Listing of parse.cgi

#!/usr/local/bin/perl5 -T

$index = ($0 =~ /index.cgi/i);
$heading_two = 4;
$heading_one = 2;
$keywords = "Jens Johansson home page";
$footnotes_used = 0;
$unix = ($ENV{'COMSPEC'} eq "" );
$user_agent = $ENV{'HTTP_USER_AGENT'};
$lynx = ($user_agent =~ /lynx/i);
$msie = ($user_agent =~ /msie/i);
if ($lynx) {
  $heading_two = 2;
  $heading_one = 1;
}

#$smiley = "";
#$smiley = "<moronic giggle>";
$smiley = "<g>";
$hspace = 15;
$spacerzzz = "<img src=\"/stuff/2x2.gif\" alt=\"\" hspace=$hspace>";
$title = "n/a";
$fontname = "arial,helvetica";

srand;                           # randomize

$* = 1;                    # multiline string matching

if ($index) {
   open (IN, "./index.par"); 
   $in_file = "index.par";
} else {
   if ($ENV{'PATH_INFO'} eq "") {
      open (IN, "-");
      $stdin_infile = 1;
   } else {

      $arguments = $ENV{'PATH_INFO'};

      if ($ENV{'CONTENT_LENGTH'} !~ /^\s*0\s*$/) {
         read(STDIN, $buffer, $ENV{'CONTENT_LENGTH'});
         @postdata = split(/&/, $buffer);
         foreach $pair (@postdata) {
            ($post_name, $post_value) = split(/=/, $pair);
            $post_value =~ tr/+/ /;
            $post_value =~ s/%([a-fA-F0-9][a-fA-F0-9])/pack("C", hex($1))/eg;
            if ($post_name =~ /COOKIE/i && $post_name !~ /SHOWCOOKIES/i) {
               $post_arguments .= "^" . $post_value;
            } else {
               $post_arguments .= "^" . $post_name . "_" . $post_value;
            }
        }
        $post_arguments =~ s/^\^//;
     }

      $arguments =~ s/(^[^\^]*)\^?//;
      $in_file = $1;
      open (IN, $in_file);

      @set_cookies = (split(/\^/, $arguments), split(/\^/, $post_arguments));
      foreach (@ARGV) {
         push (@set_cookies, split(/\^/) );
      }
      foreach (@set_cookies) {if (/^fc(.+)$/) {push(@fake_cookies, $1); undef $_; }}
   }
}

($in_dev, $in_ino, $in_mode, $in_nlink, $in_uid, $in_gid, $in_rdev,
$in_size, $in_atime, $in_mtime, $in_ctime, $in_blksize, $in_blocks) 
= stat(IN);

 ($sec, $min, $hour, $mday, $mon, $year, $wday, $yday, $isdst) 
= localtime($in_mtime);

$in_modtime = sprintf("%s %d, %d at %02d:%02d",
('Jan', 'Feb', 'Mar', 'Apr', 'May', 'Jun',
'Jul', 'Aug', 'Sep', 'Oct', 'Nov', 'Dec')[$mon],
$mday, $year+1900, $hour, $min);

$in_file =~ s/\#.*$//;
$in_file =~ s/\^.*$//;

while (<IN>) {

   if (/^\s*\#\# \/omit \#\#/i) {$convert_omit = 0; next;}
   if ($convert_omit) {next; }

   if (/^\s*\#\# \/graphical \#\#/i) {$omit_g = 0; next;}
   if ($omit_g && ($lynx) ) {next; }

   if (/^\s*\#\# \/nographical \#\#/i) {$omit_ng = 0; next;}
   if ($omit_ng && (!$lynx) ) {next; }

   if (/\s*^\#\# fakecookies? (.*) \#\#$/) {
      push(@fake_cookies, split(/\^/, $1));
      next;
   }
   if (/^\s*\#\# polyexpand (.*)\s*\#\#/i) {
      my $p = $1; my $pe = $p; my @lines;
      $pe =~ s/([^A-Za-z0-9])/sprintf("%%%02X", ord($&))/eg;
      unless ($lynx) { #arrgghh
         while ($p =~ /(.{40,80}?\|)(.+)/ or $p =~ /(.{50,100}?) (.+)/) {
            push(@lines, $1); $p = $2;
         }
         push(@lines, $p);
         $_ = "<pre>";
         foreach my $line (@lines) {
            foreach ($line) {s/^\s+//; s/\s+$//; }
            $_ .= "\t<a href=\"/polymath.cgi?string=$pe\">$line</a>\n";
         }
         $_ .= "</pre>\n";
      } else {
         $_ = "<p><pre>\t<a href=\"/polymath.cgi?string=$pe\">$p</a>\n</pre>";
      }
   }
   if (/^\s*\#\# body (.*)\#\#/i) {$variable_body_tag_content = $1; next;}
   if (/^\s*\#\# title (.*)\#\#/i) {$title = $1; next;}
   if (/^\s*\#\# (\/)?polyexpand-collect \#\#/i) {
      if ($1 eq "/") {
         $polyexpand_collect = 0;
      } else {
         $polyexpand_collect = 1;
         $polyexpand_collect_string = "";
      }
      next;
   }
   if ($polyexpand_collect) {$polyexpand_collect_string .= $_; }
   if (/^\s*\#\# polyexpand-emit \#\#/i) {
      $polyexpand_collect_string =~ s/\#.+$//g;
      $polyexpand_collect_string =~ s/\s+/ /g;
      $_ = "
<form action = \"polymath.cgi\" method=post>
<input type=hidden name=string value=\"$polyexpand_collect_string\">
<input type=submit value=\"play that\"></form>
";

