| 1 |
<?php |
| 2 |
/** |
| 3 |
* Schema Input Validator Class |
| 4 |
* |
| 5 |
* Provides comprehensive input validation and sanitization for schema data |
| 6 |
* including JSON schema validation, XSS protection, and data integrity checks. |
| 7 |
* |
| 8 |
* @package ThinkRank\SEO |
| 9 |
* @since 1.0.0 |
| 10 |
*/ |
| 11 |
|
| 12 |
declare(strict_types=1); |
| 13 |
|
| 14 |
namespace ThinkRank\SEO; |
| 15 |
|
| 16 |
// Prevent direct access |
| 17 |
if (!defined('ABSPATH')) { |
| 18 |
exit; |
| 19 |
} |
| 20 |
|
| 21 |
/** |
| 22 |
* Schema Input Validator Class |
| 23 |
* |
| 24 |
* Handles validation and sanitization of all schema-related inputs |
| 25 |
* with comprehensive security measures and data integrity checks. |
| 26 |
* |
| 27 |
* @since 1.0.0 |
| 28 |
*/ |
| 29 |
class Schema_Input_Validator { |
| 30 |
|
| 31 |
/** |
| 32 |
* Allowed schema types with their validation rules |
| 33 |
* |
| 34 |
* @since 1.0.0 |
| 35 |
* @var array |
| 36 |
*/ |
| 37 |
private array $allowed_schema_types = [ |
| 38 |
'Article' => [ |
| 39 |
'required_fields' => ['@type', 'headline', 'author'], |
| 40 |
'optional_fields' => ['description', 'datePublished', 'dateModified', 'image', 'url'], |
| 41 |
'max_length' => ['headline' => 110, 'description' => 160] |
| 42 |
], |
| 43 |
'BlogPosting' => [ |
| 44 |
'required_fields' => ['@type', 'headline', 'author'], |
| 45 |
'optional_fields' => ['description', 'datePublished', 'dateModified', 'image', 'url'], |
| 46 |
'max_length' => ['headline' => 110, 'description' => 160] |
| 47 |
], |
| 48 |
'TechnicalArticle' => [ |
| 49 |
'required_fields' => ['@type', 'headline', 'author'], |
| 50 |
'optional_fields' => ['description', 'datePublished', 'dateModified', 'image', 'url', 'dependencies', 'proficiencyLevel'], |
| 51 |
'max_length' => ['headline' => 110, 'description' => 160] |
| 52 |
], |
| 53 |
'NewsArticle' => [ |
| 54 |
'required_fields' => ['@type', 'headline', 'author'], |
| 55 |
'optional_fields' => ['description', 'datePublished', 'dateModified', 'image', 'url', 'dateline'], |
| 56 |
'max_length' => ['headline' => 110, 'description' => 160] |
| 57 |
], |
| 58 |
'ScholarlyArticle' => [ |
| 59 |
'required_fields' => ['@type', 'headline', 'author'], |
| 60 |
'optional_fields' => ['description', 'datePublished', 'dateModified', 'image', 'url', 'citation', 'abstract'], |
| 61 |
'max_length' => ['headline' => 110, 'description' => 160] |
| 62 |
], |
| 63 |
'Report' => [ |
| 64 |
'required_fields' => ['@type', 'headline', 'author'], |
| 65 |
'optional_fields' => ['description', 'datePublished', 'dateModified', 'image', 'url'], |
| 66 |
'max_length' => ['headline' => 110, 'description' => 160] |
| 67 |
], |
| 68 |
'Organization' => [ |
| 69 |
'required_fields' => ['@type', 'name'], |
| 70 |
'optional_fields' => ['description', 'url', 'logo', 'address', 'contactPoint'], |
| 71 |
'max_length' => ['name' => 100, 'description' => 160] |
| 72 |
], |
| 73 |
'LocalBusiness' => [ |
| 74 |
'required_fields' => ['@type', 'name', 'address'], |
| 75 |
'optional_fields' => ['description', 'url', 'telephone', 'openingHours'], |
| 76 |
'max_length' => ['name' => 100, 'description' => 160] |
| 77 |
], |
| 78 |
'Product' => [ |
| 79 |
'required_fields' => ['@type', 'name'], |
| 80 |
'optional_fields' => ['description', 'image', 'brand', 'offers'], |
| 81 |
'max_length' => ['name' => 100, 'description' => 160] |
| 82 |
], |
| 83 |
'WebSite' => [ |
| 84 |
'required_fields' => ['@type', 'name', 'url'], |
| 85 |
'optional_fields' => ['description', 'potentialAction'], |
| 86 |
'max_length' => ['name' => 100, 'description' => 160] |
| 87 |
], |
| 88 |
'FAQPage' => [ |
| 89 |
'required_fields' => ['@type', 'mainEntity'], |
| 90 |
'optional_fields' => ['name', 'description'], |
| 91 |
'max_length' => ['name' => 100, 'description' => 160] |
| 92 |
], |
| 93 |
'SoftwareApplication' => [ |
| 94 |
'required_fields' => ['@type', 'name'], |
| 95 |
'optional_fields' => ['description', 'applicationCategory', 'operatingSystem'], |
| 96 |
'max_length' => ['name' => 100, 'description' => 160] |
| 97 |
], |
| 98 |
'Event' => [ |
| 99 |
'required_fields' => ['@type', 'name', 'startDate'], |
| 100 |
'optional_fields' => ['description', 'location', 'organizer', 'endDate', 'eventStatus', 'eventAttendanceMode', 'url'], |
| 101 |
'max_length' => ['name' => 100, 'description' => 160] |
| 102 |
], |
| 103 |
'Person' => [ |
| 104 |
'required_fields' => ['@type', 'name'], |
| 105 |
'optional_fields' => ['description', 'url', 'image', 'jobTitle'], |
| 106 |
'max_length' => ['name' => 100, 'description' => 160] |
| 107 |
], |
| 108 |
'HowTo' => [ |
| 109 |
'required_fields' => ['@type', 'name'], |
| 110 |
'optional_fields' => ['description', 'totalTime', 'prepTime', 'difficulty', 'estimatedCost', 'supply', 'tool', 'step', 'yield', 'image', 'video'], |
| 111 |
'max_length' => ['name' => 100, 'description' => 160] |
