imagify / inc / classes / class-imagify-files-scan.php

class-imagify-files-scan.php in Imagify Image Optimization: Optimize Images | Compress & Convert to WebP/AVIF 2.2.9, at inc/classes/class-imagify-files-scan.php

717 lines 21.1 KB

Raw Download Zip

1	<?php
2
3	/**
4	* Class handling everything that is related to "custom folders optimization".
5	*
6	* @since 1.7
7	* @author Grégory Viguier
8	*/
9	class Imagify_Files_Scan {
10
11	/**
12	* Class version.
13	*
14	* @var string
15	* @since 1.7
16	* @author Grégory Viguier
17	*/
18	const VERSION = '1.1.1';
19
20	/**
21	* Get files (optimizable by Imagify) recursively from a specific folder.
22	*
23	* @since 1.7
24	* @access public
25	* @author Grégory Viguier
26	*
27	* @param string $folder An absolute path to a folder.
28	* @return array\|object An array of absolute paths. A WP_Error object on error.
29	*/
30	public static function get_files_from_folder( $folder ) {
31	$filesystem = imagify_get_filesystem();
32
33	// Formate and validate the folder path.
34	if ( ! is_string( $folder ) ) {
35	return new WP_Error( 'invalid_folder', __( 'Invalid folder.', 'imagify' ) );
36	}
37
38	$folder = realpath( $folder );
39
40	if ( ! $folder ) {
41	return new WP_Error( 'folder_not_exists', __( 'This folder does not exist.', 'imagify' ) );
42	}
43
44	if ( ! $filesystem->is_dir( $folder ) ) {
45	return new WP_Error( 'not_a_folder', __( 'This file is not a folder.', 'imagify' ) );
46	}
47
48	if ( self::is_path_forbidden( trailingslashit( $folder ) ) ) {
49	return new WP_Error( 'folder_forbidden', __( 'This folder is not allowed.', 'imagify' ) );
50	}
51
52	// Finally we made all our validations.
53	if ( $filesystem->is_site_root( $folder ) ) {
54	// For the site's root, we don't look in sub-folders.
55	$dir = new DirectoryIterator( $folder );
56	$dir = new Imagify_Files_Iterator( $dir, false );
57	$images = [];
58
59	foreach ( new IteratorIterator( $dir ) as $file ) {
60	$images[] = $file->getPathname();
61	}
62
63	return $images;
64	}
65
66	/**
67	* 4096 stands for FilesystemIterator::SKIP_DOTS, which was introduced in php 5.3.0.
68	* 8192 stands for FilesystemIterator::UNIX_PATHS, which was introduced in php 5.3.0.
69	*/
70	$dir = new RecursiveDirectoryIterator( $folder, 4096 \| 8192 );
71	$dir = new Imagify_Files_Recursive_Iterator( $dir );
72	$images = new RecursiveIteratorIterator( $dir );
73	$images = array_keys( iterator_to_array( $images ) );
74
75	return $images;
76	}
77
78
79	/** ----------------------------------------------------------------------------------------- */
80	/** FORBIDDEN FOLDERS AND FILES ============================================================= */
81	/** ----------------------------------------------------------------------------------------- */
82
83	/**
84	* Tell if a path is autorized.
85	* When testing a folder, the path MUST have a trailing slash.
86	*
87	* @since 1.7.1
88	* @since 1.8 The path must have a trailing slash if for a folder.
89	* @access public
90	* @author Grégory Viguier
91	*
92	* @param string $file_path A file or folder absolute path.
93	* @return bool
94	*/
95	public static function is_path_autorized( $file_path ) {
96	return ! self::is_path_forbidden( $file_path );
97	}
98
99	/**
100	* Tell if a path is forbidden.
101	* When testing a folder, the path MUST have a trailing slash.
102	*
103	* @since 1.7
104	* @since 1.8 The path must have a trailing slash if for a folder.
105	* @access public
106	* @author Grégory Viguier
107	*
108	* @param string $file_path A file or folder absolute path.
109	* @return bool
110	*/
111	public static function is_path_forbidden( $file_path ) {
112	static $folders;
113
114	$filesystem = imagify_get_filesystem();
115
116	if ( self::is_filename_forbidden( $filesystem->file_name( $file_path ) ) ) {
117	return true;
118	}
119
120	if ( $filesystem->is_symlinked( $file_path ) ) {
121	// Files outside the site's folder are forbidden.
