PluginProbe
MxChat – AI Chatbot & Content Generation for WordPress / 2.2.7
MxChat – AI Chatbot & Content Generation for WordPress v2.2.7
3.2.21 3.2.20 3.2.19 3.2.18 3.2.17 3.2.16 3.2.15 3.2.14 3.2.12 3.2.13 3.2.11 3.2.10 3.2.9 3.2.8 3.2.7 3.2.6 3.2.5 3.2.4 3.2.3 3.2.2 3.2.1 2.0.3 2.0.4 2.0.5 2.0.6 All 152 releases
← All changes | admin/class-knowledge-manager.php +1637 -6473 3.2.12 → 2.2.7 View file →
@@ -15,73 +15,51 @@
15 15
16 16 /**
17 17 * Constructor - Register hooks for content processing
18 18 */
19 -public function __construct() {
20 - $this->options = get_option('mxchat_options', array());
21 - $this->mxchat_init_hooks();
19 + public function __construct() {
20 + $this->options = get_option('mxchat_options', array());
21 + $this->mxchat_init_hooks();
22 + }
22 23
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'));
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($this, 'mxchat_ajax_dismiss_completed_status')); // NEW
36 + add_action('wp_ajax_mxchat_get_content_list', array($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($this, 'mxchat_delete_from_pinecone_by_url'), 10, 1);
63 40
64 - // WordPress post management hooks
65 - add_action('pre_post_update', array($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($this, 'mxchat_handle_product_delete'));
41 +
42 + // Cron handlers for background processing
43 + add_action('mxchat_process_sitemap_urls', array($this, 'mxchat_process_sitemap_urls_cron'), 10, 5);
44 + add_action('mxchat_process_pdf_pages', array($this, 'mxchat_process_pdf_pages_cron'), 10, 5);
45 +
46 + // WordPress post management hooks
47 + add_action('post_updated', array($this, 'mxchat_handle_post_update'), 10, 3);
48 + add_action('before_delete_post', array($this, 'mxchat_handle_post_delete'));
49 + add_action('wp_trash_post', array($this, 'mxchat_handle_post_delete'));
50 + add_action('wp_ajax_mxchat_save_inline_prompt', array($this, 'mxchat_save_inline_prompt'));
51 + add_action('admin_post_mxchat_delete_pinecone_prompt', array($this, 'mxchat_handle_pinecone_prompt_delete'));
52 + add_action('wp_ajax_mxchat_delete_pinecone_prompt', array($this, 'ajax_mxchat_delete_pinecone_prompt'));
53 +
54 + // WooCommerce product hooks (if WooCommerce is active)
55 + if (class_exists('WooCommerce')) {
56 + add_action('save_post_product', array($this, 'mxchat_handle_product_change'), 10, 3);
57 + add_action('wp_trash_post', array($this, 'mxchat_handle_product_delete'));
58 + add_action('before_delete_post', array($this, 'mxchat_handle_product_delete'));
59 + }
60 +
82 61 }
83 -}
84 62
85 63 /**
86 64 * Get current options (refreshed)
87 65 */
@@ -91,9 +69,315 @@
91 69 }
92 70 return $this->options;
93 71 }
94 72
73 +
74 + /**
75 + * Handle manual batch processing via AJAX
76 + */
77 +public function ajax_manual_batch_process() {
78 + try {
79 + // Verify nonce and permissions
80 + check_ajax_referer('mxchat_status_nonce', 'nonce');
81 +
82 + if (!current_user_can('manage_options')) {
83 + wp_send_json_error('Unauthorized access');
84 + }
85 +
86 + $process_type = sanitize_text_field($_POST['process_type'] ?? '');
87 + $url = sanitize_text_field($_POST['url'] ?? '');
88 +
89 + if (empty($process_type) || empty($url)) {
90 + wp_send_json_error('Missing required parameters');
91 + }
92 +
93 + $processed = 0;
94 +
95 + if ($process_type === 'pdf') {
96 + $processed = $this->mxchat_manual_process_pdf_batch($url);
97 + } elseif ($process_type === 'sitemap') {
98 + $processed = $this->mxchat_manual_process_sitemap_batch($url);
99 + }
100 +
101 + if ($processed > 0) {
102 + wp_send_json_success(array(
103 + 'message' => "Processed {$processed} items successfully",
104 + 'processed' => $processed
105 + ));
106 + } else {
107 + wp_send_json_error('No items were processed');
108 + }
109 +
110 + } catch (Exception $e) {
111 + //error_log('Manual batch process error: ' . $e->getMessage());
112 + wp_send_json_error('Processing failed: ' . $e->getMessage());
113 + }
114 +}
95 115
116 +/**
117 + * Process a small PDF batch manually - DIRECT PROCESSING
118 + */
119 +private function mxchat_manual_process_pdf_batch($pdf_url) {
120 + try {
121 + $status_key = sanitize_key('mxchat_pdf_status_' . md5($pdf_url));
122 + $status = get_transient($status_key);
123 +
124 + if (!$status || $status['status'] !== 'processing') {
125 + //error_log('Manual PDF: No processing status found');
126 + return 0;
127 + }
128 +
129 + //error_log('Manual PDF: Starting direct processing for ' . $pdf_url);
130 +
131 + // Get current progress
132 + $current_page = $status['processed_pages'] ?? 0;
133 + $total_pages = $status['total_pages'] ?? 0;
134 +
135 + if ($current_page >= $total_pages) {
136 + //error_log('Manual PDF: Already completed');
137 + return 0;
138 + }
139 +
140 + // Try to download the PDF again for processing
141 + $response = wp_remote_get($pdf_url, array('timeout' => 30));
142 +
143 + if (is_wp_error($response)) {
144 + //error_log('Manual PDF: Failed to download PDF: ' . $response->get_error_message());
145 + return 0;
146 + }
147 +
148 + $pdf_content = wp_remote_retrieve_body($response);
149 + if (empty($pdf_content)) {
150 + //error_log('Manual PDF: Empty PDF content');
151 + return 0;
152 + }
153 +
154 + // Save PDF temporarily
155 + $upload_dir = wp_upload_dir();
156 + $temp_pdf_path = trailingslashit($upload_dir['path']) . 'temp_manual_' . time() . '.pdf';
157 + file_put_contents($temp_pdf_path, $pdf_content);
158 +
159 + // Process 2 pages directly
160 + $processed = $this->mxchat_process_pdf_pages_direct($temp_pdf_path, $pdf_url, $current_page, 5);
161 +
162 + // Clean up temp file
163 + if (file_exists($temp_pdf_path)) {
164 + wp_delete_file($temp_pdf_path);
165 + }
166 +
167 + //error_log('Manual PDF: Processed ' . $processed . ' pages');
168 + return $processed;
169 +
170 + } catch (Exception $e) {
171 + //error_log('Manual PDF batch error: ' . $e->getMessage());
172 + return 0;
173 + }
174 +}
175 +
176 +/**
177 + * Process PDF pages directly without cron
178 + */
179 +private function mxchat_process_pdf_pages_direct($pdf_path, $pdf_url, $start_page, $batch_size) {
180 + try {
181 + if (!file_exists($pdf_path)) {
182 + //error_log('Direct PDF: File not found at ' . $pdf_path);
183 + return 0;
184 + }
185 +
186 + $parser = new \Smalot\PdfParser\Parser();
187 + $pdf = $parser->parseFile($pdf_path);
188 + $pages = $pdf->getPages();
189 +
190 + $status_key = sanitize_key('mxchat_pdf_status_' . md5($pdf_url));
191 + $status = get_transient($status_key);
192 +
193 + if (!$status) {
194 + return 0;
195 + }
196 +
197 + $options = get_option('mxchat_options');
198 + $api_key = $options['api_key'] ?? '';
199 +
200 + if (empty($api_key)) {
201 + //error_log('Direct PDF: No API key');
202 + return 0;
203 + }
204 +
205 + $processed = 0;
206 + $end_page = min($start_page + $batch_size, count($pages));
207 +
208 + for ($i = $start_page; $i < $end_page; $i++) {
209 + try {
210 + $page_number = $i + 1;
211 + $text = $pages[$i]->getText();
212 +
213 + if (empty($text)) {
214 + //error_log('Direct PDF: Empty text on page ' . $page_number);
215 + continue;
216 + }
217 +
218 + $sanitized_content = $this->mxchat_sanitize_content_for_api($text);
219 + if (empty($sanitized_content)) {
220 + //error_log('Direct PDF: No content after sanitization on page ' . $page_number);
221 + continue;
222 + }
223 +
224 + // Generate embedding
225 + $embedding_vector = $this->mxchat_generate_embedding($sanitized_content);
226 + if (is_string($embedding_vector)) {
227 + //error_log('Direct PDF: Embedding failed on page ' . $page_number . ': ' . $embedding_vector);
228 + continue;
229 + }
230 +
231 + // Create metadata
232 + $metadata = array(
233 + 'document_type' => 'pdf',
234 + 'total_pages' => count($pages),
235 + 'current_page' => $page_number,
236 + 'source_url' => $pdf_url
237 + );
238 +
239 + $content_with_metadata = wp_json_encode($metadata) . "\n---\n" . $sanitized_content;
240 + $page_url = esc_url($pdf_url . "#page=" . $page_number);
241 +
242 + // Store in database
243 + $db_result = MxChat_Utils::submit_content_to_db($content_with_metadata, $page_url, $api_key);
244 +
245 + if (is_wp_error($db_result)) {
246 + //error_log('Direct PDF: DB error on page ' . $page_number . ': ' . $db_result->get_error_message());
247 + continue;
248 + }
249 +
250 + $processed++;
251 + //error_log('Direct PDF: Successfully processed page ' . $page_number);
252 +
253 + // Update status
254 + $status['processed_pages'] = $i + 1;
255 + $status['last_update'] = time();
256 + $status['percentage'] = round(($status['processed_pages'] / $status['total_pages']) * 100);
257 + set_transient($status_key, $status, DAY_IN_SECONDS);
258 +
259 + } catch (Exception $e) {
260 + //error_log('Direct PDF: Error processing page ' . ($i + 1) . ': ' . $e->getMessage());
261 + continue;
262 + }
263 + }
264 +
265 + // Check if completed
266 + if ($status['processed_pages'] >= $status['total_pages']) {
267 + $status['status'] = 'complete';
268 + set_transient($status_key, $status, DAY_IN_SECONDS);
269 + //error_log('Direct PDF: Processing completed');
270 + }
271 +
272 + return $processed;
273 +
274 + } catch (Exception $e) {
275 + //error_log('Direct PDF processing error: ' . $e->getMessage());
276 + return 0;
277 + }
278 +}
279 +
280 +/**
281 + * Process a small sitemap batch manually - DIRECT PROCESSING
282 + */
283 +private function mxchat_manual_process_sitemap_batch($sitemap_url) {
284 + try {
285 + $status_key = sanitize_key('mxchat_sitemap_status_' . md5($sitemap_url));
286 + $status = get_transient($status_key);
287 +
288 + if (!$status || $status['status'] !== 'processing') {
289 + return 0;
290 + }
291 +
292 + //error_log('Manual Sitemap: Starting direct processing for ' . $sitemap_url);
293 +
294 + // Re-fetch the sitemap to get URLs
295 + $response = wp_remote_get($sitemap_url, array('timeout' => 30));
296 + if (is_wp_error($response)) {
297 + //error_log('Manual Sitemap: Failed to fetch sitemap');
298 + return 0;
299 + }
300 +
301 + $sitemap_content = wp_remote_retrieve_body($response);
302 + $xml = simplexml_load_string($sitemap_content);
303 +
304 + if (!$xml) {
305 + //error_log('Manual Sitemap: Invalid XML');
306 + return 0;
307 + }
308 +
309 + $urls = array();
310 + foreach ($xml->url as $url_element) {
311 + $urls[] = (string)$url_element->loc;
312 + }
313 +
314 + $current_processed = $status['processed_urls'] ?? 0;
315 + $batch_size = 5;
316 + $processed = 0;
317 +
318 + // Process next 2 URLs
319 + for ($i = $current_processed; $i < min($current_processed + $batch_size, count($urls)); $i++) {
320 + $url = $urls[$i];
321 +
322 + if ($this->mxchat_process_single_url_direct($url)) {
323 + $processed++;
324 + }
325 +
326 + // Update status
327 + $status['processed_urls'] = $i + 1;
328 + $status['last_update'] = time();
329 + $status['percentage'] = round(($status['processed_urls'] / $status['total_urls']) * 100);
330 + set_transient($status_key, $status, DAY_IN_SECONDS);
331 + }
332 +
333 + // Check if completed
334 + if ($status['processed_urls'] >= $status['total_urls']) {
335 + $status['status'] = 'complete';
336 + set_transient($status_key, $status, DAY_IN_SECONDS);
337 + }
338 +
339 + //error_log('Manual Sitemap: Processed ' . $processed . ' URLs');
340 + return $processed;
341 +
342 + } catch (Exception $e) {
343 + //error_log('Manual sitemap batch error: ' . $e->getMessage());
344 + return 0;
345 + }
346 +}
347 +
348 +/**
349 + * Process a single URL directly
350 + */
351 +private function mxchat_process_single_url_direct($url) {
352 + try {
353 + $response = wp_remote_get($url, array('timeout' => 30));
354 + if (is_wp_error($response)) {
355 + return false;
356 + }
357 +
358 + $html = wp_remote_retrieve_body($response);
359 + $content = $this->mxchat_extract_main_content($html);
360 + $sanitized = $this->mxchat_sanitize_content_for_api($content);
361 +
362 + if (empty($sanitized)) {
363 + return false;
364 + }
365 +
366 + $options = get_option('mxchat_options');
367 + $api_key = $options['api_key'] ?? '';
368 +
369 + $result = MxChat_Utils::submit_content_to_db($sanitized, $url, $api_key);
370 +
371 + return !is_wp_error($result);
372 +
373 + } catch (Exception $e) {
374 + //error_log('Single URL processing error: ' . $e->getMessage());
375 + return false;
376 + }
377 +}
378 +
379 +
96 380 // ========================================
97 381 // MAIN CONTENT SUBMISSION HANDLERS
98 382 // ========================================
99 383
@@ -109,18 +393,13 @@
109 393 wp_die(esc_html__('Nonce verification failed.', 'mxchat'));
110 394 }
111 395
112 396 // 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']));
397 + $article_content = sanitize_textarea_field($_POST['article_content']);
115 398 $article_url = isset($_POST['article_url']) ? esc_url_raw($_POST['article_url']) : '';
116 399
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');
400 + // Get API key for submission
401 + $options = get_option('mxchat_options');
123 402 $selected_model = $options['embedding_model'] ?? 'text-embedding-ada-002';
124 403
125 404 if (strpos($selected_model, 'voyage') === 0) {
126 405 $api_key = $options['voyage_api_key'] ?? '';
@@ -138,10 +417,10 @@
138 417 wp_safe_redirect(esc_url(admin_url('admin.php?page=mxchat-prompts')));
139 418 exit;
140 419 }
141 420
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);
421 + // Use centralized utility function for storage
422 + $result = MxChat_Utils::submit_content_to_db($article_content, $article_url, $api_key);
144 423
145 424 if (is_wp_error($result)) {
146 425 set_transient('mxchat_admin_notice_error',
147 426 esc_html__('Error storing content: ', 'mxchat') . $result->get_error_message(),
@@ -156,23 +435,17 @@
156 435
157 436 wp_safe_redirect(esc_url(admin_url('admin.php?page=mxchat-prompts')));
158 437 exit;
159 438 }
160 -
161 439 public function mxchat_is_pdf_url($url, $response) {
162 440 $content_type = wp_remote_retrieve_header($response, 'content-type');
163 441 $file_extension = strtolower(pathinfo($url, PATHINFO_EXTENSION));
164 442
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' || $has_pdf_disposition;
443 + return strpos($content_type, 'pdf') !== false || $file_extension === 'pdf';
170 444 }
171 -
172 -
173 -public function mxchat_handle_pdf_for_knowledge_base($pdf_url, $response, $bot_id = 'default') {
445 +public function mxchat_handle_pdf_for_knowledge_base($pdf_url, $response) {
174 446 if (!current_user_can('manage_options')) {
447 + //error_log(esc_html__('Unauthorized PDF processing attempt', 'mxchat'));
175 448 return false;
176 449 }
177 450
178 451 $pdf_url = esc_url_raw($pdf_url);
@@ -178,8 +451,9 @@
178 451 $pdf_url = esc_url_raw($pdf_url);
179 452 $upload_dir = wp_upload_dir();
180 453
181 454 if (isset($upload_dir['error']) && $upload_dir['error'] !== false) {
455 + //error_log(sprintf(esc_html__('Upload directory error: %s', 'mxchat'), esc_html($upload_dir['error'])));
182 456 return false;
183 457 }
184 458
185 459 $pdf_filename = sanitize_file_name('mxchat_kb_' . time() . '.pdf');
@@ -186,12 +460,14 @@
186 460 $pdf_path = trailingslashit($upload_dir['path']) . $pdf_filename;
187 461
188 462 $response_body = wp_remote_retrieve_body($response);
189 463 if (empty($response_body)) {
464 + //error_log(esc_html__('Empty PDF response body', 'mxchat'));
190 465 return false;
191 466 }
192 467
193 468 if (!wp_mkdir_p(dirname($pdf_path))) {
