - $transcript_content .= "User: " . ($session_data->user_identifier ?: 'Guest') . "\n";
1297 - $transcript_content .= "Email: " . ($session_data->user_email ?: 'Not provided') . "\n";
1298 - $transcript_content .= "Started: " . $session_data->created_at . "\n";
1299 - }
1300 -
1301 - $transcript_content .= "\n" . str_repeat("=", 50) . "\n\n";
1302 -
1303 - // Add messages
1304 - foreach ($messages as $msg) {
1305 - // 'agent' rows are live-agent (human) replies — label them as such in
1306 - // the emailed transcript, same distinction the Transcripts viewer draws.
1307 - $role_label = ($msg->role === 'user') ? 'User' : (($msg->role === 'agent') ? 'Live Agent' : 'Assistant');
1308 - $transcript_content .= "[{$msg->timestamp}] {$role_label}:\n";
1309 - $transcript_content .= $msg->message . "\n\n";
1310 - }
1311 -
1312 - // Create temporary file for attachment using WP_Filesystem
1313 - $upload_dir = wp_upload_dir();
1314 - // basename() is the SECOND independent control on this write
1315 - // (plan-mxchat-20260731-d42bec). The validator above already rejects any id
1316 - // containing a path separator; this survives someone loosening it later.
1317 - $temp_file = $upload_dir['basedir'] . '/' . basename('mxchat-transcript-' . $session_id . '.txt');
1318 - global $wp_filesystem;
1319 - if (empty($wp_filesystem)) {
1320 - require_once ABSPATH . 'wp-admin/includes/file.php';
1321 - WP_Filesystem();
1322 - }
1323 - $wp_filesystem->put_contents($temp_file, $transcript_content, FS_CHMOD_FILE);
1324 -
1325 - // Prepare email
1326 - $subject = sprintf('[%s] Chat Transcript - Session %s', get_bloginfo('name'), substr($session_id, 0, 8));
1327 -
1328 - $message = "Please find attached the full chat transcript.\n\n";
1329 - $message .= "Session ID: {$session_id}\n";
1330 -
1331 - if ($session_data) {
1332 - $message .= "User: " . ($session_data->user_identifier ?: 'Guest') . "\n";
1333 - $message .= "Email: " . ($session_data->user_email ?: 'Not provided') . "\n";
1334 - }
1335 -
1336 - $message .= "\nView online: " . admin_url('admin.php?page=mxchat-transcripts');
1337 -
1338 - // Send email with attachment
1339 - $attachments = array($temp_file);
1340 - $result = wp_mail($to, $subject, $message, '', $attachments);
1341 -
1342 - // Clean up temporary file
1343 - if (file_exists($temp_file)) {
1344 - unlink($temp_file);
1345 - }
1346 -
1347 - return $result;
1348 -}
1349 -
1350 -
1351 -
1352 -public function mxchat_handle_save_email_and_response() {
1353 - //error_log('[DEBUG] ---------- mxchat_handle_save_email_and_response START ----------');
1354 - //error_log('DEBUG: POST data: ' . print_r($_POST, true));
1355 -
1356 - nocache_headers();
1357 -
1358 - // Validate nonce
1359 - if (!isset($_POST['nonce']) || !MxChat_Integrator::mxchat_verify_chat_send_nonce($_POST['nonce'])) {
1360 - //error_log(esc_html__('[ERROR] Invalid nonce in mxchat_handle_save_email_and_response', 'mxchat'));
1361 - wp_send_json_error(['message' => esc_html__('Invalid nonce.', 'mxchat')]);
1362 - wp_die();
1363 - }
1364 -
1365 - $session_id = isset($_POST['session_id']) ? MxChat_Utils::sanitize_session_id(wp_unslash($_POST['session_id'])) : '';
1366 - $email = isset($_POST['email']) ? sanitize_email($_POST['email']) : '';
1367 - $name = isset($_POST['name']) ? sanitize_text_field($_POST['name']) : '';
1368 -
1369 - //error_log("[DEBUG] handle_save_email_and_response -> session_id: {$session_id}, email: {$email}, name: {$name}");
1370 -
1371 - if (empty($session_id) || $session_id === 'null' || empty($email)) {
1372 - //error_log("[ERROR] Missing session_id or email: session_id={$session_id}, email={$email}");
1373 - wp_send_json_error(['message' => esc_html__('Session ID or email is missing.', 'mxchat')]);
1374 - wp_die();
1375 - }
1376 -
1377 - // Validate name if provided (check if name field is enabled and name is required)
1378 - $options = get_option('mxchat_options', []);
1379 - $name_field_enabled = isset($options['enable_name_field']) &&
1380 - ($options['enable_name_field'] === '1' || $options['enable_name_field'] === 'on');
1381 -
1382 - if ($name_field_enabled && (empty($name) || strlen(trim($name)) < 2 || strlen(trim($name)) > 100)) {
1383 - //error_log("[ERROR] Invalid name: {$name} (enabled: {$name_field_enabled})");
1384 - wp_send_json_error(['message' => esc_html__('Name must be between 2 and 100 characters.', 'mxchat')]);
1385 - wp_die();
1386 - }
1387 -
1388 - // 1) Always store email in wp_options
1389 - $email_option_key = "mxchat_email_{$session_id}";
1390 - update_option($email_option_key, $email, 'no');
1391 - //error_log("[DEBUG] handle_save_email_and_response -> updated option: {$email_option_key} => {$email}");
1392 -
1393 - // Store name in wp_options if provided
1394 - if (!empty($name)) {
1395 - $name_option_key = "mxchat_name_{$session_id}";
1396 - update_option($name_option_key, $name, 'no');
1397 - //error_log("[DEBUG] handle_save_email_and_response -> updated option: {$name_option_key} => {$name}");
1398 - }
1399 -
1400 - // 2) (Optional) Also store in DB if a row already exists
129 +private function mxchat_save_chat_message($session_id, $role, $message) {
1401 130 global $wpdb;
1402 131 $table_name = $wpdb->prefix . 'mxchat_chat_transcripts';
1403 132
1404 - // Make sure we have a valid placeholder in prepare
1405 - $sql = $wpdb->prepare("SELECT COUNT(*) FROM {$table_name} WHERE session_id = %s", $session_id);
1406 - $session_count = $wpdb->get_var($sql);
133 + $user_id = is_user_logged_in() ? get_current_user_id() : 0;
134 + $user_identifier = MxChat_User::mxchat_get_user_identifier();
135 + $user_email = MxChat_User::mxchat_get_user_email();
1407 136
1408 - //error_log("[DEBUG] handle_save_email_and_response -> session_count for {$session_id}: {$session_count} (SQL: {$sql})");
1409 -
1410 - if ($session_count) {
1411 - // Update both user_email and user_name if row(s) exist
1412 - if (!empty($name)) {
1413 - $update_sql = $wpdb->prepare(
1414 - "UPDATE {$table_name} SET user_email = %s, user_name = %s WHERE session_id = %s",
1415 - $email,
1416 - $name,
1417 - $session_id
1418 - );
1419 - } else {
1420 - $update_sql = $wpdb->prepare(
1421 - "UPDATE {$table_name} SET user_email = %s WHERE session_id = %s",
1422 - $email,
1423 - $session_id
1424 - );
1425 - }
1426 - $wpdb->query($update_sql);
1427 - //error_log("[DEBUG] handle_save_email_and_response -> DB updated: {$update_sql}");
1428 - } else {
1429 - //error_log("[INFO] handle_save_email_and_response -> No DB entry for {$session_id}, so email/name is only in wp_options.");
1430 - }
1431 -
1432 - // Provide success response (same as original)
1433 - $bot_message = __('Thanks for providing your email! You can continue chatting now.', 'mxchat');
1434 - //error_log("[DEBUG] handle_save_email_and_response -> success, returning bot_message: {$bot_message}");
1435 - wp_send_json_success(['message' => $bot_message]);
1436 - wp_die();
137 + $wpdb->insert($table_name, [
138 + 'user_id' => $user_id,
139 + 'user_identifier' => $user_identifier,
140 + 'user_email' => $user_email,
141 + 'session_id' => $session_id,
142 + 'role' => $role,
143 + 'message' => $message,
144 + 'timestamp' => current_time('mysql', 1)
145 + ]);
1437 146 }
1438 147
1439 -public function mxchat_check_email_provided() {
1440 - //error_log('[DEBUG] ---------- mxchat_check_email_provided START ----------');
1441 -
1442 - nocache_headers();
1443 -
1444 - if (!isset($_POST['nonce']) || !MxChat_Integrator::mxchat_verify_chat_send_nonce($_POST['nonce'])) {
1445 - //error_log('[ERROR] Invalid nonce in mxchat_check_email_provided');
1446 - wp_send_json_error(['message' => esc_html__('Invalid nonce', 'mxchat')]);
1447 - }
1448 -
1449 - $session_id = isset($_POST['session_id']) ? MxChat_Utils::sanitize_session_id(wp_unslash($_POST['session_id'])) : '';
1450 - if (empty($session_id) || $session_id === 'null') {
1451 - //error_log('[ERROR] No session ID provided in mxchat_check_email_provided');
1452 - wp_send_json_error(['message' => esc_html__('No session ID provided', 'mxchat')]);
1453 - }
1454 -
1455 - // Check if the user is logged in
1456 - if (is_user_logged_in()) {
1457 - $current_user = wp_get_current_user();
1458 - //error_log("[DEBUG] User is logged in as {$current_user->user_email}");
1459 -
1460 - // Get user's display name for logged in users
1461 - $user_name = !empty($current_user->display_name) ? $current_user->display_name :
1462 - (!empty($current_user->first_name) ? $current_user->first_name : '');
1463 -
1464 - $response_data = ['logged_in' => true, 'email' => $current_user->user_email];
1465 - if (!empty($user_name)) {
1466 - $response_data['name'] = $user_name;
1467 - }
1468 -
1469 - wp_send_json_success($response_data);
1470 - }
1471 -
1472 - // Check if name field is required
1473 - $options = get_option('mxchat_options', []);
1474 - $name_field_enabled = isset($options['enable_name_field']) &&
1475 - ($options['enable_name_field'] === '1' || $options['enable_name_field'] === 'on');
1476 -
1477 - $email_option_key = "mxchat_email_{$session_id}";
1478 - $stored_email = get_option($email_option_key, '');
1479 -
1480 - // Check for stored name
1481 - $name_option_key = "mxchat_name_{$session_id}";
1482 - $stored_name = get_option($name_option_key, '');
1483 -
1484 - //error_log("[DEBUG] mxchat_check_email_provided -> Checking email option: {$email_option_key}, found: {$stored_email}");
1485 - //error_log("[DEBUG] mxchat_check_email_provided -> Checking name option: {$name_option_key}, found: {$stored_name}, required: " . ($name_field_enabled ? 'yes' : 'no'));
1486 -
1487 - // Check if we have email and name (if name is required)
1488 - $has_required_info = !empty($stored_email);
1489 -
1490 - if ($name_field_enabled) {
1491 - $has_required_info = $has_required_info && !empty($stored_name);
1492 - }
1493 -
1494 - if ($has_required_info) {
1495 - //error_log("[DEBUG] mxchat_check_email_provided -> Required info found, returning success");
1496 -
1497 - $response_data = ['email' => $stored_email];
1498 - if (!empty($stored_name)) {
1499 - $response_data['name'] = $stored_name;
1500 - }
1501 -
1502 - wp_send_json_success($response_data);
1503 - } else {
1504 - //error_log("[DEBUG] mxchat_check_email_provided -> Required info missing, returning error");
1505 - wp_send_json_error(['message' => esc_html__('No email found', 'mxchat')]);
1506 - }
1507 -}
1508 -
1509 -/**
1510 - * Send error response in appropriate format based on streaming mode
1511 - * ADDED: Helper method to consistently handle errors in both streaming and non-streaming modes
1512 - *
1513 - * @param string $error_message The error message to display
1514 - * @param string $error_code Optional error code for debugging
1515 - */
1516 -private function send_error_response($error_message, $error_code = 'api_error') {
1517 - if ($this->is_streaming) {
1518 - echo "data: " . json_encode([
1519 - 'error' => true,
1520 - 'error_message' => $error_message,
1521 - 'error_code' => $error_code,
1522 - 'text' => $error_message,
1523 - 'message' => $error_message
1524 - ]) . "\n\n";
1525 - echo "data: [DONE]\n\n";
1526 - flush();
1527 - } else {
1528 - wp_send_json_error([
1529 - 'error_message' => $error_message,
1530 - 'error_code' => $error_code
1531 - ]);
1532 - }
1533 - wp_die();
1534 -}
1535 -
1536 148 public function mxchat_handle_chat_request() {
1537 149 global $wpdb;
1538 150
1539 - // Debug: Log incoming bot_id
1540 - $bot_id = isset($_POST['bot_id']) ? sanitize_key($_POST['bot_id']) : 'default';
1541 - //error_log("=== MXCHAT DEBUG: Starting chat request ===");
1542 - //error_log("MXCHAT DEBUG: Bot ID received: " . $bot_id);
1543 -
1544 - // Get bot-specific options
1545 - $bot_options = $this->get_bot_options($bot_id);
1546 - $current_options = !empty($bot_options) ? $bot_options : $this->options;
1547 -
1548 - // Check if this is a streaming request
1549 - // Allow force_streaming_test parameter to bypass the setting check (for admin compatibility testing)
1550 - $force_streaming_test = isset($_POST['force_streaming_test']) && $_POST['force_streaming_test'] === '1' && current_user_can('administrator');
1551 - $is_streaming = isset($_POST['action']) && $_POST['action'] === 'mxchat_stream_chat' &&
1552 - ($force_streaming_test || (isset($current_options['enable_streaming_toggle']) && $current_options['enable_streaming_toggle'] === 'on'));
1553 -
1554 - // ADDED: Store streaming state in class property for use in private methods
1555 - $this->is_streaming = $is_streaming;
1556 -
1557 - // NOTE: Streaming headers are now set later via setup_streaming_headers()
1558 - // This allows actions/forms to return JSON responses without header conflicts
1559 -
1560 - // Check if MX Chat Moderation is active
1561 - if (class_exists('MX_Chat_Moderation')) {
1562 - // Get user email and IP
1563 - $user_email = '';
1564 - $user_ip = $_SERVER['REMOTE_ADDR'];
1565 -
1566 - // If user is logged in, get their email
1567 - if (is_user_logged_in()) {
1568 - $current_user = wp_get_current_user();
1569 - $user_email = $current_user->user_email;
1570 - }
1571 -
1572 - // Create ban handler instance
1573 - $ban_handler = new MX_Chat_Ban_Handler();
1574 -
1575 - // Check if user is banned by IP
1576 - if ($ban_handler->check_ban($user_ip, 'ip')) {
1577 - wp_send_json([
1578 - 'success' => false,
1579 - 'message' => esc_html__('Access denied. Your IP address has been banned.', 'mxchat'),
1580 - 'status' => 'banned'
1581 - ]);
1582 - wp_die();
1583 - }
1584 -
1585 - // If user is logged in, also check email
1586 - if (!empty($user_email) && $ban_handler->check_ban($user_email, 'email')) {
1587 - wp_send_json([
1588 - 'success' => false,
1589 - 'message' => esc_html__('Access denied. Your email address has been banned.', 'mxchat'),
1590 - 'status' => 'banned'
1591 - ]);
1592 - wp_die();
1593 - }
1594 - }
1595 -
1596 - $this->fallbackResponse = ['text' => '', 'html' => '', 'images' => []];
1597 - $this->productCardHtml = '';
1598 - $this->videoEmbedHtml = '';
1599 - // Reset the per-turn function-calling UI capture (plan 48a57a).
1600 - $this->fc_ui_html = '';
1601 - $this->fc_ui_images = array();
1602 - $this->fc_ui_captured = false;
1603 -
1604 - // Get the actual WordPress user ID if logged in
1605 - $is_logged_in = is_user_logged_in();
1606 - if ($is_logged_in) {
1607 - $user_id = get_current_user_id(); // This will get the actual WordPress user ID
1608 - } else {
1609 - // For logged-out users, use your existing identifier method
1610 - $user_id = $this->mxchat_get_user_identifier();
1611 - }
1612 -
1613 151 // Get and sanitize the user identifier
152 + $user_id = $this->mxchat_get_user_identifier();
1614 153 $user_id = sanitize_key($user_id);
1615 154
1616 - // Check rate limit using new settings structure
1617 - $rate_limit_result = $this->check_rate_limit();
155 + // Manage rate limiting
156 + $rate_limit_transient_key = 'mxchat_chat_limit_' . $user_id;
157 + $chat_count = get_transient($rate_limit_transient_key);
158 + $session_transient_key = 'mxchat_chat_session_' . $user_id;
159 + $session_id = get_transient($session_transient_key);
1618 160
1619 - if ($rate_limit_result !== true) {
1620 - wp_send_json([
1621 - 'success' => false,
1622 - 'message' => $rate_limit_result['message'],
1623 - 'status' => 'rate_limit_exceeded'
1624 - ]);
1625 - wp_die();
161 + if ($chat_count === false) {
162 + $chat_count = 0;
1626 163 }
1627 164
1628 - // Rest of your existing code...
1629 - $session_id = isset($_POST['session_id']) ? MxChat_Utils::sanitize_session_id(wp_unslash($_POST['session_id'])) : '';
1630 -
1631 - // Treat the literal strings 'null' / 'undefined' as missing too. Browser edge cases
1632 - // (Safari ITP, private mode, cross-origin iframes with partitioned storage) can cause
1633 - // the frontend FormData.append() to stringify a null session_id into the literal
1634 - // "null", which would otherwise pass empty() and pollute the transcripts table with
1635 - // ghost sessions that group every visitor's first message under one row.
1636 - if ($session_id === 'null' || $session_id === 'undefined') {
1637 - $session_id = '';
165 + if ($session_id === false) {
166 + $session_id = uniqid('mxchat_chat_', true);
167 + set_transient($session_transient_key, $session_id, DAY_IN_SECONDS); // Store session ID for a day
1638 168 }
1639 169
1640 - if (empty($session_id)) {
1641 - wp_send_json_error(esc_html__('Session ID is missing.', 'mxchat'));
1642 - wp_die();
1643 - }
170 + $rate_limit_option = isset($this->options['rate_limit']) ? $this->options['rate_limit'] : 'unlimited';
1644 171
1645 - // Update session owner if it changed (e.g. IP changed due to network switch)
1646 - // The session ID itself is the authentication — if the client has it, they own it
1647 - $current_user_identifier = MxChat_User::mxchat_get_user_identifier();
1648 - $session_owner = get_option("mxchat_session_owner_{$session_id}");
172 + // Check if rate limit is not 'unlimited'
173 + if ($rate_limit_option !== 'unlimited') {
174 + $rate_limit = intval($rate_limit_option);
1649 175
1650 - if (!$session_owner || $session_owner !== $current_user_identifier) {
1651 - update_option("mxchat_session_owner_{$session_id}", $current_user_identifier, 'no');
1652 - }
1653 -
1654 - // Validate and sanitize the incoming message
1655 - if (empty($_POST['message'])) {
1656 - wp_send_json_error(esc_html__('No message received.', 'mxchat'));
1657 - wp_die();
1658 - }
1659 -
1660 - // Enforce the configurable max input length (plan a3fae2 part C). 0 = unlimited.
1661 - // Server-side guard backing the textarea's client-side maxlength (which is bypassable).
1662 - // Reads the global core setting and measures characters (mb_strlen on the unslashed
1663 - // raw POST), matching the maxlength semantics.
1664 - $mxchat_max_input_length = isset($this->options['max_input_length']) ? intval($this->options['max_input_length']) : 0;
1665 - if ($mxchat_max_input_length > 0) {
1666 - $mxchat_incoming_raw = is_string($_POST['message']) ? wp_unslash($_POST['message']) : '';
1667 - if (mb_strlen($mxchat_incoming_raw) > $mxchat_max_input_length) {
1668 - wp_send_json([
1669 - 'success' => false,
1670 - /* translators: %d: maximum allowed characters */
1671 - 'message' => sprintf(esc_html__('Your message is too long. Please keep it under %d characters.', 'mxchat'), $mxchat_max_input_length),
1672 - 'status' => 'message_too_long'
1673 - ]);
176 + if ($chat_count >= $rate_limit) {
177 + wp_send_json_error('Rate limit exceeded. Please try again later.');
1674 178 wp_die();
1675 179 }
180 + set_transient($rate_limit_transient_key, $chat_count + 1, DAY_IN_SECONDS);
1676 181 }
1677 182
1678 -
1679 - // Track originating page for first message in session
1680 - $table_name = $wpdb->prefix . 'mxchat_chat_transcripts';
1681 -
1682 - // Check if originating page columns exist
1683 - $columns_exist = $wpdb->get_var("SHOW COLUMNS FROM $table_name LIKE 'originating_page_url'");
1684 -
1685 - if ($columns_exist) {
1686 - // Check if this session already has messages
1687 - $message_count = $wpdb->get_var($wpdb->prepare(
1688 - "SELECT COUNT(*) FROM $table_name WHERE session_id = %s",
1689 - $session_id
1690 - ));
1691 -
1692 - // If this is the first message in the session
1693 - if ($message_count == 0) {
1694 - // Get originating page from JavaScript (preferred) or HTTP_REFERER (fallback)
1695 - $originating_url = '';
1696 - $originating_title = '';
1697 -
1698 - // Try to get from POST data first (sent by JavaScript)
1699 - if (isset($_POST['current_page_url'])) {
1700 - $originating_url = esc_url_raw($_POST['current_page_url']);
1701 - $originating_title = isset($_POST['current_page_title'])
1702 - ? sanitize_text_field($_POST['current_page_title'])
1703 - : '';
1704 - }
1705 - // Fallback to HTTP_REFERER if not provided by JavaScript
1706 - else if (isset($_SERVER['HTTP_REFERER'])) {
1707 - $originating_url = esc_url_raw($_SERVER['HTTP_REFERER']);
1708 - }
1709 -
1710 - // Generate title if we have URL but no title
1711 - if ($originating_url && empty($originating_title)) {
1712 - $parsed_url = parse_url($originating_url);
1713 - $path = isset($parsed_url['path']) ? trim($parsed_url['path'], '/') : '';
1714 -
1715 - if (empty($path) || $path === 'index.php' || $path === 'index.html') {
1716 - $originating_title = 'Homepage';
1717 - } else {
1718 - // Clean up the path to make a readable title
1719 - $originating_title = str_replace(['-', '_', '/', '.php', '.html'], ' ', $path);
1720 - $originating_title = ucwords(trim($originating_title));
1721 - }
1722 - }
1723 -
1724 - // Store for later use when saving the message
1725 - $this->pending_originating_page = [
1726 - 'url' => $originating_url,
1727 - 'title' => $originating_title
1728 - ];
1729 - }
1730 - }
1731 -
1732 -
1733 -
1734 - // Get page context if provided
1735 - $page_context = null;
1736 - if (isset($_POST['page_context']) && !empty($_POST['page_context'])) {
1737 - $page_context_raw = stripslashes($_POST['page_context']);
1738 - $page_context = json_decode($page_context_raw, true);
1739 -
1740 - // Validate page context structure
1741 - if (is_array($page_context) &&
1742 - isset($page_context['url']) &&
1743 - isset($page_context['title']) &&
1744 - isset($page_context['content'])) {
1745 -
1746 - // Sanitize page context
1747 - $page_context['url'] = esc_url_raw($page_context['url']);
1748 - $page_context['title'] = sanitize_text_field($page_context['title']);
1749 - $page_context['content'] = wp_kses_post($page_context['content']);
1750 - } else {
1751 - $page_context = null;
1752 - }
1753 - }
1754 -
1755 - // Modify the message sanitization to preserve PHP tags in code blocks
1756 - $allowed_tags = [
1757 - 'pre' => [],
1758 - 'code' => ['class' => true],
1759 - 'span' => ['class' => true],
1760 - 'div' => ['class' => true],
1761 - ];
1762 -
1763 - // First preserve code blocks
1764 - $message = preg_replace_callback('/<pre><code.*?>.*?<\/code><\/pre>/s', function($matches) {
1765 - return htmlspecialchars_decode($matches[0]);
1766 - }, $_POST['message']);
1767 -
1768 - // Then apply sanitization
1769 - $message = wp_kses($message, $allowed_tags);
1770 -
1771 - // Preserve code blocks from markdown conversion
1772 - $message = preg_replace('/```(\w+)?\s*([\s\S]+?)```/s', '<pre><code class="$1">$2</code></pre>', $message);
1773 - $message = apply_filters('mxchat_filter_message', $message, 'prompt', $session_id);
1774 -
1775 - // ===== SIMPLIFIED TESTING PANEL INITIALIZATION =====
1776 - // Always initialize testing data for admins (no toggle needed)
1777 - $testing_data = null;
1778 - if (current_user_can('administrator')) {
1779 - // For vision messages, use the original user message for the query display
1780 - $query_for_testing = $message;
1781 - if (isset($_POST['vision_processed']) && $_POST['vision_processed'] && isset($_POST['original_user_message'])) {
1782 - $query_for_testing = sanitize_textarea_field($_POST['original_user_message']);
1783 - }
1784 -
1785 - $testing_data = [
1786 - 'query' => $query_for_testing,
1787 - 'timestamp' => time(),
1788 - 'top_matches' => [],
1789 - 'action_matches' => [], // Initialize action matches array
1790 - 'page_context' => $page_context, // Include page context in testing data
1791 - 'is_vision' => isset($_POST['vision_processed']) && $_POST['vision_processed'],
1792 - 'bot_id' => $bot_id // Include bot ID in testing data
1793 - ];
1794 -
1795 - // Get similarity threshold from bot options or default options
1796 - $similarity_threshold = isset($current_options['similarity_threshold'])
1797 - ? ((int) $current_options['similarity_threshold']) / 100
1798 - : 0.35;
1799 -
1800 - $testing_data['similarity_threshold'] = $similarity_threshold;
1801 -
1802 - // Determine knowledge base type using bot-specific config
1803 - $bot_pinecone_config = $this->get_bot_pinecone_config($bot_id);
1804 - $use_pinecone = isset($bot_pinecone_config['use_pinecone']) ? $bot_pinecone_config['use_pinecone'] : false;
1805 - $testing_data['knowledge_base_type'] = $use_pinecone ? 'Pinecone' : 'WordPress Database';
1806 - }
1807 - // ===== END SIMPLIFIED TESTING INITIALIZATION =====
1808 -
1809 - // Add debug before and after:
1810 - //error_log('MxChat Core: About to call mxchat_pre_process_message filter with message: ' . $message);
1811 - $pre_processed_result = apply_filters('mxchat_pre_process_message', $message, $user_id, $session_id);
1812 - //error_log('MxChat Core: Filter returned: ' . (is_array($pre_processed_result) ? 'array' : $pre_processed_result));
1813 -
1814 -
1815 - // If the pre-processing returned a result (not the original message), use it directly
1816 - if (is_array($pre_processed_result) && isset($pre_processed_result['text'])) {
1817 - // Save the AI response
1818 - $this->mxchat_save_chat_message($session_id, 'bot', $pre_processed_result['text']);
1819 -
1820 - // Save HTML content if provided
1821 - if (!empty($pre_processed_result['html'])) {
1822 - $this->mxchat_save_chat_message($session_id, 'bot', $pre_processed_result['html']);
1823 - }
1824 -
1825 - // Add testing data if admin
1826 - $response_data = [
1827 - 'text' => $pre_processed_result['text'],
1828 - 'html' => $pre_processed_result['html'] ?? '',
1829 - 'session_id' => $session_id
1830 - ];
1831 -
1832 - if ($testing_data !== null) {
1833 - $response_data['testing_data'] = $testing_data;
1834 - }
1835 -
1836 - wp_send_json($response_data);
1837 - wp_die();
1838 - }
1839 -
1840 - // Save the user's message - handle vision processed messages differently
1841 - if (isset($_POST['vision_processed']) && $_POST['vision_processed'] && isset($_POST['original_user_message'])) {
1842 - // For vision messages, save the original user message with image indicator
1843 - $original_message = sanitize_textarea_field($_POST['original_user_message']);
1844 - if (isset($_POST['vision_images_count']) && $_POST['vision_images_count'] > 0) {
1845 - $image_count = intval($_POST['vision_images_count']);
1846 - $original_message .= " [{$image_count} image(s)]";
1847 - }
1848 - $this->mxchat_save_chat_message($session_id, 'user', $original_message);
1849 - } else {
1850 - // Regular message - save as normal
1851 - $this->mxchat_save_chat_message($session_id, 'user', $message);
1852 - }
1853 -
1854 -
1855 - if (is_email($message)) {
1856 - // Add the email to Loops
1857 - $this->add_email_to_loops($message);
1858 -
1859 - // Get the user's success message instruction using current_options
1860 - $user_success_message = $current_options['email_capture_response'] ?? __('Thank you for providing your email! You\'ve been added to our list.', 'mxchat');
1861 -
1862 - // Set instruction for AI using the user's success message
1863 - $this->current_action_instruction = $user_success_message;
1864 -
1865 - // Clear the email capture transient since we got the email
1866 - delete_transient('mxchat_email_capture_' . $user_id);
1867 - }
1868 -
1869 - // Check if we're in an email capture flow but user hasn't provided email yet
1870 - elseif (get_transient('mxchat_email_capture_' . $user_id)) {
1871 - // Check if the message contains an email (not the whole message being an email)
1872 - if (preg_match('/\b[A-Za-z0-9._%+-]+@[A-Za-z0-9.-]+\.[A-Z|a-z]{2,}\b/', $message, $matches)) {
1873 - $extracted_email = $matches[0];
1874 -
1875 - // Add the extracted email to Loops
1876 - $this->add_email_to_loops($extracted_email);
1877 -
1878 - // Get the user's success message instruction using current_options
1879 - $user_success_message = $current_options['email_capture_response'] ?? __('Thank you for providing your email! You\'ve been added to our list.', 'mxchat');
1880 -
1881 - // Set instruction for AI using the user's success message
1882 - $this->current_action_instruction = $user_success_message;
1883 -
1884 - // Clear the email capture transient since we got the email
1885 - delete_transient('mxchat_email_capture_' . $user_id);
1886 - }
1887 - // If no email found but we're in capture mode, remind them
1888 - else {
1889 - // Get the original instruction to remind them using current_options
1890 - $original_instruction = $current_options['triggered_phrase_response'] ?? __("Please provide your email address.", 'mxchat');
1891 - $this->current_action_instruction = $original_instruction;
1892 - }
1893 - }
1894 -
1895 - $intent_info = '';
1896 -
1897 - // Check chat mode
1898 - $chat_mode = get_option("mxchat_mode_{$session_id}", 'ai');
1899 -
1900 - // Handle agent mode
1901 - // Handle agent mode
1902 - if ($chat_mode === 'agent') {
1903 - // First, check for switch intent before doing anything else
1904 - $intent_matched = $this->mxchat_check_intent_and_invoke_callback($message, $user_id, $session_id);
1905 -
1906 - // Capture action analysis for testing panel after intent check
1907 - if ($testing_data !== null && isset($this->last_action_analysis) && !empty($this->last_action_analysis)) {
1908 - $testing_data['action_matches'] = $this->last_action_analysis;
1909 - }
1910 -
1911 - // Around line 506, in the agent mode handling section:
1912 - if ($intent_matched && !empty($this->fallbackResponse['text'])) {
1913 - // Update chat mode first
1914 - update_option("mxchat_mode_{$session_id}", 'ai');
1915 -
1916 - // Clear any existing PDF context to start fresh
1917 - $this->clear_pdf_transients($session_id);
1918 -
1919 - // Prepare clean switch response with explicit chat_mode
1920 - $response_data = [
1921 - 'text' => $this->fallbackResponse['text'],
1922 - 'html' => $this->fallbackResponse['html'] ?? '',
1923 - 'session_id' => $session_id,
1924 - 'chat_mode' => 'ai' // EXPLICITLY SET THIS
1925 - ];
1926 -
1927 - if ($testing_data !== null) {
1928 - $response_data['testing_data'] = $testing_data;
1929 - }
1930 -
1931 - // Save the mode switch message
1932 - $this->mxchat_save_chat_message($session_id, 'system', esc_html__('Switched to AI chat mode', 'mxchat'));
1933 - $this->mxchat_save_chat_message($session_id, 'bot', $this->fallbackResponse['text']);
1934 -
1935 - // Send response and exit
1936 - wp_send_json($response_data);
1937 - wp_die();
1938 - } elseif (!$intent_matched) {
1939 - // No intent matched, handle live agent message
1940 - try {
1941 - $this->mxchat_send_user_message_to_agent($message, $user_id, $session_id);
1942 -
1943 - $agent_response = [
1944 - 'status' => 'waiting_for_agent',
1945 - 'message' => esc_html__('Message sent to live agent.', 'mxchat')
1946 - ];
1947 -
1948 - if ($testing_data !== null) {
1949 - $agent_response['testing_data'] = $testing_data;
1950 - }
1951 -
1952 - wp_send_json_success($agent_response);
1953 - } catch (\Exception $e) {
1954 - wp_send_json_error(esc_html__('Failed to send message to agent', 'mxchat'));
1955 - }
1956 - wp_die();
1957 - }
1958 - }
1959 -
1960 - // Step 1: Check for new PDF URL in the message
1961 - if (!isset($_POST['vision_processed']) && preg_match('/https?:\/\/[^\s"]+/i', $message, $matches)) {
1962 - $new_pdf_url = $matches[0];
1963 -
1964 - // Check if this is likely a PDF-related request
1965 - $pdf_keywords = ['pdf', 'document', 'read', 'analyze'];
1966 - $is_pdf_request = false;
1967 -
1968 - foreach ($pdf_keywords as $keyword) {
1969 - if (stripos($message, $keyword) !== false) {
1970 - $is_pdf_request = true;
1971 - break;
1972 - }
1973 - }
1974 -
1975 - // If it looks like a PDF request or we're waiting for a PDF URL
1976 - if ($is_pdf_request || get_transient('mxchat_waiting_for_pdf_url_' . $session_id)) {
1977 - // Validate HTTPS
1978 - if (wp_http_validate_url($new_pdf_url) && parse_url($new_pdf_url, PHP_URL_SCHEME) === 'https') {
1979 - // Extract filename from URL
1980 - $pdf_filename = basename(parse_url($new_pdf_url, PHP_URL_PATH));
1981 -
1982 - // Clear previous PDF transients
1983 - $this->clear_pdf_transients($session_id);
1984 -
1985 - // Process new PDF using current_options
1986 - $max_pages = $current_options['pdf_max_pages'] ?? 69;
1987 - $embeddings = $this->fetch_and_split_pdf_pages($new_pdf_url, $max_pages);
1988 -
1989 - if ($embeddings === 'too_many_pages') {
1990 - $error_text = sprintf(
1991 - $current_options['pdf_intent_error_text'] ??
1992 - esc_html__("The provided PDF exceeds the maximum allowed limit of %d pages. Please provide a smaller document.", 'mxchat'),
1993 - $max_pages
1994 - );
1995 - $this->fallbackResponse['text'] = $error_text;
1996 - } elseif ($embeddings) {
1997 - // Store new PDF information
1998 - $pdf_filename = basename(parse_url($new_pdf_url, PHP_URL_PATH));
1999 -
2000 - // If the filename is generic, create a more descriptive one
2001 - if (in_array($pdf_filename, ['results_download.php', 'download.php', 'view.php', 'pdf.php']) ||
2002 - strpos($pdf_filename, '.php') !== false) {
2003 - $pdf_filename = 'Document_' . date('Y-m-d_H-i') . '.pdf';
2004 - }
2005 -
2006 - set_transient('mxchat_pdf_url_' . $session_id, $new_pdf_url, HOUR_IN_SECONDS);
2007 - set_transient('mxchat_pdf_filename_' . $session_id, $pdf_filename, HOUR_IN_SECONDS);
2008 - set_transient('mxchat_pdf_embeddings_' . $session_id, $embeddings, HOUR_IN_SECONDS);
2009 - set_transient('mxchat_include_pdf_in_context_' . $session_id, true, HOUR_IN_SECONDS);
2010 -
2011 - $success_text = $current_options['pdf_intent_success_text'] ??
2012 - esc_html__("I've processed the new PDF '{$pdf_filename}'. What questions do you have about it?", 'mxchat');
2013 -
2014 - $pdf_response = [
2015 - 'success' => true,
2016 - 'message' => $success_text,
2017 - 'data' => [
2018 - 'filename' => $pdf_filename
2019 - ]
2020 - ];
2021 -
2022 - if ($testing_data !== null) {
2023 - $pdf_response['testing_data'] = $testing_data;
2024 - }
2025 -
2026 - wp_send_json($pdf_response);
2027 - wp_die();
2028 - } else {
2029 - $error_text = $current_options['pdf_intent_error_text'] ??
2030 - esc_html__("Sorry, I couldn't process the PDF. Please ensure it's a valid file.", 'mxchat');
2031 - $this->fallbackResponse['text'] = $error_text;
2032 - }
2033 -
2034 - $pdf_error_response = [
2035 - 'success' => false,
2036 - 'message' => $this->fallbackResponse['text']
2037 - ];
2038 -
2039 - if ($testing_data !== null) {
2040 - $pdf_error_response['testing_data'] = $testing_data;
2041 - }
2042 -
2043 - wp_send_json($pdf_error_response);
2044 - wp_die();
2045 - }
2046 - }
2047 - }
2048 -
2049 -
2050 - // Step 2: Detect intent and handle intent-based responses
2051 - $intent_result = $this->mxchat_check_intent_and_invoke_callback($message, $user_id, $session_id);
2052 -
2053 - // Capture action analysis for testing panel after intent check
2054 - if ($testing_data !== null && isset($this->last_action_analysis) && !empty($this->last_action_analysis)) {
2055 - $testing_data['action_matches'] = $this->last_action_analysis;
2056 - }
2057 -
2058 - // Step 3: Handle the intent result appropriately
2059 - if ($intent_result !== false) {
2060 - // Intent was matched - ALWAYS send as JSON response, never streaming
2061 -
2062 - if (is_array($intent_result) && (isset($intent_result['text']) || isset($intent_result['html']))) {
2063 - // Intent returned a direct response array
2064 - $response_data = [
2065 - 'text' => $intent_result['text'] ?? '',
2066 - 'html' => $intent_result['html'] ?? '',
2067 - 'session_id' => $session_id
2068 - ];
2069 -
2070 - // IMPORTANT: Include chat_mode if present (for WhatsApp, Slack, etc.)
2071 - if (isset($intent_result['chat_mode'])) {
2072 - $response_data['chat_mode'] = $intent_result['chat_mode'];
2073 - }
2074 -
2075 - if ($testing_data !== null) {
2076 - $response_data['testing_data'] = $testing_data;
2077 - }
2078 -
2079 - wp_send_json($response_data);
2080 - wp_die();
2081 - } else if ($intent_result === true && (!empty($this->fallbackResponse['text']) || !empty($this->fallbackResponse['html']))) {
2082 - // Intent returned true and set fallbackResponse
2083 -
2084 - // SAVE TO TRANSCRIPT
2085 - if (!empty($this->fallbackResponse['text'])) {
2086 - $this->mxchat_save_chat_message($session_id, 'bot', $this->fallbackResponse['text']);
2087 - }
2088 - // Save action HTML (product cards, featured products, etc.) so it renders in transcripts
2089 - if (!empty($this->fallbackResponse['html'])) {
2090 - $this->mxchat_save_chat_message($session_id, 'bot', $this->fallbackResponse['html']);
2091 - }
2092 -
2093 - $response_data = [
2094 - 'text' => $this->fallbackResponse['text'] ?? '',
2095 - 'html' => $this->fallbackResponse['html'] ?? '',
2096 - 'session_id' => $session_id
2097 - ];
2098 -
2099 - if (isset($this->fallbackResponse['chat_mode'])) {
2100 - $response_data['chat_mode'] = $this->fallbackResponse['chat_mode'];
2101 - }
2102 -
2103 - if ($testing_data !== null) {
2104 - $response_data['testing_data'] = $testing_data;
2105 - }
2106 -
2107 - wp_send_json($response_data);
2108 - wp_die();
2109 - }
2110 - }
2111 -
2112 - // If we get here, no intent matched OR the intent didn't provide a usable response
2113 -
2114 - // Step 4: Generate AI response
2115 - // Get session start timestamp - when persistence is OFF, only include messages from this page load
2116 - $session_start_timestamp = isset($_POST['session_start_timestamp']) ? intval($_POST['session_start_timestamp']) : 0;
2117 - $conversation_history = $this->mxchat_fetch_conversation_history_for_ai($session_id, $session_start_timestamp);
2118 - $this->mxchat_increment_chat_count();
2119 -
2120 - // Generate embedding for the user's query - USE BOT-SPECIFIC API KEY
2121 - $api_key = $current_options['api_key'] ?? $this->options['api_key'];
2122 - $user_message_embedding = $this->mxchat_generate_embedding($message, $api_key);
2123 -
2124 - // Check if the embedding generation returned an error
2125 - if (is_array($user_message_embedding) && isset($user_message_embedding['error'])) {
2126 - $error_message = $user_message_embedding['error'];
2127 - $error_code = $user_message_embedding['error_code'] ?? 'embedding_error';
2128 -
2129 - // FIXED: Send error in appropriate format based on streaming mode
2130 - if ($is_streaming) {
2131 - echo "data: " . json_encode([
2132 - 'error' => true,
2133 - 'error_message' => $error_message,
2134 - 'error_code' => $error_code,
2135 - 'text' => $error_message,
2136 - 'message' => $error_message
2137 - ]) . "\n\n";
2138 - echo "data: [DONE]\n\n";
2139 - flush();
2140 - } else {
2141 - wp_send_json_error([
2142 - 'error_message' => $error_message,
2143 - 'error_code' => $error_code
2144 - ]);
2145 - }
2146 - wp_die();
2147 - }
2148 -
2149 - // Check if the embedding is valid
2150 - if (!is_array($user_message_embedding) || empty($user_message_embedding)) {
2151 - $error_message = esc_html__('Unable to process your message. The embedding service is not responding correctly.', 'mxchat');
2152 -
2153 - // FIXED: Send error in appropriate format based on streaming mode
2154 - if ($is_streaming) {
2155 - echo "data: " . json_encode([
2156 - 'error' => true,
2157 - 'error_message' => $error_message,
2158 - 'error_code' => 'invalid_embedding',
2159 - 'text' => $error_message,
2160 - 'message' => $error_message
2161 - ]) . "\n\n";
2162 - echo "data: [DONE]\n\n";
2163 - flush();
2164 - } else {
2165 - wp_send_json_error([
2166 - 'error_message' => $error_message,
2167 - 'error_code' => 'invalid_embedding'
2168 - ]);
2169 - }
2170 - wp_die();
2171 - }
2172 -
2173 - // Build context with both knowledge base and PDF content if available
2174 - $context_content = "User asked: '{$message}'\n\n";
2175 -
2176 - // Add action instruction if present (add this right after the above line)
2177 - if (!empty($this->current_action_instruction)) {
2178 - $context_content .= "===== SPECIAL INSTRUCTION =====\n";
2179 - $context_content .= "IMPORTANT: " . $this->current_action_instruction . "\n";
2180 - $context_content .= "Respond naturally and conversationally while following this instruction.\n";
2181 - $context_content .= "===== END SPECIAL INSTRUCTION =====\n\n";
2182 -
2183 - // Clear the instruction after using it
2184 - $this->current_action_instruction = null;
2185 - }
2186 -
2187 -
2188 - // Add page context if available and contextual awareness is enabled using current_options
2189 - if ($page_context && isset($current_options['contextual_awareness_toggle']) && $current_options['contextual_awareness_toggle'] === 'on') {
2190 - $context_content .= "===== CURRENT PAGE CONTEXT =====\n";
2191 - $context_content .= "Page URL: " . $page_context['url'] . "\n";
2192 - $context_content .= "Page Title: " . $page_context['title'] . "\n";
2193 - $context_content .= "Page Content: " . $page_context['content'] . "\n";
2194 - $context_content .= "===== END CURRENT PAGE CONTEXT =====\n\n";
2195 - }
2196 -
2197 - // Get relevant content from knowledge base - PASS BOT_ID and MESSAGE for Vector Store
2198 - $relevant_content = $this->mxchat_find_relevant_content($user_message_embedding, $bot_id, $message);
2199 -
2200 - // NEW: Also extract URLs from system instructions (only if citation links enabled)
2201 - // Use fresh options to ensure we get the latest setting value
2202 - $fresh_options = get_option('mxchat_options', []);
2203 - $citation_links_enabled = isset($fresh_options['citation_links_toggle']) ? ($fresh_options['citation_links_toggle'] === 'on') : true;
2204 -
2205 - $system_instructions = $this->get_system_instructions($bot_id, $session_id);
2206 - if ($citation_links_enabled && !empty($system_instructions)) {
2207 - preg_match_all(
2208 - '#\bhttps?://[^\s<>"\']+#i',
2209 - $system_instructions,
2210 - $system_instruction_urls
2211 - );
2212 -
2213 - if (!empty($system_instruction_urls[0])) {
2214 - // Merge with existing valid URLs
2215 - $this->current_valid_urls = array_merge(
2216 - $this->current_valid_urls,
2217 - $system_instruction_urls[0]
2218 - );
2219 - // Remove duplicates
2220 - $this->current_valid_urls = array_unique($this->current_valid_urls);
2221 -
2222 - //error_log("Added " . count($system_instruction_urls[0]) . " URLs from system instructions");
2223 - }
2224 - }
2225 -
2226 -// ===== CAPTURE REAL SIMILARITY DATA FOR ADMINS =====
2227 -if ($testing_data !== null && $this->last_similarity_analysis !== null) {
2228 - // Update testing data with the REAL similarity analysis
2229 - $testing_data['top_matches'] = $this->last_similarity_analysis['top_matches'];
2230 - $testing_data['total_documents_checked'] = $this->last_similarity_analysis['total_checked'] ?? 0;
2231 - $testing_data['knowledge_base_type'] = $this->last_similarity_analysis['knowledge_base_type'];
2232 - $testing_data['sources_used'] = $this->last_similarity_analysis['sources_used'] ?? 0;
2233 - $testing_data['total_chunks_used'] = $this->last_similarity_analysis['total_chunks_used'] ?? 0;
2234 -}
2235 -// ===== END SIMILARITY DATA CAPTURE =====
2236 -
2237 -// NEW: Add valid URLs to testing data for admin panel display (AFTER similarity data)
2238 -if ($testing_data !== null && !empty($this->current_valid_urls)) {
2239 - $testing_data['approved_urls'] = array_values($this->current_valid_urls);
2240 - //error_log("Added " . count($this->current_valid_urls) . " approved URLs to testing data");
2241 -}
2242 -
2243 - $kb_block = !empty($relevant_content)
2244 - ? "===== OFFICIAL KNOWLEDGE DATABASE CONTENT =====\n" . $relevant_content . "\n===== END OF OFFICIAL KNOWLEDGE DATABASE CONTENT =====\n\n"
2245 - : "===== NO RELEVANT CONTENT FOUND IN KNOWLEDGE DATABASE =====\n";
2246 -
2247 - // {context} placeholder (plan 59bc1b): when the resolved instructions
2248 - // carry the token, the KB block is injected at that spot by
2249 - // get_system_instructions() (every provider handler re-calls it) and is
2250 - // NOT appended here — otherwise the block would ride twice.
2251 - // $system_instructions above was resolved while context_kb_block was
2252 - // still null, so the literal token is still visible for this check.
2253 - if (!empty($system_instructions) && stripos($system_instructions, '{context}') !== false) {
2254 - $this->context_kb_block = $kb_block;
2255 - } else {
2256 - $context_content .= $kb_block;
2257 - }
2258 -
2259 - // NEW: Add approved URLs list to context for AI (only if citation links enabled)
2260 - if ($citation_links_enabled && !empty($this->current_valid_urls)) {
2261 - $context_content .= "===== APPROVED URLS FOR CITATIONS =====\n";
2262 - $context_content .= "You may ONLY use these exact URLs in your response:\n";
2263 - foreach ($this->current_valid_urls as $url) {
2264 - $context_content .= "- " . $url . "\n";
2265 - }
2266 - $context_content .= "\nCRITICAL: Do NOT create, modify, extend, or invent any other URLs. ";
2267 - $context_content .= "===== END APPROVED URLS =====\n\n";
2268 - }
2269 -
2270 - // Check for and include PDF content
2271 - $pdf_url = get_transient('mxchat_pdf_url_' . $session_id);
2272 - $pdf_embeddings = get_transient('mxchat_pdf_embeddings_' . $session_id);
2273 - $pdf_filename = get_transient('mxchat_pdf_filename_' . $session_id);
2274 - if ($pdf_url && $pdf_embeddings && get_transient('mxchat_include_pdf_in_context_' . $session_id)) {
2275 - $relevant_pdf_pages = $this->find_relevant_pdf_pages($user_message_embedding, $pdf_embeddings);
2276 - if (!empty($relevant_pdf_pages)) {
2277 - $context_content .= "Relevant content from PDF document '{$pdf_filename}':\n";
2278 - foreach ($relevant_pdf_pages as $page_data) {
2279 - $context_content .= "Page {$page_data['page_number']} of '{$pdf_filename}': {$page_data['text']}\n";
2280 - }
2281 - $context_content .= "\n";
2282 - }
2283 - }
2284 -
2285 - // Check for and include Word content
2286 - $word_url = get_transient('mxchat_word_url_' . $session_id);
2287 - $word_embeddings = get_transient('mxchat_word_embeddings_' . $session_id);
2288 - $word_filename = get_transient('mxchat_word_filename_' . $session_id);
2289 - if ($word_url && $word_embeddings && get_transient('mxchat_include_word_in_context_' . $session_id)) {
2290 - $relevant_word_chunks = $this->word_handler->mxchat_find_relevant_word_chunks($user_message_embedding, $word_embeddings);
2291 - if (!empty($relevant_word_chunks)) {
2292 - $context_content .= "Relevant content from Word document '{$word_filename}':\n";
2293 - foreach ($relevant_word_chunks as $chunk_data) {
2294 - $context_content .= "Section {$chunk_data['chunk_number']} of '{$word_filename}': {$chunk_data['text']}\n";
2295 - }
2296 - $context_content .= "\n";
2297 - }
2298 - }
2299 -
2300 - $context_content = apply_filters('mxchat_prepare_context', $context_content, $session_id);
2301 -
2302 - // Extract model from current options for bot-specific model support
2303 - $selected_model = isset($current_options['model']) ? $current_options['model'] : 'gpt-5.6-sol';
2304 -
2305 - // ===== Native function-calling fallback (plan-mxchat-20260617-a41dee) =====
2306 - // Intents already missed (we're past the intent router). If function
2307 - // calling is enabled and the active model is tool-capable, let the model
2308 - // SELECT and run registered callbacks as tools — independent of intents,
2309 - // works with zero Actions. The tool round is buffered; the final answer is
2310 - // emitted via the SAME envelopes the normal path uses. Default-off, so
2311 - // existing installs never enter this branch.
2312 - if ($this->mxchat_fc_should_run($selected_model)) {
2313 - $fc_outcome = $this->mxchat_fc_attempt(
2314 - $message,
2315 - $context_content,
2316 - $conversation_history,
2317 - $selected_model,
2318 - $current_options,
2319 - $session_id,
2320 - $user_id
2321 - );
2322 - if (is_array($fc_outcome) && !empty($fc_outcome['handled'])) {
2323 - $fc_text = isset($fc_outcome['text']) ? $fc_outcome['text'] : '';
2324 - if (!empty($this->current_valid_urls)) {
2325 - $fc_text = $this->validate_and_clean_urls($fc_text, $this->current_valid_urls, $session_id, $bot_id);
2326 - }
2327 - // plan-mxchat-20260617-48a57a — surface any UI element a tool
2328 - // produced (generated image / product card / image gallery) so the
2329 - // widget RENDERS it, instead of emitting only the model's text.
2330 - // The html was already saved to the transcript in
2331 - // mxchat_fc_execute_tool (or by the callback itself for self-saving
2332 - // core tools), so we persist ONLY the model's caption text here.
2333 - $fc_html = isset($this->fc_ui_html) ? $this->fc_ui_html : '';
2334 -
2335 - if ($fc_text !== '') {
2336 - $this->mxchat_save_chat_message($session_id, 'bot', $fc_text, null, null);
2337 - }
2338 -
2339 - // A video-backed KB source queued during retrieval (03ba33) must
2340 - // surface on the FC path too — the FC envelopes below are the ONLY
2341 - // exit for this turn, so append it to the html channel and persist
2342 - // it (tool html was already saved in mxchat_fc_execute_tool; the
2343 - // video embed has no other save point on this path).
2344 - if (!empty($this->videoEmbedHtml)) {
2345 - $fc_html .= $this->videoEmbedHtml;
2346 - $this->mxchat_save_chat_message($session_id, 'bot', $this->videoEmbedHtml);
2347 - }
2348 -
2349 - if ($is_streaming) {
2350 - // The frontend SSE reader routes any event carrying text/html
2351 - // to handleNonStreamResponse(), which renders text + html in a
2352 - // single bot message — so emit one complete event (mirrors the
2353 - // intent path's text/html envelope).
2354 - $sse = array('session_id' => $session_id);
2355 - if ($fc_text !== '') $sse['text'] = $fc_text;
2356 - if ($fc_html !== '') $sse['html'] = $fc_html;
2357 - if ($fc_text === '' && $fc_html === '') $sse['text'] = $this->mxchat_fc_giveup_text();
2358 - echo "data: " . wp_json_encode($sse) . "\n\n";
2359 - echo "data: [DONE]\n\n";
2360 - flush();
2361 - } else {
2362 - $fc_response_data = array('text' => $fc_text, 'html' => $fc_html, 'session_id' => $session_id);
2363 - if ($testing_data !== null) {
2364 - $fc_response_data['testing_data'] = $testing_data;
2365 - }
2366 - wp_send_json($fc_response_data);
2367 - }
2368 - wp_die();
2369 - }
2370 - }
2371 - // ===== end function-calling fallback =====
2372 -
2373 - // Streaming + a queued video embed (03ba33): the provider handlers own the
2374 - // token stream and the [DONE] terminator, so the embed rides a dedicated
2375 - // append_html SSE event emitted BEFORE the stream starts. The client
2376 - // stashes it and appends it as its own bot bubble after [DONE] — old
2377 - // cached widget JS simply ignores the unknown key (no content/text/html/
2378 - // error field, so no branch matches). Transcript save happens after the
2379 - // stream completes, so history order matches the live order (text, then
2380 - // embed).
2381 - if ($is_streaming && !empty($this->videoEmbedHtml)) {
2382 - echo "data: " . wp_json_encode(array(
2383 - 'append_html' => $this->videoEmbedHtml,
2384 - 'session_id' => $session_id,
2385 - )) . "\n\n";
2386 - flush();
2387 - }
2388 -
2389 - $response = $this->mxchat_generate_response(
2390 - $context_content,
2391 - $current_options['api_key'] ?? $this->options['api_key'],
2392 - $current_options['xai_api_key'] ?? $this->options['xai_api_key'],
2393 - $current_options['claude_api_key'] ?? $this->options['claude_api_key'],
2394 - $current_options['deepseek_api_key'] ?? $this->options['deepseek_api_key'],
2395 - $current_options['gemini_api_key'] ?? $this->options['gemini_api_key'],
2396 - $current_options['openrouter_api_key'] ?? $this->options['openrouter_api_key'],
2397 - $conversation_history,
2398 - $is_streaming,
2399 - $session_id,
2400 - $testing_data,
2401 - $selected_model
2402 - );
2403 -
2404 - // Handle streaming vs non-streaming responses
2405 - if ($is_streaming) {
2406 - // Check if streaming actually happened or if it fell back to regular response
2407 - if ($response === true) {
2408 - // Persist the video embed AFTER the provider saved the streamed
2409 - // text, so history replays in the same order the visitor saw
2410 - // (text bubble, then embed bubble). See 03ba33.
2411 - if (!empty($this->videoEmbedHtml)) {
2412 - $this->mxchat_save_chat_message($session_id, 'bot', $this->videoEmbedHtml);
2413 - }
2414 - wp_die();
2415 - }
2416 - // If we get here, streaming fell back to regular response, continue
2417 - // But if there's an error, we need to send it as SSE format since headers are already set
2418 - if (is_array($response) && isset($response['error'])) {
2419 - $error_message = $response['error'];
2420 - $error_code = $response['error_code'] ?? 'api_error';
2421 - // Send error in SSE format that the client JS can handle
2422 - echo "data: " . json_encode([
2423 - 'error' => true,
2424 - 'error_message' => $error_message,
2425 - 'error_code' => $error_code,
2426 - 'text' => $error_message, // Also include as text for fallback handling
2427 - 'message' => $error_message
2428 - ]) . "\n\n";
2429 - echo "data: [DONE]\n\n";
2430 - flush();
2431 - wp_die();
2432 - }
2433 - }
2434 -
2435 - // Check if the response is an error array (non-streaming mode)
2436 - if (is_array($response) && isset($response['error'])) {
2437 - wp_send_json_error([
2438 - 'error_message' => $response['error'],
2439 - 'error_code' => $response['error_code'] ?? 'api_error'
2440 - ]);
2441 - wp_die();
2442 - }
2443 -
2444 - // DEBUG: Check what we have
2445 - //error_log("=== BEFORE URL VALIDATION ===");
2446 - //error_log("current_valid_urls is empty? " . (empty($this->current_valid_urls) ? 'YES' : 'NO'));
2447 - //error_log("current_valid_urls count: " . count($this->current_valid_urls));
2448 - //error_log("current_valid_urls content: " . print_r($this->current_valid_urls, true));
2449 -
2450 - // If we get here, the response is valid text - now validate URLs
2451 - if (!empty($this->current_valid_urls)) {
2452 - //error_log("CALLING validate_and_clean_urls");
2453 - $response = $this->validate_and_clean_urls($response, $this->current_valid_urls, $session_id, $bot_id);
2454 - } else {
2455 - //error_log("SKIPPING validation - current_valid_urls is empty");
2456 - }
2457 - // ===== END URL VALIDATION =====
2458 -
2459 - // Prepare RAG context data for storage (only include documents used for context)
2460 - $rag_context_for_storage = null;
2461 - $has_rag_data = $this->last_similarity_analysis !== null && !empty($this->last_similarity_analysis['top_matches']);
2462 - $has_action_data = isset($this->last_action_analysis) && !empty($this->last_action_analysis);
2463 -
2464 - if ($has_rag_data || $has_action_data) {
2465 - $rag_context_for_storage = [];
2466 -
2467 - // Add RAG/source data if available
2468 - if ($has_rag_data) {
2469 - $rag_context_for_storage['top_matches'] = $this->last_similarity_analysis['top_matches'];
2470 - $rag_context_for_storage['approved_urls'] = $this->current_valid_urls ?? [];
2471 - $rag_context_for_storage['similarity_threshold'] = $this->last_similarity_analysis['threshold_used'] ?? 0.35;
2472 - $rag_context_for_storage['knowledge_base_type'] = $this->last_similarity_analysis['knowledge_base_type'] ?? 'WordPress Database';
2473 - $rag_context_for_storage['total_documents_checked'] = $this->last_similarity_analysis['total_checked'] ?? 0;
2474 - $rag_context_for_storage['sources_used'] = $this->last_similarity_analysis['sources_used'] ?? 0;
2475 - $rag_context_for_storage['total_chunks_used'] = $this->last_similarity_analysis['total_chunks_used'] ?? 0;
2476 - }
2477 -
2478 - // Add action analysis data if available
2479 - if ($has_action_data) {
2480 - $rag_context_for_storage['action_analysis'] = $this->last_action_analysis;
2481 - }
2482 - }
2483 -
2484 - // Save the cleaned response with RAG context
2485 - $this->mxchat_save_chat_message($session_id, 'bot', $response, null, $rag_context_for_storage);
2486 -
2487 - // Step 5: Save additional content if available
2488 - if (!empty($this->productCardHtml)) {
2489 - $this->mxchat_save_chat_message($session_id, 'bot', $this->productCardHtml);
2490 - }
2491 -
2492 - if (!empty($this->fallbackResponse['html'])) {
2493 - $this->mxchat_save_chat_message($session_id, 'bot', $this->fallbackResponse['html']);
2494 - }
2495 -
2496 - if (!empty($this->videoEmbedHtml)) {
2497 - $this->mxchat_save_chat_message($session_id, 'bot', $this->videoEmbedHtml);
2498 - }
2499 -
2500 - // Step 6: Return the response
2501 - // DEBUG: Check if newlines exist in the response
2502 - //error_log("=== MXCHAT NON-STREAMING RESPONSE DEBUG ===");
2503 - //error_log("Response has newlines: " . (strpos($response, "\n") !== false ? 'YES' : 'NO'));
2504 - //error_log("Response first 500 chars: " . substr($response, 0, 500));
2505 -
2506 - // Product cards and action html keep their existing either/or precedence;
2507 - // a queued video embed (03ba33) is APPENDED so it can coexist with both.
2508 - $additional_html = !empty($this->productCardHtml) ? $this->productCardHtml : ($this->fallbackResponse['html'] ?? '');
2509 - if (!empty($this->videoEmbedHtml)) {
2510 - $additional_html .= $this->videoEmbedHtml;
2511 - }
2512 -
2513 - $response_data = [
2514 - 'text' => $response,
2515 - 'html' => $additional_html,
2516 - 'session_id' => $session_id
2517 - ];
2518 -
2519 - // Include vectorstore error info for admin debugging (only visible to admins via testing_data)
2520 - if (!empty($this->last_vectorstore_error) && $testing_data !== null) {
2521 - $testing_data['vectorstore_error'] = $this->last_vectorstore_error;
2522 - }
2523 -
2524 - // Also pass it as a top-level field so JS can show a better error message to admins
2525 - if (!empty($this->last_vectorstore_error) && current_user_can('manage_options')) {
2526 - $response_data['vectorstore_error'] = $this->last_vectorstore_error;
2527 - }
2528 -
2529 - // Always add testing data for admins (no toggle needed)
2530 - if ($testing_data !== null) {
2531 - $response_data['testing_data'] = $testing_data;
2532 - }
2533 -
2534 - wp_send_json($response_data);
183 + // Validate and sanitize the incoming message
184 + if (!isset($_POST['message'])) {
185 + wp_send_json_error('No message received');
2535 186 wp_die();
2536 -}
2537 -
2538 -/**
2539 - * Get bot-specific options for multi-bot functionality
2540 - * Falls back to default options if bot_id is 'default' or multi-bot add-on is not active
2541 - */
2542 -// Also debug the bot options retrieval
2543 -private function get_bot_options($bot_id = 'default') {
2544 - //error_log("MXCHAT DEBUG: get_bot_options called for bot: " . $bot_id);
2545 -
2546 - // The admin Testing tab renders the real widget as bot_id "testing", which
2547 - // is not a registered multi-bot. It must resolve the DEFAULT bot's config
2548 - // so the Testing chat behaves exactly like the front-end (same precedent
2549 - // as the Actions enabled_bots check).
2550 - if ($bot_id === 'testing') {
2551 - $bot_id = 'default';
2552 187 }
2553 188
2554 - if ($bot_id === 'default' || !class_exists('MxChat_Multi_Bot_Manager')) {
2555 - //error_log("MXCHAT DEBUG: Using default options (no multi-bot or bot is 'default')");
2556 - return array();
2557 - }
2558 -
2559 - $bot_options = apply_filters('mxchat_get_bot_options', array(), $bot_id);
2560 -
2561 - if (!empty($bot_options)) {
2562 - //error_log("MXCHAT DEBUG: Got bot-specific options from filter");
2563 - if (isset($bot_options['similarity_threshold'])) {
2564 - //error_log(" - similarity_threshold: " . $bot_options['similarity_threshold']);
2565 - }
2566 - }
2567 -
2568 - return is_array($bot_options) ? $bot_options : array();
2569 -}
2570 -
2571 -/**
2572 - * Get bot-specific Pinecone configuration
2573 - * Used in the knowledge retrieval functions
2574 - */
2575 -// Also add debugging to your get_bot_pinecone_config function
2576 -private function get_bot_pinecone_config($bot_id = 'default') {
2577 - //error_log("MXCHAT DEBUG: get_bot_pinecone_config called for bot: " . $bot_id);
2578 -
2579 - // Admin Testing tab bot → resolve the DEFAULT bot's backend. Without this,
2580 - // on a multi-bot + Pinecone site the filter below gets an unknown bot id
2581 - // with an EMPTY default, returns array(), and the dispatcher silently
2582 - // searches the WordPress DB while the front-end searches Pinecone — the
2583 - // Testing panel then reports similarity results from a different KB.
2584 - if ($bot_id === 'testing') {
2585 - $bot_id = 'default';
2586 - }
2587 -
2588 - // If default bot or multi-bot add-on not active, use default Pinecone config
2589 - if ($bot_id === 'default' || !class_exists('MxChat_Multi_Bot_Manager')) {
2590 - //error_log("MXCHAT DEBUG: Using default Pinecone config (no multi-bot or bot is 'default')");
2591 - $addon_options = get_option('mxchat_pinecone_addon_options', array());
2592 - $config = array(
2593 - 'use_pinecone' => (isset($addon_options['mxchat_use_pinecone']) && $addon_options['mxchat_use_pinecone'] === '1'),
2594 - 'api_key' => $addon_options['mxchat_pinecone_api_key'] ?? '',
2595 - 'host' => $addon_options['mxchat_pinecone_host'] ?? '',
2596 - 'namespace' => $addon_options['mxchat_pinecone_namespace'] ?? ''
2597 - );
2598 - //error_log("MXCHAT DEBUG: Default config - use_pinecone: " . ($config['use_pinecone'] ? 'true' : 'false'));
2599 - return $config;
2600 - }
2601 -
2602 - //error_log("MXCHAT DEBUG: Calling filter 'mxchat_get_bot_pinecone_config' for bot: " . $bot_id);
2603 -
2604 - // Hook for multi-bot add-on to provide bot-specific Pinecone config
2605 - $bot_pinecone_config = apply_filters('mxchat_get_bot_pinecone_config', array(), $bot_id);
2606 -
2607 - if (!empty($bot_pinecone_config)) {
2608 - //error_log("MXCHAT DEBUG: Got bot-specific config from filter");
2609 - //error_log(" - use_pinecone: " . (isset($bot_pinecone_config['use_pinecone']) ? ($bot_pinecone_config['use_pinecone'] ? 'true' : 'false') : 'not set'));
2610 - //error_log(" - host: " . ($bot_pinecone_config['host'] ?? 'not set'));
2611 - //error_log(" - namespace: " . ($bot_pinecone_config['namespace'] ?? 'not set'));
2612 - } else {
2613 - //error_log("MXCHAT DEBUG: Filter returned empty config!");
2614 - }
2615 -
2616 - return is_array($bot_pinecone_config) ? $bot_pinecone_config : array();
2617 -}
2618 -
2619 -
2620 -// Updated function to check intents and invoke the callback function
2621 -private function mxchat_check_intent_and_invoke_callback($message, $user_id, $session_id) {
2622 - global $wpdb;
2623 - $chat_mode = get_option("mxchat_mode_{$session_id}", 'ai');
2624 -
2625 - // Get the current bot_id
2626 - $current_bot_id = $this->get_current_bot_id($session_id);
2627 -
2628 - // Generate the user embedding
2629 - $user_embedding = $this->mxchat_generate_embedding($message, $this->options['api_key']);
2630 -
2631 - // Check if embedding generation returned an error
2632 - if (is_array($user_embedding) && isset($user_embedding['error'])) {
2633 - $error_message = $user_embedding['error'];
2634 - $error_code = $user_embedding['error_code'] ?? 'embedding_error';
2635 -
2636 - // FIXED: Send error in appropriate format based on streaming mode
2637 - if ($this->is_streaming) {
2638 - echo "data: " . json_encode([
2639 - 'error' => true,
2640 - 'error_message' => $error_message,
2641 - 'error_code' => $error_code,
2642 - 'text' => $error_message,
2643 - 'message' => $error_message
2644 - ]) . "\n\n";
2645 - echo "data: [DONE]\n\n";
2646 - flush();
2647 - } else {
2648 - wp_send_json_error([
2649 - 'error_message' => $error_message,
2650 - 'error_code' => $error_code
2651 - ]);
2652 - }
189 + $message = sanitize_text_field($_POST['message']);
190 + if (empty($message)) {
191 + wp_send_json_error('Message is empty or invalid.');
2653 192 wp_die();
2654 193 }
2655 194
2656 - // Check if embedding is valid
2657 - if (!is_array($user_embedding) || empty($user_embedding)) {
2658 - $error_message = esc_html__('Unable to process your message. The embedding service is not responding correctly.', 'mxchat');
195 + // Initialize the variable with the original message
196 + $message_with_order_details = $message;
2659 197
2660 - // FIXED: Send error in appropriate format based on streaming mode
2661 - if ($this->is_streaming) {
2662 - echo "data: " . json_encode([
2663 - 'error' => true,
2664 - 'error_message' => $error_message,
2665 - 'error_code' => 'invalid_embedding',
2666 - 'text' => $error_message,
2667 - 'message' => $error_message
2668 - ]) . "\n\n";
2669 - echo "data: [DONE]\n\n";
2670 - flush();
2671 - } else {
2672 - wp_send_json_error([
2673 - 'error_message' => $error_message,
2674 - 'error_code' => 'invalid_embedding'
2675 - ]);
2676 - }
2677 - wp_die();
2678 - }
198 + // Check if the user asked about orders
199 + if (MxChat_WooCommerce::mxchat_is_order_related_query($message)) {
200 + $order_details = MxChat_WooCommerce::mxchat_fetch_user_orders_details();
2679 201
2680 - // Fetch intents from the database
2681 - $table_name = $wpdb->prefix . 'mxchat_intents';
2682 - if ($chat_mode === 'agent') {
2683 - $query = $wpdb->prepare(
2684 - "SELECT * FROM $table_name WHERE callback_function = %s AND (enabled = 1 OR enabled IS NULL)",
2685 - 'mxchat_handle_switch_to_chatbot_intent'
2686 - );
2687 - $intents = $wpdb->get_results($query);
2688 - } else {
2689 - $intents = $wpdb->get_results("SELECT * FROM $table_name WHERE enabled = 1 OR enabled IS NULL");
2690 - }
2691 -
2692 - if (empty($intents)) {
2693 - return false;
2694 - }
2695 -
2696 - // Prefetch individual phrase vectors from wp_mxchat_intent_phrases (grouped by intent_id)
2697 - $phrases_table = $wpdb->prefix . 'mxchat_intent_phrases';
2698 - $phrases_by_intent = [];
2699 - if ($wpdb->get_var("SHOW TABLES LIKE '$phrases_table'") === $phrases_table) {
2700 - $all_phrases = $wpdb->get_results("SELECT intent_id, phrase, embedding_vector FROM $phrases_table");
2701 - foreach ($all_phrases as $p) {
2702 - $phrases_by_intent[$p->intent_id][] = $p;
2703 - }
2704 - }
2705 -
2706 - $highest_similarity = -INF;
2707 - $matched_intent = null;
2708 -
2709 - // Array to store action analysis for testing panel
2710 - $action_analysis = [];
2711 -
2712 - foreach ($intents as $intent) {
2713 - // Additional check for enabled state
2714 - $is_enabled = isset($intent->enabled) ? (bool)$intent->enabled : true;
2715 - if (!$is_enabled) {
2716 - continue;
2717 - }
2718 -
2719 - // Check if this action is enabled for the current bot
2720 - if (!$this->is_action_enabled_for_bot($intent, $current_bot_id)) {
2721 - continue;
2722 - }
2723 -
2724 - $best_similarity = -INF;
2725 - $matched_phrase_text = '';
2726 -
2727 - // Check legacy embedding vector (existing behavior)
2728 - $intent_embedding_serialized = $intent->embedding_vector;
2729 - $intent_embedding = $intent_embedding_serialized
2730 - ? unserialize($intent_embedding_serialized, ['allowed_classes' => false])
2731 - : null;
2732 -
2733 - if (is_array($intent_embedding) && !empty($intent_embedding)) {
2734 - $legacy_similarity = $this->mxchat_calculate_cosine_similarity($user_embedding, $intent_embedding);
2735 - if ($legacy_similarity > $best_similarity) {
2736 - $best_similarity = $legacy_similarity;
2737 - $matched_phrase_text = 'legacy';
202 + // If order details are available, append them to the user's message
203 + if (!empty($order_details)) {
204 + $message_with_order_details = $message . "\n\n" . $order_details;
2738 205 }
2739 206 }
2740 207
2741 - // Check individual phrase vectors
2742 - if (isset($phrases_by_intent[$intent->id])) {
2743 - foreach ($phrases_by_intent[$intent->id] as $phrase_row) {
2744 - $phrase_embedding = $phrase_row->embedding_vector
2745 - ? unserialize($phrase_row->embedding_vector, ['allowed_classes' => false])
2746 - : null;
2747 - if (!is_array($phrase_embedding)) {
2748 - continue;
2749 - }
2750 - $phrase_similarity = $this->mxchat_calculate_cosine_similarity($user_embedding, $phrase_embedding);
2751 - if ($phrase_similarity > $best_similarity) {
2752 - $best_similarity = $phrase_similarity;
2753 - $matched_phrase_text = $phrase_row->phrase;
2754 - }
2755 - }
2756 - }
2757 208
2758 - // Skip if no valid embedding was found at all
2759 - if ($best_similarity === -INF) {
2760 - continue;
2761 - }
209 + // Save the combined message to the database
210 + $this->mxchat_save_chat_message($session_id, 'user', $message_with_order_details);
2762 211
2763 - $similarity = $best_similarity;
2764 - $intent_threshold = isset($intent->similarity_threshold) ? $intent->similarity_threshold : 0.85;
2765 -
2766 - // Store action analysis data for testing panel
2767 - $action_analysis[] = [
2768 - 'intent_label' => $intent->intent_label,
2769 - 'callback_function' => $intent->callback_function,
2770 - 'similarity' => round($similarity, 4),
2771 - 'similarity_percentage' => round($similarity * 100, 2),
2772 - 'threshold' => $intent_threshold,
2773 - 'threshold_percentage' => round($intent_threshold * 100, 2),
2774 - 'above_threshold' => $similarity >= $intent_threshold,
2775 - 'matched_phrase' => $matched_phrase_text,
2776 - 'triggered' => false // Will be updated below if this intent is triggered
2777 - ];
2778 -
2779 - if ($similarity >= $intent_threshold && $similarity > $highest_similarity) {
2780 - $highest_similarity = $similarity;
2781 - $matched_intent = $intent;
2782 - }
2783 - }
2784 -
2785 - // Mark the triggered action if any
2786 - if ($matched_intent) {
2787 - foreach ($action_analysis as &$action) {
2788 - if ($action['intent_label'] === $matched_intent->intent_label) {
2789 - $action['triggered'] = true;
2790 - break;
2791 - }
2792 - }
2793 - }
2794 -
2795 - // Sort actions by similarity (highest first) and store for testing panel
2796 - usort($action_analysis, function($a, $b) {
2797 - return $b['similarity'] <=> $a['similarity'];
2798 - });
2799 -
2800 - // Store action analysis for testing panel capture
2801 - $this->last_action_analysis = $action_analysis;
2802 -
2803 - // Around line 715 in your mxchat_check_intent_and_invoke_callback function
2804 - if ($matched_intent) {
2805 - // If the callback is a method on this instance (core callback), call it directly
2806 - if (method_exists($this, $matched_intent->callback_function)) {
2807 - $callback_result = call_user_func(
2808 - [$this, $matched_intent->callback_function],
2809 - $message,
2810 - $user_id,
2811 - $session_id,
2812 - $matched_intent,
2813 - $user_context ?? null
2814 - );
2815 - } else {
2816 - // Otherwise, use apply_filters for add-on callbacks
2817 - $callback_result = apply_filters(
2818 - $matched_intent->callback_function,
2819 - false,
2820 - $message,
2821 - $user_id,
2822 - $session_id,
2823 - $matched_intent
2824 - );
2825 - }
2826 -
2827 - // Handle the callback result properly
2828 - if ($callback_result !== false) {
2829 - // If callback returned an array with chat_mode, use it directly
2830 - if (is_array($callback_result) && isset($callback_result['chat_mode'])) {
2831 - $this->fallbackResponse = $callback_result;
2832 - return $callback_result; // Return the full array
2833 - } else {
2834 - $this->fallbackResponse = $callback_result;
2835 - return true;
2836 - }
2837 - }
2838 - }
2839 -
2840 - return false;
2841 -}
2842 -
2843 -/**
2844 - * Check if an action is enabled for a specific bot
2845 - */
2846 -private function is_action_enabled_for_bot($intent, $bot_id) {
2847 - // If enabled_bots column doesn't exist or is null, assume it's enabled for all bots (backward compatibility)
2848 - if (!isset($intent->enabled_bots) || empty($intent->enabled_bots)) {
2849 - return true;
2850 - }
2851 -
2852 - $enabled_bots = json_decode($intent->enabled_bots, true);
2853 -
2854 - // If JSON decode fails or returns empty array, assume enabled for all (backward compatibility)
2855 - if (!is_array($enabled_bots) || empty($enabled_bots)) {
2856 - return true;
2857 - }
2858 -
2859 - // Admin testing tab uses bot_id "testing" — treat it as "default" so all
2860 - // default-bot actions are testable from the admin panel
2861 - if ($bot_id === 'testing') {
2862 - $bot_id = 'default';
2863 - }
2864 -
2865 - // Check if the current bot is in the enabled bots list
2866 - return in_array($bot_id, $enabled_bots);
2867 -}
2868 -
2869 -// Helper function to clear PDF and Word document related transients
2870 -private function clear_pdf_transients($session_id) {
2871 - // PDF transients
2872 - delete_transient('mxchat_pdf_url_' . $session_id);
2873 - delete_transient('mxchat_pdf_embeddings_' . $session_id);
2874 - delete_transient('mxchat_include_pdf_in_context_' . $session_id);
2875 - delete_transient('mxchat_waiting_for_pdf_url_' . $session_id);
2876 -
2877 - // Word document transients
2878 - delete_transient('mxchat_word_url_' . $session_id);
2879 - delete_transient('mxchat_word_filename_' . $session_id);
2880 - delete_transient('mxchat_word_embeddings_' . $session_id);
2881 - delete_transient('mxchat_include_word_in_context_' . $session_id);
2882 - delete_transient('mxchat_waiting_for_word_' . $session_id);
2883 -}
2884 -
2885 -
2886 -
2887 -//verified good
2888 -public function mxchat_handle_email_capture($message, $user_id, $session_id) {
2889 - // Get the user's original instruction/message
2890 - $user_instruction = esc_html($this->options['triggered_phrase_response'] ?? esc_html__("Please provide your email address.", 'mxchat'));
2891 -
2892 - // Set instruction for AI - just pass along what the user wanted to say
2893 - $this->current_action_instruction = $user_instruction;
2894 -
2895 - // Set the transient to track email capture flow
2896 - set_transient('mxchat_email_capture_' . $user_id, true, 5 * MINUTE_IN_SECONDS);
2897 -
2898 - // Return false to let the AI generate the response
2899 - return false;
2900 -}
2901 -
2902 -public function mxchat_generate_image($message, $user_id, $session_id) {
2903 - //error_log("Starting image generation for message: " . $message);
2904 -
2905 - // Prepare a prompt for OpenAI image generation
2906 - $prompt = esc_html__('Create an image of ', 'mxchat') . sanitize_text_field($message);
2907 -
2908 - // Opt-in routing: when 'custom_provider_for_images' is on, route image gen
2909 - // through the configured Custom (OpenAI-compatible) /images/generations route.
