| @@ -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 |
| @@ -303,8 +331,91 @@ | ||
| 303 | 331 | } |
| 304 | 332 | } |
| 305 | 333 | |
| 306 | 334 | /** |
| 335 | + * Save settings, then refresh what a new canonical scheme invalidates. | |
| 336 | + * | |
| 337 | + * The static sitemap files are written with the scheme in force when they | |
| 338 | + * were built, and nothing else rebuilds them until a post or term changes. | |
| 339 | + * So a change of scheme left every `<loc>` on the old one while canonical | |
| 340 | + * and og:url had already moved (#736). Every writer (the settings route, | |
| 341 | + * the robots route, the MCP abilities, an import) lands here. | |
| 342 | + * | |
| 343 | + * @since 2.7.0 | |
| 344 | + * | |
| 345 | + * @param string $context_type Context type. | |
| 346 | + * @param int|null $context_id Context ID. | |
| 347 | + * @param array $settings Settings to save. | |
| 348 | + * @return bool | |
| 349 | + */ | |
| 350 | + public function save_settings(string $context_type, ?int $context_id, array $settings): bool { | |
| 351 | + if (!self::touches_canonical_scheme($context_type, $context_id, $settings)) { | |
| 352 | + return parent::save_settings($context_type, $context_id, $settings); | |
| 353 | + } | |
| 354 | + | |
| 355 | + $before = Url_Scheme::preference(); | |
| 356 | + $saved = parent::save_settings($context_type, $context_id, $settings); | |
| 357 | + | |
| 358 | + if ($saved) { | |
| 359 | + $this->on_canonical_scheme_saved($before); | |
| 360 | + } | |
| 361 | + | |
| 362 | + return $saved; | |
| 363 | + } | |
| 364 | + | |
| 365 | + /** | |
| 366 | + * Whether a save can change the site-wide canonical scheme. | |
| 367 | + * | |
| 368 | + * @since 2.7.0 | |
| 369 | + * | |
| 370 | + * @param string $context_type Context type. | |
| 371 | + * @param int|null $context_id Context ID. | |
| 372 | + * @param array $settings Settings being saved. | |
| 373 | + * @return bool | |
| 374 | + */ | |
| 375 | + public static function touches_canonical_scheme(string $context_type, ?int $context_id, array $settings): bool { | |
| 376 | + return 'site' === sanitize_key($context_type) | |
| 377 | + && empty($context_id) | |
| 378 | + && array_key_exists('canonical_scheme', $settings); | |
| 379 | + } | |
| 380 | + | |
| 381 | + /** | |
| 382 | + * Rebuild the static sitemaps when the effective scheme changed. | |
| 383 | + * | |
| 384 | + * Compares the effective preference, filter included, so a site whose | |
| 385 | + * scheme is pinned by `thinkrank_canonical_scheme` does not rebuild on a | |
| 386 | + * stored value that changes nothing it publishes. | |
| 387 | + * | |
| 388 | + * @since 2.7.0 | |
| 389 | + * | |
| 390 | + * @param string $before Effective scheme before the save. | |
| 391 | + * @return void | |
| 392 | + */ | |
| 393 | + protected function on_canonical_scheme_saved(string $before): void { | |
| 394 | + // The preference is cached for the request; the save just changed it. | |
| 395 | + Url_Scheme::reset(); | |
| 396 | + | |
| 397 | + if (Url_Scheme::preference() === $before) { | |
| 398 | + return; | |
| 399 | + } | |
| 400 | + | |
| 401 | + $this->schedule_sitemap_rebuild(); | |
| 402 | + } | |
| 403 | + | |
| 404 | + /** | |
| 405 | + * Queue a settings-driven sitemap rebuild. | |
| 406 | + * | |
| 407 | + * Debounced and run after the response, like any other settings change | |
