PluginProbe
MxChat – AI Chatbot & Content Generation for WordPress / 2.5.6
MxChat – AI Chatbot & Content Generation for WordPress v2.5.6
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 | admin/class-knowledge-manager.php +4950 -8837 3.2.16 → 2.5.6 View file →
@@ -1,8838 +1,4951 @@
1 -<?php
2 -/**
3 - * File: admin/class-knowledge-manager.php
4 - *
5 - * Handles all knowledge base content processing for MxChat
6 - * Including PDF, sitemap, content processing, and WordPress post management
7 - */
8 -if (!defined('ABSPATH')) {
9 - exit; // Exit if accessed directly
10 -}
11 -
12 -class MxChat_Knowledge_Manager {
13 -
14 - private $options;
15 -
16 - // Post IDs whose vectors were already deleted by mxchat_handle_status_transition this
17 - // request, so the transient-based branch in mxchat_handle_post_update can skip the
18 - // redundant (idempotent but network-visible) second deletion.
19 - private $transition_deleted_posts = array();
20 -
21 - /**
22 - * Constructor - Register hooks for content processing
23 - */
24 -public function __construct() {
25 - $this->options = get_option('mxchat_options', array());
26 - $this->mxchat_init_hooks();
27 -
28 - $this->mxchat_init_role_hooks();
29 -}
30 -
31 -/**
32 - * Initialize WordPress hooks for content processing
33 - *
34 - */
35 -private function mxchat_init_hooks() {
36 - // Admin post handlers for form submissions
37 - add_action('admin_post_mxchat_submit_content', array($this, 'mxchat_handle_content_submission'));
38 - add_action('admin_post_mxchat_submit_sitemap', array($this, 'mxchat_handle_sitemap_submission'));
39 - add_action('admin_post_mxchat_submit_pdf_file', array($this, 'mxchat_handle_pdf_file_submission'));
40 - add_action('admin_post_mxchat_submit_youtube', array($this, 'mxchat_handle_youtube_submission'));
41 - add_action('admin_post_mxchat_stop_processing', array($this, 'mxchat_stop_processing'));
42 -
43 - // AJAX handlers for real-time processing and status updates
44 - add_action('wp_ajax_mxchat_get_status_updates', array($this, 'mxchat_ajax_get_status_updates'));
45 - add_action('wp_ajax_mxchat_dismiss_completed_status', array($this, 'mxchat_ajax_dismiss_completed_status'));
46 - add_action('wp_ajax_mxchat_get_content_list', array($this, 'ajax_mxchat_get_content_list'));
47 - add_action('wp_ajax_mxchat_process_selected_content', array($this, 'ajax_mxchat_process_selected_content'));
48 - add_action('wp_ajax_mxchat_save_inline_prompt', array($this, 'mxchat_save_inline_prompt'));
49 - add_action('admin_post_mxchat_delete_pinecone_prompt', array($this, 'mxchat_handle_pinecone_prompt_delete'));
50 - add_action('wp_ajax_mxchat_delete_pinecone_prompt', array($this, 'ajax_mxchat_delete_pinecone_prompt'));
51 - add_action('wp_ajax_mxchat_delete_chunks_by_url', array($this, 'ajax_mxchat_delete_chunks_by_url'));
52 - add_action('wp_ajax_mxchat_delete_wordpress_prompt', array($this, 'ajax_mxchat_delete_wordpress_prompt'));
53 - add_action('wp_ajax_mxchat_bulk_delete_knowledge', array($this, 'ajax_mxchat_bulk_delete_knowledge'));
54 - add_action('wp_ajax_mxchat_update_role_restriction', array($this, 'ajax_mxchat_update_role_restriction'));
55 -
56 - // Queue-based processing AJAX handlers
57 - add_action('wp_ajax_mxchat_get_next_queue_item', array($this, 'ajax_mxchat_get_next_queue_item'));
58 - add_action('wp_ajax_mxchat_process_queue_item', array($this, 'ajax_mxchat_process_queue_item'));
59 - add_action('wp_ajax_mxchat_get_queue_status', array($this, 'ajax_mxchat_get_queue_status'));
60 - add_action('wp_ajax_mxchat_clear_queue', array($this, 'ajax_mxchat_clear_queue'));
61 - add_action('wp_ajax_mxchat_retry_failed', array($this, 'ajax_mxchat_retry_failed'));
62 - add_action('wp_ajax_mxchat_get_recent_entries', array($this, 'ajax_mxchat_get_recent_entries'));
63 - add_action('wp_ajax_mxchat_detect_sitemaps', array($this, 'ajax_mxchat_detect_sitemaps'));
64 - add_action('wp_ajax_mxchat_refresh_pinecone_entries', array($this, 'ajax_mxchat_refresh_pinecone_entries'));
65 - add_action('wp_ajax_mxchat_paginate_entries', array($this, 'ajax_mxchat_paginate_entries'));
66 - add_action('wp_ajax_mxchat_get_entry_content', array($this, 'ajax_mxchat_get_entry_content'));
67 - add_action('wp_ajax_mxchat_save_entry_content', array($this, 'ajax_mxchat_save_entry_content'));
68 - add_action('wp_ajax_mxchat_inspect_entry', array($this, 'ajax_mxchat_inspect_entry'));
69 -
70 - // WordPress post management hooks
71 - add_action('pre_post_update', array($this, 'mxchat_store_pre_update_status'), 10, 2);
72 - add_action('post_updated', array($this, 'mxchat_handle_post_update'), 10, 3);
73 - add_action('before_delete_post', array($this, 'mxchat_handle_post_delete'));
74 - add_action('wp_trash_post', array($this, 'mxchat_handle_post_delete'));
75 - // Authoritative unpublish detection: core hands this hook the REAL previous status, so
76 - // removal no longer depends on the mxchat_prev_status_* transients (evictable by persistent
77 - // object caches, never written by paths that bypass wp_update_post, e.g. plugins flipping
78 - // post_status directly and calling wp_transition_post_status themselves).
79 - add_action('transition_post_status', array($this, 'mxchat_handle_status_transition'), 10, 3);
80 -
81 - // ACF hook - fires AFTER ACF fields are saved, ensuring ACF data is available
82 - // Priority 20 to run after ACF's own save (which runs at priority 10)
83 - add_action('acf/save_post', array($this, 'mxchat_handle_acf_save'), 20);
84 -
85 - // One-time cleanup for vectors orphaned by unpublishes that predate the
86 - // transition_post_status handler (plan 816fb1): wp mxchat prune-unpublished
87 - if (defined('WP_CLI') && WP_CLI) {
88 - WP_CLI::add_command('mxchat prune-unpublished', array($this, 'cli_prune_unpublished'));
89 - }
90 -
91 - add_action('wp_ajax_mxchat_mark_queue_complete', array($this, 'ajax_mxchat_mark_queue_complete'));
92 -
93 - // WooCommerce product hooks (if WooCommerce is active)
94 - if (class_exists('WooCommerce')) {
95 - add_action('pre_post_update', array($this, 'mxchat_store_pre_update_status'), 10, 2);
96 - add_action('save_post_product', array($this, 'mxchat_handle_product_change'), 10, 3);
97 - add_action('wp_trash_post', array($this, 'mxchat_handle_product_delete'));
98 - add_action('before_delete_post', array($this, 'mxchat_handle_product_delete'));
99 - }
100 -}
101 -
102 - /**
103 - * Get current options (refreshed)
104 - */
105 - private function mxchat_get_options() {
106 - if (empty($this->options)) {
107 - $this->options = get_option('mxchat_options', array());
108 - }
109 - return $this->options;
110 - }
111 -
112 -
113 - // ========================================
114 - // MAIN CONTENT SUBMISSION HANDLERS
115 - // ========================================
116 -
117 -public function mxchat_handle_content_submission() {
118 - // Check if the form was submitted and the user has permission.
119 - if (!isset($_POST['submit_content']) || !current_user_can('manage_options')) {
120 - return;
121 - }
122 -
123 - // Verify the nonce.
124 - $nonce = isset($_POST['mxchat_submit_content_nonce']) ? sanitize_text_field(wp_unslash($_POST['mxchat_submit_content_nonce'])) : '';
125 - if (!wp_verify_nonce($nonce, 'mxchat_submit_content_action')) {
126 - wp_die(esc_html__('Nonce verification failed.', 'mxchat'));
127 - }
128 -
129 - // Sanitize the inputs.
130 - // Use wp_kses_post to allow safe HTML and wp_unslash to remove WordPress added slashes
131 - $article_content = wp_kses_post(wp_unslash($_POST['article_content']));
132 - $article_url = isset($_POST['article_url']) ? esc_url_raw($_POST['article_url']) : '';
133 -
134 - // Get bot_id from form submission
135 - $bot_id = isset($_POST['bot_id']) ? sanitize_key($_POST['bot_id']) : 'default';
136 -
137 - // Get bot-specific options and API key
138 - $bot_options = $this->get_bot_options($bot_id);
139 - $options = !empty($bot_options) ? $bot_options : get_option('mxchat_options');
140 - $selected_model = $options['embedding_model'] ?? 'text-embedding-ada-002';
141 -
142 - if (strpos($selected_model, 'voyage') === 0) {
143 - $api_key = $options['voyage_api_key'] ?? '';
144 - } elseif (strpos($selected_model, 'gemini-embedding') === 0) {
145 - $api_key = $options['gemini_api_key'] ?? '';
146 - } else {
147 - $api_key = $options['api_key'] ?? '';
148 - }
149 -
150 - if (empty($api_key)) {
151 - set_transient('mxchat_admin_notice_error',
152 - esc_html__('API key is not configured. Please add your API key in the settings before submitting content.', 'mxchat'),
153 - 30
154 - );
155 - wp_safe_redirect(esc_url(admin_url('admin.php?page=mxchat-prompts')));
156 - exit;
157 - }
158 -
159 - // Use centralized utility function with bot_id
160 - $result = MxChat_Utils::submit_content_to_db($article_content, $article_url, $api_key, null, $bot_id);
161 -
162 - if (is_wp_error($result)) {
163 - set_transient('mxchat_admin_notice_error',
164 - esc_html__('Error storing content: ', 'mxchat') . $result->get_error_message(),
165 - 30
166 - );
167 - } else {
168 - set_transient('mxchat_admin_notice_success',
169 - esc_html__('Content successfully submitted!', 'mxchat'),
170 - 30
171 - );
172 - }
173 -
174 - wp_safe_redirect(esc_url(admin_url('admin.php?page=mxchat-prompts')));
175 - exit;
176 -}
177 -
178 -/**
179 - * Handle the "YouTube" KB import source (admin-post form submission).
180 - *
181 - * Per-video description mode:
182 - * - auto: fetch oEmbed metadata (reliable) + best-effort captions transcript.
183 - * If no usable transcript, index the metadata anyway, tell the admin,
184 - * and bounce back with the manual box pre-filled (never fail silently).
185 - * - manual: the admin's own description is what gets indexed; metadata rides along.
186 - *
187 - * The row is stored with content_type 'youtube' and source_url = the canonical
188 - * watch URL, so re-importing the same video UPDATES the entry (source_url
189 - * duplicate handling in MxChat_Utils::store_in_wordpress_db) — that is also the
190 - * "augment a metadata-only entry" path.
191 - */
192 -public function mxchat_handle_youtube_submission() {
193 - if (!isset($_POST['submit_youtube']) || !current_user_can('manage_options')) {
194 - wp_die(esc_html__('Unauthorized access', 'mxchat'));
195 - }
196 -
197 - check_admin_referer('mxchat_submit_youtube_action', 'mxchat_submit_youtube_nonce');
198 -
199 - $redirect_url = admin_url('admin.php?page=mxchat-prompts');
200 -
201 - $youtube_url = isset($_POST['youtube_url']) ? esc_url_raw(wp_unslash($_POST['youtube_url'])) : '';
202 - $video_id = MxChat_Utils::parse_youtube_id($youtube_url);
203 -
204 - if (empty($video_id)) {
205 - set_transient('mxchat_admin_notice_error',
206 - esc_html__('That does not look like a link to a single YouTube video. Please paste a watch, youtu.be, or Shorts URL.', 'mxchat'),
207 - 30
208 - );
209 - wp_safe_redirect(esc_url($redirect_url));
210 - exit;
211 - }
212 -
213 - $canonical_url = 'https://www.youtube.com/watch?v=' . $video_id;
214 -
215 - $description_mode = (isset($_POST['youtube_description_mode']) && $_POST['youtube_description_mode'] === 'manual') ? 'manual' : 'auto';
216 - $manual_description = isset($_POST['youtube_description']) ? trim(wp_kses_post(wp_unslash($_POST['youtube_description']))) : '';
217 -
218 - $bot_id = isset($_POST['bot_id']) ? sanitize_key($_POST['bot_id']) : 'default';
219 -
220 - // Resolve the embedding API key exactly like the sibling handlers.
221 - $bot_options = $this->get_bot_options($bot_id);
222 - $options = !empty($bot_options) ? $bot_options : get_option('mxchat_options');
223 - $selected_model = $options['embedding_model'] ?? 'text-embedding-ada-002';
224 -
225 - if (strpos($selected_model, 'voyage') === 0) {
226 - $api_key = $options['voyage_api_key'] ?? '';
227 - $provider_name = 'Voyage AI';
228 - } elseif (strpos($selected_model, 'gemini-embedding') === 0) {
229 - $api_key = $options['gemini_api_key'] ?? '';
230 - $provider_name = 'Google Gemini';
231 - } else {
232 - $api_key = $options['api_key'] ?? '';
233 - $provider_name = 'OpenAI';
234 - }
235 -
236 - if (empty($api_key)) {
237 - set_transient('mxchat_admin_notice_error',
238 - sprintf(
239 - esc_html__('%s API key is not configured. Please add your API key in the settings before submitting content.', 'mxchat'),
240 - $provider_name
241 - ),
242 - 30
243 - );
244 - wp_safe_redirect(esc_url($redirect_url));
245 - exit;
246 - }
247 -
248 - // Metadata is fetched in BOTH modes — it is the reliable half of auto, and in
249 - // manual mode it enriches the indexed text with the real title/channel.
250 - $meta = $this->mxchat_fetch_youtube_oembed($video_id);
251 - $video_title = isset($meta['title']) ? sanitize_text_field($meta['title']) : '';
252 - $video_channel = isset($meta['author_name']) ? sanitize_text_field($meta['author_name']) : '';
253 -
254 - $header_lines = 'YouTube Video: ' . ($video_title !== '' ? $video_title : $canonical_url) . "\n";
255 - if ($video_channel !== '') {
256 - $header_lines .= 'Channel: ' . $video_channel . "\n";
257 - }
258 - $header_lines .= 'URL: ' . $canonical_url . "\n\n";
259 -
260 - $transcript_missing = false;
261 -
262 - if ($description_mode === 'manual') {
263 - if ($manual_description === '') {
264 - set_transient('mxchat_admin_notice_error',
265 - esc_html__('Please write a description for the video, or switch to auto-fetch.', 'mxchat'),
266 - 30
267 - );
268 - wp_safe_redirect(esc_url($redirect_url));
269 - exit;
270 - }
271 - $indexed_text = $header_lines . $manual_description;
272 - } else {
273 - $transcript = $this->mxchat_fetch_youtube_transcript($video_id);
274 -
275 - if (strlen($transcript) >= 200) {
276 - $indexed_text = $header_lines . $transcript;
277 - } else {
278 - // Graceful fallback: captions disabled / blocked / no speech. Auto
279 - // reliably gets metadata; it does NOT guarantee a transcript.
280 - $transcript_missing = true;
281 -
282 - if ($video_title === '' && $video_channel === '') {
283 - // Both halves failed — nothing meaningful to index.
284 - set_transient('mxchat_admin_notice_error',
285 - esc_html__('Could not retrieve any information for that video (no metadata and no captions). Please check the URL, or use the manual description option.', 'mxchat'),
286 - 30
287 - );
288 - wp_safe_redirect(esc_url($redirect_url));
289 - exit;
290 - }
291 -
292 - $indexed_text = $header_lines . sprintf(
293 - /* translators: 1: video title, 2: channel name */
294 - __('A YouTube video titled "%1$s" from the channel %2$s.', 'mxchat'),
295 - $video_title !== '' ? $video_title : $canonical_url,
296 - $video_channel !== '' ? $video_channel : 'YouTube'
297 - );
298 - }
299 - }
300 -
301 - $result = MxChat_Utils::submit_content_to_db($indexed_text, $canonical_url, $api_key, null, $bot_id, 'youtube');
302 -
303 - if (is_wp_error($result)) {
304 - set_transient('mxchat_admin_notice_error',
305 - esc_html__('Error storing video in the knowledge base: ', 'mxchat') . $result->get_error_message(),
306 - 30
307 - );
308 - wp_safe_redirect(esc_url($redirect_url));
309 - exit;
310 - }
311 -
312 - if ($transcript_missing) {
313 - set_transient('mxchat_admin_notice_success',
314 - esc_html__('Video indexed from its title and channel — no captions were available for a transcript. The form below is pre-filled: write your own description and import again to improve matching (it updates the same entry).', 'mxchat'),
315 - 30
316 - );
317 - // Bounce back with prefill args so the page reopens the YouTube form in
318 - // manual mode with the URL + fetched title ready to augment.
319 - $redirect_url = add_query_arg(array(
320 - 'mxchat_yt_prefill' => '1',
321 - 'yt_url' => rawurlencode($canonical_url),
322 - 'yt_title' => rawurlencode($video_title),
323 - ), $redirect_url);
324 - } else {
325 - set_transient('mxchat_admin_notice_success',
326 - esc_html__('YouTube video successfully added to the knowledge base!', 'mxchat'),
327 - 30
328 - );
329 - }
330 -
331 - wp_safe_redirect(esc_url_raw($redirect_url));
332 - exit;
333 -}
334 -
335 -/**
336 - * Fetch YouTube oEmbed metadata for a video (no API key required).
337 - * Returns the decoded array (title, author_name, thumbnail_url, ...) or array().
338 - */
339 -private function mxchat_fetch_youtube_oembed($video_id) {
340 - $oembed_url = 'https://www.youtube.com/oembed?url=' . rawurlencode('https://www.youtube.com/watch?v=' . $video_id) . '&format=json';
341 - $response = wp_remote_get($oembed_url, array('timeout' => 15));
342 - if (is_wp_error($response) || wp_remote_retrieve_response_code($response) !== 200) {
343 - return array();
344 - }
345 - $data = json_decode(wp_remote_retrieve_body($response), true);
346 - return is_array($data) ? $data : array();
347 -}
348 -
349 -/**
350 - * Best-effort captions transcript for a video. Deliberately ISOLATED: this uses
351 - * YouTube's unofficial timedtext route (the caption track list embedded in the
352 - * watch page), which YouTube has broken before and will break again. Every
353 - * failure mode returns '' so a break degrades to the metadata-only import path
354 - * instead of erroring the whole submission. Do not let anything in here throw.
355 - */
356 -private function mxchat_fetch_youtube_transcript($video_id) {
357 - $watch_url = 'https://www.youtube.com/watch?v=' . $video_id . '&hl=en';
358 -
359 - // First try the honest ingest UA; some responses omit the player config for
360 - // bot UAs, so retry once with a browser UA before giving up.
361 - $user_agents = array(
362 - mxchat_ingest_user_agent(),
363 - 'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/126.0.0.0 Safari/537.36',
364 - );
365 -
366 - $tracks = array();
367 - foreach ($user_agents as $ua) {
368 - $response = wp_remote_get($watch_url, array(
369 - 'timeout' => 20,
370 - 'user-agent' => $ua,
371 - ));
372 - if (is_wp_error($response) || wp_remote_retrieve_response_code($response) !== 200) {
373 - continue;
374 - }
375 - $body = wp_remote_retrieve_body($response);
376 - if (!is_string($body) || $body === '' || !preg_match('/"captionTracks":(\[.*?\])(?=,")/s', $body, $m)) {
377 - continue;
378 - }
379 - $decoded = json_decode($m[1], true);
380 - if (is_array($decoded) && !empty($decoded)) {
381 - $tracks = $decoded;
382 - break;
383 - }
384 - }
385 -
386 - if (empty($tracks)) {
387 - return '';
388 - }
389 -
390 - // Prefer an English track, else take the first offered.
391 - $chosen = null;
392 - foreach ($tracks as $track) {
393 - if (isset($track['languageCode']) && strpos($track['languageCode'], 'en') === 0) {
394 - $chosen = $track;
395 - break;
396 - }
397 - }
398 - if ($chosen === null) {
399 - $chosen = $tracks[0];
400 - }
401 - if (empty($chosen['baseUrl']) || !is_string($chosen['baseUrl'])) {
402 - return '';
403 - }
404 -
405 - $timedtext = wp_remote_get($chosen['baseUrl'], array('timeout' => 20));
406 - if (is_wp_error($timedtext) || wp_remote_retrieve_response_code($timedtext) !== 200) {
407 - return '';
408 - }
409 - $xml = wp_remote_retrieve_body($timedtext);
410 - if (!is_string($xml) || strpos($xml, '<text') === false) {
411 - return '';
412 - }
413 -
414 - // <text start=".." dur="..">caption</text> — strip tags, decode the
415 - // double-encoded entities timedtext ships, collapse whitespace.
416 - $text = preg_replace('/<[^>]+>/', ' ', $xml);
417 - $text = html_entity_decode(html_entity_decode($text, ENT_QUOTES | ENT_HTML5, 'UTF-8'), ENT_QUOTES | ENT_HTML5, 'UTF-8');
418 - $text = trim(preg_replace('/\s+/u', ' ', $text));
419 -
420 - return $text;
421 -}
422 -
423 -public function mxchat_is_pdf_url($url, $response) {
424 - $content_type = wp_remote_retrieve_header($response, 'content-type');
425 - $file_extension = strtolower(pathinfo($url, PATHINFO_EXTENSION));
426 -
427 - // Check Content-Disposition header for .pdf filename (Google Drive sends this)
428 - $disposition = wp_remote_retrieve_header($response, 'content-disposition');
429 - $has_pdf_disposition = ! empty($disposition) && stripos($disposition, '.pdf') !== false;
430 -
431 - return strpos($content_type, 'pdf') !== false || $file_extension === 'pdf' || $has_pdf_disposition;
432 -}
433 -
434 -
435 -public function mxchat_handle_pdf_for_knowledge_base($pdf_url, $response, $bot_id = 'default') {
436 - if (!current_user_can('manage_options')) {
437 - return false;
438 - }
439 -
440 - $pdf_url = esc_url_raw($pdf_url);
441 - $upload_dir = wp_upload_dir();
442 -
443 - if (isset($upload_dir['error']) && $upload_dir['error'] !== false) {
444 - return false;
445 - }
446 -
447 - $pdf_filename = sanitize_file_name('mxchat_kb_' . time() . '.pdf');
448 - $pdf_path = trailingslashit($upload_dir['path']) . $pdf_filename;
449 -
450 - $response_body = wp_remote_retrieve_body($response);
451 - if (empty($response_body)) {
452 - return false;
453 - }
454 -
455 - if (!wp_mkdir_p(dirname($pdf_path))) {
456 - return false;
457 - }
458 -
459 - try {
460 - file_put_contents($pdf_path, $response_body);
461 -
462 - if (!file_exists($pdf_path)) {
463 - throw new Exception(__('Failed to save PDF file', 'mxchat'));
464 - }
465 -
466 - $total_pages = $this->mxchat_validate_and_count_pdf_pages($pdf_path);
467 -
468 - if ($total_pages === false || $total_pages < 1) {
469 - throw new Exception(__('Invalid PDF: Unable to parse or no pages found', 'mxchat'));
470 - }
471 -
472 - // Create unique queue ID
473 - $queue_id = 'pdf_' . md5($pdf_url . time());
474 -
475 - // Create array of pages to process
476 - $pages = array();
477 - for ($i = 1; $i <= $total_pages; $i++) {
478 - $pages[] = array(
479 - 'pdf_path' => $pdf_path,
480 - 'pdf_url' => $pdf_url,
481 - 'page_number' => $i,
482 - 'total_pages' => $total_pages
483 - );
484 - }
485 -
486 - // Add pages to queue
487 - $queued_count = $this->mxchat_add_to_queue($queue_id, 'pdf_page', $pages, $bot_id);
488 -
489 - if ($queued_count === 0) {
490 - wp_delete_file($pdf_path);
491 - throw new Exception(__('Failed to add PDF pages to processing queue', 'mxchat'));
492 - }
493 -
494 - // Store queue metadata
495 - $this->mxchat_set_queue_meta($queue_id, 'source_url', $pdf_url);
496 - $this->mxchat_set_queue_meta($queue_id, 'queue_type', 'pdf');
497 - $this->mxchat_set_queue_meta($queue_id, 'total_items', $total_pages);
498 - $this->mxchat_set_queue_meta($queue_id, 'bot_id', $bot_id);
499 - $this->mxchat_set_queue_meta($queue_id, 'pdf_path', $pdf_path);
500 - $this->mxchat_set_queue_meta($queue_id, 'created_at', current_time('mysql'));
501 -
502 - // Store queue ID in transient for status tracking
503 - set_transient('mxchat_active_queue_pdf', $queue_id, DAY_IN_SECONDS);
504 - set_transient('mxchat_last_pdf_url', $pdf_url, DAY_IN_SECONDS);
505 -
506 - return 'queued';
507 -
508 - } catch (Exception $e) {
509 - if (file_exists($pdf_path)) {
510 - wp_delete_file($pdf_path);
511 - }
512 - return $e->getMessage();
513 - }
514 -}
515 -
516 -/**
517 - * Handle direct PDF file upload from the knowledge base page
518 - */
519 -public function mxchat_handle_pdf_file_submission() {
520 - if (!isset($_POST['submit_pdf_file']) || !current_user_can('manage_options')) {
521 - wp_die(esc_html__('Unauthorized access', 'mxchat'));
522 - }
523 -
524 - check_admin_referer('mxchat_submit_pdf_file_action', 'mxchat_submit_pdf_file_nonce');
525 -
526 - $redirect_url = admin_url('admin.php?page=mxchat-prompts');
527 -
528 - // Validate file upload
529 - if (empty($_FILES['pdf_file']) || $_FILES['pdf_file']['error'] !== UPLOAD_ERR_OK) {
530 - $error_code = isset($_FILES['pdf_file']['error']) ? $_FILES['pdf_file']['error'] : UPLOAD_ERR_NO_FILE;
531 - $error_messages = array(
532 - UPLOAD_ERR_INI_SIZE => __('The uploaded file exceeds the server upload_max_filesize limit.', 'mxchat'),
533 - UPLOAD_ERR_FORM_SIZE => __('The uploaded file exceeds the form MAX_FILE_SIZE limit.', 'mxchat'),
534 - UPLOAD_ERR_PARTIAL => __('The file was only partially uploaded.', 'mxchat'),
535 - UPLOAD_ERR_NO_FILE => __('No file was uploaded. Please select a PDF file.', 'mxchat'),
536 - UPLOAD_ERR_NO_TMP_DIR => __('Server missing temporary folder.', 'mxchat'),
537 - UPLOAD_ERR_CANT_WRITE => __('Server failed to write file to disk.', 'mxchat'),
538 - );
539 - $error_msg = isset($error_messages[$error_code]) ? $error_messages[$error_code] : __('Unknown upload error.', 'mxchat');
540 - set_transient('mxchat_admin_notice_error', $error_msg, 30);
541 - wp_safe_redirect(esc_url($redirect_url));
542 - exit;
543 - }
544 -
545 - $file = $_FILES['pdf_file'];
546 -
547 - // Validate MIME type
548 - $finfo = finfo_open(FILEINFO_MIME_TYPE);
549 - $mime_type = finfo_file($finfo, $file['tmp_name']);
550 - finfo_close($finfo);
551 -
552 - if ($mime_type !== 'application/pdf') {
553 - set_transient('mxchat_admin_notice_error',
554 - esc_html__('Invalid file type. Only PDF files are accepted.', 'mxchat'),
555 - 30
556 - );
557 - wp_safe_redirect(esc_url($redirect_url));
558 - exit;
559 - }
560 -
561 - // Validate extension
562 - $ext = strtolower(pathinfo($file['name'], PATHINFO_EXTENSION));
563 - if ($ext !== 'pdf') {
564 - set_transient('mxchat_admin_notice_error',
565 - esc_html__('Invalid file extension. Only .pdf files are accepted.', 'mxchat'),
566 - 30
567 - );
568 - wp_safe_redirect(esc_url($redirect_url));
569 - exit;
570 - }
571 -
572 - $bot_id = isset($_POST['bot_id']) ? sanitize_key($_POST['bot_id']) : 'default';
573 - $original_filename = sanitize_file_name($file['name']);
574 -
575 - $upload_dir = wp_upload_dir();
576 - if (isset($upload_dir['error']) && $upload_dir['error'] !== false) {
577 - set_transient('mxchat_admin_notice_error',
578 - esc_html__('WordPress upload directory is not writable.', 'mxchat'),
579 - 30
580 - );
581 - wp_safe_redirect(esc_url($redirect_url));
582 - exit;
583 - }
584 -
585 - $pdf_filename = sanitize_file_name('mxchat_kb_' . time() . '.pdf');
586 - $pdf_path = trailingslashit($upload_dir['path']) . $pdf_filename;
587 -
588 - if (!wp_mkdir_p(dirname($pdf_path))) {
589 - set_transient('mxchat_admin_notice_error',
590 - esc_html__('Failed to create upload directory.', 'mxchat'),
591 - 30
592 - );
593 - wp_safe_redirect(esc_url($redirect_url));
594 - exit;
595 - }
596 -
597 - // Move uploaded file
598 - if (!move_uploaded_file($file['tmp_name'], $pdf_path)) {
599 - set_transient('mxchat_admin_notice_error',
600 - esc_html__('Failed to save uploaded PDF file.', 'mxchat'),
601 - 30
602 - );
603 - wp_safe_redirect(esc_url($redirect_url));
604 - exit;
605 - }
606 -
607 - try {
608 - $total_pages = $this->mxchat_validate_and_count_pdf_pages($pdf_path);
609 -
610 - if ($total_pages === false || $total_pages < 1) {
611 - throw new Exception(__('Invalid PDF: Unable to parse or no pages found', 'mxchat'));
612 - }
613 -
614 - // Use original filename as the source identifier
615 - $source_label = 'upload://' . $original_filename;
616 -
617 - $queue_id = 'pdf_' . md5($source_label . time());
618 -
619 - $pages = array();
620 - for ($i = 1; $i <= $total_pages; $i++) {
621 - $pages[] = array(
622 - 'pdf_path' => $pdf_path,
623 - 'pdf_url' => $source_label,
624 - 'page_number' => $i,
625 - 'total_pages' => $total_pages,
626 - );
627 - }
628 -
629 - $queued_count = $this->mxchat_add_to_queue($queue_id, 'pdf_page', $pages, $bot_id);
630 -
631 - if ($queued_count === 0) {
632 - wp_delete_file($pdf_path);
633 - throw new Exception(__('Failed to add PDF pages to processing queue', 'mxchat'));
634 - }
635 -
636 - $this->mxchat_set_queue_meta($queue_id, 'source_url', $source_label);
637 - $this->mxchat_set_queue_meta($queue_id, 'queue_type', 'pdf');
638 - $this->mxchat_set_queue_meta($queue_id, 'total_items', $total_pages);
639 - $this->mxchat_set_queue_meta($queue_id, 'bot_id', $bot_id);
640 - $this->mxchat_set_queue_meta($queue_id, 'pdf_path', $pdf_path);
641 - $this->mxchat_set_queue_meta($queue_id, 'created_at', current_time('mysql'));
642 -
643 - set_transient('mxchat_active_queue_pdf', $queue_id, DAY_IN_SECONDS);
644 - set_transient('mxchat_last_pdf_url', $source_label, DAY_IN_SECONDS);
645 -
646 - set_transient('mxchat_admin_notice_success',
647 - sprintf(
648 - esc_html__('PDF "%s" (%d pages) queued for processing. Processing will start automatically.', 'mxchat'),
649 - esc_html($original_filename),
650 - $total_pages
651 - ),
652 - 30
653 - );
654 -
655 - } catch (Exception $e) {
656 - if (file_exists($pdf_path)) {
657 - wp_delete_file($pdf_path);
658 - }
659 - set_transient('mxchat_admin_notice_error',
660 - esc_html__('Failed to process uploaded PDF: ', 'mxchat') . esc_html($e->getMessage()),
661 - 30
662 - );
663 - }
664 -
665 - wp_safe_redirect(esc_url($redirect_url));
666 - exit;
667 -}
668 -
669 -/**
670 - * Validate PDF and count pages with multiple parser attempts
671 - */
672 -private function mxchat_validate_and_count_pdf_pages($pdf_path) {
673 - // Method 1: Try with Smalot PDF Parser (your current method)
674 - try {
675 - mxchat_load_pdf_parser();
676 - $parser = new \Smalot\PdfParser\Parser();
677 - $pdf = $parser->parseFile($pdf_path);
678 - $pages = $pdf->getPages();
679 - $page_count = count($pages);
680 -
681 - if ($page_count > 0) {
682 - //error_log('PDF parsed successfully with Smalot parser: ' . $page_count . ' pages');
683 - return $page_count;
684 - }
685 - } catch (Exception $e) {
686 - //error_log('Smalot PDF parser failed: ' . $e->getMessage());
687 - }
688 -
689 - // Method 2: Try with pdfinfo command (if available)
690 - if (function_exists('shell_exec') && !$this->mxchat_is_shell_disabled()) {
691 - try {
692 - $command = 'pdfinfo ' . escapeshellarg($pdf_path) . ' 2>&1';
693 - $output = shell_exec($command);
694 -
695 - if ($output && preg_match('/Pages:\s*(\d+)/', $output, $matches)) {
696 - $page_count = intval($matches[1]);
697 - if ($page_count > 0) {
698 - //error_log('PDF parsed successfully with pdfinfo: ' . $page_count . ' pages');
699 - return $page_count;
700 - }
701 - }
702 - } catch (Exception $e) {
703 - //error_log('pdfinfo command failed: ' . $e->getMessage());
704 - }
705 - }
706 -
707 - // Method 3: Try to repair PDF and parse again
708 - try {
709 - $repaired_path = $this->mxchat_attempt_pdf_repair($pdf_path);
710 - if ($repaired_path && $repaired_path !== $pdf_path) {
711 - mxchat_load_pdf_parser();
712 - $parser = new \Smalot\PdfParser\Parser();
713 - $pdf = $parser->parseFile($repaired_path);
714 - $pages = $pdf->getPages();
715 - $page_count = count($pages);
716 -
717 - if ($page_count > 0) {
718 - // Replace original with repaired version
719 - copy($repaired_path, $pdf_path);
720 - unlink($repaired_path);
721 - //error_log('PDF repaired and parsed successfully: ' . $page_count . ' pages');
722 - return $page_count;
723 - }
724 -
725 - // Clean up repaired file if it didn't work
726 - unlink($repaired_path);
727 - }
728 - } catch (Exception $e) {
729 - //error_log('PDF repair attempt failed: ' . $e->getMessage());
730 - }
731 -
732 - // Method 4: Manual PDF structure analysis (basic page count)
733 - try {
734 - $page_count = $this->mxchat_manual_pdf_page_count($pdf_path);
735 - if ($page_count > 0) {
736 - //error_log('PDF page count determined manually: ' . $page_count . ' pages');
737 - return $page_count;
738 - }
739 - } catch (Exception $e) {
740 - //error_log('Manual PDF analysis failed: ' . $e->getMessage());
741 - }
742 -
743 - //error_log('All PDF parsing methods failed for: ' . $pdf_path);
744 - return false;
745 -}
746 -
747 -/**
748 - * Check if shell_exec is disabled
749 - */
750 -private function mxchat_is_shell_disabled() {
751 - $disabled = explode(',', ini_get('disable_functions'));
752 - return in_array('shell_exec', $disabled);
753 -}
754 -
755 -/**
756 - * Attempt to repair PDF using basic methods
757 - */
758 -private function mxchat_attempt_pdf_repair($pdf_path) {
759 - try {
760 - $content = file_get_contents($pdf_path);
761 - if (!$content) {
762 - return false;
763 - }
764 -
765 - // Check if PDF starts with proper header
766 - if (substr($content, 0, 4) !== '%PDF') {
767 - // Try to find PDF header in the content
768 - $header_pos = strpos($content, '%PDF');
769 - if ($header_pos !== false && $header_pos < 1024) {
770 - // Remove junk before PDF header
771 - $content = substr($content, $header_pos);
772 - $repaired_path = $pdf_path . '.repaired';
773 - file_put_contents($repaired_path, $content);
774 - return $repaired_path;
775 - }
776 - }
777 -
778 - // Check for EOF marker
779 - $content = rtrim($content);
780 - if (!preg_match('/%%EOF\s*$/', $content)) {
781 - // Add EOF marker if missing
782 - $content .= "\n%%EOF";
783 - $repaired_path = $pdf_path . '.repaired';
784 - file_put_contents($repaired_path, $content);
785 - return $repaired_path;
786 - }
787 -
788 - } catch (Exception $e) {
789 - //error_log('PDF repair error: ' . $e->getMessage());
790 - }
791 -
792 - return false;
793 -}
794 -
795 -/**
796 - * Manual PDF page counting by analyzing PDF structure
797 - */
798 -private function mxchat_manual_pdf_page_count($pdf_path) {
799 - try {
800 - $content = file_get_contents($pdf_path);
801 - if (!$content) {
802 - return 0;
803 - }
804 -
805 - // Method 1: Count /Type /Page objects
806 - $page_count = preg_match_all('/\/Type\s*\/Page[^s]/', $content);
807 - if ($page_count > 0) {
808 - return $page_count;
809 - }
810 -
811 - // Method 2: Look for /Count in pages object
812 - if (preg_match('/\/Type\s*\/Pages.*?\/Count\s+(\d+)/', $content, $matches)) {
813 - return intval($matches[1]);
814 - }
815 -
816 - // Method 3: Count page references
817 - $page_count = preg_match_all('/\d+\s+0\s+obj\s*<<[^>]*\/Type\s*\/Page/', $content);
818 - if ($page_count > 0) {
819 - return $page_count;
820 - }
821 -
822 - } catch (Exception $e) {
823 - //error_log('Manual PDF analysis error: ' . $e->getMessage());
824 - }
825 -
826 - return 0;
827 -}
828 -
829 -
830 -public function mxchat_save_inline_prompt() {
831 - // DEBUG: Log what we're receiving
832 - //error_log('=== MXCHAT DEBUG ===');
833 - //error_log('POST data: ' . print_r($_POST, true));
834 - //error_log('Nonce from POST: ' . ($_POST['_ajax_nonce'] ?? 'NOT FOUND'));
835 -
836 - // Check for nonce security
837 - check_ajax_referer('mxchat_save_inline_nonce', '_ajax_nonce');
838 -
839 - // If we get here, nonce passed
840 - //error_log('Nonce verification PASSED');
841 -
842 - // Verify permissions
843 - if (!current_user_can('manage_options')) {
844 - wp_send_json_error(esc_html__('Permission denied.', 'mxchat'));
845 - return;
846 - }
847 -
848 - global $wpdb;
849 - $table_name = $wpdb->prefix . 'mxchat_system_prompt_content';
850 -
851 - // Validate and sanitize input data
852 - $prompt_id = isset($_POST['id']) ? absint($_POST['id']) : 0;
853 - $article_content = isset($_POST['article_content']) ? wp_kses_post(wp_unslash($_POST['article_content'])) : '';
854 - $article_url = isset($_POST['article_url']) ? esc_url_raw($_POST['article_url']) : '';
855 -
856 - if ($prompt_id > 0 && !empty($article_content)) {
857 - // Re-generate the embedding vector for the updated content
858 - $embedding_vector = $this->mxchat_generate_embedding($article_content);
859 - if (is_array($embedding_vector)) {
860 - // Serialize the embedding vector before storing it
861 - $embedding_vector_serialized = serialize($embedding_vector);
862 - // Update the prompt in the database
863 - $updated = $wpdb->update(
864 - $table_name,
865 - array(
866 - 'article_content' => $article_content,
867 - 'embedding_vector' => $embedding_vector_serialized,
868 - 'source_url' => $article_url,
869 - ),
870 - array('id' => $prompt_id),
871 - array('%s', '%s', '%s'),
872 - array('%d')
873 - );
874 - if ($updated !== false) {
875 - wp_send_json_success();
876 - } else {
877 - MxChat_Admin::mxchat_log_debug('knowledge_error', 'Database update failed when saving knowledge entry');
878 - wp_send_json_error(esc_html__('Database update failed.', 'mxchat'));
879 - }
880 - } else {
881 - MxChat_Admin::mxchat_log_debug('embedding_error', 'Embedding generation failed for knowledge entry');
882 - wp_send_json_error(esc_html__('Embedding generation failed.', 'mxchat'));
883 - }
884 - } else {
885 - wp_send_json_error(esc_html__('Invalid data.', 'mxchat'));
886 - }
887 -}
888 -
889 -
890 -/**
891 - * AJAX: Get full content for editing — reassembles chunks if needed.
892 - * Works for both WordPress DB and Pinecone entries.
893 - */
894 -public function ajax_mxchat_get_entry_content() {
895 - check_ajax_referer('mxchat_edit_entry_nonce', 'nonce');
896 -
897 - if ( ! current_user_can('manage_options') ) {
898 - wp_send_json_error( array( 'message' => 'Permission denied.' ) );
899 - }
900 -
901 - $source_url = isset($_POST['source_url']) ? sanitize_text_field( wp_unslash($_POST['source_url']) ) : '';
902 - $entry_id = isset($_POST['entry_id']) ? absint($_POST['entry_id']) : 0;
903 - $data_source = isset($_POST['data_source']) ? sanitize_key($_POST['data_source']) : 'wordpress';
904 - $bot_id = isset($_POST['bot_id']) ? sanitize_key($_POST['bot_id']) : 'default';
905 -
906 - if ( $data_source === 'pinecone' ) {
907 - // Pinecone: fetch vectors by source_url, reassemble chunks
908 - $content = $this->get_pinecone_entry_content( $source_url, $entry_id, $bot_id );
909 - } else {
910 - // WordPress DB
911 - $content = $this->get_wordpress_entry_content( $source_url, $entry_id );
912 - }
913 -
914 - if ( is_wp_error( $content ) ) {
915 - wp_send_json_error( array( 'message' => $content->get_error_message() ) );
916 - }
917 -
918 - wp_send_json_success( $content );
919 -}
920 -
921 -/**
922 - * Get content from WordPress DB — reassembles chunks by source_url.
923 - */
924 -private function get_wordpress_entry_content( $source_url, $entry_id ) {
925 - global $wpdb;
926 - $table = $wpdb->prefix . 'mxchat_system_prompt_content';
927 -
928 - // If we have a source_url, check for chunks
929 - if ( ! empty( $source_url ) && strpos( $source_url, 'mxchat://' ) !== 0 ) {
930 - $rows = $wpdb->get_results( $wpdb->prepare(
931 - "SELECT id, article_content, source_url, content_type FROM {$table} WHERE source_url = %s ORDER BY id ASC",
932 - $source_url
933 - ) );
934 -
935 - if ( $rows && count( $rows ) > 1 ) {
936 - // Multiple rows = chunked. Reassemble.
937 - $chunks = array();
938 - foreach ( $rows as $row ) {
939 - $parsed = MxChat_Chunker::parse_stored_chunk( $row->article_content );
940 - $index = isset( $parsed['metadata']['chunk_index'] ) ? intval( $parsed['metadata']['chunk_index'] ) : count( $chunks );
941 - $chunks[ $index ] = $parsed['text'];
942 - }
943 - ksort( $chunks );
944 - return array(
945 - 'content' => implode( "\n\n", $chunks ),
946 - 'source_url' => $source_url,
947 - 'is_chunked' => true,
948 - 'chunk_count' => count( $chunks ),
949 - 'content_type' => $rows[0]->content_type,
950 - );
951 - } elseif ( $rows && count( $rows ) === 1 ) {
952 - $parsed = MxChat_Chunker::parse_stored_chunk( $rows[0]->article_content );
953 - return array(
954 - 'content' => $parsed['text'],
955 - 'source_url' => $source_url,
956 - 'entry_id' => $rows[0]->id,
957 - 'is_chunked' => false,
958 - 'content_type' => $rows[0]->content_type,
959 - );
960 - }
961 - }
962 -
963 - // Fallback: fetch by ID
964 - if ( $entry_id > 0 ) {
965 - $row = $wpdb->get_row( $wpdb->prepare(
966 - "SELECT id, article_content, source_url, content_type FROM {$table} WHERE id = %d",
967 - $entry_id
968 - ) );
969 - if ( $row ) {
970 - $parsed = MxChat_Chunker::parse_stored_chunk( $row->article_content );
971 - return array(
972 - 'content' => $parsed['text'],
973 - 'source_url' => $row->source_url,
974 - 'entry_id' => $row->id,
975 - 'is_chunked' => false,
976 - 'content_type' => $row->content_type,
977 - );
978 - }
979 - }
980 -
981 - return new WP_Error( 'not_found', 'Entry not found.' );
982 -}
983 -
984 -/**
985 - * Get content from Pinecone — fetches vectors by source_url, reassembles chunks.
986 - */
987 -private function get_pinecone_entry_content( $source_url, $entry_id, $bot_id ) {
988 - if ( ! class_exists('MxChat_Pinecone_Manager') ) {
989 - return new WP_Error( 'pinecone_unavailable', 'Pinecone manager not available.' );
990 - }
991 -
992 - // Get Pinecone config
993 - if ( $bot_id === 'default' || ! class_exists('MxChat_Multi_Bot_Manager') ) {
994 - $pinecone_options = get_option('mxchat_pinecone_addon_options');
995 - $api_key = $pinecone_options['mxchat_pinecone_api_key'] ?? '';
996 - $host = $pinecone_options['mxchat_pinecone_host'] ?? '';
997 - $namespace = $pinecone_options['mxchat_pinecone_namespace'] ?? '';
998 - } else {
999 - $bot_config = apply_filters('mxchat_get_bot_pinecone_config', array(), $bot_id);
1000 - $api_key = $bot_config['api_key'] ?? '';
1001 - $host = $bot_config['host'] ?? '';
1002 - $namespace = $bot_config['namespace'] ?? '';
1003 - }
1004 -
1005 - if ( empty($host) || empty($api_key) ) {
1006 - return new WP_Error( 'pinecone_config', 'Pinecone not configured.' );
1007 - }
1008 -
1009 - // List vectors with the source_url prefix
1010 - $base_id = md5( $source_url );
1011 - $vector_ids = array( $base_id );
1012 -
1013 - // Find chunk vectors
1014 - $list_url = "https://{$host}/vectors/list";
1015 - $list_body = array( 'prefix' => $base_id . '_chunk_', 'limit' => 100 );
1016 - if ( ! empty($namespace) ) {
1017 - $list_body['namespace'] = $namespace;
1018 - }
1019 -
1020 - $list_resp = wp_remote_post( $list_url, array(
1021 - 'headers' => array( 'Api-Key' => $api_key, 'Content-Type' => 'application/json' ),
1022 - 'body' => wp_json_encode( $list_body ),
1023 - 'timeout' => 15,
1024 - ) );
1025 -
1026 - if ( ! is_wp_error($list_resp) ) {
1027 - $list_data = json_decode( wp_remote_retrieve_body($list_resp), true );
1028 - if ( ! empty($list_data['vectors']) ) {
1029 - foreach ( $list_data['vectors'] as $v ) {
1030 - $vector_ids[] = $v['id'];
1031 - }
1032 - }
1033 - }
1034 -
1035 - // Fetch vectors with metadata
1036 - $fetch_url = "https://{$host}/vectors/fetch";
1037 - $fetch_body = array( 'ids' => $vector_ids );
1038 - if ( ! empty($namespace) ) {
1039 - $fetch_body['namespace'] = $namespace;
1040 - }
1041 -
1042 - $fetch_resp = wp_remote_post( $fetch_url, array(
1043 - 'headers' => array( 'Api-Key' => $api_key, 'Content-Type' => 'application/json' ),
1044 - 'body' => wp_json_encode( $fetch_body ),
1045 - 'timeout' => 15,
1046 - ) );
1047 -
1048 - if ( is_wp_error($fetch_resp) ) {
1049 - return new WP_Error( 'pinecone_fetch', 'Failed to fetch from Pinecone.' );
1050 - }
1051 -
1052 - $fetch_data = json_decode( wp_remote_retrieve_body($fetch_resp), true );
1053 - $vectors = $fetch_data['vectors'] ?? array();
1054 -
1055 - if ( empty($vectors) ) {
1056 - return new WP_Error( 'not_found', 'Entry not found in Pinecone.' );
1057 - }
1058 -
1059 - // Reassemble chunks
1060 - $chunks = array();
1061 - $content_type = 'content';
1062 - foreach ( $vectors as $vid => $vector ) {
1063 - $meta = $vector['metadata'] ?? array();
1064 - $text = $meta['text'] ?? '';
1065 - $index = $meta['chunk_index'] ?? 0;
1066 - $content_type = $meta['type'] ?? 'content';
1067 - $chunks[ intval($index) ] = $text;
1068 - }
1069 - ksort( $chunks );
1070 -
1071 - return array(
1072 - 'content' => implode( "\n\n", $chunks ),
1073 - 'source_url' => $source_url,
1074 - 'is_chunked' => count($chunks) > 1,
1075 - 'chunk_count' => count($chunks),
1076 - 'content_type' => $content_type,
1077 - );
1078 -}
1079 -
1080 -/**
1081 - * AJAX: Inspect a knowledge entry — returns the per-chunk stored text + metadata
1082 - * WITHOUT collapsing it, so a site owner can see exactly what was indexed for an
1083 - * entry (plan-mxchat-20260628-d8cb4b). READ-ONLY: never re-embeds or mutates.
1084 - */
1085 -public function ajax_mxchat_inspect_entry() {
1086 - check_ajax_referer('mxchat_inspect_entry_nonce', 'nonce');
1087 -
1088 - if ( ! current_user_can('manage_options') ) {
1089 - wp_send_json_error( array( 'message' => esc_html__('Permission denied.', 'mxchat') ) );
1090 - }
1091 -
1092 - $source_url = isset($_POST['source_url']) ? sanitize_text_field( wp_unslash($_POST['source_url']) ) : '';
1093 - $entry_id = isset($_POST['entry_id']) ? absint($_POST['entry_id']) : 0;
1094 - $data_source = isset($_POST['data_source']) ? sanitize_key($_POST['data_source']) : 'wordpress';
1095 - $bot_id = isset($_POST['bot_id']) ? sanitize_key($_POST['bot_id']) : 'default';
1096 -
1097 - if ( $data_source === 'pinecone' ) {
1098 - $result = $this->inspect_pinecone_entry( $source_url, $entry_id, $bot_id );
1099 - } else {
1100 - $result = $this->inspect_wordpress_entry( $source_url, $entry_id );
1101 - }
1102 -
1103 - if ( is_wp_error( $result ) ) {
1104 - wp_send_json_error( array( 'message' => $result->get_error_message() ) );
1105 - }
1106 -
1107 - wp_send_json_success( $result );
1108 -}
1109 -
1110 -/**
1111 - * Read-only inspector for WordPress-DB entries. Mirrors get_wordpress_entry_content()
1112 - * but returns each STORED chunk's exact text + length (no implode), plus the assembled
1113 - * embedded text. This shows what is actually in the index, not a re-derivation from the post.
1114 - */
1115 -private function inspect_wordpress_entry( $source_url, $entry_id ) {
1116 - global $wpdb;
1117 - $table = $wpdb->prefix . 'mxchat_system_prompt_content';
1118 -
1119 - $rows = array();
1120 -
1121 - // Group by the real stored source_url — this INCLUDES "mxchat://" manual
1122 - // Direct Content entries (the spec's manual-entry case), which share one
1123 - // source_url across their chunk rows. Only the synthetic "_ungrouped_<id>"
1124 - // display key (invented by the table view for rows with no source_url) is
1125 - // excluded; those fall through to the entry_id lookup below.
1126 - if ( ! empty( $source_url ) && strpos( $source_url, '_ungrouped_' ) !== 0 ) {
1127 - $rows = $wpdb->get_results( $wpdb->prepare(
1128 - "SELECT id, article_content, source_url, content_type FROM {$table} WHERE source_url = %s ORDER BY id ASC",
1129 - $source_url
1130 - ) );
1131 - }
1132 -
1133 - // Fallback / manual "Direct Content" entries: fetch the single row by id.
1134 - if ( empty( $rows ) && $entry_id > 0 ) {
1135 - $row = $wpdb->get_row( $wpdb->prepare(
1136 - "SELECT id, article_content, source_url, content_type FROM {$table} WHERE id = %d",
1137 - $entry_id
1138 - ) );
1139 - if ( $row ) {
1140 - $rows = array( $row );
1141 - }
1142 - }
1143 -
1144 - if ( empty( $rows ) ) {
1145 - return new WP_Error( 'not_found', esc_html__('Entry not found in the local knowledge database.', 'mxchat') );
1146 - }
1147 -
1148 - $chunks = array();
1149 - $content_type = '';
1150 - foreach ( $rows as $row ) {
1151 - $parsed = MxChat_Chunker::parse_stored_chunk( $row->article_content );
1152 - $text = isset( $parsed['text'] ) ? $parsed['text'] : '';
1153 - $index = isset( $parsed['metadata']['chunk_index'] ) ? intval( $parsed['metadata']['chunk_index'] ) : count( $chunks );
1154 - $content_type = $row->content_type;
1155 - $chunks[] = array(
1156 - 'index' => $index,
1157 - 'text' => $text,
1158 - 'length' => function_exists('mb_strlen') ? mb_strlen( $text ) : strlen( $text ),
1159 - 'row_id' => intval( $row->id ),
1160 - );
1161 - }
1162 -
1163 - usort( $chunks, function( $a, $b ) { return $a['index'] - $b['index']; } );
1164 -
1165 - $assembled = implode( "\n\n", wp_list_pluck( $chunks, 'text' ) );
1166 -
1167 - return array(
1168 - 'store' => 'wordpress',
1169 - 'source_url' => $source_url,
1170 - 'content_type' => $content_type,
1171 - 'is_chunked' => count( $chunks ) > 1,
1172 - 'chunk_count' => count( $chunks ),
1173 - 'assembled' => $assembled,
1174 - 'assembled_length' => function_exists('mb_strlen') ? mb_strlen( $assembled ) : strlen( $assembled ),
1175 - 'chunks' => array_values( $chunks ),
1176 - // WP-DB storage carries no separate vector metadata; surface that fact
1177 - // rather than letting the owner guess (the spec's taxonomy question).
1178 - 'metadata' => array(),
1179 - 'metadata_note' => esc_html__('Stored in the local WordPress database. Only the assembled text shown here is embedded — there are no separate vector metadata fields (e.g. taxonomy terms are not stored unless they were injected into the text itself).', 'mxchat'),
1180 - );
1181 -}
1182 -
1183 -/**
1184 - * Read-only inspector for Pinecone entries. Mirrors get_pinecone_entry_content()
1185 - * but keeps each vector's text + metadata instead of imploding, so the owner can
1186 - * confirm exactly which metadata fields (text/source_url/type/last_updated/created_at/bot_id)
1187 - * are present per chunk. READ-ONLY.
1188 - */
1189 -private function inspect_pinecone_entry( $source_url, $entry_id, $bot_id ) {
1190 - if ( ! class_exists('MxChat_Pinecone_Manager') ) {
1191 - return new WP_Error( 'pinecone_unavailable', esc_html__('Pinecone manager not available.', 'mxchat') );
1192 - }
1193 -
1194 - if ( $bot_id === 'default' || ! class_exists('MxChat_Multi_Bot_Manager') ) {
1195 - $pinecone_options = get_option('mxchat_pinecone_addon_options');
1196 - $api_key = $pinecone_options['mxchat_pinecone_api_key'] ?? '';
1197 - $host = $pinecone_options['mxchat_pinecone_host'] ?? '';
1198 - $namespace = $pinecone_options['mxchat_pinecone_namespace'] ?? '';
1199 - } else {
1200 - $bot_config = apply_filters('mxchat_get_bot_pinecone_config', array(), $bot_id);
1201 - $api_key = $bot_config['api_key'] ?? '';
1202 - $host = $bot_config['host'] ?? '';
1203 - $namespace = $bot_config['namespace'] ?? '';
1204 - }
1205 -
1206 - if ( empty($host) || empty($api_key) ) {
1207 - return new WP_Error( 'pinecone_config', esc_html__('Pinecone not configured.', 'mxchat') );
1208 - }
1209 -
1210 - $base_id = md5( $source_url );
1211 - $vector_ids = array( $base_id );
1212 -
1213 - $list_url = "https://{$host}/vectors/list";
1214 - $list_body = array( 'prefix' => $base_id . '_chunk_', 'limit' => 100 );
1215 - if ( ! empty($namespace) ) {
1216 - $list_body['namespace'] = $namespace;
1217 - }
1218 -
1219 - $list_resp = wp_remote_post( $list_url, array(
1220 - 'headers' => array( 'Api-Key' => $api_key, 'Content-Type' => 'application/json' ),
1221 - 'body' => wp_json_encode( $list_body ),
1222 - 'timeout' => 15,
1223 - ) );
1224 -
1225 - if ( ! is_wp_error($list_resp) ) {
1226 - $list_data = json_decode( wp_remote_retrieve_body($list_resp), true );
1227 - if ( ! empty($list_data['vectors']) ) {
1228 - foreach ( $list_data['vectors'] as $v ) {
1229 - $vector_ids[] = $v['id'];
1230 - }
1231 - }
1232 - }
1233 -
1234 - $fetch_url = "https://{$host}/vectors/fetch";
1235 - $fetch_body = array( 'ids' => $vector_ids );
1236 - if ( ! empty($namespace) ) {
1237 - $fetch_body['namespace'] = $namespace;
1238 - }
1239 -
1240 - $fetch_resp = wp_remote_post( $fetch_url, array(
1241 - 'headers' => array( 'Api-Key' => $api_key, 'Content-Type' => 'application/json' ),
1242 - 'body' => wp_json_encode( $fetch_body ),
1243 - 'timeout' => 15,
1244 - ) );
1245 -
1246 - if ( is_wp_error($fetch_resp) ) {
1247 - return new WP_Error( 'pinecone_fetch', esc_html__('Failed to fetch from Pinecone.', 'mxchat') );
1248 - }
1249 -
1250 - $fetch_data = json_decode( wp_remote_retrieve_body($fetch_resp), true );
1251 - $vectors = $fetch_data['vectors'] ?? array();
1252 -
1253 - if ( empty($vectors) ) {
1254 - return new WP_Error( 'not_found', esc_html__('Entry not found in Pinecone.', 'mxchat') );
1255 - }
1256 -
1257 - // Whitelisted metadata fields the spec calls out — shown so devs can confirm
1258 - // what is (and is NOT) stored per vector.
1259 - $meta_fields = array( 'text', 'source_url', 'type', 'last_updated', 'created_at', 'bot_id', 'chunk_index', 'total_chunks' );
1260 - $chunks = array();
1261 - $content_type = '';
1262 - foreach ( $vectors as $vid => $vector ) {
1263 - $meta = isset($vector['metadata']) && is_array($vector['metadata']) ? $vector['metadata'] : array();
1264 - $text = $meta['text'] ?? '';
1265 - $index = isset($meta['chunk_index']) ? intval($meta['chunk_index']) : count($chunks);
1266 - $content_type = $meta['type'] ?? $content_type;
1267 -
1268 - $clean_meta = array();
1269 - foreach ( $meta_fields as $field ) {
1270 - if ( array_key_exists( $field, $meta ) && $field !== 'text' ) {
1271 - $clean_meta[ $field ] = is_scalar( $meta[ $field ] ) ? (string) $meta[ $field ] : wp_json_encode( $meta[ $field ] );
1272 - }
1273 - }
1274 -
1275 - $chunks[] = array(
1276 - 'index' => $index,
1277 - 'text' => $text,
1278 - 'length' => function_exists('mb_strlen') ? mb_strlen( $text ) : strlen( $text ),
1279 - 'vector_id' => (string) $vid,
1280 - 'metadata' => $clean_meta,
1281 - );
1282 - }
1283 -
1284 - usort( $chunks, function( $a, $b ) { return $a['index'] - $b['index']; } );
1285 -
1286 - $assembled = implode( "\n\n", wp_list_pluck( $chunks, 'text' ) );
1287 -
1288 - return array(
1289 - 'store' => 'pinecone',
1290 - 'source_url' => $source_url,
1291 - 'content_type' => $content_type,
1292 - 'is_chunked' => count( $chunks ) > 1,
1293 - 'chunk_count' => count( $chunks ),
1294 - 'assembled' => $assembled,
1295 - 'assembled_length' => function_exists('mb_strlen') ? mb_strlen( $assembled ) : strlen( $assembled ),
1296 - 'chunks' => array_values( $chunks ),
1297 - 'metadata' => array(),
1298 - 'metadata_note' => esc_html__('Stored in Pinecone. Each chunk above lists the vector metadata fields actually present — if a field you expect (such as taxonomy terms) is missing here, it was not stored as metadata and is only searchable if it appears in the embedded text.', 'mxchat'),
1299 - );
1300 -}
1301 -
1302 -/**
1303 - * AJAX: Save edited content — re-chunks and re-embeds as needed.
1304 - * Works for both WordPress DB and Pinecone entries.
1305 - */
1306 -public function ajax_mxchat_save_entry_content() {
1307 - check_ajax_referer('mxchat_edit_entry_nonce', 'nonce');
1308 -
1309 - if ( ! current_user_can('manage_options') ) {
1310 - wp_send_json_error( array( 'message' => 'Permission denied.' ) );
1311 - }
1312 -
1313 - $source_url = isset($_POST['source_url']) ? sanitize_text_field( wp_unslash($_POST['source_url']) ) : '';
1314 - $entry_id = isset($_POST['entry_id']) ? absint($_POST['entry_id']) : 0;
1315 - $content = isset($_POST['content']) ? wp_kses_post( wp_unslash($_POST['content']) ) : '';
1316 - $data_source = isset($_POST['data_source']) ? sanitize_key($_POST['data_source']) : 'wordpress';
1317 - $bot_id = isset($_POST['bot_id']) ? sanitize_key($_POST['bot_id']) : 'default';
1318 - $content_type = isset($_POST['content_type']) ? sanitize_key($_POST['content_type']) : 'content';
1319 -
1320 - if ( empty($content) ) {
1321 - wp_send_json_error( array( 'message' => 'Content cannot be empty.' ) );
1322 - }
1323 -
1324 - // Get the embedding API key
1325 - $options = get_option('mxchat_options', array());
1326 - $api_key = '';
1327 -
1328 - if ( $bot_id !== 'default' && class_exists('MxChat_Multi_Bot_Manager') ) {
1329 - $bot_options = apply_filters('mxchat_get_bot_options', array(), $bot_id);
1330 - $api_key = $bot_options['api_key'] ?? '';
1331 - }
1332 - if ( empty($api_key) ) {
1333 - $api_key = $options['api_key'] ?? '';
1334 - }
1335 -
1336 - global $wpdb;
1337 - $table = $wpdb->prefix . 'mxchat_system_prompt_content';
1338 -
1339 - // If source_url is empty but we have an entry_id, look it up
1340 - if ( empty($source_url) && $entry_id > 0 && $data_source === 'wordpress' ) {
1341 - $row = $wpdb->get_row( $wpdb->prepare( "SELECT source_url FROM {$table} WHERE id = %d", $entry_id ) );
1342 - if ( $row && ! empty($row->source_url) ) {
1343 - $source_url = $row->source_url;
1344 - }
1345 - }
1346 -
1347 - // For manual entries (no source_url or mxchat:// prefix), delete the old entry by ID first
1348 - // so submit_content_to_db creates a replacement instead of a duplicate
1349 - // Also treat legacy mxchat.ai source URLs as manual — old bug assigned the site URL to manual entries
1350 - $is_legacy_manual = !empty($source_url) && strpos($source_url, 'mxchat.ai') !== false && strpos($source_url, 'mxchat://') !== 0;
1351 - if ( $entry_id > 0 && (empty($source_url) || strpos($source_url, 'mxchat://') === 0 || $is_legacy_manual) ) {
1352 - $wpdb->delete( $table, array( 'id' => $entry_id ), array( '%d' ) );
1353 - // Clear legacy URL so submit_content_to_db generates a unique mxchat:// identifier
1354 - // instead of reusing the shared URL (which would mass-delete other entries with the same URL)
1355 - if ( $is_legacy_manual ) {
1356 - $source_url = '';
1357 - }
1358 - }
1359 -
1360 - // Use the existing submit_content_to_db which handles chunking, Pinecone, and WP DB
1361 - $vector_id = ! empty($source_url) ? md5($source_url) : md5('mxchat_manual_' . $entry_id);
1362 -
1363 - // submit_content_to_db already handles: delete old chunks → re-chunk → re-embed → store
1364 - $result = MxChat_Utils::submit_content_to_db( $content, $source_url, $api_key, $vector_id, $bot_id, $content_type );
1365 -
1366 - if ( is_wp_error($result) ) {
1367 - wp_send_json_error( array( 'message' => $result->get_error_message() ) );
1368 - }
1369 -
1370 - wp_send_json_success( array( 'message' => 'Content saved and re-embedded successfully.' ) );
1371 -}
1372 -
1373 -public function mxchat_get_pdf_processing_status($pdf_url) {
1374 - $pdf_url = esc_url_raw($pdf_url);
1375 - $status = get_transient(sanitize_key('mxchat_pdf_status_' . md5($pdf_url)));
1376 -
1377 - if (!$status || !is_array($status)) {
1378 - return false;
1379 - }
1380 -
1381 - // Check for stalled processing (no updates for 5 minutes)
1382 - if ($status['status'] === 'processing' && (time() - absint($status['last_update'])) > 300) {
1383 - $status['status'] = 'error';
1384 - $status['error'] = __('PDF processing appears to be stalled. No updates for over 5 minutes.', 'mxchat');
1385 -
1386 - // Save the updated status
1387 - set_transient(
1388 - sanitize_key('mxchat_pdf_status_' . md5($pdf_url)),
1389 - array_map('sanitize_text_field', $status),
1390 - DAY_IN_SECONDS
1391 - );
1392 - }
1393 -
1394 - $result = array(
1395 - 'total_pages' => absint($status['total_pages']),
1396 - 'processed_pages' => absint($status['processed_pages']),
1397 - 'failed_pages' => absint($status['failed_pages'] ?? 0),
1398 - 'percentage' => ($status['total_pages'] > 0)
1399 - ? round((absint($status['processed_pages']) / absint($status['total_pages'])) * 100)
1400 - : 0,
1401 - 'status' => sanitize_text_field($status['status']),
1402 - 'last_update' => human_time_diff(absint($status['last_update']), time()) . ' ' . esc_html__('ago', 'mxchat'),
1403 - 'failed_pages_list' => isset($status['failed_pages_list']) ? $status['failed_pages_list'] : array(),
1404 - 'completion_summary' => isset($status['completion_summary']) ? $status['completion_summary'] : null
1405 - );
1406 -
1407 - // Add error message if present
1408 - if (isset($status['error']) && !empty($status['error'])) {
1409 - $result['error'] = sanitize_text_field($status['error']);
1410 - }
1411 -
1412 - return $result;
1413 -}
1414 -
1415 -
1416 -public function mxchat_handle_sitemap_submission() {
1417 - // Check if the form was submitted and verify permissions
1418 - if (!isset($_POST['submit_sitemap']) || !current_user_can('manage_options')) {
1419 - wp_die(esc_html__('Unauthorized access', 'mxchat'));
1420 - }
1421 -
1422 - // Verify nonce
1423 - check_admin_referer('mxchat_submit_sitemap_action', 'mxchat_submit_sitemap_nonce');
1424 -
1425 - // Validate URL
1426 - if (!isset($_POST['sitemap_url']) || empty($_POST['sitemap_url'])) {
1427 - set_transient('mxchat_admin_notice_error',
1428 - esc_html__('Please provide a valid URL.', 'mxchat'),
1429 - 30
1430 - );
1431 - wp_safe_redirect(esc_url(admin_url('admin.php?page=mxchat-prompts')));
1432 - exit;
1433 - }
1434 -
1435 - $submitted_url = esc_url_raw($_POST['sitemap_url']);
1436 -
1437 - // Convert Google Drive sharing URLs to direct download URLs
1438 - if ( strpos($submitted_url, 'drive.google.com') !== false ) {
1439 - $file_id = '';
1440 - if ( preg_match('/[?&]id=([a-zA-Z0-9_-]+)/', $submitted_url, $m) ) {
1441 - $file_id = $m[1];
1442 - } elseif ( preg_match('#/file/d/([a-zA-Z0-9_-]+)#', $submitted_url, $m) ) {
1443 - $file_id = $m[1];
1444 - }
1445 - if ( ! empty($file_id) ) {
1446 - $submitted_url = 'https://drive.google.com/uc?export=download&id=' . $file_id;
1447 - }
1448 - }
1449 -
1450 - // Get bot_id from form submission
1451 - $bot_id = isset($_POST['bot_id']) ? sanitize_key($_POST['bot_id']) : 'default';
1452 -
1453 - // Get bot-specific options and validate API key
1454 - $bot_options = $this->get_bot_options($bot_id);
1455 - $options = !empty($bot_options) ? $bot_options : get_option('mxchat_options');
1456 - $selected_model = $options['embedding_model'] ?? 'text-embedding-ada-002';
1457 -
1458 - if (strpos($selected_model, 'voyage') === 0) {
1459 - $api_key = $options['voyage_api_key'] ?? '';
1460 - $provider_name = 'Voyage AI';
1461 - } elseif (strpos($selected_model, 'gemini-embedding') === 0) {
1462 - $api_key = $options['gemini_api_key'] ?? '';
1463 - $provider_name = 'Google Gemini';
1464 - } else {
1465 - $api_key = $options['api_key'] ?? '';
1466 - $provider_name = 'OpenAI';
1467 - }
1468 -
1469 - if (empty($api_key)) {
1470 - $error_message = sprintf(
1471 - esc_html__('%s API key is not configured. Please add your API key in the settings before submitting content.', 'mxchat'),
1472 - $provider_name
1473 - );
1474 - set_transient('mxchat_admin_notice_error', $error_message, 30);
1475 - wp_safe_redirect(esc_url(admin_url('admin.php?page=mxchat-prompts')));
1476 - exit;
1477 - }
1478 -
1479 - // Fetch URL — send an honest, versioned MXChat crawler UA (not a spoofed
1480 - // browser). Stale browser UAs are exactly what WAFs like SiteGround's
1481 - // ModSecurity flag as scrapers, 403-ing the fetch (including PDFs served
1482 - // from the site's own media library, which route through this same call).
1483 - // See mxchat_ingest_user_agent(). Accept is kept for content negotiation;
1484 - // the browser-only Accept-Language fingerprint is dropped so it stays
1485 - // coherent with a bot identity.
