| 1 |
<?php |
| 2 |
/** |
| 3 |
* Download REST bridge — proxies the /sites/{id}/rewind/downloads |
| 4 |
* collection and its per-download status. |
| 5 |
* |
| 6 |
* @package automattic/jetpack-backup-plugin |
| 7 |
*/ |
| 8 |
|
| 9 |
namespace Automattic\Jetpack\Backup\V0005\REST; |
| 10 |
|
| 11 |
use Automattic\Jetpack\Connection\Client; |
| 12 |
use WP_Error; |
| 13 |
use WP_REST_Request; |
| 14 |
use WP_REST_Server; |
| 15 |
|
| 16 |
if ( ! defined( 'ABSPATH' ) ) { |
| 17 |
exit( 0 ); |
| 18 |
} |
| 19 |
|
| 20 |
/** |
| 21 |
* Download endpoints powering the Download screen: |
| 22 |
* - POST /jetpack/v4/backups/download/{rewindId} → ask WPCOM to build the archive |
| 23 |
* - GET /jetpack/v4/backups/download/{rewindId}/status → poll progress |
| 24 |
*/ |
| 25 |
class Download_Bridge { |
| 26 |
|
| 27 |
/** |
| 28 |
* Register routes. |
| 29 |
* |
| 30 |
* @return void |
| 31 |
*/ |
| 32 |
public static function register_routes() { |
| 33 |
register_rest_route( |
| 34 |
'jetpack/v4', |
| 35 |
'/backups/download/(?P<rewind_id>[A-Za-z0-9.\-]+)', |
| 36 |
array( |
| 37 |
'methods' => WP_REST_Server::CREATABLE, |
| 38 |
'callback' => array( __CLASS__, 'initiate_download' ), |
| 39 |
'permission_callback' => array( Rest_Controller::class, 'permission_check' ), |
| 40 |
'args' => array( |
| 41 |
'rewind_id' => array( |
| 42 |
'type' => 'string', |
| 43 |
'required' => true, |
| 44 |
), |
| 45 |
'types' => array( |
| 46 |
'type' => 'object', |
| 47 |
// Values must be booleans. WordPress validates |
| 48 |
// `object` with `rest_is_object()`, which is just |
| 49 |
// `is_array()`, so without this a JSON list of |
| 50 |
// category names passes and reaches WPCOM as a list |
| 51 |
// whose numeric keys it reads as category names. |
| 52 |
// This rejects that with a 400 before the callback |
| 53 |
// runs; `Rest_Controller::named_types()` makes the |
| 54 |
// shape guarantee that a value check cannot. |
| 55 |
'additionalProperties' => array( 'type' => 'boolean' ), |
| 56 |
), |
| 57 |
// Opaque `/rewind/backup/ls` entry ids, only meaningful |
| 58 |
// alongside `types: { paths: true }` — see `initiate_download()`. |
| 59 |
'include_path_list' => array( |
| 60 |
'type' => 'array', |
| 61 |
'items' => array( 'type' => 'string' ), |
| 62 |
), |
| 63 |
'exclude_path_list' => array( |
| 64 |
'type' => 'array', |
| 65 |
'items' => array( 'type' => 'string' ), |
| 66 |
), |
| 67 |
), |
| 68 |
) |
| 69 |
); |
| 70 |
|
| 71 |
register_rest_route( |
| 72 |
'jetpack/v4', |
| 73 |
'/backups/download/(?P<rewind_id>[A-Za-z0-9.\-]+)/status', |
| 74 |
array( |
| 75 |
'methods' => WP_REST_Server::READABLE, |
| 76 |
'callback' => array( __CLASS__, 'get_download_status' ), |
| 77 |
'permission_callback' => array( Rest_Controller::class, 'permission_check' ), |
| 78 |
'args' => array( |
| 79 |
'rewind_id' => array( |
| 80 |
'type' => 'string', |
| 81 |
'required' => true, |
| 82 |
), |
| 83 |
'download_id' => array( |
| 84 |
'type' => 'integer', |
| 85 |
'required' => true, |
| 86 |
), |
| 87 |
), |
| 88 |
) |
| 89 |
); |
| 90 |
} |
| 91 |
|
| 92 |
/** |
| 93 |
* Initiate a backup download. |
| 94 |
* |
| 95 |
* Proxies POST wpcom/v2 /sites/{id}/rewind/downloads. The previous |
| 96 |
* target — a `prepare-download` path under the backup — is not a |
| 97 |
* registered route and answered `rest_no_route` on every site tested, |
| 98 |
* so the Download screen could never have worked. |
| 99 |
* |
| 100 |
* The rewind id travels in the request **body**, in full. It is not a |
| 101 |
* path segment here, and truncating its decimal suffix would address |
| 102 |
* a different backup than the one the reader picked. |
| 103 |
* |
| 104 |
