fix: 修复补跑误含 cninfo 三步与时区不一致 (P1-2/P1-3)

- P1-2: 提取 DEFAULT_NEWS_STEPS(全链路去 report+cninfo),定时/补跑/默认三处统一;
  启动补跑不再误执行 cninfo 公告管道
- P1-3: 新增 scheduler/timeutil.py(调度时区统一入口,SCHEDULE_TZ 可覆盖,
  默认 Asia/Shanghai);run_scheduler 定时/补跑/--once、crawler 补跑保护、
  reporter 兜底日期全部改用调度时区,避免系统时区非上海时日期错位
- 新增 6 个测试;全量 282 passed

tests/test_incremental.py

@@ -8,6 +8,7 @@
 from __future__ import annotations
 
 import json
+from datetime import UTC
 from pathlib import Path
 from types import SimpleNamespace
 
@@ -306,8 +307,7 @@ def test_crawler_today_runs_normally(monkeypatch: pytest.MonkeyPatch) -> None:
 
 def test_report_step_uses_date_str(monkeypatch: pytest.MonkeyPatch) -> None:
     """report 步骤必须使用传入的 date_str,而非 date.today()(P0-4)。"""
-    from scheduler import pipeline
-    from scheduler import reporter
+    from scheduler import pipeline, reporter
 
     received: list[str] = []
 
@@ -346,3 +346,106 @@ def test_pipeline_backfill_skips_crawler_keeps_rest(
     assert all(s.success for s in result.steps)
     state = pipeline._load_pipeline_state(state_path)
     assert state["20260101"]["crawler"]["status"] == "ok"
+
+
+# --------------------------------------------------------------------------- #
+# P1-2: 默认步骤不含 cninfo 独立管道 / P1-3: 调度时区统一
+# --------------------------------------------------------------------------- #
+
+def test_default_news_steps_excludes_cninfo_and_report() -> None:
+    """新闻链路默认步骤不含 report 与 cninfo 三步(P1-2)。"""
+    from scheduler import DEFAULT_NEWS_STEPS
+
+    assert "report" not in DEFAULT_NEWS_STEPS
+    assert "cninfo_crawl" not in DEFAULT_NEWS_STEPS
+    assert "cninfo_extract" not in DEFAULT_NEWS_STEPS
+    assert "cninfo_pdf" not in DEFAULT_NEWS_STEPS
+    # 新闻链路核心步骤齐全
+    assert {"crawler", "xwlb", "extractor", "dedup", "llm",
+            "embedding", "qdrant"} <= set(DEFAULT_NEWS_STEPS)
+
+
+def test_run_pipeline_default_steps_excludes_cninfo(
+    tmp_path: Path, monkeypatch: pytest.MonkeyPatch,
+) -> None:
+    """run_pipeline 默认 steps 不含 cninfo(只跑新闻链路)。"""
+    from scheduler import pipeline
+    from scheduler.timeutil import today_str
+
+    calls: list[str] = []
+
+    def _fake_run(cmd, timeout=None):  # noqa: ARG001
+        name = next(c.split(".")[-1] for c in cmd if "scripts.run_" in c)
+        calls.append(name)
+        return SimpleNamespace(returncode=0)
+
+    monkeypatch.setattr(pipeline.subprocess, "run", _fake_run)
+    state_path = tmp_path / "state.json"
+    today = today_str()  # 今天:避免 crawler 补跑保护跳过
+    result = pipeline.run_pipeline(
+        today, state_path=state_path,
+    )
+    assert result.all_success is True
+    assert calls == ["run_crawler", "run_xwlb", "run_extractor", "run_dedup",
+                     "run_event_extraction", "run_embedding", "run_qdrant_ingest"]
+    assert "cninfo" not in " ".join(calls)
+
+
+def test_scheduler_daemon_backfill_uses_default_news_steps(
+    monkeypatch: pytest.MonkeyPatch,
+) -> None:
+    """守护进程补跑 steps 使用 DEFAULT_NEWS_STEPS(不含 cninfo 三步)。"""
+    import inspect
+
+    import scripts.run_scheduler as rs
+    from scheduler import DEFAULT_NEWS_STEPS
+
+    # 验证补跑逻辑引用的常量:通过源码断言 + 常量内容双重保证
+    src = inspect.getsource(rs._daemon)
+    assert "list(DEFAULT_NEWS_STEPS)" in src
+    assert "cninfo_crawl" not in DEFAULT_NEWS_STEPS
+
+
+def test_today_str_matches_schedule_timezone(monkeypatch: pytest.MonkeyPatch) -> None:
+    """today_str 按调度时区返回 YYYYMMDD(P1-3)。"""
+    from datetime import datetime
+    from zoneinfo import ZoneInfo
+
+    from scheduler.timeutil import today_str
+
+    s = today_str()
+    assert len(s) == 8 and s.isdigit()
+    # 与调度时区(默认 Asia/Shanghai)当前日期一致
+    expect = datetime.now(ZoneInfo("Asia/Shanghai")).strftime("%Y%m%d")
+    assert s == expect
+
+
+def test_today_str_respects_env_override(monkeypatch: pytest.MonkeyPatch) -> None:
+    """SCHEDULE_TZ 环境变量可覆盖调度时区(P1-3)。"""
+    from datetime import datetime
+
+    from scheduler.timeutil import today_str
+
+    # 覆盖为 UTC 后,today_str 应返回 UTC 日期(而非默认 Asia/Shanghai)
+    monkeypatch.setenv("SCHEDULE_TZ", "UTC")
+    s = today_str()
+    expect = datetime.now(UTC).strftime("%Y%m%d")
+    assert s == expect
+
+
+def test_once_uses_today_str_when_date_missing(monkeypatch: pytest.MonkeyPatch) -> None:
+    """--once 不带 --date 时按调度时区取今天(P1-3)。"""
+    import scripts.run_scheduler as rs
+
+    captured: dict[str, str] = {}
+
+    def _fake_pipeline(date_str, steps=None, resume=False):  # noqa: ARG001
+        captured["date"] = date_str
+
+    monkeypatch.setattr(rs, "run_pipeline", _fake_pipeline)
+    from types import SimpleNamespace
+
+    args = SimpleNamespace(steps=None, resume=False, date=None)
+    rc = rs._once(args)
+    assert rc == 0
+    assert captured["date"] == rs.today_str()