1486 - $response = wp_remote_get($submitted_url, array(
1487 - 'timeout' => 30,
1488 - 'sslverify' => false,
1489 - 'user-agent' => mxchat_ingest_user_agent(),
1490 - 'headers' => array(
1491 - 'Accept' => 'text/html,application/xhtml+xml,application/xml;q=0.9,*/*;q=0.8',
1492 - ),
1493 - ));
1494 -
1495 - if (is_wp_error($response) || wp_remote_retrieve_response_code($response) !== 200) {
1496 - $error_message = is_wp_error($response) ? $response->get_error_message() : 'HTTP Status: ' . wp_remote_retrieve_response_code($response);
1497 - set_transient('mxchat_admin_notice_error',
1498 - sprintf(
1499 - esc_html__('Failed to fetch the URL: %s', 'mxchat'),
1500 - esc_html($error_message)
1501 - ),
1502 - 30
1503 - );
1504 - wp_safe_redirect(esc_url(admin_url('admin.php?page=mxchat-prompts')));
1505 - exit;
1506 - }
1507 -
1508 - $content_type = wp_remote_retrieve_header($response, 'content-type');
1509 - $body_content = wp_remote_retrieve_body($response);
1510 -
1511 - if (empty($body_content)) {
1512 - set_transient('mxchat_admin_notice_error',
1513 - esc_html__('Empty response received from URL.', 'mxchat'),
1514 - 30
1515 - );
1516 - wp_safe_redirect(esc_url(admin_url('admin.php?page=mxchat-prompts')));
1517 - exit;
1518 - }
1519 -
1520 - // Handle PDF URL
1521 - if ($this->mxchat_is_pdf_url($submitted_url, $response)) {
1522 - $result = $this->mxchat_handle_pdf_for_knowledge_base($submitted_url, $response, $bot_id);
1523 -
1524 - if ($result === 'queued') {
1525 - set_transient('mxchat_admin_notice_success',
1526 - esc_html__('PDF queued for processing. Processing will start automatically.', 'mxchat'),
1527 - 30
1528 - );
1529 - } else {
1530 - set_transient('mxchat_admin_notice_error',
1531 - esc_html__('Failed to queue PDF processing: ', 'mxchat') . esc_html($result),
1532 - 30
1533 - );
1534 - }
1535 -
1536 - wp_safe_redirect(esc_url(admin_url('admin.php?page=mxchat-prompts')));
1537 - exit;
1538 - }
1539 -
1540 - // Handle Sitemap XML
1541 - if (strpos($content_type, 'xml') !== false || strpos($body_content, '<urlset') !== false) {
1542 - libxml_use_internal_errors(true);
1543 - $xml = simplexml_load_string($body_content);
1544 - $xml_errors = libxml_get_errors();
1545 - libxml_clear_errors();
1546 -
1547 - if ($xml === false || !empty($xml_errors)) {
1548 - set_transient('mxchat_admin_notice_error',
1549 - esc_html__('Invalid sitemap XML. Please provide a valid sitemap.', 'mxchat'),
1550 - 30
1551 - );
1552 - wp_safe_redirect(esc_url(admin_url('admin.php?page=mxchat-prompts')));
1553 - exit;
1554 - }
1555 -
1556 - $result = $this->mxchat_handle_sitemap_for_knowledge_base($xml, $submitted_url, $bot_id);
1557 -
1558 - if ($result === 'queued') {
1559 - set_transient('mxchat_admin_notice_success',
1560 - esc_html__('Sitemap queued for processing. Processing will start automatically.', 'mxchat'),
1561 - 30
1562 - );
1563 - } else {
1564 - set_transient('mxchat_admin_notice_error',
1565 - esc_html__('Failed to queue sitemap processing. Please check the status below for details.', 'mxchat'),
1566 - 30
1567 - );
1568 - }
1569 -
1570 - wp_safe_redirect(esc_url(admin_url('admin.php?page=mxchat-prompts')));
1571 - exit;
1572 - }
1573 -
1574 - // Handle Regular URL (single page)
1575 - $page_content = $this->mxchat_extract_main_content($body_content);
1576 - $sanitized_content = $this->mxchat_sanitize_content_for_api($page_content);
1577 -
1578 - //error_log('[MXCHAT-URL-DEBUG] URL: ' . $submitted_url);
1579 - //error_log('[MXCHAT-URL-DEBUG] Extracted page_content length: ' . strlen($page_content));
1580 - //error_log('[MXCHAT-URL-DEBUG] Sanitized content length: ' . strlen($sanitized_content));
1581 - //error_log('[MXCHAT-URL-DEBUG] Content preview: ' . substr($sanitized_content, 0, 500));
1582 -
1583 - if (empty($sanitized_content)) {
1584 - set_transient('mxchat_admin_notice_error',
1585 - esc_html__('No valid content found on the provided URL.', 'mxchat'),
1586 - 30
1587 - );
1588 - wp_safe_redirect(esc_url(admin_url('admin.php?page=mxchat-prompts')));
1589 - exit;
1590 - }
1591 -
1592 - // For single URLs, process immediately using submit_content_to_db
1593 - // This handles chunking automatically for large content
1594 - $db_result = MxChat_Utils::submit_content_to_db(
1595 - $sanitized_content,
1596 - $submitted_url,
1597 - $api_key,
1598 - null,
1599 - $bot_id,
1600 - 'url' // content_type
1601 - );
1602 -
1603 - if (is_wp_error($db_result)) {
1604 - $error_message = esc_html__('Failed to store content in database: ', 'mxchat') . esc_html($db_result->get_error_message());
1605 - set_transient('mxchat_admin_notice_error', $error_message, 30);
1606 - } else {
1607 - $success_message = esc_html__('URL content successfully submitted!', 'mxchat');
1608 - set_transient('mxchat_admin_notice_success', $success_message, 30);
1609 - }
1610 -
1611 - wp_safe_redirect(esc_url(admin_url('admin.php?page=mxchat-prompts')));
1612 - exit;
1613 -}
1614 -
1615 -
1616 -public function mxchat_get_single_url_status() {
1617 - $status = get_transient('mxchat_single_url_status');
1618 - if (!$status) {
1619 - return null;
1620 - }
1621 -
1622 - // Add human-readable time
1623 - if (isset($status['timestamp'])) {
1624 - $status['human_time'] = human_time_diff(strtotime($status['timestamp']), current_time('timestamp')) . ' ' . __('ago', 'mxchat');
1625 - }
1626 -
1627 - return $status;
1628 -}
1629 -
1630 -public function mxchat_handle_sitemap_for_knowledge_base($xml, $sitemap_url, $bot_id = 'default') {
1631 - if (!current_user_can('manage_options')) {
1632 - return false;
1633 - }
1634 -
1635 - try {
1636 - $sitemap_url = esc_url_raw($sitemap_url);
1637 -
1638 - if (!$xml || !is_object($xml)) {
1639 - throw new Exception(__('Invalid XML object provided', 'mxchat'));
1640 - }
1641 -
1642 - // Get bot-specific embedding API for validation
1643 - $bot_options = $this->get_bot_options($bot_id);
1644 - $options = !empty($bot_options) ? $bot_options : get_option('mxchat_options');
1645 -
1646 - // Test the embedding API before processing
1647 - $test_phrase = "Test embedding generation for MxChat";
1648 - $test_result = $this->mxchat_generate_embedding($test_phrase, $bot_id);
1649 -
1650 - if (is_string($test_result)) {
1651 - throw new Exception(__('Embedding API validation failed: ', 'mxchat') . $test_result);
1652 - }
1653 -
1654 - if (!is_array($test_result)) {
1655 - throw new Exception(__('Embedding API returned unexpected result type. Please check your configuration.', 'mxchat'));
1656 - }
1657 -
1658 - // Extract URLs from sitemap
1659 - $urls = array();
1660 - foreach ($xml->url as $url_element) {
1661 - $url = esc_url_raw((string)$url_element->loc);
1662 - if ($url) {
1663 - $urls[] = array('url' => $url);
1664 - }
1665 - }
1666 -
1667 - $total_urls = count($urls);
1668 -
1669 - if ($total_urls < 1) {
1670 - throw new Exception(__('No valid URLs found in sitemap', 'mxchat'));
1671 - }
1672 -
1673 - // Create unique queue ID
1674 - $queue_id = 'sitemap_' . md5($sitemap_url . time());
1675 -
1676 - // Add URLs to queue
1677 - $queued_count = $this->mxchat_add_to_queue($queue_id, 'url', $urls, $bot_id);
1678 -
1679 - if ($queued_count === 0) {
1680 - throw new Exception(__('Failed to add URLs to processing queue', 'mxchat'));
1681 - }
1682 -
1683 - // Store queue metadata
1684 - $this->mxchat_set_queue_meta($queue_id, 'source_url', $sitemap_url);
1685 - $this->mxchat_set_queue_meta($queue_id, 'queue_type', 'sitemap');
1686 - $this->mxchat_set_queue_meta($queue_id, 'total_items', $total_urls);
1687 - $this->mxchat_set_queue_meta($queue_id, 'bot_id', $bot_id);
1688 - $this->mxchat_set_queue_meta($queue_id, 'created_at', current_time('mysql'));
1689 -
1690 - // Store queue ID in transient for status tracking
1691 - set_transient('mxchat_active_queue_sitemap', $queue_id, DAY_IN_SECONDS);
1692 - set_transient('mxchat_last_sitemap_url', $sitemap_url, DAY_IN_SECONDS);
1693 -
1694 - return 'queued';
1695 -
1696 - } catch (Exception $e) {
1697 - $error_message = $e->getMessage();
1698 - //error_log(sprintf(esc_html__('Error preparing sitemap for processing: %s', 'mxchat'), esc_html($error_message)));
1699 -
1700 - return $error_message;
1701 - }
1702 -
1703 -}
1704 -
1705 -/**
1706 - * Remove shortcode tags but preserve the content inside them
1707 - * Example: [vc_column]Hello World[/vc_column] becomes "Hello World"
1708 - *
1709 - * @param string $content The content containing shortcodes
1710 - * @return string Content with shortcode tags removed but inner content preserved
1711 - */
1712 -private function strip_shortcode_tags_preserve_content($content) {
1713 - // Single-pass regex removes all shortcode brackets: [tag], [tag attr="val"], [/tag], [tag /]
1714 - // Content between tags is inherently preserved since only brackets are targeted
1715 - $result = preg_replace('/\[\/?\w[\w-]*[^\]]*\]/', '', $content);
1716 - return ($result !== null) ? $result : $content;
1717 -}
1718 -
1719 -public function mxchat_sanitize_content_for_api($content) {
1720 - //error_log('[MXCHAT-SANITIZE] Original content preview: ' . substr($content, 0, 500) . '...');
1721 -
1722 - // Remove shortcode tags but PRESERVE content inside them
1723 - $content = $this->strip_shortcode_tags_preserve_content($content);
1724 -
1725 - // Remove script, style tags, and HTML comments
1726 - $content = preg_replace('/<script\b[^>]*>(.*?)<\/script>/is', '', $content);
1727 - $content = preg_replace('/<style\b[^>]*>(.*?)<\/style>/is', '', $content);
1728 - $content = preg_replace('/<!--(.|\s)*?-->/', '', $content);
1729 -
1730 - // Remove all HTML tags and decode HTML entities
1731 - $content = wp_strip_all_tags($content);
1732 - $content = html_entity_decode($content, ENT_QUOTES | ENT_HTML5);
1733 -
1734 - // Normalize whitespace but preserve paragraph breaks
1735 - // First, normalize line endings to \n
1736 - $content = str_replace(["\r\n", "\r"], "\n", $content);
1737 - // Replace multiple spaces/tabs with single space, but preserve newlines
1738 - $content = preg_replace('/[ \t]+/', ' ', $content);
1739 - // Replace 3+ newlines with 2 newlines (max 2 blank lines)
1740 - $content = preg_replace('/\n{3,}/', "\n\n", $content);
1741 - // Trim each line
1742 - $lines = explode("\n", $content);
1743 - $lines = array_map('trim', $lines);
1744 - $content = implode("\n", $lines);
1745 - // Final trim
1746 - $content = trim($content);
1747 -
1748 - // Remove control characters (which can cause database issues) but preserve \n (0x0A) and \r (0x0D)
1749 - $content = preg_replace('/[\x00-\x08\x0B\x0C\x0E-\x1F\x7F]/u', '', $content);
1750 -
1751 - // Remove NULL bytes which can cause database errors
1752 - $content = str_replace("\0", "", $content);
1753 -
1754 - // Ensure valid UTF-8 encoding
1755 - $content = wp_check_invalid_utf8($content);
1756 -
1757 - // Remove any extremely long strings without spaces (often garbage)
1758 - $content = preg_replace('/\S{300,}/', ' ', $content);
1759 -
1760 - // Replace problematic characters that often cause database issues
1761 - $content = preg_replace('/[\x{10000}-\x{10FFFF}]/u', '', $content); // Remove emoji and other high Unicode characters
1762 -
1763 - // Replace any remaining potentially problematic characters with spaces
1764 - // BUT preserve newlines by temporarily replacing them
1765 - $content = str_replace("\n", "NEWLINE_PLACEHOLDER", $content);
1766 - $content = preg_replace('/[^\p{L}\p{N}\p{P}\p{Z}\p{Sm}]/u', ' ', $content);
1767 - $content = str_replace("NEWLINE_PLACEHOLDER", "\n", $content);
1768 -
1769 - // Limit to reasonable length if needed
1770 - $max_length = 65000; // Just under MySQL TEXT field limit
1771 - if (strlen($content) > $max_length) {
1772 - $content = substr($content, 0, $max_length);
1773 - }
1774 -
1775 - //error_log('[MXCHAT-SANITIZE] Sanitized content preview: ' . substr($content, 0, 500) . '...');
1776 - return $content;
1777 -}
1778 -public function mxchat_extract_main_content($html) {
1779 - if (empty($html)) {
1780 - return '';
1781 - }
1782 - try {
1783 - $dom = new DOMDocument;
1784 - libxml_use_internal_errors(true); // Suppress HTML parsing errors
1785 - @$dom->loadHTML(mb_convert_encoding($html, 'HTML-ENTITIES', 'UTF-8'), LIBXML_HTML_NOIMPLIED | LIBXML_HTML_NODEFDTD);
1786 - $xpath = new DOMXPath($dom);
1787 -
1788 - // For debugging purposes
1789 - $debugEnabled = true; // Set to true to enable debugging output
1790 - $debug = function($message) use ($debugEnabled) {
1791 - if ($debugEnabled) {
1792 - //error_log('[MXCHAT-EXTRACT-DEBUG] ' . $message);
1793 - }
1794 - };
1795 -
1796 - // Direct targeting for Gerow theme posts
1797 - $post_text = $xpath->query('//div[contains(@class, "post-text")]');
1798 - if ($post_text && $post_text->length > 0) {
1799 - $debug("Found post-text directly");
1800 - $content = '';
1801 - foreach ($post_text as $node) {
1802 - $content .= $dom->saveHTML($node);
1803 - }
1804 - if (!empty($content)) {
1805 - $debug("Returning post-text content");
1806 - return $content;
1807 - }
1808 - }
1809 -
1810 - // Try to get the blog details content which contains the post-text
1811 - $blog_details = $xpath->query('//div[contains(@class, "blog-details-content")]');
1812 - if ($blog_details && $blog_details->length > 0) {
1813 - $debug("Found blog-details-content");
1814 - $content = '';
1815 - foreach ($blog_details as $node) {
1816 - $content .= $dom->saveHTML($node);
1817 - }
1818 - if (!empty($content)) {
1819 - $debug("Returning blog-details-content");
1820 - return $content;
1821 - }
1822 - }
1823 -
1824 - // Try to get the article which contains the blog details
1825 - $article = $xpath->query('//article[contains(@class, "blog-details-wrap")]');
1826 - if ($article && $article->length > 0) {
1827 - $debug("Found article with blog-details-wrap");
1828 - $content = '';
1829 - foreach ($article as $node) {
1830 - $content .= $dom->saveHTML($node);
1831 - }
1832 - if (!empty($content)) {
1833 - $debug("Returning article content");
1834 - return $content;
1835 - }
1836 - }
1837 -
1838 - // Try even broader with the blog-item-wrap
1839 - $blog_item = $xpath->query('//div[contains(@class, "blog-item-wrap")]');
1840 - if ($blog_item && $blog_item->length > 0) {
1841 - $debug("Found blog-item-wrap");
1842 - $content = '';
1843 - foreach ($blog_item as $node) {
1844 - $content .= $dom->saveHTML($node);
1845 - }
1846 - if (!empty($content)) {
1847 - $debug("Returning blog-item-wrap content");
1848 - return $content;
1849 - }
1850 - }
1851 -
1852 - // Specific Gerow theme path
1853 - $gerow_path = $xpath->query('//section[contains(@class, "blog-area")]//div[contains(@class, "post-text")]');
1854 - if ($gerow_path && $gerow_path->length > 0) {
1855 - $debug("Found Gerow theme path to post-text");
1856 - $content = '';
1857 - foreach ($gerow_path as $node) {
1858 - $content .= $dom->saveHTML($node);
1859 - }
1860 - if (!empty($content)) {
1861 - $debug("Returning Gerow post-text content");
1862 - return $content;
1863 - }
1864 - }
1865 -
1866 - // Generic blog post selectors
1867 - $selectors = [
1868 - // Blog post specific selectors
1869 - '//div[contains(@class, "post-text")]',
1870 - '//article[contains(@class, "blog-post-item")]//div[contains(@class, "post-text")]',
1871 - '//div[contains(@class, "blog-details-content")]',
1872 - '//article[contains(@class, "blog-details-wrap")]',
1873 - '//div[contains(@class, "entry-content")]',
1874 - '//div[contains(@class, "blog-content")]',
1875 - '//div[contains(@class, "blog-item-wrap")]',
1876 -
1877 - // More general content selectors
1878 - '//div[contains(@class, "page__content")]',
1879 - '//div[contains(@class, "elementor-widget-container")]',
1880 - '//div[contains(@class, "elementor-text-editor")]',
1881 - '//div[contains(@class, "elementor-widget-text-editor")]',
1882 - '//*[contains(@class, "entry-content")]',
1883 - '//*[contains(@class, "post-content")]',
1884 - '//*[contains(@class, "article-content")]',
1885 - '//*[@id="content"]',
1886 - '//*[@id="main-content"]',
1887 - '//section[contains(@class, "blog-area")]',
1888 - '//article',
1889 - '//main',
1890 - '//div[contains(@class, "content")]'
1891 - ];
1892 -
1893 - // First handle Elementor content - get only leaf widget containers to avoid duplicates
1894 - $debug("Checking for Elementor content");
1895 - // Get widget containers that are direct children of widgets (not nested inside other widget containers)
1896 - $elementor_widgets = $xpath->query('//div[contains(@class, "elementor-widget")]//div[contains(@class, "elementor-widget-container")]');
1897 - if ($elementor_widgets && $elementor_widgets->length > 0) {
1898 - $debug("Found Elementor widgets");
1899 - $seen_content = array(); // Track seen content to avoid duplicates
1900 - $combined_content = '';
1901 - foreach ($elementor_widgets as $widget) {
1902 - $widget_content = $dom->saveHTML($widget);
1903 - if (!empty($widget_content)) {
1904 - // Create a hash of the content to detect duplicates
1905 - $content_hash = md5($widget_content);
1906 - if (!isset($seen_content[$content_hash])) {
1907 - $seen_content[$content_hash] = true;
1908 - $combined_content .= $widget_content;
1909 - }
1910 - }
1911 - }
1912 - if (!empty($combined_content)) {
1913 - $debug("Returning Elementor content");
1914 - return $combined_content;
1915 - }
1916 - }
1917 -
1918 - // Try standard selectors one by one
1919 - foreach ($selectors as $selector) {
1920 - $debug("Trying selector: " . $selector);
1921 - $nodes = $xpath->query($selector);
1922 - if ($nodes && $nodes->length > 0) {
1923 - $debug("Found " . $nodes->length . " matches for selector: " . $selector);
1924 - // Only take the FIRST matching node to avoid duplicate content
1925 - // (pages often have nested or multiple containers with same class)
1926 - $content = $dom->saveHTML($nodes->item(0));
1927 - if (!empty($content)) {
1928 - $debug("Returning content from selector: " . $selector . " (first match only)");
1929 - return $content;
1930 - }
1931 - }
1932 - }
1933 -
1934 - // Manual regex fallback for post-text if DOM methods fail
1935 - $debug("Trying regex fallback");
1936 - if (preg_match('/<div class="post-text">(.*?)<\/div>\s*<\/div>\s*<\/div>/s', $html, $matches)) {
1937 - $debug("Found post-text via regex");
1938 - return '<div class="post-text">' . $matches[1] . '</div>';
1939 - }
1940 -
1941 - // Try to extract the blog section as a whole
1942 - $blog_section = $xpath->query('//section[contains(@class, "blog-area")]');
1943 - if ($blog_section && $blog_section->length > 0) {
1944 - $debug("Found blog-area section");
1945 - $content = '';
1946 - foreach ($blog_section as $node) {
1947 - $content .= $dom->saveHTML($node);
1948 - }
1949 - if (!empty($content)) {
1950 - $debug("Returning blog-area section content");
1951 - return $content;
1952 - }
1953 - }
1954 -
1955 - // Generic container selectors for non-CMS sites (like .asp pages)
1956 - $debug("Trying generic container selectors");
1957 - $generic_selectors = [
1958 - '//div[@id="main"]',
1959 - '//div[@id="wrapper"]',
1960 - '//div[@id="page"]',
1961 - '//div[@id="site-content"]',
1962 - '//div[contains(@class, "main-content")]',
1963 - '//div[contains(@class, "page-content")]',
1964 - '//div[contains(@class, "site-content")]',
1965 - ];
1966 -
1967 - foreach ($generic_selectors as $selector) {
1968 - $debug("Trying generic selector: " . $selector);
1969 - $nodes = $xpath->query($selector);
1970 - if ($nodes && $nodes->length > 0) {
1971 - $content = $dom->saveHTML($nodes->item(0));
1972 - if (!empty($content)) {
1973 - $debug("Returning content from generic selector: " . $selector);
1974 - return $content;
1975 - }
1976 - }
1977 - }
1978 -
1979 - // Paragraph-based content detection - find regions with substantial text
1980 - $debug("Trying paragraph-based content detection");
1981 - $paragraphs = $xpath->query('//p[string-length(normalize-space()) > 50]');
1982 - if ($paragraphs && $paragraphs->length >= 3) {
1983 - $debug("Found " . $paragraphs->length . " substantial paragraphs");
1984 - // Collect all substantial paragraphs and their content
1985 - $paragraph_content = '';
1986 - foreach ($paragraphs as $p) {
1987 - $paragraph_content .= $dom->saveHTML($p) . "\n";
1988 - }
1989 - if (!empty($paragraph_content)) {
1990 - $debug("Returning paragraph-based content");
1991 - return $paragraph_content;
1992 - }
1993 - }
1994 -
1995 - // Improved body fallback - strip nav/header/footer elements first
1996 - $debug("Using improved body fallback");
1997 - $body = $dom->getElementsByTagName('body');
1998 - if ($body->length > 0) {
1999 - // Clone the body to avoid modifying the original DOM
2000 - $body_clone = $body->item(0)->cloneNode(true);
2001 -
2002 - // Remove common non-content elements by tag name
2003 - $remove_tags = ['nav', 'header', 'footer', 'aside', 'script', 'style', 'noscript'];
2004 - foreach ($remove_tags as $tag) {
2005 - $elements = $body_clone->getElementsByTagName($tag);
2006 - // Iterate backwards to safely remove elements
2007 - for ($i = $elements->length - 1; $i >= 0; $i--) {
2008 - $el = $elements->item($i);
2009 - if ($el && $el->parentNode) {
2010 - $el->parentNode->removeChild($el);
2011 - }
2012 - }
2013 - }
2014 -
2015 - // Remove elements with common non-content class names using XPath on the cloned body
2016 - $temp_dom = new DOMDocument();
2017 - @$temp_dom->appendChild($temp_dom->importNode($body_clone, true));
2018 - $temp_xpath = new DOMXPath($temp_dom);
2019 -
2020 - $remove_class_patterns = [
2021 - '//*[contains(@class, "nav")]',
2022 - '//*[contains(@class, "menu")]',
2023 - '//*[contains(@class, "sidebar")]',
2024 - '//*[contains(@class, "footer")]',
2025 - '//*[contains(@class, "header")]',
2026 - '//*[contains(@id, "nav")]',
2027 - '//*[contains(@id, "menu")]',
2028 - '//*[contains(@id, "sidebar")]',
2029 - '//*[contains(@id, "footer")]',
2030 - '//*[contains(@id, "header")]',
2031 - ];
2032 -
2033 - foreach ($remove_class_patterns as $pattern) {
2034 - $elements = $temp_xpath->query($pattern);
2035 - if ($elements) {
2036 - for ($i = $elements->length - 1; $i >= 0; $i--) {
2037 - $el = $elements->item($i);
2038 - if ($el && $el->parentNode) {
2039 - $el->parentNode->removeChild($el);
2040 - }
2041 - }
2042 - }
2043 - }
2044 -
2045 - $cleaned_content = $temp_dom->saveHTML();
2046 - if (!empty($cleaned_content)) {
2047 - $debug("Returning cleaned body content");
2048 - return $cleaned_content;
2049 - }
2050 - }
2051 -
2052 - // Last resort: return the original HTML
2053 - $debug("Returning original HTML");
2054 - return $html;
2055 - } catch (Exception $e) {
2056 - //error_log('[MXCHAT-ERROR] Content extraction failed: ' . $e->getMessage());
2057 - return $html; // Return original HTML if parsing fails
2058 - } finally {
2059 - libxml_clear_errors();
2060 - }
2061 -}
2062 -public function mxchat_get_sitemap_processing_status($sitemap_url) {
2063 - $sitemap_url = esc_url_raw($sitemap_url);
2064 - $status_key = sanitize_key('mxchat_sitemap_status_' . md5($sitemap_url));
2065 - $status = get_transient($status_key);
2066 -
2067 - if (!$status || !is_array($status)) {
2068 - return false;
2069 - }
2070 -
2071 - // Auto-complete check: if all URLs are processed but status isn't complete
2072 - if (isset($status['processed_urls']) && isset($status['total_urls']) &&
2073 - $status['processed_urls'] >= $status['total_urls'] &&
2074 - isset($status['status']) && $status['status'] !== 'complete' &&
2075 - $status['status'] !== 'error') {
2076 -
2077 - // Mark as complete
2078 - $status['status'] = 'complete';
2079 - $status['processed_urls'] = $status['total_urls']; // Ensure exact match
2080 -
2081 - // Update the transient with the corrected status
2082 - set_transient($status_key, $status, DAY_IN_SECONDS);
2083 - }
2084 -
2085 - return array(
2086 - 'total_urls' => absint($status['total_urls']),
2087 - 'processed_urls' => absint($status['processed_urls']),
2088 - 'failed_urls' => absint($status['failed_urls'] ?? 0),
2089 - 'percentage' => ($status['total_urls'] > 0)
2090 - ? round((absint($status['processed_urls']) / absint($status['total_urls'])) * 100)
2091 - : 0,
2092 - 'status' => sanitize_text_field($status['status']),
2093 - 'last_update' => human_time_diff(absint($status['last_update']), time()) . ' ' . esc_html__('ago', 'mxchat'),
2094 - 'error' => isset($status['error']) ? sanitize_text_field($status['error']) : '',
2095 - 'last_error' => isset($status['last_error']) ? sanitize_text_field($status['last_error']) : '',
2096 - 'failed_urls_list' => isset($status['failed_urls_list']) ? $status['failed_urls_list'] : array()
2097 - );
2098 -}
2099 -
2100 -public function mxchat_ajax_get_status_updates() {
2101 - try {
2102 - // Verify the request
2103 - check_ajax_referer('mxchat_status_nonce', 'nonce');
2104 -
2105 - // Get active queue IDs
2106 - $sitemap_queue_id = get_transient('mxchat_active_queue_sitemap');
2107 - $pdf_queue_id = get_transient('mxchat_active_queue_pdf');
2108 -
2109 - $sitemap_status = false;
2110 - $pdf_status = false;
2111 -
2112 - // Get sitemap queue status
2113 - if ($sitemap_queue_id) {
2114 - $sitemap_status = $this->mxchat_get_queue_status_data($sitemap_queue_id, 'sitemap');
2115 - }
2116 -
2117 - // Get PDF queue status
2118 - if ($pdf_queue_id) {
2119 - $pdf_status = $this->mxchat_get_queue_status_data($pdf_queue_id, 'pdf');
2120 - }
2121 -
2122 - $is_active_processing =
2123 - ($sitemap_status && $sitemap_status['status'] === 'processing') ||
2124 - ($pdf_status && $pdf_status['status'] === 'processing');
2125 -
2126 - // Return JSON response with the status data
2127 - wp_send_json(array(
2128 - 'pdf_status' => $pdf_status,
2129 - 'sitemap_status' => $sitemap_status,
2130 - 'is_processing' => $is_active_processing,
2131 - 'sitemap_queue_id' => $sitemap_queue_id,
2132 - 'pdf_queue_id' => $pdf_queue_id
2133 - ));
2134 -
2135 - } catch (Exception $e) {
2136 - //error_log('MxChat Status Update Error: ' . $e->getMessage());
2137 -
2138 - wp_send_json_error(array(
2139 - 'message' => 'Error getting status updates: ' . $e->getMessage(),
2140 - 'status' => 'error'
2141 - ));
2142 - }
2143 -}
2144 -
2145 -/**
2146 - * Helper function to get queue status data
2147 - */
2148 -private function mxchat_get_queue_status_data($queue_id, $type = 'sitemap') {
2149 - global $wpdb;
2150 - $table_name = $wpdb->prefix . 'mxchat_processing_queue';
2151 -
2152 - // Get counts by status
2153 - $counts = $wpdb->get_results($wpdb->prepare(
2154 - "SELECT status, COUNT(*) as count
2155 - FROM $table_name
2156 - WHERE queue_id = %s
2157 - GROUP BY status",
2158 - $queue_id
2159 - ), OBJECT_K);
2160 -
2161 - $total = 0;
2162 - $completed = 0;
2163 - $failed = 0;
2164 - $processing = 0;
2165 - $pending = 0;
2166 -
2167 - foreach ($counts as $status => $data) {
2168 - $count = absint($data->count);
2169 - $total += $count;
2170 -
2171 - switch ($status) {
2172 - case 'completed':
2173 - $completed = $count;
2174 - break;
2175 - case 'failed':
2176 - $failed = $count;
2177 - break;
2178 - case 'processing':
2179 - $processing = $count;
2180 - break;
2181 - case 'pending':
2182 - $pending = $count;
2183 - break;
2184 - }
2185 - }
2186 -
2187 - if ($total === 0) {
2188 - return false;
2189 - }
2190 -
2191 - // Calculate percentage
2192 - $percentage = round((($completed + $failed) / $total) * 100);
2193 -
2194 - // Get failed items details (limit to 50)
2195 - $failed_items = array();
2196 - if ($failed > 0) {
2197 - $failed_results = $wpdb->get_results($wpdb->prepare(
2198 - "SELECT item_type, item_data, error_message, attempts, completed_at
2199 - FROM $table_name
2200 - WHERE queue_id = %s
2201 - AND status = 'failed'
2202 - AND attempts >= max_attempts
2203 - ORDER BY id DESC
2204 - LIMIT 50",
2205 - $queue_id
2206 - ));
2207 -
2208 - foreach ($failed_results as $item) {
2209 - $data = json_decode($item->item_data, true);
2210 - $url = $type === 'pdf' ? ($data['pdf_url'] ?? '') : ($data['url'] ?? '');
2211 -
2212 - $failed_items[] = array(
2213 - 'url' => $url,
2214 - 'page' => $type === 'pdf' ? ($data['page_number'] ?? 0) : 0,
2215 - 'error' => $item->error_message,
2216 - 'retries' => $item->attempts,
2217 - 'time' => strtotime($item->completed_at)
2218 - );
2219 - }
2220 - }
2221 -
2222 - // Get queue metadata
2223 - $source_url = $this->mxchat_get_queue_meta($queue_id, 'source_url');
2224 -
2225 - // Determine if queue is complete
2226 - $is_complete = ($pending === 0 && $processing === 0);
2227 -
2228 - // Get last update time
2229 - $last_update = $wpdb->get_var($wpdb->prepare(
2230 - "SELECT MAX(UNIX_TIMESTAMP(COALESCE(completed_at, started_at, created_at)))
2231 - FROM $table_name
2232 - WHERE queue_id = %s",
2233 - $queue_id
2234 - ));
2235 -
2236 - $last_update_text = $last_update ? human_time_diff($last_update, time()) . ' ' . __('ago', 'mxchat') : __('Just now', 'mxchat');
2237 -
2238 - // Format based on type
2239 - if ($type === 'pdf') {
2240 - return array(
2241 - 'total_pages' => $total,
2242 - 'processed_pages' => $completed + $failed,
2243 - 'failed_pages' => $failed,
2244 - 'percentage' => $percentage,
2245 - 'status' => $is_complete ? 'complete' : 'processing',
2246 - 'last_update' => $last_update_text,
2247 - 'failed_pages_list' => $failed_items,
2248 - 'pdf_url' => $source_url,
2249 - 'queue_id' => $queue_id
2250 - );
2251 - } else {
2252 - return array(
2253 - 'total_urls' => $total,
2254 - 'processed_urls' => $completed + $failed,
2255 - 'failed_urls' => $failed,
2256 - 'percentage' => $percentage,
2257 - 'status' => $is_complete ? 'complete' : 'processing',
2258 - 'last_update' => $last_update_text,
2259 - 'failed_urls_list' => $failed_items,
2260 - 'sitemap_url' => $source_url,
2261 - 'queue_id' => $queue_id
2262 - );
2263 - }
2264 -}
2265 -
2266 -/**
2267 - * Public method to get processing status for both sitemap and PDF queues
2268 - * Used by admin pages to display processing status
2269 - *
2270 - * @return array Array with 'sitemap_status', 'pdf_status', and 'is_processing' keys
2271 - */
2272 -public function mxchat_get_processing_statuses() {
2273 - $sitemap_queue_id = get_transient('mxchat_active_queue_sitemap');
2274 - $pdf_queue_id = get_transient('mxchat_active_queue_pdf');
2275 -
2276 - $sitemap_status = false;
2277 - $pdf_status = false;
2278 -
2279 - if ($sitemap_queue_id) {
2280 - $sitemap_status = $this->mxchat_get_queue_status_data($sitemap_queue_id, 'sitemap');
2281 - }
2282 -
2283 - if ($pdf_queue_id) {
2284 - $pdf_status = $this->mxchat_get_queue_status_data($pdf_queue_id, 'pdf');
2285 - }
2286 -
2287 - $is_processing =
2288 - ($sitemap_status && $sitemap_status['status'] === 'processing') ||
2289 - ($pdf_status && $pdf_status['status'] === 'processing');
2290 -
2291 - return array(
2292 - 'sitemap_status' => $sitemap_status,
2293 - 'pdf_status' => $pdf_status,
2294 - 'is_processing' => $is_processing
2295 - );
2296 -}
2297 -
2298 -/**
2299 - * AJAX handler to get recent knowledge entries for real-time table updates
2300 - * UPDATED: Now supports both WordPress DB and Pinecone data sources
2301 - */
2302 -public function ajax_mxchat_get_recent_entries() {
2303 - check_ajax_referer('mxchat_entries_nonce', 'nonce');
2304 -
2305 - if (!current_user_can('manage_options')) {
2306 - wp_send_json_error(array('message' => 'Unauthorized'));
2307 - return;
2308 - }
2309 -
2310 - global $wpdb;
2311 - $table_name = $wpdb->prefix . 'mxchat_system_prompt_content';
2312 -
2313 - // Get parameters
2314 - $last_id = isset($_POST['last_id']) ? absint($_POST['last_id']) : 0;
2315 - $limit = isset($_POST['limit']) ? min(absint($_POST['limit']), 50) : 10;
2316 - $bot_id = isset($_POST['bot_id']) ? sanitize_text_field($_POST['bot_id']) : 'default';
2317 -
2318 - // Check if Pinecone is enabled for this bot
2319 - $pinecone_manager = $this->mxchat_get_pinecone_manager();
2320 - $pinecone_options = $pinecone_manager ? $pinecone_manager->mxchat_get_bot_pinecone_options($bot_id) : array();
2321 - $use_pinecone = ($pinecone_options['mxchat_use_pinecone'] ?? '0') === '1';
2322 - $has_pinecone_api = !empty($pinecone_options['mxchat_pinecone_api_key']);
2323 -
2324 - if ($use_pinecone && $has_pinecone_api) {
2325 - // PINECONE DATA SOURCE - Get grouped entry count (not raw vector count)
2326 - // Use mxchat_fetch_pinecone_records which returns total_unique_entries
2327 - $records = $pinecone_manager->mxchat_fetch_pinecone_records($pinecone_options, '', 1, 10, $bot_id, '');
2328 - $total_count = $records['total'] ?? 0;
2329 -
2330 - // For Pinecone, we don't return individual entries during polling
2331 - // (entries are already displayed on page load via mxchat_fetch_pinecone_records)
2332 - // We just return the updated count
2333 - wp_send_json_success(array(
2334 - 'entries' => array(),
2335 - 'total_count' => absint($total_count),
2336 - 'max_id' => $last_id,
2337 - 'data_source' => 'pinecone'
2338 - ));
2339 - return;
2340 - }
2341 -
2342 - // WORDPRESS DB DATA SOURCE
2343 - // Build query to get entries newer than last_id
2344 - $where_clauses = array('1=1');
2345 - $where_values = array();
2346 -
2347 - if ($last_id > 0) {
2348 - $where_clauses[] = 'id > %d';
2349 - $where_values[] = $last_id;
2350 - }
2351 -
2352 - // Note: WordPress DB table doesn't have bot_id column
2353 - // Multi-bot filtering is handled via Pinecone namespaces
2354 -
2355 - $where_sql = implode(' AND ', $where_clauses);
2356 -
2357 - // Get recent entries
2358 - $query = "SELECT id, article_content, source_url, timestamp
2359 - FROM $table_name
2360 - WHERE $where_sql
2361 - ORDER BY id DESC
2362 - LIMIT %d";
2363 -
2364 - $where_values[] = $limit;
2365 -
2366 - $entries = $wpdb->get_results($wpdb->prepare($query, $where_values));
2367 -
2368 - // Get total count of GROUPED entries (by source_url) - matches pagination display
2369 - // Count unique source_urls (excluding mxchat:// internal refs) + count of ungrouped rows
2370 - $total_count = $wpdb->get_var(
2371 - "SELECT (SELECT COUNT(DISTINCT source_url) FROM {$table_name} WHERE source_url != '' AND source_url NOT LIKE 'mxchat://%') +
2372 - (SELECT COUNT(*) FROM {$table_name} WHERE source_url = '' OR source_url IS NULL OR source_url LIKE 'mxchat://%')"
2373 - );
2374 -
2375 - // Format entries for response
2376 - $formatted_entries = array();
2377 - $preview_length = 150;
2378 - foreach ($entries as $entry) {
2379 - // Parse chunk metadata using the proper chunker method (same as initial page load)
2380 - if (class_exists('MxChat_Chunker')) {
2381 - $chunk_meta = MxChat_Chunker::parse_stored_chunk($entry->article_content);
2382 - $display_content = $chunk_meta['text'];
2383 - $chunk_metadata = $chunk_meta['metadata'];
2384 - } else {
2385 - $display_content = $entry->article_content;
2386 - $chunk_metadata = array();
2387 - }
2388 -
2389 - $content_preview = mb_strlen($display_content) > $preview_length
2390 - ? mb_substr($display_content, 0, $preview_length) . '...'
2391 - : $display_content;
2392 -
2393 - $formatted_entries[] = array(
2394 - 'id' => $entry->id,
2395 - 'preview' => esc_html($content_preview),
2396 - 'full_content' => wp_kses_post(wpautop($display_content)),
2397 - 'content_length' => mb_strlen($display_content),
2398 - 'preview_length' => $preview_length,
2399 - 'source_url' => $entry->source_url,
2400 - 'has_link' => !empty($entry->source_url) && strpos($entry->source_url, 'mxchat://') !== 0,
2401 - 'chunk_metadata' => $chunk_metadata,
2402 - 'bot_id' => $entry->bot_id ?? 'default',
2403 - 'edit_nonce' => wp_create_nonce('mxchat_edit_entry_nonce'),
2404 - 'delete_nonce' => wp_create_nonce('mxchat_delete_prompt_nonce')
2405 - );
2406 - }
2407 -
2408 - wp_send_json_success(array(
2409 - 'entries' => $formatted_entries,
2410 - 'total_count' => absint($total_count),
2411 - 'max_id' => !empty($entries) ? $entries[0]->id : $last_id,
2412 - 'data_source' => 'wordpress'
2413 - ));
2414 -}
2415 -
2416 -/**
2417 - * Get Pinecone total count from stats API
2418 - * Helper function for ajax_mxchat_get_recent_entries
2419 - */
2420 -private function mxchat_get_pinecone_count_from_stats($pinecone_options) {
2421 - $api_key = $pinecone_options['mxchat_pinecone_api_key'] ?? '';
2422 - $host = $pinecone_options['mxchat_pinecone_host'] ?? '';
2423 - $namespace = $pinecone_options['mxchat_pinecone_namespace'] ?? '';
2424 -
2425 - if (empty($api_key) || empty($host)) {
2426 - return 0;
2427 - }
2428 -
2429 - try {
2430 - $stats_url = "https://{$host}/describe_index_stats";
2431 -
2432 - $response = wp_remote_post($stats_url, array(
2433 - 'headers' => array(
2434 - 'Api-Key' => $api_key,
2435 - 'Content-Type' => 'application/json'
2436 - ),
2437 - 'body' => '{}',
2438 - 'timeout' => 10
2439 - ));
2440 -
2441 - if (!is_wp_error($response) && wp_remote_retrieve_response_code($response) === 200) {
2442 - $body = wp_remote_retrieve_body($response);
2443 - $stats_data = json_decode($body, true);
2444 -
2445 - // If namespace is specified, get count from that specific namespace
2446 - if (!empty($namespace) && isset($stats_data['namespaces'][$namespace]['vectorCount'])) {
2447 - return intval($stats_data['namespaces'][$namespace]['vectorCount']);
2448 - }
2449 -
2450 - // If no namespace specified or namespace not found in response, use total
2451 - return intval($stats_data['totalVectorCount'] ?? 0);
2452 - }
2453 -
2454 - return 0;
2455 -
2456 - } catch (Exception $e) {
2457 - return 0;
2458 - }
2459 -}
2460 -
2461 -/**
2462 - * AJAX handler to refresh Pinecone entries table via AJAX
2463 - * Returns the table HTML for updating the UI without a full page reload
2464 - */
2465 -public function ajax_mxchat_refresh_pinecone_entries() {
2466 - check_ajax_referer('mxchat_entries_nonce', 'nonce');
2467 -
2468 - if (!current_user_can('manage_options')) {
2469 - wp_send_json_error(array('message' => 'Unauthorized'));
2470 - return;
2471 - }
2472 -
2473 - $bot_id = isset($_POST['bot_id']) ? sanitize_text_field($_POST['bot_id']) : 'default';
2474 - $page = isset($_POST['page']) ? absint($_POST['page']) : 1;
2475 - $per_page = 25;
2476 - $search_query = isset($_POST['search']) ? sanitize_text_field($_POST['search']) : '';
2477 - $content_type_filter = isset($_POST['content_type']) ? sanitize_key($_POST['content_type']) : '';
2478 -
2479 - // Get Pinecone manager and options
2480 - $pinecone_manager = $this->mxchat_get_pinecone_manager();
2481 - if (!$pinecone_manager) {
2482 - wp_send_json_error(array('message' => 'Pinecone manager not available'));
2483 - return;
2484 - }
2485 -
2486 - $pinecone_options = $pinecone_manager->mxchat_get_bot_pinecone_options($bot_id);
2487 - $use_pinecone = ($pinecone_options['mxchat_use_pinecone'] ?? '0') === '1';
2488 - $pinecone_api_key = $pinecone_options['mxchat_pinecone_api_key'] ?? '';
2489 -
2490 - if (!$use_pinecone || empty($pinecone_api_key)) {
2491 - wp_send_json_error(array('message' => 'Pinecone not configured'));
2492 - return;
2493 - }
2494 -
2495 - // Fetch records from Pinecone
2496 - $records = $pinecone_manager->mxchat_fetch_pinecone_records($pinecone_options, $search_query, $page, $per_page, $bot_id, $content_type_filter);
2497 - $prompts = $records['data'] ?? array();
2498 - $total_records = $records['total'] ?? 0;
2499 -
2500 - // Preprocess Pinecone records — set chunk_metadata and display_content
2501 - // (matches admin-knowledge-page.php preprocessing)
2502 - foreach ($prompts as $prompt) {
2503 - if (isset($prompt->chunk_index) && $prompt->chunk_index !== null) {
2504 - $prompt->chunk_metadata = array(
2505 - 'chunk_index' => intval($prompt->chunk_index),
2506 - 'total_chunks' => isset($prompt->total_chunks) ? intval($prompt->total_chunks) : null,
2507 - 'is_chunked' => isset($prompt->is_chunked) ? (bool) $prompt->is_chunked : true,
2508 - 'source_url' => $prompt->source_url ?? ''
2509 - );
2510 - $prompt->display_content = $prompt->article_content;
2511 - } else {
2512 - $prompt->chunk_metadata = array();
2513 - $prompt->display_content = $prompt->article_content ?? '';
2514 - }
2515 - }
2516 -
2517 - // Group prompts by source_url
2518 - $grouped_prompts = array();
2519 - foreach ($prompts as $prompt) {
2520 - $source_url = '';
2521 - if (!empty($prompt->chunk_metadata['source_url'])) {
2522 - $source_url = $prompt->chunk_metadata['source_url'];
2523 - } elseif (!empty($prompt->source_url)) {
2524 - $source_url = $prompt->source_url;
2525 - }
2526 -
2527 - if (!empty($source_url)) {
2528 - if (!isset($grouped_prompts[$source_url])) {
2529 - $grouped_prompts[$source_url] = array();
2530 - }
2531 - $grouped_prompts[$source_url][] = $prompt;
2532 - } else {
2533 - $grouped_prompts['_ungrouped_' . $prompt->id] = array($prompt);
2534 - }
2535 - }
2536 -
2537 - // Sort each group by chunk_index
2538 - foreach ($grouped_prompts as $source_url => &$group) {
2539 - usort($group, function($a, $b) {
2540 - $index_a = isset($a->chunk_metadata['chunk_index']) ? intval($a->chunk_metadata['chunk_index']) : 0;
2541 - $index_b = isset($b->chunk_metadata['chunk_index']) ? intval($b->chunk_metadata['chunk_index']) : 0;
2542 - return $index_a - $index_b;
2543 - });
2544 - }
2545 - unset($group);
2546 -
2547 - // Build HTML for the table rows - must match admin-knowledge-page.php structure exactly
2548 - ob_start();
2549 - $display_index = 0;
2550 - $current_page = $page;
2551 - $data_source = 'pinecone';
2552 - $current_bot_id = $bot_id;
2553 - $preview_length = 150;
2554 -
2555 - if (empty($grouped_prompts)) {
2556 - echo '<tr><td colspan="4" style="padding: 40px; text-align: center; color: var(--mxch-text-muted);">';
2557 - esc_html_e('No knowledge entries found in Pinecone.', 'mxchat');
2558 - echo '</td></tr>';
2559 - } else {
2560 - foreach ($grouped_prompts as $source_url => $group) {
2561 - $chunk_count = count($group);
2562 - $first_prompt = $group[0];
2563 - $display_index++;
2564 -
2565 - if ($chunk_count > 1) {
2566 - // Multiple chunks - show grouped row with expand button
2567 - $group_id = 'group-' . md5($source_url);
2568 - ?>
2569 - <tr id="prompt-<?php echo esc_attr($first_prompt->id); ?>"
2570 - class="mxchat-chunk-group-header"
2571 - data-source="<?php echo esc_attr($data_source); ?>"
2572 - data-group-id="<?php echo esc_attr($group_id); ?>"
2573 - style="border-bottom: 1px solid var(--mxch-card-border); background: rgba(33, 150, 243, 0.02);">
2574 - <td style="padding: 12px 16px; text-align: center;">
2575 - <input type="checkbox"
2576 - class="mxchat-entry-checkbox"
2577 - data-entry-id="<?php echo esc_attr($first_prompt->id); ?>"
2578 - data-source="<?php echo esc_attr($data_source); ?>"
2579 - data-source-url="<?php echo esc_attr($source_url); ?>"
2580 - data-is-group="true"
2581 - data-chunk-count="<?php echo esc_attr($chunk_count); ?>">
2582 - </td>
2583 - <td style="padding: 12px 16px; font-size: 13px;">
2584 - <?php echo esc_html($display_index + (($current_page - 1) * $per_page)); ?>
2585 - </td>
2586 - <td class="mxchat-content-cell" style="padding: 12px 16px; font-size: 13px;">
2587 - <div class="mxchat-chunk-group-info">
2588 - <button type="button" class="mxchat-chunk-toggle" data-group-id="<?php echo esc_attr($group_id); ?>">
2589 - <span class="dashicons dashicons-arrow-right-alt2"></span>
2590 - </button>
2591 - <span class="mxchat-chunk-badge"><?php echo esc_html($chunk_count); ?> <?php esc_html_e('chunks', 'mxchat'); ?></span>
2592 - <span class="mxchat-chunk-preview">
2593 - <?php
2594 - $parent_content = isset($first_prompt->display_content) ? $first_prompt->display_content : (isset($first_prompt->article_content) ? $first_prompt->article_content : '');
2595 - $content_preview = mb_substr($parent_content, 0, 100);
2596 - echo esc_html($content_preview . '...');
2597 - ?>
2598 - </span>
2599 - </div>
2600 - </td>
2601 - <td class="mxchat-url-cell" style="padding: 12px 16px; font-size: 13px;">
2602 - <?php if (!empty($source_url) && strpos($source_url, 'mxchat://') !== 0 && strpos($source_url, '_ungrouped_') !== 0) : ?>
2603 - <a href="<?php echo esc_url($source_url); ?>" target="_blank" style="color: var(--mxch-primary); text-decoration: none;">
2604 - <span class="dashicons dashicons-external" style="font-size: 14px;"></span>
2605 - <?php esc_html_e('View Source', 'mxchat'); ?>
2606 - </a>
2607 - <?php else : ?>
2608 - <span style="color: var(--mxch-text-muted);"><?php esc_html_e('Manual Content', 'mxchat'); ?></span>
2609 - <?php endif; ?>
2610 - </td>
2611 - <td class="mxchat-actions-cell" style="padding: 12px 16px; white-space: nowrap;">
2612 - <?php if ($data_source !== 'pinecone') : ?>
2613 - <button type="button"
2614 - class="mxch-btn mxch-btn-ghost mxch-btn-sm mxchat-edit-entry-btn"
2615 - data-source-url="<?php echo esc_attr($source_url); ?>"
2616 - data-entry-id="<?php echo esc_attr($first_prompt->id); ?>"
2617 - data-data-source="<?php echo esc_attr($data_source); ?>"
2618 - data-bot-id="<?php echo esc_attr($current_bot_id); ?>"
2619 - data-nonce="<?php echo wp_create_nonce('mxchat_edit_entry_nonce'); ?>"
2620 - title="<?php esc_attr_e('Edit content', 'mxchat'); ?>">
2621 - <span class="dashicons dashicons-edit" style="font-size: 14px;"></span>
2622 - </button>
2623 - <?php endif; ?>
2624 - <button type="button"
2625 - class="mxch-btn mxch-btn-ghost mxch-btn-sm delete-button-group"
2626 - data-source-url="<?php echo esc_attr($source_url); ?>"
2627 - data-chunk-count="<?php echo esc_attr($chunk_count); ?>"
2628 - data-data-source="<?php echo esc_attr($data_source); ?>"
2629 - data-bot-id="<?php echo esc_attr($current_bot_id); ?>"
2630 - data-nonce="<?php echo wp_create_nonce('mxchat_delete_chunks_nonce'); ?>"
2631 - style="color: var(--mxch-error);"
2632 - title="<?php esc_attr_e('Delete all chunks', 'mxchat'); ?>">
2633 - <span class="dashicons dashicons-trash" style="font-size: 14px;"></span>
2634 - </button>
2635 - </td>
2636 - </tr>
2637 - <?php
2638 - // Render hidden chunk rows
2639 - foreach ($group as $chunk_index => $chunk) {
2640 - $meta_chunk_index = isset($chunk->chunk_metadata['chunk_index']) ? intval($chunk->chunk_metadata['chunk_index']) : $chunk_index;
2641 - $meta_total_chunks = isset($chunk->chunk_metadata['total_chunks']) ? intval($chunk->chunk_metadata['total_chunks']) : $chunk_count;
2642 - $content = isset($chunk->display_content) ? $chunk->display_content : (isset($chunk->article_content) ? $chunk->article_content : '');
2643 - $content_preview = mb_strlen($content) > $preview_length
2644 - ? mb_substr($content, 0, $preview_length) . '...'
2645 - : $content;
2646 - ?>
2647 - <tr id="prompt-<?php echo esc_attr($chunk->id); ?>"
2648 - class="mxchat-chunk-row <?php echo esc_attr($group_id); ?>"
2649 - data-source="<?php echo esc_attr($data_source); ?>"
2650 - style="display: none; background: #f8f9fa; border-bottom: 1px solid var(--mxch-card-border);">
2651 - <td style="padding: 12px 16px; text-align: center;">
2652 - <!-- Checkbox column placeholder for chunks (managed by group) -->
2653 - </td>
2654 - <td style="padding: 12px 16px 12px 30px; font-size: 13px;">
2655 - <!-- Hidden ID column for chunks -->
2656 - </td>
2657 - <td class="mxchat-content-cell" style="padding: 12px 16px; font-size: 13px;">
2658 - <div class="mxchat-accordion-wrapper">
2659 - <div class="mxchat-content-preview">
2660 - <span class="mxchat-chunk-indicator" style="margin-right: 10px; color: var(--mxch-text-secondary); font-size: 12px;">
2661 - <?php printf(esc_html__('Chunk %d of %d', 'mxchat'), $meta_chunk_index + 1, $meta_total_chunks); ?>
2662 - </span>
2663 - <span class="preview-text"><?php echo esc_html($content_preview); ?></span>
2664 - <?php if (mb_strlen($content) > $preview_length) : ?>
2665 - <button class="mxchat-expand-toggle" type="button">
2666 - <span class="dashicons dashicons-arrow-down-alt2"></span>
2667 - </button>
2668 - <?php endif; ?>
2669 - </div>
2670 - <div class="mxchat-content-full" style="display: none;">
2671 - <div class="content-view">
2672 - <?php
2673 - if (preg_match('/[\x{0590}-\x{05FF}]/u', $content)) {
2674 - echo '<div dir="rtl" lang="he" class="rtl-content">';
2675 - echo wp_kses_post(wpautop($content));
2676 - echo '</div>';
2677 - } else {
2678 - echo wp_kses_post(wpautop($content));
2679 - }
2680 - ?>
2681 - </div>
2682 - </div>
2683 - </div>
2684 - </td>
2685 - <td class="mxchat-url-cell" style="padding: 12px 16px; font-size: 13px;">
2686 - <span class="mxchat-chunk-label" style="color: var(--mxch-text-muted);"><?php esc_html_e('Same as parent', 'mxchat'); ?></span>
2687 - </td>
2688 - <td class="mxchat-actions-cell" style="padding: 12px 16px;">
2689 - <span class="mxchat-chunk-label" style="color: var(--mxch-text-muted);"><?php esc_html_e('Managed by group', 'mxchat'); ?></span>
2690 - </td>
2691 - </tr>
2692 - <?php
2693 - }
2694 - } else {
2695 - // Single entry - display normally with accordion
2696 - $prompt = $first_prompt;
2697 - $content = isset($prompt->display_content) ? $prompt->display_content : (isset($prompt->article_content) ? $prompt->article_content : '');
2698 - $content_preview = mb_strlen($content) > $preview_length
2699 - ? mb_substr($content, 0, $preview_length) . '...'
2700 - : $content;
2701 - ?>
2702 - <tr id="prompt-<?php echo esc_attr($prompt->id); ?>"
2703 - data-source="<?php echo esc_attr($data_source); ?>"
2704 - style="border-bottom: 1px solid var(--mxch-card-border); background: rgba(33, 150, 243, 0.02);">
2705 - <td style="padding: 12px 16px; text-align: center;">
2706 - <input type="checkbox"
2707 - class="mxchat-entry-checkbox"
2708 - data-entry-id="<?php echo esc_attr($prompt->id); ?>"
2709 - data-source="<?php echo esc_attr($data_source); ?>"
2710 - data-source-url="<?php echo esc_attr($prompt->source_url ?? ''); ?>"
2711 - data-is-group="false"
2712 - data-chunk-count="1">
2713 - </td>
2714 - <td style="padding: 12px 16px; font-size: 13px;">
2715 - <?php echo esc_html($display_index + (($current_page - 1) * $per_page)); ?>
2716 - </td>
2717 - <td class="mxchat-content-cell" style="padding: 12px 16px; font-size: 13px;">
2718 - <div class="mxchat-accordion-wrapper">
2719 - <div class="mxchat-content-preview">
2720 - <span class="preview-text"><?php echo esc_html($content_preview); ?></span>
2721 - <?php if (mb_strlen($content) > $preview_length) : ?>
2722 - <button class="mxchat-expand-toggle" type="button">
2723 - <span class="dashicons dashicons-arrow-down-alt2"></span>
2724 - </button>
2725 - <?php endif; ?>
2726 - </div>
2727 - <div class="mxchat-content-full" style="display: none;">
2728 - <div class="content-view">
2729 - <?php
2730 - if (preg_match('/[\x{0590}-\x{05FF}]/u', $content)) {
2731 - echo '<div dir="rtl" lang="he" class="rtl-content">';
2732 - echo wp_kses_post(wpautop($content));
2733 - echo '</div>';
2734 - } else {
2735 - echo wp_kses_post(wpautop($content));
2736 - }
2737 - ?>
2738 - </div>
2739 - </div>
2740 - </div>
2741 - </td>
2742 - <td class="mxchat-url-cell" style="padding: 12px 16px; font-size: 13px;">
2743 - <?php
2744 - $actual_source = $source_url;
2745 - if (strpos($source_url, '_ungrouped_') === 0) {
2746 - $actual_source = $prompt->source_url ?? '';
2747 - }
2748 - if (!empty($actual_source) && strpos($actual_source, 'mxchat://') !== 0) : ?>
2749 - <a href="<?php echo esc_url($actual_source); ?>" target="_blank" style="color: var(--mxch-primary); text-decoration: none;">
2750 - <span class="dashicons dashicons-external" style="font-size: 14px;"></span>
2751 - <?php esc_html_e('View', 'mxchat'); ?>
2752 - </a>
2753 - <?php else : ?>
2754 - <span style="color: var(--mxch-text-muted);"><?php esc_html_e('Manual', 'mxchat'); ?></span>
2755 - <?php endif; ?>
2756 - </td>
2757 - <td style="padding: 12px 16px;">
2758 - <button type="button" class="mxch-btn mxch-btn-ghost mxch-btn-sm delete-button-ajax" data-vector-id="<?php echo esc_attr($prompt->id); ?>" data-bot-id="<?php echo esc_attr($current_bot_id); ?>" data-nonce="<?php echo wp_create_nonce('mxchat_delete_pinecone_prompt_nonce'); ?>" style="color: var(--mxch-error);">
2759 - <span class="dashicons dashicons-trash" style="font-size: 14px;"></span>
2760 - </button>
2761 - </td>
2762 - </tr>
2763 - <?php
2764 - }
2765 - }
2766 - }
2767 - $html = ob_get_clean();
2768 -
2769 - // Generate pagination HTML for Pinecone
2770 - $total_pages = ceil($total_records / $per_page);
2771 - $pagination_html = '';
2772 - if ($total_pages > 1) {
2773 - $pagination_html = '<div class="mxchat-ajax-pagination" data-current-page="' . esc_attr($page) . '" data-total-pages="' . esc_attr($total_pages) . '" data-search="' . esc_attr($search_query) . '" data-content-type="' . esc_attr($content_type_filter) . '">';
2774 -
2775 - // Previous button
2776 - if ($page > 1) {
2777 - $pagination_html .= '<a href="#" class="mxchat-page-link" data-page="' . ($page - 1) . '">' . esc_html__('&laquo; Previous', 'mxchat') . '</a> ';
2778 - }
2779 -
2780 - // Page numbers
2781 - $start_page = max(1, $page - 2);
2782 - $end_page = min($total_pages, $page + 2);
2783 -
2784 - if ($start_page > 1) {
2785 - $pagination_html .= '<a href="#" class="mxchat-page-link" data-page="1">1</a> ';
2786 - if ($start_page > 2) {
2787 - $pagination_html .= '<span class="mxchat-page-dots">...</span> ';
2788 - }
2789 - }
2790 -
2791 - for ($i = $start_page; $i <= $end_page; $i++) {
2792 - if ($i == $page) {
2793 - $pagination_html .= '<span class="mxchat-page-current">' . $i . '</span> ';
2794 - } else {
2795 - $pagination_html .= '<a href="#" class="mxchat-page-link" data-page="' . $i . '">' . $i . '</a> ';
2796 - }
2797 - }
2798 -
2799 - if ($end_page < $total_pages) {
2800 - if ($end_page < $total_pages - 1) {
2801 - $pagination_html .= '<span class="mxchat-page-dots">...</span> ';
2802 - }
2803 - $pagination_html .= '<a href="#" class="mxchat-page-link" data-page="' . $total_pages . '">' . $total_pages . '</a> ';
2804 - }
2805 -
2806 - // Next button
2807 - if ($page < $total_pages) {
2808 - $pagination_html .= '<a href="#" class="mxchat-page-link" data-page="' . ($page + 1) . '">' . esc_html__('Next &raquo;', 'mxchat') . '</a>';
2809 - }
2810 -
2811 - $pagination_html .= '</div>';
2812 - }
2813 -
2814 - wp_send_json_success(array(
2815 - 'html' => $html,
2816 - 'pagination_html' => $pagination_html,
2817 - 'total_count' => $total_records,
2818 - 'total_pages' => $total_pages,
2819 - 'page' => $page,
2820 - 'per_page' => $per_page,
2821 - 'data_source' => 'pinecone'
2822 - ));
2823 -}
2824 -
2825 -/**
2826 - * AJAX handler for pagination - handles both WordPress DB and Pinecone data sources
2827 - * Returns paginated entries without requiring a full page reload
2828 - */
2829 -public function ajax_mxchat_paginate_entries() {
2830 - check_ajax_referer('mxchat_entries_nonce', 'nonce');
2831 -
2832 - if (!current_user_can('manage_options')) {
2833 - wp_send_json_error(array('message' => 'Unauthorized'));
2834 - return;
2835 - }
2836 -
2837 - $bot_id = isset($_POST['bot_id']) ? sanitize_text_field($_POST['bot_id']) : 'default';
2838 - $page = isset($_POST['page']) ? absint($_POST['page']) : 1;
2839 - $search_query = isset($_POST['search']) ? sanitize_text_field($_POST['search']) : '';
2840 - $content_type_filter = isset($_POST['content_type']) ? sanitize_key($_POST['content_type']) : '';
2841 - $per_page = 25;
2842 -
2843 - // Check if Pinecone is enabled for this bot
2844 - $pinecone_manager = $this->mxchat_get_pinecone_manager();
2845 - $pinecone_options = $pinecone_manager ? $pinecone_manager->mxchat_get_bot_pinecone_options($bot_id) : array();
2846 - $use_pinecone = ($pinecone_options['mxchat_use_pinecone'] ?? '0') === '1';
2847 - $has_pinecone_api = !empty($pinecone_options['mxchat_pinecone_api_key']);
2848 -
2849 - if ($use_pinecone && $has_pinecone_api) {
2850 - // Delegate to Pinecone pagination handler (pass search params)
2851 - $_POST['page'] = $page;
2852 - $_POST['search'] = $search_query;
2853 - $_POST['content_type'] = $content_type_filter;
2854 - $this->ajax_mxchat_refresh_pinecone_entries();
2855 - return;
2856 - }
2857 -
2858 - // WordPress DB pagination - MUST match initial page load logic exactly
2859 - global $wpdb;
2860 - $table_name = $wpdb->prefix . 'mxchat_system_prompt_content';
2861 - $offset = ($page - 1) * $per_page;
2862 -
2863 - // Build WHERE clause for search and content type filtering
2864 - $where_clauses = array();
2865 - $where_values = array();
2866 -
2867 - if ($search_query) {
2868 - $where_clauses[] = "article_content LIKE %s";
2869 - $where_values[] = '%' . $wpdb->esc_like($search_query) . '%';
2870 - }
2871 -
2872 - if ($content_type_filter) {
2873 - switch ($content_type_filter) {
2874 - case 'manual':
2875 - $where_clauses[] = "(source_url = '' OR source_url IS NULL OR source_url LIKE 'mxchat://%')";
2876 - break;
2877 - case 'pdf':
2878 - $where_clauses[] = "source_url LIKE '%.pdf'";
2879 - break;
2880 - case 'url':
2881 - $where_clauses[] = "source_url != '' AND source_url IS NOT NULL AND source_url NOT LIKE 'mxchat://%' AND source_url NOT LIKE '%.pdf'";
2882 - break;
2883 - }
2884 - }
2885 -
2886 - $where_sql = !empty($where_clauses) ? 'WHERE ' . implode(' AND ', $where_clauses) : '';
2887 -
2888 - // Count grouped entries with filters applied
2889 - if (!empty($where_values)) {
2890 - $count_args = array_merge($where_values, $where_values);
2891 - $count_query = $wpdb->prepare(
2892 - "SELECT (SELECT COUNT(DISTINCT source_url) FROM {$table_name} {$where_sql} AND source_url != '' AND source_url NOT LIKE 'mxchat://%') +
2893 - (SELECT COUNT(*) FROM {$table_name} {$where_sql} AND (source_url = '' OR source_url IS NULL OR source_url LIKE 'mxchat://%'))",
2894 - ...$count_args
2895 - );
2896 - $total_records = $wpdb->get_var($count_query);
2897 - } else if (!empty($where_sql)) {
2898 - // Content type filter only (no search), no prepared values needed
2899 - $total_records = $wpdb->get_var(
2900 - "SELECT (SELECT COUNT(DISTINCT source_url) FROM {$table_name} {$where_sql} AND source_url != '' AND source_url NOT LIKE 'mxchat://%') +
2901 - (SELECT COUNT(*) FROM {$table_name} {$where_sql} AND (source_url = '' OR source_url IS NULL OR source_url LIKE 'mxchat://%'))"
2902 - );
2903 - } else {
2904 - // No filters
2905 - $total_records = $wpdb->get_var(
2906 - "SELECT (SELECT COUNT(DISTINCT source_url) FROM {$table_name} WHERE source_url != '' AND source_url NOT LIKE 'mxchat://%') +
2907 - (SELECT COUNT(*) FROM {$table_name} WHERE source_url = '' OR source_url IS NULL OR source_url LIKE 'mxchat://%')"
2908 - );
2909 - }
2910 - $total_pages = ceil($total_records / $per_page);
2911 -
2912 - // Step 1: Get unique source_urls for this page (ordered by latest timestamp) with filters
2913 - if (!empty($where_values)) {
2914 - $query_args = array_merge($where_values, array($per_page, $offset));
2915 - $urls_query = $wpdb->prepare(
2916 - "SELECT source_url, MAX(timestamp) as latest_ts FROM {$table_name}
2917 - {$where_sql}
2918 - GROUP BY source_url ORDER BY latest_ts DESC LIMIT %d OFFSET %d",
2919 - ...$query_args
2920 - );
2921 - } else if (!empty($where_sql)) {
2922 - $urls_query = $wpdb->prepare(
2923 - "SELECT source_url, MAX(timestamp) as latest_ts FROM {$table_name}
2924 - {$where_sql}
2925 - GROUP BY source_url ORDER BY latest_ts DESC LIMIT %d OFFSET %d",
2926 - $per_page, $offset
2927 - );
2928 - } else {
2929 - $urls_query = $wpdb->prepare(
2930 - "SELECT source_url, MAX(timestamp) as latest_ts FROM {$table_name}
2931 - GROUP BY source_url ORDER BY latest_ts DESC LIMIT %d OFFSET %d",
2932 - $per_page, $offset
2933 - );
2934 - }
2935 - $page_urls = $wpdb->get_results($urls_query);
2936 -
2937 - // Step 2: Build list of source_urls to fetch
2938 - $url_list = array();
2939 - $url_order_map = array();
2940 - $order_index = 0;
2941 - foreach ($page_urls as $url_row) {
2942 - $url = $url_row->source_url;
2943 - $url_list[] = $url;
2944 - $url_order_map[$url] = $order_index++;
2945 - }
2946 -
2947 - // Step 3: Fetch all rows for these source_urls (with search filter if applicable)
2948 - $prompts = array();
2949 - if (!empty($url_list)) {
2950 - $placeholders = implode(',', array_fill(0, count($url_list), '%s'));
2951 - if ($search_query) {
2952 - // Include search filter in the final fetch
2953 - $prompts_query = $wpdb->prepare(
2954 - "SELECT id, article_content, source_url, timestamp, role_restriction
2955 - FROM {$table_name}
2956 - WHERE source_url IN ($placeholders) AND article_content LIKE %s
2957 - ORDER BY timestamp DESC",
2958 - ...array_merge($url_list, ['%' . $wpdb->esc_like($search_query) . '%'])
2959 - );
2960 - } else {
2961 - $prompts_query = $wpdb->prepare(
2962 - "SELECT id, article_content, source_url, timestamp, role_restriction
2963 - FROM {$table_name}
2964 - WHERE source_url IN ($placeholders)
2965 - ORDER BY timestamp DESC",
2966 - $url_list
2967 - );
2968 - }
2969 - $prompts = $wpdb->get_results($prompts_query);
2970 - }
2971 -
2972 - // Group prompts by source_url for chunk display
2973 - $grouped_prompts = array();
2974 - foreach ($prompts as $prompt) {
2975 - $source_url = $prompt->source_url ?? '';
2976 -
2977 - // Parse chunk metadata using the proper chunker method (same as initial page load)
2978 - if (class_exists('MxChat_Chunker')) {
2979 - $chunk_meta = MxChat_Chunker::parse_stored_chunk($prompt->article_content);
2980 - $prompt->chunk_metadata = $chunk_meta['metadata'];
2981 - $prompt->display_content = $chunk_meta['text'];
2982 - } else {
2983 - $prompt->chunk_metadata = array();
2984 - $prompt->display_content = $prompt->article_content;
2985 - }
2986 -
2987 - if (!empty($source_url) && strpos($source_url, 'mxchat://') !== 0) {
2988 - if (!isset($grouped_prompts[$source_url])) {
2989 - $grouped_prompts[$source_url] = array();
2990 - }
2991 - $grouped_prompts[$source_url][] = $prompt;
2992 - } else {
2993 - // Ungrouped entries
2994 - $grouped_prompts['_ungrouped_' . $prompt->id] = array($prompt);
2995 - }
2996 - }
2997 -
2998 - // Sort groups by the original URL order (newest first)
2999 - uksort($grouped_prompts, function($a, $b) use ($url_order_map) {
3000 - $order_a = isset($url_order_map[$a]) ? $url_order_map[$a] : PHP_INT_MAX;
3001 - $order_b = isset($url_order_map[$b]) ? $url_order_map[$b] : PHP_INT_MAX;
3002 - return $order_a - $order_b;
3003 - });
3004 -
3005 - // Sort each group internally by chunk_index
3006 - foreach ($grouped_prompts as $source_url => &$group) {
3007 - usort($group, function($a, $b) {
3008 - $index_a = isset($a->chunk_metadata['chunk_index']) ? intval($a->chunk_metadata['chunk_index']) : 0;
3009 - $index_b = isset($b->chunk_metadata['chunk_index']) ? intval($b->chunk_metadata['chunk_index']) : 0;
3010 - return $index_a - $index_b;
3011 - });
3012 - }
3013 - unset($group);
3014 -
3015 - // Build HTML for the table rows
3016 - ob_start();
3017 - $display_index = 0;
3018 - $current_page = $page;
3019 - $data_source = 'wordpress';
3020 - $current_bot_id = $bot_id;
3021 - $preview_length = 150;
3022 -
3023 - if (empty($grouped_prompts)) {
3024 - echo '<tr><td colspan="5" style="padding: 40px; text-align: center; color: var(--mxch-text-muted);">';
3025 - esc_html_e('No knowledge entries found. Use the Import Options to add content.', 'mxchat');
3026 - echo '</td></tr>';
3027 - } else {
3028 - foreach ($grouped_prompts as $source_url => $group) {
3029 - $chunk_count = count($group);
3030 - $first_prompt = $group[0];
3031 - $display_index++;
3032 -
3033 - if ($chunk_count > 1) {
3034 - // Multiple chunks - show grouped row with expand button
3035 - $group_id = 'group-' . md5($source_url);
3036 - ?>
3037 - <tr id="prompt-<?php echo esc_attr($first_prompt->id); ?>"
3038 - class="mxchat-chunk-group-header"
3039 - data-source="<?php echo esc_attr($data_source); ?>"
3040 - data-group-id="<?php echo esc_attr($group_id); ?>"
3041 - style="border-bottom: 1px solid var(--mxch-card-border);">
3042 - <td style="padding: 12px 16px; text-align: center;">
3043 - <input type="checkbox"
3044 - class="mxchat-entry-checkbox"
3045 - data-entry-id="<?php echo esc_attr($first_prompt->id); ?>"
3046 - data-source="<?php echo esc_attr($data_source); ?>"
3047 - data-source-url="<?php echo esc_attr($source_url); ?>"
3048 - data-is-group="true"
3049 - data-chunk-count="<?php echo esc_attr($chunk_count); ?>">
3050 - </td>
3051 - <td style="padding: 12px 16px; font-size: 13px;">
3052 - <?php echo esc_html($first_prompt->id); ?>
3053 - </td>
3054 - <td class="mxchat-content-cell" style="padding: 12px 16px; font-size: 13px;">
3055 - <div class="mxchat-chunk-group-info">
3056 - <button type="button" class="mxchat-chunk-toggle" data-group-id="<?php echo esc_attr($group_id); ?>">
3057 - <span class="dashicons dashicons-arrow-right-alt2"></span>
3058 - </button>
3059 - <span class="mxchat-chunk-badge"><?php echo esc_html($chunk_count); ?> <?php esc_html_e('chunks', 'mxchat'); ?></span>
3060 - <span class="mxchat-chunk-preview">
3061 - <?php
3062 - $parent_content = isset($first_prompt->display_content) ? $first_prompt->display_content : $first_prompt->article_content;
3063 - $content_preview = mb_substr($parent_content, 0, 100);
3064 - echo esc_html($content_preview . '...');
3065 - ?>
3066 - </span>
3067 - </div>
3068 - </td>
3069 - <td class="mxchat-url-cell" style="padding: 12px 16px; font-size: 13px;">
3070 - <?php if (!empty($source_url) && strpos($source_url, 'mxchat://') !== 0 && strpos($source_url, '_ungrouped_') !== 0) : ?>
3071 - <a href="<?php echo esc_url($source_url); ?>" target="_blank" style="color: var(--mxch-primary); text-decoration: none;">
3072 - <span class="dashicons dashicons-external" style="font-size: 14px;"></span>
3073 - <?php esc_html_e('View Source', 'mxchat'); ?>
3074 - </a>
3075 - <?php else : ?>
3076 - <span style="color: var(--mxch-text-muted);"><?php esc_html_e('Manual Content', 'mxchat'); ?></span>
3077 - <?php endif; ?>
3078 - </td>
3079 - <td class="mxchat-actions-cell" style="padding: 12px 16px; white-space: nowrap;">
3080 - <?php if ($data_source !== 'pinecone') : ?>
3081 - <button type="button"
3082 - class="mxch-btn mxch-btn-ghost mxch-btn-sm mxchat-edit-entry-btn"
3083 - data-source-url="<?php echo esc_attr($source_url); ?>"
3084 - data-entry-id="<?php echo esc_attr($first_prompt->id); ?>"
3085 - data-data-source="<?php echo esc_attr($data_source); ?>"
3086 - data-bot-id="<?php echo esc_attr($current_bot_id); ?>"
3087 - data-nonce="<?php echo wp_create_nonce('mxchat_edit_entry_nonce'); ?>"
3088 - title="<?php esc_attr_e('Edit content', 'mxchat'); ?>">
3089 - <span class="dashicons dashicons-edit" style="font-size: 14px;"></span>
3090 - </button>
3091 - <?php endif; ?>
3092 - <button type="button"
3093 - class="mxch-btn mxch-btn-ghost mxch-btn-sm delete-button-group"
3094 - data-source-url="<?php echo esc_attr($source_url); ?>"
3095 - data-chunk-count="<?php echo esc_attr($chunk_count); ?>"
3096 - data-data-source="<?php echo esc_attr($data_source); ?>"
3097 - data-bot-id="<?php echo esc_attr($current_bot_id); ?>"
3098 - data-nonce="<?php echo wp_create_nonce('mxchat_delete_chunks_nonce'); ?>"
3099 - style="color: var(--mxch-error);"
3100 - title="<?php esc_attr_e('Delete all chunks', 'mxchat'); ?>">
3101 - <span class="dashicons dashicons-trash" style="font-size: 14px;"></span>
3102 - </button>
3103 - </td>
3104 - </tr>
3105 - <?php
3106 - // Render hidden chunk rows
3107 - foreach ($group as $chunk_index => $chunk) {
3108 - $meta_chunk_index = isset($chunk->chunk_metadata['chunk_index']) ? intval($chunk->chunk_metadata['chunk_index']) : $chunk_index;
3109 - $meta_total_chunks = isset($chunk->chunk_metadata['total_chunks']) ? intval($chunk->chunk_metadata['total_chunks']) : $chunk_count;
3110 - $content = isset($chunk->display_content) ? $chunk->display_content : $chunk->article_content;
3111 - $content_preview = mb_strlen($content) > $preview_length
3112 - ? mb_substr($content, 0, $preview_length) . '...'
