On 29/7/04 3:39 pm, Alexander Bergolth <[EMAIL PROTECTED]> wrote:
> Hi!
>
> Since attribute values should be UTF-8 encoded, shouldn't Net::LDIF
Some attributes aren't UTF-8, eg jpegPhoto and userPassword both use OCTET
STRING. Certificates and CRLs (etc) are BER.
> convert values that are "character-strings" explicitly to UTF8?
>
> --- LDIF.pm.orig 2004-07-29 15:45:25.000000000 +0200
> +++ LDIF.pm 2004-07-29 15:44:54.000000000 +0200
> @@ -296,7 +296,9 @@
> my $ln = $lower ? lc $attr : $attr;
> if ($v =~ /(^[ :]|[\x00-\x1f\x7f-\xff])/) {
> require MIME::Base64;
> - $ln .= ":: " . MIME::Base64::encode($v,"");
> + use Encode;
> + $ln .= ":: " .MIME::Base64::encode(
> + Encode::is_utf8($v)? encode_utf8($v) : $v,"");
> }
> else {
> $ln .= ": " . $v;
>
> Otherwise a perl-string that contains only code points up to 0xFF (e.g.
> a string in Latin1) will be internally represented as 8 bit characters
> and will be converted to a Base64 converted Latin1 string, not a Base64
> converted UTF-8 string.
>
> The same applies to all method that communicate with the ldap-server.
> Shouldn't they convert perl character strings to UTF-8 octet-strings
> (like above) when sending data to the server?
No, particularly if you still trying to support LDAPv2.
Cheers,
Chris
