Hacker News Reader

Comment by diveanon | Hacker News Reader

Something that I have always wished for is a more human readable dsl for regex.

I don't really mind developing regex, but revisiting a complex regex years later without docs is extremely painful.

I have a (mostly abandoned, since I couldn't find an organization to cbe early adopter in the time I gave myself to work on it) side project with this aim. I spent a lot of time thinking about language design, from the perspective of enabling adoption of a new language when an existing one has so much network effect:

https://www.slideshare.net/AurSaraf/re3-modern-regex-syntax-...

https://github.com/sonoflilit/re2

Perl's regex engine actually supports defining regex in human readable groups[0] which I think is pretty nice.

In my regex engine[1], I'm adding support for a blended mix of plain regex syntax and EBNF-like definition syntax, so you can do something similar.

[0] https://stackoverflow.com/a/1917982

[1] https://github.com/hexops/zorex

There are a few. E.g.:

https://verbalexpressions.github.io/

nice link, thanks for sharing

Did you look at SNOBOL/REBOL/SPITBOL/RED?

https://www.red-lang.org/2013/11/041-introducing-parse.html

Rebol family of languages is not related to Snobol. They also have PEG parsing engines, not regex ones.

Indeed, my bad. I mis-remembered there being an evolutionary link - but I don'ttthink the is one, there's SNOBOL4/SPITBOL, and REBOL/Red.

I agree, I would think there should be a better DSL at the finite state machines/automaton/grammar level, but designed such that regular expressions could be implemented in trivially and in a readable manner.

Reply on news.ycombinator.com