2910 - if (!empty($this->options['custom_provider_for_images']) && $this->options['custom_provider_for_images'] === 'on') {
2911 - $image_response = $this->mxchat_generate_custom_image($prompt);
2912 - } else {
2913 - // Use the existing OpenAI API key
2914 - $openai_api_key = sanitize_text_field($this->options['api_key']);
2915 - // Call OpenAI GPT Image to generate an image
2916 - $image_response = $this->mxchat_generate_openai_image($prompt, $openai_api_key);
2917 - }
2918 -
2919 - // Check if the response contains an image URL
2920 - if (isset($image_response['imageUrl'])) {
2921 - $image_url = esc_url_raw($image_response['imageUrl']);
2922 -
2923 - // Construct the HTML with a CSS class instead of inline styles
2924 - $response_html = '<img src="' . esc_url($image_url) . '" alt="' . esc_attr__('Generated Image', 'mxchat') . '" class="mxchat-generated-image" />';
2925 - $response_text = esc_html__('Here is the image I generated:', 'mxchat');
2926 -
2927 - // Save the bot message with both text and HTML
2928 - $this->mxchat_save_chat_message($session_id, 'bot', $response_text);
2929 - $this->mxchat_save_chat_message($session_id, 'bot', $response_html);
2930 -
2931 - // Set the fallback response for the chat handler
2932 - $this->fallbackResponse = [
2933 - 'text' => $response_text,
2934 - 'html' => $response_html,
2935 - 'images' => [$image_url]
2936 - ];
2937 -
2938 - // For debugging/verification - Use json_encode to verify what's being set
2939 - //error_log("Image generation successful - fallbackResponse set: " . json_encode($this->fallbackResponse));
2940 -
2941 - // Return the response directly instead of relying on the property
2942 - return $this->fallbackResponse;
2943 - } else {
2944 - $response_text = esc_html__("I'm sorry, but I couldn't generate an image based on your request.", 'mxchat');
2945 -
2946 - // Save the error message
2947 - $this->mxchat_save_chat_message($session_id, 'bot', $response_text);
2948 -
2949 - // Set the fallback response for the chat handler
2950 - $this->fallbackResponse = [
2951 - 'text' => $response_text,
2952 - 'html' => '',
2953 - 'images' => []
2954 - ];
2955 -
2956 - //error_log("DALL-E image generation error: " . esc_html($image_response['error'] ?? 'Unknown error.'));
2957 - //error_log("Error fallbackResponse set: " . json_encode($this->fallbackResponse));
2958 -
2959 - // Return the response directly instead of relying on the property
2960 - return $this->fallbackResponse;
2961 - }
2962 -}
2963 -
2964 -public function mxchat_generate_gemini_image($message, $user_id, $session_id) {
2965 - $prompt = esc_html__('Create an image of ', 'mxchat') . sanitize_text_field($message);
2966 -
2967 - $gemini_api_key = sanitize_text_field($this->options['gemini_api_key'] ?? '');
2968 - if (empty($gemini_api_key)) {
2969 - $response_text = esc_html__("Gemini API key is not configured.", 'mxchat');
2970 - $this->mxchat_save_chat_message($session_id, 'bot', $response_text);
2971 - return ['text' => $response_text, 'html' => '', 'images' => []];
2972 - }
2973 -
2974 - $image_response = $this->mxchat_generate_imagen_image($prompt, $gemini_api_key);
2975 -
2976 - if (isset($image_response['imageUrl'])) {
2977 - $image_url = esc_url_raw($image_response['imageUrl']);
2978 -
2979 - $response_html = '<img src="' . esc_url($image_url) . '" alt="' . esc_attr__('Generated Image', 'mxchat') . '" class="mxchat-generated-image" />';
2980 - $response_text = esc_html__('Here is the image I generated:', 'mxchat');
2981 -
2982 - $this->mxchat_save_chat_message($session_id, 'bot', $response_text);
2983 - $this->mxchat_save_chat_message($session_id, 'bot', $response_html);
2984 -
2985 - $this->fallbackResponse = [
2986 - 'text' => $response_text,
2987 - 'html' => $response_html,
2988 - 'images' => [$image_url]
2989 - ];
2990 -
2991 - return $this->fallbackResponse;
2992 - } else {
2993 - $response_text = esc_html__("I'm sorry, but I couldn't generate an image based on your request.", 'mxchat');
2994 -
2995 - $this->mxchat_save_chat_message($session_id, 'bot', $response_text);
2996 -
2997 - $this->fallbackResponse = [
2998 - 'text' => $response_text,
2999 - 'html' => '',
3000 - 'images' => []
3001 - ];
3002 -
3003 - return $this->fallbackResponse;
3004 - }
3005 -}
3006 -
3007 -private function mxchat_save_generated_image($base64_data, $mime_type = 'image/png', $prefix = 'mxchat-generated') {
3008 - // Map the real mime type to a matching file extension so the saved file's
3009 - // extension always agrees with its bytes. A mismatch (e.g. Imagen returning
3010 - // webp bytes that were written into a ".png" file) makes the browser refuse
3011 - // to render the image even though the file saved successfully and the bot
3012 - // reported success — that was the Gemini/Imagen "image never renders" bug.
3013 - // OpenAI + custom-provider paths pass 'image/png' explicitly, so they are
3014 - // unaffected; this only matters for providers that return another type.
3015 - $mime_to_ext = [
3016 - 'image/jpeg' => 'jpg',
3017 - 'image/jpg' => 'jpg',
3018 - 'image/png' => 'png',
3019 - 'image/webp' => 'webp',
3020 - 'image/gif' => 'gif',
3021 - ];
3022 - $mime_type = strtolower(trim((string) $mime_type));
3023 - if (isset($mime_to_ext[$mime_type])) {
3024 - $extension = $mime_to_ext[$mime_type];
3025 - } else {
3026 - // Unknown/unsupported type: fall back to png and normalize the stored
3027 - // mime so the attachment record and the file extension stay consistent.
3028 - $extension = 'png';
3029 - $mime_type = 'image/png';
3030 - }
3031 - $filename = sanitize_file_name($prefix . '-' . wp_generate_uuid4() . '.' . $extension);
3032 - $decoded = base64_decode($base64_data);
3033 -
3034 - if ($decoded === false) {
3035 - return new \WP_Error('decode_failed', esc_html__('Failed to decode image data.', 'mxchat'));
3036 - }
3037 -
3038 - $upload = wp_upload_bits($filename, null, $decoded);
3039 -
3040 - if (!empty($upload['error'])) {
3041 - return new \WP_Error('upload_failed', $upload['error']);
3042 - }
3043 -
3044 - $attach_id = wp_insert_attachment([
3045 - 'post_mime_type' => $mime_type,
3046 - 'post_title' => $prefix,
3047 - 'post_content' => '',
3048 - 'post_status' => 'inherit',
3049 - ], $upload['file']);
3050 -
3051 - if (is_wp_error($attach_id)) {
3052 - return $attach_id;
3053 - }
3054 -
3055 - require_once ABSPATH . 'wp-admin/includes/image.php';
3056 - $metadata = wp_generate_attachment_metadata($attach_id, $upload['file']);
3057 - wp_update_attachment_metadata($attach_id, $metadata);
3058 -
3059 - return esc_url_raw(wp_get_attachment_url($attach_id));
3060 -}
3061 -
3062 -private function mxchat_generate_openai_image($prompt, $api_key, $model = 'gpt-image-1', $timeout = 60) {
3063 - $api_url = 'https://api.openai.com/v1/images/generations';
3064 - $body = json_encode([
3065 - 'prompt' => sanitize_text_field($prompt),
3066 - 'n' => 1,
3067 - 'size' => '1024x1024',
3068 - 'quality' => 'medium',
3069 - 'output_format' => 'png',
3070 - 'model' => sanitize_text_field($model),
3071 - ]);
3072 -
3073 - $args = [
3074 - 'body' => $body,
3075 - 'headers' => [
3076 - 'Content-Type' => 'application/json',
3077 - 'Authorization' => 'Bearer ' . sanitize_text_field($api_key),
3078 - ],
3079 - 'method' => 'POST',
3080 - 'timeout' => absint($timeout),
3081 - ];
3082 -
3083 - $response = wp_remote_post($api_url, $args);
3084 -
3085 - if (is_wp_error($response)) {
3086 - return ['error' => esc_html__('Error generating image: ', 'mxchat') . $response->get_error_message()];
3087 - }
3088 -
3089 - $response_body = json_decode(wp_remote_retrieve_body($response), true);
3090 -
3091 - $b64 = $response_body['data'][0]['b64_json'] ?? $response_body['data'][0]['b64'] ?? null;
3092 - if ($b64) {
3093 - $saved_url = $this->mxchat_save_generated_image($b64, 'image/png', 'mxchat-openai');
3094 - if (is_wp_error($saved_url)) {
3095 - return ['error' => $saved_url->get_error_message()];
3096 - }
3097 - return ['imageUrl' => $saved_url];
3098 - } else {
3099 - return ['error' => esc_html__('Failed to generate image.', 'mxchat')];
3100 - }
3101 -}
3102 -
3103 -/**
3104 - * Generate an image via a Custom (OpenAI-compatible) provider's /images/generations route.
3105 - * Only called when the opt-in 'custom_provider_for_images' setting is on.
3106 - */
3107 -private function mxchat_generate_custom_image($prompt, $timeout = 90) {
3108 - $cfg = $this->mxchat_resolve_custom_provider();
3109 - if (empty($cfg['base_url'])) {
3110 - return ['error' => esc_html__('Custom provider Base URL is not configured.', 'mxchat')];
3111 - }
3112 - $url = $cfg['base_url'] . '/images/generations';
3113 - if (!empty($cfg['api_version'])) {
3114 - $url .= (strpos($url, '?') === false ? '?' : '&') . 'api-version=' . rawurlencode($cfg['api_version']);
3115 - }
3116 - $body = wp_json_encode([
3117 - 'prompt' => sanitize_text_field($prompt),
3118 - 'n' => 1,
3119 - 'size' => '1024x1024',
3120 - 'model' => $cfg['model'],
3121 - ]);
3122 - $response = wp_remote_post($url, [
3123 - 'headers' => $this->mxchat_custom_provider_assoc_headers($cfg),
3124 - 'body' => $body,
3125 - 'method' => 'POST',
3126 - 'timeout' => absint($timeout),
3127 - ]);
3128 - if (is_wp_error($response)) {
3129 - return ['error' => esc_html__('Error generating image (custom provider): ', 'mxchat') . $response->get_error_message()];
3130 - }
3131 - $resp = json_decode(wp_remote_retrieve_body($response), true);
3132 - // Try b64 first (matches OpenAI shape), then url-based fallback.
3133 - $b64 = $resp['data'][0]['b64_json'] ?? $resp['data'][0]['b64'] ?? null;
3134 - if ($b64) {
3135 - $saved = $this->mxchat_save_generated_image($b64, 'image/png', 'mxchat-custom');
3136 - if (is_wp_error($saved)) {
3137 - return ['error' => $saved->get_error_message()];
3138 - }
3139 - return ['imageUrl' => $saved];
3140 - }
3141 - $remote_url = $resp['data'][0]['url'] ?? null;
3142 - if ($remote_url) {
3143 - return ['imageUrl' => esc_url_raw($remote_url)];
3144 - }
3145 - $err_msg = $this->extract_provider_error($resp, esc_html__('Custom provider did not return an image.', 'mxchat'));
3146 - return ['error' => esc_html($err_msg)];
3147 -}
3148 -
3149 -private function mxchat_generate_imagen_image($prompt, $api_key, $timeout = 60) {
3150 - $api_url = 'https://generativelanguage.googleapis.com/v1beta/models/imagen-4.0-generate-001:predict';
3151 -
3152 - $body = json_encode([
3153 - 'instances' => [['prompt' => sanitize_text_field($prompt)]],
3154 - 'parameters' => [
3155 - 'sampleCount' => 1,
3156 - 'aspectRatio' => '1:1',
3157 - ],
3158 - ]);
3159 -
3160 - $args = [
3161 - 'body' => $body,
3162 - 'headers' => [
3163 - 'Content-Type' => 'application/json',
3164 - 'x-goog-api-key' => sanitize_text_field($api_key),
3165 - ],
3166 - 'method' => 'POST',
3167 - 'timeout' => absint($timeout),
3168 - ];
3169 -
3170 - $response = wp_remote_post($api_url, $args);
3171 -
3172 - if (is_wp_error($response)) {
3173 - return ['error' => esc_html__('Error generating image: ', 'mxchat') . $response->get_error_message()];
3174 - }
3175 -
3176 - $response_body = json_decode(wp_remote_retrieve_body($response), true);
3177 -
3178 - $b64 = $response_body['predictions'][0]['bytesBase64Encoded'] ?? $response_body['predictions'][0]['imageBytes'] ?? null;
3179 - if ($b64) {
3180 - $mime = $response_body['predictions'][0]['mimeType'] ?? 'image/png';
3181 - $saved_url = $this->mxchat_save_generated_image($b64, $mime, 'mxchat-gemini');
3182 - if (is_wp_error($saved_url)) {
3183 - return ['error' => $saved_url->get_error_message()];
3184 - }
3185 - return ['imageUrl' => $saved_url];
3186 - } else {
3187 - return ['error' => esc_html__('Failed to generate image.', 'mxchat')];
3188 - }
3189 -}
3190 -
3191 -/**
3192 - * Handle web search requests.
3193 - *
3194 - * Sends the refined search query to the Brave Search API and uses the
3195 - * results to generate a conversational response with the AI model.
3196 - *
3197 - * @since 1.0.0
3198 - * @param string $message The user's search query.
3199 - * @param string $user_id The user identifier.
3200 - * @param string $session_id The current session ID.
3201 - * @return array Response array containing text with embedded HTML links
3202 - */
3203 -public function mxchat_handle_search_request($message, $user_id, $session_id) {
3204 - // Step 1: Interpret and refine the search query
3205 - $refined_search_query = $this->mxchat_interpret_search_query($message);
3206 - if (empty($refined_search_query)) {
3207 - return array(
3208 - 'text' => esc_html__('I apologize, but could you please rephrase your search request?', 'mxchat'),
3209 - 'html' => ''
3210 - );
3211 - }
3212 -
3213 - // Retrieve and validate API settings
3214 - $options = get_option('mxchat_options');
3215 - $api_key = isset($options['brave_api_key']) ? sanitize_text_field($options['brave_api_key']) : '';
3216 - $results_count = isset($options['brave_results_count']) ? absint($options['brave_results_count']) : 5;
3217 -
3218 - if (empty($api_key)) {
3219 - return array(
3220 - 'text' => esc_html__('Search functionality is temporarily unavailable. Please try again later.', 'mxchat'),
3221 - 'html' => ''
3222 - );
3223 - }
3224 -
3225 - // Build the API request URL
3226 - $api_url = add_query_arg(
3227 - array(
3228 - 'q' => rawurlencode($refined_search_query),
3229 - 'count' => $results_count,
3230 - 'text_decorations' => 'true',
3231 - 'rich_data' => 'true',
3232 - ),
3233 - 'https://api.search.brave.com/res/v1/web/search'
3234 - );
3235 -
3236 - // Attempt to retrieve cached results first
3237 - $transient_key = 'mxchat_search_' . md5($refined_search_query);
3238 - $results = get_transient($transient_key);
3239 -
3240 - if (false === $results) {
3241 - // SECURITY FIX: Changed to wp_safe_remote_get
3242 - $response = wp_safe_remote_get(
3243 - $api_url,
3244 - array(
3245 - 'headers' => array(
3246 - 'Accept' => 'application/json',
3247 - 'Accept-Encoding' => 'gzip',
3248 - 'X-Subscription-Token'=> $api_key,
3249 - ),
3250 - 'timeout' => 10,
3251 - )
3252 - );
3253 -
3254 - if (is_wp_error($response)) {
3255 - return array(
3256 - 'text' => esc_html__('I encountered an error while searching. Please try again.', 'mxchat'),
3257 - 'html' => ''
3258 - );
3259 - }
3260 -
3261 - $results = json_decode(wp_remote_retrieve_body($response), true);
3262 -
3263 - if (json_last_error() !== JSON_ERROR_NONE) {
3264 - return array(
3265 - 'text' => esc_html__('I received an invalid response from the search service.', 'mxchat'),
3266 - 'html' => ''
3267 - );
3268 - }
3269 -
3270 - // Cache results for one hour
3271 - set_transient($transient_key, $results, HOUR_IN_SECONDS);
3272 - }
3273 -
3274 - // Process results
3275 - if (!empty($results['web']['results']) && is_array($results['web']['results'])) {
3276 - // Create a more straightforward summary with HTML links
3277 - $search_results_text = '';
3278 -
3279 - // Add a simple intro
3280 - $search_results_text .= sprintf(
3281 - esc_html__("Here's what I found about '%s':", 'mxchat'),
3282 - esc_html($refined_search_query)
3283 - );
3284 -
3285 - // Add the top results with HTML links
3286 - foreach (array_slice($results['web']['results'], 0, 5) as $result) {
3287 - $title = isset($result['title']) ? wp_strip_all_tags($result['title']) : '';
3288 - $url = isset($result['url']) ? esc_url($result['url']) : '';
3289 - $description = isset($result['description']) ? wp_strip_all_tags($result['description']) : '';
3290 -
3291 - // Add a line break after the intro
3292 - $search_results_text .= '<br><br>';
3293 -
3294 - // Add title as a link
3295 - $search_results_text .= sprintf(
3296 - '<a href="%s" target="_blank" rel="noopener noreferrer">%s</a><br>',
3297 - $url,
3298 - $title
3299 - );
3300 -
3301 - // Add a condensed description
3302 - $search_results_text .= sprintf("%s", $description);
3303 - }
3304 -
3305 - // Save to chat history
3306 - $this->mxchat_save_chat_message($session_id, 'bot', $search_results_text);
3307 -
3308 - // Return the formatted text with embedded HTML links
3309 - return array(
3310 - 'text' => $search_results_text,
3311 - 'html' => ''
3312 - );
3313 - } else {
3314 - return array(
3315 - 'text' => sprintf(
3316 - esc_html__('I searched for "%s" but couldn\'t find any relevant results. Would you like to try different search terms?', 'mxchat'),
3317 - esc_html($refined_search_query)
3318 - ),
3319 - 'html' => ''
3320 - );
3321 - }
3322 -}
3323 -
3324 -//very good
3325 -/**
3326 - * Handle image search requests from the chatbot
3327 - *
3328 - * @param string $message The user's search query
3329 - * @param int $user_id The user's ID
3330 - * @param string $session_id The chat session ID
3331 - * @return array Response array with text and HTML content
3332 - */
3333 -public function mxchat_handle_image_search_request($message, $user_id, $session_id) {
3334 - // Step 1: Interpret the search query using the user's selected AI model
3335 - $refined_search_query = $this->mxchat_interpret_search_query($message);
3336 -
3337 - // If no query was interpreted, return a fallback message
3338 - if (empty($refined_search_query)) {
3339 - return array(
3340 - 'text' => __("I'm sorry, I couldn't interpret your search query. Please specify what you'd like to see images of.", 'mxchat'),
3341 - 'html' => "",
3342 - );
3343 - }
3344 -
3345 - // Brave API URL
3346 - $api_url = 'https://api.search.brave.com/res/v1/images/search';
3347 -
3348 - // Retrieve Brave API settings
3349 - $options = get_option('mxchat_options');
3350 - $api_key = isset($options['brave_api_key']) ? sanitize_text_field($options['brave_api_key']) : '';
3351 -
3352 - if (empty($api_key)) {
3353 - return array(
3354 - 'text' => __("API key is not configured. Please set it in the Brave Search Settings.", 'mxchat'),
3355 - 'html' => "",
3356 - );
3357 - }
3358 -
3359 - $image_count = isset($options['brave_image_count']) ? intval($options['brave_image_count']) : 4;
3360 - $safe_search = isset($options['brave_safe_search']) ? sanitize_text_field($options['brave_safe_search']) : 'strict';
3361 -
3362 - // Append query parameters based on settings
3363 - $api_url = add_query_arg([
3364 - 'q' => rawurlencode($refined_search_query),
3365 - 'count' => $image_count,
3366 - 'safesearch' => $safe_search,
3367 - ], $api_url);
3368 -
3369 - // Implement caching
3370 - $transient_key = 'mxchat_image_search_' . md5($refined_search_query);
3371 - $body = get_transient($transient_key);
3372 -
3373 - if (false === $body) {
3374 - $args = [
3375 - 'headers' => [
3376 - 'Accept' => 'application/json',
3377 - 'Accept-Encoding' => 'gzip',
3378 - 'X-Subscription-Token' => $api_key,
3379 - ],
3380 - 'timeout' => 10,
3381 - ];
3382 -
3383 - // SECURITY FIX: Changed to wp_safe_remote_get
3384 - $response = wp_safe_remote_get($api_url, $args);
3385 -
3386 - if (is_wp_error($response)) {
3387 - return array(
3388 - 'text' => __("I'm sorry, I couldn't retrieve any images based on your request.", 'mxchat'),
3389 - 'html' => "",
3390 - );
3391 - }
3392 -
3393 - $body = json_decode(wp_remote_retrieve_body($response), true);
3394 - set_transient($transient_key, $body, HOUR_IN_SECONDS);
3395 - }
3396 -
3397 - // Process the API response
3398 - if (isset($body['results']) && is_array($body['results']) && count($body['results']) > 0) {
3399 - $html_output = '<div class="mxchat-image-gallery">';
3400 -
3401 - // Get the configured image count (1-6)
3402 - $display_count = isset($options['brave_image_count']) ? intval($options['brave_image_count']) : 4;
3403 - $display_count = min($display_count, count($body['results'])); // Make sure we don't exceed available images
3404 -
3405 - // Use only the requested number of images
3406 - for ($i = 0; $i < $display_count; $i++) {
3407 - $image = $body['results'][$i];
3408 - $image_url = isset($image['url']) ? esc_url($image['url']) : '';
3409 - $thumbnail_url = isset($image['thumbnail']['src']) ? esc_url($image['thumbnail']['src']) : '';
3410 - $title = isset($image['title']) ? esc_html($image['title']) : esc_html__('Image', 'mxchat');
3411 -
3412 - if ($image_url && $thumbnail_url) {
3413 - $html_output .= '<div class="mxchat-image-item">';
3414 - $html_output .= '<strong class="mxchat-image-title">' . $title . '</strong>';
3415 - $html_output .= '<a href="' . $image_url . '" target="_blank" rel="noopener noreferrer" class="mxchat-image-link">';
3416 - $html_output .= '<img src="' . $thumbnail_url . '" alt="' . $title . '" class="mxchat-image-thumbnail">';
3417 - $html_output .= '</a></div>';
3418 - }
3419 - }
3420 -
3421 - $html_output .= '</div>';
3422 -
3423 - // Create response text
3424 - $response_text = sprintf(__("Here are some images of %s:", 'mxchat'), $refined_search_query);
3425 -
3426 - // Save both response text and HTML to chat history
3427 - $this->mxchat_save_chat_message($session_id, 'bot', $response_text);
3428 - $this->mxchat_save_chat_message($session_id, 'bot', $html_output);
3429 -
3430 - // Return the combined response
3431 - return array(
3432 - 'text' => $response_text,
3433 - 'html' => $html_output,
3434 - );
3435 - } else {
3436 - $response_text = __("I'm sorry, I couldn't retrieve any images based on your request.", 'mxchat');
3437 -
3438 - // Save the error message to chat history
3439 - $this->mxchat_save_chat_message($session_id, 'bot', $response_text);
3440 -
3441 - return array(
3442 - 'text' => $response_text,
3443 - 'html' => "",
3444 - );
3445 - }
3446 -}
3447 -
3448 -/**
3449 - * Interpret the search query using the user's selected AI model
3450 - *
3451 - * @param string $user_query The original query from the user
3452 - * @return string The refined search query
3453 - */
3454 -public function mxchat_interpret_search_query($user_query) {
3455 - $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');
3456 -
3457 - // Get options and determine the selected model
3458 - $options = $this->options ?? get_option('mxchat_options');
3459 - $selected_model = isset($options['model']) ? $options['model'] : 'gpt-5.6-sol';
3460 -
3461 - // Custom (OpenAI-compatible) provider routes by model id, not prefix.
3462 - if ($selected_model === 'custom-provider') {
3463 - return $this->interpret_query_with_custom($user_query, $system_prompt);
3464 - }
3465 -
3466 - // Extract model prefix to determine the provider
3467 - $model_parts = explode('-', $selected_model);
3468 - $provider = strtolower($model_parts[0]);
3469 -
3470 - // Determine which API key to use based on the provider
3471 - switch ($provider) {
3472 - case 'gemini':
3473 - $api_key = isset($options['gemini_api_key']) ? sanitize_text_field($options['gemini_api_key']) : '';
3474 - if (empty($api_key)) {
3475 - return sanitize_text_field($user_query); // Default to original query if API key missing
3476 - }
3477 - return $this->interpret_query_with_gemini($user_query, $system_prompt, $api_key, $selected_model);
3478 -
3479 - case 'claude':
3480 - $api_key = isset($options['claude_api_key']) ? sanitize_text_field($options['claude_api_key']) : '';
3481 - if (empty($api_key)) {
3482 - return sanitize_text_field($user_query);
3483 - }
3484 - return $this->interpret_query_with_claude($user_query, $system_prompt, $api_key, $selected_model);
3485 -
3486 - case 'grok':
3487 - $api_key = isset($options['xai_api_key']) ? sanitize_text_field($options['xai_api_key']) : '';
3488 - if (empty($api_key)) {
3489 - return sanitize_text_field($user_query);
3490 - }
3491 - return $this->interpret_query_with_xai($user_query, $system_prompt, $api_key, $selected_model);
3492 -
3493 - case 'deepseek':
3494 - $api_key = isset($options['deepseek_api_key']) ? sanitize_text_field($options['deepseek_api_key']) : '';
3495 - if (empty($api_key)) {
3496 - return sanitize_text_field($user_query);
3497 - }
3498 - return $this->interpret_query_with_deepseek($user_query, $system_prompt, $api_key, $selected_model);
3499 -
3500 - case 'gpt':
3501 - default:
3502 - // Default to OpenAI for custom models or unrecognized prefixes
3503 - $api_key = isset($options['api_key']) ? sanitize_text_field($options['api_key']) : '';
3504 - if (empty($api_key)) {
3505 - return sanitize_text_field($user_query);
3506 - }
3507 - return $this->interpret_query_with_openai($user_query, $system_prompt, $api_key, $selected_model);
3508 - }
3509 -}
3510 -
3511 -/**
3512 - * Interpret query against the configured Custom (OpenAI-compatible) provider.
3513 - * Uses the same base URL + auth scheme as the chat dispatcher.
3514 - */
3515 -private function interpret_query_with_custom($user_query, $system_prompt) {
3516 - $cfg = $this->mxchat_resolve_custom_provider();
3517 - if (empty($cfg['base_url'])) {
3518 - return sanitize_text_field($user_query);
3519 - }
3520 - // plan-mxchat-20260715-7124f4: a custom OpenAI-compatible endpoint pointed at
3521 - // a gpt-5-class model rejects temperature!=1 and the legacy max_tokens key.
3522 - // Byte-identical for ordinary custom models (temperature kept, max_tokens
3523 - // used); only gpt-5-class custom models change (best-effort — custom
3524 - // endpoints vary).
3525 - $token_key = $this->mxchat_openai_token_param_for($cfg['model']);
3526 - $payload = [
3527 - 'model' => $cfg['model'],
3528 - 'messages' => [
3529 - ['role' => 'system', 'content' => $system_prompt],
3530 - ['role' => 'user', 'content' => sanitize_text_field($user_query)],
3531 - ],
3532 - $token_key => 20,
3533 - ];
3534 - if ($this->mxchat_openai_supports_temperature_for($cfg['model'])) {
3535 - $payload['temperature'] = 0.2;
3536 - }
3537 - $args = [
3538 - 'headers' => $this->mxchat_custom_provider_assoc_headers($cfg),
3539 - 'body' => wp_json_encode($payload),
3540 - 'method' => 'POST',
3541 - 'timeout' => 15,
3542 - ];
3543 - $response = wp_remote_post($cfg['chat_url'], $args);
3544 - if (is_wp_error($response)) {
3545 - return sanitize_text_field($user_query);
3546 - }
3547 - $body = json_decode(wp_remote_retrieve_body($response), true);
3548 - return isset($body['choices'][0]['message']['content'])
3549 - ? sanitize_text_field(trim($body['choices'][0]['message']['content']))
3550 - : sanitize_text_field($user_query);
3551 -}
3552 -
3553 -/**
3554 - * Convert the colon-style header list returned by mxchat_resolve_custom_provider
3555 - * into the assoc-array form wp_remote_post expects.
3556 - */
3557 -private function mxchat_custom_provider_assoc_headers($cfg) {
3558 - $headers = ['Content-Type' => 'application/json'];
3559 - if (!empty($cfg['api_key'])) {
3560 - if (($cfg['auth_scheme'] ?? 'bearer') === 'api-key') {
3561 - $headers['api-key'] = $cfg['api_key'];
3562 - } else {
3563 - $headers['Authorization'] = 'Bearer ' . $cfg['api_key'];
3564 - }
3565 - }
3566 - return $headers;
3567 -}
3568 -
3569 -/**
3570 - * Interpret query using OpenAI models
3571 - */
3572 -private function interpret_query_with_openai($user_query, $system_prompt, $api_key, $model = 'gpt-5.6-sol') {
3573 - $url = 'https://api.openai.com/v1/chat/completions';
3574 - // plan-mxchat-20260715-7124f4: the default chat model is a gpt-5-family id
3575 - // and every gpt-5* rejects both a non-default temperature and the legacy
3576 - // max_tokens key (400). This call swallowed the 400 and silently degraded to
3577 - // the raw query on every gpt-5 install, quietly disabling product/image
3578 - // search-query interpretation. Derive capability from the core catalog
3579 - // (dcb71c) so this tracks future model adds; strpos fallback for a
3580 - // partial-upgrade window where the catalog method isn't loaded.
3581 - $token_key = $this->mxchat_openai_token_param_for($model);
3582 - $payload = [
3583 - 'model' => $model,
3584 - 'messages' => [
3585 - ['role' => 'system', 'content' => $system_prompt],
3586 - ['role' => 'user', 'content' => sanitize_text_field($user_query)],
3587 - ],
3588 - $token_key => 20,
3589 - ];
3590 - if ($this->mxchat_openai_supports_temperature_for($model)) {
3591 - $payload['temperature'] = 0.2;
3592 - }
3593 - $args = [
3594 - 'headers' => [
3595 - 'Authorization' => 'Bearer ' . $api_key,
3596 - 'Content-Type' => 'application/json',
3597 - ],
3598 - 'body' => wp_json_encode($payload),
3599 - 'method' => 'POST',
3600 - 'timeout' => 15,
3601 - ];
3602 -
3603 - $response = wp_remote_post($url, $args);
3604 - if (is_wp_error($response)) {
3605 - return sanitize_text_field($user_query);
3606 - }
3607 -
3608 - $body = json_decode(wp_remote_retrieve_body($response), true);
3609 - return isset($body['choices'][0]['message']['content'])
3610 - ? sanitize_text_field(trim($body['choices'][0]['message']['content']))
3611 - : sanitize_text_field($user_query);
3612 -}
3613 -
3614 -/**
3615 - * Anthropic removed temperature/top_p/top_k starting with Opus 4.7 (the API
3616 - * returns 400 if sent) — add new flagship model ids here. (We don't send
3617 - * top_p/top_k in any Claude body, so the list only needs to gate temperature
3618 - * stripping. We never send a `thinking` param either, which is required for
3619 - * claude-fable-5: it rejects an explicit thinking "disabled" — omit only.)
3620 - */
3621 -private function mxchat_claude_omits_temperature($model) {
3622 - // plan-mxchat-20260714-dcb71c: derive from the core model catalog (single
3623 - // source of truth). Every caller here passes a Claude model, so
3624 - // !supports_temperature() reproduces the old 4-id in_array() result exactly.
3625 - // Frozen list kept as fallback for a partial-upgrade window where the
3626 - // catalog method isn't loaded.
3627 - if (class_exists('MxChat_Model_Catalog') && method_exists('MxChat_Model_Catalog', 'supports_temperature')) {
3628 - return !MxChat_Model_Catalog::supports_temperature($model);
3629 - }
3630 - $no_temp = array('claude-opus-5', 'claude-opus-4-7', 'claude-opus-4-8', 'claude-fable-5', 'claude-sonnet-5');
3631 - return in_array($model, $no_temp, true);
3632 -}
3633 -
3634 -/**
3635 - * plan-mxchat-20260715-7124f4: OpenAI completion-token key for this model,
3636 - * sourced from the core catalog (gpt-5* → max_completion_tokens; else
3637 - * max_tokens). strpos fallback for a partial-upgrade window where the catalog
3638 - * method isn't loaded.
3639 - *
3640 - * @param string $model OpenAI(-compatible) model id.
3641 - * @return string 'max_completion_tokens' | 'max_tokens'
3642 - */
3643 -private function mxchat_openai_token_param_for($model) {
3644 - if (class_exists('MxChat_Model_Catalog') && method_exists('MxChat_Model_Catalog', 'openai_token_param')) {
3645 - return MxChat_Model_Catalog::openai_token_param($model);
3646 - }
3647 - return strpos((string) $model, 'gpt-5') === 0 ? 'max_completion_tokens' : 'max_tokens';
3648 -}
3649 -
3650 -/**
3651 - * plan-mxchat-20260715-7124f4: whether a NON-default temperature may be sent to
3652 - * this OpenAI(-compatible) model. gpt-5* accept only the default (1) — sending
3653 - * any other value 400s. Sourced from the core catalog; strpos fallback for a
3654 - * partial-upgrade window.
3655 - *
3656 - * @param string $model OpenAI(-compatible) model id.
3657 - * @return bool
3658 - */
3659 -private function mxchat_openai_supports_temperature_for($model) {
3660 - if (class_exists('MxChat_Model_Catalog') && method_exists('MxChat_Model_Catalog', 'supports_temperature')) {
3661 - return MxChat_Model_Catalog::supports_temperature($model);
3662 - }
3663 - return strpos((string) $model, 'gpt-5') !== 0;
3664 -}
3665 -
3666 -/**
3667 - * plan-mxchat-20260714-dcb71c: per-surface reasoning_effort, sourced from the
3668 - * core model catalog so a model add propagates automatically. The fallback is
3669 - * the frozen pre-dcb71c inline ladder, used only if the catalog method is
3670 - * unavailable (a partial-upgrade window). Byte-identical to the old inline
3671 - * blocks by construction — proven by the dcb71c equivalence harness.
3672 - *
3673 - * @param string $model Chat model id.
3674 - * @param string $context 'chat' | 'websearch'.
3675 - * @return string|null Effort to send, or null to omit the param.
3676 - */
3677 -private function mxchat_reasoning_effort_for($model, $context) {
3678 - if (class_exists('MxChat_Model_Catalog') && method_exists('MxChat_Model_Catalog', 'reasoning_effort_for')) {
3679 - return MxChat_Model_Catalog::reasoning_effort_for($model, $context);
3680 - }
3681 - return $this->mxchat_reasoning_effort_fallback($model, $context);
3682 -}
3683 -
3684 -private function mxchat_reasoning_effort_fallback($model, $context) {
3685 - if (strpos($model, 'gpt-5') !== 0) {
3686 - return null;
3687 - }
3688 - if ($context === 'websearch') {
3689 - $no_reasoning_web = array('gpt-5.2', 'gpt-5.3-chat-latest', 'gpt-5.4-mini', 'gpt-5.4-nano');
3690 - if (in_array($model, $no_reasoning_web, true)) return null;
3691 - if ($model === 'gpt-5.1-2025-11-13') return 'low';
3692 - if ($model === 'gpt-5.5') return 'low';
3693 - if ($model === 'gpt-5.4') return 'low';
3694 - if (in_array($model, array('gpt-5.6-sol', 'gpt-5.6-terra', 'gpt-5.6-luna'), true)) return 'low';
3695 - return null;
3696 - }
3697 - // 'chat'
3698 - // gpt-5.1/5.3-chat-latest stay listed after their 2026-08-10 retirement:
3699 - // unmigrated bot-level / add-on-saved ids must keep routing correctly
3700 - // until every surface is swept (plan e46b8f).
3701 - $no_reasoning_models = array('gpt-5.2', 'gpt-5.1-chat-latest', 'gpt-5.3-chat-latest', 'gpt-5.4-mini', 'gpt-5.4-nano');
3702 - if (in_array($model, $no_reasoning_models, true)) return null;
3703 - if ($model === 'gpt-5.1-2025-11-13') return 'low';
3704 - if ($model === 'gpt-5.5') return 'none';
3705 - if ($model === 'gpt-5.4') return 'none';
3706 - if (in_array($model, array('gpt-5.6-sol', 'gpt-5.6-terra', 'gpt-5.6-luna'), true)) return 'low';
3707 - return 'minimal';
3708 -}
3709 -
3710 -/**
3711 - * Interpret query using Claude models
3712 - */
3713 -private function interpret_query_with_claude($user_query, $system_prompt, $api_key, $model) {
3714 - // Anthropic retired claude-opus-4-20250514 / claude-sonnet-4-20250514 on 2026-06-15.
3715 - // Read-time rescue: remap a saved dead ID to the current equivalent before the API call.
3716 - if ($model === 'claude-opus-4-20250514') { $model = 'claude-opus-4-8'; }
3717 - elseif ($model === 'claude-sonnet-4-20250514') { $model = 'claude-sonnet-4-6'; }
3718 - $url = 'https://api.anthropic.com/v1/messages';
3719 -
3720 - $payload = [
3721 - 'model' => $model,
3722 - 'system' => $system_prompt,
3723 - 'messages' => [
3724 - ['role' => 'user', 'content' => sanitize_text_field($user_query)]
3725 - ],
3726 - 'max_tokens' => 20,
3727 - 'temperature' => 0.2,
3728 - ];
3729 - if ($this->mxchat_claude_omits_temperature($model)) { unset($payload['temperature']); }
3730 -
3731 - $args = [
3732 - 'headers' => [
3733 - 'Content-Type' => 'application/json',
3734 - 'x-api-key' => $api_key,
3735 - 'anthropic-version' => '2023-06-01',
3736 - ],
3737 - 'body' => wp_json_encode($payload),
3738 - 'method' => 'POST',
3739 - 'timeout' => 15,
3740 - ];
3741 -
3742 - $response = wp_remote_post($url, $args);
3743 - if (is_wp_error($response)) {
3744 - return sanitize_text_field($user_query);
3745 - }
3746 -
3747 - $body = json_decode(wp_remote_retrieve_body($response), true);
3748 - // claude-fable-5 prepends a thinking block to content — take the first
3749 - // TEXT block, not content[0].
3750 - foreach ((array) ($body['content'] ?? array()) as $block) {
3751 - if (isset($block['type'], $block['text']) && $block['type'] === 'text' && trim($block['text']) !== '') {
3752 - return sanitize_text_field(trim($block['text']));
3753 - }
3754 - }
3755 -
3756 - return sanitize_text_field($user_query);
3757 -}
3758 -
3759 -/**
3760 - * Interpret query using Gemini models
3761 - */
3762 -private function interpret_query_with_gemini($user_query, $system_prompt, $api_key, $model) {
3763 - if ($model === 'gemini-3-pro-preview') {
3764 - $model = 'gemini-3.1-pro-preview';
3765 - }
3766 - // Use v1beta for preview models, v1 for stable models
3767 - $api_version = (strpos($model, 'preview') !== false || strpos($model, 'exp') !== false) ? 'v1beta' : 'v1';
3768 -
3769 - $url = "https://generativelanguage.googleapis.com/{$api_version}/models/{$model}:generateContent?key=" . urlencode($api_key);
3770 -
3771 - $args = [
3772 - 'headers' => [
3773 - 'Content-Type' => 'application/json',
3774 - ],
3775 - 'body' => wp_json_encode([
3776 - 'contents' => [
3777 - [
3778 - 'role' => 'user',
3779 - 'parts' => [
3780 - ['text' => $system_prompt . "\n\nQuery: " . sanitize_text_field($user_query)]
3781 - ]
3782 - ]
3783 - ],
3784 - 'generationConfig' => [
3785 - 'temperature' => 0.2,
3786 - 'maxOutputTokens' => 20,
3787 - ],
3788 - ]),
3789 - 'method' => 'POST',
3790 - 'timeout' => 15,
3791 - ];
3792 -
3793 - $response = wp_remote_post($url, $args);
3794 - if (is_wp_error($response)) {
3795 - return sanitize_text_field($user_query);
3796 - }
3797 -
3798 - $body = json_decode(wp_remote_retrieve_body($response), true);
3799 - if (!empty($body['candidates'][0]['content']['parts'][0]['text'])) {
3800 - return sanitize_text_field(trim($body['candidates'][0]['content']['parts'][0]['text']));
3801 - }
3802 -
3803 - return sanitize_text_field($user_query);
3804 -}
3805 -
3806 -/**
3807 - * Interpret query using X.AI (Grok) models
3808 - */
3809 -private function interpret_query_with_xai($user_query, $system_prompt, $api_key, $model) {
3810 - $url = 'https://api.xai.com/v1/chat/completions';
3811 -
3812 - $args = [
3813 - 'headers' => [
3814 - 'Content-Type' => 'application/json',
3815 - 'Authorization' => 'Bearer ' . $api_key,
3816 - ],
3817 - 'body' => wp_json_encode([
3818 - 'model' => $model,
3819 - 'messages' => [
3820 - ['role' => 'system', 'content' => $system_prompt],
3821 - ['role' => 'user', 'content' => sanitize_text_field($user_query)],
3822 - ],
3823 - 'temperature' => 0.2,
3824 - 'max_tokens' => 20,
3825 - ]),
3826 - 'method' => 'POST',
3827 - 'timeout' => 15,
3828 - ];
3829 -
3830 - $response = wp_remote_post($url, $args);
3831 - if (is_wp_error($response)) {
3832 - return sanitize_text_field($user_query);
3833 - }
3834 -
3835 - $body = json_decode(wp_remote_retrieve_body($response), true);
3836 - if (isset($body['choices'][0]['message']['content'])) {
3837 - return sanitize_text_field(trim($body['choices'][0]['message']['content']));
3838 - }
3839 -
3840 - return sanitize_text_field($user_query);
3841 -}
3842 -
3843 -/**
3844 - * Interpret query using DeepSeek models
3845 - */
3846 -private function interpret_query_with_deepseek($user_query, $system_prompt, $api_key, $model) {
3847 - $url = 'https://api.deepseek.com/v1/chat/completions';
3848 -
3849 - $args = [
3850 - 'headers' => [
3851 - 'Content-Type' => 'application/json',
3852 - 'Authorization' => 'Bearer ' . $api_key,
3853 - ],
3854 - 'body' => wp_json_encode([
3855 - 'model' => $model,
3856 - 'messages' => [
3857 - ['role' => 'system', 'content' => $system_prompt],
3858 - ['role' => 'user', 'content' => sanitize_text_field($user_query)],
3859 - ],
3860 - 'temperature' => 0.2,
3861 - 'max_tokens' => 20,
3862 - // DeepSeek V4 defaults to thinking mode ON (temperature ignored,
3863 - // reasoning burns the 20-token budget); keep the legacy
3864 - // deepseek-chat semantics = non-thinking.
3865 - 'thinking' => ['type' => 'disabled'],
3866 - ]),
3867 - 'method' => 'POST',
3868 - 'timeout' => 15,
3869 - ];
3870 -
3871 - $response = wp_remote_post($url, $args);
3872 - if (is_wp_error($response)) {
3873 - return sanitize_text_field($user_query);
3874 - }
3875 -
3876 - $body = json_decode(wp_remote_retrieve_body($response), true);
3877 - if (isset($body['choices'][0]['message']['content'])) {
3878 - return sanitize_text_field(trim($body['choices'][0]['message']['content']));
3879 - }
3880 -
3881 - return sanitize_text_field($user_query);
3882 -}
3883 -
3884 -//very good
3885 -private function add_email_to_loops($email) {
3886 - // Sanitize the email
3887 - $email = sanitize_email($email);
3888 -
3889 - // Retrieve and sanitize options
3890 - $api_key = isset($this->options['loops_api_key']) ? sanitize_text_field($this->options['loops_api_key']) : '';
3891 - $mailing_list_id = isset($this->options['loops_mailing_list']) ? sanitize_text_field($this->options['loops_mailing_list']) : '';
3892 -
3893 - // Check for missing API key or mailing list ID
3894 - if (empty($api_key) || empty($mailing_list_id)) {
3895 - //error_log(esc_html__('Loops API key or mailing list ID is missing.', 'mxchat'));
3896 - return;
3897 - }
3898 -
3899 - $data = array(
3900 - 'email' => $email,
3901 - 'subscribed' => true,
3902 - 'source' => __('MxChat AI Chatbot', 'mxchat'),
3903 - 'mailingLists' => array($mailing_list_id => true),
3904 - );
3905 -
3906 - $url = 'https://app.loops.so/api/v1/contacts/create';
3907 - $args = array(
3908 - 'body' => wp_json_encode($data),
3909 - 'headers' => array(
3910 - 'Authorization' => 'Bearer ' . $api_key,
3911 - 'Content-Type' => 'application/json',
3912 - ),
3913 - 'method' => 'POST',
3914 - 'timeout' => 45,
3915 - );
3916 -
3917 - $response = wp_remote_post($url, $args);
3918 -
3919 - // Handle errors in the API request
3920 - if (is_wp_error($response)) {
3921 - //error_log(esc_html__('Error adding email to Loops: ', 'mxchat') . $response->get_error_message());
3922 - return;
3923 - }
3924 -
3925 - // Check for non-200 HTTP responses
3926 - $response_code = wp_remote_retrieve_response_code($response);
3927 - if ($response_code != 200) {
3928 - $response_body = wp_remote_retrieve_body($response);
3929 - //error_log(esc_html__('Loops API responded with code ', 'mxchat') . $response_code . ': ' . $response_body);
3930 - }
3931 -}
3932 -
3933 -public function mxchat_handle_pdf_discussion($message, $user_id, $session_id) {
3934 - // Get the maximum number of pages allowed from admin settings
3935 - $max_pages = isset($this->options['pdf_max_pages']) ? intval($this->options['pdf_max_pages']) : 69;
3936 -
3937 - // Retrieve options for dynamic texts
3938 - $trigger_text = $this->options['pdf_intent_trigger_text'] ?? __("Please provide the URL to the PDF you'd like to discuss.", 'mxchat');
3939 - $success_text = $this->options['pdf_intent_success_text'] ?? __("I've processed the PDF. What questions do you have about it?", 'mxchat');
3940 - $error_text = $this->options['pdf_intent_error_text'] ?? __("Sorry, I couldn't process the PDF. Please ensure it's a valid file.", 'mxchat');
3941 -
3942 - // Check for explicit request for new PDF
3943 - $new_pdf_requested = stripos($message, 'new') !== false ||
3944 - stripos($message, 'another') !== false ||
3945 - stripos($message, 'different') !== false;
3946 -
3947 - // If user mentions adding/reading a PDF, set waiting flag
3948 - if (stripos($message, 'pdf') !== false ||
3949 - stripos($message, 'document') !== false ||
3950 - stripos($message, 'read') !== false) {
3951 - set_transient('mxchat_waiting_for_pdf_url_' . $session_id, true, HOUR_IN_SECONDS);
3952 - $this->fallbackResponse['text'] = $trigger_text;
3953 - return;
3954 - }
3955 -
3956 - // If we're waiting for a URL or user requested new PDF
3957 - if ($new_pdf_requested || get_transient('mxchat_waiting_for_pdf_url_' . $session_id)) {
3958 - if (preg_match('/https?:\/\/[^\s"]+/i', $message, $matches)) {
3959 - // Process URL... (rest of your existing URL processing code)
3960 - } else {
3961 - $this->fallbackResponse['text'] = $trigger_text;
3962 - }
3963 - return;
3964 - }
3965 -
3966 - // Default to proceeding with conversation if no specific PDF action is needed
3967 - $this->fallbackResponse['text'] = '';
3968 -}
3969 -
3970 -
3971 -/**
3972 - * Enhanced fetch_and_split_pdf_pages with SSRF protection
3973 - */
3974 -private function fetch_and_split_pdf_pages($pdf_source, $max_pages) {
3975 - // CLEAR DEBUG LOGGING
3976 - //error_log("=== MXCHAT PDF PROCESSING START ===");
3977 - //error_log("PDF Source: " . $pdf_source);
3978 - //error_log("Max Pages: " . $max_pages);
3979 - //error_log("Session ID: " . ($this->session_id ?? 'not set'));
3980 -
3981 - // Check if Advanced Claude Toolbar is available and enabled
3982 - $claude_available = function_exists('mxchatACT_is_advanced_claude_enabled');
3983 - $claude_enabled = $claude_available ? mxchatACT_is_advanced_claude_enabled() : false;
3984 -
3985 - //error_log("Claude Function Available: " . ($claude_available ? 'YES' : 'NO'));
3986 - //error_log("Claude Enabled: " . ($claude_enabled ? 'YES' : 'NO'));
3987 -
3988 - if ($claude_available && $claude_enabled) {
3989 - //error_log("🚀 ATTEMPTING CLAUDE PROCESSING...");
3990 -
3991 - // Attempt Claude processing first
3992 - $claude_result = apply_filters('mxchat_process_pdf_advanced', false, $pdf_source, $max_pages, $this->session_id);
3993 -
3994 - if ($claude_result !== false && is_array($claude_result) && !empty($claude_result)) {
3995 - //error_log("✅ CLAUDE PROCESSING SUCCESSFUL!");
3996 - //error_log("Claude returned " . count($claude_result) . " processed pages");
3997 -
3998 - // Log first page details for verification
3999 - if (isset($claude_result[0])) {
4000 - $first_page = $claude_result[0];
4001 - //error_log("First page enhanced: " . (isset($first_page['enhanced']) && $first_page['enhanced'] ? 'YES' : 'NO'));
4002 - //error_log("Processing method: " . ($first_page['processing_method'] ?? 'not set'));
4003 - //error_log("First page text preview: " . substr($first_page['text'] ?? '', 0, 100) . "...");
4004 - }
4005 -
4006 - //error_log("=== MXCHAT PDF PROCESSING END (CLAUDE) ===");
4007 - return $claude_result;
4008 - } else {
4009 - //error_log("❌ CLAUDE PROCESSING FAILED or returned invalid result");
4010 - //error_log("Claude result type: " . gettype($claude_result));
4011 - if (is_array($claude_result)) {
4012 - //error_log("Claude result count: " . count($claude_result));
4013 - }
4014 - }
4015 - }
4016 -
4017 - // Fallback to basic processing
4018 - //error_log("🔄 FALLING BACK TO BASIC PDF PROCESSING...");
4019 -
4020 - $upload_dir = wp_upload_dir();
4021 - $temp_file = null;
4022 -
4023 - try {
4024 - // Your existing basic processing code here...
4025 - // (I'll include the key parts with debug logging)
4026 -
4027 - if (filter_var($pdf_source, FILTER_VALIDATE_URL)) {
4028 - //error_log("Downloading PDF from URL...");
4029 -
4030 - // SECURITY FIX: Validate URL before processing
4031 - if (!$this->mxchat_is_safe_pdf_url($pdf_source)) {
4032 - //error_log("❌ SECURITY: Blocked unsafe PDF URL");
4033 - return false;
4034 - }
4035 -
4036 - $temp_file = wp_tempnam($pdf_source);
4037 -
4038 - // SECURITY FIX: Changed from wp_remote_get to wp_safe_remote_get
4039 - // Route through the shared MXChat crawler identity (plan bae78f/b6d93c) so
4040 - // every remote-content fetch presents one honest, versioned, filterable,
4041 - // allowlistable User-Agent. function_exists guard keeps the front-end/nopriv
4042 - // path safe if the helper (in the always-loaded main file) is ever unavailable.
4043 - $response = wp_safe_remote_get($pdf_source, [
4044 - 'timeout' => 60,
4045 - 'headers' => ['User-Agent' => function_exists('mxchat_ingest_user_agent') ? mxchat_ingest_user_agent() : 'MxChat PDF Processor']
4046 - ]);
4047 -
4048 - if (is_wp_error($response) || wp_remote_retrieve_response_code($response) !== 200) {
4049 - $error_message = is_wp_error($response) ? $response->get_error_message() : 'HTTP ' . wp_remote_retrieve_response_code($response);
4050 - //error_log("❌ BASIC PROCESSING: Failed to download PDF: " . $error_message);
4051 - return false;
4052 - }
4053 -
4054 - global $wp_filesystem;
4055 - if (empty($wp_filesystem)) {
4056 - require_once ABSPATH . 'wp-admin/includes/file.php';
4057 - WP_Filesystem();
4058 - }
4059 - $wp_filesystem->put_contents($temp_file, wp_remote_retrieve_body($response), FS_CHMOD_FILE);
4060 - //error_log("✅ PDF downloaded successfully");
4061 - } else {
4062 - $temp_file = $pdf_source;
4063 - //error_log("Using local PDF file: " . $temp_file);
4064 - }
4065 -
4066 - // Parse PDF
4067 - //error_log("Parsing PDF with basic parser...");
4068 - mxchat_load_pdf_parser();
4069 - $parser = new \Smalot\PdfParser\Parser();
4070 - $pdf = $parser->parseFile($temp_file);
4071 - $pages = $pdf->getPages();
4072 -
4073 - //error_log("PDF contains " . count($pages) . " pages");
4074 -
4075 - if (count($pages) > $max_pages) {
4076 - //error_log("❌ BASIC PROCESSING: Too many pages (" . count($pages) . " > " . $max_pages . ")");
4077 - if (filter_var($pdf_source, FILTER_VALIDATE_URL) && $temp_file) {
4078 - unlink($temp_file);
4079 - }
4080 - return 'too_many_pages';
4081 - }
4082 -
4083 - $embeddings = [];
4084 - $processed_pages = 0;
4085 -
4086 - foreach ($pages as $page_number => $page) {
4087 - $text = $page->getText();
4088 -
4089 - if (empty(trim($text))) {
4090 - //error_log("Skipping empty page: " . ($page_number + 1));
4091 - continue;
4092 - }
4093 -
4094 - $text = $this->mxchat_clean_text($text);
4095 -
4096 - $embedding = $this->mxchat_generate_embedding(
4097 - __("Page ", 'mxchat') . ($page_number + 1) . ": " . $text,
4098 - $this->options['api_key']
4099 - );
4100 -
4101 - if ($embedding) {
4102 - $embeddings[] = [
4103 - 'page_number' => $page_number + 1,
4104 - 'embedding' => $embedding,
4105 - 'text' => $text,
4106 - 'enhanced' => false, // CLEARLY MARK AS BASIC
4107 - 'processing_method' => 'basic_pdf_parser'
4108 - ];
4109 - $processed_pages++;
4110 - }
4111 - }
4112 -
4113 - //error_log("✅ BASIC PROCESSING COMPLETE: " . $processed_pages . " pages processed");
4114 -
4115 - // Cleanup
4116 - if (filter_var($pdf_source, FILTER_VALIDATE_URL) && $temp_file && file_exists($temp_file)) {
4117 - unlink($temp_file);
4118 - }
4119 -
4120 - //error_log("=== MXCHAT PDF PROCESSING END (BASIC) ===");
4121 - return $embeddings;
4122 -
4123 - } catch (\Exception $e) {
4124 - //error_log("❌ BASIC PROCESSING ERROR: " . $e->getMessage());
4125 - if (filter_var($pdf_source, FILTER_VALIDATE_URL) && $temp_file && file_exists($temp_file)) {
4126 - unlink($temp_file);
4127 - }
4128 - //error_log("=== MXCHAT PDF PROCESSING END (ERROR) ===");
4129 - return false;
4130 - }
4131 -}
4132 -
4133 -
4134 -/**
4135 - * Validate PDF URL for security
4136 - * Prevents SSRF attacks by blocking dangerous URLs
4137 - */
4138 -
4139 -private function mxchat_is_safe_pdf_url($url) {
4140 - // Use WordPress core function for comprehensive validation
4141 - // This blocks localhost, private IPs, and reserved IP ranges
4142 - $validated_url = wp_http_validate_url($url);
4143 -
4144 - if ($validated_url === false) {
4145 - return false;
4146 - }
4147 -
4148 - // Additional check: only allow HTTP/HTTPS schemes
4149 - $parsed = parse_url($url);
4150 - if (!isset($parsed['scheme']) || !in_array($parsed['scheme'], ['http', 'https'], true)) {
4151 - return false;
4152 - }
4153 -
4154 - return true;
4155 -}
4156 -
4157 -
4158 -private function mxchat_clean_text($text) {
4159 - // Remove excessive whitespace
4160 - $text = preg_replace('/\s+/', ' ', $text);
4161 -
4162 - // Remove control characters except newlines and tabs
4163 - $text = preg_replace('/[\x00-\x08\x0B\x0C\x0E-\x1F\x7F]/', '', $text);
4164 -
4165 - // Normalize line endings
4166 - $text = str_replace(["\r\n", "\r"], "\n", $text);
4167 -
4168 - // Trim whitespace
4169 - $text = trim($text);
4170 -
4171 - return $text;
4172 -}
4173 -
4174 -private function find_relevant_pdf_pages($query_embedding, $embeddings) {
4175 - //error_log(esc_html__("find_relevant_pdf_pages called.", 'mxchat'));
4176 -
4177 - $most_relevant = null;
4178 - $highest_similarity = -INF;
4179 -
4180 - foreach ($embeddings as $page_data) {
4181 - $similarity = $this->mxchat_calculate_cosine_similarity($query_embedding, $page_data['embedding']);
4182 -
4183 - if ($similarity > $highest_similarity) {
4184 - $highest_similarity = $similarity;
4185 - $most_relevant = $page_data['page_number'];
4186 - }
4187 - }
4188 -
4189 - if (!is_null($most_relevant)) {
4190 - $page_numbers = range(max(1, $most_relevant - 1), min(count($embeddings), $most_relevant + 1));
4191 - return array_filter($embeddings, function ($page) use ($page_numbers) {
4192 - return in_array($page['page_number'], $page_numbers);
4193 - });
4194 - }
4195 -
4196 - return [];
4197 -}
4198 -
4199 -
4200 -public function handle_pdf_upload() {
4201 - if (!isset($_POST['nonce']) || !MxChat_Integrator::mxchat_verify_chat_send_nonce(wp_unslash((string) $_POST['nonce']))) {
4202 - wp_send_json_error(array('message' => esc_html__('Invalid nonce.', 'mxchat')), 403);
4203 - }
4204 -
4205 - if (!isset($_FILES['pdf_file']) || !isset($_POST['session_id'])) {
4206 - wp_send_json_error(esc_html__('Missing required parameters.', 'mxchat'));
4207 - return;
4208 - }
4209 -
4210 - // SECURITY FIX: Check if PDF uploads are enabled in settings
4211 - $options = get_option('mxchat_options', array());
4212 - $show_pdf_button = isset($options['show_pdf_upload_button']) ? $options['show_pdf_upload_button'] : 'on';
4213 -
4214 - if ($show_pdf_button !== 'on') {
4215 - wp_send_json_error(esc_html__('PDF uploads are currently disabled.', 'mxchat'));
4216 - return;
4217 - }
4218 -
4219 - $file = $_FILES['pdf_file'];
4220 - $session_id = MxChat_Utils::sanitize_session_id(wp_unslash($_POST['session_id']));
4221 - $original_filename = sanitize_text_field($file['name']);
4222 -
4223 - // Update session owner if it changed (e.g. IP changed due to network switch)
4224 - $current_user_identifier = MxChat_User::mxchat_get_user_identifier();
4225 - $session_owner = get_option("mxchat_session_owner_{$session_id}");
4226 -
4227 - if (!$session_owner || $session_owner !== $current_user_identifier) {
4228 - update_option("mxchat_session_owner_{$session_id}", $current_user_identifier, 'no');
4229 - }
4230 -
4231 - $file_type = wp_check_filetype($file['name'], ['pdf' => 'application/pdf']);
4232 - if ($file_type['type'] !== 'application/pdf') {
4233 - wp_send_json_error(esc_html__('Invalid file type. Only PDF files are allowed.', 'mxchat'));
4234 - return;
4235 - }
4236 -
4237 - $upload_dir = wp_upload_dir();
4238 -
4239 - // SECURITY FIX: Generate random filename without exposing session_id
4240 - $random_string = wp_generate_password(20, false, false); // 20 char alphanumeric string
4241 - $pdf_filename = 'mxchat_' . $random_string . '_' . time() . '.pdf';
4242 - $pdf_path = $upload_dir['path'] . '/' . $pdf_filename;
4243 -
4244 - if (!move_uploaded_file($file['tmp_name'], $pdf_path)) {
4245 - wp_send_json_error(esc_html__('Failed to upload file.', 'mxchat'));
4246 - return;
4247 - }
4248 -
4249 - $this->clear_pdf_transients($session_id);
4250 -
4251 - $max_pages = isset($this->options['pdf_max_pages']) ? intval($this->options['pdf_max_pages']) : 69;
4252 - $embeddings = $this->fetch_and_split_pdf_pages($pdf_path, $max_pages);
4253 -
4254 - if ($embeddings === 'too_many_pages') {
4255 - unlink($pdf_path);
4256 - $error_message = sprintf(
4257 - $this->options['pdf_intent_error_text'] ??
4258 - esc_html__("The provided PDF exceeds the maximum allowed limit of %d pages. Please provide a smaller document.", 'mxchat'),
4259 - $max_pages
4260 - );
4261 - wp_send_json_error($error_message);
4262 - return;
4263 - }
4264 -
4265 - if ($embeddings === false || empty($embeddings)) {
4266 - unlink($pdf_path);
4267 - $error_message = $this->options['pdf_intent_error_text'] ??
4268 - esc_html__('The uploaded PDF appears to be empty or contains unsupported content.', 'mxchat');
4269 - wp_send_json_error($error_message);
4270 - return;
4271 - }
4272 -
4273 - if (!empty($embeddings)) {
4274 - // Store the mapping between session and the random filename
4275 - set_transient('mxchat_pdf_url_' . $session_id, $pdf_path, HOUR_IN_SECONDS);
4276 - set_transient('mxchat_pdf_filename_' . $session_id, $original_filename, HOUR_IN_SECONDS);
4277 - set_transient('mxchat_pdf_embeddings_' . $session_id, $embeddings, HOUR_IN_SECONDS);
4278 - set_transient('mxchat_include_pdf_in_context_' . $session_id, true, HOUR_IN_SECONDS);
4279 -
4280 - $success_message = $this->options['pdf_intent_success_text'] ??
4281 - esc_html__("I've processed the PDF. What questions do you have about it?", 'mxchat');
4282 -
4283 - wp_send_json_success([
4284 - 'message' => $success_message,
4285 - 'filename' => $original_filename
4286 - ]);
4287 - return;
4288 - }
4289 -
4290 - unlink($pdf_path);
4291 - $error_message = $this->options['pdf_intent_error_text'] ??
4292 - esc_html__('Sorry, I couldn\'t process the PDF. Please ensure it\'s a valid file.', 'mxchat');
4293 - wp_send_json_error($error_message);
4294 - return;
4295 -}
4296 -public function handle_pdf_remove() {
4297 - if (!isset($_POST['nonce']) || !MxChat_Integrator::mxchat_verify_chat_send_nonce(wp_unslash((string) $_POST['nonce']))) {
4298 - wp_send_json_error(array('message' => esc_html__('Invalid nonce.', 'mxchat')), 403);
4299 - }
4300 -
4301 - if (empty($_POST['session_id'])) {
4302 - wp_send_json_error(esc_html__('Session ID missing.', 'mxchat'));
212 + // Generate and validate the embedding
213 + $user_message_embedding = $this->mxchat_generate_embedding($message_with_order_details, $this->options['api_key']);
214 + if (!is_array($user_message_embedding)) {
215 + wp_send_json_error('Error processing your message.');
4303 216 wp_die();
4304 217 }
4305 218
4306 - $session_id = MxChat_Utils::sanitize_session_id(wp_unslash($_POST['session_id']));
4307 - if ($session_id === '') {
4308 - wp_send_json_error(esc_html__('Session ID missing.', 'mxchat'));
4309 - wp_die();
4310 - }
219 + // Find relevant content based on embedding
220 + $relevant_content = $this->mxchat_find_relevant_content($user_message_embedding);
4311 221
4312 - // Session-ownership bookkeeping (plan-mxchat-20260731-d42bec).
4313 - //
4314 - // Be clear about what this does and does not do. It mirrors the history
4315 - // endpoint's rule exactly, as directed, INCLUDING its changed-IP tolerance:
4316 - // possession of the session id IS the credential, so a mismatched identifier
4317 - // re-owns the session instead of being refused. That means this does NOT
4318 - // refuse a caller who supplies someone else's session id — it keeps the two
4319 - // endpoints agreeing about who owns a session, and records the owner so a
4320 - // future stricter policy has trustworthy data to enforce against.
