← All changes | includes/seo/class-schema-input-validator.php +174 -14 2.0.0 → 2.14.1 View file →

		@@ -118,8 +118,48 @@
118	118	'VideoObject' => [
119	119	'required_fields' => ['@type', 'name', 'thumbnailUrl', 'uploadDate'],
120	120	'optional_fields' => ['description', 'contentUrl', 'embedUrl', 'duration', 'url'],
121	121	'max_length' => ['name' => 110, 'description' => 160]
	122	+ ],
	123	+ // Offered by the metabox Schema Type dropdown and registered in
	124	+ // Schema_Factory, but missing here — so a Review could be generated and
	125	+ // never deployed (#462). Field list mirrors Schema_Factory::Review.
	126	+ 'Review' => [
	127	+ 'required_fields' => ['@type', 'itemReviewed', 'reviewRating', 'author'],
	128	+ 'optional_fields' => ['reviewBody', 'datePublished', 'publisher', 'name', 'url'],
	129	+ 'max_length' => ['name' => 110, 'reviewBody' => 500]
	130	+ ],
	131	+ // The WebPage family. #624 made all four selectable in the metabox and
	132	+ // taught the generate/validate/optimize/preview routes, Schema_Builder,
	133	+ // Schema_Factory and Schema_Graph about them — but not this array, and
	134	+ // deploy_schema_markup() gates every entry on it. So each one generated
	135	+ // cleanly, validated at a score of 100, reported deployment_ready, then
	136	+ // failed deployment with "Invalid schema type: AboutPage" and no row
	137	+ // written. Exactly the #462 Review bug, one array and two releases
	138	+ // later, which is why SchemaInputValidatorCoverageTest now scans the
	139	+ // dropdown instead of trusting this list to be extended by hand.
	140	+ //
	141	+ // `name` and `url` are the two populate_webpage_schema() always sets;
	142	+ // the rest are conditional, so they are optional here.
	143	+ 'WebPage' => [
	144	+ 'required_fields' => ['@type', 'name', 'url'],
	145	+ 'optional_fields' => ['description', 'datePublished', 'dateModified', 'isPartOf', 'breadcrumb', 'primaryImageOfPage', 'inLanguage', 'mainEntity'],
	146	+ 'max_length' => ['name' => 110, 'description' => 160]
	147	+ ],
	148	+ 'AboutPage' => [
	149	+ 'required_fields' => ['@type', 'name', 'url'],
	150	+ 'optional_fields' => ['description', 'datePublished', 'dateModified', 'isPartOf', 'breadcrumb', 'primaryImageOfPage', 'inLanguage', 'mainEntity'],
	151	+ 'max_length' => ['name' => 110, 'description' => 160]
	152	+ ],
	153	+ 'ContactPage' => [
	154	+ 'required_fields' => ['@type', 'name', 'url'],
	155	+ 'optional_fields' => ['description', 'datePublished', 'dateModified', 'isPartOf', 'breadcrumb', 'primaryImageOfPage', 'inLanguage', 'mainEntity'],
	156	+ 'max_length' => ['name' => 110, 'description' => 160]
	157	+ ],
	158	+ 'ProfilePage' => [
	159	+ 'required_fields' => ['@type', 'name', 'url'],
	160	+ 'optional_fields' => ['description', 'datePublished', 'dateModified', 'isPartOf', 'breadcrumb', 'primaryImageOfPage', 'inLanguage', 'mainEntity'],
	161	+ 'max_length' => ['name' => 110, 'description' => 160]
122	162	]
123	163	];
124	164
125	165	/**
		@@ -226,9 +266,9 @@
226	266	$result['errors'] = array_merge($result['errors'], $field_validation['errors']);
227	267	}
228	268
229	269	// 6. Validate data types and formats
230		- $format_validation = $this->validate_data_formats($sanitized_data, $schema_type);
	270	+ $format_validation = $this->validate_data_formats($sanitized_data, $schema_type, $schema_data);
231	271	if (!$format_validation['valid']) {
232	272	$result['errors'] = array_merge($result['errors'], $format_validation['errors']);
233	273	}
234	274	$result['warnings'] = array_merge($result['warnings'], $format_validation['warnings']);
		@@ -363,15 +403,29 @@
363	403	$result['errors'][] = 'Invalid @context value. Must be "https://schema.org"';
364	404	$result['valid'] = false;
365	405	}
366	406
367		- // Check for required @type
	407	+ // Check for required @type.
	408	+ // `@type` may be an array — "@type": ["Product","Offer"] is valid
	409	+ // JSON-LD. Comparing an array against a string emitted an "Array to
	410	+ // string conversion" warning and always failed (#468), so match if the
	411	+ // expected type appears anywhere in the list.
