diff --git a/docs/MVP-WorkOrders.md b/docs/MVP-WorkOrders.md index da92ef6..8162244 100644 --- a/docs/MVP-WorkOrders.md +++ b/docs/MVP-WorkOrders.md @@ -470,6 +470,27 @@ test: 记录双平台默认规模真实验收 fix: 完善失败重试与跳过边界 ``` +完成记录： + +```text +完成日期：2026-07-03 +相关 commit：fix: 完善失败重试与跳过边界 +验证命令： +- .venv/bin/python -m pytest tests/integration/test_failure_tolerance.py tests/unit/test_report_stats.py tests/integration/test_task_creation.py tests/unit/test_comment_pagination.py tests/unit/test_ai_schema.py -q +- .venv/bin/python -m pytest tests/unit tests/integration -q +- .venv/bin/python -m pytest tests/unit tests/integration --cov=app --cov-branch --cov-report=term-missing +- curl -f http://localhost:8000/health +验收结论： +- 连续 429 超限测试覆盖并断言退避序列 1s→2s→4s，错误类型为 rate_limited 且不泄露 API Key。 +- 单个内容条目评论抓取失败时记录 failed item，后续内容继续处理，任务可成功。 +- 所有内容条目失败时任务 status=failed，error_stage/error_type/error_message 可见。 +- AI 全批失败已有测试覆盖：评论标记 failed，任务 analysis_status=insufficient。 +- 报告摘要失败时 item/hotspot 报告仍创建，并使用默认总结文案。 +- 已有 running 任务时拒绝新建任务，返回 400。 +- 未发现 httpx.AsyncClient 使用。 +遗留问题：W04 发现的 SQLite WAL/SHM deleted 句柄问题转入 WO-06。 +``` + ### WO-06 SQLite 数据库稳定性与恢复策略 优先级：P0 diff --git a/tests/integration/test_failure_tolerance.py b/tests/integration/test_failure_tolerance.py index c27e73a..e49f01a 100644 --- a/tests/integration/test_failure_tolerance.py +++ b/tests/integration/test_failure_tolerance.py @@ -34,6 +34,24 @@ class OneFailedOneSuccessfulItemPlatform: return [CommentData(source_comment_id="c1", content="继续成功")] +class AllItemsFailPlatform: + def fetch_hotspots(self, *, limit): + from app.platforms.base import HotspotData + + return [HotspotData(source_hot_id="h1", title="热点一", rank=1)] + + def search_items_by_hotspot(self, keyword, *, limit): + from app.platforms.base import ContentItemData + + return [ + ContentItemData(source_item_id="bad-one", item_type="video", title="失败内容一"), + ContentItemData(source_item_id="bad-two", item_type="video", title="失败内容二"), + ] + + def fetch_comments(self, source_item_id, *, limit): + raise PlatformAPIError("comments exhausted", error_type="rate_limited", status_code=429) + + def test_hotspot_failure_marks_task_failed(monkeypatch): with make_test_client() as (_client, engine): monkeypatch.setattr("app.services.task_service.build_platform", lambda _platform: FailingHotspotPlatform()) @@ -102,3 +120,30 @@ def test_failed_content_item_is_recorded_and_following_item_continues(monkeypatc assert failed_item.error_stage == "crawl_comments" assert failed_item.error_type == "rate_limited" assert successful_item.status == "success" + + +def test_task_fails_with_visible_reason_when_all_content_items_fail(monkeypatch): + with make_test_client() as (_client, engine): + monkeypatch.setattr("app.services.task_service.build_platform", lambda _platform: AllItemsFailPlatform()) + + with Session(engine) as session: + task = create_task( + session, + CreateTaskRequest(platform="douyin", hotspot_limit=1, item_limit_per_hotspot=2, comment_limit_per_item=10), + submit_background=False, + ) + task_id = task.id + run_task(task_id, session_factory=lambda: session) + + persisted = session.get(Task, task_id) + failed_items = session.query(ContentItem).filter_by(task_id=task_id, status="failed").all() + + assert persisted.status == "failed" + assert persisted.total_items_count == 2 + assert persisted.processed_items_count == 2 + assert persisted.successful_items_count == 0 + assert persisted.failed_items_count == 2 + assert persisted.error_stage == "crawl_comments" + assert persisted.error_type == "rate_limited" + assert persisted.error_message == "comments exhausted" + assert len(failed_items) == 2 diff --git a/tests/unit/test_report_stats.py b/tests/unit/test_report_stats.py index 4bdb878..e90390b 100644 --- a/tests/unit/test_report_stats.py +++ b/tests/unit/test_report_stats.py @@ -5,7 +5,7 @@ from sqlalchemy.orm import Session from app.db import Base from app.models import Comment, ContentItem, Hotspot, Task -from app.services.report_service import build_comment_metrics, generate_item_report +from app.services.report_service import DEFAULT_SUMMARY, build_comment_metrics, generate_hotspot_report, generate_item_report def test_build_comment_metrics_counts_sentiments_and_top_labels(): @@ -74,3 +74,55 @@ def test_generate_item_report_persists_markdown_and_default_summary(): assert "样本评论数量" in report.markdown_content finally: engine.dispose() + + +def test_generate_hotspot_report_persists_default_summary_when_ai_summary_fails(): + engine = create_engine("sqlite:///:memory:", connect_args={"check_same_thread": False}) + Base.metadata.create_all(engine) + try: + with Session(engine) as session: + task = Task(platform="xiaohongshu", status="success") + session.add(task) + session.flush() + hotspot = Hotspot(task_id=task.id, platform="xiaohongshu", title="热点", raw_data="{}") + session.add(hotspot) + session.flush() + item = ContentItem( + task_id=task.id, + hotspot_id=hotspot.id, + platform="xiaohongshu", + source_item_id="n1", + item_type="note", + title="笔记", + status="success", + raw_data="{}", + ) + session.add(item) + session.flush() + session.add( + Comment( + task_id=task.id, + hotspot_id=hotspot.id, + content_item_id=item.id, + platform="xiaohongshu", + source_comment_id="c1", + content="好评", + sentiment="positive", + labels='["认可"]', + raw_data="{}", + ) + ) + session.commit() + + report = generate_hotspot_report( + session, + hotspot.id, + summary_provider=lambda *_args, **_kwargs: (_ for _ in ()).throw(RuntimeError("summary failed")), + ) + + assert report.report_type == "hotspot" + assert report.hotspot_id == hotspot.id + assert report.summary == DEFAULT_SUMMARY + assert DEFAULT_SUMMARY in report.markdown_content + finally: + engine.dispose()