Add a test

Author: dmontagu

tests/evals/test_dataset.py

@@ -7,6 +7,7 @@
 from typing import Any
 
 import pytest
+import yaml
 from dirty_equals import HasRepr, IsNumber
 from inline_snapshot import snapshot
 from pydantic import BaseModel, TypeAdapter
@@ -819,10 +820,29 @@ async def test_serialization_to_yaml(example_dataset: Dataset[TaskInput, TaskOut
     # Test loading back
     loaded_dataset = Dataset[TaskInput, TaskOutput, TaskMetadata].from_file(yaml_path)
     assert len(loaded_dataset.cases) == 2
+    assert loaded_dataset.name == 'example'
     assert loaded_dataset.cases[0].name == 'case1'
     assert loaded_dataset.cases[0].inputs.query == 'What is 2+2?'
 
 
+async def test_deserializing_without_name(
+    example_dataset: Dataset[TaskInput, TaskOutput, TaskMetadata], tmp_path: Path
+):
+    """Test serializing a dataset to YAML."""
+    # Save the dataset
+    yaml_path = tmp_path / 'test_cases.yaml'
+    example_dataset.to_file(yaml_path)
+
+    # Rewrite the file _without_ a name to test deserializing a name-less file
+    obj = yaml.safe_load(yaml_path.read_text())
+    obj.pop('name', None)
+    yaml_path.write_text(yaml.dump(obj))
+
+    # Test loading results in the name coming from the filename stem
+    loaded_dataset = Dataset[TaskInput, TaskOutput, TaskMetadata].from_file(yaml_path)
+    assert loaded_dataset.name == 'test_cases'
+
+
 async def test_serialization_to_json(example_dataset: Dataset[TaskInput, TaskOutput, TaskMetadata], tmp_path: Path):
     """Test serializing a dataset to JSON."""
     json_path = tmp_path / 'test_cases.json'
@@ -854,6 +874,7 @@ def test_serialization_errors(tmp_path: Path):
 async def test_from_text():
     """Test creating a dataset from text."""
     dataset_dict = {
+        'name': 'my dataset',
         'cases': [
             {
                 'name': '1',
@@ -873,6 +894,7 @@ async def test_from_text():
     }
 
     loaded_dataset = Dataset[TaskInput, TaskOutput, TaskMetadata].from_text(json.dumps(dataset_dict))
+    assert loaded_dataset.name == 'my dataset'
     assert loaded_dataset.cases == snapshot(
         [
             Case(