| 408 | + * that alters what the sitemap publishes. | |
| 409 | + * | |
| 410 | + * @since 2.7.0 | |
| 411 | + * @return void | |
| 412 | + */ | |
| 413 | + protected function schedule_sitemap_rebuild(): void { | |
| 414 | + (new Sitemap_Generator(false))->schedule_regeneration(); | |
| 415 | + } | |
| 416 | + | |
| 417 | + /** | |
| 307 | 418 | * Build the icon derivatives for a newly chosen favicon. |
| 308 | 419 | * |
| 309 | 420 | * Runs on save, which is the only moment the choice changes and the only |
| 310 | 421 | * place image work belongs — resolving a size on the front end must stay a |
| @@ -397,11 +508,12 @@ | ||
| 397 | 508 | * Attachment ID behind a configured icon URL, or 0 when it is not ours. |
| 398 | 509 | * |
| 399 | 510 | * attachment_url_to_postid() matches _wp_attached_file, which holds the |
| 400 | 511 | * ORIGINAL upload path, so the URL of a generated derivative |
| 401 | - * (`logo-512.png`) returns 0 — and that is exactly what the media picker | |
| 402 | - * hands back when the user chooses a size. Strip the dimension suffix and | |
| 403 | - * try the original once. | |
| 512 | + * (`logo-512x512.png`) returns 0 — and that is exactly what the media | |
| 513 | + * picker hands back when the user chooses a size. Attachment_Lookup falls | |
| 514 | + * back to the original behind it; the fallback started here and moved | |
| 515 | + * there when every other image lookup turned out to need it (#847). | |
| 404 | 516 | * |
| 405 | 517 | * Shared with SEO_Manager's site-icon filter so both sides of the feature |
| 406 | 518 | * agree on which attachment a configured URL means. |
| 407 | 519 | * |
| @@ -410,21 +522,9 @@ | ||
| 410 | 522 | * @param string $url Configured icon URL. |
| 411 | 523 | * @return int Attachment ID, or 0. |
| 412 | 524 | */ |
| 413 | 525 | public static function icon_attachment_id(string $url): int { |
| 414 | - $attachment_id = (int) attachment_url_to_postid($url); | |
| 415 | - | |
| 416 | - if ($attachment_id) { | |
| 417 | - return $attachment_id; | |
| 418 | - } | |
| 419 | - | |
| 420 | - $original = preg_replace('/-\d+x\d+(?=\.[a-zA-Z0-9]+$)/', '', $url); | |
| 421 | - | |
| 422 | - if (is_string($original) && $original !== $url) { | |
| 423 | - return (int) attachment_url_to_postid($original); | |
| 424 | - } | |
| 425 | - | |
| 426 | - return 0; | |
| 526 | + return Attachment_Lookup::id_from_url($url); | |
| 427 | 527 | } |
| 428 | 528 | |
| 429 | 529 | /** |
| 430 | 530 | * Which ICON_SIZES derivatives this attachment still needs. |
| @@ -1178,17 +1278,19 @@ | ||
| 1178 | 1278 | $home_text = $settings['breadcrumb_home_text'] ?? 'Home'; |
| 1179 | 1279 | if (empty($home_text)) { |
| 1180 | 1280 | $optimization['warnings'][] = 'Empty home text reduces accessibility for screen readers'; |
| 1181 | 1281 | $optimization['score'] -= 15; |
| 1182 | - } elseif (strlen($home_text) > 20) { | |
| 1183 | - $optimization['suggestions'][] = 'Keep home text concise (current: ' . strlen($home_text) . ' chars)'; | |
| 1282 | + } elseif (mb_strlen($home_text) > 20) { | |
| 1283 | + // mb_strlen: this number is shown to the user as "chars" (#687). | |
| 1284 | + $optimization['suggestions'][] = 'Keep home text concise (current: ' . mb_strlen($home_text) . ' chars)'; | |
| 1184 | 1285 | $optimization['score'] -= 5; |
| 1185 | 1286 | } |
| 1186 | 1287 | |
| 1187 | 1288 | // Check prefix usage |
