From 5768bf362262f78376af14517c4921941986192e Mon Sep 17 00:00:00 2001 From: Alex Date: Tue, 10 May 2022 13:16:57 +0200 Subject: First implementation of K2V (#293) **Specification:** View spec at [this URL](https://git.deuxfleurs.fr/Deuxfleurs/garage/src/branch/k2v/doc/drafts/k2v-spec.md) - [x] Specify the structure of K2V triples - [x] Specify the DVVS format used for causality detection - [x] Specify the K2V index (just a counter of number of values per partition key) - [x] Specify single-item endpoints: ReadItem, InsertItem, DeleteItem - [x] Specify index endpoint: ReadIndex - [x] Specify multi-item endpoints: InsertBatch, ReadBatch, DeleteBatch - [x] Move to JSON objects instead of tuples - [x] Specify endpoints for polling for updates on single values (PollItem) **Implementation:** - [x] Table for K2V items, causal contexts - [x] Indexing mechanism and table for K2V index - [x] Make API handlers a bit more generic - [x] K2V API endpoint - [x] K2V API router - [x] ReadItem - [x] InsertItem - [x] DeleteItem - [x] PollItem - [x] ReadIndex - [x] InsertBatch - [x] ReadBatch - [x] DeleteBatch **Testing:** - [x] Just a simple Python script that does some requests to check visually that things are going right (does not contain parsing of results or assertions on returned values) - [x] Actual tests: - [x] Adapt testing framework - [x] Simple test with InsertItem + ReadItem - [x] Test with several Insert/Read/DeleteItem + ReadIndex - [x] Test all combinations of return formats for ReadItem - [x] Test with ReadBatch, InsertBatch, DeleteBatch - [x] Test with PollItem - [x] Test error codes - [ ] Fix most broken stuff - [x] test PollItem broken randomly - [x] when invalid causality tokens are given, errors should be 4xx not 5xx **Improvements:** - [x] Descending range queries - [x] Specify - [x] Implement - [x] Add test - [x] Batch updates to index counter - [x] Put K2V behind `k2v` feature flag Co-authored-by: Alex Auvolat Reviewed-on: https://git.deuxfleurs.fr/Deuxfleurs/garage/pulls/293 Co-authored-by: Alex Co-committed-by: Alex --- src/garage/tests/s3/list.rs | 615 +++++++++++++++++++++++++++++ src/garage/tests/s3/mod.rs | 6 + src/garage/tests/s3/multipart.rs | 415 +++++++++++++++++++ src/garage/tests/s3/objects.rs | 266 +++++++++++++ src/garage/tests/s3/simple.rs | 31 ++ src/garage/tests/s3/streaming_signature.rs | 185 +++++++++ src/garage/tests/s3/website.rs | 324 +++++++++++++++ 7 files changed, 1842 insertions(+) create mode 100644 src/garage/tests/s3/list.rs create mode 100644 src/garage/tests/s3/mod.rs create mode 100644 src/garage/tests/s3/multipart.rs create mode 100644 src/garage/tests/s3/objects.rs create mode 100644 src/garage/tests/s3/simple.rs create mode 100644 src/garage/tests/s3/streaming_signature.rs create mode 100644 src/garage/tests/s3/website.rs (limited to 'src/garage/tests/s3') diff --git a/src/garage/tests/s3/list.rs b/src/garage/tests/s3/list.rs new file mode 100644 index 00000000..bb03f250 --- /dev/null +++ b/src/garage/tests/s3/list.rs @@ -0,0 +1,615 @@ +use crate::common; + +const KEYS: [&str; 8] = ["a", "a/a", "a/b", "a/c", "a/d/a", "a/é", "b", "c"]; +const KEYS_MULTIPART: [&str; 5] = ["a", "a", "c", "c/a", "c/b"]; + +#[tokio::test] +async fn test_listobjectsv2() { + let ctx = common::context(); + let bucket = ctx.create_bucket("listobjectsv2"); + + for k in KEYS { + ctx.client + .put_object() + .bucket(&bucket) + .key(k) + .send() + .await + .unwrap(); + } + + { + // Scoping the variable to avoid reusing it + // in a following assert due to copy paste + let r = ctx + .client + .list_objects_v2() + .bucket(&bucket) + .send() + .await + .unwrap(); + + assert_eq!(r.contents.unwrap().len(), 8); + assert!(r.common_prefixes.is_none()); + } + + //@FIXME aws-sdk-s3 automatically checks max-key values. + // If we set it to zero, it drops it, and it is probably + // the same behavior on values bigger than 1000. + // Boto and awscli do not perform these tests, we should write + // our own minimal library to bypass AWS SDK's tests and be + // sure that we behave correctly. + + { + // With 2 elements + let r = ctx + .client + .list_objects_v2() + .bucket(&bucket) + .max_keys(2) + .send() + .await + .unwrap(); + + assert_eq!(r.contents.unwrap().len(), 2); + assert!(r.common_prefixes.is_none()); + assert!(r.next_continuation_token.is_some()); + } + + { + // With pagination + let mut cnt = 0; + let mut next = None; + let last_idx = KEYS.len() - 1; + + for i in 0..KEYS.len() { + let r = ctx + .client + .list_objects_v2() + .bucket(&bucket) + .set_continuation_token(next) + .max_keys(1) + .send() + .await + .unwrap(); + + cnt += 1; + next = r.next_continuation_token; + + assert_eq!(r.contents.unwrap().len(), 1); + assert!(r.common_prefixes.is_none()); + if i != last_idx { + assert!(next.is_some()); + } + } + assert_eq!(cnt, KEYS.len()); + } + + { + // With a delimiter + let r = ctx + .client + .list_objects_v2() + .bucket(&bucket) + .delimiter("/") + .send() + .await + .unwrap(); + + assert_eq!