← All changes | includes/frontend/class-schema-graph.php +159 -5 2.6.0 → 2.7.0 View file →

		@@ -87,8 +87,26 @@
87	87	'Book', 'Movie', 'Service', 'ImageObject', 'VideoObject',
88	88	];
89	89
90	90	/**
	91	+ * Types that may carry a `breadcrumb` property.
	92	+ *
	93	+ * Schema.org limits `breadcrumb` to WebPage and its subtypes. Attaching it
	94	+ * to a BlogPosting, Product or Recipe primary fails validation with
	95	+ * "Unexpected property" on every URL; the standalone BreadcrumbList node is
	96	+ * emitted either way (#693).
	97	+ *
	98	+ * @since 2.7.0
	99	+ * @var array<int,string>
	100	+ */
	101	+ private const BREADCRUMB_TYPES = [
	102	+ 'WebPage', 'AboutPage', 'CheckoutPage', 'CollectionPage', 'ContactPage',
	103	+ 'FAQPage', 'ItemPage', 'MedicalWebPage', 'ProfilePage', 'QAPage',
	104	+ 'RealEstateListing', 'SearchResultsPage', 'MediaGallery', 'ImageGallery',
	105	+ 'VideoGallery',
	106	+ ];
	107	+
	108	+ /**
91	109	* Gutenberg FAQ block name.
92	110	*/
93	111	private const FAQ_BLOCK = 'thinkrank/faq';
94	112
		@@ -149,8 +167,16 @@
149	167	*/
150	168	private static ?self $instance = null;
151	169
152	170	/**
	171	+ * Memoised master switch, or null when it has not been read this request.
	172	+ *
	173	+ * @since 2.7.0
	174	+ * @var bool|null
	175	+ */
	176	+ private static ?bool $master_switch_on = null;
	177	+
	178	+ /**
153	179	* Competing page-level entities: ['rank' => int, 'schema' => array, 'type' => string].
154	180	*
155	181	* @var array<int,array>
156	182	*/
		@@ -232,8 +258,10 @@
232	258	* @return void
233	259	*/
234	260	public static function reset(): void {
235	261	self::$instance = null;
	262	+ // Or a test that seeds the switch inherits the previous test's answer.
	263	+ self::$master_switch_on = null;
236	264	}
237	265
238	266	/**
239	267	* Register a candidate for the page's single page-level entity.
		@@ -302,8 +330,61 @@
302	330	return (is_string($actual) && $actual !== '') ? $actual : $declared;
303	331	}
304	332
305	333	/**
	334	+ * Whether a page entity may carry a `breadcrumb` property.
	335	+ *
	336	+ * Reading the node's own `@type` key is not enough, and every case it
	337	+ * misses ends with a real WebPage losing a valid property:
	338	+ *
	339	+ * - A deployed schema whose stored JSON omits `@type` carries the type in
	340	+ * the `schema_type` column instead. `effective_type()` already resolves
	341	+ * that, which is why the node's `@id` reads `#webpage` even though the
	342	+ * node itself has no `@type` — so the resolved value is what has to be
	343	+ * consulted here too.
	344	+ * - JSON-LD permits several types on one node. `["WebPage", "FAQPage"]` is
	345	+ * a WebPage, but a strict in_array() against the array as a whole is
	346	+ * false, so the trail would be dropped from a page that may carry it.
	347	+ * - A list naming nothing usable (`[]`, `[null]`) is the first case wearing
	348	+ * the second's clothes, and resolves the same way.
	349	+ *
	350	+ * @since 2.7.0
	351	+ * @param array $node The page entity.
	352	+ * @param string $resolved_type Type the graph resolved for it.
	353	+ * @return bool
	354	+ */
	355	+ private function allows_breadcrumb(array $node, string $resolved_type): bool {
	356	+ $declared = $node['@type'] ?? '';
	357	+
	358	+ // One path for both shapes. Splitting them invites the list branch to
	359	+ // grow its own idea of what an absent type means, which is the mistake
	360	+ // being corrected here in the first place.
