#!/usr/bin/perl -w

# mathinhtml is copyright (C) 2003 Michael J Miller
# This program is free software; you can redistribute it and/or modify it
# under the terms of the GNU General Public License.  This program is 
# distributed without any warranty (not even the implied warranty of 
# merchantability or fitness for a particular purpose).

use strict;
use File::Basename;
my $programname="mathinhtml v0.51 (16-Feb-2003)";

#######################################################################
# This section establishes the rules by which mathinhtml translates 
# the contents of an element into the requested image/text.

my %rules;

$rules{"TEX"}={
   input1 => "\\nopagenumbers _CONTENTS_ \\bye",
   action1 => "tex -interaction=batchmode tempfile.1",
   action2 => "dvips -E -q tempfile.dvi -o tempfile.ps",
   density => 500, 
   attrib => 'ALIGN="CENTER"'
};

$rules{"EQ"}={%{$rules{"TEX"}}};
$rules{"EQ"}{"input1"}="\\nopagenumbers \$\$ _CONTENTS_ \$\$ \\bye";

my $ps_setup="set term postscript eps enhanced 24; set output 'tempfile.ps';";
$rules{"GNUPLOT"}={
   input1 => "$ps_setup _CONTENTS_ ",
   action1 => "gnuplot tempfile.1",
   density => 200, 
   attrib => 'ALIGN="TOP"'
};

$rules{"GR"}={%{$rules{"GNUPLOT"}}};
$rules{"GR"}{"input1"}="$ps_setup; set zeroaxis; set xtics axis;".
             "set ytics axis; set noborder; set nokey; _CONTENTS_ ";

$rules{"EQN"}={ 
   input1 => "_CONTENTS_ ", 
   action1 => "tex2html tempfile.1" 
};

my @tempfiles=qw/ tempfile.1 tempfile.dvi tempfile.log tempfile.ps /;

###############################################################
# This section defines the subroutines "translate" (which translates 
# an element to its corresponding image/text) and "replace" (which 
# replaces an element by its translation).

my %archive;  # This hash catalogs those image files that have already been
              # created, so that existing images will not be recreated.

my $tagnames=join '|', keys %rules;

sub translate {
   my $key=$_[0];
   my ($tagname, $contents)=split /\t/, $key;

   for (my $i=1; defined $rules{$tagname}{"input$i"}; $i++){
      my $input=$rules{$tagname}{"input$i"};
      $input =~ s/_CONTENTS_/$contents/g;
      open FILE, "> tempfile.$i";  print FILE "$input\n";  close FILE;
  }

   print "Translating $key\n";
   my $output;
   for (my $i=1; defined $rules{$tagname}{"action$i"}; $i++){
           $output=qx/ $rules{$tagname}{"action$i"} /;
   }


   if (-s "tempfile.ps"){
       my $num=0;  while (-s "IMG_$num.png") { $num++ }; 
       my $imgname="IMG_$num.png";
       my $density=$rules{$tagname}{"density"};

       system "convert -density $density tempfile.ps $imgname";

       if (-s "$imgname.0") {  #This is a workaround for a dvips bug. 
	   rename "$imgname.0", $imgname; unlink "$imgname.1";
       };

       my ($width, $height);
       open FILE, "< $imgname";  seek (FILE, 16, 0);
       read (FILE, $width, 4);  read (FILE, $height, 4);  close FILE;
       $width=unpack 'N', $width;  $height=unpack 'N', $height;

       $archive{$key}= {filename=>$imgname, width=>$width, height=>$height };
   }
   else { 
      $output =~ s/\s+/ /g;  $archive{$key}=$output 
   };

   unlink @tempfiles;
}



sub replace {
   my $qstr=q/(?:"[^"]*?"|'[^']*?')/;  # Matches a quoted string

   my ($tagname, $attributes, $contents)=
      $_[0] =~ m|^<($tagnames)((?:\s+\w+=$qstr)*)\s*>(.*?)</\1>$|is;
   $contents =~ s/\s+/ /g;
   my $key="$tagname\t$contents";

   if (! exists $archive{$key}){ &translate($key) };

   if (ref $archive{$key}){
      for ($contents) {
         # replace "forbidden" html characters
         s/&/&amp;/g;  s/</&lt;/g;  s/>/&gt;/g;  s/"/&quot;/g;  s/'/&#39;/g;
         # replace backslash with space (may make TeX contents more readable)
         s/\\/ /g;
      };
      $attributes=" SRC=\"$archive{$key}{'filename'}\"".
                  " WIDTH=\"".($archive{$key}{'width'}/4)."\"".
                  " HEIGHT=\"".($archive{$key}{'height'}/4)."\"".
                  " ALT=\"$contents\" TITLE=\"$contents\"".
                  " $rules{$tagname}{'attrib'} $attributes";
      while ($attributes =~ s/\s(\w+)=$qstr(.*?)\s\1=($qstr)/ $1=$3 $2 /is){};
      for ($attributes){ s/\s+/ /g; s/ $// };
      return "<IMG $attributes>";
   }
   else { return $archive{$key} };
}


##############################################################
# This section defines the subroutines readarchive and writearchive,
# which synchronize the hash %archive with the file archive.log.
# The format of the file archive.log is tab-separated lines of
#    filename   width   height   tagname   contents   output

sub readarchive {
   if (open(ARCHIVELOG, "< archive.log")) {
      while ( <ARCHIVELOG> ){
         chomp;
         my ($filename, $width, $height, $tagname, $contents, $output)=
	     split /\t/;
         if ($filename ne "" && -s $filename) { 
            $archive{"$tagname\t$contents"}=
               { filename => $filename, width => $width, height => $height };
         }
         elsif (defined($output)) {  
            $archive{"$tagname\t$contents"}=$output 
         };
      };
   };
   close(ARCHIVELOG);
}

sub writearchive {
   open(ARCHIVELOG, "> archive.log");
   foreach my $key (keys %archive){
      if (ref $archive{$key}){
         print ARCHIVELOG "$archive{$key}{'filename'}\t".
                        "$archive{$key}{'width'}\t".
                        "$archive{$key}{'height'}\t$key\n"
      }
      else {
         print ARCHIVELOG "\t\t\t$key\t$archive{$key}\n";
      };
   };
   close ARCHIVELOG;
}

#################################################################
# This is the main executable.

print "\n$programname\nCopyright (C) 2003 Michael J Miller\n".
      "This program comes with absolutely no warranty.\n\n";

chdir dirname $ARGV[0];
&readarchive;

my $filename=basename $ARGV[0];

open(SOURCEFILE, "< $filename");  undef $/;  $_=<>;  close SOURCEFILE;
s|</HEAD>|\n<META NAME="translator" CONTENT="$programname">\n</HEAD>|i;
my @sections= m/(.*?)(<P>|<BR>|<LI>|<TD>|.$)/igs;

$filename =~ s/\.mih$//;
open(HTMLFILE, "> $filename.html");

foreach (@sections){
   s|<($tagnames)[\s>].*?</\1>|&replace($&)|iesg;
   print HTMLFILE "$_";
}
close HTMLFILE;

&writearchive;