需要 Go 1.11 及以上版本;使用云存储或 SCP 前需自行准备对应认证配置;部分存储后端需单独引入对应子包
go get github.com/viant/afsREADME 未说明,纯 Go 库无硬件要求
Go 语言统一抽象多种存储系统的库,用同一套 API 操作本地、云存储和 SCP
Go 语言实现的抽象文件存储库,用统一 API 屏蔽本地磁盘、S3、SCP、tar、zip 等不同存储后端,省去为每种存储单独写适配代码。
核心能力
- 用 URL scheme 区分存储后端,统一 CRUD 接口
- 支持 Copy/Move 跨存储系统直接传输数据
- 内置 tar/zip 归档遍历与批量上传器
- 提供 matcher 按扩展名、路径、gitignore 规则过滤
- 支持流式读写与内容替换 modifier
容易踩的坑
- v1.30.1最新正式版2026-10-26 · 2026-10-26 22:56 UTC
Changes since v1.30.0
- Add a scratchpad storage provider with per-request user isolation.
- Add scoped mappings between custom resource schemes and filesystem roots.
- Preserve file/directory semantics in the io filesystem adapter and correct embedded filesystem ordering.
- Avoid adding virtual root entries when rewriting tar archives.
- Fix closing a zero-byte opened pipe and extend matcher/nested-walk regression coverage.
Full comparison: https://github.com/viant/afs/compare/v1.30.0...v1.30.1
Target commit:
b2fe50a. - v1.30.02025-12-23 · 2025-12-23 15:46 UTC
这个版本没有写更新说明。
- v1.29.02025-12-02 · 2025-12-02 21:53 UTC
这个版本没有写更新说明。
- v1.28.02025-11-07 · 2025-11-07 21:02 UTC
这个版本没有写更新说明。
- v1.27.02025-11-07 · 2025-11-07 14:26 UTC
查看更新说明
added url.ToFileURL
用户点评
登录用户的真实使用记录,可按版本和硬件筛选还没有人点评。第一个说出真实体验的人最有价值。
afs - abstract file storage
Please refer to CHANGELOG.md if you encounter breaking changes.
- Motivation
- Introduction
- Usage
- Matchers
- Content modifiers
- Streaming data
- Options
- Storage Implementations
- Testing mode
- Storage Manager
- GoCover
- License
- Credits and Acknowledgements
Motivation
When dealing with various storage systems, like cloud storage, SCP, container or local file system, using shared API for typical storage operation provides an excellent simplification. What's more, the ability to simulate storage-related errors like Auth or EOF allows you to test an app error handling.
Introduction
This library uses a storage manager abstraction to provide an implementation for a specific storage system with following
- CRUD Operation:
List(ctx context.Context, URL string, options ...Option) ([]Object, error)
Walk(ctx context.Context, URL string, handler OnVisit, options ...Option) error
Open(ctx context.Context, object Object, options ...Option) (io.ReadCloser, error)
OpenURL(ctx context.Context, URL string, options ...Option) (io.ReadCloser, error)
Upload(ctx context.Context, URL string, mode os.FileMode, reader io.Reader, options ...Option) error
Create(ctx context.Context, URL string, mode os.FileMode, isDir bool, options ...Option) error
Delete(ctx context.Context, URL string, options ...Option) error
- Batch uploader:
type Upload func(ctx context.Context, parent string, info os.FileInfo, reader io.Reader) error
Uploader(ctx context.Context, URL string, options ...Option) (Upload, io.Closer, error)
- Utilities:
Copy(ctx context.Context, sourceURL, destURL string, options ...Option) error
Move(ctx context.Context, sourceURL, destURL string, options ...Option) error
NewWriter(ctx context.Context, URL string, mode os.FileMode, options ...storage.Option) (io.WriteCloser, error)
DownloadWithURL(ctx context.Context, URL string, options ...Option) ([]byte, error)
Download(ctx context.Context, object Object, options ...Option) ([]byte, error)
URL scheme is used to identify storage system, or alternatively relative/absolute path can be used for local file storage. By default, all operations using the same baseURL share the same corresponding storage manager instance. For example, instead supplying SCP auth details for all operations, auth option can be used only once.