3113 - : $content;
3114 - ?>
3115 - <tr id="prompt-<?php echo esc_attr($chunk->id); ?>"
3116 - class="mxchat-chunk-row <?php echo esc_attr($group_id); ?>"
3117 - data-source="<?php echo esc_attr($data_source); ?>"
3118 - style="display: none; background: #f8f9fa; border-bottom: 1px solid var(--mxch-card-border);">
3119 - <td style="padding: 12px 16px; text-align: center;">
3120 - <!-- Checkbox column placeholder for chunks (managed by group) -->
3121 - </td>
3122 - <td style="padding: 12px 16px 12px 30px; font-size: 13px;">
3123 - <!-- Hidden ID column for chunks -->
3124 - </td>
3125 - <td class="mxchat-content-cell" style="padding: 12px 16px; font-size: 13px;">
3126 - <div class="mxchat-accordion-wrapper">
3127 - <div class="mxchat-content-preview">
3128 - <span class="mxchat-chunk-indicator" style="margin-right: 10px; color: var(--mxch-text-secondary); font-size: 12px;">
3129 - <?php printf(esc_html__('Chunk %d of %d', 'mxchat'), $meta_chunk_index + 1, $meta_total_chunks); ?>
3130 - </span>
3131 - <span class="preview-text"><?php echo esc_html($content_preview); ?></span>
3132 - <?php if (mb_strlen($content) > $preview_length) : ?>
3133 - <button class="mxchat-expand-toggle" type="button">
3134 - <span class="dashicons dashicons-arrow-down-alt2"></span>
3135 - </button>
3136 - <?php endif; ?>
3137 - </div>
3138 - <div class="mxchat-content-full" style="display: none;">
3139 - <div class="content-view">
3140 - <?php
3141 - if (preg_match('/[\x{0590}-\x{05FF}]/u', $content)) {
3142 - echo '<div dir="rtl" lang="he" class="rtl-content">';
3143 - echo wp_kses_post(wpautop($content));
3144 - echo '</div>';
3145 - } else {
3146 - echo wp_kses_post(wpautop($content));
3147 - }
3148 - ?>
3149 - </div>
3150 - </div>
3151 - </div>
3152 - </td>
3153 - <td class="mxchat-url-cell" style="padding: 12px 16px; font-size: 13px;">
3154 - <span class="mxchat-chunk-label" style="color: var(--mxch-text-muted);"><?php esc_html_e('Same as parent', 'mxchat'); ?></span>
3155 - </td>
3156 - <td class="mxchat-actions-cell" style="padding: 12px 16px;">
3157 - <span class="mxchat-chunk-label" style="color: var(--mxch-text-muted);"><?php esc_html_e('Managed by group', 'mxchat'); ?></span>
3158 - </td>
3159 - </tr>
3160 - <?php
3161 - }
3162 - } else {
3163 - // Single entry - display normally with accordion
3164 - $prompt = $first_prompt;
3165 - $content = isset($prompt->display_content) ? $prompt->display_content : $prompt->article_content;
3166 - $content_preview = mb_strlen($content) > $preview_length
3167 - ? mb_substr($content, 0, $preview_length) . '...'
3168 - : $content;
3169 - ?>
3170 - <tr id="prompt-<?php echo esc_attr($prompt->id); ?>"
3171 - data-source="<?php echo esc_attr($data_source); ?>"
3172 - style="border-bottom: 1px solid var(--mxch-card-border);">
3173 - <td style="padding: 12px 16px; text-align: center;">
3174 - <input type="checkbox"
3175 - class="mxchat-entry-checkbox"
3176 - data-entry-id="<?php echo esc_attr($prompt->id); ?>"
3177 - data-source="<?php echo esc_attr($data_source); ?>"
3178 - data-source-url="<?php echo esc_attr($source_url); ?>"
3179 - data-is-group="false">
3180 - </td>
3181 - <td style="padding: 12px 16px; font-size: 13px;">
3182 - <?php echo esc_html($prompt->id); ?>
3183 - </td>
3184 - <td class="mxchat-content-cell" style="padding: 12px 16px; font-size: 13px;">
3185 - <div class="mxchat-accordion-wrapper">
3186 - <div class="mxchat-content-preview">
3187 - <span class="preview-text"><?php echo esc_html($content_preview); ?></span>
3188 - <?php if (mb_strlen($content) > $preview_length) : ?>
3189 - <button class="mxchat-expand-toggle" type="button">
3190 - <span class="dashicons dashicons-arrow-down-alt2"></span>
3191 - </button>
3192 - <?php endif; ?>
3193 - </div>
3194 - <div class="mxchat-content-full" style="display: none;">
3195 - <div class="content-view">
3196 - <?php
3197 - if (preg_match('/[\x{0590}-\x{05FF}]/u', $content)) {
3198 - echo '<div dir="rtl" lang="he" class="rtl-content">';
3199 - echo wp_kses_post(wpautop($content));
3200 - echo '</div>';
3201 - } else {
3202 - echo wp_kses_post(wpautop($content));
3203 - }
3204 - ?>
3205 - </div>
3206 - </div>
3207 - </div>
3208 - </td>
3209 - <td class="mxchat-url-cell" style="padding: 12px 16px; font-size: 13px;">
3210 - <?php
3211 - $actual_source = $source_url;
3212 - if (strpos($source_url, '_ungrouped_') === 0) {
3213 - $actual_source = $prompt->source_url ?? '';
3214 - }
3215 - if (!empty($actual_source) && strpos($actual_source, 'mxchat://') !== 0) : ?>
3216 - <a href="<?php echo esc_url($actual_source); ?>" target="_blank" style="color: var(--mxch-primary); text-decoration: none;">
3217 - <span class="dashicons dashicons-external" style="font-size: 14px;"></span>
3218 - <?php esc_html_e('View', 'mxchat'); ?>
3219 - </a>
3220 - <?php else : ?>
3221 - <span style="color: var(--mxch-text-muted);"><?php esc_html_e('Manual', 'mxchat'); ?></span>
3222 - <?php endif; ?>
3223 - </td>
3224 - <td style="padding: 12px 16px; white-space: nowrap;">
3225 - <button type="button"
3226 - class="mxch-btn mxch-btn-ghost mxch-btn-sm mxchat-edit-entry-btn"
3227 - data-source-url="<?php echo esc_attr($prompt->source_url ?? ''); ?>"
3228 - data-entry-id="<?php echo esc_attr($prompt->id); ?>"
3229 - data-data-source="<?php echo esc_attr($data_source); ?>"
3230 - data-bot-id="<?php echo esc_attr($current_bot_id); ?>"
3231 - data-nonce="<?php echo wp_create_nonce('mxchat_edit_entry_nonce'); ?>"
3232 - title="<?php esc_attr_e('Edit content', 'mxchat'); ?>">
3233 - <span class="dashicons dashicons-edit" style="font-size: 14px;"></span>
3234 - </button>
3235 - <button type="button" class="mxch-btn mxch-btn-ghost mxch-btn-sm delete-button-wordpress" data-entry-id="<?php echo esc_attr($prompt->id); ?>" data-bot-id="<?php echo esc_attr($current_bot_id); ?>" data-nonce="<?php echo wp_create_nonce('mxchat_delete_wordpress_prompt_nonce'); ?>" style="color: var(--mxch-error);">
3236 - <span class="dashicons dashicons-trash" style="font-size: 14px;"></span>
3237 - </button>
3238 - </td>
3239 - </tr>
3240 - <?php
3241 - }
3242 - }
3243 - }
3244 - $html = ob_get_clean();
3245 -
3246 - // Generate pagination HTML (include search/filter data for subsequent pages)
3247 - $pagination_html = '';
3248 - if ($total_pages > 1) {
3249 - $pagination_html = '<div class="mxchat-ajax-pagination" data-current-page="' . esc_attr($page) . '" data-total-pages="' . esc_attr($total_pages) . '" data-search="' . esc_attr($search_query) . '" data-content-type="' . esc_attr($content_type_filter) . '">';
3250 -
3251 - // Previous button
3252 - if ($page > 1) {
3253 - $pagination_html .= '<a href="#" class="mxchat-page-link" data-page="' . ($page - 1) . '">' . esc_html__('&laquo; Previous', 'mxchat') . '</a> ';
3254 - }
3255 -
3256 - // Page numbers
3257 - $start_page = max(1, $page - 2);
3258 - $end_page = min($total_pages, $page + 2);
3259 -
3260 - if ($start_page > 1) {
3261 - $pagination_html .= '<a href="#" class="mxchat-page-link" data-page="1">1</a> ';
3262 - if ($start_page > 2) {
3263 - $pagination_html .= '<span class="mxchat-page-dots">...</span> ';
3264 - }
3265 - }
3266 -
3267 - for ($i = $start_page; $i <= $end_page; $i++) {
3268 - if ($i == $page) {
3269 - $pagination_html .= '<span class="mxchat-page-current">' . $i . '</span> ';
3270 - } else {
3271 - $pagination_html .= '<a href="#" class="mxchat-page-link" data-page="' . $i . '">' . $i . '</a> ';
3272 - }
3273 - }
3274 -
3275 - if ($end_page < $total_pages) {
3276 - if ($end_page < $total_pages - 1) {
3277 - $pagination_html .= '<span class="mxchat-page-dots">...</span> ';
3278 - }
3279 - $pagination_html .= '<a href="#" class="mxchat-page-link" data-page="' . $total_pages . '">' . $total_pages . '</a> ';
3280 - }
3281 -
3282 - // Next button
3283 - if ($page < $total_pages) {
3284 - $pagination_html .= '<a href="#" class="mxchat-page-link" data-page="' . ($page + 1) . '">' . esc_html__('Next &raquo;', 'mxchat') . '</a>';
3285 - }
3286 -
3287 - $pagination_html .= '</div>';
3288 - }
3289 -
3290 - wp_send_json_success(array(
3291 - 'html' => $html,
3292 - 'pagination_html' => $pagination_html,
3293 - 'total_count' => $total_records,
3294 - 'total_pages' => $total_pages,
3295 - 'page' => $page,
3296 - 'per_page' => $per_page,
3297 - 'data_source' => 'wordpress'
3298 - ));
3299 -}
3300 -
3301 -/**
3302 - * AJAX handler to detect available sitemaps on the site
3303 - * Optimized for speed - only checks primary sitemap indexes first
3304 - */
3305 -public function ajax_mxchat_detect_sitemaps() {
3306 - check_ajax_referer('mxchat_detect_sitemaps_nonce', 'nonce');
3307 -
3308 - if (!current_user_can('manage_options')) {
3309 - wp_send_json_error(array('message' => 'Unauthorized'));
3310 - return;
3311 - }
3312 -
3313 - $site_url = get_site_url();
3314 - $sitemaps = array();
3315 - $found_index = false;
3316 -
3317 - // Only check the main sitemap index files first (much faster)
3318 - // These are the primary entry points that contain sub-sitemaps
3319 - $primary_indexes = array(
3320 - 'sitemap_index.xml' => 'Yoast SEO / Rank Math', // Yoast & Rank Math
3321 - 'wp-sitemap.xml' => 'WordPress Core', // WordPress Core
3322 - 'sitemap.xml' => 'Standard', // Generic/AIOSEO
3323 - );
3324 -
3325 - foreach ($primary_indexes as $path => $source) {
3326 - $url = trailingslashit($site_url) . $path;
3327 -
3328 - $response = wp_remote_head($url, array(
3329 - 'timeout' => 10,
3330 - 'sslverify' => false,
3331 - 'redirection' => 1,
3332 - 'user-agent' => mxchat_ingest_user_agent(),
3333 - ));
3334 -
3335 - if (!is_wp_error($response) && wp_remote_retrieve_response_code($response) === 200) {
3336 - // Found a sitemap index - parse it to get sub-sitemaps
3337 - $sub_sitemaps = $this->parse_sitemap_index($url);
3338 - if (!empty($sub_sitemaps)) {
3339 - $sitemaps[] = array(
3340 - 'url' => $url,
3341 - 'type' => 'index',
3342 - 'source' => $source,
3343 - 'sub_sitemaps' => $sub_sitemaps
3344 - );
3345 - $found_index = true;
3346 - // Found a valid index, no need to check others
3347 - break;
3348 - }
3349 - }
3350 - }
3351 -
3352 - // If no sitemap index found, check for standalone sitemaps
3353 - if (!$found_index) {
3354 - $standalone_sitemaps = array(
3355 - 'post-sitemap.xml' => array('type' => 'content', 'source' => 'Yoast SEO'),
3356 - 'page-sitemap.xml' => array('type' => 'content', 'source' => 'Yoast SEO'),
3357 - );
3358 -
3359 - foreach ($standalone_sitemaps as $path => $info) {
3360 - $url = trailingslashit($site_url) . $path;
3361 -
3362 - $response = wp_remote_head($url, array(
3363 - 'timeout' => 2,
3364 - 'sslverify' => false
3365 - ));
3366 -
3367 - if (!is_wp_error($response) && wp_remote_retrieve_response_code($response) === 200) {
3368 - $sitemaps[] = array(
3369 - 'url' => $url,
3370 - 'type' => $info['type'],
3371 - 'source' => $info['source'],
3372 - 'url_count' => 0 // Skip URL count for speed
3373 - );
3374 - }
3375 - }
3376 - }
3377 -
3378 - wp_send_json_success(array(
3379 - 'sitemaps' => $sitemaps,
3380 - 'site_url' => $site_url
3381 - ));
3382 -}
3383 -
3384 -/**
3385 - * Parse a sitemap index to get sub-sitemaps
3386 - * Optimized: doesn't fetch URL count for each sub-sitemap (too slow)
3387 - */
3388 -private function parse_sitemap_index($url) {
3389 - $sub_sitemaps = array();
3390 -
3391 - $response = wp_remote_get($url, array(
3392 - 'timeout' => 30,
3393 - 'sslverify' => false,
3394 - 'user-agent' => mxchat_ingest_user_agent(),
3395 - 'headers' => array(
3396 - 'Accept' => 'text/html,application/xhtml+xml,application/xml;q=0.9,*/*;q=0.8',
3397 - ),
3398 - ));
3399 -
3400 - if (is_wp_error($response)) {
3401 - return $sub_sitemaps;
3402 - }
3403 -
3404 - $body = wp_remote_retrieve_body($response);
3405 - if (empty($body)) {
3406 - return $sub_sitemaps;
3407 - }
3408 -
3409 - // Suppress XML errors
3410 - libxml_use_internal_errors(true);
3411 - $xml = simplexml_load_string($body);
3412 - libxml_clear_errors();
3413 -
3414 - if ($xml === false) {
3415 - return $sub_sitemaps;
3416 - }
3417 -
3418 - // Check if it's a sitemap index (contains <sitemap> elements)
3419 - if (isset($xml->sitemap)) {
3420 - foreach ($xml->sitemap as $sitemap) {
3421 - $loc = (string) $sitemap->loc;
3422 - if (!empty($loc)) {
3423 - // Try to determine the type from the URL
3424 - $type = 'content';
3425 - if (strpos($loc, 'category') !== false || strpos($loc, 'tag') !== false || strpos($loc, 'taxonomy') !== false) {
3426 - $type = 'taxonomy';
3427 - } elseif (strpos($loc, 'author') !== false || strpos($loc, 'user') !== false) {
3428 - $type = 'author';
3429 - }
3430 -
3431 - // Skip URL count - too slow to fetch for each sitemap
3432 - $sub_sitemaps[] = array(
3433 - 'url' => $loc,
3434 - 'type' => $type,
3435 - 'url_count' => 0, // Don't fetch - takes too long
3436 - 'name' => basename(parse_url($loc, PHP_URL_PATH))
3437 - );
3438 - }
3439 - }
3440 - }
3441 -
3442 - return $sub_sitemaps;
3443 -}
3444 -
3445 -/**
3446 - * Get URL count from a sitemap
3447 - */
3448 -private function get_sitemap_url_count($url) {
3449 - $response = wp_remote_get($url, array(
3450 - 'timeout' => 30,
3451 - 'sslverify' => false,
3452 - 'user-agent' => mxchat_ingest_user_agent(),
3453 - 'headers' => array(
3454 - 'Accept' => 'text/html,application/xhtml+xml,application/xml;q=0.9,*/*;q=0.8',
3455 - ),
3456 - ));
3457 -
3458 - if (is_wp_error($response)) {
3459 - return 0;
3460 - }
3461 -
3462 - $body = wp_remote_retrieve_body($response);
3463 - if (empty($body)) {
3464 - return 0;
3465 - }
3466 -
3467 - // Count <url> or <loc> elements
3468 - $count = preg_match_all('/<url>/i', $body, $matches);
3469 - return $count ?: 0;
3470 -}
3471 -
3472 -/**
3473 - * Get sitemaps declared in robots.txt
3474 - */
3475 -private function get_sitemaps_from_robots($site_url) {
3476 - $sitemaps = array();
3477 - $robots_url = trailingslashit($site_url) . 'robots.txt';
3478 -
3479 - $response = wp_remote_get($robots_url, array(
3480 - 'timeout' => 15,
3481 - 'sslverify' => false,
3482 - 'user-agent' => mxchat_ingest_user_agent(),
3483 - ));
3484 -
3485 - if (is_wp_error($response)) {
3486 - return $sitemaps;
3487 - }
3488 -
3489 - $body = wp_remote_retrieve_body($response);
3490 - if (empty($body)) {
3491 - return $sitemaps;
3492 - }
3493 -
3494 - // Find Sitemap: declarations
3495 - if (preg_match_all('/^Sitemap:\s*(.+)$/mi', $body, $matches)) {
3496 - foreach ($matches[1] as $sitemap_url) {
3497 - $sitemap_url = trim($sitemap_url);
3498 - if (filter_var($sitemap_url, FILTER_VALIDATE_URL)) {
3499 - $sitemaps[] = $sitemap_url;
3500 - }
3501 - }
3502 - }
3503 -
3504 - return $sitemaps;
3505 -}
3506 -
3507 -public function mxchat_stop_processing() {
3508 - // Verify permissions
3509 - if (!current_user_can('manage_options')) {
3510 - wp_die(esc_html__('Unauthorized access', 'mxchat'));
3511 - }
3512 -
3513 - // Verify nonce
3514 - check_admin_referer('mxchat_stop_processing_action', 'mxchat_stop_processing_nonce');
3515 -
3516 - global $wpdb;
3517 - $table_name = $wpdb->prefix . 'mxchat_processing_queue';
3518 -
3519 - // Get active queue IDs
3520 - $sitemap_queue_id = get_transient('mxchat_active_queue_sitemap');
3521 - $pdf_queue_id = get_transient('mxchat_active_queue_pdf');
3522 -
3523 - // Delete all pending items from active queues
3524 - if ($sitemap_queue_id) {
3525 - $wpdb->delete(
3526 - $table_name,
3527 - array(
3528 - 'queue_id' => $sitemap_queue_id,
3529 - 'status' => 'pending'
3530 - ),
3531 - array('%s', '%s')
3532 - );
3533 -
3534 - delete_transient('mxchat_active_queue_sitemap');
3535 - delete_transient('mxchat_last_sitemap_url');
3536 - }
3537 -
3538 - if ($pdf_queue_id) {
3539 - // Get PDF path before deleting
3540 - $pdf_path = $this->mxchat_get_queue_meta($pdf_queue_id, 'pdf_path');
3541 -
3542 - $wpdb->delete(
3543 - $table_name,
3544 - array(
3545 - 'queue_id' => $pdf_queue_id,
3546 - 'status' => 'pending'
3547 - ),
3548 - array('%s', '%s')
3549 - );
3550 -
3551 - // Delete PDF file
3552 - if ($pdf_path && file_exists($pdf_path)) {
3553 - wp_delete_file($pdf_path);
3554 - }
3555 -
3556 - delete_transient('mxchat_active_queue_pdf');
3557 - delete_transient('mxchat_last_pdf_url');
3558 - }
3559 -
3560 - // Redirect back with a success message
3561 - set_transient('mxchat_admin_notice_success',
3562 - esc_html__('Processing has been stopped successfully.', 'mxchat'),
3563 - 30
3564 - );
3565 - wp_safe_redirect(admin_url('admin.php?page=mxchat-prompts'));
3566 - exit;
3567 -}
3568 -
3569 -/**
3570 - * Get content list for processing
3571 - */
3572 -public function ajax_mxchat_get_content_list() {
3573 - // Verify the nonce
3574 - check_ajax_referer('mxchat_content_selector_nonce', 'nonce');
3575 -
3576 - if (!current_user_can('manage_options')) {
3577 - wp_send_json_error(__('Unauthorized access', 'mxchat'));
3578 - }
3579 -
3580 - $page = isset($_GET['page']) ? absint($_GET['page']) : 1;
3581 - $per_page = isset($_GET['per_page']) ? absint($_GET['per_page']) : 100;
3582 - $search = isset($_GET['search']) ? sanitize_text_field($_GET['search']) : '';
3583 - $post_type = isset($_GET['post_type']) ? sanitize_text_field($_GET['post_type']) : 'all';
3584 - $post_status = isset($_GET['post_status']) ? sanitize_text_field($_GET['post_status']) : 'publish';
3585 - $processed_filter = isset($_GET['processed_filter']) ? sanitize_text_field($_GET['processed_filter']) : 'all';
3586 -
3587 - // Build query args
3588 - $args = array(
3589 - 'posts_per_page' => $per_page,
3590 - 'paged' => $page,
3591 - 'post_status' => $post_status !== 'all' ? $post_status : array('publish', 'draft', 'pending'),
3592 - 'orderby' => 'date',
3593 - 'order' => 'DESC',
3594 - );
3595 -
3596 - // Handle post types - IMPROVED VERSION
3597 - if ($post_type !== 'all') {
3598 - $args['post_type'] = $post_type;
3599 - } else {
3600 - // Get all available post types that might contain content
3601 - $all_post_types = array();
3602 -
3603 - // First get all public post types
3604 - $public_types = get_post_types(array('public' => true), 'names');
3605 - $all_post_types = array_merge($all_post_types, $public_types);
3606 -
3607 - // Add common forum/community post types
3608 - $forum_types = array('topic', 'reply', 'forum', 'wpforo_topic', 'wpforo_post');
3609 - foreach ($forum_types as $forum_type) {
3610 - if (post_type_exists($forum_type)) {
3611 - $all_post_types[] = $forum_type;
3612 - }
3613 - }
3614 -
3615 - // Add other commonly used post types
3616 - $common_types = array('product', 'job_listing', 'event', 'portfolio');
3617 - foreach ($common_types as $common_type) {
3618 - if (post_type_exists($common_type)) {
3619 - $all_post_types[] = $common_type;
3620 - }
3621 - }
3622 -
3623 - // Remove duplicates and ensure we have at least some post types
3624 - $all_post_types = array_unique($all_post_types);
3625 -
3626 - if (empty($all_post_types)) {
3627 - // Fallback to basic post types
3628 - $all_post_types = array('post', 'page');
3629 - }
3630 -
3631 - $args['post_type'] = $all_post_types;
3632 -
3633 - // Debug logging to see what post types are being queried
3634 - //error_log('MxChat Debug: Querying post types: ' . implode(', ', $all_post_types));
3635 - }
3636 -
3637 - if (!empty($search)) {
3638 - $args['s'] = $search;
3639 - }
3640 -
3641 - // Get processed data from storage
3642 - $processed_data = array();
3643 -
3644 - $pinecone_options = get_option('mxchat_pinecone_addon_options', array());
3645 - $use_pinecone = ($pinecone_options['mxchat_use_pinecone'] ?? '0') === '1';
3646 -
3647 - if ($use_pinecone && !empty($pinecone_options['mxchat_pinecone_api_key'])) {
3648 - // Get fresh data from Pinecone - no caching
3649 - $processed_data = $this->mxchat_get_pinecone_processed_content($pinecone_options);
3650 - } else {
3651 - // WordPress DB checking with better URL matching for all post types
3652 - global $wpdb;
3653 - $table_name = $wpdb->prefix . 'mxchat_system_prompt_content';
3654 - $processed_items = $wpdb->get_results("SELECT id, source_url, article_content, timestamp FROM {$table_name}");
3655 -
3656 - // Group items by source_url to count chunks
3657 - $url_chunk_counts = array();
3658 - $url_latest_timestamp = array();
3659 - $url_first_id = array();
3660 -
3661 - if (!empty($processed_items)) {
3662 - foreach ($processed_items as $item) {
3663 - $url = $item->source_url;
3664 - if (empty($url)) continue;
3665 -
3666 - // Count chunks per URL
3667 - if (!isset($url_chunk_counts[$url])) {
3668 - $url_chunk_counts[$url] = 0;
3669 - $url_latest_timestamp[$url] = $item->timestamp;
3670 - $url_first_id[$url] = $item->id;
3671 - }
3672 - $url_chunk_counts[$url]++;
3673 -
3674 - // Track latest timestamp
3675 - if (strtotime($item->timestamp) > strtotime($url_latest_timestamp[$url])) {
3676 - $url_latest_timestamp[$url] = $item->timestamp;
3677 - }
3678 - }
3679 -
3680 - // Now build processed_data with chunk counts
3681 - foreach ($url_chunk_counts as $url => $chunk_count) {
3682 - $post_id = $this->mxchat_url_to_post_id_improved($url);
3683 -
3684 - if ($post_id) {
3685 - $processed_data[$post_id] = array(
3686 - 'db_id' => $url_first_id[$url],
3687 - 'timestamp' => $url_latest_timestamp[$url],
3688 - 'url' => $url,
3689 - 'source' => 'wordpress',
3690 - 'chunk_count' => $chunk_count
3691 - );
3692 - }
3693 - }
3694 - }
3695 - }
3696 -
3697 - // Get processed IDs as a simple array for in_array checks
3698 - $processed_ids = array_keys($processed_data);
3699 -
3700 - // Handle processed/unprocessed filter
3701 - if ($processed_filter === 'processed' && !empty($processed_ids)) {
3702 - $args['post__in'] = $processed_ids;
3703 - } elseif ($processed_filter === 'unprocessed' && !empty($processed_ids)) {
3704 - $args['post__not_in'] = $processed_ids;
3705 - }
3706 -
3707 - // Run the query
3708 - $query = new WP_Query($args);
3709 - $content_items = array();
3710 -
3711 - if ($query->have_posts()) {
3712 - while ($query->have_posts()) {
3713 - $query->the_post();
3714 - $id = get_the_ID();
3715 - $post_date = get_the_date();
3716 - $excerpt = wp_trim_words(get_the_excerpt(), 20, '...');
3717 - $word_count = str_word_count(strip_tags(get_the_content()));
3718 -
3719 - $is_processed = in_array($id, $processed_ids);
3720 - $processed_date = '';
3721 - $db_record_id = 0;
3722 - $data_source = 'none';
3723 -
3724 - if ($is_processed && isset($processed_data[$id])) {
3725 - $item_data = $processed_data[$id];
3726 - $data_source = $item_data['source'];
3727 -
3728 - if ($data_source === 'wordpress' && isset($item_data['timestamp'])) {
3729 - // WordPress DB format
3730 - $timestamp = strtotime($item_data['timestamp']);
3731 - $processed_date = human_time_diff($timestamp, current_time('timestamp')) . ' ago';
3732 - $db_record_id = $item_data['db_id'];
3733 - } elseif ($data_source === 'pinecone') {
3734 - // Pinecone format
3735 - $processed_date = $item_data['processed_date'];
3736 - $db_record_id = $item_data['db_id'];
3737 - }
3738 - }
3739 -
3740 - // Get chunk count for this item
3741 - $chunk_count = 0;
3742 - if ($is_processed && isset($processed_data[$id]['chunk_count'])) {
3743 - $chunk_count = intval($processed_data[$id]['chunk_count']);
3744 - }
3745 -
3746 - $content_items[] = array(
3747 - 'id' => $id,
3748 - 'title' => get_the_title(),
3749 - 'permalink' => get_permalink(),
3750 - 'date' => $post_date,
3751 - 'type' => get_post_type(),
3752 - 'status' => get_post_status(),
3753 - 'excerpt' => $excerpt,
3754 - 'word_count' => $word_count,
3755 - 'already_processed' => $is_processed,
3756 - 'processed_date' => $processed_date,
3757 - 'db_record_id' => $db_record_id,
3758 - 'data_source' => $data_source,
3759 - 'chunk_count' => $chunk_count
3760 - );
3761 - }
3762 - wp_reset_postdata();
3763 - }
3764 -
3765 - $response = array(
3766 - 'items' => $content_items,
3767 - 'total' => $query->found_posts,
3768 - 'total_pages' => $query->max_num_pages,
3769 - 'current_page' => $page,
3770 - 'processed_count' => count($processed_ids)
3771 - );
3772 -
3773 - wp_send_json_success($response);
3774 - exit;
3775 -}
3776 -
3777 -
3778 -/**
3779 - * This function handles various WooCommerce URL formats and permalink structures
3780 - */
3781 -private function mxchat_url_to_post_id_improved($url) {
3782 - // First try the standard WordPress function
3783 - $post_id = url_to_postid($url);
3784 -
3785 - if ($post_id > 0) {
3786 - return $post_id;
3787 - }
3788 -
3789 - // If that fails, try more aggressive URL matching
3790 - // Remove trailing slashes and query parameters for better matching
3791 - $clean_url = rtrim($url, '/');
3792 - $clean_url = strtok($clean_url, '?'); // Remove query parameters
3793 -
3794 - // Try again with cleaned URL
3795 - $post_id = url_to_postid($clean_url);
3796 - if ($post_id > 0) {
3797 - return $post_id;
3798 - }
3799 -
3800 - // For bbPress forum topics, try extracting slug from URL
3801 - if (strpos($url, '/topic/') !== false || strpos($url, '/forums/') !== false) {
3802 - // Handle bbPress URLs: /forums/topic/topic-name/
3803 - if (preg_match('/\/forums\/topic\/([^\/\?]+)/', $url, $matches)) {
3804 - $topic_slug = $matches[1];
3805 -
3806 - // Look up topic by slug
3807 - $topic = get_page_by_path($topic_slug, OBJECT, 'topic');
3808 - if ($topic) {
3809 - return $topic->ID;
3810 - }
3811 -
3812 - // Alternative method: query by post_name
3813 - global $wpdb;
3814 - $post_id = $wpdb->get_var($wpdb->prepare(
3815 - "SELECT ID FROM {$wpdb->posts} WHERE post_name = %s AND post_type = 'topic' AND post_status IN ('publish', 'closed')",
3816 - $topic_slug
3817 - ));
3818 -
3819 - if ($post_id) {
3820 - return intval($post_id);
3821 - }
3822 - }
3823 -
3824 - // Handle simpler topic URLs: /topic/topic-name/
3825 - if (preg_match('/\/topic\/([^\/\?]+)/', $url, $matches)) {
3826 - $topic_slug = $matches[1];
3827 -
3828 - global $wpdb;
3829 - $post_id = $wpdb->get_var($wpdb->prepare(
3830 - "SELECT ID FROM {$wpdb->posts} WHERE post_name = %s AND post_type = 'topic' AND post_status IN ('publish', 'closed')",
3831 - $topic_slug
3832 - ));
3833 -
3834 - if ($post_id) {
3835 - return intval($post_id);
3836 - }
3837 - }
3838 - }
3839 -
3840 - // For WooCommerce products
3841 - if (strpos($url, '/product/') !== false || strpos($url, 'product=') !== false) {
3842 - // Extract product slug from various URL formats
3843 - $product_slug = '';
3844 -
3845 - // Handle pretty permalinks: /product/product-name/
3846 - if (preg_match('/\/product\/([^\/\?]+)/', $url, $matches)) {
3847 - $product_slug = $matches[1];
3848 - }
3849 - // Handle query parameters: ?product=product-name
3850 - elseif (preg_match('/[\?&]product=([^&]+)/', $url, $matches)) {
3851 - $product_slug = $matches[1];
3852 - }
3853 -
3854 - if (!empty($product_slug)) {
3855 - // Look up product by slug
3856 - $product = get_page_by_path($product_slug, OBJECT, 'product');
3857 - if ($product) {
3858 - return $product->ID;
3859 - }
3860 -
3861 - // Alternative method: query by post_name
3862 - global $wpdb;
3863 - $post_id = $wpdb->get_var($wpdb->prepare(
3864 - "SELECT ID FROM {$wpdb->posts} WHERE post_name = %s AND post_type = 'product' AND post_status = 'publish'",
3865 - $product_slug
3866 - ));
3867 -
3868 - if ($post_id) {
3869 - return intval($post_id);
3870 - }
3871 - }
3872 - }
3873 -
3874 - // Generic approach: try to extract slug and match against all post types
3875 - $parsed_url = wp_parse_url($clean_url);
3876 - $path = $parsed_url['path'] ?? '';
3877 -
3878 - if (!empty($path)) {
3879 - // Get the last part of the path as potential slug
3880 - $path_parts = array_filter(explode('/', trim($path, '/')));
3881 - $potential_slug = end($path_parts);
3882 -
3883 - if (!empty($potential_slug)) {
3884 - global $wpdb;
3885 -
3886 - // Try to find any post with this slug
3887 - $post_id = $wpdb->get_var($wpdb->prepare(
3888 - "SELECT ID FROM {$wpdb->posts}
3889 - WHERE post_name = %s
3890 - AND post_status IN ('publish', 'closed', 'private')
3891 - AND post_type NOT IN ('revision', 'attachment', 'nav_menu_item')
3892 - ORDER BY CASE
3893 - WHEN post_type = 'post' THEN 1
3894 - WHEN post_type = 'page' THEN 2
3895 - WHEN post_type = 'topic' THEN 3
3896 - WHEN post_type = 'product' THEN 4
3897 - ELSE 5
3898 - END
3899 - LIMIT 1",
3900 - $potential_slug
3901 - ));
3902 -
3903 - if ($post_id) {
3904 - return intval($post_id);
3905 - }
3906 - }
3907 - }
3908 -
3909 - // ADDITIONAL: Try direct database lookup by URL variations
3910 - global $wpdb;
3911 - $table_name = $wpdb->prefix . 'mxchat_system_prompt_content';
3912 -
3913 - // Try variations of the URL (with/without trailing slash, http/https)
3914 - $url_variations = array(
3915 - $url,
3916 - rtrim($url, '/'),
3917 - $url . '/',
3918 - str_replace('http://', 'https://', $url),
3919 - str_replace('https://', 'http://', $url),
3920 - str_replace('http://', 'https://', rtrim($url, '/')),
3921 - str_replace('https://', 'http://', rtrim($url, '/'))
3922 - );
3923 -
3924 - // Remove duplicates
3925 - $url_variations = array_unique($url_variations);
3926 -
3927 - foreach ($url_variations as $variation) {
3928 - $existing_record = $wpdb->get_row($wpdb->prepare(
3929 - "SELECT id, source_url FROM $table_name WHERE source_url = %s",
3930 - $variation
3931 - ));
3932 -
3933 - if ($existing_record) {
3934 - // Try to get post ID from this stored URL
3935 - $stored_post_id = url_to_postid($existing_record->source_url);
3936 - if ($stored_post_id > 0) {
3937 - return $stored_post_id;
3938 - }
3939 - }
3940 - }
3941 -
3942 - return 0; // No match found
3943 -}
3944 -/**
3945 - * Process selected content via AJAX
3946 - */
3947 -public function ajax_mxchat_process_selected_content() {
3948 - // Basic request validation
3949 - if (!check_ajax_referer('mxchat_content_selector_nonce', 'nonce', false)) {
3950 - wp_send_json_error('Invalid nonce');
3951 - exit;
3952 - }
3953 -
3954 - if (!current_user_can('manage_options')) {
3955 - wp_send_json_error('Unauthorized access');
3956 - exit;
3957 - }
3958 -
3959 - // Get post IDs - safely parse the array
3960 - $post_ids = array();
3961 - if (isset($_POST['post_ids']) && is_array($_POST['post_ids'])) {
3962 - foreach ($_POST['post_ids'] as $id) {
3963 - $post_ids[] = absint($id);
3964 - }
3965 - }
3966 -
3967 - if (empty($post_ids)) {
3968 - wp_send_json_error('No content selected');
3969 - exit;
3970 - }
3971 -
3972 - // Get bot_id from request
3973 - $bot_id = isset($_POST['bot_id']) ? sanitize_key($_POST['bot_id']) : 'default';
3974 -
3975 - // ACF→PDF extraction is opt-in per import batch. Persist the last-used value so users
3976 - // don't re-check on every batch; the default is OFF for installs that haven't set it.
3977 - $extract_acf_pdfs = !empty($_POST['extract_acf_pdfs']) && $_POST['extract_acf_pdfs'] !== 'false';
3978 - $mxchat_options = get_option('mxchat_options', array());
3979 - if (!is_array($mxchat_options)) {
3980 - $mxchat_options = array();
3981 - }
3982 - $prior_default = !empty($mxchat_options['acf_pdf_extract_default']);
3983 - if ($prior_default !== $extract_acf_pdfs) {
3984 - $mxchat_options['acf_pdf_extract_default'] = $extract_acf_pdfs ? 1 : 0;
3985 - update_option('mxchat_options', $mxchat_options);
3986 - }
3987 -
3988 - // Process only ONE post at a time to avoid request size issues
3989 - $post_id = reset($post_ids);
3990 - $post = get_post($post_id);
3991 -
3992 - if (!$post) {
3993 - wp_send_json_error('Post not found');
3994 - exit;
3995 - }
3996 -
3997 - // Allow developers to modify post data before processing into knowledge base
3998 - $post = apply_filters('mxchat_before_process_post', $post, $bot_id);
3999 -
4000 - // Get content including title, short description (for WooCommerce), and main content
4001 - $content = $post->post_title . "\n\n";
4002 -
4003 - // Add short description if it exists (WooCommerce products use post_excerpt for short description)
4004 - if (!empty($post->post_excerpt)) {
4005 - // Remove shortcode tags but preserve content inside them
4006 - $clean_excerpt = $this->strip_shortcode_tags_preserve_content($post->post_excerpt);
4007 - $content .= "Short Description: " . wp_strip_all_tags($clean_excerpt) . "\n\n";
4008 - }
4009 -
4010 - // Add main content - remove shortcode tags but preserve content inside them
4011 - $clean_content = $this->strip_shortcode_tags_preserve_content($post->post_content);
4012 - $content .= wp_strip_all_tags($clean_content);
4013 -
4014 - // ADD WOOCOMMERCE PRODUCT DATA (pricing, stock, categories, custom tabs)
4015 - if (get_post_type($post_id) === 'product' && class_exists('WooCommerce')) {
4016 - $product = wc_get_product($post_id);
4017 -
4018 - if ($product) {
4019 - // Get pricing information
4020 - $regular_price = $product->get_regular_price();
4021 - $sale_price = $product->get_sale_price();
4022 - $price = $product->get_price();
4023 - $sku = $product->get_sku();
4024 -
4025 - // Get currency symbol
4026 - $currency_symbol = get_woocommerce_currency_symbol();
4027 -
4028 - // Add pricing information
4029 - $content .= "\n";
4030 - if (!empty($regular_price)) {
4031 - $content .= "Price: " . $currency_symbol . $regular_price . "\n";
4032 - } elseif (!empty($price)) {
4033 - $content .= "Price: " . $currency_symbol . $price . "\n";
4034 - }
4035 -
4036 - if (!empty($sale_price) && $sale_price !== $regular_price) {
4037 - $content .= "Sale Price: " . $currency_symbol . $sale_price . "\n";
4038 - }
4039 -
4040 - // Handle variable products - show price range
4041 - if ($product->is_type('variable')) {
4042 - $min_price = $product->get_variation_price('min');
4043 - $max_price = $product->get_variation_price('max');
4044 - if ($min_price !== $max_price) {
4045 - $content .= "Price Range: " . $currency_symbol . $min_price . " - " . $currency_symbol . $max_price . "\n";
4046 - }
4047 - }
4048 -
4049 - if (!empty($sku)) {
4050 - $content .= "SKU: " . $sku . "\n";
4051 - }
4052 -
4053 - // Get product categories
4054 - $categories = wp_get_post_terms($post_id, 'product_cat', array('fields' => 'names'));
4055 - if (!empty($categories) && !is_wp_error($categories)) {
4056 - $content .= "Categories: " . implode(', ', $categories) . "\n";
4057 - }
4058 - }
4059 -
4060 - // Get Custom Product Tabs (supports "Custom Product Tabs for WooCommerce" by Code Parrots)
4061 - $custom_tabs = get_post_meta($post_id, 'yikes_woo_products_tabs', true);
4062 - if (!empty($custom_tabs) && is_array($custom_tabs)) {
4063 - foreach ($custom_tabs as $tab) {
4064 - $tab_title = isset($tab['title']) ? $tab['title'] : (isset($tab['tab_title']) ? $tab['tab_title'] : '');
4065 - $tab_content = isset($tab['content']) ? $tab['content'] : '';
4066 -
4067 - if (!empty($tab_title) && !empty($tab_content)) {
4068 - $content .= "\n" . $tab_title . ": " . wp_strip_all_tags($tab_content) . "\n";
4069 - }
4070 - }
4071 - }
4072 -
4073 - // Also check for reusable/saved tabs applied to this product
4074 - $applied_saved_tabs = get_post_meta($post_id, 'yikes_woo_reusable_products_tabs_applied', true);
4075 - if (!empty($applied_saved_tabs) && is_array($applied_saved_tabs)) {
4076 - $saved_tabs = get_option('yikes_woo_reusable_products_tabs', array());
4077 - if (!empty($saved_tabs) && is_array($saved_tabs)) {
4078 - foreach ($applied_saved_tabs as $saved_tab_id) {
4079 - if (isset($saved_tabs[$saved_tab_id])) {
4080 - $tab = $saved_tabs[$saved_tab_id];
4081 - $tab_title = isset($tab['title']) ? $tab['title'] : (isset($tab['tab_title']) ? $tab['tab_title'] : '');
4082 - $tab_content = isset($tab['content']) ? $tab['content'] : '';
4083 -
4084 - if (!empty($tab_title) && !empty($tab_content)) {
4085 - $content .= "\n" . $tab_title . ": " . wp_strip_all_tags($tab_content) . "\n";
4086 - }
4087 - }
4088 - }
4089 - }
4090 - }
4091 - }
4092 -
4093 - // ADD ACF FIELDS SUPPORT
4094 - $acf_fields = $this->mxchat_get_acf_fields_for_post($post_id);
4095 - $pdf_extracted_count = 0;
4096 - if (!empty($acf_fields)) {
4097 - $acf_content_parts = array();
4098 - $pdf_attachment_ids = array();
4099 -
4100 - foreach ($acf_fields as $field_name => $field_value) {
4101 - $formatted_value = $this->mxchat_format_acf_field_value($field_value, $field_name, $post_id);
4102 -
4103 - if (!empty($formatted_value)) {
4104 - $field_label = ucwords(str_replace('_', ' ', $field_name));
4105 - $acf_content_parts[] = $field_label . ": " . $formatted_value;
4106 - }
4107 -
4108 - // Walk this field's value tree for any PDF attachment references and queue them for extraction.
4109 - // Only when the user opted into ACF→PDF extraction for this batch; otherwise the ACF text
4110 - // still lands in the KB but the heavier PDF parsing is skipped.
4111 - if ($extract_acf_pdfs) {
4112 - $this->mxchat_collect_pdf_attachment_ids_from_acf_value($field_value, $pdf_attachment_ids);
4113 - }
4114 - }
4115 -
4116 - if (!empty($acf_content_parts)) {
4117 - $content .= "\n\n" . implode("\n", $acf_content_parts);
4118 - }
4119 -
4120 - // Extract text from each unique PDF found in ACF fields and append as a labeled section
4121 - if ($extract_acf_pdfs && !empty($pdf_attachment_ids)) {
4122 - $pdf_attachment_ids = array_unique(array_filter(array_map('intval', $pdf_attachment_ids)));
4123 - $pdf_sections = array();
4124 - foreach ($pdf_attachment_ids as $att_id) {
4125 - $pdf_text = $this->mxchat_extract_pdf_text_by_attachment_id($att_id);
4126 - if (!empty($pdf_text)) {
4127 - $pdf_title = get_the_title($att_id);
4128 - $pdf_url = wp_get_attachment_url($att_id);
4129 - $header = 'PDF Attachment';
4130 - if (!empty($pdf_title)) {
4131 - $header .= ': ' . $pdf_title;
4132 - }
4133 - if (!empty($pdf_url)) {
4134 - $header .= ' (' . $pdf_url . ')';
4135 - }
4136 - $pdf_sections[] = $header . "\n" . $pdf_text;
4137 - $pdf_extracted_count++;
4138 - }
4139 - }
4140 - if (!empty($pdf_sections)) {
4141 - $content .= "\n\nAttached PDFs:\n" . implode("\n\n", $pdf_sections);
4142 - }
4143 - }
4144 - }
4145 -
4146 - // ADD CUSTOM POST META SUPPORT (whitelisted non-ACF meta fields)
4147 - $custom_meta = $this->mxchat_get_whitelisted_post_meta($post_id);
4148 - if (!empty($custom_meta)) {
4149 - $meta_content_parts = array();
4150 -
4151 - foreach ($custom_meta as $meta_key => $meta_value) {
4152 - // Convert meta key to readable label
4153 - $meta_label = ucwords(str_replace(array('_', '-'), ' ', $meta_key));
4154 - $meta_content_parts[] = $meta_label . ": " . $meta_value;
4155 - }
4156 -
4157 - if (!empty($meta_content_parts)) {
4158 - $content .= "\n\n" . implode("\n", $meta_content_parts);
4159 - }
4160 - }
4161 -
4162 - // Debug logging for WordPress Import content
4163 - //error_log('[MXCHAT-WP-IMPORT-DEBUG] Post ID: ' . $post_id . ' Title: ' . $post->post_title);
4164 - //error_log('[MXCHAT-WP-IMPORT-DEBUG] Raw post_content length: ' . strlen($post->post_content));
4165 - //error_log('[MXCHAT-WP-IMPORT-DEBUG] Final content length: ' . strlen($content));
4166 - //error_log('[MXCHAT-WP-IMPORT-DEBUG] Content preview: ' . substr($content, 0, 300));
4167 -
4168 - // Note: Removed 10,000 char limit - chunking now handles large content properly
4169 -
4170 - // Get bot-specific API key
4171 - $bot_options = $this->get_bot_options($bot_id);
4172 - $options = !empty($bot_options) ? $bot_options : get_option('mxchat_options');
4173 - $selected_model = $options['embedding_model'] ?? 'text-embedding-ada-002';
4174 -
4175 - if (strpos($selected_model, 'voyage') === 0) {
4176 - $api_key = $options['voyage_api_key'] ?? '';
4177 - $provider_name = 'Voyage AI';
4178 - } elseif (strpos($selected_model, 'gemini-embedding') === 0) {
4179 - $api_key = $options['gemini_api_key'] ?? '';
4180 - $provider_name = 'Google Gemini';
4181 - } else {
4182 - $api_key = $options['api_key'] ?? '';
4183 - $provider_name = 'OpenAI';
4184 - }
4185 -
4186 - if (empty($api_key)) {
4187 - MxChat_Admin::mxchat_log_debug('api_error', $provider_name . ' API key not configured for knowledge processing');
4188 - wp_send_json_error($provider_name . ' API key not configured');
4189 - exit;
4190 - }
4191 -
4192 - $source_url = get_permalink($post_id);
4193 - $vector_id = md5($source_url); // Vector ID for Pinecone
4194 -
4195 - // Check for existing content in bot-specific storage
4196 - $is_update = false;
4197 -
4198 - // Get bot-specific Pinecone configuration
4199 - $bot_pinecone_config = $this->get_bot_pinecone_config($bot_id);
4200 - $use_pinecone = !empty($bot_pinecone_config) && ($bot_pinecone_config['use_pinecone'] ?? false);
4201 -
4202 - if ($use_pinecone && !empty($bot_pinecone_config['api_key'])) {
4203 - // Check Pinecone for this bot
4204 - $pinecone_data = $this->mxchat_get_pinecone_processed_content($bot_pinecone_config);
4205 - if (isset($pinecone_data[$post_id])) {
4206 - $is_update = true;
4207 - }
4208 - } else {
4209 - // Check WordPress DB (same as before since it's shared)
4210 - global $wpdb;
4211 - $table_name = $wpdb->prefix . 'mxchat_system_prompt_content';
4212 - $existing_record = $wpdb->get_row($wpdb->prepare(
4213 - "SELECT id FROM $table_name WHERE source_url = %s",
4214 - $source_url
4215 - ));
4216 -
4217 - if ($existing_record) {
4218 - $is_update = true;
4219 - }
4220 - }
4221 -
4222 - // UPDATED 2.5.6: Determine content type based on post_type
4223 - $post_type = $post->post_type;
4224 - $content_type = 'content'; // Default fallback
4225 -
4226 - // Map WordPress post types to content types
4227 - switch ($post_type) {
4228 - case 'post':
4229 - $content_type = 'post';
4230 - break;
4231 - case 'page':
4232 - $content_type = 'page';
4233 - break;
4234 - case 'product':
4235 - $content_type = 'product';
4236 - break;
4237 - default:
4238 - // For custom post types, use the post type name
4239 - $content_type = sanitize_key($post_type);
4240 - break;
4241 - }
4242 -
4243 - // Use the centralized utility function with bot_id and content_type
4244 - $result = MxChat_Utils::submit_content_to_db(
4245 - $content,
4246 - $source_url,
4247 - $api_key,
4248 - $vector_id,
4249 - $bot_id,
4250 - $content_type
4251 - );
4252 -
4253 - if (is_wp_error($result)) {
4254 - MxChat_Admin::mxchat_log_debug('storage_error', 'Knowledge storage failed: ' . $result->get_error_message(), array('source_url' => $source_url));
4255 - wp_send_json_error('Storage failed: ' . $result->get_error_message());
4256 - exit;
4257 - }
4258 -
4259 - // Automatically apply role restriction based on tags
4260 - $this->apply_role_restriction_to_post($post_id, $source_url);
4261 -
4262 - $operation_type = $is_update ? 'update' : 'new';
4263 -
4264 - // Count ACF fields for debugging
4265 - $acf_field_count = count($acf_fields);
4266 -
4267 - // Success response with minimal data
4268 - wp_send_json_success(array(
4269 - 'message' => $operation_type === 'update' ? 'Content updated successfully' : 'Content processed successfully',
4270 - 'post_id' => $post_id,
4271 - 'title' => $post->post_title,
4272 - 'operation_type' => $operation_type,
4273 - 'vector_id' => $vector_id,
4274 - 'acf_fields_found' => $acf_field_count,
4275 - 'pdf_extracted_count' => (int) $pdf_extracted_count,
4276 - 'content_preview' => substr($content, 0, 100) . '...',
4277 - 'bot_id' => $bot_id
4278 - ));
4279 - exit;
4280 -}
4281 -
4282 -private function apply_role_restriction_to_post($post_id, $source_url) {
4283 - // Get tag-role mappings
4284 - $mappings = get_option('mxchat_tag_role_mappings', array());
4285 -
4286 - if (empty($mappings)) {
4287 - return; // No mappings, leave as public
4288 - }
4289 -
4290 - // Get all tags for the post
4291 - $post_tags = wp_get_post_tags($post_id, array('fields' => 'slugs'));
4292 -
4293 - if (empty($post_tags)) {
4294 - return; // No tags, leave as public
4295 - }
4296 -
4297 - // Determine the highest role restriction based on tags
4298 - $highest_role = 'public';
4299 - $role_hierarchy = array(
4300 - 'public' => 0,
4301 - 'logged_in' => 1,
4302 - 'subscriber' => 2,
4303 - 'contributor' => 3,
4304 - 'author' => 4,
4305 - 'editor' => 5,
4306 - 'administrator' => 6
4307 - );
4308 -
4309 - foreach ($post_tags as $tag_slug) {
4310 - if (isset($mappings[$tag_slug])) {
4311 - $role = $mappings[$tag_slug];
4312 - if (isset($role_hierarchy[$role]) && $role_hierarchy[$role] > $role_hierarchy[$highest_role]) {
4313 - $highest_role = $role;
4314 - }
4315 - }
4316 - }
4317 -
4318 - // If no restricted tags found, return (leave as public)
4319 - if ($highest_role === 'public') {
4320 - return;
4321 - }
4322 -
4323 - // Update the role restriction in the database
4324 - global $wpdb;
4325 -
4326 - // Check if using Pinecone
4327 - $pinecone_options = get_option('mxchat_pinecone_addon_options', array());
4328 - $use_pinecone = ($pinecone_options['mxchat_use_pinecone'] ?? '0') === '1';
4329 -
4330 - if ($use_pinecone && !empty($pinecone_options['mxchat_pinecone_api_key'])) {
4331 - // Update Pinecone role restriction
4332 - $roles_table = $wpdb->prefix . 'mxchat_pinecone_roles';
4333 - $vector_id = md5($source_url);
4334 -
4335 - $wpdb->replace(
4336 - $roles_table,
4337 - array(
4338 - 'vector_id' => $vector_id,
4339 - 'role_restriction' => $highest_role,
4340 - 'updated_at' => current_time('mysql')
4341 - ),
4342 - array('%s', '%s', '%s')
4343 - );
4344 - } else {
4345 - // Update WordPress DB
4346 - $table_name = $wpdb->prefix . 'mxchat_system_prompt_content';
4347 -
4348 - $wpdb->update(
4349 - $table_name,
4350 - array('role_restriction' => $highest_role),
4351 - array('source_url' => $source_url),
4352 - array('%s'),
4353 - array('%s')
4354 - );
4355 - }
4356 -}
4357 -
4358 -public function mxchat_get_public_post_types() {
4359 - // Get all public post types
4360 - $post_types = get_post_types(array('public' => true), 'objects');
4361 - $post_type_options = array();
4362 -
4363 - foreach ($post_types as $post_type) {
4364 - $post_type_options[$post_type->name] = $post_type->label;
4365 - }
4366 -
4367 - // Also include common forum/community post types that might not be marked as public
4368 - $additional_types = array(
4369 - 'topic' => 'Forum Topics (bbPress)',
4370 - 'reply' => 'Forum Replies (bbPress)',
4371 - 'forum' => 'Forums (bbPress)',
4372 - 'wpforo_topic' => 'wpForo Topics',
4373 - 'wpforo_post' => 'wpForo Posts'
4374 - );
4375 -
4376 - foreach ($additional_types as $type_name => $type_label) {
4377 - if (post_type_exists($type_name) && !isset($post_type_options[$type_name])) {
4378 - $post_type_options[$type_name] = $type_label;
4379 - }
4380 - }
4381 -
4382 - return $post_type_options;
4383 -}
4384 -
4385 -/**
4386 - * Retrieves processed content from Pinecone API
4387 - */
4388 -public function mxchat_get_pinecone_processed_content($pinecone_options) {
4389 - $api_key = $pinecone_options['mxchat_pinecone_api_key'] ?? '';
4390 - $host = $pinecone_options['mxchat_pinecone_host'] ?? '';
4391 -
4392 - if (empty($api_key) || empty($host)) {
4393 - return array();
4394 - }
4395 -
4396 - $pinecone_data = array();
4397 -
4398 - try {
4399 - // Always get fresh data from Pinecone
4400 - $pinecone_data = $this->mxchat_scan_pinecone_for_processed_content($pinecone_options);
4401 -
4402 - // Method 2: Final fallback - try stats endpoint (if available)
4403 - if (empty($pinecone_data)) {
4404 - $stats_url = "https://{$host}/describe_index_stats";
4405 -
4406 - $response = wp_remote_post($stats_url, array(
4407 - 'headers' => array(
4408 - 'Api-Key' => $api_key,
4409 - 'Content-Type' => 'application/json'
4410 - ),
4411 - 'body' => json_encode(array()),
4412 - 'timeout' => 30
4413 - ));
4414 -
4415 - if (!is_wp_error($response) && wp_remote_retrieve_response_code($response) === 200) {
4416 - $body = wp_remote_retrieve_body($response);
4417 - $stats_data = json_decode($body, true);
4418 - }
4419 - }
4420 -
4421 - } catch (Exception $e) {
4422 - // Log error but return fresh data only
4423 - }
4424 -
4425 - return $pinecone_data;
4426 -}
4427 -public function mxchat_fetch_pinecone_vectors_by_ids($pinecone_options, $vector_ids) {
4428 - //error_log('=== DEBUG: Starting mxchat_fetch_pinecone_vectors_by_ids ===');
4429 -
4430 - $api_key = $pinecone_options['mxchat_pinecone_api_key'] ?? '';
4431 - $host = $pinecone_options['mxchat_pinecone_host'] ?? '';
4432 -
4433 - if (empty($api_key) || empty($host) || empty($vector_ids)) {
4434 - //error_log('DEBUG: Missing parameters for fetch by IDs');
4435 - return array();
4436 - }
4437 -
4438 - try {
4439 - $fetch_url = "https://{$host}/vectors/fetch";
4440 - //error_log('DEBUG: Fetch URL: ' . $fetch_url);
4441 - //error_log('DEBUG: Fetching ' . count($vector_ids) . ' vector IDs');
4442 -
4443 - // Pinecone fetch API allows fetching specific vectors by ID
4444 - $fetch_data = array(
4445 - 'ids' => array_values($vector_ids)
4446 - );
4447 -
4448 - $response = wp_remote_post($fetch_url, array(
4449 - 'headers' => array(
4450 - 'Api-Key' => $api_key,
4451 - 'Content-Type' => 'application/json'
4452 - ),
4453 - 'body' => json_encode($fetch_data),
4454 - 'timeout' => 30
4455 - ));
4456 -
4457 - if (is_wp_error($response)) {
4458 - //error_log('DEBUG: Fetch by IDs WP error: ' . $response->get_error_message());
4459 - return array();
4460 - }
4461 -
4462 - $response_code = wp_remote_retrieve_response_code($response);
4463 - //error_log('DEBUG: Fetch response code: ' . $response_code);
4464 -
4465 - if ($response_code !== 200) {
4466 - $error_body = wp_remote_retrieve_body($response);
4467 - //error_log('DEBUG: Fetch failed with body: ' . $error_body);
4468 - return array();
4469 - }
4470 -
4471 - $body = wp_remote_retrieve_body($response);
4472 - $data = json_decode($body, true);
4473 -
4474 - //error_log('DEBUG: Fetch response structure: ' . print_r(array_keys($data), true));
4475 -
4476 - if (!isset($data['vectors'])) {
4477 - //error_log('DEBUG: No vectors key in response');
4478 - return array();
4479 - }
4480 -
4481 - $processed_data = array();
4482 -
4483 - foreach ($data['vectors'] as $vector_id => $vector_data) {
4484 - $metadata = $vector_data['metadata'] ?? array();
4485 - $source_url = $metadata['source_url'] ?? '';
4486 -
4487 - if (!empty($source_url)) {
4488 - $post_id = url_to_postid($source_url);
4489 - if ($post_id) {
4490 - $created_at = $metadata['created_at'] ?? '';
4491 - $processed_date = 'Recently';
4492 -
4493 - if (!empty($created_at)) {
4494 - $timestamp = is_numeric($created_at) ? $created_at : strtotime($created_at);
4495 - if ($timestamp) {
4496 - $processed_date = human_time_diff($timestamp, current_time('timestamp')) . ' ago';
4497 - }
4498 - }
4499 -
4500 - $processed_data[$post_id] = array(
4501 - 'db_id' => $vector_id,
4502 - 'processed_date' => $processed_date,
4503 - 'url' => $source_url,
4504 - 'source' => 'pinecone',
4505 - 'timestamp' => $timestamp ?? current_time('timestamp')
4506 - );
4507 - }
4508 - }
4509 - }
4510 -
4511 - //error_log('DEBUG: Processed ' . count($processed_data) . ' vectors from fetch');
4512 - return $processed_data;
4513 -
4514 - } catch (Exception $e) {
4515 - //error_log('DEBUG: Exception in fetch_pinecone_vectors_by_ids: ' . $e->getMessage());
4516 - return array();
4517 - }
4518 -}
4519 -
4520 -/**
4521 - * Get embedding dimensions based on the selected model.
4522 - */
4523 -private function mxchat_get_embedding_dimensions() {
4524 - $options = get_option('mxchat_options', array());
4525 - $selected_model = $options['embedding_model'] ?? 'text-embedding-ada-002';
4526 -
4527 - $model_dimensions = array(
4528 - 'text-embedding-ada-002' => 1536,
4529 - 'text-embedding-3-small' => 1536,
4530 - 'text-embedding-3-large' => 3072,
4531 - 'voyage-2' => 1024,
4532 - 'voyage-large-2' => 1536,
4533 - 'voyage-3-large' => 2048,
4534 - 'gemini-embedding-001' => 1536,
4535 - );
4536 -
4537 - if (strpos($selected_model, 'voyage-3-large') === 0) {
4538 - $custom_dimensions = $options['voyage_output_dimension'] ?? 2048;
4539 - return intval($custom_dimensions);
4540 - }
4541 -
4542 - if (strpos($selected_model, 'gemini-embedding') === 0) {
4543 - $custom_dimensions = $options['gemini_output_dimension'] ?? 1536;
4544 - return intval($custom_dimensions);
4545 - }
4546 -
4547 - return $model_dimensions[$selected_model] ?? 1536;
4548 -}
4549 -
4550 -/**
4551 - * Scan Pinecone for processed content
4552 - */
4553 -public function mxchat_scan_pinecone_for_processed_content($pinecone_options) {
4554 - $api_key = $pinecone_options['mxchat_pinecone_api_key'] ?? '';
4555 - $host = $pinecone_options['mxchat_pinecone_host'] ?? '';
4556 -
4557 - if (empty($api_key) || empty($host)) {
4558 - return array();
4559 - }
4560 -
4561 - try {
4562 - // Use multiple random vectors to get better coverage
4563 - $all_matches = array();
4564 - $seen_ids = array();
4565 -
4566 - // Get correct dimensions for the configured embedding model
4567 - $dimensions = $this->mxchat_get_embedding_dimensions();
4568 -
4569 - // Try 3 different random vectors to get better coverage
4570 - for ($i = 0; $i < 3; $i++) {
4571 - $query_url = "https://{$host}/query";
4572 -
4573 - // Generate a random unit vector instead of zeros
4574 - $random_vector = array();
4575 - for ($j = 0; $j < $dimensions; $j++) {
4576 - $random_vector[] = (rand(-1000, 1000) / 1000.0);
4577 - }
4578 -
4579 - // Normalize the vector to unit length
4580 - $magnitude = sqrt(array_sum(array_map(function($x) { return $x * $x; }, $random_vector)));
4581 - if ($magnitude > 0) {
4582 - $random_vector = array_map(function($x) use ($magnitude) { return $x / $magnitude; }, $random_vector);
4583 - }
4584 -
4585 - $query_data = array(
4586 - 'includeMetadata' => true,
4587 - 'includeValues' => false,
4588 - 'topK' => 10000,
4589 - 'vector' => $random_vector
4590 - );
4591 -
4592 - $response = wp_remote_post($query_url, array(
4593 - 'headers' => array(
4594 - 'Api-Key' => $api_key,
4595 - 'Content-Type' => 'application/json'
4596 - ),
4597 - 'body' => json_encode($query_data),
4598 - 'timeout' => 30
4599 - ));
4600 -
4601 - if (is_wp_error($response)) {
4602 - continue;
4603 - }
4604 -
4605 - $response_code = wp_remote_retrieve_response_code($response);
4606 -
4607 - if ($response_code !== 200) {
4608 - continue;
4609 - }
4610 -
4611 - $body = wp_remote_retrieve_body($response);
4612 - $data = json_decode($body, true);
4613 -
4614 - if (isset($data['matches'])) {
4615 - foreach ($data['matches'] as $match) {
4616 - $match_id = $match['id'] ?? '';
4617 - if (!empty($match_id) && !isset($seen_ids[$match_id])) {
4618 - $all_matches[] = $match;
4619 - $seen_ids[$match_id] = true;
4620 - }
4621 - }
4622 - }
4623 - }
4624 -
4625 - // Convert matches to processed data format, grouping by URL to count chunks
4626 - $processed_data = array();
4627 - $url_chunk_counts = array();
4628 -
4629 - foreach ($all_matches as $match) {
4630 - $metadata = $match['metadata'] ?? array();
4631 - $source_url = $metadata['source_url'] ?? '';
4632 - $match_id = $match['id'] ?? '';
4633 -
4634 - if (!empty($source_url) && !empty($match_id)) {
4635 - $post_id = url_to_postid($source_url);
4636 - if ($post_id) {
4637 - // Count chunks per post_id
4638 - if (!isset($url_chunk_counts[$post_id])) {
4639 - $url_chunk_counts[$post_id] = 0;
4640 - }
4641 - $url_chunk_counts[$post_id]++;
4642 -
4643 - $created_at = $metadata['created_at'] ?? '';
4644 - $processed_date = 'Recently';
4645 -
4646 - if (!empty($created_at)) {
4647 - $timestamp = is_numeric($created_at) ? $created_at : strtotime($created_at);
4648 - if ($timestamp) {
4649 - $processed_date = human_time_diff($timestamp, current_time('timestamp')) . ' ago';
4650 - }
4651 - }
4652 -
4653 - // Only store if not already set, or update with newer timestamp
4654 - if (!isset($processed_data[$post_id]) ||
4655 - ($timestamp ?? 0) > ($processed_data[$post_id]['timestamp'] ?? 0)) {
4656 - $processed_data[$post_id] = array(
4657 - 'db_id' => $match_id,
4658 - 'processed_date' => $processed_date,
4659 - 'url' => $source_url,
4660 - 'source' => 'pinecone',
4661 - 'timestamp' => $timestamp ?? current_time('timestamp')
4662 - );
4663 - }
4664 - }
4665 - }
4666 - }
4667 -
4668 - // Add chunk counts to processed data
4669 - foreach ($url_chunk_counts as $post_id => $chunk_count) {
4670 - if (isset($processed_data[$post_id])) {
4671 - $processed_data[$post_id]['chunk_count'] = $chunk_count;
4672 - }
4673 - }
4674 -
4675 - return $processed_data;
4676 -
4677 - } catch (Exception $e) {
4678 - return array();
4679 - }
4680 -}
4681 -/**
4682 - * Generate embeddings from input text for MXChat with bot support
4683 - */
4684 -private function mxchat_generate_embedding($text, $bot_id = 'default') {
4685 - // Enable detailed logging for debugging
4686 - //error_log('[MXCHAT-EMBED] Starting embedding generation for bot: ' . $bot_id . '. Text length: ' . strlen($text) . ' bytes');
4687 - //error_log('[MXCHAT-EMBED] Text preview: ' . substr($text, 0, 100) . '...');
4688 -
4689 - // Get bot-specific options
4690 - $bot_options = $this->get_bot_options($bot_id);
4691 - $options = !empty($bot_options) ? $bot_options : get_option('mxchat_options');
4692 -
4693 - // Opt-in: when the custom provider is selected for embeddings, index through
4694 - // the same custom endpoint the query path uses so stored vectors and query
4695 - // vectors share a model. Returns the vector array on success, or an error
4696 - // string on failure (this function's existing failure contract).
