[Bugfix][Bench] Clean up synthetic video files and writers (#59529)

Signed-off-by: Prudhvivuda <prudhvi12042001@gmail.com>
Co-authored-by: OpenAI Codex <noreply@openai.com>

tests/benchmarks/test_random_multimodal_dataset_video.py

@@ -2,8 +2,11 @@
 # SPDX-FileCopyrightText: Copyright contributors to the vLLM project
 
 import os
+import tempfile
+from pathlib import Path
 from tempfile import NamedTemporaryFile
 from typing import Any, cast
+from unittest.mock import Mock
 
 import cv2
 import pybase64 as base64
@@ -61,11 +64,17 @@ def test_map_config_to_modality(video_dataset: RandomMultiModalDataset):
 
 
 @pytest.mark.benchmark
-def test_generate_mm_item_video(video_dataset: RandomMultiModalDataset):
-    """Test generating multimodal items for video configurations."""
+def test_generate_mm_item_video(
+    video_dataset: RandomMultiModalDataset,
+    tmp_path: Path,
+    monkeypatch: pytest.MonkeyPatch,
+):
+    """Generate a decodable video without leaving temporary files behind."""
+    monkeypatch.setattr(tempfile, "tempdir", str(tmp_path))
     # Test video item generation
     video_config = (64, 48, 8)  # height, width, num_frames
     result = video_dataset.generate_mm_item(video_config)
+    assert not list(tmp_path.iterdir())
 
     # Check the result structure matches OpenAI API format
     assert isinstance(result, dict)
@@ -105,6 +114,33 @@ def test_generate_mm_item_video(video_dataset: RandomMultiModalDataset):
             os.unlink(temp_path)
 
 
+@pytest.mark.benchmark
+@pytest.mark.parametrize("writer_opened", [False, True], ids=["open", "write"])
+def test_generate_synthetic_video_cleans_up_on_error(
+    video_dataset: RandomMultiModalDataset,
+    tmp_path: Path,
+    monkeypatch: pytest.MonkeyPatch,
+    writer_opened: bool,
+):
+    """Release the writer and remove temporary storage when encoding fails."""
+    monkeypatch.setattr(tempfile, "tempdir", str(tmp_path))
+    writer = Mock(spec=cv2.VideoWriter)
+    writer.isOpened.return_value = writer_opened
+    writer.write.side_effect = RuntimeError("Failed to write video frame")
+    monkeypatch.setattr(cv2, "VideoWriter", lambda *args, **kwargs: writer)
+
+    error = (
+        "Failed to write video frame"
+        if writer_opened
+        else "Failed to create video writer"
+    )
+    with pytest.raises(RuntimeError, match=error):
+        video_dataset.generate_synthetic_video(48, 64, 8)
+
+    assert not list(tmp_path.iterdir())
+    writer.release.assert_called_once_with()
+
+
 @pytest.mark.benchmark
 def test_generate_mm_item_image(video_dataset: RandomMultiModalDataset):
     """Test generating multimodal items for image configurations."""

vllm/benchmarks/datasets/datasets.py

@@ -25,7 +25,7 @@ from dataclasses import dataclass, replace
 from functools import cache
 from io import BytesIO
 from pathlib import Path
-from tempfile import NamedTemporaryFile
+from tempfile import TemporaryDirectory
 from typing import Any, cast
 
 import numpy as np
@@ -975,32 +975,26 @@ class RandomMultiModalDataset(RandomDataset):
             dtype=np.uint8,
         )
 
-        # Create a temporary video file in memory
         fourcc = cv2.VideoWriter_fourcc(*"mp4v")
         fps = 30  # frames per second
 
-        with NamedTemporaryFile(suffix=".mp4", delete=False) as temp_file:
-            temp_path = temp_file.name
+        with TemporaryDirectory() as temp_dir:
+            temp_path = Path(temp_dir) / "video.mp4"
 
-            # Create video writer
             video_writer = cv2.VideoWriter(
-                temp_path, fourcc=fourcc, fps=fps, frameSize=(width, height)
+                str(temp_path), fourcc=fourcc, fps=fps, frameSize=(width, height)
             )
 
-            if not video_writer.isOpened():
-                raise RuntimeError("Failed to create video writer")
+            try:
+                if not video_writer.isOpened():
+                    raise RuntimeError("Failed to create video writer")
 
-            for frame in random_pixels:
-                video_writer.write(frame)
+                for frame in random_pixels:
+                    video_writer.write(frame)
+            finally:
+                video_writer.release()
 
-            video_writer.release()
-            temp_file.close()
-
-            # Read the video file content
-            with open(temp_path, "rb") as f:
-                video_content = f.read()
-
-            return {"bytes": video_content}
+            return {"bytes": temp_path.read_bytes()}
 
     def map_config_to_modality(self, config: tuple[int, int, int]) -> str:
         """Map the configuration to the modality."""