getconf 1.4.1

getconf, a versatile configuration lib for Python projects

The getconf project provides simple configuration helpers for Python programs.

It provides a simple API to read from various configuration files and environment variables:

import getconf
config = getconf.ConfigGetter('myproj', ['/etc/myproj.conf'])
db_host = config.get('db.host', 'localhost')
db_port = config.getint('db.port', 5432)

Beyond this API, getconf aims at unifying configuration setup across development and production systems, respecting the standard procedures in each system:

• Allow userspace configuration on development systems

• Allow multiple different configurations for continuous integration systems

• Use standard configuration space in /etc on traditional production servers

• Handle environment-based configuration for cloud-based platforms

getconf supports Python 2.6, 2.7, 3.3, 3.4 and is distributed under the two-clause BSD license.

Links

• Package on PyPI: http://pypi.python.org/pypi/getconf/

• Doc on ReadTheDocs: http://readthedocs.org/docs/getconf/

• Source on GitHub: http://github.com/Polyconseil/getconf/

Installation

Intall the package from PyPI, using pip:

pip install getconf

Or from GitHub:

git clone git://github.com/Polyconseil/getconf

getconf has no external dependancy beyond Python.

Introduction

Note

Please refer to the full doc for reference and advanced usage.

All configuration values are accessed through the getconf.ConfigGetter object:

import getconf
config = getconf.ConfigGetter('myproj', ['/etc/myproj/settings.ini', './local_settings.ini'])

The above line declares:

• Use the myproj namespace (explained later; this is mostly used for environment-based configuration, as a prefix for environment variables)

• Look, in turn, at /etc/myproj/settings.ini (for production) and ./local_settings.ini (for development); the latter overriding the former.

Once the getconf.ConfigGetter has been configured, it can be used to retrieve settings:

debug = config.getbool('debug', False)
db_host = config.get('db.host', 'localhost')
db_port = config.getint('db.port', 5432)
allowed_hosts = config.getlist('django.allowed_hosts', ['*'])

All settings have a type (default is text), and accept a default value. They use namespaces (think ‘sections’) for easier reading.

With the above setup, getconf will try to provide db.host by inspecting the following options in order (it stops at the first defined value):

• From the environment variable MYPROJ_DB_HOST, if defined

• From the host key in the [db] section of ./local_settings.ini

• From the host key in the [db] section of /etc/myproj/settings.ini

• From the default provided value, 'localhost'

Features

Env-based configuration files

An extra configuration file/directory/glob can be provided through MYPROJ_CONFIG; it takes precedence over other files

Default options

An extra dictionary can be provided as ConfigGetter(defaults=some_dict); it is used after configuration files and environment variables.

It should be a dict mapping a section name to a dict of key => value:

>>> config = ConfigGetter('myproj', defaults={'db': {'host': 'localhost'}})
>>> config.get('db.host')
'localhost'

Typed getters

getconf can convert options into a few standard types:

config.getbool('db.enabled', False)
config.getint('db.port', 5432)
config.getlist('db.tables')  # Expects a comma-separated list
config.getfloat('db.auto_vacuum_scale_factor', 0.2)

Concepts

getconf relies on a few key concepts:

namespace

Each ConfigGetter works within a specific namespace (its first argument).

Its goal is to avoid mistakes while reading the environment: with ConfigGetter(namespace='myproj'), only environment variables beginning with MYPROJ_ will be read

Sections

The configuration options for a project often grow quite a lot; to restrict complexity, getconf splits values into sections, similar to Python’s configparser module.

Section are handled differently depending on the actual configuration source:

• section.key is mapped to MYPROJ_SECTION_KEY for environment variables

• section.key is mapped to [section] key = in configuration files

• section.key is mapped to defaults['section']['key'] in the defaults dict.

Default section

Some settings are actually “globals” for a projet. This is handled by unset section names:

• key is mapped to MYPROJ_KEY for environment variables

• key is mapped to [DEFAULT] key = in configuration files

• key is mapped to defaults['DEFAULT']['key'] in the defaults dict.