Name

SQL::Abstract - Generate SQL from Raku data structures

Synopsis

use SQL::Abstract;

my $abstract = SQL::Abstract.new(:placeholders<dbi>);
my $query = $abstract.select('table', <foo bar>, :id(3));
my $result = $dbh.query($result.sql, $result.arguments);

my $join = { :left<books>, :right<authors>, :using<author_id> };
my $result = $abstract.select($join, ['books.name', 'authors.name'], { 'cost' => { '<' => 10 }});

Description

SQL::Abstract abstracts the generation of SQL queries. Fundamentally its functionality has three components

An AST to represent SQL queries
A set of helpers to convert standard raku datatypes into such an AST
A renderer to turn that AST into an SQL query

It should be able to represent any SELECT, UPDATE, INSERT, or DELETE query that is valid in both Postgresql and SQLite. This subset should be generally portable to other databases as well.

Class SQL::Abstract

This is the main class of the module.

new(:$placeholders!, Bool :$quoting)

This creates a new SQL::Abstract object. It has one mandatory name argument, $placeholders, which takes one of the following values:

'dbi'/SQL::Abstract::Placeholders::DBI
This will use DBI style (?, ?) placeholders
'postgres'/SQL::Abstract::Placeholders::Postgres
This will use Postgres style ($1, $2) placeholders.

It also takes an optional argument :$quoting, if enabled table and column names will always be quoted.

select

method select(Source(Any) $source, Column::List(Any) $columns = *, Conditions(Any) $where?, Common(Any) :$common,
Distinction(Any) :$distinct, GroupBy(Any) :$group-by, Conditions(Any) :$having, Window::Clauses(Any) :$windows,
Compound(Pair) :$compound, OrderBy(Any) :$order-by, Int :$limit, Int :$offset, Locking(Any) :$locking)

This will generate a SELECT query. It will select $columns from $source, filtering by $conditions.

my $join = { :left<books>, :right<authors>, :using<author_id> };
my $result = $abstract.select($join, ['books.name', 'authors.name'], { :cost{ '<' => 10 } });
# SELECT books.name, authors.name FROM books INNER JOIN authors USING (author_id) WHERE cost < 10

my @columns = [ 'name', :sum{ :function<count>, :arguments(*) } ];
my $counts = $$abstract.select('artists', @columns, { :name(like => 'A%') }, :group-by<name>, :order-by(:sum<desc>));
# SELECT name, COUNT(*) as sum FROM artists WHERE name LIKE 'A%' GROUP BY name ORDER BY sum DESC

update

method update(Table(Any) $target, Assigns(Any) $assigns, Conditions(Any) $where?,
Common(Any) :$common, Source(Any) :$from, Column::List(Any) :$returning)

This will update $target by assigning the columns and values from $set if they match $where, returning $returning.

insert

Map insertion

method insert(Table(Any) $target, Assigns(Any) $values, Common(Any) :$common,
Overriding(Str) :$overriding, Conflicts(Any) :$conflicts, Column::List(Any) :$returning)

Inserts the values in $values into the table $target, returning the columns in $returning

$abstract.insert('artists', { :name<Metallica> }, :returning(*));
# INSERT INTO artists (name) VALUES ('Metallica') RETURNING *

List insertions

method insert(Table(Any) $target, Identifiers(Any) $columns, Rows(List) $rows, Common(Any) :$common,
Overriding(Str) :$overriding, Conflicts(Any) :$conflicts, Column::List(Any) :$returning)

Insert into $target, assigning each of the values in Rows to a new row in the table. This way one can insert a multitude of rows into a table.

$abstract.insert('artists', ['name'], [ ['Metallica'], ['Motörhead'] ], :returning(*));
# INSERT INTO artists (name) VALUES ('Metallica'), ('Motörhead') RETURNING *

$abstract.insert('artists', List, [ [ 'Metallica'], ], :returning<id>);
# INSERT INTO artists VALUES ('Metallica') RETURNING id

Select insertion

method insert(Table(Any) $target, Identifiers(Any) $columns, Select(Map) $select, Common(Any) :$common,
Overriding(Str) :$overriding, Conflicts(Any) :$conflicts, Column::List(Any) :$returning)

This selects from a (usually different) table, and inserts the values into the table.

