|
|
/*
|
|
|
* Minio Cloud Storage, (C) 2017, 2018 Minio, Inc.
|
|
|
*
|
|
|
* Licensed under the Apache License, Version 2.0 (the "License");
|
|
|
* you may not use this file except in compliance with the License.
|
|
|
* You may obtain a copy of the License at
|
|
|
*
|
|
|
* http://www.apache.org/licenses/LICENSE-2.0
|
|
|
*
|
|
|
* Unless required by applicable law or agreed to in writing, software
|
|
|
* distributed under the License is distributed on an "AS IS" BASIS,
|
|
|
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
|
|
* See the License for the specific language governing permissions and
|
|
|
* limitations under the License.
|
|
|
*/
|
|
|
|
|
|
package oss
|
|
|
|
|
|
import (
|
|
|
"context"
|
|
|
"encoding/xml"
|
|
|
"fmt"
|
|
|
"io"
|
|
|
"net/http"
|
|
|
"strconv"
|
|
|
"strings"
|
|
|
|
|
|
"github.com/aliyun/aliyun-oss-go-sdk/oss"
|
|
|
"github.com/dustin/go-humanize"
|
|
|
|
|
|
"github.com/minio/cli"
|
|
|
miniogopolicy "github.com/minio/minio-go/pkg/policy"
|
|
|
minio "github.com/minio/minio/cmd"
|
|
|
"github.com/minio/minio/cmd/logger"
|
|
|
"github.com/minio/minio/pkg/auth"
|
|
|
"github.com/minio/minio/pkg/hash"
|
|
|
"github.com/minio/minio/pkg/policy"
|
|
|
"github.com/minio/minio/pkg/policy/condition"
|
|
|
)
|
|
|
|
|
|
const (
|
|
|
ossS3MinPartSize = 5 * humanize.MiByte
|
|
|
ossMaxParts = 1000
|
|
|
ossMaxKeys = 1000
|
|
|
ossBackend = "oss"
|
|
|
)
|
|
|
|
|
|
func init() {
|
|
|
const ossGatewayTemplate = `NAME:
|
|
|
{{.HelpName}} - {{.Usage}}
|
|
|
|
|
|
USAGE:
|
|
|
{{.HelpName}} {{if .VisibleFlags}}[FLAGS]{{end}} [ENDPOINT]
|
|
|
{{if .VisibleFlags}}
|
|
|
FLAGS:
|
|
|
{{range .VisibleFlags}}{{.}}
|
|
|
{{end}}{{end}}
|
|
|
ENDPOINT:
|
|
|
OSS server endpoint. Default ENDPOINT is https://oss.aliyuncs.com
|
|
|
|
|
|
ENVIRONMENT VARIABLES:
|
|
|
ACCESS:
|
|
|
MINIO_ACCESS_KEY: Username or access key of OSS storage.
|
|
|
MINIO_SECRET_KEY: Password or secret key of OSS storage.
|
|
|
|
|
|
BROWSER:
|
|
|
MINIO_BROWSER: To disable web browser access, set this value to "off".
|
|
|
|
|
|
DOMAIN:
|
|
|
MINIO_DOMAIN: To enable virtual-host-style requests, set this value to Minio host domain name.
|
|
|
|
|
|
CACHE:
|
|
|
MINIO_CACHE_DRIVES: List of mounted drives or directories delimited by ";".
|
|
|
MINIO_CACHE_EXCLUDE: List of cache exclusion patterns delimited by ";".
|
|
|
MINIO_CACHE_EXPIRY: Cache expiry duration in days.
|
|
|
MINIO_CACHE_MAXUSE: Maximum permitted usage of the cache in percentage (0-100).
|
|
|
|
|
|
EXAMPLES:
|
|
|
1. Start minio gateway server for Aliyun OSS backend.
|
|
|
$ export MINIO_ACCESS_KEY=accesskey
|
|
|
$ export MINIO_SECRET_KEY=secretkey
|
|
|
$ {{.HelpName}}
|
|
|
|
|
|
2. Start minio gateway server for Aliyun OSS backend on custom endpoint.
|
|
|
$ export MINIO_ACCESS_KEY=Q3AM3UQ867SPQQA43P2F
|
|
|
$ export MINIO_SECRET_KEY=zuf+tfteSlswRu7BJ86wekitnifILbZam1KYY3TG
|
|
|
$ {{.HelpName}} https://oss.example.com
|
|
|
|
|
|
3. Start minio gateway server for Aliyun OSS backend with edge caching enabled.
|
|
|
$ export MINIO_ACCESS_KEY=accesskey
|
|
|
$ export MINIO_SECRET_KEY=secretkey
|
|
|
$ export MINIO_CACHE_DRIVES="/mnt/drive1;/mnt/drive2;/mnt/drive3;/mnt/drive4"
|
|
|
$ export MINIO_CACHE_EXCLUDE="bucket1/*;*.png"
|
|
|
$ export MINIO_CACHE_EXPIRY=40
|
|
|
$ export MINIO_CACHE_MAXUSE=80
|
|
|
$ {{.HelpName}}
|
|
|
`
|
|
|
|
|
|
minio.RegisterGatewayCommand(cli.Command{
|
|
|
Name: "oss",
|
|
|
Usage: "Alibaba Cloud (Aliyun) Object Storage Service (OSS)",
|
|
|
Action: ossGatewayMain,
|
|
|
CustomHelpTemplate: ossGatewayTemplate,
|
|
|
HideHelpCommand: true,
|
|
|
})
|
|
|
}
|
|
|
|
|
|
// Handler for 'minio gateway oss' command line.
|
|
|
func ossGatewayMain(ctx *cli.Context) {
|
|
|
if ctx.Args().First() == "help" {
|
|
|
cli.ShowCommandHelpAndExit(ctx, ossBackend, 1)
|
|
|
}
|
|
|
|
|
|
// Validate gateway arguments.
|
|
|
host := ctx.Args().First()
|
|
|
logger.FatalIf(minio.ValidateGatewayArguments(ctx.GlobalString("address"), host), "Invalid argument")
|
|
|
|
|
|
minio.StartGateway(ctx, &OSS{host})
|
|
|
}
|
|
|
|
|
|
// OSS implements Gateway.
|
|
|
type OSS struct {
|
|
|
host string
|
|
|
}
|
|
|
|
|
|
// Name implements Gateway interface.
|
|
|
func (g *OSS) Name() string {
|
|
|
return ossBackend
|
|
|
}
|
|
|
|
|
|
// NewGatewayLayer implements Gateway interface and returns OSS ObjectLayer.
|
|
|
func (g *OSS) NewGatewayLayer(creds auth.Credentials) (minio.ObjectLayer, error) {
|
|
|
var err error
|
|
|
|
|
|
// Regions and endpoints
|
|
|
// https://www.alibabacloud.com/help/doc-detail/31837.htm
|
|
|
if g.host == "" {
|
|
|
g.host = "https://oss.aliyuncs.com"
|
|
|
}
|
|
|
|
|
|
// Initialize oss client object.
