UploadPartCopy and improvements to CopyObject #188

Merged
lx merged 4 commits from improve-copies into main 2022-01-13 13:18:32 +00:00
11 changed files with 623 additions and 108 deletions

View file

@ -21,7 +21,7 @@ Not implemented:
## Endpoint implementation ## Endpoint implementation
All APIs that are not mentionned are not implemented and will return a 400 bad request. All APIs that are not mentionned are not implemented and will return a 501 Not Implemented.
| Endpoint | Status | | Endpoint | Status |
|------------------------------|----------------------------------| |------------------------------|----------------------------------|
@ -48,6 +48,7 @@ All APIs that are not mentionned are not implemented and will return a 400 bad r
| PutObject | Implemented | | PutObject | Implemented |
| PutBucketWebsite | Partially implemented (see below)| | PutBucketWebsite | Partially implemented (see below)|
| UploadPart | Implemented | | UploadPart | Implemented |
| UploadPartCopy | Implemented |
- **GetBucketVersioning:** Stub implementation (Garage does not yet support versionning so this always returns - **GetBucketVersioning:** Stub implementation (Garage does not yet support versionning so this always returns

View file

@ -30,7 +30,7 @@ your motivations for doing so in the PR message.
| | [*GetBucketCors*](https://git.deuxfleurs.fr/Deuxfleurs/garage/issues/138) | | | [*GetBucketCors*](https://git.deuxfleurs.fr/Deuxfleurs/garage/issues/138) |
| | [*PutBucketCors*](https://git.deuxfleurs.fr/Deuxfleurs/garage/issues/138) | | | [*PutBucketCors*](https://git.deuxfleurs.fr/Deuxfleurs/garage/issues/138) |
| | [*DeleteBucketCors*](https://git.deuxfleurs.fr/Deuxfleurs/garage/issues/138) | | | [*DeleteBucketCors*](https://git.deuxfleurs.fr/Deuxfleurs/garage/issues/138) |
| | [*UploadPartCopy*](https://git.deuxfleurs.fr/Deuxfleurs/garage/issues/160) | | | UploadPartCopy |
| | [*GetBucketWebsite*](https://git.deuxfleurs.fr/Deuxfleurs/garage/issues/77) | | | [*GetBucketWebsite*](https://git.deuxfleurs.fr/Deuxfleurs/garage/issues/77) |
| | [*PutBucketWebsite*](https://git.deuxfleurs.fr/Deuxfleurs/garage/issues/77) | | | [*PutBucketWebsite*](https://git.deuxfleurs.fr/Deuxfleurs/garage/issues/77) |
| | DeleteBucketWebsite | | | DeleteBucketWebsite |