* Returns `{ id }` for the React layer (the WPCOM key is `downloadId`). |
| 105 |
* |
| 106 |
* @param WP_REST_Request $request The REST request. |
| 107 |
* @return \WP_REST_Response|WP_Error |
| 108 |
*/ |
| 109 |
public static function initiate_download( WP_REST_Request $request ) { |
| 110 |
$blog_id = Rest_Controller::get_blog_id_or_error(); |
| 111 |
if ( is_wp_error( $blog_id ) ) { |
| 112 |
return $blog_id; |
| 113 |
} |
| 114 |
$rewind_id = (string) $request->get_param( 'rewind_id' ); |
| 115 |
$types = $request->get_param( 'types' ); |
| 116 |
|
| 117 |
$named_types = Rest_Controller::named_types( $types ); |
| 118 |
$include = self::path_list( $request, 'include_path_list' ); |
| 119 |
$exclude = self::path_list( $request, 'exclude_path_list' ); |
| 120 |
|
| 121 |
// Nothing upstream checks this pairing: VaultPress reads the path |
| 122 |
// lists only for the `paths` type, so a list beside any other |
| 123 |
// category answers 200 with a *full-site* archive. |
| 124 |
// |
| 125 |
// `has_param()` too, for both keys: `path_list()` trims a blank list |
| 126 |
// away, and a caller that named files must not fall through as one |
| 127 |
// that named none. |
| 128 |
if ( $include || $exclude |
| 129 |
|| $request->has_param( 'include_path_list' ) |
| 130 |
|| $request->has_param( 'exclude_path_list' ) ) { |
| 131 |
if ( array( 'paths' ) !== array_keys( $named_types ) ) { |
| 132 |
return new WP_Error( |
| 133 |
'path_list_needs_paths_type', |
| 134 |
__( 'A file selection can only be downloaded on its own.', 'jetpack-backup-pkg' ), |
| 135 |
array( 'status' => 400 ) |
| 136 |
); |
| 137 |
} |
| 138 |
} |
| 139 |
|
| 140 |
if ( ! $include && ! $exclude && isset( $named_types['paths'] ) ) { |
| 141 |
// The mirror image, and just as silent upstream: `paths` with |
| 142 |
// nothing to scope it by yields an archive of everything. |
| 143 |
return new WP_Error( |
| 144 |
'paths_type_needs_path_list', |
| 145 |
__( 'No files were named for this download.', 'jetpack-backup-pkg' ), |
| 146 |
array( 'status' => 400 ) |
| 147 |
); |
| 148 |
} |
| 149 |
|
| 150 |
// A supplied `types` that names nothing is refused rather than |
| 151 |
// dropped. Omitting the key is not "download nothing" — WPCOM |
| 152 |
// reads an absent `types` as every category, so forwarding an |
| 153 |
// empty selection as an omission would hand back the full archive |
| 154 |
// the caller had just excluded. `/rewind/downloads` has no |
| 155 |
// server-side guard of its own, unlike the v2 restore route. |
| 156 |
if ( Rest_Controller::request_names_no_types( $request ) ) { |
| 157 |
return new WP_Error( |
| 158 |
'no_types_selected', |
| 159 |
__( 'Select at least one item to download.', 'jetpack-backup-pkg' ), |
| 160 |
array( 'status' => 400 ) |
| 161 |
); |
| 162 |
} |
| 163 |
|
| 164 |
$body = array( 'rewindId' => $rewind_id ); |
| 165 |
// Absent when the caller named no categories at all, which is how |
| 166 |
// a whole-archive download is spelled upstream. |
| 167 |
if ( ! empty( $named_types ) ) { |
| 168 |
$body['types'] = $named_types; |
| 169 |
} |
| 170 |
// Arrays, not the comma-joined string upstream also takes: that branch |
| 171 |
// sanitises the whole string before splitting, so `"a, b"` arrives as `" b"`. |
| 172 |
if ( $include ) { |
| 173 |
$body['include_path_list'] = $include; |
| 174 |
} |
| 175 |
if ( $exclude ) { |
| 176 |
$body['exclude_path_list'] = $exclude; |
| 177 |
} |
| 178 |
|
| 179 |
$response = Client::wpcom_json_api_request_as_user( |
| 180 |
sprintf( '/sites/%d/rewind/downloads', $blog_id ), |
| 181 |
'v2', |
| 182 |
array( 'method' => 'POST' ), |
| 183 |
$body, |
| 184 |
'wpcom' |
| 185 |
); |
| 186 |
|
| 187 |
