| @@ -318,8 +318,36 @@ | ||
| 318 | 318 | * @var bool |
| 319 | 319 | */ |
| 320 | 320 | private static bool $icon_sizes_listener_registered = false; |
| 321 | 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 $robots_sync_listener_registered = false; | |
| 329 | + | |
| 330 | + /** | |
| 331 | + * Flag set when a plugin change may have altered the sitemap set. | |
| 332 | + * | |
| 333 | + * @since 2.14.0 | |
| 334 | + * @var string | |
| 335 | + */ | |
| 336 | + public const ROBOTS_RESYNC_OPTION = 'thinkrank_robots_txt_resync_pending'; | |
| 337 | + | |
| 338 | + /** | |
| 339 | + * Flag set when a plugin change may have altered the sitemap index. | |
| 340 | + * | |
| 341 | + * Separate from ROBOTS_RESYNC_OPTION because the two files exist | |
| 342 | + * independently: that flag is only set when a physical robots.txt exists, | |
| 343 | + * and the sitemap index needs rebuilding whether or not it does. | |
| 344 | + * | |
| 345 | + * @since 2.15.0 | |
| 346 | + * @var string | |
| 347 | + */ | |
| 348 | + public const SITEMAP_RESYNC_OPTION = 'thinkrank_sitemap_contributors_changed'; | |
| 349 | + | |
| 322 | 350 | public function __construct() { |
| 323 | 351 | parent::__construct('site_identity'); |
| 324 | 352 | |
| 325 | 353 | if (!self::$icon_sizes_listener_registered) { |
| @@ -328,11 +356,130 @@ | ||
| 328 | 356 | // Admin only: resizing is not front-end work, and admin traffic is |
| 329 | 357 | // enough to run a one-time backfill promptly. |
| 330 | 358 | add_action('admin_init', [self::class, 'maybe_backfill_icon_sizes']); |
| 331 | 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 | + } | |
| 332 | 383 | } |
| 333 | 384 | |
| 334 | 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 | + /** | |
| 335 | 482 | * Save settings, then refresh what a new canonical scheme invalidates. |
| 336 | 483 | * |
| 337 | 484 | * The static sitemap files are written with the scheme in force when they |
| 338 | 485 | * were built, and nothing else rebuilds them until a post or term changes. |
| @@ -508,11 +655,12 @@ | ||
| 508 | 655 | * Attachment ID behind a configured icon URL, or 0 when it is not ours. |
| 509 | 656 | * |
| 510 | 657 | * attachment_url_to_postid() matches _wp_attached_file, which holds the |
| 511 | 658 | * ORIGINAL upload path, so the URL of a generated derivative |
| 512 | - * (`logo-512.png`) returns 0 — and that is exactly what the media picker | |
| 513 | - * hands back when the user chooses a size. Strip the dimension suffix and | |
| 514 | - * 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). | |
| 515 | 663 | * |
| 516 | 664 | * Shared with SEO_Manager's site-icon filter so both sides of the feature |
| 517 | 665 | * agree on which attachment a configured URL means. |
| 518 | 666 | * |
| @@ -521,21 +669,9 @@ | ||
| 521 | 669 | * @param string $url Configured icon URL. |
| 522 | 670 | * @return int Attachment ID, or 0. |
| 523 | 671 | */ |
| 524 | 672 | public static function icon_attachment_id(string $url): int { |
| 525 | - $attachment_id = (int) attachment_url_to_postid($url); | |
| 526 | - | |
| 527 | - if ($attachment_id) { | |
| 528 | - return $attachment_id; | |
| 529 | - } | |
| 530 | - | |
| 531 | - $original = preg_replace('/-\d+x\d+(?=\.[a-zA-Z0-9]+$)/', '', $url); | |
| 532 | - | |
| 533 | - if (is_string($original) && $original !== $url) { | |
| 534 | - return (int) attachment_url_to_postid($original); | |
| 535 | - } | |
| 536 | - | |
| 537 | - return 0; | |
| 673 | + return Attachment_Lookup::id_from_url($url); | |
| 538 | 674 | } |
| 539 | 675 | |
| 540 | 676 | /** |
| 541 | 677 | * Which ICON_SIZES derivatives this attachment still needs. |
| @@ -825,9 +961,20 @@ | ||
| 825 | 961 | // here rather than stored: the textarea holds the user's body, with |
| 826 | 962 | // the fenced block stripped out of every read and re-applied on every |
