| 1 |
<?php |
| 2 |
|
| 3 |
/** |
| 4 |
* Image SEO Manager Class |
| 5 |
* |
| 6 |
* Manages image-specific SEO settings including automatic |
| 7 |
* ALT and TITLE attribute management. |
| 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 |
// Prevent direct access |
| 19 |
if (!defined('ABSPATH')) { |
| 20 |
exit; |
| 21 |
} |
| 22 |
|
| 23 |
/** |
| 24 |
* Image SEO Manager Class |
| 25 |
* |
| 26 |
* Handles image attribute optimization and settings management. |
| 27 |
* |
| 28 |
* @since 1.0.0 |
| 29 |
*/ |
| 30 |
class Image_SEO_Manager extends Abstract_SEO_Manager { |
| 31 |
|
| 32 |
/** |
| 33 |
* Accepted values for the `alt_source` setting. |
| 34 |
* |
| 35 |
* Single source of truth for the schema's enum, the REST arg constraint and |
| 36 |
* validate_settings(), so the three cannot disagree about what is legal. |
| 37 |
* |
| 38 |
* @since 1.29.1 |
| 39 |
* @var string[] |
| 40 |
*/ |
| 41 |
public const ALT_SOURCES = ['template', 'ai']; |
| 42 |
|
| 43 |
/** |
| 44 |
* Memoized site separator symbol. |
| 45 |
* |
| 46 |
* Resolved once per request rather than on every image processed during |
| 47 |
* the_content, since the separator is a site-wide option. |
| 48 |
* |
| 49 |
* @since 1.16.0 |
| 50 |
* @var string|null |
| 51 |
*/ |
| 52 |
private ?string $separator = null; |
| 53 |
|
| 54 |
/** |
| 55 |
* Attachment IDs whose cache purge is being held back, or null when not. |
| 56 |
* |
| 57 |
* A bulk fill writes alt text onto hundreds of attachments that between |
| 58 |
* them appear on a handful of pages. Resolving usage per attachment would |
| 59 |
* run one unindexed postmeta scan per image; holding the IDs and resolving |
| 60 |
* the whole batch in one query at the end is the same answer for a |
| 61 |
* fraction of the work. |
| 62 |
* |
| 63 |
* @since 2.12.0 |
| 64 |
* @var int[]|null |
| 65 |
*/ |
| 66 |
private ?array $deferred_alt_purge = null; |
| 67 |
|
| 68 |
/** |
| 69 |
* What the most recent purge_alt_caches() call did. |
| 70 |
* |
| 71 |
* Read by the MCP abilities, which have to report the cache outcome |
| 72 |
* alongside the write and cannot see inside fill_attachment_alt(). |
| 73 |
* |
| 74 |
* @since 2.12.0 |
| 75 |
* @var array{posts: int[], warnings: string[]} |
| 76 |
*/ |
| 77 |
private array $last_alt_cache_purge = ['posts' => [], 'warnings' => []]; |
| 78 |
|
| 79 |
/** |
| 80 |
* Constructor |
| 81 |
* |
| 82 |
* @since 1.0.0 |
| 83 |
*/ |
| 84 |
public function __construct() { |
| 85 |
parent::__construct('image_seo'); |
| 86 |
} |
| 87 |
|
| 88 |
/** |
| 89 |
* Validate SEO settings (implements interface) |
| 90 |
* |
| 91 |
* @since 1.0.0 |
| 92 |
* |
| 93 |
* @param array $settings Settings array to validate |
| 94 |
* @return array Validation results |
| 95 |
*/ |
| 96 |
public function validate_settings(array $settings): array { |
| 97 |
$validation = [ |
| 98 |
'valid' => true, |
| 99 |
'errors' => [], |
| 100 |
'warnings' => [], |
| 101 |
'suggestions' => [], |
| 102 |
'score' => 100 |
| 103 |
]; |
| 104 |
|
| 105 |
$boolean_fields = ['add_missing_alt', 'add_missing_title', 'save_alt_to_media', 'auto_fill_on_upload', 'media_alt_overwrite']; |
| 106 |
foreach ($boolean_fields as $field) { |
| 107 |
if (isset($settings[$field]) && !is_bool($settings[$field])) { |
| 108 |
$validation['errors'][] = sprintf('%s must be a boolean value', $field); |
| 109 |
$validation['valid'] = false; |
| 110 |
} |
| 111 |
} |
| 112 |
|
| 113 |
$string_fields = ['alt_format', 'title_format']; |
| 114 |
foreach ($string_fields as $field) { |
| 115 |
if (isset($settings[$field]) && !is_string($settings[$field])) { |
| 116 |
$validation['errors'][] = sprintf('%s must be a string', $field); |
| 117 |
$validation['valid'] = false; |
| 118 |
} |
| 119 |
} |
| 120 |
|
| 121 |
// The schema declares alt_source as an enum but nothing used to check |
| 122 |
// it, so any string persisted. The consumer falls back to the template |
| 123 |
// path on an unknown value, which hid the drift rather than surfacing |
| 124 |
// it — the settings screen just had no option to select (#323). |
| 125 |
if (isset($settings['alt_source']) && !in_array($settings['alt_source'], self::ALT_SOURCES, true)) { |
| 126 |
$validation['errors'][] = sprintf( |
| 127 |
'alt_source must be one of: %s', |
| 128 |
implode(', ', self::ALT_SOURCES) |
| 129 |
); |
| 130 |
$validation['valid'] = false; |
| 131 |
} |
| 132 |
|
| 133 |
return $validation; |
| 134 |
} |
| 135 |
|
| 136 |
/** |
| 137 |
* Get output data for frontend rendering (implements interface) |
| 138 |
* |
| 139 |
* @since 1.0.0 |
| 140 |
* |
| 141 |
* @param string $context_type The context type |
| 142 |
* @param int|null $context_id Optional. Context ID |
| 143 |
* @return array Output data ready for frontend rendering |
| 144 |
*/ |
| 145 |
public function get_output_data(string $context_type, ?int $context_id): array { |
| 146 |
return $this->get_settings($context_type, $context_id); |
| 147 |
} |
| 148 |
|
| 149 |
