[Pyrex] Structures containing pointers to Python objects

[Joachim Saul]

* Konrad Hinsen [2004-02-01 13:47]:
> Suppose I use the NumPy interface code from the Pyrex FAQ:
>
> cdef extern from "Numeric/arrayobject.h":
>
>     struct PyArray_Descr:
>         int type_num, elsize
>         char type
>
>     ctypedef class Numeric.ArrayType [object PyArrayObject]:
>         cdef char *data
>         cdef int nd
>         cdef int *dimensions, *strides
>         cdef object base
>         cdef PyArray_Descr *descr
>         cdef int flags
>
> How do I then write a declaration for a struct that contains a pointer
> to a NumPy array? Pyrex won't accept "ArrayType" or "ArrayType *"
> inside a struct declaration, nor "PyArrayObject *".

Do you actually mean a struct or a C class? For the latter, what I
do is something like

cdef public class cTrace [type Trace_Type, object TraceObject]:

    cdef ArrayType _data    # note that it is *not* declared public

    property data:

        def __get__(self):
            return self._data
        def __set__(self, ArrayType data):
            self._data = data

This is partly a workaround, because it is currently not possible
to declare

    cdef public ArrayType data

and instead one would have to write

    cdef public object data

I therefore use the property 'data' to access the private '_data';
the __set__ takes care of the type checking and in the real-life
code performs some additional tasks. In all examples, [_]data is
actually a pointer to the array.

Is this what you were looking for?

Cheers,
Joachim