ablocks / includes / classes / page-cache / store.php

store.php in aBlocks – Gutenberg Blocks, User Dashboard Builder, Popup Builder, Form Builder & Animation Builder 2.11.0, at includes/classes/page-cache/store.php

637 lines 18.6 KB

Raw Download Zip

1	<?php
2	namespace ABlocks\Classes\PageCache;
3
4	if ( ! defined( 'ABSPATH' ) ) {
5	exit;
6	}
7
8	use ABlocks\Classes\FileUpload;
9	use ABlocks\Classes\CacheBackend;
10
11	/**
12	* Page Cache — filesystem layout, reads, writes and purges.
13	*
14	* Files live at:
15	*
16	* uploads/ablocks_uploads/page-cache/{host}/{url-path}/index.html
17	* index.html.gz
18	*
19	* The layout mirrors the URL path rather than hashing it. Hashing would be
20	* simpler here, but nginx cannot compute a hash in a `try_files` rule, so a hash
21	* layout permanently forecloses the server-level tier. Mirroring costs nothing
22	* now and is the only layout all three tiers can share.
23	*
24	* Because nginx matches against its already-decoded `$uri`, the on-disk path
25	* must be the decoded URL path — including UTF-8 slugs. That rules out
26	* sanitising by transliteration or hashing unsafe segments; instead we reject
27	* anything that cannot be represented safely. See {@see Store::sanitize_segment()}.
28	*
29	* Writes are direct filesystem calls rather than WP_Filesystem. This is
30	* deliberate: WP_Filesystem cannot do an atomic rename, may fall back to an FTP
31	* transport that prompts for credentials, and requires loading wp-admin includes
32	* on a frontend request. A page cache needs write-then-rename so a concurrent
33	* reader never observes a half-written page. Every path written here is
34	* confined to the plugin's own uploads subdirectory and is validated against it
35	* before use.
36	*/
37	class Store {
38
39	const DIR_NAME = 'page-cache';
40	const FILE_NAME = 'index.html';
41
42	/**
43	* Resolved base directory, memoised per request.
44	*
45	* @var string\|null
46	*/
47	private static $base_dir = null;
48
49	/**
50	* Absolute path to the page-cache root, without a trailing slash.
51	*
52	* @return string
53	*/
54	public static function base_dir() {
55	if ( null === self::$base_dir ) {
56	$upload = new FileUpload();
57	self::$base_dir = untrailingslashit( $upload->get_upload_dir() ) . '/' . self::DIR_NAME;
58	}
59	return self::$base_dir;
60	}
61
62	/**
63	* Absolute path to the cache file for a host + URL path, or null when the
64	* request cannot be represented safely on disk.
65	*
66	* @param string $host Host name.
67	* @param string $path URL path (decoded or encoded; decoded here).
68	* @param string $variant Optional variant suffix, e.g. 'mobile'.
69	* @return string\|null
70	*/
71	public static function file_path( $host, $path, $variant = '' ) {
72	$host = self::sanitize_host( $host );
73	if ( '' === $host ) {
74	return null;
75	}
76
77	$segments = self::sanitize_path( $path );
78	if ( null === $segments ) {
79	return null;
80	}
81
82	$file = self::FILE_NAME;
83	if ( '' !== $variant ) {
84	$variant = preg_replace( '/[^a-z0-9-]/', '', strtolower( $variant ) );
85	if ( '' !== $variant ) {
86	$file = 'index-' . $variant . '.html';
87	}
88	}
89
90	$dir = self::base_dir() . '/' . $host;
91	if ( ! empty( $segments ) ) {
92	$dir .= '/' . implode( '/', $segments );
93	}
94
95	$full = $dir . '/' . $file;
96
97	// Final containment assertion. The segment rules below already make
98	// traversal impossible, but this is the check that actually matters, so
99	// it is enforced independently of them rather than trusted to them.
100	if ( ! self::is_inside_base( $full ) ) {
101	return null;
102	}
103
104	return $full;
105	}
106
107	/**
108	* Cache file path for the current request, or null when not representable.
109	*
110	* @return string\|null
111	*/
112	public static function current_file_path() {
113	$host = isset( $_SERVER['HTTP_HOST'] ) ? wp_unslash( $_SERVER['HTTP_HOST'] ) : '';
114	return self::file_path( is_string( $host ) ? $host : '', Rules::request_path(), self::current_variant() );
115	}
116
117	/**
118	* Variant suffix for the current request ('' or 'mobile').