View file

@ -116,21 +116,7 @@ if [ -z "$SKIP_DUCK" ]; then
done done
fi fi
rm /tmp/garage.{1..3}.{rnd,b64} # Advanced testing via S3API
if [ -z "$SKIP_AWS" ]; then
echo "🧪 Website Testing"
echo "<h1>hello world</h1>" > /tmp/garage-index.html
aws s3 cp /tmp/garage-index.html s3://eprouvette/index.html
[ `curl -s -o /dev/null -w "%{http_code}" --header "Host: eprouvette.garage.tld" http://127.0.0.1:3921/ ` == 404 ]
garage -c /tmp/config.1.toml bucket website --allow eprouvette
[ `curl -s -o /dev/null -w "%{http_code}" --header "Host: eprouvette.garage.tld" http://127.0.0.1:3921/ ` == 200 ]
garage -c /tmp/config.1.toml bucket website --deny eprouvette
[ `curl -s -o /dev/null -w "%{http_code}" --header "Host: eprouvette.garage.tld" http://127.0.0.1:3921/ ` == 404 ]
aws s3 rm s3://eprouvette/index.html
rm /tmp/garage-index.html
fi
if [ -z "$SKIP_AWS" ]; then if [ -z "$SKIP_AWS" ]; then
echo "🔌 Test S3API" echo "🔌 Test S3API"
@ -265,8 +251,61 @@ if [ -z "$SKIP_AWS" ]; then
aws s3api abort-multipart-upload --bucket eprouvette --key $key --upload-id $uid; aws s3api abort-multipart-upload --bucket eprouvette --key $key --upload-id $uid;
echo "Deleted ${key}:${uid}" echo "Deleted ${key}:${uid}"
done done
# Test for UploadPartCopy
aws s3 cp "/tmp/garage.3.rnd" "s3://eprouvette/copy_part_source"
UPLOAD_ID=$(aws s3api create-multipart-upload --bucket eprouvette --key test_multipart | jq -r .UploadId)
PART1=$(aws s3api upload-part \
--bucket eprouvette --key test_multipart \
--upload-id $UPLOAD_ID --part-number 1 \
--body /tmp/garage.2.rnd | jq .ETag)
PART2=$(aws s3api upload-part-copy \
--bucket eprouvette --key test_multipart \
--upload-id $UPLOAD_ID --part-number 2 \
--copy-source "/eprouvette/copy_part_source" \
--copy-source-range "bytes=500-5000500" \
| jq .CopyPartResult.ETag)
PART3=$(aws s3api upload-part \
--bucket eprouvette --key test_multipart \
--upload-id $UPLOAD_ID --part-number 3 \
--body /tmp/garage.3.rnd | jq .ETag)
cat >/tmp/garage.multipart_struct <<EOF
{
"Parts": [
{
"ETag": $PART1,
"PartNumber": 1
},
{
"ETag": $PART2,
"PartNumber": 2
},
{
"ETag": $PART3,
"PartNumber": 3
}
]
}
EOF
aws s3api complete-multipart-upload \
--bucket eprouvette --key test_multipart --upload-id $UPLOAD_ID \
--multipart-upload file:///tmp/garage.multipart_struct
aws s3 cp "s3://eprouvette/test_multipart" /tmp/garage.test_multipart
cat /tmp/garage.2.rnd <(tail -c +501 /tmp/garage.3.rnd | head -c 5000001) /tmp/garage.3.rnd > /tmp/garage.test_multipart_reference
diff /tmp/garage.test_multipart /tmp/garage.test_multipart_reference >/tmp/garage.test_multipart_diff 2>&1
aws s3 rm "s3://eprouvette/copy_part_source"
aws s3 rm "s3://eprouvette/test_multipart"
rm /tmp/garage.multipart_struct
rm /tmp/garage.test_multipart
rm /tmp/garage.test_multipart_reference
rm /tmp/garage.test_multipart_diff
fi fi
rm /tmp/garage.{1..3}.{rnd,b64}
if [ -z "$SKIP_AWS" ]; then if [ -z "$SKIP_AWS" ]; then
echo "🪣 Test bucket logic " echo "🪣 Test bucket logic "
AWS_ACCESS_KEY_ID=`cat /tmp/garage.s3 |cut -d' ' -f1` AWS_ACCESS_KEY_ID=`cat /tmp/garage.s3 |cut -d' ' -f1`
@ -282,6 +321,19 @@ if [ -z "$SKIP_AWS" ]; then
[ $(aws s3 ls | wc -l) == 1 ] [ $(aws s3 ls | wc -l) == 1 ]
fi fi
if [ -z "$SKIP_AWS" ]; then
echo "🧪 Website Testing"
echo "<h1>hello world</h1>" > /tmp/garage-index.html
aws s3 cp /tmp/garage-index.html s3://eprouvette/index.html
[ `curl -s -o /dev/null -w "%{http_code}" --header "Host: eprouvette.garage.tld" http://127.0.0.1:3921/ ` == 404 ]
garage -c /tmp/config.1.toml bucket website --allow eprouvette
[ `curl -s -o /dev/null -w "%{http_code}" --header "Host: eprouvette.garage.tld" http://127.0.0.1:3921/ ` == 200 ]
garage -c /tmp/config.1.toml bucket website --deny eprouvette
[ `curl -s -o /dev/null -w "%{http_code}" --header "Host: eprouvette.garage.tld" http://127.0.0.1:3921/ ` == 404 ]
aws s3 rm s3://eprouvette/index.html
rm /tmp/garage-index.html
fi
echo "🏁 Teardown" echo "🏁 Teardown"
AWS_ACCESS_KEY_ID=`cat /tmp/garage.s3 |cut -d' ' -f1` AWS_ACCESS_KEY_ID=`cat /tmp/garage.s3 |cut -d' ' -f1`
AWS_SECRET_ACCESS_KEY=`cat /tmp/garage.s3 |cut -d' ' -f2` AWS_SECRET_ACCESS_KEY=`cat /tmp/garage.s3 |cut -d' ' -f2`

View file

@ -156,19 +156,24 @@ async fn handler_inner(garage: Arc<Garage>, req: Request<Body>) -> Result<Respon
.await .await
} }
Endpoint::CopyObject { key, .. } => { Endpoint::CopyObject { key, .. } => {
let copy_source = req.headers().get("x-amz-copy-source").unwrap().to_str()?; handle_copy(garage, &api_key, &req, bucket_id, &key).await
let copy_source = percent_encoding::percent_decode_str(copy_source).decode_utf8()?; }
let (source_bucket, source_key) = parse_bucket_key(&copy_source, None)?; Endpoint::UploadPartCopy {
let source_bucket_id = key,
resolve_bucket(&garage, &source_bucket.to_string(), &api_key).await?; part_number,
if !api_key.allow_read(&source_bucket_id) { upload_id,
return Err(Error::Forbidden(format!( ..
"Reading from bucket {} not allowed for this key", } => {
source_bucket handle_upload_part_copy(
))); garage,
} &api_key,
let source_key = source_key.ok_or_bad_request("No source key specified")?; &req,
handle_copy(garage, &req, bucket_id, &key, source_bucket_id, source_key).await bucket_id,
&key,
part_number,
&upload_id,
)
.await
} }
Endpoint::PutObject { key, .. } => { Endpoint::PutObject { key, .. } => {
handle_put(garage, req, bucket_id, &key, content_sha256).await handle_put(garage, req, bucket_id, &key, content_sha256).await
@ -321,7 +326,7 @@ async fn handle_request_without_bucket(
} }
#[allow(clippy::ptr_arg)] #[allow(clippy::ptr_arg)]
async fn resolve_bucket( pub async fn resolve_bucket(
garage: &Garage, garage: &Garage,
bucket_name: &String, bucket_name: &String,
api_key: &Key, api_key: &Key,
@ -347,7 +352,7 @@ async fn resolve_bucket(
/// ///
/// S3 internally manages only buckets and keys. This function splits /// S3 internally manages only buckets and keys. This function splits
/// an HTTP path to get the corresponding bucket name and key. /// an HTTP path to get the corresponding bucket name and key.
fn parse_bucket_key<'a>( pub fn parse_bucket_key<'a>(
path: &'a str, path: &'a str,
host_bucket: Option<&'a str>, host_bucket: Option<&'a str>,
) -> Result<(&'a str, Option<&'a str>), Error> { ) -> Result<(&'a str, Option<&'a str>), Error> {

