| @@ -40,8 +40,36 @@ | ||
| 40 | 40 | */ |
| 41 | 41 | class Site_Identity_Manager extends Abstract_SEO_Manager { |
| 42 | 42 | |
| 43 | 43 | /** |
| 44 | + * The per-context title formats as ThinkRank ships them. | |
| 45 | + * | |
| 46 | + * These are not in get_default_settings(): the admin screen seeds them on | |
| 47 | + * first save, so on a real install they are stored values, indistinguishable | |
| 48 | + * from a template the user typed. The migration needs to tell those two | |
| 49 | + * apart — it may overwrite a shipped default with an imported template, and | |
| 50 | + * must never overwrite a choice the user made — so this is the record of | |
| 51 | + * what "untouched" looks like. | |
| 52 | + * | |
| 53 | + * Keep in step with getDefaultSettings() in | |
| 54 | + * src/admin/components/essential-seo/SiteIdentityTab.js. SiteIdentityTitleFormatDefaultsTest | |
| 55 | + * fails when the two drift. | |
| 56 | + * | |
| 57 | + * @since 2.8.0 | |
| 58 | + * @var array<string, string> | |
| 59 | + */ | |
| 60 | + public const TITLE_FORMAT_DEFAULTS = [ | |
| 61 | + 'homepage_title' => '%site_title% %sep% %site_description%', | |
| 62 | + 'post_title' => '%post_title% %sep% %site_title%', | |
| 63 | + 'page_title' => '%page_title% %sep% %site_title%', | |
| 64 | + 'category_title' => '%category_title% %sep% %site_title%', | |
| 65 | + 'tag_title' => '%tag_title% %sep% %site_title%', | |
| 66 | + 'author_title' => '%author_name% %sep% %site_title%', | |
| 67 | + 'search_title' => 'Search Results for "%search_term%" %sep% %site_title%', | |
| 68 | + 'archive_title' => '%archive_title% %sep% %site_title%', | |
| 69 | + ]; | |
| 70 | + | |
| 71 | + /** | |
| 44 | 72 | * WordPress filesystem instance |
| 45 | 73 | * |
| 46 | 74 | * @since 1.0.0 |
| 47 | 75 | * @var \WP_Filesystem_Base|null |
| @@ -290,8 +318,36 @@ | ||
| 290 | 318 | * @var bool |
| 291 | 319 | */ |
| 292 | 320 | private static bool $icon_sizes_listener_registered = false; |
| 293 | 321 | |
| 322 | + /** | |
| 323 | + * Whether the robots.txt resync listener is registered for this request. | |
| 324 | + * | |
| 325 | + * @since 2.14.0 | |
| 326 | + * @var bool | |
| 327 | + */ | |
| 328 | + private static bool $robots_sync_listener_registered = false; | |
| 329 | + | |
| 330 | + /** | |
| 331 | + * Flag set when a plugin change may have altered the sitemap set. | |
| 332 | + * | |
| 333 | + * @since 2.14.0 | |
| 334 | + * @var string | |
| 335 | + */ | |
| 336 | + public const ROBOTS_RESYNC_OPTION = 'thinkrank_robots_txt_resync_pending'; | |
| 337 | + | |
| 338 | + /** | |
| 339 | + * Flag set when a plugin change may have altered the sitemap index. | |
| 340 | + * | |
| 341 | + * Separate from ROBOTS_RESYNC_OPTION because the two files exist | |
| 342 | + * independently: that flag is only set when a physical robots.txt exists, | |
| 343 | + * and the sitemap index needs rebuilding whether or not it does. | |
| 344 | + * | |
| 345 | + * @since 2.15.0 | |
| 346 | + * @var string | |
| 347 | + */ | |
| 348 | + public const SITEMAP_RESYNC_OPTION = 'thinkrank_sitemap_contributors_changed'; | |
| 349 | + | |
| 294 | 350 | public function __construct() { |
| 295 | 351 | parent::__construct('site_identity'); |
| 296 | 352 | |
| 297 | 353 | if (!self::$icon_sizes_listener_registered) { |
| @@ -300,11 +356,130 @@ | ||
| 300 | 356 | // Admin only: resizing is not front-end work, and admin traffic is |
| 301 | 357 | // enough to run a one-time backfill promptly. |
| 302 | 358 | add_action('admin_init', [self::class, 'maybe_backfill_icon_sizes']); |
| 303 | 359 | } |
| 360 | + | |
| 361 | + if (!self::$robots_sync_listener_registered) { | |
| 362 | + self::$robots_sync_listener_registered = true; | |
| 363 | + | |
| 364 | + // A physical robots.txt bypasses PHP entirely, so composing the | |
| 365 | + // Sitemap block at render time fixes the served output only on | |
| 366 | + // sites with no file. Activating or deactivating a sitemap | |
| 367 | + // contributor changes the set, and until #835 nothing rewrote the | |
| 368 | + // file: the deactivated plugin's sitemap stayed advertised, serving | |
| 369 | + // HTML to anything that followed it. | |
| 370 | + add_action('activated_plugin', [self::class, 'flag_robots_txt_resync']); | |
| 371 | + add_action('deactivated_plugin', [self::class, 'flag_robots_txt_resync']); | |
| 372 | + add_action('init', [self::class, 'maybe_resync_robots_txt'], 99); | |
