fix(data-compare): avoid unsafe key inference

This commit is contained in:
jearhe authored and GitHub committed 2026-08-18 10:51:26 +08:00
1 parent 665e5920c9
commit fab5c75fd2
4 files changed
+109 -6

No files matched your search

@@ -14,6 +14,7 @@ import { databaseOptionsForConnection, fetchNamespaceOptionsForConnection } from
import { isSchemaAware } from "@/lib/database/databaseCapabilities";
import { copyToClipboard } from "@/lib/common/clipboard";
import type { DataCompareCellValue, DataCompareModifiedRow, DataCompareResult, DataCompareRow, DataCompareSyncPlan, DataCompareSyncPlanTableOptions } from "@/lib/dataGrid/dataCompare";
import { inferCompareKeyColumns } from "@/lib/dataGrid/dataCompare";
import type { ColumnInfo, DatabaseType } from "@/types/database";
import * as api from "@/lib/backend/api";
import { executeWithProductionSqlGuard } from "@/lib/database/productionExecutionGuard";
@@ -394,9 +395,7 @@ async function loadColumnsWithCache(cache: Map<string, CompareColumn[]>, connect
async function inferKeyColumnsForTable(table: string, sourceColumnCache?: Map<string, CompareColumn[]>): Promise<string[]> {
if (!sourceConnectionId.value || !sourceDatabase.value || !sourceSchema.value || !table) return [];
const columns = sourceColumnCache ? await loadColumnsWithCache(sourceColumnCache, sourceConnectionId.value, sourceDatabase.value, sourceSchema.value, table) : (((await api.getColumns(sourceConnectionId.value, sourceDatabase.value, sourceSchema.value, table)) as CompareColumn[]) ?? []);
const primaryKeys = columns.filter((column) => column.is_primary_key).map((column) => column.name);
if (primaryKeys.length > 0) return primaryKeys;
return columns.slice(0, 1).map((column) => column.name);
return inferCompareKeyColumns(columns);
}
async function inferKeyColumns() {
@@ -0,0 +1,36 @@
import { describe, expect, it } from "vitest";
import { inferCompareKeyColumns } from "../dataCompare";
describe("inferCompareKeyColumns", () => {
it("returns all primary-key columns when a primary key exists", () => {
expect(
inferCompareKeyColumns([
{ name: "id", is_primary_key: true },
{ name: "name", is_primary_key: false },
]),
).toEqual(["id"]);
});
it("returns a composite primary key as-is", () => {
expect(
inferCompareKeyColumns([
{ name: "tenant_id", is_primary_key: true },
{ name: "user_id", is_primary_key: true },
{ name: "name", is_primary_key: false },
]),
).toEqual(["tenant_id", "user_id"]);
});
it("returns an empty array instead of falling back to the first column when there is no primary key", () => {
expect(
inferCompareKeyColumns([
{ name: "category", is_primary_key: false },
{ name: "name", is_primary_key: false },
]),
).toEqual([]);
});
it("returns an empty array for an empty column list", () => {
expect(inferCompareKeyColumns([])).toEqual([]);
});
});
@@ -118,3 +118,24 @@ export interface DataCompareSyncPlan {
syncStatements: string[];
syncSql: string;
}
/**
* Safely infers the columns to use as comparison keys for a table.
*
* Only primary-key columns are trusted: a primary key is unique by
* definition, so comparing on it can never produce duplicate-key failures.
* Composite primary keys are returned as-is.
*
* When no primary key exists an empty array is returned so the caller can ask
* the user to pick matching columns explicitly. Falling back to the first
* column would silently select an arbitrary, possibly non-unique field
* (category, status, date, ...) and make the whole table comparison fail with
* an unhelpful duplicate-key error.
*/
export function inferCompareKeyColumns(columns: Pick<ColumnInfo, "name" | "is_primary_key">[]): string[] {
const primaryKeys: string[] = [];
for (const column of columns) {
if (column.is_primary_key) primaryKeys.push(column.name);
}
return primaryKeys;
}