← All changes | includes/seo/class-schema-builder.php +47 -20 2.7.0 → 2.12.0 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);
		@@ -722,12 +728,25 @@
722	728	return $schema;
723	729	}
724	730
725	731	/**
726		- * Truncate text to specified length
727		- * PRESERVED: Exact same method logic from original Schema_Generator
	732	+ * Normalise text for JSON-LD and truncate it to a length.
728	733	*
	734	+ * Every description, headline and reviewBody this class emits passes
	735	+ * through here, so this is the one place the deployed nodes get the same
	736	+ * treatment the automatic ones have had since #766. Before, it only
	737	+ * stripped tags: the block editor stores `&` as `&`, which was then
	738	+ * published literally (nothing decodes JSON-LD), and the cut landed
	739	+ * mid-word with "..." glued on. A deployed node outranks the automatic
	740	+ * one, so deploying a page type made its description worse than not
	741	+ * deploying anything.
	742	+ *
	743	+ * Seo_Text::trim_to_length() is multibyte-aware, which the #473 fix here
	744	+ * already required (a byte cut mojibaked every non-Latin description),
	745	+ * and falls back to a character cut for scripts without spaces.
	746	+ *
729	747	* @since 1.0.0
	748	+ * @since 2.10.0 Decodes entities and cuts on a word boundary.
730	749	*
731	750	* @param string $text Text to truncate
732	751	* @param int $length Maximum length
733	752	* @return string Truncated text
		@@ -732,19 +751,12 @@
732	751	* @param int $length Maximum length
733	752	* @return string Truncated text
734	753	*/
735	754	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, $length - 3)) . '...';
	755	+ return \ThinkRank\Core\Seo_Text::trim_to_length(
	756	+ \ThinkRank\Core\Seo_Text::normalize_schema_text($text),
	757	+ $length
	758	+ );
747	759	}
748	760
749	761	/**
750	762	* Normalise a date into ISO 8601 with a timezone offset.
		@@ -827,17 +839,18 @@
827	839	'@type' => 'ImageObject',
828	840	'url' => $image_url
829	841	];
830	842
831		- // Try to get image dimensions if it's a WordPress attachment
832		- $attachment_id = attachment_url_to_postid($image_url);
	843	+ // Try to get image dimensions if it's a WordPress attachment — those
	844	+ // of the file this URL names, which need not be the original upload.
	845	+ $attachment_id = Attachment_Lookup::id_from_url($image_url);
833	846	if ($attachment_id) {
834		- $image_data = wp_get_attachment_image_src($attachment_id, 'full');
	847	+ $image_file = Attachment_Lookup::describe($attachment_id, $image_url);
835	848	// SVGs report 0x0 — omit the dimensions rather than emitting
836	849	// zeroes, which invalidate the ImageObject.
837		- if ($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) $image_data[2];
	850	+ if ($image_file['width'] > 0 && $image_file['height'] > 0) {
	851	+ $image_schema['width'] = $image_file['width'];
	852	+ $image_schema['height'] = $image_file['height'];
840	853	}
841	854	}
842	855
843	856	return $image_schema;
		@@ -1055,9 +1068,13 @@
1055	1068	$schema['description'] = $this->truncate_text($data['site_data']['website_description'], 160);
1056	1069	} elseif (!empty($data['content'])) {
1057	1070	$schema['description'] = $this->truncate_text($data['content'], 160);
1058	1071	} else {
1059		- $schema['description'] = get_bloginfo('description');
	1072	+ // The tagline is stored esc_html()'d by sanitize_option(), so an
	1073	+ // ampersand in it arrives as `&amp;`.
	1074	+ $schema['description'] = \ThinkRank\Core\Seo_Text::normalize_schema_text(
	1075	+ (string) get_bloginfo('description')
	1076	+ );
1060	1077	}
1061	1078
1062	1079	// Author - use organization or person data
1063	1080	if (!empty($data['site_data']['organization_name'])) {
		@@ -1289,8 +1306,18 @@
1289	1306	*/
1290	1307	private function populate_local_business_schema(array $schema, array $data, string $context): array {
1291	1308	// Get business data from Site Identity Business Info (single source of truth)
1292	1309	$business_data = $this->get_business_data_from_site_identity();
	1310	+
	1311	+ // The business type chosen in Local SEO is what this node IS. It was
	1312	+ // read into $business_data and then never used, so every site published
	1313	+ // "@type": "LocalBusiness" whatever it had picked, and the ~150-type
	1314	+ // selector (#623) changed nothing but a meta tag. create_base_schema()
	1315	+ // has already put @type in place, so reassigning keeps key order and a
	1316	+ // site on the default publishes exactly what it did before.
	1317	+ $schema['@type'] = \ThinkRank\Config\Local_Business_Types_Config::schema_type(
	1318	+ $business_data['business_type'] ?? ''
	1319	+ );
1293	1320
1294	1321	// Required properties - use business name from Business Info.
1295	1322	// `name` is required for LocalBusiness, so an empty saved value must fall
1296	1323	// through to the next source rather than emit "".