| 1188 | 1289 | $prefix = $settings['breadcrumb_prefix'] ?? ''; |
| 1189 | - if (!empty($prefix) && strlen($prefix) > 50) { | |
| 1190 | - $optimization['suggestions'][] = 'Breadcrumb prefix is quite long (' . strlen($prefix) . ' chars) - consider shortening'; | |
| 1290 | + if (!empty($prefix) && mb_strlen($prefix) > 50) { | |
| 1291 | + // mb_strlen: this number is shown to the user as "chars" (#687). | |
| 1292 | + $optimization['suggestions'][] = 'Breadcrumb prefix is quite long (' . mb_strlen($prefix) . ' chars) - consider shortening'; | |
| 1191 | 1293 | $optimization['score'] -= 5; |
| 1192 | 1294 | } |
| 1193 | 1295 | |
| 1194 | 1296 | // Current page display |
| @@ -1319,13 +1421,16 @@ | ||
| 1319 | 1421 | } |
| 1320 | 1422 | |
| 1321 | 1423 | // Additional logo analysis for local images |
| 1322 | 1424 | if (!empty($logo_url) && filter_var($logo_url, FILTER_VALIDATE_URL)) { |
| 1323 | - $attachment_id = attachment_url_to_postid($logo_url); | |
| 1425 | + $attachment_id = Attachment_Lookup::id_from_url($logo_url); | |
| 1324 | 1426 | if ($attachment_id) { |
| 1325 | 1427 | $image_meta = wp_get_attachment_metadata($attachment_id); |
| 1326 | - $width = isset($image_meta['width']) ? (int) $image_meta['width'] : 0; | |
| 1327 | - $height = isset($image_meta['height']) ? (int) $image_meta['height'] : 0; | |
| 1428 | + // The configured file's own size — a logo picked at a generated | |
| 1429 | + // size is not as large as the upload behind it. | |
| 1430 | + $logo_file = Attachment_Lookup::describe($attachment_id, $logo_url); | |
| 1431 | + $width = $logo_file['width']; | |
| 1432 | + $height = $logo_file['height']; | |
| 1328 | 1433 | |
| 1329 | 1434 | // SVG logos store 0x0 metadata — no dimension/ratio analysis |
| 1330 | 1435 | // is possible (and dividing by 0 is fatal). |
| 1331 | 1436 | if ($image_meta && $width > 0 && $height > 0) { |
| @@ -1428,11 +1533,12 @@ | ||
| 1428 | 1533 | $optimization['score'] -= 15; |
| 1429 | 1534 | } |
| 1430 | 1535 | } |
| 1431 | 1536 | |
| 1432 | - // Business type validation | |
| 1433 | - if (empty($settings['business_type'])) { | |
| 1434 | - $optimization['suggestions'][] = 'Select a specific business type for better schema markup'; | |
| 1537 | + // Business type validation (shared rule, one message — #622). | |
| 1538 | + $business_type = $this->business_type_status($settings); | |
| 1539 | + if ('suggestion' === $business_type['status']) { | |
| 1540 | + $optimization['suggestions'][] = $business_type['message']; | |
| 1435 | 1541 | $optimization['score'] -= 5; |
| 1436 | 1542 | } |
| 1437 | 1543 | |
| 1438 | 1544 | // Email validation |
| @@ -1962,24 +2068,17 @@ | ||
| 1962 | 2068 | 'icon' => '✗' |
| 1963 | 2069 | ]; |
| 1964 | 2070 | } |
| 1965 | 2071 | |
| 1966 | - // Business Type validation | |
| 1967 | - if (!empty($settings['business_type']) && $settings['business_type'] !== 'LocalBusiness') { | |
| 1968 | - $field_details[] = [ | |
| 1969 | - 'field' => 'business_type', | |
| 1970 | - 'label' => 'Business type is selected for proper schema markup.', | |
| 1971 | - 'status' => 'valid', | |
| 1972 | - 'icon' => '✓' | |
| 1973 | - ]; | |
| 1974 | - } else { | |
| 1975 | - $field_details[] = [ | |
| 1976 | - 'field' => 'business_type', | |
| 1977 | - 'label' => 'Specific business type selection recommended for better schema markup.', | |
| 1978 | - 'status' => 'suggestion', | |
