← All changes | app/Services/SEOService.php +265 -15 3.0.7 → 3.0.16 View file →

		@@ -233,8 +233,30 @@
233	233	return wp_strip_all_tags(strip_tags($text));
234	234	}
235	235
236	236	/**
	237	+ * Expand SEO meta placeholders for taxonomy terms.
	238	+ *
	239	+ * The destination / activity / category meta forms tell the operator to
	240	+ * "Use {name} as placeholder" in the meta title, description and keywords.
	241	+ * Those values are stored verbatim (the term may be renamed later), so the
	242	+ * token must be substituted here, at render time, with the term's name.
	243	+ * No-op when the value contains no token, so plain values are untouched.
	244	+ *
	245	+ * @param string $value Stored meta value, possibly containing {name}
	246	+ * @param string $name Term name to substitute
	247	+ * @return string Value with {name} replaced
	248	+ */
	249	+ private function expandTermTokens(string $value, string $name): string
	250	+ {
	251	+ if ($value === '' || strpos($value, '{') === false) {
	252	+ return $value;
	253	+ }
	254	+
	255	+ return str_replace(['{name}', '{Name}'], $name, $value);
	256	+ }
	257	+
	258	+ /**
237	259	* Validate and sanitize URL
238	260	*
239	261	* @param string $url URL to validate
240	262	* @return string Validated URL
		@@ -413,13 +435,15 @@
413	435	$metadata = \is_array($maybe) ? $maybe : [];
414	436	}
415	437	}
416	438
	439	+ $name = (string) ($destination->name ?? '');
417	440	$title = $metadata['seo_title'] ?? $destination->name ?? '';
418		- $this->seoData['title'] = $this->sanitizeText((string) $title);
	441	+ $this->seoData['title'] = $this->sanitizeText($this->expandTermTokens((string) $title, $name));
419	442	$descRaw = $metadata['seo_description'] ?? $destination->description ?? '';
420		- $this->seoData['description'] = $this->sanitizeText(\wp_trim_words(\wp_strip_all_tags((string) $descRaw), 45, '…'));
421		- $this->seoData['keywords'] = $this->sanitizeText((string) ($metadata['seo_keywords'] ?? ''));
	443	+ $descRaw = $this->expandTermTokens((string) $descRaw, $name);
	444	+ $this->seoData['description'] = $this->sanitizeText(\wp_trim_words(\wp_strip_all_tags($descRaw), 45, '…'));
	445	+ $this->seoData['keywords'] = $this->sanitizeText($this->expandTermTokens((string) ($metadata['seo_keywords'] ?? ''), $name));
422	446
423	447	// Get featured image or gallery image
424	448	$this->seoData['image'] = $destination->featured_image_url ??
425	449	($destination->gallery_images[0] ?? '');
		@@ -452,12 +476,14 @@
452	476	$metadata = \is_array($maybe) ? $maybe : [];
453	477	}
454	478	}
455	479
456		- $this->seoData['title'] = $this->sanitizeText((string) ($metadata['seo_title'] ?? $activity->name ?? ''));
	480	+ $name = (string) ($activity->name ?? '');
	481	+ $this->seoData['title'] = $this->sanitizeText($this->expandTermTokens((string) ($metadata['seo_title'] ?? $activity->name ?? ''), $name));
457	482	$descRaw = $metadata['seo_description'] ?? $activity->description ?? '';
458		- $this->seoData['description'] = $this->sanitizeText(\wp_trim_words(\wp_strip_all_tags((string) $descRaw), 45, '…'));
459		- $this->seoData['keywords'] = $this->sanitizeText((string) ($metadata['seo_keywords'] ?? ''));
	483	+ $descRaw = $this->expandTermTokens((string) $descRaw, $name);
	484	+ $this->seoData['description'] = $this->sanitizeText(\wp_trim_words(\wp_strip_all_tags($descRaw), 45, '…'));
	485	+ $this->seoData['keywords'] = $this->sanitizeText($this->expandTermTokens((string) ($metadata['seo_keywords'] ?? ''), $name));
460	486
461	487	// Get featured image or gallery image
462	488	$this->seoData['image'] = $activity->featured_image_url ??
