MessagePack

MessagePack is an efficient binary serialization format. It lets you exchange data among multiple languages like JSON but it's faster and smaller. For example, small integers (like flags or error code) are encoded into a single byte, and typical short strings only require an extra byte in addition to the strings themselves.

If you ever wished to use JSON for convenience (storing an image with metadata) but could not for technical reasons (binary data, size, speed...), MessagePack is a perfect replacement.

require 'msgpack'
msg = [1,2,3].to_msgpack  #=> "\x93\x01\x02\x03"
MessagePack.unpack(msg)   #=> [1,2,3]

Add msgpack to your Gemfile to install with Bundler:

# Gemfile
gem 'msgpack'

Or, use RubyGems to install:

gem install msgpack

Or, build msgpack-ruby and install from a checked-out msgpack-ruby repository:

bundle
rake
gem install --local pkg/msgpack

• Create REST API returing MessagePack using Rails + RABL
• Store objects efficiently serialized by msgpack on memcached or Redis
  • In fact Redis supports msgpack in EVAL-scripts
• Upload data in efficient format from mobile devices such as smartphones
  • MessagePack works on iPhone/iPad and Android. See also Objective-C and Java implementations
• Design a portable protocol to communicate with embedded devices
  • Check also Fluentd which is a log collector which uses msgpack for the log format (they say it uses JSON but actually it's msgpack, which is compatible with JSON)
• Exchange objects between software components written in different languages
  • You'll need a flexible but efficient format so that components exchange objects while keeping compatibility

MessagePack for Ruby should run on x86, ARM, PowerPC, SPARC and other CPU architectures.

And it works with MRI (CRuby) and Rubinius. Patches to improve portability are highly welcomed.

Use MessagePack.pack or to_msgpack:

require 'msgpack'
msg = MessagePack.pack(obj)  # or
msg = obj.to_msgpack
File.binwrite('mydata.msgpack', msg)

Packer provides advanced API to serialize objects in streaming style:

# serialize a 2-element array [e1, e2]
pk = MessagePack::Packer.new(io)
pk.write_array_header(2).write(e1).write(e2).flush

See API reference for details.

Use MessagePack.unpack:

require 'msgpack'
msg = File.binread('mydata.msgpack')
obj = MessagePack.unpack(msg)

Unpacker provides advanced API to deserialize objects in streaming style:

# deserialize objects from an IO
u = MessagePack::Unpacker.new(io)
u.each do |obj|
  # ...
end

or event-driven style which works well with EventMachine:

# event-driven deserialization
def on_read(data)
  @u ||= MessagePack::Unpacker.new
  @u.feed_each(data) {|obj|
     # ...
  }
end

See API reference for details.

By default, symbols are serialized as strings:

packed = :symbol.to_msgpack     # => "\xA6symbol"
MessagePack.unpack(packed)      # => "symbol"

This can be customized by registering an extension type for them:

MessagePack::DefaultFactory.register_type(0x00, Symbol)

# symbols now survive round trips
packed = :symbol.to_msgpack     # => "\xc7\x06\x00symbol"
MessagePack.unpack(packed)      # => :symbol

The extension type for symbols is configurable like any other extension type. For example, to customize how symbols are packed you can just redefine Symbol#to_msgpack_ext. Doing this gives you an option to prevent symbols from being serialized altogether by throwing an exception:

class Symbol
  def to_msgpack_ext
    raise "Serialization of symbols prohibited"
  end
end

MessagePack::DefaultFactory.register_type(0x00, Symbol)

[1, :symbol, 'string'].to_msgpack  # => RuntimeError: Serialization of symbols prohibited

There are the timestamp extension type in MessagePack, but it is not registered by default.

To map Ruby's Time to MessagePack's timestamp for the default factory:

MessagePack::DefaultFactory.register_type(
  MessagePack::Timestamp::TYPE, # or just -1
  Time,
  packer: MessagePack::Time::Packer,
  unpacker: MessagePack::Time::Unpacker
)

See API reference for details.

Packer and Unpacker support Extension types of MessagePack.

# register how to serialize custom class at first
pk = MessagePack::Packer.new(io)
pk.register_type(0x01, MyClass1, :to_msgpack_ext) # equal to pk.register_type(0x01, MyClass)
pk.register_type(0x02, MyClass2){|obj| obj.how_to_serialize() } # blocks also available

# almost same API for unpacker
uk = MessagePack::Unpacker.new()
uk.register_type(0x01, MyClass1, :from_msgpack_ext)
uk.register_type(0x02){|data| MyClass2.create_from_serialized_data(data) }

MessagePack::Factory is to create packer and unpacker which have same extension types.

factory = MessagePack::Factory.new
factory.register_type(0x01, MyClass1) # same with next line
factory.register_type(0x01, MyClass1, packer: :to_msgpack_ext, unpacker: :from_msgpack_ext)
pk = factory.packer(options_for_packer)
uk = factory.unpacker(options_for_unpacker)

For MessagePack.pack and MessagePack.unpack, default packer/unpacker refer MessagePack::DefaultFactory. Call MessagePack::DefaultFactory.register_type to enable types process globally.

MessagePack::DefaultFactory.register_type(0x03, MyClass3)
MessagePack.unpack(data_with_ext_typeid_03) #=> MyClass3 instance

Alternatively, extension types can call the packer or unpacker recursively to generate the extension data:

Point = Struct.new(:x, :y)
factory = MessagePack::Factory.new
factory.register_type(
  0x01,
  Point,
  packer: ->(point, packer) {
    packer.write(point.x)
    packer.write(point.y)
  },
  unpacker: ->(unpacker) {
    x = unpacker.read
    y = unpacker.read
    Point.new(x, y)
  },
  recursive: true,
)
factory.load(factory.dump(Point.new(12, 34))) # => #<struct Point x=12, y=34>

Creating Packer and Unpacker objects is expensive. For best performance it is preferable to re-use these objects.

MessagePack::Factory#pool makes that easier:

factory = MessagePack::Factory.new
factory.register_type(
  0x01,
  Point,
  packer: ->(point, packer) {
    packer.write(point.x)
    packer.write(point.y)
  },
  unpacker: ->(unpacker) {
    x = unpacker.read
    y = unpacker.read
    Point.new(x, y)
  },
  recursive: true,
)
pool = factory.pool(5) # The pool size should match the number of threads expected to use the factory concurrently.

pool.load(pool.dump(Point.new(12, 34))) # => #<struct Point x=12, y=34>

MessagePack for Ruby provides a buffer API so that you can read or write data by hand, not via Packer or Unpacker API.

This MessagePack::Buffer is backed with a fixed-length shared memory pool which is very fast for small data (<= 4KB), and has zero-copy capability which significantly affects performance to handle large binary data.

Before building msgpack, you need to install bundler and dependencies.

gem install bundler
bundle install

Then, you can run the tasks as follows:

bundle exec rake build

bundle exec rake spec

bundle exec rake doc

To build -java gems for JRuby, run:

rake build:java

If this directory has Gemfile.lock (generated with MRI), remove it beforehand.

Online documentation (https://ruby.msgpack.org) is generated from the gh-pages branch. To update documents in gh-pages branch:

bundle exec rake doc
git checkout gh-pages
cp doc/* ./ -a

• Author
  • Sadayuki Furuhashi frsyuki@gmail.com
• Copyright
  • Copyright (c) 2008-2015 Sadayuki Furuhashi
• License
  • Apache License, Version 2.0