How to recursively search for tar files and then search them (with xargs)

[genivia-inc]

Thinking it might be useful to add question #359 and answer here for those who have similar questions.

Question

I'm searching for a phrase inside files that are themselves inside tar files which are scattered throughout a number of sub directories.

$ ugrep -rl --include="*.tar" 'phrase' .

What I get in output is:

dir1/tarfile.tar
dir2/othertarfile.tar
dir2/anothertar.tar
dir3/tartar.tar

What I want is not to see the name of the file inside the tar file that the phrase was in. For example:

dir1/tarfile.tar has a file in it called /home/user/thefile.txt and it contains the phrase "phrase" that I'm searching for. I want to see the name of the "/home/user/thefile.txt" as well as the name of the dir1/tarfile.tar file.

Answer

Use xargs with ugrep as follows.

If you want to recursively search for tar files only and then list the files in those that match a pattern:

$ ug -rl -Otar "" | xargs ug -lz "PATTERN"

This runs a ugrep to look for tar files recursively with ug -rl -Otar "" and a second ugrep to list each of the tar contents with ug -lz "PATTERN" to match PATTERN in files stored in tar files.

I am using ug here to pretty print and sort the tar files (option --sort). With ugrep use the following to sort tar files by name (or specify a sort key):

$ ugrep --sort -rl -Otar "" | xargs ugrep -lz "PATTERN"

Note that the problem with using --include="*.tar" (same as -Otar) is that you only select tar files to search, which also only searches tar files inside tar files when you use option -z. Option -z must be used to search inside tar, tarballs, zip files and compressed files.

[genivia-inc]

Note: see answer in the question post.