Line data Source code
1 : //! Azure Blob Storage wrapper
2 :
3 : use std::borrow::Cow;
4 : use std::collections::HashMap;
5 : use std::env;
6 : use std::io;
7 : use std::num::NonZeroU32;
8 : use std::pin::Pin;
9 : use std::str::FromStr;
10 : use std::sync::Arc;
11 : use std::time::Duration;
12 : use std::time::SystemTime;
13 :
14 : use super::REMOTE_STORAGE_PREFIX_SEPARATOR;
15 : use anyhow::Result;
16 : use azure_core::request_options::{MaxResults, Metadata, Range};
17 : use azure_core::RetryOptions;
18 : use azure_identity::DefaultAzureCredential;
19 : use azure_storage::StorageCredentials;
20 : use azure_storage_blobs::blob::CopyStatus;
21 : use azure_storage_blobs::prelude::ClientBuilder;
22 : use azure_storage_blobs::{blob::operations::GetBlobBuilder, prelude::ContainerClient};
23 : use bytes::Bytes;
24 : use futures::future::Either;
25 : use futures::stream::Stream;
26 : use futures_util::StreamExt;
27 : use futures_util::TryStreamExt;
28 : use http_types::{StatusCode, Url};
29 : use tokio_util::sync::CancellationToken;
30 : use tracing::debug;
31 :
32 : use crate::{
33 : error::Cancelled, s3_bucket::RequestKind, AzureConfig, ConcurrencyLimiter, Download,
34 : DownloadError, Listing, ListingMode, RemotePath, RemoteStorage, StorageMetadata,
35 : TimeTravelError, TimeoutOrCancel,
36 : };
37 :
38 : pub struct AzureBlobStorage {
39 : client: ContainerClient,
40 : prefix_in_container: Option<String>,
41 : max_keys_per_list_response: Option<NonZeroU32>,
42 : concurrency_limiter: ConcurrencyLimiter,
43 : // Per-request timeout. Accessible for tests.
44 : pub timeout: Duration,
45 : }
46 :
47 : impl AzureBlobStorage {
48 6 : pub fn new(azure_config: &AzureConfig, timeout: Duration) -> Result<Self> {
49 6 : debug!(
50 0 : "Creating azure remote storage for azure container {}",
51 : azure_config.container_name
52 : );
53 :
54 6 : let account = env::var("AZURE_STORAGE_ACCOUNT").expect("missing AZURE_STORAGE_ACCOUNT");
55 :
56 : // If the `AZURE_STORAGE_ACCESS_KEY` env var has an access key, use that,
57 : // otherwise try the token based credentials.
58 6 : let credentials = if let Ok(access_key) = env::var("AZURE_STORAGE_ACCESS_KEY") {
59 6 : StorageCredentials::access_key(account.clone(), access_key)
60 : } else {
61 0 : let token_credential = DefaultAzureCredential::default();
62 0 : StorageCredentials::token_credential(Arc::new(token_credential))
63 : };
64 :
65 : // we have an outer retry
66 6 : let builder = ClientBuilder::new(account, credentials).retry(RetryOptions::none());
67 6 :
68 6 : let client = builder.container_client(azure_config.container_name.to_owned());
69 :
70 6 : let max_keys_per_list_response =
71 6 : if let Some(limit) = azure_config.max_keys_per_list_response {
72 : Some(
73 2 : NonZeroU32::new(limit as u32)
74 2 : .ok_or_else(|| anyhow::anyhow!("max_keys_per_list_response can't be 0"))?,
75 : )
76 : } else {
77 4 : None
78 : };
79 :
80 6 : Ok(AzureBlobStorage {
81 6 : client,
82 6 : prefix_in_container: azure_config.prefix_in_container.to_owned(),
83 6 : max_keys_per_list_response,
84 6 : concurrency_limiter: ConcurrencyLimiter::new(azure_config.concurrency_limit.get()),
85 6 : timeout,
86 6 : })
87 6 : }
88 :
89 107 : pub fn relative_path_to_name(&self, path: &RemotePath) -> String {
