| 1 |
<?php |
| 2 |
|
| 3 |
/** |
| 4 |
* Site Identity Manager Class |
| 5 |
* |
| 6 |
* Comprehensive site identity management with title formats, separators, |
| 7 |
* breadcrumb navigation, robots.txt management, and site identity optimization. |
| 8 |
* Implements 2025 SEO best practices with real industry-standard algorithms. |
| 9 |
* |
| 10 |
* @package ThinkRank |
| 11 |
* @subpackage SEO |
| 12 |
* @since 1.0.0 |
| 13 |
*/ |
| 14 |
|
| 15 |
declare(strict_types=1); |
| 16 |
|
| 17 |
namespace ThinkRank\SEO; |
| 18 |
|
| 19 |
// Prevent direct access |
| 20 |
if (!defined('ABSPATH')) { |
| 21 |
exit; |
| 22 |
} |
| 23 |
|
| 24 |
// Ensure dependencies are loaded |
| 25 |
if (!class_exists('ThinkRank\\SEO\\Abstract_SEO_Manager')) { |
| 26 |
require_once THINKRANK_PLUGIN_DIR . 'includes/seo/class-abstract-seo-manager.php'; |
| 27 |
} |
| 28 |
|
| 29 |
if (!interface_exists('ThinkRank\\SEO\\Interfaces\\SEO_Manager_Interface')) { |
| 30 |
require_once THINKRANK_PLUGIN_DIR . 'includes/seo/interfaces/class-seo-manager-interface.php'; |
| 31 |
} |
| 32 |
|
| 33 |
/** |
| 34 |
* Site Identity Manager Class |
| 35 |
* |
| 36 |
* Manages all aspects of site identity including title formats, breadcrumbs, |
| 37 |
* robots.txt, and global SEO settings with context-aware optimization. |
| 38 |
* |
| 39 |
* @since 1.0.0 |
| 40 |
*/ |
| 41 |
class Site_Identity_Manager extends Abstract_SEO_Manager { |
| 42 |
|
| 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 |
/** |
| 72 |
* WordPress filesystem instance |
| 73 |
* |
| 74 |
* @since 1.0.0 |
| 75 |
* @var \WP_Filesystem_Base|null |
| 76 |
*/ |
| 77 |
private $filesystem = null; |
| 78 |
|
| 79 |
/** |
| 80 |
* Title format templates with dynamic placeholders |
| 81 |
* |
| 82 |
* @since 1.0.0 |
| 83 |
* @var array |
| 84 |
*/ |
| 85 |
private array $title_templates = [ |
| 86 |
'default' => '%title% %separator% %sitename%', |
| 87 |
'reverse' => '%sitename% %separator% %title%', |
| 88 |
'title_only' => '%title%', |
| 89 |
'sitename_only' => '%sitename%', |
| 90 |
'custom' => '%title% %separator% %sitename% %separator% %tagline%', |
| 91 |
'category' => '%title% %separator% %category% %separator% %sitename%', |
| 92 |
'author' => '%title% %separator% %author% %separator% %sitename%', |
| 93 |
'date' => '%title% %separator% %date% %separator% %sitename%', |
| 94 |
'search' => 'Search Results for "%searchterm%" %separator% %sitename%', |
| 95 |
'404' => 'Page Not Found %separator% %sitename%' |
| 96 |
]; |
| 97 |
|
| 98 |
/** |
| 99 |
* Available title separators with their specifications |
| 100 |
* |
| 101 |
* @since 1.0.0 |
| 102 |
* @var array |
| 103 |
*/ |
| 104 |
public static array $title_separators = [ |
| 105 |
'pipe' => [ |
| 106 |
'symbol' => '|', |
| 107 |
'name' => 'Pipe', |
| 108 |
'description' => 'Vertical bar separator (most common)', |
| 109 |
'seo_score' => 10 |
| 110 |
], |
| 111 |
'dash' => [ |
| 112 |
'symbol' => '-', |
| 113 |
'name' => 'Dash', |
| 114 |
'description' => 'Hyphen separator (clean and readable)', |
| 115 |
'seo_score' => 9 |
| 116 |
], |
| 117 |
'bullet' => [ |
| 118 |
'symbol' => '•', |
| 119 |
'name' => 'Bullet', |
| 120 |
'description' => 'Bullet point separator (modern)', |
| 121 |
'seo_score' => 8 |
| 122 |
], |
| 123 |
'colon' => [ |
| 124 |
'symbol' => ':', |
| 125 |
'name' => 'Colon', |
| 126 |
'description' => 'Colon separator (formal)', |
| 127 |
'seo_score' => 7 |
| 128 |
], |
| 129 |
'greater' => [ |
| 130 |
'symbol' => '>', |
| 131 |
'name' => 'Greater Than', |
| 132 |
'description' => 'Arrow-like separator (hierarchical)', |
| 133 |
'seo_score' => 6 |
| 134 |
], |
| 135 |
'tilde' => [ |
| 136 |
'symbol' => '~', |
| 137 |
'name' => 'Tilde', |
| 138 |
'description' => 'Wave separator (unique)', |
| 139 |
'seo_score' => 5 |
| 140 |
] |
| 141 |
]; |
| 142 |
|
| 143 |
/** |
| 144 |
* Get the currently active title separator symbol |
| 145 |
* |
| 146 |
* @since 1.0.0 |
| 147 |
* @return string Separator symbol |
| 148 |
*/ |
| 149 |
public static function get_active_separator_symbol(): string { |
| 150 |
$manager = new self(); |
| 151 |
$settings = $manager->get_settings('site'); |
| 152 |
$separator_key = $settings['title_separator'] ?? 'pipe'; |
| 153 |
|
| 154 |
return self::$title_separators[$separator_key]['symbol'] ?? '|'; |
| 155 |
} |
| 156 |
|
| 157 |
/** |
| 158 |
* Breadcrumb types and their configurations |
| 159 |
* |
| 160 |
* @since 1.0.0 |
| 161 |
* @var array |
| 162 |
*/ |
| 163 |
private array $breadcrumb_types = [ |
| 164 |
'hierarchical' => [ |
| 165 |
'name' => 'Hierarchical', |
| 166 |
'description' => 'Based on page hierarchy and categories', |
| 167 |
'schema_type' => 'BreadcrumbList', |
| 168 |
'seo_value' => 10 |
| 169 |
], |
| 170 |
'taxonomy' => [ |
| 171 |
'name' => 'Taxonomy-based', |
| 172 |
'description' => 'Based on post categories and tags', |
| 173 |
'schema_type' => 'BreadcrumbList', |
| 174 |
'seo_value' => 9 |
| 175 |
], |
| 176 |
'path' => [ |
| 177 |
'name' => 'URL Path', |
| 178 |
'description' => 'Based on URL structure', |
| 179 |
'schema_type' => 'BreadcrumbList', |
| 180 |
'seo_value' => 8 |
| 181 |
], |
| 182 |
'custom' => [ |
| 183 |
'name' => 'Custom', |
| 184 |
'description' => 'Manually defined breadcrumb structure', |
| 185 |
'schema_type' => 'BreadcrumbList', |
| 186 |
'seo_value' => 7 |
| 187 |
] |
| 188 |
]; |
| 189 |
|
| 190 |
/** |
| 191 |
* Robots.txt directives and their specifications |
| 192 |
* |
| 193 |
* @since 1.0.0 |
| 194 |
* @var array |
| 195 |
*/ |
| 196 |
private array $robots_directives = [ |
| 197 |
'user_agent' => [ |
| 198 |
'required' => true, |
| 199 |
'description' => 'Specifies which web crawler the rules apply to', |
| 200 |
'examples' => ['*', 'Googlebot', 'Bingbot', 'Yandexbot'] |
| 201 |
], |
| 202 |
'disallow' => [ |
| 203 |
'required' => false, |
| 204 |
'description' => 'Specifies paths that should not be crawled', |
| 205 |
'examples' => ['/admin/', '/wp-admin/', '/wp-includes/', '/private/'] |
| 206 |
], |
| 207 |
'allow' => [ |
| 208 |
'required' => false, |
| 209 |
'description' => 'Specifies paths that should be crawled (overrides disallow)', |
| 210 |
'examples' => ['/wp-admin/admin-ajax.php', '/wp-content/uploads/'] |
| 211 |
], |
| 212 |
'sitemap' => [ |
| 213 |
'required' => false, |
| 214 |
'description' => 'Specifies the location of XML sitemaps', |
| 215 |
'examples' => ['/sitemap.xml', '/sitemap_index.xml'] |
| 216 |
], |
| 217 |
'crawl_delay' => [ |
| 218 |
'required' => false, |
| 219 |
'description' => 'Specifies delay between requests (in seconds)', |
| 220 |
'examples' => ['1', '5', '10'] |
| 221 |
] |
| 222 |
]; |
| 223 |
|
| 224 |
/** |
| 225 |
* Site identity elements configuration |
| 226 |
* |
| 227 |
* @since 1.0.0 |
| 228 |
* @var array |
| 229 |
*/ |
| 230 |
private array $identity_elements = [ |
| 231 |
'logo' => [ |
| 232 |
'type' => 'image', |
| 233 |
'required' => false, |
| 234 |
'description' => 'Site logo for branding and schema markup', |
| 235 |
'recommended_size' => '600x60', |
| 236 |
'max_size' => '2MB' |
| 237 |
], |
| 238 |
'favicon' => [ |
| 239 |
'type' => 'image', |
| 240 |
'required' => false, |
| 241 |
'description' => 'Site favicon for browser tabs', |
| 242 |
'recommended_size' => '32x32', |
| 243 |
'formats' => ['ico', 'png'] |
| 244 |
], |
| 245 |
'apple_touch_icon' => [ |
| 246 |
'type' => 'image', |
| 247 |
'required' => false, |
| 248 |
'description' => 'Apple touch icon for iOS devices', |
| 249 |
'recommended_size' => '180x180', |
| 250 |
'format' => 'png' |
| 251 |
], |
| 252 |
'site_name' => [ |
| 253 |
'type' => 'text', |
| 254 |
'required' => true, |
| 255 |
'description' => 'Official site name for branding', |
| 256 |
'max_length' => 60 |
| 257 |
], |
| 258 |
'tagline' => [ |
| 259 |
'type' => 'text', |
| 260 |
'required' => false, |
| 261 |
'description' => 'Site tagline or slogan', |
| 262 |
'max_length' => 160 |
| 263 |
], |
| 264 |
'description' => [ |
| 265 |
'type' => 'text', |
| 266 |
'required' => false, |
| 267 |
'description' => 'Site description for meta tags', |
| 268 |
'max_length' => 160 |
| 269 |
] |
| 270 |
]; |
| 271 |
|
| 272 |
/** |
| 273 |
* Constructor |
| 274 |
* |
| 275 |
* @since 1.0.0 |
| 276 |
*/ |
| 277 |
/** |
| 278 |
* The square derivatives wp_site_icon() asks for. |
| 279 |
* |
| 280 |
* Core generates these only through its own Site Icon crop flow, so an |
| 281 |
* image chosen as a ThinkRank favicon straight from the media library has |
| 282 |
* none of them and every sizes="" declaration is a near miss (#571). |
| 283 |
* |
| 284 |
* @since 2.3.1 |
| 285 |
* @var int[] |
| 286 |
*/ |
| 287 |
public const ICON_SIZES = [32, 180, 192, 270]; |
| 288 |
|
| 289 |
/** |
| 290 |
* Transient holding resolved icon URLs, keyed by configured URL and size. |
| 291 |
* |
| 292 |
* The site-icon filter runs in wp_head on every FRONT-END request, and |
| 293 |
* resolving a URL to its attachment costs an uncached postmeta query. The |
| 294 |
* mapping only changes when the icon setting does, so it is cached here and |
| 295 |
* dropped on save. |
| 296 |
* |
| 297 |
* @since 2.3.1 |
| 298 |
* @var string |
| 299 |
*/ |
| 300 |
public const ICON_URL_TRANSIENT = 'thinkrank_site_icon_urls'; |
| 301 |
|
| 302 |
/** |
| 303 |
* Marker for the one-time derivative backfill on existing installs. |
| 304 |
* |
| 305 |
* @since 2.3.1 |
| 306 |
* @var string |
| 307 |
*/ |
| 308 |
public const ICON_BACKFILL_OPTION = 'thinkrank_site_icon_sizes_backfilled'; |
| 309 |
|
| 310 |
/** |
| 311 |
* Whether the icon-derivative listener has been registered this request. |
| 312 |
* |
| 313 |
* Static because `thinkrank_seo_settings_saved` is a global hook — one |
| 314 |
* listener serves every instance, and this class is constructed on the |
| 315 |
* front end as well as in admin. |
| 316 |
* |
| 317 |
* @since 2.3.1 |
| 318 |
* @var bool |
| 319 |
*/ |
| 320 |
private static bool $icon_sizes_listener_registered = false; |
| 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 $robots_sync_listener_registered = false; |
| 329 |
|
| 330 |
/** |
| 331 |
* Flag set when a plugin change may have altered the sitemap set. |
| 332 |
* |
| 333 |
* @since 2.14.0 |
| 334 |
* @var string |
| 335 |
*/ |
| 336 |
public const ROBOTS_RESYNC_OPTION = 'thinkrank_robots_txt_resync_pending'; |
| 337 |
|
| 338 |
/** |
| 339 |
* Flag set when a plugin change may have altered the sitemap index. |
| 340 |
* |
| 341 |
* Separate from ROBOTS_RESYNC_OPTION because the two files exist |
| 342 |
* independently: that flag is only set when a physical robots.txt exists, |
| 343 |
* and the sitemap index needs rebuilding whether or not it does. |
| 344 |
* |
| 345 |
* @since 2.15.0 |
| 346 |
* @var string |
| 347 |
*/ |
| 348 |
public const SITEMAP_RESYNC_OPTION = 'thinkrank_sitemap_contributors_changed'; |
| 349 |
|
| 350 |
public function __construct() { |
| 351 |
parent::__construct('site_identity'); |
| 352 |
|
| 353 |
if (!self::$icon_sizes_listener_registered) { |
| 354 |
self::$icon_sizes_listener_registered = true; |
| 355 |
add_action('thinkrank_seo_settings_saved', [$this, 'generate_icon_sizes_on_save'], 10, 2); |
| 356 |
// Admin only: resizing is not front-end work, and admin traffic is |
| 357 |
// enough to run a one-time backfill promptly. |
| 358 |
add_action('admin_init', [self::class, 'maybe_backfill_icon_sizes']); |
| 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 |
} |
| 383 |
} |
| 384 |
|
| 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 |
/** |
| 482 |
* Save settings, then refresh what a new canonical scheme invalidates. |
| 483 |
* |
| 484 |
* The static sitemap files are written with the scheme in force when they |
| 485 |
* were built, and nothing else rebuilds them until a post or term changes. |
| 486 |
* So a change of scheme left every `<loc>` on the old one while canonical |
| 487 |
* and og:url had already moved (#736). Every writer (the settings route, |
| 488 |
* the robots route, the MCP abilities, an import) lands here. |
| 489 |
* |
| 490 |
* @since 2.7.0 |
| 491 |
* |
| 492 |
* @param string $context_type Context type. |
| 493 |
* @param int|null $context_id Context ID. |
| 494 |
* @param array $settings Settings to save. |
| 495 |
* @return bool |
| 496 |
*/ |
| 497 |
public function save_settings(string $context_type, ?int $context_id, array $settings): bool { |
| 498 |
if (!self::touches_canonical_scheme($context_type, $context_id, $settings)) { |
| 499 |
return parent::save_settings($context_type, $context_id, $settings); |
| 500 |
} |
| 501 |
|
| 502 |
$before = Url_Scheme::preference(); |
| 503 |
$saved = parent::save_settings($context_type, $context_id, $settings); |
| 504 |
|
| 505 |
if ($saved) { |
| 506 |
$this->on_canonical_scheme_saved($before); |
| 507 |
} |
| 508 |
|
| 509 |
return $saved; |
| 510 |
} |
| 511 |
|
| 512 |
/** |
| 513 |
* Whether a save can change the site-wide canonical scheme. |
| 514 |
* |
| 515 |
* @since 2.7.0 |
| 516 |
* |
| 517 |
* @param string $context_type Context type. |
| 518 |
* @param int|null $context_id Context ID. |
| 519 |
* @param array $settings Settings being saved. |
| 520 |
* @return bool |
| 521 |
*/ |
| 522 |
public static function touches_canonical_scheme(string $context_type, ?int $context_id, array $settings): bool { |
| 523 |
return 'site' === sanitize_key($context_type) |
| 524 |
&& empty($context_id) |
| 525 |
&& array_key_exists('canonical_scheme', $settings); |
| 526 |
} |
| 527 |
|
| 528 |
/** |
| 529 |
* Rebuild the static sitemaps when the effective scheme changed. |
| 530 |
* |
| 531 |
* Compares the effective preference, filter included, so a site whose |
| 532 |
* scheme is pinned by `thinkrank_canonical_scheme` does not rebuild on a |
| 533 |
* stored value that changes nothing it publishes. |
| 534 |
* |
| 535 |
* @since 2.7.0 |
| 536 |
* |
| 537 |
* @param string $before Effective scheme before the save. |
| 538 |
* @return void |
| 539 |
*/ |
| 540 |
protected function on_canonical_scheme_saved(string $before): void { |
| 541 |
// The preference is cached for the request; the save just changed it. |
| 542 |
Url_Scheme::reset(); |
| 543 |
|
| 544 |
if (Url_Scheme::preference() === $before) { |
| 545 |
return; |
| 546 |
} |
| 547 |
|
| 548 |
$this->schedule_sitemap_rebuild(); |
| 549 |
} |
| 550 |
|
| 551 |
/** |
| 552 |
* Queue a settings-driven sitemap rebuild. |
| 553 |
* |
| 554 |
* Debounced and run after the response, like any other settings change |
| 555 |
* that alters what the sitemap publishes. |
| 556 |
* |
| 557 |
* @since 2.7.0 |
| 558 |
* @return void |
| 559 |
*/ |
| 560 |
protected function schedule_sitemap_rebuild(): void { |
| 561 |
(new Sitemap_Generator(false))->schedule_regeneration(); |
| 562 |
} |
| 563 |
|
| 564 |
/** |
| 565 |
* Build the icon derivatives for a newly chosen favicon. |
| 566 |
* |
| 567 |
* Runs on save, which is the only moment the choice changes and the only |
| 568 |
* place image work belongs — resolving a size on the front end must stay a |
| 569 |
* lookup. Failure is silent by design: a missing derivative degrades to the |
| 570 |
* next best file, so a site whose host cannot resize still renders an icon. |
| 571 |
* |
| 572 |
* @since 2.3.1 |
| 573 |
* |
| 574 |
* @param string $manager_type Settings category that was saved. |
| 575 |
* @param array $settings The settings that were written. |
| 576 |
* @return void |
| 577 |
*/ |
| 578 |
public function generate_icon_sizes_on_save(string $manager_type, array $settings): void { |
| 579 |
if ('site_identity' !== $manager_type) { |
| 580 |
return; |
| 581 |
} |
| 582 |
|
| 583 |
// The choice, or the derivatives behind it, may have just changed. |
| 584 |
delete_transient(self::ICON_URL_TRANSIENT); |
| 585 |
|
| 586 |
foreach (['favicon_url', 'apple_touch_icon_url'] as $key) { |
| 587 |
if (empty($settings[$key]) || !is_string($settings[$key])) { |
| 588 |
continue; |
| 589 |
} |
| 590 |
|
| 591 |
$attachment_id = self::icon_attachment_id($settings[$key]); |
| 592 |
|
| 593 |
if ($attachment_id) { |
| 594 |
self::ensure_icon_sizes($attachment_id); |
| 595 |
} |
| 596 |
} |
| 597 |
|
| 598 |
// Dropped again after the resizes finish. Resizing is not instant, and a |
| 599 |
// front-end request arriving mid-generation would otherwise repopulate |
| 600 |
// the transient with the pre-derivative URLs and pin them for the full |
| 601 |
// TTL — leaving the sizes= declarations untrue until the next save. |
| 602 |
delete_transient(self::ICON_URL_TRANSIENT); |
| 603 |
} |
| 604 |
|
| 605 |
/** |
| 606 |
* Build the derivatives for a site that configured its icons before this |
| 607 |
* existed. |
| 608 |
* |
| 609 |
* generate_icon_sizes_on_save() only fires on a settings write, so every |
| 610 |
* site with an icon already chosen would keep serving whatever |
| 611 |
* wp_get_attachment_image_url() could find — in practice the 150x150 |
| 612 |
* thumbnail behind a sizes="32x32" declaration — until someone happened to |
| 613 |
* re-save Site Identity. That is the bug this is meant to fix, so the |
| 614 |
* derivatives are built once on upgrade instead of waiting for a save. |
| 615 |
* |
| 616 |
* Guarded by its own option rather than the plugin version so it runs once |
| 617 |
* and stays cheap: the check is a single autoloaded read on requests after |
| 618 |
* the first. |
| 619 |
* |
| 620 |
* @since 2.3.1 |
| 621 |
* |
| 622 |
* @return void |
| 623 |
*/ |
| 624 |
public static function maybe_backfill_icon_sizes(): void { |
| 625 |
if (get_option(self::ICON_BACKFILL_OPTION)) { |
| 626 |
return; |
| 627 |
} |
| 628 |
|
| 629 |
// Written before the work, not after: a host that cannot resize must |
| 630 |
// not retry on every admin request forever. |
| 631 |
update_option(self::ICON_BACKFILL_OPTION, time(), true); |
| 632 |
|
| 633 |
$settings = (new self())->get_settings('site'); |
| 634 |
|
| 635 |
if (!is_array($settings)) { |
| 636 |
return; |
| 637 |
} |
| 638 |
|
| 639 |
foreach (['favicon_url', 'apple_touch_icon_url'] as $key) { |
| 640 |
if (empty($settings[$key]) || !is_string($settings[$key])) { |
| 641 |
continue; |
| 642 |
} |
| 643 |
|
| 644 |
$attachment_id = self::icon_attachment_id($settings[$key]); |
| 645 |
|
| 646 |
if ($attachment_id) { |
| 647 |
self::ensure_icon_sizes($attachment_id); |
| 648 |
} |
| 649 |
} |
| 650 |
|
| 651 |
delete_transient(self::ICON_URL_TRANSIENT); |
| 652 |
} |
| 653 |
|
| 654 |
/** |
| 655 |
* Attachment ID behind a configured icon URL, or 0 when it is not ours. |
| 656 |
* |
| 657 |
* attachment_url_to_postid() matches _wp_attached_file, which holds the |
| 658 |
* ORIGINAL upload path, so the URL of a generated derivative |
| 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). |
| 663 |
* |
| 664 |
* Shared with SEO_Manager's site-icon filter so both sides of the feature |
| 665 |
* agree on which attachment a configured URL means. |
| 666 |
* |
| 667 |
* @since 2.3.1 |
| 668 |
* |
| 669 |
* @param string $url Configured icon URL. |
| 670 |
* @return int Attachment ID, or 0. |
| 671 |
*/ |
| 672 |
public static function icon_attachment_id(string $url): int { |
| 673 |
return Attachment_Lookup::id_from_url($url); |
| 674 |
} |
| 675 |
|
| 676 |
/** |
| 677 |
* Which ICON_SIZES derivatives this attachment still needs. |
| 678 |
* |
| 679 |
* Split out from the generation so the decision can be asserted on its |
| 680 |
* own: whether a size is skipped because it already exists or because it |
| 681 |
* would upscale is invisible once both answers are "nothing was built". |
| 682 |
* |
| 683 |
* A source is measured by its SHORTER edge — a 400x40 banner cannot yield |
| 684 |
* a true 192x192 — and anything reporting no dimensions at all (SVGs) is |
| 685 |
* left alone. |
| 686 |
* |
| 687 |
* @since 2.3.1 |
| 688 |
* |
| 689 |
* @param array $meta Attachment metadata. |
| 690 |
* @return array<string, array{width: int, height: int, crop: bool}> Sizes to build. |
| 691 |
*/ |
| 692 |
public static function missing_icon_sizes(array $meta): array { |
| 693 |
$source = min((int) ($meta['width'] ?? 0), (int) ($meta['height'] ?? 0)); |
| 694 |
|
| 695 |
if ($source < 1) { |
| 696 |
return []; |
| 697 |
} |
| 698 |
|
| 699 |
$wanted = []; |
| 700 |
foreach (self::ICON_SIZES as $size) { |
| 701 |
// Never upscale: a stretched source behind an accurate sizes="" |
| 702 |
// label is worse than the honest near miss it would replace. |
| 703 |
if (isset($meta['sizes']["site_icon-{$size}"]) || $size > $source) { |
| 704 |
continue; |
| 705 |
} |
| 706 |
|
| 707 |
$wanted["site_icon-{$size}"] = ['width' => $size, 'height' => $size, 'crop' => true]; |
| 708 |
} |
| 709 |
|
| 710 |
return $wanted; |
| 711 |
} |
| 712 |
|
| 713 |
/** |
| 714 |
* Generate whatever ICON_SIZES derivatives this attachment is missing. |
| 715 |
* |
| 716 |
* Only the missing ones, and never one larger than the source: upscaling a |
| 717 |
* small favicon would put a blurrier file behind an accurate sizes="" label |
| 718 |
* than the honest near-miss it replaced. |
| 719 |
* |
| 720 |
* @since 2.3.1 |
| 721 |
* |
| 722 |
* @param int $attachment_id Attachment to build derivatives for. |
| 723 |
* @return string[] Size names generated, empty when there was nothing to do. |
| 724 |
*/ |
| 725 |
public static function ensure_icon_sizes(int $attachment_id): array { |
| 726 |
$meta = wp_get_attachment_metadata($attachment_id); |
| 727 |
|
| 728 |
if (!is_array($meta)) { |
| 729 |
return []; |
| 730 |
} |
| 731 |
|
| 732 |
$wanted = self::missing_icon_sizes($meta); |
| 733 |
|
| 734 |
if (empty($wanted)) { |
| 735 |
return []; |
| 736 |
} |
| 737 |
|
| 738 |
$file = get_attached_file($attachment_id); |
| 739 |
|
| 740 |
if (!$file || !file_exists($file)) { |