4321 - //
4322 - // What actually protects another visitor's upload here is that session ids
4323 - // are 128-bit CSPRNG values (plan-0c17b5) and therefore not guessable. If we
4324 - // ever want a real boundary on this endpoint, it has to be decided for the
4325 - // history endpoint at the same time.
4326 - $current_user_identifier = MxChat_User::mxchat_get_user_identifier();
4327 - $session_owner = get_option("mxchat_session_owner_{$session_id}");
4328 - if (!$session_owner || $session_owner !== $current_user_identifier) {
4329 - update_option("mxchat_session_owner_{$session_id}", $current_user_identifier, 'no');
4330 - }
222 + // Fetch conversation history from the database
223 + $conversation_history = $this->mxchat_fetch_conversation_history_for_ajax($session_id);
4331 224
4332 - $pdf_path = get_transient('mxchat_pdf_url_' . $session_id);
225 + // Increment the chat count
226 + $this->mxchat_increment_chat_count();
4333 227
4334 - if ($pdf_path && file_exists($pdf_path)) {
4335 - unlink($pdf_path);
4336 - }
228 + // Generate a response from the AI model
229 + $response = $this->mxchat_generate_response($relevant_content, $this->options['api_key'], $conversation_history);
4337 230
4338 - $this->clear_pdf_transients($session_id);
231 + // Save the bot response to the database
232 + $this->mxchat_save_chat_message($session_id, 'bot', $response);
4339 233
4340 - wp_send_json_success([
4341 - 'message' => esc_html__('PDF removed successfully.', 'mxchat')
4342 - ]);
4343 - wp_die();
4344 -}
234 + // Send the response back to the client
235 + wp_send_json(['message' => $response]);
4345 236
4346 -
4347 -function mxchat_fetch_new_messages() {
4348 - $session_id = MxChat_Utils::sanitize_session_id(wp_unslash($_POST['session_id']));
4349 - $last_seen_id = sanitize_text_field($_POST['last_seen_id']);
4350 - $persistence_enabled = $_POST['persistence_enabled'] === 'true';
4351 - $initial_timestamp = isset($_POST['initial_timestamp']) ? intval($_POST['initial_timestamp']) : 0;
4352 -
4353 - if (empty($session_id)) {
4354 - //error_log(esc_html__('Fetch new messages error: Session ID missing.', 'mxchat'));
4355 - wp_send_json_error(['message' => esc_html__('Session ID missing.', 'mxchat')]);
4356 - wp_die();
4357 - }
4358 -
4359 - $history = get_option("mxchat_history_{$session_id}", []);
4360 -
4361 - //error_log("MxChat WhatsApp DEBUG: Fetch new messages for session {$session_id}");
4362 - //error_log("MxChat WhatsApp DEBUG: last_seen_id = " . var_export($last_seen_id, true));
4363 - //error_log("MxChat WhatsApp DEBUG: History count = " . count($history));
4364 - //error_log("MxChat WhatsApp DEBUG: Full history = " . print_r($history, true));
4365 -
4366 - $new_messages = array_filter($history, function ($message) use ($last_seen_id, $persistence_enabled, $initial_timestamp) {
4367 - //error_log("MxChat WhatsApp DEBUG: Checking message - ID: " . ($message['id'] ?? 'NO_ID') . ", Role: " . ($message['role'] ?? 'NO_ROLE'));
4368 -
4369 - // If persistence is enabled, show all new messages
4370 - if ($persistence_enabled) {
4371 - $has_id = !empty($message['id']);
4372 - $is_agent = $message['role'] === 'agent';
4373 -
4374 - // If last_seen_id is empty, 'NaN', or invalid, show all agent messages
4375 - if (empty($last_seen_id) || $last_seen_id === 'NaN' || $last_seen_id === 'undefined') {
4376 - $is_newer = true;
4377 - } else {
4378 - $is_newer = strcmp($message['id'] ?? '', $last_seen_id) > 0;
4379 - }
4380 -
4381 - //error_log("MxChat WhatsApp DEBUG: has_id={$has_id}, is_newer={$is_newer}, is_agent={$is_agent}");
4382 -
4383 - return $has_id && $is_newer && $is_agent;
4384 - }
4385 -
4386 - // If persistence is disabled, only show messages after initial timestamp
4387 - return !empty($message['id']) &&
4388 - $message['role'] === 'agent' &&
4389 - $message['timestamp'] > $initial_timestamp;
4390 - });
4391 -
4392 - //error_log("MxChat WhatsApp DEBUG: Filtered messages count = " . count($new_messages));
4393 -
4394 - // Include current chat mode so frontend can detect agent→AI transitions
4395 - $chat_mode = get_option("mxchat_mode_{$session_id}", 'ai');
4396 -
4397 - wp_send_json_success([
4398 - 'new_messages' => array_values($new_messages),
4399 - 'chat_mode' => $chat_mode
4400 - ]);
4401 237 wp_die();
4402 238 }
4403 -public function mxchat_live_agent_handover($message, $user_id, $session_id) {
4404 - // First check if live agents are available.
4405 - // Outside the SLACK availability schedule this behaves exactly like the
4406 - // manual toggle being off — same away message, same stay-in-AI-mode path
4407 - // (plans 8ccaa2 + 99d7a4: each channel owns its own schedule). The schedule
4408 - // normally stops the tool being offered at all; this is the backstop for
4409 - // any path that calls the handover directly.
4410 - $live_agent_available = $this->options['live_agent_status'] ?? 'off';
4411 - $within_hours = !class_exists('MxChat_Live_Agent_Schedule')
4412 - || MxChat_Live_Agent_Schedule::is_within_hours('slack');
4413 - if ($live_agent_available !== 'on' || !$within_hours) {
4414 - $away_message = $this->options['live_agent_away_message'] ?? 'Sorry, live agents are currently unavailable. I can continue helping you as an AI assistant.';
4415 - $this->fallbackResponse = [
4416 - 'text' => $away_message,
4417 - 'html' => '',
4418 - 'images' => [],
4419 - 'chat_mode' => 'ai'
4420 - ];
4421 - wp_send_json([
4422 - 'text' => $away_message,
4423 - 'html' => '',
4424 - 'chat_mode' => 'ai',
4425 - 'session_id' => $session_id
4426 - ]);
4427 - wp_die();
4428 - }
4429 239
4430 - $slack_bot_token = $this->options['live_agent_bot_token'] ?? '';
4431 240
4432 - if (empty($slack_bot_token)) {
4433 - return false;
4434 - }
4435 -
4436 - // Check if channel already exists for this session
4437 - $channel_id = get_option("mxchat_channel_{$session_id}", '');
4438 -
4439 - // Shared-channel mode (plan 9f7756): when a shared handoff channel is
4440 - // configured and this session doesn't already own a per-conversation
4441 - // channel, the handoff posts into the shared channel as a new thread
4442 - // (or into the session's existing thread on a re-handover). Any failure
4443 - // to reach the shared channel falls back to per-conversation creation
4444 - // below, so a misconfigured channel never drops a handoff.
4445 - $shared_channel_setting = trim($this->options['live_agent_shared_channel'] ?? '');
4446 - $shared_thread_ts = get_option("mxchat_thread_{$session_id}", '');
4447 - $use_shared_channel = ($shared_channel_setting !== '' && empty($channel_id));
4448 -
4449 - if (empty($channel_id) && !$use_shared_channel) {
4450 - $channel_id = $this->mxchat_create_conversation_channel($session_id);
4451 - if (empty($channel_id)) {
4452 - return false; // Failed to create channel
4453 - }
4454 - }
4455 -
4456 - // Get recent chat history
4457 - $history = get_option("mxchat_history_{$session_id}", []);
4458 - $recent_history = array_slice($history, -5);
4459 -
4460 - // Format conversation context
4461 - $conversation_context = "";
4462 - if (!empty($recent_history)) {
4463 - $conversation_context = "*Recent Conversation:*\n";
4464 - foreach ($recent_history as $hist_message) {
4465 - $role_display = $hist_message['role'] === 'user' ? 'User' : 'AI';
4466 - $conversation_context .= ">{$role_display}: {$hist_message['content']}\n";
4467 - }
4468 - $conversation_context .= "\n";
4469 - }
4470 -
4471 - update_option("mxchat_mode_{$session_id}", 'agent');
4472 -
4473 - // Send message to channel
4474 - $channel_message = "🔔 *New Live Agent Request*\n\n";
4475 - $channel_message .= "*Session ID:* `{$session_id}`\n";
4476 - $channel_message .= "*User ID:* `{$user_id}`\n";
4477 -
4478 - // Surface the captured visitor identity so the agent knows who they're talking to —
4479 - // guest User IDs are 0, but the pre-chat gate / login / transcript often has name+email (plan-e2195b).
4480 - $visitor = $this->mxchat_get_visitor_identity($session_id);
4481 - if (!empty($visitor['name']) && !empty($visitor['email'])) {
4482 - $channel_message .= "*Visitor:* {$visitor['name']} <{$visitor['email']}>\n";
4483 - } elseif (!empty($visitor['email'])) {
4484 - $channel_message .= "*Visitor:* <{$visitor['email']}>\n";
4485 - } elseif (!empty($visitor['name'])) {
4486 - $channel_message .= "*Visitor:* {$visitor['name']}\n";
4487 - }
4488 - $channel_message .= "\n";
4489 -
4490 - if (!empty($conversation_context)) {
4491 - $channel_message .= $conversation_context;
4492 - }
4493 -
4494 - $channel_message .= "*Current Message:*\n{$message}\n\n";
4495 - if ($use_shared_channel) {
4496 - $channel_message .= "_Reply in this thread - replies here go to the user. `!endchat` in this thread ends the chat._";
4497 - } else {
4498 - $channel_message .= "_Reply directly in this channel - all messages will go to the user_";
4499 - }
4500 -
4501 - if ($use_shared_channel) {
4502 - $posted = $this->mxchat_post_shared_handoff($session_id, $channel_message, $shared_thread_ts);
4503 - if (!$posted) {
4504 - // Shared channel unreachable (wrong name/ID, bot not invited,
4505 - // archived...). Fall back to the per-conversation flow so the
4506 - // visitor still reaches an agent; the settings page surfaces the
4507 - // recorded error to the admin.
4508 - $use_shared_channel = false;
4509 - $channel_id = $this->mxchat_create_conversation_channel($session_id);
4510 - if (empty($channel_id)) {
4511 - return false;
4512 - }
4513 - $channel_message = str_replace(
4514 - "_Reply in this thread - replies here go to the user. `!endchat` in this thread ends the chat._",
4515 - "_Reply directly in this channel - all messages will go to the user_",
4516 - $channel_message
4517 - );
4518 - }
4519 - }
4520 -
4521 - if (!$use_shared_channel) {
4522 - $handoff_post = wp_remote_post('https://slack.com/api/chat.postMessage', [
4523 - 'headers' => [
4524 - 'Content-Type' => 'application/json',
4525 - 'Authorization' => 'Bearer ' . $slack_bot_token
4526 - ],
4527 - 'body' => json_encode([
4528 - 'channel' => $channel_id,
4529 - 'text' => $channel_message,
4530 - 'mrkdwn' => true
4531 - ])
4532 - ]);
4533 - // Re-handover edge (plan 7458a7): the stored mxchat_channel_ may point
4534 - // at a channel archived by the auto-archive toggle (or deleted by an
4535 - // admin). Slack answers is_archived / channel_not_found — clear the
4536 - // stale option, mint a fresh channel, and re-post ONCE so the handoff
4537 - // is never silently dropped.
4538 - if (!is_wp_error($handoff_post)) {
4539 - $handoff_data = json_decode(wp_remote_retrieve_body($handoff_post), true);
4540 - $handoff_err = isset($handoff_data['error']) ? $handoff_data['error'] : '';
4541 - if (isset($handoff_data['ok']) && !$handoff_data['ok'] && in_array($handoff_err, array('is_archived', 'channel_not_found'), true)) {
4542 - delete_option("mxchat_channel_{$session_id}");
4543 - $channel_id = $this->mxchat_create_conversation_channel($session_id);
4544 - if (!empty($channel_id)) {
4545 - wp_remote_post('https://slack.com/api/chat.postMessage', [
4546 - 'headers' => [
4547 - 'Content-Type' => 'application/json',
4548 - 'Authorization' => 'Bearer ' . $slack_bot_token
4549 - ],
4550 - 'body' => json_encode([
4551 - 'channel' => $channel_id,
4552 - 'text' => $channel_message,
4553 - 'mrkdwn' => true
4554 - ])
4555 - ]);
4556 - }
4557 - }
4558 - }
4559 - }
4560 -
4561 - $success_message = $this->options['live_agent_notification_message'] ?? 'Live agent has been notified.';
4562 - $this->mxchat_save_chat_message($session_id, 'bot', $success_message);
4563 -
4564 - $this->fallbackResponse = [
4565 - 'text' => $success_message,
4566 - 'html' => '',
4567 - 'images' => [],
4568 - 'chat_mode' => 'agent'
4569 - ];
4570 -
4571 - wp_send_json([
4572 - 'success' => true,
4573 - 'text' => $success_message,
4574 - 'html' => '',
4575 - 'chat_mode' => 'agent',
4576 - 'session_id' => $session_id,
4577 - 'fallbackResponse' => $this->fallbackResponse
4578 - ]);
4579 - wp_die();
241 +private function mxchat_get_user_identifier() {
242 + return MxChat_User::mxchat_get_user_identifier();
4580 243 }
4581 244
4582 -/**
4583 - * Archive a session's per-conversation chat- channel after !endchat / session
4584 - * cleanup (plan 7458a7). HARD GUARDS, in order: the toggle must be on
4585 - * (default off = zero change for existing installs); a session with
4586 - * mxchat_thread_ set is a 9f7756 SHARED-channel session and is never
4587 - * archived; only the channel this session owns via mxchat_channel_ is
4588 - * archived, and only when it matches the channel the caller is acting on.
4589 - * Best-effort by design — a failed archive is logged and never blocks the
4590 - * mode flip or cleanup.
4591 - *
4592 - * @param string $session_id
4593 - * @param string $event_channel_id Channel the caller is acting on.
4594 - */
4595 -private function mxchat_maybe_archive_conversation_channel($session_id, $event_channel_id) {
4596 - $toggle = $this->options['live_agent_archive_on_end_toggle'] ?? 'off';
4597 - if ($toggle !== 'on') {
4598 - return;
4599 - }
4600 - if (get_option("mxchat_thread_{$session_id}", '') !== '') {
4601 - return; // shared-channel session — the shared channel is NEVER archived
4602 - }
4603 - $owned_channel = get_option("mxchat_channel_{$session_id}", '');
4604 - if ($owned_channel === '' || $owned_channel !== $event_channel_id) {
4605 - return;
4606 - }
4607 - $slack_bot_token = $this->options['live_agent_bot_token'] ?? '';
4608 - if (empty($slack_bot_token)) {
4609 - return;
4610 - }
4611 - $response = wp_remote_post('https://slack.com/api/conversations.archive', [
4612 - 'headers' => [
4613 - 'Content-Type' => 'application/json',
4614 - 'Authorization' => 'Bearer ' . $slack_bot_token
4615 - ],
4616 - 'body' => json_encode(['channel' => $owned_channel])
4617 - ]);
4618 - if (is_wp_error($response)) {
4619 - error_log('MxChat: conversations.archive request failed: ' . $response->get_error_message());
4620 - return;
4621 - }
4622 - $data = json_decode(wp_remote_retrieve_body($response), true);
4623 - if (empty($data['ok'])) {
4624 - error_log('MxChat: conversations.archive returned error: ' . (isset($data['error']) ? $data['error'] : 'unknown'));
4625 - }
4626 -}
4627 245
4628 -/**
4629 - * Create a dedicated per-conversation Slack channel for a session and invite
4630 - * the configured agents. Extracted from mxchat_live_agent_handover so the
4631 - * shared-channel mode (plan 9f7756) can reuse it as its fallback path.
4632 - *
4633 - * @param string $session_id
4634 - * @return string Channel ID, or '' on failure.
4635 - */
4636 -private function mxchat_create_conversation_channel($session_id) {
4637 - $slack_bot_token = $this->options['live_agent_bot_token'] ?? '';
4638 - if (empty($slack_bot_token)) {
4639 - return '';
4640 - }
4641 246
4642 - $channel_id = '';
4643 - $channel_name = $this->generate_channel_name($session_id);
247 + private function mxchat_generate_embedding($text, $api_key) {
248 + $endpoint = 'https://api.openai.com/v1/embeddings';
4644 249
4645 - $response = wp_remote_post('https://slack.com/api/conversations.create', [
4646 - 'headers' => [
4647 - 'Content-Type' => 'application/json',
4648 - 'Authorization' => 'Bearer ' . $slack_bot_token
4649 - ],
4650 - 'body' => json_encode([
4651 - 'name' => $channel_name,
4652 - 'is_private' => false // Public channel - anyone in workspace can join
4653 - ])
4654 - ]);
4655 -
4656 - if (!is_wp_error($response)) {
4657 - $response_data = json_decode(wp_remote_retrieve_body($response), true);
4658 -
4659 - if (isset($response_data['ok']) && $response_data['ok']) {
4660 - $channel_id = $response_data['channel']['id'];
4661 - update_option("mxchat_channel_{$session_id}", $channel_id);
4662 -
4663 - // Auto-invite agents to the channel
4664 - $agent_user_ids = $this->options['live_agent_user_ids'] ?? '';
4665 -
4666 - if (!empty($agent_user_ids)) {
4667 - // Parse user IDs (one per line)
4668 - $user_ids = array_filter(array_map('trim', explode("\n", $agent_user_ids)));
4669 -
4670 - foreach ($user_ids as $user_id_to_invite) {
4671 - wp_remote_post('https://slack.com/api/conversations.invite', [
4672 - 'headers' => [
4673 - 'Content-Type' => 'application/json',
4674 - 'Authorization' => 'Bearer ' . $slack_bot_token
4675 - ],
4676 - 'body' => json_encode([
4677 - 'channel' => $channel_id,
4678 - 'users' => $user_id_to_invite
4679 - ])
4680 - ]);
4681 - }
4682 - }
4683 - }
4684 - }
4685 -
4686 - return $channel_id;
4687 -}
4688 -
4689 -/**
4690 - * Post a handoff (or a re-handover) into the configured shared channel.
4691 - * First post per session becomes the conversation's thread root; its ts is
4692 - * stored in mxchat_thread_{session} and every later message rides that
4693 - * thread. Records the Slack error for the settings page on failure so the
4694 - * caller can fall back to per-conversation creation.
4695 - *
4696 - * @param string $session_id
4697 - * @param string $text Fully-built handoff message.
4698 - * @param string $thread_ts Existing thread root for this session, '' if none.
4699 - * @return bool True when the message reached the shared channel.
4700 - */
4701 -private function mxchat_post_shared_handoff($session_id, $text, $thread_ts = '') {
4702 - $slack_bot_token = $this->options['live_agent_bot_token'] ?? '';
4703 - $configured = trim($this->options['live_agent_shared_channel'] ?? '');
4704 - if (empty($slack_bot_token) || $configured === '') {
4705 - return false;
4706 - }
4707 -
4708 - // Posting by #name works once the bot is a member; the response carries
4709 - // the real channel ID, cached so the inbound webhook and user-relay
4710 - // don't depend on how the admin wrote the setting.
4711 - $cache = get_option('mxchat_slack_shared_channel_id', array());
4712 - $target = (is_array($cache) && ($cache['configured'] ?? '') === $configured && !empty($cache['id']))
4713 - ? $cache['id']
4714 - : ltrim($configured, '#');
4715 -
4716 - $body = [
4717 - 'channel' => $target,
4718 - 'text' => $text,
4719 - 'mrkdwn' => true
4720 - ];
4721 - if ($thread_ts !== '') {
4722 - $body['thread_ts'] = $thread_ts;
4723 - }
4724 -
4725 - $response = wp_remote_post('https://slack.com/api/chat.postMessage', [
4726 - 'headers' => [
4727 - 'Content-Type' => 'application/json',
4728 - 'Authorization' => 'Bearer ' . $slack_bot_token
4729 - ],
4730 - 'body' => json_encode($body)
4731 - ]);
4732 -
4733 - if (is_wp_error($response)) {
4734 - update_option('mxchat_slack_shared_channel_error', array(
4735 - 'error' => $response->get_error_message(),
4736 - 'configured' => $configured,
4737 - 'time' => time(),
4738 - ), false);
4739 - return false;
4740 - }
4741 -
4742 - $data = json_decode(wp_remote_retrieve_body($response), true);
4743 - if (empty($data['ok'])) {
4744 - update_option('mxchat_slack_shared_channel_error', array(
4745 - 'error' => $data['error'] ?? 'unknown_error',
4746 - 'configured' => $configured,
4747 - 'time' => time(),
4748 - ), false);
4749 - return false;
4750 - }
4751 -
4752 - delete_option('mxchat_slack_shared_channel_error');
4753 -
4754 - if (!empty($data['channel'])) {
4755 - update_option('mxchat_slack_shared_channel_id', array(
4756 - 'configured' => $configured,
4757 - 'id' => $data['channel'],
4758 - ), false);
4759 - }
4760 - if ($thread_ts === '' && !empty($data['ts'])) {
4761 - update_option("mxchat_thread_{$session_id}", $data['ts'], 'no');
4762 - }
4763 -
4764 - return true;
4765 -}
4766 -
4767 -private function generate_channel_name($session_id) {
4768 - $email = null;
4769 - $name = null;
4770 -
4771 - // 1. First priority: Check if user is logged in and get their info
4772 - if (is_user_logged_in()) {
4773 - $current_user = wp_get_current_user();
4774 - if (!empty($current_user->user_email)) {
4775 - $email = $current_user->user_email;
4776 - //error_log("[DEBUG] Using logged-in user email for channel: {$email}");
4777 - }
4778 - if (!empty($current_user->display_name)) {
4779 - $name = $current_user->display_name;
4780 - //error_log("[DEBUG] Using logged-in user name for channel: {$name}");
4781 - }
4782 - }
4783 -
4784 - // 2. Second priority: Check for saved email/name from "require email to chat" option
4785 - if (empty($email)) {
4786 - $email_option_key = "mxchat_email_{$session_id}";
4787 - $saved_email = get_option($email_option_key);
4788 - if (!empty($saved_email)) {
4789 - $email = $saved_email;
4790 - //error_log("[DEBUG] Using saved email from session for channel: {$email}");
4791 - }
4792 - }
4793 -
4794 - if (empty($name)) {
4795 - $name_option_key = "mxchat_name_{$session_id}";
4796 - $saved_name = get_option($name_option_key);
4797 - if (!empty($saved_name)) {
4798 - $name = $saved_name;
4799 - //error_log("[DEBUG] Using saved name from session for channel: {$name}");
4800 - }
4801 - }
4802 -
4803 - // 3. Third priority: Check existing chat transcript for email/name
4804 - if (empty($email) || empty($name)) {
4805 - global $wpdb;
4806 - $table_name = $wpdb->prefix . 'mxchat_chat_transcripts';
4807 - $existing_data = $wpdb->get_row($wpdb->prepare(
4808 - "SELECT user_email, user_name FROM $table_name WHERE session_id = %s AND (user_email IS NOT NULL OR user_name IS NOT NULL) LIMIT 1",
4809 - $session_id
4810 - ));
4811 -
4812 - if ($existing_data) {
4813 - if (empty($email) && !empty($existing_data->user_email)) {
4814 - $email = $existing_data->user_email;
4815 - //error_log("[DEBUG] Using email from chat transcript for channel: {$email}");
4816 - }
4817 - if (empty($name) && !empty($existing_data->user_name)) {
4818 - $name = $existing_data->user_name;
4819 - //error_log("[DEBUG] Using name from chat transcript for channel: {$name}");
4820 - }
4821 - }
4822 - }
4823 -
4824 - // 4. Generate channel name based on priority: Name > Email > Session ID
4825 - $channel_name = '';
4826 -
4827 - if (!empty($name)) {
4828 - // Convert name to valid Slack channel name
4829 - $base_name = strtolower(trim($name));
4830 - // Replace spaces and invalid characters
4831 - $base_name = preg_replace('/[^a-z0-9\s]/', '', $base_name);
4832 - $base_name = preg_replace('/\s+/', '-', $base_name);
4833 - $base_name = trim($base_name, '-');
4834 -
4835 - // Get last 4 characters of session ID for uniqueness
4836 - $session_suffix = substr($session_id, -4);
4837 - $channel_name = 'chat-' . $base_name . '-' . strtolower($session_suffix);
4838 -
4839 - // Slack channel names have a 21 character limit
4840 - if (strlen($channel_name) > 21) {
4841 - // Calculate available space for name (21 - 'chat-' - '-' - session_suffix)
4842 - $available_space = 21 - 5 - 1 - strlen($session_suffix); // 'chat-' = 5, '-' = 1
4843 - $truncated_name = substr($base_name, 0, $available_space);
4844 - $truncated_name = rtrim($truncated_name, '-'); // Remove trailing hyphen
4845 - $channel_name = 'chat-' . $truncated_name . '-' . strtolower($session_suffix);
4846 - }
4847 -
4848 - //error_log("[DEBUG] Using name for channel: {$channel_name} (from name: {$name})");
4849 -
4850 - } elseif (!empty($email)) {
4851 - // Convert email to valid Slack channel name (your existing logic)
4852 - $channel_name = 'chat-' . strtolower(str_replace(['@', '.', '+', '_'], ['-at-', '-', '-plus-', '-'], $email));
4853 - // Remove any remaining invalid characters
4854 - $channel_name = preg_replace('/[^a-z0-9\-]/', '', $channel_name);
4855 - // Ensure it doesn't end with a hyphen
4856 - $channel_name = rtrim($channel_name, '-');
4857 - // Slack channel names have a 21 character limit, so truncate if needed
4858 - if (strlen($channel_name) > 21) {
4859 - $channel_name = substr($channel_name, 0, 21);
4860 - $channel_name = rtrim($channel_name, '-'); // Remove trailing hyphen if truncation created one
4861 - }
4862 -
4863 - //error_log("[DEBUG] Using email for channel: {$channel_name} (from email: {$email})");
4864 -
4865 - } else {
4866 - // Fallback to session ID if no name or email found
4867 - $channel_name = 'chat-' . strtolower(preg_replace('/[^a-zA-Z0-9]/', '', $session_id));
4868 - //error_log("[DEBUG] No name or email found, using session ID for channel: {$channel_name}");
4869 - }
4870 -
4871 - // Final validation - ensure channel name meets Slack requirements
4872 - if (strlen($channel_name) > 21) {
4873 - $channel_name = substr($channel_name, 0, 21);
4874 - $channel_name = rtrim($channel_name, '-');
4875 - }
4876 -
4877 - //error_log("[DEBUG] Generated channel name: {$channel_name}");
4878 - return $channel_name;
4879 -}
4880 -
4881 -/**
4882 - * Telegram Live Agent Handover
4883 - * Creates a forum topic in the Telegram group and notifies agents
4884 - */
4885 -public function mxchat_telegram_live_agent_handover($message, $user_id, $session_id) {
4886 - // Check if Telegram agents are available. Telegram has its OWN availability
4887 - // schedule, independent of Slack's (plan 99d7a4 — each Integrations tab
4888 - // owns its scheduler). Backstop only; the tool is normally withheld
4889 - // off-hours.
4890 - $telegram_available = $this->options['telegram_status'] ?? 'off';
4891 - $within_hours = !class_exists('MxChat_Live_Agent_Schedule')
4892 - || MxChat_Live_Agent_Schedule::is_within_hours('telegram');
4893 - if ($telegram_available !== 'on' || !$within_hours) {
4894 - $away_message = $this->options['telegram_away_message'] ?? 'Sorry, live agents are currently unavailable. I can continue helping you as an AI assistant.';
4895 - $this->fallbackResponse = [
4896 - 'text' => $away_message,
4897 - 'html' => '',
4898 - 'images' => [],
4899 - 'chat_mode' => 'ai'
4900 - ];
4901 - wp_send_json([
4902 - 'text' => $away_message,
4903 - 'html' => '',
4904 - 'chat_mode' => 'ai',
4905 - 'session_id' => $session_id
250 + $body = wp_json_encode([
251 + 'input' => $text,
252 + 'model' => 'text-embedding-ada-002'
4906 253 ]);
4907 - wp_die();
4908 - }
4909 254
4910 - $telegram_bot_token = $this->options['telegram_bot_token'] ?? '';
4911 - $telegram_group_id = $this->options['telegram_group_id'] ?? '';
4912 -
4913 - if (empty($telegram_bot_token) || empty($telegram_group_id)) {
4914 - return false;
4915 - }
4916 -
4917 - // Check if topic already exists for this session
4918 - $topic_id = get_option("mxchat_telegram_topic_{$session_id}", '');
4919 -
4920 - if (empty($topic_id)) {
4921 - // Generate topic name
4922 - $topic_name = $this->generate_telegram_topic_name($session_id);
4923 -
4924 - // Random icon color (Telegram forum topic colors)
4925 - $icon_colors = [0x6FB9F0, 0xFFD67E, 0xCB86DB, 0x8EEE98, 0xFF93B2, 0xFB6F5F];
4926 - $icon_color = $icon_colors[array_rand($icon_colors)];
4927 -
4928 - // Create forum topic
4929 - $response = wp_remote_post("https://api.telegram.org/bot{$telegram_bot_token}/createForumTopic", [
4930 - 'headers' => ['Content-Type' => 'application/json'],
4931 - 'body' => json_encode([
4932 - 'chat_id' => $telegram_group_id,
4933 - 'name' => $topic_name,
4934 - 'icon_color' => $icon_color
4935 - ])
4936 - ]);
4937 -
4938 - if (!is_wp_error($response)) {
4939 - $response_body = wp_remote_retrieve_body($response);
4940 - $response_data = json_decode($response_body, true);
4941 -
4942 - if (isset($response_data['ok']) && $response_data['ok']) {
4943 - $topic_id = $response_data['result']['message_thread_id'];
4944 - update_option("mxchat_telegram_topic_{$session_id}", $topic_id);
4945 - update_option("mxchat_telegram_group_{$session_id}", $telegram_group_id);
4946 - }
4947 - }
4948 -
4949 - if (empty($topic_id)) {
4950 - return false; // Failed to create topic
4951 - }
4952 - }
4953 -
4954 - // Get recent chat history
4955 - $history = get_option("mxchat_history_{$session_id}", []);
4956 - $recent_history = array_slice($history, -5);
4957 -
4958 - // Format conversation context for Telegram (HTML format)
4959 - $conversation_context = "";
4960 - if (!empty($recent_history)) {
4961 - $conversation_context = "<b>Recent Conversation:</b>\n";
4962 - foreach ($recent_history as $hist_message) {
4963 - $role_display = $hist_message['role'] === 'user' ? '👤 User' : '🤖 AI';
4964 - $escaped_content = htmlspecialchars($hist_message['content'], ENT_QUOTES, 'UTF-8');
4965 - $conversation_context .= "{$role_display}: {$escaped_content}\n";
4966 - }
4967 - $conversation_context .= "\n";
4968 - }
4969 -
4970 - // Get user info
4971 - $user_email = get_option("mxchat_email_{$session_id}", 'Not provided');
4972 - $user_name = get_option("mxchat_name_{$session_id}", 'Anonymous');
4973 -
4974 - // Update session mode
4975 - update_option("mxchat_mode_{$session_id}", 'agent');
4976 -
4977 - // Send initial message to topic
4978 - $escaped_message = htmlspecialchars($message, ENT_QUOTES, 'UTF-8');
4979 - $topic_message = "🔔 <b>New Live Agent Request</b>\n\n";
4980 - $topic_message .= "<b>Session ID:</b> <code>{$session_id}</code>\n";
4981 - $topic_message .= "<b>User:</b> {$user_name}\n";
4982 - $topic_message .= "<b>Email:</b> {$user_email}\n\n";
4983 -
4984 - if (!empty($conversation_context)) {
4985 - $topic_message .= $conversation_context;
4986 - }
4987 -
4988 - $topic_message .= "<b>Current Message:</b>\n{$escaped_message}\n\n";
4989 - $topic_message .= "<i>Reply in this topic - messages will be sent to the user</i>\n";
4990 - $topic_message .= "<i>Type #close, #end, #disconnect, or #done to end the session</i>";
4991 -
4992 - wp_remote_post("https://api.telegram.org/bot{$telegram_bot_token}/sendMessage", [
4993 - 'headers' => ['Content-Type' => 'application/json'],
4994 - 'body' => json_encode([
4995 - 'chat_id' => $telegram_group_id,
4996 - 'message_thread_id' => $topic_id,
4997 - 'text' => $topic_message,
4998 - 'parse_mode' => 'HTML'
4999 - ])
5000 - ]);
5001 -
5002 - $success_message = $this->options['telegram_notification_message'] ?? "I've notified a support agent. Please allow a moment for them to respond.";
5003 - $this->mxchat_save_chat_message($session_id, 'bot', $success_message);
5004 -
5005 - $this->fallbackResponse = [
5006 - 'text' => $success_message,
5007 - 'html' => '',
5008 - 'images' => [],
5009 - 'chat_mode' => 'agent'
5010 - ];
5011 -
5012 - wp_send_json([
5013 - 'success' => true,
5014 - 'text' => $success_message,
5015 - 'html' => '',
5016 - 'chat_mode' => 'agent',
5017 - 'session_id' => $session_id,
5018 - 'fallbackResponse' => $this->fallbackResponse
5019 - ]);
5020 - wp_die();
5021 -}
5022 -
5023 -/**
5024 - * Generate topic name for Telegram forum
5025 - */
5026 -private function generate_telegram_topic_name($session_id) {
5027 - $name = null;
5028 - $email = null;
5029 -
5030 - // Check logged in user
5031 - if (is_user_logged_in()) {
5032 - $current_user = wp_get_current_user();
5033 - if (!empty($current_user->display_name)) {
5034 - $name = $current_user->display_name;
5035 - }
5036 - if (!empty($current_user->user_email)) {
5037 - $email = $current_user->user_email;
5038 - }
5039 - }
5040 -
5041 - // Check session data
5042 - if (empty($name)) {
5043 - $name = get_option("mxchat_name_{$session_id}");
5044 - }
5045 - if (empty($email)) {
5046 - $email = get_option("mxchat_email_{$session_id}");
5047 - }
5048 -
5049 - // Generate topic name
5050 - $session_suffix = substr($session_id, -6);
5051 -
5052 - if (!empty($name)) {
5053 - // Clean name for topic (max 128 chars in Telegram)
5054 - $clean_name = preg_replace('/[^\p{L}\p{N}\s\-]/u', '', $name);
5055 - $clean_name = trim($clean_name);
5056 - if (strlen($clean_name) > 50) {
5057 - $clean_name = substr($clean_name, 0, 50);
5058 - }
5059 - return "Chat - {$clean_name} ({$session_suffix})";
5060 - } elseif (!empty($email)) {
5061 - // Use email prefix
5062 - $email_prefix = explode('@', $email)[0];
5063 - if (strlen($email_prefix) > 30) {
5064 - $email_prefix = substr($email_prefix, 0, 30);
5065 - }
5066 - return "Chat - {$email_prefix} ({$session_suffix})";
5067 - }
5068 -
5069 - return "Chat - {$session_suffix}";
5070 -}
5071 -
5072 -/**
5073 - * Send user message to Telegram agent
5074 - */
5075 -public function mxchat_send_user_message_to_telegram_agent($message, $user_id, $session_id) {
5076 - $telegram_bot_token = $this->options['telegram_bot_token'] ?? '';
5077 - $topic_id = get_option("mxchat_telegram_topic_{$session_id}", '');
5078 - $group_id = get_option("mxchat_telegram_group_{$session_id}", '');
5079 -
5080 - if (empty($telegram_bot_token) || empty($topic_id) || empty($group_id)) {
5081 - return false;
5082 - }
5083 -
5084 - $escaped_message = htmlspecialchars($message, ENT_QUOTES, 'UTF-8');
5085 - $user_message = "👤 <b>User:</b> {$escaped_message}";
5086 -
5087 - $response = wp_remote_post("https://api.telegram.org/bot{$telegram_bot_token}/sendMessage", [
5088 - 'headers' => ['Content-Type' => 'application/json'],
5089 - 'body' => json_encode([
5090 - 'chat_id' => $group_id,
5091 - 'message_thread_id' => $topic_id,
5092 - 'text' => $user_message,
5093 - 'parse_mode' => 'HTML'
5094 - ])
5095 - ]);
5096 -
5097 - return !is_wp_error($response);
5098 -}
5099 -
5100 -/**
5101 - * Handle incoming Telegram webhook
5102 - */
5103 -public function handle_telegram_webhook(WP_REST_Request $request) {
5104 - $body = $request->get_body();
5105 - $data = json_decode($body, true);
5106 -
5107 - //error_log('[MxChat Telegram DEBUG] Webhook received: ' . $body);
5108 -
5109 - // Handle message events from forum topics
5110 - if (isset($data['message'])) {
5111 - $message_data = $data['message'];
5112 -
5113 - // Skip if not from a forum topic
5114 - if (!isset($message_data['message_thread_id'])) {
5115 - //error_log('[MxChat Telegram DEBUG] Skipped: No message_thread_id (not a forum topic message)');
5116 - return new WP_REST_Response(['ok' => true]);
5117 - }
5118 -
5119 - // Skip bot messages
5120 - if (isset($message_data['from']['is_bot']) && $message_data['from']['is_bot']) {
5121 - //error_log('[MxChat Telegram DEBUG] Skipped: Message from bot');
5122 - return new WP_REST_Response(['ok' => true]);
5123 - }
5124 -
5125 - $chat_id = $message_data['chat']['id'] ?? '';
5126 - $topic_id = $message_data['message_thread_id'];
5127 - $message_text = $message_data['text'] ?? '';
5128 - $message_id = $message_data['message_id'] ?? '';
5129 - $from = $message_data['from'] ?? [];
5130 - $agent_name = trim(($from['first_name'] ?? '') . ' ' . ($from['last_name'] ?? ''));
5131 - if (empty($agent_name)) {
5132 - $agent_name = $from['username'] ?? 'Agent';
5133 - }
5134 -
5135 - //error_log("[MxChat Telegram DEBUG] Parsed: chat_id={$chat_id}, topic_id={$topic_id}, agent={$agent_name}, text={$message_text}");
5136 -
5137 - // Skip empty messages
5138 - if (empty($message_text)) {
5139 - //error_log('[MxChat Telegram DEBUG] Skipped: Empty message text');
5140 - return new WP_REST_Response(['ok' => true]);
5141 - }
5142 -
5143 - // Find session ID by topic ID - cast to string for comparison
5144 - global $wpdb;
5145 - $topic_id_str = strval($topic_id);
5146 - $session_option = $wpdb->get_var(
5147 - $wpdb->prepare(
5148 - "SELECT option_name FROM {$wpdb->options}
5149 - WHERE option_name LIKE %s
5150 - AND option_value = %s",
5151 - 'mxchat_telegram_topic_%',
5152 - $topic_id_str
5153 - )
5154 - );
5155 -
5156 - //error_log("[MxChat Telegram DEBUG] Looking for topic_id={$topic_id_str} in options, found: " . ($session_option ?: 'NULL'));
5157 -
5158 - if ($session_option) {
5159 - $session_id = str_replace('mxchat_telegram_topic_', '', $session_option);
5160 - //error_log("[MxChat Telegram DEBUG] Session ID: {$session_id}");
5161 -
5162 - // Verify the group ID matches
5163 - $stored_group_id = get_option("mxchat_telegram_group_{$session_id}", '');
5164 - //error_log("[MxChat Telegram DEBUG] Stored group_id={$stored_group_id}, received chat_id={$chat_id}");
5165 -
5166 - if (strval($stored_group_id) != strval($chat_id)) {
5167 - //error_log('[MxChat Telegram DEBUG] Skipped: Group ID mismatch');
5168 - return new WP_REST_Response(['ok' => true]);
5169 - }
5170 -
5171 - // Check for closure commands
5172 - $lower_text = strtolower(trim($message_text));
5173 - if (in_array($lower_text, ['#close', '#end', '#disconnect', '#done'])) {
5174 - //error_log("[MxChat Telegram DEBUG] Closure command received: {$lower_text}");
5175 - // End the live agent session
5176 - update_option("mxchat_mode_{$session_id}", 'ai');
5177 -
5178 - // Save disconnect message
5179 - $disconnect_message = "Live agent session ended. You're now chatting with the AI assistant.";
5180 - $this->mxchat_save_chat_message($session_id, 'bot', $disconnect_message);
5181 -
5182 - // Notify in Telegram
5183 - $telegram_bot_token = $this->options['telegram_bot_token'] ?? '';
5184 - if (!empty($telegram_bot_token)) {
5185 - wp_remote_post("https://api.telegram.org/bot{$telegram_bot_token}/sendMessage", [
5186 - 'headers' => ['Content-Type' => 'application/json'],
5187 - 'body' => json_encode([
5188 - 'chat_id' => $chat_id,
5189 - 'message_thread_id' => $topic_id,
5190 - 'text' => "✅ Session closed. User returned to AI chatbot.",
5191 - 'parse_mode' => 'HTML'
5192 - ])
5193 - ]);
5194 -
5195 - // Optionally close the topic
5196 - wp_remote_post("https://api.telegram.org/bot{$telegram_bot_token}/closeForumTopic", [
5197 - 'headers' => ['Content-Type' => 'application/json'],
5198 - 'body' => json_encode([
5199 - 'chat_id' => $chat_id,
5200 - 'message_thread_id' => $topic_id
5201 - ])
5202 - ]);
5203 - }
5204 -
5205 - return new WP_REST_Response(['ok' => true]);
5206 - }
5207 -
5208 - // Deduplicate messages
5209 - $message_key = md5($session_id . $message_id . $message_text);
5210 - $processed_messages = get_transient('mxchat_telegram_messages_' . $session_id) ?: [];
5211 -
5212 - if (in_array($message_key, $processed_messages)) {
5213 - //error_log('[MxChat Telegram DEBUG] Skipped: Duplicate message');
5214 - return new WP_REST_Response(['ok' => true]);
5215 - }
5216 -
5217 - $processed_messages[] = $message_key;
5218 - if (count($processed_messages) > 50) {
5219 - $processed_messages = array_slice($processed_messages, -50);
5220 - }
5221 - set_transient('mxchat_telegram_messages_' . $session_id, $processed_messages, HOUR_IN_SECONDS);
5222 -
5223 - // Save the agent message - format with agent name prefix for proper parsing
5224 - $formatted_message = "Agent: {$agent_name} - {$message_text}";
5225 - //error_log("[MxChat Telegram DEBUG] Saving agent message: {$formatted_message}");
5226 -
5227 - $this->mxchat_save_chat_message($session_id, 'agent', $formatted_message);
5228 -
5229 - // Verify the message was saved to history
5230 - $history = get_option("mxchat_history_{$session_id}", []);
5231 - $last_message = end($history);
5232 - //error_log("[MxChat Telegram DEBUG] History after save - count: " . count($history) . ", last message role: " . ($last_message['role'] ?? 'none'));
5233 -
5234 - // Send confirmation back to Telegram
5235 - $telegram_bot_token = $this->options['telegram_bot_token'] ?? '';
5236 - if (!empty($telegram_bot_token)) {
5237 - $confirm_key = 'mxchat_telegram_confirm_' . $message_key;
5238 - if (!get_transient($confirm_key)) {
5239 - wp_remote_post("https://api.telegram.org/bot{$telegram_bot_token}/sendMessage", [
5240 - 'headers' => ['Content-Type' => 'application/json'],
5241 - 'body' => json_encode([
5242 - 'chat_id' => $chat_id,
5243 - 'message_thread_id' => $topic_id,
5244 - 'text' => "✅ <i>Message sent to user</i>",
5245 - 'parse_mode' => 'HTML',
5246 - 'reply_to_message_id' => $message_id
5247 - ])
5248 - ]);
5249 - set_transient($confirm_key, true, 300);
5250 - }
5251 - }
5252 - } else {
5253 - //error_log("[MxChat Telegram DEBUG] No session found for topic_id={$topic_id}");
5254 - }
5255 - } else {
5256 - //error_log('[MxChat Telegram DEBUG] No message in webhook data');
5257 - }
5258 -
5259 - return new WP_REST_Response(['ok' => true]);
5260 -}
5261 -
5262 -public function mxchat_send_user_message_to_agent($message, $user_id, $session_id) {
5263 - // Check if this is a Telegram agent session
5264 - $telegram_topic_id = get_option("mxchat_telegram_topic_{$session_id}", '');
5265 - if (!empty($telegram_topic_id)) {
5266 - return $this->mxchat_send_user_message_to_telegram_agent($message, $user_id, $session_id);
5267 - }
5268 -
5269 - // Otherwise, try Slack
5270 - $slack_bot_token = $this->options['live_agent_bot_token'] ?? '';
5271 -
5272 - // Shared-channel session: the conversation lives in a thread of the
5273 - // shared channel (plan 9f7756); relay user messages into that thread.
5274 - $thread_ts = get_option("mxchat_thread_{$session_id}", '');
5275 - if (!empty($thread_ts)) {
5276 - $cache = get_option('mxchat_slack_shared_channel_id', array());
5277 - $channel_id = is_array($cache) ? ($cache['id'] ?? '') : '';
5278 - } else {
5279 - $channel_id = get_option("mxchat_channel_{$session_id}", '');
5280 - }
5281 -
5282 - if (empty($slack_bot_token) || empty($channel_id)) {
5283 - return false;
5284 - }
5285 -
5286 - $user_message = "💬 *User:* {$message}";
5287 -
5288 - $body = [
5289 - 'channel' => $channel_id,
5290 - 'text' => $user_message,
5291 - 'mrkdwn' => true
5292 - ];
5293 - if (!empty($thread_ts)) {
5294 - $body['thread_ts'] = $thread_ts;
5295 - }
5296 -
5297 - $response = wp_remote_post('https://slack.com/api/chat.postMessage', [
5298 - 'headers' => [
5299 - 'Content-Type' => 'application/json',
5300 - 'Authorization' => 'Bearer ' . $slack_bot_token
5301 - ],
5302 - 'body' => json_encode($body)
5303 - ]);
5304 -
5305 - return !is_wp_error($response);
5306 -}
5307 -public function handle_slack_interaction(WP_REST_Request $request) {
5308 - //error_log('Received Slack interaction');
5309 -
5310 - $payload = json_decode($request->get_param('payload'), true);
5311 - //error_log('Payload: ' . print_r($payload, true));
5312 -
5313 - // Handle button click
5314 - if ($payload['type'] === 'block_actions' && $payload['actions'][0]['action_id'] === 'reply_to_user') {
5315 - $session_id = $payload['actions'][0]['value'];
5316 - $trigger_id = $payload['trigger_id'];
5317 -
5318 - // Get Bot Token from settings
5319 - $slack_token = $this->options['live_agent_bot_token'] ?? '';
5320 -
5321 - if (empty($slack_token)) {
5322 - //error_log('Slack Bot Token not configured');
5323 - return new WP_REST_Response(['error' => esc_html__('Bot token not configured', 'mxchat')], 400);
5324 - }
5325 - $response = wp_remote_post('https://slack.com/api/views.open', [
255 + $args = [
256 + 'body' => $body,
5326 257 'headers' => [
5327 258 'Content-Type' => 'application/json',
5328 - 'Authorization' => 'Bearer ' . $slack_token
259 + 'Authorization' => 'Bearer ' . $api_key,
5329 260 ],
5330 - 'body' => json_encode([
5331 - 'trigger_id' => $trigger_id,
5332 - 'view' => [
5333 - 'type' => 'modal',
5334 - 'callback_id' => 'reply_modal',
5335 - 'title' => [
5336 - 'type' => 'plain_text',
5337 - 'text' => __('Reply to User', 'mxchat')
5338 - ],
5339 - 'submit' => [
5340 - 'type' => 'plain_text',
5341 - 'text' => __('Send', 'mxchat')
5342 - ],
5343 - 'close' => [
5344 - 'type' => 'plain_text',
5345 - 'text' => __('Cancel', 'mxchat')
5346 - ],
5347 - 'blocks' => [
5348 - [
5349 - 'type' => 'input',
5350 - 'block_id' => 'reply_block',
5351 - 'label' => [
5352 - 'type' => 'plain_text',
5353 - 'text' => sprintf(__('Reply to session: %s', 'mxchat'), $session_id)
5354 - ],
5355 - 'element' => [
5356 - 'type' => 'plain_text_input',
5357 - 'action_id' => 'message',
5358 - 'multiline' => true,
5359 - 'placeholder' => [
5360 - 'type' => 'plain_text',
5361 - 'text' => __('Type your message here...', 'mxchat')
5362 - ]
5363 - ]
5364 - ]
5365 - ],
5366 - 'private_metadata' => $session_id
5367 - ]
5368 - ])
5369 - ]);
5370 -
5371 - //error_log('Views.open response: ' . print_r($response, true));
5372 -
5373 - // Return immediate acknowledgment
5374 - return new WP_REST_Response(['ok' => true]);
5375 - }
5376 -
5377 - // Handle modal submission
5378 -// Handle modal submission
5379 -if ($payload['type'] === 'view_submission') {
5380 - $session_id = $payload['view']['private_metadata'];
5381 - $message = $payload['view']['state']['values']['reply_block']['message']['value'];
5382 -
5383 - // Save the message (keep the message_id but don't include in response)
5384 - $this->mxchat_save_chat_message($session_id, 'agent', $message);
5385 -
5386 - // Keep the original response format for Slack
5387 - return new WP_REST_Response([
5388 - 'response_action' => 'clear'
5389 - ]);
5390 -}
5391 -
5392 - // Default acknowledgment
5393 - return new WP_REST_Response(['ok' => true]);
5394 -}
5395 -public function mxchat_handle_agent_response(WP_REST_Request $request) {
5396 - //error_log('Received agent response request');
5397 - //error_log('Request data: ' . print_r($request->get_params(), true));
5398 - // //error_log('Raw body: ' . file_get_contents('php://input'));
5399 -
5400 - // Get the data from Slack's slash command format
5401 - $command_text = $request->get_param('text');
5402 - // //error_log('Command text: ' . $command_text);
5403 -
5404 - if (empty($command_text)) {
5405 - //error_log(esc_html__('Agent response error: No command text received', 'mxchat'));
5406 - return new WP_REST_Response([
5407 - 'error' => esc_html__('Command text is required. Format: /reply session_id message', 'mxchat')
5408 - ], 400);
5409 - }
5410 -
5411 - // Split the command text into session_id and message
5412 - $parts = explode(' ', $command_text, 2);
5413 - if (count($parts) !== 2) {
5414 - //error_log('Agent response error: Invalid command format');
5415 - return new WP_REST_Response([
5416 - 'error' => esc_html__('Invalid format. Use: /reply session_id message', 'mxchat')
5417 - ], 400);
5418 - }
5419 -
5420 - $session_id = sanitize_text_field($parts[0]);
5421 - $message = sanitize_text_field($parts[1]);
5422 -
5423 - //error_log("Processing agent response - Session ID: $session_id, Message: $message");
5424 -
5425 - // Save the message
5426 - $message_id = $this->mxchat_save_chat_message($session_id, 'agent', $message);
5427 -
5428 - if (!$message_id) {
5429 - // //error_log('Failed to save agent message');
5430 - return new WP_REST_Response([
5431 - 'error' => esc_html__('Failed to save message', 'mxchat')
5432 - ], 500);
5433 - }
5434 -
5435 - // Return success response in Slack's expected format
5436 - return new WP_REST_Response([
5437 - 'response_type' => 'in_channel',
5438 - 'text' => esc_html__("Message sent successfully to session $session_id", 'mxchat')
5439 - ], 200);
5440 -}
5441 -public function mxchat_handle_switch_to_chatbot_intent($message, $user_id, $session_id) {
5442 - // Update mode to AI
5443 - update_option("mxchat_mode_{$session_id}", 'ai');
5444 -
5445 - // Clear any existing PDF context to start fresh
5446 - $this->clear_pdf_transients($session_id);
5447 -
5448 - // Set the response with explicit chat_mode
5449 - $this->fallbackResponse = [
5450 - 'text' => esc_html__('You are now chatting with the AI chatbot.', 'mxchat'),
5451 - 'html' => '',
5452 - 'images' => [],
5453 - 'chat_mode' => 'ai' // Ensure this is set
5454 - ];
5455 -
5456 - // Return the complete response array instead of just true
5457 - return $this->fallbackResponse;
5458 -}
5459 -
5460 -/**
5461 - * Normalize Slack mrkdwn before relaying an agent's message to the web visitor.
5462 - * Slack's Events API auto-wraps URLs as <https://url> or <https://url|Label>, wraps
5463 - * mentions as <@U…>/<#C…|name>, and HTML-escapes &, <, >. Relayed raw, the visitor
5464 - * sees a broken/doubled link with a trailing > (plan-e2195b). Unwrap links FIRST, then
5465 - * unescape entities LAST so extracted URLs (which can contain &amp;) are not corrupted.
5466 - */
5467 -private function normalize_slack_text($text) {
5468 - if (!is_string($text) || $text === '') {
5469 - return $text;
5470 - }
5471 -
5472 - $text = preg_replace_callback('/<([^>|]+)(?:\|([^>]*))?>/', function ($m) {
5473 - $target = $m[1];
5474 - $label = isset($m[2]) ? $m[2] : '';
5475 -
5476 - // User/channel mentions: <@U…> or <#C…|name> — prefer the human label, else drop the id.
5477 - if (isset($target[0]) && ($target[0] === '@' || $target[0] === '#')) {
5478 - return $label !== '' ? $label : '';
5479 - }
5480 - // mailto:/tel: — strip the scheme for display.
5481 - if (stripos($target, 'mailto:') === 0) {
5482 - $addr = substr($target, 7);
5483 - return ($label !== '' && $label !== $addr) ? "{$label} ({$addr})" : $addr;
5484 - }
5485 - if (stripos($target, 'tel:') === 0) {
5486 - $num = substr($target, 4);
5487 - return ($label !== '' && $label !== $num) ? "{$label} ({$num})" : $num;
5488 - }
5489 - // Regular URL: <url|Label> -> "Label (url)"; bare <url> -> "url".
5490 - if ($label !== '' && $label !== $target) {
5491 - return "{$label} ({$target})";
5492 - }
5493 - return $target;
5494 - }, $text);
5495 -
5496 - // Entity-unescape LAST (after link extraction) so &amp; inside URLs is repaired too.
5497 - $text = str_replace(array('&amp;', '&lt;', '&gt;'), array('&', '<', '>'), $text);
5498 -
5499 - return $text;
5500 -}
5501 -
5502 -/**
5503 - * Resolve the visitor's name + email for a session, mirroring generate_channel_name()'s
5504 - * priority order: logged-in user, then the pre-chat gate options (mxchat_email_/mxchat_name_),
5505 - * then the chat transcript. Returns ['name' => ..., 'email' => ...] (either may be ''). plan-e2195b.
5506 - */
5507 -private function mxchat_get_visitor_identity($session_id) {
5508 - $email = '';
5509 - $name = '';
5510 -
5511 - if (is_user_logged_in()) {
5512 - $current_user = wp_get_current_user();
5513 - if (!empty($current_user->user_email)) { $email = $current_user->user_email; }
5514 - if (!empty($current_user->display_name)) { $name = $current_user->display_name; }
5515 - }
5516 -
5517 - if (empty($email)) {
5518 - $saved_email = get_option("mxchat_email_{$session_id}", '');
5519 - if (!empty($saved_email)) { $email = $saved_email; }
5520 - }
5521 - if (empty($name)) {
5522 - $saved_name = get_option("mxchat_name_{$session_id}", '');
5523 - if (!empty($saved_name)) { $name = $saved_name; }
5524 - }
5525 -
5526 - if (empty($email) || empty($name)) {
5527 - global $wpdb;
5528 - $table_name = $wpdb->prefix . 'mxchat_chat_transcripts';
5529 - $existing_data = $wpdb->get_row($wpdb->prepare(
5530 - "SELECT user_email, user_name FROM $table_name WHERE session_id = %s AND (user_email IS NOT NULL OR user_name IS NOT NULL) LIMIT 1",
5531 - $session_id
5532 - ));
5533 - if ($existing_data) {
5534 - if (empty($email) && !empty($existing_data->user_email)) { $email = $existing_data->user_email; }
5535 - if (empty($name) && !empty($existing_data->user_name)) { $name = $existing_data->user_name; }
5536 - }
5537 - }
5538 -
5539 - return array('name' => $name, 'email' => $email);
5540 -}
5541 -
5542 -public function handle_slack_messages(WP_REST_Request $request) {
5543 - // Log the incoming request for debugging
5544 - //error_log('Slack events request received: ' . $request->get_body());
5545 -
5546 - $body = $request->get_body();
5547 - $data = json_decode($body, true);
5548 -
5549 - // Handle Slack URL verification
5550 - if (isset($data['type']) && $data['type'] === 'url_verification') {
5551 - //error_log('Slack URL verification challenge: ' . $data['challenge']);
5552 - return new WP_REST_Response($data['challenge'], 200, ['Content-Type' => 'text/plain']);
5553 - }
5554 -
5555 - // IMPORTANT: Handle Slack's event deduplication
5556 - if (isset($data['event_id'])) {
5557 - $event_id = $data['event_id'];
5558 - $processed_events = get_transient('mxchat_slack_events') ?: [];
5559 -
5560 - // Check if we've already processed this event
5561 - if (in_array($event_id, $processed_events)) {
5562 - //error_log("Duplicate event detected: $event_id");
5563 - return new WP_REST_Response(['ok' => true]);
5564 - }
5565 -
5566 - // Add this event to processed list
5567 - $processed_events[] = $event_id;
5568 - // Keep only last 100 events to prevent memory issues
5569 - if (count($processed_events) > 100) {
5570 - $processed_events = array_slice($processed_events, -100);
5571 - }
5572 - // Store for 1 hour
5573 - set_transient('mxchat_slack_events', $processed_events, HOUR_IN_SECONDS);
5574 - }
5575 -
5576 - // Handle message events
5577 - if (isset($data['event']) && $data['event']['type'] === 'message') {
5578 - $event = $data['event'];
5579 -
5580 - // Skip bot messages and messages with subtypes (like bot_message)
5581 - if (isset($event['bot_id']) || isset($event['subtype'])) {
5582 - return new WP_REST_Response(['ok' => true]);
5583 - }
5584 -
5585 - // Threaded replies: in shared-channel mode every conversation lives in
5586 - // a thread rooted at its handoff message — route those to their session
5587 - // by thread root (plan 9f7756). Any other threaded reply (e.g. under a
5588 - // per-conversation channel's confirmation message) finds no session and
5589 - // is skipped, exactly as before.
5590 - if (isset($event['thread_ts']) && $event['thread_ts'] !== $event['ts']) {
5591 - return $this->mxchat_route_shared_thread_reply($event);
5592 - }
5593 -
5594 - $channel_id = $event['channel'];
5595 - $message_text = $event['text'] ?? '';
5596 - $message_ts = $event['ts'] ?? '';
5597 -
5598 - // Find session ID by looking for matching channel
5599 - global $wpdb;
5600 - $session_option = $wpdb->get_var(
5601 - $wpdb->prepare(
5602 - "SELECT option_name FROM {$wpdb->options}
5603 - WHERE option_name LIKE 'mxchat_channel_%'
5604 - AND option_value = %s",
5605 - $channel_id
5606 - )
5607 - );
5608 -
5609 - if ($session_option) {
5610 - $session_id = str_replace('mxchat_channel_', '', $session_option);
5611 -
5612 - // Create a unique key for this specific message
5613 - $message_key = md5($session_id . $message_ts . $message_text);
5614 - $processed_messages = get_transient('mxchat_processed_messages_' . $session_id) ?: [];
5615 -
5616 - // Check if we've already processed this exact message
5617 - if (in_array($message_key, $processed_messages)) {
5618 - //error_log("Duplicate message detected for session $session_id");
5619 - return new WP_REST_Response(['ok' => true]);
5620 - }
5621 -
5622 - // Add to processed messages
5623 - $processed_messages[] = $message_key;
5624 - // Keep only last 50 messages per session
5625 - if (count($processed_messages) > 50) {
5626 - $processed_messages = array_slice($processed_messages, -50);
5627 - }
5628 - set_transient('mxchat_processed_messages_' . $session_id, $processed_messages, HOUR_IN_SECONDS);
5629 -
5630 - $slack_bot_token = $this->options['live_agent_bot_token'] ?? '';
5631 -
5632 - // Handle agent ending the chat — transfer back to AI
5633 - // Format: "!endchat" or "!endchat <custom message to user>"
5634 - if (preg_match('/^!endchat\b/i', trim($message_text))) {
5635 - update_option("mxchat_mode_{$session_id}", 'ai');
5636 -
5637 - // Extract custom message after !endchat, or use empty string
5638 - $custom_message = trim(preg_replace('/^!endchat\s*/i', '', trim($message_text)));
5639 -
5640 - // Send the agent's custom farewell message if provided
5641 - if (!empty($custom_message)) {
5642 - $this->mxchat_save_chat_message($session_id, 'agent', $this->normalize_slack_text($custom_message));
5643 - }
5644 -
5645 - // Confirm in Slack channel
5646 - if (!empty($slack_bot_token)) {
5647 - wp_remote_post('https://slack.com/api/chat.postMessage', [
5648 - 'headers' => [
5649 - 'Content-Type' => 'application/json',
5650 - 'Authorization' => 'Bearer ' . $slack_bot_token
5651 - ],
5652 - 'body' => json_encode([
5653 - 'channel' => $channel_id,
5654 - 'text' => "✅ *Chat ended.* User has been transferred back to AI mode.",
5655 - 'mrkdwn' => true
5656 - ])
5657 - ]);
5658 - }
5659 -
5660 - // Auto-archive the ended conversation's channel (plan 7458a7).
5661 - // Toggle-gated, best-effort — never blocks the mode flip.
5662 - $this->mxchat_maybe_archive_conversation_channel($session_id, $channel_id);
5663 -
5664 - return new WP_REST_Response(['ok' => true]);
5665 - }
5666 -
5667 - // Save the agent message (normalize Slack link/entity formatting first — plan-e2195b)
5668 - $this->mxchat_save_chat_message($session_id, 'agent', $this->normalize_slack_text($message_text));
5669 -
5670 - // Send confirmation back to Slack (only once)
5671 - if (!empty($slack_bot_token)) {
5672 - // Use a transient to prevent duplicate confirmations
5673 - $confirm_key = 'mxchat_confirm_' . $message_key;
5674 - if (!get_transient($confirm_key)) {
5675 - wp_remote_post('https://slack.com/api/chat.postMessage', [
5676 - 'headers' => [
5677 - 'Content-Type' => 'application/json',
5678 - 'Authorization' => 'Bearer ' . $slack_bot_token
5679 - ],
5680 - 'body' => json_encode([
5681 - 'channel' => $channel_id,
5682 - 'text' => "✅ _Message sent to user_",
5683 - 'thread_ts' => $event['ts'] // Reply in thread
5684 - ])
5685 - ]);
5686 - // Set transient to prevent duplicate confirmations
5687 - set_transient($confirm_key, true, 300); // 5 minutes
5688 - }
5689 - }
5690 - }
5691 - }
5692 -
5693 - return new WP_REST_Response(['ok' => true]);
5694 -}
5695 -
5696 -/**
5697 - * Route an agent's threaded Slack reply to the session whose shared-channel
5698 - * conversation is rooted at that thread (plan 9f7756). Sessions are keyed by
5699 - * the thread root ts stored in mxchat_thread_{session}, so two visitors in
5700 - * the same shared channel can never cross-wire. Unknown threads are ignored.
5701 - *
5702 - * @param array $event Slack message event (has thread_ts !== ts).
5703 - * @return WP_REST_Response
5704 - */
5705 -private function mxchat_route_shared_thread_reply($event) {
5706 - $thread_root = $event['thread_ts'] ?? '';
5707 - $message_text = $event['text'] ?? '';
5708 - $message_ts = $event['ts'] ?? '';
5709 - $channel_id = $event['channel'] ?? '';
5710 -
5711 - if ($thread_root === '') {
5712 - return new WP_REST_Response(['ok' => true]);
5713 - }
5714 -
5715 - // Find the session owning this thread root (same reverse-lookup shape as
5716 - // the per-conversation channel mapping).
5717 - global $wpdb;
5718 - $session_option = $wpdb->get_var(
5719 - $wpdb->prepare(
5720 - "SELECT option_name FROM {$wpdb->options}
5721 - WHERE option_name LIKE 'mxchat_thread_%'
5722 - AND option_value = %s",
5723 - $thread_root
5724 - )
5725 - );
5726 -
5727 - if (!$session_option) {
5728 - // Not a shared-channel conversation thread (e.g. a reply under a
5729 - // per-conversation confirmation) — ignore, as before.
5730 - return new WP_REST_Response(['ok' => true]);
5731 - }
5732 -
5733 - $session_id = str_replace('mxchat_thread_', '', $session_option);
5734 -
5735 - // Per-message dedupe — same transient pattern as the top-level handler.
5736 - $message_key = md5($session_id . $message_ts . $message_text);
5737 - $processed_messages = get_transient('mxchat_processed_messages_' . $session_id) ?: [];
5738 - if (in_array($message_key, $processed_messages)) {
5739 - return new WP_REST_Response(['ok' => true]);
5740 - }
5741 - $processed_messages[] = $message_key;
5742 - if (count($processed_messages) > 50) {
5743 - $processed_messages = array_slice($processed_messages, -50);
5744 - }
5745 - set_transient('mxchat_processed_messages_' . $session_id, $processed_messages, HOUR_IN_SECONDS);
5746 -
5747 - $slack_bot_token = $this->options['live_agent_bot_token'] ?? '';
5748 -
5749 - // Agent ending the chat from inside the thread — same command contract as
5750 - // per-conversation channels: "!endchat" or "!endchat <farewell>".
5751 - if (preg_match('/^!endchat\b/i', trim($message_text))) {
5752 - update_option("mxchat_mode_{$session_id}", 'ai');
5753 -
5754 - $custom_message = trim(preg_replace('/^!endchat\s*/i', '', trim($message_text)));
5755 - if (!empty($custom_message)) {
5756 - $this->mxchat_save_chat_message($session_id, 'agent', $this->normalize_slack_text($custom_message));
5757 - }
5758 -
5759 - if (!empty($slack_bot_token) && $channel_id !== '') {
5760 - wp_remote_post('https://slack.com/api/chat.postMessage', [
5761 - 'headers' => [
5762 - 'Content-Type' => 'application/json',
5763 - 'Authorization' => 'Bearer ' . $slack_bot_token
5764 - ],
5765 - 'body' => json_encode([
5766 - 'channel' => $channel_id,
5767 - 'text' => "✅ *Chat ended.* User has been transferred back to AI mode.",
5768 - 'thread_ts' => $thread_root,
5769 - 'mrkdwn' => true
5770 - ])
5771 - ]);
5772 - }
5773 -
5774 - return new WP_REST_Response(['ok' => true]);
5775 - }
5776 -
5777 - // Save the agent message for the widget (normalized like the channel path).
5778 - $this->mxchat_save_chat_message($session_id, 'agent', $this->normalize_slack_text($message_text));
5779 -
5780 - // Confirmation stays inside the conversation's thread.
5781 - if (!empty($slack_bot_token) && $channel_id !== '') {
5782 - $confirm_key = 'mxchat_confirm_' . $message_key;
5783 - if (!get_transient($confirm_key)) {
5784 - wp_remote_post('https://slack.com/api/chat.postMessage', [
5785 - 'headers' => [
5786 - 'Content-Type' => 'application/json',
5787 - 'Authorization' => 'Bearer ' . $slack_bot_token
5788 - ],
5789 - 'body' => json_encode([
5790 - 'channel' => $channel_id,
5791 - 'text' => "✅ _Message sent to user_",
5792 - 'thread_ts' => $thread_root
5793 - ])
5794 - ]);
5795 - set_transient($confirm_key, true, 300);
5796 - }
5797 - }
5798 -
5799 - return new WP_REST_Response(['ok' => true]);
5800 -}
5801 -
5802 -// For the word upload handler
5803 -public function mxchat_handle_word_upload() {
5804 - // Delegate to word handler
5805 - $this->word_handler->mxchat_handle_word_upload();
5806 -}
5807 -
5808 -// For the word removal handler
5809 -public function mxchat_handle_word_remove() {
5810 - // Delegate to word handler
5811 - $this->word_handler->mxchat_handle_word_remove();
5812 -}
5813 -
5814 -// For the word status check
5815 -public function mxchat_check_word_status() {
5816 - // Delegate to word handler
5817 - $this->word_handler->mxchat_check_word_status();
5818 -}
5819 -
5820 -
5821 -private function mxchat_get_user_identifier() {
5822 - return MxChat_User::mxchat_get_user_identifier();
5823 -}
5824 -
5825 -private function mxchat_generate_embedding($text, $api_key) {
5826 - try {
5827 - // Get options and selected model
5828 - $options = get_option('mxchat_options');
5829 - $selected_model = $options['embedding_model'] ?? 'text-embedding-ada-002';
5830 -
5831 - // Opt-in: route embeddings through the Custom (OpenAI-compatible) provider.
5832 - // Off by default so existing sites see byte-identical behavior.