4697 - if (isset($options['custom_provider_for_embeddings']) && $options['custom_provider_for_embeddings'] === 'on') {
4698 - if (!class_exists('MxChat_Utils')) {
4699 - require_once dirname(__FILE__) . '/../includes/class-mxchat-utils.php';
4700 - }
4701 - return MxChat_Utils::generate_embedding_custom($text, $options);
4702 - }
4703 -
4704 - $selected_model = $options['embedding_model'] ?? 'text-embedding-ada-002';
4705 - //error_log('[MXCHAT-EMBED] Selected embedding model for bot ' . $bot_id . ': ' . $selected_model);
4706 -
4707 - // Determine provider and endpoint
4708 - if (strpos($selected_model, 'voyage') === 0) {
4709 - $api_key = $options['voyage_api_key'] ?? '';
4710 - $endpoint = 'https://api.voyageai.com/v1/embeddings';
4711 - $provider_name = 'Voyage AI';
4712 - //error_log('[MXCHAT-EMBED] Using Voyage AI API for bot ' . $bot_id);
4713 - } elseif (strpos($selected_model, 'gemini-embedding') === 0) {
4714 - $api_key = $options['gemini_api_key'] ?? '';
4715 - $endpoint = 'https://generativelanguage.googleapis.com/v1beta/models/' . $selected_model . ':embedContent';
4716 - $provider_name = 'Google Gemini';
4717 - //error_log('[MXCHAT-EMBED] Using Google Gemini API for bot ' . $bot_id);
4718 - } else {
4719 - $api_key = $options['api_key'] ?? '';
4720 - $endpoint = 'https://api.openai.com/v1/embeddings';
4721 - $provider_name = 'OpenAI';
4722 - //error_log('[MXCHAT-EMBED] Using OpenAI API for bot ' . $bot_id);
4723 - }
4724 -
4725 - //error_log('[MXCHAT-EMBED] Using endpoint: ' . $endpoint);
4726 -
4727 - if (empty($api_key)) {
4728 - $error_message = sprintf('Missing %s API key for bot %s. Please configure your API key in the bot settings.', $provider_name, $bot_id);
4729 - //error_log('[MXCHAT-EMBED] Error: ' . $error_message);
4730 - return $error_message;
4731 - }
4732 -
4733 - // Check if text is too long (for OpenAI, roughly estimate tokens as words/0.75)
4734 - $estimated_tokens = ceil(str_word_count($text) / 0.75);
4735 - //error_log('[MXCHAT-EMBED] Estimated token count: ~' . $estimated_tokens);
4736 -
4737 - if ($estimated_tokens > 8000 && strpos($selected_model, 'voyage') === false && strpos($selected_model, 'gemini-embedding') === false) {
4738 - //error_log('[MXCHAT-EMBED] Warning: Text may exceed OpenAI token limits (8K for most models)');
4739 - // Consider truncating text here
4740 - }
4741 -
4742 - // Prepare request body based on provider
4743 - if (strpos($selected_model, 'gemini-embedding') === 0) {
4744 - // Gemini API format
4745 - $request_body = array(
4746 - 'model' => 'models/' . $selected_model,
4747 - 'content' => array(
4748 - 'parts' => array(
4749 - array('text' => $text)
4750 - )
4751 - )
4752 - );
4753 -
4754 - // Set output dimensionality to 1536 for consistency with other models
4755 - $request_body['outputDimensionality'] = 1536;
4756 - } else {
4757 - // OpenAI/Voyage API format
4758 - $request_body = array(
4759 - 'model' => $selected_model,
4760 - 'input' => $text
4761 - );
4762 -
4763 - // Add output_dimension for voyage-3-large model
4764 - if ($selected_model === 'voyage-3-large') {
4765 - $request_body['output_dimension'] = 2048;
4766 - }
4767 - }
4768 -
4769 - //error_log('[MXCHAT-EMBED] Request prepared with model: ' . $selected_model);
4770 -
4771 - // Prepare headers based on provider
4772 - if (strpos($selected_model, 'gemini-embedding') === 0) {
4773 - // Gemini uses API key as query parameter
4774 - $endpoint .= '?key=' . $api_key;
4775 - $headers = array(
4776 - 'Content-Type' => 'application/json'
4777 - );
4778 - } else {
4779 - // OpenAI/Voyage use Bearer token
4780 - $headers = array(
4781 - 'Authorization' => 'Bearer ' . $api_key,
4782 - 'Content-Type' => 'application/json'
4783 - );
4784 - }
4785 -
4786 - // Make API request
4787 - //error_log('[MXCHAT-EMBED] Sending API request to: ' . $endpoint);
4788 - $response = wp_remote_post($endpoint, array(
4789 - 'body' => wp_json_encode($request_body),
4790 - 'headers' => $headers,
4791 - 'timeout' => 60 // Increased timeout for large inputs
4792 - ));
4793 -
4794 - // Handle wp_remote_post errors
4795 - if (is_wp_error($response)) {
4796 - $error_message = $response->get_error_message();
4797 - //error_log('[MXCHAT-EMBED] WP Remote Post Error: ' . $error_message);
4798 - return 'Connection error: ' . $error_message;
4799 - }
4800 -
4801 - // Get and check HTTP response code
4802 - $http_code = wp_remote_retrieve_response_code($response);
4803 - //error_log('[MXCHAT-EMBED] API Response Code: ' . $http_code);
4804 -
4805 - if ($http_code !== 200) {
4806 - $error_body = wp_remote_retrieve_body($response);
4807 - //error_log('[MXCHAT-EMBED] API Error Response Body: ' . $error_body);
4808 -
4809 - // Try to parse error for more details
4810 - $error_json = json_decode($error_body, true);
4811 - if (json_last_error() === JSON_ERROR_NONE && isset($error_json['error'])) {
4812 - $error_type = $error_json['error']['type'] ?? 'unknown';
4813 - $error_message = $error_json['error']['message'] ?? 'No message';
4814 - //error_log('[MXCHAT-EMBED] API Error Type: ' . $error_type);
4815 - //error_log('[MXCHAT-EMBED] API Error Message: ' . $error_message);
4816 -
4817 - // Customize error message for common API errors
4818 - if ($error_type === 'invalid_request_error' && strpos($error_message, 'API key') !== false) {
4819 - $error_message = sprintf('Invalid %s API key for bot %s. Please check your API key in the bot settings.', $provider_name, $bot_id);
4820 - } elseif ($error_type === 'authentication_error') {
4821 - $error_message = sprintf('%s authentication failed for bot %s. Please verify your API key in the bot settings.', $provider_name, $bot_id);
4822 - }
4823 -
4824 - //error_log('[MXCHAT-EMBED] Returning error: ' . $error_message);
4825 - return $error_message;
4826 - }
4827 -
4828 - $error_message = sprintf("API Error (HTTP %d): Unable to generate embedding for bot %s", $http_code, $bot_id);
4829 - //error_log('[MXCHAT-EMBED] Returning error: ' . $error_message);
4830 - return $error_message;
4831 - }
4832 -
4833 - // Parse response body
4834 - $response_body = wp_remote_retrieve_body($response);
4835 - //error_log('[MXCHAT-EMBED] Received response length: ' . strlen($response_body) . ' bytes');
4836 -
4837 - $response_data = json_decode($response_body, true);
4838 -
4839 - if (json_last_error() !== JSON_ERROR_NONE) {
4840 - $error = json_last_error_msg();
4841 - //error_log('[MXCHAT-EMBED] JSON Parse Error: ' . $error);
4842 - //error_log('[MXCHAT-EMBED] Response preview: ' . substr($response_body, 0, 200));
4843 - return "Failed to parse API response: $error";
4844 - }
4845 -
4846 - // Handle different response formats based on provider
4847 - if (strpos($selected_model, 'gemini-embedding') === 0) {
4848 - // Gemini API response format
4849 - if (isset($response_data['embedding']['values'])) {
4850 - $embedding_dimensions = count($response_data['embedding']['values']);
4851 - //error_log('[MXCHAT-EMBED] Successfully extracted Gemini embedding with ' . $embedding_dimensions . ' dimensions');
4852 -
4853 - // Check if embedding dimensions are as expected (should be 1536)
4854 - if ($embedding_dimensions !== 1536) {
4855 - //error_log('[MXCHAT-EMBED] Warning: Unexpected Gemini embedding dimensions: ' . $embedding_dimensions);
4856 - }
4857 -
4858 - return $response_data['embedding']['values'];
4859 - } else {
4860 - //error_log('[MXCHAT-EMBED] Error: No embedding found in Gemini response');
4861 - //error_log('[MXCHAT-EMBED] Response structure: ' . wp_json_encode(array_keys($response_data)));
4862 -
4863 - if (isset($response_data['error'])) {
4864 - $error_message = "Gemini API Error in response: " . wp_json_encode($response_data['error']);
4865 - //error_log('[MXCHAT-EMBED] ' . $error_message);
4866 - return $error_message;
4867 - }
4868 -
4869 - $error_message = "Invalid Gemini API response format: No embedding found";
4870 - //error_log('[MXCHAT-EMBED] ' . $error_message);
4871 - return $error_message;
4872 - }
4873 - } else {
4874 - // OpenAI/Voyage API response format
4875 - if (isset($response_data['data'][0]['embedding'])) {
4876 - $embedding_dimensions = count($response_data['data'][0]['embedding']);
4877 - //error_log('[MXCHAT-EMBED] Successfully extracted embedding with ' . $embedding_dimensions . ' dimensions');
4878 -
4879 - // Check if embedding dimensions are as expected
4880 - if (($selected_model === 'text-embedding-ada-002' && $embedding_dimensions !== 1536) ||
4881 - ($selected_model === 'voyage-3-large' && $embedding_dimensions !== 2048)) {
4882 - //error_log('[MXCHAT-EMBED] Warning: Unexpected embedding dimensions');
4883 - }
4884 -
4885 - return $response_data['data'][0]['embedding'];
4886 - } else {
4887 - //error_log('[MXCHAT-EMBED] Error: No embedding found in response');
4888 - //error_log('[MXCHAT-EMBED] Response structure: ' . wp_json_encode(array_keys($response_data)));
4889 -
4890 - if (isset($response_data['error'])) {
4891 - $error_message = "API Error in response: " . wp_json_encode($response_data['error']);
4892 - //error_log('[MXCHAT-EMBED] ' . $error_message);
4893 - return $error_message;
4894 - }
4895 -
4896 - $error_message = "Invalid API response format: No embedding found";
4897 - //error_log('[MXCHAT-EMBED] ' . $error_message);
4898 - return $error_message;
4899 - }
4900 - }
4901 -}
4902 -
4903 -/**
4904 - * Get bot-specific options for multi-bot functionality
4905 - * Falls back to default options if bot_id is 'default' or multi-bot add-on is not active
4906 - */
4907 -private function get_bot_options($bot_id = 'default') {
4908 - //error_log("MXCHAT DEBUG: get_bot_options called for bot: " . $bot_id);
4909 -
4910 - if ($bot_id === 'default' || !class_exists('MxChat_Multi_Bot_Manager')) {
4911 - //error_log("MXCHAT DEBUG: Using default options (no multi-bot or bot is 'default')");
4912 - return array();
4913 - }
4914 -
4915 - $bot_options = apply_filters('mxchat_get_bot_options', array(), $bot_id);
4916 -
4917 - if (!empty($bot_options)) {
4918 - //error_log("MXCHAT DEBUG: Got bot-specific options from filter");
4919 - if (isset($bot_options['similarity_threshold'])) {
4920 - //error_log(" - similarity_threshold: " . $bot_options['similarity_threshold']);
4921 - }
4922 - }
4923 -
4924 - return is_array($bot_options) ? $bot_options : array();
4925 -}
4926 -
4927 -/**
4928 - * Get bot-specific Pinecone configuration
4929 - * Used in the knowledge retrieval functions
4930 - */
4931 -// Also add debugging to your get_bot_pinecone_config function
4932 -private function get_bot_pinecone_config($bot_id = 'default') {
4933 - //error_log("MXCHAT DEBUG: get_bot_pinecone_config called for bot: " . $bot_id);
4934 -
4935 - // If default bot or multi-bot add-on not active, use default Pinecone config
4936 - if ($bot_id === 'default' || !class_exists('MxChat_Multi_Bot_Manager')) {
4937 - //error_log("MXCHAT DEBUG: Using default Pinecone config (no multi-bot or bot is 'default')");
4938 - $addon_options = get_option('mxchat_pinecone_addon_options', array());
4939 - $config = array(
4940 - 'use_pinecone' => (isset($addon_options['mxchat_use_pinecone']) && $addon_options['mxchat_use_pinecone'] === '1'),
4941 - 'api_key' => $addon_options['mxchat_pinecone_api_key'] ?? '',
4942 - 'host' => $addon_options['mxchat_pinecone_host'] ?? '',
4943 - 'namespace' => $addon_options['mxchat_pinecone_namespace'] ?? ''
4944 - );
4945 - //error_log("MXCHAT DEBUG: Default config - use_pinecone: " . ($config['use_pinecone'] ? 'true' : 'false'));
4946 - return $config;
4947 - }
4948 -
4949 - //error_log("MXCHAT DEBUG: Calling filter 'mxchat_get_bot_pinecone_config' for bot: " . $bot_id);
4950 -
4951 - // Hook for multi-bot add-on to provide bot-specific Pinecone config
4952 - $bot_pinecone_config = apply_filters('mxchat_get_bot_pinecone_config', array(), $bot_id);
4953 -
4954 - if (!empty($bot_pinecone_config)) {
4955 - //error_log("MXCHAT DEBUG: Got bot-specific config from filter");
4956 - //error_log(" - use_pinecone: " . (isset($bot_pinecone_config['use_pinecone']) ? ($bot_pinecone_config['use_pinecone'] ? 'true' : 'false') : 'not set'));
4957 - //error_log(" - host: " . ($bot_pinecone_config['host'] ?? 'not set'));
4958 - //error_log(" - namespace: " . ($bot_pinecone_config['namespace'] ?? 'not set'));
4959 - } else {
4960 - //error_log("MXCHAT DEBUG: Filter returned empty config!");
4961 - }
4962 -
4963 - return is_array($bot_pinecone_config) ? $bot_pinecone_config : array();
4964 -}
4965 -
4966 -
4967 -public function mxchat_ajax_dismiss_completed_status() {
4968 - try {
4969 - // Verify the request
4970 - check_ajax_referer('mxchat_status_nonce', 'nonce');
4971 -
4972 - if (!current_user_can('manage_options')) {
4973 - wp_send_json_error('Unauthorized access');
4974 - exit;
4975 - }
4976 -
4977 - $card_type = isset($_POST['card_type']) ? sanitize_text_field($_POST['card_type']) : '';
4978 -
4979 - if ($card_type === 'pdf') {
4980 - // Clear PDF status
4981 - $pdf_url = get_transient('mxchat_last_pdf_url');
4982 - if ($pdf_url) {
4983 - delete_transient('mxchat_pdf_status_' . md5($pdf_url));
4984 - delete_transient('mxchat_last_pdf_url');
4985 - }
4986 - } elseif ($card_type === 'sitemap') {
4987 - // Clear sitemap status
4988 - $sitemap_url = get_transient('mxchat_last_sitemap_url');
4989 - if ($sitemap_url) {
4990 - delete_transient('mxchat_sitemap_status_' . md5($sitemap_url));
4991 - delete_transient('mxchat_last_sitemap_url');
4992 - }
4993 - }
4994 -
4995 - wp_send_json_success(array('message' => 'Status dismissed successfully'));
4996 -
4997 - } catch (Exception $e) {
4998 - wp_send_json_error(array('message' => 'Error dismissing status: ' . $e->getMessage()));
4999 - }
5000 -}
5001 -
5002 -/**
5003 - * Render completed status cards on page load
5004 - * This ensures completed processing status persists through page refreshes
5005 - */
5006 -public function mxchat_render_completed_status_cards() {
5007 - $output = '';
5008 -
5009 - // Check for completed PDF status
5010 - $pdf_url = get_transient('mxchat_last_pdf_url');
5011 - if ($pdf_url) {
5012 - $pdf_status = $this->mxchat_get_pdf_processing_status($pdf_url);
5013 - if ($pdf_status && ($pdf_status['status'] === 'complete' || $pdf_status['status'] === 'error')) {
5014 - $output .= $this->mxchat_render_pdf_status_card($pdf_status, $pdf_url);
5015 - }
5016 - }
5017 -
5018 - // Check for completed sitemap status
5019 - $sitemap_url = get_transient('mxchat_last_sitemap_url');
5020 - if ($sitemap_url) {
5021 - $sitemap_status = $this->mxchat_get_sitemap_processing_status($sitemap_url);
5022 - if ($sitemap_status && ($sitemap_status['status'] === 'complete' || $sitemap_status['status'] === 'error')) {
5023 - $output .= $this->mxchat_render_sitemap_status_card($sitemap_status, $sitemap_url);
5024 - }
5025 - }
5026 -
5027 - return $output;
5028 -}
5029 -
5030 -/**
5031 - * Render PDF status card HTML
5032 - */
5033 -private function mxchat_render_pdf_status_card($status, $pdf_url) {
5034 - $html = '<div class="mxchat-status-card" data-card-type="pdf">';
5035 - $html .= '<div class="mxchat-status-header">';
5036 - $html .= '<h4>' . esc_html__('PDF Processing Status', 'mxchat') . '</h4>';
5037 -
5038 - // Add dismiss button for completed status
5039 - if ($status['status'] === 'complete' || $status['status'] === 'error') {
5040 - $html .= '<button type="button" class="mxchat-dismiss-button">' . esc_html__('Dismiss', 'mxchat') . '</button>';
5041 - }
5042 -
5043 - // Process Batch button for processing status
5044 - if ($status['status'] === 'processing') {
5045 - $html .= '<button type="button" class="mxchat-manual-batch-btn"
5046 - data-process-type="pdf"
5047 - data-url="' . esc_attr($pdf_url) . '">
5048 - ' . esc_html__('Process Batch', 'mxchat') . '</button>';
5049 - }
5050 -
5051 - // Add status badges
5052 - if ($status['status'] === 'error') {
5053 - $html .= '<span class="mxchat-status-badge mxchat-status-failed">' . esc_html__('Error', 'mxchat') . '</span>';
5054 - } elseif ($status['status'] === 'complete') {
5055 - if ($status['failed_pages'] > 0) {
5056 - $html .= '<span class="mxchat-status-badge mxchat-status-warning">' .
5057 - sprintf(esc_html__('Completed with %d failures', 'mxchat'), $status['failed_pages']) . '</span>';
5058 - } else {
5059 - $html .= '<span class="mxchat-status-badge mxchat-status-success">' . esc_html__('Complete', 'mxchat') . '</span>';
5060 - }
5061 - }
5062 -
5063 - $html .= '</div>'; // End header
5064 -
5065 - // Progress bar
5066 - $html .= '<div class="mxchat-progress-bar">';
5067 - $html .= '<div class="mxchat-progress-fill" style="width: ' . esc_attr($status['percentage']) . '%"></div>';
5068 - $html .= '</div>';
5069 -
5070 - // Status details
5071 - $html .= '<div class="mxchat-status-details">';
5072 - $html .= '<p>' . sprintf(
5073 - esc_html__('Progress: %d of %d pages (%d%%)', 'mxchat'),
5074 - $status['processed_pages'],
5075 - $status['total_pages'],
5076 - $status['percentage']
5077 - ) . '</p>';
5078 -
5079 - // Show failed pages count if any
5080 - if ($status['failed_pages'] > 0) {
5081 - $html .= '<p><strong>' . esc_html__('Failed pages:', 'mxchat') . '</strong> ' . esc_html($status['failed_pages']) . '</p>';
5082 - }
5083 -
5084 - $html .= '<p><strong>' . esc_html__('Status:', 'mxchat') . '</strong> ' . esc_html(ucfirst($status['status'])) . '</p>';
5085 - $html .= '<p><strong>' . esc_html__('Last update:', 'mxchat') . '</strong> ' . esc_html($status['last_update']) . '</p>';
5086 -
5087 - // Add completion summary if available AND it's an array
5088 - if (isset($status['completion_summary']) && is_array($status['completion_summary']) && !empty($status['completion_summary'])) {
5089 - $summary = $status['completion_summary'];
5090 - $html .= '<div class="mxchat-completion-summary">';
5091 - $html .= '<h5>' . esc_html__('Processing Summary', 'mxchat') . '</h5>';
5092 - $html .= '<p><strong>' . esc_html__('Total Pages:', 'mxchat') . '</strong> ' . esc_html($summary['total_pages']) . '</p>';
5093 - $html .= '<p><strong>' . esc_html__('Successful:', 'mxchat') . '</strong> ' . esc_html($summary['successful_pages']) . '</p>';
5094 - $html .= '<p><strong>' . esc_html__('Failed:', 'mxchat') . '</strong> ' . esc_html($summary['failed_pages']) . '</p>';
5095 - $html .= '<p><strong>' . esc_html__('Completed:', 'mxchat') . '</strong> ' . esc_html($summary['completion_time']) . '</p>';
5096 - $html .= '</div>';
5097 - }
5098 -
5099 - // Add failed pages list if any AND it's an array
5100 - if (isset($status['failed_pages_list']) && is_array($status['failed_pages_list']) && !empty($status['failed_pages_list'])) {
5101 - $html .= $this->mxchat_render_failed_pages_list($status['failed_pages_list']);
5102 - }
5103 -
5104 - // Add error message if any
5105 - if (isset($status['error']) && !empty($status['error'])) {
5106 - $html .= '<div class="mxchat-error-notice">';
5107 - $html .= '<p class="error">' . esc_html($status['error']) . '</p>';
5108 - $html .= '</div>';
5109 - }
5110 -
5111 - $html .= '</div>'; // End details
5112 - $html .= '</div>'; // End card
5113 -
5114 - return $html;
5115 -}
5116 -/**
5117 - * Render sitemap status card HTML
5118 - */
5119 -private function mxchat_render_sitemap_status_card($status, $sitemap_url) {
5120 - $html = '<div class="mxchat-status-card" data-card-type="sitemap">';
5121 - $html .= '<div class="mxchat-status-header">';
5122 - $html .= '<h4>' . esc_html__('Sitemap Processing Status', 'mxchat') . '</h4>';
5123 -
5124 - // Add dismiss button for completed status
5125 - if ($status['status'] === 'complete' || $status['status'] === 'error') {
5126 - $html .= '<button type="button" class="mxchat-dismiss-button">' . esc_html__('Dismiss', 'mxchat') . '</button>';
5127 - }
5128 -
5129 - // Process Batch button for processing status
5130 - if ($status['status'] === 'processing') {
5131 - $html .= '<button type="button" class="mxchat-manual-batch-btn"
5132 - data-process-type="sitemap"
5133 - data-url="' . esc_attr($sitemap_url) . '">
5134 - ' . esc_html__('Process Batch', 'mxchat') . '</button>';
5135 - }
5136 -
5137 - // Add status badges
5138 - if ($status['status'] === 'error') {
5139 - $html .= '<span class="mxchat-status-badge mxchat-status-failed">' . esc_html__('Error', 'mxchat') . '</span>';
5140 - } elseif ($status['status'] === 'complete') {
5141 - if ($status['failed_urls'] > 0) {
5142 - $html .= '<span class="mxchat-status-badge mxchat-status-warning">' .
5143 - sprintf(esc_html__('Completed with %d failures', 'mxchat'), $status['failed_urls']) . '</span>';
5144 - } else {
5145 - $html .= '<span class="mxchat-status-badge mxchat-status-success">' . esc_html__('Complete', 'mxchat') . '</span>';
5146 - }
5147 - }
5148 -
5149 - $html .= '</div>'; // End header
5150 -
5151 - // Progress bar
5152 - $html .= '<div class="mxchat-progress-bar">';
5153 - $html .= '<div class="mxchat-progress-fill" style="width: ' . esc_attr($status['percentage']) . '%"></div>';
5154 - $html .= '</div>';
5155 -
5156 - // Status details
5157 - $html .= '<div class="mxchat-status-details">';
5158 - $html .= '<p>' . sprintf(
5159 - esc_html__('Progress: %d of %d URLs (%d%%)', 'mxchat'),
5160 - $status['processed_urls'],
5161 - $status['total_urls'],
5162 - $status['percentage']
5163 - ) . '</p>';
5164 -
5165 - // Show failed URLs count if any
5166 - if ($status['failed_urls'] > 0) {
5167 - $html .= '<p><strong>' . esc_html__('Failed URLs:', 'mxchat') . '</strong> ' . esc_html($status['failed_urls']) . '</p>';
5168 - }
5169 -
5170 - $html .= '<p><strong>' . esc_html__('Status:', 'mxchat') . '</strong> ' . esc_html(ucfirst($status['status'])) . '</p>';
5171 - $html .= '<p><strong>' . esc_html__('Last update:', 'mxchat') . '</strong> ' . esc_html($status['last_update']) . '</p>';
5172 -
5173 - // Add completion summary if available AND it's an array
5174 - if (isset($status['completion_summary']) && is_array($status['completion_summary']) && !empty($status['completion_summary'])) {
5175 - $summary = $status['completion_summary'];
5176 - $html .= '<div class="mxchat-completion-summary">';
5177 - $html .= '<h5>' . esc_html__('Processing Summary', 'mxchat') . '</h5>';
5178 - $html .= '<p><strong>' . esc_html__('Total URLs:', 'mxchat') . '</strong> ' . esc_html($summary['total_urls']) . '</p>';
5179 - $html .= '<p><strong>' . esc_html__('Successful:', 'mxchat') . '</strong> ' . esc_html($summary['successful_urls']) . '</p>';
5180 - $html .= '<p><strong>' . esc_html__('Failed:', 'mxchat') . '</strong> ' . esc_html($summary['failed_urls']) . '</p>';
5181 - $html .= '<p><strong>' . esc_html__('Completed:', 'mxchat') . '</strong> ' . esc_html($summary['completion_time']) . '</p>';
5182 - $html .= '</div>';
5183 - }
5184 -
5185 - // Add error messages if any (but not the failed URLs list)
5186 - if (!empty($status['error']) || !empty($status['last_error'])) {
5187 - $html .= '<div class="mxchat-error-notice">';
5188 -
5189 - if (!empty($status['error'])) {
5190 - $html .= '<p class="error">' . esc_html($status['error']) . '</p>';
5191 - }
5192 -
5193 - if (!empty($status['last_error'])) {
5194 - $html .= '<p class="last-error">' . esc_html__('Last error:', 'mxchat') . ' ' . esc_html($status['last_error']) . '</p>';
5195 - }
5196 -
5197 - $html .= '</div>';
5198 - }
5199 -
5200 - $html .= '</div>'; // End details
5201 - $html .= '</div>'; // End card
5202 -
5203 - return $html;
5204 -}
5205 -
5206 -
5207 -/**
5208 - * Render failed pages list
5209 - */
5210 -private function mxchat_render_failed_pages_list($failed_pages_list) {
5211 - // Validate that $failed_pages_list is an array and not empty
5212 - if (!is_array($failed_pages_list) || empty($failed_pages_list)) {
5213 - return '';
5214 - }
5215 -
5216 - $html = '<div class="mxchat-error-notice">';
5217 - $html .= '<div class="mxchat-failed-pages-container">';
5218 - $html .= '<h5>' . sprintf(esc_html__('Failed Pages (%d)', 'mxchat'), count($failed_pages_list)) . '</h5>';
5219 - $html .= '<details>';
5220 - $html .= '<summary>' . esc_html__('Show Failed Pages', 'mxchat') . '</summary>';
5221 - $html .= '<div class="mxchat-failed-pages-list">';
5222 -
5223 - // Create table for failed pages
5224 - $html .= '<table class="widefat striped">';
5225 - $html .= '<thead><tr>';
5226 - $html .= '<th>' . esc_html__('Page', 'mxchat') . '</th>';
5227 - $html .= '<th>' . esc_html__('Error', 'mxchat') . '</th>';
5228 - $html .= '<th>' . esc_html__('Retries', 'mxchat') . '</th>';
5229 - $html .= '<th>' . esc_html__('Time', 'mxchat') . '</th>';
5230 - $html .= '</tr></thead><tbody>';
5231 -
5232 - // Sort failed pages by most recent
5233 - $sorted_failed_pages = $failed_pages_list;
5234 - usort($sorted_failed_pages, function($a, $b) {
5235 - return ($b['time'] ?? 0) - ($a['time'] ?? 0);
5236 - });
5237 -
5238 - foreach ($sorted_failed_pages as $item) {
5239 - // Ensure $item is an array before accessing its elements
5240 - if (!is_array($item)) {
5241 - continue;
5242 - }
5243 -
5244 - $time_ago = isset($item['time']) ? human_time_diff($item['time'], current_time('timestamp')) . ' ' . esc_html__('ago', 'mxchat') : 'Unknown';
5245 - $html .= '<tr>';
5246 - $html .= '<td>' . esc_html__('Page', 'mxchat') . ' ' . esc_html($item['page'] ?? 'Unknown') . '</td>';
5247 - $html .= '<td style="word-break: break-word;">' . esc_html($item['error'] ?? 'Unknown error') . '</td>';
5248 - $html .= '<td>' . esc_html($item['retries'] ?? 'N/A') . '</td>';
5249 - $html .= '<td>' . esc_html($time_ago) . '</td>';
5250 - $html .= '</tr>';
5251 - }
5252 -
5253 - $html .= '</tbody></table>';
5254 - $html .= '</div></details></div></div>';
5255 -
5256 - return $html;
5257 -}
5258 -
5259 -/**
5260 - * Render failed URLs list
5261 - */
5262 -private function mxchat_render_failed_urls_list($failed_urls_list) {
5263 - // Validate that $failed_urls_list is an array and not empty
5264 - if (!is_array($failed_urls_list) || empty($failed_urls_list)) {
5265 - return '';
5266 - }
5267 -
5268 - $html = '<div class="mxchat-failed-urls-container">';
5269 - $html .= '<h5>' . sprintf(esc_html__('Failed URLs (%d)', 'mxchat'), count($failed_urls_list)) . '</h5>';
5270 - $html .= '<details>';
5271 - $html .= '<summary>' . esc_html__('Show Failed URLs', 'mxchat') . '</summary>';
5272 - $html .= '<div class="mxchat-failed-urls-list">';
5273 -
5274 - // Create table for failed URLs
5275 - $html .= '<table class="widefat striped">';
5276 - $html .= '<thead><tr>';
5277 - $html .= '<th>' . esc_html__('URL', 'mxchat') . '</th>';
5278 - $html .= '<th>' . esc_html__('Error', 'mxchat') . '</th>';
5279 - $html .= '<th>' . esc_html__('Retries', 'mxchat') . '</th>';
5280 - $html .= '<th>' . esc_html__('Time', 'mxchat') . '</th>';
5281 - $html .= '</tr></thead><tbody>';
5282 -
5283 - // Sort failed URLs by most recent
5284 - $sorted_failed_urls = $failed_urls_list;
5285 - usort($sorted_failed_urls, function($a, $b) {
5286 - return ($b['time'] ?? 0) - ($a['time'] ?? 0);
5287 - });
5288 -
5289 - // Show up to 50 failed URLs
5290 - $display_urls = array_slice($sorted_failed_urls, 0, 50);
5291 -
5292 - foreach ($display_urls as $item) {
5293 - // Ensure $item is an array before accessing its elements
5294 - if (!is_array($item)) {
5295 - continue;
5296 - }
5297 -
5298 - $url = $item['url'] ?? '';
5299 - $time_ago = isset($item['time']) ? human_time_diff($item['time'], current_time('timestamp')) . ' ' . esc_html__('ago', 'mxchat') : 'Unknown';
5300 -
5301 - // Truncate URL for display
5302 - $display_url = strlen($url) > 50 ? substr($url, 0, 47) . '...' : $url;
5303 -
5304 - $html .= '<tr>';
5305 - $html .= '<td style="word-break: break-all;">';
5306 - if (!empty($url)) {
5307 - $html .= '<a href="' . esc_url($url) . '" target="_blank" rel="noopener noreferrer">' . esc_html($display_url) . '</a>';
5308 - } else {
5309 - $html .= esc_html__('Unknown URL', 'mxchat');
5310 - }
5311 - $html .= '</td>';
5312 - $html .= '<td style="word-break: break-word;">' . esc_html($item['error'] ?? 'Unknown error') . '</td>';
5313 - $html .= '<td>' . esc_html($item['retries'] ?? 'N/A') . '</td>';
5314 - $html .= '<td>' . esc_html($time_ago) . '</td>';
5315 - $html .= '</tr>';
5316 - }
5317 -
5318 - $html .= '</tbody></table>';
5319 -
5320 - if (count($failed_urls_list) > 50) {
5321 - $html .= '<div class="mxchat-failed-urls-more">+ ' .
5322 - (count($failed_urls_list) - 50) .
5323 - ' ' . esc_html__('more failed URLs not shown', 'mxchat') . '</div>';
5324 - }
5325 -
5326 - $html .= '</div></details></div>';
5327 -
5328 - return $html;
5329 -}
5330 -
5331 -/**
5332 - * Get all ACF fields for a specific post, excluding any fields the user has disabled
5333 - */
5334 -public function mxchat_get_acf_fields_for_post($post_id) {
5335 - if (!function_exists('get_fields')) {
5336 - return array();
5337 - }
5338 -
5339 - $fields = get_fields($post_id);
5340 - if (!$fields || !is_array($fields)) {
5341 - return array();
5342 - }
5343 -
5344 - // Get excluded fields from settings
5345 - $excluded_fields = get_option('mxchat_acf_excluded_fields', array());
5346 - if (!empty($excluded_fields) && is_array($excluded_fields)) {
5347 - foreach ($excluded_fields as $excluded_field) {
5348 - if (isset($fields[$excluded_field])) {
5349 - unset($fields[$excluded_field]);
5350 - }
5351 - }
5352 - }
5353 -
5354 - return $fields;
5355 -}
5356 -
5357 -/**
5358 - * Get all registered ACF field groups and their fields for the settings UI
5359 - */
5360 -public function mxchat_get_all_acf_fields() {
5361 - if (!function_exists('acf_get_field_groups') || !function_exists('acf_get_fields')) {
5362 - return array();
5363 - }
5364 -
5365 - $all_fields = array();
5366 - $field_groups = acf_get_field_groups();
5367 -
5368 - if (!empty($field_groups)) {
5369 - foreach ($field_groups as $group) {
5370 - $group_fields = acf_get_fields($group['key']);
5371 - if (!empty($group_fields)) {
5372 - $all_fields[$group['title']] = array();
5373 - foreach ($group_fields as $field) {
5374 - $all_fields[$group['title']][] = array(
5375 - 'name' => $field['name'],
5376 - 'label' => $field['label'],
5377 - 'type' => $field['type']
5378 - );
5379 - }
5380 - }
5381 - }
5382 - }
5383 -
5384 - return $all_fields;
5385 -}
5386 -
5387 -/**
5388 - * Get whitelisted custom post meta for a given post
5389 - * This allows non-ACF meta fields (like OptionTree, theme meta boxes) to be included in embeddings
5390 - */
5391 -public function mxchat_get_whitelisted_post_meta($post_id) {
5392 - $whitelist = get_option('mxchat_custom_meta_whitelist', '');
5393 -
5394 - if (empty($whitelist)) {
5395 - return array();
5396 - }
5397 -
5398 - // Parse the whitelist - one meta key per line
5399 - $meta_keys = array_filter(array_map('trim', explode("\n", $whitelist)));
5400 -
5401 - if (empty($meta_keys)) {
5402 - return array();
5403 - }
5404 -
5405 - $result = array();
5406 -
5407 - foreach ($meta_keys as $key) {
5408 - // Skip empty keys
5409 - if (empty($key)) {
5410 - continue;
5411 - }
5412 -
5413 - $value = get_post_meta($post_id, $key, true);
5414 -
5415 - // Only include non-empty string values
5416 - if (!empty($value) && is_string($value)) {
5417 - $result[$key] = $value;
5418 - } elseif (!empty($value) && is_array($value)) {
5419 - // Handle array values by joining them
5420 - $flat_value = $this->mxchat_flatten_meta_array($value);
5421 - if (!empty($flat_value)) {
5422 - $result[$key] = $flat_value;
5423 - }
5424 - }
5425 - }
5426 -
5427 - return $result;
5428 -}
5429 -
5430 -/**
5431 - * Flatten array meta values into a readable string
5432 - */
5433 -private function mxchat_flatten_meta_array($array, $depth = 0) {
5434 - if ($depth > 3) {
5435 - return ''; // Prevent infinite recursion
5436 - }
5437 -
5438 - $parts = array();
5439 -
5440 - foreach ($array as $key => $value) {
5441 - if (is_string($value) && !empty($value)) {
5442 - $parts[] = $value;
5443 - } elseif (is_array($value)) {
5444 - $nested = $this->mxchat_flatten_meta_array($value, $depth + 1);
5445 - if (!empty($nested)) {
5446 - $parts[] = $nested;
5447 - }
5448 - }
5449 - }
5450 -
5451 - return implode(', ', $parts);
5452 -}
5453 -
5454 -/**
5455 - * Format ACF field values for content extraction
5456 - */
5457 -public function mxchat_format_acf_field_value($value, $field_name = '', $post_id = 0) {
5458 - if (empty($value)) {
5459 - return '';
5460 - }
5461 -
5462 - // Handle WP_Post objects first (THIS IS THE KEY FIX)
5463 - if ($value instanceof WP_Post) {
5464 - return $value->post_title ?: '';
5465 - }
5466 -
5467 - // Handle other WP objects
5468 - if (is_object($value)) {
5469 - if (isset($value->post_title)) {
5470 - return $value->post_title;
5471 - } elseif (isset($value->display_name)) {
5472 - return $value->display_name;
5473 - } elseif (isset($value->name)) {
5474 - return $value->name;
5475 - } elseif (method_exists($value, '__toString')) {
5476 - try {
5477 - return (string) $value;
5478 - } catch (Exception $e) {
5479 - return '';
5480 - }
5481 - }
5482 - // For any other objects, return empty string
5483 - return '';
5484 - }
5485 -
5486 - // Handle different ACF field types
5487 - if (is_array($value)) {
5488 - // Check if it's an image/file field
5489 - if (isset($value['url'])) {
5490 - // Image field - return alt text, title, or caption
5491 - if (!empty($value['alt'])) {
5492 - return $value['alt'];
5493 - } elseif (!empty($value['title'])) {
5494 - return $value['title'];
5495 - } elseif (!empty($value['caption'])) {
5496 - return $value['caption'];
5497 - } else {
5498 - return ''; // Don't include just the URL
5499 - }
5500 - }
5501 -
5502 - // Check if it's a post object or relationship field
5503 - if (isset($value['post_title'])) {
5504 - return $value['post_title'];
5505 - }
5506 -
5507 - // Check if it's a user field
5508 - if (isset($value['display_name'])) {
5509 - return $value['display_name'];
5510 - }
5511 -
5512 - // Check if it's a taxonomy term
5513 - if (isset($value['name']) && isset($value['taxonomy'])) {
5514 - return $value['name'];
5515 - }
5516 -
5517 - // Check if it's a select field with label
5518 - if (isset($value['label'])) {
5519 - return $value['label'];
5520 - }
5521 -
5522 - // Check for repeater field or flexible content
5523 - if (is_numeric(key($value))) {
5524 - $sub_values = array();
5525 - foreach ($value as $sub_item) {
5526 - if (is_array($sub_item)) {
5527 - // For repeater/flexible content, extract text values
5528 - $sub_text = $this->mxchat_extract_text_from_acf_array($sub_item);
5529 - if (!empty($sub_text)) {
5530 - $sub_values[] = $sub_text;
5531 - }
5532 - } elseif ($sub_item instanceof WP_Post) {
5533 - // Handle WP_Post objects in arrays
5534 - $sub_values[] = $sub_item->post_title ?: '';
5535 - } else {
5536 - $sub_values[] = (string) $sub_item;
5537 - }
5538 - }
5539 - return implode(', ', array_filter($sub_values));
5540 - }
5541 -
5542 - // For other arrays, try to extract meaningful text
5543 - $text_values = array();
5544 - foreach ($value as $key => $val) {
5545 - if (is_string($val) && !empty(trim($val))) {
5546 - $text_values[] = trim($val);
5547 - } elseif ($val instanceof WP_Post) {
5548 - // Handle WP_Post objects in associative arrays
5549 - $text_values[] = $val->post_title ?: '';
5550 - } elseif (is_array($val) && isset($val['post_title'])) {
5551 - $text_values[] = $val['post_title'];
5552 - } elseif (is_array($val) && isset($val['name'])) {
5553 - $text_values[] = $val['name'];
5554 - }
5555 - }
5556 -
5557 - return implode(', ', array_filter($text_values));
5558 - }
5559 -
5560 - // Handle boolean values
5561 - if (is_bool($value)) {
5562 - return $value ? 'Yes' : 'No';
5563 - }
5564 -
5565 - // Handle numeric values
5566 - if (is_numeric($value)) {
5567 - return (string) $value;
5568 - }
5569 -
5570 - // Handle string values
5571 - if (is_string($value)) {
5572 - return trim($value);
5573 - }
5574 -
5575 - // For anything else that we can't handle, return empty string
5576 - // This prevents the "Object could not be converted to string" error
5577 - return '';
5578 -}
5579 -
5580 -/**
5581 - * Extract text from complex ACF array structures
5582 - */
5583 -private function mxchat_extract_text_from_acf_array($array) {
5584 - if (!is_array($array)) {
5585 - return '';
5586 - }
5587 -
5588 - $text_parts = array();
5589 -
5590 - foreach ($array as $key => $value) {
5591 - if (is_string($value) && !empty(trim($value))) {
5592 - // Skip keys that are likely to be IDs or technical values
5593 - if (!is_numeric($value) || strlen($value) > 10) {
5594 - $text_parts[] = trim($value);
5595 - }
5596 - } elseif ($value instanceof WP_Post) {
5597 - // Handle WP_Post objects
5598 - $text_parts[] = $value->post_title ?: '';
5599 - } elseif (is_array($value)) {
5600 - if (isset($value['post_title'])) {
5601 - $text_parts[] = $value['post_title'];
5602 - } elseif (isset($value['name'])) {
5603 - $text_parts[] = $value['name'];
5604 - } elseif (isset($value['label'])) {
5605 - $text_parts[] = $value['label'];
5606 - }
5607 - } elseif (is_object($value)) {
5608 - // Handle other objects safely
5609 - if (isset($value->post_title)) {
5610 - $text_parts[] = $value->post_title;
5611 - } elseif (isset($value->name)) {
5612 - $text_parts[] = $value->name;
5613 - } elseif (isset($value->display_name)) {
5614 - $text_parts[] = $value->display_name;
5615 - }
5616 - }
5617 - }
5618 -
5619 - return implode(', ', array_filter($text_parts));
5620 -}
5621 -
5622 -/**
5623 - * Walk an ACF field value tree and collect attachment IDs for any value that
5624 - * resolves to a PDF in the WordPress media library. Handles the three shapes
5625 - * ACF returns for File/Image/URL fields (array with ID+url, integer attachment ID,
5626 - * plain URL string), and recurses through repeater/group/flexible content.
5627 - *
5628 - * @param mixed $value The ACF field value (any depth)
5629 - * @param array $out Accumulator (passed by reference) for attachment IDs
5630 - * @param int $depth Recursion guard
5631 - */
5632 -private function mxchat_collect_pdf_attachment_ids_from_acf_value($value, &$out, $depth = 0) {
5633 - if ($depth > 6) {
5634 - return; // prevent runaway recursion on circular/very-deep structures
5635 - }
5636 -
5637 - if (empty($value)) {
5638 - return;
5639 - }
5640 -
5641 - // Array shapes: ACF File/Image return value=array; repeaters/groups are arrays of arrays
5642 - if (is_array($value)) {
5643 - // Direct File/Image-style array (has 'url' and usually 'ID' + 'mime_type')
5644 - $looks_like_attachment = isset($value['url']) || isset($value['ID']) || isset($value['id']);
5645 - if ($looks_like_attachment) {
5646 - $att_id = 0;
5647 - if (!empty($value['ID']) && is_numeric($value['ID'])) {
5648 - $att_id = (int) $value['ID'];
5649 - } elseif (!empty($value['id']) && is_numeric($value['id'])) {
5650 - $att_id = (int) $value['id'];
5651 - } elseif (!empty($value['url']) && is_string($value['url'])) {
5652 - $att_id = (int) attachment_url_to_postid($value['url']);
5653 - }
5654 -
5655 - $is_pdf = false;
5656 - if (!empty($value['mime_type']) && $value['mime_type'] === 'application/pdf') {
5657 - $is_pdf = true;
5658 - } elseif (!empty($value['subtype']) && strtolower((string) $value['subtype']) === 'pdf') {
5659 - $is_pdf = true;
5660 - } elseif (!empty($value['url']) && is_string($value['url']) && $this->mxchat_url_looks_like_pdf($value['url'])) {
5661 - $is_pdf = true;
5662 - } elseif ($att_id && get_post_mime_type($att_id) === 'application/pdf') {
5663 - $is_pdf = true;
5664 - }
5665 -
5666 - if ($is_pdf && $att_id && get_post_mime_type($att_id) === 'application/pdf') {
5667 - $out[] = $att_id;
5668 - }
5669 - // An array node that represents one attachment doesn't contain other
5670 - // attachments inside it — done with this branch.
5671 - return;
5672 - }
5673 -
5674 - // Recurse: repeater rows, flexible-content layouts, groups, etc.
5675 - foreach ($value as $sub) {
5676 - $this->mxchat_collect_pdf_attachment_ids_from_acf_value($sub, $out, $depth + 1);
5677 - }
5678 - return;
5679 - }
5680 -
5681 - // Plain numeric attachment ID (ACF File field set to "Return: ID")
5682 - if (is_numeric($value)) {
5683 - $att_id = (int) $value;
5684 - if ($att_id > 0 && get_post_mime_type($att_id) === 'application/pdf') {
5685 - $out[] = $att_id;
5686 - }
5687 - return;
5688 - }
5689 -
5690 - // Plain string — URL pointing at a PDF (ACF File field set to "Return: URL", or a custom URL/text field)
5691 - if (is_string($value)) {
5692 - $trimmed = trim($value);
5693 - if ($trimmed !== '' && $this->mxchat_url_looks_like_pdf($trimmed)) {
5694 - $att_id = (int) attachment_url_to_postid($trimmed);
5695 - if ($att_id > 0 && get_post_mime_type($att_id) === 'application/pdf') {
5696 - $out[] = $att_id;
5697 - }
5698 - }
5699 - return;
5700 - }
5701 -}
5702 -
5703 -/**
5704 - * Heuristic: does this URL/string look like a PDF reference?
5705 - * Tolerates query strings and fragments (#page=2).
5706 - */
5707 -private function mxchat_url_looks_like_pdf($url) {
5708 - if (!is_string($url) || $url === '') {
5709 - return false;
5710 - }
5711 - // Strip query + fragment before checking extension
5712 - $path = preg_replace('/[?#].*$/', '', $url);
5713 - return (bool) preg_match('/\.pdf$/i', $path);
5714 -}
5715 -
5716 -/**
5717 - * Extract text from a PDF attachment by ID using the bundled Smalot parser.
5718 - * Reads the file directly from disk via get_attached_file (no HTTP fetch).
5719 - * Result is cached on the attachment as post_meta keyed by file mtime so we
5720 - * only parse the same PDF once unless the file changes on disk.
5721 - *
5722 - * @param int $attachment_id
5723 - * @return string Extracted plain text, or '' on failure.
5724 - */
5725 -private function mxchat_extract_pdf_text_by_attachment_id($attachment_id) {
5726 - $attachment_id = (int) $attachment_id;
5727 - if ($attachment_id <= 0) {
5728 - return '';
5729 - }
5730 - if (get_post_mime_type($attachment_id) !== 'application/pdf') {
5731 - return '';
5732 - }
5733 -
5734 - $pdf_path = get_attached_file($attachment_id);
5735 - if (empty($pdf_path) || !file_exists($pdf_path) || !is_readable($pdf_path)) {
5736 - return '';
5737 - }
5738 -
5739 - // Raw-file size cap. Parsing very large PDFs can OOM the request; skip with a log entry
5740 - // and let the rest of the ACF content land in the KB. Filterable for users who need it bigger.
5741 - $default_max_bytes = 25 * 1024 * 1024;
5742 - $max_bytes = (int) apply_filters('mxchat_acf_pdf_max_bytes', $default_max_bytes, $attachment_id, $pdf_path);
5743 - if ($max_bytes > 0) {
5744 - $file_size = @filesize($pdf_path);
5745 - if ($file_size !== false && $file_size > $max_bytes) {
5746 - error_log(sprintf(
5747 - '[mxchat] ACF PDF skipped (over size cap): attachment %d "%s" %d bytes > cap %d',
5748 - $attachment_id,
5749 - basename($pdf_path),
5750 - $file_size,
5751 - $max_bytes
5752 - ));
5753 - return '';
5754 - }
5755 - }
5756 -
5757 - $mtime = @filemtime($pdf_path);
5758 - $cache_meta_key = '_mxchat_acf_pdf_text_v1';
5759 - $cached = get_post_meta($attachment_id, $cache_meta_key, true);
5760 - if (is_array($cached) && isset($cached['mtime'], $cached['text']) && (int) $cached['mtime'] === (int) $mtime) {
5761 - return (string) $cached['text'];
5762 - }
5763 -
5764 - $text = '';
5765 - try {
5766 - if (function_exists('mxchat_load_pdf_parser')) {
5767 - mxchat_load_pdf_parser();
5768 - }
5769 - if (!class_exists('\\Smalot\\PdfParser\\Parser')) {
5770 - return '';
5771 - }
5772 - $parser = new \Smalot\PdfParser\Parser();
5773 - $pdf = $parser->parseFile($pdf_path);
5774 - $pages = $pdf->getPages();
5775 - $page_texts = array();
5776 - foreach ($pages as $page) {
5777 - $page_text = '';
5778 - try {
5779 - $page_text = $page->getText();
5780 - } catch (\Exception $e) {
5781 - $page_text = '';
5782 - }
5783 - if (!empty($page_text)) {
5784 - $page_texts[] = $page_text;
5785 - }
5786 - }
5787 - $text = trim(implode("\n\n", $page_texts));
5788 - } catch (\Exception $e) {
5789 - error_log('[mxchat] ACF PDF extraction failed for attachment ' . $attachment_id . ': ' . $e->getMessage());
5790 - return '';
5791 - } catch (\Throwable $e) {
5792 - error_log('[mxchat] ACF PDF extraction error for attachment ' . $attachment_id . ': ' . $e->getMessage());
5793 - return '';
5794 - }
5795 -
5796 - // Cap per-PDF text to avoid blowing up the embedding payload on enormous PDFs.
5797 - // The chunker downstream will still split this into multiple vectors.
5798 - $max_len = (int) apply_filters('mxchat_acf_pdf_text_max_length', 50000);
5799 - if ($max_len > 0 && strlen($text) > $max_len) {
5800 - $text = substr($text, 0, $max_len);
5801 - }
5802 -
5803 - update_post_meta($attachment_id, $cache_meta_key, array(
5804 - 'mtime' => (int) $mtime,
5805 - 'text' => $text,
5806 - ));
5807 -
5808 - return $text;
5809 -}
5810 -
5811 -/**
5812 - * Handle ACF save - fires after ACF fields are saved
5813 - * This ensures ACF field data is available when syncing to knowledge base
5814 - */
5815 -public function mxchat_handle_acf_save($post_id) {
5816 - // Skip if not a valid post
5817 - if (!$post_id || $post_id === 'options') {
5818 - return;
5819 - }
5820 -
5821 - // Skip autosaves and revisions
5822 - if (defined('DOING_AUTOSAVE') && DOING_AUTOSAVE || wp_is_post_revision($post_id)) {
5823 - return;
5824 - }
5825 -
5826 - $post = get_post($post_id);
5827 - if (!$post) {
5828 - return;
5829 - }
5830 -
5831 - $post_type = $post->post_type;
5832 -
5833 - // Check if sync is enabled for this post type
5834 - $should_sync = false;
5835 -
5836 - if ($post_type === 'post' && get_option('mxchat_auto_sync_posts') === '1') {
5837 - $should_sync = true;
5838 - } else if ($post_type === 'page' && get_option('mxchat_auto_sync_pages') === '1') {
5839 - $should_sync = true;
5840 - } else if ($post_type === 'product' && class_exists('WooCommerce')) {
5841 - // WooCommerce products - check if WooCommerce integration is enabled
5842 - $options = get_option('mxchat_options', array());
5843 - if (isset($options['enable_woocommerce_integration']) &&
5844 - ($options['enable_woocommerce_integration'] === '1' || $options['enable_woocommerce_integration'] === 'on')) {
5845 - $should_sync = true;
5846 - }
5847 - } else {
5848 - // Check custom post types
5849 - $option_name = 'mxchat_auto_sync_' . $post_type;
5850 - if (get_option($option_name) === '1') {
5851 - $should_sync = true;
5852 - }
5853 - }
5854 -
5855 - if (!$should_sync) {
5856 - return;
5857 - }
5858 -
5859 - // Only process published posts
5860 - if ($post->post_status !== 'publish') {
5861 - return;
5862 - }
5863 -
5864 - // Check if this post has any ACF fields - if not, no need to re-sync
5865 - $acf_fields = $this->mxchat_get_acf_fields_for_post($post_id);
5866 - if (empty($acf_fields)) {
5867 - return;
5868 - }
5869 -
5870 - // Use a transient to prevent duplicate processing (post_updated may have already run)
5871 - $transient_key = 'mxchat_acf_synced_' . $post_id;
5872 - if (get_transient($transient_key)) {
5873 - return;
5874 - }
5875 - set_transient($transient_key, true, 60); // Prevent re-processing for 60 seconds
5876 -
5877 - // Re-run the sync with ACF data now available
5878 - // We pass $update=true since this is effectively an update with ACF data
5879 - $this->mxchat_handle_post_update($post_id, $post, true);
5880 -}
5881 -
5882 -public function mxchat_handle_post_update($post_id, $post, $update) {
5883 - // Basic validation checks
5884 - if (defined('DOING_AUTOSAVE') && DOING_AUTOSAVE || wp_is_post_revision($post_id)) {
5885 - return;
5886 - }
5887 -
5888 - $post_type = $post->post_type;
5889 -
5890 - // Check if sync is enabled for this post type
5891 - $should_sync = false;
5892 -
5893 - // Check built-in post types first
5894 - if ($post_type === 'post' && get_option('mxchat_auto_sync_posts') === '1') {
5895 - $should_sync = true;
5896 - } else if ($post_type === 'page' && get_option('mxchat_auto_sync_pages') === '1') {
5897 - $should_sync = true;
5898 - } else {
5899 - // Check custom post types
5900 - $option_name = 'mxchat_auto_sync_' . $post_type;
5901 - if (get_option($option_name) === '1') {
5902 - $should_sync = true;
5903 - }
5904 - }
5905 -
5906 - if (!$should_sync) {
5907 - return;
5908 - }
5909 -
5910 - // Check if we have stored the previous status and URL in our transients
5911 - $previous_status_key = 'mxchat_prev_status_' . $post_id;
5912 - $previous_status = get_transient($previous_status_key);
5913 -
5914 - $previous_url_key = 'mxchat_prev_url_' . $post_id;
5915 - $previous_url = get_transient($previous_url_key);
5916 -
5917 - // If the post was previously published but is now not published, remove from knowledge base
5918 - if ($previous_status === 'publish' && $post->post_status !== 'publish') {
5919 - // Use the stored URL from when it was published, or fall back to current permalink
5920 - $source_url = $previous_url ?: get_permalink($post_id);
5921 -
5922 - // mxchat_handle_status_transition already deleted for this post earlier in this
5923 - // request (it fires first inside wp_insert_post); skip the redundant round-trip.
5924 - if ($source_url && empty($this->transition_deleted_posts[$post_id])) {
5925 - // Chunk-aware deletion (routes to Pinecone or WP DB and removes base + all chunks)
5926 - MxChat_Utils::delete_chunks_for_url($source_url, 'default');
5927 - }
5928 -
5929 - // Clean up the transients and exit early
5930 - delete_transient($previous_status_key);
5931 - delete_transient($previous_url_key);
5932 - return;
5933 - }
5934 -
5935 - // Slug/permalink rename while still published: delete the old vectors before upserting new ones.
5936 - // Without this, md5(old_url) vectors (base + chunks) would be orphaned under the stale URL.
5937 - if ($post->post_status === 'publish' && !empty($previous_url)) {
5938 - $current_url = get_permalink($post_id);
5939 - if ($current_url && $current_url !== $previous_url) {
5940 - MxChat_Utils::delete_chunks_for_url($previous_url, 'default');
5941 - }
5942 - }
5943 -
5944 - // Store the current status for next time (if this is an update)
5945 - if ($update) {
5946 - set_transient($previous_status_key, $post->post_status, DAY_IN_SECONDS);
5947 -
5948 - // If the post is currently published, also store its URL
5949 - if ($post->post_status === 'publish') {
5950 - $current_url = get_permalink($post_id);
5951 - set_transient($previous_url_key, $current_url, DAY_IN_SECONDS);
5952 - }
5953 - }
5954 -
5955 - // Only process currently published content for adding/updating
5956 - if ($post->post_status === 'publish') {
5957 - // Get the source URL
5958 - $source_url = get_permalink($post_id);
5959 -
5960 - // Get content with proper formatting (matching ajax_mxchat_process_selected_content)
5961 - $title = get_the_title($post_id);
5962 - $content = get_post_field('post_content', $post_id);
5963 - $excerpt = get_post_field('post_excerpt', $post_id);
5964 -
5965 - // Remove shortcode tags but preserve content inside them
5966 - $content = $this->strip_shortcode_tags_preserve_content($content);
5967 - $excerpt = $this->strip_shortcode_tags_preserve_content($excerpt);
5968 -
5969 - // Strip tags but preserve structure (don't use 'the_content' filter as it may re-add shortcodes)
5970 - $content = wp_strip_all_tags($content);
5971 -
5972 - // Combine title, short description (if exists), and content
5973 - $final_content = $title . "\n\n";
5974 -
5975 - // Add short description if it exists (WooCommerce products use post_excerpt for short description)
5976 - if (!empty($excerpt)) {
5977 - $final_content .= "Short Description: " . wp_strip_all_tags($excerpt) . "\n\n";
5978 - }
5979 -
5980 - $final_content .= $content;
5981 -
5982 - // For WooCommerce products, include pricing and product details
5983 - if ($post_type === 'product' && class_exists('WooCommerce')) {
5984 - $product = wc_get_product($post_id);
5985 -
5986 - if ($product) {
5987 - // Get pricing information
5988 - $regular_price = $product->get_regular_price();
5989 - $sale_price = $product->get_sale_price();
5990 - $price = $product->get_price();
5991 - $sku = $product->get_sku();
5992 -
5993 - // Get currency symbol
5994 - $currency_symbol = get_woocommerce_currency_symbol();
5995 -
5996 - // Add pricing information
5997 - $final_content .= "\n";
5998 - if (!empty($regular_price)) {
5999 - $final_content .= "Price: " . $currency_symbol . $regular_price . "\n";
6000 - } elseif (!empty($price)) {
6001 - $final_content .= "Price: " . $currency_symbol . $price . "\n";
6002 - }
6003 -
6004 - if (!empty($sale_price) && $sale_price !== $regular_price) {
6005 - $final_content .= "Sale Price: " . $currency_symbol . $sale_price . "\n";
6006 - }
6007 -
6008 - // Handle variable products - show price range
6009 - if ($product->is_type('variable')) {
6010 - $min_price = $product->get_variation_price('min');
6011 - $max_price = $product->get_variation_price('max');
6012 - if ($min_price !== $max_price) {
6013 - $final_content .= "Price Range: " . $currency_symbol . $min_price . " - " . $currency_symbol . $max_price . "\n";
6014 - }
6015 - }
6016 -
6017 - if (!empty($sku)) {
6018 - $final_content .= "SKU: " . $sku . "\n";
6019 - }
6020 -
6021 - // Get product categories
6022 - $categories = wp_get_post_terms($post_id, 'product_cat', array('fields' => 'names'));
6023 - if (!empty($categories) && !is_wp_error($categories)) {
6024 - $final_content .= "Categories: " . implode(', ', $categories) . "\n";
6025 - }
6026 - }
6027 - }
6028 -
6029 - // For custom post types like job_listing, include additional fields
6030 - if ($post_type === 'job_listing') {
6031 - // Add job-specific meta if available
6032 - $job_location = get_post_meta($post_id, '_job_location', true);
6033 - if (!empty($job_location)) {
6034 - $final_content .= "\n\nLocation: " . $job_location;
6035 - }
6036 -
6037 - // Get job type terms
6038 - $job_types = get_the_terms($post_id, 'job_listing_type');
6039 - if (!empty($job_types) && !is_wp_error($job_types)) {
6040 - $types = array();
6041 - foreach ($job_types as $type) {
6042 - $types[] = $type->name;
6043 - }
6044 - $final_content .= "\n\nJob Type: " . implode(', ', $types);
6045 - }
6046 -
6047 - // Get company name if available
6048 - $company_name = get_post_meta($post_id, '_company_name', true);
6049 - if (!empty($company_name)) {
6050 - $final_content .= "\n\nCompany: " . $company_name;
6051 - }
6052 - }
6053 -
6054 - // ADD ACF FIELDS SUPPORT (matches ajax_mxchat_process_selected_content behavior)
6055 - $acf_fields = $this->mxchat_get_acf_fields_for_post($post_id);
6056 - if (!empty($acf_fields)) {
6057 - $acf_content_parts = array();
6058 - $pdf_attachment_ids = array();
6059 -
6060 - foreach ($acf_fields as $field_name => $field_value) {
6061 - $formatted_value = $this->mxchat_format_acf_field_value($field_value, $field_name, $post_id);
6062 - if (!empty($formatted_value)) {
6063 - // Convert field name to readable label
6064 - $field_label = ucwords(str_replace(['_', '-'], ' ', $field_name));
6065 - $acf_content_parts[] = $field_label . ": " . $formatted_value;
6066 - }
6067 -
6068 - $this->mxchat_collect_pdf_attachment_ids_from_acf_value($field_value, $pdf_attachment_ids);
6069 - }
6070 -
6071 - if (!empty($acf_content_parts)) {
6072 - $final_content .= "\n\n" . implode("\n", $acf_content_parts);
6073 - }
6074 -
6075 - // Gate the auto-sync PDF-extraction loop behind an opt-in option.
6076 - // Mirrors the per-batch checkbox the manual content selector has; the
6077 - // 25 MB size cap lives in the shared extractor so it applies in both
6078 - // paths regardless. Default OFF — re-parsing every ACF PDF on every
6079 - // editor save is expensive and most sites don't want it.
6080 - $autosync_extract_acf_pdfs = get_option('mxchat_auto_sync_acf_pdfs', '0') === '1';
6081 - if ($autosync_extract_acf_pdfs && !empty($pdf_attachment_ids)) {
6082 - $pdf_attachment_ids = array_unique(array_filter(array_map('intval', $pdf_attachment_ids)));
6083 - $pdf_sections = array();
6084 - foreach ($pdf_attachment_ids as $att_id) {
6085 - $pdf_text = $this->mxchat_extract_pdf_text_by_attachment_id($att_id);
6086 - if (!empty($pdf_text)) {
6087 - $pdf_title = get_the_title($att_id);
6088 - $pdf_url = wp_get_attachment_url($att_id);
6089 - $header = 'PDF Attachment';
6090 - if (!empty($pdf_title)) {
6091 - $header .= ': ' . $pdf_title;
6092 - }
6093 - if (!empty($pdf_url)) {
6094 - $header .= ' (' . $pdf_url . ')';
6095 - }
6096 - $pdf_sections[] = $header . "\n" . $pdf_text;
6097 - }
6098 - }
6099 - if (!empty($pdf_sections)) {
6100 - $final_content .= "\n\nAttached PDFs:\n" . implode("\n\n", $pdf_sections);
6101 - }
6102 - }
6103 - }
6104 -
6105 - // ADD CUSTOM POST META SUPPORT (whitelisted non-ACF meta fields)
6106 - $custom_meta = $this->mxchat_get_whitelisted_post_meta($post_id);
6107 - if (!empty($custom_meta)) {
6108 - $meta_content_parts = array();
6109 -
6110 - foreach ($custom_meta as $meta_key => $meta_value) {
6111 - // Convert meta key to readable label
6112 - $meta_label = ucwords(str_replace(array('_', '-'), ' ', $meta_key));
6113 - $meta_content_parts[] = $meta_label . ": " . $meta_value;
6114 - }
6115 -
6116 - if (!empty($meta_content_parts)) {
6117 - $final_content .= "\n\n" . implode("\n", $meta_content_parts);
6118 - }
6119 - }
6120 -
6121 - // Get API key with proper model detection
6122 - $options = get_option('mxchat_options');
6123 - $selected_model = $options['embedding_model'] ?? 'text-embedding-ada-002';
6124 -
6125 - if (strpos($selected_model, 'voyage') === 0) {
6126 - $api_key = $options['voyage_api_key'] ?? '';
6127 - } elseif (strpos($selected_model, 'gemini-embedding') === 0) {
6128 - $api_key = $options['gemini_api_key'] ?? '';
6129 - } else {
6130 - $api_key = $options['api_key'] ?? '';
6131 - }
6132 -
6133 - if (empty($api_key)) {
6134 - return;
6135 - }
6136 -
6137 - // Use the centralized utility function for storage
6138 - $result = MxChat_Utils::submit_content_to_db(
6139 - $final_content,
6140 - $source_url,
6141 - $api_key,
6142 - md5($source_url) // Vector ID for Pinecone
6143 - );
6144 -
6145 - // After successful storage, apply role restriction based on tags
6146 - if (!is_wp_error($result)) {
6147 - $this->apply_role_restriction_to_post($post_id, $source_url);
6148 - }
6149 - }
6150 -
6151 - // Clean up the stored previous status if not used above
6152 - if ($previous_status !== 'publish' || $post->post_status === 'publish') {
6153 - delete_transient($previous_status_key);
6154 - delete_transient($previous_url_key);
6155 - }
6156 -}
6157 -
6158 -/**
6159 - * Store the post status and URL before update to detect status transitions
6160 - * This runs before the post is actually updated in the database
6161 - */
6162 -public function mxchat_store_pre_update_status($post_id, $data) {
6163 - // Get the current post from database (before update)
6164 - $current_post = get_post($post_id);
6165 -
6166 - if ($current_post) {
6167 - // Store the current status temporarily
6168 - $status_key = 'mxchat_prev_status_' . $post_id;
6169 - set_transient($status_key, $current_post->post_status, HOUR_IN_SECONDS);
6170 -
6171 - // If the post is currently published, also store its URL
6172 - if ($current_post->post_status === 'publish') {
6173 - $url_key = 'mxchat_prev_url_' . $post_id;
6174 - $current_url = get_permalink($post_id);
6175 - set_transient($url_key, $current_url, HOUR_IN_SECONDS);
6176 - }
6177 - }
6178 -}
6179 -
6180 -/**
6181 - * Whether auto-sync is enabled for a post type (mirrors the checks used by the
6182 - * update/delete handlers; kept as one helper so new call sites cannot drift).
6183 - */
6184 -private function mxchat_is_auto_sync_enabled($post_type) {
6185 - if ($post_type === 'post') {
6186 - return get_option('mxchat_auto_sync_posts') === '1';
6187 - }
6188 - if ($post_type === 'page') {
6189 - return get_option('mxchat_auto_sync_pages') === '1';
6190 - }
6191 - return get_option('mxchat_auto_sync_' . $post_type) === '1';
6192 -}
6193 -
6194 -/**
6195 - * Remove a post's vectors the moment it leaves 'publish', using the authoritative
6196 - * old status core passes to transition_post_status — no transient involved (plan 816fb1).
6197 - *
6198 - * Covers status changes that never route through wp_update_post (scheduled-expiry
6199 - * plugins and others that flip post_status directly and call wp_transition_post_status),
6200 - * where neither pre_post_update nor post_updated fires and the old detection missed.
6201 - */
6202 -public function mxchat_handle_status_transition($new_status, $old_status, $post) {
6203 - if (!($post instanceof WP_Post) || wp_is_post_revision($post->ID)) {
6204 - return;
6205 - }
6206 - // Only the publish -> not-publish edge matters here.
6207 - if ($old_status !== 'publish' || $new_status === 'publish') {
6208 - return;
6209 - }
6210 - // Trash is handled by mxchat_handle_post_delete (wp_trash_post) with pre-trash URL
6211 - // resolution; skip to avoid a second network round-trip per trash.
6212 - if ($new_status === 'trash') {
6213 - return;
6214 - }
6215 - if (!$this->mxchat_is_auto_sync_enabled($post->post_type)) {
6216 - return;
6217 - }
6218 -
6219 - $urls = array();
6220 -
6221 - // The DB may already hold the new status when this fires, so get_permalink() on the
6222 - // live post could build a draft-style URL whose md5 misses the stored vector IDs.
6223 - // Reconstruct the published permalink from a clone instead.
6224 - $published_clone = clone $post;
6225 - $published_clone->post_status = 'publish';
6226 - $published_url = get_permalink($published_clone);
6227 - if ($published_url) {
6228 - $urls[] = $published_url;
6229 - }
6230 -
6231 - // Honour the pre-update capture when present (covers a slug change in the same save).
6232 - $previous_url = get_transient('mxchat_prev_url_' . $post->ID);
6233 - if (!empty($previous_url)) {
6234 - $urls[] = $previous_url;
6235 - }
6236 -
6237 - foreach (array_unique($urls) as $url) {
6238 - MxChat_Utils::delete_chunks_for_url($url, 'default');
6239 - }
6240 -
6241 - if (!empty($urls)) {
6242 - $this->transition_deleted_posts[$post->ID] = true;
6243 - }
6244 -}
6245 -
6246 -/**
6247 - * WP-CLI: remove knowledge-base entries left behind by posts that were unpublished,
6248 - * trashed, or made private before the transition_post_status handler existed.
6249 - *
6250 - * Walks every auto-synced post type's non-published posts, reconstructs each one's
6251 - * published-era permalink, and deletes its vectors (routes to Pinecone or the WP table).
6252 - * Deletion is idempotent, so never-indexed posts are a cheap no-op.
6253 - *
6254 - * ## OPTIONS
6255 - *
6256 - * [--dry-run]
6257 - * : Report what would be removed without deleting anything.
6258 - *
6259 - * ## EXAMPLES
6260 - *
6261 - * wp mxchat prune-unpublished --dry-run
6262 - * wp mxchat prune-unpublished
6263 - */
6264 -public function cli_prune_unpublished($args, $assoc_args) {
6265 - global $wpdb;
6266 - $dry_run = !empty($assoc_args['dry-run']);
6267 - $table = $wpdb->prefix . 'mxchat_system_prompt_content';
6268 -
6269 - $candidate_types = array_merge(array('post', 'page'), array_values(get_post_types(array('_builtin' => false), 'names')));
6270 - $synced_types = array();
6271 - foreach ($candidate_types as $type) {
6272 - if ($this->mxchat_is_auto_sync_enabled($type)) {
6273 - $synced_types[] = $type;
6274 - }
6275 - }
6276 - if (empty($synced_types)) {
6277 - WP_CLI::success('No post types have auto-sync enabled; nothing to prune.');
6278 - return;
6279 - }
6280 -
6281 - $scanned = 0;
6282 - $pruned = 0;
6283 - $paged = 1;
6284 - do {
6285 - $query = new WP_Query(array(
6286 - 'post_type' => $synced_types,
6287 - 'post_status' => array('draft', 'pending', 'private', 'future', 'trash'),
6288 - 'posts_per_page' => 100,
6289 - 'paged' => $paged,
6290 - 'fields' => 'ids',
6291 - ));
6292 - foreach ($query->posts as $post_id) {
6293 - $post = get_post($post_id);
6294 - if (!$post) {
6295 - continue;
6296 - }
6297 - $scanned++;
6298 -
6299 - // Rebuild the permalink the post had while published: publish-status clone,
6300 - // with wp_trash_post's __trashed slug suffix stripped for trashed posts.
6301 - $clone = clone $post;
6302 - $clone->post_status = 'publish';
6303 - if (substr($clone->post_name, -9) === '__trashed') {
6304 - $clone->post_name = substr($clone->post_name, 0, -9);
6305 - }
6306 - $url = get_permalink($clone);
6307 - if (!$url) {
6308 - continue;
6309 - }
6310 -
6311 - // Local-table row count is exact in WordPress-DB mode; in Pinecone mode it
6312 - // reads 0 but the delete below still routes to Pinecone and is idempotent.
6313 - $local_rows = (int) $wpdb->get_var($wpdb->prepare(
6314 - "SELECT COUNT(*) FROM {$table} WHERE source_url = %s", $url
6315 - ));
6316 -
6317 - if ($dry_run) {
6318 - if ($local_rows > 0) {
6319 - WP_CLI::log(sprintf('Would remove %d row(s): %s (post %d, %s)', $local_rows, $url, $post_id, $post->post_status));
6320 - $pruned += $local_rows;
6321 - }
6322 - continue;
6323 - }
6324 -
6325 - MxChat_Utils::delete_chunks_for_url($url, 'default');
6326 - if ($local_rows > 0) {
6327 - WP_CLI::log(sprintf('Removed %d row(s): %s (post %d, %s)', $local_rows, $url, $post_id, $post->post_status));
6328 - $pruned += $local_rows;
6329 - }
6330 - }
6331 - $more = $paged < $query->max_num_pages;
6332 - $paged++;
6333 - } while ($more);
6334 -
6335 - WP_CLI::success(sprintf(
6336 - '%s %d local knowledge row(s) across %d non-published post(s) scanned.%s',
6337 - $dry_run ? 'Would remove' : 'Removed',
6338 - $pruned,
6339 - $scanned,
6340 - ' (Pinecone-mode deletions are not counted locally.)'
6341 - ));
6342 -}
6343 -
6344 -public function mxchat_handle_post_delete($post_id) {
6345 - // Get post data before it's deleted
6346 - $post = get_post($post_id);
6347 -
6348 - // Basic validation
6349 - if (!$post || wp_is_post_revision($post_id)) {
6350 - return;
6351 - }
6352 -
6353 - $post_type = $post->post_type;
6354 -
6355 - // Check if sync is enabled for this post type
6356 - $should_sync = false;
6357 -
6358 - // Check built-in post types first
6359 - if ($post_type === 'post' && get_option('mxchat_auto_sync_posts') === '1') {
6360 - $should_sync = true;
6361 - } else if ($post_type === 'page' && get_option('mxchat_auto_sync_pages') === '1') {
6362 - $should_sync = true;
6363 - } else {
6364 - // Check custom post types
6365 - $option_name = 'mxchat_auto_sync_' . $post_type;
6366 - if (get_option($option_name) === '1') {
6367 - $should_sync = true;
6368 - }
6369 - }
6370 -
6371 - if (!$should_sync) {
6372 - return;
6373 - }
6374 -
6375 - // Resolve the pre-trash URL. wp_trash_post renames the slug with "__trashed" before firing
6376 - // this hook, so get_permalink() here would return the trashed URL and md5() would miss the
6377 - // real vector IDs stored under the original URL.
6378 - $source_url = $this->mxchat_resolve_pre_trash_url($post_id);
6379 - if (!$source_url) {
6380 - //error_log('MXChat: Failed to resolve source URL for post ' . $post_id);
6381 - return;
6382 - }
6383 -
6384 - // Use chunk-aware deletion (handles both chunked and non-chunked content)
6385 - $delete_result = MxChat_Utils::delete_chunks_for_url($source_url, 'default');
6386 -
6387 - if (is_wp_error($delete_result)) {
6388 - //error_log('MXChat: Chunk-aware deletion failed for URL: ' . $source_url . ' - ' . $delete_result->get_error_message());
6389 - }
6390 -
6391 - delete_transient('mxchat_prev_url_' . $post_id);
6392 - delete_transient('mxchat_prev_status_' . $post_id);
6393 -}
6394 -
6395 -/**
6396 - * Resolve the source URL for a post being trashed/deleted.
6397 - *
6398 - * Why: wp_trash_post appends "__trashed" to the slug before the wp_trash_post action fires, so
6399 - * get_permalink() returns a URL whose md5() won't match the vector IDs stored in Pinecone or
6400 - * the source_url rows in the WP DB. Prefer the URL captured by mxchat_store_pre_update_status
6401 - * (runs on pre_post_update, before the rename); fall back to stripping the __trashed suffix.