| 741 |
return []; |
| 742 |
} |
| 743 |
|
| 744 |
$editor = wp_get_image_editor($file); |
| 745 |
|
| 746 |
if (is_wp_error($editor)) { |
| 747 |
return []; |
| 748 |
} |
| 749 |
|
| 750 |
$generated = $editor->multi_resize($wanted); |
| 751 |
|
| 752 |
if (empty($generated)) { |
| 753 |
return []; |
| 754 |
} |
| 755 |
|
| 756 |
$meta['sizes'] = array_merge($meta['sizes'] ?? [], $generated); |
| 757 |
wp_update_attachment_metadata($attachment_id, $meta); |
| 758 |
|
| 759 |
return array_keys($generated); |
| 760 |
} |
| 761 |
|
| 762 |
/** |
| 763 |
* Initialize WordPress filesystem |
| 764 |
* |
| 765 |
* @since 1.0.0 |
| 766 |
* @return bool True if filesystem is initialized, false otherwise |
| 767 |
*/ |
| 768 |
private function init_filesystem(): bool { |
| 769 |
if ($this->filesystem !== null) { |
| 770 |
return true; |
| 771 |
} |
| 772 |
|
| 773 |
global $wp_filesystem; |
| 774 |
|
| 775 |
if (!function_exists('WP_Filesystem')) { |
| 776 |
require_once ABSPATH . 'wp-admin/includes/file.php'; |
| 777 |
} |
| 778 |
|
| 779 |
$credentials = request_filesystem_credentials('', '', false, false, null); |
| 780 |
if (!WP_Filesystem($credentials)) { |
| 781 |
return false; |
| 782 |
} |
| 783 |
|
| 784 |
$this->filesystem = $wp_filesystem; |
| 785 |
return true; |
| 786 |
} |
| 787 |
|
| 788 |
/** |
| 789 |
* Check if directory is writable using WP_Filesystem |
| 790 |
* |
| 791 |
* @since 1.0.0 |
| 792 |
* @param string $path Directory path to check |
| 793 |
* @return bool True if writable, false otherwise |
| 794 |
*/ |
| 795 |
private function is_directory_writable(string $path): bool { |
| 796 |
if (!$this->init_filesystem()) { |
| 797 |
return false; |
| 798 |
} |
| 799 |
|
| 800 |
return $this->filesystem->is_writable($path); |
| 801 |
} |
| 802 |
|
| 803 |
/** |
| 804 |
* Check if file is writable using WP_Filesystem |
| 805 |
* |
| 806 |
* @since 1.0.0 |
| 807 |
* @param string $file File path to check |
| 808 |
* @return bool True if writable, false otherwise |
| 809 |
*/ |
| 810 |
private function is_file_writable(string $file): bool { |
| 811 |
if (!$this->init_filesystem()) { |
| 812 |
return false; |
| 813 |
} |
| 814 |
|
| 815 |
return $this->filesystem->is_writable($file); |
| 816 |
} |
| 817 |
public function generate_title(string $template_name = 'default', array $data = [], string $context = 'site'): string { |
| 818 |
// Get template |
| 819 |
$template = $this->title_templates[$template_name] ?? $this->title_templates['default']; |
| 820 |
|
| 821 |
// Get site settings |
| 822 |
$settings = $this->get_settings('site'); |
| 823 |
$separator = $this->get_title_separator($settings['title_separator'] ?? 'pipe'); |
| 824 |
|
| 825 |
// Prepare placeholder data |
| 826 |
$placeholders = $this->prepare_title_placeholders($data, $context, $settings); |
| 827 |
|
| 828 |
// Replace placeholders |
| 829 |
$title = $this->replace_title_placeholders($template, $placeholders, $separator); |
| 830 |
|
| 831 |
// Clean and optimize title |
| 832 |
$title = $this->optimize_title($title, $context); |
| 833 |
|
| 834 |
return $title; |
| 835 |
} |
| 836 |
|
| 837 |
/** |
| 838 |
* Generate breadcrumb navigation with schema markup |
| 839 |
* |
| 840 |
* @since 1.0.0 |
| 841 |
* |
| 842 |
* @param string $type Breadcrumb type |
| 843 |
* @param array $options Breadcrumb options |
| 844 |
* @return array Breadcrumb data with schema markup |
| 845 |
*/ |
| 846 |
public function generate_breadcrumbs(string $type = 'hierarchical', array $options = []): array { |
| 847 |
$breadcrumbs = [ |
| 848 |
'items' => [], |
| 849 |
'schema' => [], |
| 850 |
'html' => '', |
| 851 |
'type' => $type, |
| 852 |
'count' => 0 |
| 853 |
]; |
| 854 |
|
| 855 |
// Get breadcrumb settings |
| 856 |
$settings = $this->get_settings('site'); |
| 857 |
$breadcrumb_settings = $settings['breadcrumbs'] ?? []; |
| 858 |
|
| 859 |
// Generate breadcrumb items based on type |
| 860 |
switch ($type) { |
| 861 |
case 'hierarchical': |
| 862 |
$breadcrumbs['items'] = $this->generate_hierarchical_breadcrumbs($options); |
| 863 |
break; |
| 864 |
case 'taxonomy': |
| 865 |
$breadcrumbs['items'] = $this->generate_taxonomy_breadcrumbs($options); |
| 866 |
break; |
| 867 |
case 'path': |
| 868 |
$breadcrumbs['items'] = $this->generate_path_breadcrumbs($options); |
| 869 |
break; |
| 870 |
case 'custom': |
| 871 |
$breadcrumbs['items'] = $this->generate_custom_breadcrumbs($options); |
| 872 |
break; |
| 873 |
} |
| 874 |
|
| 875 |
// Generate schema markup |
| 876 |
$breadcrumbs['schema'] = $this->generate_breadcrumb_schema($breadcrumbs['items']); |
| 877 |
|
| 878 |
// Generate HTML output |
| 879 |
$breadcrumbs['html'] = $this->generate_breadcrumb_html($breadcrumbs['items'], $breadcrumb_settings); |
| 880 |
|
| 881 |
// Set count |
| 882 |
$breadcrumbs['count'] = count($breadcrumbs['items']); |
| 883 |
|
| 884 |
return $breadcrumbs; |
| 885 |
} |
| 886 |
|
| 887 |
/** |
| 888 |
* Generate and manage robots.txt content |
| 889 |
* |
| 890 |
* @since 1.0.0 |
| 891 |
* |
| 892 |
* @param array $custom_rules Optional custom rules to add |
| 893 |
* @return array Robots.txt data and validation |
| 894 |
*/ |
| 895 |
public function generate_robots_txt(array $custom_rules = []): array { |
| 896 |
$robots_data = [ |
| 897 |
'content' => '', |
| 898 |
'rules' => [], |
| 899 |
'validation' => [], |
| 900 |
'file_exists' => false, |
| 901 |
'writable' => false |
| 902 |
]; |
| 903 |
|
| 904 |
// Check if robots.txt file exists and is writable |
| 905 |
$robots_file = ABSPATH . 'robots.txt'; |
| 906 |
$robots_data['file_exists'] = file_exists($robots_file); |
| 907 |
$robots_data['writable'] = $this->is_directory_writable(dirname($robots_file)); |
| 908 |
|
| 909 |
// Get site settings |
| 910 |
$settings = $this->get_settings('site'); |
| 911 |
|
| 912 |
// Generate default rules (pass full settings so sitemap_url is available) |
| 913 |
$default_rules = $this->generate_default_robots_rules($settings); |
| 914 |
|
| 915 |
// Merge with custom rules |
| 916 |
$all_rules = array_merge($default_rules, $custom_rules); |
| 917 |
|
| 918 |
// Validate rules |
| 919 |
$robots_data['validation'] = $this->validate_robots_rules($all_rules); |
| 920 |
|
| 921 |
// Generate robots.txt content |
| 922 |
$robots_data['content'] = $this->build_robots_txt_content($all_rules); |
| 923 |
$robots_data['rules'] = $all_rules; |
| 924 |
|
| 925 |
return $robots_data; |
| 926 |
} |
| 927 |
|
| 928 |
/** |
| 929 |
* Resolve the robots.txt that should actually be served. |
| 930 |
* |
| 931 |
* The Robots.txt textarea (`robots_txt_content`) is the source of truth the |
| 932 |
* admin sees and edits; per the UI, an empty value means "auto-generate". |
| 933 |
* Both the virtual `robots_txt` filter and the physical file are rendered |
| 934 |
* through here so what is served always matches what the textarea shows — |
| 935 |
* previously the served output was regenerated from rules and silently |
| 936 |
* ignored any manual edit. |
| 937 |
* |
| 938 |
* @since 1.20.0 |
| 939 |
* @return string Robots.txt body, always newline-terminated. |
| 940 |
*/ |
| 941 |
public function render_robots_txt(): string { |
| 942 |
$settings = $this->get_settings('site'); |
| 943 |
|
| 944 |
// A site-wide crawl block — "Allow Search Engines" off, or WordPress's |
| 945 |
// "Discourage search engines" (Settings → Reading, blog_public=0) — must |
| 946 |
// win over any custom robots.txt content. Otherwise a stored override |
| 947 |
// that permits crawling would silently defeat the block on every serving |
| 948 |
// and persistence path. When blocked, force the generated output, which |
| 949 |
// resolves to `User-agent: * / Disallow: /` via generate_default_robots_rules(). |
| 950 |
$allow_search = $settings['allow_search_engines'] ?? true; |
| 951 |
$fully_blocked = empty($allow_search) || !get_option('blog_public'); |
| 952 |
|
| 953 |
$custom = trim((string) ($settings['robots_txt_content'] ?? '')); |
| 954 |
// A user edit may still carry the old header if it was stored before the |
| 955 |
// header/body split — strip it so we don't emit two headers. |
| 956 |
$body = ($custom !== '' && !$fully_blocked) |
| 957 |
? $this->strip_robots_header($custom) |
| 958 |
: trim($this->generate_robots_txt()['content']); |
| 959 |
|
| 960 |
// The per-agent AI directives are machine-owned, so they are composed |
| 961 |
// here rather than stored: the textarea holds the user's body, with |
| 962 |
// the fenced block stripped out of every read and re-applied on every |
| 963 |
// render. A site-wide block already disallows everyone, so adding the |
| 964 |
// per-agent group there would be noise restating the same refusal. |
| 965 |
// Composed here rather than read from storage, for the same reason as |
| 966 |
// the AI block below: the set of sitemaps an install publishes is a |
| 967 |
// runtime fact. `robots_txt_content` is a snapshot of it taken at the |
| 968 |
// last save, and nothing invalidated that snapshot, so deactivating a |
| 969 |
// sitemap provider left its URL advertised and serving HTML (#835). |
| 970 |
// Composing it on every render means the advertisement agrees with what |
| 971 |
// the install publishes, in both directions, with no cache to expire. |
| 972 |
if (!$fully_blocked) { |
| 973 |
$body = $this->apply_sitemap_block($body); |
| 974 |
} |
| 975 |
|
| 976 |
if (!$fully_blocked) { |
| 977 |
$body = $this->apply_ai_crawler_block($body, $settings); |
| 978 |
} |
| 979 |
|
| 980 |
if ($body === '') { |
| 981 |
return ''; |
| 982 |
} |
| 983 |
|
| 984 |
return $this->robots_txt_header() . $body . "\n"; |
| 985 |
} |
| 986 |
|
| 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 |
/** |
| 1067 |
* Resolve the robots.txt actually served to crawlers, with its origin. |
| 1068 |
* |
| 1069 |
* Lets an API/MCP consumer see the effective output without crawling the |
| 1070 |
* URL. Mirrors serving precedence: a physical robots.txt in the web root is |
| 1071 |
* served verbatim by the web server; otherwise the rendered content (custom |
| 1072 |
* override or generated defaults) is served through the `robots_txt` filter. |
| 1073 |
* |
| 1074 |
* @since 1.20.0 |
| 1075 |
* @return array{content: string, is_default: bool, source: string} Effective |
| 1076 |
* robots.txt, whether it is ThinkRank's generated default (vs. a |
| 1077 |
* custom override), and where it originates from. |
| 1078 |
*/ |
| 1079 |
public function get_effective_robots_txt(): array { |
| 1080 |
// A real file in the web root wins — the web server serves it directly. |
| 1081 |
$robots_file = ABSPATH . 'robots.txt'; |
| 1082 |
if (file_exists($robots_file) && is_readable($robots_file)) { |
| 1083 |
// phpcs:ignore WordPress.WP.AlternativeFunctions.file_get_contents_file_get_contents -- Reading a public web-root file; WP_Filesystem is not available on front-end requests. |
| 1084 |
return [ |
| 1085 |
// phpcs:ignore WordPress.WP.AlternativeFunctions.file_get_contents_file_get_contents -- reads a local file the plugin just located; WP_Filesystem would need credentials on some hosts. |
| 1086 |
'content' => (string) file_get_contents($robots_file), |
| 1087 |
'is_default' => false, |
| 1088 |
'source' => 'file', |
| 1089 |
]; |
| 1090 |
} |
| 1091 |
|
| 1092 |
$settings = $this->get_settings('site'); |
| 1093 |
|
| 1094 |
// Management disabled — WordPress serves its own core default. |
| 1095 |
if (empty($settings['robots_txt_enabled'])) { |
| 1096 |
return [ |
| 1097 |
'content' => '', |
| 1098 |
'is_default' => true, |
| 1099 |
'source' => 'wordpress', |
| 1100 |
]; |
| 1101 |
} |
| 1102 |
|
| 1103 |
// A non-empty stored override replaces the generated defaults. |
| 1104 |
$custom = trim((string) ($settings['robots_txt_content'] ?? '')); |
| 1105 |
|
| 1106 |
return [ |
| 1107 |
'content' => $this->render_robots_txt(), |
| 1108 |
'is_default' => $custom === '', |
| 1109 |
'source' => $custom === '' ? 'generated' : 'custom', |
| 1110 |
]; |
| 1111 |
} |
| 1112 |
|
| 1113 |
/** |
| 1114 |
* Describe how /robots.txt is actually delivered, and whether that still |
| 1115 |
* matches the saved settings. |
| 1116 |
* |
| 1117 |
* The admin screen edits settings, but a physical robots.txt in the web root |
| 1118 |
* is served directly by the web server and bypasses the `robots_txt` filter |
| 1119 |
* entirely. When those two drift, the editor is showing content no crawler |
| 1120 |
* ever sees — the conflict this exists to surface. |
| 1121 |
* |
| 1122 |
* @since 1.31.0 |
| 1123 |
* |
| 1124 |
* @return array{content: string, source: string, is_default: bool, in_sync: bool, out_of_sync_reason: string, url: string} |
| 1125 |
* The served content and its origin, whether it still reflects the |
| 1126 |
* body the editor is showing, why it does not when it does not |
| 1127 |
* ('file_drift' or 'crawl_blocked'), and the public URL it is |
| 1128 |
* served from. |
| 1129 |
*/ |
| 1130 |
public function get_robots_txt_delivery(): array { |
| 1131 |
$effective = $this->get_effective_robots_txt(); |
| 1132 |
$settings = $this->get_settings('site'); |
| 1133 |
|
| 1134 |
// Compare bodies, not raw strings: the auto-generated header carries a |
| 1135 |
// regeneration timestamp that always differs and means nothing here. |
| 1136 |
// The AI crawler block is composed at render time on both sides, so it |
| 1137 |
// is identical by construction and comparing it would only ever report |
| 1138 |
// a false drift the admin cannot act on. |
| 1139 |
$served = $this->strip_ai_crawler_block($this->strip_robots_header($effective['content'])); |
| 1140 |
|
| 1141 |
// Measure against the body the editor is displaying — get_served_robots_body() |
| 1142 |
// — not against render_robots_txt(). Two things made the old comparison |
| 1143 |
// report "in sync" while the screen showed rules no crawler receives: |
| 1144 |
// a physical file was compared to a freshly rendered body rather than |
| 1145 |
// to the stored override the textarea shows, and a site-wide crawl |
| 1146 |
// block makes render_robots_txt() return the generated "Disallow: /" |
| 1147 |
// on both sides of the comparison, so it always matched. |
| 1148 |
$expected = $this->get_served_robots_body(); |
| 1149 |
|
| 1150 |
// Management off: WordPress serves its own default and the editor is not |
| 1151 |
// claiming anything is live, so there is nothing to be out of sync with. |
| 1152 |
$managed = !empty($settings['robots_txt_enabled']); |
| 1153 |
$in_sync = !$managed || $served === $expected; |
| 1154 |
|
| 1155 |
$reason = ''; |
| 1156 |
if (!$in_sync) { |
| 1157 |
// A crawl block is a deliberate override, not a stale file, and the |
| 1158 |
// admin needs to be told which of the two they are looking at. |
| 1159 |
$blocked = empty($settings['allow_search_engines'] ?? true) || !get_option('blog_public'); |
| 1160 |
$reason = $blocked ? 'crawl_blocked' : 'file_drift'; |
| 1161 |
} |
| 1162 |
|
| 1163 |
return [ |
| 1164 |
'content' => $effective['content'], |
| 1165 |
'source' => $effective['source'], |
| 1166 |
'is_default' => $effective['is_default'], |
| 1167 |
'in_sync' => $in_sync, |
| 1168 |
'out_of_sync_reason' => $reason, |
| 1169 |
'url' => home_url('/robots.txt'), |
| 1170 |
]; |
| 1171 |
} |
| 1172 |
|
| 1173 |
/** |
| 1174 |
* Keep the physical robots.txt file in step with the saved settings. |
| 1175 |
* |
| 1176 |
* When management is enabled the physical file is the source of truth the |
| 1177 |
* web server serves, so this makes sure it exists and matches the effective |
| 1178 |
* content — creating it if missing. When management is disabled it removes |
| 1179 |
* any existing file so WordPress serves its default again. Callers invoke |
| 1180 |
* this after saving robots settings so a plain Save both creates and |
| 1181 |
* refreshes the file without a separate "Generate" step. |
| 1182 |
* |
| 1183 |
* @since 1.20.0 |
| 1184 |
* @return bool True if the file was written or removed as intended. |
| 1185 |
*/ |
| 1186 |
public function sync_robots_txt_file(): bool { |
| 1187 |
$robots_file = ABSPATH . 'robots.txt'; |
| 1188 |
$settings = $this->get_settings('site'); |
| 1189 |
|
| 1190 |
// Management turned off: drop any existing file so WordPress serves its |
| 1191 |
// default again, rather than leaving a stale ThinkRank file behind. |
| 1192 |
if (empty($settings['robots_txt_enabled'])) { |
| 1193 |
if (file_exists($robots_file) && $this->init_filesystem()) { |
| 1194 |
$this->filesystem->delete($robots_file); |
| 1195 |
} |
| 1196 |
return true; |
| 1197 |
} |
| 1198 |
|
| 1199 |
$content = $this->render_robots_txt(); |
| 1200 |
if ($content === '') { |
| 1201 |
return false; |
| 1202 |
} |
| 1203 |
|
| 1204 |
// Run the effective content through the standard robots_txt filter so |
| 1205 |
// lines added by other integrations (ThinkRank Pro's News/Video |
| 1206 |
// Publisher Sitemaps at priority 999, and any third-party plugin) are |
| 1207 |
// baked into the physical file. A physical robots.txt bypasses core's |
| 1208 |
// do_robots()/robots_txt filter entirely, so without this those lines |
| 1209 |
// are silently dropped. ThinkRank's own filter_robots_txt callback just |
| 1210 |
// re-returns this same content (it calls render_robots_txt(), which does |
| 1211 |
// not re-apply the filter), so there is no recursion or double-append. |
| 1212 |
// phpcs:ignore WordPress.NamingConventions.PrefixAllGlobals.NonPrefixedHooknameFound -- WPML/core hook, not ours to name. |
| 1213 |
$content = (string) apply_filters('robots_txt', $content, (bool) get_option('blog_public')); |
| 1214 |
if ($content === '') { |
| 1215 |
return false; |
| 1216 |
} |
| 1217 |
|
| 1218 |
// write_robots_txt() creates the file when absent and overwrites it |
| 1219 |
// otherwise, so this covers both first-time creation and re-sync. |
| 1220 |
$result = $this->write_robots_txt($content); |
| 1221 |
return !empty($result['success']); |
| 1222 |
} |
| 1223 |
|
| 1224 |
/** |
| 1225 |
* Write robots.txt content to filesystem |
| 1226 |
* |
| 1227 |
* @since 1.0.0 |
| 1228 |
* |
| 1229 |
* @param string $content Robots.txt content to write |
| 1230 |
* @return array Write operation result |
| 1231 |
*/ |
| 1232 |
public function write_robots_txt(string $content): array { |
| 1233 |
$result = [ |
| 1234 |
'success' => false, |
| 1235 |
'message' => '', |
| 1236 |
'file_path' => '', |
| 1237 |
'permissions' => [] |
| 1238 |
]; |
| 1239 |
|
| 1240 |
$robots_file = ABSPATH . 'robots.txt'; |
| 1241 |
|
| 1242 |
// Security: Validate file path to prevent path traversal attacks |
| 1243 |
$real_robots_file = realpath(dirname($robots_file)) . DIRECTORY_SEPARATOR . basename($robots_file); |
| 1244 |
$allowed_dir = realpath(ABSPATH); |
| 1245 |
|
| 1246 |
if (!$allowed_dir || strpos(dirname($real_robots_file), $allowed_dir) !== 0) { |
| 1247 |
$result['message'] = 'Invalid file path detected for security reasons.'; |
| 1248 |
return $result; |
| 1249 |
} |
| 1250 |
|
| 1251 |
$result['file_path'] = $robots_file; |
| 1252 |
|
| 1253 |
// Check directory permissions |
| 1254 |
$result['permissions'] = [ |
| 1255 |
'directory_writable' => $this->is_directory_writable(ABSPATH), |
| 1256 |
'file_exists' => file_exists($robots_file), |
| 1257 |
'file_writable' => file_exists($robots_file) ? $this->is_file_writable($robots_file) : null |
| 1258 |
]; |
| 1259 |
|
| 1260 |
// Check if we can write to the directory |
| 1261 |
if (!$result['permissions']['directory_writable']) { |
| 1262 |
$result['message'] = 'WordPress root directory is not writable. Please check file permissions.'; |
| 1263 |
return $result; |
| 1264 |
} |
| 1265 |
|
| 1266 |
// Check if existing file is writable (if it exists) |
| 1267 |
if ($result['permissions']['file_exists'] && !$result['permissions']['file_writable']) { |
| 1268 |
$result['message'] = 'Existing robots.txt file is not writable. Please check file permissions.'; |
| 1269 |
return $result; |
| 1270 |
} |
| 1271 |
|
| 1272 |
try { |
| 1273 |
// Write new content using WP_Filesystem |
| 1274 |
if (!$this->init_filesystem()) { |
| 1275 |
$result['message'] = 'Could not initialize WordPress filesystem.'; |
| 1276 |
return $result; |
| 1277 |
} |
| 1278 |
|
| 1279 |
$write_success = $this->filesystem->put_contents($robots_file, $content, FS_CHMOD_FILE); |
| 1280 |
|
| 1281 |
if ($write_success) { |
| 1282 |
$result['success'] = true; |
| 1283 |
$result['message'] = 'Robots.txt file written successfully.'; |
| 1284 |
$result['bytes_written'] = strlen($content); |
| 1285 |
} else { |
| 1286 |
$result['message'] = 'Failed to write robots.txt file.'; |
| 1287 |
} |
| 1288 |
} catch (\Exception $e) { |
| 1289 |
$result['message'] = 'Error writing robots.txt file: ' . $e->getMessage(); |
| 1290 |
} |
| 1291 |
|
| 1292 |
return $result; |
| 1293 |
} |
| 1294 |
|
| 1295 |
/** |
| 1296 |
* Optimize site identity data with comprehensive analysis |
| 1297 |
* |
| 1298 |
* @since 1.0.0 |
| 1299 |
* |
| 1300 |
* @param array $identity_data Site identity data to optimize |
| 1301 |
* @param array $options Optimization options including section focus |
| 1302 |
* @return array Optimized identity data with validation |
| 1303 |
*/ |
| 1304 |
public function optimize_site_identity(array $identity_data, array $options = []): array { |
| 1305 |
$optimization = [ |
| 1306 |
'optimized_data' => [], |
| 1307 |
'validation' => [], |
| 1308 |
'suggestions' => [], |
| 1309 |
'warnings' => [], |
| 1310 |
'improvements' => [], |
| 1311 |
'score' => 0, |
| 1312 |
'section_scores' => [] |
| 1313 |
]; |
| 1314 |
|
| 1315 |
// Determine optimization focus |
| 1316 |
$focus = $options['focus'] ?? 'all'; |
| 1317 |
|
| 1318 |
