core.php — Media Cleaner: Clean your WordPress! 6.1.5

media-cleaner / classes / core.php

media-cleaner / classes Last commit date

parsers 5 years ago admin.php 5 years ago core.php 5 years ago engine.php 5 years ago init.php 5 years ago parsers.php 5 years ago rest.php 5 years ago support.php 5 years ago ui.php 5 years ago

core.php

1186 lines

1	<?php
2
3	class Meow_WPMC_Core {
4
5	public $admin = null;
6	public $is_rest = false;
7	public $is_cli = false;
8	public $is_pro = false;
9	public $engine = null;
10	public $catch_timeout = true; // This will halt the plugin before reaching the PHP timeout.
11	private $regex_file = '/[A-Za-z0-9-_,.\s]+[.]{1}(MIMETYPES)/';
12	public $types = "jpg\|jpeg\|jpe\|gif\|png\|tiff\|bmp\|csv\|pdf\|xls\|xlsx\|doc\|docx\|odt\|wpd\|rtf\|tiff\|mp3\|mp4\|mov\|wav\|lua";
13	public $current_method = 'media';
14	private $refcache = array();
15	public $servername = null; // meowapps.com (site URL without http/https)
16	public $site_url = null; // https://meowapps.com
17	public $upload_path = null; // /www/wp-content/uploads (path to uploads)
18	public $upload_url = null; // wp-content/uploads (uploads without domain)
19	private $check_content = null;
20	private $debug_logs = null;
21	private $multilingual = false;
22	private $languages = array();
23	private $allow_usage = null;
24	private $allow_setup = null;
25
26	public function __construct() {
27	add_action( 'plugins_loaded', array( $this, 'init' ) );
28	}
29
30	function init() {
31
32	// Variables
33	$this->site_url = get_site_url();
34	$this->multilingual = $this->is_multilingual();
35	$this->languages = $this->get_languages();
36	$this->current_method = get_option( 'wpmc_method', 'media' );
37	$this->regex_file = str_replace( "MIMETYPES", $this->types, $this->regex_file );
38	$this->servername = str_replace( 'http://', '', str_replace( 'https://', '', $this->site_url ) );
39	$uploaddir = wp_upload_dir();
40	$this->upload_path = $uploaddir['basedir'];
41	$this->upload_url = substr( $uploaddir['baseurl'], 1 + strlen( $this->site_url ) );
42	$this->check_content = get_option( 'wpmc_content', true );
43	$this->debug_logs = get_option( 'wpmc_debuglogs', false );
44	$this->is_rest = MeowCommon_Helpers::is_rest();
45	$this->is_cli = defined( 'WP_CLI' ) && WP_CLI;
46	global $wpmc;
47	$wpmc = $this;
48
49	// Check the roles
50	$this->allow_usage = apply_filters( 'wpmc_allow_usage', current_user_can( 'administrator' ) );
51	$this->allow_setup = apply_filters( 'wpmc_allow_setup', current_user_can( 'manage_options' ) );
52	if ( !$this->is_cli && !$this->allow_usage ) {
53	return;
54	}
55
56	// Language
57	load_plugin_textdomain( WPMC_DOMAIN, false, basename( WPMC_PATH ) . '/languages' );
58
59	// Admin
60	$this->admin = new Meow_WPMC_Admin( $this->allow_setup );
61
62	// Advanced core
63	if ( class_exists( 'MeowPro_WPMC_Core' ) ) {
64	new MeowPro_WPMC_Core( $this );
65	}
66
67	// Install hooks and engine on;y if they might be used
68	if ( is_admin() \|\| $this->is_rest \|\| $this->is_cli ) {
69	add_action( 'wpmc_initialize_parsers', array( $this, 'initialize_parsers' ), 10, 0 );
70	add_filter( 'wp_unique_filename', array( $this, 'wp_unique_filename' ), 10, 3 );
71	$this->engine = new Meow_WPMC_Engine( $this, $this->admin );
72	}
73
74	// Only for REST
75	if ( $this->is_rest ) {
76	new Meow_WPMC_Rest( $this, $this->admin );
77	}
78
79	if ( is_admin() ) {
80	new Meow_WPMC_UI( $this, $this->admin );
81	}
82	}
83
84	function initialize_parsers() {
85	include_once( 'parsers.php' );
86	new Meow_WPMC_Parsers();
87	}
88
89	function deepsleep( $seconds ) {
90	$start_time = time();
91	while( true ) {
92	if ( ( time() - $start_time ) > $seconds ) {
93	return false;
94	}
95	get_post( array( 'posts_per_page' => 50 ) );
96	}
97	}
98
99	private $start_time;
100	private $time_elapsed = 0;
101	private $item_scan_avg_time = 0;
102	private $wordpress_init_time = 0.5;
103	private $max_execution_time;
104	private $items_checked = 0;
105	private $items_count = 0;
106
107	function get_max_execution_time() {
108	if ( isset( $this->max_execution_time ) )
109	return $this->max_execution_time;
110
111	$this->max_execution_time = ini_get( "max_execution_time" );
112	if ( empty( $this->max_execution_time ) \|\| $this->max_execution_time < 5 )
113	$this->max_execution_time = 30;
114
115	return $this->max_execution_time;
116	}
117
118	function timeout_check_start( $count ) {
119	$this->start_time = time();
120	$this->items_count = $count;
121	$this->get_max_execution_time();
122	}
123
124	function timeout_get_elapsed() {
125	return $this->time_elapsed . 'ms';
126	}
127
128	function timeout_check() {
129	$this->time_elapsed = time() - $this->start_time;
130	$this->time_remaining = $this->max_execution_time - $this->wordpress_init_time - $this->time_elapsed;
131	if ( $this->catch_timeout ) {
132	if ( $this->time_remaining - $this->item_scan_avg_time < 0 ) {
133	error_log("Media Cleaner Timeout! Check the Media Cleaner logs for more info.");
134	$this->log( "😵 Timeout! Some info for debug:" );
135	$this->log( "🍀 Elapsed time: $this->time_elapsed" );
136	$this->log( "🍀 WP init time: $this->wordpress_init_time" );
137	$this->log( "🍀 Remaining time: $this->time_remaining" );