   }
   if (/^\s*\#\# literal (.*)\#\#/i) {$convert_literal = 1; $literal_opt = $1; next;}
   if (/^\s*\#\# \/literal \#\#/i) {$convert_literal = 0; next;}

   if (/^\s*\#\# graphical \#\#/i) {$omit_g = 1; next;}
   if (/^\s*\#\# nographical \#\#/i) {$omit_ng = 1; next;}
   if (/^\s*\#\# omit \#\#/i) {$convert_omit = 1; next;}
   if ($convert_literal) {
      s/([^A-Za-z0-9\t])/sprintf("&#%03d;", ord($&))/eg;
      if ($literal_opt =~ /tab/) {s/\t/$spacerzzz/g; }
      unless ($literal_opt =~ /nobr/) {s/$/<br>/;} 
   }

   if (/^\s*\#\#/) {next; }
   s/\#\#.*$//;
   $page .= $_;
}
close(IN);

foreach (split (/;/, $ENV{'HTTP_COOKIE'}) ) {
   /\s*(.*)=(.*)/;
   $cookie{$1} = $2;
}

if ($cookie{'fontname'}) {$fontname = $cookie{'fontname'}; }

#
# headings etc
#

$hed_tag = 0; $seq_tag = 0; 

$tmsgt = scalar(localtime);

if ($tmsgt =~ /2001/ && $tmsgt =~ /Sep 1[12]/) {
$tmsg = "<font size=+2><font color=red><br><br><center>
Phone and email service spotty. I am alive and OK, at least as of
21:33 local time (EDT, Sep 11). Lots of other people 
here were not so fortunate.
Let's hope nothing more happens.
I appreciate your concern.
<br><br></center>
</font></font>\n";
$page = $tmsg . $page;
}

if ($tmsgt =~ /Sep 11/) {
$tmsg = "<font color=red><br><br><center>
Let's all hope for a calm and peaceful September 11 this year
<br><br></center>
</font>\n";
$page = $tmsg . $page;
}

if ($page =~ /^\/\/\/\s*(.*)\s*$/) {$title = $1;} 

$page =~ s/^\/\/\/(.*)$/<h${heading_one}>$1<\/h${heading_one}>/g;

while ($page 
 =~ s/^\/\/\s*(.*)\s*$/<a name=\"HED_TAG_$hed_tag\"><\/a><h${heading_two}><u>$1<\/u><\/h${heading_two}>/
) {$headings[$hed_tag] = $1; $hed_tag++; $nheadings++;}

if (1) { 

#Pragma: no-cache
#Cache-Control: no-cache
#Expires: Thursday, 01-Jan-1970 00:00:00 GMT

   $html_head = 
"Content-type: text/html
Keywords: $keywords
$doc_specific_header";

   unless($html_head =~ /^Cache-control/) {
      $html_head .= "Cache-Control: no-cache\n";
   }

   foreach (@set_cookies) {
      if (/^$/) {next; }
      /(.+)_(.*)/;
      $c1 = $1; $c2 = $2;
      if ($c1 eq "HILITE") {next; }
#
# special cookie 'timestamp' when set will aqcuire local time 
#   (in seconds since midnight 1/1/1970)
#
      if ($_ eq "timestamp") {$c1= "timestamp"; $c2 = $timestamp = time(); }
      $html_head .= 
         "Set-Cookie: $c1=$c2; expires=Monday, 18-Jan-2038 22:14:07 GMT; path=/\n";
      if ($c1 eq 'showcookies') {
        if ($c2 eq 'yes') {$showcookies = 1; } else
          {$showcookies = -1; }

      }
   }
} else {
   $html_head = "";
}

if ( (($cookie{'showcookies'} eq 'yes') + $showcookies) > 0 ) {
   $toppage .= "<center><font size=-1>First, me saw cookie: ";
   foreach (keys %cookie) {
      $toppage .= "'$_' = '$cookie{$_}'; "; 
   }
   $toppage .= " Then, me set cookie: ";

   foreach (@set_cookies) {
      if (/^$/) {next; }
      /(.+)_(.*)/;
      $c1 = $1; $c2 = $2;
      $toppage .= "'$c1' = '$c2'; "; 
   }

   foreach (@fake_cookies) {
      /(.+)_(.*)/;
      $c1 = $1; $c2 = $2;
      $toppage .= "'$c1' = '$c2' (fake cookie); "; 
   }

   $toppage .= "<br>You can turn this annoying cookie text off ";
   if ($in_file) {
      $toppage .= "<a href=\"$in_file^showcookies_normal\">here</a>, ";
   } else {
      $toppage .= "by going <a href=\"about.par?showcookies_normal\">here</a> ('about' page), ";
   } 
   $toppage .= "if you want!<br></font></center>";
}

foreach (@set_cookies, @fake_cookies) {

   if (/^VARNISH/) {
      $varnish_cookie = 1;
      next;
   }

   if (/^$/) {next; }
   /(.+)_(.*)/; 

   $c1 = $1; $c2 = $2;
   if ($_ eq "timestamp") {$c1= "timestamp"; $c2 = $timestamp; }
   $cookie{$c1} = $c2;

#
# special argument "HILITE_char-string" will not set a cookie per se.
# instead, a word will be marked for subsequent highligting.
#

   if ($c1 eq "HILITE") {
      $hilite = $c2;
      $hilite =~ s/-/\\s/;
      next;
   }

}

if ($cookie{'sick'} eq 'yes') {$cookie{'back'} = 'white'; }

if ($cookie{'fontname'}) {$fontname = $cookie{'fontname'}; } # again!

