| @@ -266,8 +266,16 @@ | ||
| 266 | 266 | // LLMs_Txt_Manager for the static file) guarantees an explicit UTF-8 |
| 267 | 267 | // charset. Priority 8 keeps it ahead of redirect_canonical(). |
| 268 | 268 | add_action('template_redirect', [$this, 'maybe_serve_llms_txt'], 8); |
| 269 | 269 | |
| 270 | + // Serve the sitemap from PHP on sites whose web root cannot be written. | |
| 271 | + // ThinkRank publishes sitemaps as real files, so where that is possible | |
| 272 | + // the web server answers first and this never runs; where it is not, | |
| 273 | + // this is the only thing that answers at all, and without it the | |
| 274 | + // feature was simply unavailable (#752). Same priority 8, and for the | |
| 275 | + // same reason: ahead of redirect_canonical(). | |
| 276 | + add_action('template_redirect', [$this, 'maybe_serve_sitemap'], 8); | |
| 277 | + | |
| 270 | 278 | // Take WordPress core's own sitemap offline while ThinkRank's is active. |
| 271 | 279 | // Two sitemap indexes on one site is a crawl conflict: core keeps |
| 272 | 280 | // /wp-sitemap.xml served and injects its own "Sitemap:" line into |
| 273 | 281 | // robots.txt (WP_Sitemaps::add_robots, priority 0). Until now that line |
| @@ -369,8 +377,16 @@ | ||
| 369 | 377 | } |
| 370 | 378 | |
| 371 | 379 | // Initialize Global SEO Schema Output and store reference |
| 372 | 380 | $this->global_seo_schema = new Global_SEO_Schema_Output(); |
| 381 | + // Let schema reuse the description this class already resolves, so the | |
| 382 | + // JSON-LD and the meta/og/twitter tags cannot disagree about what the | |
| 383 | + // page is (#766). Passed as a callback rather than a value: schema is | |
| 384 | + // built during wp_head, by which point the request context this | |
| 385 | + // resolution depends on is set, and it must not be captured earlier. | |
| 386 | + $this->global_seo_schema->set_description_resolver( | |
| 387 | + fn (): string => (string) $this->get_meta_description() | |
| 388 | + ); | |
| 373 | 389 | $this->global_seo_schema->init(); |
| 374 | 390 | } |
| 375 | 391 | |
| 376 | 392 | /** |
| @@ -691,9 +707,86 @@ | ||
| 691 | 707 | * |
| 692 | 708 | * @param string $title Resolved title. |
| 693 | 709 | * @return string Title with the page indicator, when there is one. |
| 694 | 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 | + * `<em>` 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 | + | |
| 695 | 786 | public static function with_page_suffix(string $title): string { |
| 787 | + $title = self::strip_title_tags($title); | |
| 788 | + | |
| 696 | 789 | $page = self::current_page_number(); |
| 697 | 790 | |
| 698 | 791 | if ($page <= 1 || '' === $title) { |
| 699 | 792 | return $title; |
| @@ -784,11 +877,13 @@ | ||
| 784 | 877 | // Output main ThinkRank SEO header comment (only once) |
| 785 | 878 | self::note_opening_comment(); |
| 786 | 879 | |
| 787 | 880 | // Ensure description is within optimal length (150-160 characters) |
| 788 | - if (strlen($description) > 160) { | |
| 789 | - $description = wp_trim_words($description, 25, '...'); | |
| 790 | - } | |
| 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); | |
| 791 | 886 | |
| 792 | 887 | echo "<!-- ThinkRank SEO Meta Description -->\n"; |
| 793 | 888 | echo '<meta name="description" content="' . esc_attr($description) . '" />' . "\n"; |
| 794 | 889 | echo "<!-- /ThinkRank SEO Meta Description -->\n"; |
| @@ -1300,9 +1395,9 @@ | ||
| 1300 | 1395 | * |
| 1301 | 1396 | * @param array $og_tags Open Graph tags array |
| 1302 | 1397 | * @return void |
| 1303 | 1398 | */ |
