fix(stability): 補強 scheduler 例外處理與 vendor 模板路徑

ADR-017 Phase 3f-3：移除 scheduler 裸 except，P1 任務失敗改走 EventRouter notify_failure 同步入口；清理 docker-compose vendor 死 mount；修正 vendor_bp template_folder 指向專案 web/templates。
2026-04-29 21:39:25 +08:00
parent 13fa165ee2
commit 9528d6c23e
4 changed files with 104 additions and 43 deletions
--- a/docker-compose.yml
+++ b/docker-compose.yml
@@ -79,14 +79,6 @@ services:
      - ./auto_import_index.html:/app/auto_import_index.html:ro
      - ./settings.html:/app/settings.html:ro
      - ./system_settings.html:/app/system_settings.html:ro
-      # 廠商缺貨系統
-      - ./vendor_routes.py:/app/vendor_routes.py:ro
-      - ./vendor_stockout_index.html:/app/vendor_stockout_index.html:ro
-      - ./vendor_stockout_import.html:/app/vendor_stockout_import.html:ro
-      - ./vendor_stockout_list.html:/app/vendor_stockout_list.html:ro
-      - ./vendor_stockout_send_email.html:/app/vendor_stockout_send_email.html:ro
-      - ./vendor_stockout_history.html:/app/vendor_stockout_history.html:ro
-      - ./vendor_management.html:/app/vendor_management.html:ro
      # 其他根目錄路由
      - ./auto_import_routes.py:/app/auto_import_routes.py:ro
      - ./crawler_management_routes.py:/app/crawler_management_routes.py:ro
--- a/routes/vendor_routes.py
+++ b/routes/vendor_routes.py
@@ -23,9 +23,8 @@ from services.logger_manager import SystemLogger
 sys_log = SystemLogger("VendorRoutes").get_logger()

 # 建立 Blueprint
-# V-Fix (2026-01-23): 指定正確的 template_folder，解決 TemplateNotFound 錯誤
-import os
-_base_dir = os.path.dirname(os.path.abspath(__file__))
+# V-Fix: 指定正確的 template_folder，解決 TemplateNotFound 錯誤
+_base_dir = os.path.dirname(os.path.dirname(os.path.abspath(__file__)))
 _template_folder = os.path.join(_base_dir, 'web', 'templates')
 vendor_bp = Blueprint('vendor', __name__, url_prefix='/vendor-stockout', template_folder=_template_folder)

--- a/scheduler.py
+++ b/scheduler.py
@@ -240,7 +240,8 @@ def run_momo_task():
                        logging.warning(f"[Crawler] [MOMO]   ⚠️ 頁面載入超時 (15s)，強制停止載入並嘗試解析內容...")
                        try:
                            driver.execute_script("window.stop();")
-                        except: pass
+                        except Exception as e:
+                            logging.exception(f"[Crawler] [MOMO] window.stop() 失敗但繼續 | Category: {cat_name} | Error: {e}")

                    # V-Opt: eager 模式下，get 會很快返回，這裡稍微等待一下 DOM 穩定
                    time.sleep(1)
@@ -251,7 +252,8 @@ def run_momo_task():
                        for j in range(1, 3): # V-Opt: 再減少一次滾動 (4 -> 3)，通常前兩屏即包含大部分熱銷商品
                            driver.execute_script(f"window.scrollTo(0, {j * 1500});") # 加大滾動距離
                            time.sleep(0.5)
-                    except Exception: pass
+                    except Exception as e:
+                        logging.exception(f"[Crawler] [MOMO] 滾動頁面失敗但繼續 | Category: {cat_name} | Error: {e}")
                    
