PluginProbe
ThinkRank AI SEO – AI SEO Plugin for WordPress: Schema, XML Sitemaps, Meta Tags, Search Console & Local SEO / 2.10.0
ThinkRank AI SEO – AI SEO Plugin for WordPress: Schema, XML Sitemaps, Meta Tags, Search Console & Local SEO v2.10.0
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 trunk 1.0.0 1.0.1 All 51 releases
← All changes | includes/frontend/class-seo-manager.php +1686 -145 1.30.0 → 2.10.0 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($image_url) . "\" />\n";
1168 1798
1169 1799 // Get image dimensions and alt text
@@ -1192,8 +1822,30 @@
1192 1822 echo "<meta property=\"og:image:alt\" content=\"" . esc_attr($image_alt) . "\" />\n";
1193 1823 }
1194 1824 }
1195 1825
1826 + // Alternatives, same as the enhanced emitter above. This path only
1827 + // runs when the Social Meta Manager is unavailable, but the issue
1828 + // reported against it (#636) and a site that lands here should not
1829 + // silently lose a feature it switched on.
1830 + if (!empty($primary_og_image)) {
1831 + $social_settings = $this->social_manager
1832 + ? $this->social_manager->get_settings(
1833 + $this->current_context === 'homepage' ? 'site' : $this->current_context,
1834 + $this->current_post_id
1835 + )
1836 + : [];
1837 +
1838 + if (!empty($social_settings['og_multiple_images'])) {
1839 + self::output_extra_og_images(
1840 + \ThinkRank\SEO\Social_Images::additional(
1841 + (int) $this->current_post_id,
1842 + $primary_og_image
1843 + )
1844 + );
1845 + }
1846 + }
1847 +
1196 1848 // Add article specific tags for posts only
1197 1849 if ($og_type === 'article') {
1198 1850 echo '<meta property="article:published_time" content="' . esc_attr(get_the_date('c', $this->current_post_id)) . '" />' . "\n";
1199 1851 echo '<meta property="article:modified_time" content="' . esc_attr(get_the_modified_date('c', $this->current_post_id)) . '" />' . "\n";
@@ -1221,8 +1873,21 @@
1221 1873 *
1222 1874 * @return void
1223 1875 */
1224 1876 public function output_twitter_card_tags(): void {
1877 + // Per-content-type Twitter card switch; see output_open_graph_tags().
1878 + if (!\ThinkRank\SEO\Content_Type_Settings::is_enabled_for_current(
1879 + \ThinkRank\SEO\Content_Type_Settings::FEATURE_TWITTER,
1880 + true
1881 + )) {
1882 + return;
1883 + }
1884 +
1885 + // Same reasoning as the Open Graph block: nothing on a 404 is shareable.
1886 + if ($this->current_context === '404') {
1887 + return;
1888 + }
1889 +
1225 1890 // Priority 1: Try Social Meta Manager (Social Media tab settings)
1226 1891 if ($this->social_manager) {
1227 1892 // Map context for Social Meta Manager (homepage -> site for site-wide settings)
1228 1893 $social_context = $this->current_context === 'homepage' ? 'site' : $this->current_context;
@@ -1269,8 +1934,14 @@
1269 1934 $twitter_title_override = \ThinkRank\SEO\Pattern_Resolver::resolve_value((string) get_post_meta($pid, '_thinkrank_twitter_title', true), $pid);
1270 1935 $twitter_description_override = \ThinkRank\SEO\Pattern_Resolver::resolve_value((string) get_post_meta($pid, '_thinkrank_twitter_description', true), $pid);
1271 1936 $og_title_override = \ThinkRank\SEO\Pattern_Resolver::resolve_value((string) get_post_meta($pid, '_thinkrank_og_title', true), $pid);
1272 1937 $og_description_override = \ThinkRank\SEO\Pattern_Resolver::resolve_value((string) get_post_meta($pid, '_thinkrank_og_description', true), $pid);
1938 + } elseif ($this->current_term_id) {
1939 + $tid = $this->current_term_id;
1940 + $twitter_title_override = \ThinkRank\SEO\Pattern_Resolver::resolve_term_value((string) get_term_meta($tid, '_thinkrank_twitter_title', true), $tid);
1941 + $twitter_description_override = \ThinkRank\SEO\Pattern_Resolver::resolve_term_value((string) get_term_meta($tid, '_thinkrank_twitter_description', true), $tid);
1942 + $og_title_override = \ThinkRank\SEO\Pattern_Resolver::resolve_term_value((string) get_term_meta($tid, '_thinkrank_og_title', true), $tid);
1943 + $og_description_override = \ThinkRank\SEO\Pattern_Resolver::resolve_term_value((string) get_term_meta($tid, '_thinkrank_og_description', true), $tid);
1273 1944 }
1274 1945
1275 1946 // Title cascade: Twitter override > OG override > Global SEO > Site Identity > default
1276 1947 $title = '';
@@ -1295,10 +1966,14 @@
1295 1966 $description = $og_description_override;
1296 1967 } else {
1297 1968 $description = $this->get_meta_description();
1298 1969 }
1299 - if (!$description) {
1300 - $description = is_singular() ? wp_trim_words(get_the_excerpt(), 30) : get_bloginfo('description');
1970 + // Skipped for a protected post: core answers get_the_excerpt() with its
1971 + // "There is no excerpt because this is a protected post." placeholder,
1972 + // so this is not a leak — but publishing that sentence as the social
1973 + // description is worse than publishing none (#363).
1974 + if (!$description && !$this->is_content_password_protected()) {
1975 + $description = is_singular() ? \ThinkRank\Core\Seo_Text::trim_words(get_the_excerpt(), 30) : get_bloginfo('description');
1301 1976 }
1302 1977
1303 1978 // Determine card type based on image availability
1304 1979 $card_type = 'summary';
@@ -1307,9 +1982,9 @@
1307 1982 }
1308 1983
1309 1984 echo "<!-- ThinkRank SEO Twitter Card Meta Tags -->\n";
1310 1985 echo '<meta name="twitter:card" content="' . esc_attr($card_type) . '" />' . "\n";
1311 - echo "<meta name=\"twitter:title\" content=\"" . esc_attr($title) . "\" />\n";
1986 + echo "<meta name=\"twitter:title\" content=\"" . esc_attr(self::strip_title_tags($title)) . "\" />\n";
1312 1987 echo "<meta name=\"twitter:description\" content=\"" . esc_attr($description) . "\" />\n";
1313 1988
1314 1989 // Add Twitter image with proper fallback priority
1315 1990 $twitter_image_url = $this->get_twitter_image_with_fallback();
@@ -1359,11 +2034,18 @@
1359 2034 }
1360 2035
1361 2036 if (empty($canonical_url)) {
1362 2037 $canonical_url = $this->current_post_id ? get_permalink($this->current_post_id) : get_permalink();
2038 +
2039 + // Core's rel_canonical() keeps the page number; this replaced
2040 + // it with a bare permalink, so every <!--nextpage--> sub-page
2041 + // and every /comment-page-N/ canonicalised to page 1 — a
2042 + // regression against core behaviour (#397). A custom canonical
2043 + // is left exactly as the user typed it.
2044 + $canonical_url = self::with_singular_page($canonical_url);
1363 2045 }
1364 2046 } else {
1365 - $canonical_url = $this->get_non_singular_canonical_url();
2047 + $canonical_url = self::get_non_singular_canonical_url();
1366 2048 }
1367 2049
1368 2050 /**
1369 2051 * Filter the canonical URL before output.
@@ -1377,14 +2059,115 @@
1377 2059 if (empty($canonical_url)) {
1378 2060 return;
1379 2061 }
1380 2062
2063 + // After the filter, so a canonical an add-on supplied is normalized
2064 + // too — and a cross-domain one is left alone, since Url_Scheme only
2065 + // touches URLs on this site's own host.