| 1304 | - private function output_social_og_tags(array $og_tags): void { | |
| 1399 | + private function output_social_og_tags(array $og_tags, array $extra_images = []): void { | |
| 1305 | 1400 | // Honor the thinkrank_og_type filter here too — this "Enhanced" path is |
| 1306 | 1401 | // the active OG emitter, so add-ons (e.g. Pro's WooCommerce module which |
| 1307 | 1402 | // sets 'product' on product pages) must be applied to it, not only to |
| 1308 | 1403 | // output_open_graph_tags(). |
| @@ -1346,12 +1441,62 @@ | ||
| 1346 | 1441 | echo '<meta property="' . esc_attr($property) . '" content="' . $this->esc_meta_value($property, $content) . '" />' . "\n"; |
| 1347 | 1442 | } |
| 1348 | 1443 | } |
| 1349 | 1444 | |
| 1445 | + // Alternatives, after the primary and everything belonging to it. | |
| 1446 | + // Order is the whole point: a consumer reads og:image tags in document | |
| 1447 | + // order and treats the first as primary, and a structured property | |
| 1448 | + // attaches to the most recently declared image — so each alternative's | |
| 1449 | + // companions have to follow its own URL, not be grouped at the end. | |
| 1450 | + self::output_extra_og_images($extra_images); | |
| 1451 | + | |
| 1350 | 1452 | echo "<!-- /ThinkRank SEO Open Graph Tags -->\n"; |
| 1351 | 1453 | } |
| 1352 | 1454 | |
| 1353 | 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 | + /** | |
| 1354 | 1499 | * Output social media Twitter Card tags from Social Meta Manager |
| 1355 | 1500 | * |
| 1356 | 1501 | * @param array $twitter_tags Twitter Card tags array |
| 1357 | 1502 | * @return void |
| @@ -1514,9 +1659,12 @@ | ||
| 1514 | 1659 | // The Social Meta Manager ran, so it owns Open Graph output. If OG is |
| 1515 | 1660 | // toggled off, emit nothing — do NOT fall through to the basic |
| 1516 | 1661 | // emitter (which would re-add a full OG block despite the toggle). |
| 1517 | 1662 | if (!empty($social_data['og_enabled'])) { |
| 1518 | - $this->output_social_og_tags($social_data['og_tags']); | |
| 1663 | + $this->output_social_og_tags( | |
| 1664 | + $social_data['og_tags'], | |
| 1665 | + $social_data['og_extra_images'] ?? [] | |
| 1666 | + ); | |
| 1519 | 1667 | } |
| 1520 | 1668 | return; |
| 1521 | 1669 | } |
| 1522 | 1670 | |
| @@ -1584,9 +1732,9 @@ | ||
| 1584 | 1732 | // "There is no excerpt because this is a protected post." placeholder, |
| 1585 | 1733 | // so this is not a leak — but publishing that sentence as the social |
| 1586 | 1734 | // description is worse than publishing none (#363). |
| 1587 | 1735 | if (!$description && !$this->is_content_password_protected()) { |
| 1588 | - $description = is_singular() ? wp_trim_words(get_the_excerpt(), 30) : get_bloginfo('description'); | |
| 1736 | + $description = is_singular() ? \ThinkRank\Core\Seo_Text::trim_words(get_the_excerpt(), 30) : get_bloginfo('description'); | |
| 1589 | 1737 | } |
| 1590 | 1738 | |
| 1591 | 1739 | $url = is_singular() ? get_permalink() : home_url(); |
| 1592 | 1740 | $site_name = $this->site_identity_data && !empty($this->site_identity_data['identity']['site_name']) |
| @@ -1614,11 +1762,11 @@ | ||
| 1614 | 1762 | $og_type = apply_filters('thinkrank_og_type', $og_type); |
| 1615 | 1763 | |
| 1616 | 1764 | echo "<!-- ThinkRank SEO Open Graph Meta Tags -->\n"; |
| 1617 | 1765 | echo "<meta property=\"og:type\" content=\"" . esc_attr($og_type) . "\" />\n"; |
| 1618 | - 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"; | |
| 1619 | 1767 | echo "<meta property=\"og:description\" content=\"" . esc_attr($description) . "\" />\n"; |
