| 1 |
<?php |
| 2 |
/** |
| 3 |
* Robots Meta Manager Class |
| 4 |
* |
| 5 |
* Universal robots meta tag generation and validation for search engine optimization. |
| 6 |
* Implements 2025 search engine guidelines with real robots meta specifications, |
| 7 |
* canonical URL management, and advanced robots directives handling. |
| 8 |
* |
| 9 |
* @package ThinkRank |
| 10 |
* @subpackage SEO |
| 11 |
* @since 1.0.0 |
| 12 |
*/ |
| 13 |
|
| 14 |
declare(strict_types=1); |
| 15 |
|
| 16 |
namespace ThinkRank\SEO; |
| 17 |
|
| 18 |
/** |
| 19 |
* Robots Meta Manager Class |
| 20 |
* |
| 21 |
* Generates and validates robots meta tags for search engine optimization. |
| 22 |
* Provides context-aware robots meta generation with canonical URL management |
| 23 |
* and advanced robots directives handling. |
| 24 |
* |
| 25 |
* @since 1.0.0 |
| 26 |
*/ |
| 27 |
class Robots_Meta_Manager extends Abstract_SEO_Manager { |
| 28 |
|
| 29 |
/** |
| 30 |
* Supported robots directives with their specifications |
| 31 |
* |
| 32 |
* @since 1.0.0 |
| 33 |
* @var array |
| 34 |
*/ |
| 35 |
private array $robots_directives = [ |
| 36 |
// Index directives |
| 37 |
'index' => [ |
| 38 |
'type' => 'index', |
| 39 |
'opposite' => 'noindex', |
| 40 |
'description' => 'Allow search engines to index this page', |
| 41 |
'default' => true, |
| 42 |
'search_engines' => ['google', 'bing', 'yahoo', 'yandex', 'baidu'] |
| 43 |
], |
| 44 |
'noindex' => [ |
| 45 |
'type' => 'index', |
| 46 |
'opposite' => 'index', |
| 47 |
'description' => 'Prevent search engines from indexing this page', |
| 48 |
'default' => false, |
| 49 |
'search_engines' => ['google', 'bing', 'yahoo', 'yandex', 'baidu'] |
| 50 |
], |
| 51 |
|
| 52 |
// Follow directives |
| 53 |
'follow' => [ |
| 54 |
'type' => 'follow', |
| 55 |
'opposite' => 'nofollow', |
| 56 |
'description' => 'Allow search engines to follow links on this page', |
| 57 |
'default' => true, |
| 58 |
'search_engines' => ['google', 'bing', 'yahoo', 'yandex', 'baidu'] |
| 59 |
], |
| 60 |
'nofollow' => [ |
| 61 |
'type' => 'follow', |
| 62 |
'opposite' => 'follow', |
| 63 |
'description' => 'Prevent search engines from following links on this page', |
| 64 |
'default' => false, |
| 65 |
'search_engines' => ['google', 'bing', 'yahoo', 'yandex', 'baidu'] |
| 66 |
], |
| 67 |
|
| 68 |
// Archive directives |
| 69 |
'archive' => [ |
| 70 |
'type' => 'archive', |
| 71 |
'opposite' => 'noarchive', |
| 72 |
'description' => 'Allow search engines to show cached versions of this page', |
| 73 |
'default' => true, |
| 74 |
'search_engines' => ['google', 'bing', 'yahoo'] |
| 75 |
], |
| 76 |
'noarchive' => [ |
| 77 |
'type' => 'archive', |
| 78 |
'opposite' => 'archive', |
| 79 |
'description' => 'Prevent search engines from showing cached versions of this page', |
| 80 |
'default' => false, |
| 81 |
'search_engines' => ['google', 'bing', 'yahoo'] |
| 82 |
], |
| 83 |
|
| 84 |
// Snippet directives |
| 85 |
'snippet' => [ |
| 86 |
'type' => 'snippet', |
| 87 |
'opposite' => 'nosnippet', |
| 88 |
'description' => 'Allow search engines to show text snippets of this page', |
| 89 |
'default' => true, |
| 90 |
'search_engines' => ['google', 'bing'] |
| 91 |
], |
| 92 |
'nosnippet' => [ |
| 93 |
'type' => 'snippet', |
| 94 |
'opposite' => 'snippet', |
| 95 |
'description' => 'Prevent search engines from showing text snippets of this page', |
| 96 |
'default' => false, |
| 97 |
'search_engines' => ['google', 'bing'] |
| 98 |
], |
| 99 |
|
| 100 |
// Image directives |
| 101 |
'noimageindex' => [ |
| 102 |
'type' => 'image', |
| 103 |
'opposite' => null, |
| 104 |
'description' => 'Prevent search engines from indexing images on this page', |
| 105 |
'default' => false, |
| 106 |
'search_engines' => ['google'] |
| 107 |
], |
| 108 |
|
| 109 |
// Translation directives |
| 110 |
'notranslate' => [ |
| 111 |
'type' => 'translate', |
| 112 |
'opposite' => null, |
| 113 |
'description' => 'Prevent search engines from offering translation of this page', |
| 114 |
'default' => false, |
| 115 |
'search_engines' => ['google'] |
| 116 |
], |
| 117 |
|
| 118 |
// Video directives |
| 119 |
'novideoindex' => [ |
| 120 |
'type' => 'video', |
| 121 |
'opposite' => null, |
| 122 |
'description' => 'Prevent search engines from indexing videos on this page', |
| 123 |
'default' => false, |
| 124 |
'search_engines' => ['google'] |
| 125 |
] |
| 126 |
]; |
| 127 |
|
| 128 |
/** |
| 129 |
* Advanced robots directives with parameters |
| 130 |
* |
| 131 |
* @since 1.0.0 |
| 132 |
* @var array |
| 133 |
*/ |
| 134 |
private array $advanced_directives = [ |
| 135 |
'max-snippet' => [ |
| 136 |
'type' => 'parameterized', |
| 137 |
'description' => 'Maximum number of characters to show in text snippets', |
| 138 |
'default_value' => -1, // -1 means no limit |
| 139 |
'min_value' => -1, |
| 140 |
'max_value' => 320, |
| 141 |
'search_engines' => ['google'] |
| 142 |
], |
| 143 |
'max-image-preview' => [ |
| 144 |
'type' => 'parameterized', |
| 145 |
'description' => 'Maximum size of image preview', |
| 146 |
'default_value' => 'large', |
| 147 |
'allowed_values' => ['none', 'standard', 'large'], |
| 148 |
'search_engines' => ['google'] |
| 149 |
], |
| 150 |
'max-video-preview' => [ |
| 151 |
'type' => 'parameterized', |
| 152 |
'description' => 'Maximum duration of video preview in seconds', |
| 153 |
'default_value' => -1, // -1 means no limit |
| 154 |
'min_value' => -1, |
| 155 |
'max_value' => 3600, // 1 hour |
| 156 |
'search_engines' => ['google'] |
| 157 |
], |
| 158 |
'unavailable_after' => [ |
| 159 |
'type' => 'date', |
| 160 |
'description' => 'Date after which the page should not be indexed', |
| 161 |
'format' => 'RFC 850', |
| 162 |
'search_engines' => ['google'] |
| 163 |
] |
| 164 |
]; |
| 165 |
|
| 166 |
/** |
| 167 |
