#!/usr/bin/perl -w
#
# latex2utf8txt, Jun 18 2005 Peter Csizmadia
#

use strict;

if($#ARGV < 0) {
	print <<EOF;
Usage: latex2utf8txt file.tex
EOF
	exit 0;
}

#my $outenc = shift @ARGV;

my $paragraph = "";

sub convert {
	my $s = shift;
	$_ = $s;
	s/\\DeclareFontFamily\{\w+\}\{\w+\}\{\w*\}//g;
	s/\\DeclareFontShape\{\w+\}\{\w+\}\{\w*\}\{\w*\}\{[^\}]*\}\{[^\}]*\}//g;
	s/\\DeclareMathAlphabet\{[^\}]*\}\{[^\}]*\}\{[^\}]*\}\{[^\}]*\}\{[^\}]*\}//g;
	s/\\catchline\{[^\}]*\}\{[^\}]*\}\{[^\}]*\}\{[^\}]*\}\{[^\}]*\}//g;
	s/\\(markboth)\{(.*)\}\{(.*)\}$/$2\n$3/g;
	s/\\(title|author|address)\{(.*)\}$/$2/g;
	s/\\keywords\{([^\}]*)\}/$1/g;
	s/\\(section|subsection|subsubsection)\{(.*)\}$/$2/g;
	s/\\epsfig\{[^\}]*\}//g;
	s/\\psfig\{[^\}]*\}//g;
	s/\{(\\bf|\\it) (\w+)\}/$2/g;
	s/^``(\w)/"$1/g;
	s/ ``(\w)/ "$1/g;
	s/([\w\.])''$/$1"/g;
	s/([\w\.])'' /$1" /g;
	# incomplete
	s/\{\\'a\}|\\'a/á/g;
	s/\{\\'A\}|\\'A/Á/g;
	s/\{\\`a\}|\\`a/à/g;
	s/\{\\`A\}|\\`A/À/g;
	s/\{\\^a\}|\\^a/â/g;
	s/\{\\^A\}|\\^A/Â/g;
	s/\{\\'e\}|\\'e/é/g;
	s/\{\\'E\}|\\'E/É/g;
	s/\{\\`e\}|\\`e/è/g;
	s/\{\\`E\}|\\`E/È/g;
	s/\{\\^e\}|\\^e/ê/g;
	s/\{\\^E\}|\\^E/Ê/g;
	s/\{\\'\\i\}|\\'\\i/í/g;
	s/\{\\'I\}|\\'I/Í/g;
	s/\{\\`\\i\}|\\`\\i/ì/g;
	s/\{\\`I\}|\\`I/Ì/g;
	s/\{\\^\\i\}|\\^\\i/î/g;
	s/\{\\^I\}|\\^I/Î/g;
	s/\{\\'o\}|\\'o/ó/g;
	s/\{\\'O\}|\\'O/Ó/g;
	s/\{\\`o\}|\\`o/ò/g;
	s/\{\\`O\}|\\`O/Ò/g;
	s/\{\\"o\}|\\"o/ö/g;
	s/\{\\"O\}|\\"O/Ö/g;
	s/\{\\~o\}|\\~o/õ/g;
	s/\{\\~O\}|\\~O/Õ/g;
	s/\{\\^o\}|\\^o/ô/g;
	s/\{\\^O\}|\\^O/Ô/g;
	s/\{\\H o\}|\\H o/ő/g;
	s/\{\\H O\}|\\H O/Ő/g;
	s/\{\\'u\}|\\'u/ú/g;
	s/\{\\'U\}|\\'U/Ú/g;
	s/\{\\`u\}|\\`u/ù/g;
	s/\{\\`U\}|\\`U/Ù/g;
	s/\{\\"u\}|\\"u/ü/g;
	s/\{\\"U\}|\\"U/Ü/g;
	s/\{\\^u\}|\\^u/û/g;
	s/\{\\^U\}|\\^U/Û/g;
	s/\{\\H u\}|\\H u/ű/g;
	s/\{\\H U\}|\\H U/Ű/g;
	return $_;
}

while(<>) {
	chomp;
	s/^\\documentclass\{.*//;
	s/^\\usepackage\{.*\}//;
	s/^\\def\\.*//;
	s/^\\maketitle$//;
	s/^\\begin\{abstract\}$//;
	s/^\\end\{abstract\}$//;
	s/^\\begin\{document\}$//;
	s/^\\end\{document\}$//;
	s/^%.*//g;
	s/^[ \t]+//g;
	s/[ \t]+$//g;
	if(/^$/) {
		if(!($paragraph eq "")) {
			printf("%s\n\n", convert($paragraph));
			$paragraph = "";
		}
	} elsif(/^\\begin\{[a-z]+\}$/ || /^\\end\{[a-z]+\}$/) {
		my $line = $_;
		if(!($paragraph eq "")) {
			printf("%s\n", convert($paragraph));
			$paragraph = "";
		}
		printf("%s\n", $line);
	} elsif(/^\\item / || /^\\bibitem\{/) {
		my $line = $_;
		if(!($paragraph eq "")) {
			printf("%s\n", convert($paragraph));
			$paragraph = "";
		}
		$paragraph = $line;
	} elsif($paragraph eq "") {
		$paragraph = $_;
	} else {
		$paragraph .= " ".$_;
	}
}
if(!($paragraph eq "")) {
	printf("%s\n", convert($paragraph));
}