#!/usr/bin/perl
#
# Walk Apache cache and report stats in the form
#	total	class
# where the class depends on the type of data:
#	HOST	hostname of URL
#	TYPE	content-type
#

use strict qw(vars);

use cs::Misc;
use cs::Source;
use cs::Upd;
use cs::URL;
use cs::RFC822;
use cs::Hier;
use File::Find;

$::Blocksize=1024;

$::Usage="Usage: $::cmd class [dirs...]
	class	HOST or TYPE
";

my($badopts)=0;

if (! @ARGV)
	{ warn "$::cmd: missing class\n";
	  $badopts=1;
	}
else
{ $::Type=uc(shift(@ARGV));
  @ARGV='.' unless @ARGV;
}

die $::Usage if $badopts;

$cs::Upd::This=new cs::Upd main::STDERR;     # default upd structure

undef %::Tally;

find(sub {
	   return 0 if ! stat;
	   return 1 if -d _;
	   return 0 if ! -f _;
	   $::Silent || out($File::Find::name);
	   my($s)=cs::Source::open($_);
	   if (! defined $s)
		{ warn "$::cmd: can't open $File::Find::name: possible error: $!\n";
		  return 0;
		}

	  { my($line);

	    return 0 if ! defined ($line=$s->GetLine()) || ! length $line;

	    my($urlhdr);

	    return 0 if ! defined ($urlhdr=$s->GetLine()) || ! length $urlhdr;

	    return 0 if ! defined ($line=$s->GetLine()) || ! length $line;
	    $s->UnGet($urlhdr);
	  }

	  my($h);

	  return 0 if ! defined ($h=new cs::RFC822 $s);
	  undef $s;

	  tally($h,$File::Find::name);

	  1;
	},
     @ARGV);

out('');

for (sort keys %::Tally)
	{ print "$::Tally{$_}\t$_\n";
	}

exit 0;

sub tally($)
	{ my($h,$path)=@_;
	  ## warn "h=".cs::Hier::h2a($h,1)."\n";
	  my($cl)=scalar($h->Hdr(CONTENT_LENGTH));
	  my($size)=int(($cl+$::Blocksize/2)/$::Blocksize);

	  return if $size < 1;

	  my($key);

	  if ($::Type eq HOST)
		{ my($url)=$h->Hdr(X_URL);

		  if (! defined $url
		   || ! length $url)
			{ warn "$::cmd: no URL for $path\n";
			}

		  my($u)=new cs::URL $url;

		  $key=lc($url->HostPart());
		}
	  elsif ($::Type eq TYPE)
		{ $key=lc($h->Hdr(CONTENT_TYPE));
		}
	  else
	  { die "$::cmd: unsupported class \"$::Type\"\n";
	  }

	  $::Tally{$key}+=$size;
	}