#!/usr/bin/perl -w
#
# escapechars, Jun 18 2005 Peter Csizmadia
# (@since Aug 31 2004)
#

use strict;

if($#ARGV < 1) {
	print <<EOF;
Usage: escapechars INPUT_ENCODING OUTPUT_FORMAT [files...]

Input encodings: utf8, latin1, latin2 etc.

Output formats:
\txml - use ampersand Unicode escape sequences
\ttex - use accented characters, partially implemented

Example: escapechars utf8 xml file.utf8.html > file.html
EOF
	exit 0;
}

my $inenc = shift @ARGV;
my $outenc = shift @ARGV;

unshift(@ARGV, '-') unless @ARGV;

while($ARGV = shift) {
	open(ARGV, $ARGV) or die "cannot open file";
	binmode(ARGV, ":encoding(".$inenc.")")
		or die "cannot set input encoding";
	if($outenc eq "xml") {
		while(defined(my $c = getc(ARGV))) {
			if(ord($c) < 128) {
				print $c;
			} else {
				print xmlunicode($c);
			}
		}
	} elsif($outenc eq "tex") {
		# incomplete, only hungarian accents are implemented
		while(<ARGV>) {
			s/\x{e1}/\{\\'a\}/g;
			s/\x{c1}/\{\\'A\}/g;
			s/\x{e9}/\{\\'e\}/g;
			s/\x{c9}/\{\\'E\}/g;
			s/\x{ed}/\{\\'\\i\}/g;
			s/\x{cd}/\{\\'I\}/g;
			s/\x{f3}/\{\\'o\}/g;
			s/\x{d3}/\{\\'O\}/g;
			s/\x{fa}/\{\\'u\}/g;
			s/\x{da}/\{\\'U\}/g;
			s/\x{f6}/\{\\"o\}/g;
			s/\x{d6}/\{\\"O\}/g;
			s/\x{fc}/\{\\"u\}/g;
			s/\x{dc}/\{\\"U\}/g;
			s/\x{f5}/\{\\~o\}/g;
			s/\x{d5}/\{\\~O\}/g;
			s/\x{fb}/\{\\^u\}/g;
			s/\x{db}/\{\\^U\}/g;
			s/\x{151}/\{\\H o\}/g;
			s/\x{150}/\{\\H O\}/g;
			s/\x{171}/\{\\H u\}/g;
			s/\x{170}/\{\\H U\}/g;
			print;
		}
	} else {
		print STDERR $outenc.": unknown output format\n";
		exit 1;
	}
	close(ARGV);
}

sub xmlunicode {
	my ($c) = @_;
	my $k = ord($c);
	return sprintf("&#%d;", $k);
}