From 9375e240cc08cfcf3c5e74a1cbd5c87509e61bd7 Mon Sep 17 00:00:00 2001
From: can <can@anyscale.com>
Date: Tue, 19 Nov 2024 20:54:43 +0000
Subject: [PATCH] [data][dashboard] fix confusing dataset operator name

Signed-off-by: can <can@anyscale.com>
---
 .../dashboard/client/src/components/DataOverviewTable.tsx  | 2 +-
 .../client/src/pages/data/DataOverview.component.test.tsx  | 7 ++++---
 python/ray/dashboard/client/src/type/data.ts               | 1 +
 python/ray/dashboard/modules/data/tests/test_data_head.py  | 2 ++
 python/ray/data/_internal/execution/streaming_executor.py  | 1 +
 python/ray/data/tests/test_stats.py                        | 1 +
 6 files changed, 10 insertions(+), 4 deletions(-)

diff --git a/python/ray/dashboard/client/src/components/DataOverviewTable.tsx b/python/ray/dashboard/client/src/components/DataOverviewTable.tsx
index c52bdee0c28b..c3538ba9ccd4 100644
--- a/python/ray/dashboard/client/src/components/DataOverviewTable.tsx
+++ b/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
diff --git a/python/ray/dashboard/client/src/pages/data/DataOverview.component.test.tsx b/python/ray/dashboard/client/src/pages/data/DataOverview.component.test.tsx
index 12cd19792928..107cd2037cfc 100644
--- a/python/ray/dashboard/client/src/pages/data/DataOverview.component.test.tsx
+++ b/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();
diff --git a/python/ray/dashboard/client/src/type/data.ts b/python/ray/dashboard/client/src/type/data.ts
index dbbfc5889238..64d91a006568 100644
--- a/python/ray/dashboard/client/src/type/data.ts
+++ b/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 = {
diff --git a/python/ray/dashboard/modules/data/tests/test_data_head.py b/python/ray/dashboard/modules/data/tests/test_data_head.py
index 650079360a8b..4b18e6e482fd 100644
--- a/python/ray/dashboard/modules/data/tests/test_data_head.py
+++ b/python/ray/dashboard/modules/data/tests/test_data_head.py
@@ -13,6 +13,7 @@
 DATA_HEAD_URLS = {"GET": "http://localhost:8265/api/data/datasets/{job_id}"}
 
 DATA_SCHEMA = [
+    "name",
     "state",
     "progress",
     "total",
@@ -66,6 +67,7 @@ def test_get_datasets():
     op0 = operators[0]
     assert sorted(op0.keys()) == sorted(OPERATOR_SCHEMA)
     assert op0["operator"] == "Input0"
+    assert op0["name"] == "Input"
     assert op0["progress"] == 20
     assert op0["total"] == 20
     assert op0["state"] == "FINISHED"
diff --git a/python/ray/data/_internal/execution/streaming_executor.py b/python/ray/data/_internal/execution/streaming_executor.py
index ca48d7766c35..a4276e2bafe6 100644
--- a/python/ray/data/_internal/execution/streaming_executor.py
+++ b/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,
diff --git a/python/ray/data/tests/test_stats.py b/python/ray/data/tests/test_stats.py
index cb0d31f22774..a41e060fb3f5 100644
--- a/python/ray/data/tests/test_stats.py
+++ b/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"