Sqlize-pg is a SQL query builder for Python. It's main goals are:
- speed: because fast is good
- transparency: do not hide the true nature of SQL
- mutability: we should be able to mutate the query
This library is primarily developed for use with Postgresql and no efforts have been invested into testing or using with other database backends.
Sqlize-pg can be installed using pip
or easy_install
as usual:
pip install sqlize-pg
This section will provide a brief introduction to sqlize-pg. The examples are all doctested, so rest assured that they work as expected.
The basic concept is to instantiate an object representing some type of query, optionally manipulate attributes on it to fine-tune the clauses, and finally convert the query into SQL string by coercing it into string.
Note that the queries are meant to be used with placeholder values, and no
quoting is performed by sqlize-pg. The generated SQL strings are intended to be
used with psycopg2.extensions.cursor.execute()
, and similar methods.
A basic select looks like this:
>>> import sqlize_pg as sql >>> q = sql.Select('*', sets='foo')
Note that we call tables 'sets' to avoid the clash with Python's from
keyword.
To convert the query to SQL, we simply coerce it into a str
:
>>> str(q) 'SELECT * FROM foo;'
You can select multiple things:
>>> str(sql.Select(['foo', 'bar'], sets='foo')) 'SELECT foo, bar FROM foo;'
You can also select from mutliple tables:
>>> str(sql.Select('*', sets=['foo', 'bar'])) 'SELECT * FROM foo , bar;'
If you want to restrict your select, all common clauses are available:
>>> str(sql.Select('*', ['foo', 'bar'], where='a = %s', group='foo', ... order='-bar', limit=10, offset=20)) 'SELECT * FROM foo , bar WHERE a = %s GROUP BY foo ORDER BY bar DESC LIMIT 10 OFFSET 20;'
So far it looks like a rather complicated way of writing SQL. The real power, though, comes from the fact that every aspect of the query object can be tweaked.:
>>> q = sql.Select() >>> str(q) 'SELECT *;' >>> q.what = 'foo' >>> q.sets = 'this' >>> q.sets.join('other', sql.INNER) <sqlize_pg.builder.From object at ...> >>> q.where = 'bar = %s' >>> q.limit = 2 >>> str(q) 'SELECT foo FROM this INNER JOIN other WHERE bar = %s LIMIT 2;'
Now let's take a look at individual clauses.
The where
attribute is represented by a sqlize_pg.builder.Where
object,
which supports a few handy operators for adding conditions:
>>> q = sql.Select() >>> q.where = 'foo = %s' >>> q.where &= 'bar = %s' >>> q.where |= 'foo = bar' >>> str(q) 'SELECT * WHERE foo = %s AND bar = %s OR foo = bar;'
The &=
and |=
have method aliases. Main advantage is that methods are
chainable. The above example can be rewritten as:
>>> q = sql.Select() >>> q.where = 'foo = %s' >>> q.where.and_('bar = %s').or_('foo = bar') <sqlize_pg.builder.Where object at ...> >>> str(q) 'SELECT * WHERE foo = %s AND bar = %s OR foo = bar;'
Note the underscore. We can't use method names that look like built-in operators.
The sets
attribute is represented by a sqlize_pg.builder.From
object. It
has a few utility methods which you can use to add and join other tables:
>>> q = sql.Select() >>> q.sets = 'foo' >>> q.sets.append('bar') <sqlize_pg.builder.From object at ...> >>> str(q) 'SELECT * FROM foo , bar;' >>> q = sql.Select() >>> q.sets = 'foo' >>> q.sets.join('bar', sql.NATURAL) <sqlize_pg.builder.From object at ...> >>> str(q) 'SELECT * FROM foo NATURAL JOIN bar;'
There is no direct support for aggregates. Instead, you write raw SQL.:
>>> q = sql.Select('COUNT(*) as count', sets='foo', group='bar') >>> str(q) 'SELECT COUNT(*) as count FROM foo GROUP BY bar;'
This is intentional. We wanted sqlize-pg to be as true to SQL as possible, and not get in your way.
Apart from selecting, sqlize-pg supports inserts, updates, deletion, and replacement.
Inserts look like this:
>>> q = sql.Insert('foo', '%s, %s, %s') >>> str(q) 'INSERT INTO foo VALUES (%s, %s, %s);'
You can also specify columns:
>>> q = sql.Insert('foo', '%s, %s, %s', ('foo', 'bar', 'baz')) >>> str(q) 'INSERT INTO foo (foo, bar, baz) VALUES (%s, %s, %s);'
If you omit the values, the query will contain named placeholders:
>>> q = sql.Insert('foo', cols=('foo', 'bar', 'baz')) >>> str(q) 'INSERT INTO foo (foo, bar, baz) VALUES (:foo, :bar, :baz);'
Replacing is similar to inserting, but uses Replace
class instead:
>>> q = sql.Replace('foo', constraints=['id'], cols=['id', 'name']) >>> str(q) 'INSERT INTO foo (id, name) VALUES (%(id)s, %(name)s) ON CONFLICT (id) DO UPDATE SET id = %(id)s, name = %(name)s;'
The update query looks like this:
>>> q = sql.Update('foo', 'bar = %s', baz='%s') >>> str(q) 'UPDATE foo SET baz = %s WHERE bar = %s;'
The second argument is the same as where
in Select()
. It can be
modified after initialization:
>>> q = sql.Update('foo', baz='%s') >>> q.where &= 'foo = %s' >>> q.where |= 'bar = %s' >>> str(q) 'UPDATE foo SET baz = %s WHERE foo = %s OR bar = %s;'
Any keyword arguments passed to Update()
will be converted to SET
clauses.
Deleting rows can be accomplished using the Delete()
class.:
>>> q = sql.Delete('foo', 'bar = %s') >>> str(q) 'DELETE FROM foo WHERE bar = %s;'
As with Update()
, the second argument is a where
clause, and can be
manipulated.
Unfortunately, there are currently no docs apart from this introduction. I hope that codebase is not too difficult to follow, though, so if you can't wait, you can peek into the source files.
TODO
Report all bugs and feature requests to our `issue tracker`_.
_issue tracker: https://github.com/Outernet-Project/sqlize-pg/issues