| 112 |
], |
| 113 |
'BreadcrumbList' => [ |
| 114 |
'required_fields' => ['@type', 'itemListElement'], |
| 115 |
'optional_fields' => ['name', 'description', 'numberOfItems'], |
| 116 |
'max_length' => ['name' => 100, 'description' => 160] |
| 117 |
], |
| 118 |
'VideoObject' => [ |
| 119 |
'required_fields' => ['@type', 'name', 'thumbnailUrl', 'uploadDate'], |
| 120 |
'optional_fields' => ['description', 'contentUrl', 'embedUrl', 'duration', 'url'], |
| 121 |
'max_length' => ['name' => 110, 'description' => 160] |
| 122 |
], |
| 123 |
// Offered by the metabox Schema Type dropdown and registered in |
| 124 |
// Schema_Factory, but missing here — so a Review could be generated and |
| 125 |
// never deployed (#462). Field list mirrors Schema_Factory::Review. |
| 126 |
'Review' => [ |
| 127 |
'required_fields' => ['@type', 'itemReviewed', 'reviewRating', 'author'], |
| 128 |
'optional_fields' => ['reviewBody', 'datePublished', 'publisher', 'name', 'url'], |
| 129 |
'max_length' => ['name' => 110, 'reviewBody' => 500] |
| 130 |
] |
| 131 |
]; |
| 132 |
|
| 133 |
/** |
| 134 |
* Dangerous HTML tags and attributes to strip |
| 135 |
* |
| 136 |
* @since 1.0.0 |
| 137 |
* @var array |
| 138 |
*/ |
| 139 |
private array $dangerous_tags = [ |
| 140 |
'script', 'iframe', 'object', 'embed', 'form', 'input', 'button', |
| 141 |
'link', 'meta', 'style', 'base', 'frame', 'frameset' |
| 142 |
]; |
| 143 |
|
| 144 |
/** |
| 145 |
* Allowed URL protocols |
| 146 |
* |
| 147 |
* @since 1.0.0 |
| 148 |
* @var array |
| 149 |
*/ |
| 150 |
private array $allowed_protocols = ['http', 'https', 'mailto', 'tel']; |
| 151 |
|
| 152 |
/** |
| 153 |
* Maximum allowed JSON depth to prevent JSON bomb attacks |
| 154 |
* |
| 155 |
* @since 1.0.0 |
| 156 |
* @var int |
| 157 |
*/ |
| 158 |
private const MAX_JSON_DEPTH = 10; |
| 159 |
|
| 160 |
/** |
| 161 |
* Maximum payload size in bytes (500KB) |
| 162 |
* |
| 163 |
* @since 1.0.0 |
| 164 |
* @var int |
| 165 |
*/ |
| 166 |
private const MAX_PAYLOAD_SIZE = 512000; |
| 167 |
|
| 168 |
/** |
| 169 |
* Maximum array size (number of elements) |
| 170 |
* |
| 171 |
* @since 1.0.0 |
| 172 |
* @var int |
| 173 |
*/ |
| 174 |
private const MAX_ARRAY_SIZE = 100; |
| 175 |
|
| 176 |
/** |
| 177 |
* Maximum string length for any single field |
| 178 |
* |
| 179 |
* @since 1.0.0 |
| 180 |
* @var int |
| 181 |
*/ |
| 182 |
private const MAX_STRING_LENGTH = 10000; |
| 183 |
|
| 184 |
/** |
| 185 |
* Validate and sanitize schema data |
| 186 |
* |
| 187 |
* @since 1.0.0 |
| 188 |
* |
| 189 |
* @param array $schema_data Raw schema data |
| 190 |
* @param string $schema_type Schema type |
| 191 |
* @return array Validation result with sanitized data |
| 192 |
*/ |
| 193 |
public function validate_schema_data(array $schema_data, string $schema_type): array { |
| 194 |
$result = [ |
| 195 |
'valid' => false, |
| 196 |
'sanitized_data' => [], |
| 197 |
'errors' => [], |
| 198 |
'warnings' => [] |
| 199 |
]; |
| 200 |
|
| 201 |
try { |
| 202 |
// 1. Validate payload size to prevent DoS attacks |
| 203 |
$size_validation = $this->validate_payload_size($schema_data); |
| 204 |
if (!$size_validation['valid']) { |
| 205 |
$result['errors'] = array_merge($result['errors'], $size_validation['errors']); |
| 206 |
return $result; |
| 207 |
} |
| 208 |
|
| 209 |
// 2. Validate schema type |
| 210 |
if (!$this->is_valid_schema_type($schema_type)) { |
| 211 |
$result['errors'][] = "Invalid schema type: {$schema_type}"; |
| 212 |
return $result; |
| 213 |
} |
| 214 |
|
| 215 |
// 3. Validate JSON structure and depth |
| 216 |
$structure_validation = $this->validate_json_structure($schema_data, $schema_type); |
| 217 |
if (!$structure_validation['valid']) { |
| 218 |
$result['errors'] = array_merge($result['errors'], $structure_validation['errors']); |
| 219 |
return $result; |
| 220 |
} |
| 221 |
|
| 222 |
// 3.1. Validate JSON depth to prevent JSON bomb attacks |
| 223 |
if (!$this->validate_json_depth($schema_data)) { |
| 224 |
$result['errors'][] = 'Schema data exceeds maximum allowed depth (' . self::MAX_JSON_DEPTH . ' levels)'; |
| 225 |
return $result; |
| 226 |
} |
| 227 |
|
| 228 |
// 4. Sanitize all input data |
| 229 |
$sanitized_data = $this->sanitize_schema_data($schema_data); |
| 230 |
|
| 231 |
// 5. Validate required fields |
| 232 |
$field_validation = $this->validate_required_fields($sanitized_data, $schema_type); |
| 233 |
if (!$field_validation['valid']) { |
| 234 |
$result['errors'] = array_merge($result['errors'], $field_validation['errors']); |
| 235 |
} |
| 236 |
|
| 237 |
// 6. Validate data types and formats |
| 238 |
$format_validation = $this->validate_data_formats($sanitized_data, $schema_type); |
| 239 |
if (!$format_validation['valid']) { |
| 240 |
$result['errors'] = array_merge($result['errors'], $format_validation['errors']); |
| 241 |
} |
| 242 |
