opendal/services/azblob/
core.rs

1// Licensed to the Apache Software Foundation (ASF) under one
2// or more contributor license agreements.  See the NOTICE file
3// distributed with this work for additional information
4// regarding copyright ownership.  The ASF licenses this file
5// to you under the Apache License, Version 2.0 (the
6// "License"); you may not use this file except in compliance
7// with the License.  You may obtain a copy of the License at
8//
9//   http://www.apache.org/licenses/LICENSE-2.0
10//
11// Unless required by applicable law or agreed to in writing,
12// software distributed under the License is distributed on an
13// "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
14// KIND, either express or implied.  See the License for the
15// specific language governing permissions and limitations
16// under the License.
17
18use std::fmt;
19use std::fmt::Debug;
20use std::fmt::Formatter;
21use std::sync::Arc;
22use std::time::Duration;
23
24use base64::prelude::BASE64_STANDARD;
25use base64::Engine;
26use bytes::Bytes;
27use constants::X_MS_META_PREFIX;
28use http::header::HeaderName;
29use http::header::CONTENT_LENGTH;
30use http::header::CONTENT_TYPE;
31use http::header::IF_MATCH;
32use http::header::IF_MODIFIED_SINCE;
33use http::header::IF_NONE_MATCH;
34use http::header::IF_UNMODIFIED_SINCE;
35use http::HeaderValue;
36use http::Request;
37use http::Response;
38use reqsign::AzureStorageCredential;
39use reqsign::AzureStorageLoader;
40use reqsign::AzureStorageSigner;
41use serde::Deserialize;
42use serde::Serialize;
43use uuid::Uuid;
44
45use crate::raw::*;
46use crate::*;
47
48pub mod constants {
49    // Indicates the Blob Storage version that was used to execute the request
50    pub const X_MS_VERSION: &str = "x-ms-version";
51
52    pub const X_MS_BLOB_TYPE: &str = "x-ms-blob-type";
53    pub const X_MS_COPY_SOURCE: &str = "x-ms-copy-source";
54    pub const X_MS_BLOB_CACHE_CONTROL: &str = "x-ms-blob-cache-control";
55    pub const X_MS_BLOB_CONDITION_APPENDPOS: &str = "x-ms-blob-condition-appendpos";
56    pub const X_MS_META_PREFIX: &str = "x-ms-meta-";
57
58    // indicates the version of the blob, and it can be used in subsequent requests to access the blob.
59    pub const X_MS_VERSION_ID: &str = "x-ms-version-id";
60
61    // Server-side encryption with customer-provided headers
62    pub const X_MS_ENCRYPTION_KEY: &str = "x-ms-encryption-key";
63    pub const X_MS_ENCRYPTION_KEY_SHA256: &str = "x-ms-encryption-key-sha256";
64    pub const X_MS_ENCRYPTION_ALGORITHM: &str = "x-ms-encryption-algorithm";
65}
66
67pub struct AzblobCore {
68    pub info: Arc<AccessorInfo>,
69    pub container: String,
70    pub root: String,
71    pub endpoint: String,
72    pub encryption_key: Option<HeaderValue>,
73    pub encryption_key_sha256: Option<HeaderValue>,
74    pub encryption_algorithm: Option<HeaderValue>,
75    pub loader: AzureStorageLoader,
76    pub signer: AzureStorageSigner,
77}
78
79impl Debug for AzblobCore {
80    fn fmt(&self, f: &mut Formatter<'_>) -> fmt::Result {
81        f.debug_struct("AzblobCore")
82            .field("container", &self.container)
83            .field("root", &self.root)
84            .field("endpoint", &self.endpoint)
85            .finish_non_exhaustive()
86    }
87}
88
89impl AzblobCore {
90    async fn load_credential(&self) -> Result<AzureStorageCredential> {
91        let cred = self
92            .loader
93            .load()
94            .await
95            .map_err(new_request_credential_error)?;
96
97        if let Some(cred) = cred {
98            Ok(cred)
99        } else {
100            Err(Error::new(
101                ErrorKind::ConfigInvalid,
102                "no valid credential found",
103            ))
104        }
105    }
106
107    pub async fn sign_query<T>(&self, req: &mut Request<T>) -> Result<()> {
108        let cred = self.load_credential().await?;
109
110        self.signer
111            .sign_query(req, Duration::from_secs(3600), &cred)
112            .map_err(new_request_sign_error)
113    }
114
115    pub async fn sign<T>(&self, req: &mut Request<T>) -> Result<()> {
116        let cred = self.load_credential().await?;
117        // Insert x-ms-version header for normal requests.
