What it does:
- Iterate over store ids 1-7000, looking for stock levels of the specified VMPs.
- For stores that aren't found, record these in
output/stores_notfound.txt - Write the stock as JSON to per-VMP files.
- Combine these files into a single CSV.
- When run via Github actions, write the result back to the repo.
It takes about 15 mins to scrape one VMP. Each job in a workflow can run for 6 hours. So without modification, you can reliably track up to ~18 VMPs.
The master data is recorded at outputs/stock_levels_all.csv;
for convenience, heat maps showing low stock by PCN are also generated
as png files in outputs/, named for the VMP.