https://github.com/pykong/weatherwiz

A chatbot for conversational weather forecasts, built with Rasa, Python and love ❤️.
https://github.com/pykong/weatherwiz

brightsky chatbot iubh poetry python rasa weather weather-forecast

Last synced: 21 days ago
JSON representation

A chatbot for conversational weather forecasts, built with Rasa, Python and love ❤️.

• Host: GitHub
• URL: https://github.com/pykong/weatherwiz
• Owner: pykong
• License: mit
• Created: 2024-01-21T18:39:04.000Z (10 months ago)
• Default Branch: main
• Last Pushed: 2024-01-21T18:53:01.000Z (10 months ago)
• Last Synced: 2024-05-14T00:06:24.842Z (6 months ago)
• Topics: brightsky, chatbot, iubh, poetry, python, rasa, weather, weather-forecast
• Language: Python
• Homepage:
• Size: 9.26 MB
• Stars: 2
• Watchers: 1
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md
  • License: LICENSE
  • Citation: CITATION.cff

Awesome Lists containing this project

README

        



    

        

        

    





    

    

    

    

    

    

    

        

    

    

    





    

        

    



# WeatherWiz

WeatherWiz is a chatbot for weather forecasts.

It is built with [Rasa](https://rasa.com/), Python and love ❤️ and proudly **features**:

- a fancy frontend build with [chatroom.js](https://github.com/scalableminds/chatroom)

- using free APIs (_no signup or API key required_):

  - **real weather data** from [Bright Sky](https://brightsky.dev/docs/#/) API

    - forecasts up to seven days

    - historical data from 01.01.2010

  - geocoding via [Nominatim](https://nominatim.org/) API

  - user IP address as a location fallback via [ipinfo.io](https://ipinfo.io/) API

- slots for place and time for keeping conversation context

- a [dateparser](https://dateparser.readthedocs.io/en/latest/) for date and time extraction

- high test coverage via Pytest and Rasa's tests

- full Dockerization

## Getting started

To start WeatherWiz you will need:

- a working [Docker engine](https://docs.docker.com/engine/install/)

- Internet access (for setup and API access)

- at least 16GB RAM (32GB recommended)

Talking to WeatherWiz is as easy as:

1. Clone the project

2. Open a shell in the project folder

3. In the shell, run `docker compose up`

4. Open [localhost:8000](http://localhost:8000) in your browser

5. Enjoy talking to WeatherWiz... 🙂

## Contributing

WeatherWiz suggests the following development setup:

1. **Docker** to run the application

2. **Poetry** for dependency management

3. **Ruff** for formatting and linting

4. A type checker, preferably [pyright](https://github.com/microsoft/pyright)

5. **Pytest** for unit testing

Configuration is located in the `pyproject.toml`, at the root of the project.

To contribute:

1. Fork this repository

2. Do your changes, including:

   - test coverage

   - documentation

3. Open a PR

You may also feel free to open any issue.

## Current limitations

- WeatherWiz will only answer questions regarding rain, temperature and wind

- answer format remains generic

- failure of API access is not handled gracefully

## Links

- [dateparser](https://dateparser.readthedocs.io/en/latest/)

- [DateparserEntityExtractor](https://rasahq.github.io/rasa-nlu-examples/docs/extractors/dateparser/)

- [chatroom.js](https://github.com/scalableminds/chatroom)

- [Bright Sky](https://brightsky.dev/docs/#/)

- [Nominatim](https://nominatim.org/)

- [ipinfo.io](https://ipinfo.io/)