View file

@ -54,6 +54,10 @@ pub enum Error {
#[error(display = "Tried to delete a non-empty bucket")] #[error(display = "Tried to delete a non-empty bucket")]
BucketNotEmpty, BucketNotEmpty,
/// Precondition failed (e.g. x-amz-copy-source-if-match)
#[error(display = "At least one of the preconditions you specified did not hold")]
PreconditionFailed,
// Category: bad request // Category: bad request
/// The request contained an invalid UTF-8 sequence in its path or in other parameters /// The request contained an invalid UTF-8 sequence in its path or in other parameters
#[error(display = "Invalid UTF-8: {}", _0)] #[error(display = "Invalid UTF-8: {}", _0)]
@ -115,6 +119,7 @@ impl Error {
match self { match self {
Error::NoSuchKey | Error::NoSuchBucket | Error::NoSuchUpload => StatusCode::NOT_FOUND, Error::NoSuchKey | Error::NoSuchBucket | Error::NoSuchUpload => StatusCode::NOT_FOUND,
Error::BucketNotEmpty | Error::BucketAlreadyExists => StatusCode::CONFLICT, Error::BucketNotEmpty | Error::BucketAlreadyExists => StatusCode::CONFLICT,
Error::PreconditionFailed => StatusCode::PRECONDITION_FAILED,
Error::Forbidden(_) => StatusCode::FORBIDDEN, Error::Forbidden(_) => StatusCode::FORBIDDEN,
Error::InternalError( Error::InternalError(
GarageError::Timeout GarageError::Timeout
@ -137,6 +142,7 @@ impl Error {
Error::NoSuchUpload => "NoSuchUpload", Error::NoSuchUpload => "NoSuchUpload",
Error::BucketAlreadyExists => "BucketAlreadyExists", Error::BucketAlreadyExists => "BucketAlreadyExists",
Error::BucketNotEmpty => "BucketNotEmpty", Error::BucketNotEmpty => "BucketNotEmpty",
Error::PreconditionFailed => "PreconditionFailed",
Error::Forbidden(_) => "AccessDenied", Error::Forbidden(_) => "AccessDenied",
Error::AuthorizationHeaderMalformed(_) => "AuthorizationHeaderMalformed", Error::AuthorizationHeaderMalformed(_) => "AuthorizationHeaderMalformed",
Error::NotImplemented(_) => "NotImplemented", Error::NotImplemented(_) => "NotImplemented",

