#!/usr/bin/perl -w
#
# Script:	filestats
#
# Description:	Print stats on files in a tree, inc. average size,
#				distribution and counts for file types
#				(Note: undoubtably not the most elegant solution)
#
# Author:	Ade Rixon, ar@rsi.co.uk
#
# History:	v0.2,	24/9/98
#

use File::Basename;
use File::Find;

$idxcnt = 0;
$bytecount = 0;

usage() if (@ARGV == 0);

find(\&wanted, @ARGV) or usage();	# process the dirs

# output stats
$mean = int($bytecount/$filecount);
print "Average file size is: $bytecount/$filecount = ", $mean, " bytes\n";
print "\nDistribution of file sizes:\n";
print "Size (K)\tNo. of files\n";
for ($i=0; $i <= scalar(@filesizes); $i++)
{
	if ($filesizes[$i])
	{
		printf "%8d - %d\t%12d\n", $i*10, $i*10+9, $filesizes[$i];
	}
}
print "\nFile report by type (extension):\n";
print "Type\tNo. of files\tSize (bytes)\n";
$extsize = 0;
foreach $ext (sort keys %extidx)
{
	printf "%-8s%8d\t%s\n", $ext, $extfiles[$extidx{$ext}], $extbytes[$extidx{$ext}];
	$extcnt += $extfiles[$extidx{$ext}];
	$extsize += $extbytes[$extidx{$ext}];
}
printf "\nTotals\t%8d\t%s\n", $extcnt, $extsize;

exit;

sub usage {
	die("Usage: $0 <directories to search>\n");
}

# subroutine for find
sub wanted
{
	my $file = $_;
	
	return 0 if (! -f $file);

	#($dev,$ino,$mode,$nlink,$uid,$gid,$rdev,$size) = stat($file);
	$size = -s $file;
	# update total bytes, no. of files and array of sizes in K
	$bytecount += $size;
	$filecount++;
	$filesizes[int($size/10240)]++;
	
	# get filename extension
	$ext = (fileparse($file, '\.\w\w?\w?\w?$'))[2];
	$ext = 'other' if ($ext eq "");	# default
	$ext = lc($ext);	# ignore case
	# %extidx gives indexes into array for each file ext
	# @extfiles, @extbytes contain file and byte count for each ext
	if (!defined($extidx{$ext}))
	{
		# initialise new array element
		$extidx{$ext} = $idxcnt++;	# index
		$extbytes[$extidx{$ext}] = 0;	# byte count
	}
	$extbytes[$extidx{$ext}] += $size;
	$extfiles[$extidx{$ext}]++;
}