PluginProbe
ThinkRank AI SEO – AI SEO Plugin for WordPress: Schema, XML Sitemaps, Meta Tags, Search Console & Local SEO / 2.13.0
ThinkRank AI SEO – AI SEO Plugin for WordPress: Schema, XML Sitemaps, Meta Tags, Search Console & Local SEO v2.13.0
2.13.0 2.12.0 2.11.0 2.10.0 2.9.0 2.8.0 2.7.0 2.6.0 2.5.0 2.4.0 2.3.0 2.2.0 2.1.1 2.1.0 2.0.2 2.0.1 2.0.0 1.32.0 1.31.0 1.30.0 1.29.0 1.28.0 1.27.0 1.26.0 1.25.0 All 54 releases
← All changes | includes/frontend/class-seo-manager.php +1707 -148 1.30.0 → 2.13.0 View file →
@@ -41,8 +41,16 @@
41 41 */
42 42 private array $current_metadata = [];
43 43
44 44 /**
45 + * Term ID of the archive being rendered, when the request is a term archive.
46 + *
47 + * @since 2.0.1
48 + * @var int|null
49 + */
50 + private ?int $current_term_id = null;
51 +
52 + /**
45 53 * Site Identity Manager instance
46 54 *
47 55 * @var \ThinkRank\SEO\Site_Identity_Manager|null
48 56 */
@@ -48,8 +56,29 @@
48 56 */
49 57 private ?\ThinkRank\SEO\Site_Identity_Manager $site_identity_manager = null;
50 58
51 59 /**
60 + * Resolved icon URLs, keyed by "<md5 of configured URL>:<size>".
61 + *
62 + * wp_site_icon() renders four tags per page and each one resolves the same
63 + * setting, so without this the lookup is four rounds of
64 + * attachment_url_to_postid() — an uncached postmeta query apiece — for one
65 + * answer. Loaded from, and persisted to, a transient: this filter runs in
66 + * wp_head on every FRONT-END request, and the mapping only changes when the
67 + * icon setting does.
68 + *
69 + * @var array<string, string>|null Null until loaded.
70 + */
71 + private ?array $icon_urls = null;
72 +
73 + /**
74 + * Whether $icon_urls gained an entry that is not in the transient yet.
75 + *
76 + * @var bool
77 + */
78 + private bool $icon_urls_dirty = false;
79 +
80 + /**
52 81 * Social Meta Manager instance
53 82 *
54 83 * @var \ThinkRank\SEO\Social_Meta_Manager|null
55 84 */
@@ -83,8 +112,16 @@
83 112 */
84 113 private ?\ThinkRank\SEO\Image_SEO_Manager $image_seo_manager = null;
85 114
86 115 /**
116 + * External Links Manager instance
117 + *
118 + * @since 2.5.0
119 + * @var \ThinkRank\SEO\External_Links_Manager|null
120 + */
121 + private ?\ThinkRank\SEO\External_Links_Manager $external_links_manager = null;
122 +
123 + /**
87 124 * Current page context
88 125 *
89 126 * @var string
90 127 */
@@ -90,8 +127,39 @@
90 127 */
91 128 private string $current_context = 'site';
92 129
93 130 /**
131 + * Whether the opening "Search Engine Optimization by ThinkRank" comment has
132 + * already been printed for this request.
133 + *
134 + * Shared across the request rather than kept as a local `static` inside the
135 + * emitter, because the closing comment is printed from a different method
136 + * (and the opening one can also come from Author_Archives_Manager). Without
137 + * that, output_closing_comment() decided on its own always-false local
138 + * static and emitted an orphan `<!-- /ThinkRank SEO -->` on every page whose
139 + * meta description was empty.
140 + *
141 + * @since 2.0.1
142 + * @var bool
143 + */
144 + private static bool $opening_comment_output = false;
145 +
146 + /**
147 + * Memoised "should core's sitemap be disabled" flag. Null until resolved.
148 + *
149 + * @var bool|null
150 + */
151 + private ?bool $thinkrank_sitemap_enabled = null;
152 +
153 + /**
154 + * Memoised public URL of the sitemap ThinkRank publishes. Empty until
155 + * should_disable_core_sitemap() has resolved, and while it resolves false.
156 + *
157 + * @var string
158 + */
159 + private string $thinkrank_sitemap_url = '';
160 +
161 + /**
94 162 * Initialize SEO manager
95 163 *
96 164 * @return void
97 165 */
@@ -107,17 +175,22 @@
107 175
108 176 // Initialize Global SEO Schema Output
109 177 $this->initialize_global_seo_schema();
110 178
111 - // Initialize Google Analytics Tracking Manager
112 - $this->initialize_google_analytics_tracking();
113 -
114 179 // Initialize Image SEO Manager
115 180 $this->initialize_image_seo_manager();
116 181
182 + // Initialize External Links Manager (rel=nofollow / target=_blank)
183 + $this->initialize_external_links_manager();
184 +
117 185 // Initialize current post and context data first
118 186 add_action('wp', [$this, 'initialize_current_context']);
119 187
188 + // ...then let it be corrected if the request turns into a 404 later.
189 + // Late, so every set_404() on this hook has already run; still well
190 + // before wp_head, which the template fires.
191 + add_action('template_redirect', [$this, 'recheck_404_context'], 999);
192 +
120 193 // Use HIGH PRIORITY hooks to override other SEO plugins
121 194 // Priority 1-5 ensures ThinkRank runs before other SEO plugins
122 195
123 196 // Override WordPress title with HIGH priority
@@ -148,11 +221,27 @@
148 221 // the page would emit two <link rel="canonical"> tags on singular views.
149 222 remove_action('wp_head', 'rel_canonical');
150 223 add_action('wp_head', [$this, 'output_canonical_url'], 6);
151 224
225 + // Silence the Bricks theme's own SEO + Open Graph output so a Bricks
226 + // site doesn't ship two of every tag. Bricks is a THEME, so it loads
227 + // after plugins: at this point BRICKS_VERSION is not yet defined and a
228 + // `defined()` guard here would always be false. Registering the filters
229 + // unconditionally is correct and free — the hooks only ever fire from
230 + // inside Bricks itself (#257). This mirrors the core rel_canonical and
231 + // wp_robots removals above: one producer per tag.
232 + add_filter('bricks/frontend/disable_seo', '__return_true');
233 + add_filter('bricks/frontend/disable_opengraph', '__return_true');
234 +
152 235 // Add Site Identity specific outputs
153 236 add_action('wp_head', [$this, 'output_site_schema_markup'], 7);
154 237 add_action('wp_head', [$this, 'output_breadcrumb_schema'], 8);
238 + // Late enough that Global_SEO_Schema_Output (priority 15) has registered.
239 + add_action('wp_head', [$this, 'output_schema_graph'], 20);
240 + // Tell the graph it has a renderer, so a body producer asking whether
241 + // its FAQ was absorbed can trigger collection itself when a block theme
242 + // renders the post content ahead of wp_head.
243 + Schema_Graph::instance()->schedule_render();
155 244
156 245 // Add closing comment (runs last)
157 246 add_action('wp_head', [$this, 'output_closing_comment'], 99);
158 247
@@ -169,8 +258,46 @@
169 258
170 259 // Add robots.txt filter hook
171 260 add_filter('robots_txt', [$this, 'filter_robots_txt'], 10, 2);
172 261
262 + // Serve /llms.txt from PHP when the request reaches WordPress. A
263 + // published llms.txt is a physical file, so the web server normally
264 + // answers it — with `text/plain` and no charset, which renders UTF-8
265 + // content as mojibake. This route (plus the .htaccess block written by
266 + // LLMs_Txt_Manager for the static file) guarantees an explicit UTF-8
267 + // charset. Priority 8 keeps it ahead of redirect_canonical().
268 + add_action('template_redirect', [$this, 'maybe_serve_llms_txt'], 8);
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', [$this, 'maybe_serve_sitemap'], 8);
277 +
278 + // Take WordPress core's own sitemap offline while ThinkRank's is active.
279 + // Two sitemap indexes on one site is a crawl conflict: core keeps
280 + // /wp-sitemap.xml served and injects its own "Sitemap:" line into
281 + // robots.txt (WP_Sitemaps::add_robots, priority 0). Until now that line
282 + // only disappeared as a side effect of filter_robots_txt() replacing the
283 + // whole filter output, which does not happen when robots.txt management
284 + // is off, when Site Identity is disabled, or when another SEO plugin
285 + // claims the filter first — and it never took /wp-sitemap.xml itself
286 + // offline, so crawlers could still find and follow the duplicate index.
287 + add_filter('wp_sitemaps_enabled', [$this, 'filter_wp_sitemaps_enabled']);
288 +
289 + // …and point the URLs core owned at our sitemap, rather than letting
290 + // them dead-end. Disabling core's sitemap does not unhook the two core
291 + // paths that route /sitemap.xml: WP_Rewrite::rewrite_rules() adds the
292 + // `sitemap\.xml` rule unconditionally, and redirect_canonical() 301s any
293 + // request carrying the `sitemap` query var to /wp-sitemap.xml without
294 + // consulting wp_sitemaps_enabled — which then 404s. Runs before both
295 + // redirect_canonical() and WP_Sitemaps::render_sitemaps() (priority 10),
296 + // and after a Pro redirect rule (priority 1) so a user-defined redirect
297 + // for these URLs still wins.
298 + add_action('template_redirect', [$this, 'redirect_core_sitemap_requests'], 9);
299 +
173 300 // Keep an existing physical robots.txt in step with WordPress's
174 301 // "Discourage search engines" toggle (blog_public). A physical file
175 302 // bypasses core's robots_txt filter, so flipping blog_public after the
176 303 // file was written would otherwise leave the previous crawl policy served
@@ -180,8 +307,17 @@
180 307 // Serve the Site Identity favicon through core's site-icon pipeline so
181 308 // wp_site_icon() outputs it on the front-end (and previews pick it up)
182 309 add_filter('get_site_icon_url', [$this, 'filter_site_icon_url'], 10, 2);
183 310
311 + // Rewrite outbound anchors (rel=nofollow / target=_blank). Runs at
312 + // the very end of the_content, after core's formatting AND after the
313 + // image filter above, so it sees the markup the visitor will get. The
314 + // stored post_content is never touched — turning the settings off
315 + // restores the author's markup exactly.
316 + add_filter('the_content', [$this, 'filter_external_links'], 100000);
317 + add_filter('the_excerpt', [$this, 'filter_external_links'], 100000);
318 + add_filter('widget_text_content', [$this, 'filter_external_links'], 100000);
319 +
184 320 // Process image SEO in content
185 321 add_filter('the_content', [$this, 'filter_content_images'], 99999);
186 322 add_filter('post_thumbnail_html', [$this, 'filter_content_images'], 11, 2);
187 323 add_filter('woocommerce_single_product_image_thumbnail_html', [$this, 'filter_content_images'], 11);
@@ -241,39 +377,73 @@
241 377 }
242 378
243 379 // Initialize Global SEO Schema Output and store reference
244 380 $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 + );
245 389 $this->global_seo_schema->init();
246 390 }
247 391
248 392 /**
249 - * Initialize Google Analytics Tracking Manager
393 + * Initialize Image SEO Manager
250 394 *
251 395 * @return void
252 396 */
253 - private function initialize_google_analytics_tracking(): void {
254 - if (!class_exists('ThinkRank\\Frontend\\Google_Analytics_Tracking_Manager')) {
255 - require_once THINKRANK_PLUGIN_DIR . 'includes/frontend/class-google-analytics-tracking-manager.php';
397 + private function initialize_image_seo_manager(): void {
398 + if (!class_exists('ThinkRank\\SEO\\Image_SEO_Manager')) {
399 + require_once THINKRANK_PLUGIN_DIR . 'includes/seo/class-image-seo-manager.php';
256 400 }
257 401
258 - // Initialize Google Analytics Tracking Manager
259 - new \ThinkRank\Frontend\Google_Analytics_Tracking_Manager();
402 + $this->image_seo_manager = new \ThinkRank\SEO\Image_SEO_Manager();
260 403 }
261 404
262 405 /**
263 - * Initialize Image SEO Manager
406 + * Initialize External Links Manager
264 407 *
408 + * @since 2.5.0
265 409 * @return void
266 410 */
267 - private function initialize_image_seo_manager(): void {
268 - if (!class_exists('ThinkRank\\SEO\\Image_SEO_Manager')) {
269 - require_once THINKRANK_PLUGIN_DIR . 'includes/seo/class-image-seo-manager.php';
411 + private function initialize_external_links_manager(): void {
412 + if (!class_exists('ThinkRank\\SEO\\External_Links_Manager')) {
413 + require_once THINKRANK_PLUGIN_DIR . 'includes/seo/class-external-links-manager.php';
270 414 }
271 415
272 - $this->image_seo_manager = new \ThinkRank\SEO\Image_SEO_Manager();
416 + $this->external_links_manager = new \ThinkRank\SEO\External_Links_Manager();
273 417 }
274 418
275 419 /**
420 + * Filter rendered content to annotate external links
421 + *
422 + * @since 2.5.0
423 + * @param mixed $content Content to filter; passed through untouched when
424 + * it is not a string.
425 + * @return mixed Filtered content.
426 + */
427 + public function filter_external_links($content) {
428 + // No return type: a filter value another plugin hands through as null
429 + // or an object belongs to whoever set it, and coercing it to '' would
430 + // silently drop their content on the floor.
431 + if (!is_string($content) || $content === '' || !$this->external_links_manager) {
432 + return $content;
433 + }
434 +
435 + // Feeds carry the same markup to a reader we do not control; leave
436 + // them as authored rather than annotating for a context that has no
437 + // browser tab to open.
438 + if (is_feed()) {
439 + return $content;
440 + }
441 +
442 + return $this->external_links_manager->process_content($content);
443 + }
444 +
445 + /**
276 446 * Filter content to inject image SEO attributes
277 447 *
278 448 * @since 1.0.0
279 449 * @param string $content Content to filter
@@ -325,18 +495,74 @@
325 495 $this->current_metadata = $this->get_post_seo_metadata($post_id);
326 496 }
327 497 }
328 498
499 + // Term archives. Category, tag and custom-taxonomy pages store their SEO
500 + // title and description as term meta — written by the term UI, by the
501 + // abilities API and by the Yoast/RankMath/AIOSEO/SEOPress importer — but
502 + // nothing here ever read them, so the whole title/description cascade
503 + // fell through to the theme default and no description tag was printed
504 + // at all. Term robots was fixed for the same reason in 1.31.0 (#290);
505 + // this is the title and description half (#386).