5833 - if (!empty($options['custom_provider_for_embeddings']) && $options['custom_provider_for_embeddings'] === 'on') {
5834 - return $this->mxchat_generate_embedding_custom($text);
5835 - }
5836 -
5837 - // Determine endpoint and API key based on model
5838 - if (strpos($selected_model, 'voyage') === 0) {
5839 - $endpoint = 'https://api.voyageai.com/v1/embeddings';
5840 - $api_key = $options['voyage_api_key'] ?? '';
5841 -
5842 - // Check if Voyage API key is missing
5843 - if (empty($api_key)) {
5844 - //error_log('Voyage API key is missing');
5845 - return [
5846 - 'error' => esc_html__('Voyage AI API key is not configured', 'mxchat'),
5847 - 'error_code' => 'missing_voyage_api_key'
5848 - ];
5849 - }
5850 - } elseif (strpos($selected_model, 'gemini-embedding') === 0) {
5851 - $endpoint = 'https://generativelanguage.googleapis.com/v1beta/models/' . $selected_model . ':embedContent';
5852 - $api_key = $options['gemini_api_key'] ?? '';
5853 -
5854 - // Check if Gemini API key is missing
5855 - if (empty($api_key)) {
5856 - //error_log('Gemini API key is missing');
5857 - return [
5858 - 'error' => esc_html__('Google Gemini API key is not configured', 'mxchat'),
5859 - 'error_code' => 'missing_gemini_api_key'
5860 - ];
5861 - }
5862 - } else {
5863 - $endpoint = 'https://api.openai.com/v1/embeddings';
5864 - // Use the passed API key for OpenAI
5865 -
5866 - // Check if OpenAI API key is missing
5867 - if (empty($api_key)) {
5868 - //error_log('OpenAI API key is missing');
5869 - return [
5870 - 'error' => esc_html__('OpenAI API key is not configured', 'mxchat'),
5871 - 'error_code' => 'missing_openai_api_key'
5872 - ];
5873 - }
5874 - }
5875 -
5876 - // Check if text is empty
5877 - if (empty($text)) {
5878 - //error_log('Empty text provided for embedding generation');
5879 - return [
5880 - 'error' => esc_html__('No text provided for embedding generation', 'mxchat'),
5881 - 'error_code' => 'empty_embedding_text'
5882 - ];
5883 - }
5884 -
5885 - // Prepare request body based on provider
5886 - if (strpos($selected_model, 'gemini-embedding') === 0) {
5887 - // Gemini API format
5888 - $request_body = [
5889 - 'model' => 'models/' . $selected_model,
5890 - 'content' => [
5891 - 'parts' => [
5892 - ['text' => $text]
5893 - ]
5894 - ],
5895 - 'outputDimensionality' => 1536
5896 - ];
5897 -
5898 - // Prepare headers for Gemini (API key as query parameter)
5899 - $endpoint .= '?key=' . $api_key;
5900 - $headers = [
5901 - 'Content-Type' => 'application/json'
5902 - ];
5903 - } else {
5904 - // OpenAI/Voyage API format
5905 - $request_body = [
5906 - 'input' => $text,
5907 - 'model' => $selected_model
5908 - ];
5909 -
5910 - // Add output_dimension for voyage-3-large
5911 - if ($selected_model === 'voyage-3-large') {
5912 - $request_body['output_dimension'] = 2048;
5913 - }
5914 -
5915 - // Prepare headers for OpenAI/Voyage
5916 - $headers = [
5917 - 'Content-Type' => 'application/json',
5918 - 'Authorization' => 'Bearer ' . $api_key
5919 - ];
5920 - }
5921 -
5922 - // Prepare request arguments
5923 - $args = [
5924 - 'body' => wp_json_encode($request_body),
5925 - 'headers' => $headers,
5926 261 'timeout' => 60,
5927 262 'redirection' => 5,
5928 263 'blocking' => true,
5929 264 'httpversion' => '1.0',
@@ -5928,2060 +263,85 @@
5928 263 'blocking' => true,
5929 264 'httpversion' => '1.0',
5930 265 'sslverify' => true,
5931 266 ];
5932 -
5933 - // Make the request
267 +
5934 268 $response = wp_remote_post($endpoint, $args);
5935 -
5936 - // Handle WordPress errors
269 +
5937 270 if (is_wp_error($response)) {
5938 - $error_message = $response->get_error_message();
5939 - //error_log('Embedding Generation Error: ' . $error_message);
5940 - return [
5941 - 'error' => esc_html__('Connection error when generating embeddings: ', 'mxchat') . esc_html($error_message),
5942 - 'error_code' => 'embedding_connection_error'
5943 - ];
271 + return null;
5944 272 }
5945 -
5946 - // Check HTTP status code
5947 - $status_code = wp_remote_retrieve_response_code($response);
5948 - if ($status_code !== 200) {
5949 - $response_body = json_decode(wp_remote_retrieve_body($response), true);
5950 -
5951 - $error_message = $this->extract_provider_error($response_body, 'HTTP Error ' . $status_code);
5952 273
5953 - $error_type = isset($response_body['error']['type'])
5954 - ? $response_body['error']['type']
5955 - : 'unknown';
5956 -
5957 - //error_log('Embedding API HTTP Error: ' . $status_code . ' - ' . $error_message);
5958 -
5959 - // Handle specific error types
5960 - switch ($error_type) {
5961 - case 'invalid_request_error':
5962 - if (strpos($error_message, 'API key') !== false) {
5963 - return [
5964 - 'error' => esc_html__('Invalid API key for embedding generation. Please check your API key configuration.', 'mxchat'),
5965 - 'error_code' => 'embedding_invalid_api_key'
5966 - ];
5967 - }
5968 - break;
5969 -
5970 - case 'authentication_error':
5971 - return [
5972 - 'error' => esc_html__('Authentication failed for embedding generation. Please check your API key.', 'mxchat'),
5973 - 'error_code' => 'embedding_auth_error'
5974 - ];
5975 -
5976 - case 'rate_limit_exceeded':
5977 - return [
5978 - 'error' => esc_html__('Rate limit exceeded for embedding generation. Please try again later.', 'mxchat'),
5979 - 'error_code' => 'embedding_rate_limit'
5980 - ];
5981 -
5982 - case 'quota_exceeded':
5983 - return [
5984 - 'error' => esc_html__('API quota exceeded for embedding generation. Please check your billing details.', 'mxchat'),
5985 - 'error_code' => 'embedding_quota_exceeded'
5986 - ];
5987 - }
5988 -
5989 - // Generic error fallback
5990 - return [
5991 - 'error' => esc_html__('Embedding API error - check embedding API key.: ', 'mxchat') . esc_html($error_message),
5992 - 'error_code' => 'embedding_api_error',
5993 - 'status_code' => $status_code
5994 - ];
5995 - }
5996 -
5997 274 $response_body = json_decode(wp_remote_retrieve_body($response), true);
5998 -
5999 - // Handle different response formats based on provider
6000 - if (strpos($selected_model, 'gemini-embedding') === 0) {
6001 - // Gemini API response format
6002 - if (isset($response_body['embedding']['values']) && is_array($response_body['embedding']['values'])) {
6003 - return $response_body['embedding']['values'];
6004 - } else {
6005 - //error_log('Invalid Gemini embedding response: ' . wp_json_encode($response_body));
6006 - return [
6007 - 'error' => esc_html__('Received invalid embedding data from the Gemini API.', 'mxchat'),
6008 - 'error_code' => 'invalid_gemini_embedding_response'
6009 - ];
6010 - }
6011 - } else {
6012 - // OpenAI/Voyage API response format
6013 - if (isset($response_body['data'][0]['embedding']) && is_array($response_body['data'][0]['embedding'])) {
6014 - return $response_body['data'][0]['embedding'];
6015 - } else {
6016 - //error_log('Invalid embedding response: ' . wp_json_encode($response_body));
6017 - return [
6018 - 'error' => esc_html__('Received invalid embedding data from the API.', 'mxchat'),
6019 - 'error_code' => 'invalid_embedding_response'
6020 - ];
6021 - }
6022 - }
6023 - } catch (Exception $e) {
6024 - //error_log('Embedding Exception: ' . $e->getMessage());
6025 - return [
6026 - 'error' => esc_html__('System error when generating embeddings: ', 'mxchat') . esc_html($e->getMessage()),
6027 - 'error_code' => 'embedding_exception'
6028 - ];
6029 - }
6030 -}
6031 275
6032 -
6033 -/**
6034 - * Generate embedding via a Custom (OpenAI-compatible) provider's /embeddings route.
6035 - * Only called when the opt-in 'custom_provider_for_embeddings' setting is on.
6036 - * Returns a numeric array (the embedding vector) on success, or ['error','error_code'] on failure.
6037 - */
6038 -private function mxchat_generate_embedding_custom($text) {
6039 - if (empty($text)) {
6040 - return ['error' => esc_html__('No text provided for embedding generation', 'mxchat'), 'error_code' => 'empty_embedding_text'];
6041 - }
6042 - $cfg = $this->mxchat_resolve_custom_provider();
6043 - if (empty($cfg['base_url'])) {
6044 - return ['error' => esc_html__('Custom provider Base URL is not configured.', 'mxchat'), 'error_code' => 'missing_custom_provider_base_url'];
6045 - }
6046 -
6047 - $options = get_option('mxchat_options');
6048 - $embed_url = $cfg['base_url'] . '/embeddings';
6049 - if (!empty($cfg['api_version'])) {
6050 - $embed_url .= (strpos($embed_url, '?') === false ? '?' : '&') . 'api-version=' . rawurlencode($cfg['api_version']);
6051 - }
6052 - $model = isset($options['custom_provider_embedding_model']) && trim((string) $options['custom_provider_embedding_model']) !== ''
6053 - ? trim((string) $options['custom_provider_embedding_model'])
6054 - : $cfg['model'];
6055 -
6056 - $response = wp_remote_post($embed_url, [
6057 - 'headers' => $this->mxchat_custom_provider_assoc_headers($cfg),
6058 - 'body' => wp_json_encode(['input' => $text, 'model' => $model]),
6059 - 'timeout' => 60,
6060 - ]);
6061 - if (is_wp_error($response)) {
6062 - return [
6063 - 'error' => esc_html__('Connection error when generating embeddings (custom provider): ', 'mxchat') . esc_html($response->get_error_message()),
6064 - 'error_code' => 'embedding_custom_connection_error',
6065 - ];
6066 - }
6067 - $status = wp_remote_retrieve_response_code($response);
6068 - $body = json_decode(wp_remote_retrieve_body($response), true);
6069 - if ($status !== 200) {
6070 - $msg = $this->extract_provider_error($body, 'HTTP ' . $status);
6071 - return [
6072 - 'error' => esc_html__('Custom embedding endpoint error: ', 'mxchat') . esc_html($msg),
6073 - 'error_code' => 'embedding_custom_api_error',
6074 - 'status_code' => $status,
6075 - ];
6076 - }
6077 - if (isset($body['data'][0]['embedding']) && is_array($body['data'][0]['embedding'])) {
6078 - return $body['data'][0]['embedding'];
6079 - }
6080 - return [
6081 - 'error' => esc_html__('Invalid embedding response from custom provider.', 'mxchat'),
6082 - 'error_code' => 'embedding_custom_invalid_response',
6083 - ];
6084 -}
6085 -
6086 -private function mxchat_find_relevant_content($user_embedding, $bot_id = 'default', $user_query = '') {
6087 - //error_log("MXCHAT DEBUG: find_relevant_content called with bot_id: " . $bot_id);
6088 -
6089 - // Check for OpenAI Vector Store first (takes priority when enabled)
6090 - $bot_vectorstore_config = $this->get_bot_vectorstore_config($bot_id);
6091 -
6092 - if ($bot_vectorstore_config['use_vectorstore']) {
6093 - // Get current model to verify it's an OpenAI model
6094 - $bot_options = $this->get_bot_options($bot_id);
6095 - $mxchat_options = get_option('mxchat_options', array());
6096 - $current_options = !empty($bot_options) ? $bot_options : $mxchat_options;
6097 - $selected_model = $current_options['model'] ?? 'gpt-5.6-sol';
6098 -
6099 - if ($this->is_openai_chat_model($selected_model)) {
6100 - //error_log("MXCHAT DEBUG: Using OpenAI Vector Store for knowledge retrieval");
6101 - return $this->find_relevant_content_openai_vectorstore($user_query, $bot_id, $bot_vectorstore_config);
276 + if (isset($response_body['data'][0]['embedding']) && is_array($response_body['data'][0]['embedding'])) {
277 + return $response_body['data'][0]['embedding'];
6102 278 } else {
6103 - //error_log("MXCHAT DEBUG: Vector Store enabled but model is not OpenAI (" . $selected_model . "), skipping Vector Store");
279 + return null;
6104 280 }
6105 281 }
6106 282
6107 - // Get bot-specific Pinecone configuration
6108 - $bot_pinecone_config = $this->get_bot_pinecone_config($bot_id);
6109 -
6110 - // Debug: Log the Pinecone configuration
6111 - //error_log("MXCHAT DEBUG: Pinecone config for bot '$bot_id':");
6112 - //error_log(" - use_pinecone: " . ($bot_pinecone_config['use_pinecone'] ? 'true' : 'false'));
6113 - //error_log(" - api_key: " . (empty($bot_pinecone_config['api_key']) ? 'EMPTY' : 'SET (hidden)'));
6114 - //error_log(" - host: " . ($bot_pinecone_config['host'] ?? 'NOT SET'));
6115 - //error_log(" - namespace: " . ($bot_pinecone_config['namespace'] ?? 'NOT SET'));
6116 -
6117 - // Determine whether to use Pinecone based on bot configuration
6118 - $use_pinecone = isset($bot_pinecone_config['use_pinecone']) ? $bot_pinecone_config['use_pinecone'] : false;
6119 -
6120 - //error_log("MXCHAT DEBUG: Using " . ($use_pinecone ? "Pinecone" : "WordPress Database") . " for knowledge retrieval");
6121 -
6122 - if ($use_pinecone) {
6123 - return $this->find_relevant_content_pinecone($user_embedding, $bot_id, $bot_pinecone_config);
6124 - } else {
6125 - return $this->find_relevant_content_wordpress($user_embedding, $bot_id, $user_query);
6126 - }
6127 -}
6128 -
6129 -private function find_relevant_content_wordpress($user_embedding, $bot_id = 'default', $user_query = '') {
283 +private function mxchat_find_relevant_content($user_embedding) {
6130 284 global $wpdb;
6131 285 $system_prompt_table = $wpdb->prefix . 'mxchat_system_prompt_content';
6132 - // Initialize similarity analysis storage
6133 - $this->last_similarity_analysis = [
6134 - 'knowledge_base_type' => 'WordPress Database',
6135 - 'bot_id' => $bot_id,
6136 - 'top_matches' => [],
6137 - 'threshold_used' => 0,
6138 - 'total_checked' => 0
6139 - ];
6140 286
6141 - // NEW: Initialize valid URLs array
6142 - $valid_urls = [];
287 + // Define a cache key for embeddings
288 + $cache_key = 'mxchat_system_prompt_embeddings';
6143 289
6144 - // Get bot-specific options for similarity threshold
6145 - $bot_options = $this->get_bot_options($bot_id);
6146 - $current_options = !empty($bot_options) ? $bot_options : $this->options;
290 + // Attempt to get the embeddings from the cache
291 + $embeddings = wp_cache_get($cache_key, 'mxchat_system_prompts');
6147 292
6148 - // Get knowledge manager instance for role checking
6149 - $knowledge_manager = MxChat_Knowledge_Manager::get_instance();
293 + if ($embeddings === false) {
294 + // Cache miss, query the database and cache the results
295 + $query = "SELECT id, embedding_vector FROM {$system_prompt_table}";
296 + $embeddings = $wpdb->get_results($query);
6150 297
6151 - // Get base similarity threshold from bot options or default options
6152 - $similarity_threshold = isset($current_options['similarity_threshold'])
6153 - ? ((int) $current_options['similarity_threshold']) / 100
6154 - : 0.35;
6155 - $this->last_similarity_analysis['threshold_used'] = $similarity_threshold;
6156 -
6157 - // Precompute bot_filter once, outside the streaming loop
6158 - $bot_filter = '';
6159 - if ($bot_id !== 'default') {
6160 - $column_exists = $wpdb->get_var("SHOW COLUMNS FROM {$system_prompt_table} LIKE 'bot_metadata'");
6161 - if ($column_exists) {
6162 - $bot_filter = $wpdb->prepare(" AND (bot_metadata = %s OR bot_metadata IS NULL OR bot_metadata = '')", $bot_id);
298 + if ($embeddings === null || empty($embeddings)) {
299 + error_log("No embeddings found in the database.");
300 + return null; // Return null to handle no embeddings gracefully
6163 301 }
6164 - }
6165 302
6166 - // Hybrid keyword boost (plan-38ffa1, default OFF). Runs a ranked keyword
6167 - // query alongside the vector scan and fuses the two lists by reciprocal
6168 - // rank, so exact-token queries (SKUs, error codes, names) hit even when
6169 - // their embedding similarity is semantic mush. The keyword leg runs FIRST
6170 - // so the vector scan below can record true cosine similarity for its hits
6171 - // (the display keeps cosine % as the anchor).
6172 - $hybrid_enabled = get_option('mxchat_hybrid_keyword_toggle', 'off') === 'on'
6173 - && trim((string) $user_query) !== '';
6174 - $keyword_hits = array(); // ranked + access-filtered, max 20
6175 - $keyword_ids = array(); // id => keyword rank (1-based)
6176 - $keyword_similarities = array(); // id => cosine recorded during the scan
6177 - if ($hybrid_enabled) {
6178 - $keyword_hits = $this->mxchat_hybrid_keyword_search($user_query, $system_prompt_table, $bot_filter, $knowledge_manager);
6179 - foreach ($keyword_hits as $kw_i => $kw_hit) {
6180 - $keyword_ids[$kw_hit['id']] = $kw_i + 1;
6181 - }
303 + // Cache the results if successful
304 + wp_cache_set($cache_key, $embeddings, 'mxchat_system_prompts', 3600); // Cache for 1 hour
6182 305 }
6183 306
6184 - // ===== STREAMING TOP-K PASS =====
6185 - // Stream rows in small batches, compute cosine similarity per row, and keep only:
6186 - // - top 10 by raw similarity (for the testing/debug display panel)
6187 - // - candidates above threshold with access (capped) for context assembly
6188 - // This bounds peak memory regardless of knowledge base size and avoids loading
6189 - // article_content for every row. article_content is fetched in Phase 2 for winners only.
6190 - $batch_size = 250;
6191 - $max_candidates = 200; // safety cap, well above rag_sources_limit * max_chunks_per_source
6192 - $top_display = [];
6193 - $candidates = [];
6194 - $total_checked = 0;
6195 - $offset = 0;
307 + $most_relevant_id = null;
308 + $highest_similarity = -INF;
6196 309
6197 - do {
6198 - $batch = $wpdb->get_results($wpdb->prepare(
6199 - "SELECT id, embedding_vector, source_url, role_restriction
6200 - FROM {$system_prompt_table}
6201 - WHERE 1=1 {$bot_filter}
6202 - LIMIT %d OFFSET %d",
6203 - $batch_size,
6204 - $offset
6205 - ));
310 + foreach ($embeddings as $embedding) {
311 + $database_embedding = maybe_unserialize($embedding->embedding_vector);
6206 312
6207 - if (empty($batch)) {
6208 - break;
6209 - }
313 + // Debugging: Log the embeddings
314 + // if (!is_array($database_embedding)) {
315 + // error_log("Invalid database embedding format for ID {$embedding->id}: " . print_r($database_embedding, true));
316 + // continue;
317 + // }
6210 318
6211 - foreach ($batch as $row) {
6212 - $database_embedding = $row->embedding_vector
6213 - ? unserialize($row->embedding_vector, ['allowed_classes' => false])
6214 - : null;
6215 -
6216 - if (!is_array($database_embedding) || !is_array($user_embedding)) {
6217 - unset($database_embedding);
6218 - continue;
6219 - }
6220 -
319 + if (is_array($user_embedding)) {
6221 320 $similarity = $this->mxchat_calculate_cosine_similarity($user_embedding, $database_embedding);
6222 - unset($database_embedding);
6223 321
6224 - $role_restriction = $row->role_restriction ?? 'public';
6225 - $has_access = $knowledge_manager->mxchat_user_has_content_access($role_restriction);
6226 - $source_url = $row->source_url ?? '';
322 + // Debugging: Log the similarity score
323 + // error_log("Calculated similarity for ID {$embedding->id}: {$similarity}");
6227 324
6228 - // Maintain top 10 display buffer (insert-if-beats-worst)
6229 - if (count($top_display) < 10) {
6230 - $top_display[] = [
6231 - 'id' => $row->id,
6232 - 'similarity' => $similarity,
6233 - 'source_url' => $source_url,
6234 - 'role_restriction' => $role_restriction,
6235 - 'has_access' => $has_access,
6236 - ];
6237 - usort($top_display, function ($a, $b) {
6238 - return $b['similarity'] <=> $a['similarity'];
6239 - });
6240 - } elseif ($similarity > $top_display[9]['similarity']) {
6241 - $top_display[9] = [
6242 - 'id' => $row->id,
6243 - 'similarity' => $similarity,
6244 - 'source_url' => $source_url,
6245 - 'role_restriction' => $role_restriction,
6246 - 'has_access' => $has_access,
6247 - ];
6248 - usort($top_display, function ($a, $b) {
6249 - return $b['similarity'] <=> $a['similarity'];
6250 - });
325 + if ($similarity > $highest_similarity) {
326 + $highest_similarity = $similarity;
327 + $most_relevant_id = $embedding->id;
6251 328 }
6252 -
6253 - // Record cosine for keyword-leg hits so fusion/display can anchor
6254 - // on the true similarity % even for below-threshold rescues.
6255 - if ($hybrid_enabled && isset($keyword_ids[$row->id])) {
6256 - $keyword_similarities[$row->id] = $similarity;
6257 - }
6258 -
6259 - // Track candidates for context assembly (above threshold + has access)
6260 - if ($similarity >= $similarity_threshold && $has_access) {
6261 - $candidates[] = [
6262 - 'id' => $row->id,
6263 - 'similarity' => $similarity,
6264 - 'source_url' => $source_url,
6265 - ];
6266 - }
6267 -
6268 - $total_checked++;
6269 - }
6270 -
6271 - unset($batch);
6272 -
6273 - // Trim candidates periodically to cap memory during long scans
6274 - if (count($candidates) > $max_candidates) {
6275 - usort($candidates, function ($a, $b) {
6276 - return $b['similarity'] <=> $a['similarity'];
6277 - });
6278 - $candidates = array_slice($candidates, 0, $max_candidates);
6279 - }
6280 -
6281 - $offset += $batch_size;
6282 - } while (true);
6283 -
6284 - if ($total_checked === 0) {
6285 - $this->current_valid_urls = [];
6286 - return '';
6287 - }
6288 -
6289 - // Final candidates sort (best first)
6290 - if (count($candidates) > 1) {
6291 - usort($candidates, function ($a, $b) {
6292 - return $b['similarity'] <=> $a['similarity'];
6293 - });
6294 - }
6295 -
6296 - // ===== HYBRID FUSION (plan-38ffa1) =====
6297 - // Reciprocal-rank fusion over the top-20 of each leg (k=60 standard).
6298 - // Rank-based, so the incomparable score scales (cosine 0-1 vs FULLTEXT
6299 - // relevance) never need calibrating. A below-threshold vector row can
6300 - // enter via a strong keyword rank — that is the point of the feature.
6301 - // Every candidate gets a 'rank_score' the downstream source ordering
6302 - // uses: with hybrid OFF it is exactly the cosine similarity, so the
6303 - // legacy path is byte-identical.
6304 - $fused_rank_map = array(); // id => 1-based fused rank
6305 - $matched_via_map = array(); // id => 'vector' | 'keyword' | 'both'
6306 - if (!$hybrid_enabled) {
6307 - foreach ($candidates as &$cand_ref) {
6308 - $cand_ref['rank_score'] = $cand_ref['similarity'];
6309 - }
6310 - unset($cand_ref);
6311 - } else {
6312 - $rrf_k = 60;
6313 - $fused = array();
6314 - foreach (array_slice($candidates, 0, 20) as $leg_rank => $cand) {
6315 - $fused[$cand['id']] = array(
6316 - 'id' => $cand['id'],
6317 - 'similarity' => $cand['similarity'],
6318 - 'source_url' => $cand['source_url'],
6319 - 'rrf' => 1 / ($rrf_k + $leg_rank + 1),
6320 - 'via' => 'vector',
6321 - );
6322 - }
6323 - foreach ($keyword_hits as $leg_rank => $hit) {
6324 - $rrf = 1 / ($rrf_k + $leg_rank + 1);
6325 - if (isset($fused[$hit['id']])) {
6326 - $fused[$hit['id']]['rrf'] += $rrf;
6327 - $fused[$hit['id']]['via'] = 'both';
6328 - } else {
6329 - $fused[$hit['id']] = array(
6330 - 'id' => $hit['id'],
6331 - 'similarity' => $keyword_similarities[$hit['id']] ?? 0.0,
6332 - 'source_url' => $hit['source_url'],
6333 - 'rrf' => $rrf,
6334 - 'via' => 'keyword',
6335 - );
6336 - }
6337 - }
6338 - uasort($fused, function ($a, $b) {
6339 - return $b['rrf'] <=> $a['rrf'];
6340 - });
6341 -
6342 - // Vector candidates beyond the top-20 leg keep flowing to the prompt
6343 - // builders after the fused block, in their vector order — the result
6344 - // count/shape downstream stays unchanged.
6345 - $tail = array_slice($candidates, 20);
6346 - $candidates = array();
6347 - $rank = 0;
6348 - foreach ($fused as $f) {
6349 - $rank++;
6350 - $fused_rank_map[$f['id']] = $rank;
6351 - $matched_via_map[$f['id']] = $f['via'];
6352 - $candidates[] = array(
6353 - 'id' => $f['id'],
6354 - 'similarity' => $f['similarity'],
6355 - 'source_url' => $f['source_url'],
6356 - 'rank_score' => $f['rrf'],
6357 - );
6358 - }
6359 - foreach ($tail as $cand) {
6360 - // Below any fused rrf (min possible fused rrf is 1/(60+40)=0.01;
6361 - // similarity * 1e-6 <= 1e-6), preserving relative vector order.
6362 - $cand['rank_score'] = $cand['similarity'] * 1e-6;
6363 - $candidates[] = $cand;
6364 - }
6365 - if (count($candidates) > $max_candidates) {
6366 - $candidates = array_slice($candidates, 0, $max_candidates);
6367 - }
6368 - }
6369 -
6370 - // ===== PHASE 2: FETCH ARTICLE CONTENT ONLY FOR WINNERS =====
6371 - // Gather unique IDs we actually need (top_display + candidates) and pull
6372 - // article_content in bounded IN() batches. This avoids loading content for
6373 - // every row during the similarity scan.
6374 - $needed_ids = [];
6375 - foreach ($top_display as $item) {
6376 - $needed_ids[$item['id']] = true;
6377 - }
6378 - foreach ($candidates as $item) {
6379 - $needed_ids[$item['id']] = true;
6380 - }
6381 - $needed_ids = array_keys($needed_ids);
6382 -
6383 - $content_map = [];
6384 - if (!empty($needed_ids)) {
6385 - foreach (array_chunk($needed_ids, 250) as $chunk_ids) {
6386 - $placeholders = implode(',', array_fill(0, count($chunk_ids), '%d'));
6387 - $rows = $wpdb->get_results($wpdb->prepare(
6388 - "SELECT id, article_content FROM {$system_prompt_table} WHERE id IN ($placeholders)",
6389 - ...$chunk_ids
6390 - ));
6391 - foreach ($rows as $r) {
6392 - $content_map[$r->id] = $r->article_content;
6393 - }
6394 - unset($rows);
6395 - }
6396 - }
6397 -
6398 - // Build the all_similarities display array from the top 10
6399 - $all_similarities = [];
6400 - foreach ($top_display as $item) {
6401 - $article_content_for_parse = $content_map[$item['id']] ?? '';
6402 - $parsed_for_display = MxChat_Chunker::parse_stored_chunk($article_content_for_parse);
6403 - $is_chunk = $parsed_for_display['is_chunked'];
6404 - $chunk_meta = $parsed_for_display['metadata'];
6405 -
6406 - if (!empty($item['source_url']) && $item['source_url'] !== '#') {
6407 - $source_display = $item['source_url'];
6408 329 } else {
6409 - $content_preview = strip_tags($article_content_for_parse);
6410 - $content_preview = preg_replace('/\s+/', ' ', $content_preview);
6411 - $source_display = substr(trim($content_preview), 0, 50) . '...';
330 + // error_log("User embedding is not an array. Embedding data: " . print_r($user_embedding, true));
6412 331 }
6413 -
6414 - $all_similarities[] = [
6415 - 'document_id' => $item['id'],
6416 - 'similarity' => $item['similarity'],
6417 - 'similarity_percentage' => round($item['similarity'] * 100, 2),
6418 - 'above_threshold' => $item['similarity'] >= $similarity_threshold,
6419 - 'source_display' => $source_display,
6420 - 'content_preview' => substr(strip_tags($parsed_for_display['text'] ?? ''), 0, 100) . '...',
6421 - 'used_for_context' => false,
6422 - 'role_restriction' => $item['role_restriction'],
6423 - 'has_access' => $item['has_access'],
6424 - 'filtered_out' => !$item['has_access'],
6425 - 'is_chunk' => $is_chunk,
6426 - 'chunk_index' => $is_chunk ? ($chunk_meta['chunk_index'] ?? 0) : null,
6427 - 'total_chunks' => $is_chunk ? ($chunk_meta['total_chunks'] ?? 1) : null
6428 - ];
6429 332 }
6430 333
6431 - // Build url_groups from candidates for chunk reassembly
6432 - $url_groups = array();
6433 - foreach ($candidates as $cand) {
6434 - $article_content = $content_map[$cand['id']] ?? '';
6435 - $parsed = MxChat_Chunker::parse_stored_chunk($article_content);
6436 - $is_chunked = $parsed['is_chunked'];
6437 - $chunk_index = $parsed['metadata']['chunk_index'] ?? 0;
6438 - $text_content = $parsed['text'];
6439 -
6440 - $source_url = $cand['source_url'];
6441 - $group_key = !empty($source_url) ? $source_url : '_manual_' . $cand['id'];
6442 -
6443 - if (!isset($url_groups[$group_key])) {
6444 - $url_groups[$group_key] = array(
6445 - 'source_url' => $source_url,
6446 - 'best_score' => 0,
6447 - 'is_chunked' => $is_chunked,
6448 - 'chunks' => array(),
6449 - 'single_text' => '',
6450 - 'single_id' => null
6451 - );
6452 - }
6453 -
6454 - // rank_score == similarity with hybrid off (byte-identical ordering);
6455 - // with hybrid on it carries the fused rank so keyword rescues sort up.
6456 - $cand_rank_score = $cand['rank_score'] ?? $cand['similarity'];
6457 - if ($cand_rank_score > $url_groups[$group_key]['best_score']) {
6458 - $url_groups[$group_key]['best_score'] = $cand_rank_score;
6459 - }
6460 -
6461 - if ($is_chunked) {
6462 - $url_groups[$group_key]['is_chunked'] = true;
6463 - $url_groups[$group_key]['chunks'][] = array(
6464 - 'id' => $cand['id'],
6465 - 'score' => $cand['similarity'],
6466 - 'chunk_index' => $chunk_index,
6467 - 'text' => $text_content
6468 - );
6469 - } else {
6470 - $url_groups[$group_key]['single_text'] = $text_content;
6471 - $url_groups[$group_key]['single_id'] = $cand['id'];
6472 - }
334 + if ($most_relevant_id !== null) {
335 + // Fetch content with product links
336 + return $this->fetch_content_with_product_links($most_relevant_id);
6473 337 }
6474 338
6475 - // Hybrid display augmentation (plan-38ffa1, Maxwell's approval note):
6476 - // make sure every fused-top-10 row appears in the debug panel — a
6477 - // keyword-only rescue may sit below the vector top-10 buffer — and stamp
6478 - // matched_via + fused_rank on every row. Cosine % stays the anchor; no
6479 - // raw RRF numbers surface.
6480 - if ($hybrid_enabled) {
6481 - $displayed_ids = array();
6482 - foreach ($all_similarities as $disp_item) {
6483 - $displayed_ids[$disp_item['document_id']] = true;
6484 - }
6485 - $kw_info_by_id = array();
6486 - foreach ($keyword_hits as $hit) {
6487 - $kw_info_by_id[$hit['id']] = $hit;
6488 - }
6489 - foreach ($fused_rank_map as $fused_id => $fused_rank) {
6490 - if ($fused_rank > 10 || isset($displayed_ids[$fused_id])) {
6491 - continue;
6492 - }
6493 - $aug_content = $content_map[$fused_id] ?? '';
6494 - $aug_parsed = MxChat_Chunker::parse_stored_chunk($aug_content);
6495 - $aug_hit = $kw_info_by_id[$fused_id] ?? array();
6496 - $aug_similarity = $keyword_similarities[$fused_id] ?? 0.0;
6497 - $aug_source_url = $aug_hit['source_url'] ?? '';
6498 - if (!empty($aug_source_url) && $aug_source_url !== '#') {
6499 - $aug_source_display = $aug_source_url;
6500 - } else {
6501 - $aug_preview = preg_replace('/\s+/', ' ', strip_tags($aug_content));
6502 - $aug_source_display = substr(trim($aug_preview), 0, 50) . '...';
6503 - }
6504 - $all_similarities[] = [
6505 - 'document_id' => $fused_id,
6506 - 'similarity' => $aug_similarity,
6507 - 'similarity_percentage' => round($aug_similarity * 100, 2),
6508 - 'above_threshold' => $aug_similarity >= $similarity_threshold,
6509 - 'source_display' => $aug_source_display,
6510 - 'content_preview' => substr(strip_tags($aug_parsed['text'] ?? ''), 0, 100) . '...',
6511 - 'used_for_context' => false,
6512 - 'role_restriction' => $aug_hit['role_restriction'] ?? 'public',
6513 - 'has_access' => $aug_hit['has_access'] ?? true,
6514 - 'filtered_out' => false,
6515 - 'is_chunk' => $aug_parsed['is_chunked'],
6516 - 'chunk_index' => $aug_parsed['is_chunked'] ? ($aug_parsed['metadata']['chunk_index'] ?? 0) : null,
6517 - 'total_chunks' => $aug_parsed['is_chunked'] ? ($aug_parsed['metadata']['total_chunks'] ?? 1) : null,
6518 - ];
6519 - }
6520 - foreach ($all_similarities as &$disp_ref) {
6521 - $disp_ref['matched_via'] = $matched_via_map[$disp_ref['document_id']] ?? null;
6522 - $disp_ref['fused_rank'] = $fused_rank_map[$disp_ref['document_id']] ?? null;
6523 - }
6524 - unset($disp_ref);
6525 - }
6526 -
6527 - // Sort for the testing/debug display: fused rank when hybrid is on
6528 - // (nulls last, cosine as tie-break), raw similarity otherwise.
6529 - if ($hybrid_enabled) {
6530 - usort($all_similarities, function ($a, $b) {
6531 - $ar = $a['fused_rank'] ?? PHP_INT_MAX;
6532 - $br = $b['fused_rank'] ?? PHP_INT_MAX;
6533 - if ($ar !== $br) {
6534 - return $ar <=> $br;
6535 - }
6536 - return $b['similarity'] <=> $a['similarity'];
6537 - });
6538 - } else {
6539 - usort($all_similarities, function ($a, $b) {
6540 - return $b['similarity'] <=> $a['similarity'];
6541 - });
6542 - }
6543 -
6544 - // Sort URL groups by best score (highest first)
6545 - uasort($url_groups, function($a, $b) {
6546 - return $b['best_score'] <=> $a['best_score'];
6547 - });
6548 -
6549 - // Get RAG sources limit from options (default 6, min 3, max 10)
6550 - $rag_sources_limit = isset($current_options['rag_sources_limit']) ? intval($current_options['rag_sources_limit']) : 3;
6551 - if ($rag_sources_limit < 3) $rag_sources_limit = 3;
6552 - if ($rag_sources_limit > 10) $rag_sources_limit = 10;
6553 -
6554 - // Take top N unique URLs based on user setting
6555 - $top_urls = array_slice($url_groups, 0, $rag_sources_limit, true);
6556 -
6557 - // Track which document IDs are used for context
6558 - $used_document_ids = [];
6559 - foreach ($top_urls as $group) {
6560 - if ($group['is_chunked']) {
6561 - foreach ($group['chunks'] as $chunk) {
6562 - $used_document_ids[] = $chunk['id'];
6563 - }
6564 - } elseif ($group['single_id']) {
6565 - $used_document_ids[] = $group['single_id'];
6566 - }
6567 - }
6568 -
6569 - // Update the all_similarities array to mark which were actually used
6570 - foreach ($all_similarities as &$similarity_item) {
6571 - $similarity_item['used_for_context'] = in_array($similarity_item['document_id'], $used_document_ids);
6572 - }
6573 -
6574 - // Store top 10 for testing panel
6575 - $this->last_similarity_analysis['top_matches'] = array_slice($all_similarities, 0, 10);
6576 - $this->last_similarity_analysis['total_checked'] = $total_checked;
6577 -
6578 - // Initialize final content
6579 - $content = '';
6580 - $matches_used = 0;
6581 - $total_chunks_used = 0;
6582 - $max_total_chunks = isset($current_options['rag_chunks_limit']) ? intval($current_options['rag_chunks_limit']) : 15;
6583 - if ($max_total_chunks < 8) $max_total_chunks = 8;
6584 - if ($max_total_chunks > 20) $max_total_chunks = 20;
6585 - $max_chunks_per_source = 5; // Cap per individual source to limit token usage
6586 -
6587 - // Check if citation links are enabled (default to 'on' for backwards compatibility)
6588 - // Use fresh options to ensure we get the latest setting value
6589 - $fresh_options = get_option('mxchat_options', []);
6590 - $citation_links_enabled = isset($fresh_options['citation_links_toggle']) ? ($fresh_options['citation_links_toggle'] === 'on') : true;
6591 -
6592 - // Build content from top sources
6593 - foreach ($top_urls as $group_key => $group) {
6594 - $source_url = $group['source_url']; // Use actual source_url, not the group key
6595 -
6596 - // Stop if we've hit the total chunk limit
6597 - if ($total_chunks_used >= $max_total_chunks) {
6598 - break;
6599 - }
6600 -
6601 - $full_text = '';
6602 - $chunks_in_this_source = 1; // Default for non-chunked content
6603 -
6604 - if ($group['is_chunked']) {
6605 - // Calculate how many chunks we can still use (respect both total and per-source caps)
6606 - $chunks_remaining = min($max_chunks_per_source, $max_total_chunks - $total_chunks_used);
6607 -
6608 - // Fetch chunks for this URL with limit
6609 - $full_text = $this->reassemble_chunks_from_wordpress($source_url, $chunks_remaining, $chunks_in_this_source);
6610 -
6611 - // If fetching all chunks fails, fall back to matched chunks
6612 - if (empty($full_text)) {
6613 - // Sort matched chunks by index and concatenate
6614 - usort($group['chunks'], function($a, $b) {
6615 - return $a['chunk_index'] <=> $b['chunk_index'];
6616 - });
6617 -
6618 - $chunk_texts = array();
6619 - $chunks_in_this_source = 0;
6620 - foreach ($group['chunks'] as $chunk) {
6621 - if ($total_chunks_used + $chunks_in_this_source >= $max_total_chunks) {
6622 - break;
6623 - }
6624 - $chunk_texts[] = $chunk['text'];
6625 - $chunks_in_this_source++;
6626 - }
6627 - $full_text = implode("\n\n", $chunk_texts);
6628 - }
6629 - } else {
6630 - $full_text = $group['single_text'];
6631 - $chunks_in_this_source = 1;
6632 - }
6633 -
6634 - if (!empty($full_text)) {
6635 - // Strip URLs from content if citation links are disabled
6636 - if (!$citation_links_enabled) {
6637 - $full_text = preg_replace('#\bhttps?://[^\s<>"\']+#i', '', $full_text);
6638 - $full_text = preg_replace('/\s+/', ' ', trim($full_text)); // Clean up extra spaces
6639 - }
6640 -
6641 - // Use numbered reference for URL-based entries, plain info label for manual entries
6642 - // Manual entries are stored with an internal mxchat:// placeholder URL — never expose them as citations
6643 - if (!empty($source_url) && $source_url !== '#' && strpos($source_url, 'mxchat://') !== 0) {
6644 - $matches_used++;
6645 - $content .= "## Reference " . $matches_used . " ##\n";
6646 - $content .= $full_text . "\n\n";
6647 -
6648 - // Only include citation URLs if citation links are enabled
6649 - if ($citation_links_enabled) {
6650 - $valid_urls[] = $source_url;
6651 - $content .= "URL: " . $source_url . "\n\n";
6652 - }
6653 -
6654 - // Video-backed source → queue the consent-safe embed (03ba33)
6655 - $this->maybe_queue_youtube_embed($source_url, $full_text);
6656 - } else {
6657 - // Manual entry — no reference number, no citation
6658 - $content .= "## Information ##\n";
6659 - $content .= $full_text . "\n\n";
6660 - }
6661 -
6662 - // Extract any URLs from the text content itself (only if citation links enabled)
6663 - if ($citation_links_enabled) {
6664 - preg_match_all(
6665 - '#\bhttps?://[^\s<>"\']+#i',
6666 - $full_text,
6667 - $content_urls
6668 - );
6669 - if (!empty($content_urls[0])) {
6670 - $valid_urls = array_merge($valid_urls, $content_urls[0]);
6671 - }
6672 - }
6673 -
6674 - $total_chunks_used += $chunks_in_this_source;
6675 - }
6676 - }
6677 -
6678 - // NEW: Store unique valid URLs for validation
6679 - $this->current_valid_urls = array_unique($valid_urls);
6680 -
6681 - // Store sources and chunks counts for testing/transcript display
6682 - $this->last_similarity_analysis['sources_used'] = $matches_used;
6683 - $this->last_similarity_analysis['total_chunks_used'] = $total_chunks_used;
6684 -
6685 - // Allow add-ons to act on similarity results (e.g. WooCommerce product card display)
6686 - do_action('mxchat_similarity_results', $this->last_similarity_analysis['top_matches'], $bot_id);
6687 -
6688 - // Add response guidelines
6689 - if (empty($top_urls)) {
6690 - // No matched sources: return empty so the prompt assembler's
6691 - // "NO RELEVANT CONTENT FOUND IN KNOWLEDGE DATABASE" branch fires —
6692 - // a no-info sentence wrapped in OFFICIAL KNOWLEDGE markers reads to
6693 - // the model as authoritative content (plan d7daf8).
6694 - $content = '';
6695 - } else {
6696 - // Build response guidelines based on citation links setting
6697 - $content .= "\n## Response Guidelines ##\n" .
6698 - "You are an AI Chatbot. Answer naturally and helpfully using only the information from the references above. " .
6699 - "Be conversational and friendly, but never mention your knowledge base or training data. " .
6700 - "If you don't have specific information or are uncertain about any details, it's always " .
6701 - "better to honestly say you don't know rather than making up or guessing at answers. " .
6702 - "When information is incomplete, let them know you are unsure.\n\n";
6703 -
6704 - // Only add hyperlink instructions if citation links are enabled
6705 - if ($citation_links_enabled) {
6706 - $content .= "CRITICAL: When creating hyperlinks, always use proper markdown format with descriptive text: " .
6707 - "[descriptive text](url). NEVER use empty brackets like [](url). The text in brackets must describe what the link is about. " .
6708 - "Only cite references that have a URL. Do not cite or add source labels to Information sections that have no URL.";
6709 - } else {
6710 - $content .= "IMPORTANT: Do not include any citation links, source URLs, or hyperlinks in your responses. " .
6711 - "Simply provide helpful answers based on the reference information without citing sources.";
6712 - }
6713 - }
6714 -
6715 - return trim($content);
339 + error_log("No relevant content found. Most relevant ID was null.");
340 + return null; // Return null if no relevant content is found
6716 341 }
6717 342
6718 -/**
6719 - * plan-mxchat-20260717-03ba33 — if a KB source used for context is a single
6720 - * YouTube video, queue ONE consent-safe embed for the response html channel.
6721 - * Called from BOTH retrieval builders (WordPress DB + Pinecone) inside their
6722 - * real-URL winner branch, in ranked order — so the first (best) video wins and
6723 - * later matches are ignored. Only KB/admin-ingested sources ever reach this
6724 - * point; a URL a visitor pastes in chat never does.
6725 - */
6726 -private function maybe_queue_youtube_embed($source_url, $full_text) {
6727 - if (!empty($this->videoEmbedHtml)) {
6728 - return; // one video per response
6729 - }
6730 - $video_id = MxChat_Utils::parse_youtube_id($source_url);
6731 - if (empty($video_id)) {
6732 - return;
6733 - }
6734 - // Ingestion writes "YouTube Video: {title}" / "Channel: {name}" / "URL: …"
6735 - // header lines into the indexed text. NOTE: when citation links are
6736 - // disabled the winner loop collapses ALL whitespace to single spaces
6737 - // before this runs, so the title must be terminated by the next header
6738 - // label, not by end-of-line. Fall back to a generic label when absent
6739 - // (e.g. a YouTube watch page imported through the plain URL source).
6740 - $title = '';
6741 - if (preg_match('/YouTube Video:\s*(.+?)(?=\s+Channel:\s|\s+URL:\s|\r|\n|$)/i', (string) $full_text, $m)) {
6742 - $title = trim(mb_substr(trim($m[1]), 0, 140));
6743 - if (preg_match('#^https?://#i', $title)) {
6744 - $title = ''; // header carried the URL, not a real title
6745 - }
6746 - }
6747 - $this->videoEmbedHtml = $this->build_youtube_embed_html($video_id, $title, $source_url);
6748 -}
6749 343
6750 -/**
6751 - * Consent-safe click-to-load YouTube facade. No Google iframe is created until
6752 - * the visitor taps play (chat-script.js swaps the facade for a
6753 - * youtube-nocookie.com iframe). The caption always carries a plain "Watch on
6754 - * YouTube" link, which is also the graceful degrade on strict-CSP sites where
6755 - * third-party frames are blocked.
6756 - */
6757 -private function build_youtube_embed_html($video_id, $title, $watch_url) {
6758 - $video_id = preg_replace('/[^A-Za-z0-9_-]/', '', (string) $video_id);
6759 - if ($video_id === '') {
6760 - return '';
6761 - }
6762 - $thumb = 'https://i.ytimg.com/vi/' . $video_id . '/hqdefault.jpg';
6763 - $label = ($title !== '') ? $title : __('YouTube video', 'mxchat');
6764 -
6765 - $html = '<div class="mxchat-youtube-embed" data-video-id="' . esc_attr($video_id) . '">';
6766 - $html .= '<button type="button" class="mxchat-youtube-facade" aria-label="' . esc_attr(sprintf(__('Play video: %s', 'mxchat'), $label)) . '">';
6767 - $html .= '<img class="mxchat-youtube-thumb" src="' . esc_url($thumb) . '" alt="' . esc_attr($label) . '" loading="lazy" />';
6768 - $html .= '<span class="mxchat-youtube-play" aria-hidden="true"><svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24" width="22" height="22" fill="currentColor" aria-hidden="true"><path d="M8 5v14l11-7z"/></svg></span>';
6769 - $html .= '</button>';
6770 - $html .= '<div class="mxchat-youtube-caption">';
6771 - $html .= '<span class="mxchat-youtube-title">' . esc_html($label) . '</span>';
6772 - $html .= '<a class="mxchat-youtube-link" href="' . esc_url($watch_url) . '" target="_blank" rel="noopener noreferrer">' . esc_html__('Watch on YouTube', 'mxchat') . '</a>';
6773 - $html .= '</div>';
6774 - $html .= '</div>';
6775 - return $html;
6776 -}
6777 -
6778 -/**
6779 - * Fetch and reassemble chunks for a URL from WordPress database
6780 - *
6781 - * @param string $source_url The source URL to fetch chunks for
6782 - * @param int $max_chunks Maximum number of chunks to return (0 = unlimited)
6783 - * @param int &$chunk_count Reference to store the actual number of chunks returned
6784 - * @return string Reassembled content from chunks
6785 - */
6786 -private function reassemble_chunks_from_wordpress($source_url, $max_chunks = 0, &$chunk_count = 0) {
6787 - global $wpdb;
6788 - $table = $wpdb->prefix . 'mxchat_system_prompt_content';
6789 -
6790 - // Fetch all rows with this source_url
6791 - $rows = $wpdb->get_results($wpdb->prepare(
6792 - "SELECT article_content FROM {$table}
6793 - WHERE source_url = %s
6794 - ORDER BY id ASC",
6795 - $source_url
6796 - ));
6797 -
6798 - if (empty($rows)) {
6799 - $chunk_count = 0;
6800 - return '';
6801 - }
6802 -
6803 - // Parse and sort chunks by index
6804 - $chunks = array();
6805 - foreach ($rows as $row) {
6806 - $parsed = MxChat_Chunker::parse_stored_chunk($row->article_content);
6807 -
6808 - if ($parsed['is_chunked']) {
6809 - $chunk_index = $parsed['metadata']['chunk_index'] ?? 0;
6810 - $chunks[$chunk_index] = $parsed['text'];
6811 - } else {
6812 - // Non-chunked content - just return it
6813 - $chunks[] = $parsed['text'];
6814 - }
6815 - }
6816 -
6817 - // Sort by chunk index
6818 - ksort($chunks);
6819 -
6820 - // Apply chunk limit if specified
6821 - if ($max_chunks > 0 && count($chunks) > $max_chunks) {
6822 - $chunks = array_slice($chunks, 0, $max_chunks, true);
6823 - }
6824 -
6825 - // Store actual chunk count
6826 - $chunk_count = count($chunks);
6827 -
6828 - // Reassemble content
6829 - return implode("\n\n", $chunks);
6830 -}
6831 -
6832 -private function find_relevant_content_pinecone($user_embedding, $bot_id = 'default', $bot_config = null) {
6833 - global $wpdb;
6834 -
6835 - //error_log("MXCHAT DEBUG: find_relevant_content_pinecone called");
6836 - //error_log(" - bot_id: " . $bot_id);
6837 - //error_log(" - user_embedding is array: " . (is_array($user_embedding) ? 'yes' : 'no'));
6838 - //error_log(" - user_embedding count: " . (is_array($user_embedding) ? count($user_embedding) : 'N/A'));
6839 -
6840 - // Use bot-specific config or fall back to default
6841 - if ($bot_config === null) {
6842 - $bot_config = $this->get_bot_pinecone_config($bot_id);
6843 - }
6844 -
6845 - $api_key = $bot_config['api_key'] ?? '';
6846 - $host = $bot_config['host'] ?? '';
6847 - $namespace = $bot_config['namespace'] ?? '';
6848 -
6849 - //error_log("MXCHAT DEBUG: Pinecone query parameters:");
6850 - //error_log(" - API Key: " . (empty($api_key) ? 'EMPTY - ERROR!' : 'Present (length: ' . strlen($api_key) . ')'));
6851 - //error_log(" - Host: " . (empty($host) ? 'EMPTY - ERROR!' : $host));
6852 - //error_log(" - Namespace: " . (empty($namespace) ? 'EMPTY (will use default)' : $namespace));
6853 -
6854 - // Initialize similarity analysis storage
6855 - $this->last_similarity_analysis = [
6856 - 'knowledge_base_type' => 'Pinecone',
6857 - 'bot_id' => $bot_id,
6858 - 'namespace' => $namespace,
6859 - 'top_matches' => [],
6860 - 'threshold_used' => 0,
6861 - 'total_checked' => 0
6862 - ];
6863 -
6864 - // NEW: Initialize valid URLs array
6865 - $valid_urls = [];
6866 -
6867 - if (empty($host) || empty($api_key)) {
6868 - //error_log("MXCHAT DEBUG ERROR: Missing Pinecone host or API key!");
6869 - //error_log(" - Host empty: " . (empty($host) ? 'YES' : 'NO'));
6870 - //error_log(" - API key empty: " . (empty($api_key) ? 'YES' : 'NO'));
6871 - // Store empty array for valid URLs since we can't proceed
6872 - $this->current_valid_urls = [];
6873 - return '';
6874 - }
6875 -
6876 - // Get knowledge manager instance for role checking
6877 - $knowledge_manager = MxChat_Knowledge_Manager::get_instance();
6878 -
6879 - // Get the similarity threshold from the bot options or main options
6880 - $bot_options = $this->get_bot_options($bot_id);
6881 - $current_options = !empty($bot_options) ? $bot_options : get_option('mxchat_options', []);
6882 -
6883 - $similarity_threshold = isset($current_options['similarity_threshold'])
6884 - ? ((int) $current_options['similarity_threshold']) / 100
6885 - : 0.35;
6886 -
6887 - $this->last_similarity_analysis['threshold_used'] = $similarity_threshold;
6888 -
6889 - // Prepare the query request for Pinecone
6890 - $api_endpoint = "https://{$host}/query";
6891 -
6892 - $request_body = array(
6893 - 'vector' => $user_embedding,
6894 - 'topK' => 50, // Increased for chunked content grouping - need more candidates to find top N unique URLs
6895 - 'includeMetadata' => true,
6896 - 'includeValues' => true
6897 - );
6898 -
6899 - // Add namespace if specified for this bot
6900 - if (!empty($namespace)) {
6901 - $request_body['namespace'] = $namespace;
6902 - }
6903 -
6904 - //error_log("MXCHAT DEBUG: About to call Pinecone API");
6905 - //error_log(" - Endpoint: " . $api_endpoint);
6906 - //error_log(" - Namespace in request: " . (!empty($namespace) ? $namespace : 'NOT SET'));
6907 -
6908 - $response = wp_remote_post($api_endpoint, array(
6909 - 'headers' => array(
6910 - 'Api-Key' => $api_key,
6911 - 'accept' => 'application/json',
6912 - 'content-type' => 'application/json'
6913 - ),
6914 - 'body' => wp_json_encode($request_body),
6915 - 'timeout' => 30
6916 - ));
6917 -
6918 - if (is_wp_error($response)) {
6919 - //error_log("MXCHAT DEBUG ERROR: WP Error in Pinecone request: " . $response->get_error_message());
6920 - // Store empty array for valid URLs
6921 - $this->current_valid_urls = [];
6922 - return '';
6923 - }
6924 -
6925 - $response_code = wp_remote_retrieve_response_code($response);
6926 - //error_log("MXCHAT DEBUG: Pinecone response code: " . $response_code);
6927 -
6928 - if ($response_code !== 200) {
6929 - $response_body = wp_remote_retrieve_body($response);
6930 - //error_log("MXCHAT DEBUG ERROR: Pinecone API error response: " . substr($response_body, 0, 500));
6931 - // Store empty array for valid URLs
6932 - $this->current_valid_urls = [];
6933 - return '';
6934 - }
6935 -
6936 - // ADD DETAILED DEBUG SECTION HERE
6937 - $response_body = wp_remote_retrieve_body($response);
6938 - //error_log("MXCHAT DEBUG: Raw Pinecone response length: " . strlen($response_body));
6939 -
6940 - $results = json_decode($response_body, true);
6941 -
6942 - if (json_last_error() !== JSON_ERROR_NONE) {
6943 - //error_log("MXCHAT DEBUG ERROR: JSON decode error: " . json_last_error_msg());
6944 - //error_log("MXCHAT DEBUG: First 500 chars of response: " . substr($response_body, 0, 500));
6945 - // Store empty array for valid URLs
6946 - $this->current_valid_urls = [];
6947 - return '';
6948 - }
6949 -
6950 - //error_log("MXCHAT DEBUG: Pinecone response structure:");
6951 - //error_log(" - Has 'matches' key: " . (isset($results['matches']) ? 'yes' : 'no'));
6952 - //error_log(" - Has 'namespace' key: " . (isset($results['namespace']) ? 'yes (' . $results['namespace'] . ')' : 'no'));
6953 -
6954 - if (empty($results['matches'])) {
6955 - //error_log("MXCHAT DEBUG: No matches found in Pinecone response");
6956 - //error_log("MXCHAT DEBUG: Response keys: " . implode(', ', array_keys($results)));
6957 - // Store empty array for valid URLs
6958 - $this->current_valid_urls = [];
6959 - return '';
6960 - }
6961 -
6962 - //error_log("MXCHAT DEBUG: Found " . count($results['matches']) . " matches in Pinecone");
6963 -
6964 - // Log first match details for debugging
6965 - if (!empty($results['matches'][0])) {
6966 - $first_match = $results['matches'][0];
6967 - //error_log("MXCHAT DEBUG: First match details:");
6968 - //error_log(" - Score: " . ($first_match['score'] ?? 'no score'));
6969 - //error_log(" - Has metadata: " . (isset($first_match['metadata']) ? 'yes' : 'no'));
6970 - if (isset($first_match['metadata'])) {
6971 - //error_log(" - Metadata keys: " . implode(', ', array_keys($first_match['metadata'])));
6972 - }
6973 - }
6974 -
6975 - // Initialize the final content
6976 - $content = '';
6977 - $matches_used = 0;
6978 - $matches_used_for_context = [];
6979 - $total_chunks_used = 0;
6980 - $max_total_chunks = isset($current_options['rag_chunks_limit']) ? intval($current_options['rag_chunks_limit']) : 15;
6981 - if ($max_total_chunks < 8) $max_total_chunks = 8;
6982 - if ($max_total_chunks > 20) $max_total_chunks = 20;
6983 - $max_chunks_per_source = 5; // Cap per individual source to limit token usage
6984 -
6985 - // Check if citation links are enabled (default to 'on' for backwards compatibility)
6986 - // Use fresh options to ensure we get the latest setting value
6987 - $fresh_options = get_option('mxchat_options', []);
6988 - $citation_links_enabled = isset($fresh_options['citation_links_toggle']) ? ($fresh_options['citation_links_toggle'] === 'on') : true;
6989 -
6990 - // NEW CHUNKING LOGIC: Group results by source_url for chunk reassembly
6991 - $url_groups = array();
6992 -
6993 - foreach ($results['matches'] as $index => $match) {
6994 - // Skip if similarity is below threshold
6995 - if ($match['score'] < $similarity_threshold) {
6996 - continue;
6997 - }
6998 -
6999 - $metadata = $match['metadata'] ?? array();
7000 - $source_url = $metadata['source_url'] ?? '';
7001 - $match_id = $match['id'] ?? '';
7002 -
7003 - // LAZY ROLE CHECK: Only check role for content we're actually considering
7004 - $role_restriction = $this->get_single_vector_role($match_id, $metadata);
7005 - $has_access = $knowledge_manager->mxchat_user_has_content_access($role_restriction);
7006 -
7007 - // Skip if user doesn't have access
7008 - if (!$has_access) {
7009 - continue;
7010 - }
7011 -
7012 - // Use a unique key for manual entries without a source URL
7013 - $group_key = !empty($source_url) ? $source_url : '_manual_' . $match_id;
7014 -
7015 - // Group by source URL (or unique key for manual entries)
7016 - if (!isset($url_groups[$group_key])) {
7017 - $url_groups[$group_key] = array(
7018 - 'source_url' => $source_url,
7019 - 'best_score' => 0,
7020 - 'is_chunked' => isset($metadata['is_chunked']) && $metadata['is_chunked'],
7021 - 'chunks' => array(),
7022 - 'single_text' => ''
7023 - );
7024 - }
7025 -
7026 - // Track best score for this group
7027 - if ($match['score'] > $url_groups[$group_key]['best_score']) {
7028 - $url_groups[$group_key]['best_score'] = $match['score'];
7029 - }
7030 -
7031 - // Store chunk info or single text
7032 - if ($url_groups[$group_key]['is_chunked']) {
7033 - $url_groups[$group_key]['chunks'][] = array(
7034 - 'id' => $match_id,
7035 - 'score' => $match['score'],
7036 - 'chunk_index' => $metadata['chunk_index'] ?? 0,
7037 - 'text' => $metadata['text'] ?? ''
7038 - );
7039 - } else {
7040 - // Non-chunked content - just store the text
7041 - $url_groups[$group_key]['single_text'] = $metadata['text'] ?? '';
7042 - $url_groups[$group_key]['single_id'] = $match_id;
7043 - }
7044 - }
7045 -
7046 - // Sort URL groups by best score (highest first)
7047 - uasort($url_groups, function($a, $b) {
7048 - return $b['best_score'] <=> $a['best_score'];
7049 - });
7050 -
7051 - // Get RAG sources limit from options (default 6, min 3, max 10)
7052 - $rag_sources_limit = isset($current_options['rag_sources_limit']) ? intval($current_options['rag_sources_limit']) : 3;
7053 - if ($rag_sources_limit < 3) $rag_sources_limit = 3;
7054 - if ($rag_sources_limit > 10) $rag_sources_limit = 10;
7055 -
7056 - // Take top N unique URLs based on user setting
7057 - $top_urls = array_slice($url_groups, 0, $rag_sources_limit, true);
7058 -
7059 - // Track which match IDs are actually used for context
7060 - foreach ($top_urls as $group) {
7061 - if ($group['is_chunked']) {
7062 - foreach ($group['chunks'] as $chunk) {
7063 - $matches_used_for_context[] = $chunk['id'];
7064 - }
7065 - } elseif (!empty($group['single_id'])) {
7066 - $matches_used_for_context[] = $group['single_id'];
7067 - }
7068 - }
7069 -
7070 - // Build content from top sources
7071 - foreach ($top_urls as $group_key => $group) {
7072 - $source_url = $group['source_url']; // Use actual source_url, not the group key
7073 -
7074 - // Stop if we've hit the total chunk limit
7075 - if ($total_chunks_used >= $max_total_chunks) {
7076 - break;
7077 - }
7078 -
7079 - $full_text = '';
7080 - $chunks_in_this_source = 1; // Default for non-chunked content
7081 -
7082 - if ($group['is_chunked']) {
7083 - // Calculate how many chunks we can still use (respect both total and per-source caps)
7084 - $chunks_remaining = min($max_chunks_per_source, $max_total_chunks - $total_chunks_used);
7085 -
7086 - // Fetch chunks for this URL with limit
7087 - $full_text = $this->reassemble_chunks_from_pinecone($source_url, $bot_config, $chunks_remaining, $chunks_in_this_source);
7088 -
7089 - // If fetching all chunks fails, fall back to matched chunks
7090 - if (empty($full_text)) {
7091 - // Sort matched chunks by index and concatenate
7092 - usort($group['chunks'], function($a, $b) {
7093 - return $a['chunk_index'] <=> $b['chunk_index'];
7094 - });
7095 -
7096 - $chunk_texts = array();
7097 - $chunks_in_this_source = 0;
7098 - foreach ($group['chunks'] as $chunk) {
7099 - if ($total_chunks_used + $chunks_in_this_source >= $max_total_chunks) {
7100 - break;
7101 - }
7102 - $chunk_texts[] = $chunk['text'];
7103 - $chunks_in_this_source++;
7104 - }
7105 - $full_text = implode("\n\n", $chunk_texts);
7106 - }
7107 - } else {
7108 - $full_text = $group['single_text'];
7109 - $chunks_in_this_source = 1;
7110 - }
7111 -
7112 - if (!empty($full_text)) {
7113 - // Strip URLs from content if citation links are disabled
7114 - if (!$citation_links_enabled) {
7115 - $full_text = preg_replace('#\bhttps?://[^\s<>"\']+#i', '', $full_text);
7116 - $full_text = preg_replace('/\s+/', ' ', trim($full_text)); // Clean up extra spaces
7117 - }
7118 -
7119 - // Use numbered reference for URL-based entries, plain info label for manual entries
7120 - // Manual entries are stored with an internal mxchat:// placeholder URL — never expose them as citations
7121 - if (!empty($source_url) && $source_url !== '#' && strpos($source_url, 'mxchat://') !== 0) {
7122 - $matches_used++;
7123 - $content .= "## Reference " . $matches_used . " ##\n";
7124 - $content .= $full_text . "\n\n";
7125 -
7126 - // Only include citation URLs if citation links are enabled
7127 - if ($citation_links_enabled) {
7128 - $valid_urls[] = $source_url;
7129 - $content .= "URL: " . $source_url . "\n\n";
7130 - }
7131 -
7132 - // Video-backed source → queue the consent-safe embed (03ba33)
7133 - $this->maybe_queue_youtube_embed($source_url, $full_text);
7134 - } else {
7135 - // Manual entry — no reference number, no citation. Count it as a USED
7136 - // source (plan-mxchat-20260622-c1fe6a): without this, manual/Direct-Content
7137 - // entries (empty or mxchat:// source_url) never increment $matches_used, so
7138 - // the gate below (`if ($matches_used === 0)`) discards manual-only context on
7139 - // the Pinecone backend and the model is told "No reference information was
7140 - // found" — even though the testing panel reports used_for_context:true. It
7141 - // also corrects the cosmetic sources_used:0 the panel/transcript showed. The
7142 - // sibling local/WP-DB builder gates on empty($top_urls), so it never had this
7143 - // bug; this brings Pinecone to parity. Manual entries are still uncited (not
7144 - // added to $valid_urls, no "URL:" line).
7145 - $matches_used++;
7146 - $content .= "## Information ##\n";
7147 - $content .= $full_text . "\n\n";
7148 - }
7149 -
7150 - // Extract any URLs from the text content itself (only if citation links enabled)
7151 - if ($citation_links_enabled) {
7152 - preg_match_all(
7153 - '#\bhttps?://[^\s<>"\']+#i',
7154 - $full_text,
7155 - $content_urls
7156 - );
7157 - if (!empty($content_urls[0])) {
7158 - $valid_urls = array_merge($valid_urls, $content_urls[0]);
7159 - }
7160 - }
7161 -
7162 - $total_chunks_used += $chunks_in_this_source;
7163 - }
7164 - }
7165 -
7166 - // Process ALL matches for testing data (top 10) - with role checking for testing display
7167 - $all_matches = [];
7168 - foreach ($results['matches'] as $index => $match) {
7169 - if ($index >= 10) break; // Limit to top 10 for testing
7170 -
7171 - $match_id = $match['id'] ?? '';
7172 -
7173 - // Check role access for testing display (use cache if available)
7174 - $role_restriction = $this->get_single_vector_role($match_id, $match['metadata']);
7175 - $has_access = $knowledge_manager->mxchat_user_has_content_access($role_restriction);
7176 -
7177 - $source_display = '';
7178 - if (!empty($match['metadata']['source_url'])) {
7179 - $source_display = $match['metadata']['source_url'];
7180 - } else {
7181 - $content_preview = strip_tags($match['metadata']['text'] ?? '');
7182 - $content_preview = preg_replace('/\s+/', ' ', $content_preview);
7183 - $source_display = substr(trim($content_preview), 0, 50) . '...';
7184 - }
7185 -
7186 - $match_id_for_display = $match['id'] ?? $index;
7187 -
7188 - // Check for chunk metadata in Pinecone
7189 - $is_chunk = isset($match['metadata']['is_chunked']) && $match['metadata']['is_chunked'];
7190 - $chunk_index = isset($match['metadata']['chunk_index']) ? intval($match['metadata']['chunk_index']) : null;
7191 - $total_chunks = isset($match['metadata']['total_chunks']) ? intval($match['metadata']['total_chunks']) : null;
7192 -
7193 - // Also detect chunk from vector ID pattern: {hash}_chunk_{index}
7194 - if (!$is_chunk && MxChat_Chunker::is_chunk_vector_id($match_id_for_display)) {
7195 - $is_chunk = true;
7196 - }
7197 -
7198 - $all_matches[] = [
7199 - 'document_id' => $match_id_for_display,
7200 - 'similarity' => $match['score'],
7201 - 'similarity_percentage' => round($match['score'] * 100, 2),
7202 - 'above_threshold' => $match['score'] >= $similarity_threshold,
7203 - 'source_display' => $source_display,
7204 - 'content_preview' => substr(strip_tags($match['metadata']['text'] ?? ''), 0, 100) . '...',
7205 - 'used_for_context' => in_array($match_id_for_display, $matches_used_for_context),
7206 - 'role_restriction' => $role_restriction,
7207 - 'has_access' => $has_access,
7208 - 'filtered_out' => !$has_access,
7209 - 'is_chunk' => $is_chunk,
7210 - 'chunk_index' => $chunk_index,
7211 - 'total_chunks' => $total_chunks
7212 - ];
7213 - }
7214 -
7215 - // Store for testing panel
7216 - $this->last_similarity_analysis['top_matches'] = $all_matches;
7217 - $this->last_similarity_analysis['total_checked'] = count($results['matches']);
7218 - $this->last_similarity_analysis['sources_used'] = $matches_used;
7219 - $this->last_similarity_analysis['total_chunks_used'] = $total_chunks_used;
7220 -
7221 - // NEW: Store unique valid URLs for validation
7222 - $this->current_valid_urls = array_unique($valid_urls);
7223 -
7224 - // Allow add-ons to act on similarity results (e.g. WooCommerce product card display)
7225 - do_action('mxchat_similarity_results', $this->last_similarity_analysis['top_matches'], $bot_id);
7226 -
7227 - // Add response guidelines
7228 - if ($matches_used === 0) {
7229 - // Empty return → assembler's NO RELEVANT CONTENT branch (plan d7daf8).
7230 - $content = '';
7231 - } else {
7232 - // Build response guidelines based on citation links setting
7233 - $content .= "\n## Response Guidelines ##\n" .
7234 - "You are an AI Chatbot. Answer naturally and helpfully using only the information from the references above. " .
7235 - "Be conversational and friendly, but never mention your knowledge base or training data. " .
7236 - "If you don't have specific information or are uncertain about any details, it's always " .
7237 - "better to honestly say you don't know rather than making up or guessing at answers. " .
7238 - "When information is incomplete, let them know you are unsure.\n\n";
7239 -
7240 - // Only add hyperlink instructions if citation links are enabled
7241 - if ($citation_links_enabled) {
7242 - $content .= "CRITICAL: When creating hyperlinks, always use proper markdown format with descriptive text: " .
7243 - "[descriptive text](url). NEVER use empty brackets like [](url). The text in brackets must describe what the link is about. " .
7244 - "Only cite references that have a URL. Do not cite or add source labels to Information sections that have no URL.";
7245 - } else {
7246 - $content .= "IMPORTANT: Do not include any citation links, source URLs, or hyperlinks in your responses. " .
