| 1 | use std::{ |
| 2 | borrow::Cow, |
| 3 | collections::{HashMap, VecDeque}, |
| 4 | io::Read as _, |
| 5 | panic::AssertUnwindSafe, |
| 6 | sync::Arc, |
| 7 | time::Duration, |
| 8 | }; |
| 9 | |
| 10 | use futures_util::{ |
| 11 | FutureExt, StreamExt, |
| 12 | future::{Shared, WeakShared}, |
| 13 | }; |
| 14 | use matrix_sdk::{ |
| 15 | Client, Room, SendOutsideWasm, SyncOutsideWasm, |
| 16 | media::{MediaFormat, MediaRequestParameters, MediaThumbnailSettings}, |
| 17 | room::reply::{EnforceThread, Reply as SdkReply}, |
| 18 | ruma::{ |
| 19 | api::{ |
| 20 | Metadata, OutgoingRequest, |
| 21 | auth_scheme::SendAccessToken, |
| 22 | client::{authenticated_media, media}, |
| 23 | }, |
| 24 | events::room::{ |
| 25 | EncryptedFileHash, EncryptedFileHashAlgorithm, EncryptedFileInfo, MediaSource, |
| 26 | message::{ |
| 27 | AddMentions, Relation, ReplyWithinThread, RoomMessageEventContentWithoutRelation, |
| 28 | }, |
| 29 | }, |
| 30 | }, |
| 31 | }; |
| 32 | use matrix_sdk_crypto::{AttachmentDecryptor, MediaEncryptionInfo}; |
| 33 | use ruma::{MxcUri, OwnedEventId}; |
| 34 | use tokio::sync::Mutex; |
| 35 | |
| 36 | use super::{executor::EngineBoxFuture, url_preview::UrlPreviews}; |
| 37 | use crate::{ |
| 38 | AttachmentEncryption, AttachmentUpload, Error, ErrorKind, LinkPreview, Media, MediaFile, |
| 39 | MediaId, MediaKind, MediaLocation, MediaSize, |
| 40 | normalise::{MediaRegistration, MediaScope, MediaSourceInfo}, |
| 41 | upload::prepare_upload, |
| 42 | }; |
| 43 | |
| 44 | /// a stalled download gives up after this long without a byte |
| 45 | const DOWNLOAD_IDLE: Duration = Duration::from_secs(60); |
| 46 | /// the sdk client gives up on any request after thirty seconds, far too soon for a video |
| 47 | const DOWNLOAD_TOTAL: Duration = Duration::from_secs(60 * 60); |
| 48 | |
| 49 | type FlightFuture = EngineBoxFuture<'static, Result<Media, Error>>; |
| 50 | type Flight = Shared<FlightFuture>; |
| 51 | |
| 52 | /// owns everything needed to turn a media id or an mxc uri into bytes, so the |
| 53 | /// handle can answer media queries without a trip through the engine |
| 54 | pub(crate) struct MediaService { |
| 55 | client: Client, |
| 56 | registry: Mutex<MediaRegistry>, |
| 57 | flights: Mutex<HashMap<String, WeakShared<FlightFuture>>>, |
| 58 | max_bytes: usize, |
| 59 | context_limit: usize, |
| 60 | url_previews: UrlPreviews, |
| 61 | } |
| 62 | |
| 63 | impl MediaService { |
| 64 | pub(crate) fn new(client: Client, max_bytes: usize, context_limit: usize) -> Self { |
| 65 | Self { |
| 66 | url_previews: UrlPreviews::new(client.clone()), |
| 67 | client, |
| 68 | registry: Mutex::new(MediaRegistry::default()), |
| 69 | flights: Mutex::new(HashMap::new()), |
| 70 | max_bytes, |
| 71 | context_limit, |
| 72 | } |
| 73 | } |
| 74 | |
| 75 | /// one event context session's media, bounded by `EngineConfig::max_context_media` |
| 76 | pub(crate) fn context_scope(&self) -> MediaScope { |
| 77 | MediaScope::bounded(self.context_limit) |
| 78 | } |
| 79 | |
| 80 | pub(crate) async fn url_preview(&self, url: &str) -> Result<Option<LinkPreview>, Error> { |
| 81 | self.url_previews.preview(url).await |
| 82 | } |
| 83 | |
| 84 | pub(crate) async fn register(&self, scope: &MediaScope, media: Vec<MediaRegistration>) { |
| 85 | if !media.is_empty() { |
| 86 | self.registry.lock().await.insert_all(scope, media); |
| 87 | } |
| 88 | } |
| 89 | |
| 90 | pub(crate) async fn release(&self, scope: &MediaScope) { |
| 91 | self.registry.lock().await.release(scope); |
| 92 | } |
| 93 | |
| 94 | pub(crate) async fn clear(&self) { |
| 95 | self.registry.lock().await.clear(); |
| 96 | } |
| 97 | |
| 98 | pub(crate) async fn timeline_media( |
| 99 | &self, |
| 100 | media_id: &MediaId, |
| 101 | size: MediaSize, |
| 102 | ) -> Result<Media, Error> { |
| 103 | let chosen = choose_source(self.registered(media_id).await?, size)?; |
| 104 | if chosen |
| 105 | .declared_size |
| 106 | .is_some_and(|size| size > self.max_bytes as u64) |
| 107 | { |
| 108 | return Err(too_large("media exceeds the configured byte limit")); |
| 109 | } |
| 110 | self.fetch_or_file(chosen.request, chosen.mimetype).await |
| 111 | } |
| 112 | |