// Section-specific optimization |
| 1319 |
if ($focus === 'title_formats' || $focus === 'all') { |
| 1320 |
$title_optimization = $this->optimize_title_formats($identity_data); |
| 1321 |
$optimization['section_scores']['title_formats'] = $title_optimization['score']; |
| 1322 |
$optimization['suggestions'] = array_merge($optimization['suggestions'], $title_optimization['suggestions']); |
| 1323 |
$optimization['warnings'] = array_merge($optimization['warnings'], $title_optimization['warnings']); |
| 1324 |
} |
| 1325 |
|
| 1326 |
if ($focus === 'breadcrumbs' || $focus === 'all') { |
| 1327 |
$breadcrumb_optimization = $this->optimize_breadcrumbs($identity_data); |
| 1328 |
$optimization['section_scores']['breadcrumbs'] = $breadcrumb_optimization['score']; |
| 1329 |
$optimization['suggestions'] = array_merge($optimization['suggestions'], $breadcrumb_optimization['suggestions']); |
| 1330 |
$optimization['warnings'] = array_merge($optimization['warnings'], $breadcrumb_optimization['warnings']); |
| 1331 |
} |
| 1332 |
|
| 1333 |
if ($focus === 'robots_txt' || $focus === 'all') { |
| 1334 |
$robots_optimization = $this->optimize_robots_txt($identity_data); |
| 1335 |
$optimization['section_scores']['robots_txt'] = $robots_optimization['score']; |
| 1336 |
$optimization['suggestions'] = array_merge($optimization['suggestions'], $robots_optimization['suggestions']); |
| 1337 |
$optimization['warnings'] = array_merge($optimization['warnings'], $robots_optimization['warnings']); |
| 1338 |
} |
| 1339 |
|
| 1340 |
if ($focus === 'site_assets' || $focus === 'all') { |
| 1341 |
$assets_optimization = $this->optimize_site_assets($identity_data); |
| 1342 |
$optimization['section_scores']['site_assets'] = $assets_optimization['score']; |
| 1343 |
$optimization['suggestions'] = array_merge($optimization['suggestions'], $assets_optimization['suggestions']); |
| 1344 |
$optimization['warnings'] = array_merge($optimization['warnings'], $assets_optimization['warnings']); |
| 1345 |
} |
| 1346 |
|
| 1347 |
// Legacy element-by-element optimization for basic site info |
| 1348 |
if ($focus === 'site_info' || $focus === 'all') { |
| 1349 |
foreach ($this->identity_elements as $element => $config) { |
| 1350 |
if (isset($identity_data[$element])) { |
| 1351 |
$element_optimization = $this->optimize_identity_element( |
| 1352 |
$element, |
| 1353 |
$identity_data[$element], |
| 1354 |
$config |
| 1355 |
); |
| 1356 |
|
| 1357 |
$optimization['optimized_data'][$element] = $element_optimization['optimized_value']; |
| 1358 |
$optimization['validation'][$element] = $element_optimization['validation']; |
| 1359 |
$optimization['suggestions'] = array_merge( |
| 1360 |
$optimization['suggestions'], |
| 1361 |
$element_optimization['suggestions'] |
| 1362 |
); |
| 1363 |
} |
| 1364 |
} |
| 1365 |
} |
| 1366 |
|
| 1367 |
// Calculate overall optimization score |
| 1368 |
if (!empty($optimization['section_scores'])) { |
| 1369 |
$optimization['score'] = (int) round(array_sum($optimization['section_scores']) / count($optimization['section_scores'])); |
| 1370 |
} else { |
| 1371 |
$optimization['score'] = $this->calculate_identity_score($optimization['validation']); |
| 1372 |
} |
| 1373 |
|
| 1374 |
// Store optimization results in seo_analysis table |
| 1375 |
$this->store_optimization_results($optimization, $focus); |
| 1376 |
|
| 1377 |
return $optimization; |
| 1378 |
} |
| 1379 |
|
| 1380 |
/** |
| 1381 |
* Optimize title formats with enhanced rules |
| 1382 |
* |
| 1383 |
* @since 1.0.0 |
| 1384 |
* |
| 1385 |
* @param array $settings Title format settings |
| 1386 |
* @return array Optimization results |
| 1387 |
*/ |
| 1388 |
public function optimize_title_formats(array $settings): array { |
| 1389 |
$optimization = [ |
| 1390 |
'score' => 100, |
| 1391 |
'suggestions' => [], |
| 1392 |
'warnings' => [], |
| 1393 |
'improvements' => [] |
| 1394 |
]; |
| 1395 |
|
| 1396 |
// Check separator choice (applies to every context template). |
| 1397 |
$separator = $settings['title_separator'] ?? 'pipe'; |
| 1398 |
$separator_data = self::$title_separators[$separator] ?? null; |
| 1399 |
if ($separator_data) { |
| 1400 |
$seo_score = $separator_data['seo_score'] ?? 5; |
| 1401 |
if ($seo_score < 8) { |
| 1402 |
$optimization['suggestions'][] = "Consider using '|' or '-' separators for better SEO performance"; |
| 1403 |
$optimization['score'] -= (10 - $seo_score); |
| 1404 |
} |
| 1405 |
} |
| 1406 |
|
| 1407 |
// Analyze the per-context templates the Title Formats UI actually edits |
| 1408 |
// and the front end actually renders — not the legacy `title_template` |
| 1409 |
// enum, which this screen never sets. |
| 1410 |
$context_labels = [ |
| 1411 |
'homepage_title' => 'Homepage', |
| 1412 |
'post_title' => 'Post', |
| 1413 |
'page_title' => 'Page', |
| 1414 |
'category_title' => 'Category', |
| 1415 |
'tag_title' => 'Tag', |
| 1416 |
'author_title' => 'Author', |
| 1417 |
'search_title' => 'Search', |
| 1418 |
'archive_title' => 'Archive', |
| 1419 |
]; |
| 1420 |
|
| 1421 |
$configured = 0; |
| 1422 |
foreach ($context_labels as $key => $label) { |
| 1423 |
$template = isset($settings[$key]) ? trim((string) $settings[$key]) : ''; |
| 1424 |
if ($template === '') { |
| 1425 |
continue; // Unconfigured — the front end falls back for this context. |
| 1426 |
} |
| 1427 |
$configured++; |
| 1428 |
|
| 1429 |
// Brand recognition: the title should carry the site name. |
| 1430 |
if (strpos($template, '%site_title%') === false && strpos($template, '%site_name%') === false) { |
| 1431 |
$optimization['suggestions'][] = "{$label} title has no site name — add %site_title% for brand recognition"; |
| 1432 |
$optimization['score'] -= 5; |
| 1433 |
} |
| 1434 |
|
| 1435 |
// Length check against the ~60-char guideline, measured on the |
| 1436 |
// resolved title for THIS context (with representative sample data). |
| 1437 |
$sample_length = strlen($this->generate_sample_title($settings, $key)); |
| 1438 |
if ($sample_length > 60) { |
| 1439 |
$optimization['warnings'][] = "{$label} title renders about {$sample_length} characters (over the 60-character limit)"; |
| 1440 |
$optimization['score'] -= 10; |
| 1441 |
} elseif ($sample_length > 0 && $sample_length < 20) { |
| 1442 |
$optimization['suggestions'][] = "{$label} title renders only about {$sample_length} characters — consider adding more context"; |
| 1443 |
$optimization['score'] -= 3; |
| 1444 |
} |
| 1445 |
} |
| 1446 |
|
| 1447 |
// No context templates set at all — ThinkRank won't control any titles. |
| 1448 |
if ($configured === 0) { |
| 1449 |
$optimization['suggestions'][] = 'No title formats are configured — set templates so ThinkRank controls your page titles'; |
| 1450 |
$optimization['score'] -= 10; |
| 1451 |
} |
| 1452 |
|
| 1453 |
$optimization['score'] = max(0, min(100, $optimization['score'])); |
| 1454 |
|
| 1455 |
return $optimization; |
| 1456 |
} |
| 1457 |
|
| 1458 |
/** |
| 1459 |
* Optimize breadcrumb settings with UX best practices |
| 1460 |
* |
| 1461 |
* @since 1.0.0 |
| 1462 |
* |
| 1463 |
* @param array $settings Breadcrumb settings |
| 1464 |
* @return array Optimization results |
| 1465 |
*/ |
| 1466 |
public function optimize_breadcrumbs(array $settings): array { |
| 1467 |
$optimization = [ |
| 1468 |
'score' => 100, |
| 1469 |
'suggestions' => [], |
| 1470 |
'warnings' => [], |
| 1471 |
'improvements' => [] |
| 1472 |
]; |
| 1473 |
|
| 1474 |
// Check if breadcrumbs are enabled |
| 1475 |
if (!($settings['breadcrumbs_enabled'] ?? true)) { |
| 1476 |
$optimization['suggestions'][] = 'Enable breadcrumbs to improve user navigation and SEO (recommended by Google)'; |
| 1477 |
$optimization['score'] = 20; // Major penalty for disabled breadcrumbs |
| 1478 |
return $optimization; |
| 1479 |
} |
| 1480 |
|
| 1481 |
// Validate breadcrumb type |
| 1482 |
$type = $settings['breadcrumb_type'] ?? 'hierarchical'; |
| 1483 |
$type_scores = [ |
| 1484 |
'hierarchical' => 100, |
| 1485 |
'category_based' => 90, |
| 1486 |
'simple' => 70, |
| 1487 |
'custom' => 80 |
| 1488 |
]; |
| 1489 |
|
| 1490 |
$type_score = $type_scores[$type] ?? 60; |
| 1491 |
$optimization['score'] = min($optimization['score'], $type_score); |
| 1492 |
|
| 1493 |
if ($type === 'simple') { |
| 1494 |
$optimization['suggestions'][] = 'Consider hierarchical breadcrumbs for better site structure representation'; |
| 1495 |
} |
| 1496 |
|
| 1497 |
// Validate separator choice |
| 1498 |
$separator = $settings['breadcrumb_separator'] ?? '›'; |
| 1499 |
$separator_ux = [ |
| 1500 |
'›' => ['score' => 100, 'note' => 'Clear directional indicator'], |
| 1501 |
'>' => ['score' => 95, 'note' => 'Simple and effective'], |
| 1502 |
'/' => ['score' => 85, 'note' => 'Familiar but can confuse with URLs'], |
| 1503 |
'|' => ['score' => 75, 'note' => 'Less intuitive for navigation'], |
| 1504 |
'»' => ['score' => 90, 'note' => 'Distinctive double arrow'] |
| 1505 |
]; |
| 1506 |
|
| 1507 |
$sep_data = $separator_ux[$separator] ?? ['score' => 50, 'note' => 'Unusual choice']; |
| 1508 |
$optimization['score'] = min($optimization['score'], $sep_data['score']); |
| 1509 |
|
| 1510 |
if ($sep_data['score'] < 95) { |
| 1511 |
$optimization['suggestions'][] = "Separator '{$separator}': {$sep_data['note']}"; |
| 1512 |
} |
| 1513 |
|
| 1514 |
// Validate home text |
| 1515 |
$home_text = $settings['breadcrumb_home_text'] ?? 'Home'; |
| 1516 |
if (empty($home_text)) { |
| 1517 |
$optimization['warnings'][] = 'Empty home text reduces accessibility for screen readers'; |
| 1518 |
$optimization['score'] -= 15; |
| 1519 |
} elseif (mb_strlen($home_text) > 20) { |
| 1520 |
// mb_strlen: this number is shown to the user as "chars" (#687). |
| 1521 |
$optimization['suggestions'][] = 'Keep home text concise (current: ' . mb_strlen($home_text) . ' chars)'; |
| 1522 |
$optimization['score'] -= 5; |
| 1523 |
} |
| 1524 |
|
| 1525 |
// Check prefix usage |
| 1526 |
$prefix = $settings['breadcrumb_prefix'] ?? ''; |
| 1527 |
if (!empty($prefix) && mb_strlen($prefix) > 50) { |
| 1528 |
// mb_strlen: this number is shown to the user as "chars" (#687). |
| 1529 |
$optimization['suggestions'][] = 'Breadcrumb prefix is quite long (' . mb_strlen($prefix) . ' chars) - consider shortening'; |
| 1530 |
$optimization['score'] -= 5; |
| 1531 |
} |
| 1532 |
|
| 1533 |
// Current page display |
| 1534 |
if (!($settings['show_current_page'] ?? true)) { |
| 1535 |
$optimization['suggestions'][] = 'Show current page in breadcrumbs for better user orientation'; |
| 1536 |
$optimization['score'] -= 10; |
| 1537 |
} |
| 1538 |
|
| 1539 |
return $optimization; |
| 1540 |
} |
| 1541 |
|
| 1542 |
/** |
| 1543 |
* Optimize robots.txt settings with technical SEO best practices |
| 1544 |
* |
| 1545 |
* @since 1.0.0 |
| 1546 |
* |
| 1547 |
* @param array $settings Robots.txt settings |
| 1548 |
* @return array Optimization results |
| 1549 |
*/ |
| 1550 |
public function optimize_robots_txt(array $settings): array { |
| 1551 |
$optimization = [ |
| 1552 |
'score' => 100, |
| 1553 |
'suggestions' => [], |
| 1554 |
'warnings' => [], |
| 1555 |
'improvements' => [] |
| 1556 |
]; |
| 1557 |
|
| 1558 |
// Check if robots.txt management is enabled |
| 1559 |
if (!($settings['robots_txt_enabled'] ?? true)) { |
| 1560 |
$optimization['suggestions'][] = 'Enable robots.txt management for better SEO control and automated updates'; |
| 1561 |
$optimization['score'] = 30; |
| 1562 |
return $optimization; |
| 1563 |
} |
| 1564 |
|
| 1565 |
// Critical: Search engine access |
| 1566 |
if (!($settings['allow_search_engines'] ?? true)) { |
| 1567 |
$optimization['warnings'][] = 'CRITICAL: Search engines are blocked - your site will not be indexed by Google, Bing, etc.'; |
| 1568 |
$optimization['score'] = 10; // Severe penalty |
| 1569 |
} |
| 1570 |
|
| 1571 |
// Sitemap URL validation (now from sitemap settings) |
| 1572 |
$sitemap_urls = $this->get_sitemap_urls_for_robots(); |
| 1573 |
if (empty($sitemap_urls)) { |
| 1574 |
$optimization['suggestions'][] = 'Enable sitemap generation to include sitemap URLs in robots.txt'; |
| 1575 |
$optimization['score'] -= 15; |
| 1576 |
} else { |
| 1577 |
// Validate first sitemap accessibility (representative check) |
| 1578 |
$first_sitemap = $sitemap_urls[0]; |
| 1579 |
$sitemap_response = wp_remote_head($first_sitemap, ['timeout' => 10]); |
| 1580 |
if (is_wp_error($sitemap_response) || wp_remote_retrieve_response_code($sitemap_response) !== 200) { |
| 1581 |
$optimization['warnings'][] = 'Primary sitemap URL is not accessible - check sitemap generation'; |
| 1582 |
$optimization['score'] -= 10; |
| 1583 |
} |
| 1584 |
} |
| 1585 |
|
| 1586 |
// File system permissions |
| 1587 |
$robots_file = ABSPATH . 'robots.txt'; |
| 1588 |
$robots_dir = dirname($robots_file); |
| 1589 |
|
| 1590 |
if (!$this->is_directory_writable($robots_dir)) { |
| 1591 |
$optimization['warnings'][] = 'WordPress root directory is not writable - robots.txt cannot be managed automatically'; |
| 1592 |
$optimization['score'] -= 15; |
| 1593 |
} elseif (file_exists($robots_file) && !$this->is_file_writable($robots_file)) { |
| 1594 |
$optimization['warnings'][] = 'Existing robots.txt file is not writable - cannot update automatically'; |
| 1595 |
$optimization['score'] -= 10; |
| 1596 |
} |
| 1597 |
|
| 1598 |
// Content analysis |
| 1599 |
$custom_content = $settings['robots_txt_content'] ?? ''; |
| 1600 |
if (!empty($custom_content)) { |
| 1601 |
// Check for dangerous patterns |
| 1602 |
if (preg_match('/User-agent:\s*\*\s*\n\s*Disallow:\s*\/\s*$/m', $custom_content)) { |
| 1603 |
$optimization['warnings'][] = 'Blocking all content for all crawlers - this will prevent search engine indexing'; |
| 1604 |
$optimization['score'] -= 30; |
| 1605 |
} |
| 1606 |
|
| 1607 |
// Check for sitemap declaration in content |
| 1608 |
if (!empty($sitemap_urls) && strpos($custom_content, 'Sitemap:') === false) { |
| 1609 |
$optimization['suggestions'][] = 'Sitemap URLs are automatically included in generated robots.txt'; |
| 1610 |
$optimization['score'] -= 5; |
| 1611 |
} |
| 1612 |
} |
| 1613 |
|
| 1614 |
return $optimization; |
| 1615 |
} |
| 1616 |
|
| 1617 |
/** |
| 1618 |
* Optimize site assets (logo, favicon, apple touch icon) |
| 1619 |
* |
| 1620 |
* @since 1.0.0 |
| 1621 |
* |
| 1622 |
* @param array $settings Site assets settings |
| 1623 |
* @return array Optimization results |
| 1624 |
*/ |
| 1625 |
public function optimize_site_assets(array $settings): array { |
| 1626 |
$optimization = [ |
| 1627 |
'score' => 100, |
| 1628 |
'suggestions' => [], |
| 1629 |
'warnings' => [], |
| 1630 |
'improvements' => [] |
| 1631 |
]; |
| 1632 |
|
| 1633 |
// Check site logo |
| 1634 |
$logo_url = $settings['logo_url'] ?? ''; |
| 1635 |
if (empty($logo_url)) { |
| 1636 |
$optimization['suggestions'][] = 'Add a site logo for better branding and professional appearance'; |
| 1637 |
$optimization['score'] -= 20; |
| 1638 |
} else { |
| 1639 |
// Validate logo URL and dimensions |
| 1640 |
if (!filter_var($logo_url, FILTER_VALIDATE_URL)) { |
| 1641 |
$optimization['warnings'][] = 'Logo URL format is invalid'; |
| 1642 |
$optimization['score'] -= 15; |
| 1643 |
} |
| 1644 |
} |
| 1645 |
|
| 1646 |
// Check favicon |
| 1647 |
$favicon_url = $settings['favicon_url'] ?? ''; |
| 1648 |
if (empty($favicon_url)) { |
| 1649 |
$optimization['suggestions'][] = 'Add a favicon for better browser tab identification'; |
| 1650 |
$optimization['score'] -= 15; |
| 1651 |
} |
| 1652 |
|
| 1653 |
// Check Apple touch icon |
| 1654 |
$apple_icon_url = $settings['apple_touch_icon_url'] ?? ''; |
| 1655 |
if (empty($apple_icon_url)) { |
| 1656 |
$optimization['suggestions'][] = 'Add an Apple touch icon for better iOS device experience'; |
| 1657 |
$optimization['score'] -= 10; |
| 1658 |
} |
| 1659 |
|
| 1660 |
// Additional logo analysis for local images |
| 1661 |
if (!empty($logo_url) && filter_var($logo_url, FILTER_VALIDATE_URL)) { |
| 1662 |
$attachment_id = Attachment_Lookup::id_from_url($logo_url); |
| 1663 |
if ($attachment_id) { |
| 1664 |
$image_meta = wp_get_attachment_metadata($attachment_id); |
| 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 = $logo_file['height']; |
| 1670 |
|
| 1671 |
// SVG logos store 0x0 metadata — no dimension/ratio analysis |
| 1672 |
// is possible (and dividing by 0 is fatal). |
| 1673 |
if ($image_meta && $width > 0 && $height > 0) { |
| 1674 |
if ($width < 112 || $height < 112) { |
| 1675 |
$optimization['warnings'][] = "Logo dimensions ({$width}x{$height}) are below recommended minimum (112x112)"; |
| 1676 |
$optimization['score'] -= 10; |
| 1677 |
} |
| 1678 |
|
| 1679 |
if ($width > 1920 || $height > 1920) { |
| 1680 |
$optimization['suggestions'][] = "Logo dimensions ({$width}x{$height}) are very large - consider optimizing for faster loading"; |
| 1681 |
$optimization['score'] -= 5; |
| 1682 |
} |
| 1683 |
|
| 1684 |
// Aspect ratio check |
| 1685 |
$ratio = $width / $height; |
| 1686 |
if ($ratio < 0.5 || $ratio > 2.0) { |
| 1687 |
$optimization['suggestions'][] = 'Logo aspect ratio should be between 1:2 and 2:1 for optimal display'; |
| 1688 |
$optimization['score'] -= 5; |
| 1689 |
} |
| 1690 |
} |
| 1691 |
} |
| 1692 |
} |
| 1693 |
|
| 1694 |
return $optimization; |
| 1695 |
} |
| 1696 |
|
| 1697 |
/** |
| 1698 |
* Optimize local SEO settings for better local search visibility |
| 1699 |
* |
| 1700 |
* @since 1.0.0 |
| 1701 |
* |
| 1702 |
* @param array $settings Local SEO settings |
| 1703 |
* @return array Optimization results |
| 1704 |
*/ |
| 1705 |
public function optimize_local_seo(array $settings): array { |
| 1706 |
$optimization = [ |
| 1707 |
'score' => 100, |
| 1708 |
'suggestions' => [], |
| 1709 |
'warnings' => [], |
| 1710 |
'improvements' => [], |
| 1711 |
'optimized_data' => [] |
| 1712 |
]; |
| 1713 |
|
| 1714 |
// Check if local SEO is enabled |
| 1715 |
if (empty($settings['local_seo_enabled'])) { |
| 1716 |
$optimization['warnings'][] = 'Local SEO is disabled - enable it to improve local search visibility'; |
| 1717 |
$optimization['score'] -= 20; |
| 1718 |
return $optimization; |
| 1719 |
} |
| 1720 |
|
| 1721 |
// Validate business name (required for local SEO) |
| 1722 |
if (empty($settings['business_name'])) { |
| 1723 |
$optimization['warnings'][] = 'Business name is required for local SEO'; |
| 1724 |
$optimization['score'] -= 25; |
| 1725 |
} else { |
| 1726 |
// Optimize business name |
| 1727 |
$optimized_name = $this->optimize_business_name($settings['business_name']); |
| 1728 |
if ($optimized_name !== $settings['business_name']) { |
| 1729 |
$optimization['optimized_data']['business_name'] = $optimized_name; |
| 1730 |
$optimization['suggestions'][] = 'Business name optimized for better local search visibility'; |
| 1731 |
} |
| 1732 |
} |
| 1733 |
|
| 1734 |
// Validate complete address (NAP consistency) |
| 1735 |
$address_score = $this->validate_business_address($settings, $optimization); |
| 1736 |
$optimization['score'] -= (100 - $address_score); |
| 1737 |
|
| 1738 |
// Validate phone number |
| 1739 |
if (empty($settings['business_phone'])) { |
| 1740 |
$optimization['warnings'][] = 'Business phone number is missing - important for local SEO and NAP consistency'; |
| 1741 |
$optimization['score'] -= 15; |
| 1742 |
} else { |
| 1743 |
$optimized_phone = $this->optimize_phone_number($settings['business_phone']); |
| 1744 |
if ($optimized_phone !== $settings['business_phone']) { |
| 1745 |
$optimization['optimized_data']['business_phone'] = $optimized_phone; |
| 1746 |
$optimization['suggestions'][] = 'Phone number formatted for better consistency'; |
| 1747 |
} |
| 1748 |
} |
| 1749 |
|
| 1750 |
// Validate business hours |
| 1751 |
if (empty($settings['business_hours']) || !is_array($settings['business_hours'])) { |
| 1752 |
$optimization['suggestions'][] = 'Add business hours to improve local search visibility and customer experience'; |
| 1753 |
$optimization['score'] -= 10; |
| 1754 |
} else { |
| 1755 |
$hours_validation = $this->validate_business_hours($settings['business_hours']); |
| 1756 |
if (!$hours_validation['valid']) { |
| 1757 |
$optimization['warnings'] = array_merge($optimization['warnings'], $hours_validation['warnings']); |
| 1758 |
$optimization['score'] -= $hours_validation['penalty']; |
| 1759 |
} |
| 1760 |
} |
| 1761 |
|
| 1762 |
// Check for geo-coordinates |
| 1763 |
if (empty($settings['business_latitude']) || empty($settings['business_longitude'])) { |
| 1764 |
$optimization['suggestions'][] = 'Add latitude and longitude coordinates for precise location targeting'; |
| 1765 |
$optimization['score'] -= 10; |
| 1766 |
} else { |
| 1767 |
// Validate coordinates |
| 1768 |
if (!$this->validate_coordinates($settings['business_latitude'], $settings['business_longitude'])) { |
| 1769 |
$optimization['warnings'][] = 'Invalid latitude or longitude coordinates'; |
| 1770 |
$optimization['score'] -= 15; |
| 1771 |
} |
| 1772 |
} |
| 1773 |
|
| 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']; |
| 1778 |
$optimization['score'] -= 5; |
| 1779 |
} |
| 1780 |
|
| 1781 |
// Email validation |
| 1782 |
if (!empty($settings['business_email']) && !is_email($settings['business_email'])) { |
| 1783 |
$optimization['warnings'][] = 'Business email format is invalid'; |
| 1784 |
$optimization['score'] -= 10; |
| 1785 |
} |
| 1786 |
|
| 1787 |
// Local SEO best practices |
| 1788 |
$this->add_local_seo_best_practices($optimization, $settings); |
| 1789 |
|
| 1790 |
return $optimization; |
| 1791 |
} |
| 1792 |
|
| 1793 |
/** |
| 1794 |
* Optimize business name for local SEO |
| 1795 |
* |
| 1796 |
* @param string $business_name Original business name |
| 1797 |