| 373 | + | |
| 374 | + // The sitemap index is a second static file listing the same | |
| 375 | + // contributors, with its own rebuild path. #835 / #859 resynced | |
| 376 | + // robots.txt only, so after Pro was deactivated the index kept | |
| 377 | + // advertising news-sitemap.xml, which then served the home page | |
| 378 | + // as HTML (#920). | |
| 379 | + add_action('activated_plugin', [self::class, 'flag_sitemap_resync']); | |
| 380 | + add_action('deactivated_plugin', [self::class, 'flag_sitemap_resync']); | |
| 381 | + add_action('init', [self::class, 'maybe_resync_sitemap'], 99); | |
| 382 | + } | |
| 304 | 383 | } |
| 305 | 384 | |
| 306 | 385 | /** |
| 386 | + * Note that the set of sitemap contributors may have changed. | |
| 387 | + * | |
| 388 | + * Deliberately unconditional about which plugin: a contributor is anything | |
| 389 | + * hooking `thinkrank_additional_sitemaps`, which is resolved at runtime and | |
| 390 | + * cannot be inspected for a plugin that is on its way out. | |
| 391 | + * | |
| 392 | + * The rewrite is not done here. `deactivated_plugin` fires inside the | |
| 393 | + * request that deactivated it, while that plugin's filters are still | |
| 394 | + * attached, so rendering now still sees the sitemap that is going away — | |
| 395 | + * measured, not assumed: the first version of this fix wrote the | |
| 396 | + * deactivated plugin's sitemap straight back into the file. The next | |
| 397 | + * request has the real plugin set loaded, so the work waits for it. | |
| 398 | + * | |
| 399 | + * @since 2.14.0 | |
| 400 | + * @return void | |
| 401 | + */ | |
| 402 | + public static function flag_robots_txt_resync(): void { | |
| 403 | + if (!file_exists(ABSPATH . 'robots.txt')) { | |
| 404 | + return; | |
| 405 | + } | |
| 406 | + | |
| 407 | + update_option(self::ROBOTS_RESYNC_OPTION, 1, false); | |
| 408 | + } | |
| 409 | + | |
| 410 | + /** | |
| 411 | + * Rewrite the physical robots.txt once, on the request after a change. | |
| 412 | + * | |
| 413 | + * @since 2.14.0 | |
| 414 | + * @return void | |
| 415 | + */ | |
| 416 | + public static function maybe_resync_robots_txt(): void { | |
| 417 | + if (!get_option(self::ROBOTS_RESYNC_OPTION)) { | |
| 418 | + return; | |
| 419 | + } | |
| 420 | + | |
| 421 | + // Cleared first, so a render that fatals cannot retry on every request | |
| 422 | + // for the rest of the site's life. | |
| 423 | + delete_option(self::ROBOTS_RESYNC_OPTION); | |
| 424 | + | |
| 425 | + if (!file_exists(ABSPATH . 'robots.txt')) { | |
| 426 | + return; | |
| 427 | + } | |
| 428 | + | |
| 429 | + (new self())->sync_robots_txt_file(); | |
| 430 | + } | |
| 431 | + | |
| 432 | + /** | |
| 433 | + * Note that the set of sitemap contributors may have changed. | |
| 434 | + * | |
| 435 | + * Unconditional, unlike flag_robots_txt_resync(): the sitemap files exist | |
| 436 | + * whether or not robots.txt does. The rebuild waits for the next request | |
| 437 | + * for the same reason as the robots.txt one, since `deactivated_plugin` | |
| 438 | + * still runs with the outgoing plugin's `thinkrank_additional_sitemaps` | |
| 439 | + * callback attached. | |
| 440 | + * | |
| 441 | + * @since 2.15.0 | |
| 442 | + * @return void | |
| 443 | + */ | |
| 444 | + public static function flag_sitemap_resync(): void { | |
| 445 | + update_option(self::SITEMAP_RESYNC_OPTION, 1, false); | |
| 446 | + } | |
| 447 | + | |
| 448 | + /** | |
| 449 | + * Queue a sitemap rebuild once, on the request after a contributor change. | |
| 450 | + * | |
| 451 | + * Goes through schedule_regeneration(), the debounced and lock-protected | |
| 452 | + * path a sitemap settings save uses, so a burst of plugin changes (a bulk | |
| 453 | + * deactivate, say) still produces one rebuild. That path also drops the | |
| 454 | + * cached dynamic documents, so sites serving the sitemap from PHP drop the | |
| 455 | + * entry as well. | |
| 456 | + * | |
| 457 | + * @since 2.15.0 | |
| 458 | + * @return void | |
| 459 | + */ | |
| 460 | + public static function maybe_resync_sitemap(): void { | |
| 461 | + if (!get_option(self::SITEMAP_RESYNC_OPTION)) { | |
| 462 | + return; | |
| 463 | + } | |
| 464 | + | |
| 465 | + // Cleared first, so a rebuild that fatals cannot be retried on every | |
| 466 | + // request for the rest of the site's life. | |
| 467 | + delete_option(self::SITEMAP_RESYNC_OPTION); | |
| 468 | + | |
| 469 | + $generator = new Sitemap_Generator(false); | |
