← All changes | includes/class-mxchat-utils.php +328 -70 2.0.5 → 2.2.1 View file →

		@@ -4,112 +4,370 @@
4	4	}
5	5
6	6	class MxChat_Utils {
7	7
8		- /**
9		- * Submit or update content (and its embedding) in the database.
10		- *
11		- * If the source_url already exists, update that row.
12		- * Otherwise, insert a new row.
13		- *
14		- * @param string $content The content to be embedded.
15		- * @param string $source_url The source URL of the content.
16		- * @param string $api_key The API key used for generating embeddings.
17		- */
18		- public static function submit_content_to_db($content, $source_url, $api_key) {
19		- global $wpdb;
20		- $table_name = $wpdb->prefix . 'mxchat_system_prompt_content';
	8	+/**
	9	+ * Submit or update content (and its embedding) in the database.
	10	+ * Stores in Pinecone if enabled, otherwise stores in WordPress DB.
	11	+ *
	12	+ * @param string $content The content to be embedded.
	13	+ * @param string $source_url The source URL of the content.
	14	+ * @param string $api_key The API key used for generating embeddings.
	15	+ * @param string $vector_id Optional vector ID for Pinecone (if not provided, will use md5 of URL)
	16	+ * @return bool|WP_Error True on success, WP_Error on failure
	17	+ */
	18	+public static function submit_content_to_db($content, $source_url, $api_key, $vector_id = null) {
	19	+ global $wpdb;
	20	+ $table_name = $wpdb->prefix . 'mxchat_system_prompt_content';
	21	+
	22	+ error_log('[MXCHAT-DB] Starting database submission for URL: ' . $source_url);
	23	+ error_log('[MXCHAT-DB] Content length: ' . strlen($content) . ' bytes');
	24	+
	25	+ // Sanitize the source URL
	26	+ $source_url = esc_url_raw($source_url);
	27	+
	28	+ // Just ensure UTF-8 validity without aggressive escaping
	29	+ $safe_content = wp_check_invalid_utf8($content);
	30	+ // Remove only null bytes and other control characters, but preserve newlines (\n = \x0A) and carriage returns (\r = \x0D)
	31	+ $safe_content = preg_replace('/[\x00-\x08\x0B\x0C\x0E-\x1F\x7F]/u', '', $safe_content);
21	32
22		- // Sanitize the source URL
23		- $source_url = esc_url_raw($source_url);
24	33
25		- // Generate the embedding using the API key
26		- $embedding_vector = self::generate_embedding($content, $api_key);
	34	+ // Generate the embedding using the API key
	35	+ $embedding_vector = self::generate_embedding($content, $api_key);
	36	+
	37	+ if (!is_array($embedding_vector)) {
	38	+ error_log('[MXCHAT-DB] Error: Embedding generation failed');
	39	+ return new WP_Error('embedding_failed', 'Failed to generate embedding for content');
	40	+ }
	41	+
	42	+ error_log('[MXCHAT-DB] Embedding generated successfully');
	43	+
	44	+ // Check if Pinecone is enabled and configured
	45	+ if (self::is_pinecone_enabled()) {
	46	+ error_log('[MXCHAT-DB] Pinecone is enabled - using Pinecone storage');
	47	+ // Store in Pinecone only
	48	+ return self::store_in_pinecone_only($embedding_vector, $content, $source_url, $vector_id);
	49	+ } else {
	50	+ error_log('[MXCHAT-DB] Pinecone not enabled - using WordPress storage');
	51	+ // Store in WordPress database only
	52	+ $embedding_vector_serialized = maybe_serialize($embedding_vector);
	53	+ return self::store_in_wordpress_db($safe_content, $source_url, $embedding_vector_serialized, $table_name);
	54	+ }
	55	+}
27	56
28		- if (is_array($embedding_vector)) {
