Skip to content


Subversion checkout URL

You can clone with
Download ZIP
branch: master
Fetching contributors…

Cannot retrieve contributors at this time

260 lines (199 sloc) 11.665 kB
MongoDBx::Class - Flexible ORM for MongoDB databases
use MongoDBx::Class;
# create a new instance of the module and load a model schema
my $dbx = MongoDBx::Class->new(namespace => 'MyApp::Model::DB');
# connect to a MongoDB server
my $conn = $dbx->connect(host => 'localhost', port => 27017);
# be safe by default
$conn->safe(1); # we could've also just passed "safe => 1" to $dbx->connect() above
# get a MongoDB database
my $db = $conn->get_database('people');
# insert a person
my $person = $db->insert({ name => 'Some Guy', birth_date => '1984-06-12', _class => 'Person' });
print "Created person ".$person->name." (".$person->id.")\n";
$person->update({ name => 'Some Smart Guy' });
WARNING: MongoDBx::Class is still in beta status. Please do not rely on
it yet for production use.
MongoDBx::Class is a flexible object relational mapper (ORM) for MongoDB
databases. Given a schema-like collection of document classes,
MongoDBx::Class expands MongoDB objects (hash-refs in Perl) from the
database into objects of those document classes, and collapses such
objects back to the database.
MongoDBx::Class takes advantage of the fact that Perl's MongoDB driver
is Moose-based to extend and tweak the driver's behavior, instead of
wrapping it. This means MongoDBx::Class does not define its own syntax,
so you simply use it exactly as you would the MongoDB driver directly.
That said, MongoDBx::Class adds some sugar that enhances and simplifies
the syntax unobtrusively (either use it or don't). Thus, it is
relatively easy to convert your current MongoDB applications to
MongoDBx::Class. A collection in MongoDBx::Class
"isa('MongoDB::Collection')", a database in MongoDBx::Class
"isa('MongoDB::Database')", etc.
As opposed to other ORMs (even non-MongoDB ones), MongoDBx::Class
attempts to stay as close as possible to MongoDB's non-schematic nature.
While most ORMs enforce using a single collection (or table in the SQL
world) for every object class, MongoDBx::Class allows you to store
documents of different classes in different collections (and even
databases). A collection can hold documents of many different classes.
Not only that, as MongoDBx::Class is Moose based, you can easily create
very flexible schemas by using concepts such as inheritance and roles.
For example, say you have a collection called 'people' with documents
representing, well, people, but these people can either be teachers or
students. Also, students may assume the role "hall monitor". With
MongoDBx::Class, you can create a common base class, say "People", and
two more classes that extend it - "Teacher" and "Student" with
attributes that are only relevant to each one. You also create a role
called "HallMonitor", possibly with some methods of its own. You can
save all these "people documents" into a single MongoDB collection, and
when fetching documents from that collection, they will be properly
expanded to their correct classes (though you will have to apply roles
yourself - at least for now).
As MongoDB is rather young, there aren't many options out there, though
CPAN has some pretty good ones, and will probably have more as MongoDB
popularity rises.
The first MongoDB ORM in CPAN was Mongoose, and while it's a very good
ORM, MongoDBx::Class was mainly written to overcome some limitations of
Mongoose. The biggest of these limitations is that in order to provide a
more comfortable syntax than MongoDB's native syntax, Mongoose makes the
unfortunate decision of being implemented as a singleton, meaning only
one instance of a Mongoose-based schema can be used in an application.
That essentially kills multithreaded applications. Say you have a
Plack-based (doesn't have to be Plack-based though) web application
deployed via Starman (or any other web server for that matter), which is
a pre-forking web server - you're pretty much doomed. As MongoDB's
driver states, it doesn't support connection pooling, so every fork has
to have its own connection to the MongoDB server. Mongoose being a
singleton means your threads will not have a connection to the server,
and you're screwed. MongoDBx::Class does not suffer this limitation. You
can start as many connections as you like. If you're running in a
pre-forking environment, you don't have to worry about it at all.
Other differences from Mongoose include:
* Mongoose creates its own syntax, MongoDBx::Class doesn't, you use
MongoDB's syntax directly.
* A document class in Mongoose is connected to a single collection
only, and a collection can only have documents of that class.
MongoDBx::Class doesn't have that limitation. Do what you like.
* Mongoose has limited support for multiple database usage. With
MongoDBx::Class, you can use as many databases as you want.
* MongoDBx::Class is way faster. While I haven't performed any real
benchmarks, an application converted from Mongoose to
MongoDBx::Class showed an increase of speed in orders of magnitude.
