| 1 |
<?php |
| 2 |
/** |
| 3 |
* Guard against images that cannot be decoded within the available memory. |
| 4 |
* |
| 5 |
* Background |
| 6 |
* ---------- |
| 7 |
* GD holds images UNCOMPRESSED — width x height x 4 bytes — so the size on disk says |
| 8 |
* nothing about the cost of processing one. A 6336x9504 Pexels photo is 1.7MB as a JPEG |
| 9 |
* and 229.7MB as a GD bitmap. Decoding it under a 256M memory_limit is an instant fatal |
| 10 |
* inside imagecreatefromstring(), which WordPress surfaces as "There has been a critical |
| 11 |
* error on this website" and a HTTP 500 — killing a kit import page with no usable error. |
| 12 |
* |
| 13 |
* Imagick is far more frugal and can spill to a disk-backed pixel cache, which is why this |
| 14 |
* only bites on hosts without it. |
| 15 |
* |
| 16 |
* This class is deliberately free of WordPress dependencies: the sizing arithmetic and the |
| 17 |
* URL checks are pure functions, so they can be reasoned about - and exercised - on their own. |
| 18 |
* |
| 19 |
* @package WDesignKit |
| 20 |
* @since 2.6.2 |
| 21 |
*/ |
| 22 |
|
| 23 |
if ( ! defined( 'ABSPATH' ) ) { |
| 24 |
exit; |
| 25 |
} |
| 26 |
|
| 27 |
if ( ! class_exists( 'Wdkit_Image_Guard' ) ) { |
| 28 |
|
| 29 |
/** |
| 30 |
* Image memory-safety helpers. |
| 31 |
*/ |
| 32 |
class Wdkit_Image_Guard { |
| 33 |
|
| 34 |
/** |
| 35 |
* Bytes per pixel for a decoded truecolour bitmap (RGBA). |
| 36 |
*/ |
| 37 |
const BYTES_PER_PIXEL = 4; |
| 38 |
|
| 39 |
/** |
| 40 |
* Multiplier applied to the decoded source size to approximate real peak usage. |
| 41 |
* |
| 42 |
* A resize needs the decoded source AND a destination bitmap live at the same time, |
| 43 |
* plus the compressed file contents that were read to produce the source. 1.6x of the |
| 44 |
* source bitmap tracks observed peaks closely enough to keep us on the safe side |
| 45 |
* without rejecting images that would actually have succeeded. |
| 46 |
*/ |
| 47 |
const PEAK_OVERHEAD = 1.6; |
| 48 |
|
| 49 |
/** |
| 50 |
* Image file extensions we are willing to sideload. |
| 51 |
*/ |
| 52 |
const IMAGE_EXTENSIONS = 'jpg|jpeg|jpe|gif|png|webp|avif|bmp'; |
| 53 |
|
| 54 |
/** |
| 55 |
* Video file extensions a media control can point at. |
| 56 |
* |
| 57 |
* Kept out of IMAGE_EXTENSIONS for the same reason SVG is: that list governs what may be |
| 58 |
* *decoded*, and none of these ever are. Only the extensions WordPress already allows |
| 59 |
* with no user (mp4/m4v, webm, mov/qt, ogv), so a background sideload of one succeeds |
| 60 |
* where an SVG's cannot. |
| 61 |
*/ |
| 62 |
const VIDEO_EXTENSIONS = 'mp4|m4v|webm|mov|qt|ogv'; |
| 63 |
|
| 64 |
/** |
| 65 |
* Everything a media control can legitimately point at, SVG and video included. |
| 66 |
* |
| 67 |
* SVG and video are kept apart from IMAGE_EXTENSIONS because that list governs what may |
| 68 |
* be *decoded* - neither is ever rasterised, so the memory arithmetic does not apply to |
| 69 |
* them. This list is the other question: "is this a media file we import at all", which |
| 70 |
* is what filename_from_url() and is_foreign_media() need. Leaving video out of it meant |
| 71 |
* filename_from_url() returned '' for every video, and callers read that as "cannot |
