#!/bin/bash
#
#   Compare the content of a local and a remote directories
#
# V0.1 - 2014-11-23
#    Missing: SSH port management

#-------------------------------------------------------------
# Define usage
#-------------------------------------------------------------

usage() {
    cat <<-'_usage_statement_'

compareDirs.sh compares the top level content of 2 local or remote directories using du.
    
Usage: compareDirs.sh [[user@]host1:]dir1 [[user@]host2:]dir2
    
Note: SSH port management is missing.

_usage_statement_
}

#-------------------------------------------------------------
# Analyse les arguments de la ligne de commande.
#-------------------------------------------------------------

echo -e "\n$(basename $0) $@ \n"

[[ $# != 2 ]] && {
   usage
   exit -1
}

function getDirFromParam {
   idx=`expr index "$1" ':' `
   if [ $idx == 0 ]
   then
      echo $1
   else
      echo ${1:$idx}
   fi
}

function getHostFromParam {
   idx=`expr index "$1" ':' `
   if [ $idx != 0 ]
   then
      ((idx--))
      echo ${1:0:$idx}
   fi
}

function checkHostAndDir {
   dir=$1
   host=$2
   if [ $host ]
   then
      # echo "$host - $dir --> REMOTE"
      if ( ! ssh $host "pwd" > /dev/null 2>&1 )
      then
         echo  "ERROR: $host unreachable!"
         return  1
      elif ( ! ssh $host "cd $dir" > /dev/null 2>&1 )
      then
         echo  "ERROR: $dir unreachable on $host!" 
         return  2
      fi
   else 
      # echo "$dir  --> Local"
      [ -d "${dir}" ] || {
         echo -e "Error $dir must be a directory!"  
         return  3
      }
   fi
   return  0
}

dir1=$(getDirFromParam $1)
host1=$(getHostFromParam $1)
dir2=$(getDirFromParam $2)
host2=$(getHostFromParam $2)

# echo -e "dir1 : $dir1 - host1 : $host1\ndir2 : $dir2 - host2 : $host2"

checkHostAndDir "$dir1" $host1
[ $? -eq 0 ] || exit -1
checkHostAndDir "$dir2" $host2
[ $? -eq 0 ] || exit -1

#-------------------------------------------------------------
# VARS
#-------------------------------------------------------------

tmp_report_1="/tmp/`basename $0`-local-`date +%F-%T`.log"
tmp_report_2="/tmp/`basename $0`-remote-`date +%F-%T`.log"

#-------------------------------------------------------------
# Recupération sur erreur
#-------------------------------------------------------------
function traitement_erreur {
   echo  
   echo  "---------------------"
   echo  ">>> Erreur $? >>>>"
   echo  "---------------------"
   echo  
   rm -f $tmp_report_1 $tmp_report_2
   exit 1
}
trap traitement_erreur SIGHUP SIGINT SIGTERM SIGKILL

#-------------------------------------------------------------
# fonction de bilan de synchro
# déclarée ici pour pouvoir être utilisée par le traitement d'erreur qui suit
#-------------------------------------------------------------

function local_report {
   function numFormat { 
      echo $1  | sed ':a;s/\B[0-9]\{3\}\>/.&/;ta' 
   }
	function stats_folder() {
	  # Attention : le nom du dossier peut contenir des blancs
	  # il faut donc le traiter entre ""
	  local folder="$1"
	  local items=`find "$folder" -iname '*' | wc -l`
	  local taille=0
	  echo  -e "   $folder" 
	  # echo  -e "   $(basename $folder)" 
	  if [ $items -gt 0 ]; then
		 taille=$(du -bc "$folder" | grep total | sed -e "s/[ \t]*total/ /g")
       taille_f=$(numFormat $taille)
       items_f=$(numFormat $items)
		 echo -e "      $items_f items, $taille_f octets" 
	  else 
		 echo vide
	  fi
	}
   local curdir=$(pwd)
   local dir="$(readlink -f "$1")"
	echo -e "LOCAL report:"
   echo -e "$dir"
   cd "$dir"
	for f in * ; do
		# Attention : le nom du dossier peut contenir des blancs
		# il faut donc le passer entre ""
      #echo "---f: $f"
		[ -d "$f" ] && stats_folder "$f"
	done
	echo -e "Total: " 
	stats_folder "$dir"
   # echo "--- pwd: $(pwd) "
   cd "$curdir"
	echo
}

function remote_report {
   local dir="$1"
   local host="$2"
   ssh $host "
      function numFormat { 
         echo \$1  | sed ':a;s/\\B[0-9]\\{3\}\\>/.&/;ta' 
      }
      function stats_folder() {
         # Attention : le nom du dossier peut contenir des blancs
         # il faut donc le traiter entre \"\"
         local folder=\"\$1\"
         local items=\`find \"\$folder\" -iname '*' | wc -l\`
	      local taille=0 
         echo  -e \"   \$folder\" 
         if test \$items -ne 0; then
		      taille=\$(du -bc \"\$folder\" | grep total | sed -e \"s/[ \t]*total/ /g\")
            taille_f=\$(numFormat \$taille)
            items_f=\$(numFormat \$items)
            echo -e \"      \$items_f items, \$taille_f octets\" 
         else 
            echo vide
         fi
      }

      dir=$(readlink -f "$dir")
      echo \"REMOTE report: \"
      echo $dir
      cd $dir
      for folder in * ; do
         # Attention : le nom du dossier peut contenir des blancs
         # il faut donc le passer entre \"\"
         [ -d \"\$folder\" ] && stats_folder \"\$folder\"
      done
      echo -e \"Total: \" 
      stats_folder \"$dir\"
      echo
   "
}
echo

# set -x

if [ $host1 ]
then
   remote_report "$dir1" $host1 >> $tmp_report_1
else 
   local_report "$dir1" >> $tmp_report_1
fi

if [ $host2 ]
then
   remote_report "$dir2" $host2 >> $tmp_report_2
else 
   local_report "$dir2" >> $tmp_report_2
fi

sdiff -atw 140 $tmp_report_1 $tmp_report_2
rm -f $tmp_report_1 $tmp_report_2