AWS (GOFAKE)S3

AWS S3 fake server and testing library for comprehensive S3 integration testing. This tool can be used to run a test server, for example, to support testing AWS Lambda functions that interact with S3. It also serves as a straightforward and convenient S3 mock and test server for various development needs.

Intended Use

GOFAKE)S3 is primarily designed for:

• Local development of S3-dependent AWS Lambda functions.
• Testing implementations with AWS S3 access.
• Facilitating browser-based direct uploads to S3 in a local testing environment.

When Not to Use (GOFAKE)S3?

(GOFAKE)S3 should not be used as a production service. Its primary purpose is to aid in development and testing:

• (GOFAKE)S3 is not designed for production-level data storage or handling.
• It lacks the robustness required for safe, persistent access to production data.
• The tool is still under development with significant portions of the AWS S3 API yet to be implemented. Consequently, breaking changes are expected.

For production environments, consider more established solutions. Some recommended alternatives can be found in the "Similar Notable Projects" section below.

How to use it?

Example (aws-sdk-go version 1)

// fake s3
backend := s3mem.New()
faker := gofakes3.New(backend)
ts := httptest.NewServer(faker.Server())
defer ts.Close()

// configure S3 client
s3Config := &aws.Config{
	Credentials:      credentials.NewStaticCredentials("YOUR-ACCESSKEYID", "YOUR-SECRETACCESSKEY", ""),
	Endpoint:         aws.String(ts.URL),
	Region:           aws.String("eu-central-1"),
	DisableSSL:       aws.Bool(true),
	S3ForcePathStyle: aws.Bool(true),
}
newSession := session.New(s3Config)

s3Client := s3.New(newSession)
cparams := &s3.CreateBucketInput{
	Bucket: aws.String("newbucket"),
}

// Create a new bucket using the CreateBucket call.
_, err := s3Client.CreateBucket(cparams)
if err != nil {
	// Message from an error.
	fmt.Println(err.Error())
	return
}

// Upload a new object "testobject" with the string "Hello World!" to our "newbucket".
_, err = s3Client.PutObject(&s3.PutObjectInput{
	Body:   strings.NewReader(`{"configuration": {"main_color": "#333"}, "screens": []}`),
	Bucket: aws.String("newbucket"),
	Key:    aws.String("test.txt"),
})

// ... accessing of test.txt through any S3 client would now be possible

Example for V2 (aws-sdk-go-v2)

backend := s3mem.New()
faker := gofakes3.New(backend)
ts := httptest.NewServer(faker.Server())
defer ts.Close()

// Difference in configuring the client

// Setup a new config
cfg, _ := config.LoadDefaultConfig(
	context.TODO(),
    config.WithCredentialsProvider(credentials.NewStaticCredentialsProvider("KEY", "SECRET", "SESSION")),
    config.WithHTTPClient(&http.Client{
        Transport: &http.Transport{
            TLSClientConfig: &tls.Config{InsecureSkipVerify: true},
        },
    }),
    config.WithEndpointResolverWithOptions(
        aws.EndpointResolverWithOptionsFunc(func(_, _ string, _ ...interface{}) (aws.Endpoint, error) {
            return aws.Endpoint{URL: ts.URL}, nil
        }),
    ),
)

// Create an Amazon S3 v2 client, important to use o.UsePathStyle
// alternatively change local DNS settings, e.g., in /etc/hosts
// to support requests to https://.127.0.0.1:32947/...
client := s3.NewFromConfig(cfg, func(o *s3.Options) {
	o.UsePathStyle = true
})

Please feel free to check it out and to provide useful feedback (using github issues), but be aware, this software is used internally and for the local development only. Thus, it has no demand for correctness, performance or security.

There are two ways to run locally: using DNS, or using S3 path mode.

S3 path mode is the most flexible and least restrictive, but it does require that you are able to modify your client code.In Go, the modification would look like so:

config := aws.Config{}
config.WithS3ForcePathStyle(true)

S3 path mode works over the network by default for all bucket names.

If you are unable to modify the code, DNS mode can be used, but it comes with further restrictions and requires you to be able to modify your local DNS resolution.

If using localhost as your endpoint, you will need to add the following to /etc/hosts for every bucket you want to fake:

127.0.0.1 .localhost

It is trickier if you want other machines to be able to use your fake S3 server as you need to be able to modify their DNS resolution as well.

Exemplary usage

Lambda Example

var AWS   = require('aws-sdk')

var ep = new AWS.Endpoint('https://localhost:9000');
var s3 = new AWS.S3({endpoint: ep});

exports.handle = function (e, ctx) {
  s3.createBucket({
    Bucket: '',
  }, function(err, data) {
    if (err) return console.log(err, err.stack);
    ctx.succeed(data)
  });
}

Upload Example

  
    
  
  

  <form enctype="multipart/form-data" method="post" action="https://localhost:9000/<bucket-name>/">
    Key to upload:
    <input value="user/user1/test/<filename>" name="key" type="input"><br>
    <input value="public-read" name="acl" type="hidden">
    <input value="14365123651274" name="x-amz-meta-uuid" type="hidden">
    <input value="AES256" name="x-amz-server-side-encryption" type="hidden">
    <input value="AKIAIOSFODNN7EXAMPLE/20151229/us-east-1/s3/aws4_request" name="X-Amz-Credential" type="text">
    <input value="AWS4-HMAC-SHA256" name="X-Amz-Algorithm" type="text">
    <input value="20151229T000000Z" name="X-Amz-Date" type="text">

    Tags for File:
    <input value="" name="x-amz-meta-tag" type="input"><br>
    <input value="<Base64-encoded policy string>" name="Policy" type="hidden">
    <input value="<signature-value>" name="X-Amz-Signature" type="hidden">
    File:
    <input name="file" type="file"> <br>
    
    <input value="Upload to Amazon S3" type="submit">
  </form>

Similar notable projects

• https://github.com/minio/minio not similar but powerfull 😉
• https://github.com/andrewgaul/s3proxy by @andrewgaul

Contributors

A big thank you to all the contributors, especially Blake @shabbyrobe who pushed this little project to the next level!

Help wanted