mountpoint_s3_client/s3_crt_client/
get_object.rs1use std::ops::Deref;
2use std::os::unix::prelude::OsStrExt;
3use std::pin::Pin;
4use std::sync::Arc;
5use std::sync::atomic::{AtomicU64, Ordering};
6use std::task::{Context, Poll};
7
8use async_trait::async_trait;
9use bytes::Bytes;
10use futures::channel::mpsc::UnboundedReceiver;
11use futures::stream::FusedStream;
12use futures::{Stream, StreamExt};
13use mountpoint_s3_crt::http::request_response::{Header, Headers};
14use mountpoint_s3_crt::s3::client::{MetaRequest, MetaRequestResult};
15use pin_project::pin_project;
16use tracing::trace;
17
18use crate::error_metadata::ClientErrorMetadata;
19use crate::object_client::{
20 Checksum, ChecksumMode, ClientBackpressureHandle, GetBodyPart, GetObjectError, GetObjectParams, GetObjectResponse,
21 ObjectChecksumError, ObjectClientError, ObjectClientResult, ObjectMetadata,
22};
23
24use super::{CancellingMetaRequest, ResponseHeadersError, S3CrtClient, S3Operation, S3RequestError, parse_checksum};
25
26impl S3CrtClient {
27 pub(super) async fn get_object(
30 &self,
31 bucket: &str,
32 key: &str,
33 params: &GetObjectParams,
34 ) -> Result<S3GetObjectResponse, ObjectClientError<GetObjectError, S3RequestError>> {
35 let requested_checksums = params.checksum_mode.as_ref() == Some(&ChecksumMode::Enabled);
36 let next_offset = params.range.as_ref().map(|r| r.start).unwrap_or(0);
37 let (event_sender, mut event_receiver) = futures::channel::mpsc::unbounded();
38 let meta_request = {
39 let span =
40 request_span!(self.inner, "get_object", bucket, key, range=?params.range, if_match=?params.if_match);
41
42 let mut message = self
43 .inner
44 .new_request_template("GET", bucket)
45 .map_err(S3RequestError::construction_failure)?;
46
47 message
49 .set_header(&Header::new("accept", "*/*"))
50 .map_err(S3RequestError::construction_failure)?;
51
52 if requested_checksums {
53 message
55 .set_header(&Header::new("x-amz-checksum-mode", "enabled"))
56 .map_err(S3RequestError::construction_failure)?;
57 }
58
59 if let Some(etag) = params.if_match.as_ref() {
60 message
62 .set_header(&Header::new("If-Match", etag.as_str()))
63 .map_err(S3RequestError::construction_failure)?;
64 }
65
66 if let Some(range) = params.range.as_ref() {
67 let range_value = format!("bytes={}-{}", range.start, range.end.saturating_sub(1));
69 message
70 .set_header(&Header::new("Range", range_value))
71 .map_err(S3RequestError::construction_failure)?;
72 }
73
74 let key = format!("/{key}");
75 message
76 .set_request_path(key)
77 .map_err(S3RequestError::construction_failure)?;
78
79 let mut options = message.into_options(S3Operation::GetObject);
80 options.part_size(self.inner.read_part_size as u64);
81
82 let mut headers_sender = Some(event_sender.clone());
83 let part_sender = event_sender.clone();
84
85 self.inner.meta_request_with_callbacks(
86 options,
87 span,
88 |_| (),
89 move |headers, status| {
90 if (200..300).contains(&status) {
93 if let Some(headers_sender) = headers_sender.take() {
96 _ = headers_sender.unbounded_send(S3GetObjectEvent::Headers(headers.clone()));
97 }
98 }
99 },
100 move |offset, data| {
101 let owned_buffer = data
102 .to_owned_buffer()
103 .expect("buffers returned from GetObject can always be acquired");
104 let bytes = Bytes::from_owner(owned_buffer);
105 let body_part = GetBodyPart { offset, data: bytes };
106 _ = part_sender.unbounded_send(S3GetObjectEvent::BodyPart(body_part));
107 },
108 parse_get_object_error,
109 move |result| {
110 if let Err(e) = result {
111 _ = event_sender.unbounded_send(S3GetObjectEvent::Error(e));
112 }
113 event_sender.close_channel();
114 },
115 )?
