Explore our sites

Stack Exchange
tour help careers 2.0
Take the 2-minute tour ×
Stack Overflow is a question and answer site for professional and enthusiast programmers. It's 100% free, no registration required.
up vote 1 down vote favorite

I am attempting to subclass a numpy structured array so that I can add special methods to the subclass. Everything works fine until I retrieve a single index from the array. When this occurs, an object of type numpy.void is returned instead of the subclass's type. (Actually, whether subclassing or not, retrieving a single index from a structured array returns an object of type numpy.void.) Why is this? And how can I ensure that an instance of my class is returned? I assume that overriding __getitem__ would be the way to go, but I'm not familiar enough with ndarray subclassing to be confident that I would not screw up something else. Please advise.

Here is an example of the behavior I am describing:


import numpy as np

# The ndarray subclass
class Foo(np.ndarray):

    # Do something special that uses a field of the structured array
    def bar(self):
        return self['BAR']


def main():
    # Set up the structured array
    arr = np.arange((3+2)*2,dtype=np.float64).view(dtype=np.dtype([('BAR',np.float64,3),('other',np.float64,2)]))

    # Get a Foo instance using the data
    obj = arr.view(Foo) 

    print 'type(obj):     ',type(obj)      # As expected: Foo object
    print 'type(obj[:1]): ',type(obj[:1])  # As expected: Foo object
    print 'type(obj[0]):  ',type(obj[0])   # Why numpy.void???

    print 'obj.bar():'                     # As expected
    print obj.bar()                        # As expected
    print 'obj[:1].bar():',obj[:1].bar()   # As expected
    print 'obj[0].bar(): ',obj[0].bar()    # Causes exception: AttributeError: 'numpy.void' object has no attribute 'bar'


if __name__=="__main__":
    main()

The output is as follows:


type(obj):      <class '__main__.Foo'>
type(obj[:1]):  <class '__main__.Foo'>
type(obj[0]):   <type 'numpy.void'>
obj.bar():
[[ 0.  1.  2.]
 [ 5.  6.  7.]]
obj[:1].bar(): [[ 0.  1.  2.]]
Traceback (most recent call last):
  File "/Library/Frameworks/Python.framework/Versions/7.3/lib/python2.7/site-packages/ipdb/__main__.py", line 138, in main
    pdb._runscript(mainpyfile)
  File "/Library/Frameworks/Python.framework/Versions/7.3/lib/python2.7/pdb.py", line 1233, in _runscript
    self.run(statement)
  File "/Library/Frameworks/Python.framework/Versions/7.3/lib/python2.7/bdb.py", line 387, in run
    exec cmd in globals, locals
  File "", line 1, in 
  File "scratch.py", line 1, in 
    import numpy as np
  File "scratch.py", line 25, in main
    print 'obj[0].bar(): ',obj[0].bar()    # Causes exception: AttributeError: 'numpy.void' object has no attribute 'bar'
AttributeError: 'numpy.void' object has no attribute 'bar'
obj[0].bar():  Uncaught exception. Entering post mortem debugging
Running 'cont' or 'step' will restart the program
share|improve this question
add comment

2 Answers

up vote 1 down vote

The error has nothing to do with superclass, it comes from view and custom dtype:

>>> arr = np.arange((3+2)*2,dtype=np.float64).view(dtype=np.dtype([('BAR',np.float64,3),('other',np.float64,2)]))
>>> arr[0]
([0.0, 1.0, 2.0], [3.0, 4.0])
>>> type(arr[0])
<type 'numpy.void'>
share|improve this answer
 
That is what I meant by "(Actually, whether subclassing or not, retrieving a single index from a structured array returns an object of type numpy.void.)" However, my context is subclassing, and I would like to know how best to alter my code to give an object that maintains the Foo interface. –  benpro Nov 18 at 20:25
add comment
up vote 0 down vote

Try:

print 'type(obj[[0]]):  ',type(obj[[0]])   # Why numpy.void???
print 'obj[[0]].bar(): ',obj[[0]].bar()

You are seeing the difference between accessing with a scalar index and a list (array) one.

From the user guide for record arrays

Here we have created a one-dimensional array of length 2. Each element of this array is a record that contains three items, a 32-bit integer, a 32-bit float, and a string of length 10 or less. If we index this array at the second position we get the second record:

obj[0] returns a record, not any array. obj[[0]] returns an array with one record.

Likewise, x=np.array([0,1,2,3]); type(x[0]) is not an array; it's a scalar (dtype).

share|improve this answer
add comment

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.