| 1 |
<?php |
| 2 |
/** |
| 3 |
* ActivityPub HTTP Class. |
| 4 |
* |
| 5 |
* @package Activitypub |
| 6 |
*/ |
| 7 |
|
| 8 |
namespace Activitypub; |
| 9 |
|
| 10 |
use Activitypub\Collection\Actors; |
| 11 |
|
| 12 |
/** |
| 13 |
* ActivityPub HTTP Class |
| 14 |
* |
| 15 |
* @author Matthias Pfefferle |
| 16 |
*/ |
| 17 |
class Http { |
| 18 |
/** |
| 19 |
* Send a POST Request with the needed HTTP Headers |
| 20 |
* |
| 21 |
* @param string $url The URL endpoint. |
| 22 |
* @param string $body The Post Body. |
| 23 |
* @param int|null $user_id The WordPress User-ID, or null to sign with the Application key. |
| 24 |
* |
| 25 |
* @return array|\WP_Error The POST Response or an WP_Error. |
| 26 |
*/ |
| 27 |
public static function post( $url, $body, $user_id ) { |
| 28 |
/** |
| 29 |
* Fires before an HTTP POST request is made. |
| 30 |
* |
| 31 |
* @param string $url The URL endpoint. |
| 32 |
* @param string $body The POST body. |
| 33 |
* @param int|null $user_id The WordPress User ID, or null when signing with the Application key. |
| 34 |
*/ |
| 35 |
\do_action( 'activitypub_pre_http_post', $url, $body, $user_id ); |
| 36 |
|
| 37 |
/** |
| 38 |
* Filters the HTTP headers user agent string. |
| 39 |
* |
| 40 |
* @param string $user_agent The user agent string. |
| 41 |
* @param string $url The request URL. |
| 42 |
*/ |
| 43 |
$user_agent = \apply_filters( 'http_headers_useragent', 'WordPress/' . get_masked_wp_version() . '; ' . \get_bloginfo( 'url' ), $url ); |
| 44 |
|
| 45 |
/** |
| 46 |
* Filters the timeout duration for remote POST requests in ActivityPub. |
| 47 |
* |
| 48 |
* @param int $timeout The timeout value in seconds. Default 10 seconds. |
| 49 |
*/ |
| 50 |
$timeout = \apply_filters( 'activitypub_remote_post_timeout', 10 ); |
| 51 |
|
| 52 |
/* |
| 53 |
* Get the private key for signing the request. If a user ID is provided, |
| 54 |
* get the user's private key; otherwise, use the application's private key. |
| 55 |
*/ |
| 56 |
$private_key = null === $user_id ? Application::get_private_key() : Actors::get_private_key( $user_id ); |
| 57 |
|
| 58 |
$args = array( |
| 59 |
'timeout' => $timeout, |
| 60 |
'limit_response_size' => 1048576, |
| 61 |
'redirection' => 3, |
| 62 |
'user-agent' => "$user_agent; ActivityPub", |
| 63 |
'headers' => array( |
| 64 |
'Accept' => 'application/activity+json', |
| 65 |
'Content-Type' => 'application/activity+json', |
| 66 |
'Date' => \gmdate( 'D, d M Y H:i:s T' ), |
| 67 |
), |
| 68 |
'body' => $body, |
| 69 |
'key_id' => \json_decode( $body )->actor . '#main-key', |
| 70 |
'private_key' => $private_key, |
| 71 |
'user_id' => $user_id, |
| 72 |
); |
| 73 |
|
| 74 |
$response = \wp_safe_remote_post( $url, $args ); |
| 75 |
$code = \wp_remote_retrieve_response_code( $response ); |
| 76 |
|
| 77 |
if ( $code >= 400 ) { |
| 78 |
$response = new \WP_Error( |
| 79 |
$code, |
| 80 |
\__( 'Failed HTTP Request', 'activitypub' ), |
| 81 |
array( |
| 82 |
'status' => $code, |
| 83 |
'response' => $response, |
| 84 |
) |
| 85 |
); |
| 86 |
} |
| 87 |
|
| 88 |
/** |
| 89 |
* Action to save the response of the remote POST request. |
| 90 |
* |
| 91 |
* @param array|\WP_Error $response The response of the remote POST request. |
| 92 |
* @param string $url The URL endpoint. |
| 93 |
* @param string $body The Post Body. |
| 94 |
* @param int|null $user_id The WordPress User-ID, or null when signing with the Application key. |
| 95 |
*/ |
| 96 |
\do_action( 'activitypub_safe_remote_post_response', $response, $url, $body, $user_id ); |
| 97 |