116 };
117
118 let headers = match event_receiver.next().await {
119 Some(S3GetObjectEvent::Headers(headers)) => headers,
120 Some(S3GetObjectEvent::Error(e)) => {
121 return Err(e);
122 }
123 event => {
124 trace!(?event, "unexpected GetObject event while waiting for headers");
126 return Err(S3RequestError::internal_failure(ResponseHeadersError::MissingHeaders).into());
127 }
128 };
129
130 let backpressure_handle = if self.inner.enable_backpressure {
131 let read_window_end_offset =
132 Arc::new(AtomicU64::new(next_offset + self.inner.initial_read_window_size as u64));
133 Some(S3BackpressureHandle {
134 read_window_end_offset,
135 meta_request: meta_request.clone(),
136 })
137 } else {
138 None
139 };
140 Ok(S3GetObjectResponse {
141 meta_request,
142 event_receiver,
143 requested_checksums,
144 backpressure_handle,
145 headers,
146 next_offset,
147 })
148 }
149}
150
151#[derive(Debug)]
152enum S3GetObjectEvent {
153 Headers(Headers),
154 BodyPart(GetBodyPart),
155 Error(ObjectClientError<GetObjectError, S3RequestError>),
156}
157
158#[derive(Clone, Debug)]
159pub struct S3BackpressureHandle {
160 read_window_end_offset: Arc<AtomicU64>,
163 meta_request: MetaRequest,
164}
165
166impl ClientBackpressureHandle for S3BackpressureHandle {
167 fn increment_read_window(&mut self, len: usize) {
168 self.read_window_end_offset.fetch_add(len as u64, Ordering::SeqCst);
169 self.meta_request.increment_read_window(len as u64);
170 }
171
172 fn ensure_read_window(&mut self, desired_end_offset: u64) {
173 let diff = desired_end_offset.saturating_sub(self.read_window_end_offset()) as usize;
174 self.increment_read_window(diff);
175 }
176
177 fn read_window_end_offset(&self) -> u64 {
178 self.read_window_end_offset.load(Ordering::SeqCst)
179 }
180}
181
182#[derive(Debug)]
188#[pin_project]
189pub struct S3GetObjectResponse {
190 meta_request: CancellingMetaRequest,
191 #[pin]
192 event_receiver: UnboundedReceiver<S3GetObjectEvent>,
193 requested_checksums: bool,
194 backpressure_handle: Option<S3BackpressureHandle>,
195 headers: Headers,
196 next_offset: u64,
198}
199
200#[cfg_attr(not(docsrs), async_trait)]
201impl GetObjectResponse for S3GetObjectResponse {
202 type BackpressureHandle = S3BackpressureHandle;
203 type ClientError = S3RequestError;
204
205 fn backpressure_handle(&mut self) -> Option<&mut Self::BackpressureHandle> {
206 self.backpressure_handle.as_mut()
207 }
208
209 fn get_object_metadata(&self) -> ObjectMetadata {
210 self.headers
211 .iter()
212 .filter_map(|(key, value)| {
213 let metadata_header = key.to_str()?.strip_prefix("x-amz-meta-")?;
214 let value = value.to_str()?;
215 Some((metadata_header.to_string(), value.to_string()))
216 })
217 .collect()
218 }
219
220 fn get_object_checksum(&self) -> Result<Checksum, ObjectChecksumError> {
221 if !self.requested_checksums {
222 return Err(ObjectChecksumError::DidNotRequestChecksums);
223 }
224
225 parse_checksum(&self.headers).map_err(|e| ObjectChecksumError::HeadersError(Box::new(e)))
226 }
227}
228
229impl Stream for S3GetObjectResponse {
230 type Item = ObjectClientResult<GetBodyPart, GetObjectError, S3RequestError>;
231
232 fn poll_next(self: Pin<&mut Self>, cx: &mut Context) -> Poll<Option<Self::Item>> {
233 if self.event_receiver.is_terminated() {
234 return Poll::Ready(None);
235 }
236
237 let this = self.project();
238 match this.event_receiver.poll_next(cx) {
239 Poll::Ready(None) => Poll::Ready(None),