(r.contents.unwrap().len(), 3); + assert_eq!(r.common_prefixes.unwrap().len(), 1); + } + + { + // With a delimiter and pagination + let mut cnt_pfx = 0; + let mut cnt_key = 0; + let mut next = None; + + for _i in 0..KEYS.len() { + let r = ctx + .client + .list_objects_v2() + .bucket(&bucket) + .set_continuation_token(next) + .delimiter("/") + .max_keys(1) + .send() + .await + .unwrap(); + + next = r.next_continuation_token; + match (r.contents, r.common_prefixes) { + (Some(k), None) if k.len() == 1 => cnt_key += 1, + (None, Some(pfx)) if pfx.len() == 1 => cnt_pfx += 1, + _ => unreachable!("logic error"), + }; + if next.is_none() { + break; + } + } + assert_eq!(cnt_key, 3); + assert_eq!(cnt_pfx, 1); + } + + { + // With a prefix + let r = ctx + .client + .list_objects_v2() + .bucket(&bucket) + .prefix("a/") + .send() + .await + .unwrap(); + + assert_eq!(r.contents.unwrap().len(), 5); + assert!(r.common_prefixes.is_none()); + } + + { + // With a prefix and a delimiter + let r = ctx + .client + .list_objects_v2() + .bucket(&bucket) + .prefix("a/") + .delimiter("/") + .send() + .await + .unwrap(); + + assert_eq!(r.contents.unwrap().len(), 4); + assert_eq!(r.common_prefixes.unwrap().len(), 1); + } + + { + // With a prefix, a delimiter and max_key + let r = ctx + .client + .list_objects_v2() + .bucket(&bucket) + .prefix("a/") + .delimiter("/") + .max_keys(1) + .send() + .await + .unwrap(); + + assert_eq!(r.contents.as_ref().unwrap().len(), 1); + assert_eq!( + r.contents + .unwrap() + .first() + .unwrap() + .key + .as_ref() + .unwrap() + .as_str(), + "a/a" + ); + assert!(r.common_prefixes.is_none()); + } + { + // With start_after before all keys + let r = ctx + .client + .list_objects_v2() + .bucket(&bucket) + .start_after("Z") + .send() + .await + .unwrap(); + + assert_eq!(r.contents.unwrap().len(), 8); + assert!(r.common_prefixes.is_none()); + } + { + // With start_after after all keys + let r = ctx + .client + .list_objects_v2() + .bucket(&bucket) + .start_after("c") + .send() + .await + .unwrap(); + + assert!(r.contents.is_none()); + assert!(r.common_prefixes.is_none()); + } +} + +#[tokio::test] +async fn test_listobjectsv1() { + let ctx = common::context(); + let bucket = ctx.create_bucket("listobjects"); + + for k in KEYS { + ctx.client + .put_object() + .bucket(&bucket) + .key(k) + .send() + .await + .unwrap(); + } + + { + let r = ctx + .client + .list_objects() + .bucket(&bucket) + .send() + .await + .unwrap(); + + assert_eq!(r.contents.unwrap().len(), 8); + assert!(r.common_prefixes.is_none()); + } + + { + // With 2 elements + let r = ctx + .client + .list_objects() + .bucket(&bucket) + .max_keys(2) + .send() + .await + .unwrap(); + + assert_eq!(r.contents.unwrap().len(), 2); + assert!(r.common_prefixes.is_none()); + assert!(r.next_marker.is_some()); + } + + { + // With pagination + let mut cnt = 0; + let mut next = None; + let last_idx = KEYS.len() - 1; + + for i in 0..KEYS.len() { + let r = ctx + .client + .list_objects() + .bucket(&bucket) + .set_marker(next) + .max_keys(1) + .send() + .await + .unwrap(); + + cnt += 1; + next = r.next_marker; + + assert_eq!(r.contents.unwrap().len(), 1); + assert!(r.common_prefixes.is_none()); + if i != last_idx { + assert!(next.is_some()); + } + } + assert_eq!(cnt, KEYS.len()); + } + + { + // With a delimiter + let r = ctx + .client + .list_objects() + .bucket(&bucket) + .delimiter("/") + .send() + .await + .unwrap(); + + assert_eq!(r.contents.unwrap().len(), 3); + assert_eq!(r.common_prefixes.unwrap().len(), 1); + } + + { + // With a delimiter and pagination + let mut cnt_pfx = 0; + let mut cnt_key = 0; + let mut next = None; + + for _i in 0..KEYS.len() { + let r = ctx + .client + .list_objects() + .bucket(&bucket) + .delimiter("/") + .set_marker(next) + .max_keys(1) + .send() + .await + .unwrap(); + + next = r.next_marker; + match (r.contents, r.common_prefixes) { + (Some(k), None) if k.len() == 1 => cnt_key += 1, + (None, Some(pfx)) if pfx.len() == 1 => cnt_pfx += 1, + _ => unreachable!("logic error"), + }; + if next.is_none() { + break; + } + } + assert_eq!(cnt_key, 3); + // We have no optimization to skip the whole prefix + // on listobjectsv1 so we return the same one 5 times, + // for each element. It is up to the client to merge its result. + // This is compliant with AWS spec. + assert_eq!(cnt_pfx, 5); + } + + { + // With a prefix + let r = ctx + .client + .list_objects() + .bucket(&bucket) + .prefix("a/") + .send() + .await + .unwrap(); + + assert_eq!(r.contents.unwrap().len(), 5); + assert!(r.common_prefixes.is_none()); + } + + { + // With a prefix and a delimiter + let r = ctx + .client + .list_objects() + .bucket(&bucket) + .prefix("a/") + .delimiter("/") + .send() + .await + .unwrap(); + + assert_eq!(r.contents.unwrap().len(), 4); + assert_eq!(r.common_prefixes.unwrap().len(), 1); + } + + { + // With a prefix, a delimiter and max_key + let r = ctx + .client + .list_objects() + .bucket(&bucket) + .prefix("a/") + .delimiter("/") + .max_keys(1) + .send() + .await + .unwrap(); + + assert_eq!