|
|
|
client, err := oss.New(g.host, creds.AccessKey, creds.SecretKey)
|
|
|
if err != nil {
|
|
|
return nil, err
|
|
|
}
|
|
|
|
|
|
return &ossObjects{
|
|
|
Client: client,
|
|
|
}, nil
|
|
|
}
|
|
|
|
|
|
// Production - oss is production ready.
|
|
|
func (g *OSS) Production() bool {
|
|
|
return true
|
|
|
}
|
|
|
|
|
|
// appendS3MetaToOSSOptions converts metadata meant for S3 PUT/COPY
|
|
|
// object into oss.Option.
|
|
|
//
|
|
|
// S3 user-metadata is translated to OSS metadata by removing the
|
|
|
// `X-Amz-Meta-` prefix and converted into `X-Oss-Meta-`.
|
|
|
//
|
|
|
// Header names are canonicalized as in http.Header.
|
|
|
func appendS3MetaToOSSOptions(ctx context.Context, opts []oss.Option, s3Metadata map[string]string) ([]oss.Option, error) {
|
|
|
if opts == nil {
|
|
|
opts = make([]oss.Option, 0, len(s3Metadata))
|
|
|
}
|
|
|
for k, v := range s3Metadata {
|
|
|
k = http.CanonicalHeaderKey(k)
|
|
|
|
|
|
switch {
|
|
|
case strings.HasPrefix(k, "X-Amz-Meta-"):
|
|
|
metaKey := k[len("X-Amz-Meta-"):]
|
|
|
// NOTE(timonwong): OSS won't allow headers with underscore(_).
|
|
|
if strings.Contains(metaKey, "_") {
|
|
|
logger.LogIf(ctx, minio.UnsupportedMetadata{})
|
|
|
return nil, minio.UnsupportedMetadata{}
|
|
|
}
|
|
|
opts = append(opts, oss.Meta(metaKey, v))
|
|
|
case k == "X-Amz-Acl":
|
|
|
// Valid values: public-read, private, and public-read-write
|
|
|
opts = append(opts, oss.ObjectACL(oss.ACLType(v)))
|
|
|
case k == "X-Amz-Server-Side-Encryption":
|
|
|
opts = append(opts, oss.ServerSideEncryption(v))
|
|
|
case k == "X-Amz-Copy-Source-If-Match":
|
|
|
opts = append(opts, oss.CopySourceIfMatch(v))
|
|
|
case k == "X-Amz-Copy-Source-If-None-Match":
|
|
|
opts = append(opts, oss.CopySourceIfNoneMatch(v))
|
|
|
case k == "X-Amz-Copy-Source-If-Unmodified-Since":
|
|
|
if v, err := http.ParseTime(v); err == nil {
|
|
|
opts = append(opts, oss.CopySourceIfUnmodifiedSince(v))
|
|
|
}
|
|
|
case k == "X-Amz-Copy-Source-If-Modified-Since":
|
|
|
if v, err := http.ParseTime(v); err == nil {
|
|
|
opts = append(opts, oss.CopySourceIfModifiedSince(v))
|
|
|
}
|
|
|
case k == "Accept-Encoding":
|
|
|
opts = append(opts, oss.AcceptEncoding(v))
|
|
|
case k == "Cache-Control":
|
|
|
opts = append(opts, oss.CacheControl(v))
|
|
|
case k == "Content-Disposition":
|
|
|
opts = append(opts, oss.ContentDisposition(v))
|
|
|
case k == "Content-Encoding":
|
|
|
opts = append(opts, oss.ContentEncoding(v))
|
|
|
case k == "Content-Length":
|
|
|
if v, err := strconv.ParseInt(v, 10, 64); err == nil {
|
|
|
opts = append(opts, oss.ContentLength(v))
|
|
|
}
|
|
|
case k == "Content-MD5":
|
|
|
opts = append(opts, oss.ContentMD5(v))
|
|
|
case k == "Content-Type":
|
|
|
opts = append(opts, oss.ContentType(v))
|
|
|
case k == "Expires":
|
|
|
if v, err := http.ParseTime(v); err == nil {
|
|
|
opts = append(opts, oss.Expires(v))
|
|
|
}
|
|
|
}
|
|
|
}
|
|
|
|
|
|
return opts, nil
|
|
|
}
|
|
|
|
|
|
// ossMetaToS3Meta converts OSS metadata to S3 metadata.
|
|
|
// It is the reverse of appendS3MetaToOSSOptions.
|
|
|
func ossHeaderToS3Meta(header http.Header) map[string]string {
|
|
|
// Decoding technique for each key is used here is as follows
|
|
|
// Each '_' is converted to '-'
|
|
|
// Each '__' is converted to '_'
|
|
|
// With this basic assumption here are some of the expected
|
|
|
// translations for these keys.
|
|
|
// i: 'x_s3cmd__attrs' -> o: 'x-s3cmd_attrs' (mixed)
|
|
|
// i: 'x____test____value' -> o: 'x__test__value' (double '_')
|
|
|
decodeKey := func(key string) string {
|
|
|
tokens := strings.Split(key, "__")
|
|
|
for i := range tokens {
|
|
|
tokens[i] = strings.Replace(tokens[i], "_", "-", -1)
|
|
|
}
|
|
|
return strings.Join(tokens, "_")
|
|
|
}
|
|
|
|
|
|
s3Metadata := make(map[string]string)
|
|
|
for k := range header {
|
|
|
k = http.CanonicalHeaderKey(k)
|
|
|
switch {
|
|
|
case strings.HasPrefix(k, oss.HTTPHeaderOssMetaPrefix):
|
|
|
// Add amazon s3 meta prefix
|
|
|
metaKey := k[len(oss.HTTPHeaderOssMetaPrefix):]
|
|
|
metaKey = "X-Amz-Meta-" + decodeKey(metaKey)
|
|
|
metaKey = http.CanonicalHeaderKey(metaKey)
|
|
|
s3Metadata[metaKey] = header.Get(k)
|
|
|
case k == "Cache-Control":
|
|
|
fallthrough
|
|
|
case k == "Content-Encoding":
|
|
|
fallthrough
|
|
|
case k == "Content-Disposition":
|
|
|
fallthrough
|
|
|
case k == "Content-Length":
|
|
|
fallthrough
|
|
|
case k == "Content-MD5":
|
|
|
fallthrough
|
|
|
case k == "Content-Type":
|
|
|
s3Metadata[k] = header.Get(k)
|
|
|
}
|
|
|
}
|
|
|
|
|
|
return s3Metadata
|
|
|
}
|
|
|
|
|
|
// ossToObjectError converts OSS errors to minio object layer errors.
|
|
|
func ossToObjectError(err error, params ...string) error {
|
|
|
if err == nil {
|
|
|
return nil
|
|
|
}
|
|
|
|
|
|
bucket := ""
|
|
|
object := ""
|
|
|
uploadID := ""
|
|
|
switch len(params) {
|
|
|
case 3:
|
|
|
uploadID = params[2]
|
|
|
fallthrough
|
|
|
case 2:
|
|
|
object = params[1]
|
|
|
fallthrough
|
|
|
case 1:
|
|
|
bucket = params[0]
|
|
|
}
|
|
|
|
|
|
ossErr, ok := err.(oss.ServiceError)
|
|
|
if !ok {
|
|
|
// We don't interpret non OSS errors. As oss errors will
|
|
|
// have StatusCode to help to convert to object errors.