View file

@ -1,6 +1,11 @@
use std::sync::Arc; use std::sync::Arc;
use std::time::{Duration, SystemTime, UNIX_EPOCH};
use futures::TryFutureExt;
use md5::{Digest as Md5Digest, Md5};
use hyper::{Body, Request, Response}; use hyper::{Body, Request, Response};
use serde::Serialize;
use garage_table::*; use garage_table::*;
use garage_util::data::*; use garage_util::data::*;
@ -8,63 +13,50 @@ use garage_util::time::*;
use garage_model::block_ref_table::*; use garage_model::block_ref_table::*;
use garage_model::garage::Garage; use garage_model::garage::Garage;
use garage_model::key_table::Key;
use garage_model::object_table::*; use garage_model::object_table::*;
use garage_model::version_table::*; use garage_model::version_table::*;
use crate::api_server::{parse_bucket_key, resolve_bucket};
use crate::error::*; use crate::error::*;
use crate::s3_put::get_headers; use crate::s3_put::{decode_upload_id, get_headers};
use crate::s3_xml; use crate::s3_xml::{self, xmlns_tag};
pub async fn handle_copy( pub async fn handle_copy(
garage: Arc<Garage>, garage: Arc<Garage>,
api_key: &Key,
req: &Request<Body>, req: &Request<Body>,
dest_bucket_id: Uuid, dest_bucket_id: Uuid,
dest_key: &str, dest_key: &str,
source_bucket_id: Uuid,
source_key: &str,
) -> Result<Response<Body>, Error> { ) -> Result<Response<Body>, Error> {
let source_object = garage let copy_precondition = CopyPreconditionHeaders::parse(req)?;
.object_table
.get(&source_bucket_id, &source_key.to_string())
.await?
.ok_or(Error::NoSuchKey)?;
let source_last_v = source_object let source_object = get_copy_source(&garage, api_key, req).await?;
.versions()
.iter()
.rev()
.find(|v| v.is_complete())
.ok_or(Error::NoSuchKey)?;
let source_last_state = match &source_last_v.state { let (source_version, source_version_data, source_version_meta) =
ObjectVersionState::Complete(x) => x, extract_source_info(&source_object)?;
_ => unreachable!(),
};
// Check precondition, e.g. x-amz-copy-source-if-match
copy_precondition.check(source_version, &source_version_meta.etag)?;
// Generate parameters for copied object
let new_uuid = gen_uuid(); let new_uuid = gen_uuid();
let new_timestamp = now_msec(); let new_timestamp = now_msec();
// Implement x-amz-metadata-directive: REPLACE // Implement x-amz-metadata-directive: REPLACE
let old_meta = match source_last_state {
ObjectVersionData::DeleteMarker => {
return Err(Error::NoSuchKey);
}
ObjectVersionData::Inline(meta, _bytes) => meta,
ObjectVersionData::FirstBlock(meta, _fbh) => meta,
};
let new_meta = match req.headers().get("x-amz-metadata-directive") { let new_meta = match req.headers().get("x-amz-metadata-directive") {
Some(v) if v == hyper::header::HeaderValue::from_static("REPLACE") => ObjectVersionMeta { Some(v) if v == hyper::header::HeaderValue::from_static("REPLACE") => ObjectVersionMeta {
headers: get_headers(req)?, headers: get_headers(req)?,
size: old_meta.size, size: source_version_meta.size,
etag: old_meta.etag.clone(), etag: source_version_meta.etag.clone(),
}, },
_ => old_meta.clone(), _ => source_version_meta.clone(),
}; };
let etag = new_meta.etag.to_string(); let etag = new_meta.etag.to_string();
// Save object copy // Save object copy
match source_last_state { match source_version_data {
ObjectVersionData::DeleteMarker => unreachable!(), ObjectVersionData::DeleteMarker => unreachable!(),
ObjectVersionData::Inline(_meta, bytes) => { ObjectVersionData::Inline(_meta, bytes) => {
let dest_object_version = ObjectVersion { let dest_object_version = ObjectVersion {
@ -86,7 +78,7 @@ pub async fn handle_copy(
// Get block list from source version // Get block list from source version
let source_version = garage let source_version = garage
.version_table .version_table
.get(&source_last_v.uuid, &EmptyKey) .get(&source_version.uuid, &EmptyKey)
.await?; .await?;
let source_version = source_version.ok_or(Error::NoSuchKey)?; let source_version = source_version.ok_or(Error::NoSuchKey)?;
@ -156,13 +148,468 @@ pub async fn handle_copy(
} }
let last_modified = msec_to_rfc3339(new_timestamp); let last_modified = msec_to_rfc3339(new_timestamp);
let result = s3_xml::CopyObjectResult { let result = CopyObjectResult {
last_modified: s3_xml::Value(last_modified), last_modified: s3_xml::Value(last_modified),
etag: s3_xml::Value(etag), etag: s3_xml::Value(format!("\"{}\"", etag)),
}; };
let xml = s3_xml::to_xml_with_header(&result)?; let xml = s3_xml::to_xml_with_header(&result)?;
Ok(Response::builder() Ok(Response::builder()
.header("Content-Type", "application/xml") .header("Content-Type", "application/xml")
.header("x-amz-version-id", hex::encode(new_uuid))
.header(
"x-amz-copy-source-version-id",
hex::encode(source_version.uuid),
)
.body(Body::from(xml))?) .body(Body::from(xml))?)
} }
pub async fn handle_upload_part_copy(
garage: Arc<Garage>,
api_key: &Key,
req: &Request<Body>,
dest_bucket_id: Uuid,
dest_key: &str,
part_number: u64,
upload_id: &str,
) -> Result<Response<Body>, Error> {
let copy_precondition = CopyPreconditionHeaders::parse(req)?;
let dest_version_uuid = decode_upload_id(upload_id)?;
let dest_key = dest_key.to_string();
let (source_object, dest_object) = futures::try_join!(
get_copy_source(&garage, api_key, req),
garage
.object_table
.get(&dest_bucket_id, &dest_key)
.map_err(Error::from),
)?;
let dest_object = dest_object.ok_or(Error::NoSuchKey)?;
let (source_object_version, source_version_data, source_version_meta) =
extract_source_info(&source_object)?;
// Check precondition on source, e.g. x-amz-copy-source-if-match
copy_precondition.check(source_object_version, &source_version_meta.etag)?;
// Check source range is valid