122	return true;
123	}
124
125	if ( ! isset( $folders ) ) {
126	$folders = self::get_forbidden_folders();
127	$folders = array_map( 'strtolower', $folders );
128	$folders = array_flip( $folders );
129	}
130
131	$file_path = self::normalize_path_for_comparison( $file_path );
132
133	if ( isset( $folders[ $file_path ] ) ) {
134	return true;
135	}
136
137	$delim = Imagify_Filesystem::PATTERN_DELIMITER;
138
139	foreach ( self::get_forbidden_folder_patterns() as $pattern ) {
140	if ( preg_match( $delim . '^' . $pattern . $delim, $file_path ) ) {
141	return true;
142	}
143	}
144
145	foreach ( $folders as $folder => $i ) {
146	if ( strpos( $file_path, $folder ) === 0 ) {
147	return true;
148	}
149	}
150
151	return false;
152	}
153
154	/**
155	* Get the list of folders where Imagify won't look for files to optimize.
156	*
157	* @since 1.7
158	* @access public
159	* @author Grégory Viguier
160	*
161	* @return array A list of absolute paths.
162	*/
163	public static function get_forbidden_folders() {
164	static $folders;
165
166	if ( isset( $folders ) ) {
167	return $folders;
168	}
169
170	$filesystem = imagify_get_filesystem();
171	$site_root = $filesystem->get_site_root();
172	$abspath = $filesystem->get_abspath();
173	$folders = [
174	// Server.
175	$site_root . 'cgi-bin', // `cgi-bin`
176	// WordPress.
177	$abspath . 'wp-admin',
178	$abspath . WPINC,
179	WP_CONTENT_DIR . '/mu-plugins', // MU plugins.
180	WP_CONTENT_DIR . '/upgrade', // Upgrade.
181	// Plugins.
182	WP_CONTENT_DIR . '/bps-backup', // BulletProof Security.
183	self::get_ewww_tools_path(), // EWWW: /wp-content/ewww.
184	WP_CONTENT_DIR . '/ngg', // NextGen Gallery.
185	WP_CONTENT_DIR . '/ngg_styles', // NextGen Gallery.
186	WP_CONTENT_DIR . '/w3tc-config', // W3 Total Cache.
187	WP_CONTENT_DIR . '/wfcache', // WP Fastest Cache.
188	WP_CONTENT_DIR . '/wp-rocket-config', // WP Rocket.
189	Imagify_Custom_Folders::get_backup_dir_path(), // Imagify "Custom folders" backup: /imagify-backup.
190	IMAGIFY_PATH, // Imagify plugin: /wp-content/plugins/imagify.
191	self::get_shortpixel_path(), // ShortPixel: /wp-content/uploads/ShortpixelBackups.
192	];
193
194	if ( ! is_multisite() ) {
195	$uploads_dir = $filesystem->get_upload_basedir( true );
196	$ngg_galleries = self::get_ngg_galleries_path();
197
198	if ( $ngg_galleries ) {
199	$folders[] = $ngg_galleries; // NextGen Gallery: /wp-content/gallery.
200	}
201
202	$folders[] = $uploads_dir . 'formidable'; // Formidable Forms: /wp-content/uploads/formidable.
203	$folders[] = get_imagify_backup_dir_path( true ); // Imagify Media Library backup: /wp-content/uploads/backup.
204	$folders[] = self::get_wc_logs_path(); // WooCommerce Logs: /wp-content/uploads/wc-logs.
205	$folders[] = $uploads_dir . 'woocommerce_uploads'; // WooCommerce uploads: /wp-content/uploads/woocommerce_uploads.
206	}
207
208	$folders = array_map( [ $filesystem, 'normalize_dir_path' ], $folders );
209
210	/**
211	* Add folders to the list of forbidden ones.
212	*
213	* @since 1.7
214	* @author Grégory Viguier
215	*
216	* @param array $added_folders List of absolute paths.
217	* @param array $folders List of folders already forbidden.
