← All changes | includes/seo/class-schema-management-system.php +135 -43 2.4.0 → 2.12.0 View file →

		@@ -201,13 +201,43 @@
201	201	'WebPage' => [
202	202	'name' => 'WebPage',
203	203	'description' => 'Individual web pages',
204	204	'required_properties' => ['name', 'url'],
	205	+ // 'post' as well as 'page': the per-page selector reaches this for
	206	+ // any post type, and a registry limited to 'page' silently produced
	207	+ // nothing for the rest (#624).
205	208	'recommended_properties' => ['description', 'author', 'datePublished', 'breadcrumb'],
206	209	'rich_snippets' => ['webpage', 'breadcrumb'],
207		- 'context_types' => ['page'],
	210	+ 'context_types' => ['page', 'post'],
208	211	'priority' => 'medium'
209	212	],
	213	+ 'AboutPage' => [
	214	+ 'name' => 'AboutPage',
	215	+ 'description' => 'A page describing the organisation or person behind the site',
	216	+ 'required_properties' => ['name', 'url'],
	217	+ 'recommended_properties' => ['description', 'author', 'datePublished', 'breadcrumb'],
	218	+ 'rich_snippets' => ['webpage', 'breadcrumb'],
	219	+ 'context_types' => ['page', 'post'],
	220	+ 'priority' => 'medium'
	221	+ ],
	222	+ 'ContactPage' => [
	223	+ 'name' => 'ContactPage',
	224	+ 'description' => 'A page giving contact details',
	225	+ 'required_properties' => ['name', 'url'],
	226	+ 'recommended_properties' => ['description', 'author', 'datePublished', 'breadcrumb'],
	227	+ 'rich_snippets' => ['webpage', 'breadcrumb'],
	228	+ 'context_types' => ['page', 'post'],
	229	+ 'priority' => 'medium'
	230	+ ],
	231	+ 'ProfilePage' => [
	232	+ 'name' => 'ProfilePage',
	233	+ 'description' => 'A page about a single person or organisation',
	234	+ 'required_properties' => ['name', 'url'],
	235	+ 'recommended_properties' => ['description', 'author', 'datePublished', 'breadcrumb'],
	236	+ 'rich_snippets' => ['webpage', 'breadcrumb'],
	237	+ 'context_types' => ['page', 'post'],
	238	+ 'priority' => 'medium'
	239	+ ],
210	240	'FAQPage' => [
211	241	'name' => 'FAQPage',
212	242	'description' => 'Frequently Asked Questions pages',
213	243	'required_properties' => ['mainEntity'],
		@@ -397,9 +427,25 @@
397	427	'business_phone', 'business_email', 'business_hours',
398	428	'business_latitude', 'business_longitude', 'business_price_range',
399	429	];
400	430
401		- if (empty(array_intersect_key($settings, array_flip($business_keys)))) {
	431	+ // Which deployed types a Site Identity key can invalidate. The business
	432	+ // block feeds LocalBusiness and Organization; alternate_name feeds the
	433	+ // WebSite node, which had no entry here at all — so editing it left the
	434	+ // deployed schema showing the previous value until something else
	435	+ // happened to redeploy (#692).
	436	+ $refresh_types = [];
	437	+
	438	+ if (!empty(array_intersect_key($settings, array_flip($business_keys)))) {
	439	+ $refresh_types[] = 'LocalBusiness';
	440	+ $refresh_types[] = 'Organization';
	441	+ }
	442	+
	443	+ if (array_key_exists('alternate_name', $settings)) {
	444	+ $refresh_types[] = 'WebSite';
	445	+ }
	446	+
	447	+ if (empty($refresh_types)) {
402	448	return;
403	449	}
404	450
405	451	$schema_settings = $this->get_settings($context_type, $context_id);
		@@ -409,9 +455,9 @@
409	455
410	456	// Only refresh types that are actually deployed, so this never adds a
411	457	// type the admin did not enable.
412	458	$deployed = array_keys((array) $this->get_deployed_schemas($context_type, $context_id));
413		- $affected = array_values(array_intersect($deployed, ['LocalBusiness', 'Organization']));
	459	+ $affected = array_values(array_intersect($deployed, $refresh_types));
414	460
415	461	if (empty($affected)) {
416	462	return;
417	463	}
		@@ -1583,12 +1629,23 @@
1583	1629	} elseif ($context_id && in_array($context_type, ['post', 'page', 'product'], true)) {
1584	1630	// Post/page/product data
1585	1631	$post = get_post($context_id);
1586	1632	if ($post) {
	1633	+ // Resolve the featured image's URL to its ID here, where the ID
	1634	+ // is in hand, so the schema builder does not query for an
	1635	+ // attachment it was just given (#847). Offered as a hint rather
	1636	+ // than asserted: `post_thumbnail_url` can swap the URL for one
	1637	+ // the featured image does not own.