let source_range = match req.headers().get("x-amz-copy-source-range") {
Some(range) => {
let range_str = range.to_str()?;
let mut ranges = http_range::HttpRange::parse(range_str, source_version_meta.size)
.map_err(|e| (e, source_version_meta.size))?;
if ranges.len() != 1 {
return Err(Error::BadRequest(
"Invalid x-amz-copy-source-range header: exactly 1 range must be given".into(),
));
} else {
ranges.pop().unwrap()
}
}
None => http_range::HttpRange {
start: 0,
length: source_version_meta.size,
},
};
// Check destination version is indeed in uploading state
if !dest_object
.versions()
.iter()
.any(|v| v.uuid == dest_version_uuid && v.is_uploading())
{
return Err(Error::NoSuchUpload);
}
// Check source version is not inlined
match source_version_data {
ObjectVersionData::DeleteMarker => unreachable!(),
ObjectVersionData::Inline(_meta, _bytes) => {
// This is only for small files, we don't bother handling this.
// (in AWS UploadPartCopy works for parts at least 5MB which
// is never the case of an inline object)
return Err(Error::BadRequest(
"Source object is too small (minimum part size is 5Mb)".into(),
));
}
ObjectVersionData::FirstBlock(_meta, _first_block_hash) => (),
};
// Fetch source versin with its block list,
// and destination version to check part hasn't yet been uploaded
let (source_version, dest_version) = futures::try_join!(
garage
.version_table
.get(&source_object_version.uuid, &EmptyKey),
garage.version_table.get(&dest_version_uuid, &EmptyKey),
)?;
let source_version = source_version.ok_or(Error::NoSuchKey)?;
// Check this part number hasn't yet been uploaded
if let Some(dv) = dest_version {
if dv.has_part_number(part_number) {
return Err(Error::BadRequest(format!(
"Part number {} has already been uploaded",
part_number
)));
}
}
// We want to reuse blocks from the source version as much as possible.
// However, we still need to get the data from these blocks
// because we need to know it to calculate the MD5sum of the part
// which is used as its ETag.
// First, calculate what blocks we want to keep,
// and the subrange of the block to take, if the bounds of the
// requested range are in the middle.
let (range_begin, range_end) = (source_range.start, source_range.start + source_range.length);
let mut blocks_to_copy = vec![];
let mut current_offset = 0;
let mut size_to_copy = 0;
for (_bk, block) in source_version.blocks.items().iter() {
let (block_begin, block_end) = (current_offset, current_offset + block.size);
if block_begin < range_end && block_end > range_begin {
let subrange_begin = if block_begin < range_begin {
Some(range_begin - block_begin)
} else {
None
};
let subrange_end = if block_end > range_end {
Some(range_end - block_begin)
} else {
None
};
let range_to_copy = match (subrange_begin, subrange_end) {
(Some(b), Some(e)) => Some(b as usize..e as usize),
(None, Some(e)) => Some(0..e as usize),
(Some(b), None) => Some(b as usize..block.size as usize),
(None, None) => None,
};
size_to_copy += range_to_copy
.as_ref()
.map(|x| x.len() as u64)
.unwrap_or(block.size);
blocks_to_copy.push((block.hash, range_to_copy));
}
current_offset = block_end;
}
if size_to_copy < 1024 * 1024 {
return Err(Error::BadRequest(format!(
"Not enough data to copy: {} bytes (minimum: 1MB)",
size_to_copy
)));
}
// Now, actually copy the blocks
let mut md5hasher = Md5::new();
let mut block = Some(
garage
.block_manager
.rpc_get_block(&blocks_to_copy[0].0)
.await?,
);
let mut current_offset = 0;
for (i, (block_hash, range_to_copy)) in blocks_to_copy.iter().enumerate() {
let (current_block, subrange_hash) = match range_to_copy.clone() {
Some(r) => {
let subrange = block.take().unwrap()[r].to_vec();
let hash = blake2sum(&subrange);
(subrange, hash)
}
None => (block.take().unwrap(), *block_hash),
};
md5hasher.update(&current_block[..]);
let mut version = Version::new(dest_version_uuid, dest_bucket_id, dest_key.clone(), false);
version.blocks.put(
VersionBlockKey {
part_number,
offset: current_offset,
},
VersionBlock {
hash: subrange_hash,
size: current_block.len() as u64,
},
);
current_offset += current_block.len() as u64;
let block_ref = BlockRef {
block: subrange_hash,
version: dest_version_uuid,
deleted: false.into(),
};
let next_block_hash = blocks_to_copy.get(i + 1).map(|(h, _)| *h);
let garage2 = garage.clone();
let garage3 = garage.clone();
let is_subrange = range_to_copy.is_some();
let (_, _, _, next_block) = futures::try_join!(
// Thing 1: if we are taking a subrange of the source block,
// we need to insert that subrange as a new block.
async move {
if is_subrange {
garage2
.block_manager
.rpc_put_block(subrange_hash, current_block)
.await
} else {
Ok(())
}
},
// Thing 2: we need to insert the block in the version
garage.version_table.insert(&version),
// Thing 3: we need to add a block reference
garage.block_ref_table.insert(&block_ref),
// Thing 4: we need to prefetch the next block
async move {
match next_block_hash {
Some(h) => Ok(Some(garage3.block_manager.rpc_get_block(&h).await?)),
None => Ok(None),
}
},
)?;
block = next_block;
}
let data_md5sum = md5hasher.finalize();
let etag = hex::encode(data_md5sum);
// Put the part's ETag in the Versiontable
let mut version = Version::new(dest_version_uuid, dest_bucket_id, dest_key.clone(), false);
version.parts_etags.put(part_number, etag.clone());
garage.version_table.insert(&version).await?;
// LGTM
let resp_xml = s3_xml::to_xml_with_header(&CopyPartResult {
xmlns: (),
etag: s3_xml::Value(format!("\"{}\"", etag)),
last_modified: s3_xml::Value(msec_to_rfc3339(source_object_version.timestamp)),
})?;
Ok(Response::builder()
.header("Content-Type", "application/xml")
.header(
"x-amz-copy-source-version-id",
hex::encode(source_object_version.uuid),
