feat(ops): 执行日志与 cookie 失效遥测入库

新增 PostgreSQL 运维审计层，本地 run.json 仍是权威源，数据库做跨机器可查、可长期留存的关系型镜像。数据面： - ops_workflow_runs / ops_workflow_run_steps：每次运行与每次步骤尝试一行，保留状态、模式、耗时、exit_code、错误与 journal 路径 - ops_account_cookie_events：登录态观测事件流，只存 cookie 指纹，不落明文值 - ops_account_cookie_lifetimes / ops_account_cookie_stats：一份 cookie 的实际寿命与账号级平均/中位/最短/最长，用于发现失效规律；同时记录 cookie 自带的声明过期时间，可与实际寿命对比区分自然到期与服务端提前吊销采集点： - RunIndex.index_journal 镜像运行记录，覆盖 CLI、控制台与调度器全部路径 - CookieGate 的 gate 判定与 AccountKeepaliveManager 的周期探活（默认 10 分钟）共同提供 cookie 观测，探活决定失效时间的精度可用性： - 新增 gyxx audit runs|cookies|cookie-stats|schema，支持 --json - 首次写入自动幂等建表，建表用 advisory lock 串行化，避免并发 DDL 死锁 - 审计写入永不抛出异常：连续失败熔断冷却并落 logs/ops-audit 日志，数据库不可用不影响工作流执行
2026-10-19 11:01:14 +08:00
1 parent cf85985753
commit 2ef635be0e
@@ -0,0 +1,359 @@
+from __future__ import annotations
+
+import json
+from datetime import date, datetime, timezone
+from pathlib import Path
+from typing import Any
+
+import pytest
+
+from gyxx_flow.ops.audit import (
+    SCHEMA_SQL,
+    AuditStore,
+    split_step_key,
+    summarize_cookies,
+)
+from gyxx_flow.ops.index import RunIndex
+
+
+DOUYIN_COOKIES: list[dict[str, Any]] = [
+    {
+        "name": "sessionid",
+        "value": "AAA",
+        "domain": ".jinritemai.com",
+        "expires": 1766000000.0,
+    },
+    {
+        "name": "sid_tt",
+        "value": "BBB",
+        "domain": ".jinritemai.com",
+        "expires": 1766100000.0,
+    },
+    {
+        "name": "unrelated",
+        "value": "CCC",
+        "domain": ".example.com",
+        "expires": 1700000000.0,
+    },
+    {
+        "name": "session",
+        "value": "DDD",
+        "domain": ".jinritemai.com",
+        "expires": -1,
+    },
+]
+
+
+class _FakeCursor:
+    def __init__(self, log: list[Any], rows: list[dict[str, Any]]) -> None:
+        self._log = log
+        self._rows = rows
+
+    def __enter__(self) -> "_FakeCursor":
+        return self
+
+    def __exit__(self, *_: object) -> bool:
+        return False
+
+    def execute(self, sql: str, params: Any = None) -> None:
+        self._log.append(("execute", sql, params))
+
+    def executemany(self, sql: str, rows: Any) -> None:
+        self._log.append(("executemany", sql, list(rows)))
+
+    def fetchall(self) -> list[dict[str, Any]]:
+        return self._rows
+
+    def fetchone(self) -> dict[str, Any] | None:
+        # 默认模拟「ops_audit_schema 尚不存在」，从而走首次建表分支。
+        return None
+
+    def close(self) -> None:
+        return None
+
+
+class _FakeConnection:
+    def __init__(self, log: list[Any], rows: list[dict[str, Any]] | None = None) -> None:
+        self._log = log
+        self._rows = rows or []
+
+    def cursor(self) -> _FakeCursor:
+        return _FakeCursor(self._log, self._rows)
+
+    def commit(self) -> None:
+        self._log.append(("commit",))
+
+    def close(self) -> None:
+        self._log.append(("close",))
+
+
+def _store_with_fake(
+    tmp_path: Path,
+    *,
+    environment: dict[str, str] | None = None,
+    rows: list[dict[str, Any]] | None = None,
+) -> tuple[AuditStore, list[Any]]:
+    log: list[Any] = []
+    store = AuditStore(
+        tmp_path,
+        environment=environment or {"GYXX_POSTGRES_DSN": "postgresql://fake/audit"},
+        connect=lambda: _FakeConnection(log, rows),
+    )
+    return store, log
+
+
+# ----------------------------------------------------------------------
+# cookie 快照
+# ----------------------------------------------------------------------
+
+
+def test_summarize_cookies_keeps_only_required_cookies() -> None:
+    snapshot = summarize_cookies(
+        DOUYIN_COOKIES,
+        required_domains=("jinritemai.com",),
+        required_names=("sessionid", "sid_tt", "session"),
+    )
+
+    assert snapshot.cookie_count == 4
+    assert snapshot.required_count == 3
+    # 最早的必需 cookie 过期时间：1766000000，而不是无关 cookie 的 1700000000
+    assert snapshot.declared_expires_at == datetime.fromtimestamp(
+        1766000000.0, tz=timezone.utc
+    )
+
+
+def test_summarize_cookies_ignores_session_cookies_for_expiry() -> None:
+    snapshot = summarize_cookies(
+        DOUYIN_COOKIES,
