urbanonymous / api_demo_v2

API created with Python - FastAPI. Handles conversational sms messaging using Twilio and Rasa.

api_demo_v2

This API is the result of a hiring exercise.

The purpose of this API is to allow an user to subscribe to updates of the service.

Technologies used:

• Python
• FastAPI
• Twilio
• Rasa
• MongoDB
• Docker
• Docker Compose

Use Case overview

User obtains a token to access to our API

User subscribes to updates of the service using the API, sending the token and a valid phone number.

Server sends a confirmation message to the requested phone number, to validate that the user wants to receive updates from the service.

User confirms to receive updates by sms

Server sends an update after 15s and other after 35s

User unsubscribes

Server sends a confirmation message

API routes

The API has 3 routes:

• POST /login - Returns a JWT token for interacting with the API.
• GET /me - Returns the information about the token.
• GET /run - Initiates a sms conversation with the phone number sent.

For more documentation, read the OpenAPI spec at http://localhost:8080/docs

Start the API

To start the API you need to execute the following command: make

That will build the docker images needed by executing docker-compose build and start the API container at the port 8080

The build process takes around 3-5 minutes, as it needs to download all the required software.

After that, the containers are going to start, but the API will wait 70s to let Rasa initialize.

Rasa has a model already pretrained! :)

The api will be located at http://localhost:8080 The default username is spicy The default password is soup

Clean environment

To remove the container and images of this API, run the following commands:

make stop

make clean

Project structure

Files related to application logic are inside the backend/app directory. Application parts are:

api
├── auth                    - auth related deps and routes.
│   ├── deps                - dependencies for routes with auth.
│   └── auth                - auth related routes.
├── core                    - internal API objects.
│   ├── config.py           - settings for this api.
│   ├── database.py         - mongodb client and utils.
│   ├── message_handler.py  - process messages and apply bussiness logic.
│   ├── message_listener.py - listens to sms from Twilio.
│   ├── rasa_adapter.py     - connects to rasa and expose helpers.
│   └── twilio_adapter.py   - connects to twilio and expose helpers.
├── routes                  - web routes
│   └── subscriptions.py    - subscriptions related routes.
├── schemas                 - pydantic models for this api.
├── main.py                 - FastAPI application creation.
└── pre_start.py            - database population.

Files related to the conversational AI are inside the rasa directory.

TODO

• Implement logging system and connect it with external providers like Sentry