6402 - */
6403 -private function mxchat_resolve_pre_trash_url($post_id) {
6404 - $previous_url = get_transient('mxchat_prev_url_' . $post_id);
6405 - if (!empty($previous_url)) {
6406 - return $previous_url;
6407 - }
6408 -
6409 - $current = get_permalink($post_id);
6410 - if (!$current) {
6411 - return '';
6412 - }
6413 - return preg_replace('#__trashed(/?)$#', '$1', $current);
6414 -}
6415 -
6416 -
6417 -
6418 -public function mxchat_handle_product_change($post_id, $post, $update) {
6419 - if ($post->post_type !== 'product') {
6420 - return;
6421 - }
6422 -
6423 - if ($post->post_status === 'publish') {
6424 - add_action('shutdown', function() use ($post_id) {
6425 - $product = wc_get_product($post_id);
6426 - if ($product) {
6427 - $this->mxchat_store_product_embedding($product);
6428 - }
6429 - });
6430 - }
6431 -}
6432 -
6433 -/**
6434 - * Store WooCommerce product embeddings
6435 - */
6436 -private function mxchat_store_product_embedding($product) {
6437 - if (!isset($this->options['enable_woocommerce_integration']) ||
6438 - !in_array($this->options['enable_woocommerce_integration'], ['1', 'on'])) {
6439 - return;
6440 - }
6441 -
6442 - $source_url = get_permalink($product->get_id());
6443 - $product_id = $product->get_id();
6444 -
6445 - // Build product content
6446 - $title = $product->get_name();
6447 - $description = $product->get_description();
6448 - $short_description = $product->get_short_description();
6449 - $regular_price = $product->get_regular_price();
6450 - $sale_price = $product->get_sale_price();
6451 - $price = $product->get_price();
6452 - $sku = $product->get_sku();
6453 -
6454 - // Get currency symbol
6455 - $currency_symbol = get_woocommerce_currency_symbol();
6456 -
6457 - // Format content consistently
6458 - $content = $title . "\n\n";
6459 -
6460 - if (!empty($short_description)) {
6461 - $content .= "Short Description: " . wp_strip_all_tags($short_description) . "\n\n";
6462 - }
6463 -
6464 - if (!empty($description)) {
6465 - $content .= wp_strip_all_tags($description) . "\n\n";
6466 - }
6467 -
6468 - // Add pricing information
6469 - if (!empty($regular_price)) {
6470 - $content .= "Price: " . $currency_symbol . $regular_price . "\n";
6471 - } elseif (!empty($price)) {
6472 - $content .= "Price: " . $currency_symbol . $price . "\n";
6473 - }
6474 -
6475 - if (!empty($sale_price) && $sale_price !== $regular_price) {
6476 - $content .= "Sale Price: " . $currency_symbol . $sale_price . "\n";
6477 - }
6478 -
6479 - // Handle variable products - show price range
6480 - if ($product->is_type('variable')) {
6481 - $min_price = $product->get_variation_price('min');
6482 - $max_price = $product->get_variation_price('max');
6483 - if ($min_price !== $max_price) {
6484 - $content .= "Price Range: " . $currency_symbol . $min_price . " - " . $currency_symbol . $max_price . "\n";
6485 - }
6486 - }
6487 -
6488 - if (!empty($sku)) {
6489 - $content .= "SKU: " . $sku . "\n";
6490 - }
6491 -
6492 - // Get product categories
6493 - $categories = wp_get_post_terms($product_id, 'product_cat', array('fields' => 'names'));
6494 - if (!empty($categories) && !is_wp_error($categories)) {
6495 - $content .= "Categories: " . implode(', ', $categories) . "\n";
6496 - }
6497 -
6498 - // Get Custom Product Tabs (supports "Custom Product Tabs for WooCommerce" by Code Parrots)
6499 - $custom_tabs = get_post_meta($product_id, 'yikes_woo_products_tabs', true);
6500 - if (!empty($custom_tabs) && is_array($custom_tabs)) {
6501 - foreach ($custom_tabs as $tab) {
6502 - $tab_title = isset($tab['title']) ? $tab['title'] : (isset($tab['tab_title']) ? $tab['tab_title'] : '');
6503 - $tab_content = isset($tab['content']) ? $tab['content'] : '';
6504 -
6505 - if (!empty($tab_title) && !empty($tab_content)) {
6506 - $content .= "\n" . $tab_title . ": " . wp_strip_all_tags($tab_content) . "\n";
6507 - }
6508 - }
6509 - }
6510 -
6511 - // Also check for reusable/saved tabs applied to this product
6512 - $applied_saved_tabs = get_post_meta($product_id, 'yikes_woo_reusable_products_tabs_applied', true);
6513 - if (!empty($applied_saved_tabs) && is_array($applied_saved_tabs)) {
6514 - // Get the saved tabs option
6515 - $saved_tabs = get_option('yikes_woo_reusable_products_tabs', array());
6516 - if (!empty($saved_tabs) && is_array($saved_tabs)) {
6517 - foreach ($applied_saved_tabs as $saved_tab_id) {
6518 - if (isset($saved_tabs[$saved_tab_id])) {
6519 - $tab = $saved_tabs[$saved_tab_id];
6520 - $tab_title = isset($tab['title']) ? $tab['title'] : (isset($tab['tab_title']) ? $tab['tab_title'] : '');
6521 - $tab_content = isset($tab['content']) ? $tab['content'] : '';
6522 -
6523 - if (!empty($tab_title) && !empty($tab_content)) {
6524 - $content .= "\n" . $tab_title . ": " . wp_strip_all_tags($tab_content) . "\n";
6525 - }
6526 - }
6527 - }
6528 - }
6529 - }
6530 -
6531 - // Get API key with proper model detection
6532 - $options = get_option('mxchat_options');
6533 - $selected_model = $options['embedding_model'] ?? 'text-embedding-ada-002';
6534 -
6535 - if (strpos($selected_model, 'voyage') === 0) {
6536 - $api_key = $options['voyage_api_key'] ?? '';
6537 - } elseif (strpos($selected_model, 'gemini-embedding') === 0) {
6538 - $api_key = $options['gemini_api_key'] ?? '';
6539 - } else {
6540 - $api_key = $options['api_key'] ?? '';
6541 - }
6542 -
6543 - if (empty($api_key)) {
6544 - //error_log('MxChat Auto-sync: No API key configured for embedding model');
6545 - return;
6546 - }
6547 -
6548 - // Use the centralized utility function for storage
6549 - $result = MxChat_Utils::submit_content_to_db(
6550 - $content,
6551 - $source_url,
6552 - $api_key,
6553 - md5($source_url) // Vector ID for Pinecone
6554 - );
6555 -
6556 - // After successful storage, apply role restriction based on tags
6557 - if (!is_wp_error($result)) {
6558 - $this->apply_role_restriction_to_post($product_id, $source_url);
6559 - }
6560 -
6561 - if (is_wp_error($result)) {
6562 - //error_log('MxChat WooCommerce sync failed for product ' . $product_id . ': ' . $result->get_error_message());
6563 - }
6564 -}
6565 -
6566 -public function mxchat_handle_product_delete($post_id) {
6567 - if (get_post_type($post_id) !== 'product') {
6568 - return;
6569 - }
6570 -
6571 - $source_url = $this->mxchat_resolve_pre_trash_url($post_id);
6572 - if (!$source_url) {
6573 - return;
6574 - }
6575 -
6576 - // Chunk-aware deletion (routes to Pinecone or WP DB and removes base + all chunks)
6577 - MxChat_Utils::delete_chunks_for_url($source_url, 'default');
6578 -
6579 - delete_transient('mxchat_prev_url_' . $post_id);
6580 - delete_transient('mxchat_prev_status_' . $post_id);
6581 -}
6582 -
6583 -/**
6584 - * Handle individual Pinecone content deletion
6585 - */
6586 -public function mxchat_handle_pinecone_prompt_delete() {
6587 - // Check permissions
6588 - if (!current_user_can('manage_options')) {
6589 - wp_die(esc_html__('You do not have sufficient permissions.', 'mxchat'));
6590 - }
6591 -
6592 - // Verify nonce
6593 - if (!isset($_GET['_wpnonce']) || !wp_verify_nonce($_GET['_wpnonce'], 'mxchat_delete_pinecone_prompt_nonce')) {
6594 - wp_die(esc_html__('Security check failed.', 'mxchat'));
6595 - }
6596 -
6597 - $vector_id = isset($_GET['vector_id']) ? sanitize_text_field($_GET['vector_id']) : '';
6598 -
6599 - if (empty($vector_id)) {
6600 - set_transient('mxchat_admin_notice_error',
6601 - esc_html__('Invalid vector ID.', 'mxchat'),
6602 - 30
6603 - );
6604 - wp_safe_redirect(admin_url('admin.php?page=mxchat-prompts'));
6605 - exit;
6606 - }
6607 -
6608 - // Get Pinecone settings
6609 - $pinecone_options = get_option('mxchat_pinecone_addon_options', array());
6610 - $use_pinecone = ($pinecone_options['mxchat_use_pinecone'] ?? '0') === '1';
6611 -
6612 - if (!$use_pinecone || empty($pinecone_options['mxchat_pinecone_api_key'])) {
6613 - set_transient('mxchat_admin_notice_error',
6614 - esc_html__('Pinecone is not properly configured.', 'mxchat'),
6615 - 30
6616 - );
6617 - wp_safe_redirect(admin_url('admin.php?page=mxchat-prompts'));
6618 - exit;
6619 - }
6620 -
6621 - // Delete from Pinecone
6622 - $pinecone_manager = MxChat_Pinecone_Manager::get_instance();
6623 - $result = $pinecone_manager->mxchat_delete_from_pinecone_by_vector_id(
6624 - $vector_id,
6625 - $pinecone_options['mxchat_pinecone_api_key'],
6626 - $pinecone_options['mxchat_pinecone_host']
6627 - );
6628 -
6629 - if ($result['success']) {
6630 - // No cache clearing needed since we removed caching
6631 - set_transient('mxchat_admin_notice_success',
6632 - esc_html__('Entry deleted successfully from Pinecone.', 'mxchat'),
6633 - 30
6634 - );
6635 - } else {
6636 - set_transient('mxchat_admin_notice_error',
6637 - esc_html__('Failed to delete entry: ', 'mxchat') . $result['message'],
6638 - 30
6639 - );
6640 - }
6641 -
6642 - wp_safe_redirect(admin_url('admin.php?page=mxchat-prompts'));
6643 - exit;
6644 -}
6645 -/**
6646 - * Handle individual Pinecone content deletion via AJAX
6647 - */
6648 -public function ajax_mxchat_delete_pinecone_prompt() {
6649 - // Verify nonce and permissions
6650 - if (!check_ajax_referer('mxchat_delete_pinecone_prompt_nonce', 'nonce', false)) {
6651 - wp_send_json_error('Invalid nonce');
6652 - exit;
6653 - }
6654 -
6655 - if (!current_user_can('manage_options')) {
6656 - wp_send_json_error('Unauthorized access');
6657 - exit;
6658 - }
6659 -
6660 - $vector_id = isset($_POST['vector_id']) ? sanitize_text_field($_POST['vector_id']) : '';
6661 - $bot_id = isset($_POST['bot_id']) ? sanitize_text_field($_POST['bot_id']) : 'default';
6662 -
6663 - if (empty($vector_id)) {
6664 - wp_send_json_error('Missing vector ID');
6665 - exit;
6666 - }
6667 -
6668 - // Get bot-specific Pinecone settings
6669 - $pinecone_manager = MxChat_Pinecone_Manager::get_instance();
6670 - $pinecone_options = $pinecone_manager->mxchat_get_bot_pinecone_options($bot_id);
6671 -
6672 - $use_pinecone = ($pinecone_options['mxchat_use_pinecone'] ?? '0') === '1';
6673 -
6674 - if (!$use_pinecone || empty($pinecone_options['mxchat_pinecone_api_key'])) {
6675 - wp_send_json_error('Pinecone is not properly configured for bot: ' . $bot_id);
6676 - exit;
6677 - }
6678 -
6679 - // Delete from the correct Pinecone index
6680 - $result = $pinecone_manager->mxchat_delete_from_pinecone_by_vector_id(
6681 - $vector_id,
6682 - $pinecone_options['mxchat_pinecone_api_key'],
6683 - $pinecone_options['mxchat_pinecone_host']
6684 - );
6685 -
6686 - if ($result['success']) {
6687 - // No cache clearing needed since we removed caching
6688 - wp_send_json_success(array(
6689 - 'message' => 'Entry deleted successfully from Pinecone',
6690 - 'vector_id' => $vector_id,
6691 - 'bot_id' => $bot_id
6692 - ));
6693 - } else {
6694 - MxChat_Admin::mxchat_log_debug('pinecone_error', 'Failed to delete from Pinecone: ' . $result['message'], array('vector_id' => $vector_id, 'bot_id' => $bot_id));
6695 - wp_send_json_error('Failed to delete from Pinecone: ' . $result['message']);
6696 - }
6697 -
6698 - exit;
6699 -}
6700 -
6701 -/**
6702 - * Handle deletion of all chunks for a given source URL via AJAX
6703 - * Follows the same pattern as ajax_mxchat_delete_pinecone_prompt
6704 - */
6705 -public function ajax_mxchat_delete_chunks_by_url() {
6706 - // Verify nonce and permissions
6707 - if (!check_ajax_referer('mxchat_delete_chunks_nonce', 'nonce', false)) {
6708 - wp_send_json_error('Invalid nonce');
6709 - exit;
6710 - }
6711 -
6712 - if (!current_user_can('manage_options')) {
6713 - wp_send_json_error('Unauthorized access');
6714 - exit;
6715 - }
6716 -
6717 - $source_url = isset($_POST['source_url']) ? esc_url_raw($_POST['source_url']) : '';
6718 - $data_source = isset($_POST['data_source']) ? sanitize_text_field($_POST['data_source']) : 'wordpress';
6719 - $bot_id = isset($_POST['bot_id']) ? sanitize_text_field($_POST['bot_id']) : 'default';
6720 -
6721 - if (empty($source_url)) {
6722 - wp_send_json_error('Missing source URL');
6723 - exit;
6724 - }
6725 -
6726 - // Generate the base vector ID from the source URL (same as how chunks are created)
6727 - $base_vector_id = md5($source_url);
6728 -
6729 - if ($data_source === 'pinecone') {
6730 - // Get bot-specific Pinecone settings (same as working delete function)
6731 - $pinecone_manager = MxChat_Pinecone_Manager::get_instance();
6732 - $pinecone_options = $pinecone_manager->mxchat_get_bot_pinecone_options($bot_id);
6733 -
6734 - $use_pinecone = ($pinecone_options['mxchat_use_pinecone'] ?? '0') === '1';
6735 -
6736 - if (!$use_pinecone || empty($pinecone_options['mxchat_pinecone_api_key'])) {
6737 - wp_send_json_error('Pinecone is not properly configured for bot: ' . $bot_id);
6738 - exit;
6739 - }
6740 -
6741 - $api_key = $pinecone_options['mxchat_pinecone_api_key'];
6742 - $host = $pinecone_options['mxchat_pinecone_host'];
6743 - $namespace = $pinecone_options['mxchat_pinecone_namespace'] ?? '';
6744 -
6745 - // Collect all vector IDs to delete
6746 - $vectors_to_delete = array();
6747 -
6748 - // Add the original single-vector ID (for non-chunked content)
6749 - $vectors_to_delete[] = $base_vector_id;
6750 -
6751 - // Use Pinecone list API to find all chunk vectors with this prefix
6752 - // NOTE: Pinecone List API is a GET request with query parameters, not POST
6753 - $prefix = $base_vector_id . '_chunk_';
6754 -
6755 - $query_params = array(
6756 - 'prefix' => $prefix,
6757 - 'limit' => 100
6758 - );
6759 -
6760 - if (!empty($namespace)) {
6761 - $query_params['namespace'] = $namespace;
6762 - }
6763 -
6764 - $list_url = "https://{$host}/vectors/list?" . http_build_query($query_params);
6765 -
6766 - $list_response = wp_remote_get($list_url, array(
6767 - 'headers' => array(
6768 - 'Api-Key' => $api_key,
6769 - 'accept' => 'application/json'
6770 - ),
6771 - 'timeout' => 30
6772 - ));
6773 -
6774 - if (!is_wp_error($list_response)) {
6775 - $list_body_response = wp_remote_retrieve_body($list_response);
6776 - $list_data = json_decode($list_body_response, true);
6777 - if (!empty($list_data['vectors'])) {
6778 - foreach ($list_data['vectors'] as $vector) {
6779 - if (isset($vector['id'])) {
6780 - $vectors_to_delete[] = $vector['id'];
6781 - }
6782 - }
6783 - }
6784 - }
6785 -
6786 - if (empty($vectors_to_delete)) {
6787 - wp_send_json_success(array(
6788 - 'message' => 'No vectors found to delete',
6789 - 'source_url' => $source_url
6790 - ));
6791 - exit;
6792 - }
6793 -
6794 - // Delete all vectors using the same endpoint as the working function
6795 - $delete_url = "https://{$host}/vectors/delete";
6796 -
6797 - $delete_body = array(
6798 - 'ids' => $vectors_to_delete
6799 - );
6800 -
6801 - if (!empty($namespace)) {
6802 - $delete_body['namespace'] = $namespace;
6803 - }
6804 -
6805 - $delete_response = wp_remote_post($delete_url, array(
6806 - 'headers' => array(
6807 - 'Api-Key' => $api_key,
6808 - 'accept' => 'application/json',
6809 - 'content-type' => 'application/json'
6810 - ),
6811 - 'body' => wp_json_encode($delete_body),
6812 - 'timeout' => 30
6813 - ));
6814 -
6815 - if (is_wp_error($delete_response)) {
6816 - MxChat_Admin::mxchat_log_debug('pinecone_error', 'Failed to delete chunks from Pinecone: ' . $delete_response->get_error_message(), array('source_url' => $source_url));
6817 - wp_send_json_error('Failed to delete from Pinecone: ' . $delete_response->get_error_message());
6818 - exit;
6819 - }
6820 -
6821 - $response_code = wp_remote_retrieve_response_code($delete_response);
6822 -
6823 - if ($response_code !== 200) {
6824 - MxChat_Admin::mxchat_log_debug('pinecone_error', 'Pinecone API error (HTTP ' . $response_code . ')', array('source_url' => $source_url));
6825 - wp_send_json_error('Pinecone API error (HTTP ' . $response_code . ')');
6826 - exit;
6827 - }
6828 -
6829 - wp_send_json_success(array(
6830 - 'message' => 'All chunks deleted successfully from Pinecone',
6831 - 'source_url' => $source_url,
6832 - 'deleted_count' => count($vectors_to_delete)
6833 - ));
6834 -
6835 - } else {
6836 - // WordPress database deletion
6837 - global $wpdb;
6838 - $table_name = $wpdb->prefix . 'mxchat_system_prompt_content';
6839 -
6840 - $result = $wpdb->delete(
6841 - $table_name,
6842 - array('source_url' => $source_url),
6843 - array('%s')
6844 - );
6845 -
6846 - if ($result === false) {
6847 - MxChat_Admin::mxchat_log_debug('database_error', 'Failed to delete from database: ' . $wpdb->last_error, array('source_url' => $source_url));
6848 - wp_send_json_error('Failed to delete from database: ' . $wpdb->last_error);
6849 - exit;
6850 - }
6851 -
6852 - wp_send_json_success(array(
6853 - 'message' => 'All chunks deleted successfully from database',
6854 - 'source_url' => $source_url,
6855 - 'deleted_count' => $result
6856 - ));
6857 - }
6858 -
6859 - exit;
6860 -}
6861 -
6862 -/**
6863 - * Handle individual WordPress database content deletion via AJAX
6864 - * Mirrors the Pinecone delete handler but for WordPress database entries
6865 - */
6866 -public function ajax_mxchat_delete_wordpress_prompt() {
6867 - // Verify nonce and permissions
6868 - if (!check_ajax_referer('mxchat_delete_wordpress_prompt_nonce', 'nonce', false)) {
6869 - wp_send_json_error('Invalid nonce');
6870 - exit;
6871 - }
6872 -
6873 - if (!current_user_can('manage_options')) {
6874 - wp_send_json_error('Unauthorized access');
6875 - exit;
6876 - }
6877 -
6878 - $entry_id = isset($_POST['entry_id']) ? intval($_POST['entry_id']) : 0;
6879 -
6880 - if (empty($entry_id)) {
6881 - wp_send_json_error('Missing entry ID');
6882 - exit;
6883 - }
6884 -
6885 - global $wpdb;
6886 - $table_name = $wpdb->prefix . 'mxchat_system_prompt_content';
6887 -
6888 - // Clear cache for this entry
6889 - wp_cache_delete('prompt_' . $entry_id, 'mxchat_prompts');
6890 -
6891 - // Delete from database
6892 - $result = $wpdb->delete(
6893 - $table_name,
6894 - array('id' => $entry_id),
6895 - array('%d')
6896 - );
6897 -
6898 - if ($result !== false) {
6899 - wp_send_json_success(array(
6900 - 'message' => 'Entry deleted successfully',
6901 - 'entry_id' => $entry_id
6902 - ));
6903 - } else {
6904 - wp_send_json_error('Failed to delete entry from database');
6905 - }
6906 -
6907 - exit;
6908 -}
6909 -
6910 -/**
6911 - * Handle bulk deletion of knowledge entries via AJAX
6912 - * Supports both Pinecone and WordPress database entries
6913 - */
6914 -public function ajax_mxchat_bulk_delete_knowledge() {
6915 - // Verify nonce and permissions
6916 - if (!check_ajax_referer('mxchat_bulk_delete_knowledge_nonce', 'nonce', false)) {
6917 - wp_send_json_error('Invalid nonce');
6918 - exit;
6919 - }
6920 -
6921 - if (!current_user_can('manage_options')) {
6922 - wp_send_json_error('Unauthorized access');
6923 - exit;
6924 - }
6925 -
6926 - $entries = isset($_POST['entries']) ? $_POST['entries'] : array();
6927 - $bot_id = isset($_POST['bot_id']) ? sanitize_text_field($_POST['bot_id']) : 'default';
6928 -
6929 - if (empty($entries) || !is_array($entries)) {
6930 - wp_send_json_error('No entries provided');
6931 - exit;
6932 - }
6933 -
6934 - // Extend execution time — bulk Pinecone operations can take a while
6935 - if (function_exists('set_time_limit')) {
6936 - set_time_limit(120);
6937 - }
6938 -
6939 - $success_ids = array();
6940 - $failed_ids = array();
6941 - $errors = array();
6942 -
6943 - global $wpdb;
6944 - $table_name = $wpdb->prefix . 'mxchat_system_prompt_content';
6945 -
6946 - // Get Pinecone manager for Pinecone deletions
6947 - $pinecone_manager = MxChat_Pinecone_Manager::get_instance();
6948 - $pinecone_options = $pinecone_manager->mxchat_get_bot_pinecone_options($bot_id);
6949 - $use_pinecone = ($pinecone_options['mxchat_use_pinecone'] ?? '0') === '1';
6950 -
6951 - $api_key = $pinecone_options['mxchat_pinecone_api_key'] ?? '';
6952 - $host = $pinecone_options['mxchat_pinecone_host'] ?? '';
6953 -
6954 - // =============================================
6955 - // PHASE 1: Collect all Pinecone vector IDs
6956 - // and separate WordPress entries
6957 - // =============================================
6958 - $pinecone_entry_ids = array(); // entry IDs that are pinecone-sourced
6959 - $wordpress_entries = array(); // entries for WordPress DB deletion
6960 - $all_vector_ids = array(); // all pinecone vector IDs to delete in one batch
6961 -
6962 - foreach ($entries as $entry) {
6963 - $entry_id = sanitize_text_field($entry['id'] ?? '');
6964 - $source = sanitize_text_field($entry['source'] ?? 'wordpress');
6965 - $source_url = isset($entry['sourceUrl']) ? esc_url_raw($entry['sourceUrl']) : '';
6966 - $is_group = isset($entry['isGroup']) && ($entry['isGroup'] === true || $entry['isGroup'] === 'true');
6967 -
6968 - if (empty($entry_id)) {
6969 - continue;
6970 - }
6971 -
6972 - if ($source === 'pinecone') {
6973 - if (!$use_pinecone || empty($api_key)) {
6974 - $failed_ids[] = $entry_id;
6975 - $errors[] = "Pinecone not configured for entry: $entry_id";
6976 - continue;
6977 - }
6978 -
6979 - $pinecone_entry_ids[] = $entry_id;
6980 -
6981 - if ($is_group && !empty($source_url)) {
6982 - // Grouped/chunked entry: collect base ID + chunk IDs via List API
6983 - $base_vector_id = md5($source_url);
6984 - $all_vector_ids[] = $base_vector_id;
6985 -
6986 - $list_url = 'https://' . $host . '/vectors/list?prefix=' . urlencode($base_vector_id . '_chunk_') . '&limit=100';
6987 - $list_response = wp_remote_get($list_url, array(
6988 - 'headers' => array(
6989 - 'Api-Key' => $api_key,
6990 - 'accept' => 'application/json'
6991 - ),
6992 - 'timeout' => 30
6993 - ));
6994 -
6995 - if (!is_wp_error($list_response)) {
6996 - $list_body = json_decode(wp_remote_retrieve_body($list_response), true);
6997 - if (!empty($list_body['vectors']) && is_array($list_body['vectors'])) {
6998 - foreach ($list_body['vectors'] as $vector) {
6999 - if (isset($vector['id'])) {
7000 - $all_vector_ids[] = $vector['id'];
7001 - }
7002 - }
7003 - }
7004 - }
7005 - } else {
7006 - // Single entry: the entry_id IS the vector ID
7007 - $all_vector_ids[] = $entry_id;
7008 - }
7009 - } else {
7010 - $wordpress_entries[] = $entry;
7011 - }
7012 - }
7013 -
7014 - // =============================================
7015 - // PHASE 2: Single batch delete to Pinecone
7016 - // =============================================
7017 - if (!empty($all_vector_ids)) {
7018 - $all_vector_ids = array_values(array_unique($all_vector_ids));
7019 - $pinecone_success = true;
7020 - $batches = array_chunk($all_vector_ids, 100);
7021 -
7022 - foreach ($batches as $batch) {
7023 - $delete_response = wp_remote_post("https://{$host}/vectors/delete", array(
7024 - 'headers' => array(
7025 - 'Api-Key' => $api_key,
7026 - 'accept' => 'application/json',
7027 - 'content-type' => 'application/json'
7028 - ),
7029 - 'body' => wp_json_encode(array('ids' => $batch)),
7030 - 'timeout' => 60
7031 - ));
7032 -
7033 - if (is_wp_error($delete_response)) {
7034 - $pinecone_success = false;
7035 - $errors[] = 'Pinecone batch deletion failed: ' . $delete_response->get_error_message();
7036 - MxChat_Admin::mxchat_log_debug('pinecone_error', 'Bulk delete batch failed: ' . $delete_response->get_error_message());
7037 - } else {
7038 - $response_code = wp_remote_retrieve_response_code($delete_response);
7039 - if ($response_code !== 200) {
7040 - $pinecone_success = false;
7041 - $response_body = wp_remote_retrieve_body($delete_response);
7042 - $errors[] = "Pinecone API error (HTTP $response_code)";
7043 - MxChat_Admin::mxchat_log_debug('pinecone_error', 'Bulk delete batch failed (HTTP ' . $response_code . ')', array('response' => substr($response_body, 0, 200)));
7044 - }
7045 - }
7046 - }
7047 -
7048 - // Mark all pinecone entries based on batch result
7049 - foreach ($pinecone_entry_ids as $eid) {
7050 - if ($pinecone_success) {
7051 - $success_ids[] = $eid;
7052 - } else {
7053 - $failed_ids[] = $eid;
7054 - }
7055 - }
7056 - }
7057 -
7058 - // =============================================
7059 - // PHASE 3: WordPress database deletions
7060 - // =============================================
7061 - foreach ($wordpress_entries as $entry) {
7062 - $entry_id = sanitize_text_field($entry['id'] ?? '');
7063 - $source_url = isset($entry['sourceUrl']) ? esc_url_raw($entry['sourceUrl']) : '';
7064 - $is_group = isset($entry['isGroup']) && ($entry['isGroup'] === true || $entry['isGroup'] === 'true');
7065 -
7066 - if (empty($entry_id)) {
7067 - continue;
7068 - }
7069 -
7070 - try {
7071 - if ($is_group && !empty($source_url)) {
7072 - $result = $wpdb->delete(
7073 - $table_name,
7074 - array('source_url' => $source_url),
7075 - array('%s')
7076 - );
7077 - } else {
7078 - wp_cache_delete('prompt_' . $entry_id, 'mxchat_prompts');
7079 - $result = $wpdb->delete(
7080 - $table_name,
7081 - array('id' => intval($entry_id)),
7082 - array('%d')
7083 - );
7084 - }
7085 -
7086 - if ($result !== false) {
7087 - $success_ids[] = $entry_id;
7088 - } else {
7089 - $failed_ids[] = $entry_id;
7090 - $errors[] = "Database error for entry: $entry_id";
7091 - }
7092 - } catch (Exception $e) {
7093 - $failed_ids[] = $entry_id;
7094 - $errors[] = $e->getMessage();
7095 - }
7096 - }
7097 -
7098 - wp_send_json_success(array(
7099 - 'success_ids' => $success_ids,
7100 - 'failed_ids' => $failed_ids,
7101 - 'errors' => $errors,
7102 - 'total_processed' => count($success_ids) + count($failed_ids)
7103 - ));
7104 -
7105 - exit;
7106 -}
7107 -
7108 -/**
7109 - * Get hierarchical roles for dropdown
7110 - */
7111 -public function mxchat_get_role_options() {
7112 - return array(
7113 - 'public' => __('Public (Everyone)', 'mxchat'),
7114 - 'logged_in' => __('Logged In Users', 'mxchat'),
7115 - 'subscriber' => __('Subscribers & Above', 'mxchat'),
7116 - 'contributor' => __('Contributors & Above', 'mxchat'),
7117 - 'author' => __('Authors & Above', 'mxchat'),
7118 - 'editor' => __('Editors & Above', 'mxchat'),
7119 - 'administrator' => __('Administrators Only', 'mxchat')
7120 - );
7121 -}
7122 -
7123 -/**
7124 - * Check if user has access to content based on role restriction
7125 - */
7126 -public function mxchat_user_has_content_access($role_restriction) {
7127 - // Public content is always accessible
7128 - if ($role_restriction === 'public' || empty($role_restriction)) {
7129 - return true;
7130 - }
7131 -
7132 - // Check if user is logged in for logged_in restriction
7133 - if ($role_restriction === 'logged_in') {
7134 - return is_user_logged_in();
7135 - }
7136 -
7137 - // If not logged in, no access to role-restricted content
7138 - if (!is_user_logged_in()) {
7139 - return false;
7140 - }
7141 -
7142 - $user = wp_get_current_user();
7143 - $user_roles = $user->roles;
7144 -
7145 - if (empty($user_roles)) {
7146 - return false;
7147 - }
7148 -
7149 - // Define role hierarchy (higher number = higher access)
7150 - $hierarchy = array(
7151 - 'subscriber' => 1,
7152 - 'contributor' => 2,
7153 - 'author' => 3,
7154 - 'editor' => 4,
7155 - 'administrator' => 5
7156 - );
7157 -
7158 - // Get required level
7159 - $required_level = isset($hierarchy[$role_restriction]) ? $hierarchy[$role_restriction] : 0;
7160 -
7161 - // Check if user has required level or higher
7162 - foreach ($user_roles as $user_role) {
7163 - $user_level = isset($hierarchy[$user_role]) ? $hierarchy[$user_role] : 0;
7164 - if ($user_level >= $required_level) {
7165 - return true;
7166 - }
7167 - }
7168 -
7169 - return false;
7170 -}
7171 -
7172 -/**
7173 - * Handle role restriction updates via AJAX
7174 - * Removed cache clearing call since we removed caching
7175 - */
7176 -public function ajax_mxchat_update_role_restriction() {
7177 - // Verify nonce and permissions
7178 - if (!check_ajax_referer('mxchat_update_role_nonce', 'nonce', false)) {
7179 - wp_send_json_error('Invalid nonce');
7180 - exit;
7181 - }
7182 -
7183 - if (!current_user_can('manage_options')) {
7184 - wp_send_json_error('Unauthorized access');
7185 - exit;
7186 - }
7187 -
7188 - $entry_id = isset($_POST['entry_id']) ? sanitize_text_field($_POST['entry_id']) : '';
7189 - $role_restriction = isset($_POST['role_restriction']) ? sanitize_text_field($_POST['role_restriction']) : 'public';
7190 - $data_source = isset($_POST['data_source']) ? sanitize_text_field($_POST['data_source']) : 'wordpress';
7191 -
7192 - if (empty($entry_id)) {
7193 - wp_send_json_error('Invalid entry ID');
7194 - exit;
7195 - }
7196 -
7197 - // Get knowledge manager instance to validate role restriction
7198 - $knowledge_manager = MxChat_Knowledge_Manager::get_instance();
7199 - $valid_roles = array_keys($knowledge_manager->mxchat_get_role_options());
7200 - if (!in_array($role_restriction, $valid_roles)) {
7201 - wp_send_json_error('Invalid role restriction');
7202 - exit;
7203 - }
7204 -
7205 - global $wpdb;
7206 -
7207 - if ($data_source === 'pinecone') {
7208 - // Handle Pinecone role restriction (stored separately in WordPress table)
7209 - $roles_table = $wpdb->prefix . 'mxchat_pinecone_roles';
7210 -
7211 - // Use REPLACE to insert or update the role restriction
7212 - $result = $wpdb->replace(
7213 - $roles_table,
7214 - array(
7215 - 'vector_id' => $entry_id,
7216 - 'role_restriction' => $role_restriction,
7217 - 'updated_at' => current_time('mysql')
7218 - ),
7219 - array('%s', '%s', '%s')
7220 - );
7221 -
7222 - // No cache clearing needed since we removed caching
7223 -
7224 - } else {
7225 - // Handle WordPress database role restriction (existing functionality)
7226 - $table_name = $wpdb->prefix . 'mxchat_system_prompt_content';
7227 -
7228 - $result = $wpdb->update(
7229 - $table_name,
7230 - array('role_restriction' => $role_restriction),
7231 - array('id' => absint($entry_id)),
7232 - array('%s'),
7233 - array('%d')
7234 - );
7235 - }
7236 -
7237 - if ($result === false) {
7238 - wp_send_json_error('Database update failed: ' . $wpdb->last_error);
7239 - exit;
7240 - }
7241 -
7242 - wp_send_json_success(array(
7243 - 'message' => 'Role restriction updated successfully',
7244 - 'role_restriction' => $role_restriction,
7245 - 'data_source' => $data_source,
7246 - 'entry_id' => $entry_id
7247 - ));
7248 - exit;
7249 -}
7250 -
7251 -// ========================================
7252 -// ROLE-BASED CONTENT RESTRICTIONS - NEW FUNCTIONS
7253 -// Add these to your MxChat_Knowledge_Manager class
7254 -// ========================================
7255 -
7256 -/**
7257 - * Initialize role-based content hooks
7258 - * Add this call to your __construct() or mxchat_init_hooks() method
7259 - */
7260 -private function mxchat_init_role_hooks() {
7261 - // AJAX handlers for tag-role mappings
7262 - add_action('wp_ajax_mxchat_add_tag_role_mapping', array($this, 'ajax_add_tag_role_mapping'));
7263 - add_action('wp_ajax_mxchat_delete_tag_role_mapping', array($this, 'ajax_delete_tag_role_mapping'));
7264 - add_action('wp_ajax_mxchat_get_tag_role_mappings', array($this, 'ajax_get_tag_role_mappings'));
7265 - add_action('wp_ajax_mxchat_bulk_update_tag_roles', array($this, 'ajax_bulk_update_tag_roles'));
7266 -
7267 - // Hook to automatically update role restrictions when tags are added/removed
7268 - add_action('set_object_terms', array($this, 'handle_tag_change'), 10, 6);
7269 -
7270 - // Hook to apply role restrictions on auto-sync
7271 - add_action('mxchat_content_stored', array($this, 'apply_role_restriction_after_storage'), 10, 2);
7272 -}
7273 -
7274 -/**
7275 - * Add tag-role mapping via AJAX
7276 - */
7277 -public function ajax_add_tag_role_mapping() {
7278 - // Verify nonce and permissions
7279 - check_ajax_referer('mxchat_prompts_setting_nonce', 'nonce');
7280 -
7281 - if (!current_user_can('manage_options')) {
7282 - wp_send_json_error('Unauthorized access');
7283 - exit;
7284 - }
7285 -
7286 - $tag_input = isset($_POST['tag_slug']) ? sanitize_text_field($_POST['tag_slug']) : '';
7287 - $role_restriction = isset($_POST['role_restriction']) ? sanitize_text_field($_POST['role_restriction']) : 'public';
7288 -
7289 - if (empty($tag_input)) {
7290 - wp_send_json_error('Please enter a tag name or slug');
7291 - exit;
7292 - }
7293 -
7294 - // Validate role restriction
7295 - $valid_roles = array_keys($this->mxchat_get_role_options());
7296 - if (!in_array($role_restriction, $valid_roles)) {
7297 - wp_send_json_error('Invalid role restriction');
7298 - exit;
7299 - }
7300 -
7301 - // Resolve the tag by slug first, then fall back to its display name, so users can
7302 - // enter either "premium-content" or "Premium Content". (plan b8bcf5 — the field is
7303 - // labeled by name but previously validated by slug only, producing the confusing
7304 - // "Tag does not exist in WordPress" error when a real tag's name was typed.)
7305 - $term = get_term_by('slug', $tag_input, 'post_tag');
7306 - if (!$term) {
7307 - $term = get_term_by('name', $tag_input, 'post_tag');
7308 - }
7309 - if (!$term) {
7310 - wp_send_json_error('No tag with that name or slug exists yet. Create it under Posts → Tags first, then enter its name or slug.');
7311 - exit;
7312 - }
7313 -
7314 - // Always key the mapping by the RESOLVED slug — apply_role_restriction_to_post()
7315 - // compares against each post's tag slugs, so the stored key must be a slug,
7316 - // never the raw (possibly display-name) input.
7317 - $tag_slug = $term->slug;
7318 -
7319 - // Get existing mappings
7320 - $mappings = get_option('mxchat_tag_role_mappings', array());
7321 -
7322 - // Check if mapping already exists
7323 - if (isset($mappings[$tag_slug])) {
7324 - wp_send_json_error('Mapping for this tag already exists');
7325 - exit;
7326 - }
7327 -
7328 - // Add new mapping
7329 - $mappings[$tag_slug] = $role_restriction;
7330 - update_option('mxchat_tag_role_mappings', $mappings);
7331 -
7332 - wp_send_json_success(array(
7333 - 'message' => 'Tag-role mapping added successfully',
7334 - 'tag_slug' => $tag_slug,
7335 - 'role_restriction' => $role_restriction
7336 - ));
7337 - exit;
7338 -}
7339 -
7340 -/**
7341 - * Delete tag-role mapping via AJAX
7342 - */
7343 -public function ajax_delete_tag_role_mapping() {
7344 - // Verify nonce and permissions
7345 - check_ajax_referer('mxchat_prompts_setting_nonce', 'nonce');
7346 -
7347 - if (!current_user_can('manage_options')) {
7348 - wp_send_json_error('Unauthorized access');
7349 - exit;
7350 - }
7351 -
7352 - $tag_slug = isset($_POST['tag_slug']) ? sanitize_text_field($_POST['tag_slug']) : '';
7353 -
7354 - if (empty($tag_slug)) {
7355 - wp_send_json_error('Tag slug is required');
7356 - exit;
7357 - }
7358 -
7359 - // Get existing mappings
7360 - $mappings = get_option('mxchat_tag_role_mappings', array());
7361 -
7362 - // Check if mapping exists
7363 - if (!isset($mappings[$tag_slug])) {
7364 - wp_send_json_error('Mapping does not exist');
7365 - exit;
7366 - }
7367 -
7368 - // Remove mapping
7369 - unset($mappings[$tag_slug]);
7370 - update_option('mxchat_tag_role_mappings', $mappings);
7371 -
7372 - wp_send_json_success(array(
7373 - 'message' => 'Tag-role mapping deleted successfully',
7374 - 'tag_slug' => $tag_slug
7375 - ));
7376 - exit;
7377 -}
7378 -
7379 -/**
7380 - * Get all tag-role mappings via AJAX
7381 - */
7382 -public function ajax_get_tag_role_mappings() {
7383 - // Verify nonce and permissions
7384 - check_ajax_referer('mxchat_prompts_setting_nonce', 'nonce');
7385 -
7386 - if (!current_user_can('manage_options')) {
7387 - wp_send_json_error('Unauthorized access');
7388 - exit;
7389 - }
7390 -
7391 - // Get mappings
7392 - $mappings = get_option('mxchat_tag_role_mappings', array());
7393 - $role_options = $this->mxchat_get_role_options();
7394 -
7395 - $formatted_mappings = array();
7396 -
7397 - foreach ($mappings as $tag_slug => $role_restriction) {
7398 - // Get tag object
7399 - $term = get_term_by('slug', $tag_slug, 'post_tag');
7400 -
7401 - // Count posts with this tag
7402 - $post_count = 0;
7403 - if ($term) {
7404 - $post_count = $term->count;
7405 - }
7406 -
7407 - $formatted_mappings[] = array(
7408 - 'tag_slug' => $tag_slug,
7409 - 'role_restriction' => $role_restriction,
7410 - 'role_label' => $role_options[$role_restriction] ?? $role_restriction,
7411 - 'post_count' => $post_count
7412 - );
7413 - }
7414 -
7415 - wp_send_json_success(array(
7416 - 'mappings' => $formatted_mappings
7417 - ));
7418 - exit;
7419 -}
7420 -
7421 -/**
7422 - * Bulk update role restrictions for all existing content with mapped tags
7423 - */
7424 -public function ajax_bulk_update_tag_roles() {
7425 - // Verify nonce and permissions
7426 - check_ajax_referer('mxchat_prompts_setting_nonce', 'nonce');
7427 -
7428 - if (!current_user_can('manage_options')) {
7429 - wp_send_json_error('Unauthorized access');
7430 - exit;
7431 - }
7432 -
7433 - // Get mappings
7434 - $mappings = get_option('mxchat_tag_role_mappings', array());
7435 -
7436 - if (empty($mappings)) {
7437 - wp_send_json_error('No tag-role mappings found');
7438 - exit;
7439 - }
7440 -
7441 - global $wpdb;
7442 -
7443 - // Check if using Pinecone
7444 - $pinecone_options = get_option('mxchat_pinecone_addon_options', array());
7445 - $use_pinecone = ($pinecone_options['mxchat_use_pinecone'] ?? '0') === '1';
7446 -
7447 - $updated_count = 0;
7448 - $details = array();
7449 -
7450 - foreach ($mappings as $tag_slug => $role_restriction) {
7451 - // Get all posts with this tag
7452 - $posts = get_posts(array(
7453 - 'tag' => $tag_slug,
7454 - 'post_type' => 'any',
7455 - 'posts_per_page' => -1,
7456 - 'fields' => 'ids',
7457 - 'post_status' => 'publish'
7458 - ));
7459 -
7460 - if (empty($posts)) {
7461 - continue;
7462 - }
7463 -
7464 - $tag_updated = 0;
7465 -
7466 - foreach ($posts as $post_id) {
7467 - $source_url = get_permalink($post_id);
7468 - if (!$source_url) {
7469 - continue;
7470 - }
7471 -
7472 - if ($use_pinecone && !empty($pinecone_options['mxchat_pinecone_api_key'])) {
7473 - // Update Pinecone role restriction
7474 - $roles_table = $wpdb->prefix . 'mxchat_pinecone_roles';
7475 - $vector_id = md5($source_url);
7476 -
7477 - $result = $wpdb->replace(
7478 - $roles_table,
7479 - array(
7480 - 'vector_id' => $vector_id,
7481 - 'role_restriction' => $role_restriction,
7482 - 'updated_at' => current_time('mysql')
7483 - ),
7484 - array('%s', '%s', '%s')
7485 - );
7486 - } else {
7487 - // Update WordPress DB
7488 - $table_name = $wpdb->prefix . 'mxchat_system_prompt_content';
7489 -
7490 - $result = $wpdb->update(
7491 - $table_name,
7492 - array('role_restriction' => $role_restriction),
7493 - array('source_url' => $source_url),
7494 - array('%s'),
7495 - array('%s')
7496 - );
7497 - }
7498 -
7499 - if ($result !== false) {
7500 - $tag_updated++;
7501 - $updated_count++;
7502 - }
7503 - }
7504 -
7505 - if ($tag_updated > 0) {
7506 - $details[] = sprintf(
7507 - 'Tag "%s" (%s): %d posts updated',
7508 - $tag_slug,
7509 - $role_restriction,
7510 - $tag_updated
7511 - );
7512 - }
7513 - }
7514 -
7515 - wp_send_json_success(array(
7516 - 'message' => 'Bulk update completed',
7517 - 'updated_count' => $updated_count,
7518 - 'tags_processed' => count($mappings),
7519 - 'details' => $details
7520 - ));
7521 - exit;
7522 -}
7523 -
7524 -/**
7525 - * Handle tag changes on posts (when tags are added or removed)
7526 - */
7527 -public function handle_tag_change($object_id, $terms, $tt_ids, $taxonomy, $append, $old_tt_ids) {
7528 - // Only process post tags
7529 - if ($taxonomy !== 'post_tag') {
7530 - return;
7531 - }
7532 -
7533 - // Get tag-role mappings
7534 - $mappings = get_option('mxchat_tag_role_mappings', array());
7535 -
7536 - if (empty($mappings)) {
7537 - return;
7538 - }
7539 -
7540 - // Get the post's URL
7541 - $source_url = get_permalink($object_id);
7542 - if (!$source_url) {
7543 - return;
7544 - }
7545 -
7546 - // Determine the highest role restriction based on tags
7547 - $highest_role = 'public';
7548 - $role_hierarchy = array(
7549 - 'public' => 0,
7550 - 'logged_in' => 1,
7551 - 'subscriber' => 2,
7552 - 'contributor' => 3,
7553 - 'author' => 4,
7554 - 'editor' => 5,
7555 - 'administrator' => 6
7556 - );
7557 -
7558 - // Get all current tags for the post
7559 - $current_tags = wp_get_post_tags($object_id, array('fields' => 'slugs'));
7560 -
7561 - // Find the highest role restriction among the tags
7562 - foreach ($current_tags as $tag_slug) {
7563 - if (isset($mappings[$tag_slug])) {
7564 - $role = $mappings[$tag_slug];
7565 - if (isset($role_hierarchy[$role]) && $role_hierarchy[$role] > $role_hierarchy[$highest_role]) {
7566 - $highest_role = $role;
7567 - }
7568 - }
7569 - }
7570 -
7571 - // Update the role restriction in the database
7572 - global $wpdb;
7573 -
7574 - // Check if using Pinecone
7575 - $pinecone_options = get_option('mxchat_pinecone_addon_options', array());
7576 - $use_pinecone = ($pinecone_options['mxchat_use_pinecone'] ?? '0') === '1';
7577 -
7578 - if ($use_pinecone && !empty($pinecone_options['mxchat_pinecone_api_key'])) {
7579 - // Update Pinecone role restriction
7580 - $roles_table = $wpdb->prefix . 'mxchat_pinecone_roles';
7581 - $vector_id = md5($source_url);
7582 -
7583 - $wpdb->replace(
7584 - $roles_table,
7585 - array(
7586 - 'vector_id' => $vector_id,
7587 - 'role_restriction' => $highest_role,
7588 - 'updated_at' => current_time('mysql')
7589 - ),
7590 - array('%s', '%s', '%s')
7591 - );
7592 - } else {
7593 - // Update WordPress DB
7594 - $table_name = $wpdb->prefix . 'mxchat_system_prompt_content';
7595 -
7596 - $wpdb->update(
7597 - $table_name,
7598 - array('role_restriction' => $highest_role),
7599 - array('source_url' => $source_url),
7600 - array('%s'),
7601 - array('%s')
7602 - );
7603 - }
7604 -}
7605 -
7606 -/**
7607 - * Apply role restriction after content is stored (for auto-sync)
7608 - */
7609 -public function apply_role_restriction_after_storage($post_id, $source_url) {
7610 - // Get tag-role mappings
7611 - $mappings = get_option('mxchat_tag_role_mappings', array());
7612 -
7613 - if (empty($mappings)) {
7614 - return;
7615 - }
7616 -
7617 - // Get all tags for the post
7618 - $post_tags = wp_get_post_tags($post_id, array('fields' => 'slugs'));
7619 -
7620 - if (empty($post_tags)) {
7621 - return;
7622 - }
7623 -
7624 - // Determine the highest role restriction based on tags
7625 - $highest_role = 'public';
7626 - $role_hierarchy = array(
7627 - 'public' => 0,
7628 - 'logged_in' => 1,
7629 - 'subscriber' => 2,
7630 - 'contributor' => 3,
7631 - 'author' => 4,
7632 - 'editor' => 5,
7633 - 'administrator' => 6
7634 - );
7635 -
7636 - foreach ($post_tags as $tag_slug) {
7637 - if (isset($mappings[$tag_slug])) {
7638 - $role = $mappings[$tag_slug];
7639 - if (isset($role_hierarchy[$role]) && $role_hierarchy[$role] > $role_hierarchy[$highest_role]) {
7640 - $highest_role = $role;
7641 - }
7642 - }
7643 - }
7644 -
7645 - // If no restricted tags found, return (leave as public)
7646 - if ($highest_role === 'public') {
7647 - return;
7648 - }
7649 -
7650 - // Update the role restriction
7651 - global $wpdb;
7652 -
7653 - // Check if using Pinecone
7654 - $pinecone_options = get_option('mxchat_pinecone_addon_options', array());
7655 - $use_pinecone = ($pinecone_options['mxchat_use_pinecone'] ?? '0') === '1';
7656 -
7657 - if ($use_pinecone && !empty($pinecone_options['mxchat_pinecone_api_key'])) {
7658 - // Update Pinecone role restriction
7659 - $roles_table = $wpdb->prefix . 'mxchat_pinecone_roles';
7660 - $vector_id = md5($source_url);
7661 -
7662 - $wpdb->replace(
7663 - $roles_table,
7664 - array(
7665 - 'vector_id' => $vector_id,
7666 - 'role_restriction' => $highest_role,
7667 - 'updated_at' => current_time('mysql')
7668 - ),
7669 - array('%s', '%s', '%s')
7670 - );
7671 - } else {
7672 - // Update WordPress DB
7673 - $table_name = $wpdb->prefix . 'mxchat_system_prompt_content';
7674 -
7675 - $wpdb->update(
7676 - $table_name,
7677 - array('role_restriction' => $highest_role),
7678 - array('source_url' => $source_url),
7679 - array('%s'),
7680 - array('%s')
7681 - );
7682 - }
7683 -}
7684 -
7685 -
7686 - // ========================================
7687 - // HELPER METHODS
7688 - // ========================================
7689 -
7690 - /**
7691 - * Check if user has required permissions for content processing
7692 - */
7693 - private function mxchat_check_user_permissions() {
7694 - if (!current_user_can('manage_options')) {
7695 - wp_die(esc_html__('You do not have sufficient permissions.', 'mxchat'));
7696 - }
7697 - }
7698 -
7699 - /**
7700 - * Validate nonce for security
7701 - */
7702 - private function mxchat_validate_nonce($nonce_name, $nonce_action) {
7703 - if (!isset($_POST[$nonce_name]) || !wp_verify_nonce($_POST[$nonce_name], $nonce_action)) {
7704 - wp_die(esc_html__('Security check failed.', 'mxchat'));
7705 - }
7706 - }
7707 -
7708 - /**
7709 - * Get embedding API credentials
7710 - */
7711 - private function mxchat_get_embedding_credentials() {
7712 - $embedding_model = $this->options['embedding_model'] ?? 'text-embedding-ada-002';
7713 -
7714 - if (strpos($embedding_model, 'text-embedding-') !== false) {
7715 - return array(
7716 - 'type' => 'openai',
7717 - 'api_key' => $this->options['api_key'] ?? ''
7718 - );
7719 - } elseif (strpos($embedding_model, 'voyage-') !== false) {
7720 - return array(
7721 - 'type' => 'voyage',
7722 - 'api_key' => $this->options['voyage_api_key'] ?? ''
7723 - );
7724 - } elseif (strpos($embedding_model, 'gemini-embedding-') !== false) {
7725 - return array(
7726 - 'type' => 'gemini',
7727 - 'api_key' => $this->options['gemini_api_key'] ?? ''
7728 - );
7729 - }
7730 -
7731 - return array('type' => 'unknown', 'api_key' => '');
7732 - }
7733 -
7734 - /**
7735 - * Log processing errors
7736 - */
7737 - private function mxchat_log_processing_error($operation, $error_message) {
7738 - //error_log("MxChat Knowledge Processing {$operation} Error: " . $error_message);
7739 - }
7740 -
7741 - /**
7742 - * Set admin notice transient
7743 - */
7744 - private function mxchat_set_admin_notice($type, $message) {
7745 - set_transient("mxchat_admin_notice_{$type}", $message, 30);
7746 - }
7747 -
7748 - /**
7749 - * Get Pinecone manager instance for vector operations
7750 - */
7751 - private function mxchat_get_pinecone_manager() {
7752 - return MxChat_Pinecone_Manager::get_instance();
7753 - }
7754 -
7755 -
7756 - // ========================================
7757 -// DATABASE QUEUE TABLE MANAGEMENT
7758 -// ========================================
7759 -
7760 -/**
7761 - * Create queue table on plugin activation
7762 - * Call this from your plugin activation hook
7763 - */
7764 -public function mxchat_create_queue_table() {
7765 - global $wpdb;
7766 -
7767 - $table_name = $wpdb->prefix . 'mxchat_processing_queue';
7768 - $charset_collate = $wpdb->get_charset_collate();
7769 -
7770 - $sql = "CREATE TABLE IF NOT EXISTS $table_name (
7771 - id bigint(20) unsigned NOT NULL AUTO_INCREMENT,
7772 - queue_id varchar(64) NOT NULL,
7773 - item_type varchar(20) NOT NULL,
7774 - item_data longtext NOT NULL,
7775 - status varchar(20) NOT NULL DEFAULT 'pending',
7776 - bot_id varchar(50) NOT NULL DEFAULT 'default',
7777 - priority int(11) NOT NULL DEFAULT 0,
7778 - attempts int(11) NOT NULL DEFAULT 0,
7779 - max_attempts int(11) NOT NULL DEFAULT 3,
7780 - error_message text DEFAULT NULL,
7781 - created_at datetime NOT NULL,
7782 - started_at datetime DEFAULT NULL,
7783 - completed_at datetime DEFAULT NULL,
7784 - PRIMARY KEY (id),
7785 - KEY queue_id (queue_id),
7786 - KEY status (status),
7787 - KEY item_type (item_type),
7788 - KEY priority (priority)
7789 - ) $charset_collate;";
7790 -
7791 - require_once(ABSPATH . 'wp-admin/includes/upgrade.php');
7792 - dbDelta($sql);
7793 -
7794 - // Also create a meta table for queue metadata
7795 - $meta_table = $wpdb->prefix . 'mxchat_queue_meta';
7796 -
7797 - $meta_sql = "CREATE TABLE IF NOT EXISTS $meta_table (
7798 - id bigint(20) unsigned NOT NULL AUTO_INCREMENT,
7799 - queue_id varchar(64) NOT NULL,
7800 - meta_key varchar(255) NOT NULL,
7801 - meta_value longtext,
7802 - PRIMARY KEY (id),
7803 - KEY queue_id (queue_id),
7804 - KEY meta_key (meta_key)
7805 - ) $charset_collate;";
7806 -
7807 - dbDelta($meta_sql);
7808 -}
7809 -
7810 -/**
7811 - * Add items to the processing queue
7812 - *
7813 - * @param string $queue_id Unique identifier for this queue batch
7814 - * @param string $item_type Type of item (url, pdf_page)
7815 - * @param array $items Array of items to queue
7816 - * @param string $bot_id Bot ID for processing
7817 - * @return int Number of items queued
7818 - */
7819 -private function mxchat_add_to_queue($queue_id, $item_type, $items, $bot_id = 'default') {
7820 - global $wpdb;
7821 - $table_name = $wpdb->prefix . 'mxchat_processing_queue';
7822 -
7823 - $queued_count = 0;
7824 - $priority = 0;
7825 -
7826 - foreach ($items as $item) {
7827 - $result = $wpdb->insert(
7828 - $table_name,
7829 - array(
7830 - 'queue_id' => $queue_id,
7831 - 'item_type' => $item_type,
7832 - 'item_data' => wp_json_encode($item),
7833 - 'status' => 'pending',
7834 - 'bot_id' => $bot_id,
7835 - 'priority' => $priority,
7836 - 'attempts' => 0,
7837 - 'max_attempts' => 3,
7838 - 'created_at' => current_time('mysql')
7839 - ),
7840 - array('%s', '%s', '%s', '%s', '%s', '%d', '%d', '%d', '%s')
7841 - );
7842 -
7843 - if ($result) {
7844 - $queued_count++;
7845 - }
7846 -
7847 - $priority++; // Process in order
7848 - }
7849 -
7850 - return $queued_count;
7851 -}
7852 -
7853 -/**
7854 - * Store queue metadata (total counts, source URL, etc.)
7855 - */
7856 -private function mxchat_set_queue_meta($queue_id, $meta_key, $meta_value) {
7857 - global $wpdb;
7858 - $meta_table = $wpdb->prefix . 'mxchat_queue_meta';
7859 -
7860 - // Check if meta exists
7861 - $existing = $wpdb->get_var($wpdb->prepare(
7862 - "SELECT id FROM $meta_table WHERE queue_id = %s AND meta_key = %s",
7863 - $queue_id,
7864 - $meta_key
7865 - ));
7866 -
7867 - if ($existing) {
7868 - // Update
7869 - $wpdb->update(
7870 - $meta_table,
7871 - array('meta_value' => maybe_serialize($meta_value)),
7872 - array('queue_id' => $queue_id, 'meta_key' => $meta_key),
7873 - array('%s'),
7874 - array('%s', '%s')
7875 - );
7876 - } else {
7877 - // Insert
7878 - $wpdb->insert(
7879 - $meta_table,
7880 - array(
7881 - 'queue_id' => $queue_id,
7882 - 'meta_key' => $meta_key,
7883 - 'meta_value' => maybe_serialize($meta_value)
7884 - ),
7885 - array('%s', '%s', '%s')
7886 - );
7887 - }
7888 -}
7889 -
7890 -/**
7891 - * Get queue metadata
7892 - */
7893 -private function mxchat_get_queue_meta($queue_id, $meta_key) {
7894 - global $wpdb;
7895 - $meta_table = $wpdb->prefix . 'mxchat_queue_meta';
7896 -
7897 - $value = $wpdb->get_var($wpdb->prepare(
7898 - "SELECT meta_value FROM $meta_table WHERE queue_id = %s AND meta_key = %s",
7899 - $queue_id,
7900 - $meta_key
7901 - ));
7902 -
7903 - return maybe_unserialize($value);
7904 -}
7905 -
7906 -// ========================================
7907 -// AJAX QUEUE PROCESSING HANDLERS
7908 -// ========================================
7909 -
7910 -/**
7911 - * AJAX: Get next item from queue to process
7912 - */
7913 -public function ajax_mxchat_get_next_queue_item() {
7914 - // Verify nonce and permissions
7915 - check_ajax_referer('mxchat_queue_nonce', 'nonce');
7916 -
7917 - if (!current_user_can('manage_options')) {
7918 - wp_send_json_error('Unauthorized access');
7919 - }
7920 -
7921 - $queue_id = isset($_POST['queue_id']) ? sanitize_text_field($_POST['queue_id']) : '';
7922 -
7923 - if (empty($queue_id)) {
7924 - wp_send_json_error('Missing queue ID');
7925 - }
7926 -
7927 - global $wpdb;
7928 - $table_name = $wpdb->prefix . 'mxchat_processing_queue';
7929 -
7930 - // Get next pending item with retry logic for failed items
7931 - $next_item = $wpdb->get_row($wpdb->prepare(
7932 - "SELECT * FROM $table_name
7933 - WHERE queue_id = %s
7934 - AND status IN ('pending', 'failed')
7935 - AND attempts < max_attempts
7936 - ORDER BY priority ASC, id ASC
7937 - LIMIT 1",
7938 - $queue_id
7939 - ));
7940 -
7941 - if (!$next_item) {
7942 - // No more items - queue complete
7943 - wp_send_json_success(array(
7944 - 'complete' => true,
7945 - 'message' => 'Queue processing complete'
7946 - ));
7947 - }
7948 -
7949 - // Mark item as processing
7950 - $wpdb->update(
7951 - $table_name,
7952 - array(
7953 - 'status' => 'processing',
7954 - 'started_at' => current_time('mysql'),
7955 - 'attempts' => $next_item->attempts + 1
7956 - ),
7957 - array('id' => $next_item->id),
7958 - array('%s', '%s', '%d'),
7959 - array('%d')
7960 - );
7961 -
7962 - wp_send_json_success(array(
7963 - 'complete' => false,
7964 - 'item' => array(
7965 - 'id' => $next_item->id,
7966 - 'type' => $next_item->item_type,
7967 - 'data' => json_decode($next_item->item_data, true),
7968 - 'bot_id' => $next_item->bot_id,
7969 - 'attempt' => $next_item->attempts + 1
7970 - )
7971 - ));
7972 -}
7973 -
7974 -/**
7975 - * AJAX: Process a single queue item
7976 - */
7977 -public function ajax_mxchat_process_queue_item() {
7978 - // Verify nonce and permissions
7979 - check_ajax_referer('mxchat_queue_nonce', 'nonce');
7980 -
7981 - if (!current_user_can('manage_options')) {
7982 - wp_send_json_error('Unauthorized access');
7983 - }
7984 -
7985 - $item_id = isset($_POST['item_id']) ? absint($_POST['item_id']) : 0;
7986 - $item_type = isset($_POST['item_type']) ? sanitize_text_field($_POST['item_type']) : '';
7987 - $item_data = isset($_POST['item_data']) ? $_POST['item_data'] : array();
7988 - $bot_id = isset($_POST['bot_id']) ? sanitize_key($_POST['bot_id']) : 'default';
7989 -
7990 - if (empty($item_id) || empty($item_type)) {
7991 - wp_send_json_error('Missing item data');
7992 - }
7993 -
7994 - global $wpdb;
7995 - $table_name = $wpdb->prefix . 'mxchat_processing_queue';
7996 -
7997 - // Process based on item type
7998 - try {
7999 - set_time_limit(60); // Give processing 60 seconds
8000 -
8001 - $result = false;
8002 - $error_message = '';
8003 -
8004 - // Read item directly from DB to get queue_id and preserve special chars in item_data
8005 - // (POST round-trip through JS mangles characters like apostrophes in URLs)
8006 - $db_item = $wpdb->get_row($wpdb->prepare(
8007 - "SELECT queue_id, item_data FROM $table_name WHERE id = %d",
8008 - $item_id
8009 - ));
8010 - $item_queue_id = $db_item ? $db_item->queue_id : '';
8011 - if ($db_item && !empty($db_item->item_data)) {
8012 - $db_data = json_decode($db_item->item_data, true);
8013 - if (is_array($db_data)) {
8014 - $item_data = $db_data;
8015 - }
8016 - }
8017 -
8018 - switch ($item_type) {
8019 - case 'url':
8020 - $result = $this->mxchat_process_queue_url($item_data, $bot_id, $item_queue_id);
8021 - break;
8022 -
8023 - case 'pdf_page':
8024 - $result = $this->mxchat_process_queue_pdf_page($item_data, $bot_id);
8025 - break;
8026 -
8027 - default:
8028 - throw new Exception('Unknown item type: ' . $item_type);
8029 - }
8030 -
8031 - if (is_wp_error($result)) {
8032 - $error_code = $result->get_error_code();
8033 - // Content errors (empty page, sanitization) are permanent — retrying won't help
8034 - $permanent_codes = array('empty_page', 'empty_after_sanitization', 'no_api_key', 'page_not_found');
8035 - if (in_array($error_code, $permanent_codes)) {
8036 - // Mark as permanently failed — set attempts = max_attempts so it won't be retried
8037 - $current_item = $wpdb->get_row($wpdb->prepare(
8038 - "SELECT max_attempts FROM $table_name WHERE id = %d", $item_id
8039 - ));
8040 - $wpdb->update(
8041 - $table_name,
8042 - array(
8043 - 'status' => 'failed',
8044 - 'error_message' => $result->get_error_message(),
8045 - 'attempts' => $current_item ? $current_item->max_attempts : 3
8046 - ),
8047 - array('id' => $item_id),
8048 - array('%s', '%s', '%d'),
8049 - array('%d')
8050 - );
8051 - wp_send_json_error(array(
8052 - 'message' => $result->get_error_message(),
8053 - 'permanent_failure' => true,
8054 - 'item_id' => $item_id
8055 - ));
8056 - return;
8057 - }
8058 - throw new Exception($result->get_error_message());
8059 - }
8060 -
8061 - if ($result === false) {
8062 - throw new Exception('Processing returned false - item may be empty or invalid');
8063 - }
8064 -
8065 - // Mark as completed
8066 - $wpdb->update(
8067 - $table_name,
8068 - array(
8069 - 'status' => 'completed',
8070 - 'completed_at' => current_time('mysql'),
8071 - 'error_message' => null
8072 - ),
8073 - array('id' => $item_id),
8074 - array('%s', '%s', '%s'),
8075 - array('%d')
8076 - );
8077 -
8078 - wp_send_json_success(array(
8079 - 'processed' => true,
8080 - 'item_id' => $item_id,
8081 - 'message' => 'Item processed successfully'
8082 - ));
8083 -
8084 - } catch (Exception $e) {
8085 - $error_message = $e->getMessage();
8086 -
8087 - // Get current attempt count
8088 - $item = $wpdb->get_row($wpdb->prepare(
8089 - "SELECT attempts, max_attempts FROM $table_name WHERE id = %d",
8090 - $item_id
8091 - ));
8092 -
8093 - // Check if we've exhausted retries
8094 - if ($item && $item->attempts >= $item->max_attempts) {
8095 - // Permanently failed
8096 - $wpdb->update(
8097 - $table_name,
8098 - array(
8099 - 'status' => 'failed',
8100 - 'error_message' => $error_message
8101 - ),
8102 - array('id' => $item_id),
8103 - array('%s', '%s'),
8104 - array('%d')
8105 - );
8106 -
8107 - wp_send_json_error(array(
8108 - 'message' => 'Item failed after maximum attempts: ' . $error_message,
8109 - 'permanent_failure' => true,
8110 - 'item_id' => $item_id
8111 - ));
8112 - } else {
8113 - // Mark for retry
8114 - $wpdb->update(
8115 - $table_name,
8116 - array(
8117 - 'status' => 'failed',
8118 - 'error_message' => $error_message
8119 - ),
8120 - array('id' => $item_id),
8121 - array('%s', '%s'),
8122 - array('%d')
8123 - );
8124 -
8125 - wp_send_json_error(array(
8126 - 'message' => 'Item processing failed, will retry: ' . $error_message,
8127 - 'can_retry' => true,
8128 - 'item_id' => $item_id,
8129 - 'attempts' => $item ? $item->attempts : 0
8130 - ));
8131 - }
8132 - }
8133 -}
8134 -
8135 -/**
8136 - * Process a URL from the queue
8137 - */
8138 -private function mxchat_process_queue_url($item_data, $bot_id = 'default', $queue_id = '') {
8139 - $url = isset($item_data['url']) ? $item_data['url'] : '';
8140 -
8141 - if (empty($url)) {
8142 - return new WP_Error('invalid_url', 'URL is empty');
8143 - }
8144 -
8145 - // Get bot-specific API key early (needed for both paths)
8146 - $bot_options = $this->get_bot_options($bot_id);
8147 - $options = !empty($bot_options) ? $bot_options : get_option('mxchat_options');
8148 - $selected_model = $options['embedding_model'] ?? 'text-embedding-ada-002';
8149 -
8150 - if (strpos($selected_model, 'voyage') === 0) {
8151 - $api_key = $options['voyage_api_key'] ?? '';
8152 - } elseif (strpos($selected_model, 'gemini-embedding') === 0) {
8153 - $api_key = $options['gemini_api_key'] ?? '';
8154 - } else {
8155 - $api_key = $options['api_key'] ?? '';
8156 - }
8157 -
8158 - if (empty($api_key)) {
8159 - return new WP_Error('no_api_key', 'API key not configured for bot: ' . $bot_id);
8160 - }
8161 -
8162 - // Check if this is a WooCommerce product URL and WooCommerce is active
8163 - $is_product_url = (strpos($url, '/product/') !== false || strpos($url, '/shop/') !== false);
8164 - $content_type = $is_product_url ? 'product' : 'url';
8165 -
8166 - // Try to get WooCommerce product data if it's a product URL
8167 - if ($is_product_url && class_exists('WooCommerce')) {
8168 - $product_content = $this->mxchat_extract_woocommerce_product_content($url);
8169 -
8170 - if (!empty($product_content)) {
8171 - // Successfully extracted WooCommerce product data with pricing
8172 - $result = MxChat_Utils::submit_content_to_db(
8173 - $product_content,
8174 - $url,
8175 - $api_key,
8176 - null,
8177 - $bot_id,
8178 - 'product'
8179 - );
8180 - return $result;
8181 - }
8182 - // If WooCommerce extraction failed, fall through to HTML extraction
8183 - }
8184 -
8185 - // Fetch URL content (fallback for non-products or when WooCommerce extraction fails)
8186 - $is_likely_pdf = (strtolower(pathinfo(parse_url($url, PHP_URL_PATH) ?: '', PATHINFO_EXTENSION)) === 'pdf');
8187 - $response = wp_remote_get($url, array(
8188 - 'timeout' => $is_likely_pdf ? 120 : 30,
8189 - 'redirection' => 5,
8190 - 'user-agent' => mxchat_ingest_user_agent(),
8191 - ));
8192 -
8193 - if (is_wp_error($response)) {
8194 - return $response;
8195 - }
8196 -
8197 - $response_code = wp_remote_retrieve_response_code($response);
8198 - if ($response_code !== 200) {
8199 - return new WP_Error('http_error', 'HTTP ' . $response_code . ' error for: ' . $url);
8200 - }
8201 -
8202 - // Check if URL is a PDF — expand into per-page queue items using the standard PDF pipeline
8203 - if ($this->mxchat_is_pdf_url($url, $response)) {
8204 - return $this->mxchat_expand_pdf_to_queue($url, $response, $bot_id, $queue_id);
8205 - }
8206 -
8207 - $html = wp_remote_retrieve_body($response);
8208 -
8209 - if (empty($html)) {
8210 - return new WP_Error('empty_response', 'Empty response body');
8211 - }
8212 -
8213 - // Extract and sanitize content
8214 - $content = $this->mxchat_extract_main_content($html);
8215 - $sanitized = $this->mxchat_sanitize_content_for_api($content);
8216 -
8217 - if (empty($sanitized)) {
8218 - // Not an error - just no content found (maybe a redirect or empty page)
8219 - return false;
8220 - }
8221 -
8222 - // Submit to database with content_type
8223 - $result = MxChat_Utils::submit_content_to_db(
8224 - $sanitized,
8225 - $url,
8226 - $api_key,
8227 - null,
8228 - $bot_id,
8229 - $content_type
8230 - );
8231 -
8232 - return $result;
8233 -}
8234 -
8235 -/**
8236 - * Expand a PDF URL into per-page queue items using the standard PDF pipeline.
8237 - * Called when a sitemap URL turns out to be a PDF — downloads, parses page count,
8238 - * and adds pdf_page items to the same queue so they process with full progress tracking.
8239 - */
8240 -private function mxchat_expand_pdf_to_queue($pdf_url, $response, $bot_id = 'default', $queue_id = '') {
8241 - set_time_limit(120); // PDFs need extra time for download + parsing
8242 -
8243 - $upload_dir = wp_upload_dir();
8244 - $pdf_filename = sanitize_file_name('mxchat_kb_' . md5($pdf_url) . '.pdf');
8245 - $pdf_path = trailingslashit($upload_dir['path']) . $pdf_filename;
8246 -
8247 - $response_body = wp_remote_retrieve_body($response);
8248 - if (empty($response_body)) {
8249 - return new WP_Error('empty_pdf', 'Empty PDF response for: ' . $pdf_url);
8250 - }
8251 -
8252 - if (!wp_mkdir_p(dirname($pdf_path))) {
8253 - return new WP_Error('dir_error', 'Failed to create upload directory');
8254 - }
8255 -
8256 - file_put_contents($pdf_path, $response_body);
8257 -
8258 - if (!file_exists($pdf_path)) {
8259 - return new WP_Error('save_error', 'Failed to save PDF file');
8260 - }
8261 -
8262 - try {
8263 - $total_pages = $this->mxchat_validate_and_count_pdf_pages($pdf_path);
8264 -
8265 - if ($total_pages === false || $total_pages < 1) {
8266 - wp_delete_file($pdf_path);
8267 - return new WP_Error('no_pages', 'PDF has no pages: ' . $pdf_url);
8268 - }
8269 -
8270 - // Build per-page items identical to mxchat_handle_pdf_for_knowledge_base
8271 - $pages = array();
8272 - for ($i = 1; $i <= $total_pages; $i++) {
8273 - $pages[] = array(
8274 - 'pdf_path' => $pdf_path,
8275 - 'pdf_url' => $pdf_url,
8276 - 'page_number' => $i,
8277 - 'total_pages' => $total_pages
8278 - );
8279 - }
8280 -
8281 - // Add pdf_page items to the SAME queue so the JS picks them up automatically
8282 - if (!empty($queue_id)) {
8283 - $queued_count = $this->mxchat_add_to_queue($queue_id, 'pdf_page', $pages, $bot_id);
8284 - } else {
8285 - // Fallback: create a new PDF queue (shouldn't happen in sitemap flow)
8286 - $new_queue_id = 'pdf_' . md5($pdf_url . time());
8287 - $queued_count = $this->mxchat_add_to_queue($new_queue_id, 'pdf_page', $pages, $bot_id);
8288 - $this->mxchat_set_queue_meta($new_queue_id, 'source_url', $pdf_url);
8289 - $this->mxchat_set_queue_meta($new_queue_id, 'queue_type', 'pdf');
8290 - $this->mxchat_set_queue_meta($new_queue_id, 'total_items', $total_pages);
8291 - $this->mxchat_set_queue_meta($new_queue_id, 'bot_id', $bot_id);
8292 - $this->mxchat_set_queue_meta($new_queue_id, 'pdf_path', $pdf_path);
8293 - $this->mxchat_set_queue_meta($new_queue_id, 'created_at', current_time('mysql'));
8294 - }
8295 -
8296 - if ($queued_count === 0) {
8297 - wp_delete_file($pdf_path);
8298 - return new WP_Error('queue_error', 'Failed to add PDF pages to queue');
8299 - }
8300 -
8301 - // Return true so the original URL item is marked complete
8302 - // The new pdf_page items will be processed in subsequent batches
8303 - return true;
8304 -
8305 - } catch (Exception $e) {
8306 - if (file_exists($pdf_path)) {
8307 - wp_delete_file($pdf_path);
8308 - }
8309 - return new WP_Error('pdf_parse_error', 'Error parsing PDF: ' . $e->getMessage());
8310 - }
8311 -}
8312 -
8313 -/**
8314 - * Legacy: Process a PDF URL inline during sitemap queue processing.
8315 - * @deprecated Use mxchat_expand_pdf_to_queue instead — kept for reference only.