+        required_domains=("jinritemai.com",),
+        required_names=("session",),
+    )
+
+    assert snapshot.required_count == 1
+    assert snapshot.declared_expires_at is None
+
+
+def test_summarize_cookies_fingerprint_tracks_value_changes_only() -> None:
+    baseline = summarize_cookies(DOUYIN_COOKIES, required_domains=("jinritemai.com",))
+    reordered = summarize_cookies(
+        list(reversed(DOUYIN_COOKIES)), required_domains=("jinritemai.com",)
+    )
+    replaced = [dict(cookie) for cookie in DOUYIN_COOKIES]
+    replaced[0]["value"] = "ROTATED"
+
+    assert baseline.fingerprint  # 非空
+    assert baseline.fingerprint == reordered.fingerprint
+    assert baseline.fingerprint != summarize_cookies(
+        replaced, required_domains=("jinritemai.com",)
+    ).fingerprint
+
+
+def test_summarize_cookies_without_requirements_counts_everything() -> None:
+    snapshot = summarize_cookies(DOUYIN_COOKIES)
+
+    assert snapshot.required_count == 4
+    assert snapshot.declared_expires_at == datetime.fromtimestamp(
+        1700000000.0, tz=timezone.utc
+    )
+
+
+def test_summarize_cookies_has_no_plaintext_value_in_fingerprint() -> None:
+    snapshot = summarize_cookies(DOUYIN_COOKIES, required_domains=("jinritemai.com",))
+
+    assert "AAA" not in snapshot.fingerprint
+    assert len(snapshot.fingerprint) == 32
+
+
+def test_split_step_key_handles_retry_suffix() -> None:
+    assert split_step_key("sync.attempt-2") == ("sync", 2)
+    assert split_step_key("collect_collaborators") == ("collect_collaborators", 1)
+    assert split_step_key("a.attempt-x") == ("a.attempt-x", 1)
+
+
+# ----------------------------------------------------------------------
+# 写入装配
+# ----------------------------------------------------------------------
+
+
+def test_record_run_upserts_run_and_steps(tmp_path: Path) -> None:
+    store, log = _store_with_fake(tmp_path)
+
+    ok = store.record_run(
+        {
+            "workflow_id": "content.metrics.daily",
+            "run_id": "content.metrics.daily__20261018__abc",
+            "business_date": "2026-10-18",
+            "status": "success",
+            "mode": "dry_run",
+            "shadow": False,
+            "started_at": "2026-10-19T01:41:43+00:00",
+            "ended_at": "2026-10-19T01:41:48+00:00",
+            "error": None,
+            "journal_path": "runs/content.metrics.daily/run.json",
+        },
+        {
+            "sync.attempt-1": {
+                "status": "success",
+                "exit_code": 0,
+                "started_at": "2026-10-19T01:41:44+00:00",
+                "ended_at": "2026-10-19T01:41:46+00:00",
+            },
+            "collect.attempt-1": {"status": "failed", "exit_code": 4, "error": "boom"},
+        },
+    )
+
+    assert ok is True
+    statements = [entry for entry in log if entry[0] == "execute"]
+    assert any("CREATE TABLE IF NOT EXISTS ops_workflow_runs" in item[1] for item in statements)
+
+    insert = next(item for item in statements if "INSERT INTO ops_workflow_runs" in item[1])
+    params = insert[2]
+    assert params["business_date"] == date(2026, 10, 18)
+    assert params["duration_seconds"] == 5
+    assert params["step_total"] == 2
+    assert params["step_success"] == 1
+    assert params["step_failed"] == 1
+
+    batch = next(item for item in log if item[0] == "executemany")
+    rows = {item["step_id"]: item for item in batch[2]}
+    assert rows["sync"]["attempt"] == 1
+    assert rows["sync"]["duration_seconds"] == 2
+    assert rows["collect"]["exit_code"] == 4
+    assert rows["collect"]["error"] == "boom"
+
+
+def test_record_run_rejects_incomplete_payload(tmp_path: Path) -> None:
+    store, log = _store_with_fake(tmp_path)
+
+    assert store.record_run({"workflow_id": "x"}) is False
+    assert not [entry for entry in log if entry[0] == "execute"]
+
+
+def test_record_cookie_event_writes_value_free_snapshot(tmp_path: Path) -> None:
+    store, log = _store_with_fake(tmp_path)
+    snapshot = summarize_cookies(DOUYIN_COOKIES, required_domains=("jinritemai.com",))
+
+    ok = store.record_cookie_event(