func main() {
ctx := context.Background()
{
//auth with first call
fs := afs.New()
defer fs.Close()
keyAuth, err := scp.LocalhostKeyAuth("")
if err != nil {
log.Fatal(err)
}
reader1, err := fs.OpenURL(ctx, "scp://host1:22/myfolder/asset.txt", keyAuth)
if err != nil {
log.Fatal(err)
}
...
reader2, err := fs.OpenURL(ctx, "scp://host1:22/myfolder/asset.txt", keyAuth)
}
{
//auth per baseURL
fs := afs.New()
err = fs.Init(ctx, "scp://host1:22/", keyAuth)
if err != nil {
log.Fatal(err)
}
defer fs.Destroy("scp://host1:22/")
reader, err := fs.OpenURL(ctx, "scp://host1:22/myfolder/asset.txt")
}
}
Usage
Downloading location content
func main() {
fs := afs.New()
ctx := context.Background()
objects, err := fs.List(ctx, "/tmp/folder")
if err != nil {
log.Fatal(err)
}
for _, object := range objects {
fmt.Printf("%v %v\n", object.Name(), object.URL())
if object.IsDir() {
continue
}
reader, err := fs.Open(ctx, object)
if err != nil {
log.Fatal(err)
}
data, err := ioutil.ReadAll(reader)
if err != nil {
log.Fatal(err)
}
fmt.Printf("%s\n", data)
}
}
Uploading Content
func main() {
fs := afs.New()
ctx := context.Background()
keyAuth, err := scp.LocalhostKeyAuth("")
if err != nil {
log.Fatal(err)
}
err = fs.Init(ctx, "scp://127.0.0.1:22/", keyAuth)
if err != nil {
log.Fatal(err)
}
err = fs.Upload(ctx, "scp://127.0.0.1:22/folder/asset.txt", 0644, strings.NewReader("test me"))
if err != nil {
log.Fatal(err)
}
ok, err := fs.Exists(ctx, "scp://127.0.0.1:22/folder/asset.txt")
if err != nil {
log.Fatal(err)
}
fmt.Printf("has file: %v\n", ok)
_ = fs.Delete(ctx, "scp://127.0.0.1:22/folder/asset.txt")
}
Uploading Content With Writer
func main() {
fs := afs.New()
ctx := context.Background()
keyAuth, err := scp.LocalhostKeyAuth("")
if err != nil {
log.Fatal(err)
}
err = fs.Init(ctx, "scp://127.0.0.1:22/", keyAuth)
if err != nil {
log.Fatal(err)
}
writer = fs.NewWriter(ctx, "scp://127.0.0.1:22/folder/asset.txt", 0644)
_, err := writer.Write([]byte("test me")))
if err != nil {
log.Fatal(err)
}
err = writer.Close()
if err != nil {
log.Fatal(err)
}
ok, err := fs.Exists(ctx, "scp://127.0.0.1:22/folder/asset.txt")
if err != nil {
log.Fatal(err)
}
fmt.Printf("has file: %v\n", ok)
_ = fs.Delete(ctx, "scp://127.0.0.1:22/folder/asset.txt")
}
Data Copy
func main() {
fs := afs.New()
ctx := context.Background()
keyAuth, err := scp.LocalhostKeyAuth("")
if err != nil {
log.Fatal(err)
}
err = fs.Copy(ctx, "s3://mybucket/myfolder", "scp://127.0.0.1/tmp", option.NewSource(), option.NewDest(keyAuth))
if err != nil {
log.Fatal(err)
}
}
Archiving content
func main() {
secretPath := path.Join(os.Getenv("HOME"), ".secret", "gcp-e2e.json")
auth, err := gs.NewJwtConfig(option.NewLocation(secretPath))
if err != nil {
return
}
sourceURL := "mylocalPath/"
destURL := "gs:mybucket/test.zip/zip://localhost/dir1"
fs := afs.New()
ctx := context.Background()
err = fs.Copy(ctx, sourceURL, destURL, option.NewDest(auth))
if err != nil {
log.Fatal(err)
}
}
Archive Walker
Walker can be created for tar or zip archive.
