[data][dashboard] fix confusing dataset operator name

Signed-off-by: can <[email protected]>

python/ray/dashboard/client/src/components/DataOverviewTable.tsx

@@ -193,7 +193,7 @@ const DataRow = ({
       </TableCell>
       <TableCell align="left">
         {isDatasetRow && datasetMetrics.dataset}
-        {isOperatorRow && operatorMetrics.operator}
+        {isOperatorRow && operatorMetrics.name}
       </TableCell>
       <TableCell align="right" style={{ width: 200 }}>
         <TaskProgressBar

python/ray/dashboard/client/src/pages/data/DataOverview.component.test.tsx

@@ -36,6 +36,7 @@ describe("DataOverview", () => {
         operators: [
           {
             operator: "test_ds1_op1",
+            name: "test_ds1_op",
             state: "RUNNING",
             progress: 99,
             total: 101,
@@ -104,11 +105,11 @@ describe("DataOverview", () => {
     expect(screen.getByText("70/80")).toBeVisible();
 
     // Operator dropdown
-    expect(screen.queryByText("test_ds1_op1")).toBeNull();
+    expect(screen.queryByText("test_ds1_op")).toBeNull();
     await user.click(screen.getByTitle("Expand Dataset test_ds1"));
-    expect(screen.getByText("test_ds1_op1")).toBeVisible();
+    expect(screen.getByText("test_ds1_op")).toBeVisible();
     await user.click(screen.getByTitle("Collapse Dataset test_ds1"));
-    expect(screen.queryByText("test_ds1_op1")).toBeNull();
+    expect(screen.queryByText("test_ds1_op")).toBeNull();
 
     // Second Dataset
     expect(screen.getByText("test_ds2")).toBeVisible();

python/ray/dashboard/client/src/type/data.ts

@@ -12,6 +12,7 @@ export type DatasetMetrics = DataMetrics & {
 
 export type OperatorMetrics = DataMetrics & {
   operator: string;
+  name: string;
 };
 
 export type DataMetrics = {

python/ray/dashboard/modules/data/tests/test_data_head.py

@@ -32,6 +32,7 @@
 ] + DATA_SCHEMA
 
 OPERATOR_SCHEMA = [
+    "name",
     "operator",
 ] + DATA_SCHEMA
 
@@ -64,12 +65,23 @@ def test_get_datasets():
     operators = dataset["operators"]
     assert len(operators) == 2
     op0 = operators[0]
+    op1 = operators[1]
     assert sorted(op0.keys()) == sorted(OPERATOR_SCHEMA)
-    assert op0["operator"] == "Input0"
-    assert op0["progress"] == 20
-    assert op0["total"] == 20
-    assert op0["state"] == "FINISHED"
-    assert operators[1]["operator"] == "ReadRange->MapBatches(<lambda>)1"
+    assert sorted(op1.keys()) == sorted(OPERATOR_SCHEMA)
+    assert {
+        "operator": "Input0",
+        "name": "Input",
+        "state": "FINISHED",
+        "progress": 20,
+        "total": 20,
+    }.items() <= op0.items()
+    assert {
+        "operator": "ReadRange->MapBatches(<lambda>)1",
+        "name": "ReadRange->MapBatches(<lambda>)",
+        "state": "FINISHED",
+        "progress": 20,
+        "total": 20,
+    }.items() <= op1.items()
 
     ds.map_batches(lambda x: x).materialize()
     data = requests.get(DATA_HEAD_URLS["GET"].format(job_id=job_id)).json()
@@ -83,4 +95,4 @@ def test_get_datasets():
 
 
 if __name__ == "__main__":
-    sys.exit(pytest.main(["-v", __file__]))
+    sys.exit(pytest.main(["-vv", __file__]))

python/ray/data/_internal/execution/streaming_executor.py

@@ -399,6 +399,7 @@ def _get_state_dict(self, state):
             "end_time": time.time() if state != "RUNNING" else None,
             "operators": {
                 f"{op.name}{i}": {
+                    "name": op.name,
                     "progress": op_state.num_completed_tasks,
                     "total": op.num_outputs_total(),
                     "state": state,

python/ray/data/tests/test_stats.py

@@ -1648,6 +1648,7 @@ def test_stats_actor_datasets(ray_start_cluster):
     assert "Input0" in operators
     assert "ReadRange->MapBatches(<lambda>)1" in operators
     for value in operators.values():
+        assert value["name"] in ["Input", "ReadRange->MapBatches(<lambda>)"]
         assert value["progress"] == 20
         assert value["total"] == 20
         assert value["state"] == "FINISHED"