	1638	+ $thumbnail_url = get_the_post_thumbnail_url($post->ID, 'full');
	1639	+
	1640	+ if ($thumbnail_url) {
	1641	+ Attachment_Lookup::id_from_url((string) $thumbnail_url, (int) get_post_thumbnail_id($post->ID));
	1642	+ }
	1643	+
1587	1644	$content_data = [
1588	1645	'title' => $post->post_title,
1589	1646	'url' => get_permalink($post->ID),
1590		- 'excerpt' => $post->post_excerpt ?: wp_trim_words($post->post_content, 30),
	1647	+ 'excerpt' => $post->post_excerpt ?: \ThinkRank\Core\Seo_Text::trim_words($post->post_content, 30),
1591	1648	'content' => $post->post_content,
1592	1649	'author' => [
1593	1650	'name' => get_the_author_meta('display_name', $post->post_author),
1594	1651	'url' => get_author_posts_url($post->post_author)
		@@ -1598,9 +1655,9 @@
1598	1655	// Google rejects as "Invalid value in field datePublished"
1599	1656	// and drops the Article rich result (#465).
1600	1657	'date' => get_the_date('c', $post),
1601	1658	'modified' => get_the_modified_date('c', $post),
1602		- 'image' => get_the_post_thumbnail_url($post->ID, 'full'),
	1659	+ 'image' => $thumbnail_url,
1603	1660	'focus_keywords' => Focus_Keywords::get($post->ID),
1604	1661	'business_data' => $this->get_business_data_from_local_seo(),
1605	1662	'site_data' => $this->get_site_data_for_schema(),
1606	1663	'social_data' => $this->get_social_data_for_schema()
		@@ -1993,12 +2050,10 @@
1993	2050
1994	2051	/**
1995	2052	* Get deployed schemas for frontend integration
1996	2053	*
1997		- * PERFORMANCE OPTIMIZED: This method now uses:
1998		- * 1. Schema caching layer (90% reduction in database queries)
1999		- * 2. Window function approach instead of correlated subquery (80-90% query performance improvement)
2000		- * 3. Composite index: idx_context_schema_active
	2054	+ * Returns the newest active, deployed row of each schema type for the
	2055	+ * context. Results are cached per context (see Schema_Cache_Manager).
2001	2056	*
2002	2057	* @since 1.0.0
2003	2058	*
2004	2059	* @param string $context_type Context type
		@@ -2021,48 +2076,57 @@
2021	2076
2022	2077	// Use existing seo_schema table
2023	2078	$table_name = $wpdb->prefix . 'thinkrank_seo_schema';
2024	2079
2025		- // OPTIMIZED QUERY: Use window function approach to eliminate correlated subquery
2026		- // This leverages the new composite index: idx_context_schema_active (context_type, schema_type, is_active, created_at DESC)
	2080	+ // The newest row per type used to be picked with ROW_NUMBER() OVER
	2081	+ // (PARTITION BY schema_type ...). Window functions need MySQL 8.0 /
	2082	+ // MariaDB 10.2, and WordPress still runs on MySQL 5.7, where that is
	2083	+ // a syntax error on every page view and no deployed schema is ever
	2084	+ // output. A row is the newest of its type when no other row of the
	2085	+ // same context and type outranks it, so NOT EXISTS keeps the
	2086	+ // greatest-per-group in the database, and schema_data — JSON, and
	2087	+ // large — is only transferred for the rows that are output.
	2088	+ //
	2089	+ // `<=>` is NULL-safe equality: the site context stores context_id
	2090	+ // as NULL, and `n.context_id = s.context_id` is never true for it.
	2091	+ // schema_id breaks a same-second tie, which the window function
	2092	+ // left to chance.
