django based mini framework inspired from sinatra. fully compatible with django.
Python HTML
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Failed to load latest commit information.


Issues you can help with right now: 'Stories in Ready'

Build Status: build-status Code Health

What is importd?

Slides of a talk I gave about importd:

Django is awesome, but starting a new project in it is a pain. importd is inspired from ruby's sinatra. Hello world django project:

from importd import d

def idx(request):
    return "index.html"

def post(request, post_id):
    return "post.html", {"post_id": post_id}

if __name__ == "__main__":

To run it:

$ python

This will start the debug server.

To run it in production:

$ gunicorn foo:d


Settings Framework

Managing settings in django is done via a file. Then people put a to override. This does not scale too well, we end up having very big settings file with almost no structure, and there are many issues because of lack of synchronization of among developer's machines.

importd has some methods to hopefully make this simpler and more standardized.

First of all there is no Setting customization are of two kinds, picking different things for development and prod, eg you want to activate statsd for prod, but debug_toolbar for development. Both these should be checked in so there is no scope of people not getting some setting accidentally. Then there are setting customization for not storing some things in version control system, say passwords and access tokens and keys. These should be managed via environment variable.

And then there is also a concern of exposing settings to template. We have a template context processor, which can expost whole settings to templates, but that is uncomfortable to many. You may want to expose only a small subset of things you describe in settings, and you want to do this with minimal fuss.


With that in mind importd has env(), which simply reads data from enironment. So in your you can do:

from importd import d, env
    DEBUG=not env("IS_PROD", False),
    db=env("DB_URL", "mysql://root:@localhost/dbname")

It is highly recommended you include envdir in your project. May be someday importd will auto detect envdir and set it up.

env is pretty smart, it takes default= and factory=. If default is passed, the string value of environment variable is converted to the type() of default. You can overwrite this behaviour by passing your own factory, or you can disable this behaviour altogether by passing factory=importd.NotSet.

env() also treats booleans by converting strings like "False/off/no" (case insensitive) to python's False value (and non empty string to True as bool() does).


With .debug() you can set some setting to have different values based on DEBUG.

from importd import d, debug
    DEBUG=not env("IS_PROD", False),
        'django_statsd.clients.toolbar', prod='django_statsd.clients.normal'

This will set STATSD_CLIENT to appropriate value based on if we are in debug mode or not. This is as simple as putting an if condition, but it gets repeated so many times, its worth using this shortcut. Also this way things stay in same place, you do not look for up and down the settings file, and in to see if the variable has been overwritten.


This lets you "expose" a setting for access in templates. You should not use "django.core.context_processors.settings" as a TEMPLATE_CONTEXT_PROCESSORS, instead use "importd.esettings" context preprocessor, and in templates you will have access to esettings variable.

To mark a variable as exposed you have to do this:

from importd import d, e

    SOME_VAR=e("its value"),

This will make SOME_VAR available in settings as well as in esettings.

importd.s` parameter

This lets you re-use settings variables. In settings file we define variables and reuse them when needed. In importd you can reuse defined settings variables.

This will set TEMPLATE_DEBUG settings variable to DEBUG value. s will raise ImproperlyConfiguredError exception if you will try to use it inside of key value.

Above example will raise ImproperlyConfiguredError.

d(debug={}) parameter

Some settings are only needed in debug environment, or need to be overwritten, you can use the debug= keyword argument to set things up.

from importd import d

    SOME_VAR="this is prod value",
        SOME_VAR="this is debug value"

You can also use importd.NotSet as a value in debug dict, and the setting will be removed altogether in the approprite environment (debug or prod).


Above method will open envdir directory in current directory and will load all environment variable inside this directory. If path is realpath i.e. full path then importd will try to look into specified path. If relative path specified into path then importd will look relative to current directory.

It is recommended to call it just after importing d.

debug:/prod: prefix for INSTALLED_APPS etc

It is a common pattern that some apps are only needed in debug environment, say devserver, or debug_toolbar. And since order of apps in INSTALLED_APPS, and middelware etc is important, we end up copying the whole INSTALLED_APPS, MIDDLEWARE_CLASSES etc for prod and dev, and this then tend to diverge since they are in different locations. Not good.

from importd import d, env
    DEBUG=env("IS_PROD", True),



Notice the debug: prefix in devserver and debug_toolbar. Depending on the value of DEBUG, these lines would be included or not. importd looks for strings in MIDDLEWARE_CLASSES, INSTALLED_APPS and TEMPLATE_CONTEXT_PROCESSORS.

Similarly if something starts with prod:, it is only included in production environment.

Backward Incompatibile Change

d.main() used to be not required, now it is.


  • fully compatible with django
  • supports smarturls
  • most of regularly used django functions and classes available in d. namespace, eg d.HttpResponse, d.render_to_response, d.get_object_or_404 etc
  • automatically maps "templates" folder in directory to serve templates
  • automatically maps "static" folder in to serve static content
  • management commands still available: $ python shell
  • wsgi compliant
  • gunicorn support
  • works seamlessly with fhurl (
  • Auto Add django-debug-toolbar (Needs to add it manually to INSTALLED_APPS)
  • Auto SECRET_KEY: If no SECRET_KEY on settings, try to read SECRET_KEY from ./secret.txt , if no ./secret.txt generate a random string then write it to ./secret.txt and finally return it as SECRET_KEY.
  • Auto Add coffin/django-jinja (jinja2 integration)
  • Support for livereload


$ pip install importd




Contribution Guide

To view this file, or any restructuredtext file locally before comitting on git, install restview from pypi.

Pull Requests: If you fork this repository to send pull request, please create a branch for your work instead of working directly on master. This way your master will track my master, and in case the pull request is rejected, or delayed, your master stays clean. This also makes easy to send more than one pull requests from your fork.


  • BSD