← All changes | admin/class-knowledge-manager.php +4525 -8406 3.2.12 → 2.4.6 View file →

		@@ -1,8407 +1,4526 @@
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($this, 'mxchat_handle_sitemap_submission'));
34		- add_action('admin_post_mxchat_submit_pdf_file', array($this, 'mxchat_handle_pdf_file_submission'));
35		- add_action('admin_post_mxchat_stop_processing', array($this, 'mxchat_stop_processing'));
36		-
37		- // AJAX handlers for real-time processing and status updates
38		- add_action('wp_ajax_mxchat_get_status_updates', array($this, 'mxchat_ajax_get_status_updates'));
39		- add_action('wp_ajax_mxchat_dismiss_completed_status', array($this, 'mxchat_ajax_dismiss_completed_status'));
40		- add_action('wp_ajax_mxchat_get_content_list', array($this, 'ajax_mxchat_get_content_list'));
41		- add_action('wp_ajax_mxchat_process_selected_content', array($this, 'ajax_mxchat_process_selected_content'));
42		- add_action('wp_ajax_mxchat_save_inline_prompt', array($this, 'mxchat_save_inline_prompt'));
43		- add_action('admin_post_mxchat_delete_pinecone_prompt', array($this, 'mxchat_handle_pinecone_prompt_delete'));
44		- add_action('wp_ajax_mxchat_delete_pinecone_prompt', array($this, 'ajax_mxchat_delete_pinecone_prompt'));
45		- add_action('wp_ajax_mxchat_delete_chunks_by_url', array($this, 'ajax_mxchat_delete_chunks_by_url'));
46		- add_action('wp_ajax_mxchat_delete_wordpress_prompt', array($this, 'ajax_mxchat_delete_wordpress_prompt'));
47		- add_action('wp_ajax_mxchat_bulk_delete_knowledge', array($this, 'ajax_mxchat_bulk_delete_knowledge'));
48		- add_action('wp_ajax_mxchat_update_role_restriction', array($this, 'ajax_mxchat_update_role_restriction'));
49		-
50		- // Queue-based processing AJAX handlers
51		- add_action('wp_ajax_mxchat_get_next_queue_item', array($this, 'ajax_mxchat_get_next_queue_item'));
52		- add_action('wp_ajax_mxchat_process_queue_item', array($this, 'ajax_mxchat_process_queue_item'));
53		- add_action('wp_ajax_mxchat_get_queue_status', array($this, 'ajax_mxchat_get_queue_status'));
54		- add_action('wp_ajax_mxchat_clear_queue', array($this, 'ajax_mxchat_clear_queue'));
55		- add_action('wp_ajax_mxchat_retry_failed', array($this, 'ajax_mxchat_retry_failed'));
56		- add_action('wp_ajax_mxchat_get_recent_entries', array($this, 'ajax_mxchat_get_recent_entries'));
57		- add_action('wp_ajax_mxchat_detect_sitemaps', array($this, 'ajax_mxchat_detect_sitemaps'));
58		- add_action('wp_ajax_mxchat_refresh_pinecone_entries', array($this, 'ajax_mxchat_refresh_pinecone_entries'));
59		- add_action('wp_ajax_mxchat_paginate_entries', array($this, 'ajax_mxchat_paginate_entries'));
60		- add_action('wp_ajax_mxchat_get_entry_content', array($this, 'ajax_mxchat_get_entry_content'));
61		- add_action('wp_ajax_mxchat_save_entry_content', array($this, 'ajax_mxchat_save_entry_content'));
62		- add_action('wp_ajax_mxchat_inspect_entry', array($this, 'ajax_mxchat_inspect_entry'));
63		-
64		- // WordPress post management hooks
65		- add_action('pre_post_update', array($this, 'mxchat_store_pre_update_status'), 10, 2);
66		- add_action('post_updated', array($this, 'mxchat_handle_post_update'), 10, 3);
67		- add_action('before_delete_post', array($this, 'mxchat_handle_post_delete'));
68		- add_action('wp_trash_post', array($this, 'mxchat_handle_post_delete'));
69		-
70		- // ACF hook - fires AFTER ACF fields are saved, ensuring ACF data is available
71		- // Priority 20 to run after ACF's own save (which runs at priority 10)
72		- add_action('acf/save_post', array($this, 'mxchat_handle_acf_save'), 20);
73		-
74		- add_action('wp_ajax_mxchat_mark_queue_complete', array($this, 'ajax_mxchat_mark_queue_complete'));
75		-
76		- // WooCommerce product hooks (if WooCommerce is active)
77		- if (class_exists('WooCommerce')) {
78		- add_action('pre_post_update', array($this, 'mxchat_store_pre_update_status'), 10, 2);
79		- add_action('save_post_product', array($this, 'mxchat_handle_product_change'), 10, 3);
80		- add_action('wp_trash_post', array($this, 'mxchat_handle_product_delete'));
81		- add_action('before_delete_post', array($this, 'mxchat_handle_product_delete'));
82		- }
83		-}
84		-
85		- /**
86		- * Get current options (refreshed)
87		- */
88		- private function mxchat_get_options() {
89		- if (empty($this->options)) {
90		- $this->options = get_option('mxchat_options', array());
91		- }
92		- return $this->options;
93		- }
94		-
95		-
96		- // ========================================
97		- // MAIN CONTENT SUBMISSION HANDLERS
98		- // ========================================
99		-
100		-public function mxchat_handle_content_submission() {
101		- // Check if the form was submitted and the user has permission.
102		- if (!isset($_POST['submit_content']) || !current_user_can('manage_options')) {
103		- return;
104		- }
105		-
106		- // Verify the nonce.
107		- $nonce = isset($_POST['mxchat_submit_content_nonce']) ? sanitize_text_field(wp_unslash($_POST['mxchat_submit_content_nonce'])) : '';
108		- if (!wp_verify_nonce($nonce, 'mxchat_submit_content_action')) {
109		- wp_die(esc_html__('Nonce verification failed.', 'mxchat'));
110		- }
111		-
112		- // Sanitize the inputs.
113		- // Use wp_kses_post to allow safe HTML and wp_unslash to remove WordPress added slashes
114		- $article_content = wp_kses_post(wp_unslash($_POST['article_content']));
115		- $article_url = isset($_POST['article_url']) ? esc_url_raw($_POST['article_url']) : '';
116		-
117		- // Get bot_id from form submission
118		- $bot_id = isset($_POST['bot_id']) ? sanitize_key($_POST['bot_id']) : 'default';
119		-
120		- // Get bot-specific options and API key
121		- $bot_options = $this->get_bot_options($bot_id);
122		- $options = !empty($bot_options) ? $bot_options : get_option('mxchat_options');
123		- $selected_model = $options['embedding_model'] ?? 'text-embedding-ada-002';
124		-
125		- if (strpos($selected_model, 'voyage') === 0) {
126		- $api_key = $options['voyage_api_key'] ?? '';
127		- } elseif (strpos($selected_model, 'gemini-embedding') === 0) {
128		- $api_key = $options['gemini_api_key'] ?? '';
129		- } else {
130		- $api_key = $options['api_key'] ?? '';
131		- }
132		-
133		- if (empty($api_key)) {
134		- set_transient('mxchat_admin_notice_error',
135		- esc_html__('API key is not configured. Please add your API key in the settings before submitting content.', 'mxchat'),
136		- 30
137		- );
138		- wp_safe_redirect(esc_url(admin_url('admin.php?page=mxchat-prompts')));
139		- exit;
140		- }
141		-
142		- // Use centralized utility function with bot_id
143		- $result = MxChat_Utils::submit_content_to_db($article_content, $article_url, $api_key, null, $bot_id);
144		-
145		- if (is_wp_error($result)) {
146		- set_transient('mxchat_admin_notice_error',
147		- esc_html__('Error storing content: ', 'mxchat') . $result->get_error_message(),
148		- 30
149		- );
150		- } else {
151		- set_transient('mxchat_admin_notice_success',
152		- esc_html__('Content successfully submitted!', 'mxchat'),
153		- 30
154		- );
155		- }
156		-
157		- wp_safe_redirect(esc_url(admin_url('admin.php?page=mxchat-prompts')));
158		- exit;
159		-}
160		-
161		-public function mxchat_is_pdf_url($url, $response) {
162		- $content_type = wp_remote_retrieve_header($response, 'content-type');
163		- $file_extension = strtolower(pathinfo($url, PATHINFO_EXTENSION));
164		-
165		- // Check Content-Disposition header for .pdf filename (Google Drive sends this)
166		- $disposition = wp_remote_retrieve_header($response, 'content-disposition');
167		- $has_pdf_disposition = ! empty($disposition) && stripos($disposition, '.pdf') !== false;
168		-
169		- return strpos($content_type, 'pdf') !== false || $file_extension === 'pdf' || $has_pdf_disposition;
170		-}
171		-
172		-
173		-public function mxchat_handle_pdf_for_knowledge_base($pdf_url, $response, $bot_id = 'default') {
174		- if (!current_user_can('manage_options')) {
175		- return false;
176		- }
177		-
178		- $pdf_url = esc_url_raw($pdf_url);
179		- $upload_dir = wp_upload_dir();
180		-
181		- if (isset($upload_dir['error']) && $upload_dir['error'] !== false) {
182		- return false;
183		- }
184		-
185		- $pdf_filename = sanitize_file_name('mxchat_kb_' . time() . '.pdf');
186		- $pdf_path = trailingslashit($upload_dir['path']) . $pdf_filename;
187		-
188		- $response_body = wp_remote_retrieve_body($response);
189		- if (empty($response_body)) {
190		- return false;
191		- }
192		-
193		- if (!wp_mkdir_p(dirname($pdf_path))) {
194		- return false;
195		- }
196		-
197		- try {
198		- file_put_contents($pdf_path, $response_body);
199		-
200		- if (!file_exists($pdf_path)) {
201		- throw new Exception(__('Failed to save PDF file', 'mxchat'));
202		- }
203		-
204		- $total_pages = $this->mxchat_validate_and_count_pdf_pages($pdf_path);
205		-
206		- if ($total_pages === false || $total_pages < 1) {
207		- throw new Exception(__('Invalid PDF: Unable to parse or no pages found', 'mxchat'));
208		- }
209		-
210		- // Create unique queue ID
211		- $queue_id = 'pdf_' . md5($pdf_url . time());
212		-
213		- // Create array of pages to process
214		- $pages = array();
215		- for ($i = 1; $i <= $total_pages; $i++) {
216		- $pages[] = array(
217		- 'pdf_path' => $pdf_path,
218		- 'pdf_url' => $pdf_url,
219		- 'page_number' => $i,
220		- 'total_pages' => $total_pages
221		- );
222		- }
223		-
224		- // Add pages to queue
225		- $queued_count = $this->mxchat_add_to_queue($queue_id, 'pdf_page', $pages, $bot_id);
226		-
227		- if ($queued_count === 0) {
228		- wp_delete_file($pdf_path);
229		- throw new Exception(__('Failed to add PDF pages to processing queue', 'mxchat'));
230		- }
231		-
232		- // Store queue metadata
233		- $this->mxchat_set_queue_meta($queue_id, 'source_url', $pdf_url);
234		- $this->mxchat_set_queue_meta($queue_id, 'queue_type', 'pdf');
235		- $this->mxchat_set_queue_meta($queue_id, 'total_items', $total_pages);
236		- $this->mxchat_set_queue_meta($queue_id, 'bot_id', $bot_id);
237		- $this->mxchat_set_queue_meta($queue_id, 'pdf_path', $pdf_path);
238		- $this->mxchat_set_queue_meta($queue_id, 'created_at', current_time('mysql'));
239		-
240		- // Store queue ID in transient for status tracking
241		- set_transient('mxchat_active_queue_pdf', $queue_id, DAY_IN_SECONDS);
242		- set_transient('mxchat_last_pdf_url', $pdf_url, DAY_IN_SECONDS);
243		-
244		- return 'queued';
245		-
246		- } catch (Exception $e) {
247		- if (file_exists($pdf_path)) {
248		- wp_delete_file($pdf_path);
249		- }
250		- return $e->getMessage();
251		- }
252		-}
253		-
254		-/**
255		- * Handle direct PDF file upload from the knowledge base page
256		- */
257		-public function mxchat_handle_pdf_file_submission() {
258		- if (!isset($_POST['submit_pdf_file']) || !current_user_can('manage_options')) {
259		- wp_die(esc_html__('Unauthorized access', 'mxchat'));
260		- }
261		-
262		- check_admin_referer('mxchat_submit_pdf_file_action', 'mxchat_submit_pdf_file_nonce');
263		-
264		- $redirect_url = admin_url('admin.php?page=mxchat-prompts');
265		-
266		- // Validate file upload
267		- if (empty($_FILES['pdf_file']) || $_FILES['pdf_file']['error'] !== UPLOAD_ERR_OK) {
268		- $error_code = isset($_FILES['pdf_file']['error']) ? $_FILES['pdf_file']['error'] : UPLOAD_ERR_NO_FILE;
269		- $error_messages = array(
270		- UPLOAD_ERR_INI_SIZE => __('The uploaded file exceeds the server upload_max_filesize limit.', 'mxchat'),
271		- UPLOAD_ERR_FORM_SIZE => __('The uploaded file exceeds the form MAX_FILE_SIZE limit.', 'mxchat'),
272		- UPLOAD_ERR_PARTIAL => __('The file was only partially uploaded.', 'mxchat'),
273		- UPLOAD_ERR_NO_FILE => __('No file was uploaded. Please select a PDF file.', 'mxchat'),
274		- UPLOAD_ERR_NO_TMP_DIR => __('Server missing temporary folder.', 'mxchat'),
275		- UPLOAD_ERR_CANT_WRITE => __('Server failed to write file to disk.', 'mxchat'),
276		- );
277		- $error_msg = isset($error_messages[$error_code]) ? $error_messages[$error_code] : __('Unknown upload error.', 'mxchat');
278		- set_transient('mxchat_admin_notice_error', $error_msg, 30);
279		- wp_safe_redirect(esc_url($redirect_url));
280		- exit;
281		- }
282		-
283		- $file = $_FILES['pdf_file'];
284		-
285		- // Validate MIME type
286		- $finfo = finfo_open(FILEINFO_MIME_TYPE);
287		- $mime_type = finfo_file($finfo, $file['tmp_name']);
288		- finfo_close($finfo);
289		-
290		- if ($mime_type !== 'application/pdf') {
291		- set_transient('mxchat_admin_notice_error',
292		- esc_html__('Invalid file type. Only PDF files are accepted.', 'mxchat'),
293		- 30
294		- );
295		- wp_safe_redirect(esc_url($redirect_url));
296		- exit;
297		- }
298		-
299		- // Validate extension
300		- $ext = strtolower(pathinfo($file['name'], PATHINFO_EXTENSION));
301		- if ($ext !== 'pdf') {
302		- set_transient('mxchat_admin_notice_error',
303		- esc_html__('Invalid file extension. Only .pdf files are accepted.', 'mxchat'),
304		- 30
305		- );
306		- wp_safe_redirect(esc_url($redirect_url));
307		- exit;
308		- }
309		-
310		- $bot_id = isset($_POST['bot_id']) ? sanitize_key($_POST['bot_id']) : 'default';
311		- $original_filename = sanitize_file_name($file['name']);
312		-
313		- $upload_dir = wp_upload_dir();
314		- if (isset($upload_dir['error']) && $upload_dir['error'] !== false) {
315		- set_transient('mxchat_admin_notice_error',
316		- esc_html__('WordPress upload directory is not writable.', 'mxchat'),
317		- 30
318		- );
319		- wp_safe_redirect(esc_url($redirect_url));
320		- exit;
321		- }
322		-
323		- $pdf_filename = sanitize_file_name('mxchat_kb_' . time() . '.pdf');
324		- $pdf_path = trailingslashit($upload_dir['path']) . $pdf_filename;
325		-
326		- if (!wp_mkdir_p(dirname($pdf_path))) {
327		- set_transient('mxchat_admin_notice_error',
328		- esc_html__('Failed to create upload directory.', 'mxchat'),
329		- 30
330		- );
331		- wp_safe_redirect(esc_url($redirect_url));
332		- exit;
333		- }
334		-
335		- // Move uploaded file
336		- if (!move_uploaded_file($file['tmp_name'], $pdf_path)) {
337		- set_transient('mxchat_admin_notice_error',
338		- esc_html__('Failed to save uploaded PDF file.', 'mxchat'),
339		- 30
340		- );
341		- wp_safe_redirect(esc_url($redirect_url));
342		- exit;
343		- }
344		-
345		- try {
346		- $total_pages = $this->mxchat_validate_and_count_pdf_pages($pdf_path);
347		-
348		- if ($total_pages === false || $total_pages < 1) {
349		- throw new Exception(__('Invalid PDF: Unable to parse or no pages found', 'mxchat'));
350		- }
351		-
352		- // Use original filename as the source identifier
353		- $source_label = 'upload://' . $original_filename;
354		-
355		- $queue_id = 'pdf_' . md5($source_label . time());
356		-
357		- $pages = array();
358		- for ($i = 1; $i <= $total_pages; $i++) {
359		- $pages[] = array(
360		- 'pdf_path' => $pdf_path,
361		- 'pdf_url' => $source_label,
362		- 'page_number' => $i,
363		- 'total_pages' => $total_pages,
364		- );
365		- }
366		-
367		- $queued_count = $this->mxchat_add_to_queue($queue_id, 'pdf_page', $pages, $bot_id);
368		-
369		- if ($queued_count === 0) {
370		- wp_delete_file($pdf_path);
371		- throw new Exception(__('Failed to add PDF pages to processing queue', 'mxchat'));
372		- }
373		-
374		- $this->mxchat_set_queue_meta($queue_id, 'source_url', $source_label);
375		- $this->mxchat_set_queue_meta($queue_id, 'queue_type', 'pdf');
376		- $this->mxchat_set_queue_meta($queue_id, 'total_items', $total_pages);
377		- $this->mxchat_set_queue_meta($queue_id, 'bot_id', $bot_id);
378		- $this->mxchat_set_queue_meta($queue_id, 'pdf_path', $pdf_path);
379		- $this->mxchat_set_queue_meta($queue_id, 'created_at', current_time('mysql'));
380		-
381		- set_transient('mxchat_active_queue_pdf', $queue_id, DAY_IN_SECONDS);
382		- set_transient('mxchat_last_pdf_url', $source_label, DAY_IN_SECONDS);
383		-
384		- set_transient('mxchat_admin_notice_success',
385		- sprintf(
386		- esc_html__('PDF "%s" (%d pages) queued for processing. Processing will start automatically.', 'mxchat'),
387		- esc_html($original_filename),
388		- $total_pages
389		- ),
390		- 30
391		- );
392		-
393		- } catch (Exception $e) {
394		- if (file_exists($pdf_path)) {
395		- wp_delete_file($pdf_path);
396		- }
397		- set_transient('mxchat_admin_notice_error',
398		- esc_html__('Failed to process uploaded PDF: ', 'mxchat') . esc_html($e->getMessage()),
399		- 30
400		- );
401		- }
402		-
403		- wp_safe_redirect(esc_url($redirect_url));
404		- exit;
405		-}
406		-
407		-/**
408		- * Validate PDF and count pages with multiple parser attempts
409		- */
410		-private function mxchat_validate_and_count_pdf_pages($pdf_path) {
411		- // Method 1: Try with Smalot PDF Parser (your current method)
412		- try {
413		- mxchat_load_pdf_parser();
414		- $parser = new \Smalot\PdfParser\Parser();
415		- $pdf = $parser->parseFile($pdf_path);
416		- $pages = $pdf->getPages();
417		- $page_count = count($pages);
418		-
419		- if ($page_count > 0) {
420		- //error_log('PDF parsed successfully with Smalot parser: ' . $page_count . ' pages');
421		- return $page_count;
422		- }
423		- } catch (Exception $e) {
424		- //error_log('Smalot PDF parser failed: ' . $e->getMessage());
425		- }
426		-
427		- // Method 2: Try with pdfinfo command (if available)
428		- if (function_exists('shell_exec') && !$this->mxchat_is_shell_disabled()) {
429		- try {
430		- $command = 'pdfinfo ' . escapeshellarg($pdf_path) . ' 2>&1';
431		- $output = shell_exec($command);
432		-
433		- if ($output && preg_match('/Pages:\s*(\d+)/', $output, $matches)) {
434		- $page_count = intval($matches[1]);
435		- if ($page_count > 0) {
436		- //error_log('PDF parsed successfully with pdfinfo: ' . $page_count . ' pages');
437		- return $page_count;
438		- }
439		- }
440		- } catch (Exception $e) {
441		- //error_log('pdfinfo command failed: ' . $e->getMessage());
442		- }
443		- }
444		-
445		- // Method 3: Try to repair PDF and parse again
446		- try {
447		- $repaired_path = $this->mxchat_attempt_pdf_repair($pdf_path);
448		- if ($repaired_path && $repaired_path !== $pdf_path) {
449		- mxchat_load_pdf_parser();
450		- $parser = new \Smalot\PdfParser\Parser();
451		- $pdf = $parser->parseFile($repaired_path);
452		- $pages = $pdf->getPages();
453		- $page_count = count($pages);
454		-
455		- if ($page_count > 0) {
456		- // Replace original with repaired version
457		- copy($repaired_path, $pdf_path);
458		- unlink($repaired_path);
459		- //error_log('PDF repaired and parsed successfully: ' . $page_count . ' pages');
460		- return $page_count;
461		- }
462		-
463		- // Clean up repaired file if it didn't work
464		- unlink($repaired_path);
465		- }
466		- } catch (Exception $e) {
467		- //error_log('PDF repair attempt failed: ' . $e->getMessage());
468		- }
469		-
470		- // Method 4: Manual PDF structure analysis (basic page count)
471		- try {
472		- $page_count = $this->mxchat_manual_pdf_page_count($pdf_path);
473		- if ($page_count > 0) {
474		- //error_log('PDF page count determined manually: ' . $page_count . ' pages');
475		- return $page_count;
476		- }
477		- } catch (Exception $e) {
478		- //error_log('Manual PDF analysis failed: ' . $e->getMessage());
479		- }
480		-
481		- //error_log('All PDF parsing methods failed for: ' . $pdf_path);
482		- return false;
483		-}
484		-
485		-/**
486		- * Check if shell_exec is disabled
487		- */
488		-private function mxchat_is_shell_disabled() {
489		- $disabled = explode(',', ini_get('disable_functions'));
490		- return in_array('shell_exec', $disabled);
491		-}
492		-
493		-/**
494		- * Attempt to repair PDF using basic methods
495		- */
496		-private function mxchat_attempt_pdf_repair($pdf_path) {
497		- try {
498		- $content = file_get_contents($pdf_path);
499		- if (!$content) {
500		- return false;
501		- }
502		-
503		- // Check if PDF starts with proper header
504		- if (substr($content, 0, 4) !== '%PDF') {
505		- // Try to find PDF header in the content
506		- $header_pos = strpos($content, '%PDF');
507		- if ($header_pos !== false && $header_pos < 1024) {
508		- // Remove junk before PDF header
509		- $content = substr($content, $header_pos);
510		- $repaired_path = $pdf_path . '.repaired';
511		- file_put_contents($repaired_path, $content);
512		- return $repaired_path;
513		- }
514		- }
515		-
516		- // Check for EOF marker
517		- $content = rtrim($content);
518		- if (!preg_match('/%%EOF\s*$/', $content)) {
519		- // Add EOF marker if missing
520		- $content .= "\n%%EOF";
521		- $repaired_path = $pdf_path . '.repaired';
522		- file_put_contents($repaired_path, $content);
523		- return $repaired_path;
524		- }
525		-
526		- } catch (Exception $e) {
527		- //error_log('PDF repair error: ' . $e->getMessage());
528		- }
529		-
530		- return false;
531		-}
532		-
533		-/**
534		- * Manual PDF page counting by analyzing PDF structure
535		- */
536		-private function mxchat_manual_pdf_page_count($pdf_path) {
537		- try {
538		- $content = file_get_contents($pdf_path);
539		- if (!$content) {
540		- return 0;
541		- }
542		-
543		- // Method 1: Count /Type /Page objects
544		- $page_count = preg_match_all('/\/Type\s*\/Page[^s]/', $content);
545		- if ($page_count > 0) {
546		- return $page_count;
547		- }
548		-
549		- // Method 2: Look for /Count in pages object
550		- if (preg_match('/\/Type\s*\/Pages.*?\/Count\s+(\d+)/', $content, $matches)) {
551		- return intval($matches[1]);
552		- }
553		-
554		- // Method 3: Count page references
555		- $page_count = preg_match_all('/\d+\s+0\s+obj\s*<<[^>]*\/Type\s*\/Page/', $content);
556		- if ($page_count > 0) {
557		- return $page_count;
558		- }
559		-
560		- } catch (Exception $e) {
561		- //error_log('Manual PDF analysis error: ' . $e->getMessage());
562		- }
563		-
564		- return 0;
565		-}
566		-
567		-
568		-public function mxchat_save_inline_prompt() {
569		- // DEBUG: Log what we're receiving
570		- //error_log('=== MXCHAT DEBUG ===');
571		- //error_log('POST data: ' . print_r($_POST, true));
572		- //error_log('Nonce from POST: ' . ($_POST['_ajax_nonce'] ?? 'NOT FOUND'));
573		-
574		- // Check for nonce security
575		- check_ajax_referer('mxchat_save_inline_nonce', '_ajax_nonce');
576		-
577		- // If we get here, nonce passed
578		- //error_log('Nonce verification PASSED');
579		-
580		- // Verify permissions
581		- if (!current_user_can('manage_options')) {
582		- wp_send_json_error(esc_html__('Permission denied.', 'mxchat'));
583		- return;
584		- }
585		-
586		- global $wpdb;
587		- $table_name = $wpdb->prefix . 'mxchat_system_prompt_content';
588		-
589		- // Validate and sanitize input data
590		- $prompt_id = isset($_POST['id']) ? absint($_POST['id']) : 0;
591		- $article_content = isset($_POST['article_content']) ? wp_kses_post(wp_unslash($_POST['article_content'])) : '';
592		- $article_url = isset($_POST['article_url']) ? esc_url_raw($_POST['article_url']) : '';
593		-
594		- if ($prompt_id > 0 && !empty($article_content)) {
595		- // Re-generate the embedding vector for the updated content
596		- $embedding_vector = $this->mxchat_generate_embedding($article_content);
597		- if (is_array($embedding_vector)) {
598		- // Serialize the embedding vector before storing it
599		- $embedding_vector_serialized = serialize($embedding_vector);
600		- // Update the prompt in the database
601		- $updated = $wpdb->update(
602		- $table_name,
603		- array(
604		- 'article_content' => $article_content,
605		- 'embedding_vector' => $embedding_vector_serialized,
606		- 'source_url' => $article_url,
607		- ),
608		- array('id' => $prompt_id),
609		- array('%s', '%s', '%s'),
610		- array('%d')
611		- );
612		- if ($updated !== false) {
613		- wp_send_json_success();
614		- } else {
615		- MxChat_Admin::mxchat_log_debug('knowledge_error', 'Database update failed when saving knowledge entry');
616		- wp_send_json_error(esc_html__('Database update failed.', 'mxchat'));
617		- }
618		- } else {
619		- MxChat_Admin::mxchat_log_debug('embedding_error', 'Embedding generation failed for knowledge entry');
620		- wp_send_json_error(esc_html__('Embedding generation failed.', 'mxchat'));
621		- }
622		- } else {
623		- wp_send_json_error(esc_html__('Invalid data.', 'mxchat'));
624		- }
625		-}
626		-
627		-
628		-/**
629		- * AJAX: Get full content for editing — reassembles chunks if needed.
630		- * Works for both WordPress DB and Pinecone entries.
631		- */
632		-public function ajax_mxchat_get_entry_content() {
633		- check_ajax_referer('mxchat_edit_entry_nonce', 'nonce');
634		-
635		- if ( ! current_user_can('manage_options') ) {
636		- wp_send_json_error( array( 'message' => 'Permission denied.' ) );
637		- }
638		-
639		- $source_url = isset($_POST['source_url']) ? sanitize_text_field( wp_unslash($_POST['source_url']) ) : '';
640		- $entry_id = isset($_POST['entry_id']) ? absint($_POST['entry_id']) : 0;
641		- $data_source = isset($_POST['data_source']) ? sanitize_key($_POST['data_source']) : 'wordpress';
642		- $bot_id = isset($_POST['bot_id']) ? sanitize_key($_POST['bot_id']) : 'default';
643		-
644		- if ( $data_source === 'pinecone' ) {
645		- // Pinecone: fetch vectors by source_url, reassemble chunks
646		- $content = $this->get_pinecone_entry_content( $source_url, $entry_id, $bot_id );
647		- } else {
648		- // WordPress DB
649		- $content = $this->get_wordpress_entry_content( $source_url, $entry_id );
650		- }
651		-
652		- if ( is_wp_error( $content ) ) {
653		- wp_send_json_error( array( 'message' => $content->get_error_message() ) );
654		- }
655		-
656		- wp_send_json_success( $content );
657		-}
658		-
659		-/**
660		- * Get content from WordPress DB — reassembles chunks by source_url.
661		- */
662		-private function get_wordpress_entry_content( $source_url, $entry_id ) {
663		- global $wpdb;
664		- $table = $wpdb->prefix . 'mxchat_system_prompt_content';
665		-
666		- // If we have a source_url, check for chunks
667		- if ( ! empty( $source_url ) && strpos( $source_url, 'mxchat://' ) !== 0 ) {
668		- $rows = $wpdb->get_results( $wpdb->prepare(
669		- "SELECT id, article_content, source_url, content_type FROM {$table} WHERE source_url = %s ORDER BY id ASC",
670		- $source_url
671		- ) );
672		-
673		- if ( $rows && count( $rows ) > 1 ) {
674		- // Multiple rows = chunked. Reassemble.
675		- $chunks = array();
676		- foreach ( $rows as $row ) {
677		- $parsed = MxChat_Chunker::parse_stored_chunk( $row->article_content );
678		- $index = isset( $parsed['metadata']['chunk_index'] ) ? intval( $parsed['metadata']['chunk_index'] ) : count( $chunks );
679		- $chunks[ $index ] = $parsed['text'];
680		- }
681		- ksort( $chunks );
682		- return array(
683		- 'content' => implode( "\n\n", $chunks ),
684		- 'source_url' => $source_url,
685		- 'is_chunked' => true,
686		- 'chunk_count' => count( $chunks ),
687		- 'content_type' => $rows[0]->content_type,
688		- );
689		- } elseif ( $rows && count( $rows ) === 1 ) {
690		- $parsed = MxChat_Chunker::parse_stored_chunk( $rows[0]->article_content );
691		- return array(
692		- 'content' => $parsed['text'],
693		- 'source_url' => $source_url,
694		- 'entry_id' => $rows[0]->id,
695		- 'is_chunked' => false,
696		- 'content_type' => $rows[0]->content_type,
697		- );
698		- }
699		- }
700		-
701		- // Fallback: fetch by ID
702		- if ( $entry_id > 0 ) {
703		- $row = $wpdb->get_row( $wpdb->prepare(
704		- "SELECT id, article_content, source_url, content_type FROM {$table} WHERE id = %d",
705		- $entry_id
706		- ) );
707		- if ( $row ) {
708		- $parsed = MxChat_Chunker::parse_stored_chunk( $row->article_content );
709		- return array(
710		- 'content' => $parsed['text'],
711		- 'source_url' => $row->source_url,
712		- 'entry_id' => $row->id,
713		- 'is_chunked' => false,
714		- 'content_type' => $row->content_type,
715		- );
716		- }
717		- }
718		-
719		- return new WP_Error( 'not_found', 'Entry not found.' );
720		-}
721		-
722		-/**
723		- * Get content from Pinecone — fetches vectors by source_url, reassembles chunks.
724		- */
725		-private function get_pinecone_entry_content( $source_url, $entry_id, $bot_id ) {
726		- if ( ! class_exists('MxChat_Pinecone_Manager') ) {
727		- return new WP_Error( 'pinecone_unavailable', 'Pinecone manager not available.' );
728		- }
729		-
730		- // Get Pinecone config
731		- if ( $bot_id === 'default' || ! class_exists('MxChat_Multi_Bot_Manager') ) {
732		- $pinecone_options = get_option('mxchat_pinecone_addon_options');
733		- $api_key = $pinecone_options['mxchat_pinecone_api_key'] ?? '';
734		- $host = $pinecone_options['mxchat_pinecone_host'] ?? '';
735		- $namespace = $pinecone_options['mxchat_pinecone_namespace'] ?? '';
736		- } else {
737		- $bot_config = apply_filters('mxchat_get_bot_pinecone_config', array(), $bot_id);
738		- $api_key = $bot_config['api_key'] ?? '';
739		- $host = $bot_config['host'] ?? '';
740		- $namespace = $bot_config['namespace'] ?? '';
741		- }
742		-
743		- if ( empty($host) || empty($api_key) ) {
744		- return new WP_Error( 'pinecone_config', 'Pinecone not configured.' );
745		- }
746		-
747		- // List vectors with the source_url prefix
748		- $base_id = md5( $source_url );
749		- $vector_ids = array( $base_id );
750		-
751		- // Find chunk vectors
752		- $list_url = "https://{$host}/vectors/list";
753		- $list_body = array( 'prefix' => $base_id . '_chunk_', 'limit' => 100 );
754		- if ( ! empty($namespace) ) {
755		- $list_body['namespace'] = $namespace;
756		- }
757		-
758		- $list_resp = wp_remote_post( $list_url, array(
759		- 'headers' => array( 'Api-Key' => $api_key, 'Content-Type' => 'application/json' ),
760		- 'body' => wp_json_encode( $list_body ),
761		- 'timeout' => 15,
762		- ) );
763		-
764		- if ( ! is_wp_error($list_resp) ) {
765		- $list_data = json_decode( wp_remote_retrieve_body($list_resp), true );
766		- if ( ! empty($list_data['vectors']) ) {
767		- foreach ( $list_data['vectors'] as $v ) {
768		- $vector_ids[] = $v['id'];
769		- }
770		- }
771		- }
772		-
773		- // Fetch vectors with metadata
774		- $fetch_url = "https://{$host}/vectors/fetch";
775		- $fetch_body = array( 'ids' => $vector_ids );
776		- if ( ! empty($namespace) ) {
777		- $fetch_body['namespace'] = $namespace;
778		- }
779		-
780		- $fetch_resp = wp_remote_post( $fetch_url, array(
781		- 'headers' => array( 'Api-Key' => $api_key, 'Content-Type' => 'application/json' ),
782		- 'body' => wp_json_encode( $fetch_body ),
783		- 'timeout' => 15,
784		- ) );
785		-
786		- if ( is_wp_error($fetch_resp) ) {
787		- return new WP_Error( 'pinecone_fetch', 'Failed to fetch from Pinecone.' );
788		- }
789		-
790		- $fetch_data = json_decode( wp_remote_retrieve_body($fetch_resp), true );
791		- $vectors = $fetch_data['vectors'] ?? array();
792		-
793		- if ( empty($vectors) ) {
794		- return new WP_Error( 'not_found', 'Entry not found in Pinecone.' );
795		- }
796		-
797		- // Reassemble chunks
798		- $chunks = array();
799		- $content_type = 'content';
800		- foreach ( $vectors as $vid => $vector ) {
801		- $meta = $vector['metadata'] ?? array();
802		- $text = $meta['text'] ?? '';
803		- $index = $meta['chunk_index'] ?? 0;
804		- $content_type = $meta['type'] ?? 'content';
805		- $chunks[ intval($index) ] = $text;
806		- }
807		- ksort( $chunks );
808		-
809		- return array(
810		- 'content' => implode( "\n\n", $chunks ),
811		- 'source_url' => $source_url,
812		- 'is_chunked' => count($chunks) > 1,
813		- 'chunk_count' => count($chunks),
814		- 'content_type' => $content_type,
815		- );
816		-}
817		-
818		-/**
819		- * AJAX: Inspect a knowledge entry — returns the per-chunk stored text + metadata
820		- * WITHOUT collapsing it, so a site owner can see exactly what was indexed for an
821		- * entry (plan-mxchat-20260628-d8cb4b). READ-ONLY: never re-embeds or mutates.
822		- */
823		-public function ajax_mxchat_inspect_entry() {
824		- check_ajax_referer('mxchat_inspect_entry_nonce', 'nonce');
825		-
826		- if ( ! current_user_can('manage_options') ) {
827		- wp_send_json_error( array( 'message' => esc_html__('Permission denied.', 'mxchat') ) );
828		- }
829		-
830		- $source_url = isset($_POST['source_url']) ? sanitize_text_field( wp_unslash($_POST['source_url']) ) : '';
831		- $entry_id = isset($_POST['entry_id']) ? absint($_POST['entry_id']) : 0;
832		- $data_source = isset($_POST['data_source']) ? sanitize_key($_POST['data_source']) : 'wordpress';
833		- $bot_id = isset($_POST['bot_id']) ? sanitize_key($_POST['bot_id']) : 'default';
834		-
835		- if ( $data_source === 'pinecone' ) {
836		- $result = $this->inspect_pinecone_entry( $source_url, $entry_id, $bot_id );
837		- } else {
838		- $result = $this->inspect_wordpress_entry( $source_url, $entry_id );
839		- }
840		-
841		- if ( is_wp_error( $result ) ) {
842		- wp_send_json_error( array( 'message' => $result->get_error_message() ) );
843		- }
844		-
845		- wp_send_json_success( $result );
846		-}
847		-
848		-/**
849		- * Read-only inspector for WordPress-DB entries. Mirrors get_wordpress_entry_content()
850		- * but returns each STORED chunk's exact text + length (no implode), plus the assembled
851		- * embedded text. This shows what is actually in the index, not a re-derivation from the post.
852		- */
853		-private function inspect_wordpress_entry( $source_url, $entry_id ) {
854		- global $wpdb;
855		- $table = $wpdb->prefix . 'mxchat_system_prompt_content';
856		-
857		- $rows = array();
858		-
859		- // Group by the real stored source_url — this INCLUDES "mxchat://" manual
860		- // Direct Content entries (the spec's manual-entry case), which share one
861		- // source_url across their chunk rows. Only the synthetic "_ungrouped_<id>"
862		- // display key (invented by the table view for rows with no source_url) is
863		- // excluded; those fall through to the entry_id lookup below.
864		- if ( ! empty( $source_url ) && strpos( $source_url, '_ungrouped_' ) !== 0 ) {
865		- $rows = $wpdb->get_results( $wpdb->prepare(
866		- "SELECT id, article_content, source_url, content_type FROM {$table} WHERE source_url = %s ORDER BY id ASC",
867		- $source_url
868		- ) );
869		- }
870		-
871		- // Fallback / manual "Direct Content" entries: fetch the single row by id.
872		- if ( empty( $rows ) && $entry_id > 0 ) {
873		- $row = $wpdb->get_row( $wpdb->prepare(
874		- "SELECT id, article_content, source_url, content_type FROM {$table} WHERE id = %d",
875		- $entry_id
876		- ) );
877		- if ( $row ) {
878		- $rows = array( $row );
879		- }
880		- }
881		-
882		- if ( empty( $rows ) ) {
883		- return new WP_Error( 'not_found', esc_html__('Entry not found in the local knowledge database.', 'mxchat') );
884		- }
885		-
886		- $chunks = array();
887		- $content_type = '';
888		- foreach ( $rows as $row ) {
889		- $parsed = MxChat_Chunker::parse_stored_chunk( $row->article_content );
890		- $text = isset( $parsed['text'] ) ? $parsed['text'] : '';
891		- $index = isset( $parsed['metadata']['chunk_index'] ) ? intval( $parsed['metadata']['chunk_index'] ) : count( $chunks );
892		- $content_type = $row->content_type;
893		- $chunks[] = array(
894		- 'index' => $index,
895		- 'text' => $text,
896		- 'length' => function_exists('mb_strlen') ? mb_strlen( $text ) : strlen( $text ),
897		- 'row_id' => intval( $row->id ),
898		- );
899		- }
900		-
901		- usort( $chunks, function( $a, $b ) { return $a['index'] - $b['index']; } );
902		-
903		- $assembled = implode( "\n\n", wp_list_pluck( $chunks, 'text' ) );
904		-
905		- return array(
906		- 'store' => 'wordpress',
907		- 'source_url' => $source_url,
908		- 'content_type' => $content_type,
909		- 'is_chunked' => count( $chunks ) > 1,
910		- 'chunk_count' => count( $chunks ),
911		- 'assembled' => $assembled,
912		- 'assembled_length' => function_exists('mb_strlen') ? mb_strlen( $assembled ) : strlen( $assembled ),
913		- 'chunks' => array_values( $chunks ),
914		- // WP-DB storage carries no separate vector metadata; surface that fact
915		- // rather than letting the owner guess (the spec's taxonomy question).
916		- 'metadata' => array(),
917		- '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'),
918		- );
919		-}
920		-
921		-/**
922		- * Read-only inspector for Pinecone entries. Mirrors get_pinecone_entry_content()
923		- * but keeps each vector's text + metadata instead of imploding, so the owner can
924		- * confirm exactly which metadata fields (text/source_url/type/last_updated/created_at/bot_id)
925		- * are present per chunk. READ-ONLY.
926		- */
927		-private function inspect_pinecone_entry( $source_url, $entry_id, $bot_id ) {
928		- if ( ! class_exists('MxChat_Pinecone_Manager') ) {
929		- return new WP_Error( 'pinecone_unavailable', esc_html__('Pinecone manager not available.', 'mxchat') );
930		- }
931		-
932		- if ( $bot_id === 'default' || ! class_exists('MxChat_Multi_Bot_Manager') ) {
933		- $pinecone_options = get_option('mxchat_pinecone_addon_options');
934		- $api_key = $pinecone_options['mxchat_pinecone_api_key'] ?? '';
935		- $host = $pinecone_options['mxchat_pinecone_host'] ?? '';
936		- $namespace = $pinecone_options['mxchat_pinecone_namespace'] ?? '';
937		- } else {
938		- $bot_config = apply_filters('mxchat_get_bot_pinecone_config', array(), $bot_id);
939		- $api_key = $bot_config['api_key'] ?? '';
940		- $host = $bot_config['host'] ?? '';
941		- $namespace = $bot_config['namespace'] ?? '';
942		- }
943		-
944		- if ( empty($host) || empty($api_key) ) {
945		- return new WP_Error( 'pinecone_config', esc_html__('Pinecone not configured.', 'mxchat') );
946		- }
947		-
948		- $base_id = md5( $source_url );
949		- $vector_ids = array( $base_id );
950		-
951		- $list_url = "https://{$host}/vectors/list";
952		- $list_body = array( 'prefix' => $base_id . '_chunk_', 'limit' => 100 );
953		- if ( ! empty($namespace) ) {
954		- $list_body['namespace'] = $namespace;
955		- }
956		-
957		- $list_resp = wp_remote_post( $list_url, array(
958		- 'headers' => array( 'Api-Key' => $api_key, 'Content-Type' => 'application/json' ),
959		- 'body' => wp_json_encode( $list_body ),
960		- 'timeout' => 15,
961		- ) );
962		-
963		- if ( ! is_wp_error($list_resp) ) {
964		- $list_data = json_decode( wp_remote_retrieve_body($list_resp), true );
965		- if ( ! empty($list_data['vectors']) ) {
966		- foreach ( $list_data['vectors'] as $v ) {
967		- $vector_ids[] = $v['id'];
968		- }
969		- }
970		- }
971		-
972		- $fetch_url = "https://{$host}/vectors/fetch";
973		- $fetch_body = array( 'ids' => $vector_ids );
974		- if ( ! empty($namespace) ) {
975		- $fetch_body['namespace'] = $namespace;
976		- }
977		-
978		- $fetch_resp = wp_remote_post( $fetch_url, array(
979		- 'headers' => array( 'Api-Key' => $api_key, 'Content-Type' => 'application/json' ),
980		- 'body' => wp_json_encode( $fetch_body ),
981		- 'timeout' => 15,
982		- ) );
983		-
984		- if ( is_wp_error($fetch_resp) ) {
985		- return new WP_Error( 'pinecone_fetch', esc_html__('Failed to fetch from Pinecone.', 'mxchat') );
986		- }
987		-
988		- $fetch_data = json_decode( wp_remote_retrieve_body($fetch_resp), true );
989		- $vectors = $fetch_data['vectors'] ?? array();
990		-
991		- if ( empty($vectors) ) {
992		- return new WP_Error( 'not_found', esc_html__('Entry not found in Pinecone.', 'mxchat') );
993		- }
994		-
995		- // Whitelisted metadata fields the spec calls out — shown so devs can confirm
996		- // what is (and is NOT) stored per vector.
997		- $meta_fields = array( 'text', 'source_url', 'type', 'last_updated', 'created_at', 'bot_id', 'chunk_index', 'total_chunks' );
998		- $chunks = array();
999		- $content_type = '';
1000		- foreach ( $vectors as $vid => $vector ) {
1001		- $meta = isset($vector['metadata']) && is_array($vector['metadata']) ? $vector['metadata'] : array();
1002		- $text = $meta['text'] ?? '';
1003		- $index = isset($meta['chunk_index']) ? intval($meta['chunk_index']) : count($chunks);
1004		- $content_type = $meta['type'] ?? $content_type;
1005		-
1006		- $clean_meta = array();
1007		- foreach ( $meta_fields as $field ) {
1008		- if ( array_key_exists( $field, $meta ) && $field !== 'text' ) {
1009		- $clean_meta[ $field ] = is_scalar( $meta[ $field ] ) ? (string) $meta[ $field ] : wp_json_encode( $meta[ $field ] );
1010		- }
1011		- }
1012		-
1013		- $chunks[] = array(
1014		- 'index' => $index,
1015		- 'text' => $text,
1016		- 'length' => function_exists('mb_strlen') ? mb_strlen( $text ) : strlen( $text ),
1017		- 'vector_id' => (string) $vid,
1018		- 'metadata' => $clean_meta,
1019		- );
1020		- }
1021		-
1022		- usort( $chunks, function( $a, $b ) { return $a['index'] - $b['index']; } );
1023		-
1024		- $assembled = implode( "\n\n", wp_list_pluck( $chunks, 'text' ) );
1025		-
1026		- return array(
1027		- 'store' => 'pinecone',
1028		- 'source_url' => $source_url,
1029		- 'content_type' => $content_type,
1030		- 'is_chunked' => count( $chunks ) > 1,
1031		- 'chunk_count' => count( $chunks ),
1032		- 'assembled' => $assembled,
1033		- 'assembled_length' => function_exists('mb_strlen') ? mb_strlen( $assembled ) : strlen( $assembled ),
1034		- 'chunks' => array_values( $chunks ),
1035		- 'metadata' => array(),
1036		- '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'),
1037		- );
1038		-}
1039		-
1040		-/**
1041		- * AJAX: Save edited content — re-chunks and re-embeds as needed.
1042		- * Works for both WordPress DB and Pinecone entries.
1043		- */
1044		-public function ajax_mxchat_save_entry_content() {
1045		- check_ajax_referer('mxchat_edit_entry_nonce', 'nonce');
1046		-
1047		- if ( ! current_user_can('manage_options') ) {
1048		- wp_send_json_error( array( 'message' => 'Permission denied.' ) );
1049		- }
1050		-
1051		- $source_url = isset($_POST['source_url']) ? sanitize_text_field( wp_unslash($_POST['source_url']) ) : '';
1052		- $entry_id = isset($_POST['entry_id']) ? absint($_POST['entry_id']) : 0;
1053		- $content = isset($_POST['content']) ? wp_kses_post( wp_unslash($_POST['content']) ) : '';
1054		- $data_source = isset($_POST['data_source']) ? sanitize_key($_POST['data_source']) : 'wordpress';
1055		- $bot_id = isset($_POST['bot_id']) ? sanitize_key($_POST['bot_id']) : 'default';
1056		- $content_type = isset($_POST['content_type']) ? sanitize_key($_POST['content_type']) : 'content';
1057		-
1058		- if ( empty($content) ) {
1059		- wp_send_json_error( array( 'message' => 'Content cannot be empty.' ) );
1060		- }
1061		-
1062		- // Get the embedding API key
1063		- $options = get_option('mxchat_options', array());
1064		- $api_key = '';
1065		-
1066		- if ( $bot_id !== 'default' && class_exists('MxChat_Multi_Bot_Manager') ) {
1067		- $bot_options = apply_filters('mxchat_get_bot_options', array(), $bot_id);
1068		- $api_key = $bot_options['api_key'] ?? '';
1069		- }
1070		- if ( empty($api_key) ) {
1071		- $api_key = $options['api_key'] ?? '';
1072		- }
1073		-
1074		- global $wpdb;
1075		- $table = $wpdb->prefix . 'mxchat_system_prompt_content';
1076		-
1077		- // If source_url is empty but we have an entry_id, look it up
1078		- if ( empty($source_url) && $entry_id > 0 && $data_source === 'wordpress' ) {
1079		- $row = $wpdb->get_row( $wpdb->prepare( "SELECT source_url FROM {$table} WHERE id = %d", $entry_id ) );
1080		- if ( $row && ! empty($row->source_url) ) {
1081		- $source_url = $row->source_url;
1082		- }
1083		- }
1084		-
1085		- // For manual entries (no source_url or mxchat:// prefix), delete the old entry by ID first
1086		- // so submit_content_to_db creates a replacement instead of a duplicate
1087		- // Also treat legacy mxchat.ai source URLs as manual — old bug assigned the site URL to manual entries
1088		- $is_legacy_manual = !empty($source_url) && strpos($source_url, 'mxchat.ai') !== false && strpos($source_url, 'mxchat://') !== 0;
1089		- if ( $entry_id > 0 && (empty($source_url) || strpos($source_url, 'mxchat://') === 0 || $is_legacy_manual) ) {
1090		- $wpdb->delete( $table, array( 'id' => $entry_id ), array( '%d' ) );
1091		- // Clear legacy URL so submit_content_to_db generates a unique mxchat:// identifier
1092		- // instead of reusing the shared URL (which would mass-delete other entries with the same URL)
1093		- if ( $is_legacy_manual ) {
1094		- $source_url = '';
1095		- }
1096		- }
1097		-
1098		- // Use the existing submit_content_to_db which handles chunking, Pinecone, and WP DB
1099		- $vector_id = ! empty($source_url) ? md5($source_url) : md5('mxchat_manual_' . $entry_id);
1100		-
1101		- // submit_content_to_db already handles: delete old chunks → re-chunk → re-embed → store
1102		- $result = MxChat_Utils::submit_content_to_db( $content, $source_url, $api_key, $vector_id, $bot_id, $content_type );
1103		-
1104		- if ( is_wp_error($result) ) {
1105		- wp_send_json_error( array( 'message' => $result->get_error_message() ) );
1106		- }
1107		-
1108		- wp_send_json_success( array( 'message' => 'Content saved and re-embedded successfully.' ) );
1109		-}
1110		-
1111		-public function mxchat_get_pdf_processing_status($pdf_url) {
1112		- $pdf_url = esc_url_raw($pdf_url);
1113		- $status = get_transient(sanitize_key('mxchat_pdf_status_' . md5($pdf_url)));
1114		-
1115		- if (!$status || !is_array($status)) {
1116		- return false;
1117		- }
1118		-
1119		- // Check for stalled processing (no updates for 5 minutes)
1120		- if ($status['status'] === 'processing' && (time() - absint($status['last_update'])) > 300) {
1121		- $status['status'] = 'error';
1122		- $status['error'] = __('PDF processing appears to be stalled. No updates for over 5 minutes.', 'mxchat');
1123		-
1124		- // Save the updated status
1125		- set_transient(
1126		- sanitize_key('mxchat_pdf_status_' . md5($pdf_url)),
1127		- array_map('sanitize_text_field', $status),
1128		- DAY_IN_SECONDS
1129		- );
1130		- }
1131		-
1132		- $result = array(
1133		- 'total_pages' => absint($status['total_pages']),
1134		- 'processed_pages' => absint($status['processed_pages']),
1135		- 'failed_pages' => absint($status['failed_pages'] ?? 0),
1136		- 'percentage' => ($status['total_pages'] > 0)
1137		- ? round((absint($status['processed_pages']) / absint($status['total_pages'])) * 100)
1138		- : 0,
1139		- 'status' => sanitize_text_field($status['status']),
1140		- 'last_update' => human_time_diff(absint($status['last_update']), time()) . ' ' . esc_html__('ago', 'mxchat'),
1141		- 'failed_pages_list' => isset($status['failed_pages_list']) ? $status['failed_pages_list'] : array(),
1142		- 'completion_summary' => isset($status['completion_summary']) ? $status['completion_summary'] : null
1143		- );
1144		-
1145		- // Add error message if present
1146		- if (isset($status['error']) && !empty($status['error'])) {
1147		- $result['error'] = sanitize_text_field($status['error']);
1148		- }
1149		-
1150		- return $result;
1151		-}
1152		-
1153		-
1154		-public function mxchat_handle_sitemap_submission() {
1155		- // Check if the form was submitted and verify permissions
1156		- if (!isset($_POST['submit_sitemap']) || !current_user_can('manage_options')) {
1157		- wp_die(esc_html__('Unauthorized access', 'mxchat'));
1158		- }
1159		-
1160		- // Verify nonce
1161		- check_admin_referer('mxchat_submit_sitemap_action', 'mxchat_submit_sitemap_nonce');
1162		-
1163		- // Validate URL
1164		- if (!isset($_POST['sitemap_url']) || empty($_POST['sitemap_url'])) {
1165		- set_transient('mxchat_admin_notice_error',
1166		- esc_html__('Please provide a valid URL.', 'mxchat'),
1167		- 30
1168		- );
1169		- wp_safe_redirect(esc_url(admin_url('admin.php?page=mxchat-prompts')));
1170		- exit;
1171		- }
1172		-
1173		- $submitted_url = esc_url_raw($_POST['sitemap_url']);
1174		-
1175		- // Convert Google Drive sharing URLs to direct download URLs
1176		- if ( strpos($submitted_url, 'drive.google.com') !== false ) {
1177		- $file_id = '';
1178		- if ( preg_match('/[?&]id=([a-zA-Z0-9_-]+)/', $submitted_url, $m) ) {
1179		- $file_id = $m[1];
1180		- } elseif ( preg_match('#/file/d/([a-zA-Z0-9_-]+)#', $submitted_url, $m) ) {
1181		- $file_id = $m[1];
1182		- }
1183		- if ( ! empty($file_id) ) {
1184		- $submitted_url = 'https://drive.google.com/uc?export=download&id=' . $file_id;
1185		- }
1186		- }
1187		-
1188		- // Get bot_id from form submission
1189		- $bot_id = isset($_POST['bot_id']) ? sanitize_key($_POST['bot_id']) : 'default';
1190		-
1191		- // Get bot-specific options and validate API key
1192		- $bot_options = $this->get_bot_options($bot_id);
1193		- $options = !empty($bot_options) ? $bot_options : get_option('mxchat_options');
1194		- $selected_model = $options['embedding_model'] ?? 'text-embedding-ada-002';
1195		-
1196		- if (strpos($selected_model, 'voyage') === 0) {
1197		- $api_key = $options['voyage_api_key'] ?? '';
1198		- $provider_name = 'Voyage AI';
1199		- } elseif (strpos($selected_model, 'gemini-embedding') === 0) {
1200		- $api_key = $options['gemini_api_key'] ?? '';
1201		- $provider_name = 'Google Gemini';
1202		- } else {
1203		- $api_key = $options['api_key'] ?? '';
1204		- $provider_name = 'OpenAI';
1205		- }
1206		-
1207		- if (empty($api_key)) {
1208		- $error_message = sprintf(
1209		- esc_html__('%s API key is not configured. Please add your API key in the settings before submitting content.', 'mxchat'),
1210		- $provider_name
1211		- );
1212		- set_transient('mxchat_admin_notice_error', $error_message, 30);
1213		- wp_safe_redirect(esc_url(admin_url('admin.php?page=mxchat-prompts')));
1214		- exit;
1215		- }
1216		-
1217		- // Fetch URL — use browser-like headers so servers with bot protection don't block us
1218		- $response = wp_remote_get($submitted_url, array(
1219		- 'timeout' => 30,
1220		- 'sslverify' => false,
1221		- 'user-agent' => 'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/131.0.0.0 Safari/537.36',
1222		- 'headers' => array(
1223		- 'Accept' => 'text/html,application/xhtml+xml,application/xml;q=0.9,*/*;q=0.8',
1224		- 'Accept-Language' => 'en-US,en;q=0.9',
1225		- ),
1226		- ));
1227		-
1228		- if (is_wp_error($response) || wp_remote_retrieve_response_code($response) !== 200) {
1229		- $error_message = is_wp_error($response) ? $response->get_error_message() : 'HTTP Status: ' . wp_remote_retrieve_response_code($response);
1230		- set_transient('mxchat_admin_notice_error',
1231		- sprintf(
1232		- esc_html__('Failed to fetch the URL: %s', 'mxchat'),
1233		- esc_html($error_message)
1234		- ),
1235		- 30
1236		- );
1237		- wp_safe_redirect(esc_url(admin_url('admin.php?page=mxchat-prompts')));
1238		- exit;
1239		- }
1240		-
1241		- $content_type = wp_remote_retrieve_header($response, 'content-type');
1242		- $body_content = wp_remote_retrieve_body($response);
1243		-
1244		- if (empty($body_content)) {
1245		- set_transient('mxchat_admin_notice_error',
1246		- esc_html__('Empty response received from URL.', 'mxchat'),
1247		- 30
1248		- );
1249		- wp_safe_redirect(esc_url(admin_url('admin.php?page=mxchat-prompts')));
1250		- exit;
1251		- }
1252		-
1253		- // Handle PDF URL
1254		- if ($this->mxchat_is_pdf_url($submitted_url, $response)) {
1255		- $result = $this->mxchat_handle_pdf_for_knowledge_base($submitted_url, $response, $bot_id);
1256		-
1257		- if ($result === 'queued') {
1258		- set_transient('mxchat_admin_notice_success',
1259		- esc_html__('PDF queued for processing. Processing will start automatically.', 'mxchat'),
1260		- 30
1261		- );
1262		- } else {
1263		- set_transient('mxchat_admin_notice_error',
1264		- esc_html__('Failed to queue PDF processing: ', 'mxchat') . esc_html($result),
1265		- 30
1266		- );
1267		- }
1268		-
1269		- wp_safe_redirect(esc_url(admin_url('admin.php?page=mxchat-prompts')));
1270		- exit;
1271		- }
1272		-
1273		- // Handle Sitemap XML
1274		- if (strpos($content_type, 'xml') !== false || strpos($body_content, '<urlset') !== false) {
1275		- libxml_use_internal_errors(true);
1276		- $xml = simplexml_load_string($body_content);
1277		- $xml_errors = libxml_get_errors();
1278		- libxml_clear_errors();
1279		-
1280		- if ($xml === false || !empty($xml_errors)) {
1281		- set_transient('mxchat_admin_notice_error',
1282		- esc_html__('Invalid sitemap XML. Please provide a valid sitemap.', 'mxchat'),
1283		- 30
1284		- );
1285		- wp_safe_redirect(esc_url(admin_url('admin.php?page=mxchat-prompts')));
1286		- exit;
1287		- }
1288		-
1289		- $result = $this->mxchat_handle_sitemap_for_knowledge_base($xml, $submitted_url, $bot_id);
1290		-
1291		- if ($result === 'queued') {
1292		- set_transient('mxchat_admin_notice_success',
1293		- esc_html__('Sitemap queued for processing. Processing will start automatically.', 'mxchat'),
1294		- 30
1295		- );
1296		- } else {
1297		- set_transient('mxchat_admin_notice_error',
1298		- esc_html__('Failed to queue sitemap processing. Please check the status below for details.', 'mxchat'),
1299		- 30
1300		- );
1301		- }
1302		-
1303		- wp_safe_redirect(esc_url(admin_url('admin.php?page=mxchat-prompts')));
1304		- exit;
1305		- }
1306		-
1307		- // Handle Regular URL (single page)
1308		- $page_content = $this->mxchat_extract_main_content($body_content);
1309		- $sanitized_content = $this->mxchat_sanitize_content_for_api($page_content);
1310		-
1311		- //error_log('[MXCHAT-URL-DEBUG] URL: ' . $submitted_url);
1312		- //error_log('[MXCHAT-URL-DEBUG] Extracted page_content length: ' . strlen($page_content));
1313		- //error_log('[MXCHAT-URL-DEBUG] Sanitized content length: ' . strlen($sanitized_content));
1314		- //error_log('[MXCHAT-URL-DEBUG] Content preview: ' . substr($sanitized_content, 0, 500));
1315		-
1316		- if (empty($sanitized_content)) {
1317		- set_transient('mxchat_admin_notice_error',
1318		- esc_html__('No valid content found on the provided URL.', 'mxchat'),
1319		- 30
1320		- );
1321		- wp_safe_redirect(esc_url(admin_url('admin.php?page=mxchat-prompts')));
1322		- exit;
1323		- }
1324		-
1325		- // For single URLs, process immediately using submit_content_to_db
1326		- // This handles chunking automatically for large content
1327		- $db_result = MxChat_Utils::submit_content_to_db(
1328		- $sanitized_content,
1329		- $submitted_url,
1330		- $api_key,
1331		- null,
1332		- $bot_id,
1333		- 'url' // content_type
1334		- );
1335		-
1336		- if (is_wp_error($db_result)) {
1337		- $error_message = esc_html__('Failed to store content in database: ', 'mxchat') . esc_html($db_result->get_error_message());
1338		- set_transient('mxchat_admin_notice_error', $error_message, 30);
1339		- } else {
1340		- $success_message = esc_html__('URL content successfully submitted!', 'mxchat');
1341		- set_transient('mxchat_admin_notice_success', $success_message, 30);
1342		- }
1343		-
1344		- wp_safe_redirect(esc_url(admin_url('admin.php?page=mxchat-prompts')));
1345		- exit;
1346		-}
1347		-
1348		-
1349		-public function mxchat_get_single_url_status() {
1350		- $status = get_transient('mxchat_single_url_status');
1351		- if (!$status) {
1352		- return null;
1353		- }
1354		-
1355		- // Add human-readable time
1356		- if (isset($status['timestamp'])) {
1357		- $status['human_time'] = human_time_diff(strtotime($status['timestamp']), current_time('timestamp')) . ' ' . __('ago', 'mxchat');
1358		- }
1359		-
1360		- return $status;
1361		-}
1362		-
1363		-public function mxchat_handle_sitemap_for_knowledge_base($xml, $sitemap_url, $bot_id = 'default') {
1364		- if (!current_user_can('manage_options')) {
1365		- return false;
1366		- }
1367		-
1368		- try {
1369		- $sitemap_url = esc_url_raw($sitemap_url);
1370		-
1371		- if (!$xml || !is_object($xml)) {
1372		- throw new Exception(__('Invalid XML object provided', 'mxchat'));
1373		- }
1374		-
1375		- // Get bot-specific embedding API for validation
1376		- $bot_options = $this->get_bot_options($bot_id);
1377		- $options = !empty($bot_options) ? $bot_options : get_option('mxchat_options');
1378		-
1379		- // Test the embedding API before processing
1380		- $test_phrase = "Test embedding generation for MxChat";
1381		- $test_result = $this->mxchat_generate_embedding($test_phrase, $bot_id);
1382		-
1383		- if (is_string($test_result)) {
1384		- throw new Exception(__('Embedding API validation failed: ', 'mxchat') . $test_result);
1385		- }
1386		-
1387		- if (!is_array($test_result)) {
1388		- throw new Exception(__('Embedding API returned unexpected result type. Please check your configuration.', 'mxchat'));
1389		- }
1390		-
1391		- // Extract URLs from sitemap
1392		- $urls = array();
1393		- foreach ($xml->url as $url_element) {
1394		- $url = esc_url_raw((string)$url_element->loc);
1395		- if ($url) {
1396		- $urls[] = array('url' => $url);
1397		- }
1398		- }
1399		-
1400		- $total_urls = count($urls);
1401		-
1402		- if ($total_urls < 1) {
1403		- throw new Exception(__('No valid URLs found in sitemap', 'mxchat'));
1404		- }
1405		-
1406		- // Create unique queue ID
1407		- $queue_id = 'sitemap_' . md5($sitemap_url . time());
1408		-
1409		- // Add URLs to queue
1410		- $queued_count = $this->mxchat_add_to_queue($queue_id, 'url', $urls, $bot_id);
1411		-
1412		- if ($queued_count === 0) {
1413		- throw new Exception(__('Failed to add URLs to processing queue', 'mxchat'));
1414		- }
1415		-
1416		- // Store queue metadata
1417		- $this->mxchat_set_queue_meta($queue_id, 'source_url', $sitemap_url);
1418		- $this->mxchat_set_queue_meta($queue_id, 'queue_type', 'sitemap');
1419		- $this->mxchat_set_queue_meta($queue_id, 'total_items', $total_urls);
1420		- $this->mxchat_set_queue_meta($queue_id, 'bot_id', $bot_id);
1421		- $this->mxchat_set_queue_meta($queue_id, 'created_at', current_time('mysql'));
1422		-
1423		- // Store queue ID in transient for status tracking
1424		- set_transient('mxchat_active_queue_sitemap', $queue_id, DAY_IN_SECONDS);
1425		- set_transient('mxchat_last_sitemap_url', $sitemap_url, DAY_IN_SECONDS);
1426		-
1427		- return 'queued';
1428		-
1429		- } catch (Exception $e) {
1430		- $error_message = $e->getMessage();
1431		- //error_log(sprintf(esc_html__('Error preparing sitemap for processing: %s', 'mxchat'), esc_html($error_message)));
1432		-
1433		- return $error_message;
1434		- }
1435		-
1436		-}
1437		-
1438		-/**
1439		- * Remove shortcode tags but preserve the content inside them
1440		- * Example: [vc_column]Hello World[/vc_column] becomes "Hello World"
1441		- *
1442		- * @param string $content The content containing shortcodes
1443		- * @return string Content with shortcode tags removed but inner content preserved
1444		- */
1445		-private function strip_shortcode_tags_preserve_content($content) {
1446		- // Single-pass regex removes all shortcode brackets: [tag], [tag attr="val"], [/tag], [tag /]
1447		- // Content between tags is inherently preserved since only brackets are targeted
1448		- $result = preg_replace('/\[\/?\w[\w-]*[^\]]*\]/', '', $content);
1449		- return ($result !== null) ? $result : $content;
1450		-}
1451		-
1452		-public function mxchat_sanitize_content_for_api($content) {
1453		- //error_log('[MXCHAT-SANITIZE] Original content preview: ' . substr($content, 0, 500) . '...');
1454		-
1455		- // Remove shortcode tags but PRESERVE content inside them
1456		- $content = $this->strip_shortcode_tags_preserve_content($content);
1457		-
1458		- // Remove script, style tags, and HTML comments
1459		- $content = preg_replace('/<script\b[^>]*>(.*?)<\/script>/is', '', $content);
1460		- $content = preg_replace('/<style\b[^>]*>(.*?)<\/style>/is', '', $content);
1461		- $content = preg_replace('/<!--(.|\s)*?-->/', '', $content);
1462		-
1463		- // Remove all HTML tags and decode HTML entities
1464		- $content = wp_strip_all_tags($content);
1465		- $content = html_entity_decode($content, ENT_QUOTES | ENT_HTML5);
1466		-
1467		- // Normalize whitespace but preserve paragraph breaks
1468		- // First, normalize line endings to \n
1469		- $content = str_replace(["\r\n", "\r"], "\n", $content);
1470		- // Replace multiple spaces/tabs with single space, but preserve newlines
1471		- $content = preg_replace('/[ \t]+/', ' ', $content);
1472		- // Replace 3+ newlines with 2 newlines (max 2 blank lines)
1473		- $content = preg_replace('/\n{3,}/', "\n\n", $content);
1474		- // Trim each line
1475		- $lines = explode("\n", $content);
1476		- $lines = array_map('trim', $lines);
1477		- $content = implode("\n", $lines);
1478		- // Final trim
1479		- $content = trim($content);
1480		-
1481		- // Remove control characters (which can cause database issues) but preserve \n (0x0A) and \r (0x0D)
1482		- $content = preg_replace('/[\x00-\x08\x0B\x0C\x0E-\x1F\x7F]/u', '', $content);
1483		-
1484		- // Remove NULL bytes which can cause database errors
1485		- $content = str_replace("\0", "", $content);
1486		-
1487		- // Ensure valid UTF-8 encoding
1488		- $content = wp_check_invalid_utf8($content);
1489		-
