← All changes | includes/class-mxchat-utils.php +5 -57 3.2.3 → 3.2.2 View file →

		@@ -5,51 +5,8 @@
5	5
6	6	class MxChat_Utils {
7	7
8	8	/**
9		- * Centralized embedding model registry. Single source of truth for dimensions
10		- * and provider, so model-switch protection logic doesn't drift across files.
11		- */
12		-public static function embedding_model_registry() {
13		- return array(
14		- 'text-embedding-ada-002' => array('dims' => 1536, 'provider' => 'openai', 'label' => 'Ada 2'),
15		- 'text-embedding-3-small' => array('dims' => 1536, 'provider' => 'openai', 'label' => 'TE3 Small'),
16		- 'text-embedding-3-large' => array('dims' => 3072, 'provider' => 'openai', 'label' => 'TE3 Large'),
17		- 'voyage-3-large' => array('dims' => 2048, 'provider' => 'voyage', 'label' => 'Voyage-3 Large'),
18		- 'gemini-embedding-001' => array('dims' => 1536, 'provider' => 'gemini', 'label' => 'Gemini Embedding'),
19		- );
20		-}
21		-
22		-public static function embedding_model_dimensions($model) {
23		- $registry = self::embedding_model_registry();
24		- return isset($registry[$model]) ? (int) $registry[$model]['dims'] : 0;
25		-}
26		-
27		-public static function embedding_model_label($model) {
28		- $registry = self::embedding_model_registry();
29		- return isset($registry[$model]) ? $registry[$model]['label'] : $model;
30		-}
31		-
32		-/**
33		- * Returns the model that was last used to actually write embeddings into the
34		- * KB. Differs from the user-selected setting once a switch has happened but
35		- * no re-embed has occurred yet — that's the mismatch state we warn about.
36		- */
37		-public static function get_active_embedding_model() {
38		- return get_option('mxchat_active_embedding_model', '');
39		-}
40		-
41		-/**
42		- * Stamp the model that produced the most recent successful embedding. Called
43		- * from generate_embedding() right after the API responds with a valid vector.
44		- */
45		-public static function stamp_active_embedding_model($model) {
46		- if (!empty($model) && $model !== self::get_active_embedding_model()) {
47		- update_option('mxchat_active_embedding_model', $model, false);
48		- }
49		-}
50		-
51		-/**
52	9	* UPDATED: Submit or update content (and its embedding) in the database.
53	10	* Stores in Pinecone if enabled, otherwise stores in WordPress DB.
54	11	*
55	12	* @param string $content The content to be embedded.
		@@ -255,10 +212,8 @@
255	212	$max_attempts = 3;
256	213	$current_content = $safe_content;
257	214	$result = false;
258	215
259		- $active_model = self::get_active_embedding_model();
260		-
261	216	while ($attempt <= $max_attempts && $result === false) {
262	217	try {
263	218	if ($existing_id) {
264	219	//error_log('[MXCHAT-DB] Found existing entry (ID: ' . $existing_id . '). Updating... (Attempt ' . $attempt . ')');
		@@ -271,13 +226,12 @@
271	226	'article_content' => $current_content,
272	227	'embedding_vector' => $embedding_vector_serialized,
273	228	'source_url' => $source_url,
274	229	'content_type' => $content_type,
275		- 'embedding_model' => $active_model,
276	230	'timestamp' => current_time('mysql'),
277	231	),
278	232	array('id' => $existing_id),
279		- array('%s','%s','%s','%s','%s','%s','%s'),
	233	+ array('%s','%s','%s','%s','%s','%s'),
280	234	array('%d')
281	235	);
282	236	} else {
283	237	//error_log('[MXCHAT-DB] No existing entry found. Inserting new row... (Attempt ' . $attempt . ')');
		@@ -291,12 +245,11 @@
291	245	'article_content' => $current_content,
292	246	'embedding_vector' => $embedding_vector_serialized,
293	247	'source_url' => $source_url, // Now unique for manual content
294	248	'content_type' => $content_type,
295		- 'embedding_model' => $active_model,
296	249	'timestamp' => current_time('mysql'),
297	250	),
298		- array('%s','%s','%s','%s','%s','%s','%s')
	251	+ array('%s','%s','%s','%s','%s','%s')
299	252	);
300	253	}
301	254
302	255	if ($result === false) {
		@@ -418,10 +371,9 @@
418	371	'source_url' => $url, // Can be empty for manual content
419	372	'type' => $content_type, // Now supports: post, page, pdf, url, manual, product, etc.
420	373	'last_updated' => time(),
421	374	'created_at' => time(), // Add creation timestamp
422		- 'bot_id' => $bot_id, // Add bot identification
423		- 'embedding_model' => self::get_active_embedding_model() // 3.2.3: track which model produced this vector
	375	+ 'bot_id' => $bot_id // Add bot identification
424	376	);
425	377
426	378	$vector_data = array(
427	379	'id' => $vector_id,
		@@ -569,9 +521,8 @@
569	521	// Handle different response formats based on provider
570	522	if (strpos($selected_model, 'gemini-embedding') === 0) {
571	523	// Gemini API response format
572	524	if (isset($response_body['embedding']['values']) && is_array($response_body['embedding']['values'])) {
573		- self::stamp_active_embedding_model($selected_model);
574	525	return $response_body['embedding']['values'];
575	526	} else {
576	527	//error_log('Invalid response received from Gemini embedding API for bot ' . $bot_id . ': ' . wp_json_encode($response_body));
577	528	return null;
		@@ -578,9 +529,8 @@
578	529	}
579	530	} else {
580	531	// OpenAI/Voyage API response format
581	532	if (isset($response_body['data'][0]['embedding']) && is_array($response_body['data'][0]['embedding'])) {
582		- self::stamp_active_embedding_model($selected_model);
583	533	return $response_body['data'][0]['embedding'];
584	534	} else {
585	535	//error_log('Invalid response received from embedding API for bot ' . $bot_id . ': ' . wp_json_encode($response_body));
586	536	return null;
		@@ -736,10 +686,9 @@
736	686	'total_chunks' => $chunk_metadata['total_chunks'],
737	687	'parent_url_hash' => $chunk_metadata['parent_url_hash'],
738	688	'last_updated' => time(),
739	689	'created_at' => time(),
740		- 'bot_id' => $bot_id,
741		- 'embedding_model' => self::get_active_embedding_model()
	690	+ 'bot_id' => $bot_id
742	691	);
743	692
744	693	$vector_data = array(
745	694	'id' => $vector_id,
		@@ -792,12 +741,11 @@
792	741	'article_content' => $content_with_metadata,
793	742	'embedding_vector' => $embedding_vector_serialized,
794	743	'source_url' => $source_url,
795	744	'content_type' => $content_type,
796		- 'embedding_model' => self::get_active_embedding_model(),
797	745	'timestamp' => current_time('mysql')
798	746	),
799		- array('%s', '%s', '%s', '%s', '%s', '%s', '%s')
	747	+ array('%s', '%s', '%s', '%s', '%s', '%s')
800	748	);
801	749
802	750	if ($result === false) {
803	751	return new WP_Error('database_failed', 'Failed to insert chunk: ' . $wpdb->last_error);