|
|
|
return err
|
|
|
}
|
|
|
|
|
|
switch ossErr.Code {
|
|
|
case "BucketAlreadyExists":
|
|
|
err = minio.BucketAlreadyOwnedByYou{Bucket: bucket}
|
|
|
case "BucketNotEmpty":
|
|
|
err = minio.BucketNotEmpty{Bucket: bucket}
|
|
|
case "InvalidBucketName":
|
|
|
err = minio.BucketNameInvalid{Bucket: bucket}
|
|
|
case "NoSuchBucket":
|
|
|
err = minio.BucketNotFound{Bucket: bucket}
|
|
|
case "NoSuchKey":
|
|
|
if object != "" {
|
|
|
err = minio.ObjectNotFound{Bucket: bucket, Object: object}
|
|
|
} else {
|
|
|
err = minio.BucketNotFound{Bucket: bucket}
|
|
|
}
|
|
|
case "InvalidObjectName":
|
|
|
err = minio.ObjectNameInvalid{Bucket: bucket, Object: object}
|
|
|
case "AccessDenied":
|
|
|
err = minio.PrefixAccessDenied{Bucket: bucket, Object: object}
|
|
|
case "NoSuchUpload":
|
|
|
err = minio.InvalidUploadID{UploadID: uploadID}
|
|
|
case "EntityTooSmall":
|
|
|
err = minio.PartTooSmall{}
|
|
|
case "SignatureDoesNotMatch":
|
|
|
err = minio.SignatureDoesNotMatch{}
|
|
|
case "InvalidPart":
|
|
|
err = minio.InvalidPart{}
|
|
|
}
|
|
|
|
|
|
return err
|
|
|
}
|
|
|
|
|
|
// ossObjects implements gateway for Aliyun Object Storage Service.
|
|
|
type ossObjects struct {
|
|
|
minio.GatewayUnsupported
|
|
|
Client *oss.Client
|
|
|
}
|
|
|
|
|
|
// Shutdown saves any gateway metadata to disk
|
|
|
// if necessary and reload upon next restart.
|
|
|
func (l *ossObjects) Shutdown(ctx context.Context) error {
|
|
|
return nil
|
|
|
}
|
|
|
|
|
|
// StorageInfo is not relevant to OSS backend.
|
|
|
func (l *ossObjects) StorageInfo(ctx context.Context) (si minio.StorageInfo) {
|
|
|
return
|
|
|
}
|
|
|
|
|
|
// ossIsValidBucketName verifies whether a bucket name is valid.
|
|
|
func ossIsValidBucketName(bucket string) bool {
|
|
|
// dot is not allowed in bucket name
|
|
|
if strings.Contains(bucket, ".") {
|
|
|
return false
|
|
|
}
|
|
|
if !minio.IsValidBucketName(bucket) {
|
|
|
return false
|
|
|
}
|
|
|
return true
|
|
|
}
|
|
|
|
|
|
// MakeBucketWithLocation creates a new container on OSS backend.
|
|
|
func (l *ossObjects) MakeBucketWithLocation(ctx context.Context, bucket, location string) error {
|
|
|
if !ossIsValidBucketName(bucket) {
|
|
|
logger.LogIf(ctx, minio.BucketNameInvalid{Bucket: bucket})
|
|
|
return minio.BucketNameInvalid{Bucket: bucket}
|
|
|
}
|
|
|
|
|
|
err := l.Client.CreateBucket(bucket)
|
|
|
logger.LogIf(ctx, err)
|
|
|
return ossToObjectError(err, bucket)
|
|
|
}
|
|
|
|
|
|
// ossGeBucketInfo gets bucket metadata.
|
|
|
func ossGeBucketInfo(ctx context.Context, client *oss.Client, bucket string) (bi minio.BucketInfo, err error) {
|
|
|
bgir, err := client.GetBucketInfo(bucket)
|
|
|
if err != nil {
|
|
|
logger.LogIf(ctx, err)
|
|
|
return bi, ossToObjectError(err, bucket)
|
|
|
}
|
|
|
|
|
|
return minio.BucketInfo{
|
|
|
Name: bgir.BucketInfo.Name,
|
|
|
Created: bgir.BucketInfo.CreationDate,
|
|
|
}, nil
|
|
|
}
|
|
|
|
|
|
// GetBucketInfo gets bucket metadata.
|
|
|
func (l *ossObjects) GetBucketInfo(ctx context.Context, bucket string) (bi minio.BucketInfo, err error) {
|
|
|
return ossGeBucketInfo(ctx, l.Client, bucket)
|
|
|
}
|
|
|
|
|
|
// ListBuckets lists all OSS buckets.
|
|
|
func (l *ossObjects) ListBuckets(ctx context.Context) (buckets []minio.BucketInfo, err error) {
|
|
|
marker := oss.Marker("")
|
|
|
for {
|
|
|
lbr, err := l.Client.ListBuckets(marker)
|
|
|
if err != nil {
|
|
|
logger.LogIf(ctx, err)
|
|
|
return nil, ossToObjectError(err)
|
|
|
}
|
|
|
|
|
|
for _, bi := range lbr.Buckets {
|
|
|
buckets = append(buckets, minio.BucketInfo{
|
|
|
Name: bi.Name,
|
|
|
Created: bi.CreationDate,
|
|
|
})
|
|
|
}
|
|
|
|
|
|
marker = oss.Marker(lbr.NextMarker)
|
|
|
if !lbr.IsTruncated {
|
|
|
break
|
|
|
}
|
|
|
}
|
|
|
|
|
|
return buckets, nil
|
|
|
}
|
|
|
|
|
|
// DeleteBucket deletes a bucket on OSS.
|
|
|
func (l *ossObjects) DeleteBucket(ctx context.Context, bucket string) error {
|
|
|
err := l.Client.DeleteBucket(bucket)
|
|
|
if err != nil {
|
|
|
logger.LogIf(ctx, err)
|
|
|
return ossToObjectError(err, bucket)
|
|
|
}
|
|
|
return nil
|
|
|
}
|
|
|
|
|
|
// fromOSSClientObjectProperties converts oss ObjectProperties to ObjectInfo.
|
|
|
func fromOSSClientObjectProperties(bucket string, o oss.ObjectProperties) minio.ObjectInfo {
|
|
|
// NOTE(timonwong): No Content-Type and user defined metadata.
|
|
|
// https://www.alibabacloud.com/help/doc-detail/31965.htm
|
|
|
|
|
|
return minio.ObjectInfo{
|
|
|
Bucket: bucket,
|
|
|
Name: o.Key,
|
|
|
ModTime: o.LastModified,
|
|
|
Size: o.Size,
|
|
|
ETag: minio.ToS3ETag(o.ETag),
|
|
|
}
|
|
|
}
|
|
|
|
|
|
// fromOSSClientListObjectsResult converts oss ListBucketResult to ListObjectsInfo.