8316 - */
8317 -private function mxchat_process_pdf_url_inline($pdf_url, $response, $api_key, $bot_id = 'default') {
8318 - set_time_limit(120); // PDFs need more time — downloading + parsing all pages
8319 -
8320 - $upload_dir = wp_upload_dir();
8321 - $pdf_filename = sanitize_file_name('mxchat_kb_' . md5($pdf_url) . '.pdf');
8322 - $pdf_path = trailingslashit($upload_dir['path']) . $pdf_filename;
8323 -
8324 - $response_body = wp_remote_retrieve_body($response);
8325 - if (empty($response_body)) {
8326 - return new WP_Error('empty_pdf', 'Empty PDF response');
8327 - }
8328 -
8329 - if (!wp_mkdir_p(dirname($pdf_path))) {
8330 - return new WP_Error('dir_error', 'Failed to create upload directory');
8331 - }
8332 -
8333 - file_put_contents($pdf_path, $response_body);
8334 -
8335 - if (!file_exists($pdf_path)) {
8336 - return new WP_Error('save_error', 'Failed to save PDF file');
8337 - }
8338 -
8339 - try {
8340 - mxchat_load_pdf_parser();
8341 - $parser = new \Smalot\PdfParser\Parser();
8342 - $pdf = $parser->parseFile($pdf_path);
8343 - $pages = $pdf->getPages();
8344 - $total_pages = count($pages);
8345 -
8346 - if ($total_pages < 1) {
8347 - wp_delete_file($pdf_path);
8348 - return new WP_Error('no_pages', 'PDF has no pages');
8349 - }
8350 -
8351 - $processed = 0;
8352 - $skipped_pages = array();
8353 -
8354 - for ($i = 0; $i < $total_pages; $i++) {
8355 - $page_num = $i + 1;
8356 - $text = $pages[$i]->getText();
8357 - if (empty($text)) {
8358 - $skipped_pages[] = 'Page ' . $page_num . ': No text could be extracted — page may contain only images, links, or non-standard encoding';
8359 - continue;
8360 - }
8361 -
8362 - $sanitized = $this->mxchat_sanitize_content_for_api($text);
8363 - if (empty($sanitized)) {
8364 - $skipped_pages[] = 'Page ' . $page_num . ': Text was extracted but contained only special characters, control codes, or unsupported content';
8365 - continue;
8366 - }
8367 -
8368 - $metadata = array(
8369 - 'document_type' => 'pdf',
8370 - 'total_pages' => $total_pages,
8371 - 'current_page' => $page_num,
8372 - 'source_url' => $pdf_url,
8373 - );
8374 -
8375 - $content_with_metadata = wp_json_encode($metadata) . "\n---\n" . $sanitized;
8376 - $page_url = esc_url($pdf_url . '#page=' . $page_num);
8377 -
8378 - MxChat_Utils::submit_content_to_db(
8379 - $content_with_metadata,
8380 - $page_url,
8381 - $api_key,
8382 - null,
8383 - $bot_id,
8384 - 'pdf'
8385 - );
8386 -
8387 - $processed++;
8388 - }
8389 -
8390 - // Clean up the temp PDF file
8391 - wp_delete_file($pdf_path);
8392 -
8393 - if (!empty($skipped_pages)) {
8394 - error_log('MxChat PDF: Skipped ' . count($skipped_pages) . ' of ' . $total_pages . ' pages: ' . implode('; ', $skipped_pages));
8395 - }
8396 -
8397 - return $processed > 0 ? true : false;
8398 -
8399 - } catch (Exception $e) {
8400 - if (file_exists($pdf_path)) {
8401 - wp_delete_file($pdf_path);
8402 - }
8403 - return new WP_Error('pdf_parse_error', 'Error parsing PDF: ' . $e->getMessage());
8404 - }
8405 -}
8406 -
8407 -/**
8408 - * Extract WooCommerce product content including pricing
8409 - *
8410 - * @param string $url The product URL
8411 - * @return string|false Product content with pricing, or false if not found
8412 - */
8413 -private function mxchat_extract_woocommerce_product_content($url) {
8414 - // Try to get product ID from URL
8415 - $product_id = url_to_postid($url);
8416 -
8417 - // If url_to_postid fails, try to extract from URL pattern
8418 - if (!$product_id) {
8419 - $product_slug = '';
8420 -
8421 - // Handle pretty permalinks: /product/product-name/
8422 - if (preg_match('/\/product\/([^\/\?]+)/', $url, $matches)) {
8423 - $product_slug = $matches[1];
8424 - }
8425 -
8426 - if (!empty($product_slug)) {
8427 - $product_post = get_page_by_path($product_slug, OBJECT, 'product');
8428 - if ($product_post) {
8429 - $product_id = $product_post->ID;
8430 - }
8431 - }
8432 - }
8433 -
8434 - if (!$product_id) {
8435 - return false;
8436 - }
8437 -
8438 - // Get WooCommerce product object
8439 - $product = wc_get_product($product_id);
8440 -
8441 - if (!$product) {
8442 - return false;
8443 - }
8444 -
8445 - // Build product content with pricing (similar to mxchat_store_product_embedding)
8446 - $title = $product->get_name();
8447 - $description = $product->get_description();
8448 - $short_description = $product->get_short_description();
8449 - $sku = $product->get_sku();
8450 -
8451 - // Get pricing information
8452 - $regular_price = $product->get_regular_price();
8453 - $sale_price = $product->get_sale_price();
8454 - $price = $product->get_price(); // Current active price
8455 -
8456 - // Get currency symbol
8457 - $currency_symbol = get_woocommerce_currency_symbol();
8458 -
8459 - // Format content
8460 - $content = $title . "\n\n";
8461 -
8462 - if (!empty($short_description)) {
8463 - $content .= "Short Description: " . wp_strip_all_tags($short_description) . "\n\n";
8464 - }
8465 -
8466 - if (!empty($description)) {
8467 - $content .= wp_strip_all_tags($description) . "\n\n";
8468 - }
8469 -
8470 - // Add pricing information
8471 - if (!empty($regular_price)) {
8472 - $content .= "Price: " . $currency_symbol . $regular_price . "\n";
8473 - } elseif (!empty($price)) {
8474 - $content .= "Price: " . $currency_symbol . $price . "\n";
8475 - }
8476 -
8477 - if (!empty($sale_price) && $sale_price !== $regular_price) {
8478 - $content .= "Sale Price: " . $currency_symbol . $sale_price . "\n";
8479 - }
8480 -
8481 - // Handle variable products - show price range
8482 - if ($product->is_type('variable')) {
8483 - $min_price = $product->get_variation_price('min');
8484 - $max_price = $product->get_variation_price('max');
8485 - if ($min_price !== $max_price) {
8486 - $content .= "Price Range: " . $currency_symbol . $min_price . " - " . $currency_symbol . $max_price . "\n";
8487 - }
8488 - }
8489 -
8490 - if (!empty($sku)) {
8491 - $content .= "SKU: " . $sku . "\n";
8492 - }
8493 -
8494 - // Get product categories
8495 - $categories = wp_get_post_terms($product_id, 'product_cat', array('fields' => 'names'));
8496 - if (!empty($categories) && !is_wp_error($categories)) {
8497 - $content .= "Categories: " . implode(', ', $categories) . "\n";
8498 - }
8499 -
8500 - // Get Custom Product Tabs (supports "Custom Product Tabs for WooCommerce" by Code Parrots)
8501 - $custom_tabs = get_post_meta($product_id, 'yikes_woo_products_tabs', true);
8502 - if (!empty($custom_tabs) && is_array($custom_tabs)) {
8503 - foreach ($custom_tabs as $tab) {
8504 - $tab_title = isset($tab['title']) ? $tab['title'] : (isset($tab['tab_title']) ? $tab['tab_title'] : '');
8505 - $tab_content = isset($tab['content']) ? $tab['content'] : '';
8506 -
8507 - if (!empty($tab_title) && !empty($tab_content)) {
8508 - $content .= "\n" . $tab_title . ": " . wp_strip_all_tags($tab_content) . "\n";
8509 - }
8510 - }
8511 - }
8512 -
8513 - // Also check for reusable/saved tabs applied to this product
8514 - $applied_saved_tabs = get_post_meta($product_id, 'yikes_woo_reusable_products_tabs_applied', true);
8515 - if (!empty($applied_saved_tabs) && is_array($applied_saved_tabs)) {
8516 - $saved_tabs = get_option('yikes_woo_reusable_products_tabs', array());
8517 - if (!empty($saved_tabs) && is_array($saved_tabs)) {
8518 - foreach ($applied_saved_tabs as $saved_tab_id) {
8519 - if (isset($saved_tabs[$saved_tab_id])) {
8520 - $tab = $saved_tabs[$saved_tab_id];
8521 - $tab_title = isset($tab['title']) ? $tab['title'] : (isset($tab['tab_title']) ? $tab['tab_title'] : '');
8522 - $tab_content = isset($tab['content']) ? $tab['content'] : '';
8523 -
8524 - if (!empty($tab_title) && !empty($tab_content)) {
8525 - $content .= "\n" . $tab_title . ": " . wp_strip_all_tags($tab_content) . "\n";
8526 - }
8527 - }
8528 - }
8529 - }
8530 - }
8531 -
8532 - return $this->mxchat_sanitize_content_for_api($content);
8533 -}
8534 -
8535 -/**
8536 - * Process a PDF page from the queue
8537 - */
8538 -private function mxchat_process_queue_pdf_page($item_data, $bot_id = 'default') {
8539 - $pdf_path = isset($item_data['pdf_path']) ? $item_data['pdf_path'] : '';
8540 - $pdf_url = isset($item_data['pdf_url']) ? $item_data['pdf_url'] : '';
8541 - $page_number = isset($item_data['page_number']) ? absint($item_data['page_number']) : 0;
8542 - $total_pages = isset($item_data['total_pages']) ? absint($item_data['total_pages']) : 0;
8543 -
8544 - if (empty($pdf_path) || !file_exists($pdf_path)) {
8545 - return new WP_Error('pdf_not_found', 'PDF file not found: ' . $pdf_path);
8546 - }
8547 -
8548 - if ($page_number < 1) {
8549 - return new WP_Error('invalid_page', 'Invalid page number');
8550 - }
8551 -
8552 - try {
8553 - mxchat_load_pdf_parser();
8554 - $parser = new \Smalot\PdfParser\Parser();
8555 - $pdf = $parser->parseFile($pdf_path);
8556 - $pages = $pdf->getPages();
8557 -
8558 - if (!isset($pages[$page_number - 1])) {
8559 - return new WP_Error('page_not_found', 'Page ' . $page_number . ' not found in PDF');
8560 - }
8561 -
8562 - $text = $pages[$page_number - 1]->getText();
8563 -
8564 - if (empty($text)) {
8565 - return new WP_Error('empty_page', 'Page ' . $page_number . ': No text could be extracted — page may contain only images, links, or non-standard encoding');
8566 - }
8567 -
8568 - $sanitized = $this->mxchat_sanitize_content_for_api($text);
8569 -
8570 - if (empty($sanitized)) {
8571 - return new WP_Error('empty_after_sanitization', 'Page ' . $page_number . ': Text was extracted but contained only special characters, control codes, or unsupported content that was removed during cleanup');
8572 - }
8573 -
8574 - // Create metadata
8575 - $metadata = array(
8576 - 'document_type' => 'pdf',
8577 - 'total_pages' => $total_pages,
8578 - 'current_page' => $page_number,
8579 - 'source_url' => $pdf_url
8580 - );
8581 -
8582 - $content_with_metadata = wp_json_encode($metadata) . "\n---\n" . $sanitized;
8583 - $page_url = esc_url($pdf_url . "#page=" . $page_number);
8584 -
8585 - // Get bot-specific API key
8586 - $bot_options = $this->get_bot_options($bot_id);
8587 - $options = !empty($bot_options) ? $bot_options : get_option('mxchat_options');
8588 - $selected_model = $options['embedding_model'] ?? 'text-embedding-ada-002';
8589 -
8590 - if (strpos($selected_model, 'voyage') === 0) {
8591 - $api_key = $options['voyage_api_key'] ?? '';
8592 - } elseif (strpos($selected_model, 'gemini-embedding') === 0) {
8593 - $api_key = $options['gemini_api_key'] ?? '';
8594 - } else {
8595 - $api_key = $options['api_key'] ?? '';
8596 - }
8597 -
8598 - if (empty($api_key)) {
8599 - return new WP_Error('no_api_key', 'API key not configured for bot: ' . $bot_id);
8600 - }
8601 -
8602 - // Submit to database - UPDATED 2.5.6: Added content_type 'pdf'
8603 - $result = MxChat_Utils::submit_content_to_db(
8604 - $content_with_metadata,
8605 - $page_url,
8606 - $api_key,
8607 - null,
8608 - $bot_id,
8609 - 'pdf'
8610 - );
8611 -
8612 - return $result;
8613 -
8614 - } catch (Exception $e) {
8615 - return new WP_Error('pdf_parse_error', 'Error parsing PDF: ' . $e->getMessage());
8616 - }
8617 -}
8618 -
8619 -/**
8620 - * AJAX: Get queue processing status
8621 - */
8622 -public function ajax_mxchat_get_queue_status() {
8623 - // Verify nonce and permissions
8624 - check_ajax_referer('mxchat_queue_nonce', 'nonce');
8625 -
8626 - if (!current_user_can('manage_options')) {
8627 - wp_send_json_error('Unauthorized access');
8628 - }
8629 -
8630 - $queue_id = isset($_POST['queue_id']) ? sanitize_text_field($_POST['queue_id']) : '';
8631 -
8632 - if (empty($queue_id)) {
8633 - wp_send_json_error('Missing queue ID');
8634 - }
8635 -
8636 - global $wpdb;
8637 - $table_name = $wpdb->prefix . 'mxchat_processing_queue';
8638 -
8639 - // Get counts by status
8640 - $counts = $wpdb->get_results($wpdb->prepare(
8641 - "SELECT status, COUNT(*) as count
8642 - FROM $table_name
8643 - WHERE queue_id = %s
8644 - GROUP BY status",
8645 - $queue_id
8646 - ), OBJECT_K);
8647 -
8648 - $total = 0;
8649 - $completed = 0;
8650 - $failed = 0;
8651 - $processing = 0;
8652 - $pending = 0;
8653 -
8654 - foreach ($counts as $status => $data) {
8655 - $count = absint($data->count);
8656 - $total += $count;
8657 -
8658 - switch ($status) {
8659 - case 'completed':
8660 - $completed = $count;
8661 - break;
8662 - case 'failed':
8663 - $failed = $count;
8664 - break;
8665 - case 'processing':
8666 - $processing = $count;
8667 - break;
8668 - case 'pending':
8669 - $pending = $count;
8670 - break;
8671 - }
8672 - }
8673 -
8674 - // Calculate percentage
8675 - $percentage = $total > 0 ? round((($completed + $failed) / $total) * 100) : 0;
8676 -
8677 - // Get failed items details (include all failed items, not just those that exhausted retries)
8678 - $failed_items = array();
8679 - if ($failed > 0) {
8680 - $failed_items = $wpdb->get_results($wpdb->prepare(
8681 - "SELECT item_type, item_data, error_message, attempts
8682 - FROM $table_name
8683 - WHERE queue_id = %s
8684 - AND status = 'failed'
8685 - ORDER BY id DESC
8686 - LIMIT 50",
8687 - $queue_id
8688 - ));
8689 - }
8690 -
8691 - // Get queue metadata
8692 - $source_url = $this->mxchat_get_queue_meta($queue_id, 'source_url');
8693 - $queue_type = $this->mxchat_get_queue_meta($queue_id, 'queue_type');
8694 -
8695 - // Determine if queue is complete
8696 - $is_complete = ($pending === 0 && $processing === 0);
8697 -
8698 - wp_send_json_success(array(
8699 - 'queue_id' => $queue_id,
8700 - 'queue_type' => $queue_type,
8701 - 'source_url' => $source_url,
8702 - 'total' => $total,
8703 - 'completed' => $completed,
8704 - 'failed' => $failed,
8705 - 'processing' => $processing,
8706 - 'pending' => $pending,
8707 - 'percentage' => $percentage,
8708 - 'is_complete' => $is_complete,
8709 - 'failed_items' => $failed_items,
8710 - 'status' => $is_complete ? 'complete' : 'processing'
8711 - ));
8712 -}
8713 -
8714 -/**
8715 - * AJAX: Clear completed queue
8716 - */
8717 -public function ajax_mxchat_clear_queue() {
8718 - // Verify nonce and permissions
8719 - check_ajax_referer('mxchat_queue_nonce', 'nonce');
8720 -
8721 - if (!current_user_can('manage_options')) {
8722 - wp_send_json_error('Unauthorized access');
8723 - }
8724 -
8725 - $queue_id = isset($_POST['queue_id']) ? sanitize_text_field($_POST['queue_id']) : '';
8726 -
8727 - if (empty($queue_id)) {
8728 - wp_send_json_error('Missing queue ID');
8729 - }
8730 -
8731 - global $wpdb;
8732 - $table_name = $wpdb->prefix . 'mxchat_processing_queue';
8733 - $meta_table = $wpdb->prefix . 'mxchat_queue_meta';
8734 -
8735 - // Delete queue items
8736 - $wpdb->delete(
8737 - $table_name,
8738 - array('queue_id' => $queue_id),
8739 - array('%s')
8740 - );
8741 -
8742 - // Delete queue metadata
8743 - $wpdb->delete(
8744 - $meta_table,
8745 - array('queue_id' => $queue_id),
8746 - array('%s')
8747 - );
8748 -
8749 - wp_send_json_success(array(
8750 - 'message' => 'Queue cleared successfully'
8751 - ));
8752 -}
8753 -
8754 -/**
8755 - * AJAX: Retry failed items in queue
8756 - */
8757 -public function ajax_mxchat_retry_failed() {
8758 - // Verify nonce and permissions
8759 - check_ajax_referer('mxchat_queue_nonce', 'nonce');
8760 -
8761 - if (!current_user_can('manage_options')) {
8762 - wp_send_json_error('Unauthorized access');
8763 - }
8764 -
8765 - $queue_id = isset($_POST['queue_id']) ? sanitize_text_field($_POST['queue_id']) : '';
8766 -
8767 - if (empty($queue_id)) {
8768 - wp_send_json_error('Missing queue ID');
8769 - }
8770 -
8771 - global $wpdb;
8772 - $table_name = $wpdb->prefix . 'mxchat_processing_queue';
8773 -
8774 - // Reset failed items to pending and reset attempt count
8775 - $updated = $wpdb->update(
8776 - $table_name,
8777 - array(
8778 - 'status' => 'pending',
8779 - 'attempts' => 0,
8780 - 'error_message' => null
8781 - ),
8782 - array(
8783 - 'queue_id' => $queue_id,
8784 - 'status' => 'failed'
8785 - ),
8786 - array('%s', '%d', '%s'),
8787 - array('%s', '%s')
8788 - );
8789 -
8790 - wp_send_json_success(array(
8791 - 'message' => 'Reset ' . $updated . ' failed items for retry',
8792 - 'reset_count' => $updated
8793 - ));
8794 -}
8795 -
8796 -
8797 -public function ajax_mxchat_mark_queue_complete() {
8798 - check_ajax_referer('mxchat_queue_nonce', 'nonce');
8799 -
8800 - if (!current_user_can('manage_options')) {
8801 - wp_send_json_error('Unauthorized access');
8802 - }
8803 -
8804 - $queue_id = isset($_POST['queue_id']) ? sanitize_text_field($_POST['queue_id']) : '';
8805 -
8806 - if (empty($queue_id)) {
8807 - wp_send_json_error('Missing queue ID');
8808 - }
8809 -
8810 - // Clear active queue transients
8811 - if (strpos($queue_id, 'sitemap_') === 0) {
8812 - delete_transient('mxchat_active_queue_sitemap');
8813 - } else if (strpos($queue_id, 'pdf_') === 0) {
8814 - delete_transient('mxchat_active_queue_pdf');
8815 - }
8816 -
8817 - wp_send_json_success(array('message' => 'Queue marked as complete'));
8818 -}
8819 -
8820 -
8821 - // ========================================
8822 - // STATIC ACCESS METHODS
8823 - // ========================================
8824 -
8825 - /**
8826 - * Get singleton instance
8827 - */
8828 - public static function get_instance() {
8829 - static $instance = null;
8830 - if ($instance === null) {
8831 - $instance = new self();
8832 - }
8833 - return $instance;
8834 - }
8835 -}
8836 -
8837 -// Initialize the Knowledge manager
1 +<?php
2 +/**
3 + * File: admin/class-knowledge-manager.php
4 + *
5 + * Handles all knowledge base content processing for MxChat
6 + * Including PDF, sitemap, content processing, and WordPress post management
7 + */
8 +if (!defined('ABSPATH')) {
9 + exit; // Exit if accessed directly
10 +}
11 +
12 +class MxChat_Knowledge_Manager {
13 +
14 + private $options;
15 +
16 + /**
17 + * Constructor - Register hooks for content processing
18 + */
19 +public function __construct() {
20 + $this->options = get_option('mxchat_options', array());
21 + $this->mxchat_init_hooks();
22 +
23 + $this->mxchat_init_role_hooks();
24 +}
25 +
26 +/**
27 + * Initialize WordPress hooks for content processing
28 + *
29 + */
30 +private function mxchat_init_hooks() {
31 + // Admin post handlers for form submissions
32 + add_action('admin_post_mxchat_submit_content', array($this, 'mxchat_handle_content_submission'));
33 + add_action('admin_post_mxchat_submit_sitemap', array($this, 'mxchat_handle_sitemap_submission'));
34 + add_action('admin_post_mxchat_stop_processing', array($this, 'mxchat_stop_processing'));
35 +
36 + // AJAX handlers for real-time processing and status updates
37 + add_action('wp_ajax_mxchat_get_status_updates', array($this, 'mxchat_ajax_get_status_updates'));
38 + add_action('wp_ajax_mxchat_dismiss_completed_status', array($this, 'mxchat_ajax_dismiss_completed_status'));
39 + add_action('wp_ajax_mxchat_get_content_list', array($this, 'ajax_mxchat_get_content_list'));
40 + add_action('wp_ajax_mxchat_process_selected_content', array($this, 'ajax_mxchat_process_selected_content'));
41 + add_action('wp_ajax_mxchat_save_inline_prompt', array($this, 'mxchat_save_inline_prompt'));
42 + add_action('admin_post_mxchat_delete_pinecone_prompt', array($this, 'mxchat_handle_pinecone_prompt_delete'));
43 + add_action('wp_ajax_mxchat_delete_pinecone_prompt', array($this, 'ajax_mxchat_delete_pinecone_prompt'));
44 + add_action('wp_ajax_mxchat_update_role_restriction', array($this, 'ajax_mxchat_update_role_restriction'));
45 +
46 + // Queue-based processing AJAX handlers
47 + add_action('wp_ajax_mxchat_get_next_queue_item', array($this, 'ajax_mxchat_get_next_queue_item'));
48 + add_action('wp_ajax_mxchat_process_queue_item', array($this, 'ajax_mxchat_process_queue_item'));
49 + add_action('wp_ajax_mxchat_get_queue_status', array($this, 'ajax_mxchat_get_queue_status'));
50 + add_action('wp_ajax_mxchat_clear_queue', array($this, 'ajax_mxchat_clear_queue'));
51 + add_action('wp_ajax_mxchat_retry_failed', array($this, 'ajax_mxchat_retry_failed'));
52 +
53 + // Hook for content deletion
54 + add_action('mxchat_delete_content', array($this, 'mxchat_delete_from_pinecone_by_url'), 10, 1);
55 +
56 + // WordPress post management hooks
57 + add_action('pre_post_update', array($this, 'mxchat_store_pre_update_status'), 10, 2);
58 + add_action('post_updated', array($this, 'mxchat_handle_post_update'), 10, 3);
59 + add_action('before_delete_post', array($this, 'mxchat_handle_post_delete'));
60 + add_action('wp_trash_post', array($this, 'mxchat_handle_post_delete'));
61 +
62 + add_action('wp_ajax_mxchat_mark_queue_complete', array($this, 'ajax_mxchat_mark_queue_complete'));
63 +
64 + // WooCommerce product hooks (if WooCommerce is active)
65 + if (class_exists('WooCommerce')) {
66 + add_action('pre_post_update', array($this, 'mxchat_store_pre_update_status'), 10, 2);
67 + add_action('save_post_product', array($this, 'mxchat_handle_product_change'), 10, 3);
68 + add_action('wp_trash_post', array($this, 'mxchat_handle_product_delete'));
69 + add_action('before_delete_post', array($this, 'mxchat_handle_product_delete'));
70 + }
71 +}
72 +
73 + /**
74 + * Get current options (refreshed)
75 + */
76 + private function mxchat_get_options() {
77 + if (empty($this->options)) {
78 + $this->options = get_option('mxchat_options', array());
79 + }
80 + return $this->options;
81 + }
82 +
83 +
84 + // ========================================
85 + // MAIN CONTENT SUBMISSION HANDLERS
86 + // ========================================
87 +
88 +public function mxchat_handle_content_submission() {
89 + // Check if the form was submitted and the user has permission.
90 + if (!isset($_POST['submit_content']) || !current_user_can('manage_options')) {
91 + return;
92 + }
93 +
94 + // Verify the nonce.
95 + $nonce = isset($_POST['mxchat_submit_content_nonce']) ? sanitize_text_field(wp_unslash($_POST['mxchat_submit_content_nonce'])) : '';
96 + if (!wp_verify_nonce($nonce, 'mxchat_submit_content_action')) {
97 + wp_die(esc_html__('Nonce verification failed.', 'mxchat'));
98 + }
99 +
100 + // Sanitize the inputs.
101 + // Use wp_kses_post to allow safe HTML and wp_unslash to remove WordPress added slashes
102 + $article_content = wp_kses_post(wp_unslash($_POST['article_content']));
103 + $article_url = isset($_POST['article_url']) ? esc_url_raw($_POST['article_url']) : '';
104 +
105 + // Get bot_id from form submission
106 + $bot_id = isset($_POST['bot_id']) ? sanitize_key($_POST['bot_id']) : 'default';
107 +
108 + // Get bot-specific options and API key
109 + $bot_options = $this->get_bot_options($bot_id);
110 + $options = !empty($bot_options) ? $bot_options : get_option('mxchat_options');
111 + $selected_model = $options['embedding_model'] ?? 'text-embedding-ada-002';
112 +
113 + if (strpos($selected_model, 'voyage') === 0) {
114 + $api_key = $options['voyage_api_key'] ?? '';
115 + } elseif (strpos($selected_model, 'gemini-embedding') === 0) {
116 + $api_key = $options['gemini_api_key'] ?? '';
117 + } else {
118 + $api_key = $options['api_key'] ?? '';
119 + }
120 +
121 + if (empty($api_key)) {
122 + set_transient('mxchat_admin_notice_error',
123 + esc_html__('API key is not configured. Please add your API key in the settings before submitting content.', 'mxchat'),
124 + 30
125 + );
126 + wp_safe_redirect(esc_url(admin_url('admin.php?page=mxchat-prompts')));
127 + exit;
128 + }
129 +
130 + // Use centralized utility function with bot_id
131 + $result = MxChat_Utils::submit_content_to_db($article_content, $article_url, $api_key, null, $bot_id);
132 +
133 + if (is_wp_error($result)) {
134 + set_transient('mxchat_admin_notice_error',
135 + esc_html__('Error storing content: ', 'mxchat') . $result->get_error_message(),
136 + 30
137 + );
138 + } else {
139 + set_transient('mxchat_admin_notice_success',
140 + esc_html__('Content successfully submitted!', 'mxchat'),
141 + 30
142 + );
143 + }
144 +
145 + wp_safe_redirect(esc_url(admin_url('admin.php?page=mxchat-prompts')));
146 + exit;
147 +}
148 +
149 +public function mxchat_is_pdf_url($url, $response) {
150 + $content_type = wp_remote_retrieve_header($response, 'content-type');
151 + $file_extension = strtolower(pathinfo($url, PATHINFO_EXTENSION));
152 +
153 + return strpos($content_type, 'pdf') !== false || $file_extension === 'pdf';
154 +}
155 +
156 +
157 +public function mxchat_handle_pdf_for_knowledge_base($pdf_url, $response, $bot_id = 'default') {
158 + if (!current_user_can('manage_options')) {
159 + return false;
160 + }
161 +
162 + $pdf_url = esc_url_raw($pdf_url);
163 + $upload_dir = wp_upload_dir();
164 +
165 + if (isset($upload_dir['error']) && $upload_dir['error'] !== false) {
166 + return false;
167 + }
168 +
169 + $pdf_filename = sanitize_file_name('mxchat_kb_' . time() . '.pdf');
170 + $pdf_path = trailingslashit($upload_dir['path']) . $pdf_filename;
171 +
172 + $response_body = wp_remote_retrieve_body($response);
173 + if (empty($response_body)) {
174 + return false;
175 + }
176 +
177 + if (!wp_mkdir_p(dirname($pdf_path))) {
178 + return false;
179 + }
180 +
181 + try {
182 + file_put_contents($pdf_path, $response_body);
183 +
184 + if (!file_exists($pdf_path)) {
185 + throw new Exception(__('Failed to save PDF file', 'mxchat'));
186 + }
187 +
188 + $total_pages = $this->mxchat_validate_and_count_pdf_pages($pdf_path);
189 +
190 + if ($total_pages === false || $total_pages < 1) {
191 + throw new Exception(__('Invalid PDF: Unable to parse or no pages found', 'mxchat'));
192 + }
193 +
194 + // Create unique queue ID
195 + $queue_id = 'pdf_' . md5($pdf_url . time());
196 +
197 + // Create array of pages to process
198 + $pages = array();
199 + for ($i = 1; $i <= $total_pages; $i++) {
200 + $pages[] = array(
201 + 'pdf_path' => $pdf_path,
202 + 'pdf_url' => $pdf_url,
203 + 'page_number' => $i,
204 + 'total_pages' => $total_pages
205 + );
206 + }
207 +
208 + // Add pages to queue
209 + $queued_count = $this->mxchat_add_to_queue($queue_id, 'pdf_page', $pages, $bot_id);
210 +
211 + if ($queued_count === 0) {
212 + wp_delete_file($pdf_path);
213 + throw new Exception(__('Failed to add PDF pages to processing queue', 'mxchat'));
214 + }
215 +
216 + // Store queue metadata
217 + $this->mxchat_set_queue_meta($queue_id, 'source_url', $pdf_url);
218 + $this->mxchat_set_queue_meta($queue_id, 'queue_type', 'pdf');
219 + $this->mxchat_set_queue_meta($queue_id, 'total_items', $total_pages);
220 + $this->mxchat_set_queue_meta($queue_id, 'bot_id', $bot_id);
221 + $this->mxchat_set_queue_meta($queue_id, 'pdf_path', $pdf_path);
222 + $this->mxchat_set_queue_meta($queue_id, 'created_at', current_time('mysql'));
223 +
224 + // Store queue ID in transient for status tracking
225 + set_transient('mxchat_active_queue_pdf', $queue_id, DAY_IN_SECONDS);
226 + set_transient('mxchat_last_pdf_url', $pdf_url, DAY_IN_SECONDS);
227 +
228 + return 'queued';
229 +
230 + } catch (Exception $e) {
231 + if (file_exists($pdf_path)) {
232 + wp_delete_file($pdf_path);
233 + }
234 + return $e->getMessage();
235 + }
236 +}
237 +
238 +/**
239 + * Validate PDF and count pages with multiple parser attempts
240 + */
241 +private function mxchat_validate_and_count_pdf_pages($pdf_path) {
242 + // Method 1: Try with Smalot PDF Parser (your current method)
243 + try {
244 + $parser = new \Smalot\PdfParser\Parser();
245 + $pdf = $parser->parseFile($pdf_path);
246 + $pages = $pdf->getPages();
247 + $page_count = count($pages);
248 +
249 + if ($page_count > 0) {
250 + //error_log('PDF parsed successfully with Smalot parser: ' . $page_count . ' pages');
251 + return $page_count;
252 + }
253 + } catch (Exception $e) {
254 + //error_log('Smalot PDF parser failed: ' . $e->getMessage());
255 + }
256 +
257 + // Method 2: Try with pdfinfo command (if available)
258 + if (function_exists('shell_exec') && !$this->mxchat_is_shell_disabled()) {
259 + try {
260 + $command = 'pdfinfo ' . escapeshellarg($pdf_path) . ' 2>&1';
261 + $output = shell_exec($command);
262 +
263 + if ($output && preg_match('/Pages:\s*(\d+)/', $output, $matches)) {
264 + $page_count = intval($matches[1]);
265 + if ($page_count > 0) {
266 + //error_log('PDF parsed successfully with pdfinfo: ' . $page_count . ' pages');
267 + return $page_count;
268 + }
269 + }
270 + } catch (Exception $e) {
271 + //error_log('pdfinfo command failed: ' . $e->getMessage());
272 + }
273 + }
274 +
275 + // Method 3: Try to repair PDF and parse again
276 + try {
277 + $repaired_path = $this->mxchat_attempt_pdf_repair($pdf_path);
278 + if ($repaired_path && $repaired_path !== $pdf_path) {
279 + $parser = new \Smalot\PdfParser\Parser();
280 + $pdf = $parser->parseFile($repaired_path);
281 + $pages = $pdf->getPages();
282 + $page_count = count($pages);
283 +
284 + if ($page_count > 0) {
285 + // Replace original with repaired version
286 + copy($repaired_path, $pdf_path);
287 + unlink($repaired_path);
288 + //error_log('PDF repaired and parsed successfully: ' . $page_count . ' pages');
289 + return $page_count;
290 + }
291 +
292 + // Clean up repaired file if it didn't work
293 + unlink($repaired_path);
294 + }
295 + } catch (Exception $e) {
296 + //error_log('PDF repair attempt failed: ' . $e->getMessage());
297 + }
298 +
299 + // Method 4: Manual PDF structure analysis (basic page count)
300 + try {
301 + $page_count = $this->mxchat_manual_pdf_page_count($pdf_path);
302 + if ($page_count > 0) {
303 + //error_log('PDF page count determined manually: ' . $page_count . ' pages');
304 + return $page_count;
305 + }
306 + } catch (Exception $e) {
307 + //error_log('Manual PDF analysis failed: ' . $e->getMessage());
308 + }
309 +
310 + //error_log('All PDF parsing methods failed for: ' . $pdf_path);
311 + return false;
312 +}
313 +
314 +/**
315 + * Check if shell_exec is disabled
316 + */
317 +private function mxchat_is_shell_disabled() {
318 + $disabled = explode(',', ini_get('disable_functions'));
319 + return in_array('shell_exec', $disabled);
320 +}
321 +
322 +/**
323 + * Attempt to repair PDF using basic methods
324 + */
325 +private function mxchat_attempt_pdf_repair($pdf_path) {
326 + try {
327 + $content = file_get_contents($pdf_path);
328 + if (!$content) {
329 + return false;
330 + }
331 +
332 + // Check if PDF starts with proper header
333 + if (substr($content, 0, 4) !== '%PDF') {
334 + // Try to find PDF header in the content
335 + $header_pos = strpos($content, '%PDF');
336 + if ($header_pos !== false && $header_pos < 1024) {
337 + // Remove junk before PDF header
338 + $content = substr($content, $header_pos);
339 + $repaired_path = $pdf_path . '.repaired';
340 + file_put_contents($repaired_path, $content);
341 + return $repaired_path;
342 + }
343 + }
344 +
345 + // Check for EOF marker
346 + $content = rtrim($content);
347 + if (!preg_match('/%%EOF\s*$/', $content)) {
348 + // Add EOF marker if missing
349 + $content .= "\n%%EOF";
350 + $repaired_path = $pdf_path . '.repaired';
351 + file_put_contents($repaired_path, $content);
352 + return $repaired_path;
353 + }
354 +
355 + } catch (Exception $e) {
356 + //error_log('PDF repair error: ' . $e->getMessage());
357 + }
358 +
359 + return false;
360 +}
361 +
362 +/**
363 + * Manual PDF page counting by analyzing PDF structure
364 + */
365 +private function mxchat_manual_pdf_page_count($pdf_path) {
366 + try {
367 + $content = file_get_contents($pdf_path);
368 + if (!$content) {
369 + return 0;
370 + }
371 +
372 + // Method 1: Count /Type /Page objects
373 + $page_count = preg_match_all('/\/Type\s*\/Page[^s]/', $content);
374 + if ($page_count > 0) {
375 + return $page_count;
376 + }
377 +
378 + // Method 2: Look for /Count in pages object
379 + if (preg_match('/\/Type\s*\/Pages.*?\/Count\s+(\d+)/', $content, $matches)) {
380 + return intval($matches[1]);
381 + }
382 +
383 + // Method 3: Count page references
384 + $page_count = preg_match_all('/\d+\s+0\s+obj\s*<<[^>]*\/Type\s*\/Page/', $content);
385 + if ($page_count > 0) {
386 + return $page_count;
387 + }
388 +
389 + } catch (Exception $e) {
390 + //error_log('Manual PDF analysis error: ' . $e->getMessage());
391 + }
392 +
393 + return 0;
394 +}
395 +
396 +
397 +public function mxchat_save_inline_prompt() {
398 + // DEBUG: Log what we're receiving
399 + //error_log('=== MXCHAT DEBUG ===');
400 + //error_log('POST data: ' . print_r($_POST, true));
401 + //error_log('Nonce from POST: ' . ($_POST['_ajax_nonce'] ?? 'NOT FOUND'));
402 +
403 + // Check for nonce security
404 + check_ajax_referer('mxchat_save_inline_nonce', '_ajax_nonce');
405 +
406 + // If we get here, nonce passed
407 + //error_log('Nonce verification PASSED');
408 +
409 + // Verify permissions
410 + if (!current_user_can('manage_options')) {
411 + wp_send_json_error(esc_html__('Permission denied.', 'mxchat'));
412 + return;
413 + }
414 +
415 + global $wpdb;
416 + $table_name = $wpdb->prefix . 'mxchat_system_prompt_content';
417 +
418 + // Validate and sanitize input data
419 + $prompt_id = isset($_POST['id']) ? absint($_POST['id']) : 0;
420 + $article_content = isset($_POST['article_content']) ? wp_kses_post(wp_unslash($_POST['article_content'])) : '';
421 + $article_url = isset($_POST['article_url']) ? esc_url_raw($_POST['article_url']) : '';
422 +
423 + if ($prompt_id > 0 && !empty($article_content)) {
424 + // Re-generate the embedding vector for the updated content
425 + $embedding_vector = $this->mxchat_generate_embedding($article_content);
426 + if (is_array($embedding_vector)) {
427 + // Serialize the embedding vector before storing it
428 + $embedding_vector_serialized = serialize($embedding_vector);
429 + // Update the prompt in the database
430 + $updated = $wpdb->update(
431 + $table_name,
432 + array(
433 + 'article_content' => $article_content,
434 + 'embedding_vector' => $embedding_vector_serialized,
435 + 'source_url' => $article_url,
436 + ),
437 + array('id' => $prompt_id),
438 + array('%s', '%s', '%s'),
439 + array('%d')
440 + );
441 + if ($updated !== false) {
442 + wp_send_json_success();
443 + } else {
444 + wp_send_json_error(esc_html__('Database update failed.', 'mxchat'));
445 + }
446 + } else {
447 + wp_send_json_error(esc_html__('Embedding generation failed.', 'mxchat'));
448 + }
449 + } else {
450 + wp_send_json_error(esc_html__('Invalid data.', 'mxchat'));
451 + }
452 +}
453 +
454 +
455 +public function mxchat_get_pdf_processing_status($pdf_url) {
456 + $pdf_url = esc_url_raw($pdf_url);
457 + $status = get_transient(sanitize_key('mxchat_pdf_status_' . md5($pdf_url)));
458 +
459 + if (!$status || !is_array($status)) {
460 + return false;
461 + }
462 +
463 + // Check for stalled processing (no updates for 5 minutes)
464 + if ($status['status'] === 'processing' && (time() - absint($status['last_update'])) > 300) {
465 + $status['status'] = 'error';
466 + $status['error'] = __('PDF processing appears to be stalled. No updates for over 5 minutes.', 'mxchat');
467 +
468 + // Save the updated status
469 + set_transient(
470 + sanitize_key('mxchat_pdf_status_' . md5($pdf_url)),
471 + array_map('sanitize_text_field', $status),
472 + DAY_IN_SECONDS
473 + );
474 + }
475 +
476 + $result = array(
477 + 'total_pages' => absint($status['total_pages']),
478 + 'processed_pages' => absint($status['processed_pages']),
479 + 'failed_pages' => absint($status['failed_pages'] ?? 0),
480 + 'percentage' => ($status['total_pages'] > 0)
481 + ? round((absint($status['processed_pages']) / absint($status['total_pages'])) * 100)
482 + : 0,
483 + 'status' => sanitize_text_field($status['status']),
484 + 'last_update' => human_time_diff(absint($status['last_update']), time()) . ' ' . esc_html__('ago', 'mxchat'),
485 + 'failed_pages_list' => isset($status['failed_pages_list']) ? $status['failed_pages_list'] : array(),
486 + 'completion_summary' => isset($status['completion_summary']) ? $status['completion_summary'] : null
487 + );
488 +
489 + // Add error message if present
490 + if (isset($status['error']) && !empty($status['error'])) {
491 + $result['error'] = sanitize_text_field($status['error']);
492 + }
493 +
494 + return $result;
495 +}
496 +
497 +
498 +public function mxchat_handle_sitemap_submission() {
499 + // Check if the form was submitted and verify permissions
500 + if (!isset($_POST['submit_sitemap']) || !current_user_can('manage_options')) {
501 + wp_die(esc_html__('Unauthorized access', 'mxchat'));
502 + }
503 +
504 + // Verify nonce
505 + check_admin_referer('mxchat_submit_sitemap_action', 'mxchat_submit_sitemap_nonce');
506 +
507 + // Validate URL
508 + if (!isset($_POST['sitemap_url']) || empty($_POST['sitemap_url'])) {
509 + set_transient('mxchat_admin_notice_error',
510 + esc_html__('Please provide a valid URL.', 'mxchat'),
511 + 30
512 + );
513 + wp_safe_redirect(esc_url(admin_url('admin.php?page=mxchat-prompts')));
514 + exit;
515 + }
516 +
517 + $submitted_url = esc_url_raw($_POST['sitemap_url']);
518 +
519 + // Get bot_id from form submission
520 + $bot_id = isset($_POST['bot_id']) ? sanitize_key($_POST['bot_id']) : 'default';
521 +
522 + // Get bot-specific options and validate API key
523 + $bot_options = $this->get_bot_options($bot_id);
524 + $options = !empty($bot_options) ? $bot_options : get_option('mxchat_options');
525 + $selected_model = $options['embedding_model'] ?? 'text-embedding-ada-002';
526 +
527 + if (strpos($selected_model, 'voyage') === 0) {
528 + $api_key = $options['voyage_api_key'] ?? '';
529 + $provider_name = 'Voyage AI';
530 + } elseif (strpos($selected_model, 'gemini-embedding') === 0) {
531 + $api_key = $options['gemini_api_key'] ?? '';
532 + $provider_name = 'Google Gemini';
533 + } else {
534 + $api_key = $options['api_key'] ?? '';
535 + $provider_name = 'OpenAI';
536 + }
537 +
538 + if (empty($api_key)) {
539 + $error_message = sprintf(
540 + esc_html__('%s API key is not configured. Please add your API key in the settings before submitting content.', 'mxchat'),
541 + $provider_name
542 + );
543 + set_transient('mxchat_admin_notice_error', $error_message, 30);
544 + wp_safe_redirect(esc_url(admin_url('admin.php?page=mxchat-prompts')));
545 + exit;
546 + }
547 +
548 + // Fetch URL
549 + $response = wp_remote_get($submitted_url, array('timeout' => 30));
550 +
551 + if (is_wp_error($response) || wp_remote_retrieve_response_code($response) !== 200) {
552 + $error_message = is_wp_error($response) ? $response->get_error_message() : 'HTTP Status: ' . wp_remote_retrieve_response_code($response);
553 + set_transient('mxchat_admin_notice_error',
554 + sprintf(
555 + esc_html__('Failed to fetch the URL: %s', 'mxchat'),
556 + esc_html($error_message)
557 + ),
558 + 30
559 + );
560 + wp_safe_redirect(esc_url(admin_url('admin.php?page=mxchat-prompts')));
561 + exit;
562 + }
563 +
564 + $content_type = wp_remote_retrieve_header($response, 'content-type');
565 + $body_content = wp_remote_retrieve_body($response);
566 +
567 + if (empty($body_content)) {
568 + set_transient('mxchat_admin_notice_error',
569 + esc_html__('Empty response received from URL.', 'mxchat'),
570 + 30
571 + );
572 + wp_safe_redirect(esc_url(admin_url('admin.php?page=mxchat-prompts')));
573 + exit;
574 + }
575 +
576 + // Handle PDF URL
577 + if ($this->mxchat_is_pdf_url($submitted_url, $response)) {
578 + $result = $this->mxchat_handle_pdf_for_knowledge_base($submitted_url, $response, $bot_id);
579 +
580 + if ($result === 'queued') {
581 + set_transient('mxchat_admin_notice_success',
582 + esc_html__('PDF queued for processing. Processing will start automatically.', 'mxchat'),
583 + 30
584 + );
585 + } else {
586 + set_transient('mxchat_admin_notice_error',
587 + esc_html__('Failed to queue PDF processing: ', 'mxchat') . esc_html($result),
588 + 30
589 + );
590 + }
591 +
592 + wp_safe_redirect(esc_url(admin_url('admin.php?page=mxchat-prompts')));
593 + exit;
594 + }
595 +
596 + // Handle Sitemap XML
597 + if (strpos($content_type, 'xml') !== false || strpos($body_content, '<urlset') !== false) {
598 + libxml_use_internal_errors(true);
599 + $xml = simplexml_load_string($body_content);
600 + $xml_errors = libxml_get_errors();
601 + libxml_clear_errors();
602 +
603 + if ($xml === false || !empty($xml_errors)) {
604 + set_transient('mxchat_admin_notice_error',
605 + esc_html__('Invalid sitemap XML. Please provide a valid sitemap.', 'mxchat'),
606 + 30
607 + );
608 + wp_safe_redirect(esc_url(admin_url('admin.php?page=mxchat-prompts')));
609 + exit;
610 + }
611 +
612 + $result = $this->mxchat_handle_sitemap_for_knowledge_base($xml, $submitted_url, $bot_id);
613 +
614 + if ($result === 'queued') {
615 + set_transient('mxchat_admin_notice_success',
616 + esc_html__('Sitemap queued for processing. Processing will start automatically.', 'mxchat'),
617 + 30
618 + );
619 + } else {
620 + set_transient('mxchat_admin_notice_error',
621 + esc_html__('Failed to queue sitemap processing. Please check the status below for details.', 'mxchat'),
622 + 30
623 + );
624 + }
625 +
626 + wp_safe_redirect(esc_url(admin_url('admin.php?page=mxchat-prompts')));
627 + exit;
628 + }
629 +
630 + // Handle Regular URL (single page)
631 + $page_content = $this->mxchat_extract_main_content($body_content);
632 + $sanitized_content = $this->mxchat_sanitize_content_for_api($page_content);
633 +
634 + if (empty($sanitized_content)) {
635 + set_transient('mxchat_admin_notice_error',
636 + esc_html__('No valid content found on the provided URL.', 'mxchat'),
637 + 30
638 + );
639 + wp_safe_redirect(esc_url(admin_url('admin.php?page=mxchat-prompts')));
640 + exit;
641 + }
642 +
643 + // For single URLs, process immediately (not queued)
644 + $embedding_vector = $this->mxchat_generate_embedding($sanitized_content, $bot_id);
645 +
646 + if (is_string($embedding_vector)) {
647 + $error_message = esc_html__('Failed to generate embedding: ', 'mxchat') . esc_html($embedding_vector);
648 + set_transient('mxchat_admin_notice_error', $error_message, 30);
649 + wp_safe_redirect(esc_url(admin_url('admin.php?page=mxchat-prompts')));
650 + exit;
651 + }
652 +
653 + if (is_array($embedding_vector)) {
654 + $db_result = MxChat_Utils::submit_content_to_db(
655 + $sanitized_content,
656 + $submitted_url,
657 + $api_key,
658 + null,
659 + $bot_id
660 + );
661 +
662 + if (is_wp_error($db_result)) {
663 + $error_message = esc_html__('Failed to store content in database: ', 'mxchat') . esc_html($db_result->get_error_message());
664 + set_transient('mxchat_admin_notice_error', $error_message, 30);
665 + } else {
666 + $success_message = esc_html__('URL content successfully submitted!', 'mxchat');
667 + set_transient('mxchat_admin_notice_success', $success_message, 30);
668 + }
669 + } else {
670 + $error_message = esc_html__('Failed to generate embedding: Unexpected result type. Please check your API key and try again.', 'mxchat');
671 + set_transient('mxchat_admin_notice_error', $error_message, 30);
672 + }
673 +
674 + wp_safe_redirect(esc_url(admin_url('admin.php?page=mxchat-prompts')));
675 + exit;
676 +}
677 +
678 +
679 +public function mxchat_get_single_url_status() {
680 + $status = get_transient('mxchat_single_url_status');
681 + if (!$status) {
682 + return null;
683 + }
684 +
685 + // Add human-readable time
686 + if (isset($status['timestamp'])) {
687 + $status['human_time'] = human_time_diff(strtotime($status['timestamp']), current_time('timestamp')) . ' ' . __('ago', 'mxchat');
688 + }
689 +
690 + return $status;
691 +}
692 +
693 +public function mxchat_handle_sitemap_for_knowledge_base($xml, $sitemap_url, $bot_id = 'default') {
694 + if (!current_user_can('manage_options')) {
695 + return false;
696 + }
697 +
698 + try {
699 + $sitemap_url = esc_url_raw($sitemap_url);
700 +
701 + if (!$xml || !is_object($xml)) {
702 + throw new Exception(__('Invalid XML object provided', 'mxchat'));
703 + }
704 +
705 + // Get bot-specific embedding API for validation
706 + $bot_options = $this->get_bot_options($bot_id);
707 + $options = !empty($bot_options) ? $bot_options : get_option('mxchat_options');
708 +
709 + // Test the embedding API before processing
710 + $test_phrase = "Test embedding generation for MxChat";
711 + $test_result = $this->mxchat_generate_embedding($test_phrase, $bot_id);
712 +
713 + if (is_string($test_result)) {
714 + throw new Exception(__('Embedding API validation failed: ', 'mxchat') . $test_result);
715 + }
716 +
717 + if (!is_array($test_result)) {
718 + throw new Exception(__('Embedding API returned unexpected result type. Please check your configuration.', 'mxchat'));
719 + }
720 +
721 + // Extract URLs from sitemap
722 + $urls = array();
723 + foreach ($xml->url as $url_element) {
724 + $url = esc_url_raw((string)$url_element->loc);
725 + if ($url) {
726 + $urls[] = array('url' => $url);
727 + }
728 + }
729 +
730 + $total_urls = count($urls);
731 +
732 + if ($total_urls < 1) {
733 + throw new Exception(__('No valid URLs found in sitemap', 'mxchat'));
734 + }
735 +
736 + // Create unique queue ID
737 + $queue_id = 'sitemap_' . md5($sitemap_url . time());
738 +
739 + // Add URLs to queue
740 + $queued_count = $this->mxchat_add_to_queue($queue_id, 'url', $urls, $bot_id);
741 +
742 + if ($queued_count === 0) {
743 + throw new Exception(__('Failed to add URLs to processing queue', 'mxchat'));
744 + }
745 +
746 + // Store queue metadata
747 + $this->mxchat_set_queue_meta($queue_id, 'source_url', $sitemap_url);
748 + $this->mxchat_set_queue_meta($queue_id, 'queue_type', 'sitemap');
749 + $this->mxchat_set_queue_meta($queue_id, 'total_items', $total_urls);
750 + $this->mxchat_set_queue_meta($queue_id, 'bot_id', $bot_id);
751 + $this->mxchat_set_queue_meta($queue_id, 'created_at', current_time('mysql'));
752 +
753 + // Store queue ID in transient for status tracking
754 + set_transient('mxchat_active_queue_sitemap', $queue_id, DAY_IN_SECONDS);
755 + set_transient('mxchat_last_sitemap_url', $sitemap_url, DAY_IN_SECONDS);
756 +
757 + return 'queued';
758 +
759 + } catch (Exception $e) {
760 + $error_message = $e->getMessage();
761 + error_log(sprintf(esc_html__('Error preparing sitemap for processing: %s', 'mxchat'), esc_html($error_message)));
762 +
763 + return $error_message;
764 + }
765 +
766 +}
767 +
768 +public function mxchat_sanitize_content_for_api($content) {
769 + //error_log('[MXCHAT-SANITIZE] Original content preview: ' . substr($content, 0, 500) . '...');
770 +
771 + // Strip WordPress shortcodes FIRST (WPBakery, Elementor, Woodmart, etc.)
772 + // This must be done before stripping HTML tags, otherwise the brackets are removed
773 + $content = strip_shortcodes($content);
774 +
775 + // Additional regex-based shortcode removal as a safety net
776 + // This catches any remaining shortcodes that strip_shortcodes() might have missed
777 + $content = preg_replace('/\[(\[?)([a-zA-Z0-9_-]+)(?![\w-])([^\]\/]*(?:\/(?!\])[^\]\/]*)*?)(?:(\/)\]|\](?:([^\[]*(?:\[(?!\/\2\])[^\[]*)*)\[\/\2\])?)(\]?)/s', '', $content);
778 +
779 + // Remove script, style tags, and HTML comments
780 + $content = preg_replace('/<script\b[^>]*>(.*?)<\/script>/is', '', $content);
781 + $content = preg_replace('/<style\b[^>]*>(.*?)<\/style>/is', '', $content);
782 + $content = preg_replace('/<!--(.|\s)*?-->/', '', $content);
783 +
784 + // Remove all HTML tags and decode HTML entities
785 + $content = wp_strip_all_tags($content);
786 + $content = html_entity_decode($content, ENT_QUOTES | ENT_HTML5);
787 +
788 + // Normalize whitespace but preserve paragraph breaks
789 + // First, normalize line endings to \n
790 + $content = str_replace(["\r\n", "\r"], "\n", $content);
791 + // Replace multiple spaces/tabs with single space, but preserve newlines
792 + $content = preg_replace('/[ \t]+/', ' ', $content);
793 + // Replace 3+ newlines with 2 newlines (max 2 blank lines)
794 + $content = preg_replace('/\n{3,}/', "\n\n", $content);
795 + // Trim each line
796 + $lines = explode("\n", $content);
797 + $lines = array_map('trim', $lines);
798 + $content = implode("\n", $lines);
799 + // Final trim
800 + $content = trim($content);
801 +
802 + // Remove control characters (which can cause database issues) but preserve \n (0x0A) and \r (0x0D)
803 + $content = preg_replace('/[\x00-\x08\x0B\x0C\x0E-\x1F\x7F]/u', '', $content);
804 +
805 + // Remove NULL bytes which can cause database errors
806 + $content = str_replace("\0", "", $content);
807 +
808 + // Ensure valid UTF-8 encoding
809 + $content = wp_check_invalid_utf8($content);
810 +
811 + // Remove any extremely long strings without spaces (often garbage)
812 + $content = preg_replace('/\S{300,}/', ' ', $content);
813 +
814 + // Replace problematic characters that often cause database issues
815 + $content = preg_replace('/[\x{10000}-\x{10FFFF}]/u', '', $content); // Remove emoji and other high Unicode characters
816 +
817 + // Replace any remaining potentially problematic characters with spaces
818 + // BUT preserve newlines by temporarily replacing them
819 + $content = str_replace("\n", "NEWLINE_PLACEHOLDER", $content);
820 + $content = preg_replace('/[^\p{L}\p{N}\p{P}\p{Z}\p{Sm}]/u', ' ', $content);
821 + $content = str_replace("NEWLINE_PLACEHOLDER", "\n", $content);
822 +
823 + // Limit to reasonable length if needed
824 + $max_length = 65000; // Just under MySQL TEXT field limit
825 + if (strlen($content) > $max_length) {
826 + $content = substr($content, 0, $max_length);
827 + }
828 +
829 + //error_log('[MXCHAT-SANITIZE] Sanitized content preview: ' . substr($content, 0, 500) . '...');
830 + return $content;
831 +}
832 +public function mxchat_extract_main_content($html) {
833 + if (empty($html)) {
834 + return '';
835 + }
836 + try {
837 + $dom = new DOMDocument;
838 + libxml_use_internal_errors(true); // Suppress HTML parsing errors
839 + @$dom->loadHTML(mb_convert_encoding($html, 'HTML-ENTITIES', 'UTF-8'), LIBXML_HTML_NOIMPLIED | LIBXML_HTML_NODEFDTD);
840 + $xpath = new DOMXPath($dom);
841 +
842 + // For debugging purposes
843 + $debugEnabled = false; // Set to true to enable debugging output
844 + $debug = function($message) use ($debugEnabled) {
845 + if ($debugEnabled) {
846 + //error_log('[MXCHAT-DEBUG] ' . $message);
847 + }
848 + };
849 +
850 + // Direct targeting for Gerow theme posts
851 + $post_text = $xpath->query('//div[contains(@class, "post-text")]');
852 + if ($post_text && $post_text->length > 0) {
853 + $debug("Found post-text directly");
854 + $content = '';
855 + foreach ($post_text as $node) {
856 + $content .= $dom->saveHTML($node);
857 + }
858 + if (!empty($content)) {
859 + $debug("Returning post-text content");
860 + return $content;
861 + }
862 + }
863 +
864 + // Try to get the blog details content which contains the post-text
865 + $blog_details = $xpath->query('//div[contains(@class, "blog-details-content")]');
866 + if ($blog_details && $blog_details->length > 0) {
867 + $debug("Found blog-details-content");
868 + $content = '';
869 + foreach ($blog_details as $node) {
870 + $content .= $dom->saveHTML($node);
871 + }
872 + if (!empty($content)) {
873 + $debug("Returning blog-details-content");
874 + return $content;
875 + }
876 + }
877 +
878 + // Try to get the article which contains the blog details
879 + $article = $xpath->query('//article[contains(@class, "blog-details-wrap")]');
880 + if ($article && $article->length > 0) {
881 + $debug("Found article with blog-details-wrap");
882 + $content = '';
883 + foreach ($article as $node) {
884 + $content .= $dom->saveHTML($node);
885 + }
886 + if (!empty($content)) {
887 + $debug("Returning article content");
888 + return $content;
889 + }
890 + }
891 +
892 + // Try even broader with the blog-item-wrap
893 + $blog_item = $xpath->query('//div[contains(@class, "blog-item-wrap")]');
894 + if ($blog_item && $blog_item->length > 0) {
895 + $debug("Found blog-item-wrap");
896 + $content = '';
897 + foreach ($blog_item as $node) {
898 + $content .= $dom->saveHTML($node);
899 + }
900 + if (!empty($content)) {
901 + $debug("Returning blog-item-wrap content");
902 + return $content;
903 + }
904 + }
905 +
906 + // Specific Gerow theme path
907 + $gerow_path = $xpath->query('//section[contains(@class, "blog-area")]//div[contains(@class, "post-text")]');
908 + if ($gerow_path && $gerow_path->length > 0) {
909 + $debug("Found Gerow theme path to post-text");
910 + $content = '';
911 + foreach ($gerow_path as $node) {
912 + $content .= $dom->saveHTML($node);
913 + }
914 + if (!empty($content)) {
915 + $debug("Returning Gerow post-text content");
916 + return $content;
917 + }
918 + }
919 +
920 + // Generic blog post selectors
921 + $selectors = [
922 + // Blog post specific selectors
923 + '//div[contains(@class, "post-text")]',
924 + '//article[contains(@class, "blog-post-item")]//div[contains(@class, "post-text")]',
925 + '//div[contains(@class, "blog-details-content")]',
926 + '//article[contains(@class, "blog-details-wrap")]',
927 + '//div[contains(@class, "entry-content")]',
928 + '//div[contains(@class, "blog-content")]',
929 + '//div[contains(@class, "blog-item-wrap")]',
930 +
931 + // More general content selectors
932 + '//div[contains(@class, "page__content")]',
933 + '//div[contains(@class, "elementor-widget-container")]',
934 + '//div[contains(@class, "elementor-text-editor")]',
935 + '//div[contains(@class, "elementor-widget-text-editor")]',
936 + '//*[contains(@class, "entry-content")]',
937 + '//*[contains(@class, "post-content")]',
938 + '//*[contains(@class, "article-content")]',
939 + '//*[@id="content"]',
940 + '//*[@id="main-content"]',
941 + '//section[contains(@class, "blog-area")]',
942 + '//article',
943 + '//main',
944 + '//div[contains(@class, "content")]'
945 + ];
946 +
947 + // First handle Elementor content
948 + $debug("Checking for Elementor content");
949 + $elementor_widgets = $xpath->query('//div[contains(@class, "elementor-element")]//div[contains(@class, "elementor-widget-container")]');
950 + if ($elementor_widgets && $elementor_widgets->length > 0) {
951 + $debug("Found Elementor widgets");
952 + $combined_content = '';
953 + foreach ($elementor_widgets as $widget) {
954 + $widget_content = $dom->saveHTML($widget);
955 + if (!empty($widget_content)) {
956 + $combined_content .= $widget_content;
957 + }
958 + }
959 + if (!empty($combined_content)) {
960 + $debug("Returning Elementor content");
961 + return $combined_content;
962 + }
963 + }
964 +
965 + // Try standard selectors one by one
966 + foreach ($selectors as $selector) {
967 + $debug("Trying selector: " . $selector);
968 + $nodes = $xpath->query($selector);
969 + if ($nodes && $nodes->length > 0) {
970 + $debug("Found matches for selector: " . $selector);
971 + $content = '';
972 + foreach ($nodes as $node) {
973 + $content .= $dom->saveHTML($node);
974 + }
975 + if (!empty($content)) {
976 + $debug("Returning content from selector: " . $selector);
977 + return $content;
978 + }
979 + }
980 + }
981 +
982 + // Manual regex fallback for post-text if DOM methods fail
983 + $debug("Trying regex fallback");
984 + if (preg_match('/<div class="post-text">(.*?)<\/div>\s*<\/div>\s*<\/div>/s', $html, $matches)) {
985 + $debug("Found post-text via regex");
986 + return '<div class="post-text">' . $matches[1] . '</div>';
987 + }
988 +
989 + // Try to extract the blog section as a whole
990 + $blog_section = $xpath->query('//section[contains(@class, "blog-area")]');
991 + if ($blog_section && $blog_section->length > 0) {
992 + $debug("Found blog-area section");
993 + $content = '';
994 + foreach ($blog_section as $node) {
995 + $content .= $dom->saveHTML($node);
996 + }
997 + if (!empty($content)) {
998 + $debug("Returning blog-area section content");
999 + return $content;
1000 + }
1001 + }
1002 +
1003 + // Fallback: Return the body content if no specific selector matches
1004 + $debug("Using body fallback");
1005 + $body = $dom->getElementsByTagName('body');
1006 + if ($body->length > 0) {
1007 + return $dom->saveHTML($body->item(0));
1008 + }
1009 +
1010 + // Last resort: return the original HTML
1011 + $debug("Returning original HTML");
1012 + return $html;
1013 + } catch (Exception $e) {
1014 + //error_log('[MXCHAT-ERROR] Content extraction failed: ' . $e->getMessage());
1015 + return $html; // Return original HTML if parsing fails
1016 + } finally {
1017 + libxml_clear_errors();
1018 + }
1019 +}
1020 +public function mxchat_get_sitemap_processing_status($sitemap_url) {
1021 + $sitemap_url = esc_url_raw($sitemap_url);
1022 + $status_key = sanitize_key('mxchat_sitemap_status_' . md5($sitemap_url));
1023 + $status = get_transient($status_key);
1024 +
1025 + if (!$status || !is_array($status)) {
1026 + return false;
1027 + }
1028 +
1029 + // Auto-complete check: if all URLs are processed but status isn't complete
1030 + if (isset($status['processed_urls']) && isset($status['total_urls']) &&
1031 + $status['processed_urls'] >= $status['total_urls'] &&
1032 + isset($status['status']) && $status['status'] !== 'complete' &&
1033 + $status['status'] !== 'error') {
1034 +
1035 + // Mark as complete
1036 + $status['status'] = 'complete';
1037 + $status['processed_urls'] = $status['total_urls']; // Ensure exact match
1038 +
1039 + // Update the transient with the corrected status
1040 + set_transient($status_key, $status, DAY_IN_SECONDS);
1041 + }
1042 +
1043 + return array(
1044 + 'total_urls' => absint($status['total_urls']),
1045 + 'processed_urls' => absint($status['processed_urls']),
1046 + 'failed_urls' => absint($status['failed_urls'] ?? 0),
1047 + 'percentage' => ($status['total_urls'] > 0)
1048 + ? round((absint($status['processed_urls']) / absint($status['total_urls'])) * 100)
1049 + : 0,
1050 + 'status' => sanitize_text_field($status['status']),
1051 + 'last_update' => human_time_diff(absint($status['last_update']), time()) . ' ' . esc_html__('ago', 'mxchat'),
1052 + 'error' => isset($status['error']) ? sanitize_text_field($status['error']) : '',
1053 + 'last_error' => isset($status['last_error']) ? sanitize_text_field($status['last_error']) : '',
1054 + 'failed_urls_list' => isset($status['failed_urls_list']) ? $status['failed_urls_list'] : array()
1055 + );
1056 +}
1057 +
1058 +public function mxchat_ajax_get_status_updates() {
1059 + try {
1060 + // Verify the request
1061 + check_ajax_referer('mxchat_status_nonce', 'nonce');
1062 +
1063 + // Get active queue IDs
1064 + $sitemap_queue_id = get_transient('mxchat_active_queue_sitemap');
1065 + $pdf_queue_id = get_transient('mxchat_active_queue_pdf');
1066 +
1067 + $sitemap_status = false;
1068 + $pdf_status = false;
1069 +
1070 + // Get sitemap queue status
1071 + if ($sitemap_queue_id) {
1072 + $sitemap_status = $this->mxchat_get_queue_status_data($sitemap_queue_id, 'sitemap');
1073 + }
1074 +
1075 + // Get PDF queue status
1076 + if ($pdf_queue_id) {
1077 + $pdf_status = $this->mxchat_get_queue_status_data($pdf_queue_id, 'pdf');
1078 + }
1079 +
1080 + $is_active_processing =
1081 + ($sitemap_status && $sitemap_status['status'] === 'processing') ||
1082 + ($pdf_status && $pdf_status['status'] === 'processing');
1083 +
1084 + // Return JSON response with the status data
1085 + wp_send_json(array(
1086 + 'pdf_status' => $pdf_status,
1087 + 'sitemap_status' => $sitemap_status,
1088 + 'is_processing' => $is_active_processing,
1089 + 'sitemap_queue_id' => $sitemap_queue_id,
1090 + 'pdf_queue_id' => $pdf_queue_id
1091 + ));
1092 +
1093 + } catch (Exception $e) {
1094 + error_log('MxChat Status Update Error: ' . $e->getMessage());
1095 +
1096 + wp_send_json_error(array(
1097 + 'message' => 'Error getting status updates: ' . $e->getMessage(),
1098 + 'status' => 'error'
1099 + ));
1100 + }
1101 +}
1102 +
1103 +/**
1104 + * Helper function to get queue status data
1105 + */
1106 +private function mxchat_get_queue_status_data($queue_id, $type = 'sitemap') {
1107 + global $wpdb;
1108 + $table_name = $wpdb->prefix . 'mxchat_processing_queue';
1109 +
1110 + // Get counts by status
1111 + $counts = $wpdb->get_results($wpdb->prepare(
1112 + "SELECT status, COUNT(*) as count
1113 + FROM $table_name
1114 + WHERE queue_id = %s
1115 + GROUP BY status",
1116 + $queue_id
1117 + ), OBJECT_K);
1118 +
1119 + $total = 0;
1120 + $completed = 0;
1121 + $failed = 0;
1122 + $processing = 0;
1123 + $pending = 0;
1124 +
1125 + foreach ($counts as $status => $data) {
1126 + $count = absint($data->count);
1127 + $total += $count;
1128 +
1129 + switch ($status) {
1130 + case 'completed':
1131 + $completed = $count;
1132 + break;
1133 + case 'failed':
1134 + $failed = $count;
1135 + break;
1136 + case 'processing':
1137 + $processing = $count;
1138 + break;
1139 + case 'pending':
1140 + $pending = $count;
1141 + break;
1142 + }
1143 + }
1144 +
1145 + if ($total === 0) {
1146 + return false;
1147 + }
1148 +
1149 + // Calculate percentage
1150 + $percentage = round((($completed + $failed) / $total) * 100);
1151 +
1152 + // Get failed items details (limit to 50)
1153 + $failed_items = array();
1154 + if ($failed > 0) {
1155 + $failed_results = $wpdb->get_results($wpdb->prepare(
1156 + "SELECT item_type, item_data, error_message, attempts, completed_at
1157 + FROM $table_name
1158 + WHERE queue_id = %s
1159 + AND status = 'failed'
1160 + AND attempts >= max_attempts
1161 + ORDER BY id DESC
1162 + LIMIT 50",
1163 + $queue_id
1164 + ));
1165 +
1166 + foreach ($failed_results as $item) {
1167 + $data = json_decode($item->item_data, true);
1168 + $url = $type === 'pdf' ? ($data['pdf_url'] ?? '') : ($data['url'] ?? '');
1169 +
1170 + $failed_items[] = array(
1171 + 'url' => $url,
1172 + 'page' => $type === 'pdf' ? ($data['page_number'] ?? 0) : 0,
1173 + 'error' => $item->error_message,
1174 + 'retries' => $item->attempts,
1175 + 'time' => strtotime($item->completed_at)
1176 + );
1177 + }
1178 + }
1179 +
1180 + // Get queue metadata
1181 + $source_url = $this->mxchat_get_queue_meta($queue_id, 'source_url');
1182 +
1183 + // Determine if queue is complete
1184 + $is_complete = ($pending === 0 && $processing === 0);
1185 +
1186 + // Get last update time
1187 + $last_update = $wpdb->get_var($wpdb->prepare(
1188 + "SELECT MAX(UNIX_TIMESTAMP(COALESCE(completed_at, started_at, created_at)))
1189 + FROM $table_name
1190 + WHERE queue_id = %s",
1191 + $queue_id
1192 + ));
1193 +
1194 + $last_update_text = $last_update ? human_time_diff($last_update, time()) . ' ' . __('ago', 'mxchat') : __('Just now', 'mxchat');
1195 +
1196 + // Format based on type
1197 + if ($type === 'pdf') {
1198 + return array(
1199 + 'total_pages' => $total,
1200 + 'processed_pages' => $completed + $failed,
1201 + 'failed_pages' => $failed,
1202 + 'percentage' => $percentage,
1203 + 'status' => $is_complete ? 'complete' : 'processing',
1204 + 'last_update' => $last_update_text,
1205 + 'failed_pages_list' => $failed_items,
1206 + 'pdf_url' => $source_url,
1207 + 'queue_id' => $queue_id
1208 + );
1209 + } else {
1210 + return array(
1211 + 'total_urls' => $total,
1212 + 'processed_urls' => $completed + $failed,
1213 + 'failed_urls' => $failed,
1214 + 'percentage' => $percentage,
1215 + 'status' => $is_complete ? 'complete' : 'processing',
1216 + 'last_update' => $last_update_text,
1217 + 'failed_urls_list' => $failed_items,
1218 + 'sitemap_url' => $source_url,
1219 + 'queue_id' => $queue_id
1220 + );
1221 + }
1222 +}
1223 +
1224 +public function mxchat_stop_processing() {
1225 + // Verify permissions
1226 + if (!current_user_can('manage_options')) {
1227 + wp_die(esc_html__('Unauthorized access', 'mxchat'));
1228 + }
1229 +
1230 + // Verify nonce
1231 + check_admin_referer('mxchat_stop_processing_action', 'mxchat_stop_processing_nonce');
1232 +
1233 + global $wpdb;
1234 + $table_name = $wpdb->prefix . 'mxchat_processing_queue';
1235 +
1236 + // Get active queue IDs
1237 + $sitemap_queue_id = get_transient('mxchat_active_queue_sitemap');
1238 + $pdf_queue_id = get_transient('mxchat_active_queue_pdf');
1239 +
1240 + // Delete all pending items from active queues
1241 + if ($sitemap_queue_id) {
1242 + $wpdb->delete(
1243 + $table_name,
1244 + array(
1245 + 'queue_id' => $sitemap_queue_id,
1246 + 'status' => 'pending'
1247 + ),
1248 + array('%s', '%s')
1249 + );
1250 +
1251 + delete_transient('mxchat_active_queue_sitemap');
1252 + delete_transient('mxchat_last_sitemap_url');
1253 + }
1254 +
1255 + if ($pdf_queue_id) {
1256 + // Get PDF path before deleting
1257 + $pdf_path = $this->mxchat_get_queue_meta($pdf_queue_id, 'pdf_path');
1258 +
1259 + $wpdb->delete(
1260 + $table_name,
1261 + array(
1262 + 'queue_id' => $pdf_queue_id,
1263 + 'status' => 'pending'
1264 + ),
1265 + array('%s', '%s')
1266 + );
1267 +
1268 + // Delete PDF file
1269 + if ($pdf_path && file_exists($pdf_path)) {
1270 + wp_delete_file($pdf_path);
1271 + }
1272 +
1273 + delete_transient('mxchat_active_queue_pdf');
1274 + delete_transient('mxchat_last_pdf_url');
1275 + }
1276 +
1277 + // Redirect back with a success message
1278 + set_transient('mxchat_admin_notice_success',
1279 + esc_html__('Processing has been stopped successfully.', 'mxchat'),
1280 + 30
1281 + );
1282 + wp_safe_redirect(admin_url('admin.php?page=mxchat-prompts'));
1283 + exit;
1284 +}
1285 +
1286 +/**
1287 + * Get content list for processing
1288 + */
1289 +public function ajax_mxchat_get_content_list() {
1290 + // Verify the nonce
1291 + check_ajax_referer('mxchat_content_selector_nonce', 'nonce');
1292 +
1293 + if (!current_user_can('manage_options')) {
1294 + wp_send_json_error(__('Unauthorized access', 'mxchat'));
1295 + }
1296 +
1297 + $page = isset($_GET['page']) ? absint($_GET['page']) : 1;
1298 + $per_page = isset($_GET['per_page']) ? absint($_GET['per_page']) : 50;
1299 + $search = isset($_GET['search']) ? sanitize_text_field($_GET['search']) : '';
1300 + $post_type = isset($_GET['post_type']) ? sanitize_text_field($_GET['post_type']) : 'all';
1301 + $post_status = isset($_GET['post_status']) ? sanitize_text_field($_GET['post_status']) : 'publish';
1302 + $processed_filter = isset($_GET['processed_filter']) ? sanitize_text_field($_GET['processed_filter']) : 'all';
1303 +
1304 + // Build query args
1305 + $args = array(
1306 + 'posts_per_page' => $per_page,
1307 + 'paged' => $page,
1308 + 'post_status' => $post_status !== 'all' ? $post_status : array('publish', 'draft', 'pending'),
1309 + 'orderby' => 'date',
1310 + 'order' => 'DESC',
1311 + );
1312 +
1313 + // Handle post types - IMPROVED VERSION
1314 + if ($post_type !== 'all') {
1315 + $args['post_type'] = $post_type;
1316 + } else {
1317 + // Get all available post types that might contain content
1318 + $all_post_types = array();
1319 +
1320 + // First get all public post types
1321 + $public_types = get_post_types(array('public' => true), 'names');
1322 + $all_post_types = array_merge($all_post_types, $public_types);
1323 +
1324 + // Add common forum/community post types
1325 + $forum_types = array('topic', 'reply', 'forum', 'wpforo_topic', 'wpforo_post');
1326 + foreach ($forum_types as $forum_type) {
1327 + if (post_type_exists($forum_type)) {
1328 + $all_post_types[] = $forum_type;
1329 + }
1330 + }
1331 +
1332 + // Add other commonly used post types
1333 + $common_types = array('product', 'job_listing', 'event', 'portfolio');
1334 + foreach ($common_types as $common_type) {
1335 + if (post_type_exists($common_type)) {
1336 + $all_post_types[] = $common_type;
1337 + }
1338 + }
1339 +
1340 + // Remove duplicates and ensure we have at least some post types
1341 + $all_post_types = array_unique($all_post_types);
1342 +
1343 + if (empty($all_post_types)) {
1344 + // Fallback to basic post types
1345 + $all_post_types = array('post', 'page');
