Sanitize masked arrays for SeqInfo (#64)

src/genomicranges/SeqInfo.py

@@ -2,7 +2,8 @@
 from warnings import warn
 
 import biocutils as ut
-import numpy
+
+from .utils import _sanitize_vec
 
 __author__ = "jkanche"
 __copyright__ = "jkanche"
@@ -19,7 +20,7 @@ def _validate_seqnames(seqnames):
 
 
 def _validate_seqlengths(seqlengths, num_seqs):
-    if not ut.is_list_of_type(seqlengths, (int, numpy.ndarray), ignore_none=True):
+    if not ut.is_list_of_type(seqlengths, int, ignore_none=True):
         raise ValueError("'seqlengths' should be a list of integers.")
 
     if num_seqs != len(seqlengths):
@@ -31,7 +32,7 @@ def _validate_seqlengths(seqlengths, num_seqs):
 
 
 def _validate_is_circular(is_circular, num_seqs):
-    if not ut.is_list_of_type(is_circular, (bool, numpy.ndarray), ignore_none=True):
+    if not ut.is_list_of_type(is_circular, bool, ignore_none=True):
         raise ValueError("'is_circular' should be a list of booleans.")
 
     if num_seqs != len(is_circular):
@@ -175,6 +176,8 @@ def _flatten_incoming(self, values, expected) -> List:
                     output.append(None)
             return output
 
+        values = _sanitize_vec(values)
+
         if isinstance(values, list):
             return values
 

src/genomicranges/utils.py

@@ -60,6 +60,14 @@ def sanitize_strand_vector(
     )
 
 
+def _sanitize_vec(x: Sequence):
+    if isinstance(x, np.ma.MaskedArray):
+        x.filled(fill_value=None)
+        return x.tolist()
+
+    return list(x)
+
+
 def _sanitize_strand_search_ops(query_strand, subject_strand):
     query_strand = REV_STRAND_MAP[query_strand]
     subject_strand = REV_STRAND_MAP[subject_strand]

tests/test_SeqInfo.py

@@ -64,6 +64,17 @@ def test_create_seqInfo_numpy():
     assert str(ex.value).find("list of strings") >= 0
 
 
+def test_create_seqInfo_numpy_masked():
+    si = SeqInfo(
+        ["chrA", "chrB", "chrC"],
+        np.ma.MaskedArray([10, None, 2200], mask=[0, 1, 0]),
+        np.array([None, True, False]),
+        ["hg19", "hg38", None],
+    )
+
+    assert isinstance(si, SeqInfo)
+
+
 def test_create_empty():
     si = SeqInfo.empty()