Poshan Didi Bot

This is the README for the Poshan Didi telegram bot. This repository has two python applications in it:

• A Flask-based application that provides a very lightweight API for accessing users and messages captured by the system
• A Telegram bot that serves the Poshan Didi bot

It is likely bad form to have both of these in the same repo, but this is research and it is the easiest way to share the db.py file between the two.

Scripts are written with Python 3.6+

Getting started

Setup virtualenv

Make a Python 3.6+ virtualenv and enter it.

With virtualenvwrapper:

mkvirtualenv --no-site-packages -p python3.6 poshan-didi

Install requirements

pip install -r requirements.txt

Local settings

For local settings, we use the simple-settings library. Note: This is used by the telegram bot and the Flask server because of the database credentials.

To get a local settings file first cp settings_sample.py localsettings.py.

When running code you should use a command-line or environment variable such as:

$ python script.py --settings=localsettings

or

$ SIMPLE_SETTINGS=localsettings python script.py

Customizing settings

You will need to specify the TELEGRAM_TOKEN for your bot, the NURSE_CHAT_ID that corresponds to the Telegram account you want to escalate messages up to, and--optionally--the DB_SQLALCHEMY_CONNECTION string for the database you want to use to store users and messages.

The TELEGRAM_TOKEN comes from the BotFather when you initially create your telegram bot.

The NURSE_CHAT_ID is the Telegram chat_id for the Telegram account that you want to escalate messages to. If you do not know this ahead of time, you can leave it as is and start the bot (after completing the steps below). Once the bot is started, send a message from the account that you want to use for the nurse. The message will be processed as a user message because the chat_id does not match the previously set NURSE_CHAT_ID. Check the log file or the database to find the chat_id for this account. Set the NURSE_CHAT_ID, restart the bot, and test that escalation is happening correctly.

Setup the database

Before running the bot or the flask server, it is necessary to create the database. At the moment, this is a sqlite flat file, but you can modify this easily. Use the DB_SQLALCHEMY_CONNECTION settings parameter to put in a valid SQLAlchemy connection string.

Before running the aplpication for the first time, import the db module from a standalone terminal and run the reset_db() command. For example:

$ SIMPLE_SETTINGS=localsettings ipy

In [1]: from db import Database

In [2]: Database().reset_db()

Running the Flask server

Development mode

To run the Flask server in development mode, which allows auto-reloading based on changes and additional debug output, use the following command:

$ SIMPLE_SETTINGS=localsettings FLASK_APP=app.py FLASK_DEBUG=1 python -m flask run

Production

TBD

Running the telegram bot

Building the NLU models

The telegram bot we built relies on the RASA-NLU package to do very basic NLU on incoming messages to determine the intent of incoming messages. This is overkill for initial purposes, but lays the ground work for complex NLU. The library requires that we build models based on the training data before it can be used. To do that, run the following command:

$ python -m rasa_nlu.train -c nlu_config.yml --data data/nlu-training-data.md -o models --fixed_model_name nlu --project current --verbose

This process should complete relatively quickly and will produce a folder called models that contains the training data from the NLU library.

Running the bot

Once you have built the models and setup the database, you can run the bot, use the following command:

$ SIMPLE_SETTINGS=localsettings python poshan_didi_bot.py