|
|
|
func fromOSSClientListObjectsResult(bucket string, lor oss.ListObjectsResult) minio.ListObjectsInfo {
|
|
|
objects := make([]minio.ObjectInfo, len(lor.Objects))
|
|
|
for i, oi := range lor.Objects {
|
|
|
objects[i] = fromOSSClientObjectProperties(bucket, oi)
|
|
|
}
|
|
|
|
|
|
prefixes := make([]string, len(lor.CommonPrefixes))
|
|
|
copy(prefixes, lor.CommonPrefixes)
|
|
|
|
|
|
return minio.ListObjectsInfo{
|
|
|
IsTruncated: lor.IsTruncated,
|
|
|
NextMarker: lor.NextMarker,
|
|
|
Objects: objects,
|
|
|
Prefixes: prefixes,
|
|
|
}
|
|
|
}
|
|
|
|
|
|
// ossListObjects lists all blobs in OSS bucket filtered by prefix.
|
|
|
func ossListObjects(ctx context.Context, client *oss.Client, bucket, prefix, marker, delimiter string, maxKeys int) (loi minio.ListObjectsInfo, err error) {
|
|
|
buck, err := client.Bucket(bucket)
|
|
|
if err != nil {
|
|
|
logger.LogIf(ctx, err)
|
|
|
return loi, ossToObjectError(err, bucket)
|
|
|
}
|
|
|
|
|
|
// maxKeys should default to 1000 or less.
|
|
|
if maxKeys == 0 || maxKeys > ossMaxKeys {
|
|
|
maxKeys = ossMaxKeys
|
|
|
}
|
|
|
|
|
|
lor, err := buck.ListObjects(oss.Prefix(prefix), oss.Marker(marker), oss.Delimiter(delimiter), oss.MaxKeys(maxKeys))
|
|
|
if err != nil {
|
|
|
logger.LogIf(ctx, err)
|
|
|
return loi, ossToObjectError(err, bucket)
|
|
|
}
|
|
|
|
|
|
return fromOSSClientListObjectsResult(bucket, lor), nil
|
|
|
}
|
|
|
|
|
|
// ossListObjectsV2 lists all blobs in OSS bucket filtered by prefix.
|
|
|
func ossListObjectsV2(ctx context.Context, client *oss.Client, bucket, prefix, continuationToken, delimiter string, maxKeys int,
|
|
|
fetchOwner bool, startAfter string) (loi minio.ListObjectsV2Info, err error) {
|
|
|
// fetchOwner is not supported and unused.
|
|
|
marker := continuationToken
|
|
|
if marker == "" {
|
|
|
marker = startAfter
|
|
|
}
|
|
|
|
|
|
resultV1, err := ossListObjects(ctx, client, bucket, prefix, marker, delimiter, maxKeys)
|
|
|
if err != nil {
|
|
|
return loi, err
|
|
|
}
|
|
|
|
|
|
return minio.ListObjectsV2Info{
|
|
|
Objects: resultV1.Objects,
|
|
|
Prefixes: resultV1.Prefixes,
|
|
|
ContinuationToken: continuationToken,
|
|
|
NextContinuationToken: resultV1.NextMarker,
|
|
|
IsTruncated: resultV1.IsTruncated,
|
|
|
}, nil
|
|
|
}
|
|
|
|
|
|
// ListObjects lists all blobs in OSS bucket filtered by prefix.
|
|
|
func (l *ossObjects) ListObjects(ctx context.Context, bucket, prefix, marker, delimiter string, maxKeys int) (loi minio.ListObjectsInfo, err error) {
|
|
|
return ossListObjects(ctx, l.Client, bucket, prefix, marker, delimiter, maxKeys)
|
|
|
}
|
|
|
|
|
|
// ListObjectsV2 lists all blobs in OSS bucket filtered by prefix
|
|
|
func (l *ossObjects) ListObjectsV2(ctx context.Context, bucket, prefix, continuationToken, delimiter string, maxKeys int,
|
|
|
fetchOwner bool, startAfter string) (loi minio.ListObjectsV2Info, err error) {
|
|
|
return ossListObjectsV2(ctx, l.Client, bucket, prefix, continuationToken, delimiter, maxKeys, fetchOwner, startAfter)
|
|
|
}
|
|
|
|
|
|
// ossGetObject reads an object on OSS. Supports additional
|
|
|
// parameters like offset and length which are synonymous with
|
|
|
// HTTP Range requests.
|
|
|
//
|
|
|
// startOffset indicates the starting read location of the object.
|
|
|
// length indicates the total length of the object.
|
|
|
func ossGetObject(ctx context.Context, client *oss.Client, bucket, key string, startOffset, length int64, writer io.Writer, etag string) error {
|
|
|
if length < 0 && length != -1 {
|
|
|
logger.LogIf(ctx, fmt.Errorf("Invalid argument"))
|
|
|
return ossToObjectError(fmt.Errorf("Invalid argument"), bucket, key)
|
|
|
}
|
|
|
|
|
|
bkt, err := client.Bucket(bucket)
|
|
|
if err != nil {
|
|
|
logger.LogIf(ctx, err)
|
|
|
return ossToObjectError(err, bucket, key)
|
|
|
}
|
|
|
|
|
|
var opts []oss.Option
|
|
|
if startOffset >= 0 && length >= 0 {
|
|
|
opts = append(opts, oss.Range(startOffset, startOffset+length-1))
|
|
|
}
|
|
|
|
|
|
object, err := bkt.GetObject(key, opts...)
|
|
|
if err != nil {
|
|
|
logger.LogIf(ctx, err)
|
|
|
return ossToObjectError(err, bucket, key)
|
|
|
}
|
|
|
defer object.Close()
|
|
|
|
|
|
if _, err := io.Copy(writer, object); err != nil {
|
|
|
logger.LogIf(ctx, err)
|
|
|
return ossToObjectError(err, bucket, key)
|
|
|
}
|
|
|
return nil
|
|
|
}
|
|
|
|
|
|
// GetObjectNInfo - returns object info and locked object ReadCloser
|
|
|
func (l *ossObjects) GetObjectNInfo(ctx context.Context, bucket, object string, rs *minio.HTTPRangeSpec, h http.Header, lockType minio.LockType, opts minio.ObjectOptions) (gr *minio.GetObjectReader, err error) {
|
|
|
var objInfo minio.ObjectInfo
|
|
|
objInfo, err = l.GetObjectInfo(ctx, bucket, object, opts)
|
|
|
if err != nil {
|
|
|
return nil, err
|
|
|
}
|
|
|
|
|
|
var startOffset, length int64
|
|
|
startOffset, length, err = rs.GetOffsetLength(objInfo.Size)
|
|
|
if err != nil {
|
|
|
return nil, err
|
|
|
}
|
|
|
|
|
|
pr, pw := io.Pipe()
|
|
|
go func() {
|
|
|
err := l.GetObject(ctx, bucket, object, startOffset, length, pw, objInfo.ETag, opts)
|
|
|
pw.CloseWithError(err)
|
|
|
}()
|
|
|
// Setup cleanup function to cause the above go-routine to
|
|
|
// exit in case of partial read
|
|
|
pipeCloser := func() { pr.Close() }
|
|
|
return minio.NewGetObjectReaderFromReader(pr, objInfo, pipeCloser), nil
|
|
|
}
|
|
|
|
|
|
// GetObject reads an object on OSS. Supports additional
|
|
|
// parameters like offset and length which are synonymous with
|
|
|
// HTTP Range requests.