138	$this->log( "🍀 Scan time per item: $this->item_scan_avg_time" );
139	$this->log( "🍀 PHP max_execution_time: $this->max_execution_time" );
140	header("HTTP/1.0 408 Request Timeout");
141	exit;
142	}
143	}
144	}
145
146	function timeout_check_additem() {
147	$this->items_checked++;
148	$this->time_elapsed = time() - $this->start_time;
149	$this->item_scan_avg_time = ceil( ( $this->time_elapsed / $this->items_checked ) * 10 ) / 10;
150	}
151
152	// This checks if a new uploaded filename isn't the same one as a currently
153	// filename in the trash (that would cause issues)
154	function wp_unique_filename( $filename, $ext, $dir ) {
155	$fullpath = trailingslashit( $dir ) . $filename;
156	$relativepath = $this->clean_uploaded_filename( $fullpath );
157	$trashfilepath = trailingslashit( $this->get_trashdir() ) . $relativepath;
158	if ( file_exists( $trashfilepath ) ) {
159	$path_parts = pathinfo( $fullpath );
160	$filename_noext = $path_parts['filename'];
161	$new_filename = $filename_noext . '-' . date('Ymd-His', time()) . '.' . $path_parts['extension'];
162	//error_log( 'POTENTIALLY TRASH PATH: ' . $trashfilepath );
163	//error_log( 'POTENTIALLY NEW FILE: ' . $new_filename );
164	return $new_filename;
165	}
166	return $filename;
167	}
168
169	function array_to_ids_or_urls( &$meta, &$ids, &$urls ) {
170	foreach ( $meta as $k => $m ) {
171	if ( is_numeric( $m ) ) {
172	// Probably a Media ID
173	if ( $m > 0 )
174	array_push( $ids, $m );
175	}
176	else if ( is_array( $m ) ) {
177	// If it's an array with a width, probably that the index is the Media ID
178	if ( isset( $m['width'] ) && is_numeric( $k ) ) {
179	if ( $k > 0 )
180	array_push( $ids, $k );
181	}
182	}
183	else if ( !empty( $m ) ) {
184	// If it's a string, maybe it's a file (with an extension)
185	if ( preg_match( $this->regex_file, $m ) )
186	array_push( $urls, $m );
187	}
188	}
189	}
190
191	function get_favicon() {
192	// Yoast SEO plugin
193	$vals = get_option( 'wpseo_titles' );
194	if ( !empty( $vals ) ) {
195	$url = $vals['company_logo'];
196	if ( $this->is_url( $url ) )
197	return $this->clean_url( $url );
198	}
199	}
200
201	function get_urls_from_html( $html ) {
202	if ( empty( $html ) )
203	return array();
204
205	// Proposal/fix by @copytrans
206	// Discussion: https://wordpress.org/support/topic/bug-in-core-php/#post-11647775
207	$html = mb_convert_encoding( $html, 'HTML-ENTITIES', 'UTF-8' );
208
209	// Resolve src-set and shortcodes
210	if ( !get_option( 'wpmc_shortcodes_disabled', false ) )
211	$html = do_shortcode( $html );
212
213	// TODO: Since WP 5.5, wp_filter_content_tags should be used instead of wp_make_content_images_responsive.
214	$html = function_exists( 'wp_filter_content_tags' ) ? wp_filter_content_tags( $html ) : wp_make_content_images_responsive( $html );
215
216	// Create the DOM Document
217	if ( !class_exists("DOMDocument") ) {
218	error_log( 'Media Cleaner: The DOM extension for PHP is not installed.' );
219	throw new Error( 'The DOM extension for PHP is not installed.' );
220	}
221
222	libxml_use_internal_errors(true);
223	$dom = new DOMDocument();
224	@$dom->loadHTML( $html );
225	libxml_clear_errors();
226	$results = array();
227
228	// <meta> tags in <head> area
229	$metas = $dom->getElementsByTagName( 'meta' );
230	foreach ( $metas as $meta ) {
231	$property = $meta->getAttribute( 'property' );
232	if ( $property == 'og:image' \|\| $property == 'og:image:secure_url' \|\| $property == 'twitter:image' ) {
233	$url = $meta->getAttribute( 'content' );
234	if ( $this->is_url( $url ) ) {
235	$src = $this->clean_url( $url );
236	if ( !empty( $src ) ) {
237	array_push( $results, $src );
238	}
239	}
240	}
241	}
242
243	// IFrames (by Mike Meinz)
244	$iframes = $dom->getElementsByTagName( 'iframe' );
245	foreach( $iframes as $iframe ) {
246	$iframe_src = $iframe->getAttribute( 'src' );
247	// Ignore if the iframe src is not on this server
248	if ( ( strpos( $iframe_src, $this->servername ) !== false) \|\| ( substr( $iframe_src, 0, 1 ) == "/" ) ) {
249	// Create a new DOM Document to hold iframe
250	$iframe_doc = new DOMDocument();
251	// Load the url's contents into the DOM
252	libxml_use_internal_errors( true ); // ignore html formatting problems
253	$rslt = $iframe_doc->loadHTMLFile( $iframe_src );
254	libxml_clear_errors();
255	libxml_use_internal_errors( false );
256	if ( $rslt ) {
257	// Get the resulting html
258	$iframe_html = $iframe_doc->saveHTML();
259	if ( $iframe_html !== false ) {
260	// Scan for links in the iframe
261	$iframe_urls = $this->get_urls_from_html( $iframe_html ); // Recursion
262	if ( !empty( $iframe_urls ) ) {
263	$results = array_merge( $results, $iframe_urls );
264	}
265	}
266	}
267	else {
268	$this->log( '🚫 Failed to load iframe: ' . $iframe_src );
269	}
270	}
271	}
272
273
274	// Images: src, srcset
275	$imgs = $dom->getElementsByTagName( 'img' );
276	foreach ( $imgs as $img ) {
277	//error_log($img->getAttribute('src'));
278	$src = $this->clean_url( $img->getAttribute('src') );
279	array_push( $results, $src );
280	$srcset = $img->getAttribute('srcset');
281	if ( !empty( $srcset ) ) {
282	$setImgs = explode( ',', trim( $srcset ) );
283	foreach ( $setImgs as $setImg ) {
