← All changes | admin/class-knowledge-manager.php +98 -80 3.2.2 → 3.1.7 View file →

		@@ -59,8 +59,11 @@
59	59	add_action('wp_ajax_mxchat_paginate_entries', array($this, 'ajax_mxchat_paginate_entries'));
60	60	add_action('wp_ajax_mxchat_get_entry_content', array($this, 'ajax_mxchat_get_entry_content'));
61	61	add_action('wp_ajax_mxchat_save_entry_content', array($this, 'ajax_mxchat_save_entry_content'));
62	62
	63	+ // Hook for content deletion
	64	+ add_action('mxchat_delete_content', array($this, 'mxchat_delete_from_pinecone_by_url'), 10, 1);
	65	+
63	66	// WordPress post management hooks
64	67	add_action('pre_post_update', array($this, 'mxchat_store_pre_update_status'), 10, 2);
65	68	add_action('post_updated', array($this, 'mxchat_handle_post_update'), 10, 3);
66	69	add_action('before_delete_post', array($this, 'mxchat_handle_post_delete'));
		@@ -990,18 +993,10 @@
990	993	wp_safe_redirect(esc_url(admin_url('admin.php?page=mxchat-prompts')));
991	994	exit;
992	995	}
993	996
994		- // Fetch URL — use browser-like headers so servers with bot protection don't block us
995		- $response = wp_remote_get($submitted_url, array(
996		- 'timeout' => 30,
997		- 'sslverify' => false,
998		- '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',
999		- 'headers' => array(
1000		- 'Accept' => 'text/html,application/xhtml+xml,application/xml;q=0.9,*/*;q=0.8',
1001		- 'Accept-Language' => 'en-US,en;q=0.9',
1002		- ),
1003		- ));
	997	+ // Fetch URL
	998	+ $response = wp_remote_get($submitted_url, array('timeout' => 30));
1004	999
1005	1000	if (is_wp_error($response) || wp_remote_retrieve_response_code($response) !== 200) {
1006	1001	$error_message = is_wp_error($response) ? $response->get_error_message() : 'HTTP Status: ' . wp_remote_retrieve_response_code($response);
1007	1002	set_transient('mxchat_admin_notice_error',
		@@ -2835,12 +2830,11 @@
2835	2830	foreach ($primary_indexes as $path => $source) {
2836	2831	$url = trailingslashit($site_url) . $path;
2837	2832
2838	2833	$response = wp_remote_head($url, array(
2839		- 'timeout' => 10,
	2834	+ 'timeout' => 3, // Short timeout
2840	2835	'sslverify' => false,
2841		- 'redirection' => 1,
2842		- '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',
	2836	+ 'redirection' => 1
2843	2837	));
2844	2838
2845	2839	if (!is_wp_error($response) && wp_remote_retrieve_response_code($response) === 200) {
2846	2840	// Found a sitemap index - parse it to get sub-sitemaps
		@@ -2898,15 +2892,10 @@
2898	2892	private function parse_sitemap_index($url) {
2899	2893	$sub_sitemaps = array();
2900	2894
2901	2895	$response = wp_remote_get($url, array(
2902		- 'timeout' => 30,
2903		- 'sslverify' => false,
2904		- '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',
2905		- 'headers' => array(
2906		- 'Accept' => 'text/html,application/xhtml+xml,application/xml;q=0.9,*/*;q=0.8',
2907		- 'Accept-Language' => 'en-US,en;q=0.9',
2908		- ),
	2896	+ 'timeout' => 5,
	2897	+ 'sslverify' => false
2909	2898	));
2910	2899
2911	2900	if (is_wp_error($response)) {
2912	2901	return $sub_sitemaps;
		@@ -2957,15 +2946,10 @@
2957	2946	* Get URL count from a sitemap
2958	2947	*/
2959	2948	private function get_sitemap_url_count($url) {
2960	2949	$response = wp_remote_get($url, array(
2961		- 'timeout' => 30,
2962		- 'sslverify' => false,
2963		- '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',
2964		- 'headers' => array(
2965		- 'Accept' => 'text/html,application/xhtml+xml,application/xml;q=0.9,*/*;q=0.8',
2966		- 'Accept-Language' => 'en-US,en;q=0.9',
2967		- ),
	2950	+ 'timeout' => 10,
	2951	+ 'sslverify' => false
2968	2952	));
2969	2953
2970	2954	if (is_wp_error($response)) {
2971	2955	return 0;
		@@ -2988,11 +2972,10 @@
2988	2972	$sitemaps = array();
2989	2973	$robots_url = trailingslashit($site_url) . 'robots.txt';
2990	2974
2991	2975	$response = wp_remote_get($robots_url, array(
2992		- 'timeout' => 15,
2993		- 'sslverify' => false,
