templately / modules / full-site-import / Cleanup / TmpDirsTask.php

TmpDirsTask.php in Templately – Elementor & Gutenberg Template Library: 6500+ Free & Pro Ready Templates And Cloud! 3.8.0, at modules/full-site-import/Cleanup/TmpDirsTask.php

162 lines 5.1 KB

Raw Download Zip

1	<?php
2	/**
3	* Reclaim abandoned import working directories.
4	*
5	* @package Templately
6	*/
7
8	namespace Templately\Modules\FullSiteImport\Cleanup;
9
10	use Templately\Modules\FullSiteImport\Utils\SessionData;
11	use Templately\Modules\Utilities\Cleanup\CleanupTask;
12	use Templately\Modules\Utilities\Cleanup\Context;
13	use Templately\Modules\Utilities\Cleanup\Remover;
14	use Templately\Modules\Utilities\Cleanup\RetentionKind;
15	use Templately\Modules\Utilities\Cleanup\Scanner;
16	use Templately\Modules\Utilities\Cleanup\TaskResult;
17	use WP_Error;
18
19	/**
20	* THE DISK LEAK. This is the headline defect spec 052 exists to close.
21	*
22	* `uploads/templately/tmp/{session_id}/` holds an extracted template pack —
23	* routinely 100–500 MB of XML, JSON and media. Before this task, nothing ever
24	* removed one unless an import FINISHED, and even then only on a 1-in-20
25	* random chance. An import that failed or was abandoned left its pack on disk
26	* permanently, and the scheduled cleanup made it worse by deleting the session
27	* record, so nothing referenced the directory any more.
28	*
29	* A directory is reclaimable when it is EITHER orphaned (no session record at
30	* all — the common case after the old record-only cleanup ran) OR its session
31	* has passed the retention window. An active import is never touched, and that
32	* is re-checked immediately before each removal rather than when the candidate
33	* list was built.
34	*/
35	class TmpDirsTask implements CleanupTask {
36
37	const SUBDIR = 'tmp';
38
39	public function descriptor(): array {
40	return [
41	'id' => 'fsi-tmp-dirs',
42	'label' => __( 'Abandoned import files', 'templately' ),
43	'group' => 'imports',
44	'scope' => 'files',
45	'retention_kind' => RetentionKind::AGE,
46	'destructive' => true,
47	'schedulable' => true,
48	];
49	}
50
51	public function estimate( Context $context ): TaskResult {
52	return $this->sweep( $context->with_dry_run( true ) );
53	}
54
55	public function run( Context $context ): TaskResult {
56	return $this->sweep( $context );
57	}
58
59	/**
60	* One pass over the working directories.
61	*/
62	protected function sweep( Context $context ): TaskResult {
63	$result = TaskResult::empty();
64	$root = $this->root();
65
66	if ( ! is_dir( $root ) ) {
67	// Nothing has ever been imported on this site. Not an error.
68	return $result;
69	}
70
71	$entries = @scandir( $root ); // phpcs:ignore WordPress.PHP.NoSilencedErrors.Discouraged
72	if ( false === $entries ) {
73	return $result->fail( 'Could not read the import working directory.' );
74	}
75
76	$threshold = $context->policy()->age_threshold( $this->descriptor()['id'] );
77
78	// Collected ONCE. This only answers "does a record exist for this
79	// directory"; whether that record is still ACTIVE is re-read per item
80	// immediately before removal, so freshness is not traded away for it.
81	// Re-querying every session row per directory turned a sweep of a few
82	// hundred leftovers into a few hundred full-table reads.
83	$known = $this->known_session_ids();
84
85	foreach ( $entries as $entry ) {
86	if ( '.' === $entry \|\| '..' === $entry ) {
87	continue;
88	}
89
90	$path = $root . DIRECTORY_SEPARATOR . $entry;
91	if ( ! is_dir( $path ) \|\| is_link( $path ) ) {
92	continue;
93	}
94
95	// Budget is checked per ITEM, not per task: one of these directories
96	// can hold thousands of files, so a task that only checked once at
97	// the start would overshoot and be killed mid-delete.
98	if ( ! $context->has_budget( 2.0 ) ) {
99	return $result->stopped_on_budget();
100	}
101
102	if ( ! $this->is_reclaimable( $entry, $threshold, $known ) ) {
103	continue;
104	}
105
106	$outcome = Remover::delete_directory( $path, $context );
107
108	if ( $outcome instanceof WP_Error ) {
109	$result->fail( sprintf( '%s: %s', $entry, $outcome->get_error_message() ) );
110	continue;
111	}
112
113	$result->merge( $outcome );
114	}
115
116	return $result;
117	}
118
119	/**
120	* Whether this directory may be removed.
121	*
122	* Re-reads the session state at call time — an import that started while the
123	* sweep was walking the directory must be exempt from that moment on.
124	*/
125	protected function is_reclaimable( string $entry, int $threshold, array $known = [] ): bool {
126	if ( ! isset( $known[ $entry ] ) ) {
127	// Orphaned: no record owns this directory. Reclaim it — this is the
128	// state the old record-only cleanup left behind on every site.
129	return true;
130	}
131
132	// Re-read at call time, NOT from the collected snapshot: an import that
133	// started while this sweep was walking the directory must be exempt from
134	// that moment on.
135	return ! SessionData::is_active( $entry, $threshold );
136	}
137
138	/**
139	* Session ids that currently have a record, as a lookup map.
140	*
141	* @return array<string, true>
142	*/
143	protected function known_session_ids(): array {
144	$known = [];
145
146	foreach ( array_keys( SessionData::get_all_data() ) as $session_id ) {
147	$known[ (string) $session_id ] = true;
148	}
149
150	return $known;
151	}
152
153	/**
154	* The working-directory root, resolved through the engine so the base-dir
155	* test seam redirects it. Reading `wp_upload_dir()` directly here would make
156	* every test in this suite run against the developer's real uploads folder.
157	*/
158	protected function root(): string {
159	return Scanner::get_base_dir() . DIRECTORY_SEPARATOR . static::SUBDIR;
160	}
161	}
162