284	$finalSetImg = explode( ' ', trim( $setImg ) );
285	if ( is_array( $finalSetImg ) ) {
286	array_push( $results, $this->clean_url( $finalSetImg[0] ) );
287	}
288	}
289	}
290	}
291
292	// Videos: src
293	$videos = $dom->getElementsByTagName( 'video' );
294	foreach ( $videos as $video ) {
295	//error_log($video->getAttribute('src'));
296	$src = $this->clean_url( $video->getAttribute('src') );
297	array_push( $results, $src );
298	}
299
300	// Links, href
301	$urls = $dom->getElementsByTagName( 'a' );
302	foreach ( $urls as $url ) {
303	$url_href = $url->getAttribute('href'); // mm change
304	if ( $this->is_url( $url_href ) ) { // mm change
305	$src = $this->clean_url( $url_href ); // mm change
306	if ( !empty( $src ) )
307	array_push( $results, $src );
308	}
309	}
310
311	// <link> tags in <head> area
312	$urls = $dom->getElementsByTagName( 'link' );
313	foreach ( $urls as $url ) {
314	$url_href = $url->getAttribute( 'href' );
315	if ( $this->is_url( $url_href ) ) {
316	$src = $this->clean_url( $url_href );
317	if ( !empty( $src ) ) {
318	array_push( $results, $src );
319	}
320	}
321	}
322
323	// PDF
324	preg_match_all( "/((https?:\/\/)?[^\\&\#\[\] \"\?]+\.pdf)/", $html, $res );
325	if ( !empty( $res ) && isset( $res[1] ) && count( $res[1] ) > 0 ) {
326	foreach ( $res[1] as $url ) {
327	if ( $this->is_url($url) )
328	array_push( $results, $this->clean_url( $url ) );
329	}
330	}
331
332	// Background images
333	preg_match_all( "/url\(\'?\"?((https?:\/\/)?[^\\&\#\[\] \"\?]+\.(jpe?g\|gif\|png))\'?\"?/", $html, $res );
334	if ( !empty( $res ) && isset( $res[1] ) && count( $res[1] ) > 0 ) {
335	foreach ( $res[1] as $url ) {
336	if ( $this->is_url($url) )
337	array_push( $results, $this->clean_url( $url ) );
338	}
339	}
340
341	return $results;
342	}
343
344	// Parse a meta, visit all the arrays, look for the attributes, fill $ids and $urls arrays
345	function get_from_meta( $meta, $lookFor, &$ids, &$urls ) {
346	foreach ( $meta as $key => $value ) {
347	if ( is_object( $value ) \|\| is_array( $value ) )
348	$this->get_from_meta( $value, $lookFor, $ids, $urls );
349	else if ( in_array( $key, $lookFor ) ) {
350	if ( empty( $value ) )
351	continue;
352	else if ( is_numeric( $value ) ) {
353	// It this an ID?
354	array_push( $ids, $value );
355	}
356	else {
357	if ( $this->is_url( $value ) ) {
358	// Is this an URL?
359	array_push( $urls, $this->clean_url( $value ) );
360	}
361	else {
362	// Is this an array of IDs, encoded as a string? (like "20,13")
363	$pieces = explode( ',', $value );
364	foreach ( $pieces as $pval ) {
365	if ( is_numeric( $pval ) ) {
366	array_push( $ids, $pval );
367	}
368	}
369	}
370	}
371	}
372	}
373	}
374
375	function get_images_from_themes( &$ids, &$urls ) {
376	// USE CURRENT THEME AND WP API
377	$ch = get_custom_header();
378	if ( !empty( $ch ) && !empty( $ch->url ) ) {
379	array_push( $urls, $this->clean_url( $ch->url ) );
380	}
381	if ( $this->is_url( $ch->thumbnail_url ) ) {
382	array_push( $urls, $this->clean_url( $ch->thumbnail_url ) );
383	}
384	if ( !empty( $ch ) && !empty( $ch->attachment_id ) ) {
385	array_push( $ids, $ch->attachment_id );
386	}
387	$cl = get_custom_logo();
388	if ( $this->is_url( $cl ) ) {
389	$urls = array_merge( $this->get_urls_from_html( $cl ), $urls );
390	}
391	$custom_logo = get_theme_mod( 'custom_logo' );
392	if ( !empty( $custom_logo ) && is_numeric( $custom_logo ) ) {
393	array_push( $ids, (int)$custom_logo );
394	}
395	$si = get_site_icon_url();
396	if ( $this->is_url( $si ) ) {
397	array_push( $urls, $this->clean_url( $si ) );
398	}
399	$si_id = get_option( 'site_icon' );
400	if ( !empty( $si_id ) && is_numeric( $si_id ) ) {
401	array_push( $ids, (int)$si_id );
402	}
403	$cd = get_background_image();
404	if ( $this->is_url( $cd ) ) {
405	array_push( $urls, $this->clean_url( $cd ) );
406	}
407	$photography_hero_image = get_theme_mod( 'photography_hero_image' );
408	if ( !empty( $photography_hero_image ) ) {
409	array_push( $ids, $photography_hero_image );
410	}
411	$author_profile_picture = get_theme_mod( 'author_profile_picture' );
412	if ( !empty( $author_profile_picture ) ) {
413	array_push( $ids, $author_profile_picture );
414	}
415	if ( function_exists ( 'get_uploaded_header_images' ) ) {
416	$header_images = get_uploaded_header_images();
417	if ( !empty( $header_images ) ) {
418	foreach ( $header_images as $hi ) {
419	if ( !empty ( $hi['attachment_id'] ) ) {
420	array_push( $ids, $hi['attachment_id'] );
421	}
422	}
423	}
424	}
425	}
426
427	function logs_directory_check() {
428	if ( !file_exists( WPMC_PATH . '/logs/' ) ) {
429	mkdir( WPMC_PATH . '/logs/', 0777 );
430	}
431	}
432
433	function log( $data = null, $force = false ) {
434	if ( !$this->debug_logs && !$force )
435	return;
436	$this->logs_directory_check();
437	$fh = @fopen( WPMC_PATH . '/logs/media-cleaner.log', 'a' );
438	if ( !$fh )
439	return false;
440	$date = date( "Y-m-d H:i:s" );
441	if ( is_null( $data ) )
442	fwrite( $fh, "\n" );
443	else
444	fwrite( $fh, "$date: {$data}\n" );
445	fclose( $fh );
446	return true;
447	}
448
449	/**
450	*
451	* HELPERS
452	*
453	*/
454
455	function get_trashdir() {
456	return trailingslashit( $this->upload_path ) . 'wpmc-trash';
457	}
458