218	*/
219	$added_folders = apply_filters( 'imagify_add_forbidden_folders', [], $folders );
220	$added_folders = array_filter( (array) $added_folders );
221	$added_folders = array_filter( $added_folders, 'is_string' );
222
223	if ( ! $added_folders ) {
224	return $folders;
225	}
226
227	$added_folders = array_map( [ $filesystem, 'normalize_dir_path' ], $added_folders );
228
229	$folders = array_merge( $folders, $added_folders );
230	$folders = array_flip( array_flip( $folders ) );
231
232	return $folders;
233	}
234
235	/**
236	* Get the list of folder patterns where Imagify won't look for files to optimize. This is meant for paths that are dynamic.
237	* `^` will be prepended to each pattern (aka, the pattern must match an absolute path).
238	* Pattern delimiter is `Imagify_Filesystem::PATTERN_DELIMITER`.
239	* Paths tested against these patterns are lower-cased.
240	*
241	* @since 1.7
242	* @access public
243	* @author Grégory Viguier
244	*
245	* @return array A list of regex patterns.
246	*/
247	public static function get_forbidden_folder_patterns() {
248	static $folders;
249
250	if ( isset( $folders ) ) {
251	return $folders;
252	}
253
254	$folders = [];
255
256	// Media Library: /wp\-content/uploads/(sites/\d+/)?\d{4}/\d{2}/.
257	$folders[] = self::get_media_library_pattern();
258
259	if ( is_multisite() ) {
260	/**
261	* On multisite we can't exclude Imagify's library backup folders, or any other folder located in the uploads folders (created by other plugins): there are too many ways it can fail.
262	* Only exception we're aware of so far is NextGen Gallery, because it provides a clear pattern to use.
263	*/
264	$ngg_galleries = self::get_ngg_galleries_multisite_pattern();
265
266	if ( $ngg_galleries ) {
267	// NextGen Gallery: /wp\-content/uploads/sites/\d+/nggallery/.
268	$folders[] = $ngg_galleries;
269	}
270	}
271
272	/**
273	* Add folder patterns to the list of forbidden ones.
274	* Don't forget to use `Imagify_Files_Scan::normalize_path_for_regex( $path )`!
275	*
276	* @since 1.7
277	* @author Grégory Viguier
278	*
279	* @param array $added_folders List of patterns.
280	* @param array $folders List of patterns already forbidden.
281	*/
282	$added_folders = apply_filters( 'imagify_add_forbidden_folder_patterns', [], $folders );
283	$added_folders = array_filter( (array) $added_folders );
284	$added_folders = array_filter( $added_folders, 'is_string' );
285
286	if ( ! $added_folders ) {
287	return $folders;
288	}
289
290	$folders = array_merge( $folders, $added_folders );
291	$folders = array_flip( array_flip( $folders ) );
292
293	return $folders;
294	}
295
296	/**
297	* Tell if a file/folder name is forbidden.
298	*
299	* @since 1.7
300	* @access public
301	* @author Grégory Viguier
302	*
303	* @param string $file_name A file or folder name.
304	* @return bool
305	*/
306	public static function is_filename_forbidden( $file_name ) {
307	static $file_names;
308
309	if ( ! isset( $file_names ) ) {
310	$file_names = array_flip( self::get_forbidden_file_names() );
311	}
312
313	return isset( $file_names[ strtolower( $file_name ) ] );
314	}
315
316	/**
317	* Get the list of file names that Imagify won't optimize.
318	* It can contain folder names. Names are case-lowered.
319	*
320	* @since 1.7
321	* @access public
322	* @author Grégory Viguier
323	*
324	* @return array A list of file names
325	*/
326	public static function get_forbidden_file_names() {
327	static $file_names;
328
329	if ( isset( $file_names ) ) {
330	return $file_names;
331	}
332
333	$file_names = [
334	'.',
335	'..',
336	'.DS_Store',
337	'.git',
338	'.svn',
339	'backup',
340	'backups',
341	'cache',
342	'lang',
343	'langs',
344	'languages',
345	'node_modules',
346	'Thumbs.db',
347	];
348	$file_names = array_map( 'strtolower', $file_names );
349
350	/**
351	* Add file names to the list of forbidden ones.
352	*
353	* @since 1.7
354	* @author Grégory Viguier
355	*
356	* @param array $added_file_names List of file names.
357	* @param array $file_names List of file names already forbidden.
