I have seen many repositories on recipe-finder websites, but none of them were based on ingredients search. And all of them used some or the other API like edamam etc. In-order to make it a challenge, I decided to not use an API.
Instead of searching for a dish according to it's name, based on ingredients input by the user, I'm suggesting possible dishes that can be made using these ingredients. For this project I am using a mixture of web scraped recipes off of cookbooks.com and other websites which I parsed using some python code script and the Recipe Dataset (over 2M) Food publicly available on kaggle. https://www.kaggle.com/datasets/wilmerarltstrmberg/recipe-dataset-over-2m/data Since the dataset file size is very big I couldn't include it here, you can clone my repo and download the dataset from above link. Make sure to name the file as recipes_data.csv I am instead using a smaller version of the dataset within 25mb. Also I implemented the integration with computer vision model for detecting ingredients with a single top view image
For local deployment running the flask app is enough, but to deploy on netlify, I used Gunicorn. The gunicorn.conf file serves for the important configurations and location of my flask app. I added the Gunicorn command to a startup script or Procfile. For local testing of gunicorn on windows, you can refer to https://stackoverflow.com/questions/62788628/modulenotfounderror-no-module-named-fcntl Basically u can use waitress for running on windows.