ablocks / includes / classes / images / unused-scanner.php

unused-scanner.php in aBlocks – Gutenberg Blocks, User Dashboard Builder, Popup Builder, Form Builder & Animation Builder 2.11.1, at includes/classes/images/unused-scanner.php

236 lines 6.6 KB

Raw Download Zip

1	<?php
2	namespace ABlocks\Classes\Images;
3
4	if ( ! defined( 'ABSPATH' ) ) {
5	exit;
6	}
7
8	/**
9	* Find attachments that appear to be unreferenced.
10	*
11	* ## Read this before trusting the result
12	*
13	* "Unused" cannot be determined reliably in WordPress. An image can be
14	* referenced from post content, post meta, options, theme mods, widgets, menus,
15	* customizer settings, page-builder JSON, a CSS file, another plugin's custom
16	* table, or an external site hot-linking it. A scanner sees some of those and
17	* cannot see the rest.
18	*
19	* So this deliberately answers a narrower question — "did we find any evidence
20	* this is used?" — and treats absence of evidence as a suspicion, never as
21	* proof. Everything it produces is a candidate for human review, which is why
22	* removal quarantines rather than deletes. See {@see Quarantine}.
23	*
24	* Guards that keep false positives survivable:
25	*
26	* - Attachments newer than a grace period are never listed. A freshly uploaded
27	* image is routinely unreferenced for the minutes between uploading it and
28	* using it, and that window is exactly when someone runs a cleanup.
29	* - Featured images, site icon, custom logo and header are checked explicitly.
30	* - Both the full URL and the bare filename are searched, so a resized variant
31	* or a relative reference still counts as a use.
32	*/
33	class UnusedScanner {
34
35	/**
36	* Attachments younger than this are never considered unused.
37	*/
38	const GRACE_DAYS = 30;
39
40	/**
41	* Scan a batch of attachments.
42	*
43	* @param int $limit Attachments to examine.
44	* @param int $offset Where to resume from.
45	* @return array{items:array, scanned:int, done:bool}
46	*/
47	public static function scan( $limit = 50, $offset = 0 ) {
48	$limit = max( 1, min( 200, (int) $limit ) );
49	$offset = max( 0, (int) $offset );
50
51	$query = new \WP_Query(
52	[
53	'post_type' => 'attachment',
54	'post_status' => 'inherit',
55	'post_mime_type' => 'image',
56	'posts_per_page' => $limit,
57	'offset' => $offset,
58	'orderby' => 'ID',
59	'order' => 'ASC',
60	'fields' => 'ids',
61	'no_found_rows' => false,
62	]
63	);
64
65	$grace = time() - ( (int) apply_filters( 'ablocks/images/unused_grace_days', self::GRACE_DAYS ) * DAY_IN_SECONDS );
66	$items = [];
67
68	foreach ( $query->posts as $id ) {
69	$id = (int) $id;
70
71	if ( get_post_time( 'U', true, $id ) > $grace ) {
72	continue;
73	}
74
75	$evidence = self::find_usage( $id );
76	if ( ! empty( $evidence ) ) {
77	continue;
78	}
79
80	$items[] = [
81	'id' => $id,
82	'title' => get_the_title( $id ),
83	'url' => wp_get_attachment_url( $id ),
84	'thumb' => wp_get_attachment_image_url( $id, 'thumbnail' ),
85	'bytes' => self::size_of( $id ),
86	'date' => get_the_date( 'Y-m-d', $id ),
87	];
88	}//end foreach
89
90	return [
91	'items' => $items,
92	'scanned' => $offset + count( $query->posts ),
93	'total' => (int) $query->found_posts,
94	'done' => count( $query->posts ) < $limit,
95	];
96	}
97
98	/**
99	* Look for any evidence that an attachment is in use.
100	*
101	* Returns as soon as it finds something: the caller only needs to know
102	* whether evidence exists, and stopping early keeps the scan cheap.
103	*
104	* @param int $attachment_id Attachment ID.
105	* @return string[] Reasons it appears used; empty when none were found.
106	*/
