Searching a pattern in file names with number

Question

I have a large number of files in a folder each with the date as the filename.
For example

20130101
20130102
20130103
.
.
.
20130131

similarly for other months 20130201 to 20130230, the pattern is [Year][Month][Day]. I need to make a tar file of the first fifteen files from JAN to APRIL (i.e 2013[01-04][01-15]), since double digit number range is not allowed in grep. How should I search this pattern?

Solution 1

With zsh:

tar zcf ~/file.tar.gz 20130<1-4><1-15>

If you have to use bash:

shopt -s extglob
tar zcf ~/file.tar.gz 20130[1-4]@(0?|1[0-5])

Solution 2

Use find to find the filenames:

find . -maxdepth 1 -name "20130[1-4]0?" -o -name "20130[1-4]1[0-5]" 

check if that is the correct set, and use the output as input for cpio:

find . -maxdepth 1 -name "20130[1-4]0?" -o -name "20130[1-4]1[0-5]" | cpio --create --format=ustar -O file.tar

@richard pointed out this could traverse in sub-directories (if a sub-directory matches the pattern) (and that shell expanded the pattern, corrected).

Expanding on the commandline runs the risk of the commandline becoming too long, which might happen if you had a file every few seconds for every day (20130101-00005, 20130101-00007. Piping the filename into cpio does not have problem.

Solution 3

The grep bit is.

grep -zE "^20130[1-4](0[1-9]|1[0-5])$"

Comments

[ctrl-alt-delor]

Note it is not find that is doing the work here, but the shell. Warning will recurse directories, use echo instead of find to not do this.

[Aditya Cherla]

would'nt the space after question mark give an error?

[ctrl-alt-delor]

Beware though this should work it is fragile: e.g. if the year was 2016 it would not work.

[ctrl-alt-delor]

@Aditya why? it needs the space.

[Aditya Cherla]

@richard Sorry! my mistake just looked at an example of it

[ctrl-alt-delor]

Maybe call these first in your script shopt -s nullglob; shopt -u failglob, to stop it failing when a file is not found.

[Zelda]

@richard, You right about the shell doing the work, but not about the recursion in directories and displaying unless there are folders with that filename pattern. That is different when you use find . -name "somepattern"

[bsd]

std grep will work as well. Since this particular pattern happens to want months 01-04, the first 0 can be included with the year. So, grep '20130[1-4][01..15]' will match the specified range.

[Stéphane Chazelas]

@bdowning, no, [01..15], same as [015.] matches any of those 4 characters 0, 1, ., or 5.

[ctrl-alt-delor]

Yes that is what I meant only recurses directories that match the pattern, not all directories.