$titlebullet = "·";  # ¡¢£¤¥¦§¨©ª«¬­®¯°±²³´µ¶·¸¹º»¼½¾¿×Ø÷ø
$bodybullet = "•";
if ($lynx) {$titlebullet = "--"; $bodybullet = "*"; }

$prepage = "<html><head>
<LINK REL=\"SHORTCUT ICON\" HREF=\"http://www.panix.com/~jens/favicon.ico\">
<meta http-equiv=\"imagetoolbar\" content=\"no\" />
<title>&#9835;Jens Johansson " . $titlebullet . " " .
$title . "</title></head>
";

#
# formatting
#

if ($varnish_cookie) {
   $font_spec = "<font size=-1 face=\"$fontname\">";
} else {
   $font_spec = "<font face=\"$fontname\">";
}

$font_spec_end = "</font>";

$fn_glob_open = "<font color=#004000>";

if ($in_file) { $bkg_switch_file = $in_file; }
else {$bkg_switch_file = "about.par"; }

if (! $lynx) {
   if ($cookie{'marg'} eq 'wide') {
      $bkg_switch = "<a href=\"/parse.cgi/$bkg_switch_file?marg_normal\">[ margins ]</a>\n";
   } else {
      $bkg_switch = "<a href=\"/parse.cgi/$bkg_switch_file?marg_wide\">[ no margins ]</a>\n";
   } 
}

if ( 
   $variable_body_tag_content !~ /=#[0-9a-fA-F]{6,6}/ && $cookie{'sick'} ne 'yes'
   && ( ! $lynx )
) {
   if ($cookie{'back'} eq 'white') {
      $body_attr = " text=#000000 bgcolor=#FFFFFF";
      $bkg_switch .= "<a href=\"/parse.cgi/$bkg_switch_file?back_normal\">[ view as white text on black backgound ]</a>";
   } else {
      $body_attr = " bgcolor=#444444 text=#FFFFFF link=#AAAAFF alink=#FFAA00 vlink=#FFAAAA";
      $fn_glob_open = "<font color=#30FF30>";
      $bkg_switch .= "<a href=\"/parse.cgi/$bkg_switch_file?back_white\">[ view as black text on white background ]</a>";
   }
}

$bkg_switch = "<br><font size=-2>" . $bkg_switch . "</font>";

if ($cookie{'sick'} eq 'yes') {
   $body_attr = " text=#000000 background=\"/stuff/sick_bkb.gif\"";
   $font_spec = "<font size=+3>";
   $font_spec_end = "</font>";
}

$fn_glob_close = "</font>";

$body_attr_fn = $body_attr;

$body_attr .= " " . $variable_body_tag_content;

unless ($cookie{'marg'} eq 'wide') {
   if ($user_agent =~ /MSIE/i) {
      $body_attr1 .= " \nbottommargin=60 topmargin=60 leftmargin=60 rightmargin=60";
   } else {
      $body_attr1 .= " \nmarginwidth=60 marginheight=60";
   }
}

unless ($varnish_cookie) {
   $toppage_header = "<center><a href=\"/parse.cgi/index.par\">[ home ]</a>
   <a href=\"/parse.cgi/sect-index-brief.par\">[ sections ]</a>
   <a href=\"/parse.cgi/word-index.par\">[ references ]</a>
   <a href=\"/parse.cgi/records.par\">[ cd:s ]</a>
   <a href=\"/parse.cgi/about.par\">[ about ]</a>
   <a href=\"/parse.cgi/links.par\">[ links ]</a>
   <a href=\"http://www.heptagon.se\">[ heptagon ]</a>
   " . $bkg_switch . "
   </center>
   <br><br>";
} else {
   $toppage_header = "";
}

$toppage = "<body" . $body_attr . $body_attr1 . ">\n" . $font_spec . "
<!-- pages generated with utmost pride by a suite of custom perl 
scripts written by the 'great' Jens himself..  -->\n" .
$toppage_header . $toppage;

if ($cookie{'sick'} eq 'yes') {
   $toppage .= "<font size=+4>
<center>THE WHOLE SITE IS BEING DISPLAYED IN UGLY MODE!!<br>
<blink><A href=\"about.par?sick_no#sick\">CLICK HERE TO RETURN TO NORMAL MODE</a><br></blink>
<A href=\"about.par?sick_no#sick\">
<img width=150 height=150 align=middle src=\"/stuff/sick.gif\" alt=\"\"></a></font></center>";
}

unless ($varnish_cookie) {
   $postpage = "<p><!--endnotes--><hr><p><center> ";
   unless ($stdin_infile) {
      $postpage .= "Page updated " . $in_modtime . " " . $bodybullet . " ";
   }

   open (IN, "footer.par"); while (<IN>) {$footer .= $_; } close(IN);

   $postpage .= "Email: <a href=\"mailto:jens\@panix.com\">jens\@panix.com</a></center>" .
   $footer . $font_spec_end . "</body></html>";
} else {
   $postpage = "<p><!--endnotes-->" . $font_spec_end . "</body></html>";
}

