BotHub SDK (Python)

This package provide components to works with BotHub.Studio, which is a chatbot hosting service.

With bothub-cli, you can deploy a new chatbot with just four lines of commands.

Installation

To install bothub:

$ pip install bothub

The bothub package works on python2 and python3(recommended) both.

Getting Started

You can build an echo chatbot simply by subclassing BaseBot class and overriding handle_message method.

# -*- coding: utf-8 -*-

from bothub_client.bot import BaseBot
from bothub_client.decorators import channel

class Bot(BaseBot):
    """Represent a Bot logic which interacts with a user.

    BaseBot superclass have methods belows:

    * Send message
      * self.send_message(message, chat_id=None, channel=None, extra=None)
    * Data Storage
      * self.set_project_data(data)
      * self.get_project_data()
      * self.set_user_data(data, user_id=None, channel=None)
      * self.get_user_data(user_id=None, channel=None)
      * self.nlu(vendor) -> NluClient

    When you omit user_id and channel argument, it regarded as a user
    who triggered a bot.
    """

    @channel()
    def default_handler(self, event, context):
        self.send_message(event['content'])

When a bot receives a message from an user, it triggers handle_message method with event and context object.

An event is a dict which contains following items:

• content: A message text received.
• channel: Which channel (messenger platform) sent a message.
• sender: Who sent a message. {"id": <user-id>, "name": "<username>}
• chat_id: Chatroom ID where message is sent. It can be a 1:1 chatroom or group chatroom.
• location: Location information if possible {"longitude": <float>, "latitude": <float>}
• postback: A postback data.
• new_joined: A boolean which indicates this bot was invited to some chatroom or not.
• raw_data: A raw data itself messenger platforms sent.

You can respond to this message with various tools we provides.

API Reference

Messaging

To send a message, use a self.send_message method with a message you want to send.

self.send_message('hello')

In most cases, you may omit user_id and channel arguments. Then it replies to whom sent a message to your bot. Put values to those arguments when you want to specify a receiver.

You can send a message with rich controls like ‘quick replies’ or ‘buttons’ using Message object.

from bothub_client.messages import Message

message = Message(event).add_quick_reply('Go ahead')\
                        .add_quick_reply('Never mind')\
                        .set_text('May I reserve the seat?')
self.send_message(message)

Message class provides these methods:

• set_text(text)
• add_url_button(text, url):
• add_postback_button(text, payload)
• add_quick_reply(text, payload=None, image_url=None)
• add_location_request(text) _{* Kakao, Slack is not supported.}
• add_keyboard_button(text) _{* Facebook is not supported.}
• add_image(url, width, height) _{* Facebook, Telegram is not supported.}

Storage

To store/retreive some property data, we provides following methods:

• Project level

  • self.set_project_data(data): set data to a project
  • self.get_project_data(): get data from a project

• User level

  • self.set_user_data(data, user_id=None, channel=None): set user data
  • self.get_user_data(user_id=None, channel=None): get user data

data should be a dict. An existing properties not included in data will be ignored, not be deleted.

If user_id and channel is None, it regarded as a message sender.

NLU Integeration

You can use nlu method to perform NLU after setup NLU integration at BotHub.Studio.

There are two styles to request to NLU service. (eg. to use API.ai)

First, use event object to construct message and session_id.

def handle_message(self, event, context):
   response = self.nlu('apiai').ask(event=event)
   self.send_message(response.next_message)

Or, put explicit message and session_id by yourself.

def handle_message(self, event, context):
   response = self.nlu('apiai').ask(message='hello', session_id='customer1')
   self.send_message(response.next_message)

ask method returns a NluResponse object which contains attributes like:

• raw_response: A raw response which NLU service returns.
• action: A NluAction class object to identify intent and required parameters.
• next_message: Next message text to respond NLU service recommend.

A NluAction object contains attributes like:

• intent: Intent name
• parameters: parameter dict
• completed: A boolean indicates whether action completed

For incompleted action, you need to reply to user with next_message attribute of a NluResponse instance to complete action.