119	*
120	* @return string
121	*/
122	public static function current_variant() {
123	$enabled = (bool) \ABlocks\Helper::get_settings( 'perf_page_cache_mobile_variant', false );
124	if ( ! $enabled ) {
125	return '';
126	}
127	$agent = isset( $_SERVER['HTTP_USER_AGENT'] ) ? wp_unslash( $_SERVER['HTTP_USER_AGENT'] ) : '';
128	if ( ! is_string( $agent ) \|\| '' === $agent ) {
129	return '';
130	}
131	return preg_match( '/Mobile\|Android\|iPhone\|iPad\|iPod\|Opera Mini\|IEMobile/i', $agent ) ? 'mobile' : '';
132	}
133
134	/**
135	* Write a page to the cache atomically.
136	*
137	* @param string $file Absolute target path from {@see Store::file_path()}.
138	* @param string $html Rendered HTML.
139	* @param bool $gzip Also write a .gz sibling for gzip_static.
140	* @return bool
141	*/
142	public static function write( $file, $html, $gzip = true ) {
143	if ( empty( $file ) \|\| ! self::is_inside_base( $file ) ) {
144	return false;
145	}
146
147	// Mirror into a persistent object cache when one exists. On multi-server
148	// setups the filesystem is per-node, so a page written on one server is a
149	// miss on every other; an object cache is shared. Files are still written
150	// regardless, because the nginx tier can only ever serve from disk.
151	self::mirror_to_object_cache( $file, $html );
152
153	$dir = dirname( $file );
154	if ( ! is_dir( $dir ) && ! wp_mkdir_p( $dir ) ) {
155	return false;
156	}
157
158	self::protect_base_dir();
159
160	if ( ! self::atomic_put( $file, $html ) ) {
161	return false;
162	}
163
164	$gz_file = $file . '.gz';
165	if ( $gzip && function_exists( 'gzencode' ) ) {
166	$encoded = gzencode( $html, 6 );
167	if ( false !== $encoded ) {
168	// A failed .gz write is not fatal — the plain file still serves.
169	self::atomic_put( $gz_file, $encoded );
170	}
171	} elseif ( file_exists( $gz_file ) ) {
172	// Setting turned off after entries were written with a sibling: drop
173	// the stale .gz, or gzip_static would keep serving old content.
174	wp_delete_file( $gz_file );
175	}
176
177	return true;
178	}
179
180	/**
181	* Object-cache key for a cache file path.
182	*
183	* @param string $file Absolute path.
184	* @return string
185	*/
186	public static function object_key( $file ) {
187	return 'page_' . md5( wp_normalize_path( (string) $file ) );
188	}
189
190	/**
191	* Copy a rendered page into the object cache, when one is persistent.
192	*
193	* A no-op without Redis/Memcached: storing pages in transients would put
194	* whole HTML documents in `wp_options` and read them back with a database
195	* query, which is slower than the filesystem it is meant to replace.
196	*
197	* @param string $file Absolute path the page was written to.
198	* @param string $html Rendered HTML.
199	* @return bool
200	*/
201	private static function mirror_to_object_cache( $file, $html ) {
202	if ( ! CacheBackend::is_persistent() ) {
203	return false;
204	}
205
206	$ttl = (int) \ABlocks\Helper::get_settings( 'perf_page_cache_ttl', 0 );
207	$ttl = $ttl > 0 ? $ttl * HOUR_IN_SECONDS : DAY_IN_SECONDS;
208
209	return CacheBackend::set( self::object_key( $file ), $html, $ttl, false );
210	}
211
212	/**
213	* Atomically write a file anywhere inside the plugin's uploads directory.
214	*
215	* Exposed so sibling features (the style consolidator) share one
216	* write-then-rename implementation rather than each rolling their own. The
217	* containment check is deliberately against the plugin's uploads root rather
218	* than the page-cache subdirectory, since callers legitimately write to
219	* neighbouring directories — but nothing outside that root.
220	*
221	* @param string $file Absolute target path.
222	* @param string $contents Payload.
223	* @return bool
224	*/
225	public static function put_file( $file, $contents ) {
226	$upload = new FileUpload();
227	$root = wp_normalize_path( untrailingslashit( $upload->get_upload_dir() ) );
228	$path = wp_normalize_path( (string) $file );
229
230	if ( false !== strpos( $path, '../' ) \|\| 0 !== strpos( $path, $root . '/' ) ) {
231	return false;
232	}
233
234	return self::atomic_put( $file, $contents );
235	}
236
237	/**
238	* Delete one cache entry and its siblings.
239	*
240	* @param string $file Absolute path.