|
|
|
//
|
|
|
// startOffset indicates the starting read location of the object.
|
|
|
// length indicates the total length of the object.
|
|
|
func (l *ossObjects) GetObject(ctx context.Context, bucket, key string, startOffset, length int64, writer io.Writer, etag string, opts minio.ObjectOptions) error {
|
|
|
return ossGetObject(ctx, l.Client, bucket, key, startOffset, length, writer, etag)
|
|
|
}
|
|
|
|
|
|
func translatePlainError(err error) error {
|
|
|
errString := err.Error()
|
|
|
|
|
|
switch errString {
|
|
|
case "oss: service returned without a response body (404 Not Found)":
|
|
|
return oss.ServiceError{Code: "NoSuchKey"}
|
|
|
case "oss: service returned without a response body (400 Bad Request)":
|
|
|
return oss.ServiceError{Code: "AccessDenied"}
|
|
|
}
|
|
|
|
|
|
return err
|
|
|
}
|
|
|
|
|
|
// ossGetObjectInfo reads object info and replies back ObjectInfo.
|
|
|
func ossGetObjectInfo(ctx context.Context, client *oss.Client, bucket, object string) (objInfo minio.ObjectInfo, err error) {
|
|
|
bkt, err := client.Bucket(bucket)
|
|
|
if err != nil {
|
|
|
logger.LogIf(ctx, err)
|
|
|
return objInfo, ossToObjectError(err, bucket, object)
|
|
|
}
|
|
|
|
|
|
header, err := bkt.GetObjectDetailedMeta(object)
|
|
|
if err != nil {
|
|
|
logger.LogIf(ctx, translatePlainError(err))
|
|
|
return objInfo, ossToObjectError(translatePlainError(err), bucket, object)
|
|
|
}
|
|
|
|
|
|
// Build S3 metadata from OSS metadata
|
|
|
userDefined := ossHeaderToS3Meta(header)
|
|
|
|
|
|
modTime, _ := http.ParseTime(header.Get("Last-Modified"))
|
|
|
size, _ := strconv.ParseInt(header.Get("Content-Length"), 10, 64)
|
|
|
|
|
|
return minio.ObjectInfo{
|
|
|
Bucket: bucket,
|
|
|
Name: object,
|
|
|
ModTime: modTime,
|
|
|
Size: size,
|
|
|
ETag: minio.ToS3ETag(header.Get("ETag")),
|
|
|
UserDefined: userDefined,
|
|
|
ContentType: header.Get("Content-Type"),
|
|
|
ContentEncoding: header.Get("Content-Encoding"),
|
|
|
}, nil
|
|
|
}
|
|
|
|
|
|
// GetObjectInfo reads object info and replies back ObjectInfo.
|
|
|
func (l *ossObjects) GetObjectInfo(ctx context.Context, bucket, object string, opts minio.ObjectOptions) (objInfo minio.ObjectInfo, err error) {
|
|
|
return ossGetObjectInfo(ctx, l.Client, bucket, object)
|
|
|
}
|
|
|
|
|
|
// ossPutObject creates a new object with the incoming data.
|
|
|
func ossPutObject(ctx context.Context, client *oss.Client, bucket, object string, data *hash.Reader, metadata map[string]string) (objInfo minio.ObjectInfo, err error) {
|
|
|
bkt, err := client.Bucket(bucket)
|
|
|
if err != nil {
|
|
|
logger.LogIf(ctx, err)
|
|
|
return objInfo, ossToObjectError(err, bucket, object)
|
|
|
}
|
|
|
|
|
|
// Build OSS metadata
|
|
|
opts, err := appendS3MetaToOSSOptions(ctx, nil, metadata)
|
|
|
if err != nil {
|
|
|
return objInfo, ossToObjectError(err, bucket, object)
|
|
|
}
|
|
|
|
|
|
err = bkt.PutObject(object, data, opts...)
|
|
|
if err != nil {
|
|
|
logger.LogIf(ctx, err)
|
|
|
return objInfo, ossToObjectError(err, bucket, object)
|
|
|
}
|
|
|
|
|
|
return ossGetObjectInfo(ctx, client, bucket, object)
|
|
|
}
|
|
|
|
|
|
// PutObject creates a new object with the incoming data.
|
|
|
func (l *ossObjects) PutObject(ctx context.Context, bucket, object string, r *minio.PutObjReader, metadata map[string]string, opts minio.ObjectOptions) (objInfo minio.ObjectInfo, err error) {
|
|
|
data := r.Reader
|
|
|
|
|
|
return ossPutObject(ctx, l.Client, bucket, object, data, metadata)
|
|
|
}
|
|
|
|
|
|
// CopyObject copies an object from source bucket to a destination bucket.
|
|
|
func (l *ossObjects) CopyObject(ctx context.Context, srcBucket, srcObject, dstBucket, dstObject string, srcInfo minio.ObjectInfo, srcOpts, dstOpts minio.ObjectOptions) (objInfo minio.ObjectInfo, err error) {
|
|
|
bkt, err := l.Client.Bucket(srcBucket)
|
|
|
if err != nil {
|
|
|
logger.LogIf(ctx, err)
|
|
|
return objInfo, ossToObjectError(err, srcBucket, srcObject)
|
|
|
}
|
|
|
|
|
|
opts := make([]oss.Option, 0, len(srcInfo.UserDefined)+1)
|
|
|
// Set this header such that following CopyObject() always sets the right metadata on the destination.
|
|
|
// metadata input is already a trickled down value from interpreting x-oss-metadata-directive at
|
|
|
// handler layer. So what we have right now is supposed to be applied on the destination object anyways.
|
|
|
// So preserve it by adding "REPLACE" directive to save all the metadata set by CopyObject API.
|
|
|
opts = append(opts, oss.MetadataDirective(oss.MetaReplace))
|
|
|
|
|
|
// Build OSS metadata
|
|
|
opts, err = appendS3MetaToOSSOptions(ctx, opts, srcInfo.UserDefined)
|
|
|
if err != nil {
|
|
|
return objInfo, ossToObjectError(err, srcBucket, srcObject)
|
|
|
}
|
|
|
|
|
|
if _, err = bkt.CopyObjectTo(dstBucket, dstObject, srcObject, opts...); err != nil {
|
|
|
logger.LogIf(ctx, err)
|
|
|
return objInfo, ossToObjectError(err, srcBucket, srcObject)
|
|
|
}
|
|
|
return l.GetObjectInfo(ctx, dstBucket, dstObject, dstOpts)
|
|
|
}
|
|
|
|
|
|
// DeleteObject deletes a blob in bucket.