459	function get_trashurl() {
460	return trailingslashit( $this->upload_url ) . 'wpmc-trash';
461	}
462
463	/**
464	*
465	* I18N RELATED HELPERS
466	*
467	*/
468
469	function is_multilingual() {
470	return function_exists( 'icl_object_id' );
471	}
472
473	function get_languages() {
474	$results = array();
475	if ( $this->is_multilingual() ) {
476	$languages = icl_get_languages();
477	foreach ( $languages as $language ) {
478	if ( isset( $language['code'] ) ) {
479	array_push( $results, $language['code'] );
480	}
481	else if ( isset( $language['language_code'] ) ) {
482	array_push( $results, $language['language_code'] );
483	}
484	}
485	}
486	return $results;
487	}
488
489	function get_translated_media_ids( $mediaId ) {
490	$translated_ids = array();
491	foreach ( $this->languages as $language ) {
492	$id = apply_filters( 'wpml_object_id', $mediaId, 'attachment', false, $language );
493	if ( !empty( $id ) ) {
494	array_push( $translated_ids, $id );
495	}
496	}
497	return $translated_ids;
498	}
499
500	/**
501	*
502	* DELETE / SCANNING / RESET
503	*
504	*/
505
506	function recover_file( $path ) {
507	$originalPath = trailingslashit( $this->upload_path ) . $path;
508	$trashPath = trailingslashit( $this->get_trashdir() ) . $path;
509	if ( !file_exists( $trashPath ) ) {
510	$this->log( "🚫 The file $originalPath actually does not exist in the trash." );
511	return true;
512	}
513	$path_parts = pathinfo( $originalPath );
514	if ( !file_exists( $path_parts['dirname'] ) && !wp_mkdir_p( $path_parts['dirname'] ) ) {
515	die( 'Failed to create folder.' );
516	}
517	if ( !rename( $trashPath, $originalPath ) ) {
518	die( 'Failed to move the file.' );
519	}
520	return true;
521	}
522
523	function recover( $id ) {
524	global $wpdb;
525	$table_name = $wpdb->prefix . "mclean_scan";
526	$issue = $wpdb->get_row( $wpdb->prepare( "SELECT * FROM $table_name WHERE id = %d", $id ), OBJECT );
527	$issue->path = stripslashes( $issue->path );
528
529	// Files
530	if ( $issue->type == 0 ) {
531	$this->recover_file( $issue->path );
532	$wpdb->query( $wpdb->prepare( "UPDATE $table_name SET deleted = 0 WHERE id = %d", $id ) );
533	$this->log( "�
534	Recovered {$issue->path}." );
535	return true;
536	}
537	// Media
538	else if ( $issue->type == 1 ) {
539
540	// If there is no file attached, doesn't handle the files
541	$fullpath = get_attached_file( $issue->postId );
542	if ( empty( $fullpath ) ) {
543	$this->log( "🚫 Media #{$issue->postId} does not have attached file anymore." );
544	error_log( "Media #{$issue->postId} does not have attached file anymore." );
545	return false;
546	}
547
548	$paths = $this->get_paths_from_attachment( $issue->postId );
549	foreach ( $paths as $path ) {
550	if ( !$this->recover_file( $path ) ) {
551	$this->log( "🚫 Could not recover $path." );
552	error_log( "Media Cleaner: Could not recover $path." );
553	}
554	}
555	if ( !wp_untrash_post( $issue->postId ) ) {
556	$this->log( "🚫 Failed to Untrash Post {$issue->postId} (but deleted it from Cleaner DB)." );
557	error_log( "Media Cleaner: Failed to Untrash Post {$issue->postId} (but deleted it from Cleaner DB)." );
558	return false;
559	}
560	$wpdb->query( $wpdb->prepare( "UPDATE $table_name SET deleted = 0 WHERE id = %d", $id ) );
561	$this->log( "�
562	Recovered Media #{$issue->postId}." );
563	return true;
564	}
565	}
566
567	function trash_file( $fileIssuePath ) {
568	$originalPath = trailingslashit( $this->upload_path ) . $fileIssuePath;
569	$trashPath = trailingslashit( $this->get_trashdir() ) . $fileIssuePath;
570	$path_parts = pathinfo( $trashPath );
571
572	try {
573	if ( !file_exists( $path_parts['dirname'] ) && !wp_mkdir_p( $path_parts['dirname'] ) ) {
574	$this->log( "🚫 Could not create the trash directory for Media Cleaner." );
575	error_log( "Media Cleaner: Could not create the trash directory." );
576	return false;
577	}
578	// Rename the file (move). 'is_dir' is just there for security (no way we should move a whole directory)
579	if ( is_dir( $originalPath ) ) {
580	$this->log( "🚫 Attempted to delete a directory instead of a file ($originalPath). Can't do that." );
581	error_log( "Media Cleaner: Attempted to delete a directory instead of a file ($originalPath). Can't do that." );
582	return false;
583	}
584	if ( !file_exists( $originalPath ) ) {
585	$this->log( "🚫 The file $originalPath actually does not exist." );
586	error_log( "Media Cleaner: The file $originalPath actually does not exist." );
587	return true;
588	}
589	if ( !@rename( $originalPath, $trashPath ) ) {
590	error_log( "Media Cleaner: Unknown error occured while trying to delete a file ($originalPath)." );
591	return false;
592	}
593	}
594	catch ( Exception $e ) {
595	return false;
596	}
597	$this->clean_dir( dirname( $originalPath ) );
598	return true;
599	}
600
601	function ignore( $id, $ignore ) {
602	global $wpdb;
603	$table_name = $wpdb->prefix . "mclean_scan";
604	$issue = $wpdb->get_row( $wpdb->prepare( "SELECT * FROM $table_name WHERE id = %d", $id ), OBJECT );
605	if ( !$ignore )
606	$wpdb->query( $wpdb->prepare( "UPDATE $table_name SET ignored = 0 WHERE id = %d", $id ) );
607	else {
608	if ( (int) $issue->deleted ) // If it is in trash, recover it
609	$this->recover( $id );