2066 + $canonical_url = \ThinkRank\SEO\Url_Scheme::apply($canonical_url);
2067 +
1381 2068 echo "<!-- ThinkRank SEO Canonical URL -->\n";
1382 2069 echo "<link rel=\"canonical\" href=\"" . esc_url($canonical_url) . "\" />\n";
1383 2070 echo "<!-- /ThinkRank SEO Canonical URL -->\n";
2071 +
2072 + $this->output_pagination_links();
1384 2073 }
1385 2074
1386 2075 /**
2076 + * Emit rel="prev" / rel="next" on a paginated archive.
2077 + *
2078 + * Nothing emitted these at all (#397). Google stopped using them as an
2079 + * indexing signal in 2019, so this is not an SEO win with Google — Bing
2080 + * still reads them, and they are the standard way to describe a sequence,
2081 + * which is what the pages are.
2082 + *
2083 + * @since 2.0.1
2084 + *
2085 + * @return void
2086 + */
2087 + private function output_pagination_links(): void {
2088 + // Page 1 still wants a rel="next" when there is a page 2, so only
2089 + // singular views are skipped outright.
2090 + if (is_singular()) {
2091 + return;
2092 + }
2093 +
2094 + global $wp_query;
2095 +
2096 + $total = $wp_query ? (int) $wp_query->max_num_pages : 0;
2097 +
2098 + if ($total < 2) {
2099 + return;
2100 + }
2101 +
2102 + $base = self::get_non_singular_canonical_url();
2103 +
2104 + if ('' === $base) {
2105 + return;
2106 + }
2107 +
2108 + // get_non_singular_canonical_url() already carries the current page —
2109 + // strip it back to page 1 before building the neighbours.
2110 + $current = self::current_page_number();
2111 + $base = self::without_pagination($base);
2112 +
2113 + if ($current > 1) {
2114 + printf(
2115 + "<link rel=\"prev\" href=\"%s\" />\n",
2116 + esc_url(\ThinkRank\SEO\Url_Scheme::apply(self::with_pagination($base, $current - 1)))
2117 + );
2118 + }
2119 +
2120 + if ($current < $total) {
2121 + printf(
2122 + "<link rel=\"next\" href=\"%s\" />\n",
2123 + esc_url(\ThinkRank\SEO\Url_Scheme::apply(self::with_pagination($base, $current + 1)))
2124 + );
2125 + }
2126 + }
2127 +
2128 + /**
2129 + * The rewrite base WordPress uses for page numbers ('page' by default).
2130 + *
2131 + * @since 2.0.1
2132 + *
2133 + * @return string
2134 + */
2135 + private static function pagination_base(): string {
2136 + global $wp_rewrite;
2137 +
2138 + return $wp_rewrite && $wp_rewrite->pagination_base ? $wp_rewrite->pagination_base : 'page';
2139 + }
2140 +
2141 + /**
2142 + * Append the sub-page or comment-page number to a singular canonical.
2143 + *
2144 + * @since 2.0.1
2145 + *
2146 + * @param string $url Permalink.
2147 + * @return string Permalink with the current page appended, when there is one.
2148 + */
2149 + public static function with_singular_page(string $url): string {
2150 + global $wp_rewrite;
2151 +
2152 + $page = (int) get_query_var('page');
2153 +
2154 + if ($page > 1) {
2155 + return $wp_rewrite && $wp_rewrite->using_permalinks()
2156 + ? trailingslashit($url) . user_trailingslashit($page, 'single_paged')
2157 + : add_query_arg('page', $page, $url);
2158 + }
2159 +
2160 + $comment_page = (int) get_query_var('cpage');
2161 +
2162 + if ($comment_page > 1) {
2163 + return get_comments_pagenum_link($comment_page);
2164 + }
2165 +
2166 + return $url;
2167 + }
2168 +
2169 + /**
1387 2170 * Build the canonical URL for non-singular contexts.
1388 2171 *
1389 2172 * Covers the blog home, post type / taxonomy / author / date archives.
1390 2173 * Search results and 404 pages get no canonical (they are noindexed).
@@ -1392,9 +2175,9 @@
1392 2175 * self-referential rather than pointing at page 1.
1393 2176 *
1394 2177 * @return string Canonical URL or '' when none applies
1395 2178 */
1396 - private function get_non_singular_canonical_url(): string {
2179 + public static function get_non_singular_canonical_url(): string {
1397 2180 if (is_404() || is_search()) {
1398 2181 return '';
1399 2182 }
1400 2183
@@ -1425,17 +2208,89 @@
1425 2208 return '';
1426 2209 }
1427 2210
1428 2211 // Point paginated archives at their own page, not page 1.
2212 + return self::with_pagination($canonical_url, (int) get_query_var('paged'));
2213 + }
2214 +
2215 + /**
2216 + * Append a page number to a URL the way WordPress does.
2217 + *
2218 + * Extracted so the archive canonical is not the only thing that knows how
2219 + * to build a paged URL: the schema graph derived its @id from the
2220 + * un-paginated link, so every page of an archive claimed the same node
2221 + * identity, and the singular canonical dropped the page entirely (#397).
2222 + *
2223 + * @since 2.0.1
2224 + *
2225 + * @param string $url Base URL.
2226 + * @param int $page Page number; 1 or less returns the URL unchanged.
2227 + * @return string
2228 + */
2229 + public static function with_pagination(string $url, int $page): string {
2230 + if ($page <= 1 || '' === $url) {
2231 + return $url;
2232 + }
2233 +
2234 + global $wp_rewrite;
2235 +
2236 + if ($wp_rewrite && $wp_rewrite->using_permalinks()) {
2237 + return trailingslashit($url) . user_trailingslashit(
2238 + $wp_rewrite->pagination_base . '/' . $page,
2239 + 'paged'
2240 + );
2241 + }
2242 +
2243 + return add_query_arg('paged', $page, $url);
2244 + }
2245 +
2246 + /**
2247 + * Strip a page number from a URL, whichever form it takes.
2248 + *
2249 + * The inverse of with_pagination(). Pretty permalinks carry the page as a
2250 + * /page/N/ path segment, plain permalinks as a `paged` query arg, and a
2251 + * regex over the path alone silently left the latter in place — so
2252 + * rel="prev" on page 2 pointed at page 2 (#397 review).
2253 + *
2254 + * @since 2.0.1
2255 + *
2256 + * @param string $url URL that may carry a page number.
2257 + * @return string URL for page 1.
2258 + */
2259 + public static function without_pagination(string $url): string {
2260 + if ('' === $url) {
2261 + return $url;
2262 + }
2263 +
2264 + $url = remove_query_arg('paged', $url);
2265 +
2266 + return (string) preg_replace(
2267 + '#/' . preg_quote(self::pagination_base(), '#') . '/\d+/?$#',
2268 + '/',
2269 + $url
2270 + );
2271 + }
2272 +
2273 + /**
2274 + * The page number of the current request, archive or multi-page post.
2275 + *
2276 + * `paged` counts archive pages; `page` counts the <!--nextpage--> parts of
2277 + * a single post. They are never both set.
2278 + *
2279 + * @since 2.0.1
2280 + *
2281 + * @return int Page number, 1 when this is the first page.
2282 + */
2283 + public static function current_page_number(): int {
1429 2284 $paged = (int) get_query_var('paged');
2285 +
1430 2286 if ($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);
2287 + return $paged;
1435 2288 }
1436 2289
1437 - return $canonical_url;
2290 + $page = (int) get_query_var('page');
2291 +
2292 + return $page > 1 ? $page : 1;
1438 2293 }
1439 2294
1440 2295
1441 2296 /**
@@ -1443,12 +2298,13 @@
1443 2298 *
1444 2299 * @return bool True if has ThinkRank metadata
1445 2300 */
1446 2301 private function has_thinkrank_metadata(): bool {
1447 - if (!is_singular()) {
1448 - return false;
1449 - }
1450 -
2302 + // Populated by initialize_current_context() for singular views and for
2303 + // term archives, and left empty everywhere else — so the emptiness
2304 + // check is the whole test. The `!is_singular()` early return this
2305 + // replaced is what made every stored term title and description inert:
2306 + // the entire title/description cascade hangs off this method (#386).
