Add resumable US policy news collectors and provenance

Collect official crypto and US policy documents with deterministic screening, full bodies and attachments, bounded retries, and durable history cursors. Preserve publication precision and source identity while keeping new collectors disabled.

Add real Congress, GovInfo, DOJ and Federal Register fixtures, offline capture replay, isolated database validation, and rollout instructions. Validate 982 backend tests, frontend checks, and 57 deployment control tests.

backend/src/alphapulse/ai/tools.py

@@ -22,6 +22,7 @@ from alphapulse.api.calendar import calendar_data
 from alphapulse.api.market import call_binance, client, tradable_symbols
 from alphapulse.api.news import news_filters
 from alphapulse.core.db import SessionLocal
+from alphapulse.core.us_news import SOURCES as US_SOURCES
 from alphapulse.models import ContextVersion, NewsItem, SkillVersion
 from alphapulse.models.market import INTERVALS
 
@@ -34,6 +35,8 @@ SOURCE_NAMES = {
     "coindesk": "CoinDesk", "fed": "美联储·政策与讲话", "bls": "美国劳工统计局·CPI/PPI/就业", "bea": "美国经济分析局",
 }
 
+SOURCE_NAMES.update({code: spec.name for code, spec in US_SOURCES.items()})
+
 
 @dataclass
 class ToolOutcome:
@@ -59,7 +62,7 @@ def dump(data: Any) -> str:
 
 class SearchNews(BaseModel):
     query: Annotated[str, Field(description="关键字，按标题和正文字面匹配；比特币/Bitcoin、以太坊/Ethereum、美联储、CPI/PPI/非农/GDP/PCE 等完整主题词会同时匹配中英文；其他词请分别使用原文语言检索")]
-    sources: Annotated[list[Literal["wscn_live", "wscn_news", "coindesk", "fed", "bls", "bea"]] | None, Field(description="限定来源，不填为全部")] = None
+    sources: Annotated[list[Literal["wscn_live", "wscn_news", "coindesk", "fed", "bls", "bea", "sec", "cftc", "treasury", "ofac", "fincen", "whitehouse", "congress", "federal_register", "occ", "doj", "ecb", "boe", "eurostat", "ons", "destatis", "insee"]] | None, Field(description="限定来源，不填为全部")] = None
     start: Annotated[datetime | None, Field(description="开始时间（ISO 8601，带时区），如 2026-10-01T00:00:00+08:00")] = None
     end: Annotated[datetime | None, Field(description="结束时间（ISO 8601，带时区）")] = None
     limit: Annotated[int, Field(ge=1, le=20)] = 10
@@ -70,11 +73,11 @@ async def search_news(args: SearchNews) -> ToolOutcome:
         conditions = news_filters(args.sources, args.query, args.start, args.end)
         total = await session.scalar(select(func.count()).select_from(NewsItem).where(*conditions))
         rows = (await session.execute(
-            select(NewsItem.id, NewsItem.source_code, NewsItem.published_at, NewsItem.title, NewsItem.summary, func.left(NewsItem.content_text, 200))
+            select(NewsItem.id, NewsItem.source_code, NewsItem.published_at, NewsItem.title, NewsItem.summary, func.left(NewsItem.content_text, 200), NewsItem.extra)
             .where(*conditions).order_by(NewsItem.published_at.desc()).limit(args.limit)
         )).all()
     items = [
-        {"id": r[0], "source": SOURCE_NAMES.get(r[1], r[1]), "published_at": local(r[2]), "title": r[3], "brief": r[4] or r[5]}
+        {"id": r[0], "source": SOURCE_NAMES.get(r[1], r[1]), "published_at": r[6].get("published_date") if r[6].get("published_at_precision") == "date" else local(r[2]), "title": r[3], "brief": r[4] or r[5]}
         for r in rows
     ]
     return ToolOutcome(dump({"total": total, "returned": len(items), "items": items}))
@@ -90,10 +93,13 @@ async def get_news(args: GetNews) -> ToolOutcome:
     if news is None:
         return ToolOutcome(dump({"error": f"新闻 {args.id} 不存在"}))
     return ToolOutcome(dump({
-        "id": news.id, "source": SOURCE_NAMES.get(news.source_code, news.source_code), "published_at": local(news.published_at),
+        "id": news.id, "source": SOURCE_NAMES.get(news.source_code, news.source_code), "published_at": news.extra.get("published_date") if news.extra.get("published_at_precision") == "date" else local(news.published_at),
         "title": news.title, "url": news.url, "is_paywalled": news.is_paywalled,
         "content": news.content_text[:NEWS_BODY_LIMIT], "truncated": len(news.content_text) > NEWS_BODY_LIMIT,
         "content_scope": news.extra.get("content_scope", "article"),
+        "published_at_precision": news.extra.get("published_at_precision"),
+        "published_date": news.extra.get("published_date"),
+        "attachments": news.extra.get("attachments", []),
     }))