* Search engine specific robots directives |
| 168 |
* |
| 169 |
* @since 1.0.0 |
| 170 |
* @var array |
| 171 |
*/ |
| 172 |
private array $search_engine_specific = [ |
| 173 |
'google' => ['googlebot'], |
| 174 |
'bing' => ['bingbot', 'msnbot'], |
| 175 |
'yahoo' => ['slurp'], |
| 176 |
'yandex' => ['yandexbot'], |
| 177 |
'baidu' => ['baiduspider'], |
| 178 |
'duckduckgo' => ['duckduckbot'] |
| 179 |
]; |
| 180 |
|
| 181 |
/** |
| 182 |
* Constructor |
| 183 |
* |
| 184 |
* @since 1.0.0 |
| 185 |
*/ |
| 186 |
public function __construct() { |
| 187 |
parent::__construct('robots_meta'); |
| 188 |
} |
| 189 |
|
| 190 |
/** |
| 191 |
* Generate robots meta tags for content |
| 192 |
* |
| 193 |
* @since 1.0.0 |
| 194 |
* |
| 195 |
* @param array $data Content data for robots generation |
| 196 |
* @param string $context Context type ('site', 'post', 'page', etc.) |
| 197 |
* @return array Generated robots meta tags |
| 198 |
*/ |
| 199 |
public function generate_robots_meta(array $data, string $context = 'site'): array { |
| 200 |
$robots_meta = []; |
| 201 |
|
| 202 |
// Get robots settings for context |
| 203 |
$settings = $this->get_settings($context, $data['context_id'] ?? null); |
| 204 |
|
| 205 |
// Generate main robots directive |
| 206 |
$main_directive = $this->build_main_robots_directive($settings, $context, $data); |
| 207 |
if (!empty($main_directive)) { |
| 208 |
$robots_meta['robots'] = $main_directive; |
| 209 |
} |
| 210 |
|
| 211 |
// Generate search engine specific directives |
| 212 |
$specific_directives = $this->build_search_engine_specific_directives($settings, $context, $data); |
| 213 |
$robots_meta = array_merge($robots_meta, $specific_directives); |
| 214 |
|
| 215 |
// Generate canonical URL |
| 216 |
$canonical_url = $this->generate_canonical_url($data, $context, $settings); |
| 217 |
if (!empty($canonical_url)) { |
| 218 |
$robots_meta['canonical'] = $canonical_url; |
| 219 |
} |
| 220 |
|
| 221 |
// Add other meta directives |
| 222 |
$other_meta = $this->generate_other_meta_directives($settings, $context, $data); |
| 223 |
$robots_meta = array_merge($robots_meta, $other_meta); |
| 224 |
|
| 225 |
return $robots_meta; |
| 226 |
} |
| 227 |
|
| 228 |
/** |
| 229 |
* Validate robots meta settings |
| 230 |
* |
| 231 |
* @since 1.0.0 |
| 232 |
* |
| 233 |
* @param array $robots_meta Robots meta to validate |
| 234 |
* @return array Validation results with errors, warnings, and suggestions |
| 235 |
*/ |
| 236 |
public function validate_robots_meta(array $robots_meta): array { |
| 237 |
$validation = [ |
| 238 |
'valid' => true, |
| 239 |
'errors' => [], |
| 240 |
'warnings' => [], |
| 241 |
'suggestions' => [], |
| 242 |
'score' => 100 |
| 243 |
]; |
| 244 |
|
| 245 |
// Validate main robots directive |
| 246 |
if (isset($robots_meta['robots'])) { |
| 247 |
$robots_validation = $this->validate_robots_directive($robots_meta['robots']); |
| 248 |
$validation = array_merge_recursive($validation, $robots_validation); |
| 249 |
} |
| 250 |
|
| 251 |
// Validate canonical URL |
| 252 |
if (isset($robots_meta['canonical'])) { |
| 253 |
$canonical_validation = $this->validate_canonical_url($robots_meta['canonical']); |
| 254 |
$validation = array_merge_recursive($validation, $canonical_validation); |
| 255 |
} |
| 256 |
|
| 257 |
// Validate advanced directives |
| 258 |
foreach ($this->advanced_directives as $directive => $spec) { |
| 259 |
if (isset($robots_meta[$directive])) { |
| 260 |
$directive_validation = $this->validate_advanced_directive($directive, $robots_meta[$directive], $spec); |
| 261 |
$validation = array_merge_recursive($validation, $directive_validation); |
| 262 |
} |
| 263 |
} |
| 264 |
|
| 265 |
// Check for conflicting directives |
| 266 |
$conflict_validation = $this->check_directive_conflicts($robots_meta); |
| 267 |
$validation = array_merge_recursive($validation, $conflict_validation); |
| 268 |
|
| 269 |
// Calculate overall score |
| 270 |
$validation['score'] = $this->calculate_robots_score($validation); |
| 271 |
|
| 272 |
// Add SEO suggestions |
| 273 |
$validation = $this->add_robots_seo_suggestions($robots_meta, $validation); |
| 274 |
|
| 275 |
return $validation; |
| 276 |
} |
| 277 |
|
| 278 |
/** |
| 279 |
* Generate canonical URL for content |
| 280 |
* |
| 281 |
* @since 1.0.0 |
| 282 |
* |
| 283 |
* @param array $data Content data |
| 284 |
* @param string $context Context type |
| 285 |
* @param array $settings Robots settings |
| 286 |
* @return string Canonical URL |
| 287 |
*/ |
| 288 |
public function generate_canonical_url(array $data, string $context, array $settings = []): string { |
| 289 |
// Check if custom canonical is set |
| 290 |
if (!empty($settings['custom_canonical'])) { |
| 291 |
return $this->validate_and_clean_url($settings['custom_canonical']); |
| 292 |
} |
| 293 |
|
| 294 |
// Auto-generate canonical based on context |
| 295 |
switch ($context) { |
| 296 |
case 'site': |
| 297 |
return home_url(); |
| 298 |
case 'post': |
| 299 |
case 'page': |
| 300 |
case 'product': |
| 301 |
if (!empty($data['url'])) { |
| 302 |
return $this->clean_canonical_url($data['url']); |
| 303 |
} |
| 304 |
break; |
| 305 |
} |
| 306 |
|
| 307 |
// Fallback to current URL |
| 308 |
return $this->get_current_canonical_url(); |
| 309 |
} |
| 310 |
|
| 311 |
/** |
| 312 |
* Check for duplicate content and suggest canonical URLs |
| 313 |
* |
| 314 |
* @since 1.0.0 |
| 315 |
* |
| 316 |
* @param string $url URL to check |
| 317 |
* @param string $context Context type |
| 318 |
* @return array Duplicate content analysis |
| 319 |
*/ |
| 320 |
public function analyze_duplicate_content(string $url, string $context): array { |
| 321 |
$analysis = [ |
| 322 |
'has_duplicates' => false, |
| 323 |
'duplicate_urls' => [], |
| 324 |
'suggested_canonical' => $url, |
| 325 |
