vigilante / includes / class-htaccess-recovery.php

class-htaccess-recovery.php in Vigilant – 100% Free Security Suite: Firewall, 2FA, Login, Headers, Scanner… 2.10.4, at includes/class-htaccess-recovery.php

686 lines 23.1 KB

Raw Download Zip

1	<?php
2	/**
3	* Header settings recovery
4	*
5	* Offers back the Security Headers configuration that the 2.9.8 migration
6	* wiped, reading it from the .htaccess snapshot Vigilante_Htaccess_Manager
7	* preserved before the first rewrite that would have overwritten it.
8	*
9	* Deliberately recovers SETTINGS, never the file. Restoring a stored .htaccess
10	* wholesale would also put back the WordPress block and every rule the host,
11	* the cache plugin or the CDN had added since, undoing work that has nothing to
12	* do with Vigilant, and on a site that has changed permalinks or moved to a
13	* different folder it would break routing outright. Reading our own block and
14	* writing the settings back leaves the file to be regenerated the normal way,
15	* through the validated write path, and fixes the settings screen too: a file
16	* restore would have left the screen showing factory values, so the next save
17	* would have wiped everything again.
18	*
19	* @package Vigilante
20	* @since 2.10.0
21	*/
22
23	// Prevent direct access
24	if ( ! defined( 'ABSPATH' ) ) {
25	exit;
26	}
27
28	/**
29	* Class Vigilante_Htaccess_Recovery
30	*/
31	class Vigilante_Htaccess_Recovery {
32
33	/**
34	* Set once the owner has restored or dismissed the offer.
35	*/
36	const DISMISSED_OPTION = 'vigilante_htaccess_recovery_done';
37
38	/**
39	* The security_headers section as it was just before restoring, so the
40	* restore itself can be undone.
41	*/
42	const UNDO_OPTION = 'vigilante_htaccess_recovery_undo';
43
44	/**
45	* Markers of the block that is read. Nothing outside them is ever parsed.
46	*/
47	/**
48	* The one-off snapshot of the .htaccess taken before the first rewrite that
49	* would have overwritten the owner's real header configuration.
50	*/
51	const SNAPSHOT_OPTION = 'vigilante_htaccess_pre_migration';
52
53	const BLOCK_START = '# BEGIN Vigilante Security Headers';
54	const BLOCK_END = '# END Vigilante Security Headers';
55
56	/**
57	* Values each header is allowed to carry back into the settings.
58	*
59	* The snapshot is a file on disk that anything could have edited, so it is
60	* treated as untrusted input: a value that is not on these lists is dropped
61	* rather than written into the options. Same lists the generator enforces.
62	*
63	* @return array
64	*/
65	private static function allowed() {
66	return array(
67	'x_frame_options' => array( 'SAMEORIGIN', 'DENY' ),
68	'referrer_policy' => array(
69	'no-referrer',
70	'no-referrer-when-downgrade',
71	'origin',
72	'origin-when-cross-origin',
73	'same-origin',
74	'strict-origin',
75	'strict-origin-when-cross-origin',
76	'unsafe-url',
77	),
78	'opener_policy' => array( 'unsafe-none', 'same-origin-allow-popups', 'same-origin' ),
79	'embedder_policy' => array( 'unsafe-none', 'require-corp', 'credentialless' ),
80	'resource_policy' => array( 'same-site', 'same-origin', 'cross-origin' ),
81	);
82	}
83
84	/**
85	* Whether there is a recovery to offer.
86	*
87	* @return bool
88	*/
89	public static function is_available() {
90	if ( get_option( self::DISMISSED_OPTION ) ) {
91	return false;
92	}
93
94	$recovered = self::get_recovered_settings();
95
96	return ! empty( $recovered );
97	}
98
99	/**
100	* Whether a restore can still be taken back.