2994		- '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',
	2976	+ 'timeout' => 5,
	2977	+ 'sslverify' => false
2995	2978	));
2996	2979
2997	2980	if (is_wp_error($response)) {
2998	2981	return $sitemaps;
		@@ -5181,33 +5164,40 @@
5181	5164	// If the post was previously published but is now not published, remove from knowledge base
5182	5165	if ($previous_status === 'publish' && $post->post_status !== 'publish') {
5183	5166	// Use the stored URL from when it was published, or fall back to current permalink
5184	5167	$source_url = $previous_url ?: get_permalink($post_id);
	5168	+
	5169	+ if ($source_url) {
	5170	+ // Check if Pinecone is enabled
	5171	+ $pinecone_options = get_option('mxchat_pinecone_addon_options', array());
	5172	+ $use_pinecone = ($pinecone_options['mxchat_use_pinecone'] ?? '0') === '1';
5185	5173
5186		- if ($source_url) {
5187		- // Chunk-aware deletion (routes to Pinecone or WP DB and removes base + all chunks)
5188		- MxChat_Utils::delete_chunks_for_url($source_url, 'default');
	5174	+ if ($use_pinecone && !empty($pinecone_options['mxchat_pinecone_api_key'])) {
	5175	+ // Delete from Pinecone
	5176	+ $this->mxchat_delete_from_pinecone_by_url($source_url, $pinecone_options);
	5177	+ } else {
	5178	+ // Delete from WordPress DB
	5179	+ global $wpdb;
	5180	+ $table_name = $wpdb->prefix . 'mxchat_system_prompt_content';
	5181	+
	5182	+ $result = $wpdb->delete(
	5183	+ $table_name,
	5184	+ array('source_url' => $source_url),
	5185	+ array('%s')
	5186	+ );
	5187	+ }
5189	5188	}
5190		-
	5189	+
5191	5190	// Clean up the transients and exit early
5192	5191	delete_transient($previous_status_key);
5193	5192	delete_transient($previous_url_key);
5194	5193	return;
5195	5194	}
5196		-
5197		- // Slug/permalink rename while still published: delete the old vectors before upserting new ones.
5198		- // Without this, md5(old_url) vectors (base + chunks) would be orphaned under the stale URL.
5199		- if ($post->post_status === 'publish' && !empty($previous_url)) {
5200		- $current_url = get_permalink($post_id);
5201		- if ($current_url && $current_url !== $previous_url) {
5202		- MxChat_Utils::delete_chunks_for_url($previous_url, 'default');
5203		- }
5204		- }
5205		-
	5195	+
5206	5196	// Store the current status for next time (if this is an update)
5207	5197	if ($update) {
5208	5198	set_transient($previous_status_key, $post->post_status, DAY_IN_SECONDS);
5209		-
	5199	+
5210	5200	// If the post is currently published, also store its URL
5211	5201	if ($post->post_status === 'publish') {
5212	5202	$current_url = get_permalink($post_id);
5213	5203	set_transient($previous_url_key, $current_url, DAY_IN_SECONDS);
		@@ -5437,14 +5427,12 @@
5437	5427	if (!$should_sync) {
5438	5428	return;
5439	5429	}
5440	5430
5441		- // Resolve the pre-trash URL. wp_trash_post renames the slug with "__trashed" before firing
5442		- // this hook, so get_permalink() here would return the trashed URL and md5() would miss the
5443		- // real vector IDs stored under the original URL.
5444		- $source_url = $this->mxchat_resolve_pre_trash_url($post_id);
	5431	+ // Get the URL before post is deleted
	5432	+ $source_url = get_permalink($post_id);
5445	5433	if (!$source_url) {
5446		- //error_log('MXChat: Failed to resolve source URL for post ' . $post_id);
	5434	+ //error_log('MXChat: Failed to get permalink for post ' . $post_id);
5447	5435	return;
5448	5436	}
5449	5437
5450	5438	// Use chunk-aware deletion (handles both chunked and non-chunked content)
		@@ -5452,36 +5440,56 @@
5452	5440
5453	5441	if (is_wp_error($delete_result)) {
5454	5442	//error_log('MXChat: Chunk-aware deletion failed for URL: ' . $source_url . ' - ' . $delete_result->get_error_message());
5455	5443	}
	5444	+}
5456	5445
5457		- delete_transient('mxchat_prev_url_' . $post_id);
5458		- delete_transient('mxchat_prev_status_' . $post_id);
5459		-}
5460	5446
5461		-/**
5462		- * Resolve the source URL for a post being trashed/deleted.