#
# add subheading links (only for quite long pages, maybe??)
#

# length($page) > xxxxxxx

if ((! $lynx) 
    && (    ($cookie{'headings'} eq 'on')
         || ( ($cookie{'fakeheadings'} ne '') && $cookie{'headings'} ne 'off') )
        ) { 
   for ($h=0; $h < $nheadings; $h++) { 
      $subheadings .= "\n<a href=\"$in_file#HED_TAG_$h\">[ $headings[$h] ]</a>\n";
   }
   $fontsize=-1; if ($nheadings > 8) {$fontsize=-2; }
   $page = "<font size=$fontsize>" . $subheadings . "</font><p>" . $page;
}

$page = $toppage . $page . $postpage;

#
# do comment clobbering first! 
# in case some [] tags etc have been commented out
#

# '\#\#' comments
$page =~ s/^\#\#(.*)$//g;
$page =~ s/\#\#(.*)\#\#//g;

# '#' comments
$page =~ s/^\#(.*)$/<!-- $1 -->/g;

#
# now do repetitive mangling
#

#
#    <!--#include "filename"-->               like <!--#include -->
#

while (
   ($page =~ s/<!--#include "([^>]*)"-->/{PLACEHOLDER}/)
) { 
   $include = "";
   open (IN, $1); while (<IN>) {$include .= $_; } close(IN);
   $page =~ s/{PLACEHOLDER}/$include/;
}

#
#     do freaky dynamic-page stuff:
#
#     {a|b|c|d}          output string a,b,c or d with equal probability
#     {?0.1a|?0.3b|c|d}  p(a)=0.1, p(b)=0.3, p(c)=p(d)=(1-(0.1+0.3))/2
#     {@string a}        output a only if string is contained in remote_host 
#     {@string a|b}         same -- b if NOT
#     {=cookie value a}  output a only if cookie has value value
#         the value 'normal' implies; no cookie present means 'yes'
#     
#     {=cookie value dummyarg} output VALUE of cookie if exists.
#              ^^^^^ that is, use the *actual* string 'v a l u e' 
#
#     {=cookie value a|b}   same -- b if NOT
#
#     {=timestamp number a|b}
#     output a if cookie 'timestamp' is lower than 'number'
#     {=gettimestamp dummy_arg dummy_arg}
#               output value of cookie 'timestamp' as text (GMT)
#

#Sat Nov 06 09:26:33 1999
#while ($page =~ s/{([^}]*)}/{PLACEHOLDER}/) {
#   @curly_args = split(/\|/, $1);

while (
   ($status, @result) = &excise($page, "{", "}"), $status
) {
#   @curly_args = split(/\|/, $result[2]);
   @curly_args = split_outer($result[2], "|", "[{", "]}");

#Sat Nov 06 09:26:33 1999
   
   foreach (@curly_args) { 
      s/^\\\\\t/<br><img src="\/stuff\/2x2.gif" alt=\"\" hspace=$hspace>/g ;
      s/^\\\\/<br>/g ;
      s/^\\\t/<p><img src="\/stuff\/2x2.gif" alt=\"\" hspace=$hspace>/g ;
      s/^\\/<p>/g ;
      s/\s/ /g; 
   }
   if ($curly_args[0] =~ /^@(\S*)\s+(.*)/) { # remote_host replacement
      $repl = $2;
      if ($curly_args[1] ne "") {$alt  = $curly_args[1]; } else {$alt = ""; }
      if ($ENV{"REMOTE_HOST"} =~ /$1/) 
         {} else
         {$repl = $alt; }
   } elsif ($curly_args[0] =~ /^\=(\S+)\s+(\S+)\s+(.*)/) { # cookie stuff

      $c1 = $1; $c2 = $2; $c3 = $3;

      $repl = $c3;

      if (($c2 eq 'normal') && ($cookie{$c1} eq "")) {
         $cookie{$c1} = 'normal';
      }

      if ($curly_args[1] ne "") {$alt  = $curly_args[1]; } else {$alt = ""; }

      if (($c2 eq 'value') && ($cookie{$c1} ne "") && ($cookie{$c1} ne "normal")) {

         $repl = $cookie{$c1};

      } elsif ($c1 eq "gettimestamp") { # output the timestamp cookie!

          if ($cookie{'lastvisit'} eq 'on') 
            {$repl = gmtime($cookie{'timestamp'}) . " GMT"; } 
            else {$repl = ""; }

      } elsif ($c1 eq "timestamp") { # special case! timestamp dependent!!

         if($cookie{'lastvisit'} eq 'on' && $cookie{'timestamp'} < $c2)
            {} else
            {$repl = $alt; }

      } else {
         if ($cookie{$c1} eq $c2)
            {} else
            {$repl = $alt; }
      }
      

   } else

#     ok, probability stuff
   {
      undef %curly_prob; $t = 0; $n = 0; $m = 0; 
      foreach $arg (@curly_args) {
         $m++;
         if ($arg =~ /^\?([\d.]+)(.*)/) {
            $arg = $2; $curly_prob{$arg} = $1;
            $t += $1; 
         } else {
            $curly_prob{$arg} = 0; $n++; 
         }
      }      
      $limit = 0;
      if ($t > 1) {                      # wrongly specified probabilities!
         foreach $arg (@curly_args) {    # set all equal..
            $limit += 1 / $m;
            $curly_prob{$arg} = $limit;
         }
      } else {
         foreach $arg (@curly_args) {
            if ($curly_prob{$arg} == 0)
               { $limit += (1-$t) / $n; } else
               { $limit += $curly_prob{$arg}; }
            $curly_prob{$arg} = $limit;
         }
      }
      $random = rand;

