| 1 |
<?php |
| 2 |
/** |
| 3 |
* WebFinger class file. |
| 4 |
* |
| 5 |
* @package Activitypub |
| 6 |
*/ |
| 7 |
|
| 8 |
namespace Activitypub; |
| 9 |
|
| 10 |
use Activitypub\Activity\Actor; |
| 11 |
use Activitypub\Collection\Actors; |
| 12 |
use Activitypub\Collection\Remote_Actors; |
| 13 |
|
| 14 |
/** |
| 15 |
* ActivityPub WebFinger Class. |
| 16 |
* |
| 17 |
* @author Matthias Pfefferle |
| 18 |
* |
| 19 |
* @see https://webfinger.net/ |
| 20 |
*/ |
| 21 |
class Webfinger { |
| 22 |
/** |
| 23 |
* Check whether a value looks like an `acct` identifier. |
| 24 |
* |
| 25 |
* Accepts any of: |
| 26 |
* |
| 27 |
* - `user@host` — bare WebFinger handle. |
| 28 |
* - `@user@host` — Mastodon display form with a leading `@`. |
| 29 |
* - `acct:user@host` — full RFC 7565 URI form. |
| 30 |
* |
| 31 |
* The host/local-part pattern follows `ACTIVITYPUB_USERNAME_REGEXP`. |
| 32 |
* |
| 33 |
* @since 8.3.0 |
| 34 |
* |
| 35 |
* @param mixed $value The candidate value. |
| 36 |
* @return bool True if the value matches the acct identifier pattern. |
| 37 |
*/ |
| 38 |
public static function is_acct( $value ) { |
| 39 |
if ( ! \is_string( $value ) || '' === $value ) { |
| 40 |
return false; |
| 41 |
} |
| 42 |
|
| 43 |
return (bool) \preg_match( '/^(?:acct:)?@?' . ACTIVITYPUB_USERNAME_REGEXP . '$/i', $value ); |
| 44 |
} |
| 45 |
|
| 46 |
/** |
| 47 |
* Returns a users WebFinger "resource". |
| 48 |
* |
| 49 |
* @param int $user_id The WordPress user id. |
| 50 |
* |
| 51 |
* @return string The user-resource. |
| 52 |
*/ |
| 53 |
public static function get_user_resource( $user_id ) { |
| 54 |
$user = Actors::get_by_id( $user_id ); |
| 55 |
if ( ! $user || \is_wp_error( $user ) ) { |
| 56 |
return ''; |
| 57 |
} |
| 58 |
|
| 59 |
return $user->get_webfinger(); |
| 60 |
} |
| 61 |
|
| 62 |
/** |
| 63 |
* Resolve a WebFinger resource. |
| 64 |
* |
| 65 |
* @param string $uri The WebFinger Resource. |
| 66 |
* |
| 67 |
* @return string|\WP_Error The URL or WP_Error. |
| 68 |
*/ |
| 69 |
public static function resolve( $uri ) { |
| 70 |
$data = self::get_data( $uri ); |
| 71 |
|
| 72 |
if ( \is_wp_error( $data ) ) { |
| 73 |
return $data; |
| 74 |
} |
| 75 |
|
| 76 |
if ( ! \is_array( $data ) || empty( $data['links'] ) || ! \is_array( $data['links'] ) ) { |
| 77 |
return new \WP_Error( |
| 78 |
'webfinger_missing_links', |
| 79 |
\__( 'No valid Link elements found.', 'activitypub' ), |
| 80 |
array( |
| 81 |
'status' => 400, |
| 82 |
'data' => $data, |
| 83 |
) |
| 84 |
); |
| 85 |
} |
| 86 |
|
| 87 |
foreach ( $data['links'] as $link ) { |
| 88 |
if ( |
| 89 |
isset( $link['rel'], $link['href'], $link['type'] ) && |
| 90 |
'self' === $link['rel'] && |
| 91 |
\is_string( $link['href'] ) && |
| 92 |
( |
| 93 |
'application/activity+json' === $link['type'] || |
| 94 |
'application/ld+json; profile="https://www.w3.org/ns/activitystreams"' === $link['type'] |
| 95 |
) |
| 96 |
) { |
| 97 |
return $link['href']; |
| 98 |
} |
| 99 |
} |
| 100 |
|
| 101 |
return new \WP_Error( |
| 102 |
'webfinger_url_no_activitypub', |
| 103 |
\__( 'The Site supports WebFinger but not ActivityPub', 'activitypub' ), |
| 104 |
array( |
| 105 |
'status' => 400, |
| 106 |
'data' => $data, |
| 107 |
) |
| 108 |
); |
| 109 |
} |
| 110 |
|
| 111 |
/** |
| 112 |
* Transform a URI to an acct <identifier>@<host>. |
| 113 |
* |
| 114 |