368	412	if (!isset($schema_data['@type'])) {
369	413	$result['errors'][] = 'Missing required @type field';
370	414	$result['valid'] = false;
371		- } elseif ($schema_data['@type'] !== $schema_type) {
372		- $result['errors'][] = "Schema @type '{$schema_data['@type']}' does not match expected type '{$schema_type}'";
373		- $result['valid'] = false;
	415	+ } else {
	416	+ $declared_types = is_array($schema_data['@type'])
	417	+ ? array_map('strval', $schema_data['@type'])
	418	+ : [(string) $schema_data['@type']];
	419	+
	420	+ if (!in_array($schema_type, $declared_types, true)) {
	421	+ $result['errors'][] = sprintf(
	422	+ "Schema @type '%s' does not match expected type '%s'",
	423	+ implode(', ', $declared_types),
	424	+ $schema_type
	425	+ );
	426	+ $result['valid'] = false;
	427	+ }
374	428	}
375	429
376	430	return $result;
377	431	}
		@@ -528,14 +582,45 @@
528	582	* @since 1.0.0
529	583	*
530	584	* @param array $schema_data Schema data
531	585	* @param string $schema_type Schema type
	586	+ * @param array $raw_data Schema data as submitted, before sanitization.
532	587	* @return array Validation result
533	588	*/
534		- private function validate_data_formats(array $schema_data, string $schema_type): array {
	589	+ private function validate_data_formats(array $schema_data, string $schema_type, array $raw_data = []): array {
535	590	$result = ['valid' => true, 'errors' => [], 'warnings' => []];
536	591
537	592	foreach ($schema_data as $field => $value) {
	593	+ // sameAs is a list, so its members never reached the string branch
	594	+ // below and free text entered in a social-profile field saved
	595	+ // cleanly, then shipped as invalid structured data (#480).
	596	+ if (is_array($value) && in_array($field, ['url', 'sameAs', 'logo', 'image'], true)) {
	597	+ // Validated after sanitization, but reported as entered:
	598	+ // esc_url_raw() turns "not a url" into "http://not%20a%20url",
	599	+ // which the user never typed (#949 review).
	600	+ $items = $this->url_candidates($value);
	601	+ $raw_items = is_array($raw_data[$field] ?? null) ? $this->url_candidates($raw_data[$field]) : [];
	602	+ if (count($raw_items) !== count($items)) {
	603	+ // A non-string member sanitized into a string would shift
	604	+ // the positions; fall back rather than name the wrong one.
	605	+ $raw_items = [];
	606	+ }
	607	+
	608	+ foreach ($items as $index => $item) {
	609	+ if ('' === trim($item)) {
	610	+ continue;
	611	+ }
	612	+
	613	+ if (!$this->is_valid_url($item)) {
	614	+ $shown = $raw_items[$index] ?? $item;
	615	+ $result['errors'][] = "Invalid URL format for field: {$field} ({$shown})";
	616	+ $result['valid'] = false;
	617	+ }
	618	+ }
	619	+
	620	+ continue;
	621	+ }
	622	+
538	623	if (is_string($value)) {
539	624	// Validate URLs
540	625	if (in_array($field, ['url', 'sameAs', 'logo', 'image'], true) && !empty($value)) {
541	626	if (!$this->is_valid_url($value)) {
		@@ -564,8 +649,47 @@
564	649	return $result;
565	650	}
566	651
567	652	/**
	653	+ * The URL strings inside an array-valued URL field.
	654	+ *
	655	+ * The field is either a list (`sameAs`, several `image` URLs) or a single
	656	+ * node such as the `ImageObject` Schema_Builder emits for a configured
	657	+ * logo. Checking every member of a node URL-checked its `@type` and
	658	+ * `width`, so "ImageObject" failed as an invalid URL and the deploy route
	659	+ * skipped — then retired — the site's Organization (#949). Only a node's
	660	+ * `url` / `contentUrl` are URLs; a list may hold strings or such nodes.
	661	+ *
	662	+ * @since 2.14.1
	663	+ *
	664	+ * @param array $value Array-valued URL field.
	665	+ * @return string[] URL strings to validate.
	666	+ */
	667	+ private function url_candidates(array $value): array {
	668	+ $node_urls = static function (array $node): array {
	669	+ return array_values(array_filter(
	670	+ [$node['url'] ?? null, $node['contentUrl'] ?? null],
	671	+ 'is_string'
	672	+ ));
	673	+ };
	674	+
	675	+ if (array_values($value) !== $value) {
	676	+ return $node_urls($value);
	677	+ }
	678	+
	679	+ $urls = [];
	680	+ foreach ($value as $item) {
	681	+ if (is_string($item)) {
	682	+ $urls[] = $item;
	683	+ } elseif (is_array($item)) {
	684	+ $urls = array_merge($urls, $node_urls($item));
	685	+ }
	686	+ }
	687	+
	688	+ return $urls;
	689	+ }
	690	+
	691	+ /**
568	692	* Validate content lengths
569	693	*
570	694	* @since 1.0.0
571	695	*
		@@ -721,14 +845,25 @@
721	845	$result['errors'][] = 'Invalid user or user not logged in';
722	846	return $result;
723	847	}
724	848
725		- // Check operation-specific permissions
	849	+ // Check operation-specific permissions.
