Skip to main content

Optimized app config (+Django) from environment vars and/or files

Project description

django-confenv purpose - make django (or any other Python project config) persistent, while real setup is done from file(s) with simple “env=value” format or OS environment variables. Also, this technics is known as Twelve-factor methodology or 12factor method.

NOTE, that, despite name, you can use django-confenv in ANY Python project, without activating any django-related stuff.

django-confenv is heavily reconstructed version of django-environ by Daniele Faraglia and other authors . Differences from that code are:

  • higher speed (sometimes much higher)
  • smaller code
  • smaller footprint
  • django-related stuff moved to module
  • more functionality in .env file(s) search/selection
  • some improvements/speedups in ‘api’
  • separated basic tests for base and django functionality

django-confenv is (C) 2019-2021 Vitaly Protsko <> Released under GPLv3 . If you need this code under other license, please, contact the author.

Quick usage

Get Env

# base functionality class
from confenv import Env

for Django projects

# Env with django species
from confenv.django import Env

Next, “configure” Env itself

# Set exception to be raised in case of error
# For Django it is set to ImproperlyConfigured
#Env.exception = Exception

# Get default configuration file name from named
#   environemt variable
Env.defconf_key = 'BIGPROJECT_CONFNAME'

# Set env file name to be read, instead of name,
#   defined in env var named in Env.defconf_key
# filename + '.env' will also be checked, if
#   file with plain filename not found
# File will be searched in directory where calling
#   Env class instantiation program resides if
#   Env.filepath is None
# Default: '.env'
Env.filename = 'default.conf'

# Search for file Env.filename or file '.env' in this
#   path, if set.
# If name in Env.filename is absolute, filepath will
#   not be prepended (will be ignored).
# If this file path is not absolute the directory where
#   calling Env class instantiation program resides will
#   be prepended.
# Default: None
Env.filepath = '..'

# Set dict to operate on
# Default: os.environ = dict()

Collect all changeable values in file, parseable also by shell:

# default.conf

# put settings in environment variables

# you can use simple variable substitution. Remember, that
#   this interpreter is not a shell. This works only if you
#   start value from '$<varname>'. In that case variable
#   will be substituted by value of the variable, found in
#   previous asignments or in dictionary. Resolving
#   recursion depth is not limited.
You can define
type and default value for parameters you use in you project by defining a “scheme” directly as parameters, or as a dictionary. For example, to define variable DEBUG as boolean parameter with default ‘False’ value, you can use either form:
# directly in parameters
env = Env(DEBUG=(bool, False), CACHEDIR=(str, ))

# or as a prepared list of kwargs:
kwl = {
  'DEBUG': (bool, False),
  'CACHEDIR': (str, ),

env = Env(**kwl)

Full list of supported types you can find later in this file.

# now use it in your program

#   If you dont pass the default value to converter
#   or miss it in Env class's parameters ("schema"),
#   Env will require the variable to exist in OS
#   environment (precisely, in dict),
#   otherwise Env.exception will be thrown.

# this will assign True to variable debug, if
#   environment variable DEBUG value in
#   ('true', 'on', 'ok', 'y', 'yes', 'yea', '1')
#   Other value or variable absence will assign
#   False value.
DEBUG = env('DEBUG')

# you can directly point Env to conversion you
#   need to be done with value of env var by
#   calling corresponding method directly
SERVER = env.url('SERVER', '')

# this will assign instance of ParseResult from
#   urllib.parse with pre-parsed URL, for the default
#   value in example above it will be
#   ParseResult(scheme='http', netloc='', path='', params='', query='', fragment='')
#   NB: all other conversions return expected type,
#     not an instance of side class

# use substituted var, but back it with default value
CACHE = env('CACHEDIR', default='/var/cache')

# complex json is also not a problem
PARAMS = env.json('PARAMETERS', "{'par1':'val1', 'par2': {'def': 1, 'set': 2}, 'par3': [1, 2]}")

# this will assign default dictionary to
#   variable PARAMS if PARAMETERS is absent in
#   env file or in OS environment variables

Supported types

Here is all supported data types collected in example .env file and code to use it in your programs.

# myappconf.env

# NB: Quotes for us are optional, value counted
#   from character after the equal sign, except
#   when surrounded by quotes, in which case
#   they are stripped.
#   There is limitation: variable assignments
#   must be written in one line. Continuations
#   are not supported.