                    # === V9.95: 改為容器優先的爬取策略，提高穩定性 ===
                    # 1. 找出所有可能的商品容器
@@ -467,18 +469,16 @@ def run_momo_task():
        _save_stats('momo_task', stats)
        # ADR-012 Phase 2: 走 EventRouter（Hermes L1 翻譯 + 三層式訊息）
        try:
-            from services.event_router import dispatch as _dispatch
-            _dispatch({
-                "source": "Scheduler.MOMOCrawler",
-                "event_type": "crawler_timeout",
-                "severity": "alert",
-                "title": "MOMO 爬蟲任務中斷",
-                "status": "任務失敗",
-                "impact": "P1 - 熱銷商品監控中斷",
-                "summary": str(e)[:200],
-                "trace": _tb.format_exc(),
-                "payload": {"task_name": "run_momo_task"},
-            })
+            from services.event_router import notify_failure
+            notify_failure(
+                task_name="run_momo_task",
+                error=e,
+                source="Scheduler.MOMOCrawler",
+                event_type="crawler_timeout",
+                priority="P1",
+                title="MOMO 爬蟲任務中斷",
+                trace=_tb.format_exc(),
+            )
        except Exception as _router_e:
            logging.error(f"[Crawler] [MOMO] event_router 失敗: {_router_e}")
    finally:
@@ -579,12 +579,14 @@ def run_edm_task(lpn_code="O1K5FBOqsvN"):
                        brand_text = ""
                        try:
                            brand_text = item.find_element(By.CSS_SELECTOR, ".brand").text.strip()
-                        except: pass
+                        except Exception as e:
+                            logging.exception(f"[Crawler] [EDM] 解析品牌失敗但繼續 | Block: {i+1} | Error: {e}")
                        
                        name_text = ""
                        try:
                            name_text = item.find_element(By.CSS_SELECTOR, ".brand2").text.strip()
-                        except: pass
+                        except Exception as e:
+                            logging.exception(f"[Crawler] [EDM] 解析品名失敗但繼續 | Block: {i+1} | Error: {e}")
                        
                        name = f"{brand_text} {name_text}".strip()
                        
@@ -650,7 +652,8 @@ def run_edm_task(lpn_code="O1K5FBOqsvN"):
                            disc_el = item.find_element(By.CSS_SELECTOR, ".discount span")
                            if disc_el:
                                discount_text = disc_el.text.strip() + "折"
-                        except: pass
+                        except Exception as e:
+                            logging.exception(f"[Crawler] [EDM] 解析折扣失敗但繼續 | i_code: {i_code} | Error: {e}")

                        # V9.66: 解析倒數組數
                        # HTML: <div class="last">倒數：<span id="gdsStock_1">92</span><small>組</small></div>
@@ -887,7 +890,8 @@ def run_festival_task(lpn_code="O7ylWfihYUM"):
                logging.warning("[Crawler] [Festival] ⚠️ 頁面載入超時 (120s)，嘗試停止載入並繼續解析...")
                try:
                    driver.execute_script("window.stop();")
-                except: pass
+                except Exception as e:
+                    logging.exception(f"[Crawler] [Festival] window.stop() 失敗但繼續 | Error: {e}")

            # V-Fix: 增加初始等待時間，確保頁面上的 Vue.js 框架有足夠時間初始化並掛載懶加載事件
            time.sleep(10)
@@ -1219,7 +1223,8 @@ def run_promo_event_task(lpn_code, page_type, activity_name):
                logging.warning(f"[Crawler] [{page_type.upper()}] ⚠️ 頁面載入超時 (120s)，嘗試停止載入並繼續解析...")
                try:
                    driver.execute_script("window.stop();")
-                except: pass
+                except Exception as e:
+                    logging.exception(f"[Crawler] [{page_type.upper()}] window.stop() 失敗但繼續 | Error: {e}")