90 107 : assert_eq!(std::path::MAIN_SEPARATOR, REMOTE_STORAGE_PREFIX_SEPARATOR);
91 107 : let path_string = path
92 107 : .get_path()
93 107 : .as_str()
94 107 : .trim_end_matches(REMOTE_STORAGE_PREFIX_SEPARATOR);
95 107 : match &self.prefix_in_container {
96 107 : Some(prefix) => {
97 107 : if prefix.ends_with(REMOTE_STORAGE_PREFIX_SEPARATOR) {
98 107 : prefix.clone() + path_string
99 : } else {
100 0 : format!("{prefix}{REMOTE_STORAGE_PREFIX_SEPARATOR}{path_string}")
101 : }
102 : }
103 0 : None => path_string.to_string(),
104 : }
105 107 : }
106 :
107 53 : fn name_to_relative_path(&self, key: &str) -> RemotePath {
108 53 : let relative_path =
109 53 : match key.strip_prefix(self.prefix_in_container.as_deref().unwrap_or_default()) {
110 53 : Some(stripped) => stripped,
111 : // we rely on Azure to return properly prefixed paths
112 : // for requests with a certain prefix
113 0 : None => panic!(
114 0 : "Key {key} does not start with container prefix {:?}",
115 0 : self.prefix_in_container
116 0 : ),
117 : };
118 53 : RemotePath(
119 53 : relative_path
120 53 : .split(REMOTE_STORAGE_PREFIX_SEPARATOR)
121 53 : .collect(),
122 53 : )
123 53 : }
124 :
125 7 : async fn download_for_builder(
126 7 : &self,
127 7 : builder: GetBlobBuilder,
128 7 : cancel: &CancellationToken,
129 7 : ) -> Result<Download, DownloadError> {
130 0 : let kind = RequestKind::Get;
131 :
132 0 : let _permit = self.permit(kind, cancel).await?;
133 0 : let cancel_or_timeout = crate::support::cancel_or_timeout(self.timeout, cancel.clone());
134 0 : let cancel_or_timeout_ = crate::support::cancel_or_timeout(self.timeout, cancel.clone());
135 0 :
136 0 : let mut etag = None;
137 0 : let mut last_modified = None;
138 0 : let mut metadata = HashMap::new();
139 0 :
140 0 : let download = async {
141 0 : let response = builder
142 0 : // convert to concrete Pageable
143 0 : .into_stream()
144 0 : // convert to TryStream
145 0 : .into_stream()
146 0 : .map_err(to_download_error);
147 0 :
148 0 : // apply per request timeout
149 0 : let response = tokio_stream::StreamExt::timeout(response, self.timeout);
150 0 :
151 0 : // flatten
152 0 : let response = response.map(|res| match res {
153 0 : Ok(res) => res,
154 0 : Err(_elapsed) => Err(DownloadError::Timeout),
155 0 : });
156 0 :
157 0 : let mut response = Box::pin(response);
158 :
159 0 : let Some(part) = response.next().await else {
160 0 : return Err(DownloadError::Other(anyhow::anyhow!(
161 0 : "Azure GET response contained no response body"
162 0 : )));
163 : };
164 0 : let part = part?;
165 0 : if etag.is_none() {
166 0 : etag = Some(part.blob.properties.etag);
167 0 : }
168 0 : if last_modified.is_none() {
169 0 : last_modified = Some(part.blob.properties.last_modified.into());
170 0 : }
171 0 : if let Some(blob_meta) = part.blob.metadata {
172 0 : metadata.extend(blob_meta.iter().map(|(k, v)| (k.to_owned(), v.to_owned())));
173 0 : }
174 :
175 : // unwrap safety: if these were None, bufs would be empty and we would have returned an error already
176 0 : let etag = etag.unwrap();
177 0 : let last_modified = last_modified.unwrap();
178 0 :
179 0 : let tail_stream = response
180 0 : .map(|part| match part {
181 0 : Ok(part) => Either::Left(part.data.map(|r| r.map_err(io::Error::other))),
182 0 : Err(e) => {
183 0 : Either::Right(futures::stream::once(async { Err(io::Error::other(e)) }))
