bug fixes

eva/udfs/ndarray/open.py

@@ -16,7 +16,6 @@
 import numpy as np
 import pandas as pd
 
-from eva.executor.executor_utils import ExecutorError
 from eva.udfs.abstract.abstract_udf import AbstractUDF
 
 
@@ -42,11 +41,8 @@ def _open(row: pd.Series) -> np.ndarray:
             if path_str in self._data_cache:
                 data = self._data_cache[path_str]
             else:
-                try:
-                    data = cv2.imread(path_str)
-                except Exception as e:
-                    raise ExecutorError(str(e))
-
+                data = cv2.imread(path_str)
+                assert data is not None, f"Failed to open file {path_str}"
             self._data_cache[path_str] = data
 
             return data

test/integration_tests/test_load_executor.py

@@ -412,7 +412,7 @@ def test_should_load_csv_in_table(self):
         actual_batch.sort()
 
         # assert the batches are equal
-        expected_batch = create_dummy_csv_batches()
+        expected_batch = next(create_dummy_csv_batches())
         expected_batch.modify_column_alias("myvideocsv")
         self.assertEqual(actual_batch, expected_batch)
 
@@ -448,7 +448,7 @@ def test_should_load_csv_with_columns_in_table(self):
 
         # assert the batches are equal
         select_columns = ["id", "frame_id", "video_id", "dataset_name"]
-        expected_batch = create_dummy_csv_batches(target_columns=select_columns)
+        expected_batch = next(create_dummy_csv_batches(target_columns=select_columns))
         expected_batch.modify_column_alias("myvideocsv")
         self.assertEqual(actual_batch, expected_batch)
 

test/udfs/ndarray/test_open.py

@@ -60,5 +60,5 @@ def test_open_same_path_should_use_cache(self):
             mock_cv2.imread.assert_not_called()
 
     def test_open_path_should_raise_error(self):
-        with self.assertRaises(ExecutorError):
+        with self.assertRaises(AssertionError):
             self.open_instance(pd.DataFrame(["incorrect_path"]))