| 1 |
<?php |
| 2 |
/** |
| 3 |
* Metasync_Dimension_Injector |
| 4 |
* Scans HTML for <img> tags missing width/height and injects them |
| 5 |
* based on file metadata to prevent Cumulative Layout Shift (CLS). |
| 6 |
* |
| 7 |
* @package Search Atlas SEO |
| 8 |
* @copyright Copyright (C) 2021-2025, Search Atlas Group - [email protected] |
| 9 |
* @since 2.6.0 |
| 10 |
*/ |
| 11 |
|
| 12 |
if (!defined('ABSPATH')) { |
| 13 |
exit; |
| 14 |
} |
| 15 |
|
| 16 |
class Metasync_Dimension_Injector { |
| 17 |
|
| 18 |
/** In-memory cache to avoid repeated file lookups within a request. */ |
| 19 |
private array $cache = []; |
| 20 |
|
| 21 |
/** Transient key prefix for persisted image dimensions. */ |
| 22 |
private const CACHE_PREFIX = 'ms_dim_'; |
| 23 |
|
| 24 |
/** How long resolved dimensions stay cached, in seconds (7 days). */ |
| 25 |
private const CACHE_TTL = 604800; |
| 26 |
|
| 27 |
/** Transient key prefix for failed remote lookups (negative cache). */ |
| 28 |
private const FAIL_PREFIX = 'ms_dimfail_'; |
| 29 |
|
| 30 |
/** How long a failed remote lookup stays cached, in seconds (1 hour). */ |
| 31 |
private const FAIL_TTL = 3600; |
| 32 |
|
| 33 |
/** Cap on bytes fetched from a remote image header sniff. */ |
| 34 |
private const REMOTE_MAX_BYTES = 65536; |
| 35 |
|
| 36 |
public function __construct() { |
| 37 |
add_filter('the_content', [$this, 'inject_dimensions'], 30); |
| 38 |
add_filter('post_thumbnail_html', [$this, 'inject_dimensions'], 30); |
| 39 |
add_filter('widget_text', [$this, 'inject_dimensions'], 30); |
| 40 |
} |
| 41 |
|
| 42 |
/** |
| 43 |
* Find <img> tags missing width or height and inject dimensions. |
| 44 |
*/ |
| 45 |
public function inject_dimensions(string $content): string { |
| 46 |
if (empty($content) || is_admin() || is_feed()) { |
| 47 |
return $content; |
| 48 |
} |
| 49 |
|
| 50 |
return preg_replace_callback('/<img\s[^>]+>/i', function ($matches) { |
| 51 |
$tag = $matches[0]; |
| 52 |
|
| 53 |
$has_width = preg_match('/\swidth\s*=/i', $tag); |
| 54 |
$has_height = preg_match('/\sheight\s*=/i', $tag); |
| 55 |
|
| 56 |
// Both already present - nothing to do |
| 57 |
if ($has_width && $has_height) { |
| 58 |
return $tag; |
| 59 |
} |
| 60 |
|
| 61 |
$dims = $this->get_dimensions($tag); |
| 62 |
if (!$dims) { |
| 63 |
return $tag; |
| 64 |
} |
| 65 |
|
| 66 |
if (!$has_width) { |
| 67 |
$tag = $this->add_attribute($tag, 'width', (string) $dims['width']); |
| 68 |
} |
| 69 |
if (!$has_height) { |
| 70 |
$tag = $this->add_attribute($tag, 'height', (string) $dims['height']); |
| 71 |
} |
| 72 |
|
| 73 |
return $tag; |
| 74 |
}, $content); |
| 75 |
} |
| 76 |
|
| 77 |
/** |
| 78 |
* Try to determine image dimensions from multiple sources. |
| 79 |
*/ |
| 80 |
private function get_dimensions(string $img_tag): ?array { |
| 81 |
if (!preg_match('/src=["\']([^"\']+)["\']/i', $img_tag, $m)) { |
| 82 |
return null; |
| 83 |
} |
| 84 |
$src = $m[1]; |
| 85 |
|
| 86 |
// Request-scoped cache: avoids repeated work within a single render. |
| 87 |
if (isset($this->cache[$src])) { |
| 88 |
return $this->cache[$src]; |
| 89 |
} |
| 90 |
|
| 91 |
// Persistent cache: skips disk I/O for images resolved on a prior request. |
| 92 |
$cached = $this->get_cached_dimensions($src); |
| 93 |
if ($cached !== null) { |
| 94 |
$this->cache[$src] = $cached; |
| 95 |
return $cached; |
| 96 |
} |
| 97 |
|
| 98 |
$dims = null; |
| 99 |
|
| 100 |
// Strategy 1: Try to find WordPress attachment by URL |
| 101 |
$dims = $this->get_dims_from_attachment($src); |
| 102 |
|
| 103 |
// Strategy 2: Try to read the local file directly |
| 104 |
if (!$dims) { |
| 105 |
$dims = $this->get_dims_from_local_file($src); |