* @return string Optimized business name |
| 1798 |
*/ |
| 1799 |
private function optimize_business_name(string $business_name): string { |
| 1800 |
// Remove excessive punctuation and normalize spacing |
| 1801 |
$optimized = preg_replace('/[^\w\s\-&.,]/', '', $business_name); |
| 1802 |
$optimized = preg_replace('/\s+/', ' ', $optimized); |
| 1803 |
$optimized = trim($optimized); |
| 1804 |
|
| 1805 |
// Ensure proper capitalization |
| 1806 |
$optimized = ucwords(strtolower($optimized)); |
| 1807 |
|
| 1808 |
return $optimized; |
| 1809 |
} |
| 1810 |
|
| 1811 |
/** |
| 1812 |
* Validate business address components |
| 1813 |
* |
| 1814 |
* @param array $settings Business settings |
| 1815 |
* @param array &$optimization Optimization results (passed by reference) |
| 1816 |
* @return int Address completeness score (0-100) |
| 1817 |
*/ |
| 1818 |
private function validate_business_address(array $settings, array &$optimization): int { |
| 1819 |
$score = 100; |
| 1820 |
$required_fields = ['business_address', 'business_city', 'business_state', 'business_country']; |
| 1821 |
$missing_fields = []; |
| 1822 |
|
| 1823 |
foreach ($required_fields as $field) { |
| 1824 |
if (empty($settings[$field])) { |
| 1825 |
$missing_fields[] = str_replace('business_', '', $field); |
| 1826 |
$score -= 20; |
| 1827 |
} |
| 1828 |
} |
| 1829 |
|
| 1830 |
if (!empty($missing_fields)) { |
| 1831 |
$optimization['warnings'][] = 'Missing address components: ' . implode(', ', $missing_fields) . ' - important for NAP consistency'; |
| 1832 |
} |
| 1833 |
|
| 1834 |
// Postal code is recommended but not required |
| 1835 |
if (empty($settings['business_postal_code'])) { |
| 1836 |
$optimization['suggestions'][] = 'Add postal code for more precise location targeting'; |
| 1837 |
$score -= 5; |
| 1838 |
} |
| 1839 |
|
| 1840 |
return max(0, $score); |
| 1841 |
} |
| 1842 |
|
| 1843 |
/** |
| 1844 |
* Optimize phone number format for consistency |
| 1845 |
* |
| 1846 |
* @param string $phone_number Original phone number |
| 1847 |
* @return string Optimized phone number |
| 1848 |
*/ |
| 1849 |
private function optimize_phone_number(string $phone_number): string { |
| 1850 |
// Remove all non-numeric characters except + for international numbers |
| 1851 |
$cleaned = preg_replace('/[^\d+]/', '', $phone_number); |
| 1852 |
|
| 1853 |
// If it's a US number (10 digits), format as (XXX) XXX-XXXX |
| 1854 |
if (preg_match('/^(\d{10})$/', $cleaned, $matches)) { |
| 1855 |
return '(' . substr($matches[1], 0, 3) . ') ' . substr($matches[1], 3, 3) . '-' . substr($matches[1], 6); |
| 1856 |
} |
| 1857 |
|
| 1858 |
// If it's a US number with country code, format as +1 (XXX) XXX-XXXX |
| 1859 |
if (preg_match('/^1(\d{10})$/', $cleaned, $matches)) { |
| 1860 |
return '+1 (' . substr($matches[1], 0, 3) . ') ' . substr($matches[1], 3, 3) . '-' . substr($matches[1], 6); |
| 1861 |
} |
| 1862 |
|
| 1863 |
// For international numbers, keep the + and return as-is |
| 1864 |
return $cleaned; |
| 1865 |
} |
| 1866 |
|
| 1867 |
/** |
| 1868 |
* Validate business hours format and completeness |
| 1869 |
* |
| 1870 |
* @param array $business_hours Business hours array |
| 1871 |
* @return array Validation results |
| 1872 |
*/ |
| 1873 |
private function validate_business_hours(array $business_hours): array { |
| 1874 |
$validation = [ |
| 1875 |
'valid' => true, |
| 1876 |
'warnings' => [], |
| 1877 |
'penalty' => 0 |
| 1878 |
]; |
| 1879 |
|
| 1880 |
$days = ['monday', 'tuesday', 'wednesday', 'thursday', 'friday', 'saturday', 'sunday']; |
| 1881 |
$open_days = 0; |
| 1882 |
|
| 1883 |
foreach ($days as $day) { |
| 1884 |
if (!isset($business_hours[$day])) { |
| 1885 |
continue; |
| 1886 |
} |
| 1887 |
|
| 1888 |
$day_data = $business_hours[$day]; |
| 1889 |
|
| 1890 |
if (empty($day_data['closed'])) { |
| 1891 |
$open_days++; |
| 1892 |
|
| 1893 |
// Validate time format |
| 1894 |
if (empty($day_data['open']) || empty($day_data['close'])) { |
| 1895 |
$validation['warnings'][] = "Missing opening or closing time for {$day}"; |
| 1896 |
$validation['penalty'] += 2; |
| 1897 |
} else { |
| 1898 |
// Validate time format (HH:MM) |
| 1899 |
if (!preg_match('/^\d{2}:\d{2}$/', $day_data['open']) || !preg_match('/^\d{2}:\d{2}$/', $day_data['close'])) { |
| 1900 |
$validation['warnings'][] = "Invalid time format for {$day} (use HH:MM format)"; |
| 1901 |
$validation['penalty'] += 2; |
| 1902 |
} |
| 1903 |
} |
| 1904 |
} |
| 1905 |
} |
| 1906 |
|
| 1907 |
if ($open_days === 0) { |
| 1908 |
$validation['warnings'][] = 'No business hours specified - all days marked as closed'; |
| 1909 |
$validation['penalty'] += 10; |
| 1910 |
} |
| 1911 |
|
| 1912 |
if ($validation['penalty'] > 0) { |
| 1913 |
$validation['valid'] = false; |
| 1914 |
} |
| 1915 |
|
| 1916 |
return $validation; |
| 1917 |
} |
| 1918 |
|
| 1919 |
/** |
| 1920 |
* Validate latitude and longitude coordinates |
| 1921 |
* |
| 1922 |
* @param string $latitude Latitude coordinate |
| 1923 |
* @param string $longitude Longitude coordinate |
| 1924 |
* @return bool True if coordinates are valid |
| 1925 |
*/ |
| 1926 |
private function validate_coordinates(string $latitude, string $longitude): bool { |
| 1927 |
$lat = floatval($latitude); |
| 1928 |
$lng = floatval($longitude); |
| 1929 |
|
| 1930 |
// Validate latitude range (-90 to 90) |
| 1931 |
if ($lat < -90 || $lat > 90) { |
| 1932 |
return false; |
| 1933 |
} |
| 1934 |
|
| 1935 |
// Validate longitude range (-180 to 180) |
| 1936 |
if ($lng < -180 || $lng > 180) { |
| 1937 |
return false; |
| 1938 |
} |
| 1939 |
|
| 1940 |
return true; |
| 1941 |
} |
| 1942 |
|
| 1943 |
/** |
| 1944 |
* Add local SEO best practices suggestions |
| 1945 |
* |
| 1946 |
* @param array &$optimization Optimization results (passed by reference) |
| 1947 |
* @param array $settings Business settings |
| 1948 |
* @return void |
| 1949 |
*/ |
| 1950 |
private function add_local_seo_best_practices(array &$optimization, array $settings): void { |
| 1951 |
// Check for Google My Business integration |
| 1952 |
if (empty($settings['google_my_business_url'])) { |
| 1953 |
$optimization['suggestions'][] = 'Consider adding your Google My Business profile URL for better local visibility'; |
| 1954 |
} |
| 1955 |
|
| 1956 |
// Check for social media profiles |
| 1957 |
$social_platforms = ['facebook_url', 'twitter_url', 'instagram_url', 'linkedin_url']; |
| 1958 |
$has_social = false; |
| 1959 |
foreach ($social_platforms as $platform) { |
| 1960 |
if (!empty($settings[$platform])) { |
| 1961 |
$has_social = true; |
| 1962 |
break; |
| 1963 |
} |
| 1964 |
} |
| 1965 |
|
| 1966 |
if (!$has_social) { |
| 1967 |
$optimization['suggestions'][] = 'Add social media profiles to improve local business credibility'; |
| 1968 |
} |
| 1969 |
|
| 1970 |
// Check for business description |
| 1971 |
if (empty($settings['business_description'])) { |
| 1972 |
$optimization['suggestions'][] = 'Add a business description for better context in local search results'; |
| 1973 |
} |
| 1974 |
|
| 1975 |
// Service area suggestions |
| 1976 |
if (empty($settings['service_areas'])) { |
| 1977 |
$optimization['suggestions'][] = 'Define service areas if your business serves multiple locations'; |
| 1978 |
} |
| 1979 |
} |
| 1980 |
|
| 1981 |
/** |
| 1982 |
* Generate a sample rendered title for a given context template, so the |
| 1983 |
* optimizer can measure the length users will actually see. |
| 1984 |
* |
| 1985 |
* Resolves the per-context template (e.g. `post_title`) with representative |
| 1986 |
* sample values for the same variable tokens the front-end renderer fills |
| 1987 |
* in (see SEO_Manager::get_title_placeholders()). |
| 1988 |
* |
| 1989 |
* @since 1.0.0 |
| 1990 |
* |
| 1991 |
* @param array $settings Title format settings |
| 1992 |
* @param string $context_key Per-context template key (e.g. 'post_title') |
| 1993 |
* @return string Resolved sample title (empty string when the template is unset) |
| 1994 |
*/ |
| 1995 |
private function generate_sample_title(array $settings, string $context_key = 'post_title'): string { |
| 1996 |
$template = isset($settings[$context_key]) ? trim((string) $settings[$context_key]) : ''; |
| 1997 |
if ($template === '') { |
| 1998 |
return ''; |
| 1999 |
} |
| 2000 |
|
| 2001 |
$separator = $settings['title_separator'] ?? 'pipe'; |
| 2002 |
$separator_symbol = self::$title_separators[$separator]['symbol'] ?? '|'; |
| 2003 |
|
| 2004 |
$site_name = $settings['site_name'] ?? ''; |
| 2005 |
if ($site_name === '') { |
| 2006 |
$site_name = get_bloginfo('name') ?: 'Your Site Name'; |
| 2007 |
} |
| 2008 |
$site_description = $settings['site_description'] ?? ''; |
| 2009 |
if ($site_description === '') { |
| 2010 |
$site_description = get_bloginfo('description') ?: 'Your Site Description'; |
| 2011 |
} |
| 2012 |
$tagline = $settings['tagline'] ?? ''; |
| 2013 |
if ($tagline === '') { |
| 2014 |
$tagline = $site_description; |
| 2015 |
} |
| 2016 |
|
| 2017 |
// Representative sample values for the variable tokens the front end |
| 2018 |
// substitutes per request. Keys mirror get_title_placeholders(). |
| 2019 |
$sample_data = [ |
| 2020 |
'%site_title%' => $site_name, |
| 2021 |
'%site_name%' => $site_name, |
| 2022 |
'%site_description%' => $site_description, |
| 2023 |
'%tagline%' => $tagline, |
| 2024 |
'%sep%' => ' ' . $separator_symbol . ' ', |
| 2025 |
'%separator%' => ' ' . $separator_symbol . ' ', |
| 2026 |
'%post_title%' => 'How to Optimize Your Website for Better SEO Results', |
| 2027 |
'%page_title%' => 'About Our Company', |
| 2028 |
'%category_title%' => 'SEO Tips', |
| 2029 |
'%category%' => 'SEO Tips', |
| 2030 |
'%tag_title%' => 'On-Page SEO', |
| 2031 |
'%tag%' => 'On-Page SEO', |
| 2032 |
'%author_name%' => 'Jane Doe', |
| 2033 |
'%author%' => 'Jane Doe', |
| 2034 |
'%search_term%' => 'keyword research', |
| 2035 |
'%search_phrase%' => 'keyword research', |
| 2036 |
'%archive_title%' => 'July 2026', |
| 2037 |
'%date%' => gmdate('F Y'), |
| 2038 |
]; |
| 2039 |
|
| 2040 |
$title = str_replace(array_keys($sample_data), array_values($sample_data), $template); |
| 2041 |
|
| 2042 |
// Collapse whitespace left by any empty/unresolved tokens, then trim. |
| 2043 |
$title = preg_replace('/\s+/', ' ', $title); |
| 2044 |
|
| 2045 |
return trim($title); |
| 2046 |
} |
| 2047 |
|
| 2048 |
/** |
| 2049 |
* Store optimization results in seo_analysis table |
| 2050 |
* |
| 2051 |
* @since 1.0.0 |
| 2052 |
* |
| 2053 |
* @param array $optimization Optimization results |
| 2054 |
* @param string $focus Optimization focus section |
| 2055 |
* @return void |
| 2056 |
*/ |
| 2057 |
private function store_optimization_results(array $optimization, string $focus): void { |
| 2058 |
global $wpdb; |
| 2059 |
|
| 2060 |
$table_name = $wpdb->prefix . 'thinkrank_seo_analysis'; |
| 2061 |
|
| 2062 |
// Only store if we have meaningful results |
| 2063 |
if (empty($optimization['suggestions']) && empty($optimization['warnings'])) { |
| 2064 |
return; |
| 2065 |
} |
| 2066 |
|
| 2067 |
$analysis_type = 'site_identity_rule_optimization'; |
| 2068 |
if ($focus !== 'all') { |
| 2069 |
$analysis_type .= '_' . $focus; |
| 2070 |
} |
| 2071 |
|
| 2072 |
// phpcs:ignore WordPress.DB.DirectDatabaseQuery.DirectQuery, PluginCheck.Security.DirectDB.UnescapedDBParameter -- Site identity analysis storage requires direct database access |
| 2073 |
$wpdb->insert( |
| 2074 |
$table_name, |
| 2075 |
[ |
| 2076 |
'context_type' => 'site', |
| 2077 |
'context_id' => null, |
| 2078 |
'analysis_type' => $analysis_type, |
| 2079 |
'analysis_data' => wp_json_encode($optimization), |
| 2080 |
'score' => $optimization['score'], |
| 2081 |
'status' => 'completed', |
| 2082 |
'recommendations' => wp_json_encode($optimization['suggestions']), |
| 2083 |
'validation_errors' => wp_json_encode($optimization['warnings']), |
| 2084 |
'analyzed_by' => get_current_user_id() |
| 2085 |
], |
| 2086 |
['%s', '%d', '%s', '%s', '%d', '%s', '%s', '%s', '%d'] |
| 2087 |
); |
| 2088 |
} |
| 2089 |
|
| 2090 |
/** |
| 2091 |
* Validate SEO settings (implements interface) |
| 2092 |
* |
| 2093 |
* @since 1.0.0 |
| 2094 |
* |
| 2095 |
* @param array $settings Settings array to validate |
| 2096 |
* @param string $tab_context Optional tab context for specific validation |
| 2097 |
* @return array Validation results |
| 2098 |
*/ |
| 2099 |
public function validate_settings(array $settings, string $tab_context = ''): array { |
| 2100 |
// If tab context is provided, use tab-specific validation |
| 2101 |
if (!empty($tab_context)) { |
| 2102 |
return $this->get_tab_specific_validation($settings, $tab_context); |
| 2103 |
} |
| 2104 |
|
| 2105 |
// Default comprehensive validation for backward compatibility |
| 2106 |
$validation = [ |
| 2107 |
'valid' => true, |
| 2108 |
'errors' => [], |
| 2109 |
'warnings' => [], |
| 2110 |
'suggestions' => [], |
| 2111 |
'score' => 100 |
| 2112 |
]; |
| 2113 |
|
| 2114 |
// Validate title template |
| 2115 |
if (isset($settings['title_template'])) { |
| 2116 |
if (!isset($this->title_templates[$settings['title_template']])) { |
| 2117 |
$validation['errors'][] = 'Invalid title template specified'; |
| 2118 |
$validation['valid'] = false; |
| 2119 |
} |
| 2120 |
} |
| 2121 |
|
| 2122 |
// Validate title separator |
| 2123 |
if (isset($settings['title_separator'])) { |
| 2124 |
if (!isset(self::$title_separators[$settings['title_separator']])) { |
| 2125 |
$validation['errors'][] = __('Invalid title separator specified.', 'thinkrank'); |
| 2126 |
$validation['valid'] = false; |
| 2127 |
} |
| 2128 |
} |
| 2129 |
|
| 2130 |
// Validate site name |
| 2131 |
if (isset($settings['site_name'])) { |
| 2132 |
if (empty($settings['site_name'])) { |
| 2133 |
$validation['errors'][] = 'Site name is required'; |
| 2134 |
$validation['valid'] = false; |
| 2135 |
} elseif (strlen($settings['site_name']) > 60) { |
| 2136 |
$validation['warnings'][] = 'Site name is longer than 60 characters, may be truncated'; |
| 2137 |
} |
| 2138 |
} |
| 2139 |
|
| 2140 |
// Validate site description |
| 2141 |
if (isset($settings['site_description']) && !empty($settings['site_description'])) { |
| 2142 |
if (strlen($settings['site_description']) > 160) { |
| 2143 |
$validation['warnings'][] = 'Site description is longer than 160 characters, may be truncated'; |
| 2144 |
} elseif (strlen($settings['site_description']) < 120) { |
| 2145 |
$validation['suggestions'][] = 'Consider making site description longer (120-160 characters)'; |
| 2146 |
} |
| 2147 |
} |
| 2148 |
|
| 2149 |
// Validate logo URL |
| 2150 |
if (isset($settings['logo_url']) && !empty($settings['logo_url'])) { |
| 2151 |
if (!filter_var($settings['logo_url'], FILTER_VALIDATE_URL)) { |
| 2152 |
$validation['errors'][] = 'Logo URL must be a valid URL'; |
| 2153 |
$validation['valid'] = false; |
| 2154 |
} |
| 2155 |
} |
| 2156 |
|
| 2157 |
// Validate breadcrumb settings |
| 2158 |
if (isset($settings['breadcrumb_type'])) { |
| 2159 |
if (!isset($this->breadcrumb_types[$settings['breadcrumb_type']])) { |
| 2160 |
$validation['errors'][] = 'Invalid breadcrumb type specified'; |
| 2161 |
$validation['valid'] = false; |
| 2162 |
} |
| 2163 |
} |
| 2164 |
|
| 2165 |
// Validate robots.txt settings |
| 2166 |
if (isset($settings['robots_txt_enabled']) && $settings['robots_txt_enabled']) { |
| 2167 |
if (!$this->is_directory_writable(ABSPATH)) { |
| 2168 |
$validation['warnings'][] = 'WordPress root directory is not writable, robots.txt cannot be automatically managed'; |
| 2169 |
} |
| 2170 |
} |
| 2171 |
|
| 2172 |
// Validate local SEO settings if enabled |
| 2173 |
if (isset($settings['local_seo_enabled']) && $settings['local_seo_enabled']) { |
| 2174 |
$local_seo_validation = $this->validate_local_seo_settings($settings); |
| 2175 |
$validation['errors'] = array_merge($validation['errors'], $local_seo_validation['errors']); |
| 2176 |
$validation['warnings'] = array_merge($validation['warnings'], $local_seo_validation['warnings']); |
| 2177 |
$validation['suggestions'] = array_merge($validation['suggestions'], $local_seo_validation['suggestions']); |
| 2178 |
|
| 2179 |
if (!$local_seo_validation['valid']) { |
| 2180 |
$validation['valid'] = false; |
| 2181 |
} |
| 2182 |
} |
| 2183 |
|
| 2184 |
// Calculate validation score |
| 2185 |
$validation['score'] = $this->calculate_validation_score($validation); |
| 2186 |
|
| 2187 |
// Add detailed field validation breakdown for generic validation |
| 2188 |
$validation['field_details'] = $this->get_detailed_field_validation($settings, ''); |
| 2189 |
|
| 2190 |
return $validation; |
| 2191 |
} |
| 2192 |
|
| 2193 |
/** |
| 2194 |
* Get tab-specific validation |
| 2195 |
* |
| 2196 |
* @since 1.0.0 |
| 2197 |
* |
| 2198 |
* @param array $settings Settings array to validate |
| 2199 |
* @param string $tab_context Tab context for specific validation |
| 2200 |
* @return array Tab-specific validation results |
| 2201 |
*/ |
| 2202 |
private function get_tab_specific_validation(array $settings, string $tab_context): array { |
| 2203 |
$validation = [ |
| 2204 |
'valid' => true, |
| 2205 |
'errors' => [], |
| 2206 |
'warnings' => [], |
| 2207 |
'suggestions' => [], |
| 2208 |
'score' => 100 |
| 2209 |
]; |
| 2210 |
|
| 2211 |
// Get tab-specific field details |
| 2212 |
$field_details = $this->get_detailed_field_validation($settings, $tab_context); |
| 2213 |
|
| 2214 |
// Convert field details to validation format |
| 2215 |
foreach ($field_details as $field) { |
| 2216 |
switch ($field['status']) { |
| 2217 |
case 'error': |
| 2218 |
$validation['errors'][] = $field['label']; |
| 2219 |
$validation['valid'] = false; |
| 2220 |
$validation['score'] -= 20; |
| 2221 |
break; |
| 2222 |
case 'warning': |
| 2223 |
$validation['warnings'][] = $field['label']; |
| 2224 |
$validation['score'] -= 10; |
| 2225 |
break; |
| 2226 |
case 'suggestion': |
| 2227 |
$validation['suggestions'][] = $field['label']; |
| 2228 |
$validation['score'] -= 5; |
| 2229 |
break; |
| 2230 |
} |
| 2231 |
} |
| 2232 |
|
| 2233 |
// Ensure score doesn't go below 0 |
| 2234 |
$validation['score'] = max(0, $validation['score']); |
| 2235 |
|
| 2236 |
// Add field details for frontend display |
| 2237 |
$validation['field_details'] = $field_details; |
| 2238 |
|
| 2239 |
return $validation; |
| 2240 |
} |
| 2241 |
|
| 2242 |
/** |
| 2243 |
* Get detailed field validation breakdown |
| 2244 |
* |
| 2245 |
* @since 1.0.0 |
| 2246 |
* |
| 2247 |
* @param array $settings Settings array to validate |
| 2248 |
* @param string $tab_context Tab context for specific validation |
| 2249 |
* @return array Detailed field validation results |
| 2250 |
*/ |
| 2251 |
private function get_detailed_field_validation(array $settings, string $tab_context = ''): array { |
| 2252 |
$field_details = []; |
| 2253 |
|
| 2254 |
// Return tab-specific validation based on context |
| 2255 |
switch ($tab_context) { |
| 2256 |
case 'local-seo': |
| 2257 |
return $this->get_business_info_validation($settings); |
| 2258 |
case 'hero-section': |
| 2259 |
return $this->get_hero_section_validation($settings); |
| 2260 |
case 'title-formats': |
| 2261 |
return $this->get_title_formats_validation($settings); |
| 2262 |
case 'breadcrumbs': |
| 2263 |
return $this->get_breadcrumbs_validation($settings); |
| 2264 |
default: |
| 2265 |
// Default basic info validation |
| 2266 |
return $this->get_basic_info_validation($settings); |
| 2267 |
} |
| 2268 |
} |
| 2269 |
|
| 2270 |
/** |
| 2271 |
* Get Business Info specific validation |
| 2272 |
* |
| 2273 |
* @since 1.0.0 |
| 2274 |
* |
| 2275 |
* @param array $settings Settings array to validate |
| 2276 |
* @return array Business Info validation results |
| 2277 |
*/ |
| 2278 |
private function get_business_info_validation(array $settings): array { |
| 2279 |
$field_details = []; |
| 2280 |
|
| 2281 |
// Check if Local SEO is enabled |
| 2282 |
if (empty($settings['local_seo_enabled'])) { |
| 2283 |
$field_details[] = [ |
| 2284 |
'field' => 'local_seo_enabled', |
| 2285 |
'label' => 'Local SEO is disabled. Enable to configure business information.', |
| 2286 |
'status' => 'warning', |
| 2287 |
'icon' => '⚠' |
| 2288 |
]; |
| 2289 |
return $field_details; |
| 2290 |
} |
| 2291 |
|
| 2292 |
// Business Name validation |
| 2293 |
if (!empty($settings['business_name'])) { |
| 2294 |
$field_details[] = [ |
| 2295 |
'field' => 'business_name', |
| 2296 |
'label' => 'Business name is properly configured.', |
| 2297 |
'status' => 'valid', |
| 2298 |
'icon' => '✓' |
| 2299 |
]; |
| 2300 |
} else { |
| 2301 |
$field_details[] = [ |
| 2302 |
'field' => 'business_name', |
| 2303 |
'label' => 'Business name is required for local SEO.', |
| 2304 |
'status' => 'error', |
| 2305 |
'icon' => '✗' |
| 2306 |
]; |
| 2307 |
} |
| 2308 |
|
| 2309 |
// Business Type validation — see business_type_status() for why there |
| 2310 |
// is exactly one rule here now (#622). |
| 2311 |
$business_type = $this->business_type_status($settings); |
| 2312 |
$field_details[] = [ |
| 2313 |
'field' => 'business_type', |
| 2314 |
'label' => $business_type['message'], |
| 2315 |
'status' => $business_type['status'], |
| 2316 |
'icon' => 'valid' === $business_type['status'] ? '✓' : '⚠', |
| 2317 |
]; |
| 2318 |
|
| 2319 |
// Address validation (NAP consistency) |
| 2320 |
$address_fields = ['business_address', 'business_city', 'business_state', 'business_country']; |
| 2321 |
$address_complete = true; |
| 2322 |
foreach ($address_fields as $field) { |
| 2323 |
if (empty($settings[$field])) { |
| 2324 |
$address_complete = false; |
| 2325 |
break; |
| 2326 |
} |
| 2327 |
} |
| 2328 |
|
| 2329 |
if ($address_complete) { |
| 2330 |
$field_details[] = [ |