7247 - "Simply provide helpful answers based on the reference information without citing sources.";
7248 - }
7249 - }
7250 -
7251 - return trim($content);
7252 -}
7253 -
7254 -/**
7255 - * Get role restriction for a single vector (with caching)
7256 - */
7257 -private function get_single_vector_role($vector_id, $metadata = array()) {
7258 - global $wpdb;
7259 -
7260 - if (empty($vector_id)) {
7261 - return 'public';
7262 - }
7263 -
7264 - // Check cache first
7265 - $cache_key = 'mxchat_vector_role_' . $vector_id;
7266 - $cached_role = wp_cache_get($cache_key, 'mxchat_vector_roles');
7267 -
7268 - if ($cached_role !== false) {
7269 - return $cached_role;
7270 - }
7271 -
7272 - $role_restriction = 'public';
7273 -
7274 - // First try Pinecone metadata
7275 - if (!empty($metadata['role_restriction'])) {
7276 - $role_restriction = $metadata['role_restriction'];
7277 - } else {
7278 - // Check WordPress table for user-modified roles
7279 - $roles_table = $wpdb->prefix . 'mxchat_pinecone_roles';
7280 - $stored_role = $wpdb->get_var($wpdb->prepare(
7281 - "SELECT role_restriction FROM {$roles_table} WHERE vector_id = %s",
7282 - $vector_id
7283 - ));
7284 -
7285 - if ($stored_role) {
7286 - $role_restriction = $stored_role;
7287 - }
7288 - }
7289 -
7290 - // Cache individual role for 1 hour
7291 - wp_cache_set($cache_key, $role_restriction, 'mxchat_vector_roles', 3600);
7292 -
7293 - return $role_restriction;
7294 -}
7295 -
7296 -/**
7297 - * Fetch and reassemble all chunks for a URL from Pinecone
7298 - *
7299 - * @param string $source_url The source URL to fetch chunks for
7300 - * @param array $bot_config Bot-specific Pinecone configuration
7301 - * @return string Reassembled content from all chunks
7302 - */
7303 -private function reassemble_chunks_from_pinecone($source_url, $bot_config, $max_chunks = 0, &$chunk_count = 0) {
7304 - $api_key = $bot_config['api_key'] ?? '';
7305 - $host = $bot_config['host'] ?? '';
7306 - $namespace = $bot_config['namespace'] ?? '';
7307 -
7308 - if (empty($host) || empty($api_key)) {
7309 - $chunk_count = 0;
7310 - return '';
7311 - }
7312 -
7313 - $base_hash = md5($source_url);
7314 -
7315 - // Use Pinecone list API to find all chunk vectors with this prefix
7316 - $list_url = "https://{$host}/vectors/list";
7317 -
7318 - // Limit to max_chunks if specified, otherwise fetch up to 100
7319 - $fetch_limit = ($max_chunks > 0 && $max_chunks < 100) ? $max_chunks : 100;
7320 -
7321 - $list_body = array(
7322 - 'prefix' => $base_hash . '_chunk_',
7323 - 'limit' => $fetch_limit
7324 - );
7325 -
7326 - if (!empty($namespace)) {
7327 - $list_body['namespace'] = $namespace;
7328 - }
7329 -
7330 - $list_response = wp_remote_post($list_url, array(
7331 - 'headers' => array(
7332 - 'Api-Key' => $api_key,
7333 - 'accept' => 'application/json',
7334 - 'content-type' => 'application/json'
7335 - ),
7336 - 'body' => wp_json_encode($list_body),
7337 - 'timeout' => 30
7338 - ));
7339 -
7340 - if (is_wp_error($list_response)) {
7341 - //error_log('[MXCHAT-CHUNK] List API error: ' . $list_response->get_error_message());
7342 - return '';
7343 - }
7344 -
7345 - $list_data = json_decode(wp_remote_retrieve_body($list_response), true);
7346 -
7347 - if (empty($list_data['vectors'])) {
7348 - //error_log('[MXCHAT-CHUNK] No chunk vectors found for URL: ' . $source_url);
7349 - return '';
7350 - }
7351 -
7352 - // Extract vector IDs
7353 - $vector_ids = array();
7354 - foreach ($list_data['vectors'] as $vector) {
7355 - if (isset($vector['id'])) {
7356 - $vector_ids[] = $vector['id'];
7357 - }
7358 - }
7359 -
7360 - if (empty($vector_ids)) {
7361 - return '';
7362 - }
7363 -
7364 - // Fetch all chunk content
7365 - $fetch_url = "https://{$host}/vectors/fetch";
7366 -
7367 - $fetch_body = array(
7368 - 'ids' => $vector_ids
7369 - );
7370 -
7371 - if (!empty($namespace)) {
7372 - $fetch_body['namespace'] = $namespace;
7373 - }
7374 -
7375 - $fetch_response = wp_remote_post($fetch_url, array(
7376 - 'headers' => array(
7377 - 'Api-Key' => $api_key,
7378 - 'accept' => 'application/json',
7379 - 'content-type' => 'application/json'
7380 - ),
7381 - 'body' => wp_json_encode($fetch_body),
7382 - 'timeout' => 30
7383 - ));
7384 -
7385 - if (is_wp_error($fetch_response)) {
7386 - //error_log('[MXCHAT-CHUNK] Fetch API error: ' . $fetch_response->get_error_message());
7387 - return '';
7388 - }
7389 -
7390 - $fetch_data = json_decode(wp_remote_retrieve_body($fetch_response), true);
7391 -
7392 - if (empty($fetch_data['vectors'])) {
7393 - return '';
7394 - }
7395 -
7396 - // Sort chunks by index and reassemble
7397 - $chunks = array();
7398 - foreach ($fetch_data['vectors'] as $id => $vector) {
7399 - $metadata = $vector['metadata'] ?? array();
7400 - $chunk_index = $metadata['chunk_index'] ?? 0;
7401 - $text = $metadata['text'] ?? '';
7402 -
7403 - // Store chunk with its index
7404 - $chunks[$chunk_index] = $text;
7405 - }
7406 -
7407 - // Sort by chunk index
7408 - ksort($chunks);
7409 -
7410 - // Apply chunk limit if specified
7411 - if ($max_chunks > 0 && count($chunks) > $max_chunks) {
7412 - $chunks = array_slice($chunks, 0, $max_chunks, true);
7413 - }
7414 -
7415 - // Store actual chunk count
7416 - $chunk_count = count($chunks);
7417 -
7418 - // Reassemble content
7419 - return implode("\n\n", $chunks);
7420 -}
7421 -
7422 -/**
7423 - * Search for relevant content using OpenAI Vector Store (File Search)
7424 - *
7425 - * @param string $user_query The user's query text
7426 - * @param string $bot_id The bot ID
7427 - * @param array $vectorstore_config Vector Store configuration
7428 - * @return string Formatted context string with references
7429 - */
7430 -private function find_relevant_content_openai_vectorstore($user_query, $bot_id = 'default', $vectorstore_config = array()) {
7431 - //error_log("MXCHAT DEBUG: find_relevant_content_openai_vectorstore called");
7432 - //error_log(" - bot_id: " . $bot_id);
7433 - //error_log(" - user_query length: " . strlen($user_query));
7434 -
7435 - // Get OpenAI API key
7436 - $mxchat_options = get_option('mxchat_options', array());
7437 - $api_key = $mxchat_options['api_key'] ?? '';
7438 -
7439 - // Reset vectorstore error tracking
7440 - $this->last_vectorstore_error = null;
7441 -
7442 - if (empty($api_key)) {
7443 - //error_log("MXCHAT DEBUG ERROR: OpenAI API key not configured");
7444 - $this->last_vectorstore_error = 'Vector Store search failed: OpenAI API key is not configured.';
7445 - $this->current_valid_urls = [];
7446 - return '';
7447 - }
7448 -
7449 - // Get Vector Store configuration
7450 - if (empty($vectorstore_config)) {
7451 - $vectorstore_config = $this->get_bot_vectorstore_config($bot_id);
7452 - }
7453 -
7454 - $vectorstore_ids_string = $vectorstore_config['vectorstore_ids'] ?? '';
7455 - $max_results = $vectorstore_config['max_results'] ?? 5;
7456 -
7457 - if (empty($vectorstore_ids_string)) {
7458 - //error_log("MXCHAT DEBUG ERROR: No Vector Store IDs configured");
7459 - $this->last_vectorstore_error = 'Vector Store search failed: No Vector Store IDs are configured for this bot.';
7460 - $this->current_valid_urls = [];
7461 - return '';
7462 - }
7463 -
7464 - // Parse Vector Store IDs
7465 - $vectorstore_ids = array_map('trim', explode(',', $vectorstore_ids_string));
7466 - $vectorstore_ids = array_filter($vectorstore_ids); // Remove empty values
7467 -
7468 - //error_log("MXCHAT DEBUG: Vector Store IDs: " . implode(', ', $vectorstore_ids));
7469 - //error_log("MXCHAT DEBUG: Max results: " . $max_results);
7470 -
7471 - // Initialize similarity analysis storage
7472 - $this->last_similarity_analysis = [
7473 - 'knowledge_base_type' => 'OpenAI Vector Store',
7474 - 'bot_id' => $bot_id,
7475 - 'vectorstore_ids' => $vectorstore_ids,
7476 - 'top_matches' => [],
7477 - 'threshold_used' => 0,
7478 - 'total_checked' => 0
7479 - ];
7480 -
7481 - $valid_urls = [];
7482 -
7483 - // Get the selected model
7484 - $bot_options = $this->get_bot_options($bot_id);
7485 - $current_options = !empty($bot_options) ? $bot_options : $mxchat_options;
7486 - $selected_model = $current_options['model'] ?? 'gpt-5.6-sol';
7487 -
7488 - // Verify it's an OpenAI model
7489 - if (!$this->is_openai_chat_model($selected_model)) {
7490 - //error_log("MXCHAT DEBUG ERROR: Vector Store search requires OpenAI model. Current: " . $selected_model);
7491 - $this->last_vectorstore_error = 'Vector Store search requires an OpenAI model. Current model: ' . $selected_model;
7492 - $this->current_valid_urls = [];
7493 - return '';
7494 - }
7495 -
7496 - // Use OpenAI Responses API with file_search tool
7497 - $request_body = array(
7498 - 'model' => $selected_model,
7499 - 'input' => $user_query,
7500 - 'tools' => array(
7501 - array(
7502 - 'type' => 'file_search',
7503 - 'vector_store_ids' => $vectorstore_ids,
7504 - 'max_num_results' => intval($max_results)
7505 - )
7506 - ),
7507 - 'include' => array('output[*].file_search_call.search_results')
7508 - );
7509 -
7510 - //error_log("MXCHAT VECTORSTORE: ========== REQUEST START ==========");
7511 - //error_log("MXCHAT VECTORSTORE: Model: " . $selected_model);
7512 - //error_log("MXCHAT VECTORSTORE: Query: " . substr($user_query, 0, 200));
7513 - //error_log("MXCHAT VECTORSTORE: Vector Store IDs: " . implode(', ', $vectorstore_ids));
7514 - //error_log("MXCHAT VECTORSTORE: Max Results: " . $max_results);
7515 - //error_log("MXCHAT VECTORSTORE: Request body: " . wp_json_encode($request_body));
7516 -
7517 - $response = wp_remote_post('https://api.openai.com/v1/responses', array(
7518 - 'headers' => array(
7519 - 'Authorization' => 'Bearer ' . $api_key,
7520 - 'Content-Type' => 'application/json'
7521 - ),
7522 - 'body' => wp_json_encode($request_body),
7523 - 'timeout' => 60
7524 - ));
7525 -
7526 - if (is_wp_error($response)) {
7527 - //error_log("MXCHAT VECTORSTORE ERROR: WP Error: " . $response->get_error_message());
7528 - $this->last_vectorstore_error = 'Vector Store API request failed: ' . $response->get_error_message();
7529 - $this->current_valid_urls = [];
7530 - return '';
7531 - }
7532 -
7533 - $response_code = wp_remote_retrieve_response_code($response);
7534 - //error_log("MXCHAT VECTORSTORE: Response code: " . $response_code);
7535 -
7536 - $response_body = wp_remote_retrieve_body($response);
7537 - //error_log("MXCHAT VECTORSTORE: Raw response (first 2000 chars): " . substr($response_body, 0, 2000));
7538 -
7539 - if ($response_code !== 200) {
7540 - //error_log("MXCHAT VECTORSTORE ERROR: API error response: " . $response_body);
7541 - $decoded_error = json_decode($response_body, true);
7542 - $api_error_detail = $this->extract_provider_error($decoded_error, '');
7543 - $this->last_vectorstore_error = 'Vector Store API returned HTTP ' . $response_code . ($api_error_detail ? ': ' . $api_error_detail : '');
7544 - $this->current_valid_urls = [];
7545 - return '';
7546 - }
7547 - $result = json_decode($response_body, true);
7548 -
7549 - if (json_last_error() !== JSON_ERROR_NONE) {
7550 - //error_log("MXCHAT VECTORSTORE ERROR: JSON decode error: " . json_last_error_msg());
7551 - $this->last_vectorstore_error = 'Vector Store response could not be parsed: ' . json_last_error_msg();
7552 - $this->current_valid_urls = [];
7553 - return '';
7554 - }
7555 -
7556 - // Debug: Log the structure of the result
7557 - //error_log("MXCHAT VECTORSTORE: Result keys: " . implode(', ', array_keys($result)));
7558 - if (isset($result['output'])) {
7559 - //error_log("MXCHAT VECTORSTORE: Output count: " . count($result['output']));
7560 - foreach ($result['output'] as $idx => $out) {
7561 - //error_log("MXCHAT VECTORSTORE: Output[$idx] type: " . ($out['type'] ?? 'unknown'));
7562 - //error_log("MXCHAT VECTORSTORE: Output[$idx] keys: " . implode(', ', array_keys($out)));
7563 - }
7564 - } else {
7565 - //error_log("MXCHAT VECTORSTORE: No 'output' key in result!");
7566 - }
7567 -
7568 - // Extract file search results from the response
7569 - $content = '';
7570 - $matches_used = 0;
7571 - $all_matches = [];
7572 -
7573 - // The Responses API returns output array with tool results
7574 - if (isset($result['output']) && is_array($result['output'])) {
7575 - foreach ($result['output'] as $output_item) {
7576 - // Look for file_search_call results
7577 - if (isset($output_item['type']) && $output_item['type'] === 'file_search_call') {
7578 - //error_log("MXCHAT VECTORSTORE: Found file_search_call output item");
7579 - //error_log("MXCHAT VECTORSTORE: file_search_call keys: " . implode(', ', array_keys($output_item)));
7580 -
7581 - // Check for search_results in the output item directly
7582 - $search_results = $output_item['search_results'] ?? $output_item['results'] ?? [];
7583 - //error_log("MXCHAT VECTORSTORE: Search results count: " . count($search_results));
7584 -
7585 - if (empty($search_results)) {
7586 - //error_log("MXCHAT VECTORSTORE: No search results found in file_search_call");
7587 - //error_log("MXCHAT VECTORSTORE: file_search_call content: " . wp_json_encode($output_item));
7588 - }
7589 -
7590 - foreach ($search_results as $index => $search_result) {
7591 - $filename = $search_result['filename'] ?? '';
7592 - $score = $search_result['score'] ?? 0;
7593 - $text_content = '';
7594 -
7595 - // Extract text content from the result
7596 - // The text can be directly on the result OR nested under content array
7597 - if (isset($search_result['text']) && !empty($search_result['text'])) {
7598 - // Direct text field (OpenAI's actual format)
7599 - $text_content = $search_result['text'];
7600 - //error_log("MXCHAT VECTORSTORE: Found text directly on result[$index], length: " . strlen($text_content));
7601 - } elseif (isset($search_result['content']) && is_array($search_result['content'])) {
7602 - // Nested content array format
7603 - foreach ($search_result['content'] as $content_item) {
7604 - if (isset($content_item['text'])) {
7605 - $text_content .= $content_item['text'] . "\n";
7606 - }
7607 - }
7608 - //error_log("MXCHAT VECTORSTORE: Found text in content array for result[$index], length: " . strlen($text_content));
7609 - } else {
7610 - //error_log("MXCHAT VECTORSTORE: No text found for result[$index]. Keys: " . implode(', ', array_keys($search_result)));
7611 - }
7612 -
7613 - if (!empty($text_content)) {
7614 - $content .= "## Reference " . ($matches_used + 1) . " ##\n";
7615 - $content .= trim($text_content) . "\n\n";
7616 -
7617 - if (!empty($filename)) {
7618 - $content .= "Source: " . $filename . "\n\n";
7619 - }
7620 -
7621 - // Extract URLs from content
7622 - preg_match_all(
7623 - '#\bhttps?://[^\s<>"\']+#i',
7624 - $text_content,
7625 - $content_urls
7626 - );
7627 - if (!empty($content_urls[0])) {
7628 - $valid_urls = array_merge($valid_urls, $content_urls[0]);
7629 - }
7630 -
7631 - $matches_used++;
7632 - }
7633 -
7634 - // Store for similarity analysis
7635 - $all_matches[] = [
7636 - 'document_id' => $filename ?: ('result_' . $index),
7637 - 'similarity' => $score,
7638 - 'similarity_percentage' => round($score * 100, 2),
7639 - 'above_threshold' => true,
7640 - 'source_display' => $filename,
7641 - 'content_preview' => substr(strip_tags($text_content), 0, 100) . '...',
7642 - 'used_for_context' => true,
7643 - 'role_restriction' => 'public',
7644 - 'has_access' => true,
7645 - 'filtered_out' => false
7646 - ];
7647 - }
7648 - }
7649 -
7650 - // Also check for message content with annotations (citations)
7651 - if (isset($output_item['type']) && $output_item['type'] === 'message') {
7652 - if (isset($output_item['content']) && is_array($output_item['content'])) {
7653 - foreach ($output_item['content'] as $content_block) {
7654 - if (isset($content_block['annotations']) && is_array($content_block['annotations'])) {
7655 - foreach ($content_block['annotations'] as $annotation) {
7656 - if (isset($annotation['filename'])) {
7657 - $filename = $annotation['filename'];
7658 - $score = $annotation['score'] ?? 0;
7659 - $text_content = '';
7660 -
7661 - if (isset($annotation['content']) && is_array($annotation['content'])) {
7662 - foreach ($annotation['content'] as $ann_content) {
7663 - if (isset($ann_content['text'])) {
7664 - $text_content .= $ann_content['text'] . "\n";
7665 - }
7666 - }
7667 - }
7668 -
7669 - if (!empty($text_content) && $matches_used < $max_results) {
7670 - $content .= "## Reference " . ($matches_used + 1) . " ##\n";
7671 - $content .= trim($text_content) . "\n\n";
7672 - $content .= "Source: " . $filename . "\n\n";
7673 -
7674 - preg_match_all(
7675 - '#\bhttps?://[^\s<>"\']+#i',
7676 - $text_content,
7677 - $content_urls
7678 - );
7679 - if (!empty($content_urls[0])) {
7680 - $valid_urls = array_merge($valid_urls, $content_urls[0]);
7681 - }
7682 -
7683 - $matches_used++;
7684 -
7685 - $all_matches[] = [
7686 - 'document_id' => $filename,
7687 - 'similarity' => $score,
7688 - 'similarity_percentage' => round($score * 100, 2),
7689 - 'above_threshold' => true,
7690 - 'source_display' => $filename,
7691 - 'content_preview' => substr(strip_tags($text_content), 0, 100) . '...',
7692 - 'used_for_context' => true,
7693 - 'role_restriction' => 'public',
7694 - 'has_access' => true,
7695 - 'filtered_out' => false
7696 - ];
7697 - }
7698 - }
7699 - }
7700 - }
7701 - }
7702 - }
7703 - }
7704 - }
7705 - }
7706 -
7707 - // Store for testing panel
7708 - $this->last_similarity_analysis['top_matches'] = $all_matches;
7709 - $this->last_similarity_analysis['total_checked'] = count($all_matches);
7710 -
7711 - // Store unique valid URLs for validation
7712 - $this->current_valid_urls = array_unique($valid_urls);
7713 -
7714 - // Allow add-ons to act on similarity results (e.g. WooCommerce product card display)
7715 - do_action('mxchat_similarity_results', $this->last_similarity_analysis['top_matches'], $bot_id);
7716 -
7717 - //error_log("MXCHAT VECTORSTORE: ========== SEARCH COMPLETE ==========");
7718 - //error_log("MXCHAT VECTORSTORE: Matches used: " . $matches_used);
7719 - //error_log("MXCHAT VECTORSTORE: All matches count: " . count($all_matches));
7720 - //error_log("MXCHAT VECTORSTORE: Content length: " . strlen($content));
7721 - if ($matches_used > 0) {
7722 - //error_log("MXCHAT VECTORSTORE: Content preview: " . substr($content, 0, 500));
7723 - }
7724 -
7725 - // Check if citation links are enabled
7726 - $citation_links_enabled = ($mxchat_options['citation_links_toggle'] ?? 'on') === 'on';
7727 -
7728 - // Add response guidelines
7729 - if ($matches_used === 0) {
7730 - //error_log("MXCHAT VECTORSTORE: No matches found - returning empty reference message");
7731 - // Empty return → assembler's NO RELEVANT CONTENT branch (plan d7daf8).
7732 - $content = '';
7733 - } else {
7734 - // Build response guidelines based on citation links setting
7735 - $content .= "\n## Response Guidelines ##\n" .
7736 - "You are an AI Chatbot. Answer naturally and helpfully using only the information from the references above. " .
7737 - "Be conversational and friendly, but never mention your knowledge base or training data. " .
7738 - "If you don't have specific information or are uncertain about any details, it's always " .
7739 - "better to honestly say you don't know rather than making up or guessing at answers. " .
7740 - "When information is incomplete, let them know you are unsure.\n\n";
7741 -
7742 - // Only add hyperlink instructions if citation links are enabled
7743 - if ($citation_links_enabled) {
7744 - $content .= "CRITICAL: When creating hyperlinks, always use proper markdown format with descriptive text: " .
7745 - "[descriptive text](url). NEVER use empty brackets like [](url). The text in brackets must describe what the link is about.";
7746 - } else {
7747 - $content .= "IMPORTANT: Do not include any citation links, source URLs, or hyperlinks in your responses. " .
7748 - "Simply provide helpful answers based on the reference information without citing sources.";
7749 - }
7750 - }
7751 -
7752 - //error_log("MXCHAT DEBUG: Vector Store search complete. Matches used: " . $matches_used);
7753 -
7754 - return trim($content);
7755 -}
7756 -
7757 -/**
7758 - * Check if the given model is an OpenAI chat model
7759 - *
7760 - * @param string $model The model ID
7761 - * @return bool True if it's an OpenAI model
7762 - */
7763 -private function is_openai_chat_model($model) {
7764 - $openai_prefixes = array('gpt-', 'o1-', 'o3-');
7765 - foreach ($openai_prefixes as $prefix) {
7766 - if (strpos($model, $prefix) === 0) {
7767 - return true;
7768 - }
7769 - }
7770 - return false;
7771 -}
7772 -
7773 -/**
7774 - * Get bot-specific Vector Store configuration
7775 - *
7776 - * @param string $bot_id The bot ID
7777 - * @return array Configuration array
7778 - */
7779 -private function get_bot_vectorstore_config($bot_id = 'default') {
7780 - // Admin Testing tab bot → resolve the DEFAULT bot's backend (see
7781 - // get_bot_pinecone_config). This getter already passes the real default
7782 - // config into the filter, so it was not broken — normalized anyway so the
7783 - // Testing bot can never drift from the front-end default.
7784 - if ($bot_id === 'testing') {
7785 - $bot_id = 'default';
7786 - }
7787 -
7788 - $vectorstore_options = get_option('mxchat_openai_vectorstore_options', array());
7789 -
7790 - // Default global settings
7791 - $default_config = array(
7792 - 'use_vectorstore' => ($vectorstore_options['mxchat_use_openai_vectorstore'] ?? '0') === '1',
7793 - 'vectorstore_ids' => $vectorstore_options['mxchat_vectorstore_ids'] ?? '',
7794 - 'max_results' => $vectorstore_options['mxchat_vectorstore_max_results'] ?? 5
7795 - );
7796 -
7797 - // Allow multi-bot plugin to override with bot-specific settings
7798 - $bot_config = apply_filters('mxchat_get_bot_vectorstore_config', $default_config, $bot_id);
7799 -
7800 - // Preserve max_results from global settings if not set in bot config
7801 - if (!isset($bot_config['max_results'])) {
7802 - $bot_config['max_results'] = $default_config['max_results'];
7803 - }
7804 -
7805 - return $bot_config;
7806 -}
7807 -
7808 -private function mxchat_find_relevant_products($user_embedding) {
7809 - //error_log('MXChat Vector Search: Starting product search...');
7810 -
7811 - // Retrieve the add-on settings from the database
7812 - $addon_options = get_option('mxchat_pinecone_addon_options', array());
7813 -
7814 - // Determine whether Pinecone is enabled
7815 - $use_pinecone = (isset($addon_options['mxchat_use_pinecone']) && $addon_options['mxchat_use_pinecone'] === '1') ? 1 : 0;
7816 -
7817 - //error_log('Pinecone enabled flag: ' . $use_pinecone);
7818 -
7819 - if ($use_pinecone === 1) {
7820 - //error_log('MXChat Vector Search: Using Pinecone database for products');
7821 - return $this->find_relevant_products_pinecone($user_embedding);
7822 - } else {
7823 - //error_log('MXChat Vector Search: Using WordPress database for products');
7824 - return $this->find_relevant_products_wordpress($user_embedding);
7825 - }
7826 -}
7827 -private function find_relevant_products_wordpress($user_embedding) {
7828 - global $wpdb;
7829 - $system_prompt_table = $wpdb->prefix . 'mxchat_system_prompt_content';
7830 -
7831 - if (!is_array($user_embedding)) {
7832 - return '';
7833 - }
7834 -
7835 - // Streaming top-K pass: scan rows in small batches, keep only the top 3
7836 - // results above the similarity threshold. Peak memory is bounded by
7837 - // $batch_size embedding rows plus a 3-element top list.
7838 - $batch_size = 250;
7839 - $similarity_threshold = 0.85;
7840 - $top_k = 3;
7841 - $top_results = [];
7842 - $offset = 0;
7843 -
7844 - do {
7845 - $batch = $wpdb->get_results($wpdb->prepare(
7846 - "SELECT id, embedding_vector
7847 - FROM {$system_prompt_table}
7848 - LIMIT %d OFFSET %d",
7849 - $batch_size,
7850 - $offset
7851 - ));
7852 -
7853 - if (empty($batch)) {
7854 - break;
7855 - }
7856 -
7857 - foreach ($batch as $row) {
7858 - $database_embedding = $row->embedding_vector
7859 - ? unserialize($row->embedding_vector, ['allowed_classes' => false])
7860 - : null;
7861 -
7862 - if (!is_array($database_embedding)) {
7863 - unset($database_embedding);
7864 - continue;
7865 - }
7866 -
7867 - $similarity = $this->mxchat_calculate_cosine_similarity($user_embedding, $database_embedding);
7868 - unset($database_embedding);
7869 -
7870 - if ($similarity < $similarity_threshold) {
7871 - continue;
7872 - }
7873 -
7874 - // Insert into bounded top-K (kept sorted descending)
7875 - if (count($top_results) < $top_k) {
7876 - $top_results[] = ['id' => $row->id, 'similarity' => $similarity];
7877 - usort($top_results, function ($a, $b) {
7878 - return $b['similarity'] <=> $a['similarity'];
7879 - });
7880 - } elseif ($similarity > $top_results[$top_k - 1]['similarity']) {
7881 - $top_results[$top_k - 1] = ['id' => $row->id, 'similarity' => $similarity];
7882 - usort($top_results, function ($a, $b) {
7883 - return $b['similarity'] <=> $a['similarity'];
7884 - });
7885 - }
7886 - }
7887 -
7888 - unset($batch);
7889 - $offset += $batch_size;
7890 - } while (true);
7891 -
7892 - if (empty($top_results)) {
7893 - return '';
7894 - }
7895 -
7896 - $content = '';
7897 - foreach ($top_results as $result) {
7898 - $chunk_content = $this->fetch_content_with_product_links($result['id']);
7899 - $content .= $chunk_content . "\n\n";
7900 - }
7901 -
7902 - return trim($content);
7903 -}
7904 -
7905 -
7906 -private function find_relevant_products_pinecone($user_embedding) {
7907 - //error_log('Starting Pinecone product search...');
7908 -
7909 - $options = get_option('mxchat_pinecone_addon_options', array());
7910 - $api_key = $options['mxchat_pinecone_api_key'] ?? '';
7911 - $host = $options['mxchat_pinecone_host'] ?? '';
7912 -
7913 - if (empty($host) || empty($api_key)) {
7914 - //error_log('Pinecone credentials not properly configured for product search');
7915 - return '';
7916 - }
7917 -
7918 - $similarity_threshold = 0.85;
7919 - $api_endpoint = "https://{$host}/query";
7920 -
7921 - $request_body = array(
7922 - 'vector' => $user_embedding,
7923 - 'topK' => 5,
7924 - 'includeMetadata' => true,
7925 - 'includeValues' => true,
7926 - 'filter' => array(
7927 - 'type' => 'product'
7928 - )
7929 - );
7930 -
7931 - //error_log('Sending request to Pinecone with body: ' . wp_json_encode($request_body));
7932 -
7933 - $response = wp_remote_post($api_endpoint, array(
7934 - 'headers' => array(
7935 - 'Api-Key' => $api_key,
7936 - 'accept' => 'application/json',
7937 - 'content-type' => 'application/json'
7938 - ),
7939 - 'body' => wp_json_encode($request_body),
7940 - 'timeout' => 30
7941 - ));
7942 -
7943 - if (is_wp_error($response)) {
7944 - //error_log('Pinecone product query error: ' . $response->get_error_message());
7945 - return '';
7946 - }
7947 -
7948 - $response_code = wp_remote_retrieve_response_code($response);
7949 - //error_log('Pinecone response code: ' . $response_code);
7950 -
7951 - if ($response_code !== 200) {
7952 - //error_log('Pinecone API error during product search: ' . wp_remote_retrieve_body($response));
7953 - return '';
7954 - }
7955 -
7956 - $results = json_decode(wp_remote_retrieve_body($response), true);
7957 - //error_log('Pinecone raw response: ' . wp_remote_retrieve_body($response));
7958 -
7959 - if (empty($results['matches'])) {
7960 - //error_log('No matches found in Pinecone response');
7961 - return '';
7962 - }
7963 -
7964 - $content = '';
7965 - foreach ($results['matches'] as $match) {
7966 - if ($match['score'] < $similarity_threshold) {
7967 - //error_log("Match below threshold: " . $match['score']);
7968 - continue;
7969 - }
7970 -
7971 - if (!empty($match['metadata']['text'])) {
7972 - $content .= $match['metadata']['text'];
7973 - if (!empty($match['metadata']['source_url'])) {
7974 - $content .= "\n\nFor more details, check out this product: " . esc_url($match['metadata']['source_url']);
7975 - }
7976 - $content .= "\n\n";
7977 - }
7978 - }
7979 -
7980 - return trim($content);
7981 -}
7982 -
7983 -
7984 344 private function fetch_content_with_product_links($most_relevant_id) {
7985 345 global $wpdb;
7986 346 $system_prompt_table = $wpdb->prefix . 'mxchat_system_prompt_content';
7987 347
@@ -8000,3492 +360,41 @@
8000 360
8001 361 return null;
8002 362 }
8003 363
8004 -/**
8005 - * Get system instructions for a specific bot or default
8006 - * Checks for multi-bot add-on and uses bot-specific instructions if available
8007 - * Automatically strips URLs if citation links are disabled
8008 - * Replaces {visitor_name} placeholder with actual visitor name if available
8009 - *
8010 - * @param string $bot_id The bot ID to get instructions for
8011 - * @param string $session_id Optional session ID to lookup visitor name
8012 - */
8013 -private function get_system_instructions($bot_id = 'default', $session_id = '') {
8014 - $instructions = '';
8015 364
8016 - // Check if multi-bot add-on is active
8017 - if (class_exists('MxChat_Multi_Bot_Core_Manager') && $bot_id !== 'default') {
8018 - // Get bot-specific options from multi-bot add-on
8019 - $bot_options = apply_filters('mxchat_get_bot_options', array(), $bot_id);
8020 -
8021 - // If bot has custom system instructions, use those
8022 - if (!empty($bot_options['system_prompt_instructions'])) {
8023 - $instructions = $bot_options['system_prompt_instructions'];
8024 - }
365 + private function mxchat_generate_response($relevant_content, $api_key, $conversation_history) {
366 + if (!$relevant_content) {
367 + return "I'm sorry, I couldn't find relevant information on that topic.";
8025 368 }
8026 369
8027 - // Fall back to default system instructions
8028 - if (empty($instructions)) {
8029 - $instructions = isset($this->options['system_prompt_instructions']) ? $this->options['system_prompt_instructions'] : '';
8030 - }
370 + $system_prompt_instructions = isset($this->options['system_prompt_instructions']) ? $this->options['system_prompt_instructions'] : '';
8031 371
8032 - // Check if citation links are disabled - if so, strip URLs from instructions
8033 - $fresh_options = get_option('mxchat_options', []);
8034 - $citation_links_enabled = isset($fresh_options['citation_links_toggle']) ? ($fresh_options['citation_links_toggle'] === 'on') : true;
8035 -
8036 - if (!$citation_links_enabled && !empty($instructions)) {
8037 - $instructions = preg_replace('#\bhttps?://[^\s<>"\']+#i', '', $instructions);
8038 - $instructions = preg_replace('/\s+/', ' ', trim($instructions)); // Clean up extra spaces
8039 - }
8040 -
8041 - // Replace {visitor_name} placeholder with actual visitor name if available
8042 - if (!empty($instructions) && !empty($session_id) && stripos($instructions, '{visitor_name}') !== false) {
8043 - $name_option_key = "mxchat_name_{$session_id}";
8044 - $visitor_name = get_option($name_option_key, '');
8045 -
8046 - if (!empty($visitor_name)) {
8047 - $instructions = str_ireplace('{visitor_name}', sanitize_text_field($visitor_name), $instructions);
8048 - } else {
8049 - // Remove placeholder if no name is available
8050 - $instructions = str_ireplace('{visitor_name}', '', $instructions);
8051 - $instructions = preg_replace('/\s{2,}/', ' ', trim($instructions)); // Clean up extra spaces
8052 - }
8053 - }
8054 -
8055 - // {context} placeholder (plan 59bc1b): inject the assembled knowledge-base
8056 - // block where the owner placed the token. Runs after the URL-strip and
8057 - // {visitor_name} handling and before the developer filter, so filtered
8058 - // instructions already show the final prompt. Only active once the KB
8059 - // assembly has stashed the block (context_kb_block non-null) — the early
8060 - // URL-extraction call happens before assembly and leaves the token alone.
8061 - if ($this->context_kb_block !== null && !empty($instructions) && stripos($instructions, '{context}') !== false) {
8062 - $pos = stripos($instructions, '{context}');
8063 - $instructions = substr($instructions, 0, $pos)
8064 - . rtrim($this->context_kb_block) . "\n"
8065 - . substr($instructions, $pos + strlen('{context}'));
8066 - // Additional occurrences are stripped — never duplicate the KB block.
8067 - $instructions = str_ireplace('{context}', '', $instructions);
8068 - }
8069 -
8070 - // Allow developers to filter system instructions and process shortcodes
8071 - $instructions = apply_filters('mxchat_system_instructions', $instructions, $bot_id, $session_id);
8072 - $instructions = do_shortcode($instructions);
8073 -
8074 - return $instructions;
8075 -}
8076 -/**
8077 - * Get the current bot ID from session or request context
8078 - */
8079 -private function get_current_bot_id($session_id = '') {
8080 - // First, check if bot_id is passed in the current request
8081 - if (isset($_POST['bot_id']) && !empty($_POST['bot_id'])) {
8082 - return sanitize_key($_POST['bot_id']);
8083 - }
8084 -
8085 - // If not in POST, try to get it from session data
8086 - if (!empty($session_id)) {
8087 - $bot_id = get_option("mxchat_session_bot_{$session_id}", '');
8088 - if (!empty($bot_id)) {
8089 - return $bot_id;
8090 - }
8091 - }
8092 -
8093 - // Fall back to default
8094 - return 'default';
8095 -}
8096 -/* ====================================================================== *
8097 - * Native function-calling loop (plan-mxchat-20260617-a41dee)
8098 - *
8099 - * Model-driven tool use. The model is offered MxChat's enabled callbacks as
8100 - * tools (sourced from MxChat_Tool_Registry, the single source the admin AI
8101 - * Tools checklist also reads). When the model calls a tool, the matching
8102 - * callback runs through its EXISTING permission checks, its output is fed
8103 - * back, and the loop continues up to a depth cap. INDEPENDENT of the
8104 - * intent→callback router — it runs only after intents miss, and works with
8105 - * ZERO Actions created.
8106 - *
8107 - * Entered ONLY when: function calling is enabled + the active model is
8108 - * tool-capable + at least one tool is enabled. Default-off, so existing
8109 - * installs never enter this branch (byte-for-byte unchanged behavior). The
8110 - * tool round is buffered (non-streaming) per the plan; the final answer is
8111 - * emitted via the same SSE/JSON envelopes the normal path uses.
8112 - * ====================================================================== */
8113 -
8114 -/** Gate: should the function-calling loop handle this turn? */
8115 -private function mxchat_fc_should_run($selected_model) {
8116 - if (!class_exists('MxChat_Tool_Registry') || !MxChat_Tool_Registry::is_enabled()) {
8117 - return false;
8118 - }
8119 - if (class_exists('MxChat_Model_Catalog') && !MxChat_Model_Catalog::supports_tools($selected_model)) {
8120 - return false;
8121 - }
8122 - $tools = MxChat_Tool_Registry::enabled_tools();
8123 - return !empty($tools);
8124 -}
8125 -
8126 -private function mxchat_fc_log($msg) {
8127 - if (defined('MXCHAT_DEV_MODE') && MXCHAT_DEV_MODE) {
8128 - error_log('[MxChat FC] ' . $msg);
8129 - }
8130 -}
8131 -
8132 -/**
8133 - * Resolve provider transport details. Returns null when FC can't run for this
8134 - * model/config (missing key, unsupported provider) so the caller falls back to
8135 - * the normal path. OpenAI/xAI/DeepSeek/OpenRouter/Custom share the
8136 - * OpenAI-compatible 'openai' family; Claude and Gemini are distinct.
8137 - */
8138 -private function mxchat_fc_resolve_provider($selected_model, $opts) {
8139 - // Anthropic retired claude-opus-4-20250514 / claude-sonnet-4-20250514 on 2026-06-15.
8140 - // Read-time rescue: remap a saved dead ID to the current equivalent before the API call.
8141 - if ($selected_model === 'claude-opus-4-20250514') { $selected_model = 'claude-opus-4-8'; }
8142 - elseif ($selected_model === 'claude-sonnet-4-20250514') { $selected_model = 'claude-sonnet-4-6'; }
8143 - if ($selected_model === 'openrouter') {
8144 - $model = isset($opts['openrouter_selected_model']) ? $opts['openrouter_selected_model'] : '';
8145 - $key = isset($opts['openrouter_api_key']) ? $opts['openrouter_api_key'] : '';
8146 - if ($model === '' || $key === '') return null;
8147 - return array('family'=>'openai','model'=>$model,'url'=>'https://openrouter.ai/api/v1/chat/completions',
8148 - 'headers'=>array('Content-Type'=>'application/json','Authorization'=>'Bearer '.$key),'tag'=>'openai');
8149 - }
8150 - $prefix = strtolower(explode('-', $selected_model)[0]);
8151 - switch ($prefix) {
8152 - case 'gpt': case 'o1': case 'o3': case 'o4':
8153 - $key = isset($opts['api_key']) ? $opts['api_key'] : '';
8154 - if ($key === '') return null;
8155 - return array('family'=>'openai','model'=>$selected_model,'url'=>'https://api.openai.com/v1/chat/completions',
8156 - 'headers'=>array('Content-Type'=>'application/json','Authorization'=>'Bearer '.$key),'tag'=>'openai');
8157 - case 'claude':
8158 - $key = isset($opts['claude_api_key']) ? $opts['claude_api_key'] : '';
8159 - if ($key === '') return null;
8160 - return array('family'=>'anthropic','model'=>$selected_model,'url'=>'https://api.anthropic.com/v1/messages',
8161 - 'headers'=>array('Content-Type'=>'application/json','x-api-key'=>$key,'anthropic-version'=>'2023-06-01'),'tag'=>'anthropic');
8162 - case 'gemini':
8163 - $key = isset($opts['gemini_api_key']) ? $opts['gemini_api_key'] : '';
8164 - if ($key === '') return null;
8165 - return array('family'=>'gemini','model'=>$selected_model,'key'=>$key,'tag'=>'gemini');
8166 - case 'grok': case 'xai':
8167 - $key = isset($opts['xai_api_key']) ? $opts['xai_api_key'] : '';
8168 - if ($key === '') return null;
8169 - return array('family'=>'openai','model'=>$selected_model,'url'=>'https://api.x.ai/v1/chat/completions',
8170 - 'headers'=>array('Content-Type'=>'application/json','Authorization'=>'Bearer '.$key),'tag'=>'xai');
8171 - case 'deepseek':
8172 - $key = isset($opts['deepseek_api_key']) ? $opts['deepseek_api_key'] : '';
8173 - if ($key === '') return null;
8174 - return array('family'=>'openai','model'=>$selected_model,'url'=>'https://api.deepseek.com/v1/chat/completions',
8175 - 'headers'=>array('Content-Type'=>'application/json','Authorization'=>'Bearer '.$key),'tag'=>'openai');
8176 - case 'custom':
8177 - $base = isset($opts['custom_provider_base_url']) ? rtrim($opts['custom_provider_base_url'], '/') : '';
8178 - $key = isset($opts['custom_provider_api_key']) ? $opts['custom_provider_api_key'] : '';
8179 - $model = isset($opts['custom_provider_model']) ? $opts['custom_provider_model'] : '';
8180 - if ($base === '' || $model === '') return null;
8181 - $url = (strpos($base, 'chat/completions') !== false) ? $base : $base . '/chat/completions';
8182 - $headers = array('Content-Type'=>'application/json');
8183 - if ($key !== '') $headers['Authorization'] = 'Bearer '.$key;
8184 - return array('family'=>'openai','model'=>$model,'url'=>$url,'headers'=>$headers,'tag'=>'openai');
8185 - }
8186 - return null;
8187 -}
8188 -
8189 -/**
8190 - * Top-level function-calling attempt. Returns:
8191 - * ['handled'=>true, 'text'=>'<final answer>'] when the model used ≥1 tool
8192 - * ['handled'=>false] otherwise (caller falls back
8193 - * to the normal streamed path)
8194 - */
8195 -private function mxchat_fc_attempt($message, $relevant_content, $conversation_history, $selected_model, $opts, $session_id, $user_id) {
8196 - $prov = $this->mxchat_fc_resolve_provider($selected_model, $opts);
8197 - if (!$prov) {
8198 - return array('handled' => false);
8199 - }
8200 - $tools = MxChat_Tool_Registry::enabled_tools();
8201 - if (empty($tools)) {
8202 - return array('handled' => false);
8203 - }
8204 -
8205 - $bot_id = $this->get_current_bot_id($session_id);
8206 - $system = $this->get_system_instructions($bot_id, $session_id);
8207 -
8208 - // Force callbacks into return-mode (some echo SSE directly when streaming);
8209 - // we buffer the whole tool round, then emit once. Restored in finally.
8210 - $prev_streaming = $this->is_streaming;
8211 - $this->is_streaming = false;
8212 - try {
8213 - if ($prov['family'] === 'anthropic') {
8214 - return $this->mxchat_fc_loop_anthropic($prov, $system, $relevant_content, $conversation_history, $tools, $message, $user_id, $session_id);
8215 - } elseif ($prov['family'] === 'gemini') {
8216 - return $this->mxchat_fc_loop_gemini($prov, $system, $relevant_content, $conversation_history, $tools, $message, $user_id, $session_id);
8217 - }
8218 - return $this->mxchat_fc_loop_openai($prov, $system, $relevant_content, $conversation_history, $tools, $message, $user_id, $session_id);
8219 - } catch (\Throwable $e) {
8220 - $this->mxchat_fc_log('attempt threw: ' . $e->getMessage());
8221 - return array('handled' => false);
8222 - } finally {
8223 - $this->is_streaming = $prev_streaming;
8224 - }
8225 -}
8226 -
8227 -/** Normalize MxChat history rows to [{role:user|assistant, content}]. */
8228 -private function mxchat_fc_normalize_history($conversation_history) {
8229 - $out = array();
8230 - if (!is_array($conversation_history)) return $out;
8231 - foreach ($conversation_history as $m) {
8232 - if (!is_array($m) || !isset($m['role']) || !isset($m['content'])) continue;
8233 - $role = $m['role'];
8234 - if ($role === 'bot' || $role === 'agent') $role = 'assistant';
8235 - if (!in_array($role, array('user', 'assistant'), true)) $role = 'user';
8236 - $out[] = array('role' => $role, 'content' => (string) $m['content']);
8237 - }
8238 - return $out;
8239 -}
8240 -
8241 -/** Execute the matched callback for a tool call. Returns ['ok'=>bool,'content'=>string]. */
8242 -private function mxchat_fc_execute_tool($tool_name, $args, $orig_message, $user_id, $session_id) {
8243 - $tool = MxChat_Tool_Registry::tool_by_name($tool_name, true); // enabled-only
8244 - if (!$tool) {
8245 - return array('ok' => false, 'content' => 'This tool is not available or not enabled.');
8246 - }
8247 - $fn = $tool['callback'];
8248 -
8249 - // MxChat callbacks are message-driven: hand them the model's `query`
8250 - // (falling back to the original user message).
8251 - $query = '';
8252 - if (is_array($args) && isset($args['query']) && is_string($args['query'])) {
8253 - $query = $args['query'];
8254 - }
8255 - if ($query === '') $query = $orig_message;
8256 -
8257 - // Synthetic intent row (matches wp_mxchat_intents columns → no undefined-prop warnings).
8258 - $synthetic_intent = (object) array(
8259 - 'id' => 0, 'intent_label' => $tool['label'], 'phrases' => '',
8260 - 'embedding_vector' => '', 'callback_function' => $fn,
8261 - 'similarity_threshold' => 0.0, 'enabled' => 1, 'enabled_bots' => null,
8262 - );
8263 -
8264 - try {
8265 - if (!empty($tool['is_addon'])) {
8266 - $result = apply_filters($fn, false, $query, $user_id, $session_id, $synthetic_intent);
8267 - } elseif (method_exists($this, $fn)) {
8268 - $result = call_user_func(array($this, $fn), $query, $user_id, $session_id, $synthetic_intent, null);
8269 - } else {
8270 - return array('ok' => false, 'content' => 'Tool implementation not found.');
8271 - }
8272 - } catch (\Throwable $e) {
8273 - $this->mxchat_fc_log("tool {$fn} threw: " . $e->getMessage());
8274 - return array('ok' => false, 'content' => 'The tool failed to run.');
8275 - }
8276 -
8277 - // plan-mxchat-20260617-48a57a — surface UI-bearing tool output.
8278 - // If the callback produced a UI element (generated image, product card, image
8279 - // gallery), its html MUST reach the FRONTEND as a real rendered bot message —
8280 - // NOT be stripped to text and handed to the model to paraphrase (that was the
8281 - // bug: under function calling, UI-bearing actions rendered nothing). Capture
8282 - // the html here; the FC outcome handler emits it in the response envelope.
8283 - $ui = $this->mxchat_fc_ui_payload_from($result);
8284 - if ($ui['html'] !== '' || !empty($ui['images'])) {
8285 - if ($ui['html'] !== '') {
8286 - $this->fc_ui_html .= ($this->fc_ui_html !== '' ? "\n" : '') . $ui['html'];
8287 - }
8288 - if (!empty($ui['images']) && is_array($ui['images'])) {
8289 - $this->fc_ui_images = array_merge($this->fc_ui_images, $ui['images']);
8290 - }
8291 - $this->fc_ui_captured = true;
8292 -
8293 - // Persist the html to the transcript ONLY if the callback did not already
8294 - // do so itself. Core image/search callbacks self-save (text + html);
8295 - // add-on callbacks (e.g. woo product cards) return html for the caller to
8296 - // save. ui_self_saves carries this from the registry; default by source
8297 - // (core self-saves, add-on does not) when a tool predates the flag.
8298 - $self_saves = array_key_exists('ui_self_saves', $tool)
8299 - ? !empty($tool['ui_self_saves'])
8300 - : empty($tool['is_addon']);
8301 - if ($ui['html'] !== '' && !$self_saves) {
8302 - $this->mxchat_save_chat_message($session_id, 'bot', $ui['html']);
8303 - }
8304 -
8305 - // Hand the MODEL a short acknowledgment (never the raw or stripped html)
8306 - // so the loop can add a one-line caption without trying to re-describe a
8307 - // visual it cannot see and without duplicating the displayed element.
8308 - $summary = isset($ui['text']) ? trim((string) $ui['text']) : '';
8309 - $ack = __('[A visual result has already been shown to the user in the chat. Do not repeat or describe it in detail — reply with at most a brief one-line caption.]', 'mxchat');
8310 - $content = $summary !== '' ? ($ack . ' ' . $summary) : $ack;
8311 - $this->mxchat_fc_log("executed {$fn} → [ui payload surfaced] " . substr($content, 0, 120));
8312 - return array('ok' => true, 'content' => $content);
8313 - }
8314 -
8315 - $content = $this->mxchat_fc_stringify_result($result);
8316 - $this->mxchat_fc_log("executed {$fn} → " . substr($content, 0, 160));
8317 - return array('ok' => true, 'content' => $content);
8318 -}
8319 -
8320 -/**
8321 - * Extract a UI payload (html + images + text) from a tool callback's return,
8322 - * falling back to $this->fallbackResponse for callbacks that return true after
8323 - * setting it. plan-mxchat-20260617-48a57a.
8324 - *
8325 - * @return array{html:string,images:array,text:string}
8326 - */
8327 -private function mxchat_fc_ui_payload_from($result) {
8328 - $src = null;
8329 - if (is_array($result)) {
8330 - $src = $result;
8331 - } elseif ($result === true && isset($this->fallbackResponse) && is_array($this->fallbackResponse)) {
8332 - $src = $this->fallbackResponse;
8333 - }
8334 - $html = (is_array($src) && isset($src['html']) && is_string($src['html'])) ? $src['html'] : '';
8335 - $images = (is_array($src) && isset($src['images']) && is_array($src['images'])) ? $src['images'] : array();
8336 - $text = (is_array($src) && isset($src['text'])) ? (string) $src['text'] : '';
8337 - return array('html' => $html, 'images' => $images, 'text' => $text);
8338 -}
8339 -
8340 -/** Coerce a callback's return (string|array|true|false) into a tool-result string. */
8341 -private function mxchat_fc_stringify_result($result) {
8342 - if (is_string($result)) {
8343 - return $result === '' ? 'No result.' : $result;
8344 - }
8345 - if ($result === true) {
8346 - // Callbacks that set fallbackResponse and return true.
8347 - $fb = isset($this->fallbackResponse) ? $this->fallbackResponse : null;
8348 - if (is_array($fb)) {
8349 - if (!empty($fb['text'])) return (string) $fb['text'];
8350 - if (!empty($fb['html'])) return wp_strip_all_tags((string) $fb['html']);
8351 - }
8352 - return 'Done.';
8353 - }
8354 - if ($result === false || $result === null) {
8355 - return 'No result.';
8356 - }
8357 - if (is_array($result)) {
8358 - if (isset($result['text']) && $result['text'] !== '') return (string) $result['text'];
8359 - if (isset($result['html']) && $result['html'] !== '') return wp_strip_all_tags((string) $result['html']);
8360 - $json = wp_json_encode($result);
8361 - return $json !== false ? $json : 'No result.';
8362 - }
8363 - return (string) $result;
8364 -}
8365 -
8366 -/** HTTP code + decoded body for a function-calling request. */
8367 -private function mxchat_fc_post($url, $body, $headers, $tag) {
8368 - $args = array(
8369 - 'body' => wp_json_encode($body),
8370 - 'headers' => $headers,
8371 - 'timeout' => 60,
8372 - 'redirection' => 5,
8373 - 'blocking' => true,
8374 - 'httpversion' => '1.0',
8375 - 'sslverify' => true,
8376 - );
8377 - $response = $this->mxchat_provider_call_with_retry($url, $args, $tag);
8378 - if (is_wp_error($response)) {
8379 - return array('code' => 0, 'data' => null, 'error' => $response->get_error_message());
8380 - }
8381 - $code = (int) wp_remote_retrieve_response_code($response);
8382 - $data = json_decode(wp_remote_retrieve_body($response), true);
8383 - return array('code' => $code, 'data' => $data, 'error' => null);
8384 -}
8385 -
8386 -/* ---------------- OpenAI-compatible loop (OpenAI/xAI/DeepSeek/OpenRouter/Custom) -------------- */
8387 -private function mxchat_fc_loop_openai($prov, $system, $relevant_content, $conversation_history, $tools, $orig_message, $user_id, $session_id) {
8388 - $messages = array();
8389 - $messages[] = array('role' => 'system', 'content' => $system . ' ' . $relevant_content);
8390 - foreach ($this->mxchat_fc_normalize_history($conversation_history) as $m) {
8391 - $messages[] = $m;
8392 - }
8393 -
8394 - $depth = MxChat_Tool_Registry::max_depth();
8395 - $budget = MxChat_Tool_Registry::max_tool_calls_per_turn();
8396 - $tool_schema = MxChat_Tool_Registry::to_openai_tools($tools);
8397 - $used_tool = false;
8398 - $calls_made = 0;
8399 -
8400 - for ($step = 0; $step <= $depth; $step++) {
8401 - $offer_tools = ($step < $depth) && !empty($tool_schema);
8402 - $body = array('model' => $prov['model'], 'messages' => $messages, 'temperature' => 1, 'stream' => false);
8403 - if (strpos($prov['url'], 'api.deepseek.com') !== false) {
8404 - // DeepSeek V4 defaults to thinking mode ON; tool loops want fast
8405 - // deterministic non-thinking turns (legacy deepseek-chat semantics).
8406 - $body['thinking'] = array('type' => 'disabled');
8407 - }
8408 - if ($offer_tools) {
8409 - $body['tools'] = $tool_schema;
8410 - $body['tool_choice'] = 'auto';
8411 - }
8412 - $r = $this->mxchat_fc_post($prov['url'], $body, $prov['headers'], $prov['tag']);
8413 - if ($r['code'] !== 200 || !is_array($r['data'])) {
8414 - $this->mxchat_fc_log('openai call failed: code=' . $r['code'] . ' err=' . ($r['error'] ?? ''));
8415 - return $used_tool ? array('handled' => true, 'text' => $this->mxchat_fc_giveup_text()) : array('handled' => false);
8416 - }
8417 - $msg = isset($r['data']['choices'][0]['message']) ? $r['data']['choices'][0]['message'] : null;
8418 - if (!$msg) {
8419 - return $used_tool ? array('handled' => true, 'text' => $this->mxchat_fc_giveup_text()) : array('handled' => false);
8420 - }
8421 - $tool_calls = isset($msg['tool_calls']) && is_array($msg['tool_calls']) ? $msg['tool_calls'] : array();
8422 - if (empty($tool_calls)) {
8423 - $text = isset($msg['content']) ? trim((string) $msg['content']) : '';
8424 - if (!$used_tool) return array('handled' => false); // model never used a tool → normal path
8425 - return array('handled' => true, 'text' => ($text !== '' ? $text : $this->mxchat_fc_giveup_text()));
8426 - }
8427 - // Append the assistant tool-call turn verbatim, then a tool result per call.
8428 - $used_tool = true;
8429 - $messages[] = $msg;
8430 - foreach ($tool_calls as $tc) {
8431 - if ($calls_made >= $budget) break;
8432 - $calls_made++;
8433 - $name = isset($tc['function']['name']) ? $tc['function']['name'] : '';
8434 - $args = array();
8435 - if (isset($tc['function']['arguments'])) {
8436 - $decoded = json_decode($tc['function']['arguments'], true);
8437 - if (is_array($decoded)) $args = $decoded;
8438 - }
8439 - $exec = $this->mxchat_fc_execute_tool($name, $args, $orig_message, $user_id, $session_id);
8440 - $messages[] = array(
8441 - 'role' => 'tool',
8442 - 'tool_call_id' => isset($tc['id']) ? $tc['id'] : '',
8443 - 'content' => $exec['content'],
8444 - );
8445 - }
8446 - }
8447 - return $used_tool ? array('handled' => true, 'text' => $this->mxchat_fc_giveup_text()) : array('handled' => false);
8448 -}
8449 -
8450 -/* ---------------- Anthropic Claude loop ---------------- */
8451 -private function mxchat_fc_loop_anthropic($prov, $system, $relevant_content, $conversation_history, $tools, $orig_message, $user_id, $session_id) {
8452 - $messages = $this->mxchat_fc_normalize_history($conversation_history);
8453 - $messages[] = array('role' => 'user', 'content' => $relevant_content);
8454 -
8455 - $depth = MxChat_Tool_Registry::max_depth();
8456 - $budget = MxChat_Tool_Registry::max_tool_calls_per_turn();
8457 - $tool_schema = MxChat_Tool_Registry::to_anthropic_tools($tools);
8458 - $omit_temp = $this->mxchat_claude_omits_temperature($prov['model']);
8459 - $used_tool = false;
8460 - $calls_made = 0;
8461 -
8462 - for ($step = 0; $step <= $depth; $step++) {
8463 - $offer_tools = ($step < $depth) && !empty($tool_schema);
8464 - $body = array('model' => $prov['model'], 'max_tokens' => 1024, 'temperature' => 0.8,
8465 - 'messages' => $messages, 'system' => $system);
8466 - if ($omit_temp) unset($body['temperature']);
8467 - if ($offer_tools) {
8468 - $body['tools'] = $tool_schema;
8469 - $body['tool_choice'] = array('type' => 'auto');
8470 - }
8471 - $r = $this->mxchat_fc_post($prov['url'], $body, $prov['headers'], $prov['tag']);
8472 - if ($r['code'] !== 200 || !is_array($r['data'])) {
8473 - $this->mxchat_fc_log('anthropic call failed: code=' . $r['code'] . ' err=' . ($r['error'] ?? ''));
8474 - return $used_tool ? array('handled' => true, 'text' => $this->mxchat_fc_giveup_text()) : array('handled' => false);
8475 - }
8476 - $content = isset($r['data']['content']) && is_array($r['data']['content']) ? $r['data']['content'] : array();
8477 - $tool_uses = array();
8478 - $text_out = '';
8479 - foreach ($content as $block) {
8480 - if (!isset($block['type'])) continue;
8481 - if ($block['type'] === 'tool_use') {
8482 - $tool_uses[] = $block;
8483 - } elseif ($block['type'] === 'text' && isset($block['text'])) {
8484 - $text_out .= $block['text'];
8485 - }
8486 - }
8487 - if (empty($tool_uses)) {
8488 - if (!$used_tool) return array('handled' => false);
8489 - $text_out = trim($text_out);
8490 - return array('handled' => true, 'text' => ($text_out !== '' ? $text_out : $this->mxchat_fc_giveup_text()));
8491 - }
8492 - // Append the assistant turn (the full content array), then a user turn of tool_result blocks.
8493 - $used_tool = true;
8494 - $messages[] = array('role' => 'assistant', 'content' => $content);
8495 - $results = array();
8496 - foreach ($tool_uses as $tu) {
8497 - if ($calls_made >= $budget) break;
8498 - $calls_made++;
8499 - $name = isset($tu['name']) ? $tu['name'] : '';
8500 - $args = isset($tu['input']) && is_array($tu['input']) ? $tu['input'] : array();
8501 - $exec = $this->mxchat_fc_execute_tool($name, $args, $orig_message, $user_id, $session_id);
8502 - $results[] = array(
8503 - 'type' => 'tool_result',
8504 - 'tool_use_id' => isset($tu['id']) ? $tu['id'] : '',
8505 - 'content' => $exec['content'],
8506 - );
8507 - }
8508 - $messages[] = array('role' => 'user', 'content' => $results);
8509 - }
8510 - return $used_tool ? array('handled' => true, 'text' => $this->mxchat_fc_giveup_text()) : array('handled' => false);
8511 -}
8512 -
8513 -/* ---------------- Google Gemini loop ---------------- */
8514 -private function mxchat_fc_loop_gemini($prov, $system, $relevant_content, $conversation_history, $tools, $orig_message, $user_id, $session_id) {
8515 - $contents = array();
8516 - $contents[] = array('role' => 'user', 'parts' => array(array('text' => '[System Instructions] ' . $system . ' ' . $relevant_content)));
8517 - $contents[] = array('role' => 'model', 'parts' => array(array('text' => 'I understand and will follow these instructions.')));
8518 - foreach ($this->mxchat_fc_normalize_history($conversation_history) as $m) {
8519 - $contents[] = array('role' => ($m['role'] === 'assistant' ? 'model' : 'user'),
8520 - 'parts' => array(array('text' => $m['content'])));
8521 - }
8522 -
8523 - $depth = MxChat_Tool_Registry::max_depth();
8524 - $budget = MxChat_Tool_Registry::max_tool_calls_per_turn();
8525 - $tool_schema = MxChat_Tool_Registry::to_gemini_tools($tools);
8526 - // Function calling (tools + functionDeclarations + toolConfig) is a v1beta feature on the
8527 - // Generative Language REST API. The v1 endpoint silently ignores the tools array, so a
8528 - // non-preview model (e.g. gemini-2.5-pro, gemini-3.5-flash, gemini-3.1-flash-lite) would
8529 - // just answer in text and never emit a tool call. Always use v1beta for the FC loop —
8530 - // confirmed against Google's function-calling docs (their REST example targets
8531 - // v1beta/models/gemini-3.5-flash:generateContent). v1beta is a superset, so every model
8532 - // reachable on v1 is also reachable here.
8533 - $api_version = 'v1beta';
8534 - $url = 'https://generativelanguage.googleapis.com/' . $api_version . '/models/' . $prov['model'] . ':generateContent?key=' . $prov['key'];
8535 - $headers = array('Content-Type' => 'application/json');
8536 - $used_tool = false;
8537 - $calls_made = 0;
8538 -
8539 - for ($step = 0; $step <= $depth; $step++) {
8540 - $offer_tools = ($step < $depth) && !empty($tool_schema);
8541 - $body = array(
8542 - 'contents' => $contents,
8543 - 'generationConfig' => array('temperature' => 0.7, 'topP' => 0.95, 'topK' => 40, 'maxOutputTokens' => 8192),
8544 - );
8545 - if ($offer_tools) {
8546 - $body['tools'] = $tool_schema;
8547 - $body['toolConfig'] = array('functionCallingConfig' => array('mode' => 'AUTO'));
8548 - }
8549 - $r = $this->mxchat_fc_post($url, $body, $headers, 'gemini');
8550 - if ($r['code'] !== 200 || !is_array($r['data']) || isset($r['data']['error'])) {
8551 - $this->mxchat_fc_log('gemini call failed: code=' . $r['code'] . ' err=' . ($r['error'] ?? ''));
8552 - return $used_tool ? array('handled' => true, 'text' => $this->mxchat_fc_giveup_text()) : array('handled' => false);
8553 - }
8554 - $parts = isset($r['data']['candidates'][0]['content']['parts']) && is_array($r['data']['candidates'][0]['content']['parts'])
8555 - ? $r['data']['candidates'][0]['content']['parts'] : array();
8556 - $fn_calls = array();
8557 - $text_out = '';
8558 - foreach ($parts as $p) {
8559 - if (isset($p['functionCall'])) {
8560 - $fn_calls[] = $p['functionCall'];
8561 - } elseif (isset($p['text'])) {
8562 - $text_out .= $p['text'];
8563 - }
8564 - }
8565 - if (empty($fn_calls)) {
8566 - if (!$used_tool) return array('handled' => false);
8567 - $text_out = trim($text_out);
8568 - return array('handled' => true, 'text' => ($text_out !== '' ? $text_out : $this->mxchat_fc_giveup_text()));
8569 - }
8570 - // Append the model turn (its parts) then a user turn of functionResponse parts.
8571 - $used_tool = true;
8572 - $contents[] = array('role' => 'model', 'parts' => $parts);
8573 - $resp_parts = array();
8574 - foreach ($fn_calls as $fcall) {
8575 - if ($calls_made >= $budget) break;
8576 - $calls_made++;
8577 - $name = isset($fcall['name']) ? $fcall['name'] : '';
8578 - $args = isset($fcall['args']) && is_array($fcall['args']) ? $fcall['args'] : array();
8579 - $exec = $this->mxchat_fc_execute_tool($name, $args, $orig_message, $user_id, $session_id);
8580 - $fr = array('name' => $name, 'response' => array('result' => $exec['content']));
8581 - // Gemini 3 function calls carry a unique id; echo the matching id back in the
8582 - // functionResponse so the model maps the result to the right call (Google REST
8583 - // guidance). Older models omit the id — then we send none, exactly as before.
8584 - if (isset($fcall['id']) && $fcall['id'] !== '') { $fr['id'] = $fcall['id']; }
8585 - $resp_parts[] = array('functionResponse' => $fr);
8586 - }
8587 - $contents[] = array('role' => 'user', 'parts' => $resp_parts);
8588 - }
8589 - return $used_tool ? array('handled' => true, 'text' => $this->mxchat_fc_giveup_text()) : array('handled' => false);
8590 -}
8591 -
8592 -private function mxchat_fc_giveup_text() {
8593 - return esc_html__('I looked into that but could not put together a final answer. Please try rephrasing your request.', 'mxchat');
8594 -}
8595 -
8596 -private function mxchat_generate_response($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, $selected_model = 'gpt-5.6-sol') {
8597 - try {
8598 - if (!$relevant_content) {
8599 - $error_response = [
8600 - 'error' => esc_html__("I couldn't find relevant information on that topic.", 'mxchat'),
8601 - 'error_code' => 'no_relevant_content'
8602 - ];
8603 -
8604 - if ($testing_data !== null) {
8605 - $error_response['testing_data'] = $testing_data;
8606 - }
8607 -
8608 - return $error_response;
8609 - }
8610 -
8611 - if (!is_array($conversation_history)) {
8612 - $conversation_history = array();
8613 - }
8614 -
8615 - // Check if this is an OpenRouter model
8616 - if ($selected_model === 'openrouter') {
8617 - // Get the actual OpenRouter model from options
8618 - $openrouter_selected_model = $this->options['openrouter_selected_model'] ?? '';
8619 -
8620 - if (empty($openrouter_selected_model)) {
8621 - $error_response = [
8622 - 'error' => esc_html__('No OpenRouter model selected. Please select a model in settings.', 'mxchat'),
8623 - 'error_code' => 'no_openrouter_model_selected'
8624 - ];
8625 - if ($testing_data !== null) {
8626 - $error_response['testing_data'] = $testing_data;
8627 - }
8628 - return $error_response;
8629 - }
8630 -
8631 - if (empty($openrouter_api_key)) {
8632 - $error_response = [
8633 - 'error' => esc_html__('OpenRouter API key is not configured', 'mxchat'),
8634 - 'error_code' => 'missing_openrouter_api_key'
8635 - ];
8636 - if ($testing_data !== null) {
8637 - $error_response['testing_data'] = $testing_data;
8638 - }
8639 - return $error_response;
8640 - }
8641 -
8642 - if ($streaming) {
8643 - return $this->mxchat_generate_response_openrouter_stream(
8644 - $openrouter_selected_model,
8645 - $openrouter_api_key,
8646 - $conversation_history,
8647 - $relevant_content,
8648 - $session_id,
8649 - $testing_data
8650 - );
8651 - } else {
8652 - $response = $this->mxchat_generate_response_openrouter(
8653 - $openrouter_selected_model,
8654 - $openrouter_api_key,
8655 - $conversation_history,
8656 - $relevant_content,
8657 - $session_id
8658 - );
8659 - }
8660 -
8661 - if (is_array($response) && isset($response['error'])) {
8662 - if ($testing_data !== null) {
8663 - $response['testing_data'] = $testing_data;
8664 - }
8665 - return $response;
8666 - }
8667 -
8668 - return $response;
8669 - }
8670 -
8671 - // Extract model prefix to determine the provider
8672 - $model_parts = explode('-', $selected_model);
8673 - $provider = strtolower($model_parts[0]);
8674 -
8675 - // Handle model selection based on provider prefix
8676 - switch ($provider) {
8677 - case 'gemini':
8678 - if (empty($gemini_api_key)) {
8679 - $error_response = [
8680 - 'error' => esc_html__('Google Gemini API key is not configured', 'mxchat'),
8681 - 'error_code' => 'missing_gemini_api_key'
8682 - ];
8683 - if ($testing_data !== null) {
8684 - $error_response['testing_data'] = $testing_data;
8685 - }
8686 - return $error_response;
8687 - }
8688 - $response = $this->mxchat_generate_response_gemini(
8689 - $selected_model,
8690 - $gemini_api_key,
8691 - $conversation_history,
8692 - $relevant_content,
8693 - $session_id
8694 - );
8695 - break;
8696 -
8697 - case 'claude':
8698 - if (empty($claude_api_key)) {
8699 - $error_response = [
8700 - 'error' => esc_html__('Claude API key is not configured', 'mxchat'),
8701 - 'error_code' => 'missing_claude_api_key'
8702 - ];
8703 - if ($testing_data !== null) {
8704 - $error_response['testing_data'] = $testing_data;
8705 - }
8706 - return $error_response;
8707 - }
8708 - if ($streaming) {
8709 - return $this->mxchat_generate_response_claude_stream(
8710 - $selected_model,
8711 - $claude_api_key,
8712 - $conversation_history,
8713 - $relevant_content,
8714 - $session_id,
8715 - $testing_data
8716 - );
8717 - } else {
8718 - $response = $this->mxchat_generate_response_claude(
8719 - $selected_model,
8720 - $claude_api_key,
8721 - $conversation_history,
8722 - $relevant_content,
8723 - $session_id
8724 - );
8725 - }
8726 - break;
8727 -
8728 - case 'grok':
8729 - if (empty($xai_api_key)) {
8730 - $error_response = [
8731 - 'error' => esc_html__('X.AI API key is not configured', 'mxchat'),
8732 - 'error_code' => 'missing_xai_api_key'
8733 - ];
8734 - if ($testing_data !== null) {
8735 - $error_response['testing_data'] = $testing_data;
8736 - }
8737 - return $error_response;
8738 - }
8739 - if ($streaming) {
8740 - return $this->mxchat_generate_response_xai_stream(
8741 - $selected_model,
8742 - $xai_api_key,
8743 - $conversation_history,
8744 - $relevant_content,
8745 - $session_id,
8746 - $testing_data
8747 - );
8748 - } else {
8749 - $response = $this->mxchat_generate_response_xai(
8750 - $selected_model,
8751 - $xai_api_key,
8752 - $conversation_history,
8753 - $relevant_content,
8754 - $session_id
8755 - );
8756 - }
8757 - break;
8758 -
8759 - case 'deepseek':
8760 - if (empty($deepseek_api_key)) {
8761 - $error_response = [
8762 - 'error' => esc_html__('DeepSeek API key is not configured', 'mxchat'),
8763 - 'error_code' => 'missing_deepseek_api_key'
8764 - ];
8765 - if ($testing_data !== null) {
8766 - $error_response['testing_data'] = $testing_data;
8767 - }
8768 - return $error_response;
8769 - }
8770 - if ($streaming) {
8771 - return $this->mxchat_generate_response_deepseek_stream(
8772 - $selected_model,
8773 - $deepseek_api_key,
8774 - $conversation_history,
8775 - $relevant_content,
8776 - $session_id,
8777 - $testing_data
8778 - );
8779 - } else {
8780 - $response = $this->mxchat_generate_response_deepseek(
8781 - $selected_model,
8782 - $deepseek_api_key,
8783 - $conversation_history,
8784 - $relevant_content,
8785 - $session_id
8786 - );
8787 - }
8788 - break;
8789 -
8790 - case 'custom':
8791 - // Custom (OpenAI-compatible) provider — Ollama, LM Studio, vLLM, llama.cpp, Azure OpenAI
8792 - $cp_base_url = isset($this->options['custom_provider_base_url']) ? trim((string) $this->options['custom_provider_base_url']) : '';
8793 - if (empty($cp_base_url)) {
8794 - $error_response = [
8795 - 'error' => esc_html__('Custom provider is not configured. Set Base URL in MxChat → API Keys → Custom Provider.', 'mxchat'),
8796 - 'error_code' => 'missing_custom_provider_base_url'
8797 - ];
8798 - if ($testing_data !== null) {
8799 - $error_response['testing_data'] = $testing_data;
8800 - }
8801 - return $error_response;
8802 - }
8803 - if ($streaming) {
8804 - return $this->mxchat_generate_response_custom_stream(
8805 - $selected_model,
8806 - $conversation_history,
8807 - $relevant_content,
8808 - $session_id,
8809 - $testing_data
8810 - );
8811 - } else {
8812 - $response = $this->mxchat_generate_response_custom(
8813 - $selected_model,
8814 - $conversation_history,
8815 - $relevant_content
8816 - );
8817 - }
8818 - break;
8819 -
8820 - case 'gpt':
8821 - case 'o1':
8822 - if (empty($api_key)) {
8823 - $error_response = [
8824 - 'error' => esc_html__('OpenAI API key is not configured', 'mxchat'),
8825 - 'error_code' => 'missing_openai_api_key'
8826 - ];
8827 - if ($testing_data !== null) {
8828 - $error_response['testing_data'] = $testing_data;
8829 - }
8830 - return $error_response;
8831 - }
8832 -
8833 - // Check if web search is enabled for this OpenAI model
8834 - $web_search_enabled = isset($this->options['enable_web_search']) && $this->options['enable_web_search'] === 'on';
8835 - // Models that don't support web search
8836 - $unsupported_web_search_models = array('gpt-4.1-nano');
8837 - $model_supports_web_search = !in_array($selected_model, $unsupported_web_search_models);
8838 -
8839 - if ($web_search_enabled && $model_supports_web_search) {
8840 - // Use Responses API (required for some models, or when web search is enabled)
8841 - return $this->mxchat_generate_response_openai_web_search(
8842 - $selected_model,
8843 - $api_key,
8844 - $conversation_history,
8845 - $relevant_content,
8846 - $session_id,
8847 - $testing_data,
8848 - $streaming
8849 - );
8850 - } elseif ($streaming) {
8851 - return $this->mxchat_generate_response_openai_stream(
8852 - $selected_model,
8853 - $api_key,
8854 - $conversation_history,
8855 - $relevant_content,
8856 - $session_id,
8857 - $testing_data
8858 - );
8859 - } else {
8860 - $response = $this->mxchat_generate_response_openai(
8861 - $selected_model,
8862 - $api_key,
8863 - $conversation_history,
8864 - $relevant_content,
8865 - $session_id
8866 - );
8867 - }
8868 - break;
8869 -
8870 - default:
8871 - if (empty($api_key)) {
8872 - $error_response = [
8873 - 'error' => esc_html__('OpenAI API key is not configured', 'mxchat'),
8874 - 'error_code' => 'missing_openai_api_key'
8875 - ];
8876 - if ($testing_data !== null) {
8877 - $error_response['testing_data'] = $testing_data;
8878 - }
8879 - return $error_response;
8880 - }
8881 -
8882 - // Check if web search is enabled (default case also handles OpenAI models)
8883 - $web_search_enabled = isset($this->options['enable_web_search']) && $this->options['enable_web_search'] === 'on';
8884 - $unsupported_web_search_models = array('gpt-4.1-nano');
8885 - $model_supports_web_search = !in_array($selected_model, $unsupported_web_search_models);
8886 -
8887 - if ($web_search_enabled && $model_supports_web_search) {
8888 - return $this->mxchat_generate_response_openai_web_search(
8889 - $selected_model,
8890 - $api_key,
8891 - $conversation_history,
8892 - $relevant_content,
8893 - $session_id,
8894 - $testing_data,
8895 - $streaming
8896 - );
8897 - } elseif ($streaming) {
8898 - return $this->mxchat_generate_response_openai_stream(
8899 - $selected_model,
8900 - $api_key,
8901 - $conversation_history,
8902 - $relevant_content,
8903 - $session_id,
8904 - $testing_data
8905 - );
8906 - } else {
8907 - $response = $this->mxchat_generate_response_openai(
8908 - $selected_model,
8909 - $api_key,
8910 - $conversation_history,
8911 - $relevant_content,
8912 - $session_id
8913 - );
8914 - }
8915 - break;
8916 - }
8917 -
8918 - if (is_array($response) && isset($response['error'])) {
8919 - if ($testing_data !== null) {
8920 - $response['testing_data'] = $testing_data;
8921 - }
8922 - return $response;
8923 - }
8924 -
8925 - return $response;
8926 -
8927 - } catch (Exception $e) {
8928 - $error_response = [
8929 - 'error' => sprintf(esc_html__('An error occurred: %s', 'mxchat'), esc_html($e->getMessage())),
8930 - 'error_code' => 'system_exception',
8931 - 'exception_details' => $e->getMessage()
8932 - ];
8933 -
8934 - if ($testing_data !== null) {
8935 - $error_response['testing_data'] = $testing_data;
8936 - }
8937 -
8938 - return $error_response;
8939 - }
8940 -}
8941 -private function mxchat_generate_response_openrouter_stream($selected_model, $openrouter_api_key, $conversation_history, $relevant_content, $session_id, $testing_data = null) {
8942 - try {
8943 - $bot_id = $this->get_current_bot_id($session_id);
8944 - $system_prompt_instructions = $this->get_system_instructions($bot_id, $session_id);
8945 -
8946 - if (!is_array($conversation_history)) {
8947 - $conversation_history = array();
8948 - }
8949 -
8950 - $formatted_conversation = array();
8951 -
8952 - $formatted_conversation[] = array(
8953 - 'role' => 'system',
8954 - 'content' => $system_prompt_instructions . " " . $relevant_content
8955 - );
8956 -
8957 - foreach ($conversation_history as $message) {
8958 - if (is_array($message) && isset($message['role']) && isset($message['content'])) {
8959 - $role = $message['role'];
8960 - if ($role === 'bot' || $role === 'agent') {
8961 - $role = 'assistant';
8962 - }
8963 - if (!in_array($role, ['system', 'assistant', 'user'])) {
8964 - $role = 'user';
8965 - }
8966 - $formatted_conversation[] = array(
8967 - 'role' => $role,
8968 - 'content' => $message['content']
8969 - );
8970 - }
8971 - }
8972 -
8973 - if (headers_sent() || !function_exists('curl_init')) {
8974 - $regular_response = $this->mxchat_generate_response_openrouter(
8975 - $selected_model,
8976 - $openrouter_api_key,
8977 - $conversation_history,
8978 - $relevant_content,
8979 - $session_id
8980 - );
8981 -
8982 - // Save bot response to transcript
8983 - if (!empty($regular_response) && !empty($session_id)) {
8984 - $this->mxchat_save_chat_message($session_id, 'bot', $regular_response);
8985 - }
8986 -
8987 - $response_data = [
8988 - 'text' => $regular_response,
8989 - 'html' => '',
8990 - 'session_id' => $session_id
8991 - ];
8992 -
8993 - if ($testing_data !== null) {
8994 - $response_data['testing_data'] = $testing_data;
8995 - }
8996 -
8997 - header('Content-Type: application/json');
8998 - echo json_encode($response_data);
8999 - return true;
9000 - }
9001 -
9002 - $body = json_encode([
9003 - 'model' => $selected_model,
9004 - 'messages' => $formatted_conversation,
9005 - 'temperature' => 1,
9006 - 'stream' => true
9007 - ]);
9008 -
9009 - // V2 retry-on-initial-connect: setup_streaming_headers is now lazy-fired
9010 - // inside WRITEFUNCTION on first byte of a successful upstream.