469 + //error_log(sprintf(esc_html__('Failed to create directory for PDF: %s', 'mxchat'), esc_html($pdf_path)));
194 470 return false;
195 471 }
196 472
197 473 try {
@@ -200,915 +476,303 @@
200 476 if (!file_exists($pdf_path)) {
201 477 throw new Exception(__('Failed to save PDF file', 'mxchat'));
202 478 }
203 479
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 - }
480 + $parser = new \Smalot\PdfParser\Parser();
481 + $pdf = $parser->parseFile($pdf_path);
482 + $total_pages = absint(count($pdf->getPages()));
209 483
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' => $total_pages
221 - );
484 + if ($total_pages < 1) {
485 + throw new Exception(__('Invalid PDF: no pages found', 'mxchat'));
222 486 }
223 487
224 - // Add pages to queue
225 - $queued_count = $this->mxchat_add_to_queue($queue_id, 'pdf_page', $pages, $bot_id);
488 + wp_schedule_single_event(time(), 'mxchat_process_pdf_pages', array(
489 + 'pdf_path' => $pdf_path,
490 + 'pdf_url' => $pdf_url,
491 + 'total_pages' => $total_pages,
492 + 'batch_size' => absint(15),
493 + 'batch_pause' => absint(10)
494 + ));
226 495
227 - if ($queued_count === 0) {
228 - wp_delete_file($pdf_path);
229 - throw new Exception(__('Failed to add PDF pages to processing queue', 'mxchat'));
230 - }
496 + $status_data = array(
497 + 'total_pages' => $total_pages,
498 + 'processed_pages' => 0,
499 + 'status' => 'processing',
500 + 'last_update' => time()
501 + );
231 502
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'));
503 + set_transient(
504 + sanitize_key('mxchat_pdf_status_' . md5($pdf_url)),
505 + array_map('sanitize_text_field', $status_data),
506 + DAY_IN_SECONDS
507 + );
239 508
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);
509 + return __('scheduled', 'mxchat');
243 510
244 - return 'queued';
245 -
246 511 } catch (Exception $e) {
512 + //error_log(sprintf(esc_html__('Error preparing PDF for processing: %s', 'mxchat'), esc_html($e->getMessage())));
247 513 if (file_exists($pdf_path)) {
248 514 wp_delete_file($pdf_path);
249 515 }
250 - return $e->getMessage();
516 + return false;
251 517 }
252 518 }
253 519
254 -/**
255 - * Handle direct PDF file upload from the knowledge base page
256 - */
257 -public function mxchat_handle_pdf_file_submission() {
258 - if (!isset($_POST['submit_pdf_file']) || !current_user_can('manage_options')) {
259 - wp_die(esc_html__('Unauthorized access', 'mxchat'));
260 - }
520 +public function mxchat_process_pdf_pages_cron($pdf_path, $pdf_url, $total_pages, $batch_size, $batch_pause) {
521 + // Validate inputs
522 + $pdf_path = sanitize_text_field($pdf_path);
523 + $pdf_url = esc_url_raw($pdf_url);
524 + $total_pages = absint($total_pages);
525 + $batch_size = absint($batch_size);
526 + $batch_pause = absint($batch_pause);
261 527
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 -
345 528 try {
346 - $total_pages = $this->mxchat_validate_and_count_pdf_pages($pdf_path);
347 -
348 - if ($total_pages === false || $total_pages < 1) {
349 - throw new Exception(__('Invalid PDF: Unable to parse or no pages found', 'mxchat'));
529 + if (!file_exists($pdf_path)) {
530 + throw new Exception(sprintf('PDF file not found at path: %s', esc_html($pdf_path)));
350 531 }
351 532
352 - // Use original filename as the source identifier
353 - $source_label = 'upload://' . $original_filename;
533 + $parser = new \Smalot\PdfParser\Parser();
534 + $pdf = $parser->parseFile($pdf_path);
535 + $pages = $pdf->getPages();
354 536
355 - $queue_id = 'pdf_' . md5($source_label . time());
537 + // Get current progress
538 + $status_key = sanitize_key('mxchat_pdf_status_' . md5($pdf_url));
539 + $status = get_transient($status_key);
356 540
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' => $total_pages,
364 - );
541 + if (!$status || !is_array($status)) {
542 + throw new Exception('Invalid status data retrieved from transient');
365 543 }
366 544
367 - $queued_count = $this->mxchat_add_to_queue($queue_id, 'pdf_page', $pages, $bot_id);
545 + // Initialize failed pages list if it doesn't exist
546 + if (!isset($status['failed_pages_list']) || !is_array($status['failed_pages_list'])) {
547 + $status['failed_pages_list'] = [];
548 + }
368 549
369 - if ($queued_count === 0) {
370 - wp_delete_file($pdf_path);
371 - throw new Exception(__('Failed to add PDF pages to processing queue', 'mxchat'));
550 + $start_page = absint($status['processed_pages']);
551 + $end_page = min($start_page + $batch_size, $total_pages);
552 + $options = get_option('mxchat_options');
553 +
554 + if (empty($options['api_key'])) {
555 + throw new Exception('API key is missing or invalid');
372 556 }
373 557
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'));
558 + $successful_pages = 0;
559 + $failed_pages = 0;
380 560
381 - set_transient('mxchat_active_queue_pdf', $queue_id, DAY_IN_SECONDS);
382 - set_transient('mxchat_last_pdf_url', $source_label, DAY_IN_SECONDS);
561 + for ($i = $start_page; $i < $end_page; $i++) {
562 + $page_number = $i + 1;
563 + $max_retries = 3;
564 + $retry_count = 0;
565 + $page_processed = false;
566 + $last_error = '';
383 567
384 - set_transient('mxchat_admin_notice_success',
385 - sprintf(
386 - esc_html__('PDF "%s" (%d pages) queued for processing. Processing will start automatically.', 'mxchat'),
387 - esc_html($original_filename),
388 - $total_pages
389 - ),
390 - 30
391 - );
568 + while (!$page_processed && $retry_count < $max_retries) {
569 + try {
570 + $text = $pages[$i]->getText();
571 +
572 + if (empty($text)) {
573 + throw new Exception("Empty text on page {$page_number}");
574 + }
575 +
576 + $sanitized_content = $this->mxchat_sanitize_content_for_api($text);
392 577
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 - }
578 + if (empty($sanitized_content)) {
579 + throw new Exception("No valid content after sanitization on page {$page_number}");
580 + }
402 581
403 - wp_safe_redirect(esc_url($redirect_url));
404 - exit;
405 -}
582 + $embedding_vector = $this->mxchat_generate_embedding($sanitized_content);
583 +
584 + if (is_string($embedding_vector)) {
585 + throw new Exception("Embedding generation failed: " . $embedding_vector);
586 + }
587 +
588 + if (!is_array($embedding_vector)) {
589 + throw new Exception("Embedding generation returned unexpected result type: " . gettype($embedding_vector));
590 + }
591 +
592 + $metadata = array(
593 + 'document_type' => 'pdf',
594 + 'total_pages' => $total_pages,
595 + 'current_page' => $page_number,
596 + 'prev_page' => $i > 0 ? $i : null,
597 + 'next_page' => $i < ($total_pages - 1) ? $i + 2 : null,
598 + 'source_url' => $pdf_url
599 + );
406 600
407 -/**
408 - * Validate PDF and count pages with multiple parser attempts
409 - */
410 -private function mxchat_validate_and_count_pdf_pages($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 - }
601 + $content_with_metadata = wp_json_encode($metadata) . "\n---\n" . $sanitized_content;
602 + $page_url = esc_url($pdf_url . "#page=" . $page_number);
426 603
427 - // Method 2: Try with pdfinfo command (if available)
428 - if (function_exists('shell_exec') && !$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;
604 + $db_result = MxChat_Utils::submit_content_to_db($content_with_metadata, $page_url, $options['api_key']);
605 +
606 + if (is_wp_error($db_result)) {
607 + throw new Exception("Database submission failed: " . $db_result->get_error_message());
608 + }
609 +
610 + // Success!
611 + $page_processed = true;
612 + $successful_pages++;
613 +
614 + } catch (Exception $e) {
615 + $retry_count++;
616 + $last_error = $e->getMessage();
617 +
618 + //error_log("PDF page {$page_number} failed (attempt {$retry_count}/{$max_retries}): " . $last_error);
619 +
620 + if ($retry_count < $max_retries) {
621 + // Wait before retry (exponential backoff: 1s, 2s, 4s)
622 + sleep(pow(2, $retry_count - 1));
623 + }
438 624 }
439 625 }
440 - } catch (Exception $e) {
441 - //error_log('pdfinfo command failed: ' . $e->getMessage());
442 - }
443 - }
444 626
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 $page_count;
627 + // If page still not processed after all retries, mark as failed
628 + if (!$page_processed) {
629 + $failed_pages++;
630 + $status['failed_pages_list'][] = [
631 + 'page' => $page_number,
632 + 'error' => $last_error,
633 + 'time' => time(),
634 + 'retries' => $max_retries
635 + ];
636 +
637 + // Limit failed pages list to prevent memory issues
638 + if (count($status['failed_pages_list']) > 50) {
639 + $status['failed_pages_list'] = array_slice($status['failed_pages_list'], -50);
640 + }
641 +
642 + //error_log("PDF page {$page_number} permanently failed after {$max_retries} attempts: " . $last_error);
461 643 }
644 +
645 + // Update progress
646 + $status['processed_pages'] = absint($page_number);
647 + $status['last_update'] = time();
648 + $status['failed_pages'] = absint($status['failed_pages'] ?? 0) + ($page_processed ? 0 : 1);
462 649
463 - // Clean up repaired file if it didn't work
464 - unlink($repaired_path);
650 + set_transient($status_key, array_map('sanitize_text_field', $status), DAY_IN_SECONDS);
465 651 }
466 - } catch (Exception $e) {
467 - //error_log('PDF repair attempt failed: ' . $e->getMessage());
468 - }
469 652
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 $repaired_path;
653 + // Schedule next batch if needed
654 + if ($end_page < $total_pages) {
655 + wp_schedule_single_event(time() + $batch_pause, 'mxchat_process_pdf_pages', array(
656 + 'pdf_path' => $pdf_path,
657 + 'pdf_url' => $pdf_url,
658 + 'total_pages' => $total_pages,
659 + 'batch_size' => $batch_size,
660 + 'batch_pause' => $batch_pause
661 + ));
662 + } else {
663 + // Processing complete
664 + $status['status'] = 'complete';
665 + $status['processed_pages'] = $total_pages;
666 +
667 + // Add completion summary
668 + $status['completion_summary'] = [
669 + 'total_pages' => $total_pages,
670 + 'successful_pages' => $total_pages - absint($status['failed_pages'] ?? 0),
671 + 'failed_pages' => absint($status['failed_pages'] ?? 0),
672 + 'completion_time' => current_time('mysql')
673 + ];
674 +
675 + // Save the completed status (don't delete it - let user dismiss manually)
676 + set_transient($status_key, array_map('sanitize_text_field', $status), DAY_IN_SECONDS);
677 +
678 + // Clean up the temporary PDF file
679 + if (file_exists($pdf_path)) {
680 + wp_delete_file($pdf_path);
513 681 }
682 +
683 + // DON'T delete the status transients here - let user dismiss manually
514 684 }
515 685
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;
686 + } catch (\Exception $e) {
687 + //error_log(sprintf('[MXCHAT-PDF] Error processing PDF: %s', $e->getMessage()));
688 +
689 + $status_key = sanitize_key('mxchat_pdf_status_' . md5($pdf_url));
690 + $status = get_transient($status_key);
691 +
692 + if (!$status || !is_array($status)) {
693 + $status = array(
694 + 'total_pages' => $total_pages,
695 + 'processed_pages' => 0,
696 + 'status' => 'error',
697 + 'error' => sanitize_text_field($e->getMessage()),
698 + 'last_update' => time()
699 + );
700 + } else {
701 + $status['status'] = 'error';
702 + $status['error'] = sanitize_text_field($e->getMessage());
703 + $status['last_update'] = time();
524 704 }
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 (!$content) {
540 - return 0;
705 +
706 + set_transient($status_key, array_map('sanitize_text_field', $status), DAY_IN_SECONDS);
707 +
708 + if (file_exists($pdf_path)) {
709 + wp_delete_file($pdf_path);
541 710 }
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: ' . $e->getMessage());
562 711 }
563 -
564 - return 0;
565 712 }
566 713
567 -
568 714 public function mxchat_save_inline_prompt() {
569 715 // DEBUG: Log what we're receiving
570 - //error_log('=== MXCHAT DEBUG ===');
571 - //error_log('POST data: ' . print_r($_POST, true));
572 - //error_log('Nonce from POST: ' . ($_POST['_ajax_nonce'] ?? 'NOT FOUND'));
716 + error_log('=== MXCHAT DEBUG ===');
717 + error_log('POST data: ' . print_r($_POST, true));
718 + error_log('Nonce from POST: ' . ($_POST['_ajax_nonce'] ?? 'NOT FOUND'));
573 719
574 720 // Check for nonce security
575 721 check_ajax_referer('mxchat_save_inline_nonce', '_ajax_nonce');
576 722
577 723 // 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 -}
724 + error_log('Nonce verification PASSED');
626 725
726 + // Verify permissions
727 + if (!current_user_can('manage_options')) {
728 + wp_send_json_error(esc_html__('Permission denied.', 'mxchat'));
729 + return;
730 + }
627 731
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');
732 + global $wpdb;
733 + $table_name = $wpdb->prefix . 'mxchat_system_prompt_content';
634 734
635 - if ( ! current_user_can('manage_options') ) {
636 - wp_send_json_error( array( 'message' => 'Permission denied.' ) );
637 - }
735 + // Validate and sanitize input data
736 + $prompt_id = isset($_POST['id']) ? absint($_POST['id']) : 0;
737 + $article_content = isset($_POST['article_content']) ? sanitize_textarea_field($_POST['article_content']) : '';
738 + $article_url = isset($_POST['article_url']) ? esc_url_raw($_POST['article_url']) : '';
638 739
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($_POST['bot_id']) : 'default';
740 + if ($prompt_id > 0 && !empty($article_content)) {
741 + // Re-generate the embedding vector for the updated content
742 + $embedding_vector = $this->mxchat_generate_embedding($article_content);
643 743
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, $entry_id );
650 - }
744 + if (is_array($embedding_vector)) {
745 + // Serialize the embedding vector before storing it
746 + $embedding_vector_serialized = serialize($embedding_vector);
651 747
652 - if ( is_wp_error( $content ) ) {
653 - wp_send_json_error( array( 'message' => $content->get_error_message() ) );
654 - }
748 + // Update the prompt in the database
749 + $updated = $wpdb->update(
750 + $table_name,
751 + array(
752 + 'article_content' => $article_content,
753 + 'embedding_vector' => $embedding_vector_serialized,
754 + 'source_url' => $article_url,
755 + ),
756 + array('id' => $prompt_id),
757 + array('%s', '%s', '%s'),
758 + array('%d')
759 + );
655 760
656 - wp_send_json_success( $content );
657 -}
761 + if ($updated !== false) {
762 + wp_send_json_success();
763 + } else {
764 + wp_send_json_error(esc_html__('Database update failed.', 'mxchat'));
765 + }
766 + } else {
767 + wp_send_json_error(esc_html__('Embedding generation failed.', 'mxchat'));
768 + }
769 + } else {
770 + wp_send_json_error(esc_html__('Invalid data.', 'mxchat'));
771 + }
772 + }
658 773
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';
665 774
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' => $result->get_error_message() ) );
1106 - }
1107 -
1108 - wp_send_json_success( array( 'message' => 'Content saved and re-embedded successfully.' ) );
1109 -}
1110 -
1111 775 public function mxchat_get_pdf_processing_status($pdf_url) {
1112 776 $pdf_url = esc_url_raw($pdf_url);
1113 777 $status = get_transient(sanitize_key('mxchat_pdf_status_' . md5($pdf_url)));
1114 778
@@ -1151,18 +815,24 @@
1151 815 }
1152 816
1153 817
1154 818 public function mxchat_handle_sitemap_submission() {
819 + // Start logging the submission process
820 + //error_log('[MXCHAT-URL] ===== Starting URL submission process =====');
821 +
1155 822 // Check if the form was submitted and verify permissions