* @see https://swicg.github.io/activitypub-webfinger/#reverse-discovery |
| 115 |
* |
| 116 |
* @param string $uri The URI (acct:, mailto:, http:, https:). |
| 117 |
* |
| 118 |
* @return string|\WP_Error Error or acct URI. |
| 119 |
*/ |
| 120 |
public static function uri_to_acct( $uri ) { |
| 121 |
$data = self::get_data( $uri ); |
| 122 |
|
| 123 |
if ( \is_wp_error( $data ) ) { |
| 124 |
return $data; |
| 125 |
} |
| 126 |
|
| 127 |
// Check if subject is an acct URI. |
| 128 |
if ( |
| 129 |
isset( $data['subject'] ) && |
| 130 |
\is_string( $data['subject'] ) && |
| 131 |
\str_starts_with( $data['subject'], 'acct:' ) |
| 132 |
) { |
| 133 |
return $data['subject']; |
| 134 |
} |
| 135 |
|
| 136 |
// Search for an acct URI in the aliases. |
| 137 |
if ( isset( $data['aliases'] ) && \is_array( $data['aliases'] ) ) { |
| 138 |
foreach ( $data['aliases'] as $alias ) { |
| 139 |
if ( \is_string( $alias ) && \str_starts_with( $alias, 'acct:' ) ) { |
| 140 |
return $alias; |
| 141 |
} |
| 142 |
} |
| 143 |
} |
| 144 |
|
| 145 |
return new \WP_Error( |
| 146 |
'webfinger_url_no_acct', |
| 147 |
\__( 'No acct URI found.', 'activitypub' ), |
| 148 |
array( |
| 149 |
'status' => 400, |
| 150 |
'data' => $data, |
| 151 |
) |
| 152 |
); |
| 153 |
} |
| 154 |
|
| 155 |
/** |
| 156 |
* Convert a URI string to an identifier and its host. |
| 157 |
* Automatically adds acct: if it's missing. |
| 158 |
* |
| 159 |
* @param string $url The URI (acct:, mailto:, http:, https:). |
| 160 |
* |
| 161 |
* @return \WP_Error|array Error reaction or array with identifier and host as values. |
| 162 |
*/ |
| 163 |
public static function get_identifier_and_host( $url ) { |
| 164 |
if ( ! $url ) { |
| 165 |
return new \WP_Error( |
| 166 |
'webfinger_invalid_identifier', |
| 167 |
\__( 'Invalid Identifier', 'activitypub' ), |
| 168 |
array( |
| 169 |
'status' => 400, |
| 170 |
'data' => $url, |
| 171 |
) |
| 172 |
); |
| 173 |
} |
| 174 |
|
| 175 |
// Remove leading @. |
| 176 |
$url = \ltrim( $url, '@' ); |
| 177 |
|
| 178 |
if ( \str_starts_with( $url, '//' ) ) { |
| 179 |
/* |
| 180 |
* A scheme-relative URL is a URL reference, not a handle: treating it as one would |
| 181 |
* read the host off the first `@` in its path. |
| 182 |
*/ |
| 183 |
$identifier = $url; |
| 184 |
$scheme = ''; |
| 185 |
} elseif ( |
| 186 |
// Scheme grammar per RFC 3986: a digit or hyphen in it must not read as a handle. |
| 187 |
! \preg_match( '/^([a-zA-Z][a-zA-Z0-9+.\-]*):/', $url, $match ) |
| 188 |
) { |
| 189 |
$identifier = 'acct:' . $url; |
| 190 |
$scheme = 'acct'; |
| 191 |
} else { |
| 192 |
$identifier = $url; |
| 193 |
|
| 194 |
// Schemes are case-insensitive, and the switch below compares them exactly. |
| 195 |
$scheme = \strtolower( $match[1] ); |
| 196 |
} |
| 197 |
|
| 198 |
$host = null; |
| 199 |
|
| 200 |
switch ( $scheme ) { |
| 201 |
case 'acct': |
| 202 |
case 'mailto': |
| 203 |
case 'xmpp': |
| 204 |
// Split on the last `@`: a local part may contain one, the host may not. |
| 205 |
if ( \strrpos( $identifier, '@' ) !== false ) { |
| 206 |
$host = \substr( $identifier, \strrpos( $identifier, '@' ) + 1 ); |
| 207 |
|
| 208 |
/* |
| 209 |
* Cut anything a query or fragment starts. `wp_parse_url()` does this for the |
| 210 |
* URL forms below, but this branch takes the host off the string itself, so |
| 211 |