'recommendations' => [] |
| 326 |
]; |
| 327 |
|
| 328 |
// Check for common duplicate patterns |
| 329 |
$duplicate_patterns = $this->detect_duplicate_patterns($url); |
| 330 |
if (!empty($duplicate_patterns)) { |
| 331 |
$analysis['has_duplicates'] = true; |
| 332 |
$analysis['duplicate_urls'] = $duplicate_patterns; |
| 333 |
$analysis['suggested_canonical'] = $this->suggest_canonical_from_duplicates($duplicate_patterns); |
| 334 |
$analysis['recommendations'] = $this->get_duplicate_content_recommendations($duplicate_patterns); |
| 335 |
} |
| 336 |
|
| 337 |
return $analysis; |
| 338 |
} |
| 339 |
|
| 340 |
/** |
| 341 |
* Validate SEO settings (implements interface) |
| 342 |
* |
| 343 |
* @since 1.0.0 |
| 344 |
* |
| 345 |
* @param array $settings Settings array to validate |
| 346 |
* @return array Validation results |
| 347 |
*/ |
| 348 |
public function validate_settings(array $settings): array { |
| 349 |
$validation = [ |
| 350 |
'valid' => true, |
| 351 |
'errors' => [], |
| 352 |
'warnings' => [], |
| 353 |
'suggestions' => [] |
| 354 |
]; |
| 355 |
|
| 356 |
// Validate robots index setting |
| 357 |
if (isset($settings['robots_index'])) { |
| 358 |
if (!in_array($settings['robots_index'], ['index', 'noindex'], true)) { |
| 359 |
$validation['errors'][] = 'robots_index must be either "index" or "noindex"'; |
| 360 |
$validation['valid'] = false; |
| 361 |
} |
| 362 |
} |
| 363 |
|
| 364 |
// Validate robots follow setting |
| 365 |
if (isset($settings['robots_follow'])) { |
| 366 |
if (!in_array($settings['robots_follow'], ['follow', 'nofollow'], true)) { |
| 367 |
$validation['errors'][] = 'robots_follow must be either "follow" or "nofollow"'; |
| 368 |
$validation['valid'] = false; |
| 369 |
} |
| 370 |
} |
| 371 |
|
| 372 |
// Validate advanced directives |
| 373 |
foreach ($this->advanced_directives as $directive => $spec) { |
| 374 |
if (isset($settings[$directive])) { |
| 375 |
$directive_validation = $this->validate_advanced_directive($directive, $settings[$directive], $spec); |
| 376 |
if (!$directive_validation['valid']) { |
| 377 |
$validation['errors'] = array_merge($validation['errors'], $directive_validation['errors']); |
| 378 |
$validation['valid'] = false; |
| 379 |
} |
| 380 |
$validation['warnings'] = array_merge($validation['warnings'], $directive_validation['warnings']); |
| 381 |
} |
| 382 |
} |
| 383 |
|
| 384 |
// Validate canonical URL |
| 385 |
if (isset($settings['custom_canonical']) && !empty($settings['custom_canonical'])) { |
| 386 |
if (!filter_var($settings['custom_canonical'], FILTER_VALIDATE_URL)) { |
| 387 |
$validation['errors'][] = 'custom_canonical must be a valid URL'; |
| 388 |
$validation['valid'] = false; |
| 389 |
} |
| 390 |
} |
| 391 |
|
| 392 |
// Check for conflicting settings |
| 393 |
if (isset($settings['robots_index'], $settings['robots_follow'])) { |
| 394 |
if ($settings['robots_index'] === 'noindex' && $settings['robots_follow'] === 'follow') { |
| 395 |
$validation['warnings'][] = 'Using "noindex, follow" - consider using "noindex, nofollow" for better crawl budget optimization'; |
| 396 |
} |
| 397 |
} |
| 398 |
|
| 399 |
return $validation; |
| 400 |
} |
| 401 |
|
| 402 |
/** |
| 403 |
* Get output data for frontend rendering (implements interface) |
| 404 |
* |
| 405 |
* @since 1.0.0 |
| 406 |
* |
| 407 |
* @param string $context_type The context type |
| 408 |
* @param int|null $context_id Optional. Context ID |
| 409 |
* @return array Output data ready for frontend rendering |
| 410 |
*/ |
| 411 |
public function get_output_data(string $context_type, ?int $context_id): array { |
| 412 |
$settings = $this->get_settings($context_type, $context_id); |
| 413 |
$output = [ |
| 414 |
'robots_meta' => [], |
| 415 |
'meta_tags' => [], |
| 416 |
'enabled' => false |
| 417 |
]; |
| 418 |
|
| 419 |
// Check if robots meta is enabled |
| 420 |
if (empty($settings['robots_enabled'])) { |
| 421 |
return $output; |
| 422 |
} |
| 423 |
|
| 424 |
$output['enabled'] = true; |
| 425 |
|
| 426 |
// Extract content data |
| 427 |
$content_data = $this->extract_robots_content_data($context_type, $context_id); |
| 428 |
|
| 429 |
// Generate robots meta |
| 430 |
$output['robots_meta'] = $this->generate_robots_meta($content_data, $context_type); |
| 431 |
|
| 432 |
// Convert to HTML meta tags |
| 433 |
$output['meta_tags'] = $this->convert_robots_to_meta_tags($output['robots_meta']); |
| 434 |
|
| 435 |
return $output; |
| 436 |
} |
| 437 |
|
| 438 |
/** |
| 439 |
* Get default settings for a context type (implements interface) |
| 440 |
* |
| 441 |
* @since 1.0.0 |
| 442 |
* |
| 443 |
* @param string $context_type The context type to get defaults for |
| 444 |
* @return array Default settings array |
| 445 |
*/ |
| 446 |
public function get_default_settings(string $context_type): array { |
| 447 |
$defaults = [ |
| 448 |
'robots_enabled' => true, |
| 449 |
'robots_index' => 'index', |
| 450 |
'robots_follow' => 'follow', |
| 451 |
'robots_archive' => 'archive', |
| 452 |
'robots_snippet' => 'snippet', |
| 453 |
'max_snippet' => -1, |
| 454 |
'max_image_preview' => 'large', |
| 455 |
'max_video_preview' => -1, |
| 456 |
'custom_canonical' => '', |
| 457 |
'auto_canonical' => true, |
| 458 |
'noindex_search' => true, |
| 459 |
'noindex_archives' => false |
| 460 |
]; |
| 461 |
|
| 462 |
// Context-specific defaults |
| 463 |
switch ($context_type) { |
| 464 |
case 'site': |
| 465 |
// Site-wide defaults are already set above |
| 466 |
break; |
| 467 |
case 'post': |
| 468 |
// Posts should generally be indexed |
| 469 |
$defaults['robots_index'] = 'index'; |
| 470 |
break; |
| 471 |
case 'page': |
| 472 |
// Pages should generally be indexed |
| 473 |
$defaults['robots_index'] = 'index'; |
| 474 |
break; |
| 475 |
case 'product': |
| 476 |
// Products should definitely be indexed |