| 106 |
} |
| 107 |
|
| 108 |
// Strategy 3: For external images, try getimagesize with URL (slower) |
| 109 |
if (!$dims) { |
| 110 |
$dims = $this->get_dims_from_remote($src); |
| 111 |
} |
| 112 |
|
| 113 |
if ($dims) { |
| 114 |
$this->cache[$src] = $dims; |
| 115 |
$this->set_cached_dimensions($src, $dims); |
| 116 |
} |
| 117 |
|
| 118 |
return $dims; |
| 119 |
} |
| 120 |
|
| 121 |
/** |
| 122 |
* Build the transient key for a given image src. |
| 123 |
*/ |
| 124 |
private static function cache_key(string $src): string { |
| 125 |
return self::CACHE_PREFIX . md5($src); |
| 126 |
} |
| 127 |
|
| 128 |
/** |
| 129 |
* Read previously resolved dimensions from the persistent transient cache. |
| 130 |
* Returns null on a cache miss or a malformed/partial cached value. |
| 131 |
*/ |
| 132 |
private function get_cached_dimensions(string $src): ?array { |
| 133 |
$cached = get_transient(self::cache_key($src)); |
| 134 |
|
| 135 |
if (is_array($cached) |
| 136 |
&& isset($cached['width'], $cached['height']) |
| 137 |
&& (int) $cached['width'] > 0 |
| 138 |
&& (int) $cached['height'] > 0 |
| 139 |
) { |
| 140 |
return [ |
| 141 |
'width' => (int) $cached['width'], |
| 142 |
'height' => (int) $cached['height'], |
| 143 |
]; |
| 144 |
} |
| 145 |
|
| 146 |
return null; |
| 147 |
} |
| 148 |
|
| 149 |
/** |
| 150 |
* Persist resolved dimensions so subsequent requests avoid file I/O. |
| 151 |
*/ |
| 152 |
private function set_cached_dimensions(string $src, array $dims): void { |
| 153 |
set_transient( |
| 154 |
self::cache_key($src), |
| 155 |
[ |
| 156 |
'width' => (int) $dims['width'], |
| 157 |
'height' => (int) $dims['height'], |
| 158 |
], |
| 159 |
self::CACHE_TTL |
| 160 |
); |
| 161 |
} |
| 162 |
|
| 163 |
/** |
| 164 |
* Purge cached dimensions for every size URL of an attachment. |
| 165 |
* |
| 166 |
* Hooked on `delete_attachment` and `wp_update_attachment_metadata` so the |
| 167 |
* cache can never outlive (or contradict) the image: deleting an image and |
| 168 |
* re-uploading a different one under the same filename, or regenerating |
| 169 |
* thumbnails to new dimensions, would otherwise serve stale width/height |
| 170 |
* for up to CACHE_TTL. Registered unconditionally (even when dimension |
| 171 |
* injection is disabled) so stale entries are always cleaned up. |
| 172 |
* |
| 173 |
* @param int $attachment_id Attachment whose cached dimensions to clear. |
| 174 |
*/ |
| 175 |
public static function purge_attachment_cache($attachment_id): void { |
| 176 |
$attachment_id = (int) $attachment_id; |
| 177 |
if ($attachment_id <= 0) { |
| 178 |
return; |
| 179 |
} |
| 180 |
|
| 181 |
foreach (self::collect_attachment_urls($attachment_id) as $url) { |
| 182 |
delete_transient(self::cache_key($url)); |
| 183 |
} |
| 184 |
} |
| 185 |
|
| 186 |
/** |
| 187 |
* `wp_update_attachment_metadata` filter wrapper: purge the cache, then |
| 188 |
* return the metadata untouched so the filter chain is unaffected. |
| 189 |
* |
| 190 |
* @param mixed $data Attachment metadata being saved. |
| 191 |
* @param int $attachment_id Attachment ID. |
| 192 |
* @return mixed The unmodified $data. |
| 193 |
*/ |
| 194 |
public static function purge_on_metadata_update($data, $attachment_id) { |
| 195 |
self::purge_attachment_cache($attachment_id); |
| 196 |
return $data; |
| 197 |
} |
| 198 |
|
| 199 |
/** |
| 200 |
* Collect the full-size URL plus every registered sub-size URL for an |
| 201 |
* attachment. Sub-size files live in the same directory as the full-size |
| 202 |
* file, so each is the full URL with its basename swapped for the size |
| 203 |
* filename. |
| 204 |
* |
| 205 |