$result['warnings'] = array_merge($result['warnings'], $format_validation['warnings']); |
| 243 |
|
| 244 |
// 7. Validate content length limits |
| 245 |
$length_validation = $this->validate_content_lengths($sanitized_data, $schema_type); |
| 246 |
if (!$length_validation['valid']) { |
| 247 |
$result['warnings'] = array_merge($result['warnings'], $length_validation['warnings']); |
| 248 |
} |
| 249 |
|
| 250 |
$result['valid'] = empty($result['errors']); |
| 251 |
$result['sanitized_data'] = $sanitized_data; |
| 252 |
|
| 253 |
} catch (\Exception $e) { |
| 254 |
$result['errors'][] = 'Schema validation failed: ' . $e->getMessage(); |
| 255 |
} |
| 256 |
|
| 257 |
return $result; |
| 258 |
} |
| 259 |
|
| 260 |
/** |
| 261 |
* Validate payload size to prevent DoS attacks |
| 262 |
* |
| 263 |
* @since 1.0.0 |
| 264 |
* |
| 265 |
* @param array $schema_data Schema data to validate |
| 266 |
* @return array Validation result |
| 267 |
*/ |
| 268 |
private function validate_payload_size(array $schema_data): array { |
| 269 |
$result = ['valid' => true, 'errors' => []]; |
| 270 |
|
| 271 |
// Calculate approximate payload size |
| 272 |
$payload_size = strlen(wp_json_encode($schema_data)); |
| 273 |
|
| 274 |
if ($payload_size > self::MAX_PAYLOAD_SIZE) { |
| 275 |
$result['errors'][] = sprintf( |
| 276 |
'Payload size (%s) exceeds maximum allowed size (%s)', |
| 277 |
size_format($payload_size), |
| 278 |
size_format(self::MAX_PAYLOAD_SIZE) |
| 279 |
); |
| 280 |
$result['valid'] = false; |
| 281 |
} |
| 282 |
|
| 283 |
// Validate array sizes and string lengths recursively |
| 284 |
$structure_validation = $this->validate_data_structure($schema_data); |
| 285 |
if (!$structure_validation['valid']) { |
| 286 |
$result['errors'] = array_merge($result['errors'], $structure_validation['errors']); |
| 287 |
$result['valid'] = false; |
| 288 |
} |
| 289 |
|
| 290 |
return $result; |
| 291 |
} |
| 292 |
|
| 293 |
/** |
| 294 |
* Validate data structure (arrays and strings) |
| 295 |
* |
| 296 |
* @since 1.0.0 |
| 297 |
* |
| 298 |
* @param mixed $data Data to validate |
| 299 |
* @param string $path Current path for error reporting |
| 300 |
* @return array Validation result |
| 301 |
*/ |
| 302 |
private function validate_data_structure($data, string $path = ''): array { |
| 303 |
$result = ['valid' => true, 'errors' => []]; |
| 304 |
|
| 305 |
if (is_array($data)) { |
| 306 |
// Check array size |
| 307 |
if (count($data) > self::MAX_ARRAY_SIZE) { |
| 308 |
$result['errors'][] = sprintf( |
| 309 |
'Array at path "%s" contains %d elements, maximum allowed is %d', |
| 310 |
$path ?: 'root', |
| 311 |
count($data), |
| 312 |
self::MAX_ARRAY_SIZE |
| 313 |
); |
| 314 |
$result['valid'] = false; |
| 315 |
} |
| 316 |
|
| 317 |
// Recursively validate nested data |
| 318 |
foreach ($data as $key => $value) { |
| 319 |
$current_path = $path ? "{$path}.{$key}" : $key; |
| 320 |
$nested_validation = $this->validate_data_structure($value, $current_path); |
| 321 |
if (!$nested_validation['valid']) { |
| 322 |
$result['errors'] = array_merge($result['errors'], $nested_validation['errors']); |
| 323 |
$result['valid'] = false; |
| 324 |
} |
| 325 |
} |
| 326 |
} elseif (is_string($data)) { |
| 327 |
// Check string length |
| 328 |
if (strlen($data) > self::MAX_STRING_LENGTH) { |
| 329 |
$result['errors'][] = sprintf( |
| 330 |
'String at path "%s" is %d characters, maximum allowed is %d', |
| 331 |
$path ?: 'value', |
| 332 |
strlen($data), |
| 333 |
self::MAX_STRING_LENGTH |
| 334 |
); |
| 335 |
$result['valid'] = false; |
| 336 |
} |
| 337 |
} |
| 338 |
|
| 339 |
return $result; |
| 340 |
} |
| 341 |
|
| 342 |
/** |
| 343 |
* Validate schema type |
| 344 |
* |
| 345 |
* @since 1.0.0 |
| 346 |
* |
| 347 |
* @param string $schema_type Schema type to validate |
| 348 |
* @return bool Validation result |
| 349 |
*/ |
| 350 |
private function is_valid_schema_type(string $schema_type): bool { |
| 351 |
return isset($this->allowed_schema_types[$schema_type]); |
| 352 |
} |
| 353 |
|
| 354 |
/** |
| 355 |
* Validate JSON structure |
| 356 |
* |
| 357 |
* @since 1.0.0 |
| 358 |
* |
| 359 |
* @param array $schema_data Schema data |
| 360 |
* @param string $schema_type Schema type |
| 361 |
* @return array Validation result |
| 362 |
*/ |
| 363 |
private function validate_json_structure(array $schema_data, string $schema_type): array { |
| 364 |
$result = ['valid' => true, 'errors' => []]; |
| 365 |
|
| 366 |
// Check for required @context |
| 367 |
if (!isset($schema_data['@context'])) { |
| 368 |
$result['errors'][] = 'Missing required @context field'; |
| 369 |
$result['valid'] = false; |
| 370 |
} elseif ($schema_data['@context'] !== 'https://schema.org') { |
| 371 |
$result['errors'][] = 'Invalid @context value. Must be "https://schema.org"'; |
| 372 |
$result['valid'] = false; |
| 373 |
} |
| 374 |
|
| 375 |
