Chunked downloads

[TrevorHinesley]

For getting large files from a user's Dropbox, their API allows you to specify a range of bytes--is this possible in the current dropbox_api gem's DSL?

[Jesus]

Sorry for the late reply.

I couldn't find the option to download a specific range of bytes anywhere in their API documentation. Could you link to any resource related to this?

If this is actually possible I'd be interested in implementing this feature.

[TrevorHinesley]

https://www.dropbox.com/developers-v1/core/docs#files-GET Look at the Notes section

[Jesus]

That link is for Dropbox API v1, this library implements API v2.

However, I could find this in the docs from v2:

As with content-upload endpoints, arguments are passed in the Dropbox-API-Arg request header or arg URL parameter. The response body contains file content, so the result will appear as JSON in the Dropbox-API-Result response header. These endpoints are also on the content.dropboxapi.com domain. These endpoints also support HTTP GET along with ETag-based caching (If-None-Match) and HTTP range requests.

Source: https://www.dropbox.com/developers/documentation/http/documentation#formats

So, you're right and it should be possible to download a range of bytes. I'm not sure yet about the details on how this is meant to work.

I'll see if we can have support for this feature soon.

[apurvis]

https://developer.mozilla.org/en-US/docs/Web/HTTP/Range_requests

[gbt5]

Here's my monkeypatch:

module DropboxApi::Endpoints
  class Base
    def process_response(raw_response)
      # Official Dropbox documentation for HTTP error codes:
      # https://www.dropbox.com/developers/documentation/http/documentation#error-handling
      case raw_response.status
      when 200, 206, 409
        # Status code 409 is "Endpoint-specific error". We need to look at
        # the response body to build an exception.
        build_result(raw_response.env[:api_result])
      when 429
        error = DropboxApi::Errors::TooManyRequestsError.build(
          raw_response.env[:api_result]["error_summary"],
          raw_response.env[:api_result]["error"]["reason"]
        )

        error.retry_after = raw_response.headers["retry-after"].to_i

        raise error
      else
        raise DropboxApi::Errors::HttpError,
          "HTTP #{raw_response.status}: #{raw_response.body}"
      end
    end
  end

  class ContentChunkedDownload < DropboxApi::Endpoints::Base
    def initialize(builder)
      @connection = builder.build("https://content.dropboxapi.com") do |c|
        c.response :decode_result
      end
    end

    def build_request(params)
      body = nil

      range_from = params.delete(:range_from)
      range_to   = params.delete(:range_to)
      range = "bytes=#{range_from}-#{range_to}" unless range_from.nil? || range_from.nil?

      headers = {
        'Dropbox-API-Arg' => JSON.dump(params),
        'Content-Type' => ''
      }

      headers.merge!({ 'Range' => range }) unless range.nil?

      return body, headers
    end

    def perform_request(params)
      response = get_response(params)
      api_result = process_response response

      # TODO: Stream response, current implementation will fail with very large
      #       files.
      yield response.body if block_given?

      api_result
    end
  end
end

module DropboxApi::Endpoints::Files
  class DownloadChunked < DropboxApi::Endpoints::ContentChunkedDownload
    Method      = :post
    Path        = "/2/files/download".freeze
    ResultType  = DropboxApi::Metadata::File
    ErrorType   = DropboxApi::Errors::DownloadError

    # Download a file from a users Dropbox.
    #
    # @param path [String] The path of the file to download.
    add_endpoint :download_chunk do |path, from, to, &block|
      perform_request({:path => path}.merge(range_from: from, range_to: to), &block)
    end
  end
end