118        req.headers_mut().insert(
119            HeaderName::from_static(constants::X_MS_VERSION),
120            // 2022-11-02 is the version supported by Azurite V3 and
121            // used by Azure Portal, We use this version to make
122            // sure most our developer happy.
123            //
124            // In the future, we could allow users to configure this value.
125            HeaderValue::from_static("2022-11-02"),
126        );
127        self.signer.sign(req, &cred).map_err(new_request_sign_error)
128    }
129
130    async fn batch_sign<T>(&self, req: &mut Request<T>) -> Result<()> {
131        let cred = self.load_credential().await?;
132        self.signer.sign(req, &cred).map_err(new_request_sign_error)
133    }
134
135    #[inline]
136    pub async fn send(&self, req: Request<Buffer>) -> Result<Response<Buffer>> {
137        self.info.http_client().send(req).await
138    }
139
140    pub fn insert_sse_headers(&self, mut req: http::request::Builder) -> http::request::Builder {
141        if let Some(v) = &self.encryption_key {
142            let mut v = v.clone();
143            v.set_sensitive(true);
144
145            req = req.header(HeaderName::from_static(constants::X_MS_ENCRYPTION_KEY), v)
146        }
147
148        if let Some(v) = &self.encryption_key_sha256 {
149            let mut v = v.clone();
150            v.set_sensitive(true);
151
152            req = req.header(
153                HeaderName::from_static(constants::X_MS_ENCRYPTION_KEY_SHA256),
154                v,
155            )
156        }
157
158        if let Some(v) = &self.encryption_algorithm {
159            let mut v = v.clone();
160            v.set_sensitive(true);
161
162            req = req.header(
163                HeaderName::from_static(constants::X_MS_ENCRYPTION_ALGORITHM),
164                v,
165            )
166        }
167
168        req
169    }
170}
171
172impl AzblobCore {
173    fn build_path_url(&self, path: &str) -> String {
174        format!(
175            "{}/{}/{}",
176            self.endpoint,
177            self.container,
178            percent_encode_path(&build_abs_path(&self.root, path))
179        )
180    }
181
182    pub fn azblob_get_blob_request(
183        &self,
184        path: &str,
185        range: BytesRange,
186        args: &OpRead,
187    ) -> Result<Request<Buffer>> {
188        let mut url = self.build_path_url(path);
189
190        if let Some(override_content_disposition) = args.override_content_disposition() {
191            url.push_str(&format!(
192                "?rscd={}",
193                percent_encode_path(override_content_disposition)
194            ));
195        }
196
197        let mut req = Request::get(&url);
198
199        // Set SSE headers.
200        req = self.insert_sse_headers(req);
201
202        if !range.is_full() {
203            req = req.header(http::header::RANGE, range.to_header());
204        }
205
206        if let Some(if_none_match) = args.if_none_match() {
207            req = req.header(IF_NONE_MATCH, if_none_match);
208        }
209
210        if let Some(if_match) = args.if_match() {
211            req = req.header(IF_MATCH, if_match);
212        }
213
214        if let Some(if_modified_since) = args.if_modified_since() {
215            req = req.header(
216                IF_MODIFIED_SINCE,
217                format_datetime_into_http_date(if_modified_since),
218            );
219        }
220
221        if let Some(if_unmodified_since) = args.if_unmodified_since() {
222            req = req.header(
223                IF_UNMODIFIED_SINCE,
224                format_datetime_into_http_date(if_unmodified_since),
225            );
226        }
227
228        let req = req.body(Buffer::new()).map_err(new_request_build_error)?;
229
230        Ok(req)
231    }
232
233    pub async fn azblob_get_blob(
234        &self,
235        path: &str,
236        range: BytesRange,
237        args: &OpRead,
238    ) -> Result<Response<HttpBody>> {
239        let mut req = self.azblob_get_blob_request(path, range, args)?;
240
241        self.sign(&mut req).await?;
242
243        self.info.http_client().fetch(req).await
244    }
245
246    pub fn azblob_put_blob_request(
247        &self,
248        path: &str,
249        size: Option<u64>,
250        args: &OpWrite,
251        body: Buffer,
252    ) -> Result<Request<Buffer>> {
253        let mut req = Request::put(self.build_path_url(path));
254
255        req = req.header(
256            HeaderName::from_static(constants::X_MS_BLOB_TYPE),
257            "BlockBlob",
258        );
259
260        if let Some(size) = size {
261            req = req.header(CONTENT_LENGTH, size)
262        }
263
264        if let Some(ty) = args.content_type() {
265            req = req.header(CONTENT_TYPE, ty)
266        }
267
268        // Specify the wildcard character (*) to perform the operation only if
269        // the resource does not exist, and fail the operation if it does exist.