| 1979 | - 'icon' => '⚠' | |
| 1980 | - ]; | |
| 1981 | - } | |
| 2072 | + // Business Type validation — see business_type_status() for why there | |
| 2073 | + // is exactly one rule here now (#622). | |
| 2074 | + $business_type = $this->business_type_status($settings); | |
| 2075 | + $field_details[] = [ | |
| 2076 | + 'field' => 'business_type', | |
| 2077 | + 'label' => $business_type['message'], | |
| 2078 | + 'status' => $business_type['status'], | |
| 2079 | + 'icon' => 'valid' === $business_type['status'] ? '✓' : '⚠', | |
| 2080 | + ]; | |
| 1982 | 2081 | |
| 1983 | 2082 | // Address validation (NAP consistency) |
| 1984 | 2083 | $address_fields = ['business_address', 'business_city', 'business_state', 'business_country']; |
| 1985 | 2084 | $address_complete = true; |
| @@ -2663,11 +2762,15 @@ | ||
| 2663 | 2762 | } else { |
| 2664 | 2763 | $validation['suggestions'][] = 'Add business hours to improve local search visibility'; |
| 2665 | 2764 | } |
| 2666 | 2765 | |
| 2667 | - // Validate business type | |
| 2668 | - if (empty($settings['business_type'])) { | |
| 2669 | - $validation['suggestions'][] = 'Select a specific business type for better schema markup'; | |
| 2766 | + // Business type, through the shared rule (#622). This is the only place | |
| 2767 | + // it is reported on the generic path: validate_settings() with no tab | |
| 2768 | + // context attaches basic-info field details, not business-info ones, so | |
| 2769 | + // without this the setting would go unreported there entirely. | |
| 2770 | + $business_type = $this->business_type_status($settings); | |
| 2771 | + if ('suggestion' === $business_type['status']) { | |
| 2772 | + $validation['suggestions'][] = $business_type['message']; | |
| 2670 | 2773 | } |
| 2671 | 2774 | |
| 2672 | 2775 | return $validation; |
| 2673 | 2776 | } |
| @@ -2759,8 +2862,47 @@ | ||
| 2759 | 2862 | * @since 2.0.1 |
| 2760 | 2863 | * |
| 2761 | 2864 | * @return string[] |
| 2762 | 2865 | */ |
| 2866 | + /** | |
| 2867 | + * The stored alternate name(s), shaped for schema output. | |
| 2868 | + * | |
| 2869 | + * schema.org and Google both allow `alternateName` to carry one value or | |
| 2870 | + * several, and the store already round-trips either shape, so this accepts | |
| 2871 | + * both and normalises: null when there is nothing to publish, a bare string | |
| 2872 | + * for one name, a list for more. Emitting a one-element array would be | |
| 2873 | + * valid but noisier than it needs to be. | |
| 2874 | + * | |
| 2875 | + * Shared because both WebSite producers need it and must agree — a property | |
| 2876 | + * added to one and not the other is how #688 happened. | |
| 2877 | + * | |
| 2878 | + * @since 2.7.0 | |
| 2879 | + * | |
| 2880 | + * @param mixed $value Stored alternate_name value. | |
| 2881 | + * @return string|string[]|null | |
| 2882 | + */ | |
| 2883 | + public static function alternate_name_for_schema($value) { | |
| 2884 | + $names = []; | |
| 2885 | + | |
| 2886 | + foreach ((array) $value as $name) { | |
| 2887 | + if (!is_scalar($name)) { | |
| 2888 | + continue; | |
| 2889 | + } | |
| 2890 | + | |
| 2891 | + $name = trim((string) $name); | |
| 2892 | + | |
| 2893 | + if ('' !== $name && !in_array($name, $names, true)) { | |
| 2894 | + $names[] = $name; | |
| 2895 | + } | |
| 2896 | + } | |
| 2897 | + | |
| 2898 | + if (empty($names)) { | |
| 2899 | + return null; | |
