Regex not matching file using `find` despite being valid

Question

I am attempting to run this find command to match files with a camelcased name, with the goal being to put a space between adjacent lower and upper case letters, but it's not matching anything:

find -E . -regex "([a-z])([A-Z])"

An example of a file I'm trying to match is

./250 - ErosPhiliaAgape.mp3

I've tested this regex on this file here and it matches successfully.

What is it I'm doing wrong with my find command?

What operating system are you using? The `-E` is not a standard option. — terdon, Dec 23 '16 at 16:06
Yeah, I found it. In future, please remember to always mention your OS. GNU `find` (Linux) is not the same as BSD `find` (OSX`) which is not the same as POSIX `find`. Many of the standard tools behave differently in different *nix systems. — terdon, Dec 23 '16 at 16:14

score 4 · Accepted Answer · answered Dec 23 '16 at 15:28

4

You're probably looking for something like:

find . -name "*[a-z][A-Z]*"

answered Dec 23 '16 at 15:28

SYN

2,793
12
19

That works, but when I place the matched characters in a capture group like so "*([a-z])([A-Z])*" it doesn't match. – Dec 23 '16 at 15:31
1

Find won't handle capture groups. See http://unix.stackexchange.com/questions/139331/can-regex-capture-groups-be-used-in-gnu-find-command – SYN Dec 23 '16 at 15:34
@ChrisWilson that's because the regex needs to match the entire path. The `-name` does not take regular expressions, [it takes globs](http://unix.stackexchange.com/a/46444/22222), and that only needs top match the name. – terdon Dec 23 '16 at 16:32
@SYN actually, [as I just learned](http://unix.stackexchange.com/questions/332400/regex-not-matching-file-using-find-despite-being-valid/332405?noredirect=1#comment585019_332405), GNU `find` *can* handle capture groups. Dunno about BSD/OSX `find`. In fact, that is very nicely demonstrated in the accepted answer to the question you linked to. – terdon Dec 23 '16 at 16:34

score 3 · Answer 2 · edited Apr 13 '17 at 12:36

3

Find's -regex matches the entire path, not just the file name. That means that to find /path/to/foo, you need -regex'.*foo', and not justfoo`. You want something like:

find  . -E -regex ".*[a-z][A-Z].*"

It would be much simpler to use globs and the simpler -name as suggested by SYN, however, instead of regexes.

edited Apr 13 '17 at 12:36

Community

1

answered Dec 23 '16 at 16:13

terdon

234,489
66
447
667

@steeldriver wow. That's completely new to me, thanks! – terdon Dec 23 '16 at 16:29

score 1 · Answer 3 · answered Dec 23 '16 at 21:01

This happens because ([a-z])([A-Z]) does not match ./250 - ErosPhiliaAgape.mp3. In fact, ([a-z])([A-Z]) can only match exactly 2 characters -- -regex is an anchored match over the entire path. If you want to use -regex to search for a file whose name contains a regex, you can write it like this (BSD/macOS find syntax):

find -E . -regex '.*/[^/]*([a-z])([A-Z])[^/]*'

The GNU find equivalent would be something like

find . -regextype posix-extended -regex '.*/[^/]*([a-z])([A-Z])[^/]*'

Regex not matching file using `find` despite being valid

3 Answers3