post past :: james murty

Amazon has announced beta support for the Copy Object operation in S3. This feature was pre-announced in March.

The copy functionality allows you to copy objects within or between your S3 buckets, and optionally to replace the metadata associated with the object in the process. The single new operation makes it possible to copy, move, and rename your S3 objects, and you can even update an object's metadata by copying the object in-place.

The feature is only in beta release at this stage and you cannot yet copy objects between the US and EU (Europe) locations.

I have added support for this operation to the JetS3t library toolkit, so check out the latest version from CVS if you want to try it out.

For those using the example code from Programming Amazon Web Services (affiliate link), here is a method you can add to the S3 class in S3.rb to implement the copy feature:

def copy_object(source_bucket_name, source_object_key,
  dest_bucket_name, dest_object_key, acl=nil, new_metadata=nil)

  headers = {}

  # Identify the source object
  headers['x-amz-copy-source'] = CGI::escape(
    source_bucket_name + '/' + source_object_key)

  # Copy metadata from original object, or replace the metadata.
  if new_metadata.nil?
    headers['x-amz-metadata-directive'] = 'COPY'
  else
    headers['x-amz-metadata-directive'] = 'REPLACE'
    headers.merge!(new_metadata)
  end

  # The Content-Length header must always be set.
  headers['Content-Length'] = '0'

  # Set the canned policy, may be: 'private', 'public-read',
  # 'public-read-write', 'authenticated-read'
  headers['x-amz-acl'] = acl if acl

  uri = generate_s3_uri(dest_bucket_name, dest_object_key)
  do_rest('PUT', uri, nil, headers)
  return true
end

The following command uses the copy_object method to copy an object named Object.txt from one bucket to another and make the resulting object publicly-accessible:

s3.copy_object('from-bucket', 'Object.txt',
  'to-bucket', 'Object.txt', 'public-read')

The following code will replace an object's metadata in-place without creating a new object. This allows us to set new values for the content type and "example-name" metadata items:

metadata = {
  'Content-Type' => 'text/xml',
  'x-amz-meta-example-name' => 'Example Value'}

s3.copy_object('from-bucket', 'Object.txt',
  'from-bucket', 'Object.txt', 'private', metadata)