| 113 | pub(crate) async fn media_location( |
| 114 | &self, |
| 115 | media_id: &MediaId, |
| 116 | size: MediaSize, |
| 117 | ) -> Result<MediaLocation, Error> { |
| 118 | let media = self.registered(media_id).await?; |
| 119 | let chosen = choose_source(media.clone(), size)?; |
| 120 | if is_local(&chosen.request.source) { |
| 121 | return Ok(MediaLocation::Local); |
| 122 | } |
| 123 | if !matches!(chosen.request.format, MediaFormat::Thumbnail(_)) { |
| 124 | return Ok(MediaLocation::File( |
| 125 | media_file( |
| 126 | &self.client, |
| 127 | chosen.request, |
| 128 | chosen.mimetype, |
| 129 | chosen.declared_size, |
| 130 | ) |
| 131 | .await?, |
| 132 | )); |
| 133 | } |
| 134 | let url = media_http_request(&self.client, &chosen.request) |
| 135 | .await? |
| 136 | .uri() |
| 137 | .to_string(); |
| 138 | let fallback = MediaRequestParameters { |
| 139 | source: media.full.source, |
| 140 | format: MediaFormat::File, |
| 141 | }; |
| 142 | Ok(MediaLocation::ServerThumbnail { |
| 143 | url, |
| 144 | fallback: media_file( |
| 145 | &self.client, |
| 146 | fallback, |
| 147 | media.full.mimetype, |
| 148 | media.full.declared_size, |
| 149 | ) |
| 150 | .await?, |
| 151 | }) |
| 152 | } |
| 153 | |
| 154 | async fn registered(&self, media_id: &MediaId) -> Result<MediaRegistration, Error> { |
| 155 | if let Some(uri) = media_id.as_plain() { |
| 156 | return plain_registration(media_id, uri); |
| 157 | } |
| 158 | self.registry |
| 159 | .lock() |
| 160 | .await |
| 161 | .get(media_id) |
| 162 | .cloned() |
| 163 | .ok_or_else(|| { |
| 164 | Error::new( |
| 165 | ErrorKind::MediaUnknown, |
| 166 | "media is not registered in a retained timeline", |
| 167 | ) |
| 168 | }) |
| 169 | } |
| 170 | |
| 171 | /// plenty of servers won't thumbnail everything, and the file itself is the next best |
| 172 | /// thing. the byte cap still holds either way |
| 173 | async fn fetch_or_file( |
| 174 | &self, |
| 175 | request: MediaRequestParameters, |
| 176 | mimetype: Option<String>, |
| 177 | ) -> Result<Media, Error> { |
| 178 | if !matches!(request.format, MediaFormat::Thumbnail(_)) { |
| 179 | return self.fetch(request, mimetype).await; |
| 180 | } |
| 181 | match self.fetch(request.clone(), mimetype.clone()).await { |
| 182 | Err(error) if error.kind == ErrorKind::MediaUnavailable => { |
| 183 | let file = MediaRequestParameters { |
| 184 | source: request.source, |
| 185 | format: MediaFormat::File, |
| 186 | }; |
| 187 | self.fetch(file, mimetype).await |
| 188 | } |
| 189 | result => result, |
| 190 | } |
| 191 | } |
| 192 | |
| 193 | async fn fetch( |
| 194 | &self, |
| 195 | request: MediaRequestParameters, |
| 196 | mimetype: Option<String>, |
| 197 | ) -> Result<Media, Error> { |
| 198 | let key = flight_key(&request); |
| 199 | let flight = { |
| 200 | let mut flights = self.flights.lock().await; |
| 201 | flights.retain(|_, flight| flight.upgrade().is_some()); |
| 202 | match flights.get(&key).and_then(WeakShared::upgrade) { |
| 203 | Some(flight) => flight, |
| 204 | None => { |
| 205 | let prepared = AssertUnwindSafe(fetch( |
| 206 | self.client.clone(), |
| 207 | request, |
| 208 | mimetype, |
| 209 | self.max_bytes, |
| 210 | )) |
| 211 | .catch_unwind() |
| 212 | .map(|result| { |
| 213 | result.unwrap_or_else(|_| { |
| 214 | Err(Error::new(ErrorKind::Other, "media fetch panicked")) |
| 215 | }) |
| 216 | }); |
| 217 | let boxed: FlightFuture = Box::pin(prepared); |
| 218 | let flight: Flight = boxed.shared(); |
| 219 | if let Some(weak) = flight.downgrade() { |
| 220 | flights.insert(key, weak); |
| 221 | } |
| 222 | flight |
| 223 | } |
| 224 | } |
| 225 | }; |
| 226 | flight.await |
| 227 | } |
| 228 | } |
| 229 | |
| 230 | struct ChosenSource { |
| 231 | request: MediaRequestParameters, |
| 232 | mimetype: Option<String>, |
| 233 | declared_size: Option<u64>, |
| 234 | } |
| 235 | |
| 236 | fn plain_registration(media_id: &MediaId, uri: &MxcUri) -> Result<MediaRegistration, Error> { |
| 237 | if !uri.is_valid() { |
| 238 | return Err(Error::new(ErrorKind::Invalid, "mxc uri is malformed")); |
| 239 | } |
| 240 | Ok(MediaRegistration { |
| 241 | id: media_id.clone(), |
| 242 | kind: MediaKind::Image, |
| 243 | full: MediaSourceInfo { |