358	*/
359	$added_file_names = apply_filters( 'imagify_add_forbidden_file_names', [], $file_names );
360
361	if ( ! $added_file_names \|\| ! is_array( $added_file_names ) ) {
362	return $file_names;
363	}
364
365	$added_file_names = array_filter( $added_file_names, 'is_string' );
366	$added_file_names = array_map( 'strtolower', $added_file_names );
367
368	$file_names = array_merge( $file_names, $added_file_names );
369	$file_names = array_flip( array_flip( $file_names ) );
370
371	return $file_names;
372	}
373
374
375	/** ----------------------------------------------------------------------------------------- */
376	/** PLACEHOLDERS ============================================================================ */
377	/** ----------------------------------------------------------------------------------------- */
378
379	/**
380	* Add a placeholder to a path.
381	*
382	* @since 1.7
383	* @access public
384	* @author Grégory Viguier
385	*
386	* @param string $file_path An absolute path.
387	* @return string A "placeholdered" path.
388	*/
389	public static function add_placeholder( $file_path ) {
390	$file_path = wp_normalize_path( $file_path );
391	$locations = self::get_placeholder_paths();
392
393	foreach ( $locations as $placeholder => $location_path ) {
394	if ( strpos( $file_path, $location_path ) === 0 ) {
395	return preg_replace( '@^' . preg_quote( $location_path, '@' ) . '@', $placeholder, $file_path );
396	}
397	}
398
399	// Should not happen.
400	return $file_path;
401	}
402
403	/**
404	* Change a path with a placeholder into a real path or URL.
405	*
406	* @since 1.7
407	* @access public
408	* @author Grégory Viguier
409	*
410	* @param string $file_path A path with a placeholder.
411	* @param string $type What to return: 'path' or 'url'.
412	* @return string An absolute path or a URL.
413	*/
414	public static function remove_placeholder( $file_path, $type = 'path' ) {
415	if ( 'path' === $type ) {
416	$locations = self::get_placeholder_paths();
417	} else {
418	$locations = self::get_placeholder_urls();
419	}
420
421	foreach ( $locations as $placeholder => $location_path ) {
422	if ( strpos( $file_path, $placeholder ) === 0 ) {
423	return preg_replace( '@^' . preg_quote( $placeholder, '@' ) . '@', $location_path, $file_path );
424	}
425	}
426
427	// Should not happen.
428	return $file_path;
429	}
430
431	/**
432	* Get array of pairs of placeholder => corresponding path.
433	*
434	* @since 1.7
435	* @access public
436	* @author Grégory Viguier
437	*
438	* @return array
439	*/
440	public static function get_placeholder_paths() {
441	static $replacements;
442
443	if ( isset( $replacements ) ) {
444	return $replacements;
445	}
446
447	$filesystem = imagify_get_filesystem();
448	$replacements = [
449	'{{PLUGINS}}/' => WP_PLUGIN_DIR,
450	'{{MU_PLUGINS}}/' => WPMU_PLUGIN_DIR,
451	'{{THEMES}}/' => WP_CONTENT_DIR . '/themes',
452	'{{UPLOADS}}/' => $filesystem->get_main_upload_basedir(),
453	'{{CONTENT}}/' => WP_CONTENT_DIR,
454	'{{ROOT}}/' => $filesystem->get_site_root(),
455	];
456	$replacements = array_map( [ $filesystem, 'normalize_dir_path' ], $replacements );
457
458	return $replacements;
459	}
460
461	/**
462	* Get array of pairs of placeholder => corresponding URL.
463	*
464	* @since 1.7
465	* @access public
466	* @author Grégory Viguier
467	*
468	* @return array
469	*/
470	public static function get_placeholder_urls() {
471	static $replacements;
472
473	if ( isset( $replacements ) ) {
474	return $replacements;
475	}
476
477	$filesystem = imagify_get_filesystem();
478	$replacements = [
479	'{{PLUGINS}}/' => plugins_url( '/' ),
480	'{{MU_PLUGINS}}/' => plugins_url( '/', WPMU_PLUGIN_DIR . '/.' ),
481	'{{THEMES}}/' => content_url( 'themes/' ),
482	'{{UPLOADS}}/' => $filesystem->get_main_upload_baseurl(),
483	'{{CONTENT}}/' => content_url( '/' ),
484	'{{ROOT}}/' => $filesystem->get_site_root_url(),
485	];
486
487	return $replacements;
488	}
489
490	/**
491	* A file_exists() for paths with a placeholder.