610	$wpdb->query( $wpdb->prepare( "UPDATE $table_name SET ignored = 1 WHERE id = %d", $id ) );
611	}
612	return true;
613	}
614
615	function endsWith( $haystack, $needle )
616	{
617	$length = strlen( $needle );
618	if ( $length == 0 )
619	return true;
620	return ( substr( $haystack, -$length ) === $needle );
621	}
622
623	function clean_dir( $dir ) {
624	if ( !file_exists( $dir ) )
625	return;
626	else if ( $this->endsWith( $dir, 'uploads' ) )
627	return;
628	$found = array_diff( scandir( $dir ), array( '.', '..' ) );
629	if ( count( $found ) < 1 ) {
630	if ( rmdir( $dir ) ) {
631	$this->clean_dir( dirname( $dir ) );
632	}
633	}
634	}
635
636	function delete( $id ) {
637	global $wpdb;
638	$table_name = $wpdb->prefix . "mclean_scan";
639	$issue = $wpdb->get_row( $wpdb->prepare( "SELECT * FROM $table_name WHERE id = %d", $id ), OBJECT );
640	$regex = "^(.)(\\s\\(\\+.)$";
641	$issue->path = preg_replace( '/' . $regex . '/i', '$1', $issue->path ); // remove " (+ 6 files)" from path
642
643	// Make sure there isn't a media DB entry
644	if ( $issue->type == 0 ) {
645	$attachmentid = $this->find_media_id_from_file( $issue->path, true );
646	if ( $attachmentid ) {
647	$this->log( "🚫 Issue listed as filesystem but Media {$attachmentid} exists." );
648	}
649	}
650
651	if ( $issue->type == 0 ) {
652
653	if ( $issue->deleted == 0 ) {
654	// Move file to the trash
655	if ( $this->trash_file( $issue->path ) )
656	$wpdb->query( $wpdb->prepare( "UPDATE $table_name SET deleted = 1, ignored = 0 WHERE id = %d", $id ) );
657	return true;
658	}
659	else {
660	// Delete file from the trash
661	$trashPath = trailingslashit( $this->get_trashdir() ) . $issue->path;
662	if ( !unlink( $trashPath ) ) {
663	$this->log( "🚫 Failed to delete the file." );
664	error_log( "Media Cleaner: Failed to delete the file." );
665	}
666	$wpdb->query( $wpdb->prepare( "DELETE FROM $table_name WHERE id = %d", $id ) );
667	$this->clean_dir( dirname( $trashPath ) );
668	return true;
669	}
670	}
671
672	if ( $issue->type == 1 ) {
673	if ( $issue->deleted == 0 && MEDIA_TRASH ) {
674	// Move Media to trash
675	// Let's copy the images to the trash so that it can be recovered.
676	$paths = $this->get_paths_from_attachment( $issue->postId );
677	foreach ( $paths as $path ) {
678	if ( !$this->trash_file( $path ) ) {
679	$this->log( "🚫 Could not trash $path." );
680	error_log( "Media Cleaner: Could not trash $path." );
681	}
682	}
683	wp_delete_attachment( $issue->postId, false );
684	$wpdb->query( $wpdb->prepare( "UPDATE $table_name SET deleted = 1, ignored = 0 WHERE id = %d", $id ) );
685	return true;
686	}
687	else {
688	// Trash Media definitely by recovering it (to be like a normal Media) and remove it through the
689	// standard WordPress workflow
690	if ( MEDIA_TRASH )
691	$this->recover( $id );
692	wp_delete_attachment( $issue->postId, true );
693	$wpdb->query( $wpdb->prepare( "DELETE FROM $table_name WHERE id = %d", $id ) );
694	return true;
695	}
696	}
697	return false;
698	}
699
700	/**
701	*
702	* SCANNING / RESET
703	*
704	*/
705
706	function add_reference_url( $urlOrUrls, $type, $origin = null, $extra = null ) {
707	$urlOrUrls = !is_array( $urlOrUrls ) ? array( $urlOrUrls ) : $urlOrUrls;
708	foreach ( $urlOrUrls as $url ) {
709	// With files, we need both filename without resolution and filename with resolution, it's important
710	// to make sure the original file is not deleted if a size exists for it.
711	// With media, all URLs should be without resolution to make sure it matches Media.
712	if ( $this->current_method == 'files' )
713	$this->add_reference( null, $url, $type, $origin );
714	$this->add_reference( 0, $this->clean_url_from_resolution( $url ), $type, $origin );
715	}
716	}
717
718	function add_reference_id( $idOrIds, $type, $origin = null, $extra = null ) {
719	$idOrIds = !is_array( $idOrIds ) ? array( $idOrIds ) : $idOrIds;
720	foreach ( $idOrIds as $id ) {
721	$this->add_reference( $id, "", $type, $origin );
722	if ( $this->multilingual ) {
723	$translatedIds = $this->get_translated_media_ids( (int)$id );
724
725	// Test for WPML
726	// if ( $id === '350') {
727	// $translatedIds = $this->get_translated_media_ids( (int)$id );
728	// $count = count($translatedIds);
729	// error_log( "${id} => ${count}" );
730	// }
731
732	if ( !empty( $translatedIds ) ) {
733	foreach ( $translatedIds as $translatedId ) {
734	$this->add_reference( $translatedId, "", $type, $origin );
735	}
736	}
737	}
738	}
739	}
740
741	private $cached_ids = array();
742	private $cached_urls = array();
743
744	// The references are actually not being added directly in the DB, they are being pushed
745	// into a cache ($this->refcache).
746	private function add_reference( $id, $url, $type, $origin = null, $extra = null ) {
747	if ( !empty( $id ) ) {
748	if ( !in_array( $id, $this->cached_ids ) ) {
749	array_push( $this->cached_ids, $id );
750	array_push( $this->refcache, array( 'id' => $id, 'url' => null, 'type' => $type, 'origin' => $origin ) );
751	}
752	}
753	if ( !empty( $url ) ) {
754	// The URL shouldn't contain http, https, javascript at the beginning (and there are probably many more cases)
755	// The URL must be cleaned before being passed as a reference.
