Clarify news risk evidence validation

backend/src/alphapulse/api/risk.py

@@ -58,7 +58,10 @@ async def enable(body: SettingsIn, session: DB):
         provider = await provider_for(session, settings)
     except ValueError as error:
         raise HTTPException(400, str(error)) from error
-    fixture = {'news': [{'id': 1, 'title': '测试消息：暂无新事件', 'summary': '', 'content': ''}], 'candidates': []}
+    fixture = {'news': [{'id': 1, 'title': '虚构测试新闻：示例交易平台发布维护通知',
+        'summary': '示例交易平台计划进行例行维护，期间暂停充提服务。',
+        'content': '本条为虚构测试样例，不代表真实市场事件。示例交易平台计划进行例行维护，'
+                   '期间暂停充提服务。通知未提及资产损失，恢复时间尚未公布。'}], 'candidates': []}
     from alphapulse.news_risk.service import validate_labels
     try:
         result, _ = await complete(provider, SYSTEM, fixture, Labels)

backend/src/alphapulse/news_risk/service.py

@@ -15,7 +15,10 @@ SYSTEM = '''你是加密市场新闻风险分析员，只使用输入证据。
 重大传闻可以高星，但必须标为待核实；正文不足用 unknown，不能冒充无风险。
 分别记录方向、证实状态、依据和不确定性。不得声称已独立核实；转载数量不等于独立证实。
 official 仅限输入明确来自相关官方机构的公告，媒体转述不能当作官方原文。
-summary 用中文简要归纳；evidence 是输入原文的短引用；entities 使用统一名称。
+summary 用中文简要归纳；entities 使用统一名称。
+evidence 必须逐字复制对应新闻 title、summary 或 content 中一段非空、连续的原文（最多500字）。
+只返回引用文字本身，不加外层引号、说明、翻译或省略号，不拼接多处片段；不能引用其他新闻或候选事件。
+即使没有明显风险或信息不足，也必须从该条新闻复制原文作为 evidence；分析说明写入 reasoning。
 只有相同实体、事件类型、发生时段及具体事实一致才归并到候选 event_id，否则 null。
 事件 followup/refutation 使用候选 id；更正或辟谣可降低当前星级。不要仅因同名机构合并不同事件。
 所有输出都是研究判断，不下单、不修改任何交易设置。'''
@@ -34,7 +37,8 @@ class Label(BaseModel):
     confidence: Literal['low', 'medium', 'high']
     reasoning: str = Field(min_length=1, max_length=2000)
     uncertainty: str = Field(min_length=1, max_length=1000)
-    evidence: str = Field(min_length=1, max_length=500)
+    evidence: str = Field(min_length=1, max_length=500,
+        description='逐字复制本条新闻 title、summary 或 content 中一段连续原文；不加外层引号，不改写、翻译或拼接')
     event_id: int | None
     same_event_as: int | None = Field(default=None, description="同批较早输入的新闻 id，仅同一事件时填写")
 
@@ -100,8 +104,15 @@ def validate_labels(result, inputs, candidates):
                 or positions[item.same_event_as] >= positions[item.id]):
             raise ValueError("同批事件引用必须指向较早的输入，且不能同时指定已有事件")
         source = inputs[item.id]
-        if item.evidence not in '\n'.join(str(source.get(key, '')) for key in ('title', 'summary', 'content')):
-            raise ValueError('模型证据并非输入原文引用')
+        texts = [source.get(key) or '' for key in ('title', 'summary', 'content')]
+        evidence = item.evidence.strip()
+        # Accept presentation quotes only; the stored evidence must remain a literal source substring.
+        if not any(evidence in text for text in texts) and len(evidence) >= 2 and (evidence[0], evidence[-1]) in (
+                ('“', '”'), ('「', '」'), ('‘', '’'), ('"', '"'), ("'", "'")):
+            evidence = evidence[1:-1].strip()
+        if not evidence or not any(evidence in text for text in texts):
+            raise ValueError(f'模型证据并非输入原文引用（新闻 ID {item.id}）；请逐字复制该条新闻中的连续原文，不要改写或拼接')
+        item.evidence = evidence
         if item.event_id is not None and item.event_id not in candidates:
             raise ValueError('模型引用未知候选事件')
         if item.verification == 'insufficient' and item.direction != 'unknown':

backend/tests/test_news_risk.py

@@ -27,6 +27,66 @@ def test_extreme_unverified_event_preserves_uncertainty(risk):
     assert item.stars == 5 and item.verification == 'unverified'
 