1451 2307 return !empty($this->current_metadata['title']) || !empty($this->current_metadata['description']);
1452 2308 }
1453 2309
1454 2310 /**
@@ -1582,12 +2438,21 @@
1582 2438
1583 2439 // Get excerpt
1584 2440 $post = get_post($this->current_post_id);
1585 2441 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%'] = $excerpt;
2442 + // An authored post_excerpt is written for public consumption, so
2443 + // it stays. Falling back to the body does not: for a protected
2444 + // post that derivation leaks the gated content through any
2445 + // template containing %excerpt%, and this branch runs BEFORE the
2446 + // derive-from-content priority below, so guarding only that one
2447 + // would leave this path open (#363).
2448 + if (!empty($post->post_excerpt)) {
2449 + $placeholders['%excerpt%'] = $post->post_excerpt;
2450 + } elseif (!$this->is_content_password_protected($post->ID)) {
2451 + $placeholders['%excerpt%'] = \ThinkRank\SEO\Pattern_Resolver::derive_excerpt(
2452 + \ThinkRank\SEO\Builder_Content::visible_content($post)
2453 + );
2454 + }
1590 2455 }
1591 2456
1592 2457 // Get author
1593 2458 $author_id = get_post_field('post_author', $this->current_post_id);
@@ -1647,8 +2512,17 @@
1647 2512 $settings = $this->site_identity_manager->get_settings('site');
1648 2513
1649 2514 switch ($this->current_context) {
1650 2515 case 'homepage':
2516 + // detect_current_context() collapses the static posts page into
2517 + // 'homepage', so it rendered the front page's title template and
2518 + // the two pages shipped the same <title> — a duplicate title on
2519 + // the site's two most-linked URLs (#397 review). It is a page,
2520 + // and it has its own name, so it gets the page template.
2521 + if (self::is_static_posts_page()) {
2522 + return $settings['page_title'] ?? $settings['homepage_title'] ?? null;
2523 + }
2524 +
1651 2525 return $settings['homepage_title'] ?? null;
1652 2526 case 'post':
1653 2527 return $settings['post_title'] ?? null;
1654 2528 case 'page':
@@ -1679,12 +2553,18 @@
1679 2553 $settings = $this->site_identity_manager->get_settings('site');
1680 2554 $separator = $this->get_title_separator($settings['title_separator'] ?? 'pipe');
1681 2555
1682 2556 $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'),
2557 + // first_non_empty(), not `??`: Site Identity persists these as ''
2558 + // rather than leaving them unset, and '' is not null — so the
2559 + // null-coalesce stopped dead on the empty string and the WordPress
2560 + // fallback was unreachable. A site with a tagline set in Settings →
2561 + // General rendered "%site_description%" as nothing (#398). This is
2562 + // the same reasoning first_non_empty()'s own docblock records.
2563 + '%site_title%' => $this->first_non_empty($settings['site_name'] ?? '', get_bloginfo('name')),
2564 + '%site_name%' => $this->first_non_empty($settings['site_name'] ?? '', get_bloginfo('name')),
2565 + '%site_description%' => $this->first_non_empty($settings['site_description'] ?? '', get_bloginfo('description')),
2566 + '%tagline%' => $this->first_non_empty($settings['tagline'] ?? '', get_bloginfo('description')),
1687 2567 '%separator%' => ' ' . $separator . ' ',
1688 2568 '%sep%' => ' ' . $separator . ' ',
1689 2569 '%date%' => gmdate('F Y'),
1690 2570 ];
@@ -1737,10 +2617,26 @@
1737 2617 $placeholders['%search_term%'] = get_search_query();
1738 2618 break;
1739 2619
1740 2620 case 'archive':
1741 - $placeholders['%archive_title%'] = get_the_archive_title();
2621 + // Stripped: get_the_archive_title() wraps its subject in a
2622 + // <span>, and this placeholder feeds the document <title> as
2623 + // well as og:title and twitter:title — a date archive rendered
2624 + // as "Month: <span>August 2026</span> | Site".
2625 + $placeholders['%archive_title%'] = self::archive_subject();
1742 2626 break;
2627 +
2628 + case 'homepage':
2629 + // The page template resolved for a static posts page needs the
2630 + // page's own name; without it %title%/%page_title% would render
2631 + // empty and collapse back to the site title.
2632 + if (self::is_static_posts_page()) {
2633 + $posts_page_title = get_the_title((int) get_option('page_for_posts'));
2634 + $placeholders['%title%'] = $posts_page_title;
2635 + $placeholders['%page_title%'] = $posts_page_title;
2636 + $placeholders['%post_title%'] = $posts_page_title;
2637 + }
2638 + break;
1743 2639 }
1744 2640
1745 2641 return $placeholders;
1746 2642 }
@@ -1745,8 +2641,19 @@
1745 2641 return $placeholders;
1746 2642 }
1747 2643
1748 2644 /**
2645 + * Whether this request is a static posts page rather than the front page.
2646 + *
2647 + * @since 2.0.1
2648 + *
2649 + * @return bool
2650 + */
2651 + private static function is_static_posts_page(): bool {
2652 + return is_home() && !is_front_page() && (int) get_option('page_for_posts') > 0;
2653 + }
2654 +
2655 + /**
1749 2656 * Process title template with placeholders
1750 2657 *
1751 2658 * @param string $template Template string
1752 2659 * @param array $placeholders Placeholder values
@@ -1783,8 +2690,42 @@
1783 2690 return \ThinkRank\SEO\Site_Identity_Manager::$title_separators[$separator_type]['symbol'] ?? \ThinkRank\SEO\Site_Identity_Manager::$title_separators['pipe']['symbol'];
1784 2691 }
1785 2692
1786 2693 /**
2694 + * Whether a post's body must not be read for a public surface.
2695 + *
2696 + * Deriving metadata from `post_content` publishes that content to everyone
2697 + * who requests the URL — and to every crawler and link-preview unfurler
2698 + * that reads og:description — while the page itself still shows only the
2699 + * password form, so the leak is invisible to the site owner (#363).
2700 + *
2701 + * This is the one thing every content reader should call before touching
2702 + * `post_content` for output. It mirrors core: a visitor who has already
2703 + * entered the correct password sees the body anyway, so nothing is hidden
2704 + * from them here either.
2705 + *
2706 + * @param int|null $post_id Optional. Post ID. Defaults to the current post.
2707 + * @return bool True when the body is password-gated for this visitor.
2708 + */
2709 + private function is_content_password_protected(?int $post_id = null): bool {
2710 + $post_id = $post_id ?? $this->current_post_id;
2711 +
2712 + if (!$post_id) {
2713 + return false;
2714 + }
2715 +
2716 + $post = get_post($post_id);
2717 +
2718 + if (!$post) {
2719 + return false;
2720 + }
2721 +
2722 + // Guarded for the same reason the schema path guards it: this class is
2723 + // also exercised outside a full front-end request.
2724 + return function_exists('post_password_required') && post_password_required($post);
2725 + }
2726 +
2727 + /**
1787 2728 * Get meta description with fallback system
1788 2729 * Priority: Post-specific metadata > Global SEO templates > Site Identity templates > WordPress defaults
1789 2730 *
1790 2731 * @return string|null Meta description or null if none available
@@ -1817,13 +2758,23 @@
1817 2758 return $default_description;
1818 2759 }
1819 2760 }
1820 2761
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);
2762 + // Fourth priority: Generate from content for posts/pages.
2763 + // Never for a password-protected post — deriving the description from a
2764 + // gated body published its first ~25 words in the page head, and the
2765 + // same value is reused for og:description and twitter:description, so
2766 + // one unguarded read leaked through three tags (#363).
2767 + if (is_singular() && $this->current_post_id && !$this->is_content_password_protected()) {
2768 + // Not the raw column: a Bricks page discards `post_content`, so
2769 + // whatever is still stored there is invisible — and this one value
2770 + // becomes the meta, og: and twitter: descriptions (#651).