1490		- // Remove any extremely long strings without spaces (often garbage)
1491		- $content = preg_replace('/\S{300,}/', ' ', $content);
1492		-
1493		- // Replace problematic characters that often cause database issues
1494		- $content = preg_replace('/[\x{10000}-\x{10FFFF}]/u', '', $content); // Remove emoji and other high Unicode characters
1495		-
1496		- // Replace any remaining potentially problematic characters with spaces
1497		- // BUT preserve newlines by temporarily replacing them
1498		- $content = str_replace("\n", "NEWLINE_PLACEHOLDER", $content);
1499		- $content = preg_replace('/[^\p{L}\p{N}\p{P}\p{Z}\p{Sm}]/u', ' ', $content);
1500		- $content = str_replace("NEWLINE_PLACEHOLDER", "\n", $content);
1501		-
1502		- // Limit to reasonable length if needed
1503		- $max_length = 65000; // Just under MySQL TEXT field limit
1504		- if (strlen($content) > $max_length) {
1505		- $content = substr($content, 0, $max_length);
1506		- }
1507		-
1508		- //error_log('[MXCHAT-SANITIZE] Sanitized content preview: ' . substr($content, 0, 500) . '...');
1509		- return $content;
1510		-}
1511		-public function mxchat_extract_main_content($html) {
1512		- if (empty($html)) {
1513		- return '';
1514		- }
1515		- try {
1516		- $dom = new DOMDocument;
1517		- libxml_use_internal_errors(true); // Suppress HTML parsing errors
1518		- @$dom->loadHTML(mb_convert_encoding($html, 'HTML-ENTITIES', 'UTF-8'), LIBXML_HTML_NOIMPLIED | LIBXML_HTML_NODEFDTD);
1519		- $xpath = new DOMXPath($dom);
1520		-
1521		- // For debugging purposes
1522		- $debugEnabled = true; // Set to true to enable debugging output
1523		- $debug = function($message) use ($debugEnabled) {
1524		- if ($debugEnabled) {
1525		- //error_log('[MXCHAT-EXTRACT-DEBUG] ' . $message);
1526		- }
1527		- };
1528		-
1529		- // Direct targeting for Gerow theme posts
1530		- $post_text = $xpath->query('//div[contains(@class, "post-text")]');
1531		- if ($post_text && $post_text->length > 0) {
1532		- $debug("Found post-text directly");
1533		- $content = '';
1534		- foreach ($post_text as $node) {
1535		- $content .= $dom->saveHTML($node);
1536		- }
1537		- if (!empty($content)) {
1538		- $debug("Returning post-text content");
1539		- return $content;
1540		- }
1541		- }
1542		-
1543		- // Try to get the blog details content which contains the post-text
1544		- $blog_details = $xpath->query('//div[contains(@class, "blog-details-content")]');
1545		- if ($blog_details && $blog_details->length > 0) {
1546		- $debug("Found blog-details-content");
1547		- $content = '';
1548		- foreach ($blog_details as $node) {
1549		- $content .= $dom->saveHTML($node);
1550		- }
1551		- if (!empty($content)) {
1552		- $debug("Returning blog-details-content");
1553		- return $content;
1554		- }
1555		- }
1556		-
1557		- // Try to get the article which contains the blog details
1558		- $article = $xpath->query('//article[contains(@class, "blog-details-wrap")]');
1559		- if ($article && $article->length > 0) {
1560		- $debug("Found article with blog-details-wrap");
1561		- $content = '';
1562		- foreach ($article as $node) {
1563		- $content .= $dom->saveHTML($node);
1564		- }
1565		- if (!empty($content)) {
1566		- $debug("Returning article content");
1567		- return $content;
1568		- }
1569		- }
1570		-
1571		- // Try even broader with the blog-item-wrap
1572		- $blog_item = $xpath->query('//div[contains(@class, "blog-item-wrap")]');
1573		- if ($blog_item && $blog_item->length > 0) {
1574		- $debug("Found blog-item-wrap");
1575		- $content = '';
1576		- foreach ($blog_item as $node) {
1577		- $content .= $dom->saveHTML($node);
1578		- }
1579		- if (!empty($content)) {
1580		- $debug("Returning blog-item-wrap content");
1581		- return $content;
1582		- }
1583		- }
1584		-
1585		- // Specific Gerow theme path
1586		- $gerow_path = $xpath->query('//section[contains(@class, "blog-area")]//div[contains(@class, "post-text")]');
1587		- if ($gerow_path && $gerow_path->length > 0) {
1588		- $debug("Found Gerow theme path to post-text");
1589		- $content = '';
1590		- foreach ($gerow_path as $node) {
1591		- $content .= $dom->saveHTML($node);
1592		- }
1593		- if (!empty($content)) {
1594		- $debug("Returning Gerow post-text content");
1595		- return $content;
1596		- }
1597		- }
1598		-
1599		- // Generic blog post selectors
1600		- $selectors = [
1601		- // Blog post specific selectors
1602		- '//div[contains(@class, "post-text")]',
1603		- '//article[contains(@class, "blog-post-item")]//div[contains(@class, "post-text")]',
1604		- '//div[contains(@class, "blog-details-content")]',
1605		- '//article[contains(@class, "blog-details-wrap")]',
1606		- '//div[contains(@class, "entry-content")]',
1607		- '//div[contains(@class, "blog-content")]',
1608		- '//div[contains(@class, "blog-item-wrap")]',
1609		-
1610		- // More general content selectors
1611		- '//div[contains(@class, "page__content")]',
1612		- '//div[contains(@class, "elementor-widget-container")]',
1613		- '//div[contains(@class, "elementor-text-editor")]',
1614		- '//div[contains(@class, "elementor-widget-text-editor")]',
1615		- '//*[contains(@class, "entry-content")]',
1616		- '//*[contains(@class, "post-content")]',
1617		- '//*[contains(@class, "article-content")]',
1618		- '//*[@id="content"]',
1619		- '//*[@id="main-content"]',
1620		- '//section[contains(@class, "blog-area")]',
1621		- '//article',
1622		- '//main',
1623		- '//div[contains(@class, "content")]'
1624		- ];
1625		-
1626		- // First handle Elementor content - get only leaf widget containers to avoid duplicates
1627		- $debug("Checking for Elementor content");
1628		- // Get widget containers that are direct children of widgets (not nested inside other widget containers)
1629		- $elementor_widgets = $xpath->query('//div[contains(@class, "elementor-widget")]//div[contains(@class, "elementor-widget-container")]');
1630		- if ($elementor_widgets && $elementor_widgets->length > 0) {
1631		- $debug("Found Elementor widgets");
1632		- $seen_content = array(); // Track seen content to avoid duplicates
1633		- $combined_content = '';
1634		- foreach ($elementor_widgets as $widget) {
1635		- $widget_content = $dom->saveHTML($widget);
1636		- if (!empty($widget_content)) {
1637		- // Create a hash of the content to detect duplicates
1638		- $content_hash = md5($widget_content);
1639		- if (!isset($seen_content[$content_hash])) {
1640		- $seen_content[$content_hash] = true;
1641		- $combined_content .= $widget_content;
1642		- }
1643		- }
1644		- }
1645		- if (!empty($combined_content)) {
1646		- $debug("Returning Elementor content");
1647		- return $combined_content;
1648		- }
1649		- }
1650		-
1651		- // Try standard selectors one by one
1652		- foreach ($selectors as $selector) {
1653		- $debug("Trying selector: " . $selector);
1654		- $nodes = $xpath->query($selector);
1655		- if ($nodes && $nodes->length > 0) {
1656		- $debug("Found " . $nodes->length . " matches for selector: " . $selector);
1657		- // Only take the FIRST matching node to avoid duplicate content
1658		- // (pages often have nested or multiple containers with same class)
1659		- $content = $dom->saveHTML($nodes->item(0));
1660		- if (!empty($content)) {
1661		- $debug("Returning content from selector: " . $selector . " (first match only)");
1662		- return $content;
1663		- }
1664		- }
1665		- }
1666		-
1667		- // Manual regex fallback for post-text if DOM methods fail
1668		- $debug("Trying regex fallback");
1669		- if (preg_match('/<div class="post-text">(.*?)<\/div>\s*<\/div>\s*<\/div>/s', $html, $matches)) {
1670		- $debug("Found post-text via regex");
1671		- return '<div class="post-text">' . $matches[1] . '</div>';
1672		- }
1673		-
1674		- // Try to extract the blog section as a whole
1675		- $blog_section = $xpath->query('//section[contains(@class, "blog-area")]');
1676		- if ($blog_section && $blog_section->length > 0) {
1677		- $debug("Found blog-area section");
1678		- $content = '';
1679		- foreach ($blog_section as $node) {
1680		- $content .= $dom->saveHTML($node);
1681		- }
1682		- if (!empty($content)) {
1683		- $debug("Returning blog-area section content");
1684		- return $content;
1685		- }
1686		- }
1687		-
1688		- // Generic container selectors for non-CMS sites (like .asp pages)
1689		- $debug("Trying generic container selectors");
1690		- $generic_selectors = [
1691		- '//div[@id="main"]',
1692		- '//div[@id="wrapper"]',
1693		- '//div[@id="page"]',
1694		- '//div[@id="site-content"]',
1695		- '//div[contains(@class, "main-content")]',
1696		- '//div[contains(@class, "page-content")]',
1697		- '//div[contains(@class, "site-content")]',
1698		- ];
1699		-
1700		- foreach ($generic_selectors as $selector) {
1701		- $debug("Trying generic selector: " . $selector);
1702		- $nodes = $xpath->query($selector);
1703		- if ($nodes && $nodes->length > 0) {
1704		- $content = $dom->saveHTML($nodes->item(0));
1705		- if (!empty($content)) {
1706		- $debug("Returning content from generic selector: " . $selector);
1707		- return $content;
1708		- }
1709		- }
1710		- }
1711		-
1712		- // Paragraph-based content detection - find regions with substantial text
1713		- $debug("Trying paragraph-based content detection");
1714		- $paragraphs = $xpath->query('//p[string-length(normalize-space()) > 50]');
1715		- if ($paragraphs && $paragraphs->length >= 3) {
1716		- $debug("Found " . $paragraphs->length . " substantial paragraphs");
1717		- // Collect all substantial paragraphs and their content
1718		- $paragraph_content = '';
1719		- foreach ($paragraphs as $p) {
1720		- $paragraph_content .= $dom->saveHTML($p) . "\n";
1721		- }
1722		- if (!empty($paragraph_content)) {
1723		- $debug("Returning paragraph-based content");
1724		- return $paragraph_content;
1725		- }
1726		- }
1727		-
1728		- // Improved body fallback - strip nav/header/footer elements first
1729		- $debug("Using improved body fallback");
1730		- $body = $dom->getElementsByTagName('body');
1731		- if ($body->length > 0) {
1732		- // Clone the body to avoid modifying the original DOM
1733		- $body_clone = $body->item(0)->cloneNode(true);
1734		-
1735		- // Remove common non-content elements by tag name
1736		- $remove_tags = ['nav', 'header', 'footer', 'aside', 'script', 'style', 'noscript'];
1737		- foreach ($remove_tags as $tag) {
1738		- $elements = $body_clone->getElementsByTagName($tag);
1739		- // Iterate backwards to safely remove elements
1740		- for ($i = $elements->length - 1; $i >= 0; $i--) {
1741		- $el = $elements->item($i);
1742		- if ($el && $el->parentNode) {
1743		- $el->parentNode->removeChild($el);
1744		- }
1745		- }
1746		- }
1747		-
1748		- // Remove elements with common non-content class names using XPath on the cloned body
1749		- $temp_dom = new DOMDocument();
1750		- @$temp_dom->appendChild($temp_dom->importNode($body_clone, true));
1751		- $temp_xpath = new DOMXPath($temp_dom);
1752		-
1753		- $remove_class_patterns = [
1754		- '//*[contains(@class, "nav")]',
1755		- '//*[contains(@class, "menu")]',
1756		- '//*[contains(@class, "sidebar")]',
1757		- '//*[contains(@class, "footer")]',
1758		- '//*[contains(@class, "header")]',
1759		- '//*[contains(@id, "nav")]',
1760		- '//*[contains(@id, "menu")]',
1761		- '//*[contains(@id, "sidebar")]',
1762		- '//*[contains(@id, "footer")]',
1763		- '//*[contains(@id, "header")]',
1764		- ];
1765		-
1766		- foreach ($remove_class_patterns as $pattern) {
1767		- $elements = $temp_xpath->query($pattern);
1768		- if ($elements) {
1769		- for ($i = $elements->length - 1; $i >= 0; $i--) {
1770		- $el = $elements->item($i);
1771		- if ($el && $el->parentNode) {
1772		- $el->parentNode->removeChild($el);
1773		- }
1774		- }
1775		- }
1776		- }
1777		-
1778		- $cleaned_content = $temp_dom->saveHTML();
1779		- if (!empty($cleaned_content)) {
1780		- $debug("Returning cleaned body content");
1781		- return $cleaned_content;
1782		- }
1783		- }
1784		-
1785		- // Last resort: return the original HTML
1786		- $debug("Returning original HTML");
1787		- return $html;
1788		- } catch (Exception $e) {
1789		- //error_log('[MXCHAT-ERROR] Content extraction failed: ' . $e->getMessage());
1790		- return $html; // Return original HTML if parsing fails
1791		- } finally {
1792		- libxml_clear_errors();
1793		- }
1794		-}
1795		-public function mxchat_get_sitemap_processing_status($sitemap_url) {
1796		- $sitemap_url = esc_url_raw($sitemap_url);
1797		- $status_key = sanitize_key('mxchat_sitemap_status_' . md5($sitemap_url));
1798		- $status = get_transient($status_key);
1799		-
1800		- if (!$status || !is_array($status)) {
1801		- return false;
1802		- }
1803		-
1804		- // Auto-complete check: if all URLs are processed but status isn't complete
1805		- if (isset($status['processed_urls']) && isset($status['total_urls']) &&
1806		- $status['processed_urls'] >= $status['total_urls'] &&
1807		- isset($status['status']) && $status['status'] !== 'complete' &&
1808		- $status['status'] !== 'error') {
1809		-
1810		- // Mark as complete
1811		- $status['status'] = 'complete';
1812		- $status['processed_urls'] = $status['total_urls']; // Ensure exact match
1813		-
1814		- // Update the transient with the corrected status
1815		- set_transient($status_key, $status, DAY_IN_SECONDS);
1816		- }
1817		-
1818		- return array(
1819		- 'total_urls' => absint($status['total_urls']),
1820		- 'processed_urls' => absint($status['processed_urls']),
1821		- 'failed_urls' => absint($status['failed_urls'] ?? 0),
1822		- 'percentage' => ($status['total_urls'] > 0)
1823		- ? round((absint($status['processed_urls']) / absint($status['total_urls'])) * 100)
1824		- : 0,
1825		- 'status' => sanitize_text_field($status['status']),
1826		- 'last_update' => human_time_diff(absint($status['last_update']), time()) . ' ' . esc_html__('ago', 'mxchat'),
1827		- 'error' => isset($status['error']) ? sanitize_text_field($status['error']) : '',
1828		- 'last_error' => isset($status['last_error']) ? sanitize_text_field($status['last_error']) : '',
1829		- 'failed_urls_list' => isset($status['failed_urls_list']) ? $status['failed_urls_list'] : array()
1830		- );
1831		-}
1832		-
1833		-public function mxchat_ajax_get_status_updates() {
1834		- try {
1835		- // Verify the request
1836		- check_ajax_referer('mxchat_status_nonce', 'nonce');
1837		-
1838		- // Get active queue IDs
1839		- $sitemap_queue_id = get_transient('mxchat_active_queue_sitemap');
1840		- $pdf_queue_id = get_transient('mxchat_active_queue_pdf');
1841		-
1842		- $sitemap_status = false;
1843		- $pdf_status = false;
1844		-
1845		- // Get sitemap queue status
1846		- if ($sitemap_queue_id) {
1847		- $sitemap_status = $this->mxchat_get_queue_status_data($sitemap_queue_id, 'sitemap');
1848		- }
1849		-
1850		- // Get PDF queue status
1851		- if ($pdf_queue_id) {
1852		- $pdf_status = $this->mxchat_get_queue_status_data($pdf_queue_id, 'pdf');
1853		- }
1854		-
1855		- $is_active_processing =
1856		- ($sitemap_status && $sitemap_status['status'] === 'processing') ||
1857		- ($pdf_status && $pdf_status['status'] === 'processing');
1858		-
1859		- // Return JSON response with the status data
1860		- wp_send_json(array(
1861		- 'pdf_status' => $pdf_status,
1862		- 'sitemap_status' => $sitemap_status,
1863		- 'is_processing' => $is_active_processing,
1864		- 'sitemap_queue_id' => $sitemap_queue_id,
1865		- 'pdf_queue_id' => $pdf_queue_id
1866		- ));
1867		-
1868		- } catch (Exception $e) {
1869		- //error_log('MxChat Status Update Error: ' . $e->getMessage());
1870		-
1871		- wp_send_json_error(array(
1872		- 'message' => 'Error getting status updates: ' . $e->getMessage(),
1873		- 'status' => 'error'
1874		- ));
1875		- }
1876		-}
1877		-
1878		-/**
1879		- * Helper function to get queue status data
1880		- */
1881		-private function mxchat_get_queue_status_data($queue_id, $type = 'sitemap') {
1882		- global $wpdb;
1883		- $table_name = $wpdb->prefix . 'mxchat_processing_queue';
1884		-
1885		- // Get counts by status
1886		- $counts = $wpdb->get_results($wpdb->prepare(
1887		- "SELECT status, COUNT(*) as count
1888		- FROM $table_name
1889		- WHERE queue_id = %s
1890		- GROUP BY status",
1891		- $queue_id
1892		- ), OBJECT_K);
1893		-
1894		- $total = 0;
1895		- $completed = 0;
1896		- $failed = 0;
1897		- $processing = 0;
1898		- $pending = 0;
1899		-
1900		- foreach ($counts as $status => $data) {
1901		- $count = absint($data->count);
1902		- $total += $count;
1903		-
1904		- switch ($status) {
1905		- case 'completed':
1906		- $completed = $count;
1907		- break;
1908		- case 'failed':
1909		- $failed = $count;
1910		- break;
1911		- case 'processing':
1912		- $processing = $count;
1913		- break;
1914		- case 'pending':
1915		- $pending = $count;
1916		- break;
1917		- }
1918		- }
1919		-
1920		- if ($total === 0) {
1921		- return false;
1922		- }
1923		-
1924		- // Calculate percentage
1925		- $percentage = round((($completed + $failed) / $total) * 100);
1926		-
1927		- // Get failed items details (limit to 50)
1928		- $failed_items = array();
1929		- if ($failed > 0) {
1930		- $failed_results = $wpdb->get_results($wpdb->prepare(
1931		- "SELECT item_type, item_data, error_message, attempts, completed_at
1932		- FROM $table_name
1933		- WHERE queue_id = %s
1934		- AND status = 'failed'
1935		- AND attempts >= max_attempts
1936		- ORDER BY id DESC
1937		- LIMIT 50",
1938		- $queue_id
1939		- ));
1940		-
1941		- foreach ($failed_results as $item) {
1942		- $data = json_decode($item->item_data, true);
1943		- $url = $type === 'pdf' ? ($data['pdf_url'] ?? '') : ($data['url'] ?? '');
1944		-
1945		- $failed_items[] = array(
1946		- 'url' => $url,
1947		- 'page' => $type === 'pdf' ? ($data['page_number'] ?? 0) : 0,
1948		- 'error' => $item->error_message,
1949		- 'retries' => $item->attempts,
1950		- 'time' => strtotime($item->completed_at)
1951		- );
1952		- }
1953		- }
1954		-
1955		- // Get queue metadata
1956		- $source_url = $this->mxchat_get_queue_meta($queue_id, 'source_url');
1957		-
1958		- // Determine if queue is complete
1959		- $is_complete = ($pending === 0 && $processing === 0);
1960		-
1961		- // Get last update time
1962		- $last_update = $wpdb->get_var($wpdb->prepare(
1963		- "SELECT MAX(UNIX_TIMESTAMP(COALESCE(completed_at, started_at, created_at)))
1964		- FROM $table_name
1965		- WHERE queue_id = %s",
1966		- $queue_id
1967		- ));
1968		-
1969		- $last_update_text = $last_update ? human_time_diff($last_update, time()) . ' ' . __('ago', 'mxchat') : __('Just now', 'mxchat');
1970		-
1971		- // Format based on type
1972		- if ($type === 'pdf') {
1973		- return array(
1974		- 'total_pages' => $total,
1975		- 'processed_pages' => $completed + $failed,
1976		- 'failed_pages' => $failed,
1977		- 'percentage' => $percentage,
1978		- 'status' => $is_complete ? 'complete' : 'processing',
1979		- 'last_update' => $last_update_text,
1980		- 'failed_pages_list' => $failed_items,
1981		- 'pdf_url' => $source_url,
1982		- 'queue_id' => $queue_id
1983		- );
1984		- } else {
1985		- return array(
1986		- 'total_urls' => $total,
1987		- 'processed_urls' => $completed + $failed,
1988		- 'failed_urls' => $failed,
1989		- 'percentage' => $percentage,
1990		- 'status' => $is_complete ? 'complete' : 'processing',
1991		- 'last_update' => $last_update_text,
1992		- 'failed_urls_list' => $failed_items,
1993		- 'sitemap_url' => $source_url,
1994		- 'queue_id' => $queue_id
1995		- );
1996		- }
1997		-}
1998		-
1999		-/**
2000		- * Public method to get processing status for both sitemap and PDF queues
2001		- * Used by admin pages to display processing status
2002		- *
2003		- * @return array Array with 'sitemap_status', 'pdf_status', and 'is_processing' keys
2004		- */
2005		-public function mxchat_get_processing_statuses() {
2006		- $sitemap_queue_id = get_transient('mxchat_active_queue_sitemap');
2007		- $pdf_queue_id = get_transient('mxchat_active_queue_pdf');
2008		-
2009		- $sitemap_status = false;
2010		- $pdf_status = false;
2011		-
2012		- if ($sitemap_queue_id) {
2013		- $sitemap_status = $this->mxchat_get_queue_status_data($sitemap_queue_id, 'sitemap');
2014		- }
2015		-
2016		- if ($pdf_queue_id) {
2017		- $pdf_status = $this->mxchat_get_queue_status_data($pdf_queue_id, 'pdf');
2018		- }
2019		-
2020		- $is_processing =
2021		- ($sitemap_status && $sitemap_status['status'] === 'processing') ||
2022		- ($pdf_status && $pdf_status['status'] === 'processing');
2023		-
2024		- return array(
2025		- 'sitemap_status' => $sitemap_status,
2026		- 'pdf_status' => $pdf_status,
2027		- 'is_processing' => $is_processing
2028		- );
2029		-}
2030		-
2031		-/**
2032		- * AJAX handler to get recent knowledge entries for real-time table updates
2033		- * UPDATED: Now supports both WordPress DB and Pinecone data sources
2034		- */
2035		-public function ajax_mxchat_get_recent_entries() {
2036		- check_ajax_referer('mxchat_entries_nonce', 'nonce');
2037		-
2038		- if (!current_user_can('manage_options')) {
2039		- wp_send_json_error(array('message' => 'Unauthorized'));
2040		- return;
2041		- }
2042		-
2043		- global $wpdb;
2044		- $table_name = $wpdb->prefix . 'mxchat_system_prompt_content';
2045		-
2046		- // Get parameters
2047		- $last_id = isset($_POST['last_id']) ? absint($_POST['last_id']) : 0;
2048		- $limit = isset($_POST['limit']) ? min(absint($_POST['limit']), 50) : 10;
2049		- $bot_id = isset($_POST['bot_id']) ? sanitize_text_field($_POST['bot_id']) : 'default';
2050		-
2051		- // Check if Pinecone is enabled for this bot
2052		- $pinecone_manager = $this->mxchat_get_pinecone_manager();
2053		- $pinecone_options = $pinecone_manager ? $pinecone_manager->mxchat_get_bot_pinecone_options($bot_id) : array();
2054		- $use_pinecone = ($pinecone_options['mxchat_use_pinecone'] ?? '0') === '1';
2055		- $has_pinecone_api = !empty($pinecone_options['mxchat_pinecone_api_key']);
2056		-
2057		- if ($use_pinecone && $has_pinecone_api) {
2058		- // PINECONE DATA SOURCE - Get grouped entry count (not raw vector count)
2059		- // Use mxchat_fetch_pinecone_records which returns total_unique_entries
2060		- $records = $pinecone_manager->mxchat_fetch_pinecone_records($pinecone_options, '', 1, 10, $bot_id, '');
2061		- $total_count = $records['total'] ?? 0;
2062		-
2063		- // For Pinecone, we don't return individual entries during polling
2064		- // (entries are already displayed on page load via mxchat_fetch_pinecone_records)
2065		- // We just return the updated count
2066		- wp_send_json_success(array(
2067		- 'entries' => array(),
2068		- 'total_count' => absint($total_count),
2069		- 'max_id' => $last_id,
2070		- 'data_source' => 'pinecone'
2071		- ));
2072		- return;
2073		- }
2074		-
2075		- // WORDPRESS DB DATA SOURCE
2076		- // Build query to get entries newer than last_id
2077		- $where_clauses = array('1=1');
2078		- $where_values = array();
2079		-
2080		- if ($last_id > 0) {
2081		- $where_clauses[] = 'id > %d';
2082		- $where_values[] = $last_id;
2083		- }
2084		-
2085		- // Note: WordPress DB table doesn't have bot_id column
2086		- // Multi-bot filtering is handled via Pinecone namespaces
2087		-
2088		- $where_sql = implode(' AND ', $where_clauses);
2089		-
2090		- // Get recent entries
2091		- $query = "SELECT id, article_content, source_url, timestamp
2092		- FROM $table_name
2093		- WHERE $where_sql
2094		- ORDER BY id DESC
2095		- LIMIT %d";
2096		-
2097		- $where_values[] = $limit;
2098		-
2099		- $entries = $wpdb->get_results($wpdb->prepare($query, $where_values));
2100		-
2101		- // Get total count of GROUPED entries (by source_url) - matches pagination display
2102		- // Count unique source_urls (excluding mxchat:// internal refs) + count of ungrouped rows
2103		- $total_count = $wpdb->get_var(
2104		- "SELECT (SELECT COUNT(DISTINCT source_url) FROM {$table_name} WHERE source_url != '' AND source_url NOT LIKE 'mxchat://%') +
2105		- (SELECT COUNT(*) FROM {$table_name} WHERE source_url = '' OR source_url IS NULL OR source_url LIKE 'mxchat://%')"
2106		- );
2107		-
2108		- // Format entries for response
2109		- $formatted_entries = array();
2110		- $preview_length = 150;
2111		- foreach ($entries as $entry) {
2112		- // Parse chunk metadata using the proper chunker method (same as initial page load)
2113		- if (class_exists('MxChat_Chunker')) {
2114		- $chunk_meta = MxChat_Chunker::parse_stored_chunk($entry->article_content);
2115		- $display_content = $chunk_meta['text'];
2116		- $chunk_metadata = $chunk_meta['metadata'];
2117		- } else {
2118		- $display_content = $entry->article_content;
2119		- $chunk_metadata = array();
2120		- }
2121		-
2122		- $content_preview = mb_strlen($display_content) > $preview_length
2123		- ? mb_substr($display_content, 0, $preview_length) . '...'
2124		- : $display_content;
2125		-
2126		- $formatted_entries[] = array(
2127		- 'id' => $entry->id,
2128		- 'preview' => esc_html($content_preview),
2129		- 'full_content' => wp_kses_post(wpautop($display_content)),
2130		- 'content_length' => mb_strlen($display_content),
2131		- 'preview_length' => $preview_length,
2132		- 'source_url' => $entry->source_url,
2133		- 'has_link' => !empty($entry->source_url) && strpos($entry->source_url, 'mxchat://') !== 0,
2134		- 'chunk_metadata' => $chunk_metadata,
2135		- 'bot_id' => $entry->bot_id ?? 'default',
2136		- 'edit_nonce' => wp_create_nonce('mxchat_edit_entry_nonce'),
2137		- 'delete_nonce' => wp_create_nonce('mxchat_delete_prompt_nonce')
2138		- );
2139		- }
2140		-
2141		- wp_send_json_success(array(
2142		- 'entries' => $formatted_entries,
2143		- 'total_count' => absint($total_count),
2144		- 'max_id' => !empty($entries) ? $entries[0]->id : $last_id,
2145		- 'data_source' => 'wordpress'
2146		- ));
2147		-}
2148		-
2149		-/**
2150		- * Get Pinecone total count from stats API
2151		- * Helper function for ajax_mxchat_get_recent_entries
2152		- */
2153		-private function mxchat_get_pinecone_count_from_stats($pinecone_options) {
2154		- $api_key = $pinecone_options['mxchat_pinecone_api_key'] ?? '';
2155		- $host = $pinecone_options['mxchat_pinecone_host'] ?? '';
2156		- $namespace = $pinecone_options['mxchat_pinecone_namespace'] ?? '';
2157		-
2158		- if (empty($api_key) || empty($host)) {
2159		- return 0;
2160		- }
2161		-
2162		- try {
2163		- $stats_url = "https://{$host}/describe_index_stats";
2164		-
2165		- $response = wp_remote_post($stats_url, array(
2166		- 'headers' => array(
2167		- 'Api-Key' => $api_key,
2168		- 'Content-Type' => 'application/json'
2169		- ),
2170		- 'body' => '{}',
2171		- 'timeout' => 10
2172		- ));
2173		-
2174		- if (!is_wp_error($response) && wp_remote_retrieve_response_code($response) === 200) {
2175		- $body = wp_remote_retrieve_body($response);
2176		- $stats_data = json_decode($body, true);
2177		-
2178		- // If namespace is specified, get count from that specific namespace
2179		- if (!empty($namespace) && isset($stats_data['namespaces'][$namespace]['vectorCount'])) {
2180		- return intval($stats_data['namespaces'][$namespace]['vectorCount']);
2181		- }
2182		-
2183		- // If no namespace specified or namespace not found in response, use total
2184		- return intval($stats_data['totalVectorCount'] ?? 0);
2185		- }
2186		-
2187		- return 0;
2188		-
2189		- } catch (Exception $e) {
2190		- return 0;
2191		- }
2192		-}
2193		-
2194		-/**
2195		- * AJAX handler to refresh Pinecone entries table via AJAX
2196		- * Returns the table HTML for updating the UI without a full page reload
2197		- */
2198		-public function ajax_mxchat_refresh_pinecone_entries() {
2199		- check_ajax_referer('mxchat_entries_nonce', 'nonce');
2200		-
2201		- if (!current_user_can('manage_options')) {
2202		- wp_send_json_error(array('message' => 'Unauthorized'));
2203		- return;
2204		- }
2205		-
2206		- $bot_id = isset($_POST['bot_id']) ? sanitize_text_field($_POST['bot_id']) : 'default';
2207		- $page = isset($_POST['page']) ? absint($_POST['page']) : 1;
2208		- $per_page = 25;
2209		- $search_query = isset($_POST['search']) ? sanitize_text_field($_POST['search']) : '';
2210		- $content_type_filter = isset($_POST['content_type']) ? sanitize_key($_POST['content_type']) : '';
2211		-
2212		- // Get Pinecone manager and options
2213		- $pinecone_manager = $this->mxchat_get_pinecone_manager();
2214		- if (!$pinecone_manager) {
2215		- wp_send_json_error(array('message' => 'Pinecone manager not available'));
2216		- return;
2217		- }
2218		-
2219		- $pinecone_options = $pinecone_manager->mxchat_get_bot_pinecone_options($bot_id);
2220		- $use_pinecone = ($pinecone_options['mxchat_use_pinecone'] ?? '0') === '1';
2221		- $pinecone_api_key = $pinecone_options['mxchat_pinecone_api_key'] ?? '';
2222		-
2223		- if (!$use_pinecone || empty($pinecone_api_key)) {
2224		- wp_send_json_error(array('message' => 'Pinecone not configured'));
2225		- return;
2226		- }
2227		-
2228		- // Fetch records from Pinecone
2229		- $records = $pinecone_manager->mxchat_fetch_pinecone_records($pinecone_options, $search_query, $page, $per_page, $bot_id, $content_type_filter);
2230		- $prompts = $records['data'] ?? array();
2231		- $total_records = $records['total'] ?? 0;
2232		-
2233		- // Preprocess Pinecone records — set chunk_metadata and display_content
2234		- // (matches admin-knowledge-page.php preprocessing)
2235		- foreach ($prompts as $prompt) {
2236		- if (isset($prompt->chunk_index) && $prompt->chunk_index !== null) {
2237		- $prompt->chunk_metadata = array(
2238		- 'chunk_index' => intval($prompt->chunk_index),
2239		- 'total_chunks' => isset($prompt->total_chunks) ? intval($prompt->total_chunks) : null,
2240		- 'is_chunked' => isset($prompt->is_chunked) ? (bool) $prompt->is_chunked : true,
2241		- 'source_url' => $prompt->source_url ?? ''
2242		- );
2243		- $prompt->display_content = $prompt->article_content;
2244		- } else {
2245		- $prompt->chunk_metadata = array();
2246		- $prompt->display_content = $prompt->article_content ?? '';
2247		- }
2248		- }
2249		-
2250		- // Group prompts by source_url
2251		- $grouped_prompts = array();
2252		- foreach ($prompts as $prompt) {
2253		- $source_url = '';
2254		- if (!empty($prompt->chunk_metadata['source_url'])) {
2255		- $source_url = $prompt->chunk_metadata['source_url'];
2256		- } elseif (!empty($prompt->source_url)) {
2257		- $source_url = $prompt->source_url;
2258		- }
2259		-
2260		- if (!empty($source_url)) {
2261		- if (!isset($grouped_prompts[$source_url])) {
2262		- $grouped_prompts[$source_url] = array();
2263		- }
2264		- $grouped_prompts[$source_url][] = $prompt;
2265		- } else {
2266		- $grouped_prompts['_ungrouped_' . $prompt->id] = array($prompt);
2267		- }
2268		- }
2269		-
2270		- // Sort each group by chunk_index
2271		- foreach ($grouped_prompts as $source_url => &$group) {
2272		- usort($group, function($a, $b) {
2273		- $index_a = isset($a->chunk_metadata['chunk_index']) ? intval($a->chunk_metadata['chunk_index']) : 0;
2274		- $index_b = isset($b->chunk_metadata['chunk_index']) ? intval($b->chunk_metadata['chunk_index']) : 0;
2275		- return $index_a - $index_b;
2276		- });
2277		- }
2278		- unset($group);
2279		-
2280		- // Build HTML for the table rows - must match admin-knowledge-page.php structure exactly
2281		- ob_start();
2282		- $display_index = 0;
2283		- $current_page = $page;
2284		- $data_source = 'pinecone';
2285		- $current_bot_id = $bot_id;
2286		- $preview_length = 150;
2287		-
2288		- if (empty($grouped_prompts)) {
2289		- echo '<tr><td colspan="4" style="padding: 40px; text-align: center; color: var(--mxch-text-muted);">';
2290		- esc_html_e('No knowledge entries found in Pinecone.', 'mxchat');
2291		- echo '</td></tr>';
2292		- } else {
2293		- foreach ($grouped_prompts as $source_url => $group) {
2294		- $chunk_count = count($group);
2295		- $first_prompt = $group[0];
2296		- $display_index++;
2297		-
2298		- if ($chunk_count > 1) {
2299		- // Multiple chunks - show grouped row with expand button
2300		- $group_id = 'group-' . md5($source_url);
2301		- ?>
2302		- <tr id="prompt-<?php echo esc_attr($first_prompt->id); ?>"
2303		- class="mxchat-chunk-group-header"
2304		- data-source="<?php echo esc_attr($data_source); ?>"
2305		- data-group-id="<?php echo esc_attr($group_id); ?>"
2306		- style="border-bottom: 1px solid var(--mxch-card-border); background: rgba(33, 150, 243, 0.02);">
2307		- <td style="padding: 12px 16px; text-align: center;">
2308		- <input type="checkbox"
2309		- class="mxchat-entry-checkbox"
2310		- data-entry-id="<?php echo esc_attr($first_prompt->id); ?>"
2311		- data-source="<?php echo esc_attr($data_source); ?>"
2312		- data-source-url="<?php echo esc_attr($source_url); ?>"
2313		- data-is-group="true"
2314		- data-chunk-count="<?php echo esc_attr($chunk_count); ?>">
2315		- </td>
2316		- <td style="padding: 12px 16px; font-size: 13px;">
2317		- <?php echo esc_html($display_index + (($current_page - 1) * $per_page)); ?>
2318		- </td>
2319		- <td class="mxchat-content-cell" style="padding: 12px 16px; font-size: 13px;">
2320		- <div class="mxchat-chunk-group-info">
2321		- <button type="button" class="mxchat-chunk-toggle" data-group-id="<?php echo esc_attr($group_id); ?>">
2322		- <span class="dashicons dashicons-arrow-right-alt2"></span>
2323		- </button>
2324		- <span class="mxchat-chunk-badge"><?php echo esc_html($chunk_count); ?> <?php esc_html_e('chunks', 'mxchat'); ?></span>
2325		- <span class="mxchat-chunk-preview">
2326		- <?php
2327		- $parent_content = isset($first_prompt->display_content) ? $first_prompt->display_content : (isset($first_prompt->article_content) ? $first_prompt->article_content : '');
2328		- $content_preview = mb_substr($parent_content, 0, 100);
2329		- echo esc_html($content_preview . '...');
2330		- ?>
2331		- </span>
2332		- </div>
2333		- </td>
2334		- <td class="mxchat-url-cell" style="padding: 12px 16px; font-size: 13px;">
2335		- <?php if (!empty($source_url) && strpos($source_url, 'mxchat://') !== 0 && strpos($source_url, '_ungrouped_') !== 0) : ?>
2336		- <a href="<?php echo esc_url($source_url); ?>" target="_blank" style="color: var(--mxch-primary); text-decoration: none;">
2337		- <span class="dashicons dashicons-external" style="font-size: 14px;"></span>
2338		- <?php esc_html_e('View Source', 'mxchat'); ?>
2339		- </a>
2340		- <?php else : ?>
2341		- <span style="color: var(--mxch-text-muted);"><?php esc_html_e('Manual Content', 'mxchat'); ?></span>
2342		- <?php endif; ?>
2343		- </td>
2344		- <td class="mxchat-actions-cell" style="padding: 12px 16px; white-space: nowrap;">
2345		- <?php if ($data_source !== 'pinecone') : ?>
2346		- <button type="button"
2347		- class="mxch-btn mxch-btn-ghost mxch-btn-sm mxchat-edit-entry-btn"
2348		- data-source-url="<?php echo esc_attr($source_url); ?>"
2349		- data-entry-id="<?php echo esc_attr($first_prompt->id); ?>"
2350		- data-data-source="<?php echo esc_attr($data_source); ?>"
2351		- data-bot-id="<?php echo esc_attr($current_bot_id); ?>"
2352		- data-nonce="<?php echo wp_create_nonce('mxchat_edit_entry_nonce'); ?>"
2353		- title="<?php esc_attr_e('Edit content', 'mxchat'); ?>">
2354		- <span class="dashicons dashicons-edit" style="font-size: 14px;"></span>
2355		- </button>
2356		- <?php endif; ?>
2357		- <button type="button"
2358		- class="mxch-btn mxch-btn-ghost mxch-btn-sm delete-button-group"
2359		- data-source-url="<?php echo esc_attr($source_url); ?>"
2360		- data-chunk-count="<?php echo esc_attr($chunk_count); ?>"
2361		- data-data-source="<?php echo esc_attr($data_source); ?>"
2362		- data-bot-id="<?php echo esc_attr($current_bot_id); ?>"
2363		- data-nonce="<?php echo wp_create_nonce('mxchat_delete_chunks_nonce'); ?>"
2364		- style="color: var(--mxch-error);"
2365		- title="<?php esc_attr_e('Delete all chunks', 'mxchat'); ?>">
2366		- <span class="dashicons dashicons-trash" style="font-size: 14px;"></span>
2367		- </button>
2368		- </td>
2369		- </tr>
2370		- <?php
2371		- // Render hidden chunk rows
2372		- foreach ($group as $chunk_index => $chunk) {
2373		- $meta_chunk_index = isset($chunk->chunk_metadata['chunk_index']) ? intval($chunk->chunk_metadata['chunk_index']) : $chunk_index;
2374		- $meta_total_chunks = isset($chunk->chunk_metadata['total_chunks']) ? intval($chunk->chunk_metadata['total_chunks']) : $chunk_count;
2375		- $content = isset($chunk->display_content) ? $chunk->display_content : (isset($chunk->article_content) ? $chunk->article_content : '');
2376		- $content_preview = mb_strlen($content) > $preview_length
2377		- ? mb_substr($content, 0, $preview_length) . '...'
2378		- : $content;
2379		- ?>
2380		- <tr id="prompt-<?php echo esc_attr($chunk->id); ?>"
2381		- class="mxchat-chunk-row <?php echo esc_attr($group_id); ?>"
2382		- data-source="<?php echo esc_attr($data_source); ?>"
2383		- style="display: none; background: #f8f9fa; border-bottom: 1px solid var(--mxch-card-border);">
2384		- <td style="padding: 12px 16px; text-align: center;">
2385		- <!-- Checkbox column placeholder for chunks (managed by group) -->
2386		- </td>
2387		- <td style="padding: 12px 16px 12px 30px; font-size: 13px;">
2388		- <!-- Hidden ID column for chunks -->
2389		- </td>
2390		- <td class="mxchat-content-cell" style="padding: 12px 16px; font-size: 13px;">
2391		- <div class="mxchat-accordion-wrapper">
2392		- <div class="mxchat-content-preview">
2393		- <span class="mxchat-chunk-indicator" style="margin-right: 10px; color: var(--mxch-text-secondary); font-size: 12px;">
2394		- <?php printf(esc_html__('Chunk %d of %d', 'mxchat'), $meta_chunk_index + 1, $meta_total_chunks); ?>
2395		- </span>
2396		- <span class="preview-text"><?php echo esc_html($content_preview); ?></span>
2397		- <?php if (mb_strlen($content) > $preview_length) : ?>
2398		- <button class="mxchat-expand-toggle" type="button">
2399		- <span class="dashicons dashicons-arrow-down-alt2"></span>
2400		- </button>
2401		- <?php endif; ?>
2402		- </div>
2403		- <div class="mxchat-content-full" style="display: none;">
2404		- <div class="content-view">
2405		- <?php
2406		- if (preg_match('/[\x{0590}-\x{05FF}]/u', $content)) {
2407		- echo '<div dir="rtl" lang="he" class="rtl-content">';
2408		- echo wp_kses_post(wpautop($content));
2409		- echo '</div>';
2410		- } else {
2411		- echo wp_kses_post(wpautop($content));
2412		- }
2413		- ?>
2414		- </div>
2415		- </div>
2416		- </div>
2417		- </td>
2418		- <td class="mxchat-url-cell" style="padding: 12px 16px; font-size: 13px;">
2419		- <span class="mxchat-chunk-label" style="color: var(--mxch-text-muted);"><?php esc_html_e('Same as parent', 'mxchat'); ?></span>
2420		- </td>
2421		- <td class="mxchat-actions-cell" style="padding: 12px 16px;">
2422		- <span class="mxchat-chunk-label" style="color: var(--mxch-text-muted);"><?php esc_html_e('Managed by group', 'mxchat'); ?></span>
2423		- </td>
2424		- </tr>
2425		- <?php
2426		- }
2427		- } else {
2428		- // Single entry - display normally with accordion
2429		- $prompt = $first_prompt;
2430		- $content = isset($prompt->display_content) ? $prompt->display_content : (isset($prompt->article_content) ? $prompt->article_content : '');
2431		- $content_preview = mb_strlen($content) > $preview_length
2432		- ? mb_substr($content, 0, $preview_length) . '...'
2433		- : $content;
2434		- ?>
2435		- <tr id="prompt-<?php echo esc_attr($prompt->id); ?>"
2436		- data-source="<?php echo esc_attr($data_source); ?>"
2437		- style="border-bottom: 1px solid var(--mxch-card-border); background: rgba(33, 150, 243, 0.02);">
2438		- <td style="padding: 12px 16px; text-align: center;">
2439		- <input type="checkbox"
2440		- class="mxchat-entry-checkbox"
2441		- data-entry-id="<?php echo esc_attr($prompt->id); ?>"
2442		- data-source="<?php echo esc_attr($data_source); ?>"
2443		- data-source-url="<?php echo esc_attr($prompt->source_url ?? ''); ?>"
2444		- data-is-group="false"
2445		- data-chunk-count="1">
2446		- </td>
2447		- <td style="padding: 12px 16px; font-size: 13px;">
2448		- <?php echo esc_html($display_index + (($current_page - 1) * $per_page)); ?>
2449		- </td>
2450		- <td class="mxchat-content-cell" style="padding: 12px 16px; font-size: 13px;">
2451		- <div class="mxchat-accordion-wrapper">
2452		- <div class="mxchat-content-preview">
2453		- <span class="preview-text"><?php echo esc_html($content_preview); ?></span>
2454		- <?php if (mb_strlen($content) > $preview_length) : ?>
2455		- <button class="mxchat-expand-toggle" type="button">
2456		- <span class="dashicons dashicons-arrow-down-alt2"></span>
2457		- </button>
2458		- <?php endif; ?>
2459		- </div>
2460		- <div class="mxchat-content-full" style="display: none;">
2461		- <div class="content-view">
2462		- <?php
2463		- if (preg_match('/[\x{0590}-\x{05FF}]/u', $content)) {
2464		- echo '<div dir="rtl" lang="he" class="rtl-content">';
2465		- echo wp_kses_post(wpautop($content));
2466		- echo '</div>';
2467		- } else {
2468		- echo wp_kses_post(wpautop($content));
2469		- }
2470		- ?>
2471		- </div>
2472		- </div>
2473		- </div>
2474		- </td>
2475		- <td class="mxchat-url-cell" style="padding: 12px 16px; font-size: 13px;">
2476		- <?php
2477		- $actual_source = $source_url;
2478		- if (strpos($source_url, '_ungrouped_') === 0) {
2479		- $actual_source = $prompt->source_url ?? '';
2480		- }
2481		- if (!empty($actual_source) && strpos($actual_source, 'mxchat://') !== 0) : ?>
2482		- <a href="<?php echo esc_url($actual_source); ?>" target="_blank" style="color: var(--mxch-primary); text-decoration: none;">
2483		- <span class="dashicons dashicons-external" style="font-size: 14px;"></span>
2484		- <?php esc_html_e('View', 'mxchat'); ?>
2485		- </a>
2486		- <?php else : ?>
2487		- <span style="color: var(--mxch-text-muted);"><?php esc_html_e('Manual', 'mxchat'); ?></span>
2488		- <?php endif; ?>
2489		- </td>
2490		- <td style="padding: 12px 16px;">
2491		- <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);">
2492		- <span class="dashicons dashicons-trash" style="font-size: 14px;"></span>
2493		- </button>
2494		- </td>
2495		- </tr>
2496		- <?php
2497		- }
2498		- }
2499		- }
2500		- $html = ob_get_clean();
2501		-
2502		- // Generate pagination HTML for Pinecone
2503		- $total_pages = ceil($total_records / $per_page);
2504		- $pagination_html = '';
2505		- if ($total_pages > 1) {
2506		- $pagination_html = '<div class="mxchat-ajax-pagination" data-current-page="' . esc_attr($page) . '" data-total-pages="' . esc_attr($total_pages) . '" data-search="' . esc_attr($search_query) . '" data-content-type="' . esc_attr($content_type_filter) . '">';
2507		-
2508		- // Previous button
2509		- if ($page > 1) {
2510		- $pagination_html .= '<a href="#" class="mxchat-page-link" data-page="' . ($page - 1) . '">' . esc_html__('&laquo; Previous', 'mxchat') . '</a> ';
2511		- }
2512		-
2513		- // Page numbers
2514		- $start_page = max(1, $page - 2);
2515		- $end_page = min($total_pages, $page + 2);
2516		-
2517		- if ($start_page > 1) {
2518		- $pagination_html .= '<a href="#" class="mxchat-page-link" data-page="1">1</a> ';
2519		- if ($start_page > 2) {
2520		- $pagination_html .= '<span class="mxchat-page-dots">...</span> ';
2521		- }
2522		- }
2523		-
2524		- for ($i = $start_page; $i <= $end_page; $i++) {
2525		- if ($i == $page) {
2526		- $pagination_html .= '<span class="mxchat-page-current">' . $i . '</span> ';
2527		- } else {
2528		- $pagination_html .= '<a href="#" class="mxchat-page-link" data-page="' . $i . '">' . $i . '</a> ';
2529		- }
2530		- }
2531		-
2532		- if ($end_page < $total_pages) {
2533		- if ($end_page < $total_pages - 1) {
2534		- $pagination_html .= '<span class="mxchat-page-dots">...</span> ';
2535		- }
2536		- $pagination_html .= '<a href="#" class="mxchat-page-link" data-page="' . $total_pages . '">' . $total_pages . '</a> ';
2537		- }
2538		-
2539		- // Next button
2540		- if ($page < $total_pages) {
2541		- $pagination_html .= '<a href="#" class="mxchat-page-link" data-page="' . ($page + 1) . '">' . esc_html__('Next &raquo;', 'mxchat') . '</a>';
2542		- }
2543		-
2544		- $pagination_html .= '</div>';
2545		- }
2546		-
2547		- wp_send_json_success(array(
2548		- 'html' => $html,
2549		- 'pagination_html' => $pagination_html,
2550		- 'total_count' => $total_records,
2551		- 'total_pages' => $total_pages,
2552		- 'page' => $page,
2553		- 'per_page' => $per_page,
2554		- 'data_source' => 'pinecone'
2555		- ));
2556		-}
2557		-
2558		-/**
2559		- * AJAX handler for pagination - handles both WordPress DB and Pinecone data sources
2560		- * Returns paginated entries without requiring a full page reload
2561		- */
2562		-public function ajax_mxchat_paginate_entries() {
2563		- check_ajax_referer('mxchat_entries_nonce', 'nonce');
2564		-
2565		- if (!current_user_can('manage_options')) {
2566		- wp_send_json_error(array('message' => 'Unauthorized'));
2567		- return;
2568		- }
2569		-
2570		- $bot_id = isset($_POST['bot_id']) ? sanitize_text_field($_POST['bot_id']) : 'default';
2571		- $page = isset($_POST['page']) ? absint($_POST['page']) : 1;
2572		- $search_query = isset($_POST['search']) ? sanitize_text_field($_POST['search']) : '';
2573		- $content_type_filter = isset($_POST['content_type']) ? sanitize_key($_POST['content_type']) : '';
2574		- $per_page = 25;
2575		-
2576		- // Check if Pinecone is enabled for this bot
2577		- $pinecone_manager = $this->mxchat_get_pinecone_manager();
2578		- $pinecone_options = $pinecone_manager ? $pinecone_manager->mxchat_get_bot_pinecone_options($bot_id) : array();
2579		- $use_pinecone = ($pinecone_options['mxchat_use_pinecone'] ?? '0') === '1';
2580		- $has_pinecone_api = !empty($pinecone_options['mxchat_pinecone_api_key']);
2581		-
2582		- if ($use_pinecone && $has_pinecone_api) {
2583		- // Delegate to Pinecone pagination handler (pass search params)
2584		- $_POST['page'] = $page;
2585		- $_POST['search'] = $search_query;
2586		- $_POST['content_type'] = $content_type_filter;
2587		- $this->ajax_mxchat_refresh_pinecone_entries();
2588		- return;
2589		- }
2590		-
2591		- // WordPress DB pagination - MUST match initial page load logic exactly
2592		- global $wpdb;
2593		- $table_name = $wpdb->prefix . 'mxchat_system_prompt_content';
2594		- $offset = ($page - 1) * $per_page;
2595		-
2596		- // Build WHERE clause for search and content type filtering
2597		- $where_clauses = array();
2598		- $where_values = array();
2599		-
2600		- if ($search_query) {
2601		- $where_clauses[] = "article_content LIKE %s";
2602		- $where_values[] = '%' . $wpdb->esc_like($search_query) . '%';
2603		- }
2604		-
2605		- if ($content_type_filter) {
2606		- switch ($content_type_filter) {
2607		- case 'manual':
2608		- $where_clauses[] = "(source_url = '' OR source_url IS NULL OR source_url LIKE 'mxchat://%')";
2609		- break;
2610		- case 'pdf':
2611		- $where_clauses[] = "source_url LIKE '%.pdf'";
2612		- break;
2613		- case 'url':
2614		- $where_clauses[] = "source_url != '' AND source_url IS NOT NULL AND source_url NOT LIKE 'mxchat://%' AND source_url NOT LIKE '%.pdf'";
2615		- break;
2616		- }
2617		- }
2618		-
2619		- $where_sql = !empty($where_clauses) ? 'WHERE ' . implode(' AND ', $where_clauses) : '';
2620		-
2621		- // Count grouped entries with filters applied
2622		- if (!empty($where_values)) {
2623		- $count_args = array_merge($where_values, $where_values);
2624		- $count_query = $wpdb->prepare(
2625		- "SELECT (SELECT COUNT(DISTINCT source_url) FROM {$table_name} {$where_sql} AND source_url != '' AND source_url NOT LIKE 'mxchat://%') +
2626		- (SELECT COUNT(*) FROM {$table_name} {$where_sql} AND (source_url = '' OR source_url IS NULL OR source_url LIKE 'mxchat://%'))",
2627		- ...$count_args
2628		- );
2629		- $total_records = $wpdb->get_var($count_query);
2630		- } else if (!empty($where_sql)) {
2631		- // Content type filter only (no search), no prepared values needed
2632		- $total_records = $wpdb->get_var(
2633		- "SELECT (SELECT COUNT(DISTINCT source_url) FROM {$table_name} {$where_sql} AND source_url != '' AND source_url NOT LIKE 'mxchat://%') +
2634		- (SELECT COUNT(*) FROM {$table_name} {$where_sql} AND (source_url = '' OR source_url IS NULL OR source_url LIKE 'mxchat://%'))"
2635		- );
2636		- } else {
2637		- // No filters
2638		- $total_records = $wpdb->get_var(
2639		- "SELECT (SELECT COUNT(DISTINCT source_url) FROM {$table_name} WHERE source_url != '' AND source_url NOT LIKE 'mxchat://%') +
2640		- (SELECT COUNT(*) FROM {$table_name} WHERE source_url = '' OR source_url IS NULL OR source_url LIKE 'mxchat://%')"
2641		- );
2642		- }
2643		- $total_pages = ceil($total_records / $per_page);
2644		-
2645		- // Step 1: Get unique source_urls for this page (ordered by latest timestamp) with filters
2646		- if (!empty($where_values)) {
2647		- $query_args = array_merge($where_values, array($per_page, $offset));
2648		- $urls_query = $wpdb->prepare(
2649		- "SELECT source_url, MAX(timestamp) as latest_ts FROM {$table_name}
2650		- {$where_sql}
2651		- GROUP BY source_url ORDER BY latest_ts DESC LIMIT %d OFFSET %d",
2652		- ...$query_args
2653		- );
2654		- } else if (!empty($where_sql)) {
2655		- $urls_query = $wpdb->prepare(
2656		- "SELECT source_url, MAX(timestamp) as latest_ts FROM {$table_name}
2657		- {$where_sql}
2658		- GROUP BY source_url ORDER BY latest_ts DESC LIMIT %d OFFSET %d",
2659		- $per_page, $offset
2660		- );
2661		- } else {
2662		- $urls_query = $wpdb->prepare(
2663		- "SELECT source_url, MAX(timestamp) as latest_ts FROM {$table_name}
2664		- GROUP BY source_url ORDER BY latest_ts DESC LIMIT %d OFFSET %d",
2665		- $per_page, $offset
2666		- );
2667		- }
2668		- $page_urls = $wpdb->get_results($urls_query);
2669		-
2670		- // Step 2: Build list of source_urls to fetch
2671		- $url_list = array();
2672		- $url_order_map = array();
2673		- $order_index = 0;
2674		- foreach ($page_urls as $url_row) {
2675		- $url = $url_row->source_url;
2676		- $url_list[] = $url;
2677		- $url_order_map[$url] = $order_index++;
2678		- }
2679		-
2680		- // Step 3: Fetch all rows for these source_urls (with search filter if applicable)
2681		- $prompts = array();
2682		- if (!empty($url_list)) {
2683		- $placeholders = implode(',', array_fill(0, count($url_list), '%s'));
2684		- if ($search_query) {
2685		- // Include search filter in the final fetch
2686		- $prompts_query = $wpdb->prepare(
2687		- "SELECT id, article_content, source_url, timestamp, role_restriction
2688		- FROM {$table_name}
2689		- WHERE source_url IN ($placeholders) AND article_content LIKE %s
2690		- ORDER BY timestamp DESC",
2691		- ...array_merge($url_list, ['%' . $wpdb->esc_like($search_query) . '%'])
2692		- );
2693		- } else {
2694		- $prompts_query = $wpdb->prepare(
2695		- "SELECT id, article_content, source_url, timestamp, role_restriction
2696		- FROM {$table_name}
2697		- WHERE source_url IN ($placeholders)
2698		- ORDER BY timestamp DESC",
2699		- $url_list
2700		- );
2701		- }
2702		- $prompts = $wpdb->get_results($prompts_query);
2703		- }
2704		-
2705		- // Group prompts by source_url for chunk display
2706		- $grouped_prompts = array();
2707		- foreach ($prompts as $prompt) {
2708		- $source_url = $prompt->source_url ?? '';
2709		-
2710		- // Parse chunk metadata using the proper chunker method (same as initial page load)
2711		- if (class_exists('MxChat_Chunker')) {
2712		- $chunk_meta = MxChat_Chunker::parse_stored_chunk($prompt->article_content);
2713		- $prompt->chunk_metadata = $chunk_meta['metadata'];
2714		- $prompt->display_content = $chunk_meta['text'];
2715		- } else {
2716		- $prompt->chunk_metadata = array();
2717		- $prompt->display_content = $prompt->article_content;
2718		- }
2719		-
2720		- if (!empty($source_url) && strpos($source_url, 'mxchat://') !== 0) {
2721		- if (!isset($grouped_prompts[$source_url])) {
2722		- $grouped_prompts[$source_url] = array();
2723		- }
2724		- $grouped_prompts[$source_url][] = $prompt;
2725		- } else {
2726		- // Ungrouped entries
2727		- $grouped_prompts['_ungrouped_' . $prompt->id] = array($prompt);
2728		- }
2729		- }
2730		-
2731		- // Sort groups by the original URL order (newest first)
2732		- uksort($grouped_prompts, function($a, $b) use ($url_order_map) {
2733		- $order_a = isset($url_order_map[$a]) ? $url_order_map[$a] : PHP_INT_MAX;
2734		- $order_b = isset($url_order_map[$b]) ? $url_order_map[$b] : PHP_INT_MAX;
2735		- return $order_a - $order_b;
2736		- });
2737		-
2738		- // Sort each group internally by chunk_index
2739		- foreach ($grouped_prompts as $source_url => &$group) {
2740		- usort($group, function($a, $b) {
2741		- $index_a = isset($a->chunk_metadata['chunk_index']) ? intval($a->chunk_metadata['chunk_index']) : 0;
2742		- $index_b = isset($b->chunk_metadata['chunk_index']) ? intval($b->chunk_metadata['chunk_index']) : 0;
2743		- return $index_a - $index_b;
2744		- });
2745		- }
2746		- unset($group);
2747		-
2748		- // Build HTML for the table rows
2749		- ob_start();
2750		- $display_index = 0;
2751		- $current_page = $page;
2752		- $data_source = 'wordpress';
2753		- $current_bot_id = $bot_id;
2754		- $preview_length = 150;
2755		-
2756		- if (empty($grouped_prompts)) {
2757		- echo '<tr><td colspan="5" style="padding: 40px; text-align: center; color: var(--mxch-text-muted);">';
2758		- esc_html_e('No knowledge entries found. Use the Import Options to add content.', 'mxchat');
2759		- echo '</td></tr>';
2760		- } else {
2761		- foreach ($grouped_prompts as $source_url => $group) {
2762		- $chunk_count = count($group);
2763		- $first_prompt = $group[0];
2764		- $display_index++;
2765		-
2766		- if ($chunk_count > 1) {
2767		- // Multiple chunks - show grouped row with expand button
2768		- $group_id = 'group-' . md5($source_url);
2769		- ?>
2770		- <tr id="prompt-<?php echo esc_attr($first_prompt->id); ?>"
2771		- class="mxchat-chunk-group-header"
2772		- data-source="<?php echo esc_attr($data_source); ?>"
2773		- data-group-id="<?php echo esc_attr($group_id); ?>"
2774		- style="border-bottom: 1px solid var(--mxch-card-border);">
2775		- <td style="padding: 12px 16px; text-align: center;">
2776		- <input type="checkbox"
2777		- class="mxchat-entry-checkbox"
2778		- data-entry-id="<?php echo esc_attr($first_prompt->id); ?>"
2779		- data-source="<?php echo esc_attr($data_source); ?>"
2780		- data-source-url="<?php echo esc_attr($source_url); ?>"
2781		- data-is-group="true"
2782		- data-chunk-count="<?php echo esc_attr($chunk_count); ?>">
2783		- </td>
2784		- <td style="padding: 12px 16px; font-size: 13px;">
2785		- <?php echo esc_html($first_prompt->id); ?>
2786		- </td>
2787		- <td class="mxchat-content-cell" style="padding: 12px 16px; font-size: 13px;">
2788		- <div class="mxchat-chunk-group-info">
2789		- <button type="button" class="mxchat-chunk-toggle" data-group-id="<?php echo esc_attr($group_id); ?>">
2790		- <span class="dashicons dashicons-arrow-right-alt2"></span>
2791		- </button>
2792		- <span class="mxchat-chunk-badge"><?php echo esc_html($chunk_count); ?> <?php esc_html_e('chunks', 'mxchat'); ?></span>
2793		- <span class="mxchat-chunk-preview">
2794		- <?php
2795		- $parent_content = isset($first_prompt->display_content) ? $first_prompt->display_content : $first_prompt->article_content;
2796		- $content_preview = mb_substr($parent_content, 0, 100);
2797		- echo esc_html($content_preview . '...');
2798		- ?>
2799		- </span>
2800		- </div>
2801		- </td>
2802		- <td class="mxchat-url-cell" style="padding: 12px 16px; font-size: 13px;">
2803		- <?php if (!empty($source_url) && strpos($source_url, 'mxchat://') !== 0 && strpos($source_url, '_ungrouped_') !== 0) : ?>
2804		- <a href="<?php echo esc_url($source_url); ?>" target="_blank" style="color: var(--mxch-primary); text-decoration: none;">
2805		- <span class="dashicons dashicons-external" style="font-size: 14px;"></span>
2806		- <?php esc_html_e('View Source', 'mxchat'); ?>
2807		- </a>
2808		- <?php else : ?>
2809		- <span style="color: var(--mxch-text-muted);"><?php esc_html_e('Manual Content', 'mxchat'); ?></span>
2810		- <?php endif; ?>
2811		- </td>
2812		- <td class="mxchat-actions-cell" style="padding: 12px 16px; white-space: nowrap;">
2813		- <?php if ($data_source !== 'pinecone') : ?>
2814		- <button type="button"
2815		- class="mxch-btn mxch-btn-ghost mxch-btn-sm mxchat-edit-entry-btn"
2816		- data-source-url="<?php echo esc_attr($source_url); ?>"
2817		- data-entry-id="<?php echo esc_attr($first_prompt->id); ?>"
2818		- data-data-source="<?php echo esc_attr($data_source); ?>"
2819		- data-bot-id="<?php echo esc_attr($current_bot_id); ?>"
2820		- data-nonce="<?php echo wp_create_nonce('mxchat_edit_entry_nonce'); ?>"
2821		- title="<?php esc_attr_e('Edit content', 'mxchat'); ?>">
2822		- <span class="dashicons dashicons-edit" style="font-size: 14px;"></span>
2823		- </button>
2824		- <?php endif; ?>
2825		- <button type="button"
2826		- class="mxch-btn mxch-btn-ghost mxch-btn-sm delete-button-group"
2827		- data-source-url="<?php echo esc_attr($source_url); ?>"
2828		- data-chunk-count="<?php echo esc_attr($chunk_count); ?>"
2829		- data-data-source="<?php echo esc_attr($data_source); ?>"
2830		- data-bot-id="<?php echo esc_attr($current_bot_id); ?>"
2831		- data-nonce="<?php echo wp_create_nonce('mxchat_delete_chunks_nonce'); ?>"
2832		- style="color: var(--mxch-error);"
2833		- title="<?php esc_attr_e('Delete all chunks', 'mxchat'); ?>">
2834		- <span class="dashicons dashicons-trash" style="font-size: 14px;"></span>
2835		- </button>
2836		- </td>
2837		- </tr>
2838		- <?php
2839		- // Render hidden chunk rows
2840		- foreach ($group as $chunk_index => $chunk) {
2841		- $meta_chunk_index = isset($chunk->chunk_metadata['chunk_index']) ? intval($chunk->chunk_metadata['chunk_index']) : $chunk_index;
2842		- $meta_total_chunks = isset($chunk->chunk_metadata['total_chunks']) ? intval($chunk->chunk_metadata['total_chunks']) : $chunk_count;
2843		- $content = isset($chunk->display_content) ? $chunk->display_content : $chunk->article_content;
2844		- $content_preview = mb_strlen($content) > $preview_length
2845		- ? mb_substr($content, 0, $preview_length) . '...'
2846		- : $content;
2847		- ?>
2848		- <tr id="prompt-<?php echo esc_attr($chunk->id); ?>"
2849		- class="mxchat-chunk-row <?php echo esc_attr($group_id); ?>"
2850		- data-source="<?php echo esc_attr($data_source); ?>"
2851		- style="display: none; background: #f8f9fa; border-bottom: 1px solid var(--mxch-card-border);">
2852		- <td style="padding: 12px 16px; text-align: center;">
2853		- <!-- Checkbox column placeholder for chunks (managed by group) -->
2854		- </td>
2855		- <td style="padding: 12px 16px 12px 30px; font-size: 13px;">
2856		- <!-- Hidden ID column for chunks -->
2857		- </td>
2858		- <td class="mxchat-content-cell" style="padding: 12px 16px; font-size: 13px;">
2859		- <div class="mxchat-accordion-wrapper">
2860		- <div class="mxchat-content-preview">
2861		- <span class="mxchat-chunk-indicator" style="margin-right: 10px; color: var(--mxch-text-secondary); font-size: 12px;">
2862		- <?php printf(esc_html__('Chunk %d of %d', 'mxchat'), $meta_chunk_index + 1, $meta_total_chunks); ?>
2863		- </span>
2864		- <span class="preview-text"><?php echo esc_html($content_preview); ?></span>
2865		- <?php if (mb_strlen($content) > $preview_length) : ?>
2866		- <button class="mxchat-expand-toggle" type="button">
2867		- <span class="dashicons dashicons-arrow-down-alt2"></span>
2868		- </button>
2869		- <?php endif; ?>
2870		- </div>
2871		- <div class="mxchat-content-full" style="display: none;">
2872		- <div class="content-view">
2873		- <?php
2874		- if (preg_match('/[\x{0590}-\x{05FF}]/u', $content)) {
2875		- echo '<div dir="rtl" lang="he" class="rtl-content">';
2876		- echo wp_kses_post(wpautop($content));
2877		- echo '</div>';
2878		- } else {
2879		- echo wp_kses_post(wpautop($content));
2880		- }
2881		- ?>
2882		- </div>
2883		- </div>
2884		- </div>
2885		- </td>
2886		- <td class="mxchat-url-cell" style="padding: 12px 16px; font-size: 13px;">
2887		- <span class="mxchat-chunk-label" style="color: var(--mxch-text-muted);"><?php esc_html_e('Same as parent', 'mxchat'); ?></span>
2888		- </td>
2889		- <td class="mxchat-actions-cell" style="padding: 12px 16px;">
2890		- <span class="mxchat-chunk-label" style="color: var(--mxch-text-muted);"><?php esc_html_e('Managed by group', 'mxchat'); ?></span>
2891		- </td>
2892		- </tr>
2893		- <?php
2894		- }
2895		- } else {
2896		- // Single entry - display normally with accordion
2897		- $prompt = $first_prompt;
2898		- $content = isset($prompt->display_content) ? $prompt->display_content : $prompt->article_content;
2899		- $content_preview = mb_strlen($content) > $preview_length
2900		- ? mb_substr($content, 0, $preview_length) . '...'