	2093	+ $args = [$context_type];
2027	2094	if (null === $context_id) {
2028		- // phpcs:ignore WordPress.DB.DirectDatabaseQuery.DirectQuery,WordPress.DB.DirectDatabaseQuery.NoCaching, PluginCheck.Security.DirectDB.UnescapedDBParameter -- Schema retrieval requires direct database access
2029		- $sql = sprintf(
2030		- 'SELECT schema_type, schema_data FROM (SELECT schema_type, schema_data, ROW_NUMBER() OVER (PARTITION BY schema_type ORDER BY created_at DESC) as rn FROM %s WHERE context_type = %%s AND context_id IS NULL AND is_active = 1 AND validation_status IN (\'deployed\', \'valid\')) ranked WHERE rn = 1 ORDER BY schema_type',
2031		- $table_name
2032		- );
2033		- // phpcs:ignore WordPress.DB.DirectDatabaseQuery.DirectQuery, WordPress.DB.DirectDatabaseQuery.NoCaching, PluginCheck.Security.DirectDB.UnescapedDBParameter -- Schema retrieval requires direct database access
2034		- $deployed_schemas = $wpdb->get_results(
2035		- $wpdb->prepare(
2036		- // phpcs:ignore WordPress.DB.PreparedSQL.NotPrepared, PluginCheck.Security.DirectDB.UnescapedDBParameter -- SQL is properly prepared with placeholders
2037		- $sql,
2038		- $context_type
2039		- ),
2040		- ARRAY_A
2041		- );
	2095	+ $context_where = 's.context_id IS NULL';
2042	2096	} else {
2043		- // phpcs:ignore WordPress.DB.DirectDatabaseQuery.DirectQuery,WordPress.DB.DirectDatabaseQuery.NoCaching, PluginCheck.Security.DirectDB.UnescapedDBParameter -- Schema retrieval requires direct database access
2044		- $sql = sprintf(
2045		- 'SELECT schema_type, schema_data FROM (SELECT schema_type, schema_data, ROW_NUMBER() OVER (PARTITION BY schema_type ORDER BY created_at DESC) as rn FROM %s WHERE context_type = %%s AND context_id = %%d AND is_active = 1 AND validation_status IN (\'deployed\', \'valid\')) ranked WHERE rn = 1 ORDER BY schema_type',
2046		- $table_name
2047		- );
2048		- // phpcs:ignore WordPress.DB.DirectDatabaseQuery.DirectQuery, WordPress.DB.DirectDatabaseQuery.NoCaching, PluginCheck.Security.DirectDB.UnescapedDBParameter -- Schema retrieval requires direct database access
2049		- $deployed_schemas = $wpdb->get_results(
2050		- $wpdb->prepare(
2051		- // phpcs:ignore WordPress.DB.PreparedSQL.NotPrepared, PluginCheck.Security.DirectDB.UnescapedDBParameter -- SQL is properly prepared with placeholders
2052		- $sql,
2053		- $context_type,
2054		- $context_id
2055		- ),
2056		- ARRAY_A
2057		- );
	2097	+ $context_where = 's.context_id = %d';
	2098	+ $args[] = $context_id;
2058	2099	}
2059	2100
	2101	+ $sql = sprintf(
	2102	+ 'SELECT s.schema_type, s.schema_data FROM %1$s s'
	2103	+ . ' WHERE s.context_type = %%s AND %2$s AND s.is_active = 1 AND s.validation_status IN (\'deployed\', \'valid\')'
	2104	+ . ' AND NOT EXISTS ('
	2105	+ . 'SELECT 1 FROM %1$s n'
	2106	+ . ' WHERE n.context_type = s.context_type AND n.context_id <=> s.context_id AND n.schema_type = s.schema_type'
	2107	+ . ' AND n.is_active = 1 AND n.validation_status IN (\'deployed\', \'valid\')'
	2108	+ . ' AND (n.created_at > s.created_at OR (n.created_at = s.created_at AND n.schema_id > s.schema_id))'
	2109	+ . ')'
	2110	+ . ' ORDER BY s.schema_type',
	2111	+ $table_name,
	2112	+ $context_where
	2113	+ );
	2114	+
	2115	+ // phpcs:ignore WordPress.DB.DirectDatabaseQuery.DirectQuery, WordPress.DB.DirectDatabaseQuery.NoCaching, PluginCheck.Security.DirectDB.UnescapedDBParameter -- Schema retrieval requires direct database access
	2116	+ $deployed_schemas = $wpdb->get_results(
	2117	+ $wpdb->prepare(
	2118	+ // phpcs:ignore WordPress.DB.PreparedSQL.NotPrepared, PluginCheck.Security.DirectDB.UnescapedDBParameter -- SQL is properly prepared with placeholders
	2119	+ $sql,
	2120	+ ...$args
	2121	+ ),
	2122	+ ARRAY_A
	2123	+ );
	2124	+
2060	2125	// Deliberately no early return on an empty result: it has to reach the
2061	2126	// cache write below. Most URLs have no deployed schema, so gating the
2062	2127	// write on a non-empty result made the majority of front-end requests
2063		- // permanent cache misses, re-running a ROW_NUMBER() OVER (PARTITION BY
2064		- // ...) query with two filesorts on every pageview (#392).