463	489	($activity->gallery_images[0] ?? '');
		@@ -490,12 +516,14 @@
490	516	$metadata = \is_array($maybe) ? $maybe : [];
491	517	}
492	518	}
493	519
494		- $this->seoData['title'] = $this->sanitizeText((string) ($metadata['seo_title'] ?? $category->name ?? ''));
	520	+ $name = (string) ($category->name ?? '');
	521	+ $this->seoData['title'] = $this->sanitizeText($this->expandTermTokens((string) ($metadata['seo_title'] ?? $category->name ?? ''), $name));
495	522	$descRaw = $metadata['seo_description'] ?? $category->description ?? '';
496		- $this->seoData['description'] = $this->sanitizeText(\wp_trim_words(\wp_strip_all_tags((string) $descRaw), 45, '…'));
497		- $this->seoData['keywords'] = $this->sanitizeText((string) ($metadata['seo_keywords'] ?? ''));
	523	+ $descRaw = $this->expandTermTokens((string) $descRaw, $name);
	524	+ $this->seoData['description'] = $this->sanitizeText(\wp_trim_words(\wp_strip_all_tags($descRaw), 45, '…'));
	525	+ $this->seoData['keywords'] = $this->sanitizeText($this->expandTermTokens((string) ($metadata['seo_keywords'] ?? ''), $name));
498	526
499	527	// Get featured image or gallery image
500	528	$this->seoData['image'] = $category->featured_image_url ??
501	529	($category->gallery_images[0] ?? '');
		@@ -602,8 +630,46 @@
602	630	return $raw !== '' ? $this->validateUrl($raw) : '';
603	631	}
604	632
605	633	/**
	634	+ * The page's language as WordPress reports it, reduced to what hreflang
	635	+ * and Open Graph accept: language + optional region (`de-DE`, `en-US`,
	636	+ * `ca`). Read at render time through get_locale()/the `locale` filter, so
	637	+ * a German install, WPML and Polylang per-page languages all resolve
	638	+ * correctly. WordPress variant locales such as `de_DE_formal` ("Deutsch
	639	+ * (Sie)") or `pt_PT_ao90` carry a third segment that is not a region —
	640	+ * Google ignores an hreflang like `de-DE-formal` and Facebook rejects
	641	+ * `de_DE_formal` — so only the first two segments are kept. Falls back to
	642	+ * `en-US` when WP has no locale, which keeps existing English sites
	643	+ * byte-identical.
	644	+ */
	645	+ private function languageTag(): string
	646	+ {
	647	+ $locale = (string) get_locale();
	648	+ if ($locale === '') {
	649	+ $locale = str_replace('-', '_', (string) get_bloginfo('language'));
	650	+ }
	651	+
	652	+ $parts = preg_split('/[_-]/', $locale) ?: [];
	653	+ $language = strtolower((string) ($parts[0] ?? ''));
	654	+ if (!preg_match('/^[a-z]{2,3}$/', $language)) {
	655	+ return 'en-US';
	656	+ }
	657	+
	658	+ $region = strtoupper((string) ($parts[1] ?? ''));
	659	+
	660	+ return preg_match('/^[A-Z]{2}$/', $region) ? $language . '-' . $region : $language;
	661	+ }
	662	+
	663	+ /**
	664	+ * Same language in Open Graph form (`de_DE`, `en_US`).
	665	+ */
	666	+ private function ogLocale(): string
	667	+ {
	668	+ return str_replace('-', '_', $this->languageTag());
	669	+ }
	670	+
	671	+ /**
606	672	* Output basic meta tags
607	673	*/
608	674	private function outputBasicMetaTags(): void
609	675	{
		@@ -620,9 +686,9 @@
620	686	* Output Open Graph meta tags
621	687	*/
622	688	private function outputOpenGraphTags(): void
623	689	{
624		- echo '<meta property="og:locale" content="en_US">' . "\n";
	690	+ echo '<meta property="og:locale" content="' . esc_attr($this->ogLocale()) . '">' . "\n";
625	691	echo '<meta property="og:site_name" content="' . esc_attr(get_bloginfo('name')) . '">' . "\n";
626	692	echo '<meta property="og:title" content="' . esc_attr($this->seoData['title']) . '">' . "\n";
627	693	echo '<meta property="og:description" content="' . esc_attr($this->truncateText($this->seoData['description'], 160)) . '">' . "\n";
628	694	echo '<meta property="og:type" content="' . esc_attr($this->seoData['type']) . '">' . "\n";
		@@ -658,18 +724,81 @@
658	724	}
659	725	}
660	726
661	727	/**
	728	+ * The robots directive for this page.