240 Poll::Ready(Some(S3GetObjectEvent::BodyPart(part))) => {
241 *this.next_offset = part.offset + part.data.len() as u64;
242 Poll::Ready(Some(Ok(part)))
243 }
244 Poll::Ready(Some(S3GetObjectEvent::Headers(_))) => {
245 unreachable!("headers are only sent once and received before returning the stream")
246 }
247 Poll::Ready(Some(S3GetObjectEvent::Error(e))) => Poll::Ready(Some(Err(e))),
248 Poll::Pending => {
249 if let Some(handle) = &this.backpressure_handle
254 && *this.next_offset >= handle.read_window_end_offset()
255 {
256 let err = ObjectClientError::from(S3RequestError::EmptyReadWindow);
257 return Poll::Ready(Some(Err(err)));
258 }
259 Poll::Pending
260 }
261 }
262 }
263}
264
265fn parse_get_object_error(result: &MetaRequestResult) -> Option<GetObjectError> {
266 let client_error_metadata = ClientErrorMetadata::from_meta_request_result(result);
267 match result.response_status {
268 404 => {
269 let body = result.error_response_body.as_ref()?;
270 let root = xmltree::Element::parse(body.as_bytes()).ok()?;
271 let error_code = root.get_child("Code")?;
272 let error_str = error_code.get_text()?;
273 match error_str.deref() {
274 "NoSuchBucket" => Some(GetObjectError::NoSuchBucket(client_error_metadata)),
275 "NoSuchKey" => Some(GetObjectError::NoSuchKey(client_error_metadata)),
276 _ => None,
277 }
278 }
279 412 => Some(GetObjectError::PreconditionFailed(client_error_metadata)),
280 _ => None,
281 }
282}
283
284#[cfg(test)]
285mod tests {
286 use std::ffi::{OsStr, OsString};
287
288 use super::*;
289
290 fn make_result(response_status: i32, body: impl Into<OsString>) -> MetaRequestResult {
291 MetaRequestResult {
292 response_status,
293 crt_error: 1i32.into(),
294 error_response_headers: None,
295 error_response_body: Some(body.into()),
296 }
297 }
298
299 #[test]
300 fn parse_404_no_such_key() {
301 let body = br#"<?xml version="1.0" encoding="UTF-8"?><Error><Code>NoSuchKey</Code><Message>The specified key does not exist.</Message><Key>not-a-real-key</Key><RequestId>NTKJWKHQBYNS73A9</RequestId><HostId>Nc9kWNrf4kGoq5NIUnQ4t7u04ZZXGm/i463v+jwCI8sIrZBqeYI8uffLHQ+/qusdMWNuUwqeXHU=</HostId></Error>"#;
302 let result = make_result(404, OsStr::from_bytes(&body[..]));
303 let result = parse_get_object_error(&result);
304 assert!(matches!(result, Some(GetObjectError::NoSuchKey(_))));
305 }
306
307 #[test]
308 fn parse_404_no_such_bucket() {
309 let body = br#"<?xml version="1.0" encoding="UTF-8"?><Error><Code>NoSuchBucket</Code><Message>The specified bucket does not exist</Message><BucketName>amzn-s3-demo-bucket</BucketName><RequestId>4VAGDP5HMYTDNB3Y</RequestId><HostId>JMgGqpVKIaaTieG68IODiV2piWw/q9VCTowGvWP36BEz6oIVEXiesn8cDE5ph7if0gpY5WU1Wc8=</HostId></Error>"#;
310 let result = make_result(404, OsStr::from_bytes(&body[..]));
311 let result = parse_get_object_error(&result);
312 assert!(matches!(result, Some(GetObjectError::NoSuchBucket(_))));
313 }
314
315 #[test]
316 fn parse_403_glacier_storage_class() {
317 let body = br#"<?xml version="1.0" encoding="UTF-8"?><Error><Code>InvalidObjectState</Code><Message>The action is not valid for the object's storage class</Message><RequestId>9FEFFF118E15B86F</RequestId><HostId>WVQ5kzhiT+oiUfDCOiOYv8W4Tk9eNcxWi/MK+hTS/av34Xy4rBU3zsavf0aaaaa</HostId></Error>"#;
318 let result = make_result(403, OsStr::from_bytes(&body[..]));
319 let result = parse_get_object_error(&result);
320 assert_eq!(result, None);
321 }
322}