| 72 |
* handle this" - the same way SVGs were being skipped before they were added here. |
| 73 |
*/ |
| 74 |
const MEDIA_EXTENSIONS = self::IMAGE_EXTENSIONS . '|svg|' . self::VIDEO_EXTENSIONS; |
| 75 |
|
| 76 |
/** |
| 77 |
* Extensions the deferred-media pass owns, end to end. |
| 78 |
* |
| 79 |
* One list, because three separate ones drifted apart and each divergence was a bug: |
| 80 |
* the collector deferred `avif` but the id-blanking pass did not recognise it, so an AVIF |
| 81 |
* kept a dangling id and rendered empty for the whole deferred window. Anything added |
| 82 |
* here must be deferrable, blankable AND remappable - those three passes have to agree or |
| 83 |
* an id is cleared that nothing will ever restore. |
| 84 |
* |
| 85 |
* SVG is deliberately absent: it is imported synchronously (the background pass runs with |
| 86 |
* no user, where `svg` is not in get_allowed_mime_types()) and already carries a real id. |
| 87 |
* |
| 88 |
* @since 2.7.1 |
| 89 |
*/ |
| 90 |
const DEFERRABLE_EXTENSIONS = 'png|jpe?g|webp|gif|avif|' . self::VIDEO_EXTENSIONS; |
| 91 |
|
| 92 |
/** |
| 93 |
* The same list as a plain array, for in_array() call sites. |
| 94 |
* |
| 95 |
* @since 2.7.1 |
| 96 |
* |
| 97 |
* @return array<string> |
| 98 |
*/ |
| 99 |
public static function deferrable_extensions() { |
| 100 |
return array( 'png', 'jpg', 'jpeg', 'webp', 'gif', 'avif', 'mp4', 'm4v', 'webm', 'mov', 'qt', 'ogv' ); |
| 101 |
} |
| 102 |
|
| 103 |
/** |
| 104 |
* Whether an image of the given dimensions can be decoded in the memory available. |
| 105 |
* |
| 106 |
* Pure arithmetic — no WordPress, no filesystem, no globals — so it is directly |
| 107 |
* unit-testable. |
| 108 |
* |
| 109 |
* @param int $width Image width in pixels. |
| 110 |
* @param int $height Image height in pixels. |
| 111 |
* @param int $available_bytes Bytes of headroom left. 0 or less means "unlimited". |
| 112 |
* @param int $bytes_per_pixel Bytes per decoded pixel. Default 4 (RGBA). |
| 113 |
* @param float $overhead Peak multiplier. Default self::PEAK_OVERHEAD. |
| 114 |
* @return bool True when it is safe to decode, false when it should be skipped. |
| 115 |
*/ |
| 116 |
public static function decode_fits( $width, $height, $available_bytes, $bytes_per_pixel = self::BYTES_PER_PIXEL, $overhead = self::PEAK_OVERHEAD ) { |
| 117 |
|
| 118 |
$width = (int) $width; |
| 119 |
$height = (int) $height; |
| 120 |
$available_bytes = (int) $available_bytes; |
| 121 |
|
| 122 |
// Unknown or non-raster dimensions: not ours to judge, let WordPress proceed. |
| 123 |
if ( $width < 1 || $height < 1 ) { |
| 124 |
return true; |
| 125 |
} |
| 126 |
|
| 127 |
// No ceiling (memory_limit = -1): nothing to protect against. |
| 128 |
if ( $available_bytes <= 0 ) { |
| 129 |
return true; |
| 130 |
} |
| 131 |
|
| 132 |
$bytes_per_pixel = (int) $bytes_per_pixel > 0 ? (int) $bytes_per_pixel : self::BYTES_PER_PIXEL; |
| 133 |
$overhead = (float) $overhead > 0 ? (float) $overhead : self::PEAK_OVERHEAD; |
| 134 |
|
| 135 |
// Multiply in float space: 6336 * 9504 * 4 * 1.6 overflows nothing here, but very |
| 136 |
// large synthetic dimensions would wrap a 32-bit int. |
