v0.16.9: Unified job polling and test coverage improvements

.github/workflows/ci.yml

@@ -115,10 +115,16 @@ jobs:
           cd web-ui
           npm run type-check
 
-      - name: Run tests
+      - name: Run tests with coverage
         run: |
           cd web-ui
-          npm test -- --run
+          npm test -- --run --coverage
+
+      - name: Upload coverage report
+        uses: actions/upload-artifact@v4
+        with:
+          name: web-ui-coverage
+          path: web-ui/coverage
 
   # ---------------------------------------------------------
   # E2E tests (depends on server tests + web-ui)

server/app/api_routes/routes/jobs.py

@@ -22,7 +22,6 @@
 from app.models.job import Job, JobStatus
 from app.schemas.job import JobListItem, JobListResponse, JobResultsResponse
 from app.services.storage.factory import get_storage_service
-from app.services.video_summary_service import derive_video_summary
 from app.settings import settings
 
 logger = logging.getLogger(__name__)
@@ -231,21 +230,18 @@ async def get_job(job_id: UUID, db: Session = Depends(get_db)) -> JobResultsResp
 
     # Clean Break: All completed jobs return result_url + summary
     # No more inline results for any job type
+    # v0.16.8: Summary comes from job.summary (pre-computed by worker)
 
-    # Load results from storage to derive summary
+    # Verify results file exists
     try:
-        results_path = job.output_path
-        file_path = storage.load_file(results_path)
-        with open(file_path, "r") as f:
-            results = json.load(f)
+        storage.load_file(job.output_path)
     except FileNotFoundError as err:
         raise HTTPException(status_code=404, detail="Results file not found") from err
-    except json.JSONDecodeError as err:
-        raise HTTPException(status_code=500, detail="Invalid results file") from err
 
     # Return result_url and summary for all completed jobs
     result_url = storage.get_signed_url(job.output_path)
-    summary = derive_video_summary(results)
+    # v0.16.8: Use pre-computed summary from job.summary (set by worker)
+    summary = json.loads(job.summary) if job.summary else None
     return JobResultsResponse(
         job_id=job.job_id,
         status=job.status.value,

server/app/workers/worker.py

@@ -676,11 +676,15 @@ def _finalize_job(self, job_id: str, meta: dict, results: dict) -> None:
             if job.job_type in ("video", "video_multi"):
                 job.progress = 100
             # Discussion #354: Pre-compute summary for /v1/jobs hot path
-            summary_dict = derive_video_summary(output_data)
+            # v0.16.8: Use plugin-provided summary if available (decoupled from server)
+            summary_dict = output_data.get("summary")
+            if not summary_dict:
+                # Fallback for plugins that don't provide summary
+                summary_dict = derive_video_summary(output_data)
             job.summary = json.dumps(summary_dict)
             db.commit()
             send_job_completed(str(job.job_id))
-            logger.info(f"Job {job.job_id} completed successfully via Ray")
+            logger.info(f"Job {job_id} completed successfully via Ray")
         finally:
             db.close()
 
@@ -987,7 +991,11 @@ def cb(current_frame: int, total: int = total_frames) -> None:
             if job.job_type in ("video", "video_multi"):
                 job.progress = 100
             # Discussion #354: Pre-compute summary for /v1/jobs hot path
-            summary_dict = derive_video_summary(output_data)
+            # v0.16.8: Use plugin-provided summary if available (decoupled from server)
+            summary_dict = output_data.get("summary")
+            if not summary_dict:
+                # Fallback for plugins that don't provide summary
+                summary_dict = derive_video_summary(output_data)
             job.summary = json.dumps(summary_dict)
             db.commit()
 

server/tests/api/routes/test_jobs_unified.py

@@ -116,8 +116,9 @@ def test_get_job_completed(client, session, storage):
     """Test GET /v1/jobs/{id} for completed job with results.
 
     Clean Break: Completed jobs return result_url, not inline results.
+    v0.16.8: Summary is pre-computed by worker and stored in job.summary.
     """
-    # Create a completed job
+    # Create a completed job with pre-computed summary (set by worker)
     job_id = uuid4()
     output_path = f"image/output/{job_id}.json"
     job = Job(
@@ -127,6 +128,8 @@ def test_get_job_completed(client, session, storage):
         input_path="image/input/test.png",
         output_path=output_path,
         job_type="image",
+        # v0.16.8: Pre-computed summary from worker (plugin provides this)
+        summary=json.dumps({"text_length": 14, "word_count": 2}),
     )
     session.add(job)
     session.commit()
@@ -180,8 +183,9 @@ def test_get_job_image_type(client, session, storage):
     """Test GET /v1/jobs/{id} for image job.
 
     Clean Break: Image jobs also use result_url for consistency.
+    v0.16.8: Summary is pre-computed by worker and stored in job.summary.
     """
-    # Create a completed image job
+    # Create a completed image job with pre-computed summary
     job_id = uuid4()
     output_path = f"image/output/{job_id}.json"
     job = Job(
@@ -191,6 +195,8 @@ def test_get_job_image_type(client, session, storage):
         input_path="image/input/test.png",
         output_path=output_path,
         job_type="image",
+        # v0.16.8: Pre-computed summary from worker
+        summary=json.dumps({"text_length": 10, "word_count": 2}),
     )
     session.add(job)
     session.commit()
@@ -215,8 +221,9 @@ def test_get_job_video_type(client, session, storage):
     """Test GET /v1/jobs/{id} for video job.
 