184 : }
185 0 : })
186 0 : .flatten();
187 0 : let stream = part
188 0 : .data
189 0 : .map(|r| r.map_err(io::Error::other))
190 0 : .chain(sync_wrapper::SyncStream::new(tail_stream));
191 0 : //.chain(SyncStream::from_pin(Box::pin(tail_stream)));
192 0 :
193 0 : let download_stream = crate::support::DownloadStream::new(cancel_or_timeout_, stream);
194 0 :
195 0 : Ok(Download {
196 0 : download_stream: Box::pin(download_stream),
197 0 : etag,
198 0 : last_modified,
199 0 : metadata: Some(StorageMetadata(metadata)),
200 0 : })
201 0 : };
202 :
203 : tokio::select! {
204 : bufs = download => bufs,
205 : cancel_or_timeout = cancel_or_timeout => match cancel_or_timeout {
206 : TimeoutOrCancel::Timeout => Err(DownloadError::Timeout),
207 : TimeoutOrCancel::Cancel => Err(DownloadError::Cancelled),
208 : },
209 : }
210 0 : }
211 :
212 108 : async fn permit(
213 108 : &self,
214 108 : kind: RequestKind,
215 108 : cancel: &CancellationToken,
216 108 : ) -> Result<tokio::sync::SemaphorePermit<'_>, Cancelled> {
217 0 : let acquire = self.concurrency_limiter.acquire(kind);
218 :
219 : tokio::select! {
220 : permit = acquire => Ok(permit.expect("never closed")),
221 : _ = cancel.cancelled() => Err(Cancelled),
222 : }
223 0 : }
224 : }
225 :
226 0 : fn to_azure_metadata(metadata: StorageMetadata) -> Metadata {
227 0 : let mut res = Metadata::new();
228 0 : for (k, v) in metadata.0.into_iter() {
229 0 : res.insert(k, v);
230 0 : }
231 0 : res
232 0 : }
233 :
234 0 : fn to_download_error(error: azure_core::Error) -> DownloadError {
235 0 : if let Some(http_err) = error.as_http_error() {
236 0 : match http_err.status() {
237 0 : StatusCode::NotFound => DownloadError::NotFound,
238 0 : StatusCode::BadRequest => DownloadError::BadInput(anyhow::Error::new(error)),
239 0 : _ => DownloadError::Other(anyhow::Error::new(error)),
240 : }
241 : } else {
242 0 : DownloadError::Other(error.into())
243 : }
244 0 : }
245 :
246 : impl RemoteStorage for AzureBlobStorage {
247 6 : async fn list(
248 6 : &self,
249 6 : prefix: Option<&RemotePath>,
250 6 : mode: ListingMode,
251 6 : max_keys: Option<NonZeroU32>,
252 6 : cancel: &CancellationToken,
253 6 : ) -> anyhow::Result<Listing, DownloadError> {
254 0 : let _permit = self.permit(RequestKind::List, cancel).await?;
255 :
256 0 : let op = async {
257 0 : // get the passed prefix or if it is not set use prefix_in_bucket value
258 0 : let list_prefix = prefix
259 0 : .map(|p| self.relative_path_to_name(p))
260 0 : .or_else(|| self.prefix_in_container.clone())
261 0 : .map(|mut p| {
262 : // required to end with a separator
263 : // otherwise request will return only the entry of a prefix
264 0 : if matches!(mode, ListingMode::WithDelimiter)
265 0 : && !p.ends_with(REMOTE_STORAGE_PREFIX_SEPARATOR)
266 0 : {
267 0 : p.push(REMOTE_STORAGE_PREFIX_SEPARATOR);
268 0 : }
269 0 : p
270 0 : });
271 0 :
272 0 : let mut builder = self.client.list_blobs();
273 0 :
274 0 : if let ListingMode::WithDelimiter = mode {
275 0 : builder = builder.delimiter(REMOTE_STORAGE_PREFIX_SEPARATOR.to_string());
276 0 : }
277 :
278 0 : if let Some(prefix) = list_prefix {
279 0 : builder = builder.prefix(Cow::from(prefix.to_owned()));
280 0 : }
281 :
282 0 : if let Some(limit) = self.max_keys_per_list_response {