| 244 | source: MediaSource::Plain(uri.to_owned()), |
| 245 | mimetype: None, |
| 246 | declared_size: None, |
| 247 | }, |
| 248 | thumbnail: None, |
| 249 | }) |
| 250 | } |
| 251 | |
| 252 | fn choose_source(media: MediaRegistration, size: MediaSize) -> Result<ChosenSource, Error> { |
| 253 | let (source, format, mimetype, declared_size) = match size { |
| 254 | MediaSize::Full => ( |
| 255 | media.full.source, |
| 256 | MediaFormat::File, |
| 257 | media.full.mimetype, |
| 258 | media.full.declared_size, |
| 259 | ), |
| 260 | MediaSize::Thumbnail { width, height } => match media.thumbnail { |
| 261 | Some(thumbnail) => ( |
| 262 | thumbnail.source, |
| 263 | MediaFormat::File, |
| 264 | thumbnail.mimetype, |
| 265 | thumbnail.declared_size, |
| 266 | ), |
| 267 | // the server can only resize what it can read |
| 268 | None if matches!(media.kind, MediaKind::Image | MediaKind::Video) |
| 269 | && matches!(media.full.source, MediaSource::Plain(_)) => |
| 270 | { |
| 271 | // the server picks the thumbnail's type, so the original's would be a lie |
| 272 | let format = if is_local(&media.full.source) { |
| 273 | MediaFormat::File |
| 274 | } else { |
| 275 | thumbnail_format(width, height) |
| 276 | }; |
| 277 | (media.full.source, format, None, None) |
| 278 | } |
| 279 | None => { |
| 280 | return Err(Error::new( |
| 281 | ErrorKind::MediaUnavailable, |
| 282 | "the requested media size is unavailable", |
| 283 | )); |
| 284 | } |
| 285 | }, |
| 286 | }; |
| 287 | Ok(ChosenSource { |
| 288 | request: MediaRequestParameters { source, format }, |
| 289 | mimetype, |
| 290 | declared_size, |
| 291 | }) |
| 292 | } |
| 293 | |
| 294 | async fn media_file( |
| 295 | client: &Client, |
| 296 | request: MediaRequestParameters, |
| 297 | mimetype: Option<String>, |
| 298 | declared_size: Option<u64>, |
| 299 | ) -> Result<MediaFile, Error> { |
| 300 | let url = media_http_request(client, &request) |
| 301 | .await? |
| 302 | .uri() |
| 303 | .to_string(); |
| 304 | let encryption = match &request.source { |
| 305 | MediaSource::Plain(_) => None, |
| 306 | MediaSource::Encrypted(file) => { |
| 307 | let (EncryptedFileInfo::V2(info), Some(EncryptedFileHash::Sha256(sha256))) = ( |
| 308 | &file.info, |
| 309 | file.hashes.get(&EncryptedFileHashAlgorithm::Sha256), |
| 310 | ) else { |
| 311 | return Err(Error::new( |
| 312 | ErrorKind::MediaUnavailable, |
| 313 | "encrypted media is not v2 with a sha256 hash", |
| 314 | )); |
| 315 | }; |
| 316 | Some(AttachmentEncryption { |
| 317 | key: info.k.encode(), |
| 318 | iv: info.iv.encode(), |
| 319 | sha256: sha256.encode(), |
| 320 | }) |
| 321 | } |
| 322 | }; |
| 323 | Ok(MediaFile { |
| 324 | url, |
| 325 | mimetype, |
| 326 | declared_size, |
| 327 | encryption, |
| 328 | }) |
| 329 | } |
| 330 | |
| 331 | /// servers without spec v1.11 animated thumbnails ignore the flag and send a still frame |
| 332 | fn thumbnail_format(width: u32, height: u32) -> MediaFormat { |
| 333 | let mut settings = MediaThumbnailSettings::new(width.into(), height.into()); |
| 334 | settings.animated = true; |
| 335 | MediaFormat::Thumbnail(settings) |
| 336 | } |
| 337 | |
| 338 | fn is_local(source: &MediaSource) -> bool { |
| 339 | let uri = match source { |
| 340 | MediaSource::Plain(uri) => uri, |
| 341 | MediaSource::Encrypted(file) => &file.url, |
| 342 | }; |
| 343 | uri.server_name() |
| 344 | .is_ok_and(|server| server.as_str() == "send-queue.localhost") |
| 345 | } |
| 346 | |
| 347 | /// plain and encrypted sources share a cache key in the sdk, but must never share a flight |
| 348 | fn flight_key(request: &MediaRequestParameters) -> String { |
| 349 | let uri = match &request.source { |
| 350 | MediaSource::Plain(uri) => format!("plain|{uri}"), |
| 351 | MediaSource::Encrypted(file) => format!("encrypted|{}", file.url), |
| 352 | }; |
| 353 | match &request.format { |
| 354 | MediaFormat::File => uri.to_owned(), |
| 355 | MediaFormat::Thumbnail(settings) => format!( |
| 356 | "{uri}|{}|{}x{}|{}", |
| 357 | settings.method, settings.width, settings.height, settings.animated |
| 358 | ), |
| 359 | } |
| 360 | } |
| 361 | |
| 362 | fn too_large(message: &'static str) -> Error { |