            time.sleep(10)
            logging.info(f"[Crawler] [{page_type.upper()}] 📄 頁面標題: {driver.title}")
@@ -1876,18 +1881,16 @@ def run_auto_import_task():
        # ADR-012 Phase 2: 改走 EventRouter（Hermes L1 翻譯 + 降級鏈）
        # LINE 通道保留（event_router 只處理 Telegram）
        try:
-            from services.event_router import dispatch as _dispatch
-            _dispatch({
-                "source": "Scheduler.AutoImport",
-                "event_type": "db_connection_error" if "translate host" in str(e).lower() or "operational" in str(e).lower() else "import_failure",
-                "severity": "alert",
-                "title": "當日業績自動匯入異常",
-                "status": "匯入失敗",
-                "impact": "P1 - 當日業績未更新",
-                "summary": str(e)[:200],
-                "trace": _tb.format_exc(),
-                "payload": {"task_name": "run_auto_import_task"},
-            })
+            from services.event_router import notify_failure
+            notify_failure(
+                task_name="run_auto_import_task",
+                error=e,
+                source="Scheduler.AutoImport",
+                event_type="db_connection_error" if "translate host" in str(e).lower() or "operational" in str(e).lower() else "import_failure",
+                priority="P1",
+                title="當日業績自動匯入異常",
+                trace=_tb.format_exc(),
+            )
        except Exception as _router_e:
            logging.error(f"[Scheduler] [AutoImport] event_router 失敗: {_router_e}")

@@ -2331,4 +2334,4 @@ def run_monthly_report_task():
 if __name__ == "__main__":
    # 此檔案現在由 app.py 導入並由其主執行緒管理排程。
    # 若需獨立測試，可在此處臨時加入調用程式碼。
-    logging.info("[Scheduler] [Main] 此為排程任務定義檔，請由主程式 app.py 執行。")
+    logging.info("[Scheduler] [Main] 此為排程任務定義檔，請由主程式 app.py 執行。")
--- a/services/event_router.py
+++ b/services/event_router.py
@@ -5,6 +5,8 @@
 #
 import asyncio
 import logging
+import threading
+import traceback
 from typing import Any, Dict, Optional

 from services.ai_orchestrator import AIOrchestrator
@@ -103,6 +105,71 @@ async def dispatch(event: Dict[str, Any], admin_chat_ids: Optional[list] = None)
        }


+def _run_coroutine_in_thread(coro) -> Dict[str, Any]:
+    result = {}
+
+    def runner():
+        try:
+            result["value"] = asyncio.run(coro)
+        except Exception as e:
+            result["value"] = {
+                "tier": "unknown",
+                "sent": 0,
+                "errors": [str(e)],
+                "latency_ms": 0,
+                "payload": None,
+            }
+
+    thread = threading.Thread(target=runner, daemon=True)
+    thread.start()
+    thread.join(timeout=15)
+    if thread.is_alive():
+        return {
+            "tier": "unknown",
+            "sent": 0,
+            "errors": ["dispatch_sync timed out"],
+            "latency_ms": 15000,
+            "payload": None,
+        }
+    return result["value"]
+
+
+def dispatch_sync(event: Dict[str, Any], admin_chat_ids: Optional[list] = None) -> Dict[str, Any]:
+    """同步環境使用的 EventRouter 入口。"""
+    try:
+        asyncio.get_running_loop()
+    except RuntimeError:
+        return asyncio.run(dispatch(event, admin_chat_ids=admin_chat_ids))
+    return _run_coroutine_in_thread(dispatch(event, admin_chat_ids=admin_chat_ids))
+
+
+def notify_failure(
+    task_name: str,
+    error: Exception,
+    *,
+    source: Optional[str] = None,
+    event_type: str = "scheduler_task_failure",
+    priority: str = "P2",
+    title: Optional[str] = None,
+    trace: Optional[str] = None,
+    payload: Optional[Dict[str, Any]] = None,
+) -> Dict[str, Any]:
+    """排程/背景任務失敗的同步通知 helper。"""
+    severity = "alert" if priority in {"P1", "P2"} else "warning"
+    event = {
+        "source": source or f"Scheduler.{task_name}",
+        "event_type": event_type,
+        "severity": severity,
+        "title": title or f"{task_name} 任務異常",
+        "status": "任務失敗",
+        "impact": f"{priority} - 背景任務需要檢查",
+        "summary": str(error)[:200],
+        "trace": trace or "".join(traceback.format_exception(type(error), error, error.__traceback__)),
+        "payload": {"task_name": task_name, **(payload or {})},
+    }
+    return dispatch_sync(event)
+
+
 def _classify(event: Dict[str, Any]) -> str:
    sev = event.get("severity", "info")
    has_trace = bool(event.get("trace"))