| 470 | + $settings = $generator->get_settings('site'); | |
| 471 | + | |
| 472 | + // A disabled sitemap has no files to correct. Enabling it later builds | |
| 473 | + // from the contributors present at that time. | |
| 474 | + if (empty($settings['enabled'])) { | |
| 475 | + return; | |
| 476 | + } | |
| 477 | + | |
| 478 | + $generator->schedule_regeneration(); | |
| 479 | + } | |
| 480 | + | |
| 481 | + /** | |
| 307 | 482 | * Save settings, then refresh what a new canonical scheme invalidates. |
| 308 | 483 | * |
| 309 | 484 | * The static sitemap files are written with the scheme in force when they |
| 310 | 485 | * were built, and nothing else rebuilds them until a post or term changes. |
| @@ -480,11 +655,12 @@ | ||
| 480 | 655 | * Attachment ID behind a configured icon URL, or 0 when it is not ours. |
| 481 | 656 | * |
| 482 | 657 | * attachment_url_to_postid() matches _wp_attached_file, which holds the |
| 483 | 658 | * ORIGINAL upload path, so the URL of a generated derivative |
| 484 | - * (`logo-512.png`) returns 0 — and that is exactly what the media picker | |
| 485 | - * hands back when the user chooses a size. Strip the dimension suffix and | |
| 486 | - * try the original once. | |
| 659 | + * (`logo-512x512.png`) returns 0 — and that is exactly what the media | |
| 660 | + * picker hands back when the user chooses a size. Attachment_Lookup falls | |
| 661 | + * back to the original behind it; the fallback started here and moved | |
| 662 | + * there when every other image lookup turned out to need it (#847). | |
| 487 | 663 | * |
| 488 | 664 | * Shared with SEO_Manager's site-icon filter so both sides of the feature |
| 489 | 665 | * agree on which attachment a configured URL means. |
| 490 | 666 | * |
| @@ -493,21 +669,9 @@ | ||
| 493 | 669 | * @param string $url Configured icon URL. |
| 494 | 670 | * @return int Attachment ID, or 0. |
| 495 | 671 | */ |
| 496 | 672 | public static function icon_attachment_id(string $url): int { |
| 497 | - $attachment_id = (int) attachment_url_to_postid($url); | |
| 498 | - | |
| 499 | - if ($attachment_id) { | |
| 500 | - return $attachment_id; | |
| 501 | - } | |
| 502 | - | |
| 503 | - $original = preg_replace('/-\d+x\d+(?=\.[a-zA-Z0-9]+$)/', '', $url); | |
| 504 | - | |
| 505 | - if (is_string($original) && $original !== $url) { | |
| 506 | - return (int) attachment_url_to_postid($original); | |
| 507 | - } | |
| 508 | - | |
| 509 | - return 0; | |
| 673 | + return Attachment_Lookup::id_from_url($url); | |
| 510 | 674 | } |
| 511 | 675 | |
| 512 | 676 | /** |
| 513 | 677 | * Which ICON_SIZES derivatives this attachment still needs. |
| @@ -797,9 +961,20 @@ | ||
| 797 | 961 | // here rather than stored: the textarea holds the user's body, with |
| 798 | 962 | // the fenced block stripped out of every read and re-applied on every |
| 799 | 963 | // render. A site-wide block already disallows everyone, so adding the |
| 800 | 964 | // per-agent group there would be noise restating the same refusal. |
| 965 | + // Composed here rather than read from storage, for the same reason as | |
| 966 | + // the AI block below: the set of sitemaps an install publishes is a | |
| 967 | + // runtime fact. `robots_txt_content` is a snapshot of it taken at the | |
| 968 | + // last save, and nothing invalidated that snapshot, so deactivating a | |
| 969 | + // sitemap provider left its URL advertised and serving HTML (#835). | |
| 970 | + // Composing it on every render means the advertisement agrees with what | |
| 971 | + // the install publishes, in both directions, with no cache to expire. | |
| 801 | 972 | if (!$fully_blocked) { |
| 973 | + $body = $this->apply_sitemap_block($body); | |
| 974 | + } | |
| 975 | + | |
| 976 | + if (!$fully_blocked) { | |
| 802 | 977 | $body = $this->apply_ai_crawler_block($body, $settings); |
| 803 | 978 | } |
| 804 | 979 | |
| 805 | 980 | if ($body === '') { |
| @@ -809,8 +984,87 @@ | ||
| 809 | 984 | return $this->robots_txt_header() . $body . "\n"; |
| 810 | 985 | } |
| 811 | 986 | |
| 812 | 987 | /** |
| 988 | + * Replace the generated Sitemap block with the one this install publishes. | |
| 989 | + * | |
| 990 | + * @since 2.14.0 | |
| 991 | + * @param string $body Robots.txt body, without the header. | |
| 992 | + * @return string | |
| 993 | + */ | |
| 994 | + private function apply_sitemap_block(string $body): string { | |
| 995 | + $stripped = $this->strip_generated_sitemap_block($body); | |