101	*
102	* @since 2.10.0
103	* @return bool
104	*/
105	public static function has_undo() {
106	$previous = get_option( self::UNDO_OPTION );
107
108	return is_array( $previous ) && ! empty( $previous );
109	}
110
111	/**
112	* Preserve the .htaccess once, when the file and the settings disagree.
113	*
114	* Detecting the damage by the shape of the stored section does not work, and
115	* finding that out cost a day: register_setting() hangs
116	* Vigilante_Settings::validate_options() on this option as a sanitize
117	* callback, and validate_section() fills in a default for every key the input
118	* is missing. So the broken 2.9.8 migration, which passed two keys, did not
119	* leave a two-key section behind: it left a complete one with factory values
120	* in the twelve keys it dropped, which is indistinguishable from a site that
121	* simply never customised its headers.
122	*
123	* What is distinguishable is the desync that Albert reported: the file still
124	* describes what the owner chose while the options already say factory. So
125	* that is what is looked at here. The file is the older truth, and this is
126	* the last moment it exists.
127	*
128	* Called only when the installed version has changed, never on an ordinary
129	* save: a save legitimately leaves the file describing the previous values
130	* for the instant before it is rewritten, and capturing there would burn the
131	* single slot on a difference the owner made on purpose.
132	*
133	* @since 2.10.0
134	* @param string $content The .htaccess about to be rewritten.
135	* @param Vigilante_Settings $settings Settings instance.
136	*/
137	public static function maybe_capture( $content, $settings ) {
138	if ( '' === (string) $content \|\| false !== get_option( self::SNAPSHOT_OPTION, false ) ) {
139	return;
140	}
141
142	if ( '' === self::extract_block( $content ) ) {
143	return;
144	}
145
146	/*
147	* Compared as emitted directives, never as settings arrays. The two do
148	* not have the same shape and never will: an unsafe-none COEP is not
149	* written at all, the CSP carries a report_uri the header does not, and a
150	* boolean directive comes back from the file as a flag. Comparing the
151	* arrays reported a difference on a perfectly healthy site.
152	*
153	* What is comparable is what the server is being told. So the block on
154	* disk is measured against the block this configuration would write right
155	* now: if they say different things, the file is describing a
156	* configuration that is no longer stored anywhere.
157	*/
158	require_once VIGILANTE_INCLUDES_DIR . 'class-security-headers.php';
159
160	// Solo nuestro bloque de cabeceras: el fichero entero trae también las
161	// directivas del bloque de Protección (unset X-Powered-By, unset Server),
162	// que el generador de cabeceras no produce y que harían diferir a
163	// cualquier sitio sano que tenga el cortafuegos activo.
164	$on_disk = self::directives( self::extract_block( $content ) );
165	$would = self::directives( ( new Vigilante_Security_Headers( $settings ) )->generate_rules_content() );
166
167	if ( empty( $on_disk ) \|\| $on_disk === $would ) {
168	return;
169	}
170
171	update_option(
172	self::SNAPSHOT_OPTION,
173	array(
174	'content' => (string) $content,
175	'time' => time(),
176	'version' => VIGILANTE_VERSION,
177	),
178	false
179	);
180	}
181
182	/**
183	* Every header directive a chunk of .htaccess sets, normalised for comparison.
184	*
185	* Comments, blank lines, indentation and order are all dropped: two blocks
186	* that tell the server the same things compare equal even if one was written
187	* by an older version with a different timestamp in its header comment.
188	*
189	* @since 2.10.0
190	* @param string $text Block or full file.
191	* @return array
192	*/
193	private static function directives( $text ) {
194	$lines = array();
195
196	if ( preg_match_all( '/^\sHeader\s+always\s+(set\|unset)\s+([A-Za-z0-9-]+)(?:\s+"([^"])")?/mi', (string) $text, $matches, PREG_SET_ORDER ) ) {
197	foreach ( $matches as $match ) {
198	$lines[] = strtolower( $match[1] ) . ' ' . strtolower( $match[2] ) . ' ' . ( isset( $match[3] ) ? trim( $match[3] ) : '' );
199	}
200	}
201
202	sort( $lines );
203
204	return $lines;
205	}
206
207	/**
208	* The stored snapshot, or an empty array.