)
.body(Body::from(resp_xml))?)
}
async fn get_copy_source(
garage: &Garage,
api_key: &Key,
req: &Request<Body>,
) -> Result<Object, Error> {
let copy_source = req.headers().get("x-amz-copy-source").unwrap().to_str()?;
let copy_source = percent_encoding::percent_decode_str(copy_source).decode_utf8()?;
let (source_bucket, source_key) = parse_bucket_key(&copy_source, None)?;
let source_bucket_id = resolve_bucket(garage, &source_bucket.to_string(), api_key).await?;
if !api_key.allow_read(&source_bucket_id) {
return Err(Error::Forbidden(format!(
"Reading from bucket {} not allowed for this key",
source_bucket
)));
}
let source_key = source_key.ok_or_bad_request("No source key specified")?;
let source_object = garage
.object_table
.get(&source_bucket_id, &source_key.to_string())
.await?
.ok_or(Error::NoSuchKey)?;
Ok(source_object)
}
fn extract_source_info(
source_object: &Object,
) -> Result<(&ObjectVersion, &ObjectVersionData, &ObjectVersionMeta), Error> {
let source_version = source_object
.versions()
.iter()
.rev()
.find(|v| v.is_complete())
.ok_or(Error::NoSuchKey)?;
let source_version_data = match &source_version.state {
ObjectVersionState::Complete(x) => x,
_ => unreachable!(),
};
let source_version_meta = match source_version_data {
ObjectVersionData::DeleteMarker => {
return Err(Error::NoSuchKey);
}
ObjectVersionData::Inline(meta, _bytes) => meta,
ObjectVersionData::FirstBlock(meta, _fbh) => meta,
};
Ok((source_version, source_version_data, source_version_meta))
}
struct CopyPreconditionHeaders {
copy_source_if_match: Option<Vec<String>>,
copy_source_if_modified_since: Option<SystemTime>,
copy_source_if_none_match: Option<Vec<String>>,
copy_source_if_unmodified_since: Option<SystemTime>,
}
impl CopyPreconditionHeaders {
fn parse(req: &Request<Body>) -> Result<Self, Error> {
Ok(Self {
copy_source_if_match: req
.headers()
.get("x-amz-copy-source-if-match")
.map(|x| x.to_str())
.transpose()?
.map(|x| {
x.split(',')
.map(|m| m.trim().trim_matches('"').to_string())
.collect::<Vec<_>>()
}),
copy_source_if_modified_since: req
.headers()
.get("x-amz-copy-source-if-modified-since")
.map(|x| x.to_str())
.transpose()?
.map(|x| httpdate::parse_http_date(x))
.transpose()
.ok_or_bad_request("Invalid date in x-amz-copy-source-if-modified-since")?,
copy_source_if_none_match: req
.headers()
.get("x-amz-copy-source-if-none-match")
.map(|x| x.to_str())
.transpose()?
.map(|x| {
x.split(',')
.map(|m| m.trim().trim_matches('"').to_string())
.collect::<Vec<_>>()
}),
copy_source_if_unmodified_since: req
.headers()
.get("x-amz-copy-source-if-unmodified-since")
.map(|x| x.to_str())
.transpose()?
.map(|x| httpdate::parse_http_date(x))
.transpose()
.ok_or_bad_request("Invalid date in x-amz-copy-source-if-unmodified-since")?,
})
}
fn check(&self, v: &ObjectVersion, etag: &str) -> Result<(), Error> {
let v_date = UNIX_EPOCH + Duration::from_millis(v.timestamp);
let ok = match (
&self.copy_source_if_match,
&self.copy_source_if_unmodified_since,
&self.copy_source_if_none_match,
&self.copy_source_if_modified_since,
) {
// TODO I'm not sure all of the conditions are evaluated correctly here
// If we have both if-match and if-unmodified-since,
// basically we don't care about if-unmodified-since,
// because in the spec it says that if if-match evaluates to
// true but if-unmodified-since evaluates to false,
// the copy is still done.
(Some(im), _, None, None) => im.iter().any(|x| x == etag || x == "*"),
(None, Some(ius), None, None) => v_date <= *ius,
// If we have both if-none-match and if-modified-since,
// then both of the two conditions must evaluate to true
(None, None, Some(inm), Some(ims)) => {
!inm.iter().any(|x| x == etag || x == "*") && v_date > *ims
}
(None, None, Some(inm), None) => !inm.iter().any(|x| x == etag || x == "*"),
(None, None, None, Some(ims)) => v_date > *ims,
(None, None, None, None) => true,
_ => {
return Err(Error::BadRequest(
"Invalid combination of x-amz-copy-source-if-xxxxx headers".into(),
))
}
};
if ok {
Ok(())
} else {
Err(Error::PreconditionFailed)
}
}
}
#[derive(Debug, Serialize, PartialEq)]
pub struct CopyObjectResult {
#[serde(rename = "LastModified")]
pub last_modified: s3_xml::Value,
#[serde(rename = "ETag")]
pub etag: s3_xml::Value,
}
#[derive(Debug, Serialize, PartialEq)]
pub struct CopyPartResult {
#[serde(serialize_with = "xmlns_tag")]
pub xmlns: (),
#[serde(rename = "LastModified")]
pub last_modified: s3_xml::Value,
#[serde(rename = "ETag")]
pub etag: s3_xml::Value,
}
#[cfg(test)]
mod tests {
use super::*;
use crate::s3_xml::to_xml_with_header;
#[test]
fn copy_object_result() -> Result<(), Error> {
let copy_result = CopyObjectResult {
last_modified: s3_xml::Value(msec_to_rfc3339(0)),
etag: s3_xml::Value("\"9b2cf535f27731c974343645a3985328\"".to_string()),
};
assert_eq!(
to_xml_with_header(&copy_result)?,
"<?xml version=\"1.0\" encoding=\"UTF-8\"?>\
<CopyObjectResult>\
<LastModified>1970-01-01T00:00:00.000Z</LastModified>\
<ETag>&quot;9b2cf535f27731c974343645a3985328&quot;</ETag>\
</CopyObjectResult>\
"
);
Ok(())
}
#[test]
fn serialize_copy_part_result() -> Result<(), Error> {
let expected_retval = "<?xml version=\"1.0\" encoding=\"UTF-8\"?>\
<CopyPartResult xmlns=\"http://s3.amazonaws.com/doc/2006-03-01/\">\
<LastModified>2011-04-11T20:34:56.000Z</LastModified>\
<ETag>&quot;9b2cf535f27731c974343645a3985328&quot;</ETag>\
</CopyPartResult>";
let v = CopyPartResult {
xmlns: (),
last_modified: s3_xml::Value("2011-04-11T20:34:56.000Z".into()),
etag: s3_xml::Value("\"9b2cf535f27731c974343645a3985328\"".into()),
};
println!("{}", to_xml_with_header(&v)?);
assert_eq!(to_xml_with_header(&v)?, expected_retval);
Ok(())
}
}

