| @@ -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 |
| @@ -1178,17 +1289,19 @@ | ||
| 1178 | 1289 | $home_text = $settings['breadcrumb_home_text'] ?? 'Home'; |
| 1179 | 1290 | if (empty($home_text)) { |
| 1180 | 1291 | $optimization['warnings'][] = 'Empty home text reduces accessibility for screen readers'; |
| 1181 | 1292 | $optimization['score'] -= 15; |
| 1182 | - } elseif (strlen($home_text) > 20) { | |
| 1183 | - $optimization['suggestions'][] = 'Keep home text concise (current: ' . strlen($home_text) . ' chars)'; | |
| 1293 | + } elseif (mb_strlen($home_text) > 20) { | |
| 1294 | + // mb_strlen: this number is shown to the user as "chars" (#687). | |
| 1295 | + $optimization['suggestions'][] = 'Keep home text concise (current: ' . mb_strlen($home_text) . ' chars)'; | |
| 1184 | 1296 | $optimization['score'] -= 5; |
| 1185 | 1297 | } |
| 1186 | 1298 | |
| 1187 | 1299 | // Check prefix usage |
| 1188 | 1300 | $prefix = $settings['breadcrumb_prefix'] ?? ''; |
| 1189 | - if (!empty($prefix) && strlen($prefix) > 50) { | |
| 1190 | - $optimization['suggestions'][] = 'Breadcrumb prefix is quite long (' . strlen($prefix) . ' chars) - consider shortening'; | |
| 1301 | + if (!empty($prefix) && mb_strlen($prefix) > 50) { | |
| 1302 | + // mb_strlen: this number is shown to the user as "chars" (#687). | |
| 1303 | + $optimization['suggestions'][] = 'Breadcrumb prefix is quite long (' . mb_strlen($prefix) . ' chars) - consider shortening'; | |
| 1191 | 1304 | $optimization['score'] -= 5; |
| 1192 | 1305 | } |
| 1193 | 1306 | |
| 1194 | 1307 | // Current page display |
| @@ -2759,8 +2872,47 @@ | ||
| 2759 | 2872 | * @since 2.0.1 |
| 2760 | 2873 | * |
| 2761 | 2874 | * @return string[] |
| 2762 | 2875 | */ |
| 2876 | + /** | |
| 2877 | + * The stored alternate name(s), shaped for schema output. | |
| 2878 | + * | |
| 2879 | + * schema.org and Google both allow `alternateName` to carry one value or | |
| 2880 | + * several, and the store already round-trips either shape, so this accepts | |
| 2881 | + * both and normalises: null when there is nothing to publish, a bare string | |
| 2882 | + * for one name, a list for more. Emitting a one-element array would be | |
| 2883 | + * valid but noisier than it needs to be. | |
| 2884 | + * | |
| 2885 | + * Shared because both WebSite producers need it and must agree — a property | |
| 2886 | + * added to one and not the other is how #688 happened. | |
| 2887 | + * | |
| 2888 | + * @since 2.7.0 | |
| 2889 | + * | |
| 2890 | + * @param mixed $value Stored alternate_name value. | |
| 2891 | + * @return string|string[]|null | |
| 2892 | + */ | |
| 2893 | + public static function alternate_name_for_schema($value) { | |
| 2894 | + $names = []; | |
| 2895 | + | |
| 2896 | + foreach ((array) $value as $name) { | |
| 2897 | + if (!is_scalar($name)) { | |
| 2898 | + continue; | |
| 2899 | + } | |
| 2900 | + | |
| 2901 | + $name = trim((string) $name); | |
| 2902 | + | |
| 2903 | + if ('' !== $name && !in_array($name, $names, true)) { | |
| 2904 | + $names[] = $name; | |
| 2905 | + } | |
| 2906 | + } | |
| 2907 | + | |
| 2908 | + if (empty($names)) { | |
| 2909 | + return null; | |
| 2910 | + } | |
| 2911 | + | |
| 2912 | + return 1 === count($names) ? $names[0] : $names; | |
| 2913 | + } | |
| 2914 | + | |
| 2763 | 2915 | protected function additional_setting_keys(): array { |
| 2764 | 2916 | return [ |
| 2765 | 2917 | // Title formats, one per context. |
| 2766 | 2918 | 'homepage_title', 'post_title', 'page_title', 'category_title', |
| @@ -2811,8 +2963,18 @@ | ||
| 2811 | 2963 | if (array_key_exists('ai_crawler_rules', $sanitized)) { |