1156 823 if (!isset($_POST['submit_sitemap']) || !current_user_can('manage_options')) {
824 + //error_log('[MXCHAT-URL] Error: Unauthorized access or form not submitted properly');
1157 825 wp_die(esc_html__('Unauthorized access', 'mxchat'));
1158 826 }
1159 827
1160 828 // Verify nonce
829 + //error_log('[MXCHAT-URL] Verifying nonce');
1161 830 check_admin_referer('mxchat_submit_sitemap_action', 'mxchat_submit_sitemap_nonce');
1162 831
1163 832 // Validate URL
1164 833 if (!isset($_POST['sitemap_url']) || empty($_POST['sitemap_url'])) {
834 + //error_log('[MXCHAT-URL] Error: Empty or missing URL');
1165 835 set_transient('mxchat_admin_notice_error',
1166 836 esc_html__('Please provide a valid URL.', 'mxchat'),
1167 837 30
1168 838 );
@@ -1170,28 +840,12 @@
1170 840 exit;
1171 841 }
1172 842
1173 843 $submitted_url = esc_url_raw($_POST['sitemap_url']);
844 + //error_log('[MXCHAT-URL] Processing URL: ' . $submitted_url);
1174 845
1175 - // Convert Google Drive sharing URLs to direct download URLs
1176 - if ( strpos($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');
846 + // Validate API key first
847 + $options = get_option('mxchat_options');
1194 848 $selected_model = $options['embedding_model'] ?? 'text-embedding-ada-002';
1195 849
1196 850 if (strpos($selected_model, 'voyage') === 0) {
1197 851 $api_key = $options['voyage_api_key'] ?? '';
@@ -1208,26 +862,21 @@
1208 862 $error_message = sprintf(
1209 863 esc_html__('%s API key is not configured. Please add your API key in the settings before submitting content.', 'mxchat'),
1210 864 $provider_name
1211 865 );
866 + //error_log('[MXCHAT-URL] Error: ' . $error_message);
1212 867 set_transient('mxchat_admin_notice_error', $error_message, 30);
868 + //error_log('[MXCHAT-URL] Set error transient: ' . $error_message);
1213 869 wp_safe_redirect(esc_url(admin_url('admin.php?page=mxchat-prompts')));
1214 870 exit;
1215 871 }
1216 872
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 - ));
873 + //error_log('[MXCHAT-URL] Fetching URL content');
874 + $response = wp_remote_get($submitted_url, array('timeout' => 30));
1227 875
1228 876 if (is_wp_error($response) || wp_remote_retrieve_response_code($response) !== 200) {
1229 877 $error_message = is_wp_error($response) ? $response->get_error_message() : 'HTTP Status: ' . wp_remote_retrieve_response_code($response);
878 + //error_log('[MXCHAT-URL] Error fetching URL: ' . $error_message);
1230 879 set_transient('mxchat_admin_notice_error',
1231 880 sprintf(
1232 881 esc_html__('Failed to fetch the URL: %s', 'mxchat'),
1233 882 esc_html($error_message)
@@ -1238,11 +887,13 @@
1238 887 exit;
1239 888 }
1240 889
1241 890 $content_type = wp_remote_retrieve_header($response, 'content-type');
891 + //error_log('[MXCHAT-URL] Content type: ' . $content_type);
1242 892 $body_content = wp_remote_retrieve_body($response);
1243 893
1244 894 if (empty($body_content)) {
895 + //error_log('[MXCHAT-URL] Error: Empty response body');
1245 896 set_transient('mxchat_admin_notice_error',
1246 897 esc_html__('Empty response received from URL.', 'mxchat'),
1247 898 30
1248 899 );
@@ -1248,21 +899,29 @@
1248 899 );
1249 900 wp_safe_redirect(esc_url(admin_url('admin.php?page=mxchat-prompts')));
1250 901 exit;
1251 902 }
903 + //error_log('[MXCHAT-URL] Retrieved body content length: ' . strlen($body_content) . ' bytes');
1252 904
1253 905 // Handle PDF URL
1254 906 if ($this->mxchat_is_pdf_url($submitted_url, $response)) {
1255 - $result = $this->mxchat_handle_pdf_for_knowledge_base($submitted_url, $response, $bot_id);
907 + //error_log('[MXCHAT-URL] Detected PDF URL, handling PDF for knowledge base');
908 + $result = $this->mxchat_handle_pdf_for_knowledge_base($submitted_url, $response);
909 + //error_log('[MXCHAT-URL] PDF handling result: ' . $result);
1256 910
1257 - if ($result === 'queued') {
1258 - set_transient('mxchat_admin_notice_success',
1259 - esc_html__('PDF queued for processing. Processing will start automatically.', 'mxchat'),
911 + if ($result === 'scheduled') {
912 + set_transient(
913 + 'mxchat_last_pdf_url',
914 + sanitize_text_field($submitted_url),
915 + DAY_IN_SECONDS
916 + );
917 + set_transient('mxchat_admin_notice_info',
918 + esc_html__('PDF processing has started in the background. You can check the progress in the Knowledge Base section.', 'mxchat'),
1260 919 30
1261 920 );
1262 921 } else {
1263 922 set_transient('mxchat_admin_notice_error',
1264 - esc_html__('Failed to queue PDF processing: ', 'mxchat') . esc_html($result),
923 + esc_html__('Failed to start PDF processing: ', 'mxchat') . esc_html($result),
1265 924 30
1266 925 );
1267 926 }
1268 927
@@ -1271,8 +930,9 @@
1271 930 }
1272 931
1273 932 // Handle Sitemap XML
1274 933 if (strpos($content_type, 'xml') !== false || strpos($body_content, '<urlset') !== false) {
934 + //error_log('[MXCHAT-URL] Detected XML content, processing as sitemap');
1275 935 libxml_use_internal_errors(true);
1276 936 $xml = simplexml_load_string($body_content);
1277 937 $xml_errors = libxml_get_errors();
1278 938 libxml_clear_errors();
@@ -1277,8 +937,15 @@
1277 937 $xml_errors = libxml_get_errors();
1278 938 libxml_clear_errors();
1279 939
1280 940 if ($xml === false || !empty($xml_errors)) {
941 + //error_log('[MXCHAT-URL] Error: Invalid XML format');
942 + if (!empty($xml_errors)) {
943 + foreach ($xml_errors as $error) {
944 + //error_log('[MXCHAT-URL] XML Error: ' . $error->message);
945 + }
946 + }
947 +
1281 948 set_transient('mxchat_admin_notice_error',
1282 949 esc_html__('Invalid sitemap XML. Please provide a valid sitemap.', 'mxchat'),
1283 950 30
1284 951 );
@@ -1285,18 +952,27 @@
1285 952 wp_safe_redirect(esc_url(admin_url('admin.php?page=mxchat-prompts')));
1286 953 exit;
1287 954 }
1288 955
1289 - $result = $this->mxchat_handle_sitemap_for_knowledge_base($xml, $submitted_url, $bot_id);
956 + //error_log('[MXCHAT-URL] Valid XML found, handling sitemap for knowledge base');
957 + $result = $this->mxchat_handle_sitemap_for_knowledge_base($xml, $submitted_url);
958 + //error_log('[MXCHAT-URL] Sitemap handling result: ' . $result);
1290 959
1291 - if ($result === 'queued') {
1292 - set_transient('mxchat_admin_notice_success',
1293 - esc_html__('Sitemap queued for processing. Processing will start automatically.', 'mxchat'),
960 + if ($result === 'scheduled') {
961 + set_transient(
962 + 'mxchat_last_sitemap_url',
963 + sanitize_text_field($submitted_url),
964 + DAY_IN_SECONDS
965 + );
966 + set_transient('mxchat_admin_notice_info',
967 + esc_html__('Sitemap processing has started in the background. You can check the progress in the Knowledge Base section.', 'mxchat'),
1294 968 30
1295 969 );
1296 970 } else {
971 + // Return to the admin page without a redirect for better error display
972 + // The error is already stored in the sitemap status transient
1297 973 set_transient('mxchat_admin_notice_error',
1298 - esc_html__('Failed to queue sitemap processing. Please check the status below for details.', 'mxchat'),
974 + esc_html__('Failed to start sitemap processing. Please check the status below for details.', 'mxchat'),
1299 975 30
1300 976 );
1301 977 }
1302 978
@@ -1303,50 +979,127 @@
1303 979 wp_safe_redirect(esc_url(admin_url('admin.php?page=mxchat-prompts')));
1304 980 exit;
1305 981 }
1306 982
1307 - // Handle Regular URL (single page)
983 + // Handle Regular URL
984 + //error_log('[MXCHAT-URL] Processing as regular webpage');
1308 985 $page_content = $this->mxchat_extract_main_content($body_content);
986 + //error_log('[MXCHAT-URL] Extracted content length: ' . strlen($page_content) . ' bytes');
987 +
1309 988 $sanitized_content = $this->mxchat_sanitize_content_for_api($page_content);
989 + //error_log('[MXCHAT-URL] Sanitized content length: ' . strlen($sanitized_content) . ' bytes');
1310 990
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 991 if (empty($sanitized_content)) {
992 + //error_log('[MXCHAT-URL] Error: No valid content after sanitization');
993 +
994 + // Set both transients - the error notice and the URL status
1317 995 set_transient('mxchat_admin_notice_error',
1318 996 esc_html__('No valid content found on the provided URL.', 'mxchat'),
1319 997 30
1320 998 );
999 +
1000 + // Set URL status transient
1001 + set_transient('mxchat_single_url_status', [
1002 + 'url' => $submitted_url,
1003 + 'timestamp' => current_time('mysql'),
1004 + 'status' => 'failed',
1005 + 'error' => esc_html__('No valid content found on the provided URL.', 'mxchat')
1006 + ], DAY_IN_SECONDS);
1007 +
1321 1008 wp_safe_redirect(esc_url(admin_url('admin.php?page=mxchat-prompts')));
1322 1009 exit;
1323 1010 }
1324 1011
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 - );
1012 + //error_log('[MXCHAT-URL] Generating embedding for content');
1013 + $embedding_vector = $this->mxchat_generate_embedding($sanitized_content);
1335 1014
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());
1015 + // Check if embedding_vector is a string (error message)
1016 + if (is_string($embedding_vector)) {
1017 + //error_log('[MXCHAT-URL] Error generating embedding: ' . $embedding_vector);
1018 + $error_message = esc_html__('Failed to generate embedding: ', 'mxchat') . esc_html($embedding_vector);
1019 + //error_log('[MXCHAT-URL] Setting error transient: ' . $error_message);
1020 +
1021 + // Set both transients
1338 1022 set_transient('mxchat_admin_notice_error', $error_message, 30);
1339 - } else {
1023 +
1024 + // Set URL status transient
1025 + set_transient('mxchat_single_url_status', [
1026 + 'url' => $submitted_url,
1027 + 'timestamp' => current_time('mysql'),
1028 + 'status' => 'failed',
1029 + 'error' => $error_message
1030 + ], DAY_IN_SECONDS);
1031 +
1032 + wp_safe_redirect(esc_url(admin_url('admin.php?page=mxchat-prompts')));
1033 + exit;
1034 + }
1035 +
1036 + if (is_array($embedding_vector)) {
1037 + //error_log('[MXCHAT-URL] Successfully generated embedding with ' . count($embedding_vector) . ' dimensions');
1038 +
1039 + $db_result = MxChat_Utils::submit_content_to_db(
1040 + $sanitized_content,
1041 + $submitted_url,
1042 + $api_key
1043 + );
1044 +
1045 + if (is_wp_error($db_result)) {
1046 + //error_log('[MXCHAT-URL] Error: Failed to store content in database: ' . $db_result->get_error_message());
1047 + $error_message = esc_html__('Failed to store content in database: ', 'mxchat') . esc_html($db_result->get_error_message());
1048 + //error_log('[MXCHAT-URL] Setting error transient: ' . $error_message);
1049 +
1050 + // Set both transients
1051 + set_transient('mxchat_admin_notice_error', $error_message, 30);
1052 +
1053 + // Set URL status transient
1054 + set_transient('mxchat_single_url_status', [
1055 + 'url' => $submitted_url,
1056 + 'timestamp' => current_time('mysql'),
1057 + 'status' => 'failed',
1058 + 'error' => $error_message
1059 + ], DAY_IN_SECONDS);
1060 +
1061 + wp_safe_redirect(esc_url(admin_url('admin.php?page=mxchat-prompts')));
1062 + exit;
1063 + }
1064 +
1065 + //error_log('[MXCHAT-URL] Successfully stored content in database');
1340 1066 $success_message = esc_html__('URL content successfully submitted!', 'mxchat');
1067 + //error_log('[MXCHAT-URL] Setting success transient: ' . $success_message);
1068 +
1069 + // Set both transients
1341 1070 set_transient('mxchat_admin_notice_success', $success_message, 30);
1071 +
1072 + // Set URL status transient with success
1073 + set_transient('mxchat_single_url_status', [
1074 + 'url' => $submitted_url,
1075 + 'timestamp' => current_time('mysql'),
1076 + 'status' => 'complete',
1077 + 'content_length' => strlen($sanitized_content),
1078 + 'embedding_dimensions' => count($embedding_vector)
1079 + ], DAY_IN_SECONDS);
1080 +
1081 + } else {
1082 + //error_log('[MXCHAT-URL] Error: Failed to generate embedding. Unexpected result type: ' . gettype($embedding_vector));
1083 + $error_message = esc_html__('Failed to generate embedding: Unexpected result type. Please check your API key and try again.', 'mxchat');
1084 + //error_log('[MXCHAT-URL] Setting error transient: ' . $error_message);
1085 +
1086 + // Set both transients
1087 + set_transient('mxchat_admin_notice_error', $error_message, 30);
1088 +
1089 + // Set URL status transient
1090 + set_transient('mxchat_single_url_status', [
1091 + 'url' => $submitted_url,
1092 + 'timestamp' => current_time('mysql'),
1093 + 'status' => 'failed',
1094 + 'error' => $error_message
1095 + ], DAY_IN_SECONDS);
1342 1096 }
1343 1097
1098 + //error_log('[MXCHAT-URL] ===== Completed URL submission process =====');
1344 1099 wp_safe_redirect(esc_url(admin_url('admin.php?page=mxchat-prompts')));
1345 1100 exit;
1346 1101 }
1347 -
1348 -
1349 1102 public function mxchat_get_single_url_status() {
1350 1103 $status = get_transient('mxchat_single_url_status');
1351 1104 if (!$status) {
1352 1105 return null;
@@ -1358,11 +1111,13 @@
1358 1111 }
1359 1112
1360 1113 return $status;
1361 1114 }
1362 -
1363 -public function mxchat_handle_sitemap_for_knowledge_base($xml, $sitemap_url, $bot_id = 'default') {
1115 +public function mxchat_handle_sitemap_for_knowledge_base($xml, $sitemap_url) {
1116 + // Clear any single URL status when starting sitemap processing
1117 + delete_transient('mxchat_single_url_status');
1364 1118 if (!current_user_can('manage_options')) {
1119 + //error_log(esc_html__('Unauthorized sitemap processing attempt', 'mxchat'));
1365 1120 return false;
1366 1121 }
1367 1122
1368 1123 try {
@@ -1371,96 +1126,331 @@
1371 1126 if (!$xml || !is_object($xml)) {
1372 1127 throw new Exception(__('Invalid XML object provided', 'mxchat'));
1373 1128 }
1374 1129
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');
1130 + // Add embedding validation before processing
1131 + // Test embedding with a small sample text to verify API key is working
1132 + $test_result = $this->mxchat_generate_embedding("This is a test to verify the embedding API key is working.");
1378 1133
1379 - // Test the embedding API before processing
1380 - $test_phrase = "Test embedding generation for MxChat";
1381 - $test_result = $this->mxchat_generate_embedding($test_phrase, $bot_id);
1382 -
1134 + // Check if test_result is a string (error message) rather than an array (valid embedding)
1383 1135 if (is_string($test_result)) {
1136 + //error_log('[MXCHAT-URL] Embedding API validation failed: ' . $test_result);
1137 +
1138 + // Store the error in the status transient so it can be displayed later
1139 + $status_data = array(
1140 + 'total_urls' => 0,
1141 + 'processed_urls' => 0,
1142 + 'status' => 'error',
1143 + 'error' => __('Embedding API validation failed: ', 'mxchat') . $test_result,
1144 + 'last_update' => time()
1145 + );
1146 +
1147 + set_transient(
1148 + sanitize_key('mxchat_sitemap_status_' . md5($sitemap_url)),
1149 + array_map('sanitize_text_field', $status_data),
1150 + DAY_IN_SECONDS
1151 + );
1152 +
1384 1153 throw new Exception(__('Embedding API validation failed: ', 'mxchat') . $test_result);
1385 1154 }
1386 1155
1156 + // Make sure it's an array (valid embedding)
1387 1157 if (!is_array($test_result)) {
1158 + //error_log('[MXCHAT-URL] Embedding API returned unexpected result type: ' . gettype($test_result));
1388 1159 throw new Exception(__('Embedding API returned unexpected result type. Please check your configuration.', 'mxchat'));
1389 1160 }
1390 1161
1391 - // Extract URLs from sitemap
1392 - $urls = array();
1162 + $urls = [];
