Development notes

Notes only relevant for myself in 3 years or anyone else who looks at the code.

Apologies

It turned out needlessly messy, in particular because

a) I tried to use it as a “testbed” for some techniques (e.g. there is a lot of dependency inversion in certain parts) and b) I tried to go for zero dependencies for the package itself (i.e. excluding development/testing tools like pytest). No real reason why.

“Indexed Jsonable Store”

As a consequence of (b) above, I wrote a document database-like thing (with greatly reduced functionality however) under utils/indexed_jsonable_store. The goal here was to have a compressed and randomly accessible database of JSON FoodData entries so that e.g. tests run faster than they would if the entire original FoodData JSON had to be read in each time, without blowing up my hard drive with a huge uncompressed SQLite database or anything like that. Had I been fine with dependencies, I could’ve used one of the existing document databases offering compression instead.

README.md pre-commit hook

The project’s README.md is generated from the GitHub Pages markdown using an (also extremely messy for now) pre-commit hook. Run pre-commit install to install it.

Included tools

The package comes with command-line tools to export data, improve the reference data and do some simple querying:

$ fooddata-vegattributes --help
usage: fooddata-vegattributes [-h] {generate,input-ref,annotate-ref,list-by-veg-and-fdc-categories} ...

positional arguments:
  {generate,input-ref,annotate-ref,list-by-veg-and-fdc-categories}
    generate            generate vegattributes JSON
    input-ref           interactively input reference data
    annotate-ref        annotate reference data
    list-by-veg-and-fdc-categories
                        list foods by veg and FDC categories

optional arguments:
  -h, --help            show this help message and exit

Most of them are either self-explanatory or explained in their respective help texts.

Only for annotate-ref I should probably mention that “annotate” here simply means appending the food description to lines of the reference data CSV that don’t have descriptions yet. The descriptions aren’t used for anything other than making the reference data file human-readable and this tool is purely for convenience so you can just put in FDC IDs and attributes but don’t have to then manually copy descriptions as well.

TODO

More to follow at some point…