* In Mongoose, your document class attributes are expected to be
read-write (i.e. "is => 'rw'" in Moose), otherwise expansion will
fail. This is not the case with MongoDBx::Class, your attributes can
safely be read-only.
Another ORM for MongoDB is Mongrel, which doesn't use Moose and is thus
lighter (though as MongoDB is already Moose-based, I see no benefit
here). It uses Oogly for data validation (while Moose has its own type
validation), and seems to define its own syntax as well. Unfortunately,
documentation is currently lacking, and I haven't given it a try, so I
can't draw specific comparisons here.
Even before Mongoose was born, you could use MongoDB as a backend for
KiokuDB, by using KiokuDB::Backend::MongoDB. However, KiokuDB is
considered a database of its own and uses some conventions which doesn't
fit well with MongoDB. Mongoose::Intro already gives a pretty convincing
case when and why you should or shouldn't want to use KiokuDB.
There are a few caveats and important facts to take note of when using
MongoDBx::Class as of today:
* It's alpha software. This is an early release, and bugs are found
(and fixed) all the time. Don't rely on it for production use yet.
You have been warned.
* MongoDBx::Class's flexibility is dependant on its ability to
recognize which class a document in a MongoDB collection expands to.
Currently, MongoDBx::Class requires every document to have an
attribute called "_class" that contains the name of the document
class to use. This isn't very comfortable, but works. I'm still
thinking of ways to expand documents without this. This pretty much
means that you will have to perform some preparations to use
existing MongoDB database with MongoDBx::Class - you will have to
update every document in the database with the "_class" attribute.
* References (representing joins) are expected to be in the DBRef
format, as defined in
<>. If your
database references aren't in this format, you'll have to convert
them first.
* The '_id' attribute of all your documents has to be an internally
generated MongoDB::OID. This limitation may or may not be lifted in
the future.
To start using MongoDBx::Class, please read MongoDBx::Class::Tutorial.
It also contains a list of frequently asked questions.
A string representing the namespace of the MongoDB schema used (e.g.
"MyApp::Schema"). Your document classes, structurally speaking, should
be descendants of this namespace (e.g. "MyApp::Schema::Article",
A hash-ref of document classes found when loading the schema.
new( namespace => $namespace )
Creates a new instance of this module. Requires the namespace of the
database schema to use. The schema will be immediately loaded, but no
connection to a MongoDB server is made yet.
connect( %options )
Initiates a new connection to a MongoDB server running on a certain host
and listening to a certain port. %options is the hash of attributes that
can be passed to "new()" in MongoDB::Connection, plus the 'safe'
attribute from MongoDBx::Class::Connection. You're mostly expected to
provide the 'host' and 'port' options. If a host is not provided,
'localhost' is used. If a port is not provided, 27017 (MongoDB's default
port) is used. Returns a MongoDBx::Class::Connection object.
NOTE: Since version 0.7, the created connection object isn't saved in
the top MongoDBx::Class object, but only returned, in order to be more
like how connection is made in MongoDB (and to allow multiple
connections). This change breaks backwords compatibility.
The following methods are only to be used internally.
Automatically called when creating a new instance of this module. This
loads the schema and saves a hash-ref of document classes found in the
object. Automatic loading courtesy of Module::Pluggable.
* Improve the tests.
* Make the "isa" option in MongoDBx::Class::Moose's relationship
types consistent. Either use the full package names or the short
class names.
* Add support for the AnyMongo driver.
* Add support for document attributes which are not to be saved in
the database.
* Try to find a way to not require documents to have the _class
Ido Perlmuter, "<ido at>"
Please report any bugs or feature requests to "bug-mongodbx-class at", or through the web interface at
<>. I will
be notified, and then you'll automatically be notified of progress on
your bug as I make changes.
You can find documentation for this module with the perldoc command.
perldoc MongoDBx::Class
You can also look for information at:
* RT: CPAN's request tracker
* AnnoCPAN: Annotated CPAN documentation
* CPAN Ratings
* Search CPAN
MongoDB, Mongoose, Mongrel, KiokuDB::Backend::MongoDB.
Rodrigo de Oliveira, author of Mongoose, whose code greatly assisted me
in writing MongoDBx::Class.
Copyright 2010-2011 Ido Perlmuter.
This program is free software; you can redistribute it and/or modify it
under the terms of either: the GNU General Public License as published
by the Free Software Foundation; or the Artistic License.
See for more information.
Jump to Line
Something went wrong with that request. Please try again.