2771 + $described = get_post($this->current_post_id);
2772 + $post_content = $described instanceof \WP_Post
2773 + ? \ThinkRank\SEO\Builder_Content::visible_content($described)
2774 + : get_post_field('post_content', $this->current_post_id);
1824 2775 if ($post_content) {
1825 - $excerpt = wp_trim_words(wp_strip_all_tags($post_content), 25, '...');
2776 + $excerpt = \ThinkRank\SEO\Pattern_Resolver::derive_excerpt((string) $post_content);
1826 2777 if (!empty($excerpt)) {
1827 2778 return $excerpt;
1828 2779 }
1829 2780 }
@@ -1867,11 +2818,13 @@
1867 2818 if ($description === '') {
1868 2819 return null;
1869 2820 }
1870 2821
1871 - if (strlen($description) > 160) {
1872 - $description = wp_trim_words($description, 25, '...');
1873 - }
2822 + // Measure and cut in CHARACTERS. strlen() counts bytes, so a Thai or
2823 + // CJK description tripped this limit at a third of its length, and
2824 + // wp_trim_words() then cut by a unit the locale chooses — 25 words in
2825 + // English, 25 characters in Thai (#687).
2826 + $description = \ThinkRank\Core\Seo_Text::trim_to_length($description);
1874 2827
1875 2828 return $description;
1876 2829 }
1877 2830
@@ -1918,11 +2871,13 @@
1918 2871 $description = preg_replace('/\s+/', ' ', $description);
1919 2872 $description = trim($description);
1920 2873
1921 2874 // Ensure description doesn't exceed recommended length (160 characters)
1922 - if (strlen($description) > 160) {
1923 - $description = wp_trim_words($description, 25, '...');
1924 - }
2875 + // Measure and cut in CHARACTERS. strlen() counts bytes, so a Thai or
2876 + // CJK description tripped this limit at a third of its length, and
2877 + // wp_trim_words() then cut by a unit the locale chooses — 25 words in
2878 + // English, 25 characters in Thai (#687).
2879 + $description = \ThinkRank\Core\Seo_Text::trim_to_length($description);
1925 2880
1926 2881 return $description;
1927 2882 }
1928 2883
@@ -1936,8 +2891,19 @@
1936 2891 public function output_site_schema_markup(): void {
1937 2892 $has_schema_manager_output = false;
1938 2893 $has_website_schema = false;
1939 2894
2895 + // The master switch on Essential SEO -> Schema Manager. Until #461 this
2896 + // was never read here, so turning schema off left every deployed entity
2897 + // on the page. Read it once and bail before touching the graph.
2898 + if ($this->schema_manager) {
2899 + $schema_settings = $this->schema_manager->get_settings('site', null);
2900 +
2901 + if (isset($schema_settings['enabled']) && !$schema_settings['enabled']) {
2902 + return;
2903 + }
2904 + }
2905 +
1940 2906 // PRIORITY 1: Always output site-wide schemas (Organization, Website, LocalBusiness, Person)
1941 2907 if ($this->schema_manager) {
1942 2908 $site_wide_schemas = $this->schema_manager->get_deployed_schemas('site', null);
1943 2909
@@ -1942,9 +2908,9 @@
1942 2908 $site_wide_schemas = $this->schema_manager->get_deployed_schemas('site', null);
1943 2909
1944 2910 if (!empty($site_wide_schemas)) {
1945 2911 foreach ($site_wide_schemas as $schema_type => $schema_info) {
1946 - $this->output_schema_markup($schema_info['data'], $schema_type, 'Schema Manager');
2912 + Schema_Graph::instance()->add_supporting($schema_info['data'], (string) $schema_type);
1947 2913 }
1948 2914 $has_schema_manager_output = true;
1949 2915 $has_website_schema = isset($site_wide_schemas['WebSite']);
1950 2916 }
@@ -1965,9 +2931,9 @@
1965 2931 */
1966 2932 $website_schema = apply_filters('thinkrank_website_schema', $website_schema);
1967 2933
1968 2934 if (!empty($website_schema)) {
1969 - $this->output_schema_markup($website_schema, 'WebSite', 'Site Identity');
2935 + Schema_Graph::instance()->add_supporting($website_schema, 'WebSite');
1970 2936 }
1971 2937 }
1972 2938
1973 2939 // PRIORITY 2: Also output page-specific schemas (Article, HowTo, FAQ, etc.) on individual posts/pages
@@ -1978,10 +2944,19 @@
1978 2944
1979 2945 $page_specific_schemas = $this->schema_manager->get_deployed_schemas($context_type, $context_id);
1980 2946
1981 2947 if (!empty($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
2948 + // Every deployed schema is rendered, on every plan. How many a
2949 + // page carries is decided when schemas are activated in the
2950 + // editor, not trimmed here by plan (#673).
2951 +
2952 + /**
2953 + * Filter the page-specific schemas rendered on the current page.
2954 + *
2955 + * @param array $page_specific_schemas Deployed schemas keyed by schema type.
2956 + * @param string $context_type Context type (post, page, product, site).
2957 + * @param int $context_id Post ID.
2958 + */
1984 2959 $page_specific_schemas = apply_filters(
1985 2960 'thinkrank_page_schemas_to_render',
1986 2961 $page_specific_schemas,
1987 2962 $context_type,
@@ -1987,21 +2962,37 @@
1987 2962 $context_type,
1988 2963 $context_id
1989 2964 );
1990 2965
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 - }
2966 + // A deployed node is a snapshot from Deploy time and outranks
2967 + // the automatic node, so page and article types would publish
2968 + // a frozen excerpt instead of the description the head
2969 + // resolves. Give them the live one, as the automatic node has.
2970 + $context_post = get_post($context_id);
1996 2971
1997 2972 foreach ($page_specific_schemas as $schema_type => $schema_info) {
1998 - $this->output_schema_markup($schema_info['data'], $schema_type, 'Schema Manager');
2973 + $node = $schema_info['data'];
2974 +
2975 + if ($this->global_seo_schema && $context_post instanceof \WP_Post) {
2976 + $node = $this->global_seo_schema->refresh_deployed_description($node, (string) $schema_type, $context_post);
2977 + }
2978 +
2979 + Schema_Graph::instance()->add_primary($node, (string) $schema_type, 'schema_manager');
1999 2980 }
2000 2981 $has_schema_manager_output = true;
2001 2982 }
2002 2983 }
2003 2984
2985 + // Absorb FAQ content from the post body (FAQ block / Elementor widget)
2986 + // so it merges into the graph's single FAQPage instead of each producer
2987 + // emitting its own competing one.
2988 + if (is_singular()) {
2989 + $queried_post = get_post();
2990 + if ($queried_post instanceof \WP_Post) {
2991 + Schema_Graph::instance()->collect_post_faq($queried_post);
2992 + }
2993 + }
2994 +
2004 2995 // Skip Site Identity fallback if any Schema Manager schemas were output
2005 2996 if ($has_schema_manager_output) {
2006 2997 return;
2007 2998 }
@@ -2020,9 +3011,9 @@
2020 3011
2021 3012 $schema = $this->generate_organization_schema($settings);
2022 3013
2023 3014 if ($schema) {
2024 - $this->output_schema_markup($schema, 'Organization', 'Site Identity');
3015 + Schema_Graph::instance()->add_supporting($schema, 'Organization');
2025 3016 }
2026 3017 }
2027 3018
2028 3019 /**
@@ -2044,42 +3035,50 @@
2044 3035 'url' => home_url('/'),
2045 3036 ];
2046 3037
2047 3038 $description = !empty($settings['site_description']) ? $settings['site_description'] : get_bloginfo('description');
3039 + // The tagline is stored esc_html()'d by sanitize_option(), so a site
3040 + // called "Fish & Chips" published `&amp;` literally in its WebSite
3041 + // node; nothing decodes JSON-LD downstream.
3042 + $description = \ThinkRank\Core\Seo_Text::normalize_schema_text((string) $description);
2048 3043 if (!empty($description)) {
2049 3044 $schema['description'] = $description;
2050 3045 }
2051 3046
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 - ];
3047 + // Site Identity has accepted an alternate name since the setup wizard
3048 + // shipped, and the MCP ability describes it as "published as schema
3049 + // alternateName" — but no producer ever read it, so the promise was
3050 + // false and every imported Yoast/Rank Math value sat unused (#692).
