#!/bin/sh
#
# dupremover - Recursive Duplicate Remover
#
# (c) 2011 by Sebastian Felis
#
# You are free to use, modify, restribute without warranty while keeping it
# free of charge. It would be nice to get an feedback note if you are using it.
#
# History
# 2011-01-14 Initial Version
#
 
SCRIPT="$0"
SRC=
COPY=
LOGLEVEL=0 # 2 DEBUG, 1 VERBOSE, 0 INFO, -1 WARN, -2 ERR
DELETE=0
DELETE_HARDLINK=0
MOVE=0
MOVEDST=".dupremover"
TRY=0
 
help() {
  echo "$(basename "$SCRIPT") [OPTIONS] SRC COPY
 
Compares files of SRC directory with COPY directory recursivly via md5sum and
deletes or moves duplicated files of the COPY directory.
 
\t-d|--delete
\t\tDelete the copy file
\t-m|--move
\t\tMoves the copy file to SRC/$MOVEDST
\t-t|--try
\t\tTry run. Do not perform any action
\t-H|--delete-hardlink
\t\tDelete multiple hardlink file
\t-L|--log-level LEVEL
\t\tSet log level: 2: DEBUG, 1: VERBOSE, 0: INFO, -1: WARN, -2: ERR
"
}
 
while [ -n "$1" ]; do
  case "$1" in
    -d|--delete)
      DELETE=1
      ;;
    -m|--move)
      MOVE=1
      ;;
    -t|--try)
      TRY=1
      ;;
    -H|--delete-hardlink)
      DELETE_HARDLINK=1
      ;;
    -L|--log-level)
      shift
      LOGLEVEL=$1
      ;;
    -h|--help)
      help
      exit 0
      ;;
    *)
      if [ -z "$SRC" ]; then
        SRC=$1
      elif [ -z "$COPY" ]; then
        COPY=$1
      fi
      ;;
  esac
  shift
done
 
if [ -z "$COPY" ]; then
  help
  exit 1
fi
 
log() {
  if [ $1 -eq 2 ]; then
    LEVEL=DEBUG
  elif [ $1 -eq 1 ]; then
    LEVEL=VERBOSE
  elif [ $1 -eq 0 ]; then
    LEVEL=INFO
  elif [ $1 -eq -1 ]; then
    LEVEL=WARN
  elif [ $1 -eq -2 ]; then
    LEVEL=ERR
  else
    LEVEL=UNKNOWN
  fi
  if [ $1 -le $LOGLEVEL ]; then
    echo $LEVEL: $2
  fi
}
 
action() {
  local FILE="$1"
  local DIR="$2"
  if [ $MOVE -ne 0 ]; then
    if [ ! -f "$FILE" ]; then
      log 1 "Skip moving of directory $FILE"
      return 0
    fi
    local DST="$SRC/$MOVEDST$DIR"
    log 0 "Move duplicate $FILE to $DST"
    if [ $TRY -eq 0 ]; then
      mkdir -p "$DST"
      mv "$FILE" "$DST"
      if [ $? -ne 0 ]; then
        log -2 "Could not move $FILE to $DST"
      fi
    fi
  elif [ $DELETE -ne 0 ]; then
    log 0 "Delete duplicate $FILE"
    if [ $TRY -eq 0 ]; then
      rm -rf "$FILE" > /dev/null
      if [ $? -ne 0 ]; then
        log -2 "Could not delete $FILE"
      fi
    fi
  else
    log 0 "Do nothing for duplicate $FILE"
  fi
}
 
checkemptydir() {
  local DIR="$1"
  if [ ! -d "$DIR" ]; then
    log -1 "$DIR is not a directory"
    exit
  fi
 
  COUNT=$(ls -a "$DIR" | wc -l)
  if [ $COUNT -eq 2 ]; then
    action "$DIR"
  fi
}
 
compare() {
  local DIR="$1"
  if [ ! -d "$SRC$DIR" ]; then
    log -2 "Skip non directory $_DIR"
  fi
  ls -a "$SRC$DIR" | while read NAME; do
    if [ "$NAME" = "." -o "$NAME" = ".." ]; then
      continue
    fi
    SRCFILE="$SRC$DIR/$NAME"
    COPYFILE="$COPY$DIR/$NAME"
    if [ ! -e "$COPYFILE" ]; then
      log 1 "File $NAME only in source $SRC$DIR"
      continue
    fi
    if [ -d "$SRCFILE" ]; then
      log 1 "Read directory $SRCFILE"
      compare "$DIR/$NAME"
      checkemptydir "$COPYFILE"
      continue
    fi
    if [ $(stat -c%i "$SRCFILE") -eq $(stat -c%i "$COPYFILE") ]; then
      if [ $(ls -ld "$SRCFILE" | awk '{print $2}') -gt 1 -a $DELETE_HARDLINK -ne 0 ]; then
        log -1 "File $COPYFILE is duplicated hardlink"
      else
        log 0 "Skip file $DIR/$NAME with same inode. Use -H to delete hardlinks"
        continue
      fi
    fi
    SRCSIZE=$(stat -c%s "$SRCFILE")
    COPYSIZE=$(stat -c%s "$COPYFILE")
    if [ $SRCSIZE -ne $COPYSIZE ]; then
      log 2 "File size mismatch of {$SRC|$COPY}$DIR/$NAME $SRCSIZE != $COPYSIZE"
      continue
    fi
    SRCMD5=$(md5sum "$SRCFILE" | awk '{print $1}')
    COPYMD5=$(md5sum "$COPYFILE" | awk '{print $1}')
    if [ "$SRCMD5" != "$COPYMD5" ]; then
      log 2 "MD5 mismatch of {$SRC|$COPY}$DIR/$NAME"
      continue
    fi
    log 1 "{$SRC|$COPY}$DIR/$NAME:$SRCSIZE are equal"
    action "$COPYFILE" "$DIR"
  done
  # Get files only in the copy directory
  ls -a "$COPY$DIR" | while read NAME; do
    if [ "$NAME" = "." -o "$NAME" = ".." ]; then
      continue
    fi
    SRCFILE="$SRC$DIR/$NAME"
    COPYFILE="$COPY$DIR/$NAME"
    if [ ! -e "$SRCFILE" ]; then
      log 1 "File $NAME only in copy $COPY$DIR"
      continue
    fi
  done
}
 
compare