Ruby Protocol Buffers

Build Status

Protocol Buffers are a way of encoding structured data in an efficient yet extensible format. Google uses Protocol Buffers for almost all of its internal RPC protocols and file formats.

This library has two components: a compiler to turn .proto definitions into Ruby modules (extension .pb.rb), and a runtime to use protocol buffers defined by these modules.

The compiler relies on Google's C++ based compiler (protoc) for much of the heavy lifting -- this has huge advantages in ensuring compatibility and correctness. If you don't need cross-language interoperability you can create Message classes directly in ruby, in which case protoc is not needed. See "Writing Message Classes Directly" below.

This library is heavily optimized for encoding and decoding speed.

Because this is a tool for generating code, the RDoc documentation is a bit unusual. See the text in the ProtocolBuffers::Message class for details on what code is generated.

Installation

$ gem install ruby-protocol-buffers

If you want to compile .proto files to ruby, you'll need protoc version >= 2.2 (the Google Protocol Buffer compiler) installed in the environment where you will be compiling them. You do not need protoc installed to use the generated .pb.rb files.

For greater performance, consider installing the varint gem as well. This optional gem builds a small C extension to make parsing protocol buffers faster. If your application uses a Gemfile, add varint to the Gemfile alongside ruby-protocol-buffers.

Example

Given the file test.proto:

package Test;

message MyMessage
{
  optional string myField = 1;
}

Compile it to ruby using the command:

$ ruby-protoc test.proto

Then it can be used from ruby code:

require 'test.pb'
msg = Test::MyMessage.new(:myField => 'zomgkittenz')
open("test_msg", "wb") do |f|
  msg.serialize(f)
end
encoded = msg.serialize_to_string # or msg.to_s
Test::MyMessage.parse(encoded) == msg # true

Writing Message Classes Directly

Protocol Buffer definitions are often shared between applications written in different programming languages, and so are normally defined in .proto files and translated to ruby using the ruby-protoc binary.

However, it's quite simple to write ProtocolBuffers::Message classes directly when a .proto file isn't needed.

require 'protocol_buffers'

class User < ProtocolBuffers::Message
  required :string, :name, 1
  required :string, :email, 2
  optional :int32, :logins, 3
end

class Group < ProtocolBuffers::Message
  repeated User, :users, 1
  repeated Group, :subgroups, 2

  module GroupType
    include ProtocolBuffers::Enum
    Study = 1
    Play = 2
  end

  optional GroupType, :group_type, 3
end

This code is essentially equivalent to the code ruby-protoc will generate if given this .proto file:

message User
{
  required string name = 1;
  required string email = 2;
  optional int32 logins = 3;
}

message Group
{
  repeated User users = 1;
  repeated Group subgroups = 2;

  enum GroupType {
    Study = 1;
    Play = 2;
  }

  optional GroupType group_type = 3;
}

Using a hand-written Message subclass is the same as using a Message class generated by ruby-protoc.

group = Group.new(:group_type => Group::GroupType::Play)
group.users << User.new(:name => 'test user', :email => '[email protected]')
open("group1.test", "wb") do |f|
  group.serialize(f)
end

Features

Supported Features

  • messages, enums, field types, all basic protobuf features
  • packages
  • imports
  • nested types
  • passing on unknown fields when re-serializing a message
  • groups
  • RPC stubbing

Unsupported Features

  • extensions
  • packed option (could be useful)
  • accessing custom options

Probably Never to be Supported

  • the unsupported options (java_*, optimize_for, message_set_wire_format, deprecated)

Authors

Brian Palmer (http://github.com/codekitchen)

Source

http://github.com/mozy/ruby-protocol-buffers

License

See the LICENSE file included with the distribution for licensing and copyright details.