492	*
493	* @since 1.7
494	* @access public
495	* @author Grégory Viguier
496	*
497	* @param string $file_path The file path.
498	* @return bool
499	*/
500	public static function placeholder_path_exists( $file_path ) {
501	return imagify_get_filesystem()->is_readable( self::remove_placeholder( $file_path ) );
502	}
503
504
505	/** ----------------------------------------------------------------------------------------- */
506	/** PATHS =================================================================================== */
507	/** ----------------------------------------------------------------------------------------- */
508
509	/**
510	* Get the path to NextGen galleries on monosites.
511	*
512	* @since 1.7
513	* @access public
514	* @author Grégory Viguier
515	*
516	* @return string\|bool An absolute path. False if it can't be retrieved.
517	*/
518	public static function get_ngg_galleries_path() {
519	$galleries_path = get_site_option( 'ngg_options' );
520
521	if ( empty( $galleries_path['gallerypath'] ) ) {
522	return false;
523	}
524
525	$filesystem = imagify_get_filesystem();
526	$galleries_path = $filesystem->normalize_dir_path( $galleries_path['gallerypath'] );
527	$galleries_path = trim( $galleries_path, '/' ); // Something like `wp-content/gallery`.
528
529	$ngg_root = defined( 'NGG_GALLERY_ROOT_TYPE' ) ? NGG_GALLERY_ROOT_TYPE : 'site';
530
531	if ( $galleries_path && 'content' === $ngg_root ) {
532	$ngg_root = $filesystem->normalize_dir_path( WP_CONTENT_DIR );
533	$ngg_root = trim( $ngg_root, '/' ); // Something like `abs-path/to/wp-content`.
534
535	$exploded_root = explode( '/', $ngg_root );
536	$exploded_galleries = explode( '/', $galleries_path );
537	$first_gallery_dirname = reset( $exploded_galleries );
538	$last_root_dirname = end( $exploded_root );
539
540	if ( $last_root_dirname === $first_gallery_dirname ) {
541	array_shift( $exploded_galleries );
542	$galleries_path = implode( '/', $exploded_galleries );
543	}
544	}
545
546	if ( 'content' === $ngg_root ) {
547	$ngg_root = $filesystem->normalize_dir_path( WP_CONTENT_DIR );
548	} else {
549	$ngg_root = $filesystem->get_abspath();
550	}
551
552	if ( strpos( $galleries_path, $ngg_root ) !== 0 ) {
553	$galleries_path = $ngg_root . $galleries_path;
554	}
555
556	return $galleries_path . '/';
557	}
558
559	/**
560	* Get the path to WooCommerce logs on monosites.
561	*
562	* @since 1.7
563	* @access public
564	* @author Grégory Viguier
565	*
566	* @return string An absolute path.
567	*/
568	public static function get_wc_logs_path() {
569	if ( defined( 'WC_LOG_DIR' ) ) {
570	return WC_LOG_DIR;
571	}
572
573	return imagify_get_filesystem()->get_upload_basedir( true ) . 'wc-logs/';
574	}
575
576	/**
577	* Get the path to EWWW optimization tools.
578	* It is the same for all sites on multisite.
579	*
580	* @since 1.7
581	* @access public
582	* @author Grégory Viguier
583	*
584	* @return string An absolute path.
585	*/
586	public static function get_ewww_tools_path() {
587	if ( defined( 'EWWW_IMAGE_OPTIMIZER_TOOL_PATH' ) ) {
588	return EWWW_IMAGE_OPTIMIZER_TOOL_PATH;
589	}
590
591	return WP_CONTENT_DIR . '/ewww/';
592	}
593
594	/**
595	* Get the path to ShortPixel backup folder.
596	* It is the same for all sites on multisite (and yes, you'll get a surprise if your upload base dir -aka uploads/sites/12/- is not 2 folders deeper than theuploads folder).
597	*
598	* @since 1.8
599	* @access public
600	* @author Grégory Viguier
601	*
602	* @return string An absolute path.