if ( is_wp_error( $response ) ) { |
| 188 |
return Rest_Controller::transport_error( $response, 'download_initiate_failed' ); |
| 189 |
} |
| 190 |
|
| 191 |
// Cast because `wp_remote_retrieve_response_code()` hands back |
| 192 |
// whatever the transport put there, and a numeric string fails the |
| 193 |
// strict comparison below — routing a perfectly good response into |
| 194 |
// the failure branch, where `upstream_error()`'s clamp then reports |
| 195 |
// it as a 500. Same reasoning at every bridge; the long version is |
| 196 |
// on `Rest_Controller::upstream_error()`. |
| 197 |
$status_code = (int) wp_remote_retrieve_response_code( $response ); |
| 198 |
if ( 200 !== $status_code ) { |
| 199 |
return Rest_Controller::upstream_error( |
| 200 |
$response, |
| 201 |
'download_initiate_failed', |
| 202 |
__( 'Could not start the backup download.', 'jetpack-backup-pkg' ) |
| 203 |
); |
| 204 |
} |
| 205 |
|
| 206 |
$body = json_decode( wp_remote_retrieve_body( $response ), true ); |
| 207 |
$download_id = is_array( $body ) && isset( $body['downloadId'] ) ? (int) $body['downloadId'] : 0; |
| 208 |
if ( ! $download_id ) { |
| 209 |
return new WP_Error( |
| 210 |
'download_initiate_failed', |
| 211 |
__( 'Download response missing download id.', 'jetpack-backup-pkg' ), |
| 212 |
array( 'status' => 500 ) |
| 213 |
); |
| 214 |
} |
| 215 |
|
| 216 |
return rest_ensure_response( array( 'id' => $download_id ) ); |
| 217 |
} |
| 218 |
|
| 219 |
/** |
| 220 |
* One path-list parameter as a clean list of entries. |
| 221 |
* |
| 222 |
* Entries are trimmed and empties dropped, and the result is a PHP |
| 223 |
* list so `wp_json_encode()` emits a JSON array rather than an object. |
| 224 |
* |
| 225 |
* @param WP_REST_Request $request The REST request. |
| 226 |
* @param string $key Parameter name. |
| 227 |
* @return string[] The entries, empty when the parameter is absent or names nothing. |
| 228 |
*/ |
| 229 |
private static function path_list( WP_REST_Request $request, $key ) { |
| 230 |
$value = $request->get_param( $key ); |
| 231 |
if ( ! is_array( $value ) ) { |
| 232 |
return array(); |
| 233 |
} |
| 234 |
|
| 235 |
$entries = array(); |
| 236 |
foreach ( $value as $entry ) { |
| 237 |
if ( ! is_scalar( $entry ) ) { |
| 238 |
continue; |
| 239 |
} |
| 240 |
$entry = trim( (string) $entry ); |
| 241 |
if ( '' !== $entry ) { |
| 242 |
$entries[] = $entry; |
| 243 |
} |
| 244 |
} |
| 245 |
return $entries; |
| 246 |
} |
| 247 |
|
| 248 |
/** |
| 249 |
* Poll download status. |
| 250 |
* |
| 251 |
* Proxies GET wpcom/v2 /sites/{id}/rewind/downloads/{downloadId}. The |
| 252 |
* rewind id is not part of the upstream path — it is kept on our own |
| 253 |
* route because the client keys its poll cache on (rewindId, |
| 254 |
* downloadId), and because it keeps the two download routes |
| 255 |
* symmetrical. |
| 256 |
* |
| 257 |
* The response is projected rather than forwarded, the way restore |
| 258 |
* status already is. WPCOM's payload carries **no status field**: it |
| 259 |
* attaches keys by branch — `url` and `validUntil` once the archive |
| 260 |
* is ready, `error` when it failed, and `progress` only while the |
| 261 |
* archive is still being built. Leaving that shape for the client to |
| 262 |
* interpret is what left the React layer testing three status strings |
| 263 |
* that never appear in the payload. |
| 264 |
* |
| 265 |
* @param WP_REST_Request $request The REST request. |
| 266 |
* @return \WP_REST_Response|WP_Error |
| 267 |
*/ |
| 268 |
public static function get_download_status( WP_REST_Request $request ) { |
| 269 |
$blog_id = Rest_Controller::get_blog_id_or_error(); |
| 270 |
if ( is_wp_error( $blog_id ) ) { |