29		- $embedding_vector_serialized = maybe_serialize($embedding_vector);
	57	+/**
	58	+ * Check if Pinecone is enabled and properly configured
	59	+ */
	60	+private static function is_pinecone_enabled() {
	61	+ $pinecone_options = get_option('mxchat_pinecone_addon_options');
	62	+
	63	+ if (empty($pinecone_options)) {
	64	+ return false;
	65	+ }
	66	+
	67	+ $enabled_check = !empty($pinecone_options['mxchat_use_pinecone']) && $pinecone_options['mxchat_use_pinecone'] !== '0';
	68	+ $api_key_check = !empty($pinecone_options['mxchat_pinecone_api_key']);
	69	+ $host_check = !empty($pinecone_options['mxchat_pinecone_host']);
	70	+
	71	+ return $enabled_check && $api_key_check && $host_check;
	72	+}
30	73
31		- // Check if a row already exists for this URL
32		- $existing_id = $wpdb->get_var(
33		- $wpdb->prepare(
34		- "SELECT id FROM {$table_name} WHERE source_url = %s LIMIT 1",
35		- $source_url
36		- )
37		- );
	74	+/**
	75	+ * Store content in Pinecone only
	76	+ */
	77	+private static function store_in_pinecone_only($embedding_vector, $content, $source_url, $vector_id = null) {
	78	+ error_log('[MXCHAT-PINECONE] ===== Using Pinecone-only storage =====');
	79	+
	80	+ $pinecone_options = get_option('mxchat_pinecone_addon_options');
	81	+
	82	+ $result = self::store_in_pinecone_main(
	83	+ $embedding_vector,
	84	+ $content,
	85	+ $source_url,
	86	+ $pinecone_options['mxchat_pinecone_api_key'],
	87	+ $pinecone_options['mxchat_pinecone_environment'] ?? '',
	88	+ $pinecone_options['mxchat_pinecone_index'] ?? '',
	89	+ $vector_id
	90	+ );
	91	+
	92	+ if (is_wp_error($result)) {
	93	+ error_log('[MXCHAT-PINECONE] Pinecone storage failed: ' . $result->get_error_message());
	94	+ return $result;
	95	+ }
	96	+
	97	+ error_log('[MXCHAT-PINECONE] Pinecone storage completed successfully');
	98	+ return true;
	99	+}
38	100
	101	+/**
	102	+ * Store content in WordPress database with progressive fallback
	103	+ */
	104	+private static function store_in_wordpress_db($safe_content, $source_url, $embedding_vector_serialized, $table_name) {
	105	+ global $wpdb;
	106	+
	107	+ error_log('[MXCHAT-DB] ===== Using WordPress-only storage =====');
	108	+
	109	+ // Check if a row already exists for this URL
	110	+ $existing_id = $wpdb->get_var(
	111	+ $wpdb->prepare(
	112	+ "SELECT id FROM {$table_name} WHERE source_url = %s LIMIT 1",
	113	+ $source_url
	114	+ )
	115	+ );
	116	+
	117	+ // Progressive fallback mechanism for problematic content
	118	+ $attempt = 1;
	119	+ $max_attempts = 3;
	120	+ $current_content = $safe_content;
	121	+ $result = false;
	122	+
	123	+ while ($attempt <= $max_attempts && $result === false) {
	124	+ try {
39	125	if ($existing_id) {
	126	+ error_log('[MXCHAT-DB] Found existing entry (ID: ' . $existing_id . '). Updating... (Attempt ' . $attempt . ')');
	127	+
40	128	// Update the existing row
41		- $wpdb->update(
	129	+ $result = $wpdb->update(
42	130	$table_name,
43	131	array(
44		- 'article_content' => $content,
	132	+ 'url' => $source_url,
	133	+ 'article_content' => $current_content,
45	134	'embedding_vector' => $embedding_vector_serialized,
46		- 'timestamp' => current_time('mysql'), // Remove if your table doesn't have a timestamp column
	135	+ 'source_url' => $source_url,
	136	+ 'timestamp' => current_time('mysql'),
47	137	),
48	138	array('id' => $existing_id),
49		- array('%s','%s','%s'),
