test: 记录双平台小规模真实验收

tests/integration/test_failure_tolerance.py

@@ -1,5 +1,6 @@
 from sqlalchemy.orm import Session
 
+from app.models import ContentItem, Task
 from app.platforms.base import PlatformAPIError
 from app.schemas import CreateTaskRequest
 from app.services.task_service import create_task, run_task
@@ -11,6 +12,28 @@ class FailingHotspotPlatform:
         raise PlatformAPIError("hotspot failed", error_type="api_error", status_code=500)
 
 
+class OneFailedOneSuccessfulItemPlatform:
+    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-item", item_type="video", title="失败内容"),
+            ContentItemData(source_item_id="good-item", item_type="video", title="成功内容"),
+        ]
+
+    def fetch_comments(self, source_item_id, *, limit):
+        from app.platforms.base import CommentData
+
+        if source_item_id == "bad-item":
+            raise PlatformAPIError("comments failed", error_type="rate_limited", status_code=429)
+        return [CommentData(source_comment_id="c1", content="继续成功")]
+
+
 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())
@@ -49,3 +72,33 @@ def test_unexpected_task_exception_marks_task_failed(monkeypatch):
         assert task.error_stage == "system"
         assert task.error_type == "unexpected_error"
         assert task.error_message == "boom"
+
+
+def test_failed_content_item_is_recorded_and_following_item_continues(monkeypatch):
+    with make_test_client() as (_client, engine):
+        monkeypatch.setattr("app.services.task_service.build_platform", lambda _platform: OneFailedOneSuccessfulItemPlatform())
+        monkeypatch.setattr("app.services.task_service.build_ai_dependencies", lambda: (lambda _prompt: "[]", None))
+        monkeypatch.setattr("app.services.ai_service.time.sleep", lambda _seconds: None)
+
+        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_item = session.query(ContentItem).filter_by(task_id=task_id, source_item_id="bad-item").one()
+            successful_item = session.query(ContentItem).filter_by(task_id=task_id, source_item_id="good-item").one()
+
+        assert persisted.status == "success"
+        assert persisted.total_items_count == 2
+        assert persisted.processed_items_count == 2
+        assert persisted.successful_items_count == 1
+        assert persisted.failed_items_count == 1
+        assert failed_item.status == "failed"
+        assert failed_item.error_stage == "crawl_comments"
+        assert failed_item.error_type == "rate_limited"
+        assert successful_item.status == "success"

tests/unit/test_comment_pagination.py

@@ -38,7 +38,8 @@ def test_tikhub_client_retries_429_with_exponential_backoff(monkeypatch):
 
 
 def test_tikhub_client_raises_structured_error_after_retries(monkeypatch):
-    monkeypatch.setattr("app.platforms.base.time.sleep", lambda _seconds: None)
+    sleeps = []
+    monkeypatch.setattr("app.platforms.base.time.sleep", sleeps.append)
     transport = SequenceTransport([httpx.Response(429, json={"message": "Too Many Requests"}) for _ in range(4)])
     http_client = httpx.Client(transport=httpx.MockTransport(transport))
     client = TikHubClient(base_url="https://api.test", api_key="secret-token", http_client=http_client)
@@ -49,3 +50,4 @@ def test_tikhub_client_raises_structured_error_after_retries(monkeypatch):
     assert exc_info.value.error_type == "rate_limited"
     assert exc_info.value.status_code == 429
     assert "secret-token" not in str(exc_info.value)
+    assert sleeps == [1, 2, 4]