| 827 | 963 | // render. A site-wide block already disallows everyone, so adding the |
| 828 | 964 | // per-agent group there would be noise restating the same refusal. |
| 965 | + // Composed here rather than read from storage, for the same reason as | |
| 966 | + // the AI block below: the set of sitemaps an install publishes is a | |
| 967 | + // runtime fact. `robots_txt_content` is a snapshot of it taken at the | |
| 968 | + // last save, and nothing invalidated that snapshot, so deactivating a | |
| 969 | + // sitemap provider left its URL advertised and serving HTML (#835). | |
| 970 | + // Composing it on every render means the advertisement agrees with what | |
| 971 | + // the install publishes, in both directions, with no cache to expire. | |
| 829 | 972 | if (!$fully_blocked) { |
| 973 | + $body = $this->apply_sitemap_block($body); | |
| 974 | + } | |
| 975 | + | |
| 976 | + if (!$fully_blocked) { | |
| 830 | 977 | $body = $this->apply_ai_crawler_block($body, $settings); |
| 831 | 978 | } |
| 832 | 979 | |
| 833 | 980 | if ($body === '') { |
| @@ -837,8 +984,87 @@ | ||
| 837 | 984 | return $this->robots_txt_header() . $body . "\n"; |
| 838 | 985 | } |
| 839 | 986 | |
| 840 | 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 | + /** | |
| 841 | 1067 | * Resolve the robots.txt actually served to crawlers, with its origin. |
| 842 | 1068 | * |
| 843 | 1069 | * Lets an API/MCP consumer see the effective output without crawling the |
| 844 | 1070 | * URL. Mirrors serving precedence: a physical robots.txt in the web root is |
| @@ -1410,9 +1636,9 @@ | ||
| 1410 | 1636 | $optimization['suggestions'][] = 'Add a site logo for better branding and professional appearance'; |
| 1411 | 1637 | $optimization['score'] -= 20; |
| 1412 | 1638 | } else { |
| 1413 | 1639 | // Validate logo URL and dimensions |
| 1414 | - if (!filter_var($logo_url, FILTER_VALIDATE_URL)) { | |
| 1640 | + if (!\ThinkRank\Core\Url_Validator::is_http_url($logo_url)) { | |
| 1415 | 1641 | $optimization['warnings'][] = 'Logo URL format is invalid'; |
| 1416 | 1642 | $optimization['score'] -= 15; |
| 1417 | 1643 | } |
| 1418 | 1644 | } |
| @@ -1431,14 +1657,17 @@ | ||
| 1431 | 1657 | $optimization['score'] -= 10; |
| 1432 | 1658 | } |
| 1433 | 1659 | |
| 1434 | 1660 | // Additional logo analysis for local images |
| 1435 | - if (!empty($logo_url) && filter_var($logo_url, FILTER_VALIDATE_URL)) { | |
| 1436 | - $attachment_id = attachment_url_to_postid($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); | |
| 1437 | 1663 | if ($attachment_id) { |
| 1438 | 1664 | $image_meta = wp_get_attachment_metadata($attachment_id); |
| 1439 | - $width = isset($image_meta['width']) ? (int) $image_meta['width'] : 0; | |
| 1440 | - $height = isset($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 = $logo_file['height']; | |
| 1441 | 1670 | |
| 1442 | 1671 | // SVG logos store 0x0 metadata — no dimension/ratio analysis |
| 1443 | 1672 | // is possible (and dividing by 0 is fatal). |
| 1444 | 1673 | if ($image_meta && $width > 0 && $height > 0) { |
| @@ -1916,12 +2145,37 @@ | ||
| 1916 | 2145 | $validation['suggestions'][] = 'Consider making site description longer (120-160 characters)'; |
| 1917 | 2146 | } |
| 1918 | 2147 | } |
| 1919 | 2148 | |
| 1920 | - // Validate logo URL | |
| 1921 | - if (isset($settings['logo_url']) && !empty($settings['logo_url'])) { | |
| 1922 | - if (!filter_var($settings['logo_url'], FILTER_VALIDATE_URL)) { | |
| 1923 | - $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); | |
| 1924 | 2178 | $validation['valid'] = false; |
| 1925 | 2179 | } |
| 1926 | 2180 | } |
| 1927 | 2181 | |
| @@ -2359,9 +2613,9 @@ | ||
| 2359 | 2613 | } |
| 2360 | 2614 | |
| 2361 | 2615 | // CTA URL validation |
| 2362 | 2616 | if (!empty($settings['hero_cta_url'])) { |
| 2363 | - 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'])) { | |