// Check for required @type. |
| 376 |
// `@type` may be an array — "@type": ["Product","Offer"] is valid |
| 377 |
// JSON-LD. Comparing an array against a string emitted an "Array to |
| 378 |
// string conversion" warning and always failed (#468), so match if the |
| 379 |
// expected type appears anywhere in the list. |
| 380 |
if (!isset($schema_data['@type'])) { |
| 381 |
$result['errors'][] = 'Missing required @type field'; |
| 382 |
$result['valid'] = false; |
| 383 |
} else { |
| 384 |
$declared_types = is_array($schema_data['@type']) |
| 385 |
? array_map('strval', $schema_data['@type']) |
| 386 |
: [(string) $schema_data['@type']]; |
| 387 |
|
| 388 |
if (!in_array($schema_type, $declared_types, true)) { |
| 389 |
$result['errors'][] = sprintf( |
| 390 |
"Schema @type '%s' does not match expected type '%s'", |
| 391 |
implode(', ', $declared_types), |
| 392 |
$schema_type |
| 393 |
); |
| 394 |
$result['valid'] = false; |
| 395 |
} |
| 396 |
} |
| 397 |
|
| 398 |
return $result; |
| 399 |
} |
| 400 |
|
| 401 |
/** |
| 402 |
* Sanitize schema data recursively |
| 403 |
* |
| 404 |
* @since 1.0.0 |
| 405 |
* |
| 406 |
* @param mixed $data Data to sanitize |
| 407 |
* @param string $field_key Current field key for context-aware sanitization |
| 408 |
* @return mixed Sanitized data |
| 409 |
*/ |
| 410 |
private function sanitize_schema_data($data, string $field_key = '') { |
| 411 |
if (is_array($data)) { |
| 412 |
$sanitized = []; |
| 413 |
foreach ($data as $key => $value) { |
| 414 |
$sanitized_key = $this->sanitize_key($key); |
| 415 |
$sanitized[$sanitized_key] = $this->sanitize_schema_data($value, $sanitized_key); |
| 416 |
} |
| 417 |
return $sanitized; |
| 418 |
} |
| 419 |
|
| 420 |
if (is_string($data)) { |
| 421 |
return $this->sanitize_string_value($data, $field_key); |
| 422 |
} |
| 423 |
|
| 424 |
if (is_numeric($data)) { |
| 425 |
return $this->sanitize_numeric_value($data); |
| 426 |
} |
| 427 |
|
| 428 |
if (is_bool($data)) { |
| 429 |
return $data; |
| 430 |
} |
| 431 |
|
| 432 |
// For other types, convert to string and sanitize |
| 433 |
return $this->sanitize_string_value((string) $data, $field_key); |
| 434 |
} |
| 435 |
|
| 436 |
/** |
| 437 |
* Sanitize array key |
| 438 |
* |
| 439 |
* @since 1.0.0 |
| 440 |
* |
| 441 |
* @param string $key Array key |
| 442 |
* @return string Sanitized key |
| 443 |
*/ |
| 444 |
private function sanitize_key($key): string { |
| 445 |
// Ensure key is a string first |
| 446 |
if (!is_string($key)) { |
| 447 |
return (string) $key; |
| 448 |
} |
| 449 |
|
| 450 |
// For schema data, preserve the original key names to maintain case sensitivity |
| 451 |
// Schema.org properties are case-sensitive (e.g., startDate, not startdate) |
| 452 |
// Only do basic validation without changing the case |
| 453 |
if (preg_match('/^[a-zA-Z@][a-zA-Z0-9@_-]*$/', $key)) { |
| 454 |
return $key; // Return as-is if it's a valid schema property name |
| 455 |
} |
| 456 |
|
| 457 |
// Fallback to WordPress sanitization for invalid keys |
| 458 |
return sanitize_key($key); |
| 459 |
} |
| 460 |
|
| 461 |
/** |
| 462 |
* Sanitize string value with context-aware sanitization |
| 463 |
* |
| 464 |
* @since 1.0.0 |
| 465 |
* |
| 466 |
* @param string $value String value |
| 467 |
* @param string $field_name Field name for context-aware sanitization |
| 468 |
* @return string Sanitized value |
| 469 |
*/ |
| 470 |
private function sanitize_string_value(string $value, string $field_name = ''): string { |
| 471 |
// Handle URLs differently to preserve valid URL structure |
| 472 |
if (in_array($field_name, ['url', 'sameAs', 'logo', 'image', 'mainEntityOfPage'], true)) { |
| 473 |
return esc_url_raw($value); |
| 474 |
} |
| 475 |
|
| 476 |
// Handle email fields |
| 477 |
if (in_array($field_name, ['email'], true)) { |
| 478 |
return sanitize_email($value); |
| 479 |
} |
| 480 |
|
| 481 |
// Handle description fields that may contain basic HTML |
| 482 |
if (in_array($field_name, ['description', 'text', 'articleBody'], true)) { |
| 483 |
// Allow basic HTML but strip dangerous tags |
| 484 |
$allowed_html = [ |
| 485 |
'p' => [], |
| 486 |
'br' => [], |
| 487 |
'strong' => [], |
| 488 |
'em' => [], |
| 489 |
'b' => [], |
| 490 |
'i' => [] |
| 491 |
]; |
| 492 |
$value = wp_kses($value, $allowed_html); |
| 493 |
} else { |
| 494 |
// For other fields, remove all HTML tags |
| 495 |
$value = wp_strip_all_tags($value); |
| 496 |
} |
| 497 |
|
| 498 |
// Sanitize for database storage. Note: escaping is intentionally NOT done |
| 499 |
// here. This value is stored and later emitted as JSON-LD inside a |
| 500 |
// <script type="application/ld+json"> block, where wp_json_encode() is the |
| 501 |
// correct encoder. Running esc_html() on input would persist HTML entities |
| 502 |
// (e.g. "Ben & Jerry's" -> "Ben & Jerry's") into the structured |
| 503 |