| 477 |
$defaults['robots_index'] = 'index'; |
| 478 |
$defaults['max_image_preview'] = 'large'; // Important for product images |
| 479 |
break; |
| 480 |
} |
| 481 |
|
| 482 |
return $defaults; |
| 483 |
} |
| 484 |
|
| 485 |
/** |
| 486 |
* Get settings schema definition (implements interface) |
| 487 |
* |
| 488 |
* @since 1.0.0 |
| 489 |
* |
| 490 |
* @param string $context_type The context type to get schema for |
| 491 |
* @return array Settings schema definition |
| 492 |
*/ |
| 493 |
public function get_settings_schema(string $context_type): array { |
| 494 |
return [ |
| 495 |
'robots_enabled' => [ |
| 496 |
'type' => 'boolean', |
| 497 |
'title' => 'Enable Robots Meta', |
| 498 |
'description' => 'Generate robots meta tags for search engine crawling control', |
| 499 |
'default' => true |
| 500 |
], |
| 501 |
'robots_index' => [ |
| 502 |
'type' => 'string', |
| 503 |
'title' => 'Index Directive', |
| 504 |
'description' => 'Control whether search engines should index this content', |
| 505 |
'enum' => ['index', 'noindex'], |
| 506 |
'default' => 'index' |
| 507 |
], |
| 508 |
'robots_follow' => [ |
| 509 |
'type' => 'string', |
| 510 |
'title' => 'Follow Directive', |
| 511 |
'description' => 'Control whether search engines should follow links on this content', |
| 512 |
'enum' => ['follow', 'nofollow'], |
| 513 |
'default' => 'follow' |
| 514 |
], |
| 515 |
'robots_archive' => [ |
| 516 |
'type' => 'string', |
| 517 |
'title' => 'Archive Directive', |
| 518 |
'description' => 'Control whether search engines should show cached versions', |
| 519 |
'enum' => ['archive', 'noarchive'], |
| 520 |
'default' => 'archive' |
| 521 |
], |
| 522 |
'robots_snippet' => [ |
| 523 |
'type' => 'string', |
| 524 |
'title' => 'Snippet Directive', |
| 525 |
'description' => 'Control whether search engines should show text snippets', |
| 526 |
'enum' => ['snippet', 'nosnippet'], |
| 527 |
'default' => 'snippet' |
| 528 |
], |
| 529 |
'max_snippet' => [ |
| 530 |
'type' => 'integer', |
| 531 |
'title' => 'Max Snippet Length', |
| 532 |
'description' => 'Maximum number of characters in text snippets (-1 for no limit)', |
| 533 |
'minimum' => -1, |
| 534 |
'maximum' => 320, |
| 535 |
'default' => -1 |
| 536 |
], |
| 537 |
'max_image_preview' => [ |
| 538 |
'type' => 'string', |
| 539 |
'title' => 'Max Image Preview', |
| 540 |
'description' => 'Maximum size of image preview in search results', |
| 541 |
'enum' => ['none', 'standard', 'large'], |
| 542 |
'default' => 'large' |
| 543 |
], |
| 544 |
'max_video_preview' => [ |
| 545 |
'type' => 'integer', |
| 546 |
'title' => 'Max Video Preview', |
| 547 |
'description' => 'Maximum duration of video preview in seconds (-1 for no limit)', |
| 548 |
'minimum' => -1, |
| 549 |
'maximum' => 3600, |
| 550 |
'default' => -1 |
| 551 |
], |
| 552 |
'custom_canonical' => [ |
| 553 |
'type' => 'string', |
| 554 |
'title' => 'Custom Canonical URL', |
| 555 |
'description' => 'Custom canonical URL for this content (overrides auto-generated)', |
| 556 |
'format' => 'uri', |
| 557 |
'default' => '' |
| 558 |
], |
| 559 |
'auto_canonical' => [ |
| 560 |
'type' => 'boolean', |
| 561 |
'title' => 'Auto-generate Canonical', |
| 562 |
'description' => 'Automatically generate canonical URLs', |
| 563 |
'default' => true |
| 564 |
], |
| 565 |
'noindex_search' => [ |
| 566 |
'type' => 'boolean', |
| 567 |
'title' => 'Noindex Search Results', |
| 568 |
'description' => 'Prevent indexing of search result pages', |
| 569 |
'default' => true |
| 570 |
], |
| 571 |
'noindex_archives' => [ |
| 572 |
'type' => 'boolean', |
| 573 |
'title' => 'Noindex Archive Pages', |
| 574 |
'description' => 'Prevent indexing of archive pages (categories, tags, etc.)', |
| 575 |
'default' => false |
| 576 |
] |
| 577 |
]; |
| 578 |
} |
| 579 |
|
| 580 |
/** |
| 581 |
* Extract content data for robots meta generation |
| 582 |
* |
| 583 |
* @since 1.0.0 |
| 584 |
* |
| 585 |
* @param string $context_type The context type |
| 586 |
* @param int|null $context_id Optional. Context ID |
| 587 |
* @return array Extracted content data |
| 588 |
*/ |
| 589 |
private function extract_robots_content_data(string $context_type, ?int $context_id): array { |
| 590 |
$data = [ |
| 591 |
'context_type' => $context_type, |
| 592 |
'context_id' => $context_id, |
| 593 |
'url' => '', |
| 594 |
'is_search' => false, |
| 595 |
'is_archive' => false, |
| 596 |
'is_404' => false, |
| 597 |
'is_private' => false, |
| 598 |
'post_status' => '', |
| 599 |
'post_password' => false |
| 600 |
]; |
| 601 |
|
| 602 |
if ($context_type === 'site') { |
| 603 |
$data['url'] = home_url(); |
| 604 |
} elseif ($context_id && in_array($context_type, ['post', 'page', 'product'], true)) { |
| 605 |
$post = get_post($context_id); |
| 606 |
if ($post) { |
| 607 |
$data['url'] = get_permalink($post); |
| 608 |
$data['post_status'] = $post->post_status; |
| 609 |
$data['post_password'] = !empty($post->post_password); |
| 610 |
$data['is_private'] = $post->post_status === 'private'; |
| 611 |
} |
| 612 |
} |
| 613 |
|
| 614 |
// Detect special page types |
| 615 |
if (is_search()) { |
| 616 |
$data['is_search'] = true; |
| 617 |
} |
| 618 |
if (is_archive() || is_category() || is_tag() || is_tax()) { |
| 619 |
$data['is_archive'] = true; |
| 620 |
} |
| 621 |
if (is_404()) { |
| 622 |
$data['is_404'] = true; |
| 623 |
} |
| 624 |
|
| 625 |
return $data; |
| 626 |
} |
| 627 |
|
| 628 |
/** |
| 629 |
* Build main robots directive |
| 630 |
* |
| 631 |
* @since 1.0.0 |
| 632 |
* |
| 633 |
* @param array $settings Robots settings |
| 634 |
* @param string $context Context type |
| 635 |
* @param array $data Content data |
| 636 |
* @return string Robots directive |
| 637 |
*/ |
| 638 |
private function build_main_robots_directive(array $settings, string $context, array $data): string { |
