djangobot

Bridge between Slack and Channels-based Django apps


Keywords
channels, django, slack, twisted
License
MIT
Install
pip install djangobot==0.1

Documentation

Djangobot

Djangobot is a bridge between Slack and a Channels-enabled Django app.

Specifically, it is a protocol server that produces and consumes messages for channels-based apps.

It is built atop autobahn and twisted.

Installation & Usage

To install, simply use pip.

$ pip install djangobot

Djangobot should be installed in the virtual environment of your application as it needs to be able to import one of the specified channel layers defined in your settings.py.

Then, assuming your django project is named myapp, the ASGI file is named asgi.py and you've created a channel layer in it named channel_layer, run this command:

$ DJANGOBOT_TOKEN=[your slack token] djangobot myapp.asgi:channel_layer

In production, you'll want to keep this process alive with supervisor, circus or a similar tool.

What's it doing?

When beginning djangobot, it will:

  1. Connect to the Slack API and request users and channels for your team.
  2. Initiate a Real-Time Messaging Connection.
  3. Forward any RTM events onto the slack.{type} channel. For example, message events (whose type is message) are sent along the slack.message channel.
  4. Send any messages on the slack.send channel back to slack.

Receiving events in your application

In your routing.py, you'll need to specify consumer functions to handle events on the slack channels like so:

channel_routing = {
    'websocket.receive': 'path.to.my.consumer',
    'websocket.connect': 'path.to.another.consumer',
    'websocket.disconnect': 'path.to.yet.another.consumer',

    # Slack channels
    'slack.message': 'function.to.handle.slack.messages',
    'slack.hello': 'handle.when.djangobot.connects',
    # and so forth

Sending messages to slack

To send messages to slack, simply send a dictionary at least a text key. You may optionally include the channel on which to post the message. This can be the human-readable version or a channel id. Note that djangobot necessarily posts messages as the user tied to your Slack API token.

For example:

import channels

channels.Channel('slack.send').send({'text': 'Why hello there!', 'channel': 'general'})

Of course, part of the beauty of channels is that this can be done from anywhere.

Why is this useful?

This simply bridges your slack team to your production application in real-time. On it's own, it does nothing else. Implementing actual features is up to you. Off the top of my head, some ideas:

  1. Make Slack a logging destination.
  2. 2FA to approve certain tasks.
  3. Chat through Slack to users.

Contributing

  1. Fork this repository.
  2. Create a branch with your feature or bug fix.
  3. Work on it, push commits.
  4. Submit a Pull Request.

Todo

  1. Testing: I would appreciate help testing twisted clients.
  2. Setting up the reply channel: Right now both djangobot and applications must hard-code the slack.send outgoing channel name which isn't ideal.
  3. Logging: Djangobot could logger.debug a lot more.