2901		- : $content;
2902		- ?>
2903		- <tr id="prompt-<?php echo esc_attr($prompt->id); ?>"
2904		- data-source="<?php echo esc_attr($data_source); ?>"
2905		- style="border-bottom: 1px solid var(--mxch-card-border);">
2906		- <td style="padding: 12px 16px; text-align: center;">
2907		- <input type="checkbox"
2908		- class="mxchat-entry-checkbox"
2909		- data-entry-id="<?php echo esc_attr($prompt->id); ?>"
2910		- data-source="<?php echo esc_attr($data_source); ?>"
2911		- data-source-url="<?php echo esc_attr($source_url); ?>"
2912		- data-is-group="false">
2913		- </td>
2914		- <td style="padding: 12px 16px; font-size: 13px;">
2915		- <?php echo esc_html($prompt->id); ?>
2916		- </td>
2917		- <td class="mxchat-content-cell" style="padding: 12px 16px; font-size: 13px;">
2918		- <div class="mxchat-accordion-wrapper">
2919		- <div class="mxchat-content-preview">
2920		- <span class="preview-text"><?php echo esc_html($content_preview); ?></span>
2921		- <?php if (mb_strlen($content) > $preview_length) : ?>
2922		- <button class="mxchat-expand-toggle" type="button">
2923		- <span class="dashicons dashicons-arrow-down-alt2"></span>
2924		- </button>
2925		- <?php endif; ?>
2926		- </div>
2927		- <div class="mxchat-content-full" style="display: none;">
2928		- <div class="content-view">
2929		- <?php
2930		- if (preg_match('/[\x{0590}-\x{05FF}]/u', $content)) {
2931		- echo '<div dir="rtl" lang="he" class="rtl-content">';
2932		- echo wp_kses_post(wpautop($content));
2933		- echo '</div>';
2934		- } else {
2935		- echo wp_kses_post(wpautop($content));
2936		- }
2937		- ?>
2938		- </div>
2939		- </div>
2940		- </div>
2941		- </td>
2942		- <td class="mxchat-url-cell" style="padding: 12px 16px; font-size: 13px;">
2943		- <?php
2944		- $actual_source = $source_url;
2945		- if (strpos($source_url, '_ungrouped_') === 0) {
2946		- $actual_source = $prompt->source_url ?? '';
2947		- }
2948		- if (!empty($actual_source) && strpos($actual_source, 'mxchat://') !== 0) : ?>
2949		- <a href="<?php echo esc_url($actual_source); ?>" target="_blank" style="color: var(--mxch-primary); text-decoration: none;">
2950		- <span class="dashicons dashicons-external" style="font-size: 14px;"></span>
2951		- <?php esc_html_e('View', 'mxchat'); ?>
2952		- </a>
2953		- <?php else : ?>
2954		- <span style="color: var(--mxch-text-muted);"><?php esc_html_e('Manual', 'mxchat'); ?></span>
2955		- <?php endif; ?>
2956		- </td>
2957		- <td style="padding: 12px 16px; white-space: nowrap;">
2958		- <button type="button"
2959		- class="mxch-btn mxch-btn-ghost mxch-btn-sm mxchat-edit-entry-btn"
2960		- data-source-url="<?php echo esc_attr($prompt->source_url ?? ''); ?>"
2961		- data-entry-id="<?php echo esc_attr($prompt->id); ?>"
2962		- data-data-source="<?php echo esc_attr($data_source); ?>"
2963		- data-bot-id="<?php echo esc_attr($current_bot_id); ?>"
2964		- data-nonce="<?php echo wp_create_nonce('mxchat_edit_entry_nonce'); ?>"
2965		- title="<?php esc_attr_e('Edit content', 'mxchat'); ?>">
2966		- <span class="dashicons dashicons-edit" style="font-size: 14px;"></span>
2967		- </button>
2968		- <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);">
2969		- <span class="dashicons dashicons-trash" style="font-size: 14px;"></span>
2970		- </button>
2971		- </td>
2972		- </tr>
2973		- <?php
2974		- }
2975		- }
2976		- }
2977		- $html = ob_get_clean();
2978		-
2979		- // Generate pagination HTML (include search/filter data for subsequent pages)
2980		- $pagination_html = '';
2981		- if ($total_pages > 1) {
2982		- $pagination_html = '<div class="mxchat-ajax-pagination" data-current-page="' . esc_attr($page) . '" data-total-pages="' . esc_attr($total_pages) . '" data-search="' . esc_attr($search_query) . '" data-content-type="' . esc_attr($content_type_filter) . '">';
2983		-
2984		- // Previous button
2985		- if ($page > 1) {
2986		- $pagination_html .= '<a href="#" class="mxchat-page-link" data-page="' . ($page - 1) . '">' . esc_html__('&laquo; Previous', 'mxchat') . '</a> ';
2987		- }
2988		-
2989		- // Page numbers
2990		- $start_page = max(1, $page - 2);
2991		- $end_page = min($total_pages, $page + 2);
2992		-
2993		- if ($start_page > 1) {
2994		- $pagination_html .= '<a href="#" class="mxchat-page-link" data-page="1">1</a> ';
2995		- if ($start_page > 2) {
2996		- $pagination_html .= '<span class="mxchat-page-dots">...</span> ';
2997		- }
2998		- }
2999		-
3000		- for ($i = $start_page; $i <= $end_page; $i++) {
3001		- if ($i == $page) {
3002		- $pagination_html .= '<span class="mxchat-page-current">' . $i . '</span> ';
3003		- } else {
3004		- $pagination_html .= '<a href="#" class="mxchat-page-link" data-page="' . $i . '">' . $i . '</a> ';
3005		- }
3006		- }
3007		-
3008		- if ($end_page < $total_pages) {
3009		- if ($end_page < $total_pages - 1) {
3010		- $pagination_html .= '<span class="mxchat-page-dots">...</span> ';
3011		- }
3012		- $pagination_html .= '<a href="#" class="mxchat-page-link" data-page="' . $total_pages . '">' . $total_pages . '</a> ';
3013		- }
3014		-
3015		- // Next button
3016		- if ($page < $total_pages) {
3017		- $pagination_html .= '<a href="#" class="mxchat-page-link" data-page="' . ($page + 1) . '">' . esc_html__('Next &raquo;', 'mxchat') . '</a>';
3018		- }
3019		-
3020		- $pagination_html .= '</div>';
3021		- }
3022		-
3023		- wp_send_json_success(array(
3024		- 'html' => $html,
3025		- 'pagination_html' => $pagination_html,
3026		- 'total_count' => $total_records,
3027		- 'total_pages' => $total_pages,
3028		- 'page' => $page,
3029		- 'per_page' => $per_page,
3030		- 'data_source' => 'wordpress'
3031		- ));
3032		-}
3033		-
3034		-/**
3035		- * AJAX handler to detect available sitemaps on the site
3036		- * Optimized for speed - only checks primary sitemap indexes first
3037		- */
3038		-public function ajax_mxchat_detect_sitemaps() {
3039		- check_ajax_referer('mxchat_detect_sitemaps_nonce', 'nonce');
3040		-
3041		- if (!current_user_can('manage_options')) {
3042		- wp_send_json_error(array('message' => 'Unauthorized'));
3043		- return;
3044		- }
3045		-
3046		- $site_url = get_site_url();
3047		- $sitemaps = array();
3048		- $found_index = false;
3049		-
3050		- // Only check the main sitemap index files first (much faster)
3051		- // These are the primary entry points that contain sub-sitemaps
3052		- $primary_indexes = array(
3053		- 'sitemap_index.xml' => 'Yoast SEO / Rank Math', // Yoast & Rank Math
3054		- 'wp-sitemap.xml' => 'WordPress Core', // WordPress Core
3055		- 'sitemap.xml' => 'Standard', // Generic/AIOSEO
3056		- );
3057		-
3058		- foreach ($primary_indexes as $path => $source) {
3059		- $url = trailingslashit($site_url) . $path;
3060		-
3061		- $response = wp_remote_head($url, array(
3062		- 'timeout' => 10,
3063		- 'sslverify' => false,
3064		- 'redirection' => 1,
3065		- 'user-agent' => 'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/131.0.0.0 Safari/537.36',
3066		- ));
3067		-
3068		- if (!is_wp_error($response) && wp_remote_retrieve_response_code($response) === 200) {
3069		- // Found a sitemap index - parse it to get sub-sitemaps
3070		- $sub_sitemaps = $this->parse_sitemap_index($url);
3071		- if (!empty($sub_sitemaps)) {
3072		- $sitemaps[] = array(
3073		- 'url' => $url,
3074		- 'type' => 'index',
3075		- 'source' => $source,
3076		- 'sub_sitemaps' => $sub_sitemaps
3077		- );
3078		- $found_index = true;
3079		- // Found a valid index, no need to check others
3080		- break;
3081		- }
3082		- }
3083		- }
3084		-
3085		- // If no sitemap index found, check for standalone sitemaps
3086		- if (!$found_index) {
3087		- $standalone_sitemaps = array(
3088		- 'post-sitemap.xml' => array('type' => 'content', 'source' => 'Yoast SEO'),
3089		- 'page-sitemap.xml' => array('type' => 'content', 'source' => 'Yoast SEO'),
3090		- );
3091		-
3092		- foreach ($standalone_sitemaps as $path => $info) {
3093		- $url = trailingslashit($site_url) . $path;
3094		-
3095		- $response = wp_remote_head($url, array(
3096		- 'timeout' => 2,
3097		- 'sslverify' => false
3098		- ));
3099		-
3100		- if (!is_wp_error($response) && wp_remote_retrieve_response_code($response) === 200) {
3101		- $sitemaps[] = array(
3102		- 'url' => $url,
3103		- 'type' => $info['type'],
3104		- 'source' => $info['source'],
3105		- 'url_count' => 0 // Skip URL count for speed
3106		- );
3107		- }
3108		- }
3109		- }
3110		-
3111		- wp_send_json_success(array(
3112		- 'sitemaps' => $sitemaps,
3113		- 'site_url' => $site_url
3114		- ));
3115		-}
3116		-
3117		-/**
3118		- * Parse a sitemap index to get sub-sitemaps
3119		- * Optimized: doesn't fetch URL count for each sub-sitemap (too slow)
3120		- */
3121		-private function parse_sitemap_index($url) {
3122		- $sub_sitemaps = array();
3123		-
3124		- $response = wp_remote_get($url, array(
3125		- 'timeout' => 30,
3126		- 'sslverify' => false,
3127		- 'user-agent' => 'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/131.0.0.0 Safari/537.36',
3128		- 'headers' => array(
3129		- 'Accept' => 'text/html,application/xhtml+xml,application/xml;q=0.9,*/*;q=0.8',
3130		- 'Accept-Language' => 'en-US,en;q=0.9',
3131		- ),
3132		- ));
3133		-
3134		- if (is_wp_error($response)) {
3135		- return $sub_sitemaps;
3136		- }
3137		-
3138		- $body = wp_remote_retrieve_body($response);
3139		- if (empty($body)) {
3140		- return $sub_sitemaps;
3141		- }
3142		-
3143		- // Suppress XML errors
3144		- libxml_use_internal_errors(true);
3145		- $xml = simplexml_load_string($body);
3146		- libxml_clear_errors();
3147		-
3148		- if ($xml === false) {
3149		- return $sub_sitemaps;
3150		- }
3151		-
3152		- // Check if it's a sitemap index (contains <sitemap> elements)
3153		- if (isset($xml->sitemap)) {
3154		- foreach ($xml->sitemap as $sitemap) {
3155		- $loc = (string) $sitemap->loc;
3156		- if (!empty($loc)) {
3157		- // Try to determine the type from the URL
3158		- $type = 'content';
3159		- if (strpos($loc, 'category') !== false || strpos($loc, 'tag') !== false || strpos($loc, 'taxonomy') !== false) {
3160		- $type = 'taxonomy';
3161		- } elseif (strpos($loc, 'author') !== false || strpos($loc, 'user') !== false) {
3162		- $type = 'author';
3163		- }
3164		-
3165		- // Skip URL count - too slow to fetch for each sitemap
3166		- $sub_sitemaps[] = array(
3167		- 'url' => $loc,
3168		- 'type' => $type,
3169		- 'url_count' => 0, // Don't fetch - takes too long
3170		- 'name' => basename(parse_url($loc, PHP_URL_PATH))
3171		- );
3172		- }
3173		- }
3174		- }
3175		-
3176		- return $sub_sitemaps;
3177		-}
3178		-
3179		-/**
3180		- * Get URL count from a sitemap
3181		- */
3182		-private function get_sitemap_url_count($url) {
3183		- $response = wp_remote_get($url, array(
3184		- 'timeout' => 30,
3185		- 'sslverify' => false,
3186		- 'user-agent' => 'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/131.0.0.0 Safari/537.36',
3187		- 'headers' => array(
3188		- 'Accept' => 'text/html,application/xhtml+xml,application/xml;q=0.9,*/*;q=0.8',
3189		- 'Accept-Language' => 'en-US,en;q=0.9',
3190		- ),
3191		- ));
3192		-
3193		- if (is_wp_error($response)) {
3194		- return 0;
3195		- }
3196		-
3197		- $body = wp_remote_retrieve_body($response);
3198		- if (empty($body)) {
3199		- return 0;
3200		- }
3201		-
3202		- // Count <url> or <loc> elements
3203		- $count = preg_match_all('/<url>/i', $body, $matches);
3204		- return $count ?: 0;
3205		-}
3206		-
3207		-/**
3208		- * Get sitemaps declared in robots.txt
3209		- */
3210		-private function get_sitemaps_from_robots($site_url) {
3211		- $sitemaps = array();
3212		- $robots_url = trailingslashit($site_url) . 'robots.txt';
3213		-
3214		- $response = wp_remote_get($robots_url, array(
3215		- 'timeout' => 15,
3216		- 'sslverify' => false,
3217		- 'user-agent' => 'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/131.0.0.0 Safari/537.36',
3218		- ));
3219		-
3220		- if (is_wp_error($response)) {
3221		- return $sitemaps;
3222		- }
3223		-
3224		- $body = wp_remote_retrieve_body($response);
3225		- if (empty($body)) {
3226		- return $sitemaps;
3227		- }
3228		-
3229		- // Find Sitemap: declarations
3230		- if (preg_match_all('/^Sitemap:\s*(.+)$/mi', $body, $matches)) {
3231		- foreach ($matches[1] as $sitemap_url) {
3232		- $sitemap_url = trim($sitemap_url);
3233		- if (filter_var($sitemap_url, FILTER_VALIDATE_URL)) {
3234		- $sitemaps[] = $sitemap_url;
3235		- }
3236		- }
3237		- }
3238		-
3239		- return $sitemaps;
3240		-}
3241		-
3242		-public function mxchat_stop_processing() {
3243		- // Verify permissions
3244		- if (!current_user_can('manage_options')) {
3245		- wp_die(esc_html__('Unauthorized access', 'mxchat'));
3246		- }
3247		-
3248		- // Verify nonce
3249		- check_admin_referer('mxchat_stop_processing_action', 'mxchat_stop_processing_nonce');
3250		-
3251		- global $wpdb;
3252		- $table_name = $wpdb->prefix . 'mxchat_processing_queue';
3253		-
3254		- // Get active queue IDs
3255		- $sitemap_queue_id = get_transient('mxchat_active_queue_sitemap');
3256		- $pdf_queue_id = get_transient('mxchat_active_queue_pdf');
3257		-
3258		- // Delete all pending items from active queues
3259		- if ($sitemap_queue_id) {
3260		- $wpdb->delete(
3261		- $table_name,
3262		- array(
3263		- 'queue_id' => $sitemap_queue_id,
3264		- 'status' => 'pending'
3265		- ),
3266		- array('%s', '%s')
3267		- );
3268		-
3269		- delete_transient('mxchat_active_queue_sitemap');
3270		- delete_transient('mxchat_last_sitemap_url');
3271		- }
3272		-
3273		- if ($pdf_queue_id) {
3274		- // Get PDF path before deleting
3275		- $pdf_path = $this->mxchat_get_queue_meta($pdf_queue_id, 'pdf_path');
3276		-
3277		- $wpdb->delete(
3278		- $table_name,
3279		- array(
3280		- 'queue_id' => $pdf_queue_id,
3281		- 'status' => 'pending'
3282		- ),
3283		- array('%s', '%s')
3284		- );
3285		-
3286		- // Delete PDF file
3287		- if ($pdf_path && file_exists($pdf_path)) {
3288		- wp_delete_file($pdf_path);
3289		- }
3290		-
3291		- delete_transient('mxchat_active_queue_pdf');
3292		- delete_transient('mxchat_last_pdf_url');
3293		- }
3294		-
3295		- // Redirect back with a success message
3296		- set_transient('mxchat_admin_notice_success',
3297		- esc_html__('Processing has been stopped successfully.', 'mxchat'),
3298		- 30
3299		- );
3300		- wp_safe_redirect(admin_url('admin.php?page=mxchat-prompts'));
3301		- exit;
3302		-}
3303		-
3304		-/**
3305		- * Get content list for processing
3306		- */
3307		-public function ajax_mxchat_get_content_list() {
3308		- // Verify the nonce
3309		- check_ajax_referer('mxchat_content_selector_nonce', 'nonce');
3310		-
3311		- if (!current_user_can('manage_options')) {
3312		- wp_send_json_error(__('Unauthorized access', 'mxchat'));
3313		- }
3314		-
3315		- $page = isset($_GET['page']) ? absint($_GET['page']) : 1;
3316		- $per_page = isset($_GET['per_page']) ? absint($_GET['per_page']) : 100;
3317		- $search = isset($_GET['search']) ? sanitize_text_field($_GET['search']) : '';
3318		- $post_type = isset($_GET['post_type']) ? sanitize_text_field($_GET['post_type']) : 'all';
3319		- $post_status = isset($_GET['post_status']) ? sanitize_text_field($_GET['post_status']) : 'publish';
3320		- $processed_filter = isset($_GET['processed_filter']) ? sanitize_text_field($_GET['processed_filter']) : 'all';
3321		-
3322		- // Build query args
3323		- $args = array(
3324		- 'posts_per_page' => $per_page,
3325		- 'paged' => $page,
3326		- 'post_status' => $post_status !== 'all' ? $post_status : array('publish', 'draft', 'pending'),
3327		- 'orderby' => 'date',
3328		- 'order' => 'DESC',
3329		- );
3330		-
3331		- // Handle post types - IMPROVED VERSION
3332		- if ($post_type !== 'all') {
3333		- $args['post_type'] = $post_type;
3334		- } else {
3335		- // Get all available post types that might contain content
3336		- $all_post_types = array();
3337		-
3338		- // First get all public post types
3339		- $public_types = get_post_types(array('public' => true), 'names');
3340		- $all_post_types = array_merge($all_post_types, $public_types);
3341		-
3342		- // Add common forum/community post types
3343		- $forum_types = array('topic', 'reply', 'forum', 'wpforo_topic', 'wpforo_post');
3344		- foreach ($forum_types as $forum_type) {
3345		- if (post_type_exists($forum_type)) {
3346		- $all_post_types[] = $forum_type;
3347		- }
3348		- }
3349		-
3350		- // Add other commonly used post types
3351		- $common_types = array('product', 'job_listing', 'event', 'portfolio');
3352		- foreach ($common_types as $common_type) {
3353		- if (post_type_exists($common_type)) {
3354		- $all_post_types[] = $common_type;
3355		- }
3356		- }
3357		-
3358		- // Remove duplicates and ensure we have at least some post types
3359		- $all_post_types = array_unique($all_post_types);
3360		-
3361		- if (empty($all_post_types)) {
3362		- // Fallback to basic post types
3363		- $all_post_types = array('post', 'page');
3364		- }
3365		-
3366		- $args['post_type'] = $all_post_types;
3367		-
3368		- // Debug logging to see what post types are being queried
3369		- //error_log('MxChat Debug: Querying post types: ' . implode(', ', $all_post_types));
3370		- }
3371		-
3372		- if (!empty($search)) {
3373		- $args['s'] = $search;
3374		- }
3375		-
3376		- // Get processed data from storage
3377		- $processed_data = array();
3378		-
3379		- $pinecone_options = get_option('mxchat_pinecone_addon_options', array());
3380		- $use_pinecone = ($pinecone_options['mxchat_use_pinecone'] ?? '0') === '1';
3381		-
3382		- if ($use_pinecone && !empty($pinecone_options['mxchat_pinecone_api_key'])) {
3383		- // Get fresh data from Pinecone - no caching
3384		- $processed_data = $this->mxchat_get_pinecone_processed_content($pinecone_options);
3385		- } else {
3386		- // WordPress DB checking with better URL matching for all post types
3387		- global $wpdb;
3388		- $table_name = $wpdb->prefix . 'mxchat_system_prompt_content';
3389		- $processed_items = $wpdb->get_results("SELECT id, source_url, article_content, timestamp FROM {$table_name}");
3390		-
3391		- // Group items by source_url to count chunks
3392		- $url_chunk_counts = array();
3393		- $url_latest_timestamp = array();
3394		- $url_first_id = array();
3395		-
3396		- if (!empty($processed_items)) {
3397		- foreach ($processed_items as $item) {
3398		- $url = $item->source_url;
3399		- if (empty($url)) continue;
3400		-
3401		- // Count chunks per URL
3402		- if (!isset($url_chunk_counts[$url])) {
3403		- $url_chunk_counts[$url] = 0;
3404		- $url_latest_timestamp[$url] = $item->timestamp;
3405		- $url_first_id[$url] = $item->id;
3406		- }
3407		- $url_chunk_counts[$url]++;
3408		-
3409		- // Track latest timestamp
3410		- if (strtotime($item->timestamp) > strtotime($url_latest_timestamp[$url])) {
3411		- $url_latest_timestamp[$url] = $item->timestamp;
3412		- }
3413		- }
3414		-
3415		- // Now build processed_data with chunk counts
3416		- foreach ($url_chunk_counts as $url => $chunk_count) {
3417		- $post_id = $this->mxchat_url_to_post_id_improved($url);
3418		-
3419		- if ($post_id) {
3420		- $processed_data[$post_id] = array(
3421		- 'db_id' => $url_first_id[$url],
3422		- 'timestamp' => $url_latest_timestamp[$url],
3423		- 'url' => $url,
3424		- 'source' => 'wordpress',
3425		- 'chunk_count' => $chunk_count
3426		- );
3427		- }
3428		- }
3429		- }
3430		- }
3431		-
3432		- // Get processed IDs as a simple array for in_array checks
3433		- $processed_ids = array_keys($processed_data);
3434		-
3435		- // Handle processed/unprocessed filter
3436		- if ($processed_filter === 'processed' && !empty($processed_ids)) {
3437		- $args['post__in'] = $processed_ids;
3438		- } elseif ($processed_filter === 'unprocessed' && !empty($processed_ids)) {
3439		- $args['post__not_in'] = $processed_ids;
3440		- }
3441		-
3442		- // Run the query
3443		- $query = new WP_Query($args);
3444		- $content_items = array();
3445		-
3446		- if ($query->have_posts()) {
3447		- while ($query->have_posts()) {
3448		- $query->the_post();
3449		- $id = get_the_ID();
3450		- $post_date = get_the_date();
3451		- $excerpt = wp_trim_words(get_the_excerpt(), 20, '...');
3452		- $word_count = str_word_count(strip_tags(get_the_content()));
3453		-
3454		- $is_processed = in_array($id, $processed_ids);
3455		- $processed_date = '';
3456		- $db_record_id = 0;
3457		- $data_source = 'none';
3458		-
3459		- if ($is_processed && isset($processed_data[$id])) {
3460		- $item_data = $processed_data[$id];
3461		- $data_source = $item_data['source'];
3462		-
3463		- if ($data_source === 'wordpress' && isset($item_data['timestamp'])) {
3464		- // WordPress DB format
3465		- $timestamp = strtotime($item_data['timestamp']);
3466		- $processed_date = human_time_diff($timestamp, current_time('timestamp')) . ' ago';
3467		- $db_record_id = $item_data['db_id'];
3468		- } elseif ($data_source === 'pinecone') {
3469		- // Pinecone format
3470		- $processed_date = $item_data['processed_date'];
3471		- $db_record_id = $item_data['db_id'];
3472		- }
3473		- }
3474		-
3475		- // Get chunk count for this item
3476		- $chunk_count = 0;
3477		- if ($is_processed && isset($processed_data[$id]['chunk_count'])) {
3478		- $chunk_count = intval($processed_data[$id]['chunk_count']);
3479		- }
3480		-
3481		- $content_items[] = array(
3482		- 'id' => $id,
3483		- 'title' => get_the_title(),
3484		- 'permalink' => get_permalink(),
3485		- 'date' => $post_date,
3486		- 'type' => get_post_type(),
3487		- 'status' => get_post_status(),
3488		- 'excerpt' => $excerpt,
3489		- 'word_count' => $word_count,
3490		- 'already_processed' => $is_processed,
3491		- 'processed_date' => $processed_date,
3492		- 'db_record_id' => $db_record_id,
3493		- 'data_source' => $data_source,
3494		- 'chunk_count' => $chunk_count
3495		- );
3496		- }
3497		- wp_reset_postdata();
3498		- }
3499		-
3500		- $response = array(
3501		- 'items' => $content_items,
3502		- 'total' => $query->found_posts,
3503		- 'total_pages' => $query->max_num_pages,
3504		- 'current_page' => $page,
3505		- 'processed_count' => count($processed_ids)
3506		- );
3507		-
3508		- wp_send_json_success($response);
3509		- exit;
3510		-}
3511		-
3512		-
3513		-/**
3514		- * This function handles various WooCommerce URL formats and permalink structures
3515		- */
3516		-private function mxchat_url_to_post_id_improved($url) {
3517		- // First try the standard WordPress function
3518		- $post_id = url_to_postid($url);
3519		-
3520		- if ($post_id > 0) {
3521		- return $post_id;
3522		- }
3523		-
3524		- // If that fails, try more aggressive URL matching
3525		- // Remove trailing slashes and query parameters for better matching
3526		- $clean_url = rtrim($url, '/');
3527		- $clean_url = strtok($clean_url, '?'); // Remove query parameters
3528		-
3529		- // Try again with cleaned URL
3530		- $post_id = url_to_postid($clean_url);
3531		- if ($post_id > 0) {
3532		- return $post_id;
3533		- }
3534		-
3535		- // For bbPress forum topics, try extracting slug from URL
3536		- if (strpos($url, '/topic/') !== false || strpos($url, '/forums/') !== false) {
3537		- // Handle bbPress URLs: /forums/topic/topic-name/
3538		- if (preg_match('/\/forums\/topic\/([^\/\?]+)/', $url, $matches)) {
3539		- $topic_slug = $matches[1];
3540		-
3541		- // Look up topic by slug
3542		- $topic = get_page_by_path($topic_slug, OBJECT, 'topic');
3543		- if ($topic) {
3544		- return $topic->ID;
3545		- }
3546		-
3547		- // Alternative method: query by post_name
3548		- global $wpdb;
3549		- $post_id = $wpdb->get_var($wpdb->prepare(
3550		- "SELECT ID FROM {$wpdb->posts} WHERE post_name = %s AND post_type = 'topic' AND post_status IN ('publish', 'closed')",
3551		- $topic_slug
3552		- ));
3553		-
3554		- if ($post_id) {
3555		- return intval($post_id);
3556		- }
3557		- }
3558		-
3559		- // Handle simpler topic URLs: /topic/topic-name/
3560		- if (preg_match('/\/topic\/([^\/\?]+)/', $url, $matches)) {
3561		- $topic_slug = $matches[1];
3562		-
3563		- global $wpdb;
3564		- $post_id = $wpdb->get_var($wpdb->prepare(
3565		- "SELECT ID FROM {$wpdb->posts} WHERE post_name = %s AND post_type = 'topic' AND post_status IN ('publish', 'closed')",
3566		- $topic_slug
3567		- ));
3568		-
3569		- if ($post_id) {
3570		- return intval($post_id);
3571		- }
3572		- }
3573		- }
3574		-
3575		- // For WooCommerce products
3576		- if (strpos($url, '/product/') !== false || strpos($url, 'product=') !== false) {
3577		- // Extract product slug from various URL formats
3578		- $product_slug = '';
3579		-
3580		- // Handle pretty permalinks: /product/product-name/
3581		- if (preg_match('/\/product\/([^\/\?]+)/', $url, $matches)) {
3582		- $product_slug = $matches[1];
3583		- }
3584		- // Handle query parameters: ?product=product-name
3585		- elseif (preg_match('/[\?&]product=([^&]+)/', $url, $matches)) {
3586		- $product_slug = $matches[1];
3587		- }
3588		-
3589		- if (!empty($product_slug)) {
3590		- // Look up product by slug
3591		- $product = get_page_by_path($product_slug, OBJECT, 'product');
3592		- if ($product) {
3593		- return $product->ID;
3594		- }
3595		-
3596		- // Alternative method: query by post_name
3597		- global $wpdb;
3598		- $post_id = $wpdb->get_var($wpdb->prepare(
3599		- "SELECT ID FROM {$wpdb->posts} WHERE post_name = %s AND post_type = 'product' AND post_status = 'publish'",
3600		- $product_slug
3601		- ));
3602		-
3603		- if ($post_id) {
3604		- return intval($post_id);
3605		- }
3606		- }
3607		- }
3608		-
3609		- // Generic approach: try to extract slug and match against all post types
3610		- $parsed_url = wp_parse_url($clean_url);
3611		- $path = $parsed_url['path'] ?? '';
3612		-
3613		- if (!empty($path)) {
3614		- // Get the last part of the path as potential slug
3615		- $path_parts = array_filter(explode('/', trim($path, '/')));
3616		- $potential_slug = end($path_parts);
3617		-
3618		- if (!empty($potential_slug)) {
3619		- global $wpdb;
3620		-
3621		- // Try to find any post with this slug
3622		- $post_id = $wpdb->get_var($wpdb->prepare(
3623		- "SELECT ID FROM {$wpdb->posts}
3624		- WHERE post_name = %s
3625		- AND post_status IN ('publish', 'closed', 'private')
3626		- AND post_type NOT IN ('revision', 'attachment', 'nav_menu_item')
3627		- ORDER BY CASE
3628		- WHEN post_type = 'post' THEN 1
3629		- WHEN post_type = 'page' THEN 2
3630		- WHEN post_type = 'topic' THEN 3
3631		- WHEN post_type = 'product' THEN 4
3632		- ELSE 5
3633		- END
3634		- LIMIT 1",
3635		- $potential_slug
3636		- ));
3637		-
3638		- if ($post_id) {
3639		- return intval($post_id);
3640		- }
3641		- }
3642		- }
3643		-
3644		- // ADDITIONAL: Try direct database lookup by URL variations
3645		- global $wpdb;
3646		- $table_name = $wpdb->prefix . 'mxchat_system_prompt_content';
3647		-
3648		- // Try variations of the URL (with/without trailing slash, http/https)
3649		- $url_variations = array(
3650		- $url,
3651		- rtrim($url, '/'),
3652		- $url . '/',
3653		- str_replace('http://', 'https://', $url),
3654		- str_replace('https://', 'http://', $url),
3655		- str_replace('http://', 'https://', rtrim($url, '/')),
3656		- str_replace('https://', 'http://', rtrim($url, '/'))
3657		- );
3658		-
3659		- // Remove duplicates
3660		- $url_variations = array_unique($url_variations);
3661		-
3662		- foreach ($url_variations as $variation) {
3663		- $existing_record = $wpdb->get_row($wpdb->prepare(
3664		- "SELECT id, source_url FROM $table_name WHERE source_url = %s",
3665		- $variation
3666		- ));
3667		-
3668		- if ($existing_record) {
3669		- // Try to get post ID from this stored URL
3670		- $stored_post_id = url_to_postid($existing_record->source_url);
3671		- if ($stored_post_id > 0) {
3672		- return $stored_post_id;
3673		- }
3674		- }
3675		- }
3676		-
3677		- return 0; // No match found
3678		-}
3679		-/**
3680		- * Process selected content via AJAX
3681		- */
3682		-public function ajax_mxchat_process_selected_content() {
3683		- // Basic request validation
3684		- if (!check_ajax_referer('mxchat_content_selector_nonce', 'nonce', false)) {
3685		- wp_send_json_error('Invalid nonce');
3686		- exit;
3687		- }
3688		-
3689		- if (!current_user_can('manage_options')) {
3690		- wp_send_json_error('Unauthorized access');
3691		- exit;
3692		- }
3693		-
3694		- // Get post IDs - safely parse the array
3695		- $post_ids = array();
3696		- if (isset($_POST['post_ids']) && is_array($_POST['post_ids'])) {
3697		- foreach ($_POST['post_ids'] as $id) {
3698		- $post_ids[] = absint($id);
3699		- }
3700		- }
3701		-
3702		- if (empty($post_ids)) {
3703		- wp_send_json_error('No content selected');
3704		- exit;
3705		- }
3706		-
3707		- // Get bot_id from request
3708		- $bot_id = isset($_POST['bot_id']) ? sanitize_key($_POST['bot_id']) : 'default';
3709		-
3710		- // ACF→PDF extraction is opt-in per import batch. Persist the last-used value so users
3711		- // don't re-check on every batch; the default is OFF for installs that haven't set it.
3712		- $extract_acf_pdfs = !empty($_POST['extract_acf_pdfs']) && $_POST['extract_acf_pdfs'] !== 'false';
3713		- $mxchat_options = get_option('mxchat_options', array());
3714		- if (!is_array($mxchat_options)) {
3715		- $mxchat_options = array();
3716		- }
3717		- $prior_default = !empty($mxchat_options['acf_pdf_extract_default']);
3718		- if ($prior_default !== $extract_acf_pdfs) {
3719		- $mxchat_options['acf_pdf_extract_default'] = $extract_acf_pdfs ? 1 : 0;
3720		- update_option('mxchat_options', $mxchat_options);
3721		- }
3722		-
3723		- // Process only ONE post at a time to avoid request size issues
3724		- $post_id = reset($post_ids);
3725		- $post = get_post($post_id);
3726		-
3727		- if (!$post) {
3728		- wp_send_json_error('Post not found');
3729		- exit;
3730		- }
3731		-
3732		- // Allow developers to modify post data before processing into knowledge base
3733		- $post = apply_filters('mxchat_before_process_post', $post, $bot_id);
3734		-
3735		- // Get content including title, short description (for WooCommerce), and main content
3736		- $content = $post->post_title . "\n\n";
3737		-
3738		- // Add short description if it exists (WooCommerce products use post_excerpt for short description)
3739		- if (!empty($post->post_excerpt)) {
3740		- // Remove shortcode tags but preserve content inside them
3741		- $clean_excerpt = $this->strip_shortcode_tags_preserve_content($post->post_excerpt);
3742		- $content .= "Short Description: " . wp_strip_all_tags($clean_excerpt) . "\n\n";
3743		- }
3744		-
3745		- // Add main content - remove shortcode tags but preserve content inside them
3746		- $clean_content = $this->strip_shortcode_tags_preserve_content($post->post_content);
3747		- $content .= wp_strip_all_tags($clean_content);
3748		-
3749		- // ADD WOOCOMMERCE PRODUCT DATA (pricing, stock, categories, custom tabs)
3750		- if (get_post_type($post_id) === 'product' && class_exists('WooCommerce')) {
3751		- $product = wc_get_product($post_id);
3752		-
3753		- if ($product) {
3754		- // Get pricing information
3755		- $regular_price = $product->get_regular_price();
3756		- $sale_price = $product->get_sale_price();
3757		- $price = $product->get_price();
3758		- $sku = $product->get_sku();
3759		-
3760		- // Get currency symbol
3761		- $currency_symbol = get_woocommerce_currency_symbol();
3762		-
3763		- // Add pricing information
3764		- $content .= "\n";
3765		- if (!empty($regular_price)) {
3766		- $content .= "Price: " . $currency_symbol . $regular_price . "\n";
3767		- } elseif (!empty($price)) {
3768		- $content .= "Price: " . $currency_symbol . $price . "\n";
3769		- }
3770		-
3771		- if (!empty($sale_price) && $sale_price !== $regular_price) {
3772		- $content .= "Sale Price: " . $currency_symbol . $sale_price . "\n";
3773		- }
3774		-
3775		- // Handle variable products - show price range
3776		- if ($product->is_type('variable')) {
3777		- $min_price = $product->get_variation_price('min');
3778		- $max_price = $product->get_variation_price('max');
3779		- if ($min_price !== $max_price) {
3780		- $content .= "Price Range: " . $currency_symbol . $min_price . " - " . $currency_symbol . $max_price . "\n";
3781		- }
3782		- }
3783		-
3784		- if (!empty($sku)) {
3785		- $content .= "SKU: " . $sku . "\n";
3786		- }
3787		-
3788		- // Get product categories
3789		- $categories = wp_get_post_terms($post_id, 'product_cat', array('fields' => 'names'));
3790		- if (!empty($categories) && !is_wp_error($categories)) {
3791		- $content .= "Categories: " . implode(', ', $categories) . "\n";
3792		- }
3793		- }
3794		-
3795		- // Get Custom Product Tabs (supports "Custom Product Tabs for WooCommerce" by Code Parrots)
3796		- $custom_tabs = get_post_meta($post_id, 'yikes_woo_products_tabs', true);
3797		- if (!empty($custom_tabs) && is_array($custom_tabs)) {
3798		- foreach ($custom_tabs as $tab) {
3799		- $tab_title = isset($tab['title']) ? $tab['title'] : (isset($tab['tab_title']) ? $tab['tab_title'] : '');
3800		- $tab_content = isset($tab['content']) ? $tab['content'] : '';
3801		-
3802		- if (!empty($tab_title) && !empty($tab_content)) {
3803		- $content .= "\n" . $tab_title . ": " . wp_strip_all_tags($tab_content) . "\n";
3804		- }
3805		- }
3806		- }
3807		-
3808		- // Also check for reusable/saved tabs applied to this product
3809		- $applied_saved_tabs = get_post_meta($post_id, 'yikes_woo_reusable_products_tabs_applied', true);
3810		- if (!empty($applied_saved_tabs) && is_array($applied_saved_tabs)) {
3811		- $saved_tabs = get_option('yikes_woo_reusable_products_tabs', array());
3812		- if (!empty($saved_tabs) && is_array($saved_tabs)) {
3813		- foreach ($applied_saved_tabs as $saved_tab_id) {
3814		- if (isset($saved_tabs[$saved_tab_id])) {
3815		- $tab = $saved_tabs[$saved_tab_id];
3816		- $tab_title = isset($tab['title']) ? $tab['title'] : (isset($tab['tab_title']) ? $tab['tab_title'] : '');
3817		- $tab_content = isset($tab['content']) ? $tab['content'] : '';
3818		-
3819		- if (!empty($tab_title) && !empty($tab_content)) {
3820		- $content .= "\n" . $tab_title . ": " . wp_strip_all_tags($tab_content) . "\n";
3821		- }
3822		- }
3823		- }
3824		- }
3825		- }
3826		- }
3827		-
3828		- // ADD ACF FIELDS SUPPORT
3829		- $acf_fields = $this->mxchat_get_acf_fields_for_post($post_id);
3830		- $pdf_extracted_count = 0;
3831		- if (!empty($acf_fields)) {
3832		- $acf_content_parts = array();
3833		- $pdf_attachment_ids = array();
3834		-
3835		- foreach ($acf_fields as $field_name => $field_value) {
3836		- $formatted_value = $this->mxchat_format_acf_field_value($field_value, $field_name, $post_id);
3837		-
3838		- if (!empty($formatted_value)) {
3839		- $field_label = ucwords(str_replace('_', ' ', $field_name));
3840		- $acf_content_parts[] = $field_label . ": " . $formatted_value;
3841		- }
3842		-
3843		- // Walk this field's value tree for any PDF attachment references and queue them for extraction.
3844		- // Only when the user opted into ACF→PDF extraction for this batch; otherwise the ACF text
3845		- // still lands in the KB but the heavier PDF parsing is skipped.
3846		- if ($extract_acf_pdfs) {
3847		- $this->mxchat_collect_pdf_attachment_ids_from_acf_value($field_value, $pdf_attachment_ids);
3848		- }
3849		- }
3850		-
3851		- if (!empty($acf_content_parts)) {
3852		- $content .= "\n\n" . implode("\n", $acf_content_parts);
3853		- }
3854		-
3855		- // Extract text from each unique PDF found in ACF fields and append as a labeled section
3856		- if ($extract_acf_pdfs && !empty($pdf_attachment_ids)) {
3857		- $pdf_attachment_ids = array_unique(array_filter(array_map('intval', $pdf_attachment_ids)));
3858		- $pdf_sections = array();
3859		- foreach ($pdf_attachment_ids as $att_id) {
3860		- $pdf_text = $this->mxchat_extract_pdf_text_by_attachment_id($att_id);
3861		- if (!empty($pdf_text)) {
3862		- $pdf_title = get_the_title($att_id);
3863		- $pdf_url = wp_get_attachment_url($att_id);
3864		- $header = 'PDF Attachment';
3865		- if (!empty($pdf_title)) {
3866		- $header .= ': ' . $pdf_title;
3867		- }
3868		- if (!empty($pdf_url)) {
3869		- $header .= ' (' . $pdf_url . ')';
3870		- }
3871		- $pdf_sections[] = $header . "\n" . $pdf_text;
3872		- $pdf_extracted_count++;
3873		- }
3874		- }
3875		- if (!empty($pdf_sections)) {
3876		- $content .= "\n\nAttached PDFs:\n" . implode("\n\n", $pdf_sections);
3877		- }
3878		- }
3879		- }
3880		-
3881		- // ADD CUSTOM POST META SUPPORT (whitelisted non-ACF meta fields)
3882		- $custom_meta = $this->mxchat_get_whitelisted_post_meta($post_id);
3883		- if (!empty($custom_meta)) {
3884		- $meta_content_parts = array();
3885		-
3886		- foreach ($custom_meta as $meta_key => $meta_value) {
3887		- // Convert meta key to readable label
3888		- $meta_label = ucwords(str_replace(array('_', '-'), ' ', $meta_key));
3889		- $meta_content_parts[] = $meta_label . ": " . $meta_value;
3890		- }
3891		-
3892		- if (!empty($meta_content_parts)) {
3893		- $content .= "\n\n" . implode("\n", $meta_content_parts);
3894		- }
3895		- }
3896		-
3897		- // Debug logging for WordPress Import content
3898		- //error_log('[MXCHAT-WP-IMPORT-DEBUG] Post ID: ' . $post_id . ' Title: ' . $post->post_title);
3899		- //error_log('[MXCHAT-WP-IMPORT-DEBUG] Raw post_content length: ' . strlen($post->post_content));
3900		- //error_log('[MXCHAT-WP-IMPORT-DEBUG] Final content length: ' . strlen($content));
3901		- //error_log('[MXCHAT-WP-IMPORT-DEBUG] Content preview: ' . substr($content, 0, 300));
3902		-
3903		- // Note: Removed 10,000 char limit - chunking now handles large content properly
3904		-
3905		- // Get bot-specific API key
3906		- $bot_options = $this->get_bot_options($bot_id);
3907		- $options = !empty($bot_options) ? $bot_options : get_option('mxchat_options');
3908		- $selected_model = $options['embedding_model'] ?? 'text-embedding-ada-002';
3909		-
3910		- if (strpos($selected_model, 'voyage') === 0) {
3911		- $api_key = $options['voyage_api_key'] ?? '';
3912		- $provider_name = 'Voyage AI';
3913		- } elseif (strpos($selected_model, 'gemini-embedding') === 0) {
3914		- $api_key = $options['gemini_api_key'] ?? '';
3915		- $provider_name = 'Google Gemini';
3916		- } else {
3917		- $api_key = $options['api_key'] ?? '';
3918		- $provider_name = 'OpenAI';
3919		- }
3920		-
3921		- if (empty($api_key)) {
3922		- MxChat_Admin::mxchat_log_debug('api_error', $provider_name . ' API key not configured for knowledge processing');
3923		- wp_send_json_error($provider_name . ' API key not configured');
3924		- exit;
3925		- }
3926		-
3927		- $source_url = get_permalink($post_id);
3928		- $vector_id = md5($source_url); // Vector ID for Pinecone
3929		-
3930		- // Check for existing content in bot-specific storage
3931		- $is_update = false;
3932		-
3933		- // Get bot-specific Pinecone configuration
3934		- $bot_pinecone_config = $this->get_bot_pinecone_config($bot_id);
3935		- $use_pinecone = !empty($bot_pinecone_config) && ($bot_pinecone_config['use_pinecone'] ?? false);
3936		-
3937		- if ($use_pinecone && !empty($bot_pinecone_config['api_key'])) {
3938		- // Check Pinecone for this bot
3939		- $pinecone_data = $this->mxchat_get_pinecone_processed_content($bot_pinecone_config);
3940		- if (isset($pinecone_data[$post_id])) {
3941		- $is_update = true;
3942		- }
3943		- } else {
3944		- // Check WordPress DB (same as before since it's shared)
3945		- global $wpdb;
3946		- $table_name = $wpdb->prefix . 'mxchat_system_prompt_content';
3947		- $existing_record = $wpdb->get_row($wpdb->prepare(
3948		- "SELECT id FROM $table_name WHERE source_url = %s",
3949		- $source_url
3950		- ));
3951		-
3952		- if ($existing_record) {
3953		- $is_update = true;
3954		- }
3955		- }
3956		-
3957		- // UPDATED 2.5.6: Determine content type based on post_type
3958		- $post_type = $post->post_type;
3959		- $content_type = 'content'; // Default fallback
3960		-
3961		- // Map WordPress post types to content types
3962		- switch ($post_type) {
3963		- case 'post':
3964		- $content_type = 'post';
3965		- break;
3966		- case 'page':
3967		- $content_type = 'page';
3968		- break;
3969		- case 'product':
3970		- $content_type = 'product';
3971		- break;
3972		- default:
3973		- // For custom post types, use the post type name
3974		- $content_type = sanitize_key($post_type);
3975		- break;
3976		- }
3977		-
3978		- // Use the centralized utility function with bot_id and content_type
3979		- $result = MxChat_Utils::submit_content_to_db(
3980		- $content,
3981		- $source_url,
3982		- $api_key,
3983		- $vector_id,
3984		- $bot_id,
3985		- $content_type
3986		- );
3987		-
3988		- if (is_wp_error($result)) {
3989		- MxChat_Admin::mxchat_log_debug('storage_error', 'Knowledge storage failed: ' . $result->get_error_message(), array('source_url' => $source_url));
3990		- wp_send_json_error('Storage failed: ' . $result->get_error_message());
3991		- exit;
3992		- }
3993		-
3994		- // Automatically apply role restriction based on tags
3995		- $this->apply_role_restriction_to_post($post_id, $source_url);
3996		-
3997		- $operation_type = $is_update ? 'update' : 'new';
3998		-
3999		- // Count ACF fields for debugging
4000		- $acf_field_count = count($acf_fields);
4001		-
4002		- // Success response with minimal data
4003		- wp_send_json_success(array(
4004		- 'message' => $operation_type === 'update' ? 'Content updated successfully' : 'Content processed successfully',
4005		- 'post_id' => $post_id,
4006		- 'title' => $post->post_title,
4007		- 'operation_type' => $operation_type,
4008		- 'vector_id' => $vector_id,
4009		- 'acf_fields_found' => $acf_field_count,
4010		- 'pdf_extracted_count' => (int) $pdf_extracted_count,
4011		- 'content_preview' => substr($content, 0, 100) . '...',
4012		- 'bot_id' => $bot_id
4013		- ));
4014		- exit;
4015		-}
4016		-
4017		-private function apply_role_restriction_to_post($post_id, $source_url) {
4018		- // Get tag-role mappings
4019		- $mappings = get_option('mxchat_tag_role_mappings', array());
4020		-
4021		- if (empty($mappings)) {
4022		- return; // No mappings, leave as public
4023		- }
4024		-
4025		- // Get all tags for the post
4026		- $post_tags = wp_get_post_tags($post_id, array('fields' => 'slugs'));
4027		-
4028		- if (empty($post_tags)) {
4029		- return; // No tags, leave as public
4030		- }
4031		-
4032		- // Determine the highest role restriction based on tags
4033		- $highest_role = 'public';
4034		- $role_hierarchy = array(
4035		- 'public' => 0,
4036		- 'logged_in' => 1,
4037		- 'subscriber' => 2,
4038		- 'contributor' => 3,
4039		- 'author' => 4,
4040		- 'editor' => 5,
4041		- 'administrator' => 6
4042		- );
4043		-
4044		- foreach ($post_tags as $tag_slug) {
4045		- if (isset($mappings[$tag_slug])) {
4046		- $role = $mappings[$tag_slug];
4047		- if (isset($role_hierarchy[$role]) && $role_hierarchy[$role] > $role_hierarchy[$highest_role]) {
4048		- $highest_role = $role;
4049		- }
4050		- }
4051		- }
4052		-
4053		- // If no restricted tags found, return (leave as public)
4054		- if ($highest_role === 'public') {
4055		- return;
4056		- }
4057		-
4058		- // Update the role restriction in the database
4059		- global $wpdb;
4060		-
4061		- // Check if using Pinecone
4062		- $pinecone_options = get_option('mxchat_pinecone_addon_options', array());
4063		- $use_pinecone = ($pinecone_options['mxchat_use_pinecone'] ?? '0') === '1';
4064		-
4065		- if ($use_pinecone && !empty($pinecone_options['mxchat_pinecone_api_key'])) {
4066		- // Update Pinecone role restriction
4067		- $roles_table = $wpdb->prefix . 'mxchat_pinecone_roles';
4068		- $vector_id = md5($source_url);
4069		-
4070		- $wpdb->replace(
4071		- $roles_table,
4072		- array(
4073		- 'vector_id' => $vector_id,
4074		- 'role_restriction' => $highest_role,
4075		- 'updated_at' => current_time('mysql')
4076		- ),
4077		- array('%s', '%s', '%s')
4078		- );
4079		- } else {
4080		- // Update WordPress DB
4081		- $table_name = $wpdb->prefix . 'mxchat_system_prompt_content';
4082		-
4083		- $wpdb->update(
4084		- $table_name,
4085		- array('role_restriction' => $highest_role),
4086		- array('source_url' => $source_url),
4087		- array('%s'),
4088		- array('%s')
4089		- );
4090		- }
4091		-}
4092		-
4093		-public function mxchat_get_public_post_types() {
4094		- // Get all public post types
4095		- $post_types = get_post_types(array('public' => true), 'objects');
4096		- $post_type_options = array();
4097		-
4098		- foreach ($post_types as $post_type) {
4099		- $post_type_options[$post_type->name] = $post_type->label;
4100		- }
4101		-
4102		- // Also include common forum/community post types that might not be marked as public
4103		- $additional_types = array(
4104		- 'topic' => 'Forum Topics (bbPress)',
4105		- 'reply' => 'Forum Replies (bbPress)',
4106		- 'forum' => 'Forums (bbPress)',
4107		- 'wpforo_topic' => 'wpForo Topics',
4108		- 'wpforo_post' => 'wpForo Posts'
4109		- );
4110		-
4111		- foreach ($additional_types as $type_name => $type_label) {
4112		- if (post_type_exists($type_name) && !isset($post_type_options[$type_name])) {
4113		- $post_type_options[$type_name] = $type_label;
4114		- }
4115		- }
4116		-
4117		- return $post_type_options;
4118		-}
4119		-
4120		-/**
4121		- * Retrieves processed content from Pinecone API
4122		- */
4123		-public function mxchat_get_pinecone_processed_content($pinecone_options) {
4124		- $api_key = $pinecone_options['mxchat_pinecone_api_key'] ?? '';
4125		- $host = $pinecone_options['mxchat_pinecone_host'] ?? '';
4126		-
4127		- if (empty($api_key) || empty($host)) {
4128		- return array();
4129		- }
4130		-
4131		- $pinecone_data = array();
4132		-
4133		- try {
4134		- // Always get fresh data from Pinecone
4135		- $pinecone_data = $this->mxchat_scan_pinecone_for_processed_content($pinecone_options);
4136		-
4137		- // Method 2: Final fallback - try stats endpoint (if available)
4138		- if (empty($pinecone_data)) {
4139		- $stats_url = "https://{$host}/describe_index_stats";
4140		-
4141		- $response = wp_remote_post($stats_url, array(
4142		- 'headers' => array(
4143		- 'Api-Key' => $api_key,
4144		- 'Content-Type' => 'application/json'
4145		- ),
4146		- 'body' => json_encode(array()),
4147		- 'timeout' => 30
4148		- ));
4149		-
4150		- if (!is_wp_error($response) && wp_remote_retrieve_response_code($response) === 200) {
4151		- $body = wp_remote_retrieve_body($response);
4152		- $stats_data = json_decode($body, true);
4153		- }
4154		- }
4155		-
4156		- } catch (Exception $e) {
4157		- // Log error but return fresh data only
4158		- }
4159		-
4160		- return $pinecone_data;
4161		-}
4162		-public function mxchat_fetch_pinecone_vectors_by_ids($pinecone_options, $vector_ids) {
4163		- //error_log('=== DEBUG: Starting mxchat_fetch_pinecone_vectors_by_ids ===');
4164		-
4165		- $api_key = $pinecone_options['mxchat_pinecone_api_key'] ?? '';
4166		- $host = $pinecone_options['mxchat_pinecone_host'] ?? '';
4167		-
4168		- if (empty($api_key) || empty($host) || empty($vector_ids)) {
4169		- //error_log('DEBUG: Missing parameters for fetch by IDs');
4170		- return array();
4171		- }
4172		-
4173		- try {
4174		- $fetch_url = "https://{$host}/vectors/fetch";
4175		- //error_log('DEBUG: Fetch URL: ' . $fetch_url);
4176		- //error_log('DEBUG: Fetching ' . count($vector_ids) . ' vector IDs');
4177		-
4178		- // Pinecone fetch API allows fetching specific vectors by ID
4179		- $fetch_data = array(
4180		- 'ids' => array_values($vector_ids)
4181		- );
4182		-
4183		- $response = wp_remote_post($fetch_url, array(
4184		- 'headers' => array(
4185		- 'Api-Key' => $api_key,
4186		- 'Content-Type' => 'application/json'
4187		- ),
4188		- 'body' => json_encode($fetch_data),
4189		- 'timeout' => 30
4190		- ));
4191		-
4192		- if (is_wp_error($response)) {
4193		- //error_log('DEBUG: Fetch by IDs WP error: ' . $response->get_error_message());
4194		- return array();
4195		- }
4196		-
4197		- $response_code = wp_remote_retrieve_response_code($response);
4198		- //error_log('DEBUG: Fetch response code: ' . $response_code);
4199		-
4200		- if ($response_code !== 200) {
4201		- $error_body = wp_remote_retrieve_body($response);
4202		- //error_log('DEBUG: Fetch failed with body: ' . $error_body);
4203		- return array();
4204		- }
4205		-
4206		- $body = wp_remote_retrieve_body($response);
4207		- $data = json_decode($body, true);
4208		-
4209		- //error_log('DEBUG: Fetch response structure: ' . print_r(array_keys($data), true));
4210		-
4211		- if (!isset($data['vectors'])) {
4212		- //error_log('DEBUG: No vectors key in response');
4213		- return array();
4214		- }
4215		-
4216		- $processed_data = array();
4217		-
4218		- foreach ($data['vectors'] as $vector_id => $vector_data) {
4219		- $metadata = $vector_data['metadata'] ?? array();
4220		- $source_url = $metadata['source_url'] ?? '';
4221		-
4222		- if (!empty($source_url)) {
4223		- $post_id = url_to_postid($source_url);
4224		- if ($post_id) {
4225		- $created_at = $metadata['created_at'] ?? '';
4226		- $processed_date = 'Recently';
4227		-
4228		- if (!empty($created_at)) {
4229		- $timestamp = is_numeric($created_at) ? $created_at : strtotime($created_at);
4230		- if ($timestamp) {
4231		- $processed_date = human_time_diff($timestamp, current_time('timestamp')) . ' ago';
4232		- }
4233		- }
4234		-
4235		- $processed_data[$post_id] = array(
4236		- 'db_id' => $vector_id,
4237		- 'processed_date' => $processed_date,
4238		- 'url' => $source_url,
4239		- 'source' => 'pinecone',
4240		- 'timestamp' => $timestamp ?? current_time('timestamp')
4241		- );
4242		- }
4243		- }
4244		- }
4245		-
4246		- //error_log('DEBUG: Processed ' . count($processed_data) . ' vectors from fetch');
4247		- return $processed_data;
4248		-
4249		- } catch (Exception $e) {
4250		- //error_log('DEBUG: Exception in fetch_pinecone_vectors_by_ids: ' . $e->getMessage());
4251		- return array();
4252		- }
4253		-}
4254		-
4255		-/**
4256		- * Get embedding dimensions based on the selected model.
4257		- */
4258		-private function mxchat_get_embedding_dimensions() {
4259		- $options = get_option('mxchat_options', array());
4260		- $selected_model = $options['embedding_model'] ?? 'text-embedding-ada-002';
4261		-
4262		- $model_dimensions = array(
4263		- 'text-embedding-ada-002' => 1536,
4264		- 'text-embedding-3-small' => 1536,
4265		- 'text-embedding-3-large' => 3072,
4266		- 'voyage-2' => 1024,
4267		- 'voyage-large-2' => 1536,
4268		- 'voyage-3-large' => 2048,
4269		- 'gemini-embedding-001' => 1536,
4270		- );
4271		-
4272		- if (strpos($selected_model, 'voyage-3-large') === 0) {
4273		- $custom_dimensions = $options['voyage_output_dimension'] ?? 2048;
4274		- return intval($custom_dimensions);
4275		- }
4276		-
4277		- if (strpos($selected_model, 'gemini-embedding') === 0) {
4278		- $custom_dimensions = $options['gemini_output_dimension'] ?? 1536;
4279		- return intval($custom_dimensions);
4280		- }
4281		-
4282		- return $model_dimensions[$selected_model] ?? 1536;
4283		-}
4284		-
4285		-/**
4286		- * Scan Pinecone for processed content
4287		- */
4288		-public function mxchat_scan_pinecone_for_processed_content($pinecone_options) {
4289		- $api_key = $pinecone_options['mxchat_pinecone_api_key'] ?? '';
4290		- $host = $pinecone_options['mxchat_pinecone_host'] ?? '';
4291		-
4292		- if (empty($api_key) || empty($host)) {
4293		- return array();
4294		- }
4295		-
4296		- try {
4297		- // Use multiple random vectors to get better coverage
4298		- $all_matches = array();
4299		- $seen_ids = array();
4300		-
4301		- // Get correct dimensions for the configured embedding model
4302		- $dimensions = $this->mxchat_get_embedding_dimensions();
4303		-
4304		- // Try 3 different random vectors to get better coverage
4305		- for ($i = 0; $i < 3; $i++) {
4306		- $query_url = "https://{$host}/query";
4307		-
4308		- // Generate a random unit vector instead of zeros
4309		- $random_vector = array();
4310		- for ($j = 0; $j < $dimensions; $j++) {
4311		- $random_vector[] = (rand(-1000, 1000) / 1000.0);
4312		- }
4313		-
4314		- // Normalize the vector to unit length
4315		- $magnitude = sqrt(array_sum(array_map(function($x) { return $x * $x; }, $random_vector)));
4316		- if ($magnitude > 0) {
4317		- $random_vector = array_map(function($x) use ($magnitude) { return $x / $magnitude; }, $random_vector);
4318		- }
4319		-
4320		- $query_data = array(
4321		- 'includeMetadata' => true,
4322		- 'includeValues' => false,
4323		- 'topK' => 10000,
4324		- 'vector' => $random_vector
4325		- );
4326		-
4327		- $response = wp_remote_post($query_url, array(
4328		- 'headers' => array(
4329		- 'Api-Key' => $api_key,
4330		- 'Content-Type' => 'application/json'
4331		- ),
4332		- 'body' => json_encode($query_data),
4333		- 'timeout' => 30
4334		- ));
4335		-
4336		- if (is_wp_error($response)) {
4337		- continue;
4338		- }
4339		-
4340		- $response_code = wp_remote_retrieve_response_code($response);
4341		-
4342		- if ($response_code !== 200) {
4343		- continue;
4344		- }
4345		-
4346		- $body = wp_remote_retrieve_body($response);
4347		- $data = json_decode($body, true);
4348		-
4349		- if (isset($data['matches'])) {
4350		- foreach ($data['matches'] as $match) {
4351		- $match_id = $match['id'] ?? '';
4352		- if (!empty($match_id) && !isset($seen_ids[$match_id])) {
4353		- $all_matches[] = $match;
4354		- $seen_ids[$match_id] = true;
4355		- }
4356		- }
4357		- }
4358		- }
4359		-
4360		- // Convert matches to processed data format, grouping by URL to count chunks
4361		- $processed_data = array();
4362		- $url_chunk_counts = array();
4363		-
4364		- foreach ($all_matches as $match) {
4365		- $metadata = $match['metadata'] ?? array();
4366		- $source_url = $metadata['source_url'] ?? '';
4367		- $match_id = $match['id'] ?? '';
4368		-
4369		- if (!empty($source_url) && !empty($match_id)) {
4370		- $post_id = url_to_postid($source_url);
4371		- if ($post_id) {
4372		- // Count chunks per post_id
4373		- if (!isset($url_chunk_counts[$post_id])) {
4374		- $url_chunk_counts[$post_id] = 0;
4375		- }
4376		- $url_chunk_counts[$post_id]++;
4377		-
4378		- $created_at = $metadata['created_at'] ?? '';
4379		- $processed_date = 'Recently';
4380		-
4381		- if (!empty($created_at)) {
4382		- $timestamp = is_numeric($created_at) ? $created_at : strtotime($created_at);
4383		- if ($timestamp) {
4384		- $processed_date = human_time_diff($timestamp, current_time('timestamp')) . ' ago';
4385		- }
4386		- }
4387		-
4388		- // Only store if not already set, or update with newer timestamp
4389		- if (!isset($processed_data[$post_id]) ||
4390		- ($timestamp ?? 0) > ($processed_data[$post_id]['timestamp'] ?? 0)) {
4391		- $processed_data[$post_id] = array(
4392		- 'db_id' => $match_id,
4393		- 'processed_date' => $processed_date,
4394		- 'url' => $source_url,
4395		- 'source' => 'pinecone',
4396		- 'timestamp' => $timestamp ?? current_time('timestamp')
4397		- );
4398		- }
4399		- }
4400		- }
4401		- }
4402		-
4403		- // Add chunk counts to processed data
4404		- foreach ($url_chunk_counts as $post_id => $chunk_count) {
4405		- if (isset($processed_data[$post_id])) {
4406		- $processed_data[$post_id]['chunk_count'] = $chunk_count;
4407		- }
4408		- }
4409		-
4410		- return $processed_data;
4411		-
4412		- } catch (Exception $e) {
4413		- return array();
4414		- }
4415		-}
4416		-/**
4417		- * Generate embeddings from input text for MXChat with bot support
4418		- */
4419		-private function mxchat_generate_embedding($text, $bot_id = 'default') {
4420		- // Enable detailed logging for debugging
4421		- //error_log('[MXCHAT-EMBED] Starting embedding generation for bot: ' . $bot_id . '. Text length: ' . strlen($text) . ' bytes');
4422		- //error_log('[MXCHAT-EMBED] Text preview: ' . substr($text, 0, 100) . '...');
4423		-
4424		- // Get bot-specific options
4425		- $bot_options = $this->get_bot_options($bot_id);
4426		- $options = !empty($bot_options) ? $bot_options : get_option('mxchat_options');
4427		-
4428		- // Opt-in: when the custom provider is selected for embeddings, index through
4429		- // the same custom endpoint the query path uses so stored vectors and query
4430		- // vectors share a model. Returns the vector array on success, or an error
4431		- // string on failure (this function's existing failure contract).
4432		- if (isset($options['custom_provider_for_embeddings']) && $options['custom_provider_for_embeddings'] === 'on') {
4433		- if (!class_exists('MxChat_Utils')) {
4434		- require_once dirname(__FILE__) . '/../includes/class-mxchat-utils.php';
4435		- }
4436		- return MxChat_Utils::generate_embedding_custom($text, $options);
4437		- }
4438		-
4439		- $selected_model = $options['embedding_model'] ?? 'text-embedding-ada-002';
4440		- //error_log('[MXCHAT-EMBED] Selected embedding model for bot ' . $bot_id . ': ' . $selected_model);
4441		-
4442		- // Determine provider and endpoint
4443		- if (strpos($selected_model, 'voyage') === 0) {
4444		- $api_key = $options['voyage_api_key'] ?? '';
4445		- $endpoint = 'https://api.voyageai.com/v1/embeddings';
4446		- $provider_name = 'Voyage AI';
4447		- //error_log('[MXCHAT-EMBED] Using Voyage AI API for bot ' . $bot_id);
4448		- } elseif (strpos($selected_model, 'gemini-embedding') === 0) {
4449		- $api_key = $options['gemini_api_key'] ?? '';
4450		- $endpoint = 'https://generativelanguage.googleapis.com/v1beta/models/' . $selected_model . ':embedContent';
4451		- $provider_name = 'Google Gemini';
4452		- //error_log('[MXCHAT-EMBED] Using Google Gemini API for bot ' . $bot_id);
4453		- } else {
4454		- $api_key = $options['api_key'] ?? '';
4455		- $endpoint = 'https://api.openai.com/v1/embeddings';
4456		- $provider_name = 'OpenAI';
4457		- //error_log('[MXCHAT-EMBED] Using OpenAI API for bot ' . $bot_id);
4458		- }
4459		-
4460		- //error_log('[MXCHAT-EMBED] Using endpoint: ' . $endpoint);
4461		-
4462		- if (empty($api_key)) {
4463		- $error_message = sprintf('Missing %s API key for bot %s. Please configure your API key in the bot settings.', $provider_name, $bot_id);
4464		- //error_log('[MXCHAT-EMBED] Error: ' . $error_message);
4465		- return $error_message;
4466		- }
4467		-
4468		- // Check if text is too long (for OpenAI, roughly estimate tokens as words/0.75)
4469		- $estimated_tokens = ceil(str_word_count($text) / 0.75);
4470		- //error_log('[MXCHAT-EMBED] Estimated token count: ~' . $estimated_tokens);
4471		-
4472		- if ($estimated_tokens > 8000 && strpos($selected_model, 'voyage') === false && strpos($selected_model, 'gemini-embedding') === false) {
4473		- //error_log('[MXCHAT-EMBED] Warning: Text may exceed OpenAI token limits (8K for most models)');
4474		- // Consider truncating text here
4475		- }
4476		-
4477		- // Prepare request body based on provider
4478		- if (strpos($selected_model, 'gemini-embedding') === 0) {
4479		- // Gemini API format
4480		- $request_body = array(
4481		- 'model' => 'models/' . $selected_model,
4482		- 'content' => array(
4483		- 'parts' => array(
4484		- array('text' => $text)
4485		- )
4486		- )
4487		- );
4488		-
4489		- // Set output dimensionality to 1536 for consistency with other models
4490		- $request_body['outputDimensionality'] = 1536;
4491		- } else {
4492		- // OpenAI/Voyage API format
4493		- $request_body = array(
4494		- 'model' => $selected_model,
4495		- 'input' => $text
4496		- );
4497		-
4498		- // Add output_dimension for voyage-3-large model
4499		- if ($selected_model === 'voyage-3-large') {
4500		- $request_body['output_dimension'] = 2048;
4501		- }
4502		- }
4503		-
4504		- //error_log('[MXCHAT-EMBED] Request prepared with model: ' . $selected_model);
4505		-
4506		- // Prepare headers based on provider
4507		- if (strpos($selected_model, 'gemini-embedding') === 0) {
4508		- // Gemini uses API key as query parameter
4509		- $endpoint .= '?key=' . $api_key;
4510		- $headers = array(
4511		- 'Content-Type' => 'application/json'
4512		- );
4513		- } else {
4514		- // OpenAI/Voyage use Bearer token
4515		- $headers = array(
4516		- 'Authorization' => 'Bearer ' . $api_key,
4517		- 'Content-Type' => 'application/json'
4518		- );
4519		- }
4520		-
4521		- // Make API request
4522		- //error_log('[MXCHAT-EMBED] Sending API request to: ' . $endpoint);
4523		- $response = wp_remote_post($endpoint, array(
4524		- 'body' => wp_json_encode($request_body),
4525		- 'headers' => $headers,
4526		- 'timeout' => 60 // Increased timeout for large inputs
4527		- ));
4528		-
4529		- // Handle wp_remote_post errors
4530		- if (is_wp_error($response)) {
4531		- $error_message = $response->get_error_message();
4532		- //error_log('[MXCHAT-EMBED] WP Remote Post Error: ' . $error_message);
4533		- return 'Connection error: ' . $error_message;
4534		- }
4535		-
4536		- // Get and check HTTP response code
4537		- $http_code = wp_remote_retrieve_response_code($response);
4538		- //error_log('[MXCHAT-EMBED] API Response Code: ' . $http_code);
4539		-
4540		- if ($http_code !== 200) {
4541		- $error_body = wp_remote_retrieve_body($response);
4542		- //error_log('[MXCHAT-EMBED] API Error Response Body: ' . $error_body);
4543		-
4544		- // Try to parse error for more details
4545		- $error_json = json_decode($error_body, true);
4546		- if (json_last_error() === JSON_ERROR_NONE && isset($error_json['error'])) {
4547		- $error_type = $error_json['error']['type'] ?? 'unknown';
4548		- $error_message = $error_json['error']['message'] ?? 'No message';
4549		- //error_log('[MXCHAT-EMBED] API Error Type: ' . $error_type);
4550		- //error_log('[MXCHAT-EMBED] API Error Message: ' . $error_message);
4551		-
4552		- // Customize error message for common API errors
4553		- if ($error_type === 'invalid_request_error' && strpos($error_message, 'API key') !== false) {
4554		- $error_message = sprintf('Invalid %s API key for bot %s. Please check your API key in the bot settings.', $provider_name, $bot_id);
4555		- } elseif ($error_type === 'authentication_error') {
4556		- $error_message = sprintf('%s authentication failed for bot %s. Please verify your API key in the bot settings.', $provider_name, $bot_id);
4557		- }
4558		-
4559		- //error_log('[MXCHAT-EMBED] Returning error: ' . $error_message);
4560		- return $error_message;
4561		- }
4562		-
4563		- $error_message = sprintf("API Error (HTTP %d): Unable to generate embedding for bot %s", $http_code, $bot_id);
4564		- //error_log('[MXCHAT-EMBED] Returning error: ' . $error_message);
4565		- return $error_message;
4566		- }
4567		-
4568		- // Parse response body
4569		- $response_body = wp_remote_retrieve_body($response);
4570		- //error_log('[MXCHAT-EMBED] Received response length: ' . strlen($response_body) . ' bytes');
4571		-
4572		- $response_data = json_decode($response_body, true);
4573		-
4574		- if (json_last_error() !== JSON_ERROR_NONE) {
4575		- $error = json_last_error_msg();
4576		- //error_log('[MXCHAT-EMBED] JSON Parse Error: ' . $error);
4577		- //error_log('[MXCHAT-EMBED] Response preview: ' . substr($response_body, 0, 200));
4578		- return "Failed to parse API response: $error";
4579		- }
4580		-
4581		- // Handle different response formats based on provider
4582		- if (strpos($selected_model, 'gemini-embedding') === 0) {
4583		- // Gemini API response format
4584		- if (isset($response_data['embedding']['values'])) {
4585		- $embedding_dimensions = count($response_data['embedding']['values']);
4586		- //error_log('[MXCHAT-EMBED] Successfully extracted Gemini embedding with ' . $embedding_dimensions . ' dimensions');
4587		-
4588		- // Check if embedding dimensions are as expected (should be 1536)
4589		- if ($embedding_dimensions !== 1536) {
4590		- //error_log('[MXCHAT-EMBED] Warning: Unexpected Gemini embedding dimensions: ' . $embedding_dimensions);
4591		- }
4592		-
4593		- return $response_data['embedding']['values'];
4594		- } else {
4595		- //error_log('[MXCHAT-EMBED] Error: No embedding found in Gemini response');
4596		- //error_log('[MXCHAT-EMBED] Response structure: ' . wp_json_encode(array_keys($response_data)));
4597		-
4598		- if (isset($response_data['error'])) {
4599		- $error_message = "Gemini API Error in response: " . wp_json_encode($response_data['error']);
4600		- //error_log('[MXCHAT-EMBED] ' . $error_message);
4601		- return $error_message;
4602		- }
4603		-
4604		- $error_message = "Invalid Gemini API response format: No embedding found";
4605		- //error_log('[MXCHAT-EMBED] ' . $error_message);
4606		- return $error_message;
4607		- }
4608		- } else {
4609		- // OpenAI/Voyage API response format
4610		- if (isset($response_data['data'][0]['embedding'])) {
4611		- $embedding_dimensions = count($response_data['data'][0]['embedding']);
4612		- //error_log('[MXCHAT-EMBED] Successfully extracted embedding with ' . $embedding_dimensions . ' dimensions');
4613		-
4614		- // Check if embedding dimensions are as expected
4615		- if (($selected_model === 'text-embedding-ada-002' && $embedding_dimensions !== 1536) ||
4616		- ($selected_model === 'voyage-3-large' && $embedding_dimensions !== 2048)) {
4617		- //error_log('[MXCHAT-EMBED] Warning: Unexpected embedding dimensions');
4618		- }
4619		-
4620		- return $response_data['data'][0]['embedding'];
4621		- } else {
4622		- //error_log('[MXCHAT-EMBED] Error: No embedding found in response');
4623		- //error_log('[MXCHAT-EMBED] Response structure: ' . wp_json_encode(array_keys($response_data)));
4624		-
4625		- if (isset($response_data['error'])) {
4626		- $error_message = "API Error in response: " . wp_json_encode($response_data['error']);
4627		- //error_log('[MXCHAT-EMBED] ' . $error_message);
4628		- return $error_message;
4629		- }
4630		-
4631		- $error_message = "Invalid API response format: No embedding found";
4632		- //error_log('[MXCHAT-EMBED] ' . $error_message);
4633		- return $error_message;
4634		- }
4635		- }
4636		-}
4637		-
4638		-/**
4639		- * Get bot-specific options for multi-bot functionality
4640		- * Falls back to default options if bot_id is 'default' or multi-bot add-on is not active
4641		- */
4642		-private function get_bot_options($bot_id = 'default') {
4643		- //error_log("MXCHAT DEBUG: get_bot_options called for bot: " . $bot_id);
4644		-
4645		- if ($bot_id === 'default' || !class_exists('MxChat_Multi_Bot_Manager')) {
4646		- //error_log("MXCHAT DEBUG: Using default options (no multi-bot or bot is 'default')");
4647		- return array();
4648		- }
4649		-
4650		- $bot_options = apply_filters('mxchat_get_bot_options', array(), $bot_id);
4651		-
4652		- if (!empty($bot_options)) {
4653		- //error_log("MXCHAT DEBUG: Got bot-specific options from filter");
4654		- if (isset($bot_options['similarity_threshold'])) {
4655		- //error_log(" - similarity_threshold: " . $bot_options['similarity_threshold']);
4656		- }
4657		- }
4658		-
4659		- return is_array($bot_options) ? $bot_options : array();
4660		-}
4661		-
4662		-/**
4663		- * Get bot-specific Pinecone configuration
4664		- * Used in the knowledge retrieval functions
4665		- */
4666		-// Also add debugging to your get_bot_pinecone_config function
4667		-private function get_bot_pinecone_config($bot_id = 'default') {
4668		- //error_log("MXCHAT DEBUG: get_bot_pinecone_config called for bot: " . $bot_id);
4669		-
4670		- // If default bot or multi-bot add-on not active, use default Pinecone config
4671		- if ($bot_id === 'default' || !class_exists('MxChat_Multi_Bot_Manager')) {
4672		- //error_log("MXCHAT DEBUG: Using default Pinecone config (no multi-bot or bot is 'default')");
4673		- $addon_options = get_option('mxchat_pinecone_addon_options', array());
4674		- $config = array(
4675		- 'use_pinecone' => (isset($addon_options['mxchat_use_pinecone']) && $addon_options['mxchat_use_pinecone'] === '1'),
4676		- 'api_key' => $addon_options['mxchat_pinecone_api_key'] ?? '',
4677		- 'host' => $addon_options['mxchat_pinecone_host'] ?? '',
4678		- 'namespace' => $addon_options['mxchat_pinecone_namespace'] ?? ''
4679		- );
4680		- //error_log("MXCHAT DEBUG: Default config - use_pinecone: " . ($config['use_pinecone'] ? 'true' : 'false'));
4681		- return $config;
4682		- }
4683		-
4684		- //error_log("MXCHAT DEBUG: Calling filter 'mxchat_get_bot_pinecone_config' for bot: " . $bot_id);
4685		-
4686		- // Hook for multi-bot add-on to provide bot-specific Pinecone config
4687		- $bot_pinecone_config = apply_filters('mxchat_get_bot_pinecone_config', array(), $bot_id);
4688		-
4689		- if (!empty($bot_pinecone_config)) {
4690		- //error_log("MXCHAT DEBUG: Got bot-specific config from filter");
4691		- //error_log(" - use_pinecone: " . (isset($bot_pinecone_config['use_pinecone']) ? ($bot_pinecone_config['use_pinecone'] ? 'true' : 'false') : 'not set'));
4692		- //error_log(" - host: " . ($bot_pinecone_config['host'] ?? 'not set'));
4693		- //error_log(" - namespace: " . ($bot_pinecone_config['namespace'] ?? 'not set'));
4694		- } else {
4695		- //error_log("MXCHAT DEBUG: Filter returned empty config!");
4696		- }
4697		-
4698		- return is_array($bot_pinecone_config) ? $bot_pinecone_config : array();
4699		-}
4700		-
4701		-
4702		-public function mxchat_ajax_dismiss_completed_status() {
4703		- try {
4704		- // Verify the request
4705		- check_ajax_referer('mxchat_status_nonce', 'nonce');
4706		-
4707		- if (!current_user_can('manage_options')) {
4708		- wp_send_json_error('Unauthorized access');
4709		- exit;
4710		- }
4711		-
4712		- $card_type = isset($_POST['card_type']) ? sanitize_text_field($_POST['card_type']) : '';
4713		-
4714		- if ($card_type === 'pdf') {
4715		- // Clear PDF status
4716		- $pdf_url = get_transient('mxchat_last_pdf_url');
4717		- if ($pdf_url) {
4718		- delete_transient('mxchat_pdf_status_' . md5($pdf_url));
4719		- delete_transient('mxchat_last_pdf_url');
4720		- }
4721		- } elseif ($card_type === 'sitemap') {
4722		- // Clear sitemap status
4723		- $sitemap_url = get_transient('mxchat_last_sitemap_url');
4724		- if ($sitemap_url) {
4725		- delete_transient('mxchat_sitemap_status_' . md5($sitemap_url));
4726		- delete_transient('mxchat_last_sitemap_url');
4727		- }
4728		- }
4729		-
4730		- wp_send_json_success(array('message' => 'Status dismissed successfully'));
4731		-
4732		- } catch (Exception $e) {
4733		- wp_send_json_error(array('message' => 'Error dismissing status: ' . $e->getMessage()));
4734		- }
4735		-}
4736		-
4737		-/**
4738		- * Render completed status cards on page load
4739		- * This ensures completed processing status persists through page refreshes
4740		- */
4741		-public function mxchat_render_completed_status_cards() {
4742		- $output = '';
4743		-
4744		- // Check for completed PDF status
4745		- $pdf_url = get_transient('mxchat_last_pdf_url');
4746		- if ($pdf_url) {
4747		- $pdf_status = $this->mxchat_get_pdf_processing_status($pdf_url);
4748		- if ($pdf_status && ($pdf_status['status'] === 'complete' || $pdf_status['status'] === 'error')) {
4749		- $output .= $this->mxchat_render_pdf_status_card($pdf_status, $pdf_url);
4750		- }
4751		- }
4752		-
4753		- // Check for completed sitemap status
4754		- $sitemap_url = get_transient('mxchat_last_sitemap_url');
4755		- if ($sitemap_url) {
4756		- $sitemap_status = $this->mxchat_get_sitemap_processing_status($sitemap_url);
4757		- if ($sitemap_status && ($sitemap_status['status'] === 'complete' || $sitemap_status['status'] === 'error')) {
4758		- $output .= $this->mxchat_render_sitemap_status_card($sitemap_status, $sitemap_url);
4759		- }
4760		- }
4761		-
4762		- return $output;
4763		-}
4764		-
4765		-/**
4766		- * Render PDF status card HTML
4767		- */
4768		-private function mxchat_render_pdf_status_card($status, $pdf_url) {
4769		- $html = '<div class="mxchat-status-card" data-card-type="pdf">';
4770		- $html .= '<div class="mxchat-status-header">';
4771		- $html .= '<h4>' . esc_html__('PDF Processing Status', 'mxchat') . '</h4>';
4772		-
4773		- // Add dismiss button for completed status
4774		- if ($status['status'] === 'complete' || $status['status'] === 'error') {
4775		- $html .= '<button type="button" class="mxchat-dismiss-button">' . esc_html__('Dismiss', 'mxchat') . '</button>';
4776		- }
4777		-
4778		- // Process Batch button for processing status
4779		- if ($status['status'] === 'processing') {
4780		- $html .= '<button type="button" class="mxchat-manual-batch-btn"
4781		- data-process-type="pdf"
4782		- data-url="' . esc_attr($pdf_url) . '">
4783		- ' . esc_html__('Process Batch', 'mxchat') . '</button>';
4784		- }
4785		-
4786		- // Add status badges
4787		- if ($status['status'] === 'error') {
4788		- $html .= '<span class="mxchat-status-badge mxchat-status-failed">' . esc_html__('Error', 'mxchat') . '</span>';
4789		- } elseif ($status['status'] === 'complete') {
4790		- if ($status['failed_pages'] > 0) {
4791		- $html .= '<span class="mxchat-status-badge mxchat-status-warning">' .
