jetpack / jetpack_vendor / automattic / jetpack-backup / src / rest / class-rest-controller.php

class-rest-controller.php in Jetpack – WP Security, Backup, Speed, & Growth 16.3-a.5, at jetpack_vendor/automattic/jetpack-backup/src/rest/class-rest-controller.php

446 lines 17.6 KB

Raw Download Zip

1	<?php
2	/**
3	* REST controller for the modernized Backup dashboard.
4	*
5	* @package automattic/jetpack-backup-plugin
6	*/
7
8	namespace Automattic\Jetpack\Backup\V0005\REST;
9
10	use Automattic\Jetpack\Backup\V0005\Jetpack_Backup;
11	use Automattic\Jetpack\Connection\Manager as Connection_Manager;
12	use Jetpack_Options;
13	use WP_Error;
14	use WP_REST_Request;
15
16	if ( ! defined( 'ABSPATH' ) ) {
17	exit( 0 );
18	}
19
20	/**
21	* Registers REST routes that back the modernized dashboard.
22	*
23	* Each bridge class declares its routes via `register_routes()` and uses
24	* the shared `permission_check()` helper for the `manage_options` gate.
25	* Routes only register when `is_modernized()` is true, so the
26	* legacy plugin is byte-identical when it is false.
27	*/
28	class Rest_Controller {
29
30	/**
31	* Every category WPCOM's rewind endpoints recognize in a `types` map.
32	*
33	* The first six are the whole-site checklist the Restore and Download
34	* screens render.
35	*
36	* `paths` covers the granular download shape only — `types: { paths:
37	* true }`, paired with `include_path_list` / `exclude_path_list`.
38	* Nothing sends it yet (C3), but leaving it out would make the file
39	* browser's granular download fail closed with a confusing 400 the day
40	* it is wired up.
41	*
42	* It does not carry over to granular restore, whatever that ends up
43	* spelling: the v1 route took `types: 'paths'` as a bare string, which
44	* is not a map at all and would never reach this allowlist. Granular
45	* restore has to establish its own shape against the v2 route before
46	* anything here can claim to cover it.
47	*
48	* This list has to grow if VaultPress adds a category. WPCOM's own
49	* route deliberately does not allowlist, so that it stays open to new
50	* types; we can afford to be stricter because we also own the UI that
51	* produces the values, and here a value that names nothing is the
52	* dangerous case rather than merely a useless one.
53	*
54	* @var string[]
55	*/
56	private const CATEGORIES = array(
57	'themes',
58	'plugins',
59	'roots',
60	'contents',
61	'sqls',
62	'uploads',
63	'paths',
64	);
65
66	/**
67	* Hook entry point. Registers all bridge routes if `is_modernized()` is true.
68	*
69	* @return void
70	*/
71	public static function register_routes() {
72	if ( ! Jetpack_Backup::is_modernized() ) {
73	return;
74	}
75
76	Capabilities_Bridge::register_routes();
77	Activity_Log_Bridge::register_routes();
78	File_Browser_Bridge::register_routes();
79	Download_Bridge::register_routes();
80	Restore_Bridge::register_routes();
81	}
82
83	/**
84	* Permission check shared by every modernized-dashboard route.
85	*
86	* Mirrors the activity-log package's pattern: `manage_options` is
87	* necessary but not sufficient — every bridge eventually proxies a
88	* WPCOM endpoint that's user-gated, so a site admin who isn't
89	* personally WPCOM-linked needs a clearer error than the opaque
90	* "Only Administrators can query…" WPCOM returns.
91	*
92	* @return bool\|WP_Error True when the current user can call the bridges, WP_Error otherwise.
93	*/
94	public static function permission_check() {
95	if ( ! current_user_can( 'manage_options' ) ) {
96	return false;
97	}
98
99	if ( ! ( new Connection_Manager() )->is_user_connected() ) {
100	return new WP_Error(
101	'user_not_connected',
102	__( 'Your WordPress.com account is not connected to this site.', 'jetpack-backup-pkg' ),
103	array( 'status' => 403 )
104	);
105	}
106
107	return true;
108	}
109
110	/**
111	* Returns the site's WPCOM blog id, or a `not_connected` WP_Error
112	* when the site hasn't been registered yet. Shared across the bridges
113	* so the `sprintf( '/sites/%d/…', $blog_id )` upstream path is never
114	* built with an empty id.
115	*
116	* @return int\|WP_Error Blog id, or WP_Error when not connected.
117	*/
118	public static function get_blog_id_or_error() {
119	$blog_id = (int) Jetpack_Options::get_option( 'id' );
120	if ( ! $blog_id ) {
121	return new WP_Error(
122	'not_connected',
123	__( 'This site is not connected to Jetpack.', 'jetpack-backup-pkg' ),
124	array( 'status' => 412 )
125	);
126	}
127	return $blog_id;
128	}
129
130	/**
131	* Rebuild a restore/download `types` parameter as a named map.
