Structured identifiers

Write five recognisers, Aadhaar with its check digit, PAN, Indian mobile, IFSC and UPI id, and get the given tests green. The artifact is piiwall/patterns.py and a patterns row in your results table where Aadhaar precision is above 0.9.

Try first

Take the 195 false positives from milestone 1 (add a --show-fp flag to the CLI if you want them listed, ten lines of code). Sort them into piles by hand: phone numbers, account numbers, 12-digit traps, order ids. Write down which pile each of these ideas would remove: a length check, a first-digit check, a check digit, a look at the surrounding words.

Then, for PAN, which has no check digit, write down the strongest structural test you can think of. Only then read on.