270        if args.if_not_exists() {
271            req = req.header(IF_NONE_MATCH, "*");
272        }
273
274        if let Some(v) = args.if_none_match() {
275            req = req.header(IF_NONE_MATCH, v);
276        }
277
278        if let Some(cache_control) = args.cache_control() {
279            req = req.header(constants::X_MS_BLOB_CACHE_CONTROL, cache_control);
280        }
281
282        // Set SSE headers.
283        req = self.insert_sse_headers(req);
284
285        if let Some(user_metadata) = args.user_metadata() {
286            for (key, value) in user_metadata {
287                req = req.header(format!("{X_MS_META_PREFIX}{key}"), value)
288            }
289        }
290
291        // Set body
292        let req = req.body(body).map_err(new_request_build_error)?;
293
294        Ok(req)
295    }
296
297    pub async fn azblob_put_blob(
298        &self,
299        path: &str,
300        size: Option<u64>,
301        args: &OpWrite,
302        body: Buffer,
303    ) -> Result<Response<Buffer>> {
304        let mut req = self.azblob_put_blob_request(path, size, args, body)?;
305
306        self.sign(&mut req).await?;
307        self.send(req).await
308    }
309
310    /// For appendable object, it could be created by `put` an empty blob
311    /// with `x-ms-blob-type` header set to `AppendBlob`.
312    /// And it's just initialized with empty content.
313    ///
314    /// If want to append content to it, we should use the following method `azblob_append_blob_request`.
315    ///
316    /// # Notes
317    ///
318    /// Appendable blob's custom header could only be set when it's created.
319    ///
320    /// The following custom header could be set:
321    /// - `content-type`
322    /// - `x-ms-blob-cache-control`
323    ///
324    /// # Reference
325    ///
326    /// https://learn.microsoft.com/en-us/rest/api/storageservices/put-blob
327    fn azblob_init_appendable_blob_request(
328        &self,
329        path: &str,
330        args: &OpWrite,
331    ) -> Result<Request<Buffer>> {
332        let mut req = Request::put(self.build_path_url(path));
333
334        // Set SSE headers.
335        req = self.insert_sse_headers(req);
336
337        // The content-length header must be set to zero
338        // when creating an appendable blob.
339        req = req.header(CONTENT_LENGTH, 0);
340        req = req.header(
341            HeaderName::from_static(constants::X_MS_BLOB_TYPE),
342            "AppendBlob",
343        );
344
345        if let Some(ty) = args.content_type() {
346            req = req.header(CONTENT_TYPE, ty)
347        }
348
349        if let Some(cache_control) = args.cache_control() {
350            req = req.header(constants::X_MS_BLOB_CACHE_CONTROL, cache_control);
351        }
352
353        let req = req.body(Buffer::new()).map_err(new_request_build_error)?;
354
355        Ok(req)
356    }
357
358    pub async fn azblob_init_appendable_blob(
359        &self,
360        path: &str,
361        args: &OpWrite,
362    ) -> Result<Response<Buffer>> {
363        let mut req = self.azblob_init_appendable_blob_request(path, args)?;
364
365        self.sign(&mut req).await?;
366        self.send(req).await
367    }
368
369    /// Append content to an appendable blob.
370    /// The content will be appended to the end of the blob.
371    ///
372    /// # Notes
373    ///
374    /// - The maximum size of the content could be appended is 4MB.
375    /// - `Append Block` succeeds only if the blob already exists.