|
| 98 |
return $response; |
| 99 |
} |
| 100 |
|
| 101 |
/** |
| 102 |
* Send a GET Request with the needed HTTP Headers. |
| 103 |
* |
| 104 |
* @param string $url The URL endpoint. |
| 105 |
* @param array $args Optional. Additional arguments to customize the request. |
| 106 |
* - 'headers': Array of headers to override defaults. |
| 107 |
* @param bool|int $cached Optional. Whether to cache the response, or the cache duration in seconds for |
| 108 |
* successful responses. Failed responses use a fixed short backoff duration. Default false. |
| 109 |
* |
| 110 |
* @return array|\WP_Error The GET Response or a WP_Error. |
| 111 |
*/ |
| 112 |
public static function get( $url, $args = array(), $cached = false ) { |
| 113 |
// Backward compatibility: if $args is boolean/int, it's the old $cached parameter. |
| 114 |
if ( ! \is_array( $args ) ) { |
| 115 |
\_deprecated_argument( |
| 116 |
__METHOD__, |
| 117 |
'7.9.0', |
| 118 |
\esc_html__( 'The $cached parameter should now be passed as the third argument.', 'activitypub' ) |
| 119 |
); |
| 120 |
$cached = $args; |
| 121 |
$args = array(); |
| 122 |
} |
| 123 |
|
| 124 |
/** |
| 125 |
* Fires before an HTTP GET request is made. |
| 126 |
* |
| 127 |
* @param string $url The URL endpoint. |
| 128 |
*/ |
| 129 |
\do_action( 'activitypub_pre_http_get', $url ); |
| 130 |
|
| 131 |
$transient_key = self::generate_cache_key( $url ); |
| 132 |
|
| 133 |
// Check cache only if caching is requested. |
| 134 |
if ( $cached ) { |
| 135 |
$response = \get_transient( $transient_key ); |
| 136 |
|
| 137 |
if ( $response ) { |
| 138 |
/** |
| 139 |
* Action to save the response of the remote GET request. |
| 140 |
* |
| 141 |
* @param array|\WP_Error $response The response of the remote GET request. |
| 142 |
* @param string $url The URL endpoint. |
| 143 |
*/ |
| 144 |
\do_action( 'activitypub_safe_remote_get_response', $response, $url ); |
| 145 |
|
| 146 |
return $response; |
| 147 |
} |
| 148 |
} |
| 149 |
|
| 150 |
/** |
| 151 |
* Filters the HTTP headers user agent string. |
| 152 |
* |
| 153 |
* @param string $user_agent The user agent string. |
| 154 |
* @param string $url The request URL. |
| 155 |
*/ |
| 156 |
$user_agent = \apply_filters( 'http_headers_useragent', 'WordPress/' . get_masked_wp_version() . '; ' . \get_bloginfo( 'url' ), $url ); |
| 157 |
|
| 158 |
/** |
| 159 |
* Filters the timeout duration for remote GET requests in ActivityPub. |
| 160 |
* |
| 161 |
* @param int $timeout The timeout value in seconds. Default 10 seconds. |
| 162 |
*/ |
| 163 |
$timeout = \apply_filters( 'activitypub_remote_get_timeout', 10 ); |
| 164 |
|
| 165 |
$defaults = array( |
| 166 |
'timeout' => $timeout, |
| 167 |
'limit_response_size' => 1048576, |
| 168 |
'redirection' => 3, |
| 169 |
'user-agent' => "$user_agent; ActivityPub", |
| 170 |
'headers' => array( |
| 171 |
'Accept' => 'application/activity+json', |
| 172 |
'Content-Type' => 'application/activity+json', |
| 173 |
'Date' => \gmdate( 'D, d M Y H:i:s T' ), |
| 174 |
), |
| 175 |
'key_id' => Application::get_key_id(), |
| 176 |
'private_key' => Application::get_private_key(), |
| 177 |
); |
| 178 |
|
| 179 |
$args = \wp_parse_args( $args, $defaults ); |
| 180 |
$args['headers'] = \wp_parse_args( $args['headers'], $defaults['headers'] ); |
| 181 |
|
| 182 |
$response = \wp_safe_remote_get( $url, $args ); |
| 183 |
$code = \wp_remote_retrieve_response_code( $response ); |
| 184 |
|
| 185 |
if ( \is_wp_error( $response ) || $code >= 400 ) { |
| 186 |