(r.contents.as_ref().unwrap().len(), 1); + assert_eq!( + r.contents + .unwrap() + .first() + .unwrap() + .key + .as_ref() + .unwrap() + .as_str(), + "a/a" + ); + assert!(r.common_prefixes.is_none()); + } + { + // With marker before all keys + let r = ctx + .client + .list_objects() + .bucket(&bucket) + .marker("Z") + .send() + .await + .unwrap(); + + assert_eq!(r.contents.unwrap().len(), 8); + assert!(r.common_prefixes.is_none()); + } + { + // With start_after after all keys + let r = ctx + .client + .list_objects() + .bucket(&bucket) + .marker("c") + .send() + .await + .unwrap(); + + assert!(r.contents.is_none()); + assert!(r.common_prefixes.is_none()); + } +} + +#[tokio::test] +async fn test_listmultipart() { + let ctx = common::context(); + let bucket = ctx.create_bucket("listmultipartuploads"); + + for k in KEYS_MULTIPART { + ctx.client + .create_multipart_upload() + .bucket(&bucket) + .key(k) + .send() + .await + .unwrap(); + } + + { + // Default + let r = ctx + .client + .list_multipart_uploads() + .bucket(&bucket) + .send() + .await + .unwrap(); + + assert_eq!(r.uploads.unwrap().len(), 5); + assert!(r.common_prefixes.is_none()); + } + { + // With pagination + let mut next = None; + let mut upnext = None; + let last_idx = KEYS_MULTIPART.len() - 1; + + for i in 0..KEYS_MULTIPART.len() { + let r = ctx + .client + .list_multipart_uploads() + .bucket(&bucket) + .set_key_marker(next) + .set_upload_id_marker(upnext) + .max_uploads(1) + .send() + .await + .unwrap(); + + next = r.next_key_marker; + upnext = r.next_upload_id_marker; + + assert_eq!(r.uploads.unwrap().len(), 1); + assert!(r.common_prefixes.is_none()); + if i != last_idx { + assert!(next.is_some()); + } + } + } + { + // With delimiter + let r = ctx + .client + .list_multipart_uploads() + .bucket(&bucket) + .delimiter("/") + .send() + .await + .unwrap(); + + assert_eq!(r.uploads.unwrap().len(), 3); + assert_eq!(r.common_prefixes.unwrap().len(), 1); + } + { + // With delimiter and pagination + let mut next = None; + let mut upnext = None; + let mut upcnt = 0; + let mut pfxcnt = 0; + let mut loopcnt = 0; + + while loopcnt < KEYS_MULTIPART.len() { + let r = ctx + .client + .list_multipart_uploads() + .bucket(&bucket) + .delimiter("/") + .max_uploads(1) + .set_key_marker(next) + .set_upload_id_marker(upnext) + .send() + .await + .unwrap(); + + next = r.next_key_marker; + upnext = r.next_upload_id_marker; + + loopcnt += 1; + upcnt += r.uploads.unwrap_or_default().len(); + pfxcnt += r.common_prefixes.unwrap_or_default().len(); + + if next.is_none() { + break; + } + } + + assert_eq!(upcnt + pfxcnt, loopcnt); + assert_eq!(upcnt, 3); + assert_eq!(pfxcnt, 1); + } + { + // With prefix + let r = ctx + .client + .list_multipart_uploads() + .bucket(&bucket) + .prefix("c") + .send() + .await + .unwrap(); + + assert_eq!(r.uploads.unwrap().len(), 3); + assert!(r.common_prefixes.is_none()); + } + { + // With prefix and delimiter + let r = ctx + .client + .list_multipart_uploads() + .bucket(&bucket) + .prefix("c") + .delimiter("/") + .send() + .await + .unwrap(); + + assert_eq!(r.uploads.unwrap().len(), 1); + assert_eq!(r.common_prefixes.unwrap().len(), 1); + } + { + // With prefix, delimiter and max keys + let r = ctx + .client + .list_multipart_uploads() + .bucket(&bucket) + .prefix("c") + .delimiter("/") + .max_uploads(1) + .send() + .await + .unwrap(); + + assert_eq!(r.uploads.unwrap().len(), 1); + assert!(r.common_prefixes.is_none()); + } + { + // With starting token before the first element + let r = ctx + .client + .list_multipart_uploads() + .bucket(&bucket) + .key_marker("ZZZZZ") + .send() + .await + .unwrap(); + + assert_eq!(r.uploads.unwrap().len(), 5); + assert!(r.common_prefixes.is_none()); + } + { + // With starting token after the last element + let r = ctx + .client + .list_multipart_uploads() + .bucket(&bucket) + .key_marker("d") + .send() + .await + .unwrap(); + + assert!(r.uploads.is_none()); + assert!(r.common_prefixes.is_none()); + } +} diff --git a/src/garage/tests/s3/mod.rs b/src/garage/tests/s3/mod.rs new file mode 100644 index 00000000..623eb665 --- /dev/null +++ b/src/garage/tests/s3/mod.rs @@ -0,0 +1,6 @@ +mod list; +mod multipart; +mod objects; +mod simple; +mod streaming_signature; +mod website; diff --git a/src/garage/tests/s3/multipart.rs b/src/garage/tests/s3/multipart.rs new file mode 100644 index 00000000..895a2993 --- /dev/null +++ b/src/garage/tests/s3/multipart.rs @@ -0,0 +1,415 @@ +use crate::common; +use aws_sdk_s3::model::{CompletedMultipartUpload, CompletedPart}; +use aws_sdk_s3::types::ByteStream; + +const SZ_5MB: usize = 5 * 1024 * 1024; +const SZ_10MB: usize = 10 * 1024 * 1024; + +#[tokio::test] +async fn test_uploadlistpart() { + let ctx = common::context(); + let bucket = ctx.create_bucket("uploadpart"); + + let u1 = vec![0xee; SZ_5MB]; + let u2 = vec![0x11; SZ_5MB]; + + let up = ctx + .client + .create_multipart_upload() + .bucket(&bucket) + .key("a") + .send() + .await + .unwrap(); + let uid = up.upload_id.as_ref().unwrap(); + + assert!