#!/bin/sh

# Copyright (C) 2010, Jeremy Laidman
# under the GNU General Public License.
# For details: http://www.gnu.org/licenses/gpl.txt

# Version 0.2 - 23 Sep 2010

# look for files with the same name and content (verified using md5 checksum)
# to a target directory and replace the file with a hardlink to the
# target file

# give me source and target directory roots
# roots must be corresponding, so that
# files are in the same place, otherwise they
# won't be found

DEBUG=0
VERBOSE=0

die() {
	echo "$@" >&2
	exit 1
}

same_file() {
	# we do a sum first, cos it's quicker
	# if the same, we do an md5sum to be sure
	S1=`sum "$1"`
	S2=`sum "$2"`
	[ 0$DEBUG -gt 2 ] && echo "Sums: $S1 $S2" >&2
	if [ "$S1" = "$S2" ]; then
		S1=`md5sum "$1" | { read A B; echo $A; }`
		S2=`md5sum "$2" | { read A B; echo $A; }`
		[ 0$DEBUG -gt 2 ] && echo "Sums: $S1 $S2" >&2
		if [ "$S1" = "$S2" ]; then
			[ 0$DEBUG -gt 2 ] && echo "Same file" >&2
			return 0
		fi
	fi
	return 1
}

same_inode() {
	# we compare the inode numbers of the two files
	# and return true if they are the same
	I1=`ls -i "$1" | { read A B; echo $A; }`
	I2=`ls -i "$2" | { read A B; echo $A; }`
	[ 0$DEBUG -gt 2 ] && echo "Inodes: $I1 $I2" >&2
	[ "$I1" -a "$I2" ] || die "Unable to find inodes for '$1' and '$2'"
	if [ "$I1" = "$I2" ]; then
		[ 0$DEBUG -gt 2 ] && echo "Same inode" >&2
		return 0
	fi
	return 1
}

linkify_it() {
	F1="$1"	# source
	F2="$2"	# destination
	TF=`dirname "$F1"`/.linkify.`basename "$F1"`.$$
	mv "$F1" "$TF"		|| die "Unable to rename file, bailing: '$F1'"
	ln "$F2" "$F1"		|| die "Unable to create symlink, bailing: '$F1'"
	touch --reference="$F2" "$TF"	|| die "Unable to set timestamp on new directory entry, bailing: '$F2'"
	chmod --reference="$F2" "$TF"	|| die "Unable to set permissions on new directory entry, bailing: '$F2'"
	chown --reference="$F2" "$TF"	|| die "Unable to set ownership on new directory entry, bailing: '$F2'"
	rm -f "$TF"		|| die "Unable to clean up temp file, bailing: '$T1'"
}

usage() {
	printf "Usage: $0 [ -d | -h ] source-path dest-path\n\t-d = debug\n\t-h = help\n"
}

# get switches
while true; do
	case "$1" in
		-d|--debug)	DEBUG=`expr 0$DEBUG + 1`; shift;;
		-v|--verbose)	VERBOSE=`expr 0$VERBOSE + 1`; shift;;
		-h|--help)	usage; exit 0;;
		-*)		die "Invalid switch: $1";;
		*)		break;;
	esac
done

# now we should only have the two paths left
[ "$2" ] || die "Specify source and target directory"
[ "$3" ] && die "Specify source and target directory"

SRCDIR="$1"
DSTDIR="$2"

[ 0$DEBUG -gt 0 ] && VERBOSE=1	# debug implies verbose
[ 0$DEBUG -gt 0 ] && echo "DEBUG is $DEBUG, SRC=$SRCDIR, DST=$DSTDIR" >&2

[ -d "$SRCDIR" ] || die "Invalid source directory $SRCDIR"
[ -d "$DSTDIR" ] || die "Invalid target directory $DSTDIR"
( cd "$SRCDIR" ) || die "Unable to cd to source directory"

FILECOUNT=0
LINKCOUNT=0

[ 0$VERBOSE -gt 0 ] && echo "Linkifying '$SRCDIR' using '$DSTDIR'"
( cd "$DSTDIR"; find . -type f ) | while read FILE; do
	[ 0$VERBOSE -gt 0 ] && printf "$LINKCOUNT/$FILECOUNT\r"
	FILECOUNT=`expr $FILECOUNT + 1`
	[ 0$DEBUG -gt 2 ] && echo "Maybe $FILE" >&2
	if [ -f "$SRCDIR/$FILE" ]; then
		[ 0$DEBUG -gt 1 ] && echo "Candidate: $FILE" >&2
		if same_inode "$DSTDIR/$FILE" "$SRCDIR/$FILE"; then
			[ 0$DEBUG -gt 0 ] && echo "Same inode, already been linkified: '$FILE'" >&2
		else
			[ 0$DEBUG -gt 1 ] && echo "Different inode detected." >&2
			if same_file "$DSTDIR/$FILE" "$SRCDIR/$FILE"; then
				[ 0$DEBUG -gt 0 ] && echo "Same file detected with different inodes" >&2
				[ 0$VERBOSE -gt 0 ] && echo "'$FILE'..." >&2
				linkify_it "$SRCDIR/$FILE" "$DSTDIR/$FILE"
				LINKCOUNT=`expr $LINKCOUNT + 1`
			fi
		fi
	fi
done
[ 0$VERBOSE -gt 0 ] && printf "\n"	# newline after ticker