1393 1163 foreach ($xml->url as $url_element) {
1394 1164 $url = esc_url_raw((string)$url_element->loc);
1395 1165 if ($url) {
1396 - $urls[] = array('url' => $url);
1166 + $urls[] = $url;
1397 1167 }
1398 1168 }
1399 1169
1400 - $total_urls = count($urls);
1170 + $total_urls = absint(count($urls));
1401 1171
1402 1172 if ($total_urls < 1) {
1403 1173 throw new Exception(__('No valid URLs found in sitemap', 'mxchat'));
1404 1174 }
1405 1175
1406 - // Create unique queue ID
1407 - $queue_id = 'sitemap_' . md5($sitemap_url . time());
1176 + wp_schedule_single_event(time(), 'mxchat_process_sitemap_urls', array(
1177 + 'urls' => $urls,
1178 + 'sitemap_url' => $sitemap_url,
1179 + 'total_urls' => $total_urls,
1180 + 'batch_size' => absint(10),
1181 + 'batch_pause' => absint(5)
1182 + ));
1408 1183
1409 - // Add URLs to queue
1410 - $queued_count = $this->mxchat_add_to_queue($queue_id, 'url', $urls, $bot_id);
1184 + $status_data = array(
1185 + 'total_urls' => $total_urls,
1186 + 'processed_urls' => 0,
1187 + 'status' => 'processing',
1188 + 'last_update' => time()
1189 + );
1411 1190
1412 - if ($queued_count === 0) {
1413 - throw new Exception(__('Failed to add URLs to processing queue', 'mxchat'));
1414 - }
1191 + set_transient(
1192 + sanitize_key('mxchat_sitemap_status_' . md5($sitemap_url)),
1193 + array_map('sanitize_text_field', $status_data),
1194 + DAY_IN_SECONDS
1195 + );
1415 1196
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'));
1197 + return __('scheduled', 'mxchat');
1422 1198
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) {
1199 + } catch (\Exception $e) {
1430 1200 $error_message = $e->getMessage();
1431 1201 //error_log(sprintf(esc_html__('Error preparing sitemap for processing: %s', 'mxchat'), esc_html($error_message)));
1432 1202
1203 + // Store the sitemap URL and error in transients so they can be displayed
1204 + set_transient(
1205 + 'mxchat_last_sitemap_url',
1206 + sanitize_text_field($sitemap_url),
1207 + DAY_IN_SECONDS
1208 + );
1209 +
1210 + $status_data = array(
1211 + 'total_urls' => 0,
1212 + 'processed_urls' => 0,
1213 + 'status' => 'error',
1214 + 'error' => $error_message,
1215 + 'last_update' => time()
1216 + );
1217 +
1218 + set_transient(
1219 + sanitize_key('mxchat_sitemap_status_' . md5($sitemap_url)),
1220 + array_map('sanitize_text_field', $status_data),
1221 + DAY_IN_SECONDS
1222 + );
1223 +
1433 1224 return $error_message;
1434 1225 }
1435 -
1436 1226 }
1437 1227
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 : $content;
1228 +public function mxchat_process_sitemap_urls_cron($urls, $sitemap_url, $total_urls, $batch_size, $batch_pause) {
1229 + // Validate inputs
1230 + $sitemap_url = esc_url_raw($sitemap_url);
1231 + $total_urls = absint($total_urls);
1232 + $batch_size = absint($batch_size);
1233 + $batch_pause = absint($batch_pause);
1234 +
1235 + if (!is_array($urls) || empty($urls)) {
1236 + return;
1237 + }
1238 +
1239 + try {
1240 + $status_key = sanitize_key('mxchat_sitemap_status_' . md5($sitemap_url));
1241 + $status = get_transient($status_key);
1242 +
1243 + if (!$status || !is_array($status)) {
1244 + throw new Exception('Invalid status data retrieved from transient');
1245 + }
1246 +
1247 + // Initialize failed_urls array if it doesn't exist
1248 + if (!isset($status['failed_urls_list']) || !is_array($status['failed_urls_list'])) {
1249 + $status['failed_urls_list'] = [];
1250 + }
1251 +
1252 + $start_url = absint($status['processed_urls']);
1253 + $end_url = min($start_url + $batch_size, $total_urls);
1254 +
1255 + // Track batch statistics
1256 + $batch_stats = [
1257 + 'processed' => 0,
1258 + 'failed' => 0,
1259 + 'last_error' => '',
1260 + 'embedding_errors' => 0
1261 + ];
1262 +
1263 + // Check embedding configuration with first URL (only on first batch)
1264 + if ($start_url === 0) {
1265 + $test_url = esc_url_raw($urls[0]);
1266 + $test_response = wp_remote_get($test_url);
1267 +
1268 + if (!is_wp_error($test_response) && wp_remote_retrieve_response_code($test_response) === 200) {
1269 + $test_html = wp_remote_retrieve_body($test_response);
1270 + $test_content = $this->mxchat_extract_main_content($test_html);
1271 + $test_sanitized = $this->mxchat_sanitize_content_for_api($test_content);
1272 +
1273 + if (!empty($test_sanitized)) {
1274 + $test_embedding = $this->mxchat_generate_embedding($test_sanitized);
1275 +
1276 + if (is_string($test_embedding)) {
1277 + throw new Exception('Embedding generation failed: ' . $test_embedding);
1278 + }
1279 +
1280 + if (!is_array($test_embedding)) {
1281 + throw new Exception('Embedding generation returned unexpected result type: ' . gettype($test_embedding));
1282 + }
1283 + }
1284 + }
1285 + }
1286 +
1287 + for ($i = $start_url; $i < $end_url; $i++) {
1288 + $page_url = esc_url_raw($urls[$i]);
1289 + $max_retries = 3;
1290 + $retry_count = 0;
1291 + $url_processed = false;
1292 + $last_error = '';
1293 +
1294 + while (!$url_processed && $retry_count < $max_retries) {
1295 + try {
1296 + // Attempt to fetch the URL
1297 + $page_response = wp_remote_get($page_url, array('timeout' => 30));
1298 +
1299 + if (is_wp_error($page_response)) {
1300 + throw new Exception('HTTP request failed: ' . $page_response->get_error_message());
1301 + }
1302 +
1303 + $response_code = wp_remote_retrieve_response_code($page_response);
1304 + if ($response_code !== 200) {
1305 + throw new Exception('HTTP Status: ' . $response_code);
1306 + }
1307 +
1308 + $page_html = wp_remote_retrieve_body($page_response);
1309 +
1310 + if (empty($page_html)) {
1311 + throw new Exception('Empty response body');
1312 + }
1313 +
1314 + $page_content = $this->mxchat_extract_main_content($page_html);
1315 + $sanitized_content = $this->mxchat_sanitize_content_for_api($page_content);
1316 +
1317 + if (empty($sanitized_content)) {
1318 + throw new Exception('No valid content found after processing');
1319 + }
1320 +
1321 + $embedding_vector = $this->mxchat_generate_embedding($sanitized_content);
1322 +
1323 + if (is_string($embedding_vector)) {
1324 + throw new Exception('Embedding generation failed: ' . $embedding_vector);
1325 + }
1326 +
1327 + if (!is_array($embedding_vector)) {
1328 + throw new Exception('Embedding generation returned unexpected result type: ' . gettype($embedding_vector));
1329 + }
1330 +
1331 + // Submit to database
1332 + $options = get_option('mxchat_options');
1333 + $submission_result = MxChat_Utils::submit_content_to_db($sanitized_content, $page_url, $options['api_key']);
1334 +
1335 + if (is_wp_error($submission_result)) {
1336 + throw new Exception('Database submission failed: ' . $submission_result->get_error_message());
1337 + }
1338 +
1339 + // Success!
1340 + $url_processed = true;
1341 + $batch_stats['processed']++;
1342 +
1343 + } catch (Exception $e) {
1344 + $retry_count++;
1345 + $last_error = $e->getMessage();
1346 +
1347 + //error_log("URL {$page_url} failed (attempt {$retry_count}/{$max_retries}): " . $last_error);
1348 +
1349 + // Track embedding errors specifically
1350 + if (strpos($last_error, 'Embedding') !== false) {
1351 + $batch_stats['embedding_errors']++;
1352 + }
1353 +
1354 + if ($retry_count < $max_retries) {
1355 + // Wait before retry (exponential backoff: 1s, 2s, 4s)
1356 + sleep(pow(2, $retry_count - 1));
1357 + }
1358 + }
1359 + }
1360 +
1361 + // If URL still not processed after all retries, mark as failed
1362 + if (!$url_processed) {
1363 + $batch_stats['failed']++;
1364 + $batch_stats['last_error'] = $last_error;
1365 +
1366 + // Add to failed URLs list
1367 + $status['failed_urls_list'][] = [
1368 + 'url' => $page_url,
1369 + 'error' => $last_error,
1370 + 'time' => time(),
1371 + 'retries' => $max_retries
1372 + ];
1373 +
1374 + // Limit the number of failed URLs we store to prevent transient size issues
1375 + if (count($status['failed_urls_list']) > 100) {
1376 + $status['failed_urls_list'] = array_slice($status['failed_urls_list'], -100);
1377 + }
1378 + }
1379 +
1380 + // Update progress
1381 + $status['processed_urls'] = absint($i + 1);
1382 + $status['last_update'] = time();
1383 + $status['failed_urls'] = absint($status['failed_urls'] ?? 0) + ($url_processed ? 0 : 1);
1384 + $status['last_error'] = $batch_stats['last_error'];
1385 +
1386 + set_transient($status_key, $status, DAY_IN_SECONDS);
1387 +
1388 + // If we have too many consecutive embedding errors, stop processing
1389 + if ($batch_stats['embedding_errors'] >= 10) {
1390 + throw new Exception('Too many consecutive embedding failures detected. Please check your API configuration.');
1391 + }
1392 + }
1393 +
1394 + // If all URLs in this batch failed, stop processing
1395 + if ($batch_stats['processed'] === 0 && $batch_stats['failed'] > 0) {
1396 + $status['status'] = 'error';
1397 + $status['error'] = sprintf(
1398 + 'Processing stopped: %d consecutive failures in batch. Last error: %s',
1399 + $batch_stats['failed'],
1400 + $batch_stats['last_error']
1401 + );
1402 + set_transient($status_key, $status, DAY_IN_SECONDS);
1403 + return;
1404 + }
1405 +
1406 + // Update final progress
1407 + $status['processed_urls'] = min($end_url, $total_urls);
1408 + $status['last_update'] = time();
1409 + set_transient($status_key, $status, DAY_IN_SECONDS);
1410 +
1411 + // Check if we've processed all URLs
1412 + if ($end_url >= $total_urls) {
1413 + // All URLs have been processed - mark as complete
1414 + $status['status'] = 'complete';
1415 + $status['processed_urls'] = $total_urls;
1416 +
1417 + // Add completion summary
1418 + $status['completion_summary'] = [
1419 + 'total_urls' => $total_urls,
1420 + 'successful_urls' => $total_urls - absint($status['failed_urls'] ?? 0),
1421 + 'failed_urls' => absint($status['failed_urls'] ?? 0),
1422 + 'completion_time' => current_time('mysql')
1423 + ];
1424 +
1425 + // Save the completed status (don't delete it - let user dismiss manually)
1426 + set_transient($status_key, $status, DAY_IN_SECONDS);
1427 +
1428 + // DON'T delete the status transients here - let user dismiss manually
1429 + } else {
1430 + // Schedule next batch
1431 + wp_schedule_single_event(time() + $batch_pause, 'mxchat_process_sitemap_urls', array(
1432 + 'urls' => $urls,
1433 + 'sitemap_url' => $sitemap_url,
1434 + 'total_urls' => $total_urls,
1435 + 'batch_size' => $batch_size,
1436 + 'batch_pause' => $batch_pause,
1437 + ));
1438 + }
1439 + } catch (\Exception $e) {
1440 + $status['status'] = 'error';
1441 + $status['error'] = $e->getMessage();
1442 + set_transient($status_key, $status, DAY_IN_SECONDS);
1443 + }
1450 1444 }
1451 -
1452 1445 public function mxchat_sanitize_content_for_api($content) {
1453 1446 //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 -
1447 +
1458 1448 // Remove script, style tags, and HTML comments
1459 1449 $content = preg_replace('/<script\b[^>]*>(.*?)<\/script>/is', '', $content);
1460 1450 $content = preg_replace('/<style\b[^>]*>(.*?)<\/style>/is', '', $content);
1461 1451 $content = preg_replace('/<!--(.|\s)*?-->/', '', $content);
1462 -
1452 +
1463 1453 // Remove all HTML tags and decode HTML entities
1464 1454 $content = wp_strip_all_tags($content);
1465 1455 $content = html_entity_decode($content, ENT_QUOTES | ENT_HTML5);
1466 1456
@@ -1518,12 +1508,12 @@
1518 1508 @$dom->loadHTML(mb_convert_encoding($html, 'HTML-ENTITIES', 'UTF-8'), LIBXML_HTML_NOIMPLIED | LIBXML_HTML_NODEFDTD);
1519 1509 $xpath = new DOMXPath($dom);
1520 1510
1521 1511 // For debugging purposes
1522 - $debugEnabled = true; // Set to true to enable debugging output
1512 + $debugEnabled = false; // Set to true to enable debugging output
1523 1513 $debug = function($message) use ($debugEnabled) {
1524 1514 if ($debugEnabled) {
1525 - //error_log('[MXCHAT-EXTRACT-DEBUG] ' . $message);
1515 + //error_log('[MXCHAT-DEBUG] ' . $message);
1526 1516 }
1527 1517 };
1528 1518
1529 1519 // Direct targeting for Gerow theme posts
@@ -1622,25 +1612,18 @@
1622 1612 '//main',
1623 1613 '//div[contains(@class, "content")]'
1624 1614 ];
1625 1615
1626 - // First handle Elementor content - get only leaf widget containers to avoid duplicates
1616 + // First handle Elementor content
1627 1617 $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")]');
1618 + $elementor_widgets = $xpath->query('//div[contains(@class, "elementor-element")]//div[contains(@class, "elementor-widget-container")]');
1630 1619 if ($elementor_widgets && $elementor_widgets->length > 0) {
1631 1620 $debug("Found Elementor widgets");
1632 - $seen_content = array(); // Track seen content to avoid duplicates
1633 1621 $combined_content = '';
1634 1622 foreach ($elementor_widgets as $widget) {
1635 1623 $widget_content = $dom->saveHTML($widget);
1636 1624 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 - }
1625 + $combined_content .= $widget_content;
1643 1626 }
1644 1627 }
1645 1628 if (!empty($combined_content)) {
1646 1629 $debug("Returning Elementor content");
@@ -1652,14 +1635,15 @@
1652 1635 foreach ($selectors as $selector) {
1653 1636 $debug("Trying selector: " . $selector);
1654 1637 $nodes = $xpath->query($selector);
1655 1638 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));
1639 + $debug("Found matches for selector: " . $selector);
1640 + $content = '';
1641 + foreach ($nodes as $node) {
1642 + $content .= $dom->saveHTML($node);
1643 + }
1660 1644 if (!empty($content)) {
1661 - $debug("Returning content from selector: " . $selector . " (first match only)");
1645 + $debug("Returning content from selector: " . $selector);
1662 1646 return $content;
1663 1647 }
1664 1648 }
1665 1649 }
@@ -1683,106 +1667,16 @@
1683 1667 $debug("Returning blog-area section content");
1684 1668 return $content;
1685 1669 }
1686 1670 }
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");
1671 +
1672 + // Fallback: Return the body content if no specific selector matches
1673 + $debug("Using body fallback");
1730 1674 $body = $dom->getElementsByTagName('body');
1731 1675 if ($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 - }
1676 + return $dom->saveHTML($body->item(0));
1783 1677 }
1784 -
1678 +
1785 1679 // Last resort: return the original HTML
1786 1680 $debug("Returning original HTML");
1787 1681 return $html;
1788 1682 } catch (Exception $e) {
@@ -1834,41 +1728,47 @@
1834 1728 try {
1835 1729 // Verify the request
1836 1730 check_ajax_referer('mxchat_status_nonce', 'nonce');
1837 1731
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');
1732 + // Get the status just like in your admin page
1733 + $pdf_url = get_transient('mxchat_last_pdf_url');
1734 + $sitemap_url = get_transient('mxchat_last_sitemap_url');
1841 1735
1842 - $sitemap_status = false;
1843 - $pdf_status = false;
1736 + $pdf_status = $pdf_url ? $this->mxchat_get_pdf_processing_status($pdf_url) : false;
1737 + $sitemap_status = $sitemap_url ? $this->mxchat_get_sitemap_processing_status($sitemap_url) : false;
1844 1738
1845 - // Get sitemap queue status
1846 - if ($sitemap_queue_id) {
1847 - $sitemap_status = $this->mxchat_get_queue_status_data($sitemap_queue_id, 'sitemap');
1739 + // Add the PDF URL to the status object
1740 + if ($pdf_status && $pdf_url) {
1741 + $pdf_status['pdf_url'] = $pdf_url;
1848 1742 }
1849 1743
1850 - // Get PDF queue status
1851 - if ($pdf_queue_id) {
1852 - $pdf_status = $this->mxchat_get_queue_status_data($pdf_queue_id, 'pdf');
1853 - }
1744 + // Set the current PDF URL for the manual batch processing button