| 639 |
$directives = []; |
| 640 |
|
| 641 |
// Determine index directive |
| 642 |
$index_directive = $this->determine_index_directive($settings, $context, $data); |
| 643 |
if ($index_directive) { |
| 644 |
$directives[] = $index_directive; |
| 645 |
} |
| 646 |
|
| 647 |
// Determine follow directive |
| 648 |
$follow_directive = $this->determine_follow_directive($settings, $context, $data); |
| 649 |
if ($follow_directive) { |
| 650 |
$directives[] = $follow_directive; |
| 651 |
} |
| 652 |
|
| 653 |
// Add other directives |
| 654 |
$other_directives = $this->get_other_robots_directives($settings, $context, $data); |
| 655 |
$directives = array_merge($directives, $other_directives); |
| 656 |
|
| 657 |
// Add advanced directives |
| 658 |
$advanced_directives = $this->get_advanced_robots_directives($settings); |
| 659 |
$directives = array_merge($directives, $advanced_directives); |
| 660 |
|
| 661 |
return implode(', ', array_filter($directives)); |
| 662 |
} |
| 663 |
|
| 664 |
/** |
| 665 |
* Determine index directive based on settings and context |
| 666 |
* |
| 667 |
* @since 1.0.0 |
| 668 |
* |
| 669 |
* @param array $settings Robots settings |
| 670 |
* @param string $context Context type |
| 671 |
* @param array $data Content data |
| 672 |
* @return string Index directive |
| 673 |
*/ |
| 674 |
private function determine_index_directive(array $settings, string $context, array $data): string { |
| 675 |
// Check for explicit noindex conditions |
| 676 |
if ($data['is_search'] && !empty($settings['noindex_search'])) { |
| 677 |
return 'noindex'; |
| 678 |
} |
| 679 |
|
| 680 |
if ($data['is_archive'] && !empty($settings['noindex_archives'])) { |
| 681 |
return 'noindex'; |
| 682 |
} |
| 683 |
|
| 684 |
if ($data['is_404']) { |
| 685 |
return 'noindex'; |
| 686 |
} |
| 687 |
|
| 688 |
if ($data['is_private'] || $data['post_password']) { |
| 689 |
return 'noindex'; |
| 690 |
} |
| 691 |
|
| 692 |
if ($data['post_status'] === 'draft' || $data['post_status'] === 'pending') { |
| 693 |
return 'noindex'; |
| 694 |
} |
| 695 |
|
| 696 |
// Use setting or default |
| 697 |
return $settings['robots_index'] ?? 'index'; |
| 698 |
} |
| 699 |
|
| 700 |
/** |
| 701 |
* Determine follow directive based on settings and context |
| 702 |
* |
| 703 |
* @since 1.0.0 |
| 704 |
* |
| 705 |
* @param array $settings Robots settings |
| 706 |
* @param string $context Context type |
| 707 |
* @param array $data Content data |
| 708 |
* @return string Follow directive |
| 709 |
*/ |
| 710 |
private function determine_follow_directive(array $settings, string $context, array $data): string { |
| 711 |
// If noindex, consider nofollow for crawl budget optimization |
| 712 |
$index_directive = $this->determine_index_directive($settings, $context, $data); |
| 713 |
if ($index_directive === 'noindex') { |
| 714 |
// For noindex pages, default to nofollow unless explicitly set to follow |
| 715 |
if (!isset($settings['robots_follow']) || $settings['robots_follow'] !== 'follow') { |
| 716 |
return 'nofollow'; |
| 717 |
} |
| 718 |
} |
| 719 |
|
| 720 |
// Use setting or default |
| 721 |
return $settings['robots_follow'] ?? 'follow'; |
| 722 |
} |
| 723 |
|
| 724 |
/** |
| 725 |
* Get other robots directives |
| 726 |
* |
| 727 |
* @since 1.0.0 |
| 728 |
* |
| 729 |
* @param array $settings Robots settings |
| 730 |
* @param string $context Context type |
| 731 |
* @param array $data Content data |
| 732 |
* @return array Other directives |
| 733 |
*/ |
| 734 |
private function get_other_robots_directives(array $settings, string $context, array $data): array { |
| 735 |
$directives = []; |
| 736 |
|
| 737 |
// Archive directive |
| 738 |
if (isset($settings['robots_archive']) && $settings['robots_archive'] === 'noarchive') { |
| 739 |
$directives[] = 'noarchive'; |
| 740 |
} |
| 741 |
|
| 742 |
// Snippet directive |
| 743 |
if (isset($settings['robots_snippet']) && $settings['robots_snippet'] === 'nosnippet') { |
| 744 |
$directives[] = 'nosnippet'; |
| 745 |
} |
| 746 |
|
| 747 |
// Image index directive |
| 748 |
if (!empty($settings['noimageindex'])) { |
| 749 |
$directives[] = 'noimageindex'; |
| 750 |
} |
| 751 |
|
| 752 |
// Video index directive |
| 753 |
if (!empty($settings['novideoindex'])) { |
| 754 |
$directives[] = 'novideoindex'; |
| 755 |
} |
| 756 |
|
| 757 |
// Translation directive |
| 758 |
if (!empty($settings['notranslate'])) { |
| 759 |
$directives[] = 'notranslate'; |
| 760 |
} |
| 761 |
|
| 762 |
return $directives; |
| 763 |
} |
| 764 |
|
| 765 |
/** |
| 766 |
* Get advanced robots directives with parameters |
| 767 |
* |
| 768 |
* @since 1.0.0 |
| 769 |
* |
| 770 |
* @param array $settings Robots settings |
| 771 |
* @return array Advanced directives |
| 772 |
*/ |
| 773 |
private function get_advanced_robots_directives(array $settings): array { |
| 774 |
$directives = []; |
| 775 |
|
| 776 |
// Max snippet |
| 777 |
if (isset($settings['max_snippet']) && $settings['max_snippet'] !== -1) { |
| 778 |
$directives[] = 'max-snippet:' . (int) $settings['max_snippet']; |
| 779 |
} |
| 780 |
|
| 781 |
// Max image preview |
| 782 |
if (isset($settings['max_image_preview']) && $settings['max_image_preview'] !== 'large') { |
| 783 |
$directives[] = 'max-image-preview:' . $settings['max_image_preview']; |
| 784 |
} |
| 785 |
|
| 786 |
// Max video preview |
| 787 |
if (isset($settings['max_video_preview']) && $settings['max_video_preview'] !== -1) { |
| 788 |
$directives[] = 'max-video-preview:' . (int) $settings['max_video_preview']; |
| 789 |
} |
| 790 |
|
| 791 |
// Unavailable after |
| 792 |
if (!empty($settings['unavailable_after'])) { |
| 793 |
$directives[] = 'unavailable_after:' . $settings['unavailable_after']; |
| 794 |
} |
| 795 |
|
| 796 |
return $directives; |
| 797 |
