← All changes | includes/frontend/class-seo-manager.php +146 -5 2.7.0 → 2.12.0 View file →

		@@ -266,8 +266,16 @@
266	266	// LLMs_Txt_Manager for the static file) guarantees an explicit UTF-8
267	267	// charset. Priority 8 keeps it ahead of redirect_canonical().
268	268	add_action('template_redirect', [$this, 'maybe_serve_llms_txt'], 8);
269	269
	270	+ // Serve the sitemap from PHP on sites whose web root cannot be written.
	271	+ // ThinkRank publishes sitemaps as real files, so where that is possible
	272	+ // the web server answers first and this never runs; where it is not,
	273	+ // this is the only thing that answers at all, and without it the
	274	+ // feature was simply unavailable (#752). Same priority 8, and for the
	275	+ // same reason: ahead of redirect_canonical().
	276	+ add_action('template_redirect', [$this, 'maybe_serve_sitemap'], 8);
	277	+
270	278	// Take WordPress core's own sitemap offline while ThinkRank's is active.
271	279	// Two sitemap indexes on one site is a crawl conflict: core keeps
272	280	// /wp-sitemap.xml served and injects its own "Sitemap:" line into
273	281	// robots.txt (WP_Sitemaps::add_robots, priority 0). Until now that line
		@@ -369,8 +377,16 @@
369	377	}
370	378
371	379	// Initialize Global SEO Schema Output and store reference
372	380	$this->global_seo_schema = new Global_SEO_Schema_Output();
	381	+ // Let schema reuse the description this class already resolves, so the
	382	+ // JSON-LD and the meta/og/twitter tags cannot disagree about what the
	383	+ // page is (#766). Passed as a callback rather than a value: schema is
	384	+ // built during wp_head, by which point the request context this
	385	+ // resolution depends on is set, and it must not be captured earlier.
	386	+ $this->global_seo_schema->set_description_resolver(
	387	+ fn (): string => (string) $this->get_meta_description()
	388	+ );
373	389	$this->global_seo_schema->init();
374	390	}
375	391
376	392	/**
		@@ -1783,12 +1799,16 @@
1783	1799	// Get image dimensions and alt text
1784	1800	$image_id = get_post_thumbnail_id($this->current_post_id);
1785	1801	$image_meta = wp_get_attachment_metadata($image_id);
1786	1802	if ($image_meta) {
	1803	+ // The `large` file being published, not the original it
	1804	+ // was generated from: the metadata's own width and height
	1805	+ // describe an image this tag does not point at (#847).
	1806	+ $image_file = \ThinkRank\SEO\Attachment_Lookup::describe((int) $image_id, (string) $image_url);
1787	1807	// SVGs (and other vector uploads) report 0x0 — emitting
1788	1808	// those as og:image dimensions is invalid, so skip them.
1789		- $og_width = isset($image_meta['width']) ? (int) $image_meta['width'] : 0;
1790		- $og_height = isset($image_meta['height']) ? (int) $image_meta['height'] : 0;
	1809	+ $og_width = $image_file['width'];
	1810	+ $og_height = $image_file['height'];
1791	1811	if ($og_width > 0 && $og_height > 0) {
1792	1812	echo "<meta property=\"og:image:width\" content=\"" . esc_attr($og_width) . "\" />\n";
1793	1813	echo "<meta property=\"og:image:height\" content=\"" . esc_attr($og_height) . "\" />\n";
1794	1814	}
		@@ -1793,9 +1813,9 @@
1793	1813	echo "<meta property=\"og:image:height\" content=\"" . esc_attr($og_height) . "\" />\n";
1794	1814	}
1795	1815	// Derive the real mime type instead of hardcoding image/jpeg,
1796	1816	// which mislabels PNG/WebP featured images.