1745 + $current_pdf_url = $pdf_url;
1854 1746
1747 + // Check for true processing status, not just presence of status
1855 1748 $is_active_processing =
1856 - ($sitemap_status && $sitemap_status['status'] === 'processing') ||
1857 - ($pdf_status && $pdf_status['status'] === 'processing');
1749 + ($sitemap_status && isset($sitemap_status['status']) && $sitemap_status['status'] === 'processing') ||
1750 + ($pdf_status && isset($pdf_status['status']) && $pdf_status['status'] === 'processing');
1858 1751
1752 + // Get single URL status, but only if no processing is active
1753 + $single_url_status = !$is_active_processing ? $this->mxchat_get_single_url_status() : false;
1754 +
1755 + // REMOVED: Auto-clearing of completed status - now only done via dismiss button
1756 +
1859 1757 // Return JSON response with the status data
1860 1758 wp_send_json(array(
1861 1759 'pdf_status' => $pdf_status,
1862 1760 'sitemap_status' => $sitemap_status,
1761 + 'single_url_status' => $single_url_status,
1863 1762 'is_processing' => $is_active_processing,
1864 - 'sitemap_queue_id' => $sitemap_queue_id,
1865 - 'pdf_queue_id' => $pdf_queue_id
1763 + 'current_pdf_url' => $current_pdf_url
1866 1764 ));
1867 1765
1868 1766 } catch (Exception $e) {
1767 + // Log the error
1869 1768 //error_log('MxChat Status Update Error: ' . $e->getMessage());
1870 1769
1770 + // Return a friendly error response
1871 1771 wp_send_json_error(array(
1872 1772 'message' => 'Error getting status updates: ' . $e->getMessage(),
1873 1773 'status' => 'error'
1874 1774 ));
@@ -1873,1373 +1773,8 @@
1873 1773 'status' => 'error'
1874 1774 ));
1875 1775 }
1876 1776 }
1877 -
1878 -/**
1879 - * Helper function to get queue status data
1880 - */
1881 -private function mxchat_get_queue_status_data($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 1777 public function mxchat_stop_processing() {
3243 1778 // Verify permissions
3244 1779 if (!current_user_can('manage_options')) {
3245 1780 wp_die(esc_html__('Unauthorized access', 'mxchat'));
@@ -3247,52 +1782,28 @@
3247 1782
3248 1783 // Verify nonce
3249 1784 check_admin_referer('mxchat_stop_processing_action', 'mxchat_stop_processing_nonce');
3250 1785
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');
1786 + // Get the last sitemap URL and clear its transient
1787 + $sitemap_url = get_transient('mxchat_last_sitemap_url');
1788 + if ($sitemap_url) {
1789 + delete_transient('mxchat_sitemap_status_' . md5($sitemap_url));
3270 1790 delete_transient('mxchat_last_sitemap_url');
3271 1791 }
3272 -
3273 - if ($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');
1792 +
1793 + // Get the last PDF URL and clear its transient
1794 + $pdf_url = get_transient('mxchat_last_pdf_url');
1795 + if ($pdf_url) {
1796 + delete_transient('mxchat_pdf_status_' . md5($pdf_url));
3292 1797 delete_transient('mxchat_last_pdf_url');
3293 1798 }
3294 1799
1800 + // Unschedule any pending sitemap events
1801 + $timestamp = wp_next_scheduled('mxchat_process_sitemap_urls');
1802 + if ($timestamp) {
1803 + wp_unschedule_event($timestamp, 'mxchat_process_sitemap_urls');
1804 + }
1805 +
3295 1806 // Redirect back with a success message
3296 1807 set_transient('mxchat_admin_notice_success',
3297 1808 esc_html__('Processing has been stopped successfully.', 'mxchat'),
3298 1809 30
@@ -3299,12 +1810,8 @@
3299 1810 );
3300 1811 wp_safe_redirect(admin_url('admin.php?page=mxchat-prompts'));
3301 1812 exit;
3302 1813 }
3303 -
3304 -/**
3305 - * Get content list for processing
3306 - */
3307 1814 public function ajax_mxchat_get_content_list() {
3308 1815 // Verify the nonce
3309 1816 check_ajax_referer('mxchat_content_selector_nonce', 'nonce');
3310 1817
@@ -3312,9 +1819,9 @@
3312 1819 wp_send_json_error(__('Unauthorized access', 'mxchat'));
3313 1820 }
3314 1821
3315 1822 $page = isset($_GET['page']) ? absint($_GET['page']) : 1;
3316 - $per_page = isset($_GET['per_page']) ? absint($_GET['per_page']) : 100;
1823 + $per_page = isset($_GET['per_page']) ? absint($_GET['per_page']) : 20;
3317 1824 $search = isset($_GET['search']) ? sanitize_text_field($_GET['search']) : '';
3318 1825 $post_type = isset($_GET['post_type']) ? sanitize_text_field($_GET['post_type']) : 'all';
3319 1826 $post_status = isset($_GET['post_status']) ? sanitize_text_field($_GET['post_status']) : 'publish';
3320 1827 $processed_filter = isset($_GET['processed_filter']) ? sanitize_text_field($_GET['processed_filter']) : 'all';
@@ -3327,47 +1834,20 @@
3327 1834 'orderby' => 'date',
3328 1835 'order' => 'DESC',
3329 1836 );
3330 1837
3331 - // Handle post types - IMPROVED VERSION
1838 + // Handle post types
3332 1839 if ($post_type !== 'all') {
3333 1840 $args['post_type'] = $post_type;
3334 1841 } else {
3335 - // Get all available post types that might contain content
3336 - $all_post_types = array();
1842 + // Default to post and page if we can't get post types
1843 + $args['post_type'] = array('post', 'page');
3337 1844
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 - }
1845 + // Try to get public post types
1846 + $public_types = $this->mxchat_get_public_post_types();
1847 + if (is_array($public_types) && !empty($public_types)) {
1848 + $args['post_type'] = array_keys($public_types);
3348 1849 }
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(', ', $all_post_types));
3370 1850 }
3371 1851
3372 1852 if (!empty($search)) {
3373 1853 $args['s'] = $search;
@@ -3372,58 +1852,37 @@
3372 1852 if (!empty($search)) {
3373 1853 $args['s'] = $search;
3374 1854 }
3375 1855
3376 - // Get processed data from storage
1856 + // ================================
1857 + // FIXED: Check only the ACTIVE storage method
1858 + // ================================
1859 +
3377 1860 $processed_data = array();
3378 1861
1862 + $pinecone_manager = MxChat_Pinecone_Manager::get_instance();
3379 1863 $pinecone_options = get_option('mxchat_pinecone_addon_options', array());
1864 + $pinecone_manager->mxchat_refresh_after_new_content($pinecone_options);
3380 1865 $use_pinecone = ($pinecone_options['mxchat_use_pinecone'] ?? '0') === '1';
3381 1866
3382 1867 if ($use_pinecone && !empty($pinecone_options['mxchat_pinecone_api_key'])) {
3383 - // Get fresh data from Pinecone - no caching
1868 + // ONLY check Pinecone if it's enabled
3384 1869 $processed_data = $this->mxchat_get_pinecone_processed_content($pinecone_options);
3385 1870 } else {
3386 - // WordPress DB checking with better URL matching for all post types
1871 + // ONLY check WordPress DB if Pinecone is not enabled
3387 1872 global $wpdb;
3388 1873 $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 -
1874 + $processed_items = $wpdb->get_results("SELECT id, source_url, timestamp FROM {$table_name}");
1875 +
3396 1876 if (!empty($processed_items)) {
3397 1877 foreach ($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 -
1878 + $post_id = url_to_postid($item->source_url);
3419 1879 if ($post_id) {
3420 1880 $processed_data[$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
1881 + 'db_id' => $item->id,
1882 + 'timestamp' => $item->timestamp,
1883 + 'url' => $item->source_url,
1884 + 'source' => 'wordpress'
3426 1885 );
3427 1886 }
3428 1887 }
3429 1888 }
@@ -3428,8 +1887,10 @@
3428 1887 }
3429 1888 }
3430 1889 }
3431 1890
1891 + // ================================
1892 +
3432 1893 // Get processed IDs as a simple array for in_array checks
3433 1894 $processed_ids = array_keys($processed_data);
3434 1895
3435 1896 // Handle processed/unprocessed filter
@@ -3471,14 +1932,8 @@
3471 1932 $db_record_id = $item_data['db_id'];
3472 1933 }
3473 1934 }
3474 1935
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 1936 $content_items[] = array(
3482 1937 'id' => $id,
3483 1938 'title' => get_the_title(),
3484 1939 'permalink' => get_permalink(),
@@ -3489,10 +1944,9 @@
3489 1944 'word_count' => $word_count,
3490 1945 'already_processed' => $is_processed,
3491 1946 'processed_date' => $processed_date,
3492 1947 'db_record_id' => $db_record_id,
3493 - 'data_source' => $data_source,
3494 - 'chunk_count' => $chunk_count
1948 + 'data_source' => $data_source
3495 1949 );
3496 1950 }
3497 1951 wp_reset_postdata();
3498 1952 }
@@ -3508,178 +1962,8 @@
3508 1962 wp_send_json_success($response);
3509 1963 exit;
3510 1964 }
3511 1965
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 1966 public function ajax_mxchat_process_selected_content() {
3683 1967 // Basic request validation
3684 1968 if (!check_ajax_referer('mxchat_content_selector_nonce', 'nonce', false)) {
3685 1969 wp_send_json_error('Invalid nonce');
@@ -3703,24 +1987,8 @@
3703 1987 wp_send_json_error('No content selected');
3704 1988 exit;
3705 1989 }
3706 1990
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 1991 // Process only ONE post at a time to avoid request size issues
3724 1992 $post_id = reset($post_ids);
3725 1993 $post = get_post($post_id);
3726 1994
@@ -3727,185 +1995,15 @@
3727 1995 if (!$post) {
3728 1996 wp_send_json_error('Post not found');
3729 1997 exit;
3730 1998 }
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');
1999 +
2000 + // Get minimal content
2001 + $content = $post->post_title . "\n\n" . wp_strip_all_tags($post->post_content);
2002 + $content = substr($content, 0, 10000); // Limit content size
2003 +
2004 + // Get API key with proper model detection
2005 + $options = get_option('mxchat_options');
3908 2006 $selected_model = $options['embedding_model'] ?? 'text-embedding-ada-002';
3909 2007
3910 2008 if (strpos($selected_model, 'voyage') === 0) {
3911 2009 $api_key = $options['voyage_api_key'] ?? '';
@@ -3918,9 +2016,8 @@
3918 2016 $provider_name = 'OpenAI';
3919 2017 }
3920 2018
3921 2019 if (empty($api_key)) {
3922 - MxChat_Admin::mxchat_log_debug('api_error', $provider_name . ' API key not configured for knowledge processing');
3923 2020 wp_send_json_error($provider_name . ' API key not configured');
3924 2021 exit;
3925 2022 }
3926 2023
@@ -3926,23 +2023,26 @@
3926 2023
3927 2024 $source_url = get_permalink($post_id);
3928 2025 $vector_id = md5($source_url); // Vector ID for Pinecone
3929 2026
3930 - // Check for existing content in bot-specific storage
2027 + // ================================
2028 + // FIXED: Check for existing content in ONLY the active storage method
2029 + // ================================
2030 +
3931 2031 $is_update = false;
3932 2032
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);
2033 + // Check if Pinecone is enabled
2034 + $pinecone_options = get_option('mxchat_pinecone_addon_options', array());
2035 + $use_pinecone = ($pinecone_options['mxchat_use_pinecone'] ?? '0') === '1';
3936 2036
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);
2037 + if ($use_pinecone && !empty($pinecone_options['mxchat_pinecone_api_key'])) {
2038 + // ONLY check Pinecone if it's enabled
2039 + $pinecone_data = $this->mxchat_get_pinecone_processed_content($pinecone_options);
3940 2040 if (isset($pinecone_data[$post_id])) {
3941 2041 $is_update = true;
3942 2042 }
3943 2043 } else {
3944 - // Check WordPress DB (same as before since it's shared)
2044 + // ONLY check WordPress DB if Pinecone is not enabled
3945 2045 global $wpdb;
3946 2046 $table_name = $wpdb->prefix . 'mxchat_system_prompt_content';
3947 2047 $existing_record = $wpdb->get_row($wpdb->prepare(
3948 2048 "SELECT id FROM $table_name WHERE source_url = %s",
@@ -3952,54 +2052,54 @@
3952 2052 if ($existing_record) {
3953 2053 $is_update = true;
3954 2054 }
3955 2055 }
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
2056 +
2057 + // Use the centralized utility function for storage
3979 2058 $result = MxChat_Utils::submit_content_to_db(
3980 - $content,
3981 - $source_url,
2059 + $content,
2060 + $source_url,
3982 2061 $api_key,
3983 - $vector_id,
3984 - $bot_id,
3985 - $content_type
2062 + $vector_id
3986 2063 );
3987 -
2064 +
3988 2065 if (is_wp_error($result)) {
3989 - MxChat_Admin::mxchat_log_debug('storage_error', 'Knowledge storage failed: ' . $result->get_error_message(), array('source_url' => $source_url));
3990 2066 wp_send_json_error('Storage failed: ' . $result->get_error_message());
3991 2067 exit;
3992 2068 }
3993 -
3994 - // Automatically apply role restriction based on tags
3995 - $this->apply_role_restriction_to_post($post_id, $source_url);
3996 -
2069 +
2070 + // ================================
2071 + // UPDATE: Only update caches if Pinecone is enabled
2072 + // ================================
2073 +
2074 + if ($use_pinecone && !empty($pinecone_options['mxchat_pinecone_api_key'])) {
2075 + // Update vector ID cache for improved fetching
2076 + $this->mxchat_update_pinecone_vector_cache($vector_id);
2077 +
2078 + // Update local processed content cache for immediate UI feedback
2079 + $pinecone_cache = get_option('mxchat_pinecone_processed_cache', array());
2080 + $pinecone_cache[$post_id] = array(
2081 + 'db_id' => $vector_id,
2082 + 'processed_date' => 'Just now',
2083 + 'url' => $source_url,
2084 + 'source' => 'pinecone',
2085 + 'timestamp' => current_time('timestamp')
2086 + );
2087 + update_option('mxchat_pinecone_processed_cache', $pinecone_cache);
2088 +
2089 + // Also update the general processed content cache
2090 + $processed_cache = get_option('mxchat_processed_content_cache', array());
2091 + $processed_cache[$post_id] = array(
2092 + 'db_id' => $vector_id,
2093 + 'timestamp' => current_time('timestamp'),
2094 + 'url' => $source_url,
2095 + 'source' => 'pinecone'
2096 + );
2097 + update_option('mxchat_processed_content_cache', $processed_cache);
2098 + }
2099 +
3997 2100 $operation_type = $is_update ? 'update' : 'new';
3998 2101
3999 - // Count ACF fields for debugging
4000 - $acf_field_count = count($acf_fields);
4001 -
4002 2102 // Success response with minimal data
4003 2103 wp_send_json_success(array(
4004 2104 'message' => $operation_type === 'update' ? 'Content updated successfully' : 'Content processed successfully',
4005 2105 'post_id' => $post_id,
@@ -4004,95 +2104,27 @@
4004 2104 'message' => $operation_type === 'update' ? 'Content updated successfully' : 'Content processed successfully',
4005 2105 'post_id' => $post_id,
4006 2106 'title' => $post->post_title,
4007 2107 '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
2108 + 'vector_id' => $vector_id, // Include vector ID for debugging
2109 + 'cache_updated' => $use_pinecone // Indicate if cache was updated
4013 2110 ));
4014 2111 exit;
4015 2112 }
4016 2113
4017 -private function apply_role_restriction_to_post($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' => $source_url),
4087 - array('%s'),
4088 - array('%s')
4089 - );
4090 - }
4091 -}
4092 2114
2115 +
2116 + /**
2117 + * Updates cache with new vector ID if absent
2118 + */
2119 + public function mxchat_update_pinecone_vector_cache($vector_id) {
2120 + $cached_ids = get_option('mxchat_pinecone_vector_ids_cache', array());
2121 + if (!in_array($vector_id, $cached_ids)) {
2122 + $cached_ids[] = $vector_id;
2123 + update_option('mxchat_pinecone_vector_ids_cache', $cached_ids);
2124 + }
2125 + }
4093 2126 public function mxchat_get_public_post_types() {
4094 - // Get all public post types
4095 2127 $post_types = get_post_types(array('public' => true), 'objects');
4096 2128 $post_type_options = array();
4097 2129
4098 2130 foreach ($post_types as $post_type) {
@@ -4098,45 +2130,51 @@
4098 2130 foreach ($post_types as $post_type) {
4099 2131 $post_type_options[$post_type->name] = $post_type->label;