|
|
|
func (l *ossObjects) DeleteObject(ctx context.Context, bucket, object string) error {
|
|
|
bkt, err := l.Client.Bucket(bucket)
|
|
|
if err != nil {
|
|
|
logger.LogIf(ctx, err)
|
|
|
return ossToObjectError(err, bucket, object)
|
|
|
}
|
|
|
|
|
|
err = bkt.DeleteObject(object)
|
|
|
if err != nil {
|
|
|
logger.LogIf(ctx, err)
|
|
|
return ossToObjectError(err, bucket, object)
|
|
|
}
|
|
|
return nil
|
|
|
}
|
|
|
|
|
|
// fromOSSClientListMultipartsInfo converts oss ListMultipartUploadResult to ListMultipartsInfo
|
|
|
func fromOSSClientListMultipartsInfo(lmur oss.ListMultipartUploadResult) minio.ListMultipartsInfo {
|
|
|
uploads := make([]minio.MultipartInfo, len(lmur.Uploads))
|
|
|
for i, um := range lmur.Uploads {
|
|
|
uploads[i] = minio.MultipartInfo{
|
|
|
Object: um.Key,
|
|
|
UploadID: um.UploadID,
|
|
|
Initiated: um.Initiated,
|
|
|
}
|
|
|
}
|
|
|
|
|
|
commonPrefixes := make([]string, len(lmur.CommonPrefixes))
|
|
|
copy(commonPrefixes, lmur.CommonPrefixes)
|
|
|
|
|
|
return minio.ListMultipartsInfo{
|
|
|
KeyMarker: lmur.KeyMarker,
|
|
|
UploadIDMarker: lmur.UploadIDMarker,
|
|
|
NextKeyMarker: lmur.NextKeyMarker,
|
|
|
NextUploadIDMarker: lmur.NextUploadIDMarker,
|
|
|
MaxUploads: lmur.MaxUploads,
|
|
|
IsTruncated: lmur.IsTruncated,
|
|
|
Uploads: uploads,
|
|
|
Prefix: lmur.Prefix,
|
|
|
Delimiter: lmur.Delimiter,
|
|
|
CommonPrefixes: commonPrefixes,
|
|
|
}
|
|
|
}
|
|
|
|
|
|
// ListMultipartUploads lists all multipart uploads.
|
|
|
func (l *ossObjects) ListMultipartUploads(ctx context.Context, bucket, prefix, keyMarker, uploadIDMarker, delimiter string, maxUploads int) (lmi minio.ListMultipartsInfo, err error) {
|
|
|
bkt, err := l.Client.Bucket(bucket)
|
|
|
if err != nil {
|
|
|
logger.LogIf(ctx, err)
|
|
|
return lmi, ossToObjectError(err, bucket)
|
|
|
}
|
|
|
|
|
|
lmur, err := bkt.ListMultipartUploads(oss.Prefix(prefix), oss.KeyMarker(keyMarker), oss.UploadIDMarker(uploadIDMarker),
|
|
|
oss.Delimiter(delimiter), oss.MaxUploads(maxUploads))
|
|
|
if err != nil {
|
|
|
logger.LogIf(ctx, err)
|
|
|
return lmi, ossToObjectError(err, bucket)
|
|
|
}
|
|
|
|
|
|
return fromOSSClientListMultipartsInfo(lmur), nil
|
|
|
}
|
|
|
|
|
|
// NewMultipartUpload upload object in multiple parts.
|
|
|
func (l *ossObjects) NewMultipartUpload(ctx context.Context, bucket, object string, metadata map[string]string, o minio.ObjectOptions) (uploadID string, err error) {
|
|
|
bkt, err := l.Client.Bucket(bucket)
|
|
|
if err != nil {
|
|
|
logger.LogIf(ctx, err)
|
|
|
return uploadID, ossToObjectError(err, bucket, object)
|
|
|
}
|
|
|
|
|
|
// Build OSS metadata
|
|
|
opts, err := appendS3MetaToOSSOptions(ctx, nil, metadata)
|
|
|
if err != nil {
|
|
|
return uploadID, ossToObjectError(err, bucket, object)
|
|
|
}
|
|
|
|
|
|
lmur, err := bkt.InitiateMultipartUpload(object, opts...)
|
|
|
if err != nil {
|
|
|
logger.LogIf(ctx, err)
|
|
|
return uploadID, ossToObjectError(err, bucket, object)
|
|
|
}
|
|
|
|
|
|
return lmur.UploadID, nil
|
|
|
}
|
|
|
|
|
|
// PutObjectPart puts a part of object in bucket.
|
|
|
func (l *ossObjects) PutObjectPart(ctx context.Context, bucket, object, uploadID string, partID int, r *minio.PutObjReader, opts minio.ObjectOptions) (pi minio.PartInfo, err error) {
|
|
|
data := r.Reader
|
|
|
bkt, err := l.Client.Bucket(bucket)
|
|
|
if err != nil {
|
|
|
logger.LogIf(ctx, err)
|
|
|
return pi, ossToObjectError(err, bucket, object)
|
|
|
}
|
|
|
|
|
|
imur := oss.InitiateMultipartUploadResult{
|
|
|
Bucket: bucket,
|
|
|
Key: object,
|
|
|
UploadID: uploadID,
|
|
|
}
|
|
|
size := data.Size()
|
|
|
up, err := bkt.UploadPart(imur, data, size, partID)
|
|
|
if err != nil {
|
|
|
logger.LogIf(ctx, err)
|
|
|
return pi, ossToObjectError(err, bucket, object)
|
|
|
}
|
|
|
|
|
|
return minio.PartInfo{
|
|
|
Size: size,
|
|
|
ETag: minio.ToS3ETag(up.ETag),
|
|
|
// NOTE(timonwong): LastModified is not supported
|
|
|
PartNumber: up.PartNumber,
|
|
|
}, nil
|
|
|
}
|
|
|
|
|
|
func ossBuildListObjectPartsParams(uploadID string, partNumberMarker, maxParts int) map[string]interface{} {
|
|
|
return map[string]interface{}{
|
|
|
"uploadId": uploadID,
|
|
|
"part-number-marker": strconv.Itoa(partNumberMarker),
|
|
|
"max-parts": strconv.Itoa(maxParts),
|
|
|
}
|
|
|
}
|
|
|
|
|
|
// fromOSSClientListPartsInfo converts OSS ListUploadedPartsResult to ListPartsInfo
|
|
|
func fromOSSClientListPartsInfo(lupr oss.ListUploadedPartsResult, partNumberMarker int) minio.ListPartsInfo {
|
|
|
parts := make([]minio.PartInfo, len(lupr.UploadedParts))
|
|
|
for i, up := range lupr.UploadedParts {
|
|
|
parts[i] = minio.PartInfo{
|
|
|
PartNumber: up.PartNumber,
|
|
|
LastModified: up.LastModified,
|
|
|
ETag: minio.ToS3ETag(up.ETag),
|
|
|
Size: int64(up.Size),
|
|
|
}
|
|
|
}
|
|
|
|
|
|
nextPartNumberMarker, _ := strconv.Atoi(lupr.NextPartNumberMarker)
|
|
|
return minio.ListPartsInfo{
|
|
|
Bucket: lupr.Bucket,
|
|
|
Object: lupr.Key,
|
|
|
UploadID: lupr.UploadID,
|
|
|
PartNumberMarker: partNumberMarker,
|
|
|
NextPartNumberMarker: nextPartNumberMarker,
|
|
|
MaxParts: lupr.MaxParts,
|
|
|
IsTruncated: lupr.IsTruncated,
|
|
|
Parts: parts,
|
|
|
}
|
|
|
}
|
|
|
|
|
|
func ossListObjectParts(client *oss.Client, bucket, object, uploadID string, partNumberMarker, maxParts int) (lupr oss.ListUploadedPartsResult, err error) {
|
|
|
params := ossBuildListObjectPartsParams(uploadID, partNumberMarker, maxParts)
|
|
|
resp, err := client.Conn.Do("GET", bucket, object, params, nil, nil, 0, nil)
|
|
|
if err != nil {
|
|
|
return lupr, err
|
|
|
}
|
|
|
|
|
|
// always drain output (response body)
|
|
|
defer minio.CloseResponse(resp.Body)
|
|
|
|
|
|
err = xml.NewDecoder(resp.Body).Decode(&lupr)
|
|
|
if err != nil {
|
|
|
return lupr, err
|
|
|
}
|
|
|
return lupr, nil
|
|
|
}
|
|
|
|
|
|
// CopyObjectPart creates a part in a multipart upload by copying
|
|
|
// existing object or a part of it.
