Introduction

This is a Ruby driver for the 10gen Mongo DB. For more information about Mongo, see www.mongodb.org.

Note: this driver is still alpha quality. The API will change, as may the data saved to the database (especially primary key values). Do not use this for any production data yet.

Start by reading the XGen::Mongo::Driver::Mongo and XGen::Mongo::Driver::DB documentation, then move on to XGen::Mongo::Driver::Collection and XGen::Mongo::Driver::Cursor.

A quick code sample:

require 'mongo'

include XGen::Mongo::Driver

db = Mongo.new('localhost').db('sample-db')
coll = db.collection('test')

coll.clear
3.times { |i| coll.insert({'a' => i+1}) }
puts "There are #{coll.count()} records. Here they are:"
coll.find().each { |doc| puts doc.inspect }

This driver also includes an implementation of a GridStore class, a Ruby interface to Mongo’s GridFS storage. NOTE: the GridStore code may be moved to a separate project.

Installation

Install the “mongo” gem by typing

$ gem sources -a http://gems.github.com
$ sudo gem install mongodb-mongo-ruby-driver

The first line tells RubyGems to add the GitHub gem repository. You only need to run this command once.

From the GitHub source

The source code is available at github.com/mongodb/mongo-ruby-driver. You can either clone the git repository or download a tarball or zip file. Once you have the source, you can use it from wherever you downloaded it or you can install it as a gem from the source by typing

$ rake gem:install

Demo

You can see and run the examples if you’ve downloaded the source. Mongo must be running, of course.

$ ruby examples/simple.rb

See also the test code, especially tests/test_db_api.rb.

For the GridFS class GridStore, see the tests.

The Driver

Here is some simple example code:

require 'rubygems'        # not required for Ruby 1.9
require 'mongo'

include XGen::Mongo::Driver
db = Mongo.new.db('my-db-name')
things = db.collection('things')

things.clear
things.insert('a' => 42)
things.insert('a' => 99, 'b' => Time.now)
puts things.count                               # => 2
puts things.find('a' => 42).next_object.inspect # {"a"=>42}

GridStore

The GridStore class is a Ruby implementation of Mongo’s GridFS file storage system. An instance of GridStore is like an IO object. See the rdocs for details.

Note that the GridStore class is not automatically required when you require ‘mongo’. You need to require ‘mongo/gridfs’.

Example code:

GridStore.open(database, 'filename', 'w') { |f|
  f.puts "Hello, world!"
}
GridStore.open(database, 'filename, 'r') { |f|
  puts f.read         # => Hello, world!\n
}
GridStore.open(database, 'filename', 'w+') { |f|
  f.puts "But wait, there's more!"
}
GridStore.open(database, 'filename, 'r') { |f|
  puts f.read         # => Hello, world!\nBut wait, there's more!\n
}

Notes

String Encoding

The BSON (“Binary JSON”) format used to communicate with Mongo requires that strings be UTF-8 (en.wikipedia.org/wiki/UTF-8).

Ruby 1.9 has built-in character encoding support. All strings sent to Mongo and received from Mongo are converted to UTF-8 when necessary, and strings read from Mongo will have their character encodings set to UTF-8.

When used with Ruby 1.8, the bytes in each string are written to and read from Mongo as-is. If the string is ASCII all is well, because ASCII is a subset of UTF-8. If the string is not ASCII then it may not be a well-formed UTF-8 string.

The DB class

Primary Key Factories

A basic Mongo driver is not responsible for creating primary keys or knowing how to interpret them. You can tell the Ruby Mongo driver how to create primary keys by passing in the :pk option to the Mongo#db method.

include XGen::Mongo::Driver
db = Mongo.new.db('dbname', :pk => MyPKFactory.new)

A primary key factory object must respond to :create_pk, which should take a hash and return a hash which merges the original hash with any primary key fields the factory wishes to inject. NOTE: if the object already has a primary key, the factory should not inject a new key; this means that the object is being used in a repsert but it already exists. The idea here is that when ever a record is inserted, the :pk object’s create_pk method will be called and the new hash returned will be inserted.