4792		- sprintf(esc_html__('Completed with %d failures', 'mxchat'), $status['failed_pages']) . '</span>';
4793		- } else {
4794		- $html .= '<span class="mxchat-status-badge mxchat-status-success">' . esc_html__('Complete', 'mxchat') . '</span>';
4795		- }
4796		- }
4797		-
4798		- $html .= '</div>'; // End header
4799		-
4800		- // Progress bar
4801		- $html .= '<div class="mxchat-progress-bar">';
4802		- $html .= '<div class="mxchat-progress-fill" style="width: ' . esc_attr($status['percentage']) . '%"></div>';
4803		- $html .= '</div>';
4804		-
4805		- // Status details
4806		- $html .= '<div class="mxchat-status-details">';
4807		- $html .= '<p>' . sprintf(
4808		- esc_html__('Progress: %d of %d pages (%d%%)', 'mxchat'),
4809		- $status['processed_pages'],
4810		- $status['total_pages'],
4811		- $status['percentage']
4812		- ) . '</p>';
4813		-
4814		- // Show failed pages count if any
4815		- if ($status['failed_pages'] > 0) {
4816		- $html .= '<p><strong>' . esc_html__('Failed pages:', 'mxchat') . '</strong> ' . esc_html($status['failed_pages']) . '</p>';
4817		- }
4818		-
4819		- $html .= '<p><strong>' . esc_html__('Status:', 'mxchat') . '</strong> ' . esc_html(ucfirst($status['status'])) . '</p>';
4820		- $html .= '<p><strong>' . esc_html__('Last update:', 'mxchat') . '</strong> ' . esc_html($status['last_update']) . '</p>';
4821		-
4822		- // Add completion summary if available AND it's an array
4823		- if (isset($status['completion_summary']) && is_array($status['completion_summary']) && !empty($status['completion_summary'])) {
4824		- $summary = $status['completion_summary'];
4825		- $html .= '<div class="mxchat-completion-summary">';
4826		- $html .= '<h5>' . esc_html__('Processing Summary', 'mxchat') . '</h5>';
4827		- $html .= '<p><strong>' . esc_html__('Total Pages:', 'mxchat') . '</strong> ' . esc_html($summary['total_pages']) . '</p>';
4828		- $html .= '<p><strong>' . esc_html__('Successful:', 'mxchat') . '</strong> ' . esc_html($summary['successful_pages']) . '</p>';
4829		- $html .= '<p><strong>' . esc_html__('Failed:', 'mxchat') . '</strong> ' . esc_html($summary['failed_pages']) . '</p>';
4830		- $html .= '<p><strong>' . esc_html__('Completed:', 'mxchat') . '</strong> ' . esc_html($summary['completion_time']) . '</p>';
4831		- $html .= '</div>';
4832		- }
4833		-
4834		- // Add failed pages list if any AND it's an array
4835		- if (isset($status['failed_pages_list']) && is_array($status['failed_pages_list']) && !empty($status['failed_pages_list'])) {
4836		- $html .= $this->mxchat_render_failed_pages_list($status['failed_pages_list']);
4837		- }
4838		-
4839		- // Add error message if any
4840		- if (isset($status['error']) && !empty($status['error'])) {
4841		- $html .= '<div class="mxchat-error-notice">';
4842		- $html .= '<p class="error">' . esc_html($status['error']) . '</p>';
4843		- $html .= '</div>';
4844		- }
4845		-
4846		- $html .= '</div>'; // End details
4847		- $html .= '</div>'; // End card
4848		-
4849		- return $html;
4850		-}
4851		-/**
4852		- * Render sitemap status card HTML
4853		- */
4854		-private function mxchat_render_sitemap_status_card($status, $sitemap_url) {
4855		- $html = '<div class="mxchat-status-card" data-card-type="sitemap">';
4856		- $html .= '<div class="mxchat-status-header">';
4857		- $html .= '<h4>' . esc_html__('Sitemap Processing Status', 'mxchat') . '</h4>';
4858		-
4859		- // Add dismiss button for completed status
4860		- if ($status['status'] === 'complete' || $status['status'] === 'error') {
4861		- $html .= '<button type="button" class="mxchat-dismiss-button">' . esc_html__('Dismiss', 'mxchat') . '</button>';
4862		- }
4863		-
4864		- // Process Batch button for processing status
4865		- if ($status['status'] === 'processing') {
4866		- $html .= '<button type="button" class="mxchat-manual-batch-btn"
4867		- data-process-type="sitemap"
4868		- data-url="' . esc_attr($sitemap_url) . '">
4869		- ' . esc_html__('Process Batch', 'mxchat') . '</button>';
4870		- }
4871		-
4872		- // Add status badges
4873		- if ($status['status'] === 'error') {
4874		- $html .= '<span class="mxchat-status-badge mxchat-status-failed">' . esc_html__('Error', 'mxchat') . '</span>';
4875		- } elseif ($status['status'] === 'complete') {
4876		- if ($status['failed_urls'] > 0) {
4877		- $html .= '<span class="mxchat-status-badge mxchat-status-warning">' .
4878		- sprintf(esc_html__('Completed with %d failures', 'mxchat'), $status['failed_urls']) . '</span>';
4879		- } else {
4880		- $html .= '<span class="mxchat-status-badge mxchat-status-success">' . esc_html__('Complete', 'mxchat') . '</span>';
4881		- }
4882		- }
4883		-
4884		- $html .= '</div>'; // End header
4885		-
4886		- // Progress bar
4887		- $html .= '<div class="mxchat-progress-bar">';
4888		- $html .= '<div class="mxchat-progress-fill" style="width: ' . esc_attr($status['percentage']) . '%"></div>';
4889		- $html .= '</div>';
4890		-
4891		- // Status details
4892		- $html .= '<div class="mxchat-status-details">';
4893		- $html .= '<p>' . sprintf(
4894		- esc_html__('Progress: %d of %d URLs (%d%%)', 'mxchat'),
4895		- $status['processed_urls'],
4896		- $status['total_urls'],
4897		- $status['percentage']
4898		- ) . '</p>';
4899		-
4900		- // Show failed URLs count if any
4901		- if ($status['failed_urls'] > 0) {
4902		- $html .= '<p><strong>' . esc_html__('Failed URLs:', 'mxchat') . '</strong> ' . esc_html($status['failed_urls']) . '</p>';
4903		- }
4904		-
4905		- $html .= '<p><strong>' . esc_html__('Status:', 'mxchat') . '</strong> ' . esc_html(ucfirst($status['status'])) . '</p>';
4906		- $html .= '<p><strong>' . esc_html__('Last update:', 'mxchat') . '</strong> ' . esc_html($status['last_update']) . '</p>';
4907		-
4908		- // Add completion summary if available AND it's an array
4909		- if (isset($status['completion_summary']) && is_array($status['completion_summary']) && !empty($status['completion_summary'])) {
4910		- $summary = $status['completion_summary'];
4911		- $html .= '<div class="mxchat-completion-summary">';
4912		- $html .= '<h5>' . esc_html__('Processing Summary', 'mxchat') . '</h5>';
4913		- $html .= '<p><strong>' . esc_html__('Total URLs:', 'mxchat') . '</strong> ' . esc_html($summary['total_urls']) . '</p>';
4914		- $html .= '<p><strong>' . esc_html__('Successful:', 'mxchat') . '</strong> ' . esc_html($summary['successful_urls']) . '</p>';
4915		- $html .= '<p><strong>' . esc_html__('Failed:', 'mxchat') . '</strong> ' . esc_html($summary['failed_urls']) . '</p>';
4916		- $html .= '<p><strong>' . esc_html__('Completed:', 'mxchat') . '</strong> ' . esc_html($summary['completion_time']) . '</p>';
4917		- $html .= '</div>';
4918		- }
4919		-
4920		- // Add error messages if any (but not the failed URLs list)
4921		- if (!empty($status['error']) || !empty($status['last_error'])) {
4922		- $html .= '<div class="mxchat-error-notice">';
4923		-
4924		- if (!empty($status['error'])) {
4925		- $html .= '<p class="error">' . esc_html($status['error']) . '</p>';
4926		- }
4927		-
4928		- if (!empty($status['last_error'])) {
4929		- $html .= '<p class="last-error">' . esc_html__('Last error:', 'mxchat') . ' ' . esc_html($status['last_error']) . '</p>';
4930		- }
4931		-
4932		- $html .= '</div>';
4933		- }
4934		-
4935		- $html .= '</div>'; // End details
4936		- $html .= '</div>'; // End card
4937		-
4938		- return $html;
4939		-}
4940		-
4941		-
4942		-/**
4943		- * Render failed pages list
4944		- */
4945		-private function mxchat_render_failed_pages_list($failed_pages_list) {
4946		- // Validate that $failed_pages_list is an array and not empty
4947		- if (!is_array($failed_pages_list) || empty($failed_pages_list)) {
4948		- return '';
4949		- }
4950		-
4951		- $html = '<div class="mxchat-error-notice">';
4952		- $html .= '<div class="mxchat-failed-pages-container">';
4953		- $html .= '<h5>' . sprintf(esc_html__('Failed Pages (%d)', 'mxchat'), count($failed_pages_list)) . '</h5>';
4954		- $html .= '<details>';
4955		- $html .= '<summary>' . esc_html__('Show Failed Pages', 'mxchat') . '</summary>';
4956		- $html .= '<div class="mxchat-failed-pages-list">';
4957		-
4958		- // Create table for failed pages
4959		- $html .= '<table class="widefat striped">';
4960		- $html .= '<thead><tr>';
4961		- $html .= '<th>' . esc_html__('Page', 'mxchat') . '</th>';
4962		- $html .= '<th>' . esc_html__('Error', 'mxchat') . '</th>';
4963		- $html .= '<th>' . esc_html__('Retries', 'mxchat') . '</th>';
4964		- $html .= '<th>' . esc_html__('Time', 'mxchat') . '</th>';
4965		- $html .= '</tr></thead><tbody>';
4966		-
4967		- // Sort failed pages by most recent
4968		- $sorted_failed_pages = $failed_pages_list;
4969		- usort($sorted_failed_pages, function($a, $b) {
4970		- return ($b['time'] ?? 0) - ($a['time'] ?? 0);
4971		- });
4972		-
4973		- foreach ($sorted_failed_pages as $item) {
4974		- // Ensure $item is an array before accessing its elements
4975		- if (!is_array($item)) {
4976		- continue;
4977		- }
4978		-
4979		- $time_ago = isset($item['time']) ? human_time_diff($item['time'], current_time('timestamp')) . ' ' . esc_html__('ago', 'mxchat') : 'Unknown';
4980		- $html .= '<tr>';
4981		- $html .= '<td>' . esc_html__('Page', 'mxchat') . ' ' . esc_html($item['page'] ?? 'Unknown') . '</td>';
4982		- $html .= '<td style="word-break: break-word;">' . esc_html($item['error'] ?? 'Unknown error') . '</td>';
4983		- $html .= '<td>' . esc_html($item['retries'] ?? 'N/A') . '</td>';
4984		- $html .= '<td>' . esc_html($time_ago) . '</td>';
4985		- $html .= '</tr>';
4986		- }
4987		-
4988		- $html .= '</tbody></table>';
4989		- $html .= '</div></details></div></div>';
4990		-
4991		- return $html;
4992		-}
4993		-
4994		-/**
4995		- * Render failed URLs list
4996		- */
4997		-private function mxchat_render_failed_urls_list($failed_urls_list) {
4998		- // Validate that $failed_urls_list is an array and not empty
4999		- if (!is_array($failed_urls_list) || empty($failed_urls_list)) {
5000		- return '';
5001		- }
5002		-
5003		- $html = '<div class="mxchat-failed-urls-container">';
5004		- $html .= '<h5>' . sprintf(esc_html__('Failed URLs (%d)', 'mxchat'), count($failed_urls_list)) . '</h5>';
5005		- $html .= '<details>';
5006		- $html .= '<summary>' . esc_html__('Show Failed URLs', 'mxchat') . '</summary>';
5007		- $html .= '<div class="mxchat-failed-urls-list">';
5008		-
5009		- // Create table for failed URLs
5010		- $html .= '<table class="widefat striped">';
5011		- $html .= '<thead><tr>';
5012		- $html .= '<th>' . esc_html__('URL', 'mxchat') . '</th>';
5013		- $html .= '<th>' . esc_html__('Error', 'mxchat') . '</th>';
5014		- $html .= '<th>' . esc_html__('Retries', 'mxchat') . '</th>';
5015		- $html .= '<th>' . esc_html__('Time', 'mxchat') . '</th>';
5016		- $html .= '</tr></thead><tbody>';
5017		-
5018		- // Sort failed URLs by most recent
5019		- $sorted_failed_urls = $failed_urls_list;
5020		- usort($sorted_failed_urls, function($a, $b) {
5021		- return ($b['time'] ?? 0) - ($a['time'] ?? 0);
5022		- });
5023		-
5024		- // Show up to 50 failed URLs
5025		- $display_urls = array_slice($sorted_failed_urls, 0, 50);
5026		-
5027		- foreach ($display_urls as $item) {
5028		- // Ensure $item is an array before accessing its elements
5029		- if (!is_array($item)) {
5030		- continue;
5031		- }
5032		-
5033		- $url = $item['url'] ?? '';
5034		- $time_ago = isset($item['time']) ? human_time_diff($item['time'], current_time('timestamp')) . ' ' . esc_html__('ago', 'mxchat') : 'Unknown';
5035		-
5036		- // Truncate URL for display
5037		- $display_url = strlen($url) > 50 ? substr($url, 0, 47) . '...' : $url;
5038		-
5039		- $html .= '<tr>';
5040		- $html .= '<td style="word-break: break-all;">';
5041		- if (!empty($url)) {
5042		- $html .= '<a href="' . esc_url($url) . '" target="_blank" rel="noopener noreferrer">' . esc_html($display_url) . '</a>';
5043		- } else {
5044		- $html .= esc_html__('Unknown URL', 'mxchat');
5045		- }
5046		- $html .= '</td>';
5047		- $html .= '<td style="word-break: break-word;">' . esc_html($item['error'] ?? 'Unknown error') . '</td>';
5048		- $html .= '<td>' . esc_html($item['retries'] ?? 'N/A') . '</td>';
5049		- $html .= '<td>' . esc_html($time_ago) . '</td>';
5050		- $html .= '</tr>';
5051		- }
5052		-
5053		- $html .= '</tbody></table>';
5054		-
5055		- if (count($failed_urls_list) > 50) {
5056		- $html .= '<div class="mxchat-failed-urls-more">+ ' .
5057		- (count($failed_urls_list) - 50) .
5058		- ' ' . esc_html__('more failed URLs not shown', 'mxchat') . '</div>';
5059		- }
5060		-
5061		- $html .= '</div></details></div>';
5062		-
5063		- return $html;
5064		-}
5065		-
5066		-/**
5067		- * Get all ACF fields for a specific post, excluding any fields the user has disabled
5068		- */
5069		-public function mxchat_get_acf_fields_for_post($post_id) {
5070		- if (!function_exists('get_fields')) {
5071		- return array();
5072		- }
5073		-
5074		- $fields = get_fields($post_id);
5075		- if (!$fields || !is_array($fields)) {
5076		- return array();
5077		- }
5078		-
5079		- // Get excluded fields from settings
5080		- $excluded_fields = get_option('mxchat_acf_excluded_fields', array());
5081		- if (!empty($excluded_fields) && is_array($excluded_fields)) {
5082		- foreach ($excluded_fields as $excluded_field) {
5083		- if (isset($fields[$excluded_field])) {
5084		- unset($fields[$excluded_field]);
5085		- }
5086		- }
5087		- }
5088		-
5089		- return $fields;
5090		-}
5091		-
5092		-/**
5093		- * Get all registered ACF field groups and their fields for the settings UI
5094		- */
5095		-public function mxchat_get_all_acf_fields() {
5096		- if (!function_exists('acf_get_field_groups') || !function_exists('acf_get_fields')) {
5097		- return array();
5098		- }
5099		-
5100		- $all_fields = array();
5101		- $field_groups = acf_get_field_groups();
5102		-
5103		- if (!empty($field_groups)) {
5104		- foreach ($field_groups as $group) {
5105		- $group_fields = acf_get_fields($group['key']);
5106		- if (!empty($group_fields)) {
5107		- $all_fields[$group['title']] = array();
5108		- foreach ($group_fields as $field) {
5109		- $all_fields[$group['title']][] = array(
5110		- 'name' => $field['name'],
5111		- 'label' => $field['label'],
5112		- 'type' => $field['type']
5113		- );
5114		- }
5115		- }
5116		- }
5117		- }
5118		-
5119		- return $all_fields;
5120		-}
5121		-
5122		-/**
5123		- * Get whitelisted custom post meta for a given post
5124		- * This allows non-ACF meta fields (like OptionTree, theme meta boxes) to be included in embeddings
5125		- */
5126		-public function mxchat_get_whitelisted_post_meta($post_id) {
5127		- $whitelist = get_option('mxchat_custom_meta_whitelist', '');
5128		-
5129		- if (empty($whitelist)) {
5130		- return array();
5131		- }
5132		-
5133		- // Parse the whitelist - one meta key per line
5134		- $meta_keys = array_filter(array_map('trim', explode("\n", $whitelist)));
5135		-
5136		- if (empty($meta_keys)) {
5137		- return array();
5138		- }
5139		-
5140		- $result = array();
5141		-
5142		- foreach ($meta_keys as $key) {
5143		- // Skip empty keys
5144		- if (empty($key)) {
5145		- continue;
5146		- }
5147		-
5148		- $value = get_post_meta($post_id, $key, true);
5149		-
5150		- // Only include non-empty string values
5151		- if (!empty($value) && is_string($value)) {
5152		- $result[$key] = $value;
5153		- } elseif (!empty($value) && is_array($value)) {
5154		- // Handle array values by joining them
5155		- $flat_value = $this->mxchat_flatten_meta_array($value);
5156		- if (!empty($flat_value)) {
5157		- $result[$key] = $flat_value;
5158		- }
5159		- }
5160		- }
5161		-
5162		- return $result;
5163		-}
5164		-
5165		-/**
5166		- * Flatten array meta values into a readable string
5167		- */
5168		-private function mxchat_flatten_meta_array($array, $depth = 0) {
5169		- if ($depth > 3) {
5170		- return ''; // Prevent infinite recursion
5171		- }
5172		-
5173		- $parts = array();
5174		-
5175		- foreach ($array as $key => $value) {
5176		- if (is_string($value) && !empty($value)) {
5177		- $parts[] = $value;
5178		- } elseif (is_array($value)) {
5179		- $nested = $this->mxchat_flatten_meta_array($value, $depth + 1);
5180		- if (!empty($nested)) {
5181		- $parts[] = $nested;
5182		- }
5183		- }
5184		- }
5185		-
5186		- return implode(', ', $parts);
5187		-}
5188		-
5189		-/**
5190		- * Format ACF field values for content extraction
5191		- */
5192		-public function mxchat_format_acf_field_value($value, $field_name = '', $post_id = 0) {
5193		- if (empty($value)) {
5194		- return '';
5195		- }
5196		-
5197		- // Handle WP_Post objects first (THIS IS THE KEY FIX)
5198		- if ($value instanceof WP_Post) {
5199		- return $value->post_title ?: '';
5200		- }
5201		-
5202		- // Handle other WP objects
5203		- if (is_object($value)) {
5204		- if (isset($value->post_title)) {
5205		- return $value->post_title;
5206		- } elseif (isset($value->display_name)) {
5207		- return $value->display_name;
5208		- } elseif (isset($value->name)) {
5209		- return $value->name;
5210		- } elseif (method_exists($value, '__toString')) {
5211		- try {
5212		- return (string) $value;
5213		- } catch (Exception $e) {
5214		- return '';
5215		- }
5216		- }
5217		- // For any other objects, return empty string
5218		- return '';
5219		- }
5220		-
5221		- // Handle different ACF field types
5222		- if (is_array($value)) {
5223		- // Check if it's an image/file field
5224		- if (isset($value['url'])) {
5225		- // Image field - return alt text, title, or caption
5226		- if (!empty($value['alt'])) {
5227		- return $value['alt'];
5228		- } elseif (!empty($value['title'])) {
5229		- return $value['title'];
5230		- } elseif (!empty($value['caption'])) {
5231		- return $value['caption'];
5232		- } else {
5233		- return ''; // Don't include just the URL
5234		- }
5235		- }
5236		-
5237		- // Check if it's a post object or relationship field
5238		- if (isset($value['post_title'])) {
5239		- return $value['post_title'];
5240		- }
5241		-
5242		- // Check if it's a user field
5243		- if (isset($value['display_name'])) {
5244		- return $value['display_name'];
5245		- }
5246		-
5247		- // Check if it's a taxonomy term
5248		- if (isset($value['name']) && isset($value['taxonomy'])) {
5249		- return $value['name'];
5250		- }
5251		-
5252		- // Check if it's a select field with label
5253		- if (isset($value['label'])) {
5254		- return $value['label'];
5255		- }
5256		-
5257		- // Check for repeater field or flexible content
5258		- if (is_numeric(key($value))) {
5259		- $sub_values = array();
5260		- foreach ($value as $sub_item) {
5261		- if (is_array($sub_item)) {
5262		- // For repeater/flexible content, extract text values
5263		- $sub_text = $this->mxchat_extract_text_from_acf_array($sub_item);
5264		- if (!empty($sub_text)) {
5265		- $sub_values[] = $sub_text;
5266		- }
5267		- } elseif ($sub_item instanceof WP_Post) {
5268		- // Handle WP_Post objects in arrays
5269		- $sub_values[] = $sub_item->post_title ?: '';
5270		- } else {
5271		- $sub_values[] = (string) $sub_item;
5272		- }
5273		- }
5274		- return implode(', ', array_filter($sub_values));
5275		- }
5276		-
5277		- // For other arrays, try to extract meaningful text
5278		- $text_values = array();
5279		- foreach ($value as $key => $val) {
5280		- if (is_string($val) && !empty(trim($val))) {
5281		- $text_values[] = trim($val);
5282		- } elseif ($val instanceof WP_Post) {
5283		- // Handle WP_Post objects in associative arrays
5284		- $text_values[] = $val->post_title ?: '';
5285		- } elseif (is_array($val) && isset($val['post_title'])) {
5286		- $text_values[] = $val['post_title'];
5287		- } elseif (is_array($val) && isset($val['name'])) {
5288		- $text_values[] = $val['name'];
5289		- }
5290		- }
5291		-
5292		- return implode(', ', array_filter($text_values));
5293		- }
5294		-
5295		- // Handle boolean values
5296		- if (is_bool($value)) {
5297		- return $value ? 'Yes' : 'No';
5298		- }
5299		-
5300		- // Handle numeric values
5301		- if (is_numeric($value)) {
5302		- return (string) $value;
5303		- }
5304		-
5305		- // Handle string values
5306		- if (is_string($value)) {
5307		- return trim($value);
5308		- }
5309		-
5310		- // For anything else that we can't handle, return empty string
5311		- // This prevents the "Object could not be converted to string" error
5312		- return '';
5313		-}
5314		-
5315		-/**
5316		- * Extract text from complex ACF array structures
5317		- */
5318		-private function mxchat_extract_text_from_acf_array($array) {
5319		- if (!is_array($array)) {
5320		- return '';
5321		- }
5322		-
5323		- $text_parts = array();
5324		-
5325		- foreach ($array as $key => $value) {
5326		- if (is_string($value) && !empty(trim($value))) {
5327		- // Skip keys that are likely to be IDs or technical values
5328		- if (!is_numeric($value) || strlen($value) > 10) {
5329		- $text_parts[] = trim($value);
5330		- }
5331		- } elseif ($value instanceof WP_Post) {
5332		- // Handle WP_Post objects
5333		- $text_parts[] = $value->post_title ?: '';
5334		- } elseif (is_array($value)) {
5335		- if (isset($value['post_title'])) {
5336		- $text_parts[] = $value['post_title'];
5337		- } elseif (isset($value['name'])) {
5338		- $text_parts[] = $value['name'];
5339		- } elseif (isset($value['label'])) {
5340		- $text_parts[] = $value['label'];
5341		- }
5342		- } elseif (is_object($value)) {
5343		- // Handle other objects safely
5344		- if (isset($value->post_title)) {
5345		- $text_parts[] = $value->post_title;
5346		- } elseif (isset($value->name)) {
5347		- $text_parts[] = $value->name;
5348		- } elseif (isset($value->display_name)) {
5349		- $text_parts[] = $value->display_name;
5350		- }
5351		- }
5352		- }
5353		-
5354		- return implode(', ', array_filter($text_parts));
5355		-}
5356		-
5357		-/**
5358		- * Walk an ACF field value tree and collect attachment IDs for any value that
5359		- * resolves to a PDF in the WordPress media library. Handles the three shapes
5360		- * ACF returns for File/Image/URL fields (array with ID+url, integer attachment ID,
5361		- * plain URL string), and recurses through repeater/group/flexible content.
5362		- *
5363		- * @param mixed $value The ACF field value (any depth)
5364		- * @param array $out Accumulator (passed by reference) for attachment IDs
5365		- * @param int $depth Recursion guard
5366		- */
5367		-private function mxchat_collect_pdf_attachment_ids_from_acf_value($value, &$out, $depth = 0) {
5368		- if ($depth > 6) {
5369		- return; // prevent runaway recursion on circular/very-deep structures
5370		- }
5371		-
5372		- if (empty($value)) {
5373		- return;
5374		- }
5375		-
5376		- // Array shapes: ACF File/Image return value=array; repeaters/groups are arrays of arrays
5377		- if (is_array($value)) {
5378		- // Direct File/Image-style array (has 'url' and usually 'ID' + 'mime_type')
5379		- $looks_like_attachment = isset($value['url']) || isset($value['ID']) || isset($value['id']);
5380		- if ($looks_like_attachment) {
5381		- $att_id = 0;
5382		- if (!empty($value['ID']) && is_numeric($value['ID'])) {
5383		- $att_id = (int) $value['ID'];
5384		- } elseif (!empty($value['id']) && is_numeric($value['id'])) {
5385		- $att_id = (int) $value['id'];
5386		- } elseif (!empty($value['url']) && is_string($value['url'])) {
5387		- $att_id = (int) attachment_url_to_postid($value['url']);
5388		- }
5389		-
5390		- $is_pdf = false;
5391		- if (!empty($value['mime_type']) && $value['mime_type'] === 'application/pdf') {
5392		- $is_pdf = true;
5393		- } elseif (!empty($value['subtype']) && strtolower((string) $value['subtype']) === 'pdf') {
5394		- $is_pdf = true;
5395		- } elseif (!empty($value['url']) && is_string($value['url']) && $this->mxchat_url_looks_like_pdf($value['url'])) {
5396		- $is_pdf = true;
5397		- } elseif ($att_id && get_post_mime_type($att_id) === 'application/pdf') {
5398		- $is_pdf = true;
5399		- }
5400		-
5401		- if ($is_pdf && $att_id && get_post_mime_type($att_id) === 'application/pdf') {
5402		- $out[] = $att_id;
5403		- }
5404		- // An array node that represents one attachment doesn't contain other
5405		- // attachments inside it — done with this branch.
5406		- return;
5407		- }
5408		-
5409		- // Recurse: repeater rows, flexible-content layouts, groups, etc.
5410		- foreach ($value as $sub) {
5411		- $this->mxchat_collect_pdf_attachment_ids_from_acf_value($sub, $out, $depth + 1);
5412		- }
5413		- return;
5414		- }
5415		-
5416		- // Plain numeric attachment ID (ACF File field set to "Return: ID")
5417		- if (is_numeric($value)) {
5418		- $att_id = (int) $value;
5419		- if ($att_id > 0 && get_post_mime_type($att_id) === 'application/pdf') {
5420		- $out[] = $att_id;
5421		- }
5422		- return;
5423		- }
5424		-
5425		- // Plain string — URL pointing at a PDF (ACF File field set to "Return: URL", or a custom URL/text field)
5426		- if (is_string($value)) {
5427		- $trimmed = trim($value);
5428		- if ($trimmed !== '' && $this->mxchat_url_looks_like_pdf($trimmed)) {
5429		- $att_id = (int) attachment_url_to_postid($trimmed);
5430		- if ($att_id > 0 && get_post_mime_type($att_id) === 'application/pdf') {
5431		- $out[] = $att_id;
5432		- }
5433		- }
5434		- return;
5435		- }
5436		-}
5437		-
5438		-/**
5439		- * Heuristic: does this URL/string look like a PDF reference?
5440		- * Tolerates query strings and fragments (#page=2).
5441		- */
5442		-private function mxchat_url_looks_like_pdf($url) {
5443		- if (!is_string($url) || $url === '') {
5444		- return false;
5445		- }
5446		- // Strip query + fragment before checking extension
5447		- $path = preg_replace('/[?#].*$/', '', $url);
5448		- return (bool) preg_match('/\.pdf$/i', $path);
5449		-}
5450		-
5451		-/**
5452		- * Extract text from a PDF attachment by ID using the bundled Smalot parser.
5453		- * Reads the file directly from disk via get_attached_file (no HTTP fetch).
5454		- * Result is cached on the attachment as post_meta keyed by file mtime so we
5455		- * only parse the same PDF once unless the file changes on disk.
5456		- *
5457		- * @param int $attachment_id
5458		- * @return string Extracted plain text, or '' on failure.
5459		- */
5460		-private function mxchat_extract_pdf_text_by_attachment_id($attachment_id) {
5461		- $attachment_id = (int) $attachment_id;
5462		- if ($attachment_id <= 0) {
5463		- return '';
5464		- }
5465		- if (get_post_mime_type($attachment_id) !== 'application/pdf') {
5466		- return '';
5467		- }
5468		-
5469		- $pdf_path = get_attached_file($attachment_id);
5470		- if (empty($pdf_path) || !file_exists($pdf_path) || !is_readable($pdf_path)) {
5471		- return '';
5472		- }
5473		-
5474		- // Raw-file size cap. Parsing very large PDFs can OOM the request; skip with a log entry
5475		- // and let the rest of the ACF content land in the KB. Filterable for users who need it bigger.
5476		- $default_max_bytes = 25 * 1024 * 1024;
5477		- $max_bytes = (int) apply_filters('mxchat_acf_pdf_max_bytes', $default_max_bytes, $attachment_id, $pdf_path);
5478		- if ($max_bytes > 0) {
5479		- $file_size = @filesize($pdf_path);
5480		- if ($file_size !== false && $file_size > $max_bytes) {
5481		- error_log(sprintf(
5482		- '[mxchat] ACF PDF skipped (over size cap): attachment %d "%s" %d bytes > cap %d',
5483		- $attachment_id,
5484		- basename($pdf_path),
5485		- $file_size,
5486		- $max_bytes
5487		- ));
5488		- return '';
5489		- }
5490		- }
5491		-
5492		- $mtime = @filemtime($pdf_path);
5493		- $cache_meta_key = '_mxchat_acf_pdf_text_v1';
5494		- $cached = get_post_meta($attachment_id, $cache_meta_key, true);
5495		- if (is_array($cached) && isset($cached['mtime'], $cached['text']) && (int) $cached['mtime'] === (int) $mtime) {
5496		- return (string) $cached['text'];
5497		- }
5498		-
5499		- $text = '';
5500		- try {
5501		- if (function_exists('mxchat_load_pdf_parser')) {
5502		- mxchat_load_pdf_parser();
5503		- }
5504		- if (!class_exists('\\Smalot\\PdfParser\\Parser')) {
5505		- return '';
5506		- }
5507		- $parser = new \Smalot\PdfParser\Parser();
5508		- $pdf = $parser->parseFile($pdf_path);
5509		- $pages = $pdf->getPages();
5510		- $page_texts = array();
5511		- foreach ($pages as $page) {
5512		- $page_text = '';
5513		- try {
5514		- $page_text = $page->getText();
5515		- } catch (\Exception $e) {
5516		- $page_text = '';
5517		- }
5518		- if (!empty($page_text)) {
5519		- $page_texts[] = $page_text;
5520		- }
5521		- }
5522		- $text = trim(implode("\n\n", $page_texts));
5523		- } catch (\Exception $e) {
5524		- error_log('[mxchat] ACF PDF extraction failed for attachment ' . $attachment_id . ': ' . $e->getMessage());
5525		- return '';
5526		- } catch (\Throwable $e) {
5527		- error_log('[mxchat] ACF PDF extraction error for attachment ' . $attachment_id . ': ' . $e->getMessage());
5528		- return '';
5529		- }
5530		-
5531		- // Cap per-PDF text to avoid blowing up the embedding payload on enormous PDFs.
5532		- // The chunker downstream will still split this into multiple vectors.
5533		- $max_len = (int) apply_filters('mxchat_acf_pdf_text_max_length', 50000);
5534		- if ($max_len > 0 && strlen($text) > $max_len) {
5535		- $text = substr($text, 0, $max_len);
5536		- }
5537		-
5538		- update_post_meta($attachment_id, $cache_meta_key, array(
5539		- 'mtime' => (int) $mtime,
5540		- 'text' => $text,
5541		- ));
5542		-
5543		- return $text;
5544		-}
5545		-
5546		-/**
5547		- * Handle ACF save - fires after ACF fields are saved
5548		- * This ensures ACF field data is available when syncing to knowledge base
5549		- */
5550		-public function mxchat_handle_acf_save($post_id) {
5551		- // Skip if not a valid post
5552		- if (!$post_id || $post_id === 'options') {
5553		- return;
5554		- }
5555		-
5556		- // Skip autosaves and revisions
5557		- if (defined('DOING_AUTOSAVE') && DOING_AUTOSAVE || wp_is_post_revision($post_id)) {
5558		- return;
5559		- }
5560		-
5561		- $post = get_post($post_id);
5562		- if (!$post) {
5563		- return;
5564		- }
5565		-
5566		- $post_type = $post->post_type;
5567		-
5568		- // Check if sync is enabled for this post type
5569		- $should_sync = false;
5570		-
5571		- if ($post_type === 'post' && get_option('mxchat_auto_sync_posts') === '1') {
5572		- $should_sync = true;
5573		- } else if ($post_type === 'page' && get_option('mxchat_auto_sync_pages') === '1') {
5574		- $should_sync = true;
5575		- } else if ($post_type === 'product' && class_exists('WooCommerce')) {
5576		- // WooCommerce products - check if WooCommerce integration is enabled
5577		- $options = get_option('mxchat_options', array());
5578		- if (isset($options['enable_woocommerce_integration']) &&
5579		- ($options['enable_woocommerce_integration'] === '1' || $options['enable_woocommerce_integration'] === 'on')) {
5580		- $should_sync = true;
5581		- }
5582		- } else {
5583		- // Check custom post types
5584		- $option_name = 'mxchat_auto_sync_' . $post_type;
5585		- if (get_option($option_name) === '1') {
5586		- $should_sync = true;
5587		- }
5588		- }
5589		-
5590		- if (!$should_sync) {
5591		- return;
5592		- }
5593		-
5594		- // Only process published posts
5595		- if ($post->post_status !== 'publish') {
5596		- return;
5597		- }
5598		-
5599		- // Check if this post has any ACF fields - if not, no need to re-sync
5600		- $acf_fields = $this->mxchat_get_acf_fields_for_post($post_id);
5601		- if (empty($acf_fields)) {
5602		- return;
5603		- }
5604		-
5605		- // Use a transient to prevent duplicate processing (post_updated may have already run)
5606		- $transient_key = 'mxchat_acf_synced_' . $post_id;
5607		- if (get_transient($transient_key)) {
5608		- return;
5609		- }
5610		- set_transient($transient_key, true, 60); // Prevent re-processing for 60 seconds
5611		-
5612		- // Re-run the sync with ACF data now available
5613		- // We pass $update=true since this is effectively an update with ACF data
5614		- $this->mxchat_handle_post_update($post_id, $post, true);
5615		-}
5616		-
5617		-public function mxchat_handle_post_update($post_id, $post, $update) {
5618		- // Basic validation checks
5619		- if (defined('DOING_AUTOSAVE') && DOING_AUTOSAVE || wp_is_post_revision($post_id)) {
5620		- return;
5621		- }
5622		-
5623		- $post_type = $post->post_type;
5624		-
5625		- // Check if sync is enabled for this post type
5626		- $should_sync = false;
5627		-
5628		- // Check built-in post types first
5629		- if ($post_type === 'post' && get_option('mxchat_auto_sync_posts') === '1') {
5630		- $should_sync = true;
5631		- } else if ($post_type === 'page' && get_option('mxchat_auto_sync_pages') === '1') {
5632		- $should_sync = true;
5633		- } else {
5634		- // Check custom post types
5635		- $option_name = 'mxchat_auto_sync_' . $post_type;
5636		- if (get_option($option_name) === '1') {
5637		- $should_sync = true;
5638		- }
5639		- }
5640		-
5641		- if (!$should_sync) {
5642		- return;
5643		- }
5644		-
5645		- // Check if we have stored the previous status and URL in our transients
5646		- $previous_status_key = 'mxchat_prev_status_' . $post_id;
5647		- $previous_status = get_transient($previous_status_key);
5648		-
5649		- $previous_url_key = 'mxchat_prev_url_' . $post_id;
5650		- $previous_url = get_transient($previous_url_key);
5651		-
5652		- // If the post was previously published but is now not published, remove from knowledge base
5653		- if ($previous_status === 'publish' && $post->post_status !== 'publish') {
5654		- // Use the stored URL from when it was published, or fall back to current permalink
5655		- $source_url = $previous_url ?: get_permalink($post_id);
5656		-
5657		- if ($source_url) {
5658		- // Chunk-aware deletion (routes to Pinecone or WP DB and removes base + all chunks)
5659		- MxChat_Utils::delete_chunks_for_url($source_url, 'default');
5660		- }
5661		-
5662		- // Clean up the transients and exit early
5663		- delete_transient($previous_status_key);
5664		- delete_transient($previous_url_key);
5665		- return;
5666		- }
5667		-
5668		- // Slug/permalink rename while still published: delete the old vectors before upserting new ones.
5669		- // Without this, md5(old_url) vectors (base + chunks) would be orphaned under the stale URL.
5670		- if ($post->post_status === 'publish' && !empty($previous_url)) {
5671		- $current_url = get_permalink($post_id);
5672		- if ($current_url && $current_url !== $previous_url) {
5673		- MxChat_Utils::delete_chunks_for_url($previous_url, 'default');
5674		- }
5675		- }
5676		-
5677		- // Store the current status for next time (if this is an update)
5678		- if ($update) {
5679		- set_transient($previous_status_key, $post->post_status, DAY_IN_SECONDS);
5680		-
5681		- // If the post is currently published, also store its URL
5682		- if ($post->post_status === 'publish') {
5683		- $current_url = get_permalink($post_id);
5684		- set_transient($previous_url_key, $current_url, DAY_IN_SECONDS);
5685		- }
5686		- }
5687		-
5688		- // Only process currently published content for adding/updating
5689		- if ($post->post_status === 'publish') {
5690		- // Get the source URL
5691		- $source_url = get_permalink($post_id);
5692		-
5693		- // Get content with proper formatting (matching ajax_mxchat_process_selected_content)
5694		- $title = get_the_title($post_id);
5695		- $content = get_post_field('post_content', $post_id);
5696		- $excerpt = get_post_field('post_excerpt', $post_id);
5697		-
5698		- // Remove shortcode tags but preserve content inside them
5699		- $content = $this->strip_shortcode_tags_preserve_content($content);
5700		- $excerpt = $this->strip_shortcode_tags_preserve_content($excerpt);
5701		-
5702		- // Strip tags but preserve structure (don't use 'the_content' filter as it may re-add shortcodes)
5703		- $content = wp_strip_all_tags($content);
5704		-
5705		- // Combine title, short description (if exists), and content
5706		- $final_content = $title . "\n\n";
5707		-
5708		- // Add short description if it exists (WooCommerce products use post_excerpt for short description)
5709		- if (!empty($excerpt)) {
5710		- $final_content .= "Short Description: " . wp_strip_all_tags($excerpt) . "\n\n";
5711		- }
5712		-
5713		- $final_content .= $content;
5714		-
5715		- // For WooCommerce products, include pricing and product details
5716		- if ($post_type === 'product' && class_exists('WooCommerce')) {
5717		- $product = wc_get_product($post_id);
5718		-
5719		- if ($product) {
5720		- // Get pricing information
5721		- $regular_price = $product->get_regular_price();
5722		- $sale_price = $product->get_sale_price();
5723		- $price = $product->get_price();
5724		- $sku = $product->get_sku();
5725		-
5726		- // Get currency symbol
5727		- $currency_symbol = get_woocommerce_currency_symbol();
5728		-
5729		- // Add pricing information
5730		- $final_content .= "\n";
5731		- if (!empty($regular_price)) {
5732		- $final_content .= "Price: " . $currency_symbol . $regular_price . "\n";
5733		- } elseif (!empty($price)) {
5734		- $final_content .= "Price: " . $currency_symbol . $price . "\n";
5735		- }
5736		-
5737		- if (!empty($sale_price) && $sale_price !== $regular_price) {
5738		- $final_content .= "Sale Price: " . $currency_symbol . $sale_price . "\n";
5739		- }
5740		-
5741		- // Handle variable products - show price range
5742		- if ($product->is_type('variable')) {
5743		- $min_price = $product->get_variation_price('min');
5744		- $max_price = $product->get_variation_price('max');
5745		- if ($min_price !== $max_price) {
5746		- $final_content .= "Price Range: " . $currency_symbol . $min_price . " - " . $currency_symbol . $max_price . "\n";
5747		- }
5748		- }
5749		-
5750		- if (!empty($sku)) {
5751		- $final_content .= "SKU: " . $sku . "\n";
5752		- }
5753		-
5754		- // Get product categories
5755		- $categories = wp_get_post_terms($post_id, 'product_cat', array('fields' => 'names'));
5756		- if (!empty($categories) && !is_wp_error($categories)) {
5757		- $final_content .= "Categories: " . implode(', ', $categories) . "\n";
5758		- }
5759		- }
5760		- }
5761		-
5762		- // For custom post types like job_listing, include additional fields
5763		- if ($post_type === 'job_listing') {
5764		- // Add job-specific meta if available
5765		- $job_location = get_post_meta($post_id, '_job_location', true);
5766		- if (!empty($job_location)) {
5767		- $final_content .= "\n\nLocation: " . $job_location;
5768		- }
5769		-
5770		- // Get job type terms
5771		- $job_types = get_the_terms($post_id, 'job_listing_type');
5772		- if (!empty($job_types) && !is_wp_error($job_types)) {
5773		- $types = array();
5774		- foreach ($job_types as $type) {
5775		- $types[] = $type->name;
5776		- }
5777		- $final_content .= "\n\nJob Type: " . implode(', ', $types);
5778		- }
5779		-
5780		- // Get company name if available
5781		- $company_name = get_post_meta($post_id, '_company_name', true);
5782		- if (!empty($company_name)) {
5783		- $final_content .= "\n\nCompany: " . $company_name;
5784		- }
5785		- }
5786		-
5787		- // ADD ACF FIELDS SUPPORT (matches ajax_mxchat_process_selected_content behavior)
5788		- $acf_fields = $this->mxchat_get_acf_fields_for_post($post_id);
5789		- if (!empty($acf_fields)) {
5790		- $acf_content_parts = array();
5791		- $pdf_attachment_ids = array();
5792		-
5793		- foreach ($acf_fields as $field_name => $field_value) {
5794		- $formatted_value = $this->mxchat_format_acf_field_value($field_value, $field_name, $post_id);
5795		- if (!empty($formatted_value)) {
5796		- // Convert field name to readable label
5797		- $field_label = ucwords(str_replace(['_', '-'], ' ', $field_name));
5798		- $acf_content_parts[] = $field_label . ": " . $formatted_value;
5799		- }
5800		-
5801		- $this->mxchat_collect_pdf_attachment_ids_from_acf_value($field_value, $pdf_attachment_ids);
5802		- }
5803		-
5804		- if (!empty($acf_content_parts)) {
5805		- $final_content .= "\n\n" . implode("\n", $acf_content_parts);
5806		- }
5807		-
5808		- // Gate the auto-sync PDF-extraction loop behind an opt-in option.
5809		- // Mirrors the per-batch checkbox the manual content selector has; the
5810		- // 25 MB size cap lives in the shared extractor so it applies in both
5811		- // paths regardless. Default OFF — re-parsing every ACF PDF on every
5812		- // editor save is expensive and most sites don't want it.
5813		- $autosync_extract_acf_pdfs = get_option('mxchat_auto_sync_acf_pdfs', '0') === '1';
5814		- if ($autosync_extract_acf_pdfs && !empty($pdf_attachment_ids)) {
5815		- $pdf_attachment_ids = array_unique(array_filter(array_map('intval', $pdf_attachment_ids)));
5816		- $pdf_sections = array();
5817		- foreach ($pdf_attachment_ids as $att_id) {
5818		- $pdf_text = $this->mxchat_extract_pdf_text_by_attachment_id($att_id);
5819		- if (!empty($pdf_text)) {
5820		- $pdf_title = get_the_title($att_id);
5821		- $pdf_url = wp_get_attachment_url($att_id);
5822		- $header = 'PDF Attachment';
5823		- if (!empty($pdf_title)) {
5824		- $header .= ': ' . $pdf_title;
5825		- }
5826		- if (!empty($pdf_url)) {
5827		- $header .= ' (' . $pdf_url . ')';
5828		- }
5829		- $pdf_sections[] = $header . "\n" . $pdf_text;
5830		- }
5831		- }
5832		- if (!empty($pdf_sections)) {
5833		- $final_content .= "\n\nAttached PDFs:\n" . implode("\n\n", $pdf_sections);
5834		- }
5835		- }
5836		- }
5837		-
5838		- // ADD CUSTOM POST META SUPPORT (whitelisted non-ACF meta fields)
5839		- $custom_meta = $this->mxchat_get_whitelisted_post_meta($post_id);
5840		- if (!empty($custom_meta)) {
5841		- $meta_content_parts = array();
5842		-
5843		- foreach ($custom_meta as $meta_key => $meta_value) {
5844		- // Convert meta key to readable label
5845		- $meta_label = ucwords(str_replace(array('_', '-'), ' ', $meta_key));
5846		- $meta_content_parts[] = $meta_label . ": " . $meta_value;
5847		- }
5848		-
5849		- if (!empty($meta_content_parts)) {
5850		- $final_content .= "\n\n" . implode("\n", $meta_content_parts);
5851		- }
5852		- }
5853		-
5854		- // Get API key with proper model detection
5855		- $options = get_option('mxchat_options');
5856		- $selected_model = $options['embedding_model'] ?? 'text-embedding-ada-002';
5857		-
5858		- if (strpos($selected_model, 'voyage') === 0) {
5859		- $api_key = $options['voyage_api_key'] ?? '';
5860		- } elseif (strpos($selected_model, 'gemini-embedding') === 0) {
5861		- $api_key = $options['gemini_api_key'] ?? '';
5862		- } else {
5863		- $api_key = $options['api_key'] ?? '';
5864		- }
5865		-
5866		- if (empty($api_key)) {
5867		- return;
5868		- }
5869		-
5870		- // Use the centralized utility function for storage
5871		- $result = MxChat_Utils::submit_content_to_db(
5872		- $final_content,
5873		- $source_url,
5874		- $api_key,
5875		- md5($source_url) // Vector ID for Pinecone
5876		- );
5877		-
5878		- // After successful storage, apply role restriction based on tags
5879		- if (!is_wp_error($result)) {
5880		- $this->apply_role_restriction_to_post($post_id, $source_url);
5881		- }
5882		- }
5883		-
5884		- // Clean up the stored previous status if not used above
5885		- if ($previous_status !== 'publish' || $post->post_status === 'publish') {
5886		- delete_transient($previous_status_key);
5887		- delete_transient($previous_url_key);
5888		- }
5889		-}
5890		-
5891		-/**
5892		- * Store the post status and URL before update to detect status transitions
5893		- * This runs before the post is actually updated in the database
5894		- */
5895		-public function mxchat_store_pre_update_status($post_id, $data) {
5896		- // Get the current post from database (before update)
5897		- $current_post = get_post($post_id);
5898		-
5899		- if ($current_post) {
5900		- // Store the current status temporarily
5901		- $status_key = 'mxchat_prev_status_' . $post_id;
5902		- set_transient($status_key, $current_post->post_status, HOUR_IN_SECONDS);
5903		-
5904		- // If the post is currently published, also store its URL
5905		- if ($current_post->post_status === 'publish') {
5906		- $url_key = 'mxchat_prev_url_' . $post_id;
5907		- $current_url = get_permalink($post_id);
5908		- set_transient($url_key, $current_url, HOUR_IN_SECONDS);
5909		- }
5910		- }
5911		-}
5912		-
5913		-public function mxchat_handle_post_delete($post_id) {
5914		- // Get post data before it's deleted
5915		- $post = get_post($post_id);
5916		-
5917		- // Basic validation
5918		- if (!$post || wp_is_post_revision($post_id)) {
5919		- return;
5920		- }
5921		-
5922		- $post_type = $post->post_type;
5923		-
5924		- // Check if sync is enabled for this post type
5925		- $should_sync = false;
5926		-
5927		- // Check built-in post types first
5928		- if ($post_type === 'post' && get_option('mxchat_auto_sync_posts') === '1') {
5929		- $should_sync = true;
5930		- } else if ($post_type === 'page' && get_option('mxchat_auto_sync_pages') === '1') {
5931		- $should_sync = true;
5932		- } else {
5933		- // Check custom post types
5934		- $option_name = 'mxchat_auto_sync_' . $post_type;
5935		- if (get_option($option_name) === '1') {
5936		- $should_sync = true;
5937		- }
5938		- }
5939		-
5940		- if (!$should_sync) {
5941		- return;
5942		- }
5943		-
5944		- // Resolve the pre-trash URL. wp_trash_post renames the slug with "__trashed" before firing
5945		- // this hook, so get_permalink() here would return the trashed URL and md5() would miss the
5946		- // real vector IDs stored under the original URL.
5947		- $source_url = $this->mxchat_resolve_pre_trash_url($post_id);
5948		- if (!$source_url) {
5949		- //error_log('MXChat: Failed to resolve source URL for post ' . $post_id);
5950		- return;
5951		- }
5952		-
5953		- // Use chunk-aware deletion (handles both chunked and non-chunked content)
5954		- $delete_result = MxChat_Utils::delete_chunks_for_url($source_url, 'default');
5955		-
5956		- if (is_wp_error($delete_result)) {
5957		- //error_log('MXChat: Chunk-aware deletion failed for URL: ' . $source_url . ' - ' . $delete_result->get_error_message());
5958		- }
5959		-
5960		- delete_transient('mxchat_prev_url_' . $post_id);
5961		- delete_transient('mxchat_prev_status_' . $post_id);
5962		-}
5963		-
5964		-/**
5965		- * Resolve the source URL for a post being trashed/deleted.
5966		- *
5967		- * Why: wp_trash_post appends "__trashed" to the slug before the wp_trash_post action fires, so
5968		- * get_permalink() returns a URL whose md5() won't match the vector IDs stored in Pinecone or
5969		- * the source_url rows in the WP DB. Prefer the URL captured by mxchat_store_pre_update_status
5970		- * (runs on pre_post_update, before the rename); fall back to stripping the __trashed suffix.