|
|
|
func (l *ossObjects) CopyObjectPart(ctx context.Context, srcBucket, srcObject, destBucket, destObject, uploadID string,
|
|
|
partID int, startOffset, length int64, srcInfo minio.ObjectInfo, srcOpts, dstOpts minio.ObjectOptions) (p minio.PartInfo, err error) {
|
|
|
|
|
|
bkt, err := l.Client.Bucket(destBucket)
|
|
|
if err != nil {
|
|
|
logger.LogIf(ctx, err)
|
|
|
return p, ossToObjectError(err, destBucket)
|
|
|
}
|
|
|
|
|
|
// Build OSS metadata
|
|
|
opts, err := appendS3MetaToOSSOptions(ctx, nil, srcInfo.UserDefined)
|
|
|
if err != nil {
|
|
|
return p, ossToObjectError(err, srcBucket, srcObject)
|
|
|
}
|
|
|
|
|
|
completePart, err := bkt.UploadPartCopy(oss.InitiateMultipartUploadResult{
|
|
|
Key: destObject,
|
|
|
UploadID: uploadID,
|
|
|
}, srcBucket, srcObject, startOffset, length, partID, opts...)
|
|
|
|
|
|
if err != nil {
|
|
|
logger.LogIf(ctx, err)
|
|
|
return p, ossToObjectError(err, srcBucket, srcObject)
|
|
|
}
|
|
|
|
|
|
p.PartNumber = completePart.PartNumber
|
|
|
p.ETag = minio.ToS3ETag(completePart.ETag)
|
|
|
return p, nil
|
|
|
}
|
|
|
|
|
|
// ListObjectParts returns all object parts for specified object in specified bucket
|
|
|
func (l *ossObjects) ListObjectParts(ctx context.Context, bucket, object, uploadID string, partNumberMarker, maxParts int) (lpi minio.ListPartsInfo, err error) {
|
|
|
lupr, err := ossListObjectParts(l.Client, bucket, object, uploadID, partNumberMarker, maxParts)
|
|
|
if err != nil {
|
|
|
logger.LogIf(ctx, err)
|
|
|
return lpi, ossToObjectError(err, bucket, object, uploadID)
|
|
|
}
|
|
|
|
|
|
return fromOSSClientListPartsInfo(lupr, partNumberMarker), nil
|
|
|
}
|
|
|
|
|
|
// AbortMultipartUpload aborts a ongoing multipart upload.
|
|
|
func (l *ossObjects) AbortMultipartUpload(ctx context.Context, bucket, object, uploadID string) error {
|
|
|
bkt, err := l.Client.Bucket(bucket)
|
|
|
if err != nil {
|
|
|
logger.LogIf(ctx, err)
|
|
|
return ossToObjectError(err, bucket, object)
|
|
|
}
|
|
|
|
|
|
err = bkt.AbortMultipartUpload(oss.InitiateMultipartUploadResult{
|
|
|
Bucket: bucket,
|
|
|
Key: object,
|
|
|
UploadID: uploadID,
|
|
|
})
|
|
|
if err != nil {
|
|
|
logger.LogIf(ctx, err)
|
|
|
return ossToObjectError(err, bucket, object)
|
|
|
}
|
|
|
return nil
|
|
|
}
|
|
|
|
|
|
// CompleteMultipartUpload completes ongoing multipart upload and finalizes object.
|
|
|
func (l *ossObjects) CompleteMultipartUpload(ctx context.Context, bucket, object, uploadID string, uploadedParts []minio.CompletePart, opts minio.ObjectOptions) (oi minio.ObjectInfo, err error) {
|
|
|
client := l.Client
|
|
|
bkt, err := client.Bucket(bucket)
|
|
|
if err != nil {
|
|
|
logger.LogIf(ctx, err)
|
|
|
return oi, ossToObjectError(err, bucket, object)
|
|
|
}
|
|
|
|
|
|
// Error out if uploadedParts except last part sizing < 5MiB.
|
|
|
// NOTE(timonwong): Actually, OSS wont't throw EntityTooSmall error, doing this check just for mint :(
|
|
|
var partNumberMarker int
|
|
|
lupr := oss.ListUploadedPartsResult{IsTruncated: true}
|
|
|
for lupr.IsTruncated {
|
|
|
lupr, err = ossListObjectParts(client, bucket, object, uploadID, partNumberMarker, ossMaxParts)
|
|
|
if err != nil {
|
|
|
logger.LogIf(ctx, err)
|
|
|
return oi, ossToObjectError(err, bucket, object, uploadID)
|
|
|
}
|
|
|
|
|
|
uploadedParts := lupr.UploadedParts
|
|
|
if !lupr.IsTruncated {
|
|
|
if len(uploadedParts) < 1 {
|
|
|
uploadedParts = nil
|
|
|
} else {
|
|
|
uploadedParts = uploadedParts[:len(uploadedParts)-1]
|
|
|
}
|
|
|
}
|
|
|
|
|
|
for _, part := range uploadedParts {
|
|
|
if part.Size < ossS3MinPartSize {
|
|
|
logger.LogIf(ctx, minio.PartTooSmall{
|
|
|
PartNumber: part.PartNumber,
|
|
|
PartSize: int64(part.Size),
|
|
|
PartETag: minio.ToS3ETag(part.ETag),
|
|
|
})
|
|
|
return oi, minio.PartTooSmall{
|
|
|
PartNumber: part.PartNumber,
|
|
|
PartSize: int64(part.Size),
|
|
|
PartETag: minio.ToS3ETag(part.ETag),
|
|
|
}
|
|
|
}
|
|
|
}
|
|
|
|
|
|
partNumberMarker, _ = strconv.Atoi(lupr.NextPartNumberMarker)
|
|
|
}
|
|
|
|
|
|
imur := oss.InitiateMultipartUploadResult{
|
|
|
Bucket: bucket,
|
|
|
Key: object,
|
|
|
UploadID: uploadID,
|
|
|
}
|
|
|
parts := make([]oss.UploadPart, len(uploadedParts))
|
|
|
for i, up := range uploadedParts {
|
|
|
parts[i] = oss.UploadPart{
|
|
|
PartNumber: up.PartNumber,
|
|
|
ETag: strings.TrimSuffix(up.ETag, "-1"), // Trim "-1" suffix in ETag as PutObjectPart().
