The pipeline fails with the following status:

Error: Process completed with exit code 137

A quick scan through the internet points to the tests being interrupted by a SIGKILL (9) being sent from the ubuntu based runner. This is likely caused by excessive memory usage, but need to investigate further why this is occurring and what has changed since 70172ec where it was working fine.

It was discovered that singular and plural forms of ingredients were being written to the database. In order to avoid this, we will lemmatize the tokens.

I do not want to redirect user directly to the source website, but take them to a unique recipe page with more information regarding the recipe.

With nuxt you can do this relatively easily https://nuxtjs.org/guide/routing/

Right now the URL is hardcoded in the application. This should be dynamic whether in development, production, etc.

Right now axios is loading all recipes from api. I am not sure whether this will cause problems when scaling. Need to look into dynamically loading recipes as the user scrolls down to the bottom of the page.

After switching from orm package to raw sqlalchemy async support, I ran into the issue of exceptions being raised when a unique constraint was broken.

The prior functionality was to except the Unique constraint exception and move along with the next ingredient, but now it is breaking the add recipe endpoint:

ERROR: duplicate key value violates unique constraint "ingredients_ingredient_key"

A cleaner solution would be to first check if ingredient exists before trying to add it.

Update the readme to give an overview of the what the project entails.

Currently, I do not have anything in-place regarding database migrations. I am completely removing the database volume and recreating the tables.

Goal:

• Research best practices
• Check out https://alembic.sqlalchemy.org/en/latest/
• Apply migrations

Documentation needed for frontend build or another make command will be usefull

cd frontend 
yarn install
nuxt generate

Another things backend failing unless you build images locally since backend/api/start.sh needed to run

it gives this error

 File "/opt/mychef/app/main.py", line 4, in <module>
    from app.api.v1.router import router
  File "/opt/mychef/app/api/v1/router.py", line 3, in <module>
    from .endpoints import ingredients, recipes, sources
  File "/opt/mychef/app/api/v1/endpoints/ingredients.py", line 12, in <module>
    from app.db.repositories.ingredients import IngredientsRepository
ModuleNotFoundError: No module named 'app.db.repositories'
[2022-03-23 19:13:18 +0000] [12] [INFO] Worker exiting (pid: 12)
[2022-03-23 19:13:18 +0000] [1] [INFO] Shutting down: Master
[2022-03-23 19:13:18 +0000] [1] [INFO] Reason: Worker failed to boot.```

When I comment that line it worked fine and backend updated. before that i tried like couple times and every time scraper failed Recipe already exist.
I guess in my first run scraper interrupted and when I restart make scrape it fails on same recipe and this does not allow to update further recipes with backend

Many of the dependencies have become outdated and I ran into trouble build and running the app in clean environment.

May want to also consider switching images to debian to avoid long build times with alpine.

The picture size of the recipe cards can change based on the dimensions of the image. Need to confine the size to a defined dimension.

Right now we load a blank nlp model and attach it to the app. Instead we should return None if the model is not found and return an error when a recipe post request is attempted, which should consequently stop the spider as well.

Add base frontend to application.

The following should be done:

• add web service to docker-compose
• display recipes and pictures of meals in a grid format (Vue + Bootstrap)
  • used Vuetify instead of bootstrap: https://vuetifyjs.com/en/
• display ingredients on either hover or click
  • this crowds the overlay container -> in the future will display a Match %
• open a ticket to add user ingredient input #10
• open a ticket to improve recipe grid for mobile, medium and large windows #11

I think it is a good idea to refactor front end to use nuxt instead of vue for three main reasons:

• SEO => vuxt prerenders html
• Routing => routing comes with /pages directory
• State Management => nuxt comes with vuex

The UI should show that no recipes were found when the frontend does not receive recipes from the api.

Need a way to display available ingredients when the user is inputting data. Right now we only store a list of ingredients as a JSON array via the recipe endpoint.

Tasks:

• Write tests
• Create a schema and DB model for ingredients
• Create an endpoint for reading/writing ingredients
• Write ingredients that are not in DB before storing recipe

Update docker service for the api to utilize gunicorn for multiple workers etc.

Follow advice from https://fastapi.tiangolo.com/deployment/

Link to docker image https://github.com/tiangolo/uvicorn-gunicorn-fastapi-docker

Improve UX by dynamically changing the grid dimensions of the recipe list:

Goal:

• Create an adaptive grid for small, medium, and large screens

Docs: https://docs.meilisearch.com/

Make scraper an optional service through using thedepends_on keyword in docker-compose.

Before how I started service:

docker-compose up -d which started all service

New way:

docker compose up ui which will start db and api because ui depends on these

Also make the entrypoint of scraper scrapy.

Prompt user for ingredients.

Goal:

• At a minimum the app should accept user recipe keywords as input and store

Horizon:

• Assist ingredient input with completion -> will need to create a table in DB with unique ingredients
• Rank recipes by Match %

• Add mypy dependency to image
• Add static typing to API
• Add static typing to Scraper

I think it is a good idea to refactor front end to use nuxt instead of vue for three main reasons:

• SEO => vuxt prerenders html
• Routing => routing comes with /pages directory
• State Management => nuxt comes with vuex

https://nuxtjs.org/