Update NIH award Format in the award number normalizer

[tsande16]

normalizeAwardNumber() does not capture award numbers that have the format [A-Z][0-9][A-Z], most activity codes are 1 letter followed by 2 numbers or two letters followed by 1 number; however there are cases of letter, number, letter e.g. P2C HD042854

Note: P2C activity codes make up a very small portion of award numbers. Only 2 were found in 1 year's worth of data from PMC

ref: https://www.era.nih.gov/files/Deciphering-NIH-Application.pdf
https://grants.nih.gov/grants/funding/ac_search_results.htm

To test: mvn verify

[markpatton]

I wonder if that regexp needs to be quite so strict and complicated? Instead of adding a special that there can be a letter at the end of the initial part, the rules could just be loosened to match numbers and letters. But I may be missing some of the point.

Tests pass for me.

[tsande16]

@markpatton definitely something to look into. I agree it's complicated. Back in aug/sept I ran tests on 4-5k nih award numbers and 4-5k non-award numbers. The tests were successful. This particular case occurred because the activity code is not very common (2 within ~9.5k records).

Since non-nih award numbers are all over the place, my initial idea was to try and capture as exactly as possible so as to not perform modifications to non-nih award numbers.

The normalization probably wouldn't affect non-nih award numbers since they are only removing leading/trailing whitespace, making all characters uppercase. The only thing that could potentially cause a problem is the removal of leading a zeros. Loosening the regex could then match with a non-nih grant and remove a leading zero, and there 1,401 grants in pass_grants (STAGE) with at least 1 leading zero. More testing on this would need to be performed - I can create a ticket to look into investigate this?

[markpatton]

@tsande16 Things look good now. It was just sort of a comment about watching that regex grow in size, when we get more types of awards numbers to deal with, we may want to revisit.

[tsande16]

@markpatton that makes sense. It could definitely grow unwieldy.