| 363 | Error::new(ErrorKind::MediaTooLarge, message) |
| 364 | } |
| 365 | |
| 366 | pub(crate) async fn download_file( |
| 367 | client: &Client, |
| 368 | source: MediaSource, |
| 369 | mimetype: Option<String>, |
| 370 | max_bytes: usize, |
| 371 | ) -> Result<Media, Error> { |
| 372 | let request = MediaRequestParameters { |
| 373 | source, |
| 374 | format: MediaFormat::File, |
| 375 | }; |
| 376 | fetch(client.clone(), request, mimetype, max_bytes).await |
| 377 | } |
| 378 | |
| 379 | async fn fetch( |
| 380 | client: Client, |
| 381 | request: MediaRequestParameters, |
| 382 | mimetype: Option<String>, |
| 383 | max_bytes: usize, |
| 384 | ) -> Result<Media, Error> { |
| 385 | if let Some(cached) = cached_media(&client, &request).await? { |
| 386 | if cached.len() > max_bytes { |
| 387 | if let Ok(store) = client.media_store().lock().await { |
| 388 | let _ = store.remove_media_content(&request).await; |
| 389 | } |
| 390 | return Err(too_large("cached media exceeds the configured byte limit")); |
| 391 | } |
| 392 | // plain and encrypted share a cache key, so a plain fetch of an encrypted file's uri |
| 393 | // leaves ciphertext here. its hash gives it away |
| 394 | if is_ciphertext(&cached, &request.source) { |
| 395 | let bytes = maybe_decrypt(cached, &request, max_bytes)?; |
| 396 | cache(&client, &request, &bytes).await; |
| 397 | return Ok(media(bytes, mimetype)); |
| 398 | } |
| 399 | return Ok(media(cached, mimetype)); |
| 400 | } |
| 401 | let (bytes, served_mimetype) = download_bounded(&client, &request, max_bytes).await?; |
| 402 | // the sdk caches plaintext under the encrypted key, so we do too or nobody can read anything |
| 403 | let bytes = maybe_decrypt(bytes, &request, max_bytes)?; |
| 404 | cache(&client, &request, &bytes).await; |
| 405 | Ok(media(bytes, mimetype.or(served_mimetype))) |
| 406 | } |
| 407 | |
| 408 | async fn cache(client: &Client, request: &MediaRequestParameters, bytes: &[u8]) { |
| 409 | if let Ok(store) = client.media_store().lock().await { |
| 410 | let _ = store |
| 411 | .add_media_content( |
| 412 | request, |
| 413 | bytes.to_vec(), |
| 414 | matrix_sdk::media::store::IgnoreMediaRetentionPolicy::No, |
| 415 | ) |
| 416 | .await; |
| 417 | } |
| 418 | } |
| 419 | |
| 420 | fn is_ciphertext(bytes: &[u8], source: &MediaSource) -> bool { |
| 421 | use sha2::{Digest, Sha256}; |
| 422 | let MediaSource::Encrypted(file) = source else { |
| 423 | return false; |
| 424 | }; |
| 425 | file.hashes |
| 426 | .get(&ruma::events::room::EncryptedFileHashAlgorithm::Sha256) |
| 427 | .is_some_and(|hash| hash.as_bytes() == Sha256::digest(bytes).as_slice()) |
| 428 | } |
| 429 | |
| 430 | fn media(bytes: Vec<u8>, mimetype: Option<String>) -> Media { |
| 431 | let mimetype = mimetype |
| 432 | .or_else(|| sniff_image(&bytes).map(str::to_owned)) |
| 433 | .unwrap_or_else(|| "application/octet-stream".to_owned()); |
| 434 | Media { |
| 435 | bytes: Arc::from(bytes), |
| 436 | mimetype, |
| 437 | } |
| 438 | } |
| 439 | |
| 440 | /// the cache forgets content types, and avatars never declare one. images are the common case |
| 441 | fn sniff_image(bytes: &[u8]) -> Option<&'static str> { |
| 442 | match bytes { |
| 443 | [0x89, b'P', b'N', b'G', ..] => Some("image/png"), |
| 444 | [0xff, 0xd8, 0xff, ..] => Some("image/jpeg"), |
| 445 | [b'G', b'I', b'F', b'8', ..] => Some("image/gif"), |
| 446 | [ |
| 447 | b'R', |
| 448 | b'I', |
| 449 | b'F', |
| 450 | b'F', |
| 451 | _, |
| 452 | _, |
| 453 | _, |
| 454 | _, |
| 455 | b'W', |
| 456 | b'E', |
| 457 | b'B', |
| 458 | b'P', |
| 459 | .., |
| 460 | ] => Some("image/webp"), |
| 461 | _ => None, |
| 462 | } |
| 463 | } |
| 464 | |
| 465 | async fn cached_media( |
| 466 | client: &Client, |
| 467 | request: &MediaRequestParameters, |
| 468 | ) -> Result<Option<Vec<u8>>, Error> { |
| 469 | fn cache_failed( |
| 470 | source: impl std::error::Error + SendOutsideWasm + SyncOutsideWasm + 'static, |
| 471 | ) -> Error { |
| 472 | Error::with_source(ErrorKind::Store, "reading media cache failed", source) |
| 473 | } |
| 474 | client |
| 475 | .media_store() |
| 476 | .lock() |
| 477 | .await |
| 478 | .map_err(cache_failed)? |
| 479 | .get_media_content(request) |
| 480 | .await |