| 996 | + $urls = $this->get_sitemap_urls_for_robots(); | |
| 997 | + | |
| 998 | + if (empty($urls)) { | |
| 999 | + return $stripped; | |
| 1000 | + } | |
| 1001 | + | |
| 1002 | + $block = ''; | |
| 1003 | + foreach ($urls as $url) { | |
| 1004 | + $block .= 'Sitemap: ' . $url . "\n"; | |
| 1005 | + } | |
| 1006 | + | |
| 1007 | + if ('' === trim($stripped)) { | |
| 1008 | + return trim($block); | |
| 1009 | + } | |
| 1010 | + | |
| 1011 | + // The grammar build_robots_txt_content() writes: one blank line before | |
| 1012 | + // the block, none inside it. A blank line terminates a record in the | |
| 1013 | + // robots.txt grammar, so a line between every directive is invalid. | |
| 1014 | + return rtrim($stripped) . "\n\n" . trim($block); | |
| 1015 | + } | |
| 1016 | + | |
| 1017 | + /** | |
| 1018 | + * Remove the plugin-written Sitemap block from a stored body. | |
| 1019 | + * | |
| 1020 | + * Only the trailing run of `Sitemap:` lines is removed, which is the exact | |
| 1021 | + * shape `build_robots_txt_content()` writes: a blank line, then nothing but | |
| 1022 | + * `Sitemap:` lines to the end of the body. A `Sitemap:` line anywhere else | |
| 1023 | + * was typed by the site owner and is left exactly where they put it, which | |
| 1024 | + * is why this cannot simply strip every matching line. | |
| 1025 | + * | |
| 1026 | + * @since 2.14.0 | |
| 1027 | + * @param string $body Robots.txt body. | |
| 1028 | + * @return string | |
| 1029 | + */ | |
| 1030 | + private function strip_generated_sitemap_block(string $body): string { | |
| 1031 | + $lines = preg_split('/\R/', $body); | |
| 1032 | + | |
| 1033 | + if (!is_array($lines)) { | |
| 1034 | + return $body; | |
| 1035 | + } | |
| 1036 | + | |
| 1037 | + $cut = count($lines); | |
| 1038 | + | |
| 1039 | + // Walk back over the trailing block: sitemap lines, and the blank lines | |
| 1040 | + // that separate or pad it. Anything else ends the block. | |
| 1041 | + for ($i = count($lines) - 1; $i >= 0; $i--) { | |
| 1042 | + $line = trim($lines[$i]); | |
| 1043 | + | |
| 1044 | + if ('' === $line) { | |
| 1045 | + $cut = $i; | |
| 1046 | + continue; | |
| 1047 | + } | |
| 1048 | + | |
| 1049 | + if (0 === stripos($line, 'sitemap:')) { | |
| 1050 | + $cut = $i; | |
| 1051 | + continue; | |
| 1052 | + } | |
| 1053 | + | |
| 1054 | + break; | |
| 1055 | + } | |
| 1056 | + | |
| 1057 | + if ($cut >= count($lines)) { | |
| 1058 | + return $body; | |
| 1059 | + } | |
| 1060 | + | |
| 1061 | + // Nothing but sitemap lines in the whole body means there is no owner | |
| 1062 | + // content to keep. | |
| 1063 | + return rtrim(implode("\n", array_slice($lines, 0, $cut))); | |
| 1064 | + } | |
| 1065 | + | |
| 1066 | + /** | |
| 813 | 1067 | * Resolve the robots.txt actually served to crawlers, with its origin. |
| 814 | 1068 | * |
| 815 | 1069 | * Lets an API/MCP consumer see the effective output without crawling the |
| 816 | 1070 | * URL. Mirrors serving precedence: a physical robots.txt in the web root is |
| @@ -1382,9 +1636,9 @@ | ||
| 1382 | 1636 | $optimization['suggestions'][] = 'Add a site logo for better branding and professional appearance'; |
| 1383 | 1637 | $optimization['score'] -= 20; |
| 1384 | 1638 | } else { |
| 1385 | 1639 | // Validate logo URL and dimensions |
| 1386 | - if (!filter_var($logo_url, FILTER_VALIDATE_URL)) { | |
| 1640 | + if (!\ThinkRank\Core\Url_Validator::is_http_url($logo_url)) { | |
| 1387 | 1641 | $optimization['warnings'][] = 'Logo URL format is invalid'; |
| 1388 | 1642 | $optimization['score'] -= 15; |
| 1389 | 1643 | } |
| 1390 | 1644 | } |
| @@ -1403,14 +1657,17 @@ | ||
| 1403 | 1657 | $optimization['score'] -= 10; |
| 1404 | 1658 | } |
| 1405 | 1659 | |
| 1406 | 1660 | // Additional logo analysis for local images |
| 1407 | - if (!empty($logo_url) && filter_var($logo_url, FILTER_VALIDATE_URL)) { | |
| 1408 | - $attachment_id = attachment_url_to_postid($logo_url); | |
| 1661 | + if (!empty($logo_url) && \ThinkRank\Core\Url_Validator::is_http_url($logo_url)) { | |
| 1662 | + $attachment_id = Attachment_Lookup::id_from_url($logo_url); | |
| 1409 | 1663 | if ($attachment_id) { |
| 1410 | 1664 | $image_meta = wp_get_attachment_metadata($attachment_id); |
| 1411 | - $width = isset($image_meta['width']) ? (int) $image_meta['width'] : 0; | |
| 1412 | - $height = isset($image_meta['height']) ? (int) $image_meta['height'] : 0; | |
