=head1 NAME Redis::Object - Use Redis with an ORMish interface =head1 DESCRIPTION Implements a scaled down ORM-like interface to access Redis as a database. If you want to use Redis as a cache, use L instead. =head1 SYNOPSIS package MyRedisDatabase; use Moose; extends qw/ Redis::Object /; has tables => ( isa => 'ArrayRef[Str]', is => 'ro', default => sub { [ qw/ SomeTable / ] } ); __PACKAGE__->make_immutable; package MyRedisDatabase::SomeTable; use Moose; with qw/ Redis::Object::Table /; has attrib1 => ( isa => 'Str', is => 'rw', default => 'Something' ); has attrib2 => ( isa => 'Int', is => 'rw' ); has attrib3 => ( isa => 'HashRef', is => 'rw' ); has attrib4 => ( isa => 'ArrayRef', is => 'rw' ); sub INDEX_ATTRIBUTES { qw/ attrib1 / } __PACKAGE__->make_immutable; package main; # init database my $db = MyRedisDatabase->new( server => '127.0.0.1:6379' ); # create item my $item = $db->create( SomeTable => { attrib1 => "Hello", attrib2 => 123, attrib3 => { something => "serializeable" }, attrib4 => [ 1..99 ] } ); print "Created ". $item->id; # fetch item by id my $item = $db->find( SomeTable => $id ); print $item->attrib1. "\n"; # search items my $result = $db->search( SomeTable => { attrib1 => "Hello", attrib2 => 123 } ); while( my $item = $result->next ) { print "Found ". $item->id. "\n"; } # update item $item->attrib1( "bla" ); $db->update( $item, { attrib2 => 333 } ); $item->update( { attrib1 => "Hallo" } ); # remove an item $db->remove( $item ); $item->remove; # clear a table (remvoe all entries!) $db->truncate( 'SomeTable' ); =head1 YOU SHOULD KNOW =head2 Searching / Sorting Redis is more than a simple key-value store - but it is no relational database, by any means. So limit your expectations towards complex searching or sorting. This interface implements searching by primary key (an integer ID, which is automatically assigened to each "row" in the database), searching indexed String values with compare- and prefix-search. All search capability aside from this results in a full "table" scan. =head2 Indices This interface allows you to define certain columes as indexed. Those columes should always be strings - not numbers, nor even more complex data strucutres. Those strings you can search with wildcars, such as "word*" or "w*rd*" =head2 Structure This interface will store your instances, represented by L-objects, in a distinct strucuture. Do not try to use this interface with pre-existing data! The structure relates to the L attributes of your classes. Assuming the following table-class: package MyDB::MyTable; use Moose; with qw/ Redis::Object::Table /; has somekey => ( isa => "Str", is => "rw", required => 1 ); has otherkey => ( isa => "Int", is => "rw", required => 1 ); sub INDEX_ATTRIBUTES { qw/ somekey / } The resulting "rows" would look something like this # contains the an ID timestamp, used for certain lookups mytable:1:_ # contains the values of both attributres mytable:1:somekey mytable:1:otherkey # indexed key "somekey" for fast lookup mytable:1:_:somekey:The_Value There is also a special key/value per table, which contains an incrementing integer for the primary key mytable:_id =cut