756	if ( substr( $url, 0, 5 ) === "http:" \|\| substr( $url, 0, 6 ) === "https:" \|\| substr( $url, 0, 11 ) === "javascript:" ) {
757	return;
758	}
759	if ( !in_array( $url, $this->cached_urls ) ) {
760	array_push( $this->cached_urls, $url );
761	array_push( $this->refcache, array( 'id' => null, 'url' => $url, 'type' => $type, 'origin' => $origin ) );
762	}
763	}
764	}
765
766	// The cache containing the references is wrote to the DB.
767	function write_references() {
768	global $wpdb;
769	$table = $wpdb->prefix . "mclean_refs";
770	$values = array();
771	$place_holders = array();
772	$query = "INSERT INTO $table (mediaId, mediaUrl, originType) VALUES ";
773	foreach ( $this->refcache as $value ) {
774	if ( !is_null( $value['id'] ) ) {
775	array_push( $values, $value['id'], $value['type'] );
776	$place_holders[] = "('%d',NULL,'%s')";
777	if ( $this->debug_logs ) {
778	$this->log( "＋ Media #{$value['id']} (as ID)" );
779	}
780	}
781	else if ( !is_null( $value['url'] ) ) {
782	array_push( $values, $value['url'], $value['type'] );
783	$place_holders[] = "(NULL,'%s','%s')";
784	if ( $this->debug_logs ) {
785	$this->log( "＋ {$value['url']}" );
786	}
787	}
788	}
789	if ( !empty( $values ) ) {
790	$query .= implode( ', ', $place_holders );
791	$prepared = $wpdb->prepare( "$query ", $values );
792	$wpdb->query( $prepared );
793	}
794	$this->refcache = array();
795	}
796
797	function check_is_ignore( $file ) {
798	global $wpdb;
799	$table_name = $wpdb->prefix . "mclean_scan";
800	$count = $wpdb->get_var( "SELECT COUNT(*)
801	FROM $table_name
802	WHERE ignored = 1
803	AND path LIKE '%". esc_sql( $wpdb->esc_like( $file ) ) . "%'" );
804	if ( $count > 0 ) {
805	$this->log( "🚫 Could not trash $file." );
806	}
807	return ($count > 0);
808	}
809
810	function find_media_id_from_file( $file, $doLog ) {
811	global $wpdb;
812	$postmeta_table_name = $wpdb->prefix . 'postmeta';
813	$file = $this->clean_uploaded_filename( $file );
814	$sql = $wpdb->prepare( "SELECT post_id
815	FROM {$postmeta_table_name}
816	WHERE meta_key = '_wp_attached_file'
817	AND meta_value = %s", $file
818	);
819	$ret = $wpdb->get_var( $sql );
820	if ( $doLog ) {
821	if ( empty( $ret ) )
822	$this->log( "🚫 File $file not found as _wp_attached_file (Library)." );
823	else {
824	$this->log( "�
825	File $file found as Media $ret." );
826	}
827	}
828
829	return $ret;
830	}
831
832	function get_image_sizes() {
833	$sizes = array();
834	global $_wp_additional_image_sizes;
835	foreach ( get_intermediate_image_sizes() as $s ) {
836	$crop = false;
837	if ( isset( $_wp_additional_image_sizes[$s] ) ) {
838	$width = intval( $_wp_additional_image_sizes[$s]['width'] );
839	$height = intval( $_wp_additional_image_sizes[$s]['height'] );
840	$crop = $_wp_additional_image_sizes[$s]['crop'];
841	} else {
842	$width = get_option( $s.'_size_w' );
843	$height = get_option( $s.'_size_h' );
844	$crop = get_option( $s.'_crop' );
845	}
846	$sizes[$s] = array( 'width' => $width, 'height' => $height, 'crop' => $crop );
847	}
848	return $sizes;
849	}
850
851	function clean_url_from_resolution( $url ) {
852	$pattern = '/[_-]\d+x\d+(?=\.[a-z]{3,4}$)/';
853	$url = preg_replace( $pattern, '', $url );
854	return $url;
855	}
856
857	function is_url( $url ) {
858	return ( (
859	!empty( $url ) ) &&
860	is_string( $url ) &&
861	strlen( $url ) > 4 && (
862	strtolower( substr( $url, 0, 4) ) == 'http' \|\| $url[0] == '/'
863	)
864	);
865	}
866
867	function clean_url_from_resolution_ref( &$url ) {
868	$url = $this->clean_url_from_resolution( $url );
869	}
870
871	// From a url to the shortened and cleaned url (for example '2013/02/file.png')
872	function clean_url( $url ) {
873	// if ( is_array( $url ) ) {
874	// error_log( print_r( $url, 1 ) );
875	// }
876	$dirIndex = strpos( $url, $this->upload_url );
877	if ( empty( $url ) \|\| $dirIndex === false ) {
878	$finalUrl = null;
879	}
880	else {
881	$finalUrl = urldecode( substr( $url, 1 + strlen( $this->upload_url ) + $dirIndex ) );
882	}
883	return $finalUrl;
884	}
885
886	// From a fullpath to the shortened and cleaned path (for example '2013/02/file.png')
887	// Original version by Jordy
888	// function clean_uploaded_filename( $fullpath ) {
889	// $basedir = $this->upload_path;
890	// $file = str_replace( $basedir, '', $fullpath );
891	// $file = str_replace( "./", "", $file );
892	// $file = trim( $file, "/" );
893	// return $file;
894	// }
895
896	// From a fullpath to the shortened and cleaned path (for example '2013/02/file.png')
897	// Faster version, more difficult to read, by Mike Meinz
898	function clean_uploaded_filename( $fullpath ) {
899	$dirIndex = strpos( $fullpath, $this->upload_url );
900	if ( $dirIndex == false ) {
901	$file = $fullpath;
902	}
903	else {
904	// Remove first part of the path leaving yyyy/mm/filename.ext
905	$file = substr( $fullpath, 1 + strlen( $this->upload_url ) + $dirIndex );
906	}
907	if ( substr( $file, 0, 2 ) == './' ) {
908	$file = substr( $file, 2 );
909	}
910	if ( substr( $file, 0, 1 ) == '/' ) {
911	$file = substr( $file, 1 );
912	}
913	return $file;
914	}
915
916	/*
917	Check if the file or the Media ID is used in the install.
918	That file or ID will be checked against the database of references created by the plugin
919	by the parsers.