# bool variable, accepting values
#   true on ok y yes yea 1
#   as True, any other value as False.
#   You can use "export" keyword before variable
#   name and maintain one place of configuration
#   for application and accompanying shell scripts
export DEBUG=no

#   lines with "unset" keyword are silently ignored

# str variable
#   Any sequence of characters. This is effectively
#   the same, as unicode type, look below for
#   declaration example(s)
SERVERNAME="Our service server"

# bytes variable
#   Sequence of any characters, that can be
#   decoded by .encode('utf8') method. Or you can
#   pass encoding directly to convertor method
#   with parameter: env.bytes('VAR', encoding='utf16')
WELCOMEMSG=Добро пожаловать!

# integer variable

# float variable

# list variable

# tuple variable
#   value type in this case will be str

# dict variable #1
#   In this case all values are strings and
#   here is no need to declare "schema" for
#   this variant.

# dict variable #2
#   This dictionary variant needs declaration
#   to properly convert values, it can be done
#   in Env class instantiation parameters.

# json variable
#   This type can be used for complex setup
#   of something (like menu) or for any other
#   kind of structured (initialization?) data.
MENUEXTRA={"ExtraItem1": {"display": "&Cook Coffee", "handler": "cooker", "allow": "ALL"}}

# url variable
#   This can be used for pointing to any kind
#   of resources, allowed schemes are as in
#   urllib.
There is additional convertors for django applications.
They pesent only in Env, imported from confenv.django module.
# djangosite.env

# Applications list
#   We can add applications to django's standard
#   list in this way.

# Database URL
#   This variable value can be parsed as database
#   configuration for django project. Env will
#   automatically select appropriate django driver
#   for database type pointed by url scheme.
#   Recognized schemes are:
#     postgres, postgresql, pgsql, postgis, mysql,
#     mysql2, mysql-connector, mysqlgis, mssql, oracle,
#     pyodbc, redshift, spatialite, sqlite, ldap

# Cache URL
#   You can point django to cache resource(-s) as url
#   Recognized schemes are:
#     dbcache, dummycache, filecache, locmemcache,
#     memcache, pymemcache, rediscache, redis

# E-Mail URL
#   Django's e-mail submitting parameters
#   Recognized schemes are:
#     smtp, smtps, smtp+tls, smtp+ssl, consolemail,
#     filemail, memorymail, dummymail

# Search URL
#   This otional feature uses drivers from django-haystack
#   to find that needle.
#   Recognized schemes are:
#     elasticsearch, elasticsearch2, solr, whoosh,
#     xapian, simple
You can use all power of Env without any types declarations.
Env instance has methods for direct variable conversion. But, I’m shure, you want stricter value type definitions, that gives more chances in bug hunting. Also, you always can use direct call to instance to get plain value backed by default.
NB: Here is nuance with naming variable type in declarations

and in call to convertors. Types from this list:

str, bool, int, float, list, tuple, dict

you can point directly, as they are build-in and this identifiers are known for interpreter. Types

url, json, unicode, bytes

you can use only by quoting their names, as shown below. Althrough, you can quote names of all types, including ones from first list, if you do not remember well which ones are built-in. ;-)

from confenv import Env
Env.filename = 'myappconf'

kwl = {
  # if you dont give default value, you can get an exception
  #   from Env in case if variable not present in
  'DEBUG': (bool, False),

  # you can completely omit declaration of variables
  #   with str value type - it is default
  #'SERVERNAME': (str, ),
  #   or, which is equivalent, you can declare it as
  'SERVERNAME': ('unicode', ),

  # we'll comment out this declaration to demonstrate direct
  #   convertor calls later
  #'WELCOMEMSG': ('bytes', 'Welcome !'),

  'PORT': (int, ),
  'MAXAVGLOAD': (float, ),

  # quoted type name also works
  'LISTENIP': ('list', ),

  # items for this tuple are of type str
  'ENDPOINTS': (tuple, ),

  # default type of values str
  'USERACL': (dict, {'ALL': 'deny'} ),

  # default value type is str
  'COEFFICIENT': ({'value': str, 'cast': {'a': float, 'b': int} }, {'result': 'NaN'}),

  'MENUEXTRA': ('json', ),
  'EXTLOGO': ('url', ),

env = Env(**kwl)

