How to test if every element in a numpy array is masked

Question

How do I test if every element in a numpy array is masked? Here's what I'd like to do:

x = #is a maksed numpy array
masked_min = numpy.ma.min(x)
if masked_min IS NOT A MASKED ELEMENT:
  #do some stuff only if masked_min is a value

In practice I see this:

>>> x = numpy.ma.array(numpy.array([1,2,3]),mask=[True,True,True])
>>> masked_min = numpy.ma.min(x)
masked

Testing for masked is not helpful:

>>> numpy.ma.sum(x) == numpy.ma.masked
masked

Answer 1

If by "test if an entire array is masked in numpy" you mean whether every element is masked, since the mask itself an array, you could use .mask.all():

>>> x = numpy.ma.array(numpy.array([1,2,3]),mask=[True,True,True])
>>> x
masked_array(data = [-- -- --],
             mask = [ True  True  True],
       fill_value = 999999)

>>> x.mask
array([ True,  True,  True], dtype=bool)
>>> x.mask.all()
True

or maybe .count(), but that's axis-dependent. OTOH if you really only need to test whether the result of your min call is "masked", you can do that directly:

>>> numpy.ma.min(x)
masked
>>> type(_)
<class 'numpy.ma.core.MaskedConstant'>
>>> isinstance(numpy.ma.min(x), numpy.ma.core.MaskedConstant)
True
>>> numpy.ma.min(x) is numpy.ma.masked
True

[There might be easier ways to do this; I seldom use masked arrays myself.]

Answer 2

I think the most Pythonic way would be to just do what you want, and then catch any errors that result if the entire array is masked.

For example:

x = numpy.ma.array(numpy.array([1,2,3]),mask=[True,True,True])
try: 
    my_min = numpy.min(x[~x.mask])
    # Continue with my_min
except ValueError: 
    # Bail or what have you
    print 'Masks all around!'