920	*/
921	public function reference_exists( $file, $mediaId ) {
922	global $wpdb;
923	$table = $wpdb->prefix . "mclean_refs";
924	$row = null;
925	if ( !empty( $mediaId ) ) {
926	$row = $wpdb->get_row( $wpdb->prepare( "SELECT originType FROM $table WHERE mediaId = %d", $mediaId ) );
927	if ( !empty( $row ) ) {
928	$origin = $row->originType === 'MEDIA LIBRARY' ? 'Media Library' : 'content';
929	$this->log( "�
930	Media #{$mediaId} used by {$origin}" );
931	return $row->originType;
932	}
933	}
934	if ( !empty( $file ) ) {
935	$row = $wpdb->get_row( $wpdb->prepare( "SELECT originType FROM $table WHERE mediaUrl = %s", $file ) );
936	if ( !empty( $row ) ) {
937	$origin = $row->originType === 'MEDIA LIBRARY' ? 'Media Library' : 'content';
938	$this->log( "�
939	File {$file} used by {$origin}" );
940	return $row->originType;
941	}
942	}
943	return false;
944	}
945
946	function get_paths_from_attachment( $attachmentId ) {
947	$paths = array();
948	$fullpath = get_attached_file( $attachmentId );
949	if ( empty( $fullpath ) ) {
950	error_log( 'Media Cleaner: Could not find attached file for Media ID ' . $attachmentId );
951	return array();
952	}
953	$mainfile = $this->clean_uploaded_filename( $fullpath );
954	array_push( $paths, $mainfile );
955	$baseUp = pathinfo( $mainfile );
956	$filespath = trailingslashit( $this->upload_path ) . trailingslashit( $baseUp['dirname'] );
957	$meta = wp_get_attachment_metadata( $attachmentId );
958	if ( isset( $meta['original_image'] ) ) {
959	$original_image = $this->clean_uploaded_filename( $filespath . $meta['original_image'] );
960	array_push( $paths, $original_image );
961	}
962	$isImage = isset( $meta, $meta['width'], $meta['height'] );
963	$sizes = $this->get_image_sizes();
964	if ( $isImage && isset( $meta['sizes'] ) ) {
965	foreach ( $meta['sizes'] as $name => $attr ) {
966	if ( isset( $attr['file'] ) ) {
967	$file = $this->clean_uploaded_filename( $filespath . $attr['file'] );
968	array_push( $paths, $file );
969	}
970	}
971	}
972	return $paths;
973	}
974
975	function is_media_ignored( $attachmentId ) {
976	global $wpdb;
977	$table_name = $wpdb->prefix . "mclean_scan";
978	$issue = $wpdb->get_row( $wpdb->prepare( "SELECT * FROM $table_name WHERE postId = %d", $attachmentId ), OBJECT );
979	//error_log( $attachmentId );
980	//error_log( print_r( $issue, 1 ) );
981	if ( $issue && $issue->ignored )
982	return true;
983	return false;
984	}
985
986	function check_media( $attachmentId, $checkOnly = false ) {
987
988	// Is Media ID ignored, consider as used.
989	if ( $this->is_media_ignored( $attachmentId ) ) {
990	return true;
991	}
992
993	$size = 0;
994	$countfiles = 0;
995	$check_broken_media = !$this->check_content;
996	$fullpath = get_attached_file( $attachmentId );
997	$is_broken = !file_exists( $fullpath );
998
999	// It's a broken-only scan
1000	if ( $check_broken_media && !$is_broken ) {
1001	$is_considered_used = apply_filters( 'wpmc_check_media', true, $attachmentId, false );
1002	return $is_considered_used;
1003	}
1004
1005	// Let's analyze the usage of each path (thumbnails included) for this Media ID.
1006	$issue = 'NO_CONTENT';
1007	$paths = $this->get_paths_from_attachment( $attachmentId );
1008	foreach ( $paths as $path ) {
1009
1010	// If it's found in the content, we stop the scan right away
1011	if ( $this->check_content && $this->reference_exists( $path, $attachmentId ) ) {
1012	$is_considered_used = apply_filters( 'wpmc_check_media', true, $attachmentId, false );
1013	if ( $is_considered_used ) {
1014	return true;
1015	}
1016	}
1017
1018	// Let's count the size of the files for later, in case it's unused
1019	$filepath = trailingslashit( $this->upload_path ) . $path;
1020	if ( file_exists( $filepath ) )
1021	$size += filesize( $filepath );
1022	$countfiles++;
1023	}
1024
1025	// This Media ID seems not in used (or broken)
1026	// Let's double-check through the filter (overridable by users)
1027	$is_considered_used = apply_filters( 'wpmc_check_media', false, $attachmentId, $is_broken );
1028	if ( !$is_considered_used ) {
1029	if ( $is_broken ) {
1030	$this->log( "🚫 File {$fullpath} does not exist." );
1031	$issue = 'ORPHAN_MEDIA';
1032	}
1033	if ( !$checkOnly ) {
1034	global $wpdb;
1035	$table_name = $wpdb->prefix . "mclean_scan";
1036	$mainfile = $this->clean_uploaded_filename( $fullpath );
1037	$wpdb->insert( $table_name,
1038	array(
1039	'time' => current_time('mysql'),
1040	'type' => 1,
1041	'size' => $size,
1042	'path' => $mainfile . ( $countfiles > 0 ? ( " (+ " . $countfiles . " files)" ) : "" ),
1043	'postId' => $attachmentId,
1044	'issue' => $issue
1045	)
1046	);
1047	}
1048	}
1049	return $is_considered_used;
1050	}
