← All changes | includes/frontend/class-schema-graph.php +206 -14 2.6.0 → 2.10.0 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 $schema Schema array.
		@@ -560,11 +641,24 @@
560	641	if (!is_array($block)) {
561	642	continue;
562	643	}
563	644
564		- if (($block['blockName'] ?? '') === self::FAQ_BLOCK) {
565		- $attrs = $block['attrs'] ?? [];
	645	+ $block_name = (string) ($block['blockName'] ?? '');
	646	+ $attrs = is_array($block['attrs'] ?? null) ? $block['attrs'] : [];
566	647
	648	+ // A leftover Rank Math FAQ block is absorbed as if it were ours, so
	649	+ // an unmigrated post contributes its questions to the single
	650	+ // FAQPage rather than to nothing at all (#777). The fallback stays
	651	+ // silent while Rank Math is active and still emitting its own.
	652	+ if (\ThinkRank\Integrations\Rank_Math_Blocks::is_source_block($block_name)) {
	653	+ $fallback = \ThinkRank\Integrations\Rank_Math_Blocks::schema_fallback($block_name, $attrs);
	654	+ if (null !== $fallback) {
	655	+ $block_name = $fallback['name'];
	656	+ $attrs = $fallback['attrs'];
	657	+ }
	658	+ }
	659	+
	660	+ if ($block_name === self::FAQ_BLOCK) {
567	661	// Mirrors Blocks_Manager: schema is on unless explicitly disabled.
568	662	$disabled = array_key_exists('outputSchema', $attrs) && false === $attrs['outputSchema'];
569	663
570	664	if (!$disabled) {
		@@ -788,8 +882,59 @@
788	882	return !empty($this->primary_candidates) || !empty($this->supporting) || !empty($this->faq_entities);
789	883	}
790	884
791	885	/**
	886	+ * Whether ThinkRank may emit structured data for this request at all.
	887	+ *
	888	+ * The master switch and the matrix's per-content-type Schema switch, asked
	889	+ * once. #461 put the master switch inside output_site_schema_markup(),
	890	+ * which is one of four producers; the other three never learned about it,
	891	+ * so turning Schema off removed the deployed rows and left the live
	892	+ * generator running — the page emitted *more* types with the switch off
	893	+ * than with it on (#688).
	894	+ *
	895	+ * Static because the producers that need it do not share a base class: the
	896	+ * three graph producers converge on render(), but Blocks_Manager emits its
	897	+ * own script tag from a content filter and never touches the graph, so it
	898	+ * has to ask the same question independently.
	899	+ *
	900	+ * @since 2.7.0
	901	+ * @return bool True when structured data may be emitted.
	902	+ */
	903	+ public static function output_allowed(): bool {
	904	+ // Memoised because inject_block_schema() asks once per matching block,
	905	+ // and Schema_Management_System's constructor builds a schema builder and
	906	+ // a cache manager and registers listeners — it is not something to spin
	907	+ // up per block. The switch is site-wide, so it cannot change within a
	908	+ // request; the per-content-type check below is query-dependent and stays
	909	+ // live.
	910	+ if (null === self::$master_switch_on) {
	911	+ self::$master_switch_on = true;
	912	+
	913	+ if (class_exists('ThinkRank\\SEO\\Schema_Management_System')) {
	914	+ $settings = (new \ThinkRank\SEO\Schema_Management_System())->get_settings('site', null);
	915	+
	916	+ // Absent means "not configured", which every other reader treats
	917	+ // as enabled; only a value that is present and off disables.
	918	+ self::$master_switch_on = !(array_key_exists('enabled', $settings) && empty($settings['enabled']));
	919	+ }
	920	+ }
	921	+
	922	+ if (!self::$master_switch_on) {
	923	+ return false;
	924	+ }
	925	+
	926	+ if (class_exists('ThinkRank\\SEO\\Content_Type_Settings')) {
	927	+ return \ThinkRank\SEO\Content_Type_Settings::is_enabled_for_current(
	928	+ \ThinkRank\SEO\Content_Type_Settings::FEATURE_SCHEMA,
	929	+ true
	930	+ );
	931	+ }
	932	+
	933	+ return true;
	934	+ }
	935	+
	936	+ /**
792	937	* Assemble and emit the graph. Safe to call more than once.
793	938	*
794	939	* @since 1.32.0
795	940	* @return void
		@@ -798,8 +943,14 @@
798	943	if ($this->rendered || !$this->has_nodes()) {
799	944	return;
800	945	}
801	946
	947	+ // Every graph producer converges here, so this is the one place the
	948	+ // master switch has to hold for all of them (#688).