283 0 : builder = builder.max_results(MaxResults::new(limit));
284 0 : }
285 :
286 0 : let response = builder.into_stream();
287 0 : let response = response.into_stream().map_err(to_download_error);
288 0 : let response = tokio_stream::StreamExt::timeout(response, self.timeout);
289 0 : let response = response.map(|res| match res {
290 0 : Ok(res) => res,
291 0 : Err(_elapsed) => Err(DownloadError::Timeout),
292 0 : });
293 0 :
294 0 : let mut response = std::pin::pin!(response);
295 0 :
296 0 : let mut res = Listing::default();
297 0 :
298 0 : let mut max_keys = max_keys.map(|mk| mk.get());
299 0 : while let Some(entry) = response.next().await {
300 0 : let entry = entry?;
301 0 : let prefix_iter = entry
302 0 : .blobs
303 0 : .prefixes()
304 0 : .map(|prefix| self.name_to_relative_path(&prefix.name));
305 0 : res.prefixes.extend(prefix_iter);
306 0 :
307 0 : let blob_iter = entry
308 0 : .blobs
309 0 : .blobs()
310 0 : .map(|k| self.name_to_relative_path(&k.name));
311 :
312 0 : for key in blob_iter {
313 0 : res.keys.push(key);
314 :
315 0 : if let Some(mut mk) = max_keys {
316 0 : assert!(mk > 0);
317 0 : mk -= 1;
318 0 : if mk == 0 {
319 0 : return Ok(res); // limit reached
320 0 : }
321 0 : max_keys = Some(mk);
322 0 : }
323 : }
324 : }
325 :
326 0 : Ok(res)
327 0 : };
328 :
329 : tokio::select! {
330 : res = op => res,
331 : _ = cancel.cancelled() => Err(DownloadError::Cancelled),
332 : }
333 0 : }
334 :
335 0 : async fn upload(
336 0 : &self,
337 0 : from: impl Stream<Item = std::io::Result<Bytes>> + Send + Sync + 'static,
338 0 : data_size_bytes: usize,
339 0 : to: &RemotePath,
340 0 : metadata: Option<StorageMetadata>,
341 0 : cancel: &CancellationToken,
342 0 : ) -> anyhow::Result<()> {
343 0 : let _permit = self.permit(RequestKind::Put, cancel).await?;
344 :
345 0 : let op = async {
346 0 : let blob_client = self.client.blob_client(self.relative_path_to_name(to));
347 0 :
348 0 : let from: Pin<Box<dyn Stream<Item = std::io::Result<Bytes>> + Send + Sync + 'static>> =
349 0 : Box::pin(from);
350 0 :
351 0 : let from = NonSeekableStream::new(from, data_size_bytes);
352 0 :
353 0 : let body = azure_core::Body::SeekableStream(Box::new(from));
354 0 :
355 0 : let mut builder = blob_client.put_block_blob(body);
356 :
357 0 : if let Some(metadata) = metadata {
358 0 : builder = builder.metadata(to_azure_metadata(metadata));
359 0 : }
360 :
361 0 : let fut = builder.into_future();
362 0 : let fut = tokio::time::timeout(self.timeout, fut);
363 0 :
364 0 : match fut.await {
365 0 : Ok(Ok(_response)) => Ok(()),
366 0 : Ok(Err(azure)) => Err(azure.into()),
367 0 : Err(_timeout) => Err(TimeoutOrCancel::Cancel.into()),
368 : }
369 0 : };
370 :
371 : tokio::select! {
372 : res = op => res,
373 : _ = cancel.cancelled() => Err(TimeoutOrCancel::Cancel.into()),
374 : }
375 0 : }
376 :
377 2 : async fn download(
378 2 : &self,
379 2 : from: &RemotePath,
380 2 : cancel: &CancellationToken,
381 2 : ) -> Result<Download, DownloadError> {
382 0 : let blob_client = self.client.blob_client(self.relative_path_to_name(from));
383 0 :
384 0 : let builder = blob_client.get();
385 0 :
386 0 : self.download_for_builder(builder, cancel).await
387 0 : }
388 :
389 5 : async fn download_byte_range(
390 5 : &self,
391 5 : from: &RemotePath,
392 5 : start_inclusive: u64,
