BUG: Fix IntervalTree handling of NaN (#23353)

Author: jschendel

doc/source/whatsnew/v0.24.0.txt

@@ -1133,6 +1133,7 @@ Interval
 - Bug in the ``IntervalIndex`` repr where a trailing comma was missing after the list of intervals (:issue:`20611`)
 - Bug in :class:`Interval` where scalar arithmetic operations did not retain the ``closed`` value (:issue:`22313`)
 - Bug in :class:`IntervalIndex` where indexing with datetime-like values raised a ``KeyError`` (:issue:`20636`)
+- Bug in ``IntervalTree`` where data containing ``NaN`` triggered a warning and resulted in incorrect indexing queries with :class:`IntervalIndex` (:issue:`23352`)
 
 Indexing
 ^^^^^^^^

pandas/_libs/intervaltree.pxi.in

@@ -72,6 +72,12 @@ cdef class IntervalTree(IntervalMixin):
 
         self.closed = closed
 
+        # GH 23352: ensure no nan in nodes
+        mask = ~np.isnan(self.left)
+        self.left = self.left[mask]
+        self.right = self.right[mask]
+        indices = indices[mask]
+
         node_cls = NODE_CLASSES[str(self.dtype), closed]
         self.root = node_cls(self.left, self.right, indices, leaf_size)
 

pandas/tests/indexes/interval/test_interval_tree.py

@@ -14,10 +14,22 @@ def dtype(request):
     return request.param
 
 
-@pytest.fixture(scope='class')
-def tree(dtype):
-    left = np.arange(5, dtype=dtype)
-    return IntervalTree(left, left + 2)
+@pytest.fixture(params=[1, 2, 10])
+def leaf_size(request):
+    return request.param
+
+
+@pytest.fixture(params=[
+    np.arange(5, dtype='int64'),
+    np.arange(5, dtype='int32'),
+    np.arange(5, dtype='uint64'),
+    np.arange(5, dtype='float64'),
+    np.arange(5, dtype='float32'),
+    np.array([0, 1, 2, 3, 4, np.nan], dtype='float64'),
+    np.array([0, 1, 2, 3, 4, np.nan], dtype='float32')])
+def tree(request, leaf_size):
+    left = request.param
+    return IntervalTree(left, left + 2, leaf_size=leaf_size)
 
 
 class TestIntervalTree(object):