| 1665 | + // The configured file's own size — a logo picked at a generated | |
| 1666 | + // size is not as large as the upload behind it. | |
| 1667 | + $logo_file = Attachment_Lookup::describe($attachment_id, $logo_url); | |
| 1668 | + $width = $logo_file['width']; | |
| 1669 | + $height = $logo_file['height']; | |
| 1413 | 1670 | |
| 1414 | 1671 | // SVG logos store 0x0 metadata — no dimension/ratio analysis |
| 1415 | 1672 | // is possible (and dividing by 0 is fatal). |
| 1416 | 1673 | if ($image_meta && $width > 0 && $height > 0) { |
| @@ -1513,11 +1770,12 @@ | ||
| 1513 | 1770 | $optimization['score'] -= 15; |
| 1514 | 1771 | } |
| 1515 | 1772 | } |
| 1516 | 1773 | |
| 1517 | - // Business type validation | |
| 1518 | - if (empty($settings['business_type'])) { | |
| 1519 | - $optimization['suggestions'][] = 'Select a specific business type for better schema markup'; | |
| 1774 | + // Business type validation (shared rule, one message — #622). | |
| 1775 | + $business_type = $this->business_type_status($settings); | |
| 1776 | + if ('suggestion' === $business_type['status']) { | |
| 1777 | + $optimization['suggestions'][] = $business_type['message']; | |
| 1520 | 1778 | $optimization['score'] -= 5; |
| 1521 | 1779 | } |
| 1522 | 1780 | |
| 1523 | 1781 | // Email validation |
| @@ -1887,12 +2145,37 @@ | ||
| 1887 | 2145 | $validation['suggestions'][] = 'Consider making site description longer (120-160 characters)'; |
| 1888 | 2146 | } |
| 1889 | 2147 | } |
| 1890 | 2148 | |
| 1891 | - // Validate logo URL | |
| 1892 | - if (isset($settings['logo_url']) && !empty($settings['logo_url'])) { | |
| 1893 | - if (!filter_var($settings['logo_url'], FILTER_VALIDATE_URL)) { | |
| 1894 | - $validation['errors'][] = 'Logo URL must be a valid URL'; | |
| 2149 | + // Image and link URLs. These are written into src and href | |
| 2150 | + // attributes (the schema logo, the admin previews, the hero section), | |
| 2151 | + // so only web URLs are accepted. is_valid() takes any scheme, and | |
| 2152 | + // "javascript://%0Aalert(1)" passed it and was stored as | |
| 2153 | + // "javascript://alert(1)" once sanitize_text_field() dropped the %0A. | |
| 2154 | + // The logo and default social image must be absolute: they are | |
| 2155 | + // published in schema and Open Graph, which require it. The others | |
| 2156 | + // may also be a path on this site. | |
| 2157 | + $url_fields = [ | |
| 2158 | + 'logo_url' => ['Logo URL', false], | |
| 2159 | + 'default_social_image' => ['Default social image URL', false], | |
| 2160 | + 'favicon_url' => ['Favicon URL', true], | |
| 2161 | + 'apple_touch_icon_url' => ['Apple touch icon URL', true], | |
| 2162 | + 'hero_background_image' => ['Hero background image URL', true], | |
| 2163 | + 'hero_cta_url' => ['Call-to-action URL', true], | |
| 2164 | + ]; | |
| 2165 | + foreach ($url_fields as $key => [$label, $allow_path]) { | |
| 2166 | + if (!isset($settings[$key]) || '' === $settings[$key] || null === $settings[$key]) { | |
| 2167 | + continue; | |
| 2168 | + } | |
| 2169 | + | |
| 2170 | + $ok = $allow_path | |
| 2171 | + ? \ThinkRank\Core\Url_Validator::is_http_url_or_path($settings[$key]) | |
| 2172 | + : \ThinkRank\Core\Url_Validator::is_http_url($settings[$key]); | |
| 2173 | + | |
| 2174 | + if (!$ok) { | |
| 2175 | + $validation['errors'][] = $allow_path | |
| 2176 | + ? sprintf('%s must be an http or https URL, or a path starting with /', $label) | |
| 2177 | + : sprintf('%s must be an http or https URL', $label); | |
| 1895 | 2178 | $validation['valid'] = false; |
| 1896 | 2179 | } |
| 1897 | 2180 | } |
| 1898 | 2181 | |
| @@ -2047,24 +2330,17 @@ | ||
| 2047 | 2330 | 'icon' => '✗' |
| 2048 | 2331 | ]; |
| 2049 | 2332 | } |
| 2050 | 2333 | |
| 2051 | - // Business Type validation | |
| 2052 | - if (!empty($settings['business_type']) && $settings['business_type'] !== 'LocalBusiness') { | |
| 2053 | - $field_details[] = [ | |
| 2054 | - 'field' => 'business_type', | |
| 2055 | - 'label' => 'Business type is selected for proper schema markup.', | |
| 2056 | - 'status' => 'valid', | |
| 2057 | - 'icon' => '✓' | |
| 2058 | - ]; | |
| 2059 | - } else { | |
| 2060 | - $field_details[] = [ | |
| 2061 | - 'field' => 'business_type', | |
| 2062 | - 'label' => 'Specific business type selection recommended for better schema markup.', | |
| 2063 | - 'status' => 'suggestion', | |
| 2064 | - 'icon' => '⚠' | |
| 2065 | - ]; | |
| 2066 | - } | |