393 5 : end_exclusive: Option<u64>,
394 5 : cancel: &CancellationToken,
395 5 : ) -> Result<Download, DownloadError> {
396 0 : let blob_client = self.client.blob_client(self.relative_path_to_name(from));
397 0 :
398 0 : let mut builder = blob_client.get();
399 :
400 0 : let range: Range = if let Some(end_exclusive) = end_exclusive {
401 0 : (start_inclusive..end_exclusive).into()
402 : } else {
403 0 : (start_inclusive..).into()
404 : };
405 0 : builder = builder.range(range);
406 0 :
407 0 : self.download_for_builder(builder, cancel).await
408 0 : }
409 :
410 44 : async fn delete(&self, path: &RemotePath, cancel: &CancellationToken) -> anyhow::Result<()> {
411 0 : self.delete_objects(std::array::from_ref(path), cancel)
412 0 : .await
413 0 : }
414 :
415 47 : async fn delete_objects<'a>(
416 47 : &self,
417 47 : paths: &'a [RemotePath],
418 47 : cancel: &CancellationToken,
419 47 : ) -> anyhow::Result<()> {
420 0 : let _permit = self.permit(RequestKind::Delete, cancel).await?;
421 :
422 0 : let op = async {
423 : // TODO batch requests are also not supported by the SDK
424 : // https://github.com/Azure/azure-sdk-for-rust/issues/1068
425 : // https://github.com/Azure/azure-sdk-for-rust/issues/1249
426 0 : for path in paths {
427 0 : let blob_client = self.client.blob_client(self.relative_path_to_name(path));
428 0 :
429 0 : let request = blob_client.delete().into_future();
430 :
431 0 : let res = tokio::time::timeout(self.timeout, request).await;
432 :
433 0 : match res {
434 0 : Ok(Ok(_response)) => continue,
435 0 : Ok(Err(e)) => {
436 0 : if let Some(http_err) = e.as_http_error() {
437 0 : if http_err.status() == StatusCode::NotFound {
438 0 : continue;
439 0 : }
440 0 : }
441 0 : return Err(e.into());
442 : }
443 0 : Err(_elapsed) => return Err(TimeoutOrCancel::Timeout.into()),
444 : }
445 : }
446 :
447 0 : Ok(())
448 0 : };
449 :
450 : tokio::select! {
451 : res = op => res,
452 : _ = cancel.cancelled() => Err(TimeoutOrCancel::Cancel.into()),
453 : }
454 0 : }
455 :
456 1 : async fn copy(
457 1 : &self,
458 1 : from: &RemotePath,
459 1 : to: &RemotePath,
460 1 : cancel: &CancellationToken,
461 1 : ) -> anyhow::Result<()> {
462 0 : let _permit = self.permit(RequestKind::Copy, cancel).await?;
463 :
464 0 : let timeout = tokio::time::sleep(self.timeout);
465 0 :
466 0 : let mut copy_status = None;
467 0 :
468 0 : let op = async {
469 0 : let blob_client = self.client.blob_client(self.relative_path_to_name(to));
470 :
471 0 : let source_url = format!(
472 0 : "{}/{}",
473 0 : self.client.url()?,
474 0 : self.relative_path_to_name(from)
475 : );
476 :
477 0 : let builder = blob_client.copy(Url::from_str(&source_url)?);
478 0 : let copy = builder.into_future();
479 :
480 0 : let result = copy.await?;
481 :
482 0 : copy_status = Some(result.copy_status);
483 : loop {
484 0 : match copy_status.as_ref().expect("we always set it to Some") {
485 : CopyStatus::Aborted => {
486 0 : anyhow::bail!("Received abort for copy from {from} to {to}.");
487 : }
488 : CopyStatus::Failed => {
489 0 : anyhow::bail!("Received failure response for copy from {from} to {to}.");
490 : }
491 0 : CopyStatus::Success => return Ok(()),
492 0 : CopyStatus::Pending => (),
493 0 : }
494 0 : // The copy is taking longer. Waiting a second and then re-trying.