1346 + }
1347 +
1348 + $args['post_type'] = $all_post_types;
1349 +
1350 + // Debug logging to see what post types are being queried
1351 + //error_log('MxChat Debug: Querying post types: ' . implode(', ', $all_post_types));
1352 + }
1353 +
1354 + if (!empty($search)) {
1355 + $args['s'] = $search;
1356 + }
1357 +
1358 + // Get processed data from storage
1359 + $processed_data = array();
1360 +
1361 + $pinecone_options = get_option('mxchat_pinecone_addon_options', array());
1362 + $use_pinecone = ($pinecone_options['mxchat_use_pinecone'] ?? '0') === '1';
1363 +
1364 + if ($use_pinecone && !empty($pinecone_options['mxchat_pinecone_api_key'])) {
1365 + // Get fresh data from Pinecone - no caching
1366 + $processed_data = $this->mxchat_get_pinecone_processed_content($pinecone_options);
1367 + } else {
1368 + // WordPress DB checking with better URL matching for all post types
1369 + global $wpdb;
1370 + $table_name = $wpdb->prefix . 'mxchat_system_prompt_content';
1371 + $processed_items = $wpdb->get_results("SELECT id, source_url, timestamp FROM {$table_name}");
1372 +
1373 + if (!empty($processed_items)) {
1374 + foreach ($processed_items as $item) {
1375 + // Use improved URL matching that works for all post types
1376 + $post_id = $this->mxchat_url_to_post_id_improved($item->source_url);
1377 +
1378 + if ($post_id) {
1379 + $processed_data[$post_id] = array(
1380 + 'db_id' => $item->id,
1381 + 'timestamp' => $item->timestamp,
1382 + 'url' => $item->source_url,
1383 + 'source' => 'wordpress'
1384 + );
1385 + }
1386 + }
1387 + }
1388 + }
1389 +
1390 + // Get processed IDs as a simple array for in_array checks
1391 + $processed_ids = array_keys($processed_data);
1392 +
1393 + // Handle processed/unprocessed filter
1394 + if ($processed_filter === 'processed' && !empty($processed_ids)) {
1395 + $args['post__in'] = $processed_ids;
1396 + } elseif ($processed_filter === 'unprocessed' && !empty($processed_ids)) {
1397 + $args['post__not_in'] = $processed_ids;
1398 + }
1399 +
1400 + // Run the query
1401 + $query = new WP_Query($args);
1402 + $content_items = array();
1403 +
1404 + if ($query->have_posts()) {
1405 + while ($query->have_posts()) {
1406 + $query->the_post();
1407 + $id = get_the_ID();
1408 + $post_date = get_the_date();
1409 + $excerpt = wp_trim_words(get_the_excerpt(), 20, '...');
1410 + $word_count = str_word_count(strip_tags(get_the_content()));
1411 +
1412 + $is_processed = in_array($id, $processed_ids);
1413 + $processed_date = '';
1414 + $db_record_id = 0;
1415 + $data_source = 'none';
1416 +
1417 + if ($is_processed && isset($processed_data[$id])) {
1418 + $item_data = $processed_data[$id];
1419 + $data_source = $item_data['source'];
1420 +
1421 + if ($data_source === 'wordpress' && isset($item_data['timestamp'])) {
1422 + // WordPress DB format
1423 + $timestamp = strtotime($item_data['timestamp']);
1424 + $processed_date = human_time_diff($timestamp, current_time('timestamp')) . ' ago';
1425 + $db_record_id = $item_data['db_id'];
1426 + } elseif ($data_source === 'pinecone') {
1427 + // Pinecone format
1428 + $processed_date = $item_data['processed_date'];
1429 + $db_record_id = $item_data['db_id'];
1430 + }
1431 + }
1432 +
1433 + $content_items[] = array(
1434 + 'id' => $id,
1435 + 'title' => get_the_title(),
1436 + 'permalink' => get_permalink(),
1437 + 'date' => $post_date,
1438 + 'type' => get_post_type(),
1439 + 'status' => get_post_status(),
1440 + 'excerpt' => $excerpt,
1441 + 'word_count' => $word_count,
1442 + 'already_processed' => $is_processed,
1443 + 'processed_date' => $processed_date,
1444 + 'db_record_id' => $db_record_id,
1445 + 'data_source' => $data_source
1446 + );
1447 + }
1448 + wp_reset_postdata();
1449 + }
1450 +
1451 + $response = array(
1452 + 'items' => $content_items,
1453 + 'total' => $query->found_posts,
1454 + 'total_pages' => $query->max_num_pages,
1455 + 'current_page' => $page,
1456 + 'processed_count' => count($processed_ids)
1457 + );
1458 +
1459 + wp_send_json_success($response);
1460 + exit;
1461 +}
1462 +
1463 +
1464 +/**
1465 + * This function handles various WooCommerce URL formats and permalink structures
1466 + */
1467 +private function mxchat_url_to_post_id_improved($url) {
1468 + // First try the standard WordPress function
1469 + $post_id = url_to_postid($url);
1470 +
1471 + if ($post_id > 0) {
1472 + return $post_id;
1473 + }
1474 +
1475 + // If that fails, try more aggressive URL matching
1476 + // Remove trailing slashes and query parameters for better matching
1477 + $clean_url = rtrim($url, '/');
1478 + $clean_url = strtok($clean_url, '?'); // Remove query parameters
1479 +
1480 + // Try again with cleaned URL
1481 + $post_id = url_to_postid($clean_url);
1482 + if ($post_id > 0) {
1483 + return $post_id;
1484 + }
1485 +
1486 + // For bbPress forum topics, try extracting slug from URL
1487 + if (strpos($url, '/topic/') !== false || strpos($url, '/forums/') !== false) {
1488 + // Handle bbPress URLs: /forums/topic/topic-name/
1489 + if (preg_match('/\/forums\/topic\/([^\/\?]+)/', $url, $matches)) {
1490 + $topic_slug = $matches[1];
1491 +
1492 + // Look up topic by slug
1493 + $topic = get_page_by_path($topic_slug, OBJECT, 'topic');
1494 + if ($topic) {
1495 + return $topic->ID;
1496 + }
1497 +
1498 + // Alternative method: query by post_name
1499 + global $wpdb;
1500 + $post_id = $wpdb->get_var($wpdb->prepare(
1501 + "SELECT ID FROM {$wpdb->posts} WHERE post_name = %s AND post_type = 'topic' AND post_status IN ('publish', 'closed')",
1502 + $topic_slug
1503 + ));
1504 +
1505 + if ($post_id) {
1506 + return intval($post_id);
1507 + }
1508 + }
1509 +
1510 + // Handle simpler topic URLs: /topic/topic-name/
1511 + if (preg_match('/\/topic\/([^\/\?]+)/', $url, $matches)) {
1512 + $topic_slug = $matches[1];
1513 +
1514 + global $wpdb;
1515 + $post_id = $wpdb->get_var($wpdb->prepare(
1516 + "SELECT ID FROM {$wpdb->posts} WHERE post_name = %s AND post_type = 'topic' AND post_status IN ('publish', 'closed')",
1517 + $topic_slug
1518 + ));
1519 +
1520 + if ($post_id) {
1521 + return intval($post_id);
1522 + }
1523 + }
1524 + }
1525 +
1526 + // For WooCommerce products
1527 + if (strpos($url, '/product/') !== false || strpos($url, 'product=') !== false) {
1528 + // Extract product slug from various URL formats
1529 + $product_slug = '';
1530 +
1531 + // Handle pretty permalinks: /product/product-name/
1532 + if (preg_match('/\/product\/([^\/\?]+)/', $url, $matches)) {
1533 + $product_slug = $matches[1];
1534 + }
1535 + // Handle query parameters: ?product=product-name
1536 + elseif (preg_match('/[\?&]product=([^&]+)/', $url, $matches)) {
1537 + $product_slug = $matches[1];
1538 + }
1539 +
1540 + if (!empty($product_slug)) {
1541 + // Look up product by slug
1542 + $product = get_page_by_path($product_slug, OBJECT, 'product');
1543 + if ($product) {
1544 + return $product->ID;
1545 + }
1546 +
1547 + // Alternative method: query by post_name
1548 + global $wpdb;
1549 + $post_id = $wpdb->get_var($wpdb->prepare(
1550 + "SELECT ID FROM {$wpdb->posts} WHERE post_name = %s AND post_type = 'product' AND post_status = 'publish'",
1551 + $product_slug
1552 + ));
1553 +
1554 + if ($post_id) {
1555 + return intval($post_id);
1556 + }
1557 + }
1558 + }
1559 +
1560 + // Generic approach: try to extract slug and match against all post types
1561 + $parsed_url = wp_parse_url($clean_url);
1562 + $path = $parsed_url['path'] ?? '';
1563 +
1564 + if (!empty($path)) {
1565 + // Get the last part of the path as potential slug
1566 + $path_parts = array_filter(explode('/', trim($path, '/')));
1567 + $potential_slug = end($path_parts);
1568 +
1569 + if (!empty($potential_slug)) {
1570 + global $wpdb;
1571 +
1572 + // Try to find any post with this slug
1573 + $post_id = $wpdb->get_var($wpdb->prepare(
1574 + "SELECT ID FROM {$wpdb->posts}
1575 + WHERE post_name = %s
1576 + AND post_status IN ('publish', 'closed', 'private')
1577 + AND post_type NOT IN ('revision', 'attachment', 'nav_menu_item')
1578 + ORDER BY CASE
1579 + WHEN post_type = 'post' THEN 1
1580 + WHEN post_type = 'page' THEN 2
1581 + WHEN post_type = 'topic' THEN 3
1582 + WHEN post_type = 'product' THEN 4
1583 + ELSE 5
1584 + END
1585 + LIMIT 1",
1586 + $potential_slug
1587 + ));
1588 +
1589 + if ($post_id) {
1590 + return intval($post_id);
1591 + }
1592 + }
1593 + }
1594 +
1595 + // ADDITIONAL: Try direct database lookup by URL variations
1596 + global $wpdb;
1597 + $table_name = $wpdb->prefix . 'mxchat_system_prompt_content';
1598 +
1599 + // Try variations of the URL (with/without trailing slash, http/https)
1600 + $url_variations = array(
1601 + $url,
1602 + rtrim($url, '/'),
1603 + $url . '/',
1604 + str_replace('http://', 'https://', $url),
1605 + str_replace('https://', 'http://', $url),
1606 + str_replace('http://', 'https://', rtrim($url, '/')),
1607 + str_replace('https://', 'http://', rtrim($url, '/'))
1608 + );
1609 +
1610 + // Remove duplicates
1611 + $url_variations = array_unique($url_variations);
1612 +
1613 + foreach ($url_variations as $variation) {
1614 + $existing_record = $wpdb->get_row($wpdb->prepare(
1615 + "SELECT id, source_url FROM $table_name WHERE source_url = %s",
1616 + $variation
1617 + ));
1618 +
1619 + if ($existing_record) {
1620 + // Try to get post ID from this stored URL
1621 + $stored_post_id = url_to_postid($existing_record->source_url);
1622 + if ($stored_post_id > 0) {
1623 + return $stored_post_id;
1624 + }
1625 + }
1626 + }
1627 +
1628 + return 0; // No match found
1629 +}
1630 +/**
1631 + * Process selected content via AJAX
1632 + */
1633 +public function ajax_mxchat_process_selected_content() {
1634 + // Basic request validation
1635 + if (!check_ajax_referer('mxchat_content_selector_nonce', 'nonce', false)) {
1636 + wp_send_json_error('Invalid nonce');
1637 + exit;
1638 + }
1639 +
1640 + if (!current_user_can('manage_options')) {
1641 + wp_send_json_error('Unauthorized access');
1642 + exit;
1643 + }
1644 +
1645 + // Get post IDs - safely parse the array
1646 + $post_ids = array();
1647 + if (isset($_POST['post_ids']) && is_array($_POST['post_ids'])) {
1648 + foreach ($_POST['post_ids'] as $id) {
1649 + $post_ids[] = absint($id);
1650 + }
1651 + }
1652 +
1653 + if (empty($post_ids)) {
1654 + wp_send_json_error('No content selected');
1655 + exit;
1656 + }
1657 +
1658 + // Get bot_id from request
1659 + $bot_id = isset($_POST['bot_id']) ? sanitize_key($_POST['bot_id']) : 'default';
1660 +
1661 + // Process only ONE post at a time to avoid request size issues
1662 + $post_id = reset($post_ids);
1663 + $post = get_post($post_id);
1664 +
1665 + if (!$post) {
1666 + wp_send_json_error('Post not found');
1667 + exit;
1668 + }
1669 +
1670 + // Get content including title, short description (for WooCommerce), and main content
1671 + $content = $post->post_title . "\n\n";
1672 +
1673 + // Add short description if it exists (WooCommerce products use post_excerpt for short description)
1674 + if (!empty($post->post_excerpt)) {
1675 + // Strip shortcodes first (WPBakery, Elementor, etc.), then strip HTML tags
1676 + $clean_excerpt = strip_shortcodes($post->post_excerpt);
1677 + // Additional regex-based shortcode removal as a safety net
1678 + $clean_excerpt = preg_replace('/\[(\[?)([a-zA-Z0-9_-]+)(?![\w-])([^\]\/]*(?:\/(?!\])[^\]\/]*)*?)(?:(\/)\]|\](?:([^\[]*(?:\[(?!\/\2\])[^\[]*)*)\[\/\2\])?)(\]?)/s', '', $clean_excerpt);
1679 + $content .= "Short Description: " . wp_strip_all_tags($clean_excerpt) . "\n\n";
1680 + }
1681 +
1682 + // Add main content - strip shortcodes first, then strip HTML tags
1683 + $clean_content = strip_shortcodes($post->post_content);
1684 + // Additional regex-based shortcode removal as a safety net
1685 + $clean_content = preg_replace('/\[(\[?)([a-zA-Z0-9_-]+)(?![\w-])([^\]\/]*(?:\/(?!\])[^\]\/]*)*?)(?:(\/)\]|\](?:([^\[]*(?:\[(?!\/\2\])[^\[]*)*)\[\/\2\])?)(\]?)/s', '', $clean_content);
1686 + $content .= wp_strip_all_tags($clean_content);
1687 +
1688 + // ADD CUSTOM PRODUCT TABS SUPPORT (for WooCommerce products)
1689 + if (get_post_type($post_id) === 'product') {
1690 + // Get Custom Product Tabs (supports "Custom Product Tabs for WooCommerce" by Code Parrots)
1691 + $custom_tabs = get_post_meta($post_id, 'yikes_woo_products_tabs', true);
1692 + if (!empty($custom_tabs) && is_array($custom_tabs)) {
1693 + foreach ($custom_tabs as $tab) {
1694 + $tab_title = isset($tab['title']) ? $tab['title'] : (isset($tab['tab_title']) ? $tab['tab_title'] : '');
1695 + $tab_content = isset($tab['content']) ? $tab['content'] : '';
1696 +
1697 + if (!empty($tab_title) && !empty($tab_content)) {
1698 + $content .= "\n" . $tab_title . ": " . wp_strip_all_tags($tab_content) . "\n";
1699 + }
1700 + }
1701 + }
1702 +
1703 + // Also check for reusable/saved tabs applied to this product
1704 + $applied_saved_tabs = get_post_meta($post_id, 'yikes_woo_reusable_products_tabs_applied', true);
1705 + if (!empty($applied_saved_tabs) && is_array($applied_saved_tabs)) {
1706 + $saved_tabs = get_option('yikes_woo_reusable_products_tabs', array());
1707 + if (!empty($saved_tabs) && is_array($saved_tabs)) {
1708 + foreach ($applied_saved_tabs as $saved_tab_id) {
1709 + if (isset($saved_tabs[$saved_tab_id])) {
1710 + $tab = $saved_tabs[$saved_tab_id];
1711 + $tab_title = isset($tab['title']) ? $tab['title'] : (isset($tab['tab_title']) ? $tab['tab_title'] : '');
1712 + $tab_content = isset($tab['content']) ? $tab['content'] : '';
1713 +
1714 + if (!empty($tab_title) && !empty($tab_content)) {
1715 + $content .= "\n" . $tab_title . ": " . wp_strip_all_tags($tab_content) . "\n";
1716 + }
1717 + }
1718 + }
1719 + }
1720 + }
1721 + }
1722 +
1723 + // ADD ACF FIELDS SUPPORT
1724 + $acf_fields = $this->mxchat_get_acf_fields_for_post($post_id);
1725 + if (!empty($acf_fields)) {
1726 + $acf_content_parts = array();
1727 +
1728 + foreach ($acf_fields as $field_name => $field_value) {
1729 + $formatted_value = $this->mxchat_format_acf_field_value($field_value, $field_name, $post_id);
1730 +
1731 + if (!empty($formatted_value)) {
1732 + $field_label = ucwords(str_replace('_', ' ', $field_name));
1733 + $acf_content_parts[] = $field_label . ": " . $formatted_value;
1734 + }
1735 + }
1736 +
1737 + if (!empty($acf_content_parts)) {
1738 + $content .= "\n\n" . implode("\n", $acf_content_parts);
1739 + }
1740 + }
1741 +
1742 + $content = substr($content, 0, 10000); // Limit content size
1743 +
1744 + // Get bot-specific API key
1745 + $bot_options = $this->get_bot_options($bot_id);
1746 + $options = !empty($bot_options) ? $bot_options : get_option('mxchat_options');
1747 + $selected_model = $options['embedding_model'] ?? 'text-embedding-ada-002';
1748 +
1749 + if (strpos($selected_model, 'voyage') === 0) {
1750 + $api_key = $options['voyage_api_key'] ?? '';
1751 + $provider_name = 'Voyage AI';
1752 + } elseif (strpos($selected_model, 'gemini-embedding') === 0) {
1753 + $api_key = $options['gemini_api_key'] ?? '';
1754 + $provider_name = 'Google Gemini';
1755 + } else {
1756 + $api_key = $options['api_key'] ?? '';
1757 + $provider_name = 'OpenAI';
1758 + }
1759 +
1760 + if (empty($api_key)) {
1761 + wp_send_json_error($provider_name . ' API key not configured');
1762 + exit;
1763 + }
1764 +
1765 + $source_url = get_permalink($post_id);
1766 + $vector_id = md5($source_url); // Vector ID for Pinecone
1767 +
1768 + // Check for existing content in bot-specific storage
1769 + $is_update = false;
1770 +
1771 + // Get bot-specific Pinecone configuration
1772 + $bot_pinecone_config = $this->get_bot_pinecone_config($bot_id);
1773 + $use_pinecone = !empty($bot_pinecone_config) && ($bot_pinecone_config['use_pinecone'] ?? false);
1774 +
1775 + if ($use_pinecone && !empty($bot_pinecone_config['api_key'])) {
1776 + // Check Pinecone for this bot
1777 + $pinecone_data = $this->mxchat_get_pinecone_processed_content($bot_pinecone_config);
1778 + if (isset($pinecone_data[$post_id])) {
1779 + $is_update = true;
1780 + }
1781 + } else {
1782 + // Check WordPress DB (same as before since it's shared)
1783 + global $wpdb;
1784 + $table_name = $wpdb->prefix . 'mxchat_system_prompt_content';
1785 + $existing_record = $wpdb->get_row($wpdb->prepare(
1786 + "SELECT id FROM $table_name WHERE source_url = %s",
1787 + $source_url
1788 + ));
1789 +
1790 + if ($existing_record) {
1791 + $is_update = true;
1792 + }
1793 + }
1794 +
1795 + // UPDATED 2.5.6: Determine content type based on post_type
1796 + $post_type = $post->post_type;
1797 + $content_type = 'content'; // Default fallback
1798 +
1799 + // Map WordPress post types to content types
1800 + switch ($post_type) {
1801 + case 'post':
1802 + $content_type = 'post';
1803 + break;
1804 + case 'page':
1805 + $content_type = 'page';
1806 + break;
1807 + case 'product':
1808 + $content_type = 'product';
1809 + break;
1810 + default:
1811 + // For custom post types, use the post type name
1812 + $content_type = sanitize_key($post_type);
1813 + break;
1814 + }
1815 +
1816 + // Use the centralized utility function with bot_id and content_type
1817 + $result = MxChat_Utils::submit_content_to_db(
1818 + $content,
1819 + $source_url,
1820 + $api_key,
1821 + $vector_id,
1822 + $bot_id,
1823 + $content_type
1824 + );
1825 +
1826 + if (is_wp_error($result)) {
1827 + wp_send_json_error('Storage failed: ' . $result->get_error_message());
1828 + exit;
1829 + }
1830 +
1831 + // Automatically apply role restriction based on tags
1832 + $this->apply_role_restriction_to_post($post_id, $source_url);
1833 +
1834 + $operation_type = $is_update ? 'update' : 'new';
1835 +
1836 + // Count ACF fields for debugging
1837 + $acf_field_count = count($acf_fields);
1838 +
1839 + // Success response with minimal data
1840 + wp_send_json_success(array(
1841 + 'message' => $operation_type === 'update' ? 'Content updated successfully' : 'Content processed successfully',
1842 + 'post_id' => $post_id,
1843 + 'title' => $post->post_title,
1844 + 'operation_type' => $operation_type,
1845 + 'vector_id' => $vector_id,
1846 + 'acf_fields_found' => $acf_field_count,
1847 + 'content_preview' => substr($content, 0, 100) . '...',
1848 + 'bot_id' => $bot_id
1849 + ));
1850 + exit;
1851 +}
1852 +
1853 +private function apply_role_restriction_to_post($post_id, $source_url) {
1854 + // Get tag-role mappings
1855 + $mappings = get_option('mxchat_tag_role_mappings', array());
1856 +
1857 + if (empty($mappings)) {
1858 + return; // No mappings, leave as public
1859 + }
1860 +
1861 + // Get all tags for the post
1862 + $post_tags = wp_get_post_tags($post_id, array('fields' => 'slugs'));
1863 +
1864 + if (empty($post_tags)) {
1865 + return; // No tags, leave as public
1866 + }
1867 +
1868 + // Determine the highest role restriction based on tags
1869 + $highest_role = 'public';
1870 + $role_hierarchy = array(
1871 + 'public' => 0,
1872 + 'logged_in' => 1,
1873 + 'subscriber' => 2,
1874 + 'contributor' => 3,
1875 + 'author' => 4,
1876 + 'editor' => 5,
1877 + 'administrator' => 6
1878 + );
1879 +
1880 + foreach ($post_tags as $tag_slug) {
1881 + if (isset($mappings[$tag_slug])) {
1882 + $role = $mappings[$tag_slug];
1883 + if (isset($role_hierarchy[$role]) && $role_hierarchy[$role] > $role_hierarchy[$highest_role]) {
1884 + $highest_role = $role;
1885 + }
1886 + }
1887 + }
1888 +
1889 + // If no restricted tags found, return (leave as public)
1890 + if ($highest_role === 'public') {
1891 + return;
1892 + }
1893 +
1894 + // Update the role restriction in the database
1895 + global $wpdb;
1896 +
1897 + // Check if using Pinecone
1898 + $pinecone_options = get_option('mxchat_pinecone_addon_options', array());
1899 + $use_pinecone = ($pinecone_options['mxchat_use_pinecone'] ?? '0') === '1';
1900 +
1901 + if ($use_pinecone && !empty($pinecone_options['mxchat_pinecone_api_key'])) {
1902 + // Update Pinecone role restriction
1903 + $roles_table = $wpdb->prefix . 'mxchat_pinecone_roles';
1904 + $vector_id = md5($source_url);
1905 +
1906 + $wpdb->replace(
1907 + $roles_table,
1908 + array(
1909 + 'vector_id' => $vector_id,
1910 + 'role_restriction' => $highest_role,
1911 + 'updated_at' => current_time('mysql')
1912 + ),
1913 + array('%s', '%s', '%s')
1914 + );
1915 + } else {
1916 + // Update WordPress DB
1917 + $table_name = $wpdb->prefix . 'mxchat_system_prompt_content';
1918 +
1919 + $wpdb->update(
1920 + $table_name,
1921 + array('role_restriction' => $highest_role),
1922 + array('source_url' => $source_url),
1923 + array('%s'),
1924 + array('%s')
1925 + );
1926 + }
1927 +}
1928 +
1929 +public function mxchat_get_public_post_types() {
1930 + // Get all public post types
1931 + $post_types = get_post_types(array('public' => true), 'objects');
1932 + $post_type_options = array();
1933 +
1934 + foreach ($post_types as $post_type) {
1935 + $post_type_options[$post_type->name] = $post_type->label;
1936 + }
1937 +
1938 + // Also include common forum/community post types that might not be marked as public
1939 + $additional_types = array(
1940 + 'topic' => 'Forum Topics (bbPress)',
1941 + 'reply' => 'Forum Replies (bbPress)',
1942 + 'forum' => 'Forums (bbPress)',
1943 + 'wpforo_topic' => 'wpForo Topics',
1944 + 'wpforo_post' => 'wpForo Posts'
1945 + );
1946 +
1947 + foreach ($additional_types as $type_name => $type_label) {
1948 + if (post_type_exists($type_name) && !isset($post_type_options[$type_name])) {
1949 + $post_type_options[$type_name] = $type_label;
1950 + }
1951 + }
1952 +
1953 + return $post_type_options;
1954 +}
1955 +
1956 +/**
1957 + * Retrieves processed content from Pinecone API
1958 + */
1959 +public function mxchat_get_pinecone_processed_content($pinecone_options) {
1960 + $api_key = $pinecone_options['mxchat_pinecone_api_key'] ?? '';
1961 + $host = $pinecone_options['mxchat_pinecone_host'] ?? '';
1962 +
1963 + if (empty($api_key) || empty($host)) {
1964 + return array();
1965 + }
1966 +
1967 + $pinecone_data = array();
1968 +
1969 + try {
1970 + // Always get fresh data from Pinecone
1971 + $pinecone_data = $this->mxchat_scan_pinecone_for_processed_content($pinecone_options);
1972 +
1973 + // Method 2: Final fallback - try stats endpoint (if available)
1974 + if (empty($pinecone_data)) {
1975 + $stats_url = "https://{$host}/describe_index_stats";
1976 +
1977 + $response = wp_remote_post($stats_url, array(
1978 + 'headers' => array(
1979 + 'Api-Key' => $api_key,
1980 + 'Content-Type' => 'application/json'
1981 + ),
1982 + 'body' => json_encode(array()),
1983 + 'timeout' => 30
1984 + ));
1985 +
1986 + if (!is_wp_error($response) && wp_remote_retrieve_response_code($response) === 200) {
1987 + $body = wp_remote_retrieve_body($response);
1988 + $stats_data = json_decode($body, true);
1989 + }
1990 + }
1991 +
1992 + } catch (Exception $e) {
1993 + // Log error but return fresh data only
1994 + }
1995 +
1996 + return $pinecone_data;
1997 +}
1998 +public function mxchat_fetch_pinecone_vectors_by_ids($pinecone_options, $vector_ids) {
1999 + //error_log('=== DEBUG: Starting mxchat_fetch_pinecone_vectors_by_ids ===');
2000 +
2001 + $api_key = $pinecone_options['mxchat_pinecone_api_key'] ?? '';
2002 + $host = $pinecone_options['mxchat_pinecone_host'] ?? '';
2003 +
2004 + if (empty($api_key) || empty($host) || empty($vector_ids)) {
2005 + //error_log('DEBUG: Missing parameters for fetch by IDs');
2006 + return array();
2007 + }
2008 +
2009 + try {
2010 + $fetch_url = "https://{$host}/vectors/fetch";
2011 + //error_log('DEBUG: Fetch URL: ' . $fetch_url);
2012 + //error_log('DEBUG: Fetching ' . count($vector_ids) . ' vector IDs');
2013 +
2014 + // Pinecone fetch API allows fetching specific vectors by ID
2015 + $fetch_data = array(
2016 + 'ids' => array_values($vector_ids)
2017 + );
2018 +
2019 + $response = wp_remote_post($fetch_url, array(
2020 + 'headers' => array(
2021 + 'Api-Key' => $api_key,
2022 + 'Content-Type' => 'application/json'
2023 + ),
2024 + 'body' => json_encode($fetch_data),
2025 + 'timeout' => 30
2026 + ));
2027 +
2028 + if (is_wp_error($response)) {
2029 + //error_log('DEBUG: Fetch by IDs WP error: ' . $response->get_error_message());
2030 + return array();
2031 + }
2032 +
2033 + $response_code = wp_remote_retrieve_response_code($response);
2034 + //error_log('DEBUG: Fetch response code: ' . $response_code);
2035 +
2036 + if ($response_code !== 200) {
2037 + $error_body = wp_remote_retrieve_body($response);
2038 + //error_log('DEBUG: Fetch failed with body: ' . $error_body);
2039 + return array();
2040 + }
2041 +
2042 + $body = wp_remote_retrieve_body($response);
2043 + $data = json_decode($body, true);
2044 +
2045 + //error_log('DEBUG: Fetch response structure: ' . print_r(array_keys($data), true));
2046 +
2047 + if (!isset($data['vectors'])) {
2048 + //error_log('DEBUG: No vectors key in response');
2049 + return array();
2050 + }
2051 +
2052 + $processed_data = array();
2053 +
2054 + foreach ($data['vectors'] as $vector_id => $vector_data) {
2055 + $metadata = $vector_data['metadata'] ?? array();
2056 + $source_url = $metadata['source_url'] ?? '';
2057 +
2058 + if (!empty($source_url)) {
2059 + $post_id = url_to_postid($source_url);
2060 + if ($post_id) {
2061 + $created_at = $metadata['created_at'] ?? '';
2062 + $processed_date = 'Recently';
2063 +
2064 + if (!empty($created_at)) {
2065 + $timestamp = is_numeric($created_at) ? $created_at : strtotime($created_at);
2066 + if ($timestamp) {
2067 + $processed_date = human_time_diff($timestamp, current_time('timestamp')) . ' ago';
2068 + }
2069 + }
2070 +
2071 + $processed_data[$post_id] = array(
2072 + 'db_id' => $vector_id,
2073 + 'processed_date' => $processed_date,
2074 + 'url' => $source_url,
2075 + 'source' => 'pinecone',
2076 + 'timestamp' => $timestamp ?? current_time('timestamp')
2077 + );
2078 + }
2079 + }
2080 + }
2081 +
2082 + //error_log('DEBUG: Processed ' . count($processed_data) . ' vectors from fetch');
2083 + return $processed_data;
2084 +
2085 + } catch (Exception $e) {
2086 + //error_log('DEBUG: Exception in fetch_pinecone_vectors_by_ids: ' . $e->getMessage());
2087 + return array();
2088 + }
2089 +}
2090 +
2091 +/**
2092 + * Scan Pinecone for processed content
2093 + */
2094 +public function mxchat_scan_pinecone_for_processed_content($pinecone_options) {
2095 + $api_key = $pinecone_options['mxchat_pinecone_api_key'] ?? '';
2096 + $host = $pinecone_options['mxchat_pinecone_host'] ?? '';
2097 +
2098 + if (empty($api_key) || empty($host)) {
2099 + return array();
2100 + }
2101 +
2102 + try {
2103 + // Use multiple random vectors to get better coverage
2104 + $all_matches = array();
2105 + $seen_ids = array();
2106 +
2107 + // Try 3 different random vectors to get better coverage
2108 + for ($i = 0; $i < 3; $i++) {
2109 + $query_url = "https://{$host}/query";
2110 +
2111 + // Generate a random unit vector instead of zeros
2112 + $random_vector = array();
2113 + for ($j = 0; $j < 1536; $j++) {
2114 + $random_vector[] = (rand(-1000, 1000) / 1000.0);
2115 + }
2116 +
2117 + // Normalize the vector to unit length
2118 + $magnitude = sqrt(array_sum(array_map(function($x) { return $x * $x; }, $random_vector)));
2119 + if ($magnitude > 0) {
2120 + $random_vector = array_map(function($x) use ($magnitude) { return $x / $magnitude; }, $random_vector);
2121 + }
2122 +
2123 + $query_data = array(
2124 + 'includeMetadata' => true,
2125 + 'includeValues' => false,
2126 + 'topK' => 10000,
2127 + 'vector' => $random_vector
2128 + );
2129 +
2130 + $response = wp_remote_post($query_url, array(
2131 + 'headers' => array(
2132 + 'Api-Key' => $api_key,
2133 + 'Content-Type' => 'application/json'
2134 + ),
2135 + 'body' => json_encode($query_data),
2136 + 'timeout' => 30
2137 + ));
2138 +
2139 + if (is_wp_error($response)) {
2140 + continue;
2141 + }
2142 +
2143 + $response_code = wp_remote_retrieve_response_code($response);
2144 +
2145 + if ($response_code !== 200) {
2146 + continue;
2147 + }
2148 +
2149 + $body = wp_remote_retrieve_body($response);
2150 + $data = json_decode($body, true);
2151 +
2152 + if (isset($data['matches'])) {
2153 + foreach ($data['matches'] as $match) {
2154 + $match_id = $match['id'] ?? '';
2155 + if (!empty($match_id) && !isset($seen_ids[$match_id])) {
2156 + $all_matches[] = $match;
2157 + $seen_ids[$match_id] = true;
2158 + }
2159 + }
2160 + }
2161 + }
2162 +
2163 + // Convert matches to processed data format
2164 + $processed_data = array();
2165 +
2166 + foreach ($all_matches as $match) {
2167 + $metadata = $match['metadata'] ?? array();
2168 + $source_url = $metadata['source_url'] ?? '';
2169 + $match_id = $match['id'] ?? '';
2170 +
2171 + if (!empty($source_url) && !empty($match_id)) {
2172 + $post_id = url_to_postid($source_url);
2173 + if ($post_id) {
2174 + $created_at = $metadata['created_at'] ?? '';
2175 + $processed_date = 'Recently';
2176 +
2177 + if (!empty($created_at)) {
2178 + $timestamp = is_numeric($created_at) ? $created_at : strtotime($created_at);
2179 + if ($timestamp) {
2180 + $processed_date = human_time_diff($timestamp, current_time('timestamp')) . ' ago';
2181 + }
2182 + }
2183 +
2184 + $processed_data[$post_id] = array(
2185 + 'db_id' => $match_id,
2186 + 'processed_date' => $processed_date,
2187 + 'url' => $source_url,
2188 + 'source' => 'pinecone',
2189 + 'timestamp' => $timestamp ?? current_time('timestamp')
2190 + );
2191 + }
2192 + }
2193 + }
2194 +
2195 + return $processed_data;
2196 +
2197 + } catch (Exception $e) {
2198 + return array();
2199 + }
2200 +}
2201 +/**
2202 + * Generate embeddings from input text for MXChat with bot support
2203 + */
2204 +private function mxchat_generate_embedding($text, $bot_id = 'default') {
2205 + // Enable detailed logging for debugging
2206 + //error_log('[MXCHAT-EMBED] Starting embedding generation for bot: ' . $bot_id . '. Text length: ' . strlen($text) . ' bytes');
2207 + //error_log('[MXCHAT-EMBED] Text preview: ' . substr($text, 0, 100) . '...');
2208 +
2209 + // Get bot-specific options
2210 + $bot_options = $this->get_bot_options($bot_id);
2211 + $options = !empty($bot_options) ? $bot_options : get_option('mxchat_options');
2212 +
2213 + $selected_model = $options['embedding_model'] ?? 'text-embedding-ada-002';
2214 + //error_log('[MXCHAT-EMBED] Selected embedding model for bot ' . $bot_id . ': ' . $selected_model);
2215 +
2216 + // Determine provider and endpoint
2217 + if (strpos($selected_model, 'voyage') === 0) {
2218 + $api_key = $options['voyage_api_key'] ?? '';
2219 + $endpoint = 'https://api.voyageai.com/v1/embeddings';
2220 + $provider_name = 'Voyage AI';
2221 + //error_log('[MXCHAT-EMBED] Using Voyage AI API for bot ' . $bot_id);
2222 + } elseif (strpos($selected_model, 'gemini-embedding') === 0) {
2223 + $api_key = $options['gemini_api_key'] ?? '';
2224 + $endpoint = 'https://generativelanguage.googleapis.com/v1beta/models/' . $selected_model . ':embedContent';
2225 + $provider_name = 'Google Gemini';
2226 + //error_log('[MXCHAT-EMBED] Using Google Gemini API for bot ' . $bot_id);
2227 + } else {
2228 + $api_key = $options['api_key'] ?? '';
2229 + $endpoint = 'https://api.openai.com/v1/embeddings';
2230 + $provider_name = 'OpenAI';
2231 + //error_log('[MXCHAT-EMBED] Using OpenAI API for bot ' . $bot_id);
2232 + }
2233 +
2234 + //error_log('[MXCHAT-EMBED] Using endpoint: ' . $endpoint);
2235 +
2236 + if (empty($api_key)) {
2237 + $error_message = sprintf('Missing %s API key for bot %s. Please configure your API key in the bot settings.', $provider_name, $bot_id);
2238 + //error_log('[MXCHAT-EMBED] Error: ' . $error_message);
2239 + return $error_message;
2240 + }
2241 +
2242 + // Check if text is too long (for OpenAI, roughly estimate tokens as words/0.75)
2243 + $estimated_tokens = ceil(str_word_count($text) / 0.75);
2244 + //error_log('[MXCHAT-EMBED] Estimated token count: ~' . $estimated_tokens);
2245 +
2246 + if ($estimated_tokens > 8000 && strpos($selected_model, 'voyage') === false && strpos($selected_model, 'gemini-embedding') === false) {
2247 + //error_log('[MXCHAT-EMBED] Warning: Text may exceed OpenAI token limits (8K for most models)');
2248 + // Consider truncating text here
2249 + }
2250 +
2251 + // Prepare request body based on provider
2252 + if (strpos($selected_model, 'gemini-embedding') === 0) {
2253 + // Gemini API format
2254 + $request_body = array(
2255 + 'model' => 'models/' . $selected_model,
2256 + 'content' => array(
2257 + 'parts' => array(
2258 + array('text' => $text)
2259 + )
2260 + )
2261 + );
2262 +
2263 + // Set output dimensionality to 1536 for consistency with other models
2264 + $request_body['outputDimensionality'] = 1536;
2265 + } else {
2266 + // OpenAI/Voyage API format
2267 + $request_body = array(
2268 + 'model' => $selected_model,
2269 + 'input' => $text
2270 + );
2271 +
2272 + // Add output_dimension for voyage-3-large model
2273 + if ($selected_model === 'voyage-3-large') {
2274 + $request_body['output_dimension'] = 2048;
2275 + }
2276 + }
2277 +
2278 + //error_log('[MXCHAT-EMBED] Request prepared with model: ' . $selected_model);
2279 +
2280 + // Prepare headers based on provider
2281 + if (strpos($selected_model, 'gemini-embedding') === 0) {
2282 + // Gemini uses API key as query parameter
2283 + $endpoint .= '?key=' . $api_key;
2284 + $headers = array(
2285 + 'Content-Type' => 'application/json'
2286 + );
2287 + } else {
2288 + // OpenAI/Voyage use Bearer token
2289 + $headers = array(
2290 + 'Authorization' => 'Bearer ' . $api_key,
2291 + 'Content-Type' => 'application/json'
2292 + );
2293 + }
2294 +
2295 + // Make API request
2296 + //error_log('[MXCHAT-EMBED] Sending API request to: ' . $endpoint);
2297 + $response = wp_remote_post($endpoint, array(
2298 + 'body' => wp_json_encode($request_body),
2299 + 'headers' => $headers,
2300 + 'timeout' => 60 // Increased timeout for large inputs
2301 + ));
2302 +
2303 + // Handle wp_remote_post errors
2304 + if (is_wp_error($response)) {
2305 + $error_message = $response->get_error_message();
2306 + //error_log('[MXCHAT-EMBED] WP Remote Post Error: ' . $error_message);
2307 + return 'Connection error: ' . $error_message;
2308 + }
2309 +
2310 + // Get and check HTTP response code
2311 + $http_code = wp_remote_retrieve_response_code($response);
2312 + //error_log('[MXCHAT-EMBED] API Response Code: ' . $http_code);
2313 +
2314 + if ($http_code !== 200) {
2315 + $error_body = wp_remote_retrieve_body($response);
2316 + //error_log('[MXCHAT-EMBED] API Error Response Body: ' . $error_body);
2317 +
2318 + // Try to parse error for more details
2319 + $error_json = json_decode($error_body, true);
2320 + if (json_last_error() === JSON_ERROR_NONE && isset($error_json['error'])) {
2321 + $error_type = $error_json['error']['type'] ?? 'unknown';
2322 + $error_message = $error_json['error']['message'] ?? 'No message';
2323 + //error_log('[MXCHAT-EMBED] API Error Type: ' . $error_type);
2324 + //error_log('[MXCHAT-EMBED] API Error Message: ' . $error_message);
2325 +
2326 + // Customize error message for common API errors
2327 + if ($error_type === 'invalid_request_error' && strpos($error_message, 'API key') !== false) {
2328 + $error_message = sprintf('Invalid %s API key for bot %s. Please check your API key in the bot settings.', $provider_name, $bot_id);
2329 + } elseif ($error_type === 'authentication_error') {
2330 + $error_message = sprintf('%s authentication failed for bot %s. Please verify your API key in the bot settings.', $provider_name, $bot_id);
2331 + }
2332 +
2333 + //error_log('[MXCHAT-EMBED] Returning error: ' . $error_message);
2334 + return $error_message;
2335 + }
2336 +
2337 + $error_message = sprintf("API Error (HTTP %d): Unable to generate embedding for bot %s", $http_code, $bot_id);
2338 + //error_log('[MXCHAT-EMBED] Returning error: ' . $error_message);
2339 + return $error_message;
2340 + }
2341 +
2342 + // Parse response body
2343 + $response_body = wp_remote_retrieve_body($response);
2344 + //error_log('[MXCHAT-EMBED] Received response length: ' . strlen($response_body) . ' bytes');
2345 +
2346 + $response_data = json_decode($response_body, true);
2347 +
2348 + if (json_last_error() !== JSON_ERROR_NONE) {
2349 + $error = json_last_error_msg();
2350 + //error_log('[MXCHAT-EMBED] JSON Parse Error: ' . $error);
2351 + //error_log('[MXCHAT-EMBED] Response preview: ' . substr($response_body, 0, 200));
2352 + return "Failed to parse API response: $error";
2353 + }
2354 +
2355 + // Handle different response formats based on provider
2356 + if (strpos($selected_model, 'gemini-embedding') === 0) {
2357 + // Gemini API response format
2358 + if (isset($response_data['embedding']['values'])) {
2359 + $embedding_dimensions = count($response_data['embedding']['values']);
2360 + //error_log('[MXCHAT-EMBED] Successfully extracted Gemini embedding with ' . $embedding_dimensions . ' dimensions');
2361 +
2362 + // Check if embedding dimensions are as expected (should be 1536)
2363 + if ($embedding_dimensions !== 1536) {
2364 + //error_log('[MXCHAT-EMBED] Warning: Unexpected Gemini embedding dimensions: ' . $embedding_dimensions);
2365 + }
2366 +
2367 + return $response_data['embedding']['values'];
2368 + } else {
2369 + //error_log('[MXCHAT-EMBED] Error: No embedding found in Gemini response');
2370 + //error_log('[MXCHAT-EMBED] Response structure: ' . wp_json_encode(array_keys($response_data)));
2371 +
2372 + if (isset($response_data['error'])) {
2373 + $error_message = "Gemini API Error in response: " . wp_json_encode($response_data['error']);
2374 + //error_log('[MXCHAT-EMBED] ' . $error_message);
2375 + return $error_message;
2376 + }
2377 +
2378 + $error_message = "Invalid Gemini API response format: No embedding found";
2379 + //error_log('[MXCHAT-EMBED] ' . $error_message);
2380 + return $error_message;
2381 + }
2382 + } else {
2383 + // OpenAI/Voyage API response format
2384 + if (isset($response_data['data'][0]['embedding'])) {
2385 + $embedding_dimensions = count($response_data['data'][0]['embedding']);
2386 + //error_log('[MXCHAT-EMBED] Successfully extracted embedding with ' . $embedding_dimensions . ' dimensions');
2387 +
2388 + // Check if embedding dimensions are as expected
2389 + if (($selected_model === 'text-embedding-ada-002' && $embedding_dimensions !== 1536) ||
2390 + ($selected_model === 'voyage-3-large' && $embedding_dimensions !== 2048)) {
2391 + //error_log('[MXCHAT-EMBED] Warning: Unexpected embedding dimensions');
2392 + }
2393 +
2394 + return $response_data['data'][0]['embedding'];
2395 + } else {
2396 + //error_log('[MXCHAT-EMBED] Error: No embedding found in response');
2397 + //error_log('[MXCHAT-EMBED] Response structure: ' . wp_json_encode(array_keys($response_data)));
2398 +
2399 + if (isset($response_data['error'])) {
2400 + $error_message = "API Error in response: " . wp_json_encode($response_data['error']);
2401 + //error_log('[MXCHAT-EMBED] ' . $error_message);
2402 + return $error_message;
2403 + }
2404 +
2405 + $error_message = "Invalid API response format: No embedding found";
2406 + //error_log('[MXCHAT-EMBED] ' . $error_message);
2407 + return $error_message;
2408 + }
2409 + }
2410 +}
2411 +
2412 +/**
2413 + * Get bot-specific options for multi-bot functionality
2414 + * Falls back to default options if bot_id is 'default' or multi-bot add-on is not active
2415 + */
2416 +private function get_bot_options($bot_id = 'default') {
2417 + //error_log("MXCHAT DEBUG: get_bot_options called for bot: " . $bot_id);
2418 +
2419 + if ($bot_id === 'default' || !class_exists('MxChat_Multi_Bot_Manager')) {
2420 + //error_log("MXCHAT DEBUG: Using default options (no multi-bot or bot is 'default')");
2421 + return array();
2422 + }
2423 +
2424 + $bot_options = apply_filters('mxchat_get_bot_options', array(), $bot_id);
2425 +
2426 + if (!empty($bot_options)) {
2427 + //error_log("MXCHAT DEBUG: Got bot-specific options from filter");
2428 + if (isset($bot_options['similarity_threshold'])) {
2429 + //error_log(" - similarity_threshold: " . $bot_options['similarity_threshold']);
2430 + }
2431 + }
2432 +
2433 + return is_array($bot_options) ? $bot_options : array();
2434 +}
2435 +
2436 +/**
2437 + * Get bot-specific Pinecone configuration
2438 + * Used in the knowledge retrieval functions
2439 + */
2440 +// Also add debugging to your get_bot_pinecone_config function
2441 +private function get_bot_pinecone_config($bot_id = 'default') {
2442 + //error_log("MXCHAT DEBUG: get_bot_pinecone_config called for bot: " . $bot_id);
2443 +
2444 + // If default bot or multi-bot add-on not active, use default Pinecone config
2445 + if ($bot_id === 'default' || !class_exists('MxChat_Multi_Bot_Manager')) {
2446 + //error_log("MXCHAT DEBUG: Using default Pinecone config (no multi-bot or bot is 'default')");
2447 + $addon_options = get_option('mxchat_pinecone_addon_options', array());
2448 + $config = array(
2449 + 'use_pinecone' => (isset($addon_options['mxchat_use_pinecone']) && $addon_options['mxchat_use_pinecone'] === '1'),
2450 + 'api_key' => $addon_options['mxchat_pinecone_api_key'] ?? '',
2451 + 'host' => $addon_options['mxchat_pinecone_host'] ?? '',
2452 + 'namespace' => $addon_options['mxchat_pinecone_namespace'] ?? ''
2453 + );
2454 + //error_log("MXCHAT DEBUG: Default config - use_pinecone: " . ($config['use_pinecone'] ? 'true' : 'false'));
2455 + return $config;
2456 + }
2457 +
2458 + //error_log("MXCHAT DEBUG: Calling filter 'mxchat_get_bot_pinecone_config' for bot: " . $bot_id);
2459 +
2460 + // Hook for multi-bot add-on to provide bot-specific Pinecone config
2461 + $bot_pinecone_config = apply_filters('mxchat_get_bot_pinecone_config', array(), $bot_id);
2462 +
2463 + if (!empty($bot_pinecone_config)) {
2464 + //error_log("MXCHAT DEBUG: Got bot-specific config from filter");
2465 + //error_log(" - use_pinecone: " . (isset($bot_pinecone_config['use_pinecone']) ? ($bot_pinecone_config['use_pinecone'] ? 'true' : 'false') : 'not set'));
2466 + //error_log(" - host: " . ($bot_pinecone_config['host'] ?? 'not set'));
2467 + //error_log(" - namespace: " . ($bot_pinecone_config['namespace'] ?? 'not set'));
2468 + } else {
2469 + //error_log("MXCHAT DEBUG: Filter returned empty config!");
2470 + }
2471 +
2472 + return is_array($bot_pinecone_config) ? $bot_pinecone_config : array();
2473 +}
2474 +
2475 +
2476 +public function mxchat_ajax_dismiss_completed_status() {
2477 + try {
2478 + // Verify the request
2479 + check_ajax_referer('mxchat_status_nonce', 'nonce');
2480 +
2481 + if (!current_user_can('manage_options')) {
2482 + wp_send_json_error('Unauthorized access');
2483 + exit;
2484 + }
2485 +
2486 + $card_type = isset($_POST['card_type']) ? sanitize_text_field($_POST['card_type']) : '';
2487 +
2488 + if ($card_type === 'pdf') {
2489 + // Clear PDF status
2490 + $pdf_url = get_transient('mxchat_last_pdf_url');
2491 + if ($pdf_url) {
2492 + delete_transient('mxchat_pdf_status_' . md5($pdf_url));
2493 + delete_transient('mxchat_last_pdf_url');
2494 + }
2495 + } elseif ($card_type === 'sitemap') {
2496 + // Clear sitemap status
2497 + $sitemap_url = get_transient('mxchat_last_sitemap_url');
2498 + if ($sitemap_url) {
2499 + delete_transient('mxchat_sitemap_status_' . md5($sitemap_url));
2500 + delete_transient('mxchat_last_sitemap_url');
2501 + }
2502 + }
2503 +
2504 + wp_send_json_success(array('message' => 'Status dismissed successfully'));
2505 +
2506 + } catch (Exception $e) {
2507 + wp_send_json_error(array('message' => 'Error dismissing status: ' . $e->getMessage()));
2508 + }
2509 +}
2510 +
2511 +/**
2512 + * Render completed status cards on page load
2513 + * This ensures completed processing status persists through page refreshes
2514 + */
2515 +public function mxchat_render_completed_status_cards() {
2516 + $output = '';
2517 +
2518 + // Check for completed PDF status
2519 + $pdf_url = get_transient('mxchat_last_pdf_url');
2520 + if ($pdf_url) {
2521 + $pdf_status = $this->mxchat_get_pdf_processing_status($pdf_url);
2522 + if ($pdf_status && ($pdf_status['status'] === 'complete' || $pdf_status['status'] === 'error')) {
2523 + $output .= $this->mxchat_render_pdf_status_card($pdf_status, $pdf_url);
2524 + }
2525 + }
2526 +
2527 + // Check for completed sitemap status
2528 + $sitemap_url = get_transient('mxchat_last_sitemap_url');
2529 + if ($sitemap_url) {
2530 + $sitemap_status = $this->mxchat_get_sitemap_processing_status($sitemap_url);
2531 + if ($sitemap_status && ($sitemap_status['status'] === 'complete' || $sitemap_status['status'] === 'error')) {
2532 + $output .= $this->mxchat_render_sitemap_status_card($sitemap_status, $sitemap_url);
2533 + }
2534 + }
2535 +
2536 + return $output;
2537 +}
2538 +
2539 +/**
2540 + * Render PDF status card HTML
2541 + */
2542 +private function mxchat_render_pdf_status_card($status, $pdf_url) {
2543 + $html = '<div class="mxchat-status-card" data-card-type="pdf">';
2544 + $html .= '<div class="mxchat-status-header">';
2545 + $html .= '<h4>' . esc_html__('PDF Processing Status', 'mxchat') . '</h4>';
2546 +
2547 + // Add dismiss button for completed status
2548 + if ($status['status'] === 'complete' || $status['status'] === 'error') {
2549 + $html .= '<button type="button" class="mxchat-dismiss-button">' . esc_html__('Dismiss', 'mxchat') . '</button>';
2550 + }
2551 +
2552 + // Process Batch button for processing status
2553 + if ($status['status'] === 'processing') {
2554 + $html .= '<button type="button" class="mxchat-manual-batch-btn"
2555 + data-process-type="pdf"
2556 + data-url="' . esc_attr($pdf_url) . '">
2557 + ' . esc_html__('Process Batch', 'mxchat') . '</button>';
2558 + }
2559 +
2560 + // Add status badges
2561 + if ($status['status'] === 'error') {
2562 + $html .= '<span class="mxchat-status-badge mxchat-status-failed">' . esc_html__('Error', 'mxchat') . '</span>';
2563 + } elseif ($status['status'] === 'complete') {
2564 + if ($status['failed_pages'] > 0) {
2565 + $html .= '<span class="mxchat-status-badge mxchat-status-warning">' .
2566 + sprintf(esc_html__('Completed with %d failures', 'mxchat'), $status['failed_pages']) . '</span>';
2567 + } else {
2568 + $html .= '<span class="mxchat-status-badge mxchat-status-success">' . esc_html__('Complete', 'mxchat') . '</span>';
2569 + }
2570 + }
2571 +
2572 + $html .= '</div>'; // End header
2573 +
2574 + // Progress bar
2575 + $html .= '<div class="mxchat-progress-bar">';
2576 + $html .= '<div class="mxchat-progress-fill" style="width: ' . esc_attr($status['percentage']) . '%"></div>';
2577 + $html .= '</div>';
2578 +
2579 + // Status details
2580 + $html .= '<div class="mxchat-status-details">';
2581 + $html .= '<p>' . sprintf(
2582 + esc_html__('Progress: %d of %d pages (%d%%)', 'mxchat'),
2583 + $status['processed_pages'],
2584 + $status['total_pages'],
2585 + $status['percentage']
2586 + ) . '</p>';
2587 +
2588 + // Show failed pages count if any
2589 + if ($status['failed_pages'] > 0) {
2590 + $html .= '<p><strong>' . esc_html__('Failed pages:', 'mxchat') . '</strong> ' . esc_html($status['failed_pages']) . '</p>';
2591 + }
2592 +
2593 + $html .= '<p><strong>' . esc_html__('Status:', 'mxchat') . '</strong> ' . esc_html(ucfirst($status['status'])) . '</p>';
2594 + $html .= '<p><strong>' . esc_html__('Last update:', 'mxchat') . '</strong> ' . esc_html($status['last_update']) . '</p>';
2595 +
2596 + // Add completion summary if available AND it's an array
2597 + if (isset($status['completion_summary']) && is_array($status['completion_summary']) && !empty($status['completion_summary'])) {
2598 + $summary = $status['completion_summary'];
2599 + $html .= '<div class="mxchat-completion-summary">';
2600 + $html .= '<h5>' . esc_html__('Processing Summary', 'mxchat') . '</h5>';
2601 + $html .= '<p><strong>' . esc_html__('Total Pages:', 'mxchat') . '</strong> ' . esc_html($summary['total_pages']) . '</p>';
2602 + $html .= '<p><strong>' . esc_html__('Successful:', 'mxchat') . '</strong> ' . esc_html($summary['successful_pages']) . '</p>';
2603 + $html .= '<p><strong>' . esc_html__('Failed:', 'mxchat') . '</strong> ' . esc_html($summary['failed_pages']) . '</p>';
2604 + $html .= '<p><strong>' . esc_html__('Completed:', 'mxchat') . '</strong> ' . esc_html($summary['completion_time']) . '</p>';
2605 + $html .= '</div>';
2606 + }
2607 +
2608 + // Add failed pages list if any AND it's an array
2609 + if (isset($status['failed_pages_list']) && is_array($status['failed_pages_list']) && !empty($status['failed_pages_list'])) {
2610 + $html .= $this->mxchat_render_failed_pages_list($status['failed_pages_list']);
2611 + }
2612 +
2613 + // Add error message if any
2614 + if (isset($status['error']) && !empty($status['error'])) {
2615 + $html .= '<div class="mxchat-error-notice">';
2616 + $html .= '<p class="error">' . esc_html($status['error']) . '</p>';
2617 + $html .= '</div>';
2618 + }
2619 +
2620 + $html .= '</div>'; // End details
2621 + $html .= '</div>'; // End card
2622 +
2623 + return $html;
2624 +}
2625 +/**
2626 + * Render sitemap status card HTML
2627 + */
2628 +private function mxchat_render_sitemap_status_card($status, $sitemap_url) {
2629 + $html = '<div class="mxchat-status-card" data-card-type="sitemap">';
2630 + $html .= '<div class="mxchat-status-header">';
2631 + $html .= '<h4>' . esc_html__('Sitemap Processing Status', 'mxchat') . '</h4>';
2632 +
2633 + // Add dismiss button for completed status
2634 + if ($status['status'] === 'complete' || $status['status'] === 'error') {
2635 + $html .= '<button type="button" class="mxchat-dismiss-button">' . esc_html__('Dismiss', 'mxchat') . '</button>';
2636 + }
2637 +
2638 + // Process Batch button for processing status
2639 + if ($status['status'] === 'processing') {
2640 + $html .= '<button type="button" class="mxchat-manual-batch-btn"
2641 + data-process-type="sitemap"
2642 + data-url="' . esc_attr($sitemap_url) . '">
2643 + ' . esc_html__('Process Batch', 'mxchat') . '</button>';
2644 + }
2645 +
2646 + // Add status badges
2647 + if ($status['status'] === 'error') {
2648 + $html .= '<span class="mxchat-status-badge mxchat-status-failed">' . esc_html__('Error', 'mxchat') . '</span>';
2649 + } elseif ($status['status'] === 'complete') {
2650 + if ($status['failed_urls'] > 0) {
2651 + $html .= '<span class="mxchat-status-badge mxchat-status-warning">' .
2652 + sprintf(esc_html__('Completed with %d failures', 'mxchat'), $status['failed_urls']) . '</span>';
2653 + } else {
2654 + $html .= '<span class="mxchat-status-badge mxchat-status-success">' . esc_html__('Complete', 'mxchat') . '</span>';
2655 + }
2656 + }
2657 +
2658 + $html .= '</div>'; // End header
2659 +
2660 + // Progress bar
2661 + $html .= '<div class="mxchat-progress-bar">';
2662 + $html .= '<div class="mxchat-progress-fill" style="width: ' . esc_attr($status['percentage']) . '%"></div>';
2663 + $html .= '</div>';
2664 +
2665 + // Status details
2666 + $html .= '<div class="mxchat-status-details">';
2667 + $html .= '<p>' . sprintf(
2668 + esc_html__('Progress: %d of %d URLs (%d%%)', 'mxchat'),
2669 + $status['processed_urls'],
2670 + $status['total_urls'],
2671 + $status['percentage']
2672 + ) . '</p>';
2673 +
2674 + // Show failed URLs count if any
2675 + if ($status['failed_urls'] > 0) {
2676 + $html .= '<p><strong>' . esc_html__('Failed URLs:', 'mxchat') . '</strong> ' . esc_html($status['failed_urls']) . '</p>';
2677 + }
2678 +
2679 + $html .= '<p><strong>' . esc_html__('Status:', 'mxchat') . '</strong> ' . esc_html(ucfirst($status['status'])) . '</p>';
2680 + $html .= '<p><strong>' . esc_html__('Last update:', 'mxchat') . '</strong> ' . esc_html($status['last_update']) . '</p>';
2681 +
2682 + // Add completion summary if available AND it's an array
2683 + if (isset($status['completion_summary']) && is_array($status['completion_summary']) && !empty($status['completion_summary'])) {
2684 + $summary = $status['completion_summary'];
2685 + $html .= '<div class="mxchat-completion-summary">';
2686 + $html .= '<h5>' . esc_html__('Processing Summary', 'mxchat') . '</h5>';
2687 + $html .= '<p><strong>' . esc_html__('Total URLs:', 'mxchat') . '</strong> ' . esc_html($summary['total_urls']) . '</p>';
2688 + $html .= '<p><strong>' . esc_html__('Successful:', 'mxchat') . '</strong> ' . esc_html($summary['successful_urls']) . '</p>';
2689 + $html .= '<p><strong>' . esc_html__('Failed:', 'mxchat') . '</strong> ' . esc_html($summary['failed_urls']) . '</p>';
2690 + $html .= '<p><strong>' . esc_html__('Completed:', 'mxchat') . '</strong> ' . esc_html($summary['completion_time']) . '</p>';
2691 + $html .= '</div>';
2692 + }
2693 +
2694 + // Add error messages if any (but not the failed URLs list)
2695 + if (!empty($status['error']) || !empty($status['last_error'])) {
2696 + $html .= '<div class="mxchat-error-notice">';
2697 +
2698 + if (!empty($status['error'])) {
2699 + $html .= '<p class="error">' . esc_html($status['error']) . '</p>';
2700 + }
2701 +
2702 + if (!empty($status['last_error'])) {
2703 + $html .= '<p class="last-error">' . esc_html__('Last error:', 'mxchat') . ' ' . esc_html($status['last_error']) . '</p>';
2704 + }
2705 +
2706 + $html .= '</div>';
2707 + }
2708 +
2709 + $html .= '</div>'; // End details
2710 + $html .= '</div>'; // End card
2711 +
2712 + return $html;
2713 +}
2714 +
2715 +
2716 +/**
2717 + * Render failed pages list
2718 + */
2719 +private function mxchat_render_failed_pages_list($failed_pages_list) {
2720 + // Validate that $failed_pages_list is an array and not empty
2721 + if (!is_array($failed_pages_list) || empty($failed_pages_list)) {
2722 + return '';
2723 + }
2724 +
2725 + $html = '<div class="mxchat-error-notice">';
2726 + $html .= '<div class="mxchat-failed-pages-container">';
2727 + $html .= '<h5>' . sprintf(esc_html__('Failed Pages (%d)', 'mxchat'), count($failed_pages_list)) . '</h5>';
2728 + $html .= '<details>';
2729 + $html .= '<summary>' . esc_html__('Show Failed Pages', 'mxchat') . '</summary>';
2730 + $html .= '<div class="mxchat-failed-pages-list">';
2731 +
2732 + // Create table for failed pages
2733 + $html .= '<table class="widefat striped">';
2734 + $html .= '<thead><tr>';
2735 + $html .= '<th>' . esc_html__('Page', 'mxchat') . '</th>';
2736 + $html .= '<th>' . esc_html__('Error', 'mxchat') . '</th>';
2737 + $html .= '<th>' . esc_html__('Retries', 'mxchat') . '</th>';
2738 + $html .= '<th>' . esc_html__('Time', 'mxchat') . '</th>';
2739 + $html .= '</tr></thead><tbody>';
2740 +
2741 + // Sort failed pages by most recent
2742 + $sorted_failed_pages = $failed_pages_list;
2743 + usort($sorted_failed_pages, function($a, $b) {
2744 + return ($b['time'] ?? 0) - ($a['time'] ?? 0);
2745 + });
2746 +
2747 + foreach ($sorted_failed_pages as $item) {
2748 + // Ensure $item is an array before accessing its elements
2749 + if (!is_array($item)) {
2750 + continue;
2751 + }
2752 +
2753 + $time_ago = isset($item['time']) ? human_time_diff($item['time'], current_time('timestamp')) . ' ' . esc_html__('ago', 'mxchat') : 'Unknown';
2754 + $html .= '<tr>';
2755 + $html .= '<td>' . esc_html__('Page', 'mxchat') . ' ' . esc_html($item['page'] ?? 'Unknown') . '</td>';
2756 + $html .= '<td style="word-break: break-word;">' . esc_html($item['error'] ?? 'Unknown error') . '</td>';
2757 + $html .= '<td>' . esc_html($item['retries'] ?? 'N/A') . '</td>';
2758 + $html .= '<td>' . esc_html($time_ago) . '</td>';
2759 + $html .= '</tr>';
2760 + }
2761 +
2762 + $html .= '</tbody></table>';
2763 + $html .= '</div></details></div></div>';
2764 +
2765 + return $html;
2766 +}
2767 +
2768 +/**
2769 + * Render failed URLs list
2770 + */
2771 +private function mxchat_render_failed_urls_list($failed_urls_list) {
2772 + // Validate that $failed_urls_list is an array and not empty
2773 + if (!is_array($failed_urls_list) || empty($failed_urls_list)) {
2774 + return '';
2775 + }
2776 +
2777 + $html = '<div class="mxchat-failed-urls-container">';
2778 + $html .= '<h5>' . sprintf(esc_html__('Failed URLs (%d)', 'mxchat'), count($failed_urls_list)) . '</h5>';
2779 + $html .= '<details>';
2780 + $html .= '<summary>' . esc_html__('Show Failed URLs', 'mxchat') . '</summary>';
2781 + $html .= '<div class="mxchat-failed-urls-list">';
2782 +
2783 + // Create table for failed URLs
2784 + $html .= '<table class="widefat striped">';
2785 + $html .= '<thead><tr>';
2786 + $html .= '<th>' . esc_html__('URL', 'mxchat') . '</th>';
2787 + $html .= '<th>' . esc_html__('Error', 'mxchat') . '</th>';
2788 + $html .= '<th>' . esc_html__('Retries', 'mxchat') . '</th>';
2789 + $html .= '<th>' . esc_html__('Time', 'mxchat') . '</th>';
2790 + $html .= '</tr></thead><tbody>';
2791 +
2792 + // Sort failed URLs by most recent
2793 + $sorted_failed_urls = $failed_urls_list;
2794 + usort($sorted_failed_urls, function($a, $b) {
2795 + return ($b['time'] ?? 0) - ($a['time'] ?? 0);
2796 + });
2797 +
2798 + // Show up to 50 failed URLs
2799 + $display_urls = array_slice($sorted_failed_urls, 0, 50);
2800 +
2801 + foreach ($display_urls as $item) {
2802 + // Ensure $item is an array before accessing its elements
2803 + if (!is_array($item)) {
2804 + continue;
2805 + }
2806 +
2807 + $url = $item['url'] ?? '';
2808 + $time_ago = isset($item['time']) ? human_time_diff($item['time'], current_time('timestamp')) . ' ' . esc_html__('ago', 'mxchat') : 'Unknown';
2809 +
2810 + // Truncate URL for display
2811 + $display_url = strlen($url) > 50 ? substr($url, 0, 47) . '...' : $url;
2812 +
2813 + $html .= '<tr>';
2814 + $html .= '<td style="word-break: break-all;">';
2815 + if (!empty($url)) {
2816 + $html .= '<a href="' . esc_url($url) . '" target="_blank" rel="noopener noreferrer">' . esc_html($display_url) . '</a>';
2817 + } else {
2818 + $html .= esc_html__('Unknown URL', 'mxchat');
2819 + }
2820 + $html .= '</td>';
2821 + $html .= '<td style="word-break: break-word;">' . esc_html($item['error'] ?? 'Unknown error') . '</td>';
2822 + $html .= '<td>' . esc_html($item['retries'] ?? 'N/A') . '</td>';
2823 + $html .= '<td>' . esc_html($time_ago) . '</td>';
2824 + $html .= '</tr>';
2825 + }
2826 +
2827 + $html .= '</tbody></table>';
2828 +
2829 + if (count($failed_urls_list) > 50) {
2830 + $html .= '<div class="mxchat-failed-urls-more">+ ' .
2831 + (count($failed_urls_list) - 50) .
2832 + ' ' . esc_html__('more failed URLs not shown', 'mxchat') . '</div>';
2833 + }
2834 +
2835 + $html .= '</div></details></div>';
2836 +
2837 + return $html;
2838 +}
2839 +
2840 +/**
2841 + * Get all ACF fields for a specific post
2842 + */
2843 +public function mxchat_get_acf_fields_for_post($post_id) {
2844 + if (!function_exists('get_fields')) {
2845 + return array();
2846 + }
2847 +
2848 + $fields = get_fields($post_id);
2849 + if (!$fields || !is_array($fields)) {
2850 + return array();
2851 + }
2852 +
2853 + return $fields;
2854 +}
2855 +
2856 +/**
2857 + * Format ACF field values for content extraction
2858 + */
2859 +public function mxchat_format_acf_field_value($value, $field_name = '', $post_id = 0) {
2860 + if (empty($value)) {
2861 + return '';
2862 + }
2863 +
2864 + // Handle WP_Post objects first (THIS IS THE KEY FIX)
2865 + if ($value instanceof WP_Post) {
2866 + return $value->post_title ?: '';
2867 + }
2868 +
2869 + // Handle other WP objects
2870 + if (is_object($value)) {
2871 + if (isset($value->post_title)) {
2872 + return $value->post_title;
2873 + } elseif (isset($value->display_name)) {
2874 + return $value->display_name;
2875 + } elseif (isset($value->name)) {
2876 + return $value->name;
2877 + } elseif (method_exists($value, '__toString')) {
2878 + try {
2879 + return (string) $value;
2880 + } catch (Exception $e) {
2881 + return '';
2882 + }
2883 + }
2884 + // For any other objects, return empty string
2885 + return '';
2886 + }
2887 +
2888 + // Handle different ACF field types
2889 + if (is_array($value)) {
2890 + // Check if it's an image/file field
2891 + if (isset($value['url'])) {
2892 + // Image field - return alt text, title, or caption
2893 + if (!empty($value['alt'])) {
2894 + return $value['alt'];
2895 + } elseif (!empty($value['title'])) {
2896 + return $value['title'];
2897 + } elseif (!empty($value['caption'])) {
2898 + return $value['caption'];
2899 + } else {
2900 + return ''; // Don't include just the URL
2901 + }
2902 + }
2903 +
2904 + // Check if it's a post object or relationship field
2905 + if (isset($value['post_title'])) {
2906 + return $value['post_title'];
2907 + }
2908 +
2909 + // Check if it's a user field
2910 + if (isset($value['display_name'])) {
2911 + return $value['display_name'];
2912 + }
2913 +
2914 + // Check if it's a taxonomy term
2915 + if (isset($value['name']) && isset($value['taxonomy'])) {
2916 + return $value['name'];
2917 + }
2918 +
2919 + // Check if it's a select field with label
2920 + if (isset($value['label'])) {
2921 + return $value['label'];
2922 + }
2923 +
2924 + // Check for repeater field or flexible content
2925 + if (is_numeric(key($value))) {
2926 + $sub_values = array();
2927 + foreach ($value as $sub_item) {
2928 + if (is_array($sub_item)) {
2929 + // For repeater/flexible content, extract text values
2930 + $sub_text = $this->mxchat_extract_text_from_acf_array($sub_item);
2931 + if (!empty($sub_text)) {
2932 + $sub_values[] = $sub_text;
2933 + }
2934 + } elseif ($sub_item instanceof WP_Post) {
2935 + // Handle WP_Post objects in arrays
2936 + $sub_values[] = $sub_item->post_title ?: '';
2937 + } else {
2938 + $sub_values[] = (string) $sub_item;
2939 + }
2940 + }
2941 + return implode(', ', array_filter($sub_values));
2942 + }
2943 +
2944 + // For other arrays, try to extract meaningful text
2945 + $text_values = array();
2946 + foreach ($value as $key => $val) {
2947 + if (is_string($val) && !empty(trim($val))) {
2948 + $text_values[] = trim($val);
2949 + } elseif ($val instanceof WP_Post) {
2950 + // Handle WP_Post objects in associative arrays
2951 + $text_values[] = $val->post_title ?: '';
2952 + } elseif (is_array($val) && isset($val['post_title'])) {
2953 + $text_values[] = $val['post_title'];
2954 + } elseif (is_array($val) && isset($val['name'])) {
2955 + $text_values[] = $val['name'];
2956 + }
2957 + }
2958 +
2959 + return implode(', ', array_filter($text_values));
2960 + }
2961 +
2962 + // Handle boolean values
2963 + if (is_bool($value)) {
2964 + return $value ? 'Yes' : 'No';
2965 + }
2966 +
2967 + // Handle numeric values
2968 + if (is_numeric($value)) {
2969 + return (string) $value;
2970 + }
2971 +
2972 + // Handle string values
2973 + if (is_string($value)) {
2974 + return trim($value);
2975 + }
2976 +
2977 + // For anything else that we can't handle, return empty string
2978 + // This prevents the "Object could not be converted to string" error
2979 + return '';
2980 +}
2981 +
2982 +/**
2983 + * Extract text from complex ACF array structures
2984 + */
2985 +private function mxchat_extract_text_from_acf_array($array) {
2986 + if (!is_array($array)) {
2987 + return '';
2988 + }
2989 +
2990 + $text_parts = array();
2991 +
2992 + foreach ($array as $key => $value) {
2993 + if (is_string($value) && !empty(trim($value))) {
2994 + // Skip keys that are likely to be IDs or technical values
2995 + if (!is_numeric($value) || strlen($value) > 10) {
2996 + $text_parts[] = trim($value);
2997 + }
2998 + } elseif ($value instanceof WP_Post) {
2999 + // Handle WP_Post objects
3000 + $text_parts[] = $value->post_title ?: '';
3001 + } elseif (is_array($value)) {
3002 + if (isset($value['post_title'])) {
3003 + $text_parts[] = $value['post_title'];
3004 + } elseif (isset($value['name'])) {
3005 + $text_parts[] = $value['name'];
3006 + } elseif (isset($value['label'])) {
3007 + $text_parts[] = $value['label'];
3008 + }
3009 + } elseif (is_object($value)) {
3010 + // Handle other objects safely
3011 + if (isset($value->post_title)) {
3012 + $text_parts[] = $value->post_title;
3013 + } elseif (isset($value->name)) {
3014 + $text_parts[] = $value->name;
3015 + } elseif (isset($value->display_name)) {
3016 + $text_parts[] = $value->display_name;
3017 + }
3018 + }
3019 + }
3020 +
3021 + return implode(', ', array_filter($text_parts));
3022 +}
3023 +
3024 +public function mxchat_handle_post_update($post_id, $post, $update) {
3025 + // Basic validation checks
3026 + if (defined('DOING_AUTOSAVE') && DOING_AUTOSAVE || wp_is_post_revision($post_id)) {
3027 + return;
3028 + }
3029 +
3030 + $post_type = $post->post_type;
3031 +
3032 + // Check if sync is enabled for this post type
3033 + $should_sync = false;
3034 +
3035 + // Check built-in post types first
3036 + if ($post_type === 'post' && get_option('mxchat_auto_sync_posts') === '1') {
3037 + $should_sync = true;
3038 + } else if ($post_type === 'page' && get_option('mxchat_auto_sync_pages') === '1') {
3039 + $should_sync = true;
3040 + } else {
3041 + // Check custom post types
3042 + $option_name = 'mxchat_auto_sync_' . $post_type;
3043 + if (get_option($option_name) === '1') {
3044 + $should_sync = true;
3045 + }
3046 + }
3047 +
3048 + if (!$should_sync) {
3049 + return;
3050 + }
3051 +
3052 + // Check if we have stored the previous status and URL in our transients
3053 + $previous_status_key = 'mxchat_prev_status_' . $post_id;
3054 + $previous_status = get_transient($previous_status_key);
3055 +
3056 + $previous_url_key = 'mxchat_prev_url_' . $post_id;
3057 + $previous_url = get_transient($previous_url_key);
3058 +
3059 + // If the post was previously published but is now not published, remove from knowledge base
3060 + if ($previous_status === 'publish' && $post->post_status !== 'publish') {
3061 + // Use the stored URL from when it was published, or fall back to current permalink
3062 + $source_url = $previous_url ?: get_permalink($post_id);
3063 +
3064 + if ($source_url) {
3065 + // Check if Pinecone is enabled
3066 + $pinecone_options = get_option('mxchat_pinecone_addon_options', array());
3067 + $use_pinecone = ($pinecone_options['mxchat_use_pinecone'] ?? '0') === '1';
3068 +
3069 + if ($use_pinecone && !empty($pinecone_options['mxchat_pinecone_api_key'])) {
3070 + // Delete from Pinecone
3071 + $this->mxchat_delete_from_pinecone_by_url($source_url, $pinecone_options);
3072 + } else {
3073 + // Delete from WordPress DB
3074 + global $wpdb;
3075 + $table_name = $wpdb->prefix . 'mxchat_system_prompt_content';
3076 +
3077 + $result = $wpdb->delete(
3078 + $table_name,
3079 + array('source_url' => $source_url),
3080 + array('%s')
3081 + );
3082 + }
3083 + }
3084 +
3085 + // Clean up the transients and exit early
3086 + delete_transient($previous_status_key);
3087 + delete_transient($previous_url_key);
3088 + return;
3089 + }
3090 +
3091 + // Store the current status for next time (if this is an update)
3092 + if ($update) {
3093 + set_transient($previous_status_key, $post->post_status, DAY_IN_SECONDS);
3094 +
3095 + // If the post is currently published, also store its URL
3096 + if ($post->post_status === 'publish') {
3097 + $current_url = get_permalink($post_id);
3098 + set_transient($previous_url_key, $current_url, DAY_IN_SECONDS);
3099 + }
3100 + }
3101 +
3102 + // Only process currently published content for adding/updating
3103 + if ($post->post_status === 'publish') {
3104 + // Get the source URL
3105 + $source_url = get_permalink($post_id);
3106 +
3107 + // Get content with proper formatting (matching ajax_mxchat_process_selected_content)
3108 + $title = get_the_title($post_id);
3109 + $content = get_post_field('post_content', $post_id);
3110 + $excerpt = get_post_field('post_excerpt', $post_id);
3111 +
3112 + // Strip shortcodes first (removes WPBakery, Elementor, etc.)