| 2900 | + } | |
| 2901 | + | |
| 2902 | + return 1 === count($names) ? $names[0] : $names; | |
| 2903 | + } | |
| 2904 | + | |
| 2763 | 2905 | protected function additional_setting_keys(): array { |
| 2764 | 2906 | return [ |
| 2765 | 2907 | // Title formats, one per context. |
| 2766 | 2908 | 'homepage_title', 'post_title', 'page_title', 'category_title', |
| @@ -2811,12 +2953,96 @@ | ||
| 2811 | 2953 | if (array_key_exists('ai_crawler_rules', $sanitized)) { |
| 2812 | 2954 | $sanitized['ai_crawler_rules'] = AI_Crawlers::normalize_rules($sanitized['ai_crawler_rules']); |
| 2813 | 2955 | } |
| 2814 | 2956 | |
| 2957 | + // Same reasoning one key up, for the scheme override (#638). Anything | |
| 2958 | + // that is not one of the three modes means "follow WordPress", and is | |
| 2959 | + // stored as that rather than kept verbatim — otherwise get-site-identity | |
| 2960 | + // -settings would report a scheme the site does not actually publish. | |
| 2961 | + if (array_key_exists('canonical_scheme', $sanitized)) { | |
| 2962 | + $sanitized['canonical_scheme'] = in_array($sanitized['canonical_scheme'], Url_Scheme::MODES, true) | |
| 2963 | + ? $sanitized['canonical_scheme'] | |
| 2964 | + : Url_Scheme::AUTOMATIC; | |
| 2965 | + } | |
| 2966 | + | |
| 2967 | + // Same reasoning again for the business type. It goes straight into | |
| 2968 | + // LocalBusiness schema, so a type that is not in the schema.org | |
| 2969 | + // vocabulary is invalid structured data — and storing it verbatim would | |
| 2970 | + // have get-site-identity-settings report a type the site cannot | |
| 2971 | + // actually publish. An empty value keeps meaning "not set"; anything | |
| 2972 | + // else unrecognised falls back to the general-purpose root (#623). | |
| 2973 | + if (array_key_exists('business_type', $sanitized)) { | |
| 2974 | + $type = (string) $sanitized['business_type']; | |
| 2975 | + | |
| 2976 | + if ('' !== $type && !\ThinkRank\Config\Local_Business_Types_Config::is_valid($type)) { | |
| 2977 | + $type = \ThinkRank\Config\Local_Business_Types_Config::ROOT; | |
| 2978 | + } | |
| 2979 | + | |
| 2980 | + $sanitized['business_type'] = $type; | |
| 2981 | + } | |
| 2982 | + | |
| 2815 | 2983 | return $sanitized; |
| 2816 | 2984 | } |
| 2817 | 2985 | |
| 2818 | 2986 | /** |
| 2987 | + * schema.org's general-purpose LocalBusiness type. | |
| 2988 | + * | |
| 2989 | + * The default, the first option in the control, and a valid answer in its | |
| 2990 | + * own right — which is the whole point of #622. | |
| 2991 | + * | |
| 2992 | + * @since 2.10.0 | |
| 2993 | + * @var string | |
| 2994 | + */ | |
| 2995 | + private const GENERAL_BUSINESS_TYPE = 'LocalBusiness'; | |
| 2996 | + | |
| 2997 | + /** | |
| 2998 | + * The one rule for whether a business type needs the user's attention. | |
| 2999 | + * | |
| 3000 | + * There were three, with two wordings and two different conditions. Two | |
| 3001 | + * fired when the value was empty; the third fired when it WAS | |
| 3002 | + * `LocalBusiness` — which is the default, the first option in the control | |
| 3003 | + * and a perfectly valid schema.org type. So the warning appeared out of the | |
| 3004 | + * box for every site, could not be cleared without choosing a type that | |
| 3005 | + * might be inaccurate, and on an empty value it appeared three times in two | |
| 3006 | + * different phrasings, which is why it was reported as showing twice (#622). | |
