TempDB

Spawn temporary PostgreSQL clusters for use in unit tests. This is alpha quality software and the API may change between version.

Usage

The following example shows how to create a temporary database with pytest.

import psycopg2
import pytest

from contextlib import closing
from tempdb import find_postgres_bin_dir, PostgresFactory


@pytest.fixture
def conn():
    # Try to discover a directory where PostgreSQL is installed. If you
    # have installed it in a non-standard location you must replace
    # pg_bin_dir with the path to the binary directory for your
    # installation.
    #
    # It is possible to provide a specific version as an argument if there
    # are multiple versions to choose from. The highest available version
    # is chosen by default.
    pg_bin_dir = find_postgres_bin_dir()
    if pg_bin_dir is None:
        pytest.skip("Unable to locate a PostgreSQL installation")

    # The factory is bound to a particular PostgreSQL version and can be
    # used to create an arbitrary number of clusters
    factory = PostgresFactory(pg_bin_dir)

    # A cluster is an actual running instance of PostgreSQL. By default a
    # cluster doesn't have any databases apart from the defaults. It is
    # possible to create new databases using the method
    # .create_database(name, template=None)
    with closing(factory.create_temporary_cluster()) as cluster:
        tmp_db = cluster.create_database("tmp")
        yield psycopg2.connect(tmp_db.dsn)


def test_create_tables(conn):
    with conn.cursor() as c:
        c.execute("""
            CREATE TABLE test(
                id SERIAL,
                name VARCHAR
            )
        """)
        c.execute("INSERT INTO test(name) VALUES (%s), (%s)", [
            "Abel",
            "Cain",
        ])
        c.execute("SELECT name FROM test ORDER BY id")

        assert [name for name, in c.fetchall()] == [
            "Abel",
            "Cain",
        ]

Changelog

Version 0.1.0

Released on 3rd June, 2019

• Initial release