209	*
210	* @return array
211	*/
212	public static function get_snapshot() {
213	$snapshot = get_option( self::SNAPSHOT_OPTION );
214
215	return ( is_array( $snapshot ) && ! empty( $snapshot['content'] ) ) ? $snapshot : array();
216	}
217
218	/**
219	* The raw Vigilant header block from the snapshot, for the owner to read.
220	*
221	* @return string
222	*/
223	public static function get_raw_block() {
224	$snapshot = self::get_snapshot();
225
226	if ( empty( $snapshot ) ) {
227	return '';
228	}
229
230	return self::extract_block( $snapshot['content'] );
231	}
232
233	/**
234	* Cut out our own block. Everything outside the markers is ignored.
235	*
236	* @param string $content Full .htaccess content.
237	* @return string
238	*/
239	private static function extract_block( $content ) {
240	$start = strpos( $content, self::BLOCK_START );
241
242	if ( false === $start ) {
243	return '';
244	}
245
246	$end = strpos( $content, self::BLOCK_END, $start );
247
248	if ( false === $end ) {
249	return '';
250	}
251
252	return substr( $content, $start, ( $end - $start ) + strlen( self::BLOCK_END ) );
253	}
254
255	/**
256	* Read every "Header always set" line of the block into name => value.
257	*
258	* @param string $block Block content.
259	* @return array
260	*/
261	private static function read_headers( $block ) {
262	$headers = array();
263
264	if ( ! preg_match_all( '/^\sHeader\s+always\s+set\s+([A-Za-z0-9-]+)\s+"([^"])"/mi', $block, $matches, PREG_SET_ORDER ) ) {
265	return $headers;
266	}
267
268	foreach ( $matches as $match ) {
269	$headers[ strtolower( $match[1] ) ] = trim( $match[2] );
270	}
271
272	return $headers;
273	}
274
275	/**
276	* The settings the snapshot describes, ready to be compared or written.
277	*
278	* @return array Section-shaped array, empty when there is nothing to offer.
279	*/
280	public static function get_recovered_settings() {
281	return self::settings_from_block( self::get_raw_block() );
282	}
283
284	/**
285	* The settings a given .htaccess content describes, reading only our block.
286	*
287	* @since 2.10.0
288	* @param string $content Full .htaccess content.
289	* @return array
290	*/
291	public static function settings_from_content( $content ) {
292	return self::settings_from_block( self::extract_block( (string) $content ) );
293	}
294
295	/**
296	* Turn one Vigilant header block into a settings array.
297	*
298	* @param string $block Block content.
299	* @return array
300	*/
301	private static function settings_from_block( $block ) {
302	if ( '' === $block ) {
303	return array();
304	}
305
306	$headers = self::read_headers( $block );
307
308	if ( empty( $headers ) ) {
309	return array();
310	}
311
312	$allowed = self::allowed();
313	$recovered = array();
314
315	if ( isset( $headers['x-frame-options'] )
316	&& in_array( strtoupper( $headers['x-frame-options'] ), $allowed['x_frame_options'], true )
317	) {
318	$recovered['x_frame_options'] = strtoupper( $headers['x-frame-options'] );
319	}
320
321	if ( isset( $headers['referrer-policy'] )
322	&& in_array( strtolower( $headers['referrer-policy'] ), $allowed['referrer_policy'], true )
323	) {
324	$recovered['referrer_policy'] = strtolower( $headers['referrer-policy'] );
325	}
326
327	$hsts = self::read_hsts( $headers );
328
329	if ( ! empty( $hsts ) ) {
330	$recovered['hsts'] = $hsts;
331	}
332
333	$csp = self::read_csp( $headers );
334
335	if ( ! empty( $csp ) ) {
336	$recovered['csp'] = $csp;
337	}
338
339	$cross_origin = self::read_cross_origin( $headers, $allowed );
340
341	if ( ! empty( $cross_origin ) ) {
342	$recovered['cross_origin_policies'] = $cross_origin;
343	}
344
345	return $recovered;
346	}
347
348	/**
349	* Strict-Transport-Security back into the hsts sub-array.