506 + if (is_category() || is_tag() || is_tax()) {
507 + $queried = get_queried_object();
508 + if ($queried instanceof \WP_Term) {
509 + $this->current_term_id = $queried->term_id;
510 + $this->current_metadata = $this->get_term_seo_metadata($queried->term_id);
511 + }
512 + }
513 +
329 514 // Load site identity data
330 515 $this->load_site_identity_data();
331 516 }
332 517
333 518 /**
519 + * Drop the request's post identity once it has become a 404.
520 + *
521 + * `initialize_current_context()` runs on `wp`, but a request can be turned
522 + * into a 404 after that: `set_404()` on `template_redirect` is the ordinary
523 + * way to refuse a URL that did resolve to a real post, and both core and
524 + * plugins do it — ThinkRank Pro's Markdown for AI refuses an ineligible
525 + * `.md` URL that way. The snapshot still said `post`/`page` and still held
526 + * the post id and its metadata, so the error page shipped that post's meta
527 + * description, focus keywords and — where the social emitters got that far
528 + * — its og:description and twitter:description, all of which a request that
529 + * was a 404 from the start never prints (#655).
530 + *
531 + * Clearing the snapshot rather than special-casing each emitter is what
532 + * makes every consumer agree, including the ones that read
533 + * `$current_metadata` without ever asking what the context is.
534 + *
535 + * @since 2.3.1
536 + *
537 + * @return void
538 + */
539 + public function recheck_404_context(): void {
540 + if (!is_404() || '404' === $this->current_context) {
541 + return;
542 + }
543 +
544 + $this->current_context = '404';
545 + $this->current_post_id = null;
546 + $this->current_term_id = null;
547 + $this->current_metadata = [];
548 + }
549 +
550 + /**
334 551 * Detect current page context
335 552 *
336 553 * @return string Current context type
337 554 */
338 555 private function detect_current_context(): string {
556 + // 404 first: a not-found request matches none of the branches below and
557 + // used to fall through to 'site', which handed crawlers the homepage's
558 + // social identity for an error page. It gets its own context so the
559 + // social layer can skip it, matching get_non_singular_canonical_url(),
560 + // which already suppresses the canonical for 404 and search.
561 + if (is_404()) {
562 + return '404';
563 + }
564 +
339 565 if (is_home() || is_front_page()) {
340 566 return 'homepage';
341 567 } elseif (is_single()) {
342 568 return 'post';
@@ -392,8 +618,37 @@
392 618 ];
393 619 }
394 620
395 621 /**
622 + * Get SEO metadata for a term.
623 + *
624 + * Mirrors get_post_seo_metadata(): the stored values may carry variable
625 + * tags, so they are resolved against the term's own values. Focus keyword
626 + * and score have no term equivalent on the frontend and stay empty.
627 + *
628 + * @since 2.0.1
629 + *
630 + * @param int $term_id Term ID.
631 + * @return array SEO metadata.
632 + */
633 + private function get_term_seo_metadata(int $term_id): array {
634 + $title = get_term_meta($term_id, '_thinkrank_seo_title', true);
635 + $description = get_term_meta($term_id, '_thinkrank_meta_description', true);
636 +
637 + return [
638 + 'title' => $title
639 + ? \ThinkRank\SEO\Pattern_Resolver::resolve_term_value((string) $title, $term_id)
640 + : '',
641 + 'description' => $description
642 + ? \ThinkRank\SEO\Pattern_Resolver::resolve_term_value((string) $description, $term_id)
643 + : '',
644 + 'focus_keyword' => '',
645 + 'focus_keywords' => [],
646 + 'seo_score' => '',
647 + ];
648 + }
649 +
650 + /**
396 651 * Resolve the effective SEO title for the current request.
397 652 *
398 653 * Same priority chain as override_document_title() — post-specific
399 654 * ThinkRank metadata (resolved _thinkrank_seo_title) > Global SEO
@@ -418,17 +673,23 @@
418 673 * @param string $title Original title
419 674 * @return string Modified title
420 675 */
421 676 public function override_document_title($title): string {
677 + // A content type with metas switched off keeps whatever title the theme
678 + // and WordPress produce (#660).
679 + if (!$this->metas_enabled()) {
680 + return $title;
681 + }
682 +
422 683 // First priority: Post-specific ThinkRank metadata
423 684 if ($this->has_thinkrank_metadata() && !empty($this->current_metadata['title'])) {
424 - return $this->current_metadata['title'];
685 + return self::with_page_suffix($this->current_metadata['title']);
425 686 }
426 687
427 688 // Second priority: Global SEO templates, Third priority: Site Identity templates
428 689 $generated_title = $this->generate_context_title();
429 690 if ($generated_title) {
430 - return $generated_title;
691 + return self::with_page_suffix($generated_title);
431 692 }
432 693
433 694 return $title;
434 695 }
@@ -433,8 +694,117 @@
433 694 return $title;
434 695 }
435 696
436 697 /**
698 + * Append a page indicator to a title on page 2 and beyond.
699 + *
700 + * This filter short-circuits pre_get_document_title at priority 1, which
701 + * drops the " – Page 2" core would otherwise add — so every page of an
702 + * archive, and every part of a multi-page post, shared one <title> (#397).
703 + * The templates have no %page% token, so the suffix is added here rather
704 + * than asking every site to edit its title format.
705 + *
706 + * @since 2.0.1
707 + *
708 + * @param string $title Resolved title.
709 + * @return string Title with the page indicator, when there is one.
710 + */
711 + /**
712 + * The archive's subject, without the label WordPress prefixes it with.
713 + *
714 + * `get_the_archive_title()` returns "Month: September 2026", "Archives:
715 + * Recipes", "Category: Uncategorized" — the label is core's, aimed at an
716 + * archive heading on the page, and it reads badly in a browser tab, an
717 + * og:title or a search result. Category, tag and author contexts already
718 + * avoid it by using the raw name; the generic archive context did not, so
719 + * date, custom-post-type and custom-taxonomy archives carried it (#640).
720 + *
721 + * Removed through core's own `get_the_archive_title_prefix` filter rather
722 + * than by matching the prefix text, because that text is translated and
723 + * differs per archive type — a string comparison would work in English and
724 + * silently stop working everywhere else.
725 + *
726 + * A site that wants a prefix can put one in its title template, where it is
727 + * visible and editable, instead of inheriting one it cannot see.
728 + *
729 + * @since 2.7.0
730 + *
731 + * @return string Archive subject, with markup and the core prefix removed.
732 + */
733 + private static function archive_subject(): string {
734 + $drop_prefix = static function (): string {
735 + return '';
736 + };
737 +
738 + add_filter('get_the_archive_title_prefix', $drop_prefix, 99);
739 +
740 + $title = (string) get_the_archive_title();
741 +
742 + remove_filter('get_the_archive_title_prefix', $drop_prefix, 99);
743 +
744 + // The <span> core wraps the subject in survives the prefix filter.
745 + return trim(wp_strip_all_tags($title));
746 + }
747 +
748 + /**
749 + * Remove HTML from a title that is about to be emitted.
750 + *
751 + * A title carrying markup is broken twice over, in two different ways, and
752 + * both were reaching real pages: inside `<title>` the tags render literally,
753 + * because that element is RCDATA and never parses them; inside `og:title`
754 + * and `twitter:title` they are attribute-escaped, so the reader sees
755 + * `&lt;em&gt;` as visible text (#640).
756 + *
757 + * Applied at the point of emission rather than at each source, so it covers
758 + * every branch that can produce a title — post meta, Global SEO templates,
759 + * Site Identity templates — without each having to remember.
760 + *
761 + * Unconditional rather than a setting: there is no title for which markup is
762 + * the correct output. The filter is the escape hatch for anyone who
763 + * disagrees, and lets a site keep entities it deliberately encoded.
764 + *
765 + * @since 2.7.0
766 + *
767 + * @param string $title Title about to be emitted.
768 + * @return string Title with any markup removed.
769 + */
770 + public static function strip_title_tags(string $title): string {
771 + /**
772 + * Filter whether HTML is stripped from generated titles.
773 + *
774 + * @since 2.7.0
775 + *
776 + * @param bool $strip Whether to strip. Default true.
777 + * @param string $title The title being emitted.
778 + */
779 + if (!apply_filters('thinkrank_strip_title_tags', true, $title)) {
780 + return $title;
781 + }
782 +
783 + return trim(wp_strip_all_tags($title));
784 + }
785 +
786 + public static function with_page_suffix(string $title): string {
787 + $title = self::strip_title_tags($title);
788 +
789 + $page = self::current_page_number();
790 +
791 + if ($page <= 1 || '' === $title) {
792 + return $title;
793 + }
794 +
795 + $separator = class_exists('\ThinkRank\SEO\Site_Identity_Manager')
796 + ? \ThinkRank\SEO\Site_Identity_Manager::get_active_separator_symbol()
797 + : '|';
798 +
799 + return $title . ' ' . $separator . ' ' . sprintf(
800 + /* translators: %d: page number. */
801 + __('Page %d', 'thinkrank'),
802 + $page
803 + );
804 + }
805 +
806 + /**
437 807 * Override WordPress wp_title (HIGH PRIORITY)
438 808 * Priority: Post-specific metadata > Global SEO templates > Site Identity templates
439 809 *
440 810 * @param string $title Original title
@@ -441,18 +811,24 @@
441 811 * @param string $sep Title separator
442 812 * @return string Modified title
443 813 */
444 814 public function override_wp_title(string $title, string $sep = ''): string {
815 + if (!$this->metas_enabled()) {
816 + return $title;
817 + }
818 +
445 819 // First priority: Post-specific ThinkRank metadata
446 820 if ($this->has_thinkrank_metadata() && !empty($this->current_metadata['title'])) {
447 821 $site_name = get_bloginfo('name');
448 - return $this->current_metadata['title'] . ($sep ? " $sep " : ' | ') . $site_name;
822 + return self::with_page_suffix(
823 + $this->current_metadata['title'] . ($sep ? " $sep " : ' | ') . $site_name
824 + );
449 825 }
450 826
451 827 // Second priority: Global SEO templates, Third priority: Site Identity templates
452 828 $generated_title = $this->generate_context_title();
453 829 if ($generated_title) {
454 - return $generated_title;
830 + return self::with_page_suffix($generated_title);
455 831 }
456 832
457 833 return $title;
458 834 }
@@ -457,8 +833,25 @@
457 833 return $title;
458 834 }
459 835
460 836 /**
837 + * Whether ThinkRank owns the title and meta description for this request.
838 + *
839 + * Metas are on site-wide by default; the per-content-type matrix can switch
840 + * them off for one content type, in which case ThinkRank stops overriding
841 + * the document title and prints no meta description (#660).
842 + *
843 + * @since 2.5.0
844 + * @return bool
845 + */
846 + private function metas_enabled(): bool {
847 + return \ThinkRank\SEO\Content_Type_Settings::is_enabled_for_current(
848 + \ThinkRank\SEO\Content_Type_Settings::FEATURE_META,
849 + true
850 + );
851 + }
852 +
853 + /**
461 854 * Output meta description (HIGH PRIORITY)
462 855 * Priority: Post-specific metadata > Global SEO templates > Site Identity templates > WordPress defaults
463 856 *
464 857 * Author archives are skipped entirely: Author_Archives_Manager owns that
@@ -473,22 +866,24 @@
473 866 if (is_author()) {
474 867 return;
475 868 }
476 869
870 + if (!$this->metas_enabled()) {
871 + return;
872 + }
873 +
477 874 $description = $this->get_meta_description();
478 875
479 876 if ($description) {
480 877 // Output main ThinkRank SEO header comment (only once)
481 - static $header_output = false;
482 - if (!$header_output) {
483 - echo "<!-- Search Engine Optimization by ThinkRank - https://thinkrank.ai/ -->\n";
484 - $header_output = true;
485 - }
878 + self::note_opening_comment();
486 879
487 880 // Ensure description is within optimal length (150-160 characters)
488 - if (strlen($description) > 160) {
489 - $description = wp_trim_words($description, 25, '...');
490 - }
881 + // Measure and cut in CHARACTERS. strlen() counts bytes, so a Thai or
882 + // CJK description tripped this limit at a third of its length, and
883 + // wp_trim_words() then cut by a unit the locale chooses — 25 words in
884 + // English, 25 characters in Thai (#687).
885 + $description = \ThinkRank\Core\Seo_Text::trim_to_length($description);
491 886
492 887 echo "<!-- ThinkRank SEO Meta Description -->\n";
493 888 echo '<meta name="description" content="' . esc_attr($description) . '" />' . "\n";
494 889 echo "<!-- /ThinkRank SEO Meta Description -->\n";
@@ -529,13 +924,41 @@
529 924
530 925 // Output generator meta tag
531 926 echo '<meta name="generator" content="ThinkRank ' . esc_attr(THINKRANK_VERSION) . '" />' . "\n";
532 927
533 - // Output viewport meta tag if not already present
534 - if (!has_action('wp_head', 'wp_site_icon') || !wp_is_mobile()) {
535 - echo '<meta name="viewport" content="width=device-width, initial-scale=1.0" />' . "\n";
928 + // No viewport tag here. The viewport is the theme's responsibility and
929 + // every modern theme ships one, so emitting our own only ever produced a
930 + // second <meta name="viewport"> in the document. The old guard could not
931 + // prevent that either: has_action() returns the registered priority
932 + // (truthy), so its first operand was always false, and !wp_is_mobile() is
933 + // true for every desktop request.
934 + echo "<!-- /ThinkRank SEO Meta Tags -->\n";
935 + }
936 +
937 + /**
938 + * Build the basic robots directive list from a set of robots flags.
939 + *
940 + * Shared by the search/404 branch of get_robots_meta_content() so those
941 + * pages resolve their directives through the same rules as everything else
942 + * rather than a hardcoded literal.
943 + *
944 + * @since 2.5.0
945 + * @param array $settings Robots flags (index/noindex/nofollow/...).
946 + * @return string[] Directives.
947 + */
948 + private static function build_robots_directives(array $settings): array {
949 + $robots = [];
950 +
951 + $robots[] = !empty($settings['noindex']) ? 'noindex' : 'index';
952 + $robots[] = !empty($settings['nofollow']) ? 'nofollow' : 'follow';
953 +
954 + foreach (['noarchive', 'noimageindex', 'nosnippet'] as $directive) {
955 + if (!empty($settings[$directive])) {
956 + $robots[] = $directive;
957 + }
536 958 }
537 - echo "<!-- /ThinkRank SEO Meta Tags -->\n";
959 +
960 + return $robots;
538 961 }
539 962
540 963 /**
541 964 * Get robots meta content based on context and settings
@@ -544,13 +967,22 @@
544 967 */
545 968 private function get_robots_meta_content(): string {
546 969 $robots = [];
547 970
548 - // 404 and search results must never be indexed, regardless of the
549 - // configured global/post-type directives. Links are still followed so
550 - // crawlers can discover the rest of the site.