| 137 |
$needed = (float) $width * (float) $height * (float) $bytes_per_pixel * $overhead; |
| 138 |
|
| 139 |
return $needed <= (float) $available_bytes; |
| 140 |
} |
| 141 |
|
| 142 |
/** |
| 143 |
* Extract a usable filename from an image URL, ignoring any query string. |
| 144 |
* |
| 145 |
* Elementor's Import_Images::import() does `basename( $url )` and then bails silently |
| 146 |
* when wp_check_filetype() finds no extension. A sized CDN URL such as |
| 147 |
* |
| 148 |
* https://images.pexels.com/photos/1/pexels-photo-1.jpeg?w=1920&auto=compress |
| 149 |
* |
| 150 |
* has the basename "pexels-photo-1.jpeg?w=1920&auto=compress", which ends in |
| 151 |
* "compress" — so Elementor silently imports nothing, no attachment is created, and |
| 152 |
* every widget that resolves its image through an attachment ID renders blank. |
| 153 |
* |
| 154 |
* WordPress core gets this right in media_sideload_image() by matching `[^\?]+` up to |
| 155 |
* the query string. This mirrors that so callers can localise such URLs before |
| 156 |
* Elementor ever sees them. |
| 157 |
* |
| 158 |
* @param string $url Image URL. |
| 159 |
* @return string Bare filename including extension, or '' when the URL is not an image. |
| 160 |
*/ |
| 161 |
public static function filename_from_url( $url ) { |
| 162 |
if ( ! is_string( $url ) || '' === trim( $url ) ) { |
| 163 |
return ''; |
| 164 |
} |
| 165 |
|
| 166 |
// Match the path portion only — [^\?] stops at the query string, exactly as |
| 167 |
// media_sideload_image() does. |
| 168 |
// |
| 169 |
// MEDIA_EXTENSIONS, not IMAGE_EXTENSIONS: an SVG is a file we import like any other, |
| 170 |
// even though it is never decoded and so plays no part in the memory arithmetic. |
| 171 |
// Leaving it out here returned an empty filename for every SVG, and callers that |
| 172 |
// treat that as "cannot handle this" skipped logos and icons entirely. |
| 173 |
if ( ! preg_match( '/[^\?]+\.(?:' . self::MEDIA_EXTENSIONS . ')\b/i', $url, $matches ) ) { |
| 174 |
return ''; |
| 175 |
} |
| 176 |
|
| 177 |
$filename = $matches[0]; |
| 178 |
|
| 179 |
// Take the last path segment without depending on WordPress's wp_basename(). |
| 180 |
$filename = preg_replace( '#^.*/#', '', str_replace( '\\', '/', $filename ) ); |
| 181 |
|
| 182 |
return (string) $filename; |
| 183 |
} |
| 184 |
|
| 185 |
/** |
| 186 |
* Is this media reference still pointing off-site? |
| 187 |
* |
| 188 |
* Template content arrives holding whatever URL the media had wherever it came from, |
| 189 |
* and alongside it that site's attachment ID - which means nothing here. Controls left |
| 190 |
* in that state render nothing at all whenever the control resolves through the ID |
| 191 |
* rather than the URL: Elementor inlines an SVG icon by reading the file off the |
| 192 |
* attachment, and addon widgets fetch their thumbnail by ID, so both come out empty. |
| 193 |
* Worse, a source ID can collide with a real local post - IDs seen in practice landed |
| 194 |
* on a revision and on an Elementor template. |
| 195 |
* |
| 196 |
* Two things make this narrow enough to act on: |
| 197 |
* |
| 198 |
* - The URL must be absolute http(s). Relative and data: URLs are not ours to judge. |