350	*
351	* @param array $headers Parsed headers.
352	* @return array
353	*/
354	private static function read_hsts( $headers ) {
355	if ( ! isset( $headers['strict-transport-security'] ) ) {
356	return array();
357	}
358
359	$value = strtolower( $headers['strict-transport-security'] );
360
361	if ( ! preg_match( '/max-age\s=\s(\d+)/', $value, $match ) ) {
362	return array();
363	}
364
365	return array(
366	'enabled' => true,
367	'max_age' => min( absint( $match[1] ), YEAR_IN_SECONDS * 2 ),
368	'include_subdomains' => ( false !== strpos( $value, 'includesubdomains' ) ),
369	'preload' => ( false !== strpos( $value, 'preload' ) ),
370	);
371	}
372
373	/**
374	* Content-Security-Policy back into the csp sub-array.
375	*
376	* @param array $headers Parsed headers.
377	* @return array
378	*/
379	private static function read_csp( $headers ) {
380	$report_only = isset( $headers['content-security-policy-report-only'] );
381	$key = $report_only ? 'content-security-policy-report-only' : 'content-security-policy';
382
383	if ( ! isset( $headers[ $key ] ) \|\| '' === $headers[ $key ] ) {
384	return array();
385	}
386
387	$directives = array();
388
389	foreach ( explode( ';', $headers[ $key ] ) as $chunk ) {
390	$chunk = trim( $chunk );
391
392	if ( '' === $chunk ) {
393	continue;
394	}
395
396	$parts = preg_split( '/\s+/', $chunk, 2 );
397	$name = strtolower( trim( $parts[0] ) );
398
399	// Directive names are a fixed vocabulary; anything else is noise.
400	if ( ! preg_match( '/^[a-z0-9-]+$/', $name ) ) {
401	continue;
402	}
403
404	// A valueless directive (upgrade-insecure-requests) is a flag.
405	$directives[ $name ] = isset( $parts[1] ) ? trim( $parts[1] ) : true;
406	}
407
408	if ( empty( $directives ) ) {
409	return array();
410	}
411
412	return array(
413	'enabled' => true,
414	'report_only' => $report_only,
415	'directives' => $directives,
416	);
417	}
418
419	/**
420	* The three cross-origin policies.
421	*
422	* @param array $headers Parsed headers.
423	* @param array $allowed Allowed values.
424	* @return array
425	*/
426	private static function read_cross_origin( $headers, $allowed ) {
427	$map = array(
428	'cross-origin-opener-policy' => 'opener_policy',
429	'cross-origin-embedder-policy' => 'embedder_policy',
430	'cross-origin-resource-policy' => 'resource_policy',
431	);
432
433	$out = array();
434
435	foreach ( $map as $header => $key ) {
436	if ( ! isset( $headers[ $header ] ) ) {
437	continue;
438	}
439
440	$value = strtolower( $headers[ $header ] );
441
442	if ( in_array( $value, $allowed[ $key ], true ) ) {
443	$out[ $key ] = $value;
444	}
445	}
446
447	return $out;
448	}
449
450	/**
451	* What the owner is shown before deciding: setting, value now, value found.
452	*
453	* Only rows that would actually change are returned, so the list is the
454	* change itself rather than a dump of the whole section.
455	*
456	* @param Vigilante_Settings $settings Settings instance.
457	* @return array List of array( label, current, recovered ).