376    ///
377    /// # Reference
378    ///
379    /// https://learn.microsoft.com/en-us/rest/api/storageservices/append-block
380    fn azblob_append_blob_request(
381        &self,
382        path: &str,
383        position: u64,
384        size: u64,
385        body: Buffer,
386    ) -> Result<Request<Buffer>> {
387        let url = format!("{}?comp=appendblock", &self.build_path_url(path));
388
389        let mut req = Request::put(&url);
390
391        // Set SSE headers.
392        req = self.insert_sse_headers(req);
393
394        req = req.header(CONTENT_LENGTH, size);
395
396        req = req.header(constants::X_MS_BLOB_CONDITION_APPENDPOS, position);
397
398        let req = req.body(body).map_err(new_request_build_error)?;
399
400        Ok(req)
401    }
402
403    pub async fn azblob_append_blob(
404        &self,
405        path: &str,
406        position: u64,
407        size: u64,
408        body: Buffer,
409    ) -> Result<Response<Buffer>> {
410        let mut req = self.azblob_append_blob_request(path, position, size, body)?;
411
412        self.sign(&mut req).await?;
413        self.send(req).await
414    }
415
416    pub fn azblob_put_block_request(
417        &self,
418        path: &str,
419        block_id: Uuid,
420        size: Option<u64>,
421        args: &OpWrite,
422        body: Buffer,
423    ) -> Result<Request<Buffer>> {
424        // To be written as part of a blob, a block must have been successfully written to the server in an earlier Put Block operation.
425        // refer to https://learn.microsoft.com/en-us/rest/api/storageservices/put-block?tabs=microsoft-entra-id
426        let url = QueryPairsWriter::new(&self.build_path_url(path))
427            .push("comp", "block")
428            .push(
429                "blockid",
430                &percent_encode_path(&BASE64_STANDARD.encode(block_id.as_bytes())),
431            )
432            .finish();
433
434        let mut req = Request::put(&url);
435        // Set SSE headers.
436        req = self.insert_sse_headers(req);
437
438        if let Some(cache_control) = args.cache_control() {
439            req = req.header(constants::X_MS_BLOB_CACHE_CONTROL, cache_control);
440        }
441        if let Some(size) = size {
442            req = req.header(CONTENT_LENGTH, size)
443        }
444
445        if let Some(ty) = args.content_type() {
446            req = req.header(CONTENT_TYPE, ty)
447        }
448        // Set body
449        let req = req.body(body).map_err(new_request_build_error)?;
450
451        Ok(req)
452    }
453
454    pub async fn azblob_put_block(
455        &self,
456        path: &str,
457        block_id: Uuid,
458        size: Option<u64>,
459        args: &OpWrite,
460        body: Buffer,
461    ) -> Result<Response<Buffer>> {
462        let mut req = self.azblob_put_block_request(path, block_id, size, args, body)?;
463
464        self.sign(&mut req).await?;
465        self.send(req).await
466    }
467
468    fn azblob_complete_put_block_list_request(
469        &self,
470        path: &str,
471        block_ids: Vec<Uuid>,
472        args: &OpWrite,
473    ) -> Result<Request<Buffer>> {
474        let url = format!("{}?comp=blocklist", &self.build_path_url(path));
475
476        let req = Request::put(&url);
477
478        // Set SSE headers.
479        let mut req = self.insert_sse_headers(req);
480        if let Some(cache_control) = args.cache_control() {
481            req = req.header(constants::X_MS_BLOB_CACHE_CONTROL, cache_control);
482        }
483
484        let content = quick_xml::se::to_string(&PutBlockListRequest {
485            latest: block_ids
486                .into_iter()
487                .map(|block_id| {
488                    let encoded_block_id: String = BASE64_STANDARD.encode(block_id.as_bytes());
489                    encoded_block_id
490                })
491                .collect(),
492        })
493        .map_err(new_xml_serialize_error)?;
494
495        req = req.header(CONTENT_LENGTH, content.len());
496
497        let req = req
498            .body(Buffer::from(Bytes::from(content)))
499            .map_err(new_request_build_error)?;
500
501        Ok(req)
502    }
503
504    pub async fn azblob_complete_put_block_list(
505        &self,
506        path: &str,
507        block_ids: Vec<Uuid>,
508        args: &OpWrite,
509    ) -> Result<Response<Buffer>> {
510        let mut req = self.azblob_complete_put_block_list_request(path, block_ids, args)?;
511
512        self.sign(&mut req).await?;
513
514        self.send(req).await
515    }
516
517    pub fn azblob_head_blob_request(&self, path: &str, args: &OpStat) -> Result<Request<Buffer>> {
518        let mut req = Request::head(self.build_path_url(path));
519
520        // Set SSE headers.
