How to correctly use find with regular expression?

Question

I cannot figure out the correct use of find with -regex option:

For example, given this example:

Morcheeba/Big Calm/02 Shoulder Holster.flac
Morcheeba/Big Calm/02. Shoulder Holster.flac

I only want to match the string in the form path/to/file/## filename.ext, without the dot.

I've tried with:

find Morcheeba/ -regextype egrep -iregex ".*[0-9]{2}*[a-z]*flac$"

and some variation, but I always get all files either none.

What's wrong?

Thanks in advance.

Solution

Note that -regex/-iregex (GNU extensions), like the standard -path match on the full path, not just the file name.

If you want to find files whose name (as opposed to path) starts with 2 decimal digits, following by something that is not a decimal digit and ending in .flac, you could do portably:

find . -name '[0-9][0-9][!0-9]*.[fF][lL][aA][cC]'

Note that it wouldn't match on 01.flac because the [!0-9] can't match there. Instead you could write it:

find . -name '[0-9][0-9][!0-9]*' -name '*.[fF][lL][aA][cC]'

Those use wildcard patterns, not regexps.

To use GNU find's -iregex, the equivalents would be:

find . -regextype egrep -iregex '.*/[0-9]{2}[^0-9/][^/]*\.flac'
find . -regextype egrep -iregex '.*/[0-9]{2}([^0-9/][^/]*)?\.flac'

That is, we anchor the two digits at the start, as we make sure the rest of the regexp doesn't span a /.

Comments

[jamesbtate]

I think you are mixing . and *. In your example, the second * could be a space and the third * should be a . or \.. You could use [^\.] to match any character that is not a period. Put a * after it to match 0 or more non-period characters.

[RomanPerekhrest]

can't reproduce, seems to work (even with that strange pattern)

[Stéphane Chazelas]

That would match on ./path/to/foo123.flac or ./foo/1234/whatever.flac though.

[Daniele]

Yes, I had to fix it a bit.