// data. Escape at the output boundary, not at storage. |
| 504 |
$value = sanitize_text_field($value); |
| 505 |
|
| 506 |
return trim($value); |
| 507 |
} |
| 508 |
|
| 509 |
/** |
| 510 |
* Sanitize numeric value |
| 511 |
* |
| 512 |
* @since 1.0.0 |
| 513 |
* |
| 514 |
* @param mixed $value Numeric value |
| 515 |
* @return float|int Sanitized numeric value |
| 516 |
*/ |
| 517 |
private function sanitize_numeric_value($value) { |
| 518 |
if (is_int($value) || ctype_digit((string) $value)) { |
| 519 |
return (int) $value; |
| 520 |
} |
| 521 |
|
| 522 |
return (float) $value; |
| 523 |
} |
| 524 |
|
| 525 |
/** |
| 526 |
* Validate required fields |
| 527 |
* |
| 528 |
* @since 1.0.0 |
| 529 |
* |
| 530 |
* @param array $schema_data Schema data |
| 531 |
* @param string $schema_type Schema type |
| 532 |
* @return array Validation result |
| 533 |
*/ |
| 534 |
private function validate_required_fields(array $schema_data, string $schema_type): array { |
| 535 |
$result = ['valid' => true, 'errors' => []]; |
| 536 |
$rules = $this->allowed_schema_types[$schema_type]; |
| 537 |
|
| 538 |
foreach ($rules['required_fields'] as $field) { |
| 539 |
if (!isset($schema_data[$field]) || empty($schema_data[$field])) { |
| 540 |
$result['errors'][] = "Missing required field: {$field}"; |
| 541 |
$result['valid'] = false; |
| 542 |
} |
| 543 |
} |
| 544 |
return $result; |
| 545 |
} |
| 546 |
|
| 547 |
/** |
| 548 |
* Validate data formats |
| 549 |
* |
| 550 |
* @since 1.0.0 |
| 551 |
* |
| 552 |
* @param array $schema_data Schema data |
| 553 |
* @param string $schema_type Schema type |
| 554 |
* @return array Validation result |
| 555 |
*/ |
| 556 |
private function validate_data_formats(array $schema_data, string $schema_type): array { |
| 557 |
$result = ['valid' => true, 'errors' => [], 'warnings' => []]; |
| 558 |
|
| 559 |
foreach ($schema_data as $field => $value) { |
| 560 |
// sameAs is a list, so its members never reached the string branch |
| 561 |
// below and free text entered in a social-profile field saved |
| 562 |
// cleanly, then shipped as invalid structured data (#480). |
| 563 |
if (is_array($value) && in_array($field, ['url', 'sameAs', 'logo', 'image'], true)) { |
| 564 |
foreach ($value as $item) { |
| 565 |
if (!is_string($item) || '' === trim($item)) { |
| 566 |
continue; |
| 567 |
} |
| 568 |
|
| 569 |
if (!$this->is_valid_url($item)) { |
| 570 |
$result['errors'][] = "Invalid URL format for field: {$field} ({$item})"; |
| 571 |
$result['valid'] = false; |
| 572 |
} |
| 573 |
} |
| 574 |
|
| 575 |
continue; |
| 576 |
} |
| 577 |
|
| 578 |
if (is_string($value)) { |
| 579 |
// Validate URLs |
| 580 |
if (in_array($field, ['url', 'sameAs', 'logo', 'image'], true) && !empty($value)) { |
| 581 |
if (!$this->is_valid_url($value)) { |
| 582 |
$result['errors'][] = "Invalid URL format for field: {$field}"; |
| 583 |
$result['valid'] = false; |
| 584 |
} |
| 585 |
} |
| 586 |
|
| 587 |
// Validate email addresses |
| 588 |
if (in_array($field, ['email'], true) && !empty($value)) { |
| 589 |
if (!is_email($value)) { |
| 590 |
$result['errors'][] = "Invalid email format for field: {$field}"; |
| 591 |
$result['valid'] = false; |
| 592 |
} |
| 593 |
} |
| 594 |
|
| 595 |
// Validate dates |
| 596 |
if (in_array($field, ['datePublished', 'dateModified'], true) && !empty($value)) { |
| 597 |
if (!$this->is_valid_date($value)) { |
| 598 |
$result['warnings'][] = "Invalid date format for field: {$field}. Use ISO 8601 format."; |
| 599 |
} |
| 600 |
} |
| 601 |
} |
| 602 |
} |
| 603 |
|
| 604 |
return $result; |
| 605 |
} |
| 606 |
|
| 607 |
/** |
| 608 |
* Validate content lengths |
| 609 |
* |
| 610 |
* @since 1.0.0 |
| 611 |
* |
| 612 |
* @param array $schema_data Schema data |
| 613 |
* @param string $schema_type Schema type |
| 614 |
* @return array Validation result |
| 615 |
*/ |
| 616 |
private function validate_content_lengths(array $schema_data, string $schema_type): array { |
| 617 |
$result = ['valid' => true, 'warnings' => []]; |
| 618 |
$rules = $this->allowed_schema_types[$schema_type]; |
| 619 |
|
| 620 |
if (isset($rules['max_length'])) { |
| 621 |
foreach ($rules['max_length'] as $field => $max_length) { |
| 622 |
if (isset($schema_data[$field]) && is_string($schema_data[$field])) { |
| 623 |
$length = strlen($schema_data[$field]); |
| 624 |
if ($length > $max_length) { |
| 625 |
$result['warnings'][] = "Field '{$field}' exceeds recommended length of {$max_length} characters (current: {$length})"; |
| 626 |
} |
| 627 |
} |
| 628 |
} |
| 629 |
} |
| 630 |
|
| 631 |
return $result; |
| 632 |
} |
| 633 |
|
| 634 |
/** |
| 635 |
* Validate URL format and protocol |
| 636 |
* |
| 637 |
* @since 1.0.0 |
| 638 |
* |
| 639 |
* @param string $url URL to validate |
| 640 |
* @return bool Validation result |
| 641 |
*/ |
| 642 |
private function is_valid_url(string $url): bool { |
| 643 |