| 481 | .map_err(cache_failed) |
| 482 | } |
| 483 | |
| 484 | fn maybe_decrypt( |
| 485 | bytes: Vec<u8>, |
| 486 | request: &MediaRequestParameters, |
| 487 | max_bytes: usize, |
| 488 | ) -> Result<Vec<u8>, Error> { |
| 489 | let MediaSource::Encrypted(file) = &request.source else { |
| 490 | return Ok(bytes); |
| 491 | }; |
| 492 | fn decrypt_failed( |
| 493 | source: impl std::error::Error + SendOutsideWasm + SyncOutsideWasm + 'static, |
| 494 | ) -> Error { |
| 495 | Error::with_source( |
| 496 | ErrorKind::MediaUnavailable, |
| 497 | "media decryption failed", |
| 498 | source, |
| 499 | ) |
| 500 | } |
| 501 | let info: MediaEncryptionInfo = (**file).clone().into(); |
| 502 | let mut cursor = std::io::Cursor::new(bytes); |
| 503 | let mut decryptor = AttachmentDecryptor::new(&mut cursor, info).map_err(decrypt_failed)?; |
| 504 | let mut decrypted = Vec::new(); |
| 505 | let mut chunk = [0_u8; 8192]; |
| 506 | loop { |
| 507 | let read = decryptor.read(&mut chunk).map_err(decrypt_failed)?; |
| 508 | if read == 0 { |
| 509 | return Ok(decrypted); |
| 510 | } |
| 511 | if read > max_bytes.saturating_sub(decrypted.len()) { |
| 512 | return Err(too_large( |
| 513 | "decrypted media exceeds the configured byte limit", |
| 514 | )); |
| 515 | } |
| 516 | decrypted.extend_from_slice(&chunk[..read]); |
| 517 | } |
| 518 | } |
| 519 | |
| 520 | /// the sdk's own download has no size cap, so we do it by hand |
| 521 | async fn download_bounded( |
| 522 | client: &Client, |
| 523 | request: &MediaRequestParameters, |
| 524 | max_bytes: usize, |
| 525 | ) -> Result<(Vec<u8>, Option<String>), Error> { |
| 526 | if is_local(&request.source) { |
| 527 | return Err(Error::new( |
| 528 | ErrorKind::MediaUnavailable, |
| 529 | "local media is not in the media cache", |
| 530 | )); |
| 531 | } |
| 532 | let network_failed = |
| 533 | |source| Error::with_source(ErrorKind::Network, "media download failed", source); |
| 534 | let mut refreshed = false; |
| 535 | let response = loop { |
| 536 | let http_request = media_http_request(client, request).await?; |
| 537 | let (parts, _) = http_request.into_parts(); |
| 538 | let mut builder = client |
| 539 | .http_client() |
| 540 | .get(parts.uri.to_string()) |
| 541 | .timeout(DOWNLOAD_TOTAL); |
| 542 | for (name, value) in &parts.headers { |
| 543 | builder = builder.header(name.as_str(), value.as_bytes()); |
| 544 | } |
| 545 | let response = builder.send().await.map_err(network_failed)?; |
| 546 | if response.status() != matrix_sdk::reqwest::StatusCode::UNAUTHORIZED { |
| 547 | break response; |
| 548 | } |
| 549 | if refreshed |
| 550 | || client |
| 551 | .session_tokens() |
| 552 | .is_none_or(|tokens| tokens.refresh_token.is_none()) |
| 553 | { |
| 554 | return Err(Error::new( |
| 555 | ErrorKind::Unauthenticated, |
| 556 | "media download was not authorized", |
| 557 | )); |
| 558 | } |
| 559 | client.refresh_access_token().await.map_err(|source| { |
| 560 | Error::from_refresh_token("refreshing media access token failed", source) |
| 561 | })?; |
| 562 | refreshed = true; |
| 563 | }; |
| 564 | let status = response.status(); |
| 565 | if !status.is_success() { |
| 566 | return Err(Error::new( |
| 567 | ErrorKind::MediaUnavailable, |
| 568 | format!("media download failed with status {status}"), |
| 569 | )); |
| 570 | } |
| 571 | if response |
| 572 | .content_length() |
| 573 | .is_some_and(|length| length > max_bytes as u64) |
| 574 | { |
| 575 | return Err(too_large("media exceeds the configured byte limit")); |
| 576 | } |
| 577 | let mimetype = response |
| 578 | .headers() |
| 579 | .get(http::header::CONTENT_TYPE) |
| 580 | .and_then(|value| value.to_str().ok()) |
| 581 | .map(str::to_owned); |
| 582 | let mut stream = response.bytes_stream(); |
| 583 | let mut bytes = Vec::new(); |
| 584 | loop { |
| 585 | let chunk = matrix_sdk::timeout::timeout(stream.next(), DOWNLOAD_IDLE) |
| 586 | .await |
| 587 | .map_err(|_| Error::new(ErrorKind::Network, "media download stalled"))?; |
| 588 | let Some(chunk) = chunk else { break }; |
| 589 | let chunk = chunk.map_err(network_failed)?; |
| 590 | if chunk.len() > max_bytes.saturating_sub(bytes.len()) { |
| 591 | return Err(too_large( |