$abstract.insert('artists', 'name', { :source<new_artists>, :columns<name> }, :returning(*));
# INSERT INTO artists (name) SELECT name FROM new_artists RETURNING *

delete

method delete(Table:D(Any:D) $target, Conditions(Any) $where, Common(Any) :$common,
Source(Any) :$using, Column::List(Any) :$returning)

This deletes rows from the database, optionally returning their values.

$abstract.delete('artists', { :name<Madonna> });
# DELETE FROM artists WHERE name = 'Madonna'

Helper types

SQL::Abstract uses various helper types that will generally coerce from basic datastructures:

SQL::Abstract::Identifier

This represents an identifier (typically a table name or column name, or an alias for such). It can be created from either:

Str
This will interpret a string (e.g. "foo" or "foo.bar") as an identifier.
List
This will interpret the elements of the list representing the components of the name. E.g. <bar baz> is equivalent to "bar.baz" .

SQL::Abstract::Identifiers

This takes either a list of Identifier(), or a single Identifier(). Note that a single list will be interpreted will be interpreted as a list of string identifiers, if one wants to pass a single list-from identifier the list must be nested (e.g. [ <table column>,]).

my SQL::Abstract::Identifiers() $identifiers = <name email website>;

SQL::Abstract::Source

A source is source of data, usually a table or a join. If not passed as a Source object it will upconvert from the following types:

Str
This represents the named table, e.g. "my_table".
List
This represents the named table, with the elements of the list representing the components of the table name. E.g. <bar baz> is equivalent to "bar.baz" .
Pair (Str => Identifier(Cool))
This will rename the table in the value to the name in the key.
Pair (Str => Select(Map))
This will use the result of a subquery as if it's a table.
Map
This will join two Sources, named left and right, it requires one of the following entries to join them on:
```
  * Join::Conditions() :$on

  * Identifiers() :$conditions

  * Bool :$natural

  * Bool :$cross
```
e.g. { :left<artist>, :right<album>, :using<album_id> } or { :left<artist>, :right<album>, :on{'artist.id' => 'album.artist_id'} }
The first three joiners take an optional :$type argument that can be any of "inner"/Join::Type::Inner, "left"/Join::Type::Left, "right"/Join::Type::Right or "full"/Join::Type::Full.

SQL::Abstract::Table does SQL::Abstract::Source

This role takes the same conversions as Source, but only the ones that represent a table. Unlike other sources, this can be used for mutating operations (update/insert/delete).

SQL::Abstract::Column::List

my Column::List() $columns = ('name', :number(:count(*)));
# name, COUNT(*) AS number;

This is a list of items representing a column. Each item can either be a: much like Identifiers, however it will accept not just identifiers but any expression (e.g. comparisons, function calls, etc…). If given a pair it will rename the value to the key (value AS key). A whatever-star will represent all columns.

SQL::Abstract::Conditions

my Conditions() $conditions = { :name(:like<%leon%>), :age(25..45), :country('nl'|'be'|lu') };
# name LIKE '%leon%' AND AGE BETWEEN 25 AND 45 AND country IN('nl', 'be', 'lu')

This is a pair, a list of pairs, a hash or an Expression. In the former three cases, the key (called left in the rest of this section) shall be an Identifier() designating a column name, or an Expression. The right hand side can be one of several types:

Expression

This will be used as-is

Any:U

This will check if the left expression is NULL; :left(Any) equals left IS NULL.

Pair

This will use the key as operator to compare left against another value or expression. E.g. :left('<' => 42) renders like left < 42 . The following keys are known:

=
!=
<>
<
<=
>
>=
like
not-like
distinct
not-distinct
||
->
->>
#>
#>>
*
/
%
+
-
&
|
<<
>>

Range

This will check if a value is in a certain range. E.g. :left(1..42) will render like left BETWEEN 1 AND 42.