View file

@ -121,7 +121,7 @@ pub async fn handle_list(
key: uriencode_maybe(key, query.common.urlencode_resp), key: uriencode_maybe(key, query.common.urlencode_resp),
last_modified: s3_xml::Value(msec_to_rfc3339(info.last_modified)), last_modified: s3_xml::Value(msec_to_rfc3339(info.last_modified)),
size: s3_xml::IntValue(info.size as i64), size: s3_xml::IntValue(info.size as i64),
etag: s3_xml::Value(info.etag.to_string()), etag: s3_xml::Value(format!("\"{}\"", info.etag.to_string())),
storage_class: s3_xml::Value("STANDARD".to_string()), storage_class: s3_xml::Value("STANDARD".to_string()),
}) })
.collect(), .collect(),

View file

@ -370,12 +370,15 @@ pub async fn handle_put_part(
let key = key.to_string(); let key = key.to_string();
let mut chunker = BodyChunker::new(req.into_body(), garage.config.block_size); let mut chunker = BodyChunker::new(req.into_body(), garage.config.block_size);
let (object, first_block) = let (object, version, first_block) = futures::try_join!(
futures::try_join!(garage.object_table.get(&bucket_id, &key), chunker.next(),)?; garage.object_table.get(&bucket_id, &key),
garage.version_table.get(&version_uuid, &EmptyKey),
chunker.next()
)?;
// Check object is valid and multipart block can be accepted // Check object is valid and multipart block can be accepted
let first_block = first_block.ok_or_else(|| Error::BadRequest("Empty body".to_string()))?; let first_block = first_block.ok_or_bad_request("Empty body")?;
let object = object.ok_or_else(|| Error::BadRequest("Object not found".to_string()))?; let object = object.ok_or_bad_request("Object not found")?;
if !object if !object
.versions() .versions()
@ -385,6 +388,16 @@ pub async fn handle_put_part(
return Err(Error::NoSuchUpload); return Err(Error::NoSuchUpload);
} }
// Check part hasn't already been uploaded
if let Some(v) = version {
if v.has_part_number(part_number) {
return Err(Error::BadRequest(format!(
"Part number {} has already been uploaded",
part_number
)));
}
}
// Copy block to store // Copy block to store
let version = Version::new(version_uuid, bucket_id, key, false); let version = Version::new(version_uuid, bucket_id, key, false);
let first_block_hash = blake2sum(&first_block[..]); let first_block_hash = blake2sum(&first_block[..]);
@ -519,7 +532,7 @@ pub async fn handle_complete_multipart_upload(
location: None, location: None,
bucket: s3_xml::Value(bucket_name.to_string()), bucket: s3_xml::Value(bucket_name.to_string()),
key: s3_xml::Value(key), key: s3_xml::Value(key),
etag: s3_xml::Value(etag), etag: s3_xml::Value(format!("\"{}\"", etag)),
}; };
let xml = s3_xml::to_xml_with_header(&result)?; let xml = s3_xml::to_xml_with_header(&result)?;

