← All changes | media-optimization/class-dimension-injector.php +206 -8 2.6.14 → 2.7.1 View file →

		@@ -17,8 +17,23 @@
17	17
18	18	/** In-memory cache to avoid repeated file lookups within a request. */
19	19	private array $cache = [];
20	20
	21	+ /** Transient key prefix for persisted image dimensions. */
	22	+ private const CACHE_PREFIX = 'ms_dim_';
	23	+
	24	+ /** How long resolved dimensions stay cached, in seconds (7 days). */
	25	+ private const CACHE_TTL = 604800;
	26	+
	27	+ /** Transient key prefix for failed remote lookups (negative cache). */
	28	+ private const FAIL_PREFIX = 'ms_dimfail_';
	29	+
	30	+ /** How long a failed remote lookup stays cached, in seconds (1 hour). */
	31	+ private const FAIL_TTL = 3600;
	32	+
	33	+ /** Cap on bytes fetched from a remote image header sniff. */
	34	+ private const REMOTE_MAX_BYTES = 65536;
	35	+
21	36	public function __construct() {
22	37	add_filter('the_content', [$this, 'inject_dimensions'], 30);
23	38	add_filter('post_thumbnail_html', [$this, 'inject_dimensions'], 30);
24	39	add_filter('widget_text', [$this, 'inject_dimensions'], 30);
		@@ -27,9 +42,9 @@
27	42	/**
28	43	* Find <img> tags missing width or height and inject dimensions.
29	44	*/
30	45	public function inject_dimensions(string $content): string {
31		- if (empty($content) || is_admin()) {
	46	+ if (empty($content) || is_admin() || is_feed()) {
32	47	return $content;
33	48	}
34	49
35	50	return preg_replace_callback('/<img\s[^>]+>/i', function ($matches) {
		@@ -67,12 +82,20 @@
67	82	return null;
68	83	}
69	84	$src = $m[1];
70	85
	86	+ // Request-scoped cache: avoids repeated work within a single render.
71	87	if (isset($this->cache[$src])) {
72	88	return $this->cache[$src];
73	89	}
74	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	+
75	98	$dims = null;
76	99
77	100	// Strategy 1: Try to find WordPress attachment by URL
78	101	$dims = $this->get_dims_from_attachment($src);
		@@ -88,8 +111,9 @@
88	111	}
89	112
90	113	if ($dims) {
91	114	$this->cache[$src] = $dims;
	115	+ $this->set_cached_dimensions($src, $dims);
92	116	}
93	117
94	118	return $dims;
95	119	}
		@@ -94,8 +118,117 @@
94	118	return $dims;
95	119	}
96	120
97	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	+ /**
98	231	* Look up dimensions from WP attachment metadata.
99	232	*/
100	233	private function get_dims_from_attachment(string $url): ?array {
101	234	$attachment_id = attachment_url_to_postid($url);
		@@ -127,8 +260,17 @@
127	260	}
128	261	}
129	262	}
130	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	+
131	273	// Fall back to full size
132	274	if (!empty($meta['width']) && !empty($meta['height'])) {
133	275	return [
134	276	'width' => (int) $meta['width'],
		@@ -166,25 +308,39 @@
166	308
167	309	/**
168	310	* Fetch dimensions from a remote URL.
169	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.
170	319	*/
171	320	private function get_dims_from_remote(string $url): ?array {
172		- if (strpos($url, 'http') !== 0) {
	321	+ if (!$this->is_remote_fetch_allowed($url)) {
173	322	return null;
174	323	}
175	324
	325	+ $fail_key = self::FAIL_PREFIX . md5($url);
	326	+ if (get_transient($fail_key) !== false) {
	327	+ return null;
	328	+ }
	329	+
176	330	$response = wp_remote_get($url, [
177		- 'timeout' => 3,
178		- 'headers' => ['Range' => 'bytes=0-32767'],
	331	+ 'timeout' => 3,
	332	+ 'headers' => ['Range' => 'bytes=0-' . (self::REMOTE_MAX_BYTES - 1)],
	333	+ 'limit_response_size' => self::REMOTE_MAX_BYTES,
179	334	]);
180	335
181		- if (is_wp_error($response)) {
182		- return null;
	336	+ $body = null;
	337	+ if (!is_wp_error($response)) {
	338	+ $body = wp_remote_retrieve_body($response);
183	339	}
184	340
185		- $body = wp_remote_retrieve_body($response);
186	341	if (empty($body)) {
	342	+ set_transient($fail_key, 1, self::FAIL_TTL);
187	343	return null;
188	344	}
189	345
190	346	if (!function_exists('wp_tempnam')) {
		@@ -193,15 +349,57 @@
193	349
194	350	$tmp = wp_tempnam($url);
195	351	file_put_contents($tmp, $body);
196	352	$info = @getimagesize($tmp);
197		- @unlink($tmp);
	353	+ @wp_delete_file($tmp);
198	354
199	355	if ($info && $info[0] > 0 && $info[1] > 0) {
200	356	return ['width' => $info[0], 'height' => $info[1]];
201	357	}
202	358
	359	+ set_transient($fail_key, 1, self::FAIL_TTL);
203	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;
204	402	}
205	403
206	404	/**
207	405	* Insert an attribute into an <img> tag.