| 199 |
* - The path must end in a media extension. Media controls share the {url,id} shape |
| 200 |
* with other controls, and this is what keeps a link from being mistaken for one. |
| 201 |
* |
| 202 |
* Compared without the scheme, because one site is routinely reachable over both http |
| 203 |
* and https and a mismatch there would make every local URL look foreign. |
| 204 |
* |
| 205 |
* @param string $url URL from a media control. |
| 206 |
* @param string $upload_baseurl This site's upload base URL. |
| 207 |
* @return bool True when the URL is media belonging somewhere else. |
| 208 |
*/ |
| 209 |
public static function is_foreign_media( $url, $upload_baseurl, $require_extension = true ) { |
| 210 |
if ( ! is_string( $url ) || ! is_string( $upload_baseurl ) || '' === $upload_baseurl ) { |
| 211 |
return false; |
| 212 |
} |
| 213 |
|
| 214 |
if ( ! preg_match( '#^https?://#i', $url ) ) { |
| 215 |
return false; |
| 216 |
} |
| 217 |
|
| 218 |
$strip = static function ( $value ) { |
| 219 |
return preg_replace( '#^https?://#i', '', $value ); |
| 220 |
}; |
| 221 |
|
| 222 |
// Already ours. |
| 223 |
if ( 0 === strpos( $strip( $url ), $strip( $upload_baseurl ) ) ) { |
| 224 |
return false; |
| 225 |
} |
| 226 |
|
| 227 |
// A caller that found this URL by scanning text has to prove it is media somehow, |
| 228 |
// and the extension is the only evidence available. A caller that found it in a |
| 229 |
// `{ url, id }` media control already knows, and passes false: the template CDN |
| 230 |
// serves extensionless URLs and resolves the type by Content-Type, so requiring an |
| 231 |
// extension there rejected real media and left it hotlinked forever. |
| 232 |
if ( ! $require_extension ) { |
| 233 |
return true; |
| 234 |
} |
| 235 |
|
| 236 |
$path = (string) parse_url( $url, PHP_URL_PATH ); // phpcs:ignore WordPress.WP.AlternativeFunctions.parse_url_parse_url -- no WordPress dependency by design. |
| 237 |
|
| 238 |
return (bool) preg_match( '/\.(?:' . self::MEDIA_EXTENSIONS . ')$/i', $path ); |
| 239 |
} |
| 240 |
|
| 241 |
/** |
| 242 |
* Width requested for a Pexels source image before it is downloaded. |
| 243 |
* |
| 244 |
* 1920px is wide enough for any section background a kit page uses; the originals this |
| 245 |
* caps run up to 6336x9504 (see the class docblock). |
| 246 |
*/ |
| 247 |
const PEXELS_MAX_WIDTH = 1920; |
| 248 |
|
| 249 |
/** |
| 250 |
* Cap a Pexels source image at a sane width before it is downloaded. |
| 251 |
* |
| 252 |
* Pexels photo URLs resolve to full-resolution originals by default, but accept a `w` |
| 253 |
* query parameter that returns a server-resized rendition instead. A kit page never |
| 254 |
* needs more than PEXELS_MAX_WIDTH, and asking for it up front means a smaller network |
| 255 |
* transfer AND a smaller bitmap for the decode-memory guard above to worry about, |
| 256 |
* instead of downloading the original just to discard the resolution afterward. |
| 257 |
* |
| 258 |
* Pure string handling - no WordPress dependency, unit-testable like the rest of this |
| 259 |
* class. |
| 260 |
* |
| 261 |
* @param string $url Candidate image URL. |
| 262 |