4100 2132 }
4101 2133
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 2134 return $post_type_options;
4118 2135 }
2136 +public function mxchat_get_pinecone_processed_content($pinecone_options) {
2137 + //error_log('=== DEBUG: Starting mxchat_get_pinecone_processed_content ===');
2138 +
2139 + // First check local cache for immediate updates
2140 + $cached_data = get_option('mxchat_pinecone_processed_cache', array());
2141 + //error_log('DEBUG: Found ' . count($cached_data) . ' items in local cache');
4119 2142
4120 -/**
4121 - * Retrieves processed content from Pinecone API
4122 - */
4123 -public function mxchat_get_pinecone_processed_content($pinecone_options) {
4124 2143 $api_key = $pinecone_options['mxchat_pinecone_api_key'] ?? '';
4125 2144 $host = $pinecone_options['mxchat_pinecone_host'] ?? '';
2145 +
2146 + //error_log('DEBUG: API key present: ' . (!empty($api_key) ? 'YES' : 'NO'));
2147 + //error_log('DEBUG: Host: ' . $host);
4126 2148
4127 2149 if (empty($api_key) || empty($host)) {
4128 - return array();
2150 + //error_log('DEBUG: Missing API credentials, returning cached data only');
2151 + return $cached_data;
4129 2152 }
4130 2153
4131 2154 $pinecone_data = array();
4132 2155
4133 2156 try {
4134 - // Always get fresh data from Pinecone
4135 - $pinecone_data = $this->mxchat_scan_pinecone_for_processed_content($pinecone_options);
2157 + // Method 1: Try to get vectors using cached vector IDs first
2158 + $cached_vector_ids = get_option('mxchat_pinecone_vector_ids_cache', array());
2159 + //error_log('DEBUG: Found ' . count($cached_vector_ids) . ' cached vector IDs');
4136 2160
4137 - // Method 2: Final fallback - try stats endpoint (if available)
2161 + if (!empty($cached_vector_ids)) {
2162 + //error_log('DEBUG: Trying to fetch by cached vector IDs...');
2163 + $pinecone_data = $this->mxchat_fetch_pinecone_vectors_by_ids($pinecone_options, $cached_vector_ids);
2164 + //error_log('DEBUG: Fetch by IDs returned ' . count($pinecone_data) . ' items');
2165 + }
2166 +
2167 + // Method 2: If no cached IDs or fetch failed, use scanning approach
4138 2168 if (empty($pinecone_data)) {
2169 + //error_log('DEBUG: Trying scanning approach...');
2170 + $pinecone_data = $this->mxchat_scan_pinecone_for_processed_content($pinecone_options);
2171 + //error_log('DEBUG: Scanning returned ' . count($pinecone_data) . ' items');
2172 + }
2173 +
2174 + // Method 3: Final fallback - try stats endpoint
2175 + if (empty($pinecone_data)) {
2176 + //error_log('DEBUG: Trying stats endpoint...');
4139 2177 $stats_url = "https://{$host}/describe_index_stats";
4140 2178
4141 2179 $response = wp_remote_post($stats_url, array(
4142 2180 'headers' => array(
@@ -4149,17 +2187,44 @@
4149 2187
4150 2188 if (!is_wp_error($response) && wp_remote_retrieve_response_code($response) === 200) {
4151 2189 $body = wp_remote_retrieve_body($response);
4152 2190 $stats_data = json_decode($body, true);
2191 + //error_log('DEBUG: Pinecone stats: ' . print_r($stats_data, true));
2192 + } else {
2193 + if (is_wp_error($response)) {
2194 + //error_log('DEBUG: Stats endpoint error: ' . $response->get_error_message());
2195 + } else {
2196 + //error_log('DEBUG: Stats endpoint failed with code: ' . wp_remote_retrieve_response_code($response));
2197 + }
4153 2198 }
4154 2199 }
4155 2200
4156 2201 } catch (Exception $e) {
4157 - // Log error but return fresh data only
2202 + //error_log('DEBUG: Exception in get_pinecone_processed_content: ' . $e->getMessage());
4158 2203 }
4159 2204
4160 - return $pinecone_data;
2205 + // Merge cached data with Pinecone data
2206 + $merged_data = $pinecone_data;
2207 +
2208 + foreach ($cached_data as $post_id => $cache_item) {
2209 + $cache_timestamp = $cache_item['timestamp'] ?? 0;
2210 + $time_diff = current_time('timestamp') - $cache_timestamp;
2211 +
2212 + if ($time_diff < 300) { // 5 minutes = 300 seconds
2213 + $merged_data[$post_id] = $cache_item;
2214 + } else {
2215 + if (!isset($merged_data[$post_id])) {
2216 + $merged_data[$post_id] = $cache_item;
2217 + }
2218 + }
2219 + }
2220 +
2221 + //error_log('DEBUG: Final merged data count: ' . count($merged_data));
2222 + //error_log('=== DEBUG: End mxchat_get_pinecone_processed_content ===');
2223 +
2224 + return $merged_data;
4161 2225 }
2226 +
4162 2227 public function mxchat_fetch_pinecone_vectors_by_ids($pinecone_options, $vector_ids) {
4163 2228 //error_log('=== DEBUG: Starting mxchat_fetch_pinecone_vectors_by_ids ===');
4164 2229
4165 2230 $api_key = $pinecone_options['mxchat_pinecone_api_key'] ?? '';
@@ -4251,46 +2316,16 @@
4251 2316 return array();
4252 2317 }
4253 2318 }
4254 2319
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 2320 public function mxchat_scan_pinecone_for_processed_content($pinecone_options) {
2321 + //error_log('=== DEBUG: Starting mxchat_scan_pinecone_for_processed_content ===');
2322 +
4289 2323 $api_key = $pinecone_options['mxchat_pinecone_api_key'] ?? '';
4290 2324 $host = $pinecone_options['mxchat_pinecone_host'] ?? '';
4291 2325
4292 2326 if (empty($api_key) || empty($host)) {
2327 + //error_log('DEBUG: Missing API credentials for scanning');
4293 2328 return array();
4294 2329 }
4295 2330
4296 2331 try {
@@ -4297,18 +2332,17 @@
4297 2332 // Use multiple random vectors to get better coverage
4298 2333 $all_matches = array();
4299 2334 $seen_ids = array();
4300 2335
4301 - // Get correct dimensions for the configured embedding model
4302 - $dimensions = $this->mxchat_get_embedding_dimensions();
4303 -
4304 2336 // Try 3 different random vectors to get better coverage
4305 2337 for ($i = 0; $i < 3; $i++) {
2338 + //error_log('DEBUG: Scanning attempt ' . ($i + 1) . '/3');
2339 +
4306 2340 $query_url = "https://{$host}/query";
4307 2341
4308 2342 // Generate a random unit vector instead of zeros
4309 2343 $random_vector = array();
4310 - for ($j = 0; $j < $dimensions; $j++) {
2344 + for ($j = 0; $j < 1536; $j++) {
4311 2345 $random_vector[] = (rand(-1000, 1000) / 1000.0);
4312 2346 }
4313 2347
4314 2348 // Normalize the vector to unit length
@@ -4333,14 +2367,18 @@
4333 2367 'timeout' => 30
4334 2368 ));
4335 2369
4336 2370 if (is_wp_error($response)) {
2371 + //error_log('DEBUG: Query attempt ' . ($i + 1) . ' WP error: ' . $response->get_error_message());
4337 2372 continue;
4338 2373 }
4339 2374
4340 2375 $response_code = wp_remote_retrieve_response_code($response);
2376 + //error_log('DEBUG: Query attempt ' . ($i + 1) . ' response code: ' . $response_code);
4341 2377
4342 2378 if ($response_code !== 200) {
2379 + $error_body = wp_remote_retrieve_body($response);
2380 + //error_log('DEBUG: Query attempt ' . ($i + 1) . ' failed with body: ' . substr($error_body, 0, 500));
4343 2381 continue;
4344 2382 }
4345 2383
4346 2384 $body = wp_remote_retrieve_body($response);
@@ -4346,8 +2384,9 @@
4346 2384 $body = wp_remote_retrieve_body($response);
4347 2385 $data = json_decode($body, true);
4348 2386
4349 2387 if (isset($data['matches'])) {
2388 + //error_log('DEBUG: Query attempt ' . ($i + 1) . ' returned ' . count($data['matches']) . ' matches');
4350 2389 foreach ($data['matches'] as $match) {
4351 2390 $match_id = $match['id'] ?? '';
4352 2391 if (!empty($match_id) && !isset($seen_ids[$match_id])) {
4353 2392 $all_matches[] = $match;
@@ -4353,14 +2392,18 @@
4353 2392 $all_matches[] = $match;
4354 2393 $seen_ids[$match_id] = true;
4355 2394 }
4356 2395 }
2396 + } else {
2397 + //error_log('DEBUG: Query attempt ' . ($i + 1) . ' - no matches key in response');
4357 2398 }
4358 2399 }
4359 2400
4360 - // Convert matches to processed data format, grouping by URL to count chunks
2401 + //error_log('DEBUG: Total unique matches found: ' . count($all_matches));
2402 +
2403 + // Convert matches to processed data format
4361 2404 $processed_data = array();
4362 - $url_chunk_counts = array();
2405 + $vector_ids_for_cache = array();
4363 2406
4364 2407 foreach ($all_matches as $match) {
4365 2408 $metadata = $match['metadata'] ?? array();
4366 2409 $source_url = $metadata['source_url'] ?? '';
@@ -4368,14 +2411,8 @@
4368 2411
4369 2412 if (!empty($source_url) && !empty($match_id)) {
4370 2413 $post_id = url_to_postid($source_url);
4371 2414 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 2415 $created_at = $metadata['created_at'] ?? '';
4379 2416 $processed_date = 'Recently';
4380 2417
4381 2418 if (!empty($created_at)) {
@@ -4384,322 +2421,243 @@
4384 2421 $processed_date = human_time_diff($timestamp, current_time('timestamp')) . ' ago';
4385 2422 }
4386 2423 }
4387 2424
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 - }
2425 + $processed_data[$post_id] = array(
2426 + 'db_id' => $match_id,
2427 + 'processed_date' => $processed_date,
2428 + 'url' => $source_url,
2429 + 'source' => 'pinecone',
2430 + 'timestamp' => $timestamp ?? current_time('timestamp')
2431 + );
2432 +
2433 + $vector_ids_for_cache[] = $match_id;
4399 2434 }
4400 2435 }
4401 2436 }
4402 2437
4403 - // Add chunk counts to processed data
4404 - foreach ($url_chunk_counts as $post_id => $chunk_count) {
4405 - if (isset($processed_data[$post_id])) {
4406 - $processed_data[$post_id]['chunk_count'] = $chunk_count;
4407 - }
2438 + // Update the vector IDs cache for future use
2439 + if (!empty($vector_ids_for_cache)) {
2440 + update_option('mxchat_pinecone_vector_ids_cache', $vector_ids_for_cache);
2441 + //error_log('DEBUG: Updated vector IDs cache with ' . count($vector_ids_for_cache) . ' IDs');
4408 2442 }
4409 2443
2444 + //error_log('DEBUG: Returning ' . count($processed_data) . ' processed items from scanning');
4410 2445 return $processed_data;
4411 2446
4412 2447 } catch (Exception $e) {
2448 + //error_log('DEBUG: Exception in scan_pinecone_for_processed_content: ' . $e->getMessage());
4413 2449 return array();
4414 2450 }
4415 2451 }
4416 -/**
4417 - * Generate embeddings from input text for MXChat with bot support
4418 - */
4419 -private function mxchat_generate_embedding($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 2452
4424 - // Get bot-specific options
4425 - $bot_options = $this->get_bot_options($bot_id);
4426 - $options = !empty($bot_options) ? $bot_options : get_option('mxchat_options');
2453 + /**
2454 + * Generates embeddings from input text for MXChat
2455 + */
2456 + private function mxchat_generate_embedding($text) {
2457 + // Enable detailed logging for debugging
2458 + //error_log('[MXCHAT-EMBED] Starting embedding generation. Text length: ' . strlen($text) . ' bytes');
2459 + //error_log('[MXCHAT-EMBED] Text preview: ' . substr($text, 0, 100) . '...');
4427 2460
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 - }
2461 + $options = get_option('mxchat_options');
2462 + $selected_model = $options['embedding_model'] ?? 'text-embedding-ada-002';
2463 + //error_log('[MXCHAT-EMBED] Selected embedding model: ' . $selected_model);
4438 2464
4439 - $selected_model = $options['embedding_model'] ?? 'text-embedding-ada-002';
4440 - //error_log('[MXCHAT-EMBED] Selected embedding model for bot ' . $bot_id . ': ' . $selected_model);
2465 + // Determine provider and endpoint
2466 + if (strpos($selected_model, 'voyage') === 0) {
2467 + $api_key = $options['voyage_api_key'] ?? '';
2468 + $endpoint = 'https://api.voyageai.com/v1/embeddings';
2469 + $provider_name = 'Voyage AI';
2470 + //error_log('[MXCHAT-EMBED] Using Voyage AI API');
2471 + } elseif (strpos($selected_model, 'gemini-embedding') === 0) {
2472 + $api_key = $options['gemini_api_key'] ?? '';
2473 + $endpoint = 'https://generativelanguage.googleapis.com/v1beta/models/' . $selected_model . ':embedContent';
2474 + $provider_name = 'Google Gemini';
2475 + //error_log('[MXCHAT-EMBED] Using Google Gemini API');
2476 + } else {
2477 + $api_key = $options['api_key'] ?? '';
2478 + $endpoint = 'https://api.openai.com/v1/embeddings';
2479 + $provider_name = 'OpenAI';
2480 + //error_log('[MXCHAT-EMBED] Using OpenAI API');
2481 + }
4441 2482
4442 - // Determine provider and endpoint
4443 - if (strpos($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 - }
2483 + //error_log('[MXCHAT-EMBED] Using endpoint: ' . $endpoint);
4459 2484
4460 - //error_log('[MXCHAT-EMBED] Using endpoint: ' . $endpoint);
2485 + if (empty($api_key)) {
2486 + $error_message = sprintf('Missing %s API key. Please configure your API key in the MxChat settings.', $provider_name);
2487 + //error_log('[MXCHAT-EMBED] Error: ' . $error_message);
2488 + return $error_message;
2489 + }
4461 2490
4462 - if (empty($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 - }
2491 + // Check if text is too long (for OpenAI, roughly estimate tokens as words/0.75)
2492 + $estimated_tokens = ceil(str_word_count($text) / 0.75);
2493 + //error_log('[MXCHAT-EMBED] Estimated token count: ~' . $estimated_tokens);
4467 2494
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: ~' . $estimated_tokens);
2495 + if ($estimated_tokens > 8000 && strpos($selected_model, 'voyage') === false && strpos($selected_model, 'gemini-embedding') === false) {
2496 + //error_log('[MXCHAT-EMBED] Warning: Text may exceed OpenAI token limits (8K for most models)');
2497 + // Consider truncating text here
2498 + }
4471 2499
4472 - if ($estimated_tokens > 8000 && strpos($selected_model, 'voyage') === false && strpos($selected_model, 'gemini-embedding') === false) {
4473 - //error_log('[MXCHAT-EMBED] Warning: Text may exceed OpenAI token limits (8K for most models)');
4474 - // Consider truncating text here
4475 - }
2500 + // Prepare request body based on provider
2501 + if (strpos($selected_model, 'gemini-embedding') === 0) {
2502 + // Gemini API format
2503 + $request_body = array(
2504 + 'model' => 'models/' . $selected_model,
2505 + 'content' => array(
2506 + 'parts' => array(
2507 + array('text' => $text)
2508 + )
2509 + )
2510 + );
4476 2511
4477 - // Prepare request body based on provider
4478 - if (strpos($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 - );
2512 + // Set output dimensionality to 1536 for consistency with other models
2513 + $request_body['outputDimensionality'] = 1536;
2514 + } else {
2515 + // OpenAI/Voyage API format
2516 + $request_body = array(
2517 + 'model' => $selected_model,
2518 + 'input' => $text
2519 + );
4488 2520
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' => $text
4496 - );
2521 + // Add output_dimension for voyage-3-large model
2522 + if ($selected_model === 'voyage-3-large') {
2523 + $request_body['output_dimension'] = 2048;
2524 + }
2525 + }
4497 2526
4498 - // Add output_dimension for voyage-3-large model
4499 - if ($selected_model === 'voyage-3-large') {
4500 - $request_body['output_dimension'] = 2048;
4501 - }
4502 - }
2527 + //error_log('[MXCHAT-EMBED] Request prepared with model: ' . $selected_model);
4503 2528
4504 - //error_log('[MXCHAT-EMBED] Request prepared with model: ' . $selected_model);
2529 + // Prepare headers based on provider
2530 + if (strpos($selected_model, 'gemini-embedding') === 0) {
2531 + // Gemini uses API key as query parameter
2532 + $endpoint .= '?key=' . $api_key;
2533 + $headers = array(
2534 + 'Content-Type' => 'application/json'
2535 + );
2536 + } else {
2537 + // OpenAI/Voyage use Bearer token
2538 + $headers = array(
2539 + 'Authorization' => 'Bearer ' . $api_key,
2540 + 'Content-Type' => 'application/json'
2541 + );
2542 + }
4505 2543
4506 - // Prepare headers based on provider