	139	+ array('%s','%s','%s','%s','%s'),
50	140	array('%d')
51	141	);
52	142	} else {
	143	+ error_log('[MXCHAT-DB] No existing entry found. Inserting new row... (Attempt ' . $attempt . ')');
	144	+ error_log('[MXCHAT-DB] Content sample: ' . substr($current_content, 0, 1000));
	145	+
53	146	// Insert a new row
54		- $wpdb->insert(
	147	+ $result = $wpdb->insert(
55	148	$table_name,
56	149	array(
57		- 'article_content' => $content,
	150	+ 'url' => $source_url,
	151	+ 'article_content' => $current_content,
58	152	'embedding_vector' => $embedding_vector_serialized,
59	153	'source_url' => $source_url,
60		- 'timestamp' => current_time('mysql'), // Remove if your table doesn't have a timestamp column
	154	+ 'timestamp' => current_time('mysql'),
61	155	),
62		- array('%s','%s','%s','%s')
	156	+ array('%s','%s','%s','%s','%s')
63	157	);
64	158	}
65		- } else {
66		- // If embedding generation failed, you could log or handle it here
67		- // error_log('Embedding generation failed for content from ' . $source_url);
	159	+
	160	+ if ($result === false) {
	161	+ error_log('[MXCHAT-DB] Database operation failed (Attempt ' . $attempt . '): ' . $wpdb->last_error);
	162	+
	163	+ // Progressively apply more aggressive sanitization on failure
	164	+ if ($attempt === 1) {
	165	+ // First fallback: Use a more aggressive character filter and shorten
	166	+ $current_content = preg_replace('/[^\p{L}\p{N}\s.,;:!?()-]/u', '', $current_content);
	167	+ $current_content = substr($current_content, 0, 50000);
	168	+ } else if ($attempt === 2) {
	169	+ // Second fallback: Keep only alphanumeric and basic punctuation, shorten further
	170	+ $current_content = preg_replace('/[^a-zA-Z0-9\s.,;:!?()-]/u', '', $current_content);
	171	+ $current_content = substr($current_content, 0, 30000);
	172	+ }
	173	+
	174	+ $attempt++;
	175	+ }
	176	+ } catch (Exception $e) {
	177	+ error_log('[MXCHAT-DB] Exception during database operation: ' . $e->getMessage());
	178	+ $attempt++;
68	179	}
69	180	}
	181	+
	182	+ if ($result === false) {
	183	+ error_log('[MXCHAT-DB] All database operation attempts failed');
	184	+ return new WP_Error('database_failed', 'Failed to store content in WordPress database after ' . $max_attempts . ' attempts');
	185	+ }
	186	+
	187	+ error_log('[MXCHAT-DB] WordPress database operation completed successfully (Attempt ' . ($attempt - 1) . ')');
	188	+ return true;
	189	+}
70	190
71		- /**
72		- * Generate an embedding for the given text using the specified API key.
73		- *
74		- * @param string $text The text to be embedded.
75		- * @param string $api_key The API key used for generating embeddings.
76		- * @return array|null The embedding vector or null on failure.
77		- */
78		- private static function generate_embedding($text, $api_key) {
79		- $endpoint = 'https://api.openai.com/v1/embeddings';
	191	+/**
	192	+ * Store content in Pinecone database
	193	+ */
	194	+private static function store_in_pinecone_main($embedding_vector, $content, $url, $api_key, $environment, $index_name, $vector_id = null) {
	195	+ error_log('[MXCHAT-PINECONE-MAIN] ===== Starting Pinecone storage =====');
	196	+
	197	+ $vector_id = $vector_id ?: md5($url);
	198	+ error_log('[MXCHAT-PINECONE-MAIN] Vector ID: ' . $vector_id);
	199	+
	200	+ $options = get_option('mxchat_pinecone_addon_options');
	201	+ $host = $options['mxchat_pinecone_host'] ?? '';