// Capture the served-from URL before $response is replaced by the error below. |
| 187 |
$effective_url = \is_wp_error( $response ) ? '' : self::effective_url( $response ); |
| 188 |
|
| 189 |
if ( ! $code ) { |
| 190 |
$code = 0; |
| 191 |
} |
| 192 |
$response = new \WP_Error( $code, \__( 'Failed HTTP Request', 'activitypub' ), array( 'status' => $code ) ); |
| 193 |
|
| 194 |
/* |
| 195 |
* Cache errors to prevent repeated timeout waits, but never one reached via a |
| 196 |
* cross-host redirect: cached under the requested URL's key, a one-off open |
| 197 |
* redirect on the requested host would let a transient 4xx be replayed as a |
| 198 |
* repeatable federation outage (the same caching concern as the success path |
| 199 |
* below, just with a WP_Error instead of a document). |
| 200 |
* |
| 201 |
* - Retriable errors (timeouts, 5xx): 1 minute (server may recover quickly). |
| 202 |
* - Other errors (4xx): 15 minutes (client errors are more permanent). |
| 203 |
*/ |
| 204 |
if ( $cached && ( ! $effective_url || is_same_host( $url, $effective_url ) ) ) { |
| 205 |
if ( \in_array( $code, ACTIVITYPUB_RETRY_ERROR_CODES, true ) || 0 === $code ) { |
| 206 |
$cache_duration = MINUTE_IN_SECONDS; |
| 207 |
} else { |
| 208 |
$cache_duration = 15 * MINUTE_IN_SECONDS; |
| 209 |
} |
| 210 |
|
| 211 |
\set_transient( $transient_key, $response, $cache_duration ); |
| 212 |
} |
| 213 |
|
| 214 |
return $response; |
| 215 |
} |
| 216 |
|
| 217 |
/** |
| 218 |
* Action to save the response of the remote GET request. |
| 219 |
* |
| 220 |
* @param array|\WP_Error $response The response of the remote GET request. |
| 221 |
* @param string $url The URL endpoint. |
| 222 |
*/ |
| 223 |
\do_action( 'activitypub_safe_remote_get_response', $response, $url ); |
| 224 |
|
| 225 |
/* |
| 226 |
* Never persist a response that was redirected to a different host. The cache |
| 227 |
* is keyed on the requested URL, so caching cross-origin content under that key |
| 228 |
* would let a one-off open redirect on the requested host durably associate the |
| 229 |
* other host's document with that URL — a later lookup (e.g. a public-key fetch) |
| 230 |
* would then return it. Same-host redirects (e.g. http→https) stay cacheable. |
| 231 |
*/ |
| 232 |
$effective_url = self::effective_url( $response ); |
| 233 |
if ( $effective_url && ! is_same_host( $url, $effective_url ) ) { |
| 234 |
return $response; |
| 235 |
} |
| 236 |
|
| 237 |
// Cache successful responses when caching is requested. |
| 238 |
if ( $cached ) { |
| 239 |
$cache_duration = $cached; |
| 240 |
if ( ! \is_int( $cache_duration ) ) { |
| 241 |
$cache_duration = HOUR_IN_SECONDS; |
| 242 |
} |
| 243 |
\set_transient( $transient_key, $response, $cache_duration ); |
| 244 |
} |
| 245 |
|
| 246 |
return $response; |
| 247 |
} |
| 248 |
|
| 249 |
/** |
| 250 |
* Check for URL for Tombstone. |
| 251 |
* |
| 252 |
* @param string $url The URL to check. |
| 253 |
* |
| 254 |
* @return bool True if the URL is a tombstone. |
| 255 |
*/ |
| 256 |
public static function is_tombstone( $url ) { |
| 257 |
\_deprecated_function( __METHOD__, '7.3.0', 'Activitypub\Tombstone::exists_remote' ); |
| 258 |
|
| 259 |
return Tombstone::exists_remote( $url ); |
| 260 |
} |
| 261 |
|
| 262 |
/** |
| 263 |
* Generate a cache key for the URL. |
| 264 |
* |
| 265 |
* @param string $url The URL to generate the cache key for. |
| 266 |
* |
| 267 |
* @return string The cache key. |
| 268 |
*/ |
| 269 |
public static function generate_cache_key( $url ) { |
| 270 |
return 'activitypub_http_' . \md5( $url ); |