Here is a sample primary key factory, taken from the tests:

class TestPKFactory
  def create_pk(row)
    row['_id'] ||= XGen::Mongo::Driver::ObjectID.new
    row
  end
end

Here’s a slightly more sophisticated one that handles both symbol and string keys. This is the PKFactory that comes with the MongoRecord code (an ActiveRecord-like framework for non-Rails apps) and the AR Mongo adapter code (for Rails):

class PKFactory
  def create_pk(row)
    return row if row[:_id]
    row.delete(:_id)      # in case it exists but the value is nil
    row['_id'] ||= XGen::Mongo::Driver::ObjectID.new
    row
  end
end

A database’s PK factory object may be set after you obtain it, but only once. The only reason it is changeable is so that libraries such as MongoRecord that use this driver can set the PK factory after obtaining the database but before using it for the first time.

Strict mode

Each database has an optional strict mode. If strict mode is on, then asking for a collection that does not exist will raise an error, as will asking to create a collection that already exists. Note that both these operations are completely harmless; strict mode is a programmer convenience only.

To turn on strict mode, either pass in :strict => true when obtaining a DB object or call the :strict= method:

db = XGen::Mongo::Driver::Mongo.new.db('dbname', :strict => true)
# I'm feeling lax
db.strict = false
# No, I'm not!
db.strict = true

The method DB#strict? returns the current value of that flag.

Testing

If you have the source code, you can run the tests.

$ rake test

The tests assume that the Mongo database is running on the default port. You can override the default host (localhost) and port (Mongo::DEFAULT_PORT) by using the environment variables MONGO_RUBY_DRIVER_HOST and MONGO_RUBY_DRIVER_PORT.

The project mongo-qa (github.com/mongodb/mongo-qa) contains many more Mongo driver tests that are language independent. To run thoses tests as part of the “rake test” task, download the code “next to” this directory. So, after installing the mongo-qa code you would have these two directories next to each other:

$ ls
mongo-qa
mongo-ruby-driver
$ rake test

The tests run just fine if the mongo-qa directory is not there.

Additionally, the script bin/validate is used by the mongo-qa project’s validator script.

Documentation

This documentation is available online at mongo.rubyforge.org. You can generate the documentation if you have the source by typing

$ rake rdoc

Then open the file html/index.html.

Release Notes

See the git log comments.

To Do

  • Add group_by. Need to figure out how we are going to send functions. The current thinking is that Mongo will allow a subset of JavaScript (which we would have to send as a string), but this is still under discussion.

  • Tests for update and repsert.

  • Add a way to specify a collection of databases on startup (a simple array of IP address/port numbers, perhaps, or a hash or something). The driver would then find the master and, on each subsequent command, ask that machine if it is the master before proceeding.

  • Introduce optional per-database and per-collection PKInjector.

  • More tests.

Optimizations

  • Only update message sizes once, not after every write of a value. This will require an explicit call to update_message_length in each message subclass.

  • ensure_index commands should be cached to prevent excessive communication with the database. (Or, the driver user should be informed that ensure_index is not a lightweight operation for the particular driver.)

Credits

Adrian Madrid, [email protected]

  • bin/mongo_console

  • examples/benchmarks.rb

  • examples/irb.rb

  • Modifications to examples/simple.rb

  • Found plenty of bugs and missing features.

  • Ruby 1.9 support.

  • Gem support.

  • Many other code suggestions and improvements.

License

Copyright © 2008-2009 10gen Inc.

This program is free software: you can redistribute it and/or modify it under the terms of the GNU Affero General Public License, version 3, as published by the Free Software Foundation.

This program is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU Affero General Public License for more details.

See www.gnu.org/licenses for a copy of the GNU Affero General Public License.