S3 Upload Server

Data rescue efforts often require a method for posting large files to S3 buckets for sharing. This server allows users to post files up to 5GB in size to S3 without needing AWS credentials or knowing how to use the command-line.

I'm still investigating doing multipart uploads from the browser, which would allow files larger than 5GB. I'll update this repo as progress is made.

Features

• Super-Simple S3 Uploading: Allow users to upload files to S3 with a browser link. No passing AWS credentials, no scripts or command-line experience necessary.
• "Burner" S3 Credentials Scoped to an available filepath For +5Gig files, or users that wish to use the command line, use amazon STS to create a set of credentials that will only allow uploading of a specified path.
• Optional Basic Http Authorization: Set a global username & password to limit access to the upload area with a simple user/pass combo you can pass around to trusted parties.
• Deadline Setting: Configure the server to stop accepting new uploads after a certain time. Useful to "set & forget" the server without having it accept new uploads forever.
• Configurable View Templates: Set messages & instructions using the config.json file.
• Taken-Path-Suffixing Prevent overwriting existing files by appending a numerical suffix to existing paths
• Upload Directories Set a list of directories (paths) that the uploader is allowed to upload to

S3 Requirements

In order for this to work you'll need two settings on the S3 side to be properly configured:

• An account with an access key & secret that has write access to the bucket. See sample user policies
• An S3 Bucket with a a CORS configuration that allows PUT & POST requests from your server. That would look something like this:

<?xml version="1.0" encoding="UTF-8"?>
<CORSConfiguration xmlns="http://s3.amazonaws.com/doc/2006-03-01/">
  <CORSRule>
    <AllowedOrigin>*</AllowedOrigin>
    <AllowedMethod>GET</AllowedMethod>
    <MaxAgeSeconds>3000</MaxAgeSeconds>
    <AllowedHeader>*</AllowedHeader>
  </CORSRule>
  <CORSRule>
    <AllowedOrigin>[app url]</AllowedOrigin>
    <AllowedMethod>POST</AllowedMethod>
    <AllowedMethod>PUT</AllowedMethod>
    <AllowedHeader>*</AllowedHeader>
  </CORSRule>
</CORSConfiguration>

The second AllowedOrigin should be the url of the server you're setting up, as described below. If, for example the app you posted was available at http://data-uploader.herokuapp.com, you'd set the second CORSRule AllowedOrigin to be that url, http://data-uploader.herokuapp.com.

Posting Server to Heroku

Posting this server to Heroku is the easiest way to get up & running publically. Make sure you have a free heroku account, and have installed the heroku CLI on your machine before starting.

1. Clone the repo.
2. Navigate to repo directory & run heroku create [app-name].
3. Set enviornment variables with heroku config:set AWS_REGION=[bucket region] AWS_S3_BUCKET_NAME=[bucket name] AWS_ACCESS_KEY_ID=[access key] AWS_SECRET_ACCESS_KEY=[access secret]
4. Run git push heroku master to push your code & start the server.
5. Navigate to http://[app-name].herokuapp.com in your browser & test you're uploads.

Configuring the server

The server accepts configuration in two places, a config.json file, and enviornment variables. Secrets such as the AWS_SECRET_ACCESS_KEY should always be set with enviornment variables.. If you're running this code locally it can be convenient to set these values in the config.json for testing purposes, but they should never be checked into the git repository.

Burner Credentials

To use burner credentials, first the EnableBurnerCredentials configuration option must be true in configuration. Additionally, the configured AWS account must be allowed to perform the sts:GetFederationToken action. For more info, check the sample user policies.

This will open up an endpoint that you can issue GET requests to: /burner?object_name=example.zip&dir=example_directory&format=json

• object_name is the name of the file to upload. If the requested name is already in the bucket an untaken name will be returned.
• This url will use any of the configured directories, specified by the dir param. If directories aren't specified this param will not be allowed.
• The format=json will return json of credentials only. If format is left unspecified the returned format will be an HTML page with directions on how to use the credentials.

TODO:

• Client-Side ETA for uploads
• Figure out a web-based solution for files larger than 5GB
• Multi-File Upload?
• Have site collect uploader details and save to S3 Bucket in json log files
• Calculate MD5 File Hash Client-side
• Upload Size Restrictions
• Upload Rate Limiting in GB Uploaded / Minute or something
• Make x-amz-public-read header optional for uploads