| 271 |
} |
| 272 |
|
| 273 |
/** |
| 274 |
* Requests the Data from the Object-URL or Object-Array. |
| 275 |
* |
| 276 |
* Fetched objects are self-confirmed before they are returned, the same way |
| 277 |
* Mastodon's `JsonLdHelper#fetch_resource` works: an object is trusted only when |
| 278 |
* its own `id` is the URL it was actually served from (after any redirects). If |
| 279 |
* the document served at the requested URL declares a different `id`, that id is |
| 280 |
* dereferenced from its own host and accepted only when it self-confirms. This |
| 281 |
* makes every caller safe to cache the result under its `id` — one host can never |
| 282 |
* serve an object (and its public key) under another host's id — without each |
| 283 |
* caller having to re-check the origin itself. |
| 284 |
* |
| 285 |
* @param array|string $url_or_object The Object or the Object URL. |
| 286 |
* @param bool $cached Optional. Whether the result should be cached. Default true. |
| 287 |
* |
| 288 |
* @return array|\WP_Error The Object data as array or WP_Error on failure. |
| 289 |
*/ |
| 290 |
public static function get_remote_object( $url_or_object, $cached = true ) { |
| 291 |
/** |
| 292 |
* Filters the preemptive return value of a remote object request. |
| 293 |
* |
| 294 |
* This is an explicit in-process override (used for caching and tests), not |
| 295 |
* untrusted network data, so it is returned as-is without self-confirmation. |
| 296 |
* |
| 297 |
* @param array|string|null $response The response. |
| 298 |
* @param array|string|null $url_or_object The Object or the Object URL. |
| 299 |
*/ |
| 300 |
$response = \apply_filters( 'activitypub_pre_http_get_remote_object', null, $url_or_object ); |
| 301 |
if ( null !== $response ) { |
| 302 |
return $response; |
| 303 |
} |
| 304 |
|
| 305 |
$url = object_to_uri( $url_or_object ); |
| 306 |
|
| 307 |
if ( Webfinger::is_acct( $url ) ) { |
| 308 |
$url = Webfinger::resolve( $url ); |
| 309 |
} |
| 310 |
|
| 311 |
if ( ! $url ) { |
| 312 |
return new \WP_Error( |
| 313 |
'activitypub_no_valid_actor_identifier', |
| 314 |
\__( 'The "actor" identifier is not valid', 'activitypub' ), |
| 315 |
array( |
| 316 |
'status' => 404, |
| 317 |
'object' => $url, |
| 318 |
) |
| 319 |
); |
| 320 |
} |
| 321 |
|
| 322 |
if ( \is_wp_error( $url ) ) { |
| 323 |
return $url; |
| 324 |
} |
| 325 |
|
| 326 |
$final_url = ''; |
| 327 |
$object = self::fetch_object( $url, $cached, $final_url ); |
| 328 |
|
| 329 |
if ( \is_wp_error( $object ) ) { |
| 330 |
return $object; |
| 331 |
} |
| 332 |
|
| 333 |
// Trust the document when it is served under its own id (after redirects). |
| 334 |
if ( id_matches_url( $object, $final_url ) ) { |
| 335 |
return $object; |
| 336 |
} |
| 337 |
|
| 338 |
$declared_id = isset( $object['id'] ) && \is_string( $object['id'] ) ? $object['id'] : ''; |
| 339 |
|
| 340 |
/* |
| 341 |
* An id-less object cannot be cached under an id, so it cannot be written |
| 342 |
* under another id in an id-keyed cache. Return the document as served. |
| 343 |
*/ |
| 344 |
if ( '' === $declared_id ) { |
| 345 |
return $object; |
| 346 |
} |
| 347 |
|
| 348 |
// Re-fetch the declared id from its own host and require it to self-confirm. One hop only. |
| 349 |
$object = self::fetch_object( $declared_id, $cached, $final_url ); |
| 350 |
|
| 351 |
if ( \is_wp_error( $object ) ) { |
| 352 |
return $object; |
| 353 |
} |
| 354 |
|
| 355 |
if ( ! id_matches_url( $object, $final_url ) ) { |
| 356 |
return new \WP_Error( |
| 357 |