	361	+ $named = false;
	362	+
	363	+ foreach (is_array($declared) ? $declared : [$declared] as $type) {
	364	+ if (!is_string($type) || '' === $type) {
	365	+ continue;
	366	+ }
	367	+
	368	+ $named = true;
	369	+
	370	+ if (in_array($type, self::BREADCRUMB_TYPES, true)) {
	371	+ return true;
	372	+ }
	373	+ }
	374	+
	375	+ // The node named a type, and none of them may carry a breadcrumb.
	376	+ if ($named) {
	377	+ return false;
	378	+ }
	379	+
	380	+ // It named none, so it is whatever the graph resolved for it — the same
	381	+ // value its @id was minted from. An empty list is no more informative
	382	+ // than a missing key and must not read as "definitely not a WebPage".
	383	+ return in_array($resolved_type, self::BREADCRUMB_TYPES, true);
	384	+ }
	385	+
	386	+ /**
306	387	* Register a node that does not compete for the page-level slot.
307	388	*
308	389	* @since 1.32.0
309	390	* @param array $schema Schema array.
		@@ -788,8 +869,59 @@
788	869	return !empty($this->primary_candidates) || !empty($this->supporting) || !empty($this->faq_entities);
789	870	}
790	871
791	872	/**
	873	+ * Whether ThinkRank may emit structured data for this request at all.
	874	+ *
	875	+ * The master switch and the matrix's per-content-type Schema switch, asked
	876	+ * once. #461 put the master switch inside output_site_schema_markup(),
	877	+ * which is one of four producers; the other three never learned about it,
	878	+ * so turning Schema off removed the deployed rows and left the live
	879	+ * generator running — the page emitted *more* types with the switch off
	880	+ * than with it on (#688).
	881	+ *
	882	+ * Static because the producers that need it do not share a base class: the
	883	+ * three graph producers converge on render(), but Blocks_Manager emits its
	884	+ * own script tag from a content filter and never touches the graph, so it
	885	+ * has to ask the same question independently.
	886	+ *
	887	+ * @since 2.7.0
	888	+ * @return bool True when structured data may be emitted.
	889	+ */
	890	+ public static function output_allowed(): bool {
	891	+ // Memoised because inject_block_schema() asks once per matching block,
	892	+ // and Schema_Management_System's constructor builds a schema builder and
	893	+ // a cache manager and registers listeners — it is not something to spin
	894	+ // up per block. The switch is site-wide, so it cannot change within a
	895	+ // request; the per-content-type check below is query-dependent and stays
	896	+ // live.
	897	+ if (null === self::$master_switch_on) {
	898	+ self::$master_switch_on = true;
	899	+
	900	+ if (class_exists('ThinkRank\\SEO\\Schema_Management_System')) {
	901	+ $settings = (new \ThinkRank\SEO\Schema_Management_System())->get_settings('site', null);
	902	+
	903	+ // Absent means "not configured", which every other reader treats
	904	+ // as enabled; only a value that is present and off disables.
	905	+ self::$master_switch_on = !(array_key_exists('enabled', $settings) && empty($settings['enabled']));
	906	+ }
	907	+ }
	908	+
	909	+ if (!self::$master_switch_on) {
	910	+ return false;
	911	+ }
	912	+
	913	+ if (class_exists('ThinkRank\\SEO\\Content_Type_Settings')) {
	914	+ return \ThinkRank\SEO\Content_Type_Settings::is_enabled_for_current(
	915	+ \ThinkRank\SEO\Content_Type_Settings::FEATURE_SCHEMA,
	916	+ true
	917	+ );
	918	+ }
	919	+
	920	+ return true;
	921	+ }
	922	+
	923	+ /**
792	924	* Assemble and emit the graph. Safe to call more than once.