971 + // 404 and search results are noindex/follow by default — the behaviour
972 + // that used to be hardcoded here. It is now settings-driven (#660): the
973 + // Content Type Matrix can give either its own robots directives, and an
974 + // install that never touched them resolves to exactly the old pair.
551 975 if (is_404() || is_search()) {
552 - $robots = apply_filters('thinkrank_robots_meta', ['noindex', 'follow']);
976 + $entity = is_404()
977 + ? \ThinkRank\SEO\Content_Type_Settings::ENTITY_404
978 + : \ThinkRank\SEO\Content_Type_Settings::ENTITY_SEARCH;
979 +
980 + $robots = self::build_robots_directives(
981 + \ThinkRank\SEO\Content_Type_Settings::resolve_robots_meta($entity)
982 + );
983 +
984 + $robots = apply_filters('thinkrank_robots_meta', $robots);
553 985 return implode(', ', array_unique($robots));
554 986 }
555 987
556 988 // 1. Get global robot meta settings (Base)
@@ -579,8 +1011,24 @@
579 1011 $current_settings = array_merge($current_settings, $global_seo_settings[$post_type]['robots_meta']);
580 1012 }
581 1013 }
582 1014
1015 + // 2b. Apply the per-entity directives for the non-singular content
1016 + // types the matrix covers — taxonomy archives plus author and date
1017 + // archives. Terms keep their own per-term override, applied further
1018 + // down so it still wins over the taxonomy-wide value (#660).
1019 + if (!is_singular()) {
1020 + $entity_key = \ThinkRank\SEO\Content_Type_Settings::current_entity_key();
1021 +
1022 + if ($entity_key !== null) {
1023 + $entity_settings = \ThinkRank\SEO\Content_Type_Settings::get_entity_settings($entity_key);
1024 +
1025 + if (!empty($entity_settings['robots_meta_enabled']) && is_array($entity_settings['robots_meta'] ?? null)) {
1026 + $current_settings = array_merge($current_settings, $entity_settings['robots_meta']);
1027 + }
1028 + }
1029 + }
1030 +
583 1031 // Determine Index/Noindex based on merged settings
584 1032 // Priority: if noindex is true, it overrides index
585 1033 if (!empty($current_settings['noindex'])) {
586 1034 $robots[] = 'noindex';
@@ -644,12 +1092,30 @@
644 1092 $robots = $this->apply_post_robots_override(get_the_ID(), $robots, $current_settings);
645 1093 }
646 1094
647 1095 // Check for archive pages (search is handled by the early return above)
648 - if (is_archive()) {
1096 + //
1097 + // is_home() is deliberately included: the blog listing is not an
1098 + // is_archive(), so page 2 of a term archive was noindex while page 2 of
1099 + // the blog listing was index — the same kind of page, treated two
1100 + // different ways, on the same site (#397).
1101 + if (is_archive() || is_home()) {
649 1102 // Allow indexing of category/tag archives but be more conservative
650 1103 if (is_paged()) {
651 - $robots = ['noindex', 'follow'];
1104 + /**
1105 + * Filter whether a paginated archive is set noindex.
1106 + *
1107 + * Rank Math and Yoast now index paginated archives with a
1108 + * self-referential canonical by default, so a site that wants
1109 + * that can have it without patching.
1110 + *
1111 + * @since 2.0.1
1112 + *
1113 + * @param bool $noindex Whether to noindex this paginated page.
1114 + */
1115 + if (apply_filters('thinkrank_noindex_paged_archives', true)) {
1116 + $robots = ['noindex', 'follow'];
1117 + }
652 1118 }
653 1119
654 1120 // Honor the global date-archive noindex toggle (written by the
655 1121 // Rank Math/Yoast settings importer). Author archives are handled
@@ -658,8 +1124,27 @@
658 1124 $robots = ['noindex', 'follow'];
659 1125 }
660 1126 }
661 1127
1128 + // 4. Term meta override for taxonomy archives (Overrides everything).
1129 + //
1130 + // Terms had no branch here at all — not a wrong key or a skipped
1131 + // conditional, the lookup simply did not exist — so a category, tag or
1132 + // custom-taxonomy archive saved with noindex still rendered the global
1133 + // default. The stored value read back correctly through the abilities
1134 + // API, which made the setting look applied when it never reached output.
1135 + //
1136 + // Deliberately placed *after* the archive block so it is a real
1137 + // override, matching how a per-post override is final for singular
1138 + // views. Running it earlier would let is_paged() overwrite a term's
1139 + // explicit directives on page 2 of its own archive.
1140 + if (is_category() || is_tag() || is_tax()) {
1141 + $queried = get_queried_object();
1142 + if ($queried instanceof \WP_Term) {
1143 + $robots = $this->apply_term_robots_override($queried->term_id, $robots, $current_settings);
1144 + }
1145 + }
1146 +
662 1147 // Apply filters for customization
663 1148 $robots = apply_filters('thinkrank_robots_meta', $robots);
664 1149
665 1150 // Remove duplicates and implode
@@ -678,20 +1163,69 @@
678 1163 * @param array $current_settings Effective robots flags (global + post type)
679 1164 * @return array Updated robots directive list
680 1165 */
681 1166 private function apply_post_robots_override(int $post_id, array $robots, array $current_settings): array {
682 - if (!(bool) get_post_meta($post_id, '_thinkrank_robots_meta_enabled', true)) {
1167 + return $this->apply_meta_robots_override(
1168 + (bool) get_post_meta($post_id, '_thinkrank_robots_meta_enabled', true),
1169 + (string) get_post_meta($post_id, '_thinkrank_robots_meta', true),
1170 + (string) get_post_meta($post_id, '_thinkrank_advanced_robots_meta', true),
1171 + $robots,
1172 + $current_settings
1173 + );
1174 + }
1175 +
1176 + /**
1177 + * Apply per-term robots overrides on top of the cascaded directives.
1178 + *
1179 + * The term-meta twin of apply_post_robots_override(). Terms store the same
1180 + * three keys with the same shapes — written by the update-term-seo ability
1181 + * and by the Rank Math / Yoast / AIOSEO / SEOPress importer — so the two
1182 + * paths share one engine rather than a second copy that can drift.
1183 + *
1184 + * @since 1.31.0
1185 + *
1186 + * @param int $term_id Term being rendered
1187 + * @param array $robots Directives accumulated so far
1188 + * @param array $current_settings Effective robots flags (global + post type)
1189 + * @return array Updated robots directive list
1190 + */
1191 + private function apply_term_robots_override(int $term_id, array $robots, array $current_settings): array {
1192 + return $this->apply_meta_robots_override(
1193 + (bool) get_term_meta($term_id, '_thinkrank_robots_meta_enabled', true),
1194 + (string) get_term_meta($term_id, '_thinkrank_robots_meta', true),
1195 + (string) get_term_meta($term_id, '_thinkrank_advanced_robots_meta', true),
1196 + $robots,
1197 + $current_settings
1198 + );
1199 + }
1200 +
1201 + /**
1202 + * Rebuild the robots directives from a stored override, whatever holds it.
1203 + *
1204 + * Kept free of get_post_meta()/get_term_meta() so posts and terms cannot
1205 + * diverge: term support was missing entirely because the only override
1206 + * logic lived behind a post-meta read.
1207 + *
1208 + * @since 1.31.0
1209 + *
1210 + * @param bool $enabled Whether the override is switched on
1211 + * @param string $raw_robots JSON robots flags
1212 + * @param string $raw_advanced JSON advanced directives
1213 + * @param array $robots Directives accumulated so far
1214 + * @param array $current_settings Effective robots flags (global + post type)
1215 + * @return array Updated robots directive list
1216 + */
1217 + private function apply_meta_robots_override(bool $enabled, string $raw_robots, string $raw_advanced, array $robots, array $current_settings): array {
1218 + if (!$enabled) {
683 1219 return $robots;
684 1220 }
685 1221
686 - $raw_robots = get_post_meta($post_id, '_thinkrank_robots_meta', true);
687 - $post_robots = is_string($raw_robots) && $raw_robots !== '' ? json_decode($raw_robots, true) : null;
1222 + $post_robots = $raw_robots !== '' ? json_decode($raw_robots, true) : null;
688 1223 if (!is_array($post_robots)) {
689 1224 return $robots;
690 1225 }
691 1226
692 - $raw_advanced = get_post_meta($post_id, '_thinkrank_advanced_robots_meta', true);
693 - $post_advanced = is_string($raw_advanced) && $raw_advanced !== '' ? json_decode($raw_advanced, true) : null;
1227 + $post_advanced = $raw_advanced !== '' ? json_decode($raw_advanced, true) : null;
694 1228
695 1229 $effective = array_merge($current_settings, array_intersect_key($post_robots, array_flip([
696 1230 'index', 'noindex', 'nofollow', 'noarchive', 'noimageindex', 'nosnippet',
697 1231 ])));
@@ -861,9 +1395,9 @@
861 1395 *
862 1396 * @param array $og_tags Open Graph tags array
863 1397 * @return void
864 1398 */
865 - private function output_social_og_tags(array $og_tags): void {
1399 + private function output_social_og_tags(array $og_tags, array $extra_images = []): void {
866 1400 // Honor the thinkrank_og_type filter here too — this "Enhanced" path is
867 1401 // the active OG emitter, so add-ons (e.g. Pro's WooCommerce module which
868 1402 // sets 'product' on product pages) must be applied to it, not only to
869 1403 // output_open_graph_tags().
@@ -907,12 +1441,62 @@
907 1441 echo '<meta property="' . esc_attr($property) . '" content="' . $this->esc_meta_value($property, $content) . '" />' . "\n";
908 1442 }
909 1443 }
910 1444
1445 + // Alternatives, after the primary and everything belonging to it.
1446 + // Order is the whole point: a consumer reads og:image tags in document
1447 + // order and treats the first as primary, and a structured property
1448 + // attaches to the most recently declared image — so each alternative's
1449 + // companions have to follow its own URL, not be grouped at the end.
1450 + self::output_extra_og_images($extra_images);
1451 +
911 1452 echo "<!-- /ThinkRank SEO Open Graph Tags -->\n";
912 1453 }
913 1454
914 1455 /**
1456 + * Emit the secondary og:image tags a page offers.
1457 + *
1458 + * Shared by the enhanced and basic emitters so both describe an
1459 + * alternative image the same way (#636).
1460 + *
1461 + * @since 2.7.0
1462 + *
1463 + * @param array $images Each with url, and width/height/type/alt where known.
1464 + * @return void
1465 + */
1466 + private static function output_extra_og_images(array $images): void {
1467 + foreach ($images as $image) {
1468 + $url = isset($image['url']) ? (string) $image['url'] : '';
1469 +
1470 + if ('' === $url) {
1471 + continue;
1472 + }
1473 +
1474 + echo '<meta property="og:image" content="' . esc_url($url) . '" />' . "\n";
1475 +
1476 + if (strpos($url, 'https://') === 0) {
1477 + echo '<meta property="og:image:secure_url" content="' . esc_url($url) . '" />' . "\n";
1478 + }
1479 +
1480 + // Only what is actually known: a dimension guessed for a remote
1481 + // image is a number a consumer lays a card out with before it has
1482 + // fetched the file.
1483 + if (!empty($image['width']) && !empty($image['height'])) {
1484 + echo '<meta property="og:image:width" content="' . esc_attr((string) $image['width']) . '" />' . "\n";
1485 + echo '<meta property="og:image:height" content="' . esc_attr((string) $image['height']) . '" />' . "\n";
1486 + }
1487 +
1488 + if (!empty($image['type'])) {
1489 + echo '<meta property="og:image:type" content="' . esc_attr((string) $image['type']) . '" />' . "\n";
1490 + }
1491 +
1492 + if (!empty($image['alt'])) {
1493 + echo '<meta property="og:image:alt" content="' . esc_attr((string) $image['alt']) . '" />' . "\n";
1494 + }
1495 + }
1496 + }
1497 +
1498 + /**
915 1499 * Output social media Twitter Card tags from Social Meta Manager
916 1500 *
917 1501 * @param array $twitter_tags Twitter Card tags array
918 1502 * @return void
@@ -977,8 +1561,16 @@
977 1561 *
978 1562 * @return void
979 1563 */
980 1564 public function output_platform_meta_tags(): void {
1565 + // Same reasoning as the Open Graph and Twitter emitters: an error page
1566 + // has no shareable identity, and passing '404' through as a social
1567 + // context asks the manager for settings that describe a page which does
1568 + // not exist. Guarding all three keeps them from disagreeing.
1569 + if ($this->current_context === '404') {
1570 + return;
1571 + }
1572 +
981 1573 // Try Social Meta Manager for platform tags
982 1574 if ($this->social_manager) {
983 1575 // Map context for Social Meta Manager (homepage -> site for site-wide settings)
984 1576 $social_context = $this->current_context === 'homepage' ? 'site' : $this->current_context;
@@ -1030,8 +1622,23 @@
1030 1622 *
1031 1623 * @return void
1032 1624 */
1033 1625 public function output_open_graph_tags(): void {
1626 + // Per-content-type Open Graph switch. 'inherit' (the default) keeps the
1627 + // site-wide Social Media setting, which the emitters below read (#660).
1628 + if (!\ThinkRank\SEO\Content_Type_Settings::is_enabled_for_current(
1629 + \ThinkRank\SEO\Content_Type_Settings::FEATURE_OPEN_GRAPH,
1630 + true
1631 + )) {
1632 + return;
1633 + }
1634 +
1635 + // An error page has no shareable identity. Emitting Open Graph here
1636 + // advertised the homepage as the og:url of a URL that does not exist.
1637 + if ($this->current_context === '404') {
1638 + return;
1639 + }
1640 +
1034 1641 // Priority 1: Try Social Meta Manager (Social Media tab settings)
1035 1642 if ($this->social_manager) {
1036 1643 // Map context for Social Meta Manager (homepage -> site for site-wide settings)
1037 1644 $social_context = $this->current_context === 'homepage' ? 'site' : $this->current_context;
@@ -1052,9 +1659,12 @@
1052 1659 // The Social Meta Manager ran, so it owns Open Graph output. If OG is
1053 1660 // toggled off, emit nothing — do NOT fall through to the basic
1054 1661 // emitter (which would re-add a full OG block despite the toggle).
1055 1662 if (!empty($social_data['og_enabled'])) {
1056 - $this->output_social_og_tags($social_data['og_tags']);
1663 + $this->output_social_og_tags(
1664 + $social_data['og_tags'],
1665 + $social_data['og_extra_images'] ?? []
1666 + );
1057 1667 }
1058 1668 return;
1059 1669 }
1060 1670
@@ -1082,8 +1692,21 @@
1082 1692 (string) get_post_meta($this->current_post_id, '_thinkrank_og_description', true),
1083 1693 $this->current_post_id
1084 1694 );
1085 1695 $og_image_override = get_post_meta($this->current_post_id, '_thinkrank_og_image', true);
1696 + } elseif ($this->current_term_id) {
1697 + // Terms carry the same social override keys — the abilities API
1698 + // writes them — so honour them here rather than letting the term's
1699 + // SEO title stand in for an explicit og:title.
