NAME
Catmandu::Store::DBI::Bag - implementation of a Catmandu::Bag for DBI
SYNOPSIS
my $store = Catmandu::Store::DBI->new(
data_source => "dbi:SQLite:dbname=/tmp/test.db",
bags => {
data => {
mapping => {
_id => {
column => 'id',
type => 'string',
index => 1,
unique => 1
},
author => {
type => 'string'
},
subject => {
type => 'string',
},
_data => {
column => 'data',
type => 'binary',
serialize => 'all'
}
}
}
}
);
my $bag = $store->bag('data');
#SELECT
{
#SELECT * FROM DATA WHERE author = 'Nicolas'
my $iterator = $bag->select( author => 'Nicolas' );
}
#CHAINED SELECT
{
#SELECT * FROM DATA WHERE author = 'Nicolas' AND subject = 'ICT'
my $iterator = $bag->select( author => 'Nicolas' )->select( subject => 'ICT' );
}
#COUNT
{
#SELECT * FROM DATA WHERE author = 'Nicolas'
my $iterator = $bag->select( author => 'Nicolas' );
#SELECT COUNT(*) FROM ( SELECT * FROM DATA WHERE author = 'Nicolas' )
my $count = $iterator->count();
}
#DETECT
{
#SELECT * FROM DATA WHERE author = 'Nicolas' AND subject = 'ICT' LIMIT 1
my $record = $bag->select( author => 'Nicolas' )->detect( subject => 'ICT' );
}
#NOTES
{
#This creates an iterator with a specialized SQL query:
#SELECT * FROM DATA WHERE author = 'Nicolas'
my $iterator = $bag->select( author => 'Nicolas' );
#But this does not
my $iterator2 = $iterator->select( title => "Hello world" );
#'title' does not have a corresponding table column, so it falls back to the default implementation,
#and loops over every record.
}
{
#this is faster..
my $iterator = $bag->select( author => 'Nicolas' )->select( title => 'Hello world');
#..than
my $iterator2 = $bag->select( title => 'Hello world' )->select( author => 'Nicolas' );
#reason:
# the select statement of $iterator creates a specialized query, and so reduces the amount of records to loop over.
# $iterator is a L<Catmandu::Store::DBI::Iterator>.
# the select statement of $iterator2 does not have a specialized query, so it's a generic L<Catmandu::Iterator>.
# the second select statement of $iterator2 receives this generic object as its source, and can only loop over its records.
}DESCRIPTION
Catmandu::Store::DBI::Bag provides some method overrides specific for DBI interfaces, to make querying more efficient.
METHODS
store_with_table
Equivalent to the store accessor, but ensures that the table for this bag exists.
select($key => $val)
Overrides equivalent method in Catmandu::Bag.
Either returns a generic Catmandu::Iterator or a more efficient Catmandu::Store::DBI::Iterator.
Expect the following behaviour:
the key has a corresponding table column configured
a SQL where clause is created in the background:
.. WHERE $key = $val
Chained select statements with existing table columns result in a combined where clause:
.. WHERE $key1 = $val1 AND $key2 = $val2 ..The returned object is a Catmandu::Store::DBI::Iterator, instead of the generic Catmandu::Iterator.
the key does not have a corresponding table column configured
The returned object is a generic Catmandu::Iterator.
This iterator can only loop over the records provided by the previous Catmandu::Iterable.
A few important notes:
A select statement only results in a Catmandu::Store::DBI::Iterator, when it has a mapped key, and the previous iterator is either a Catmandu::Store::DBI::Bag or a Catmandu::Store::DBI::Iterator.
As soon as the returned object is a generic Catmandu::Iterator, any following select statement with mapped columns will not make a more efficient Catmandu::Store::DBI::Iterator.
In order to make your chained statements efficient, do the following:
create indexes on the table columns
put select statements with mapped keys in front, and those with non mapped keys at the end.
To configure table columns, see Catmandu::Store::DBI.
detect($key => $val)
Overrides equivalent method in Catmandu::Bag.
Also returns first record where $key matches $val.
Works like the select method above, but adds the SQL statement 'LIMIT 1' to the current SQL query in the background.
first()
Overrides equivalent method in Catmandu::Bag.
Also returns first record using the current iterator.
The parent method uses a generator, but fetches only one record.
This method adds the SQL statement 'LIMIT 1' to the current SQL query.
count()
Overrides equivalent method in Catmandu::Bag.
When the source is a Catmandu::Store::DBI::Bag, or a Catmandu::Store::DBI::Iterator, a specialized SQL query is created:
SELECT COUNT(*) FROM TABLE WHERE (..)The select statement of the source is between the parenthesises.