793	925	*
794	926	* @since 1.32.0
795	927	* @return void
		@@ -798,8 +930,14 @@
798	930	if ($this->rendered || !$this->has_nodes()) {
799	931	return;
800	932	}
801	933
	934	+ // Every graph producer converges here, so this is the one place the
	935	+ // master switch has to hold for all of them (#688).
	936	+ if (!self::output_allowed()) {
	937	+ return;
	938	+ }
	939	+
802	940	// A 404 response represents no content, so there is nothing for
803	941	// structured data to describe. The page-level producers already skip
804	942	// this context, but the site-identity entity does not, so without this
805	943	// guard every miss — including crawlers probing URLs that never existed
		@@ -834,8 +972,18 @@
834	972	if (empty($graph)) {
835	973	return;
836	974	}
837	975
	976	+ // One pass over the assembled graph, rather than at each producer.
	977	+ // @id and url values arrive from a dozen of them — some derived from
	978	+ // WordPress, some read straight out of stored settings — and on a
	979	+ // misconfigured site that produced a single graph carrying both
	980	+ // schemes at once, with @ids that no longer matched the canonical they
	981	+ // are supposed to identify (#638). Normalizing where the graph is
	982	+ // serialized is the only place that catches all of them, including
	983	+ // nodes an add-on added through the filter above.
	984	+ $graph = \ThinkRank\SEO\Url_Scheme::apply_deep($graph);
	985	+
838	986	$json = wp_json_encode(
839	987	['@context' => self::SCHEMA_CONTEXT, '@graph' => array_values($graph)],
840	988	JSON_UNESCAPED_SLASHES | JSON_UNESCAPED_UNICODE | JSON_PRETTY_PRINT
841	989	| JSON_HEX_TAG | JSON_HEX_AMP | JSON_HEX_APOS | JSON_HEX_QUOT
		@@ -958,11 +1106,12 @@
958	1106	$primary = ['schema' => $faq, 'type' => 'FAQPage'];
959	1107	$faq = null;
960	1108	}
961	1109
962		- $nodes = [];
963		- $primary_id = '';
964		- $used_ids = [];
	1110	+ $nodes = [];
	1111	+ $primary_id = '';
	1112	+ $primary_type = '';
	1113	+ $used_ids = [];
965	1114
966	1115	if (null !== $primary) {
967	1116	$node = $primary['schema'];
968	1117
		@@ -970,9 +1119,10 @@
970	1119	// so an "Article" setting that renders BlogPosting reads #blogposting.
971	1120	$resolved_type = $this->effective_type($node, $primary['type']);
972	1121
973	1122	$node = $this->assign_id($node, $base . '#' . strtolower($resolved_type), $used_ids);
974		- $primary_id = $node['@id'];
	1123	+ $primary_id = $node['@id'];
	1124	+ $primary_type = $resolved_type;
975	1125	$nodes['primary'] = $node;
976	1126	}
977	1127
978	1128	// Entities deployed alongside the winner (Pro's Multi-Schema lets a post
		@@ -1059,9 +1209,13 @@
1059	1209	if (isset($nodes['primary'])) {
1060	1210	if ($website_id !== '' && !isset($nodes['primary']['isPartOf'])) {
1061	1211	$nodes['primary']['isPartOf'] = ['@id' => $website_id];
1062	1212	}
1063		- if ($breadcrumb_id !== '' && !isset($nodes['primary']['breadcrumb'])) {
	1213	+ if (
	1214	+ $breadcrumb_id !== ''
	1215	+ && !isset($nodes['primary']['breadcrumb'])
	1216	+ && $this->allows_breadcrumb($nodes['primary'], $primary_type)
	1217	+ ) {
1064	1218	$nodes['primary']['breadcrumb'] = ['@id' => $breadcrumb_id];
1065	1219	}
1066	1220
1067	1221	// Point publisher/author at the full nodes already in the graph.