#!/bin/bash
#
# $Id: find-upgrade-orphans.sh,v 1.4 2005/01/04 14:00:56 peterw Exp peterw $
#
# find-upgrade-orphans.sh
#
# script to find files that should be saved before re-installing linux
#
# Copyright (c) 2005 Peter Watkins, peterw@tux.org
#
# licensed under the terms of the GNU General Public License

required_apps="rpm awk cat date df echo expr find grep id mkdir od printf sed sort uniq tar wc which"

app_available() {
  # takes a single argument, an app name, and tests for its availabilty
  [ -z $(which $1 2>/dev/null) ] && return 1
  # else we found a copy
  return 0
}

hexname() {
  # use 'od' to turn a string like "/boot" into a hex string like
  # "622f6f6f00745" so that we can store per-partition files without
  # worrying about the "/" characters
  in="$1"
  raw=`printf '%s' "$in" | od -x | sed 's:^0*::;s:[ 	]::g'`
  out=""
  for bit in $raw; do
    out="${out}${bit}"
  done
  echo $out
}

print_info() {
  echo `date +%k:%M:%S` "$1"
}

# make sure we have the apps we need
missing_apps=""
for app in $required_apps; do
  app_available "$app" || missing_apps="${missing_apps} $app"
done
if [ "$missing_apps" != "" ]; then
  echo "ERROR: The following utilities are required but not on this system: $missing_apps" >/dev/stderr
  exit 4
fi
unset missing_apps

partitions_keep=""
partitions_reformat=""
partitions_all=`df | awk '{print $5}' | grep '^/'`

uid=`id -u`
if [ "$uid" != "0" ]; then
  echo "ERROR: This tool should be run by root so all filesystem assets are visible!" >/dev/stderr
  exit 3
fi
unset uid

# package test
pkgs=`rpm -qa 2>/dev/null`
if [ $? -ne 0 -o "$pkgs" = "" ]; then
  echo "ERROR: no packages found! This script currently only supports RPM-based systems" >/dev/stderr
  exit 5
fi
unset pkgs

echo
echo "Please indicate whether each of the currently mounted"
echo "partitions will be reformatted or not when you upgrade"
echo

# ask the user about each partition
for p in $partitions_all; do
  ans=""
  while [ "$ans" = "" ]; do
    printf '%20s: will this be reformatted (Y|N)? ' $p
    read ans
    if [ "$ans" = "Y" ]; then
      partitions_reformat="$partitions_reformat	$p"
    else 
      if [ "$ans" = "N" ]; then
        partitions_keep="$partitions_keep	$p"
      else
        ans=""
      fi
    fi
  done
  unset ans
done

# confirm their response
echo
echo "This tool will look for files that should be preserved on"
echo "the following partitions:"
echo "    $partitions_reformat"
echo
echo "The following partitions will be ignored by this tool:"
echo "    $partitions_keep"
echo

ans=""
while [ "$ans" = "" ]; do
  printf "Is that alright (Y|N)? "
  read ans
  if [ "$ans" = "N" ]; then
    echo "Please re-run this tool to correct this."
    exit 1
  else
    if [ "$ans" != "Y" ]; then
      ans=""
    fi
  fi
done
unset ans

echo

# make a tmpdir
made=0
tries=0
while [ $made -eq 0 -a $tries -lt 10 ]; do
  r=$RANDOM
  basedir="${TMPDIR:-/tmp}"
  tmpdir="${basedir}/dist-upgrade.$r"
  mkdir -m0700 $tmpdir
  if [ $? -eq 0 -a -d $tmpdir ]; then
    made=1
  fi
  tries=`expr $tries + 1`
done
if [ $made -ne 1 ]; then
  echo "ERROR: unable to make a safe work directory in \"$basedir\"" >/dev/stderr
  exit 2
fi
unset made
unset tries

# change to the root of the filesystem
cd /

print_info "finding all the files that belong to packages"
rpm -qal | grep '^/' | sort | uniq > $tmpdir/packaged-files

print_info "finding all the packages"
rpm -qa --queryformat '%{NAME}\n' | sort | uniq > $tmpdir/packages

print_info "finding all the package files that have been modified"
for pkg in `cat $tmpdir/packages`; do
  # note the package to help troubleshoot erroneous reports
  echo "${pkg}:" >> $tmpdir/rpm-verify-out
  # --nomtime: we don't care if the file's been 'touched' so
  # long as it has not been modified; I have observed files 
  # owned by multiple packages frequently fail the mtime test
  rpm --verify --nomtime $pkg >> $tmpdir/rpm-verify-out
done
sed 's:^[^/]*::' < $tmpdir/rpm-verify-out | grep '^/' | sort | uniq > $tmpdir/modified-files

for p in $partitions_reformat; do
  # get a version of the pathname with no slashes (ugly!)
  pname=`hexname $p`
  printf '%30s\t%s\n' "$p" $pname >> $tmpdir/pnames

  print_info "finding all the files on doomed $p partition"
  # note: this also finds devices, pipes, links, sockets, and doors
  find $p -xdev \! -type d -print > $tmpdir/$pname.all-files

  print_info "finding the files on doomed $p that are not owned by packages"
  cat $tmpdir/$pname.all-files $tmpdir/$pname.all-files $tmpdir/packaged-files | sort | uniq -c | awk ' $1 == 2 { print $2 }' > $tmpdir/$pname.orphans

  howmany=`wc -l < $tmpdir/$pname.orphans`
  total=`wc -l < $tmpdir/$pname.all-files`
  print_info "$p has $howmany orphans out of $total files"
  unset pname
done

print_info "finding the files we need to save"
# loop through these items, as some may be directories (RPM verification)
for item in `cat $tmpdir/*.orphans $tmpdir/modified-files | sort | uniq`; do
  # note: this also finds devices, pipes, links, sockets, and doors
  if [ \! -d "$item" ]; then
    echo "$item" >> $tmpdir/keep
  fi
done

print_info "calculating size of files we need to save (this could take some time)"
howmany=`wc -l < $tmpdir/keep`
howbig=`tar czf - -T $tmpdir/keep 2>/dev/null| wc -c`
howbigmb=`expr $howbig / 1024 / 1024`
echo
echo "$tmpdir/keep contains names of the $howmany files ($howbigmb MB compressed) you likely want to keep"
echo

for p in $partitions_keep; do
  freekb=`df -k "$p" | grep "$p" | awk '{print $3}'`
  freemb=`expr $freekb / 1024`
  if [ $freemb -gt $howbigmb ]; then
    echo "$p has $freemb MB free and could hold the backed-up files"
    if [ "$p" != "/tmp" ]; then
      goodpath="${p}/"
    fi
  else
    echo "$p does not have enough space for the backed-up files"
  fi
  if [ "$p" = "/tmp" ]; then
    echo "WARNING: /tmp is not recommended for important files. Some distributions flush /tmp"
    echo "         when the system boots; others purge it periodcally."
  fi
  unset freekb
  unset freemb
done

echo
echo "To back up the files, use a command like this:"
timestamp=`date +%Y%m%d%H%M`
echo "  umask 077 && tar czf ${goodpath}backups-${timestamp}.tar.gz -T $tmpdir/keep"
echo
unset timestamp
unset goodpath

print_info "All done."

exit 0