Support PCRE (look behind)

[ar1a]

with the addition of --pcre2 flag into BurntSushi/ripgrep , lookbehind is something i've been pining for. It'd be great if PCRE2 support could be added

[sharkdp]

Thank you for the feedback.

Could you please provide a few examples where advanced regex features like lookbehind could be useful in a realistic fd use case?

[ar1a]

I can't think of any examples off the top of my head but the use case I was going for was the negative lookaheads/behinds. They are pretty awkward to express without lookbehinds.

[sharkdp]

I think I am going to have to decline (for now).

fd is trying to stay simple and is only targeting "80% of the use cases" (see README), so I am not going to implement a feature without any use case at all.

[jrraymond]

A use case is finding all file which don't match an extension, ei: ^.+\.(?!py$)[^.]+$. ?! requires look-around.

[sharkdp]

@jrraymond

This use case can also be solved with fds -E, --exclude <pattern>... option:

fd pattern -E '*.py'

For example: search for all files matching the README pattern that do not have an md extension:

fd README -E '*.md'

That's (arguably) also more readable than the lookbehind.

[ouxch]

@sharkdp
a use case is to find out files with both 'mutation' and 'asset' in their names
option: ^(?=.*mutation)(?=.*asset).*$

[unusualevent]

the -E exclude thing definitely slipped past me on the first few tries to solve this. Compared to wishing I had a negative lookbehind... having a quick "exclude" glob seems so much nicer. it's possible this could be phrased better in the readme as "find all files not matching x" is a use case of it, not just "ignore" :)

[sharkdp]

@unusualevent A small PR with a documentation update would be very much appreciated.

[unusualevent]

@sharkdp I'd probably butcher the readme :/