← All changes | includes/seo/class-schema-builder.php +121 -44 2.0.2 → 2.14.3 View file →

		@@ -91,8 +91,14 @@
91	91	case 'WebSite':
92	92	$schema = $this->populate_website_schema($schema, $data, $context);
93	93	break;
94	94	case 'WebPage':
	95	+ case 'AboutPage':
	96	+ case 'ContactPage':
	97	+ case 'ProfilePage':
	98	+ // All four carry the same WebPage properties; only @type
	99	+ // differs, and create_base_schema() has already set it from the
	100	+ // requested type (#624).
95	101	$schema = $this->populate_webpage_schema($schema, $data, $context);
96	102	break;
97	103	case 'FAQPage':
98	104	$schema = $this->populate_faq_schema($schema, $data, $context);
		@@ -457,8 +463,54 @@
457	463	return '';
458	464	}
459	465
460	466	/**
	467	+ * The WebSite SearchAction for the sitelinks search box, or null when the
	468	+ * site switched it off.
	469	+ *
	470	+ * Reads the Website tab's two settings. website_enable_search absent (or
	471	+ * null) means not configured, which stays on; any other value is read as
	472	+ * a boolean, so the '' a switched-off toggle is stored as turns it off.
	473	+ *
	474	+ * website_search_url is used as the urlTemplate when it is an http(s) URL,
	475	+ * or a site path, and carries the {search_term_string} placeholder that
	476	+ * query-input names. Without the placeholder the template cannot take a
	477	+ * query, so the default WordPress search URL is used instead, as it is
	478	+ * when the field is empty.
	479	+ *
	480	+ * @since 2.15.0
	481	+ *
	482	+ * @param array $settings Schema settings or site_data.
	483	+ * @return array|null SearchAction node, or null when switched off.
	484	+ */
	485	+ public static function website_search_action(array $settings): ?array {
	486	+ $enabled = $settings['website_enable_search'] ?? null;
	487	+ if (null !== $enabled && empty($enabled)) {
	488	+ return null;
	489	+ }
	490	+
	491	+ $url_template = home_url('/?s={search_term_string}');
	492	+
	493	+ $custom = trim((string) ($settings['website_search_url'] ?? ''));
	494	+ if ('' !== $custom && false !== strpos($custom, '{search_term_string}')) {
	495	+ if (\ThinkRank\Core\Url_Validator::is_http_url($custom)) {
	496	+ $url_template = $custom;
	497	+ } elseif (\ThinkRank\Core\Url_Validator::is_http_url_or_path($custom)) {
	498	+ $url_template = home_url($custom);
	499	+ }
	500	+ }
	501	+
	502	+ return [
	503	+ '@type' => 'SearchAction',
	504	+ 'target' => [
	505	+ '@type' => 'EntryPoint',
	506	+ 'urlTemplate' => $url_template
	507	+ ],
	508	+ 'query-input' => 'required name=search_term_string'
	509	+ ];
	510	+ }
	511	+
	512	+ /**
461	513	* Populate Organization schema
462	514	* PRESERVED: Exact same method logic from original Schema_Generator
463	515	*
464	516	* @since 1.0.0
		@@ -487,11 +539,18 @@
487	539	$data['url'] ?? null,
488	540	home_url()
489	541	);
490	542
491		- // Logo from user configuration or theme customizer
492		- if (!empty($data['site_data']['organization_logo'])) {
493		- $schema['logo'] = $this->format_image_schema($data['site_data']['organization_logo']);
	543	+ // Logo: the Schema Manager's Organization logo, then the Site Identity
	544	+ // logo (the one WebSite.publisher and LocalBusiness already use, so the
	545	+ // same entity carries the same logo everywhere), then the theme's
	546	+ // custom logo. With none of them, logo is left out (#1010).
	547	+ $logo_url = $this->first_non_empty(
	548	+ $data['site_data']['organization_logo'] ?? null,
	549	+ $data['site_data']['logo_url'] ?? null
	550	+ );
	551	+ if ('' !== $logo_url) {
	552	+ $schema['logo'] = $this->format_image_schema($logo_url);
494	553	} else {
495	554	// Fallback to WordPress custom logo
496	555	$custom_logo_id = get_theme_mod('custom_logo');
497	556	if ($custom_logo_id) {
		@@ -722,12 +781,25 @@
722	781	return $schema;
723	782	}
724	783
725	784	/**
726		- * Truncate text to specified length
727		- * PRESERVED: Exact same method logic from original Schema_Generator
	785	+ * Normalise text for JSON-LD and truncate it to a length.