View file

@ -107,14 +107,6 @@ pub struct DeleteResult {
pub errors: Vec<DeleteError>, pub errors: Vec<DeleteError>,
} }
#[derive(Debug, Serialize, PartialEq)]
pub struct CopyObjectResult {
#[serde(rename = "LastModified")]
pub last_modified: Value,
#[serde(rename = "ETag")]
pub etag: Value,
}
#[derive(Debug, Serialize, PartialEq)] #[derive(Debug, Serialize, PartialEq)]
pub struct InitiateMultipartUploadResult { pub struct InitiateMultipartUploadResult {
#[serde(serialize_with = "xmlns_tag")] #[serde(serialize_with = "xmlns_tag")]
@ -426,24 +418,6 @@ mod tests {
Ok(()) Ok(())
} }
#[test]
fn copy_object_result() -> Result<(), ApiError> {
let copy_result = CopyObjectResult {
last_modified: Value(msec_to_rfc3339(0)),
etag: Value("9b2cf535f27731c974343645a3985328".to_string()),
};
assert_eq!(
to_xml_with_header(&copy_result)?,
"<?xml version=\"1.0\" encoding=\"UTF-8\"?>\
<CopyObjectResult>\
<LastModified>1970-01-01T00:00:00.000Z</LastModified>\
<ETag>9b2cf535f27731c974343645a3985328</ETag>\
</CopyObjectResult>\
"
);
Ok(())
}
#[test] #[test]
fn initiate_multipart_upload_result() -> Result<(), ApiError> { fn initiate_multipart_upload_result() -> Result<(), ApiError> {
let result = InitiateMultipartUploadResult { let result = InitiateMultipartUploadResult {
@ -471,7 +445,7 @@ mod tests {
location: Some(Value("https://garage.tld/mybucket/a/plop".to_string())), location: Some(Value("https://garage.tld/mybucket/a/plop".to_string())),
bucket: Value("mybucket".to_string()), bucket: Value("mybucket".to_string()),
key: Value("a/plop".to_string()), key: Value("a/plop".to_string()),
etag: Value("3858f62230ac3c915f300c664312c11f-9".to_string()), etag: Value("\"3858f62230ac3c915f300c664312c11f-9\"".to_string()),
}; };
assert_eq!( assert_eq!(
to_xml_with_header(&result)?, to_xml_with_header(&result)?,
@ -480,7 +454,7 @@ mod tests {
<Location>https://garage.tld/mybucket/a/plop</Location>\ <Location>https://garage.tld/mybucket/a/plop</Location>\
<Bucket>mybucket</Bucket>\ <Bucket>mybucket</Bucket>\
<Key>a/plop</Key>\ <Key>a/plop</Key>\
<ETag>3858f62230ac3c915f300c664312c11f-9</ETag>\ <ETag>&quot;3858f62230ac3c915f300c664312c11f-9&quot;</ETag>\
</CompleteMultipartUploadResult>" </CompleteMultipartUploadResult>"
); );
Ok(()) Ok(())
@ -557,7 +531,7 @@ mod tests {
contents: vec![ListBucketItem { contents: vec![ListBucketItem {
key: Value("sample.jpg".to_string()), key: Value("sample.jpg".to_string()),
last_modified: Value(msec_to_rfc3339(0)), last_modified: Value(msec_to_rfc3339(0)),
etag: Value("bf1d737a4d46a19f3bced6905cc8b902".to_string()), etag: Value("\"bf1d737a4d46a19f3bced6905cc8b902\"".to_string()),
size: IntValue(142863), size: IntValue(142863),
storage_class: Value("STANDARD".to_string()), storage_class: Value("STANDARD".to_string()),
}], }],
@ -578,7 +552,7 @@ mod tests {
<Contents>\ <Contents>\
<Key>sample.jpg</Key>\ <Key>sample.jpg</Key>\
<LastModified>1970-01-01T00:00:00.000Z</LastModified>\ <LastModified>1970-01-01T00:00:00.000Z</LastModified>\
<ETag>bf1d737a4d46a19f3bced6905cc8b902</ETag>\ <ETag>&quot;bf1d737a4d46a19f3bced6905cc8b902&quot;</ETag>\
<Size>142863</Size>\ <Size>142863</Size>\
<StorageClass>STANDARD</StorageClass>\ <StorageClass>STANDARD</StorageClass>\
</Contents>\ </Contents>\
@ -656,7 +630,7 @@ mod tests {
contents: vec![ListBucketItem { contents: vec![ListBucketItem {
key: Value("ExampleObject.txt".to_string()), key: Value("ExampleObject.txt".to_string()),
last_modified: Value(msec_to_rfc3339(0)), last_modified: Value(msec_to_rfc3339(0)),
etag: Value("599bab3ed2c697f1d26842727561fd94".to_string()), etag: Value("\"599bab3ed2c697f1d26842727561fd94\"".to_string()),
size: IntValue(857), size: IntValue(857),
storage_class: Value("REDUCED_REDUNDANCY".to_string()), storage_class: Value("REDUCED_REDUNDANCY".to_string()),
}], }],
@ -674,7 +648,7 @@ mod tests {
<Contents>\ <Contents>\
<Key>ExampleObject.txt</Key>\ <Key>ExampleObject.txt</Key>\
<LastModified>1970-01-01T00:00:00.000Z</LastModified>\ <LastModified>1970-01-01T00:00:00.000Z</LastModified>\
<ETag>599bab3ed2c697f1d26842727561fd94</ETag>\ <ETag>&quot;599bab3ed2c697f1d26842727561fd94&quot;</ETag>\
<Size>857</Size>\ <Size>857</Size>\
<StorageClass>REDUCED_REDUNDANCY</StorageClass>\ <StorageClass>REDUCED_REDUNDANCY</StorageClass>\
</Contents>\ </Contents>\
@ -704,7 +678,7 @@ mod tests {
contents: vec![ListBucketItem { contents: vec![ListBucketItem {
key: Value("happyfacex.jpg".to_string()), key: Value("happyfacex.jpg".to_string()),
last_modified: Value(msec_to_rfc3339(0)), last_modified: Value(msec_to_rfc3339(0)),
etag: Value("70ee1738b6b21e2c8a43f3a5ab0eee71".to_string()), etag: Value("\"70ee1738b6b21e2c8a43f3a5ab0eee71\"".to_string()),
size: IntValue(1111), size: IntValue(1111),
storage_class: Value("STANDARD".to_string()), storage_class: Value("STANDARD".to_string()),
}], }],
@ -724,7 +698,7 @@ mod tests {
<Contents>\ <Contents>\
<Key>happyfacex.jpg</Key>\ <Key>happyfacex.jpg</Key>\
<LastModified>1970-01-01T00:00:00.000Z</LastModified>\ <LastModified>1970-01-01T00:00:00.000Z</LastModified>\
<ETag>70ee1738b6b21e2c8a43f3a5ab0eee71</ETag>\ <ETag>&quot;70ee1738b6b21e2c8a43f3a5ab0eee71&quot;</ETag>\
<Size>1111</Size>\ <Size>1111</Size>\
<StorageClass>STANDARD</StorageClass>\ <StorageClass>STANDARD</StorageClass>\
</Contents>\ </Contents>\

View file

@ -266,10 +266,13 @@ fn canonical_header_string(headers: &HashMap<String, String>, signed_headers: &s
let mut items = headers let mut items = headers
.iter() .iter()
.filter(|(key, _)| signed_headers_vec.contains(&key.as_str())) .filter(|(key, _)| signed_headers_vec.contains(&key.as_str()))
.map(|(key, value)| key.to_lowercase() + ":" + value.trim())
.collect::<Vec<_>>(); .collect::<Vec<_>>();
items.sort(); items.sort_by(|(k1, _), (k2, _)| k1.cmp(k2));
items.join("\n") items
.iter()
.map(|(key, value)| key.to_lowercase() + ":" + value.trim())
.collect::<Vec<_>>()
.join("\n")
} }
fn canonical_query_string(uri: &hyper::Uri) -> String { fn canonical_query_string(uri: &hyper::Uri) -> String {

View file

@ -47,6 +47,20 @@ impl Version {
key, key,
} }
} }
pub fn has_part_number(&self, part_number: u64) -> bool {
let case1 = self
.parts_etags
.items()
.binary_search_by(|(k, _)| k.cmp(&part_number))
.is_ok();
let case2 = self
.blocks
.items()
.binary_search_by(|(k, _)| k.part_number.cmp(&part_number))
.is_ok();
case1 || case2
}
} }
#[derive(PartialEq, Eq, Clone, Copy, Debug, Serialize, Deserialize)] #[derive(PartialEq, Eq, Clone, Copy, Debug, Serialize, Deserialize)]