##!/usr/bin/perl

# John Hobbs 2007 - www.velvetcache.org

$file = $ARGV[0];
open(INFO, "< ".$file);
@lines = <INFO>;
close(INFO);

$size = @lines;
$counter = 0;
$size = $size / 50;

open(FILEWRITE, "> NEW".$file);
foreach(@lines) {
	$counter++;
	if(0 == ($counter % 50) || $counter == @lines) {
 	print "\rProcessing: [";
	for($i = 0; $i < ($counter / $size); $i++) {
		print "+";
	}
	for($i = 0; $i < (49 - ($counter / $size)); $i++) {
		print "-";
	}
	print "]";
	}

	# Empty paragraph removal
	$_ =~ s/<p>\s*<\/p>//mi;
	if($_ =~ m/^\s*\n$/) { 
		# If the line is just a newline or newline and spaces, scrap it.
		$_ = '';
	}
	else {
		# Remove excess spaces
		$_ =~ s/  //mi;
		# I get these alot...
		$_ =~ s/&shy;//mi;
	}
	print FILEWRITE $_;
}
close FILEWRITE;
print "\n";