| 1620 | - echo "<meta property=\"og:url\" content=\"" . esc_url($url) . "\" />\n"; | |
| 1768 | + echo "<meta property=\"og:url\" content=\"" . esc_url(\ThinkRank\SEO\Url_Scheme::apply($url)) . "\" />\n"; | |
| 1621 | 1769 | echo "<meta property=\"og:site_name\" content=\"" . esc_attr($site_name) . "\" />\n"; |
| 1622 | 1770 | /** |
| 1623 | 1771 | * Filter the og:locale value. |
| 1624 | 1772 | * |
| @@ -1635,14 +1783,17 @@ | ||
| 1635 | 1783 | $og_locale = (string) apply_filters('thinkrank_og_locale', get_locale()); |
| 1636 | 1784 | echo "<meta property=\"og:locale\" content=\"" . esc_attr($og_locale) . "\" />\n"; |
| 1637 | 1785 | |
| 1638 | 1786 | // Add OG image — per-post override > featured image |
| 1787 | + $primary_og_image = ''; | |
| 1639 | 1788 | if (is_singular() && $this->current_post_id) { |
| 1640 | 1789 | if (!empty($og_image_override)) { |
| 1790 | + $primary_og_image = (string) $og_image_override; | |
| 1641 | 1791 | echo "<meta property=\"og:image\" content=\"" . esc_url($og_image_override) . "\" />\n"; |
| 1642 | 1792 | echo "<meta property=\"og:image:secure_url\" content=\"" . esc_url($og_image_override) . "\" />\n"; |
| 1643 | 1793 | } elseif (has_post_thumbnail($this->current_post_id)) { |
| 1644 | 1794 | $image_url = get_the_post_thumbnail_url($this->current_post_id, 'large'); |
| 1795 | + $primary_og_image = (string) $image_url; | |
| 1645 | 1796 | echo "<meta property=\"og:image\" content=\"" . esc_url($image_url) . "\" />\n"; |
| 1646 | 1797 | echo "<meta property=\"og:image:secure_url\" content=\"" . esc_url($image_url) . "\" />\n"; |
| 1647 | 1798 | |
| 1648 | 1799 | // Get image dimensions and alt text |
| @@ -1648,12 +1799,16 @@ | ||
| 1648 | 1799 | // Get image dimensions and alt text |
| 1649 | 1800 | $image_id = get_post_thumbnail_id($this->current_post_id); |
| 1650 | 1801 | $image_meta = wp_get_attachment_metadata($image_id); |
| 1651 | 1802 | if ($image_meta) { |
| 1803 | + // The `large` file being published, not the original it | |
| 1804 | + // was generated from: the metadata's own width and height | |
| 1805 | + // describe an image this tag does not point at (#847). | |
| 1806 | + $image_file = \ThinkRank\SEO\Attachment_Lookup::describe((int) $image_id, (string) $image_url); | |
| 1652 | 1807 | // SVGs (and other vector uploads) report 0x0 — emitting |
| 1653 | 1808 | // those as og:image dimensions is invalid, so skip them. |
| 1654 | - $og_width = isset($image_meta['width']) ? (int) $image_meta['width'] : 0; | |
| 1655 | - $og_height = isset($image_meta['height']) ? (int) $image_meta['height'] : 0; | |
| 1809 | + $og_width = $image_file['width']; | |
| 1810 | + $og_height = $image_file['height']; | |
| 1656 | 1811 | if ($og_width > 0 && $og_height > 0) { |
| 1657 | 1812 | echo "<meta property=\"og:image:width\" content=\"" . esc_attr($og_width) . "\" />\n"; |
| 1658 | 1813 | echo "<meta property=\"og:image:height\" content=\"" . esc_attr($og_height) . "\" />\n"; |
| 1659 | 1814 | } |
| @@ -1658,9 +1813,9 @@ | ||
| 1658 | 1813 | echo "<meta property=\"og:image:height\" content=\"" . esc_attr($og_height) . "\" />\n"; |
| 1659 | 1814 | } |
| 1660 | 1815 | // Derive the real mime type instead of hardcoding image/jpeg, |
| 1661 | 1816 | // which mislabels PNG/WebP featured images. |
| 1662 | - $image_mime = get_post_mime_type($image_id); | |
| 1817 | + $image_mime = $image_file['type']; | |
| 1663 | 1818 | if ($image_mime) { |
| 1664 | 1819 | echo "<meta property=\"og:image:type\" content=\"" . esc_attr($image_mime) . "\" />\n"; |
| 1665 | 1820 | } |
| 1666 | 1821 | } |
| @@ -1671,8 +1826,30 @@ | ||