3051 + $alternate_name = \ThinkRank\SEO\Site_Identity_Manager::alternate_name_for_schema($settings['alternate_name'] ?? null);
3052 + if (null !== $alternate_name) {
3053 + $schema['alternateName'] = $alternate_name;
3054 + }
2060 3055
2061 - return $schema;
2062 - }
3056 + // The sitelinks searchbox switch was honoured only for a deployed
3057 + // WebSite row; this live fallback added potentialAction unconditionally,
3058 + // so website_enable_search = 0 still shipped the SearchAction (#688).
3059 + // Absent means not configured, which stays enabled.
3060 + $search_enabled = true;
3061 + if ($this->schema_manager) {
3062 + $schema_settings = $this->schema_manager->get_settings('site', null);
2063 3063
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;
3064 + if (array_key_exists('website_enable_search', $schema_settings)) {
3065 + $search_enabled = !empty($schema_settings['website_enable_search']);
3066 + }
2075 3067 }
2076 3068
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($schema_type) . ' Schema -->' . "\n";
3069 + if ($search_enabled) {
3070 + $schema['potentialAction'] = [
3071 + '@type' => 'SearchAction',
3072 + 'target' => [
3073 + '@type' => 'EntryPoint',
3074 + 'urlTemplate' => home_url('/?s={search_term_string}'),
3075 + ],
3076 + 'query-input' => 'required name=search_term_string',
3077 + ];
3078 + }
3079 +
3080 + return $schema;
2082 3081 }
2083 3082
2084 3083 /**
2085 3084 * Generate organization schema markup
@@ -2275,8 +3274,15 @@
2275 3274 if (!$this->site_identity_data || !$this->site_identity_data['enabled']) {
2276 3275 return;
2277 3276 }
2278 3277
3278 + // A breadcrumb trail for a URL that does not exist, or for a search
3279 + // results page, describes nothing — and the plugin already emits no
3280 + // canonical on either (#471).
3281 + if (is_404() || is_search()) {
3282 + return;
3283 + }
3284 +
2279 3285 $settings = $this->site_identity_manager->get_settings('site');
2280 3286
2281 3287 // Only output if breadcrumbs are enabled
2282 3288 if (empty($settings['breadcrumbs_enabled'])) {
@@ -2282,42 +3288,74 @@
2282 3288 if (empty($settings['breadcrumbs_enabled'])) {
2283 3289 return;
2284 3290 }
2285 3291
3292 + // Schema Manager's own breadcrumb switch. Only Site Identity's
3293 + // breadcrumbs_enabled was consulted here, so enable_breadcrumbs_schema
3294 + // = 0 removed a deployed BreadcrumbList row and left this live one
3295 + // emitting the node anyway (#688). Absent means not configured, which
3296 + // stays enabled.
3297 + if ($this->schema_manager) {
3298 + $schema_settings = $this->schema_manager->get_settings('site', null);
3299 +
3300 + if (array_key_exists('enable_breadcrumbs_schema', $schema_settings)
3301 + && empty($schema_settings['enable_breadcrumbs_schema'])) {
3302 + return;
3303 + }
3304 + }
3305 +
2286 3306 $breadcrumbs = $this->generate_breadcrumbs($settings);
2287 3307
2288 3308 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";
3309 + Schema_Graph::instance()->add_supporting($breadcrumbs['schema'], 'BreadcrumbList');
2294 3310 }
2295 3311 }
2296 3312
2297 3313 /**
3314 + * Emit everything ThinkRank collected for this request as one linked @graph.
3315 + *
3316 + * Runs after every producer has registered (site schema 7, breadcrumbs 8,
3317 + * Global SEO 15), so the graph can arbitrate between them.
3318 + *
3319 + * @since 1.32.0
3320 + * @return void
3321 + */
3322 + public function output_schema_graph(): void {
3323 + Schema_Graph::instance()->render();
3324 + }
3325 +
3326 + /**
2298 3327 * Output closing comment for ThinkRank SEO
2299 3328 *
2300 3329 * @return void
2301 3330 */
2302 3331 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 || $this->has_seo_output()) {
3332 + // Close only what was actually opened. has_seo_output() is true on
3333 + // nearly every page, so testing it here printed a closing comment with
3334 + // no matching opener whenever the meta description was empty (search
3335 + // results, author archives without a description).
3336 + if (self::$opening_comment_output) {
2306 3337 echo "<!-- /ThinkRank SEO -->\n";
2307 3338 }
2308 3339 }
2309 3340
2310 3341 /**
2311 - * Check if any SEO content has been output
3342 + * Print the opening ThinkRank comment, once per request.
2312 3343 *
2313 - * @return bool True if SEO content was output
3344 + * Public and static so Author_Archives_Manager — which prints its own meta
3345 + * description on wp_head at priority 5 — opens the block through the same
3346 + * flag the closing comment reads.
3347 + *
3348 + * @since 2.0.1
3349 + * @return void
2314 3350 */
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 && $this->site_identity_data['enabled']);
3351 + public static function note_opening_comment(): void {
3352 + if (self::$opening_comment_output) {
3353 + return;
3354 + }
3355 +
3356 + echo "<!-- Search Engine Optimization by ThinkRank - https://thinkrank.ai/ -->\n";
3357 + self::$opening_comment_output = true;
2320 3358 }
2321 3359
2322 3360 /**
2323 3361 * Display breadcrumbs HTML
@@ -2511,8 +3549,151 @@
2511 3549 return $breadcrumbs;
2512 3550 }
2513 3551
2514 3552 /**
3553 + * Serve /llms.txt through PHP so the response declares UTF-8.
3554 + *
3555 + * Cheap guard first: every other front-end request leaves without loading
3556 + * the manager.
3557 + *
3558 + * @since 1.32.0
3559 + *
3560 + * @return void
3561 + */
3562 + /**
3563 + * Serve a ThinkRank sitemap document for this request, when it is one.
3564 + *
3565 + * Only acts in dynamic delivery mode. In static mode a real file exists and
3566 + * the web server returns it without WordPress ever loading, so answering
3567 + * here as well would mean two sources for the same bytes.
3568 + *
3569 + * @since 2.9.0
3570 + *
3571 + * @return void
3572 + */
3573 + public function maybe_serve_sitemap(): void {
3574 + $filename = $this->requested_sitemap_filename();
3575 + if ('' === $filename) {
3576 + return;
3577 + }
3578 +
3579 + try {
3580 + // Read-only instance: passing false keeps it from registering a
3581 + // second copy of the auto-generation hooks.
3582 + $generator = new \ThinkRank\SEO\Sitemap_Generator(false);
3583 + $settings = $generator->get_settings('site');
3584 +
3585 + if (empty($settings['enabled'])) {
3586 + return;
3587 + }
3588 +
3589 + if ('dynamic' !== $generator->resolve_delivery_mode($settings)) {
3590 + return;
3591 + }
3592 +
3593 + if (!$generator->publishes_document_name($filename, $settings)) {
3594 + return;
3595 + }
3596 +
3597 + $xml = $generator->render_document($filename, $settings);
3598 + } catch (\Throwable $e) {
3599 + // A failed render must not replace the sitemap with a fatal. Leave
3600 + // the request alone so WordPress answers as it otherwise would.
3601 + return;
3602 + }
3603 +
3604 + if (!is_string($xml) || '' === trim($xml)) {
3605 + return;
3606 + }
3607 +
3608 + status_header(200);
3609 + header('Content-Type: application/xml; charset=UTF-8');
3610 + header('X-Robots-Tag: noindex, follow', true);
3611 +
3612 + // Built XML, escaped by the builders as they assemble it; escaping the
3613 + // document here would corrupt it.
3614 + echo $xml; // phpcs:ignore WordPress.Security.EscapeOutput.OutputNotEscaped
3615 + exit;
3616 + }
3617 +
3618 + /**
3619 + * The sitemap file name this request is asking for, if it looks like one.