(up.upload_id.is_some()); + + { + let r = ctx + .client + .list_parts() + .bucket(&bucket) + .key("a") + .upload_id(uid) + .send() + .await + .unwrap(); + + assert!(r.parts.is_none()); + } + + let p1 = ctx + .client + .upload_part() + .bucket(&bucket) + .key("a") + .upload_id(uid) + .part_number(2) + .body(ByteStream::from(u1)) + .send() + .await + .unwrap(); + + { + // ListPart on 1st element + let r = ctx + .client + .list_parts() + .bucket(&bucket) + .key("a") + .upload_id(uid) + .send() + .await + .unwrap(); + + let ps = r.parts.unwrap(); + assert_eq!(ps.len(), 1); + let fp = ps.iter().find(|x| x.part_number == 2).unwrap(); + assert!(fp.last_modified.is_some()); + assert_eq!( + fp.e_tag.as_ref().unwrap(), + "\"3366bb9dcf710d6801b5926467d02e19\"" + ); + assert_eq!(fp.size, SZ_5MB as i64); + } + + let p2 = ctx + .client + .upload_part() + .bucket(&bucket) + .key("a") + .upload_id(uid) + .part_number(1) + .body(ByteStream::from(u2)) + .send() + .await + .unwrap(); + + { + // ListPart on the 2 elements + let r = ctx + .client + .list_parts() + .bucket(&bucket) + .key("a") + .upload_id(uid) + .send() + .await + .unwrap(); + + let ps = r.parts.unwrap(); + assert_eq!(ps.len(), 2); + let fp = ps.iter().find(|x| x.part_number == 1).unwrap(); + assert!(fp.last_modified.is_some()); + assert_eq!( + fp.e_tag.as_ref().unwrap(), + "\"3c484266f9315485694556e6c693bfa2\"" + ); + assert_eq!(fp.size, SZ_5MB as i64); + } + + { + // Call pagination + let r = ctx + .client + .list_parts() + .bucket(&bucket) + .key("a") + .upload_id(uid) + .max_parts(1) + .send() + .await + .unwrap(); + + assert!(r.part_number_marker.is_none()); + assert!(r.next_part_number_marker.is_some()); + assert_eq!(r.max_parts, 1_i32); + assert!(r.is_truncated); + assert_eq!(r.key.unwrap(), "a"); + assert_eq!(r.upload_id.unwrap().as_str(), uid.as_str()); + assert_eq!(r.parts.unwrap().len(), 1); + + let r2 = ctx + .client + .list_parts() + .bucket(&bucket) + .key("a") + .upload_id(uid) + .max_parts(1) + .part_number_marker(r.next_part_number_marker.as_ref().unwrap()) + .send() + .await + .unwrap(); + + assert_eq!( + r2.part_number_marker.as_ref().unwrap(), + r.next_part_number_marker.as_ref().unwrap() + ); + assert_eq!(r2.max_parts, 1_i32); + assert!(r2.is_truncated); + assert_eq!(r2.key.unwrap(), "a"); + assert_eq!(r2.upload_id.unwrap().as_str(), uid.as_str()); + assert_eq!(r2.parts.unwrap().len(), 1); + } + + let cmp = CompletedMultipartUpload::builder() + .parts( + CompletedPart::builder() + .part_number(1) + .e_tag(p2.e_tag.unwrap()) + .build(), + ) + .parts( + CompletedPart::builder() + .part_number(2) + .e_tag(p1.e_tag.unwrap()) + .build(), + ) + .build(); + + ctx.client + .complete_multipart_upload() + .bucket(&bucket) + .key("a") + .upload_id(uid) + .multipart_upload(cmp) + .send() + .await + .unwrap(); + + // The multipart upload must not appear anymore + assert!(ctx + .client + .list_parts() + .bucket(&bucket) + .key("a") + .upload_id(uid) + .send() + .await + .is_err()); + + { + // The object must appear as a regular object + let r = ctx + .client + .head_object() + .bucket(&bucket) + .key("a") + .send() + .await + .unwrap(); + + assert_eq!(r.content_length, (SZ_5MB * 2) as i64); + } +} + +#[tokio::test] +async fn test_uploadpartcopy() { + let ctx = common::context(); + let bucket = ctx.create_bucket("uploadpartcopy"); + + let u1 = vec![0x11; SZ_10MB]; + let u2 = vec![0x22; SZ_5MB]; + let u3 = vec![0x33; SZ_5MB]; + let u4 = vec![0x44; SZ_5MB]; + let u5 = vec![0x55; SZ_5MB]; + + let overflow = 5500000 - SZ_5MB; + let mut exp_obj = u3.clone(); + exp_obj.extend(&u4[500..]); + exp_obj.extend(&u5[..overflow + 1]); + exp_obj.extend(&u2); + exp_obj.extend(&u1[500..5500000 + 1]); + + // (setup) Upload a single part object + ctx.client + .put_object() + .bucket(&bucket) + .key("source1") + .body(ByteStream::from(u1)) + .send() + .await + .unwrap(); + + // (setup) Upload a multipart object with 2 parts + { + let up = ctx + .client + .create_multipart_upload() + .bucket(&bucket) + .key("source2") + .send() + .await + .unwrap(); + let uid = up.upload_id.as_ref().unwrap(); + + let p1 = ctx + .client + .upload_part() + .bucket(&bucket) + .key("source2") + .upload_id(uid) + .part_number(1) + .body(ByteStream::from(u4)) + .send() + .await + .unwrap(); + + let p2 = ctx + .client + .upload_part() + .bucket(&bucket) + .key("source2") + .upload_id(uid) + .part_number(2) + .body(ByteStream::from(u5)) + .send() + .await + .unwrap(); + + let cmp = CompletedMultipartUpload::builder() + .parts( + CompletedPart::builder() + .part_number(1) + .e_tag(p1.e_tag.unwrap()) + .build(), + ) + .parts( + CompletedPart::builder() + .part_number(2) + .e_tag(p2.e_tag.unwrap()) + .build(), + ) + .build(); + + ctx.client + .complete_multipart_upload() + .bucket(&bucket) + .key("source2") + .upload_id(uid) + .multipart_upload(cmp) + .send() + .await + .unwrap(); + } + + // Our multipart object that does copy + let up = ctx + .client + .create_multipart_upload() + .bucket(&bucket) + .key("target") + .send() + .await + .unwrap(); + let uid = up.upload_id.as_ref().unwrap(); + + let p3 = ctx + .client + .upload_part() + .bucket(&bucket) + .key("target") + .upload_id(uid) + .part_number(3) + .body(ByteStream::from(u2)) + .send() + .await + .unwrap(); + + let p1 = ctx + .client + .upload_part() + .bucket(&bucket) + .key("target") + .upload_id(uid) + .part_number(1) + .body(ByteStream::from(u3)) + .send() + .await + .unwrap(); + + let p2 = ctx + .client + .upload_part_copy() + .bucket(&bucket) + .key("target") + .upload_id(uid) + .part_number(2) + .copy_source("uploadpartcopy/source2") + .copy_source_range("bytes=500-5500000") + .send() + .await + .unwrap(); + + let p4 = ctx + .client + .upload_part_copy() + .bucket(&bucket) + .key("target") + .upload_id(uid) + .part_number(4) + .copy_source("uploadpartcopy/source1") + .copy_source_range("bytes=500-5500000") + .send() + .await + .unwrap(); + + let cmp = CompletedMultipartUpload::builder() + .parts( + CompletedPart::builder() + .part_number(1) + .e_tag(p1.e_tag.unwrap()) + .build(), + ) + .parts( + CompletedPart::builder() + .part_number(2) + .e_tag(p2.copy_part_result.unwrap().e_tag.unwrap()) + .build(), + ) + .parts( + CompletedPart::builder() + .part_number(3) + .e_tag(p3.e_tag.unwrap()) + .build(), + ) + .parts( + CompletedPart::builder() + .part_number(4) + .e_tag(p4.copy_part_result.unwrap().e_tag.unwrap()) + .build(), + ) + .build(); + + ctx.client + .complete_multipart_upload() + .bucket(&bucket) + .key("target") + .upload_id(uid) + .multipart_upload(cmp) + .send() + .await + .unwrap(); + + // (check) Get object + + let obj = ctx + .client + .get_object() + .bucket(&bucket) + .key("target") + .send() + .await + .unwrap(); + + let real_obj = obj + .body + .collect() + .await + .expect("Error reading data") + .into_bytes(); + + assert_eq!(real_obj.len(), exp_obj.len()); + assert_eq!(real_obj, exp_obj); +} diff --git a/src/garage/tests/s3/objects.rs b/src/garage/tests/s3/objects.rs new file mode 100644 index 00000000..e1175b81 --- /dev/null +++ b/src/garage/tests/s3/objects.rs @@ -0,0 +1,266 @@ +use crate::common; +use aws_sdk_s3::model::{Delete, ObjectIdentifier}; +use aws_sdk_s3::types::ByteStream; + +const STD_KEY: &str = "hello world"; +const CTRL_KEY: &str = "\x00\x01\x02\x00"; +const UTF8_KEY: &str = "\u{211D}\u{1F923}\u{1F44B}"; +const BODY: &[u8; 62] = b"0123456789ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz"; + +#[tokio::test] +async fn test_putobject() { + let ctx = common::context(); + let bucket = ctx.create_bucket("putobject"); + + { + // Send an empty object (can serve as a directory marker) + // with a content type + let etag = "\"d41d8cd98f00b204e9800998ecf8427e\""; + let content_type = "text/csv"; + let r = ctx + .client + .put_object() + .bucket(&bucket) + .key(STD_KEY) + .content_type(content_type) + .send() + .await + .unwrap(); + + assert_eq!(r.e_tag.unwrap().as_str(), etag); + // We return a version ID here + // We should check if Amazon is returning one when versioning is not enabled + assert!(r.version_id.is_some()); + + let _version = r.version_id.unwrap(); + + let o = ctx + .client + .get_object() + .bucket(&bucket) + .key(STD_KEY) + .send() + .await + .unwrap(); + + assert_bytes_eq!(o.body, b""); + assert_eq!(o.e_tag.unwrap(), etag); + // We do not return version ID + // We should check if Amazon is returning one when versioning is not enabled + // assert_eq!(o.version_id.unwrap(), _version); + assert_eq!(o.content_type.unwrap(), content_type); + assert!(o.last_modified.is_some()); + assert_eq!(o.content_length, 0); + assert_eq!(o.parts_count, 0); + assert_eq!(o.tag_count, 0); + } + + { + // Key with control characters, + // no content type and some data + let etag = "\"49f68a5c8493ec2c0bf489821c21fc3b\""; + let data = ByteStream::from_static(b"hi"); + + let r = ctx + .client + .put_object() + .bucket(&bucket) + .key(CTRL_KEY) + .body(data) + .send() + .await + .unwrap(); + + assert_eq!(r.e_tag.unwrap().as_str(), etag); + assert!(r.version_id.is_some()); + + let o = ctx + .client + .get_object() + .bucket(&bucket) + .key(CTRL_KEY) + .send() + .await + .unwrap(); + + assert_bytes_eq!(o.body, b"hi"); + assert_eq!(o.e_tag.unwrap(), etag); + assert!(o.last_modified.is_some()); + assert_eq!(o.content_length, 2); + assert_eq!(o.parts_count, 0); + assert_eq!(o.tag_count, 0); + } + + { + // Key with UTF8 codepoints including emoji + let etag = "\"d41d8cd98f00b204e9800998ecf8427e\""; + + let r = ctx + .client + .put_object() + .bucket(&bucket) + .key(UTF8_KEY) + .send() + .await + .unwrap(); + + assert_eq!(r.e_tag.unwrap().as_str(), etag); + assert!(r.version_id.is_some()); + + let o = ctx + .client + .get_object() + .bucket(&bucket) + .key(UTF8_KEY) + .send() + .await + .unwrap(); + + assert_bytes_eq!(o.body, b""); + assert_eq!(o.e_tag.unwrap(), etag); + assert!(o.last_modified.is_some()); + assert_eq!(o.content_length, 0); + assert_eq!(o.parts_count, 0); + assert_eq!