} |
| 798 |
|
| 799 |
/** |
| 800 |
* Build search engine specific directives |
| 801 |
* |
| 802 |
* @since 1.0.0 |
| 803 |
* |
| 804 |
* @param array $settings Robots settings |
| 805 |
* @param string $context Context type |
| 806 |
* @param array $data Content data |
| 807 |
* @return array Search engine specific directives |
| 808 |
*/ |
| 809 |
private function build_search_engine_specific_directives(array $settings, string $context, array $data): array { |
| 810 |
$specific_directives = []; |
| 811 |
|
| 812 |
// Check if any search engine specific settings are configured |
| 813 |
foreach ($this->search_engine_specific as $engine => $bots) { |
| 814 |
$engine_settings = $settings[$engine] ?? []; |
| 815 |
if (!empty($engine_settings)) { |
| 816 |
foreach ($bots as $bot) { |
| 817 |
$directive = $this->build_main_robots_directive($engine_settings, $context, $data); |
| 818 |
if (!empty($directive)) { |
| 819 |
$specific_directives[$bot] = $directive; |
| 820 |
} |
| 821 |
} |
| 822 |
} |
| 823 |
} |
| 824 |
|
| 825 |
return $specific_directives; |
| 826 |
} |
| 827 |
|
| 828 |
/** |
| 829 |
* Generate other meta directives |
| 830 |
* |
| 831 |
* @since 1.0.0 |
| 832 |
* |
| 833 |
* @param array $settings Robots settings |
| 834 |
* @param string $context Context type |
| 835 |
* @param array $data Content data |
| 836 |
* @return array Other meta directives |
| 837 |
*/ |
| 838 |
private function generate_other_meta_directives(array $settings, string $context, array $data): array { |
| 839 |
$meta = []; |
| 840 |
|
| 841 |
// Refresh directive for dynamic content |
| 842 |
if (!empty($settings['refresh_interval'])) { |
| 843 |
$meta['refresh'] = (int) $settings['refresh_interval']; |
| 844 |
} |
| 845 |
|
| 846 |
// Rating directive for content rating |
| 847 |
if (!empty($settings['rating'])) { |
| 848 |
$meta['rating'] = $settings['rating']; |
| 849 |
} |
| 850 |
|
| 851 |
return $meta; |
| 852 |
} |
| 853 |
|
| 854 |
/** |
| 855 |
* Clean canonical URL |
| 856 |
* |
| 857 |
* @since 1.0.0 |
| 858 |
* |
| 859 |
* @param string $url URL to clean |
| 860 |
* @return string Cleaned canonical URL |
| 861 |
*/ |
| 862 |
private function clean_canonical_url(string $url): string { |
| 863 |
// Remove query parameters that don't affect content |
| 864 |
$parsed_url = wp_parse_url($url); |
| 865 |
|
| 866 |
if (!$parsed_url) { |
| 867 |
return $url; |
| 868 |
} |
| 869 |
|
| 870 |
// Rebuild URL without unwanted query parameters |
| 871 |
$clean_url = $parsed_url['scheme'] . '://' . $parsed_url['host']; |
| 872 |
|
| 873 |
if (!empty($parsed_url['port']) && $parsed_url['port'] !== 80 && $parsed_url['port'] !== 443) { |
| 874 |
$clean_url .= ':' . $parsed_url['port']; |
| 875 |
} |
| 876 |
|
| 877 |
if (!empty($parsed_url['path'])) { |
| 878 |
$clean_url .= $parsed_url['path']; |
| 879 |
} |
| 880 |
|
| 881 |
// Only include specific query parameters that affect content |
| 882 |
if (!empty($parsed_url['query'])) { |
| 883 |
parse_str($parsed_url['query'], $query_params); |
| 884 |
$allowed_params = ['p', 'page_id', 'cat', 'tag', 's']; // WordPress core parameters |
| 885 |
$filtered_params = array_intersect_key($query_params, array_flip($allowed_params)); |
| 886 |
|
| 887 |
if (!empty($filtered_params)) { |
| 888 |
$clean_url .= '?' . http_build_query($filtered_params); |
| 889 |
} |
| 890 |
} |
| 891 |
|
| 892 |
return $clean_url; |
| 893 |
} |
| 894 |
|
| 895 |
/** |
| 896 |
* Get current canonical URL |
| 897 |
* |
| 898 |
* @since 1.0.0 |
| 899 |
* |
| 900 |
* @return string Current canonical URL |
| 901 |
*/ |
| 902 |
private function get_current_canonical_url(): string { |
| 903 |
if (is_admin()) { |
| 904 |
return home_url(); |
| 905 |
} |
| 906 |
|
| 907 |
global $wp; |
| 908 |
$current_url = home_url(add_query_arg([], $wp->request)); |
| 909 |
return $this->clean_canonical_url($current_url); |
| 910 |
} |
| 911 |
|
| 912 |
/** |
| 913 |
* Validate and clean URL |
| 914 |
* |
| 915 |
* @since 1.0.0 |
| 916 |
* |
| 917 |
* @param string $url URL to validate and clean |
| 918 |
* @return string Validated and cleaned URL |
| 919 |
*/ |
| 920 |
private function validate_and_clean_url(string $url): string { |
| 921 |
if (!filter_var($url, FILTER_VALIDATE_URL)) { |
| 922 |
return ''; |
| 923 |
} |
| 924 |
|
| 925 |
return $this->clean_canonical_url($url); |
| 926 |
} |
| 927 |
|
| 928 |
/** |
| 929 |
* Validate robots directive |
| 930 |
* |
| 931 |
* @since 1.0.0 |
| 932 |
* |
| 933 |
* @param string $directive Robots directive to validate |
| 934 |
* @return array Validation results |
| 935 |
*/ |
| 936 |
private function validate_robots_directive(string $directive): array { |
| 937 |
$validation = [ |
| 938 |
'valid' => true, |
| 939 |
'errors' => [], |
| 940 |
'warnings' => [], |
| 941 |
'suggestions' => [] |
| 942 |
]; |
| 943 |
|
| 944 |
if (empty($directive)) { |
| 945 |
$validation['warnings'][] = 'Empty robots directive'; |
| 946 |
return $validation; |
| 947 |
} |
| 948 |
|
| 949 |
$directives = array_map('trim', explode(',', $directive)); |
| 950 |
$known_directives = array_keys($this->robots_directives); |
| 951 |
$advanced_directives = array_keys($this->advanced_directives); |
| 952 |
|
| 953 |
foreach ($directives as $single_directive) { |
| 954 |
// Check for parameterized directives |
| 955 |
if (strpos($single_directive, ':') !== false) { |
| 956 |
[$directive_name, $directive_value] = explode(':', $single_directive, 2); |
| 957 |
$directive_name = trim($directive_name); |
| 958 |
|
| 959 |
if (!in_array($directive_name, $advanced_directives, true)) { |
| 960 |
$validation['warnings'][] = "Unknown advanced directive: {$directive_name}"; |
| 961 |
} |
| 962 |
} else { |
| 963 |
// Check basic directives |
| 964 |
if (!in_array($single_directive, $known_directives, true)) { |
| 965 |