107	public static function find_usage( $attachment_id ) {
108	global $wpdb;
109
110	$attachment_id = (int) $attachment_id;
111	$found = [];
112
113	// Attached to a post. Cheap, and the most common real use.
114	$parent = (int) get_post_field( 'post_parent', $attachment_id );
115	if ( $parent && 'attachment' !== get_post_type( $parent ) ) {
116	$found[] = 'attached';
117	return $found;
118	}
119
120	// Featured image.
121	$thumb_for = $wpdb->get_var(
122	$wpdb->prepare(
123	"SELECT post_id FROM {$wpdb->postmeta} WHERE meta_key = '_thumbnail_id' AND meta_value = %d LIMIT 1",
124	$attachment_id
125	)
126	);
127	if ( $thumb_for ) {
128	$found[] = 'featured-image';
129	return $found;
130	}
131
132	// Site identity: logo, icon, header, background.
133	foreach ( [ 'site_icon', 'site_logo' ] as $option ) {
134	if ( (int) get_option( $option ) === $attachment_id ) {
135	$found[] = $option;
136	return $found;
137	}
138	}
139	foreach ( [ 'custom_logo', 'header_image_data' ] as $mod ) {
140	$value = get_theme_mod( $mod );
141	if ( is_object( $value ) && isset( $value->attachment_id ) && (int) $value->attachment_id === $attachment_id ) {
142	$found[] = 'theme-' . $mod;
143	return $found;
144	}
145	if ( is_numeric( $value ) && (int) $value === $attachment_id ) {
146	$found[] = 'theme-' . $mod;
147	return $found;
148	}
149	}
150
151	$url = wp_get_attachment_url( $attachment_id );
152	$file = get_post_meta( $attachment_id, '_wp_attached_file', true );
153
154	$needles = array_values(
155	array_unique(
156	array_filter(
157	[
158	$url,
159	$file ? basename( $file ) : '',
160	'wp-image-' . $attachment_id,
161	]
162	)
163	)
164	);
165
166	foreach ( $needles as $needle ) {
167	$like = '%' . $wpdb->esc_like( $needle ) . '%';
168
169	// Post content and excerpts, including page-builder markup, which is
170	// stored as post content or as JSON in meta.
171	$in_posts = $wpdb->get_var(
172	$wpdb->prepare(
173	"SELECT ID FROM {$wpdb->posts}
174	WHERE post_status NOT IN ('trash','auto-draft')
175	AND ( post_content LIKE %s OR post_excerpt LIKE %s )
176	LIMIT 1",
177	$like,
178	$like
179	)
180	);
181	if ( $in_posts ) {
182	$found[] = 'in-content:' . (int) $in_posts;
183	return $found;
184	}
185
186	// Meta covers ACF fields, builder payloads and most plugin storage.
187	$in_meta = $wpdb->get_var(
188	$wpdb->prepare(
189	"SELECT post_id FROM {$wpdb->postmeta}
190	WHERE meta_value LIKE %s AND post_id <> %d
191	LIMIT 1",
192	$like,
193	$attachment_id
194	)
195	);
196	if ( $in_meta ) {
197	$found[] = 'in-meta:' . (int) $in_meta;
198	return $found;
199	}
200
201	// Options catch widgets, theme mods, and plugin settings.
202	$in_options = $wpdb->get_var(
203	$wpdb->prepare(
204	"SELECT option_name FROM {$wpdb->options} WHERE option_value LIKE %s LIMIT 1",
205	$like
206	)
207	);
208	if ( $in_options ) {
209	$found[] = 'in-option:' . sanitize_key( $in_options );
210	return $found;
211	}
212	}//end foreach
213
214	// A last chance for anything with its own storage: a plugin that keeps
215	// media references in a custom table can answer here rather than having
216	// its images quarantined.
217	return (array) apply_filters( 'ablocks/images/usage_evidence', $found, $attachment_id );
218	}
219
220	/**
221	* Total bytes an attachment occupies, including generated sizes.
222	*
223	* @param int $attachment_id Attachment ID.
224	* @return int
225	*/
226	public static function size_of( $attachment_id ) {
227	$total = 0;
228	foreach ( Compressor::files_for( $attachment_id ) as $path ) {
229	if ( file_exists( $path ) ) {
230	$total += (int) filesize( $path );
231	}
232	}
233	return $total;
234	}
235	}
236