      $repl = "";
      foreach $arg (@curly_args) {
         if ($random < $curly_prob{$arg} ) {
            $repl = $arg; last;
         }
      }
  }   

#Sat Nov 06 09:26:33 1999
#   $page =~ s/{PLACEHOLDER}/${repl}/;
   $page = $result[0] . $repl . $result[1];
#Sat Nov 06 09:26:33 1999

}

#
# handle [* ] footnotes
#

while ($page =~ s/\[\*([^]]*)\]/{PLACEHOLDER}/) {

   if ($cookie{'foot'} eq 'off') {
      $repl = "";
      goto done_footnote;
   }

   $footnote_text = $1;
   $footnote_text =~ s/\s*([\x00-\xFF]*)\s*/$1/;
   if ($footnote_text =~ /([^\|]+)\|(.*)/) {
      $footnote_link = $footnote_inline = $1; $footnote_text = $2;
      $fn_open = "";
      $fn_close = "";
   } else {
      $footnote_inline = "";
      if ($cookie{'foot'} eq 'bottom') {
         $footnote_link = ++$footnote_count;
      } else {
         $footnote_link = "*";
      }
      $fn_open = "<font size=-1><sup>";
      $fn_close = "</sup></font>";
   }

   $footnote_text =~ s/\s+/ /g;

# escape special chars into into $footnote_text_w

   $footnote_text_w = $footnote_text;
   $footnote_text_w =~ s/(.)/sprintf("%02X", ord($&))/eg;
   
   if ($cookie{'foot'} eq 'bottom') {
      $fn_open = "<font size=-1><sup>";
      $fn_close = "</sup></font>";
      $repl = $fn_open . "<a href=\"$in_file#FOOT_$footnote_count\"\n";
      $repl .= ">" . $fn_glob_open . $footnote_link . $fn_glob_close . "</a>" . $fn_close;
      $footnotes_at_end .= "<br><a name=\"FOOT_$footnote_count\"></a>$footnote_count: $footnote_text\n";
   } else {
      if ($cookie{'foot'} eq 'inline') {
         if (length($footnote_inline != 0))
             {$footnote_inline .= " "; }
         $repl = " <font size=-1>$footnote_inline($footnote_text) </font>";
      } else {
         $repl = $fn_open . "<a href=\"/fn0.cgi?${footnote_text_w}\"\n";
         unless ($cookie{'foot'} eq 'href') { # .. test onClick, maybe? return false?
            $repl .= "onMouseOver=\"footnote('${footnote_text_w}', event)\" 
onMouseOut=\"footnoteout()\"";
            $footnotes_used = 1;
         }

      $repl .= ">" . $fn_glob_open . $footnote_link . $fn_glob_close . "</a>" . $fn_close;
      }
   }
 
done_footnote:

   $page =~ s/{PLACEHOLDER}/${repl}/;
}

#
# handle [] special indexed link names
#

#while ($page =~ s/([^\\])\[([\S][^]]*)\]/$1{PLACEHOLDER}/) {  
while ($page =~ s/\[([\S][^]]*)\]/{PLACEHOLDER}/) {  

# [ xxxxxx ] not transformed, for text mode button etc

 $link_text = $1;
 if ($link_text =~ /^\\(.*)$/) { # [\blabla] => [\ is escape seq
   $repl = "$1"; # just lose the []'s
 } elsif ($link_text =~ /\?(.*)/) { # [?] search engine page quick reference
   $spec = $spec0 = $1;
   $spec0 =~ s/\s*(\w+)\s.*/$1/;
   $spec =~ s/\s*(.*)/$1/;
   $spec =~ s/([^\w\. ])/sprintf("%%%02X", ord($&))/eg;
   $spec =~ tr/ /+/;
   $repl = "<a href=\"/generate-web-search.cgi/$spec\" VARNISH>internet search links for '$spec0'</a>"; 
 } else { # normal [blabla] link

# webify it into link_text_w

   $link_text_w = $link_text;
   $link_text_w =~ s/([^\w\. ])/sprintf("%%%02X", ord($&))/eg;
   $link_text_w  =~ tr/ /+/;
   $repl = "<a href=\"/generate-word-index.cgi/${link_text_w}\" name=\"SEQ_TAG_$seq_tag\" VARNISH>$link_text</a>";
   $seq_tag++;
   }
 $page =~ s/{PLACEHOLDER}/${repl}/;
}

# fix escaped [xxxxxxx xxxx ] 

#$page =~ s/([\\])(\[[\S][^]]*\])/$2/g; 

#
# finally do rest of one-time manglings
#

if (length($footnotes_at_end) != 0) {
  $footnotes_at_end = "<font size=-3>" . $footnotes_at_end . "</font>";
  $page =~ s/<!--endnotes-->/$footnotes_at_end/;
}

# records.par jpegs or gifs.. 
if ($cookie{'recimg'} eq 'jpg' && $in_file =~ /records\.par/i) {
   $page =~ s#src="/stuff/cov/(.+)\.gif"#src="/stuff/cov/$1\.jpg"#ig;
}

# blink
if ($cookie{'blink'} eq 'off') {
   $page =~ s/<\s*\/?blink\s*>//ig;
}

# not sure how lynx handles &nbsp; .. checked, seems OK!
if (1 || $user_agent =~ /(mozilla|msie|opera|arachne)/i) {

   while($page =~ s/\[( .* )\]/{PLACEHOLDER}/) {
#   while($page =~ s/\[ (.*) \]/{PLACEHOLDER}/) {