	729	+ *
	730	+ * Normally "index, follow" with the usual snippet hints, exactly as before.
	731	+ * When the operator has both dropped this content type from the sitemap and
	732	+ * opted in to noindexing excluded types, it becomes "noindex, follow" —
	733	+ * because removing a URL from a sitemap does not stop Google indexing it,
	734	+ * and this page was previously asserting the opposite.
	735	+ *
	736	+ * "follow" is kept deliberately: the page should stop being indexed, but
	737	+ * links out of it (to trips that ARE published) should still be crawled.
	738	+ */
	739	+ private function robotsDirective(): string
	740	+ {
	741	+ $indexable = 'index, follow, max-snippet:-1, max-image-preview:large, max-video-preview:-1';
	742	+
	743	+ if (!SettingsService::isEnabled('sitemap_noindex_excluded')) {
	744	+ return $indexable;
	745	+ }
	746	+
	747	+ $sitemapType = $this->sitemapTypeForPage();
	748	+ if ($sitemapType === '') {
	749	+ return $indexable;
	750	+ }
	751	+
	752	+ if (!class_exists(\Yatra\Sitemap\SitemapService::class)) {
	753	+ return $indexable;
	754	+ }
	755	+
	756	+ $sitemap = new \Yatra\Sitemap\SitemapService();
	757	+
	758	+ return $sitemap->isTypeEnabled($sitemapType) ? $indexable : 'noindex, follow';
	759	+ }
	760	+
	761	+ /**
	762	+ * Map this page onto the sitemap content type that governs it.
	763	+ *
	764	+ * The taxonomy listing pages are published by the sitemap's `archive` type
	765	+ * (the trip listing plus each taxonomy index), so they follow it.
	766	+ *
	767	+ * @return string '' when no sitemap type owns this page.
	768	+ */
	769	+ private function sitemapTypeForPage(): string
	770	+ {
	771	+ switch ($this->pageType) {
	772	+ case self::PAGE_TYPE_TRIP:
	773	+ return \Yatra\Sitemap\SitemapService::TYPE_TRIP;
	774	+ case self::PAGE_TYPE_DESTINATION:
	775	+ return \Yatra\Sitemap\SitemapService::TYPE_DESTINATION;
	776	+ case self::PAGE_TYPE_ACTIVITY:
	777	+ return \Yatra\Sitemap\SitemapService::TYPE_ACTIVITY;
	778	+ case self::PAGE_TYPE_CATEGORY:
	779	+ return \Yatra\Sitemap\SitemapService::TYPE_CATEGORY;
	780	+ case self::PAGE_TYPE_TRIP_ARCHIVE:
	781	+ case self::PAGE_TYPE_DESTINATION_LISTING:
	782	+ case self::PAGE_TYPE_ACTIVITY_LISTING:
	783	+ case self::PAGE_TYPE_CATEGORY_LISTING:
	784	+ return \Yatra\Sitemap\SitemapService::TYPE_ARCHIVE;
	785	+ default:
	786	+ return '';
	787	+ }
	788	+ }
	789	+
	790	+ /**
662	791	* Output advanced meta tags
663	792	*/
664	793	private function outputAdvancedMetaTags(): void
665	794	{
666		- echo '<meta name="robots" content="index, follow, max-snippet:-1, max-image-preview:large, max-video-preview:-1">' . "\n";
	795	+ echo '<meta name="robots" content="' . esc_attr($this->robotsDirective()) . '">' . "\n";
667	796	echo '<link rel="canonical" href="' . esc_url($this->seoData['url']) . '">' . "\n";
668	797	echo '<meta name="author" content="' . esc_attr($this->seoData['author']) . '">' . "\n";
669	798	echo '<meta name="publisher" content="' . esc_attr($this->seoData['publisher']) . '">' . "\n";
670	799	echo '<meta name="lastmod" content="' . esc_attr($this->seoData['modified_time']) . '">' . "\n";
671		- echo '<link rel="alternate" hreflang="en-US" href="' . esc_url($this->seoData['url']) . '">' . "\n";
	800	+ echo '<link rel="alternate" hreflang="' . esc_attr($this->languageTag()) . '" href="' . esc_url($this->seoData['url']) . '">' . "\n";
672	801	echo '<meta name="revisit-after" content="7 days">' . "\n";
673	802	echo '<meta name="distribution" content="global">' . "\n";
674	803	echo '<meta name="rating" content="general">' . "\n";
675	804	}
		@@ -766,9 +895,13 @@
766	895	private function outputSchemaMarkup(): void
767	896	{
768	897	$schema = $this->generateSchemaMarkup();
769	898	if (!empty($schema)) {
770		- echo '<script type="application/ld+json">' . json_encode($schema, JSON_UNESCAPED_SLASHES | JSON_UNESCAPED_UNICODE) . '</script>' . "\n";
	899	+ // JSON_HEX_TAG|JSON_HEX_AMP escape < > & as \u00xx so no string value
	900	+ // (e.g. user-submitted review text/author) can break out of this
	901	+ // <script> block — a </script> in a review would otherwise be XSS.