603	*/
604	public static function get_shortpixel_path() {
605	if ( defined( 'SHORTPIXEL_BACKUP_FOLDER' ) ) {
606	return trailingslashit( SHORTPIXEL_BACKUP_FOLDER );
607	}
608
609	$filesystem = imagify_get_filesystem();
610	$path = $filesystem->get_upload_basedir( true );
611	$path = is_main_site() ? $path : $filesystem->dir_path( $filesystem->dir_path( $path ) );
612
613	return $path . 'ShortpixelBackups/';
614	}
615
616
617	/** ----------------------------------------------------------------------------------------- */
618	/** REGEX PATTERNS ========================================================================== */
619	/** ----------------------------------------------------------------------------------------- */
620
621	/**
622	* Get the regex pattern used to match the paths to the media library.
623	* Pattern delimiter is `Imagify_Filesystem::PATTERN_DELIMITER`.
624	* Paths tested against these patterns are lower-cased.
625	*
626	* @since 1.8
627	* @access public
628	* @author Grégory Viguier
629	*
630	* @return string Something like `/wp\-content/uploads/(sites/\d+/)?\d{4}/\d{2}/`.
631	*/
632	public static function get_media_library_pattern() {
633	$filesystem = imagify_get_filesystem();
634	$uploads_dir = self::normalize_path_for_regex( $filesystem->get_main_upload_basedir() );
635
636	if ( ! is_multisite() ) {
637	if ( get_option( 'uploads_use_yearmonth_folders' ) ) {
638	// In year/month folders.
639	return $uploads_dir . '\d{4}/\d{2}/';
640	}
641
642	// Not in year/month folders.
643	return $uploads_dir . '[^/]+$';
644	}
645
646	$pattern = $filesystem->get_multisite_uploads_subdir_pattern();
647
648	if ( get_option( 'uploads_use_yearmonth_folders' ) ) {
649	// In year/month folders.
650	return $uploads_dir . '(' . $pattern . ')?\d{4}/\d{2}/';
651	}
652
653	// Not in year/month folders.
654	return $uploads_dir . '(' . $pattern . ')?[^/]+$';
655	}
656
657	/**
658	* Get the regex pattern used to match the paths to NextGen galleries on multisite.
659	* Pattern delimiter is `Imagify_Filesystem::PATTERN_DELIMITER`.
660	* Paths tested against these patterns are lower-cased.
661	*
662	* @since 1.8
663	* @access public
664	* @author Grégory Viguier
665	*
666	* @return string\|bool Something like `/wp-content/uploads/sites/\d+/nggallery/`. False if it can't be retrieved.
667	*/
668	public static function get_ngg_galleries_multisite_pattern() {
669	$galleries_path = self::get_ngg_galleries_path(); // Something like `wp-content/uploads/sites/%BLOG_ID%/nggallery/`.
670
671	if ( ! $galleries_path ) {
672	return false;
673	}
674
675	$galleries_path = self::normalize_path_for_regex( $galleries_path );
676	$galleries_path = str_replace( [ '%blog_name%', '%blog_id%' ], [ '.+', '\d+' ], $galleries_path );
677
678	return $galleries_path;
679	}
680
681
682	/** ----------------------------------------------------------------------------------------- */
683	/** NORMALIZATION TOOLS ===================================================================== */
684	/** ----------------------------------------------------------------------------------------- */
685
686	/**
687	* Normalize a file path, aiming for path comparison.
688	* The path is normalized and case-lowered.
689	*
690	* @since 1.7
691	* @since 1.8 No trailing slash anymore, because it can be used for files.
692	* @access public
693	* @author Grégory Viguier
694	*
695	* @param string $file_path The file path.
696	* @return string The normalized file path.
697	*/
698	public static function normalize_path_for_comparison( $file_path ) {
699	return strtolower( wp_normalize_path( $file_path ) );
700	}
701
702	/**
703	* Normalize a file path, aiming for use in a regex pattern.
704	* The path is normalized, case-lowered, and escaped.
705	*
706	* @since 1.8
707	* @access public
708	* @author Grégory Viguier
709	*
710	* @param string $file_path The file path.
711	* @return string The normalized file path.
712	*/
713	public static function normalize_path_for_regex( $file_path ) {
714	return preg_quote( imagify_get_filesystem()->normalize_path_for_comparison( $file_path ), Imagify_Filesystem::PATTERN_DELIMITER );
715	}
716	}
717