241	* @return bool
242	*/
243	public static function delete_file( $file ) {
244	if ( empty( $file ) \|\| ! self::is_inside_base( $file ) ) {
245	return false;
246	}
247	$deleted = false;
248	CacheBackend::delete( self::object_key( $file ) );
249	foreach ( [ $file, $file . '.gz' ] as $target ) {
250	if ( file_exists( $target ) ) {
251	wp_delete_file( $target );
252	$deleted = true;
253	}
254	}
255	return $deleted;
256	}
257
258	/**
259	* Delete every variant cached for a URL.
260	*
261	* @param string $url Absolute or relative URL.
262	* @return bool True when at least one file was removed.
263	*/
264	public static function delete_url( $url ) {
265	$parts = wp_parse_url( $url );
266	if ( empty( $parts ) ) {
267	return false;
268	}
269	$host = isset( $parts['host'] ) ? $parts['host'] : ( isset( $_SERVER['HTTP_HOST'] ) ? wp_unslash( $_SERVER['HTTP_HOST'] ) : '' );
270	$path = isset( $parts['path'] ) ? $parts['path'] : '/';
271
272	$deleted = false;
273	foreach ( [ '', 'mobile' ] as $variant ) {
274	$file = self::file_path( is_string( $host ) ? $host : '', $path, $variant );
275	if ( $file && self::delete_file( $file ) ) {
276	$deleted = true;
277	}
278	}
279	return $deleted;
280	}
281
282	/**
283	* Remove every cached page.
284	*
285	* @return int Files removed.
286	*/
287	public static function flush() {
288	$base = self::base_dir();
289	if ( ! is_dir( $base ) ) {
290	return 0;
291	}
292
293	// Only host directories are removed. Files sitting directly in the base
294	// are the guards (index.php, .htaccess) — and on Apache the .htaccess
295	// carries the tier-3 serve rules, so wiping it on every purge would
296	// quietly demote the site back to PHP-served caching.
297	$removed = 0;
298	// phpcs:ignore WordPress.PHP.NoSilencedErrors.Discouraged -- Return value is checked; a directory vanishing mid-purge is normal and must not warn.
299	$entries = @scandir( $base );
300	if ( false === $entries ) {
301	return 0;
302	}
303
304	foreach ( $entries as $entry ) {
305	if ( '.' === $entry \|\| '..' === $entry ) {
306	continue;
307	}
308	$path = $base . '/' . $entry;
309	if ( is_dir( $path ) && ! is_link( $path ) ) {
310	$removed += self::delete_tree( $path, true );
311	}
312	}
313
314	self::protect_base_dir();
315	return $removed;
316	}
317
318	/**
319	* Delete entries older than the configured TTL.
320	*
321	* @param int $ttl_hours Hours; 0 disables expiry.
322	* @return int Files removed.
323	*/
324	public static function prune( $ttl_hours = null ) {
325	if ( null === $ttl_hours ) {
326	$ttl_hours = (int) \ABlocks\Helper::get_settings( 'perf_page_cache_ttl', 0 );
327	}
328	$ttl_hours = (int) $ttl_hours;
329	if ( $ttl_hours <= 0 ) {
330	return 0;
331	}
332
333	$base = self::base_dir();
334	if ( ! is_dir( $base ) ) {
335	return 0;
336	}
337
338	$cutoff = time() - ( $ttl_hours * HOUR_IN_SECONDS );
339	$removed = 0;
340
341	foreach ( self::iterate_files( $base ) as $path ) {
342	if ( '.gz' === substr( $path, -3 ) ) {
343	continue; // Handled with its parent.
344	}
345	if ( filemtime( $path ) < $cutoff ) {
346	$removed += self::delete_file( $path ) ? 1 : 0;
347	}
348	}
349
350	return $removed;
351	}
352
353	/**
354	* Cached-page count and total size on disk.
355	*
356	* @return array{pages:int, files:int, bytes:int}
357	*/
358	public static function stats() {
359	$base = self::base_dir();
360	$stats = [
361	'pages' => 0,
362	'files' => 0,
363	'bytes' => 0,
364	];
365	if ( ! is_dir( $base ) ) {
366	return $stats;
367	}
368	foreach ( self::iterate_files( $base ) as $path ) {
369	$stats['files']++;
370	$stats['bytes'] += (int) filesize( $path );
371	if ( '.html' === substr( $path, -5 ) ) {
372	$stats['pages']++;
373	}
374	}
375	return $stats;
376	}
377
378	/**
379	* Write the directory guards.