1797		- $image_mime = get_post_mime_type($image_id);
	1817	+ $image_mime = $image_file['type'];
1798	1818	if ($image_mime) {
1799	1819	echo "<meta property=\"og:image:type\" content=\"" . esc_attr($image_mime) . "\" />\n";
1800	1820	}
1801	1821	}
		@@ -2946,10 +2966,22 @@
2946	2966	$context_type,
2947	2967	$context_id
2948	2968	);
2949	2969
	2970	+ // A deployed node is a snapshot from Deploy time and outranks
	2971	+ // the automatic node, so page and article types would publish
	2972	+ // a frozen excerpt instead of the description the head
	2973	+ // resolves. Give them the live one, as the automatic node has.
	2974	+ $context_post = get_post($context_id);
	2975	+
2950	2976	foreach ($page_specific_schemas as $schema_type => $schema_info) {
2951		- Schema_Graph::instance()->add_primary($schema_info['data'], (string) $schema_type, 'schema_manager');
	2977	+ $node = $schema_info['data'];
	2978	+
	2979	+ if ($this->global_seo_schema && $context_post instanceof \WP_Post) {
	2980	+ $node = $this->global_seo_schema->refresh_deployed_description($node, (string) $schema_type, $context_post);
	2981	+ }
	2982	+
	2983	+ Schema_Graph::instance()->add_primary($node, (string) $schema_type, 'schema_manager');
2952	2984	}
2953	2985	$has_schema_manager_output = true;
2954	2986	}
2955	2987	}
		@@ -3007,8 +3039,12 @@
3007	3039	'url' => home_url('/'),
3008	3040	];
3009	3041
3010	3042	$description = !empty($settings['site_description']) ? $settings['site_description'] : get_bloginfo('description');
	3043	+ // The tagline is stored esc_html()'d by sanitize_option(), so a site
	3044	+ // called "Fish & Chips" published `&amp;` literally in its WebSite
	3045	+ // node; nothing decodes JSON-LD downstream.
	3046	+ $description = \ThinkRank\Core\Seo_Text::normalize_schema_text((string) $description);
3011	3047	if (!empty($description)) {
3012	3048	$schema['description'] = $description;
3013	3049	}
3014	3050
		@@ -3526,8 +3562,102 @@
3526	3562	* @since 1.32.0
3527	3563	*
3528	3564	* @return void
3529	3565	*/
	3566	+ /**
	3567	+ * Serve a ThinkRank sitemap document for this request, when it is one.
	3568	+ *
	3569	+ * Only acts in dynamic delivery mode. In static mode a real file exists and
	3570	+ * the web server returns it without WordPress ever loading, so answering
	3571	+ * here as well would mean two sources for the same bytes.
	3572	+ *
	3573	+ * @since 2.9.0
	3574	+ *
	3575	+ * @return void
	3576	+ */
	3577	+ public function maybe_serve_sitemap(): void {
	3578	+ $filename = $this->requested_sitemap_filename();
	3579	+ if ('' === $filename) {
	3580	+ return;
	3581	+ }
	3582	+
	3583	+ try {
	3584	+ // Read-only instance: passing false keeps it from registering a
	3585	+ // second copy of the auto-generation hooks.
	3586	+ $generator = new \ThinkRank\SEO\Sitemap_Generator(false);
	3587	+ $settings = $generator->get_settings('site');
	3588	+
	3589	+ if (empty($settings['enabled'])) {
	3590	+ return;
	3591	+ }
	3592	+
	3593	+ if ('dynamic' !== $generator->resolve_delivery_mode($settings)) {
	3594	+ return;
	3595	+ }
	3596	+
	3597	+ if (!$generator->publishes_document_name($filename, $settings)) {
	3598	+ return;
	3599	+ }
	3600	+
	3601	+ $xml = $generator->render_document($filename, $settings);
	3602	+ } catch (\Throwable $e) {
	3603	+ // A failed render must not replace the sitemap with a fatal. Leave
	3604	+ // the request alone so WordPress answers as it otherwise would.