| 592 | "downloaded media exceeds the configured byte limit", |
| 593 | )); |
| 594 | } |
| 595 | bytes.extend_from_slice(&chunk); |
| 596 | } |
| 597 | Ok((bytes, mimetype)) |
| 598 | } |
| 599 | |
| 600 | async fn media_http_request( |
| 601 | client: &Client, |
| 602 | request: &MediaRequestParameters, |
| 603 | ) -> Result<http::Request<Vec<u8>>, Error> { |
| 604 | let versions = client.supported_versions().await.map_err(|source| { |
| 605 | Error::with_source( |
| 606 | ErrorKind::Network, |
| 607 | "reading supported server versions failed", |
| 608 | source, |
| 609 | ) |
| 610 | })?; |
| 611 | let authenticated = |
| 612 | authenticated_media::get_content::v1::Request::PATH_BUILDER.is_supported(&versions); |
| 613 | let base = client.homeserver().to_string(); |
| 614 | let token = client.access_token(); |
| 615 | let invalid_uri = |
| 616 | |source| Error::with_source(ErrorKind::Invalid, "media URI is invalid", source); |
| 617 | let (uri, thumbnail) = match (&request.source, &request.format) { |
| 618 | (MediaSource::Encrypted(file), _) => (&file.url, None), |
| 619 | (MediaSource::Plain(uri), MediaFormat::File) => (uri, None), |
| 620 | (MediaSource::Plain(uri), MediaFormat::Thumbnail(settings)) => (uri, Some(settings)), |
| 621 | }; |
| 622 | // each endpoint has its own auth input type, so no generic helper can take them all |
| 623 | macro_rules! into_http { |
| 624 | ($endpoint:expr, $token:expr) => { |
| 625 | $endpoint |
| 626 | .try_into_http_request::<Vec<u8>>(&base, $token, Cow::Borrowed(&versions)) |
| 627 | .map_err(|source| { |
| 628 | Error::with_source(ErrorKind::Other, "media request is invalid", source) |
| 629 | })? |
| 630 | }; |
| 631 | } |
| 632 | let authed = SendAccessToken::IfRequired(token.as_deref().unwrap_or("")); |
| 633 | #[allow(deprecated)] |
| 634 | let mut http_request = match (thumbnail, authenticated) { |
| 635 | (None, true) => into_http!( |
| 636 | authenticated_media::get_content::v1::Request::from_uri(uri).map_err(invalid_uri)?, |
| 637 | authed |
| 638 | ), |
| 639 | (None, false) => into_http!( |
| 640 | media::get_content::v3::Request::from_url(uri).map_err(invalid_uri)?, |
| 641 | SendAccessToken::None |
| 642 | ), |
| 643 | (Some(settings), true) => { |
| 644 | let mut endpoint = authenticated_media::get_content_thumbnail::v1::Request::from_uri( |
| 645 | uri, |
| 646 | settings.width, |
| 647 | settings.height, |
| 648 | ) |
| 649 | .map_err(invalid_uri)?; |
| 650 | endpoint.method = Some(settings.method.clone()); |
| 651 | endpoint.animated = Some(settings.animated); |
| 652 | into_http!(endpoint, authed) |
| 653 | } |
| 654 | (Some(settings), false) => { |
| 655 | let mut endpoint = media::get_content_thumbnail::v3::Request::from_url( |
| 656 | uri, |
| 657 | settings.width, |
| 658 | settings.height, |
| 659 | ) |
| 660 | .map_err(invalid_uri)?; |
| 661 | endpoint.method = Some(settings.method.clone()); |
| 662 | endpoint.animated = Some(settings.animated); |
| 663 | into_http!(endpoint, SendAccessToken::None) |
| 664 | } |
| 665 | }; |
| 666 | // legacy endpoints don't ask for a token but plenty of servers want one anyway |
| 667 | if let Some(token) = token.as_deref() { |
| 668 | let value = format!("Bearer {token}").parse().map_err(|_| { |
| 669 | Error::new(ErrorKind::Unauthenticated, "stored access token is invalid") |
| 670 | })?; |
| 671 | http_request |
| 672 | .headers_mut() |
| 673 | .insert(http::header::AUTHORIZATION, value); |
| 674 | } |
| 675 | Ok(http_request) |
| 676 | } |
| 677 | |
| 678 | pub(crate) async fn queue_upload( |
| 679 | room: Room, |
| 680 | upload: AttachmentUpload, |
| 681 | max_bytes: usize, |
| 682 | ) -> Result<(), Error> { |
| 683 | let (mut upload, mime, mut config) = prepare_upload(upload, max_bytes)?; |
| 684 | let thread_root = upload.timeline_id.thread_root().cloned(); |
| 685 | config.reply = match (upload.reply_to.take(), thread_root) { |
| 686 | (Some(event_id), thread_root) => { |
| 687 | let reply = || SdkReply { |
| 688 | event_id: event_id.clone(), |
| 689 | enforce_thread: if thread_root.is_some() { |
| 690 | EnforceThread::Threaded(ReplyWithinThread::Yes) |
| 691 | } else { |
| 692 | EnforceThread::MaybeThreaded |
| 693 | }, |
