Add __repr__ methods for DatasetColumn and DatasetColumnPair

src/evaluate/evaluator/utils.py

@@ -8,6 +8,9 @@ def __init__(self, dataset: Dataset, key: str):
         self.dataset = dataset
         self.key = key
 
+    def __repr__(self):
+        return f"DatasetColumn(key={self.key!r}, len={len(self)})"
+
     def __len__(self):
         return len(self.dataset)
 
@@ -65,6 +68,16 @@ def __init__(
         self.first_key = first_key
         self.second_key = second_key
 
+    def __repr__(self):
+        return (
+            f"DatasetColumnPair("
+            f"first_col={self.first_col!r}, "
+            f"second_col={self.second_col!r}, "
+            f"first_key={self.first_key!r}, "
+            f"second_key={self.second_key!r}, "
+            f"len={len(self)})"
+        )
+
     def __len__(self):
         return len(self.dataset)
 

tests/test_evaluator_utils.py

@@ -0,0 +1,20 @@
+from datasets import Dataset
+
+from evaluate.evaluator.utils import DatasetColumn, DatasetColumnPair
+
+
+def test_dataset_column_repr():
+    dataset = Dataset.from_dict({"text": ["a", "b", "c"]})
+    column = DatasetColumn(dataset, "text")
+
+    assert repr(column) == "DatasetColumn(key='text', len=3)"
+
+
+def test_dataset_column_pair_repr():
+    dataset = Dataset.from_dict({"text": ["a", "b", "c"], "label": [0, 1, 0]})
+    column_pair = DatasetColumnPair(dataset, "text", "label", "prediction", "reference")
+
+    assert repr(column_pair) == (
+        "DatasetColumnPair(first_col='text', second_col='label', "
+        "first_key='prediction', second_key='reference', len=3)"
+    )