Hi,
I'm trying to compile the python-uno bridge using my system python UCS4
enabled (Debian unstable) from OOo1.1.4 source.
It indeed compile but it does not work because the unicode conversion
between OOo and python are not working correctly. Since I don't know C++
and gdb it's difficult to understand what is going on.
Conversion from OOo -> python seems to be done in pyuno_util in the
function ustring2PyUnicode
For Wide unicode the code is:
OString o = OUStringToOString( str, RTL_TEXTENCODING_UCS4 );
ret = PyRef( PyUnicode_FromUnicode( (Py_UNICODE*)o.getStr()
,o.getLength() ), SAL_NO_ACQUIRE );
Conversion from python -> OOo are in pyuno_runtime in pyObject2Any.
Again for wide unicode:
// fixed for 0.9.2: OUString ctor expects the length of the byte array !
s = OUString( (sal_Char * ) PyUnicode_AsUnicode( o ),
PyUnicode_GetSize( o ) * sizeof(Py_UNICODE),
RTL_TEXTENCODING_UCS4 );
I replaced both conversions by a conversion through UTF8.
In pyuno_util:
OString o = OUStringToOString( str, RTL_TEXTENCODING_UTF8 );
PyObject* utf8 = PyUnicode_DecodeUTF8(o.getStr(), o.getLength(),NULL);
ret = PyRef( utf8, SAL_NO_ACQUIRE );
An in pyuno_runtime:
PyObject* utf8 = PyUnicode_AsUTF8String(o);
s = OUString(PyString_AsString(utf8),PyString_Size(utf8),
RTL_TEXTENCODING_UTF8);
It seems now to work ok. Since there is an additional conversion it may
be a bit slower but I don't think this is the main limiting factor.
My problem is that I have no idea if it can be considered as a bug in
pyuno, in the sal/rtl library or in my Debian system. Did somebody have
any success with UCS4 python?
--
Pierre Martineau
mail: [EMAIL PROTECTED]
Bibus home: http://bibus-biblio.sourceforge.net/
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