521        req = self.insert_sse_headers(req);
522
523        if let Some(if_none_match) = args.if_none_match() {
524            req = req.header(IF_NONE_MATCH, if_none_match);
525        }
526
527        if let Some(if_match) = args.if_match() {
528            req = req.header(IF_MATCH, if_match);
529        }
530
531        let req = req.body(Buffer::new()).map_err(new_request_build_error)?;
532
533        Ok(req)
534    }
535
536    pub async fn azblob_get_blob_properties(
537        &self,
538        path: &str,
539        args: &OpStat,
540    ) -> Result<Response<Buffer>> {
541        let mut req = self.azblob_head_blob_request(path, args)?;
542
543        self.sign(&mut req).await?;
544        self.send(req).await
545    }
546
547    fn azblob_delete_blob_request(&self, path: &str) -> Result<Request<Buffer>> {
548        let req = Request::delete(self.build_path_url(path));
549
550        req.header(CONTENT_LENGTH, 0)
551            .body(Buffer::new())
552            .map_err(new_request_build_error)
553    }
554
555    pub async fn azblob_delete_blob(&self, path: &str) -> Result<Response<Buffer>> {
556        let mut req = self.azblob_delete_blob_request(path)?;
557
558        self.sign(&mut req).await?;
559        self.send(req).await
560    }
561
562    pub async fn azblob_copy_blob(&self, from: &str, to: &str) -> Result<Response<Buffer>> {
563        let source = self.build_path_url(from);
564        let target = self.build_path_url(to);
565
566        let mut req = Request::put(&target)
567            .header(constants::X_MS_COPY_SOURCE, source)
568            .header(CONTENT_LENGTH, 0)
569            .body(Buffer::new())
570            .map_err(new_request_build_error)?;
571
572        self.sign(&mut req).await?;
573        self.send(req).await
574    }
575
576    pub async fn azblob_list_blobs(
577        &self,
578        path: &str,
579        next_marker: &str,
580        delimiter: &str,
581        limit: Option<usize>,
582    ) -> Result<Response<Buffer>> {
583        let p = build_abs_path(&self.root, path);
584        let mut url = QueryPairsWriter::new(&format!("{}/{}", self.endpoint, self.container))
585            .push("restype", "container")
586            .push("comp", "list");
587
588        if !p.is_empty() {
589            url = url.push("prefix", &percent_encode_path(&p));
590        }
591        if let Some(limit) = limit {
592            url = url.push("maxresults", &limit.to_string());
593        }
594        if !delimiter.is_empty() {
595            url = url.push("delimiter", delimiter);
596        }
597        if !next_marker.is_empty() {
598            url = url.push("marker", next_marker);
599        }
600
601        let mut req = Request::get(url.finish())
602            .body(Buffer::new())
603            .map_err(new_request_build_error)?;
604
605        self.sign(&mut req).await?;
606        self.send(req).await
607    }
608
609    pub async fn azblob_batch_delete(&self, paths: &[String]) -> Result<Response<Buffer>> {
610        let url = format!(
611            "{}/{}?restype=container&comp=batch",
612            self.endpoint, self.container
613        );
614
615        let mut multipart = Multipart::new();
616
617        for (idx, path) in paths.iter().enumerate() {
618            let mut req = self.azblob_delete_blob_request(path)?;
619            self.batch_sign(&mut req).await?;
620
621            multipart = multipart.part(
622                MixedPart::from_request(req).part_header("content-id".parse().unwrap(), idx.into()),
623            );
624        }
625
626        let req = Request::post(url);
627        let mut req = multipart.apply(req)?;
628
629        self.sign(&mut req).await?;
630        self.send(req).await
631    }
632}
633
634/// Request of PutBlockListRequest
635#[derive(Default, Debug, Serialize, Deserialize)]
636#[serde(default, rename = "BlockList", rename_all = "PascalCase")]
637pub struct PutBlockListRequest {
638    pub latest: Vec<String>,
639}
640
641#[derive(Default, Debug, Deserialize)]
642#[serde(default, rename_all = "PascalCase")]
643pub struct ListBlobsOutput {
644    pub blobs: Blobs,
645    #[serde(rename = "NextMarker")]
646    pub next_marker: Option<String>,
647}
648
649#[derive(Default, Debug, Deserialize)]
650#[serde(default, rename_all = "PascalCase")]
651pub struct Blobs {
652    pub blob: Vec<Blob>,
653    pub blob_prefix: Vec<BlobPrefix>,
654}