5971		- */
5972		-private function mxchat_resolve_pre_trash_url($post_id) {
5973		- $previous_url = get_transient('mxchat_prev_url_' . $post_id);
5974		- if (!empty($previous_url)) {
5975		- return $previous_url;
5976		- }
5977		-
5978		- $current = get_permalink($post_id);
5979		- if (!$current) {
5980		- return '';
5981		- }
5982		- return preg_replace('#__trashed(/?)$#', '$1', $current);
5983		-}
5984		-
5985		-
5986		-
5987		-public function mxchat_handle_product_change($post_id, $post, $update) {
5988		- if ($post->post_type !== 'product') {
5989		- return;
5990		- }
5991		-
5992		- if ($post->post_status === 'publish') {
5993		- add_action('shutdown', function() use ($post_id) {
5994		- $product = wc_get_product($post_id);
5995		- if ($product) {
5996		- $this->mxchat_store_product_embedding($product);
5997		- }
5998		- });
5999		- }
6000		-}
6001		-
6002		-/**
6003		- * Store WooCommerce product embeddings
6004		- */
6005		-private function mxchat_store_product_embedding($product) {
6006		- if (!isset($this->options['enable_woocommerce_integration']) ||
6007		- !in_array($this->options['enable_woocommerce_integration'], ['1', 'on'])) {
6008		- return;
6009		- }
6010		-
6011		- $source_url = get_permalink($product->get_id());
6012		- $product_id = $product->get_id();
6013		-
6014		- // Build product content
6015		- $title = $product->get_name();
6016		- $description = $product->get_description();
6017		- $short_description = $product->get_short_description();
6018		- $regular_price = $product->get_regular_price();
6019		- $sale_price = $product->get_sale_price();
6020		- $price = $product->get_price();
6021		- $sku = $product->get_sku();
6022		-
6023		- // Get currency symbol
6024		- $currency_symbol = get_woocommerce_currency_symbol();
6025		-
6026		- // Format content consistently
6027		- $content = $title . "\n\n";
6028		-
6029		- if (!empty($short_description)) {
6030		- $content .= "Short Description: " . wp_strip_all_tags($short_description) . "\n\n";
6031		- }
6032		-
6033		- if (!empty($description)) {
6034		- $content .= wp_strip_all_tags($description) . "\n\n";
6035		- }
6036		-
6037		- // Add pricing information
6038		- if (!empty($regular_price)) {
6039		- $content .= "Price: " . $currency_symbol . $regular_price . "\n";
6040		- } elseif (!empty($price)) {
6041		- $content .= "Price: " . $currency_symbol . $price . "\n";
6042		- }
6043		-
6044		- if (!empty($sale_price) && $sale_price !== $regular_price) {
6045		- $content .= "Sale Price: " . $currency_symbol . $sale_price . "\n";
6046		- }
6047		-
6048		- // Handle variable products - show price range
6049		- if ($product->is_type('variable')) {
6050		- $min_price = $product->get_variation_price('min');
6051		- $max_price = $product->get_variation_price('max');
6052		- if ($min_price !== $max_price) {
6053		- $content .= "Price Range: " . $currency_symbol . $min_price . " - " . $currency_symbol . $max_price . "\n";
6054		- }
6055		- }
6056		-
6057		- if (!empty($sku)) {
6058		- $content .= "SKU: " . $sku . "\n";
6059		- }
6060		-
6061		- // Get product categories
6062		- $categories = wp_get_post_terms($product_id, 'product_cat', array('fields' => 'names'));
6063		- if (!empty($categories) && !is_wp_error($categories)) {
6064		- $content .= "Categories: " . implode(', ', $categories) . "\n";
6065		- }
6066		-
6067		- // Get Custom Product Tabs (supports "Custom Product Tabs for WooCommerce" by Code Parrots)
6068		- $custom_tabs = get_post_meta($product_id, 'yikes_woo_products_tabs', true);
6069		- if (!empty($custom_tabs) && is_array($custom_tabs)) {
6070		- foreach ($custom_tabs as $tab) {
6071		- $tab_title = isset($tab['title']) ? $tab['title'] : (isset($tab['tab_title']) ? $tab['tab_title'] : '');
6072		- $tab_content = isset($tab['content']) ? $tab['content'] : '';
6073		-
6074		- if (!empty($tab_title) && !empty($tab_content)) {
6075		- $content .= "\n" . $tab_title . ": " . wp_strip_all_tags($tab_content) . "\n";
6076		- }
6077		- }
6078		- }
6079		-
6080		- // Also check for reusable/saved tabs applied to this product
6081		- $applied_saved_tabs = get_post_meta($product_id, 'yikes_woo_reusable_products_tabs_applied', true);
6082		- if (!empty($applied_saved_tabs) && is_array($applied_saved_tabs)) {
6083		- // Get the saved tabs option
6084		- $saved_tabs = get_option('yikes_woo_reusable_products_tabs', array());
6085		- if (!empty($saved_tabs) && is_array($saved_tabs)) {
6086		- foreach ($applied_saved_tabs as $saved_tab_id) {
6087		- if (isset($saved_tabs[$saved_tab_id])) {
6088		- $tab = $saved_tabs[$saved_tab_id];
6089		- $tab_title = isset($tab['title']) ? $tab['title'] : (isset($tab['tab_title']) ? $tab['tab_title'] : '');
6090		- $tab_content = isset($tab['content']) ? $tab['content'] : '';
6091		-
6092		- if (!empty($tab_title) && !empty($tab_content)) {
6093		- $content .= "\n" . $tab_title . ": " . wp_strip_all_tags($tab_content) . "\n";
6094		- }
6095		- }
6096		- }
6097		- }
6098		- }
6099		-
6100		- // Get API key with proper model detection
6101		- $options = get_option('mxchat_options');
6102		- $selected_model = $options['embedding_model'] ?? 'text-embedding-ada-002';
6103		-
6104		- if (strpos($selected_model, 'voyage') === 0) {
6105		- $api_key = $options['voyage_api_key'] ?? '';
6106		- } elseif (strpos($selected_model, 'gemini-embedding') === 0) {
6107		- $api_key = $options['gemini_api_key'] ?? '';
6108		- } else {
6109		- $api_key = $options['api_key'] ?? '';
6110		- }
6111		-
6112		- if (empty($api_key)) {
6113		- //error_log('MxChat Auto-sync: No API key configured for embedding model');
6114		- return;
6115		- }
6116		-
6117		- // Use the centralized utility function for storage
6118		- $result = MxChat_Utils::submit_content_to_db(
6119		- $content,
6120		- $source_url,
6121		- $api_key,
6122		- md5($source_url) // Vector ID for Pinecone
6123		- );
6124		-
6125		- // After successful storage, apply role restriction based on tags
6126		- if (!is_wp_error($result)) {
6127		- $this->apply_role_restriction_to_post($product_id, $source_url);
6128		- }
6129		-
6130		- if (is_wp_error($result)) {
6131		- //error_log('MxChat WooCommerce sync failed for product ' . $product_id . ': ' . $result->get_error_message());
6132		- }
6133		-}
6134		-
6135		-public function mxchat_handle_product_delete($post_id) {
6136		- if (get_post_type($post_id) !== 'product') {
6137		- return;
6138		- }
6139		-
6140		- $source_url = $this->mxchat_resolve_pre_trash_url($post_id);
6141		- if (!$source_url) {
6142		- return;
6143		- }
6144		-
6145		- // Chunk-aware deletion (routes to Pinecone or WP DB and removes base + all chunks)
6146		- MxChat_Utils::delete_chunks_for_url($source_url, 'default');
6147		-
6148		- delete_transient('mxchat_prev_url_' . $post_id);
6149		- delete_transient('mxchat_prev_status_' . $post_id);
6150		-}
6151		-
6152		-/**
6153		- * Handle individual Pinecone content deletion
6154		- */
6155		-public function mxchat_handle_pinecone_prompt_delete() {
6156		- // Check permissions
6157		- if (!current_user_can('manage_options')) {
6158		- wp_die(esc_html__('You do not have sufficient permissions.', 'mxchat'));
6159		- }
6160		-
6161		- // Verify nonce
6162		- if (!isset($_GET['_wpnonce']) || !wp_verify_nonce($_GET['_wpnonce'], 'mxchat_delete_pinecone_prompt_nonce')) {
6163		- wp_die(esc_html__('Security check failed.', 'mxchat'));
6164		- }
6165		-
6166		- $vector_id = isset($_GET['vector_id']) ? sanitize_text_field($_GET['vector_id']) : '';
6167		-
6168		- if (empty($vector_id)) {
6169		- set_transient('mxchat_admin_notice_error',
6170		- esc_html__('Invalid vector ID.', 'mxchat'),
6171		- 30
6172		- );
6173		- wp_safe_redirect(admin_url('admin.php?page=mxchat-prompts'));
6174		- exit;
6175		- }
6176		-
6177		- // Get Pinecone settings
6178		- $pinecone_options = get_option('mxchat_pinecone_addon_options', array());
6179		- $use_pinecone = ($pinecone_options['mxchat_use_pinecone'] ?? '0') === '1';
6180		-
6181		- if (!$use_pinecone || empty($pinecone_options['mxchat_pinecone_api_key'])) {
6182		- set_transient('mxchat_admin_notice_error',
6183		- esc_html__('Pinecone is not properly configured.', 'mxchat'),
6184		- 30
6185		- );
6186		- wp_safe_redirect(admin_url('admin.php?page=mxchat-prompts'));
6187		- exit;
6188		- }
6189		-
6190		- // Delete from Pinecone
6191		- $pinecone_manager = MxChat_Pinecone_Manager::get_instance();
6192		- $result = $pinecone_manager->mxchat_delete_from_pinecone_by_vector_id(
6193		- $vector_id,
6194		- $pinecone_options['mxchat_pinecone_api_key'],
6195		- $pinecone_options['mxchat_pinecone_host']
6196		- );
6197		-
6198		- if ($result['success']) {
6199		- // No cache clearing needed since we removed caching
6200		- set_transient('mxchat_admin_notice_success',
6201		- esc_html__('Entry deleted successfully from Pinecone.', 'mxchat'),
6202		- 30
6203		- );
6204		- } else {
6205		- set_transient('mxchat_admin_notice_error',
6206		- esc_html__('Failed to delete entry: ', 'mxchat') . $result['message'],
6207		- 30
6208		- );
6209		- }
6210		-
6211		- wp_safe_redirect(admin_url('admin.php?page=mxchat-prompts'));
6212		- exit;
6213		-}
6214		-/**
6215		- * Handle individual Pinecone content deletion via AJAX
6216		- */
6217		-public function ajax_mxchat_delete_pinecone_prompt() {
6218		- // Verify nonce and permissions
6219		- if (!check_ajax_referer('mxchat_delete_pinecone_prompt_nonce', 'nonce', false)) {
6220		- wp_send_json_error('Invalid nonce');
6221		- exit;
6222		- }
6223		-
6224		- if (!current_user_can('manage_options')) {
6225		- wp_send_json_error('Unauthorized access');
6226		- exit;
6227		- }
6228		-
6229		- $vector_id = isset($_POST['vector_id']) ? sanitize_text_field($_POST['vector_id']) : '';
6230		- $bot_id = isset($_POST['bot_id']) ? sanitize_text_field($_POST['bot_id']) : 'default';
6231		-
6232		- if (empty($vector_id)) {
6233		- wp_send_json_error('Missing vector ID');
6234		- exit;
6235		- }
6236		-
6237		- // Get bot-specific Pinecone settings
6238		- $pinecone_manager = MxChat_Pinecone_Manager::get_instance();
6239		- $pinecone_options = $pinecone_manager->mxchat_get_bot_pinecone_options($bot_id);
6240		-
6241		- $use_pinecone = ($pinecone_options['mxchat_use_pinecone'] ?? '0') === '1';
6242		-
6243		- if (!$use_pinecone || empty($pinecone_options['mxchat_pinecone_api_key'])) {
6244		- wp_send_json_error('Pinecone is not properly configured for bot: ' . $bot_id);
6245		- exit;
6246		- }
6247		-
6248		- // Delete from the correct Pinecone index
6249		- $result = $pinecone_manager->mxchat_delete_from_pinecone_by_vector_id(
6250		- $vector_id,
6251		- $pinecone_options['mxchat_pinecone_api_key'],
6252		- $pinecone_options['mxchat_pinecone_host']
6253		- );
6254		-
6255		- if ($result['success']) {
6256		- // No cache clearing needed since we removed caching
6257		- wp_send_json_success(array(
6258		- 'message' => 'Entry deleted successfully from Pinecone',
6259		- 'vector_id' => $vector_id,
6260		- 'bot_id' => $bot_id
6261		- ));
6262		- } else {
6263		- MxChat_Admin::mxchat_log_debug('pinecone_error', 'Failed to delete from Pinecone: ' . $result['message'], array('vector_id' => $vector_id, 'bot_id' => $bot_id));
6264		- wp_send_json_error('Failed to delete from Pinecone: ' . $result['message']);
6265		- }
6266		-
6267		- exit;
6268		-}
6269		-
6270		-/**
6271		- * Handle deletion of all chunks for a given source URL via AJAX
6272		- * Follows the same pattern as ajax_mxchat_delete_pinecone_prompt
6273		- */
6274		-public function ajax_mxchat_delete_chunks_by_url() {
6275		- // Verify nonce and permissions
6276		- if (!check_ajax_referer('mxchat_delete_chunks_nonce', 'nonce', false)) {
6277		- wp_send_json_error('Invalid nonce');
6278		- exit;
6279		- }
6280		-
6281		- if (!current_user_can('manage_options')) {
6282		- wp_send_json_error('Unauthorized access');
6283		- exit;
6284		- }
6285		-
6286		- $source_url = isset($_POST['source_url']) ? esc_url_raw($_POST['source_url']) : '';
6287		- $data_source = isset($_POST['data_source']) ? sanitize_text_field($_POST['data_source']) : 'wordpress';
6288		- $bot_id = isset($_POST['bot_id']) ? sanitize_text_field($_POST['bot_id']) : 'default';
6289		-
6290		- if (empty($source_url)) {
6291		- wp_send_json_error('Missing source URL');
6292		- exit;
6293		- }
6294		-
6295		- // Generate the base vector ID from the source URL (same as how chunks are created)
6296		- $base_vector_id = md5($source_url);
6297		-
6298		- if ($data_source === 'pinecone') {
6299		- // Get bot-specific Pinecone settings (same as working delete function)
6300		- $pinecone_manager = MxChat_Pinecone_Manager::get_instance();
6301		- $pinecone_options = $pinecone_manager->mxchat_get_bot_pinecone_options($bot_id);
6302		-
6303		- $use_pinecone = ($pinecone_options['mxchat_use_pinecone'] ?? '0') === '1';
6304		-
6305		- if (!$use_pinecone || empty($pinecone_options['mxchat_pinecone_api_key'])) {
6306		- wp_send_json_error('Pinecone is not properly configured for bot: ' . $bot_id);
6307		- exit;
6308		- }
6309		-
6310		- $api_key = $pinecone_options['mxchat_pinecone_api_key'];
6311		- $host = $pinecone_options['mxchat_pinecone_host'];
6312		- $namespace = $pinecone_options['mxchat_pinecone_namespace'] ?? '';
6313		-
6314		- // Collect all vector IDs to delete
6315		- $vectors_to_delete = array();
6316		-
6317		- // Add the original single-vector ID (for non-chunked content)
6318		- $vectors_to_delete[] = $base_vector_id;
6319		-
6320		- // Use Pinecone list API to find all chunk vectors with this prefix
6321		- // NOTE: Pinecone List API is a GET request with query parameters, not POST
6322		- $prefix = $base_vector_id . '_chunk_';
6323		-
6324		- $query_params = array(
6325		- 'prefix' => $prefix,
6326		- 'limit' => 100
6327		- );
6328		-
6329		- if (!empty($namespace)) {
6330		- $query_params['namespace'] = $namespace;
6331		- }
6332		-
6333		- $list_url = "https://{$host}/vectors/list?" . http_build_query($query_params);
6334		-
6335		- $list_response = wp_remote_get($list_url, array(
6336		- 'headers' => array(
6337		- 'Api-Key' => $api_key,
6338		- 'accept' => 'application/json'
6339		- ),
6340		- 'timeout' => 30
6341		- ));
6342		-
6343		- if (!is_wp_error($list_response)) {
6344		- $list_body_response = wp_remote_retrieve_body($list_response);
6345		- $list_data = json_decode($list_body_response, true);
6346		- if (!empty($list_data['vectors'])) {
6347		- foreach ($list_data['vectors'] as $vector) {
6348		- if (isset($vector['id'])) {
6349		- $vectors_to_delete[] = $vector['id'];
6350		- }
6351		- }
6352		- }
6353		- }
6354		-
6355		- if (empty($vectors_to_delete)) {
6356		- wp_send_json_success(array(
6357		- 'message' => 'No vectors found to delete',
6358		- 'source_url' => $source_url
6359		- ));
6360		- exit;
6361		- }
6362		-
6363		- // Delete all vectors using the same endpoint as the working function
6364		- $delete_url = "https://{$host}/vectors/delete";
6365		-
6366		- $delete_body = array(
6367		- 'ids' => $vectors_to_delete
6368		- );
6369		-
6370		- if (!empty($namespace)) {
6371		- $delete_body['namespace'] = $namespace;
6372		- }
6373		-
6374		- $delete_response = wp_remote_post($delete_url, array(
6375		- 'headers' => array(
6376		- 'Api-Key' => $api_key,
6377		- 'accept' => 'application/json',
6378		- 'content-type' => 'application/json'
6379		- ),
6380		- 'body' => wp_json_encode($delete_body),
6381		- 'timeout' => 30
6382		- ));
6383		-
6384		- if (is_wp_error($delete_response)) {
6385		- MxChat_Admin::mxchat_log_debug('pinecone_error', 'Failed to delete chunks from Pinecone: ' . $delete_response->get_error_message(), array('source_url' => $source_url));
6386		- wp_send_json_error('Failed to delete from Pinecone: ' . $delete_response->get_error_message());
6387		- exit;
6388		- }
6389		-
6390		- $response_code = wp_remote_retrieve_response_code($delete_response);
6391		-
6392		- if ($response_code !== 200) {
6393		- MxChat_Admin::mxchat_log_debug('pinecone_error', 'Pinecone API error (HTTP ' . $response_code . ')', array('source_url' => $source_url));
6394		- wp_send_json_error('Pinecone API error (HTTP ' . $response_code . ')');
6395		- exit;
6396		- }
6397		-
6398		- wp_send_json_success(array(
6399		- 'message' => 'All chunks deleted successfully from Pinecone',
6400		- 'source_url' => $source_url,
6401		- 'deleted_count' => count($vectors_to_delete)
6402		- ));
6403		-
6404		- } else {
6405		- // WordPress database deletion
6406		- global $wpdb;
6407		- $table_name = $wpdb->prefix . 'mxchat_system_prompt_content';
6408		-
6409		- $result = $wpdb->delete(
6410		- $table_name,
6411		- array('source_url' => $source_url),
6412		- array('%s')
6413		- );
6414		-
6415		- if ($result === false) {
6416		- MxChat_Admin::mxchat_log_debug('database_error', 'Failed to delete from database: ' . $wpdb->last_error, array('source_url' => $source_url));
6417		- wp_send_json_error('Failed to delete from database: ' . $wpdb->last_error);
6418		- exit;
6419		- }
6420		-
6421		- wp_send_json_success(array(
6422		- 'message' => 'All chunks deleted successfully from database',
6423		- 'source_url' => $source_url,
6424		- 'deleted_count' => $result
6425		- ));
6426		- }
6427		-
6428		- exit;
6429		-}
6430		-
6431		-/**
6432		- * Handle individual WordPress database content deletion via AJAX
6433		- * Mirrors the Pinecone delete handler but for WordPress database entries
6434		- */
6435		-public function ajax_mxchat_delete_wordpress_prompt() {
6436		- // Verify nonce and permissions
6437		- if (!check_ajax_referer('mxchat_delete_wordpress_prompt_nonce', 'nonce', false)) {
6438		- wp_send_json_error('Invalid nonce');
6439		- exit;
6440		- }
6441		-
6442		- if (!current_user_can('manage_options')) {
6443		- wp_send_json_error('Unauthorized access');
6444		- exit;
6445		- }
6446		-
6447		- $entry_id = isset($_POST['entry_id']) ? intval($_POST['entry_id']) : 0;
6448		-
6449		- if (empty($entry_id)) {
6450		- wp_send_json_error('Missing entry ID');
6451		- exit;
6452		- }
6453		-
6454		- global $wpdb;
6455		- $table_name = $wpdb->prefix . 'mxchat_system_prompt_content';
6456		-
6457		- // Clear cache for this entry
6458		- wp_cache_delete('prompt_' . $entry_id, 'mxchat_prompts');
6459		-
6460		- // Delete from database
6461		- $result = $wpdb->delete(
6462		- $table_name,
6463		- array('id' => $entry_id),
6464		- array('%d')
6465		- );
6466		-
6467		- if ($result !== false) {
6468		- wp_send_json_success(array(
6469		- 'message' => 'Entry deleted successfully',
6470		- 'entry_id' => $entry_id
6471		- ));
6472		- } else {
6473		- wp_send_json_error('Failed to delete entry from database');
6474		- }
6475		-
6476		- exit;
6477		-}
6478		-
6479		-/**
6480		- * Handle bulk deletion of knowledge entries via AJAX
6481		- * Supports both Pinecone and WordPress database entries
6482		- */
6483		-public function ajax_mxchat_bulk_delete_knowledge() {
6484		- // Verify nonce and permissions
6485		- if (!check_ajax_referer('mxchat_bulk_delete_knowledge_nonce', 'nonce', false)) {
6486		- wp_send_json_error('Invalid nonce');
6487		- exit;
6488		- }
6489		-
6490		- if (!current_user_can('manage_options')) {
6491		- wp_send_json_error('Unauthorized access');
6492		- exit;
6493		- }
6494		-
6495		- $entries = isset($_POST['entries']) ? $_POST['entries'] : array();
6496		- $bot_id = isset($_POST['bot_id']) ? sanitize_text_field($_POST['bot_id']) : 'default';
6497		-
6498		- if (empty($entries) || !is_array($entries)) {
6499		- wp_send_json_error('No entries provided');
6500		- exit;
6501		- }
6502		-
6503		- // Extend execution time — bulk Pinecone operations can take a while
6504		- if (function_exists('set_time_limit')) {
6505		- set_time_limit(120);
6506		- }
6507		-
6508		- $success_ids = array();
6509		- $failed_ids = array();
6510		- $errors = array();
6511		-
6512		- global $wpdb;
6513		- $table_name = $wpdb->prefix . 'mxchat_system_prompt_content';
6514		-
6515		- // Get Pinecone manager for Pinecone deletions
6516		- $pinecone_manager = MxChat_Pinecone_Manager::get_instance();
6517		- $pinecone_options = $pinecone_manager->mxchat_get_bot_pinecone_options($bot_id);
6518		- $use_pinecone = ($pinecone_options['mxchat_use_pinecone'] ?? '0') === '1';
6519		-
6520		- $api_key = $pinecone_options['mxchat_pinecone_api_key'] ?? '';
6521		- $host = $pinecone_options['mxchat_pinecone_host'] ?? '';
6522		-
6523		- // =============================================
6524		- // PHASE 1: Collect all Pinecone vector IDs
6525		- // and separate WordPress entries
6526		- // =============================================
6527		- $pinecone_entry_ids = array(); // entry IDs that are pinecone-sourced
6528		- $wordpress_entries = array(); // entries for WordPress DB deletion
6529		- $all_vector_ids = array(); // all pinecone vector IDs to delete in one batch
6530		-
6531		- foreach ($entries as $entry) {
6532		- $entry_id = sanitize_text_field($entry['id'] ?? '');
6533		- $source = sanitize_text_field($entry['source'] ?? 'wordpress');
6534		- $source_url = isset($entry['sourceUrl']) ? esc_url_raw($entry['sourceUrl']) : '';
6535		- $is_group = isset($entry['isGroup']) && ($entry['isGroup'] === true || $entry['isGroup'] === 'true');
6536		-
6537		- if (empty($entry_id)) {
6538		- continue;
6539		- }
6540		-
6541		- if ($source === 'pinecone') {
6542		- if (!$use_pinecone || empty($api_key)) {
6543		- $failed_ids[] = $entry_id;
6544		- $errors[] = "Pinecone not configured for entry: $entry_id";
6545		- continue;
6546		- }
6547		-
6548		- $pinecone_entry_ids[] = $entry_id;
6549		-
6550		- if ($is_group && !empty($source_url)) {
6551		- // Grouped/chunked entry: collect base ID + chunk IDs via List API
6552		- $base_vector_id = md5($source_url);
6553		- $all_vector_ids[] = $base_vector_id;
6554		-
6555		- $list_url = 'https://' . $host . '/vectors/list?prefix=' . urlencode($base_vector_id . '_chunk_') . '&limit=100';
6556		- $list_response = wp_remote_get($list_url, array(
6557		- 'headers' => array(
6558		- 'Api-Key' => $api_key,
6559		- 'accept' => 'application/json'
6560		- ),
6561		- 'timeout' => 30
6562		- ));
6563		-
6564		- if (!is_wp_error($list_response)) {
6565		- $list_body = json_decode(wp_remote_retrieve_body($list_response), true);
6566		- if (!empty($list_body['vectors']) && is_array($list_body['vectors'])) {
6567		- foreach ($list_body['vectors'] as $vector) {
6568		- if (isset($vector['id'])) {
6569		- $all_vector_ids[] = $vector['id'];
6570		- }
6571		- }
6572		- }
6573		- }
6574		- } else {
6575		- // Single entry: the entry_id IS the vector ID
6576		- $all_vector_ids[] = $entry_id;
6577		- }
6578		- } else {
6579		- $wordpress_entries[] = $entry;
6580		- }
6581		- }
6582		-
6583		- // =============================================
6584		- // PHASE 2: Single batch delete to Pinecone
6585		- // =============================================
6586		- if (!empty($all_vector_ids)) {
6587		- $all_vector_ids = array_values(array_unique($all_vector_ids));
6588		- $pinecone_success = true;
6589		- $batches = array_chunk($all_vector_ids, 100);
6590		-
6591		- foreach ($batches as $batch) {
6592		- $delete_response = wp_remote_post("https://{$host}/vectors/delete", array(
6593		- 'headers' => array(
6594		- 'Api-Key' => $api_key,
6595		- 'accept' => 'application/json',
6596		- 'content-type' => 'application/json'
6597		- ),
6598		- 'body' => wp_json_encode(array('ids' => $batch)),
6599		- 'timeout' => 60
6600		- ));
6601		-
6602		- if (is_wp_error($delete_response)) {
6603		- $pinecone_success = false;
6604		- $errors[] = 'Pinecone batch deletion failed: ' . $delete_response->get_error_message();
6605		- MxChat_Admin::mxchat_log_debug('pinecone_error', 'Bulk delete batch failed: ' . $delete_response->get_error_message());
6606		- } else {
6607		- $response_code = wp_remote_retrieve_response_code($delete_response);
6608		- if ($response_code !== 200) {
6609		- $pinecone_success = false;
6610		- $response_body = wp_remote_retrieve_body($delete_response);
6611		- $errors[] = "Pinecone API error (HTTP $response_code)";
6612		- MxChat_Admin::mxchat_log_debug('pinecone_error', 'Bulk delete batch failed (HTTP ' . $response_code . ')', array('response' => substr($response_body, 0, 200)));
6613		- }
6614		- }
6615		- }
6616		-
6617		- // Mark all pinecone entries based on batch result
6618		- foreach ($pinecone_entry_ids as $eid) {
6619		- if ($pinecone_success) {
6620		- $success_ids[] = $eid;
6621		- } else {
6622		- $failed_ids[] = $eid;
6623		- }
6624		- }
6625		- }
6626		-
6627		- // =============================================
6628		- // PHASE 3: WordPress database deletions
6629		- // =============================================
6630		- foreach ($wordpress_entries as $entry) {
6631		- $entry_id = sanitize_text_field($entry['id'] ?? '');
6632		- $source_url = isset($entry['sourceUrl']) ? esc_url_raw($entry['sourceUrl']) : '';
6633		- $is_group = isset($entry['isGroup']) && ($entry['isGroup'] === true || $entry['isGroup'] === 'true');
6634		-
6635		- if (empty($entry_id)) {
6636		- continue;
6637		- }
6638		-
6639		- try {
6640		- if ($is_group && !empty($source_url)) {
6641		- $result = $wpdb->delete(
6642		- $table_name,
6643		- array('source_url' => $source_url),
6644		- array('%s')
6645		- );
6646		- } else {
6647		- wp_cache_delete('prompt_' . $entry_id, 'mxchat_prompts');
6648		- $result = $wpdb->delete(
6649		- $table_name,
6650		- array('id' => intval($entry_id)),
6651		- array('%d')
6652		- );
6653		- }
6654		-
6655		- if ($result !== false) {
6656		- $success_ids[] = $entry_id;
6657		- } else {
6658		- $failed_ids[] = $entry_id;
6659		- $errors[] = "Database error for entry: $entry_id";
6660		- }
6661		- } catch (Exception $e) {
6662		- $failed_ids[] = $entry_id;
6663		- $errors[] = $e->getMessage();
6664		- }
6665		- }
6666		-
6667		- wp_send_json_success(array(
6668		- 'success_ids' => $success_ids,
6669		- 'failed_ids' => $failed_ids,
6670		- 'errors' => $errors,
6671		- 'total_processed' => count($success_ids) + count($failed_ids)
6672		- ));
6673		-
6674		- exit;
6675		-}
6676		-
6677		-/**
6678		- * Get hierarchical roles for dropdown
6679		- */
6680		-public function mxchat_get_role_options() {
6681		- return array(
6682		- 'public' => __('Public (Everyone)', 'mxchat'),
6683		- 'logged_in' => __('Logged In Users', 'mxchat'),
6684		- 'subscriber' => __('Subscribers & Above', 'mxchat'),
6685		- 'contributor' => __('Contributors & Above', 'mxchat'),
6686		- 'author' => __('Authors & Above', 'mxchat'),
6687		- 'editor' => __('Editors & Above', 'mxchat'),
6688		- 'administrator' => __('Administrators Only', 'mxchat')
6689		- );
6690		-}
6691		-
6692		-/**
6693		- * Check if user has access to content based on role restriction
6694		- */
6695		-public function mxchat_user_has_content_access($role_restriction) {
6696		- // Public content is always accessible
6697		- if ($role_restriction === 'public' || empty($role_restriction)) {
6698		- return true;
6699		- }
6700		-
6701		- // Check if user is logged in for logged_in restriction
6702		- if ($role_restriction === 'logged_in') {
6703		- return is_user_logged_in();
6704		- }
6705		-
6706		- // If not logged in, no access to role-restricted content
6707		- if (!is_user_logged_in()) {
6708		- return false;
6709		- }
6710		-
6711		- $user = wp_get_current_user();
6712		- $user_roles = $user->roles;
6713		-
6714		- if (empty($user_roles)) {
6715		- return false;
6716		- }
6717		-
6718		- // Define role hierarchy (higher number = higher access)
6719		- $hierarchy = array(
6720		- 'subscriber' => 1,
6721		- 'contributor' => 2,
6722		- 'author' => 3,
6723		- 'editor' => 4,
6724		- 'administrator' => 5
6725		- );
6726		-
6727		- // Get required level
6728		- $required_level = isset($hierarchy[$role_restriction]) ? $hierarchy[$role_restriction] : 0;
6729		-
6730		- // Check if user has required level or higher
6731		- foreach ($user_roles as $user_role) {
6732		- $user_level = isset($hierarchy[$user_role]) ? $hierarchy[$user_role] : 0;
6733		- if ($user_level >= $required_level) {
6734		- return true;
6735		- }
6736		- }
6737		-
6738		- return false;
6739		-}
6740		-
6741		-/**
6742		- * Handle role restriction updates via AJAX
6743		- * Removed cache clearing call since we removed caching
6744		- */
6745		-public function ajax_mxchat_update_role_restriction() {
6746		- // Verify nonce and permissions
6747		- if (!check_ajax_referer('mxchat_update_role_nonce', 'nonce', false)) {
6748		- wp_send_json_error('Invalid nonce');
6749		- exit;
6750		- }
6751		-
6752		- if (!current_user_can('manage_options')) {
6753		- wp_send_json_error('Unauthorized access');
6754		- exit;
6755		- }
6756		-
6757		- $entry_id = isset($_POST['entry_id']) ? sanitize_text_field($_POST['entry_id']) : '';
6758		- $role_restriction = isset($_POST['role_restriction']) ? sanitize_text_field($_POST['role_restriction']) : 'public';
6759		- $data_source = isset($_POST['data_source']) ? sanitize_text_field($_POST['data_source']) : 'wordpress';
6760		-
6761		- if (empty($entry_id)) {
6762		- wp_send_json_error('Invalid entry ID');
6763		- exit;
6764		- }
6765		-
6766		- // Get knowledge manager instance to validate role restriction
6767		- $knowledge_manager = MxChat_Knowledge_Manager::get_instance();
6768		- $valid_roles = array_keys($knowledge_manager->mxchat_get_role_options());
6769		- if (!in_array($role_restriction, $valid_roles)) {
6770		- wp_send_json_error('Invalid role restriction');
6771		- exit;
6772		- }
6773		-
6774		- global $wpdb;
6775		-
6776		- if ($data_source === 'pinecone') {
6777		- // Handle Pinecone role restriction (stored separately in WordPress table)
6778		- $roles_table = $wpdb->prefix . 'mxchat_pinecone_roles';
6779		-
6780		- // Use REPLACE to insert or update the role restriction
6781		- $result = $wpdb->replace(
6782		- $roles_table,
6783		- array(
6784		- 'vector_id' => $entry_id,
6785		- 'role_restriction' => $role_restriction,
6786		- 'updated_at' => current_time('mysql')
6787		- ),
6788		- array('%s', '%s', '%s')
6789		- );
6790		-
6791		- // No cache clearing needed since we removed caching
6792		-
6793		- } else {
6794		- // Handle WordPress database role restriction (existing functionality)
6795		- $table_name = $wpdb->prefix . 'mxchat_system_prompt_content';
6796		-
6797		- $result = $wpdb->update(
6798		- $table_name,
6799		- array('role_restriction' => $role_restriction),
6800		- array('id' => absint($entry_id)),
6801		- array('%s'),
6802		- array('%d')
6803		- );
6804		- }
6805		-
6806		- if ($result === false) {
6807		- wp_send_json_error('Database update failed: ' . $wpdb->last_error);
6808		- exit;
6809		- }
6810		-
6811		- wp_send_json_success(array(
6812		- 'message' => 'Role restriction updated successfully',
6813		- 'role_restriction' => $role_restriction,
6814		- 'data_source' => $data_source,
6815		- 'entry_id' => $entry_id
6816		- ));
6817		- exit;
6818		-}
6819		-
6820		-// ========================================
6821		-// ROLE-BASED CONTENT RESTRICTIONS - NEW FUNCTIONS
6822		-// Add these to your MxChat_Knowledge_Manager class
6823		-// ========================================
6824		-
6825		-/**
6826		- * Initialize role-based content hooks
6827		- * Add this call to your __construct() or mxchat_init_hooks() method
6828		- */
6829		-private function mxchat_init_role_hooks() {
6830		- // AJAX handlers for tag-role mappings
6831		- add_action('wp_ajax_mxchat_add_tag_role_mapping', array($this, 'ajax_add_tag_role_mapping'));
6832		- add_action('wp_ajax_mxchat_delete_tag_role_mapping', array($this, 'ajax_delete_tag_role_mapping'));
6833		- add_action('wp_ajax_mxchat_get_tag_role_mappings', array($this, 'ajax_get_tag_role_mappings'));
6834		- add_action('wp_ajax_mxchat_bulk_update_tag_roles', array($this, 'ajax_bulk_update_tag_roles'));
6835		-
6836		- // Hook to automatically update role restrictions when tags are added/removed
6837		- add_action('set_object_terms', array($this, 'handle_tag_change'), 10, 6);
6838		-
6839		- // Hook to apply role restrictions on auto-sync
6840		- add_action('mxchat_content_stored', array($this, 'apply_role_restriction_after_storage'), 10, 2);
6841		-}
6842		-
6843		-/**
6844		- * Add tag-role mapping via AJAX
6845		- */
6846		-public function ajax_add_tag_role_mapping() {
6847		- // Verify nonce and permissions
6848		- check_ajax_referer('mxchat_prompts_setting_nonce', 'nonce');
6849		-
6850		- if (!current_user_can('manage_options')) {
6851		- wp_send_json_error('Unauthorized access');
6852		- exit;
6853		- }
6854		-
6855		- $tag_input = isset($_POST['tag_slug']) ? sanitize_text_field($_POST['tag_slug']) : '';
6856		- $role_restriction = isset($_POST['role_restriction']) ? sanitize_text_field($_POST['role_restriction']) : 'public';
6857		-
6858		- if (empty($tag_input)) {
6859		- wp_send_json_error('Please enter a tag name or slug');
6860		- exit;
6861		- }
6862		-
6863		- // Validate role restriction
6864		- $valid_roles = array_keys($this->mxchat_get_role_options());
6865		- if (!in_array($role_restriction, $valid_roles)) {
6866		- wp_send_json_error('Invalid role restriction');
6867		- exit;
6868		- }
6869		-
6870		- // Resolve the tag by slug first, then fall back to its display name, so users can
6871		- // enter either "premium-content" or "Premium Content". (plan b8bcf5 — the field is
6872		- // labeled by name but previously validated by slug only, producing the confusing
6873		- // "Tag does not exist in WordPress" error when a real tag's name was typed.)
6874		- $term = get_term_by('slug', $tag_input, 'post_tag');
6875		- if (!$term) {
6876		- $term = get_term_by('name', $tag_input, 'post_tag');
6877		- }
6878		- if (!$term) {
6879		- 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.');
6880		- exit;
6881		- }
6882		-
6883		- // Always key the mapping by the RESOLVED slug — apply_role_restriction_to_post()
6884		- // compares against each post's tag slugs, so the stored key must be a slug,
6885		- // never the raw (possibly display-name) input.
6886		- $tag_slug = $term->slug;
6887		-
6888		- // Get existing mappings
6889		- $mappings = get_option('mxchat_tag_role_mappings', array());
6890		-
6891		- // Check if mapping already exists
6892		- if (isset($mappings[$tag_slug])) {
6893		- wp_send_json_error('Mapping for this tag already exists');
6894		- exit;
6895		- }
6896		-
6897		- // Add new mapping
6898		- $mappings[$tag_slug] = $role_restriction;
6899		- update_option('mxchat_tag_role_mappings', $mappings);
6900		-
6901		- wp_send_json_success(array(
6902		- 'message' => 'Tag-role mapping added successfully',
6903		- 'tag_slug' => $tag_slug,
6904		- 'role_restriction' => $role_restriction
6905		- ));
6906		- exit;
6907		-}
6908		-
6909		-/**
6910		- * Delete tag-role mapping via AJAX
6911		- */
6912		-public function ajax_delete_tag_role_mapping() {
6913		- // Verify nonce and permissions
6914		- check_ajax_referer('mxchat_prompts_setting_nonce', 'nonce');
6915		-
6916		- if (!current_user_can('manage_options')) {
6917		- wp_send_json_error('Unauthorized access');
6918		- exit;
6919		- }
6920		-
6921		- $tag_slug = isset($_POST['tag_slug']) ? sanitize_text_field($_POST['tag_slug']) : '';
6922		-
6923		- if (empty($tag_slug)) {
6924		- wp_send_json_error('Tag slug is required');
6925		- exit;
6926		- }
6927		-
6928		- // Get existing mappings
6929		- $mappings = get_option('mxchat_tag_role_mappings', array());
6930		-
6931		- // Check if mapping exists
6932		- if (!isset($mappings[$tag_slug])) {
6933		- wp_send_json_error('Mapping does not exist');
6934		- exit;
6935		- }
6936		-
6937		- // Remove mapping
6938		- unset($mappings[$tag_slug]);
6939		- update_option('mxchat_tag_role_mappings', $mappings);
6940		-
6941		- wp_send_json_success(array(
6942		- 'message' => 'Tag-role mapping deleted successfully',
6943		- 'tag_slug' => $tag_slug
6944		- ));
6945		- exit;
6946		-}
6947		-
6948		-/**
6949		- * Get all tag-role mappings via AJAX
6950		- */
6951		-public function ajax_get_tag_role_mappings() {
6952		- // Verify nonce and permissions
6953		- check_ajax_referer('mxchat_prompts_setting_nonce', 'nonce');
6954		-
6955		- if (!current_user_can('manage_options')) {
6956		- wp_send_json_error('Unauthorized access');
6957		- exit;
6958		- }
6959		-
6960		- // Get mappings
6961		- $mappings = get_option('mxchat_tag_role_mappings', array());
6962		- $role_options = $this->mxchat_get_role_options();
6963		-
6964		- $formatted_mappings = array();
6965		-
6966		- foreach ($mappings as $tag_slug => $role_restriction) {
6967		- // Get tag object
6968		- $term = get_term_by('slug', $tag_slug, 'post_tag');
6969		-
6970		- // Count posts with this tag
6971		- $post_count = 0;
6972		- if ($term) {
6973		- $post_count = $term->count;
6974		- }
6975		-
6976		- $formatted_mappings[] = array(
6977		- 'tag_slug' => $tag_slug,
6978		- 'role_restriction' => $role_restriction,
6979		- 'role_label' => $role_options[$role_restriction] ?? $role_restriction,
6980		- 'post_count' => $post_count
6981		- );
6982		- }
6983		-
6984		- wp_send_json_success(array(
6985		- 'mappings' => $formatted_mappings
6986		- ));
6987		- exit;
6988		-}
6989		-
6990		-/**
6991		- * Bulk update role restrictions for all existing content with mapped tags
6992		- */
6993		-public function ajax_bulk_update_tag_roles() {
6994		- // Verify nonce and permissions
6995		- check_ajax_referer('mxchat_prompts_setting_nonce', 'nonce');
6996		-
6997		- if (!current_user_can('manage_options')) {
6998		- wp_send_json_error('Unauthorized access');
6999		- exit;
7000		- }
7001		-
7002		- // Get mappings
7003		- $mappings = get_option('mxchat_tag_role_mappings', array());
7004		-
7005		- if (empty($mappings)) {
7006		- wp_send_json_error('No tag-role mappings found');
7007		- exit;
7008		- }
7009		-
7010		- global $wpdb;
7011		-
7012		- // Check if using Pinecone
7013		- $pinecone_options = get_option('mxchat_pinecone_addon_options', array());
7014		- $use_pinecone = ($pinecone_options['mxchat_use_pinecone'] ?? '0') === '1';
7015		-
7016		- $updated_count = 0;
7017		- $details = array();
7018		-
7019		- foreach ($mappings as $tag_slug => $role_restriction) {
7020		- // Get all posts with this tag
7021		- $posts = get_posts(array(
7022		- 'tag' => $tag_slug,
7023		- 'post_type' => 'any',
7024		- 'posts_per_page' => -1,
7025		- 'fields' => 'ids',
7026		- 'post_status' => 'publish'
7027		- ));
7028		-
7029		- if (empty($posts)) {
7030		- continue;
7031		- }
7032		-
7033		- $tag_updated = 0;
7034		-
7035		- foreach ($posts as $post_id) {
7036		- $source_url = get_permalink($post_id);
7037		- if (!$source_url) {
7038		- continue;
7039		- }
7040		-
7041		- if ($use_pinecone && !empty($pinecone_options['mxchat_pinecone_api_key'])) {
7042		- // Update Pinecone role restriction
7043		- $roles_table = $wpdb->prefix . 'mxchat_pinecone_roles';
7044		- $vector_id = md5($source_url);
7045		-
7046		- $result = $wpdb->replace(
7047		- $roles_table,
7048		- array(
7049		- 'vector_id' => $vector_id,
7050		- 'role_restriction' => $role_restriction,
7051		- 'updated_at' => current_time('mysql')
7052		- ),
7053		- array('%s', '%s', '%s')
7054		- );
7055		- } else {
7056		- // Update WordPress DB
7057		- $table_name = $wpdb->prefix . 'mxchat_system_prompt_content';
7058		-
7059		- $result = $wpdb->update(
7060		- $table_name,
7061		- array('role_restriction' => $role_restriction),
7062		- array('source_url' => $source_url),
7063		- array('%s'),
7064		- array('%s')
7065		- );
7066		- }
7067		-
7068		- if ($result !== false) {
7069		- $tag_updated++;
7070		- $updated_count++;
7071		- }
7072		- }
7073		-
7074		- if ($tag_updated > 0) {
7075		- $details[] = sprintf(
7076		- 'Tag "%s" (%s): %d posts updated',
7077		- $tag_slug,
7078		- $role_restriction,
7079		- $tag_updated
7080		- );
7081		- }
7082		- }
7083		-
7084		- wp_send_json_success(array(
7085		- 'message' => 'Bulk update completed',
7086		- 'updated_count' => $updated_count,
7087		- 'tags_processed' => count($mappings),
7088		- 'details' => $details
7089		- ));
7090		- exit;
7091		-}
7092		-
7093		-/**
7094		- * Handle tag changes on posts (when tags are added or removed)
7095		- */
7096		-public function handle_tag_change($object_id, $terms, $tt_ids, $taxonomy, $append, $old_tt_ids) {
7097		- // Only process post tags
7098		- if ($taxonomy !== 'post_tag') {
7099		- return;
7100		- }
7101		-
7102		- // Get tag-role mappings
7103		- $mappings = get_option('mxchat_tag_role_mappings', array());
7104		-
7105		- if (empty($mappings)) {
7106		- return;
7107		- }
7108		-
7109		- // Get the post's URL
7110		- $source_url = get_permalink($object_id);
7111		- if (!$source_url) {
7112		- return;
7113		- }
7114		-
7115		- // Determine the highest role restriction based on tags
7116		- $highest_role = 'public';
7117		- $role_hierarchy = array(
7118		- 'public' => 0,
7119		- 'logged_in' => 1,
7120		- 'subscriber' => 2,
7121		- 'contributor' => 3,
7122		- 'author' => 4,
7123		- 'editor' => 5,
7124		- 'administrator' => 6
7125		- );
7126		-
7127		- // Get all current tags for the post
7128		- $current_tags = wp_get_post_tags($object_id, array('fields' => 'slugs'));
7129		-
7130		- // Find the highest role restriction among the tags
7131		- foreach ($current_tags as $tag_slug) {
7132		- if (isset($mappings[$tag_slug])) {
7133		- $role = $mappings[$tag_slug];
7134		- if (isset($role_hierarchy[$role]) && $role_hierarchy[$role] > $role_hierarchy[$highest_role]) {
7135		- $highest_role = $role;
7136		- }
7137		- }
7138		- }
7139		-
7140		- // Update the role restriction in the database
7141		- global $wpdb;
7142		-
7143		- // Check if using Pinecone
7144		- $pinecone_options = get_option('mxchat_pinecone_addon_options', array());
7145		- $use_pinecone = ($pinecone_options['mxchat_use_pinecone'] ?? '0') === '1';
7146		-
7147		- if ($use_pinecone && !empty($pinecone_options['mxchat_pinecone_api_key'])) {
7148		- // Update Pinecone role restriction
7149		- $roles_table = $wpdb->prefix . 'mxchat_pinecone_roles';
7150		- $vector_id = md5($source_url);
7151		-
7152		- $wpdb->replace(
7153		- $roles_table,
7154		- array(
7155		- 'vector_id' => $vector_id,
7156		- 'role_restriction' => $highest_role,
7157		- 'updated_at' => current_time('mysql')
7158		- ),
7159		- array('%s', '%s', '%s')
7160		- );
7161		- } else {
7162		- // Update WordPress DB
7163		- $table_name = $wpdb->prefix . 'mxchat_system_prompt_content';
7164		-
7165		- $wpdb->update(
7166		- $table_name,
7167		- array('role_restriction' => $highest_role),
7168		- array('source_url' => $source_url),
7169		- array('%s'),
7170		- array('%s')
7171		- );
7172		- }
7173		-}
7174		-
7175		-/**
7176		- * Apply role restriction after content is stored (for auto-sync)
7177		- */
7178		-public function apply_role_restriction_after_storage($post_id, $source_url) {
7179		- // Get tag-role mappings
7180		- $mappings = get_option('mxchat_tag_role_mappings', array());
7181		-
7182		- if (empty($mappings)) {
7183		- return;
7184		- }
7185		-
7186		- // Get all tags for the post
7187		- $post_tags = wp_get_post_tags($post_id, array('fields' => 'slugs'));
7188		-
7189		- if (empty($post_tags)) {
7190		- return;
7191		- }
7192		-
7193		- // Determine the highest role restriction based on tags
7194		- $highest_role = 'public';
7195		- $role_hierarchy = array(
7196		- 'public' => 0,
7197		- 'logged_in' => 1,
7198		- 'subscriber' => 2,
7199		- 'contributor' => 3,
7200		- 'author' => 4,
7201		- 'editor' => 5,
7202		- 'administrator' => 6
7203		- );
7204		-
7205		- foreach ($post_tags as $tag_slug) {
7206		- if (isset($mappings[$tag_slug])) {
7207		- $role = $mappings[$tag_slug];
7208		- if (isset($role_hierarchy[$role]) && $role_hierarchy[$role] > $role_hierarchy[$highest_role]) {
7209		- $highest_role = $role;
7210		- }
7211		- }
7212		- }
7213		-
7214		- // If no restricted tags found, return (leave as public)
7215		- if ($highest_role === 'public') {
7216		- return;
7217		- }
7218		-
7219		- // Update the role restriction
7220		- global $wpdb;
7221		-
7222		- // Check if using Pinecone
7223		- $pinecone_options = get_option('mxchat_pinecone_addon_options', array());
7224		- $use_pinecone = ($pinecone_options['mxchat_use_pinecone'] ?? '0') === '1';
7225		-
7226		- if ($use_pinecone && !empty($pinecone_options['mxchat_pinecone_api_key'])) {
7227		- // Update Pinecone role restriction
7228		- $roles_table = $wpdb->prefix . 'mxchat_pinecone_roles';
7229		- $vector_id = md5($source_url);
7230		-
7231		- $wpdb->replace(
7232		- $roles_table,
7233		- array(
7234		- 'vector_id' => $vector_id,
7235		- 'role_restriction' => $highest_role,
7236		- 'updated_at' => current_time('mysql')
7237		- ),
7238		- array('%s', '%s', '%s')
7239		- );
7240		- } else {
7241		- // Update WordPress DB
7242		- $table_name = $wpdb->prefix . 'mxchat_system_prompt_content';
7243		-
7244		- $wpdb->update(
7245		- $table_name,
7246		- array('role_restriction' => $highest_role),
7247		- array('source_url' => $source_url),
7248		- array('%s'),
7249		- array('%s')
7250		- );
7251		- }
7252		-}
7253		-
7254		-
7255		- // ========================================
7256		- // HELPER METHODS
7257		- // ========================================
7258		-
7259		- /**
7260		- * Check if user has required permissions for content processing
7261		- */
7262		- private function mxchat_check_user_permissions() {
7263		- if (!current_user_can('manage_options')) {
7264		- wp_die(esc_html__('You do not have sufficient permissions.', 'mxchat'));
7265		- }
7266		- }
7267		-
7268		- /**
7269		- * Validate nonce for security
7270		- */
7271		- private function mxchat_validate_nonce($nonce_name, $nonce_action) {
7272		- if (!isset($_POST[$nonce_name]) || !wp_verify_nonce($_POST[$nonce_name], $nonce_action)) {
7273		- wp_die(esc_html__('Security check failed.', 'mxchat'));
7274		- }
7275		- }
7276		-
7277		- /**
7278		- * Get embedding API credentials
7279		- */
7280		- private function mxchat_get_embedding_credentials() {
7281		- $embedding_model = $this->options['embedding_model'] ?? 'text-embedding-ada-002';
7282		-
7283		- if (strpos($embedding_model, 'text-embedding-') !== false) {
7284		- return array(
7285		- 'type' => 'openai',
7286		- 'api_key' => $this->options['api_key'] ?? ''
7287		- );
7288		- } elseif (strpos($embedding_model, 'voyage-') !== false) {
7289		- return array(
7290		- 'type' => 'voyage',
7291		- 'api_key' => $this->options['voyage_api_key'] ?? ''
7292		- );
7293		- } elseif (strpos($embedding_model, 'gemini-embedding-') !== false) {
7294		- return array(
7295		- 'type' => 'gemini',
7296		- 'api_key' => $this->options['gemini_api_key'] ?? ''
7297		- );
7298		- }
7299		-
7300		- return array('type' => 'unknown', 'api_key' => '');
7301		- }
7302		-
7303		- /**
7304		- * Log processing errors
7305		- */
7306		- private function mxchat_log_processing_error($operation, $error_message) {
7307		- //error_log("MxChat Knowledge Processing {$operation} Error: " . $error_message);
7308		- }
7309		-
7310		- /**
7311		- * Set admin notice transient
7312		- */
7313		- private function mxchat_set_admin_notice($type, $message) {
7314		- set_transient("mxchat_admin_notice_{$type}", $message, 30);
7315		- }
7316		-
7317		- /**
7318		- * Get Pinecone manager instance for vector operations
7319		- */
7320		- private function mxchat_get_pinecone_manager() {
7321		- return MxChat_Pinecone_Manager::get_instance();
7322		- }
7323		-
7324		-
7325		- // ========================================
7326		-// DATABASE QUEUE TABLE MANAGEMENT
7327		-// ========================================
7328		-
7329		-/**
7330		- * Create queue table on plugin activation
7331		- * Call this from your plugin activation hook
7332		- */
7333		-public function mxchat_create_queue_table() {
7334		- global $wpdb;
7335		-
7336		- $table_name = $wpdb->prefix . 'mxchat_processing_queue';
7337		- $charset_collate = $wpdb->get_charset_collate();
7338		-
7339		- $sql = "CREATE TABLE IF NOT EXISTS $table_name (
7340		- id bigint(20) unsigned NOT NULL AUTO_INCREMENT,
7341		- queue_id varchar(64) NOT NULL,
7342		- item_type varchar(20) NOT NULL,
7343		- item_data longtext NOT NULL,
7344		- status varchar(20) NOT NULL DEFAULT 'pending',
7345		- bot_id varchar(50) NOT NULL DEFAULT 'default',
7346		- priority int(11) NOT NULL DEFAULT 0,
7347		- attempts int(11) NOT NULL DEFAULT 0,
7348		- max_attempts int(11) NOT NULL DEFAULT 3,
7349		- error_message text DEFAULT NULL,
7350		- created_at datetime NOT NULL,
7351		- started_at datetime DEFAULT NULL,
7352		- completed_at datetime DEFAULT NULL,
7353		- PRIMARY KEY (id),
7354		- KEY queue_id (queue_id),
7355		- KEY status (status),
7356		- KEY item_type (item_type),
7357		- KEY priority (priority)
7358		- ) $charset_collate;";
7359		-
7360		- require_once(ABSPATH . 'wp-admin/includes/upgrade.php');
7361		- dbDelta($sql);
7362		-
7363		- // Also create a meta table for queue metadata
7364		- $meta_table = $wpdb->prefix . 'mxchat_queue_meta';
7365		-
7366		- $meta_sql = "CREATE TABLE IF NOT EXISTS $meta_table (
7367		- id bigint(20) unsigned NOT NULL AUTO_INCREMENT,
7368		- queue_id varchar(64) NOT NULL,
7369		- meta_key varchar(255) NOT NULL,
7370		- meta_value longtext,
7371		- PRIMARY KEY (id),
7372		- KEY queue_id (queue_id),
7373		- KEY meta_key (meta_key)
7374		- ) $charset_collate;";
7375		-
7376		- dbDelta($meta_sql);
7377		-}
7378		-
7379		-/**
7380		- * Add items to the processing queue
7381		- *
7382		- * @param string $queue_id Unique identifier for this queue batch
7383		- * @param string $item_type Type of item (url, pdf_page)
7384		- * @param array $items Array of items to queue
7385		- * @param string $bot_id Bot ID for processing
7386		- * @return int Number of items queued
7387		- */
7388		-private function mxchat_add_to_queue($queue_id, $item_type, $items, $bot_id = 'default') {
7389		- global $wpdb;
7390		- $table_name = $wpdb->prefix . 'mxchat_processing_queue';
7391		-
7392		- $queued_count = 0;
7393		- $priority = 0;
7394		-
7395		- foreach ($items as $item) {
7396		- $result = $wpdb->insert(
7397		- $table_name,
7398		- array(
7399		- 'queue_id' => $queue_id,
7400		- 'item_type' => $item_type,
7401		- 'item_data' => wp_json_encode($item),
7402		- 'status' => 'pending',
7403		- 'bot_id' => $bot_id,
7404		- 'priority' => $priority,
7405		- 'attempts' => 0,
7406		- 'max_attempts' => 3,
7407		- 'created_at' => current_time('mysql')
7408		- ),
7409		- array('%s', '%s', '%s', '%s', '%s', '%d', '%d', '%d', '%s')
7410		- );
7411		-
7412		- if ($result) {
7413		- $queued_count++;
7414		- }
7415		-
7416		- $priority++; // Process in order
7417		- }
7418		-
7419		- return $queued_count;
7420		-}
7421		-
7422		-/**
7423		- * Store queue metadata (total counts, source URL, etc.)
7424		- */
7425		-private function mxchat_set_queue_meta($queue_id, $meta_key, $meta_value) {
7426		- global $wpdb;
7427		- $meta_table = $wpdb->prefix . 'mxchat_queue_meta';
7428		-
7429		- // Check if meta exists
7430		- $existing = $wpdb->get_var($wpdb->prepare(
7431		- "SELECT id FROM $meta_table WHERE queue_id = %s AND meta_key = %s",
7432		- $queue_id,
7433		- $meta_key
7434		- ));
7435		-
7436		- if ($existing) {
7437		- // Update
7438		- $wpdb->update(
7439		- $meta_table,
7440		- array('meta_value' => maybe_serialize($meta_value)),
7441		- array('queue_id' => $queue_id, 'meta_key' => $meta_key),
7442		- array('%s'),
7443		- array('%s', '%s')
7444		- );
7445		- } else {
7446		- // Insert
7447		- $wpdb->insert(
7448		- $meta_table,
7449		- array(
7450		- 'queue_id' => $queue_id,
7451		- 'meta_key' => $meta_key,
7452		- 'meta_value' => maybe_serialize($meta_value)
7453		- ),
7454		- array('%s', '%s', '%s')
7455		- );
7456		- }
7457		-}
7458		-
7459		-/**
7460		- * Get queue metadata
7461		- */
7462		-private function mxchat_get_queue_meta($queue_id, $meta_key) {
7463		- global $wpdb;
7464		- $meta_table = $wpdb->prefix . 'mxchat_queue_meta';
7465		-
7466		- $value = $wpdb->get_var($wpdb->prepare(
7467		- "SELECT meta_value FROM $meta_table WHERE queue_id = %s AND meta_key = %s",
7468		- $queue_id,
7469		- $meta_key
7470		- ));
7471		-
7472		- return maybe_unserialize($value);
7473		-}
7474		-
7475		-// ========================================
7476		-// AJAX QUEUE PROCESSING HANDLERS
7477		-// ========================================
7478		-
7479		-/**
7480		- * AJAX: Get next item from queue to process
7481		- */
7482		-public function ajax_mxchat_get_next_queue_item() {
7483		- // Verify nonce and permissions
7484		- check_ajax_referer('mxchat_queue_nonce', 'nonce');
7485		-
7486		- if (!current_user_can('manage_options')) {
7487		- wp_send_json_error('Unauthorized access');
7488		- }
7489		-
7490		- $queue_id = isset($_POST['queue_id']) ? sanitize_text_field($_POST['queue_id']) : '';
7491		-
7492		- if (empty($queue_id)) {
7493		- wp_send_json_error('Missing queue ID');
7494		- }
7495		-
7496		- global $wpdb;
7497		- $table_name = $wpdb->prefix . 'mxchat_processing_queue';
7498		-
7499		- // Get next pending item with retry logic for failed items
7500		- $next_item = $wpdb->get_row($wpdb->prepare(
7501		- "SELECT * FROM $table_name
7502		- WHERE queue_id = %s
7503		- AND status IN ('pending', 'failed')
7504		- AND attempts < max_attempts
7505		- ORDER BY priority ASC, id ASC
7506		- LIMIT 1",
7507		- $queue_id
7508		- ));
7509		-
7510		- if (!$next_item) {
7511		- // No more items - queue complete
7512		- wp_send_json_success(array(
7513		- 'complete' => true,
7514		- 'message' => 'Queue processing complete'
7515		- ));
7516		- }
7517		-
7518		- // Mark item as processing
7519		- $wpdb->update(
7520		- $table_name,
7521		- array(
7522		- 'status' => 'processing',
7523		- 'started_at' => current_time('mysql'),
7524		- 'attempts' => $next_item->attempts + 1
7525		- ),
7526		- array('id' => $next_item->id),
7527		- array('%s', '%s', '%d'),
7528		- array('%d')
7529		- );
7530		-
7531		- wp_send_json_success(array(
7532		- 'complete' => false,
7533		- 'item' => array(
7534		- 'id' => $next_item->id,
7535		- 'type' => $next_item->item_type,
7536		- 'data' => json_decode($next_item->item_data, true),
7537		- 'bot_id' => $next_item->bot_id,
7538		- 'attempt' => $next_item->attempts + 1
7539		- )
7540		- ));
7541		-}
7542		-
7543		-/**
7544		- * AJAX: Process a single queue item
7545		- */
7546		-public function ajax_mxchat_process_queue_item() {
7547		- // Verify nonce and permissions
7548		- check_ajax_referer('mxchat_queue_nonce', 'nonce');
7549		-
7550		- if (!current_user_can('manage_options')) {
7551		- wp_send_json_error('Unauthorized access');
7552		- }
7553		-
7554		- $item_id = isset($_POST['item_id']) ? absint($_POST['item_id']) : 0;
7555		- $item_type = isset($_POST['item_type']) ? sanitize_text_field($_POST['item_type']) : '';
7556		- $item_data = isset($_POST['item_data']) ? $_POST['item_data'] : array();
7557		- $bot_id = isset($_POST['bot_id']) ? sanitize_key($_POST['bot_id']) : 'default';
7558		-
7559		- if (empty($item_id) || empty($item_type)) {
7560		- wp_send_json_error('Missing item data');
7561		- }
7562		-
7563		- global $wpdb;
7564		- $table_name = $wpdb->prefix . 'mxchat_processing_queue';
7565		-
7566		- // Process based on item type
7567		- try {
7568		- set_time_limit(60); // Give processing 60 seconds
7569		-
7570		- $result = false;
7571		- $error_message = '';
7572		-
7573		- // Read item directly from DB to get queue_id and preserve special chars in item_data
7574		- // (POST round-trip through JS mangles characters like apostrophes in URLs)
7575		- $db_item = $wpdb->get_row($wpdb->prepare(
7576		- "SELECT queue_id, item_data FROM $table_name WHERE id = %d",
7577		- $item_id
7578		- ));
7579		- $item_queue_id = $db_item ? $db_item->queue_id : '';
7580		- if ($db_item && !empty($db_item->item_data)) {
7581		- $db_data = json_decode($db_item->item_data, true);
7582		- if (is_array($db_data)) {
7583		- $item_data = $db_data;
7584		- }
7585		- }
7586		-
7587		- switch ($item_type) {
7588		- case 'url':
7589		- $result = $this->mxchat_process_queue_url($item_data, $bot_id, $item_queue_id);
7590		- break;
7591		-
7592		- case 'pdf_page':
7593		- $result = $this->mxchat_process_queue_pdf_page($item_data, $bot_id);
7594		- break;
7595		-
7596		- default:
7597		- throw new Exception('Unknown item type: ' . $item_type);
7598		- }
7599		-
7600		- if (is_wp_error($result)) {
7601		- $error_code = $result->get_error_code();
7602		- // Content errors (empty page, sanitization) are permanent — retrying won't help
7603		- $permanent_codes = array('empty_page', 'empty_after_sanitization', 'no_api_key', 'page_not_found');
7604		- if (in_array($error_code, $permanent_codes)) {
7605		- // Mark as permanently failed — set attempts = max_attempts so it won't be retried
7606		- $current_item = $wpdb->get_row($wpdb->prepare(
7607		- "SELECT max_attempts FROM $table_name WHERE id = %d", $item_id
7608		- ));
7609		- $wpdb->update(
7610		- $table_name,
7611		- array(
7612		- 'status' => 'failed',
7613		- 'error_message' => $result->get_error_message(),
7614		- 'attempts' => $current_item ? $current_item->max_attempts : 3
7615		- ),
7616		- array('id' => $item_id),
7617		- array('%s', '%s', '%d'),
7618		- array('%d')
7619		- );
7620		- wp_send_json_error(array(
7621		- 'message' => $result->get_error_message(),
7622		- 'permanent_failure' => true,
7623		- 'item_id' => $item_id
7624		- ));
7625		- return;
7626		- }
7627		- throw new Exception($result->get_error_message());
7628		- }
7629		-
7630		- if ($result === false) {
7631		- throw new Exception('Processing returned false - item may be empty or invalid');
7632		- }
7633		-
7634		- // Mark as completed
7635		- $wpdb->update(
7636		- $table_name,
7637		- array(
7638		- 'status' => 'completed',
7639		- 'completed_at' => current_time('mysql'),
7640		- 'error_message' => null
7641		- ),
7642		- array('id' => $item_id),
7643		- array('%s', '%s', '%s'),
7644		- array('%d')
7645		- );
7646		-
7647		- wp_send_json_success(array(
7648		- 'processed' => true,
7649		- 'item_id' => $item_id,
7650		- 'message' => 'Item processed successfully'
7651		- ));
7652		-
7653		- } catch (Exception $e) {
7654		- $error_message = $e->getMessage();
7655		-
7656		- // Get current attempt count
7657		- $item = $wpdb->get_row($wpdb->prepare(
7658		- "SELECT attempts, max_attempts FROM $table_name WHERE id = %d",
7659		- $item_id
7660		- ));
7661		-
7662		- // Check if we've exhausted retries
7663		- if ($item && $item->attempts >= $item->max_attempts) {
7664		- // Permanently failed
7665		- $wpdb->update(
7666		- $table_name,
7667		- array(
7668		- 'status' => 'failed',
7669		- 'error_message' => $error_message
7670		- ),
7671		- array('id' => $item_id),
7672		- array('%s', '%s'),
7673		- array('%d')
7674		- );
7675		-
7676		- wp_send_json_error(array(
7677		- 'message' => 'Item failed after maximum attempts: ' . $error_message,
7678		- 'permanent_failure' => true,
7679		- 'item_id' => $item_id
7680		- ));
7681		- } else {
7682		- // Mark for retry
7683		- $wpdb->update(
7684		- $table_name,
7685		- array(
7686		- 'status' => 'failed',
7687		- 'error_message' => $error_message
7688		- ),
7689		- array('id' => $item_id),
7690		- array('%s', '%s'),
7691		- array('%d')
7692		- );
7693		-
7694		- wp_send_json_error(array(
7695		- 'message' => 'Item processing failed, will retry: ' . $error_message,
7696		- 'can_retry' => true,
7697		- 'item_id' => $item_id,
7698		- 'attempts' => $item ? $item->attempts : 0
7699		- ));
7700		- }
7701		- }
7702		-}
7703		-
7704		-/**
7705		- * Process a URL from the queue
7706		- */
7707		-private function mxchat_process_queue_url($item_data, $bot_id = 'default', $queue_id = '') {
7708		- $url = isset($item_data['url']) ? $item_data['url'] : '';
7709		-
7710		- if (empty($url)) {
7711		- return new WP_Error('invalid_url', 'URL is empty');
7712		- }
7713		-
7714		- // Get bot-specific API key early (needed for both paths)
7715		- $bot_options = $this->get_bot_options($bot_id);
7716		- $options = !empty($bot_options) ? $bot_options : get_option('mxchat_options');
7717		- $selected_model = $options['embedding_model'] ?? 'text-embedding-ada-002';
7718		-
7719		- if (strpos($selected_model, 'voyage') === 0) {
7720		- $api_key = $options['voyage_api_key'] ?? '';
7721		- } elseif (strpos($selected_model, 'gemini-embedding') === 0) {
7722		- $api_key = $options['gemini_api_key'] ?? '';
7723		- } else {
7724		- $api_key = $options['api_key'] ?? '';
7725		- }
7726		-
7727		- if (empty($api_key)) {
7728		- return new WP_Error('no_api_key', 'API key not configured for bot: ' . $bot_id);
7729		- }
7730		-
7731		- // Check if this is a WooCommerce product URL and WooCommerce is active
7732		- $is_product_url = (strpos($url, '/product/') !== false || strpos($url, '/shop/') !== false);
7733		- $content_type = $is_product_url ? 'product' : 'url';
7734		-
7735		- // Try to get WooCommerce product data if it's a product URL
7736		- if ($is_product_url && class_exists('WooCommerce')) {
7737		- $product_content = $this->mxchat_extract_woocommerce_product_content($url);
7738		-
7739		- if (!empty($product_content)) {
7740		- // Successfully extracted WooCommerce product data with pricing
7741		- $result = MxChat_Utils::submit_content_to_db(
7742		- $product_content,
7743		- $url,
7744		- $api_key,
7745		- null,
7746		- $bot_id,
7747		- 'product'
7748		- );
7749		- return $result;
7750		- }
7751		- // If WooCommerce extraction failed, fall through to HTML extraction
7752		- }
7753		-
7754		- // Fetch URL content (fallback for non-products or when WooCommerce extraction fails)
7755		- $is_likely_pdf = (strtolower(pathinfo(parse_url($url, PHP_URL_PATH) ?: '', PATHINFO_EXTENSION)) === 'pdf');
7756		- $response = wp_remote_get($url, array(
7757		- 'timeout' => $is_likely_pdf ? 120 : 30,
7758		- 'redirection' => 5,
7759		- 'user-agent' => 'MxChat/1.0'
7760		- ));
7761		-
7762		- if (is_wp_error($response)) {
7763		- return $response;
7764		- }
7765		-
7766		- $response_code = wp_remote_retrieve_response_code($response);
7767		- if ($response_code !== 200) {
7768		- return new WP_Error('http_error', 'HTTP ' . $response_code . ' error for: ' . $url);
7769		- }
7770		-
7771		- // Check if URL is a PDF — expand into per-page queue items using the standard PDF pipeline
7772		- if ($this->mxchat_is_pdf_url($url, $response)) {
7773		- return $this->mxchat_expand_pdf_to_queue($url, $response, $bot_id, $queue_id);
7774		- }
7775		-
7776		- $html = wp_remote_retrieve_body($response);
7777		-
7778		- if (empty($html)) {
7779		- return new WP_Error('empty_response', 'Empty response body');
7780		- }
7781		-
7782		- // Extract and sanitize content
7783		- $content = $this->mxchat_extract_main_content($html);
7784		- $sanitized = $this->mxchat_sanitize_content_for_api($content);
7785		-
7786		- if (empty($sanitized)) {
7787		- // Not an error - just no content found (maybe a redirect or empty page)
7788		- return false;
7789		- }
7790		-
7791		- // Submit to database with content_type
7792		- $result = MxChat_Utils::submit_content_to_db(
7793		- $sanitized,
7794		- $url,
7795		- $api_key,
7796		- null,
7797		- $bot_id,
7798		- $content_type
7799		- );
7800		-
7801		- return $result;
7802		-}
7803		-
7804		-/**
7805		- * Expand a PDF URL into per-page queue items using the standard PDF pipeline.
7806		- * Called when a sitemap URL turns out to be a PDF — downloads, parses page count,
7807		- * and adds pdf_page items to the same queue so they process with full progress tracking.
7808		- */
7809		-private function mxchat_expand_pdf_to_queue($pdf_url, $response, $bot_id = 'default', $queue_id = '') {
7810		- set_time_limit(120); // PDFs need extra time for download + parsing
7811		-
7812		- $upload_dir = wp_upload_dir();
7813		- $pdf_filename = sanitize_file_name('mxchat_kb_' . md5($pdf_url) . '.pdf');
7814		- $pdf_path = trailingslashit($upload_dir['path']) . $pdf_filename;
7815		-
7816		- $response_body = wp_remote_retrieve_body($response);
7817		- if (empty($response_body)) {
7818		- return new WP_Error('empty_pdf', 'Empty PDF response for: ' . $pdf_url);
7819		- }
7820		-
7821		- if (!wp_mkdir_p(dirname($pdf_path))) {
7822		- return new WP_Error('dir_error', 'Failed to create upload directory');
7823		- }
7824		-
7825		- file_put_contents($pdf_path, $response_body);
7826		-
7827		- if (!file_exists($pdf_path)) {
7828		- return new WP_Error('save_error', 'Failed to save PDF file');
7829		- }
7830		-
7831		- try {
7832		- $total_pages = $this->mxchat_validate_and_count_pdf_pages($pdf_path);
7833		-
7834		- if ($total_pages === false || $total_pages < 1) {
7835		- wp_delete_file($pdf_path);
7836		- return new WP_Error('no_pages', 'PDF has no pages: ' . $pdf_url);
7837		- }
7838		-
7839		- // Build per-page items identical to mxchat_handle_pdf_for_knowledge_base
7840		- $pages = array();
7841		- for ($i = 1; $i <= $total_pages; $i++) {
7842		- $pages[] = array(
7843		- 'pdf_path' => $pdf_path,
7844		- 'pdf_url' => $pdf_url,
7845		- 'page_number' => $i,
7846		- 'total_pages' => $total_pages
7847		- );
7848		- }
7849		-
7850		- // Add pdf_page items to the SAME queue so the JS picks them up automatically
7851		- if (!empty($queue_id)) {
7852		- $queued_count = $this->mxchat_add_to_queue($queue_id, 'pdf_page', $pages, $bot_id);
7853		- } else {
7854		- // Fallback: create a new PDF queue (shouldn't happen in sitemap flow)
7855		- $new_queue_id = 'pdf_' . md5($pdf_url . time());
7856		- $queued_count = $this->mxchat_add_to_queue($new_queue_id, 'pdf_page', $pages, $bot_id);
7857		- $this->mxchat_set_queue_meta($new_queue_id, 'source_url', $pdf_url);
7858		- $this->mxchat_set_queue_meta($new_queue_id, 'queue_type', 'pdf');
7859		- $this->mxchat_set_queue_meta($new_queue_id, 'total_items', $total_pages);
7860		- $this->mxchat_set_queue_meta($new_queue_id, 'bot_id', $bot_id);
7861		- $this->mxchat_set_queue_meta($new_queue_id, 'pdf_path', $pdf_path);
7862		- $this->mxchat_set_queue_meta($new_queue_id, 'created_at', current_time('mysql'));
7863		- }
7864		-
7865		- if ($queued_count === 0) {
7866		- wp_delete_file($pdf_path);
7867		- return new WP_Error('queue_error', 'Failed to add PDF pages to queue');
7868		- }
7869		-
7870		- // Return true so the original URL item is marked complete
7871		- // The new pdf_page items will be processed in subsequent batches
7872		- return true;
7873		-
7874		- } catch (Exception $e) {
7875		- if (file_exists($pdf_path)) {
7876		- wp_delete_file($pdf_path);
7877		- }
7878		- return new WP_Error('pdf_parse_error', 'Error parsing PDF: ' . $e->getMessage());
7879		- }
7880		-}
7881		-
7882		-/**
7883		- * Legacy: Process a PDF URL inline during sitemap queue processing.
7884		- * @deprecated Use mxchat_expand_pdf_to_queue instead — kept for reference only.
