numpy: circular definition in source code

Question

I went to look up the source to np.transpose ( source ) and the definition seems circular?

def transpose(a, axes=None):

    try:
        transpose = a.transpose
    except AttributeError:
        return _wrapit(a, 'transpose', axes)
    return transpose(axes)

If all transpose(a) does is call a.transpose then how do we look up a.transpose?

---

What part of the code is doing the transposition? All I see is referral to another transpose function.

Answer 1

Here's the code for _wrapit:

File:       /usr/local/lib/python2.7/site-packages/numpy/core/fromnumeric.py
Definition: numpy.core.fromnumeric._wrapit(obj, method, *args, **kwds)
Source:
def _wrapit(obj, method, *args, **kwds):
    try:
        wrap = obj.__array_wrap__
    except AttributeError:
        wrap = None
    result = getattr(asarray(obj), method)(*args, **kwds)
    if wrap:
        if not isinstance(result, mu.ndarray):
            result = asarray(result)
        result = wrap(result)
    return result

Answer 2

It's not actually a circular reference. a.transpose is a reference to the object's method, not the function defined by numpy. It's effectively saying "If object a already has a transpose method, then leave it alone; otherwise, use _wrapit to wrap the object a in an ndarray object".

As the ndarray class has a transpose method, forcing the Python object into that class gives the object access to the method.

This is a little outside my area of expertise, but it would appear that ndarray is defined in the C portion of the numpy code, so that would be where you'd find the actual logic behind it.