3620 + *
3621 + * Deliberately a cheap shape test. Whether the site actually publishes the
3622 + * name is settled by the caller against the generator, so that a request
3623 + * for someone else's sitemap is never answered here.
3624 + *
3625 + * @since 2.9.0
3626 + *
3627 + * @return string File name, or '' when this is not a sitemap request.
3628 + */
3629 + private function requested_sitemap_filename(): string {
3630 + if (empty($_SERVER['REQUEST_URI'])) {
3631 + return '';
3632 + }
3633 +
3634 + $path = wp_parse_url(sanitize_text_field(wp_unslash($_SERVER['REQUEST_URI'])), PHP_URL_PATH);
3635 + if (!is_string($path) || '' === $path) {
3636 + return '';
3637 + }
3638 +
3639 + // Strip the install's home path so subdirectory installs match too.
3640 + $home_path = (string) wp_parse_url(home_url('/'), PHP_URL_PATH);
3641 + if ('' !== $home_path && '/' !== $home_path && 0 === strpos($path, $home_path)) {
3642 + $path = substr($path, strlen($home_path));
3643 + }
3644 +
3645 + $candidate = strtolower(trim($path, '/'));
3646 +
3647 + // One path segment ending in .xml. Anything nested is not a file we
3648 + // publish to the web root.
3649 + if ('' === $candidate || strpos($candidate, '/') !== false) {
3650 + return '';
3651 + }
3652 +
3653 + return substr($candidate, -4) === '.xml' ? $candidate : '';
3654 + }
3655 +
3656 + public function maybe_serve_llms_txt(): void {
3657 + if (!$this->is_llms_txt_request()) {
3658 + return;
3659 + }
3660 +
3661 + if (!class_exists('ThinkRank\\SEO\\LLMs_Txt_Manager')) {
3662 + require_once THINKRANK_PLUGIN_DIR . 'includes/seo/class-llms-txt-manager.php';
3663 + }
3664 +
3665 + $manager = new \ThinkRank\SEO\LLMs_Txt_Manager();
3666 + $manager->serve_llms_txt();
3667 + }
3668 +
3669 + /**
3670 + * Whether the current request is for /llms.txt.
3671 + *
3672 + * @since 1.32.0
3673 + *
3674 + * @return bool
3675 + */
3676 + private function is_llms_txt_request(): bool {
3677 + if (empty($_SERVER['REQUEST_URI'])) {
3678 + return false;
3679 + }
3680 +
3681 + $path = wp_parse_url(sanitize_text_field(wp_unslash($_SERVER['REQUEST_URI'])), PHP_URL_PATH);
3682 + if (!is_string($path) || '' === $path) {
3683 + return false;
3684 + }
3685 +
3686 + // Strip the install's home path so subdirectory installs match too.
3687 + $home_path = (string) wp_parse_url(home_url('/'), PHP_URL_PATH);
3688 + if ('' !== $home_path && '/' !== $home_path && 0 === strpos($path, $home_path)) {
3689 + $path = substr($path, strlen($home_path));
3690 + }
3691 +
3692 + return 'llms.txt' === strtolower(trim($path, '/'));
3693 + }
3694 +
3695 + /**
2515 3696 * Filter WordPress robots.txt output
2516 3697 *
2517 3698 * @param string $output The default robots.txt output
2518 3699 * @param string $is_public Whether the site is public
@@ -2545,8 +3726,190 @@
2545 3726 return $output;
2546 3727 }
2547 3728
2548 3729 /**
3730 + * Disable WordPress core's sitemap while ThinkRank's sitemap is enabled.
3731 + *
3732 + * Prevents the site from publishing two competing sitemap indexes. Core's
3733 + * /wp-sitemap.xml is taken offline (it 404s) and, as a consequence, core
3734 + * stops adding its own "Sitemap:" directive to robots.txt — including on the
3735 + * paths where ThinkRank does not own the robots.txt output.
3736 + *
3737 + * Only ever turns core's sitemap *off*: when ThinkRank's sitemap is disabled
3738 + * the incoming value is returned untouched, so core (or another plugin
3739 + * filtering this) keeps whatever behaviour it already had.
3740 + *
3741 + * @since 1.31.0
3742 + *
3743 + * @param bool $enabled Whether core's sitemap functionality is enabled.
3744 + * @return bool Filtered value.
3745 + */
3746 + public function filter_wp_sitemaps_enabled($enabled): bool {
3747 + return $this->should_disable_core_sitemap() ? false : (bool) $enabled;
3748 + }
3749 +
3750 + /**
3751 + * Redirect the sitemap URLs core owns to the sitemap ThinkRank publishes.
3752 + *
3753 + * Only the *index* route is redirected. Core's per-type children
3754 + * (/wp-sitemap-posts-post-1.xml and friends) are genuinely gone once core is
3755 + * switched off, and a 404 is the honest answer for those; the index is the
3756 + * one URL crawlers and humans actually guess, and the one core's own
3757 + * /sitemap.xml rule funnels into.
3758 + *
3759 + * @since 1.31.0
3760 + *
3761 + * @return void
3762 + */
3763 + public function redirect_core_sitemap_requests(): void {
3764 + if ('index' !== get_query_var('sitemap')) {
3765 + return;
3766 + }
3767 +
3768 + $request_uri = isset($_SERVER['REQUEST_URI'])
3769 + ? sanitize_text_field(wp_unslash($_SERVER['REQUEST_URI']))
3770 + : '';
3771 +
3772 + $target = $this->resolve_core_sitemap_redirect(
3773 + (string) wp_parse_url($request_uri, PHP_URL_PATH)
3774 + );
3775 +
3776 + if ('' === $target) {
3777 + return;
3778 + }
3779 +
3780 + wp_safe_redirect($target, 301, 'ThinkRank');
3781 + exit;
3782 + }
3783 +
3784 + /**
3785 + * Where a request for one of core's sitemap URLs should be sent, if anywhere.
3786 + *
3787 + * Split out from the hook so the rules are testable without dispatching a
3788 + * request — the caller above is the only part that cannot be (it exits).
3789 + *
3790 + * @since 1.31.0
3791 + *
3792 + * @param string $requested_path Path of the incoming request.
3793 + * @return string Absolute URL to redirect to, or '' to leave the request alone.
3794 + */
3795 + private function resolve_core_sitemap_redirect(string $requested_path): string {
3796 + // Nothing to redirect to unless we have actually taken core offline,
3797 + // which already implies our own sitemap file is on disk.
3798 + if (!$this->should_disable_core_sitemap()) {
3799 + return '';
3800 + }
3801 +
3802 + $target = $this->thinkrank_sitemap_url;
3803 + if ('' === $target) {
3804 + return '';
3805 + }
3806 +
3807 + // Never redirect a URL to itself. A site publishing at /sitemap.xml
3808 + // normally has the web server serve that file before WordPress sees the
3809 + // request, but on a setup where the request does reach PHP this is the
3810 + // difference between a redirect and a loop.
3811 + $destination = (string) wp_parse_url($target, PHP_URL_PATH);
3812 +
3813 + if ('' !== $requested_path && untrailingslashit($requested_path) === untrailingslashit($destination)) {
3814 + return '';
3815 + }
3816 +
3817 + return $target;
3818 + }
3819 +
3820 + /**
3821 + * Whether core's sitemap should be switched off for this site.
3822 + *
3823 + * True when ThinkRank publishes its own sitemap — except in two cases where
3824 + * taking core offline would leave a URL answering nothing:
3825 + *
3826 + * 1. The site is configured to publish *at core's own URL* (the "WordPress
3827 + * Core" preset). Once the static file exists the web server serves it
3828 + * ahead of WordPress anyway, so core can be left alone.