4507 - if (strpos($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 - }
2544 + // Make API request
2545 + //error_log('[MXCHAT-EMBED] Sending API request to: ' . $endpoint);
2546 + $response = wp_remote_post($endpoint, array(
2547 + 'body' => wp_json_encode($request_body),
2548 + 'headers' => $headers,
2549 + 'timeout' => 60 // Increased timeout for large inputs
2550 + ));
4520 2551
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' => $headers,
4526 - 'timeout' => 60 // Increased timeout for large inputs
4527 - ));
2552 + // Handle wp_remote_post errors
2553 + if (is_wp_error($response)) {
2554 + $error_message = $response->get_error_message();
2555 + //error_log('[MXCHAT-EMBED] WP Remote Post Error: ' . $error_message);
2556 + return 'Connection error: ' . $error_message;
2557 + }
4528 2558
4529 - // Handle wp_remote_post errors
4530 - if (is_wp_error($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 - }
2559 + // Get and check HTTP response code
2560 + $http_code = wp_remote_retrieve_response_code($response);
2561 + //error_log('[MXCHAT-EMBED] API Response Code: ' . $http_code);
4535 2562
4536 - // Get and check HTTP response code
4537 - $http_code = wp_remote_retrieve_response_code($response);
4538 - //error_log('[MXCHAT-EMBED] API Response Code: ' . $http_code);
2563 + if ($http_code !== 200) {
2564 + $error_body = wp_remote_retrieve_body($response);
2565 + //error_log('[MXCHAT-EMBED] API Error Response Body: ' . $error_body);
4539 2566
4540 - if ($http_code !== 200) {
4541 - $error_body = wp_remote_retrieve_body($response);
4542 - //error_log('[MXCHAT-EMBED] API Error Response Body: ' . $error_body);
2567 + // Try to parse error for more details
2568 + $error_json = json_decode($error_body, true);
2569 + if (json_last_error() === JSON_ERROR_NONE && isset($error_json['error'])) {
2570 + $error_type = $error_json['error']['type'] ?? 'unknown';
2571 + $error_message = $error_json['error']['message'] ?? 'No message';
2572 + //error_log('[MXCHAT-EMBED] API Error Type: ' . $error_type);
2573 + //error_log('[MXCHAT-EMBED] API Error Message: ' . $error_message);
4543 2574
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: ' . $error_message);
2575 + // Customize error message for common API errors
2576 + if ($error_type === 'invalid_request_error' && strpos($error_message, 'API key') !== false) {
2577 + $error_message = sprintf('Invalid %s API key. Please check your API key in the MxChat settings.', $provider_name);
2578 + } elseif ($error_type === 'authentication_error') {
2579 + $error_message = sprintf('%s authentication failed. Please verify your API key in the MxChat settings.', $provider_name);
2580 + }
4551 2581
4552 - // Customize error message for common API errors
4553 - if ($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 - }
2582 + //error_log('[MXCHAT-EMBED] Returning error: ' . $error_message);
2583 + return $error_message;
2584 + }
4558 2585
4559 - //error_log('[MXCHAT-EMBED] Returning error: ' . $error_message);
4560 - return $error_message;
4561 - }
2586 + $error_message = sprintf("API Error (HTTP %d): Unable to generate embedding", $http_code);
2587 + //error_log('[MXCHAT-EMBED] Returning error: ' . $error_message);
2588 + return $error_message;
2589 + }
4562 2590
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 - }
2591 + // Parse response body
2592 + $response_body = wp_remote_retrieve_body($response);
2593 + //error_log('[MXCHAT-EMBED] Received response length: ' . strlen($response_body) . ' bytes');
4567 2594
4568 - // Parse response body
4569 - $response_body = wp_remote_retrieve_body($response);
4570 - //error_log('[MXCHAT-EMBED] Received response length: ' . strlen($response_body) . ' bytes');
2595 + $response_data = json_decode($response_body, true);
4571 2596
4572 - $response_data = json_decode($response_body, true);
2597 + if (json_last_error() !== JSON_ERROR_NONE) {
2598 + $error = json_last_error_msg();
2599 + //error_log('[MXCHAT-EMBED] JSON Parse Error: ' . $error);
2600 + //error_log('[MXCHAT-EMBED] Response preview: ' . substr($response_body, 0, 200));
2601 + return "Failed to parse API response: $error";
2602 + }
4573 2603
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: $error";
4579 - }
2604 + // Handle different response formats based on provider
2605 + if (strpos($selected_model, 'gemini-embedding') === 0) {
2606 + // Gemini API response format
2607 + if (isset($response_data['embedding']['values'])) {
2608 + $embedding_dimensions = count($response_data['embedding']['values']);
2609 + //error_log('[MXCHAT-EMBED] Successfully extracted Gemini embedding with ' . $embedding_dimensions . ' dimensions');
4580 2610
4581 - // Handle different response formats based on provider
4582 - if (strpos($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 ' . $embedding_dimensions . ' dimensions');
2611 + // Check if embedding dimensions are as expected (should be 1536)
2612 + if ($embedding_dimensions !== 1536) {
2613 + //error_log('[MXCHAT-EMBED] Warning: Unexpected Gemini embedding dimensions: ' . $embedding_dimensions);
2614 + }
4587 2615
4588 - // Check if embedding dimensions are as expected (should be 1536)
4589 - if ($embedding_dimensions !== 1536) {
4590 - //error_log('[MXCHAT-EMBED] Warning: Unexpected Gemini embedding dimensions: ' . $embedding_dimensions);
4591 - }
2616 + return $response_data['embedding']['values'];
2617 + } else {
2618 + //error_log('[MXCHAT-EMBED] Error: No embedding found in Gemini response');
2619 + //error_log('[MXCHAT-EMBED] Response structure: ' . wp_json_encode(array_keys($response_data)));
4592 2620
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)));
2621 + if (isset($response_data['error'])) {
2622 + $error_message = "Gemini API Error in response: " . wp_json_encode($response_data['error']);
2623 + //error_log('[MXCHAT-EMBED] ' . $error_message);
2624 + return $error_message;
2625 + }
4597 2626
4598 - if (isset($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 - }
2627 + $error_message = "Invalid Gemini API response format: No embedding found";
2628 + //error_log('[MXCHAT-EMBED] ' . $error_message);
2629 + return $error_message;
2630 + }
2631 + } else {
2632 + // OpenAI/Voyage API response format
2633 + if (isset($response_data['data'][0]['embedding'])) {
2634 + $embedding_dimensions = count($response_data['data'][0]['embedding']);
2635 + //error_log('[MXCHAT-EMBED] Successfully extracted embedding with ' . $embedding_dimensions . ' dimensions');
4603 2636
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 ' . $embedding_dimensions . ' dimensions');
2637 + // Check if embedding dimensions are as expected
2638 + if (($selected_model === 'text-embedding-ada-002' && $embedding_dimensions !== 1536) ||
2639 + ($selected_model === 'voyage-3-large' && $embedding_dimensions !== 2048)) {
2640 + //error_log('[MXCHAT-EMBED] Warning: Unexpected embedding dimensions');
2641 + }
4613 2642
4614 - // Check if embedding dimensions are as expected
4615 - if (($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 - }
2643 + return $response_data['data'][0]['embedding'];
2644 + } else {
2645 + //error_log('[MXCHAT-EMBED] Error: No embedding found in response');
2646 + //error_log('[MXCHAT-EMBED] Response structure: ' . wp_json_encode(array_keys($response_data)));
4619 2647
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)));
2648 + if (isset($response_data['error'])) {
2649 + $error_message = "API Error in response: " . wp_json_encode($response_data['error']);
2650 + //error_log('[MXCHAT-EMBED] ' . $error_message);
2651 + return $error_message;
2652 + }
4624 2653
4625 - if (isset($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 -
2654 + $error_message = "Invalid API response format: No embedding found";
2655 + //error_log('[MXCHAT-EMBED] ' . $error_message);
2656 + return $error_message;
2657 + }
2658 + }
2659 + }
4702 2660 public function mxchat_ajax_dismiss_completed_status() {
4703 2661 try {
4704 2662 // Verify the request
4705 2663 check_ajax_referer('mxchat_status_nonce', 'nonce');
@@ -5062,564 +3020,18 @@
5062 3020
5063 3021 return $html;
5064 3022 }
5065 3023
5066 -/**
5067 - * Get all ACF fields for a specific post, excluding any fields the user has disabled
5068 - */
5069 -public function mxchat_get_acf_fields_for_post($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 || $post_id === 'options') {
5553 - return;
5554 - }
5555 -
5556 - // Skip autosaves and revisions
3024 +public function mxchat_handle_post_update($post_id, $post, $update) {
3025 + // Basic validation checks
5557 3026 if (defined('DOING_AUTOSAVE') && DOING_AUTOSAVE || wp_is_post_revision($post_id)) {
5558 3027 return;
5559 3028 }
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
3029 +
3030 + // Only process published content
5595 3031 if ($post->post_status !== 'publish') {
5596 3032 return;
5597 3033 }
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 3034
5623 3035 $post_type = $post->post_type;
5624 3036
5625 3037 // Check if sync is enabled for this post type
@@ -5641,276 +3053,80 @@
5641 3053 if (!$should_sync) {
5642 3054 return;
5643 3055 }
5644 3056
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);
3057 + // Get content with proper formatting (matching ajax_mxchat_process_selected_content)
3058 + $title = get_the_title($post_id);
3059 + $content = get_post_field('post_content', $post_id);
5648 3060
5649 - $previous_url_key = 'mxchat_prev_url_' . $post_id;
5650 - $previous_url = get_transient($previous_url_key);
3061 + // Apply WordPress content filters to get properly formatted content
3062 + $content = apply_filters('the_content', $content);
5651 3063
5652 - // If the post was previously published but is now not published, remove from knowledge base
5653 - if ($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');
3064 + // Strip tags but preserve structure
3065 + $content = wp_strip_all_tags($content);
3066 +
3067 + // Combine title and content
3068 + $final_content = $title . "\n\n" . $content;
3069 +
3070 + // For custom post types like job_listing, include additional fields
3071 + if ($post_type === 'job_listing') {
3072 + // Add job-specific meta if available
3073 + $job_location = get_post_meta($post_id, '_job_location', true);
3074 + if (!empty($job_location)) {
3075 + $final_content .= "\n\nLocation: " . $job_location;
5660 3076 }
5661 -
5662 - // Clean up the transients and exit early
5663 - delete_transient($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 3077
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 - }
3078 + // Get job type terms
3079 + $job_types = get_the_terms($post_id, 'job_listing_type');
3080 + if (!empty($job_types) && !is_wp_error($job_types)) {
3081 + $types = array();
3082 + foreach ($job_types as $type) {
3083 + $types[] = $type->name;
5759 3084 }
3085 + $final_content .= "\n\nJob Type: " . implode(', ', $types);
5760 3086 }
5761 -
5762 - // For custom post types like job_listing, include additional fields
5763 - if ($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 = $options['embedding_model'] ?? 'text-embedding-ada-002';
5857 3087
5858 - if (strpos($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'] ?? '';
3088 + // Get company name if available
3089 + $company_name = get_post_meta($post_id, '_company_name', true);
3090 + if (!empty($company_name)) {
3091 + $final_content .= "\n\nCompany: " . $company_name;
5864 3092 }
5865 -
5866 - if (empty($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, $source_url);
5881 - }
5882 3093 }
5883 3094
5884 - // Clean up the stored previous status if not used above
5885 - if ($previous_status !== 'publish' || $post->post_status === 'publish') {
5886 - delete_transient($previous_status_key);
5887 - delete_transient($previous_url_key);
3095 + // Get the source URL
3096 + $source_url = get_permalink($post_id);
3097 +
3098 + // Get API key with proper model detection
3099 + $options = get_option('mxchat_options');
3100 + $selected_model = $options['embedding_model'] ?? 'text-embedding-ada-002';
3101 +
3102 + if (strpos($selected_model, 'voyage') === 0) {
3103 + $api_key = $options['voyage_api_key'] ?? '';
3104 + } elseif (strpos($selected_model, 'gemini-embedding') === 0) {
3105 + $api_key = $options['gemini_api_key'] ?? '';
3106 + } else {
3107 + $api_key = $options['api_key'] ?? '';
5888 3108 }
5889 -}
5890 -
5891 -/**
5892 - * Store the post status and URL before update to detect status transitions
5893 - * This runs before the post is actually updated in the database
5894 - */
5895 -public function mxchat_store_pre_update_status($post_id, $data) {
5896 - // Get the current post from database (before update)
5897 - $current_post = get_post($post_id);
5898 3109
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, $current_url, HOUR_IN_SECONDS);
5909 - }
3110 + if (empty($api_key)) {
3111 + //error_log('MxChat Auto-sync: No API key configured for embedding model');
3112 + return;
5910 3113 }
3114 +
3115 + // Use the centralized utility function for storage
3116 + $result = MxChat_Utils::submit_content_to_db(
3117 + $final_content,
3118 + $source_url,
3119 + $api_key,
3120 + md5($source_url) // Vector ID for Pinecone
3121 + );
3122 +
3123 + if (is_wp_error($result)) {
3124 + //error_log('MxChat Auto-sync failed for post ' . $post_id . ': ' . $result->get_error_message());
3125 + }
5911 3126 }
5912 3127
3128 +
5913 3129 public function mxchat_handle_post_delete($post_id) {
5914 3130 // Get post data before it's deleted
5915 3131 $post = get_post($post_id);
5916 3132
@@ -5940,51 +3156,85 @@
5940 3156 if (!$should_sync) {
5941 3157 return;
5942 3158 }
5943 3159
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);
3160 + // Get the URL before post is deleted
3161 + $source_url = get_permalink($post_id);
5948 3162 if (!$source_url) {
5949 - //error_log('MXChat: Failed to resolve source URL for post ' . $post_id);
3163 + //error_log('MXChat: Failed to get permalink for post ' . $post_id);
5950 3164 return;
5951 3165 }
5952 3166
5953 - // Use chunk-aware deletion (handles both chunked and non-chunked content)
5954 - $delete_result = MxChat_Utils::delete_chunks_for_url($source_url, 'default');
3167 + // Check if Pinecone is enabled
3168 + $pinecone_options = get_option('mxchat_pinecone_addon_options', array());
3169 + $use_pinecone = ($pinecone_options['mxchat_use_pinecone'] ?? '0') === '1';
5955 3170
5956 - if (is_wp_error($delete_result)) {
5957 - //error_log('MXChat: Chunk-aware deletion failed for URL: ' . $source_url . ' - ' . $delete_result->get_error_message());
3171 + if ($use_pinecone && !empty($pinecone_options['mxchat_pinecone_api_key'])) {
3172 + // Delete from Pinecone
3173 + $this->mxchat_delete_from_pinecone_by_url($source_url, $pinecone_options);
3174 + } else {
3175 + // Delete from WordPress DB
3176 + global $wpdb;
3177 + $table_name = $wpdb->prefix . 'mxchat_system_prompt_content';
3178 +
3179 + $result = $wpdb->delete(
3180 + $table_name,
3181 + array('source_url' => $source_url),
3182 + array('%s')
3183 + );
3184 +
3185 + if ($result === false) {
3186 + //error_log('MXChat: WordPress DB deletion failed for URL: ' . $source_url);
3187 + }
5958 3188 }
3189 +}
5959 3190
5960 - delete_transient('mxchat_prev_url_' . $post_id);
5961 - delete_transient('mxchat_prev_status_' . $post_id);
5962 -}
5963 3191
5964 -/**
5965 - * Resolve the source URL for a post being trashed/deleted.