728	786	*
	787	+ * Every description, headline and reviewBody this class emits passes
	788	+ * through here, so this is the one place the deployed nodes get the same
	789	+ * treatment the automatic ones have had since #766. Before, it only
	790	+ * stripped tags: the block editor stores `&` as `&`, which was then
	791	+ * published literally (nothing decodes JSON-LD), and the cut landed
	792	+ * mid-word with "..." glued on. A deployed node outranks the automatic
	793	+ * one, so deploying a page type made its description worse than not
	794	+ * deploying anything.
	795	+ *
	796	+ * Seo_Text::trim_to_length() is multibyte-aware, which the #473 fix here
	797	+ * already required (a byte cut mojibaked every non-Latin description),
	798	+ * and falls back to a character cut for scripts without spaces.
	799	+ *
729	800	* @since 1.0.0
	801	+ * @since 2.10.0 Decodes entities and cuts on a word boundary.
730	802	*
731	803	* @param string $text Text to truncate
732	804	* @param int $length Maximum length
733	805	* @return string Truncated text
		@@ -732,19 +804,12 @@
732	804	* @param int $length Maximum length
733	805	* @return string Truncated text
734	806	*/
735	807	private function truncate_text(string $text, int $length): string {
736		- $text = wp_strip_all_tags($text);
737		-
738		- // Multibyte-aware. strlen()/substr() count bytes, so a cut landing
739		- // mid-character produced invalid UTF-8 — wp_json_encode()'s sanity
740		- // check then replaced the tail with "?", mojibaking every non-Latin
741		- // site's description and headline (#473).
742		- if (mb_strlen($text) <= $length) {
743		- return $text;
744		- }
745		-
746		- return mb_substr($text, 0, max(0, $length - 3)) . '...';
	808	+ return \ThinkRank\Core\Seo_Text::trim_to_length(
	809	+ \ThinkRank\Core\Seo_Text::normalize_schema_text($text),
	810	+ $length
	811	+ );
747	812	}
748	813
749	814	/**
750	815	* Normalise a date into ISO 8601 with a timezone offset.
		@@ -827,17 +892,18 @@
827	892	'@type' => 'ImageObject',
828	893	'url' => $image_url
829	894	];
830	895
831		- // Try to get image dimensions if it's a WordPress attachment
832		- $attachment_id = attachment_url_to_postid($image_url);
	896	+ // Try to get image dimensions if it's a WordPress attachment — those
	897	+ // of the file this URL names, which need not be the original upload.
	898	+ $attachment_id = Attachment_Lookup::id_from_url($image_url);
833	899	if ($attachment_id) {
834		- $image_data = wp_get_attachment_image_src($attachment_id, 'full');
	900	+ $image_file = Attachment_Lookup::describe($attachment_id, $image_url);
835	901	// SVGs report 0x0 — omit the dimensions rather than emitting
836	902	// zeroes, which invalidate the ImageObject.
837		- if ($image_data && (int) $image_data[1] > 0 && (int) $image_data[2] > 0) {
838		- $image_schema['width'] = (int) $image_data[1];
839		- $image_schema['height'] = (int) $image_data[2];
	903	+ if ($image_file['width'] > 0 && $image_file['height'] > 0) {
	904	+ $image_schema['width'] = $image_file['width'];
	905	+ $image_schema['height'] = $image_file['height'];
840	906	}
841	907	}
842	908
843	909	return $image_schema;
		@@ -911,9 +977,9 @@
911	977	'organization_social_telegram'
912	978	];
913	979
914	980	foreach ($social_fields as $field) {
915		- if (!empty($settings[$field]) && filter_var($settings[$field], FILTER_VALIDATE_URL)) {
	981	+ if (!empty($settings[$field]) && \ThinkRank\Core\Url_Validator::is_valid($settings[$field])) {
916	982	$social_profiles[] = $settings[$field];
917	983	}
918	984	}
919	985
		@@ -1039,8 +1105,18 @@
1039	1105	$data['url'] ?? '',
1040	1106	home_url()
1041	1107	);
1042	1108
	1109	+ // Both WebSite producers have to carry this or the deployed node and the
	1110	+ // default one disagree about the same site — the shape of failure #688
	1111	+ // documents. The default node is generate_website_schema() (#692).