3829 + * 2. ThinkRank's own sitemap file is not on disk yet. Sitemaps here are
3830 + * static files with no dynamic route (see save_sitemap_to_file()), so
3831 + * while the file is missing core's /sitemap.xml -> /wp-sitemap.xml
3832 + * redirect is the only thing answering that URL; suppressing core would
3833 + * turn a recoverable "enabled but not generated" state into a hard 404
3834 + * for crawlers. Core is taken offline as soon as our file appears, so the
3835 + * duplicate-index conflict this filter exists to prevent cannot occur —
3836 + * two indexes are only ever reachable if both are actually published.
3837 + *
3838 + * Once our file does exist, the URLs core stops answering are handed to
3839 + * redirect_core_sitemap_requests() rather than left to 404 — which is why
3840 + * this also resolves the destination.
3841 + *
3842 + * Resolved lazily and memoised: this is consulted from an `init`-time filter
3843 + * on every request, and the underlying settings read is object-cached. The
3844 + * memoisation also keeps the file_exists() call to one per request.
3845 + *
3846 + * @since 1.31.0
3847 + *
3848 + * @return bool True when WordPress core's sitemap should be disabled.
3849 + */
3850 + private function should_disable_core_sitemap(): bool {
3851 + if ($this->thinkrank_sitemap_enabled === null) {
3852 + try {
3853 + // Read-only instance — passing false keeps it from registering a
3854 + // second copy of the save_post/term auto-generation hooks.
3855 + $generator = new \ThinkRank\SEO\Sitemap_Generator(false);
3856 + $settings = $generator->get_settings('site');
3857 +
3858 + // "Can ThinkRank actually answer its sitemap URL right now?" In
3859 + // static mode that means the file is on disk; in dynamic mode
3860 + // maybe_serve_sitemap() answers it, so there is nothing to look
3861 + // for. Keeping the file test as the only answer would have left
3862 + // core's sitemap in place on every dynamic site, which is the
3863 + // crawl conflict this suppression exists to prevent (#752).
3864 + // The #346 behaviour is unchanged: a static site with nothing
3865 + // published still falls through to core rather than 404ing.
3866 + $can_serve = 'dynamic' === $generator->resolve_delivery_mode($settings)
3867 + || $generator->primary_sitemap_file_exists($settings);
3868 +
3869 + $this->thinkrank_sitemap_enabled = !empty($settings['enabled'])
3870 + && !$this->publishes_at_core_sitemap_url($settings)
3871 + && $can_serve;
3872 +
3873 + if ($this->thinkrank_sitemap_enabled) {
3874 + $this->thinkrank_sitemap_url = $generator->get_primary_sitemap_url($settings);
3875 + }
3876 + } catch (\Exception $e) {
3877 + // Settings unreadable — leave core's sitemap alone rather than
3878 + // removing a working sitemap on the strength of a failed read.
3879 + $this->thinkrank_sitemap_enabled = false;
3880 + $this->thinkrank_sitemap_url = '';
3881 + }
3882 + }
3883 +
3884 + return $this->thinkrank_sitemap_enabled;
3885 + }
3886 +
3887 + /**
3888 + * Whether any configured sitemap URL is WordPress core's own wp-sitemap.xml.
3889 + *
3890 + * @since 1.31.0
3891 + *
3892 + * @param array $settings Sitemap settings.
3893 + * @return bool True when the site publishes at core's sitemap URL.
3894 + */
3895 + private function publishes_at_core_sitemap_url(array $settings): bool {
3896 + foreach ((array) ($settings['sitemap_urls'] ?? []) as $sitemap) {
3897 + if (!is_array($sitemap)) {
3898 + continue;
3899 + }
3900 +
3901 + $path = (string) wp_parse_url((string) ($sitemap['url'] ?? ''), PHP_URL_PATH);
3902 +
3903 + if (ltrim($path, '/') === 'wp-sitemap.xml') {
3904 + return true;
3905 + }
3906 + }
3907 +
3908 + return false;
3909 + }
3910 +
3911 + /**
2549 3912 * Re-sync the physical robots.txt when WordPress's "Discourage search
2550 3913 * engines" setting (blog_public) changes.
2551 3914 *
2552 3915 * Only acts when ThinkRank robots management is enabled AND a physical
@@ -2596,15 +3959,17 @@
2596 3959 if (empty($settings['enabled'])) {
2597 3960 return (string) $url;
2598 3961 }
2599 3962
3963 + $size = (int) $size;
3964 +
2600 3965 // Apple touch icon has its own dedicated setting
2601 - if ((int) $size === 180 && !empty($settings['apple_touch_icon_url'])) {
2602 - return esc_url($settings['apple_touch_icon_url']);
3966 + if ($size === 180 && !empty($settings['apple_touch_icon_url'])) {
3967 + return $this->resolve_icon_url((string) $settings['apple_touch_icon_url'], $size);
2603 3968 }
2604 3969
2605 3970 if (!empty($settings['favicon_url'])) {
2606 - return esc_url($settings['favicon_url']);
3971 + return $this->resolve_icon_url((string) $settings['favicon_url'], $size);
2607 3972 }
2608 3973
2609 3974 return (string) $url;
2610 3975 }
@@ -2609,8 +3974,178 @@
2609 3974 return (string) $url;
2610 3975 }
2611 3976
2612 3977 /**
3978 + * Whether breadcrumb labels should prefer the SEO title.
3979 + *
3980 + * Off unless the site turns it on, so updating the plugin never rewrites an
3981 + * existing trail.
3982 + *
3983 + * @since 2.3.1
3984 + *
3985 + * @param array $settings Breadcrumb settings.
3986 + * @return bool
3987 + */
3988 + private function breadcrumbs_use_seo_title(array $settings): bool {
3989 + return !empty($settings['breadcrumb_use_seo_title']);
3990 + }
3991 +
3992 + /**
3993 + * Label for a post in the breadcrumb trail.
3994 + *
3995 + * With the toggle on, the post's own SEO title wins — the same
3996 + * `_thinkrank_seo_title` value (variable tags resolved) the document title
3997 + * uses — so the trail under a search snippet reads the same as the snippet
3998 + * itself. Anything empty falls back to the raw post title; the global title
3999 + * pattern is deliberately NOT part of the chain, since resolving it would
4000 + * append the site name to every crumb.
4001 + *
4002 + * @since 2.3.1
4003 + *
4004 + * @param int $post_id Post ID.
4005 + * @param array $settings Breadcrumb settings.
4006 + * @return string Breadcrumb label.
4007 + */
4008 + private function get_breadcrumb_post_title(int $post_id, array $settings): string {
4009 + $title = (string) get_the_title($post_id);
4010 +
4011 + if (!$this->breadcrumbs_use_seo_title($settings)) {
4012 + return $title;
4013 + }
4014 +
4015 + $seo_title = trim((string) get_post_meta($post_id, '_thinkrank_seo_title', true));
4016 +
4017 + if ('' === $seo_title) {
4018 + return $title;
4019 + }
4020 +
4021 + $resolved = trim(\ThinkRank\SEO\Pattern_Resolver::resolve_value($seo_title, $post_id));
4022 +
4023 + return '' !== $resolved ? $resolved : $title;
4024 + }
4025 +
4026 + /**
4027 + * Label for a term in the breadcrumb trail.
4028 + *
4029 + * Term counterpart to {@see self::get_breadcrumb_post_title()}, resolving
4030 + * the term's `_thinkrank_seo_title` against its own values.
4031 + *
4032 + * @since 2.3.1
4033 + *
4034 + * @param object $term Term object.
4035 + * @param array $settings Breadcrumb settings.
4036 + * @return string Breadcrumb label.
4037 + */
4038 + private function get_breadcrumb_term_title($term, array $settings): string {
4039 + $name = (string) ($term->name ?? '');
4040 +
4041 + if (!$this->breadcrumbs_use_seo_title($settings) || empty($term->term_id)) {
4042 + return $name;
4043 + }
4044 +
4045 + $seo_title = trim((string) get_term_meta((int) $term->term_id, '_thinkrank_seo_title', true));
4046 +
4047 + if ('' === $seo_title) {
4048 + return $name;
4049 + }
4050 +
4051 + $resolved = trim(\ThinkRank\SEO\Pattern_Resolver::resolve_term_value($seo_title, (int) $term->term_id));
4052 +
4053 + return '' !== $resolved ? $resolved : $name;
4054 + }
4055 +
4056 + /**
4057 + * Resolve a configured icon URL to the derivative that fits $size.