      $repl = $1;
      $repl =~ s/ /\&nbsp\;/g;
      $page =~ s/{PLACEHOLDER}/${repl}/;
    }
} else {
}

# breaks, para marks, etc.
# \\tab => <br>+hspace, \tab => <p>+hspace
# \\ => <br>, \ => <p>

$page =~ s/^\\\\(\t+)/"<br>" . ($spacerzzz x length($1)) /eg ;
$page =~ s/^\\\\/<br>/g ;
$page =~ s/^\\(\t+)/"<p>" . ($spacerzzz x length($1)) /eg ;
$page =~ s/^\\/<p>/g ;

# word emphasis
   $page =~ s/\*([a-zA-Z][-0-9A-Za-z!?,.:;"'_ ]{1,50})\*/<i>$1<\/i>/g ;

# smileys

if ($cookie{'smiley'} eq 'on') {
   $page =~ s/(;\)|:\)|<g>)/${smiley}/g;
} else {
   $page =~ s/(;\)|:\)|<g>)//g;
}

# hey, let's change any stupid <B></B> tags to <i> !

$page =~ s/<([^b]?)b>/<$1i>/gi ;

# and as an antidote to that stupidity, let's specify <bb> to mean
# bold _really_ .. sigh

$page =~ s/<(.?)bb>/<$1b>/gi ;

# -- => typographical character

unless ($cookie{'dash'} eq 'off') {
   $page =~ s/([^!])--([^>])/$1&#151;$2/g;
}

#
# iso -> entityname translation
#
%entitymap = (
   'Å', '&Aring;', 'Ä', '&Auml;', 'Æ', '&AElig;', 'É', '&Eacute;',
   'Ñ', '&Ntilde;', 'Ö', '&Ouml;', 'Ø', '&Oslash;', 'Ü', '&Uuml;',
   'ß', '&szlig;', 'å', '&aring;', 'ä', '&auml;', 'æ', '&aelig;',
   'é', '&eacute;', 'ñ', '&ntilde;', 'ö', '&ouml;', 'ø', '&oslash;',
   'ü', '&uuml;', '®', '&reg;', '©', '&copy;', 
#  '™', '&trade;'
   '™', '(tm)'
);

if ($lynx) {$entitymap{'©'} = '(c)'; }

$entity_charclass = '[' . join('', keys %entitymap) . ']';
$page =~ s/($entity_charclass)/$entitymap{$1}/eg;

#
# hey, let's waste billions of processor cycles and megabits of 'net bandwidth! 
#
if ($cookie{'sick'} eq 'yes') {
   $page2 = ""; $in_literal = 0; $in_tag = 0;
   for ($i = 0; $i < length($page); $page2 .= $str, $i++) {
      $str = substr($page, $i, 1);   
      if ($in_literal && $str eq ";") {$in_literal = 0; next; }
      if ($in_literal) {next; }
      if ($str eq "&") { $in_literal = 1; next; }
      if ($in_tag && $str eq ">") {$in_tag = 0; next; }
      if ($in_tag) {next; }
      if ($str eq "<") { $in_tag = 1; next; }
      $str = sprintf("<font color=#%2x%2x%2x size=%d>%s<\/font>", rand(255), rand(128), rand(128), rand(4)+3, $str);
   }
   $page = $page2;
}

# this is even more useless... :)

if ($cookie{'atarifont'} eq 'yes' && ! $lynx) {
   if ($cookie{'back'} eq 'normal') {$atarifontdir = '/stuff/atari0/';} 
   else {$atarifontdir = '/stuff/atari1/';} 
   
   $page2 = ""; $in_literal = 0; $in_tag = 0;
   for ($i = 0; $i < length($page); $page2 .= $str, $i++) {
      $str = substr($page, $i, 1);   
      if ($in_literal && $str eq ";") {$in_literal = 0; next;
}
      if ($in_literal) {next; }
      if ($str eq "&") { $in_literal = 1; next; }
      if ($in_tag && $str eq ">") {$in_tag = 0; next; }
      if ($in_tag) {next; }
      if ($str eq "<") { $in_tag = 1; next; }
      $str =~ s/\s/ /;
      $str = sprintf(
         "<img width=8 height=16 src=\"%schar-%02x.gif\" alt=\"%s\" hspace=0 vspace=0>", 
         $atarifontdir, ord($str), $str);
   }
   $page = $page2;
}

# "hilite"
if ($hilite) {

#   $page =~ s/([\s>;()])($hilite)([\s'!.?,<&()])/$1<b><blink><font color=red>$2<\/font><\/blink><\/b>$3/ig;
   $page =~ s/([\s>;()])($hilite)([\s'!.?,<&()])/$1<b><font color=red>$2<\/font><\/b>$3/ig;

}

# short-hand links <http_blabla> => <a href="http_blabla">http_blabla</a>
# Sat Aug 21 14:00:51 1999 hacked to also allow <link_text|url>

while ($page =~ s/<(http:|ftp:|www\d*\.)([^>]+)>/{PLACEHOLDER}/) {
#while ($page =~ s/<(((http:|ftp:|www\d*\.)([^>]+))|(([^>]+)\.par($|\|)))>/{PLACEHOLDER}/) {
   $r_url = $1 . $2;
#   $r_url = $1;
   $r_url =~ s/^www/http:\/\/www/;
   if ($r_url =~ /(.+)\|(.*)/) {
      $replace = "<a href=\"$1\">$2<\/a>";
   } else {
      $replace = "<a href=\"$r_url\">$r_url<\/a>";
   }
   $page =~ s/{PLACEHOLDER}/$replace/;
}