	2128	+ // permanent cache misses, re-running the query on every pageview (#392).
2065	2129	$deployed_schemas = $deployed_schemas ?: [];
2066	2130
2067	2131	// Process schemas for return
2068	2132	$processed_schemas = [];
		@@ -2242,9 +2306,12 @@
2242	2306	}
2243	2307
2244	2308	$type = $schema['@type'] ?? '';
2245	2309	$type = is_array($type) ? reset($type) : $type;
2246		- $is_entity = in_array((string) $type, self::ENTITY_URL_TYPES, true);
	2310	+ // A LocalBusiness is deployed under the subtype the site chose, so the
	2311	+ // exemption has to cover every subtype, not only the literal root.
	2312	+ $is_entity = in_array((string) $type, self::ENTITY_URL_TYPES, true)
	2313	+ || \ThinkRank\Config\Local_Business_Types_Config::is_local_business($type);
2247	2314
2248	2315	if (isset($schema['url']) && !$is_entity) {
2249	2316	$schema['url'] = $permalink;
2250	2317	}
		@@ -2272,9 +2339,15 @@
2272	2339	* Covers non-ISO-8601 dates (#465) and WP locales in inLanguage, which must
2273	2340	* be a BCP-47 tag — en-US, not en_US (#473). Walks nested nodes so values
2274	2341	* inside author/publisher/@graph entries are covered too.
2275	2342	*
	2343	+ * Also decodes HTML entities in plain-text properties. Schema_Builder
	2344	+ * stored the block editor's `&amp;` as-is until 2.10.0, and nothing
	2345	+ * decodes JSON-LD downstream, so every deployed node built from post text
	2346	+ * published the entity literally.
	2347	+ *
2276	2348	* @since 1.16.0
	2349	+ * @since 2.10.0 Decodes entities in plain-text properties.
2277	2350	*
2278	2351	* @param array $schema Decoded schema data.
2279	2352	* @return array Normalised schema.
2280	2353	*/
		@@ -2283,8 +2356,16 @@
2283	2356	'datePublished', 'dateModified', 'dateCreated', 'uploadDate',
2284	2357	'startDate', 'endDate', 'validFrom', 'validThrough', 'expires',
2285	2358	];
2286	2359
	2360	+ // Plain text in schema.org. Answer/HowToStep `text` is deliberately
	2361	+ // absent: Google reads Answer.text as HTML, where an entity is correct
	2362	+ // and decoding `&lt;` would turn escaped text into live markup.
	2363	+ static $text_keys = [
	2364	+ 'name', 'headline', 'alternativeHeadline', 'description',
	2365	+ 'reviewBody', 'about', 'abstract', 'caption',
	2366	+ ];
	2367	+
2287	2368	foreach ($schema as $key => $value) {
2288	2369	if (is_array($value)) {
2289	2370	$schema[$key] = $this->normalize_stored_schema($value);
2290	2371	continue;
		@@ -2294,8 +2375,16 @@
2294	2375	$schema[$key] = str_replace('_', '-', $value);
2295	2376	continue;
2296	2377	}
2297	2378
	2379	+ // Decode only: a snapshot already truncated with an ellipsis must
	2380	+ // keep it, which the full Seo_Text::normalize_schema_text() would
	2381	+ // strip as an excerpt marker.
	2382	+ if (in_array($key, $text_keys, true) && is_string($value) && '' !== $value) {
	2383	+ $schema[$key] = \ThinkRank\Core\Seo_Text::decode_schema_entities($value);
	2384	+ continue;
	2385	+ }
	2386	+
2298	2387	if (!in_array($key, $date_keys, true) || !is_string($value) || '' === $value) {
2299	2388	continue;
2300	2389	}
2301	2390
		@@ -2598,8 +2687,11 @@
2598	2687	'site_url' => home_url(),
2599	2688	'admin_email' => get_option('admin_email'),
2600	2689	'language' => get_locale(),
2601	2690	'timezone' => get_option('timezone_string'),
	2691	+ // Read by populate_website_schema(), so the deployed WebSite node
	2692	+ // carries the same alternateName as the default one (#692).
	2693	+ 'alternate_name' => $site_identity_settings['alternate_name'] ?? '',
2602	2694	'founded_date' => $site_identity_settings['founded_date'] ?? '',
2603	2695	'founder_name' => $site_identity_settings['founder_name'] ?? '',
2604	2696	'company_type' => $site_identity_settings['company_type'] ?? 'Organization',
2605	2697	// Site Identity assets