458	*/
459	public static function get_diff( $settings ) {
460	$recovered = self::get_recovered_settings();
461
462	if ( empty( $recovered ) ) {
463	return array();
464	}
465
466	$current = $settings->get_section( 'security_headers' );
467	$rows = array();
468
469	$labels = array(
470	'x_frame_options' => __( 'X-Frame-Options', 'vigilante' ),
471	'referrer_policy' => __( 'Referrer-Policy', 'vigilante' ),
472	'hsts' => __( 'HSTS', 'vigilante' ),
473	'csp' => __( 'Content Security Policy', 'vigilante' ),
474	'cross_origin_policies' => __( 'Cross-Origin Policies', 'vigilante' ),
475	);
476
477	foreach ( $recovered as $key => $value ) {
478	$now = isset( $current[ $key ] ) ? $current[ $key ] : null;
479
480	/*
481	* Shown as what restoring would actually leave behind, which is the
482	* merge restore() performs, not the recovered value on its own.
483	*
484	* The difference is not academic. A COEP of unsafe-none is never
485	* written to the file, so it cannot be read back, and the recovered
486	* cross_origin_policies arrives with two of its three keys. Described
487	* raw, that read as "your COEP is about to disappear", when
488	* array_replace_recursive() keeps it untouched. Nothing may be written
489	* that was not shown, and nothing may be shown that will not happen.
490	*/
491	$applied = ( is_array( $value ) && is_array( $now ) )
492	? array_replace_recursive( $now, $value )
493	: $value;
494
495	/*
496	* Compared on the values, never on their descriptions. A description
497	* is a summary and summaries collide: two different policies both
498	* described as "14 directives" would have compared equal, so the CSP
499	* would have been restored without ever appearing in the list the
500	* owner approves.
501	*/
502	if ( $now == $applied ) { // phpcs:ignore WordPress.PHP.StrictComparisons.LooseComparison -- arrays compare by key/value pairs regardless of order, which is what "same configuration" means here.
503	continue;
504	}
505
506	$rows[] = array(
507	'label' => isset( $labels[ $key ] ) ? $labels[ $key ] : $key,
508	'current' => self::describe( $key, $now ),
509	'recovered' => self::describe( $key, $applied ),
510	'detail' => ( 'csp' === $key ) ? self::describe_csp_changes( $now, $applied ) : '',
511	);
512	}
513
514	return $rows;
515	}
516
517	/**
518	* Name the CSP directives that actually differ.
519	*
520	* Two policies can both be "14 directives" and mean very different things,
521	* so the count alone is not something anyone can make a decision on.
522	*
523	* @param mixed $now Current csp value.
524	* @param array $recovered Recovered csp value.
525	* @return string
526	*/
527	private static function describe_csp_changes( $now, $recovered ) {
528	$before = ( is_array( $now ) && isset( $now['directives'] ) && is_array( $now['directives'] ) ) ? $now['directives'] : array();
529	$after = ( isset( $recovered['directives'] ) && is_array( $recovered['directives'] ) ) ? $recovered['directives'] : array();
530
531	$changed = array();
532
533	foreach ( array_keys( $before + $after ) as $name ) {
534	$a = isset( $before[ $name ] ) ? $before[ $name ] : null;
535	$b = isset( $after[ $name ] ) ? $after[ $name ] : null;
536
537	if ( $a !== $b ) {
538	$changed[] = $name;
539	}
540	}
541
542	if ( empty( $changed ) ) {
543	return '';
544	}
545
546	sort( $changed );
547
548	/* translators: %s: comma-separated list of Content Security Policy directive names. */
549	return sprintf( __( 'Differs in: %s', 'vigilante' ), implode( ', ', $changed ) );
550	}
551
552	/**
553	* One-line human description of a setting value.
554	*
555	* @param string $key Setting key.
556	* @param mixed $value Setting value.