# store values from environment in program configuration items

flagDebug = env('DEBUG')
# assigned value bool False

textServer = env('SERVERNAME', 'Default service')
# assigned value str 'Our service server'

# in-line conversion
textWelcome = env.bytes('WELCOMEMSG', 'Welcome !')
# assigned value str 'Добро пожаловать!'

paramPort = env('PORT', default=4321)
# assigned value int 1234

paramNextPort = env.PORT + 1
# yes, all parameters are accessible as object attribute

paramLoad = env('MAXAVGLOAD', default=10.0)
# assigned value float 5.5

paramListen = env('LISTENIP', default=[ '' ])
# assigned value [ '', '', '' ]

progServices = env('ENDPOINTS', default=('start', 'stop'))
# assigned value tuple('start', 'read', 'calculate', 'write', 'stop')

paramACL = env('USERACL')
# assigned value {'root': 'admin', 'jdoe': 'operator', 'john': 'user'}

paramMUL = env('COEFFICIENT')
# assigned value { 'a': 10.11, 'b': 5, 'result': 'unknown' }

menuExtra = env('MENUEXTRA')
# assigned value { 'ExtraItem1': { 'display': '&Cook Coffee', 'handler': 'cooker', 'allow': 'ALL' } }

urlLogo = env('EXTLOGO')
# assigned value ParseResult(scheme='http', netloc='', path='/biglogo.jpg', params='', query='', fragment='')

config = {}
# you can use items() keys() and values() to access raw content of

for k, v in env:
  print('{}={}'.format(k, v))
# access to all 'raw' values

For django project Env can generate complete configuration dictionaries for database, cache, e-mail and search functions to include it into standard django config structures.

# This file can be "static", all changeable information
# resides in ../myappconf.env and ../djangosite.env

# get django-specific Env
from confenv.django import Env

# set path to .env's to directory, where reside
Env.filepath = '..'

# expecting, that you place example .env files 'myappconf'
# and 'djangosite' in appropriate dir
Env.filename = 'myappconf'

# django Env's version have place to hold variable names
# to read database, cache, e-mail and search URLs default
# values from
# They are class members with this default values:
#Env.defenv_db = 'DATABASE_URL'
#Env.defenv_cache = 'CACHE_URL'
#Env.defenv_email = 'EMAIL_URL'
#Env.defenv_search = 'SEARCH_URL'

# search for database config here by default
Env.defenv_db = 'MAINDB_URL'

env = Env()

# load additional configuration

DEBUG = env.bool('DEBUG', True)

# use "untyped" variable as list to extend
#   django's default
  # ... standard django's list ...
] + env.list('APPADD', [])

# ...

# databases config
  # config will be read from MAINDB_URL
  'default': env.db_url(),

  # this cache config will be read from CACHE_URL
  'default': env.cache_url(),

  # second cache config
  'quick': env.cache('MEMCACHE_URL'),

# This will require MAINMAIL key to be existent
#   in dict
EMAIL_CONFIG = env.email_url('MAINMAIL')

# django's search extension
  'default': env.search_url('SEARCHENGINE', 'simple://'),


This package can be installed from standard Python packages source

pip install django-confenv


This code is (c) Vitaly Protsko <>, under GPLv3 .

This work is based on django-environ (c) Daniele Faraglia which includes work from other authors .

Project details

Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Files for django-confenv, version 1.1.4
Filename, size File type Python version Upload date Hashes
Filename, size django_confenv-1.1.4-py3-none-any.whl (22.0 kB) File type Wheel Python version py3 Upload date Hashes View
Filename, size django-confenv-1.1.4.tar.gz (17.6 kB) File type Source Python version None Upload date Hashes View

Supported by

AWS AWS Cloud computing Datadog Datadog Monitoring Facebook / Instagram Facebook / Instagram PSF Sponsor Fastly Fastly CDN Google Google Object Storage and Download Analytics Huawei Huawei PSF Sponsor Microsoft Microsoft PSF Sponsor NVIDIA NVIDIA PSF Sponsor Pingdom Pingdom Monitoring Salesforce Salesforce PSF Sponsor Sentry Sentry Error logging StatusPage StatusPage Status page