This was a fun build. A case where my work clearly shaved at the very least an hour or two a day, was used by many, and had an elegance.

One source of business for this company was through bidding on jobs at a private auction. Winning contracts were delivered on a standard PDF document which was used to onboard the job – by keying the information into a form with over 40 fields. Anywhere from 4 to 12 of these a day would come in, and the company’s business with this auction was growing.

Using regex, python, and a library that compiled the script into an executable, I made an “app”. It was actually just a desktop shortcut. Dragging the pdf to it opened a new browser tab, loaded the form page, “read” the document text, and auto-filled as many fields as available in the document (they were not always fully filled in by the auction house). Staff could then review it, add any notes on missing information, and submit it for processing.

Someone said that RegEx is kind of like the Defense against the Dark Arts of programming. It’s powerful, but difficult. This is compounded when working with text in a pdf document. Its structure is a kind of black box of unreliable consistency. Those queries took a minute, and most look like this:

regex.search(r'(?<=Date Services Needed:.*)(?:\([2-9]\d{2}\)\ ?|[2-9]\d{2}(?:\-?|\ ?))[2-9]\d{2}[- ]?\d{4}(?=.*Physician Information:)|$', text)

Would love to do that again for somebody. Contact me if you have this situation and are interested.