chatbotmaker 0.0.13

This package automates the process of bot creation

ChatBotMaker

This project aims to automate plateform messaging where the plateform support message forwarding.

Description

This module is based around a bot class in which you inject the necessary code / objects:

• Messenger (An object that sends message or event back)
• Dispatcher (An object that contains all the logic rule)
• Database (An object that allows database interaction)

General Idea

User are in states and define what action should be executed.

A state is composed of 3 event and 1 input event (or main event):

• func(user, user_input): main function called with the user's input
• enter_func(user): called every time before the user sends text in its current state
• pre_func(user): called when entering a state (ie. during change_state)
• post_func(user): called when exiting a state (ie. during change_state)

Components

Messenger

You can create your own messenger class that should inherit the chatbotmaker.Messenger class.
It must implement a send(user_id: str, message:str) method and can implement other optional methods.

Dispatcher

The dispatcher recieves your config as a dictionnary in the following format:

{  
  'actions': {  
    'handle_name': {  
        'enter_func': lambda user: user.send_message('Hi there every time'),  
        'pref_func': lambda user: user.send_message('Hi there'),  
        'func': lambda user, user_input: user.change_state('home'),  
        'post_func': lambda user: user.send_message('You are redirected'),  
    },  
    'home': {  
        'pref_func': lambda user: user.send_message('Welcome back!'),  
        'func': 'lambda user, user_input: user.change_state(user_input)',  
    },  
    'input': {  
        'func': 'lambda user, user_input: (  
                    user.store_argument('input', user_input),  
                    user.change_state('home'),  
                )',  
    },  
    'help': Dispatcher.DEFAULT,
  }  
}

DEFAULT

You can associate function by their name to make the binding process easier. You must name your functions with {key}_{handle_name} (eg: func_help, pre_func_help, post_func_help). You then must pass the binding table dictionary to the Dispatcher constructor alongside the config __init__(self, config, env=None) . This generally are locals() or globals().

How does it work?

The user is the orm User class to whom we add some method (redirect failing attribute calls):

• send_message(message: str)
• change_state(state: str)
• get_argument(name: str)
• store_argument(name: str, value: str)
• self.messenger, self.dispatcher, self.database (the one onjected in the bot)

Database

The given database database follows the following architecture:

• users:
  • id = Column(Integer, primary_key=True)
  • fb_id = Column(String)
  • state = Column(String)
  • arguments = relationship('Argument', back_populates='user', lazy='dynamic')
• arguments:
  • id = Column(Integer, primary_key=True)
  • name = Column(String)
  • value = Column(String)
  • user_id = Column(Integer, ForeignKey('users.id'))
  • user = relationship('User', uselist=False, back_populates='arguments')

The database expects a config (sqlachemy.config) object to initialize the database. Please check sqlalchemy configuration for more information.

Custom tables / ORM classes

You can add custom tables (thus ORM classes). To do so create your own base (declarative_base() from sqlalchemy) and with it your own models . Then inject it in the database object database([...], base=my_base).

In case you want to create relationships with default classes, it is possible. A function add_relationship(class_to_add, name, value) is available in the database module.

Usage

Default components

To avoid re-inventing the wheel, some "common" components have already been coded. They are in chatbotmaker.default.

Facebook

• FacebookMessenger(authentication_token)
• facebook_route(request, facebook_check_token, bot)
  • this flask routing should be called directly from the routing function

Dev

We have the dev file containing:

• DevMessenger() # prints everythin in console

Installation

Using PIP since its a pip module repository:

42sh$ : pip install chatbotmaker

Contributing

Do no hesitate to make a pull request or launch a discussion. I am looking foreward to expand the default capabilites.

Authors and acknowledgment

Author:

Dominique MICHEL [e-mail]

Status

The project has reached its first final phase. Now there will be:

• need to think about the design and facilitate user-database integration
  • Good work done here, maybe allow any orm engiine ine the future
• need of tests (why not make a CI pipeline)
  • tests must tests more the content of the calls
  • github ci pipeline is so confusing coming from gitlab :o

Once the backend is functional and robust, i aim to make a frontend plateform to allow non-programming people to create bots too.