Merge branch 'main' into fix/rerun_flaky_tests

Author: markurtz

tests/unit/data/deserializers/test_file.py

@@ -284,7 +284,7 @@ def test_hdf5_file_deserializer_success(tmp_path):
 ###################
 
 
-@pytest.mark.skip(reason="issue: #492")
+@pytest.mark.sanity
 def test_db_file_deserializer_success(monkeypatch, tmp_path):
     import sqlite3
 
@@ -317,6 +317,7 @@ def mock_from_sql(sql, con, **kwargs):
         data=db_path,
         processor_factory=processor_factory(),
         random_seed=1,
+        sql="SELECT * FROM samples",
     )
 
     # Assert: result is of type Dataset

tests/unit/data/deserializers/test_memory.py

@@ -0,0 +1,233 @@
+import pytest
+from datasets import Dataset
+
+from guidellm.data.deserializers.deserializer import (
+    DataNotSupportedError,
+)
+from guidellm.data.deserializers.memory import (
+    InMemoryCsvDatasetDeserializer,
+    InMemoryDictDatasetDeserializer,
+    InMemoryDictListDatasetDeserializer,
+    InMemoryItemListDatasetDeserializer,
+    InMemoryJsonStrDatasetDeserializer,
+)
+
+
+@pytest.fixture
+def processor_factory():
+    return None  # Dummy processor factory for testing
+
+
+###################
+# Tests dict in memory deserializer
+###################
+
+
+@pytest.mark.smoke
+def test_in_memory_dict_deserializer_success(processor_factory):
+    deserializer = InMemoryDictDatasetDeserializer()
+
+    data = {
+        "text": ["hello", "world"],
+        "id": [1, 2],
+    }
+
+    dataset = deserializer(
+        data=data,
+        processor_factory=processor_factory,
+        random_seed=42,
+    )
+
+    assert isinstance(dataset, Dataset)
+    assert dataset["text"] == ["hello", "world"]
+    assert dataset["id"] == [1, 2]
+    assert len(dataset) == 2
+
+
+@pytest.mark.smoke
+def test_in_memory_dict_deserializer_invalid_not_dict(processor_factory):
+    deserializer = InMemoryDictDatasetDeserializer()
+
+    with pytest.raises(DataNotSupportedError):
+        deserializer(
+            data="not a dict",
+            processor_factory=processor_factory,
+            random_seed=42,
+        )
+
+
+@pytest.mark.smoke
+def test_in_memory_dict_deserializer_empty_dict(processor_factory):
+    deserializer = InMemoryDictDatasetDeserializer()
+
+    with pytest.raises(DataNotSupportedError):
+        deserializer(
+            data={},
+            processor_factory=processor_factory,
+            random_seed=42,
+        )
+
+
+@pytest.mark.smoke
+def test_in_memory_dict_deserializer_value_not_list(processor_factory):
+    deserializer = InMemoryDictDatasetDeserializer()
+
+    with pytest.raises(DataNotSupportedError):
+        deserializer(
+            data={"text": "hello"},  # value 不是 list
+            processor_factory=processor_factory,
+            random_seed=42,
+        )
+
+
+@pytest.mark.smoke
+def test_in_memory_dict_deserializer_list_length_mismatch(processor_factory):
+    deserializer = InMemoryDictDatasetDeserializer()
+
+    with pytest.raises(DataNotSupportedError):
+        deserializer(
+            data={
+                "text": ["hello", "world"],
+                "id": [1],  # diferent length
+            },
+            processor_factory=processor_factory,
+            random_seed=42,
+        )
+
+
+###################
+# Tests dict list in memory deserializer
+###################
+
+
+@pytest.mark.smoke
+def test_in_memory_dict_list_deserializer_success(processor_factory):
+    # Arrange
+    data = [
+        {"id": 1, "text": "hello"},
+        {"id": 2, "text": "world"},
+        {"id": 3, "text": "guidellm"},
+    ]
+
+    deserializer = InMemoryDictListDatasetDeserializer()
+
+    # Act
+    dataset = deserializer(
+        data=data,
+        processor_factory=processor_factory,
+        random_seed=42,
+    )
+
+    # Assert
+    assert isinstance(dataset, Dataset)
+    assert dataset["id"] == [1, 2, 3]
+    assert dataset["text"] == ["hello", "world", "guidellm"]
+    assert len(dataset) == 3
+
+
+@pytest.mark.smoke
+def test_in_memory_dict_list_deserializer_key_mismatch(processor_factory):
+    deserializer = InMemoryDictListDatasetDeserializer()
+
+    wrong_data = [
+        {"id": 1, "text": "hello"},
+        {"id": 2, "msg": "world"},  # key mismatch
+    ]
+
+    with pytest.raises(DataNotSupportedError):
+        deserializer(
+            data=wrong_data,
+            processor_factory=processor_factory,
+            random_seed=42,
+        )
+
+
+###################
+# Tests list in memory deserializer
+###################
+
+
+@pytest.mark.smoke
+def test_in_memory_item_list_deserializer_key_mismatch(processor_factory):
+    data = ["a", "b", "c"]
+
+    deserializer = InMemoryItemListDatasetDeserializer()
+
+    # Act
+    dataset = deserializer(
+        data=data,
+        processor_factory=processor_factory,
+        random_seed=42,
+    )
+
+    # Assert
+    assert isinstance(dataset, Dataset)
+    assert dataset["data"] == data
+    assert len(dataset) == 3
+
+
+@pytest.mark.smoke
+def test_in_memory_item_list_custom_column_name(processor_factory):
+    deserializer = InMemoryItemListDatasetDeserializer()
+    data = [1, 2, 3]
+
+    dataset = deserializer(
+        data=data,
+        processor_factory=processor_factory,
+        random_seed=123,
+        column_name="numbers",
+    )
+
+    assert list(dataset.column_names) == ["numbers"]
+    assert dataset["numbers"] == [1, 2, 3]
+
+
+###################
+# Tests json in memory deserializer
+###################
+
+
+@pytest.mark.parametrize(
+    ("json_input"),
+    [
+        '{"text": ["hello", "world"], "id": [1, 2]}',
+        '[{"id": 1, "text": "hello"}, {"id": 2, "text": "world"}]',
+        '["a", "b", "c"]',
+    ],
+)
+@pytest.mark.smoke
+def test_in_memory_json_deserializer_success(processor_factory, json_input):
+    deserializer = InMemoryJsonStrDatasetDeserializer()
+
+    dataset = deserializer(
+        data=json_input,
+        processor_factory=processor_factory,
+        random_seed=42,
+    )
+
+    assert isinstance(dataset, Dataset)
+    assert len(dataset) > 0
+
+
+###################
+# Tests csv in memory deserializer
+###################
+
+
+@pytest.mark.smoke
+def test_csv_file_deserializer_success(processor_factory):
+    csv_str = "id,text\n1,hello\n2,world\n"
+
+    deserializer = InMemoryCsvDatasetDeserializer()
+
+    dataset = deserializer(
+        data=csv_str,
+        processor_factory=processor_factory,
+        random_seed=43,
+    )
+
+    assert isinstance(dataset, Dataset)
+    assert {"id", "text"}.issubset(set(dataset.column_names))
+    assert dataset["id"] == ["1", "2"]
+    assert dataset["text"] == ["hello", "world"]
+    assert len(dataset) == 2