| 3007 | + * | |
| 3008 | + * The rule now: a type is expected, and any type in the vocabulary is a | |
| 3009 | + * correct answer. Only an unset value is worth prompting about. | |
| 3010 | + * `LocalBusiness` is the general-purpose answer and is accepted as one — | |
| 3011 | + * with a note that a more specific type sharpens the schema, phrased as the | |
| 3012 | + * guidance it is rather than as a fault the user has to clear. | |
| 3013 | + * | |
| 3014 | + * @since 2.10.0 | |
| 3015 | + * | |
| 3016 | + * @param array $settings Site identity settings. | |
| 3017 | + * @return array{status:string,message:string} `valid` or `suggestion`. | |
| 3018 | + */ | |
| 3019 | + private function business_type_status(array $settings): array { | |
| 3020 | + $type = trim((string) ($settings['business_type'] ?? '')); | |
| 3021 | + | |
| 3022 | + if ('' === $type) { | |
| 3023 | + return [ | |
| 3024 | + 'status' => 'suggestion', | |
| 3025 | + 'message' => __('Select a business type so your local schema describes the right kind of business.', 'thinkrank'), | |
| 3026 | + ]; | |
| 3027 | + } | |
| 3028 | + | |
| 3029 | + // The literal rather than a constant from the expanded type list (#623): | |
| 3030 | + // that lands on its own branch, and this fix must not wait on it. | |
| 3031 | + if (self::GENERAL_BUSINESS_TYPE === $type) { | |
| 3032 | + return [ | |
| 3033 | + 'status' => 'valid', | |
| 3034 | + 'message' => __('Business type is set to Local Business. A more specific type sharpens your schema, if one fits.', 'thinkrank'), | |
| 3035 | + ]; | |
| 3036 | + } | |
| 3037 | + | |
| 3038 | + return [ | |
| 3039 | + 'status' => 'valid', | |
| 3040 | + 'message' => __('Business type is selected for proper schema markup.', 'thinkrank'), | |
| 3041 | + ]; | |
| 3042 | + } | |
| 3043 | + | |
| 3044 | + /** | |
| 2819 | 3045 | * Get default settings for a context type (implements interface) |
| 2820 | 3046 | * |
| 2821 | 3047 | * @since 1.0.0 |
| 2822 | 3048 | * |
| @@ -2836,8 +3062,27 @@ | ||
| 2836 | 3062 | 'breadcrumb_home_text' => 'Home', |
| 2837 | 3063 | 'breadcrumb_separator' => '>', |
| 2838 | 3064 | 'robots_txt_enabled' => true, |
| 2839 | 3065 | 'allow_search_engines' => true, |
| 3066 | + // Answer 404 when a content selector in the URL resolved to | |
| 3067 | + // nothing (#634). On by default, unlike the other new settings | |
| 3068 | + // here: it changes no URL a visitor or a correct crawler uses, only | |
| 3069 | + // ones where WordPress resolved nothing and served the blog listing | |
| 3070 | + // at 200 anyway. | |
| 3071 | + 'query_protection' => true, | |
| 3072 | + | |
| 3073 | + // Feed controls (#635). All three off, so an upgrade changes | |
| 3074 | + // nothing about what an existing site already sends its | |
| 3075 | + // subscribers; a brand-new install is seeded with the signature and | |
| 3076 | + // the noindex on, in Activator::seed_feed_defaults(). | |
| 3077 | + 'feed_excerpt_only' => false, | |
| 3078 | + 'feed_source_link' => false, | |
| 3079 | + 'feed_noindex' => false, | |
| 3080 | + | |
| 3081 | + // The scheme self-referential URLs go out with (#638). 'automatic' | |
| 3082 | + // means substitute nothing and follow WordPress, which is what | |
| 3083 | + // every site did before the setting existed. | |
| 3084 | + 'canonical_scheme' => Url_Scheme::AUTOMATIC, | |
