#!/usr/bin/perl -T

$ENV{'PATH'} = '' ;
my $tmpfile ="/tmp/hash-$$" ;

use strict;
use LWP::UserAgent;
use Digest::MD5 qw( md5_base64 ) ;
use CGI;
my $cgi = new CGI;
my $url = $cgi->param('url')  || $ENV{'URL'} ;

my %catalog = (
	'HTML'	=>	'/usr/local/valet/sgmlib/catalog' ,
	'XHTML'	=>	'/usr/local/valet/sgmlib/xhtml1/xhtml.soc' ,
	'XML'	=>	'/usr/local/valet/sgmlib/xml.soc' ,
) ;

my $markup = $cgi->param('markup') || 'HTML' ;

my $soc = $catalog{$markup} ;
my $nsgmls = "SGML_CATALOG_FILES=$soc /usr/local/valet/bin/onsgmls -E0 " ;

my $agent = new LWP::UserAgent;
$agent->agent("Site Valet Metric") ;

my $request = new HTTP::Request 'GET' , $url ;
my $response = $agent->request($request) ;

my $lm = $response->last_modified || "(not specified)" ;

print qq(Content-Type: text/plain
Cache-Control: no-cache

Hashes for $url
) ;
Last-modified: $lm


if ( $response->is_success ) {

  if ( open ( TMP, ">$tmpfile" ) ) {
    print TMP $response->content ;
    close TMP ;
    my $in_all ;
    my $in_attr ;
    my $in_els ;
    my $in_hdrs ;
    my $in_text ;
    if ( open (VAL, "$nsgmls $tmpfile 2>/dev/null |") ) {
      for (<VAL>) {
	$in_all .= $_ ;
	$in_attr .= $_ if /^[\(\)A]/ ;
	$in_els .= $_ if /^[\(\)]/ ;
	$in_hdrs .= $_ if /^[\(\)]H[123456]/ ;
	$in_text .= $_ if /^\-/ ;
      }
      close VAL ;
      my $all = md5_base64($in_all) ;
      my $attr = md5_base64($in_attr) ;
      my $els = md5_base64($in_els) ;
      my $hdrs = md5_base64($in_hdrs) ;
      my $text = md5_base64($in_text) ;

print qq(
All Content:		$all
Elements+Attributes:	$attr
Elements:		$els
Headings:		$hdrs
Text:			$text
) ;
    } else {
      print qq(
Server error 2
      ) ;
    }
  } else {
    print qq(
Server error 1
    ) ;
  }

} else {
print qq(
Error parsing or fetching $url
) ;
}
END {
  unlink $tmpfile if -f $tmpfile ;
}