	202	+
	203	+ error_log('[MXCHAT-PINECONE-MAIN] Host from options: ' . $host);
	204	+ error_log('[MXCHAT-PINECONE-MAIN] API key length: ' . strlen($api_key));
	205	+ error_log('[MXCHAT-PINECONE-MAIN] Environment: ' . $environment);
	206	+ error_log('[MXCHAT-PINECONE-MAIN] Index name: ' . $index_name);
80	207
81		- $body = wp_json_encode([
82		- 'input' => $text,
83		- 'model' => 'text-embedding-ada-002'
84		- ]);
	208	+ if (empty($host)) {
	209	+ error_log('[MXCHAT-PINECONE-MAIN] ERROR: Host is empty');
	210	+ return new WP_Error('pinecone_config', 'Pinecone host is not configured. Please set the host in your settings.');
	211	+ }
85	212
86		- $args = [
87		- 'body' => $body,
88		- 'headers' => [
89		- 'Content-Type' => 'application/json',
90		- 'Authorization' => 'Bearer ' . $api_key,
	213	+ // Determine if this is a product URL
	214	+ $is_product = (strpos($url, '/product/') !== false || strpos($url, '/shop/') !== false);
	215	+ error_log('[MXCHAT-PINECONE-MAIN] Is product URL: ' . ($is_product ? 'YES' : 'NO'));
	216	+
	217	+ $api_endpoint = "https://{$host}/vectors/upsert";
	218	+ error_log('[MXCHAT-PINECONE-MAIN] API endpoint: ' . $api_endpoint);
	219	+
	220	+ $request_body = array(
	221	+ 'vectors' => array(
	222	+ array(
	223	+ 'id' => $vector_id,
	224	+ 'values' => $embedding_vector,
	225	+ 'metadata' => array(
	226	+ 'text' => $content,
	227	+ 'source_url' => $url,
	228	+ 'type' => $is_product ? 'product' : 'content',
	229	+ 'last_updated' => time()
	230	+ )
	231	+ )
	232	+ )
	233	+ );
	234	+
	235	+ error_log('[MXCHAT-PINECONE-MAIN] Request body prepared (embedding dimensions: ' . count($embedding_vector) . ')');
	236	+
	237	+ $response = wp_remote_post($api_endpoint, array(
	238	+ 'headers' => array(
	239	+ 'Api-Key' => $api_key,
	240	+ 'accept' => 'application/json',
	241	+ 'content-type' => 'application/json'
	242	+ ),
	243	+ 'body' => wp_json_encode($request_body),
	244	+ 'timeout' => 30,
	245	+ 'data_format' => 'body'
	246	+ ));
	247	+
	248	+ if (is_wp_error($response)) {
	249	+ error_log('[MXCHAT-PINECONE-MAIN] WordPress request error: ' . $response->get_error_message());
	250	+ return new WP_Error('pinecone_request', $response->get_error_message());
	251	+ }
	252	+
	253	+ $response_code = wp_remote_retrieve_response_code($response);
	254	+ error_log('[MXCHAT-PINECONE-MAIN] Response code: ' . $response_code);
	255	+
	256	+ if ($response_code !== 200) {
	257	+ $body = wp_remote_retrieve_body($response);
	258	+ error_log('[MXCHAT-PINECONE-MAIN] API error - Response body: ' . $body);
	259	+ return new WP_Error('pinecone_api', sprintf(
	260	+ 'Pinecone API error (HTTP %d): %s',
	261	+ $response_code,
	262	+ $body
	263	+ ));
	264	+ }
	265	+
	266	+ $response_body = wp_remote_retrieve_body($response);
	267	+ error_log('[MXCHAT-PINECONE-MAIN] Success response: ' . $response_body);
	268	+ error_log('[MXCHAT-PINECONE-MAIN] Successfully stored in Pinecone');
	269	+ error_log('[MXCHAT-PINECONE-MAIN] ===== Pinecone storage complete =====');
	270	+
	271	+ return true;
	272	+}
	273	+
	274	+/**
	275	+ * Generate an embedding for the given text using the specified API key.
	276	+ *
	277	+ * @param string $text The text to be embedded.
	278	+ * @param string $api_key The API key used for generating embeddings.
	279	+ * @return array|null The embedding vector or null on failure.