| 2840 | 3085 | 'robots_txt_content' => '', |
| 2841 | 3086 | // Empty map = every AI crawler allowed. Defaults must stay |
| 2842 | 3087 | // permissive so an upgrade never starts blocking a crawler a site |
| 2843 | 3088 | // was happily serving (#657). |
| @@ -3051,16 +3296,17 @@ | ||
| 3051 | 3296 | // Remove extra whitespace |
| 3052 | 3297 | $title = preg_replace('/\s+/', ' ', $title); |
| 3053 | 3298 | $title = trim($title); |
| 3054 | 3299 | |
| 3055 | - // Ensure title is not too long (60 characters max for SEO) | |
| 3056 | - if (strlen($title) > 60) { | |
| 3057 | - // Try to truncate at word boundary | |
| 3058 | - $title = wp_trim_words($title, 8, '...'); | |
| 3059 | - if (strlen($title) > 60) { | |
| 3060 | - $title = substr($title, 0, 57) . '...'; | |
| 3061 | - } | |
| 3062 | - } | |
| 3300 | + // Ensure title is not too long (60 characters max for SEO). | |
| 3301 | + // All three units here were wrong for non-Latin text: strlen() counts | |
| 3302 | + // BYTES so the gate fired at 20 Thai characters, wp_trim_words() counts | |
| 3303 | + // CHARACTERS on th/ja/zh_* so `8` cut the title to 8 of them, and | |
| 3304 | + // substr() cuts bytes so it split a character mid-sequence (#687). | |
| 3305 | + $title = \ThinkRank\Core\Seo_Text::trim_to_length( | |
| 3306 | + $title, | |
| 3307 | + \ThinkRank\Core\Seo_Text::TITLE_MAX_LENGTH | |
| 3308 | + ); | |
| 3063 | 3309 | |
| 3064 | 3310 | // Ensure title is not empty |
| 3065 | 3311 | if (empty($title)) { |
| 3066 | 3312 | $title = get_bloginfo('name'); |
| @@ -3450,8 +3696,29 @@ | ||
| 3450 | 3696 | * @since 1.0.0 |
| 3451 | 3697 | * @return array Array of sitemap URLs |
| 3452 | 3698 | */ |
| 3453 | 3699 | private function get_sitemap_urls_for_robots(): array { |
| 3700 | + // One wrapper over every return path below, including the #104 extras. | |
| 3701 | + // The Sitemap: line is the only absolute URL of ours in robots.txt and | |
| 3702 | + // the one a crawler follows to find everything else, so it has to carry | |
| 3703 | + // the site's scheme preference (#638). Applied here rather than where | |
| 3704 | + // the body is assembled, because that path also renders a robots.txt a | |
| 3705 | + // site owner typed themselves, and their text is not ours to rewrite. | |
| 3706 | + return array_map( | |
| 3707 | + static function (string $url): string { | |
| 3708 | + return Url_Scheme::apply($url); | |
| 3709 | + }, | |
| 3710 | + $this->collect_sitemap_urls_for_robots() | |
| 3711 | + ); | |
| 3712 | + } | |
| 3713 | + | |
| 3714 | + /** | |
| 3715 | + * The sitemap URLs robots.txt advertises, before the scheme preference. | |
| 3716 | + * | |
| 3717 | + * @since 1.0.0 | |
| 3718 | + * @return array Array of sitemap URLs | |
| 3719 | + */ | |
| 3720 | + private function collect_sitemap_urls_for_robots(): array { | |
| 3454 | 3721 | try { |
| 3455 | 3722 | // Get sitemap settings |
| 3456 | 3723 | $sitemap_generator = new \ThinkRank\SEO\Sitemap_Generator(); |
| 3457 | 3724 | $sitemap_settings = $sitemap_generator->get_settings('site'); |
| @@ -3500,9 +3767,22 @@ | ||
| 3500 | 3767 | // business sitemap and the sitemaps other plugins register both land |
| 3501 | 3768 | // here for the same reason, so they go through one list (#104). |
| 3502 | 3769 | $extra = []; |
| 3503 | 3770 | |
| 3504 | - if (file_exists(ABSPATH . 'local-sitemap.xml')) { | |
