I did some work using XML::Writer to create XML docs from fielded text file. I noticed that it only escaped four or five 'special' chars. e.g. '&' to &.

This was OK until I started encountering some data which was outside the 7bit ascii range that was not escaped and throwing some XML parsing tools. I extended the code to do a general escape of characters outside the 7bit range.

You could use a similar approach if you find that unescaped chars are causing a problem.

sub XML::Writer::_escapeLiteral
{
    # escape 'normal' characters e.g. ampersand
    if ($_[0] =~ /[<>&"']/)
    {
        $_[0] =~ s/\&/&amp;/g;        # ampersand
        $_[0] =~ s/\"/&quot;/g;        # quotes
        $_[0] =~ s/\</&lt;/g;        # left angle
        $_[0] =~ s/\>/&gt;/g;        # right angle
        $_[0] =~ s/\'/&apos;/g;        # apostrophe
        # Add more here...
    }

    # Look for occurrances of chars in the range outside normal 7 bit 
+ascii table
    # Note the character in Octal
    if ($_[0] =~ /([\177-\377])/)
    {
        my $escapedString = "";
        
        # Generic escape for all other chars
        while ($_[0] =~ /\G(.*?)([\177-\377])/gc)
        {
            # concat the escaped char in the form &#123;
            $escapedString .= sprintf ("%s&#%d;", $1, ord($2));
            #print "escapedstring = $1 \n";
        }
        
        # Concat whatever is left
        if ($_[0] =~ /\G(.*)$/g)
        {
            $escapedString .= $1;
        }
        
        return $escapedString;
    }
    
    return $_[0];
}
[download]

Enjoy!
Inman