Consider it done. The open-source AI agent that works out of the box · 想到,就能做到。开源、开箱即用的 AI Agent。
TypeScript agentai-agentai-assistantandroidclaude-codecodexelectroniosllmmacosreact-nativetypescriptwindows
/**
* github-client listPullRequestReviewThreads —— 回归测试(PR #558 follow-up)。
*
* github-client 包本身无测试基建(零依赖、只有 build 脚本),故在 @cindy/mcps(已有
* vitest 且依赖该 client)里通过 stub 全局 fetch 覆盖这个 GraphQL 方法的几处易错逻辑:
* - null-guard:不存在的 PR(pullRequest: null)抛 GithubApiError(status=404),
* 让上层 toolCatch 归类成 NOT_FOUND(本次 follow-up 一并修的映射)。
* - pageInfo 透传(分页 hasNextPage / endCursor 原样带回)。
* - commentsTruncated(totalCount > 已取回条数时置 true)。
* - comment.id 取自可为 null 的 databaseId。
*/
import { afterEach, describe, expect, it, vi } from 'vitest';
import { GithubClient, GithubApiError } from '@cindy/github-client';
function graphqlResponse(body: unknown): Response {
return new Response(JSON.stringify(body), {
status: 200,
headers: { 'content-type': 'application/json' },
});
}
function client() {
return new GithubClient({ token: 't', owner: 'makecindy', repo: 'cindy' });
}
afterEach(() => {
vi.unstubAllGlobals();
});
describe('listPullRequestReviewThreads', () => {
it('throws GithubApiError(404) when the PR does not exist (pullRequest: null)', async () => {
vi.stubGlobal('fetch', vi.fn(async () => graphqlResponse({ data: { repository: { pullRequest: null } } })));
const err = await client()
.listPullRequestReviewThreads(999999)
.then(() => null)
.catch((e) => e);
expect(err).toBeInstanceOf(GithubApiError);
expect((err as GithubApiError).status).toBe(404);
});
it('passes pageInfo (hasNextPage / endCursor) through verbatim', async () => {
vi.stubGlobal(
'fetch',
vi.fn(async () =>
graphqlResponse({
data: {
repository: {
pullRequest: {
reviewThreads: {
pageInfo: { hasNextPage: true, endCursor: 'CURSOR_XYZ' },
nodes: [],
},
},
},
},
}),
),
);
const res = await client().listPullRequestReviewThreads(1, { first: 10, after: 'PREV' });
expect(res.pageInfo).toEqual({ hasNextPage: true, endCursor: 'CURSOR_XYZ' });
expect(res.threads).toEqual([]);
});
it('sets commentsTruncated when totalCount exceeds the returned comment count', async () => {
vi.stubGlobal(
'fetch',
vi.fn(async () =>
graphqlResponse({
data: {
repository: {
pullRequest: {
reviewThreads: {
pageInfo: { hasNextPage: false, endCursor: null },
nodes: [
{
id: 'PRRT_full',
isResolved: false,
isOutdated: false,
path: 'a.ts',
line: 1,
comments: {
totalCount: 1,
nodes: [{ databaseId: 111, body: 'x', createdAt: '2026-01-01', author: { login: 'a' } }],
},
},
{
id: 'PRRT_truncated',
isResolved: false,
isOutdated: false,
path: 'b.ts',
line: 2,
comments: {
totalCount: 80, // > 已取回的 1 条 → 截断
nodes: [{ databaseId: null, body: 'y', createdAt: '2026-01-02', author: null }],
},
},
],
},
},
},
},
}),
),
);
const res = await client().listPullRequestReviewThreads(1);
expect(res.threads[0].commentsTruncated).toBe(false);
expect(res.threads[1].commentsTruncated).toBe(true);
// databaseId 为 null 时 comment.id 应原样为 null,不崩。
expect(res.threads[1].comments[0].id).toBeNull();
expect(res.threads[1].comments[0].author).toBeNull();
});
it('fails fast when owner/repo are not configured (no GraphQL call)', async () => {
const spy = vi.fn();
vi.stubGlobal('fetch', spy);
const bare = new GithubClient({ token: 't' });
await expect(bare.listPullRequestReviewThreads(1)).rejects.toThrow(/owner and repo are required/);
expect(spy).not.toHaveBeenCalled();
});
});