#!/usr/bin/perl
# -*- perl -*-
#
#  Copyright (c) 1997, 2000 DJ Delorie, All Rights Reserved.  NO WARRANTEE.
#

push(@INC, split(':', $ENV{'PATH'}));
push(@INC, "/home/apache/bin");

#open(LOCK, ">/tmp/lock");
#flock(LOCK, 2);

if ($ARGV[0] eq "see-script") {
    print "Content-type: text/plain\n\n";
    open(IN, "ses.cgi");
    print while <IN>;
    exit 0;
}

$sockaddr = "S n a4 x8";
#require "sys/socket.ph";
use Socket;
require "./common.pl";

push(@INC, "/usr/local/etc/httpd/bin");
push(@INC, split(':', $ENV{"PATH"}));
require "cgi-lib.pl";
&ReadParse;

$in{'url'} =~ s@http://http://@http://@;

print "Content-type: text/html\n\n";
print `header -sky Search Engine Simulator`;

$agent = "Delorie.com SES";

if ($in{'url'}) {
    print "<form action=\"ses.cgi\" method=GET>\n";
    $value = $in{'url'};
    $value = "http://" unless $value;
    print "<input type=text name=url size=40 value=\"$value\">\n";
    print "<input type=submit value=\" view \"></p>\n";
    print "</form>\n";

    print "<hr noshade size=2>\n";
    if ($in{'url'} !~ m@^http://([^\.]+\.)+[^\.]+@) {
	print "Sorry, I can only handle http://<i>some.host</i>/ URLs.\n";
	$u = &HtmlEncode($in{'url'});
	print "You typed in `<tt>$u</tt>'.\n";
	print `trailer`;
	exit 0;
    }

    $page = &webget($in{'url'});

    $has_tables = 1 if $page =~ /<\s*table/i;

    if ($webgot_url ne $in{'url'}) {
	print "<p><em>Note: The web server returned a \"redirect\".  Search\n";
	print "engines may not always follow redirects.  The URL below reflect the\n";
	print "page I was redirected to.</em></p>\n";
    }

    ($date) = $web_header =~ /Last-modified:\s*(.*\S)/i;
    $size = length($page);

    print "<h2><tt>$webgot_url</tt></h2>\n\n";

    $page =~ s/[\r\n\t ]+/ /g;
    $page =~ s@<img[^>]* alt="([^\"]*)"[^>]*>@$1@gi;
    $page =~ s@<img[^>]* alt='([^\']*)'[^>]*>@$1@gi;
    $page =~ s@<img[^>]* alt=([^ >]*)[^>]*>@$1@gi;

    ($title) = $page =~ m@<title>([^<]*)</title>@i;
    $title =~ s@</title>.*@@;
    if (! $title) {
	$title = $in{'url'};
	$title =~ s@.*/(.+)@$1@;
	$title = "<tt>$title</tt>";
    }

    $banner = "<b><big><a href=\"$in{'url'}\">$title</a> &nbsp; </big></b>";
    if ($date) {
	$banner .= " <small><i>$date, $size bytes</i></small><br>\n";
    } else {
	$banner .= " <small><i>$size bytes</i></small><br>\n";
    }
    $start = "<p><table border=1 cellspacing=0 cellpadding=5><tr><td>\n";
    $trailer = "<br><small><tt><a href=\"$in{'url'}\">$in{'url'}</a></tt></small>";
    $end = "</td></tr></table></p>\n";

    $summary = $page;
    $summary =~ s@.*</(title|head)>@@i;

    # From striphtml
    $summary =~ s{<!(.*?)(--.*?--\s*)*(.*?)>}{\ }gsx;
    $summary =~ s{\<([^>\'\"]|\"[^\"]*\"|\'[^\']*\')*\>}{\ }gsx; #"

    $summary =~ s/\&nbsp;/ /g;
    $summary =~ s/  +/ /g;
    $summary =~ s/^ +//g;
    $summary =~ s/^(.{1,200}\S) .*/$1/;
    $summary =~ s/\</\&lt;/g;
    $summary =~ s/\>/\&gt;/g;

    print "<p>Example 1: Summary based on all text on page</p>\n";
    print $start, $banner, $summary, $trailer, $end;

    $headers = $page;
    $htext = "";

    $i = 0;
    while ($headers =~ m@<h[1-6][^>]*>(.*?)</?(h[1-6]|tr|td|th|table)>@ig) {
        $htext .= $1;
        $htext .= " ";
	last if $i++ == 100; # just in case
    }

    $htext =~ s{<!(.*?)(--.*?--\s*)*(.*?)>}{<!$1 $3>}gsx;
    $htext =~ s@\<([^\>\'\"]|\"[^\"]*\"|\'[^\']*\')*\>@@gsx; #"
    $htext =~ s/\&nbsp;/ /g;
    $htext =~ s/  +/ /g;
    $htext =~ s/^ +//g;
    $htext =~ s/^(.{1,200}\S) .*/$1/;
    $htext =~ s/</&lt;/g;
    $htext =~ s/>/&gt;/g;

    print "<p>Example 2: Summary based on headers only</p>\n";
    print $start, $banner, $htext, $trailer, $end;


    $text = $page;
    $text =~ s/[\r\n\t\ ]+/ /g;

    # From striphtml
    $text =~ s{<!(.*?)(--.*?--\s*)*(.*?)>}{\ }gsx;
    $text =~ s{\<([^>\'\"]|\"[^\"]*\"|\'[^\']*\')*\>}{\ }gsx; #"

    $text =~ s/\&nbsp;/ /g;
    $text =~ s/  +/ /g;
    $text =~ s/^ +//g;
    $text =~ s/\</\&lt;/g;
    $text =~ s/\>/\&gt;/g;

    print "<p>Example 3: Text excerpt of page</p>\n";
    print $start, "<small><tt>", $text, "</tt></small>", $end;

    if ($has_tables) { print <<EOF;

<p>If your page uses tables, and the wrong text is showing up in these
examples (i.e. the sidebar instead of the body), I have <a
href="ses-hint.html">some helpful hints for laying out your tables
to make your main text show up first</a></p>

EOF
}

} else {

    print "<p>This service allows web authors to see what their pages will look\n";
    print "like (sort of) to a search engine.  This service ignores the META\n";
    print "tags that some search engines honor.</p>\n";

    print "<form action=\"ses.cgi\" method=GET>\n";

    print "<p>Enter the URL you want to view:<br>\n";
    $value = $in{'url'};
    $value = "http://" unless $value;
    print "<input type=text name=url size=50 value=\"$value\"></p>\n";

    print "<p><input type=submit value=\" view page \"></p>\n";
    print "</form>\n";

    print <<EOF;

<p>If your page uses tables, and the wrong text is showing up in these
examples (i.e. the sidebar instead of the body), I have <a
href="ses-hint.html">some helpful hints for laying out your tables
to make your main text show up first</a></p>

EOF

    print "<p><a href=\"ses.cgi?see-script\">See the CGI's perl source</a></p>\n";

    open(V, "viewers.html");
    print while <V>;
    close(V);
}

print `trailer`;