| 2331 |
'field' => 'business_address', |
| 2332 |
'label' => 'Complete business address is configured for NAP consistency.', |
| 2333 |
'status' => 'valid', |
| 2334 |
'icon' => '✓' |
| 2335 |
]; |
| 2336 |
} else { |
| 2337 |
$field_details[] = [ |
| 2338 |
'field' => 'business_address', |
| 2339 |
'label' => 'Complete address (street, city, state, country) required for local SEO.', |
| 2340 |
'status' => 'error', |
| 2341 |
'icon' => '✗' |
| 2342 |
]; |
| 2343 |
} |
| 2344 |
|
| 2345 |
// Phone validation |
| 2346 |
if (!empty($settings['business_phone'])) { |
| 2347 |
if ($this->validate_phone_format($settings['business_phone'])) { |
| 2348 |
$field_details[] = [ |
| 2349 |
'field' => 'business_phone', |
| 2350 |
'label' => 'Business phone number is properly formatted.', |
| 2351 |
'status' => 'valid', |
| 2352 |
'icon' => '✓' |
| 2353 |
]; |
| 2354 |
} else { |
| 2355 |
$field_details[] = [ |
| 2356 |
'field' => 'business_phone', |
| 2357 |
'label' => 'Business phone number format could be improved.', |
| 2358 |
'status' => 'warning', |
| 2359 |
'icon' => '⚠' |
| 2360 |
]; |
| 2361 |
} |
| 2362 |
} else { |
| 2363 |
$field_details[] = [ |
| 2364 |
'field' => 'business_phone', |
| 2365 |
'label' => 'Business phone number is important for local SEO and customer contact.', |
| 2366 |
'status' => 'warning', |
| 2367 |
'icon' => '⚠' |
| 2368 |
]; |
| 2369 |
} |
| 2370 |
|
| 2371 |
// Email validation |
| 2372 |
if (!empty($settings['business_email'])) { |
| 2373 |
if (is_email($settings['business_email'])) { |
| 2374 |
$field_details[] = [ |
| 2375 |
'field' => 'business_email', |
| 2376 |
'label' => 'Business email address is valid.', |
| 2377 |
'status' => 'valid', |
| 2378 |
'icon' => '✓' |
| 2379 |
]; |
| 2380 |
} else { |
| 2381 |
$field_details[] = [ |
| 2382 |
'field' => 'business_email', |
| 2383 |
'label' => 'Business email address format is invalid.', |
| 2384 |
'status' => 'error', |
| 2385 |
'icon' => '✗' |
| 2386 |
]; |
| 2387 |
} |
| 2388 |
} else { |
| 2389 |
$field_details[] = [ |
| 2390 |
'field' => 'business_email', |
| 2391 |
'label' => 'Business email address recommended for contact information.', |
| 2392 |
'status' => 'suggestion', |
| 2393 |
'icon' => '⚠' |
| 2394 |
]; |
| 2395 |
} |
| 2396 |
|
| 2397 |
// Coordinates validation |
| 2398 |
if (!empty($settings['business_latitude']) && !empty($settings['business_longitude'])) { |
| 2399 |
if ($this->validate_coordinates($settings['business_latitude'], $settings['business_longitude'])) { |
| 2400 |
$field_details[] = [ |
| 2401 |
'field' => 'business_coordinates', |
| 2402 |
'label' => 'Business coordinates are properly configured for precise location.', |
| 2403 |
'status' => 'valid', |
| 2404 |
'icon' => '✓' |
| 2405 |
]; |
| 2406 |
} else { |
| 2407 |
$field_details[] = [ |
| 2408 |
'field' => 'business_coordinates', |
| 2409 |
'label' => 'Business coordinates appear to be invalid.', |
| 2410 |
'status' => 'error', |
| 2411 |
'icon' => '✗' |
| 2412 |
]; |
| 2413 |
} |
| 2414 |
} else { |
| 2415 |
$field_details[] = [ |
| 2416 |
'field' => 'business_coordinates', |
| 2417 |
'label' => 'Business coordinates recommended for precise location targeting.', |
| 2418 |
'status' => 'suggestion', |
| 2419 |
'icon' => '⚠' |
| 2420 |
]; |
| 2421 |
} |
| 2422 |
|
| 2423 |
return $field_details; |
| 2424 |
} |
| 2425 |
|
| 2426 |
/** |
| 2427 |
* Get Basic Info validation (default) |
| 2428 |
* |
| 2429 |
* @since 1.0.0 |
| 2430 |
* |
| 2431 |
* @param array $settings Settings array to validate |
| 2432 |
* @return array Basic Info validation results |
| 2433 |
*/ |
| 2434 |
private function get_basic_info_validation(array $settings): array { |
| 2435 |
$field_details = []; |
| 2436 |
|
| 2437 |
// Site Name validation |
| 2438 |
if (!empty($settings['site_name'])) { |
| 2439 |
$field_details[] = [ |
| 2440 |
'field' => 'site_name', |
| 2441 |
'label' => 'Site name is properly configured.', |
| 2442 |
'status' => 'valid', |
| 2443 |
'icon' => '✓' |
| 2444 |
]; |
| 2445 |
} else { |
| 2446 |
$field_details[] = [ |
| 2447 |
'field' => 'site_name', |
| 2448 |
'label' => 'Site name is required.', |
| 2449 |
'status' => 'error', |
| 2450 |
'icon' => '✗' |
| 2451 |
]; |
| 2452 |
} |
| 2453 |
|
| 2454 |
// Site Description validation |
| 2455 |
if (!empty($settings['site_description'])) { |
| 2456 |
$length = strlen($settings['site_description']); |
| 2457 |
if ($length >= 120 && $length <= 160) { |
| 2458 |
$field_details[] = [ |
| 2459 |
'field' => 'site_description', |
| 2460 |
'label' => 'Site description is properly configured.', |
| 2461 |
'status' => 'valid', |
| 2462 |
'icon' => '✓' |
| 2463 |
]; |
| 2464 |
} else { |
| 2465 |
$field_details[] = [ |
| 2466 |
'field' => 'site_description', |
| 2467 |
'label' => 'Site description length could be optimized (120-160 characters recommended).', |
| 2468 |
'status' => 'warning', |
| 2469 |
'icon' => '⚠' |
| 2470 |
]; |
| 2471 |
} |
| 2472 |
} else { |
| 2473 |
$field_details[] = [ |
| 2474 |
'field' => 'site_description', |
| 2475 |
'label' => 'Site description is recommended for better SEO.', |
| 2476 |
'status' => 'warning', |
| 2477 |
'icon' => '⚠' |
| 2478 |
]; |
| 2479 |
} |
| 2480 |
|
| 2481 |
// Tagline validation |
| 2482 |
if (!empty($settings['tagline'])) { |
| 2483 |
$field_details[] = [ |
| 2484 |
'field' => 'tagline', |
| 2485 |
'label' => 'Site tagline is configured.', |
| 2486 |
'status' => 'valid', |
| 2487 |
'icon' => '✓' |
| 2488 |
]; |
| 2489 |
} else { |
| 2490 |
$field_details[] = [ |
| 2491 |
'field' => 'tagline', |
| 2492 |
'label' => 'Site tagline recommended for better branding.', |
| 2493 |
'status' => 'suggestion', |
| 2494 |
'icon' => '⚠' |
| 2495 |
]; |
| 2496 |
} |
| 2497 |
|
| 2498 |
// Default Meta Description validation |
| 2499 |
if (!empty($settings['default_meta_description'])) { |
| 2500 |
$length = strlen($settings['default_meta_description']); |
| 2501 |
if ($length >= 120 && $length <= 160) { |
| 2502 |
$field_details[] = [ |
| 2503 |
'field' => 'default_meta_description', |
| 2504 |
'label' => 'Default meta description is properly configured.', |
| 2505 |
'status' => 'valid', |
| 2506 |
'icon' => '✓' |
| 2507 |
]; |
| 2508 |
} else { |
| 2509 |
$field_details[] = [ |
| 2510 |
'field' => 'default_meta_description', |
| 2511 |
'label' => 'Default meta description length could be optimized (120-160 characters recommended).', |
| 2512 |
'status' => 'warning', |
| 2513 |
'icon' => '⚠' |
| 2514 |
]; |
| 2515 |
} |
| 2516 |
} else { |
| 2517 |
$field_details[] = [ |
| 2518 |
'field' => 'default_meta_description', |
| 2519 |
'label' => 'Default meta description recommended for pages without specific descriptions.', |
| 2520 |
'status' => 'suggestion', |
| 2521 |
'icon' => '⚠' |
| 2522 |
]; |
| 2523 |
} |
| 2524 |
|
| 2525 |
return $field_details; |
| 2526 |
} |
| 2527 |
|
| 2528 |
/** |
| 2529 |
* Get Hero Section validation |
| 2530 |
* |
| 2531 |
* @since 1.0.0 |
| 2532 |
* |
| 2533 |
* @param array $settings Settings array to validate |
| 2534 |
* @return array Hero Section validation results |
| 2535 |
*/ |
| 2536 |
private function get_hero_section_validation(array $settings): array { |
| 2537 |
$field_details = []; |
| 2538 |
|
| 2539 |
// Hero Title validation |
| 2540 |
if (!empty($settings['hero_title'])) { |
| 2541 |
$field_details[] = [ |
| 2542 |
'field' => 'hero_title', |
| 2543 |
'label' => 'Hero title is configured.', |
| 2544 |
'status' => 'valid', |
| 2545 |
'icon' => '✓' |
| 2546 |
]; |
| 2547 |
} else { |
| 2548 |
$field_details[] = [ |
| 2549 |
'field' => 'hero_title', |
| 2550 |
'label' => 'Hero title recommended for better homepage presentation.', |
| 2551 |
'status' => 'suggestion', |
| 2552 |
'icon' => '⚠' |
| 2553 |
]; |
| 2554 |
} |
| 2555 |
|
| 2556 |
// Hero Subtitle validation (correct field name) |
| 2557 |
if (!empty($settings['hero_subtitle'])) { |
| 2558 |
$field_details[] = [ |
| 2559 |
'field' => 'hero_subtitle', |
| 2560 |
'label' => 'Hero subtitle is configured.', |
| 2561 |
'status' => 'valid', |
| 2562 |
'icon' => '✓' |
| 2563 |
]; |
| 2564 |
} else { |
| 2565 |
$field_details[] = [ |
| 2566 |
'field' => 'hero_subtitle', |
| 2567 |
'label' => 'Hero subtitle recommended for better user engagement.', |
| 2568 |
'status' => 'suggestion', |
| 2569 |
'icon' => '⚠' |
| 2570 |
]; |
| 2571 |
} |
| 2572 |
|
| 2573 |
// CTA Text validation |
| 2574 |
if (!empty($settings['hero_cta_text'])) { |
| 2575 |
$field_details[] = [ |
| 2576 |
'field' => 'hero_cta_text', |
| 2577 |
'label' => 'Call-to-action text is configured.', |
| 2578 |
'status' => 'valid', |
| 2579 |
'icon' => '✓' |
| 2580 |
]; |
| 2581 |
} else { |
| 2582 |
$field_details[] = [ |
| 2583 |
'field' => 'hero_cta_text', |
| 2584 |
'label' => 'Call-to-action text recommended for better conversion.', |
| 2585 |
'status' => 'suggestion', |
| 2586 |
'icon' => '⚠' |
| 2587 |
]; |
| 2588 |
} |
| 2589 |
|
| 2590 |
// CTA URL validation |
| 2591 |
if (!empty($settings['hero_cta_url'])) { |
| 2592 |
if (filter_var($settings['hero_cta_url'], FILTER_VALIDATE_URL) || strpos($settings['hero_cta_url'], '/') === 0) { |
| 2593 |
$field_details[] = [ |
| 2594 |
'field' => 'hero_cta_url', |
| 2595 |
'label' => 'Call-to-action URL is properly configured.', |
| 2596 |
'status' => 'valid', |
| 2597 |
'icon' => '✓' |
| 2598 |
]; |
| 2599 |
} else { |
| 2600 |
$field_details[] = [ |
| 2601 |
'field' => 'hero_cta_url', |
| 2602 |
'label' => 'Call-to-action URL format appears invalid.', |
| 2603 |
'status' => 'warning', |
| 2604 |
'icon' => '⚠' |
| 2605 |
]; |
| 2606 |
} |
| 2607 |
} else { |
| 2608 |
$field_details[] = [ |
| 2609 |
'field' => 'hero_cta_url', |
| 2610 |
'label' => 'Call-to-action URL recommended for better conversion.', |
| 2611 |
'status' => 'suggestion', |
| 2612 |
'icon' => '⚠' |
| 2613 |
]; |
| 2614 |
} |
| 2615 |
|
| 2616 |
// Hero Background Image validation |
| 2617 |
if (!empty($settings['hero_background_image'])) { |
| 2618 |
$field_details[] = [ |
| 2619 |
'field' => 'hero_background_image', |
| 2620 |
'label' => 'Hero background image is configured.', |
| 2621 |
'status' => 'valid', |
| 2622 |
'icon' => '✓' |
| 2623 |
]; |
| 2624 |
} else { |
| 2625 |
$field_details[] = [ |
| 2626 |
'field' => 'hero_background_image', |
| 2627 |
'label' => 'Hero background image recommended for visual appeal.', |
| 2628 |
'status' => 'suggestion', |
| 2629 |
'icon' => '⚠' |
| 2630 |
]; |
| 2631 |
} |
| 2632 |
|
| 2633 |
// Site Logo validation (from Site Assets section) |
| 2634 |
if (!empty($settings['logo_url'])) { |
| 2635 |
if (filter_var($settings['logo_url'], FILTER_VALIDATE_URL)) { |
| 2636 |
$field_details[] = [ |
| 2637 |
'field' => 'logo_url', |
| 2638 |
'label' => 'Site logo is properly configured.', |
| 2639 |
'status' => 'valid', |
| 2640 |
'icon' => '✓' |
| 2641 |
]; |
| 2642 |
} else { |
| 2643 |
$field_details[] = [ |
| 2644 |
'field' => 'logo_url', |
| 2645 |
'label' => 'Site logo URL format appears invalid.', |
| 2646 |
'status' => 'warning', |
| 2647 |
'icon' => '⚠' |
| 2648 |
]; |
| 2649 |
} |
| 2650 |
} else { |
| 2651 |
$field_details[] = [ |
| 2652 |
'field' => 'logo_url', |
| 2653 |
'label' => 'Site logo recommended for branding and schema markup.', |
| 2654 |
'status' => 'suggestion', |
| 2655 |
'icon' => '⚠' |
| 2656 |
]; |
| 2657 |
} |
| 2658 |
|
| 2659 |
// Favicon validation |
| 2660 |
if (!empty($settings['favicon_url'])) { |
| 2661 |
$field_details[] = [ |
| 2662 |
'field' => 'favicon_url', |
| 2663 |
'label' => 'Favicon is configured.', |
| 2664 |
'status' => 'valid', |
| 2665 |
'icon' => '✓' |
| 2666 |
]; |
| 2667 |
} else { |
| 2668 |
$field_details[] = [ |
| 2669 |
'field' => 'favicon_url', |
| 2670 |
'label' => 'Favicon recommended for browser tab identification.', |
| 2671 |
'status' => 'suggestion', |
| 2672 |
'icon' => '⚠' |
| 2673 |
]; |
| 2674 |
} |
| 2675 |
|
| 2676 |
// Apple Touch Icon validation |
| 2677 |
if (!empty($settings['apple_touch_icon_url'])) { |
| 2678 |
$field_details[] = [ |
| 2679 |
'field' => 'apple_touch_icon_url', |
| 2680 |
'label' => 'Apple touch icon is configured.', |
| 2681 |
'status' => 'valid', |
| 2682 |
'icon' => '✓' |
| 2683 |
]; |
| 2684 |
} else { |
| 2685 |
$field_details[] = [ |
| 2686 |
'field' => 'apple_touch_icon_url', |
| 2687 |
'label' => 'Apple touch icon recommended for iOS devices.', |
| 2688 |
'status' => 'suggestion', |
| 2689 |
'icon' => '⚠' |
| 2690 |
]; |
| 2691 |
} |
| 2692 |
|
| 2693 |
return $field_details; |
| 2694 |
} |
| 2695 |
|
| 2696 |
/** |
| 2697 |
* Get Title Formats validation |
| 2698 |
* |
| 2699 |
* @since 1.0.0 |
| 2700 |
* |
| 2701 |
* @param array $settings Settings array to validate |
| 2702 |
* @return array Title Formats validation results |
| 2703 |
*/ |
| 2704 |
private function get_title_formats_validation(array $settings): array { |
| 2705 |
$field_details = []; |
| 2706 |
|
| 2707 |
// Title Separator validation |
| 2708 |
if (!empty($settings['title_separator'])) { |
| 2709 |
$field_details[] = [ |
| 2710 |
'field' => 'title_separator', |
| 2711 |
'label' => 'Title separator is properly configured.', |
| 2712 |
'status' => 'valid', |
| 2713 |
'icon' => '✓' |
| 2714 |
]; |
| 2715 |
} else { |
| 2716 |
$field_details[] = [ |
| 2717 |
'field' => 'title_separator', |
| 2718 |
'label' => 'Title separator is required.', |
| 2719 |
'status' => 'error', |
| 2720 |
'icon' => '✗' |
| 2721 |
]; |
| 2722 |
} |
| 2723 |
|
| 2724 |
// Homepage Title validation |
| 2725 |
if (!empty($settings['homepage_title'])) { |
| 2726 |
$field_details[] = [ |
| 2727 |
'field' => 'homepage_title', |
| 2728 |
'label' => 'Homepage title format is configured.', |
| 2729 |
'status' => 'valid', |
| 2730 |
'icon' => '✓' |
| 2731 |
]; |
| 2732 |
} else { |
| 2733 |
$field_details[] = [ |
| 2734 |
'field' => 'homepage_title', |
| 2735 |
'label' => 'Homepage title format recommended.', |
| 2736 |
'status' => 'suggestion', |
| 2737 |
'icon' => '⚠' |
| 2738 |
]; |
| 2739 |
} |
| 2740 |
|
| 2741 |
// Post Title validation |
| 2742 |
if (!empty($settings['post_title'])) { |
| 2743 |
$field_details[] = [ |
| 2744 |
'field' => 'post_title', |
| 2745 |
'label' => 'Post title format is configured.', |
| 2746 |
'status' => 'valid', |
| 2747 |
'icon' => '✓' |
| 2748 |
]; |
| 2749 |
} else { |
| 2750 |
$field_details[] = [ |
| 2751 |
'field' => 'post_title', |
| 2752 |
'label' => 'Post title format recommended.', |
| 2753 |
'status' => 'suggestion', |
| 2754 |
'icon' => '⚠' |
| 2755 |
]; |
| 2756 |
} |
| 2757 |
|
| 2758 |
// Page Title validation |
| 2759 |
if (!empty($settings['page_title'])) { |
| 2760 |
$field_details[] = [ |
| 2761 |
'field' => 'page_title', |
| 2762 |
'label' => 'Page title format is configured.', |
| 2763 |
'status' => 'valid', |
| 2764 |
'icon' => '✓' |
| 2765 |
]; |
| 2766 |
} else { |
| 2767 |
$field_details[] = [ |
| 2768 |
'field' => 'page_title', |
| 2769 |
'label' => 'Page title format recommended.', |
| 2770 |
'status' => 'suggestion', |
| 2771 |
'icon' => '⚠' |
| 2772 |
]; |
| 2773 |
} |
| 2774 |
|
| 2775 |
// Category Title validation |
| 2776 |
if (!empty($settings['category_title'])) { |
| 2777 |
$field_details[] = [ |
| 2778 |
'field' => 'category_title', |
| 2779 |
'label' => 'Category title format is configured.', |
| 2780 |
'status' => 'valid', |
| 2781 |
'icon' => '✓' |
| 2782 |
]; |
| 2783 |
} else { |
| 2784 |
$field_details[] = [ |
| 2785 |
'field' => 'category_title', |
| 2786 |
'label' => 'Category title format recommended.', |
| 2787 |
'status' => 'suggestion', |
| 2788 |
'icon' => '⚠' |
| 2789 |
]; |
| 2790 |
} |
| 2791 |
|
| 2792 |
// Search Title validation |
| 2793 |
if (!empty($settings['search_title'])) { |
| 2794 |
$field_details[] = [ |
| 2795 |
'field' => 'search_title', |
| 2796 |
'label' => 'Search title format is configured.', |
| 2797 |
'status' => 'valid', |
| 2798 |
'icon' => '✓' |
| 2799 |
]; |
| 2800 |
} else { |
| 2801 |
$field_details[] = [ |
| 2802 |
'field' => 'search_title', |
| 2803 |
'label' => 'Search title format recommended.', |
| 2804 |
'status' => 'suggestion', |
| 2805 |
'icon' => '⚠' |
| 2806 |
]; |
| 2807 |
} |
| 2808 |
|
| 2809 |
return $field_details; |
| 2810 |
} |
| 2811 |
|
| 2812 |
/** |
| 2813 |
* Get Breadcrumbs validation |
| 2814 |
* |
| 2815 |
* @since 1.0.0 |
| 2816 |
* |
| 2817 |
* @param array $settings Settings array to validate |
| 2818 |
* @return array Breadcrumbs validation results |
| 2819 |
*/ |
| 2820 |
private function get_breadcrumbs_validation(array $settings): array { |
| 2821 |
$field_details = []; |
| 2822 |
|
| 2823 |
// Breadcrumbs enabled validation |
| 2824 |
if (!empty($settings['breadcrumbs_enabled'])) { |
| 2825 |
$field_details[] = [ |
| 2826 |
'field' => 'breadcrumbs_enabled', |
| 2827 |
'label' => 'Breadcrumbs are enabled for better navigation.', |
| 2828 |
'status' => 'valid', |
| 2829 |
'icon' => '✓' |
| 2830 |
]; |
| 2831 |
|
| 2832 |
// Only validate other fields if breadcrumbs are enabled |
| 2833 |
// Breadcrumb Type validation |
| 2834 |
if (!empty($settings['breadcrumb_type'])) { |
| 2835 |
$field_details[] = [ |
| 2836 |
'field' => 'breadcrumb_type', |
| 2837 |
'label' => 'Breadcrumb type is properly configured.', |
| 2838 |
'status' => 'valid', |
| 2839 |
'icon' => '✓' |
| 2840 |
]; |
| 2841 |
} else { |
| 2842 |
$field_details[] = [ |
| 2843 |
'field' => 'breadcrumb_type', |
| 2844 |
'label' => 'Breadcrumb type selection is required.', |
| 2845 |
'status' => 'error', |
| 2846 |
'icon' => '✗' |
| 2847 |
]; |
| 2848 |
} |
| 2849 |
|
| 2850 |
// Home Text validation |
| 2851 |
if (!empty($settings['breadcrumb_home_text'])) { |
| 2852 |
$field_details[] = [ |
| 2853 |
'field' => 'breadcrumb_home_text', |
| 2854 |
'label' => 'Home breadcrumb text is configured.', |
| 2855 |
'status' => 'valid', |
| 2856 |
'icon' => '✓' |
| 2857 |
]; |
| 2858 |
} else { |
| 2859 |
$field_details[] = [ |
| 2860 |
'field' => 'breadcrumb_home_text', |
| 2861 |
'label' => 'Home breadcrumb text recommended for clarity.', |
| 2862 |
'status' => 'suggestion', |
| 2863 |
'icon' => '⚠' |
| 2864 |
]; |
| 2865 |
} |
| 2866 |
|
| 2867 |
// Breadcrumb Separator validation |
| 2868 |
if (!empty($settings['breadcrumb_separator'])) { |
| 2869 |
$field_details[] = [ |
| 2870 |
'field' => 'breadcrumb_separator', |
| 2871 |
'label' => 'Breadcrumb separator is configured.', |
| 2872 |
'status' => 'valid', |
| 2873 |
'icon' => '✓' |
| 2874 |
]; |
| 2875 |
} else { |
| 2876 |
$field_details[] = [ |
| 2877 |
'field' => 'breadcrumb_separator', |
| 2878 |
'label' => 'Breadcrumb separator recommended for better formatting.', |
| 2879 |
'status' => 'suggestion', |
| 2880 |
'icon' => '⚠' |
| 2881 |
]; |
| 2882 |
} |
| 2883 |
|
| 2884 |
// Breadcrumb Prefix validation (optional) |
| 2885 |
if (!empty($settings['breadcrumb_prefix'])) { |
| 2886 |
$field_details[] = [ |
| 2887 |
'field' => 'breadcrumb_prefix', |
| 2888 |
'label' => 'Breadcrumb prefix is configured.', |
| 2889 |
'status' => 'valid', |
| 2890 |
'icon' => '✓' |
| 2891 |
]; |
| 2892 |
} else { |
| 2893 |
$field_details[] = [ |
| 2894 |
'field' => 'breadcrumb_prefix', |
| 2895 |
'label' => 'Breadcrumb prefix is optional but can improve user guidance.', |
| 2896 |
'status' => 'suggestion', |
| 2897 |
'icon' => '⚠' |
| 2898 |
]; |
| 2899 |
} |
| 2900 |
|
| 2901 |
// Show Current Page validation |
| 2902 |
$field_details[] = [ |
| 2903 |
'field' => 'show_current_page', |
| 2904 |
'label' => isset($settings['show_current_page']) ? |
| 2905 |
'Current page display preference is configured.' : |
| 2906 |
'Current page display preference is set to default.', |
| 2907 |
'status' => 'valid', |
| 2908 |
'icon' => '✓' |
| 2909 |
]; |
| 2910 |
} else { |
| 2911 |
$field_details[] = [ |
| 2912 |
'field' => 'breadcrumbs_enabled', |
| 2913 |
'label' => 'Breadcrumbs recommended for better user experience and SEO.', |
| 2914 |
'status' => 'suggestion', |
| 2915 |
'icon' => '⚠' |
| 2916 |
]; |
| 2917 |
} |
| 2918 |
|
| 2919 |
return $field_details; |
| 2920 |
} |
| 2921 |
|
| 2922 |
/** |
| 2923 |
* Validate local SEO settings |
| 2924 |
* |
| 2925 |
* @since 1.0.0 |
| 2926 |
* |
| 2927 |
* @param array $settings Settings array to validate |
| 2928 |
* @return array Local SEO validation results |
| 2929 |
*/ |
| 2930 |
private function validate_local_seo_settings(array $settings): array { |
| 2931 |
$validation = [ |
| 2932 |
'valid' => true, |
| 2933 |
'errors' => [], |
| 2934 |
'warnings' => [], |
| 2935 |
'suggestions' => [] |
| 2936 |
]; |
| 2937 |
|
| 2938 |
// Business name is what makes the LocalBusiness schema useful, but it |
| 2939 |
// cannot be a blocking error: the toggle is what reveals the business |
| 2940 |
// fields, so requiring the name up front makes enabling Local SEO |
| 2941 |
// impossible. The frontend already skips the output while the name is |
| 2942 |
// empty (see Seo_Manager::output_local_seo_meta_tags()). |
| 2943 |
if (empty($settings['business_name'])) { |
| 2944 |
$validation['warnings'][] = 'Business name is missing - required before local business schema is output'; |
| 2945 |
} elseif (strlen($settings['business_name']) > 100) { |
| 2946 |
$validation['warnings'][] = 'Business name is very long, consider shortening for better display'; |
| 2947 |
} |
| 2948 |
|
| 2949 |
// Validate business address components (NAP consistency) |
| 2950 |
$required_address_fields = [ |
| 2951 |
'business_address' => 'Business address', |
| 2952 |
'business_city' => 'Business city', |
| 2953 |
'business_state' => 'Business state/province', |
| 2954 |
'business_country' => 'Business country' |
| 2955 |
]; |
| 2956 |
|
| 2957 |
foreach ($required_address_fields as $field => $label) { |
| 2958 |
if (empty($settings[$field])) { |
| 2959 |