# bare .par links transformed to specify parse.cgi
$page =~ s/href="([#\w.-]+\.par[^"]*)"/href="\/parse.cgi\/$1"/ig;

# here we go, window madness. needs to be tidied up..
# sdfsdf. . 3.4 .23.4 23.4 2 135 34 !!!!QHTUQWYRUWEYT!!!!
# <  VARNISH> -- no controls -- floating text window
# <  TARNISH> -- no controls, read image data for target win size if present
# 

if ($cookie{'img'} ne 'same' && (! $lynx )) {
   while ($page =~ s/<a\shref=([^>]*)>/{PLACEHOLDER}/i) {
      $repl = $1;

# of the form '"/stuff/img.gif"', '"mailto:jens@panix.com"' or '"http://w.a.q/~ww/" tarnish'

      unless ($repl =~ /mailto/) {

         if ($repl =~ /\"([^"]*)\.(gif|jpg|jpeg)\"/i) {
            $tarnish_fn = $1 . "." . $2;
         } else {
            $tarnish_fn = 0;
         }

         if ($repl =~ / ([TV])ARNISH$/i

# this is not quite done so disable it...
&& 0

) {
            $varnish = $1; if ($varnish =~ /v/i) {$varnish = 1; } else {$varnish = 2; }
         } else {
            $varnish = 0;
         }

         if ($tarnish_fn || $varnish) {
   
            $tarnish_h = 0;
            $tarnish_w = 0;
            $tarnish_ctl = 1;
         $repl =~ s/$/ target=\"newjj\" onClick=\"tarnish($tarnish_h, $tarnish_w, $tarnish_ctl)\" /i;
         $imgnew_used = 1;
         if ($tarnish_ctl != 0) {$tarnish_ctl_used = 1; }
         }
      }
      $page =~ s/{PLACEHOLDER}//;
   }

# oh my what a fucking kludge..

   $page =~ s/</</g; $page =~ s/>/>/g;

} # else replace TARNISH with '' but who cares

# moved heavy files..
$page =~ s#src="/stuff/([^"]+)"#src="http://jens.org/stuff/$1"#ig;

$page =~ s#href="/stuff/([^"]+\.(mp3|wav|mid|gif|jpg|txt|troff|ps|zip|exe|c))"#href="http://jens.org/stuff/$1"#ig;

# Lastly f we're on panix, transform '="/' into '="/~jens/'
if ($unix) {$page =~ s/="\//="\/~jens\//g;}

#
# add whatever JavaScript that is necessary
#

if ($cookie{'img'} ne 'same') {
   $javascript = "if (window.name != \"newjj\") {window.name = \"jj\";}

";
}

# stripped comment
# // open a window

if ($imgnew_used) { # JavaScript routine required
   $javascript .= "var wo = new Object();
function tarnish(width, height, controls) {
 var attributes, attributes1;
 attributes = 
    'resizable=yes,scrollbars=no,location=no,menubar=no,toolbar=no,status=no';" .
($tarnish_controls_used ? "
 if (controls) {attributes = 
   'resizable=yes,scrollbars=yes,location=yes,menubar=yes,toolbar=yes,status=yes';}" 
: "" ) . "
 if (width == 0) {width = screen.width*2/3; }
 if (height == 0) {height = screen.height*2/3; }
 attributes = attributes + ',width=' + width + ',height=' + height;
 wo = window.open('', 'newjj', attributes)
" .
# // wo.document.write('<html><head><title>newjj</title></head><body 
# ${body_attr_fn}>wait..</body></html>');
# // wo.document.close();
" wo.window.focus();
}

";
}

# 
# stripped comments..
# // Jens J's footnote handler v 1.0. (from http://www.panix.com/~jens)
# // Free use if you leave these lines intact! What a deal!
# // winobj.moveTo(event.screenX + 10, event.screenY + 10);
# // here we decode footnote encoded by perl cgi source like:
# // s/\\s/ /g; s/(.)/sprintf('%02X', ord(&#36;1))/eg;
# // (simply set text_dec to text instead, if it's plain text)
# // user could close it

if ($footnotes_used) { # JavaScript routine required
   unless ($cookie{'foot'} eq 'inline' && $cookie{'foot'} eq 'href') { 
      $javascript .= "defaultStatus = 'Move cursor over green link to see footnote.';
"
   }

   if ($msie) {
     $fn_winattr_calc = "fnx = event.screenX+10; fny = event.screenY+10;";
     $fn_winattr = "',left=' + fnx + ',top=' + fny";
   } else {
     $fn_winattr_calc = "fnx = event.screenX-window.pageXOffset+5; fny = event.screenY-window.pageYOffset+5; // NS BUG!!!?";
     $fn_winattr = "',screenX=' + fnx + ',screenY=' + fny";
   }