132	*
133	* The PHP counterpart of the client's `requireTypes`, and the reason
134	* it exists here rather than being trusted from the request: WordPress
135	* validates `'type' => 'object'` with `rest_is_object()`, which is
136	* `is_array()`. A JSON list therefore passes validation and arrives as
137	* a PHP list, whose numeric keys WPCOM reads as category names. The
138	* route schema rejects the realistic version of that, but only because
139	* the members fail a boolean check — shape itself is never asserted —
140	* so the guarantee is made here, where the payload is actually built.
141	*
142	* Only known categories with a truthy value survive, and every
143	* surviving value is normalized to `true`. Values are read with
144	* `rest_sanitize_boolean()` so a form-encoded `"false"` or `"0"` means
145	* skip rather than select.
146	*
147	* Unknown keys are dropped rather than forwarded, which is what makes
148	* `request_names_no_types()` a total guard: without it a payload naming
149	* only categories WPCOM does not recognize would satisfy the guard and
150	* be sent on, and what WPCOM does with a `types` that matches nothing
151	* is not characterized. Dropping them means such a payload names
152	* nothing, and is refused. The realistic way to get there is not an
153	* attacker — an admin who can craft the request can already omit
154	* `types` for a whole-site operation — but a future client-side typo:
155	* renaming a checklist key `sqls` to `sql` would otherwise go through
156	* silently.
157	*
158	* @param mixed $types Raw `types` parameter from the request.
159	* @return array<string, true> Named types, empty when none are selected.
160	*/
161	public static function named_types( $types ) {
162	if ( ! is_array( $types ) && ! is_object( $types ) ) {
163	return array();
164	}
165
166	$named = array();
167	foreach ( (array) $types as $key => $value ) {
168	if ( in_array( $key, self::CATEGORIES, true ) && rest_sanitize_boolean( $value ) ) {
169	$named[ $key ] = true;
170	}
171	}
172	return $named;
173	}
174
175	/**
176	* Whether the request supplied a `types` parameter that names no category.
177	*
178	* The distinction this draws is the whole point of the helper, and it
179	* is the opposite of what it looks like. An absent `types` is a
180	* valid, deliberate request for every category — WPCOM's contract is
181	* "omit it for everything" — so the mutations leave the key out for a
182	* whole-site restore or a full archive. A supplied `types` that
183	* survives into nothing is the other thing entirely: the caller tried
184	* to name categories and named none, and forwarding that as an
185	* omission would quietly upgrade "restore nothing" into "restore
186	* everything", against a live site.
187	*
188	* It takes the request rather than the value because the value cannot
189	* answer the question. `{"types": null}` is supplied and names nothing,
190	* but arrives as the same `null` an omitted key does — and the schema
191	* never sees it, since `WP_REST_Request::has_valid_params()` skips
192	* `validate_callback` for a null param. `has_param()` is the only thing
193	* that knows the key was on the wire.
194	*
195	* Nothing upstream catches it on both routes. The v2 restore route
196	* rejects a `types` naming nothing, but `/rewind/downloads` does not,
197	* so the guarantee has to be made here for the pair to behave alike.
198	*
199	* @param WP_REST_Request $request The REST request.
200	* @return bool True when the caller supplied a `types` that names no category.
201	*/
202	public static function request_names_no_types( WP_REST_Request $request ) {
203	if ( ! $request->has_param( 'types' ) ) {
204	return false;
205	}
206
207	return ! self::named_types( $request->get_param( 'types' ) );
208	}
209
210	/**
211	* Convert a transport-level failure into a bridge error.
212	*
213	* `Client::wpcom_json_api_request_as_*` answers with a `WP_Error` when
214	* the request never reached WPCOM at all — DNS, TLS, or the cURL
215	* timeout behind JETPACK-2173's "cURL error 28". Returning that error
216	* unchanged hands cURL's own text to the browser, where the dashboard
217	* renders the message verbatim in a notice; it also carries no
218	* `status`, so core answers 500 for what is really a reachability
219	* problem rather than a server fault.
220	*
221	* The raw text is preserved under `transport` rather than discarded.
222	* It is the only part a support agent can act on, and it is the same
223	* reason the non-200 branches forward WPCOM's status instead of
224	* flattening it.
225	*
226	* Always 502: telling a timeout from a refused connection would mean
227	* matching on cURL's English message text, and no caller reads the
228	* difference.
229	*
230	* @param WP_Error $error Transport error from the HTTP client.
231	* @param string $code Bridge error code for the operation that failed.