9011 -
9012 - $captured_status_code = 0;
9013 - $captured_body_pre_stream = '';
9014 - $full_response = '';
9015 - $stream_started = false;
9016 - $buffer = '';
9017 - $errno = 0;
9018 - $last_curl_error = '';
9019 - $http_code = 0;
9020 - $max_attempts = $this->mxchat_retry_enabled() ? 3 : 1;
9021 - $backoff_ms = array(0, 750, 2000);
9022 -
9023 - for ($attempt = 0; $attempt < $max_attempts; $attempt++) {
9024 - if ($attempt > 0 && $backoff_ms[$attempt] > 0) {
9025 - usleep($backoff_ms[$attempt] * 1000);
9026 - }
9027 -
9028 - $captured_status_code = 0;
9029 - $captured_body_pre_stream = '';
9030 - $full_response = '';
9031 - $stream_started = false;
9032 - $buffer = '';
9033 -
9034 - $ch = curl_init();
9035 - curl_setopt($ch, CURLOPT_URL, 'https://openrouter.ai/api/v1/chat/completions');
9036 - curl_setopt($ch, CURLOPT_RETURNTRANSFER, false);
9037 - curl_setopt($ch, CURLOPT_POST, true);
9038 - curl_setopt($ch, CURLOPT_POSTFIELDS, $body);
9039 - curl_setopt($ch, CURLOPT_HTTPHEADER, array(
9040 - 'Content-Type: application/json',
9041 - 'Authorization: Bearer ' . $openrouter_api_key,
9042 - 'HTTP-Referer: ' . home_url(),
9043 - 'X-Title: ' . get_bloginfo('name')
9044 - ));
9045 - curl_setopt($ch, CURLOPT_SSL_VERIFYPEER, true);
9046 - curl_setopt($ch, CURLOPT_TIMEOUT, 60);
9047 -
9048 - curl_setopt($ch, CURLOPT_HEADERFUNCTION, function($ch, $header) use (&$captured_status_code) {
9049 - if ($captured_status_code === 0 && preg_match('#^HTTP/\S+\s+(\d+)\b#', $header, $m)) {
9050 - $captured_status_code = (int) $m[1];
9051 - }
9052 - return strlen($header);
9053 - });
9054 -
9055 - curl_setopt($ch, CURLOPT_WRITEFUNCTION, function($ch, $data) use (&$full_response, &$stream_started, &$buffer, &$captured_status_code, &$captured_body_pre_stream, $testing_data) {
9056 - if ($captured_status_code !== 0 && $captured_status_code !== 200) {
9057 - $captured_body_pre_stream .= $data;
9058 - return strlen($data);
9059 - }
9060 -
9061 - if (!$this->streaming_headers_sent) {
9062 - $this->setup_streaming_headers();
9063 - }
9064 -
9065 - if (!$stream_started && $testing_data !== null) {
9066 - echo "data: " . json_encode(['testing_data' => $testing_data]) . "\n\n";
9067 - flush();
9068 - $stream_started = true;
9069 - }
9070 -
9071 - $buffer .= $data;
9072 - $lines = explode("\n", $buffer);
9073 - $buffer = array_pop($lines);
9074 -
9075 - foreach ($lines as $line) {
9076 - if (trim($line) === '') {
9077 - continue;
9078 - }
9079 - if (strpos($line, 'data: ') !== 0) {
9080 - continue;
9081 - }
9082 -
9083 - $json_str = substr($line, 6);
9084 -
9085 - if (trim($json_str) === '[DONE]') {
9086 - echo "data: [DONE]\n\n";
9087 - flush();
9088 - continue;
9089 - }
9090 -
9091 - $json = json_decode(trim($json_str), true);
9092 - if ($json && isset($json['choices'][0]['delta']['content'])) {
9093 - $content = $json['choices'][0]['delta']['content'];
9094 - $full_response .= $content;
9095 -
9096 - echo "data: " . json_encode(['content' => $content]) . "\n\n";
9097 - flush();
9098 - }
9099 - }
9100 -
9101 - return strlen($data);
9102 - });
9103 -
9104 - $response = curl_exec($ch);
9105 - $errno = curl_errno($ch);
9106 - $last_curl_error = curl_error($ch);
9107 - $http_code = $captured_status_code !== 0 ? $captured_status_code : (int) curl_getinfo($ch, CURLINFO_HTTP_CODE);
9108 - curl_close($ch);
9109 -
9110 - if (!$errno && $http_code === 200) {
9111 - break;
9112 - }
9113 -
9114 - $is_transient = $this->mxchat_is_transient_provider_error_raw($http_code, $captured_body_pre_stream, 'openai', $errno);
9115 - $can_retry = !$this->streaming_headers_sent
9116 - && ($attempt + 1) < $max_attempts
9117 - && $is_transient;
9118 -
9119 - if (defined('WP_DEBUG') && WP_DEBUG) {
9120 - error_log(sprintf(
9121 - '[MxChat] openrouter_stream initial-connect failure (attempt=%d/%d, status=%d, errno=%d, transient=%s, %s).',
9122 - $attempt + 1, $max_attempts, $http_code, $errno,
9123 - $is_transient ? 'yes' : 'no',
9124 - $can_retry ? 'Retrying.' : 'Giving up.'
9125 - ));
9126 - }
9127 -
9128 - if (!$can_retry) {
9129 - break;
9130 - }
9131 - }
9132 -
9133 - if (!$errno && $http_code === 200) {
9134 - if (!empty($full_response) && !empty($session_id)) {
9135 - $rag_context_for_storage = null;
9136 - $has_rag_data = $this->last_similarity_analysis !== null && !empty($this->last_similarity_analysis['top_matches']);
9137 - $has_action_data = isset($this->last_action_analysis) && !empty($this->last_action_analysis);
9138 -
9139 - if ($has_rag_data || $has_action_data) {
9140 - $rag_context_for_storage = [];
9141 -
9142 - if ($has_rag_data) {
9143 - $rag_context_for_storage['top_matches'] = $this->last_similarity_analysis['top_matches'];
9144 - $rag_context_for_storage['approved_urls'] = $this->current_valid_urls ?? [];
9145 - $rag_context_for_storage['similarity_threshold'] = $this->last_similarity_analysis['threshold_used'] ?? 0.35;
9146 - $rag_context_for_storage['knowledge_base_type'] = $this->last_similarity_analysis['knowledge_base_type'] ?? 'WordPress Database';
9147 - $rag_context_for_storage['total_documents_checked'] = $this->last_similarity_analysis['total_checked'] ?? 0;
9148 - }
9149 -
9150 - if ($has_action_data) {
9151 - $rag_context_for_storage['action_analysis'] = $this->last_action_analysis;
9152 - }
9153 - }
9154 - $this->mxchat_save_chat_message($session_id, 'bot', $full_response, null, $rag_context_for_storage);
9155 - }
9156 - return true;
9157 - }
9158 -
9159 - return $this->mxchat_stream_emit_fallback(
9160 - 'openai',
9161 - $this->mxchat_generate_response_openrouter($selected_model, $openrouter_api_key, $conversation_history, $relevant_content, $session_id),
9162 - $session_id,
9163 - $testing_data
9164 - );
9165 -
9166 - } catch (Exception $e) {
9167 - return $this->mxchat_stream_emit_fallback(
9168 - 'openai',
9169 - $this->mxchat_generate_response_openrouter($selected_model, $openrouter_api_key, $conversation_history, $relevant_content, $session_id),
9170 - $session_id,
9171 - $testing_data
9172 - );
9173 - }
9174 -}
9175 -private function mxchat_generate_response_openai_stream($selected_model, $api_key, $conversation_history, $relevant_content, $session_id, $testing_data = null) {
9176 - // OpenAI retires gpt-5.1-chat-latest / gpt-5.3-chat-latest on 2026-08-10
9177 - // (replacement gpt-5.6-sol). Read-time rescue mirrors the non-streaming
9178 - // path (plan e46b8f).
9179 - if ($selected_model === 'gpt-5.1-chat-latest' || $selected_model === 'gpt-5.3-chat-latest') { $selected_model = 'gpt-5.6-sol'; }
9180 - try {
9181 - $bot_id = $this->get_current_bot_id($session_id);
9182 -
9183 - // Get system prompt instructions using centralized function
9184 - $system_prompt_instructions = $this->get_system_instructions($bot_id, $session_id);
9185 -
9186 - // Ensure conversation_history is an array
9187 - if (!is_array($conversation_history)) {
9188 - $conversation_history = array();
9189 - }
9190 -
9191 - // Format conversation history for OpenAI
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 = $message['role'];
9202 - if ($role === 'bot' || $role === 'agent') {
9203 - $role = 'assistant';
9204 - }
9205 - if (!in_array($role, ['system', 'assistant', 'user', 'function', 'tool'])) {
9206 - $role = 'user';
9207 - }
9208 - $formatted_conversation[] = array(
9209 - 'role' => $role,
9210 - 'content' => $message['content']
9211 - );
9212 - }
9213 - }
9214 -
9215 - // Check if we can actually stream
9216 - if (headers_sent() || !function_exists('curl_init')) {
9217 - // Fallback to regular response with testing data
9218 - $regular_response = $this->mxchat_generate_response_openai(
9219 - $selected_model,
9220 - $api_key,
9221 - $conversation_history,
9222 - $relevant_content,
9223 - $session_id
9224 - );
9225 -
9226 - // Save bot response to transcript
9227 - if (!empty($regular_response) && !empty($session_id)) {
9228 - $this->mxchat_save_chat_message($session_id, 'bot', $regular_response);
9229 - }
9230 -
9231 - $response_data = [
9232 - 'text' => $regular_response,
9233 - 'html' => '',
9234 - 'session_id' => $session_id
9235 - ];
9236 -
9237 - if ($testing_data !== null) {
9238 - $response_data['testing_data'] = $testing_data;
9239 - }
9240 -
9241 - header('Content-Type: application/json');
9242 - echo json_encode($response_data);
9243 - return true;
9244 - }
9245 -
9246 - // Build request body with optimal settings for fast streaming
9247 - $request_body = [
9248 - 'model' => $selected_model,
9249 - 'messages' => $formatted_conversation,
9250 - 'temperature' => 1,
9251 - 'stream' => true
9252 - ];
9253 -
9254 - // reasoning_effort — sourced from the core model catalog (plan-dcb71c);
9255 - // frozen inline ladder lives in mxchat_reasoning_effort_fallback().
9256 - $effort = $this->mxchat_reasoning_effort_for($selected_model, 'chat');
9257 - if ($effort !== null) {
9258 - $request_body['reasoning_effort'] = $effort;
9259 - }
9260 -
9261 - $body = json_encode($request_body);
9262 -
9263 - // V2 retry-on-initial-connect: do NOT call setup_streaming_headers() here.
9264 - // It is now lazy-fired inside the WRITEFUNCTION on the first byte of a
9265 - // SUCCESSFUL upstream response, gated by the captured HTTP status.
9266 -
9267 - $captured_status_code = 0;
9268 - $captured_body_pre_stream = '';
9269 - $full_response = '';
9270 - $stream_started = false;
9271 - $buffer = '';
9272 - $errno = 0;
9273 - $last_curl_error = '';
9274 - $http_code = 0;
9275 - $max_attempts = $this->mxchat_retry_enabled() ? 3 : 1;
9276 - $backoff_ms = array(0, 750, 2000);
9277 -
9278 - for ($attempt = 0; $attempt < $max_attempts; $attempt++) {
9279 - if ($attempt > 0 && $backoff_ms[$attempt] > 0) {
9280 - usleep($backoff_ms[$attempt] * 1000);
9281 - }
9282 -
9283 - // Reset per-attempt capture state.
9284 - $captured_status_code = 0;
9285 - $captured_body_pre_stream = '';
9286 - $full_response = '';
9287 - $stream_started = false;
9288 - $buffer = '';
9289 -
9290 - $ch = curl_init();
9291 - curl_setopt($ch, CURLOPT_URL, 'https://api.openai.com/v1/chat/completions');
9292 - curl_setopt($ch, CURLOPT_RETURNTRANSFER, false);
9293 - curl_setopt($ch, CURLOPT_POST, true);
9294 - curl_setopt($ch, CURLOPT_POSTFIELDS, $body);
9295 - curl_setopt($ch, CURLOPT_HTTPHEADER, array(
9296 - 'Content-Type: application/json',
9297 - 'Authorization: Bearer ' . $api_key
9298 - ));
9299 - curl_setopt($ch, CURLOPT_SSL_VERIFYPEER, true);
9300 - curl_setopt($ch, CURLOPT_TIMEOUT, 60);
9301 -
9302 - // Capture HTTP status as soon as response headers arrive — fires before WRITEFUNCTION.
9303 - curl_setopt($ch, CURLOPT_HEADERFUNCTION, function($ch, $header) use (&$captured_status_code) {
9304 - if ($captured_status_code === 0 && preg_match('#^HTTP/\S+\s+(\d+)\b#', $header, $m)) {
9305 - $captured_status_code = (int) $m[1];
9306 - }
9307 - return strlen($header);
9308 - });
9309 -
9310 - // Buffer control for real-time streaming
9311 - curl_setopt($ch, CURLOPT_WRITEFUNCTION, function($ch, $data) use (&$full_response, &$stream_started, &$buffer, &$captured_status_code, &$captured_body_pre_stream, $testing_data) {
9312 - // V2 guard: if upstream returned non-200, buffer body for transient
9313 - // classification and DO NOT emit to client. Stream channel must NOT open.
9314 - if ($captured_status_code !== 0 && $captured_status_code !== 200) {
9315 - $captured_body_pre_stream .= $data;
9316 - return strlen($data);
9317 - }
9318 -
9319 - // Lazy-fire streaming headers on first byte of a SUCCESSFUL upstream.
9320 - // After this point streaming_headers_sent === true → retry is structurally blocked.
9321 - if (!$this->streaming_headers_sent) {
9322 - $this->setup_streaming_headers();
9323 - }
9324 -
9325 - // Send testing data as the first event if available
9326 - if (!$stream_started && $testing_data !== null) {
9327 - echo "data: " . json_encode(['testing_data' => $testing_data]) . "\n\n";
9328 - flush();
9329 - $stream_started = true;
9330 - }
9331 -
9332 - // CRITICAL FIX: Append new data to buffer
9333 - $buffer .= $data;
9334 -
9335 - // Process complete lines only
9336 - $lines = explode("\n", $buffer);
9337 -
9338 - // CRITICAL FIX: Keep the last incomplete line in the buffer
9339 - $buffer = array_pop($lines);
9340 -
9341 - foreach ($lines as $line) {
9342 - if (trim($line) === '') {
9343 - continue;
9344 - }
9345 - if (strpos($line, 'data: ') !== 0) {
9346 - continue;
9347 - }
9348 -
9349 - $json_str = substr($line, 6);
9350 -
9351 - if (trim($json_str) === '[DONE]') {
9352 - echo "data: [DONE]\n\n";
9353 - flush();
9354 - continue;
9355 - }
9356 -
9357 - $json = json_decode(trim($json_str), true);
9358 - if ($json && isset($json['choices'][0]['delta']['content'])) {
9359 - $content = $json['choices'][0]['delta']['content'];
9360 - $full_response .= $content;
9361 -
9362 - echo "data: " . json_encode(['content' => $content]) . "\n\n";
9363 - flush();
9364 - }
9365 - }
9366 -
9367 - return strlen($data);
9368 - });
9369 -
9370 - $response = curl_exec($ch);
9371 - $errno = curl_errno($ch);
9372 - $last_curl_error = curl_error($ch);
9373 - $http_code = $captured_status_code !== 0 ? $captured_status_code : (int) curl_getinfo($ch, CURLINFO_HTTP_CODE);
9374 - curl_close($ch);
9375 -
9376 - if (!$errno && $http_code === 200) {
9377 - break; // Happy path — WRITEFUNCTION already streamed everything.
9378 - }
9379 -
9380 - $is_transient = $this->mxchat_is_transient_provider_error_raw($http_code, $captured_body_pre_stream, 'openai', $errno);
9381 - $can_retry = !$this->streaming_headers_sent
9382 - && ($attempt + 1) < $max_attempts
9383 - && $is_transient;
9384 -
9385 - if (defined('WP_DEBUG') && WP_DEBUG) {
9386 - error_log(sprintf(
9387 - '[MxChat] openai_stream initial-connect failure (attempt=%d/%d, status=%d, errno=%d, transient=%s, %s).',
9388 - $attempt + 1, $max_attempts, $http_code, $errno,
9389 - $is_transient ? 'yes' : 'no',
9390 - $can_retry ? 'Retrying.' : 'Giving up.'
9391 - ));
9392 - }
9393 -
9394 - if (!$can_retry) {
9395 - break;
9396 - }
9397 - }
9398 -
9399 - // Post-loop branch.
9400 - if (!$errno && $http_code === 200) {
9401 - // Happy path — save the complete response to maintain chat persistence.
9402 - if (!empty($full_response) && !empty($session_id)) {
9403 - $rag_context_for_storage = null;
9404 - $has_rag_data = $this->last_similarity_analysis !== null && !empty($this->last_similarity_analysis['top_matches']);
9405 - $has_action_data = isset($this->last_action_analysis) && !empty($this->last_action_analysis);
9406 -
9407 - if ($has_rag_data || $has_action_data) {
9408 - $rag_context_for_storage = [];
9409 -
9410 - if ($has_rag_data) {
9411 - $rag_context_for_storage['top_matches'] = $this->last_similarity_analysis['top_matches'];
9412 - $rag_context_for_storage['approved_urls'] = $this->current_valid_urls ?? [];
9413 - $rag_context_for_storage['similarity_threshold'] = $this->last_similarity_analysis['threshold_used'] ?? 0.35;
9414 - $rag_context_for_storage['knowledge_base_type'] = $this->last_similarity_analysis['knowledge_base_type'] ?? 'WordPress Database';
9415 - $rag_context_for_storage['total_documents_checked'] = $this->last_similarity_analysis['total_checked'] ?? 0;
9416 - }
9417 -
9418 - if ($has_action_data) {
9419 - $rag_context_for_storage['action_analysis'] = $this->last_action_analysis;
9420 - }
9421 - }
9422 - $this->mxchat_save_chat_message($session_id, 'bot', $full_response, null, $rag_context_for_storage);
9423 - }
9424 -
9425 - return true;
9426 - }
9427 -
9428 - // Failure path — branch on whether SSE channel was opened.
9429 - return $this->mxchat_stream_emit_fallback(
9430 - 'openai',
9431 - $this->mxchat_generate_response_openai($selected_model, $api_key, $conversation_history, $relevant_content, $session_id),
9432 - $session_id,
9433 - $testing_data
9434 - );
9435 -
9436 - } catch (Exception $e) {
9437 - return $this->mxchat_stream_emit_fallback(
9438 - 'openai',
9439 - $this->mxchat_generate_response_openai($selected_model, $api_key, $conversation_history, $relevant_content, $session_id),
9440 - $session_id,
9441 - $testing_data
9442 - );
9443 - }
9444 -}
9445 -
9446 -/**
9447 - * Shared fallback emitter for streaming chat functions. Two outcomes:
9448 - * - streaming_headers_sent === true: SSE channel is open. Emit fallback content
9449 - * as `data: {...}\n\n` + `data: [DONE]\n\n` so the widget renders it as a
9450 - * normal bot bubble. Transcript row is persisted.
9451 - * - streaming_headers_sent === false: SSE channel never opened (retries
9452 - * exhausted on initial connect). Emit a clean JSON response — the path
9453 - * the widget would normally hit if streaming wasn't even attempted.
9454 - *
9455 - * Used by all six *_stream functions after their per-attempt retry loop.
9456 - */
9457 -private function mxchat_stream_emit_fallback($provider_hint, $regular_response, $session_id, $testing_data = null) {
9458 - $is_error_array = is_array($regular_response) && isset($regular_response['error']);
9459 -
9460 - if ($this->streaming_headers_sent) {
9461 - if ($is_error_array) {
9462 - echo "data: " . json_encode([
9463 - 'error' => true,
9464 - 'error_message' => $regular_response['error'],
9465 - 'error_code' => $regular_response['error_code'] ?? 'api_error',
9466 - 'text' => $regular_response['error'],
9467 - 'message' => $regular_response['error']
9468 - ]) . "\n\n";
9469 - echo "data: [DONE]\n\n";
9470 - flush();
9471 - return true;
9472 - }
9473 - $fallback_message = (string) $regular_response;
9474 - if (!empty($fallback_message) && !empty($session_id)) {
9475 - $this->mxchat_save_chat_message($session_id, 'bot', $fallback_message);
9476 - }
9477 - echo "data: " . json_encode(['content' => $fallback_message]) . "\n\n";
9478 - echo "data: [DONE]\n\n";
9479 - flush();
9480 - return true;
9481 - }
9482 -
9483 - // SSE channel never opened — clean JSON fallback.
9484 - if ($is_error_array) {
9485 - header('Content-Type: application/json');
9486 - echo json_encode(array(
9487 - 'error' => true,
9488 - 'error_message' => $regular_response['error'],
9489 - 'error_code' => $regular_response['error_code'] ?? 'api_error',
9490 - 'text' => $regular_response['error'],
9491 - 'message' => $regular_response['error'],
9492 - ));
9493 - return true;
9494 - }
9495 -
9496 - $fallback_message = (string) $regular_response;
9497 - if (!empty($fallback_message) && !empty($session_id)) {
9498 - $this->mxchat_save_chat_message($session_id, 'bot', $fallback_message);
9499 - }
9500 - $response_data = array(
9501 - 'text' => $fallback_message,
9502 - 'html' => '',
9503 - 'session_id' => $session_id,
9504 - );
9505 - if ($testing_data !== null) {
9506 - $response_data['testing_data'] = $testing_data;
9507 - }
9508 - header('Content-Type: application/json');
9509 - echo json_encode($response_data);
9510 - return true;
9511 -}
9512 -
9513 -/**
9514 - * Resolve custom (OpenAI-compatible) provider config from settings.
9515 - * Returns ['base_url','api_key','model','auth_scheme','api_version','chat_url','headers'].
9516 - */
9517 -private function mxchat_resolve_custom_provider() {
9518 - $base_url = isset($this->options['custom_provider_base_url']) ? rtrim(trim((string) $this->options['custom_provider_base_url']), '/') : '';
9519 - $api_key = isset($this->options['custom_provider_api_key']) ? trim((string) $this->options['custom_provider_api_key']) : '';
9520 - $model = isset($this->options['custom_provider_model']) ? trim((string) $this->options['custom_provider_model']) : '';
9521 - $auth_scheme = isset($this->options['custom_provider_auth_scheme']) ? $this->options['custom_provider_auth_scheme'] : 'bearer';
9522 - $api_version = isset($this->options['custom_provider_api_version']) ? trim((string) $this->options['custom_provider_api_version']) : '';
9523 -
9524 - $chat_url = $base_url . '/chat/completions';
9525 - if (!empty($api_version)) {
9526 - $chat_url .= (strpos($chat_url, '?') === false ? '?' : '&') . 'api-version=' . rawurlencode($api_version);
9527 - }
9528 -
9529 - $headers = array('Content-Type: application/json');
9530 - if (!empty($api_key)) {
9531 - if ($auth_scheme === 'api-key') {
9532 - $headers[] = 'api-key: ' . $api_key;
9533 - } else {
9534 - $headers[] = 'Authorization: Bearer ' . $api_key;
9535 - }
9536 - }
9537 -
9538 - return array(
9539 - 'base_url' => $base_url,
9540 - 'api_key' => $api_key,
9541 - 'model' => $model !== '' ? $model : 'default',
9542 - 'auth_scheme' => $auth_scheme,
9543 - 'api_version' => $api_version,
9544 - 'chat_url' => $chat_url,
9545 - 'headers' => $headers,
9546 - );
9547 -}
9548 -
9549 -/**
9550 - * Streaming chat completion against an OpenAI-compatible custom provider
9551 - * (Ollama, LM Studio, vLLM, llama.cpp, Azure OpenAI, etc.).
9552 - * Mirrors mxchat_generate_response_openai_stream but with parameterized URL/auth/model.
9553 - */
9554 -private function mxchat_generate_response_custom_stream($selected_model, $conversation_history, $relevant_content, $session_id, $testing_data = null) {
9555 - try {
9556 - $cfg = $this->mxchat_resolve_custom_provider();
9557 - if (empty($cfg['base_url'])) {
9558 - return array('error' => esc_html__('Custom provider Base URL is not configured.', 'mxchat'), 'error_code' => 'missing_custom_provider_base_url');
9559 - }
9560 -
9561 - $bot_id = $this->get_current_bot_id($session_id);
9562 - $system_prompt_instructions = $this->get_system_instructions($bot_id, $session_id);
9563 - if (!is_array($conversation_history)) {
9564 - $conversation_history = array();
9565 - }
9566 -
9567 - $formatted_conversation = array();
9568 - $formatted_conversation[] = array(
9569 - 'role' => 'system',
9570 - 'content' => $system_prompt_instructions . ' ' . $relevant_content,
9571 - );
9572 - foreach ($conversation_history as $message) {
9573 - if (is_array($message) && isset($message['role']) && isset($message['content'])) {
9574 - $role = $message['role'];
9575 - if ($role === 'bot' || $role === 'agent') { $role = 'assistant'; }
9576 - if (!in_array($role, array('system', 'assistant', 'user', 'function', 'tool'))) { $role = 'user'; }
9577 - $formatted_conversation[] = array('role' => $role, 'content' => $message['content']);
9578 - }
9579 - }
9580 -
9581 - if (headers_sent() || !function_exists('curl_init')) {
9582 - // No streaming capability — fall through to non-stream wrapper
9583 - $regular = $this->mxchat_generate_response_custom($selected_model, $conversation_history, $relevant_content);
9584 - if (!empty($regular) && !empty($session_id) && is_string($regular)) {
9585 - $this->mxchat_save_chat_message($session_id, 'bot', $regular);
9586 - }
9587 - $response_data = array('text' => is_string($regular) ? $regular : '', 'html' => '', 'session_id' => $session_id);
9588 - if ($testing_data !== null) { $response_data['testing_data'] = $testing_data; }
9589 - header('Content-Type: application/json');
9590 - echo json_encode($response_data);
9591 - return true;
9592 - }
9593 -
9594 - $request_body = array(
9595 - 'model' => $cfg['model'],
9596 - 'messages' => $formatted_conversation,
9597 - 'stream' => true,
9598 - );
9599 - $body = json_encode($request_body);
9600 -
9601 - // V2 retry-on-initial-connect: setup_streaming_headers is lazy-fired in WRITEFUNCTION.
9602 -
9603 - $captured_status_code = 0;
9604 - $captured_body_pre_stream = '';
9605 - $full_response = '';
9606 - $stream_started = false;
9607 - $buffer = '';
9608 - $errno = 0;
9609 - $http_code = 0;
9610 - $max_attempts = $this->mxchat_retry_enabled() ? 3 : 1;
9611 - $backoff_ms = array(0, 750, 2000);
9612 -
9613 - for ($attempt = 0; $attempt < $max_attempts; $attempt++) {
9614 - if ($attempt > 0 && $backoff_ms[$attempt] > 0) {
9615 - usleep($backoff_ms[$attempt] * 1000);
9616 - }
9617 -
9618 - $captured_status_code = 0;
9619 - $captured_body_pre_stream = '';
9620 - $full_response = '';
9621 - $stream_started = false;
9622 - $buffer = '';
9623 -
9624 - $ch = curl_init();
9625 - curl_setopt($ch, CURLOPT_URL, $cfg['chat_url']);
9626 - curl_setopt($ch, CURLOPT_RETURNTRANSFER, false);
9627 - curl_setopt($ch, CURLOPT_POST, true);
9628 - curl_setopt($ch, CURLOPT_POSTFIELDS, $body);
9629 - curl_setopt($ch, CURLOPT_HTTPHEADER, $cfg['headers']);
9630 - curl_setopt($ch, CURLOPT_SSL_VERIFYPEER, true);
9631 - curl_setopt($ch, CURLOPT_TIMEOUT, 120);
9632 -
9633 - curl_setopt($ch, CURLOPT_HEADERFUNCTION, function($ch, $header) use (&$captured_status_code) {
9634 - if ($captured_status_code === 0 && preg_match('#^HTTP/\S+\s+(\d+)\b#', $header, $m)) {
9635 - $captured_status_code = (int) $m[1];
9636 - }
9637 - return strlen($header);
9638 - });
9639 -
9640 - curl_setopt($ch, CURLOPT_WRITEFUNCTION, function($ch, $data) use (&$full_response, &$stream_started, &$buffer, &$captured_status_code, &$captured_body_pre_stream, $testing_data) {
9641 - if ($captured_status_code !== 0 && $captured_status_code !== 200) {
9642 - $captured_body_pre_stream .= $data;
9643 - return strlen($data);
9644 - }
9645 -
9646 - if (!$this->streaming_headers_sent) {
9647 - $this->setup_streaming_headers();
9648 - }
9649 -
9650 - if (!$stream_started && $testing_data !== null) {
9651 - echo "data: " . json_encode(array('testing_data' => $testing_data)) . "\n\n";
9652 - flush();
9653 - $stream_started = true;
9654 - }
9655 - $buffer .= $data;
9656 - $lines = explode("\n", $buffer);
9657 - $buffer = array_pop($lines);
9658 - foreach ($lines as $line) {
9659 - if (trim($line) === '') { continue; }
9660 - if (strpos($line, 'data: ') !== 0) { continue; }
9661 - $json_str = substr($line, 6);
9662 - if (trim($json_str) === '[DONE]') {
9663 - echo "data: [DONE]\n\n";
9664 - flush();
9665 - continue;
9666 - }
9667 - $json = json_decode(trim($json_str), true);
9668 - if ($json && isset($json['choices'][0]['delta']['content'])) {
9669 - $content = $json['choices'][0]['delta']['content'];
9670 - $full_response .= $content;
9671 - echo "data: " . json_encode(array('content' => $content)) . "\n\n";
9672 - flush();
9673 - }
9674 - }
9675 - return strlen($data);
9676 - });
9677 -
9678 - $response = curl_exec($ch);
9679 - $errno = curl_errno($ch);
9680 - $http_code = $captured_status_code !== 0 ? $captured_status_code : (int) curl_getinfo($ch, CURLINFO_HTTP_CODE);
9681 - curl_close($ch);
9682 -
9683 - if (!$errno && $http_code === 200) {
9684 - break;
9685 - }
9686 -
9687 - $is_transient = $this->mxchat_is_transient_provider_error_raw($http_code, $captured_body_pre_stream, 'openai', $errno);
9688 - $can_retry = !$this->streaming_headers_sent
9689 - && ($attempt + 1) < $max_attempts
9690 - && $is_transient;
9691 -
9692 - if (defined('WP_DEBUG') && WP_DEBUG) {
9693 - error_log(sprintf(
9694 - '[MxChat] custom_stream initial-connect failure (attempt=%d/%d, status=%d, errno=%d, transient=%s, %s).',
9695 - $attempt + 1, $max_attempts, $http_code, $errno,
9696 - $is_transient ? 'yes' : 'no',
9697 - $can_retry ? 'Retrying.' : 'Giving up.'
9698 - ));
9699 - }
9700 -
9701 - if (!$can_retry) {
9702 - break;
9703 - }
9704 - }
9705 -
9706 - if (!$errno && $http_code === 200) {
9707 - if (!empty($full_response) && !empty($session_id)) {
9708 - $this->mxchat_save_chat_message($session_id, 'bot', $full_response);
9709 - }
9710 - return true;
9711 - }
9712 -
9713 - return $this->mxchat_stream_emit_fallback(
9714 - 'openai',
9715 - $this->mxchat_generate_response_custom($selected_model, $conversation_history, $relevant_content),
9716 - $session_id,
9717 - $testing_data
9718 - );
9719 -
9720 - } catch (Exception $e) {
9721 - return array('error' => sprintf(esc_html__('Custom provider error: %s', 'mxchat'), $e->getMessage()), 'error_code' => 'custom_provider_exception');
9722 - }
9723 -}
9724 -
9725 -/**
9726 - * Non-streaming chat completion against a custom OpenAI-compatible provider.
9727 - * Returns string content on success, array['error'=>...] on failure.
9728 - */
9729 -private function mxchat_generate_response_custom($selected_model, $conversation_history, $relevant_content) {
9730 - $cfg = $this->mxchat_resolve_custom_provider();
9731 - if (empty($cfg['base_url'])) {
9732 - return array('error' => esc_html__('Custom provider Base URL is not configured.', 'mxchat'), 'error_code' => 'missing_custom_provider_base_url');
9733 - }
9734 -
9735 - $bot_id = $this->get_current_bot_id(null);
9736 - $system_prompt_instructions = $this->get_system_instructions($bot_id, null);
9737 - if (!is_array($conversation_history)) {
9738 - $conversation_history = array();
9739 - }
9740 -
9741 - $messages = array(array(
9742 - 'role' => 'system',
9743 - 'content' => $system_prompt_instructions . ' ' . $relevant_content,
9744 - ));
9745 - foreach ($conversation_history as $message) {
9746 - if (is_array($message) && isset($message['role']) && isset($message['content'])) {
9747 - $role = $message['role'];
9748 - if ($role === 'bot' || $role === 'agent') { $role = 'assistant'; }
9749 - if (!in_array($role, array('system', 'assistant', 'user', 'function', 'tool'))) { $role = 'user'; }
9750 - $messages[] = array('role' => $role, 'content' => $message['content']);
9751 - }
9752 - }
9753 -
9754 - $headers_assoc = array('Content-Type' => 'application/json');
9755 - if (!empty($cfg['api_key'])) {
9756 - if ($cfg['auth_scheme'] === 'api-key') {
9757 - $headers_assoc['api-key'] = $cfg['api_key'];
9758 - } else {
9759 - $headers_assoc['Authorization'] = 'Bearer ' . $cfg['api_key'];
9760 - }
9761 - }
9762 -
9763 - $response = $this->mxchat_provider_call_with_retry($cfg['chat_url'], array(
9764 - 'headers' => $headers_assoc,
9765 - 'body' => wp_json_encode(array(
9766 - 'model' => $cfg['model'],
9767 - 'messages' => $messages,
9768 - )),
9769 - 'timeout' => 120,
9770 - ), 'openai');
9771 -
9772 - if (is_wp_error($response)) {
9773 - return array('error' => sprintf(esc_html__('Custom provider request failed: %s', 'mxchat'), $response->get_error_message()), 'error_code' => 'custom_provider_network_error');
9774 - }
9775 - $code = (int) wp_remote_retrieve_response_code($response);
9776 - if ($code < 200 || $code >= 300) {
9777 - return array('error' => sprintf(esc_html__('Custom provider returned HTTP %d.', 'mxchat'), $code), 'error_code' => 'custom_provider_http_error');
9778 - }
9779 - $body = json_decode(wp_remote_retrieve_body($response), true);
9780 - if (isset($body['choices'][0]['message']['content'])) {
9781 - return (string) $body['choices'][0]['message']['content'];
9782 - }
9783 - return array('error' => esc_html__('Custom provider returned an unexpected response shape.', 'mxchat'), 'error_code' => 'custom_provider_response_shape');
9784 -}
9785 -
9786 -/**
9787 - * Generate response using OpenAI Responses API with web search tool
9788 - * This uses the newer Responses API which supports web search functionality
9789 - */
9790 -private function mxchat_generate_response_openai_web_search($selected_model, $api_key, $conversation_history, $relevant_content, $session_id, $testing_data = null, $streaming = false) {
9791 - // OpenAI retires gpt-5.1-chat-latest / gpt-5.3-chat-latest on 2026-08-10
9792 - // (replacement gpt-5.6-sol). Read-time rescue mirrors the chat paths
9793 - // (plan e46b8f).
9794 - if ($selected_model === 'gpt-5.1-chat-latest' || $selected_model === 'gpt-5.3-chat-latest') { $selected_model = 'gpt-5.6-sol'; }
9795 - try {
9796 - $bot_id = $this->get_current_bot_id($session_id);
9797 - $system_prompt_instructions = $this->get_system_instructions($bot_id, $session_id);
9798 -
9799 - if (!is_array($conversation_history)) {
9800 - $conversation_history = array();
9801 - }
9802 -
9803 - // Build the input for Responses API
9804 - // The Responses API uses a different format - we need to construct the input properly
9805 - $input_parts = [];
9806 -
9807 - // Add system instructions as context
9808 - $system_context = $system_prompt_instructions . "\n\n" . $relevant_content;
9809 -
9810 - // Build conversation as input items for Responses API
9811 - foreach ($conversation_history as $message) {
9812 - if (is_array($message) && isset($message['role']) && isset($message['content'])) {
9813 - $role = $message['role'];
9814 - if ($role === 'bot' || $role === 'agent') {
9815 - $role = 'assistant';
9816 - }
9817 - if (!in_array($role, ['assistant', 'user'])) {
9818 - $role = 'user';
9819 - }
9820 - $input_parts[] = [
9821 - 'type' => 'message',
9822 - 'role' => $role,
9823 - 'content' => $message['content']
9824 - ];
9825 - }
9826 - }
9827 -
9828 - // Build request body for Responses API
9829 - $request_body = [
9830 - 'model' => $selected_model,
9831 - 'input' => $input_parts,
9832 - 'instructions' => $system_context,
9833 - 'stream' => $streaming
9834 - ];
9835 -
9836 - // Only add web search tool if web search is enabled in settings
9837 - $web_search_enabled = isset($this->options['enable_web_search']) && $this->options['enable_web_search'] === 'on';
9838 - if ($web_search_enabled) {
9839 - $request_body['tools'] = [
9840 - ['type' => 'web_search']
9841 - ];
9842 - }
9843 -
9844 - // reasoning.effort — sourced from the core model catalog (plan-dcb71c),
9845 - // 'websearch' surface; frozen inline ladder in mxchat_reasoning_effort_fallback().
9846 - $effort = $this->mxchat_reasoning_effort_for($selected_model, 'websearch');
9847 - if ($effort !== null) {
9848 - $request_body['reasoning'] = ['effort' => $effort];
9849 - }
9850 -
9851 - //error_log("MXCHAT WEB SEARCH: Request body: " . json_encode($request_body));
9852 -
9853 - if ($streaming) {
9854 - return $this->mxchat_web_search_streaming_response($request_body, $api_key, $session_id, $testing_data);
9855 - } else {
9856 - return $this->mxchat_web_search_non_streaming_response($request_body, $api_key, $session_id, $testing_data);
9857 - }
9858 -
9859 - } catch (Exception $e) {
9860 - //error_log("MXCHAT WEB SEARCH ERROR: " . $e->getMessage());
9861 - return [
9862 - 'error' => sprintf(esc_html__('Web search error: %s', 'mxchat'), esc_html($e->getMessage())),
9863 - 'error_code' => 'web_search_exception'
9864 - ];
9865 - }
9866 -}
9867 -
9868 -/**
9869 - * Handle non-streaming web search response
9870 - */
9871 -private function mxchat_web_search_non_streaming_response($request_body, $api_key, $session_id, $testing_data) {
9872 - $request_body['stream'] = false;
9873 -
9874 - $response = $this->mxchat_provider_call_with_retry('https://api.openai.com/v1/responses', array(
9875 - 'headers' => array(
9876 - 'Authorization' => 'Bearer ' . $api_key,
9877 - 'Content-Type' => 'application/json'
9878 - ),
9879 - 'body' => json_encode($request_body),
9880 - 'timeout' => 90
9881 - ), 'openai');
9882 -
9883 - if (is_wp_error($response)) {
9884 - //error_log("MXCHAT WEB SEARCH ERROR: WP Error: " . $response->get_error_message());
9885 - return [
9886 - 'error' => esc_html__('Failed to connect to OpenAI web search API', 'mxchat'),
9887 - 'error_code' => 'web_search_connection_error'
9888 - ];
9889 - }
9890 -
9891 - $response_code = wp_remote_retrieve_response_code($response);
9892 - $response_body = wp_remote_retrieve_body($response);
9893 -
9894 - //error_log("MXCHAT WEB SEARCH: Response code: " . $response_code);
9895 - //error_log("MXCHAT WEB SEARCH: Response body (first 2000): " . substr($response_body, 0, 2000));
9896 -
9897 - if ($response_code !== 200) {
9898 - $error_data = json_decode($response_body, true);
9899 - $error_message = $this->extract_provider_error($error_data, 'Unknown API error');
9900 - return [
9901 - 'error' => sprintf(esc_html__('OpenAI API error: %s', 'mxchat'), esc_html($error_message)),
9902 - 'error_code' => 'web_search_api_error'
9903 - ];
9904 - }
9905 -
9906 - $result = json_decode($response_body, true);
9907 -
9908 - if (json_last_error() !== JSON_ERROR_NONE) {
9909 - return [
9910 - 'error' => esc_html__('Invalid response from OpenAI', 'mxchat'),
9911 - 'error_code' => 'web_search_json_error'
9912 - ];
9913 - }
9914 -
9915 - // Extract the response text and citations from Responses API format
9916 - $output_text = '';
9917 - $citations = [];
9918 -
9919 - if (isset($result['output'])) {
9920 - foreach ($result['output'] as $output_item) {
9921 - if ($output_item['type'] === 'message' && isset($output_item['content'])) {
9922 - foreach ($output_item['content'] as $content_item) {
9923 - if ($content_item['type'] === 'output_text') {
9924 - $output_text .= $content_item['text'];
9925 -
9926 - // Extract citations/annotations
9927 - if (isset($content_item['annotations'])) {
9928 - foreach ($content_item['annotations'] as $annotation) {
9929 - if ($annotation['type'] === 'url_citation') {
9930 - $citations[] = [
9931 - 'url' => $annotation['url'],
9932 - 'title' => $annotation['title'] ?? ''
9933 - ];
9934 - }
9935 - }
9936 - }
9937 - }
9938 - }
9939 - }
9940 - }
9941 - }
9942 -
9943 - // If we have citations, append them to the response
9944 - if (!empty($citations)) {
9945 - $output_text .= "\n\n**Sources:**\n";
9946 - $seen_urls = [];
9947 - foreach ($citations as $citation) {
9948 - if (!in_array($citation['url'], $seen_urls)) {
9949 - $seen_urls[] = $citation['url'];
9950 - $title = !empty($citation['title']) ? $citation['title'] : $citation['url'];
9951 - $output_text .= "- [" . $title . "](" . $citation['url'] . ")\n";
9952 - }
9953 - }
9954 - }
9955 -
9956 - // Transcript save is handled by the main handler (mxchat_handle_chat_request)
9957 - // which includes rag_context for the "sources" link in transcripts.
9958 -
9959 - // plan-4aa8e5: a 200 whose output carries no output_text (status
9960 - // "incomplete" with max_output_tokens exhausted, content-filter-emptied
9961 - // output, shape drift) previously fell through and returned '' — a
9962 - // silent empty bot bubble. This is the DEFAULT model path
9963 - // (the default OpenAI chat model routes through /v1/responses).
9964 - if (trim($output_text) === '') {
9965 - return $this->mxchat_empty_completion_error($result, 'OpenAI');
9966 - }
9967 -
9968 - return $output_text;
9969 -}
9970 -
9971 -/**
9972 - * Handle streaming web search response using Responses API
9973 - */
9974 -private function mxchat_web_search_streaming_response($request_body, $api_key, $session_id, $testing_data) {
9975 - $request_body['stream'] = true;
9976 -
9977 - // Check if we can stream
9978 - if (headers_sent() || !function_exists('curl_init')) {
9979 - // Fallback to non-streaming
9980 - return $this->mxchat_web_search_non_streaming_response($request_body, $api_key, $session_id, $testing_data);
9981 - }
9982 -
9983 - // Setup streaming headers
9984 - $this->setup_streaming_headers();
9985 -
9986 - $ch = curl_init();
9987 - curl_setopt($ch, CURLOPT_URL, 'https://api.openai.com/v1/responses');
9988 - curl_setopt($ch, CURLOPT_RETURNTRANSFER, false);
9989 - curl_setopt($ch, CURLOPT_POST, true);
9990 - curl_setopt($ch, CURLOPT_POSTFIELDS, json_encode($request_body));
9991 - curl_setopt($ch, CURLOPT_HTTPHEADER, array(
9992 - 'Content-Type: application/json',
9993 - 'Authorization: Bearer ' . $api_key
9994 - ));
9995 - curl_setopt($ch, CURLOPT_SSL_VERIFYPEER, true);
9996 - curl_setopt($ch, CURLOPT_TIMEOUT, 120);
9997 -
9998 - $full_response = '';
9999 - $stream_started = false;
10000 - $buffer = '';
10001 - $citations = [];
10002 - $empty_error_emitted = false;
10003 -
10004 - curl_setopt($ch, CURLOPT_WRITEFUNCTION, function($ch, $data) use (&$full_response, &$stream_started, &$buffer, &$citations, &$empty_error_emitted, $testing_data) {
10005 - // Send testing data as first event if available
10006 - if (!$stream_started && $testing_data !== null) {
10007 - echo "data: " . json_encode(['testing_data' => $testing_data]) . "\n\n";
10008 - flush();
10009 - $stream_started = true;
10010 - }
10011 -
10012 - $buffer .= $data;
10013 - $lines = explode("\n", $buffer);
10014 - $buffer = array_pop($lines);
10015 -
10016 - foreach ($lines as $line) {
10017 - if (trim($line) === '') continue;
10018 - if (strpos($line, 'data: ') !== 0) continue;
10019 -
10020 - $json_str = substr($line, 6);
10021 -
10022 - if (trim($json_str) === '[DONE]') {
10023 - // Append citations if we have any
10024 - if (!empty($citations)) {
10025 - $citation_text = "\n\n**Sources:**\n";
10026 - $seen_urls = [];
10027 - foreach ($citations as $citation) {
10028 - if (!in_array($citation['url'], $seen_urls)) {
10029 - $seen_urls[] = $citation['url'];
10030 - $title = !empty($citation['title']) ? $citation['title'] : $citation['url'];
10031 - $citation_text .= "- [" . $title . "](" . $citation['url'] . ")\n";
10032 - }
10033 - }
10034 - echo "data: " . json_encode(['content' => $citation_text]) . "\n\n";
10035 - $full_response .= $citation_text;
10036 - flush();
10037 - }
10038 - // plan-4aa8e5: zero deltas streamed → say so instead of
10039 - // closing a silent empty bubble (client renders text events).
10040 - if (trim($full_response) === '' && !$empty_error_emitted) {
10041 - $empty_error_emitted = true;
10042 - echo "data: " . json_encode(['content' => esc_html__('The AI provider returned an empty response. Please try again.', 'mxchat')]) . "\n\n";
10043 - }
10044 - echo "data: [DONE]\n\n";
10045 - flush();
10046 - continue;
10047 - }
10048 -
10049 - $json = json_decode(trim($json_str), true);
10050 - if (!$json) continue;
10051 -
10052 - // Handle Responses API streaming events
10053 - // The format is different from Chat Completions
10054 - if (isset($json['type'])) {
10055 - switch ($json['type']) {
10056 - case 'response.output_text.delta':
10057 - // Text content delta
10058 - if (isset($json['delta'])) {
10059 - $content = $json['delta'];
10060 - $full_response .= $content;
10061 - echo "data: " . json_encode(['content' => $content]) . "\n\n";
10062 - flush();
10063 - }
10064 - break;
10065 -
10066 - case 'response.output_item.done':
10067 - // Check for citations in completed items
10068 - if (isset($json['item']['content'])) {
10069 - foreach ($json['item']['content'] as $content_item) {
10070 - if (isset($content_item['annotations'])) {
10071 - foreach ($content_item['annotations'] as $annotation) {
10072 - if ($annotation['type'] === 'url_citation') {
10073 - $citations[] = [
10074 - 'url' => $annotation['url'],
10075 - 'title' => $annotation['title'] ?? ''
10076 - ];
10077 - }
10078 - }
10079 - }
10080 - }
10081 - }
10082 - break;
10083 - }
10084 - }
10085 - }
10086 -
10087 - return strlen($data);
10088 - });
10089 -
10090 - $response = curl_exec($ch);
10091 - $http_code = curl_getinfo($ch, CURLINFO_HTTP_CODE);
10092 -
10093 - if (curl_errno($ch) || $http_code !== 200) {
10094 - $curl_error = curl_error($ch);
10095 - curl_close($ch);
10096 -
10097 - //error_log("MXCHAT WEB SEARCH STREAM ERROR: HTTP $http_code, cURL error: $curl_error");
10098 -
10099 - return $this->mxchat_stream_emit_fallback(
10100 - 'web_search',
10101 - $this->mxchat_web_search_non_streaming_response($request_body, $api_key, $session_id, $testing_data),
10102 - $session_id,
10103 - $testing_data
10104 - );
10105 - }
10106 -
10107 - curl_close($ch);
10108 -
10109 - // plan-4aa8e5: the Responses API can end its stream via typed events
10110 - // without a [DONE] line — if nothing was streamed at all, close out with
10111 - // the empty-completion message instead of leaving a silent bubble.
10112 - if (trim($full_response) === '' && !$empty_error_emitted) {
10113 - echo "data: " . json_encode(['content' => esc_html__('The AI provider returned an empty response. Please try again.', 'mxchat')]) . "\n\n";
10114 - echo "data: [DONE]\n\n";
10115 - flush();
10116 - }
10117 -
10118 - // Save the complete response with RAG context so the "sources" link
10119 - // appears in transcripts — mirrors the pattern used by Claude/OpenAI streaming.
10120 - if (!empty($full_response) && !empty($session_id)) {
10121 - $rag_context_for_storage = null;
10122 - $has_rag_data = $this->last_similarity_analysis !== null && !empty($this->last_similarity_analysis['top_matches']);
10123 - $has_action_data = isset($this->last_action_analysis) && !empty($this->last_action_analysis);
10124 -
10125 - if ($has_rag_data || $has_action_data) {
10126 - $rag_context_for_storage = [];
10127 -
10128 - if ($has_rag_data) {
10129 - $rag_context_for_storage['top_matches'] = $this->last_similarity_analysis['top_matches'];
10130 - $rag_context_for_storage['approved_urls'] = $this->current_valid_urls ?? [];
10131 - $rag_context_for_storage['similarity_threshold'] = $this->last_similarity_analysis['threshold_used'] ?? 0.35;
10132 - $rag_context_for_storage['knowledge_base_type'] = $this->last_similarity_analysis['knowledge_base_type'] ?? 'WordPress Database';
10133 - $rag_context_for_storage['total_documents_checked'] = $this->last_similarity_analysis['total_checked'] ?? 0;
10134 - }
10135 -
10136 - if ($has_action_data) {
10137 - $rag_context_for_storage['action_analysis'] = $this->last_action_analysis;
10138 - }
10139 - }
10140 - $this->mxchat_save_chat_message($session_id, 'bot', $full_response, null, $rag_context_for_storage);
10141 - }
10142 -
10143 - return true;
10144 -}
10145 -
10146 -private function mxchat_generate_response_claude_stream($selected_model, $claude_api_key, $conversation_history, $relevant_content, $session_id, $testing_data = null) {
10147 - // Anthropic retired claude-opus-4-20250514 / claude-sonnet-4-20250514 on 2026-06-15.
10148 - // Read-time rescue: remap a saved dead ID to the current equivalent before the API call.
10149 - if ($selected_model === 'claude-opus-4-20250514') { $selected_model = 'claude-opus-4-8'; }
10150 - elseif ($selected_model === 'claude-sonnet-4-20250514') { $selected_model = 'claude-sonnet-4-6'; }
10151 - try {
10152 - // Get bot ID from session or request
10153 - $bot_id = $this->get_current_bot_id($session_id);
10154 -
10155 - // Get system prompt instructions using centralized function
10156 - $system_prompt_instructions = $this->get_system_instructions($bot_id, $session_id);
10157 - // Ensure conversation_history is an array
10158 - if (!is_array($conversation_history)) {
10159 - $conversation_history = array();
10160 - }
10161 -
10162 - // Clean and validate conversation history
10163 - foreach ($conversation_history as &$message) {
10164 - // Convert bot and agent roles to assistant
10165 - if ($message['role'] === 'bot' || $message['role'] === 'agent') {
10166 - $message['role'] = 'assistant';
10167 - }
10168 -
10169 - // Remove unsupported roles - Claude only supports 'assistant' and 'user'
10170 - if (!in_array($message['role'], ['assistant', 'user'])) {
10171 - $message['role'] = 'user';
10172 - }
10173 -
10174 - // Ensure content field exists
10175 - if (!isset($message['content']) || empty($message['content'])) {
10176 - $message['content'] = '';
10177 - }
10178 -
10179 - // Remove any unsupported fields
10180 - $message = array_intersect_key($message, array_flip(['role', 'content']));
10181 - }
10182 -
10183 - // Add relevant content as the latest user message
10184 - $conversation_history[] = [
10185 - 'role' => 'user',
10186 - 'content' => $relevant_content
10187 - ];
10188 -
10189 - // Prepare the request body with stream: true
10190 - $payload = [
10191 - 'model' => $selected_model,
10192 - 'messages' => $conversation_history,
10193 - 'max_tokens' => 1000,
10194 - 'temperature' => 0.8,
10195 - 'system' => $system_prompt_instructions,
10196 - 'stream' => true
10197 - ];
10198 - if ($this->mxchat_claude_omits_temperature($selected_model)) { unset($payload['temperature']); }
10199 - $body = json_encode($payload);
10200 -
10201 - // Check if we can actually stream (headers not sent, etc.)
10202 - if (headers_sent() || !function_exists('curl_init')) {
10203 - // Fallback to regular response with testing data
10204 - //error_log("MxChat: Streaming not possible, falling back to regular response");
10205 - $regular_response = $this->mxchat_generate_response_claude(
10206 - $selected_model,
10207 - $claude_api_key,
10208 - array_slice($conversation_history, 0, -1), // Remove the added content
10209 - $relevant_content,
10210 - $session_id
10211 - );
10212 -
10213 - // Save bot response to transcript
10214 - if (!empty($regular_response) && !empty($session_id)) {
10215 - $this->mxchat_save_chat_message($session_id, 'bot', $regular_response);
10216 - }
10217 -
10218 - // Return as JSON with testing data
10219 - $response_data = [
10220 - 'text' => $regular_response,
10221 - 'html' => '',
10222 - 'session_id' => $session_id
10223 - ];
10224 -
10225 - if ($testing_data !== null) {
10226 - $response_data['testing_data'] = $testing_data;
10227 - //error_log("MxChat Testing: Added testing data to Claude fallback response");
10228 - }
10229 -
10230 - // Clear any streaming headers and send JSON
10231 - if (headers_sent() === false) {
10232 - header('Content-Type: application/json');
10233 - }
10234 - echo json_encode($response_data);
10235 - return true; // Indicate we handled the response
10236 - }
10237 -
10238 - // V2 retry-on-initial-connect: setup_streaming_headers is lazy-fired in WRITEFUNCTION.
10239 -
10240 - $captured_status_code = 0;
10241 - $captured_body_pre_stream = '';
10242 - $full_response = '';
10243 - $stream_started = false;
10244 - $buffer = '';
10245 - $errno = 0;
10246 - $http_code = 0;
10247 - $max_attempts = $this->mxchat_retry_enabled() ? 3 : 1;
10248 - $backoff_ms = array(0, 750, 2000);
10249 -
10250 - for ($attempt = 0; $attempt < $max_attempts; $attempt++) {
10251 - if ($attempt > 0 && $backoff_ms[$attempt] > 0) {
10252 - usleep($backoff_ms[$attempt] * 1000);
10253 - }
10254 -
10255 - $captured_status_code = 0;
10256 - $captured_body_pre_stream = '';
10257 - $full_response = '';
10258 - $stream_started = false;
10259 - $buffer = '';
10260 -
10261 - $ch = curl_init();
10262 - curl_setopt($ch, CURLOPT_URL, 'https://api.anthropic.com/v1/messages');
10263 - curl_setopt($ch, CURLOPT_RETURNTRANSFER, false);
10264 - curl_setopt($ch, CURLOPT_POST, true);
10265 - curl_setopt($ch, CURLOPT_POSTFIELDS, $body);
10266 - curl_setopt($ch, CURLOPT_HTTPHEADER, array(
10267 - 'Content-Type: application/json',
10268 - 'x-api-key: ' . $claude_api_key,
10269 - 'anthropic-version: 2023-06-01'
10270 - ));
10271 - curl_setopt($ch, CURLOPT_SSL_VERIFYPEER, true);
10272 - curl_setopt($ch, CURLOPT_TIMEOUT, 60);
10273 -
10274 - curl_setopt($ch, CURLOPT_HEADERFUNCTION, function($ch, $header) use (&$captured_status_code) {
10275 - if ($captured_status_code === 0 && preg_match('#^HTTP/\S+\s+(\d+)\b#', $header, $m)) {
10276 - $captured_status_code = (int) $m[1];
10277 - }
10278 - return strlen($header);
10279 - });
10280 -
10281 - curl_setopt($ch, CURLOPT_WRITEFUNCTION, function($ch, $data) use (&$full_response, &$stream_started, &$buffer, &$captured_status_code, &$captured_body_pre_stream, $testing_data) {
10282 - if ($captured_status_code !== 0 && $captured_status_code !== 200) {
10283 - $captured_body_pre_stream .= $data;
10284 - return strlen($data);
10285 - }
10286 -
10287 - if (!$this->streaming_headers_sent) {
10288 - $this->setup_streaming_headers();
10289 - }
10290 -
10291 - if (!$stream_started && $testing_data !== null) {
10292 - echo "data: " . json_encode(['testing_data' => $testing_data]) . "\n\n";
10293 - flush();
10294 - $stream_started = true;
10295 - }
10296 -
10297 - $buffer .= $data;
10298 - $lines = explode("\n", $buffer);
10299 - $buffer = array_pop($lines);
10300 -
10301 - foreach ($lines as $line) {
10302 - if (trim($line) === '') {
10303 - continue;
10304 - }
10305 -
10306 - if (strpos($line, 'event: ') === 0) {
10307 - continue;
10308 - }
10309 -
10310 - if (strpos($line, 'data: ') === 0) {
10311 - $json_str = substr($line, 6);
10312 -
10313 - $json = json_decode(trim($json_str), true);
10314 - if (json_last_error() !== JSON_ERROR_NONE) {
10315 - continue;
10316 - }
10317 -
10318 - if (isset($json['type'])) {
10319 - switch ($json['type']) {
10320 - case 'content_block_delta':
10321 - if (isset($json['delta']['text'])) {
10322 - $content = $json['delta']['text'];
10323 - $full_response .= $content;
10324 - echo "data: " . json_encode(['content' => $content]) . "\n\n";
10325 - flush();
10326 - }
10327 - break;
10328 -
10329 - case 'message_stop':
10330 - echo "data: [DONE]\n\n";
10331 - flush();
10332 - break;
10333 -
10334 - case 'error':
10335 - echo "data: " . json_encode(['error' => $this->extract_provider_error($json, 'Unknown error')]) . "\n\n";
10336 - flush();
10337 - break;
10338 - }
10339 - }
10340 - }
10341 - }
10342 -
10343 - return strlen($data);
10344 - });
10345 -
10346 - $response = curl_exec($ch);
10347 - $errno = curl_errno($ch);
10348 - $http_code = $captured_status_code !== 0 ? $captured_status_code : (int) curl_getinfo($ch, CURLINFO_HTTP_CODE);
10349 - curl_close($ch);
10350 -
10351 - if (!$errno && $http_code === 200) {
10352 - break;
10353 - }
10354 -
10355 - $is_transient = $this->mxchat_is_transient_provider_error_raw($http_code, $captured_body_pre_stream, 'anthropic', $errno);
10356 - $can_retry = !$this->streaming_headers_sent
10357 - && ($attempt + 1) < $max_attempts
10358 - && $is_transient;
10359 -
10360 - if (defined('WP_DEBUG') && WP_DEBUG) {
10361 - error_log(sprintf(
10362 - '[MxChat] claude_stream initial-connect failure (attempt=%d/%d, status=%d, errno=%d, transient=%s, %s).',
10363 - $attempt + 1, $max_attempts, $http_code, $errno,
10364 - $is_transient ? 'yes' : 'no',
10365 - $can_retry ? 'Retrying.' : 'Giving up.'
10366 - ));
10367 - }
10368 -
10369 - if (!$can_retry) {
10370 - break;
10371 - }
10372 - }
10373 -
10374 - if ($errno || $http_code !== 200) {
10375 - return $this->mxchat_stream_emit_fallback(
10376 - 'anthropic',
10377 - $this->mxchat_generate_response_claude($selected_model, $claude_api_key, array_slice($conversation_history, 0, -1), $relevant_content, $session_id),
10378 - $session_id,
10379 - $testing_data
10380 - );
10381 - }
10382 -
10383 - // Save the complete response to maintain chat persistence
10384 - if (!empty($full_response) && !empty($session_id)) {
10385 - // Prepare RAG context for streaming response
10386 - $rag_context_for_storage = null;
10387 - $has_rag_data = $this->last_similarity_analysis !== null && !empty($this->last_similarity_analysis['top_matches']);
10388 - $has_action_data = isset($this->last_action_analysis) && !empty($this->last_action_analysis);
10389 -
10390 - if ($has_rag_data || $has_action_data) {
10391 - $rag_context_for_storage = [];
10392 -
10393 - if ($has_rag_data) {
10394 - $rag_context_for_storage['top_matches'] = $this->last_similarity_analysis['top_matches'];
10395 - $rag_context_for_storage['approved_urls'] = $this->current_valid_urls ?? [];
10396 - $rag_context_for_storage['similarity_threshold'] = $this->last_similarity_analysis['threshold_used'] ?? 0.35;
10397 - $rag_context_for_storage['knowledge_base_type'] = $this->last_similarity_analysis['knowledge_base_type'] ?? 'WordPress Database';
10398 - $rag_context_for_storage['total_documents_checked'] = $this->last_similarity_analysis['total_checked'] ?? 0;
10399 - }
10400 -
10401 - if ($has_action_data) {
10402 - $rag_context_for_storage['action_analysis'] = $this->last_action_analysis;
10403 - }
10404 - }
10405 - $this->mxchat_save_chat_message($session_id, 'bot', $full_response, null, $rag_context_for_storage);
10406 - }
10407 -
10408 - return true; // Indicate streaming completed successfully
10409 -
10410 - } catch (Exception $e) {
10411 - return $this->mxchat_stream_emit_fallback(
10412 - 'anthropic',
10413 - $this->mxchat_generate_response_claude($selected_model, $claude_api_key, $conversation_history, $relevant_content, $session_id),
10414 - $session_id,
10415 - $testing_data
10416 - );
10417 - }
10418 -}
10419 -private function mxchat_generate_response_xai_stream($selected_model, $xai_api_key, $conversation_history, $relevant_content, $session_id, $testing_data = null) {
10420 - try {
10421 - // Get bot ID from session or request
10422 - $bot_id = $this->get_current_bot_id($session_id);
10423 -
10424 - // Get system prompt instructions using centralized function
10425 - $system_prompt_instructions = $this->get_system_instructions($bot_id, $session_id);
10426 -
10427 - // Ensure conversation_history is an array
10428 - if (!is_array($conversation_history)) {
10429 - $conversation_history = array();
10430 - }
10431 -
10432 - // Format conversation history for X.AI (same as OpenAI format)
10433 - $formatted_conversation = array();
10434 -
10435 - $formatted_conversation[] = array(
10436 - 'role' => 'system',
10437 - 'content' => $system_prompt_instructions . " " . $relevant_content
10438 - );
10439 -
10440 - foreach ($conversation_history as $message) {
10441 - if (is_array($message) && isset($message['role']) && isset($message['content'])) {
10442 - $role = $message['role'];
10443 - if ($role === 'bot' || $role === 'agent') {
10444 - $role = 'assistant';
10445 - }
10446 - if (!in_array($role, ['system', 'assistant', 'user', 'function', 'tool'])) {
10447 - $role = 'user';
10448 - }
10449 - $formatted_conversation[] = array(
10450 - 'role' => $role,
10451 - 'content' => $message['content']
10452 - );
10453 - }
10454 - }
10455 -
10456 - // Check if we can actually stream
10457 - if (headers_sent() || !function_exists('curl_init')) {
10458 - // Fallback to regular response with testing data
10459 - //error_log("MxChat: X.AI streaming not possible, falling back to regular response");
10460 - $regular_response = $this->mxchat_generate_response_xai(
10461 - $selected_model,
10462 - $xai_api_key,
10463 - $conversation_history,
10464 - $relevant_content,
10465 - $session_id
10466 - );
10467 -
10468 - // Save bot response to transcript
10469 - if (!empty($regular_response) && !empty($session_id)) {
10470 - $this->mxchat_save_chat_message($session_id, 'bot', $regular_response);
10471 - }
10472 -
10473 - $response_data = [
10474 - 'text' => $regular_response,
10475 - 'html' => '',
10476 - 'session_id' => $session_id
10477 - ];
10478 -
10479 - if ($testing_data !== null) {
10480 - $response_data['testing_data'] = $testing_data;
10481 - //error_log("MxChat Testing: Added testing data to X.AI fallback response");
10482 - }
10483 -
10484 - header('Content-Type: application/json');
10485 - echo json_encode($response_data);
10486 - return true;
10487 - }
10488 -
10489 - // Prepare the request body with stream: true
10490 - $body = json_encode([
10491 - 'model' => $selected_model,
10492 - 'messages' => $formatted_conversation,
10493 - 'temperature' => 0.8,
10494 - 'stream' => true
10495 - ]);
10496 -
10497 - // V2 retry-on-initial-connect: setup_streaming_headers is lazy-fired in WRITEFUNCTION.
10498 -
10499 - $captured_status_code = 0;
10500 - $captured_body_pre_stream = '';
10501 - $full_response = '';
10502 - $stream_started = false;
10503 - $buffer = '';
10504 - $errno = 0;
10505 - $http_code = 0;
10506 - $max_attempts = $this->mxchat_retry_enabled() ? 3 : 1;
10507 - $backoff_ms = array(0, 750, 2000);
10508 -
10509 - for ($attempt = 0; $attempt < $max_attempts; $attempt++) {
10510 - if ($attempt > 0 && $backoff_ms[$attempt] > 0) {
10511 - usleep($backoff_ms[$attempt] * 1000);
10512 - }
10513 -
10514 - $captured_status_code = 0;
10515 - $captured_body_pre_stream = '';
10516 - $full_response = '';
10517 - $stream_started = false;
10518 - $buffer = '';
10519 -
10520 - $ch = curl_init();
10521 - curl_setopt($ch, CURLOPT_URL, 'https://api.x.ai/v1/chat/completions');
10522 - curl_setopt($ch, CURLOPT_RETURNTRANSFER, false);
10523 - curl_setopt($ch, CURLOPT_POST, true);
10524 - curl_setopt($ch, CURLOPT_POSTFIELDS, $body);
10525 - curl_setopt($ch, CURLOPT_HTTPHEADER, array(
10526 - 'Content-Type: application/json',
10527 - 'Authorization: Bearer ' . $xai_api_key
10528 - ));
10529 - curl_setopt($ch, CURLOPT_SSL_VERIFYPEER, true);
10530 - curl_setopt($ch, CURLOPT_TIMEOUT, 60);
10531 -
10532 - curl_setopt($ch, CURLOPT_HEADERFUNCTION, function($ch, $header) use (&$captured_status_code) {
10533 - if ($captured_status_code === 0 && preg_match('#^HTTP/\S+\s+(\d+)\b#', $header, $m)) {
10534 - $captured_status_code = (int) $m[1];
10535 - }
10536 - return strlen($header);
10537 - });
10538 -
10539 - curl_setopt($ch, CURLOPT_WRITEFUNCTION, function($ch, $data) use (&$full_response, &$stream_started, &$buffer, &$captured_status_code, &$captured_body_pre_stream, $testing_data) {
10540 - if ($captured_status_code !== 0 && $captured_status_code !== 200) {
10541 - $captured_body_pre_stream .= $data;
10542 - return strlen($data);
10543 - }
10544 -
10545 - if (!$this->streaming_headers_sent) {
10546 - $this->setup_streaming_headers();
10547 - }
10548 -
10549 - if (!$stream_started && $testing_data !== null) {
10550 - echo "data: " . json_encode(['testing_data' => $testing_data]) . "\n\n";
10551 - flush();
10552 - $stream_started = true;
10553 - }
10554 -
10555 - $buffer .= $data;
10556 - $lines = explode("\n", $buffer);
10557 - $buffer = array_pop($lines);
10558 -
10559 - foreach ($lines as $line) {
10560 - if (trim($line) === '') {
10561 - continue;
10562 - }
10563 - if (strpos($line, 'data: ') !== 0) {
10564 - continue;
10565 - }
10566 -
10567 - $json_str = substr($line, 6);
10568 -
10569 - if (trim($json_str) === '[DONE]') {
10570 - echo "data: [DONE]\n\n";
10571 - flush();
10572 - continue;
10573 - }
10574 -
10575 - $json = json_decode(trim($json_str), true);
10576 - if ($json && isset($json['choices'][0]['delta']['content'])) {
10577 - $content = $json['choices'][0]['delta']['content'];
10578 - $full_response .= $content;
10579 - echo "data: " . json_encode(['content' => $content]) . "\n\n";
10580 - flush();
10581 - }
10582 - }
10583 -
10584 - return strlen($data);
10585 - });
10586 -
10587 - $response = curl_exec($ch);
10588 - $errno = curl_errno($ch);
10589 - $http_code = $captured_status_code !== 0 ? $captured_status_code : (int) curl_getinfo($ch, CURLINFO_HTTP_CODE);
10590 - curl_close($ch);
10591 -
10592 - if (!$errno && $http_code === 200) {
10593 - break;
10594 - }
10595 -
10596 - $is_transient = $this->mxchat_is_transient_provider_error_raw($http_code, $captured_body_pre_stream, 'xai', $errno);
10597 - $can_retry = !$this->streaming_headers_sent
10598 - && ($attempt + 1) < $max_attempts
10599 - && $is_transient;
10600 -
10601 - if (defined('WP_DEBUG') && WP_DEBUG) {
10602 - error_log(sprintf(
10603 - '[MxChat] xai_stream initial-connect failure (attempt=%d/%d, status=%d, errno=%d, transient=%s, %s).',
10604 - $attempt + 1, $max_attempts, $http_code, $errno,
10605 - $is_transient ? 'yes' : 'no',
10606 - $can_retry ? 'Retrying.' : 'Giving up.'