* `acct:user@example.com#x` would otherwise carry `#x` into the host and miss |
| 212 |
* a domain block on `example.com`. |
| 213 |
*/ |
| 214 |
$host = \substr( $host, 0, \strcspn( $host, '?#' ) ); |
| 215 |
} |
| 216 |
break; |
| 217 |
default: |
| 218 |
$host = \wp_parse_url( $identifier, PHP_URL_HOST ); |
| 219 |
break; |
| 220 |
} |
| 221 |
|
| 222 |
if ( empty( $host ) ) { |
| 223 |
return new \WP_Error( |
| 224 |
'webfinger_invalid_identifier', |
| 225 |
\__( 'Invalid Identifier', 'activitypub' ), |
| 226 |
array( |
| 227 |
'status' => 400, |
| 228 |
'data' => $url, |
| 229 |
) |
| 230 |
); |
| 231 |
} |
| 232 |
|
| 233 |
return array( $identifier, \strtolower( $host ) ); |
| 234 |
} |
| 235 |
|
| 236 |
/** |
| 237 |
* Get the host of an identifier. |
| 238 |
* |
| 239 |
* The host half of {@see self::get_identifier_and_host()}, for callers that only need that. |
| 240 |
* Prefer this over parsing a host out directly when you need an identifier's host: a handle |
| 241 |
* has none to parse, and a URI in a scheme other than `http` can carry an `@` in its path |
| 242 |
* that would be read as one. |
| 243 |
* |
| 244 |
* Not a drop-in for `is_same_host()`, which deliberately fails closed on an identifier with |
| 245 |
* no parsable host. Giving an `acct:` keyId a host there would re-admit what 9.2.1 rejected. |
| 246 |
* |
| 247 |
* @since 9.3.0 |
| 248 |
* |
| 249 |
* @param string $uri The identifier, a URL or a handle. |
| 250 |
* |
| 251 |
* @return string The host, lowercased, or an empty string when there is none. |
| 252 |
*/ |
| 253 |
public static function get_host( $uri ) { |
| 254 |
$identifier_and_host = self::get_identifier_and_host( (string) $uri ); |
| 255 |
|
| 256 |
if ( \is_wp_error( $identifier_and_host ) ) { |
| 257 |
return ''; |
| 258 |
} |
| 259 |
|
| 260 |
/* |
| 261 |
* Folded here rather than in `get_identifier_and_host()`, whose host is also used to build |
| 262 |
* WebFinger and intent URLs: an IPv6 authority needs its brackets to stay a valid URL. |
| 263 |
*/ |
| 264 |
return fold_host( $identifier_and_host[1] ); |
| 265 |
} |
| 266 |
|
| 267 |
/** |
| 268 |
* Get the WebFinger data for a given URI. |
| 269 |
* |
| 270 |
* @param string $uri The Identifier: <identifier>@<host> or URI. |
| 271 |
* |
| 272 |
* @return \WP_Error|mixed Error reaction, or the decoded document. The remote server picks the |
| 273 |
* body, so callers check the shape before indexing it. |
| 274 |
*/ |
| 275 |
public static function get_data( $uri ) { |
| 276 |
$identifier_and_host = self::get_identifier_and_host( $uri ); |
| 277 |
|
| 278 |
if ( \is_wp_error( $identifier_and_host ) ) { |
| 279 |
return $identifier_and_host; |
| 280 |
} |
| 281 |
|
| 282 |
list( $identifier, $host ) = $identifier_and_host; |
| 283 |
|
| 284 |
$webfinger_url = \sprintf( |
| 285 |
'https://%s/.well-known/webfinger?resource=%s', |
| 286 |
$host, |
| 287 |
\rawurlencode( $identifier ) |
| 288 |
); |
| 289 |
|
| 290 |
// Use Http::get() which handles all caching (success and errors). |
| 291 |
$response = Http::get( |
| 292 |
$webfinger_url, |
| 293 |
array( 'headers' => array( 'Accept' => 'application/jrd+json' ) ), |
| 294 |
WEEK_IN_SECONDS |
| 295 |
); |
| 296 |
|
| 297 |
if ( \is_wp_error( $response ) ) { |
| 298 |
return $response; |
| 299 |
} |
| 300 |
|
| 301 |
$body = \wp_remote_retrieve_body( $response ); |
