#!/usr/bin/perl
#
# Histogram generator
# (c) 1996 Heiko W.Rupp <hwr@pilhuhn.de>
#
#
# run histogram.pl < size-file
#
# size-file is generated with a INN feed like
#
#
# size-all:*:Tf,WB:
#



while(<>) {
	chop;
	$sum=$sum+$_;
	$count++;
	$bytes=int($_);
	if ($bytes < 5001) {
		$ro=int($bytes/50)*50;
		$lt5++;
	} elsif ($bytes >5000 && $bytes < 15001) {
		$ro=int($bytes/500)*500;
	} else {
		$ro=int($bytes/5000)*5000;
	}
	$rou=sprintf("%6.0d",$ro);
	$bucks{$rou}++;
	$lt9++ if ($bytes < 9001 );
	$lt64++ if ($bytes < 64001 );

}
print "Total: ",$sum," Bytes\n";
print "Count: ",$count," Articles\n";
print $lt5," Articles with less than 5001 bytes, ";
printf "%2.2f %%\n",100*($lt5/$count);
print $lt9," Articles with less than 9001 bytes, ";
printf "%2.2f %%\n",100*($lt9/$count);
print $lt64," Articles with less than 64001 bytes, ";
printf "%2.2f %%\n",100*($lt64/$count);
print "Average: ",int($sum/$count)," Bytes/Article\n";
print "Histogram is:\n";

$maxi=0;
foreach $b (%bucks) {
	$by=$bucks{$b};
	$maxi=$by if $by > $maxi;
}

print "\n";

$maxl=$maxi/40.0;

printf "(Each + represents at least %2.2f Articles)\n",$maxl;
print "Actual sizes are round down \n\n";

foreach $b (sort %bucks) {
	next if $bucks{$b} eq "";
	print "Size: ",$b," Articles: ";
	printf "%7.0d ",$bucks{$b};
	print "+" x int($bucks{$b}/$maxl);
	print "\n";
	
}