/** |
| 150 |
* Get default settings for a context type (implements interface) |
| 151 |
* |
| 152 |
* @since 1.0.0 |
| 153 |
* |
| 154 |
* @param string $context_type The context type to get defaults for |
| 155 |
* @return array Default settings array |
| 156 |
*/ |
| 157 |
/** |
| 158 |
* Images per batch when alt text comes from the vision model. |
| 159 |
* |
| 160 |
* Each one is a paid call of a few seconds; 10 keeps a batch inside a |
| 161 |
* normal PHP timeout and keeps the spend per click predictable. |
| 162 |
* |
| 163 |
* @since 1.28.0 |
| 164 |
* @var int |
| 165 |
*/ |
| 166 |
private const AI_BATCH_LIMIT = 10; |
| 167 |
|
| 168 |
public function get_default_settings(string $context_type): array { |
| 169 |
return [ |
| 170 |
'add_missing_alt' => false, |
| 171 |
'alt_format' => '%filename%', |
| 172 |
'add_missing_title' => false, |
| 173 |
'title_format' => '%title% %separator% %sitename%', |
| 174 |
// Media Library alt persistence (writes _wp_attachment_image_alt) |
| 175 |
'save_alt_to_media' => false, |
| 176 |
'auto_fill_on_upload' => false, |
| 177 |
'media_alt_overwrite' => false, |
| 178 |
// 'template' rewrites the filename; 'ai' looks at the picture. |
| 179 |
// Defaults to template because AI costs the user money per image. |
| 180 |
'alt_source' => 'template', |
| 181 |
]; |
| 182 |
} |
| 183 |
|
| 184 |
/** |
| 185 |
* Get settings schema definition (implements interface) |
| 186 |
* |
| 187 |
* @since 1.0.0 |
| 188 |
* |
| 189 |
* @param string $context_type The context type to get schema for |
| 190 |
* @return array Settings schema definition |
| 191 |
*/ |
| 192 |
public function get_settings_schema(string $context_type): array { |
| 193 |
return [ |
| 194 |
'add_missing_alt' => [ |
| 195 |
'type' => 'boolean', |
| 196 |
'title' => __('Add Missing Alt Attributes', 'thinkrank'), |
| 197 |
'description' => __('Automatically add ALT attributes to images if they are missing.', 'thinkrank'), |
| 198 |
'default' => false |
| 199 |
], |
| 200 |
'alt_format' => [ |
| 201 |
'type' => 'string', |
| 202 |
'title' => __('Alt attribute format', 'thinkrank'), |
| 203 |
'description' => __('The format to use for automatically generated ALT attributes.', 'thinkrank'), |
| 204 |
'default' => '%filename%' |
| 205 |
], |
| 206 |
'add_missing_title' => [ |
| 207 |
'type' => 'boolean', |
| 208 |
'title' => __('Add Missing Title Attributes', 'thinkrank'), |
| 209 |
'description' => __('Automatically add TITLE attributes to images if they are missing.', 'thinkrank'), |
| 210 |
'default' => false |
| 211 |
], |
| 212 |
'title_format' => [ |
| 213 |
'type' => 'string', |
| 214 |
'title' => __('Title attribute format', 'thinkrank'), |
| 215 |
'description' => __('The format to use for automatically generated TITLE attributes.', 'thinkrank'), |
| 216 |
'default' => '%title% %separator% %sitename%' |
| 217 |
], |
| 218 |
'alt_source' => [ |
| 219 |
'type' => 'string', |
| 220 |
'title' => __('Alt text source', 'thinkrank'), |
| 221 |
'description' => __('“Template” builds alt text from the filename and title. “AI” looks at the image itself and describes what is in it — this uses your AI provider key and costs one call per image.', 'thinkrank'), |
| 222 |
'default' => 'template', |
| 223 |
'enum' => self::ALT_SOURCES |
| 224 |
], |
| 225 |
'save_alt_to_media' => [ |
| 226 |
'type' => 'boolean', |
| 227 |
'title' => __('Save alt text to the Media Library', 'thinkrank'), |
| 228 |
'description' => __('Persist generated alt text onto the attachment record so it works everywhere, not just in rendered content.', 'thinkrank'), |
| 229 |
'default' => false |
| 230 |
], |
| 231 |
'auto_fill_on_upload' => [ |
| 232 |
'type' => 'boolean', |
| 233 |
'title' => __('Fill alt text on upload', 'thinkrank'), |
| 234 |
'description' => __('When a new image is uploaded, automatically save generated alt text to the Media Library.', 'thinkrank'), |
| 235 |
'default' => false |
| 236 |
], |
| 237 |
'media_alt_overwrite' => [ |
| 238 |
'type' => 'boolean', |
| 239 |
'title' => __('Overwrite existing alt text', 'thinkrank'), |
| 240 |
'description' => __('Replace alt text that is already set, instead of only filling images that are missing it.', 'thinkrank'), |
| 241 |
'default' => false |
| 242 |
] |
| 243 |
]; |
| 244 |
} |
| 245 |
|
| 246 |
/** |
| 247 |
* Process content and inject missing image attributes |
| 248 |
* |
| 249 |
* @since 1.0.0 |
| 250 |
* @param string $content The content to process |
| 251 |
* @return string Processed content |
| 252 |
*/ |
| 253 |
public function process_content(string $content, $post_id = null): string { |
| 254 |
$settings = $this->get_settings('site'); |
| 255 |
|
| 256 |
if (empty($settings['add_missing_alt']) && empty($settings['add_missing_title'])) { |
| 257 |
return $content; |
| 258 |
} |
| 259 |
|
| 260 |
static $count = 0; |
| 261 |
$post_id ??= get_the_ID(); |
| 262 |
$id_to_pass = is_int($post_id) ? $post_id : 0; |
| 263 |
|
| 264 |
