Regex for digits in Unix find command

Question

I have this command:

find reports/ -type f -mtime +90 -regex ".*\.\(csv\|sql\|txt\|xls\|zip\)"

And I need to beef it up so the part before the file extensions matches a YYYY/MM/DD pattern, like so:

reports/2010/10/10/23.txt
reports/2010/10/10/23.xls
reports/2010/10/10/26.csv
reports/2010/10/10/26.sql
reports/2010/10/10/26.txt
reports/2010/10/10/26.xls
reports/2010/10/10/27.csv

But I'm failing to get any permutation of \d and parens escaping to work.

UPDATE: here's what worked for me based on the accepted answer below:

find reports/ -type f -mtime +90 -regex "reports/201[01]/\([1-9]\|1[012]\)/\([1-9]\|[12][0-9]\|3[01]\)/.*\.\(csv\|sql\|txt\|xls\|zip\)"

Solution 1

This is what I have used in the past:

Year: (19|20)[0-9][0-9]
Month:  0[1-9]|1[012]
Day: (0[1-9]|[12][0-9]|3[01])

You can put these together in your regex. You will, ofcourse, have to escape the brackets and pipes.

Solution 2

\d is an extension of regular expressions that is not supported by Emacs regular expressions and POSIX regular expressions (those are the flavours find supports). You can use [[:digit:]] or [0-9] instead.

Solution 3

The following is ugly and does not weed out invalid dates, but might be close enough:

find reports/ -type f -regex ".*/reports/[0-9][0-9][0-9][0-9]/[0-9][0-9]/[0-9][0-9]/[0-9][0-9]\.\(csv\|sql\|txt\|xls\|zip\)"

Solution 4

You can use the repeaters like this:

find ./ -regextype posix-egrep -iregex ".*\._[0-9]{8}-[0-9]{6}.*"

I use this to find backups of the form:

./foo._20140716-121745.OLD

Where foo is the original name and the numbers are the date and time.

(on CentOS 6.5)

P.S. -regextype posix-extended works too.

View more solutions

Comments

[Teflon Ted]

That looks good (and I'll test it in a bit) but is it possible to tighten up the ranges with something like "[0-9]{4}" instead of repeating it four times in a row?

[David J. Liszewski]

The numeric quantifier "{4}" did not seem to work with the version of regexec in the version of libc used by find on my system (libc 2.3.4). YMMV.

[zpea]

You can use [0-9], but if you can use [[:digit:]] depends on which -regextype you use. For example emacs (the default type) does not support it whereas posix-extended does. See GNU findutils manual: 8.5 Regular Expressions for the syntax descriptions linked on the bottom.