Fix BP tree corruption due to numpy uints

empress/tools.py

@@ -345,19 +345,23 @@ def shifting(bitlist, size=51):
     Parameters
     ----------
     bitlist: list of int
-        The input list of 0-1
+        The input list of bits (0 or 1). Depending on the version of iow
+        installed, the entries in this list might be of a slightly different
+        type (e.g. np.uint8).
     size: int
         The size of the buffer
 
     Returns
     -------
     list of int
-        Representation of the 0-1s as a list of int
+        Representation of the bits as a list of int. Regardless of the types in
+        the input bitlist, the entries in the output list will always have type
+        int.
 
     Raises
     ------
     ValueError
-        If any of the list values is different than 0 or 1
+        If any of the entries in bitlist is not equal to 0 or 1.
 
     References
     ----------
@@ -370,6 +374,11 @@ def shifting(bitlist, size=51):
     if not all(x in [0, 1] for x in bitlist):
         raise ValueError('Your list has values other than 0-1s')
 
+    # Convert all the bits to ints -- if they are np.uint8, etc. instead and
+    # we don't convert them, then this can cause problems: see
+    # https://github.com/biocore/empress/issues/562
+    bitlist = [int(x) for x in bitlist]
+
     values = [iter(bitlist)] * size
     ints = []
     for num in zip_longest(*values):

tests/python/test_taxonomy_utils.py

@@ -57,9 +57,9 @@ def setUp(self):
         )
 
     def _check_basic_case_worked(self, split_fm, taxcols):
-        """Checks that a given DataFrame (and list of split-up taxonomy columns)
-           matches the expected output from running split_taxonomy() on
-           self.feature_metadata.
+        """Checks that a given DataFrame (and list of split-up taxonomy
+           columns) matches the expected output from running split_taxonomy()
+           on self.feature_metadata.
         """
 
         # Let's verify that split_fm looks how we expect it to look.

tests/python/test_tools.py

@@ -583,6 +583,22 @@ def _count_bits(n):
                                     "than 0-1s"):
             tools.shifting([10])
 
+    def test_shifting_np_uint8(self):
+        # Verifies that https://github.com/biocore/empress/issues/562 is fixed.
+        # Checks that, whether the inputs are ints or np.uint8s, the output is
+        # the same.
+        bits = [1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 1, 1]
+        np_bits = [np.uint8(b) for b in bits]
+        expected = [4035]
+
+        o1 = tools.shifting(bits)
+        assert o1 == expected
+        assert type(o1[0]) is int
+
+        o2 = tools.shifting(np_bits)
+        assert o2 == expected
+        assert type(o2[0]) is int
+
     def test_filter_feature_metadata_to_tree_1_tip_filtered(self):
         ft, fi = tools.filter_feature_metadata_to_tree(
             self.tip_md, self.int_md, self.shorn_tree