#!/usr/bin/perl -w

#By Neil Watson on 08/18/01 (watson-wilson.ca)
#Rev 1.3 11/07/01

#Usage filehog <directory>
#options:
# -p = report top 10 pst files

use strict;
use warnings;
use File::Find;
use CGI qw(:standard);

#change this for the locale of final report:
my $repdir = "/var/log";

my (@field, @dupes);
my $dir = $ARGV[0];
my $pst = "";
my $args = join "", @ARGV;
if ($args =~ m/-p/) { $pst = "yes" }
my $host = `hostname`;
chomp($host); 
my @date = split(/\s+/, `date`);
chomp($date[5]);

#input
open(RAW, ">/tmp/filehog") || die "Could not open /tmp/filehog";

#calls sub wanted
find(\&wanted, $dir);

close(RAW);

#output
open(HTML, ">$repdir/hogs_$host.html") || die "Could not open $repdir/hogs_$host.html";
print HTML start_html("Space Hogs on server $host for $date[0], $date[1] $date[2] $date[5]");
print HTML h1("Space Hogs on server <i>$host</i> for $date[0], $date[1] $date[2] $date[5]")."\n\n";
print HTML h3("These people are shamefully using more than their fair share of disk space.  Hopefully, under this public scrutiny, they will change their greedy ways.");

print HTML "\n<hr>";
print HTML "\n<table>\n";
print HTML "<tr>\n";
print HTML "\t<th>Size (MB)<th>\n";
print HTML "\t<th>File Name and Location<th>\n"; 
print HTML "\t<th>File Owner<th>\n";
print HTML "<tr>\n";

#sort RAW data
if ($pst eq "yes"){
	system "sort -nr /tmp/filehog|grep -iv .pst|head>/tmp/filehog_sort";
}
else{
	system "sort -nr /tmp/filehog|head>/tmp/filehog_sort";
}

#write final report
open(RAW, "/tmp/filehog_sort") || die "Could not open /tmp/filehog_sort";
while (<RAW>) {
	chomp;
	s/^\s*//;
	@field = split(/,/, $_);
	$field[2] = `grep x:$field[2] /etc/passwd|cut -d: -f1`;
	chomp($field[2]);
	
	print HTML "<tr>\n";
	print HTML "\t<td>$field[0]<td>\n";
	print HTML "\t<td>$field[1]<td>\n";
	print HTML "\t<td>$field[2]<td>\n";
	print HTML "<tr>\n";	
}
close(RAW);

print HTML "</table>";

#call pst report
if ($pst eq "yes") { pst() }
 
print HTML end_html();
close(HTML);

system "rm -f /tmp/filehog /tmp/filehog_sort"; #remove scratch files



#subroutines:

sub wanted { # collects size, name and uid of all files
	my $uid = (lstat($_))[4];
	my $size = (lstat($_))[7];
	$size = $size/1000000;
	
	printf RAW "%5.0f", $size;
	print RAW ",$File::Find::name,$uid\n";

}


sub pst { #generates report on top 10 pst files
	system "egrep -i \.PST, /tmp/filehog|sort -nr|head>/tmp/psthog_sort";
	
	print HTML "<p>\n<hr>\n<p>\n";

	print HTML h3("Another space issue is caused by the proliferation of PST files.  If any of these files are yours and you are not actively using them please contact <a href=\"mailto:support\@yourdomain.com\">Support</a> to have them burned onto a CDROM");

	print HTML "\n<hr>";
	print HTML "\n<table>\n";
	print HTML "<tr>\n";
	print HTML "\t<th>Size (MB)<th>\n";
	print HTML "\t<th>File Name and Location<th>\n";
	print HTML "\t<th>File Owner<th>\n";
	print HTML "<tr>\n";

	#write pst file report
	open(RAW, "/tmp/psthog_sort") or die "Could not open /tmp/psthog_sort";
	while (<RAW>) {
	chomp;
        	s/^\s*//;
	        @field = split(/,/, $_);
        	$field[2] = `grep x:$field[2] /etc/passwd|cut -d: -f1`;
	        chomp($field[2]);

	        print HTML "<tr>\n";
        	print HTML "\t<td>$field[0]<td>\n";
	        print HTML "\t<td>$field[1]<td>\n";
        	print HTML "\t<td>$field[2]<td>\n";
	        print HTML "<tr>\n";
	}
	close(RAW);

	print HTML "</table>";
}