	850	+ //
	851	+ // #457 loosened the route permission_callbacks to the delegable
	852	+ // `thinkrank_schema` capability, but these handler-level checks still
	853	+ // demanded edit_posts / publish_posts / manage_options — so a role
	854	+ // granted Schema access could generate and validate but was denied on
	855	+ // deploy, bulk operations and everything site-context. That is exactly
	856	+ // the symptom #457 set out to fix (#470). A holder of thinkrank_schema
	857	+ // satisfies any schema operation; the built-in caps remain as the
	858	+ // fallback for roles that never went through the Role Manager.
	859	+ $has_schema_cap = user_can($user_id, 'thinkrank_schema');
	860	+
726	861	switch ($operation) {
727	862	case 'generate':
728	863	case 'validate':
729	864	case 'optimize':
730		- if (!user_can($user_id, 'edit_posts')) {
	865	+ if (!$has_schema_cap && !user_can($user_id, 'edit_posts')) {
731	866	$result['errors'][] = 'Insufficient permissions for schema generation/validation';
732	867	return $result;
733	868	}
734	869	break;
		@@ -733,9 +868,9 @@
733	868	}
734	869	break;
735	870
736	871	case 'deploy':
737		- if (!user_can($user_id, 'publish_posts')) {
	872	+ if (!$has_schema_cap && !user_can($user_id, 'publish_posts')) {
738	873	$result['errors'][] = 'Insufficient permissions for schema deployment';
739	874	return $result;
740	875	}
741	876	break;
		@@ -741,9 +876,9 @@
741	876	break;
742	877
743	878	case 'manage_settings':
744	879	case 'bulk_operations':
745		- if (!user_can($user_id, 'manage_options')) {
	880	+ if (!$has_schema_cap && !user_can($user_id, 'manage_options')) {
746	881	$result['errors'][] = 'Insufficient permissions for schema management';
747	882	return $result;
748	883	}
749	884	break;
		@@ -824,9 +959,14 @@
824	959	// SECURITY: Check site-level permissions for site context.
825	960	// user_can($user_id, …) rather than current_user_can() so this
826	961	// agrees with the rest of the validator outside a REST request,
827	962	// where the current user and $user_id can differ (cron, CLI).
828		- if (!$user_id || !user_can($user_id, 'manage_options')) {
	963	+ //
	964	+ // Accepts the delegable `thinkrank_schema` capability as well as
	965	+ // manage_options: /schema/settings already lets a delegated role
	966	+ // edit site schema settings, so blocking site-context generate and
	967	+ // deploy for the same role was inconsistent (#470).
	968	+ if (!$user_id || (!user_can($user_id, 'thinkrank_schema') && !user_can($user_id, 'manage_options'))) {
829	969	$result['errors'][] = 'Access denied: You need administrator privileges for site-level schema operations';
830	970	return $result;
831	971	}
832	972	}
		@@ -901,14 +1041,34 @@
901	1041	* @return array Sanitized options
902	1042	*/
903	1043	public function sanitize_options(array $options): array {
904	1044	$sanitized = [];
	1045	+ // Anything omitted here is dropped before the manager sees it, which is
	1046	+ // why apply_content_schema_settings_from_options() and the per-request
	1047	+ // schema-type opt-in were unreachable from REST (#470). The list now
	1048	+ // covers every option the generate path actually reads.
	1049	+ //
	1050	+ // `validation_level` previously allowed 'basic' and rejected 'lenient',
	1051	+ // disagreeing with Schema_Settings_Config, validate_settings() and the
	1052	+ // update-settings ability, which all use 'lenient'.
	1053	+ // `deployment_method` no longer advertises microdata/rdfa, which
	1054	+ // determine_deployment_method() hardcodes away to json_ld anyway.
905	1055	$allowed_options = [
906		- 'deployment_method' => ['json_ld', 'microdata', 'rdfa'],
907		- 'validation_level' => ['strict', 'moderate', 'basic'],
	1056	+ 'deployment_method' => ['json_ld'],
	1057	+ 'validation_level' => ['strict', 'moderate', 'lenient'],
908	1058	'include_meta' => 'boolean',
909	1059	'minify_output' => 'boolean',
910		- 'cache_duration' => 'integer'
	1060	+ 'cache_duration' => 'integer',
	1061	+ 'rich_snippets_optimization' => 'boolean',
	1062	+ 'knowledge_graph' => 'boolean',
	1063	+ 'auto_generate_schema' => 'boolean',
	1064	+ 'enable_article_schema' => 'boolean',
	1065	+ 'enable_faq_schema' => 'boolean',
	1066	+ 'enable_howto_schema' => 'boolean',
	1067	+ 'enable_product_schema' => 'boolean',
	1068	+ 'enable_local_business' => 'boolean',
	1069	+ 'enable_breadcrumbs_schema' => 'boolean',
	1070	+ 'enable_accordion_faq_schema' => 'boolean',
911	1071	];
912	1072
913	1073	foreach ($options as $key => $value) {
914	1074	$sanitized_key = sanitize_key($key);