#!/usr/bin/perl -w



print "\nThis is HTML2ConTeXt. Version 2002-05-15\n";

print "I'll try to convert your HTML file for ConTeXt.\n";

print "copyleft Henning Hraban Ramm, http://www.fiee.net/texnique/\n\n";



unless ($ARGV[0]) {die "You must name a file to convert! $!"};

my $HTMLDatei = $ARGV[0];

unless (-T $HTMLDatei) {

	print "$HTMLDatei not found!\n";

	if (-T $HTMLDatei.'.htm')  { $HTMLDatei .= ".htm"; }

	if (-T $HTMLDatei.'.html') { $HTMLDatei .= ".html"; }

} # unless

$HTMLDatei =~ s/\\/\//g;



my $Table="n";

my $Encod="win";



open (QUELLE, $HTMLDatei) or die "Can't open $HTMLDatei! $!";



my $TeXDatei = $HTMLDatei;

$TeXDatei =~ s/\.htm$/\.tex/i;

print $TeXDatei."\n";

open (ZIEL, ">".$TeXDatei) or die "Can't make $TeXDatei! $!";



while (<QUELLE>) {

# single entities and chars

	s§&(.)uml;§\\\"$1§g;

	s§&(.)acute;§\\´$1§g;

	s§&(.)grave;§\\`$1§g;

	s§&(.)circ;§\\^$1§g;

	s§&(.)ring;§\\°$1§g;

if ($Encod eq "win") {

	s§&szlig;§ß§g;

} else {

	s§&szlig;§\\ss{}§g;

} # if Encoding

	s§&(\#150|endash);§--§g;	# endash

	s§ - § -- §g;			# endash

	s§&nbsp;§~§g;	# non breaking space

	s§&quot;([^<>]*)&quot;§\\quotation{$1}§g;

	s§&(r|l)aquo;([^<>]*)&(l|r)aquo;§\\quotation{$2}§g;

	s§&\#132;([^<>]*)&quot;§\\quotation{$1}§g;

	s§&\#132;([^<>]*)$§\\quotation{$1§g; # uncompleted line

	s§\s(&quot;|\")§ \\quotation{§g;	# begin quote

	s§&quot;\s§} §g;	# end quote

	s§&quot;§\"§g;		# quote

	s§([^\\=\s])\"§$1}§g;	# end quote

	s§%§|~|\\%{}§g;	# percent

	s§&lt;§<§g;

	s§&gt;§>§g;

	s§&amp;§\&§g;

	s§&sup(.);§^$1§g;

	s§&frac(.)(.);§\\frac{$1}{$2}§g;

	s§&\#133;§ §g;

#	s§§§g;

#	s§§§g;



# TeX words and marks

	s§T<SUB>E</SUB>X§TeX§g;

	s§pdfTeX§\\pdfTeX{}§gi;

	s§ppchTeX§\\pdfTeX{}§gi;

	s§ConTeXt§\\ConTeXt{}§g;

	s§CONTEXT§\\ConTeXt{}§g;

	s§(\s)TeX§$1\\TeX{}§g;



# environments

	s§<BODY[^<>]*>§\\starttext§gi;

	s§</BODY>§\\stoptext§gi;

	s§(<BLOCKQUOTE>|<QUOTE>)§\\startquotation§gi;

	s§(<\/BLOCKQUOTE>|<\/QUOTE>)§\\stopquotation§gi;

	s§</*DIV[^<>]*>§§gi;	# delete all divs

	s§</*FONT[^<>]*(>|$)§§gi;	# delete all font tags



# Headers

	s§<H1>§\\chapter{§gi;

	s§<H2>§\\section{§gi;

	s§<H3>§\\subsection{§gi;

	s§<H4>§\\subsubsection{§gi;

	s§</H.>§}§gi;



# Links

	s§<A\s(.*)HREF=\"(.*)\">(.*)</A>§\\goto{$3}[URL($2)]§gi;

	s§<A\s(.*)NAME=\"(.*)\">(.*)</A>§\\reference[$2]{$3}§gi;



# Tables

if ($Table eq "y") {

	s§<TABLE([^<>]*)>§\\bTABLE \%$1 §gi;

	s§</TABLE>§\\eTABLE§gi;

	s§</TD>§\\eTD §gi;

	s§<TD([^<>]*)>§\\bTD §gi;

	s§</TR>§\\eTR §gi;

	s§<TR([^<>]*)>§\\bTR §gi;

} else {

	s§</*T(ABLE|D|R|BODY)[^<>]*>§§gi;	# delete all table tags

} # if Table



# Images

	s§<IMG\s([^<>]*)>§\\externalfigure[$1]§gi;

	s§<IMG\s([^"=]*)src=\"([^<>]*)\"([^<>]*)$§\\externalfigure[$2]\t\% $1 $3§gi;



# Lists

	s§<UL>§\\startitemize\[1\]§gi;

	s§<OL>§\\startitemize\[n\]§gi;

	s§<DL>§\\startitemize\[1\]§gi; # ?

	s§</.L>§\\stopitemize§gi;

	s§<LI>§\\item §gi;

	s§<DT>§\\item §gi; #

	s§<DD>§\\item §gi; #

	s§</LI>§§gi;





	s§<P[^<>]*>§§gi;

#	s§</P>§\\par§gi;

	s§</P>§\n\n§gi;

	s§<BR[^<>]*>§\n§gi;

	s§<HR[^<>]*>§\\blank §gi;



	s§<(PRE|TT|CODE)>§\\type{§gi;

	s§<(STRONG|B)>§{\\bf §gi;

	s§<(EM|I|U)>§{\\em §gi;



	s§^</([^\s]*)>$§\\stop$1§gi;

	s§^<([^\s]*)([^<>]*)>$§\\start$1\[$2\]§gi;

	s§</.*>§}§gi; # all other closing tags become }

	s§<([^\s]*)(\s)(.*)>§\\$1\[$3\]\{§gi; # all other opening tags become {

	s§<([^\s]*)>§\\$1\{§gi; # all other opening tags become {



	s§^\s*§§g;	# remove trailing spaces



	print ZIEL;

	print ".";

} # while

print "\n";



close (ZIEL);

close (QUELLE);





# \goto{text}[URL(Link)]