(o.tag_count, 0); + } +} + +#[tokio::test] +async fn test_getobject() { + let ctx = common::context(); + let bucket = ctx.create_bucket("getobject"); + + let etag = "\"46cf18a9b447991b450cad3facf5937e\""; + let data = ByteStream::from_static(BODY); + + let r = ctx + .client + .put_object() + .bucket(&bucket) + .key(STD_KEY) + .body(data) + .send() + .await + .unwrap(); + + assert_eq!(r.e_tag.unwrap().as_str(), etag); + + { + let o = ctx + .client + .get_object() + .bucket(&bucket) + .key(STD_KEY) + .range("bytes=1-9") + .send() + .await + .unwrap(); + + assert_eq!(o.content_range.unwrap().as_str(), "bytes 1-9/62"); + assert_bytes_eq!(o.body, &BODY[1..10]); + } + { + let o = ctx + .client + .get_object() + .bucket(&bucket) + .key(STD_KEY) + .range("bytes=9-") + .send() + .await + .unwrap(); + assert_eq!(o.content_range.unwrap().as_str(), "bytes 9-61/62"); + assert_bytes_eq!(o.body, &BODY[9..]); + } + { + let o = ctx + .client + .get_object() + .bucket(&bucket) + .key(STD_KEY) + .range("bytes=-5") + .send() + .await + .unwrap(); + assert_eq!(o.content_range.unwrap().as_str(), "bytes 57-61/62"); + assert_bytes_eq!(o.body, &BODY[57..]); + } +} + +#[tokio::test] +async fn test_deleteobject() { + let ctx = common::context(); + let bucket = ctx.create_bucket("deleteobject"); + + let mut to_del = Delete::builder(); + + // add content without data + for i in 0..5 { + let k = format!("k-{}", i); + ctx.client + .put_object() + .bucket(&bucket) + .key(k.to_string()) + .send() + .await + .unwrap(); + if i > 0 { + to_del = to_del.objects(ObjectIdentifier::builder().key(k).build()); + } + } + + // add content with data + for i in 0..5 { + let k = format!("l-{}", i); + let data = ByteStream::from_static(BODY); + ctx.client + .put_object() + .bucket(&bucket) + .key(k.to_string()) + .body(data) + .send() + .await + .unwrap(); + + if i > 0 { + to_del = to_del.objects(ObjectIdentifier::builder().key(k).build()); + } + } + + ctx.client + .delete_object() + .bucket(&bucket) + .key("k-0") + .send() + .await + .unwrap(); + + ctx.client + .delete_object() + .bucket(&bucket) + .key("l-0") + .send() + .await + .unwrap(); + + let r = ctx + .client + .delete_objects() + .bucket(&bucket) + .delete(to_del.build()) + .send() + .await + .unwrap(); + + assert_eq!(r.deleted.unwrap().len(), 8); + + let l = ctx + .client + .list_objects_v2() + .bucket(&bucket) + .send() + .await + .unwrap(); + + assert!(l.contents.is_none()); +} diff --git a/src/garage/tests/s3/simple.rs b/src/garage/tests/s3/simple.rs new file mode 100644 index 00000000..f54ae9ac --- /dev/null +++ b/src/garage/tests/s3/simple.rs @@ -0,0 +1,31 @@ +use crate::common; + +#[tokio::test] +async fn test_simple() { + use aws_sdk_s3::types::ByteStream; + + let ctx = common::context(); + let bucket = ctx.create_bucket("test-simple"); + + let data = ByteStream::from_static(b"Hello world!"); + + ctx.client + .put_object() + .bucket(&bucket) + .key("test") + .body(data) + .send() + .await + .unwrap(); + + let res = ctx + .client + .get_object() + .bucket(&bucket) + .key("test") + .send() + .await + .unwrap(); + + assert_bytes_eq!(res.body, b"Hello world!"); +} diff --git a/src/garage/tests/s3/streaming_signature.rs b/src/garage/tests/s3/streaming_signature.rs new file mode 100644 index 00000000..c68f7dfc --- /dev/null +++ b/src/garage/tests/s3/streaming_signature.rs @@ -0,0 +1,185 @@ +use std::collections::HashMap; + +use crate::common; +use common::custom_requester::BodySignature; +use hyper::Method; + +const STD_KEY: &str = "hello-world"; +//const CTRL_KEY: &str = "\x00\x01\x02\x00"; +const BODY: &[u8; 62] = b"0123456789ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz"; + +#[tokio::test] +async fn test_putobject_streaming() { + let ctx = common::context(); + let bucket = ctx.create_bucket("putobject-streaming"); + + { + // Send an empty object (can serve as a directory marker) + // with a content type + let etag = "\"d41d8cd98f00b204e9800998ecf8427e\""; + let content_type = "text/csv"; + let mut headers = HashMap::new(); + headers.insert("content-type".to_owned(), content_type.to_owned()); + let _ = ctx + .custom_request + .builder(bucket.clone()) + .method(Method::PUT) + .path(STD_KEY.to_owned()) + .unsigned_headers(headers) + .vhost_style(true) + .body(vec![]) + .body_signature(BodySignature::Streaming(10)) + .send() + .await + .unwrap(); + + // assert_eq!(r.e_tag.unwrap().as_str(), etag); + // We return a version ID here + // We should check if Amazon is returning one when versioning is not enabled + // assert!(r.version_id.is_some()); + + //let _version = r.version_id.unwrap(); + + let o = ctx + .client + .get_object() + .bucket(&bucket) + .key(STD_KEY) + .send() + .await + .unwrap(); + + assert_bytes_eq!(o.body, b""); + assert_eq!(o.e_tag.unwrap(), etag); + // We do not return version ID + // We should check if Amazon is returning one when versioning is not enabled + // assert_eq!(o.version_id.unwrap(), _version); + assert_eq!(o.content_type.unwrap(), content_type); + assert!(o.last_modified.is_some()); + assert_eq!(o.content_length, 0); + assert_eq!