[python][RDF] Add tests to AsRVec and FromNumpy for data of type string

Author: siliataider

bindings/pyroot/pythonizations/test/rdataframe_makenumpy.py

@@ -24,6 +24,19 @@ def test_dtypes(self):
             df = ROOT.RDF.FromNumpy(data)
             self.assertEqual(df.Mean("x").GetValue(), 2)
 
+    def test_object_dtype_strings(self):
+        """
+        Test reading numpy arrays with dtype=object containing strings
+        """
+        data = {"x": np.array(["test_string_1", "test_string_2"], dtype=object)}
+        df = ROOT.RDF.FromNumpy(data)
+        colnames = df.GetColumnNames()
+
+        self.assertIn("x", colnames)
+
+        entries = df.AsNumpy()["x"]
+        self.assertTrue(np.array_equal(entries, np.array(["test_string_1", "test_string_2"], dtype=object)))
+
     def test_multiple_columns(self):
         """
         Test reading multiple columns

bindings/pyroot/pythonizations/test/rvec_asrvec.py

@@ -49,6 +49,34 @@ def test_dtypes(self):
             self.check_memory_adoption(root_obj, np_obj)
             self.check_size(2, root_obj)
 
+
+    def test_object_dtype_strings(self):
+        """
+        Test adoption of numpy arrays with dtype=object and dtype=str containing strings
+        """
+        for dtype in [object, np.str_, '<U13']:
+            with self.subTest(dtype=dtype):
+                np_obj = np.array(["test_string_1", "test_string_2"], dtype=dtype)
+                root_obj = ROOT.VecOps.AsRVec(np_obj)
+
+                self.check_size(2, root_obj)
+                self.assertEqual(root_obj[0], np_obj[0])
+                self.assertEqual(root_obj[1], np_obj[1])
+
+
+    def test_object_dtype_mixed_types(self):
+        """
+        Test that a TypeError is raised for numpy arrays with dtype=object 
+        containing elements of different types
+        """
+        np_obj = np.array(["string", {}], dtype=object)
+
+        with self.assertRaises(TypeError) as context:
+            ROOT.VecOps.AsRVec(np_obj)
+
+        self.assertIn("All elements in the numpy array of objects must be of the same type", str(context.exception))
+
+
     def test_multidim(self):
         """
         Test adoption of multi-dimensional numpy arrays