// Use regex for high performance, but careful with HTML structure |
| 265 |
return preg_replace_callback('/<img([^>]+)>/i', function ($matches) use ($settings, &$count, $id_to_pass) { |
| 266 |
$count++; |
| 267 |
$img_tag = $matches[0]; |
| 268 |
$attributes_str = $matches[1]; |
| 269 |
|
| 270 |
// Parse attributes (keys lower-cased; quoted and unquoted values supported) |
| 271 |
$attributes = $this->parse_attributes($attributes_str); |
| 272 |
|
| 273 |
$alt_missing = !empty($settings['add_missing_alt']) && trim((string) ($attributes['alt'] ?? '')) === ''; |
| 274 |
$title_missing = !empty($settings['add_missing_title']) && trim((string) ($attributes['title'] ?? '')) === ''; |
| 275 |
|
| 276 |
// Nothing to inject on this image — skip before any source/attachment work. |
| 277 |
if (!$alt_missing && !$title_missing) { |
| 278 |
return $img_tag; |
| 279 |
} |
| 280 |
|
| 281 |
// Resolve the real image source. Lazy-load markup keeps the true URL in a |
| 282 |
// data-* attribute while `src` is empty or a placeholder/data-URI. |
| 283 |
$src = $this->resolve_image_src($attributes); |
| 284 |
|
| 285 |
// No resolvable source (spacers, tracking pixels, pure placeholders) — nothing |
| 286 |
// meaningful to describe, and nothing to derive %filename% from either. |
| 287 |
if ($src === '') { |
| 288 |
return $img_tag; |
| 289 |
} |
| 290 |
|
| 291 |
// Resolve the attachment ID only when a format that we're about to apply |
| 292 |
// actually references attachment metadata (%image_title% / %image_caption%). |
| 293 |
// attachment_url_to_postid() is a DB query, so avoid it for the common |
| 294 |
// filename-based formats and for images that need no injection. |
| 295 |
$needs_attachment = |
| 296 |
($alt_missing && $this->format_uses_attachment($settings['alt_format'] ?? '')) || |
| 297 |
($title_missing && $this->format_uses_attachment($settings['title_format'] ?? '')); |
| 298 |
$attachment_id = $needs_attachment |
| 299 |
? $this->url_to_attachment_id($src, (string) ($attributes['class'] ?? '')) |
| 300 |
: 0; |
| 301 |
|
| 302 |
// Handle ALT attribute |
| 303 |
if ($alt_missing) { |
| 304 |
$alt_val = $this->generate_attribute_value($settings['alt_format'] ?? '', $attachment_id, $id_to_pass, $count, $src); |
| 305 |
if ($alt_val !== '') { |
| 306 |
$img_tag = $this->inject_attribute($img_tag, 'alt', $alt_val, isset($attributes['alt'])); |
| 307 |
} |
| 308 |
} |
| 309 |
|
| 310 |
// Handle TITLE attribute |
| 311 |
if ($title_missing) { |
| 312 |
$title_val = $this->generate_attribute_value($settings['title_format'] ?? '', $attachment_id, $id_to_pass, $count, $src); |
| 313 |
if ($title_val !== '') { |
| 314 |
$img_tag = $this->inject_attribute($img_tag, 'title', $title_val, isset($attributes['title'])); |
| 315 |
} |
| 316 |
} |
| 317 |
|
| 318 |
return $img_tag; |
| 319 |
}, $content); |
| 320 |
} |
| 321 |
|
| 322 |
/** |
| 323 |
* Parse an <img> attribute string into a lower-cased key => value map. |
| 324 |
* |
| 325 |
* Handles double-quoted, single-quoted and unquoted attribute values so that |
| 326 |
* existing attributes (e.g. an unquoted `alt=Something`) are correctly detected |
| 327 |
* and not duplicated. Attribute names are normalised to lower-case so uppercase |
| 328 |
* markup (`SRC=`, `ALT=`) is recognised. |
| 329 |
* |
| 330 |
* @since 1.19.1 |
| 331 |
* @param string $attributes_str The raw attribute portion of the tag. |
| 332 |
* @return array<string,string> Lower-cased attribute name => value. |
| 333 |
*/ |
| 334 |
private function parse_attributes(string $attributes_str): array { |
| 335 |
$matched = preg_match_all( |
| 336 |
'/([a-zA-Z][a-zA-Z0-9:-]*)\s*=\s*(?:"([^"]*)"|\'([^\']*)\'|([^\s"\'>]+))/', |
| 337 |
$attributes_str, |
| 338 |
$matches, |
| 339 |
PREG_SET_ORDER |
| 340 |
); |
| 341 |
|
| 342 |
if (!$matched) { |
| 343 |
return []; |
| 344 |
} |
| 345 |
|
| 346 |
$attributes = []; |
| 347 |
foreach ($matches as $m) { |
| 348 |
$key = strtolower($m[1]); |
| 349 |
|
| 350 |
if (isset($m[2]) && $m[2] !== '') { |
| 351 |
$value = $m[2]; |
| 352 |
} elseif (isset($m[3]) && $m[3] !== '') { |
| 353 |
$value = $m[3]; |
| 354 |
} elseif (isset($m[4]) && $m[4] !== '') { |
| 355 |
$value = $m[4]; |
| 356 |
} else { |
| 357 |
$value = ''; |
| 358 |
} |
| 359 |
|
| 360 |
$attributes[$key] = $value; |
| 361 |
} |
| 362 |
|
| 363 |
return $attributes; |
| 364 |
} |
| 365 |
|
| 366 |
/** |
| 367 |
* Resolve a usable image source from the parsed attributes. |
| 368 |
* |
| 369 |
* Prefers `src`, but falls back to common lazy-load attributes when `src` is |
| 370 |
* empty or a `data:` URI placeholder, so generated alt/title reflect the real |
| 371 |
* image rather than a base64 blob. |
| 372 |
* |
| 373 |
* @since 1.19.1 |
| 374 |
* @param array<string,string> $attributes Parsed attributes. |
| 375 |
* @return string The resolved source URL, or '' if none is usable. |
| 376 |
*/ |
| 377 |
private function resolve_image_src(array $attributes): string { |
| 378 |
