How to grep string in tar.gz file with matched filenames?

beeth2006 · 09-28-2011, 02:44 PM

I got a log file called access.tar.gz, which has 100 text files named: a1, a2, a3, …, a100. I want to search for string "fail" in the access.tar.gz file, WITHOUT creating a temp folder by uncompress the .tar.gz file, and need to know which aX file(s) has "fail" string. How can I get the output like:

a2 <line with "fail" string>
a2 <another line with "fail" string>
a6 <line with "fail" string>
a11 <line with "fail" string>
…
( that is, the matched filenames, with matched lines ).

I can use "zgrep -a fail access.tar.gz" to get all the matched lines, WITHOUT the matched filenames, and I don't see any option in zgrep to show filenames. And I think it's very common users want to know which file has the matched lines.

How can I do it with single line of command(s) to do this without writing a script?
Thanks!

snooly · 09-28-2011, 05:00 PM

I made a few small test files, and then tarred them up with a gzip as well. When I use zgrep -a, it does seem to tell me the file name. But maybe that's because my files are very short? Approximately how long are your files?

Here's the type of output I get:

zgrep -a elephant testtar.tar.gz
a3.txt0000664000175000017500000000001111640713531010247 0ustar snoolysnoolyelephant

versions of software used:
zgrep -V
z$grep (gzip) 1.3.12

tar --version
tar (GNU tar) 1.23

beeth2006 · 09-29-2011, 09:08 AM

Yes, i think your files are short so that the matched file(s) are only shown once, not for every matched lines. But thanks!

colucix · 09-29-2011, 10:00 AM

You can try by means of a little script, like this:

Code:

#!/bin/bash
while read file
do
  result=$(tar xf access.tar.gz -O "$file" | grep fail)
  if [[ -n $result ]]
  then
    echo "$result" | awk -v file="$file" '{print file ":", $0}'
  fi
  unset result
done < <(tar tf access.tar.gz | grep -v '/$')

However it may require a lot of time if the size of the archive is huge, because it is actually unzipped and extracted, even if the output is sent to the standard output to avoid the creation of temporary files.