| 1671 | 1826 | echo "<meta property=\"og:image:alt\" content=\"" . esc_attr($image_alt) . "\" />\n"; |
| 1672 | 1827 | } |
| 1673 | 1828 | } |
| 1674 | 1829 | |
| 1830 | + // Alternatives, same as the enhanced emitter above. This path only | |
| 1831 | + // runs when the Social Meta Manager is unavailable, but the issue | |
| 1832 | + // reported against it (#636) and a site that lands here should not | |
| 1833 | + // silently lose a feature it switched on. | |
| 1834 | + if (!empty($primary_og_image)) { | |
| 1835 | + $social_settings = $this->social_manager | |
| 1836 | + ? $this->social_manager->get_settings( | |
| 1837 | + $this->current_context === 'homepage' ? 'site' : $this->current_context, | |
| 1838 | + $this->current_post_id | |
| 1839 | + ) | |
| 1840 | + : []; | |
| 1841 | + | |
| 1842 | + if (!empty($social_settings['og_multiple_images'])) { | |
| 1843 | + self::output_extra_og_images( | |
| 1844 | + \ThinkRank\SEO\Social_Images::additional( | |
| 1845 | + (int) $this->current_post_id, | |
| 1846 | + $primary_og_image | |
| 1847 | + ) | |
| 1848 | + ); | |
| 1849 | + } | |
| 1850 | + } | |
| 1851 | + | |
| 1675 | 1852 | // Add article specific tags for posts only |
| 1676 | 1853 | if ($og_type === 'article') { |
| 1677 | 1854 | echo '<meta property="article:published_time" content="' . esc_attr(get_the_date('c', $this->current_post_id)) . '" />' . "\n"; |
| 1678 | 1855 | echo '<meta property="article:modified_time" content="' . esc_attr(get_the_modified_date('c', $this->current_post_id)) . '" />' . "\n"; |
| @@ -1798,9 +1975,9 @@ | ||
| 1798 | 1975 | // "There is no excerpt because this is a protected post." placeholder, |
| 1799 | 1976 | // so this is not a leak — but publishing that sentence as the social |
| 1800 | 1977 | // description is worse than publishing none (#363). |
| 1801 | 1978 | if (!$description && !$this->is_content_password_protected()) { |
| 1802 | - $description = is_singular() ? wp_trim_words(get_the_excerpt(), 30) : get_bloginfo('description'); | |
| 1979 | + $description = is_singular() ? \ThinkRank\Core\Seo_Text::trim_words(get_the_excerpt(), 30) : get_bloginfo('description'); | |
| 1803 | 1980 | } |
| 1804 | 1981 | |
| 1805 | 1982 | // Determine card type based on image availability |
| 1806 | 1983 | $card_type = 'summary'; |
| @@ -1809,9 +1986,9 @@ | ||
| 1809 | 1986 | } |
| 1810 | 1987 | |
| 1811 | 1988 | echo "<!-- ThinkRank SEO Twitter Card Meta Tags -->\n"; |
| 1812 | 1989 | echo '<meta name="twitter:card" content="' . esc_attr($card_type) . '" />' . "\n"; |
| 1813 | - echo "<meta name=\"twitter:title\" content=\"" . esc_attr($title) . "\" />\n"; | |
| 1990 | + echo "<meta name=\"twitter:title\" content=\"" . esc_attr(self::strip_title_tags($title)) . "\" />\n"; | |
| 1814 | 1991 | echo "<meta name=\"twitter:description\" content=\"" . esc_attr($description) . "\" />\n"; |
| 1815 | 1992 | |
| 1816 | 1993 | // Add Twitter image with proper fallback priority |
| 1817 | 1994 | $twitter_image_url = $this->get_twitter_image_with_fallback(); |
| @@ -1886,8 +2063,13 @@ | ||
| 1886 | 2063 | if (empty($canonical_url)) { |
| 1887 | 2064 | return; |
| 1888 | 2065 | } |
| 1889 | 2066 | |
| 2067 | + // After the filter, so a canonical an add-on supplied is normalized | |
| 2068 | + // too — and a cross-domain one is left alone, since Url_Scheme only | |
| 2069 | + // touches URLs on this site's own host. | |
| 2070 | + $canonical_url = \ThinkRank\SEO\Url_Scheme::apply($canonical_url); | |
| 2071 | + | |
| 1890 | 2072 | echo "<!-- ThinkRank SEO Canonical URL -->\n"; |
| 1891 | 2073 | echo "<link rel=\"canonical\" href=\"" . esc_url($canonical_url) . "\" />\n"; |