7885		- */
7886		-private function mxchat_process_pdf_url_inline($pdf_url, $response, $api_key, $bot_id = 'default') {
7887		- set_time_limit(120); // PDFs need more time — downloading + parsing all pages
7888		-
7889		- $upload_dir = wp_upload_dir();
7890		- $pdf_filename = sanitize_file_name('mxchat_kb_' . md5($pdf_url) . '.pdf');
7891		- $pdf_path = trailingslashit($upload_dir['path']) . $pdf_filename;
7892		-
7893		- $response_body = wp_remote_retrieve_body($response);
7894		- if (empty($response_body)) {
7895		- return new WP_Error('empty_pdf', 'Empty PDF response');
7896		- }
7897		-
7898		- if (!wp_mkdir_p(dirname($pdf_path))) {
7899		- return new WP_Error('dir_error', 'Failed to create upload directory');
7900		- }
7901		-
7902		- file_put_contents($pdf_path, $response_body);
7903		-
7904		- if (!file_exists($pdf_path)) {
7905		- return new WP_Error('save_error', 'Failed to save PDF file');
7906		- }
7907		-
7908		- try {
7909		- mxchat_load_pdf_parser();
7910		- $parser = new \Smalot\PdfParser\Parser();
7911		- $pdf = $parser->parseFile($pdf_path);
7912		- $pages = $pdf->getPages();
7913		- $total_pages = count($pages);
7914		-
7915		- if ($total_pages < 1) {
7916		- wp_delete_file($pdf_path);
7917		- return new WP_Error('no_pages', 'PDF has no pages');
7918		- }
7919		-
7920		- $processed = 0;
7921		- $skipped_pages = array();
7922		-
7923		- for ($i = 0; $i < $total_pages; $i++) {
7924		- $page_num = $i + 1;
7925		- $text = $pages[$i]->getText();
7926		- if (empty($text)) {
7927		- $skipped_pages[] = 'Page ' . $page_num . ': No text could be extracted — page may contain only images, links, or non-standard encoding';
7928		- continue;
7929		- }
7930		-
7931		- $sanitized = $this->mxchat_sanitize_content_for_api($text);
7932		- if (empty($sanitized)) {
7933		- $skipped_pages[] = 'Page ' . $page_num . ': Text was extracted but contained only special characters, control codes, or unsupported content';
7934		- continue;
7935		- }
7936		-
7937		- $metadata = array(
7938		- 'document_type' => 'pdf',
7939		- 'total_pages' => $total_pages,
7940		- 'current_page' => $page_num,
7941		- 'source_url' => $pdf_url,
7942		- );
7943		-
7944		- $content_with_metadata = wp_json_encode($metadata) . "\n---\n" . $sanitized;
7945		- $page_url = esc_url($pdf_url . '#page=' . $page_num);
7946		-
7947		- MxChat_Utils::submit_content_to_db(
7948		- $content_with_metadata,
7949		- $page_url,
7950		- $api_key,
7951		- null,
7952		- $bot_id,
7953		- 'pdf'
7954		- );
7955		-
7956		- $processed++;
7957		- }
7958		-
7959		- // Clean up the temp PDF file
7960		- wp_delete_file($pdf_path);
7961		-
7962		- if (!empty($skipped_pages)) {
7963		- error_log('MxChat PDF: Skipped ' . count($skipped_pages) . ' of ' . $total_pages . ' pages: ' . implode('; ', $skipped_pages));
7964		- }
7965		-
7966		- return $processed > 0 ? true : false;
7967		-
7968		- } catch (Exception $e) {
7969		- if (file_exists($pdf_path)) {
7970		- wp_delete_file($pdf_path);
7971		- }
7972		- return new WP_Error('pdf_parse_error', 'Error parsing PDF: ' . $e->getMessage());
7973		- }
7974		-}
7975		-
7976		-/**
7977		- * Extract WooCommerce product content including pricing
7978		- *
7979		- * @param string $url The product URL
7980		- * @return string|false Product content with pricing, or false if not found
7981		- */
7982		-private function mxchat_extract_woocommerce_product_content($url) {
7983		- // Try to get product ID from URL
7984		- $product_id = url_to_postid($url);
7985		-
7986		- // If url_to_postid fails, try to extract from URL pattern
7987		- if (!$product_id) {
7988		- $product_slug = '';
7989		-
7990		- // Handle pretty permalinks: /product/product-name/
7991		- if (preg_match('/\/product\/([^\/\?]+)/', $url, $matches)) {
7992		- $product_slug = $matches[1];
7993		- }
7994		-
7995		- if (!empty($product_slug)) {
7996		- $product_post = get_page_by_path($product_slug, OBJECT, 'product');
7997		- if ($product_post) {
7998		- $product_id = $product_post->ID;
7999		- }
8000		- }
8001		- }
8002		-
8003		- if (!$product_id) {
8004		- return false;
8005		- }
8006		-
8007		- // Get WooCommerce product object
8008		- $product = wc_get_product($product_id);
8009		-
8010		- if (!$product) {
8011		- return false;
8012		- }
8013		-
8014		- // Build product content with pricing (similar to mxchat_store_product_embedding)
8015		- $title = $product->get_name();
8016		- $description = $product->get_description();
8017		- $short_description = $product->get_short_description();
8018		- $sku = $product->get_sku();
8019		-
8020		- // Get pricing information
8021		- $regular_price = $product->get_regular_price();
8022		- $sale_price = $product->get_sale_price();
8023		- $price = $product->get_price(); // Current active price
8024		-
8025		- // Get currency symbol
8026		- $currency_symbol = get_woocommerce_currency_symbol();
8027		-
8028		- // Format content
8029		- $content = $title . "\n\n";
8030		-
8031		- if (!empty($short_description)) {
8032		- $content .= "Short Description: " . wp_strip_all_tags($short_description) . "\n\n";
8033		- }
8034		-
8035		- if (!empty($description)) {
8036		- $content .= wp_strip_all_tags($description) . "\n\n";
8037		- }
8038		-
8039		- // Add pricing information
8040		- if (!empty($regular_price)) {
8041		- $content .= "Price: " . $currency_symbol . $regular_price . "\n";
8042		- } elseif (!empty($price)) {
8043		- $content .= "Price: " . $currency_symbol . $price . "\n";
8044		- }
8045		-
8046		- if (!empty($sale_price) && $sale_price !== $regular_price) {
8047		- $content .= "Sale Price: " . $currency_symbol . $sale_price . "\n";
8048		- }
8049		-
8050		- // Handle variable products - show price range
8051		- if ($product->is_type('variable')) {
8052		- $min_price = $product->get_variation_price('min');
8053		- $max_price = $product->get_variation_price('max');
8054		- if ($min_price !== $max_price) {
8055		- $content .= "Price Range: " . $currency_symbol . $min_price . " - " . $currency_symbol . $max_price . "\n";
8056		- }
8057		- }
8058		-
8059		- if (!empty($sku)) {
8060		- $content .= "SKU: " . $sku . "\n";
8061		- }
8062		-
8063		- // Get product categories
8064		- $categories = wp_get_post_terms($product_id, 'product_cat', array('fields' => 'names'));
8065		- if (!empty($categories) && !is_wp_error($categories)) {
8066		- $content .= "Categories: " . implode(', ', $categories) . "\n";
8067		- }
8068		-
8069		- // Get Custom Product Tabs (supports "Custom Product Tabs for WooCommerce" by Code Parrots)
8070		- $custom_tabs = get_post_meta($product_id, 'yikes_woo_products_tabs', true);
8071		- if (!empty($custom_tabs) && is_array($custom_tabs)) {
8072		- foreach ($custom_tabs as $tab) {
8073		- $tab_title = isset($tab['title']) ? $tab['title'] : (isset($tab['tab_title']) ? $tab['tab_title'] : '');
8074		- $tab_content = isset($tab['content']) ? $tab['content'] : '';
8075		-
8076		- if (!empty($tab_title) && !empty($tab_content)) {
8077		- $content .= "\n" . $tab_title . ": " . wp_strip_all_tags($tab_content) . "\n";
8078		- }
8079		- }
8080		- }
8081		-
8082		- // Also check for reusable/saved tabs applied to this product
8083		- $applied_saved_tabs = get_post_meta($product_id, 'yikes_woo_reusable_products_tabs_applied', true);
8084		- if (!empty($applied_saved_tabs) && is_array($applied_saved_tabs)) {
8085		- $saved_tabs = get_option('yikes_woo_reusable_products_tabs', array());
8086		- if (!empty($saved_tabs) && is_array($saved_tabs)) {
8087		- foreach ($applied_saved_tabs as $saved_tab_id) {
8088		- if (isset($saved_tabs[$saved_tab_id])) {
8089		- $tab = $saved_tabs[$saved_tab_id];
8090		- $tab_title = isset($tab['title']) ? $tab['title'] : (isset($tab['tab_title']) ? $tab['tab_title'] : '');
8091		- $tab_content = isset($tab['content']) ? $tab['content'] : '';
8092		-
8093		- if (!empty($tab_title) && !empty($tab_content)) {
8094		- $content .= "\n" . $tab_title . ": " . wp_strip_all_tags($tab_content) . "\n";
8095		- }
8096		- }
8097		- }
8098		- }
8099		- }
8100		-
8101		- return $this->mxchat_sanitize_content_for_api($content);
8102		-}
8103		-
8104		-/**
8105		- * Process a PDF page from the queue
8106		- */
8107		-private function mxchat_process_queue_pdf_page($item_data, $bot_id = 'default') {
8108		- $pdf_path = isset($item_data['pdf_path']) ? $item_data['pdf_path'] : '';
8109		- $pdf_url = isset($item_data['pdf_url']) ? $item_data['pdf_url'] : '';
8110		- $page_number = isset($item_data['page_number']) ? absint($item_data['page_number']) : 0;
8111		- $total_pages = isset($item_data['total_pages']) ? absint($item_data['total_pages']) : 0;
8112		-
8113		- if (empty($pdf_path) || !file_exists($pdf_path)) {
8114		- return new WP_Error('pdf_not_found', 'PDF file not found: ' . $pdf_path);
8115		- }
8116		-
8117		- if ($page_number < 1) {
8118		- return new WP_Error('invalid_page', 'Invalid page number');
8119		- }
8120		-
8121		- try {
8122		- mxchat_load_pdf_parser();
8123		- $parser = new \Smalot\PdfParser\Parser();
8124		- $pdf = $parser->parseFile($pdf_path);
8125		- $pages = $pdf->getPages();
8126		-
8127		- if (!isset($pages[$page_number - 1])) {
8128		- return new WP_Error('page_not_found', 'Page ' . $page_number . ' not found in PDF');
8129		- }
8130		-
8131		- $text = $pages[$page_number - 1]->getText();
8132		-
8133		- if (empty($text)) {
8134		- return new WP_Error('empty_page', 'Page ' . $page_number . ': No text could be extracted — page may contain only images, links, or non-standard encoding');
8135		- }
8136		-
8137		- $sanitized = $this->mxchat_sanitize_content_for_api($text);
8138		-
8139		- if (empty($sanitized)) {
8140		- 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');
8141		- }
8142		-
8143		- // Create metadata
8144		- $metadata = array(
8145		- 'document_type' => 'pdf',
8146		- 'total_pages' => $total_pages,
8147		- 'current_page' => $page_number,
8148		- 'source_url' => $pdf_url
8149		- );
8150		-
8151		- $content_with_metadata = wp_json_encode($metadata) . "\n---\n" . $sanitized;
8152		- $page_url = esc_url($pdf_url . "#page=" . $page_number);
8153		-
8154		- // Get bot-specific API key
8155		- $bot_options = $this->get_bot_options($bot_id);
8156		- $options = !empty($bot_options) ? $bot_options : get_option('mxchat_options');
8157		- $selected_model = $options['embedding_model'] ?? 'text-embedding-ada-002';
8158		-
8159		- if (strpos($selected_model, 'voyage') === 0) {
8160		- $api_key = $options['voyage_api_key'] ?? '';
8161		- } elseif (strpos($selected_model, 'gemini-embedding') === 0) {
8162		- $api_key = $options['gemini_api_key'] ?? '';
8163		- } else {
8164		- $api_key = $options['api_key'] ?? '';
8165		- }
8166		-
8167		- if (empty($api_key)) {
8168		- return new WP_Error('no_api_key', 'API key not configured for bot: ' . $bot_id);
8169		- }
8170		-
8171		- // Submit to database - UPDATED 2.5.6: Added content_type 'pdf'
8172		- $result = MxChat_Utils::submit_content_to_db(
8173		- $content_with_metadata,
8174		- $page_url,
8175		- $api_key,
8176		- null,
8177		- $bot_id,
8178		- 'pdf'
8179		- );
8180		-
8181		- return $result;
8182		-
8183		- } catch (Exception $e) {
8184		- return new WP_Error('pdf_parse_error', 'Error parsing PDF: ' . $e->getMessage());
8185		- }
8186		-}
8187		-
8188		-/**
8189		- * AJAX: Get queue processing status
8190		- */
8191		-public function ajax_mxchat_get_queue_status() {
8192		- // Verify nonce and permissions
8193		- check_ajax_referer('mxchat_queue_nonce', 'nonce');
8194		-
8195		- if (!current_user_can('manage_options')) {
8196		- wp_send_json_error('Unauthorized access');
8197		- }
8198		-
8199		- $queue_id = isset($_POST['queue_id']) ? sanitize_text_field($_POST['queue_id']) : '';
8200		-
8201		- if (empty($queue_id)) {
8202		- wp_send_json_error('Missing queue ID');
8203		- }
8204		-
8205		- global $wpdb;
8206		- $table_name = $wpdb->prefix . 'mxchat_processing_queue';
8207		-
8208		- // Get counts by status
8209		- $counts = $wpdb->get_results($wpdb->prepare(
8210		- "SELECT status, COUNT(*) as count
8211		- FROM $table_name
8212		- WHERE queue_id = %s
8213		- GROUP BY status",
8214		- $queue_id
8215		- ), OBJECT_K);
8216		-
8217		- $total = 0;
8218		- $completed = 0;
8219		- $failed = 0;
8220		- $processing = 0;
8221		- $pending = 0;
8222		-
8223		- foreach ($counts as $status => $data) {
8224		- $count = absint($data->count);
8225		- $total += $count;
8226		-
8227		- switch ($status) {
8228		- case 'completed':
8229		- $completed = $count;
8230		- break;
8231		- case 'failed':
8232		- $failed = $count;
8233		- break;
8234		- case 'processing':
8235		- $processing = $count;
8236		- break;
8237		- case 'pending':
8238		- $pending = $count;
8239		- break;
8240		- }
8241		- }
8242		-
8243		- // Calculate percentage
8244		- $percentage = $total > 0 ? round((($completed + $failed) / $total) * 100) : 0;
8245		-
8246		- // Get failed items details (include all failed items, not just those that exhausted retries)
8247		- $failed_items = array();
8248		- if ($failed > 0) {
8249		- $failed_items = $wpdb->get_results($wpdb->prepare(
8250		- "SELECT item_type, item_data, error_message, attempts
8251		- FROM $table_name
8252		- WHERE queue_id = %s
8253		- AND status = 'failed'
8254		- ORDER BY id DESC
8255		- LIMIT 50",
8256		- $queue_id
8257		- ));
8258		- }
8259		-
8260		- // Get queue metadata
8261		- $source_url = $this->mxchat_get_queue_meta($queue_id, 'source_url');
8262		- $queue_type = $this->mxchat_get_queue_meta($queue_id, 'queue_type');
8263		-
8264		- // Determine if queue is complete
8265		- $is_complete = ($pending === 0 && $processing === 0);
8266		-
8267		- wp_send_json_success(array(
8268		- 'queue_id' => $queue_id,
8269		- 'queue_type' => $queue_type,
8270		- 'source_url' => $source_url,
8271		- 'total' => $total,
8272		- 'completed' => $completed,
8273		- 'failed' => $failed,
8274		- 'processing' => $processing,
8275		- 'pending' => $pending,
8276		- 'percentage' => $percentage,
8277		- 'is_complete' => $is_complete,
8278		- 'failed_items' => $failed_items,
8279		- 'status' => $is_complete ? 'complete' : 'processing'
8280		- ));
8281		-}
8282		-
8283		-/**
8284		- * AJAX: Clear completed queue
8285		- */
8286		-public function ajax_mxchat_clear_queue() {
8287		- // Verify nonce and permissions
8288		- check_ajax_referer('mxchat_queue_nonce', 'nonce');
8289		-
8290		- if (!current_user_can('manage_options')) {
8291		- wp_send_json_error('Unauthorized access');
8292		- }
8293		-
8294		- $queue_id = isset($_POST['queue_id']) ? sanitize_text_field($_POST['queue_id']) : '';
8295		-
8296		- if (empty($queue_id)) {
8297		- wp_send_json_error('Missing queue ID');
8298		- }
8299		-
8300		- global $wpdb;
8301		- $table_name = $wpdb->prefix . 'mxchat_processing_queue';
8302		- $meta_table = $wpdb->prefix . 'mxchat_queue_meta';
8303		-
8304		- // Delete queue items
8305		- $wpdb->delete(
8306		- $table_name,
8307		- array('queue_id' => $queue_id),
8308		- array('%s')
8309		- );
8310		-
8311		- // Delete queue metadata
8312		- $wpdb->delete(
8313		- $meta_table,
8314		- array('queue_id' => $queue_id),
8315		- array('%s')
8316		- );
8317		-
8318		- wp_send_json_success(array(
8319		- 'message' => 'Queue cleared successfully'
8320		- ));
8321		-}
8322		-
8323		-/**
8324		- * AJAX: Retry failed items in queue
8325		- */
8326		-public function ajax_mxchat_retry_failed() {
8327		- // Verify nonce and permissions
8328		- check_ajax_referer('mxchat_queue_nonce', 'nonce');
8329		-
8330		- if (!current_user_can('manage_options')) {
8331		- wp_send_json_error('Unauthorized access');
8332		- }
8333		-
8334		- $queue_id = isset($_POST['queue_id']) ? sanitize_text_field($_POST['queue_id']) : '';
8335		-
8336		- if (empty($queue_id)) {
8337		- wp_send_json_error('Missing queue ID');
8338		- }
8339		-
8340		- global $wpdb;
8341		- $table_name = $wpdb->prefix . 'mxchat_processing_queue';
8342		-
8343		- // Reset failed items to pending and reset attempt count
8344		- $updated = $wpdb->update(
8345		- $table_name,
8346		- array(
8347		- 'status' => 'pending',
8348		- 'attempts' => 0,
8349		- 'error_message' => null
8350		- ),
8351		- array(
8352		- 'queue_id' => $queue_id,
8353		- 'status' => 'failed'
8354		- ),
8355		- array('%s', '%d', '%s'),
8356		- array('%s', '%s')
8357		- );
8358		-
8359		- wp_send_json_success(array(
8360		- 'message' => 'Reset ' . $updated . ' failed items for retry',
8361		- 'reset_count' => $updated
8362		- ));
8363		-}
8364		-
8365		-
8366		-public function ajax_mxchat_mark_queue_complete() {
8367		- check_ajax_referer('mxchat_queue_nonce', 'nonce');
8368		-
8369		- if (!current_user_can('manage_options')) {
8370		- wp_send_json_error('Unauthorized access');
8371		- }
8372		-
8373		- $queue_id = isset($_POST['queue_id']) ? sanitize_text_field($_POST['queue_id']) : '';
8374		-
8375		- if (empty($queue_id)) {
8376		- wp_send_json_error('Missing queue ID');
8377		- }
8378		-
8379		- // Clear active queue transients
8380		- if (strpos($queue_id, 'sitemap_') === 0) {
8381		- delete_transient('mxchat_active_queue_sitemap');
8382		- } else if (strpos($queue_id, 'pdf_') === 0) {
8383		- delete_transient('mxchat_active_queue_pdf');
8384		- }
8385		-
8386		- wp_send_json_success(array('message' => 'Queue marked as complete'));
8387		-}
8388		-
8389		-
8390		- // ========================================
8391		- // STATIC ACCESS METHODS
8392		- // ========================================
8393		-
8394		- /**
8395		- * Get singleton instance
8396		- */
8397		- public static function get_instance() {
8398		- static $instance = null;
8399		- if ($instance === null) {
8400		- $instance = new self();
8401		- }
8402		- return $instance;
8403		- }
8404		-}
8405		-
8406		-// 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	+ $this->mxchat_init_hooks();
	22	+ }
	23	+
	24	+ /**
	25	+ * Initialize WordPress hooks for content processing
	26	+ */
	27	+ private function mxchat_init_hooks() {
	28	+ // Admin post handlers for form submissions
	29	+ add_action('admin_post_mxchat_submit_content', array($this, 'mxchat_handle_content_submission'));
	30	+ add_action('admin_post_mxchat_submit_sitemap', array($this, 'mxchat_handle_sitemap_submission'));
	31	+ add_action('admin_post_mxchat_stop_processing', array($this, 'mxchat_stop_processing'));
	32	+
	33	+ // AJAX handlers for real-time processing and status updates
	34	+ add_action('wp_ajax_mxchat_get_status_updates', array($this, 'mxchat_ajax_get_status_updates'));
	35	+ add_action('wp_ajax_mxchat_dismiss_completed_status', array($this, 'mxchat_ajax_dismiss_completed_status'));
	36	+ add_action('wp_ajax_mxchat_get_content_list', array($this, 'ajax_mxchat_get_content_list'));
	37	+ add_action('wp_ajax_mxchat_process_selected_content', array($this, 'ajax_mxchat_process_selected_content'));
	38	+ add_action('wp_ajax_mxchat_manual_batch_process', array($this, 'ajax_manual_batch_process'));
	39	+ add_action('mxchat_delete_content', array($this, 'mxchat_delete_from_pinecone_by_url'), 10, 1);
	40	+
	41	+
	42	+ // Cron handlers for background processing
	43	+ add_action('mxchat_process_sitemap_urls', array($this, 'mxchat_process_sitemap_urls_cron'), 10, 5);
	44	+ add_action('mxchat_process_pdf_pages', array($this, 'mxchat_process_pdf_pages_cron'), 10, 5);
	45	+
	46	+ // WordPress post management hooks - UPDATED FOR BETTER STATUS TRACKING
	47	+ add_action('pre_post_update', array($this, 'mxchat_store_pre_update_status'), 10, 2);
	48	+ add_action('post_updated', array($this, 'mxchat_handle_post_update'), 10, 3);
	49	+ add_action('before_delete_post', array($this, 'mxchat_handle_post_delete'));
	50	+ add_action('wp_trash_post', array($this, 'mxchat_handle_post_delete'));
	51	+ add_action('wp_ajax_mxchat_save_inline_prompt', array($this, 'mxchat_save_inline_prompt'));
	52	+ add_action('admin_post_mxchat_delete_pinecone_prompt', array($this, 'mxchat_handle_pinecone_prompt_delete'));
	53	+ add_action('wp_ajax_mxchat_delete_pinecone_prompt', array($this, 'ajax_mxchat_delete_pinecone_prompt'));
	54	+ add_action('wp_ajax_mxchat_update_role_restriction', array($this, 'ajax_mxchat_update_role_restriction'));
	55	+
	56	+ // WooCommerce product hooks (if WooCommerce is active)
	57	+ if (class_exists('WooCommerce')) {
	58	+ add_action('pre_post_update', array($this, 'mxchat_store_pre_update_status'), 10, 2); // Same hook for products
	59	+ add_action('save_post_product', array($this, 'mxchat_handle_product_change'), 10, 3);
	60	+ add_action('wp_trash_post', array($this, 'mxchat_handle_product_delete'));
	61	+ add_action('before_delete_post', array($this, 'mxchat_handle_product_delete'));
	62	+ }
	63	+
	64	+ }
	65	+
	66	+ /**
	67	+ * Get current options (refreshed)
	68	+ */
	69	+ private function mxchat_get_options() {
	70	+ if (empty($this->options)) {
	71	+ $this->options = get_option('mxchat_options', array());
	72	+ }
	73	+ return $this->options;
	74	+ }
	75	+
	76	+public function ajax_manual_batch_process() {
	77	+ try {
	78	+ // Verify nonce and permissions
	79	+ check_ajax_referer('mxchat_status_nonce', 'nonce');
	80	+
	81	+ if (!current_user_can('manage_options')) {
	82	+ wp_send_json_error('Unauthorized access');
	83	+ }
	84	+
	85	+ $process_type = sanitize_text_field($_POST['process_type'] ?? '');
	86	+ $url = sanitize_text_field($_POST['url'] ?? '');
	87	+
	88	+ if (empty($process_type) || empty($url)) {
	89	+ wp_send_json_error('Missing required parameters');
	90	+ }
	91	+
	92	+ // Debug logging
	93	+ //error_log('MANUAL BATCH DEBUG: Process type: ' . $process_type);
	94	+ //error_log('MANUAL BATCH DEBUG: URL: ' . $url);
	95	+
	96	+ // FIXED: Extract bot_id from stored status instead of POST data
	97	+ $bot_id = 'default';
	98	+
	99	+ if ($process_type === 'pdf') {
	100	+ $status_key = sanitize_key('mxchat_pdf_status_' . md5($url));
	101	+ $status = get_transient($status_key);
	102	+ //error_log('MANUAL BATCH DEBUG: Status key: ' . $status_key);
	103	+ //error_log('MANUAL BATCH DEBUG: Status data: ' . print_r($status, true));
	104	+
	105	+ if ($status && isset($status['bot_id'])) {
	106	+ $bot_id = $status['bot_id'];
	107	+ //error_log('MANUAL BATCH DEBUG: Bot ID from status: ' . $bot_id);
	108	+ } else {
	109	+ //error_log('MANUAL BATCH DEBUG: No bot_id in status, using default');
	110	+ }
	111	+ } elseif ($process_type === 'sitemap') {
	112	+ $status_key = sanitize_key('mxchat_sitemap_status_' . md5($url));
	113	+ $status = get_transient($status_key);
	114	+ if ($status && isset($status['bot_id'])) {
	115	+ $bot_id = $status['bot_id'];
	116	+ }
	117	+ }
	118	+
	119	+ //error_log('MANUAL BATCH DEBUG: Final bot_id: ' . $bot_id);
	120	+
	121	+ $processed = 0;
	122	+
	123	+ if ($process_type === 'pdf') {
	124	+ $processed = $this->mxchat_manual_process_pdf_batch($url);
	125	+ //error_log('MANUAL BATCH DEBUG: PDF processing returned: ' . $processed);
	126	+ } elseif ($process_type === 'sitemap') {
	127	+ $processed = $this->mxchat_manual_process_sitemap_batch($url);
	128	+ }
	129	+
	130	+ if ($processed > 0) {
	131	+ wp_send_json_success(array(
	132	+ 'message' => "Processed {$processed} items successfully",
	133	+ 'processed' => $processed,
	134	+ 'bot_id' => $bot_id
	135	+ ));
	136	+ } else {
	137	+ // Enhanced error response with debugging info
	138	+ wp_send_json_error(array(
	139	+ 'message' => 'No items were processed',
	140	+ 'debug_info' => array(
	141	+ 'process_type' => $process_type,
	142	+ 'url' => $url,
	143	+ 'bot_id' => $bot_id,
	144	+ 'status_exists' => !empty($status),
	145	+ 'status_data' => $status
	146	+ )
	147	+ ));
	148	+ }
	149	+
	150	+ } catch (Exception $e) {
	151	+ //error_log('MANUAL BATCH DEBUG: Exception: ' . $e->getMessage());
	152	+ wp_send_json_error('Processing failed: ' . $e->getMessage());
	153	+ }
	154	+}
	155	+
	156	+private function mxchat_manual_process_pdf_batch($pdf_url) {
	157	+ try {
	158	+ //error_log('MANUAL PDF DEBUG: Starting batch processing for: ' . $pdf_url);
	159	+
	160	+ $status_key = sanitize_key('mxchat_pdf_status_' . md5($pdf_url));
	161	+ $status = get_transient($status_key);
	162	+
	163	+ //error_log('MANUAL PDF DEBUG: Status key: ' . $status_key);
	164	+ //error_log('MANUAL PDF DEBUG: Status data: ' . print_r($status, true));
	165	+
	166	+ if (!$status || $status['status'] !== 'processing') {
	167	+ //error_log('MANUAL PDF DEBUG: No processing status found or status is not processing');
	168	+ //error_log('MANUAL PDF DEBUG: Status: ' . ($status ? $status['status'] : 'NULL'));
	169	+ return 0;
	170	+ }
	171	+
	172	+ // FIXED: Extract bot_id from status
	173	+ $bot_id = $status['bot_id'] ?? 'default';
	174	+ //error_log('MANUAL PDF DEBUG: Bot ID from status: ' . $bot_id);
	175	+
	176	+ // Get current progress
	177	+ $current_page = $status['processed_pages'] ?? 0;
	178	+ $total_pages = $status['total_pages'] ?? 0;
	179	+
	180	+ //error_log('MANUAL PDF DEBUG: Current page: ' . $current_page . ', Total pages: ' . $total_pages);
	181	+
	182	+ if ($current_page >= $total_pages) {
	183	+ //error_log('MANUAL PDF DEBUG: Already completed');
	184	+ return 0;
	185	+ }
	186	+
	187	+ // Try to download the PDF again for processing
	188	+ //error_log('MANUAL PDF DEBUG: Attempting to download PDF');
	189	+ $response = wp_remote_get($pdf_url, array('timeout' => 30));
	190	+
	191	+ if (is_wp_error($response)) {
	192	+ //error_log('MANUAL PDF DEBUG: Failed to download PDF: ' . $response->get_error_message());
	193	+ return 0;
	194	+ }
	195	+
	196	+ $pdf_content = wp_remote_retrieve_body($response);
	197	+ if (empty($pdf_content)) {
	198	+ //error_log('MANUAL PDF DEBUG: Empty PDF content');
	199	+ return 0;
	200	+ }
	201	+
	202	+ //error_log('MANUAL PDF DEBUG: PDF content size: ' . strlen($pdf_content) . ' bytes');
	203	+
	204	+ // Save PDF temporarily
	205	+ $upload_dir = wp_upload_dir();
	206	+ $temp_pdf_path = trailingslashit($upload_dir['path']) . 'temp_manual_' . time() . '.pdf';
	207	+ file_put_contents($temp_pdf_path, $pdf_content);
	208	+
	209	+ //error_log('MANUAL PDF DEBUG: Temp PDF saved to: ' . $temp_pdf_path);
	210	+
	211	+ // Process 5 pages directly with bot_id
	212	+ $processed = $this->mxchat_process_pdf_pages_direct($temp_pdf_path, $pdf_url, $current_page, 5, $bot_id);
	213	+
	214	+ //error_log('MANUAL PDF DEBUG: Direct processing returned: ' . $processed);
	215	+
	216	+ // Clean up temp file
	217	+ if (file_exists($temp_pdf_path)) {
	218	+ wp_delete_file($temp_pdf_path);
	219	+ //error_log('MANUAL PDF DEBUG: Cleaned up temp file');
	220	+ }
	221	+
	222	+ return $processed;
	223	+
	224	+ } catch (Exception $e) {
	225	+ //error_log('MANUAL PDF DEBUG: Exception in manual batch: ' . $e->getMessage());
	226	+ return 0;
	227	+ }
	228	+}
	229	+
	230	+/**
	231	+ * Process PDF pages directly without cron
	232	+ */
	233	+private function mxchat_process_pdf_pages_direct($pdf_path, $pdf_url, $start_page, $batch_size, $bot_id = 'default') {
	234	+ try {
	235	+ if (!file_exists($pdf_path)) {
	236	+ //error_log('Direct PDF: File not found at ' . $pdf_path);
	237	+ return 0;
	238	+ }
	239	+
	240	+ $parser = new \Smalot\PdfParser\Parser();
	241	+ $pdf = $parser->parseFile($pdf_path);
	242	+ $pages = $pdf->getPages();
	243	+
	244	+ $status_key = sanitize_key('mxchat_pdf_status_' . md5($pdf_url));
	245	+ $status = get_transient($status_key);
	246	+
	247	+ if (!$status) {
	248	+ return 0;
	249	+ }
	250	+
	251	+ // UPDATED: Get bot-specific options
	252	+ $bot_options = $this->get_bot_options($bot_id);
	253	+ $options = !empty($bot_options) ? $bot_options : get_option('mxchat_options');
	254	+ $selected_model = $options['embedding_model'] ?? 'text-embedding-ada-002';
	255	+
	256	+ if (strpos($selected_model, 'voyage') === 0) {
	257	+ $api_key = $options['voyage_api_key'] ?? '';
	258	+ } elseif (strpos($selected_model, 'gemini-embedding') === 0) {
	259	+ $api_key = $options['gemini_api_key'] ?? '';
	260	+ } else {
	261	+ $api_key = $options['api_key'] ?? '';
	262	+ }
	263	+
	264	+ if (empty($api_key)) {
	265	+ //error_log('Direct PDF: No API key for bot: ' . $bot_id);
	266	+ return 0;
	267	+ }
	268	+
	269	+ $processed = 0;
	270	+ $end_page = min($start_page + $batch_size, count($pages));
	271	+
	272	+ for ($i = $start_page; $i < $end_page; $i++) {
	273	+ try {
	274	+ $page_number = $i + 1;
	275	+ $text = $pages[$i]->getText();
	276	+
	277	+ if (empty($text)) {
	278	+ //error_log('Direct PDF: Empty text on page ' . $page_number);
	279	+ continue;
	280	+ }
	281	+
	282	+ $sanitized_content = $this->mxchat_sanitize_content_for_api($text);
	283	+ if (empty($sanitized_content)) {
	284	+ //error_log('Direct PDF: No content after sanitization on page ' . $page_number);
	285	+ continue;
	286	+ }
	287	+
	288	+ // UPDATED: Use bot-specific embedding generation
	289	+ $embedding_vector = $this->mxchat_generate_embedding($sanitized_content, $bot_id);
	290	+ if (is_string($embedding_vector)) {
	291	+ //error_log('Direct PDF: Embedding failed on page ' . $page_number . ': ' . $embedding_vector);
	292	+ continue;
	293	+ }
	294	+
	295	+ // Create metadata
	296	+ $metadata = array(
	297	+ 'document_type' => 'pdf',
	298	+ 'total_pages' => count($pages),
	299	+ 'current_page' => $page_number,
	300	+ 'source_url' => $pdf_url
	301	+ );
	302	+
	303	+ $content_with_metadata = wp_json_encode($metadata) . "\n---\n" . $sanitized_content;
	304	+ $page_url = esc_url($pdf_url . "#page=" . $page_number);
	305	+
	306	+ // UPDATED: Pass bot_id to database submission
	307	+ $db_result = MxChat_Utils::submit_content_to_db($content_with_metadata, $page_url, $api_key, null, $bot_id);
	308	+
	309	+ if (is_wp_error($db_result)) {
	310	+ //error_log('Direct PDF: DB error on page ' . $page_number . ': ' . $db_result->get_error_message());
	311	+ continue;
	312	+ }
	313	+
	314	+ $processed++;
	315	+ //error_log('Direct PDF: Successfully processed page ' . $page_number . ' for bot: ' . $bot_id);
	316	+
	317	+ // Update status
	318	+ $status['processed_pages'] = $i + 1;
	319	+ $status['last_update'] = time();
	320	+ $status['percentage'] = round(($status['processed_pages'] / $status['total_pages']) * 100);
	321	+ set_transient($status_key, $status, DAY_IN_SECONDS);
	322	+
	323	+ } catch (Exception $e) {
	324	+ //error_log('Direct PDF: Error processing page ' . ($i + 1) . ': ' . $e->getMessage());
	325	+ continue;
	326	+ }
	327	+ }
	328	+
	329	+ // Check if completed
	330	+ if ($status['processed_pages'] >= $status['total_pages']) {
	331	+ $status['status'] = 'complete';
	332	+ set_transient($status_key, $status, DAY_IN_SECONDS);
	333	+ //error_log('Direct PDF: Processing completed for bot: ' . $bot_id);
	334	+ }
	335	+
	336	+ return $processed;
	337	+
	338	+ } catch (Exception $e) {
	339	+ //error_log('Direct PDF processing error: ' . $e->getMessage());
	340	+ return 0;
	341	+ }
	342	+}
	343	+
	344	+
	345	+/**
	346	+ * Process a small sitemap batch manually - DIRECT PROCESSING
	347	+ */
	348	+private function mxchat_manual_process_sitemap_batch($sitemap_url) {
	349	+ try {
	350	+ $status_key = sanitize_key('mxchat_sitemap_status_' . md5($sitemap_url));
	351	+ $status = get_transient($status_key);
	352	+
	353	+ if (!$status || $status['status'] !== 'processing') {
	354	+ return 0;
	355	+ }
	356	+
	357	+ // FIXED: Extract bot_id from status
	358	+ $bot_id = $status['bot_id'] ?? 'default';
	359	+
	360	+ //error_log('Manual Sitemap: Starting direct processing for ' . $sitemap_url . ' with bot: ' . $bot_id);
	361	+
	362	+ // Re-fetch the sitemap to get URLs
	363	+ $response = wp_remote_get($sitemap_url, array('timeout' => 30));
	364	+ if (is_wp_error($response)) {
	365	+ //error_log('Manual Sitemap: Failed to fetch sitemap');
	366	+ return 0;
	367	+ }
	368	+
	369	+ $sitemap_content = wp_remote_retrieve_body($response);
	370	+ $xml = simplexml_load_string($sitemap_content);
	371	+
	372	+ if (!$xml) {
	373	+ //error_log('Manual Sitemap: Invalid XML');
	374	+ return 0;
	375	+ }
	376	+
	377	+ $urls = array();
	378	+ foreach ($xml->url as $url_element) {
	379	+ $urls[] = (string)$url_element->loc;
	380	+ }
	381	+
	382	+ $current_processed = $status['processed_urls'] ?? 0;
	383	+ $batch_size = 50;
	384	+ $processed = 0;
	385	+
	386	+ // Process next batch of URLs
	387	+ for ($i = $current_processed; $i < min($current_processed + $batch_size, count($urls)); $i++) {
	388	+ $url = $urls[$i];
	389	+
	390	+ // UPDATED: Pass bot_id to single URL processing
	391	+ if ($this->mxchat_process_single_url_direct($url, $bot_id)) {
	392	+ $processed++;
	393	+ }
	394	+
	395	+ // Update status
	396	+ $status['processed_urls'] = $i + 1;
	397	+ $status['last_update'] = time();
	398	+ $status['percentage'] = round(($status['processed_urls'] / $status['total_urls']) * 100);
	399	+ set_transient($status_key, $status, DAY_IN_SECONDS);
	400	+ }
	401	+
	402	+ // Check if completed
	403	+ if ($status['processed_urls'] >= $status['total_urls']) {
	404	+ $status['status'] = 'complete';
	405	+ set_transient($status_key, $status, DAY_IN_SECONDS);
	406	+ }
	407	+
	408	+ //error_log('Manual Sitemap: Processed ' . $processed . ' URLs for bot: ' . $bot_id);
	409	+ return $processed;
	410	+
	411	+ } catch (Exception $e) {
	412	+ //error_log('Manual sitemap batch error: ' . $e->getMessage());
	413	+ return 0;
	414	+ }
	415	+}
	416	+
	417	+
	418	+/**
	419	+ * Process a single URL directly
	420	+ */
	421	+private function mxchat_process_single_url_direct($url, $bot_id = 'default') {
	422	+ try {
	423	+ $response = wp_remote_get($url, array('timeout' => 30));
	424	+ if (is_wp_error($response)) {
	425	+ //error_log('Single URL processing failed for ' . $url . ': ' . $response->get_error_message());
	426	+ return false;
	427	+ }
	428	+
	429	+ $html = wp_remote_retrieve_body($response);
	430	+ $content = $this->mxchat_extract_main_content($html);
	431	+ $sanitized = $this->mxchat_sanitize_content_for_api($content);
	432	+
	433	+ if (empty($sanitized)) {
	434	+ //error_log('Single URL processing: No content found for ' . $url);
	435	+ return false;
	436	+ }
	437	+
	438	+ // UPDATED: Get bot-specific options and API key
	439	+ $bot_options = $this->get_bot_options($bot_id);
	440	+ $options = !empty($bot_options) ? $bot_options : get_option('mxchat_options');
	441	+ $selected_model = $options['embedding_model'] ?? 'text-embedding-ada-002';
	442	+
	443	+ if (strpos($selected_model, 'voyage') === 0) {
	444	+ $api_key = $options['voyage_api_key'] ?? '';
	445	+ } elseif (strpos($selected_model, 'gemini-embedding') === 0) {
	446	+ $api_key = $options['gemini_api_key'] ?? '';
	447	+ } else {
	448	+ $api_key = $options['api_key'] ?? '';
	449	+ }
	450	+
	451	+ if (empty($api_key)) {
	452	+ //error_log('Single URL processing: No API key configured for bot: ' . $bot_id);
	453	+ return false;
	454	+ }
	455	+
	456	+ // UPDATED: Pass bot_id to database submission
	457	+ $result = MxChat_Utils::submit_content_to_db($sanitized, $url, $api_key, null, $bot_id);
	458	+
	459	+ $success = !is_wp_error($result);
	460	+
	461	+ if ($success) {
	462	+ //error_log('Single URL processing: Successfully processed ' . $url . ' for bot: ' . $bot_id);
	463	+ } else {
	464	+ //error_log('Single URL processing: Failed to store ' . $url . ' for bot: ' . $bot_id . ': ' . $result->get_error_message());
	465	+ }
	466	+
	467	+ return $success;
	468	+
	469	+ } catch (Exception $e) {
	470	+ //error_log('Single URL processing error: ' . $e->getMessage());
	471	+ return false;
	472	+ }
	473	+}
	474	+
	475	+ // ========================================
	476	+ // MAIN CONTENT SUBMISSION HANDLERS
	477	+ // ========================================
	478	+
	479	+public function mxchat_handle_content_submission() {
	480	+ // Check if the form was submitted and the user has permission.
	481	+ if (!isset($_POST['submit_content']) || !current_user_can('manage_options')) {
	482	+ return;
	483	+ }
	484	+
	485	+ // Verify the nonce.
	486	+ $nonce = isset($_POST['mxchat_submit_content_nonce']) ? sanitize_text_field(wp_unslash($_POST['mxchat_submit_content_nonce'])) : '';
	487	+ if (!wp_verify_nonce($nonce, 'mxchat_submit_content_action')) {
	488	+ wp_die(esc_html__('Nonce verification failed.', 'mxchat'));
	489	+ }
	490	+
	491	+ // Sanitize the inputs.
	492	+ $article_content = sanitize_textarea_field($_POST['article_content']);
	493	+ $article_url = isset($_POST['article_url']) ? esc_url_raw($_POST['article_url']) : '';
	494	+
	495	+ // UPDATED: Get bot_id from form submission
	496	+ $bot_id = isset($_POST['bot_id']) ? sanitize_key($_POST['bot_id']) : 'default';
	497	+
	498	+ // UPDATED: Get bot-specific options and API key
	499	+ $bot_options = $this->get_bot_options($bot_id);
	500	+ $options = !empty($bot_options) ? $bot_options : get_option('mxchat_options');
	501	+ $selected_model = $options['embedding_model'] ?? 'text-embedding-ada-002';
	502	+
	503	+ if (strpos($selected_model, 'voyage') === 0) {
	504	+ $api_key = $options['voyage_api_key'] ?? '';
	505	+ } elseif (strpos($selected_model, 'gemini-embedding') === 0) {
	506	+ $api_key = $options['gemini_api_key'] ?? '';
	507	+ } else {
	508	+ $api_key = $options['api_key'] ?? '';
	509	+ }
	510	+
	511	+ if (empty($api_key)) {
	512	+ set_transient('mxchat_admin_notice_error',
	513	+ esc_html__('API key is not configured. Please add your API key in the settings before submitting content.', 'mxchat'),
	514	+ 30
	515	+ );
	516	+ wp_safe_redirect(esc_url(admin_url('admin.php?page=mxchat-prompts')));
	517	+ exit;
	518	+ }
	519	+
	520	+ // UPDATED: Use centralized utility function with bot_id
	521	+ $result = MxChat_Utils::submit_content_to_db($article_content, $article_url, $api_key, null, $bot_id);
	522	+
	523	+ if (is_wp_error($result)) {
	524	+ set_transient('mxchat_admin_notice_error',
	525	+ esc_html__('Error storing content: ', 'mxchat') . $result->get_error_message(),
	526	+ 30
	527	+ );
	528	+ } else {
	529	+ set_transient('mxchat_admin_notice_success',
	530	+ esc_html__('Content successfully submitted!', 'mxchat'),
	531	+ 30
	532	+ );
	533	+ }
	534	+
	535	+ wp_safe_redirect(esc_url(admin_url('admin.php?page=mxchat-prompts')));
	536	+ exit;
	537	+}
	538	+
	539	+public function mxchat_is_pdf_url($url, $response) {
	540	+ $content_type = wp_remote_retrieve_header($response, 'content-type');
	541	+ $file_extension = strtolower(pathinfo($url, PATHINFO_EXTENSION));
	542	+
	543	+ return strpos($content_type, 'pdf') !== false || $file_extension === 'pdf';
	544	+}
	545	+
	546	+
	547	+public function mxchat_handle_pdf_for_knowledge_base($pdf_url, $response, $bot_id = 'default') {
	548	+ if (!current_user_can('manage_options')) {
	549	+ //error_log('[PDF DEBUG] Unauthorized PDF processing attempt');
	550	+ return false;
	551	+ }
	552	+
	553	+ //error_log('[PDF DEBUG] Starting PDF processing for bot: ' . $bot_id);
	554	+ //error_log('[PDF DEBUG] PDF URL: ' . $pdf_url);
	555	+
	556	+ $pdf_url = esc_url_raw($pdf_url);
	557	+ $upload_dir = wp_upload_dir();
	558	+
	559	+ if (isset($upload_dir['error']) && $upload_dir['error'] !== false) {
	560	+ //error_log('[PDF DEBUG] Upload directory error: ' . $upload_dir['error']);
	561	+ return false;
	562	+ }
	563	+
	564	+ $pdf_filename = sanitize_file_name('mxchat_kb_' . time() . '.pdf');
	565	+ $pdf_path = trailingslashit($upload_dir['path']) . $pdf_filename;
	566	+
	567	+ $response_body = wp_remote_retrieve_body($response);
	568	+ if (empty($response_body)) {
	569	+ //error_log('[PDF DEBUG] Empty PDF response body');
	570	+ return false;
	571	+ }
	572	+
	573	+ if (!wp_mkdir_p(dirname($pdf_path))) {
	574	+ //error_log('[PDF DEBUG] Failed to create directory for PDF: ' . $pdf_path);
	575	+ return false;
	576	+ }
	577	+
	578	+ try {
	579	+ file_put_contents($pdf_path, $response_body);
	580	+
	581	+ if (!file_exists($pdf_path)) {
	582	+ throw new Exception(__('Failed to save PDF file', 'mxchat'));
	583	+ }
	584	+
	585	+ $total_pages = $this->mxchat_validate_and_count_pdf_pages($pdf_path);
	586	+
	587	+ if ($total_pages === false || $total_pages < 1) {
	588	+ throw new Exception(__('Invalid PDF: Unable to parse or no pages found', 'mxchat'));
	589	+ }
	590	+
	591	+ //error_log('[PDF DEBUG] PDF validated successfully with ' . $total_pages . ' pages');
	592	+
	593	+ // UPDATED: Pass bot_id to PDF processing cron
	594	+ wp_schedule_single_event(time(), 'mxchat_process_pdf_pages', array(
	595	+ $pdf_path, // Position 0
	596	+ $pdf_url, // Position 1
	597	+ $total_pages, // Position 2
	598	+ absint(15), // Position 3 (batch_size)
	599	+ absint(10), // Position 4 (batch_pause)
	600	+ $bot_id // Position 5 (bot_id)
	601	+ ));
	602	+
	603	+ //error_log('[PDF DEBUG] Cron job scheduled with bot_id: ' . $bot_id);
	604	+
	605	+ // UPDATED: Store bot_id in status data
	606	+ $status_data = array(
	607	+ 'total_pages' => $total_pages,
	608	+ 'processed_pages' => 0,
	609	+ 'status' => 'processing',
	610	+ 'last_update' => time(),
	611	+ 'bot_id' => $bot_id // CRITICAL: Store the bot_id
	612	+ );
	613	+
	614	+ $status_key = sanitize_key('mxchat_pdf_status_' . md5($pdf_url));
	615	+ set_transient($status_key, $status_data, DAY_IN_SECONDS);
	616	+
	617	+ //error_log('[PDF DEBUG] Status stored with bot_id: ' . $bot_id . ' using key: ' . $status_key);
	618	+
	619	+ return __('scheduled', 'mxchat');
	620	+
	621	+ } catch (Exception $e) {
	622	+ //error_log('[PDF DEBUG] Error preparing PDF for processing: ' . $e->getMessage());
	623	+ if (file_exists($pdf_path)) {
	624	+ wp_delete_file($pdf_path);
	625	+ }
	626	+ return false;
	627	+ }
	628	+}
	629	+
	630	+/**
	631	+ * NEW: Validate PDF and count pages with multiple parser attempts
	632	+ */
	633	+private function mxchat_validate_and_count_pdf_pages($pdf_path) {
	634	+ // Method 1: Try with Smalot PDF Parser (your current method)
	635	+ try {
	636	+ $parser = new \Smalot\PdfParser\Parser();
	637	+ $pdf = $parser->parseFile($pdf_path);
	638	+ $pages = $pdf->getPages();
	639	+ $page_count = count($pages);
	640	+
	641	+ if ($page_count > 0) {
	642	+ //error_log('PDF parsed successfully with Smalot parser: ' . $page_count . ' pages');
	643	+ return $page_count;
	644	+ }
	645	+ } catch (Exception $e) {
	646	+ //error_log('Smalot PDF parser failed: ' . $e->getMessage());
	647	+ }
	648	+
	649	+ // Method 2: Try with pdfinfo command (if available)
	650	+ if (function_exists('shell_exec') && !$this->mxchat_is_shell_disabled()) {
	651	+ try {
	652	+ $command = 'pdfinfo ' . escapeshellarg($pdf_path) . ' 2>&1';
	653	+ $output = shell_exec($command);
	654	+
	655	+ if ($output && preg_match('/Pages:\s*(\d+)/', $output, $matches)) {
	656	+ $page_count = intval($matches[1]);
	657	+ if ($page_count > 0) {
	658	+ //error_log('PDF parsed successfully with pdfinfo: ' . $page_count . ' pages');
	659	+ return $page_count;
	660	+ }
	661	+ }
	662	+ } catch (Exception $e) {
	663	+ //error_log('pdfinfo command failed: ' . $e->getMessage());
	664	+ }
	665	+ }
	666	+
	667	+ // Method 3: Try to repair PDF and parse again
	668	+ try {
	669	+ $repaired_path = $this->mxchat_attempt_pdf_repair($pdf_path);
	670	+ if ($repaired_path && $repaired_path !== $pdf_path) {
	671	+ $parser = new \Smalot\PdfParser\Parser();
	672	+ $pdf = $parser->parseFile($repaired_path);
	673	+ $pages = $pdf->getPages();
	674	+ $page_count = count($pages);
	675	+
	676	+ if ($page_count > 0) {
	677	+ // Replace original with repaired version
	678	+ copy($repaired_path, $pdf_path);
	679	+ unlink($repaired_path);
	680	+ //error_log('PDF repaired and parsed successfully: ' . $page_count . ' pages');
	681	+ return $page_count;
	682	+ }
	683	+
	684	+ // Clean up repaired file if it didn't work
	685	+ unlink($repaired_path);
	686	+ }
	687	+ } catch (Exception $e) {
	688	+ //error_log('PDF repair attempt failed: ' . $e->getMessage());
	689	+ }
	690	+
	691	+ // Method 4: Manual PDF structure analysis (basic page count)
	692	+ try {
	693	+ $page_count = $this->mxchat_manual_pdf_page_count($pdf_path);
	694	+ if ($page_count > 0) {
	695	+ //error_log('PDF page count determined manually: ' . $page_count . ' pages');
	696	+ return $page_count;
	697	+ }
	698	+ } catch (Exception $e) {
	699	+ //error_log('Manual PDF analysis failed: ' . $e->getMessage());
	700	+ }
	701	+
	702	+ //error_log('All PDF parsing methods failed for: ' . $pdf_path);
	703	+ return false;
	704	+}
	705	+
	706	+/**
	707	+ * NEW: Check if shell_exec is disabled
	708	+ */
	709	+private function mxchat_is_shell_disabled() {
	710	+ $disabled = explode(',', ini_get('disable_functions'));
	711	+ return in_array('shell_exec', $disabled);
	712	+}
	713	+
	714	+/**
	715	+ * NEW: Attempt to repair PDF using basic methods
	716	+ */
	717	+private function mxchat_attempt_pdf_repair($pdf_path) {
	718	+ try {
	719	+ $content = file_get_contents($pdf_path);
	720	+ if (!$content) {
	721	+ return false;
	722	+ }
	723	+
	724	+ // Check if PDF starts with proper header
	725	+ if (substr($content, 0, 4) !== '%PDF') {
	726	+ // Try to find PDF header in the content
	727	+ $header_pos = strpos($content, '%PDF');
	728	+ if ($header_pos !== false && $header_pos < 1024) {
	729	+ // Remove junk before PDF header
	730	+ $content = substr($content, $header_pos);
	731	+ $repaired_path = $pdf_path . '.repaired';
	732	+ file_put_contents($repaired_path, $content);
	733	+ return $repaired_path;
	734	+ }
	735	+ }
	736	+
	737	+ // Check for EOF marker
	738	+ $content = rtrim($content);
	739	+ if (!preg_match('/%%EOF\s*$/', $content)) {
	740	+ // Add EOF marker if missing
	741	+ $content .= "\n%%EOF";
	742	+ $repaired_path = $pdf_path . '.repaired';
	743	+ file_put_contents($repaired_path, $content);
	744	+ return $repaired_path;
	745	+ }
	746	+
	747	+ } catch (Exception $e) {
	748	+ //error_log('PDF repair error: ' . $e->getMessage());
	749	+ }
	750	+
	751	+ return false;
	752	+}
	753	+
	754	+/**
	755	+ * NEW: Manual PDF page counting by analyzing PDF structure
	756	+ */
	757	+private function mxchat_manual_pdf_page_count($pdf_path) {
	758	+ try {
	759	+ $content = file_get_contents($pdf_path);
	760	+ if (!$content) {
	761	+ return 0;
	762	+ }
	763	+
	764	+ // Method 1: Count /Type /Page objects
	765	+ $page_count = preg_match_all('/\/Type\s*\/Page[^s]/', $content);
	766	+ if ($page_count > 0) {
	767	+ return $page_count;
	768	+ }
	769	+
	770	+ // Method 2: Look for /Count in pages object
	771	+ if (preg_match('/\/Type\s*\/Pages.*?\/Count\s+(\d+)/', $content, $matches)) {
	772	+ return intval($matches[1]);
	773	+ }
	774	+
	775	+ // Method 3: Count page references
	776	+ $page_count = preg_match_all('/\d+\s+0\s+obj\s*<<[^>]*\/Type\s*\/Page/', $content);
	777	+ if ($page_count > 0) {
	778	+ return $page_count;
	779	+ }
	780	+
	781	+ } catch (Exception $e) {
	782	+ //error_log('Manual PDF analysis error: ' . $e->getMessage());
	783	+ }
	784	+
	785	+ return 0;
	786	+}
	787	+
	788	+public function mxchat_process_pdf_pages_cron($pdf_path, $pdf_url, $total_pages, $batch_size, $batch_pause, $bot_id = 'default') {
	789	+ // ADD THIS DEBUG SECTION AT THE VERY BEGINNING
	790	+ //error_log('[PDF CRON DEBUG] ===== PDF Cron Job Started =====');
	791	+ //error_log('[PDF CRON DEBUG] Initial bot_id parameter: ' . $bot_id);
	792	+ //error_log('[PDF CRON DEBUG] Received parameters:');
	793	+ //error_log('[PDF CRON DEBUG] - pdf_path: ' . $pdf_path);
	794	+ //error_log('[PDF CRON DEBUG] - pdf_url: ' . $pdf_url);
	795	+ //error_log('[PDF CRON DEBUG] - total_pages: ' . $total_pages);
	796	+ //error_log('[PDF CRON DEBUG] - batch_size: ' . $batch_size);
	797	+ //error_log('[PDF CRON DEBUG] - batch_pause: ' . $batch_pause);
	798	+ //error_log('[PDF CRON DEBUG] - Total args received: ' . func_num_args());
	799	+ //error_log('[PDF CRON DEBUG] - All args: ' . print_r(func_get_args(), true));
	800	+
	801	+ // FIXED: Get the correct bot_id from stored status instead of relying on cron parameters
	802	+ $status_key = sanitize_key('mxchat_pdf_status_' . md5($pdf_url));
	803	+ $status = get_transient($status_key);
	804	+
	805	+ if ($status && isset($status['bot_id'])) {
	806	+ $bot_id = $status['bot_id'];
	807	+ //error_log('[PDF CRON DEBUG] Using bot_id from status: ' . $bot_id);
	808	+ } else {
	809	+ //error_log('[PDF CRON DEBUG] No bot_id in status, using default: ' . $bot_id);
	810	+ }
	811	+
	812	+ // Validate inputs
	813	+ $pdf_path = sanitize_text_field($pdf_path);
	814	+ $pdf_url = esc_url_raw($pdf_url);
	815	+ $total_pages = absint($total_pages);
	816	+ $batch_size = absint($batch_size);
	817	+ $batch_pause = absint($batch_pause);
	818	+ $bot_id = sanitize_key($bot_id);
	819	+
	820	+ try {
	821	+ if (!file_exists($pdf_path)) {
	822	+ throw new Exception(sprintf('PDF file not found at path: %s', esc_html($pdf_path)));
	823	+ }
	824	+
	825	+ // Try to parse PDF with error recovery
	826	+ $pdf = null;
	827	+ $pages = null;
	828	+
	829	+ try {
	830	+ $parser = new \Smalot\PdfParser\Parser();
	831	+ $pdf = $parser->parseFile($pdf_path);
	832	+ $pages = $pdf->getPages();
	833	+ } catch (Exception $e) {
	834	+ //error_log('Primary PDF parsing failed, attempting recovery: ' . $e->getMessage());
	835	+
	836	+ $repaired_path = $this->mxchat_attempt_pdf_repair($pdf_path);
	837	+ if ($repaired_path) {
	838	+ try {
	839	+ $parser = new \Smalot\PdfParser\Parser();
	840	+ $pdf = $parser->parseFile($repaired_path);
	841	+ $pages = $pdf->getPages();
	842	+
	843	+ copy($repaired_path, $pdf_path);
	844	+ unlink($repaired_path);
	845	+ //error_log('PDF successfully repaired and parsed');
	846	+ } catch (Exception $e2) {
	847	+ if (file_exists($repaired_path)) {
	848	+ unlink($repaired_path);
	849	+ }
	850	+ throw new Exception('PDF parsing failed even after repair attempt: ' . $e2->getMessage());
	851	+ }
	852	+ } else {
	853	+ throw new Exception('PDF parsing failed and repair was unsuccessful: ' . $e->getMessage());
	854	+ }
	855	+ }
	856	+
	857	+ if (!$pages || count($pages) === 0) {
	858	+ throw new Exception('No pages found in PDF after parsing');
	859	+ }
	860	+
	861	+ // Get current progress (already fetched above for bot_id)
	862	+ if (!$status || !is_array($status)) {
	863	+ throw new Exception('Invalid status data retrieved from transient');
	864	+ }
	865	+
	866	+ // Initialize failed pages list if it doesn't exist
	867	+ if (!isset($status['failed_pages_list']) || !is_array($status['failed_pages_list'])) {
	868	+ $status['failed_pages_list'] = [];
	869	+ }
	870	+
	871	+ $start_page = absint($status['processed_pages']);
	872	+ $end_page = min($start_page + $batch_size, $total_pages);
	873	+
	874	+ // UPDATED: Get bot-specific options using the correct bot_id
	875	+ $bot_options = $this->get_bot_options($bot_id);
	876	+ $options = !empty($bot_options) ? $bot_options : get_option('mxchat_options');
	877	+
	878	+ //error_log('[PDF CRON DEBUG] Using bot options for bot: ' . $bot_id);
	879	+
	880	+ if (empty($options['api_key'])) {
	881	+ throw new Exception('API key is missing or invalid for bot: ' . $bot_id);
	882	+ }
	883	+
	884	+ $successful_pages = 0;
	885	+ $failed_pages = 0;
	886	+
	887	+ for ($i = $start_page; $i < $end_page; $i++) {
	888	+ $page_number = $i + 1;
	889	+ $max_retries = 3;
	890	+ $retry_count = 0;
	891	+ $page_processed = false;
	892	+ $last_error = '';
	893	+
	894	+ while (!$page_processed && $retry_count < $max_retries) {
	895	+ try {
	896	+ $text = $pages[$i]->getText();
	897	+
	898	+ if (empty($text)) {
	899	+ throw new Exception("Empty text on page {$page_number}");
	900	+ }
	901	+
	902	+ $sanitized_content = $this->mxchat_sanitize_content_for_api($text);
	903	+
	904	+ if (empty($sanitized_content)) {
	905	+ throw new Exception("No valid content after sanitization on page {$page_number}");
	906	+ }
	907	+
	908	+ // UPDATED: Use bot-specific embedding generation with correct bot_id
	909	+ $embedding_vector = $this->mxchat_generate_embedding($sanitized_content, $bot_id);
	910	+
	911	+ if (is_string($embedding_vector)) {
	912	+ throw new Exception("Embedding generation failed: " . $embedding_vector);
	913	+ }
	914	+
	915	+ if (!is_array($embedding_vector)) {
	916	+ throw new Exception("Embedding generation returned unexpected result type: " . gettype($embedding_vector));
	917	+ }
	918	+
	919	+ $metadata = array(
	920	+ 'document_type' => 'pdf',
	921	+ 'total_pages' => $total_pages,
	922	+ 'current_page' => $page_number,
	923	+ 'prev_page' => $i > 0 ? $i : null,
	924	+ 'next_page' => $i < ($total_pages - 1) ? $i + 2 : null,
	925	+ 'source_url' => $pdf_url
	926	+ );
	927	+
	928	+ $content_with_metadata = wp_json_encode($metadata) . "\n---\n" . $sanitized_content;
	929	+ $page_url = esc_url($pdf_url . "#page=" . $page_number);
	930	+
	931	+ // UPDATED: Pass correct bot_id to database submission
	932	+ $db_result = MxChat_Utils::submit_content_to_db($content_with_metadata, $page_url, $options['api_key'], null, $bot_id);
	933	+
	934	+ if (is_wp_error($db_result)) {
	935	+ throw new Exception("Database submission failed: " . $db_result->get_error_message());
	936	+ }
	937	+
	938	+ // Success!
	939	+ $page_processed = true;
	940	+ $successful_pages++;
	941	+ //error_log('[PDF CRON DEBUG] Successfully processed page ' . $page_number . ' for bot: ' . $bot_id);
	942	+
	943	+ } catch (Exception $e) {
	944	+ $retry_count++;
	945	+ $last_error = $e->getMessage();
	946	+
	947	+ //error_log("PDF page {$page_number} failed (attempt {$retry_count}/{$max_retries}): " . $last_error);
	948	+
	949	+ if ($retry_count < $max_retries) {
	950	+ sleep(pow(2, $retry_count - 1));
	951	+ }
	952	+ }
	953	+ }
	954	+
	955	+ // If page still not processed after all retries, mark as failed
	956	+ if (!$page_processed) {
	957	+ $failed_pages++;
	958	+ $status['failed_pages_list'][] = [
	959	+ 'page' => $page_number,
	960	+ 'error' => $last_error,
	961	+ 'time' => time(),
	962	+ 'retries' => $max_retries
	963	+ ];
	964	+
	965	+ if (count($status['failed_pages_list']) > 50) {
	966	+ $status['failed_pages_list'] = array_slice($status['failed_pages_list'], -50);
	967	+ }
	968	+
	969	+ //error_log("PDF page {$page_number} permanently failed after {$max_retries} attempts: " . $last_error);
	970	+ }
	971	+
	972	+ // Update progress
	973	+ $status['processed_pages'] = absint($page_number);
	974	+ $status['last_update'] = time();
	975	+ $status['failed_pages'] = absint($status['failed_pages'] ?? 0) + ($page_processed ? 0 : 1);
	976	+
	977	+ set_transient($status_key, $status, DAY_IN_SECONDS);
	978	+ }
	979	+
	980	+ // Schedule next batch if needed
	981	+ if ($end_page < $total_pages) {
	982	+ // Use the same indexed array format (though bot_id still won't pass correctly)
	983	+ wp_schedule_single_event(time() + $batch_pause, 'mxchat_process_pdf_pages', array(
	984	+ $pdf_path,
	985	+ $pdf_url,
	986	+ $total_pages,
	987	+ $batch_size,
	988	+ $batch_pause,
	989	+ $bot_id // This still won't work, but we're now getting bot_id from status
	990	+ ));
	991	+ } else {
	992	+ // Processing complete
	993	+ $status['status'] = 'complete';
	994	+ $status['processed_pages'] = $total_pages;
	995	+
	996	+ $status['completion_summary'] = [
	997	+ 'total_pages' => $total_pages,
	998	+ 'successful_pages' => $total_pages - absint($status['failed_pages'] ?? 0),
	999	+ 'failed_pages' => absint($status['failed_pages'] ?? 0),
	1000	+ 'completion_time' => current_time('mysql')
	1001	+ ];
	1002	+
	1003	+ set_transient($status_key, $status, DAY_IN_SECONDS);
	1004	+
	1005	+ if (file_exists($pdf_path)) {
	1006	+ wp_delete_file($pdf_path);
	1007	+ }
	1008	+
	1009	+ //error_log('[PDF CRON DEBUG] PDF processing completed for bot: ' . $bot_id);
	1010	+ }
	1011	+
	1012	+ } catch (\Exception $e) {
	1013	+ //error_log(sprintf('[MXCHAT-PDF] Error processing PDF for bot %s: %s', $bot_id, $e->getMessage()));
	1014	+
	1015	+ $status_key = sanitize_key('mxchat_pdf_status_' . md5($pdf_url));
	1016	+ $status = get_transient($status_key);
	1017	+
	1018	+ if (!$status || !is_array($status)) {
	1019	+ $status = array(
	1020	+ 'total_pages' => $total_pages,
	1021	+ 'processed_pages' => 0,
	1022	+ 'status' => 'error',
	1023	+ 'error' => sanitize_text_field($e->getMessage()),
	1024	+ 'last_update' => time(),
	1025	+ 'bot_id' => $bot_id
	1026	+ );
	1027	+ } else {
	1028	+ $status['status'] = 'error';
	1029	+ $status['error'] = sanitize_text_field($e->getMessage());
	1030	+ $status['last_update'] = time();
	1031	+ }
	1032	+
	1033	+ set_transient($status_key, $status, DAY_IN_SECONDS);
	1034	+
	1035	+ if (file_exists($pdf_path)) {
	1036	+ wp_delete_file($pdf_path);
	1037	+ }
	1038	+ }
	1039	+}
	1040	+
	1041	+public function mxchat_save_inline_prompt() {
	1042	+ // DEBUG: Log what we're receiving
	1043	+ //error_log('=== MXCHAT DEBUG ===');
	1044	+ //error_log('POST data: ' . print_r($_POST, true));
	1045	+ //error_log('Nonce from POST: ' . ($_POST['_ajax_nonce'] ?? 'NOT FOUND'));
	1046	+
	1047	+ // Check for nonce security
	1048	+ check_ajax_referer('mxchat_save_inline_nonce', '_ajax_nonce');
	1049	+
	1050	+ // If we get here, nonce passed
	1051	+ //error_log('Nonce verification PASSED');
	1052	+
	1053	+ // Verify permissions
	1054	+ if (!current_user_can('manage_options')) {
	1055	+ wp_send_json_error(esc_html__('Permission denied.', 'mxchat'));
	1056	+ return;
	1057	+ }
	1058	+
	1059	+ global $wpdb;
	1060	+ $table_name = $wpdb->prefix . 'mxchat_system_prompt_content';
	1061	+
	1062	+ // Validate and sanitize input data - FIXED LINE BELOW
	1063	+ $prompt_id = isset($_POST['id']) ? absint($_POST['id']) : 0;
	1064	+ $article_content = isset($_POST['article_content']) ? sanitize_textarea_field(wp_unslash($_POST['article_content'])) : '';
	1065	+ $article_url = isset($_POST['article_url']) ? esc_url_raw($_POST['article_url']) : '';
	1066	+
	1067	+ if ($prompt_id > 0 && !empty($article_content)) {
	1068	+ // Re-generate the embedding vector for the updated content
	1069	+ $embedding_vector = $this->mxchat_generate_embedding($article_content);
	1070	+ if (is_array($embedding_vector)) {
	1071	+ // Serialize the embedding vector before storing it
	1072	+ $embedding_vector_serialized = serialize($embedding_vector);
	1073	+ // Update the prompt in the database
	1074	+ $updated = $wpdb->update(
	1075	+ $table_name,
	1076	+ array(
	1077	+ 'article_content' => $article_content,
	1078	+ 'embedding_vector' => $embedding_vector_serialized,
	1079	+ 'source_url' => $article_url,
	1080	+ ),
	1081	+ array('id' => $prompt_id),
	1082	+ array('%s', '%s', '%s'),
	1083	+ array('%d')
	1084	+ );
	1085	+ if ($updated !== false) {
	1086	+ wp_send_json_success();
	1087	+ } else {
	1088	+ wp_send_json_error(esc_html__('Database update failed.', 'mxchat'));
	1089	+ }
	1090	+ } else {
	1091	+ wp_send_json_error(esc_html__('Embedding generation failed.', 'mxchat'));
	1092	+ }
	1093	+ } else {
	1094	+ wp_send_json_error(esc_html__('Invalid data.', 'mxchat'));
	1095	+ }
	1096	+}
	1097	+
	1098	+
	1099	+public function mxchat_get_pdf_processing_status($pdf_url) {
	1100	+ $pdf_url = esc_url_raw($pdf_url);
	1101	+ $status = get_transient(sanitize_key('mxchat_pdf_status_' . md5($pdf_url)));
	1102	+
	1103	+ if (!$status || !is_array($status)) {
	1104	+ return false;
	1105	+ }
	1106	+
	1107	+ // Check for stalled processing (no updates for 5 minutes)
	1108	+ if ($status['status'] === 'processing' && (time() - absint($status['last_update'])) > 300) {
	1109	+ $status['status'] = 'error';
	1110	+ $status['error'] = __('PDF processing appears to be stalled. No updates for over 5 minutes.', 'mxchat');
	1111	+
	1112	+ // Save the updated status
	1113	+ set_transient(
	1114	+ sanitize_key('mxchat_pdf_status_' . md5($pdf_url)),
	1115	+ array_map('sanitize_text_field', $status),
	1116	+ DAY_IN_SECONDS
	1117	+ );
	1118	+ }
	1119	+
	1120	+ $result = array(
	1121	+ 'total_pages' => absint($status['total_pages']),
	1122	+ 'processed_pages' => absint($status['processed_pages']),
	1123	+ 'failed_pages' => absint($status['failed_pages'] ?? 0),
	1124	+ 'percentage' => ($status['total_pages'] > 0)
	1125	+ ? round((absint($status['processed_pages']) / absint($status['total_pages'])) * 100)
	1126	+ : 0,
	1127	+ 'status' => sanitize_text_field($status['status']),
	1128	+ 'last_update' => human_time_diff(absint($status['last_update']), time()) . ' ' . esc_html__('ago', 'mxchat'),
	1129	+ 'failed_pages_list' => isset($status['failed_pages_list']) ? $status['failed_pages_list'] : array(),
	1130	+ 'completion_summary' => isset($status['completion_summary']) ? $status['completion_summary'] : null
	1131	+ );
	1132	+
	1133	+ // Add error message if present
	1134	+ if (isset($status['error']) && !empty($status['error'])) {
	1135	+ $result['error'] = sanitize_text_field($status['error']);
	1136	+ }
	1137	+
	1138	+ return $result;
	1139	+}
	1140	+
	1141	+
	1142	+public function mxchat_handle_sitemap_submission() {
	1143	+ // START DEBUG
	1144	+ //error_log('[SITEMAP DEBUG] ===== Starting URL submission process =====');
	1145	+ //error_log('[SITEMAP DEBUG] POST data: ' . print_r($_POST, true));
	1146	+
	1147	+ // Get bot_id from form submission EARLY for debugging
	1148	+ $bot_id = isset($_POST['bot_id']) ? sanitize_key($_POST['bot_id']) : 'default';
	1149	+ //error_log('[SITEMAP DEBUG] Extracted bot_id: ' . $bot_id);
	1150	+ //error_log('[SITEMAP DEBUG] Class exists MxChat_Multi_Bot_Manager: ' . (class_exists('MxChat_Multi_Bot_Manager') ? 'YES' : 'NO'));
	1151	+ // END DEBUG
	1152	+
	1153	+ // Check if the form was submitted and verify permissions
	1154	+ if (!isset($_POST['submit_sitemap']) || !current_user_can('manage_options')) {
	1155	+ //error_log('[SITEMAP DEBUG] Error: Unauthorized access or form not submitted properly');
	1156	+ wp_die(esc_html__('Unauthorized access', 'mxchat'));
	1157	+ }
	1158	+
	1159	+ // Verify nonce
	1160	+ //error_log('[SITEMAP DEBUG] Verifying nonce');
	1161	+ check_admin_referer('mxchat_submit_sitemap_action', 'mxchat_submit_sitemap_nonce');
	1162	+
	1163	+ // Validate URL
	1164	+ if (!isset($_POST['sitemap_url']) || empty($_POST['sitemap_url'])) {
	1165	+ //error_log('[SITEMAP DEBUG] Error: Empty or missing URL');
	1166	+ set_transient('mxchat_admin_notice_error',
	1167	+ esc_html__('Please provide a valid URL.', 'mxchat'),
	1168	+ 30
	1169	+ );
	1170	+ wp_safe_redirect(esc_url(admin_url('admin.php?page=mxchat-prompts')));
	1171	+ exit;
	1172	+ }
	1173	+
	1174	+ $submitted_url = esc_url_raw($_POST['sitemap_url']);
	1175	+
	1176	+ // Continue processing with already extracted bot_id
	1177	+ //error_log('[SITEMAP DEBUG] Processing URL: ' . $submitted_url . ' for bot: ' . $bot_id);
	1178	+
	1179	+ // UPDATED: Get bot-specific options and validate API key
	1180	+ $bot_options = $this->get_bot_options($bot_id);
	1181	+ //error_log('[SITEMAP DEBUG] Bot options retrieved: ' . print_r($bot_options, true));
	1182	+
	1183	+ $options = !empty($bot_options) ? $bot_options : get_option('mxchat_options');
	1184	+ $selected_model = $options['embedding_model'] ?? 'text-embedding-ada-002';
	1185	+
	1186	+ //error_log('[SITEMAP DEBUG] Selected embedding model: ' . $selected_model);
	1187	+
	1188	+ if (strpos($selected_model, 'voyage') === 0) {
	1189	+ $api_key = $options['voyage_api_key'] ?? '';
	1190	+ $provider_name = 'Voyage AI';
	1191	+ } elseif (strpos($selected_model, 'gemini-embedding') === 0) {
	1192	+ $api_key = $options['gemini_api_key'] ?? '';
	1193	+ $provider_name = 'Google Gemini';
	1194	+ } else {
	1195	+ $api_key = $options['api_key'] ?? '';
	1196	+ $provider_name = 'OpenAI';
	1197	+ }
	1198	+
	1199	+ //error_log('[SITEMAP DEBUG] Provider: ' . $provider_name . ', Has API key: ' . (!empty($api_key) ? 'YES' : 'NO'));
	1200	+
	1201	+ if (empty($api_key)) {
	1202	+ $error_message = sprintf(
	1203	+ esc_html__('%s API key is not configured. Please add your API key in the settings before submitting content.', 'mxchat'),
	1204	+ $provider_name
	1205	+ );
	1206	+ //error_log('[SITEMAP DEBUG] Error: ' . $error_message);
	1207	+ set_transient('mxchat_admin_notice_error', $error_message, 30);
	1208	+ wp_safe_redirect(esc_url(admin_url('admin.php?page=mxchat-prompts')));
	1209	+ exit;
	1210	+ }
	1211	+
	1212	+ //error_log('[SITEMAP DEBUG] Fetching URL content');
	1213	+ $response = wp_remote_get($submitted_url, array('timeout' => 30));
	1214	+
	1215	+ if (is_wp_error($response) || wp_remote_retrieve_response_code($response) !== 200) {
	1216	+ $error_message = is_wp_error($response) ? $response->get_error_message() : 'HTTP Status: ' . wp_remote_retrieve_response_code($response);
	1217	+ //error_log('[SITEMAP DEBUG] Error fetching URL: ' . $error_message);
	1218	+ set_transient('mxchat_admin_notice_error',
	1219	+ sprintf(
	1220	+ esc_html__('Failed to fetch the URL: %s', 'mxchat'),
	1221	+ esc_html($error_message)
	1222	+ ),
	1223	+ 30
	1224	+ );
	1225	+ wp_safe_redirect(esc_url(admin_url('admin.php?page=mxchat-prompts')));
	1226	+ exit;
	1227	+ }
	1228	+
	1229	+ $content_type = wp_remote_retrieve_header($response, 'content-type');
	1230	+ //error_log('[SITEMAP DEBUG] Content type: ' . $content_type);
	1231	+ $body_content = wp_remote_retrieve_body($response);
	1232	+
	1233	+ if (empty($body_content)) {
	1234	+ //error_log('[SITEMAP DEBUG] Error: Empty response body');
	1235	+ set_transient('mxchat_admin_notice_error',
	1236	+ esc_html__('Empty response received from URL.', 'mxchat'),
	1237	+ 30
	1238	+ );
	1239	+ wp_safe_redirect(esc_url(admin_url('admin.php?page=mxchat-prompts')));
	1240	+ exit;
	1241	+ }
	1242	+ //error_log('[SITEMAP DEBUG] Retrieved body content length: ' . strlen($body_content) . ' bytes');
	1243	+
	1244	+ // Handle PDF URL
	1245	+ if ($this->mxchat_is_pdf_url($submitted_url, $response)) {
	1246	+ //error_log('[SITEMAP DEBUG] Detected PDF URL, handling PDF for knowledge base');
	1247	+ //error_log('[SITEMAP DEBUG] About to call PDF handler with bot_id: ' . $bot_id);
	1248	+
	1249	+ // UPDATED: Pass bot_id to PDF handler
	1250	+ $result = $this->mxchat_handle_pdf_for_knowledge_base($submitted_url, $response, $bot_id);
	1251	+ //error_log('[SITEMAP DEBUG] PDF handling result: ' . $result);
	1252	+
	1253	+ if ($result === 'scheduled') {
	1254	+ set_transient(
	1255	+ 'mxchat_last_pdf_url',
	1256	+ sanitize_text_field($submitted_url),
	1257	+ DAY_IN_SECONDS
	1258	+ );
	1259	+ // UPDATED: Store bot_id for PDF processing
	1260	+ set_transient(
	1261	+ 'mxchat_last_pdf_bot_id',
	1262	+ $bot_id,
	1263	+ DAY_IN_SECONDS
	1264	+ );
	1265	+ //error_log('[SITEMAP DEBUG] PDF processing scheduled successfully for bot: ' . $bot_id);
	1266	+ set_transient('mxchat_admin_notice_info',
	1267	+ esc_html__('PDF processing has started in the background. You can check the progress in the Knowledge Base section.', 'mxchat'),
	1268	+ 30
	1269	+ );
	1270	+ } else {
	1271	+ //error_log('[SITEMAP DEBUG] PDF processing failed: ' . $result);
	1272	+ set_transient('mxchat_admin_notice_error',
	1273	+ esc_html__('Failed to start PDF processing: ', 'mxchat') . esc_html($result),
	1274	+ 30
	1275	+ );
	1276	+ }
	1277	+
	1278	+ wp_safe_redirect(esc_url(admin_url('admin.php?page=mxchat-prompts')));
	1279	+ exit;
	1280	+ }
	1281	+
	1282	+ // Handle Sitemap XML
	1283	+ if (strpos($content_type, 'xml') !== false || strpos($body_content, '<urlset') !== false) {
	1284	+ //error_log('[SITEMAP DEBUG] Detected XML content, processing as sitemap');
	1285	+ libxml_use_internal_errors(true);
	1286	+ $xml = simplexml_load_string($body_content);
	1287	+ $xml_errors = libxml_get_errors();
	1288	+ libxml_clear_errors();
	1289	+
	1290	+ if ($xml === false || !empty($xml_errors)) {
	1291	+ //error_log('[SITEMAP DEBUG] Error: Invalid XML format');
	1292	+ if (!empty($xml_errors)) {
	1293	+ foreach ($xml_errors as $error) {
	1294	+ //error_log('[SITEMAP DEBUG] XML Error: ' . $error->message);
	1295	+ }
	1296	+ }
	1297	+
	1298	+ set_transient('mxchat_admin_notice_error',
	1299	+ esc_html__('Invalid sitemap XML. Please provide a valid sitemap.', 'mxchat'),
	1300	+ 30
	1301	+ );
	1302	+ wp_safe_redirect(esc_url(admin_url('admin.php?page=mxchat-prompts')));
	1303	+ exit;
	1304	+ }
	1305	+
	1306	+ //error_log('[SITEMAP DEBUG] Valid XML found, handling sitemap for knowledge base');
	1307	+ // UPDATED: Pass bot_id to sitemap handler
	1308	+ $result = $this->mxchat_handle_sitemap_for_knowledge_base($xml, $submitted_url, $bot_id);
	1309	+ //error_log('[SITEMAP DEBUG] Sitemap handling result: ' . $result);
	1310	+
	1311	+ if ($result === 'scheduled') {
	1312	+ set_transient(
	1313	+ 'mxchat_last_sitemap_url',
	1314	+ sanitize_text_field($submitted_url),
	1315	+ DAY_IN_SECONDS
	1316	+ );
	1317	+ // UPDATED: Store bot_id for sitemap processing
	1318	+ set_transient(
	1319	+ 'mxchat_last_sitemap_bot_id',
	1320	+ $bot_id,
	1321	+ DAY_IN_SECONDS
	1322	+ );
	1323	+ set_transient('mxchat_admin_notice_info',
	1324	+ esc_html__('Sitemap processing has started in the background. You can check the progress in the Knowledge Base section.', 'mxchat'),
	1325	+ 30
	1326	+ );
	1327	+ } else {
	1328	+ set_transient('mxchat_admin_notice_error',
	1329	+ esc_html__('Failed to start sitemap processing. Please check the status below for details.', 'mxchat'),
	1330	+ 30
	1331	+ );
	1332	+ }
	1333	+
	1334	+ wp_safe_redirect(esc_url(admin_url('admin.php?page=mxchat-prompts')));
	1335	+ exit;
	1336	+ }
	1337	+
	1338	+ // Handle Regular URL
	1339	+ //error_log('[SITEMAP DEBUG] Processing as regular webpage');
	1340	+ $page_content = $this->mxchat_extract_main_content($body_content);
	1341	+ //error_log('[SITEMAP DEBUG] Extracted content length: ' . strlen($page_content) . ' bytes');
	1342	+
	1343	+ $sanitized_content = $this->mxchat_sanitize_content_for_api($page_content);
	1344	+ //error_log('[SITEMAP DEBUG] Sanitized content length: ' . strlen($sanitized_content) . ' bytes');
	1345	+
	1346	+ if (empty($sanitized_content)) {
	1347	+ //error_log('[SITEMAP DEBUG] Error: No valid content after sanitization');
	1348	+
	1349	+ set_transient('mxchat_admin_notice_error',
	1350	+ esc_html__('No valid content found on the provided URL.', 'mxchat'),
	1351	+ 30
	1352	+ );
	1353	+
	1354	+ set_transient('mxchat_single_url_status', [
	1355	+ 'url' => $submitted_url,
	1356	+ 'timestamp' => current_time('mysql'),
	1357	+ 'status' => 'failed',
	1358	+ 'error' => esc_html__('No valid content found on the provided URL.', 'mxchat')
	1359	+ ], DAY_IN_SECONDS);
	1360	+
	1361	+ wp_safe_redirect(esc_url(admin_url('admin.php?page=mxchat-prompts')));
	1362	+ exit;
	1363	+ }
	1364	+
	1365	+ //error_log('[SITEMAP DEBUG] Generating embedding for content');
	1366	+ $embedding_vector = $this->mxchat_generate_embedding($sanitized_content, $bot_id);
	1367	+
	1368	+ if (is_string($embedding_vector)) {
	1369	+ //error_log('[SITEMAP DEBUG] Error generating embedding: ' . $embedding_vector);
	1370	+ $error_message = esc_html__('Failed to generate embedding: ', 'mxchat') . esc_html($embedding_vector);
	1371	+
	1372	+ set_transient('mxchat_admin_notice_error', $error_message, 30);
	1373	+
	1374	+ set_transient('mxchat_single_url_status', [
	1375	+ 'url' => $submitted_url,
	1376	+ 'timestamp' => current_time('mysql'),
	1377	+ 'status' => 'failed',
	1378	+ 'error' => $error_message
	1379	+ ], DAY_IN_SECONDS);
	1380	+
	1381	+ wp_safe_redirect(esc_url(admin_url('admin.php?page=mxchat-prompts')));
	1382	+ exit;
	1383	+ }
	1384	+
	1385	+ if (is_array($embedding_vector)) {
	1386	+ //error_log('[SITEMAP DEBUG] Successfully generated embedding with ' . count($embedding_vector) . ' dimensions');
	1387	+
	1388	+ // UPDATED: Pass bot_id to database submission
	1389	+ $db_result = MxChat_Utils::submit_content_to_db(
	1390	+ $sanitized_content,
	1391	+ $submitted_url,
	1392	+ $api_key,
	1393	+ null,
	1394	+ $bot_id
	1395	+ );
	1396	+
	1397	+ if (is_wp_error($db_result)) {
	1398	+ //error_log('[SITEMAP DEBUG] Error: Failed to store content in database: ' . $db_result->get_error_message());
	1399	+ $error_message = esc_html__('Failed to store content in database: ', 'mxchat') . esc_html($db_result->get_error_message());
	1400	+
	1401	+ set_transient('mxchat_admin_notice_error', $error_message, 30);
	1402	+
	1403	+ set_transient('mxchat_single_url_status', [
	1404	+ 'url' => $submitted_url,
	1405	+ 'timestamp' => current_time('mysql'),
	1406	+ 'status' => 'failed',
	1407	+ 'error' => $error_message
	1408	+ ], DAY_IN_SECONDS);
	1409	+
	1410	+ wp_safe_redirect(esc_url(admin_url('admin.php?page=mxchat-prompts')));
	1411	+ exit;
	1412	+ }
	1413	+
	1414	+ //error_log('[SITEMAP DEBUG] Successfully stored content in database');
	1415	+ $success_message = esc_html__('URL content successfully submitted!', 'mxchat');
	1416	+
	1417	+ set_transient('mxchat_admin_notice_success', $success_message, 30);
	1418	+
	1419	+ set_transient('mxchat_single_url_status', [
	1420	+ 'url' => $submitted_url,
	1421	+ 'timestamp' => current_time('mysql'),
	1422	+ 'status' => 'complete',
	1423	+ 'content_length' => strlen($sanitized_content),
	1424	+ 'embedding_dimensions' => count($embedding_vector)
	1425	+ ], DAY_IN_SECONDS);
	1426	+
	1427	+ } else {
	1428	+ //error_log('[SITEMAP DEBUG] Error: Failed to generate embedding. Unexpected result type: ' . gettype($embedding_vector));
	1429	+ $error_message = esc_html__('Failed to generate embedding: Unexpected result type. Please check your API key and try again.', 'mxchat');
	1430	+
	1431	+ set_transient('mxchat_admin_notice_error', $error_message, 30);
	1432	+
	1433	+ set_transient('mxchat_single_url_status', [
	1434	+ 'url' => $submitted_url,
	1435	+ 'timestamp' => current_time('mysql'),
	1436	+ 'status' => 'failed',
	1437	+ 'error' => $error_message
	1438	+ ], DAY_IN_SECONDS);
	1439	+ }
	1440	+
	1441	+ //error_log('[SITEMAP DEBUG] ===== Completed URL submission process =====');
	1442	+ wp_safe_redirect(esc_url(admin_url('admin.php?page=mxchat-prompts')));
	1443	+ exit;
	1444	+}
	1445	+
	1446	+public function mxchat_get_single_url_status() {
	1447	+ $status = get_transient('mxchat_single_url_status');
	1448	+ if (!$status) {
	1449	+ return null;
	1450	+ }
	1451	+
	1452	+ // Add human-readable time
	1453	+ if (isset($status['timestamp'])) {
	1454	+ $status['human_time'] = human_time_diff(strtotime($status['timestamp']), current_time('timestamp')) . ' ' . __('ago', 'mxchat');
	1455	+ }
	1456	+
	1457	+ return $status;
	1458	+}
	1459	+
	1460	+public function mxchat_handle_sitemap_for_knowledge_base($xml, $sitemap_url, $bot_id = 'default') {
	1461	+ delete_transient('mxchat_single_url_status');
	1462	+ if (!current_user_can('manage_options')) {
	1463	+ //error_log(esc_html__('Unauthorized sitemap processing attempt', 'mxchat'));
	1464	+ return false;
	1465	+ }
	1466	+
	1467	+ try {
	1468	+ $sitemap_url = esc_url_raw($sitemap_url);
	1469	+
	1470	+ if (!$xml || !is_object($xml)) {
	1471	+ throw new Exception(__('Invalid XML object provided', 'mxchat'));
	1472	+ }
	1473	+
	1474	+ // UPDATED: Get bot-specific embedding API for validation
	1475	+ $bot_options = $this->get_bot_options($bot_id);
	1476	+ $options = !empty($bot_options) ? $bot_options : get_option('mxchat_options');
	1477	+
	1478	+ // ADD THIS: Test the embedding API before processing
	1479	+ $test_phrase = "Test embedding generation for MxChat";
	1480	+ $test_result = $this->mxchat_generate_embedding($test_phrase, $bot_id);
	1481	+
	1482	+ if (is_string($test_result)) {
	1483	+ //error_log('[MXCHAT-URL] Embedding API validation failed: ' . $test_result);
	1484	+
	1485	+ $status_data = array(
	1486	+ 'total_urls' => 0,
	1487	+ 'processed_urls' => 0,
	1488	+ 'status' => 'error',
	1489	+ 'error' => __('Embedding API validation failed: ', 'mxchat') . $test_result,
	1490	+ 'last_update' => time(),
	1491	+ 'bot_id' => $bot_id // ADDED
	1492	+ );
	1493	+
	1494	+ set_transient(
	1495	+ sanitize_key('mxchat_sitemap_status_' . md5($sitemap_url)),
	1496	+ array_map('sanitize_text_field', $status_data),
	1497	+ DAY_IN_SECONDS
	1498	+ );
	1499	+
	1500	+ throw new Exception(__('Embedding API validation failed: ', 'mxchat') . $test_result);
	1501	+ }
	1502	+
	1503	+ if (!is_array($test_result)) {
	1504	+ //error_log('[MXCHAT-URL] Embedding API returned unexpected result type: ' . gettype($test_result));
	1505	+ throw new Exception(__('Embedding API returned unexpected result type. Please check your configuration.', 'mxchat'));
	1506	+ }
	1507	+
	1508	+ $urls = [];
	1509	+ foreach ($xml->url as $url_element) {
	1510	+ $url = esc_url_raw((string)$url_element->loc);
	1511	+ if ($url) {
	1512	+ $urls[] = $url;
	1513	+ }
	1514	+ }
	1515	+
	1516	+ $total_urls = absint(count($urls));
	1517	+
	1518	+ if ($total_urls < 1) {
	1519	+ throw new Exception(__('No valid URLs found in sitemap', 'mxchat'));
	1520	+ }
	1521	+
	1522	+ // UPDATED: Pass bot_id to sitemap processing cron
	1523	+ wp_schedule_single_event(time(), 'mxchat_process_sitemap_urls', array(
	1524	+ 'urls' => $urls,
	1525	+ 'sitemap_url' => $sitemap_url,
	1526	+ 'total_urls' => $total_urls,
	1527	+ 'batch_size' => absint(10),
	1528	+ 'batch_pause' => absint(5),
	1529	+ 'bot_id' => $bot_id // ADDED
	1530	+ ));
	1531	+
	1532	+ $status_data = array(
	1533	+ 'total_urls' => $total_urls,
	1534	+ 'processed_urls' => 0,
	1535	+ 'status' => 'processing',
	1536	+ 'last_update' => time(),
	1537	+ 'bot_id' => $bot_id // ADDED
	1538	+ );
	1539	+
	1540	+ set_transient(
	1541	+ sanitize_key('mxchat_sitemap_status_' . md5($sitemap_url)),
	1542	+ array_map('sanitize_text_field', $status_data),
	1543	+ DAY_IN_SECONDS
	1544	+ );
	1545	+
	1546	+ return __('scheduled', 'mxchat');
	1547	+
	1548	+ } catch (\Exception $e) {
	1549	+ $error_message = $e->getMessage();
	1550	+ //error_log(sprintf(esc_html__('Error preparing sitemap for processing: %s', 'mxchat'), esc_html($error_message)));
	1551	+
	1552	+ set_transient(
	1553	+ 'mxchat_last_sitemap_url',
	1554	+ sanitize_text_field($sitemap_url),
	1555	+ DAY_IN_SECONDS
	1556	+ );
	1557	+
	1558	+ $status_data = array(
	1559	+ 'total_urls' => 0,
	1560	+ 'processed_urls' => 0,
	1561	+ 'status' => 'error',
	1562	+ 'error' => $error_message,
	1563	+ 'last_update' => time(),
	1564	+ 'bot_id' => $bot_id // ADDED
	1565	+ );
	1566	+
	1567	+ set_transient(
	1568	+ sanitize_key('mxchat_sitemap_status_' . md5($sitemap_url)),
	1569	+ array_map('sanitize_text_field', $status_data),
	1570	+ DAY_IN_SECONDS
	1571	+ );
	1572	+
	1573	+ return $error_message;
	1574	+ }
	1575	+}
	1576	+public function mxchat_process_sitemap_urls_cron($urls, $sitemap_url, $total_urls, $batch_size, $batch_pause, $bot_id = 'default') {
	1577	+ // Validate inputs
	1578	+ $sitemap_url = esc_url_raw($sitemap_url);
	1579	+ $total_urls = absint($total_urls);
	1580	+ $batch_size = absint($batch_size);
	1581	+ $batch_pause = absint($batch_pause);
	1582	+ $bot_id = sanitize_key($bot_id);
	1583	+
	1584	+ if (!is_array($urls) || empty($urls)) {
	1585	+ return;
	1586	+ }
	1587	+
	1588	+ try {
	1589	+ $status_key = sanitize_key('mxchat_sitemap_status_' . md5($sitemap_url));
	1590	+ $status = get_transient($status_key);
	1591	+
	1592	+ if (!$status || !is_array($status)) {
	1593	+ throw new Exception('Invalid status data retrieved from transient');
	1594	+ }
	1595	+
	1596	+ // Initialize failed_urls array if it doesn't exist
	1597	+ if (!isset($status['failed_urls_list']) || !is_array($status['failed_urls_list'])) {
	1598	+ $status['failed_urls_list'] = [];
	1599	+ }
	1600	+
	1601	+ $start_url = absint($status['processed_urls']);
	1602	+ $end_url = min($start_url + $batch_size, $total_urls);
	1603	+
	1604	+ // UPDATED: Get bot-specific options
	1605	+ $bot_options = $this->get_bot_options($bot_id);
	1606	+ $options = !empty($bot_options) ? $bot_options : get_option('mxchat_options');
	1607	+
	1608	+ // Track batch statistics
	1609	+ $batch_stats = [
	1610	+ 'processed' => 0,
	1611	+ 'failed' => 0,
	1612	+ 'last_error' => '',
	1613	+ 'embedding_errors' => 0,
	1614	+ 'network_errors' => 0,
	1615	+ 'timeout_errors' => 0
	1616	+ ];
	1617	+
	1618	+ @set_time_limit(300);
	1619	+
	1620	+ for ($i = $start_url; $i < $end_url; $i++) {
	1621	+ $page_url = esc_url_raw($urls[$i]);
	1622	+ $max_retries = 5;
	1623	+ $retry_count = 0;
	1624	+ $url_processed = false;
	1625	+ $last_error = '';
	1626	+
	1627	+ if (memory_get_usage(true) > (1024 * 1024 * 100)) {
	1628	+ //error_log('MxChat: Memory usage high, taking break');
	1629	+ sleep(2);
	1630	+ }
	1631	+
	1632	+ while (!$url_processed && $retry_count < $max_retries) {
	1633	+ try {
	1634	+ $timeout = 30 + ($retry_count * 10);
	1635	+
	1636	+ $page_response = wp_remote_get($page_url, array(
	1637	+ 'timeout' => $timeout,
	1638	+ 'redirection' => 5,
	1639	+ 'user-agent' => 'MxChat/1.0'
	1640	+ ));
	1641	+
	1642	+ if (is_wp_error($page_response)) {
	1643	+ $error_msg = $page_response->get_error_message();
	1644	+
	1645	+ if (strpos($error_msg, 'timeout') !== false) {
	1646	+ $batch_stats['timeout_errors']++;
	1647	+ } else {
	1648	+ $batch_stats['network_errors']++;
	1649	+ }
	1650	+
	1651	+ throw new Exception('HTTP request failed: ' . $error_msg);
	1652	+ }
	1653	+
	1654	+ $response_code = wp_remote_retrieve_response_code($page_response);
	1655	+
	1656	+ if (!in_array($response_code, [200, 201, 202])) {
	1657	+ if ($response_code >= 400 && $response_code < 500) {
	1658	+ throw new Exception('HTTP Status: ' . $response_code . ' (permanent failure)');
	1659	+ }
	1660	+ throw new Exception('HTTP Status: ' . $response_code);
	1661	+ }
	1662	+
	1663	+ $page_html = wp_remote_retrieve_body($page_response);
	1664	+
	1665	+ if (empty($page_html)) {
	1666	+ throw new Exception('Empty response body');
	1667	+ }
	1668	+
	1669	+ $page_content = $this->mxchat_extract_main_content($page_html);
	1670	+ $sanitized_content = $this->mxchat_sanitize_content_for_api($page_content);
	1671	+
	1672	+ if (empty($sanitized_content)) {
	1673	+ //error_log("MxChat: No content found for URL: {$page_url}");
	1674	+ $url_processed = true;
	1675	+ $batch_stats['processed']++;
	1676	+ break;
	1677	+ }
	1678	+
	1679	+ // UPDATED: Use bot-specific embedding generation
	1680	+ $embedding_vector = $this->mxchat_generate_embedding($sanitized_content, $bot_id);
	1681	+
	1682	+ if (is_string($embedding_vector)) {
	1683	+ $batch_stats['embedding_errors']++;
	1684	+
	1685	+ if (strpos($embedding_vector, 'rate limit') !== false ||
	1686	+ strpos($embedding_vector, 'quota') !== false) {
	1687	+ sleep(30 + ($retry_count * 10));
	1688	+ }
	1689	+
	1690	+ throw new Exception('Embedding generation failed: ' . $embedding_vector);
	1691	+ }
	1692	+
	1693	+ if (!is_array($embedding_vector)) {
	1694	+ throw new Exception('Embedding generation returned unexpected result type: ' . gettype($embedding_vector));
	1695	+ }
	1696	+
	1697	+ // UPDATED: Pass bot_id to database submission
	1698	+ $submission_result = MxChat_Utils::submit_content_to_db($sanitized_content, $page_url, $options['api_key'], null, $bot_id);
	1699	+
	1700	+ if (is_wp_error($submission_result)) {
	1701	+ throw new Exception('Database submission failed: ' . $submission_result->get_error_message());
	1702	+ }
	1703	+
	1704	+ // Success!