| 2334 | + // Business Type validation — see business_type_status() for why there | |
| 2335 | + // is exactly one rule here now (#622). | |
| 2336 | + $business_type = $this->business_type_status($settings); | |
| 2337 | + $field_details[] = [ | |
| 2338 | + 'field' => 'business_type', | |
| 2339 | + 'label' => $business_type['message'], | |
| 2340 | + 'status' => $business_type['status'], | |
| 2341 | + 'icon' => 'valid' === $business_type['status'] ? '✓' : '⚠', | |
| 2342 | + ]; | |
| 2067 | 2343 | |
| 2068 | 2344 | // Address validation (NAP consistency) |
| 2069 | 2345 | $address_fields = ['business_address', 'business_city', 'business_state', 'business_country']; |
| 2070 | 2346 | $address_complete = true; |
| @@ -2337,9 +2613,9 @@ | ||
| 2337 | 2613 | } |
| 2338 | 2614 | |
| 2339 | 2615 | // CTA URL validation |
| 2340 | 2616 | if (!empty($settings['hero_cta_url'])) { |
| 2341 | - if (filter_var($settings['hero_cta_url'], FILTER_VALIDATE_URL) || strpos($settings['hero_cta_url'], '/') === 0) { | |
| 2617 | + if (\ThinkRank\Core\Url_Validator::is_http_url_or_path($settings['hero_cta_url'])) { | |
| 2342 | 2618 | $field_details[] = [ |
| 2343 | 2619 | 'field' => 'hero_cta_url', |
| 2344 | 2620 | 'label' => 'Call-to-action URL is properly configured.', |
| 2345 | 2621 | 'status' => 'valid', |
| @@ -2380,9 +2656,9 @@ | ||
| 2380 | 2656 | } |
| 2381 | 2657 | |
| 2382 | 2658 | // Site Logo validation (from Site Assets section) |
| 2383 | 2659 | if (!empty($settings['logo_url'])) { |
| 2384 | - if (filter_var($settings['logo_url'], FILTER_VALIDATE_URL)) { | |
| 2660 | + if (\ThinkRank\Core\Url_Validator::is_http_url($settings['logo_url'])) { | |
| 2385 | 2661 | $field_details[] = [ |
| 2386 | 2662 | 'field' => 'logo_url', |
| 2387 | 2663 | 'label' => 'Site logo is properly configured.', |
| 2388 | 2664 | 'status' => 'valid', |
| @@ -2748,11 +3024,15 @@ | ||
| 2748 | 3024 | } else { |
| 2749 | 3025 | $validation['suggestions'][] = 'Add business hours to improve local search visibility'; |
| 2750 | 3026 | } |
| 2751 | 3027 | |
| 2752 | - // Validate business type | |
| 2753 | - if (empty($settings['business_type'])) { | |
| 2754 | - $validation['suggestions'][] = 'Select a specific business type for better schema markup'; | |
| 3028 | + // Business type, through the shared rule (#622). This is the only place | |
| 3029 | + // it is reported on the generic path: validate_settings() with no tab | |
| 3030 | + // context attaches basic-info field details, not business-info ones, so | |
| 3031 | + // without this the setting would go unreported there entirely. | |
| 3032 | + $business_type = $this->business_type_status($settings); | |
| 3033 | + if ('suggestion' === $business_type['status']) { | |
| 3034 | + $validation['suggestions'][] = $business_type['message']; | |
| 2755 | 3035 | } |
| 2756 | 3036 | |
| 2757 | 3037 | return $validation; |
| 2758 | 3038 | } |
| @@ -2888,8 +3168,10 @@ | ||
| 2888 | 3168 | return [ |
| 2889 | 3169 | // Title formats, one per context. |
| 2890 | 3170 | 'homepage_title', 'post_title', 'page_title', 'category_title', |
| 2891 | 3171 | 'tag_title', 'author_title', 'search_title', 'archive_title', |
| 3172 | + // The blog-index homepage's meta description (#897). | |
| 3173 | + 'homepage_description', | |
| 2892 | 3174 | // Breadcrumbs. |
| 2893 | 3175 | 'breadcrumb_prefix', 'show_current_page', 'breadcrumb_use_seo_title', |
| 2894 | 3176 | // Identity, as written by the setup wizard and the importers. |
| 2895 | 3177 | 'alternate_name', 'identity_type', 'represents', |
| @@ -2945,12 +3227,86 @@ | ||
| 2945 | 3227 | ? $sanitized['canonical_scheme'] |
| 2946 | 3228 | : Url_Scheme::AUTOMATIC; |
| 2947 | 3229 | } |
| 2948 | 3230 | |
| 3231 | + // Same reasoning again for the business type. It goes straight into | |
| 3232 | + // LocalBusiness schema, so a type that is not in the schema.org | |
| 3233 | + // vocabulary is invalid structured data — and storing it verbatim would | |
| 3234 | + // have get-site-identity-settings report a type the site cannot | |
| 3235 | + // actually publish. An empty value keeps meaning "not set"; anything | |
| 3236 | + // else unrecognised falls back to the general-purpose root (#623). | |
| 3237 | + if (array_key_exists('business_type', $sanitized)) { | |
| 3238 | + $type = (string) $sanitized['business_type']; | |
| 3239 | + | |
| 3240 | + if ('' !== $type && !\ThinkRank\Config\Local_Business_Types_Config::is_valid($type)) { | |