| 302 |
|
| 303 |
return \json_decode( $body, true ); |
| 304 |
} |
| 305 |
|
| 306 |
/** |
| 307 |
* Get the Remote-Follow endpoint for a given URI. |
| 308 |
* |
| 309 |
* @param string $uri The WebFinger Resource URI. |
| 310 |
* |
| 311 |
* @return string|\WP_Error Error or the Remote-Follow endpoint URI. |
| 312 |
*/ |
| 313 |
public static function get_remote_follow_endpoint( $uri ) { |
| 314 |
return self::get_intent_endpoint( $uri, 'follow', true ); |
| 315 |
} |
| 316 |
|
| 317 |
/** |
| 318 |
* Generate a cache key for a given URI. |
| 319 |
* |
| 320 |
* @param string $uri A WebFinger Resource URI. |
| 321 |
* |
| 322 |
* @return string The cache key. |
| 323 |
*/ |
| 324 |
public static function generate_cache_key( $uri ) { |
| 325 |
$uri = \ltrim( $uri, '@' ); |
| 326 |
|
| 327 |
if ( \filter_var( $uri, FILTER_VALIDATE_EMAIL ) ) { |
| 328 |
$uri = 'acct:' . $uri; |
| 329 |
} |
| 330 |
|
| 331 |
return 'webfinger_' . \md5( $uri ); |
| 332 |
} |
| 333 |
|
| 334 |
/** |
| 335 |
* Infer a shortname from the Actor ID or URL. Used only for fallbacks, |
| 336 |
* we will try to use what's supplied. |
| 337 |
* |
| 338 |
* @param Actor|string $actor_or_uri The Actor or URI. |
| 339 |
* |
| 340 |
* @return string Hopefully the name of the Follower. |
| 341 |
*/ |
| 342 |
public static function guess( $actor_or_uri ) { |
| 343 |
if ( ! $actor_or_uri instanceof Actor ) { |
| 344 |
$actor = Remote_Actors::fetch_by_uri( $actor_or_uri ); |
| 345 |
if ( \is_wp_error( $actor ) ) { |
| 346 |
return extract_name_from_uri( $actor_or_uri ) . '@' . \wp_parse_url( $actor_or_uri, PHP_URL_HOST ); |
| 347 |
} |
| 348 |
|
| 349 |
$actor_or_uri = $actor; |
| 350 |
} |
| 351 |
|
| 352 |
if ( $actor_or_uri->get_preferred_username() ) { |
| 353 |
return $actor_or_uri->get_preferred_username() . '@' . \wp_parse_url( $actor_or_uri->get_id(), PHP_URL_HOST ); |
| 354 |
} |
| 355 |
|
| 356 |
return extract_name_from_uri( $actor_or_uri->get_id() ) . '@' . \wp_parse_url( $actor_or_uri->get_id(), PHP_URL_HOST ); |
| 357 |
} |
| 358 |
|
| 359 |
/** |
| 360 |
* Get the Intent endpoint for a given URI and intent. |
| 361 |
* |
| 362 |
* @since 8.0.0 |
| 363 |
* |
| 364 |
* @see https://codeberg.org/fediverse/fep/src/branch/main/fep/3b86/fep-3b86.md |
| 365 |
* |
| 366 |
* @param string $uri The WebFinger Resource URI. |
| 367 |
* @param string $intent The intent to look for. |
| 368 |
* @param bool $fallback Whether to fallback to the Remote-Follow endpoint. |
| 369 |
* |
| 370 |
* @return string|\WP_Error Error or the Intent endpoint URI (may contain `{uri}` placeholder). |
| 371 |
*/ |
| 372 |
public static function get_intent_endpoint( $uri, $intent, $fallback = false ) { |
| 373 |
$data = self::get_data( $uri ); |
| 374 |
|
| 375 |
if ( \is_wp_error( $data ) ) { |
| 376 |
return $data; |
| 377 |
} |
| 378 |
|
| 379 |
if ( empty( $data['links'] ) || ! \is_array( $data['links'] ) ) { |
| 380 |
return new \WP_Error( |
| 381 |
'webfinger_missing_links', |
| 382 |
\__( 'No valid Link elements found.', 'activitypub' ), |
| 383 |
array( |
| 384 |
'status' => 400, |
| 385 |
'data' => $data, |
| 386 |
) |
| 387 |
); |
| 388 |
} |
| 389 |
|
| 390 |
// Normalize the links with $rel as key. |
| 391 |
$links = array(); |
| 392 |
|
| 393 |
foreach ( $data['links'] as $link ) { |
| 394 |