func main() {
ctx := context.Background()
fs := afs.New()
walker := tar.NewWalker(s3afs.New())
err := fs.Copy(ctx, "/tmp/test.tar", "s3:///dest/folder/test", walker)
if err != nil {
log.Fatal(err)
}
Archive Uploader
Uploader can be created for tar or zip archive.
func main() {
ctx := context.Background()
fs := afs.New()
uploader := zip.NewBatchUploader(gsafs.New())
err := fs.Copy(ctx, "gs:///tmp/test/data", "/tmp/data.zip", uploader)
if err != nil {
log.Fatal(err)
}
}
Data Move
func main() {
fs := afs.New()
ctx := context.Background()
keyAuth, err := scp.LocalhostKeyAuth("")
if err != nil {
log.Fatal(err)
}
err = fs.Move(ctx, "/tmp/transient/app", "scp://127.0.0.1/tmp", option.NewSource(), option.NewDest(keyAuth))
if err != nil {
log.Fatal(err)
}
}
Batch Upload
func main() {
fs := afs.New()
ctx := context.Background()
upload, closer, err := fs.Uploader(ctx, "/tmp/clone")
if err != nil {
log.Fatal(err)
}
defer closer.Close()
assets := []*asset.Resource{
asset.NewFile("asset1.txt", []byte("test 1"), 0644),
asset.NewFile("asset2.txt", []byte("test 2"), 0644),
asset.NewDir("folder1", file.DefaultDirOsMode),
asset.NewFile("folder1/asset1.txt", []byte("test 3"), 0644),
asset.NewFile("folder1/asset2.txt", []byte("test 4"), 0644),
}
for _, asset := range assets {
relative := ""
var reader io.Reader
if strings.Contains(asset.Name, "/") {
relative, _ = path.Split(asset.Name)
}
if ! asset.Dir {
reader = bytes.NewReader(asset.Data)
}
err = upload(ctx, relative, asset.Info(), reader)
if err != nil {
log.Fatal(err)
}
}
}
Matchers
To filter source content you can use Matcher option. The following have been implemented.
func main() {
matcher, err := NewBasic("/data", ".avro", nil)
fs := afs.New()
ctx := context.Background()
err := fs.Copy(ctx, "/tmp/data", "s3://mybucket/data/", matcher.Match)
if err != nil {
log.Fatal(err)
}
}
Exclusion
func main() {
matcher := matcher.Basic{Exclusion:".+/data/perf/\\d+/.+"}
fs := afs.New()
ctx := context.Background()
err := fs.Copy(ctx, "/tmp/data", "s3://mybucket/data/", matcher.Match)
if err != nil {
log.Fatal(err)
}
}
OS style filepath match, with the following terms:
- '*' matches any sequence of non-Separator characters
- '?' matches any single non-Separator character
- '[' [ '^' ] { character-range } ']'
func main() {
matcher := matcher.Filepath("*.avro")
fs := afs.New()
ctx := context.Background()
err := fs.Copy(ctx, "/tmp/data", "gs://mybucket/data/", matcher)
if err != nil {
log.Fatal(err)
}
}
Ignore matcher represents matcher that matches file that are not in the ignore rules. The syntax of ignore borrows heavily from that of .gitignore; see https://git-scm.com/docs/gitignore or man gitignore for a full reference.