Map

This will be interpreted as a conjunction of the hash pairs. E.g. :left{ '>' => 3, '<' => 42 } will render like left > 3 AND left < 42 .

Junction

This will check against the values in the function. E.g. :left(1|2|4) will render like left IN (1, 2, 4).

Capture

This will be used as a literal value. E.g. :left(\'NOW()') will render like left = NOW().

Any

If none of the above options match, the value will be compared to as is (as a placeholder). :left(2) will render equivalent to left = 2.

SQL::Abstract::Assigns

This takes a list of pairs, or a hash. The keys shall be a value or an expression. E.g. :name<author>, :id(SQL::Abstract::Values::Default), :timestamp(\'NOW()')

SQL::Abstract::OrderBy

This takes a list of things to sort by. Much like Column::List this accepts identifiers and expressions, but * isn't allowed and pair values are interpreted as order modifier (e.g. :column<desc>). A hash element will be expanded as well (e.g. { :column<column_name>, :order<desc>, :nulls<last> } )

SQL::Abstract::Common

This represents a common table expression. It converts from a pair or a list of pairs, with the keys being the name and the values being either a table name, a select hash or an SQL::Abstract::Query object.

my Common() $cte = recent => { :source<users>, :columns('name', :count(:count(*)), :group-by(name) };
# WITH recent AS (SELECT name, COUNT(*) AS count FROM users GROUP BY name);
)

SQL::Abstract::Locking

This takes one or more locking clauses. A locking clause is usually taken ... strings: 'update', C'<no key update'>, 'share', 'key share', but it can also take a pair of stregth

SQL::Abstract::GroupBy

This takes a list of grouping elements. Usually these are just columns, but they may also be arbitrary expressions (inclusing lists of columns). A pair is taken as a function call with the key as function name and the value as arguments.

SQL::Abstract::Conflicts

This represents one or more upsert clause. It can either be the string 'nothing', or a pair with the columns as keys and an Assigns(Map).

my SQL::Abstract::Conflicts = <name organization> => { :$email };
# ON CONFLICT (name, organization) DO UPDATE email = ?

SQL::Abstract::Distinction

This takes Truefor a distinct row, or a Column::List for specific rows that have to be distinct.

Window::Definition

Window definiton converts from a map taking the following keys, all optional:

Identifier(Cool) :$existing
This takes the name of an existing window
Column::List(Any) :$partition-by
This lists expressions to partition the rows by, somewhat similar to a GROUP BY.
OrderBy(Any) :$order-by
The order within a frame.
Boundary(Any) :$from
This argument defines the starting boundary of the frame. This can be any of:
```
* 'preceding'

* :preceding($amount)

* 'current'

* :following($amount)
```

It defaults to 'preceding'.

Boundary(Any) :$to
This optional argument defines the ending boundary of the frame, This can be any of:
```
* :preceding($amount)

* 'current'

* :following($amount)

* 'following'
```
Mode:D(Str) :$mode = Mode::Range
The mode of the frame takes one of 'rows', 'range' or 'groups', defaulting to 'range'.
Exclusion(Str) :$exclude
The exclusion of the frame, it takes one of 'current row', 'group', 'ties' or 'no others' (the default).

my Window::Definition $d = { :partition-by<foo bar>, :from<current> :to(:following(5)), :exclude<ties> }
# PARTITION BY foo, bar RANGE BETWEEN CURRENT ROW AND 5 FOLLOWING EXCLUDE TIES

Window::Clauses

This takes one or more pairs, with the names being windows names and the values taking window definition maps.

my Windows::Clauses $clauses =
    over5 => { :frame{ :preceding(5) } },
    foo => { :partition-by<foo bar>, :mode<range>, :from<current> :to(:following(5)), :exclude<ties> };
# WINDOW
#   over5 AS (ROWS 5 PRECEDING),
#   foo as (PARTITION BY foo, bar RANGE BETWEEN CURRENT ROW AND 5 FOLLOWING EXCLUDE TIES)

Author

Leon Timmermans fawaka@gmail.com

Copyright and License

This library is free software; you can redistribute it and/or modify it under the Artistic License 2.0.