1700 + $og_title_override = \ThinkRank\SEO\Pattern_Resolver::resolve_term_value(
1701 + (string) get_term_meta($this->current_term_id, '_thinkrank_og_title', true),
1702 + $this->current_term_id
1703 + );
1704 + $og_description_override = \ThinkRank\SEO\Pattern_Resolver::resolve_term_value(
1705 + (string) get_term_meta($this->current_term_id, '_thinkrank_og_description', true),
1706 + $this->current_term_id
1707 + );
1708 + $og_image_override = get_term_meta($this->current_term_id, '_thinkrank_og_image', true);
1086 1709 }
1087 1710
1088 1711 // Get title using priority system: OG override > post-specific > Global SEO > Site Identity > default
1089 1712 $title = '';
@@ -1104,10 +1727,14 @@
1104 1727 $description = $og_description_override;
1105 1728 } else {
1106 1729 $description = $this->get_meta_description();
1107 1730 }
1108 - if (!$description) {
1109 - $description = is_singular() ? wp_trim_words(get_the_excerpt(), 30) : get_bloginfo('description');
1731 + // Skipped for a protected post: core answers get_the_excerpt() with its
1732 + // "There is no excerpt because this is a protected post." placeholder,
1733 + // so this is not a leak — but publishing that sentence as the social
1734 + // description is worse than publishing none (#363).
1735 + if (!$description && !$this->is_content_password_protected()) {
1736 + $description = is_singular() ? \ThinkRank\Core\Seo_Text::trim_words(get_the_excerpt(), 30) : get_bloginfo('description');
1110 1737 }
1111 1738
1112 1739 $url = is_singular() ? get_permalink() : home_url();
1113 1740 $site_name = $this->site_identity_data && !empty($this->site_identity_data['identity']['site_name'])
@@ -1135,11 +1762,11 @@
1135 1762 $og_type = apply_filters('thinkrank_og_type', $og_type);
1136 1763
1137 1764 echo "<!-- ThinkRank SEO Open Graph Meta Tags -->\n";
1138 1765 echo "<meta property=\"og:type\" content=\"" . esc_attr($og_type) . "\" />\n";
1139 - echo "<meta property=\"og:title\" content=\"" . esc_attr($title) . "\" />\n";
1766 + echo "<meta property=\"og:title\" content=\"" . esc_attr(self::strip_title_tags($title)) . "\" />\n";
1140 1767 echo "<meta property=\"og:description\" content=\"" . esc_attr($description) . "\" />\n";
1141 - echo "<meta property=\"og:url\" content=\"" . esc_url($url) . "\" />\n";
1768 + echo "<meta property=\"og:url\" content=\"" . esc_url(\ThinkRank\SEO\Url_Scheme::apply($url)) . "\" />\n";
1142 1769 echo "<meta property=\"og:site_name\" content=\"" . esc_attr($site_name) . "\" />\n";
1143 1770 /**
1144 1771 * Filter the og:locale value.
1145 1772 *
@@ -1156,14 +1783,17 @@
1156 1783 $og_locale = (string) apply_filters('thinkrank_og_locale', get_locale());
1157 1784 echo "<meta property=\"og:locale\" content=\"" . esc_attr($og_locale) . "\" />\n";
1158 1785
1159 1786 // Add OG image — per-post override > featured image
1787 + $primary_og_image = '';
1160 1788 if (is_singular() && $this->current_post_id) {
1161 1789 if (!empty($og_image_override)) {
1790 + $primary_og_image = (string) $og_image_override;
1162 1791 echo "<meta property=\"og:image\" content=\"" . esc_url($og_image_override) . "\" />\n";
1163 1792 echo "<meta property=\"og:image:secure_url\" content=\"" . esc_url($og_image_override) . "\" />\n";
1164 1793 } elseif (has_post_thumbnail($this->current_post_id)) {
1165 1794 $image_url = get_the_post_thumbnail_url($this->current_post_id, 'large');
1795 + $primary_og_image = (string) $image_url;
1166 1796 echo "<meta property=\"og:image\" content=\"" . esc_url($image_url) . "\" />\n";
1167 1797 echo "<meta property=\"og:image:secure_url\" content=\"" . esc_url($image_url) . "\" />\n";
1168 1798
1169 1799 // Get image dimensions and alt text
@@ -1169,12 +1799,16 @@
1169 1799 // Get image dimensions and alt text
1170 1800 $image_id = get_post_thumbnail_id($this->current_post_id);
1171 1801 $image_meta = wp_get_attachment_metadata($image_id);
1172 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);
1173 1807 // SVGs (and other vector uploads) report 0x0 — emitting
1174 1808 // those as og:image dimensions is invalid, so skip them.
1175 - $og_width = isset($image_meta['width']) ? (int) $image_meta['width'] : 0;
1176 - $og_height = isset($image_meta['height']) ? (int) $image_meta['height'] : 0;
1809 + $og_width = $image_file['width'];
1810 + $og_height = $image_file['height'];
1177 1811 if ($og_width > 0 && $og_height > 0) {
1178 1812 echo "<meta property=\"og:image:width\" content=\"" . esc_attr($og_width) . "\" />\n";
1179 1813 echo "<meta property=\"og:image:height\" content=\"" . esc_attr($og_height) . "\" />\n";
1180 1814 }
@@ -1179,9 +1813,9 @@
1179 1813 echo "<meta property=\"og:image:height\" content=\"" . esc_attr($og_height) . "\" />\n";
1180 1814 }
1181 1815 // Derive the real mime type instead of hardcoding image/jpeg,
1182 1816 // which mislabels PNG/WebP featured images.
1183 - $image_mime = get_post_mime_type($image_id);
1817 + $image_mime = $image_file['type'];
1184 1818 if ($image_mime) {
1185 1819 echo "<meta property=\"og:image:type\" content=\"" . esc_attr($image_mime) . "\" />\n";
1186 1820 }
1187 1821 }
@@ -1192,8 +1826,30 @@
1192 1826 echo "<meta property=\"og:image:alt\" content=\"" . esc_attr($image_alt) . "\" />\n";
1193 1827 }
1194 1828 }
1195 1829
1830 + // Alternatives, same as the enhanced emitter above. This path only
1831 + // runs when the Social Meta Manager is unavailable, but the issue
1832 + // reported against it (#636) and a site that lands here should not
1833 + // silently lose a feature it switched on.
1834 + if (!empty($primary_og_image)) {
1835 + $social_settings = $this->social_manager
1836 + ? $this->social_manager->get_settings(
1837 + $this->current_context === 'homepage' ? 'site' : $this->current_context,
1838 + $this->current_post_id
1839 + )
1840 + : [];
1841 +
1842 + if (!empty($social_settings['og_multiple_images'])) {
1843 + self::output_extra_og_images(
1844 + \ThinkRank\SEO\Social_Images::additional(
1845 + (int) $this->current_post_id,
1846 + $primary_og_image
1847 + )
1848 + );
1849 + }
1850 + }
1851 +
1196 1852 // Add article specific tags for posts only
1197 1853 if ($og_type === 'article') {
1198 1854 echo '<meta property="article:published_time" content="' . esc_attr(get_the_date('c', $this->current_post_id)) . '" />' . "\n";
1199 1855 echo '<meta property="article:modified_time" content="' . esc_attr(get_the_modified_date('c', $this->current_post_id)) . '" />' . "\n";
@@ -1221,8 +1877,21 @@
1221 1877 *
1222 1878 * @return void
1223 1879 */
1224 1880 public function output_twitter_card_tags(): void {
1881 + // Per-content-type Twitter card switch; see output_open_graph_tags().
1882 + if (!\ThinkRank\SEO\Content_Type_Settings::is_enabled_for_current(
1883 + \ThinkRank\SEO\Content_Type_Settings::FEATURE_TWITTER,
1884 + true
1885 + )) {
1886 + return;
1887 + }
1888 +
1889 + // Same reasoning as the Open Graph block: nothing on a 404 is shareable.
1890 + if ($this->current_context === '404') {
1891 + return;
1892 + }
1893 +
1225 1894 // Priority 1: Try Social Meta Manager (Social Media tab settings)
1226 1895 if ($this->social_manager) {
1227 1896 // Map context for Social Meta Manager (homepage -> site for site-wide settings)
1228 1897 $social_context = $this->current_context === 'homepage' ? 'site' : $this->current_context;
@@ -1269,8 +1938,14 @@
1269 1938 $twitter_title_override = \ThinkRank\SEO\Pattern_Resolver::resolve_value((string) get_post_meta($pid, '_thinkrank_twitter_title', true), $pid);
1270 1939 $twitter_description_override = \ThinkRank\SEO\Pattern_Resolver::resolve_value((string) get_post_meta($pid, '_thinkrank_twitter_description', true), $pid);
1271 1940 $og_title_override = \ThinkRank\SEO\Pattern_Resolver::resolve_value((string) get_post_meta($pid, '_thinkrank_og_title', true), $pid);
1272 1941 $og_description_override = \ThinkRank\SEO\Pattern_Resolver::resolve_value((string) get_post_meta($pid, '_thinkrank_og_description', true), $pid);
1942 + } elseif ($this->current_term_id) {
1943 + $tid = $this->current_term_id;
1944 + $twitter_title_override = \ThinkRank\SEO\Pattern_Resolver::resolve_term_value((string) get_term_meta($tid, '_thinkrank_twitter_title', true), $tid);
1945 + $twitter_description_override = \ThinkRank\SEO\Pattern_Resolver::resolve_term_value((string) get_term_meta($tid, '_thinkrank_twitter_description', true), $tid);
1946 + $og_title_override = \ThinkRank\SEO\Pattern_Resolver::resolve_term_value((string) get_term_meta($tid, '_thinkrank_og_title', true), $tid);
1947 + $og_description_override = \ThinkRank\SEO\Pattern_Resolver::resolve_term_value((string) get_term_meta($tid, '_thinkrank_og_description', true), $tid);
1273 1948 }
1274 1949
1275 1950 // Title cascade: Twitter override > OG override > Global SEO > Site Identity > default
1276 1951 $title = '';
@@ -1295,10 +1970,14 @@
1295 1970 $description = $og_description_override;
1296 1971 } else {
1297 1972 $description = $this->get_meta_description();
1298 1973 }
1299 - if (!$description) {
1300 - $description = is_singular() ? wp_trim_words(get_the_excerpt(), 30) : get_bloginfo('description');
1974 + // Skipped for a protected post: core answers get_the_excerpt() with its
1975 + // "There is no excerpt because this is a protected post." placeholder,
1976 + // so this is not a leak — but publishing that sentence as the social
1977 + // description is worse than publishing none (#363).
1978 + if (!$description && !$this->is_content_password_protected()) {
1979 + $description = is_singular() ? \ThinkRank\Core\Seo_Text::trim_words(get_the_excerpt(), 30) : get_bloginfo('description');
1301 1980 }
1302 1981
1303 1982 // Determine card type based on image availability
1304 1983 $card_type = 'summary';
@@ -1307,9 +1986,9 @@
1307 1986 }
1308 1987
1309 1988 echo "<!-- ThinkRank SEO Twitter Card Meta Tags -->\n";
1310 1989 echo '<meta name="twitter:card" content="' . esc_attr($card_type) . '" />' . "\n";
1311 - echo "<meta name=\"twitter:title\" content=\"" . esc_attr($title) . "\" />\n";
1990 + echo "<meta name=\"twitter:title\" content=\"" . esc_attr(self::strip_title_tags($title)) . "\" />\n";
1312 1991 echo "<meta name=\"twitter:description\" content=\"" . esc_attr($description) . "\" />\n";
1313 1992
1314 1993 // Add Twitter image with proper fallback priority
1315 1994 $twitter_image_url = $this->get_twitter_image_with_fallback();
@@ -1359,11 +2038,18 @@
1359 2038 }
1360 2039
1361 2040 if (empty($canonical_url)) {
1362 2041 $canonical_url = $this->current_post_id ? get_permalink($this->current_post_id) : get_permalink();
2042 +
2043 + // Core's rel_canonical() keeps the page number; this replaced
2044 + // it with a bare permalink, so every <!--nextpage--> sub-page
2045 + // and every /comment-page-N/ canonicalised to page 1 — a
2046 + // regression against core behaviour (#397). A custom canonical
2047 + // is left exactly as the user typed it.
2048 + $canonical_url = self::with_singular_page($canonical_url);
1363 2049 }
1364 2050 } else {
1365 - $canonical_url = $this->get_non_singular_canonical_url();
2051 + $canonical_url = self::get_non_singular_canonical_url();
1366 2052 }
1367 2053
1368 2054 /**
1369 2055 * Filter the canonical URL before output.
@@ -1377,14 +2063,115 @@
1377 2063 if (empty($canonical_url)) {
1378 2064 return;
1379 2065 }
1380 2066
2067 + // After the filter, so a canonical an add-on supplied is normalized
2068 + // too — and a cross-domain one is left alone, since Url_Scheme only
2069 + // touches URLs on this site's own host.
2070 + $canonical_url = \ThinkRank\SEO\Url_Scheme::apply($canonical_url);
2071 +
1381 2072 echo "<!-- ThinkRank SEO Canonical URL -->\n";
1382 2073 echo "<link rel=\"canonical\" href=\"" . esc_url($canonical_url) . "\" />\n";
1383 2074 echo "<!-- /ThinkRank SEO Canonical URL -->\n";
2075 +
2076 + $this->output_pagination_links();
1384 2077 }
1385 2078
1386 2079 /**
2080 + * Emit rel="prev" / rel="next" on a paginated archive.
2081 + *
2082 + * Nothing emitted these at all (#397). Google stopped using them as an
2083 + * indexing signal in 2019, so this is not an SEO win with Google — Bing
2084 + * still reads them, and they are the standard way to describe a sequence,
2085 + * which is what the pages are.
2086 + *
2087 + * @since 2.0.1
2088 + *
2089 + * @return void
2090 + */
2091 + private function output_pagination_links(): void {
2092 + // Page 1 still wants a rel="next" when there is a page 2, so only
2093 + // singular views are skipped outright.
2094 + if (is_singular()) {
2095 + return;
2096 + }
2097 +
2098 + global $wp_query;
2099 +
2100 + $total = $wp_query ? (int) $wp_query->max_num_pages : 0;
2101 +
2102 + if ($total < 2) {
2103 + return;
2104 + }
2105 +
2106 + $base = self::get_non_singular_canonical_url();
2107 +
2108 + if ('' === $base) {
2109 + return;
2110 + }
2111 +
2112 + // get_non_singular_canonical_url() already carries the current page —
2113 + // strip it back to page 1 before building the neighbours.