$validation['warnings'][] = "{$label} is missing - important for NAP consistency and local search"; |
| 2960 |
} |
| 2961 |
} |
| 2962 |
|
| 2963 |
// Validate postal code (recommended) |
| 2964 |
if (empty($settings['business_postal_code'])) { |
| 2965 |
$validation['suggestions'][] = 'Add postal code for more precise location targeting'; |
| 2966 |
} |
| 2967 |
|
| 2968 |
// Validate phone number |
| 2969 |
if (empty($settings['business_phone'])) { |
| 2970 |
$validation['warnings'][] = 'Business phone number is missing - important for local SEO and customer contact'; |
| 2971 |
} elseif (!$this->validate_phone_format($settings['business_phone'])) { |
| 2972 |
$validation['suggestions'][] = 'Phone number format could be improved for consistency'; |
| 2973 |
} |
| 2974 |
|
| 2975 |
// Validate email address |
| 2976 |
if (!empty($settings['business_email']) && !is_email($settings['business_email'])) { |
| 2977 |
$validation['errors'][] = 'Business email address format is invalid'; |
| 2978 |
$validation['valid'] = false; |
| 2979 |
} |
| 2980 |
|
| 2981 |
// Validate coordinates if provided |
| 2982 |
if (!empty($settings['business_latitude']) || !empty($settings['business_longitude'])) { |
| 2983 |
if (empty($settings['business_latitude']) || empty($settings['business_longitude'])) { |
| 2984 |
$validation['warnings'][] = 'Both latitude and longitude are required for geo-location'; |
| 2985 |
} elseif (!$this->validate_coordinates($settings['business_latitude'], $settings['business_longitude'])) { |
| 2986 |
$validation['errors'][] = 'Invalid latitude or longitude coordinates'; |
| 2987 |
$validation['valid'] = false; |
| 2988 |
} |
| 2989 |
} else { |
| 2990 |
$validation['suggestions'][] = 'Add latitude and longitude coordinates for precise location targeting'; |
| 2991 |
} |
| 2992 |
|
| 2993 |
// Validate business hours |
| 2994 |
if (!empty($settings['business_hours']) && is_array($settings['business_hours'])) { |
| 2995 |
$hours_validation = $this->validate_business_hours($settings['business_hours']); |
| 2996 |
if (!$hours_validation['valid']) { |
| 2997 |
$validation['warnings'] = array_merge($validation['warnings'], $hours_validation['warnings']); |
| 2998 |
} |
| 2999 |
} else { |
| 3000 |
$validation['suggestions'][] = 'Add business hours to improve local search visibility'; |
| 3001 |
} |
| 3002 |
|
| 3003 |
// Business type, through the shared rule (#622). This is the only place |
| 3004 |
// it is reported on the generic path: validate_settings() with no tab |
| 3005 |
// context attaches basic-info field details, not business-info ones, so |
| 3006 |
// without this the setting would go unreported there entirely. |
| 3007 |
$business_type = $this->business_type_status($settings); |
| 3008 |
if ('suggestion' === $business_type['status']) { |
| 3009 |
$validation['suggestions'][] = $business_type['message']; |
| 3010 |
} |
| 3011 |
|
| 3012 |
return $validation; |
| 3013 |
} |
| 3014 |
|
| 3015 |
/** |
| 3016 |
* Validate phone number format |
| 3017 |
* |
| 3018 |
* @since 1.0.0 |
| 3019 |
* |
| 3020 |
* @param string $phone_number Phone number to validate |
| 3021 |
* @return bool True if format is acceptable |
| 3022 |
*/ |
| 3023 |
private function validate_phone_format(string $phone_number): bool { |
| 3024 |
// Remove all non-numeric characters except + for international numbers |
| 3025 |
$cleaned = preg_replace('/[^\d+]/', '', $phone_number); |
| 3026 |
|
| 3027 |
// Check for common valid formats |
| 3028 |
return ( |
| 3029 |
preg_match('/^\d{10}$/', $cleaned) || // 10 digits (US) |
| 3030 |
preg_match('/^1\d{10}$/', $cleaned) || // 1 + 10 digits (US with country code) |
| 3031 |
preg_match('/^\+\d{7,15}$/', $cleaned) // International format |
| 3032 |
); |
| 3033 |
} |
| 3034 |
|
| 3035 |
/** |
| 3036 |
* Get output data for frontend rendering (implements interface) |
| 3037 |
* |
| 3038 |
* @since 1.0.0 |
| 3039 |
* |
| 3040 |
* @param string $context_type The context type |
| 3041 |
* @param int|null $context_id Optional. Context ID |
| 3042 |
* @return array Output data ready for frontend rendering |
| 3043 |
*/ |
| 3044 |
public function get_output_data(string $context_type, ?int $context_id): array { |
| 3045 |
$settings = $this->get_settings($context_type, $context_id); |
| 3046 |
|
| 3047 |
$output = [ |
| 3048 |
'title' => '', |
| 3049 |
'breadcrumbs' => [], |
| 3050 |
'identity' => [], |
| 3051 |
'robots_txt' => [], |
| 3052 |
'enabled' => $settings['enabled'] ?? true |
| 3053 |
]; |
| 3054 |
|
| 3055 |
if (!$output['enabled']) { |
| 3056 |
return $output; |
| 3057 |
} |
| 3058 |
|
| 3059 |
// Generate title for current context |
| 3060 |
$title_data = $this->extract_title_data($context_type, $context_id); |
| 3061 |
$output['title'] = $this->generate_title( |
| 3062 |
$settings['title_template'] ?? 'default', |
| 3063 |
$title_data, |
| 3064 |
$context_type |
| 3065 |
); |
| 3066 |
|
| 3067 |
// Generate breadcrumbs if enabled |
| 3068 |
if (!empty($settings['breadcrumbs_enabled'])) { |
| 3069 |
$breadcrumb_options = [ |
| 3070 |
'context_type' => $context_type, |
| 3071 |
'context_id' => $context_id |
| 3072 |
]; |
| 3073 |
$output['breadcrumbs'] = $this->generate_breadcrumbs( |
| 3074 |
$settings['breadcrumb_type'] ?? 'hierarchical', |
| 3075 |
$breadcrumb_options |
| 3076 |
); |
| 3077 |
} |
| 3078 |
|
| 3079 |
// Get site identity data |
| 3080 |
$output['identity'] = $this->get_site_identity_data($settings); |
| 3081 |
|
| 3082 |
// Get robots.txt data if enabled |
| 3083 |
if (!empty($settings['robots_txt_enabled'])) { |
| 3084 |
$output['robots_txt'] = $this->generate_robots_txt($settings['custom_robots_rules'] ?? []); |
| 3085 |
} |
| 3086 |
|
| 3087 |
return $output; |
| 3088 |
} |
| 3089 |
|
| 3090 |
/** |
| 3091 |
* Keys the Site Identity screens store beyond the 16 defaults. |
| 3092 |
* |
| 3093 |
* Title formats, breadcrumb configuration, the hero fields, the business |
| 3094 |
* block and the wizard's identity fields are all real settings written by |
| 3095 |
* this manager, none of which get_default_settings() names — it seeds only |
| 3096 |
* the values a fresh install needs. Gating on defaults alone would stop |
| 3097 |
* every one of them saving (#452). |
| 3098 |
* |
| 3099 |
* @since 2.0.1 |
| 3100 |
* |
| 3101 |
* @return string[] |
| 3102 |
*/ |
| 3103 |
/** |
| 3104 |
* The stored alternate name(s), shaped for schema output. |
| 3105 |
* |
| 3106 |
* schema.org and Google both allow `alternateName` to carry one value or |
| 3107 |
* several, and the store already round-trips either shape, so this accepts |
| 3108 |
* both and normalises: null when there is nothing to publish, a bare string |
| 3109 |
* for one name, a list for more. Emitting a one-element array would be |
| 3110 |
* valid but noisier than it needs to be. |
| 3111 |
* |
| 3112 |
* Shared because both WebSite producers need it and must agree — a property |
| 3113 |
* added to one and not the other is how #688 happened. |
| 3114 |
* |
| 3115 |
* @since 2.7.0 |
| 3116 |
* |
| 3117 |
* @param mixed $value Stored alternate_name value. |
| 3118 |
* @return string|string[]|null |
| 3119 |
*/ |
| 3120 |
public static function alternate_name_for_schema($value) { |
| 3121 |
$names = []; |
| 3122 |
|
| 3123 |
foreach ((array) $value as $name) { |
| 3124 |
if (!is_scalar($name)) { |
| 3125 |
continue; |
| 3126 |
} |
| 3127 |
|
| 3128 |
$name = trim((string) $name); |
| 3129 |
|
| 3130 |
if ('' !== $name && !in_array($name, $names, true)) { |
| 3131 |
$names[] = $name; |
| 3132 |
} |
| 3133 |
} |
| 3134 |
|
| 3135 |
if (empty($names)) { |
| 3136 |
return null; |
| 3137 |
} |
| 3138 |
|
| 3139 |
return 1 === count($names) ? $names[0] : $names; |
| 3140 |
} |
| 3141 |
|
| 3142 |
protected function additional_setting_keys(): array { |
| 3143 |
return [ |
| 3144 |
// Title formats, one per context. |
| 3145 |
'homepage_title', 'post_title', 'page_title', 'category_title', |
| 3146 |
'tag_title', 'author_title', 'search_title', 'archive_title', |
| 3147 |
// The blog-index homepage's meta description (#897). |
| 3148 |
'homepage_description', |
| 3149 |
// Breadcrumbs. |
| 3150 |
'breadcrumb_prefix', 'show_current_page', 'breadcrumb_use_seo_title', |
| 3151 |
// Identity, as written by the setup wizard and the importers. |
| 3152 |
'alternate_name', 'identity_type', 'represents', |
| 3153 |
'default_meta_description', 'default_social_image', |
| 3154 |
'social_media_accounts', |
| 3155 |
// Schema toggles that live on this screen. |
| 3156 |
'organization_schema', 'knowledge_graph', |
| 3157 |
// Robots rules composed by the Robots.txt panel. |
| 3158 |
'custom_robots_rules', |
| 3159 |
// Per-agent AI crawler allow/block map (#657). |
| 3160 |
'ai_crawler_rules', |
| 3161 |
// Hero section. |
| 3162 |
'hero_title', 'hero_subtitle', 'hero_cta_text', 'hero_cta_url', |
| 3163 |
'hero_background_image', |
| 3164 |
// Local SEO / business details. |
| 3165 |
'local_seo_enabled', 'business_type', 'business_name', |
| 3166 |
'business_address', 'business_city', 'business_state', |
| 3167 |
'business_postal_code', 'business_country', 'business_phone', |
| 3168 |
'business_email', 'business_latitude', 'business_longitude', |
| 3169 |
'business_price_range', 'business_hours', |
| 3170 |
]; |
| 3171 |
} |
| 3172 |
|
| 3173 |
/** |
| 3174 |
* Sanitize settings, normalising the AI crawler rule map. |
| 3175 |
* |
| 3176 |
* The generic array sanitizer keeps the shape but says nothing about the |
| 3177 |
* values: a payload could store `ai_crawler_rules[gptbot] = "maybe"`, or a |
| 3178 |
* slug no crawler answers to, and both would round-trip through every |
| 3179 |
* later response. Normalising here rather than in the REST handler puts it |
| 3180 |
* on the one path every writer shares — the settings route, the robots |
| 3181 |
* route and the MCP abilities all land in save_settings() (#657). |
| 3182 |
* |
| 3183 |
* @since 2.5.0 |
| 3184 |
* |
| 3185 |
* @param array $settings Settings to sanitize. |
| 3186 |
* @param string $context_type Context type. |
| 3187 |
* @return array Sanitized settings. |
| 3188 |
*/ |
| 3189 |
protected function sanitize_settings(array $settings, string $context_type = 'site'): array { |
| 3190 |
$sanitized = parent::sanitize_settings($settings, $context_type); |
| 3191 |
|
| 3192 |
if (array_key_exists('ai_crawler_rules', $sanitized)) { |
| 3193 |
$sanitized['ai_crawler_rules'] = AI_Crawlers::normalize_rules($sanitized['ai_crawler_rules']); |
| 3194 |
} |
| 3195 |
|
| 3196 |
// Same reasoning one key up, for the scheme override (#638). Anything |
| 3197 |
// that is not one of the three modes means "follow WordPress", and is |
| 3198 |
// stored as that rather than kept verbatim — otherwise get-site-identity |
| 3199 |
// -settings would report a scheme the site does not actually publish. |
| 3200 |
if (array_key_exists('canonical_scheme', $sanitized)) { |
| 3201 |
$sanitized['canonical_scheme'] = in_array($sanitized['canonical_scheme'], Url_Scheme::MODES, true) |
| 3202 |
? $sanitized['canonical_scheme'] |
| 3203 |
: Url_Scheme::AUTOMATIC; |
| 3204 |
} |
| 3205 |
|
| 3206 |
// Same reasoning again for the business type. It goes straight into |
| 3207 |
// LocalBusiness schema, so a type that is not in the schema.org |
| 3208 |
// vocabulary is invalid structured data — and storing it verbatim would |
| 3209 |
// have get-site-identity-settings report a type the site cannot |
| 3210 |
// actually publish. An empty value keeps meaning "not set"; anything |
| 3211 |
// else unrecognised falls back to the general-purpose root (#623). |
| 3212 |
if (array_key_exists('business_type', $sanitized)) { |
| 3213 |
$type = (string) $sanitized['business_type']; |
| 3214 |
|
| 3215 |
if ('' !== $type && !\ThinkRank\Config\Local_Business_Types_Config::is_valid($type)) { |
| 3216 |
$type = \ThinkRank\Config\Local_Business_Types_Config::ROOT; |
| 3217 |
} |
| 3218 |
|
| 3219 |
$sanitized['business_type'] = $type; |
| 3220 |
} |
| 3221 |
|
| 3222 |
return $sanitized; |
| 3223 |
} |
| 3224 |
|
| 3225 |
/** |
| 3226 |
* schema.org's general-purpose LocalBusiness type. |
| 3227 |
* |
| 3228 |
* The default, the first option in the control, and a valid answer in its |
| 3229 |
* own right — which is the whole point of #622. |
| 3230 |
* |
| 3231 |
* @since 2.10.0 |
| 3232 |
* @var string |
| 3233 |
*/ |
| 3234 |
private const GENERAL_BUSINESS_TYPE = 'LocalBusiness'; |
| 3235 |
|
| 3236 |
/** |
| 3237 |
* The one rule for whether a business type needs the user's attention. |
| 3238 |
* |
| 3239 |
* There were three, with two wordings and two different conditions. Two |
| 3240 |
* fired when the value was empty; the third fired when it WAS |
| 3241 |
* `LocalBusiness` — which is the default, the first option in the control |
| 3242 |
* and a perfectly valid schema.org type. So the warning appeared out of the |
| 3243 |
* box for every site, could not be cleared without choosing a type that |
| 3244 |
* might be inaccurate, and on an empty value it appeared three times in two |
| 3245 |
* different phrasings, which is why it was reported as showing twice (#622). |
| 3246 |
* |
| 3247 |
* The rule now: a type is expected, and any type in the vocabulary is a |
| 3248 |
* correct answer. Only an unset value is worth prompting about. |
| 3249 |
* `LocalBusiness` is the general-purpose answer and is accepted as one — |
| 3250 |
* with a note that a more specific type sharpens the schema, phrased as the |
| 3251 |
* guidance it is rather than as a fault the user has to clear. |
| 3252 |
* |
| 3253 |
* @since 2.10.0 |
| 3254 |
* |
| 3255 |
* @param array $settings Site identity settings. |
| 3256 |
* @return array{status:string,message:string} `valid` or `suggestion`. |
| 3257 |
*/ |
| 3258 |
private function business_type_status(array $settings): array { |
| 3259 |
$type = trim((string) ($settings['business_type'] ?? '')); |
| 3260 |
|
| 3261 |
if ('' === $type) { |
| 3262 |
return [ |
| 3263 |
'status' => 'suggestion', |
| 3264 |
'message' => __('Select a business type so your local schema describes the right kind of business.', 'thinkrank'), |
| 3265 |
]; |
| 3266 |
} |
| 3267 |
|
| 3268 |
// The literal rather than a constant from the expanded type list (#623): |
| 3269 |
// that lands on its own branch, and this fix must not wait on it. |
| 3270 |
if (self::GENERAL_BUSINESS_TYPE === $type) { |
| 3271 |
return [ |
| 3272 |
'status' => 'valid', |
| 3273 |
'message' => __('Business type is set to Local Business. A more specific type sharpens your schema, if one fits.', 'thinkrank'), |
| 3274 |
]; |
| 3275 |
} |
| 3276 |
|
| 3277 |
return [ |
| 3278 |
'status' => 'valid', |
| 3279 |
'message' => __('Business type is selected for proper schema markup.', 'thinkrank'), |
| 3280 |
]; |
| 3281 |
} |
| 3282 |
|
| 3283 |
/** |
| 3284 |
* Get default settings for a context type (implements interface) |
| 3285 |
* |
| 3286 |
* @since 1.0.0 |
| 3287 |
* |
| 3288 |
* @param string $context_type The context type to get defaults for |
| 3289 |
* @return array Default settings array |
| 3290 |
*/ |
| 3291 |
public function get_default_settings(string $context_type): array { |
| 3292 |
$defaults = [ |
| 3293 |
'enabled' => true, |
| 3294 |
'title_template' => 'default', |
| 3295 |
'title_separator' => 'pipe', |
| 3296 |
'site_name' => get_bloginfo('name'), |
| 3297 |
'site_description' => get_bloginfo('description'), |
| 3298 |
'tagline' => get_bloginfo('description'), |
| 3299 |
'breadcrumbs_enabled' => true, |
| 3300 |
'breadcrumb_type' => 'hierarchical', |
| 3301 |
'breadcrumb_home_text' => 'Home', |
| 3302 |
'breadcrumb_separator' => '>', |
| 3303 |
'robots_txt_enabled' => true, |
| 3304 |
'allow_search_engines' => true, |
| 3305 |
// Answer 404 when a content selector in the URL resolved to |
| 3306 |
// nothing (#634). On by default, unlike the other new settings |
| 3307 |
// here: it changes no URL a visitor or a correct crawler uses, only |
| 3308 |
// ones where WordPress resolved nothing and served the blog listing |
| 3309 |
// at 200 anyway. |
| 3310 |
'query_protection' => true, |
| 3311 |
|
| 3312 |
// Feed controls (#635). All three off, so an upgrade changes |
| 3313 |
// nothing about what an existing site already sends its |
| 3314 |
// subscribers; a brand-new install is seeded with the signature and |
| 3315 |
// the noindex on, in Activator::seed_feed_defaults(). |
| 3316 |
'feed_excerpt_only' => false, |
| 3317 |
'feed_source_link' => false, |
| 3318 |
'feed_noindex' => false, |
| 3319 |
|
| 3320 |
// The scheme self-referential URLs go out with (#638). 'automatic' |
| 3321 |
// means substitute nothing and follow WordPress, which is what |
| 3322 |
// every site did before the setting existed. |
| 3323 |
'canonical_scheme' => Url_Scheme::AUTOMATIC, |
| 3324 |
'robots_txt_content' => '', |
| 3325 |
// Empty map = every AI crawler allowed. Defaults must stay |
| 3326 |
// permissive so an upgrade never starts blocking a crawler a site |
| 3327 |
// was happily serving (#657). |
| 3328 |
'ai_crawler_rules' => [], |
| 3329 |
'logo_url' => '', |
| 3330 |
'favicon_url' => '', |
| 3331 |
'apple_touch_icon_url' => '' |
| 3332 |
]; |
| 3333 |
|
| 3334 |
// Context-specific defaults |
| 3335 |
switch ($context_type) { |
| 3336 |
case 'site': |
| 3337 |
// Site-wide defaults are already set above |
| 3338 |
break; |
| 3339 |
case 'post': |
| 3340 |
$defaults['title_template'] = 'default'; |
| 3341 |
$defaults['breadcrumb_type'] = 'taxonomy'; |
| 3342 |
break; |
| 3343 |
case 'page': |
| 3344 |
$defaults['title_template'] = 'default'; |
| 3345 |
$defaults['breadcrumb_type'] = 'hierarchical'; |
| 3346 |
break; |
| 3347 |
case 'product': |
| 3348 |
$defaults['title_template'] = 'category'; |
| 3349 |
$defaults['breadcrumb_type'] = 'taxonomy'; |
| 3350 |
break; |
| 3351 |
} |
| 3352 |
|
| 3353 |
return $defaults; |
| 3354 |
} |
| 3355 |
|
| 3356 |
/** |
| 3357 |
* Get settings schema definition (implements interface) |
| 3358 |
* |
| 3359 |
* @since 1.0.0 |
| 3360 |
* |
| 3361 |
* @param string $context_type The context type to get schema for |
| 3362 |
* @return array Settings schema definition |
| 3363 |
*/ |
| 3364 |
public function get_settings_schema(string $context_type): array { |
| 3365 |
return [ |
| 3366 |
'enabled' => [ |
| 3367 |
'type' => 'boolean', |
| 3368 |
'title' => 'Enable Site Identity', |
| 3369 |
'description' => 'Enable site identity management features', |
| 3370 |
'default' => true |
| 3371 |
], |
| 3372 |
'title_template' => [ |
| 3373 |
'type' => 'string', |
| 3374 |
'title' => 'Title Template', |
| 3375 |
'description' => 'Template for generating page titles', |
| 3376 |
'enum' => array_keys($this->title_templates), |
| 3377 |
'default' => 'default' |
| 3378 |
], |
| 3379 |
'title_separator' => [ |
| 3380 |
'type' => 'string', |
| 3381 |
'title' => 'Title Separator', |
| 3382 |
'description' => 'Character used to separate title elements', |
| 3383 |
'enum' => array_keys(self::$title_separators), |
| 3384 |
'default' => 'pipe' |
| 3385 |
], |
| 3386 |
'site_name' => [ |
| 3387 |
'type' => 'string', |
| 3388 |
'title' => 'Site Name', |
| 3389 |
'description' => 'Official name of the website', |
| 3390 |
'maxLength' => 60, |
| 3391 |
'default' => get_bloginfo('name') |
| 3392 |
], |
| 3393 |
'site_description' => [ |
| 3394 |
'type' => 'string', |
| 3395 |
'title' => 'Site Description', |
| 3396 |
'description' => 'Brief description of the website', |
| 3397 |
'maxLength' => 160, |
| 3398 |
'default' => get_bloginfo('description') |
| 3399 |
], |
| 3400 |
'breadcrumbs_enabled' => [ |
| 3401 |
'type' => 'boolean', |
| 3402 |
'title' => 'Enable Breadcrumbs', |
| 3403 |
'description' => 'Enable breadcrumb navigation generation', |
| 3404 |
'default' => true |
| 3405 |
], |
| 3406 |
'breadcrumb_type' => [ |
| 3407 |
'type' => 'string', |
| 3408 |
'title' => 'Breadcrumb Type', |
| 3409 |
'description' => 'Type of breadcrumb navigation to generate', |
| 3410 |
'enum' => array_keys($this->breadcrumb_types), |
| 3411 |
'default' => 'hierarchical' |
| 3412 |
], |
| 3413 |
'robots_txt_enabled' => [ |
| 3414 |
'type' => 'boolean', |
| 3415 |
'title' => 'Enable Robots.txt Management', |
| 3416 |
'description' => 'Enable automatic robots.txt generation and management', |
| 3417 |
'default' => true |
| 3418 |
], |
| 3419 |
'logo_url' => [ |
| 3420 |
'type' => 'string', |
| 3421 |
'title' => 'Logo URL', |
| 3422 |
'description' => 'URL of the site logo image', |
| 3423 |
'format' => 'uri', |
| 3424 |
'default' => '' |
| 3425 |
], |
| 3426 |
'favicon_url' => [ |
| 3427 |
'type' => 'string', |
| 3428 |
'title' => 'Favicon URL', |
| 3429 |
'description' => 'URL of the site favicon', |
| 3430 |
'format' => 'uri', |
| 3431 |
'default' => '' |
| 3432 |
] |
| 3433 |
]; |
| 3434 |
} |
| 3435 |
|
| 3436 |
/** |
| 3437 |
* Prepare title placeholders for replacement |
| 3438 |
* |
| 3439 |
* @since 1.0.0 |
| 3440 |
* |
| 3441 |
* @param array $data Content data |
| 3442 |
* @param string $context Context type |
| 3443 |
* @param array $settings Site settings |
| 3444 |
* @return array Placeholder values |
| 3445 |
*/ |
| 3446 |
private function prepare_title_placeholders(array $data, string $context, array $settings): array { |
| 3447 |
$placeholders = [ |
| 3448 |
'%title%' => $data['title'] ?? '', |
| 3449 |
// `?:` rather than `??`: these are persisted as '' rather than left |
| 3450 |
// unset, and '' is not null, so the null-coalesce never reached the |
| 3451 |
// WordPress fallback (#398). |
| 3452 |
'%sitename%' => ($settings['site_name'] ?? '') ?: get_bloginfo('name'), |
| 3453 |
'%tagline%' => ($settings['tagline'] ?? '') ?: get_bloginfo('description'), |
| 3454 |
'%separator%' => '', // Will be replaced with actual separator |
