How to wrap a c library function that returns raw image data efficiently?

[Vivek Kumar]

Hi, I am working on wrapping an existing C library using cython. The library has an API to get the get the raw image data 

bool getImage(unsigned char* image)

How can I wrap it in a such that it returns a python array which uses the same memory that was allocated in C? Following is my pyx code

def get_image(int width, int height):
 
    cdef float* buffer =

<float*>PyMem_Malloc(width * height * sizeof(unsigned char))

    cdef bint success = getImage(buffer)
   
    return bool(success), buffer // Doesn't compile since not a pyobj

[da-woods]

I'd suggest something like:

def get_image(int width, int height):

    cdef float[:,::1] mview

    arr = np.empty((width, height), dtype=np.float32)

    mview = arr

    success = getImage(&mview[0,0])

    if success:

        return arr

    else:

        return None

I'm allocating memory as a Numpy array (since that's easy to return) and using a memoryview to get a pointer to the data.

Note that I've returned "None or a numpy array" rather than a tuple of success and array, since this seems like a more natural Python interface.

--

---
You received this message because you are subscribed to the Google Groups "cython-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to cython-users...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/cython-users/a8cf51d8-76d9-4e78-85d5-b92475ae7a25n%40googlegroups.com.