#!/bin/bash
#
# sysinfo
#
# This script when run on an rpm based system, will print out a list of
# the installed packages, reverse sorted by size.  Additional information
# about dependencies that each package fulfills will be printed out, and
# optionally the rpm description of the package.
#
# The motivation for this script was to look at systems intended to be
# put into a livecd format, and find the 'low hanging fruit' of packages
# that can be removed to achieve a smaller footprint.  In general the
# output produced could be useful for any sysadmin looking for the best
# way to free up disk space by removing unneeded packages.
#


#showdesc=0
showdesc=1

#rm -f tttmp.*

if [ ! -f tttmp.rpmqalast ]; then
    rpm -qa --last > tttmp.rpmqalast
fi

instpkgs=$( cat tttmp.rpmqalast | awk '{print $1}' )

if [ ! -f tttmp.unsortedpkgsizes ]; then
    for instpkg in $instpkgs; do
	pkgsize=$( rpm -q --queryformat="%{SIZE}" $instpkg )
	pkgsizemb=$( echo "$pkgsize / 1024 / 1024" | bc -l | sed -e 's/\(.*\...\).*/\1/' )
	echo "$pkgsizemb megs --- $instpkg" >> tttmp.unsortedpkgsizes
    done
fi

if [ ! -f tttmp.sortedpkgsizes ]; then
    cat tttmp.unsortedpkgsizes | sort -n -r > tttmp.sortedpkgsizes
fi

instpkgs=$( cat tttmp.sortedpkgsizes | awk '{print $4}' )
for instpkg in $instpkgs; do
    echo "========================================================================"
    pkgsize=$( rpm -q --queryformat="%{SIZE}" $instpkg )
    pkgsizemb=$( echo "$pkgsize / 1024 / 1024" | bc -l | sed -e 's/\(.*\...\).*/\1/' )
    echo -en "(${pkgsizemb}MB) "
    shortname=$( rpm -q --queryformat="%{NAME}" $instpkg )
    echo -en "$shortname"
    if $( rpm -e --test $instpkg > tttmp.rpmetestout 2>&1 ); then
	echo -en " is removable\n"
    else
	echo -en " is NOT removable\n"
	rm -f tttmp.deppkgsizes.unsorted
	rm -f tttmp.deppkgsizes.sorted
	cat tttmp.rpmetestout | grep "is needed by (installed)" | sed -e 's/.*\(installed\)..//' | sed -e 's/\(.*\)\..*$/\1/' | sort | uniq  > tttmp.dependencies
	deps=$( cat tttmp.dependencies )
	for dep in $deps; do
#	    echo "debug: checking dependency $dep"
	    pkgsize=$( rpm -q --queryformat="%{SIZE}" $dep )
	    pkgsizemb=$( echo "$pkgsize / 1024 / 1024" | bc -l | sed -e 's/\(.*\...\).*/\1/' )
	    echo "$pkgsizemb megs --- $dep" >> tttmp.deppkgsizes.unsorted
	done
	cat tttmp.deppkgsizes.unsorted | sort -n -r | uniq > tttmp.deppkgsizes.sorted
	numdeps=$( cat tttmp.deppkgsizes.sorted | wc -l )
	topthreedeps=$( cat tttmp.deppkgsizes.sorted | head -3 | awk '{print $4}' )
	echo "------------------------------------------------------------------------"
	echo "number of packages that are dependent: ${numdeps}"
	echo "biggest deps are "
	for bigdep in $topthreedeps; do
	    shortname=$( rpm -q --queryformat="%{NAME}" $bigdep )
	    echo -en "$shortname"
	    pkgsize=$( rpm -q --queryformat="%{SIZE}" $bigdep )
	    pkgsizemb=$( echo "$pkgsize / 1024 / 1024" | bc -l | sed -e 's/\(.*\...\).*/\1/' )
	    echo -en "(${pkgsizemb}MB) \n"
	done
	echo -en "\n"
    fi

    if (($showdesc)); then
	echo "------------------------------------------------------------------------"
	echo -en "package description: $instpkg \n\n"
	rpm -q --queryformat="%{DESCRIPTION}\n" $instpkg
    fi
    echo "========================================================================"
    echo -en "\n\n\n"
done