	902	+ // Google parses the \u-escaped JSON identically.
	903	+ echo '<script type="application/ld+json">' . json_encode($schema, JSON_UNESCAPED_SLASHES | JSON_UNESCAPED_UNICODE | JSON_HEX_TAG | JSON_HEX_AMP) . '</script>' . "\n";
771	904	}
772	905	}
773	906
774	907	/**
		@@ -787,9 +920,9 @@
787	920	'url' => home_url()
788	921	],
789	922	'dateModified' => $this->seoData['modified_time'],
790	923	'datePublished' => $this->seoData['published_time'],
791		- 'inLanguage' => 'en-US',
	924	+ 'inLanguage' => $this->languageTag(),
792	925	'isPartOf' => [
793	926	'@type' => 'WebSite',
794	927	'name' => get_bloginfo('name'),
795	928	'url' => home_url()
		@@ -808,11 +941,16 @@
808	941	case self::PAGE_TYPE_CATEGORY_LISTING:
809	942	return $this->generateCollectionPageSchema($baseSchema);
810	943	case self::PAGE_TYPE_DESTINATION:
811	944	return $this->generatePlaceSchema($baseSchema);
	945	+ case self::PAGE_TYPE_TRIP:
	946	+ // A tour with ratings/reviews must be a Google-supported review
	947	+ // type (Product) with the rating + reviews nested inside it —
	948	+ // NOT a TouristTrip/Article, which Google rejects for review
	949	+ // snippets ("invalid object type for parent_node").
	950	+ return $this->generateTripProductSchema($baseSchema);
812	951	case self::PAGE_TYPE_ACTIVITY:
813	952	case self::PAGE_TYPE_CATEGORY:
814		- case self::PAGE_TYPE_TRIP:
815	953	return $this->generateArticleSchema($baseSchema);
816	954	default:
817	955	return $baseSchema;
818	956	}
		@@ -861,8 +999,120 @@
861	999	'@type' => 'Organization',
862	1000	'name' => get_bloginfo('name')
863	1001	]
864	1002	]);
	1003	+ }
	1004	+
	1005	+ /**
	1006	+ * Generate Product schema for a single trip, with rating + reviews nested
	1007	+ * INSIDE the product (the structure Google requires for review snippets).
	1008	+ *
	1009	+ * `aggregateRating` and `review` are added only when approved reviews exist —
	1010	+ * an empty aggregateRating is itself a structured-data error. Each review
	1011	+ * carries the required author / reviewRating / reviewBody / datePublished so
	1012	+ * Google no longer reports "Missing field author" or "Missing field itemReviewed".
	1013	+ */
	1014	+ private function generateTripProductSchema(array $baseSchema): array
	1015	+ {
	1016	+ $trip = $this->pageObject;
	1017	+
	1018	+ $tripId = (\is_object($trip) && isset($trip->id)) ? (int) $trip->id : 0;
	1019	+ $avg = (\is_object($trip) && \method_exists($trip, 'getAverageRating'))
	1020	+ ? (float) $trip->getAverageRating()
	1021	+ : 0.0;
	1022	+ $count = (\is_object($trip) && \method_exists($trip, 'getReviewCount'))
	1023	+ ? (int) $trip->getReviewCount()
	1024	+ : 0;
	1025	+
	1026	+ // No approved reviews → keep the generic Article (unchanged behaviour).