10607 - ));
10608 - }
10609 -
10610 - if (!$can_retry) {
10611 - break;
10612 - }
10613 - }
10614 -
10615 - if ($errno || $http_code !== 200) {
10616 - return $this->mxchat_stream_emit_fallback(
10617 - 'xai',
10618 - $this->mxchat_generate_response_xai($selected_model, $xai_api_key, $conversation_history, $relevant_content, $session_id),
10619 - $session_id,
10620 - $testing_data
10621 - );
10622 - }
10623 -
10624 - // Save the complete response to maintain chat persistence
10625 - if (!empty($full_response) && !empty($session_id)) {
10626 - // Prepare RAG context for streaming response
10627 - $rag_context_for_storage = null;
10628 - $has_rag_data = $this->last_similarity_analysis !== null && !empty($this->last_similarity_analysis['top_matches']);
10629 - $has_action_data = isset($this->last_action_analysis) && !empty($this->last_action_analysis);
10630 -
10631 - if ($has_rag_data || $has_action_data) {
10632 - $rag_context_for_storage = [];
10633 -
10634 - if ($has_rag_data) {
10635 - $rag_context_for_storage['top_matches'] = $this->last_similarity_analysis['top_matches'];
10636 - $rag_context_for_storage['approved_urls'] = $this->current_valid_urls ?? [];
10637 - $rag_context_for_storage['similarity_threshold'] = $this->last_similarity_analysis['threshold_used'] ?? 0.35;
10638 - $rag_context_for_storage['knowledge_base_type'] = $this->last_similarity_analysis['knowledge_base_type'] ?? 'WordPress Database';
10639 - $rag_context_for_storage['total_documents_checked'] = $this->last_similarity_analysis['total_checked'] ?? 0;
10640 - }
10641 -
10642 - if ($has_action_data) {
10643 - $rag_context_for_storage['action_analysis'] = $this->last_action_analysis;
10644 - }
10645 - }
10646 - $this->mxchat_save_chat_message($session_id, 'bot', $full_response, null, $rag_context_for_storage);
10647 - }
10648 -
10649 - return true; // Indicate streaming completed successfully
10650 -
10651 - } catch (Exception $e) {
10652 - return $this->mxchat_stream_emit_fallback(
10653 - 'xai',
10654 - $this->mxchat_generate_response_xai($selected_model, $xai_api_key, $conversation_history, $relevant_content),
10655 - $session_id,
10656 - $testing_data
10657 - );
10658 - }
10659 -}
10660 -private function mxchat_generate_response_deepseek_stream($selected_model, $deepseek_api_key, $conversation_history, $relevant_content, $session_id, $testing_data = null) {
10661 - try {
10662 - // Get bot ID from session or request
10663 - $bot_id = $this->get_current_bot_id($session_id);
10664 -
10665 - // Get system prompt instructions using centralized function
10666 - $system_prompt_instructions = $this->get_system_instructions($bot_id, $session_id);
10667 -
10668 - // Ensure conversation_history is an array
10669 - if (!is_array($conversation_history)) {
10670 - $conversation_history = array();
10671 - }
10672 -
10673 - // Format conversation history for DeepSeek
10674 - $formatted_conversation = array();
10675 -
10676 - $formatted_conversation[] = array(
10677 - 'role' => 'system',
10678 - 'content' => $system_prompt_instructions . " " . $relevant_content
10679 - );
10680 -
10681 - foreach ($conversation_history as $message) {
10682 - if (is_array($message) && isset($message['role']) && isset($message['content'])) {
10683 - $role = $message['role'];
10684 - if ($role === 'bot' || $role === 'agent') {
10685 - $role = 'assistant';
10686 - }
10687 - if (!in_array($role, ['system', 'assistant', 'user', 'function', 'tool'])) {
10688 - $role = 'user';
10689 - }
10690 - $formatted_conversation[] = array(
10691 - 'role' => $role,
10692 - 'content' => $message['content']
10693 - );
10694 - }
10695 - }
10696 -
10697 - // Check if we can actually stream
10698 - if (headers_sent() || !function_exists('curl_init')) {
10699 - // Fallback to regular response with testing data
10700 - //error_log("MxChat: DeepSeek streaming not possible, falling back to regular response");
10701 - $regular_response = $this->mxchat_generate_response_deepseek(
10702 - $selected_model,
10703 - $deepseek_api_key,
10704 - $conversation_history,
10705 - $relevant_content,
10706 - $session_id
10707 - );
10708 -
10709 - // Save bot response to transcript
10710 - if (!empty($regular_response) && !empty($session_id)) {
10711 - $this->mxchat_save_chat_message($session_id, 'bot', $regular_response);
10712 - }
10713 -
10714 - $response_data = [
10715 - 'text' => $regular_response,
10716 - 'html' => '',
10717 - 'session_id' => $session_id
10718 - ];
10719 -
10720 - if ($testing_data !== null) {
10721 - $response_data['testing_data'] = $testing_data;
10722 - //error_log("MxChat Testing: Added testing data to DeepSeek fallback response");
10723 - }
10724 -
10725 - header('Content-Type: application/json');
10726 - echo json_encode($response_data);
10727 - return true;
10728 - }
10729 -
10730 - // Prepare the request body with stream: true
10731 - $body = json_encode([
10732 - 'model' => $selected_model,
10733 - 'messages' => $formatted_conversation,
10734 - 'temperature' => 0.8,
10735 - 'stream' => true,
10736 - // DeepSeek V4 defaults to thinking mode ON (temperature ignored,
10737 - // long silent reasoning before the first delta); the widget wants
10738 - // the legacy deepseek-chat semantics = non-thinking.
10739 - 'thinking' => ['type' => 'disabled']
10740 - ]);
10741 -
10742 - // V2 retry-on-initial-connect: setup_streaming_headers is lazy-fired in WRITEFUNCTION.
10743 -
10744 - $captured_status_code = 0;
10745 - $captured_body_pre_stream = '';
10746 - $full_response = '';
10747 - $stream_started = false;
10748 - $buffer = '';
10749 - $errno = 0;
10750 - $http_code = 0;
10751 - $max_attempts = $this->mxchat_retry_enabled() ? 3 : 1;
10752 - $backoff_ms = array(0, 750, 2000);
10753 -
10754 - for ($attempt = 0; $attempt < $max_attempts; $attempt++) {
10755 - if ($attempt > 0 && $backoff_ms[$attempt] > 0) {
10756 - usleep($backoff_ms[$attempt] * 1000);
10757 - }
10758 -
10759 - $captured_status_code = 0;
10760 - $captured_body_pre_stream = '';
10761 - $full_response = '';
10762 - $stream_started = false;
10763 - $buffer = '';
10764 -
10765 - $ch = curl_init();
10766 - curl_setopt($ch, CURLOPT_URL, 'https://api.deepseek.com/v1/chat/completions');
10767 - curl_setopt($ch, CURLOPT_RETURNTRANSFER, false);
10768 - curl_setopt($ch, CURLOPT_POST, true);
10769 - curl_setopt($ch, CURLOPT_POSTFIELDS, $body);
10770 - curl_setopt($ch, CURLOPT_HTTPHEADER, array(
10771 - 'Content-Type: application/json',
10772 - 'Authorization: Bearer ' . $deepseek_api_key
10773 - ));
10774 - curl_setopt($ch, CURLOPT_SSL_VERIFYPEER, true);
10775 - curl_setopt($ch, CURLOPT_TIMEOUT, 60);
10776 -
10777 - curl_setopt($ch, CURLOPT_HEADERFUNCTION, function($ch, $header) use (&$captured_status_code) {
10778 - if ($captured_status_code === 0 && preg_match('#^HTTP/\S+\s+(\d+)\b#', $header, $m)) {
10779 - $captured_status_code = (int) $m[1];
10780 - }
10781 - return strlen($header);
10782 - });
10783 -
10784 - curl_setopt($ch, CURLOPT_WRITEFUNCTION, function($ch, $data) use (&$full_response, &$stream_started, &$buffer, &$captured_status_code, &$captured_body_pre_stream, $testing_data) {
10785 - if ($captured_status_code !== 0 && $captured_status_code !== 200) {
10786 - $captured_body_pre_stream .= $data;
10787 - return strlen($data);
10788 - }
10789 -
10790 - if (!$this->streaming_headers_sent) {
10791 - $this->setup_streaming_headers();
10792 - }
10793 -
10794 - if (!$stream_started && $testing_data !== null) {
10795 - echo "data: " . json_encode(['testing_data' => $testing_data]) . "\n\n";
10796 - flush();
10797 - $stream_started = true;
10798 - }
10799 -
10800 - $buffer .= $data;
10801 - $lines = explode("\n", $buffer);
10802 - $buffer = array_pop($lines);
10803 -
10804 - foreach ($lines as $line) {
10805 - if (trim($line) === '') {
10806 - continue;
10807 - }
10808 - if (strpos($line, 'data: ') !== 0) {
10809 - continue;
10810 - }
10811 -
10812 - $json_str = substr($line, 6);
10813 -
10814 - if (trim($json_str) === '[DONE]') {
10815 - echo "data: [DONE]\n\n";
10816 - flush();
10817 - continue;
10818 - }
10819 -
10820 - $json = json_decode(trim($json_str), true);
10821 - if ($json && isset($json['choices'][0]['delta']['content'])) {
10822 - $content = $json['choices'][0]['delta']['content'];
10823 - $full_response .= $content;
10824 - echo "data: " . json_encode(['content' => $content]) . "\n\n";
10825 - flush();
10826 - }
10827 - }
10828 -
10829 - return strlen($data);
10830 - });
10831 -
10832 - $response = curl_exec($ch);
10833 - $errno = curl_errno($ch);
10834 - $http_code = $captured_status_code !== 0 ? $captured_status_code : (int) curl_getinfo($ch, CURLINFO_HTTP_CODE);
10835 - curl_close($ch);
10836 -
10837 - if (!$errno && $http_code === 200) {
10838 - break;
10839 - }
10840 -
10841 - $is_transient = $this->mxchat_is_transient_provider_error_raw($http_code, $captured_body_pre_stream, 'openai', $errno);
10842 - $can_retry = !$this->streaming_headers_sent
10843 - && ($attempt + 1) < $max_attempts
10844 - && $is_transient;
10845 -
10846 - if (defined('WP_DEBUG') && WP_DEBUG) {
10847 - error_log(sprintf(
10848 - '[MxChat] deepseek_stream initial-connect failure (attempt=%d/%d, status=%d, errno=%d, transient=%s, %s).',
10849 - $attempt + 1, $max_attempts, $http_code, $errno,
10850 - $is_transient ? 'yes' : 'no',
10851 - $can_retry ? 'Retrying.' : 'Giving up.'
10852 - ));
10853 - }
10854 -
10855 - if (!$can_retry) {
10856 - break;
10857 - }
10858 - }
10859 -
10860 - if ($errno || $http_code !== 200) {
10861 - return $this->mxchat_stream_emit_fallback(
10862 - 'openai',
10863 - $this->mxchat_generate_response_deepseek($selected_model, $deepseek_api_key, $conversation_history, $relevant_content, $session_id),
10864 - $session_id,
10865 - $testing_data
10866 - );
10867 - }
10868 -
10869 - // Save the complete response to maintain chat persistence
10870 - if (!empty($full_response) && !empty($session_id)) {
10871 - // Prepare RAG context for streaming response
10872 - $rag_context_for_storage = null;
10873 - $has_rag_data = $this->last_similarity_analysis !== null && !empty($this->last_similarity_analysis['top_matches']);
10874 - $has_action_data = isset($this->last_action_analysis) && !empty($this->last_action_analysis);
10875 -
10876 - if ($has_rag_data || $has_action_data) {
10877 - $rag_context_for_storage = [];
10878 -
10879 - if ($has_rag_data) {
10880 - $rag_context_for_storage['top_matches'] = $this->last_similarity_analysis['top_matches'];
10881 - $rag_context_for_storage['approved_urls'] = $this->current_valid_urls ?? [];
10882 - $rag_context_for_storage['similarity_threshold'] = $this->last_similarity_analysis['threshold_used'] ?? 0.35;
10883 - $rag_context_for_storage['knowledge_base_type'] = $this->last_similarity_analysis['knowledge_base_type'] ?? 'WordPress Database';
10884 - $rag_context_for_storage['total_documents_checked'] = $this->last_similarity_analysis['total_checked'] ?? 0;
10885 - }
10886 -
10887 - if ($has_action_data) {
10888 - $rag_context_for_storage['action_analysis'] = $this->last_action_analysis;
10889 - }
10890 - }
10891 - $this->mxchat_save_chat_message($session_id, 'bot', $full_response, null, $rag_context_for_storage);
10892 - }
10893 -
10894 - return true; // Indicate streaming completed successfully
10895 -
10896 - } catch (Exception $e) {
10897 - return $this->mxchat_stream_emit_fallback(
10898 - 'openai',
10899 - $this->mxchat_generate_response_deepseek($selected_model, $deepseek_api_key, $conversation_history, $relevant_content),
10900 - $session_id,
10901 - $testing_data
10902 - );
10903 - }
10904 -}
10905 -
10906 -
10907 -/**
10908 - * Extract a human-readable error message from a decoded provider response body.
10909 - * Providers disagree on shape: OpenAI/Anthropic/Google nest it (error.message),
10910 - * xAI returns a plain string under 'error'. Mirrors mxchat-vision's shipped
10911 - * extract_provider_error(); deliberately hint-free in core (vision's too-small
10912 - * image hint is an upload concern that doesn't apply here).
10913 - *
10914 - * @param mixed $decoded_body Decoded JSON body (array), or whatever json_decode returned.
10915 - * @param string $fallback Message to return when no provider text is found.
10916 - * @return string
10917 - */
10918 -private function extract_provider_error($decoded_body, $fallback) {
10919 - $message = '';
10920 - if (isset($decoded_body['error']['message']) && is_string($decoded_body['error']['message']) && $decoded_body['error']['message'] !== '') {
10921 - $message = $decoded_body['error']['message'];
10922 - } elseif (isset($decoded_body['error']) && is_string($decoded_body['error']) && $decoded_body['error'] !== '') {
10923 - $message = $decoded_body['error'];
10924 - }
10925 -
10926 - if ($message === '') {
10927 - return $fallback;
10928 - }
10929 -
10930 - return $message;
10931 -}
10932 -
10933 -/**
10934 - * plan-4aa8e5: a provider 200 whose body parses to no text must never reach
10935 - * the widget as a silent empty bot bubble. Standard error shape for that
10936 - * case, preferring the body's own explanation — error.message first (the
10937 - * 950731 passthrough pattern), then the Responses API's
10938 - * incomplete_details.reason (e.g. "max_output_tokens") — before the generic
10939 - * retry message.
10940 - */
10941 -private function mxchat_empty_completion_error($decoded_body, $provider_label) {
10942 - $reason = '';
10943 - if (isset($decoded_body['error']['message']) && is_string($decoded_body['error']['message']) && $decoded_body['error']['message'] !== '') {
10944 - $reason = $decoded_body['error']['message'];
10945 - } elseif (isset($decoded_body['incomplete_details']['reason']) && is_string($decoded_body['incomplete_details']['reason']) && $decoded_body['incomplete_details']['reason'] !== '') {
10946 - $reason = sprintf(__('response incomplete: %s', 'mxchat'), $decoded_body['incomplete_details']['reason']);
10947 - }
10948 -
10949 - $message = ($reason !== '')
10950 - ? sprintf(esc_html__('%1$s returned an empty response (%2$s). Please try again.', 'mxchat'), $provider_label, esc_html($reason))
10951 - : sprintf(esc_html__('%s returned an empty response. Please try again.', 'mxchat'), $provider_label);
10952 -
10953 - return [
10954 - 'error' => $message,
10955 - 'error_code' => 'empty_completion',
10956 - 'provider' => strtolower($provider_label),
10957 - ];
10958 -}
10959 -
10960 -private function mxchat_generate_response_openrouter($selected_model, $openrouter_api_key, $conversation_history, $relevant_content, $session_id = '') {
10961 - try {
10962 - if (!is_array($conversation_history)) {
10963 - $conversation_history = array();
10964 - }
10965 -
10966 - $bot_id = $this->get_current_bot_id($session_id);
10967 - $system_prompt_instructions = $this->get_system_instructions($bot_id, $session_id);
10968 -
10969 - $formatted_conversation = array();
10970 -
10971 - $formatted_conversation[] = array(
10972 - 'role' => 'system',
10973 - 'content' => $system_prompt_instructions . " " . $relevant_content
10974 - );
10975 -
10976 - foreach ($conversation_history as $message) {
10977 - if (is_array($message) && isset($message['role']) && isset($message['content'])) {
10978 - $role = $message['role'];
10979 -
10980 - if ($role === 'bot' || $role === 'agent') {
10981 - $role = 'assistant';
10982 - }
10983 - if (!in_array($role, ['system', 'assistant', 'user'])) {
10984 - $role = 'user';
10985 - }
10986 -
10987 - $formatted_conversation[] = array(
10988 - 'role' => $role,
10989 - 'content' => $message['content']
10990 - );
10991 - }
10992 - }
10993 -
10994 - $body = json_encode([
10995 - 'model' => $selected_model,
10996 - 'messages' => $formatted_conversation,
10997 - 'temperature' => 1,
10998 - ]);
10999 -
11000 - $args = [
11001 - 'body' => $body,
11002 - 'headers' => [
11003 - 'Content-Type' => 'application/json',
11004 - 'Authorization' => 'Bearer ' . $openrouter_api_key,
11005 - 'HTTP-Referer' => home_url(),
11006 - 'X-Title' => get_bloginfo('name'),
11007 - ],
11008 - 'timeout' => 60,
11009 - 'redirection' => 5,
11010 - 'blocking' => true,
11011 - 'httpversion' => '1.0',
11012 - 'sslverify' => true,
11013 - ];
11014 -
11015 - $response = $this->mxchat_provider_call_with_retry('https://openrouter.ai/api/v1/chat/completions', $args, 'openai');
11016 -
11017 - if (is_wp_error($response)) {
11018 - $error_message = $response->get_error_message();
11019 - return [
11020 - 'error' => $this->mxchat_friendly_chat_error(0, $error_message, 'OpenRouter', $selected_model),
11021 - 'error_code' => 'openrouter_connection_error',
11022 - 'provider' => 'openrouter'
11023 - ];
11024 - }
11025 -
11026 - $status_code = wp_remote_retrieve_response_code($response);
11027 - if ($status_code !== 200) {
11028 - $response_body = wp_remote_retrieve_body($response);
11029 - $decoded_response = json_decode($response_body, true);
11030 -
11031 - $error_message = $this->extract_provider_error($decoded_response, 'HTTP Error ' . $status_code);
11032 -
11033 - return [
11034 - 'error' => esc_html__('OpenRouter API error: ', 'mxchat') . esc_html($error_message),
11035 - 'error_code' => 'openrouter_api_error',
11036 - 'provider' => 'openrouter',
11037 - 'status_code' => $status_code
11038 - ];
11039 - }
11040 -
11041 - $response_body = wp_remote_retrieve_body($response);
11042 - $decoded_response = json_decode($response_body, true);
11043 -
11044 - if (isset($decoded_response['choices'][0]['message']['content'])) {
11045 - $text = trim($decoded_response['choices'][0]['message']['content']);
11046 - if ($text !== '') {
11047 - return $text;
11048 - }
11049 - return $this->mxchat_empty_completion_error($decoded_response, 'OpenRouter');
11050 - } else {
11051 - return [
11052 - 'error' => esc_html__('Unexpected response format from OpenRouter.', 'mxchat'),
11053 - 'error_code' => 'openrouter_response_format_error',
11054 - 'provider' => 'openrouter'
11055 - ];
11056 - }
11057 - } catch (Exception $e) {
11058 - return [
11059 - 'error' => esc_html__('System error when processing OpenRouter request: ', 'mxchat') . esc_html($e->getMessage()),
11060 - 'error_code' => 'openrouter_exception',
11061 - 'provider' => 'openrouter'
11062 - ];
11063 - }
11064 -}
11065 -
11066 -/**
11067 - * Build a chat-bubble-safe message for a non-200 provider (chat) error.
11068 - *
11069 - * Visitors must NEVER see raw API internals (model names, key/billing/quota
11070 - * text). Admins (manage_options) get an actionable hint — and, for the common
11071 - * "model not available on this key" case, a direct pointer to change the model
11072 - * (the site owner can fix it in one click). Anthropic returns model-access as a
11073 - * 4xx with a message like "Claude Fable 5 is not available. Please use Opus 4.8."
11074 - *
11075 - * Provider-agnostic by design (reusable for the xai/gemini/deepseek branches),
11076 - * but Anthropic is the confirmed, reproduced case wired up here (plan 1d3b0f).
11077 - *
11078 - * @param int $http_code HTTP status from the provider.
11079 - * @param string $error_message Raw provider error.message (may be empty).
11080 - * @param string $provider_label Human provider name, e.g. 'Anthropic'.
11081 - * @param string $model The model id the failing request used. When a
11082 - * model-access failure is detected and this is
11083 - * non-empty, a persistent admin notice is armed
11084 - * (mxchat_show_model_access_notice) so the OWNER
11085 - * learns about it even when only anonymous
11086 - * visitors hit the broken bot (plan e46b8f).
11087 - * @return string Message safe to render as a chat bubble.
11088 - */
11089 -private function mxchat_friendly_chat_error($http_code, $error_message, $provider_label = '', $model = '') {
11090 - $raw = trim((string) $error_message);
11091 -
11092 - // Detect a model-access / availability problem the site owner can fix by
11093 - // choosing a different model. (Anthropic phrasing + the common API shapes.)
11094 - $low = strtolower($raw);
11095 - $is_model_access = (strpos($low, 'not available') !== false)
11096 - || (strpos($low, 'does not have access') !== false)
11097 - || (strpos($low, 'do not have access') !== false)
11098 - || (strpos($low, 'does not exist') !== false) // OpenAI: "model `x` does not exist or you do not have access"
11099 - || (strpos($low, 'model_not_found') !== false)
11100 - || (strpos($low, 'not_found_error') !== false)
11101 - || (strpos($low, 'model not found') !== false) // xAI
11102 - || (strpos($low, 'not found') !== false) // Gemini: "models/x is not found for API version ..."
11103 - || (strpos($low, 'permission_denied') !== false) // Gemini gated model
11104 - || (strpos($low, 'permission denied') !== false);
11105 -
11106 - // Arm the persistent admin notice (throttled: skip if the same model was
11107 - // flagged within the last hour — chat errors can fire per message).
11108 - if ($is_model_access && $model !== '') {
11109 - $existing = get_option('mxchat_model_access_notice');
11110 - $stale = !is_array($existing)
11111 - || !isset($existing['model'], $existing['time'])
11112 - || $existing['model'] !== $model
11113 - || (time() - (int) $existing['time']) > HOUR_IN_SECONDS;
11114 - if ($stale) {
11115 - update_option('mxchat_model_access_notice', array(
11116 - 'model' => (string) $model,
11117 - 'provider' => (string) $provider_label,
11118 - 'time' => time(),
11119 - ), false);
11120 - }
11121 - }
11122 -
11123 - if (current_user_can('manage_options')) {
11124 - if ($is_model_access) {
11125 - return $raw !== ''
11126 - ? sprintf(
11127 - /* translators: %s: raw provider error detail */
11128 - esc_html__('The selected AI model isn\'t available on your API key. Choose another model in MxChat → Settings. (Details: %s)', 'mxchat'),
11129 - $raw
11130 - )
11131 - : esc_html__('The selected AI model isn\'t available on your API key. Choose another model in MxChat → Settings.', 'mxchat');
11132 - }
11133 - return $raw !== ''
11134 - ? sprintf(
11135 - /* translators: 1: provider label, 2: raw provider error detail */
11136 - esc_html__('The AI provider (%1$s) returned an error: %2$s. Check your model and API key in MxChat → Settings.', 'mxchat'),
11137 - $provider_label !== '' ? $provider_label : esc_html__('AI', 'mxchat'),
11138 - $raw
11139 - )
11140 - : esc_html__('The AI provider returned an error. Check your model and API key in MxChat → Settings.', 'mxchat');
11141 - }
11142 -
11143 - // Visitors: friendly, generic, no internals leaked.
11144 - return esc_html__('Sorry, I\'m having trouble responding right now. Please try again in a moment.', 'mxchat');
11145 -}
11146 -
11147 -private function mxchat_generate_response_claude($selected_model, $claude_api_key, $conversation_history, $relevant_content, $session_id = '') {
11148 - // Anthropic retired claude-opus-4-20250514 / claude-sonnet-4-20250514 on 2026-06-15.
11149 - // Read-time rescue: remap a saved dead ID to the current equivalent before the API call.
11150 - if ($selected_model === 'claude-opus-4-20250514') { $selected_model = 'claude-opus-4-8'; }
11151 - elseif ($selected_model === 'claude-sonnet-4-20250514') { $selected_model = 'claude-sonnet-4-6'; }
11152 -
11153 - // Get bot ID from session or request
11154 - $bot_id = $this->get_current_bot_id($session_id);
11155 -
11156 - // Get system prompt instructions using centralized function
11157 - $system_prompt_instructions = $this->get_system_instructions($bot_id, $session_id);
11158 -
11159 - // Clean and validate conversation history
11160 - foreach ($conversation_history as &$message) {
11161 - // Convert bot and agent roles to assistant
11162 - if ($message['role'] === 'bot' || $message['role'] === 'agent') {
11163 - $message['role'] = 'assistant';
11164 - }
11165 -
11166 - // Remove unsupported roles - Claude only supports 'assistant' and 'user'
11167 - if (!in_array($message['role'], ['assistant', 'user'])) {
11168 - $message['role'] = 'user';
11169 - }
11170 -
11171 - // Ensure content field exists
11172 - if (!isset($message['content']) || empty($message['content'])) {
11173 - $message['content'] = '';
11174 - }
11175 -
11176 - // Remove any unsupported fields
11177 - $message = array_intersect_key($message, array_flip(['role', 'content']));
11178 - }
11179 -
11180 - // Add relevant content as the latest user message
11181 - $conversation_history[] = [
11182 - 'role' => 'user',
11183 - 'content' => $relevant_content
11184 - ];
11185 -
11186 - // Build request body
11187 - $payload = [
11188 - 'model' => $selected_model,
11189 - 'max_tokens' => 1000,
11190 - 'temperature' => 0.8,
11191 - 'messages' => $conversation_history,
11192 - 'system' => $system_prompt_instructions
11193 - ];
11194 - if ($this->mxchat_claude_omits_temperature($selected_model)) { unset($payload['temperature']); }
11195 - $body = json_encode($payload);
11196 -
11197 - // Set up API request
11198 - $args = [
11199 - 'body' => $body,
11200 - 'headers' => [
11201 - 'Content-Type' => 'application/json',
11202 - 'x-api-key' => $claude_api_key,
11203 - 'anthropic-version' => '2023-06-01'
11204 - ],
11205 - 'timeout' => 60,
11206 - 'redirection' => 5,
11207 - 'blocking' => true,
11208 - 'httpversion' => '1.0',
11209 - 'sslverify' => true,
11210 - ];
11211 -
11212 - // Make API request
11213 - $response = $this->mxchat_provider_call_with_retry('https://api.anthropic.com/v1/messages', $args, 'anthropic');
11214 -
11215 - // Check for WordPress errors
11216 - if (is_wp_error($response)) {
11217 - //error_log("Claude API request error: " . $response->get_error_message());
11218 - return "Sorry, there was an error connecting to the API.";
11219 - }
11220 -
11221 - // Check HTTP response code
11222 - $http_code = wp_remote_retrieve_response_code($response);
11223 - if ($http_code !== 200) {
11224 - $error_body = wp_remote_retrieve_body($response);
11225 - //error_log("Claude API HTTP error: " . $http_code . " - " . $error_body);
11226 -
11227 - // Try to extract error message from response
11228 - $error_data = json_decode($error_body, true);
11229 - $error_message = isset($error_data['error']['message']) ?
11230 - $error_data['error']['message'] :
11231 - "HTTP error " . $http_code;
11232 -
11233 - // Surface an admin-actionable message (and a model-change pointer for the
11234 - // model-access case) without leaking raw API internals to visitors. This
11235 - // is the single chokepoint for BOTH the non-streaming and streaming Claude
11236 - // paths (the stream's non-200 fallback re-enters this method). plan 1d3b0f.
11237 - return $this->mxchat_friendly_chat_error($http_code, $error_message, 'Anthropic', $selected_model);
11238 - }
11239 -
11240 - // Parse response
11241 - $response_body = json_decode(wp_remote_retrieve_body($response), true);
11242 -
11243 - // Check for JSON decode errors
11244 - if (json_last_error() !== JSON_ERROR_NONE) {
11245 - //error_log("Claude API JSON decode error: " . json_last_error_msg());
11246 - return "Sorry, there was an error processing the API response.";
11247 - }
11248 -
11249 - // Extract and validate response content. claude-fable-5 prepends a
11250 - // thinking block to content even with no thinking param — take the first
11251 - // TEXT block rather than content[0].
11252 - if (isset($response_body['content']) && is_array($response_body['content'])) {
11253 - foreach ($response_body['content'] as $block) {
11254 - // plan-4aa8e5: skip empty text blocks — a 200 whose only text
11255 - // block trims to '' must not render as a silent empty bubble.
11256 - if (isset($block['type'], $block['text']) && $block['type'] === 'text' && trim($block['text']) !== '') {
11257 - return trim($block['text']);
11258 - }
11259 - }
11260 - return $this->mxchat_empty_completion_error($response_body, 'Claude');
11261 - }
11262 -
11263 - // Log unexpected response format
11264 - //error_log("Claude API unexpected response format: " . print_r($response_body, true));
11265 - return "Sorry, I received an unexpected response format from the API.";
11266 -}
11267 -private function mxchat_generate_response_openai($selected_model, $api_key, $conversation_history, $relevant_content, $session_id = '') {
11268 - // OpenAI retires gpt-5.1-chat-latest / gpt-5.3-chat-latest on 2026-08-10
11269 - // (replacement gpt-5.6-sol). Read-time rescue for saved / bot-level ids
11270 - // that missed mxchat_migrate_deprecated_models() (plan e46b8f).
11271 - if ($selected_model === 'gpt-5.1-chat-latest' || $selected_model === 'gpt-5.3-chat-latest') { $selected_model = 'gpt-5.6-sol'; }
11272 - try {
11273 - // Ensure conversation_history is an array
11274 - if (!is_array($conversation_history)) {
11275 - $conversation_history = array();
11276 - }
11277 -
11278 - // Get bot ID from session or request. plan eb9c38: resolve the real bot
11279 - // from the session (was hardcoded '' → always default bot on multi-bot
11280 - // installs) and fix the undefined $session_id that fed get_system_instructions.
11281 - $bot_id = $this->get_current_bot_id($session_id);
11282 -
11283 - // Get system prompt instructions using centralized function
11284 - $system_prompt_instructions = $this->get_system_instructions($bot_id, $session_id);
11285 -
11286 - // Create a new array for the formatted conversation
11287 - $formatted_conversation = array();
11288 -
11289 - // Add system message first
11290 - $formatted_conversation[] = array(
11291 - 'role' => 'system',
11292 - 'content' => $system_prompt_instructions . " " . $relevant_content
11293 - );
11294 -
11295 - // Add the rest of the conversation history
11296 - foreach ($conversation_history as $message) {
11297 - if (is_array($message) && isset($message['role']) && isset($message['content'])) {
11298 - $role = $message['role'];
11299 -
11300 - // Convert roles to supported format
11301 - if ($role === 'bot' || $role === 'agent') {
11302 - $role = 'assistant';
11303 - }
11304 - if (!in_array($role, ['system', 'assistant', 'user', 'function', 'tool'])) {
11305 - $role = 'user';
11306 - }
11307 -
11308 - $formatted_conversation[] = array(
11309 - 'role' => $role,
11310 - 'content' => $message['content']
11311 - );
11312 - }
11313 - }
11314 -
11315 - // Build request body with optimal settings for fast responses
11316 - $request_body = [
11317 - 'model' => $selected_model,
11318 - 'messages' => $formatted_conversation,
11319 - 'temperature' => 1,
11320 - 'stream' => false
11321 - ];
11322 -
11323 - // reasoning_effort — sourced from the core model catalog (plan-dcb71c);
11324 - // frozen inline ladder lives in mxchat_reasoning_effort_fallback().
11325 - $effort = $this->mxchat_reasoning_effort_for($selected_model, 'chat');
11326 - if ($effort !== null) {
11327 - $request_body['reasoning_effort'] = $effort;
11328 - }
11329 -
11330 - $body = json_encode($request_body);
11331 -
11332 - $args = [
11333 - 'body' => $body,
11334 - 'headers' => [
11335 - 'Content-Type' => 'application/json',
11336 - 'Authorization' => 'Bearer ' . $api_key,
11337 - ],
11338 - 'timeout' => 60,
11339 - 'redirection' => 5,
11340 - 'blocking' => true,
11341 - 'httpversion' => '1.0',
11342 - 'sslverify' => true,
11343 - ];
11344 -
11345 - $response = $this->mxchat_provider_call_with_retry('https://api.openai.com/v1/chat/completions', $args, 'openai');
11346 -
11347 - if (is_wp_error($response)) {
11348 - $error_message = $response->get_error_message();
11349 - return [
11350 - 'error' => $this->mxchat_friendly_chat_error(0, $error_message, 'OpenAI', $selected_model),
11351 - 'error_code' => 'openai_connection_error',
11352 - 'provider' => 'openai'
11353 - ];
11354 - }
11355 -
11356 - $status_code = wp_remote_retrieve_response_code($response);
11357 - if ($status_code !== 200) {
11358 - $response_body = wp_remote_retrieve_body($response);
11359 - $decoded_response = json_decode($response_body, true);
11360 -
11361 - $error_message = isset($decoded_response['error']['message'])
11362 - ? $decoded_response['error']['message']
11363 - : 'HTTP Error ' . $status_code;
11364 -
11365 - $error_type = isset($decoded_response['error']['type'])
11366 - ? $decoded_response['error']['type']
11367 - : 'unknown';
11368 -
11369 - // Handle specific error types
11370 - switch ($error_type) {
11371 - case 'invalid_request_error':
11372 - if (strpos($error_message, 'API key') !== false) {
11373 - return [
11374 - 'error' => esc_html__('Invalid OpenAI API key. Please check your API key configuration.', 'mxchat'),
11375 - 'error_code' => 'openai_invalid_api_key',
11376 - 'provider' => 'openai'
11377 - ];
11378 - }
11379 - break;
11380 -
11381 - case 'authentication_error':
11382 - return [
11383 - 'error' => esc_html__('Authentication failed with OpenAI. Please check your API key.', 'mxchat'),
11384 - 'error_code' => 'openai_auth_error',
11385 - 'provider' => 'openai'
11386 - ];
11387 -
11388 - case 'rate_limit_exceeded':
11389 - return [
11390 - 'error' => esc_html__('OpenAI rate limit exceeded. Please try again later.', 'mxchat'),
11391 - 'error_code' => 'openai_rate_limit',
11392 - 'provider' => 'openai'
11393 - ];
11394 -
11395 - case 'quota_exceeded':
11396 - return [
11397 - 'error' => esc_html__('OpenAI API quota exceeded. Please check your billing details.', 'mxchat'),
11398 - 'error_code' => 'openai_quota_exceeded',
11399 - 'provider' => 'openai'
11400 - ];
11401 - }
11402 -
11403 - // Generic error fallback only — the typed cases above already produce
11404 - // clean messages. Route the raw-tail generic case through the leak-safe
11405 - // helper so visitors never see provider internals. plan 5da59a.
11406 - return [
11407 - 'error' => $this->mxchat_friendly_chat_error($status_code, $error_message, 'OpenAI', $selected_model),
11408 - 'error_code' => 'openai_api_error',
11409 - 'provider' => 'openai',
11410 - 'status_code' => $status_code
11411 - ];
11412 - }
11413 -
11414 - $response_body = wp_remote_retrieve_body($response);
11415 - $decoded_response = json_decode($response_body, true);
11416 -
11417 - if (isset($decoded_response['choices'][0]['message']['content'])) {
11418 - $text = trim($decoded_response['choices'][0]['message']['content']);
11419 - if ($text !== '') {
11420 - return $text;
11421 - }
11422 - return $this->mxchat_empty_completion_error($decoded_response, 'OpenAI');
11423 - } else {
11424 - return [
11425 - 'error' => esc_html__('Unexpected response format from OpenAI.', 'mxchat'),
11426 - 'error_code' => 'openai_response_format_error',
11427 - 'provider' => 'openai'
11428 - ];
11429 - }
11430 - } catch (Exception $e) {
11431 - return [
11432 - 'error' => esc_html__('System error when processing OpenAI request: ', 'mxchat') . esc_html($e->getMessage()),
11433 - 'error_code' => 'openai_exception',
11434 - 'provider' => 'openai'
11435 - ];
11436 - }
11437 -}
11438 -
11439 -private function mxchat_generate_response_xai($selected_model, $xai_api_key, $conversation_history, $relevant_content, $session_id = '') {
11440 - try {
11441 - // Get bot ID from session or request
11442 - $bot_id = $this->get_current_bot_id($session_id);
11443 -
11444 - // Get system prompt instructions using centralized function
11445 - $system_prompt_instructions = $this->get_system_instructions($bot_id, $session_id);
11446 -
11447 - // Add system prompt to relevant content
11448 372 $content_with_instructions = $system_prompt_instructions . " " . $relevant_content;
11449 373
11450 - // Prepend system instructions to the conversation history
11451 374 array_unshift($conversation_history, [
11452 375 'role' => 'system',
11453 376 'content' => "Here are your instructions: " . $content_with_instructions
11454 377 ]);
11455 378
11456 - // Ensure consistency: Replace 'bot' and 'agent' roles with supported values
11457 379 foreach ($conversation_history as &$message) {
11458 380 if ($message['role'] === 'bot') {
11459 381 $message['role'] = 'assistant';
11460 - } elseif ($message['role'] === 'agent') {
11461 - // Tag the message as coming from a live agent
11462 - $message['role'] = 'assistant';
11463 - if (!isset($message['metadata'])) {
11464 - $message['metadata'] = ['source' => 'live_agent'];
11465 - }
11466 382 }
383 + }
11467 384
11468 - // Ensure all roles are valid
11469 - if (!in_array($message['role'], ['system', 'assistant', 'user', 'function', 'tool'])) {
11470 - $message['role'] = 'user'; // Default to 'user'
11471 - }
11472 - }
385 + $api_url = 'https://api.openai.com/v1/chat/completions';
11473 386
11474 - // Build the request body
11475 387 $body = json_encode([
11476 - 'model' => $selected_model,
388 + 'model' => isset($this->options['model']) ? $this->options['model'] : 'gpt-3.5',
11477 389 'messages' => $conversation_history,
11478 - 'temperature' => 0.8,
11479 - 'stream' => false
11480 390 ]);
11481 391
11482 - // Set up the API request
11483 392 $args = [
11484 393 'body' => $body,
11485 394 'headers' => [
11486 395 'Content-Type' => 'application/json',
11487 - 'Authorization' => 'Bearer ' . $xai_api_key,
396 + 'Authorization' => 'Bearer ' . $api_key,
11488 397 ],
11489 398 'timeout' => 60,
11490 399 'redirection' => 5,
11491 400 'blocking' => true,
@@ -11492,653 +401,28 @@
11492 401 'httpversion' => '1.0',
11493 402 'sslverify' => true,
11494 403 ];
11495 404
11496 - // Make the API request
11497 - $response = $this->mxchat_provider_call_with_retry('https://api.x.ai/v1/chat/completions', $args, 'xai');
405 + $response = wp_remote_post($api_url, $args);
11498 406
11499 - // Process the response
11500 407 if (is_wp_error($response)) {
11501 - $error_message = $response->get_error_message();
11502 - //error_log('X.AI API Error: ' . $error_message);
11503 - return [
11504 - 'error' => $this->mxchat_friendly_chat_error(0, $error_message, 'X.AI', $selected_model),
11505 - 'error_code' => 'xai_connection_error',
11506 - 'provider' => 'xai'
11507 - ];
408 + return "Sorry, there was an error processing your request.";
11508 409 }
11509 410
11510 - $status_code = wp_remote_retrieve_response_code($response);
11511 - if ($status_code !== 200) {
11512 - $response_body = wp_remote_retrieve_body($response);
11513 - $decoded_response = json_decode($response_body, true);
411 + $response_body = json_decode(wp_remote_retrieve_body($response), true);
11514 412
11515 - // Log the full response for debugging
11516 - //error_log('X.AI Error Response: ' . print_r($decoded_response, true));
11517 -
11518 - // Extract error message from X.AI's specific format
11519 - $error_message = '';
11520 -
11521 - // Check for direct error string (as seen in your logs)
11522 - if (isset($decoded_response['error']) && is_string($decoded_response['error'])) {
11523 - $error_message = $decoded_response['error'];
413 + if (isset($response_body['choices'][0]['message']['content'])) {
414 + if (isset($response_body['usage'])) {
415 + $prompt_tokens = $response_body['usage']['prompt_tokens'];
416 + $total_tokens = $response_body['usage']['total_tokens'];
11524 417 }
11525 - // Check for nested error object (OpenAI style)
11526 - elseif (isset($decoded_response['error']['message'])) {
11527 - $error_message = $decoded_response['error']['message'];
11528 - }
11529 - // Check for top-level message
11530 - elseif (isset($decoded_response['message'])) {
11531 - $error_message = $decoded_response['message'];
11532 - }
11533 - // Fallback
11534 - else {
11535 - $error_message = 'HTTP Error ' . $status_code;
11536 - }
11537 -
11538 - //error_log('X.AI API HTTP Error: ' . $status_code . ' - ' . $error_message);
11539 -
11540 - // Check for API key errors using string matching
11541 - if (stripos($error_message, 'api key') !== false ||
11542 - stripos($error_message, 'incorrect api key') !== false ||
11543 - stripos($error_message, 'invalid api key') !== false) {
11544 - return [
11545 - 'error' => esc_html__('Invalid X.AI API key. Please check your API key configuration.', 'mxchat'),
11546 - 'error_code' => 'xai_invalid_api_key',
11547 - 'provider' => 'xai'
11548 - ];
11549 - }
11550 -
11551 - // Authentication errors
11552 - if ($status_code === 401 || $status_code === 403 ||
11553 - stripos($error_message, 'auth') !== false) {
11554 - return [
11555 - 'error' => esc_html__('Authentication failed with X.AI. Please check your API key.', 'mxchat') . ' ' . esc_html($error_message),
11556 - 'error_code' => 'xai_auth_error',
11557 - 'provider' => 'xai'
11558 - ];
11559 - }
11560 -
11561 - // Model errors — keep the canned category text as a prefix, but carry the
11562 - // provider's extracted reason (e.g. "Model not found: <id>") so the owner
11563 - // sees the specific model/reason instead of only the generic category.
11564 - if (stripos($error_message, 'model') !== false) {
11565 - return [
11566 - 'error' => esc_html__('Invalid model specified for X.AI. Please check your model configuration.', 'mxchat') . ' ' . esc_html($error_message),
11567 - 'error_code' => 'xai_invalid_model',
11568 - 'provider' => 'xai'
11569 - ];
11570 - }
11571 -
11572 - // Rate limit errors
11573 - if ($status_code === 429 ||
11574 - stripos($error_message, 'rate') !== false ||
11575 - stripos($error_message, 'limit') !== false) {
11576 - return [
11577 - 'error' => esc_html__('X.AI rate limit exceeded. Please try again later.', 'mxchat'),
11578 - 'error_code' => 'xai_rate_limit',
11579 - 'provider' => 'xai'
11580 - ];
11581 - }
11582 -
11583 - // Quota errors
11584 - if (stripos($error_message, 'quota') !== false ||
11585 - stripos($error_message, 'billing') !== false) {
11586 - return [
11587 - 'error' => esc_html__('X.AI API quota exceeded. Please check your billing details.', 'mxchat'),
11588 - 'error_code' => 'xai_quota_exceeded',
11589 - 'provider' => 'xai'
11590 - ];
11591 - }
11592 -
11593 - // Server errors
11594 - if ($status_code >= 500) {
11595 - return [
11596 - 'error' => esc_html__('X.AI service is currently unavailable. Please try again later.', 'mxchat'),
11597 - 'error_code' => 'xai_service_unavailable',
11598 - 'provider' => 'xai'
11599 - ];
11600 - }
11601 -
11602 - // Generic error fallback. Route the user-facing text through the
11603 - // leak-safe helper (admins get an actionable hint, visitors a generic
11604 - // fallback) instead of echoing raw provider internals. Preserve the
11605 - // structured contract (error_code/provider/status_code) for logging. plan 5da59a.
11606 - return [
11607 - 'error' => $this->mxchat_friendly_chat_error($status_code, $error_message, 'xAI', $selected_model),
11608 - 'error_code' => 'xai_api_error',
11609 - 'provider' => 'xai',
11610 - 'status_code' => $status_code
11611 - ];
11612 - }
11613 -
11614 - $response_body = wp_remote_retrieve_body($response);
11615 - $decoded_response = json_decode($response_body, true);
11616 -
11617 - if (isset($decoded_response['choices'][0]['message']['content'])) {
11618 - $text = trim($decoded_response['choices'][0]['message']['content']);
11619 - if ($text !== '') {
11620 - return $text;
11621 - }
11622 - return $this->mxchat_empty_completion_error($decoded_response, 'X.AI');
418 + return trim($response_body['choices'][0]['message']['content']);
11623 419 } else {
11624 - //error_log('X.AI API Response Format Error: ' . print_r($decoded_response, true));
11625 - return [
11626 - 'error' => esc_html__('Unexpected response format from X.AI.', 'mxchat'),
11627 - 'error_code' => 'xai_response_format_error',
11628 - 'provider' => 'xai'
11629 - ];
420 + return "Sorry, I couldn't process that request.";
11630 421 }
11631 -} catch (Exception $e) {
11632 - //error_log('X.AI Exception: ' . $e->getMessage());
11633 - return [
11634 - 'error' => esc_html__('System error when processing X.AI request: ', 'mxchat') . esc_html($e->getMessage()),
11635 - 'error_code' => 'xai_exception',
11636 - 'provider' => 'xai'
11637 - ];
11638 422 }
11639 423
11640 424
11641 -}
11642 -private function mxchat_generate_response_deepseek($selected_model, $deepseek_api_key, $conversation_history, $relevant_content, $session_id = '') {
11643 - try {
11644 - // Ensure conversation_history is an array
11645 - if (!is_array($conversation_history)) {
11646 - $conversation_history = array();
11647 - }
11648 -
11649 - // Get bot ID from session or request
11650 - $bot_id = $this->get_current_bot_id($session_id);
11651 -
11652 - // Get system prompt instructions using centralized function
11653 - $system_prompt_instructions = $this->get_system_instructions($bot_id, $session_id);
11654 -
11655 - // Create a new array for the formatted conversation
11656 - $formatted_conversation = array();
11657 -
11658 - // Add system message first
11659 - $formatted_conversation[] = array(
11660 - 'role' => 'system',
11661 - 'content' => $system_prompt_instructions . " " . $relevant_content
11662 - );
11663 -
11664 - // Add the rest of the conversation history
11665 - foreach ($conversation_history as $message) {
11666 - if (is_array($message) && isset($message['role']) && isset($message['content'])) {
11667 - $role = $message['role'];
11668 -
11669 - // Convert roles to supported format
11670 - if ($role === 'bot' || $role === 'agent') {
11671 - $role = 'assistant';
11672 - }
11673 - if (!in_array($role, ['system', 'assistant', 'user', 'function', 'tool'])) {
11674 - $role = 'user';
11675 - }
11676 -
11677 - $formatted_conversation[] = array(
11678 - 'role' => $role,
11679 - 'content' => $message['content']
11680 - );
11681 - }
11682 - }
11683 -
11684 - $body = json_encode([
11685 - 'model' => $selected_model,
11686 - 'messages' => $formatted_conversation,
11687 - 'temperature' => 0.8,
11688 - 'stream' => false,
11689 - // DeepSeek V4 defaults to thinking mode ON (temperature ignored,
11690 - // slow reasoning-first responses); the widget wants the legacy
11691 - // deepseek-chat semantics = non-thinking.
11692 - 'thinking' => ['type' => 'disabled']
11693 - ]);
11694 -
11695 - $args = [
11696 - 'body' => $body,
11697 - 'headers' => [
11698 - 'Content-Type' => 'application/json',
11699 - 'Authorization' => 'Bearer ' . $deepseek_api_key,
11700 - ],
11701 - 'timeout' => 60,
11702 - 'redirection' => 5,
11703 - 'blocking' => true,
11704 - 'httpversion' => '1.0',
11705 - 'sslverify' => true,
11706 - ];
11707 -
11708 - $response = $this->mxchat_provider_call_with_retry('https://api.deepseek.com/v1/chat/completions', $args, 'openai');
11709 -
11710 - if (is_wp_error($response)) {
11711 - $error_message = $response->get_error_message();
11712 - //error_log('DeepSeek API Error: ' . $error_message);
11713 - return [
11714 - 'error' => $this->mxchat_friendly_chat_error(0, $error_message, 'DeepSeek', $selected_model),
11715 - 'error_code' => 'deepseek_connection_error',
11716 - 'provider' => 'deepseek'
11717 - ];
11718 - }
11719 -
11720 - $status_code = wp_remote_retrieve_response_code($response);
11721 - if ($status_code !== 200) {
11722 - $response_body = wp_remote_retrieve_body($response);
11723 - $decoded_response = json_decode($response_body, true);
11724 -
11725 - $error_message = isset($decoded_response['error']['message'])
11726 - ? $decoded_response['error']['message']
11727 - : 'HTTP Error ' . $status_code;
11728 -
11729 - $error_type = isset($decoded_response['error']['type'])
11730 - ? $decoded_response['error']['type']
11731 - : 'unknown';
11732 -
11733 - //error_log('DeepSeek API HTTP Error: ' . $status_code . ' - ' . $error_message);
11734 -
11735 - // Handle specific error types
11736 - switch ($status_code) {
11737 - case 401:
11738 - return [
11739 - 'error' => esc_html__('Authentication failed with DeepSeek. Please check your API key.', 'mxchat'),
11740 - 'error_code' => 'deepseek_auth_error',
11741 - 'provider' => 'deepseek'
11742 - ];
11743 -
11744 - case 400:
11745 - if (strpos($error_message, 'API key') !== false) {
11746 - return [
11747 - 'error' => esc_html__('Invalid DeepSeek API key. Please check your API key configuration.', 'mxchat'),
11748 - 'error_code' => 'deepseek_invalid_api_key',
11749 - 'provider' => 'deepseek'
11750 - ];
11751 - }
11752 - break;
11753 -
11754 - case 429:
11755 - if (strpos($error_message, 'quota') !== false) {
11756 - return [
11757 - 'error' => esc_html__('DeepSeek API quota exceeded. Please check your billing details.', 'mxchat'),
11758 - 'error_code' => 'deepseek_quota_exceeded',
11759 - 'provider' => 'deepseek'
11760 - ];
11761 - } else {
11762 - return [
11763 - 'error' => esc_html__('DeepSeek rate limit exceeded. Please try again later.', 'mxchat'),
11764 - 'error_code' => 'deepseek_rate_limit',
11765 - 'provider' => 'deepseek'
11766 - ];
11767 - }
11768 -
11769 - case 500:
11770 - case 502:
11771 - case 503:
11772 - case 504:
11773 - return [
11774 - 'error' => esc_html__('DeepSeek service is currently unavailable. Please try again later.', 'mxchat'),
11775 - 'error_code' => 'deepseek_service_unavailable',
11776 - 'provider' => 'deepseek'
11777 - ];
11778 - }
11779 -
11780 - // Generic error fallback — leak-safe helper (see plan 5da59a / 1d3b0f).
11781 - return [
11782 - 'error' => $this->mxchat_friendly_chat_error($status_code, $error_message, 'DeepSeek', $selected_model),
11783 - 'error_code' => 'deepseek_api_error',
11784 - 'provider' => 'deepseek',
11785 - 'status_code' => $status_code
11786 - ];
11787 - }
11788 -
11789 - $response_body = wp_remote_retrieve_body($response);
11790 - $decoded_response = json_decode($response_body, true);
11791 -
11792 - if (isset($decoded_response['choices'][0]['message']['content'])) {
11793 - $text = trim($decoded_response['choices'][0]['message']['content']);
11794 - if ($text !== '') {
11795 - return $text;
11796 - }
11797 - return $this->mxchat_empty_completion_error($decoded_response, 'DeepSeek');
11798 - } else {
11799 - //error_log('DeepSeek API Response Format Error: ' . print_r($decoded_response, true));
11800 - return [
11801 - 'error' => esc_html__('Unexpected response format from DeepSeek.', 'mxchat'),
11802 - 'error_code' => 'deepseek_response_format_error',
11803 - 'provider' => 'deepseek'
11804 - ];
11805 - }
11806 - } catch (Exception $e) {
11807 - //error_log('DeepSeek Exception: ' . $e->getMessage());
11808 - return [
11809 - 'error' => esc_html__('System error when processing DeepSeek request: ', 'mxchat') . esc_html($e->getMessage()),
11810 - 'error_code' => 'deepseek_exception',
11811 - 'provider' => 'deepseek'
11812 - ];
11813 - }
11814 -}
11815 -private function mxchat_generate_response_gemini($selected_model, $gemini_api_key, $conversation_history, $relevant_content, $session_id = '') {
11816 - // Read-time remap: gemini-3-pro-preview was shut down March 9, 2026.
11817 - // Auto-rescue existing installs whose saved model is the dead ID.
11818 - if ($selected_model === 'gemini-3-pro-preview') {
11819 - $selected_model = 'gemini-3.1-pro-preview';
11820 - }
11821 - // Get bot ID from session or request
11822 - $bot_id = $this->get_current_bot_id($session_id);
11823 -
11824 - // Get system prompt instructions using centralized function
11825 - $system_prompt_instructions = $this->get_system_instructions($bot_id, $session_id);
11826 -
11827 - // Add system prompt to relevant content
11828 - $content_with_instructions = $system_prompt_instructions . " " . $relevant_content;
11829 -
11830 - // Format messages for Gemini API
11831 - $formatted_messages = [];
11832 -
11833 - // Add system message as the first user message with role prefix
11834 - // Note: Gemini doesn't have a dedicated system role, so we use a prefixed user message
11835 - $formatted_messages[] = [
11836 - 'role' => 'user',
11837 - 'parts' => [
11838 - ['text' => "[System Instructions] " . $content_with_instructions]
11839 - ]
11840 - ];
11841 -
11842 - // Add model response to acknowledge system instructions
11843 - $formatted_messages[] = [
11844 - 'role' => 'model',
11845 - 'parts' => [
11846 - ['text' => "I understand and will follow these instructions."]
11847 - ]
11848 - ];
11849 -
11850 - // Process the rest of the conversation history
11851 - $current_role = null;
11852 - $current_parts = [];
11853 -
11854 - foreach ($conversation_history as $message) {
11855 - // Skip the first system message as we already handled it
11856 - if ($message['role'] === 'system') {
11857 - continue;
11858 - }
11859 -
11860 - // Map roles to Gemini format
11861 - $gemini_role = '';
11862 - if ($message['role'] === 'user') {
11863 - $gemini_role = 'user';
11864 - } else if (in_array($message['role'], ['assistant', 'bot', 'agent'])) {
11865 - $gemini_role = 'model';
11866 - } else {
11867 - // Skip unsupported roles
11868 - continue;
11869 - }
11870 -
11871 - // If we have a new role, add the previous message
11872 - if ($current_role !== null && $current_role !== $gemini_role && !empty($current_parts)) {
11873 - $formatted_messages[] = [
11874 - 'role' => $current_role,
11875 - 'parts' => $current_parts
11876 - ];
11877 - $current_parts = [];
11878 - }
11879 -
11880 - // Set current role and add text to parts
11881 - $current_role = $gemini_role;
11882 - $current_parts[] = ['text' => $message['content']];
11883 - }
11884 -
11885 - // Add the last message if there's content
11886 - if ($current_role !== null && !empty($current_parts)) {
11887 - $formatted_messages[] = [
11888 - 'role' => $current_role,
11889 - 'parts' => $current_parts
11890 - ];
11891 - }
11892 -
11893 - // Built-in Web Search grounding for Gemini (plan 46b9ea).
11894 - // The enable_web_search toggle historically routed ONLY to OpenAI's web_search
11895 - // tool; for a Gemini chat model it was a silent no-op. Gemini grounds natively
11896 - // (and free) via the Google Search tool, so when the toggle is on we attach it
11897 - // here on the PLAIN dispatch path. The function-calling loop (mxchat_fc_loop_gemini)
11898 - // is a SEPARATE path reached only when AI Tools are active, so grounding here
11899 - // never double-fires with function calling.
11900 - $web_search_enabled = isset($this->options['enable_web_search']) && $this->options['enable_web_search'] === 'on';
11901 - // Gemini ids that do NOT support Google Search grounding (none today — every
11902 - // shipped chat model is 2.x/3.x and grounds natively). Kept as the explicit
11903 - // opt-out list mirroring the OpenAI $unsupported_web_search_models pattern.
11904 - $gemini_unsupported_grounding = array();
11905 - $grounding_active = $web_search_enabled && !in_array($selected_model, $gemini_unsupported_grounding, true);
11906 -
11907 - // Build the request body
11908 - $request_payload = [
11909 - 'contents' => $formatted_messages,
11910 - 'generationConfig' => [
11911 - 'temperature' => 0.7,
11912 - 'topP' => 0.95,
11913 - 'topK' => 40,
11914 - 'maxOutputTokens' => 8192,
11915 - ],
11916 - 'safetySettings' => [
11917 - [
11918 - 'category' => 'HARM_CATEGORY_HARASSMENT',
11919 - 'threshold' => 'BLOCK_MEDIUM_AND_ABOVE'
11920 - ],
11921 - [
11922 - 'category' => 'HARM_CATEGORY_HATE_SPEECH',
11923 - 'threshold' => 'BLOCK_MEDIUM_AND_ABOVE'
11924 - ],
11925 - [
11926 - 'category' => 'HARM_CATEGORY_SEXUALLY_EXPLICIT',
11927 - 'threshold' => 'BLOCK_MEDIUM_AND_ABOVE'
11928 - ],
11929 - [
11930 - 'category' => 'HARM_CATEGORY_DANGEROUS_CONTENT',
11931 - 'threshold' => 'BLOCK_MEDIUM_AND_ABOVE'
11932 - ]
11933 - ]
11934 - ];
11935 -
11936 - if ($grounding_active) {
11937 - // Gemini 1.5 used the older google_search_retrieval shape; 2.0+ uses the
11938 - // bare google_search tool. Branch by model family so a future 1.5 id still
11939 - // grounds (no 1.5 ships today, so this resolves to google_search). The empty
11940 - // tool config must serialize as a JSON object {}, not an array [].
11941 - if (strpos($selected_model, 'gemini-1.5') !== false) {
11942 - $request_payload['tools'] = [ ['google_search_retrieval' => new \stdClass()] ];
11943 - } else {
11944 - $request_payload['tools'] = [ ['google_search' => new \stdClass()] ];
11945 - }
11946 - }
11947 -
11948 - $body = json_encode($request_payload);
11949 -
11950 - // Prepare the API endpoint
11951 - // Use v1beta for preview models (Gemini 3, experimental), v1 for stable models.
11952 - // Grounding (the google_search tool) is a v1beta feature, so force v1beta whenever
11953 - // it's active — otherwise a stable model on v1 would silently drop the tool.
11954 - $api_version = ($grounding_active || strpos($selected_model, 'preview') !== false || strpos($selected_model, 'exp') !== false) ? 'v1beta' : 'v1';
11955 - $api_endpoint = 'https://generativelanguage.googleapis.com/' . $api_version . '/models/' . $selected_model . ':generateContent?key=' . $gemini_api_key;
11956 -
11957 - // Set up the API request
11958 - $args = [
11959 - 'body' => $body,
11960 - 'headers' => [
11961 - 'Content-Type' => 'application/json',
11962 - ],
11963 - 'timeout' => 60,
11964 - 'redirection' => 5,
11965 - 'blocking' => true,
11966 - 'httpversion' => '1.0',
11967 - 'sslverify' => true,
11968 - ];
11969 -
11970 - // Make the API request
11971 - $response = $this->mxchat_provider_call_with_retry($api_endpoint, $args, 'gemini');
11972 -
11973 - // Process the response
11974 - if (is_wp_error($response)) {
11975 - // plan b13282: route the transport-error string through the leak-safe helper
11976 - // (admin-actionable, generic for visitors) instead of echoing the raw WP HTTP
11977 - // error. http_code 0 = no HTTP response, so the helper uses the generic branch.
11978 - return $this->mxchat_friendly_chat_error(0, $response->get_error_message(), 'Gemini', $selected_model);
11979 - }
11980 -
11981 - $response_body = json_decode(wp_remote_retrieve_body($response), true);
11982 -
11983 - // Handle potential errors in the response. Gemini surfaces errors as a
11984 - // 200/non-200 body with an `error` envelope; route the user-facing text
11985 - // through the leak-safe helper (admin-actionable, no visitor leak) rather
11986 - // than echoing the raw provider message. plan 5da59a.
11987 - if (isset($response_body['error'])) {
11988 - //error_log('Gemini API Error: ' . json_encode($response_body['error']));
11989 - $gemini_error_message = isset($response_body['error']['message'])
11990 - ? $response_body['error']['message']
11991 - : 'Unknown error';
11992 - $gemini_http_code = wp_remote_retrieve_response_code($response);
11993 - return $this->mxchat_friendly_chat_error($gemini_http_code, $gemini_error_message, 'Gemini', $selected_model);
11994 - }
11995 -
11996 - // Extract the response text
11997 - if (isset($response_body['candidates'][0]['content']['parts'][0]['text'])) {
11998 - $text = trim($response_body['candidates'][0]['content']['parts'][0]['text']);
11999 - if ($text !== '') {
12000 - return $text;
12001 - }
12002 - return $this->mxchat_empty_completion_error($response_body, 'Gemini');
12003 - } else {
12004 - //error_log('Unexpected Gemini API response format: ' . json_encode($response_body));
12005 - return "Sorry, I couldn't process that request. The response format was unexpected.";
12006 - }
12007 -}
12008 -
12009 -
12010 -public function test_streaming_request() {
12011 - $options = get_option('mxchat_options', []);
12012 - $model = $options['model'] ?? 'gpt-5.6-sol';
12013 -
12014 - // Detect provider from model prefix
12015 - $provider = strtolower(explode('-', $model)[0]);
12016 -
12017 - $sample_prompt = 'Hello! Can you stream this response back to me?';
12018 - $messages = [['role' => 'user', 'content' => $sample_prompt]];
12019 - $headers = [];
12020 - $body = [];
12021 - $url = '';
12022 - $api_key = '';
12023 -
12024 - switch ($provider) {
12025 - case 'gpt':
12026 - case 'o1':
12027 - $api_key = $options['api_key'] ?? '';
12028 - if (empty($api_key)) return '❌ Missing API key for OpenAI';
12029 - $url = 'https://api.openai.com/v1/chat/completions';
12030 - $headers = [
12031 - 'Content-Type: application/json',
12032 - 'Authorization: Bearer ' . $api_key
12033 - ];
12034 - $body = [
12035 - 'model' => $model,
12036 - 'messages' => $messages,
12037 - 'stream' => true
12038 - ];
12039 - break;
12040 -
12041 - case 'claude':
12042 - $api_key = $options['claude_api_key'] ?? '';
12043 - if (empty($api_key)) return '❌ Missing API key for Claude';
12044 - $url = 'https://api.anthropic.com/v1/messages';
12045 - $headers = [
12046 - 'Content-Type: application/json',
12047 - 'x-api-key: ' . $api_key,
12048 - 'anthropic-version: 2023-06-01'
12049 - ];
12050 - $body = [
12051 - 'model' => $model,
12052 - 'messages' => $messages,
12053 - 'max_tokens' => 100,
12054 - 'stream' => true
12055 - ];
12056 - break;
12057 -
12058 - case 'grok':
12059 - $api_key = $options['xai_api_key'] ?? '';
12060 - if (empty($api_key)) return '❌ Missing API key for X.AI';
12061 - $url = 'https://api.x.ai/v1/chat/completions';
12062 - $headers = [
12063 - 'Content-Type: application/json',
12064 - 'Authorization: Bearer ' . $api_key
12065 - ];
12066 - $body = [
12067 - 'model' => $model,
12068 - 'messages' => $messages,
12069 - 'stream' => true
12070 - ];
12071 - break;
12072 -
12073 - case 'deepseek':
12074 - if (empty($deepseek_api_key)) {
12075 - $error_response = [
12076 - 'error' => esc_html__('DeepSeek API key is not configured', 'mxchat'),
12077 - 'error_code' => 'missing_deepseek_api_key'
12078 - ];
12079 - if ($testing_data !== null) {
12080 - $error_response['testing_data'] = $testing_data;
12081 - }
12082 - return $error_response;
12083 - }
12084 - if ($streaming) {
12085 - return $this->mxchat_generate_response_deepseek_stream(
12086 - $selected_model,
12087 - $deepseek_api_key,
12088 - $conversation_history,
12089 - $relevant_content,
12090 - $session_id,
12091 - $testing_data // Pass testing data
12092 - );
12093 - } else {
12094 - $response = $this->mxchat_generate_response_deepseek(
12095 - $selected_model,
12096 - $deepseek_api_key,
12097 - $conversation_history,
12098 - $relevant_content,
12099 - $session_id
12100 - );
12101 - }
12102 - break;
12103 -
12104 - case 'gemini':
12105 - $api_key = $options['gemini_api_key'] ?? '';
12106 - if (empty($api_key)) return '❌ Missing API key for Gemini';
12107 - $url = 'https://generativelanguage.googleapis.com/v1beta/models/' . $model . ':streamGenerateContent?key=' . $api_key;
12108 - $headers = ['Content-Type: application/json'];
12109 - $body = [
12110 - 'contents' => [['role' => 'user', 'parts' => [['text' => $sample_prompt]]]],
12111 - 'generationConfig' => ['temperature' => 0.7]
12112 - ];
12113 - break;
12114 -
12115 - default:
12116 - return '❌ Unsupported provider: ' . $provider;
12117 - }
12118 -
12119 - // Do the actual streaming test
12120 - $ch = curl_init($url);
12121 - curl_setopt($ch, CURLOPT_POSTFIELDS, json_encode($body));
12122 - curl_setopt($ch, CURLOPT_HTTPHEADER, $headers);
12123 - curl_setopt($ch, CURLOPT_RETURNTRANSFER, true);
12124 - curl_setopt($ch, CURLOPT_TIMEOUT, 15);
12125 - curl_setopt($ch, CURLOPT_SSL_VERIFYPEER, true);
12126 -
12127 - $response = curl_exec($ch);
12128 - $http_code = curl_getinfo($ch, CURLINFO_HTTP_CODE);
12129 - $error = curl_error($ch);
12130 - curl_close($ch);
12131 -
12132 - if ($error) return "❌ cURL error: $error";
12133 - if ($http_code !== 200) {
12134 - $error_message = json_decode($response, true)['error']['message'] ?? 'Unknown';
12135 - return "❌ HTTP $http_code: $error_message";
12136 - }
12137 -
12138 - return true;
12139 -}
12140 -
12141 425 public function mxchat_dismiss_pre_chat_message() {
12142 426 // Get and sanitize the user identifier
12143 427 $user_id = $this->mxchat_get_user_identifier();
12144 428 $user_id = sanitize_key($user_id);
@@ -12149,118 +433,11 @@
12149 433
12150 434 wp_send_json_success();
12151 435 }
12152 436
12153 -public function mxchat_check_pre_chat_message_status() {
12154 - // Get and sanitize the user identifier
12155 - $user_id = $this->mxchat_get_user_identifier();
12156 - $user_id = sanitize_key($user_id);
12157 437
12158 - // Check if the transient exists (i.e., if the message was dismissed)
12159 - $transient_key = 'mxchat_pre_chat_message_dismissed_' . $user_id;
12160 - $dismissed = get_transient($transient_key);
12161 438
12162 - // Log the result to see if it's being set correctly
12163 - //error_log("Check pre-chat message dismissed for $user_id: " . ($dismissed ? 'Yes' : 'No'));
12164 -
12165 - if ($dismissed) {
12166 - wp_send_json_success(['dismissed' => true]);
12167 - } else {
12168 - wp_send_json_success(['dismissed' => false]);
12169 - }
12170 -
12171 - wp_die();
12172 -}
12173 -
12174 -/**
12175 - * Keyword leg for hybrid retrieval (plan-38ffa1): ranked keyword query over
12176 - * the WP-DB knowledge table. FULLTEXT when the index is available, LIKE on
12177 - * the top query terms otherwise (capability detected once and cached by
12178 - * MxChat_Utils::mxchat_hybrid_detect_capability). Respects the same bot
12179 - * scoping as the vector query ($bot_filter) and the same role-restriction
12180 - * access rules as vector candidates.
12181 - *
12182 - * @return array[] Ranked hits: [id, source_url, role_restriction, has_access]
12183 - */
12184 -private function mxchat_hybrid_keyword_search($user_query, $system_prompt_table, $bot_filter, $knowledge_manager) {
12185 - global $wpdb;
12186 -
12187 - $capability = get_option('mxchat_hybrid_keyword_capability', '');
12188 - if (!in_array($capability, array('fulltext', 'like'), true)) {
12189 - $capability = MxChat_Utils::mxchat_hybrid_detect_capability();
12190 - }
12191 -
12192 - $limit = 20;
12193 - $rows = array();
12194 -
12195 - if ($capability === 'fulltext') {
12196 - $rows = $wpdb->get_results($wpdb->prepare(
12197 - "SELECT id, source_url, role_restriction,
12198 - MATCH(article_content) AGAINST (%s IN NATURAL LANGUAGE MODE) AS kw_score
12199 - FROM {$system_prompt_table}
12200 - WHERE MATCH(article_content) AGAINST (%s IN NATURAL LANGUAGE MODE) {$bot_filter}
12201 - ORDER BY kw_score DESC, id ASC
12202 - LIMIT %d",
12203 - $user_query,
12204 - $user_query,
12205 - $limit
12206 - ));
12207 - } else {
12208 - // LIKE fallback: length-weighted term scoring. Longer, rarer tokens
12209 - // (the SKU, the error code) must outrank ubiquitous short words — an
12210 - // equal-weight score lets "the" + one common word tie with the exact
12211 - // token and the tie-break pick the wrong row (caught by the 38ffa1
12212 - // verification harness). Stopwords are dropped outright.
