Labeled matches: why is this not in every regex engine?
18 points by iev
18 points by iev
Have you thought about trying to add RE# to rebar? I'd be curious to see a performance comparison between it and JavelinPattern.
They haven't updated it in a while, and probably partially because it's not apples-to-apples for many things, but they did do some work on it.
Here's there forked repo:
https://github.com/ieviev/rebar
It's fast.
Oh i never knew about JavelinPattern! This is very interesting, thanks!
There is an older version benchmarked here, but i can benchmark it again soon. It should be highly competitive, the earlier version had a few optimization bugs that made it quadratic.
I hope I am not the one one who cringed every time they went out of their way to claim that [0-9]{4}-[0-9]{2}-[0-9]{2} was a date, as if 0000-00-00 was going to show up in a document. If I had more emotionally energy, I'd for sure put spaCy up against a document containing nonsensical 10 digit strings and see how many it thinks are the 0th of Niluary in the Nilth Century
Can’t understand if it’s sarcasm here but we precisely carve every valid date in yyyy-mm-dd. even in textual forms like “December Thirteen Nineteen-Fifty-Four” is regular