557	* @return string
558	*/
559	private static function describe( $key, $value ) {
560	if ( null === $value \|\| '' === $value ) {
561	return __( 'not set', 'vigilante' );
562	}
563
564	if ( 'hsts' === $key ) {
565	if ( empty( $value['enabled'] ) ) {
566	return __( 'off', 'vigilante' );
567	}
568
569	$parts = array( 'max-age=' . absint( isset( $value['max_age'] ) ? $value['max_age'] : 0 ) );
570
571	if ( ! empty( $value['include_subdomains'] ) ) {
572	$parts[] = 'includeSubDomains';
573	}
574
575	if ( ! empty( $value['preload'] ) ) {
576	$parts[] = 'preload';
577	}
578
579	return implode( '; ', $parts );
580	}
581
582	if ( 'csp' === $key ) {
583	if ( empty( $value['enabled'] ) ) {
584	return __( 'off', 'vigilante' );
585	}
586
587	$count = ( isset( $value['directives'] ) && is_array( $value['directives'] ) ) ? count( $value['directives'] ) : 0;
588
589	/* translators: %d: number of Content Security Policy directives. */
590	return sprintf( _n( '%d directive', '%d directives', $count, 'vigilante' ), $count );
591	}
592
593	if ( 'cross_origin_policies' === $key ) {
594	$bits = array();
595
596	foreach ( array( 'opener_policy', 'embedder_policy', 'resource_policy' ) as $sub ) {
597	if ( ! empty( $value[ $sub ] ) ) {
598	$bits[] = $value[ $sub ];
599	}
600	}
601
602	return $bits ? implode( ', ', $bits ) : __( 'not set', 'vigilante' );
603	}
604
605	return is_scalar( $value ) ? (string) $value : __( 'not set', 'vigilante' );
606	}
607
608	/**
609	* Write the recovered settings, keeping an undo copy.
610	*
611	* @param Vigilante_Settings $settings Settings instance.
612	* @return true\|WP_Error
613	*/
614	public static function restore( $settings ) {
615	$recovered = self::get_recovered_settings();
616
617	if ( empty( $recovered ) ) {
618	return new WP_Error( 'nothing_to_restore', __( 'There is nothing to restore.', 'vigilante' ) );
619	}
620
621	$current = $settings->get_section( 'security_headers' );
622
623	// Keep what it looked like first, so this is reversible too.
624	update_option( self::UNDO_OPTION, $current, false );
625
626	$settings->update_section( 'security_headers', array_replace_recursive( $current, $recovered ) );
627	$settings->clear_cache();
628
629	update_option( self::DISMISSED_OPTION, 1, false );
630
631	return self::rewrite_rules();
632	}
633
634	/**
635	* Put back the settings as they were before restoring.
636	*
637	* @param Vigilante_Settings $settings Settings instance.
638	* @return true\|WP_Error
639	*/
640	public static function undo( $settings ) {
641	$previous = get_option( self::UNDO_OPTION );
642
643	if ( ! is_array( $previous ) \|\| empty( $previous ) ) {
644	return new WP_Error( 'no_undo', __( 'There is nothing to undo.', 'vigilante' ) );
645	}
646
647	$settings->update_section( 'security_headers', $previous );
648	$settings->clear_cache();
649	delete_option( self::UNDO_OPTION );
650
651	return self::rewrite_rules();
652	}
653
654	/**
655	* Stop offering the recovery, without changing any setting.
656	*/
657	public static function dismiss() {
658	update_option( self::DISMISSED_OPTION, 1, false );
659	}
660
661	/**
662	* Regenerate the .htaccess from the settings, through the normal path.
663	*
664	* @return true\|WP_Error
665	*/
666	private static function rewrite_rules() {
667	$settings = new Vigilante_Settings();
668
669	if ( ! $settings->is_module_enabled( 'security_headers' ) ) {
670	return true;
671	}
672
673	require_once VIGILANTE_INCLUDES_DIR . 'class-security-headers.php';
674
675	$result = ( new Vigilante_Security_Headers( $settings ) )->apply_rules();
676
677	// Not being on Apache is not a failure: the settings are what matters,
678	// and there is no block to write on nginx.
679	if ( is_wp_error( $result ) && 'not_apache' === $result->get_error_code() ) {
680	return true;
681	}
682
683	return is_wp_error( $result ) ? $result : true;
684	}
685	}
686