	949	+ if (!self::output_allowed()) {
	950	+ return;
	951	+ }
	952	+
802	953	// A 404 response represents no content, so there is nothing for
803	954	// structured data to describe. The page-level producers already skip
804	955	// this context, but the site-identity entity does not, so without this
805	956	// guard every miss — including crawlers probing URLs that never existed
		@@ -834,8 +985,18 @@
834	985	if (empty($graph)) {
835	986	return;
836	987	}
837	988
	989	+ // One pass over the assembled graph, rather than at each producer.
	990	+ // @id and url values arrive from a dozen of them — some derived from
	991	+ // WordPress, some read straight out of stored settings — and on a
	992	+ // misconfigured site that produced a single graph carrying both
	993	+ // schemes at once, with @ids that no longer matched the canonical they
	994	+ // are supposed to identify (#638). Normalizing where the graph is
	995	+ // serialized is the only place that catches all of them, including
	996	+ // nodes an add-on added through the filter above.
	997	+ $graph = \ThinkRank\SEO\Url_Scheme::apply_deep($graph);
	998	+
838	999	$json = wp_json_encode(
839	1000	['@context' => self::SCHEMA_CONTEXT, '@graph' => array_values($graph)],
840	1001	JSON_UNESCAPED_SLASHES | JSON_UNESCAPED_UNICODE | JSON_PRETTY_PRINT
841	1002	| JSON_HEX_TAG | JSON_HEX_AMP | JSON_HEX_APOS | JSON_HEX_QUOT
		@@ -845,9 +1006,9 @@
845	1006	return;
846	1007	}
847	1008
848	1009	echo "<!-- ThinkRank Schema Graph -->\n";
849		- echo '<script type="application/ld+json">' . "\n";
	1010	+ echo '<script type="application/ld+json" data-thinkrank="schema-graph">' . "\n";
850	1011	echo $json . "\n"; // phpcs:ignore WordPress.Security.EscapeOutput.OutputNotEscaped -- wp_json_encode with JSON_HEX_* cannot break out of the script block.
851	1012	echo '</script>' . "\n";
852	1013	echo "<!-- /ThinkRank Schema Graph -->\n";
853	1014	}
		@@ -958,11 +1119,12 @@
958	1119	$primary = ['schema' => $faq, 'type' => 'FAQPage'];
959	1120	$faq = null;
960	1121	}
961	1122
962		- $nodes = [];
963		- $primary_id = '';
964		- $used_ids = [];
	1123	+ $nodes = [];
	1124	+ $primary_id = '';
	1125	+ $primary_type = '';
	1126	+ $used_ids = [];
965	1127
966	1128	if (null !== $primary) {
967	1129	$node = $primary['schema'];
968	1130
		@@ -970,9 +1132,10 @@
970	1132	// so an "Article" setting that renders BlogPosting reads #blogposting.
971	1133	$resolved_type = $this->effective_type($node, $primary['type']);
972	1134
973	1135	$node = $this->assign_id($node, $base . '#' . strtolower($resolved_type), $used_ids);
974		- $primary_id = $node['@id'];
	1136	+ $primary_id = $node['@id'];
	1137	+ $primary_type = $resolved_type;
975	1138	$nodes['primary'] = $node;
976	1139	}
977	1140
978	1141	// Entities deployed alongside the winner (Pro's Multi-Schema lets a post
		@@ -1013,8 +1176,9 @@
1013	1176	$person_nodes = [];
1014	1177
1015	1178	foreach ($this->supporting as $index => $node) {
1016	1179	$type = $node['@type'] ?? '';
	1180	+ $slot = $this->site_level_slot($type);
1017	1181
1018	1182	if ('BreadcrumbList' === $type) {
1019	1183	$node = $this->assign_id($node, $base . '#breadcrumb', $used_ids);
1020	1184	$breadcrumb_id = $node['@id'];
		@@ -1023,9 +1187,9 @@
1023	1187	$website_id = $node['@id'];
1024	1188	} elseif ('Organization' === $type) {
1025	1189	$node = $this->assign_id($node, home_url('/#organization'), $used_ids);
1026	1190	$organization_nodes[] = $node;
1027		- } elseif (in_array($type, self::SITE_LEVEL_TYPES, true)) {
	1191	+ } elseif ('' !== $slot) {
1028	1192	// Site-level entities describe the site, not the page, so their
1029	1193	// @id must be stable across URLs. Falling through to the
1030	1194	// page-scoped branch minted a fresh identity on every URL, so
1031	1195	// one business became N entities in a crawler's graph and
		@@ -1035,9 +1199,9 @@
1035	1199	// arrive here claiming the same @id. assign_id() would resolve
1036	1200	// that collision by minting "#person-2", turning a duplicate
1037	1201	// into two competing entities that split the identity a
1038	1202	// knowledge graph is meant to consolidate (#479).