| 1892 | 2074 | echo "<!-- /ThinkRank SEO Canonical URL -->\n"; |
| 1893 | 2075 | |
| @@ -1934,9 +2116,9 @@ | ||
| 1934 | 2116 | |
| 1935 | 2117 | if ($current > 1) { |
| 1936 | 2118 | printf( |
| 1937 | 2119 | "<link rel=\"prev\" href=\"%s\" />\n", |
| 1938 | - esc_url(self::with_pagination($base, $current - 1)) | |
| 2120 | + esc_url(\ThinkRank\SEO\Url_Scheme::apply(self::with_pagination($base, $current - 1))) | |
| 1939 | 2121 | ); |
| 1940 | 2122 | } |
| 1941 | 2123 | |
| 1942 | 2124 | if ($current < $total) { |
| @@ -1941,9 +2123,9 @@ | ||
| 1941 | 2123 | |
| 1942 | 2124 | if ($current < $total) { |
| 1943 | 2125 | printf( |
| 1944 | 2126 | "<link rel=\"next\" href=\"%s\" />\n", |
| 1945 | - esc_url(self::with_pagination($base, $current + 1)) | |
| 2127 | + esc_url(\ThinkRank\SEO\Url_Scheme::apply(self::with_pagination($base, $current + 1))) | |
| 1946 | 2128 | ); |
| 1947 | 2129 | } |
| 1948 | 2130 | } |
| 1949 | 2131 | |
| @@ -2443,9 +2625,9 @@ | ||
| 2443 | 2625 | // Stripped: get_the_archive_title() wraps its subject in a |
| 2444 | 2626 | // <span>, and this placeholder feeds the document <title> as |
| 2445 | 2627 | // well as og:title and twitter:title — a date archive rendered |
| 2446 | 2628 | // as "Month: <span>August 2026</span> | Site". |
| 2447 | - $placeholders['%archive_title%'] = wp_strip_all_tags((string) get_the_archive_title()); | |
| 2629 | + $placeholders['%archive_title%'] = self::archive_subject(); | |
| 2448 | 2630 | break; |
| 2449 | 2631 | |
| 2450 | 2632 | case 'homepage': |
| 2451 | 2633 | // The page template resolved for a static posts page needs the |
| @@ -2640,11 +2822,13 @@ | ||
| 2640 | 2822 | if ($description === '') { |
| 2641 | 2823 | return null; |
| 2642 | 2824 | } |
| 2643 | 2825 | |
| 2644 | - if (strlen($description) > 160) { | |
| 2645 | - $description = wp_trim_words($description, 25, '...'); | |
| 2646 | - } | |
| 2826 | + // Measure and cut in CHARACTERS. strlen() counts bytes, so a Thai or | |
| 2827 | + // CJK description tripped this limit at a third of its length, and | |
| 2828 | + // wp_trim_words() then cut by a unit the locale chooses — 25 words in | |
| 2829 | + // English, 25 characters in Thai (#687). | |
| 2830 | + $description = \ThinkRank\Core\Seo_Text::trim_to_length($description); | |
| 2647 | 2831 | |
| 2648 | 2832 | return $description; |
| 2649 | 2833 | } |
| 2650 | 2834 | |
| @@ -2691,11 +2875,13 @@ | ||
| 2691 | 2875 | $description = preg_replace('/\s+/', ' ', $description); |
| 2692 | 2876 | $description = trim($description); |
| 2693 | 2877 | |
| 2694 | 2878 | // Ensure description doesn't exceed recommended length (160 characters) |
| 2695 | - if (strlen($description) > 160) { | |
| 2696 | - $description = wp_trim_words($description, 25, '...'); | |
| 2697 | - } | |
| 2879 | + // Measure and cut in CHARACTERS. strlen() counts bytes, so a Thai or | |
| 2880 | + // CJK description tripped this limit at a third of its length, and | |
| 2881 | + // wp_trim_words() then cut by a unit the locale chooses — 25 words in | |
| 2882 | + // English, 25 characters in Thai (#687). | |
| 2883 | + $description = \ThinkRank\Core\Seo_Text::trim_to_length($description); | |
| 2698 | 2884 | |
| 2699 | 2885 | return $description; |
| 2700 | 2886 | } |
| 2701 | 2887 | |
| @@ -2780,10 +2966,22 @@ | ||
| 2780 | 2966 | $context_type, |
| 2781 | 2967 | $context_id |
| 2782 | 2968 | ); |
| 2783 | 2969 | |
| 2970 | + // A deployed node is a snapshot from Deploy time and outranks | |
| 2971 | + // the automatic node, so page and article types would publish | |