	1027	+ // A Product is emitted ONLY when there's a real rating to nest, which is
	1028	+ // exactly what fixes the review markup — and it avoids emitting a bare
	1029	+ // Product (offers/review/rating all absent) that Google would flag as
	1030	+ // incomplete on the many tours that have no reviews yet.
	1031	+ if (!($tripId > 0 && $count > 0 && $avg > 0)) {
	1032	+ return $this->generateArticleSchema($baseSchema);
	1033	+ }
	1034	+
	1035	+ $schema = [
	1036	+ '@context' => 'https://schema.org',
	1037	+ '@type' => 'Product',
	1038	+ 'name' => $this->seoData['title'],
	1039	+ 'url' => $this->seoData['url'],
	1040	+ 'brand' => [
	1041	+ '@type' => 'Brand',
	1042	+ 'name' => get_bloginfo('name'),
	1043	+ ],
	1044	+ ];
	1045	+ if (!empty($this->seoData['description'])) {
	1046	+ $schema['description'] = $this->seoData['description'];
	1047	+ }
	1048	+ if (!empty($this->seoData['image'])) {
	1049	+ $schema['image'] = $this->seoData['image'];
	1050	+ }
	1051	+
	1052	+ $schema['aggregateRating'] = [
	1053	+ '@type' => 'AggregateRating',
	1054	+ 'ratingValue' => (string) round($avg, 1),
	1055	+ 'reviewCount' => (string) $count,
	1056	+ 'bestRating' => '5',
	1057	+ 'worstRating' => '1',
	1058	+ ];
	1059	+
	1060	+ $rows = [];
	1061	+ try {
	1062	+ $rows = (new \Yatra\Repositories\ReviewRepository())->findApprovedByTripId($tripId, 10);
	1063	+ } catch (\Throwable $e) {
	1064	+ $rows = [];
	1065	+ }
	1066	+
	1067	+ $reviews = [];
	1068	+ foreach ($rows as $row) {
	1069	+ $rating = (int) ($row->rating ?? 0);
	1070	+ if ($rating < 1 || $rating > 5) {
	1071	+ continue;
	1072	+ }
	1073	+ // Strip tags on every user-derived value — even though the emitter
	1074	+ // now \u-escapes < > &, keep the data itself clean/plain-text.
	1075	+ $authorName = trim(\wp_strip_all_tags((string) ($row->author_name ?? $row->user_display_name ?? '')));
	1076	+ if ($authorName === '') {
	1077	+ $authorName = __('Anonymous', 'yatra');
	1078	+ }
	1079	+
	1080	+ $review = [
	1081	+ '@type' => 'Review',
	1082	+ 'author' => ['@type' => 'Person', 'name' => $authorName],
	1083	+ 'reviewRating' => [
	1084	+ '@type' => 'Rating',
	1085	+ 'ratingValue' => (string) $rating,
	1086	+ 'bestRating' => '5',
	1087	+ 'worstRating' => '1',
	1088	+ ],
	1089	+ ];
	1090	+
	1091	+ $title = $this->sanitizeText(trim((string) ($row->title ?? '')));
	1092	+ if ($title !== '') {
	1093	+ $review['name'] = $title;
	1094	+ }
	1095	+ $body = $this->sanitizeText(trim(\wp_strip_all_tags((string) ($row->content ?? ''))));
	1096	+ if ($body !== '') {
	1097	+ $review['reviewBody'] = $body;
	1098	+ }
	1099	+ $created = (string) ($row->created_at ?? '');
	1100	+ if ($created !== '') {
	1101	+ $ts = strtotime($created);
	1102	+ if ($ts) {
	1103	+ $review['datePublished'] = date('Y-m-d', $ts);
	1104	+ }
	1105	+ }
	1106	+
	1107	+ $reviews[] = $review;
	1108	+ }
	1109	+
	1110	+ if (!empty($reviews)) {
	1111	+ $schema['review'] = $reviews;
	1112	+ }
	1113	+
	1114	+ return $schema;
865	1115	}
866	1116
867	1117	/**
868	1118	* Truncate text to specified length