#!/bin/sh -
#
#   make-hardlinks
#
#   Find files with identical contents (MD5-based) and save space by
#   re-creating them as hardlinks.  Tested and works on FreeBSD.
#
#   Copyright (C) 2005-2009 Oliver Fromme <olli@fromme.com> <olli@secnetix.de>
#   All rights reserved. Standard 2-clause BSD license and disclaimer applies.
#

for i in "$@"; do
	if [ ! -d "$i" ]; then
		echo "${0##*/}: ${i}: Not a directory." >&2
		echo "Usage:  ${0##*/} [<directory> ...]" >&2
		echo "If no directory is specified, start scanning in the current directory." >&2
		exit 2
	fi
done

if [ $# -lt 1 ]; then
	set .
fi

find -- "$@" -type f -not -empty -print0 | xargs -0 md5 -r \
| awk '
	BEGIN {
		a = "'\''"
		n = 0
	}

	{
		if (h[$1]) {
			x1 = h[$1]
			x2 = $2
			gsub(/'\''/, a "\\" a a, x1)
			gsub(/'\''/, a "\\" a a, x2)
			cmd[n] = sprintf("ln -f '\''%s'\'' '\''%s'\''", x1, x2)
			n++
		}
		else
			h[$1] = $2
	}

	END {
		k = 0
		for (i = 0; i < n; i++)
			if (system(cmd[i]) == 0)
				k++
			else {
				print "System command failed:" > "/dev/stderr"
				print cmd[i] > "/dev/stderr"
			}

		print k, "links created." > "/dev/stderr"
		if (k < n) {
			print n - k, "ERRORS." > "/dev/stderr"
			exit 1
		}
	}
'

#--