| 2812 | 2964 | $sanitized['ai_crawler_rules'] = AI_Crawlers::normalize_rules($sanitized['ai_crawler_rules']); |
| 2813 | 2965 | } |
| 2814 | 2966 | |
| 2967 | + // Same reasoning one key up, for the scheme override (#638). Anything | |
| 2968 | + // that is not one of the three modes means "follow WordPress", and is | |
| 2969 | + // stored as that rather than kept verbatim — otherwise get-site-identity | |
| 2970 | + // -settings would report a scheme the site does not actually publish. | |
| 2971 | + if (array_key_exists('canonical_scheme', $sanitized)) { | |
| 2972 | + $sanitized['canonical_scheme'] = in_array($sanitized['canonical_scheme'], Url_Scheme::MODES, true) | |
| 2973 | + ? $sanitized['canonical_scheme'] | |
| 2974 | + : Url_Scheme::AUTOMATIC; | |
| 2975 | + } | |
| 2976 | + | |
| 2815 | 2977 | return $sanitized; |
| 2816 | 2978 | } |
| 2817 | 2979 | |
| 2818 | 2980 | /** |
| @@ -2836,8 +2998,27 @@ | ||
| 2836 | 2998 | 'breadcrumb_home_text' => 'Home', |
| 2837 | 2999 | 'breadcrumb_separator' => '>', |
| 2838 | 3000 | 'robots_txt_enabled' => true, |
| 2839 | 3001 | 'allow_search_engines' => true, |
| 3002 | + // Answer 404 when a content selector in the URL resolved to | |
| 3003 | + // nothing (#634). On by default, unlike the other new settings | |
| 3004 | + // here: it changes no URL a visitor or a correct crawler uses, only | |
| 3005 | + // ones where WordPress resolved nothing and served the blog listing | |
| 3006 | + // at 200 anyway. | |
| 3007 | + 'query_protection' => true, | |
| 3008 | + | |
| 3009 | + // Feed controls (#635). All three off, so an upgrade changes | |
| 3010 | + // nothing about what an existing site already sends its | |
| 3011 | + // subscribers; a brand-new install is seeded with the signature and | |
| 3012 | + // the noindex on, in Activator::seed_feed_defaults(). | |
| 3013 | + 'feed_excerpt_only' => false, | |
| 3014 | + 'feed_source_link' => false, | |
| 3015 | + 'feed_noindex' => false, | |
| 3016 | + | |
| 3017 | + // The scheme self-referential URLs go out with (#638). 'automatic' | |
| 3018 | + // means substitute nothing and follow WordPress, which is what | |
| 3019 | + // every site did before the setting existed. | |
| 3020 | + 'canonical_scheme' => Url_Scheme::AUTOMATIC, | |
| 2840 | 3021 | 'robots_txt_content' => '', |
| 2841 | 3022 | // Empty map = every AI crawler allowed. Defaults must stay |
| 2842 | 3023 | // permissive so an upgrade never starts blocking a crawler a site |
| 2843 | 3024 | // was happily serving (#657). |
| @@ -3051,16 +3232,17 @@ | ||
| 3051 | 3232 | // Remove extra whitespace |
| 3052 | 3233 | $title = preg_replace('/\s+/', ' ', $title); |
| 3053 | 3234 | $title = trim($title); |
| 3054 | 3235 | |
| 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 | - } | |
| 3236 | + // Ensure title is not too long (60 characters max for SEO). | |
| 3237 | + // All three units here were wrong for non-Latin text: strlen() counts | |
| 3238 | + // BYTES so the gate fired at 20 Thai characters, wp_trim_words() counts | |
| 3239 | + // CHARACTERS on th/ja/zh_* so `8` cut the title to 8 of them, and | |
| 3240 | + // substr() cuts bytes so it split a character mid-sequence (#687). | |
| 3241 | + $title = \ThinkRank\Core\Seo_Text::trim_to_length( | |
| 3242 | + $title, | |
| 3243 | + \ThinkRank\Core\Seo_Text::TITLE_MAX_LENGTH | |
| 3244 | + ); | |
| 3063 | 3245 | |
| 3064 | 3246 | // Ensure title is not empty |
| 3065 | 3247 | if (empty($title)) { |
| 3066 | 3248 | $title = get_bloginfo('name'); |
| @@ -3450,8 +3632,29 @@ | ||
| 3450 | 3632 | * @since 1.0.0 |
| 3451 | 3633 | * @return array Array of sitemap URLs |
| 3452 | 3634 | */ |
| 3453 | 3635 | private function get_sitemap_urls_for_robots(): array { |
