#! /bin/sh

# find-eq-files -- find files that are equal
# $Id$ 
# Carlos Duarte, 990131/990216

# usage: find-eq-files [dirs] [find options]
# example: find-eq-files ~/tmp ~/sources 

dirs="$*"
test $# -eq 0 && dirs=.

p="`echo $PATH | tr : ' '`"

# finds the best file hasher program available
# output of these, must be: hash1 [hash2 ...] filename
# hash my be anything that caracterizes the file contents (length, 
# signature, etc)
# 
# syntax of list is: 
# cmd [args |script_to_change_output_to: hash1 ... filename]
sed '/^#/d'<<'eof'|
du -ab 
ls -ld |awk '{ print $(NF-4), $NF}'
md5sum 
md5 |sed 's/^.*(\([^)]*\)). \(.*\)$/\2 \1/'
cksum 
sum 
wc -c
eof
while read cmd args; do
	for d in $p; do
		if test -x $d/$cmd; then 
			set -- $args; args=
			while test $# -ne 0; do 
				case "$1" in 
				"|"*) break ;; 
				*) args="$args $1"; shift ;; 
				esac
			done
			proc=cat
			test $# -ne 0 && proc="`echo \"$*\"|cut -c2-`"
			find $dirs -type f -print |
			xargs $cmd $args |
			eval $proc
			break 2 
		fi
	done
done |
awk '{ 
	filename = $NF
	hash = ""
	for (i=1; i<NF; i++)
		hash = hash "_" $i
	x[hash] = x[hash] filename "-\n-"
}

END {
	for (i in x) {
		n = split(x[i], a, /-\n-/)
		if (n == 2)
			continue
		while (--n) {
			printf "%s ", a[n]
		}
		print ""
	}
}' | 
while read x; do
	set -- $x
	while test $# -ne 0; do 
		x=
		first=$1; shift
		echo ============
		echo $first
		for other
		do
			if cmp $first $other > /dev/null; then
				echo $other
			else
				x="$x $other"
			fi
		done
		set -- $x
	done
done |
sed '
/^===*$/!b
:b
N
$d
N
/^===*\n.*\n===*$/{
	s/^.*\n//
	bb
}'
exit