* @param int $max_width Width to request. Default self::PEXELS_MAX_WIDTH. |
| 263 |
* @return string The URL unchanged if it is not a Pexels image URL, otherwise capped. |
| 264 |
*/ |
| 265 |
public static function cap_pexels_source( $url, $max_width = self::PEXELS_MAX_WIDTH ) { |
| 266 |
if ( ! is_string( $url ) || '' === $url ) { |
| 267 |
return $url; |
| 268 |
} |
| 269 |
|
| 270 |
$parts = parse_url( $url ); // phpcs:ignore WordPress.WP.AlternativeFunctions.parse_url_parse_url -- no WordPress dependency by design. |
| 271 |
|
| 272 |
if ( ! is_array( $parts ) || empty( $parts['host'] ) ) { |
| 273 |
return $url; |
| 274 |
} |
| 275 |
|
| 276 |
$host = strtolower( (string) $parts['host'] ); |
| 277 |
|
| 278 |
// Exact host, or a real subdomain of pexels.com. A substring test (the previous |
| 279 |
// `stripos( $host, 'pexels.com' )`) also accepted `pexels.com.evil.tld` and |
| 280 |
// `notpexels.com`, which is the wrong shape for a host check even where the |
| 281 |
// consequence is only a rewritten query string - the next reader will take it for a |
| 282 |
// host allowlist. |
| 283 |
if ( 'pexels.com' !== $host && ! self::str_ends_with_ci( $host, '.pexels.com' ) ) { |
| 284 |
return $url; |
| 285 |
} |
| 286 |
|
| 287 |
$max_width = (int) $max_width > 0 ? (int) $max_width : self::PEXELS_MAX_WIDTH; |
| 288 |
|
| 289 |
$query = isset( $parts['query'] ) ? (string) $parts['query'] : ''; |
| 290 |
|
| 291 |
$params = array(); |
| 292 |
parse_str( $query, $params ); |
| 293 |
|
| 294 |
// parse_str() is lossy: it rewrites any parameter name containing a dot or a space |
| 295 |
// into an underscore, so `X-Amz-Sig.nature=abc` is re-emitted as |
| 296 |
// `X-Amz-Sig_nature=abc`. On a signed URL that invalidates the signature, the CDN |
| 297 |
// answers 403, media_sideload_image() returns WP_Error and the image is dropped with |
| 298 |
// nothing to retry it. Rather than guess which parameters matter, check whether the |
| 299 |
// round trip is lossless at all and decline to rewrite when it is not - serving the |
| 300 |
// original at full width is always the safer failure. |
| 301 |
if ( '' !== $query && ! self::query_round_trips( $query, $params ) ) { |
| 302 |
return $url; |
| 303 |
} |
| 304 |
|
| 305 |
// Already capped at or below the target width - leave it alone. `is_numeric` first: |
| 306 |
// `(int) 'abc'` is 0, which compares as "already small enough" and skipped the cap |
| 307 |
// entirely, so a junk width was a second route to a full-resolution download. |
| 308 |
if ( isset( $params['w'] ) && is_numeric( $params['w'] ) && (int) $params['w'] > 0 |
| 309 |
&& (int) $params['w'] <= $max_width |
| 310 |
) { |
| 311 |
return $url; |
| 312 |
} |
| 313 |
|
| 314 |
$params['w'] = $max_width; |
| 315 |
|
| 316 |
// `auto=compress` is what makes Pexels honour `w` on some endpoints; harmless to set |
| 317 |
// unconditionally since it only affects encoding, never dimensions. |
| 318 |
if ( empty( $params['auto'] ) ) { |
| 319 |
$params['auto'] = 'compress'; |
| 320 |
} |
| 321 |
|
| 322 |
// Reassembled from components rather than `strtok( $url, '?' )`, which kept any |
| 323 |
// `#fragment` in the base and so appended the new query *after* it - producing |
| 324 |