$validation['warnings'][] = "Unknown robots directive: {$single_directive}"; |
| 966 |
} |
| 967 |
} |
| 968 |
} |
| 969 |
|
| 970 |
// Check for conflicting directives |
| 971 |
if (in_array('index', $directives, true) && in_array('noindex', $directives, true)) { |
| 972 |
$validation['errors'][] = 'Conflicting directives: index and noindex cannot be used together'; |
| 973 |
$validation['valid'] = false; |
| 974 |
} |
| 975 |
|
| 976 |
if (in_array('follow', $directives, true) && in_array('nofollow', $directives, true)) { |
| 977 |
$validation['errors'][] = 'Conflicting directives: follow and nofollow cannot be used together'; |
| 978 |
$validation['valid'] = false; |
| 979 |
} |
| 980 |
|
| 981 |
return $validation; |
| 982 |
} |
| 983 |
|
| 984 |
/** |
| 985 |
* Validate canonical URL |
| 986 |
* |
| 987 |
* @since 1.0.0 |
| 988 |
* |
| 989 |
* @param string $canonical_url Canonical URL to validate |
| 990 |
* @return array Validation results |
| 991 |
*/ |
| 992 |
private function validate_canonical_url(string $canonical_url): array { |
| 993 |
$validation = [ |
| 994 |
'valid' => true, |
| 995 |
'errors' => [], |
| 996 |
'warnings' => [], |
| 997 |
'suggestions' => [] |
| 998 |
]; |
| 999 |
|
| 1000 |
if (empty($canonical_url)) { |
| 1001 |
$validation['warnings'][] = 'No canonical URL specified'; |
| 1002 |
return $validation; |
| 1003 |
} |
| 1004 |
|
| 1005 |
if (!filter_var($canonical_url, FILTER_VALIDATE_URL)) { |
| 1006 |
$validation['errors'][] = 'Invalid canonical URL format'; |
| 1007 |
$validation['valid'] = false; |
| 1008 |
return $validation; |
| 1009 |
} |
| 1010 |
|
| 1011 |
// Check if canonical URL is accessible |
| 1012 |
$parsed_url = wp_parse_url($canonical_url); |
| 1013 |
$current_host = wp_parse_url(home_url(), PHP_URL_HOST); |
| 1014 |
|
| 1015 |
if ($parsed_url['host'] !== $current_host) { |
| 1016 |
$validation['warnings'][] = 'Canonical URL points to external domain - ensure this is intentional'; |
| 1017 |
} |
| 1018 |
|
| 1019 |
// Check for common canonical URL issues |
| 1020 |
if (strpos($canonical_url, '?') !== false) { |
| 1021 |
$validation['suggestions'][] = 'Consider removing query parameters from canonical URL for cleaner URLs'; |
| 1022 |
} |
| 1023 |
|
| 1024 |
if (substr($canonical_url, -1) !== '/' && !pathinfo($canonical_url, PATHINFO_EXTENSION)) { |
| 1025 |
$validation['suggestions'][] = 'Consider adding trailing slash to canonical URL for consistency'; |
| 1026 |
} |
| 1027 |
|
| 1028 |
return $validation; |
| 1029 |
} |
| 1030 |
|
| 1031 |
/** |
| 1032 |
* Validate advanced directive |
| 1033 |
* |
| 1034 |
* @since 1.0.0 |
| 1035 |
* |
| 1036 |
* @param string $directive_name Directive name |
| 1037 |
* @param mixed $directive_value Directive value |
| 1038 |
* @param array $spec Directive specification |
| 1039 |
* @return array Validation results |
| 1040 |
*/ |
| 1041 |
private function validate_advanced_directive(string $directive_name, $directive_value, array $spec): array { |
| 1042 |
$validation = [ |
| 1043 |
'valid' => true, |
| 1044 |
'errors' => [], |
| 1045 |
'warnings' => [], |
| 1046 |
'suggestions' => [] |
| 1047 |
]; |
| 1048 |
|
| 1049 |
switch ($spec['type']) { |
| 1050 |
case 'parameterized': |
| 1051 |
if (isset($spec['allowed_values'])) { |
| 1052 |
if (!in_array($directive_value, $spec['allowed_values'], true)) { |
| 1053 |
$validation['errors'][] = "{$directive_name} must be one of: " . implode(', ', $spec['allowed_values']); |
| 1054 |
$validation['valid'] = false; |
| 1055 |
} |
| 1056 |
} elseif (isset($spec['min_value'], $spec['max_value'])) { |
| 1057 |
$value = (int) $directive_value; |
| 1058 |
if ($value < $spec['min_value'] || $value > $spec['max_value']) { |
| 1059 |
$validation['errors'][] = "{$directive_name} must be between {$spec['min_value']} and {$spec['max_value']}"; |
| 1060 |
$validation['valid'] = false; |
| 1061 |
} |
| 1062 |
} |
| 1063 |
break; |
| 1064 |
case 'date': |
| 1065 |
if (!strtotime($directive_value)) { |
| 1066 |
$validation['errors'][] = "{$directive_name} must be a valid date"; |
| 1067 |
$validation['valid'] = false; |
| 1068 |
} |
| 1069 |
break; |
| 1070 |
} |
| 1071 |
|
| 1072 |
return $validation; |
| 1073 |
} |
| 1074 |
|
| 1075 |
/** |
| 1076 |
* Check for directive conflicts |
| 1077 |
* |
| 1078 |
* @since 1.0.0 |
| 1079 |
* |
| 1080 |
* @param array $robots_meta Robots meta array |
| 1081 |
* @return array Validation results |
| 1082 |
*/ |
| 1083 |
private function check_directive_conflicts(array $robots_meta): array { |
| 1084 |
$validation = [ |
| 1085 |
'valid' => true, |
| 1086 |
'errors' => [], |
| 1087 |
'warnings' => [], |
| 1088 |
'suggestions' => [] |
| 1089 |
]; |
| 1090 |
|
| 1091 |
if (isset($robots_meta['robots'])) { |
| 1092 |
$directives = array_map('trim', explode(',', $robots_meta['robots'])); |
| 1093 |
|
| 1094 |
// Check for logical conflicts |
| 1095 |
if (in_array('noindex', $directives, true) && in_array('nosnippet', $directives, true)) { |
| 1096 |
$validation['suggestions'][] = 'nosnippet is redundant when noindex is used'; |
| 1097 |
} |
| 1098 |
|
| 1099 |
if (in_array('noindex', $directives, true) && in_array('noarchive', $directives, true)) { |
| 1100 |
$validation['suggestions'][] = 'noarchive is redundant when noindex is used'; |
| 1101 |
} |
| 1102 |
} |
| 1103 |
|
| 1104 |
return $validation; |
| 1105 |
} |
| 1106 |
|
| 1107 |
/** |
| 1108 |
* Calculate robots score based on validation |
| 1109 |
* |
| 1110 |
* @since 1.0.0 |
| 1111 |
* |
| 1112 |
* @param array $validation Validation results |
| 1113 |
* @return int Robots score (0-100) |
| 1114 |
*/ |
| 1115 |
private function calculate_robots_score(array $validation): int { |
| 1116 |
$score = 100; |
| 1117 |
|
| 1118 |
// Deduct points for errors and warnings |
| 1119 |
$score -= count($validation['errors']) * 20; |