(o.parts_count, 0); + assert_eq!(o.tag_count, 0); + } + + { + let etag = "\"46cf18a9b447991b450cad3facf5937e\""; + + let _ = ctx + .custom_request + .builder(bucket.clone()) + .method(Method::PUT) + //.path(CTRL_KEY.to_owned()) at the moment custom_request does not encode url so this + //fail + .path("abc".to_owned()) + .vhost_style(true) + .body(BODY.to_vec()) + .body_signature(BodySignature::Streaming(16)) + .send() + .await + .unwrap(); + + // assert_eq!(r.e_tag.unwrap().as_str(), etag); + // assert!(r.version_id.is_some()); + + let o = ctx + .client + .get_object() + .bucket(&bucket) + //.key(CTRL_KEY) + .key("abc") + .send() + .await + .unwrap(); + + assert_bytes_eq!(o.body, BODY); + assert_eq!(o.e_tag.unwrap(), etag); + assert!(o.last_modified.is_some()); + assert_eq!(o.content_length, 62); + assert_eq!(o.parts_count, 0); + assert_eq!(o.tag_count, 0); + } +} + +#[tokio::test] +async fn test_create_bucket_streaming() { + let ctx = common::context(); + let bucket = "createbucket-streaming"; + + { + // create bucket + let _ = ctx + .custom_request + .builder(bucket.to_owned()) + .method(Method::PUT) + .body_signature(BodySignature::Streaming(10)) + .send() + .await + .unwrap(); + + // test if the bucket exists and works properly + let etag = "\"d41d8cd98f00b204e9800998ecf8427e\""; + let content_type = "text/csv"; + let _ = ctx + .client + .put_object() + .bucket(bucket) + .key(STD_KEY) + .content_type(content_type) + .send() + .await + .unwrap(); + + let o = ctx + .client + .get_object() + .bucket(bucket) + .key(STD_KEY) + .send() + .await + .unwrap(); + + assert_eq!(o.e_tag.unwrap(), etag); + } +} + +#[tokio::test] +async fn test_put_website_streaming() { + let ctx = common::context(); + let bucket = ctx.create_bucket("putwebsite-streaming"); + + { + let website_config = r#" + + + err/error.html + + + home.html + +"#; + + let mut query = HashMap::new(); + query.insert("website".to_owned(), None); + let _ = ctx + .custom_request + .builder(bucket.clone()) + .method(Method::PUT) + .query_params(query) + .body(website_config.as_bytes().to_vec()) + .body_signature(BodySignature::Streaming(10)) + .send() + .await + .unwrap(); + + let o = ctx + .client + .get_bucket_website() + .bucket(&bucket) + .send() + .await + .unwrap(); + + assert_eq!(o.index_document.unwrap().suffix.unwrap(), "home.html"); + assert_eq!(o.error_document.unwrap().key.unwrap(), "err/error.html"); + } +} diff --git a/src/garage/tests/s3/website.rs b/src/garage/tests/s3/website.rs new file mode 100644 index 00000000..0570ac6a --- /dev/null +++ b/src/garage/tests/s3/website.rs @@ -0,0 +1,324 @@ +use crate::common; +use crate::common::ext::*; +use aws_sdk_s3::{ + model::{CorsConfiguration, CorsRule, ErrorDocument, IndexDocument, WebsiteConfiguration}, + types::ByteStream, +}; +use http::Request; +use hyper::{ + body::{to_bytes, Body}, + Client, +}; + +const BODY: &[u8; 16] = b"

bonjour

"; +const BODY_ERR: &[u8; 6] = b"erreur"; + +#[tokio::test] +async fn test_website() { + const BCKT_NAME: &str = "my-website"; + let ctx = common::context(); + let bucket = ctx.create_bucket(BCKT_NAME); + + let data = ByteStream::from_static(BODY); + + ctx.client + .put_object() + .bucket(&bucket) + .key("index.html") + .body(data) + .send() + .await + .unwrap(); + + let client = Client::new(); + + let req = || { + Request::builder() + .method("GET") + .uri(format!("http://127.0.0.1:{}/", ctx.garage.web_port)) + .header("Host", format!("{}.web.garage", BCKT_NAME)) + .body(Body::empty()) + .unwrap() + }; + + let mut resp = client.request(req()).await.unwrap(); + + assert_eq!(resp.status(), 404); + assert_ne!( + to_bytes(resp.body_mut()).await.unwrap().as_ref(), + BODY.as_ref() + ); /* check that we do not leak body */ + + ctx.garage + .command() + .args(["bucket", "website", "--allow", BCKT_NAME]) + .quiet() + .expect_success_status("Could not allow website on bucket"); + + resp = client.request(req()).await.unwrap(); + assert_eq!(resp.status(), 200); + assert_eq!( + to_bytes(resp.body_mut()).await.unwrap().as_ref(), + BODY.as_ref() + ); + + ctx.garage + .command() + .args(["bucket", "website", "--deny", BCKT_NAME]) + .quiet() + .expect_success_status("Could not deny website on bucket"); + + resp = client.request(req()).await.unwrap(); + assert_eq!(resp.status(), 404); + assert_ne!