232	* @return WP_Error
233	*/
234	public static function transport_error( WP_Error $error, $code ) {
235	return new WP_Error(
236	$code,
237	__( 'Could not reach WordPress.com. Check your connection and try again.', 'jetpack-backup-pkg' ),
238	array(
239	'status' => 502,
240	'transport' => array(
241	'code' => $error->get_error_code(),
242	'message' => $error->get_error_message(),
243	),
244	)
245	);
246	}
247
248	/**
249	* Longest either half of a forwarded upstream reason may be.
250	*
251	* Neither field is bounded upstream and both travel to the browser on
252	* every failed request, so a VaultPress stack trace in `message` would
253	* otherwise be copied out verbatim.
254	*
255	* @var int
256	*/
257	private const REASON_MAX_LENGTH = 200;
258
259	/**
260	* Convert a non-200 answer from WordPress.com into a bridge error.
261	*
262	* The counterpart of `transport_error()`, for the failure where the
263	* request did arrive and WordPress.com refused it. Every bridge used to
264	* spell this branch out for itself, and every spelling threw away the
265	* only part that says why: a plan problem and an expired token both
266	* reached a support agent as `restore_initiate_failed` / "Could not
267	* start the backup restore.", distinguishable only by a status code.
268	*
269	* WordPress.com's own reason is preserved under `wpcom`, deliberately
270	* shaped like `transport_error()`'s `transport` key. The client decides
271	* what to do with it: `failureMessage()` in `_helpers.ts` maps the
272	* codes whose meaning is the code itself, and renders the message for
273	* the ones — `rewind_error`, `authorization_required` — where one code
274	* spans several unrelated situations and only the sentence tells them
275	* apart.
276	*
277	* That the message can reach a reader is why the flattening and the
278	* 200-character clip below are not housekeeping. They are the whole
279	* reason it is safe to render, so do not relax them. It stays a plain
280	* string all the way out; the client escapes it by rendering it as
281	* React children.
282	*
283	* Only those two fields are forwarded, never the body — an error body
284	* is unbounded and can echo the request that produced it.
285	*
286	* The status is clamped to the failure range rather than merely tested
287	* for truthiness, and that is load-bearing. The retrieval helper hands
288	* back whatever the transport put there, and its callers must cast
289	* before comparing — a numeric-string `'200'` fails `200 !==
290	* $status_code`, which routed a perfectly good response into the
291	* failure branch. Every status comparison in this package casts now —
292	* bridges and legacy routes alike — so this function should no longer
293	* be reachable with a success code; the clamp stays because "should
294	* not" is not "cannot", and the cost of being wrong is one-sided.
295	* Forwarding a 200 would set `data.status` to 200, WordPress would
296	* serve the error envelope as HTTP 200, `apiFetch` would resolve
297	* instead of rejecting, and `apiCall()` would never throw — so a failed
298	* restore would run the mutation's `onSuccess` and report a restore
299	* that never started. A visible failure becoming an invisible false
300	* success is the worst outcome available on a destructive operation, so
301	* anything outside 4xx/5xx becomes a 500.
302	*
303	* The same clamp is what keeps junk out. `(int)` is a total function:
304	* `'2 Bad'` is 2, `3.7` is 3, `true` is 1, and a zero must never reach
305	* a `WP_Error` at all, because core hands the status to
306	* `status_header()` and a zero there emits an invalid status line.
307	*
308	* What the cast does buy, and the reason it is here rather than an
309	* `is_int()` test, is that a genuine `'404'` from a transport that
310	* reports statuses as strings now travels as 404 instead of being
311	* flattened to 500. The client is literal about the type too:
312	* `isAmbiguousFailure()` only reads a status that is already a number.
313	*
314	* @param array\|\WP_Error $response The wp_remote_* response. Non-200 by the time it gets here.
315	* @param string $code Bridge error code for the operation that failed.
316	* @param string $message Translated message for the reader.
317	* @return WP_Error
318	*/
319	public static function upstream_error( $response, $code, $message ) {
320	$status_code = (int) wp_remote_retrieve_response_code( $response );
321
322	$data = array( 'status' => $status_code >= 400 && $status_code <= 599 ? $status_code : 500 );
323
324	$reason = self::upstream_reason( wp_remote_retrieve_body( $response ) );
325	if ( ! empty( $reason ) ) {
326	$data['wpcom'] = $reason;
327	}
328
329	return new WP_Error( $code, $message, $data );
330	}
331
332	/**
333	* Read WordPress.com's own reason out of a response body.
334	*
335	* Three shapes have to be read here and they disagree about where the
336	* reason lives. A wpcom/v2 route serializes a `WP_Error` as `{ code,
337	* message, data }`. The older v1 envelope names that same token `error`
338	* and keeps prose beside it in `message`. And the restore endpoint's
339	* own `{ ok: false, error }` body puts VaultPress's sentence directly
340	* in `error`, with no token anywhere.