func mian(){
ignoreMatcher, err := matcher.NewIgnore([]string{"*.txt", ".ignore"})
//or matcher.NewIgnore(option.NewLocation(".cloudignore"))
if err != nil {
log.Fatal(err)
}
fs := afs.New()
ctx := context.Background()
objects, err := fs.List(ctx, "/tmp/folder", ignoreMatcher.Match)
if err != nil {
log.Fatal(err)
}
for _, object := range objects {
fmt.Printf("%v %v\n", object.Name(), object.URL())
if object.IsDir() {
continue
}
}
}
Modification Time Matcher represents matcher that matches file that were modified either before or after specified time.
func mian(){
before, err := toolbox.TimeAt("2 days ago in UTC")
if err != nil {
log.Fatal(err)
}
modTimeMatcher, err := matcher.NewModification(before, nil)
if err != nil {
log.Fatal(err)
}
fs := afs.New()
ctx := context.Background()
objects, err := fs.List(ctx, "/tmp/folder", modTimeMatcher.Match)
if err != nil {
log.Fatal(err)
}
for _, object := range objects {
fmt.Printf("%v %v\n", object.Name(), object.URL())
if object.IsDir() {
continue
}
}
}
Content modifiers
To modify resource content on the fly you can use Modifier option.
func main() {
fs := afs.New()
ctx := context.Background()
sourceURL := "file:/tmp/app.war/zip://localhost/WEB-INF/classes/config.properties"
destURL := "file:/tmp/app.war/zip://localhost/"
err := fs.Copy(ctx, sourceURL, destURL, modifier.Replace(map[string]string{
"${changeMe}": os.Getenv("USER"),
}))
if err != nil {
log.Fatal(err)
}
}
package main
import (
"context"
"log"
"github.com/viant/afs"
"io"
"fmt"
"io/ioutil"
"os"
"strings"
)
func modifyContent(info os.FileInfo, reader io.ReadCloser) (closer io.ReadCloser, e error) {
if strings.HasSuffix(info.Name() ,".info") {
data, err := ioutil.ReadAll(reader)
if err != nil {
return nil, err
}
_ = reader.Close()
expanded := strings.Replace(string(data), "$os.User", os.Getenv("USER"), 1)
reader = ioutil.NopCloser(strings.NewReader(expanded))
}
return reader, nil
}
func main() {
fs := afs.New()
reader ,err := fs.OpenURL(context.Background(), "s3://mybucket/meta.info", modifyContent)
if err != nil {
log.Fatal(err)
}
defer reader.Close()
content, err := ioutil.ReadAll(reader)
if err != nil {
log.Fatal(err)
}
fmt.Printf("content: %s\n", content)
}
Streaming data
Streaming data allows data reading and uploading in chunks with small memory footprint.
jwtConfig, err := gs.NewJwtConfig()
if err != nil {
log.Fatal(err)
}
ctx := context.Background()
fs := afs.New()
sourceURL := "gs://myBucket/path/myasset.gz"
reader, err := fs.OpenURL(ctx, sourceURL, jwtConfig, option.NewStream(64*1024*1024, 0))
if err != nil {
log.Fatal(err)
}
_ = os.Setenv("AWS_SDK_LOAD_CONFIG", "true")
destURL := "s3://myBucket/path/myasset.gz"
err = fs.Upload(ctx, destURL, 0644, reader, &option.Checksum{Skip:true})
if err != nil {
log.Fatal(err)
}
// or
writer = fs.NewWriter(ctx, destURL, 0644, &option.Checksum{Skip:true})
_, err = io.Copy(writer, reader)
if err != nil {
log.Fatal(err)
}
err = writer.Close()
if err != nil {
log.Fatal(err)
}
Options
To control number and position of listed resources you can yse page option.
Provider specific timeout.
Provides user/password auth.