| 3636 | + // One wrapper over every return path below, including the #104 extras. | |
| 3637 | + // The Sitemap: line is the only absolute URL of ours in robots.txt and | |
| 3638 | + // the one a crawler follows to find everything else, so it has to carry | |
| 3639 | + // the site's scheme preference (#638). Applied here rather than where | |
| 3640 | + // the body is assembled, because that path also renders a robots.txt a | |
| 3641 | + // site owner typed themselves, and their text is not ours to rewrite. | |
| 3642 | + return array_map( | |
| 3643 | + static function (string $url): string { | |
| 3644 | + return Url_Scheme::apply($url); | |
| 3645 | + }, | |
| 3646 | + $this->collect_sitemap_urls_for_robots() | |
| 3647 | + ); | |
| 3648 | + } | |
| 3649 | + | |
| 3650 | + /** | |
| 3651 | + * The sitemap URLs robots.txt advertises, before the scheme preference. | |
| 3652 | + * | |
| 3653 | + * @since 1.0.0 | |
| 3654 | + * @return array Array of sitemap URLs | |
| 3655 | + */ | |
| 3656 | + private function collect_sitemap_urls_for_robots(): array { | |
| 3454 | 3657 | try { |
| 3455 | 3658 | // Get sitemap settings |
| 3456 | 3659 | $sitemap_generator = new \ThinkRank\SEO\Sitemap_Generator(); |
| 3457 | 3660 | $sitemap_settings = $sitemap_generator->get_settings('site'); |
| @@ -3500,9 +3703,22 @@ | ||
| 3500 | 3703 | // business sitemap and the sitemaps other plugins register both land |
| 3501 | 3704 | // here for the same reason, so they go through one list (#104). |
| 3502 | 3705 | $extra = []; |
| 3503 | 3706 | |
| 3504 | - if (file_exists(ABSPATH . 'local-sitemap.xml')) { | |
| 3707 | + // Not a file test. Under dynamic delivery the local sitemap is | |
| 3708 | + // served from PHP and no file is ever written, so file_exists() | |
| 3709 | + // silently dropped a sitemap the site really does publish (#752). | |
| 3710 | + // On static sites the file is still what proves it, so both count. | |
| 3711 | + $local_sitemap_published = file_exists(ABSPATH . 'local-sitemap.xml'); | |
| 3712 | + | |
| 3713 | + if (!$local_sitemap_published && class_exists('ThinkRank\\SEO\\Sitemap_Generator')) { | |
| 3714 | + $generator = new \ThinkRank\SEO\Sitemap_Generator(false); | |
| 3715 | + | |
| 3716 | + $local_sitemap_published = 'dynamic' === $generator->resolve_delivery_mode() | |
| 3717 | + && $generator->publishes_local_sitemap(); | |
| 3718 | + } | |
| 3719 | + | |
| 3720 | + if ($local_sitemap_published) { | |
| 3505 | 3721 | $extra[] = '/local-sitemap.xml'; |
| 3506 | 3722 | } |
| 3507 | 3723 | |
| 3508 | 3724 | foreach (\ThinkRank\SEO\Sitemap_Generator::additional_sitemaps() as $path) { |
| @@ -3927,11 +4143,14 @@ | ||
| 3927 | 4143 | $optimization['validation']['valid'] = false; |
| 3928 | 4144 | } |
| 3929 | 4145 | |
| 3930 | 4146 | if (!empty($value) && isset($config['max_length'])) { |
| 3931 | - if (strlen($value) > $config['max_length']) { | |
| 4147 | + // The warning says "characters", so measure and cut in characters: | |
| 4148 | + // strlen()/substr() fired early on non-Latin values and the | |
| 4149 | + // suggested replacement was cut mid-character (#687). | |
| 4150 | + if (mb_strlen($value) > $config['max_length']) { | |
| 3932 | 4151 | $optimization['validation']['warnings'][] = "{$element} exceeds maximum length of {$config['max_length']} characters"; |
| 3933 | - $optimization['optimized_value'] = substr($value, 0, $config['max_length']); | |
| 4152 | + $optimization['optimized_value'] = \ThinkRank\Core\Seo_Text::trim_to_length($value, (int) $config['max_length']); | |
| 3934 | 4153 | } |
| 3935 | 4154 | } |
| 3936 | 4155 | |
| 3937 | 4156 | // SEO-specific optimizations |