// Basic URL validation |
| 644 |
if (!filter_var($url, FILTER_VALIDATE_URL)) { |
| 645 |
return false; |
| 646 |
} |
| 647 |
|
| 648 |
// Check allowed protocols |
| 649 |
$parsed = wp_parse_url($url); |
| 650 |
if (!isset($parsed['scheme']) || !in_array($parsed['scheme'], $this->allowed_protocols, true)) { |
| 651 |
return false; |
| 652 |
} |
| 653 |
|
| 654 |
return true; |
| 655 |
} |
| 656 |
|
| 657 |
/** |
| 658 |
* Validate date format |
| 659 |
* |
| 660 |
* @since 1.0.0 |
| 661 |
* |
| 662 |
* @param string $date Date to validate |
| 663 |
* @return bool Validation result |
| 664 |
*/ |
| 665 |
private function is_valid_date(string $date): bool { |
| 666 |
// Check ISO 8601 format |
| 667 |
$formats = [ |
| 668 |
'Y-m-d\TH:i:s\Z', |
| 669 |
'Y-m-d\TH:i:sP', |
| 670 |
'Y-m-d\TH:i:s', |
| 671 |
'Y-m-d' |
| 672 |
]; |
| 673 |
|
| 674 |
foreach ($formats as $format) { |
| 675 |
$parsed = \DateTime::createFromFormat($format, $date); |
| 676 |
if ($parsed && $parsed->format($format) === $date) { |
| 677 |
return true; |
| 678 |
} |
| 679 |
} |
| 680 |
|
| 681 |
return false; |
| 682 |
} |
| 683 |
|
| 684 |
/** |
| 685 |
* Check rate limiting for user |
| 686 |
* |
| 687 |
* @since 1.0.0 |
| 688 |
* |
| 689 |
* @param int $user_id User ID |
| 690 |
* @param string $action Action type |
| 691 |
* @param int $limit Rate limit (requests per hour) |
| 692 |
* @return bool Whether request is allowed |
| 693 |
*/ |
| 694 |
public function check_rate_limit(int $user_id, string $action, int $limit = 100): bool { |
| 695 |
// Persist the window in a transient (object cache / options) so the limit |
| 696 |
// is enforced ACROSS requests. A per-request static array — as used |
| 697 |
// previously — always starts empty on a fresh PHP process and therefore |
| 698 |
// never throttled anything. |
| 699 |
$key = 'thinkrank_schema_rl_' . $user_id . '_' . sanitize_key($action); |
| 700 |
|
| 701 |
// Serialize the read-modify-write with a MySQL named lock so concurrent |
| 702 |
// requests can't each read the same timestamp list, individually pass the |
| 703 |
// limit check, and overwrite one another — which would let bursts slip |
| 704 |
// past the configured limit. GET_LOCK is DB-level, so it serializes the |
| 705 |
// critical section regardless of where the transient is stored. |
| 706 |
global $wpdb; |
| 707 |
$lock_name = substr('tr_schema_rl_' . md5($key), 0, 64); |
| 708 |
$have_lock = ($wpdb instanceof \wpdb) |
| 709 |
? (int) $wpdb->get_var($wpdb->prepare('SELECT GET_LOCK(%s, %d)', $lock_name, 3)) === 1 |
| 710 |
: false; |
| 711 |
|
| 712 |
try { |
| 713 |
$current_time = time(); |
| 714 |
$window_start = $current_time - HOUR_IN_SECONDS; // 1 hour window |
| 715 |
|
| 716 |
$timestamps = get_transient($key); |
| 717 |
if (!is_array($timestamps)) { |
| 718 |
$timestamps = []; |
| 719 |
} |
| 720 |
|
| 721 |
// Drop entries outside the window. |
| 722 |
$timestamps = array_values(array_filter( |
| 723 |
$timestamps, |
| 724 |
static function ($timestamp) use ($window_start) { |
| 725 |
return (int) $timestamp > $window_start; |
| 726 |
} |
| 727 |
)); |
| 728 |
|
| 729 |
// Check if limit exceeded. |
| 730 |
if (count($timestamps) >= $limit) { |
| 731 |
set_transient($key, $timestamps, HOUR_IN_SECONDS); |
| 732 |
return false; |
| 733 |
} |
| 734 |
|
| 735 |
// Record this request. |
| 736 |
$timestamps[] = $current_time; |
| 737 |
set_transient($key, $timestamps, HOUR_IN_SECONDS); |
| 738 |
|
| 739 |
return true; |
| 740 |
} finally { |
| 741 |
if ($have_lock) { |
| 742 |
$wpdb->query($wpdb->prepare('SELECT RELEASE_LOCK(%s)', $lock_name)); |
| 743 |
} |
| 744 |
} |
| 745 |
} |
| 746 |
|
| 747 |
/** |
| 748 |
* Validate user permissions for schema operations |
| 749 |
* |
| 750 |
* @since 1.0.0 |
| 751 |
* |
| 752 |
* @param string $operation Operation type |
| 753 |
* @param int $user_id User ID |
| 754 |
* @return array Validation result |
| 755 |
*/ |
| 756 |
public function validate_user_permissions(string $operation, int $user_id): array { |
| 757 |
$result = ['valid' => false, 'errors' => []]; |
| 758 |
|
| 759 |
// Check if user exists and is logged in |
| 760 |
if (!$user_id || !get_userdata($user_id)) { |
| 761 |
$result['errors'][] = 'Invalid user or user not logged in'; |
| 762 |
return $result; |
| 763 |
} |
| 764 |
|
| 765 |
// Check operation-specific permissions. |
| 766 |
// |
| 767 |
// #457 loosened the route permission_callbacks to the delegable |
| 768 |
// `thinkrank_schema` capability, but these handler-level checks still |
| 769 |
// demanded edit_posts / publish_posts / manage_options — so a role |
| 770 |
// granted Schema access could generate and validate but was denied on |
| 771 |
// deploy, bulk operations and everything site-context. That is exactly |
| 772 |
// the symptom #457 set out to fix (#470). A holder of thinkrank_schema |
| 773 |