// `…/a.jpg#frag?w=1920`, which every server ignores. That silently defeated the cap |
| 325 |
// on any fragment-carrying URL. |
| 326 |
$rebuilt = ''; |
| 327 |
|
| 328 |
if ( ! empty( $parts['scheme'] ) ) { |
| 329 |
$rebuilt .= $parts['scheme'] . '://'; |
| 330 |
} |
| 331 |
|
| 332 |
if ( ! empty( $parts['user'] ) ) { |
| 333 |
$rebuilt .= $parts['user'] . ( empty( $parts['pass'] ) ? '' : ':' . $parts['pass'] ) . '@'; |
| 334 |
} |
| 335 |
|
| 336 |
$rebuilt .= $parts['host']; |
| 337 |
|
| 338 |
if ( ! empty( $parts['port'] ) ) { |
| 339 |
$rebuilt .= ':' . (int) $parts['port']; |
| 340 |
} |
| 341 |
|
| 342 |
$rebuilt .= isset( $parts['path'] ) ? $parts['path'] : ''; |
| 343 |
$rebuilt .= '?' . http_build_query( $params ); |
| 344 |
|
| 345 |
// Fragment last, where it belongs. |
| 346 |
if ( isset( $parts['fragment'] ) && '' !== $parts['fragment'] ) { |
| 347 |
$rebuilt .= '#' . $parts['fragment']; |
| 348 |
} |
| 349 |
|
| 350 |
return $rebuilt; |
| 351 |
} |
| 352 |
|
| 353 |
/** |
| 354 |
* Case-insensitive suffix test. |
| 355 |
* |
| 356 |
* str_ends_with() is PHP 8.0+; this plugin still supports 7.4. |
| 357 |
* |
| 358 |
* @since 2.7.1 |
| 359 |
* |
| 360 |
* @param string $haystack Subject. |
| 361 |
* @param string $needle Suffix to look for. |
| 362 |
* @return bool |
| 363 |
*/ |
| 364 |
private static function str_ends_with_ci( $haystack, $needle ) { |
| 365 |
$len = strlen( $needle ); |
| 366 |
|
| 367 |
if ( 0 === $len ) { |
| 368 |
return true; |
| 369 |
} |
| 370 |
|
| 371 |
return strlen( $haystack ) >= $len && 0 === substr_compare( $haystack, $needle, -$len, $len, true ); |
| 372 |
} |
| 373 |
|
| 374 |
/** |
| 375 |
* Can this query string survive parse_str() + http_build_query() unchanged? |
| 376 |
* |
| 377 |
* Compares the parameter names the parser produced against the names actually present in |
| 378 |
* the raw query. Any mismatch means rebuilding would emit a different query than it was |
| 379 |
* given - the dot/space-to-underscore rewrite is the common case - and the caller should |
| 380 |
* leave the URL alone. Name-based rather than provider-based, so an unfamiliar signing |
| 381 |
* scheme fails safe without needing to be listed. |
| 382 |
* |
| 383 |
* @since 2.7.1 |
| 384 |
* |
| 385 |
* @param string $query Raw query string. |
| 386 |
* @param array $params Result of parse_str() on that query. |
| 387 |
* @return bool True when a rewrite would preserve every parameter name. |
| 388 |
*/ |
| 389 |
private static function query_round_trips( $query, array $params ) { |
| 390 |
|
| 391 |
foreach ( explode( '&', $query ) as $pair ) { |
| 392 |
if ( '' === $pair ) { |
| 393 |
continue; |
| 394 |
} |
| 395 |
|
| 396 |
$raw_key = urldecode( strtok( $pair, '=' ) ); |
| 397 |
|
| 398 |
// PHP maps `a[b]=1` to a nested array; the bracket form is expected to differ |
| 399 |
// from the flat key, and it is not the case being guarded against. |
| 400 |
if ( false !== strpos( $raw_key, '[' ) ) { |
| 401 |
continue; |
| 402 |
} |
| 403 |
|
| 404 |
if ( ! array_key_exists( $raw_key, $params ) ) { |
| 405 |
return false; |
| 406 |
} |
| 407 |
} |
| 408 |
|
| 409 |
return true; |
| 410 |
} |
| 411 |
|
| 412 |
} |
| 413 |
} |
| 414 |
|