	3605	+ return;
	3606	+ }
	3607	+
	3608	+ if (!is_string($xml) || '' === trim($xml)) {
	3609	+ return;
	3610	+ }
	3611	+
	3612	+ status_header(200);
	3613	+ header('Content-Type: application/xml; charset=UTF-8');
	3614	+ header('X-Robots-Tag: noindex, follow', true);
	3615	+
	3616	+ // Built XML, escaped by the builders as they assemble it; escaping the
	3617	+ // document here would corrupt it.
	3618	+ echo $xml; // phpcs:ignore WordPress.Security.EscapeOutput.OutputNotEscaped
	3619	+ exit;
	3620	+ }
	3621	+
	3622	+ /**
	3623	+ * The sitemap file name this request is asking for, if it looks like one.
	3624	+ *
	3625	+ * Deliberately a cheap shape test. Whether the site actually publishes the
	3626	+ * name is settled by the caller against the generator, so that a request
	3627	+ * for someone else's sitemap is never answered here.
	3628	+ *
	3629	+ * @since 2.9.0
	3630	+ *
	3631	+ * @return string File name, or '' when this is not a sitemap request.
	3632	+ */
	3633	+ private function requested_sitemap_filename(): string {
	3634	+ if (empty($_SERVER['REQUEST_URI'])) {
	3635	+ return '';
	3636	+ }
	3637	+
	3638	+ $path = wp_parse_url(sanitize_text_field(wp_unslash($_SERVER['REQUEST_URI'])), PHP_URL_PATH);
	3639	+ if (!is_string($path) || '' === $path) {
	3640	+ return '';
	3641	+ }
	3642	+
	3643	+ // Strip the install's home path so subdirectory installs match too.
	3644	+ $home_path = (string) wp_parse_url(home_url('/'), PHP_URL_PATH);
	3645	+ if ('' !== $home_path && '/' !== $home_path && 0 === strpos($path, $home_path)) {
	3646	+ $path = substr($path, strlen($home_path));
	3647	+ }
	3648	+
	3649	+ $candidate = strtolower(trim($path, '/'));
	3650	+
	3651	+ // One path segment ending in .xml. Anything nested is not a file we
	3652	+ // publish to the web root.
	3653	+ if ('' === $candidate || strpos($candidate, '/') !== false) {
	3654	+ return '';
	3655	+ }
	3656	+
	3657	+ return substr($candidate, -4) === '.xml' ? $candidate : '';
	3658	+ }
	3659	+
3530	3660	public function maybe_serve_llms_txt(): void {
3531	3661	if (!$this->is_llms_txt_request()) {
3532	3662	return;
3533	3663	}
		@@ -3728,11 +3858,22 @@
3728	3858	// second copy of the save_post/term auto-generation hooks.
3729	3859	$generator = new \ThinkRank\SEO\Sitemap_Generator(false);
3730	3860	$settings = $generator->get_settings('site');
3731	3861
	3862	+ // "Can ThinkRank actually answer its sitemap URL right now?" In
	3863	+ // static mode that means the file is on disk; in dynamic mode
	3864	+ // maybe_serve_sitemap() answers it, so there is nothing to look
	3865	+ // for. Keeping the file test as the only answer would have left
	3866	+ // core's sitemap in place on every dynamic site, which is the
	3867	+ // crawl conflict this suppression exists to prevent (#752).
	3868	+ // The #346 behaviour is unchanged: a static site with nothing
	3869	+ // published still falls through to core rather than 404ing.
	3870	+ $can_serve = 'dynamic' === $generator->resolve_delivery_mode($settings)
	3871	+ || $generator->primary_sitemap_file_exists($settings);
	3872	+
3732	3873	$this->thinkrank_sitemap_enabled = !empty($settings['enabled'])
3733	3874	&& !$this->publishes_at_core_sitemap_url($settings)
3734		- && $generator->primary_sitemap_file_exists($settings);
	3875	+ && $can_serve;
3735	3876
3736	3877	if ($this->thinkrank_sitemap_enabled) {
3737	3878	$this->thinkrank_sitemap_url = $generator->get_primary_sitemap_url($settings);
3738	3879	}