- Source & Dest Options
Groups options by source or destination options. This options work with Copy or Move operations.
func main() {
fs := afs.New()
secretPath := path.Join(os.Getenv("HOME"), ".secret","gcp.json")
jwtConfig, err := gs.NewJwtConfig(option.NewLocation(secretPath))
if err != nil {
log.Fatal(err)
}
sourceOptions := option.NewSource(jwtConfig)
authConfig, err := s3.NewAuthConfig(option.NewLocation("aws.json"))
if err != nil {
log.Fatal(err)
}
destOptions := option.NewDest(authConfig)
err = fs.Copy(ctx, "gs://mybucket/data", "s3://mybucket/data", sourceOptions, destOptions)
}
- option.Checksum skip computing checksum if Skip is set, this option allows streaming upload in chunks
- option.Stream: download reader reads data with specified stream PartSize
Check out storage manager for additional options.
Storage Implementations
Testing fs
To unit test all storage operation all in memory you can use faker fs.
In addition you can use error options to test exception handling.
- DownloadError
func mian() {
fs := afs.NewFaker()
ctx := context.Background()
err := fs.Upload(ctx, "gs://myBucket/folder/asset.txt", 0, strings.NewReader("some data"), option.NewUploadError(io.EOF))
if err != nil {
log.Fatalf("expect upload error: %v", err)
}
}
- ReaderError
func mian() {
fs := afs.NewFaker()
ctx := context.Background()
err := fs.Upload(ctx, "gs://myBucket/folder/asset.txt", 0, strings.NewReader("some data"), option.NewDownloadError(io.EOF))
if err != nil {
log.Fatal(err)
}
_, err = fs.OpenURL(ctx, "gs://myBucket/folder/asset.txt")
if err != nil {
log.Fatalf("expect download error: %v", err)
}
}
- UploadError
func mian() {
fs := afs.NewFaker()
ctx := context.Background()
err := fs.Upload(ctx, "gs://myBucket/folder/asset.txt", 0, strings.NewReader("some data"), option.NewUploadError(io.EOF))
if err != nil {
log.Fatalf("expect upload error: %v", err)
}
}
Code generation for static or in memory go file
Generate with mem storage
package main
import (
"log"
"github.com/viant/afs/parrot
)
func mian() {
ctx := context.Background()
err := parrot.GenerateWithMem(ctx, "pathToBinaryAsset", "gen.go", false)
if err != nil {
log.Fatal(err)
}
}
Generate static data files
package main
import (
"log"
"github.com/viant/afs/parrot
)
func mian() {
ctx := context.Background()
err := parrot.Generate(ctx, "pathToBinaryAsset", "data/", false)
if err != nil {
log.Fatal(err)
}
}
Test setup utilities
Package asset defines basic utilities to quickly manage asset related unit tests.
func Test_XXX(t *testing.T) {
var useCases = []struct {
description string
location string
options []storage.Option
assets []*asset.Resource
}{
}
ctx := context.Background()
for _, useCase := range useCases {
fs := afs.New()
mgr, err := afs.Manager(useCase.location, useCase.options...)
if err != nil {
log.Fatal(err)
}
err = asset.Create(mgr, useCase.location, useCase.assets)
if err != nil {
log.Fatal(err)
}
//... actual app logic
actuals, err := asset.Load(mgr, useCase.location)
if err != nil {
log.Fatal(err)
}
for _, expect := range useCase.assets {
actual, ok := actuals[expect.Name]
if !assert.True(t, ok, useCase.description+": "+expect.Name+fmt.Sprintf(" - actuals: %v", actuals)) {
continue
}
assert.EqualValues(t, expect.Name, actual.Name, useCase.description+" "+expect.Name)
assert.EqualValues(t, expect.Mode, actual.Mode, useCase.description+" "+expect.Name)
assert.EqualValues(t, expect.Dir, actual.Dir, useCase.description+" "+expect.Name)
assert.EqualValues(t, expect.Data, actual.Data, useCase.description+" "+expect.Name)
}
_ = asset.Cleanup(mgr, useCase.location)
}
}
GoCover
License
The source code is made available under the terms of the Apache License, Version 2, as stated in the file LICENSE.
Individual files may be made available under their own specific license, all compatible with Apache License, Version 2. Please see individual files for details.
Credits and Acknowledgements
Library Author: Adrian Witas