DBIx-Squirrel

DBI, minus the boilerplate, plus a pipeline.

With DBIx::Squirrel, you still write SQL, as you do with the DBI. It takes away the friction around it: placeholders in whatever style you like, iterators over results, and a pipeline of stages written beside the query. It never hides the SQL, and it isn't an ORM.

my @names = $dbh->iterate(
    'SELECT name FROM track WHERE plays > :plays ORDER BY plays DESC',
    plays => 6000,
    pluck,              # each row's name
    sub { uc },         # in capitals
)->all;

It's a subclass of the DBI, so everything the DBI does still works. Where the two do the same thing, they give the same result.

Why use it

The DBI is the right foundation, and it's low-level on purpose. Most code written around it does the same few chores, by hand, each time. DBIx::Squirrel does them for you.

  • Placeholders you can read.

    With ?, values go in the order the marks appear. A value used twice is passed twice, and changing the SQL can mean changing the code that calls it. Named placeholders bind by name, with each value given once. They work on any driver, whatever placeholders it supports itself, so SQL written with PostgreSQL's $1 runs unchanged on SQLite. Values that don't match the placeholders croak, rather than being bound in the wrong place.

  • Rows that come out ready to use.

    There's no fetch loop to write, and no array to push onto. The stages that map, filter and expand the rows sit beside the query. You can read what a result holds next to the SQL that makes it. Rows are fetched in batches, so you can work through a large result a row at a time without holding all of it in memory. Ready-made stages do the common jobs, such as taking the first few rows, picking out a column, or handing rows out in batches.

  • Queries with names.

    A statement and its stages can become a function with a name of its own. The SQL is written in one place, and the code that uses it reads as plainly as titles_by 2.

  • Nothing to unlearn, and nothing to undo.

    Your handles, attributes, drivers and existing code all still work, because it's still the DBI. Use as much as helps, and stop there. It isn't an ORM. You write the SQL, so everything your database can do is still there, and it never runs a query you didn't write.

One step at a time

You can take it up a step at a time, and stop at any step. Each works on its own, and code written for the DBI keeps working at every one.

1. Connect with it.

use DBIx::Squirrel;

my $dbh = DBIx::Squirrel->connect('dbi:SQLite:dbname=music.db', '', '');

Your DBI code runs as it did. connect and prepare croak if they fail, so there's no or die to write.

2. Write placeholders in the style you like.

my $sth = $dbh->prepare('SELECT title FROM album WHERE artist_id = :artist AND year > :year');
$sth->execute(artist => 2, year => 1994);

my $album = $dbh->selectrow_hashref('SELECT * FROM album WHERE id = $1', 2);

?, ?1, :1, $1, :name and $name are each rewritten as ? for the DBI. A numbered or named placeholder takes one value, wherever it appears. do and the select methods need no undef before the values.

my $of_years = $dbh->selectall_arrayref('SELECT * FROM album WHERE year IN (...:years)',
    years => [1967, 1994]);

A placeholder with ... before it takes a list. It's written as a ? for each value in the array reference given.

3. Iterate over the results.

my $albums = $dbh->iterate('SELECT title, year FROM album ORDER BY year');
while (my($row) = $albums->next) {
    say "$row->[0] ($row->[1])";
}

Rows are fetched in batches, and the statement is finished when they run out. all takes the rest at once, nearly as quickly as plain DBI. first takes the next one, and single the only one, warning if there's another.

4. Add stages.

my @hits = $dbh->iterate(
    'SELECT name, plays FROM track ORDER BY plays DESC',
    sub { $_->[1] >= 5000 ? $_ : () },    # leave out the rest
    sub { "$_->[0] ($_->[1])" },          # make each a string
)->all;

Each row goes through the stages, given as $_. A stage returns nothing to drop the row, one value in its place, or several. It can also stop the iteration early. A stage made with gather can hold rows, and give them out at the end, in batches or as one value. pipe runs anything else through stages the same way.

Ready-made stages come with it, for jobs that come up again and again. They're in four modules: take and drop in DBIx::Squirrel::Stages::Flow, with take_if, drop_if, take_while and more on request, pluck and as_hash in DBIx::Squirrel::Stages::Rows, batch and collect in DBIx::Squirrel::Stages::Batch, and peek and count_into in DBIx::Squirrel::Stages::Taps.

A sink, given last, says what the call gives in place of an iterator: keep_all every result, count_items how many, keep_first the first, and others. They're in Ekorn::Pipeline::Sinks. DBIx::Squirrel::Tooling exports every stage, sink and word from one module, by tags, as :stages and :sinks, or :all:

use DBIx::Squirrel::Tooling qw(:all);

my @top_ten = $dbh->iterate(
    'SELECT name, plays FROM track ORDER BY plays DESC',
    take_if { $_->[1] >= 5000 },    # leave out the rest
    pluck('name'),
    take(10),
    keep_all,
);

my $hits = $dbh->iterate('SELECT * FROM track WHERE plays > ?', 6000, count_items);

Where Ekorn::Pipeline::Operators is in scope, | chains them too, as in $it | count_items. The reference says how to write your own stages to work alongside them.

The pipeline is Ekorn::Pipeline. It began here, and is now a distribution of its own, so that any function can take stages from its caller. DBIx::Squirrel uses it as it is, and keeps its own names for its modules of functions, such as DBIx::Squirrel::Stop. The stages that need a database, such as pluck, are DBIx::Squirrel's own.

5. Have rows as objects.

my $album = $dbh->results('SELECT * FROM album WHERE id = :id', id => 2)->single;
say $album->title, ' (', $album->year, ')';

Each row has a method for each column, and stages can work on the objects.

6. Give your databases and statements names.

use DBIx::Squirrel::Tooling qw(pluck keep_all);
use DBIx::Squirrel::Sugar
    database  => {music => ['dbi:SQLite:dbname=music.db', '', '']},
    statement => {
        titles_by => [music => 'SELECT title FROM album WHERE artist_id = ?', pluck],
    };

my @titles = titles_by 2, keep_all;

DBIx::Squirrel::Sugar makes each a function with its own name. A statement's stages are declared beside its SQL. Called, it gives an iterator over the results, or, with a sink last, the sink's result. Given pipeline, it names pipelines too, which a statement can name among its stages.

Documentation

The reference describes every feature in detail, with examples. The user guide is a tour of the package, a layer at a time, from the DBI underneath to the shortcuts on top, with recipes that work. Each module's POD, starting with DBIx::Squirrel, says the same more briefly.

Installing

cpanm DBIx::Squirrel

It needs Perl 5.36 or later.

Upgrading

This is version 3. It builds on Ekorn::Pipeline, and makes one rule of what a call with stages gives: an iterator, unless a sink given last says otherwise. Most code written for 2.x runs as it did, and Upgrading from 2.x, in the reference, lists what changes.

Version 2 was a rewrite. Code written for 1.x needs changes to run, and Upgrading from 1.x lists them. To stay on 1.x, which also runs on older Perls, install its last version:

cpanm DBIx::Squirrel@1.6.5

Its code is in the archived DBIx-Squirrel-1 repository.

The repository was called DBIx-Squirrel-2 while version 2 was in development. It took the original name, DBIx-Squirrel, before its first release.

Licence

Copyright (c) 2024-2026 by Iain Campbell. This is free software; you can redistribute it and/or modify it under the same terms as the Perl 5 programming language system itself. See the file LICENSE.

Acknowledgements

Designed and created by Iain Campbell, in collaboration with Claude, Anthropic's AI model.