	1112	+ $alternate_name = \ThinkRank\SEO\Site_Identity_Manager::alternate_name_for_schema(
	1113	+ $data['site_data']['alternate_name'] ?? null
	1114	+ );
	1115	+ if (null !== $alternate_name) {
	1116	+ $schema['alternateName'] = $alternate_name;
	1117	+ }
	1118	+
1043	1119	// Recommended properties - prioritize user-configured Website schema description
1044	1120	if (!empty($data['site_data']['website_description'])) {
1045	1121	$schema['description'] = $this->truncate_text($data['site_data']['website_description'], 160);
1046	1122	} elseif (!empty($data['content'])) {
		@@ -1045,9 +1121,13 @@
1045	1121	$schema['description'] = $this->truncate_text($data['site_data']['website_description'], 160);
1046	1122	} elseif (!empty($data['content'])) {
1047	1123	$schema['description'] = $this->truncate_text($data['content'], 160);
1048	1124	} else {
1049		- $schema['description'] = get_bloginfo('description');
	1125	+ // The tagline is stored esc_html()'d by sanitize_option(), so an
	1126	+ // ampersand in it arrives as `&amp;`.
	1127	+ $schema['description'] = \ThinkRank\Core\Seo_Text::normalize_schema_text(
	1128	+ (string) get_bloginfo('description')
	1129	+ );
1050	1130	}
1051	1131
1052	1132	// Author - use organization or person data
1053	1133	if (!empty($data['site_data']['organization_name'])) {
		@@ -1103,22 +1183,14 @@
1103	1183	}
1104	1184
1105	1185	$schema['publisher'] = $publisher;
1106	1186
1107		- // Search action for sitelinks search box (optional but recommended)
1108		- if ($data['site_data']['website_enable_search'] ?? true) {
1109		- $search_url = $this->first_non_empty(
1110		- $data['site_data']['website_search_url'] ?? '',
1111		- home_url('/?s={search_term_string}')
1112		- );
1113		- $schema['potentialAction'] = [
1114		- '@type' => 'SearchAction',
1115		- 'target' => [
1116		- '@type' => 'EntryPoint',
1117		- 'urlTemplate' => $search_url
1118		- ],
1119		- 'query-input' => 'required name=search_term_string'
1120		- ];
	1187	+ // Search action for sitelinks search box (optional but recommended).
	1188	+ // Shared with the live fallback in SEO_Manager so the deployed and
	1189	+ // live WebSite nodes read the switch and the URL the same way (#1009).
	1190	+ $search_action = self::website_search_action($data['site_data'] ?? []);
	1191	+ if (null !== $search_action) {
	1192	+ $schema['potentialAction'] = $search_action;
1121	1193	}
1122	1194
1123	1195	// Social media profiles (sameAs)
1124	1196	$social_profiles = $this->get_social_profiles();
		@@ -1280,8 +1352,18 @@
1280	1352	private function populate_local_business_schema(array $schema, array $data, string $context): array {
1281	1353	// Get business data from Site Identity Business Info (single source of truth)
1282	1354	$business_data = $this->get_business_data_from_site_identity();
1283	1355
	1356	+ // The business type chosen in Local SEO is what this node IS. It was
	1357	+ // read into $business_data and then never used, so every site published
	1358	+ // "@type": "LocalBusiness" whatever it had picked, and the ~150-type
	1359	+ // selector (#623) changed nothing but a meta tag. create_base_schema()
	1360	+ // has already put @type in place, so reassigning keeps key order and a
	1361	+ // site on the default publishes exactly what it did before.
	1362	+ $schema['@type'] = \ThinkRank\Config\Local_Business_Types_Config::schema_type(
	1363	+ $business_data['business_type'] ?? ''
	1364	+ );
	1365	+
1284	1366	// Required properties - use business name from Business Info.
1285	1367	// `name` is required for LocalBusiness, so an empty saved value must fall
1286	1368	// through to the next source rather than emit "".
1287	1369	$schema['name'] = $this->first_non_empty(
		@@ -1561,14 +1643,9 @@
1561	1643	$social_profiles,
1562	1644	static function ($url) {
1563	1645	return is_string($url)
1564	1646	&& '' !== trim($url)
1565		- && filter_var($url, FILTER_VALIDATE_URL)
1566		- && in_array(
1567		- strtolower((string) wp_parse_url($url, PHP_URL_SCHEME)),
1568		- ['http', 'https'],
1569		- true
1570		- );
	1647	+ && \ThinkRank\Core\Url_Validator::is_http_url($url);
1571	1648	}
1572	1649	)));
1573	1650
1574	1651	if (!empty($social_profiles)) {