12213 - $stopwords = array('the', 'and', 'for', 'you', 'your', 'with', 'this', 'that', 'are', 'was', 'can', 'how', 'what', 'does', 'have', 'has', 'about', 'from', 'not', 'but', 'all', 'any', 'our', 'their');
12214 - $terms = preg_split('/[^\p{L}\p{N}_-]+/u', (string) $user_query, -1, PREG_SPLIT_NO_EMPTY);
12215 - $terms = array_filter($terms, function ($t) use ($stopwords) {
12216 - return mb_strlen($t) >= 3 && !in_array(mb_strtolower($t), $stopwords, true);
12217 - });
12218 - $terms = array_values(array_unique(array_map('mb_strtolower', $terms)));
12219 - usort($terms, function ($a, $b) {
12220 - return mb_strlen($b) <=> mb_strlen($a);
12221 - });
12222 - $terms = array_slice($terms, 0, 5);
12223 - if (empty($terms)) {
12224 - return array();
12225 - }
12226 -
12227 - $score_parts = array();
12228 - $where_parts = array();
12229 - $like_params = array();
12230 - foreach ($terms as $term) {
12231 - $score_parts[] = '((article_content LIKE %s) * ' . (int) mb_strlen($term) . ')';
12232 - $where_parts[] = 'article_content LIKE %s';
12233 - $like_params[] = '%' . $wpdb->esc_like($term) . '%';
12234 - }
12235 - $sql = "SELECT id, source_url, role_restriction, ("
12236 - . implode(' + ', $score_parts)
12237 - . ") AS kw_score FROM {$system_prompt_table} WHERE ("
12238 - . implode(' OR ', $where_parts)
12239 - . ") {$bot_filter} ORDER BY kw_score DESC, id ASC LIMIT %d";
12240 - $rows = $wpdb->get_results($wpdb->prepare(
12241 - $sql,
12242 - array_merge($like_params, $like_params, array($limit))
12243 - ));
12244 - }
12245 -
12246 - $hits = array();
12247 - foreach ((array) $rows as $row) {
12248 - $role_restriction = $row->role_restriction ?? 'public';
12249 - if (!$knowledge_manager->mxchat_user_has_content_access($role_restriction)) {
12250 - continue;
12251 - }
12252 - $hits[] = array(
12253 - 'id' => (int) $row->id,
12254 - 'source_url' => $row->source_url ?? '',
12255 - 'role_restriction' => $role_restriction,
12256 - 'has_access' => true,
12257 - );
12258 - }
12259 - return $hits;
12260 -}
12261 -
12262 -private function mxchat_calculate_cosine_similarity($vectorA, $vectorB) {
439 + private function mxchat_calculate_cosine_similarity($vectorA, $vectorB) {
12263 440 if (!is_array($vectorA) || !is_array($vectorB) || empty($vectorA) || empty($vectorB)) {
12264 441 return 0;
12265 442 }
12266 443
@@ -12280,1484 +457,111 @@
12280 457
12281 458 return $dotProduct / ($normA * $normB);
12282 459 }
12283 460
461 + public function mxchat_enqueue_scripts_styles() {
462 + // Define version numbers for the styles and scripts
463 + $chat_style_version = '1.0.8'; // Replace with your actual version
464 + $chat_script_version = '1.0.8'; // Replace with your actual version
12284 465
12285 -public function mxchat_enqueue_scripts_styles($force = false) {
12286 - // Idempotency guard (plan-915355): the smart-asset-loading safety net in
12287 - // render_chatbot_shortcode() may invoke this method a second time (or on
12288 - // every shortcode render). Run the body at most once per request so the
12289 - // nonce, dynamic-settings merge, delayed transient write, and wp_footer
12290 - // loader action never happen twice.
12291 - static $did_run = false;
12292 - if ($did_run) {
12293 - return;
12294 - }
12295 -
12296 - // Smart asset loading gate (plan-915355, opt-in, default OFF — toggle in
12297 - // MxChat → Settings → Optimization → Script Loading). When enabled and the
12298 - // shared display decision says the widget won't render on this request,
12299 - // skip all front-end assets. $force (the shortcode safety net) bypasses
12300 - // the gate because at that point the widget IS rendering. Note: bail
12301 - // WITHOUT setting $did_run, so a later forced call can still enqueue.
12302 - if (!$force
12303 - && class_exists('MxChat_Public')
12304 - && MxChat_Public::is_smart_asset_loading_enabled()
12305 - && !MxChat_Public::should_load_assets()) {
12306 - return;
12307 - }
12308 -
12309 - $did_run = true;
12310 -
12311 - // Fetch options from the database first to check loading strategy
12312 - $this->options = get_option('mxchat_options');
12313 - $loading_strategy = isset($this->options['script_loading_strategy']) ? $this->options['script_loading_strategy'] : 'default';
12314 -
12315 - // Always enqueue CSS immediately
12316 - wp_enqueue_style(
12317 - 'mxchat-chat-css',
12318 - plugin_dir_url(__FILE__) . '../css/chat-style.css',
12319 - array(),
12320 - MXCHAT_VERSION
12321 - );
12322 -
12323 - // Handle script loading based on strategy
12324 - if ($loading_strategy === 'default' || $loading_strategy === 'defer') {
12325 - // Enqueue the script normally
466 + // Correct path to the script file
12326 467 wp_enqueue_script(
12327 - 'mxchat-chat-js',
12328 - plugin_dir_url(__FILE__) . '../js/chat-script.js',
12329 - array('jquery'),
12330 - MXCHAT_VERSION,
12331 - true
468 + 'mxchat-chat-js', // Handle for the script
469 + plugin_dir_url(__FILE__) . '../js/chat-script.js', // Correct path using __FILE__
470 + array('jquery'), // Dependencies
471 + $chat_script_version, // Version for cache busting
472 + true // Load script in footer
12332 473 );
12333 474
12334 - // Add defer attribute if strategy is 'defer'
12335 - if ($loading_strategy === 'defer') {
12336 - wp_script_add_data('mxchat-chat-js', 'strategy', 'defer');
12337 - }
12338 - } else {
12339 - // For delay or interaction-based loading, we'll use a custom loader
12340 - // Don't enqueue the main script - we'll load it dynamically
12341 - add_action('wp_footer', array($this, 'mxchat_output_delayed_script_loader'), 99);
12342 - }
475 + // Enqueue the CSS file similarly
476 + wp_enqueue_style(
477 + 'mxchat-chat-css', // Handle for the style
478 + plugin_dir_url(__FILE__) . '../css/chat-style.css', // Correct path using __FILE__
479 + array(), // No dependencies
480 + $chat_style_version // Version for cache busting
481 + );
12343 482
12344 - $prompts_options = get_option('mxchat_prompts_options', array());
483 + // Fetch options from the database
484 + $this->options = get_option('mxchat_options');
12345 485
12346 - // Check if AI theme is active - if so, skip inline colors in JavaScript
12347 - $theme_options = get_option('mxchat_theme_options', array());
12348 - $ai_theme_active = !empty($theme_options['active_ai_theme_css']);
12349 - $has_bot_theme_assignments = !empty($theme_options['bot_theme_assignments']);
12350 - $skip_inline_colors = $ai_theme_active || $has_bot_theme_assignments;
486 + // Prepare settings to pass to JavaScript
487 + $style_settings = array(
488 + 'ajax_url' => admin_url('admin-ajax.php'),
489 + 'nonce' => wp_create_nonce('mxchat_chat_nonce'), // Nonce for security
490 + 'rate_limit_message' => 'Rate limit exceeded. Please try again later.',
491 + 'appendWidgetToBody' => $this->options['append_to_body'] ?? 'off'
492 + );
12351 493
12352 - // Prepare settings for JavaScript
12353 - $style_settings = array(
12354 - 'ajax_url' => admin_url('admin-ajax.php'),
12355 - // The chat-send nonce is now fetched per-request from /wp-json/mxchat/v1/nonce
12356 - // (plan-6a68c9) so it never sits in cached HTML. We still emit a nonce here
12357 - // as a one-shot fallback for the first interaction on a fresh page load
12358 - // (so the very first chat-send doesn't need to wait for a REST round-trip),
12359 - // but the widget refetches before each subsequent send.
12360 - 'nonce' => wp_create_nonce('mxchat_chat_send'),
12361 - 'rest_url' => esc_url_raw(trailingslashit(rest_url('mxchat/v1'))),
12362 - 'contextual_awareness_toggle' => isset($this->options['contextual_awareness_toggle']) ? $this->options['contextual_awareness_toggle'] : 'off',
12363 - 'link_target_toggle' => $this->options['link_target_toggle'] ?? 'off',
12364 - 'complianz_toggle' => isset($this->options['complianz_toggle']) && $this->options['complianz_toggle'] === 'on',
12365 - 'user_message_bg_color' => $this->options['user_message_bg_color'] ?? '#fff',
12366 - 'user_message_font_color' => $this->options['user_message_font_color'] ?? '#212121',
12367 - 'bot_message_bg_color' => $this->options['bot_message_bg_color'] ?? '#212121',
12368 - 'bot_message_font_color' => $this->options['bot_message_font_color'] ?? '#fff',
12369 - 'top_bar_bg_color' => $this->options['top_bar_bg_color'] ?? '#212121',
12370 - 'send_button_font_color' => $this->options['send_button_font_color'] ?? '#212121',
12371 - 'close_button_color' => $this->options['close_button_color'] ?? '#fff',
12372 - 'chatbot_background_color' => $this->options['chatbot_background_color'] ?? '#212121',
12373 - 'chatbot_bg_color' => $this->options['chatbot_bg_color'] ?? '#fff',
12374 - 'icon_color' => $this->options['icon_color'] ?? '#fff',
12375 - 'chat_input_font_color' => $this->options['chat_input_font_color'] ?? '#212121',
12376 - 'chat_persistence_toggle' => $this->options['chat_persistence_toggle'] ?? 'off',
12377 - 'appendWidgetToBody' => $this->options['append_to_body'] ?? 'off',
12378 - 'live_agent_message_bg_color' => $this->options['live_agent_message_bg_color'] ?? '#ffffff',
12379 - 'live_agent_message_font_color' => $this->options['live_agent_message_font_color'] ?? '#333333',
12380 - 'mode_indicator_bg_color' => $this->options['mode_indicator_bg_color'] ?? '#767676',
12381 - 'mode_indicator_font_color' => $this->options['mode_indicator_font_color'] ?? '#ffffff',
12382 - 'toolbar_icon_color' => $this->options['toolbar_icon_color'] ?? '#212121',
12383 - 'use_pinecone' => $prompts_options['mxchat_use_pinecone'] ?? '0',
12384 - 'email_collection_enabled' => $this->options['enable_email_block'] ?? 'off', // FIXED
12385 - 'initial_email_state' => null, // Also fixed this undefined variable
12386 - 'skip_email_check' => true,
12387 - 'pinecone_enabled' => isset($prompts_options['mxchat_use_pinecone']) && $prompts_options['mxchat_use_pinecone'] === '1',
12388 - 'skip_inline_colors' => $skip_inline_colors,
12389 - 'bot_theme_assignments' => $theme_options['bot_theme_assignments'] ?? array(),
12390 - );
12391 -
12392 - // Behavior gates + labels (model, streaming, rate-limit copy, toolbar,
12393 - // print/transcript, satisfaction rating) come from the shared
12394 - // dynamic-settings method so this inline payload and the first-open
12395 - // refresh endpoint can never drift (plan-32db95).
12396 - $style_settings = array_merge($style_settings, $this->get_dynamic_widget_settings());
12397 -
12398 - // For normal/defer loading, use wp_localize_script.
12399 - // For delayed loading, nothing is localized or stored here: the delayed
12400 - // loader (mxchat_output_delayed_script_loader) rebuilds the full settings
12401 - // array inline from options and never reads any stored copy.
12402 - if ($loading_strategy === 'default' || $loading_strategy === 'defer') {
494 + // Localize the script with necessary data
12403 495 wp_localize_script('mxchat-chat-js', 'mxchatChat', $style_settings);
12404 - } else {
12405 - // Late-render fallback (plan-915355): when the shortcode safety net
12406 - // forces this method during/after wp_footer (footer widget areas, late
12407 - // builder regions), the wp_footer:99 loader action registered above may
12408 - // already be past its slot. Emit the loader inline right now; its
12409 - // emitted-once guard prevents double output if :99 still fires.
12410 - if ($force && did_action('wp_footer')) {
12411 - $this->mxchat_output_delayed_script_loader();
12412 - }
12413 496 }
12414 -}
12415 497
12416 -/**
12417 - * Output the delayed script loader for performance optimization
12418 - */
12419 -public function mxchat_output_delayed_script_loader() {
12420 - // Emitted-once guard (plan-915355): this can now be reached both via the
12421 - // wp_footer:99 action and via the late-render inline fallback in
12422 - // mxchat_enqueue_scripts_styles(). The loader must print exactly once.
12423 - static $emitted = false;
12424 - if ($emitted) {
12425 - return;
12426 - }
12427 - $emitted = true;
12428 498
12429 - $this->options = get_option('mxchat_options');
12430 - $loading_strategy = isset($this->options['script_loading_strategy']) ? $this->options['script_loading_strategy'] : 'default';
12431 - $script_url = plugin_dir_url(__FILE__) . '../js/chat-script.js?ver=' . MXCHAT_VERSION;
12432 499
12433 - // Get the stored settings
12434 - $prompts_options = get_option('mxchat_prompts_options', array());
12435 - $theme_options = get_option('mxchat_theme_options', array());
12436 - $ai_theme_active = !empty($theme_options['active_ai_theme_css']);
12437 - $has_bot_theme_assignments = !empty($theme_options['bot_theme_assignments']);
12438 - $skip_inline_colors = $ai_theme_active || $has_bot_theme_assignments;
500 + public function mxchat_reset_rate_limits() {
501 + global $wpdb;
12439 502
12440 - $style_settings = array(
12441 - 'ajax_url' => admin_url('admin-ajax.php'),
12442 - // Per-request nonce — see plan-6a68c9; widget fetches via /wp-json/mxchat/v1/nonce
12443 - // before each send. This inline value is a one-shot fallback for the first interaction.
12444 - 'nonce' => wp_create_nonce('mxchat_chat_send'),
12445 - 'rest_url' => esc_url_raw(trailingslashit(rest_url('mxchat/v1'))),
12446 - 'contextual_awareness_toggle' => isset($this->options['contextual_awareness_toggle']) ? $this->options['contextual_awareness_toggle'] : 'off',
12447 - 'link_target_toggle' => $this->options['link_target_toggle'] ?? 'off',
12448 - 'complianz_toggle' => isset($this->options['complianz_toggle']) && $this->options['complianz_toggle'] === 'on',
12449 - 'user_message_bg_color' => $this->options['user_message_bg_color'] ?? '#fff',
12450 - 'user_message_font_color' => $this->options['user_message_font_color'] ?? '#212121',
12451 - 'bot_message_bg_color' => $this->options['bot_message_bg_color'] ?? '#212121',
12452 - 'bot_message_font_color' => $this->options['bot_message_font_color'] ?? '#fff',
12453 - 'top_bar_bg_color' => $this->options['top_bar_bg_color'] ?? '#212121',
12454 - 'send_button_font_color' => $this->options['send_button_font_color'] ?? '#212121',
12455 - 'close_button_color' => $this->options['close_button_color'] ?? '#fff',
12456 - 'chatbot_background_color' => $this->options['chatbot_background_color'] ?? '#212121',
12457 - 'chatbot_bg_color' => $this->options['chatbot_bg_color'] ?? '#fff',
12458 - 'icon_color' => $this->options['icon_color'] ?? '#fff',
12459 - 'chat_input_font_color' => $this->options['chat_input_font_color'] ?? '#212121',
12460 - 'chat_persistence_toggle' => $this->options['chat_persistence_toggle'] ?? 'off',
12461 - 'appendWidgetToBody' => $this->options['append_to_body'] ?? 'off',
12462 - 'live_agent_message_bg_color' => $this->options['live_agent_message_bg_color'] ?? '#ffffff',
12463 - 'live_agent_message_font_color' => $this->options['live_agent_message_font_color'] ?? '#333333',
12464 - 'mode_indicator_bg_color' => $this->options['mode_indicator_bg_color'] ?? '#767676',
12465 - 'mode_indicator_font_color' => $this->options['mode_indicator_font_color'] ?? '#ffffff',
12466 - 'toolbar_icon_color' => $this->options['toolbar_icon_color'] ?? '#212121',
12467 - 'use_pinecone' => $prompts_options['mxchat_use_pinecone'] ?? '0',
12468 - 'email_collection_enabled' => $this->options['enable_email_block'] ?? 'off',
12469 - 'initial_email_state' => null,
12470 - 'skip_email_check' => true,
12471 - 'pinecone_enabled' => isset($prompts_options['mxchat_use_pinecone']) && $prompts_options['mxchat_use_pinecone'] === '1',
12472 - 'skip_inline_colors' => $skip_inline_colors,
12473 - 'bot_theme_assignments' => $theme_options['bot_theme_assignments'] ?? array(),
12474 - );
503 + // Define a cache key pattern for rate limits
504 + $cache_key_pattern = 'mxchat_chat_limit_%';
12475 505
12476 - // Behavior gates + labels (model, streaming, rate-limit copy, toolbar,
12477 - // print/transcript, satisfaction rating) come from the shared
12478 - // dynamic-settings method so this inline payload and the first-open
12479 - // refresh endpoint can never drift (plan-32db95).
12480 - $style_settings = array_merge($style_settings, $this->get_dynamic_widget_settings());
506 + // Retrieve all option names matching the pattern
507 + // phpcs:ignore WordPress.DB.DirectDatabaseQuery.NoCaching, WordPress.DB.DirectDatabaseQuery.DirectQuery
508 + $option_names = $wpdb->get_col("SELECT option_name FROM {$wpdb->options} WHERE option_name LIKE 'mxchat_chat_limit_%'");
12481 509
12482 - // Determine delay time based on strategy
12483 - $delay_ms = 0;
12484 - switch ($loading_strategy) {
12485 - case 'delay_1s':
12486 - $delay_ms = 1000;
12487 - break;
12488 - case 'delay_3s':
12489 - $delay_ms = 3000;
12490 - break;
12491 - case 'delay_5s':
12492 - $delay_ms = 5000;
12493 - break;
12494 - }
510 + // db call ok; no-cache ok
511 + // phpcs:ignore WordPress.DB.DirectDatabaseQuery.DirectQuery -- db call ok
512 + $wpdb->query("DELETE FROM {$wpdb->options} WHERE option_name LIKE 'mxchat_chat_limit_%'");
12495 513
12496 - ?>
12497 - <script type="text/javascript">
12498 - (function() {
12499 - var mxchatLoaded = false;
12500 - var mxchatChat = <?php echo wp_json_encode($style_settings); ?>;
12501 - window.mxchatChat = mxchatChat;
12502 -
12503 - function loadMxChatScript() {
12504 - if (mxchatLoaded) return;
12505 - mxchatLoaded = true;
12506 -
12507 - function appendChatScript() {
12508 - var script = document.createElement('script');
12509 - script.src = <?php echo wp_json_encode($script_url); ?>;
12510 - script.type = 'text/javascript';
12511 - document.body.appendChild(script);
12512 - }
12513 -
12514 - if (typeof jQuery !== 'undefined') {
12515 - appendChatScript();
12516 - } else {
12517 - var jq = document.createElement('script');
12518 - jq.src = <?php echo wp_json_encode(includes_url('js/jquery/jquery.min.js')); ?>;
12519 - jq.onload = appendChatScript;
12520 - document.body.appendChild(jq);
12521 - }
514 + // Clear the relevant cache entries
515 + foreach ($option_names as $option_name) {
516 + wp_cache_delete($option_name, 'options');
12522 517 }
12523 518
12524 - <?php if ($loading_strategy === 'on_interaction'): ?>
12525 - // Load on user interaction
12526 - var events = ['scroll', 'mousemove', 'touchstart', 'keydown', 'click'];
12527 - events.forEach(function(evt) {
12528 - window.addEventListener(evt, loadMxChatScript, {once: true, passive: true});
12529 - });
12530 - // Fallback: load after 8 seconds if no interaction
12531 - setTimeout(loadMxChatScript, 8000);
12532 - <?php else: ?>
12533 - // Load after specified delay
12534 - setTimeout(loadMxChatScript, <?php echo intval($delay_ms); ?>);
12535 - <?php endif; ?>
12536 - })();
12537 - </script>
12538 - <?php
12539 -}
12540 -
12541 -/**
12542 - * Setup the cron jobs for rate limits with guard against multiple calls
12543 - */
12544 -public function setup_rate_limit_cron_jobs() {
12545 - // Add a guard to prevent multiple rapid calls
12546 - $last_setup = get_transient('mxchat_cron_setup_guard');
12547 - if ($last_setup && (time() - $last_setup) < 60) {
12548 - // Don't run again if we ran less than 60 seconds ago
12549 - return;
519 + // Optionally, clear a general cache if you have one
520 + wp_cache_delete('mxchat_all_chat_limits', 'options');
12550 521 }
12551 -
12552 - // Set the guard
12553 - set_transient('mxchat_cron_setup_guard', time(), 300); // 5 minutes
12554 -
12555 - try {
12556 - // First, check if WordPress cron is disabled
12557 - if (defined('DISABLE_WP_CRON') && DISABLE_WP_CRON) {
12558 - //error_log('MxChat: WordPress cron is disabled (DISABLE_WP_CRON = true), using fallback system');
12559 - $this->setup_fallback_rate_limit_system();
12560 - return;
12561 - }
12562 -
12563 - // Check if cron is already scheduled - if so, don't mess with it
12564 - if (wp_next_scheduled('mxchat_reset_rate_limits')) {
12565 - //error_log('MxChat: Rate limit cron already scheduled, skipping setup');
12566 - return;
12567 - }
12568 -
12569 - // Clear any orphaned hooks (but don't loop indefinitely)
12570 - $hooks_to_clear = [
12571 - 'mxchat_reset_rate_limits',
12572 - 'mxchat_reset_hourly_rate_limits',
12573 - 'mxchat_reset_daily_rate_limits',
12574 - 'mxchat_reset_weekly_rate_limits',
12575 - 'mxchat_reset_monthly_rate_limits'
12576 - ];
12577 -
12578 - foreach ($hooks_to_clear as $hook) {
12579 - // Only clear a maximum of 3 instances to prevent infinite loops
12580 - $cleared = 0;
12581 - while (wp_next_scheduled($hook) && $cleared < 3) {
12582 - wp_clear_scheduled_hook($hook);
12583 - $cleared++;
12584 - }
12585 - }
12586 -
12587 - // Small delay after clearing
12588 - usleep(100000); // 0.1 seconds
12589 -
12590 - // Try to schedule the event
12591 - $initial_time = time() + 300; // Start in 5 minutes
12592 - $result = wp_schedule_event($initial_time, 'hourly', 'mxchat_reset_rate_limits');
12593 522
12594 - if ($result === false) {
12595 - //error_log('MxChat: Failed to schedule cron, using fallback system');
12596 - $this->setup_fallback_rate_limit_system();
12597 - } else {
12598 - if (defined('MXCHAT_DEV_MODE') && MXCHAT_DEV_MODE) {
12599 - error_log('MxChat: rate-limit reset cron event was missing and has been re-scheduled');
12600 - }
12601 - }
12602 -
12603 - } catch (Exception $e) {
12604 - //error_log('MxChat: Cron setup exception: ' . $e->getMessage());
12605 - $this->setup_fallback_rate_limit_system();
12606 - }
12607 -}
12608 523
12609 -/**
12610 - * Try alternative cron scheduling methods
12611 - */
12612 -private function try_alternative_cron_scheduling($initial_time) {
12613 - try {
12614 - // Method 1: Try with current time instead of future time
12615 - $result1 = wp_schedule_event(time(), 'hourly', 'mxchat_reset_rate_limits');
12616 - if ($result1 !== false) {
12617 - //error_log('MxChat: Alternative method 1 (current time) succeeded');
12618 - return true;
12619 - }
12620 -
12621 - // Method 2: Try with a different interval
12622 - $result2 = wp_schedule_event($initial_time, 'daily', 'mxchat_reset_rate_limits');
12623 - if ($result2 !== false) {
12624 - //error_log('MxChat: Alternative method 2 (daily interval) succeeded');
12625 - return true;
12626 - }
12627 -
12628 - // Method 3: Try wp_schedule_single_event first, then recurring
12629 - $result3 = wp_schedule_single_event($initial_time, 'mxchat_reset_rate_limits');
12630 - if ($result3 !== false) {
12631 - //error_log('MxChat: Alternative method 3 (single event) succeeded');
12632 - // Schedule the next one manually in the handler
12633 - return true;
12634 - }
12635 -
12636 - return false;
12637 -
12638 - } catch (Exception $e) {
12639 - //error_log('MxChat: Alternative cron scheduling exception: ' . $e->getMessage());
12640 - return false;
524 +private function mxchat_fetch_woocommerce_products() {
525 + // Ensure WooCommerce is active
526 + if (!class_exists('WooCommerce')) {
527 + return [];
12641 528 }
12642 -}
12643 529
12644 -/**
12645 - * Enhanced fallback rate limit system
12646 - */
12647 -private function setup_fallback_rate_limit_system() {
12648 - // Idempotence matters here: with setup_rate_limit_cron_jobs() hooked to
12649 - // admin_init, a DISABLE_WP_CRON site reaches this on every guard pass.
12650 - // Unconditionally rewriting mxchat_next_rate_limit_check to now+3600 would
12651 - // slide the deadline forward forever and the fallback reset would never
12652 - // fire. Only initialize the deadline on a genuine transition into fallback
12653 - // mode (or if it's somehow missing).
12654 - $already_active = get_option('mxchat_use_fallback_rate_limits', false);
530 + $args = array(
531 + 'post_type' => 'product',
532 + 'post_status' => 'publish',
533 + 'posts_per_page' => -1,
534 + );
12655 535
12656 - // Set a flag to use database-based rate limit cleanup
12657 - update_option('mxchat_use_fallback_rate_limits', true);
536 + $products = get_posts($args);
537 + $product_data = [];
12658 538
12659 - // Schedule a one-time check to happen on the next plugin load
12660 - if (!$already_active || !get_option('mxchat_next_rate_limit_check', 0)) {
12661 - update_option('mxchat_next_rate_limit_check', time() + 3600);
12662 - }
539 + foreach ($products as $product) {
540 + $product_id = $product->ID;
541 + $product_obj = wc_get_product($product_id);
12663 542
12664 - // Also set up a more frequent fallback check (every 4 hours)
12665 - update_option('mxchat_fallback_check_interval', 4 * 3600);
12666 -
12667 - //error_log('MxChat: Fallback rate limit system activated');
12668 -}
12669 -
12670 -/**
12671 - * Enhanced fallback check method
12672 - * NOTE: mxchat_check_fallback_rate_limits() in mxchat-basic.php is a second
12673 - * implementation of this same check — if either changes, change both.
12674 - */
12675 -public function check_fallback_rate_limits() {
12676 - $use_fallback = get_option('mxchat_use_fallback_rate_limits', false);
12677 -
12678 - if (!$use_fallback) {
12679 - return; // Regular cron is working
12680 - }
12681 -
12682 - $next_check = get_option('mxchat_next_rate_limit_check', 0);
12683 - $check_interval = get_option('mxchat_fallback_check_interval', 3600);
12684 -
12685 - if (time() >= $next_check) {
12686 - //error_log('MxChat: Running fallback rate limit cleanup');
12687 - $this->mxchat_reset_rate_limits();
12688 -
12689 - // Schedule next check
12690 - update_option('mxchat_next_rate_limit_check', time() + $check_interval);
12691 - }
12692 -}
12693 -/**
12694 - * Enhanced rate limit check that includes fallback cleanup and bot-specific rate limits
12695 - */
12696 -public function check_rate_limit() {
12697 - // Check if we need to run fallback cleanup
12698 - $use_fallback = get_option('mxchat_use_fallback_rate_limits', false);
12699 - $next_check = get_option('mxchat_next_rate_limit_check', 0);
12700 -
12701 - if ($use_fallback && time() >= $next_check) {
12702 - $this->mxchat_reset_rate_limits();
12703 - update_option('mxchat_next_rate_limit_check', time() + 3600); // Next hour
12704 - }
12705 -
12706 - // Get bot ID from current request context
12707 - $bot_id = isset($_POST['bot_id']) ? sanitize_key($_POST['bot_id']) : 'default';
12708 -
12709 - // Get bot-specific options (includes rate limits if overridden)
12710 - $bot_options = $this->get_bot_options($bot_id);
12711 - $current_options = !empty($bot_options) ? $bot_options : $this->options;
12712 -
12713 - // Use bot-specific rate limits if available, otherwise fall back to default
12714 - $rate_limits_source = isset($current_options['rate_limits']) ? $current_options['rate_limits'] : get_option('mxchat_options', [])['rate_limits'] ?? [];
12715 -
12716 - // -------------------------------------------------------------------
12717 - // Whole-chatbot global cap (independent of role). Evaluated FIRST so
12718 - // it acts as a hard ceiling across all users + all roles. Default is
12719 - // 'unlimited' so existing installs are unchanged. Counter key drops
12720 - // both <role> and <user_id> segments — single pool per bot.
12721 - // -------------------------------------------------------------------
12722 - $global_cfg = isset($current_options['rate_limits_global']) && is_array($current_options['rate_limits_global'])
12723 - ? $current_options['rate_limits_global']
12724 - : (isset(get_option('mxchat_options', [])['rate_limits_global']) ? get_option('mxchat_options', [])['rate_limits_global'] : []);
12725 - $global_limit_raw = isset($global_cfg['limit']) ? (string) $global_cfg['limit'] : 'unlimited';
12726 - $global_timeframe = isset($global_cfg['timeframe']) ? (string) $global_cfg['timeframe'] : 'daily';
12727 - if ($global_limit_raw !== '' && $global_limit_raw !== 'unlimited' && (int) $global_limit_raw >= 1) {
12728 - $bot_id_for_global = isset($_POST['bot_id']) ? sanitize_key($_POST['bot_id']) : 'default';
12729 - $safe_bot_global = preg_replace('/[^a-zA-Z0-9_]/', '_', $bot_id_for_global);
12730 - $global_option = 'mxchat_chat_limit_' . $safe_bot_global . '_global';
12731 - $global_data = get_option($global_option, ['count' => 0, 'timestamp' => time()]);
12732 - if ((int) $global_data['count'] === 0) {
12733 - $global_data['timestamp'] = time();
12734 - update_option($global_option, $global_data);
12735 - }
12736 - $now = time();
12737 - $ts = (int) $global_data['timestamp'];
12738 - $reset = false;
12739 - switch ($global_timeframe) {
12740 - case 'hourly': $reset = ($now - $ts) >= 3600; break;
12741 - case 'daily': $reset = ($now - $ts) >= 86400; break;
12742 - case 'weekly': $reset = ($now - $ts) >= 604800; break;
12743 - case 'monthly': $reset = ($now - $ts) >= 2592000; break;
12744 - }
12745 - if ($reset) {
12746 - $global_data = ['count' => 0, 'timestamp' => $now];
12747 - update_option($global_option, $global_data);
12748 - }
12749 - if ((int) $global_data['count'] >= (int) $global_limit_raw) {
12750 - $global_msg = !empty($global_cfg['message'])
12751 - ? $global_cfg['message']
12752 - : __('This chatbot has reached its message limit. Please try again later.', 'mxchat');
12753 - return [
12754 - 'error' => true,
12755 - 'message' => $this->process_rate_limit_message_html($global_msg),
12756 - ];
12757 - }
12758 - // Reserve the slot for this request. Per-role check below also increments
12759 - // its own counter — that is intentional, both ceilings apply independently.
12760 - $global_data['count']++;
12761 - update_option($global_option, $global_data);
12762 - }
12763 -
12764 - // Determine user role or if logged out
12765 - if (is_user_logged_in()) {
12766 - $user = wp_get_current_user();
12767 - $user_id = $user->ID;
12768 -
12769 - // Get the user's primary role using reset() to safely get the first element
12770 - $user_roles = $user->roles;
12771 -
12772 - // Safely get the first role regardless of array key structure
12773 - if (!empty($user_roles) && is_array($user_roles)) {
12774 - $role = reset($user_roles); // This safely gets the first element regardless of key
12775 - } else {
12776 - $role = 'subscriber'; // Default to subscriber if no role found
12777 - }
12778 - } else {
12779 - $role = 'logged_out';
12780 - // Use IP address for non-logged-in users
12781 - $user_id = $this->get_client_ip();
12782 - }
12783 -
12784 - // Check if rate limits are configured for this role
12785 - if (!isset($rate_limits_source[$role])) {
12786 - return true; // No limit set for this role
12787 - }
12788 -
12789 - $limit = $rate_limits_source[$role]['limit'];
12790 -
12791 - // If unlimited, return true immediately
12792 - if ($limit === 'unlimited') {
12793 - return true;
12794 - }
12795 -
12796 - // Get the option name for this user/role with safer naming (include bot_id for bot-specific limits)
12797 - $safe_role = preg_replace('/[^a-zA-Z0-9_]/', '_', $role);
12798 - $safe_user_id = preg_replace('/[^a-zA-Z0-9_]/', '_', $user_id);
12799 - $safe_bot_id = preg_replace('/[^a-zA-Z0-9_]/', '_', $bot_id);
12800 -
12801 - // Include bot_id in option name so each bot has separate rate limits
12802 - $option_name = 'mxchat_chat_limit_' . $safe_bot_id . '_' . $safe_role . '_' . $safe_user_id;
12803 -
12804 - // Get the counter data
12805 - $limit_data = get_option($option_name, ['count' => 0, 'timestamp' => time()]);
12806 -
12807 - // If first request or counter reset needed, set the initial timestamp
12808 - if ($limit_data['count'] === 0) {
12809 - $limit_data['timestamp'] = time();
12810 - update_option($option_name, $limit_data);
12811 - }
12812 -
12813 - // Get the timeframe
12814 - $timeframe = isset($rate_limits_source[$role]['timeframe']) ?
12815 - $rate_limits_source[$role]['timeframe'] : 'daily';
12816 -
12817 - // Check if the counter needs to be reset based on timeframe
12818 - $current_time = time();
12819 - $timestamp = $limit_data['timestamp'];
12820 - $should_reset = false;
12821 -
12822 - switch ($timeframe) {
12823 - case 'hourly':
12824 - $should_reset = ($current_time - $timestamp) >= 3600; // 1 hour
12825 - break;
12826 - case 'daily':
12827 - $should_reset = ($current_time - $timestamp) >= 86400; // 24 hours
12828 - break;
12829 - case 'weekly':
12830 - $should_reset = ($current_time - $timestamp) >= 604800; // 7 days
12831 - break;
12832 - case 'monthly':
12833 - $should_reset = ($current_time - $timestamp) >= 2592000; // 30 days
12834 - break;
12835 - }
12836 -
12837 - // Reset the counter if the timeframe has passed
12838 - if ($should_reset) {
12839 - $limit_data = ['count' => 0, 'timestamp' => $current_time];
12840 - update_option($option_name, $limit_data);
12841 - }
12842 -
12843 - // Check if user has exceeded their limit
12844 - if ($limit_data['count'] >= intval($limit)) {
12845 - // Get the custom message for this role
12846 - $message = !empty($rate_limits_source[$role]['message'])
12847 - ? $rate_limits_source[$role]['message']
12848 - : __('Rate limit exceeded. Please try again later.', 'mxchat');
12849 -
12850 - // Add timeframe information to the message if placeholders exist
12851 - $timeframe_label = '';
12852 - switch ($timeframe) {
12853 - case 'hourly':
12854 - $timeframe_label = __('hour', 'mxchat');
12855 - break;
12856 - case 'daily':
12857 - $timeframe_label = __('day', 'mxchat');
12858 - break;
12859 - case 'weekly':
12860 - $timeframe_label = __('week', 'mxchat');
12861 - break;
12862 - case 'monthly':
12863 - $timeframe_label = __('month', 'mxchat');
12864 - break;
12865 - }
12866 -
12867 - // Replace placeholders in the message
12868 - $message = str_replace(
12869 - ['{limit}', '{count}', '{remaining}', '{timeframe}'],
12870 - [intval($limit), $limit_data['count'], max(0, intval($limit) - $limit_data['count']), $timeframe_label],
12871 - $message
543 + $product_data[] = array(
544 + 'id' => $product_id,
545 + 'name' => $product_obj->get_name(),
546 + 'description' => $product_obj->get_description(),
547 + 'short_description' => $product_obj->get_short_description(),
548 + 'url' => get_permalink($product_id),
549 + 'price' => $product_obj->get_regular_price(),
550 + 'sale_price' => $product_obj->get_sale_price(),
551 + 'stock_status' => $product_obj->get_stock_status(),
552 + 'sku' => $product_obj->get_sku(),
553 + 'in_stock' => $product_obj->is_in_stock(),
554 + 'total_sales' => $product_obj->get_total_sales(),
12872 555 );
12873 -
12874 - // Process HTML links in the message
12875 - $message = $this->process_rate_limit_message_html($message);
12876 -
12877 - // Return error with the processed message
12878 - return [
12879 - 'error' => true,
12880 - 'message' => $message
12881 - ];
12882 556 }
12883 -
12884 - // Increment the counter
12885 - $limit_data['count']++;
12886 - update_option($option_name, $limit_data);
12887 -
12888 - return true;
12889 -}
12890 557
12891 -/**
12892 - * Enhanced rate limit reset with better error handling
12893 - */
12894 -public function mxchat_reset_rate_limits() {
12895 - try {
12896 - global $wpdb;
12897 - $all_options = get_option('mxchat_options', []);
12898 - $current_time = time();
12899 -
12900 - // Get rate limit options with a safer query and limit
12901 - $option_names = $wpdb->get_col(
12902 - $wpdb->prepare(
12903 - "SELECT option_name FROM {$wpdb->options}
12904 - WHERE option_name LIKE %s
12905 - LIMIT 1000",
12906 - 'mxchat_chat_limit_%'
12907 - )
12908 - );
12909 -
12910 - if (empty($option_names)) {
12911 - return;
12912 - }
12913 -
12914 - $processed_count = 0;
12915 - $max_processing_time = 30; // Maximum 30 seconds
12916 - $start_time = time();
12917 -
12918 - foreach ($option_names as $option_name) {
12919 - // Check processing time limit
12920 - if ((time() - $start_time) > $max_processing_time) {
12921 - //error_log('MxChat: Rate limit reset timeout after processing ' . $processed_count . ' entries');
12922 - break;
12923 - }
12924 -
12925 - // Parse the option name more safely
12926 - if (!preg_match('/^mxchat_chat_limit_(.+)_(.+)$/', $option_name, $matches)) {
12927 - continue;
12928 - }
12929 -
12930 - $role_and_user = $matches[1] . '_' . $matches[2];
12931 - $parts = explode('_', $role_and_user);
12932 -
12933 - if (count($parts) < 2) {
12934 - continue;
12935 - }
12936 -
12937 - // Extract role (everything except the last part which is user ID)
12938 - $user_id_part = array_pop($parts);
12939 - $role = implode('_', $parts);
12940 -
12941 - // Skip if role doesn't exist in our settings
12942 - if (!isset($all_options['rate_limits'][$role])) {
12943 - // Clean up orphaned entries
12944 - delete_option($option_name);
12945 - continue;
12946 - }
12947 -
12948 - $timeframe = $all_options['rate_limits'][$role]['timeframe'] ?? 'daily';
12949 - $limit_data = get_option($option_name);
12950 -
12951 - if (!$limit_data || !is_array($limit_data) || !isset($limit_data['timestamp'])) {
12952 - // Clean up invalid entries
12953 - delete_option($option_name);
12954 - continue;
12955 - }
12956 -
12957 - $timestamp = $limit_data['timestamp'];
12958 - $should_reset = false;
12959 -
12960 - // Determine if we should reset based on the timeframe
12961 - switch ($timeframe) {
12962 - case 'hourly':
12963 - $should_reset = ($current_time - $timestamp) >= 3600;
12964 - break;
12965 - case 'daily':
12966 - $should_reset = ($current_time - $timestamp) >= 86400;
12967 - break;
12968 - case 'weekly':
12969 - $should_reset = ($current_time - $timestamp) >= 604800;
12970 - break;
12971 - case 'monthly':
12972 - $should_reset = ($current_time - $timestamp) >= 2592000;
12973 - break;
12974 - }
12975 -
12976 - // Reset the counter if the timeframe has passed
12977 - if ($should_reset) {
12978 - delete_option($option_name);
12979 - wp_cache_delete($option_name, 'options');
12980 - $processed_count++;
12981 - }
12982 - }
12983 -
12984 - // Clean up any orphaned cache entries
12985 - wp_cache_delete('mxchat_all_chat_limits', 'options');
12986 -
12987 - //error_log("MxChat: Rate limit reset completed. Processed {$processed_count} entries.");
12988 -
12989 - } catch (Exception $e) {
12990 - //error_log('MxChat: Rate limit reset error: ' . $e->getMessage());
12991 - }
558 + return $product_data;
12992 559 }
12993 560
12994 561
12995 -/**
12996 - * Process HTML links in rate limit messages
12997 - *
12998 - * @param string $message The rate limit message
12999 - * @return string The processed message with safe HTML links
13000 - */
13001 -private function process_rate_limit_message_html($message) {
13002 - // Return original message if empty
13003 - if (empty($message)) {
13004 - return $message;
13005 - }
13006 -
13007 - // First, convert markdown links to HTML
13008 - $message = $this->convert_markdown_links($message);
13009 -
13010 - // Then, auto-convert any remaining plain URLs to links
13011 - $message = $this->auto_link_urls($message);
13012 -
13013 - // Allow basic HTML tags for links and formatting
13014 - $allowed_tags = [
13015 - 'a' => [
13016 - 'href' => true,
13017 - 'target' => true,
13018 - 'rel' => true,
13019 - 'title' => true,
13020 - 'class' => true
13021 - ],
13022 - 'strong' => [],
13023 - 'em' => [],
13024 - 'br' => [],
13025 - 'b' => [],
13026 - 'i' => [],
13027 - 'span' => ['class' => true]
13028 - ];
13029 -
13030 - // Sanitize but allow the specified HTML tags
13031 - $processed_message = wp_kses($message, $allowed_tags);
13032 -
13033 - // If wp_kses stripped everything, return the original message as plain text
13034 - if (empty($processed_message) && !empty($message)) {
13035 - // Strip all HTML and return plain text as fallback
13036 - return wp_strip_all_tags($message);
13037 - }
13038 -
13039 - return $processed_message;
13040 -}
13041 562
13042 -/**
13043 - * Convert markdown links to HTML
13044 - *
13045 - * @param string $text The text to process
13046 - * @return string The text with markdown links converted to HTML
13047 - */
13048 -private function convert_markdown_links($text) {
13049 - // Return original text if empty
13050 - if (empty($text)) {
13051 - return $text;
13052 - }
13053 -
13054 - // Pattern to match markdown links: [text](url)
13055 - $pattern = '/\[([^\]]+)\]\(([^)]+)\)/';
13056 -
13057 - $processed_text = preg_replace_callback($pattern, function($matches) {
13058 - $link_text = $matches[1];
13059 - $url = $matches[2];
13060 -
13061 - // Clean up any trailing punctuation from the URL
13062 - $url = rtrim($url, '.,;:!?');
13063 -
13064 - // Sanitize the link text and URL
13065 - $safe_text = esc_html($link_text);
13066 - $safe_url = esc_url($url);
13067 -
13068 - // Create the HTML link
13069 - return '<a href="' . $safe_url . '" target="_blank" rel="noopener noreferrer">' . $safe_text . '</a>';
13070 - }, $text);
13071 -
13072 - // If preg_replace_callback failed, return original text
13073 - if ($processed_text === null) {
13074 - return $text;
13075 - }
13076 -
13077 - return $processed_text;
13078 -}
13079 563
13080 -/**
13081 - * Auto-convert plain URLs to clickable links
13082 - *
13083 - * @param string $text The text to process
13084 - * @return string The text with URLs converted to links
13085 - */
13086 -private function auto_link_urls($text) {
13087 - // Return original text if empty
13088 - if (empty($text)) {
13089 - return $text;
13090 - }
13091 -
13092 - // Simple pattern that avoids complex lookbehinds
13093 - // This will match URLs that are not already inside href attributes or markdown links
13094 - $pattern = '/(?<!href=["\'])(?<!\]\()https?:\/\/[^\s<>"\')\]]+/i';
13095 -
13096 - $processed_text = preg_replace_callback($pattern, function($matches) {
13097 - $url = $matches[0];
13098 - // Clean up any trailing punctuation that might have been captured
13099 - $url = rtrim($url, '.,;:!?');
13100 -
13101 - // Add target="_blank" and rel="noopener noreferrer" for security
13102 - return '<a href="' . esc_url($url) . '" target="_blank" rel="noopener noreferrer">' . esc_html($url) . '</a>';
13103 - }, $text);
13104 -
13105 - // If preg_replace_callback failed, return original text
13106 - if ($processed_text === null) {
13107 - return $text;
13108 - }
13109 -
13110 - return $processed_text;
13111 -}
13112 -
13113 -
13114 -// Helper function to get client IP address
13115 -private function get_client_ip() {
13116 - // Check for shared internet/ISP IP
13117 - if (!empty($_SERVER['HTTP_CLIENT_IP'])) {
13118 - return sanitize_text_field($_SERVER['HTTP_CLIENT_IP']);
13119 - }
13120 -
13121 - // Check for IPs passing through proxies
13122 - if (!empty($_SERVER['HTTP_X_FORWARDED_FOR'])) {
13123 - // Use the first value in the comma-separated list
13124 - $forwarded_for = explode(',', sanitize_text_field($_SERVER['HTTP_X_FORWARDED_FOR']));
13125 - return trim($forwarded_for[0]);
13126 - }
13127 -
13128 - if (!empty($_SERVER['REMOTE_ADDR'])) {
13129 - return sanitize_text_field($_SERVER['REMOTE_ADDR']);
13130 - }
13131 -
13132 - // Fallback
13133 - return 'unknown';
13134 -}
13135 -
13136 -/**
13137 - * AJAX handler to get system information for testing panel
13138 - */
13139 -/**
13140 - * AJAX handler to get system information for testing panel
13141 - */
13142 -public function mxchat_get_system_info() {
13143 - // Verify nonce for security
13144 - if (!wp_verify_nonce($_POST['nonce'], 'mxchat_test_nonce')) {
13145 - wp_send_json_error(['message' => 'Invalid nonce']);
13146 - return;
13147 - }
13148 -
13149 - // Only allow admin users
13150 - if (!current_user_can('administrator')) {
13151 - wp_send_json_error(['message' => 'Unauthorized']);
13152 - return;
13153 - }
13154 -
13155 - // Get system prompt from options
13156 - $system_prompt = isset($this->options['system_prompt_instructions'])
13157 - ? $this->options['system_prompt_instructions']
13158 - : 'No system prompt configured';
13159 -
13160 - // Get selected model
13161 - $selected_model = isset($this->options['model']) ? $this->options['model'] : 'gpt-5.6-sol';
13162 -
13163 - // Check if OpenRouter is being used
13164 - $is_openrouter = ($selected_model === 'openrouter');
13165 - $openrouter_model = '';
13166 -
13167 - if ($is_openrouter) {
13168 - // Get the actual OpenRouter model that's selected
13169 - $openrouter_model = isset($this->options['openrouter_selected_model'])
13170 - ? $this->options['openrouter_selected_model']
13171 - : 'No OpenRouter model selected';
13172 -
13173 - // Update selected_model display to show both
13174 - $selected_model = 'OpenRouter: ' . $openrouter_model;
13175 - }
13176 -
13177 - // Get API key status (just check if they exist, don't expose the keys)
13178 - $api_status = [];
13179 - $api_status['openai'] = !empty($this->options['api_key']);
13180 - $api_status['claude'] = !empty($this->options['claude_api_key']);
13181 - $api_status['gemini'] = !empty($this->options['gemini_api_key']);
13182 - $api_status['xai'] = !empty($this->options['xai_api_key']);
13183 - $api_status['deepseek'] = !empty($this->options['deepseek_api_key']);
13184 - $api_status['openrouter'] = !empty($this->options['openrouter_api_key']);
13185 -
13186 - wp_send_json_success([
13187 - 'system_prompt' => $system_prompt,
13188 - 'selected_model' => $selected_model,
13189 - 'is_openrouter' => $is_openrouter,
13190 - 'openrouter_model' => $openrouter_model,
13191 - 'api_status' => $api_status
13192 - ]);
13193 -}
13194 -
13195 -/**
13196 - * AJAX handler to get similarity threshold
13197 - */
13198 -public function mxchat_get_similarity_threshold() {
13199 - // Verify nonce for security
13200 - if (!wp_verify_nonce($_POST['nonce'], 'mxchat_test_nonce')) {
13201 - wp_send_json_error(['message' => 'Invalid nonce']);
13202 - return;
13203 - }
13204 -
13205 - // Only allow admin users
13206 - if (!current_user_can('administrator')) {
13207 - wp_send_json_error(['message' => 'Unauthorized']);
13208 - return;
13209 - }
13210 -
13211 - // Get similarity threshold from main options (default 35%)
13212 - $similarity_threshold = isset($this->options['similarity_threshold'])
13213 - ? ((int) $this->options['similarity_threshold']) / 100
13214 - : 0.35;
13215 -
13216 - wp_send_json_success([
13217 - 'threshold' => $similarity_threshold,
13218 - 'threshold_percentage' => ($similarity_threshold * 100) . '%'
13219 - ]);
13220 -}
13221 -
13222 -/**
13223 - * AJAX handler to get knowledge base status
13224 - */
13225 -public function mxchat_get_kb_status() {
13226 - // Verify nonce for security
13227 - if (!wp_verify_nonce($_POST['nonce'], 'mxchat_test_nonce')) {
13228 - wp_send_json_error(['message' => 'Invalid nonce']);
13229 - return;
13230 - }
13231 -
13232 - // Only allow admin users
13233 - if (!current_user_can('administrator')) {
13234 - wp_send_json_error(['message' => 'Unauthorized']);
13235 - return;
13236 - }
13237 -
13238 - // Check OpenAI Vector Store first (takes priority)
13239 - $vectorstore_options = get_option('mxchat_openai_vectorstore_options', array());
13240 - $use_vectorstore = (isset($vectorstore_options['mxchat_use_openai_vectorstore']) && $vectorstore_options['mxchat_use_openai_vectorstore'] === '1');
13241 -
13242 - if ($use_vectorstore) {
13243 - $vectorstore_ids = $vectorstore_options['mxchat_openai_vectorstore_ids'] ?? '';
13244 - $id_count = !empty($vectorstore_ids) ? count(array_filter(array_map('trim', explode(',', $vectorstore_ids)))) : 0;
13245 -
13246 - $kb_info = [
13247 - 'type' => 'OpenAI Vector Store',
13248 - 'status' => 'Active',
13249 - 'documents' => $id_count > 0 ? $id_count . ' vector store' . ($id_count > 1 ? 's' : '') . ' configured' : 'No vector stores configured'
13250 - ];
13251 -
13252 - wp_send_json_success($kb_info);
13253 - return;
13254 - }
13255 -
13256 - // Check Pinecone vs WordPress
13257 - $addon_options = get_option('mxchat_pinecone_addon_options', array());
13258 - $use_pinecone = (isset($addon_options['mxchat_use_pinecone']) && $addon_options['mxchat_use_pinecone'] === '1');
13259 -
13260 - $kb_info = [
13261 - 'type' => $use_pinecone ? 'Pinecone' : 'WordPress Database',
13262 - 'status' => 'Active'
13263 - ];
13264 -
13265 - // Get document count
13266 - if ($use_pinecone) {
13267 - $kb_info['documents'] = 'Connected to Pinecone';
13268 - $kb_info['api_configured'] = !empty($addon_options['mxchat_pinecone_api_key']);
13269 - } else {
13270 - // Count documents in WordPress database
13271 - global $wpdb;
13272 - $table_name = $wpdb->prefix . 'mxchat_system_prompt_content';
13273 - $count = $wpdb->get_var("SELECT COUNT(*) FROM {$table_name}");
13274 - $kb_info['documents'] = $count ? $count . ' documents' : 'No documents';
13275 - }
13276 -
13277 - wp_send_json_success($kb_info);
13278 -}
13279 -
13280 -/**
13281 - * AJAX handler to start a completely fresh session (NEW - replaces old clear session)
13282 - */
13283 -public function mxchat_start_fresh_session() {
13284 - // Verify nonce for security
13285 - if (!wp_verify_nonce($_POST['nonce'], 'mxchat_test_nonce')) {
13286 - wp_send_json_error(['message' => 'Invalid nonce']);
13287 - return;
13288 - }
13289 -
13290 - // Only allow admin users
13291 - if (!current_user_can('administrator')) {
13292 - wp_send_json_error(['message' => 'Unauthorized']);
13293 - return;
13294 - }
13295 -
13296 - $old_session_id = isset($_POST['old_session_id']) ? MxChat_Utils::sanitize_session_id(wp_unslash($_POST['old_session_id'])) : '';
13297 - $new_session_id = isset($_POST['new_session_id']) ? MxChat_Utils::sanitize_session_id(wp_unslash($_POST['new_session_id'])) : '';
13298 -
13299 - if (empty($old_session_id)) {
13300 - wp_send_json_error(['message' => 'Old session ID required']);
13301 - return;
13302 - }
13303 -
13304 - // If no new session ID provided, generate one
13305 - if (empty($new_session_id)) {
13306 - // Cryptographically strong session id (plan-0c17b5). Prefix preserved
13307 - // exactly (other code pattern-matches on 'mxchat_chat_'). random_bytes
13308 - // is guaranteed on all supported PHP (7+).
13309 - $new_session_id = 'mxchat_chat_' . bin2hex(random_bytes(16));
13310 - }
13311 -
13312 - // Clear ALL data associated with the old session
13313 - $this->clear_complete_session_data($old_session_id);
13314 -
13315 - // Initialize the new session
13316 - $this->initialize_fresh_session($new_session_id);
13317 -
13318 - wp_send_json_success([
13319 - 'message' => 'Fresh session started successfully',
13320 - 'new_session_id' => $new_session_id,
13321 - 'old_session_id' => $old_session_id
13322 - ]);
13323 -}
13324 -
13325 -/**
13326 - * Clear ALL data associated with a session (ENHANCED)
13327 - */
13328 -private function clear_complete_session_data($session_id) {
13329 - // Clear chat history
13330 - delete_option("mxchat_history_{$session_id}");
13331 -
13332 - // Clear chat mode
13333 - delete_option("mxchat_mode_{$session_id}");
13334 -
13335 - // Clear any PDF/Word transients
13336 - $this->clear_pdf_transients($session_id);
13337 - if (method_exists($this, 'clear_word_transients')) {
13338 - $this->clear_word_transients($session_id);
13339 - }
13340 -
13341 - // Archive the session's per-conversation Slack channel before its option
13342 - // is deleted (plan 7458a7 — covers transcript-retention cleanup paths).
13343 - // Toggle-gated + shared-channel-guarded inside the helper; best-effort.
13344 - $stale_channel = get_option("mxchat_channel_{$session_id}", '');
13345 - if ($stale_channel !== '') {
13346 - $this->mxchat_maybe_archive_conversation_channel($session_id, $stale_channel);
13347 - }
13348 -
13349 - // Clear agent-related data
13350 - delete_option("mxchat_channel_{$session_id}");
13351 - delete_option("mxchat_thread_{$session_id}");
13352 - delete_option("mxchat_agent_name_{$session_id}");
13353 - delete_option("mxchat_email_{$session_id}");
13354 -
13355 - // Clear any recommendation flow state
13356 - delete_option("mxchat_sr_flow_state_{$session_id}");
13357 -
13358 - // Clear any cached embeddings or context
13359 - delete_transient("mxchat_context_{$session_id}");
13360 - delete_transient("mxchat_last_query_{$session_id}");
13361 -
13362 - // Clear any testing data
13363 - delete_transient("mxchat_testing_data_{$session_id}");
13364 -
13365 - // Clear any rate limiting data for this session
13366 - delete_transient("mxchat_rate_limit_{$session_id}");
13367 -
13368 - // Clear any other session-specific transients
13369 - delete_transient("mxchat_waiting_for_pdf_url_{$session_id}");
13370 - delete_transient("mxchat_include_pdf_in_context_{$session_id}");
13371 - delete_transient("mxchat_include_word_in_context_{$session_id}");
13372 -
13373 - // Clear form addon state (pending forms and submitted forms)
13374 - delete_option("mxchat_pending_form_{$session_id}");
13375 - delete_option("mxchat_submitted_forms_{$session_id}");
13376 -
13377 - //error_log("MxChat: Cleared all data for session: {$session_id}");
13378 -}
13379 -
13380 -/**
13381 - * Initialize a fresh session with default data
13382 - */
13383 -private function initialize_fresh_session($session_id) {
13384 - // Set default chat mode
13385 - update_option("mxchat_mode_{$session_id}", 'ai');
13386 -
13387 - //error_log("MxChat: Initialized fresh session: {$session_id}");
13388 -}
13389 -
13390 -/**
13391 - * Helper method to clear Word document transients (if you have Word support)
13392 - */
13393 -private function clear_word_transients($session_id) {
13394 - delete_transient('mxchat_word_url_' . $session_id);
13395 - delete_transient('mxchat_word_filename_' . $session_id);
13396 - delete_transient('mxchat_word_embeddings_' . $session_id);
13397 - delete_transient('mxchat_include_word_in_context_' . $session_id);
13398 -}
13399 -
13400 -/**
13401 - * Simplified testing data capture method (CLEANED UP)
13402 - */
13403 -private function capture_testing_data($user_embedding, $message, $session_id) {
13404 - // Only capture for admin users
13405 - if (!current_user_can('administrator')) {
13406 - return null;
13407 - }
13408 -
13409 - $testing_data = [
13410 - 'query' => $message,
13411 - 'timestamp' => time(),
13412 - 'top_matches' => [],
13413 - 'action_matches' => [] // Add action matches
13414 - ];
13415 -
13416 - // Get similarity threshold
13417 - $similarity_threshold = isset($this->options['similarity_threshold'])
13418 - ? ((int) $this->options['similarity_threshold']) / 100
13419 - : 0.35;
13420 -
13421 - $testing_data['similarity_threshold'] = $similarity_threshold;
13422 -
13423 - // Use the real similarity analysis if available
13424 - if ($this->last_similarity_analysis !== null) {
13425 - $testing_data['knowledge_base_type'] = $this->last_similarity_analysis['knowledge_base_type'];
13426 - $testing_data['top_matches'] = $this->last_similarity_analysis['top_matches'];
13427 - $testing_data['total_documents_checked'] = $this->last_similarity_analysis['total_checked'] ?? 0;
13428 - } else {
13429 - // Fallback: determine knowledge base type
13430 - $addon_options = get_option('mxchat_pinecone_addon_options', array());
13431 - $use_pinecone = (isset($addon_options['mxchat_use_pinecone']) && $addon_options['mxchat_use_pinecone'] === '1');
13432 -
13433 - $testing_data['knowledge_base_type'] = $use_pinecone ? 'Pinecone' : 'WordPress Database';
13434 - }
13435 -
13436 - // Include action analysis if available
13437 - if (isset($this->last_action_analysis) && !empty($this->last_action_analysis)) {
13438 - $testing_data['action_matches'] = $this->last_action_analysis;
13439 -
13440 - // Clear it after capturing to avoid stale data
13441 - $this->last_action_analysis = null;
13442 - }
13443 -
13444 - return $testing_data;
13445 -}
13446 -
13447 -
13448 -/**
13449 - * Track URL clicks from chatbot responses
13450 - */
13451 -public function mxchat_track_url_click() {
13452 - // Verify nonce for security
13453 - if (!isset($_POST['nonce']) || !MxChat_Integrator::mxchat_verify_chat_send_nonce($_POST['nonce'])) {
13454 - wp_send_json_error(['message' => 'Invalid nonce']);
13455 - wp_die();
13456 - }
13457 -
13458 - $session_id = isset($_POST['session_id']) ? MxChat_Utils::sanitize_session_id(wp_unslash($_POST['session_id'])) : '';
13459 - $clicked_url = isset($_POST['url']) ? esc_url_raw($_POST['url']) : '';
13460 - $message_context = isset($_POST['message_context']) ? sanitize_textarea_field($_POST['message_context']) : '';
13461 -
13462 - if (empty($session_id) || empty($clicked_url)) {
13463 - wp_send_json_error(['message' => 'Missing required data']);
13464 - wp_die();
13465 - }
13466 -
13467 - global $wpdb;
13468 - $table_name = $wpdb->prefix . 'mxchat_url_clicks';
13469 -
13470 - // Insert click tracking record
13471 - $wpdb->insert(
13472 - $table_name,
13473 - [
13474 - 'session_id' => $session_id,
13475 - 'clicked_url' => $clicked_url,
13476 - 'message_context' => $message_context,
13477 - 'click_timestamp' => current_time('mysql', 1),
13478 - 'user_ip' => $_SERVER['REMOTE_ADDR'],
13479 - 'user_agent' => $_SERVER['HTTP_USER_AGENT']
13480 - ]
13481 - );
13482 -
13483 - wp_send_json_success(['message' => 'Click tracked']);
13484 - wp_die();
13485 -}
13486 -
13487 -/**
13488 - * Get URL click analytics for a session
13489 - */
13490 -public function mxchat_get_url_clicks($session_id) {
13491 - global $wpdb;
13492 - $table_name = $wpdb->prefix . 'mxchat_url_clicks';
13493 -
13494 - $clicks = $wpdb->get_results($wpdb->prepare(
13495 - "SELECT * FROM $table_name WHERE session_id = %s ORDER BY click_timestamp ASC",
13496 - $session_id
13497 - ));
13498 -
13499 - return $clicks;
13500 -}
13501 -/**
13502 - * Track the originating page where chat was started
13503 - */
13504 -public function mxchat_track_originating_page() {
13505 - // Verify nonce
13506 - if (!isset($_POST['nonce']) || !MxChat_Integrator::mxchat_verify_chat_send_nonce($_POST['nonce'])) {
13507 - wp_send_json_error(['message' => 'Invalid nonce']);
13508 - wp_die();
13509 - }
13510 -
13511 - $session_id = isset($_POST['session_id']) ? MxChat_Utils::sanitize_session_id(wp_unslash($_POST['session_id'])) : '';
13512 - $page_url = isset($_POST['page_url']) ? esc_url_raw($_POST['page_url']) : '';
13513 - $page_title = isset($_POST['page_title']) ? sanitize_text_field($_POST['page_title']) : '';
13514 -
13515 - if (empty($session_id)) {
13516 - wp_send_json_error(['message' => 'Missing session ID']);
13517 - wp_die();
13518 - }
13519 -
13520 - global $wpdb;
13521 - $table_name = $wpdb->prefix . 'mxchat_chat_transcripts';
13522 -
13523 - // Check if we've already tracked for this session
13524 - $existing = $wpdb->get_var($wpdb->prepare(
13525 - "SELECT COUNT(*) FROM $table_name
13526 - WHERE session_id = %s
13527 - AND originating_page_url IS NOT NULL",
13528 - $session_id
13529 - ));
13530 -
13531 - if ($existing > 0) {
13532 - wp_send_json_success(['message' => 'Already tracked']);
13533 - wp_die();
13534 - }
13535 -
13536 - // Update the first message in this session with originating page info
13537 - $wpdb->query($wpdb->prepare(
13538 - "UPDATE $table_name
13539 - SET originating_page_url = %s,
13540 - originating_page_title = %s
13541 - WHERE session_id = %s
13542 - ORDER BY timestamp ASC
13543 - LIMIT 1",
13544 - $page_url,
13545 - $page_title,
13546 - $session_id
13547 - ));
13548 -
13549 - wp_send_json_success(['message' => 'Originating page tracked']);
13550 - wp_die();
13551 -}
13552 -
13553 -/**
13554 - * Validate and clean URLs from AI response
13555 - * Removes any URLs that aren't in the knowledge base
13556 - *
13557 - * @param string $response_text The AI-generated response
13558 - * @param array $valid_urls Array of URLs from the knowledge base
13559 - * @return string Cleaned response with invalid URLs removed/flagged
13560 - */
13561 -private function validate_and_clean_urls($response_text, $valid_urls, $session_id = null, $bot_id = null) {
13562 - /**
13563 - * Filter the list of URLs treated as valid (allowlisted) BEFORE the
13564 - * response URL sanitizer strips any link not in the list. Lets a site
13565 - * owner / developer whitelist links their custom function-calling tools
13566 - * return (e.g. session or speaker pages), which are otherwise absent from
13567 - * the RAG/system-prompt-derived list and get stripped to plain text.
13568 - *
13569 - * Purely additive: with no hook registered, apply_filters returns
13570 - * $valid_urls untouched, so there is zero behavior change for anyone who
13571 - * does not use the filter. Applied before the empty-check so a hooked
13572 - * allowlist can participate. (plan-mxchat-20260710-13a471)
13573 - *
13574 - * @param array $valid_urls URLs already known-valid (RAG + system prompt).
13575 - * @param string|null $session_id Current chat session id, if available.
13576 - * @param string|null $bot_id Current bot id, if available.
13577 - */
13578 - $valid_urls = apply_filters('mxchat_valid_urls', $valid_urls, $session_id, $bot_id);
13579 -
13580 - // A bad mu-plugin returning a non-array (or non-string entries) must never
13581 - // fatal the response path — coerce defensively before any use.
13582 - if (!is_array($valid_urls)) {
13583 - $valid_urls = array();
13584 - }
13585 - $valid_urls = array_values(array_filter($valid_urls, static function ($u) {
13586 - return is_string($u) && $u !== '';
13587 - }));
13588 -
13589 - // If no valid URLs provided or empty response, return as-is
13590 - if (empty($valid_urls) || empty($response_text)) {
13591 - //error_log("Validation skipped - empty valid_urls or response");
13592 - return $response_text;
13593 - }
13594 -
13595 - // Extract all URLs from the AI response
13596 - // This regex matches http:// and https:// URLs
13597 - preg_match_all(
13598 - '#\bhttps?://[^\s<>"\')\]]+#i',
13599 - $response_text,
13600 - $matches
13601 - );
13602 -
13603 - // If no URLs found in response, return as-is
13604 - if (empty($matches[0])) {
13605 - //error_log("No URLs found in response");
13606 - return $response_text;
13607 - }
13608 -
13609 - $found_urls = $matches[0];
13610 - $cleaned_response = $response_text;
13611 - $removed_count = 0;
13612 -
13613 - // Normalize valid URLs for comparison (remove trailing slashes, fragments, etc.)
13614 - $normalized_valid_urls = array_map(function($url) {
13615 - // Remove trailing slash
13616 - $url = rtrim($url, '/');
13617 - // Remove URL fragments (#section)
13618 - $url = preg_replace('/#.*$/', '', $url);
13619 - // Remove trailing punctuation that might have been captured
13620 - $url = rtrim($url, '.,;:!?');
13621 - return $url;
13622 - }, $valid_urls);
13623 -
13624 - //error_log("Normalized valid URLs: " . print_r($normalized_valid_urls, true));
13625 -
13626 - foreach ($found_urls as $found_url) {
13627 - // Clean up the found URL (remove trailing punctuation that might have been captured)
13628 - $clean_found_url = rtrim($found_url, '.,;:!?)');
13629 -
13630 - // DEBUG: Log each URL being checked
13631 - //error_log("Checking found URL: " . $found_url);
13632 -
13633 - // Normalize for comparison
13634 - $normalized_found = rtrim($clean_found_url, '/');
13635 - $normalized_found = preg_replace('/#.*$/', '', $normalized_found);
13636 -
13637 - //error_log("Normalized found URL: " . $normalized_found);
13638 -
13639 - // Check if this URL exists in our valid URLs list
13640 - $is_valid = false;
13641 -
13642 - //error_log("Starting validation checks for: " . $normalized_found);
13643 -
13644 - // First, try exact match
13645 - if (in_array($normalized_found, $normalized_valid_urls)) {
13646 - $is_valid = true;
13647 - //error_log("EXACT MATCH FOUND");
13648 - } else {
13649 - //error_log("No exact match, checking variations...");
13650 - // If no exact match, check if it's a variation (with query params, etc.)
13651 - foreach ($normalized_valid_urls as $valid_url) {
13652 - //error_log(" Comparing against valid URL: " . $valid_url);
13653 -
13654 - // Check if the found URL starts with a valid URL (handles query params)
13655 - if (strpos($normalized_found, $valid_url) === 0) {
13656 - // Check what comes after the valid URL
13657 - $remainder = substr($normalized_found, strlen($valid_url));
13658 -
13659 - // Only valid if:
13660 - // 1. Exact match (remainder is empty)
13661 - // 2. Query params (starts with ?)
13662 - // 3. Fragment (starts with #)
13663 - if (empty($remainder) || $remainder[0] === '?' || $remainder[0] === '#') {
13664 - $is_valid = true;
13665 - //error_log(" MATCH: Found URL is valid variation of base URL");
13666 - break;
13667 - } else {
13668 - //error_log(" NOT A MATCH: Found URL extends path beyond valid URL (remainder: " . $remainder . ")");
13669 - }
13670 - }
13671 - // Also check the reverse (in case valid URL has query params)
13672 - if (strpos($valid_url, $normalized_found) === 0) {
13673 - $is_valid = true;
13674 - //error_log(" MATCH: Valid URL starts with found URL");
13675 - break;
13676 - }
13677 - }
13678 -
13679 - if (!$is_valid) {
13680 - //error_log("NO MATCH FOUND - URL should be removed");
13681 - }
13682 - }
13683 -
13684 - // If URL is not valid, remove it from the response
13685 - if (!$is_valid) {
13686 - // Log the removal for debugging
13687 - //error_log("MxChat: Removed hallucinated URL: " . $found_url);
13688 - //error_log("MxChat: Valid URLs were: " . implode(', ', array_slice($normalized_valid_urls, 0, 5)));
13689 -
13690 - $removed_count++;
13691 -
13692 - // Check if URL is part of a markdown link: [text](url)
13693 - $markdown_pattern = '/\[([^\]]+)\]\(' . preg_quote($found_url, '/') . '\)/';
13694 - if (preg_match($markdown_pattern, $cleaned_response)) {
13695 - //error_log("Found markdown link, removing but keeping text");
13696 - // Remove the markdown link but keep the text
13697 - $cleaned_response = preg_replace($markdown_pattern, '$1', $cleaned_response);
13698 - }
13699 - // Check if URL is part of an HTML link: <a href="url">text</a>
13700 - else if (preg_match('/<a[^>]*href=["\']' . preg_quote($found_url, '/') . '["\'][^>]*>(.*?)<\/a>/i', $cleaned_response, $link_match)) {
13701 - //error_log("Found HTML link, removing but keeping text");
13702 - // Remove the HTML link but keep the text
13703 - $link_text = $link_match[1];
13704 - $cleaned_response = preg_replace(
13705 - '/<a[^>]*href=["\']' . preg_quote($found_url, '/') . '["\'][^>]*>.*?<\/a>/i',
13706 - $link_text,
13707 - $cleaned_response
13708 - );
13709 - }
13710 - // Otherwise just remove the bare URL
13711 - else {
13712 - //error_log("Removing bare URL");
13713 - $cleaned_response = str_replace($found_url, '', $cleaned_response);
13714 - }
13715 - }
13716 - }
13717 -
13718 - // Log summary if any URLs were removed
13719 - if ($removed_count > 0) {
13720 - //error_log("MxChat: URL Validation Summary - Removed {$removed_count} hallucinated URL(s)");
13721 - } else {
13722 - //error_log("MxChat: URL Validation Summary - No URLs removed, all were valid");
13723 - }
13724 -
13725 - // Clean up any double spaces or awkward punctuation left behind
13726 - // IMPORTANT: Only collapse horizontal whitespace (spaces/tabs), preserve newlines for markdown formatting
13727 - $cleaned_response = preg_replace('/[^\S\n]+/', ' ', $cleaned_response); // Collapse spaces/tabs but NOT newlines
13728 - $cleaned_response = preg_replace('/[^\S\n]+([.,;:!?])/', '$1', $cleaned_response); // Same for punctuation cleanup
13729 -
13730 - //error_log("Final cleaned response: " . $cleaned_response);
13731 -
13732 - return trim($cleaned_response);
13733 -}
13734 -
13735 -/**
13736 - * AJAX handler to get current chat mode for a session
13737 - */
13738 -public function mxchat_get_current_chat_mode() {
13739 - // Verify nonce for security
13740 - if (!isset($_POST['nonce']) || !MxChat_Integrator::mxchat_verify_chat_send_nonce($_POST['nonce'])) {
13741 - wp_send_json_error(['message' => 'Invalid nonce']);
13742 - wp_die();
13743 - }
13744 -
13745 - $session_id = isset($_POST['session_id']) ? MxChat_Utils::sanitize_session_id(wp_unslash($_POST['session_id'])) : '';
13746 -
13747 - if (empty($session_id)) {
13748 - wp_send_json_error(['message' => 'Session ID missing']);
13749 - wp_die();
13750 - }
13751 -
13752 - // Get the current chat mode for this session
13753 - $chat_mode = get_option("mxchat_mode_{$session_id}", 'ai');
13754 -
13755 - wp_send_json_success([
13756 - 'chat_mode' => $chat_mode
13757 - ]);
13758 - wp_die();
13759 -}
13760 564
13761 565
13762 566
13763 567 }