'activitypub_object_id_mismatch', |
| 358 |
\__( 'The object id does not match the URL it was served from', 'activitypub' ), |
| 359 |
array( 'status' => 400 ) |
| 360 |
); |
| 361 |
} |
| 362 |
|
| 363 |
return $object; |
| 364 |
} |
| 365 |
|
| 366 |
/** |
| 367 |
* Fetch and JSON-decode a single remote document. |
| 368 |
* |
| 369 |
* @param string $url The URL to fetch. Must already be resolved (not a WebFinger acct). |
| 370 |
* @param bool $cached Whether the result may be served from and written to cache. |
| 371 |
* @param string $final_url Filled by reference with the URL the document was served from, |
| 372 |
* after following any redirects. |
| 373 |
* |
| 374 |
* @return array|\WP_Error The decoded document, or WP_Error on failure. |
| 375 |
*/ |
| 376 |
private static function fetch_object( $url, $cached, &$final_url ) { |
| 377 |
$final_url = $url; |
| 378 |
|
| 379 |
if ( ! \wp_http_validate_url( $url ) ) { |
| 380 |
return new \WP_Error( |
| 381 |
'activitypub_no_valid_object_url', |
| 382 |
\__( 'The "object" is/has no valid URL', 'activitypub' ), |
| 383 |
array( |
| 384 |
'status' => 400, |
| 385 |
'object' => $url, |
| 386 |
) |
| 387 |
); |
| 388 |
} |
| 389 |
|
| 390 |
$response = self::get( $url, array(), $cached ); |
| 391 |
|
| 392 |
if ( \is_wp_error( $response ) ) { |
| 393 |
return $response; |
| 394 |
} |
| 395 |
|
| 396 |
$effective_url = self::effective_url( $response ); |
| 397 |
if ( $effective_url ) { |
| 398 |
$final_url = $effective_url; |
| 399 |
} |
| 400 |
|
| 401 |
$data = \json_decode( \wp_remote_retrieve_body( $response ), true ); |
| 402 |
|
| 403 |
if ( ! $data ) { |
| 404 |
return new \WP_Error( |
| 405 |
'activitypub_invalid_json', |
| 406 |
\__( 'No valid JSON data', 'activitypub' ), |
| 407 |
array( |
| 408 |
'status' => 400, |
| 409 |
'object' => $url, |
| 410 |
) |
| 411 |
); |
| 412 |
} |
| 413 |
|
| 414 |
return $data; |
| 415 |
} |
| 416 |
|
| 417 |
/** |
| 418 |
* Extract the effective URL a response was served from, after redirects. |
| 419 |
* |
| 420 |
* WordPress follows redirects transparently and exposes the final URL only on |
| 421 |
* the underlying Requests response object. Returns an empty string when the URL |
| 422 |
* cannot be determined, so callers fall back to the URL they requested. |
| 423 |
* |
| 424 |
* SECURITY: the redirect protections that build on this (the self-confirmation in |
| 425 |
* get_remote_object() and the cross-host cache skip in get()) fail OPEN when this |
| 426 |
* returns an empty string — self-confirmation then compares against the requested |
| 427 |
* URL, which a redirect could have bounced away from. This relies on the internal |
| 428 |
* `http_response` → Requests response `url` shape; if a future WordPress release |
| 429 |
* changes it, re-verify that this still returns the post-redirect URL. |
| 430 |
* |
| 431 |
* @param array $response A `wp_remote_get()` response array. |
| 432 |
* |
| 433 |
* @return string The final URL, or an empty string when unavailable. |
| 434 |
*/ |
| 435 |
private static function effective_url( $response ) { |
| 436 |
if ( empty( $response['http_response'] ) || ! \is_object( $response['http_response'] ) ) { |
| 437 |
return ''; |
| 438 |
} |
| 439 |
|
| 440 |
$requests_response = $response['http_response']->get_response_object(); |
| 441 |
|
| 442 |
if ( ! \is_object( $requests_response ) || empty( $requests_response->url ) ) { |
| 443 |
return ''; |
| 444 |
} |
| 445 |
|
| 446 |
return (string) $requests_response->url; |
| 447 |
} |
| 448 |
} |
| 449 |
|