| 2972 | + // a frozen excerpt instead of the description the head | |
| 2973 | + // resolves. Give them the live one, as the automatic node has. | |
| 2974 | + $context_post = get_post($context_id); | |
| 2975 | + | |
| 2784 | 2976 | foreach ($page_specific_schemas as $schema_type => $schema_info) { |
| 2785 | - Schema_Graph::instance()->add_primary($schema_info['data'], (string) $schema_type, 'schema_manager'); | |
| 2977 | + $node = $schema_info['data']; | |
| 2978 | + | |
| 2979 | + if ($this->global_seo_schema && $context_post instanceof \WP_Post) { | |
| 2980 | + $node = $this->global_seo_schema->refresh_deployed_description($node, (string) $schema_type, $context_post); | |
| 2981 | + } | |
| 2982 | + | |
| 2983 | + Schema_Graph::instance()->add_primary($node, (string) $schema_type, 'schema_manager'); | |
| 2786 | 2984 | } |
| 2787 | 2985 | $has_schema_manager_output = true; |
| 2788 | 2986 | } |
| 2789 | 2987 | } |
| @@ -2841,21 +3039,49 @@ | ||
| 2841 | 3039 | 'url' => home_url('/'), |
| 2842 | 3040 | ]; |
| 2843 | 3041 | |
| 2844 | 3042 | $description = !empty($settings['site_description']) ? $settings['site_description'] : get_bloginfo('description'); |
| 3043 | + // The tagline is stored esc_html()'d by sanitize_option(), so a site | |
| 3044 | + // called "Fish & Chips" published `&` literally in its WebSite | |
| 3045 | + // node; nothing decodes JSON-LD downstream. | |
| 3046 | + $description = \ThinkRank\Core\Seo_Text::normalize_schema_text((string) $description); | |
| 2845 | 3047 | if (!empty($description)) { |
| 2846 | 3048 | $schema['description'] = $description; |
| 2847 | 3049 | } |
| 2848 | 3050 | |
| 2849 | - $schema['potentialAction'] = [ | |
| 2850 | - '@type' => 'SearchAction', | |
| 2851 | - 'target' => [ | |
| 2852 | - '@type' => 'EntryPoint', | |
| 2853 | - 'urlTemplate' => home_url('/?s={search_term_string}'), | |
| 2854 | - ], | |
| 2855 | - 'query-input' => 'required name=search_term_string', | |
| 2856 | - ]; | |
| 3051 | + // Site Identity has accepted an alternate name since the setup wizard | |
| 3052 | + // shipped, and the MCP ability describes it as "published as schema | |
| 3053 | + // alternateName" — but no producer ever read it, so the promise was | |
| 3054 | + // false and every imported Yoast/Rank Math value sat unused (#692). | |
| 3055 | + $alternate_name = \ThinkRank\SEO\Site_Identity_Manager::alternate_name_for_schema($settings['alternate_name'] ?? null); | |
| 3056 | + if (null !== $alternate_name) { | |
| 3057 | + $schema['alternateName'] = $alternate_name; | |
| 3058 | + } | |
| 2857 | 3059 | |
| 3060 | + // The sitelinks searchbox switch was honoured only for a deployed | |
| 3061 | + // WebSite row; this live fallback added potentialAction unconditionally, | |
| 3062 | + // so website_enable_search = 0 still shipped the SearchAction (#688). | |
| 3063 | + // Absent means not configured, which stays enabled. | |
| 3064 | + $search_enabled = true; | |
| 3065 | + if ($this->schema_manager) { | |
| 3066 | + $schema_settings = $this->schema_manager->get_settings('site', null); | |
| 3067 | + | |
| 3068 | + if (array_key_exists('website_enable_search', $schema_settings)) { | |
| 3069 | + $search_enabled = !empty($schema_settings['website_enable_search']); | |
| 3070 | + } | |
| 3071 | + } | |
| 3072 | + | |
| 3073 | + if ($search_enabled) { | |
| 3074 | + $schema['potentialAction'] = [ | |
| 3075 | + '@type' => 'SearchAction', | |
| 3076 | + 'target' => [ | |
| 3077 | + '@type' => 'EntryPoint', | |
| 3078 | + 'urlTemplate' => home_url('/?s={search_term_string}'), | |
| 3079 | + ], | |