5463		- *
5464		- * Why: wp_trash_post appends "__trashed" to the slug before the wp_trash_post action fires, so
5465		- * get_permalink() returns a URL whose md5() won't match the vector IDs stored in Pinecone or
5466		- * the source_url rows in the WP DB. Prefer the URL captured by mxchat_store_pre_update_status
5467		- * (runs on pre_post_update, before the rename); fall back to stripping the __trashed suffix.
5468		- */
5469		-private function mxchat_resolve_pre_trash_url($post_id) {
5470		- $previous_url = get_transient('mxchat_prev_url_' . $post_id);
5471		- if (!empty($previous_url)) {
5472		- return $previous_url;
5473		- }
	5447	+ /**
	5448	+ * Deletes data from Pinecone using a source URL
	5449	+ */
	5450	+ public function mxchat_delete_from_pinecone_by_url($source_url, $pinecone_options) {
	5451	+ $host = $pinecone_options['mxchat_pinecone_host'] ?? '';
	5452	+ $api_key = $pinecone_options['mxchat_pinecone_api_key'] ?? '';
5474	5453
5475		- $current = get_permalink($post_id);
5476		- if (!$current) {
5477		- return '';
5478		- }
5479		- return preg_replace('#__trashed(/?)$#', '$1', $current);
5480		-}
	5454	+ if (empty($host) || empty($api_key)) {
	5455	+ //error_log('MXChat: Pinecone deletion failed - missing configuration');
	5456	+ return false;
	5457	+ }
5481	5458
	5459	+ $api_endpoint = "https://{$host}/vectors/delete";
	5460	+ $vector_id = md5($source_url);
5482	5461
	5462	+ $request_body = array(
	5463	+ 'ids' => array($vector_id)
	5464	+ );
5483	5465
	5466	+ $response = wp_remote_post($api_endpoint, array(
	5467	+ 'headers' => array(
	5468	+ 'Api-Key' => $api_key,
	5469	+ 'accept' => 'application/json',
	5470	+ 'content-type' => 'application/json'
	5471	+ ),
	5472	+ 'body' => wp_json_encode($request_body),
	5473	+ 'timeout' => 30
	5474	+ ));
	5475	+
	5476	+ if (is_wp_error($response)) {
	5477	+ //error_log('MXChat: Pinecone deletion error - ' . $response->get_error_message());
	5478	+ return false;
	5479	+ }
	5480	+
	5481	+ $response_code = wp_remote_retrieve_response_code($response);
	5482	+ if ($response_code !== 200) {
	5483	+ //error_log('MXChat: Pinecone deletion failed with status ' . $response_code);
	5484	+ return false;
	5485	+ }
	5486	+
	5487	+ return true;
	5488	+ }
	5489	+
	5490	+
	5491	+
5484	5492	public function mxchat_handle_product_change($post_id, $post, $update) {
5485	5493	if ($post->post_type !== 'product') {
5486	5494	return;
5487	5495	}
		@@ -5633,18 +5641,28 @@
5633	5641	if (get_post_type($post_id) !== 'product') {
5634	5642	return;
5635	5643	}
5636	5644
5637		- $source_url = $this->mxchat_resolve_pre_trash_url($post_id);
5638		- if (!$source_url) {
5639		- return;
5640		- }
	5645	+ $source_url = get_permalink($post_id);
5641	5646
5642		- // Chunk-aware deletion (routes to Pinecone or WP DB and removes base + all chunks)
5643		- MxChat_Utils::delete_chunks_for_url($source_url, 'default');
	5647	+ // Check if Pinecone is enabled
	5648	+ $pinecone_options = get_option('mxchat_pinecone_addon_options', array());
	5649	+ $use_pinecone = ($pinecone_options['mxchat_use_pinecone'] ?? '0') === '1';
5644	5650
5645		- delete_transient('mxchat_prev_url_' . $post_id);
5646		- delete_transient('mxchat_prev_status_' . $post_id);
	5651	+ if ($use_pinecone && !empty($pinecone_options['mxchat_pinecone_api_key'])) {
	5652	+ // Delete from Pinecone
	5653	+ $this->mxchat_delete_from_pinecone_by_url($source_url, $pinecone_options);
	5654	+ } else {
	5655	+ // Delete from WordPress DB
	5656	+ global $wpdb;
	5657	+ $table_name = $wpdb->prefix . 'mxchat_system_prompt_content';
	5658	+
	5659	+ $wpdb->delete(
	5660	+ $table_name,
	5661	+ array('source_url' => $source_url),
	5662	+ array('%s')
	5663	+ );
	5664	+ }
5647	5665	}
5648	5666
5649	5667	/**
5650	5668	* Handle individual Pinecone content deletion