* @param int $attachment_id Attachment ID. |
| 206 |
* @return string[] Image URLs (possibly empty). |
| 207 |
*/ |
| 208 |
private static function collect_attachment_urls(int $attachment_id): array { |
| 209 |
$full = wp_get_attachment_url($attachment_id); |
| 210 |
if (!is_string($full) || $full === '') { |
| 211 |
return []; |
| 212 |
} |
| 213 |
|
| 214 |
$urls = [$full]; |
| 215 |
|
| 216 |
$meta = wp_get_attachment_metadata($attachment_id); |
| 217 |
if (is_array($meta) && !empty($meta['sizes'])) { |
| 218 |
$slash = strrpos($full, '/'); |
| 219 |
$base = $slash === false ? '' : substr($full, 0, $slash + 1); |
| 220 |
foreach ($meta['sizes'] as $size_data) { |
| 221 |
if (!empty($size_data['file'])) { |
| 222 |
$urls[] = $base . $size_data['file']; |
| 223 |
} |
| 224 |
} |
| 225 |
} |
| 226 |
|
| 227 |
return $urls; |
| 228 |
} |
| 229 |
|
| 230 |
/** |
| 231 |
* Look up dimensions from WP attachment metadata. |
| 232 |
*/ |
| 233 |
private function get_dims_from_attachment(string $url): ?array { |
| 234 |
$attachment_id = attachment_url_to_postid($url); |
| 235 |
|
| 236 |
if (!$attachment_id) { |
| 237 |
// Try without size suffix (e.g., image-300x200.jpg -> image.jpg) |
| 238 |
$clean_url = preg_replace('/-\d+x\d+(\.\w+)$/', '$1', $url); |
| 239 |
$attachment_id = attachment_url_to_postid($clean_url); |
| 240 |
} |
| 241 |
|
| 242 |
if (!$attachment_id) { |
| 243 |
return null; |
| 244 |
} |
| 245 |
|
| 246 |
$meta = wp_get_attachment_metadata($attachment_id); |
| 247 |
if (!$meta) { |
| 248 |
return null; |
| 249 |
} |
| 250 |
|
| 251 |
// Check sub-sizes first |
| 252 |
if (!empty($meta['sizes'])) { |
| 253 |
$filename = wp_basename($url); |
| 254 |
foreach ($meta['sizes'] as $size_data) { |
| 255 |
if ($size_data['file'] === $filename) { |
| 256 |
return [ |
| 257 |
'width' => (int) $size_data['width'], |
| 258 |
'height' => (int) $size_data['height'], |
| 259 |
]; |
| 260 |
} |
| 261 |
} |
| 262 |
} |
| 263 |
|
| 264 |
// A sized variant with no registered sub-size (crop added after |
| 265 |
// upload, third-party sizes) must not inherit the FULL-size |
| 266 |
// dimensions — wrong width/height re-introduces the very layout |
| 267 |
// shift this module exists to prevent. Leaving the tag alone is |
| 268 |
// safer than injecting wrong values. |
| 269 |
if (preg_match('/-\d+x\d+(?=\.\w+$)/', wp_basename($url))) { |
| 270 |
return null; |
| 271 |
} |
| 272 |
|
| 273 |
// Fall back to full size |
| 274 |
if (!empty($meta['width']) && !empty($meta['height'])) { |
| 275 |
return [ |
| 276 |
'width' => (int) $meta['width'], |
| 277 |
'height' => (int) $meta['height'], |
| 278 |
]; |
| 279 |
} |
| 280 |
|
| 281 |
return null; |
| 282 |
} |
| 283 |
|
| 284 |
/** |
| 285 |
* Convert URL to local path and use getimagesize(). |
| 286 |
*/ |
| 287 |
private function get_dims_from_local_file(string $url): ?array { |
| 288 |
$upload_dir = wp_get_upload_dir(); |
| 289 |
|
| 290 |
if (strpos($url, $upload_dir['baseurl']) === false) { |
| 291 |
return null; |
| 292 |
} |
| 293 |
|
| 294 |
$relative = str_replace($upload_dir['baseurl'], '', $url); |
| 295 |
$file = $upload_dir['basedir'] . $relative; |
| 296 |
|
| 297 |
if (!file_exists($file)) { |
| 298 |
return null; |
| 299 |
} |
| 300 |
|
| 301 |
$info = @getimagesize($file); |
| 302 |
if ($info && $info[0] > 0 && $info[1] > 0) { |
| 303 |
return ['width' => $info[0], 'height' => $info[1]]; |
| 304 |
} |
| 305 |
|
| 306 |
return null; |
| 307 |
} |
| 308 |
|
| 309 |
/** |
| 310 |
* Fetch dimensions from a remote URL. |
| 311 |
* Downloads just enough bytes to read image headers. |
| 312 |
* |
| 313 |
* Server-side fetching of arbitrary <img src> values is an SSRF vector: |