| 3080 | + 'query-input' => 'required name=search_term_string', | |
| 3081 | + ]; | |
| 3082 | + } | |
| 3083 | + | |
| 2858 | 3084 | return $schema; |
| 2859 | 3085 | } |
| 2860 | 3086 | |
| 2861 | 3087 | /** |
| @@ -3066,8 +3292,22 @@ | ||
| 3066 | 3292 | if (empty($settings['breadcrumbs_enabled'])) { |
| 3067 | 3293 | return; |
| 3068 | 3294 | } |
| 3069 | 3295 | |
| 3296 | + // Schema Manager's own breadcrumb switch. Only Site Identity's | |
| 3297 | + // breadcrumbs_enabled was consulted here, so enable_breadcrumbs_schema | |
| 3298 | + // = 0 removed a deployed BreadcrumbList row and left this live one | |
| 3299 | + // emitting the node anyway (#688). Absent means not configured, which | |
| 3300 | + // stays enabled. | |
| 3301 | + if ($this->schema_manager) { | |
| 3302 | + $schema_settings = $this->schema_manager->get_settings('site', null); | |
| 3303 | + | |
| 3304 | + if (array_key_exists('enable_breadcrumbs_schema', $schema_settings) | |
| 3305 | + && empty($schema_settings['enable_breadcrumbs_schema'])) { | |
| 3306 | + return; | |
| 3307 | + } | |
| 3308 | + } | |
| 3309 | + | |
| 3070 | 3310 | $breadcrumbs = $this->generate_breadcrumbs($settings); |
| 3071 | 3311 | |
| 3072 | 3312 | if (!empty($breadcrumbs['schema'])) { |
| 3073 | 3313 | Schema_Graph::instance()->add_supporting($breadcrumbs['schema'], 'BreadcrumbList'); |
| @@ -3322,8 +3562,102 @@ | ||
| 3322 | 3562 | * @since 1.32.0 |
| 3323 | 3563 | * |
| 3324 | 3564 | * @return void |
| 3325 | 3565 | */ |
| 3566 | + /** | |
| 3567 | + * Serve a ThinkRank sitemap document for this request, when it is one. | |
| 3568 | + * | |
| 3569 | + * Only acts in dynamic delivery mode. In static mode a real file exists and | |
| 3570 | + * the web server returns it without WordPress ever loading, so answering | |
| 3571 | + * here as well would mean two sources for the same bytes. | |
| 3572 | + * | |
| 3573 | + * @since 2.9.0 | |
| 3574 | + * | |
| 3575 | + * @return void | |
| 3576 | + */ | |
| 3577 | + public function maybe_serve_sitemap(): void { | |
| 3578 | + $filename = $this->requested_sitemap_filename(); | |
| 3579 | + if ('' === $filename) { | |
| 3580 | + return; | |
| 3581 | + } | |
| 3582 | + | |
| 3583 | + try { | |
| 3584 | + // Read-only instance: passing false keeps it from registering a | |
| 3585 | + // second copy of the auto-generation hooks. | |
| 3586 | + $generator = new \ThinkRank\SEO\Sitemap_Generator(false); | |
| 3587 | + $settings = $generator->get_settings('site'); | |
| 3588 | + | |
| 3589 | + if (empty($settings['enabled'])) { | |
| 3590 | + return; | |
| 3591 | + } | |
| 3592 | + | |
| 3593 | + if ('dynamic' !== $generator->resolve_delivery_mode($settings)) { | |
| 3594 | + return; | |
| 3595 | + } | |
| 3596 | + | |
| 3597 | + if (!$generator->publishes_document_name($filename, $settings)) { | |
| 3598 | + return; | |
| 3599 | + } | |
| 3600 | + | |
| 3601 | + $xml = $generator->render_document($filename, $settings); | |
| 3602 | + } catch (\Throwable $e) { | |
| 3603 | + // A failed render must not replace the sitemap with a fatal. Leave | |
| 3604 | + // the request alone so WordPress answers as it otherwise would. | |
| 3605 | + return; | |
| 3606 | + } | |
| 3607 | + | |
| 3608 | + if (!is_string($xml) || '' === trim($xml)) { | |
| 3609 | + return; | |
| 3610 | + } | |
| 3611 | + | |
| 3612 | + status_header(200); | |
| 3613 | + header('Content-Type: application/xml; charset=UTF-8'); | |
| 3614 | + header('X-Robots-Tag: noindex, follow', true); | |
| 3615 | + | |
| 3616 | + // Built XML, escaped by the builders as they assemble it; escaping the | |