	1705	+ $url_processed = true;
	1706	+ $batch_stats['processed']++;
	1707	+
	1708	+ } catch (Exception $e) {
	1709	+ $retry_count++;
	1710	+ $last_error = $e->getMessage();
	1711	+
	1712	+ if (strpos($last_error, 'rate limit') !== false) {
	1713	+ sleep(60);
	1714	+ } elseif (strpos($last_error, 'timeout') !== false) {
	1715	+ sleep(10);
	1716	+ } elseif (strpos($last_error, 'permanent failure') !== false) {
	1717	+ break;
	1718	+ } else {
	1719	+ sleep(pow(2, $retry_count - 1));
	1720	+ }
	1721	+ }
	1722	+ }
	1723	+
	1724	+ // If URL still not processed after all retries, mark as failed
	1725	+ if (!$url_processed) {
	1726	+ $batch_stats['failed']++;
	1727	+ $batch_stats['last_error'] = $last_error;
	1728	+
	1729	+ $status['failed_urls_list'][] = [
	1730	+ 'url' => $page_url,
	1731	+ 'error' => $last_error,
	1732	+ 'time' => time(),
	1733	+ 'retries' => $max_retries
	1734	+ ];
	1735	+
	1736	+ if (count($status['failed_urls_list']) > 100) {
	1737	+ $status['failed_urls_list'] = array_slice($status['failed_urls_list'], -100);
	1738	+ }
	1739	+ }
	1740	+
	1741	+ // Update progress after each URL
	1742	+ $status['processed_urls'] = absint($i + 1);
	1743	+ $status['last_update'] = time();
	1744	+ $status['failed_urls'] = absint($status['failed_urls'] ?? 0) + ($url_processed ? 0 : 1);
	1745	+ $status['last_error'] = $batch_stats['last_error'];
	1746	+
	1747	+ set_transient($status_key, $status, DAY_IN_SECONDS);
	1748	+ }
	1749	+
	1750	+ $failure_rate = $batch_stats['failed'] / max(1, $batch_stats['processed'] + $batch_stats['failed']);
	1751	+
	1752	+ if ($batch_stats['processed'] === 0 && $batch_stats['failed'] >= 5) {
	1753	+ $status['status'] = 'error';
	1754	+ $status['error'] = sprintf(
	1755	+ 'Processing stopped after %d consecutive failures. Last error: %s',
	1756	+ $batch_stats['failed'],
	1757	+ $batch_stats['last_error']
	1758	+ );
	1759	+ set_transient($status_key, $status, DAY_IN_SECONDS);
	1760	+ return;
	1761	+ }
	1762	+
	1763	+ // Update final progress
	1764	+ $status['processed_urls'] = min($end_url, $total_urls);
	1765	+ $status['last_update'] = time();
	1766	+ $status['batch_stats'] = $batch_stats;
	1767	+ set_transient($status_key, $status, DAY_IN_SECONDS);
	1768	+
	1769	+ // Check if we've processed all URLs
	1770	+ if ($end_url >= $total_urls) {
	1771	+ // All URLs have been processed - mark as complete
	1772	+ $status['status'] = 'complete';
	1773	+ $status['processed_urls'] = $total_urls;
	1774	+
	1775	+ $status['completion_summary'] = [
	1776	+ 'total_urls' => $total_urls,
	1777	+ 'successful_urls' => $total_urls - absint($status['failed_urls'] ?? 0),
	1778	+ 'failed_urls' => absint($status['failed_urls'] ?? 0),
	1779	+ 'completion_time' => current_time('mysql'),
	1780	+ 'final_batch_stats' => $batch_stats
	1781	+ ];
	1782	+
	1783	+ set_transient($status_key, $status, DAY_IN_SECONDS);
	1784	+ } else {
	1785	+ $dynamic_pause = $batch_pause;
	1786	+
	1787	+ if ($failure_rate > 0.5) {
	1788	+ $dynamic_pause *= 3;
	1789	+ } elseif ($batch_stats['embedding_errors'] > 3) {
	1790	+ $dynamic_pause *= 2;
	1791	+ }
	1792	+
	1793	+ // UPDATED: Pass bot_id to next batch
	1794	+ wp_schedule_single_event(time() + $dynamic_pause, 'mxchat_process_sitemap_urls', array(
	1795	+ 'urls' => $urls,
	1796	+ 'sitemap_url' => $sitemap_url,
	1797	+ 'total_urls' => $total_urls,
	1798	+ 'batch_size' => $batch_size,
	1799	+ 'batch_pause' => $batch_pause,
	1800	+ 'bot_id' => $bot_id // ADDED
	1801	+ ));
	1802	+ }
	1803	+ } catch (\Exception $e) {
	1804	+ $status['last_error'] = $e->getMessage();
	1805	+ $status['error_count'] = ($status['error_count'] ?? 0) + 1;
	1806	+
	1807	+ if ($status['error_count'] >= 5) {
	1808	+ $status['status'] = 'error';
	1809	+ $status['error'] = 'Too many batch failures: ' . $e->getMessage();
	1810	+ } else {
	1811	+ // UPDATED: Pass bot_id to retry batch
	1812	+ wp_schedule_single_event(time() + 300, 'mxchat_process_sitemap_urls', array(
	1813	+ 'urls' => $urls,
	1814	+ 'sitemap_url' => $sitemap_url,
	1815	+ 'total_urls' => $total_urls,
	1816	+ 'batch_size' => max(5, $batch_size / 2),
	1817	+ 'batch_pause' => $batch_pause * 2,
	1818	+ 'bot_id' => $bot_id // ADDED
	1819	+ ));
	1820	+ }
	1821	+
	1822	+ set_transient($status_key, $status, DAY_IN_SECONDS);
	1823	+ }
	1824	+}
	1825	+
	1826	+
	1827	+public function mxchat_sanitize_content_for_api($content) {
	1828	+ //error_log('[MXCHAT-SANITIZE] Original content preview: ' . substr($content, 0, 500) . '...');
	1829	+
	1830	+ // Remove script, style tags, and HTML comments
	1831	+ $content = preg_replace('/<script\b[^>]*>(.*?)<\/script>/is', '', $content);
	1832	+ $content = preg_replace('/<style\b[^>]*>(.*?)<\/style>/is', '', $content);
	1833	+ $content = preg_replace('/<!--(.|\s)*?-->/', '', $content);
	1834	+
	1835	+ // Remove all HTML tags and decode HTML entities
	1836	+ $content = wp_strip_all_tags($content);
	1837	+ $content = html_entity_decode($content, ENT_QUOTES | ENT_HTML5);
	1838	+
	1839	+ // Normalize whitespace but preserve paragraph breaks
	1840	+ // First, normalize line endings to \n
	1841	+ $content = str_replace(["\r\n", "\r"], "\n", $content);
	1842	+ // Replace multiple spaces/tabs with single space, but preserve newlines
	1843	+ $content = preg_replace('/[ \t]+/', ' ', $content);
	1844	+ // Replace 3+ newlines with 2 newlines (max 2 blank lines)
	1845	+ $content = preg_replace('/\n{3,}/', "\n\n", $content);
	1846	+ // Trim each line
	1847	+ $lines = explode("\n", $content);
	1848	+ $lines = array_map('trim', $lines);
	1849	+ $content = implode("\n", $lines);
	1850	+ // Final trim
	1851	+ $content = trim($content);
	1852	+
	1853	+ // Remove control characters (which can cause database issues) but preserve \n (0x0A) and \r (0x0D)
	1854	+ $content = preg_replace('/[\x00-\x08\x0B\x0C\x0E-\x1F\x7F]/u', '', $content);
	1855	+
	1856	+ // Remove NULL bytes which can cause database errors
	1857	+ $content = str_replace("\0", "", $content);
	1858	+
	1859	+ // Ensure valid UTF-8 encoding
	1860	+ $content = wp_check_invalid_utf8($content);
	1861	+
	1862	+ // Remove any extremely long strings without spaces (often garbage)
	1863	+ $content = preg_replace('/\S{300,}/', ' ', $content);
	1864	+
	1865	+ // Replace problematic characters that often cause database issues
	1866	+ $content = preg_replace('/[\x{10000}-\x{10FFFF}]/u', '', $content); // Remove emoji and other high Unicode characters
	1867	+
	1868	+ // Replace any remaining potentially problematic characters with spaces
	1869	+ // BUT preserve newlines by temporarily replacing them
	1870	+ $content = str_replace("\n", "NEWLINE_PLACEHOLDER", $content);
	1871	+ $content = preg_replace('/[^\p{L}\p{N}\p{P}\p{Z}\p{Sm}]/u', ' ', $content);
	1872	+ $content = str_replace("NEWLINE_PLACEHOLDER", "\n", $content);
	1873	+
	1874	+ // Limit to reasonable length if needed
	1875	+ $max_length = 65000; // Just under MySQL TEXT field limit
	1876	+ if (strlen($content) > $max_length) {
	1877	+ $content = substr($content, 0, $max_length);
	1878	+ }
	1879	+
	1880	+ //error_log('[MXCHAT-SANITIZE] Sanitized content preview: ' . substr($content, 0, 500) . '...');
	1881	+ return $content;
	1882	+}
	1883	+public function mxchat_extract_main_content($html) {
	1884	+ if (empty($html)) {
	1885	+ return '';
	1886	+ }
	1887	+ try {
	1888	+ $dom = new DOMDocument;
	1889	+ libxml_use_internal_errors(true); // Suppress HTML parsing errors
	1890	+ @$dom->loadHTML(mb_convert_encoding($html, 'HTML-ENTITIES', 'UTF-8'), LIBXML_HTML_NOIMPLIED | LIBXML_HTML_NODEFDTD);
	1891	+ $xpath = new DOMXPath($dom);
	1892	+
	1893	+ // For debugging purposes
	1894	+ $debugEnabled = false; // Set to true to enable debugging output
	1895	+ $debug = function($message) use ($debugEnabled) {
	1896	+ if ($debugEnabled) {
	1897	+ //error_log('[MXCHAT-DEBUG] ' . $message);
	1898	+ }
	1899	+ };
	1900	+
	1901	+ // Direct targeting for Gerow theme posts
	1902	+ $post_text = $xpath->query('//div[contains(@class, "post-text")]');
	1903	+ if ($post_text && $post_text->length > 0) {
	1904	+ $debug("Found post-text directly");
	1905	+ $content = '';
	1906	+ foreach ($post_text as $node) {
	1907	+ $content .= $dom->saveHTML($node);
	1908	+ }
	1909	+ if (!empty($content)) {
	1910	+ $debug("Returning post-text content");
	1911	+ return $content;
	1912	+ }
	1913	+ }
	1914	+
	1915	+ // Try to get the blog details content which contains the post-text
	1916	+ $blog_details = $xpath->query('//div[contains(@class, "blog-details-content")]');
	1917	+ if ($blog_details && $blog_details->length > 0) {
	1918	+ $debug("Found blog-details-content");
	1919	+ $content = '';
	1920	+ foreach ($blog_details as $node) {
	1921	+ $content .= $dom->saveHTML($node);
	1922	+ }
	1923	+ if (!empty($content)) {
	1924	+ $debug("Returning blog-details-content");
	1925	+ return $content;
	1926	+ }
	1927	+ }
	1928	+
	1929	+ // Try to get the article which contains the blog details
	1930	+ $article = $xpath->query('//article[contains(@class, "blog-details-wrap")]');
	1931	+ if ($article && $article->length > 0) {
	1932	+ $debug("Found article with blog-details-wrap");
	1933	+ $content = '';
	1934	+ foreach ($article as $node) {
	1935	+ $content .= $dom->saveHTML($node);
	1936	+ }
	1937	+ if (!empty($content)) {
	1938	+ $debug("Returning article content");
	1939	+ return $content;
	1940	+ }
	1941	+ }
	1942	+
	1943	+ // Try even broader with the blog-item-wrap
	1944	+ $blog_item = $xpath->query('//div[contains(@class, "blog-item-wrap")]');
	1945	+ if ($blog_item && $blog_item->length > 0) {
	1946	+ $debug("Found blog-item-wrap");
	1947	+ $content = '';
	1948	+ foreach ($blog_item as $node) {
	1949	+ $content .= $dom->saveHTML($node);
	1950	+ }
	1951	+ if (!empty($content)) {
	1952	+ $debug("Returning blog-item-wrap content");
	1953	+ return $content;
	1954	+ }
	1955	+ }
	1956	+
	1957	+ // Specific Gerow theme path
	1958	+ $gerow_path = $xpath->query('//section[contains(@class, "blog-area")]//div[contains(@class, "post-text")]');
	1959	+ if ($gerow_path && $gerow_path->length > 0) {
	1960	+ $debug("Found Gerow theme path to post-text");
	1961	+ $content = '';
	1962	+ foreach ($gerow_path as $node) {
	1963	+ $content .= $dom->saveHTML($node);
	1964	+ }
	1965	+ if (!empty($content)) {
	1966	+ $debug("Returning Gerow post-text content");
	1967	+ return $content;
	1968	+ }
	1969	+ }
	1970	+
	1971	+ // Generic blog post selectors
	1972	+ $selectors = [
	1973	+ // Blog post specific selectors
	1974	+ '//div[contains(@class, "post-text")]',
	1975	+ '//article[contains(@class, "blog-post-item")]//div[contains(@class, "post-text")]',
	1976	+ '//div[contains(@class, "blog-details-content")]',
	1977	+ '//article[contains(@class, "blog-details-wrap")]',
	1978	+ '//div[contains(@class, "entry-content")]',
	1979	+ '//div[contains(@class, "blog-content")]',
	1980	+ '//div[contains(@class, "blog-item-wrap")]',
	1981	+
	1982	+ // More general content selectors
	1983	+ '//div[contains(@class, "page__content")]',
	1984	+ '//div[contains(@class, "elementor-widget-container")]',
	1985	+ '//div[contains(@class, "elementor-text-editor")]',
	1986	+ '//div[contains(@class, "elementor-widget-text-editor")]',
	1987	+ '//*[contains(@class, "entry-content")]',
	1988	+ '//*[contains(@class, "post-content")]',
	1989	+ '//*[contains(@class, "article-content")]',
	1990	+ '//*[@id="content"]',
	1991	+ '//*[@id="main-content"]',
	1992	+ '//section[contains(@class, "blog-area")]',
	1993	+ '//article',
	1994	+ '//main',
	1995	+ '//div[contains(@class, "content")]'
	1996	+ ];
	1997	+
	1998	+ // First handle Elementor content
	1999	+ $debug("Checking for Elementor content");
	2000	+ $elementor_widgets = $xpath->query('//div[contains(@class, "elementor-element")]//div[contains(@class, "elementor-widget-container")]');
	2001	+ if ($elementor_widgets && $elementor_widgets->length > 0) {
	2002	+ $debug("Found Elementor widgets");
	2003	+ $combined_content = '';
	2004	+ foreach ($elementor_widgets as $widget) {
	2005	+ $widget_content = $dom->saveHTML($widget);
	2006	+ if (!empty($widget_content)) {
	2007	+ $combined_content .= $widget_content;
	2008	+ }
	2009	+ }
	2010	+ if (!empty($combined_content)) {
	2011	+ $debug("Returning Elementor content");
	2012	+ return $combined_content;
	2013	+ }
	2014	+ }
	2015	+
	2016	+ // Try standard selectors one by one
	2017	+ foreach ($selectors as $selector) {
	2018	+ $debug("Trying selector: " . $selector);
	2019	+ $nodes = $xpath->query($selector);
	2020	+ if ($nodes && $nodes->length > 0) {
	2021	+ $debug("Found matches for selector: " . $selector);
	2022	+ $content = '';
	2023	+ foreach ($nodes as $node) {
	2024	+ $content .= $dom->saveHTML($node);
	2025	+ }
	2026	+ if (!empty($content)) {
	2027	+ $debug("Returning content from selector: " . $selector);
	2028	+ return $content;
	2029	+ }
	2030	+ }
	2031	+ }
	2032	+
	2033	+ // Manual regex fallback for post-text if DOM methods fail
	2034	+ $debug("Trying regex fallback");
	2035	+ if (preg_match('/<div class="post-text">(.*?)<\/div>\s*<\/div>\s*<\/div>/s', $html, $matches)) {
	2036	+ $debug("Found post-text via regex");
	2037	+ return '<div class="post-text">' . $matches[1] . '</div>';
	2038	+ }
	2039	+
	2040	+ // Try to extract the blog section as a whole
	2041	+ $blog_section = $xpath->query('//section[contains(@class, "blog-area")]');
	2042	+ if ($blog_section && $blog_section->length > 0) {
	2043	+ $debug("Found blog-area section");
	2044	+ $content = '';
	2045	+ foreach ($blog_section as $node) {
	2046	+ $content .= $dom->saveHTML($node);
	2047	+ }
	2048	+ if (!empty($content)) {
	2049	+ $debug("Returning blog-area section content");
	2050	+ return $content;
	2051	+ }
	2052	+ }
	2053	+
	2054	+ // Fallback: Return the body content if no specific selector matches
	2055	+ $debug("Using body fallback");
	2056	+ $body = $dom->getElementsByTagName('body');
	2057	+ if ($body->length > 0) {
	2058	+ return $dom->saveHTML($body->item(0));
	2059	+ }
	2060	+
	2061	+ // Last resort: return the original HTML
	2062	+ $debug("Returning original HTML");
	2063	+ return $html;
	2064	+ } catch (Exception $e) {
	2065	+ //error_log('[MXCHAT-ERROR] Content extraction failed: ' . $e->getMessage());
	2066	+ return $html; // Return original HTML if parsing fails
	2067	+ } finally {
	2068	+ libxml_clear_errors();
	2069	+ }
	2070	+}
	2071	+public function mxchat_get_sitemap_processing_status($sitemap_url) {
	2072	+ $sitemap_url = esc_url_raw($sitemap_url);
	2073	+ $status_key = sanitize_key('mxchat_sitemap_status_' . md5($sitemap_url));
	2074	+ $status = get_transient($status_key);
	2075	+
	2076	+ if (!$status || !is_array($status)) {
	2077	+ return false;
	2078	+ }
	2079	+
	2080	+ // Auto-complete check: if all URLs are processed but status isn't complete
	2081	+ if (isset($status['processed_urls']) && isset($status['total_urls']) &&
	2082	+ $status['processed_urls'] >= $status['total_urls'] &&
	2083	+ isset($status['status']) && $status['status'] !== 'complete' &&
	2084	+ $status['status'] !== 'error') {
	2085	+
	2086	+ // Mark as complete
	2087	+ $status['status'] = 'complete';
	2088	+ $status['processed_urls'] = $status['total_urls']; // Ensure exact match
	2089	+
	2090	+ // Update the transient with the corrected status
	2091	+ set_transient($status_key, $status, DAY_IN_SECONDS);
	2092	+ }
	2093	+
	2094	+ return array(
	2095	+ 'total_urls' => absint($status['total_urls']),
	2096	+ 'processed_urls' => absint($status['processed_urls']),
	2097	+ 'failed_urls' => absint($status['failed_urls'] ?? 0),
	2098	+ 'percentage' => ($status['total_urls'] > 0)
	2099	+ ? round((absint($status['processed_urls']) / absint($status['total_urls'])) * 100)
	2100	+ : 0,
	2101	+ 'status' => sanitize_text_field($status['status']),
	2102	+ 'last_update' => human_time_diff(absint($status['last_update']), time()) . ' ' . esc_html__('ago', 'mxchat'),
	2103	+ 'error' => isset($status['error']) ? sanitize_text_field($status['error']) : '',
	2104	+ 'last_error' => isset($status['last_error']) ? sanitize_text_field($status['last_error']) : '',
	2105	+ 'failed_urls_list' => isset($status['failed_urls_list']) ? $status['failed_urls_list'] : array()
	2106	+ );
	2107	+}
	2108	+
	2109	+public function mxchat_ajax_get_status_updates() {
	2110	+ try {
	2111	+ // Verify the request
	2112	+ check_ajax_referer('mxchat_status_nonce', 'nonce');
	2113	+
	2114	+ // Get the status just like in your admin page
	2115	+ $pdf_url = get_transient('mxchat_last_pdf_url');
	2116	+ $sitemap_url = get_transient('mxchat_last_sitemap_url');
	2117	+
	2118	+ $pdf_status = $pdf_url ? $this->mxchat_get_pdf_processing_status($pdf_url) : false;
	2119	+ $sitemap_status = $sitemap_url ? $this->mxchat_get_sitemap_processing_status($sitemap_url) : false;
	2120	+
	2121	+ // Add the PDF URL to the status object
	2122	+ if ($pdf_status && $pdf_url) {
	2123	+ $pdf_status['pdf_url'] = $pdf_url;
	2124	+ }
	2125	+
	2126	+ // Set the current PDF URL for the manual batch processing button
	2127	+ $current_pdf_url = $pdf_url;
	2128	+
	2129	+ // Check for true processing status, not just presence of status
	2130	+ $is_active_processing =
	2131	+ ($sitemap_status && isset($sitemap_status['status']) && $sitemap_status['status'] === 'processing') ||
	2132	+ ($pdf_status && isset($pdf_status['status']) && $pdf_status['status'] === 'processing');
	2133	+
	2134	+ // Get single URL status, but only if no processing is active
	2135	+ $single_url_status = !$is_active_processing ? $this->mxchat_get_single_url_status() : false;
	2136	+
	2137	+ // REMOVED: Auto-clearing of completed status - now only done via dismiss button
	2138	+
	2139	+ // Return JSON response with the status data
	2140	+ wp_send_json(array(
	2141	+ 'pdf_status' => $pdf_status,
	2142	+ 'sitemap_status' => $sitemap_status,
	2143	+ 'single_url_status' => $single_url_status,
	2144	+ 'is_processing' => $is_active_processing,
	2145	+ 'current_pdf_url' => $current_pdf_url
	2146	+ ));
	2147	+
	2148	+ } catch (Exception $e) {
	2149	+ // Log the error
	2150	+ //error_log('MxChat Status Update Error: ' . $e->getMessage());
	2151	+
	2152	+ // Return a friendly error response
	2153	+ wp_send_json_error(array(
	2154	+ 'message' => 'Error getting status updates: ' . $e->getMessage(),
	2155	+ 'status' => 'error'
	2156	+ ));
	2157	+ }
	2158	+}
	2159	+public function mxchat_stop_processing() {
	2160	+ // Verify permissions
	2161	+ if (!current_user_can('manage_options')) {
	2162	+ wp_die(esc_html__('Unauthorized access', 'mxchat'));
	2163	+ }
	2164	+
	2165	+ // Verify nonce
	2166	+ check_admin_referer('mxchat_stop_processing_action', 'mxchat_stop_processing_nonce');
	2167	+
	2168	+ // Get the last sitemap URL and clear its transient
	2169	+ $sitemap_url = get_transient('mxchat_last_sitemap_url');
	2170	+ if ($sitemap_url) {
	2171	+ delete_transient('mxchat_sitemap_status_' . md5($sitemap_url));
	2172	+ delete_transient('mxchat_last_sitemap_url');
	2173	+ }
	2174	+
	2175	+ // Get the last PDF URL and clear its transient
	2176	+ $pdf_url = get_transient('mxchat_last_pdf_url');
	2177	+ if ($pdf_url) {
	2178	+ delete_transient('mxchat_pdf_status_' . md5($pdf_url));
	2179	+ delete_transient('mxchat_last_pdf_url');
	2180	+ }
	2181	+
	2182	+ // Unschedule any pending sitemap events
	2183	+ $timestamp = wp_next_scheduled('mxchat_process_sitemap_urls');
	2184	+ if ($timestamp) {
	2185	+ wp_unschedule_event($timestamp, 'mxchat_process_sitemap_urls');
	2186	+ }
	2187	+
	2188	+ // Redirect back with a success message
	2189	+ set_transient('mxchat_admin_notice_success',
	2190	+ esc_html__('Processing has been stopped successfully.', 'mxchat'),
	2191	+ 30
	2192	+ );
	2193	+ wp_safe_redirect(admin_url('admin.php?page=mxchat-prompts'));
	2194	+ exit;
	2195	+}
	2196	+
	2197	+
	2198	+/**
	2199	+ * Get content list for processing
	2200	+ */
	2201	+public function ajax_mxchat_get_content_list() {
	2202	+ // Verify the nonce
	2203	+ check_ajax_referer('mxchat_content_selector_nonce', 'nonce');
	2204	+
	2205	+ if (!current_user_can('manage_options')) {
	2206	+ wp_send_json_error(__('Unauthorized access', 'mxchat'));
	2207	+ }
	2208	+
	2209	+ $page = isset($_GET['page']) ? absint($_GET['page']) : 1;
	2210	+ $per_page = isset($_GET['per_page']) ? absint($_GET['per_page']) : 50;
	2211	+ $search = isset($_GET['search']) ? sanitize_text_field($_GET['search']) : '';
	2212	+ $post_type = isset($_GET['post_type']) ? sanitize_text_field($_GET['post_type']) : 'all';
	2213	+ $post_status = isset($_GET['post_status']) ? sanitize_text_field($_GET['post_status']) : 'publish';
	2214	+ $processed_filter = isset($_GET['processed_filter']) ? sanitize_text_field($_GET['processed_filter']) : 'all';
	2215	+
	2216	+ // Build query args
	2217	+ $args = array(
	2218	+ 'posts_per_page' => $per_page,
	2219	+ 'paged' => $page,
	2220	+ 'post_status' => $post_status !== 'all' ? $post_status : array('publish', 'draft', 'pending'),
	2221	+ 'orderby' => 'date',
	2222	+ 'order' => 'DESC',
	2223	+ );
	2224	+
	2225	+ // Handle post types - IMPROVED VERSION
	2226	+ if ($post_type !== 'all') {
	2227	+ $args['post_type'] = $post_type;
	2228	+ } else {
	2229	+ // Get all available post types that might contain content
	2230	+ $all_post_types = array();
	2231	+
	2232	+ // First get all public post types
	2233	+ $public_types = get_post_types(array('public' => true), 'names');
	2234	+ $all_post_types = array_merge($all_post_types, $public_types);
	2235	+
	2236	+ // Add common forum/community post types
	2237	+ $forum_types = array('topic', 'reply', 'forum', 'wpforo_topic', 'wpforo_post');
	2238	+ foreach ($forum_types as $forum_type) {
	2239	+ if (post_type_exists($forum_type)) {
	2240	+ $all_post_types[] = $forum_type;
	2241	+ }
	2242	+ }
	2243	+
	2244	+ // Add other commonly used post types
	2245	+ $common_types = array('product', 'job_listing', 'event', 'portfolio');
	2246	+ foreach ($common_types as $common_type) {
	2247	+ if (post_type_exists($common_type)) {
	2248	+ $all_post_types[] = $common_type;
	2249	+ }
	2250	+ }
	2251	+
	2252	+ // Remove duplicates and ensure we have at least some post types
	2253	+ $all_post_types = array_unique($all_post_types);
	2254	+
	2255	+ if (empty($all_post_types)) {
	2256	+ // Fallback to basic post types
	2257	+ $all_post_types = array('post', 'page');
	2258	+ }
	2259	+
	2260	+ $args['post_type'] = $all_post_types;
	2261	+
	2262	+ // Debug logging to see what post types are being queried
	2263	+ //error_log('MxChat Debug: Querying post types: ' . implode(', ', $all_post_types));
	2264	+ }
	2265	+
	2266	+ if (!empty($search)) {
	2267	+ $args['s'] = $search;
	2268	+ }
	2269	+
	2270	+ // Get processed data from storage
	2271	+ $processed_data = array();
	2272	+
	2273	+ $pinecone_options = get_option('mxchat_pinecone_addon_options', array());
	2274	+ $use_pinecone = ($pinecone_options['mxchat_use_pinecone'] ?? '0') === '1';
	2275	+
	2276	+ if ($use_pinecone && !empty($pinecone_options['mxchat_pinecone_api_key'])) {
	2277	+ // Get fresh data from Pinecone - no caching
	2278	+ $processed_data = $this->mxchat_get_pinecone_processed_content($pinecone_options);
	2279	+ } else {
	2280	+ // WordPress DB checking with better URL matching for all post types
	2281	+ global $wpdb;
	2282	+ $table_name = $wpdb->prefix . 'mxchat_system_prompt_content';
	2283	+ $processed_items = $wpdb->get_results("SELECT id, source_url, timestamp FROM {$table_name}");
	2284	+
	2285	+ if (!empty($processed_items)) {
	2286	+ foreach ($processed_items as $item) {
	2287	+ // Use improved URL matching that works for all post types
	2288	+ $post_id = $this->mxchat_url_to_post_id_improved($item->source_url);
	2289	+
	2290	+ if ($post_id) {
	2291	+ $processed_data[$post_id] = array(
	2292	+ 'db_id' => $item->id,
	2293	+ 'timestamp' => $item->timestamp,
	2294	+ 'url' => $item->source_url,
	2295	+ 'source' => 'wordpress'
	2296	+ );
	2297	+ }
	2298	+ }
	2299	+ }
	2300	+ }
	2301	+
	2302	+ // Get processed IDs as a simple array for in_array checks
	2303	+ $processed_ids = array_keys($processed_data);
	2304	+
	2305	+ // Handle processed/unprocessed filter
	2306	+ if ($processed_filter === 'processed' && !empty($processed_ids)) {
	2307	+ $args['post__in'] = $processed_ids;
	2308	+ } elseif ($processed_filter === 'unprocessed' && !empty($processed_ids)) {
	2309	+ $args['post__not_in'] = $processed_ids;
	2310	+ }
	2311	+
	2312	+ // Run the query
	2313	+ $query = new WP_Query($args);
	2314	+ $content_items = array();
	2315	+
	2316	+ if ($query->have_posts()) {
	2317	+ while ($query->have_posts()) {
	2318	+ $query->the_post();
	2319	+ $id = get_the_ID();
	2320	+ $post_date = get_the_date();
	2321	+ $excerpt = wp_trim_words(get_the_excerpt(), 20, '...');
	2322	+ $word_count = str_word_count(strip_tags(get_the_content()));
	2323	+
	2324	+ $is_processed = in_array($id, $processed_ids);
	2325	+ $processed_date = '';
	2326	+ $db_record_id = 0;
	2327	+ $data_source = 'none';
	2328	+
	2329	+ if ($is_processed && isset($processed_data[$id])) {
	2330	+ $item_data = $processed_data[$id];
	2331	+ $data_source = $item_data['source'];
	2332	+
	2333	+ if ($data_source === 'wordpress' && isset($item_data['timestamp'])) {
	2334	+ // WordPress DB format
	2335	+ $timestamp = strtotime($item_data['timestamp']);
	2336	+ $processed_date = human_time_diff($timestamp, current_time('timestamp')) . ' ago';
	2337	+ $db_record_id = $item_data['db_id'];
	2338	+ } elseif ($data_source === 'pinecone') {
	2339	+ // Pinecone format
	2340	+ $processed_date = $item_data['processed_date'];
	2341	+ $db_record_id = $item_data['db_id'];
	2342	+ }
	2343	+ }
	2344	+
	2345	+ $content_items[] = array(
	2346	+ 'id' => $id,
	2347	+ 'title' => get_the_title(),
	2348	+ 'permalink' => get_permalink(),
	2349	+ 'date' => $post_date,
	2350	+ 'type' => get_post_type(),
	2351	+ 'status' => get_post_status(),
	2352	+ 'excerpt' => $excerpt,
	2353	+ 'word_count' => $word_count,
	2354	+ 'already_processed' => $is_processed,
	2355	+ 'processed_date' => $processed_date,
	2356	+ 'db_record_id' => $db_record_id,
	2357	+ 'data_source' => $data_source
	2358	+ );
	2359	+ }
	2360	+ wp_reset_postdata();
	2361	+ }
	2362	+
	2363	+ $response = array(
	2364	+ 'items' => $content_items,
	2365	+ 'total' => $query->found_posts,
	2366	+ 'total_pages' => $query->max_num_pages,
	2367	+ 'current_page' => $page,
	2368	+ 'processed_count' => count($processed_ids)
	2369	+ );
	2370	+
	2371	+ wp_send_json_success($response);
	2372	+ exit;
	2373	+}
	2374	+
	2375	+
	2376	+/**
	2377	+ * This function handles various WooCommerce URL formats and permalink structures
	2378	+ */
	2379	+private function mxchat_url_to_post_id_improved($url) {
	2380	+ // First try the standard WordPress function
	2381	+ $post_id = url_to_postid($url);
	2382	+
	2383	+ if ($post_id > 0) {
	2384	+ return $post_id;
	2385	+ }
	2386	+
	2387	+ // If that fails, try more aggressive URL matching
	2388	+ // Remove trailing slashes and query parameters for better matching
	2389	+ $clean_url = rtrim($url, '/');
	2390	+ $clean_url = strtok($clean_url, '?'); // Remove query parameters
	2391	+
	2392	+ // Try again with cleaned URL
	2393	+ $post_id = url_to_postid($clean_url);
	2394	+ if ($post_id > 0) {
	2395	+ return $post_id;
	2396	+ }
	2397	+
	2398	+ // For bbPress forum topics, try extracting slug from URL
	2399	+ if (strpos($url, '/topic/') !== false || strpos($url, '/forums/') !== false) {
	2400	+ // Handle bbPress URLs: /forums/topic/topic-name/
	2401	+ if (preg_match('/\/forums\/topic\/([^\/\?]+)/', $url, $matches)) {
	2402	+ $topic_slug = $matches[1];
	2403	+
	2404	+ // Look up topic by slug
	2405	+ $topic = get_page_by_path($topic_slug, OBJECT, 'topic');
	2406	+ if ($topic) {
	2407	+ return $topic->ID;
	2408	+ }
	2409	+
	2410	+ // Alternative method: query by post_name
	2411	+ global $wpdb;
	2412	+ $post_id = $wpdb->get_var($wpdb->prepare(
	2413	+ "SELECT ID FROM {$wpdb->posts} WHERE post_name = %s AND post_type = 'topic' AND post_status IN ('publish', 'closed')",
	2414	+ $topic_slug
	2415	+ ));
	2416	+
	2417	+ if ($post_id) {
	2418	+ return intval($post_id);
	2419	+ }
	2420	+ }
	2421	+
	2422	+ // Handle simpler topic URLs: /topic/topic-name/
	2423	+ if (preg_match('/\/topic\/([^\/\?]+)/', $url, $matches)) {
	2424	+ $topic_slug = $matches[1];
	2425	+
	2426	+ global $wpdb;
	2427	+ $post_id = $wpdb->get_var($wpdb->prepare(
	2428	+ "SELECT ID FROM {$wpdb->posts} WHERE post_name = %s AND post_type = 'topic' AND post_status IN ('publish', 'closed')",
	2429	+ $topic_slug
	2430	+ ));
	2431	+
	2432	+ if ($post_id) {
	2433	+ return intval($post_id);
	2434	+ }
	2435	+ }
	2436	+ }
	2437	+
	2438	+ // For WooCommerce products
	2439	+ if (strpos($url, '/product/') !== false || strpos($url, 'product=') !== false) {
	2440	+ // Extract product slug from various URL formats
	2441	+ $product_slug = '';
	2442	+
	2443	+ // Handle pretty permalinks: /product/product-name/
	2444	+ if (preg_match('/\/product\/([^\/\?]+)/', $url, $matches)) {
	2445	+ $product_slug = $matches[1];
	2446	+ }
	2447	+ // Handle query parameters: ?product=product-name
	2448	+ elseif (preg_match('/[\?&]product=([^&]+)/', $url, $matches)) {
	2449	+ $product_slug = $matches[1];
	2450	+ }
	2451	+
	2452	+ if (!empty($product_slug)) {
	2453	+ // Look up product by slug
	2454	+ $product = get_page_by_path($product_slug, OBJECT, 'product');
	2455	+ if ($product) {
	2456	+ return $product->ID;
	2457	+ }
	2458	+
	2459	+ // Alternative method: query by post_name
	2460	+ global $wpdb;
	2461	+ $post_id = $wpdb->get_var($wpdb->prepare(
	2462	+ "SELECT ID FROM {$wpdb->posts} WHERE post_name = %s AND post_type = 'product' AND post_status = 'publish'",
	2463	+ $product_slug
	2464	+ ));
	2465	+
	2466	+ if ($post_id) {
	2467	+ return intval($post_id);
	2468	+ }
	2469	+ }
	2470	+ }
	2471	+
	2472	+ // Generic approach: try to extract slug and match against all post types
	2473	+ $parsed_url = wp_parse_url($clean_url);
	2474	+ $path = $parsed_url['path'] ?? '';
	2475	+
	2476	+ if (!empty($path)) {
	2477	+ // Get the last part of the path as potential slug
	2478	+ $path_parts = array_filter(explode('/', trim($path, '/')));
	2479	+ $potential_slug = end($path_parts);
	2480	+
	2481	+ if (!empty($potential_slug)) {
	2482	+ global $wpdb;
	2483	+
	2484	+ // Try to find any post with this slug
	2485	+ $post_id = $wpdb->get_var($wpdb->prepare(
	2486	+ "SELECT ID FROM {$wpdb->posts}
	2487	+ WHERE post_name = %s
	2488	+ AND post_status IN ('publish', 'closed', 'private')
	2489	+ AND post_type NOT IN ('revision', 'attachment', 'nav_menu_item')
	2490	+ ORDER BY CASE
	2491	+ WHEN post_type = 'post' THEN 1
	2492	+ WHEN post_type = 'page' THEN 2
	2493	+ WHEN post_type = 'topic' THEN 3
	2494	+ WHEN post_type = 'product' THEN 4
	2495	+ ELSE 5
	2496	+ END
	2497	+ LIMIT 1",
	2498	+ $potential_slug
	2499	+ ));
	2500	+
	2501	+ if ($post_id) {
	2502	+ return intval($post_id);
	2503	+ }
	2504	+ }
	2505	+ }
	2506	+
	2507	+ // ADDITIONAL: Try direct database lookup by URL variations
	2508	+ global $wpdb;
	2509	+ $table_name = $wpdb->prefix . 'mxchat_system_prompt_content';
	2510	+
	2511	+ // Try variations of the URL (with/without trailing slash, http/https)
	2512	+ $url_variations = array(
	2513	+ $url,
	2514	+ rtrim($url, '/'),
	2515	+ $url . '/',
	2516	+ str_replace('http://', 'https://', $url),
	2517	+ str_replace('https://', 'http://', $url),
	2518	+ str_replace('http://', 'https://', rtrim($url, '/')),
	2519	+ str_replace('https://', 'http://', rtrim($url, '/'))
	2520	+ );
	2521	+
	2522	+ // Remove duplicates
	2523	+ $url_variations = array_unique($url_variations);
	2524	+
	2525	+ foreach ($url_variations as $variation) {
	2526	+ $existing_record = $wpdb->get_row($wpdb->prepare(
	2527	+ "SELECT id, source_url FROM $table_name WHERE source_url = %s",
	2528	+ $variation
	2529	+ ));
	2530	+
	2531	+ if ($existing_record) {
	2532	+ // Try to get post ID from this stored URL
	2533	+ $stored_post_id = url_to_postid($existing_record->source_url);
	2534	+ if ($stored_post_id > 0) {
	2535	+ return $stored_post_id;
	2536	+ }
	2537	+ }
	2538	+ }
	2539	+
	2540	+ return 0; // No match found
	2541	+}
	2542	+/**
	2543	+ * Process selected content via AJAX
	2544	+ */
	2545	+public function ajax_mxchat_process_selected_content() {
	2546	+ // Basic request validation
	2547	+ if (!check_ajax_referer('mxchat_content_selector_nonce', 'nonce', false)) {
	2548	+ wp_send_json_error('Invalid nonce');
	2549	+ exit;
	2550	+ }
	2551	+
	2552	+ if (!current_user_can('manage_options')) {
	2553	+ wp_send_json_error('Unauthorized access');
	2554	+ exit;
	2555	+ }
	2556	+
	2557	+ // Get post IDs - safely parse the array
	2558	+ $post_ids = array();
	2559	+ if (isset($_POST['post_ids']) && is_array($_POST['post_ids'])) {
	2560	+ foreach ($_POST['post_ids'] as $id) {
	2561	+ $post_ids[] = absint($id);
	2562	+ }
	2563	+ }
	2564	+
	2565	+ if (empty($post_ids)) {
	2566	+ wp_send_json_error('No content selected');
	2567	+ exit;
	2568	+ }
	2569	+
	2570	+ // UPDATED: Get bot_id from request
	2571	+ $bot_id = isset($_POST['bot_id']) ? sanitize_key($_POST['bot_id']) : 'default';
	2572	+
	2573	+ // Process only ONE post at a time to avoid request size issues
	2574	+ $post_id = reset($post_ids);
	2575	+ $post = get_post($post_id);
	2576	+
	2577	+ if (!$post) {
	2578	+ wp_send_json_error('Post not found');
	2579	+ exit;
	2580	+ }
	2581	+
	2582	+ // Get content including ACF fields
	2583	+ $content = $post->post_title . "\n\n" . wp_strip_all_tags($post->post_content);
	2584	+
	2585	+ // ADD ACF FIELDS SUPPORT
	2586	+ $acf_fields = $this->mxchat_get_acf_fields_for_post($post_id);
	2587	+ if (!empty($acf_fields)) {
	2588	+ $acf_content_parts = array();
	2589	+
	2590	+ foreach ($acf_fields as $field_name => $field_value) {
	2591	+ $formatted_value = $this->mxchat_format_acf_field_value($field_value, $field_name, $post_id);
	2592	+
	2593	+ if (!empty($formatted_value)) {
	2594	+ $field_label = ucwords(str_replace('_', ' ', $field_name));
	2595	+ $acf_content_parts[] = $field_label . ": " . $formatted_value;
	2596	+ }
	2597	+ }
	2598	+
	2599	+ if (!empty($acf_content_parts)) {
	2600	+ $content .= "\n\n" . implode("\n", $acf_content_parts);
	2601	+ }
	2602	+ }
	2603	+
	2604	+ $content = substr($content, 0, 10000); // Limit content size
	2605	+
	2606	+ // UPDATED: Get bot-specific API key
	2607	+ $bot_options = $this->get_bot_options($bot_id);
	2608	+ $options = !empty($bot_options) ? $bot_options : get_option('mxchat_options');
	2609	+ $selected_model = $options['embedding_model'] ?? 'text-embedding-ada-002';
	2610	+
	2611	+ if (strpos($selected_model, 'voyage') === 0) {
	2612	+ $api_key = $options['voyage_api_key'] ?? '';
	2613	+ $provider_name = 'Voyage AI';
	2614	+ } elseif (strpos($selected_model, 'gemini-embedding') === 0) {
	2615	+ $api_key = $options['gemini_api_key'] ?? '';
	2616	+ $provider_name = 'Google Gemini';
	2617	+ } else {
	2618	+ $api_key = $options['api_key'] ?? '';
	2619	+ $provider_name = 'OpenAI';
	2620	+ }
	2621	+
	2622	+ if (empty($api_key)) {
	2623	+ wp_send_json_error($provider_name . ' API key not configured');
	2624	+ exit;
	2625	+ }
	2626	+
	2627	+ $source_url = get_permalink($post_id);
	2628	+ $vector_id = md5($source_url); // Vector ID for Pinecone
	2629	+
	2630	+ // UPDATED: Check for existing content in bot-specific storage
	2631	+ $is_update = false;
	2632	+
	2633	+ // Get bot-specific Pinecone configuration
	2634	+ $bot_pinecone_config = $this->get_bot_pinecone_config($bot_id);
	2635	+ $use_pinecone = !empty($bot_pinecone_config) && ($bot_pinecone_config['use_pinecone'] ?? false);
	2636	+
	2637	+ if ($use_pinecone && !empty($bot_pinecone_config['api_key'])) {
	2638	+ // Check Pinecone for this bot
	2639	+ $pinecone_data = $this->mxchat_get_pinecone_processed_content($bot_pinecone_config);
	2640	+ if (isset($pinecone_data[$post_id])) {
	2641	+ $is_update = true;
	2642	+ }
	2643	+ } else {
	2644	+ // Check WordPress DB (same as before since it's shared)
	2645	+ global $wpdb;
	2646	+ $table_name = $wpdb->prefix . 'mxchat_system_prompt_content';
	2647	+ $existing_record = $wpdb->get_row($wpdb->prepare(
	2648	+ "SELECT id FROM $table_name WHERE source_url = %s",
	2649	+ $source_url
	2650	+ ));
	2651	+
	2652	+ if ($existing_record) {
	2653	+ $is_update = true;
	2654	+ }
	2655	+ }
	2656	+
	2657	+ // UPDATED: Use the centralized utility function with bot_id
	2658	+ $result = MxChat_Utils::submit_content_to_db(
	2659	+ $content,
	2660	+ $source_url,
	2661	+ $api_key,
	2662	+ $vector_id,
	2663	+ $bot_id
	2664	+ );
	2665	+
	2666	+ if (is_wp_error($result)) {
	2667	+ wp_send_json_error('Storage failed: ' . $result->get_error_message());
	2668	+ exit;
	2669	+ }
	2670	+
	2671	+ $operation_type = $is_update ? 'update' : 'new';
	2672	+
	2673	+ // Count ACF fields for debugging
	2674	+ $acf_field_count = count($acf_fields);
	2675	+
	2676	+ // Success response with minimal data
	2677	+ wp_send_json_success(array(
	2678	+ 'message' => $operation_type === 'update' ? 'Content updated successfully' : 'Content processed successfully',
	2679	+ 'post_id' => $post_id,
	2680	+ 'title' => $post->post_title,
	2681	+ 'operation_type' => $operation_type,
	2682	+ 'vector_id' => $vector_id,
	2683	+ 'acf_fields_found' => $acf_field_count,
	2684	+ 'content_preview' => substr($content, 0, 100) . '...',
	2685	+ 'bot_id' => $bot_id
	2686	+ ));
	2687	+ exit;
	2688	+}
	2689	+
	2690	+public function mxchat_get_public_post_types() {
	2691	+ // Get all public post types
	2692	+ $post_types = get_post_types(array('public' => true), 'objects');
	2693	+ $post_type_options = array();
	2694	+
	2695	+ foreach ($post_types as $post_type) {
	2696	+ $post_type_options[$post_type->name] = $post_type->label;
	2697	+ }
	2698	+
	2699	+ // Also include common forum/community post types that might not be marked as public
	2700	+ $additional_types = array(
	2701	+ 'topic' => 'Forum Topics (bbPress)',
	2702	+ 'reply' => 'Forum Replies (bbPress)',
	2703	+ 'forum' => 'Forums (bbPress)',
	2704	+ 'wpforo_topic' => 'wpForo Topics',
	2705	+ 'wpforo_post' => 'wpForo Posts'
	2706	+ );
	2707	+
	2708	+ foreach ($additional_types as $type_name => $type_label) {
	2709	+ if (post_type_exists($type_name) && !isset($post_type_options[$type_name])) {
	2710	+ $post_type_options[$type_name] = $type_label;
	2711	+ }
	2712	+ }
	2713	+
	2714	+ return $post_type_options;
	2715	+}
	2716	+
	2717	+/**
	2718	+ * Retrieves processed content from Pinecone API
	2719	+ */
	2720	+public function mxchat_get_pinecone_processed_content($pinecone_options) {
	2721	+ $api_key = $pinecone_options['mxchat_pinecone_api_key'] ?? '';
	2722	+ $host = $pinecone_options['mxchat_pinecone_host'] ?? '';
	2723	+
	2724	+ if (empty($api_key) || empty($host)) {
	2725	+ return array();
	2726	+ }
	2727	+
	2728	+ $pinecone_data = array();
	2729	+
	2730	+ try {
	2731	+ // Always get fresh data from Pinecone
	2732	+ $pinecone_data = $this->mxchat_scan_pinecone_for_processed_content($pinecone_options);
	2733	+
	2734	+ // Method 2: Final fallback - try stats endpoint (if available)
	2735	+ if (empty($pinecone_data)) {
	2736	+ $stats_url = "https://{$host}/describe_index_stats";
	2737	+
	2738	+ $response = wp_remote_post($stats_url, array(
	2739	+ 'headers' => array(
	2740	+ 'Api-Key' => $api_key,
	2741	+ 'Content-Type' => 'application/json'
	2742	+ ),
	2743	+ 'body' => json_encode(array()),
	2744	+ 'timeout' => 30
	2745	+ ));
	2746	+
	2747	+ if (!is_wp_error($response) && wp_remote_retrieve_response_code($response) === 200) {
	2748	+ $body = wp_remote_retrieve_body($response);
	2749	+ $stats_data = json_decode($body, true);
	2750	+ }
	2751	+ }
	2752	+
	2753	+ } catch (Exception $e) {
	2754	+ // Log error but return fresh data only
	2755	+ }
	2756	+
	2757	+ return $pinecone_data;
	2758	+}
	2759	+public function mxchat_fetch_pinecone_vectors_by_ids($pinecone_options, $vector_ids) {
	2760	+ //error_log('=== DEBUG: Starting mxchat_fetch_pinecone_vectors_by_ids ===');
	2761	+
	2762	+ $api_key = $pinecone_options['mxchat_pinecone_api_key'] ?? '';
	2763	+ $host = $pinecone_options['mxchat_pinecone_host'] ?? '';
	2764	+
	2765	+ if (empty($api_key) || empty($host) || empty($vector_ids)) {
	2766	+ //error_log('DEBUG: Missing parameters for fetch by IDs');
	2767	+ return array();
	2768	+ }
	2769	+
	2770	+ try {
	2771	+ $fetch_url = "https://{$host}/vectors/fetch";
	2772	+ //error_log('DEBUG: Fetch URL: ' . $fetch_url);
	2773	+ //error_log('DEBUG: Fetching ' . count($vector_ids) . ' vector IDs');
	2774	+
	2775	+ // Pinecone fetch API allows fetching specific vectors by ID
	2776	+ $fetch_data = array(
	2777	+ 'ids' => array_values($vector_ids)
	2778	+ );
	2779	+
	2780	+ $response = wp_remote_post($fetch_url, array(
	2781	+ 'headers' => array(
	2782	+ 'Api-Key' => $api_key,
	2783	+ 'Content-Type' => 'application/json'
	2784	+ ),
	2785	+ 'body' => json_encode($fetch_data),
	2786	+ 'timeout' => 30
	2787	+ ));
	2788	+
	2789	+ if (is_wp_error($response)) {
	2790	+ //error_log('DEBUG: Fetch by IDs WP error: ' . $response->get_error_message());
	2791	+ return array();
	2792	+ }
	2793	+
	2794	+ $response_code = wp_remote_retrieve_response_code($response);
	2795	+ //error_log('DEBUG: Fetch response code: ' . $response_code);
	2796	+
	2797	+ if ($response_code !== 200) {
	2798	+ $error_body = wp_remote_retrieve_body($response);
	2799	+ //error_log('DEBUG: Fetch failed with body: ' . $error_body);
	2800	+ return array();
	2801	+ }
	2802	+
	2803	+ $body = wp_remote_retrieve_body($response);
	2804	+ $data = json_decode($body, true);
	2805	+
	2806	+ //error_log('DEBUG: Fetch response structure: ' . print_r(array_keys($data), true));
	2807	+
	2808	+ if (!isset($data['vectors'])) {
	2809	+ //error_log('DEBUG: No vectors key in response');
	2810	+ return array();
	2811	+ }
	2812	+
	2813	+ $processed_data = array();
	2814	+
	2815	+ foreach ($data['vectors'] as $vector_id => $vector_data) {
	2816	+ $metadata = $vector_data['metadata'] ?? array();
	2817	+ $source_url = $metadata['source_url'] ?? '';
	2818	+
	2819	+ if (!empty($source_url)) {
	2820	+ $post_id = url_to_postid($source_url);
	2821	+ if ($post_id) {
	2822	+ $created_at = $metadata['created_at'] ?? '';
	2823	+ $processed_date = 'Recently';
	2824	+
	2825	+ if (!empty($created_at)) {
	2826	+ $timestamp = is_numeric($created_at) ? $created_at : strtotime($created_at);
	2827	+ if ($timestamp) {
	2828	+ $processed_date = human_time_diff($timestamp, current_time('timestamp')) . ' ago';
	2829	+ }
	2830	+ }
	2831	+
	2832	+ $processed_data[$post_id] = array(
	2833	+ 'db_id' => $vector_id,
	2834	+ 'processed_date' => $processed_date,
	2835	+ 'url' => $source_url,
	2836	+ 'source' => 'pinecone',
	2837	+ 'timestamp' => $timestamp ?? current_time('timestamp')
	2838	+ );
	2839	+ }
	2840	+ }
	2841	+ }
	2842	+
	2843	+ //error_log('DEBUG: Processed ' . count($processed_data) . ' vectors from fetch');
	2844	+ return $processed_data;
	2845	+
	2846	+ } catch (Exception $e) {
	2847	+ //error_log('DEBUG: Exception in fetch_pinecone_vectors_by_ids: ' . $e->getMessage());
	2848	+ return array();
	2849	+ }
	2850	+}
	2851	+
	2852	+/**
	2853	+ * Scan Pinecone for processed content
	2854	+ */
	2855	+public function mxchat_scan_pinecone_for_processed_content($pinecone_options) {
	2856	+ $api_key = $pinecone_options['mxchat_pinecone_api_key'] ?? '';
	2857	+ $host = $pinecone_options['mxchat_pinecone_host'] ?? '';
	2858	+
	2859	+ if (empty($api_key) || empty($host)) {
	2860	+ return array();
	2861	+ }
	2862	+
	2863	+ try {
	2864	+ // Use multiple random vectors to get better coverage
	2865	+ $all_matches = array();
	2866	+ $seen_ids = array();
	2867	+
	2868	+ // Try 3 different random vectors to get better coverage
	2869	+ for ($i = 0; $i < 3; $i++) {
	2870	+ $query_url = "https://{$host}/query";
	2871	+
	2872	+ // Generate a random unit vector instead of zeros
	2873	+ $random_vector = array();
	2874	+ for ($j = 0; $j < 1536; $j++) {
	2875	+ $random_vector[] = (rand(-1000, 1000) / 1000.0);
	2876	+ }
	2877	+
	2878	+ // Normalize the vector to unit length
	2879	+ $magnitude = sqrt(array_sum(array_map(function($x) { return $x * $x; }, $random_vector)));
	2880	+ if ($magnitude > 0) {
	2881	+ $random_vector = array_map(function($x) use ($magnitude) { return $x / $magnitude; }, $random_vector);
	2882	+ }
	2883	+
	2884	+ $query_data = array(
	2885	+ 'includeMetadata' => true,
	2886	+ 'includeValues' => false,
	2887	+ 'topK' => 10000,
	2888	+ 'vector' => $random_vector
	2889	+ );
	2890	+
	2891	+ $response = wp_remote_post($query_url, array(
	2892	+ 'headers' => array(
	2893	+ 'Api-Key' => $api_key,
	2894	+ 'Content-Type' => 'application/json'
	2895	+ ),
	2896	+ 'body' => json_encode($query_data),
	2897	+ 'timeout' => 30
	2898	+ ));
	2899	+
	2900	+ if (is_wp_error($response)) {
	2901	+ continue;
	2902	+ }
	2903	+
	2904	+ $response_code = wp_remote_retrieve_response_code($response);
	2905	+
	2906	+ if ($response_code !== 200) {
	2907	+ continue;
	2908	+ }
	2909	+
	2910	+ $body = wp_remote_retrieve_body($response);
	2911	+ $data = json_decode($body, true);
	2912	+
	2913	+ if (isset($data['matches'])) {
	2914	+ foreach ($data['matches'] as $match) {
	2915	+ $match_id = $match['id'] ?? '';
	2916	+ if (!empty($match_id) && !isset($seen_ids[$match_id])) {
	2917	+ $all_matches[] = $match;
	2918	+ $seen_ids[$match_id] = true;
	2919	+ }
	2920	+ }
	2921	+ }
	2922	+ }
	2923	+
	2924	+ // Convert matches to processed data format
	2925	+ $processed_data = array();
	2926	+
	2927	+ foreach ($all_matches as $match) {
	2928	+ $metadata = $match['metadata'] ?? array();
	2929	+ $source_url = $metadata['source_url'] ?? '';
	2930	+ $match_id = $match['id'] ?? '';
	2931	+
	2932	+ if (!empty($source_url) && !empty($match_id)) {
	2933	+ $post_id = url_to_postid($source_url);
	2934	+ if ($post_id) {
	2935	+ $created_at = $metadata['created_at'] ?? '';
	2936	+ $processed_date = 'Recently';
	2937	+
	2938	+ if (!empty($created_at)) {
	2939	+ $timestamp = is_numeric($created_at) ? $created_at : strtotime($created_at);
	2940	+ if ($timestamp) {
	2941	+ $processed_date = human_time_diff($timestamp, current_time('timestamp')) . ' ago';
	2942	+ }
	2943	+ }
	2944	+
	2945	+ $processed_data[$post_id] = array(
	2946	+ 'db_id' => $match_id,
	2947	+ 'processed_date' => $processed_date,
	2948	+ 'url' => $source_url,
	2949	+ 'source' => 'pinecone',
	2950	+ 'timestamp' => $timestamp ?? current_time('timestamp')
	2951	+ );
	2952	+ }
	2953	+ }
	2954	+ }
	2955	+
	2956	+ return $processed_data;
	2957	+
	2958	+ } catch (Exception $e) {
	2959	+ return array();
	2960	+ }
	2961	+}
	2962	+/**
	2963	+ * UPDATED: Generate embeddings from input text for MXChat with bot support
	2964	+ */
	2965	+private function mxchat_generate_embedding($text, $bot_id = 'default') {
	2966	+ // Enable detailed logging for debugging
	2967	+ //error_log('[MXCHAT-EMBED] Starting embedding generation for bot: ' . $bot_id . '. Text length: ' . strlen($text) . ' bytes');
	2968	+ //error_log('[MXCHAT-EMBED] Text preview: ' . substr($text, 0, 100) . '...');
	2969	+
	2970	+ // UPDATED: Get bot-specific options
	2971	+ $bot_options = $this->get_bot_options($bot_id);
	2972	+ $options = !empty($bot_options) ? $bot_options : get_option('mxchat_options');
	2973	+
	2974	+ $selected_model = $options['embedding_model'] ?? 'text-embedding-ada-002';
	2975	+ //error_log('[MXCHAT-EMBED] Selected embedding model for bot ' . $bot_id . ': ' . $selected_model);
	2976	+
	2977	+ // Determine provider and endpoint
	2978	+ if (strpos($selected_model, 'voyage') === 0) {
	2979	+ $api_key = $options['voyage_api_key'] ?? '';
	2980	+ $endpoint = 'https://api.voyageai.com/v1/embeddings';
	2981	+ $provider_name = 'Voyage AI';
	2982	+ //error_log('[MXCHAT-EMBED] Using Voyage AI API for bot ' . $bot_id);
	2983	+ } elseif (strpos($selected_model, 'gemini-embedding') === 0) {
	2984	+ $api_key = $options['gemini_api_key'] ?? '';
	2985	+ $endpoint = 'https://generativelanguage.googleapis.com/v1beta/models/' . $selected_model . ':embedContent';
	2986	+ $provider_name = 'Google Gemini';
	2987	+ //error_log('[MXCHAT-EMBED] Using Google Gemini API for bot ' . $bot_id);
	2988	+ } else {
	2989	+ $api_key = $options['api_key'] ?? '';
	2990	+ $endpoint = 'https://api.openai.com/v1/embeddings';
	2991	+ $provider_name = 'OpenAI';
	2992	+ //error_log('[MXCHAT-EMBED] Using OpenAI API for bot ' . $bot_id);
	2993	+ }
	2994	+
	2995	+ //error_log('[MXCHAT-EMBED] Using endpoint: ' . $endpoint);
	2996	+
	2997	+ if (empty($api_key)) {
	2998	+ $error_message = sprintf('Missing %s API key for bot %s. Please configure your API key in the bot settings.', $provider_name, $bot_id);
	2999	+ //error_log('[MXCHAT-EMBED] Error: ' . $error_message);
	3000	+ return $error_message;
	3001	+ }
	3002	+
	3003	+ // Check if text is too long (for OpenAI, roughly estimate tokens as words/0.75)
	3004	+ $estimated_tokens = ceil(str_word_count($text) / 0.75);
	3005	+ //error_log('[MXCHAT-EMBED] Estimated token count: ~' . $estimated_tokens);
	3006	+
	3007	+ if ($estimated_tokens > 8000 && strpos($selected_model, 'voyage') === false && strpos($selected_model, 'gemini-embedding') === false) {
	3008	+ //error_log('[MXCHAT-EMBED] Warning: Text may exceed OpenAI token limits (8K for most models)');
	3009	+ // Consider truncating text here
	3010	+ }
	3011	+
	3012	+ // Prepare request body based on provider
	3013	+ if (strpos($selected_model, 'gemini-embedding') === 0) {
	3014	+ // Gemini API format
	3015	+ $request_body = array(
	3016	+ 'model' => 'models/' . $selected_model,
	3017	+ 'content' => array(
	3018	+ 'parts' => array(
	3019	+ array('text' => $text)
	3020	+ )
	3021	+ )
	3022	+ );
	3023	+
	3024	+ // Set output dimensionality to 1536 for consistency with other models
	3025	+ $request_body['outputDimensionality'] = 1536;
	3026	+ } else {
	3027	+ // OpenAI/Voyage API format
	3028	+ $request_body = array(
	3029	+ 'model' => $selected_model,
	3030	+ 'input' => $text
	3031	+ );
	3032	+
	3033	+ // Add output_dimension for voyage-3-large model
	3034	+ if ($selected_model === 'voyage-3-large') {
	3035	+ $request_body['output_dimension'] = 2048;
	3036	+ }
	3037	+ }
	3038	+
	3039	+ //error_log('[MXCHAT-EMBED] Request prepared with model: ' . $selected_model);
	3040	+
	3041	+ // Prepare headers based on provider
	3042	+ if (strpos($selected_model, 'gemini-embedding') === 0) {
	3043	+ // Gemini uses API key as query parameter
	3044	+ $endpoint .= '?key=' . $api_key;
	3045	+ $headers = array(
	3046	+ 'Content-Type' => 'application/json'
	3047	+ );
	3048	+ } else {
	3049	+ // OpenAI/Voyage use Bearer token
	3050	+ $headers = array(
	3051	+ 'Authorization' => 'Bearer ' . $api_key,
	3052	+ 'Content-Type' => 'application/json'
	3053	+ );
	3054	+ }
	3055	+
	3056	+ // Make API request
	3057	+ //error_log('[MXCHAT-EMBED] Sending API request to: ' . $endpoint);
	3058	+ $response = wp_remote_post($endpoint, array(
	3059	+ 'body' => wp_json_encode($request_body),
	3060	+ 'headers' => $headers,
	3061	+ 'timeout' => 60 // Increased timeout for large inputs
	3062	+ ));
	3063	+
	3064	+ // Handle wp_remote_post errors
	3065	+ if (is_wp_error($response)) {
	3066	+ $error_message = $response->get_error_message();
	3067	+ //error_log('[MXCHAT-EMBED] WP Remote Post Error: ' . $error_message);
	3068	+ return 'Connection error: ' . $error_message;
	3069	+ }
	3070	+
	3071	+ // Get and check HTTP response code
	3072	+ $http_code = wp_remote_retrieve_response_code($response);
	3073	+ //error_log('[MXCHAT-EMBED] API Response Code: ' . $http_code);
	3074	+
	3075	+ if ($http_code !== 200) {
	3076	+ $error_body = wp_remote_retrieve_body($response);
	3077	+ //error_log('[MXCHAT-EMBED] API Error Response Body: ' . $error_body);
	3078	+
	3079	+ // Try to parse error for more details
	3080	+ $error_json = json_decode($error_body, true);
	3081	+ if (json_last_error() === JSON_ERROR_NONE && isset($error_json['error'])) {
	3082	+ $error_type = $error_json['error']['type'] ?? 'unknown';
	3083	+ $error_message = $error_json['error']['message'] ?? 'No message';
	3084	+ //error_log('[MXCHAT-EMBED] API Error Type: ' . $error_type);
	3085	+ //error_log('[MXCHAT-EMBED] API Error Message: ' . $error_message);
	3086	+
	3087	+ // Customize error message for common API errors
	3088	+ if ($error_type === 'invalid_request_error' && strpos($error_message, 'API key') !== false) {
	3089	+ $error_message = sprintf('Invalid %s API key for bot %s. Please check your API key in the bot settings.', $provider_name, $bot_id);
	3090	+ } elseif ($error_type === 'authentication_error') {
	3091	+ $error_message = sprintf('%s authentication failed for bot %s. Please verify your API key in the bot settings.', $provider_name, $bot_id);
	3092	+ }
	3093	+
	3094	+ //error_log('[MXCHAT-EMBED] Returning error: ' . $error_message);
	3095	+ return $error_message;
	3096	+ }
	3097	+
	3098	+ $error_message = sprintf("API Error (HTTP %d): Unable to generate embedding for bot %s", $http_code, $bot_id);
	3099	+ //error_log('[MXCHAT-EMBED] Returning error: ' . $error_message);
	3100	+ return $error_message;
	3101	+ }
	3102	+
	3103	+ // Parse response body
	3104	+ $response_body = wp_remote_retrieve_body($response);
	3105	+ //error_log('[MXCHAT-EMBED] Received response length: ' . strlen($response_body) . ' bytes');
	3106	+
	3107	+ $response_data = json_decode($response_body, true);
	3108	+
	3109	+ if (json_last_error() !== JSON_ERROR_NONE) {
	3110	+ $error = json_last_error_msg();
	3111	+ //error_log('[MXCHAT-EMBED] JSON Parse Error: ' . $error);
	3112	+ //error_log('[MXCHAT-EMBED] Response preview: ' . substr($response_body, 0, 200));
	3113	+ return "Failed to parse API response: $error";
	3114	+ }
	3115	+
	3116	+ // Handle different response formats based on provider
	3117	+ if (strpos($selected_model, 'gemini-embedding') === 0) {
	3118	+ // Gemini API response format
	3119	+ if (isset($response_data['embedding']['values'])) {
	3120	+ $embedding_dimensions = count($response_data['embedding']['values']);
	3121	+ //error_log('[MXCHAT-EMBED] Successfully extracted Gemini embedding with ' . $embedding_dimensions . ' dimensions');
	3122	+
	3123	+ // Check if embedding dimensions are as expected (should be 1536)
	3124	+ if ($embedding_dimensions !== 1536) {
	3125	+ //error_log('[MXCHAT-EMBED] Warning: Unexpected Gemini embedding dimensions: ' . $embedding_dimensions);
	3126	+ }
	3127	+
	3128	+ return $response_data['embedding']['values'];
	3129	+ } else {
	3130	+ //error_log('[MXCHAT-EMBED] Error: No embedding found in Gemini response');
	3131	+ //error_log('[MXCHAT-EMBED] Response structure: ' . wp_json_encode(array_keys($response_data)));
	3132	+
	3133	+ if (isset($response_data['error'])) {
	3134	+ $error_message = "Gemini API Error in response: " . wp_json_encode($response_data['error']);
	3135	+ //error_log('[MXCHAT-EMBED] ' . $error_message);
	3136	+ return $error_message;
	3137	+ }
	3138	+
	3139	+ $error_message = "Invalid Gemini API response format: No embedding found";
	3140	+ //error_log('[MXCHAT-EMBED] ' . $error_message);
	3141	+ return $error_message;
	3142	+ }
	3143	+ } else {
	3144	+ // OpenAI/Voyage API response format
	3145	+ if (isset($response_data['data'][0]['embedding'])) {
	3146	+ $embedding_dimensions = count($response_data['data'][0]['embedding']);
	3147	+ //error_log('[MXCHAT-EMBED] Successfully extracted embedding with ' . $embedding_dimensions . ' dimensions');
	3148	+
	3149	+ // Check if embedding dimensions are as expected
	3150	+ if (($selected_model === 'text-embedding-ada-002' && $embedding_dimensions !== 1536) ||
	3151	+ ($selected_model === 'voyage-3-large' && $embedding_dimensions !== 2048)) {
	3152	+ //error_log('[MXCHAT-EMBED] Warning: Unexpected embedding dimensions');
	3153	+ }
	3154	+
	3155	+ return $response_data['data'][0]['embedding'];
	3156	+ } else {
	3157	+ //error_log('[MXCHAT-EMBED] Error: No embedding found in response');
	3158	+ //error_log('[MXCHAT-EMBED] Response structure: ' . wp_json_encode(array_keys($response_data)));
	3159	+
	3160	+ if (isset($response_data['error'])) {
	3161	+ $error_message = "API Error in response: " . wp_json_encode($response_data['error']);
	3162	+ //error_log('[MXCHAT-EMBED] ' . $error_message);
	3163	+ return $error_message;
	3164	+ }
	3165	+
	3166	+ $error_message = "Invalid API response format: No embedding found";
	3167	+ //error_log('[MXCHAT-EMBED] ' . $error_message);
	3168	+ return $error_message;
	3169	+ }
	3170	+ }
	3171	+}
	3172	+
	3173	+/**
	3174	+ * Get bot-specific options for multi-bot functionality
	3175	+ * Falls back to default options if bot_id is 'default' or multi-bot add-on is not active
	3176	+ */
	3177	+private function get_bot_options($bot_id = 'default') {
	3178	+ //error_log("MXCHAT DEBUG: get_bot_options called for bot: " . $bot_id);
	3179	+
	3180	+ if ($bot_id === 'default' || !class_exists('MxChat_Multi_Bot_Manager')) {
	3181	+ //error_log("MXCHAT DEBUG: Using default options (no multi-bot or bot is 'default')");
	3182	+ return array();
	3183	+ }
	3184	+
	3185	+ $bot_options = apply_filters('mxchat_get_bot_options', array(), $bot_id);
	3186	+
	3187	+ if (!empty($bot_options)) {
	3188	+ //error_log("MXCHAT DEBUG: Got bot-specific options from filter");
	3189	+ if (isset($bot_options['similarity_threshold'])) {
	3190	+ //error_log(" - similarity_threshold: " . $bot_options['similarity_threshold']);
	3191	+ }
	3192	+ }
	3193	+
	3194	+ return is_array($bot_options) ? $bot_options : array();
	3195	+}
	3196	+
	3197	+/**
	3198	+ * Get bot-specific Pinecone configuration
	3199	+ * Used in the knowledge retrieval functions
	3200	+ */
	3201	+// Also add debugging to your get_bot_pinecone_config function
	3202	+private function get_bot_pinecone_config($bot_id = 'default') {
	3203	+ //error_log("MXCHAT DEBUG: get_bot_pinecone_config called for bot: " . $bot_id);
	3204	+
	3205	+ // If default bot or multi-bot add-on not active, use default Pinecone config
	3206	+ if ($bot_id === 'default' || !class_exists('MxChat_Multi_Bot_Manager')) {
	3207	+ //error_log("MXCHAT DEBUG: Using default Pinecone config (no multi-bot or bot is 'default')");
	3208	+ $addon_options = get_option('mxchat_pinecone_addon_options', array());
	3209	+ $config = array(
	3210	+ 'use_pinecone' => (isset($addon_options['mxchat_use_pinecone']) && $addon_options['mxchat_use_pinecone'] === '1'),
	3211	+ 'api_key' => $addon_options['mxchat_pinecone_api_key'] ?? '',
	3212	+ 'host' => $addon_options['mxchat_pinecone_host'] ?? '',
	3213	+ 'namespace' => $addon_options['mxchat_pinecone_namespace'] ?? ''
	3214	+ );
	3215	+ //error_log("MXCHAT DEBUG: Default config - use_pinecone: " . ($config['use_pinecone'] ? 'true' : 'false'));
	3216	+ return $config;
	3217	+ }
	3218	+
	3219	+ //error_log("MXCHAT DEBUG: Calling filter 'mxchat_get_bot_pinecone_config' for bot: " . $bot_id);
	3220	+
	3221	+ // Hook for multi-bot add-on to provide bot-specific Pinecone config
	3222	+ $bot_pinecone_config = apply_filters('mxchat_get_bot_pinecone_config', array(), $bot_id);
	3223	+
	3224	+ if (!empty($bot_pinecone_config)) {
	3225	+ //error_log("MXCHAT DEBUG: Got bot-specific config from filter");
	3226	+ //error_log(" - use_pinecone: " . (isset($bot_pinecone_config['use_pinecone']) ? ($bot_pinecone_config['use_pinecone'] ? 'true' : 'false') : 'not set'));
	3227	+ //error_log(" - host: " . ($bot_pinecone_config['host'] ?? 'not set'));
	3228	+ //error_log(" - namespace: " . ($bot_pinecone_config['namespace'] ?? 'not set'));
	3229	+ } else {
	3230	+ //error_log("MXCHAT DEBUG: Filter returned empty config!");
	3231	+ }
	3232	+
	3233	+ return is_array($bot_pinecone_config) ? $bot_pinecone_config : array();
	3234	+}
	3235	+
	3236	+
	3237	+public function mxchat_ajax_dismiss_completed_status() {
	3238	+ try {
	3239	+ // Verify the request
	3240	+ check_ajax_referer('mxchat_status_nonce', 'nonce');
	3241	+
	3242	+ if (!current_user_can('manage_options')) {
	3243	+ wp_send_json_error('Unauthorized access');
	3244	+ exit;
	3245	+ }
	3246	+
	3247	+ $card_type = isset($_POST['card_type']) ? sanitize_text_field($_POST['card_type']) : '';
	3248	+
	3249	+ if ($card_type === 'pdf') {
	3250	+ // Clear PDF status
	3251	+ $pdf_url = get_transient('mxchat_last_pdf_url');
	3252	+ if ($pdf_url) {
	3253	+ delete_transient('mxchat_pdf_status_' . md5($pdf_url));
	3254	+ delete_transient('mxchat_last_pdf_url');
	3255	+ }
	3256	+ } elseif ($card_type === 'sitemap') {
	3257	+ // Clear sitemap status
	3258	+ $sitemap_url = get_transient('mxchat_last_sitemap_url');
	3259	+ if ($sitemap_url) {
	3260	+ delete_transient('mxchat_sitemap_status_' . md5($sitemap_url));
	3261	+ delete_transient('mxchat_last_sitemap_url');
	3262	+ }
	3263	+ }
	3264	+
	3265	+ wp_send_json_success(array('message' => 'Status dismissed successfully'));
	3266	+
	3267	+ } catch (Exception $e) {
	3268	+ wp_send_json_error(array('message' => 'Error dismissing status: ' . $e->getMessage()));
	3269	+ }
	3270	+}
	3271	+
	3272	+/**
	3273	+ * Render completed status cards on page load
	3274	+ * This ensures completed processing status persists through page refreshes
	3275	+ */
	3276	+public function mxchat_render_completed_status_cards() {
	3277	+ $output = '';
	3278	+
	3279	+ // Check for completed PDF status
	3280	+ $pdf_url = get_transient('mxchat_last_pdf_url');
	3281	+ if ($pdf_url) {
	3282	+ $pdf_status = $this->mxchat_get_pdf_processing_status($pdf_url);
	3283	+ if ($pdf_status && ($pdf_status['status'] === 'complete' || $pdf_status['status'] === 'error')) {
	3284	+ $output .= $this->mxchat_render_pdf_status_card($pdf_status, $pdf_url);
	3285	+ }
	3286	+ }
	3287	+
	3288	+ // Check for completed sitemap status
	3289	+ $sitemap_url = get_transient('mxchat_last_sitemap_url');
	3290	+ if ($sitemap_url) {
	3291	+ $sitemap_status = $this->mxchat_get_sitemap_processing_status($sitemap_url);
	3292	+ if ($sitemap_status && ($sitemap_status['status'] === 'complete' || $sitemap_status['status'] === 'error')) {
	3293	+ $output .= $this->mxchat_render_sitemap_status_card($sitemap_status, $sitemap_url);
	3294	+ }
	3295	+ }
	3296	+
	3297	+ return $output;
	3298	+}
	3299	+
	3300	+/**
	3301	+ * Render PDF status card HTML
	3302	+ */
	3303	+private function mxchat_render_pdf_status_card($status, $pdf_url) {
	3304	+ $html = '<div class="mxchat-status-card" data-card-type="pdf">';
	3305	+ $html .= '<div class="mxchat-status-header">';
	3306	+ $html .= '<h4>' . esc_html__('PDF Processing Status', 'mxchat') . '</h4>';
	3307	+
	3308	+ // Add dismiss button for completed status
	3309	+ if ($status['status'] === 'complete' || $status['status'] === 'error') {
	3310	+ $html .= '<button type="button" class="mxchat-dismiss-button">' . esc_html__('Dismiss', 'mxchat') . '</button>';
	3311	+ }
	3312	+
	3313	+ // Process Batch button for processing status
	3314	+ if ($status['status'] === 'processing') {
	3315	+ $html .= '<button type="button" class="mxchat-manual-batch-btn"
	3316	+ data-process-type="pdf"
	3317	+ data-url="' . esc_attr($pdf_url) . '">
	3318	+ ' . esc_html__('Process Batch', 'mxchat') . '</button>';
	3319	+ }
	3320	+
	3321	+ // Add status badges
	3322	+ if ($status['status'] === 'error') {
	3323	+ $html .= '<span class="mxchat-status-badge mxchat-status-failed">' . esc_html__('Error', 'mxchat') . '</span>';
	3324	+ } elseif ($status['status'] === 'complete') {
	3325	+ if ($status['failed_pages'] > 0) {
	3326	+ $html .= '<span class="mxchat-status-badge mxchat-status-warning">' .