2114 + $current = self::current_page_number();
2115 + $base = self::without_pagination($base);
2116 +
2117 + if ($current > 1) {
2118 + printf(
2119 + "<link rel=\"prev\" href=\"%s\" />\n",
2120 + esc_url(\ThinkRank\SEO\Url_Scheme::apply(self::with_pagination($base, $current - 1)))
2121 + );
2122 + }
2123 +
2124 + if ($current < $total) {
2125 + printf(
2126 + "<link rel=\"next\" href=\"%s\" />\n",
2127 + esc_url(\ThinkRank\SEO\Url_Scheme::apply(self::with_pagination($base, $current + 1)))
2128 + );
2129 + }
2130 + }
2131 +
2132 + /**
2133 + * The rewrite base WordPress uses for page numbers ('page' by default).
2134 + *
2135 + * @since 2.0.1
2136 + *
2137 + * @return string
2138 + */
2139 + private static function pagination_base(): string {
2140 + global $wp_rewrite;
2141 +
2142 + return $wp_rewrite && $wp_rewrite->pagination_base ? $wp_rewrite->pagination_base : 'page';
2143 + }
2144 +
2145 + /**
2146 + * Append the sub-page or comment-page number to a singular canonical.
2147 + *
2148 + * @since 2.0.1
2149 + *
2150 + * @param string $url Permalink.
2151 + * @return string Permalink with the current page appended, when there is one.
2152 + */
2153 + public static function with_singular_page(string $url): string {
2154 + global $wp_rewrite;
2155 +
2156 + $page = (int) get_query_var('page');
2157 +
2158 + if ($page > 1) {
2159 + return $wp_rewrite && $wp_rewrite->using_permalinks()
2160 + ? trailingslashit($url) . user_trailingslashit($page, 'single_paged')
2161 + : add_query_arg('page', $page, $url);
2162 + }
2163 +
2164 + $comment_page = (int) get_query_var('cpage');
2165 +
2166 + if ($comment_page > 1) {
2167 + return get_comments_pagenum_link($comment_page);
2168 + }
2169 +
2170 + return $url;
2171 + }
2172 +
2173 + /**
1387 2174 * Build the canonical URL for non-singular contexts.
1388 2175 *
1389 2176 * Covers the blog home, post type / taxonomy / author / date archives.
1390 2177 * Search results and 404 pages get no canonical (they are noindexed).
@@ -1392,9 +2179,9 @@
1392 2179 * self-referential rather than pointing at page 1.
1393 2180 *
1394 2181 * @return string Canonical URL or '' when none applies
1395 2182 */
1396 - private function get_non_singular_canonical_url(): string {
2183 + public static function get_non_singular_canonical_url(): string {
1397 2184 if (is_404() || is_search()) {
1398 2185 return '';
1399 2186 }
1400 2187
@@ -1425,17 +2212,89 @@
1425 2212 return '';
1426 2213 }
1427 2214
1428 2215 // Point paginated archives at their own page, not page 1.
2216 + return self::with_pagination($canonical_url, (int) get_query_var('paged'));
2217 + }
2218 +
2219 + /**
2220 + * Append a page number to a URL the way WordPress does.
2221 + *
2222 + * Extracted so the archive canonical is not the only thing that knows how
2223 + * to build a paged URL: the schema graph derived its @id from the
2224 + * un-paginated link, so every page of an archive claimed the same node
2225 + * identity, and the singular canonical dropped the page entirely (#397).
2226 + *
2227 + * @since 2.0.1
2228 + *
2229 + * @param string $url Base URL.
2230 + * @param int $page Page number; 1 or less returns the URL unchanged.
2231 + * @return string
2232 + */
2233 + public static function with_pagination(string $url, int $page): string {
2234 + if ($page <= 1 || '' === $url) {
2235 + return $url;
2236 + }
2237 +
2238 + global $wp_rewrite;
2239 +
2240 + if ($wp_rewrite && $wp_rewrite->using_permalinks()) {
2241 + return trailingslashit($url) . user_trailingslashit(
2242 + $wp_rewrite->pagination_base . '/' . $page,
2243 + 'paged'
2244 + );
2245 + }
2246 +
2247 + return add_query_arg('paged', $page, $url);
2248 + }
2249 +
2250 + /**
2251 + * Strip a page number from a URL, whichever form it takes.
2252 + *
2253 + * The inverse of with_pagination(). Pretty permalinks carry the page as a
2254 + * /page/N/ path segment, plain permalinks as a `paged` query arg, and a
2255 + * regex over the path alone silently left the latter in place — so
2256 + * rel="prev" on page 2 pointed at page 2 (#397 review).
2257 + *
2258 + * @since 2.0.1
2259 + *
2260 + * @param string $url URL that may carry a page number.
2261 + * @return string URL for page 1.
2262 + */
2263 + public static function without_pagination(string $url): string {
2264 + if ('' === $url) {
2265 + return $url;
2266 + }
2267 +
2268 + $url = remove_query_arg('paged', $url);
2269 +
2270 + return (string) preg_replace(
2271 + '#/' . preg_quote(self::pagination_base(), '#') . '/\d+/?$#',
2272 + '/',
2273 + $url
2274 + );
2275 + }
2276 +
2277 + /**
2278 + * The page number of the current request, archive or multi-page post.
2279 + *
2280 + * `paged` counts archive pages; `page` counts the <!--nextpage--> parts of
2281 + * a single post. They are never both set.
2282 + *
2283 + * @since 2.0.1
2284 + *
2285 + * @return int Page number, 1 when this is the first page.
2286 + */
2287 + public static function current_page_number(): int {
1429 2288 $paged = (int) get_query_var('paged');
2289 +
1430 2290 if ($paged > 1) {
1431 - global $wp_rewrite;
1432 - $canonical_url = $wp_rewrite->using_permalinks()
1433 - ? trailingslashit($canonical_url) . user_trailingslashit($wp_rewrite->pagination_base . '/' . $paged, 'paged')
1434 - : add_query_arg('paged', $paged, $canonical_url);
2291 + return $paged;
1435 2292 }
1436 2293
1437 - return $canonical_url;
2294 + $page = (int) get_query_var('page');
2295 +
2296 + return $page > 1 ? $page : 1;
1438 2297 }
1439 2298
1440 2299
1441 2300 /**
@@ -1443,12 +2302,13 @@
1443 2302 *
1444 2303 * @return bool True if has ThinkRank metadata
1445 2304 */
1446 2305 private function has_thinkrank_metadata(): bool {
1447 - if (!is_singular()) {
1448 - return false;
1449 - }
1450 -
2306 + // Populated by initialize_current_context() for singular views and for
2307 + // term archives, and left empty everywhere else — so the emptiness
2308 + // check is the whole test. The `!is_singular()` early return this
2309 + // replaced is what made every stored term title and description inert:
2310 + // the entire title/description cascade hangs off this method (#386).
1451 2311 return !empty($this->current_metadata['title']) || !empty($this->current_metadata['description']);
1452 2312 }
1453 2313
1454 2314 /**
@@ -1582,12 +2442,21 @@
1582 2442
1583 2443 // Get excerpt
1584 2444 $post = get_post($this->current_post_id);
1585 2445 if ($post) {
1586 - $excerpt = !empty($post->post_excerpt)
1587 - ? $post->post_excerpt
1588 - : wp_trim_words(wp_strip_all_tags($post->post_content), 25, '...');
1589 - $placeholders['%excerpt%'] = $excerpt;
2446 + // An authored post_excerpt is written for public consumption, so
2447 + // it stays. Falling back to the body does not: for a protected
2448 + // post that derivation leaks the gated content through any
2449 + // template containing %excerpt%, and this branch runs BEFORE the
2450 + // derive-from-content priority below, so guarding only that one
2451 + // would leave this path open (#363).
2452 + if (!empty($post->post_excerpt)) {
2453 + $placeholders['%excerpt%'] = $post->post_excerpt;
2454 + } elseif (!$this->is_content_password_protected($post->ID)) {
2455 + $placeholders['%excerpt%'] = \ThinkRank\SEO\Pattern_Resolver::derive_excerpt(
2456 + \ThinkRank\SEO\Builder_Content::visible_content($post)
2457 + );
2458 + }
1590 2459 }
1591 2460
1592 2461 // Get author
1593 2462 $author_id = get_post_field('post_author', $this->current_post_id);
@@ -1647,8 +2516,17 @@
1647 2516 $settings = $this->site_identity_manager->get_settings('site');
1648 2517
1649 2518 switch ($this->current_context) {
1650 2519 case 'homepage':
2520 + // detect_current_context() collapses the static posts page into
2521 + // 'homepage', so it rendered the front page's title template and
2522 + // the two pages shipped the same <title> — a duplicate title on
2523 + // the site's two most-linked URLs (#397 review). It is a page,
2524 + // and it has its own name, so it gets the page template.
2525 + if (self::is_static_posts_page()) {
2526 + return $settings['page_title'] ?? $settings['homepage_title'] ?? null;
2527 + }
2528 +
1651 2529 return $settings['homepage_title'] ?? null;
1652 2530 case 'post':
1653 2531 return $settings['post_title'] ?? null;
1654 2532 case 'page':
@@ -1679,12 +2557,18 @@
1679 2557 $settings = $this->site_identity_manager->get_settings('site');
1680 2558 $separator = $this->get_title_separator($settings['title_separator'] ?? 'pipe');
1681 2559
1682 2560 $placeholders = [
1683 - '%site_title%' => $settings['site_name'] ?? get_bloginfo('name'),
1684 - '%site_name%' => $settings['site_name'] ?? get_bloginfo('name'),
1685 - '%site_description%' => $settings['site_description'] ?? get_bloginfo('description'),
1686 - '%tagline%' => $settings['tagline'] ?? get_bloginfo('description'),
2561 + // first_non_empty(), not `??`: Site Identity persists these as ''
2562 + // rather than leaving them unset, and '' is not null — so the
2563 + // null-coalesce stopped dead on the empty string and the WordPress
2564 + // fallback was unreachable. A site with a tagline set in Settings →
2565 + // General rendered "%site_description%" as nothing (#398). This is
2566 + // the same reasoning first_non_empty()'s own docblock records.
2567 + '%site_title%' => $this->first_non_empty($settings['site_name'] ?? '', get_bloginfo('name')),
2568 + '%site_name%' => $this->first_non_empty($settings['site_name'] ?? '', get_bloginfo('name')),
2569 + '%site_description%' => $this->first_non_empty($settings['site_description'] ?? '', get_bloginfo('description')),
2570 + '%tagline%' => $this->first_non_empty($settings['tagline'] ?? '', get_bloginfo('description')),
1687 2571 '%separator%' => ' ' . $separator . ' ',
1688 2572 '%sep%' => ' ' . $separator . ' ',
1689 2573 '%date%' => gmdate('F Y'),
1690 2574 ];
@@ -1737,10 +2621,26 @@
1737 2621 $placeholders['%search_term%'] = get_search_query();
1738 2622 break;
1739 2623
1740 2624 case 'archive':
1741 - $placeholders['%archive_title%'] = get_the_archive_title();
2625 + // Stripped: get_the_archive_title() wraps its subject in a
2626 + // <span>, and this placeholder feeds the document <title> as
2627 + // well as og:title and twitter:title — a date archive rendered
2628 + // as "Month: <span>August 2026</span> | Site".
2629 + $placeholders['%archive_title%'] = self::archive_subject();
1742 2630 break;
2631 +
2632 + case 'homepage':
2633 + // The page template resolved for a static posts page needs the
2634 + // page's own name; without it %title%/%page_title% would render
2635 + // empty and collapse back to the site title.
2636 + if (self::is_static_posts_page()) {
2637 + $posts_page_title = get_the_title((int) get_option('page_for_posts'));
2638 + $placeholders['%title%'] = $posts_page_title;
2639 + $placeholders['%page_title%'] = $posts_page_title;
2640 + $placeholders['%post_title%'] = $posts_page_title;
2641 + }
2642 + break;
1743 2643 }
1744 2644
1745 2645 return $placeholders;
1746 2646 }
@@ -1745,8 +2645,19 @@
1745 2645 return $placeholders;
1746 2646 }
1747 2647
1748 2648 /**
2649 + * Whether this request is a static posts page rather than the front page.
2650 + *
2651 + * @since 2.0.1
2652 + *
2653 + * @return bool
2654 + */
2655 + private static function is_static_posts_page(): bool {
2656 + return is_home() && !is_front_page() && (int) get_option('page_for_posts') > 0;
2657 + }
2658 +
2659 + /**
1749 2660 * Process title template with placeholders
1750 2661 *
1751 2662 * @param string $template Template string
1752 2663 * @param array $placeholders Placeholder values
@@ -1783,8 +2694,42 @@
1783 2694 return \ThinkRank\SEO\Site_Identity_Manager::$title_separators[$separator_type]['symbol'] ?? \ThinkRank\SEO\Site_Identity_Manager::$title_separators['pipe']['symbol'];
1784 2695 }
1785 2696
1786 2697 /**
2698 + * Whether a post's body must not be read for a public surface.
2699 + *
2700 + * Deriving metadata from `post_content` publishes that content to everyone
2701 + * who requests the URL — and to every crawler and link-preview unfurler
2702 + * that reads og:description — while the page itself still shows only the
2703 + * password form, so the leak is invisible to the site owner (#363).
2704 + *
2705 + * This is the one thing every content reader should call before touching
2706 + * `post_content` for output. It mirrors core: a visitor who has already
2707 + * entered the correct password sees the body anyway, so nothing is hidden
2708 + * from them here either.
2709 + *
2710 + * @param int|null $post_id Optional. Post ID. Defaults to the current post.
2711 + * @return bool True when the body is password-gated for this visitor.
2712 + */
2713 + private function is_content_password_protected(?int $post_id = null): bool {
2714 + $post_id = $post_id ?? $this->current_post_id;
2715 +
2716 + if (!$post_id) {
2717 + return false;
2718 + }
2719 +
2720 + $post = get_post($post_id);
2721 +
2722 + if (!$post) {
2723 + return false;
2724 + }
2725 +
2726 + // Guarded for the same reason the schema path guards it: this class is
2727 + // also exercised outside a full front-end request.
2728 + return function_exists('post_password_required') && post_password_required($post);
2729 + }
2730 +
2731 + /**
1787 2732 * Get meta description with fallback system
1788 2733 * Priority: Post-specific metadata > Global SEO templates > Site Identity templates > WordPress defaults
1789 2734 *
1790 2735 * @return string|null Meta description or null if none available
@@ -1807,8 +2752,22 @@
1807 2752 if ($archive_description) {
1808 2753 return $archive_description;
1809 2754 }
1810 2755
2756 + // The blog-index homepage's own description (Site Identity
2757 + // homepage_description, #897), in the same vocabulary as its title. A
2758 + // static front page is a page, and its description is set on it.