655
656#[derive(Default, Debug, Deserialize)]
657#[serde(default, rename_all = "PascalCase")]
658pub struct BlobPrefix {
659    pub name: String,
660}
661
662#[derive(Default, Debug, Deserialize)]
663#[serde(default, rename_all = "PascalCase")]
664pub struct Blob {
665    pub properties: Properties,
666    pub name: String,
667}
668
669#[derive(Default, Debug, Deserialize)]
670#[serde(default, rename_all = "PascalCase")]
671pub struct Properties {
672    #[serde(rename = "Content-Length")]
673    pub content_length: u64,
674    #[serde(rename = "Last-Modified")]
675    pub last_modified: String,
676    #[serde(rename = "Content-MD5")]
677    pub content_md5: String,
678    #[serde(rename = "Content-Type")]
679    pub content_type: String,
680    pub etag: String,
681}
682
683#[cfg(test)]
684mod tests {
685    use bytes::Buf;
686    use bytes::Bytes;
687    use quick_xml::de;
688
689    use super::*;
690
691    #[test]
692    fn test_parse_xml() {
693        let bs = bytes::Bytes::from(
694            r#"
695            <?xml version="1.0" encoding="utf-8"?>
696            <EnumerationResults ServiceEndpoint="https://test.blob.core.windows.net/" ContainerName="myazurebucket">
697                <Prefix>dir1/</Prefix>
698                <Delimiter>/</Delimiter>
699                <Blobs>
700                    <Blob>
701                        <Name>dir1/2f018bb5-466f-4af1-84fa-2b167374ee06</Name>
702                        <Properties>
703                            <Creation-Time>Sun, 20 Mar 2022 11:29:03 GMT</Creation-Time>
704                            <Last-Modified>Sun, 20 Mar 2022 11:29:03 GMT</Last-Modified>
705                            <Etag>0x8DA0A64D66790C3</Etag>
706                            <Content-Length>3485277</Content-Length>
707                            <Content-Type>application/octet-stream</Content-Type>
708                            <Content-Encoding />
709                            <Content-Language />
710                            <Content-CRC64 />
711                            <Content-MD5>llJ/+jOlx5GdA1sL7SdKuw==</Content-MD5>
712                            <Cache-Control />
713                            <Content-Disposition />
714                            <BlobType>BlockBlob</BlobType>
715                            <AccessTier>Hot</AccessTier>
716                            <AccessTierInferred>true</AccessTierInferred>
717                            <LeaseStatus>unlocked</LeaseStatus>
718                            <LeaseState>available</LeaseState>
719                            <ServerEncrypted>true</ServerEncrypted>
720                        </Properties>
721                        <OrMetadata />
722                    </Blob>
723                    <Blob>
724                        <Name>dir1/5b9432b2-79c0-48d8-90c2-7d3e153826ed</Name>
725                        <Properties>
726                            <Creation-Time>Tue, 29 Mar 2022 01:54:07 GMT</Creation-Time>
727                            <Last-Modified>Tue, 29 Mar 2022 01:54:07 GMT</Last-Modified>
728                            <Etag>0x8DA112702D88FE4</Etag>
729                            <Content-Length>2471869</Content-Length>
730                            <Content-Type>application/octet-stream</Content-Type>
731                            <Content-Encoding />
732                            <Content-Language />
733                            <Content-CRC64 />
734                            <Content-MD5>xmgUltSnopLSJOukgCHFtg==</Content-MD5>
735                            <Cache-Control />
736                            <Content-Disposition />
737                            <BlobType>BlockBlob</BlobType>
738                            <AccessTier>Hot</AccessTier>
739                            <AccessTierInferred>true</AccessTierInferred>
740                            <LeaseStatus>unlocked</LeaseStatus>
741                            <LeaseState>available</LeaseState>
742                            <ServerEncrypted>true</ServerEncrypted>
743                        </Properties>
744                        <OrMetadata />
745                    </Blob>
746                    <Blob>
747                        <Name>dir1/b2d96f8b-d467-40d1-bb11-4632dddbf5b5</Name>
748                        <Properties>
749                            <Creation-Time>Sun, 20 Mar 2022 11:31:57 GMT</Creation-Time>