	280	+ */
	281	+private static function generate_embedding($text, $api_key) {
	282	+ // Get options and selected model
	283	+ $options = get_option('mxchat_options');
	284	+ $selected_model = $options['embedding_model'] ?? 'text-embedding-ada-002';
	285	+
	286	+ // Determine endpoint and API key based on model
	287	+ if (strpos($selected_model, 'voyage') === 0) {
	288	+ $endpoint = 'https://api.voyageai.com/v1/embeddings';
	289	+ $api_key = $options['voyage_api_key'] ?? '';
	290	+ } elseif (strpos($selected_model, 'gemini-embedding') === 0) {
	291	+ $endpoint = 'https://generativelanguage.googleapis.com/v1beta/models/' . $selected_model . ':embedContent';
	292	+ $api_key = $options['gemini_api_key'] ?? '';
	293	+ } else {
	294	+ $endpoint = 'https://api.openai.com/v1/embeddings';
	295	+ // Use the passed API key for OpenAI
	296	+ }
	297	+
	298	+ // Prepare request body based on provider
	299	+ if (strpos($selected_model, 'gemini-embedding') === 0) {
	300	+ // Gemini API format
	301	+ $request_body = [
	302	+ 'model' => 'models/' . $selected_model,
	303	+ 'content' => [
	304	+ 'parts' => [
	305	+ ['text' => $text]
	306	+ ]
91	307	],
92		- 'timeout' => 60,
93		- 'redirection' => 5,
94		- 'blocking' => true,
95		- 'httpversion' => '1.0',
96		- 'sslverify' => true,
	308	+ 'outputDimensionality' => 1536
97	309	];
98		-
99		- $response = wp_remote_post($endpoint, $args);
100		-
101		- if (is_wp_error($response)) {
102		- // error_log('Error generating embedding: ' . $response->get_error_message());
	310	+
	311	+ // Prepare headers for Gemini (API key as query parameter)
	312	+ $endpoint .= '?key=' . $api_key;
	313	+ $headers = [
	314	+ 'Content-Type' => 'application/json'
	315	+ ];
	316	+ } else {
	317	+ // OpenAI/Voyage API format
	318	+ $request_body = [
	319	+ 'input' => $text,
	320	+ 'model' => $selected_model
	321	+ ];
	322	+
	323	+ // Add output_dimension for voyage-3-large
	324	+ if ($selected_model === 'voyage-3-large') {
	325	+ $request_body['output_dimension'] = 2048;
	326	+ }
	327	+
	328	+ // Prepare headers for OpenAI/Voyage
	329	+ $headers = [
	330	+ 'Content-Type' => 'application/json',
	331	+ 'Authorization' => 'Bearer ' . $api_key
	332	+ ];
	333	+ }
	334	+
	335	+ $args = [
	336	+ 'body' => wp_json_encode($request_body),
	337	+ 'headers' => $headers,
	338	+ 'timeout' => 60,
	339	+ 'redirection' => 5,
	340	+ 'blocking' => true,
	341	+ 'httpversion' => '1.0',
	342	+ 'sslverify' => true,
	343	+ ];
	344	+
	345	+ $response = wp_remote_post($endpoint, $args);
	346	+
	347	+ if (is_wp_error($response)) {
	348	+ error_log('Error generating embedding: ' . $response->get_error_message());
	349	+ return null;
	350	+ }
	351	+
	352	+ $response_body = json_decode(wp_remote_retrieve_body($response), true);
	353	+
	354	+ // Handle different response formats based on provider
	355	+ if (strpos($selected_model, 'gemini-embedding') === 0) {
	356	+ // Gemini API response format
	357	+ if (isset($response_body['embedding']['values']) && is_array($response_body['embedding']['values'])) {
	358	+ return $response_body['embedding']['values'];
	359	+ } else {
	360	+ error_log('Invalid response received from Gemini embedding API: ' . wp_json_encode($response_body));
103	361	return null;
104	362	}
105		-
106		- $response_body = json_decode(wp_remote_retrieve_body($response), true);
107		-
	363	+ } else {
	364	+ // OpenAI/Voyage API response format
108	365	if (isset($response_body['data'][0]['embedding']) && is_array($response_body['data'][0]['embedding'])) {
109	366	return $response_body['data'][0]['embedding'];
110	367	} else {
111		- // error_log('Invalid response received from embedding API.');
	368	+ error_log('Invalid response received from embedding API: ' . wp_json_encode($response_body));
112	369	return null;
113	370	}
114	371	}
115		-}
	372	+}
	373	+}