341	*
342	* So `error` is sometimes a token and sometimes a sentence, and the
343	* only thing separating them is shape: a `WP_Error` code has no
344	* whitespace in it, and a VaultPress refusal is a sentence. Sorting on
345	* that is what keeps the two halves honest. The client matches `code`
346	* against a list of codes it knows, so a sentence landing there would
347	* become a key that can never match — the reason lost again, in a
348	* quieter way.
349	*
350	* Sorting is all it does, though: nothing is ever discarded. When
351	* `error` holds a sentence and `message` holds another — which is
352	* what a VaultPress refusal wrapped in a generic envelope looks like —
353	* both are kept, `error` first, because that is the specific half. An
354	* earlier revision promoted `error` only when `message` was empty, and
355	* so threw away the specific reason in exactly the shape this function
356	* exists to read.
357	*
358	* `/u` on the whitespace test is not cosmetic. Without it `\s` is
359	* ASCII-only, so a sentence spaced with U+00A0 has "no whitespace" and
360	* lands in `code` — the precise outcome the sort is here to prevent.
361	*
362	* @param string\|array $body Raw response body, or one already decoded.
363	* @return array<string, string> `code` and/or `message`; empty when the body names no reason.
364	*/
365	public static function upstream_reason( $body ) {
366	$decoded = is_array( $body ) ? $body : json_decode( (string) $body, true );
367	if ( ! is_array( $decoded ) ) {
368	return array();
369	}
370
371	$token = '';
372	foreach ( array( 'code', 'error' ) as $key ) {
373	if ( isset( $decoded[ $key ] ) && is_string( $decoded[ $key ] ) && '' !== trim( $decoded[ $key ] ) ) {
374	$token = trim( $decoded[ $key ] );
375	break;
376	}
377	}
378
379	$prose = isset( $decoded['message'] ) && is_string( $decoded['message'] ) ? trim( $decoded['message'] ) : '';
380
381	$reason = array();
382	$sentences = array();
383
384	if ( '' !== $token && ! preg_match( '/\s/u', $token ) ) {
385	$reason['code'] = self::clip_reason( $token );
386	} elseif ( '' !== $token ) {
387	$sentences[] = $token;
388	}
389
390	// Guarded against the duplicate rather than assumed away: some
391	// envelopes repeat the same text in both keys, and joining it to
392	// itself would say everything twice inside a budget meant for one.
393	if ( '' !== $prose && ! in_array( $prose, $sentences, true ) ) {
394	$sentences[] = $prose;
395	}
396
397	if ( ! empty( $sentences ) ) {
398	$reason['message'] = self::clip_reason( implode( ' ', $sentences ) );
399	}
400
401	return $reason;
402	}
403
404	/**
405	* Flatten and shorten one half of an upstream reason.
406	*
407	* Newlines go first so a multi-line upstream message cannot spend the
408	* whole budget on indentation before it says anything.
409	*
410	* `mb_substr()` rather than `substr()`, and the reason is mostly not
411	* the exotic one. A byte-wise cut spends the budget in bytes, so a
412	* reason written in a script that costs three bytes a character keeps
413	* a third of what it was allotted — 67 characters of 200, in the test
414	* that pins this. The encoding damage is the smaller half: the cut
415	* lands inside a character and leaves the field invalid UTF-8, which
416	* `wp_json_encode()` does not reject — its sanity check silently
417	* rewrites the broken bytes, so the reason arrives with a `?` on the
418	* end and nothing anywhere says why.
419	*
420	* The flatten runs in Unicode mode so a non-breaking or ideographic
421	* space collapses like any other, which also means it returns null on
422	* invalid UTF-8. The ASCII pass stands behind it so the bound is still
423	* enforced in that case. Unreachable in practice — every string that
424	* gets here came out of a `json_decode()`, which refuses invalid
425	* UTF-8 outright — but a silent null would turn the whole reason into
426	* an empty string, which is a poor way to find out.
427	*
428	* @param string $value Raw upstream text.
429	* @return string
430	*/
431	private static function clip_reason( $value ) {
432	$flattened = preg_replace( '/\s+/u', ' ', $value );
433	if ( null === $flattened ) {
434	$flattened = preg_replace( '/\s+/', ' ', $value );
435	}
436
437	$value = trim( (string) $flattened );
438
439	if ( mb_strlen( $value, 'UTF-8' ) <= self::REASON_MAX_LENGTH ) {
440	return $value;
441	}
442
443	return rtrim( mb_substr( $value, 0, self::REASON_MAX_LENGTH, 'UTF-8' ) ) . '…';
444	}
445	}
446