   $javascript .= "var winobj = new Object();
var fn_open;

function footnote (text, event) {
 var text_dec = ''; 
 var fnx, fny;
 if (fn_open) {return true; } 
 fn_open = 1;
 window.status = '***  PLEASE WAIT, GENERATING FOOTNOTE  ***';

 for(i = 0; i < text.length; i+=2) {
  text_dec += String.fromCharCode(parseInt(text.substr(i, 2), 16));
 }

 var dimension = 30 + 10 * Math.sqrt((text_dec.length + 50));
 var attributes = 'resizable=yes';
 hdimension = (dimension); wdimension = (dimension * 1.15);
 $fn_winattr_calc
 attributes = attributes + ',width=' + wdimension + ',height=' + hdimension;
 attributes = attributes + ${fn_winattr};

 winobj = window.open('', 'Fn', attributes);
 winobj.document.write(
 '<HTML><HEAD><TITLE>Fn</TITLE></HEAD><BODY${body_attr_fn}>\\n' +
 '$font_spec' +
 '<font size=-1><P><center>' + text_dec + '</P></font>\\n' +
 '<font size=-2>(move the mouse to remove footnote)</font></center>' +
 '$font_spec_end');
 winobj.document.close();
 window.status = '';
 return true;
}

function footnoteout () {
 fn_open = 0;
 if (!winobj.closed)
  winobj.close();
 return true;
} 

";
}

if ($javascript && (! $lynx )) {
   $prepage .= "
<script language='JavaScript'><!--

" . $javascript .
   "//--></script>

";
}

$page =~ s/(href|action)="(\/~jens)?\/parse.cgi\/([#\w.-]+\.par)\^([^"]*)"/href="$2\/$3?$4"/ig;
$page =~ s/(href|action)="(\/~jens)?\/parse.cgi\/([#\w.-]+\.par[^"]*)"/href="$2\/$3"/ig;

#$page =~ s#/parse.cgi/#/#g;

#
# Output the processed page
#

$page = $prepage . $page;

if (1) { 
   if ($unix) {
      $html_head .= "Content-length: " . length($page) . "\n\n";
   } else {
      # arrrgghh fucking braindead DOS newlines use 2 bytes once it's piped!
      $nlcount = length($page);
      for ($i=0; $i<length($page); $i++) {if (substr($page, $i, 1) eq "\n") {$nlcount++;}}
      $html_head .= "Content-length: " . $nlcount . "\n\n";
   }
   print $html_head;
}

print $page;

# better efficiency?? instead of regex-matching which also doesn't allow nesting..
#
#  &excise($string, "{opendelim", "closedelim}") -- 
#     returns status and $string split in 3, while tracking nested open/closedelims 
#
# eg.
#  &excise("aaaaaaaaaa{bla blurk plurrk {{-}} alb}zzz", "{", "}") returns
#    (1, "aaaaaaaaaa",
#    "zzz",
#    "bla blurk plurrk {{-}} alb")
#
#
# it seems to work!?.. !
#
# but how deal with | or other delimiters in matched string?
#

sub excise() {
   local($search_in, $opendelim, $closedelim) = @_;
   local($i, $rlen, $llen, $opendelim_count);
   $lmatch = index($search_in, $opendelim);
   if ($lmatch == -1) { return (0, '', '', ''); }
   $rlen = length($closedelim); $llen = length($opendelim);
   $opendelim_count = 1;
   for ($i=$lmatch+$llen; $i<=length($search_in)-$rlen && ($opendelim_count); ) {
      if (substr($search_in, $i, $llen) eq $opendelim) 
         {$opendelim_count++; $i+=$llen; next; }
      if (substr($search_in, $i, $rlen) eq $closedelim) 
         {$opendelim_count--; $rmatch = $i; $i+=$rlen; next; }
      $i++;
   }

   if ($opendelim_count) {
#      die "matching ".$opendelim."/".$closedelim." error";
      return (0, "matching ".$opendelim."/".$closedelim." error", '' ,''); 
   }

   ( 1, 
     substr($search_in, 0, $lmatch),
     substr($search_in, $rmatch + $rlen, length($search_in)-1),
     substr($search_in, $lmatch + $llen, $rmatch-$lmatch-$llen)
   )
}

# how to deal with | or other delimiters in matched string
#
#  &split_outer($string, "split_char", "{opendelim", "closedelim}") -- 
#     returns $string split at split_char, while tracking nested open/closedelims 
#
#  &split_outer("aaaaaaaaaa{bla blurk |plurrk [-lds|dd] {{-|-}} alb}zzz", "|", "{[", "}]") 
#  returns:
#    "aaaaaaaaaa{bla blurk" 
#    "plurrk [-lds|dd] {{-|-}} alb}zzz"
#

# @resulu = 
# &split_outer("abcdef|aaa{bla blurk |plurrk [-lds|dd] {{-|-}} alb}zzz", "|", "{[", "}]");
# print "(" . join(")<br>(", @resulu) . ")";

sub split_outer() {
   local($search_in, $split_char, $opendelim, $closedelim) = @_;
   local(@result, $len, $opendelim_count, $curchar, $j);
   local($i) = 0;
   $result[$i] = $search_in;

reloop:
   $len = length($result[$i]);

   for($j = 0; $j < $len; $j++) {
      $curchar = substr($result[$i], $j, 1);

      if ($curchar eq $split_char && !$opendelim_count) {
         $result[$i+1] = substr($result[$i], $j+1);
         $result[$i] = substr($result[$i], 0, $j);
         $i++;
         goto reloop;
      }
      if (index($opendelim, $curchar) != -1) {$opendelim_count++; }
      if (index($closedelim, $curchar) != -1) {$opendelim_count--;}
   }

#print "<br>(". join(")<br>(", @result) . ")";
   @result;
}

__END__