$candidates = ['src', 'data-src', 'data-lazy-src', 'data-original', 'data-lazy']; |
| 379 |
|
| 380 |
foreach ($candidates as $attr) { |
| 381 |
$value = trim((string) ($attributes[$attr] ?? '')); |
| 382 |
|
| 383 |
if ($value === '' || stripos($value, 'data:') === 0) { |
| 384 |
continue; |
| 385 |
} |
| 386 |
|
| 387 |
return $value; |
| 388 |
} |
| 389 |
|
| 390 |
return ''; |
| 391 |
} |
| 392 |
|
| 393 |
/** |
| 394 |
* Whether a format string references attachment-only metadata tokens. |
| 395 |
* |
| 396 |
* Used to decide if an attachment lookup (a DB query) is actually needed; |
| 397 |
* filename/site/title/count tokens do not require the attachment record. |
| 398 |
* |
| 399 |
* @since 1.19.1 |
| 400 |
* @param string $format The format string. |
| 401 |
* @return bool |
| 402 |
*/ |
| 403 |
private function format_uses_attachment(string $format): bool { |
| 404 |
return strpos($format, '%image_title%') !== false |
| 405 |
|| strpos($format, '%image_caption%') !== false; |
| 406 |
} |
| 407 |
|
| 408 |
/** |
| 409 |
* Resolve an attachment ID from a source URL. |
| 410 |
* |
| 411 |
* Body images are almost always inserted at a generated size, which |
| 412 |
* `attachment_url_to_postid()` cannot match, so %image_title% and |
| 413 |
* %image_caption% resolved to nothing for them. Attachment_Lookup reads the |
| 414 |
* `wp-image-{ID}` class the editor wrote first, and caches whatever still |
| 415 |
* has to be asked of the database (#847). |
| 416 |
* |
| 417 |
* @since 1.19.1 |
| 418 |
* @param string $src Source URL. |
| 419 |
* @param string $classes The image's class attribute, for its wp-image-{ID}. |
| 420 |
* @return int Attachment ID, or 0 if not a media-library image. |
| 421 |
*/ |
| 422 |
private function url_to_attachment_id(string $src, string $classes = ''): int { |
| 423 |
return Attachment_Lookup::id_from_url($src, Attachment_Lookup::hint_from_markup($classes)); |
| 424 |
} |
| 425 |
|
| 426 |
/** |
| 427 |
* Inject (or replace an empty) alt/title attribute on a single <img> tag. |
| 428 |
* |
| 429 |
* When replacing, the pattern is anchored to a whitespace/tag boundary and |
| 430 |
* limited to one occurrence so it can never clobber a `data-alt`/`data-title` |
| 431 |
* (or any `*-alt`/`*-title`) attribute. A callback is used for the replacement |
| 432 |
* so `$` / `\` in the value are never treated as backreferences. Insertion is |
| 433 |
* case-insensitive on the tag opener so uppercase `<IMG>` is handled. |
| 434 |
* |
| 435 |
* @since 1.19.1 |
| 436 |
* @param string $img_tag The full <img> tag. |
| 437 |
* @param string $name Attribute name ('alt' or 'title'). |
| 438 |
* @param string $value Unescaped attribute value. |
| 439 |
* @param bool $replace Whether an (empty) attribute already exists to replace. |
| 440 |
* @return string The modified tag. |
| 441 |
*/ |
| 442 |
private function inject_attribute(string $img_tag, string $name, string $value, bool $replace): string { |
| 443 |
$attr = $name . '="' . esc_attr($value) . '"'; |
| 444 |
|
| 445 |
if ($replace) { |
| 446 |
return preg_replace_callback( |
| 447 |
'/(^|\s)' . preg_quote($name, '/') . '\s*=\s*(["\'])[^"\']*\2/i', |
| 448 |
static function ($m) use ($attr) { |
| 449 |
return $m[1] . $attr; |
| 450 |
}, |
| 451 |
$img_tag, |
| 452 |
1 |
| 453 |
); |
| 454 |
} |
| 455 |
|
| 456 |
return preg_replace_callback( |
| 457 |
'/<img\b/i', |
| 458 |
static function ($m) use ($attr) { |
| 459 |
return $m[0] . ' ' . $attr; |
| 460 |
}, |
| 461 |
$img_tag, |
| 462 |
1 |
| 463 |
); |
| 464 |
} |
| 465 |
|
| 466 |
/** |
| 467 |
* Generate attribute value based on format and context |
| 468 |
* |
| 469 |
* @since 1.0.0 |
| 470 |
* @param string $format The format string |
| 471 |
* @param int $attachment_id Attachment ID |
| 472 |
* @param int $post_id Current Post ID |
| 473 |
* @param int $count Image counter |
| 474 |
* @param string $src Image source URL |
| 475 |
* @return string Generated value |
| 476 |
*/ |
| 477 |
private function generate_attribute_value(string $format, int $attachment_id, int $post_id, int $count, string $src): string { |
| 478 |
$replacements = [ |
| 479 |
'%site_title%' => get_bloginfo('name'), |
| 480 |
'%sitename%' => get_bloginfo('name'), |
| 481 |
// Empty, not the site name. The segment collapsing below drops an |
| 482 |
// unresolved token together with its separator, and the default |
| 483 |
// title_format already ends in %sitename% — substituting the site |
| 484 |
// name here printed it twice ("Site Name | Site Name") on every |
| 485 |
// image processed outside the loop (widgets, page builders, FSE). |
| 486 |
'%title%' => $post_id > 0 ? get_the_title($post_id) : '', |
| 487 |
'%count%' => (string) $count, |
| 488 |
'%filename%' => '', |
| 489 |
'%image_title%' => '', |
| 490 |
'%image_caption%' => '', |
| 491 |
]; |
| 492 |
|
| 493 |
// Get filename from src |
| 494 |
if ($src) { |
| 495 |
$filename = pathinfo($src, PATHINFO_FILENAME); |
| 496 |
$replacements['%filename%'] = str_replace(['-', '_'], ' ', $filename); |