3113 + $content = strip_shortcodes($content);
3114 + $excerpt = strip_shortcodes($excerpt);
3115 +
3116 + // Additional regex-based shortcode removal as a safety net
3117 + $content = preg_replace('/\[(\[?)([a-zA-Z0-9_-]+)(?![\w-])([^\]\/]*(?:\/(?!\])[^\]\/]*)*?)(?:(\/)\]|\](?:([^\[]*(?:\[(?!\/\2\])[^\[]*)*)\[\/\2\])?)(\]?)/s', '', $content);
3118 + $excerpt = preg_replace('/\[(\[?)([a-zA-Z0-9_-]+)(?![\w-])([^\]\/]*(?:\/(?!\])[^\]\/]*)*?)(?:(\/)\]|\](?:([^\[]*(?:\[(?!\/\2\])[^\[]*)*)\[\/\2\])?)(\]?)/s', '', $excerpt);
3119 +
3120 + // Strip tags but preserve structure (don't use 'the_content' filter as it may re-add shortcodes)
3121 + $content = wp_strip_all_tags($content);
3122 +
3123 + // Combine title, short description (if exists), and content
3124 + $final_content = $title . "\n\n";
3125 +
3126 + // Add short description if it exists (WooCommerce products use post_excerpt for short description)
3127 + if (!empty($excerpt)) {
3128 + $final_content .= "Short Description: " . wp_strip_all_tags($excerpt) . "\n\n";
3129 + }
3130 +
3131 + $final_content .= $content;
3132 +
3133 + // For custom post types like job_listing, include additional fields
3134 + if ($post_type === 'job_listing') {
3135 + // Add job-specific meta if available
3136 + $job_location = get_post_meta($post_id, '_job_location', true);
3137 + if (!empty($job_location)) {
3138 + $final_content .= "\n\nLocation: " . $job_location;
3139 + }
3140 +
3141 + // Get job type terms
3142 + $job_types = get_the_terms($post_id, 'job_listing_type');
3143 + if (!empty($job_types) && !is_wp_error($job_types)) {
3144 + $types = array();
3145 + foreach ($job_types as $type) {
3146 + $types[] = $type->name;
3147 + }
3148 + $final_content .= "\n\nJob Type: " . implode(', ', $types);
3149 + }
3150 +
3151 + // Get company name if available
3152 + $company_name = get_post_meta($post_id, '_company_name', true);
3153 + if (!empty($company_name)) {
3154 + $final_content .= "\n\nCompany: " . $company_name;
3155 + }
3156 + }
3157 +
3158 + // Get API key with proper model detection
3159 + $options = get_option('mxchat_options');
3160 + $selected_model = $options['embedding_model'] ?? 'text-embedding-ada-002';
3161 +
3162 + if (strpos($selected_model, 'voyage') === 0) {
3163 + $api_key = $options['voyage_api_key'] ?? '';
3164 + } elseif (strpos($selected_model, 'gemini-embedding') === 0) {
3165 + $api_key = $options['gemini_api_key'] ?? '';
3166 + } else {
3167 + $api_key = $options['api_key'] ?? '';
3168 + }
3169 +
3170 + if (empty($api_key)) {
3171 + return;
3172 + }
3173 +
3174 + // Use the centralized utility function for storage
3175 + $result = MxChat_Utils::submit_content_to_db(
3176 + $final_content,
3177 + $source_url,
3178 + $api_key,
3179 + md5($source_url) // Vector ID for Pinecone
3180 + );
3181 +
3182 + // After successful storage, apply role restriction based on tags
3183 + if (!is_wp_error($result)) {
3184 + $this->apply_role_restriction_to_post($post_id, $source_url);
3185 + }
3186 + }
3187 +
3188 + // Clean up the stored previous status if not used above
3189 + if ($previous_status !== 'publish' || $post->post_status === 'publish') {
3190 + delete_transient($previous_status_key);
3191 + delete_transient($previous_url_key);
3192 + }
3193 +}
3194 +
3195 +/**
3196 + * Store the post status and URL before update to detect status transitions
3197 + * This runs before the post is actually updated in the database
3198 + */
3199 +public function mxchat_store_pre_update_status($post_id, $data) {
3200 + // Get the current post from database (before update)
3201 + $current_post = get_post($post_id);
3202 +
3203 + if ($current_post) {
3204 + // Store the current status temporarily
3205 + $status_key = 'mxchat_prev_status_' . $post_id;
3206 + set_transient($status_key, $current_post->post_status, HOUR_IN_SECONDS);
3207 +
3208 + // If the post is currently published, also store its URL
3209 + if ($current_post->post_status === 'publish') {
3210 + $url_key = 'mxchat_prev_url_' . $post_id;
3211 + $current_url = get_permalink($post_id);
3212 + set_transient($url_key, $current_url, HOUR_IN_SECONDS);
3213 + }
3214 + }
3215 +}
3216 +
3217 +public function mxchat_handle_post_delete($post_id) {
3218 + // Get post data before it's deleted
3219 + $post = get_post($post_id);
3220 +
3221 + // Basic validation
3222 + if (!$post || wp_is_post_revision($post_id)) {
3223 + return;
3224 + }
3225 +
3226 + $post_type = $post->post_type;
3227 +
3228 + // Check if sync is enabled for this post type
3229 + $should_sync = false;
3230 +
3231 + // Check built-in post types first
3232 + if ($post_type === 'post' && get_option('mxchat_auto_sync_posts') === '1') {
3233 + $should_sync = true;
3234 + } else if ($post_type === 'page' && get_option('mxchat_auto_sync_pages') === '1') {
3235 + $should_sync = true;
3236 + } else {
3237 + // Check custom post types
3238 + $option_name = 'mxchat_auto_sync_' . $post_type;
3239 + if (get_option($option_name) === '1') {
3240 + $should_sync = true;
3241 + }
3242 + }
3243 +
3244 + if (!$should_sync) {
3245 + return;
3246 + }
3247 +
3248 + // Get the URL before post is deleted
3249 + $source_url = get_permalink($post_id);
3250 + if (!$source_url) {
3251 + //error_log('MXChat: Failed to get permalink for post ' . $post_id);
3252 + return;
3253 + }
3254 +
3255 + // Check if Pinecone is enabled
3256 + $pinecone_options = get_option('mxchat_pinecone_addon_options', array());
3257 + $use_pinecone = ($pinecone_options['mxchat_use_pinecone'] ?? '0') === '1';
3258 +
3259 + if ($use_pinecone && !empty($pinecone_options['mxchat_pinecone_api_key'])) {
3260 + // Delete from Pinecone
3261 + $this->mxchat_delete_from_pinecone_by_url($source_url, $pinecone_options);
3262 + } else {
3263 + // Delete from WordPress DB
3264 + global $wpdb;
3265 + $table_name = $wpdb->prefix . 'mxchat_system_prompt_content';
3266 +
3267 + $result = $wpdb->delete(
3268 + $table_name,
3269 + array('source_url' => $source_url),
3270 + array('%s')
3271 + );
3272 +
3273 + if ($result === false) {
3274 + //error_log('MXChat: WordPress DB deletion failed for URL: ' . $source_url);
3275 + }
3276 + }
3277 +}
3278 +
3279 +
3280 + /**
3281 + * Deletes data from Pinecone using a source URL
3282 + */
3283 + public function mxchat_delete_from_pinecone_by_url($source_url, $pinecone_options) {
3284 + $host = $pinecone_options['mxchat_pinecone_host'] ?? '';
3285 + $api_key = $pinecone_options['mxchat_pinecone_api_key'] ?? '';
3286 +
3287 + if (empty($host) || empty($api_key)) {
3288 + //error_log('MXChat: Pinecone deletion failed - missing configuration');
3289 + return false;
3290 + }
3291 +
3292 + $api_endpoint = "https://{$host}/vectors/delete";
3293 + $vector_id = md5($source_url);
3294 +
3295 + $request_body = array(
3296 + 'ids' => array($vector_id)
3297 + );
3298 +
3299 + $response = wp_remote_post($api_endpoint, array(
3300 + 'headers' => array(
3301 + 'Api-Key' => $api_key,
3302 + 'accept' => 'application/json',
3303 + 'content-type' => 'application/json'
3304 + ),
3305 + 'body' => wp_json_encode($request_body),
3306 + 'timeout' => 30
3307 + ));
3308 +
3309 + if (is_wp_error($response)) {
3310 + //error_log('MXChat: Pinecone deletion error - ' . $response->get_error_message());
3311 + return false;
3312 + }
3313 +
3314 + $response_code = wp_remote_retrieve_response_code($response);
3315 + if ($response_code !== 200) {
3316 + //error_log('MXChat: Pinecone deletion failed with status ' . $response_code);
3317 + return false;
3318 + }
3319 +
3320 + return true;
3321 + }
3322 +
3323 +
3324 +
3325 +public function mxchat_handle_product_change($post_id, $post, $update) {
3326 + if ($post->post_type !== 'product') {
3327 + return;
3328 + }
3329 +
3330 + if ($post->post_status === 'publish') {
3331 + add_action('shutdown', function() use ($post_id) {
3332 + $product = wc_get_product($post_id);
3333 + if ($product) {
3334 + $this->mxchat_store_product_embedding($product);
3335 + }
3336 + });
3337 + }
3338 +}
3339 +
3340 +/**
3341 + * Store WooCommerce product embeddings
3342 + */
3343 +private function mxchat_store_product_embedding($product) {
3344 + if (!isset($this->options['enable_woocommerce_integration']) ||
3345 + !in_array($this->options['enable_woocommerce_integration'], ['1', 'on'])) {
3346 + return;
3347 + }
3348 +
3349 + $source_url = get_permalink($product->get_id());
3350 +
3351 + // Build product content
3352 + $title = $product->get_name();
3353 + $description = $product->get_description();
3354 + $short_description = $product->get_short_description();
3355 + $regular_price = $product->get_regular_price();
3356 + $sale_price = $product->get_sale_price();
3357 + $sku = $product->get_sku();
3358 +
3359 + // Format content consistently
3360 + $content = $title . "\n\n";
3361 +
3362 + if (!empty($description)) {
3363 + $content .= wp_strip_all_tags($description) . "\n\n";
3364 + }
3365 +
3366 + if (!empty($short_description)) {
3367 + $content .= "Short Description: " . wp_strip_all_tags($short_description) . "\n\n";
3368 + }
3369 +
3370 + $content .= "Price: $" . $regular_price . "\n";
3371 +
3372 + if (!empty($sale_price)) {
3373 + $content .= "Sale Price: $" . $sale_price . "\n";
3374 + }
3375 +
3376 + if (!empty($sku)) {
3377 + $content .= "SKU: " . $sku . "\n";
3378 + }
3379 +
3380 + // Get Custom Product Tabs (supports "Custom Product Tabs for WooCommerce" by Code Parrots)
3381 + $custom_tabs = get_post_meta($product->get_id(), 'yikes_woo_products_tabs', true);
3382 + if (!empty($custom_tabs) && is_array($custom_tabs)) {
3383 + foreach ($custom_tabs as $tab) {
3384 + $tab_title = isset($tab['title']) ? $tab['title'] : (isset($tab['tab_title']) ? $tab['tab_title'] : '');
3385 + $tab_content = isset($tab['content']) ? $tab['content'] : '';
3386 +
3387 + if (!empty($tab_title) && !empty($tab_content)) {
3388 + $content .= "\n" . $tab_title . ": " . wp_strip_all_tags($tab_content) . "\n";
3389 + }
3390 + }
3391 + }
3392 +
3393 + // Also check for reusable/saved tabs applied to this product
3394 + $applied_saved_tabs = get_post_meta($product->get_id(), 'yikes_woo_reusable_products_tabs_applied', true);
3395 + if (!empty($applied_saved_tabs) && is_array($applied_saved_tabs)) {
3396 + // Get the saved tabs option
3397 + $saved_tabs = get_option('yikes_woo_reusable_products_tabs', array());
3398 + if (!empty($saved_tabs) && is_array($saved_tabs)) {
3399 + foreach ($applied_saved_tabs as $saved_tab_id) {
3400 + if (isset($saved_tabs[$saved_tab_id])) {
3401 + $tab = $saved_tabs[$saved_tab_id];
3402 + $tab_title = isset($tab['title']) ? $tab['title'] : (isset($tab['tab_title']) ? $tab['tab_title'] : '');
3403 + $tab_content = isset($tab['content']) ? $tab['content'] : '';
3404 +
3405 + if (!empty($tab_title) && !empty($tab_content)) {
3406 + $content .= "\n" . $tab_title . ": " . wp_strip_all_tags($tab_content) . "\n";
3407 + }
3408 + }
3409 + }
3410 + }
3411 + }
3412 +
3413 + // Get API key with proper model detection
3414 + $options = get_option('mxchat_options');
3415 + $selected_model = $options['embedding_model'] ?? 'text-embedding-ada-002';
3416 +
3417 + if (strpos($selected_model, 'voyage') === 0) {
3418 + $api_key = $options['voyage_api_key'] ?? '';
3419 + } elseif (strpos($selected_model, 'gemini-embedding') === 0) {
3420 + $api_key = $options['gemini_api_key'] ?? '';
3421 + } else {
3422 + $api_key = $options['api_key'] ?? '';
3423 + }
3424 +
3425 + if (empty($api_key)) {
3426 + //error_log('MxChat Auto-sync: No API key configured for embedding model');
3427 + return;
3428 + }
3429 +
3430 + // Use the centralized utility function for storage
3431 + $result = MxChat_Utils::submit_content_to_db(
3432 + $content,
3433 + $source_url,
3434 + $api_key,
3435 + md5($source_url) // Vector ID for Pinecone
3436 + );
3437 +
3438 + // After successful storage, apply role restriction based on tags
3439 + if (!is_wp_error($result)) {
3440 + $this->apply_role_restriction_to_post($product->get_id(), $source_url);
3441 + }
3442 +
3443 + if (is_wp_error($result)) {
3444 + //error_log('MxChat WooCommerce sync failed for product ' . $product->get_id() . ': ' . $result->get_error_message());
3445 + }
3446 +}
3447 +
3448 +public function mxchat_handle_product_delete($post_id) {
3449 + if (get_post_type($post_id) !== 'product') {
3450 + return;
3451 + }
3452 +
3453 + $source_url = get_permalink($post_id);
3454 +
3455 + // Check if Pinecone is enabled
3456 + $pinecone_options = get_option('mxchat_pinecone_addon_options', array());
3457 + $use_pinecone = ($pinecone_options['mxchat_use_pinecone'] ?? '0') === '1';
3458 +
3459 + if ($use_pinecone && !empty($pinecone_options['mxchat_pinecone_api_key'])) {
3460 + // Delete from Pinecone
3461 + $this->mxchat_delete_from_pinecone_by_url($source_url, $pinecone_options);
3462 + } else {
3463 + // Delete from WordPress DB
3464 + global $wpdb;
3465 + $table_name = $wpdb->prefix . 'mxchat_system_prompt_content';
3466 +
3467 + $wpdb->delete(
3468 + $table_name,
3469 + array('source_url' => $source_url),
3470 + array('%s')
3471 + );
3472 + }
3473 +}
3474 +
3475 +/**
3476 + * Handle individual Pinecone content deletion
3477 + */
3478 +public function mxchat_handle_pinecone_prompt_delete() {
3479 + // Check permissions
3480 + if (!current_user_can('manage_options')) {
3481 + wp_die(esc_html__('You do not have sufficient permissions.', 'mxchat'));
3482 + }
3483 +
3484 + // Verify nonce
3485 + if (!isset($_GET['_wpnonce']) || !wp_verify_nonce($_GET['_wpnonce'], 'mxchat_delete_pinecone_prompt_nonce')) {
3486 + wp_die(esc_html__('Security check failed.', 'mxchat'));
3487 + }
3488 +
3489 + $vector_id = isset($_GET['vector_id']) ? sanitize_text_field($_GET['vector_id']) : '';
3490 +
3491 + if (empty($vector_id)) {
3492 + set_transient('mxchat_admin_notice_error',
3493 + esc_html__('Invalid vector ID.', 'mxchat'),
3494 + 30
3495 + );
3496 + wp_safe_redirect(admin_url('admin.php?page=mxchat-prompts'));
3497 + exit;
3498 + }
3499 +
3500 + // Get Pinecone settings
3501 + $pinecone_options = get_option('mxchat_pinecone_addon_options', array());
3502 + $use_pinecone = ($pinecone_options['mxchat_use_pinecone'] ?? '0') === '1';
3503 +
3504 + if (!$use_pinecone || empty($pinecone_options['mxchat_pinecone_api_key'])) {
3505 + set_transient('mxchat_admin_notice_error',
3506 + esc_html__('Pinecone is not properly configured.', 'mxchat'),
3507 + 30
3508 + );
3509 + wp_safe_redirect(admin_url('admin.php?page=mxchat-prompts'));
3510 + exit;
3511 + }
3512 +
3513 + // Delete from Pinecone
3514 + $pinecone_manager = MxChat_Pinecone_Manager::get_instance();
3515 + $result = $pinecone_manager->mxchat_delete_from_pinecone_by_vector_id(
3516 + $vector_id,
3517 + $pinecone_options['mxchat_pinecone_api_key'],
3518 + $pinecone_options['mxchat_pinecone_host']
3519 + );
3520 +
3521 + if ($result['success']) {
3522 + // No cache clearing needed since we removed caching
3523 + set_transient('mxchat_admin_notice_success',
3524 + esc_html__('Entry deleted successfully from Pinecone.', 'mxchat'),
3525 + 30
3526 + );
3527 + } else {
3528 + set_transient('mxchat_admin_notice_error',
3529 + esc_html__('Failed to delete entry: ', 'mxchat') . $result['message'],
3530 + 30
3531 + );
3532 + }
3533 +
3534 + wp_safe_redirect(admin_url('admin.php?page=mxchat-prompts'));
3535 + exit;
3536 +}
3537 +/**
3538 + * Handle individual Pinecone content deletion via AJAX
3539 + */
3540 +public function ajax_mxchat_delete_pinecone_prompt() {
3541 + // Verify nonce and permissions
3542 + if (!check_ajax_referer('mxchat_delete_pinecone_prompt_nonce', 'nonce', false)) {
3543 + wp_send_json_error('Invalid nonce');
3544 + exit;
3545 + }
3546 +
3547 + if (!current_user_can('manage_options')) {
3548 + wp_send_json_error('Unauthorized access');
3549 + exit;
3550 + }
3551 +
3552 + $vector_id = isset($_POST['vector_id']) ? sanitize_text_field($_POST['vector_id']) : '';
3553 + $bot_id = isset($_POST['bot_id']) ? sanitize_text_field($_POST['bot_id']) : 'default';
3554 +
3555 + if (empty($vector_id)) {
3556 + wp_send_json_error('Missing vector ID');
3557 + exit;
3558 + }
3559 +
3560 + // Get bot-specific Pinecone settings
3561 + $pinecone_manager = MxChat_Pinecone_Manager::get_instance();
3562 + $pinecone_options = $pinecone_manager->mxchat_get_bot_pinecone_options($bot_id);
3563 +
3564 + $use_pinecone = ($pinecone_options['mxchat_use_pinecone'] ?? '0') === '1';
3565 +
3566 + if (!$use_pinecone || empty($pinecone_options['mxchat_pinecone_api_key'])) {
3567 + wp_send_json_error('Pinecone is not properly configured for bot: ' . $bot_id);
3568 + exit;
3569 + }
3570 +
3571 + // Delete from the correct Pinecone index
3572 + $result = $pinecone_manager->mxchat_delete_from_pinecone_by_vector_id(
3573 + $vector_id,
3574 + $pinecone_options['mxchat_pinecone_api_key'],
3575 + $pinecone_options['mxchat_pinecone_host']
3576 + );
3577 +
3578 + if ($result['success']) {
3579 + // No cache clearing needed since we removed caching
3580 + wp_send_json_success(array(
3581 + 'message' => 'Entry deleted successfully from Pinecone',
3582 + 'vector_id' => $vector_id,
3583 + 'bot_id' => $bot_id
3584 + ));
3585 + } else {
3586 + wp_send_json_error('Failed to delete from Pinecone: ' . $result['message']);
3587 + }
3588 +
3589 + exit;
3590 +}
3591 +
3592 +/**
3593 + * Get hierarchical roles for dropdown
3594 + */
3595 +public function mxchat_get_role_options() {
3596 + return array(
3597 + 'public' => __('Public (Everyone)', 'mxchat'),
3598 + 'logged_in' => __('Logged In Users', 'mxchat'),
3599 + 'subscriber' => __('Subscribers & Above', 'mxchat'),
3600 + 'contributor' => __('Contributors & Above', 'mxchat'),
3601 + 'author' => __('Authors & Above', 'mxchat'),
3602 + 'editor' => __('Editors & Above', 'mxchat'),
3603 + 'administrator' => __('Administrators Only', 'mxchat')
3604 + );
3605 +}
3606 +
3607 +/**
3608 + * Check if user has access to content based on role restriction
3609 + */
3610 +public function mxchat_user_has_content_access($role_restriction) {
3611 + // Public content is always accessible
3612 + if ($role_restriction === 'public' || empty($role_restriction)) {
3613 + return true;
3614 + }
3615 +
3616 + // Check if user is logged in for logged_in restriction
3617 + if ($role_restriction === 'logged_in') {
3618 + return is_user_logged_in();
3619 + }
3620 +
3621 + // If not logged in, no access to role-restricted content
3622 + if (!is_user_logged_in()) {
3623 + return false;
3624 + }
3625 +
3626 + $user = wp_get_current_user();
3627 + $user_roles = $user->roles;
3628 +
3629 + if (empty($user_roles)) {
3630 + return false;
3631 + }
3632 +
3633 + // Define role hierarchy (higher number = higher access)
3634 + $hierarchy = array(
3635 + 'subscriber' => 1,
3636 + 'contributor' => 2,
3637 + 'author' => 3,
3638 + 'editor' => 4,
3639 + 'administrator' => 5
3640 + );
3641 +
3642 + // Get required level
3643 + $required_level = isset($hierarchy[$role_restriction]) ? $hierarchy[$role_restriction] : 0;
3644 +
3645 + // Check if user has required level or higher
3646 + foreach ($user_roles as $user_role) {
3647 + $user_level = isset($hierarchy[$user_role]) ? $hierarchy[$user_role] : 0;
3648 + if ($user_level >= $required_level) {
3649 + return true;
3650 + }
3651 + }
3652 +
3653 + return false;
3654 +}
3655 +
3656 +/**
3657 + * Handle role restriction updates via AJAX
3658 + * Removed cache clearing call since we removed caching
3659 + */
3660 +public function ajax_mxchat_update_role_restriction() {
3661 + // Verify nonce and permissions
3662 + if (!check_ajax_referer('mxchat_update_role_nonce', 'nonce', false)) {
3663 + wp_send_json_error('Invalid nonce');
3664 + exit;
3665 + }
3666 +
3667 + if (!current_user_can('manage_options')) {
3668 + wp_send_json_error('Unauthorized access');
3669 + exit;
3670 + }
3671 +
3672 + $entry_id = isset($_POST['entry_id']) ? sanitize_text_field($_POST['entry_id']) : '';
3673 + $role_restriction = isset($_POST['role_restriction']) ? sanitize_text_field($_POST['role_restriction']) : 'public';
3674 + $data_source = isset($_POST['data_source']) ? sanitize_text_field($_POST['data_source']) : 'wordpress';
3675 +
3676 + if (empty($entry_id)) {
3677 + wp_send_json_error('Invalid entry ID');
3678 + exit;
3679 + }
3680 +
3681 + // Get knowledge manager instance to validate role restriction
3682 + $knowledge_manager = MxChat_Knowledge_Manager::get_instance();
3683 + $valid_roles = array_keys($knowledge_manager->mxchat_get_role_options());
3684 + if (!in_array($role_restriction, $valid_roles)) {
3685 + wp_send_json_error('Invalid role restriction');
3686 + exit;
3687 + }
3688 +
3689 + global $wpdb;
3690 +
3691 + if ($data_source === 'pinecone') {
3692 + // Handle Pinecone role restriction (stored separately in WordPress table)
3693 + $roles_table = $wpdb->prefix . 'mxchat_pinecone_roles';
3694 +
3695 + // Use REPLACE to insert or update the role restriction
3696 + $result = $wpdb->replace(
3697 + $roles_table,
3698 + array(
3699 + 'vector_id' => $entry_id,
3700 + 'role_restriction' => $role_restriction,
3701 + 'updated_at' => current_time('mysql')
3702 + ),
3703 + array('%s', '%s', '%s')
3704 + );
3705 +
3706 + // No cache clearing needed since we removed caching
3707 +
3708 + } else {
3709 + // Handle WordPress database role restriction (existing functionality)
3710 + $table_name = $wpdb->prefix . 'mxchat_system_prompt_content';
3711 +
3712 + $result = $wpdb->update(
3713 + $table_name,
3714 + array('role_restriction' => $role_restriction),
3715 + array('id' => absint($entry_id)),
3716 + array('%s'),
3717 + array('%d')
3718 + );
3719 + }
3720 +
3721 + if ($result === false) {
3722 + wp_send_json_error('Database update failed: ' . $wpdb->last_error);
3723 + exit;
3724 + }
3725 +
3726 + wp_send_json_success(array(
3727 + 'message' => 'Role restriction updated successfully',
3728 + 'role_restriction' => $role_restriction,
3729 + 'data_source' => $data_source,
3730 + 'entry_id' => $entry_id
3731 + ));
3732 + exit;
3733 +}
3734 +
3735 +// ========================================
3736 +// ROLE-BASED CONTENT RESTRICTIONS - NEW FUNCTIONS
3737 +// Add these to your MxChat_Knowledge_Manager class
3738 +// ========================================
3739 +
3740 +/**
3741 + * Initialize role-based content hooks
3742 + * Add this call to your __construct() or mxchat_init_hooks() method
3743 + */
3744 +private function mxchat_init_role_hooks() {
3745 + // AJAX handlers for tag-role mappings
3746 + add_action('wp_ajax_mxchat_add_tag_role_mapping', array($this, 'ajax_add_tag_role_mapping'));
3747 + add_action('wp_ajax_mxchat_delete_tag_role_mapping', array($this, 'ajax_delete_tag_role_mapping'));
3748 + add_action('wp_ajax_mxchat_get_tag_role_mappings', array($this, 'ajax_get_tag_role_mappings'));
3749 + add_action('wp_ajax_mxchat_bulk_update_tag_roles', array($this, 'ajax_bulk_update_tag_roles'));
3750 +
3751 + // Hook to automatically update role restrictions when tags are added/removed
3752 + add_action('set_object_terms', array($this, 'handle_tag_change'), 10, 6);
3753 +
3754 + // Hook to apply role restrictions on auto-sync
3755 + add_action('mxchat_content_stored', array($this, 'apply_role_restriction_after_storage'), 10, 2);
3756 +}
3757 +
3758 +/**
3759 + * Add tag-role mapping via AJAX
3760 + */
3761 +public function ajax_add_tag_role_mapping() {
3762 + // Verify nonce and permissions
3763 + check_ajax_referer('mxchat_prompts_setting_nonce', 'nonce');
3764 +
3765 + if (!current_user_can('manage_options')) {
3766 + wp_send_json_error('Unauthorized access');
3767 + exit;
3768 + }
3769 +
3770 + $tag_slug = isset($_POST['tag_slug']) ? sanitize_text_field($_POST['tag_slug']) : '';
3771 + $role_restriction = isset($_POST['role_restriction']) ? sanitize_text_field($_POST['role_restriction']) : 'public';
3772 +
3773 + if (empty($tag_slug)) {
3774 + wp_send_json_error('Tag slug is required');
3775 + exit;
3776 + }
3777 +
3778 + // Validate role restriction
3779 + $valid_roles = array_keys($this->mxchat_get_role_options());
3780 + if (!in_array($role_restriction, $valid_roles)) {
3781 + wp_send_json_error('Invalid role restriction');
3782 + exit;
3783 + }
3784 +
3785 + // Check if tag exists in WordPress
3786 + $term = get_term_by('slug', $tag_slug, 'post_tag');
3787 + if (!$term) {
3788 + wp_send_json_error('Tag does not exist in WordPress');
3789 + exit;
3790 + }
3791 +
3792 + // Get existing mappings
3793 + $mappings = get_option('mxchat_tag_role_mappings', array());
3794 +
3795 + // Check if mapping already exists
3796 + if (isset($mappings[$tag_slug])) {
3797 + wp_send_json_error('Mapping for this tag already exists');
3798 + exit;
3799 + }
3800 +
3801 + // Add new mapping
3802 + $mappings[$tag_slug] = $role_restriction;
3803 + update_option('mxchat_tag_role_mappings', $mappings);
3804 +
3805 + wp_send_json_success(array(
3806 + 'message' => 'Tag-role mapping added successfully',
3807 + 'tag_slug' => $tag_slug,
3808 + 'role_restriction' => $role_restriction
3809 + ));
3810 + exit;
3811 +}
3812 +
3813 +/**
3814 + * Delete tag-role mapping via AJAX
3815 + */
3816 +public function ajax_delete_tag_role_mapping() {
3817 + // Verify nonce and permissions
3818 + check_ajax_referer('mxchat_prompts_setting_nonce', 'nonce');
3819 +
3820 + if (!current_user_can('manage_options')) {
3821 + wp_send_json_error('Unauthorized access');
3822 + exit;
3823 + }
3824 +
3825 + $tag_slug = isset($_POST['tag_slug']) ? sanitize_text_field($_POST['tag_slug']) : '';
3826 +
3827 + if (empty($tag_slug)) {
3828 + wp_send_json_error('Tag slug is required');
3829 + exit;
3830 + }
3831 +
3832 + // Get existing mappings
3833 + $mappings = get_option('mxchat_tag_role_mappings', array());
3834 +
3835 + // Check if mapping exists
3836 + if (!isset($mappings[$tag_slug])) {
3837 + wp_send_json_error('Mapping does not exist');
3838 + exit;
3839 + }
3840 +
3841 + // Remove mapping
3842 + unset($mappings[$tag_slug]);
3843 + update_option('mxchat_tag_role_mappings', $mappings);
3844 +
3845 + wp_send_json_success(array(
3846 + 'message' => 'Tag-role mapping deleted successfully',
3847 + 'tag_slug' => $tag_slug
3848 + ));
3849 + exit;
3850 +}
3851 +
3852 +/**
3853 + * Get all tag-role mappings via AJAX
3854 + */
3855 +public function ajax_get_tag_role_mappings() {
3856 + // Verify nonce and permissions
3857 + check_ajax_referer('mxchat_prompts_setting_nonce', 'nonce');
3858 +
3859 + if (!current_user_can('manage_options')) {
3860 + wp_send_json_error('Unauthorized access');
3861 + exit;
3862 + }
3863 +
3864 + // Get mappings
3865 + $mappings = get_option('mxchat_tag_role_mappings', array());
3866 + $role_options = $this->mxchat_get_role_options();
3867 +
3868 + $formatted_mappings = array();
3869 +
3870 + foreach ($mappings as $tag_slug => $role_restriction) {
3871 + // Get tag object
3872 + $term = get_term_by('slug', $tag_slug, 'post_tag');
3873 +
3874 + // Count posts with this tag
3875 + $post_count = 0;
3876 + if ($term) {
3877 + $post_count = $term->count;
3878 + }
3879 +
3880 + $formatted_mappings[] = array(
3881 + 'tag_slug' => $tag_slug,
3882 + 'role_restriction' => $role_restriction,
3883 + 'role_label' => $role_options[$role_restriction] ?? $role_restriction,
3884 + 'post_count' => $post_count
3885 + );
3886 + }
3887 +
3888 + wp_send_json_success(array(
3889 + 'mappings' => $formatted_mappings
3890 + ));
3891 + exit;
3892 +}
3893 +
3894 +/**
3895 + * Bulk update role restrictions for all existing content with mapped tags
3896 + */
3897 +public function ajax_bulk_update_tag_roles() {
3898 + // Verify nonce and permissions
3899 + check_ajax_referer('mxchat_prompts_setting_nonce', 'nonce');
3900 +
3901 + if (!current_user_can('manage_options')) {
3902 + wp_send_json_error('Unauthorized access');
3903 + exit;
3904 + }
3905 +
3906 + // Get mappings
3907 + $mappings = get_option('mxchat_tag_role_mappings', array());
3908 +
3909 + if (empty($mappings)) {
3910 + wp_send_json_error('No tag-role mappings found');
3911 + exit;
3912 + }
3913 +
3914 + global $wpdb;
3915 +
3916 + // Check if using Pinecone
3917 + $pinecone_options = get_option('mxchat_pinecone_addon_options', array());
3918 + $use_pinecone = ($pinecone_options['mxchat_use_pinecone'] ?? '0') === '1';
3919 +
3920 + $updated_count = 0;
3921 + $details = array();
3922 +
3923 + foreach ($mappings as $tag_slug => $role_restriction) {
3924 + // Get all posts with this tag
3925 + $posts = get_posts(array(
3926 + 'tag' => $tag_slug,
3927 + 'post_type' => 'any',
3928 + 'posts_per_page' => -1,
3929 + 'fields' => 'ids',
3930 + 'post_status' => 'publish'
3931 + ));
3932 +
3933 + if (empty($posts)) {
3934 + continue;
3935 + }
3936 +
3937 + $tag_updated = 0;
3938 +
3939 + foreach ($posts as $post_id) {
3940 + $source_url = get_permalink($post_id);
3941 + if (!$source_url) {
3942 + continue;
3943 + }
3944 +
3945 + if ($use_pinecone && !empty($pinecone_options['mxchat_pinecone_api_key'])) {
3946 + // Update Pinecone role restriction
3947 + $roles_table = $wpdb->prefix . 'mxchat_pinecone_roles';
3948 + $vector_id = md5($source_url);
3949 +
3950 + $result = $wpdb->replace(
3951 + $roles_table,
3952 + array(
3953 + 'vector_id' => $vector_id,
3954 + 'role_restriction' => $role_restriction,
3955 + 'updated_at' => current_time('mysql')
3956 + ),
3957 + array('%s', '%s', '%s')
3958 + );
3959 + } else {
3960 + // Update WordPress DB
3961 + $table_name = $wpdb->prefix . 'mxchat_system_prompt_content';
3962 +
3963 + $result = $wpdb->update(
3964 + $table_name,
3965 + array('role_restriction' => $role_restriction),
3966 + array('source_url' => $source_url),
3967 + array('%s'),
3968 + array('%s')
3969 + );
3970 + }
3971 +
3972 + if ($result !== false) {
3973 + $tag_updated++;
3974 + $updated_count++;
3975 + }
3976 + }
3977 +
3978 + if ($tag_updated > 0) {
3979 + $details[] = sprintf(
3980 + 'Tag "%s" (%s): %d posts updated',
3981 + $tag_slug,
3982 + $role_restriction,
3983 + $tag_updated
3984 + );
3985 + }
3986 + }
3987 +
3988 + wp_send_json_success(array(
3989 + 'message' => 'Bulk update completed',
3990 + 'updated_count' => $updated_count,
3991 + 'tags_processed' => count($mappings),
3992 + 'details' => $details
3993 + ));
3994 + exit;
3995 +}
3996 +
3997 +/**
3998 + * Handle tag changes on posts (when tags are added or removed)
3999 + */
4000 +public function handle_tag_change($object_id, $terms, $tt_ids, $taxonomy, $append, $old_tt_ids) {
4001 + // Only process post tags
4002 + if ($taxonomy !== 'post_tag') {
4003 + return;
4004 + }
4005 +
4006 + // Get tag-role mappings
4007 + $mappings = get_option('mxchat_tag_role_mappings', array());
4008 +
4009 + if (empty($mappings)) {
4010 + return;
4011 + }
4012 +
4013 + // Get the post's URL
4014 + $source_url = get_permalink($object_id);
4015 + if (!$source_url) {
4016 + return;
4017 + }
4018 +
4019 + // Determine the highest role restriction based on tags
4020 + $highest_role = 'public';
4021 + $role_hierarchy = array(
4022 + 'public' => 0,
4023 + 'logged_in' => 1,
4024 + 'subscriber' => 2,
4025 + 'contributor' => 3,
4026 + 'author' => 4,
4027 + 'editor' => 5,
4028 + 'administrator' => 6
4029 + );
4030 +
4031 + // Get all current tags for the post
4032 + $current_tags = wp_get_post_tags($object_id, array('fields' => 'slugs'));
4033 +
4034 + // Find the highest role restriction among the tags
4035 + foreach ($current_tags as $tag_slug) {
4036 + if (isset($mappings[$tag_slug])) {
4037 + $role = $mappings[$tag_slug];
4038 + if (isset($role_hierarchy[$role]) && $role_hierarchy[$role] > $role_hierarchy[$highest_role]) {
4039 + $highest_role = $role;
4040 + }
4041 + }
4042 + }
4043 +
4044 + // Update the role restriction in the database
4045 + global $wpdb;
4046 +
4047 + // Check if using Pinecone
4048 + $pinecone_options = get_option('mxchat_pinecone_addon_options', array());
4049 + $use_pinecone = ($pinecone_options['mxchat_use_pinecone'] ?? '0') === '1';
4050 +
4051 + if ($use_pinecone && !empty($pinecone_options['mxchat_pinecone_api_key'])) {
4052 + // Update Pinecone role restriction
4053 + $roles_table = $wpdb->prefix . 'mxchat_pinecone_roles';
4054 + $vector_id = md5($source_url);
4055 +
4056 + $wpdb->replace(
4057 + $roles_table,
4058 + array(
4059 + 'vector_id' => $vector_id,
4060 + 'role_restriction' => $highest_role,
4061 + 'updated_at' => current_time('mysql')
4062 + ),
4063 + array('%s', '%s', '%s')
4064 + );
4065 + } else {
4066 + // Update WordPress DB
4067 + $table_name = $wpdb->prefix . 'mxchat_system_prompt_content';
4068 +
4069 + $wpdb->update(
4070 + $table_name,
4071 + array('role_restriction' => $highest_role),
4072 + array('source_url' => $source_url),
4073 + array('%s'),
4074 + array('%s')
4075 + );
4076 + }
4077 +}
4078 +
4079 +/**
4080 + * Apply role restriction after content is stored (for auto-sync)
4081 + */
4082 +public function apply_role_restriction_after_storage($post_id, $source_url) {
4083 + // Get tag-role mappings
4084 + $mappings = get_option('mxchat_tag_role_mappings', array());
4085 +
4086 + if (empty($mappings)) {
4087 + return;
4088 + }
4089 +
4090 + // Get all tags for the post
4091 + $post_tags = wp_get_post_tags($post_id, array('fields' => 'slugs'));
4092 +
4093 + if (empty($post_tags)) {
4094 + return;
4095 + }
4096 +
4097 + // Determine the highest role restriction based on tags
4098 + $highest_role = 'public';
4099 + $role_hierarchy = array(
4100 + 'public' => 0,
4101 + 'logged_in' => 1,
4102 + 'subscriber' => 2,
4103 + 'contributor' => 3,
4104 + 'author' => 4,
4105 + 'editor' => 5,
4106 + 'administrator' => 6
4107 + );
4108 +
4109 + foreach ($post_tags as $tag_slug) {
4110 + if (isset($mappings[$tag_slug])) {
4111 + $role = $mappings[$tag_slug];
4112 + if (isset($role_hierarchy[$role]) && $role_hierarchy[$role] > $role_hierarchy[$highest_role]) {
4113 + $highest_role = $role;
4114 + }
4115 + }
4116 + }
4117 +
4118 + // If no restricted tags found, return (leave as public)
4119 + if ($highest_role === 'public') {
4120 + return;
4121 + }
4122 +
4123 + // Update the role restriction
4124 + global $wpdb;
4125 +
4126 + // Check if using Pinecone
4127 + $pinecone_options = get_option('mxchat_pinecone_addon_options', array());
4128 + $use_pinecone = ($pinecone_options['mxchat_use_pinecone'] ?? '0') === '1';
4129 +
4130 + if ($use_pinecone && !empty($pinecone_options['mxchat_pinecone_api_key'])) {
4131 + // Update Pinecone role restriction
4132 + $roles_table = $wpdb->prefix . 'mxchat_pinecone_roles';
4133 + $vector_id = md5($source_url);
4134 +
4135 + $wpdb->replace(
4136 + $roles_table,
4137 + array(
4138 + 'vector_id' => $vector_id,
4139 + 'role_restriction' => $highest_role,
4140 + 'updated_at' => current_time('mysql')
4141 + ),
4142 + array('%s', '%s', '%s')
4143 + );
4144 + } else {
4145 + // Update WordPress DB
4146 + $table_name = $wpdb->prefix . 'mxchat_system_prompt_content';
4147 +
4148 + $wpdb->update(
4149 + $table_name,
4150 + array('role_restriction' => $highest_role),
4151 + array('source_url' => $source_url),
4152 + array('%s'),
4153 + array('%s')
4154 + );
4155 + }
4156 +}
4157 +
4158 +
4159 + // ========================================
4160 + // HELPER METHODS
4161 + // ========================================
4162 +
4163 + /**
4164 + * Check if user has required permissions for content processing
4165 + */
4166 + private function mxchat_check_user_permissions() {
4167 + if (!current_user_can('manage_options')) {
4168 + wp_die(esc_html__('You do not have sufficient permissions.', 'mxchat'));
4169 + }
4170 + }
4171 +
4172 + /**
4173 + * Validate nonce for security
4174 + */
4175 + private function mxchat_validate_nonce($nonce_name, $nonce_action) {
4176 + if (!isset($_POST[$nonce_name]) || !wp_verify_nonce($_POST[$nonce_name], $nonce_action)) {
4177 + wp_die(esc_html__('Security check failed.', 'mxchat'));
4178 + }
4179 + }
4180 +
4181 + /**
4182 + * Get embedding API credentials
4183 + */
4184 + private function mxchat_get_embedding_credentials() {
4185 + $embedding_model = $this->options['embedding_model'] ?? 'text-embedding-ada-002';
4186 +
4187 + if (strpos($embedding_model, 'text-embedding-') !== false) {
4188 + return array(
4189 + 'type' => 'openai',
4190 + 'api_key' => $this->options['api_key'] ?? ''
4191 + );
4192 + } elseif (strpos($embedding_model, 'voyage-') !== false) {
4193 + return array(
4194 + 'type' => 'voyage',
4195 + 'api_key' => $this->options['voyage_api_key'] ?? ''
4196 + );
4197 + } elseif (strpos($embedding_model, 'gemini-embedding-') !== false) {
4198 + return array(
4199 + 'type' => 'gemini',
4200 + 'api_key' => $this->options['gemini_api_key'] ?? ''
4201 + );
4202 + }
4203 +
4204 + return array('type' => 'unknown', 'api_key' => '');
4205 + }
4206 +
4207 + /**
4208 + * Log processing errors
4209 + */
4210 + private function mxchat_log_processing_error($operation, $error_message) {
4211 + //error_log("MxChat Knowledge Processing {$operation} Error: " . $error_message);
4212 + }
4213 +
4214 + /**
4215 + * Set admin notice transient
4216 + */
4217 + private function mxchat_set_admin_notice($type, $message) {
4218 + set_transient("mxchat_admin_notice_{$type}", $message, 30);
4219 + }
4220 +
4221 + /**
4222 + * Get Pinecone manager instance for vector operations
4223 + */
4224 + private function mxchat_get_pinecone_manager() {
4225 + return MxChat_Pinecone_Manager::get_instance();
4226 + }
4227 +
4228 +
4229 + // ========================================
4230 +// DATABASE QUEUE TABLE MANAGEMENT
4231 +// ========================================
4232 +
4233 +/**
4234 + * Create queue table on plugin activation
4235 + * Call this from your plugin activation hook
4236 + */
4237 +public function mxchat_create_queue_table() {
4238 + global $wpdb;
4239 +
4240 + $table_name = $wpdb->prefix . 'mxchat_processing_queue';
4241 + $charset_collate = $wpdb->get_charset_collate();
4242 +
4243 + $sql = "CREATE TABLE IF NOT EXISTS $table_name (
4244 + id bigint(20) unsigned NOT NULL AUTO_INCREMENT,
4245 + queue_id varchar(64) NOT NULL,
4246 + item_type varchar(20) NOT NULL,
4247 + item_data longtext NOT NULL,
4248 + status varchar(20) NOT NULL DEFAULT 'pending',
4249 + bot_id varchar(50) NOT NULL DEFAULT 'default',
4250 + priority int(11) NOT NULL DEFAULT 0,
4251 + attempts int(11) NOT NULL DEFAULT 0,
4252 + max_attempts int(11) NOT NULL DEFAULT 3,
4253 + error_message text DEFAULT NULL,
4254 + created_at datetime NOT NULL,
4255 + started_at datetime DEFAULT NULL,
4256 + completed_at datetime DEFAULT NULL,
4257 + PRIMARY KEY (id),
4258 + KEY queue_id (queue_id),
4259 + KEY status (status),
4260 + KEY item_type (item_type),
4261 + KEY priority (priority)
4262 + ) $charset_collate;";
4263 +
4264 + require_once(ABSPATH . 'wp-admin/includes/upgrade.php');
4265 + dbDelta($sql);
4266 +
4267 + // Also create a meta table for queue metadata
4268 + $meta_table = $wpdb->prefix . 'mxchat_queue_meta';
4269 +
4270 + $meta_sql = "CREATE TABLE IF NOT EXISTS $meta_table (
4271 + id bigint(20) unsigned NOT NULL AUTO_INCREMENT,
4272 + queue_id varchar(64) NOT NULL,
4273 + meta_key varchar(255) NOT NULL,
4274 + meta_value longtext,
4275 + PRIMARY KEY (id),
4276 + KEY queue_id (queue_id),
4277 + KEY meta_key (meta_key)
4278 + ) $charset_collate;";
4279 +
4280 + dbDelta($meta_sql);
4281 +}
4282 +
4283 +/**
4284 + * Add items to the processing queue
4285 + *
4286 + * @param string $queue_id Unique identifier for this queue batch
4287 + * @param string $item_type Type of item (url, pdf_page)
4288 + * @param array $items Array of items to queue
4289 + * @param string $bot_id Bot ID for processing
4290 + * @return int Number of items queued
4291 + */
4292 +private function mxchat_add_to_queue($queue_id, $item_type, $items, $bot_id = 'default') {
4293 + global $wpdb;
4294 + $table_name = $wpdb->prefix . 'mxchat_processing_queue';
4295 +
4296 + $queued_count = 0;
4297 + $priority = 0;
4298 +
4299 + foreach ($items as $item) {
4300 + $result = $wpdb->insert(
4301 + $table_name,
4302 + array(
4303 + 'queue_id' => $queue_id,
4304 + 'item_type' => $item_type,
4305 + 'item_data' => wp_json_encode($item),
4306 + 'status' => 'pending',
4307 + 'bot_id' => $bot_id,
4308 + 'priority' => $priority,
4309 + 'attempts' => 0,
4310 + 'max_attempts' => 3,
4311 + 'created_at' => current_time('mysql')
4312 + ),
4313 + array('%s', '%s', '%s', '%s', '%s', '%d', '%d', '%d', '%s')
4314 + );
4315 +
4316 + if ($result) {
4317 + $queued_count++;
4318 + }
4319 +
4320 + $priority++; // Process in order
4321 + }
4322 +
4323 + return $queued_count;
4324 +}
4325 +
4326 +/**
4327 + * Store queue metadata (total counts, source URL, etc.)
4328 + */
4329 +private function mxchat_set_queue_meta($queue_id, $meta_key, $meta_value) {
4330 + global $wpdb;
4331 + $meta_table = $wpdb->prefix . 'mxchat_queue_meta';
4332 +
4333 + // Check if meta exists
4334 + $existing = $wpdb->get_var($wpdb->prepare(
4335 + "SELECT id FROM $meta_table WHERE queue_id = %s AND meta_key = %s",
4336 + $queue_id,
4337 + $meta_key
4338 + ));
4339 +
4340 + if ($existing) {
4341 + // Update
4342 + $wpdb->update(
4343 + $meta_table,
4344 + array('meta_value' => maybe_serialize($meta_value)),
4345 + array('queue_id' => $queue_id, 'meta_key' => $meta_key),
4346 + array('%s'),
4347 + array('%s', '%s')
4348 + );
4349 + } else {
4350 + // Insert
4351 + $wpdb->insert(
4352 + $meta_table,
4353 + array(
4354 + 'queue_id' => $queue_id,
4355 + 'meta_key' => $meta_key,
4356 + 'meta_value' => maybe_serialize($meta_value)
4357 + ),
4358 + array('%s', '%s', '%s')
4359 + );
4360 + }
4361 +}
4362 +
4363 +/**
4364 + * Get queue metadata
4365 + */
4366 +private function mxchat_get_queue_meta($queue_id, $meta_key) {
4367 + global $wpdb;
4368 + $meta_table = $wpdb->prefix . 'mxchat_queue_meta';
4369 +
4370 + $value = $wpdb->get_var($wpdb->prepare(
4371 + "SELECT meta_value FROM $meta_table WHERE queue_id = %s AND meta_key = %s",
4372 + $queue_id,
4373 + $meta_key
4374 + ));
4375 +
4376 + return maybe_unserialize($value);
4377 +}
4378 +
4379 +// ========================================
4380 +// AJAX QUEUE PROCESSING HANDLERS
4381 +// ========================================
4382 +
4383 +/**
4384 + * AJAX: Get next item from queue to process
4385 + */
4386 +public function ajax_mxchat_get_next_queue_item() {
4387 + // Verify nonce and permissions
4388 + check_ajax_referer('mxchat_queue_nonce', 'nonce');
4389 +
4390 + if (!current_user_can('manage_options')) {
4391 + wp_send_json_error('Unauthorized access');
4392 + }
4393 +
4394 + $queue_id = isset($_POST['queue_id']) ? sanitize_text_field($_POST['queue_id']) : '';
4395 +
4396 + if (empty($queue_id)) {
4397 + wp_send_json_error('Missing queue ID');
4398 + }
4399 +
4400 + global $wpdb;
4401 + $table_name = $wpdb->prefix . 'mxchat_processing_queue';
4402 +
4403 + // Get next pending item with retry logic for failed items
4404 + $next_item = $wpdb->get_row($wpdb->prepare(
4405 + "SELECT * FROM $table_name
4406 + WHERE queue_id = %s
4407 + AND status IN ('pending', 'failed')
4408 + AND attempts < max_attempts
4409 + ORDER BY priority ASC, id ASC
4410 + LIMIT 1",
4411 + $queue_id
4412 + ));
4413 +
4414 + if (!$next_item) {
4415 + // No more items - queue complete
4416 + wp_send_json_success(array(
4417 + 'complete' => true,
4418 + 'message' => 'Queue processing complete'
4419 + ));
4420 + }
4421 +
4422 + // Mark item as processing
4423 + $wpdb->update(
4424 + $table_name,
4425 + array(
4426 + 'status' => 'processing',
4427 + 'started_at' => current_time('mysql'),
4428 + 'attempts' => $next_item->attempts + 1
4429 + ),
4430 + array('id' => $next_item->id),
4431 + array('%s', '%s', '%d'),
4432 + array('%d')
4433 + );
4434 +
4435 + wp_send_json_success(array(
4436 + 'complete' => false,
4437 + 'item' => array(
4438 + 'id' => $next_item->id,
4439 + 'type' => $next_item->item_type,
4440 + 'data' => json_decode($next_item->item_data, true),
4441 + 'bot_id' => $next_item->bot_id,
4442 + 'attempt' => $next_item->attempts + 1
4443 + )
4444 + ));
4445 +}
4446 +
4447 +/**
4448 + * AJAX: Process a single queue item
4449 + */
4450 +public function ajax_mxchat_process_queue_item() {
4451 + // Verify nonce and permissions
4452 + check_ajax_referer('mxchat_queue_nonce', 'nonce');
4453 +
4454 + if (!current_user_can('manage_options')) {
4455 + wp_send_json_error('Unauthorized access');
4456 + }
4457 +
4458 + $item_id = isset($_POST['item_id']) ? absint($_POST['item_id']) : 0;
4459 + $item_type = isset($_POST['item_type']) ? sanitize_text_field($_POST['item_type']) : '';
4460 + $item_data = isset($_POST['item_data']) ? $_POST['item_data'] : array();
4461 + $bot_id = isset($_POST['bot_id']) ? sanitize_key($_POST['bot_id']) : 'default';
4462 +
4463 + if (empty($item_id) || empty($item_type)) {
4464 + wp_send_json_error('Missing item data');
4465 + }
4466 +
4467 + global $wpdb;
4468 + $table_name = $wpdb->prefix . 'mxchat_processing_queue';
4469 +
4470 + // Process based on item type
4471 + try {
4472 + set_time_limit(60); // Give processing 60 seconds
4473 +
4474 + $result = false;
4475 + $error_message = '';
4476 +
4477 + switch ($item_type) {
4478 + case 'url':
4479 + $result = $this->mxchat_process_queue_url($item_data, $bot_id);
4480 + break;
4481 +
4482 + case 'pdf_page':
4483 + $result = $this->mxchat_process_queue_pdf_page($item_data, $bot_id);
4484 + break;
4485 +
4486 + default:
4487 + throw new Exception('Unknown item type: ' . $item_type);
4488 + }
4489 +
4490 + if (is_wp_error($result)) {
4491 + throw new Exception($result->get_error_message());
4492 + }
4493 +
4494 + if ($result === false) {
4495 + throw new Exception('Processing returned false - item may be empty or invalid');
4496 + }
4497 +
4498 + // Mark as completed
4499 + $wpdb->update(
4500 + $table_name,
4501 + array(
4502 + 'status' => 'completed',
4503 + 'completed_at' => current_time('mysql'),
4504 + 'error_message' => null
4505 + ),
4506 + array('id' => $item_id),
4507 + array('%s', '%s', '%s'),
4508 + array('%d')
4509 + );
4510 +
4511 + wp_send_json_success(array(
4512 + 'processed' => true,
4513 + 'item_id' => $item_id,
4514 + 'message' => 'Item processed successfully'
4515 + ));
4516 +
4517 + } catch (Exception $e) {
4518 + $error_message = $e->getMessage();
4519 +
4520 + // Get current attempt count
4521 + $item = $wpdb->get_row($wpdb->prepare(
4522 + "SELECT attempts, max_attempts FROM $table_name WHERE id = %d",
4523 + $item_id
4524 + ));
4525 +
4526 + // Check if we've exhausted retries
4527 + if ($item && $item->attempts >= $item->max_attempts) {
4528 + // Permanently failed
4529 + $wpdb->update(
4530 + $table_name,
4531 + array(
4532 + 'status' => 'failed',
4533 + 'error_message' => $error_message
4534 + ),
4535 + array('id' => $item_id),
4536 + array('%s', '%s'),
4537 + array('%d')
4538 + );
4539 +
4540 + wp_send_json_error(array(
4541 + 'message' => 'Item failed after maximum attempts: ' . $error_message,
4542 + 'permanent_failure' => true,
4543 + 'item_id' => $item_id
4544 + ));
4545 + } else {
4546 + // Mark for retry
4547 + $wpdb->update(
4548 + $table_name,
4549 + array(
4550 + 'status' => 'failed',
4551 + 'error_message' => $error_message
4552 + ),
4553 + array('id' => $item_id),
4554 + array('%s', '%s'),
4555 + array('%d')
4556 + );
4557 +
4558 + wp_send_json_error(array(
4559 + 'message' => 'Item processing failed, will retry: ' . $error_message,
4560 + 'can_retry' => true,
4561 + 'item_id' => $item_id,
4562 + 'attempts' => $item ? $item->attempts : 0
4563 + ));
4564 + }
4565 + }
4566 +}
4567 +
4568 +/**
4569 + * Process a URL from the queue
4570 + */
4571 +private function mxchat_process_queue_url($item_data, $bot_id = 'default') {
4572 + $url = isset($item_data['url']) ? $item_data['url'] : '';
4573 +
4574 + if (empty($url)) {
4575 + return new WP_Error('invalid_url', 'URL is empty');
4576 + }
4577 +
4578 + // Fetch URL content
4579 + $response = wp_remote_get($url, array(
4580 + 'timeout' => 30,
4581 + 'redirection' => 5,
4582 + 'user-agent' => 'MxChat/1.0'
4583 + ));
4584 +
4585 + if (is_wp_error($response)) {
4586 + return $response;
4587 + }
4588 +
4589 + $response_code = wp_remote_retrieve_response_code($response);
4590 + if ($response_code !== 200) {
4591 + return new WP_Error('http_error', 'HTTP ' . $response_code . ' error');
4592 + }
4593 +
4594 + $html = wp_remote_retrieve_body($response);
4595 +
4596 + if (empty($html)) {
4597 + return new WP_Error('empty_response', 'Empty response body');
4598 + }
4599 +
4600 + // Extract and sanitize content
4601 + $content = $this->mxchat_extract_main_content($html);
4602 + $sanitized = $this->mxchat_sanitize_content_for_api($content);
4603 +
4604 + if (empty($sanitized)) {
4605 + // Not an error - just no content found (maybe a redirect or empty page)
4606 + return false;
4607 + }
4608 +
4609 + // Get bot-specific API key
4610 + $bot_options = $this->get_bot_options($bot_id);
4611 + $options = !empty($bot_options) ? $bot_options : get_option('mxchat_options');
4612 + $selected_model = $options['embedding_model'] ?? 'text-embedding-ada-002';
4613 +
4614 + if (strpos($selected_model, 'voyage') === 0) {
4615 + $api_key = $options['voyage_api_key'] ?? '';
4616 + } elseif (strpos($selected_model, 'gemini-embedding') === 0) {
4617 + $api_key = $options['gemini_api_key'] ?? '';
4618 + } else {
4619 + $api_key = $options['api_key'] ?? '';
4620 + }
4621 +
4622 + if (empty($api_key)) {
4623 + return new WP_Error('no_api_key', 'API key not configured for bot: ' . $bot_id);
4624 + }
4625 +
4626 + // UPDATED 2.5.6: Determine content type from URL
4627 + $content_type = 'url'; // Default for external URLs
4628 +
4629 + // Try to detect if it's a product URL
4630 + if (strpos($url, '/product/') !== false || strpos($url, '/shop/') !== false) {
4631 + $content_type = 'product';
4632 + }
4633 +
4634 + // Submit to database with content_type
4635 + $result = MxChat_Utils::submit_content_to_db(
4636 + $sanitized,
4637 + $url,
4638 + $api_key,
4639 + null,
4640 + $bot_id,
4641 + $content_type
4642 + );
4643 +
4644 + return $result;
4645 +}
4646 +
4647 +/**
4648 + * Process a PDF page from the queue
4649 + */
4650 +private function mxchat_process_queue_pdf_page($item_data, $bot_id = 'default') {
4651 + $pdf_path = isset($item_data['pdf_path']) ? $item_data['pdf_path'] : '';
4652 + $pdf_url = isset($item_data['pdf_url']) ? $item_data['pdf_url'] : '';
4653 + $page_number = isset($item_data['page_number']) ? absint($item_data['page_number']) : 0;
4654 + $total_pages = isset($item_data['total_pages']) ? absint($item_data['total_pages']) : 0;
4655 +
4656 + if (empty($pdf_path) || !file_exists($pdf_path)) {
4657 + return new WP_Error('pdf_not_found', 'PDF file not found: ' . $pdf_path);
4658 + }
4659 +
4660 + if ($page_number < 1) {
4661 + return new WP_Error('invalid_page', 'Invalid page number');
4662 + }
4663 +
4664 + try {
4665 + $parser = new \Smalot\PdfParser\Parser();
4666 + $pdf = $parser->parseFile($pdf_path);
4667 + $pages = $pdf->getPages();
4668 +
4669 + if (!isset($pages[$page_number - 1])) {
4670 + return new WP_Error('page_not_found', 'Page ' . $page_number . ' not found in PDF');
4671 + }
4672 +
4673 + $text = $pages[$page_number - 1]->getText();
4674 +
4675 + if (empty($text)) {
4676 + // Not an error - just an empty page
4677 + return false;
4678 + }
4679 +
4680 + $sanitized = $this->mxchat_sanitize_content_for_api($text);
4681 +
4682 + if (empty($sanitized)) {
4683 + return false;
4684 + }
4685 +
4686 + // Create metadata
4687 + $metadata = array(
4688 + 'document_type' => 'pdf',
4689 + 'total_pages' => $total_pages,
4690 + 'current_page' => $page_number,
4691 + 'source_url' => $pdf_url
4692 + );
4693 +
4694 + $content_with_metadata = wp_json_encode($metadata) . "\n---\n" . $sanitized;
4695 + $page_url = esc_url($pdf_url . "#page=" . $page_number);
4696 +
4697 + // Get bot-specific API key
4698 + $bot_options = $this->get_bot_options($bot_id);
4699 + $options = !empty($bot_options) ? $bot_options : get_option('mxchat_options');
4700 + $selected_model = $options['embedding_model'] ?? 'text-embedding-ada-002';
4701 +
4702 + if (strpos($selected_model, 'voyage') === 0) {
4703 + $api_key = $options['voyage_api_key'] ?? '';
4704 + } elseif (strpos($selected_model, 'gemini-embedding') === 0) {
4705 + $api_key = $options['gemini_api_key'] ?? '';
4706 + } else {
4707 + $api_key = $options['api_key'] ?? '';
4708 + }
4709 +
4710 + if (empty($api_key)) {
4711 + return new WP_Error('no_api_key', 'API key not configured for bot: ' . $bot_id);
4712 + }
4713 +
4714 + // Submit to database - UPDATED 2.5.6: Added content_type 'pdf'
4715 + $result = MxChat_Utils::submit_content_to_db(
4716 + $content_with_metadata,
4717 + $page_url,
4718 + $api_key,
4719 + null,
4720 + $bot_id,
4721 + 'pdf'
4722 + );
4723 +
4724 + return $result;
4725 +
4726 + } catch (Exception $e) {
4727 + return new WP_Error('pdf_parse_error', 'Error parsing PDF: ' . $e->getMessage());
4728 + }
4729 +}
4730 +
4731 +/**
4732 + * AJAX: Get queue processing status
4733 + */
4734 +public function ajax_mxchat_get_queue_status() {
4735 + // Verify nonce and permissions
4736 + check_ajax_referer('mxchat_queue_nonce', 'nonce');
4737 +
4738 + if (!current_user_can('manage_options')) {
4739 + wp_send_json_error('Unauthorized access');
4740 + }
4741 +
4742 + $queue_id = isset($_POST['queue_id']) ? sanitize_text_field($_POST['queue_id']) : '';
4743 +
4744 + if (empty($queue_id)) {
4745 + wp_send_json_error('Missing queue ID');
4746 + }
4747 +
4748 + global $wpdb;
4749 + $table_name = $wpdb->prefix . 'mxchat_processing_queue';
4750 +
4751 + // Get counts by status
4752 + $counts = $wpdb->get_results($wpdb->prepare(
4753 + "SELECT status, COUNT(*) as count
4754 + FROM $table_name
4755 + WHERE queue_id = %s
4756 + GROUP BY status",
4757 + $queue_id
4758 + ), OBJECT_K);
4759 +
4760 + $total = 0;
4761 + $completed = 0;
4762 + $failed = 0;
4763 + $processing = 0;
4764 + $pending = 0;
4765 +
4766 + foreach ($counts as $status => $data) {
4767 + $count = absint($data->count);
4768 + $total += $count;
4769 +
4770 + switch ($status) {
4771 + case 'completed':
4772 + $completed = $count;
4773 + break;
4774 + case 'failed':
4775 + $failed = $count;
4776 + break;
4777 + case 'processing':
4778 + $processing = $count;
4779 + break;
4780 + case 'pending':
4781 + $pending = $count;
4782 + break;
4783 + }
4784 + }
4785 +
4786 + // Calculate percentage
4787 + $percentage = $total > 0 ? round((($completed + $failed) / $total) * 100) : 0;
4788 +
4789 + // Get failed items details
4790 + $failed_items = array();
4791 + if ($failed > 0) {
4792 + $failed_items = $wpdb->get_results($wpdb->prepare(
4793 + "SELECT item_type, item_data, error_message, attempts
4794 + FROM $table_name
4795 + WHERE queue_id = %s
4796 + AND status = 'failed'
4797 + AND attempts >= max_attempts
4798 + ORDER BY id DESC
4799 + LIMIT 50",
4800 + $queue_id
4801 + ));
4802 + }
4803 +
4804 + // Get queue metadata
4805 + $source_url = $this->mxchat_get_queue_meta($queue_id, 'source_url');
4806 + $queue_type = $this->mxchat_get_queue_meta($queue_id, 'queue_type');
4807 +
4808 + // Determine if queue is complete
4809 + $is_complete = ($pending === 0 && $processing === 0);
4810 +
4811 + wp_send_json_success(array(
4812 + 'queue_id' => $queue_id,
4813 + 'queue_type' => $queue_type,
4814 + 'source_url' => $source_url,
4815 + 'total' => $total,
4816 + 'completed' => $completed,
4817 + 'failed' => $failed,
4818 + 'processing' => $processing,
4819 + 'pending' => $pending,
4820 + 'percentage' => $percentage,
4821 + 'is_complete' => $is_complete,
4822 + 'failed_items' => $failed_items,
4823 + 'status' => $is_complete ? 'complete' : 'processing'
4824 + ));
4825 +}
4826 +
4827 +/**
4828 + * AJAX: Clear completed queue
4829 + */
4830 +public function ajax_mxchat_clear_queue() {
4831 + // Verify nonce and permissions
4832 + check_ajax_referer('mxchat_queue_nonce', 'nonce');
4833 +
4834 + if (!current_user_can('manage_options')) {
4835 + wp_send_json_error('Unauthorized access');
4836 + }
4837 +
4838 + $queue_id = isset($_POST['queue_id']) ? sanitize_text_field($_POST['queue_id']) : '';
4839 +
4840 + if (empty($queue_id)) {
4841 + wp_send_json_error('Missing queue ID');
4842 + }
4843 +
4844 + global $wpdb;
4845 + $table_name = $wpdb->prefix . 'mxchat_processing_queue';
4846 + $meta_table = $wpdb->prefix . 'mxchat_queue_meta';
4847 +
4848 + // Delete queue items
4849 + $wpdb->delete(
4850 + $table_name,
4851 + array('queue_id' => $queue_id),
4852 + array('%s')
4853 + );
4854 +
4855 + // Delete queue metadata
4856 + $wpdb->delete(
4857 + $meta_table,
4858 + array('queue_id' => $queue_id),
4859 + array('%s')
4860 + );
4861 +
4862 + wp_send_json_success(array(
4863 + 'message' => 'Queue cleared successfully'
4864 + ));
4865 +}
4866 +
4867 +/**
4868 + * AJAX: Retry failed items in queue
4869 + */
4870 +public function ajax_mxchat_retry_failed() {
4871 + // Verify nonce and permissions
4872 + check_ajax_referer('mxchat_queue_nonce', 'nonce');
4873 +
4874 + if (!current_user_can('manage_options')) {
4875 + wp_send_json_error('Unauthorized access');
4876 + }
4877 +
4878 + $queue_id = isset($_POST['queue_id']) ? sanitize_text_field($_POST['queue_id']) : '';
4879 +
4880 + if (empty($queue_id)) {
4881 + wp_send_json_error('Missing queue ID');
4882 + }
4883 +
4884 + global $wpdb;
4885 + $table_name = $wpdb->prefix . 'mxchat_processing_queue';
4886 +
4887 + // Reset failed items to pending and reset attempt count
4888 + $updated = $wpdb->update(
4889 + $table_name,
4890 + array(
4891 + 'status' => 'pending',
4892 + 'attempts' => 0,
4893 + 'error_message' => null
4894 + ),
4895 + array(
4896 + 'queue_id' => $queue_id,
4897 + 'status' => 'failed'
4898 + ),
4899 + array('%s', '%d', '%s'),
4900 + array('%s', '%s')
4901 + );
4902 +
4903 + wp_send_json_success(array(
4904 + 'message' => 'Reset ' . $updated . ' failed items for retry',
4905 + 'reset_count' => $updated
4906 + ));
4907 +}
4908 +
4909 +
4910 +public function ajax_mxchat_mark_queue_complete() {
4911 + check_ajax_referer('mxchat_queue_nonce', 'nonce');
4912 +
4913 + if (!current_user_can('manage_options')) {
4914 + wp_send_json_error('Unauthorized access');
4915 + }
4916 +
4917 + $queue_id = isset($_POST['queue_id']) ? sanitize_text_field($_POST['queue_id']) : '';
4918 +
4919 + if (empty($queue_id)) {
4920 + wp_send_json_error('Missing queue ID');
4921 + }
4922 +
4923 + // Clear active queue transients
4924 + if (strpos($queue_id, 'sitemap_') === 0) {
4925 + delete_transient('mxchat_active_queue_sitemap');
4926 + } else if (strpos($queue_id, 'pdf_') === 0) {
4927 + delete_transient('mxchat_active_queue_pdf');
4928 + }
4929 +
4930 + wp_send_json_success(array('message' => 'Queue marked as complete'));
4931 +}
4932 +
4933 +
4934 + // ========================================
4935 + // STATIC ACCESS METHODS
4936 + // ========================================
4937 +
4938 + /**
4939 + * Get singleton instance
4940 + */
4941 + public static function get_instance() {
4942 + static $instance = null;
4943 + if ($instance === null) {
4944 + $instance = new self();
4945 + }
4946 + return $instance;
4947 + }
4948 +}
4949 +
4950 +// Initialize the Knowledge manager
8838 4951 $mxchat_knowledge_manager = MxChat_Knowledge_Manager::get_instance();