| 694 | add_mentions: if upload.notify_replied_to { |
| 695 | AddMentions::Yes |
| 696 | } else { |
| 697 | AddMentions::No |
| 698 | }, |
| 699 | }; |
| 700 | if let Some(thread_root) = &thread_root { |
| 701 | check_reply_thread(&room, reply(), thread_root).await?; |
| 702 | } |
| 703 | Some(reply()) |
| 704 | } |
| 705 | (None, Some(root_event_id)) => Some(SdkReply { |
| 706 | event_id: root_event_id, |
| 707 | enforce_thread: EnforceThread::Threaded(ReplyWithinThread::No), |
| 708 | add_mentions: AddMentions::No, |
| 709 | }), |
| 710 | (None, None) => None, |
| 711 | }; |
| 712 | room.send_queue() |
| 713 | .send_attachment(upload.filename, mime, upload.bytes, config) |
| 714 | .await |
| 715 | .map(drop) |
| 716 | .map_err(|source| { |
| 717 | use matrix_sdk::send_queue::RoomSendQueueError; |
| 718 | let kind = match &source { |
| 719 | RoomSendQueueError::RoomNotJoined | RoomSendQueueError::RoomDisappeared => { |
| 720 | ErrorKind::RoomUnknown |
| 721 | } |
| 722 | RoomSendQueueError::StorageError(_) => ErrorKind::Store, |
| 723 | _ => ErrorKind::MediaUnavailable, |
| 724 | }; |
| 725 | Error::with_source(kind, "media upload could not be queued", source) |
| 726 | }) |
| 727 | } |
| 728 | |
| 729 | /// the send queue builds the relation itself, so a reply that would land in another thread has to be caught before it is queued |
| 730 | async fn check_reply_thread( |
| 731 | room: &Room, |
| 732 | reply: SdkReply, |
| 733 | thread_root: &OwnedEventId, |
| 734 | ) -> Result<(), Error> { |
| 735 | let content = room |
| 736 | .make_reply_event( |
| 737 | RoomMessageEventContentWithoutRelation::text_plain(""), |
| 738 | reply, |
| 739 | ) |
| 740 | .await |
| 741 | .map_err(|source| { |
| 742 | Error::with_source(ErrorKind::Timeline, "reply operation failed", source) |
| 743 | })?; |
| 744 | if matches!( |
| 745 | content.relates_to.as_ref(), |
| 746 | Some(Relation::Thread(thread)) if &thread.event_id == thread_root |
| 747 | ) { |
| 748 | Ok(()) |
| 749 | } else { |
| 750 | Err(Error::new( |
| 751 | ErrorKind::Invalid, |
| 752 | "reply target does not belong to the selected thread", |
| 753 | )) |
| 754 | } |
| 755 | } |
| 756 | |
| 757 | /// every entry belongs to the scopes that registered it and lives until the last of them |
| 758 | /// lets go. there is no global cap, a bounded scope only ever trims itself |
| 759 | #[derive(Default)] |
| 760 | struct MediaRegistry { |
| 761 | entries: HashMap<MediaId, Entry>, |
| 762 | scopes: HashMap<u64, ScopeEntries>, |
| 763 | } |
| 764 | |
| 765 | struct Entry { |
| 766 | media: MediaRegistration, |
| 767 | owners: Vec<u64>, |
| 768 | } |
| 769 | |
| 770 | /// least recently used first. stale order slots are skipped by stamp rather than hunted down |
| 771 | #[derive(Default)] |
| 772 | struct ScopeEntries { |
| 773 | stamps: HashMap<MediaId, u64>, |
| 774 | order: VecDeque<(MediaId, u64)>, |
| 775 | next_stamp: u64, |
| 776 | limit: Option<usize>, |
| 777 | } |
| 778 | |
| 779 | impl ScopeEntries { |
| 780 | fn touch(&mut self, id: &MediaId) { |
| 781 | let Some(limit) = self.limit else { |
| 782 | self.stamps.insert(id.clone(), 0); |
| 783 | return; |
| 784 | }; |
| 785 | let stamp = self.next_stamp; |
| 786 | self.next_stamp += 1; |
| 787 | self.stamps.insert(id.clone(), stamp); |
| 788 | self.order.push_back((id.clone(), stamp)); |
| 789 | if self.order.len() > limit.saturating_mul(4) { |
| 790 | let stamps = &self.stamps; |
| 791 | self.order |
| 792 | .retain(|(id, stamp)| stamps.get(id) == Some(stamp)); |
| 793 | } |
| 794 | } |
| 795 | |
| 796 | /// what fell out past the limit |
| 797 | fn trim(&mut self) -> Vec<MediaId> { |
| 798 | let Some(limit) = self.limit else { |
| 799 | return Vec::new(); |
| 800 | }; |
| 801 | let mut dropped = Vec::new(); |
| 802 | while self.stamps.len() > limit { |
| 803 | let Some((id, stamp)) = self.order.pop_front() else { |
| 804 | break; |
| 805 | }; |
| 806 | if self.stamps.get(&id) == Some(&stamp) { |
| 807 | self.stamps.remove(&id); |
| 808 | dropped.push(id); |
| 809 | } |
| 810 | } |
| 811 | dropped |
| 812 | } |
| 813 | } |
| 814 | |
| 815 | impl MediaRegistry { |
| 816 | fn get(&mut self, id: &MediaId) -> Option<&MediaRegistration> { |