| 497 |
} |
| 498 |
|
| 499 |
// Get attachment data if ID exists |
| 500 |
if ($attachment_id) { |
| 501 |
$attachment = get_post($attachment_id); |
| 502 |
if ($attachment) { |
| 503 |
$replacements['%image_title%'] = $attachment->post_title; |
| 504 |
$replacements['%image_caption%'] = $attachment->post_excerpt; |
| 505 |
} |
| 506 |
} |
| 507 |
|
| 508 |
// Apply replacements for every token except the separator. |
| 509 |
$value = str_replace(array_keys($replacements), array_values($replacements), $format); |
| 510 |
|
| 511 |
// Split on the separator tokens, drop segments that resolved to empty, then |
| 512 |
// re-join with the separator symbol. This prevents orphaned/leading/trailing |
| 513 |
// separators such as "| Site Name" when a token (e.g. %filename%) is empty. |
| 514 |
$segments = preg_split('/%sep(?:arator)?%/', $value); |
| 515 |
$segments = array_filter( |
| 516 |
array_map('trim', $segments), |
| 517 |
static function ($segment) { |
| 518 |
return $segment !== ''; |
| 519 |
} |
| 520 |
); |
| 521 |
$value = implode(' ' . $this->get_separator() . ' ', $segments); |
| 522 |
|
| 523 |
// Clean up double spaces if any |
| 524 |
$value = preg_replace('/\s+/', ' ', $value); |
| 525 |
|
| 526 |
return trim($value); |
| 527 |
} |
| 528 |
|
| 529 |
/** |
| 530 |
* Get site separator |
| 531 |
* |
| 532 |
* @since 1.0.0 |
| 533 |
* @return string |
| 534 |
*/ |
| 535 |
private function get_separator(): string { |
| 536 |
if ($this->separator === null) { |
| 537 |
$this->separator = Site_Identity_Manager::get_active_separator_symbol(); |
| 538 |
} |
| 539 |
return $this->separator; |
| 540 |
} |
| 541 |
|
| 542 |
// ───────────────────────────────────────────────────────────────────── |
| 543 |
// Media Library alt-text persistence (writes _wp_attachment_image_alt) |
| 544 |
// ───────────────────────────────────────────────────────────────────── |
| 545 |
|
| 546 |
/** |
| 547 |
* Generate and save alt text onto a single attachment's Media Library record. |
| 548 |
* |
| 549 |
* Uses the same `alt_format` token pipeline as output injection, so the value |
| 550 |
* matches what the front-end filter would have produced. In this context |
| 551 |
* `%title%` and `%image_title%` resolve to the attachment's own title. |
| 552 |
* |
| 553 |
* @since 1.19.1 |
| 554 |
* @param int $attachment_id The attachment ID. |
| 555 |
* @param bool $overwrite When false, images that already have alt text are left untouched. |
| 556 |
* @param bool $purge Clear the caches that already rendered this image. Pass false only |
| 557 |
* when the attachment cannot yet appear on any page. |
| 558 |
* @return bool True when the attachment now has the generated alt text; false when skipped or on failure. |
| 559 |
*/ |
| 560 |
public function fill_attachment_alt(int $attachment_id, bool $overwrite = false, bool $purge = true): bool { |
| 561 |
if (!wp_attachment_is_image($attachment_id)) { |
| 562 |
return false; |
| 563 |
} |
| 564 |
|
| 565 |
$existing = (string) get_post_meta($attachment_id, '_wp_attachment_image_alt', true); |
| 566 |
|
| 567 |
// Non-destructive by default: never clobber hand-written alt text. |
| 568 |
if (!$overwrite && trim($existing) !== '') { |
| 569 |
return false; |
| 570 |
} |
| 571 |
|
| 572 |
$settings = $this->get_settings('site'); |
| 573 |
$format = $settings['alt_format'] ?? '%filename%'; |
| 574 |
$src = (string) wp_get_attachment_url($attachment_id); |
| 575 |
|
| 576 |
$value = ''; |
| 577 |
|
| 578 |
// AI describes the picture; the template can only rewrite its filename. |
| 579 |
// Falls back to the template on any failure so a provider outage |
| 580 |
// degrades to the old behaviour instead of leaving images bare. |
| 581 |
if ('ai' === ($settings['alt_source'] ?? 'template')) { |
| 582 |
$value = $this->generate_ai_alt($attachment_id); |
| 583 |
} |
| 584 |
|
| 585 |
if ('' === $value) { |
| 586 |
// Pass the attachment ID as the post context so %title% falls back to the |
| 587 |
// attachment's own title (there is no surrounding post here). |
| 588 |
$value = sanitize_text_field( |
| 589 |
$this->generate_attribute_value($format, $attachment_id, $attachment_id, 0, $src) |
| 590 |
); |
| 591 |
} |
| 592 |
|
| 593 |
if ($value === '') { |
| 594 |
return false; |
| 595 |
} |
| 596 |
|
| 597 |
if ($existing === $value) { |
| 598 |
// Already correct — treat as success without a redundant write. |
| 599 |
return true; |
| 600 |
} |
| 601 |
|
| 602 |
$written = update_post_meta($attachment_id, '_wp_attachment_image_alt', $value) !== false; |
| 603 |
|
| 604 |
if ($written) { |
| 605 |
$this->record_alt_write($attachment_id, $purge); |
| 606 |
} |
| 607 |
|
| 608 |
return $written; |
| 609 |
} |
| 610 |
|
| 611 |
/** |
| 612 |
* Note that an attachment's alt text changed, and clear what rendered it. |
| 613 |
* |
| 614 |
* @since 2.12.0 |
| 615 |
* @param int $attachment_id Attachment whose alt text was just written. |
| 616 |