| 2364 | 2618 | $field_details[] = [ |
| 2365 | 2619 | 'field' => 'hero_cta_url', |
| 2366 | 2620 | 'label' => 'Call-to-action URL is properly configured.', |
| 2367 | 2621 | 'status' => 'valid', |
| @@ -2402,9 +2656,9 @@ | ||
| 2402 | 2656 | } |
| 2403 | 2657 | |
| 2404 | 2658 | // Site Logo validation (from Site Assets section) |
| 2405 | 2659 | if (!empty($settings['logo_url'])) { |
| 2406 | - if (filter_var($settings['logo_url'], FILTER_VALIDATE_URL)) { | |
| 2660 | + if (\ThinkRank\Core\Url_Validator::is_http_url($settings['logo_url'])) { | |
| 2407 | 2661 | $field_details[] = [ |
| 2408 | 2662 | 'field' => 'logo_url', |
| 2409 | 2663 | 'label' => 'Site logo is properly configured.', |
| 2410 | 2664 | 'status' => 'valid', |
| @@ -2914,8 +3168,10 @@ | ||
| 2914 | 3168 | return [ |
| 2915 | 3169 | // Title formats, one per context. |
| 2916 | 3170 | 'homepage_title', 'post_title', 'page_title', 'category_title', |
| 2917 | 3171 | 'tag_title', 'author_title', 'search_title', 'archive_title', |
| 3172 | + // The blog-index homepage's meta description (#897). | |
| 3173 | + 'homepage_description', | |
| 2918 | 3174 | // Breadcrumbs. |
| 2919 | 3175 | 'breadcrumb_prefix', 'show_current_page', 'breadcrumb_use_seo_title', |
| 2920 | 3176 | // Identity, as written by the setup wizard and the importers. |
| 2921 | 3177 | 'alternate_name', 'identity_type', 'represents', |
| @@ -3759,11 +4015,29 @@ | ||
| 3759 | 4015 | } |
| 3760 | 4016 | } |
| 3761 | 4017 | |
| 3762 | 4018 | if ($index_url !== '') { |
| 3763 | - // The index alone — it covers the children and, on a segmented | |
| 3764 | - // install, the local business sitemap too. | |
| 3765 | - 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], $contributed); | |
| 3766 | 4040 | } |
| 3767 | 4041 | |
| 3768 | 4042 | // Fallback to default if no URLs found |
| 3769 | 4043 | if (empty($sitemap_urls)) { |
| @@ -3853,9 +4127,12 @@ | ||
| 3853 | 4127 | $validation['warnings'][] = "Path '{$value}' should start with '/'"; |
| 3854 | 4128 | } |
| 3855 | 4129 | break; |
| 3856 | 4130 | case 'sitemap': |
| 3857 | - 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)) { | |
| 3858 | 4135 | $validation['errors'][] = "Invalid sitemap URL: {$value}"; |
| 3859 | 4136 | $validation['valid'] = false; |
| 3860 | 4137 | } |
| 3861 | 4138 | break; |
| @@ -4254,25 +4531,27 @@ | ||
| 4254 | 4531 | return $optimization; |
| 4255 | 4532 | } |
| 4256 | 4533 | |
| 4257 | 4534 | // Validate URL |
| 4258 | - if (!filter_var($value, FILTER_VALIDATE_URL)) { | |
| 4259 | - $optimization['validation']['errors'][] = "{$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"; | |
| 4260 | 4537 | $optimization['validation']['valid'] = false; |
| 4261 | 4538 | return $optimization; |
| 4262 | 4539 | } |
| 4263 | 4540 | |
| 4264 | 4541 | // Check if it's a local image |
| 4265 | - $attachment_id = attachment_url_to_postid($value); | |
| 4542 | + $attachment_id = Attachment_Lookup::id_from_url($value); | |
| 4266 | 4543 | if ($attachment_id) { |
| 4267 | 4544 | $image_meta = wp_get_attachment_metadata($attachment_id); |
| 4268 | 4545 | |
| 4269 | 4546 | if ($image_meta && isset($image_meta['width'], $image_meta['height'])) { |
| 4270 | - // Check recommended size | |
| 4547 | + // Check recommended size, against the configured file itself | |
| 4548 | + // rather than the upload it may have been generated from. | |
| 4271 | 4549 | if (isset($config['recommended_size'])) { |
| 4272 | 4550 | [$rec_width, $rec_height] = explode('x', $config['recommended_size']); |
| 4551 | + $image_file = Attachment_Lookup::describe($attachment_id, $value); | |
| 4273 | 4552 | |
| 4274 | - 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) { | |
| 4275 | 4554 | $optimization['suggestions'][] = "Consider using {$config['recommended_size']} size for optimal {$element}"; |
| 4276 | 4555 | } |
| 4277 | 4556 | } |
| 4278 | 4557 | |