 
+@pytest.mark.parametrize('evidence', ['安全事件', ' 安全事件\n', '“安全事件”', '「安全事件」', '‘安全事件’', '"安全事件"', "'安全事件'"])
+def test_evidence_presentation_quotes_preserve_literal_source(risk, evidence):
+    item = label(risk).model_copy(update={'evidence': evidence})
+    risk.validate_labels(risk.Labels(items=[item]), {1: {'content': '媒体报道安全事件，规模待核实。'}}, set())
+    assert item.evidence == '安全事件'
+
+
+@pytest.mark.parametrize('evidence', ['“资产已被盗”', '安全…事件', 'security incident', '   ', '“”', '标题\n正文'])
+def test_evidence_must_be_nonempty_and_from_one_source_field(risk, evidence):
+    item = label(risk).model_copy(update={'evidence': evidence})
+    with pytest.raises(ValueError, match='模型证据并非输入原文引用'):
+        risk.validate_labels(risk.Labels(items=[item]), {1: {'title': '标题', 'summary': '正文', 'content': '安全事件'}}, set())
+
+
+def test_evidence_retains_quotes_present_in_source(risk):
+    item = label(risk).model_copy(update={'evidence': '“安全事件”'})
+    risk.validate_labels(risk.Labels(items=[item]), {1: {'content': '媒体称这是“安全事件”。'}}, set())
+    assert item.evidence == '“安全事件”'
+
+
+@pytest.mark.parametrize('valid', [True, False])
+def test_enable_validates_evidence_before_saving_settings(risk, monkeypatch, valid):
+    from types import SimpleNamespace
+    from unittest.mock import AsyncMock
+    from fastapi import FastAPI
+    import httpx
+    from alphapulse.api import risk as api
+
+    session = SimpleNamespace(merge=AsyncMock(), commit=AsyncMock())
+    provider = SimpleNamespace(model='news-only')
+    monkeypatch.setattr(api, 'provider_for', AsyncMock(return_value=provider))
+
+    async def complete(actual_provider, system, payload, schema):
+        assert actual_provider is provider
+        news = payload['news'][0]
+        assert news['content'] and news['summary']
+        evidence = '“'+news['summary']+'”' if valid else '虚构的资产损失6%'
+        return risk.Labels(items=[label(risk).model_copy(update={'evidence': evidence})]), {}
+
+    monkeypatch.setattr(api, 'complete', complete)
+    app = FastAPI()
+    app.include_router(api.router)
+    app.dependency_overrides[api.require_admin] = lambda: None
+    app.dependency_overrides[api.get_session] = lambda: session
+
+    async def check():
+        async with httpx.AsyncClient(transport=httpx.ASGITransport(app), base_url='http://test') as client:
+            response = await client.post('/api/risk/enable', json={'provider_id': 1, 'model': 'news-only'})
+            assert response.status_code == (200 if valid else 502), response.text
+            if valid:
+                saved = session.merge.call_args.args[0]
+                assert saved.enabled and saved.model == 'news-only'
+                session.commit.assert_awaited_once()
+            else:
+                assert '模型证据并非输入原文引用' in response.json()['detail']
+                session.merge.assert_not_awaited()
+                session.commit.assert_not_awaited()
+    asyncio.run(check())
+
+
 @pytest.mark.parametrize('kind', ['missing', 'duplicate', 'unknown', 'evidence', 'candidate', 'insufficient'])
 def test_invalid_results_are_rejected(risk, kind):
     item = label(risk)