2759 + if (is_front_page() && is_home() && $this->site_identity_data && $this->site_identity_data['enabled']) {
2760 + $identity = $this->site_identity_manager->get_settings('site');
2761 + $template = trim((string) ($identity['homepage_description'] ?? ''));
2762 + if ($template !== '') {
2763 + $homepage_description = trim($this->process_title_template($template, $this->get_title_placeholders()));
2764 + if ($homepage_description !== '') {
2765 + return $homepage_description;
2766 + }
2767 + }
2768 + }
2769 +
1811 2770 // Third priority: Site Identity default meta description
1812 2771 if ($this->site_identity_data && $this->site_identity_data['enabled']) {
1813 2772 $settings = $this->site_identity_manager->get_settings('site');
1814 2773 $default_description = $settings['default_meta_description'] ?? '';
@@ -1817,13 +2776,23 @@
1817 2776 return $default_description;
1818 2777 }
1819 2778 }
1820 2779
1821 - // Fourth priority: Generate from content for posts/pages
1822 - if (is_singular() && $this->current_post_id) {
1823 - $post_content = get_post_field('post_content', $this->current_post_id);
2780 + // Fourth priority: Generate from content for posts/pages.
2781 + // Never for a password-protected post — deriving the description from a
2782 + // gated body published its first ~25 words in the page head, and the
2783 + // same value is reused for og:description and twitter:description, so
2784 + // one unguarded read leaked through three tags (#363).
2785 + if (is_singular() && $this->current_post_id && !$this->is_content_password_protected()) {
2786 + // Not the raw column: a Bricks page discards `post_content`, so
2787 + // whatever is still stored there is invisible — and this one value
2788 + // becomes the meta, og: and twitter: descriptions (#651).
2789 + $described = get_post($this->current_post_id);
2790 + $post_content = $described instanceof \WP_Post
2791 + ? \ThinkRank\SEO\Builder_Content::visible_content($described)
2792 + : get_post_field('post_content', $this->current_post_id);
1824 2793 if ($post_content) {
1825 - $excerpt = wp_trim_words(wp_strip_all_tags($post_content), 25, '...');
2794 + $excerpt = \ThinkRank\SEO\Pattern_Resolver::derive_excerpt((string) $post_content);
1826 2795 if (!empty($excerpt)) {
1827 2796 return $excerpt;
1828 2797 }
1829 2798 }
@@ -1867,11 +2836,13 @@
1867 2836 if ($description === '') {
1868 2837 return null;
1869 2838 }
1870 2839
1871 - if (strlen($description) > 160) {
1872 - $description = wp_trim_words($description, 25, '...');
1873 - }
2840 + // Measure and cut in CHARACTERS. strlen() counts bytes, so a Thai or
2841 + // CJK description tripped this limit at a third of its length, and
2842 + // wp_trim_words() then cut by a unit the locale chooses — 25 words in
2843 + // English, 25 characters in Thai (#687).
2844 + $description = \ThinkRank\Core\Seo_Text::trim_to_length($description);
1874 2845
1875 2846 return $description;
1876 2847 }
1877 2848
@@ -1918,11 +2889,13 @@
1918 2889 $description = preg_replace('/\s+/', ' ', $description);
1919 2890 $description = trim($description);
1920 2891
1921 2892 // Ensure description doesn't exceed recommended length (160 characters)
1922 - if (strlen($description) > 160) {
1923 - $description = wp_trim_words($description, 25, '...');
1924 - }
2893 + // Measure and cut in CHARACTERS. strlen() counts bytes, so a Thai or
2894 + // CJK description tripped this limit at a third of its length, and
2895 + // wp_trim_words() then cut by a unit the locale chooses — 25 words in
2896 + // English, 25 characters in Thai (#687).
2897 + $description = \ThinkRank\Core\Seo_Text::trim_to_length($description);
1925 2898
1926 2899 return $description;
1927 2900 }
1928 2901
@@ -1936,8 +2909,19 @@
1936 2909 public function output_site_schema_markup(): void {
1937 2910 $has_schema_manager_output = false;
1938 2911 $has_website_schema = false;
1939 2912
2913 + // The master switch on Essential SEO -> Schema Manager. Until #461 this
2914 + // was never read here, so turning schema off left every deployed entity
2915 + // on the page. Read it once and bail before touching the graph.
2916 + if ($this->schema_manager) {
2917 + $schema_settings = $this->schema_manager->get_settings('site', null);
2918 +
2919 + if (isset($schema_settings['enabled']) && !$schema_settings['enabled']) {
2920 + return;
2921 + }
2922 + }
2923 +
1940 2924 // PRIORITY 1: Always output site-wide schemas (Organization, Website, LocalBusiness, Person)
1941 2925 if ($this->schema_manager) {
1942 2926 $site_wide_schemas = $this->schema_manager->get_deployed_schemas('site', null);
1943 2927
@@ -1942,9 +2926,9 @@
1942 2926 $site_wide_schemas = $this->schema_manager->get_deployed_schemas('site', null);
1943 2927
1944 2928 if (!empty($site_wide_schemas)) {
1945 2929 foreach ($site_wide_schemas as $schema_type => $schema_info) {
1946 - $this->output_schema_markup($schema_info['data'], $schema_type, 'Schema Manager');
2930 + Schema_Graph::instance()->add_supporting($schema_info['data'], (string) $schema_type);
1947 2931 }
1948 2932 $has_schema_manager_output = true;
1949 2933 $has_website_schema = isset($site_wide_schemas['WebSite']);
1950 2934 }
@@ -1965,9 +2949,9 @@
1965 2949 */
1966 2950 $website_schema = apply_filters('thinkrank_website_schema', $website_schema);
1967 2951
1968 2952 if (!empty($website_schema)) {
1969 - $this->output_schema_markup($website_schema, 'WebSite', 'Site Identity');
2953 + Schema_Graph::instance()->add_supporting($website_schema, 'WebSite');
1970 2954 }
1971 2955 }
1972 2956
1973 2957 // PRIORITY 2: Also output page-specific schemas (Article, HowTo, FAQ, etc.) on individual posts/pages
@@ -1978,10 +2962,19 @@
1978 2962
1979 2963 $page_specific_schemas = $this->schema_manager->get_deployed_schemas($context_type, $context_id);
1980 2964
1981 2965 if (!empty($page_specific_schemas)) {
1982 - // Apply filter for Pro to allow multiple schemas
1983 - // In free version, it's limited to 1 schema if not filtered
2966 + // Every deployed schema is rendered, on every plan. How many a
2967 + // page carries is decided when schemas are activated in the
2968 + // editor, not trimmed here by plan (#673).
2969 +
2970 + /**
2971 + * Filter the page-specific schemas rendered on the current page.
2972 + *
2973 + * @param array $page_specific_schemas Deployed schemas keyed by schema type.
2974 + * @param string $context_type Context type (post, page, product, site).
2975 + * @param int $context_id Post ID.
2976 + */
1984 2977 $page_specific_schemas = apply_filters(
1985 2978 'thinkrank_page_schemas_to_render',
1986 2979 $page_specific_schemas,
1987 2980 $context_type,
@@ -1987,21 +2980,37 @@
1987 2980 $context_type,
1988 2981 $context_id
1989 2982 );
1990 2983
1991 - // If still multiple schemas and not Pro, limit to 1 (enforcing free limit)
1992 - $is_pro = \ThinkRank\Core\Plan_Config::is_pro();
1993 - if (!$is_pro && count($page_specific_schemas) > 2) {
1994 - $page_specific_schemas = array_slice($page_specific_schemas, 0, 2, true);
1995 - }
2984 + // A deployed node is a snapshot from Deploy time and outranks
2985 + // the automatic node, so page and article types would publish
2986 + // a frozen excerpt instead of the description the head
2987 + // resolves. Give them the live one, as the automatic node has.
2988 + $context_post = get_post($context_id);
1996 2989
1997 2990 foreach ($page_specific_schemas as $schema_type => $schema_info) {
1998 - $this->output_schema_markup($schema_info['data'], $schema_type, 'Schema Manager');
2991 + $node = $schema_info['data'];
2992 +
2993 + if ($this->global_seo_schema && $context_post instanceof \WP_Post) {
2994 + $node = $this->global_seo_schema->refresh_deployed_description($node, (string) $schema_type, $context_post);
2995 + }
2996 +
2997 + Schema_Graph::instance()->add_primary($node, (string) $schema_type, 'schema_manager');
1999 2998 }
2000 2999 $has_schema_manager_output = true;
2001 3000 }
2002 3001 }
2003 3002
3003 + // Absorb FAQ content from the post body (FAQ block / Elementor widget)
3004 + // so it merges into the graph's single FAQPage instead of each producer
3005 + // emitting its own competing one.
3006 + if (is_singular()) {
3007 + $queried_post = get_post();
3008 + if ($queried_post instanceof \WP_Post) {
3009 + Schema_Graph::instance()->collect_post_faq($queried_post);
3010 + }
3011 + }
3012 +
2004 3013 // Skip Site Identity fallback if any Schema Manager schemas were output
2005 3014 if ($has_schema_manager_output) {
2006 3015 return;
2007 3016 }
@@ -2020,9 +3029,9 @@
2020 3029
2021 3030 $schema = $this->generate_organization_schema($settings);
2022 3031
2023 3032 if ($schema) {
2024 - $this->output_schema_markup($schema, 'Organization', 'Site Identity');
3033 + Schema_Graph::instance()->add_supporting($schema, 'Organization');
2025 3034 }
2026 3035 }
2027 3036
2028 3037 /**
@@ -2044,42 +3053,50 @@
2044 3053 'url' => home_url('/'),
2045 3054 ];
2046 3055
2047 3056 $description = !empty($settings['site_description']) ? $settings['site_description'] : get_bloginfo('description');
3057 + // The tagline is stored esc_html()'d by sanitize_option(), so a site
3058 + // called "Fish & Chips" published `&amp;` literally in its WebSite
3059 + // node; nothing decodes JSON-LD downstream.
3060 + $description = \ThinkRank\Core\Seo_Text::normalize_schema_text((string) $description);
2048 3061 if (!empty($description)) {
2049 3062 $schema['description'] = $description;
2050 3063 }
2051 3064
2052 - $schema['potentialAction'] = [
2053 - '@type' => 'SearchAction',
2054 - 'target' => [
2055 - '@type' => 'EntryPoint',
2056 - 'urlTemplate' => home_url('/?s={search_term_string}'),
2057 - ],
2058 - 'query-input' => 'required name=search_term_string',
2059 - ];
3065 + // Site Identity has accepted an alternate name since the setup wizard
3066 + // shipped, and the MCP ability describes it as "published as schema
3067 + // alternateName" — but no producer ever read it, so the promise was
3068 + // false and every imported Yoast/Rank Math value sat unused (#692).
3069 + $alternate_name = \ThinkRank\SEO\Site_Identity_Manager::alternate_name_for_schema($settings['alternate_name'] ?? null);
3070 + if (null !== $alternate_name) {
3071 + $schema['alternateName'] = $alternate_name;
3072 + }
2060 3073
2061 - return $schema;
2062 - }
3074 + // The sitelinks searchbox switch was honoured only for a deployed
3075 + // WebSite row; this live fallback added potentialAction unconditionally,
3076 + // so website_enable_search = 0 still shipped the SearchAction (#688).
3077 + // Absent means not configured, which stays enabled.
3078 + $search_enabled = true;
3079 + if ($this->schema_manager) {
3080 + $schema_settings = $this->schema_manager->get_settings('site', null);
2063 3081
2064 - /**
2065 - * Output schema markup with consistent formatting
2066 - *
2067 - * @param array $schema_data Schema data
2068 - * @param string $schema_type Schema type name
2069 - * @param string $source Source of schema (Schema Manager, Site Identity, etc.)
2070 - * @return void
2071 - */
2072 - private function output_schema_markup(array $schema_data, string $schema_type, string $source): void {
2073 - if (empty($schema_data)) {
2074 - return;
3082 + if (array_key_exists('website_enable_search', $schema_settings)) {
3083 + $search_enabled = !empty($schema_settings['website_enable_search']);
3084 + }
2075 3085 }
2076 3086
2077 - echo '<!-- ThinkRank ' . esc_html($source) . ': ' . esc_html($schema_type) . ' Schema -->' . "\n";
2078 - echo '<script type="application/ld+json">' . "\n";
2079 - echo wp_json_encode($schema_data, JSON_UNESCAPED_SLASHES | JSON_UNESCAPED_UNICODE | JSON_PRETTY_PRINT | JSON_HEX_TAG | JSON_HEX_AMP | JSON_HEX_APOS | JSON_HEX_QUOT) . "\n";
2080 - echo '</script>' . "\n";
2081 - echo '<!-- /ThinkRank ' . esc_html($source) . ': ' . esc_html($schema_type) . ' Schema -->' . "\n";
3087 + if ($search_enabled) {
3088 + $schema['potentialAction'] = [
3089 + '@type' => 'SearchAction',
3090 + 'target' => [
3091 + '@type' => 'EntryPoint',
3092 + 'urlTemplate' => home_url('/?s={search_term_string}'),
3093 + ],
3094 + 'query-input' => 'required name=search_term_string',
3095 + ];
3096 + }
3097 +
3098 + return $schema;
2082 3099 }
2083 3100
2084 3101 /**
2085 3102 * Generate organization schema markup
@@ -2275,8 +3292,15 @@
2275 3292 if (!$this->site_identity_data || !$this->site_identity_data['enabled']) {
2276 3293 return;
2277 3294 }
2278 3295
3296 + // A breadcrumb trail for a URL that does not exist, or for a search
3297 + // results page, describes nothing — and the plugin already emits no
3298 + // canonical on either (#471).
3299 + if (is_404() || is_search()) {
3300 + return;
3301 + }
3302 +
2279 3303 $settings = $this->site_identity_manager->get_settings('site');
2280 3304
2281 3305 // Only output if breadcrumbs are enabled
2282 3306 if (empty($settings['breadcrumbs_enabled'])) {
@@ -2282,42 +3306,74 @@
2282 3306 if (empty($settings['breadcrumbs_enabled'])) {
2283 3307 return;
2284 3308 }
2285 3309
3310 + // Schema Manager's own breadcrumb switch. Only Site Identity's
3311 + // breadcrumbs_enabled was consulted here, so enable_breadcrumbs_schema
3312 + // = 0 removed a deployed BreadcrumbList row and left this live one
3313 + // emitting the node anyway (#688). Absent means not configured, which
3314 + // stays enabled.