+        account_id="douyin-shop",
+        state="healthy",
+        snapshot=snapshot,
+        probe_status="success",
+        probe_source="keepalive",
+        workflow_id="content.metrics.daily",
+        run_id="run-1",
+    )
+
+    assert ok is True
+    insert = next(
+        item
+        for item in log
+        if item[0] == "execute" and "INSERT INTO ops_account_cookie_events" in item[1]
+    )
+    params = insert[2]
+    assert params["account_id"] == "douyin-shop"
+    assert params["state"] == "healthy"
+    assert params["fingerprint"] == snapshot.fingerprint
+    assert params["declared_expires_at"] == snapshot.declared_expires_at
+    # 明文值不得出现在任何参数里
+    assert "AAA" not in json.dumps(params, default=str)
+
+
+def test_record_cookie_event_rejects_unknown_state(tmp_path: Path) -> None:
+    store, _ = _store_with_fake(tmp_path)
+
+    with pytest.raises(ValueError):
+        store.record_cookie_event(account_id="a", state="definitely-not-valid")
+
+
+# ----------------------------------------------------------------------
+# 降级与熔断
+# ----------------------------------------------------------------------
+
+
+def test_missing_dsn_never_raises_and_logs_failure(tmp_path: Path) -> None:
+    store = AuditStore(tmp_path, environment={})
+
+    assert store.record_run(
+        {
+            "workflow_id": "content.metrics.daily",
+            "run_id": "r1",
+            "business_date": "2026-10-18",
+            "status": "success",
+        }
+    ) is False
+    assert "审计数据库配置缺失" in store.last_error
+    failure_log = tmp_path / "logs" / "ops-audit" / "audit-failures.log"
+    assert failure_log.exists()
+    assert json.loads(failure_log.read_text(encoding="utf-8").splitlines()[0])["operation"] == (
+        "record_run"
+    )
+
+
+def test_audit_can_be_disabled_without_touching_the_database(tmp_path: Path) -> None:
+    store, log = _store_with_fake(
+        tmp_path, environment={"GYXX_POSTGRES_DSN": "postgresql://fake", "GYXX_OPS_AUDIT": "off"}
+    )
+
+    assert store.enabled is False
+    assert store.record_cookie_event(account_id="a", state="healthy") is False
+    assert log == []
+
+
+def test_consecutive_failures_open_the_circuit(tmp_path: Path) -> None:
+    attempts = 0
+
+    def _boom() -> Any:
+        nonlocal attempts
+        attempts += 1
+        raise OSError("database unreachable")
+
+    store = AuditStore(
+        tmp_path,
+        environment={"GYXX_POSTGRES_DSN": "postgresql://fake", "GYXX_OPS_AUDIT_FAILURE_COOLDOWN": "60"},
+        connect=_boom,
+    )
+    payload = {
+        "workflow_id": "w",
+        "run_id": "r",
+        "business_date": "2026-10-18",
+        "status": "success",
+    }
+
+    for _ in range(3):
+        store.record_run(payload)
+    assert attempts == 3
+
+    # 熔断生效后不再尝试连接，避免每个工作流都卡在 TCP 超时上
+    assert store.record_run(payload) is False
+    assert attempts == 3
+
+
+# ----------------------------------------------------------------------
+# 与运行索引的集成
+# ----------------------------------------------------------------------
+
+
+def test_run_index_still_works_when_audit_is_unavailable(tmp_path: Path) -> None:
+    from gyxx_flow.core.context import RunContext
+    from gyxx_flow.core.layout import DataLayout
+    from gyxx_flow.core.records import RunJournal
+
+    layout = DataLayout(tmp_path)
+    context = RunContext.create("content.metrics.daily", "2026-10-18")
+    journal = RunJournal.create(layout, context, mode="dry_run")
+    journal.finalize("success")
+
+    record = RunIndex(tmp_path).index_journal(journal)
+
+    assert record.workflow_id == "content.metrics.daily"
+    assert record.status == "success"
+    # 运行索引文件仍然写出，审计失败不影响主链路
+    assert (tmp_path / "state" / "ops" / "run-index" / f"{record.run_id}.json").exists()
+
+
+def test_schema_sql_defines_both_audit_surfaces() -> None:
+    assert "ops_workflow_runs" in SCHEMA_SQL
+    assert "ops_workflow_run_steps" in SCHEMA_SQL
+    assert "ops_account_cookie_events" in SCHEMA_SQL
+    assert "ops_account_cookie_lifetimes" in SCHEMA_SQL
+    assert "ops_account_cookie_stats" in SCHEMA_SQL