This PR does too many things :(
1. Major refactoring of the dev/scripts and dev/code-generation folders.
Primarily this was removing duplicate code and cleaning up some poorly written code snippets as well as making them more idempotent so then can be re-run over and over again but still maintain the same results. This is working on my machine, but I've been having problems in CI and comparing diffs so running generators in CI will have to wait.
2. Re-Implement using the generated api routes for testing
This was a _huge_ refactor that touched damn near every test file but now we have auto-generated typed routes with inline hints and it's used for nearly every test excluding a few that use classes for better parameterization. This should greatly reduce errors when writing new tests.
3. Minor Perf improvements for the All Recipes endpoint
A. Removed redundant loops
B. Uses orjson to do the encoding directly and returns a byte response instead of relying on the default
jsonable_encoder.
4. Fix some TS type errors that cropped up for seemingly no reason half way through the PR.
See this issue https://github.com/phillipdupuis/pydantic-to-typescript/issues/28
Basically, the generated TS type is not-correct since Pydantic will automatically fill in null fields. The resulting TS type is generated with a ? to indicate it can be null even though we _know_ that i can't be.
* added api extras to other tables
genericized api extras model from recipes
added extras column to ingredient foods
added extras column to shopping lists
added extras column to shopping list items
* updated alembic version test
* made mypy happy
* added TODO on test that does nothing
* added extras tests for lists, items, and foods
* added docs for new extras
* modified alembic versions to eliminate branching
* refactor(backend): ♻️ cleanup duplicate code in http services
* refactor(backend): ♻️ refactor database away from singleton design
removed the database single and instead injected the session into a new Database class that is created during each request life-cycle. Now sessions no longer need to be passed into each method on the database
All tests pass, but there are likely some hidden breaking changes that were not discovered.
* fix venv
* disable venv cache
* fix install script
* bump poetry version
* postgres fixes
* revert install
* fix db initialization for postgres
* add postgres to docker
* refactor(backend): ♻️ cleanup unused and duplicate code in http services
* refactor(backend): remove sessions from arguments
* refactor(backend): ♻️ convert units and ingredients to use http service class
* test(backend): ✅ add unit and food tests
* lint
* update tags
* re-enable cache
* fix missing fraction in db
* fix lint
Co-authored-by: hay-kot <hay-kot@pm.me>