495 0 : // TODO estimate time based on copy_progress and adjust time based on that
496 0 : tokio::time::sleep(Duration::from_millis(1000)).await;
497 0 : let properties = blob_client.get_properties().into_future().await?;
498 0 : let Some(status) = properties.blob.properties.copy_status else {
499 0 : tracing::warn!("copy_status for copy is None!, from={from}, to={to}");
500 0 : return Ok(());
501 : };
502 0 : copy_status = Some(status);
503 : }
504 0 : };
505 :
506 : tokio::select! {
507 : res = op => res,
508 : _ = cancel.cancelled() => Err(anyhow::Error::new(TimeoutOrCancel::Cancel)),
509 : _ = timeout => {
510 : let e = anyhow::Error::new(TimeoutOrCancel::Timeout);
511 : let e = e.context(format!("Timeout, last status: {copy_status:?}"));
512 : Err(e)
513 : },
514 : }
515 0 : }
516 :
517 0 : async fn time_travel_recover(
518 0 : &self,
519 0 : _prefix: Option<&RemotePath>,
520 0 : _timestamp: SystemTime,
521 0 : _done_if_after: SystemTime,
522 0 : _cancel: &CancellationToken,
523 0 : ) -> Result<(), TimeTravelError> {
524 0 : // TODO use Azure point in time recovery feature for this
525 0 : // https://learn.microsoft.com/en-us/azure/storage/blobs/point-in-time-restore-overview
526 0 : Err(TimeTravelError::Unimplemented)
527 0 : }
528 : }
529 :
530 : pin_project_lite::pin_project! {
531 : /// Hack to work around not being able to stream once with azure sdk.
532 : ///
533 : /// Azure sdk clones streams around with the assumption that they are like
534 : /// `Arc<tokio::fs::File>` (except not supporting tokio), however our streams are not like
535 : /// that. For example for an `index_part.json` we just have a single chunk of [`Bytes`]
536 : /// representing the whole serialized vec. It could be trivially cloneable and "semi-trivially"
537 : /// seekable, but we can also just re-try the request easier.
538 : #[project = NonSeekableStreamProj]
539 : enum NonSeekableStream<S> {
540 : /// A stream wrappers initial form.
541 : ///
542 : /// Mutex exists to allow moving when cloning. If the sdk changes to do less than 1
543 : /// clone before first request, then this must be changed.
544 : Initial {
545 : inner: std::sync::Mutex<Option<tokio_util::compat::Compat<tokio_util::io::StreamReader<S, Bytes>>>>,
546 : len: usize,
547 : },
548 : /// The actually readable variant, produced by cloning the Initial variant.
549 : ///
550 : /// The sdk currently always clones once, even without retry policy.
551 : Actual {
552 : #[pin]
553 : inner: tokio_util::compat::Compat<tokio_util::io::StreamReader<S, Bytes>>,
554 : len: usize,
555 : read_any: bool,
556 : },
557 : /// Most likely unneeded, but left to make life easier, in case more clones are added.