750                            <Last-Modified>Sun, 20 Mar 2022 11:31:57 GMT</Last-Modified>
751                            <Etag>0x8DA0A653DC82981</Etag>
752                            <Content-Length>1259677</Content-Length>
753                            <Content-Type>application/octet-stream</Content-Type>
754                            <Content-Encoding />
755                            <Content-Language />
756                            <Content-CRC64 />
757                            <Content-MD5>AxTiFXHwrXKaZC5b7ZRybw==</Content-MD5>
758                            <Cache-Control />
759                            <Content-Disposition />
760                            <BlobType>BlockBlob</BlobType>
761                            <AccessTier>Hot</AccessTier>
762                            <AccessTierInferred>true</AccessTierInferred>
763                            <LeaseStatus>unlocked</LeaseStatus>
764                            <LeaseState>available</LeaseState>
765                            <ServerEncrypted>true</ServerEncrypted>
766                        </Properties>
767                        <OrMetadata />
768                    </Blob>
769                    <BlobPrefix>
770                        <Name>dir1/dir2/</Name>
771                    </BlobPrefix>
772                    <BlobPrefix>
773                        <Name>dir1/dir21/</Name>
774                    </BlobPrefix>
775                </Blobs>
776                <NextMarker />
777            </EnumerationResults>"#,
778        );
779        let out: ListBlobsOutput = de::from_reader(bs.reader()).expect("must success");
780        println!("{out:?}");
781
782        assert_eq!(
783            out.blobs
784                .blob
785                .iter()
786                .map(|v| v.name.clone())
787                .collect::<Vec<String>>(),
788            [
789                "dir1/2f018bb5-466f-4af1-84fa-2b167374ee06",
790                "dir1/5b9432b2-79c0-48d8-90c2-7d3e153826ed",
791                "dir1/b2d96f8b-d467-40d1-bb11-4632dddbf5b5"
792            ]
793        );
794        assert_eq!(
795            out.blobs
796                .blob
797                .iter()
798                .map(|v| v.properties.content_length)
799                .collect::<Vec<u64>>(),
800            [3485277, 2471869, 1259677]
801        );
802        assert_eq!(
803            out.blobs
804                .blob
805                .iter()
806                .map(|v| v.properties.content_md5.clone())
807                .collect::<Vec<String>>(),
808            [
809                "llJ/+jOlx5GdA1sL7SdKuw==".to_string(),
810                "xmgUltSnopLSJOukgCHFtg==".to_string(),
811                "AxTiFXHwrXKaZC5b7ZRybw==".to_string()
812            ]
813        );
814        assert_eq!(
815            out.blobs
816                .blob
817                .iter()
818                .map(|v| v.properties.last_modified.clone())
819                .collect::<Vec<String>>(),
820            [
821                "Sun, 20 Mar 2022 11:29:03 GMT".to_string(),
822                "Tue, 29 Mar 2022 01:54:07 GMT".to_string(),
823                "Sun, 20 Mar 2022 11:31:57 GMT".to_string()
824            ]
825        );
826        assert_eq!(
827            out.blobs
828                .blob
829                .iter()
830                .map(|v| v.properties.etag.clone())
831                .collect::<Vec<String>>(),
832            [
833                "0x8DA0A64D66790C3".to_string(),
834                "0x8DA112702D88FE4".to_string(),
835                "0x8DA0A653DC82981".to_string()
836            ]
837        );
838        assert_eq!(
839            out.blobs
840                .blob_prefix
841                .iter()
842                .map(|v| v.name.clone())
843                .collect::<Vec<String>>(),
844            ["dir1/dir2/", "dir1/dir21/"]
845        );
846    }
847
848    /// This case is copied from real environment for testing
849    /// quick-xml overlapped-lists features. By default, quick-xml
850    /// can't deserialize content with overlapped-lists.
851    ///
852    /// For example, this case list blobs in this way:
853    ///
854    /// ```xml
855    /// <Blobs>
856    ///     <Blob>xxx</Blob>
857    ///     <BlobPrefix>yyy</BlobPrefix>
858    ///     <Blob>zzz</Blob>
859    /// </Blobs>
860    /// ```
861    ///
862    /// If `overlapped-lists` feature not enabled, we will get error `duplicate field Blob`.