4058 + *
4059 + * wp_site_icon() calls get_site_icon_url() four times — 32, 192, 180 and
4060 + * 270 — and pairs the first two with a hardcoded sizes="" attribute. This
4061 + * filter used to answer all four with the same configured URL, so one
4062 + * upload was declared as every size at once: a 1536x1536 original served
4063 + * to paint a 32px tab icon, under a sizes="32x32" label that was simply
4064 + * untrue (#571).
4065 + *
4066 + * Resolution mirrors core's own get_site_icon_url(), including the
4067 + * >= 512 -> 'full' branch, so ThinkRank's override and the core pipeline
4068 + * pick the same file for the same request.
4069 + *
4070 + * An unresolvable URL (one hosted off-site) is returned unchanged. Nothing
4071 + * is knowable about its dimensions, and suppressing it instead would leave
4072 + * the page with no rel="icon" at all — a worse outcome than an approximate
4073 + * size hint.
4074 + *
4075 + * @param string $configured Configured icon URL.
4076 + * @param int $size Icon size core is asking for.
4077 + * @return string Icon URL for that size.
4078 + */
4079 + private function resolve_icon_url(string $configured, int $size): string {
4080 + $cache_key = md5($configured) . ':' . $size;
4081 + $cached = $this->icon_urls();
4082 +
4083 + if (isset($cached[$cache_key])) {
4084 + return $cached[$cache_key];
4085 + }
4086 +
4087 + $attachment_id = \ThinkRank\SEO\Site_Identity_Manager::icon_attachment_id($configured);
4088 +
4089 + if (!$attachment_id) {
4090 + $resolved = esc_url($configured);
4091 + } else {
4092 + // Mirrors core: at 512 and above the original is what is wanted, and
4093 + // asking for an intermediate size that large would only fall back to it.
4094 + $size_data = $size >= 512 ? 'full' : [$size, $size];
4095 + $url = wp_get_attachment_image_url($attachment_id, $size_data);
4096 + $resolved = $url ? esc_url($url) : esc_url($configured);
4097 + }
4098 +
4099 + $this->icon_urls[$cache_key] = $resolved;
4100 +
4101 + if (!$this->icon_urls_dirty) {
4102 + $this->icon_urls_dirty = true;
4103 + // Written once, after the response is assembled, rather than once
4104 + // per size: wp_site_icon() resolves four in a row.
4105 + add_action('shutdown', [$this, 'persist_icon_urls'], 5);
4106 + }
4107 +
4108 + return $resolved;
4109 + }
4110 +
4111 + /**
4112 + * The resolved-icon-URL map, loaded from its transient on first use.
4113 + *
4114 + * @return array<string, string>
4115 + */
4116 + private function icon_urls(): array {
4117 + if ($this->icon_urls === null) {
4118 + $stored = get_transient(\ThinkRank\SEO\Site_Identity_Manager::ICON_URL_TRANSIENT);
4119 + $this->icon_urls = is_array($stored) ? $stored : [];
4120 + }
4121 +
4122 + return $this->icon_urls;
4123 + }
4124 +
4125 + /**
4126 + * Persist newly resolved icon URLs.
4127 + *
4128 + * Public because it runs on `shutdown`. Invalidated wholesale whenever the
4129 + * site identity settings are saved, which is the only moment the icon
4130 + * choice — or the derivatives behind it — can change.
4131 + *
4132 + * @return void
4133 + */
4134 + public function persist_icon_urls(): void {
4135 + if (!$this->icon_urls_dirty || !is_array($this->icon_urls)) {
4136 + return;
4137 + }
4138 +
4139 + $this->icon_urls_dirty = false;
4140 + set_transient(
4141 + \ThinkRank\SEO\Site_Identity_Manager::ICON_URL_TRANSIENT,
4142 + $this->icon_urls,
4143 + DAY_IN_SECONDS
4144 + );
4145 + }
4146 +
4147 + /**
2613 4148 * Get breadcrumb items for current page
2614 4149 *
2615 4150 * @param array $settings Breadcrumb settings
2616 4151 * @return array Breadcrumb items
@@ -2638,9 +4173,9 @@
2638 4173 $categories = get_the_category($current_post_id);
2639 4174 if (!empty($categories)) {
2640 4175 $category = $categories[0];
2641 4176 $items[] = [
2642 - 'title' => $category->name,
4177 + 'title' => $this->get_breadcrumb_term_title($category, $settings),
2643 4178 'url' => get_category_link($category->term_id),
2644 4179 'position' => $position++
2645 4180 ];
2646 4181 }
@@ -2645,12 +4180,18 @@
2645 4180 ];
2646 4181 }
2647 4182 }
2648 4183
2649 - // Add current post
2650 - if (empty($settings['show_current_page']) || $settings['show_current_page']) {
4184 + // Add current post. `empty($x) || $x` is true for every possible
4185 + // value — an unset key, false, 0, '' and any truthy value alike —
4186 + // so the setting had no effect on the rendered breadcrumb or on
4187 + // the BreadcrumbList JSON-LD, while the admin preview honoured it
4188 + // and disagreed with live output (#398). Site_Identity_Manager
4189 + // already had the correct form: default to on, respect an
4190 + // explicit off.
4191 + if ($settings['show_current_page'] ?? true) {
2651 4192 $items[] = [
2652 - 'title' => get_the_title($current_post_id),
4193 + 'title' => $this->get_breadcrumb_post_title($current_post_id, $settings),
2653 4194 'url' => get_permalink($current_post_id),
2654 4195 'position' => $position,
2655 4196 'current' => true
2656 4197 ];
@@ -2667,9 +4208,9 @@
2667 4208 while ($parent_id) {
2668 4209 $parent = get_post($parent_id);
2669 4210 if ($parent) {
2670 4211 $parents[] = [
2671 - 'title' => get_the_title($parent->ID),
4212 + 'title' => $this->get_breadcrumb_post_title($parent->ID, $settings),
2672 4213 'url' => get_permalink($parent->ID),
2673 4214 'position' => 0 // Will be set later
2674 4215 ];
2675 4216 $parent_id = $parent->post_parent;
@@ -2687,11 +4228,11 @@
2687 4228 $items[] = $parent;
2688 4229 }
2689 4230
2690 4231 // Add current page
2691 - if (empty($settings['show_current_page']) || $settings['show_current_page']) {
4232 + if ($settings['show_current_page'] ?? true) {
2692 4233 $items[] = [
2693 - 'title' => get_the_title($current_post_id),
4234 + 'title' => $this->get_breadcrumb_post_title($current_post_id, $settings),
2694 4235 'url' => get_permalink($current_post_id),
2695 4236 'position' => $position,
2696 4237 'current' => true
2697 4238 ];
@@ -2707,9 +4248,9 @@
2707 4248 while ($parent_id) {
2708 4249 $parent = get_category($parent_id);
2709 4250 if ($parent && !is_wp_error($parent)) {
2710 4251 $parents[] = [
2711 - 'title' => $parent->name,
4252 + 'title' => $this->get_breadcrumb_term_title($parent, $settings),
2712 4253 'url' => get_category_link($parent->term_id),
2713 4254 'position' => 0 // Will be set later
2714 4255 ];
2715 4256 $parent_id = $parent->parent;
@@ -2727,11 +4268,11 @@
2727 4268 $items[] = $parent;
2728 4269 }
2729 4270
2730 4271 // Add current category
2731 - if (empty($settings['show_current_page']) || $settings['show_current_page']) {
4272 + if ($settings['show_current_page'] ?? true) {
2732 4273 $items[] = [
2733 - 'title' => $category->name,
4274 + 'title' => $this->get_breadcrumb_term_title($category, $settings),
2734 4275 'url' => get_category_link($category->term_id),
2735 4276 'position' => $position,
2736 4277 'current' => true
2737 4278 ];