558 : Cloned {
559 : len_was: usize,
560 : }
561 : }
562 : }
563 :
564 : impl<S> NonSeekableStream<S>
565 : where
566 : S: Stream<Item = std::io::Result<Bytes>> + Send + Sync + 'static,
567 : {
568 0 : fn new(inner: S, len: usize) -> NonSeekableStream<S> {
569 0 : use tokio_util::compat::TokioAsyncReadCompatExt;
570 0 :
571 0 : let inner = tokio_util::io::StreamReader::new(inner).compat();
572 0 : let inner = Some(inner);
573 0 : let inner = std::sync::Mutex::new(inner);
574 0 : NonSeekableStream::Initial { inner, len }
575 0 : }
576 : }
577 :
578 : impl<S> std::fmt::Debug for NonSeekableStream<S> {
579 0 : fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
580 0 : match self {
581 0 : Self::Initial { len, .. } => f.debug_struct("Initial").field("len", len).finish(),
582 0 : Self::Actual { len, .. } => f.debug_struct("Actual").field("len", len).finish(),
583 0 : Self::Cloned { len_was, .. } => f.debug_struct("Cloned").field("len", len_was).finish(),
584 : }
585 0 : }
586 : }
587 :
588 : impl<S> futures::io::AsyncRead for NonSeekableStream<S>
589 : where
590 : S: Stream<Item = std::io::Result<Bytes>>,
591 : {
592 0 : fn poll_read(
593 0 : self: std::pin::Pin<&mut Self>,
594 0 : cx: &mut std::task::Context<'_>,
595 0 : buf: &mut [u8],
596 0 : ) -> std::task::Poll<std::io::Result<usize>> {
597 0 : match self.project() {
598 : NonSeekableStreamProj::Actual {
599 0 : inner, read_any, ..
600 0 : } => {
601 0 : *read_any = true;
602 0 : inner.poll_read(cx, buf)
603 : }
604 : // NonSeekableStream::Initial does not support reading because it is just much easier
605 : // to have the mutex in place where one does not poll the contents, or that's how it
606 : // seemed originally. If there is a version upgrade which changes the cloning, then
607 : // that support needs to be hacked in.
608 : //
609 : // including {self:?} into the message would be useful, but unsure how to unproject.
610 0 : _ => std::task::Poll::Ready(Err(std::io::Error::new(
611 0 : std::io::ErrorKind::Other,
612 0 : "cloned or initial values cannot be read",
613 0 : ))),
614 : }
615 0 : }
616 : }
617 :
618 : impl<S> Clone for NonSeekableStream<S> {
619 : /// Weird clone implementation exists to support the sdk doing cloning before issuing the first
620 : /// request, see type documentation.
621 0 : fn clone(&self) -> Self {
622 0 : use NonSeekableStream::*;
623 0 :
624 0 : match self {
625 0 : Initial { inner, len } => {
626 0 : if let Some(inner) = inner.lock().unwrap().take() {
627 0 : Actual {
628 0 : inner,
629 0 : len: *len,
630 0 : read_any: false,
631 0 : }
632 : } else {
633 0 : Self::Cloned { len_was: *len }
634 : }
635 : }
636 0 : Actual { len, .. } => Cloned { len_was: *len },
637 0 : Cloned { len_was } => Cloned { len_was: *len_was },
638 : }
639 0 : }
640 : }
641 :
642 : #[async_trait::async_trait]
643 : impl<S> azure_core::SeekableStream for NonSeekableStream<S>
644 : where
645 : S: Stream<Item = std::io::Result<Bytes>> + Unpin + Send + Sync + 'static,
646 : {
647 0 : async fn reset(&mut self) -> azure_core::error::Result<()> {
648 0 : use NonSeekableStream::*;
649 0 :
650 0 : let msg = match self {
651 0 : Initial { inner, .. } => {
652 0 : if inner.get_mut().unwrap().is_some() {
653 0 : return Ok(());
654 0 : } else {
655 0 : "reset after first clone is not supported"
656 0 : }
657 0 : }
658 0 : Actual { read_any, .. } if !*read_any => return Ok(()),
659 0 : Actual { .. } => "reset after reading is not supported",
660 0 : Cloned { .. } => "reset after second clone is not supported",
661 0 : };
662 0 : Err(azure_core::error::Error::new(
663 0 : azure_core::error::ErrorKind::Io,
664 0 : std::io::Error::new(std::io::ErrorKind::Other, msg),
665 0 : ))
666 0 : }
667 :
668 : // Note: it is not documented if this should be the total or remaining length, total passes the
669 : // tests.
670 0 : fn len(&self) -> usize {
671 0 : use NonSeekableStream::*;
672 0 : match self {
673 0 : Initial { len, .. } => *len,
674 0 : Actual { len, .. } => *len,
675 0 : Cloned { len_was, .. } => *len_was,
676 : }
677 0 : }
678 : }
|