* @param bool $purge False when nothing can be displaying this attachment yet. |
| 617 |
* @return void |
| 618 |
*/ |
| 619 |
protected function record_alt_write(int $attachment_id, bool $purge = true): void { |
| 620 |
// Recorded during a bulk run whatever $purge says: the batch resolves |
| 621 |
// the whole list at the end, and bulk_fill_missing_alt() counts its |
| 622 |
// writes from this list. |
| 623 |
if (null !== $this->deferred_alt_purge) { |
| 624 |
$this->deferred_alt_purge[] = $attachment_id; |
| 625 |
|
| 626 |
return; |
| 627 |
} |
| 628 |
|
| 629 |
// A fresh upload cannot be on any page, so there is nothing to clear |
| 630 |
// and nothing for thinkrank_image_alt_updated to report. Listeners |
| 631 |
// that care about new files already have core's add_attachment. |
| 632 |
if (!$purge) { |
| 633 |
return; |
| 634 |
} |
| 635 |
|
| 636 |
$this->purge_alt_caches([$attachment_id]); |
| 637 |
} |
| 638 |
|
| 639 |
/** |
| 640 |
* Drop cached renderings of every page that shows these attachments. |
| 641 |
* |
| 642 |
* Alt text is written to post meta but read out of HTML that other |
| 643 |
* software has already rendered and stored, so the write alone changes |
| 644 |
* nothing a visitor sees. Elementor keeps rendered widgets for 24 hours by |
| 645 |
* default and page caches keep whole documents, which is how an agent came |
| 646 |
* to report a successful alt-text write against a page still showing the |
| 647 |
* old words (#763). |
| 648 |
* |
| 649 |
* @since 2.12.0 |
| 650 |
* @param int[] $attachment_ids Attachments whose alt text changed. |
| 651 |
* @return array{posts: int[], warnings: string[]} Posts purged, and caches left for the user to clear. |
| 652 |
*/ |
| 653 |
public function purge_alt_caches(array $attachment_ids): array { |
| 654 |
$ids = array_values(array_unique(array_filter(array_map('intval', $attachment_ids)))); |
| 655 |
|
| 656 |
if ([] === $ids) { |
| 657 |
$this->last_alt_cache_purge = ['posts' => [], 'warnings' => []]; |
| 658 |
|
| 659 |
return $this->last_alt_cache_purge; |
| 660 |
} |
| 661 |
|
| 662 |
$posts = Attachment_Usage::posts_using($ids); |
| 663 |
$purged = Cache_Purger::purge_posts($posts); |
| 664 |
|
| 665 |
foreach ($ids as $id) { |
| 666 |
/** |
| 667 |
* Fires after ThinkRank writes an image's alt text and clears the |
| 668 |
* caches it knows about. |
| 669 |
* |
| 670 |
* Builders and cache layers ThinkRank does not handle can listen |
| 671 |
* here to drop their own rendering of the affected posts. |
| 672 |
* |
| 673 |
* @since 2.12.0 |
| 674 |
* |
| 675 |
* @param int $attachment_id The attachment whose alt text changed. |
| 676 |
* @param int[] $post_ids Posts found to display that attachment. |
| 677 |
*/ |
| 678 |
do_action('thinkrank_image_alt_updated', $id, $posts); |
| 679 |
} |
| 680 |
|
| 681 |
$this->last_alt_cache_purge = [ |
| 682 |
'posts' => $purged, |
| 683 |
'warnings' => Cache_Purger::warnings(), |
| 684 |
]; |
| 685 |
|
| 686 |
return $this->last_alt_cache_purge; |
| 687 |
} |
| 688 |
|
| 689 |
/** |
| 690 |
* The outcome of the most recent alt-text cache purge in this request. |
| 691 |
* |
| 692 |
* @since 2.12.0 |
| 693 |
* @return array{posts: int[], warnings: string[]} |
| 694 |
*/ |
| 695 |
public function last_alt_cache_purge(): array { |
| 696 |
return $this->last_alt_cache_purge; |
| 697 |
} |
| 698 |
|
| 699 |
/** |
| 700 |
* Describe an attachment with the vision model. |
| 701 |
* |
| 702 |
* Never throws: alt text generation runs in batches over a whole media |
| 703 |
* library, and one unreadable image or a rate-limit blip must not abort |
| 704 |
* the run. Returns '' so the caller falls back to the template. |
| 705 |
* |
| 706 |
* @since 1.28.0 |
| 707 |
* @param int $attachment_id Attachment to describe. |
| 708 |
* @return string Alt text, or '' when unavailable. |
| 709 |
*/ |
| 710 |
private function generate_ai_alt(int $attachment_id): string { |
| 711 |
try { |
| 712 |
$vision = new \ThinkRank\AI\Vision_Client(); |
| 713 |
|
| 714 |
if (!$vision->is_available()) { |
| 715 |
return ''; |
| 716 |
} |
| 717 |
|
| 718 |
// The parent post's title disambiguates images that are visually |
| 719 |
// ambiguous on their own (a generic chart, a product on white). |
| 720 |
$context = ''; |
| 721 |
$parent = (int) get_post_field('post_parent', $attachment_id); |
| 722 |
if ($parent > 0) { |
| 723 |
$context = (string) get_the_title($parent); |
| 724 |
} |
| 725 |
|
| 726 |
return sanitize_text_field($vision->describe_attachment($attachment_id, $context)); |
| 727 |
} catch (\Throwable $e) { |
| 728 |
return ''; |
| 729 |
} |
| 730 |
} |
| 731 |
|
| 732 |
/** |
| 733 |
* Fill alt text across the Media Library in a single batch. |
| 734 |
* |
| 735 |
* Iterates images by ascending ID using offset/limit so callers can page |
| 736 |
* through large libraries without exhausting memory or hitting timeouts. |
| 737 |
* |
| 738 |
* @since 1.19.1 |
| 739 |
* @param array $args { |
| 740 |