// satisfies any schema operation; the built-in caps remain as the |
| 774 |
// fallback for roles that never went through the Role Manager. |
| 775 |
$has_schema_cap = user_can($user_id, 'thinkrank_schema'); |
| 776 |
|
| 777 |
switch ($operation) { |
| 778 |
case 'generate': |
| 779 |
case 'validate': |
| 780 |
case 'optimize': |
| 781 |
if (!$has_schema_cap && !user_can($user_id, 'edit_posts')) { |
| 782 |
$result['errors'][] = 'Insufficient permissions for schema generation/validation'; |
| 783 |
return $result; |
| 784 |
} |
| 785 |
break; |
| 786 |
|
| 787 |
case 'deploy': |
| 788 |
if (!$has_schema_cap && !user_can($user_id, 'publish_posts')) { |
| 789 |
$result['errors'][] = 'Insufficient permissions for schema deployment'; |
| 790 |
return $result; |
| 791 |
} |
| 792 |
break; |
| 793 |
|
| 794 |
case 'manage_settings': |
| 795 |
case 'bulk_operations': |
| 796 |
if (!$has_schema_cap && !user_can($user_id, 'manage_options')) { |
| 797 |
$result['errors'][] = 'Insufficient permissions for schema management'; |
| 798 |
return $result; |
| 799 |
} |
| 800 |
break; |
| 801 |
|
| 802 |
default: |
| 803 |
$result['errors'][] = "Unknown operation: {$operation}"; |
| 804 |
return $result; |
| 805 |
} |
| 806 |
|
| 807 |
// Check rate limiting |
| 808 |
$rate_limits = [ |
| 809 |
'generate' => 50, // 50 generations per hour |
| 810 |
'validate' => 100, // 100 validations per hour |
| 811 |
'deploy' => 20, // 20 deployments per hour |
| 812 |
'optimize' => 30, // 30 optimizations per hour |
| 813 |
'bulk_operations' => 5 // 5 bulk operations per hour |
| 814 |
]; |
| 815 |
|
| 816 |
$limit = $rate_limits[$operation] ?? 100; |
| 817 |
if (!$this->check_rate_limit($user_id, $operation, $limit)) { |
| 818 |
$result['errors'][] = "Rate limit exceeded for {$operation}. Please try again later."; |
| 819 |
return $result; |
| 820 |
} |
| 821 |
|
| 822 |
$result['valid'] = true; |
| 823 |
return $result; |
| 824 |
} |
| 825 |
|
| 826 |
/** |
| 827 |
* Sanitize and validate context parameters with ownership checks |
| 828 |
* |
| 829 |
* @since 1.0.0 |
| 830 |
* |
| 831 |
* @param string $context_type Context type |
| 832 |
* @param int|null $context_id Context ID |
| 833 |
* @param int|null $user_id User ID for ownership validation |
| 834 |
* @return array Validation result |
| 835 |
*/ |
| 836 |
public function validate_context_parameters(string $context_type, ?int $context_id, ?int $user_id = null): array { |
| 837 |
$result = ['valid' => false, 'errors' => [], 'sanitized_data' => []]; |
| 838 |
|
| 839 |
// Sanitize context type |
| 840 |
$context_type = sanitize_key($context_type); |
| 841 |
$allowed_types = ['site', 'post', 'page', 'product']; |
| 842 |
|
| 843 |
if (!in_array($context_type, $allowed_types, true)) { |
| 844 |
$result['errors'][] = "Invalid context type: {$context_type}"; |
| 845 |
return $result; |
| 846 |
} |
| 847 |
|
| 848 |
// Validate context ID and ownership |
| 849 |
if ($context_type !== 'site') { |
| 850 |
if (!$context_id || $context_id <= 0) { |
| 851 |
$result['errors'][] = 'Context ID is required for non-site contexts'; |
| 852 |
return $result; |
| 853 |
} |
| 854 |
|
| 855 |
$context_id = absint($context_id); |
| 856 |
$post = get_post($context_id); |
| 857 |
|
| 858 |
if (!$post) { |
| 859 |
$result['errors'][] = "Invalid context ID: {$context_id}"; |
| 860 |
return $result; |
| 861 |
} |
| 862 |
|
| 863 |
// SECURITY: Check context ownership. |
| 864 |
// Fails closed on a missing user — a security helper that waves the |
| 865 |
// check through when it cannot identify the caller is the wrong way |
| 866 |
// round. Every caller passes a real ID, so this only tightens an |
| 867 |
// unreachable path. |
| 868 |
if (!$user_id || !$this->validate_context_ownership($post, $user_id)) { |
| 869 |
$result['errors'][] = "Access denied: You don't have permission to modify this {$context_type}"; |
| 870 |
return $result; |
| 871 |
} |
| 872 |
} else { |
| 873 |
$context_id = null; // Site context doesn't use ID |
| 874 |
|
| 875 |
// SECURITY: Check site-level permissions for site context. |
| 876 |
// user_can($user_id, …) rather than current_user_can() so this |
| 877 |
// agrees with the rest of the validator outside a REST request, |
| 878 |
// where the current user and $user_id can differ (cron, CLI). |
| 879 |
// |
| 880 |
// Accepts the delegable `thinkrank_schema` capability as well as |
| 881 |
// manage_options: /schema/settings already lets a delegated role |
| 882 |
// edit site schema settings, so blocking site-context generate and |
| 883 |
// deploy for the same role was inconsistent (#470). |
| 884 |
if (!$user_id || (!user_can($user_id, 'thinkrank_schema') && !user_can($user_id, 'manage_options'))) { |
| 885 |
$result['errors'][] = 'Access denied: You need administrator privileges for site-level schema operations'; |
| 886 |
return $result; |
| 887 |
} |
| 888 |
} |
