++ed by:

4 PAUSE users
3 non-PAUSE users.

Author image Scott Smith
and 1 contributors


UR - rich declarative transactional objects


This document describes UR version v0.11.


    use UR; 

    ## no database

    class Foo { is => 'Bar', has => [qw/prop1 prop2 prop3/] };
    $o1 = Foo->create(prop1 => 111, prop2 => 222, prop3 => 333);
    @o = Foo->get(prop2 => 222, prop1 => [101,111,121], 'prop3 between' => [200, 400]);
    # returns one object


    @o = Foo->get(prop2 => 222, prop1 => [101,111,121], 'prop3 between' => [200, 400]);
    # returns zero objects
    @o = Foo->get(prop2 => 222, prop1 => [101,111,121], 'prop3 between' => [200, 400]);
    # returns one object again

    ## database

    class Animal {
        has => [
            favorite_food => { is => 'Text', doc => "what's yummy?" },
        data_source => 'MyDB1',
        table_name => 'Animal'

    class Cat { 
        is => 'Animal', 
        has => [
            feet    => { is => 'Number' },
            fur     => { is => 'Text', valid_values => [qw/fluffy scruffy/] },
        data_source => 'MyDB1',
        table_name => 'Cat'

    Cat->create(feet => 4, fur => 'fluffy', favorite_food => 'taters');

    @cats = Cat->get(favorite_food => ['taters','sea bass']);

    $c = $cats[0];
    print $c->feet,"\n";




UR is a class framework and object/relational mapper for Perl. It starts with the familiar Perl meme of the blessed hash reference as the basis for object instances, and extends its capabilities with ORM (object-relational mapping) capabilities, object cache, in-memory transactions, more formal class definitions, metadata, documentation system, iterators, command line tools, etc.

UR can handle multiple column primary and foreign keys, SQL joins involving class inheritance and relationships, and does its best to avoid querying the database unless the requested data has not been loaded before. It has support for SQLite, Oracle, Mysql and Postgres databases, and the ability to use a text file as a table.

UR uses the same syntax to define non-persistent objects, and supports in-memory transactions for both.



UR::Manual::Overview - UR from Ten Thousand Feet

UR::Manual::Tutorial - Getting started with UR

UR::Manual::Presentation - Slides for a presentation on UR

UR::Manual::Cookbook - Recepies for getting stuff working

UR::Manual::Metadata - UR's metadata system

UR::Object::Type::Initializer - Defining classes

UR::Manual::UR - UR's command line tool

Basic Entities

UR::Object - Pretty much everything is-a UR::Object

UR::Object::Type - Metadata class for Classes

UR::Object::Property - Metadata class for Properties

UR::Namespace - Manage packages and classes

UR::Context - Software transactions and More!

UR::DataSource - How and where to get data


First create a Namespace class for your application, Music.pm:

    package Music;
    use UR;

    class Music {
        is => 'UR::Namespace'


Next, define a data source representing your database, Music/DataSource/DB1.pm

    package Music::DataSource::DB1;
    use Music;
    class Music::DataSource::DB1 {
        is => ['UR::DataSource::Mysql'],
        has_constant => [
            server  => { value => 'mysql.example.com' },
            login   => { value => 'mysqluser' },
            auth    => { value => 'mysqlpasswd' },
    or to get something going quickly, SQLite has smart defaults...

    class Music::DataSource::DB1 {
        is => 'UR::DataSource::SQLite',

Create a class to represent artists, who have many CDs, in Music/Artist.pm

    package Music::Artist;
    use Music;

    class Music::Artist {
        id_by => 'artist_id',
        has => [ 
            name => { is => 'Text' },
            cds  => { is => 'Music::Cd', is_many => 1, reverse_as => 'artist' }
        data_source => 'Music::DataSource::DB1',
        table_name => 'ARTISTS',