| 3241 | + $type = \ThinkRank\Config\Local_Business_Types_Config::ROOT; | |
| 3242 | + } | |
| 3243 | + | |
| 3244 | + $sanitized['business_type'] = $type; | |
| 3245 | + } | |
| 3246 | + | |
| 2949 | 3247 | return $sanitized; |
| 2950 | 3248 | } |
| 2951 | 3249 | |
| 2952 | 3250 | /** |
| 3251 | + * schema.org's general-purpose LocalBusiness type. | |
| 3252 | + * | |
| 3253 | + * The default, the first option in the control, and a valid answer in its | |
| 3254 | + * own right — which is the whole point of #622. | |
| 3255 | + * | |
| 3256 | + * @since 2.10.0 | |
| 3257 | + * @var string | |
| 3258 | + */ | |
| 3259 | + private const GENERAL_BUSINESS_TYPE = 'LocalBusiness'; | |
| 3260 | + | |
| 3261 | + /** | |
| 3262 | + * The one rule for whether a business type needs the user's attention. | |
| 3263 | + * | |
| 3264 | + * There were three, with two wordings and two different conditions. Two | |
| 3265 | + * fired when the value was empty; the third fired when it WAS | |
| 3266 | + * `LocalBusiness` — which is the default, the first option in the control | |
| 3267 | + * and a perfectly valid schema.org type. So the warning appeared out of the | |
| 3268 | + * box for every site, could not be cleared without choosing a type that | |
| 3269 | + * might be inaccurate, and on an empty value it appeared three times in two | |
| 3270 | + * different phrasings, which is why it was reported as showing twice (#622). | |
| 3271 | + * | |
| 3272 | + * The rule now: a type is expected, and any type in the vocabulary is a | |
| 3273 | + * correct answer. Only an unset value is worth prompting about. | |
| 3274 | + * `LocalBusiness` is the general-purpose answer and is accepted as one — | |
| 3275 | + * with a note that a more specific type sharpens the schema, phrased as the | |
| 3276 | + * guidance it is rather than as a fault the user has to clear. | |
| 3277 | + * | |
| 3278 | + * @since 2.10.0 | |
| 3279 | + * | |
| 3280 | + * @param array $settings Site identity settings. | |
| 3281 | + * @return array{status:string,message:string} `valid` or `suggestion`. | |
| 3282 | + */ | |
| 3283 | + private function business_type_status(array $settings): array { | |
| 3284 | + $type = trim((string) ($settings['business_type'] ?? '')); | |
| 3285 | + | |
| 3286 | + if ('' === $type) { | |
| 3287 | + return [ | |
| 3288 | + 'status' => 'suggestion', | |
| 3289 | + 'message' => __('Select a business type so your local schema describes the right kind of business.', 'thinkrank'), | |
| 3290 | + ]; | |
| 3291 | + } | |
| 3292 | + | |
| 3293 | + // The literal rather than a constant from the expanded type list (#623): | |
| 3294 | + // that lands on its own branch, and this fix must not wait on it. | |
| 3295 | + if (self::GENERAL_BUSINESS_TYPE === $type) { | |
| 3296 | + return [ | |
| 3297 | + 'status' => 'valid', | |
| 3298 | + 'message' => __('Business type is set to Local Business. A more specific type sharpens your schema, if one fits.', 'thinkrank'), | |
| 3299 | + ]; | |
| 3300 | + } | |
| 3301 | + | |
| 3302 | + return [ | |
| 3303 | + 'status' => 'valid', | |
| 3304 | + 'message' => __('Business type is selected for proper schema markup.', 'thinkrank'), | |
| 3305 | + ]; | |
| 3306 | + } | |
| 3307 | + | |
| 3308 | + /** | |
| 2953 | 3309 | * Get default settings for a context type (implements interface) |
| 2954 | 3310 | * |
| 2955 | 3311 | * @since 1.0.0 |
| 2956 | 3312 | * |
| @@ -3659,11 +4015,29 @@ | ||
| 3659 | 4015 | } |
| 3660 | 4016 | } |
| 3661 | 4017 | |
| 3662 | 4018 | if ($index_url !== '') { |
| 3663 | - // The index alone — it covers the children and, on a segmented | |
| 3664 | - // install, the local business sitemap too. | |
| 3665 | - return [$index_url]; | |
| 4019 | + // The index covers the children and, on a segmented install, | |
| 4020 | + // the local business sitemap too. | |
| 4021 | + // | |
| 4022 | + // It does not cover a sitemap contributed through | |
| 4023 | + // `thinkrank_additional_sitemaps`: the index is built by this | |
| 4024 | + // plugin's own generator and never lists them. Returning the | |
| 4025 | + // index alone therefore left a contributed sitemap with no | |
| 4026 | + // discovery path at all — absent from robots.txt and absent | |
| 4027 | + // from the index — so Pro's News sitemap was unreachable on any | |
| 4028 | + // install with the index enabled, which is the default (#835). | |
| 4029 | + $contributed = []; | |
| 4030 | + | |