1051
1052	// Delete all issues
1053	function reset_issues( $includingIgnored = false ) {
1054	global $wpdb;
1055	$table_name = $wpdb->prefix . "mclean_scan";
1056	if ( $includingIgnored ) {
1057	$wpdb->query( "DELETE FROM $table_name WHERE deleted = 0" );
1058	}
1059	else {
1060	$wpdb->query( "DELETE FROM $table_name WHERE ignored = 0 AND deleted = 0" );
1061	}
1062	if ( file_exists( WPMC_PATH . '/logs/media-cleaner.log' ) ) {
1063	file_put_contents( WPMC_PATH . '/logs/media-cleaner.log', '' );
1064	}
1065	$table_name = $wpdb->prefix . "mclean_refs";
1066	$wpdb->query("TRUNCATE $table_name");
1067	}
1068
1069	function echo_issue( $issue ) {
1070	if ( $issue == 'NO_CONTENT' ) {
1071	_e( "Seems not use", 'media-cleaner' );
1072	}
1073	else if ( $issue == 'ORPHAN_FILE' ) {
1074	_e( "Not in Library", 'media-cleaner' );
1075	}
1076	else if ( $issue == 'ORPHAN_RETINA' ) {
1077	_e( "Orphan Retina", 'media-cleaner' );
1078	}
1079	else if ( $issue == 'ORPHAN_WEBP' ) {
1080	_e( "Orphan WebP", 'media-cleaner' );
1081	}
1082	else if ( $issue == 'ORPHAN_MEDIA' ) {
1083	_e( "No attached file", 'media-cleaner' );
1084	}
1085	else {
1086	echo $issue;
1087	}
1088	}
1089	}
1090
1091	/*
1092	INSTALL / UNINSTALL
1093	*/
1094
1095	function wpmc_init( $mainfile ) {
1096	//register_activation_hook( $mainfile, 'wpmc_install' );
1097	//register_deactivation_hook( $mainfile, 'wpmc_uninstall' );
1098	register_uninstall_hook( $mainfile, 'wpmc_uninstall' );
1099	}
1100
1101	function wpmc_install() {
1102	wpmc_create_database();
1103	}
1104
1105	function wpmc_reset () {
1106	wpmc_remove_database();
1107	wpmc_create_database();
1108	}
1109
1110	function wpmc_uninstall () {
1111	//wpmc_remove_options();
1112	//wpmc_remove_database();
1113	}
1114
1115	// Check the DB. If does not exist, let's create it.
1116	// TODO: When PHP 7 only, let's clean this and use anonymous functions.
1117	function wpmc_check_database() {
1118	global $wpdb;
1119	static $wpmc_check_database_done = false;
1120	if ( $wpmc_check_database_done ) {
1121	return true;
1122	}
1123	$table_scan = $wpdb->prefix . "mclean_refs";
1124	$table_refs = $wpdb->prefix . "mclean_scan";
1125	$db_init = !( strtolower( $wpdb->get_var( "SHOW TABLES LIKE '$table_scan'" ) ) != strtolower( $table_scan )
1126	\|\| strtolower( $wpdb->get_var( "SHOW TABLES LIKE '$table_refs'" ) ) != strtolower( $table_refs ) );
1127	if ( !$db_init ) {
1128	wpmc_create_database();
1129	$db_init = !( strtolower( $wpdb->get_var( "SHOW TABLES LIKE '$table_scan'" ) ) != strtolower( $table_scan )
1130	\|\| strtolower( $wpdb->get_var( "SHOW TABLES LIKE '$table_refs'" ) ) != strtolower( $table_refs ) );
1131	}
1132	$wpmc_check_database_done = true;
1133	}
1134
1135	function wpmc_create_database() {
1136	global $wpdb;
1137	$table_name = $wpdb->prefix . "mclean_scan";
1138	$charset_collate = $wpdb->get_charset_collate();
1139	$sql = "CREATE TABLE $table_name (
1140	id BIGINT(20) NOT NULL AUTO_INCREMENT,
1141	time DATETIME DEFAULT '0000-00-00 00:00:00' NOT NULL,
1142	type TINYINT(1) NOT NULL,
1143	postId BIGINT(20) NULL,
1144	path TINYTEXT NULL,
1145	size INT(9) NULL,
1146	ignored TINYINT(1) NOT NULL DEFAULT 0,
1147	deleted TINYINT(1) NOT NULL DEFAULT 0,
1148	issue TINYTEXT NOT NULL,
1149	PRIMARY KEY (id)
1150	) " . $charset_collate . ";" ;
1151	require_once( ABSPATH . 'wp-admin/includes/upgrade.php' );
1152	dbDelta( $sql );
1153	$sql="ALTER TABLE $table_name ADD INDEX IgnoredIndex (ignored) USING BTREE;";
1154	$wpdb->query($sql);
1155	$table_name = $wpdb->prefix . "mclean_refs";
1156	$charset_collate = $wpdb->get_charset_collate();
1157	// This key doesn't work on too many installs because of the 'Specified key was too long' issue
1158	// KEY mediaLookUp (mediaId, mediaUrl)
1159	$sql = "CREATE TABLE $table_name (
1160	id BIGINT(20) NOT NULL AUTO_INCREMENT,
1161	mediaId BIGINT(20) NULL,
1162	mediaUrl TINYTEXT NULL,
1163	originType TINYTEXT NOT NULL,
1164	PRIMARY KEY (id)
1165	) " . $charset_collate . ";";
1166	require_once( ABSPATH . 'wp-admin/includes/upgrade.php' );
1167	dbDelta( $sql );
1168	}
1169
1170	function wpmc_remove_options() {
1171	global $wpdb;
1172	$options = $wpdb->get_results( "SELECT option_name FROM $wpdb->options WHERE option_name LIKE 'wpmc_%'" );
1173	foreach( $options as $option ) {
1174	delete_option( $option->option_name );
1175	}
1176	}
1177
1178	function wpmc_remove_database() {
1179	global $wpdb;
1180	$table_name1 = $wpdb->prefix . "mclean_scan";
1181	$table_name2 = $wpdb->prefix . "mclean_refs";
1182	$table_name3 = $wpdb->prefix . "wpmcleaner";
1183	$sql = "DROP TABLE IF EXISTS $table_name1, $table_name2, $table_name3;";
1184	$wpdb->query( $sql );
1185	}
1186