5966 - *
5967 - * Why: wp_trash_post appends "__trashed" to the slug before the wp_trash_post action fires, so
5968 - * get_permalink() returns a URL whose md5() won't match the vector IDs stored in Pinecone or
5969 - * the source_url rows in the WP DB. Prefer the URL captured by mxchat_store_pre_update_status
5970 - * (runs on pre_post_update, before the rename); fall back to stripping the __trashed suffix.
5971 - */
5972 -private function mxchat_resolve_pre_trash_url($post_id) {
5973 - $previous_url = get_transient('mxchat_prev_url_' . $post_id);
5974 - if (!empty($previous_url)) {
5975 - return $previous_url;
5976 - }
3192 + /**
3193 + * Deletes data from Pinecone using a source URL
3194 + */
3195 + public function mxchat_delete_from_pinecone_by_url($source_url, $pinecone_options) {
3196 + $host = $pinecone_options['mxchat_pinecone_host'] ?? '';
3197 + $api_key = $pinecone_options['mxchat_pinecone_api_key'] ?? '';
5977 3198
5978 - $current = get_permalink($post_id);
5979 - if (!$current) {
5980 - return '';
5981 - }
5982 - return preg_replace('#__trashed(/?)$#', '$1', $current);
5983 -}
3199 + if (empty($host) || empty($api_key)) {
3200 + //error_log('MXChat: Pinecone deletion failed - missing configuration');
3201 + return false;
3202 + }
5984 3203
3204 + $api_endpoint = "https://{$host}/vectors/delete";
3205 + $vector_id = md5($source_url);
5985 3206
3207 + $request_body = array(
3208 + 'ids' => array($vector_id)
3209 + );
5986 3210
3211 + $response = wp_remote_post($api_endpoint, array(
3212 + 'headers' => array(
3213 + 'Api-Key' => $api_key,
3214 + 'accept' => 'application/json',
3215 + 'content-type' => 'application/json'
3216 + ),
3217 + 'body' => wp_json_encode($request_body),
3218 + 'timeout' => 30
3219 + ));
3220 +
3221 + if (is_wp_error($response)) {
3222 + //error_log('MXChat: Pinecone deletion error - ' . $response->get_error_message());
3223 + return false;
3224 + }
3225 +
3226 + $response_code = wp_remote_retrieve_response_code($response);
3227 + if ($response_code !== 200) {
3228 + //error_log('MXChat: Pinecone deletion failed with status ' . $response_code);
3229 + return false;
3230 + }
3231 +
3232 + return true;
3233 + }
3234 +
3235 +
3236 +
5987 3237 public function mxchat_handle_product_change($post_id, $post, $update) {
5988 3238 if ($post->post_type !== 'product') {
5989 3239 return;
5990 3240 }
@@ -6006,12 +3256,11 @@
6006 3256 if (!isset($this->options['enable_woocommerce_integration']) ||
6007 3257 !in_array($this->options['enable_woocommerce_integration'], ['1', 'on'])) {
6008 3258 return;
6009 3259 }
6010 -
3260 +
6011 3261 $source_url = get_permalink($product->get_id());
6012 - $product_id = $product->get_id();
6013 -
3262 +
6014 3263 // Build product content
6015 3264 $title = $product->get_name();
6016 3265 $description = $product->get_description();
6017 3266 $short_description = $product->get_short_description();
@@ -6016,92 +3265,35 @@
6016 3265 $description = $product->get_description();
6017 3266 $short_description = $product->get_short_description();
6018 3267 $regular_price = $product->get_regular_price();
6019 3268 $sale_price = $product->get_sale_price();
6020 - $price = $product->get_price();
6021 3269 $sku = $product->get_sku();
6022 -
6023 - // Get currency symbol
6024 - $currency_symbol = get_woocommerce_currency_symbol();
6025 -
3270 +
6026 3271 // Format content consistently
6027 3272 $content = $title . "\n\n";
6028 -
6029 - if (!empty($short_description)) {
6030 - $content .= "Short Description: " . wp_strip_all_tags($short_description) . "\n\n";
6031 - }
6032 -
3273 +
6033 3274 if (!empty($description)) {
6034 3275 $content .= wp_strip_all_tags($description) . "\n\n";
6035 3276 }
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 . $price . "\n";
3277 +
3278 + if (!empty($short_description)) {
3279 + $content .= "Short Description: " . wp_strip_all_tags($short_description) . "\n\n";
6042 3280 }
6043 -
6044 - if (!empty($sale_price) && $sale_price !== $regular_price) {
6045 - $content .= "Sale Price: " . $currency_symbol . $sale_price . "\n";
3281 +
3282 + $content .= "Price: $" . $regular_price . "\n";
3283 +
3284 + if (!empty($sale_price)) {
3285 + $content .= "Sale Price: $" . $sale_price . "\n";
6046 3286 }
6047 -
6048 - // Handle variable products - show price range
6049 - if ($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 . $max_price . "\n";
6054 - }
6055 - }
6056 -
3287 +
6057 3288 if (!empty($sku)) {
6058 3289 $content .= "SKU: " . $sku . "\n";
6059 3290 }
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 -
3291 +
6100 3292 // Get API key with proper model detection
6101 3293 $options = get_option('mxchat_options');
6102 3294 $selected_model = $options['embedding_model'] ?? 'text-embedding-ada-002';
6103 -
3295 +
6104 3296 if (strpos($selected_model, 'voyage') === 0) {
6105 3297 $api_key = $options['voyage_api_key'] ?? '';
6106 3298 } elseif (strpos($selected_model, 'gemini-embedding') === 0) {
6107 3299 $api_key = $options['gemini_api_key'] ?? '';
@@ -6107,14 +3299,14 @@
6107 3299 $api_key = $options['gemini_api_key'] ?? '';
6108 3300 } else {
6109 3301 $api_key = $options['api_key'] ?? '';
6110 3302 }
6111 -
3303 +
6112 3304 if (empty($api_key)) {
6113 3305 //error_log('MxChat Auto-sync: No API key configured for embedding model');
6114 3306 return;
6115 3307 }
6116 -
3308 +
6117 3309 // Use the centralized utility function for storage
6118 3310 $result = MxChat_Utils::submit_content_to_db(
6119 3311 $content,
6120 3312 $source_url,
@@ -6120,16 +3312,11 @@
6120 3312 $source_url,
6121 3313 $api_key,
6122 3314 md5($source_url) // Vector ID for Pinecone
6123 3315 );
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, $source_url);
6128 - }
6129 -
3316 +
6130 3317 if (is_wp_error($result)) {
6131 - //error_log('MxChat WooCommerce sync failed for product ' . $product_id . ': ' . $result->get_error_message());
3318 + //error_log('MxChat WooCommerce sync failed for product ' . $product->get_id() . ': ' . $result->get_error_message());
6132 3319 }
6133 3320 }
6134 3321
6135 3322 public function mxchat_handle_product_delete($post_id) {
@@ -6136,18 +3323,28 @@
6136 3323 if (get_post_type($post_id) !== 'product') {
6137 3324 return;
6138 3325 }
6139 3326
6140 - $source_url = $this->mxchat_resolve_pre_trash_url($post_id);
6141 - if (!$source_url) {
6142 - return;
6143 - }
3327 + $source_url = get_permalink($post_id);
6144 3328
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');
3329 + // Check if Pinecone is enabled
3330 + $pinecone_options = get_option('mxchat_pinecone_addon_options', array());
3331 + $use_pinecone = ($pinecone_options['mxchat_use_pinecone'] ?? '0') === '1';
6147 3332
6148 - delete_transient('mxchat_prev_url_' . $post_id);
6149 - delete_transient('mxchat_prev_status_' . $post_id);
3333 + if ($use_pinecone && !empty($pinecone_options['mxchat_pinecone_api_key'])) {
3334 + // Delete from Pinecone
3335 + $this->mxchat_delete_from_pinecone_by_url($source_url, $pinecone_options);
3336 + } else {
3337 + // Delete from WordPress DB
3338 + global $wpdb;
3339 + $table_name = $wpdb->prefix . 'mxchat_system_prompt_content';
3340 +
3341 + $wpdb->delete(
3342 + $table_name,
3343 + array('source_url' => $source_url),
3344 + array('%s')
3345 + );
3346 + }
6150 3347 }
6151 3348
6152 3349 /**
6153 3350 * Handle individual Pinecone content deletion
@@ -6195,9 +3392,21 @@
6195 3392 $pinecone_options['mxchat_pinecone_host']
6196 3393 );
6197 3394
6198 3395 if ($result['success']) {
6199 - // No cache clearing needed since we removed caching
3396 + // Remove from ALL caches
3397 + $pinecone_manager->mxchat_remove_from_pinecone_vector_cache($vector_id);
3398 + $pinecone_manager->mxchat_remove_from_processed_content_caches($vector_id);
3399 +
3400 + // CLEAR ALL RELEVANT CACHES
3401 + delete_transient('mxchat_pinecone_recent_1k_cache');
3402 + delete_option('mxchat_pinecone_vector_ids_cache');
3403 + delete_option('mxchat_pinecone_processed_cache');
3404 + delete_option('mxchat_processed_content_cache');
3405 +
3406 + // Also force refresh for next page load
3407 + $pinecone_manager->mxchat_refresh_after_new_content($pinecone_options);
3408 +
6200 3409 set_transient('mxchat_admin_notice_success',
6201 3410 esc_html__('Entry deleted successfully from Pinecone.', 'mxchat'),
6202 3411 30
6203 3412 );
@@ -6210,11 +3419,9 @@
6210 3419
6211 3420 wp_safe_redirect(admin_url('admin.php?page=mxchat-prompts'));
6212 3421 exit;
6213 3422 }
6214 -/**
6215 - * Handle individual Pinecone content deletion via AJAX
6216 - */
3423 +
6217 3424 public function ajax_mxchat_delete_pinecone_prompt() {
6218 3425 // Verify nonce and permissions
6219 3426 if (!check_ajax_referer('mxchat_delete_pinecone_prompt_nonce', 'nonce', false)) {
6220 3427 wp_send_json_error('Invalid nonce');
@@ -6226,9 +3433,8 @@
6226 3433 exit;
6227 3434 }
6228 3435
6229 3436 $vector_id = isset($_POST['vector_id']) ? sanitize_text_field($_POST['vector_id']) : '';
6230 - $bot_id = isset($_POST['bot_id']) ? sanitize_text_field($_POST['bot_id']) : 'default';
6231 3437
6232 3438 if (empty($vector_id)) {
6233 3439 wp_send_json_error('Missing vector ID');
6234 3440 exit;
@@ -6233,20 +3439,19 @@
6233 3439 wp_send_json_error('Missing vector ID');
6234 3440 exit;
6235 3441 }
6236 3442
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 -
3443 + // Get Pinecone settings
3444 + $pinecone_options = get_option('mxchat_pinecone_addon_options', array());
6241 3445 $use_pinecone = ($pinecone_options['mxchat_use_pinecone'] ?? '0') === '1';
6242 3446
6243 3447 if (!$use_pinecone || empty($pinecone_options['mxchat_pinecone_api_key'])) {
6244 - wp_send_json_error('Pinecone is not properly configured for bot: ' . $bot_id);
3448 + wp_send_json_error('Pinecone is not properly configured');
6245 3449 exit;
6246 3450 }
6247 3451
6248 - // Delete from the correct Pinecone index
3452 + // Delete from Pinecone
3453 + $pinecone_manager = MxChat_Pinecone_Manager::get_instance();
6249 3454 $result = $pinecone_manager->mxchat_delete_from_pinecone_by_vector_id(
6250 3455 $vector_id,
6251 3456 $pinecone_options['mxchat_pinecone_api_key'],
6252 3457 $pinecone_options['mxchat_pinecone_host']
@@ -6252,1007 +3457,32 @@
6252 3457 $pinecone_options['mxchat_pinecone_host']
6253 3458 );
6254 3459
6255 3460 if ($result['success']) {
6256 - // No cache clearing needed since we removed caching
3461 + // Remove from ALL caches
3462 + $pinecone_manager->mxchat_remove_from_pinecone_vector_cache($vector_id);
3463 + $pinecone_manager->mxchat_remove_from_processed_content_caches($vector_id);
3464 +
3465 + // CLEAR ALL RELEVANT CACHES (ADD THESE LINES)
3466 + delete_transient('mxchat_pinecone_recent_1k_cache');
3467 + delete_option('mxchat_pinecone_vector_ids_cache');
3468 + delete_option('mxchat_pinecone_processed_cache');
3469 + delete_option('mxchat_processed_content_cache');
3470 +
3471 + // Also force refresh for next page load
3472 + $pinecone_manager->mxchat_refresh_after_new_content($pinecone_options);
3473 +
6257 3474 wp_send_json_success(array(
6258 3475 'message' => 'Entry deleted successfully from Pinecone',
6259 - 'vector_id' => $vector_id,
6260 - 'bot_id' => $bot_id
3476 + 'vector_id' => $vector_id
6261 3477 ));
6262 3478 } 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 3479 wp_send_json_error('Failed to delete from Pinecone: ' . $result['message']);
6265 3480 }
6266 3481
6267 3482 exit;
6268 3483 }
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 3484
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 3485 // ========================================
7256 3486 // HELPER METHODS
7257 3487 // ========================================
7258 3488
@@ -7319,1074 +3549,8 @@
7319 3549 */
7320 3550 private function mxchat_get_pinecone_manager() {
7321 3551 return MxChat_Pinecone_Manager::get_instance();
7322 3552 }
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 3553
8390 3554 // ========================================
8391 3555 // STATIC ACCESS METHODS
8392 3556 // ========================================