| 4031 | + foreach (\ThinkRank\SEO\Sitemap_Generator::additional_sitemaps() as $path) { | |
| 4032 | + $url = home_url($path); | |
| 4033 | + | |
| 4034 | + if ($url !== $index_url && !in_array($url, $contributed, true)) { | |
| 4035 | + $contributed[] = $url; | |
| 4036 | + } | |
| 4037 | + } | |
| 4038 | + | |
| 4039 | + return array_merge([$index_url], $contributed); | |
| 3666 | 4040 | } |
| 3667 | 4041 | |
| 3668 | 4042 | // Fallback to default if no URLs found |
| 3669 | 4043 | if (empty($sitemap_urls)) { |
| @@ -3675,9 +4049,22 @@ | ||
| 3675 | 4049 | // business sitemap and the sitemaps other plugins register both land |
| 3676 | 4050 | // here for the same reason, so they go through one list (#104). |
| 3677 | 4051 | $extra = []; |
| 3678 | 4052 | |
| 3679 | - if (file_exists(ABSPATH . 'local-sitemap.xml')) { | |
| 4053 | + // Not a file test. Under dynamic delivery the local sitemap is | |
| 4054 | + // served from PHP and no file is ever written, so file_exists() | |
| 4055 | + // silently dropped a sitemap the site really does publish (#752). | |
| 4056 | + // On static sites the file is still what proves it, so both count. | |
| 4057 | + $local_sitemap_published = file_exists(ABSPATH . 'local-sitemap.xml'); | |
| 4058 | + | |
| 4059 | + if (!$local_sitemap_published && class_exists('ThinkRank\\SEO\\Sitemap_Generator')) { | |
| 4060 | + $generator = new \ThinkRank\SEO\Sitemap_Generator(false); | |
| 4061 | + | |
| 4062 | + $local_sitemap_published = 'dynamic' === $generator->resolve_delivery_mode() | |
| 4063 | + && $generator->publishes_local_sitemap(); | |
| 4064 | + } | |
| 4065 | + | |
| 4066 | + if ($local_sitemap_published) { | |
| 3680 | 4067 | $extra[] = '/local-sitemap.xml'; |
| 3681 | 4068 | } |
| 3682 | 4069 | |
| 3683 | 4070 | foreach (\ThinkRank\SEO\Sitemap_Generator::additional_sitemaps() as $path) { |
| @@ -3740,9 +4127,12 @@ | ||
| 3740 | 4127 | $validation['warnings'][] = "Path '{$value}' should start with '/'"; |
| 3741 | 4128 | } |
| 3742 | 4129 | break; |
| 3743 | 4130 | case 'sitemap': |
| 3744 | - if (!filter_var($value, FILTER_VALIDATE_URL)) { | |
| 4131 | + // Url_Validator, not the raw PHP filter: on a site with an | |
| 4132 | + // internationalised domain the site's own sitemap URL is | |
| 4133 | + // non-ASCII and the raw filter refused it. | |
| 4134 | + if (!\ThinkRank\Core\Url_Validator::is_valid($value)) { | |
| 3745 | 4135 | $validation['errors'][] = "Invalid sitemap URL: {$value}"; |
| 3746 | 4136 | $validation['valid'] = false; |
| 3747 | 4137 | } |
| 3748 | 4138 | break; |
| @@ -4141,25 +4531,27 @@ | ||
| 4141 | 4531 | return $optimization; |
| 4142 | 4532 | } |
| 4143 | 4533 | |
| 4144 | 4534 | // Validate URL |
| 4145 | - if (!filter_var($value, FILTER_VALIDATE_URL)) { | |
| 4146 | - $optimization['validation']['errors'][] = "{$element} must be a valid URL"; | |
| 4535 | + if (!\ThinkRank\Core\Url_Validator::is_http_url($value)) { | |
| 4536 | + $optimization['validation']['errors'][] = "{$element} must be an http or https URL"; | |
| 4147 | 4537 | $optimization['validation']['valid'] = false; |
| 4148 | 4538 | return $optimization; |
| 4149 | 4539 | } |
| 4150 | 4540 | |
| 4151 | 4541 | // Check if it's a local image |
| 4152 | - $attachment_id = attachment_url_to_postid($value); | |
| 4542 | + $attachment_id = Attachment_Lookup::id_from_url($value); | |
| 4153 | 4543 | if ($attachment_id) { |
| 4154 | 4544 | $image_meta = wp_get_attachment_metadata($attachment_id); |
| 4155 | 4545 | |
| 4156 | 4546 | if ($image_meta && isset($image_meta['width'], $image_meta['height'])) { |
| 4157 | - // Check recommended size | |
| 4547 | + // Check recommended size, against the configured file itself | |
| 4548 | + // rather than the upload it may have been generated from. | |
| 4158 | 4549 | if (isset($config['recommended_size'])) { |
| 4159 | 4550 | [$rec_width, $rec_height] = explode('x', $config['recommended_size']); |
| 4551 | + $image_file = Attachment_Lookup::describe($attachment_id, $value); | |
| 4160 | 4552 | |
| 4161 | - if ((int) $image_meta['width'] !== (int) $rec_width || (int) $image_meta['height'] !== (int) $rec_height) { | |
| 4553 | + if ($image_file['width'] !== (int) $rec_width || $image_file['height'] !== (int) $rec_height) { | |
| 4162 | 4554 | $optimization['suggestions'][] = "Consider using {$config['recommended_size']} size for optimal {$element}"; |
| 4163 | 4555 | } |
| 4164 | 4556 | } |
| 4165 | 4557 | |