380	*
381	* The .htaccess is defence in depth only — nginx ignores it, and nginx is a
382	* first-class target here. The real protection is the rule that we only ever
383	* cache the fully-anonymous view of an already-public URL, so a leaked file
384	* contains exactly what curl on that URL already returns. See Rules.
385	*/
386	public static function protect_base_dir() {
387	$base = self::base_dir();
388	if ( ! is_dir( $base ) && ! wp_mkdir_p( $base ) ) {
389	return;
390	}
391
392	$index = $base . '/index.php';
393	if ( ! file_exists( $index ) ) {
394	self::atomic_put( $index, "<?php\n// Silence is golden.\n" );
395	}
396
397	$htaccess = $base . '/.htaccess';
398	if ( ! file_exists( $htaccess ) ) {
399	// Tier 3 on Apache replaces this file with serve rules.
400	self::atomic_put( $htaccess, "# aBlocks page cache — served by PHP.\n<Files \"*.html\">\nRequire all denied\n</Files>\n" );
401	}
402	}
403
404	/**
405	* Write a file atomically: temp file in the same directory, then rename.
406	*
407	* Renaming is atomic within a filesystem, so a concurrent reader sees either
408	* the old file or the new one, never a partial write. The temp file is
409	* created in the destination directory rather than the system temp dir,
410	* because rename() is only atomic when both paths share a filesystem.
411	*
412	* @param string $file Target path.
413	* @param string $contents Payload.
414	* @return bool
415	*/
416	private static function atomic_put( $file, $contents ) {
417	$dir = dirname( $file );
418	if ( ! is_dir( $dir ) && ! wp_mkdir_p( $dir ) ) {
419	return false;
420	}
421
422	$tmp = tempnam( $dir, '.ablocks-cache-' );
423	if ( false === $tmp ) {
424	return false;
425	}
426
427	// phpcs:ignore WordPress.WP.AlternativeFunctions.file_system_read_file_put_contents -- WP_Filesystem cannot write-then-rename atomically; see class docblock.
428	$written = file_put_contents( $tmp, $contents, LOCK_EX );
429	if ( false === $written ) {
430	wp_delete_file( $tmp );
431	return false;
432	}
433
434	// tempnam() creates 0600; cache files must be readable by the web server
435	// user when nginx serves them directly (tier 3).
436	// phpcs:ignore WordPress.PHP.NoSilencedErrors.Discouraged -- Best-effort: a failed chmod is recoverable, and a warning printed here would land inside the HTML being cached.
437	@chmod( $tmp, defined( 'FS_CHMOD_FILE' ) ? FS_CHMOD_FILE : 0644 );
438
439	// phpcs:ignore WordPress.PHP.NoSilencedErrors.Discouraged -- Return value is checked; the warning is suppressed because concurrent writers racing on the same page is normal and must stay silent.
440	if ( ! @rename( $tmp, $file ) ) {
441	wp_delete_file( $tmp );
442	return false;
443	}
444
445	return true;
446	}
447
448	/**
449	* Recursively delete a directory's contents.
450	*
451	* @param string $dir Directory, must be inside the cache base.
452	* @param bool $remove_self Remove $dir itself as well.
453	* @return int Files removed.
454	*/
455	private static function delete_tree( $dir, $remove_self = true ) {
456	if ( ! is_dir( $dir ) \|\| ! self::is_inside_base( $dir, true ) ) {
457	return 0;
458	}
459
460	$removed = 0;
461	// phpcs:ignore WordPress.PHP.NoSilencedErrors.Discouraged -- Return value is checked; a directory vanishing mid-purge is normal and must not warn.
462	$entries = @scandir( $dir );
463	if ( false === $entries ) {
464	return 0;
465	}
466
467	foreach ( $entries as $entry ) {
468	if ( '.' === $entry \|\| '..' === $entry ) {
469	continue;
470	}
471	$path = $dir . '/' . $entry;
472	if ( is_link( $path ) ) {
473	// Never follow a symlink out of the cache tree.
474	wp_delete_file( $path );
475	$removed++;
476	continue;
477	}
478	if ( is_dir( $path ) ) {
479	$removed += self::delete_tree( $path, true );
480	continue;
481	}
482	wp_delete_file( $path );
483	$removed++;
484	}
485
486	if ( $remove_self ) {
487	// phpcs:ignore WordPress.PHP.NoSilencedErrors.Discouraged -- Best effort: a non-empty or already-removed directory is not an error here.
488	@rmdir( $dir );
489	}
490
491	return $removed;
492	}
493
494	/**
495	* Yield every file under a directory.
496	*
497	* @param string $dir Directory.