| 3455 |
'%category%' => '', |
| 3456 |
'%author%' => '', |
| 3457 |
'%date%' => '', |
| 3458 |
'%searchterm%' => '' |
| 3459 |
]; |
| 3460 |
|
| 3461 |
// Context-specific placeholders |
| 3462 |
switch ($context) { |
| 3463 |
case 'post': |
| 3464 |
case 'page': |
| 3465 |
case 'product': |
| 3466 |
if (!empty($data['context_id'])) { |
| 3467 |
$post = get_post($data['context_id']); |
| 3468 |
if ($post) { |
| 3469 |
$placeholders['%title%'] = get_the_title($post); |
| 3470 |
$placeholders['%author%'] = get_the_author_meta('display_name', $post->post_author); |
| 3471 |
$placeholders['%date%'] = get_the_date('F j, Y', $post); |
| 3472 |
|
| 3473 |
// Get primary category |
| 3474 |
$categories = get_the_category($post->ID); |
| 3475 |
if (!empty($categories)) { |
| 3476 |
$placeholders['%category%'] = $categories[0]->name; |
| 3477 |
} |
| 3478 |
} |
| 3479 |
} |
| 3480 |
break; |
| 3481 |
case 'search': |
| 3482 |
$placeholders['%searchterm%'] = get_search_query(); |
| 3483 |
break; |
| 3484 |
} |
| 3485 |
|
| 3486 |
return $placeholders; |
| 3487 |
} |
| 3488 |
|
| 3489 |
/** |
| 3490 |
* Replace title placeholders with actual values |
| 3491 |
* |
| 3492 |
* @since 1.0.0 |
| 3493 |
* |
| 3494 |
* @param string $template Title template |
| 3495 |
* @param array $placeholders Placeholder values |
| 3496 |
* @param string $separator Title separator |
| 3497 |
* @return string Processed title |
| 3498 |
*/ |
| 3499 |
private function replace_title_placeholders(string $template, array $placeholders, string $separator): string { |
| 3500 |
// Replace separator placeholder |
| 3501 |
$placeholders['%separator%'] = $separator; |
| 3502 |
|
| 3503 |
// Replace all placeholders |
| 3504 |
$title = str_replace(array_keys($placeholders), array_values($placeholders), $template); |
| 3505 |
|
| 3506 |
// Clean up empty placeholders and extra separators |
| 3507 |
$title = preg_replace('/\s*' . preg_quote($separator, '/') . '\s*' . preg_quote($separator, '/') . '\s*/', ' ' . $separator . ' ', $title); |
| 3508 |
$title = preg_replace('/^\s*' . preg_quote($separator, '/') . '\s*|\s*' . preg_quote($separator, '/') . '\s*$/', '', $title); |
| 3509 |
|
| 3510 |
return trim($title); |
| 3511 |
} |
| 3512 |
|
| 3513 |
/** |
| 3514 |
* Get title separator symbol |
| 3515 |
* |
| 3516 |
* @since 1.0.0 |
| 3517 |
* |
| 3518 |
* @param string $separator_key Separator key |
| 3519 |
* @return string Separator symbol |
| 3520 |
*/ |
| 3521 |
private function get_title_separator(string $separator_key): string { |
| 3522 |
return self::$title_separators[$separator_key]['symbol'] ?? self::$title_separators['pipe']['symbol']; |
| 3523 |
} |
| 3524 |
|
| 3525 |
/** |
| 3526 |
* Optimize title for SEO |
| 3527 |
* |
| 3528 |
* @since 1.0.0 |
| 3529 |
* |
| 3530 |
* @param string $title Title to optimize |
| 3531 |
* @param string $context Context type |
| 3532 |
* @return string Optimized title |
| 3533 |
*/ |
| 3534 |
private function optimize_title(string $title, string $context): string { |
| 3535 |
// Remove extra whitespace |
| 3536 |
$title = preg_replace('/\s+/', ' ', $title); |
| 3537 |
$title = trim($title); |
| 3538 |
|
| 3539 |
// Ensure title is not too long (60 characters max for SEO). |
| 3540 |
// All three units here were wrong for non-Latin text: strlen() counts |
| 3541 |
// BYTES so the gate fired at 20 Thai characters, wp_trim_words() counts |
| 3542 |
// CHARACTERS on th/ja/zh_* so `8` cut the title to 8 of them, and |
| 3543 |
// substr() cuts bytes so it split a character mid-sequence (#687). |
| 3544 |
$title = \ThinkRank\Core\Seo_Text::trim_to_length( |
| 3545 |
$title, |
| 3546 |
\ThinkRank\Core\Seo_Text::TITLE_MAX_LENGTH |
| 3547 |
); |
| 3548 |
|
| 3549 |
// Ensure title is not empty |
| 3550 |
if (empty($title)) { |
| 3551 |
$title = get_bloginfo('name'); |
| 3552 |
} |
| 3553 |
|
| 3554 |
return $title; |
| 3555 |
} |
| 3556 |
|
| 3557 |
/** |
| 3558 |
* Extract title data from context |
| 3559 |
* |
| 3560 |
* @since 1.0.0 |
| 3561 |
* |
| 3562 |
* @param string $context_type Context type |
| 3563 |
* @param int|null $context_id Context ID |
| 3564 |
* @return array Title data |
| 3565 |
*/ |
| 3566 |
private function extract_title_data(string $context_type, ?int $context_id): array { |
| 3567 |
$data = [ |
| 3568 |
'title' => '', |
| 3569 |
'context_type' => $context_type, |
| 3570 |
'context_id' => $context_id |
| 3571 |
]; |
| 3572 |
|
| 3573 |
switch ($context_type) { |
| 3574 |
case 'site': |
| 3575 |
$data['title'] = get_bloginfo('name'); |
| 3576 |
break; |
| 3577 |
case 'post': |
| 3578 |
case 'page': |
| 3579 |
case 'product': |
| 3580 |
if ($context_id) { |
| 3581 |
$data['title'] = get_the_title($context_id); |
| 3582 |
} |
| 3583 |
break; |
| 3584 |
case 'search': |
| 3585 |
$data['title'] = 'Search Results'; |
| 3586 |
break; |
| 3587 |
case '404': |
| 3588 |
$data['title'] = 'Page Not Found'; |
| 3589 |
break; |
| 3590 |
} |
| 3591 |
|
| 3592 |
return $data; |
| 3593 |
} |
| 3594 |
|
| 3595 |
/** |
| 3596 |
* Generate hierarchical breadcrumbs |
| 3597 |
* |
| 3598 |
* @since 1.0.0 |
| 3599 |
* |
| 3600 |
* @param array $options Breadcrumb options |
| 3601 |
* @return array Breadcrumb items |
| 3602 |
*/ |
| 3603 |
private function generate_hierarchical_breadcrumbs(array $options): array { |
| 3604 |
$breadcrumbs = []; |
| 3605 |
|
| 3606 |
// Add home breadcrumb |
| 3607 |
$breadcrumbs[] = [ |
| 3608 |
'title' => 'Home', |
| 3609 |
'url' => home_url(), |
| 3610 |
'position' => 1 |
| 3611 |
]; |
| 3612 |
|
| 3613 |
$context_type = $options['context_type'] ?? ''; |
| 3614 |
$context_id = $options['context_id'] ?? null; |
| 3615 |
|
| 3616 |
if ($context_type === 'post' || $context_type === 'page' || $context_type === 'product') { |
| 3617 |
if ($context_id) { |
| 3618 |
$post = get_post($context_id); |
| 3619 |
if ($post) { |
| 3620 |
// Add parent pages for hierarchical content |
| 3621 |
$ancestors = get_post_ancestors($post); |
| 3622 |
$ancestors = array_reverse($ancestors); |
| 3623 |
|
| 3624 |
$position = 2; |
| 3625 |
foreach ($ancestors as $ancestor_id) { |
| 3626 |
$breadcrumbs[] = [ |
| 3627 |
'title' => get_the_title($ancestor_id), |
| 3628 |
'url' => get_permalink($ancestor_id), |
| 3629 |
'position' => $position++ |
| 3630 |
]; |
| 3631 |
} |
| 3632 |
|
| 3633 |
// Add current page |
| 3634 |
$breadcrumbs[] = [ |
| 3635 |
'title' => get_the_title($post), |
| 3636 |
'url' => get_permalink($post), |
| 3637 |
'position' => $position, |
| 3638 |
'current' => true |
| 3639 |
]; |
| 3640 |
} |
| 3641 |
} |
| 3642 |
} |
| 3643 |
|
| 3644 |
return $breadcrumbs; |
| 3645 |
} |
| 3646 |
|
| 3647 |
/** |
| 3648 |
* Generate taxonomy-based breadcrumbs |
| 3649 |
* |
| 3650 |
* @since 1.0.0 |
| 3651 |
* |
| 3652 |
* @param array $options Breadcrumb options |
| 3653 |
* @return array Breadcrumb items |
| 3654 |
*/ |
| 3655 |
private function generate_taxonomy_breadcrumbs(array $options): array { |
| 3656 |
$breadcrumbs = []; |
| 3657 |
|
| 3658 |
// Add home breadcrumb |
| 3659 |
$breadcrumbs[] = [ |
| 3660 |
'title' => 'Home', |
| 3661 |
'url' => home_url(), |
| 3662 |
'position' => 1 |
| 3663 |
]; |
| 3664 |
|
| 3665 |
$context_type = $options['context_type'] ?? ''; |
| 3666 |
$context_id = $options['context_id'] ?? null; |
| 3667 |
|
| 3668 |
if (($context_type === 'post' || $context_type === 'product') && $context_id) { |
| 3669 |
$post = get_post($context_id); |
| 3670 |
if ($post) { |
| 3671 |
// Get primary category |
| 3672 |
$categories = get_the_category($post->ID); |
| 3673 |
if (!empty($categories)) { |
| 3674 |
$primary_category = $categories[0]; |
| 3675 |
|
| 3676 |
// Add category hierarchy |
| 3677 |
$category_ancestors = get_ancestors($primary_category->term_id, 'category'); |
| 3678 |
$category_ancestors = array_reverse($category_ancestors); |
| 3679 |
|
| 3680 |
$position = 2; |
| 3681 |
foreach ($category_ancestors as $ancestor_id) { |
| 3682 |
$ancestor = get_category($ancestor_id); |
| 3683 |
$breadcrumbs[] = [ |
| 3684 |
'title' => $ancestor->name, |
| 3685 |
'url' => get_category_link($ancestor_id), |
| 3686 |
'position' => $position++ |
| 3687 |
]; |
| 3688 |
} |
| 3689 |
|
| 3690 |
// Add primary category |
| 3691 |
$breadcrumbs[] = [ |
| 3692 |
'title' => $primary_category->name, |
| 3693 |
'url' => get_category_link($primary_category->term_id), |
| 3694 |
'position' => $position++ |
| 3695 |
]; |
| 3696 |
} |
| 3697 |
|
| 3698 |
// Add current post |
| 3699 |
$breadcrumbs[] = [ |
| 3700 |
'title' => get_the_title($post), |
| 3701 |
'url' => get_permalink($post), |
| 3702 |
'position' => $position, |
| 3703 |
'current' => true |
| 3704 |
]; |
| 3705 |
} |
| 3706 |
} |
| 3707 |
|
| 3708 |
return $breadcrumbs; |
| 3709 |
} |
| 3710 |
|
| 3711 |
/** |
| 3712 |
* Generate path-based breadcrumbs |
| 3713 |
* |
| 3714 |
* @since 1.0.0 |
| 3715 |
* |
| 3716 |
* @param array $options Breadcrumb options |
| 3717 |
* @return array Breadcrumb items |
| 3718 |
*/ |
| 3719 |
private function generate_path_breadcrumbs(array $options): array { |
| 3720 |
$breadcrumbs = []; |
| 3721 |
|
| 3722 |
// Add home breadcrumb |
| 3723 |
$breadcrumbs[] = [ |
| 3724 |
'title' => 'Home', |
| 3725 |
'url' => home_url(), |
| 3726 |
'position' => 1 |
| 3727 |
]; |
| 3728 |
|
| 3729 |
// Get current URL path |
| 3730 |
$current_url = home_url(add_query_arg([])); |
| 3731 |
$path = wp_parse_url($current_url, PHP_URL_PATH); |
| 3732 |
$path_parts = array_filter(explode('/', trim($path, '/'))); |
| 3733 |
|
| 3734 |
$position = 2; |
| 3735 |
$cumulative_path = ''; |
| 3736 |
|
| 3737 |
foreach ($path_parts as $part) { |
| 3738 |
$cumulative_path .= '/' . $part; |
| 3739 |
$url = home_url($cumulative_path); |
| 3740 |
|
| 3741 |
// Try to get a meaningful title |
| 3742 |
$title = ucwords(str_replace(['-', '_'], ' ', $part)); |
| 3743 |
|
| 3744 |
$breadcrumbs[] = [ |
| 3745 |
'title' => $title, |
| 3746 |
'url' => $url, |
| 3747 |
'position' => $position++, |
| 3748 |
'current' => $cumulative_path === $path |
| 3749 |
]; |
| 3750 |
} |
| 3751 |
|
| 3752 |
return $breadcrumbs; |
| 3753 |
} |
| 3754 |
|
| 3755 |
/** |
| 3756 |
* Generate custom breadcrumbs |
| 3757 |
* |
| 3758 |
* @since 1.0.0 |
| 3759 |
* |
| 3760 |
* @param array $options Breadcrumb options |
| 3761 |
* @return array Breadcrumb items |
| 3762 |
*/ |
| 3763 |
private function generate_custom_breadcrumbs(array $options): array { |
| 3764 |
// Return custom breadcrumbs if provided in options |
| 3765 |
return $options['custom_breadcrumbs'] ?? []; |
| 3766 |
} |
| 3767 |
|
| 3768 |
/** |
| 3769 |
* Generate breadcrumb schema markup |
| 3770 |
* |
| 3771 |
* @since 1.0.0 |
| 3772 |
* |
| 3773 |
* @param array $breadcrumb_items Breadcrumb items |
| 3774 |
* @return array Schema markup |
| 3775 |
*/ |
| 3776 |
private function generate_breadcrumb_schema(array $breadcrumb_items): array { |
| 3777 |
$schema = [ |
| 3778 |
'@context' => 'https://schema.org', |
| 3779 |
'@type' => 'BreadcrumbList', |
| 3780 |
'itemListElement' => [] |
| 3781 |
]; |
| 3782 |
|
| 3783 |
foreach ($breadcrumb_items as $item) { |
| 3784 |
$schema['itemListElement'][] = [ |
| 3785 |
'@type' => 'ListItem', |
| 3786 |
'position' => $item['position'], |
| 3787 |
'name' => $item['title'], |
| 3788 |
'item' => $item['url'] |
| 3789 |
]; |
| 3790 |
} |
| 3791 |
|
| 3792 |
return $schema; |
| 3793 |
} |
| 3794 |
|
| 3795 |
/** |
| 3796 |
* Generate breadcrumb HTML |
| 3797 |
* |
| 3798 |
* @since 1.0.0 |
| 3799 |
* |
| 3800 |
* @param array $breadcrumb_items Breadcrumb items |
| 3801 |
* @param array $settings Breadcrumb settings |
| 3802 |
* @return string HTML output |
| 3803 |
*/ |
| 3804 |
private function generate_breadcrumb_html(array $breadcrumb_items, array $settings): string { |
| 3805 |
if (empty($breadcrumb_items)) { |
| 3806 |
return ''; |
| 3807 |
} |
| 3808 |
|
| 3809 |
$separator = $settings['separator'] ?? '>'; |
| 3810 |
$html = '<nav class="thinkrank-breadcrumbs" aria-label="Breadcrumb">'; |
| 3811 |
$html .= '<ol class="breadcrumb-list">'; |
| 3812 |
|
| 3813 |
foreach ($breadcrumb_items as $item) { |
| 3814 |
$html .= '<li class="breadcrumb-item">'; |
| 3815 |
|
| 3816 |
if (!empty($item['current'])) { |
| 3817 |
$html .= '<span class="breadcrumb-current" aria-current="page">' . esc_html($item['title']) . '</span>'; |
| 3818 |
} else { |
| 3819 |
$html .= '<a href="' . esc_url($item['url']) . '">' . esc_html($item['title']) . '</a>'; |
| 3820 |
} |
| 3821 |
|
| 3822 |
if ($item['position'] < count($breadcrumb_items)) { |
| 3823 |
$html .= ' <span class="breadcrumb-separator">' . esc_html($separator) . '</span> '; |
| 3824 |
} |
| 3825 |
|
| 3826 |
$html .= '</li>'; |
| 3827 |
} |
| 3828 |
|
| 3829 |
$html .= '</ol>'; |
| 3830 |
$html .= '</nav>'; |
| 3831 |
|
| 3832 |
return $html; |
| 3833 |
} |
| 3834 |
|
| 3835 |
/** |
| 3836 |
* Generate default robots.txt rules |
| 3837 |
* |
| 3838 |
* @since 1.0.0 |
| 3839 |
* |
| 3840 |
* @param array $settings Robots.txt settings |
| 3841 |
* @return array Default rules |
| 3842 |
*/ |
| 3843 |
private function generate_default_robots_rules(array $settings): array { |
| 3844 |
$rules = []; |
| 3845 |
|
| 3846 |
// Full block: when the admin turns off "Allow Search Engines" or enables |
| 3847 |
// WordPress's "Discourage search engines" (Settings → Reading, stored as |
| 3848 |
// blog_public=0), serve a robots.txt that disallows everything rather |
| 3849 |
// than the default per-path rules — otherwise the toggle has no effect. |
| 3850 |
$allow_search = $settings['allow_search_engines'] ?? true; |
| 3851 |
if (empty($allow_search) || !get_option('blog_public')) { |
| 3852 |
$rules[] = ['directive' => 'user_agent', 'value' => '*']; |
| 3853 |
$rules[] = ['directive' => 'disallow', 'value' => '/']; |
| 3854 |
return $rules; |
| 3855 |
} |
| 3856 |
|
| 3857 |
// Default user agent rule |
| 3858 |
$rules[] = [ |
| 3859 |
'directive' => 'user_agent', |
| 3860 |
'value' => '*' |
| 3861 |
]; |
| 3862 |
|
| 3863 |
// WordPress core disallows. |
| 3864 |
// |
| 3865 |
// Deliberately minimal, matching Yoast/Rank Math defaults. We do NOT |
| 3866 |
// block /wp-includes/, /wp-content/plugins/, or /wp-content/themes/: |
| 3867 |
// those paths serve the CSS and JS Google must fetch to render pages, |
| 3868 |
// and blocking them causes "blocked resource" warnings and can hurt |
| 3869 |
// rankings. /wp-json/ is left crawlable for the same reason (embeds, |
| 3870 |
// oEmbed, structured previews). Only wp-admin (bar admin-ajax) and the |
| 3871 |
// handful of non-content endpoints below are disallowed. |
| 3872 |
$default_disallows = [ |
| 3873 |
'/wp-admin/', |
| 3874 |
'/xmlrpc.php', |
| 3875 |
'/readme.html', |
| 3876 |
'/license.txt', |
| 3877 |
]; |
| 3878 |
|
| 3879 |
// WooCommerce: keep cart/checkout/account and add-to-cart query URLs out |
| 3880 |
// of the index to avoid crawl noise and duplicate/session URLs (parity |
| 3881 |
// with Rank Math's WooCommerce robots defaults). |
| 3882 |
if (class_exists('WooCommerce')) { |
| 3883 |
$default_disallows[] = '/cart/'; |
| 3884 |
$default_disallows[] = '/checkout/'; |
| 3885 |
$default_disallows[] = '/my-account/'; |
| 3886 |
$default_disallows[] = '/*add-to-cart=*'; |
| 3887 |
} |
| 3888 |
|
| 3889 |
foreach ($default_disallows as $disallow) { |
| 3890 |
$rules[] = [ |
| 3891 |
'directive' => 'disallow', |
| 3892 |
'value' => $disallow |
| 3893 |
]; |
| 3894 |
} |
| 3895 |
|
| 3896 |
// Allow specific files |
| 3897 |
$default_allows = [ |
| 3898 |
'/wp-admin/admin-ajax.php', |
| 3899 |
'/wp-content/uploads/' |
| 3900 |
]; |
| 3901 |
|
| 3902 |
foreach ($default_allows as $allow) { |
| 3903 |
$rules[] = [ |
| 3904 |
'directive' => 'allow', |
| 3905 |
'value' => $allow |
| 3906 |
]; |
| 3907 |
} |
| 3908 |
|
| 3909 |
// Add sitemap URLs from sitemap settings (auto-sync) |
| 3910 |
$sitemap_urls = $this->get_sitemap_urls_for_robots(); |
| 3911 |
|
| 3912 |
foreach ($sitemap_urls as $sitemap_url) { |
| 3913 |
if (!empty($sitemap_url)) { |
| 3914 |
$rules[] = [ |
| 3915 |
'directive' => 'sitemap', |
| 3916 |
'value' => $sitemap_url |
| 3917 |
]; |
| 3918 |
} |
| 3919 |
} |
| 3920 |
|
| 3921 |
// Add crawl delay if specified |
| 3922 |
if (!empty($settings['crawl_delay'])) { |
| 3923 |
$rules[] = [ |
| 3924 |
'directive' => 'crawl_delay', |
| 3925 |
'value' => (int) $settings['crawl_delay'] |
| 3926 |
]; |
| 3927 |
} |
| 3928 |
|
| 3929 |
return $rules; |
| 3930 |
} |
| 3931 |
|
| 3932 |
/** |
| 3933 |
* Get sitemap URLs from sitemap settings for robots.txt integration |
| 3934 |
* |
| 3935 |
* @since 1.0.0 |
| 3936 |
* @return array Array of sitemap URLs |
| 3937 |
*/ |
| 3938 |
private function get_sitemap_urls_for_robots(): array { |
| 3939 |
// One wrapper over every return path below, including the #104 extras. |
| 3940 |
// The Sitemap: line is the only absolute URL of ours in robots.txt and |
| 3941 |
// the one a crawler follows to find everything else, so it has to carry |
| 3942 |
// the site's scheme preference (#638). Applied here rather than where |
| 3943 |
// the body is assembled, because that path also renders a robots.txt a |
| 3944 |
// site owner typed themselves, and their text is not ours to rewrite. |
| 3945 |
return array_map( |
| 3946 |
static function (string $url): string { |
| 3947 |
return Url_Scheme::apply($url); |
| 3948 |
}, |
| 3949 |
$this->collect_sitemap_urls_for_robots() |
| 3950 |
); |
| 3951 |
} |
| 3952 |
|
| 3953 |
/** |
| 3954 |
* The sitemap URLs robots.txt advertises, before the scheme preference. |
| 3955 |
* |
| 3956 |
* @since 1.0.0 |
| 3957 |
* @return array Array of sitemap URLs |
| 3958 |
*/ |
| 3959 |
private function collect_sitemap_urls_for_robots(): array { |
| 3960 |
try { |
| 3961 |
// Get sitemap settings |
| 3962 |
$sitemap_generator = new \ThinkRank\SEO\Sitemap_Generator(); |
| 3963 |
$sitemap_settings = $sitemap_generator->get_settings('site'); |
| 3964 |
|
| 3965 |
// If sitemap is disabled, return default |
| 3966 |
if (empty($sitemap_settings['enabled'])) { |
| 3967 |
return [home_url('/sitemap.xml')]; |
| 3968 |
} |
| 3969 |
|
| 3970 |
$sitemap_urls = []; |
| 3971 |
$site_url = home_url(); |
| 3972 |
|
| 3973 |
// Extract enabled sitemap URLs. When the index is enabled it is the |
| 3974 |
// only entry worth advertising: every child sitemap is already |
| 3975 |
// listed inside it, so naming them again in robots.txt is pure |
| 3976 |
// redundancy and drifts out of date as soon as a post type is added. |
| 3977 |
$index_url = ''; |
| 3978 |
if (!empty($sitemap_settings['sitemap_urls']) && is_array($sitemap_settings['sitemap_urls'])) { |
| 3979 |
foreach ($sitemap_settings['sitemap_urls'] as $sitemap) { |
| 3980 |
if (empty($sitemap['enabled']) || empty($sitemap['url'])) { |
| 3981 |
continue; |
| 3982 |
} |
| 3983 |
|
| 3984 |
if (($sitemap['type'] ?? '') === 'index') { |
| 3985 |
$index_url = $site_url . $sitemap['url']; |
| 3986 |
continue; |
| 3987 |
} |
| 3988 |
|
| 3989 |
$sitemap_urls[] = $site_url . $sitemap['url']; |
| 3990 |
} |
| 3991 |
} |
| 3992 |
|
| 3993 |
if ($index_url !== '') { |
| 3994 |
// The index covers the children and, on a segmented install, |
| 3995 |
// the local business sitemap too. |
| 3996 |
// |
| 3997 |
// It does not cover a sitemap contributed through |
| 3998 |
// `thinkrank_additional_sitemaps`: the index is built by this |
| 3999 |
// plugin's own generator and never lists them. Returning the |
| 4000 |
// index alone therefore left a contributed sitemap with no |
| 4001 |
// discovery path at all — absent from robots.txt and absent |
| 4002 |
// from the index — so Pro's News sitemap was unreachable on any |
| 4003 |
// install with the index enabled, which is the default (#835). |
| 4004 |
$contributed = []; |
| 4005 |
|
| 4006 |
foreach (\ThinkRank\SEO\Sitemap_Generator::additional_sitemaps() as $path) { |
| 4007 |
$url = home_url($path); |
| 4008 |
|
| 4009 |
if ($url !== $index_url && !in_array($url, $contributed, true)) { |
| 4010 |
$contributed[] = $url; |
| 4011 |
} |
| 4012 |
} |
| 4013 |
|
| 4014 |
return array_merge([$index_url], $contributed); |
| 4015 |
} |
| 4016 |
|
| 4017 |
// Fallback to default if no URLs found |
| 4018 |
if (empty($sitemap_urls)) { |
| 4019 |
$sitemap_urls[] = home_url('/sitemap.xml'); |
| 4020 |
} |
| 4021 |
|
| 4022 |
// No index on this install, so anything not already listed above has |
| 4023 |
// no other discovery path — advertise it directly. The local |
| 4024 |
// business sitemap and the sitemaps other plugins register both land |
| 4025 |
// here for the same reason, so they go through one list (#104). |
| 4026 |
$extra = []; |
| 4027 |
|
| 4028 |
// Not a file test. Under dynamic delivery the local sitemap is |
| 4029 |
// served from PHP and no file is ever written, so file_exists() |
| 4030 |
// silently dropped a sitemap the site really does publish (#752). |
| 4031 |
// On static sites the file is still what proves it, so both count. |
| 4032 |
$local_sitemap_published = file_exists(ABSPATH . 'local-sitemap.xml'); |
| 4033 |
|
| 4034 |
if (!$local_sitemap_published && class_exists('ThinkRank\\SEO\\Sitemap_Generator')) { |
| 4035 |
$generator = new \ThinkRank\SEO\Sitemap_Generator(false); |
| 4036 |
|
| 4037 |
$local_sitemap_published = 'dynamic' === $generator->resolve_delivery_mode() |
| 4038 |
&& $generator->publishes_local_sitemap(); |
| 4039 |
} |
| 4040 |
|
| 4041 |
if ($local_sitemap_published) { |
| 4042 |
$extra[] = '/local-sitemap.xml'; |
| 4043 |
} |
| 4044 |
|
| 4045 |
foreach (\ThinkRank\SEO\Sitemap_Generator::additional_sitemaps() as $path) { |
| 4046 |
$extra[] = $path; |
| 4047 |
} |
| 4048 |
|
| 4049 |
foreach ($extra as $path) { |
| 4050 |