* @type int $offset Starting offset into the image set. Default 0. |
| 741 |
* @type int $limit Batch size (clamped 1–200). Default 50. |
| 742 |
* @type bool $overwrite Overwrite existing alt text. Default false. |
| 743 |
* } |
| 744 |
* @return array { |
| 745 |
* @type int $total Total images in the library. |
| 746 |
* @type int $processed Images looked at in this batch. |
| 747 |
* @type int $updated Images whose alt text was written. |
| 748 |
* @type int $skipped Images left unchanged (already had alt / no value). |
| 749 |
* @type int $offset The offset this batch started at. |
| 750 |
* @type int $next_offset The offset to pass for the next batch. |
| 751 |
* @type int $remaining Images still to process after this batch. |
| 752 |
* @type bool $done True when the whole library has been processed. |
| 753 |
* } |
| 754 |
*/ |
| 755 |
public function bulk_fill_missing_alt(array $args = []): array { |
| 756 |
$offset = max(0, (int) ($args['offset'] ?? 0)); |
| 757 |
$limit = min(200, max(1, (int) ($args['limit'] ?? 50))); |
| 758 |
$overwrite = !empty($args['overwrite']); |
| 759 |
|
| 760 |
// In AI mode every image is a paid provider call that takes seconds, |
| 761 |
// so a 200-image batch would both surprise the user's bill and blow |
| 762 |
// past max_execution_time. Cap the batch and let the caller page — |
| 763 |
// `remaining` already drives that loop. |
| 764 |
if ('ai' === ($this->get_settings('site')['alt_source'] ?? 'template')) { |
| 765 |
$limit = min($limit, self::AI_BATCH_LIMIT); |
| 766 |
} |
| 767 |
|
| 768 |
$total = $this->count_images(); |
| 769 |
|
| 770 |
$ids = get_posts([ |
| 771 |
'post_type' => 'attachment', |
| 772 |
'post_mime_type' => 'image', |
| 773 |
// Must cover the same set count_images() counts, or the pager can |
| 774 |
// never reach the total. 'inherit' alone excluded private-status |
| 775 |
// attachments — which media-protection and membership plugins do |
| 776 |
// create — while count_images() still counted them (#322). |
| 777 |
'post_status' => ['inherit', 'private', 'publish', 'draft', 'pending', 'future'], |
| 778 |
'numberposts' => $limit, |
| 779 |
'offset' => $offset, |
| 780 |
'fields' => 'ids', |
| 781 |
'orderby' => 'ID', |
| 782 |
'order' => 'ASC', |
| 783 |
// phpcs:ignore WordPressVIPMinimum.Performance.WPQueryParams.SuppressFilters_suppress_filters -- The pager must walk the same unfiltered set count_images() counts, or it can never reach the total (#322). |
| 784 |
'suppress_filters' => true, |
| 785 |
]); |
| 786 |
|
| 787 |
$processed = 0; |
| 788 |
|
| 789 |
// Hold every purge until the batch is done: see $deferred_alt_purge. |
| 790 |
$this->deferred_alt_purge = []; |
| 791 |
|
| 792 |
try { |
| 793 |
foreach ($ids as $id) { |
| 794 |
$processed++; |
| 795 |
$this->fill_attachment_alt((int) $id, $overwrite); |
| 796 |
} |
| 797 |
} finally { |
| 798 |
$written = array_values(array_unique($this->deferred_alt_purge ?? [])); |
| 799 |
$this->deferred_alt_purge = null; |
| 800 |
$cache = $this->purge_alt_caches($written); |
| 801 |
} |
| 802 |
|
| 803 |
// Counted from the writes that actually happened, not from what |
| 804 |
// fill_attachment_alt() returned. It answers true for an image whose |
| 805 |
// stored alt already equals the generated value, which is the right |
| 806 |
// answer to "does this image have its alt text?" and the wrong one to |
| 807 |
// "how many did you change" — a re-run over a correct library reported |
| 808 |
// every image as updated while writing nothing. `updated` and |
| 809 |
// `skipped` have documented the write, not the return value, since |
| 810 |
// this method was added. |
| 811 |
$updated = count($written); |
| 812 |
$skipped = max(0, $processed - $updated); |
| 813 |
|
| 814 |
$next_offset = $offset + count($ids); |
| 815 |
|
| 816 |
// An empty batch means there is nothing left to walk, whatever the |
| 817 |
// total claims. Deriving `done` from the count alone let any drift |
| 818 |
// between the two queries strand the caller on a batch that could |
| 819 |
// never advance the offset, and the admin UI answers that by |
| 820 |
// re-requesting up to 10,000 times. |
| 821 |
$exhausted = empty($ids); |
| 822 |
$remaining = $exhausted ? 0 : max(0, $total - $next_offset); |
| 823 |
|
| 824 |
// Bulk writes change the Site SEO Analyzer's "images have alt text" coverage. |
| 825 |
if ($updated > 0) { |
| 826 |
$this->flush_analyzer_cache(); |
| 827 |
} |
| 828 |
|
| 829 |
return [ |
| 830 |
'total' => $total, |
| 831 |
'processed' => $processed, |
| 832 |
'updated' => $updated, |
| 833 |
'skipped' => $skipped, |
| 834 |
'offset' => $offset, |
| 835 |
'next_offset' => $next_offset, |
| 836 |
'remaining' => $remaining, |
| 837 |
'done' => $exhausted || $next_offset >= $total, |
| 838 |
'cache' => [ |
| 839 |
'posts_purged' => count($cache['posts']), |
| 840 |
'warnings' => $cache['warnings'], |
| 841 |
], |
| 842 |
]; |
| 843 |
} |
| 844 |
|
| 845 |
/** |
| 846 |
* Media Library alt-text coverage stats for the settings UI. |