	3327	+ sprintf(esc_html__('Completed with %d failures', 'mxchat'), $status['failed_pages']) . '</span>';
	3328	+ } else {
	3329	+ $html .= '<span class="mxchat-status-badge mxchat-status-success">' . esc_html__('Complete', 'mxchat') . '</span>';
	3330	+ }
	3331	+ }
	3332	+
	3333	+ $html .= '</div>'; // End header
	3334	+
	3335	+ // Progress bar
	3336	+ $html .= '<div class="mxchat-progress-bar">';
	3337	+ $html .= '<div class="mxchat-progress-fill" style="width: ' . esc_attr($status['percentage']) . '%"></div>';
	3338	+ $html .= '</div>';
	3339	+
	3340	+ // Status details
	3341	+ $html .= '<div class="mxchat-status-details">';
	3342	+ $html .= '<p>' . sprintf(
	3343	+ esc_html__('Progress: %d of %d pages (%d%%)', 'mxchat'),
	3344	+ $status['processed_pages'],
	3345	+ $status['total_pages'],
	3346	+ $status['percentage']
	3347	+ ) . '</p>';
	3348	+
	3349	+ // Show failed pages count if any
	3350	+ if ($status['failed_pages'] > 0) {
	3351	+ $html .= '<p><strong>' . esc_html__('Failed pages:', 'mxchat') . '</strong> ' . esc_html($status['failed_pages']) . '</p>';
	3352	+ }
	3353	+
	3354	+ $html .= '<p><strong>' . esc_html__('Status:', 'mxchat') . '</strong> ' . esc_html(ucfirst($status['status'])) . '</p>';
	3355	+ $html .= '<p><strong>' . esc_html__('Last update:', 'mxchat') . '</strong> ' . esc_html($status['last_update']) . '</p>';
	3356	+
	3357	+ // Add completion summary if available AND it's an array
	3358	+ if (isset($status['completion_summary']) && is_array($status['completion_summary']) && !empty($status['completion_summary'])) {
	3359	+ $summary = $status['completion_summary'];
	3360	+ $html .= '<div class="mxchat-completion-summary">';
	3361	+ $html .= '<h5>' . esc_html__('Processing Summary', 'mxchat') . '</h5>';
	3362	+ $html .= '<p><strong>' . esc_html__('Total Pages:', 'mxchat') . '</strong> ' . esc_html($summary['total_pages']) . '</p>';
	3363	+ $html .= '<p><strong>' . esc_html__('Successful:', 'mxchat') . '</strong> ' . esc_html($summary['successful_pages']) . '</p>';
	3364	+ $html .= '<p><strong>' . esc_html__('Failed:', 'mxchat') . '</strong> ' . esc_html($summary['failed_pages']) . '</p>';
	3365	+ $html .= '<p><strong>' . esc_html__('Completed:', 'mxchat') . '</strong> ' . esc_html($summary['completion_time']) . '</p>';
	3366	+ $html .= '</div>';
	3367	+ }
	3368	+
	3369	+ // Add failed pages list if any AND it's an array
	3370	+ if (isset($status['failed_pages_list']) && is_array($status['failed_pages_list']) && !empty($status['failed_pages_list'])) {
	3371	+ $html .= $this->mxchat_render_failed_pages_list($status['failed_pages_list']);
	3372	+ }
	3373	+
	3374	+ // Add error message if any
	3375	+ if (isset($status['error']) && !empty($status['error'])) {
	3376	+ $html .= '<div class="mxchat-error-notice">';
	3377	+ $html .= '<p class="error">' . esc_html($status['error']) . '</p>';
	3378	+ $html .= '</div>';
	3379	+ }
	3380	+
	3381	+ $html .= '</div>'; // End details
	3382	+ $html .= '</div>'; // End card
	3383	+
	3384	+ return $html;
	3385	+}
	3386	+/**
	3387	+ * Render sitemap status card HTML
	3388	+ */
	3389	+private function mxchat_render_sitemap_status_card($status, $sitemap_url) {
	3390	+ $html = '<div class="mxchat-status-card" data-card-type="sitemap">';
	3391	+ $html .= '<div class="mxchat-status-header">';
	3392	+ $html .= '<h4>' . esc_html__('Sitemap Processing Status', 'mxchat') . '</h4>';
	3393	+
	3394	+ // Add dismiss button for completed status
	3395	+ if ($status['status'] === 'complete' || $status['status'] === 'error') {
	3396	+ $html .= '<button type="button" class="mxchat-dismiss-button">' . esc_html__('Dismiss', 'mxchat') . '</button>';
	3397	+ }
	3398	+
	3399	+ // Process Batch button for processing status
	3400	+ if ($status['status'] === 'processing') {
	3401	+ $html .= '<button type="button" class="mxchat-manual-batch-btn"
	3402	+ data-process-type="sitemap"
	3403	+ data-url="' . esc_attr($sitemap_url) . '">
	3404	+ ' . esc_html__('Process Batch', 'mxchat') . '</button>';
	3405	+ }
	3406	+
	3407	+ // Add status badges
	3408	+ if ($status['status'] === 'error') {
	3409	+ $html .= '<span class="mxchat-status-badge mxchat-status-failed">' . esc_html__('Error', 'mxchat') . '</span>';
	3410	+ } elseif ($status['status'] === 'complete') {
	3411	+ if ($status['failed_urls'] > 0) {
	3412	+ $html .= '<span class="mxchat-status-badge mxchat-status-warning">' .
	3413	+ sprintf(esc_html__('Completed with %d failures', 'mxchat'), $status['failed_urls']) . '</span>';
	3414	+ } else {
	3415	+ $html .= '<span class="mxchat-status-badge mxchat-status-success">' . esc_html__('Complete', 'mxchat') . '</span>';
	3416	+ }
	3417	+ }
	3418	+
	3419	+ $html .= '</div>'; // End header
	3420	+
	3421	+ // Progress bar
	3422	+ $html .= '<div class="mxchat-progress-bar">';
	3423	+ $html .= '<div class="mxchat-progress-fill" style="width: ' . esc_attr($status['percentage']) . '%"></div>';
	3424	+ $html .= '</div>';
	3425	+
	3426	+ // Status details
	3427	+ $html .= '<div class="mxchat-status-details">';
	3428	+ $html .= '<p>' . sprintf(
	3429	+ esc_html__('Progress: %d of %d URLs (%d%%)', 'mxchat'),
	3430	+ $status['processed_urls'],
	3431	+ $status['total_urls'],
	3432	+ $status['percentage']
	3433	+ ) . '</p>';
	3434	+
	3435	+ // Show failed URLs count if any
	3436	+ if ($status['failed_urls'] > 0) {
	3437	+ $html .= '<p><strong>' . esc_html__('Failed URLs:', 'mxchat') . '</strong> ' . esc_html($status['failed_urls']) . '</p>';
	3438	+ }
	3439	+
	3440	+ $html .= '<p><strong>' . esc_html__('Status:', 'mxchat') . '</strong> ' . esc_html(ucfirst($status['status'])) . '</p>';
	3441	+ $html .= '<p><strong>' . esc_html__('Last update:', 'mxchat') . '</strong> ' . esc_html($status['last_update']) . '</p>';
	3442	+
	3443	+ // Add completion summary if available AND it's an array
	3444	+ if (isset($status['completion_summary']) && is_array($status['completion_summary']) && !empty($status['completion_summary'])) {
	3445	+ $summary = $status['completion_summary'];
	3446	+ $html .= '<div class="mxchat-completion-summary">';
	3447	+ $html .= '<h5>' . esc_html__('Processing Summary', 'mxchat') . '</h5>';
	3448	+ $html .= '<p><strong>' . esc_html__('Total URLs:', 'mxchat') . '</strong> ' . esc_html($summary['total_urls']) . '</p>';
	3449	+ $html .= '<p><strong>' . esc_html__('Successful:', 'mxchat') . '</strong> ' . esc_html($summary['successful_urls']) . '</p>';
	3450	+ $html .= '<p><strong>' . esc_html__('Failed:', 'mxchat') . '</strong> ' . esc_html($summary['failed_urls']) . '</p>';
	3451	+ $html .= '<p><strong>' . esc_html__('Completed:', 'mxchat') . '</strong> ' . esc_html($summary['completion_time']) . '</p>';
	3452	+ $html .= '</div>';
	3453	+ }
	3454	+
	3455	+ // Add error messages if any (but not the failed URLs list)
	3456	+ if (!empty($status['error']) || !empty($status['last_error'])) {
	3457	+ $html .= '<div class="mxchat-error-notice">';
	3458	+
	3459	+ if (!empty($status['error'])) {
	3460	+ $html .= '<p class="error">' . esc_html($status['error']) . '</p>';
	3461	+ }
	3462	+
	3463	+ if (!empty($status['last_error'])) {
	3464	+ $html .= '<p class="last-error">' . esc_html__('Last error:', 'mxchat') . ' ' . esc_html($status['last_error']) . '</p>';
	3465	+ }
	3466	+
	3467	+ $html .= '</div>';
	3468	+ }
	3469	+
	3470	+ $html .= '</div>'; // End details
	3471	+ $html .= '</div>'; // End card
	3472	+
	3473	+ return $html;
	3474	+}
	3475	+
	3476	+
	3477	+/**
	3478	+ * Render failed pages list
	3479	+ */
	3480	+private function mxchat_render_failed_pages_list($failed_pages_list) {
	3481	+ // Validate that $failed_pages_list is an array and not empty
	3482	+ if (!is_array($failed_pages_list) || empty($failed_pages_list)) {
	3483	+ return '';
	3484	+ }
	3485	+
	3486	+ $html = '<div class="mxchat-error-notice">';
	3487	+ $html .= '<div class="mxchat-failed-pages-container">';
	3488	+ $html .= '<h5>' . sprintf(esc_html__('Failed Pages (%d)', 'mxchat'), count($failed_pages_list)) . '</h5>';
	3489	+ $html .= '<details>';
	3490	+ $html .= '<summary>' . esc_html__('Show Failed Pages', 'mxchat') . '</summary>';
	3491	+ $html .= '<div class="mxchat-failed-pages-list">';
	3492	+
	3493	+ // Create table for failed pages
	3494	+ $html .= '<table class="widefat striped">';
	3495	+ $html .= '<thead><tr>';
	3496	+ $html .= '<th>' . esc_html__('Page', 'mxchat') . '</th>';
	3497	+ $html .= '<th>' . esc_html__('Error', 'mxchat') . '</th>';
	3498	+ $html .= '<th>' . esc_html__('Retries', 'mxchat') . '</th>';
	3499	+ $html .= '<th>' . esc_html__('Time', 'mxchat') . '</th>';
	3500	+ $html .= '</tr></thead><tbody>';
	3501	+
	3502	+ // Sort failed pages by most recent
	3503	+ $sorted_failed_pages = $failed_pages_list;
	3504	+ usort($sorted_failed_pages, function($a, $b) {
	3505	+ return ($b['time'] ?? 0) - ($a['time'] ?? 0);
	3506	+ });
	3507	+
	3508	+ foreach ($sorted_failed_pages as $item) {
	3509	+ // Ensure $item is an array before accessing its elements
	3510	+ if (!is_array($item)) {
	3511	+ continue;
	3512	+ }
	3513	+
	3514	+ $time_ago = isset($item['time']) ? human_time_diff($item['time'], current_time('timestamp')) . ' ' . esc_html__('ago', 'mxchat') : 'Unknown';
	3515	+ $html .= '<tr>';
	3516	+ $html .= '<td>' . esc_html__('Page', 'mxchat') . ' ' . esc_html($item['page'] ?? 'Unknown') . '</td>';
	3517	+ $html .= '<td style="word-break: break-word;">' . esc_html($item['error'] ?? 'Unknown error') . '</td>';
	3518	+ $html .= '<td>' . esc_html($item['retries'] ?? 'N/A') . '</td>';
	3519	+ $html .= '<td>' . esc_html($time_ago) . '</td>';
	3520	+ $html .= '</tr>';
	3521	+ }
	3522	+
	3523	+ $html .= '</tbody></table>';
	3524	+ $html .= '</div></details></div></div>';
	3525	+
	3526	+ return $html;
	3527	+}
	3528	+
	3529	+/**
	3530	+ * Render failed URLs list
	3531	+ */
	3532	+private function mxchat_render_failed_urls_list($failed_urls_list) {
	3533	+ // Validate that $failed_urls_list is an array and not empty
	3534	+ if (!is_array($failed_urls_list) || empty($failed_urls_list)) {
	3535	+ return '';
	3536	+ }
	3537	+
	3538	+ $html = '<div class="mxchat-failed-urls-container">';
	3539	+ $html .= '<h5>' . sprintf(esc_html__('Failed URLs (%d)', 'mxchat'), count($failed_urls_list)) . '</h5>';
	3540	+ $html .= '<details>';
	3541	+ $html .= '<summary>' . esc_html__('Show Failed URLs', 'mxchat') . '</summary>';
	3542	+ $html .= '<div class="mxchat-failed-urls-list">';
	3543	+
	3544	+ // Create table for failed URLs
	3545	+ $html .= '<table class="widefat striped">';
	3546	+ $html .= '<thead><tr>';
	3547	+ $html .= '<th>' . esc_html__('URL', 'mxchat') . '</th>';
	3548	+ $html .= '<th>' . esc_html__('Error', 'mxchat') . '</th>';
	3549	+ $html .= '<th>' . esc_html__('Retries', 'mxchat') . '</th>';
	3550	+ $html .= '<th>' . esc_html__('Time', 'mxchat') . '</th>';
	3551	+ $html .= '</tr></thead><tbody>';
	3552	+
	3553	+ // Sort failed URLs by most recent
	3554	+ $sorted_failed_urls = $failed_urls_list;
	3555	+ usort($sorted_failed_urls, function($a, $b) {
	3556	+ return ($b['time'] ?? 0) - ($a['time'] ?? 0);
	3557	+ });
	3558	+
	3559	+ // Show up to 50 failed URLs
	3560	+ $display_urls = array_slice($sorted_failed_urls, 0, 50);
	3561	+
	3562	+ foreach ($display_urls as $item) {
	3563	+ // Ensure $item is an array before accessing its elements
	3564	+ if (!is_array($item)) {
	3565	+ continue;
	3566	+ }
	3567	+
	3568	+ $url = $item['url'] ?? '';
	3569	+ $time_ago = isset($item['time']) ? human_time_diff($item['time'], current_time('timestamp')) . ' ' . esc_html__('ago', 'mxchat') : 'Unknown';
	3570	+
	3571	+ // Truncate URL for display
	3572	+ $display_url = strlen($url) > 50 ? substr($url, 0, 47) . '...' : $url;
	3573	+
	3574	+ $html .= '<tr>';
	3575	+ $html .= '<td style="word-break: break-all;">';
	3576	+ if (!empty($url)) {
	3577	+ $html .= '<a href="' . esc_url($url) . '" target="_blank" rel="noopener noreferrer">' . esc_html($display_url) . '</a>';
	3578	+ } else {
	3579	+ $html .= esc_html__('Unknown URL', 'mxchat');
	3580	+ }
	3581	+ $html .= '</td>';
	3582	+ $html .= '<td style="word-break: break-word;">' . esc_html($item['error'] ?? 'Unknown error') . '</td>';
	3583	+ $html .= '<td>' . esc_html($item['retries'] ?? 'N/A') . '</td>';
	3584	+ $html .= '<td>' . esc_html($time_ago) . '</td>';
	3585	+ $html .= '</tr>';
	3586	+ }
	3587	+
	3588	+ $html .= '</tbody></table>';
	3589	+
	3590	+ if (count($failed_urls_list) > 50) {
	3591	+ $html .= '<div class="mxchat-failed-urls-more">+ ' .
	3592	+ (count($failed_urls_list) - 50) .
	3593	+ ' ' . esc_html__('more failed URLs not shown', 'mxchat') . '</div>';
	3594	+ }
	3595	+
	3596	+ $html .= '</div></details></div>';
	3597	+
	3598	+ return $html;
	3599	+}
	3600	+
	3601	+/**
	3602	+ * Get all ACF fields for a specific post
	3603	+ */
	3604	+public function mxchat_get_acf_fields_for_post($post_id) {
	3605	+ if (!function_exists('get_fields')) {
	3606	+ return array();
	3607	+ }
	3608	+
	3609	+ $fields = get_fields($post_id);
	3610	+ if (!$fields || !is_array($fields)) {
	3611	+ return array();
	3612	+ }
	3613	+
	3614	+ return $fields;
	3615	+}
	3616	+
	3617	+/**
	3618	+ * Format ACF field values for content extraction
	3619	+ */
	3620	+public function mxchat_format_acf_field_value($value, $field_name = '', $post_id = 0) {
	3621	+ if (empty($value)) {
	3622	+ return '';
	3623	+ }
	3624	+
	3625	+ // Handle WP_Post objects first (THIS IS THE KEY FIX)
	3626	+ if ($value instanceof WP_Post) {
	3627	+ return $value->post_title ?: '';
	3628	+ }
	3629	+
	3630	+ // Handle other WP objects
	3631	+ if (is_object($value)) {
	3632	+ if (isset($value->post_title)) {
	3633	+ return $value->post_title;
	3634	+ } elseif (isset($value->display_name)) {
	3635	+ return $value->display_name;
	3636	+ } elseif (isset($value->name)) {
	3637	+ return $value->name;
	3638	+ } elseif (method_exists($value, '__toString')) {
	3639	+ try {
	3640	+ return (string) $value;
	3641	+ } catch (Exception $e) {
	3642	+ return '';
	3643	+ }
	3644	+ }
	3645	+ // For any other objects, return empty string
	3646	+ return '';
	3647	+ }
	3648	+
	3649	+ // Handle different ACF field types
	3650	+ if (is_array($value)) {
	3651	+ // Check if it's an image/file field
	3652	+ if (isset($value['url'])) {
	3653	+ // Image field - return alt text, title, or caption
	3654	+ if (!empty($value['alt'])) {
	3655	+ return $value['alt'];
	3656	+ } elseif (!empty($value['title'])) {
	3657	+ return $value['title'];
	3658	+ } elseif (!empty($value['caption'])) {
	3659	+ return $value['caption'];
	3660	+ } else {
	3661	+ return ''; // Don't include just the URL
	3662	+ }
	3663	+ }
	3664	+
	3665	+ // Check if it's a post object or relationship field
	3666	+ if (isset($value['post_title'])) {
	3667	+ return $value['post_title'];
	3668	+ }
	3669	+
	3670	+ // Check if it's a user field
	3671	+ if (isset($value['display_name'])) {
	3672	+ return $value['display_name'];
	3673	+ }
	3674	+
	3675	+ // Check if it's a taxonomy term
	3676	+ if (isset($value['name']) && isset($value['taxonomy'])) {
	3677	+ return $value['name'];
	3678	+ }
	3679	+
	3680	+ // Check if it's a select field with label
	3681	+ if (isset($value['label'])) {
	3682	+ return $value['label'];
	3683	+ }
	3684	+
	3685	+ // Check for repeater field or flexible content
	3686	+ if (is_numeric(key($value))) {
	3687	+ $sub_values = array();
	3688	+ foreach ($value as $sub_item) {
	3689	+ if (is_array($sub_item)) {
	3690	+ // For repeater/flexible content, extract text values
	3691	+ $sub_text = $this->mxchat_extract_text_from_acf_array($sub_item);
	3692	+ if (!empty($sub_text)) {
	3693	+ $sub_values[] = $sub_text;
	3694	+ }
	3695	+ } elseif ($sub_item instanceof WP_Post) {
	3696	+ // Handle WP_Post objects in arrays
	3697	+ $sub_values[] = $sub_item->post_title ?: '';
	3698	+ } else {
	3699	+ $sub_values[] = (string) $sub_item;
	3700	+ }
	3701	+ }
	3702	+ return implode(', ', array_filter($sub_values));
	3703	+ }
	3704	+
	3705	+ // For other arrays, try to extract meaningful text
	3706	+ $text_values = array();
	3707	+ foreach ($value as $key => $val) {
	3708	+ if (is_string($val) && !empty(trim($val))) {
	3709	+ $text_values[] = trim($val);
	3710	+ } elseif ($val instanceof WP_Post) {
	3711	+ // Handle WP_Post objects in associative arrays
	3712	+ $text_values[] = $val->post_title ?: '';
	3713	+ } elseif (is_array($val) && isset($val['post_title'])) {
	3714	+ $text_values[] = $val['post_title'];
	3715	+ } elseif (is_array($val) && isset($val['name'])) {
	3716	+ $text_values[] = $val['name'];
	3717	+ }
	3718	+ }
	3719	+
	3720	+ return implode(', ', array_filter($text_values));
	3721	+ }
	3722	+
	3723	+ // Handle boolean values
	3724	+ if (is_bool($value)) {
	3725	+ return $value ? 'Yes' : 'No';
	3726	+ }
	3727	+
	3728	+ // Handle numeric values
	3729	+ if (is_numeric($value)) {
	3730	+ return (string) $value;
	3731	+ }
	3732	+
	3733	+ // Handle string values
	3734	+ if (is_string($value)) {
	3735	+ return trim($value);
	3736	+ }
	3737	+
	3738	+ // For anything else that we can't handle, return empty string
	3739	+ // This prevents the "Object could not be converted to string" error
	3740	+ return '';
	3741	+}
	3742	+
	3743	+/**
	3744	+ * Extract text from complex ACF array structures
	3745	+ */
	3746	+private function mxchat_extract_text_from_acf_array($array) {
	3747	+ if (!is_array($array)) {
	3748	+ return '';
	3749	+ }
	3750	+
	3751	+ $text_parts = array();
	3752	+
	3753	+ foreach ($array as $key => $value) {
	3754	+ if (is_string($value) && !empty(trim($value))) {
	3755	+ // Skip keys that are likely to be IDs or technical values
	3756	+ if (!is_numeric($value) || strlen($value) > 10) {
	3757	+ $text_parts[] = trim($value);
	3758	+ }
	3759	+ } elseif ($value instanceof WP_Post) {
	3760	+ // Handle WP_Post objects
	3761	+ $text_parts[] = $value->post_title ?: '';
	3762	+ } elseif (is_array($value)) {
	3763	+ if (isset($value['post_title'])) {
	3764	+ $text_parts[] = $value['post_title'];
	3765	+ } elseif (isset($value['name'])) {
	3766	+ $text_parts[] = $value['name'];
	3767	+ } elseif (isset($value['label'])) {
	3768	+ $text_parts[] = $value['label'];
	3769	+ }
	3770	+ } elseif (is_object($value)) {
	3771	+ // Handle other objects safely
	3772	+ if (isset($value->post_title)) {
	3773	+ $text_parts[] = $value->post_title;
	3774	+ } elseif (isset($value->name)) {
	3775	+ $text_parts[] = $value->name;
	3776	+ } elseif (isset($value->display_name)) {
	3777	+ $text_parts[] = $value->display_name;
	3778	+ }
	3779	+ }
	3780	+ }
	3781	+
	3782	+ return implode(', ', array_filter($text_parts));
	3783	+}
	3784	+
	3785	+public function mxchat_handle_post_update($post_id, $post, $update) {
	3786	+ // Basic validation checks
	3787	+ if (defined('DOING_AUTOSAVE') && DOING_AUTOSAVE || wp_is_post_revision($post_id)) {
	3788	+ return;
	3789	+ }
	3790	+
	3791	+ $post_type = $post->post_type;
	3792	+
	3793	+ // Check if sync is enabled for this post type
	3794	+ $should_sync = false;
	3795	+
	3796	+ // Check built-in post types first
	3797	+ if ($post_type === 'post' && get_option('mxchat_auto_sync_posts') === '1') {
	3798	+ $should_sync = true;
	3799	+ } else if ($post_type === 'page' && get_option('mxchat_auto_sync_pages') === '1') {
	3800	+ $should_sync = true;
	3801	+ } else {
	3802	+ // Check custom post types
	3803	+ $option_name = 'mxchat_auto_sync_' . $post_type;
	3804	+ if (get_option($option_name) === '1') {
	3805	+ $should_sync = true;
	3806	+ }
	3807	+ }
	3808	+
	3809	+ if (!$should_sync) {
	3810	+ return;
	3811	+ }
	3812	+
	3813	+ // Check if we have stored the previous status and URL in our transients
	3814	+ $previous_status_key = 'mxchat_prev_status_' . $post_id;
	3815	+ $previous_status = get_transient($previous_status_key);
	3816	+
	3817	+ $previous_url_key = 'mxchat_prev_url_' . $post_id;
	3818	+ $previous_url = get_transient($previous_url_key);
	3819	+
	3820	+ // If the post was previously published but is now not published, remove from knowledge base
	3821	+ if ($previous_status === 'publish' && $post->post_status !== 'publish') {
	3822	+ // Use the stored URL from when it was published, or fall back to current permalink
	3823	+ $source_url = $previous_url ?: get_permalink($post_id);
	3824	+
	3825	+ if ($source_url) {
	3826	+ // Check if Pinecone is enabled
	3827	+ $pinecone_options = get_option('mxchat_pinecone_addon_options', array());
	3828	+ $use_pinecone = ($pinecone_options['mxchat_use_pinecone'] ?? '0') === '1';
	3829	+
	3830	+ if ($use_pinecone && !empty($pinecone_options['mxchat_pinecone_api_key'])) {
	3831	+ // Delete from Pinecone
	3832	+ $this->mxchat_delete_from_pinecone_by_url($source_url, $pinecone_options);
	3833	+ } else {
	3834	+ // Delete from WordPress DB
	3835	+ global $wpdb;
	3836	+ $table_name = $wpdb->prefix . 'mxchat_system_prompt_content';
	3837	+
	3838	+ $result = $wpdb->delete(
	3839	+ $table_name,
	3840	+ array('source_url' => $source_url),
	3841	+ array('%s')
	3842	+ );
	3843	+ }
	3844	+ }
	3845	+
	3846	+ // Clean up the transients and exit early
	3847	+ delete_transient($previous_status_key);
	3848	+ delete_transient($previous_url_key);
	3849	+ return;
	3850	+ }
	3851	+
	3852	+ // Store the current status for next time (if this is an update)
	3853	+ if ($update) {
	3854	+ set_transient($previous_status_key, $post->post_status, DAY_IN_SECONDS);
	3855	+
	3856	+ // If the post is currently published, also store its URL
	3857	+ if ($post->post_status === 'publish') {
	3858	+ $current_url = get_permalink($post_id);
	3859	+ set_transient($previous_url_key, $current_url, DAY_IN_SECONDS);
	3860	+ }
	3861	+ }
	3862	+
	3863	+ // Only process currently published content for adding/updating
	3864	+ if ($post->post_status === 'publish') {
	3865	+ // Get the source URL
	3866	+ $source_url = get_permalink($post_id);
	3867	+
	3868	+ // Get content with proper formatting (matching ajax_mxchat_process_selected_content)
	3869	+ $title = get_the_title($post_id);
	3870	+ $content = get_post_field('post_content', $post_id);
	3871	+
	3872	+ // Apply WordPress content filters to get properly formatted content
	3873	+ $content = apply_filters('the_content', $content);
	3874	+
	3875	+ // Strip tags but preserve structure
	3876	+ $content = wp_strip_all_tags($content);
	3877	+
	3878	+ // Combine title and content
	3879	+ $final_content = $title . "\n\n" . $content;
	3880	+
	3881	+ // For custom post types like job_listing, include additional fields
	3882	+ if ($post_type === 'job_listing') {
	3883	+ // Add job-specific meta if available
	3884	+ $job_location = get_post_meta($post_id, '_job_location', true);
	3885	+ if (!empty($job_location)) {
	3886	+ $final_content .= "\n\nLocation: " . $job_location;
	3887	+ }
	3888	+
	3889	+ // Get job type terms
	3890	+ $job_types = get_the_terms($post_id, 'job_listing_type');
	3891	+ if (!empty($job_types) && !is_wp_error($job_types)) {
	3892	+ $types = array();
	3893	+ foreach ($job_types as $type) {
	3894	+ $types[] = $type->name;
	3895	+ }
	3896	+ $final_content .= "\n\nJob Type: " . implode(', ', $types);
	3897	+ }
	3898	+
	3899	+ // Get company name if available
	3900	+ $company_name = get_post_meta($post_id, '_company_name', true);
	3901	+ if (!empty($company_name)) {
	3902	+ $final_content .= "\n\nCompany: " . $company_name;
	3903	+ }
	3904	+ }
	3905	+
	3906	+ // Get API key with proper model detection
	3907	+ $options = get_option('mxchat_options');
	3908	+ $selected_model = $options['embedding_model'] ?? 'text-embedding-ada-002';
	3909	+
	3910	+ if (strpos($selected_model, 'voyage') === 0) {
	3911	+ $api_key = $options['voyage_api_key'] ?? '';
	3912	+ } elseif (strpos($selected_model, 'gemini-embedding') === 0) {
	3913	+ $api_key = $options['gemini_api_key'] ?? '';
	3914	+ } else {
	3915	+ $api_key = $options['api_key'] ?? '';
	3916	+ }
	3917	+
	3918	+ if (empty($api_key)) {
	3919	+ return;
	3920	+ }
	3921	+
	3922	+ // Use the centralized utility function for storage
	3923	+ $result = MxChat_Utils::submit_content_to_db(
	3924	+ $final_content,
	3925	+ $source_url,
	3926	+ $api_key,
	3927	+ md5($source_url) // Vector ID for Pinecone
	3928	+ );
	3929	+ }
	3930	+
	3931	+ // Clean up the stored previous status if not used above
	3932	+ if ($previous_status !== 'publish' || $post->post_status === 'publish') {
	3933	+ delete_transient($previous_status_key);
	3934	+ delete_transient($previous_url_key);
	3935	+ }
	3936	+}
	3937	+
	3938	+/**
	3939	+ * Store the post status and URL before update to detect status transitions
	3940	+ * This runs before the post is actually updated in the database
	3941	+ */
	3942	+public function mxchat_store_pre_update_status($post_id, $data) {
	3943	+ // Get the current post from database (before update)
	3944	+ $current_post = get_post($post_id);
	3945	+
	3946	+ if ($current_post) {
	3947	+ // Store the current status temporarily
	3948	+ $status_key = 'mxchat_prev_status_' . $post_id;
	3949	+ set_transient($status_key, $current_post->post_status, HOUR_IN_SECONDS);
	3950	+
	3951	+ // If the post is currently published, also store its URL
	3952	+ if ($current_post->post_status === 'publish') {
	3953	+ $url_key = 'mxchat_prev_url_' . $post_id;
	3954	+ $current_url = get_permalink($post_id);
	3955	+ set_transient($url_key, $current_url, HOUR_IN_SECONDS);
	3956	+ }
	3957	+ }
	3958	+}
	3959	+
	3960	+public function mxchat_handle_post_delete($post_id) {
	3961	+ // Get post data before it's deleted
	3962	+ $post = get_post($post_id);
	3963	+
	3964	+ // Basic validation
	3965	+ if (!$post || wp_is_post_revision($post_id)) {
	3966	+ return;
	3967	+ }
	3968	+
	3969	+ $post_type = $post->post_type;
	3970	+
	3971	+ // Check if sync is enabled for this post type
	3972	+ $should_sync = false;
	3973	+
	3974	+ // Check built-in post types first
	3975	+ if ($post_type === 'post' && get_option('mxchat_auto_sync_posts') === '1') {
	3976	+ $should_sync = true;
	3977	+ } else if ($post_type === 'page' && get_option('mxchat_auto_sync_pages') === '1') {
	3978	+ $should_sync = true;
	3979	+ } else {
	3980	+ // Check custom post types
	3981	+ $option_name = 'mxchat_auto_sync_' . $post_type;
	3982	+ if (get_option($option_name) === '1') {
	3983	+ $should_sync = true;
	3984	+ }
	3985	+ }
	3986	+
	3987	+ if (!$should_sync) {
	3988	+ return;
	3989	+ }
	3990	+
	3991	+ // Get the URL before post is deleted
	3992	+ $source_url = get_permalink($post_id);
	3993	+ if (!$source_url) {
	3994	+ //error_log('MXChat: Failed to get permalink for post ' . $post_id);
	3995	+ return;
	3996	+ }
	3997	+
	3998	+ // Check if Pinecone is enabled
	3999	+ $pinecone_options = get_option('mxchat_pinecone_addon_options', array());
	4000	+ $use_pinecone = ($pinecone_options['mxchat_use_pinecone'] ?? '0') === '1';
	4001	+
	4002	+ if ($use_pinecone && !empty($pinecone_options['mxchat_pinecone_api_key'])) {
	4003	+ // Delete from Pinecone
	4004	+ $this->mxchat_delete_from_pinecone_by_url($source_url, $pinecone_options);
	4005	+ } else {
	4006	+ // Delete from WordPress DB
	4007	+ global $wpdb;
	4008	+ $table_name = $wpdb->prefix . 'mxchat_system_prompt_content';
	4009	+
	4010	+ $result = $wpdb->delete(
	4011	+ $table_name,
	4012	+ array('source_url' => $source_url),
	4013	+ array('%s')
	4014	+ );
	4015	+
	4016	+ if ($result === false) {
	4017	+ //error_log('MXChat: WordPress DB deletion failed for URL: ' . $source_url);
	4018	+ }
	4019	+ }
	4020	+}
	4021	+
	4022	+
	4023	+ /**
	4024	+ * Deletes data from Pinecone using a source URL
	4025	+ */
	4026	+ public function mxchat_delete_from_pinecone_by_url($source_url, $pinecone_options) {
	4027	+ $host = $pinecone_options['mxchat_pinecone_host'] ?? '';
	4028	+ $api_key = $pinecone_options['mxchat_pinecone_api_key'] ?? '';
	4029	+
	4030	+ if (empty($host) || empty($api_key)) {
	4031	+ //error_log('MXChat: Pinecone deletion failed - missing configuration');
	4032	+ return false;
	4033	+ }
	4034	+
	4035	+ $api_endpoint = "https://{$host}/vectors/delete";
	4036	+ $vector_id = md5($source_url);
	4037	+
	4038	+ $request_body = array(
	4039	+ 'ids' => array($vector_id)
	4040	+ );
	4041	+
	4042	+ $response = wp_remote_post($api_endpoint, array(
	4043	+ 'headers' => array(
	4044	+ 'Api-Key' => $api_key,
	4045	+ 'accept' => 'application/json',
	4046	+ 'content-type' => 'application/json'
	4047	+ ),
	4048	+ 'body' => wp_json_encode($request_body),
	4049	+ 'timeout' => 30
	4050	+ ));
	4051	+
	4052	+ if (is_wp_error($response)) {
	4053	+ //error_log('MXChat: Pinecone deletion error - ' . $response->get_error_message());
	4054	+ return false;
	4055	+ }
	4056	+
	4057	+ $response_code = wp_remote_retrieve_response_code($response);
	4058	+ if ($response_code !== 200) {
	4059	+ //error_log('MXChat: Pinecone deletion failed with status ' . $response_code);
	4060	+ return false;
	4061	+ }
	4062	+
	4063	+ return true;
	4064	+ }
	4065	+
	4066	+
	4067	+
	4068	+public function mxchat_handle_product_change($post_id, $post, $update) {
	4069	+ if ($post->post_type !== 'product') {
	4070	+ return;
	4071	+ }
	4072	+
	4073	+ if ($post->post_status === 'publish') {
	4074	+ add_action('shutdown', function() use ($post_id) {
	4075	+ $product = wc_get_product($post_id);
	4076	+ if ($product) {
	4077	+ $this->mxchat_store_product_embedding($product);
	4078	+ }
	4079	+ });
	4080	+ }
	4081	+}
	4082	+
	4083	+/**
	4084	+ * Store WooCommerce product embeddings
	4085	+ */
	4086	+private function mxchat_store_product_embedding($product) {
	4087	+ if (!isset($this->options['enable_woocommerce_integration']) ||
	4088	+ !in_array($this->options['enable_woocommerce_integration'], ['1', 'on'])) {
	4089	+ return;
	4090	+ }
	4091	+
	4092	+ $source_url = get_permalink($product->get_id());
	4093	+
	4094	+ // Build product content
	4095	+ $title = $product->get_name();
	4096	+ $description = $product->get_description();
	4097	+ $short_description = $product->get_short_description();
	4098	+ $regular_price = $product->get_regular_price();
	4099	+ $sale_price = $product->get_sale_price();
	4100	+ $sku = $product->get_sku();
	4101	+
	4102	+ // Format content consistently
	4103	+ $content = $title . "\n\n";
	4104	+
	4105	+ if (!empty($description)) {
	4106	+ $content .= wp_strip_all_tags($description) . "\n\n";
	4107	+ }
	4108	+
	4109	+ if (!empty($short_description)) {
	4110	+ $content .= "Short Description: " . wp_strip_all_tags($short_description) . "\n\n";
	4111	+ }
	4112	+
	4113	+ $content .= "Price: $" . $regular_price . "\n";
	4114	+
	4115	+ if (!empty($sale_price)) {
	4116	+ $content .= "Sale Price: $" . $sale_price . "\n";
	4117	+ }
	4118	+
	4119	+ if (!empty($sku)) {
	4120	+ $content .= "SKU: " . $sku . "\n";
	4121	+ }
	4122	+
	4123	+ // Get API key with proper model detection
	4124	+ $options = get_option('mxchat_options');
	4125	+ $selected_model = $options['embedding_model'] ?? 'text-embedding-ada-002';
	4126	+
	4127	+ if (strpos($selected_model, 'voyage') === 0) {
	4128	+ $api_key = $options['voyage_api_key'] ?? '';
	4129	+ } elseif (strpos($selected_model, 'gemini-embedding') === 0) {
	4130	+ $api_key = $options['gemini_api_key'] ?? '';
	4131	+ } else {
	4132	+ $api_key = $options['api_key'] ?? '';
	4133	+ }
	4134	+
	4135	+ if (empty($api_key)) {
	4136	+ //error_log('MxChat Auto-sync: No API key configured for embedding model');
	4137	+ return;
	4138	+ }
	4139	+
	4140	+ // Use the centralized utility function for storage
	4141	+ $result = MxChat_Utils::submit_content_to_db(
	4142	+ $content,
	4143	+ $source_url,
	4144	+ $api_key,
	4145	+ md5($source_url) // Vector ID for Pinecone
	4146	+ );
	4147	+
	4148	+ if (is_wp_error($result)) {
	4149	+ //error_log('MxChat WooCommerce sync failed for product ' . $product->get_id() . ': ' . $result->get_error_message());
	4150	+ }
	4151	+}
	4152	+
	4153	+public function mxchat_handle_product_delete($post_id) {
	4154	+ if (get_post_type($post_id) !== 'product') {
	4155	+ return;
	4156	+ }
	4157	+
	4158	+ $source_url = get_permalink($post_id);
	4159	+
	4160	+ // Check if Pinecone is enabled
	4161	+ $pinecone_options = get_option('mxchat_pinecone_addon_options', array());
	4162	+ $use_pinecone = ($pinecone_options['mxchat_use_pinecone'] ?? '0') === '1';
	4163	+
	4164	+ if ($use_pinecone && !empty($pinecone_options['mxchat_pinecone_api_key'])) {
	4165	+ // Delete from Pinecone
	4166	+ $this->mxchat_delete_from_pinecone_by_url($source_url, $pinecone_options);
	4167	+ } else {
	4168	+ // Delete from WordPress DB
	4169	+ global $wpdb;
	4170	+ $table_name = $wpdb->prefix . 'mxchat_system_prompt_content';
	4171	+
	4172	+ $wpdb->delete(
	4173	+ $table_name,
	4174	+ array('source_url' => $source_url),
	4175	+ array('%s')
	4176	+ );
	4177	+ }
	4178	+}
	4179	+
	4180	+/**
	4181	+ * Handle individual Pinecone content deletion
	4182	+ */
	4183	+public function mxchat_handle_pinecone_prompt_delete() {
	4184	+ // Check permissions
	4185	+ if (!current_user_can('manage_options')) {
	4186	+ wp_die(esc_html__('You do not have sufficient permissions.', 'mxchat'));
	4187	+ }
	4188	+
	4189	+ // Verify nonce
	4190	+ if (!isset($_GET['_wpnonce']) || !wp_verify_nonce($_GET['_wpnonce'], 'mxchat_delete_pinecone_prompt_nonce')) {
	4191	+ wp_die(esc_html__('Security check failed.', 'mxchat'));
	4192	+ }
	4193	+
	4194	+ $vector_id = isset($_GET['vector_id']) ? sanitize_text_field($_GET['vector_id']) : '';
	4195	+
	4196	+ if (empty($vector_id)) {
	4197	+ set_transient('mxchat_admin_notice_error',
	4198	+ esc_html__('Invalid vector ID.', 'mxchat'),
	4199	+ 30
	4200	+ );
	4201	+ wp_safe_redirect(admin_url('admin.php?page=mxchat-prompts'));
	4202	+ exit;
	4203	+ }
	4204	+
	4205	+ // Get Pinecone settings
	4206	+ $pinecone_options = get_option('mxchat_pinecone_addon_options', array());
	4207	+ $use_pinecone = ($pinecone_options['mxchat_use_pinecone'] ?? '0') === '1';
	4208	+
	4209	+ if (!$use_pinecone || empty($pinecone_options['mxchat_pinecone_api_key'])) {
	4210	+ set_transient('mxchat_admin_notice_error',
	4211	+ esc_html__('Pinecone is not properly configured.', 'mxchat'),
	4212	+ 30
	4213	+ );
	4214	+ wp_safe_redirect(admin_url('admin.php?page=mxchat-prompts'));
	4215	+ exit;
	4216	+ }
	4217	+
	4218	+ // Delete from Pinecone
	4219	+ $pinecone_manager = MxChat_Pinecone_Manager::get_instance();
	4220	+ $result = $pinecone_manager->mxchat_delete_from_pinecone_by_vector_id(
	4221	+ $vector_id,
	4222	+ $pinecone_options['mxchat_pinecone_api_key'],
	4223	+ $pinecone_options['mxchat_pinecone_host']
	4224	+ );
	4225	+
	4226	+ if ($result['success']) {
	4227	+ // No cache clearing needed since we removed caching
	4228	+ set_transient('mxchat_admin_notice_success',
	4229	+ esc_html__('Entry deleted successfully from Pinecone.', 'mxchat'),
	4230	+ 30
	4231	+ );
	4232	+ } else {
	4233	+ set_transient('mxchat_admin_notice_error',
	4234	+ esc_html__('Failed to delete entry: ', 'mxchat') . $result['message'],
	4235	+ 30
	4236	+ );
	4237	+ }
	4238	+
	4239	+ wp_safe_redirect(admin_url('admin.php?page=mxchat-prompts'));
	4240	+ exit;
	4241	+}
	4242	+/**
	4243	+ * Handle individual Pinecone content deletion via AJAX
	4244	+ */
	4245	+public function ajax_mxchat_delete_pinecone_prompt() {
	4246	+ // Verify nonce and permissions
	4247	+ if (!check_ajax_referer('mxchat_delete_pinecone_prompt_nonce', 'nonce', false)) {
	4248	+ wp_send_json_error('Invalid nonce');
	4249	+ exit;
	4250	+ }
	4251	+
	4252	+ if (!current_user_can('manage_options')) {
	4253	+ wp_send_json_error('Unauthorized access');
	4254	+ exit;
	4255	+ }
	4256	+
	4257	+ $vector_id = isset($_POST['vector_id']) ? sanitize_text_field($_POST['vector_id']) : '';
	4258	+ $bot_id = isset($_POST['bot_id']) ? sanitize_text_field($_POST['bot_id']) : 'default';
	4259	+
	4260	+ if (empty($vector_id)) {
	4261	+ wp_send_json_error('Missing vector ID');
	4262	+ exit;
	4263	+ }
	4264	+
	4265	+ // Get bot-specific Pinecone settings
	4266	+ $pinecone_manager = MxChat_Pinecone_Manager::get_instance();
	4267	+ $pinecone_options = $pinecone_manager->mxchat_get_bot_pinecone_options($bot_id);
	4268	+
	4269	+ $use_pinecone = ($pinecone_options['mxchat_use_pinecone'] ?? '0') === '1';
	4270	+
	4271	+ if (!$use_pinecone || empty($pinecone_options['mxchat_pinecone_api_key'])) {
	4272	+ wp_send_json_error('Pinecone is not properly configured for bot: ' . $bot_id);
	4273	+ exit;
	4274	+ }
	4275	+
	4276	+ // Delete from the correct Pinecone index
	4277	+ $result = $pinecone_manager->mxchat_delete_from_pinecone_by_vector_id(
	4278	+ $vector_id,
	4279	+ $pinecone_options['mxchat_pinecone_api_key'],
	4280	+ $pinecone_options['mxchat_pinecone_host']
	4281	+ );
	4282	+
	4283	+ if ($result['success']) {
	4284	+ // No cache clearing needed since we removed caching
	4285	+ wp_send_json_success(array(
	4286	+ 'message' => 'Entry deleted successfully from Pinecone',
	4287	+ 'vector_id' => $vector_id,
	4288	+ 'bot_id' => $bot_id
	4289	+ ));
	4290	+ } else {
	4291	+ wp_send_json_error('Failed to delete from Pinecone: ' . $result['message']);
	4292	+ }
	4293	+
	4294	+ exit;
	4295	+}
	4296	+
	4297	+/**
	4298	+ * NEW: Get hierarchical roles for dropdown
	4299	+ */
	4300	+public function mxchat_get_role_options() {
	4301	+ return array(
	4302	+ 'public' => __('Public (Everyone)', 'mxchat'),
	4303	+ 'logged_in' => __('Logged In Users', 'mxchat'),
	4304	+ 'subscriber' => __('Subscribers & Above', 'mxchat'),
	4305	+ 'contributor' => __('Contributors & Above', 'mxchat'),
	4306	+ 'author' => __('Authors & Above', 'mxchat'),
	4307	+ 'editor' => __('Editors & Above', 'mxchat'),
	4308	+ 'administrator' => __('Administrators Only', 'mxchat')
	4309	+ );
	4310	+}
	4311	+
	4312	+/**
	4313	+ * NEW: Check if user has access to content based on role restriction
	4314	+ */
	4315	+public function mxchat_user_has_content_access($role_restriction) {
	4316	+ // Public content is always accessible
	4317	+ if ($role_restriction === 'public' || empty($role_restriction)) {
	4318	+ return true;
	4319	+ }
	4320	+
	4321	+ // Check if user is logged in for logged_in restriction
	4322	+ if ($role_restriction === 'logged_in') {
	4323	+ return is_user_logged_in();
	4324	+ }
	4325	+
	4326	+ // If not logged in, no access to role-restricted content
	4327	+ if (!is_user_logged_in()) {
	4328	+ return false;
	4329	+ }
	4330	+
	4331	+ $user = wp_get_current_user();
	4332	+ $user_roles = $user->roles;
	4333	+
	4334	+ if (empty($user_roles)) {
	4335	+ return false;
	4336	+ }
	4337	+
	4338	+ // Define role hierarchy (higher number = higher access)
	4339	+ $hierarchy = array(
	4340	+ 'subscriber' => 1,
	4341	+ 'contributor' => 2,
	4342	+ 'author' => 3,
	4343	+ 'editor' => 4,
	4344	+ 'administrator' => 5
	4345	+ );
	4346	+
	4347	+ // Get required level
	4348	+ $required_level = isset($hierarchy[$role_restriction]) ? $hierarchy[$role_restriction] : 0;
	4349	+
	4350	+ // Check if user has required level or higher
	4351	+ foreach ($user_roles as $user_role) {
	4352	+ $user_level = isset($hierarchy[$user_role]) ? $hierarchy[$user_role] : 0;
	4353	+ if ($user_level >= $required_level) {
	4354	+ return true;
	4355	+ }
	4356	+ }
	4357	+
	4358	+ return false;
	4359	+}
	4360	+
	4361	+/**
	4362	+ * Handle role restriction updates via AJAX
	4363	+ * UPDATED: Removed cache clearing call since we removed caching
	4364	+ */
	4365	+public function ajax_mxchat_update_role_restriction() {
	4366	+ // Verify nonce and permissions
	4367	+ if (!check_ajax_referer('mxchat_update_role_nonce', 'nonce', false)) {
	4368	+ wp_send_json_error('Invalid nonce');
	4369	+ exit;
	4370	+ }
	4371	+
	4372	+ if (!current_user_can('manage_options')) {
	4373	+ wp_send_json_error('Unauthorized access');
	4374	+ exit;
	4375	+ }
	4376	+
	4377	+ $entry_id = isset($_POST['entry_id']) ? sanitize_text_field($_POST['entry_id']) : '';
	4378	+ $role_restriction = isset($_POST['role_restriction']) ? sanitize_text_field($_POST['role_restriction']) : 'public';
	4379	+ $data_source = isset($_POST['data_source']) ? sanitize_text_field($_POST['data_source']) : 'wordpress';
	4380	+
	4381	+ if (empty($entry_id)) {
	4382	+ wp_send_json_error('Invalid entry ID');
	4383	+ exit;
	4384	+ }
	4385	+
	4386	+ // Get knowledge manager instance to validate role restriction
	4387	+ $knowledge_manager = MxChat_Knowledge_Manager::get_instance();
	4388	+ $valid_roles = array_keys($knowledge_manager->mxchat_get_role_options());
	4389	+ if (!in_array($role_restriction, $valid_roles)) {
	4390	+ wp_send_json_error('Invalid role restriction');
	4391	+ exit;
	4392	+ }
	4393	+
	4394	+ global $wpdb;
	4395	+
	4396	+ if ($data_source === 'pinecone') {
	4397	+ // Handle Pinecone role restriction (stored separately in WordPress table)
	4398	+ $roles_table = $wpdb->prefix . 'mxchat_pinecone_roles';
	4399	+
	4400	+ // Use REPLACE to insert or update the role restriction
	4401	+ $result = $wpdb->replace(
	4402	+ $roles_table,
	4403	+ array(
	4404	+ 'vector_id' => $entry_id,
	4405	+ 'role_restriction' => $role_restriction,
	4406	+ 'updated_at' => current_time('mysql')
	4407	+ ),
	4408	+ array('%s', '%s', '%s')
	4409	+ );
	4410	+
	4411	+ // No cache clearing needed since we removed caching
	4412	+
	4413	+ } else {
	4414	+ // Handle WordPress database role restriction (existing functionality)
	4415	+ $table_name = $wpdb->prefix . 'mxchat_system_prompt_content';
	4416	+
	4417	+ $result = $wpdb->update(
	4418	+ $table_name,
	4419	+ array('role_restriction' => $role_restriction),
	4420	+ array('id' => absint($entry_id)),
	4421	+ array('%s'),
	4422	+ array('%d')
	4423	+ );
	4424	+ }
	4425	+
	4426	+ if ($result === false) {
	4427	+ wp_send_json_error('Database update failed: ' . $wpdb->last_error);
	4428	+ exit;
	4429	+ }
	4430	+
	4431	+ wp_send_json_success(array(
	4432	+ 'message' => 'Role restriction updated successfully',
	4433	+ 'role_restriction' => $role_restriction,
	4434	+ 'data_source' => $data_source,
	4435	+ 'entry_id' => $entry_id
	4436	+ ));
	4437	+ exit;
	4438	+}
	4439	+
	4440	+ // ========================================
	4441	+ // HELPER METHODS
	4442	+ // ========================================
	4443	+
	4444	+ /**
	4445	+ * Check if user has required permissions for content processing
	4446	+ */
	4447	+ private function mxchat_check_user_permissions() {
	4448	+ if (!current_user_can('manage_options')) {
	4449	+ wp_die(esc_html__('You do not have sufficient permissions.', 'mxchat'));
	4450	+ }
	4451	+ }
	4452	+
	4453	+ /**
	4454	+ * Validate nonce for security
	4455	+ */
	4456	+ private function mxchat_validate_nonce($nonce_name, $nonce_action) {
	4457	+ if (!isset($_POST[$nonce_name]) || !wp_verify_nonce($_POST[$nonce_name], $nonce_action)) {
	4458	+ wp_die(esc_html__('Security check failed.', 'mxchat'));
	4459	+ }
	4460	+ }
	4461	+
	4462	+ /**
	4463	+ * Get embedding API credentials
	4464	+ */
	4465	+ private function mxchat_get_embedding_credentials() {
	4466	+ $embedding_model = $this->options['embedding_model'] ?? 'text-embedding-ada-002';
	4467	+
	4468	+ if (strpos($embedding_model, 'text-embedding-') !== false) {
	4469	+ return array(
	4470	+ 'type' => 'openai',
	4471	+ 'api_key' => $this->options['api_key'] ?? ''
	4472	+ );
	4473	+ } elseif (strpos($embedding_model, 'voyage-') !== false) {
	4474	+ return array(
	4475	+ 'type' => 'voyage',
	4476	+ 'api_key' => $this->options['voyage_api_key'] ?? ''
	4477	+ );
	4478	+ } elseif (strpos($embedding_model, 'gemini-embedding-') !== false) {
	4479	+ return array(
	4480	+ 'type' => 'gemini',
	4481	+ 'api_key' => $this->options['gemini_api_key'] ?? ''
	4482	+ );
	4483	+ }
	4484	+
	4485	+ return array('type' => 'unknown', 'api_key' => '');
	4486	+ }
	4487	+
	4488	+ /**
	4489	+ * Log processing errors
	4490	+ */
	4491	+ private function mxchat_log_processing_error($operation, $error_message) {
	4492	+ //error_log("MxChat Knowledge Processing {$operation} Error: " . $error_message);
	4493	+ }
	4494	+
	4495	+ /**
	4496	+ * Set admin notice transient
	4497	+ */
	4498	+ private function mxchat_set_admin_notice($type, $message) {
	4499	+ set_transient("mxchat_admin_notice_{$type}", $message, 30);
	4500	+ }
	4501	+
	4502	+ /**
	4503	+ * Get Pinecone manager instance for vector operations
	4504	+ */
	4505	+ private function mxchat_get_pinecone_manager() {
	4506	+ return MxChat_Pinecone_Manager::get_instance();
	4507	+ }
	4508	+
	4509	+ // ========================================
	4510	+ // STATIC ACCESS METHODS
	4511	+ // ========================================
	4512	+
	4513	+ /**
	4514	+ * Get singleton instance
	4515	+ */
	4516	+ public static function get_instance() {
	4517	+ static $instance = null;
	4518	+ if ($instance === null) {
	4519	+ $instance = new self();
	4520	+ }
	4521	+ return $instance;
	4522	+ }
	4523	+}
	4524	+
	4525	+// Initialize the Knowledge manager
8407	4526	$mxchat_knowledge_manager = MxChat_Knowledge_Manager::get_instance();