LitS3 is a library written in C# that provides comprehensive and straightforward access to Amazon S3 for .NET developers.

Notably

• Gives you typesafe objects instead of raw XML
• Well-documented, highly optimized code
• Uses the more efficient REST API
• Currently supports most S3 features
• Works in Mono+MS (see MonoNotes)

S3Service: A one-liner API

var s3 = new S3Service
{
    AccessKeyID = Settings.Default.AccessKeyID,
    SecretAccessKey = Settings.Default.SecretAccessKey
};

s3.ForEachBucket(Console.WriteLine);

//> Bucket "mybucket"
//> Bucket "myotherbucket"
//> Bucket "lits3-demo"

s3.AddObjectString("This is file one!", "lits3-demo", "File 1.txt");

s3.ForEachObject("lits3-demo", Console.WriteLine);

//> S3Object "File 1.txt"
//> Common Prefix "MyDirectory"

Console.WriteLine(s3.GetObjectString("lits3-demo", "File 1.txt"));

//> This is file one!

s3.CopyObject("lits3-demo", "File 1.txt", "File 1 copy.txt");

s3.ForEachObject("lits3-demo", Console.WriteLine);

//> S3Object "File 1 copy.txt"
//> S3Object "File 1.txt"
//> Common Prefix "MyDirectory"

s3.ForEachObject("lits3-demo", "MyDirectory/", Console.WriteLine);

//> S3Object "Other File.txt"

What's really cool about these examples is that LitS3 is managing memory very efficiently for you. The S3Service.ForEach… methods use callbacks to ensure that the actual web requests happen transparently in the background and are properly cleaned up.

For instance, when you call ForEachObject to list the contents of a bucket with a million objects, LitS3 will automatically handle the multiple S3 requests necessary to do this, as you iterate through each object. LitS3 will not just construct an array with a million entries!

Need more flexibility?

Every method in S3Service is backed by a corresponding S3Request+S3Response class pair that, together, encapsulate a normal WebRequest+WebResponse.

This means that every S3 operation can be performed using WebRequest's asynchronous API:

var request = new GetObjectRequest(s3, "lits3-demo", "File 1.txt");

request.BeginGetResponse(result =>
{
    // comes in on a separate thread
    using (GetObjectResponse response = request.EndGetResponse(result))
    {
        StreamReader reader = new StreamReader(response.GetResponseStream());
        Console.WriteLine(reader.ReadToEnd());

        //> This is file one!
    }
}, null);

// continues immediately without blocking...

I don't see (…random S3 feature…) in S3Service, can you add it?

This is where the S3Request+S3Response classes mentioned above come in. These backing classes like AddObjectRequest, GetObjectRequest, etc. expose nearly 100% of S3's API.

For instance, use AddObjectRequest.Metadata and GetObjectResponse.Metadata if you want to get or set S3 key-value "Metadata" on your objects.

More Code Examples

What's missing

• Doesn't currently support get/set of non-canned ACLs
• Doesn't currently support get/set of server access logging
• Haven't looked at the BitTorrent feature yet, probably trivial