498	* @return \Generator
499	*/
500	private static function iterate_files( $dir ) {
501	// phpcs:ignore WordPress.PHP.NoSilencedErrors.Discouraged -- Return value is checked; a directory vanishing mid-purge is normal and must not warn.
502	$entries = @scandir( $dir );
503	if ( false === $entries ) {
504	return;
505	}
506	foreach ( $entries as $entry ) {
507	if ( '.' === $entry \|\| '..' === $entry ) {
508	continue;
509	}
510	$path = $dir . '/' . $entry;
511	if ( is_link( $path ) ) {
512	continue;
513	}
514	if ( is_dir( $path ) ) {
515	yield from self::iterate_files( $path );
516	continue;
517	}
518	if ( '.html' === substr( $path, -5 ) \|\| '.gz' === substr( $path, -3 ) ) {
519	yield $path;
520	}
521	}
522	}
523
524	/**
525	* Normalise a host for use as a directory name.
526	*
527	* HTTP_HOST is attacker-controlled, so this is strict: lowercase, and only
528	* the characters a hostname may legitimately contain. A port is kept (as
529	* `_`-joined) so :8080 dev sites do not collide with production.
530	*
531	* @param string $host Raw host.
532	* @return string Sanitised host, or '' when unusable.
533	*/
534	private static function sanitize_host( $host ) {
535	$host = strtolower( trim( (string) $host ) );
536	$host = str_replace( ':', '_', $host );
537	$host = preg_replace( '/[^a-z0-9._\-]/', '', $host );
538	$host = trim( (string) $host, '.-_' );
539	if ( '' === $host \|\| strlen( $host ) > 253 ) {
540	return '';
541	}
542	return $host;
543	}
544
545	/**
546	* Split a URL path into validated directory segments.
547	*
548	* @param string $path URL path.
549	* @return string[]\|null Segments, or null when the path is not representable.
550	*/
551	private static function sanitize_path( $path ) {
552	$path = (string) $path;
553	$path = (string) strtok( $path, '?' );
554	$path = (string) strtok( $path, '#' );
555
556	$raw = explode( '/', $path );
557	$out = [];
558
559	foreach ( $raw as $segment ) {
560	if ( '' === $segment ) {
561	continue;
562	}
563	$clean = self::sanitize_segment( $segment );
564	if ( null === $clean ) {
565	return null;
566	}
567	$out[] = $clean;
568	}
569
570	// Absurd depth is either an attack or a misconfiguration; either way it
571	// should not create a thousand-deep directory tree.
572	if ( count( $out ) > 20 ) {
573	return null;
574	}
575
576	return $out;
577	}
578
579	/**
580	* Validate one path segment.
581	*
582	* Percent-decoded first, because nginx matches on its decoded $uri and the
583	* on-disk name must agree with it. Anything that could escape the tree or
584	* confuse the filesystem is rejected outright rather than rewritten —
585	* rewriting would silently map two distinct URLs onto one file.
586	*
587	* @param string $segment Raw segment.
588	* @return string\|null Validated segment, or null to reject the whole path.
589	*/
590	private static function sanitize_segment( $segment ) {
591	$segment = rawurldecode( $segment );
592
593	if ( '' === $segment \|\| '.' === $segment \|\| '..' === $segment ) {
594	return null;
595	}
596	// Null bytes, control characters, separators and Windows-reserved chars.
597	if ( preg_match( '#[\x00-\x1F\x7F/\\\\:*?"<>\|]#', $segment ) ) {
598	return null;
599	}
600	// A leading dot would create hidden directories and can shadow guards
601	// such as .htaccess.
602	if ( '.' === $segment[0] ) {
603	return null;
604	}
605	// Individual path components are capped well below the common 255-byte
606	// filesystem limit.
607	if ( strlen( $segment ) > 200 ) {
608	return null;
609	}
610
611	return $segment;
612	}
613
614	/**
615	* Is a path contained by the cache base directory?
616	*
617	* Compares lexically on a normalised path so it works for files that do not
618	* exist yet (realpath() would return false for those).
619	*
620	* @param string $path Candidate path.
621	* @param bool $allow_base Treat the base directory itself as inside.
622	* @return bool
623	*/
624	private static function is_inside_base( $path, $allow_base = false ) {
625	$base = wp_normalize_path( self::base_dir() );
626	$path = wp_normalize_path( (string) $path );
627
628	if ( false !== strpos( $path, '../' ) ) {
629	return false;
630	}
631	if ( $allow_base && $path === $base ) {
632	return true;
633	}
634	return 0 === strpos( $path, $base . '/' );
635	}
636	}
637