$url = home_url($path); |
| 4051 |
if (!in_array($url, $sitemap_urls, true)) { |
| 4052 |
$sitemap_urls[] = $url; |
| 4053 |
} |
| 4054 |
} |
| 4055 |
|
| 4056 |
return $sitemap_urls; |
| 4057 |
} catch (\Exception $e) { |
| 4058 |
// Fallback to default on error |
| 4059 |
return [home_url('/sitemap.xml')]; |
| 4060 |
} |
| 4061 |
} |
| 4062 |
|
| 4063 |
/** |
| 4064 |
* Validate robots.txt rules |
| 4065 |
* |
| 4066 |
* @since 1.0.0 |
| 4067 |
* |
| 4068 |
* @param array $rules Rules to validate |
| 4069 |
* @return array Validation results |
| 4070 |
*/ |
| 4071 |
private function validate_robots_rules(array $rules): array { |
| 4072 |
$validation = [ |
| 4073 |
'valid' => true, |
| 4074 |
'errors' => [], |
| 4075 |
'warnings' => [], |
| 4076 |
'suggestions' => [] |
| 4077 |
]; |
| 4078 |
|
| 4079 |
$has_user_agent = false; |
| 4080 |
|
| 4081 |
foreach ($rules as $rule) { |
| 4082 |
$directive = $rule['directive'] ?? ''; |
| 4083 |
$value = $rule['value'] ?? ''; |
| 4084 |
|
| 4085 |
// Check if directive is valid |
| 4086 |
if (!isset($this->robots_directives[$directive])) { |
| 4087 |
$validation['errors'][] = "Unknown robots.txt directive: {$directive}"; |
| 4088 |
$validation['valid'] = false; |
| 4089 |
continue; |
| 4090 |
} |
| 4091 |
|
| 4092 |
// Check for required user-agent |
| 4093 |
if ($directive === 'user_agent') { |
| 4094 |
$has_user_agent = true; |
| 4095 |
} |
| 4096 |
|
| 4097 |
// Validate directive-specific rules |
| 4098 |
switch ($directive) { |
| 4099 |
case 'disallow': |
| 4100 |
case 'allow': |
| 4101 |
if (!str_starts_with($value, '/')) { |
| 4102 |
$validation['warnings'][] = "Path '{$value}' should start with '/'"; |
| 4103 |
} |
| 4104 |
break; |
| 4105 |
case 'sitemap': |
| 4106 |
if (!filter_var($value, FILTER_VALIDATE_URL)) { |
| 4107 |
$validation['errors'][] = "Invalid sitemap URL: {$value}"; |
| 4108 |
$validation['valid'] = false; |
| 4109 |
} |
| 4110 |
break; |
| 4111 |
case 'crawl_delay': |
| 4112 |
if (!is_numeric($value) || $value < 0) { |
| 4113 |
$validation['errors'][] = "Crawl delay must be a positive number"; |
| 4114 |
$validation['valid'] = false; |
| 4115 |
} |
| 4116 |
break; |
| 4117 |
} |
| 4118 |
} |
| 4119 |
|
| 4120 |
if (!$has_user_agent) { |
| 4121 |
$validation['errors'][] = 'robots.txt must include at least one User-agent directive'; |
| 4122 |
$validation['valid'] = false; |
| 4123 |
} |
| 4124 |
|
| 4125 |
return $validation; |
| 4126 |
} |
| 4127 |
|
| 4128 |
/** |
| 4129 |
* Build robots.txt content from rules |
| 4130 |
* |
| 4131 |
* @since 1.0.0 |
| 4132 |
* |
| 4133 |
* @param array $rules Robots.txt rules |
| 4134 |
* @return string Robots.txt content |
| 4135 |
*/ |
| 4136 |
private function build_robots_txt_content(array $rules): string { |
| 4137 |
// Body only — no header. The "# generated by ThinkRank SEO" + timestamp |
| 4138 |
// block is added at render time (see robots_txt_header), so it never |
| 4139 |
// gets baked into the stored/editable content and can't show a stale |
| 4140 |
// timestamp on every update. |
| 4141 |
$content = ''; |
| 4142 |
|
| 4143 |
$current_user_agent = ''; |
| 4144 |
$sitemap_started = false; |
| 4145 |
|
| 4146 |
foreach ($rules as $rule) { |
| 4147 |
$directive = $rule['directive'] ?? ''; |
| 4148 |
$value = $rule['value'] ?? ''; |
| 4149 |
|
| 4150 |
switch ($directive) { |
| 4151 |
case 'user_agent': |
| 4152 |
if ($current_user_agent !== $value) { |
| 4153 |
$content .= "\nUser-agent: {$value}\n"; |
| 4154 |
$current_user_agent = $value; |
| 4155 |
} |
| 4156 |
break; |
| 4157 |
case 'disallow': |
| 4158 |
$content .= "Disallow: {$value}\n"; |
| 4159 |
break; |
| 4160 |
case 'allow': |
| 4161 |
$content .= "Allow: {$value}\n"; |
| 4162 |
break; |
| 4163 |
case 'crawl_delay': |
| 4164 |
$content .= "Crawl-delay: {$value}\n"; |
| 4165 |
break; |
| 4166 |
case 'sitemap': |
| 4167 |
// One blank line separates the Sitemap block from the |
| 4168 |
// preceding group, and none appear inside it. A blank line |
| 4169 |
// terminates a record in the robots.txt grammar, so putting |
| 4170 |
// one between every directive was invalid formatting. |
| 4171 |
if (!$sitemap_started) { |
| 4172 |
$content .= "\n"; |
| 4173 |
$sitemap_started = true; |
| 4174 |
} |
| 4175 |
$content .= "Sitemap: {$value}\n"; |
| 4176 |
break; |
| 4177 |
} |
| 4178 |
} |
| 4179 |
|
| 4180 |
return ltrim($content, "\n"); |
| 4181 |
} |
| 4182 |
|
| 4183 |
/** |
| 4184 |
* Parse a robots.txt body back into the {directive, value} rule shape. |
| 4185 |
* |
| 4186 |
* generate_robots_txt() returns `rules` alongside `content`, but callers |
| 4187 |
* replace `content` with the body actually being served (a stored override |
| 4188 |
* or a physical file). The generated rules then described something the |
| 4189 |
* response no longer contained. Re-deriving them from the served body keeps |
| 4190 |
* the two halves of the payload describing the same document. |
| 4191 |
* |
| 4192 |
* @since 2.0.1 |
| 4193 |
* |
| 4194 |
* @param string $content Robots.txt body (header optional). |
| 4195 |
* @return array<int, array{directive: string, value: string}> Parsed rules. |
| 4196 |
*/ |
| 4197 |
public function parse_robots_txt_rules(string $content): array { |
| 4198 |
$map = [ |
| 4199 |
'user-agent' => 'user_agent', |
| 4200 |
'disallow' => 'disallow', |
| 4201 |
'allow' => 'allow', |
| 4202 |
'crawl-delay' => 'crawl_delay', |
| 4203 |
'sitemap' => 'sitemap', |
| 4204 |
]; |
| 4205 |
|
| 4206 |
$rules = []; |
| 4207 |
|
| 4208 |
foreach (preg_split('/\r\n|\r|\n/', $this->strip_robots_header($content)) as $line) { |
| 4209 |
$line = trim($line); |
| 4210 |
|
| 4211 |
// Blank lines separate groups and `#` starts a comment; neither is |
| 4212 |
// a rule. |
| 4213 |
if ($line === '' || str_starts_with($line, '#')) { |
| 4214 |
continue; |
| 4215 |
} |
| 4216 |
|
| 4217 |
$parts = explode(':', $line, 2); |
| 4218 |
if (count($parts) !== 2) { |
| 4219 |
continue; |
| 4220 |
} |
| 4221 |
|
| 4222 |
$field = strtolower(trim($parts[0])); |
| 4223 |
if (!isset($map[$field])) { |
| 4224 |
continue; |
| 4225 |
} |
| 4226 |
|
| 4227 |
$rules[] = [ |
| 4228 |
'directive' => $map[$field], |
| 4229 |
// Sitemap values are absolute URLs and contain the `:` the |
| 4230 |
// limited explode above deliberately preserved. |
| 4231 |
'value' => trim($parts[1]), |
| 4232 |
]; |
| 4233 |
} |
| 4234 |
|
| 4235 |
return $rules; |
| 4236 |
} |
| 4237 |
|
| 4238 |
/** |
| 4239 |
* Opening fence of the machine-owned AI crawler region. |
| 4240 |
* |
| 4241 |
* @since 2.5.0 |
| 4242 |
* @var string |
| 4243 |
*/ |
| 4244 |
public const AI_BLOCK_BEGIN = '# BEGIN ThinkRank AI crawlers'; |
| 4245 |
|
| 4246 |
/** |
| 4247 |
* Closing fence of the machine-owned AI crawler region. |
| 4248 |
* |
| 4249 |
* @since 2.5.0 |
| 4250 |
* @var string |
| 4251 |
*/ |
| 4252 |
public const AI_BLOCK_END = '# END ThinkRank AI crawlers'; |
| 4253 |
|
| 4254 |
/** |
| 4255 |
* Render the fenced AI crawler region for the current settings. |
| 4256 |
* |
| 4257 |
* One `User-agent:` / `Disallow: /` record per blocked crawler. Allowed |
| 4258 |
* crawlers emit nothing at all: `Disallow:` with an empty value is the |
| 4259 |
* robots.txt way of saying "allow everything", but writing eighteen such |
| 4260 |
* records to say what silence already says would triple the file and |
| 4261 |
* invite the reading that an unlisted crawler is therefore refused. |
| 4262 |
* |
| 4263 |
* @since 2.5.0 |
| 4264 |
* |
| 4265 |
* @param array $settings Site settings. |
| 4266 |
* @return string Fenced block, newline-terminated, or '' when nothing is blocked. |
| 4267 |
*/ |
| 4268 |
private function build_ai_crawler_block(array $settings): string { |
| 4269 |
$blocked = AI_Crawlers::blocked_slugs($settings['ai_crawler_rules'] ?? []); |
| 4270 |
|
| 4271 |
if (empty($blocked)) { |
| 4272 |
return ''; |
| 4273 |
} |
| 4274 |
|
| 4275 |
$agents = AI_Crawlers::all(); |
| 4276 |
|
| 4277 |
$lines = [ |
| 4278 |
self::AI_BLOCK_BEGIN, |
| 4279 |
'# Managed by ThinkRank — edits between these lines are overwritten.', |
| 4280 |
]; |
| 4281 |
|
| 4282 |
foreach ($blocked as $slug) { |
| 4283 |
$lines[] = ''; |
| 4284 |
$lines[] = 'User-agent: ' . $agents[$slug]['token']; |
| 4285 |
$lines[] = 'Disallow: /'; |
| 4286 |
} |
| 4287 |
|
| 4288 |
$lines[] = self::AI_BLOCK_END; |
| 4289 |
|
| 4290 |
return implode("\n", $lines) . "\n"; |
| 4291 |
} |
| 4292 |
|
| 4293 |
/** |
| 4294 |
* Remove the fenced AI crawler region from a robots.txt body. |
| 4295 |
* |
| 4296 |
* Tolerates a missing closing fence rather than leaving the rest of the |
| 4297 |
* file swallowed: a truncated write, or someone deleting the END line by |
| 4298 |
* hand, would otherwise make every subsequent read drop everything below |
| 4299 |
* the opening fence. |
| 4300 |
* |
| 4301 |
* @since 2.5.0 |
| 4302 |
* |
| 4303 |
* @param string $body Robots.txt body. |
| 4304 |
* @return string Body with the region removed. |
| 4305 |
*/ |
| 4306 |
public function strip_ai_crawler_block(string $body): string { |
| 4307 |
if (false === strpos($body, self::AI_BLOCK_BEGIN)) { |
| 4308 |
return $body; |
| 4309 |
} |
| 4310 |
|
| 4311 |
$pattern = '/\R*' . preg_quote(self::AI_BLOCK_BEGIN, '/') |
| 4312 |
. '.*?(?:' . preg_quote(self::AI_BLOCK_END, '/') . '|\z)\R*/s'; |
| 4313 |
|
| 4314 |
return trim((string) preg_replace($pattern, "\n\n", $body, 1)); |
| 4315 |
} |
| 4316 |
|
| 4317 |
/** |
| 4318 |
* Put the current AI crawler region into a robots.txt body. |
| 4319 |
* |
| 4320 |
* Replaces an existing region in place so the block keeps its position in |
| 4321 |
* a hand-ordered file, and appends when there is none. Everything outside |
| 4322 |
* the fences is returned untouched — that is the whole point of fencing |
| 4323 |
* it, since the body is also a free-text field the user edits. |
| 4324 |
* |
| 4325 |
* @since 2.5.0 |
| 4326 |
* |
| 4327 |
* @param string $body Robots.txt body (fences optional). |
| 4328 |
* @param array $settings Site settings. |
| 4329 |
* @return string Body carrying the current region. |
| 4330 |
*/ |
| 4331 |
private function apply_ai_crawler_block(string $body, array $settings): string { |
| 4332 |
$stripped = $this->strip_ai_crawler_block($body); |
| 4333 |
$block = $this->build_ai_crawler_block($settings); |
| 4334 |
|
| 4335 |
if ('' === $block) { |
| 4336 |
return $stripped; |
| 4337 |
} |
| 4338 |
|
| 4339 |
if ('' === trim($stripped)) { |
| 4340 |
return trim($block); |
| 4341 |
} |
| 4342 |
|
| 4343 |
return rtrim($stripped) . "\n\n" . trim($block); |
| 4344 |
} |
| 4345 |
|
| 4346 |
/** |
| 4347 |
* The auto-generated header prepended to the served robots.txt. |
| 4348 |
* |
| 4349 |
* Kept separate from the body so it is only ever added at render time with |
| 4350 |
* a fresh timestamp, never stored or shown in the editable textarea. |
| 4351 |
* |
| 4352 |
* @return string |
| 4353 |
*/ |
| 4354 |
private function robots_txt_header(): string { |
| 4355 |
return "# Robots.txt generated by ThinkRank SEO\n" |
| 4356 |
. "# " . gmdate('Y-m-d H:i:s') . " UTC\n\n"; |
| 4357 |
} |
| 4358 |
|
| 4359 |
/** |
| 4360 |
* Strip our auto-generated header from a robots.txt string. |
| 4361 |
* |
| 4362 |
* Used when surfacing existing content for editing so the header/timestamp |
| 4363 |
* doesn't round-trip back into storage. |
| 4364 |
* |
| 4365 |
* @param string $content Raw robots.txt content. |
| 4366 |
* @return string Body without the ThinkRank header. |
| 4367 |
*/ |
| 4368 |
private function strip_robots_header(string $content): string { |
| 4369 |
$pattern = '/^# Robots\.txt generated by ThinkRank SEO\r?\n# [^\r\n]* UTC\r?\n\r?\n/'; |
| 4370 |
return trim((string) preg_replace($pattern, '', $content, 1)); |
| 4371 |
} |
| 4372 |
|
| 4373 |
/** |
| 4374 |
* The body that should populate the editor for the current site. |
| 4375 |
* |
| 4376 |
* Prefers what is actually being served: the physical file if one exists |
| 4377 |
* (header stripped), otherwise the effective body. This is what the admin |
| 4378 |
* screen shows so the textarea is never blank while /robots.txt has content. |
| 4379 |
* |
| 4380 |
* @return string |
| 4381 |
*/ |
| 4382 |
public function get_served_robots_body(): string { |
| 4383 |
$settings = $this->get_settings('site'); |
| 4384 |
|
| 4385 |
// The AI block is stripped from every one of these paths. A physical |
| 4386 |
// robots.txt we wrote carries it, and the stored override is whatever |
| 4387 |
// the textarea last held — so without this the block round-trips into |
| 4388 |
// the editor, gets saved as ordinary body text, and is then appended |
| 4389 |
// to a second time on the next render. |
| 4390 |
$custom = trim((string) ($settings['robots_txt_content'] ?? '')); |
| 4391 |
if ($custom !== '') { |
| 4392 |
return $this->strip_ai_crawler_block($this->strip_robots_header($custom)); |
| 4393 |
} |
| 4394 |
|
| 4395 |
$robots_file = ABSPATH . 'robots.txt'; |
| 4396 |
if (file_exists($robots_file)) { |
| 4397 |
// phpcs:ignore WordPress.WP.AlternativeFunctions.file_get_contents_file_get_contents, WordPress.PHP.NoSilencedErrors.Discouraged -- an unreadable robots.txt is an expected state answered with an empty string. |
| 4398 |
$raw = (string) @file_get_contents($robots_file); |
| 4399 |
if ($raw !== '') { |
| 4400 |
return $this->strip_ai_crawler_block($this->strip_robots_header($raw)); |
| 4401 |
} |
| 4402 |
} |
| 4403 |
|
| 4404 |
return $this->strip_ai_crawler_block(trim($this->generate_robots_txt()['content'])); |
| 4405 |
} |
| 4406 |
private function get_site_identity_data(array $settings): array { |
| 4407 |
return [ |
| 4408 |
'site_name' => $settings['site_name'] ?? get_bloginfo('name'), |
| 4409 |
'site_description' => $settings['site_description'] ?? get_bloginfo('description'), |
| 4410 |
'tagline' => $settings['tagline'] ?? get_bloginfo('description'), |
| 4411 |
'logo_url' => $settings['logo_url'] ?? '', |
| 4412 |
'favicon_url' => $settings['favicon_url'] ?? '', |
| 4413 |
'apple_touch_icon_url' => $settings['apple_touch_icon_url'] ?? '' |
| 4414 |
]; |
| 4415 |
} |
| 4416 |
|
| 4417 |
/** |
| 4418 |
* Optimize individual identity element |
| 4419 |
* |
| 4420 |
* @since 1.0.0 |
| 4421 |
* |
| 4422 |
* @param string $element Element name |
| 4423 |
* @param mixed $value Element value |
| 4424 |
* @param array $config Element configuration |
| 4425 |
* @return array Optimization results |
| 4426 |
*/ |
| 4427 |
private function optimize_identity_element(string $element, $value, array $config): array { |
| 4428 |
$optimization = [ |
| 4429 |
'optimized_value' => $value, |
| 4430 |
'validation' => [ |
| 4431 |
'valid' => true, |
| 4432 |
'errors' => [], |
| 4433 |
'warnings' => [] |
| 4434 |
], |
| 4435 |
'suggestions' => [] |
| 4436 |
]; |
| 4437 |
|
| 4438 |
switch ($config['type']) { |
| 4439 |
case 'text': |
| 4440 |
$optimization = $this->optimize_text_element($element, $value, $config, $optimization); |
| 4441 |
break; |
| 4442 |
case 'image': |
| 4443 |
$optimization = $this->optimize_image_element($element, $value, $config, $optimization); |
| 4444 |
break; |
| 4445 |
} |
| 4446 |
|
| 4447 |
return $optimization; |
| 4448 |
} |
| 4449 |
|
| 4450 |
/** |
| 4451 |
* Optimize text identity element |
| 4452 |
* |
| 4453 |
* @since 1.0.0 |
| 4454 |
* |
| 4455 |
* @param string $element Element name |
| 4456 |
* @param string $value Element value |
| 4457 |
* @param array $config Element configuration |
| 4458 |
* @param array $optimization Current optimization |
| 4459 |
* @return array Updated optimization |
| 4460 |
*/ |
| 4461 |
private function optimize_text_element(string $element, string $value, array $config, array $optimization): array { |
| 4462 |
if (empty($value) && !empty($config['required'])) { |
| 4463 |
$optimization['validation']['errors'][] = "{$element} is required"; |
| 4464 |
$optimization['validation']['valid'] = false; |
| 4465 |
} |
| 4466 |
|
| 4467 |
if (!empty($value) && isset($config['max_length'])) { |
| 4468 |
// The warning says "characters", so measure and cut in characters: |
| 4469 |
// strlen()/substr() fired early on non-Latin values and the |
| 4470 |
// suggested replacement was cut mid-character (#687). |
| 4471 |
if (mb_strlen($value) > $config['max_length']) { |
| 4472 |
$optimization['validation']['warnings'][] = "{$element} exceeds maximum length of {$config['max_length']} characters"; |
| 4473 |
$optimization['optimized_value'] = \ThinkRank\Core\Seo_Text::trim_to_length($value, (int) $config['max_length']); |
| 4474 |
} |
| 4475 |
} |
| 4476 |
|
| 4477 |
// SEO-specific optimizations |
| 4478 |
if ($element === 'site_name' && !empty($value)) { |
| 4479 |
// Remove excessive punctuation |
| 4480 |
$optimization['optimized_value'] = preg_replace('/[!@#$%^&*()]+/', '', $value); |
| 4481 |
} |
| 4482 |
|
| 4483 |
return $optimization; |
| 4484 |
} |
| 4485 |
|
| 4486 |
/** |
| 4487 |
* Optimize image identity element |
| 4488 |
* |
| 4489 |
* @since 1.0.0 |
| 4490 |
* |
| 4491 |
* @param string $element Element name |
| 4492 |
* @param string $value Element value |
| 4493 |
* @param array $config Element configuration |
| 4494 |
* @param array $optimization Current optimization |
| 4495 |
* @return array Updated optimization |
| 4496 |
*/ |
| 4497 |
private function optimize_image_element(string $element, string $value, array $config, array $optimization): array { |
| 4498 |
if (empty($value)) { |
| 4499 |
if (!empty($config['required'])) { |
| 4500 |
$optimization['validation']['errors'][] = "{$element} is required"; |
| 4501 |
$optimization['validation']['valid'] = false; |
| 4502 |
} |
| 4503 |
return $optimization; |
| 4504 |
} |
| 4505 |
|
| 4506 |
// Validate URL |
| 4507 |
if (!filter_var($value, FILTER_VALIDATE_URL)) { |
| 4508 |
$optimization['validation']['errors'][] = "{$element} must be a valid URL"; |
| 4509 |
$optimization['validation']['valid'] = false; |
| 4510 |
return $optimization; |
| 4511 |
} |
| 4512 |
|
| 4513 |
// Check if it's a local image |
| 4514 |
$attachment_id = Attachment_Lookup::id_from_url($value); |
| 4515 |
if ($attachment_id) { |
| 4516 |
$image_meta = wp_get_attachment_metadata($attachment_id); |
| 4517 |
|
| 4518 |
if ($image_meta && isset($image_meta['width'], $image_meta['height'])) { |
| 4519 |
// Check recommended size, against the configured file itself |
| 4520 |
// rather than the upload it may have been generated from. |
| 4521 |
if (isset($config['recommended_size'])) { |
| 4522 |
[$rec_width, $rec_height] = explode('x', $config['recommended_size']); |
| 4523 |
$image_file = Attachment_Lookup::describe($attachment_id, $value); |
| 4524 |
|
| 4525 |
if ($image_file['width'] !== (int) $rec_width || $image_file['height'] !== (int) $rec_height) { |
| 4526 |
$optimization['suggestions'][] = "Consider using {$config['recommended_size']} size for optimal {$element}"; |
| 4527 |
} |
| 4528 |
} |
| 4529 |
|
| 4530 |
// Check file size |
| 4531 |
if (isset($config['max_size'])) { |
| 4532 |
$file_path = get_attached_file($attachment_id); |
| 4533 |
if ($file_path && file_exists($file_path)) { |
| 4534 |
$file_size = filesize($file_path); |
| 4535 |
$max_size_bytes = $this->parse_size_string($config['max_size']); |
| 4536 |
|
| 4537 |
if ($file_size > $max_size_bytes) { |
| 4538 |
$optimization['validation']['warnings'][] = "{$element} file size exceeds {$config['max_size']}"; |
| 4539 |
} |
| 4540 |
} |
| 4541 |
} |
| 4542 |
} |
| 4543 |
} |
| 4544 |
|
| 4545 |
return $optimization; |
| 4546 |
} |
| 4547 |
|
| 4548 |
/** |
| 4549 |
* Parse size string to bytes |
| 4550 |
* |
| 4551 |
* @since 1.0.0 |
| 4552 |
* |
| 4553 |
* @param string $size_string Size string (e.g., '2MB', '500KB') |
| 4554 |
* @return int Size in bytes |
| 4555 |
*/ |
| 4556 |
private function parse_size_string(string $size_string): int { |
| 4557 |
$size_string = strtoupper(trim($size_string)); |
| 4558 |
$size = (int) $size_string; |
| 4559 |
|
| 4560 |
if (strpos($size_string, 'KB') !== false) { |
| 4561 |
return $size * 1024; |
| 4562 |
} elseif (strpos($size_string, 'MB') !== false) { |
| 4563 |
return $size * 1024 * 1024; |
| 4564 |
} elseif (strpos($size_string, 'GB') !== false) { |
| 4565 |
return $size * 1024 * 1024 * 1024; |
| 4566 |
} |
| 4567 |
|
| 4568 |
return $size; |
| 4569 |
} |
| 4570 |
|
| 4571 |
/** |
| 4572 |
* Calculate identity optimization score |
| 4573 |
* |
| 4574 |
* @since 1.0.0 |
| 4575 |
* |
| 4576 |
* @param array $validations Element validations |
| 4577 |
* @return int Score (0-100) |
| 4578 |
*/ |
| 4579 |
private function calculate_identity_score(array $validations): int { |
| 4580 |
$total_score = 0; |
| 4581 |
$element_count = 0; |
| 4582 |
|
| 4583 |
foreach ($validations as $validation) { |
| 4584 |
$element_score = 100; |
| 4585 |
$element_score -= count($validation['errors']) * 30; |
| 4586 |
$element_score -= count($validation['warnings']) * 15; |
| 4587 |
|
| 4588 |
$total_score += max(0, $element_score); |
| 4589 |
$element_count++; |
| 4590 |
} |
| 4591 |
|
| 4592 |
return $element_count > 0 ? (int) round($total_score / $element_count) : 0; |
| 4593 |
} |
| 4594 |
|
| 4595 |
/** |
| 4596 |
* Calculate validation score |
| 4597 |
* |
| 4598 |
* @since 1.0.0 |
| 4599 |
* |
| 4600 |
* @param array $validation Validation results |
| 4601 |
* @return int Score (0-100) |
| 4602 |
*/ |
| 4603 |
private function calculate_validation_score(array $validation): int { |
| 4604 |
$score = 100; |
| 4605 |
$score -= count($validation['errors']) * 20; |
| 4606 |
$score -= count($validation['warnings']) * 10; |
| 4607 |
$score -= count($validation['suggestions']) * 5; |
| 4608 |
|
| 4609 |
return max(0, $score); |
| 4610 |
} |
| 4611 |
} |
| 4612 |
|
