[QPID-7109] Python wrapper for qpid-cpp cannot handle unicode data.

• Type: Bug
• Status: Open
• Priority: Major
• Resolution: Unresolved
• Affects Version/s: qpid-cpp-0.34
• Component/s: Python Client (Wrapped)

Environment

Red Hat Enterprise Linux Server release 6.7 (Santiago)

• Assignee: Unassigned
• Reporter: Håkan Johansson
• Created: 26/Feb/16 12:35
• Updated: 26/Feb/16 12:35

Description

The Python wrapper for the qpid-cpp library cannot handle unicode data properly. The reason for this is how unicode objects are encoded in the PyToVariant function, which is located in the "swig_python_typemaps.i" file.

PyToVariant uses the PyUnicode_AS_DATA to encode the data, but that is wrong. It returns the internal representation of the data, which usually is UTF-16, but that depends on how the Python libraries where built. The VariantToPy function correctly uses the PyUnicode_DecodeUTF8 function, but that is not compatible with the data returned from PyUnicode_AS_DATA.

The proper solution is to either use PyUnicode_EncodeUTF8 or PyUnicode_AsUTF8String, including proper error handling.