| 889 |
|
| 890 |
$result['valid'] = true; |
| 891 |
$result['sanitized_data'] = [ |
| 892 |
'context_type' => $context_type, |
| 893 |
'context_id' => $context_id |
| 894 |
]; |
| 895 |
|
| 896 |
return $result; |
| 897 |
} |
| 898 |
|
| 899 |
/** |
| 900 |
* Validate context ownership |
| 901 |
* |
| 902 |
* @since 1.0.0 |
| 903 |
* |
| 904 |
* @param \WP_Post $post Post object |
| 905 |
* @param int $user_id User ID |
| 906 |
* @return bool Whether user has permission |
| 907 |
*/ |
| 908 |
private function validate_context_ownership(\WP_Post $post, int $user_id): bool { |
| 909 |
// `edit_post` is a meta capability: map_meta_cap() already resolves |
| 910 |
// authorship, published state, and edit_others_posts for this specific |
| 911 |
// post. It is the whole check. |
| 912 |
// |
| 913 |
// Two fallbacks used to sit under it and between them defeated the |
| 914 |
// function. One granted access on authorship alone, which hands a |
| 915 |
// Contributor back a post they lost edit rights to once it published. |
| 916 |
// The other granted access to anyone holding the post type's *general* |
| 917 |
// edit_posts capability — a cap every Author and Contributor has, that |
| 918 |
// says nothing about this post — so ownership validation returned true |
| 919 |
// for every post on the site (#326). |
| 920 |
// |
| 921 |
// user_can() rather than current_user_can() so the method honours the |
| 922 |
// $user_id it was handed, matching validate_user_permissions(). |
| 923 |
return user_can($user_id, 'edit_post', $post->ID); |
| 924 |
} |
| 925 |
|
| 926 |
/** |
| 927 |
* Validate JSON depth to prevent JSON bomb attacks |
| 928 |
* |
| 929 |
* @since 1.0.0 |
| 930 |
* |
| 931 |
* @param mixed $data Data to validate |
| 932 |
* @param int $depth Current depth level |
| 933 |
* @return bool Whether depth is within limits |
| 934 |
*/ |
| 935 |
private function validate_json_depth($data, int $depth = 0): bool { |
| 936 |
if ($depth > self::MAX_JSON_DEPTH) { |
| 937 |
return false; |
| 938 |
} |
| 939 |
|
| 940 |
if (is_array($data)) { |
| 941 |
foreach ($data as $value) { |
| 942 |
if (!$this->validate_json_depth($value, $depth + 1)) { |
| 943 |
return false; |
| 944 |
} |
| 945 |
} |
| 946 |
} |
| 947 |
|
| 948 |
return true; |
| 949 |
} |
| 950 |
|
| 951 |
/** |
| 952 |
* Validate and sanitize options array |
| 953 |
* |
| 954 |
* @since 1.0.0 |
| 955 |
* |
| 956 |
* @param array $options Options array |
| 957 |
* @return array Sanitized options |
| 958 |
*/ |
| 959 |
public function sanitize_options(array $options): array { |
| 960 |
$sanitized = []; |
| 961 |
// Anything omitted here is dropped before the manager sees it, which is |
| 962 |
// why apply_content_schema_settings_from_options() and the per-request |
| 963 |
// schema-type opt-in were unreachable from REST (#470). The list now |
| 964 |
// covers every option the generate path actually reads. |
| 965 |
// |
| 966 |
// `validation_level` previously allowed 'basic' and rejected 'lenient', |
| 967 |
// disagreeing with Schema_Settings_Config, validate_settings() and the |
| 968 |
// update-settings ability, which all use 'lenient'. |
| 969 |
// `deployment_method` no longer advertises microdata/rdfa, which |
| 970 |
// determine_deployment_method() hardcodes away to json_ld anyway. |
| 971 |
$allowed_options = [ |
| 972 |
'deployment_method' => ['json_ld'], |
| 973 |
'validation_level' => ['strict', 'moderate', 'lenient'], |
| 974 |
'include_meta' => 'boolean', |
| 975 |
'minify_output' => 'boolean', |
| 976 |
'cache_duration' => 'integer', |
| 977 |
'rich_snippets_optimization' => 'boolean', |
| 978 |
'knowledge_graph' => 'boolean', |
| 979 |
'auto_generate_schema' => 'boolean', |
| 980 |
'enable_article_schema' => 'boolean', |
| 981 |
'enable_faq_schema' => 'boolean', |
| 982 |
'enable_howto_schema' => 'boolean', |
| 983 |
'enable_product_schema' => 'boolean', |
| 984 |
'enable_local_business' => 'boolean', |
| 985 |
'enable_breadcrumbs_schema' => 'boolean', |
| 986 |
]; |
| 987 |
|
| 988 |
foreach ($options as $key => $value) { |
| 989 |
$sanitized_key = sanitize_key($key); |
| 990 |
|
| 991 |
if (!isset($allowed_options[$sanitized_key])) { |
| 992 |
continue; // Skip unknown options |
| 993 |
} |
| 994 |
|
| 995 |
$rule = $allowed_options[$sanitized_key]; |
| 996 |
|
| 997 |
if (is_array($rule)) { |
| 998 |
// Enum validation |
| 999 |
if (in_array($value, $rule, true)) { |
| 1000 |
$sanitized[$sanitized_key] = $value; |
| 1001 |
} |
| 1002 |
} elseif ($rule === 'boolean') { |
| 1003 |
$sanitized[$sanitized_key] = (bool) $value; |
| 1004 |
} elseif ($rule === 'integer') { |
| 1005 |
$sanitized[$sanitized_key] = absint($value); |
| 1006 |
} else { |
| 1007 |
$sanitized[$sanitized_key] = sanitize_text_field($value); |
| 1008 |
} |
| 1009 |
} |
| 1010 |
|
| 1011 |
return $sanitized; |
| 1012 |
} |
| 1013 |
} |
| 1014 |
|
