feat(governance): automate identity and comparison coverage

run_scheduler.py

@@ -3,7 +3,7 @@
 run_scheduler.py — momo-scheduler 容器入口點
 
 排程任務清單（對齊 app.py init_scheduler + scheduler.py 全任務）：
-  每 30 分鐘：auto_import、whitepage_check
+  每 30 分鐘：revenue_automation_cycle（auto_import → comparison backfill）、whitepage_check
   每  1 小時：momo、edm、festival
   每  4 小時：competitor_price_feeder、external_offer_sync、icaim_analysis
   每  6 小時：quality_rescore、action_plan_hygiene
@@ -22,7 +22,7 @@ import logging
 import os
 import threading
 import time
-from typing import Optional, Tuple
+from typing import Callable, Optional, Tuple
 
 import schedule
 
@@ -60,6 +60,7 @@ logger = logging.getLogger(__name__)
 
 _AI_CALLS_ERROR_SPIKE_LAST_PUSH_TS = 0.0
 _OLLAMA_111_USAGE_LAST_PUSH_TS = 0.0
+_REVENUE_AUTOMATION_LOCK = threading.Lock()
 
 
 def _scheduler_stats_path() -> str:
@@ -122,6 +123,96 @@ def run_pchome_growth_momo_backfill_catchup_task():
         return {"status": "failed", "error": str(error)}
 
 
+def _launch_revenue_automation_background(
+    job_name: str,
+    target: Callable[[], object],
+) -> dict[str, str]:
+    """Launch one revenue job without blocking or overlapping the scheduler loop."""
+    if not _REVENUE_AUTOMATION_LOCK.acquire(blocking=False):
+        logger.info("[RevenueAutomation] skipped overlapping job=%s", job_name)
+        return {"status": "skipped", "reason": "revenue_automation_already_running"}
+
+    def _runner():
+        try:
+            target()
+        except Exception as error:
+            logger.error(
+                "[RevenueAutomation] background job failed job=%s error=%s",
+                job_name,
+                error,
+                exc_info=True,
+            )
+            _notify_scheduler_failure(
+                job_name,
+                error,
+                source="Scheduler.RevenueAutomation",
+                event_type="revenue_automation_background_failure",
+                title="PChome 收益自動化背景任務失敗",
+            )
+        finally:
+            _REVENUE_AUTOMATION_LOCK.release()
+
+    thread = threading.Thread(
+        target=_runner,
+        daemon=True,
+        name=f"revenue-automation-{job_name}",
+    )
+    try:
+        thread.start()
+    except Exception:
+        _REVENUE_AUTOMATION_LOCK.release()
+        raise
+    return {"status": "queued", "job": job_name}
+
+
+def run_revenue_automation_cycle_task() -> dict[str, object]:
+    """Refresh sales first, then catch up the bounded same-item mapping lane."""
+    logger.info("[RevenueAutomation] cycle started: sales acquisition -> comparison backfill")
+    run_auto_import_task()
+    backfill = run_pchome_growth_momo_backfill_catchup_task()
+    logger.info(
+        "[RevenueAutomation] cycle completed backfill_status=%s",
+        backfill.get("status"),
+    )
+    return {"status": "completed", "backfill": backfill}
+
+
+def run_revenue_automation_cycle_background_task() -> dict[str, str]:
+    return _launch_revenue_automation_background(
+        "revenue-cycle",
+        run_revenue_automation_cycle_task,
+    )
+
+
+def run_pchome_growth_momo_backfill_background_task() -> dict[str, str]:
+    return _launch_revenue_automation_background(
+        "pchome-growth-backfill",
+        run_pchome_growth_momo_backfill_task,
+    )
+
+
+def start_revenue_automation_watchdog() -> threading.Thread:
+    """Keep revenue refresh alive even when a legacy schedule job blocks run_pending."""
+    try:
+        interval_seconds = int(os.getenv("REVENUE_AUTOMATION_WATCHDOG_SECONDS", "1800"))
+    except ValueError:
+        interval_seconds = 1800
+    interval_seconds = max(60, interval_seconds)
+
+    def _watchdog():
+        while True:
+            run_revenue_automation_cycle_background_task()
+            time.sleep(interval_seconds)
+
+    thread = threading.Thread(
+        target=_watchdog,
+        daemon=True,
+        name="revenue-automation-watchdog",
+    )
+    thread.start()
+    return thread
+
+
 def _env_flag(name: str, default: bool = False) -> bool:
     raw = os.getenv(name)
     if raw is None:
@@ -185,8 +276,8 @@ def _notify_scheduler_failure(
 
 
 def _register_schedules():
-    schedule.every(30).minutes.do(run_auto_import_task)
-    logger.info("📅 每 30 分鐘：auto_import")
+    schedule.every(30).minutes.do(run_revenue_automation_cycle_background_task)
+    logger.info("📅 每 30 分鐘：revenue_automation_cycle（業績匯入 → 比價回填，非阻塞）")
 
     schedule.every(30).minutes.do(run_whitepage_check)
     logger.info("📅 每 30 分鐘：whitepage_check")
@@ -374,12 +465,9 @@ def _register_schedules():
     schedule.every().day.at("10:30").do(run_pchome_match_backfill_task)
     logger.info("📅 每日 10:30：pchome_match_backfill")
 
-    schedule.every().day.at("10:45").do(run_pchome_growth_momo_backfill_task)
+    schedule.every().day.at("10:45").do(run_pchome_growth_momo_backfill_background_task)
     logger.info("📅 每日 10:45：pchome_growth_momo_backfill（高業績商品補 MOMO 對應）")
 
-    schedule.every(30).minutes.do(run_pchome_growth_momo_backfill_catchup_task)
-    logger.info("📅 每 30 分鐘：pchome_growth_momo_backfill_catchup（錯過每日任務時自動補跑）")
-
     # Operation Ollama-First v5.0 — Phase 1 收尾：每日 23:55 LLM Token 日報
     schedule.every().day.at("23:55").do(run_daily_token_report_task)
     logger.info("📅 每日 23:55：daily_token_report")
@@ -1368,6 +1456,13 @@ if __name__ == "__main__":
     _register_schedules()
     logger.info("✅ 全部排程任務已註冊")
 
+    _revenue_watchdog_thread = start_revenue_automation_watchdog()
+    logger.info(
+        "[RevenueAutomation] watchdog=%s alive=%s",
+        _revenue_watchdog_thread.name,
+        _revenue_watchdog_thread.is_alive(),
+    )
+
     try:
         from services.openclaw_bot.embedding_worker_runtime import start_embedding_worker