Create a class to represent CDs, in Music/Cd.pm

    package Music::Cd;
    use Music;
    class Music::Cd {
        id_by => 'cd_id',
        has => [
            artist => { is => 'Music::Artist', id_by => 'artist_id' },
            title  => { is => 'Text' },
            year   => { is => 'Integer' },
            artist_name => { via => 'artist', to => 'name' },
        data_source => 'Music::DataSource::DB1',
        table_name => 'CDS',

If the database does not exist, you can run this to generate the tables and columns from the classes you've written (very experimental):

  $ cd Music
  $ ur update schema

If the database existed already, you could have done this to get it to write the last 2 classes for you:

  $ cd Music;
  $ ur update classes

Regardless, if the classes and database tables are present, you can then use these classes in your application code:

    # Using the namespace enables auto-loading of modules upon first attempt to call a method
    use Music;  
    # This would get back all Artist objects:
    my @all_artists = Music::Artist->get();

    # After the above, further requests would be cached
    # if that set were large though, you might want to iterate gradually:
    my $artist_iter = Music::Artist->create_iterator();

    # Get the first object off of the iterator
    my $first_artist = $artist_iter->next();

    # Get all the CDs published in 2007 for the first artist
    my @cds_2007 = Music::Cd->get(year => 2007, artist => $first_artist);
    # Use non-equality operators:
    my @same_some_artists = Music::Artist->get(
        'name like' => 'John%',
        'year between' => ['2004','2009']

    # This will use a JOIN with the ARTISTS table internally to filter
    # the data in the database.  @some_cds will contain Music::Cd objects.
    # As a side effect, related Artist objects will be loaded into the cache
    my @some_cds = Music::Cd->get(
        year => '2007', 
        'artist_name like' => 'Bob%' 

    # These values would be cached...
    my @artists_for_some_cds = map { $_->artist } @some_cds;
    # This will use a join to prefetch Artist objects related to the
    # objects that match the filter
    my @other_cds = Music::Cd->get(
        'title like' => '%White%',
        -hints => ['artist']
    my $other_artist_0 = $other_cds[0]->artist;  # already loaded so no query
    # create() instantiates a new object in the current "context", but does not save 
    # it in the database.  It will autogenerate its own cd_id:
    my $new_cd = Music::Cd->create(
        title => 'Cool Album',
        year  => 2009

    # Assign it to an artist; fills in the artist_id field of $new_cd
    # Save all changes in the current transaction back to the database(s)
    # which are behind the changed objects.

Environment Variables

UR uses several environment variables to do things like run with database commits disabled, watching SQL queries run, examine query plans, and control cache size, etc.

These make development and debugging fast and easy.

See UR::Env for details.


Class::Autouse Cwd Data::Dumper Date::Calc Date::Parse DBI File::Basename FindBin FreezeThaw Path::Class Scalar::Util Sub::Installer Sub::Name Sys::Hostname Text::Diff Time::HiRes XML::Simple


UR was built by the software development team at the Genome Center at Washington University School of Medicine. Incarnations of it run laboratory automation and analysis systems for high-throughput genomics.

 Scott Smith        sakoht@cpan.org         Orginal Architecture
 Anthony Brummett   brummett@cpan.org   Primary Development

 Craig Pohl
 Todd Hepler
 Ben Oberkfell
 Kevin Crouse
 Adam Dukes
 Indraniel Das
 Shin Leong
 Eddie Belter
 Ken Swanson
 Scott Abbott
 Alice Diec
 William Schroeder
 Eric Clark
 Shawn Leonard
 Lynn Carmichael
 Jason Walker
 Amy Hawkins
 Gabe Sanderson
 James Weible
 James Eldred
 Michael Kiwala
 Mark Johnson
 Kyung Kim
 Jon Schindler
 Justin Lolofie
 Chris Harris
 Jerome Peirick
 Ryan Richt
 John Osborne
 Josh McMichael
 David Dooling


Copyright (C) 2002-2009 Washington University in St. Louis, MO.

This sofware is licensed under the same terms as Perl itself. See the LICENSE file in this distribution.