Minimal rpm2cpio using only bash, grep, tail and a decompressor

rpm2cpio

#!/bin/bash
RPMHDRLGTH=$(LC_CTYPE=C grep -abom1 '.7zXZ\|]'$'\000\000''....'$'\377\377\377\377\377\377''\|BZh9\|'$'\037\213\b' "$1")
case "$RPMHDRLGTH" in
  *7zXZ) COMPRESSOR=xz ;;
  *]*)   COMPRESSOR=lzma ;;
  *BZh9) COMPRESSOR=bzip2 ;;
  *)     COMPRESSOR=gzip ;;
esac
tail -c+$[${RPMHDRLGTH%:*}+1] "$1" | $COMPRESSOR -d

Just some notes. This assumes GNU tools. Additionally it will not work with rpm5.org rpms that use xar instead of cpio internally.

For history buffs, I first wrote this back in 2012

https://web.archive.org/web/20140303100327/http://my.opera.com/ruario/blog/2012/01/15/rpm2cpio

The one in this gist is only slightly changed

These days, packages are almost exclusively XZ compressed on recent versions of RPM-based distributions. If you only need to handle files from modern versions of Fedora, SUSE, etc. then you can get away with a single line of POSIX shell:

#!/bin/sh
tail -c+`grep -Fabom1 7zXZ "$1" | cut -d: -f1` "$1" | xz -d

The above still assumes a GNU toolset. This should work on both GNU and BSD systems and even BusyBox.

#!/bin/sh
o=`strings -o "$1" | grep -m1 7zXZ | sed 's/ *\([0-9][0-9]*\) .*/\1/'`
[ `printf popular\\\novel | strings -o | sed -n '/v/s/ *\([0-9][0-9]*\) .*/\1/p'` -eq 10 ] && o=`printf %d 0$o`
tail -c+$o "$1" | xz -d

I shall leave it as an exercise as to what I am doing here and why this looks unnecessarily complex. (spoiler: it isn't) 😉