863    #[test]
864    fn test_parse_overlapped_lists() {
865        let bs = "<?xml version=\"1.0\" encoding=\"utf-8\"?><EnumerationResults ServiceEndpoint=\"https://test.blob.core.windows.net/\" ContainerName=\"test\"><Prefix>9f7075e1-84d0-45ca-8196-ab9b71a8ef97/x/</Prefix><Delimiter>/</Delimiter><Blobs><Blob><Name>9f7075e1-84d0-45ca-8196-ab9b71a8ef97/x/</Name><Properties><Creation-Time>Thu, 01 Sep 2022 07:26:49 GMT</Creation-Time><Last-Modified>Thu, 01 Sep 2022 07:26:49 GMT</Last-Modified><Etag>0x8DA8BEB55D0EA35</Etag><Content-Length>0</Content-Length><Content-Type>application/octet-stream</Content-Type><Content-Encoding /><Content-Language /><Content-CRC64 /><Content-MD5>1B2M2Y8AsgTpgAmY7PhCfg==</Content-MD5><Cache-Control /><Content-Disposition /><BlobType>BlockBlob</BlobType><AccessTier>Hot</AccessTier><AccessTierInferred>true</AccessTierInferred><LeaseStatus>unlocked</LeaseStatus><LeaseState>available</LeaseState><ServerEncrypted>true</ServerEncrypted></Properties><OrMetadata /></Blob><BlobPrefix><Name>9f7075e1-84d0-45ca-8196-ab9b71a8ef97/x/x/</Name></BlobPrefix><Blob><Name>9f7075e1-84d0-45ca-8196-ab9b71a8ef97/x/y</Name><Properties><Creation-Time>Thu, 01 Sep 2022 07:26:50 GMT</Creation-Time><Last-Modified>Thu, 01 Sep 2022 07:26:50 GMT</Last-Modified><Etag>0x8DA8BEB55D99C08</Etag><Content-Length>0</Content-Length><Content-Type>application/octet-stream</Content-Type><Content-Encoding /><Content-Language /><Content-CRC64 /><Content-MD5>1B2M2Y8AsgTpgAmY7PhCfg==</Content-MD5><Cache-Control /><Content-Disposition /><BlobType>BlockBlob</BlobType><AccessTier>Hot</AccessTier><AccessTierInferred>true</AccessTierInferred><LeaseStatus>unlocked</LeaseStatus><LeaseState>available</LeaseState><ServerEncrypted>true</ServerEncrypted></Properties><OrMetadata /></Blob></Blobs><NextMarker /></EnumerationResults>";
866
867        de::from_reader(Bytes::from(bs).reader()).expect("must success")
868    }
869
870    /// This example is from https://learn.microsoft.com/en-us/rest/api/storageservices/put-block-list?tabs=microsoft-entra-id
871    #[test]
872    fn test_serialize_put_block_list_request() {
873        let req = PutBlockListRequest {
874            latest: vec!["1".to_string(), "2".to_string(), "3".to_string()],
875        };
876
877        let actual = quick_xml::se::to_string(&req).expect("must succeed");
878
879        pretty_assertions::assert_eq!(
880            actual,
881            r#"
882            <BlockList>
883               <Latest>1</Latest>
884               <Latest>2</Latest>
885               <Latest>3</Latest>
886            </BlockList>"#
887                // Cleanup space and new line
888                .replace([' ', '\n'], "")
889                // Escape `"` by hand to address <https://github.com/tafia/quick-xml/issues/362>
890                .replace('"', "&quot;")
891        );
892
893        let bs = "<?xml version=\"1.0\" encoding=\"utf-8\"?>
894            <BlockList>
895               <Latest>1</Latest>
896               <Latest>2</Latest>
897               <Latest>3</Latest>
898            </BlockList>";
899
900        let out: PutBlockListRequest =
901            de::from_reader(Bytes::from(bs).reader()).expect("must success");
902        assert_eq!(
903            out.latest,
904            vec!["1".to_string(), "2".to_string(), "3".to_string()]
905        );
906    }
907}
opendal/services/azblob/core.rs

opendal/services/azblob/
core.rs
