From 456b1bbf6666ab2424a644b383ffb1f4cbf1fdf4 Mon Sep 17 00:00:00 2001
From: Zhicheng Wu <150873423+newclearclear@users.noreply.github.com>
Date: Thu, 19 Mar 2026 21:08:09 +0800
Subject: [PATCH] fix: row selection leaks across pages in dataset and file
 list tables (#13668)

### What problem does this PR solve?

When using pagination in the Dataset file list or File Manager,
selecting row N on page 1 would incorrectly cause row N on page 2 (and
subsequent pages) to also appear selected. This is a state pollution
bug.

### Root Cause

TanStack React Table defaults to using array indices (0, 1, 2...) as
`rowSelection` keys. With server-side (manual) pagination, each page's
rows start from index 0, so a selection like `{2: true}` on page 1 also
matches index 2 on every other page.

### Fix

- Added `getRowId: (row) => row.id` to `useReactTable` in both
`DatasetTable` and `FilesTable`, so selection state is keyed by unique
document/file IDs instead of positional indices.
- Updated the `useSelectedIds` helper to support ID-based selection keys
while maintaining backward compatibility with index-based keys.

### Type of change

- [x] Bug Fix (non-breaking change which fixes an issue)

### Files Changed

| File | Change |
|------|--------|
| `web/src/pages/dataset/dataset/dataset-table.tsx` | Added `getRowId`
to table config |
| `web/src/pages/files/files-table.tsx` | Added `getRowId` to table
config |
| `web/src/hooks/logic-hooks/use-row-selection.ts` | Updated
`useSelectedIds` to handle ID-based selection |
---
 .../hooks/logic-hooks/use-row-selection.ts    | 19 +++++++++++++++----
 .../pages/dataset/dataset/dataset-table.tsx   |  1 +
 web/src/pages/files/files-table.tsx           |  2 +-
 3 files changed, 17 insertions(+), 5 deletions(-)

diff --git a/web/src/hooks/logic-hooks/use-row-selection.ts b/web/src/hooks/logic-hooks/use-row-selection.ts
index e86fc9bb0c..af6cf37b5f 100644
--- a/web/src/hooks/logic-hooks/use-row-selection.ts
+++ b/web/src/hooks/logic-hooks/use-row-selection.ts
@@ -29,10 +29,21 @@ export function useSelectedIds<T extends Array<{ id: string }>>(
   list: T,
 ) {
   const selectedIds = useMemo(() => {
-    const indexes = Object.keys(rowSelection);
-    return list
-      .filter((x, idx) => indexes.some((y) => Number(y) === idx))
-      .map((x) => x.id);
+    // When using getRowId, rowSelection keys are IDs, not indices
+    const selectionKeys = Object.keys(rowSelection);
+
+    // Check if keys are numeric (index-based) or string IDs
+    const isIndexBased = selectionKeys.every((key) => !isNaN(Number(key)));
+
+    if (isIndexBased) {
+      // Legacy index-based selection
+      return list
+        .filter((x, idx) => selectionKeys.some((y) => Number(y) === idx))
+        .map((x) => x.id);
+    } else {
+      // ID-based selection (when getRowId is used)
+      return selectionKeys.filter((id) => list.some((item) => item.id === id));
+    }
   }, [list, rowSelection]);
 
   return { selectedIds };
diff --git a/web/src/pages/dataset/dataset/dataset-table.tsx b/web/src/pages/dataset/dataset/dataset-table.tsx
index 1cf29aadb8..950aa76727 100644
--- a/web/src/pages/dataset/dataset/dataset-table.tsx
+++ b/web/src/pages/dataset/dataset/dataset-table.tsx
@@ -114,6 +114,7 @@ export function DatasetTable({
     getFilteredRowModel: getFilteredRowModel(),
     onColumnVisibilityChange: setColumnVisibility,
     onRowSelectionChange: setRowSelection,
+    getRowId: (row) => row.id, // Use document ID instead of row index
     manualPagination: true, //we're doing manual "server-side" pagination
     state: {
       sorting,
diff --git a/web/src/pages/files/files-table.tsx b/web/src/pages/files/files-table.tsx
index abc072185e..4c6f9d3a3b 100644
--- a/web/src/pages/files/files-table.tsx
+++ b/web/src/pages/files/files-table.tsx
@@ -260,7 +260,7 @@ export function FilesTable({
     getFilteredRowModel: getFilteredRowModel(),
     onColumnVisibilityChange: setColumnVisibility,
     onRowSelectionChange: setRowSelection,
-
+    getRowId: (row) => row.id, // Use file ID instead of row index
     manualPagination: true, //we're doing manual "server-side" pagination
     enableRowSelection(row) {
       return !isKnowledgeBaseType(row.original.source_type);