1039		- $duplicate_key = $this->find_same_entity($nodes, $type, $node);
	1203	+ $duplicate_key = $this->find_same_entity($nodes, $slot, $node);
1040	1204
1041	1205	if (null !== $duplicate_key) {
1042	1206	$nodes[$duplicate_key] = $this->merge_entity($nodes[$duplicate_key], $node);
1043	1207	continue;
		@@ -1042,11 +1206,11 @@
1042	1206	$nodes[$duplicate_key] = $this->merge_entity($nodes[$duplicate_key], $node);
1043	1207	continue;
1044	1208	}
1045	1209
1046		- $node = $this->assign_id($node, home_url('/#' . strtolower($type)), $used_ids);
	1210	+ $node = $this->assign_id($node, home_url('/#' . strtolower($slot)), $used_ids);
1047	1211
1048		- if ('Person' === $type) {
	1212	+ if ('Person' === $slot) {
1049	1213	$person_nodes[] = $node;
1050	1214	}
1051	1215	} elseif (is_string($type) && $type !== '') {
1052	1216	$node = $this->assign_id($node, $base . '#' . strtolower($type), $used_ids);
		@@ -1059,9 +1223,13 @@
1059	1223	if (isset($nodes['primary'])) {
1060	1224	if ($website_id !== '' && !isset($nodes['primary']['isPartOf'])) {
1061	1225	$nodes['primary']['isPartOf'] = ['@id' => $website_id];
1062	1226	}
1063		- if ($breadcrumb_id !== '' && !isset($nodes['primary']['breadcrumb'])) {
	1227	+ if (
	1228	+ $breadcrumb_id !== ''
	1229	+ && !isset($nodes['primary']['breadcrumb'])
	1230	+ && $this->allows_breadcrumb($nodes['primary'], $primary_type)
	1231	+ ) {
1064	1232	$nodes['primary']['breadcrumb'] = ['@id' => $breadcrumb_id];
1065	1233	}
1066	1234
1067	1235	// Point publisher/author at the full nodes already in the graph.
		@@ -1135,8 +1303,32 @@
1135	1303	return ['winner' => array_shift($kept), 'siblings' => array_values($kept)];
1136	1304	}
1137	1305
1138	1306	/**
	1307	+ * The site-level entity a node's @type makes it, or '' for none.
	1308	+ *
	1309	+ * A LocalBusiness is published under the subtype the site chose in Local
	1310	+ * SEO ("Dentist", "Restaurant"), but it is still the one business entity,
	1311	+ * so every subtype shares the LocalBusiness slot: the same home-scoped
	1312	+ * `/#localbusiness` @id it always had, and the same entity for
	1313	+ * find_same_entity() to fold a per-post copy into. Matching the literal
	1314	+ * type instead would have given a Dentist a fresh page-scoped @id on every
	1315	+ * URL, the exact split #471 closed.
	1316	+ *
	1317	+ * @since 2.10.0
	1318	+ *
	1319	+ * @param mixed $type Node @type.
	1320	+ * @return string 'LocalBusiness', 'Person' or ''.
	1321	+ */
	1322	+ private function site_level_slot($type): string {
	1323	+ if (\ThinkRank\Config\Local_Business_Types_Config::is_local_business($type)) {
	1324	+ return 'LocalBusiness';
	1325	+ }
	1326	+
	1327	+ return (is_string($type) && in_array($type, self::SITE_LEVEL_TYPES, true)) ? $type : '';
	1328	+ }
	1329	+
	1330	+ /**
1139	1331	* Find an already-placed node describing the same entity as $node.
1140	1332	*
1141	1333	* Identity is `email` when both carry one — two people can share a name,
1142	1334	* but not a mailbox — and a case-insensitive `name` match otherwise. A node
		@@ -1145,9 +1337,9 @@
1145	1337	*
1146	1338	* @since 2.0.2
1147	1339	*
1148	1340	* @param array $nodes Nodes placed so far, keyed.
1149		- * @param string $type Schema type to match within.
	1341	+ * @param string $type Site-level slot to match within (see site_level_slot()).
1150	1342	* @param array $node Candidate node.
1151	1343	* @return string|null Key of the matching node, or null.
1152	1344	*/
1153	1345	private function find_same_entity(array $nodes, string $type, array $node): ?string {
		@@ -1158,9 +1350,9 @@
1158	1350	return null;
1159	1351	}
1160	1352
1161	1353	foreach ($nodes as $key => $placed) {
1162		- if (($placed['@type'] ?? '') !== $type) {
	1354	+ if ($this->site_level_slot($placed['@type'] ?? '') !== $type) {
1163	1355	continue;
1164	1356	}
1165	1357
1166	1358	$placed_email = isset($placed['email']) ? strtolower(trim((string) $placed['email'])) : '';