It's kind a funny behavior in numpy.searchsorted. The following test fail:
import numpy as np
a = np.ma.masked_array([1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16,
17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30,
31, 32, 33, 0],
mask=[False, False, False, False, False, False, False,
False, False, False, False, False, False, False,
False, False, False, False, False, False, False,
False, False, False, False, False, False, False,
False, False, False, False, False, True],
fill_value=0, dtype='uint8')
b = np.array([1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16,
17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 33],
dtype='uint8')
expected = np.array([0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13,
14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27,
28, 29, 32])
c = a.searchsorted(b)
np.testing.assert_array_equal(c, expected)
The last entry in the c array is 34 and I don't know why.
But a similar one, it pass:
aa = np.ma.masked_array([1, 2, 3, 4, 0],
mask=[False, False, False, False, True],
fill_value=0, dtype='uint8')
bb = np.array([1, 3, 4], dtype='uint8')
expectedd = np.array([0, 2, 3])
cc = aa.searchsorted(bb)
np.testing.assert_array_equal(cc, expectedd)
On numpy.array.searchsorted documentation, its description said:
Find the indices into a sorted array a such that, if the corresponding elements in v were inserted before the indices, the order of a would be preserved.
np.searchsorted doesn't yet support masked arrays (see here for a list of supported methods).
You can get the expected result by manually indexing a with the inverse of a.mask, then passing the result as the first argument to np.searchsorted:
c = np.searchsorted(a[~a.mask], b)
# or alternatively, a[~a.mask].searchsorted(b)
print(np.allclose(c, expected))
# True