|
|
|
}
|
|
|
}
|
|
|
|
|
|
_, err = bkt.CompleteMultipartUpload(imur, parts)
|
|
|
if err != nil {
|
|
|
logger.LogIf(ctx, err)
|
|
|
return oi, ossToObjectError(err, bucket, object)
|
|
|
}
|
|
|
|
|
|
return l.GetObjectInfo(ctx, bucket, object, minio.ObjectOptions{})
|
|
|
}
|
|
|
|
|
|
// SetBucketPolicy sets policy on bucket.
|
|
|
// OSS supports three types of bucket policies:
|
|
|
// oss.ACLPublicReadWrite: readwrite in minio terminology
|
|
|
// oss.ACLPublicRead: readonly in minio terminology
|
|
|
// oss.ACLPrivate: none in minio terminology
|
|
|
func (l *ossObjects) SetBucketPolicy(ctx context.Context, bucket string, bucketPolicy *policy.Policy) error {
|
|
|
policyInfo, err := minio.PolicyToBucketAccessPolicy(bucketPolicy)
|
|
|
if err != nil {
|
|
|
// This should not happen.
|
|
|
return ossToObjectError(err, bucket)
|
|
|
}
|
|
|
|
|
|
bucketPolicies := miniogopolicy.GetPolicies(policyInfo.Statements, bucket, "")
|
|
|
if len(bucketPolicies) != 1 {
|
|
|
logger.LogIf(ctx, minio.NotImplemented{})
|
|
|
return minio.NotImplemented{}
|
|
|
}
|
|
|
|
|
|
prefix := bucket + "/*" // For all objects inside the bucket.
|
|
|
for policyPrefix, bucketPolicy := range bucketPolicies {
|
|
|
if policyPrefix != prefix {
|
|
|
logger.LogIf(ctx, minio.NotImplemented{})
|
|
|
return minio.NotImplemented{}
|
|
|
}
|
|
|
|
|
|
var acl oss.ACLType
|
|
|
switch bucketPolicy {
|
|
|
case miniogopolicy.BucketPolicyNone:
|
|
|
acl = oss.ACLPrivate
|
|
|
case miniogopolicy.BucketPolicyReadOnly:
|
|
|
acl = oss.ACLPublicRead
|
|
|
case miniogopolicy.BucketPolicyReadWrite:
|
|
|
acl = oss.ACLPublicReadWrite
|
|
|
default:
|
|
|
logger.LogIf(ctx, minio.NotImplemented{})
|
|
|
return minio.NotImplemented{}
|
|
|
}
|
|
|
|
|
|
err := l.Client.SetBucketACL(bucket, acl)
|
|
|
if err != nil {
|
|
|
logger.LogIf(ctx, err)
|
|
|
return ossToObjectError(err, bucket)
|
|
|
}
|
|
|
}
|
|
|
|
|
|
return nil
|
|
|
}
|
|
|
|
|
|
// GetBucketPolicy will get policy on bucket.
|
|
|
func (l *ossObjects) GetBucketPolicy(ctx context.Context, bucket string) (*policy.Policy, error) {
|
|
|
result, err := l.Client.GetBucketACL(bucket)
|
|
|
if err != nil {
|
|
|
logger.LogIf(ctx, err)
|
|
|
return nil, ossToObjectError(err)
|
|
|
}
|
|
|
|
|
|
var readOnly, readWrite bool
|
|
|
switch result.ACL {
|
|
|
case string(oss.ACLPrivate):
|
|
|
// By default, all buckets starts with a "private" policy.
|
|
|
return nil, ossToObjectError(minio.BucketPolicyNotFound{}, bucket)
|
|
|
case string(oss.ACLPublicRead):
|
|
|
readOnly = true
|
|
|
case string(oss.ACLPublicReadWrite):
|
|
|
readWrite = true
|
|
|
default:
|
|
|
logger.LogIf(ctx, minio.NotImplemented{})
|
|
|
return nil, minio.NotImplemented{}
|
|
|
}
|
|
|
|
|
|
actionSet := policy.NewActionSet()
|
|
|
if readOnly {
|
|
|
actionSet.Add(policy.GetBucketLocationAction)
|
|
|
actionSet.Add(policy.ListBucketAction)
|
|
|
actionSet.Add(policy.GetObjectAction)
|
|
|
}
|
|
|
if readWrite {
|
|
|
actionSet.Add(policy.GetBucketLocationAction)
|
|
|
actionSet.Add(policy.ListBucketAction)
|
|
|
actionSet.Add(policy.GetObjectAction)
|
|
|
actionSet.Add(policy.ListBucketMultipartUploadsAction)
|
|
|
actionSet.Add(policy.AbortMultipartUploadAction)
|
|
|
actionSet.Add(policy.DeleteObjectAction)
|
|
|
actionSet.Add(policy.ListMultipartUploadPartsAction)
|
|
|
actionSet.Add(policy.PutObjectAction)
|
|
|
}
|
|
|
|
|
|
return &policy.Policy{
|
|
|
Version: policy.DefaultVersion,
|
|
|
Statements: []policy.Statement{
|
|
|
policy.NewStatement(
|
|
|
policy.Allow,
|
|
|
policy.NewPrincipal("*"),
|
|
|
actionSet,
|
|
|
policy.NewResourceSet(
|
|
|
policy.NewResource(bucket, ""),
|
|
|
policy.NewResource(bucket, "*"),
|
|
|
),
|
|
|
condition.NewFunctions(),
|
|
|
),
|
|
|
},
|
|
|
}, nil
|
|
|
}
|
|
|
|
|
|
// DeleteBucketPolicy deletes all policies on bucket.
|
|
|
func (l *ossObjects) DeleteBucketPolicy(ctx context.Context, bucket string) error {
|
|
|
err := l.Client.SetBucketACL(bucket, oss.ACLPrivate)
|
|
|
if err != nil {
|
|
|
logger.LogIf(ctx, err)
|
|
|
return ossToObjectError(err, bucket)
|
|
|
}
|
|
|
return nil
|
|
|
}
|
|
|
|
|
|
// IsCompressionSupported returns whether compression is applicable for this layer.
|
|
|
func (l *ossObjects) IsCompressionSupported() bool {
|
|
|
return false
|
|
|
}
|
|
|
|