3315 + if ($this->schema_manager) {
3316 + $schema_settings = $this->schema_manager->get_settings('site', null);
3317 +
3318 + if (array_key_exists('enable_breadcrumbs_schema', $schema_settings)
3319 + && empty($schema_settings['enable_breadcrumbs_schema'])) {
3320 + return;
3321 + }
3322 + }
3323 +
2286 3324 $breadcrumbs = $this->generate_breadcrumbs($settings);
2287 3325
2288 3326 if (!empty($breadcrumbs['schema'])) {
2289 - echo "<!-- ThinkRank SEO Breadcrumb Schema Markup -->\n";
2290 - echo '<script type="application/ld+json">' . "\n";
2291 - echo wp_json_encode($breadcrumbs['schema'], JSON_UNESCAPED_SLASHES | JSON_UNESCAPED_UNICODE | JSON_PRETTY_PRINT | JSON_HEX_TAG | JSON_HEX_AMP | JSON_HEX_APOS | JSON_HEX_QUOT) . "\n";
2292 - echo '</script>' . "\n";
2293 - echo "<!-- /ThinkRank SEO Breadcrumb Schema Markup -->\n";
3327 + Schema_Graph::instance()->add_supporting($breadcrumbs['schema'], 'BreadcrumbList');
2294 3328 }
2295 3329 }
2296 3330
2297 3331 /**
3332 + * Emit everything ThinkRank collected for this request as one linked @graph.
3333 + *
3334 + * Runs after every producer has registered (site schema 7, breadcrumbs 8,
3335 + * Global SEO 15), so the graph can arbitrate between them.
3336 + *
3337 + * @since 1.32.0
3338 + * @return void
3339 + */
3340 + public function output_schema_graph(): void {
3341 + Schema_Graph::instance()->render();
3342 + }
3343 +
3344 + /**
2298 3345 * Output closing comment for ThinkRank SEO
2299 3346 *
2300 3347 * @return void
2301 3348 */
2302 3349 public function output_closing_comment(): void {
2303 - // Only output if we've output any SEO content
2304 - static $header_output = false;
2305 - if ($header_output || $this->has_seo_output()) {
3350 + // Close only what was actually opened. has_seo_output() is true on
3351 + // nearly every page, so testing it here printed a closing comment with
3352 + // no matching opener whenever the meta description was empty (search
3353 + // results, author archives without a description).
3354 + if (self::$opening_comment_output) {
2306 3355 echo "<!-- /ThinkRank SEO -->\n";
2307 3356 }
2308 3357 }
2309 3358
2310 3359 /**
2311 - * Check if any SEO content has been output
3360 + * Print the opening ThinkRank comment, once per request.
2312 3361 *
2313 - * @return bool True if SEO content was output
3362 + * Public and static so Author_Archives_Manager — which prints its own meta
3363 + * description on wp_head at priority 5 — opens the block through the same
3364 + * flag the closing comment reads.
3365 + *
3366 + * @since 2.0.1
3367 + * @return void
2314 3368 */
2315 - private function has_seo_output(): bool {
2316 - // Check if we have meta description or any other SEO data
2317 - return !empty($this->get_meta_description()) ||
2318 - $this->has_thinkrank_metadata() ||
2319 - ($this->site_identity_data && $this->site_identity_data['enabled']);
3369 + public static function note_opening_comment(): void {
3370 + if (self::$opening_comment_output) {
3371 + return;
3372 + }
3373 +
3374 + echo "<!-- Search Engine Optimization by ThinkRank - https://thinkrank.ai/ -->\n";
3375 + self::$opening_comment_output = true;
2320 3376 }
2321 3377
2322 3378 /**
2323 3379 * Display breadcrumbs HTML
@@ -2511,8 +3567,151 @@
2511 3567 return $breadcrumbs;
2512 3568 }
2513 3569
2514 3570 /**
3571 + * Serve /llms.txt through PHP so the response declares UTF-8.
3572 + *
3573 + * Cheap guard first: every other front-end request leaves without loading
3574 + * the manager.
3575 + *
3576 + * @since 1.32.0
3577 + *
3578 + * @return void
3579 + */
3580 + /**
3581 + * Serve a ThinkRank sitemap document for this request, when it is one.
3582 + *
3583 + * Only acts in dynamic delivery mode. In static mode a real file exists and
3584 + * the web server returns it without WordPress ever loading, so answering
3585 + * here as well would mean two sources for the same bytes.
3586 + *
3587 + * @since 2.9.0
3588 + *
3589 + * @return void
3590 + */
3591 + public function maybe_serve_sitemap(): void {
3592 + $filename = $this->requested_sitemap_filename();
3593 + if ('' === $filename) {
3594 + return;
3595 + }
3596 +
3597 + try {
3598 + // Read-only instance: passing false keeps it from registering a
3599 + // second copy of the auto-generation hooks.
3600 + $generator = new \ThinkRank\SEO\Sitemap_Generator(false);
3601 + $settings = $generator->get_settings('site');
3602 +
3603 + if (empty($settings['enabled'])) {
3604 + return;
3605 + }
3606 +
3607 + if ('dynamic' !== $generator->resolve_delivery_mode($settings)) {
3608 + return;
3609 + }
3610 +
3611 + if (!$generator->publishes_document_name($filename, $settings)) {
3612 + return;
3613 + }
3614 +
3615 + $xml = $generator->render_document($filename, $settings);
3616 + } catch (\Throwable $e) {
3617 + // A failed render must not replace the sitemap with a fatal. Leave
3618 + // the request alone so WordPress answers as it otherwise would.
3619 + return;
3620 + }
3621 +
3622 + if (!is_string($xml) || '' === trim($xml)) {
3623 + return;
3624 + }
3625 +
3626 + status_header(200);
3627 + header('Content-Type: application/xml; charset=UTF-8');
3628 + header('X-Robots-Tag: noindex, follow', true);
3629 +
3630 + // Built XML, escaped by the builders as they assemble it; escaping the
3631 + // document here would corrupt it.
3632 + echo $xml; // phpcs:ignore WordPress.Security.EscapeOutput.OutputNotEscaped
3633 + exit;
3634 + }
3635 +
3636 + /**
3637 + * The sitemap file name this request is asking for, if it looks like one.
3638 + *
3639 + * Deliberately a cheap shape test. Whether the site actually publishes the
3640 + * name is settled by the caller against the generator, so that a request
3641 + * for someone else's sitemap is never answered here.
3642 + *
3643 + * @since 2.9.0
3644 + *
3645 + * @return string File name, or '' when this is not a sitemap request.
3646 + */
3647 + private function requested_sitemap_filename(): string {
3648 + if (empty($_SERVER['REQUEST_URI'])) {
3649 + return '';
3650 + }
3651 +
3652 + $path = wp_parse_url(sanitize_text_field(wp_unslash($_SERVER['REQUEST_URI'])), PHP_URL_PATH);
3653 + if (!is_string($path) || '' === $path) {
3654 + return '';
3655 + }
3656 +
3657 + // Strip the install's home path so subdirectory installs match too.
3658 + $home_path = (string) wp_parse_url(home_url('/'), PHP_URL_PATH);
3659 + if ('' !== $home_path && '/' !== $home_path && 0 === strpos($path, $home_path)) {
3660 + $path = substr($path, strlen($home_path));
3661 + }
3662 +
3663 + $candidate = strtolower(trim($path, '/'));
3664 +
3665 + // One path segment ending in .xml. Anything nested is not a file we
3666 + // publish to the web root.
3667 + if ('' === $candidate || strpos($candidate, '/') !== false) {
3668 + return '';
3669 + }
3670 +
3671 + return substr($candidate, -4) === '.xml' ? $candidate : '';
3672 + }
3673 +
3674 + public function maybe_serve_llms_txt(): void {
3675 + if (!$this->is_llms_txt_request()) {
3676 + return;
3677 + }
3678 +
3679 + if (!class_exists('ThinkRank\\SEO\\LLMs_Txt_Manager')) {
3680 + require_once THINKRANK_PLUGIN_DIR . 'includes/seo/class-llms-txt-manager.php';
3681 + }
3682 +
3683 + $manager = new \ThinkRank\SEO\LLMs_Txt_Manager();
3684 + $manager->serve_llms_txt();
3685 + }
3686 +
3687 + /**
3688 + * Whether the current request is for /llms.txt.
3689 + *
3690 + * @since 1.32.0
3691 + *
3692 + * @return bool
3693 + */
3694 + private function is_llms_txt_request(): bool {
3695 + if (empty($_SERVER['REQUEST_URI'])) {
3696 + return false;
3697 + }
3698 +
3699 + $path = wp_parse_url(sanitize_text_field(wp_unslash($_SERVER['REQUEST_URI'])), PHP_URL_PATH);
3700 + if (!is_string($path) || '' === $path) {
3701 + return false;
3702 + }
3703 +
3704 + // Strip the install's home path so subdirectory installs match too.
3705 + $home_path = (string) wp_parse_url(home_url('/'), PHP_URL_PATH);
3706 + if ('' !== $home_path && '/' !== $home_path && 0 === strpos($path, $home_path)) {
3707 + $path = substr($path, strlen($home_path));
3708 + }
3709 +
3710 + return 'llms.txt' === strtolower(trim($path, '/'));
3711 + }
3712 +
3713 + /**
2515 3714 * Filter WordPress robots.txt output
2516 3715 *
2517 3716 * @param string $output The default robots.txt output
2518 3717 * @param string $is_public Whether the site is public
@@ -2545,8 +3744,190 @@
2545 3744 return $output;
2546 3745 }
2547 3746
2548 3747 /**
3748 + * Disable WordPress core's sitemap while ThinkRank's sitemap is enabled.
3749 + *
3750 + * Prevents the site from publishing two competing sitemap indexes. Core's
3751 + * /wp-sitemap.xml is taken offline (it 404s) and, as a consequence, core
3752 + * stops adding its own "Sitemap:" directive to robots.txt — including on the
3753 + * paths where ThinkRank does not own the robots.txt output.
3754 + *
3755 + * Only ever turns core's sitemap *off*: when ThinkRank's sitemap is disabled
3756 + * the incoming value is returned untouched, so core (or another plugin
3757 + * filtering this) keeps whatever behaviour it already had.
3758 + *
3759 + * @since 1.31.0
3760 + *
3761 + * @param bool $enabled Whether core's sitemap functionality is enabled.
3762 + * @return bool Filtered value.
3763 + */
3764 + public function filter_wp_sitemaps_enabled($enabled): bool {
3765 + return $this->should_disable_core_sitemap() ? false : (bool) $enabled;
3766 + }
3767 +
3768 + /**
3769 + * Redirect the sitemap URLs core owns to the sitemap ThinkRank publishes.
3770 + *
3771 + * Only the *index* route is redirected. Core's per-type children
3772 + * (/wp-sitemap-posts-post-1.xml and friends) are genuinely gone once core is
3773 + * switched off, and a 404 is the honest answer for those; the index is the
3774 + * one URL crawlers and humans actually guess, and the one core's own
3775 + * /sitemap.xml rule funnels into.
3776 + *
3777 + * @since 1.31.0
3778 + *
3779 + * @return void
3780 + */
3781 + public function redirect_core_sitemap_requests(): void {
3782 + if ('index' !== get_query_var('sitemap')) {
3783 + return;
3784 + }
3785 +
3786 + $request_uri = isset($_SERVER['REQUEST_URI'])
3787 + ? sanitize_text_field(wp_unslash($_SERVER['REQUEST_URI']))
3788 + : '';
3789 +
3790 + $target = $this->resolve_core_sitemap_redirect(
3791 + (string) wp_parse_url($request_uri, PHP_URL_PATH)
3792 + );
3793 +
3794 + if ('' === $target) {
3795 + return;
3796 + }
3797 +
3798 + wp_safe_redirect($target, 301, 'ThinkRank');
3799 + exit;
3800 + }
3801 +
3802 + /**
3803 + * Where a request for one of core's sitemap URLs should be sent, if anywhere.
3804 + *
3805 + * Split out from the hook so the rules are testable without dispatching a
3806 + * request — the caller above is the only part that cannot be (it exits).
3807 + *
3808 + * @since 1.31.0
3809 + *
3810 + * @param string $requested_path Path of the incoming request.
3811 + * @return string Absolute URL to redirect to, or '' to leave the request alone.
3812 + */
3813 + private function resolve_core_sitemap_redirect(string $requested_path): string {
3814 + // Nothing to redirect to unless we have actually taken core offline,
3815 + // which already implies our own sitemap file is on disk.
3816 + if (!$this->should_disable_core_sitemap()) {
3817 + return '';
3818 + }
3819 +
3820 + $target = $this->thinkrank_sitemap_url;
3821 + if ('' === $target) {
3822 + return '';
3823 + }
3824 +
3825 + // Never redirect a URL to itself. A site publishing at /sitemap.xml
3826 + // normally has the web server serve that file before WordPress sees the
3827 + // request, but on a setup where the request does reach PHP this is the
3828 + // difference between a redirect and a loop.
3829 + $destination = (string) wp_parse_url($target, PHP_URL_PATH);
3830 +
3831 + if ('' !== $requested_path && untrailingslashit($requested_path) === untrailingslashit($destination)) {
3832 + return '';
3833 + }
3834 +
3835 + return $target;
3836 + }
3837 +
3838 + /**
3839 + * Whether core's sitemap should be switched off for this site.
3840 + *
3841 + * True when ThinkRank publishes its own sitemap — except in two cases where
3842 + * taking core offline would leave a URL answering nothing:
3843 + *
3844 + * 1. The site is configured to publish *at core's own URL* (the "WordPress
3845 + * Core" preset). Once the static file exists the web server serves it
3846 + * ahead of WordPress anyway, so core can be left alone.
3847 + * 2. ThinkRank's own sitemap file is not on disk yet. Sitemaps here are
3848 + * static files with no dynamic route (see save_sitemap_to_file()), so
3849 + * while the file is missing core's /sitemap.xml -> /wp-sitemap.xml
3850 + * redirect is the only thing answering that URL; suppressing core would
3851 + * turn a recoverable "enabled but not generated" state into a hard 404
3852 + * for crawlers. Core is taken offline as soon as our file appears, so the
3853 + * duplicate-index conflict this filter exists to prevent cannot occur —
3854 + * two indexes are only ever reachable if both are actually published.
3855 + *
3856 + * Once our file does exist, the URLs core stops answering are handed to
3857 + * redirect_core_sitemap_requests() rather than left to 404 — which is why
3858 + * this also resolves the destination.
3859 + *
3860 + * Resolved lazily and memoised: this is consulted from an `init`-time filter
3861 + * on every request, and the underlying settings read is object-cached. The
3862 + * memoisation also keeps the file_exists() call to one per request.
3863 + *
3864 + * @since 1.31.0
3865 + *
3866 + * @return bool True when WordPress core's sitemap should be disabled.
3867 + */
3868 + private function should_disable_core_sitemap(): bool {
3869 + if ($this->thinkrank_sitemap_enabled === null) {
3870 + try {
3871 + // Read-only instance — passing false keeps it from registering a
3872 + // second copy of the save_post/term auto-generation hooks.