| 817 | let entry = self.entries.get(id)?; |
| 818 | for owner in &entry.owners { |
| 819 | if let Some(scope) = self.scopes.get_mut(owner) { |
| 820 | scope.touch(id); |
| 821 | } |
| 822 | } |
| 823 | Some(&entry.media) |
| 824 | } |
| 825 | |
| 826 | fn insert_all(&mut self, scope: &MediaScope, media: Vec<MediaRegistration>) { |
| 827 | let owner = scope.nonce(); |
| 828 | let entries = self.scopes.entry(owner).or_insert_with(|| ScopeEntries { |
| 829 | limit: scope.limit(), |
| 830 | ..ScopeEntries::default() |
| 831 | }); |
| 832 | for media in media { |
| 833 | let id = media.id.clone(); |
| 834 | entries.touch(&id); |
| 835 | let entry = self.entries.entry(id).or_insert_with(|| Entry { |
| 836 | media: media.clone(), |
| 837 | owners: Vec::new(), |
| 838 | }); |
| 839 | entry.media = media; |
| 840 | if !entry.owners.contains(&owner) { |
| 841 | entry.owners.push(owner); |
| 842 | } |
| 843 | } |
| 844 | for id in entries.trim() { |
| 845 | Self::disown(&mut self.entries, &id, owner); |
| 846 | } |
| 847 | } |
| 848 | |
| 849 | fn disown(entries: &mut HashMap<MediaId, Entry>, id: &MediaId, owner: u64) { |
| 850 | let Some(entry) = entries.get_mut(id) else { |
| 851 | return; |
| 852 | }; |
| 853 | entry.owners.retain(|current| *current != owner); |
| 854 | if entry.owners.is_empty() { |
| 855 | entries.remove(id); |
| 856 | } |
| 857 | } |
| 858 | |
| 859 | fn release(&mut self, scope: &MediaScope) { |
| 860 | let Some(released) = self.scopes.remove(&scope.nonce()) else { |
| 861 | return; |
| 862 | }; |
| 863 | for id in released.stamps.keys() { |
| 864 | Self::disown(&mut self.entries, id, scope.nonce()); |
| 865 | } |
| 866 | } |
| 867 | |
| 868 | fn clear(&mut self) { |
| 869 | self.entries.clear(); |
| 870 | self.scopes.clear(); |
| 871 | } |
| 872 | } |
| 873 | |
| 874 | #[cfg(test)] |
| 875 | mod tests { |
| 876 | use super::*; |
| 877 | use crate::MediaKind; |
| 878 | |
| 879 | fn media(name: &str) -> MediaRegistration { |
| 880 | let uri = ruma::OwnedMxcUri::from(format!("mxc://example.org/{name}")); |
| 881 | MediaRegistration { |
| 882 | id: MediaId::new(format!("test#{name}")), |
| 883 | kind: MediaKind::Image, |
| 884 | full: MediaSourceInfo { |
| 885 | source: MediaSource::Plain(uri), |
| 886 | mimetype: None, |
| 887 | declared_size: None, |
| 888 | }, |
| 889 | thumbnail: None, |
| 890 | } |
| 891 | } |
| 892 | |
| 893 | fn has(registry: &mut MediaRegistry, name: &str) -> bool { |
| 894 | registry.get(&media(name).id).is_some() |
| 895 | } |
| 896 | |
| 897 | #[test] |
| 898 | fn shared_media_lives_until_every_owner_lets_go() { |
| 899 | let mut registry = MediaRegistry::default(); |
| 900 | let (timeline, context) = (MediaScope::new(), MediaScope::bounded(8)); |
| 901 | registry.insert_all(&timeline, vec![media("shared"), media("row")]); |
| 902 | registry.insert_all(&context, vec![media("shared"), media("page")]); |
| 903 | registry.release(&context); |
| 904 | assert!(has(&mut registry, "shared")); |
| 905 | assert!(has(&mut registry, "row")); |
| 906 | assert!(!has(&mut registry, "page")); |
| 907 | registry.release(&timeline); |
| 908 | assert!(!has(&mut registry, "shared")); |
| 909 | assert!(registry.entries.is_empty()); |
| 910 | } |
| 911 | |
| 912 | #[test] |
| 913 | fn heavy_context_paging_trims_only_its_own_least_recent_entries() { |
| 914 | let mut registry = MediaRegistry::default(); |
| 915 | let (timeline, context) = (MediaScope::new(), MediaScope::bounded(4)); |
| 916 | registry.insert_all(&timeline, vec![media("on-screen"), media("both")]); |
| 917 | registry.insert_all(&context, vec![media("both"), media("kept")]); |
| 918 | for page in 0..50 { |
| 919 | assert!(has(&mut registry, "kept")); |
| 920 | registry.insert_all(&context, vec![media(&format!("page-{page}"))]); |
| 921 | } |
| 922 | assert!(has(&mut registry, "on-screen")); |
| 923 | assert!(has(&mut registry, "both")); |
| 924 | assert!(has(&mut registry, "kept")); |
| 925 | assert!(!has(&mut registry, "page-0")); |
| 926 | assert!(registry.scopes[&context.nonce()].stamps.len() <= 4); |
| 927 | registry.release(&context); |
| 928 | assert!(has(&mut registry, "both")); |
| 929 | assert!(!has(&mut registry, "kept")); |
| 930 | } |
| 931 | } |