fix: 防止重复推送通知 — 添加已发送检查 + GXGP 标记已发送

- Pipeline 新增 _exclude_sent() 方法,通知前查询数据库跳过 is_sent=True 的公告
- GXGP Spider mark_sent 改为 True,配合去重逻辑防止每次爬取重复通知
This commit is contained in:
2026-05-09 16:00:04 +08:00
parent 6fbdf097b3
commit d851dafea9
4 changed files with 48 additions and 20 deletions
+22 -16
View File
@@ -45,13 +45,16 @@ async def test_pipeline_filtered_mode():
pipeline = PostCrawlPipeline(db_session=mock_db, notification_service=mock_notify)
with patch.object(pipeline, "_save_to_db", AsyncMock(return_value=2)):
with patch.object(pipeline, "_send_notifications", AsyncMock(return_value=1)):
pipe_result = await pipeline.process(
result.announcements, config
)
assert pipe_result.stored == 2
assert pipe_result.filtered == 1
assert pipe_result.notified == 1
with patch.object(pipeline, "_exclude_sent", AsyncMock(
side_effect=lambda anns: [a for a in anns if "大化" in a["title"]]
)):
with patch.object(pipeline, "_send_notifications", AsyncMock(return_value=1)):
pipe_result = await pipeline.process(
result.announcements, config
)
assert pipe_result.stored == 2
assert pipe_result.filtered == 1
assert pipe_result.notified == 1
@pytest.mark.asyncio
@@ -84,12 +87,15 @@ async def test_pipeline_all_mode():
pipeline = PostCrawlPipeline(db_session=mock_db, notification_service=mock_notify)
with patch.object(pipeline, "_save_to_db", AsyncMock(return_value=2)):
with patch.object(pipeline, "_send_notifications", AsyncMock(return_value=2)):
with patch.object(pipeline, "_mark_sent", AsyncMock(return_value=2)):
pipe_result = await pipeline.process(
result.announcements, config
)
assert pipe_result.stored == 2
assert pipe_result.filtered == 0
assert pipe_result.notified == 2
pipeline._mark_sent.assert_awaited_once()
with patch.object(pipeline, "_exclude_sent", AsyncMock(
side_effect=lambda anns: anns
)):
with patch.object(pipeline, "_send_notifications", AsyncMock(return_value=2)):
with patch.object(pipeline, "_mark_sent", AsyncMock(return_value=2)):
pipe_result = await pipeline.process(
result.announcements, config
)
assert pipe_result.stored == 2
assert pipe_result.filtered == 0
assert pipe_result.notified == 2
pipeline._mark_sent.assert_awaited_once()