A lightweight django framework for bots
Project description
django-bot
A django library that makes it easier to develop bots with a common interface for messaging platforms (eg. Slack, FB messenger) and natural langauge parsers (eg. api.ai).
Overview
BETA version: Currently django-bot only supports Slack and api.ai. Future plans include supporting more messaging platforms (Facebook Messenger, Telegram, Kik, Google assistant, Cortana, Skype, Alexa), and more natural langauge parsers (AWS Lex, wit.ai).
This library helps to maintain authenticated users and groups in the database and allows you to respond to any messages as well as initiate conversations with any of those.
Requirements and Installation
django-bot for Python works with Python 2.7, 3.4, 3.5, 3.6 and django >= 1.8, and requires PyPI to install dependencies. The message parsing and delivery is done in the background with the help of celery. It also requires slackclient and apiai python libraries for communication with the external services.
pip install django-bot
Of course, if you prefer doing things the hard way, by pulling down the source code directly into your project:
git clone https://github.com/shaileshahuja/django-bot.git
Getting started
User model
First, we need to define the model and attributes of every user communicating with the bot.
models.py
from converse.models import AbstractUser
class MyUser(AbstractUser):
credits = models.FloatField(default=0.0)
This user model is automatically created for each authenticated user. For example, if a slack team is authenticated, MyUser object will be created for each user in the team. Make sure to define defaults for all fields.
Default properties available
org: The organization this user belongs to.
messenger: Returns an implementation of converse.messengers.MessengerBase object. This messenger object can be used to send messages to the user. It exposes a consistent interface for different platforms.
email: The email address of the user, if available
name: The name of the user, if available
Organization model
You must also define a model that will be instantiated for each organization that authenticates your bot. Again, remember to define defaults for any custom fields.
models.py
from converse.models import AbstractOrganization
class Organization(AbstractOrganization):
pass
Default properties available
users: A queryset of user objects that belong to this organization
messenger: Returns an implementation of converse.messengers.MessengerBase object. This messenger object can be used to send messages to a group common to all members of the organization. In Slack, this will send a message to #general.
name: The name of the organization, if available
Actions
Then, we need to define actions that can be triggered when the user sends a message. The parser will detect the intent of the user, extract parameters and the pass action be to taken back to the calling program.
actions.py
@Executor(action="account.balance")
class CreditsQuery(ActionBase):
@property
def execute(self):
self.user.messenger.send("Please wait while we retrieve your details...")
# this method is called in the background, so it is safe to make time consuming API requests
self.user.messenger.send_text("You have ${:.2f} left in your account".format(self.user.credits),
quick_replies=[QuickReply("buy credits"), QuickReply("redeem gift")]
We also need to tell django where the action classes / methods are written.
settings.py
ACTION_MODULES = ['<list of modules where actions can be found>'] # ['x.actions']
Integrating with Slack
Create a Slack app via the developer portal, and add the following credentials to your django application.
settings.py
SLACK_CLIENT_ID = '<your slack client id>'
SLACK_CLIENT_SECRET = '<your slack client secret>'
SLACK_VERIFICATION_TOKEN = '<your slack verification token>'
Next, add this to your django URLs.
urls.py
urlpatterns = [
...,
url(r'^converse/', include('converse.urls', namespace='converse'))
]
Next, start your server (behind https, try ngrok if in development environment), and add these URLs to your Slack app.
OAuth & Permissions -> Redirect URLs: <https base url>/converse/slack/oauth
Event Subscriptions -> Request URL: <https base url>/converse/slack/webhook
Interactive Messages -> Request URL: <https base url>/converse/slack/action
Integrating with api.ai
settings.py
# right now this is the only supported NLP framework for chatbots
TEXT_PARSER = 'converse.parsers.APIAIParser'
API_AI_CLIENT_TOKEN = '<your api.ai client token>'
To match the actions in api.ai to the actions you write, make sure the name in @Executor(action="<name>") is the same as the one the ‘actions’ field in your intent. You can access the slot filling params using self.params and the conversation context using self.contexts.
Sending messages as the bot
converse.messengers.MessengerBase: This class provides the API for all messenger classes
converse.messengers.SlackMessenger: Implements this API, and so will all future implementations of other messengers.
Methods:
send: To send a plaintext message.
send_text: To send a message with quick replies.
send_image: To send an image with quick replies.
Quick replies are instant prompts for the user to click and respond. In Slack, they are sent as actions.
Example:
user.messenger.send_text("Are you sure?", quick_replies=[QuickReply("yes"), QuickReply(text="Cancel", value="No")])
Clicking on ‘yes’ will send a request back to your server with query QuickReply.value.
Implementing your own parser
Parsers are responsible for understanding the intent of the user from the text query, which receives the text to be parsed and the session id. The session id can be used to respond to queries with context. converse.parsers.APIAIParser is one such parser that connects to api.ai. You may implement your own by extending converse.parsers.ParserBase and implementing the parse method. This method receives the text query and the session id and should return a ParserResponse object.
Project details
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
Source Distribution
Built Distribution
Hashes for django_bot-0.2.1-py2.py3-none-any.whl
Algorithm | Hash digest | |
---|---|---|
SHA256 | e5f75395251a2ea9d86218b9c9d9b3c900fb7eb37cf895a8a434a9df6ca3edc3 |
|
MD5 | b748a582aef91b03c63d3fd3b1cac385 |
|
BLAKE2b-256 | 34a2888a9f12bb474f1dd2c3f1de2e2ff735e73d783a6b6209852333e6b2ff63 |