| 271 |
return $blog_id; |
| 272 |
} |
| 273 |
$download_id = (int) $request->get_param( 'download_id' ); |
| 274 |
|
| 275 |
$response = Client::wpcom_json_api_request_as_user( |
| 276 |
sprintf( '/sites/%d/rewind/downloads/%d', $blog_id, $download_id ), |
| 277 |
'v2', |
| 278 |
array(), |
| 279 |
null, |
| 280 |
'wpcom' |
| 281 |
); |
| 282 |
|
| 283 |
if ( is_wp_error( $response ) ) { |
| 284 |
return Rest_Controller::transport_error( $response, 'download_status_fetch_failed' ); |
| 285 |
} |
| 286 |
|
| 287 |
// Cast, as in `initiate_download()`. |
| 288 |
$status_code = (int) wp_remote_retrieve_response_code( $response ); |
| 289 |
if ( 200 !== $status_code ) { |
| 290 |
return Rest_Controller::upstream_error( |
| 291 |
$response, |
| 292 |
'download_status_fetch_failed', |
| 293 |
__( 'Could not fetch download status.', 'jetpack-backup-pkg' ) |
| 294 |
); |
| 295 |
} |
| 296 |
|
| 297 |
$body = json_decode( wp_remote_retrieve_body( $response ), true ); |
| 298 |
if ( ! is_array( $body ) ) { |
| 299 |
$body = array(); |
| 300 |
} |
| 301 |
|
| 302 |
$error = isset( $body['error'] ) ? (string) $body['error'] : ''; |
| 303 |
$raw_url = isset( $body['url'] ) ? (string) $body['url'] : ''; |
| 304 |
|
| 305 |
// The client puts this straight into an `<a href>`, and React does |
| 306 |
// not strip dangerous schemes — so check before handing it over. |
| 307 |
// |
| 308 |
// Deliberately a scheme check rather than `wp_http_validate_url()`, |
| 309 |
// which the file-browser bridge uses: that one is built for URLs |
| 310 |
// *this server* is about to fetch, so it also does a DNS lookup and |
| 311 |
// rejects private IPs and non-standard ports. Right there, wrong |
| 312 |
// here — this URL is only ever loaded by the browser, so those |
| 313 |
// rules could reject a perfectly good host while costing a DNS |
| 314 |
// lookup on every poll. |
| 315 |
$scheme = '' === $raw_url ? null : wp_parse_url( $raw_url, PHP_URL_SCHEME ); |
| 316 |
$url = ( 'https' === $scheme || 'http' === $scheme ) ? $raw_url : ''; |
| 317 |
|
| 318 |
// Order matters: a failed download can still carry a stale `url` |
| 319 |
// from an earlier attempt, so the error branch is checked first. |
| 320 |
if ( '' !== $error ) { |
| 321 |
$status = 'failed'; |
| 322 |
} elseif ( '' !== $url ) { |
| 323 |
$status = 'finished'; |
| 324 |
} elseif ( '' !== $raw_url ) { |
| 325 |
// A URL arrived but is not one we will hand to the browser. |
| 326 |
// Reported as a failure rather than left to fall through to |
| 327 |
// `running`, which would poll forever against a download that |
| 328 |
// is in fact finished. |
| 329 |
$status = 'failed'; |
| 330 |
$error = __( 'The download link could not be used.', 'jetpack-backup-pkg' ); |
| 331 |
} else { |
| 332 |
$status = 'running'; |
| 333 |
} |
| 334 |
|
| 335 |
return rest_ensure_response( |
| 336 |
array( |
| 337 |
'id' => isset( $body['downloadId'] ) ? (int) $body['downloadId'] : $download_id, |
| 338 |
'status' => $status, |
| 339 |
// 0-100, and absent entirely once the download leaves the |
| 340 |
// in-flight branch. Clamped rather than trusted: the client |
| 341 |
// feeds this straight to a progress bar, and the headline bug |
| 342 |
// this projection replaces was a bar being handed 10000. Making |
| 343 |
// the range true here means no future upstream change can |
| 344 |
// reproduce that symptom. |
| 345 |
'progress' => isset( $body['progress'] ) ? max( 0, min( 100, (int) $body['progress'] ) ) : 0, |
| 346 |
'url' => $url, |
| 347 |
'valid_until' => isset( $body['validUntil'] ) ? (string) $body['validUntil'] : '', |
| 348 |
'error' => $error, |
| 349 |
) |
| 350 |
); |
| 351 |
} |
| 352 |
} |
| 353 |
|