| 3771 | + // Not a file test. Under dynamic delivery the local sitemap is | |
| 3772 | + // served from PHP and no file is ever written, so file_exists() | |
| 3773 | + // silently dropped a sitemap the site really does publish (#752). | |
| 3774 | + // On static sites the file is still what proves it, so both count. | |
| 3775 | + $local_sitemap_published = file_exists(ABSPATH . 'local-sitemap.xml'); | |
| 3776 | + | |
| 3777 | + if (!$local_sitemap_published && class_exists('ThinkRank\\SEO\\Sitemap_Generator')) { | |
| 3778 | + $generator = new \ThinkRank\SEO\Sitemap_Generator(false); | |
| 3779 | + | |
| 3780 | + $local_sitemap_published = 'dynamic' === $generator->resolve_delivery_mode() | |
| 3781 | + && $generator->publishes_local_sitemap(); | |
| 3782 | + } | |
| 3783 | + | |
| 3784 | + if ($local_sitemap_published) { | |
| 3505 | 3785 | $extra[] = '/local-sitemap.xml'; |
| 3506 | 3786 | } |
| 3507 | 3787 | |
| 3508 | 3788 | foreach (\ThinkRank\SEO\Sitemap_Generator::additional_sitemaps() as $path) { |
| @@ -3927,11 +4207,14 @@ | ||
| 3927 | 4207 | $optimization['validation']['valid'] = false; |
| 3928 | 4208 | } |
| 3929 | 4209 | |
| 3930 | 4210 | if (!empty($value) && isset($config['max_length'])) { |
| 3931 | - if (strlen($value) > $config['max_length']) { | |
| 4211 | + // The warning says "characters", so measure and cut in characters: | |
| 4212 | + // strlen()/substr() fired early on non-Latin values and the | |
| 4213 | + // suggested replacement was cut mid-character (#687). | |
| 4214 | + if (mb_strlen($value) > $config['max_length']) { | |
| 3932 | 4215 | $optimization['validation']['warnings'][] = "{$element} exceeds maximum length of {$config['max_length']} characters"; |
| 3933 | - $optimization['optimized_value'] = substr($value, 0, $config['max_length']); | |
| 4216 | + $optimization['optimized_value'] = \ThinkRank\Core\Seo_Text::trim_to_length($value, (int) $config['max_length']); | |
| 3934 | 4217 | } |
| 3935 | 4218 | } |
| 3936 | 4219 | |
| 3937 | 4220 | // SEO-specific optimizations |
| @@ -3970,18 +4253,20 @@ | ||
| 3970 | 4253 | return $optimization; |
| 3971 | 4254 | } |
| 3972 | 4255 | |
| 3973 | 4256 | // Check if it's a local image |
| 3974 | - $attachment_id = attachment_url_to_postid($value); | |
| 4257 | + $attachment_id = Attachment_Lookup::id_from_url($value); | |
| 3975 | 4258 | if ($attachment_id) { |
| 3976 | 4259 | $image_meta = wp_get_attachment_metadata($attachment_id); |
| 3977 | 4260 | |
| 3978 | 4261 | if ($image_meta && isset($image_meta['width'], $image_meta['height'])) { |
| 3979 | - // Check recommended size | |
| 4262 | + // Check recommended size, against the configured file itself | |
| 4263 | + // rather than the upload it may have been generated from. | |
| 3980 | 4264 | if (isset($config['recommended_size'])) { |
| 3981 | 4265 | [$rec_width, $rec_height] = explode('x', $config['recommended_size']); |
| 4266 | + $image_file = Attachment_Lookup::describe($attachment_id, $value); | |
| 3982 | 4267 | |
| 3983 | - if ((int) $image_meta['width'] !== (int) $rec_width || (int) $image_meta['height'] !== (int) $rec_height) { | |
| 4268 | + if ($image_file['width'] !== (int) $rec_width || $image_file['height'] !== (int) $rec_height) { | |
| 3984 | 4269 | $optimization['suggestions'][] = "Consider using {$config['recommended_size']} size for optimal {$element}"; |
| 3985 | 4270 | } |
| 3986 | 4271 | } |
| 3987 | 4272 | |
