#!/bin/bash
CONFIG_FILE=machines.conf
MACHINE_PATTERN=machine=([^@]+@[^:]+\|local):.+
# example 1: machine=username1@pcname1:basedir1
# example 2: machine=local:basedir1

function get_data {
  old_IFS="$IFS"; IFS="=:@"
  index=0
  for arg in $1; do
    case $index in
      0) ;;           # "machine"
      1) user=$arg;;  # "username1"
      2) host=$arg;;  # "pcname1"
      3) dir=$arg;;   # "basedir1"
    esac
    let index++
  done
  IFS="$old_IFS"
  if [ $index -eq 3 ]; then  # local
    dir=$host; unset user host
  fi

  if test -n "$user"; then
    # Remote!
    tmpfile=.tmpfile.txt
    scp -q $0 ${user}@${host}:/tmp/
    ssh ${user}@${host} chmod a+x /tmp/$0
    echo machine=local:${dir} | ssh ${user}@${host} "cat > /tmp/machines.conf"
    ssh ${user}@${host} "cd /tmp/; /tmp/$0 -r"  # Parameter -r !
    remote_files=$( ssh ${user}@${host} "cd /tmp/; ls -tr1 *.data *.sums | tail -2" )
    ## echo DEBUG remote_files=$remote_files
    for file in $remote_files; do
      scp -q ${user}@${host}:/tmp/${file} $tmpfile
      mv $tmpfile "${user}@${host}:${file}"
    done
    return
  fi

  filename_prefix="${dir//\//:}"
  current_dir="$PWD"
  cd "$dir"
  find . -type f | cut -c3- > ${current_dir}/${filename_prefix}.files
  cd "$current_dir"; rm -f ${filename_prefix}.tmpdata
  while read file; do
    sum=$( md5sum "$dir/$file" | cut -f1 -d" " )
    date=$( stat -c '%y' "$dir/$file" | cut -c1-16 | tr ' :' '-.' )
    size=$( stat -c '%s' "$dir/$file" )
#    date=$( stat -f "%Sm" "$dir/$file" | tr " " "-" )    # OS X
#    size=$( stat -f "%z" "$dir/$file" )     # OS X
    echo "$file!$sum!$date!$size" >> ${filename_prefix}.tmpdata
  done < ${filename_prefix}.files
  
  sort ${filename_prefix}.tmpdata > ${filename_prefix}.data
  cut -f1,2 -d! < ${filename_prefix}.data > ${filename_prefix}.sums
  rm ${filename_prefix}.{files,tmpdata}
}


# Hauptprogramm
if [ ! $BASH_VERSION \> 4 ]; then
  echo "Brauche Bash >= 4.0"
  exit
fi

REMOTE_EXEC=$1  # Hier steht bei Remote-Ausfuehrung "-r" drin

i=0
while read; do
  if [[ $REPLY =~ $MACHINE_PATTERN ]]; then
    machine[$i]="$REPLY"; 
    let i++
  fi
done < $CONFIG_FILE
machines=$i

# Daten der verschiedenen Ordner sammeln
for ((i=0; i<$machines; i++)); do 
  get_data ${machine[i]}
done

if [ "$REMOTE_EXEC" = "-r" ]; then
  exit  # Auswertung nur auf lokaler Maschine
fi

# Daten auswerten
for (( i=0; i<$machines; i++)); do
  echo "pc$i = ${machine[i]}"
done
echo

declare -A hashes dates sizes count # Hash-Datenstrukturen
for ((i=0; i<$machines; i++)); do 
  # Eine Datei auslesen
  datafile=$( echo ${machine[i]} | 
              sed -e s/^machine=// -e s/^local:// -e s/$/.data/ | \
              tr "/" ":" )

  while read; do
    line=$REPLY
    filename=${line%%\!*}; tmp_rest=${line#*\!}
    hash=${tmp_rest%%\!*}; tmp_rest=${tmp_rest#*\!}
    date=${tmp_rest%%\!*}; size=${tmp_rest#*\!}

    if [ "${hashes[$filename]}" = "" ]; then
      # Datei taucht zum 1. Mal auf
      ## echo "DEBUG: Erstes Auftreten von $filename in Ordner $i"
      extra=""
      num=1
      for ((j=0; j<$i; j++)); do extra=$extra"\n"; ((num++)); done
      hashes[$filename]=$extra$hash
      dates[$filename]=$extra$date
      sizes[$filename]=$extra$size
      count[$filename]=$num
    else
      # Datei war schon da
      hashes[$filename]=${hashes[$filename]}"\n"$hash # append
      dates[$filename]=${dates[$filename]}"\n"$date   # append
      sizes[$filename]=${sizes[$filename]}"\n"$size   # append
      (( count[$filename]++ ))
    fi
  done < $datafile

  # ggf. Leerzeilen anhängen
  if [ $i -gt 0 ]; then
    for filename in "${!hashes[@]}"; do
      if [ ${count[$filename]} -lt $[i+1] ]; then
        ## echo DEBUG: i=$i, $filename, count[$filename]=${count[$filename]}, muss verlaengern
        hashes[$filename]=${hashes[$filename]}"\n"
        dates[$filename]=${dates[$filename]}"\n"
        sizes[$filename]=${sizes[$filename]}"\n"
        (( count[$filename]++ ))
      fi
    done

  
  fi
done

# Abweichungen suchen
declare -a h d s  # Arrays
treffer=0
for filename in "${!hashes[@]}"; do
  # Anzahl der verschiedenen Hash-Eintraege (bei Gleichheit: 1)
  
  #echo DEBUG $filename
  #echo ----START
  #echo -e ${hashes[$filename]}
  #echo -e ${sizes[$filename]}
  #echo -e ${dates[$filename]}
  #echo ----END
  
  num=$( echo -e ${hashes[$filename]} | sed 's/^$/MISSING/' | sort -u | wc -l )
  if [ $num \> 1 ]; then
    ((treffer++))
    printf "[%2d] %s:\n" $treffer $filename
    {
      for ((i=0; i<$machines; i++)); do echo $treffer; done
      for ((i=0; i<$machines; i++)); do echo "pc$i:"; done
      echo -e ${sizes[$filename]}  | sed 's/^$/<missing>/'
      echo -e ${dates[$filename]}  | sed 's/^$/-/'
      echo -e ${hashes[$filename]} | sed 's/^$/-/'
    } | pr -5 -t -s" " | xargs printf "[%2d] %s %10s  %-17s %s\n" 
  fi
done

rm *.data *.sums