     Issue #350: Video jobs return result_url instead of inline results.
+    v0.16.8: Summary is pre-computed by worker and stored in job.summary.
     """
-    # Create a completed video job
+    # Create a completed video job with pre-computed summary
     job_id = uuid4()
     output_path = f"video/output/{job_id}.json"
     job = Job(
@@ -226,6 +233,8 @@ def test_get_job_video_type(client, session, storage):
         input_path="video/input/test.mp4",
         output_path=output_path,
         job_type="video",
+        # v0.16.8: Pre-computed summary from worker
+        summary=json.dumps({"frame_count": 1, "detection_count": 0, "classes": []}),
     )
     session.add(job)
     session.commit()
@@ -268,27 +277,37 @@ def test_get_job_results_file_not_found(client, session):
 
 
 def test_get_job_results_invalid_json(client, session, storage):
-    """Test GET /v1/jobs/{id} when results file contains invalid JSON."""
-    # Create a completed job with invalid JSON results
+    """Test GET /v1/jobs/{id} when results file contains invalid JSON.
+
+    v0.16.8: API no longer parses the JSON file for summary - it comes from
+    job.summary (pre-computed by worker). API just checks file existence.
+    Invalid JSON would have been caught by the worker when saving results.
+    """
+    # Create a completed job with pre-computed summary
     job_id = uuid4()
+    output_path = f"image/output/{job_id}.json"
     job = Job(
         job_id=job_id,
         status=JobStatus.completed,
         plugin_id="ocr",
         input_path="image/input/test.png",
-        output_path="image/output/invalid.json",
+        output_path=output_path,
         job_type="image",
+        summary=json.dumps({"text_length": 0, "word_count": 0}),
     )
     session.add(job)
     session.commit()
 
-    # Create invalid JSON file
-    storage.save_file(BytesIO(b"invalid json"), "image/output/invalid.json")
+    # Create a file (even with invalid JSON - API doesn't parse it)
+    storage.save_file(BytesIO(b"invalid json"), output_path)
 
     response = client.get(f"/v1/jobs/{job_id}")
 
-    assert response.status_code == 500
-    assert "invalid results file" in response.json()["detail"].lower()
+    # v0.16.8: API returns 200 because file exists, summary comes from DB
+    assert response.status_code == 200
+    data = response.json()
+    assert data["result_url"] is not None
+    assert data["summary"] is not None
 
 
 # Issue #350: Artifact Pattern - video jobs return result_url, not results
@@ -298,6 +317,7 @@ def test_get_job_video_returns_result_url(client, session, storage):
     """Test GET /v1/jobs/{id} returns result_url for video jobs.
 
     Issue #350: Video jobs should return a URL for lazy loading.
+    v0.16.8: Summary is pre-computed by worker and stored in job.summary.
     """
     job_id = uuid4()
     output_path = f"video/output/{job_id}.json"
@@ -308,6 +328,10 @@ def test_get_job_video_returns_result_url(client, session, storage):
         input_path="video/input/test.mp4",
         output_path=output_path,
         job_type="video",
+        # v0.16.8: Pre-computed summary from worker
+        summary=json.dumps(
+            {"frame_count": 100, "detection_count": 200, "classes": ["player", "ball"]}
+        ),
     )
     session.add(job)
     session.commit()
@@ -338,6 +362,7 @@ def test_get_job_video_includes_summary(client, session, storage):
     """Test GET /v1/jobs/{id} includes summary for video jobs.
 
     Issue #350: Summary contains derived metadata.
+    v0.16.8: Summary is pre-computed by worker and stored in job.summary.
     """
     job_id = uuid4()
     output_path = f"video/output/{job_id}.json"
@@ -348,6 +373,10 @@ def test_get_job_video_includes_summary(client, session, storage):
         input_path="video/input/test.mp4",
         output_path=output_path,
         job_type="video",
+        # v0.16.8: Pre-computed summary from worker (matches test assertions)
+        summary=json.dumps(
+            {"frame_count": 50, "detection_count": 100, "classes": ["player", "ball"]}
+        ),
     )
     session.add(job)
     session.commit()

server/tests/app/workers/test_worker.py

@@ -629,13 +629,19 @@ def test_worker_stores_summary_on_completion(test_engine, session):
     mock_plugin_service.get_plugin_manifest.return_value = {
         "tools": [{"id": "detect", "input_types": ["video"]}]
     }
-    # Return video results with frames and detections
+    # Return video results with frames, detections, and plugin-computed summary
+    # Plugin (yolo-tracker) computes summary via compute_video_summary()
     mock_plugin_service.run_plugin_tool.return_value = {
         "total_frames": 10,
         "frames": [
             {"frame_idx": i, "detections": [{"class": "person"}, {"class": "car"}]}
             for i in range(10)
         ],
+        "summary": {
+            "frame_count": 10,
+            "detection_count": 20,
+            "classes": ["person", "car"],
+        },
     }
 
     # Run worker
@@ -651,11 +657,12 @@ def test_worker_stores_summary_on_completion(test_engine, session):
     # Discussion #354: Summary should be stored
     assert updated_job.summary is not None, "Worker should compute and store summary"
 
-    # Verify summary content
+    # Verify summary content (from plugin-provided summary)
     summary = json.loads(updated_job.summary)
     assert "frame_count" in summary
     assert "detection_count" in summary
     assert "classes" in summary
-    # 10 frames * 2 detections each = 20 total
+    # Verify plugin's summary is used (not recomputed)
+    assert summary["frame_count"] == 10
     assert summary["detection_count"] == 20
-    assert set(summary["classes"]) == {"person", "car"}
+    assert summary["classes"] == ["person", "car"]