3873 + $generator = new \ThinkRank\SEO\Sitemap_Generator(false);
3874 + $settings = $generator->get_settings('site');
3875 +
3876 + // "Can ThinkRank actually answer its sitemap URL right now?" In
3877 + // static mode that means the file is on disk; in dynamic mode
3878 + // maybe_serve_sitemap() answers it, so there is nothing to look
3879 + // for. Keeping the file test as the only answer would have left
3880 + // core's sitemap in place on every dynamic site, which is the
3881 + // crawl conflict this suppression exists to prevent (#752).
3882 + // The #346 behaviour is unchanged: a static site with nothing
3883 + // published still falls through to core rather than 404ing.
3884 + $can_serve = 'dynamic' === $generator->resolve_delivery_mode($settings)
3885 + || $generator->primary_sitemap_file_exists($settings);
3886 +
3887 + $this->thinkrank_sitemap_enabled = !empty($settings['enabled'])
3888 + && !$this->publishes_at_core_sitemap_url($settings)
3889 + && $can_serve;
3890 +
3891 + if ($this->thinkrank_sitemap_enabled) {
3892 + $this->thinkrank_sitemap_url = $generator->get_primary_sitemap_url($settings);
3893 + }
3894 + } catch (\Exception $e) {
3895 + // Settings unreadable — leave core's sitemap alone rather than
3896 + // removing a working sitemap on the strength of a failed read.
3897 + $this->thinkrank_sitemap_enabled = false;
3898 + $this->thinkrank_sitemap_url = '';
3899 + }
3900 + }
3901 +
3902 + return $this->thinkrank_sitemap_enabled;
3903 + }
3904 +
3905 + /**
3906 + * Whether any configured sitemap URL is WordPress core's own wp-sitemap.xml.
3907 + *
3908 + * @since 1.31.0
3909 + *
3910 + * @param array $settings Sitemap settings.
3911 + * @return bool True when the site publishes at core's sitemap URL.
3912 + */
3913 + private function publishes_at_core_sitemap_url(array $settings): bool {
3914 + foreach ((array) ($settings['sitemap_urls'] ?? []) as $sitemap) {
3915 + if (!is_array($sitemap)) {
3916 + continue;
3917 + }
3918 +
3919 + $path = (string) wp_parse_url((string) ($sitemap['url'] ?? ''), PHP_URL_PATH);
3920 +
3921 + if (ltrim($path, '/') === 'wp-sitemap.xml') {
3922 + return true;
3923 + }
3924 + }
3925 +
3926 + return false;
3927 + }
3928 +
3929 + /**
2549 3930 * Re-sync the physical robots.txt when WordPress's "Discourage search
2550 3931 * engines" setting (blog_public) changes.
2551 3932 *
2552 3933 * Only acts when ThinkRank robots management is enabled AND a physical
@@ -2596,15 +3977,17 @@
2596 3977 if (empty($settings['enabled'])) {
2597 3978 return (string) $url;
2598 3979 }
2599 3980
3981 + $size = (int) $size;
3982 +
2600 3983 // Apple touch icon has its own dedicated setting
2601 - if ((int) $size === 180 && !empty($settings['apple_touch_icon_url'])) {
2602 - return esc_url($settings['apple_touch_icon_url']);
3984 + if ($size === 180 && !empty($settings['apple_touch_icon_url'])) {
3985 + return $this->resolve_icon_url((string) $settings['apple_touch_icon_url'], $size);
2603 3986 }
2604 3987
2605 3988 if (!empty($settings['favicon_url'])) {
2606 - return esc_url($settings['favicon_url']);
3989 + return $this->resolve_icon_url((string) $settings['favicon_url'], $size);
2607 3990 }
2608 3991
2609 3992 return (string) $url;
2610 3993 }
@@ -2609,8 +3992,178 @@
2609 3992 return (string) $url;
2610 3993 }
2611 3994
2612 3995 /**
3996 + * Whether breadcrumb labels should prefer the SEO title.
3997 + *
3998 + * Off unless the site turns it on, so updating the plugin never rewrites an
3999 + * existing trail.
4000 + *
4001 + * @since 2.3.1
4002 + *
4003 + * @param array $settings Breadcrumb settings.
4004 + * @return bool
4005 + */
4006 + private function breadcrumbs_use_seo_title(array $settings): bool {
4007 + return !empty($settings['breadcrumb_use_seo_title']);
4008 + }
4009 +
4010 + /**
4011 + * Label for a post in the breadcrumb trail.
4012 + *
4013 + * With the toggle on, the post's own SEO title wins — the same
4014 + * `_thinkrank_seo_title` value (variable tags resolved) the document title
4015 + * uses — so the trail under a search snippet reads the same as the snippet
4016 + * itself. Anything empty falls back to the raw post title; the global title
4017 + * pattern is deliberately NOT part of the chain, since resolving it would
4018 + * append the site name to every crumb.
4019 + *
4020 + * @since 2.3.1
4021 + *
4022 + * @param int $post_id Post ID.
4023 + * @param array $settings Breadcrumb settings.
4024 + * @return string Breadcrumb label.
4025 + */
4026 + private function get_breadcrumb_post_title(int $post_id, array $settings): string {
4027 + $title = (string) get_the_title($post_id);
4028 +
4029 + if (!$this->breadcrumbs_use_seo_title($settings)) {
4030 + return $title;
4031 + }
4032 +
4033 + $seo_title = trim((string) get_post_meta($post_id, '_thinkrank_seo_title', true));
4034 +
4035 + if ('' === $seo_title) {
4036 + return $title;
4037 + }
4038 +
4039 + $resolved = trim(\ThinkRank\SEO\Pattern_Resolver::resolve_value($seo_title, $post_id));
4040 +
4041 + return '' !== $resolved ? $resolved : $title;
4042 + }
4043 +
4044 + /**
4045 + * Label for a term in the breadcrumb trail.
4046 + *
4047 + * Term counterpart to {@see self::get_breadcrumb_post_title()}, resolving
4048 + * the term's `_thinkrank_seo_title` against its own values.
4049 + *
4050 + * @since 2.3.1
4051 + *
4052 + * @param object $term Term object.
4053 + * @param array $settings Breadcrumb settings.
4054 + * @return string Breadcrumb label.
4055 + */
4056 + private function get_breadcrumb_term_title($term, array $settings): string {
4057 + $name = (string) ($term->name ?? '');
4058 +
4059 + if (!$this->breadcrumbs_use_seo_title($settings) || empty($term->term_id)) {
4060 + return $name;
4061 + }
4062 +
4063 + $seo_title = trim((string) get_term_meta((int) $term->term_id, '_thinkrank_seo_title', true));
4064 +
4065 + if ('' === $seo_title) {
4066 + return $name;
4067 + }
4068 +
4069 + $resolved = trim(\ThinkRank\SEO\Pattern_Resolver::resolve_term_value($seo_title, (int) $term->term_id));
4070 +
4071 + return '' !== $resolved ? $resolved : $name;
4072 + }
4073 +
4074 + /**
4075 + * Resolve a configured icon URL to the derivative that fits $size.
4076 + *
4077 + * wp_site_icon() calls get_site_icon_url() four times — 32, 192, 180 and
4078 + * 270 — and pairs the first two with a hardcoded sizes="" attribute. This
4079 + * filter used to answer all four with the same configured URL, so one
4080 + * upload was declared as every size at once: a 1536x1536 original served
4081 + * to paint a 32px tab icon, under a sizes="32x32" label that was simply
4082 + * untrue (#571).
4083 + *
4084 + * Resolution mirrors core's own get_site_icon_url(), including the
4085 + * >= 512 -> 'full' branch, so ThinkRank's override and the core pipeline
4086 + * pick the same file for the same request.
4087 + *
4088 + * An unresolvable URL (one hosted off-site) is returned unchanged. Nothing
4089 + * is knowable about its dimensions, and suppressing it instead would leave
4090 + * the page with no rel="icon" at all — a worse outcome than an approximate
4091 + * size hint.
4092 + *
4093 + * @param string $configured Configured icon URL.
4094 + * @param int $size Icon size core is asking for.
4095 + * @return string Icon URL for that size.
4096 + */
4097 + private function resolve_icon_url(string $configured, int $size): string {
4098 + $cache_key = md5($configured) . ':' . $size;
4099 + $cached = $this->icon_urls();
4100 +
4101 + if (isset($cached[$cache_key])) {
4102 + return $cached[$cache_key];
4103 + }
4104 +
4105 + $attachment_id = \ThinkRank\SEO\Site_Identity_Manager::icon_attachment_id($configured);
4106 +
4107 + if (!$attachment_id) {
4108 + $resolved = esc_url($configured);
4109 + } else {
4110 + // Mirrors core: at 512 and above the original is what is wanted, and
4111 + // asking for an intermediate size that large would only fall back to it.
4112 + $size_data = $size >= 512 ? 'full' : [$size, $size];
4113 + $url = wp_get_attachment_image_url($attachment_id, $size_data);
4114 + $resolved = $url ? esc_url($url) : esc_url($configured);
4115 + }
4116 +
4117 + $this->icon_urls[$cache_key] = $resolved;
4118 +
4119 + if (!$this->icon_urls_dirty) {
4120 + $this->icon_urls_dirty = true;
4121 + // Written once, after the response is assembled, rather than once
4122 + // per size: wp_site_icon() resolves four in a row.
4123 + add_action('shutdown', [$this, 'persist_icon_urls'], 5);
4124 + }
4125 +
4126 + return $resolved;
4127 + }
4128 +
4129 + /**
4130 + * The resolved-icon-URL map, loaded from its transient on first use.
4131 + *
4132 + * @return array<string, string>
4133 + */
4134 + private function icon_urls(): array {
4135 + if ($this->icon_urls === null) {
4136 + $stored = get_transient(\ThinkRank\SEO\Site_Identity_Manager::ICON_URL_TRANSIENT);
4137 + $this->icon_urls = is_array($stored) ? $stored : [];
4138 + }
4139 +
4140 + return $this->icon_urls;
4141 + }
4142 +
4143 + /**
4144 + * Persist newly resolved icon URLs.
4145 + *
4146 + * Public because it runs on `shutdown`. Invalidated wholesale whenever the
4147 + * site identity settings are saved, which is the only moment the icon
4148 + * choice — or the derivatives behind it — can change.
4149 + *
4150 + * @return void
4151 + */
4152 + public function persist_icon_urls(): void {
4153 + if (!$this->icon_urls_dirty || !is_array($this->icon_urls)) {
4154 + return;
4155 + }
4156 +
4157 + $this->icon_urls_dirty = false;
4158 + set_transient(
4159 + \ThinkRank\SEO\Site_Identity_Manager::ICON_URL_TRANSIENT,
4160 + $this->icon_urls,
4161 + DAY_IN_SECONDS
4162 + );
4163 + }
4164 +
4165 + /**
2613 4166 * Get breadcrumb items for current page
2614 4167 *
2615 4168 * @param array $settings Breadcrumb settings
2616 4169 * @return array Breadcrumb items
@@ -2638,9 +4191,9 @@
2638 4191 $categories = get_the_category($current_post_id);
2639 4192 if (!empty($categories)) {
2640 4193 $category = $categories[0];
2641 4194 $items[] = [
2642 - 'title' => $category->name,
4195 + 'title' => $this->get_breadcrumb_term_title($category, $settings),
2643 4196 'url' => get_category_link($category->term_id),
2644 4197 'position' => $position++
2645 4198 ];
2646 4199 }
@@ -2645,12 +4198,18 @@
2645 4198 ];
2646 4199 }
2647 4200 }
2648 4201
2649 - // Add current post
2650 - if (empty($settings['show_current_page']) || $settings['show_current_page']) {
4202 + // Add current post. `empty($x) || $x` is true for every possible
4203 + // value — an unset key, false, 0, '' and any truthy value alike —
4204 + // so the setting had no effect on the rendered breadcrumb or on
4205 + // the BreadcrumbList JSON-LD, while the admin preview honoured it
4206 + // and disagreed with live output (#398). Site_Identity_Manager
4207 + // already had the correct form: default to on, respect an
4208 + // explicit off.
4209 + if ($settings['show_current_page'] ?? true) {
2651 4210 $items[] = [
2652 - 'title' => get_the_title($current_post_id),
4211 + 'title' => $this->get_breadcrumb_post_title($current_post_id, $settings),
2653 4212 'url' => get_permalink($current_post_id),
2654 4213 'position' => $position,
2655 4214 'current' => true
2656 4215 ];
@@ -2667,9 +4226,9 @@
2667 4226 while ($parent_id) {
2668 4227 $parent = get_post($parent_id);
2669 4228 if ($parent) {
2670 4229 $parents[] = [
2671 - 'title' => get_the_title($parent->ID),
4230 + 'title' => $this->get_breadcrumb_post_title($parent->ID, $settings),
2672 4231 'url' => get_permalink($parent->ID),
2673 4232 'position' => 0 // Will be set later
2674 4233 ];
2675 4234 $parent_id = $parent->post_parent;
@@ -2687,11 +4246,11 @@
2687 4246 $items[] = $parent;
2688 4247 }
2689 4248
2690 4249 // Add current page
2691 - if (empty($settings['show_current_page']) || $settings['show_current_page']) {
4250 + if ($settings['show_current_page'] ?? true) {
2692 4251 $items[] = [
2693 - 'title' => get_the_title($current_post_id),
4252 + 'title' => $this->get_breadcrumb_post_title($current_post_id, $settings),
2694 4253 'url' => get_permalink($current_post_id),
2695 4254 'position' => $position,
2696 4255 'current' => true
2697 4256 ];
@@ -2707,9 +4266,9 @@
2707 4266 while ($parent_id) {
2708 4267 $parent = get_category($parent_id);
2709 4268 if ($parent && !is_wp_error($parent)) {
2710 4269 $parents[] = [
2711 - 'title' => $parent->name,
4270 + 'title' => $this->get_breadcrumb_term_title($parent, $settings),
2712 4271 'url' => get_category_link($parent->term_id),
2713 4272 'position' => 0 // Will be set later
2714 4273 ];
2715 4274 $parent_id = $parent->parent;
@@ -2727,11 +4286,11 @@
2727 4286 $items[] = $parent;
2728 4287 }
2729 4288
2730 4289 // Add current category
2731 - if (empty($settings['show_current_page']) || $settings['show_current_page']) {
4290 + if ($settings['show_current_page'] ?? true) {
2732 4291 $items[] = [
2733 - 'title' => $category->name,
4292 + 'title' => $this->get_breadcrumb_term_title($category, $settings),
2734 4293 'url' => get_category_link($category->term_id),
2735 4294 'position' => $position,
2736 4295 'current' => true
2737 4296 ];