Internationalization patch, i18n, iconv, ...

[js9s at netscape.net]

I have something like uid=someone,ou=火,dc=domain,dc=org. The ou value is in gb2312 encoding.  If I use PyString_(En/De) the whole dn string, OpenLDAP doesn't like it.  So I have to parse out the ou=, and only PyString_(En/De) that piece, which make things quite bit more complicated, and I did not spend time to figure out why. I choose a quick and easy(dirty ?) way out for myself, and get this thread started. I will certainly vote for someone who can device an effecient and clean way to do it cross platforms. 

Michael Ströder <michael at stroeder.com> wrote:

>js9s at netscape.net wrote:
>> 
>> The iconv is part of glibc 2.2.2 and above, so iconv.h should be in
>> any recent LINUX system
>
>web2ldap based on python-ldap aims to run on Win32 platform too...
>
>Anyway all character set translation should be based on Unicode
>support introduced in Python 1.6 because the codec concept looks
>very clean. There are Unicode codecs for Python based on iconv lib
>if the built-in Python codecs are not sufficient.
>
>> I did try to use PyString_(De/En)code, but it is not very well
>> suited to openldap type of applications, especially if dn has mixed
>> ascii and utf8 encodings.
>
>I'm not sure what you mean with "mixed ascii and utf8 encodings".
>
>UTF-8 is a variable length encoding of the ISO-10646 character set


>which maps the page containing the ASCII characters directly to
>ASCII.
>
>Example with my last name:
>
>>>> unicode('Str?der','iso-8859-1').encode('utf-8')
>'Str\xc3\xb6der'
>
>Ciao, Michael.
>