| 1120 |
$score -= count($validation['warnings']) * 10; |
| 1121 |
|
| 1122 |
return max(0, $score); |
| 1123 |
} |
| 1124 |
|
| 1125 |
/** |
| 1126 |
* Add robots SEO suggestions |
| 1127 |
* |
| 1128 |
* @since 1.0.0 |
| 1129 |
* |
| 1130 |
* @param array $robots_meta Robots meta array |
| 1131 |
* @param array $validation Current validation results |
| 1132 |
* @return array Updated validation results |
| 1133 |
*/ |
| 1134 |
private function add_robots_seo_suggestions(array $robots_meta, array $validation): array { |
| 1135 |
// Suggest canonical URL if missing |
| 1136 |
if (!isset($robots_meta['canonical'])) { |
| 1137 |
$validation['suggestions'][] = 'Add canonical URL to prevent duplicate content issues'; |
| 1138 |
} |
| 1139 |
|
| 1140 |
// Suggest max-image-preview for content with images |
| 1141 |
if (!isset($robots_meta['max-image-preview'])) { |
| 1142 |
$validation['suggestions'][] = 'Consider setting max-image-preview:large for better image visibility in search results'; |
| 1143 |
} |
| 1144 |
|
| 1145 |
// 2025 SEO best practices |
| 1146 |
if (isset($robots_meta['robots'])) { |
| 1147 |
$directives = array_map('trim', explode(',', $robots_meta['robots'])); |
| 1148 |
|
| 1149 |
if (!in_array('index', $directives, true) && !in_array('noindex', $directives, true)) { |
| 1150 |
$validation['suggestions'][] = 'Explicitly specify index or noindex directive for clarity'; |
| 1151 |
} |
| 1152 |
} |
| 1153 |
|
| 1154 |
return $validation; |
| 1155 |
} |
| 1156 |
|
| 1157 |
/** |
| 1158 |
* Convert robots meta to HTML meta tags |
| 1159 |
* |
| 1160 |
* @since 1.0.0 |
| 1161 |
* |
| 1162 |
* @param array $robots_meta Robots meta array |
| 1163 |
* @return array HTML meta tags |
| 1164 |
*/ |
| 1165 |
private function convert_robots_to_meta_tags(array $robots_meta): array { |
| 1166 |
$meta_tags = []; |
| 1167 |
|
| 1168 |
foreach ($robots_meta as $name => $content) { |
| 1169 |
if (!empty($content)) { |
| 1170 |
if ($name === 'canonical') { |
| 1171 |
$meta_tags[] = [ |
| 1172 |
'rel' => 'canonical', |
| 1173 |
'href' => esc_url($content) |
| 1174 |
]; |
| 1175 |
} else { |
| 1176 |
$meta_tags[] = [ |
| 1177 |
'name' => esc_attr($name), |
| 1178 |
'content' => esc_attr($content) |
| 1179 |
]; |
| 1180 |
} |
| 1181 |
} |
| 1182 |
} |
| 1183 |
|
| 1184 |
return $meta_tags; |
| 1185 |
} |
| 1186 |
|
| 1187 |
/** |
| 1188 |
* Detect duplicate content patterns |
| 1189 |
* |
| 1190 |
* @since 1.0.0 |
| 1191 |
* |
| 1192 |
* @param string $url URL to check for duplicates |
| 1193 |
* @return array Duplicate URL patterns |
| 1194 |
*/ |
| 1195 |
private function detect_duplicate_patterns(string $url): array { |
| 1196 |
$duplicates = []; |
| 1197 |
$parsed_url = wp_parse_url($url); |
| 1198 |
|
| 1199 |
if (!$parsed_url) { |
| 1200 |
return $duplicates; |
| 1201 |
} |
| 1202 |
|
| 1203 |
$base_url = $parsed_url['scheme'] . '://' . $parsed_url['host'] . ($parsed_url['path'] ?? ''); |
| 1204 |
|
| 1205 |
// Check for common duplicate patterns |
| 1206 |
$patterns = [ |
| 1207 |
$base_url . '/', // With trailing slash |
| 1208 |
rtrim($base_url, '/'), // Without trailing slash |
| 1209 |
$base_url . '?utm_source=test', // With tracking parameters |
| 1210 |
$base_url . '#section', // With fragment |
| 1211 |
$base_url . '?ref=homepage' // With referrer parameters |
| 1212 |
]; |
| 1213 |
|
| 1214 |
foreach ($patterns as $pattern) { |
| 1215 |
if ($pattern !== $url) { |
| 1216 |
$duplicates[] = $pattern; |
| 1217 |
} |
| 1218 |
} |
| 1219 |
|
| 1220 |
return $duplicates; |
| 1221 |
} |
| 1222 |
|
| 1223 |
/** |
| 1224 |
* Suggest canonical from duplicates |
| 1225 |
* |
| 1226 |
* @since 1.0.0 |
| 1227 |
* |
| 1228 |
* @param array $duplicates Array of duplicate URLs |
| 1229 |
* @return string Suggested canonical URL |
| 1230 |
*/ |
| 1231 |
private function suggest_canonical_from_duplicates(array $duplicates): string { |
| 1232 |
if (empty($duplicates)) { |
| 1233 |
return ''; |
| 1234 |
} |
| 1235 |
|
| 1236 |
// Prefer URLs without query parameters and with trailing slashes |
| 1237 |
$scored_urls = []; |
| 1238 |
|
| 1239 |
foreach ($duplicates as $url) { |
| 1240 |
$score = 0; |
| 1241 |
|
| 1242 |
// Prefer URLs without query parameters |
| 1243 |
if (strpos($url, '?') === false) { |
| 1244 |
$score += 10; |
| 1245 |
} |
| 1246 |
|
| 1247 |
// Prefer URLs with trailing slashes for directories |
| 1248 |
if (substr($url, -1) === '/' && !pathinfo($url, PATHINFO_EXTENSION)) { |
| 1249 |
$score += 5; |
| 1250 |
} |
| 1251 |
|
| 1252 |
// Prefer HTTPS |
| 1253 |
if (strpos($url, 'https://') === 0) { |
| 1254 |
$score += 3; |
| 1255 |
} |
| 1256 |
|
| 1257 |
$scored_urls[$url] = $score; |
| 1258 |
} |
| 1259 |
|
| 1260 |
arsort($scored_urls); |
| 1261 |
return array_key_first($scored_urls); |
| 1262 |
} |
| 1263 |
|
| 1264 |
/** |
| 1265 |
* Get duplicate content recommendations |
| 1266 |
* |
| 1267 |
* @since 1.0.0 |
| 1268 |
* |
| 1269 |
* @param array $duplicates Array of duplicate URLs |
| 1270 |
* @return array Recommendations |
| 1271 |
*/ |
| 1272 |
private function get_duplicate_content_recommendations(array $duplicates): array { |
| 1273 |
$recommendations = []; |
| 1274 |
|
| 1275 |
if (count($duplicates) > 1) { |
| 1276 |
$recommendations[] = 'Set up 301 redirects from duplicate URLs to the canonical version'; |
| 1277 |
$recommendations[] = 'Use consistent URL structure across your site'; |
| 1278 |
$recommendations[] = 'Implement proper canonical tags on all pages'; |
| 1279 |
} |
| 1280 |
|
| 1281 |
foreach ($duplicates as $url) { |
| 1282 |
if (strpos($url, '?') !== false) { |
| 1283 |
$recommendations[] = 'Remove unnecessary query parameters from URLs'; |
| 1284 |
break; |
| 1285 |
} |
| 1286 |
} |
| 1287 |
|
| 1288 |
return array_unique($recommendations); |
| 1289 |
} |
| 1290 |
} |
| 1291 |
|
