#!/usr/bin/perl -w
use strict;
$| = 1;

use Unicode::String qw(latin1 utf8);

####################################################################
# Fix Umlaute
# Works on UTF _and_ ISO hosts :-)
#
# -- http://www.candy.pgate.net/shop/.tpl/ger/.cms/fixumlaute_pl.txt
my $VERSION = '0.81.02';
####################################################################


#####  PROTOTYPING
sub Debug($);


#####  GLOBAL
my $DEBUG = 1; # 0, 1, 2 -- ZZ
my $KEEP_OWNER = 'nobody'; # ZZ
my $PREFIX = $ARGV[0] || 'cms_doc_';
my $FILE_COUNT = 0;
$SIG{'__WARN__'} = \&alarm_handler; # install signal handler

&main();

##################################################
sub main() {
##################################################
	opendir(D, '.') or print STDERR $!;
	while (my $f = readdir(D)) {
		next unless $f =~ /^$PREFIX.+\.html?$/;
		next if $f =~ /_bak\d+\.html?$/;

		print "- $f\n";
		my $rc = &checknfixit(file => $f);
		print STDERR "ERROR at $f\n" if $rc;
	}
	closedir(D);

	print "$FILE_COUNT files have been updated.\n";
}

##################################################
sub checknfixit (%) {
##################################################
	my %args = @_;

	my $errstr = '';

	open(IN, "<$args{'file'}"      ) or $errstr .= "(1) $! ";
	open(OUT, ">.tmp_$args{'file'}") or $errstr .= "(2) $! ";

	if ($errstr) {
		print STDERR "ERR E1000: $errstr ($args{'file'})\n";
		return 1;
	}

	my $umlaute_found = 0;
	while (my $in = <IN>) {
		if (is_ascii($in)) {
			print OUT $in; # as is
		}
		else {
			$umlaute_found++;
			my $text_iso  = (utf8($in))->latin1;
			my $text_utf8 = (latin1($text_iso))->utf8; # reverse check

			my $input = $text_iso;
			   $input = $in if $in ne $text_utf8; # Is ISO already!

			print OUT &web_enc($input); # newline included
		}
	}
	close IN;
	close OUT;

	if ($umlaute_found) {
		rename ".tmp_$args{'file'}", $args{'file'} or print STDERR "ERR (rename) $!\n";
		system("chown $KEEP_OWNER $args{'file'}") && print STDERR "ERR $!\n";
		Debug "  Converted Umlaute on $umlaute_found lines.";
		$FILE_COUNT++;
	}
	else {
		unlink ".tmp_$args{'file'}" or print STDERR $!;
		Debug "  Leaving $args{'file'} unchanged.";
	}

	0;
}

sub is_ascii($) {
	my $in = $_[0];
	for (my $i = 0; $i < length($in); $i++) {
		if (unpack('C', substr($in, $i, 1)) > 127) {
			return 0;
		}
	}
	return 1;
}

sub web_enc ($) {
	my $enc = '';
	for (my $i = 0; $i < length($_[0]); $i++) {
		my $ordno = ord substr($_[0], $i, 1);
		$enc .= $ordno > 127 ? sprintf("&#%d;", $ordno) : substr($_[0], $i, 1);
	}

	$enc =~ s/ $//;;
	$enc;
}

sub Debug($) { print "$_[0]\n"; }

sub alarm_handler () {
	#print STDERR "alarm catched!\n";
	return;
}

__END__

E1001

--
/usr/local/apache2/htdocs/www.candy.pgate.net/shop/.tpl/ger/.cms/fixumlaute.pl