( + to_bytes(resp.body_mut()).await.unwrap().as_ref(), + BODY.as_ref() + ); /* check that we do not leak body */ +} + +#[tokio::test] +async fn test_website_s3_api() { + const BCKT_NAME: &str = "my-cors"; + let ctx = common::context(); + let bucket = ctx.create_bucket(BCKT_NAME); + + let data = ByteStream::from_static(BODY); + + ctx.client + .put_object() + .bucket(&bucket) + .key("site/home.html") + .body(data) + .send() + .await + .unwrap(); + + ctx.client + .put_object() + .bucket(&bucket) + .key("err/error.html") + .body(ByteStream::from_static(BODY_ERR)) + .send() + .await + .unwrap(); + + let conf = WebsiteConfiguration::builder() + .index_document(IndexDocument::builder().suffix("home.html").build()) + .error_document(ErrorDocument::builder().key("err/error.html").build()) + .build(); + + ctx.client + .put_bucket_website() + .bucket(&bucket) + .website_configuration(conf) + .send() + .await + .unwrap(); + + let cors = CorsConfiguration::builder() + .cors_rules( + CorsRule::builder() + .id("main-rule") + .allowed_headers("*") + .allowed_methods("GET") + .allowed_methods("PUT") + .allowed_origins("*") + .build(), + ) + .build(); + + ctx.client + .put_bucket_cors() + .bucket(&bucket) + .cors_configuration(cors) + .send() + .await + .unwrap(); + + { + let cors_res = ctx + .client + .get_bucket_cors() + .bucket(&bucket) + .send() + .await + .unwrap(); + + let main_rule = cors_res.cors_rules().unwrap().iter().next().unwrap(); + + assert_eq!(main_rule.id.as_ref().unwrap(), "main-rule"); + assert_eq!( + main_rule.allowed_headers.as_ref().unwrap(), + &vec!["*".to_string()] + ); + assert_eq!( + main_rule.allowed_origins.as_ref().unwrap(), + &vec!["*".to_string()] + ); + assert_eq!( + main_rule.allowed_methods.as_ref().unwrap(), + &vec!["GET".to_string(), "PUT".to_string()] + ); + } + + let client = Client::new(); + + // Test direct requests with CORS + { + let req = Request::builder() + .method("GET") + .uri(format!("http://127.0.0.1:{}/site/", ctx.garage.web_port)) + .header("Host", format!("{}.web.garage", BCKT_NAME)) + .header("Origin", "https://example.com") + .body(Body::empty()) + .unwrap(); + + let mut resp = client.request(req).await.unwrap(); + + assert_eq!(resp.status(), 200); + assert_eq!( + resp.headers().get("access-control-allow-origin").unwrap(), + "*" + ); + assert_eq!( + to_bytes(resp.body_mut()).await.unwrap().as_ref(), + BODY.as_ref() + ); + } + + // Test ErrorDocument on 404 + { + let req = Request::builder() + .method("GET") + .uri(format!( + "http://127.0.0.1:{}/wrong.html", + ctx.garage.web_port + )) + .header("Host", format!("{}.web.garage", BCKT_NAME)) + .body(Body::empty()) + .unwrap(); + + let mut resp = client.request(req).await.unwrap(); + + assert_eq!(resp.status(), 404); + assert_eq!( + to_bytes(resp.body_mut()).await.unwrap().as_ref(), + BODY_ERR.as_ref() + ); + } + + // Test CORS with an allowed preflight request + { + let req = Request::builder() + .method("OPTIONS") + .uri(format!("http://127.0.0.1:{}/site/", ctx.garage.web_port)) + .header("Host", format!("{}.web.garage", BCKT_NAME)) + .header("Origin", "https://example.com") + .header("Access-Control-Request-Method", "PUT") + .body(Body::empty()) + .unwrap(); + + let mut resp = client.request(req).await.unwrap(); + + assert_eq!(resp.status(), 200); + assert_eq!( + resp.headers().get("access-control-allow-origin").unwrap(), + "*" + ); + assert_ne!( + to_bytes(resp.body_mut()).await.unwrap().as_ref(), + BODY.as_ref() + ); + } + + // Test CORS with a forbidden preflight request + { + let req = Request::builder() + .method("OPTIONS") + .uri(format!("http://127.0.0.1:{}/site/", ctx.garage.web_port)) + .header("Host", format!("{}.web.garage", BCKT_NAME)) + .header("Origin", "https://example.com") + .header("Access-Control-Request-Method", "DELETE") + .body(Body::empty()) + .unwrap(); + + let mut resp = client.request(req).await.unwrap(); + + assert_eq!(resp.status(), 403); + assert_ne!( + to_bytes(resp.body_mut()).await.unwrap().as_ref(), + BODY.as_ref() + ); + } + + //@TODO test CORS on the S3 endpoint. We need to handle auth manually to check it. + + // Delete cors + ctx.client + .delete_bucket_cors() + .bucket(&bucket) + .send() + .await + .unwrap(); + + // Check CORS are deleted from the API + // @FIXME check what is the expected behavior when GetBucketCors is called on a bucket without + // any CORS. + assert!(ctx + .client + .get_bucket_cors() + .bucket(&bucket) + .send() + .await + .is_err()); + + // Test CORS are not sent anymore on a previously allowed request + { + let req = Request::builder() + .method("OPTIONS") + .uri(format!("http://127.0.0.1:{}/site/", ctx.garage.web_port)) + .header("Host", format!("{}.web.garage", BCKT_NAME)) + .header("Origin", "https://example.com") + .header("Access-Control-Request-Method", "PUT") + .body(Body::empty()) + .unwrap(); + + let mut resp = client.request(req).await.unwrap(); + + assert_eq!(resp.status(), 403); + assert_ne!( + to_bytes(resp.body_mut()).await.unwrap().as_ref(), + BODY.as_ref() + ); + } + + // Disallow website from the API + ctx.client + .delete_bucket_website() + .bucket(&bucket) + .send() + .await + .unwrap(); + + // Check that the website is not served anymore + { + let req = Request::builder() + .method("GET") + .uri(format!("http://127.0.0.1:{}/site/", ctx.garage.web_port)) + .header("Host", format!("{}.web.garage", BCKT_NAME)) + .body(Body::empty()) + .unwrap(); + + let mut resp = client.request(req).await.unwrap(); + + assert_eq!(resp.status(), 404); + assert_ne!( + to_bytes(resp.body_mut()).await.unwrap().as_ref(), + BODY_ERR.as_ref() + ); + assert_ne!( + to_bytes(resp.body_mut()).await.unwrap().as_ref(), + BODY.as_ref() + ); + } +} -- cgit v1.2.3