| 847 |
* |
| 848 |
* @since 1.19.1 |
| 849 |
* @return array{total:int,with_alt:int,missing:int} |
| 850 |
*/ |
| 851 |
public function get_media_alt_stats(): array { |
| 852 |
$total = $this->count_images(); |
| 853 |
$with_alt = $this->count_images_with_alt(); |
| 854 |
|
| 855 |
return [ |
| 856 |
'total' => $total, |
| 857 |
'with_alt' => $with_alt, |
| 858 |
'missing' => max(0, $total - $with_alt), |
| 859 |
]; |
| 860 |
} |
| 861 |
|
| 862 |
/** |
| 863 |
* Auto-fill hook target — save alt text for a freshly uploaded image. |
| 864 |
* |
| 865 |
* Gated by the `save_alt_to_media` + `auto_fill_on_upload` settings so it is a |
| 866 |
* no-op unless the feature is enabled. Respects the overwrite preference. |
| 867 |
* |
| 868 |
* @since 1.19.1 |
| 869 |
* @param int $attachment_id The newly created attachment ID. |
| 870 |
* @return void |
| 871 |
*/ |
| 872 |
public function maybe_auto_fill_on_upload(int $attachment_id): void { |
| 873 |
$settings = $this->get_settings('site'); |
| 874 |
|
| 875 |
if (empty($settings['save_alt_to_media']) || empty($settings['auto_fill_on_upload'])) { |
| 876 |
return; |
| 877 |
} |
| 878 |
|
| 879 |
if (!wp_attachment_is_image($attachment_id)) { |
| 880 |
return; |
| 881 |
} |
| 882 |
|
| 883 |
// No cache purge: this fires on add_attachment, so the file was created |
| 884 |
// seconds ago and no page can be displaying it yet. Skipping the lookup |
| 885 |
// keeps a bulk media import off two unindexed postmeta scans per file. |
| 886 |
$this->fill_attachment_alt($attachment_id, !empty($settings['media_alt_overwrite']), false); |
| 887 |
} |
| 888 |
|
| 889 |
/** |
| 890 |
* Total number of image attachments in the library. |
| 891 |
* |
| 892 |
* Counted with an explicit `post_status != 'trash'` rather than through |
| 893 |
* wp_count_attachments(). The helper applies that filter internally, which |
| 894 |
* looked equivalent — but it left the two halves of get_media_alt_stats() |
| 895 |
* with different notions of which images exist, and only one of them said |
| 896 |
* so out loud. Spelling the filter out here keeps this query and |
| 897 |
* count_images_with_alt() visibly in step (#321). |
| 898 |
* |
| 899 |
* @since 1.19.1 |
| 900 |
* @return int |
| 901 |
*/ |
| 902 |
private function count_images(): int { |
| 903 |
// phpcs:disable WordPress.DB.DirectDatabaseQuery.DirectQuery, WordPress.DB.DirectDatabaseQuery.NoCaching, WordPress.DB.PreparedSQL.InterpolatedNotPrepared -- indexed COUNT; short-lived admin action |
| 904 |
return (int) $this->wpdb->get_var( |
| 905 |
"SELECT COUNT(*) FROM {$this->wpdb->posts} |
| 906 |
WHERE post_type = 'attachment' |
| 907 |
AND post_mime_type LIKE 'image/%' |
| 908 |
AND post_status != 'trash'" |
| 909 |
); |
| 910 |
// phpcs:enable WordPress.DB.DirectDatabaseQuery.DirectQuery, WordPress.DB.DirectDatabaseQuery.NoCaching, WordPress.DB.PreparedSQL.InterpolatedNotPrepared |
| 911 |
} |
| 912 |
|
| 913 |
/** |
| 914 |
* Number of image attachments that already have non-empty alt text. |
| 915 |
* |
| 916 |
* Carries the same `post_status != 'trash'` filter as count_images(), so a |
| 917 |
* trashed image can never be counted as covered against a total it is not |
| 918 |
* part of. Matches the Site SEO Analyzer's alt-text check, which applies |
| 919 |
* the same filter to both of its counts. |
| 920 |
* |
| 921 |
* @since 1.19.1 |
| 922 |
* @return int |
| 923 |
*/ |
| 924 |
private function count_images_with_alt(): int { |
| 925 |
// phpcs:disable WordPress.DB.DirectDatabaseQuery.DirectQuery, WordPress.DB.DirectDatabaseQuery.NoCaching, WordPress.DB.PreparedSQL.InterpolatedNotPrepared -- indexed COUNT via postmeta meta_key index; short-lived admin action |
| 926 |
return (int) $this->wpdb->get_var( |
| 927 |
"SELECT COUNT(DISTINCT p.ID) FROM {$this->wpdb->posts} p |
| 928 |
INNER JOIN {$this->wpdb->postmeta} pm |
| 929 |
ON pm.post_id = p.ID |
| 930 |
AND pm.meta_key = '_wp_attachment_image_alt' |
| 931 |
AND pm.meta_value != '' |
| 932 |
WHERE p.post_type = 'attachment' |
| 933 |
AND p.post_mime_type LIKE 'image/%' |
| 934 |
AND p.post_status != 'trash'" |
| 935 |
); |
| 936 |
// phpcs:enable WordPress.DB.DirectDatabaseQuery.DirectQuery, WordPress.DB.DirectDatabaseQuery.NoCaching, WordPress.DB.PreparedSQL.InterpolatedNotPrepared |
| 937 |
} |
| 938 |
|
| 939 |
/** |
| 940 |
* Bust the Site SEO Analyzer's cached result so its alt-text coverage refreshes. |
| 941 |
* |
| 942 |
* Uses the analyzer's transient key directly to avoid instantiating it here. |
| 943 |
* |
| 944 |
* @since 1.19.1 |
| 945 |
* @return void |
| 946 |
*/ |
| 947 |
private function flush_analyzer_cache(): void { |
| 948 |
// Ask the analyzer rather than duplicating its transient key here — the |
| 949 |
// literal drifted out of sync the moment anyone renamed it. |
| 950 |
if (class_exists('ThinkRank\\SEO\\SEO_Analyzer')) { |
| 951 |
(new SEO_Analyzer())->flush_cache(); |
| 952 |
} |
| 953 |
} |
| 954 |
} |
| 955 |
|