| 3617 | + // document here would corrupt it. | |
| 3618 | + echo $xml; // phpcs:ignore WordPress.Security.EscapeOutput.OutputNotEscaped | |
| 3619 | + exit; | |
| 3620 | + } | |
| 3621 | + | |
| 3622 | + /** | |
| 3623 | + * The sitemap file name this request is asking for, if it looks like one. | |
| 3624 | + * | |
| 3625 | + * Deliberately a cheap shape test. Whether the site actually publishes the | |
| 3626 | + * name is settled by the caller against the generator, so that a request | |
| 3627 | + * for someone else's sitemap is never answered here. | |
| 3628 | + * | |
| 3629 | + * @since 2.9.0 | |
| 3630 | + * | |
| 3631 | + * @return string File name, or '' when this is not a sitemap request. | |
| 3632 | + */ | |
| 3633 | + private function requested_sitemap_filename(): string { | |
| 3634 | + if (empty($_SERVER['REQUEST_URI'])) { | |
| 3635 | + return ''; | |
| 3636 | + } | |
| 3637 | + | |
| 3638 | + $path = wp_parse_url(sanitize_text_field(wp_unslash($_SERVER['REQUEST_URI'])), PHP_URL_PATH); | |
| 3639 | + if (!is_string($path) || '' === $path) { | |
| 3640 | + return ''; | |
| 3641 | + } | |
| 3642 | + | |
| 3643 | + // Strip the install's home path so subdirectory installs match too. | |
| 3644 | + $home_path = (string) wp_parse_url(home_url('/'), PHP_URL_PATH); | |
| 3645 | + if ('' !== $home_path && '/' !== $home_path && 0 === strpos($path, $home_path)) { | |
| 3646 | + $path = substr($path, strlen($home_path)); | |
| 3647 | + } | |
| 3648 | + | |
| 3649 | + $candidate = strtolower(trim($path, '/')); | |
| 3650 | + | |
| 3651 | + // One path segment ending in .xml. Anything nested is not a file we | |
| 3652 | + // publish to the web root. | |
| 3653 | + if ('' === $candidate || strpos($candidate, '/') !== false) { | |
| 3654 | + return ''; | |
| 3655 | + } | |
| 3656 | + | |
| 3657 | + return substr($candidate, -4) === '.xml' ? $candidate : ''; | |
| 3658 | + } | |
| 3659 | + | |
| 3326 | 3660 | public function maybe_serve_llms_txt(): void { |
| 3327 | 3661 | if (!$this->is_llms_txt_request()) { |
| 3328 | 3662 | return; |
| 3329 | 3663 | } |
| @@ -3524,11 +3858,22 @@ | ||
| 3524 | 3858 | // second copy of the save_post/term auto-generation hooks. |
| 3525 | 3859 | $generator = new \ThinkRank\SEO\Sitemap_Generator(false); |
| 3526 | 3860 | $settings = $generator->get_settings('site'); |
| 3527 | 3861 | |
| 3862 | + // "Can ThinkRank actually answer its sitemap URL right now?" In | |
| 3863 | + // static mode that means the file is on disk; in dynamic mode | |
| 3864 | + // maybe_serve_sitemap() answers it, so there is nothing to look | |
| 3865 | + // for. Keeping the file test as the only answer would have left | |
| 3866 | + // core's sitemap in place on every dynamic site, which is the | |
| 3867 | + // crawl conflict this suppression exists to prevent (#752). | |
| 3868 | + // The #346 behaviour is unchanged: a static site with nothing | |
| 3869 | + // published still falls through to core rather than 404ing. | |
| 3870 | + $can_serve = 'dynamic' === $generator->resolve_delivery_mode($settings) | |
| 3871 | + || $generator->primary_sitemap_file_exists($settings); | |
| 3872 | + | |
| 3528 | 3873 | $this->thinkrank_sitemap_enabled = !empty($settings['enabled']) |
| 3529 | 3874 | && !$this->publishes_at_core_sitemap_url($settings) |
| 3530 | - && $generator->primary_sitemap_file_exists($settings); | |
| 3875 | + && $can_serve; | |
| 3531 | 3876 | |
| 3532 | 3877 | if ($this->thinkrank_sitemap_enabled) { |
| 3533 | 3878 | $this->thinkrank_sitemap_url = $generator->get_primary_sitemap_url($settings); |
| 3534 | 3879 | } |
