windows-server-2003/inetcore/mshtml/tools/unicodedirection_verify.pl

#
# dirclass_verify.pl
#
# [email protected]
# 31 Jul 1998
#
# Checks the validity of the directional classifications file generated by
# dirclass_extract.pl.
#

%dirlist = ();

$last = -1;

while (<>)
{
    /^([0-9a-fA-F]{4}) (\w\w\w)/;
    $char = hex($1);
    $dir = $2;

    if ($char != $last + 1)
    {
        $error = sprintf("List out of order at character %d\n", $char);
        die($error);
    }
    if (!($dir =~ /^(UNK|LTR|RTL|ARA|WSP|NEU|SEG|BLK|ANM|ENM|ETM|ESP|CSP|FMT|CBN)$/))
    {
        $error = sprintf("Unknown classification '%s' for character %d\n", $dir, $char);
        die($error);
    }
    $dirlist{$dir}++;
    $last++;
}

$cchTotal = 0;
while (($dir, $cch) = each(%dirlist))
{
    printf("%s\t%s\n", $dir, $cch);
    $cchTotal += $cch;
}

printf("\n%d total characters\n", $cchTotal);