| 314 |
* content authors can plant internal, localhost, or cloud-metadata URLs |
| 315 |
* that this server would otherwise probe on every pageview. Remote |
| 316 |
* fetching is therefore opt-in — it only runs for hosts allowlisted via |
| 317 |
* the 'metasync_dimension_remote_hosts' filter — and failures are |
| 318 |
* negatively cached so they are not re-fetched on every request. |
| 319 |
*/ |
| 320 |
private function get_dims_from_remote(string $url): ?array { |
| 321 |
if (!$this->is_remote_fetch_allowed($url)) { |
| 322 |
return null; |
| 323 |
} |
| 324 |
|
| 325 |
$fail_key = self::FAIL_PREFIX . md5($url); |
| 326 |
if (get_transient($fail_key) !== false) { |
| 327 |
return null; |
| 328 |
} |
| 329 |
|
| 330 |
$response = wp_remote_get($url, [ |
| 331 |
'timeout' => 3, |
| 332 |
'headers' => ['Range' => 'bytes=0-' . (self::REMOTE_MAX_BYTES - 1)], |
| 333 |
'limit_response_size' => self::REMOTE_MAX_BYTES, |
| 334 |
]); |
| 335 |
|
| 336 |
$body = null; |
| 337 |
if (!is_wp_error($response)) { |
| 338 |
$body = wp_remote_retrieve_body($response); |
| 339 |
} |
| 340 |
|
| 341 |
if (empty($body)) { |
| 342 |
set_transient($fail_key, 1, self::FAIL_TTL); |
| 343 |
return null; |
| 344 |
} |
| 345 |
|
| 346 |
if (!function_exists('wp_tempnam')) { |
| 347 |
require_once ABSPATH . 'wp-admin/includes/file.php'; |
| 348 |
} |
| 349 |
|
| 350 |
$tmp = wp_tempnam($url); |
| 351 |
file_put_contents($tmp, $body); |
| 352 |
$info = @getimagesize($tmp); |
| 353 |
@unlink($tmp); |
| 354 |
|
| 355 |
if ($info && $info[0] > 0 && $info[1] > 0) { |
| 356 |
return ['width' => $info[0], 'height' => $info[1]]; |
| 357 |
} |
| 358 |
|
| 359 |
set_transient($fail_key, 1, self::FAIL_TTL); |
| 360 |
return null; |
| 361 |
} |
| 362 |
|
| 363 |
/** |
| 364 |
* A remote URL may only be fetched when explicitly allowed. |
| 365 |
* |
| 366 |
* @param string $url Absolute image URL. |
| 367 |
* @return bool True when the scheme is http(s), the URL passes core's |
| 368 |
* internal-address validation, and its host is allowlisted. |
| 369 |
*/ |
| 370 |
private function is_remote_fetch_allowed(string $url): bool { |
| 371 |
/** |
| 372 |
* Hosts whose images may be fetched server-side for dimension |
| 373 |
* sniffing. Empty by default — remote dimension fetching is opt-in, |
| 374 |
* which closes the SSRF surface until a site owner explicitly |
| 375 |
* trusts specific hosts. Checked before any URL parsing so the |
| 376 |
* default (fetch-disabled) path stays cheap. |
| 377 |
* |
| 378 |
* @param string[] $allowed_hosts List of hostnames. |
| 379 |
*/ |
| 380 |
$allowed_hosts = apply_filters('metasync_dimension_remote_hosts', []); |
| 381 |
if (empty($allowed_hosts)) { |
| 382 |
return false; |
| 383 |
} |
| 384 |
|
| 385 |
$scheme = wp_parse_url($url, PHP_URL_SCHEME); |
| 386 |
if (!in_array($scheme, ['http', 'https'], true)) { |
| 387 |
return false; |
| 388 |
} |
| 389 |
|
| 390 |
if (!wp_http_validate_url($url)) { |
| 391 |
return false; |
| 392 |
} |
| 393 |
|
| 394 |
$host = strtolower((string) wp_parse_url($url, PHP_URL_HOST)); |
| 395 |
foreach ($allowed_hosts as $allowed) { |
| 396 |
if (strtolower((string) $allowed) === $host) { |
| 397 |
return true; |
| 398 |
} |
| 399 |
} |
| 400 |
|
| 401 |
return false; |
| 402 |
} |
| 403 |
|
| 404 |
/** |
| 405 |
* Insert an attribute into an <img> tag. |
| 406 |
*/ |
| 407 |
private function add_attribute(string $tag, string $name, string $value): string { |
| 408 |
return preg_replace( |
| 409 |
'/(<img\s)/i', |
| 410 |
sprintf('$1%s="%s" ', esc_attr($name), esc_attr($value)), |
| 411 |
$tag, |
| 412 |
1 |
| 413 |
); |
| 414 |
} |
| 415 |
} |
| 416 |
|
