chunker and ui

tests/chunker/test_pipeline.py

@@ -0,0 +1,144 @@
+"""
+Tests for Pipeline — end-to-end orchestration, stats, error handling.
+
+Demonstrates: TDD (Interview Topic 8) — integration testing with mocked FFmpeg probe.
+"""
+
+from unittest.mock import MagicMock, patch
+
+import pytest
+
+from core.chunker import Pipeline
+from core.chunker.exceptions import PipelineError
+
+
+def mock_probe(duration):
+    """Create a mock ProbeResult with the given duration."""
+    result = MagicMock()
+    result.duration = duration
+    return result
+
+
+class TestPipeline:
+    @patch("core.chunker.chunker.probe_file")
+    def test_end_to_end(self, mock_pf, temp_file):
+        """Full pipeline processes a file successfully."""
+        path = temp_file(b"x" * 4096)
+        mock_pf.return_value = mock_probe(40.0)
+
+        result = Pipeline(
+            source=path,
+            chunk_duration=10.0,
+            num_workers=2,
+            processor_type="checksum",
+        ).run()
+
+        assert result.total_chunks == 4
+        assert result.processed == 4
+        assert result.failed == 0
+        assert result.elapsed_time > 0
+        assert result.chunks_in_order is True
+
+    @patch("core.chunker.chunker.probe_file")
+    def test_throughput_calculated(self, mock_pf, temp_file):
+        """Pipeline calculates throughput."""
+        path = temp_file(b"x" * 10000)
+        mock_pf.return_value = mock_probe(30.0)
+
+        result = Pipeline(source=path, chunk_duration=10.0, num_workers=2).run()
+
+        assert result.throughput_mbps > 0
+
+    @patch("core.chunker.chunker.probe_file")
+    def test_worker_stats(self, mock_pf, temp_file):
+        """Pipeline reports per-worker stats."""
+        path = temp_file(b"x" * 4000)
+        mock_pf.return_value = mock_probe(40.0)
+
+        result = Pipeline(
+            source=path, chunk_duration=10.0, num_workers=2
+        ).run()
+
+        assert len(result.worker_stats) == 2
+        for worker_id, stats in result.worker_stats.items():
+            assert "processed" in stats
+            assert "errors" in stats
+
+    def test_nonexistent_file(self):
+        """Non-existent file raises PipelineError."""
+        with pytest.raises(PipelineError):
+            Pipeline(source="/nonexistent/file.mp4").run()
+
+    @patch("core.chunker.chunker.probe_file")
+    def test_event_callback(self, mock_pf, temp_file):
+        """Pipeline emits events through callback."""
+        path = temp_file(b"x" * 2048)
+        mock_pf.return_value = mock_probe(20.0)
+        events = []
+
+        def capture(event_type, data):
+            events.append(event_type)
+
+        Pipeline(
+            source=path,
+            chunk_duration=10.0,
+            num_workers=1,
+            event_callback=capture,
+        ).run()
+
+        assert "pipeline_start" in events
+        assert "pipeline_complete" in events
+        assert "chunk_queued" in events
+
+    @patch("core.chunker.chunker.probe_file")
+    def test_simulated_decode_processor(self, mock_pf, temp_file):
+        """Pipeline works with simulated_decode processor."""
+        path = temp_file(b"x" * 2048)
+        mock_pf.return_value = mock_probe(20.0)
+
+        result = Pipeline(
+            source=path,
+            chunk_duration=10.0,
+            num_workers=2,
+            processor_type="simulated_decode",
+        ).run()
+
+        assert result.total_chunks == 2
+        assert result.failed == 0
+
+    @patch("core.chunker.chunker.probe_file")
+    def test_single_chunk_file(self, mock_pf, temp_file):
+        """Duration shorter than chunk_duration produces one chunk."""
+        path = temp_file(b"x" * 100)
+        mock_pf.return_value = mock_probe(5.0)
+
+        result = Pipeline(source=path, chunk_duration=10.0).run()
+
+        assert result.total_chunks == 1
+        assert result.processed == 1
+
+    @patch("core.chunker.chunker.probe_file")
+    def test_retries_tracked(self, mock_pf, temp_file):
+        """Pipeline result tracks total retries."""
+        path = temp_file(b"x" * 2048)
+        mock_pf.return_value = mock_probe(20.0)
+
+        result = Pipeline(source=path, chunk_duration=10.0).run()
+
+        assert result.retries >= 0  # Might be 0 if no failures
+
+    @patch("core.chunker.chunker.probe_file")
+    def test_output_dir_and_chunk_files(self, mock_pf, temp_file):
+        """Pipeline tracks output_dir and chunk_files when set."""
+        path = temp_file(b"x" * 1024)
+        mock_pf.return_value = mock_probe(10.0)
+
+        result = Pipeline(
+            source=path,
+            chunk_duration=10.0,
+            processor_type="checksum",
+        ).run()
+
+        # No output_dir set, so chunk_files should be empty
+        assert result.output_dir is None
+        assert result.chunk_files == []