if ( ! isset( $link['rel'], $link['template'] ) || ! \is_string( $link['rel'] ) || ! \is_string( $link['template'] ) ) { |
| 395 |
continue; |
| 396 |
} |
| 397 |
|
| 398 |
$template = \trim( $link['template'] ); |
| 399 |
|
| 400 |
/* |
| 401 |
* A scheme check alone would pass `//host`, `/path`, `https:///path` and the empty |
| 402 |
* string: with no colon there is nothing for `wp_kses_bad_protocol()` to strip, so the |
| 403 |
* comparison below always matches. An accepted junk template is also stored under its |
| 404 |
* rel, which stops the OStatus and FEP-3b86 fallbacks from ever being reached. |
| 405 |
*/ |
| 406 |
if ( ! get_url_authority( $template ) ) { |
| 407 |
continue; |
| 408 |
} |
| 409 |
|
| 410 |
// The list is explicit: the `wp_allowed_protocols()` default is wider and filterable. |
| 411 |
$allowed = \wp_kses_bad_protocol( $template, array( 'http', 'https' ) ); |
| 412 |
|
| 413 |
// Lowercased both sides like `wp_http_validate_url()`, or `HTTPS://` reads as a rewrite. |
| 414 |
if ( \strtolower( $allowed ) !== \strtolower( $template ) ) { |
| 415 |
continue; |
| 416 |
} |
| 417 |
|
| 418 |
$links[ \strtolower( $link['rel'] ) ] = $template; |
| 419 |
} |
| 420 |
|
| 421 |
$intent = \sanitize_text_field( $intent ); |
| 422 |
$intent = \strtolower( $intent ); |
| 423 |
|
| 424 |
if ( ! \filter_var( $intent, FILTER_VALIDATE_URL ) ) { |
| 425 |
$intent = 'https://w3id.org/fep/3b86/' . $intent; |
| 426 |
} |
| 427 |
|
| 428 |
if ( isset( $links[ $intent ] ) ) { |
| 429 |
return $links[ $intent ]; |
| 430 |
} |
| 431 |
|
| 432 |
if ( ! $fallback ) { |
| 433 |
return new \WP_Error( |
| 434 |
'webfinger_missing_intent_endpoint', |
| 435 |
\__( 'No valid Intent endpoint found.', 'activitypub' ), |
| 436 |
array( |
| 437 |
'status' => 400, |
| 438 |
'data' => $data, |
| 439 |
) |
| 440 |
); |
| 441 |
} |
| 442 |
|
| 443 |
/* |
| 444 |
* OStatus subscribe URL (deprecated but still widely supported) |
| 445 |
* |
| 446 |
* @see https://ostatus.github.io/spec/OStatus%201.0%20Draft%202.html#anchor10 |
| 447 |
*/ |
| 448 |
if ( isset( $links['http://ostatus.org/schema/1.0/subscribe'] ) ) { |
| 449 |
return $links['http://ostatus.org/schema/1.0/subscribe']; |
| 450 |
} |
| 451 |
|
| 452 |
/* |
| 453 |
* FEP-3b86 Object Intent — the generic "open this object on my home |
| 454 |
* server" link, equivalent to pasting the URL into the home server's |
| 455 |
* search box. Useful when no verb-specific intent is advertised. |
| 456 |
* |
| 457 |
* @see https://codeberg.org/fediverse/fep/src/branch/main/fep/3b86/fep-3b86.md#5-1-object-intent |
| 458 |
*/ |
| 459 |
if ( isset( $links['https://w3id.org/fep/3b86/object'] ) ) { |
| 460 |
return $links['https://w3id.org/fep/3b86/object']; |
| 461 |
} |
| 462 |
|
| 463 |
// Last-resort: construct a Mastodon-compatible authorize_interaction URL. |
| 464 |
$identifier_and_host = self::get_identifier_and_host( $uri ); |
| 465 |
|
| 466 |
if ( \is_wp_error( $identifier_and_host ) ) { |
| 467 |
return new \WP_Error( |
| 468 |
'webfinger_missing_intent_endpoint', |
| 469 |
\__( 'No valid Intent endpoint found.', 'activitypub' ), |
| 470 |
array( |
| 471 |
'status' => 400, |
| 472 |
'data' => $data, |
| 473 |
) |
| 474 |
); |
| 475 |
} |
| 476 |
|
| 477 |
return 'https://' . $identifier_and_host[1] . '/authorize_interaction?uri={uri}'; |
| 478 |
} |
| 479 |
} |
| 480 |
|
