feat: add historical news AI backfill and visible task queue

Freeze date-range previews and model settings, track per-news progress and downstream targets, and share completion-driven scheduling with realtime analysis.

Prevent overlapping retries for the same news, retain batch queue associations, and refresh selected details with session-safe pagination.

backend/tests/test_news_backfill.py

@@ -0,0 +1,63 @@
+"""Backfill selection and task progress use isolated model and storage boundaries."""
+import asyncio
+from types import SimpleNamespace as N
+from unittest.mock import AsyncMock, MagicMock
+import pytest
+
+
+def test_progress_is_scoped_and_removed_on_failure():
+    from alphapulse.news_risk.progress import execute, step, active, current
+    async def check():
+        async def task():
+            step('查询资料',tool='read_news')
+            assert len(active)==1 and current.get()['tool']=='read_news'
+            raise ValueError('Synthetic')
+        with pytest.raises(ValueError): await execute(('analysis',7),task)
+        assert active=={} and current.get() is None
+    asyncio.run(check())
+
+
+@pytest.mark.parametrize('published,revisions,expected', [(True,(2,2),True),(True,(1,2),False),(False,(2,2),False)])
+def test_all_current_packages_require_published_results(trading,published,revisions,expected):
+    from alphapulse.news_risk.backfill import classification
+    package=N(result={'summary':'ok'} if published else None,published_revision=revisions[0],revision=revisions[1])
+    session=N(scalar=AsyncMock(return_value=N(id=3,result={'packages':[]})),scalars=AsyncMock(return_value=N(all=lambda:[package])))
+    actual,_,_=asyncio.run(classification(session,N(id=1,content_hash='current')))
+    assert actual is expected
+
+
+def test_empty_active_memberships_are_incomplete(trading):
+    from alphapulse.news_risk.backfill import classification
+    session=N(scalar=AsyncMock(return_value=N(id=3,result={'old':True})),scalars=AsyncMock(return_value=N(all=lambda:[])))
+    assert asyncio.run(classification(session,N(id=1,content_hash='current')))[0] is False
+
+
+def test_expired_preview_does_not_create_work(trading,monkeypatch):
+    from alphapulse.news_risk import backfill
+    monkeypatch.setattr(backfill.redis,'get',AsyncMock(return_value=None))
+    session=MagicMock()
+    with pytest.raises(ValueError,match='预览已过期'):asyncio.run(backfill.start(session,'expired'))
+    session.add.assert_not_called()
+
+
+@pytest.mark.parametrize('modes',[[],['missing','missing'],['review','other'],['other']])
+def test_range_rejects_empty_duplicate_or_unknown_modes(trading,modes):
+    from pydantic import ValidationError
+    from alphapulse.api.risk import ReprocessIn
+    with pytest.raises(ValidationError):
+        ReprocessIn(start='2026-10-01T00:00:00Z',end='2026-11-01T00:00:00Z',modes=modes)
+
+
+@pytest.mark.parametrize('failure',[False,True])
+def test_empty_or_invalid_outlook_poll_does_not_admit_a_task(trading,monkeypatch,failure):
+    from alphapulse.news_risk import scheduler, worker
+    session=MagicMock()
+    session.__aenter__=AsyncMock(return_value=session); session.__aexit__=AsyncMock()
+    session.get=AsyncMock(return_value=N(enabled=True,start_at='configured'))
+    session.scalar=AsyncMock(return_value=None)
+    monkeypatch.setattr(scheduler,'SessionLocal',lambda:session)
+    monkeypatch.setattr(worker,'enqueue',AsyncMock())
+    check=AsyncMock(side_effect=ValueError('Synthetic configuration failure') if failure else None,return_value=False)
+    monkeypatch.setattr(worker,'outlook',check)
+    assert asyncio.run(scheduler.candidate(set(),0)) is None
+    check.assert_awaited_once_with(prepare_only=True)

backend/tests/test_news_concurrency.py

@@ -12,6 +12,7 @@ def fake_session(settings):
     session.get = AsyncMock(return_value=settings)
     session.scalars = AsyncMock(return_value=N(all=lambda: []))
     session.commit = AsyncMock()
+    session.execute = AsyncMock()
     return session
 
 
@@ -117,7 +118,7 @@ def test_history_completion_is_counted_independently_of_item_order(trading, monk
         items=[{'id':i,'hash':'current','state':'pending'} for i in (1,2,3)])
     article = N(id=30, state='done', error=None)
     session = fake_session(None)
-    session.scalar = AsyncMock(side_effect=[job, article])
+    session.scalar = AsyncMock(side_effect=[job, None, article])
     session.get = AsyncMock(side_effect=lambda model,*a,**kw: {
         RiskSettings:N(enabled=True), ReprocessJob:job,
         ArticleAnalysis:article, NewsItem:N(id=3, content_hash='current')}[model])

backend/tests/test_news_reprocess.py

@@ -21,7 +21,7 @@ def test_pause_after_claim_does_not_count_old_success(trading, monkeypatch):
     job = N(id=20, state='pending', position=0, items=[{'id':21,'hash':'abc'}], failures=[], settings={})
     old = N(id=25, state='done', error=None, result={'old_success':True})
     session = session_mock()
-    session.scalar.side_effect = [None, None, job, old]
+    session.scalar.side_effect = [None, None, job, None, old]
     session.get.side_effect = lambda model, *a, **kw: {
         RiskSettings:N(enabled=True), NewsItem:N(id=21,content_hash='abc'), ArticleAnalysis:old, ReprocessJob:job}[model]
     paused = session_mock()
@@ -59,7 +59,7 @@ def test_downstream_failure_does_not_mark_unrelated_news_failed(trading, monkeyp
 @pytest.mark.parametrize('stage', ['grouping', 'analysis', 'events', 'outlook'])
 def test_retry_resumes_only_failed_stage(trading, stage):
     from alphapulse.api.risk import reprocess_retry
-    job = N(id=10,state='failed',position=2,items=[{'id':11},{'id':12}],
+    job = N(id=10,state='failed',position=2,settings={},items=[{'id':11},{'id':12}],
         failures=[{'stage':stage,'id':11,'error':'Synthetic failure'}])
     failed = N(id=11,state='failed',quotes={})
     session = session_mock()
@@ -81,7 +81,7 @@ def test_job_outlook_retry_waits_for_its_grouping_retry(trading, monkeypatch):
     from alphapulse.news_risk import reprocess, worker, events
     session = session_mock()
     session.get.return_value = N(enabled=True)
-    session.scalar.side_effect = [None, N(quotes={'job_id':10}), N(id=10,position=0,items=[{'id':11}])]
+    session.scalar.side_effect = [None, N(quotes={'job_id':10}), N(id=10,position=0,settings={},items=[{'id':11}])]
     monkeypatch.setattr(worker,'SessionLocal',lambda:session)
     monkeypatch.setattr(events,'reading_inputs',AsyncMock(return_value=[]))
     current = AsyncMock(side_effect=AssertionError('Must not use current settings for a job retry'))