Update app.py

app.py

@@ -416,8 +416,15 @@ class TextToSQLSystem:
 
     # in class TextToSQLSystem:
 
-    def _validate_and_fix_sql(self, sql: str, question: str) -> str:
+    # in class TextToSQLSystem:
 
+    def _validate_and_fix_sql(self, sql: str, question: str) -> str:
+        """
+        (V8 / 最終可靠版)
+        一個全面、多層次的 SQL 驗證與生成引擎。
+        本函數的觸發邏輯經過強化，不再依賴模型生成的特定幻覺內容，
+        而是更主動地基於使用者問題的意圖來啟用模板。
+        """
         if not sql or not self.schema:
             self._log("SQL 修正被跳過，因輸入為空或 schema 未載入。", "WARNING")
             return sql
@@ -435,17 +442,28 @@ class TextToSQLSystem:
         # 匹配 "top 5 買家 營收", "貢獻最高的10個客戶", "業績最好的申請方" 等
         top_n_pattern = r"(?:top|前|最高|最大|最好)\s*(\d+)?\s*(?:個|名)?\s*([^ ]+?)\s*(?:的)?(?:營收|業績|貢獻|金額|sales|revenue)"
         top_n_match = re.search(top_n_pattern, question, re.IGNORECASE)
+
+        # --- 意圖 2: 查詢特定實體的報告數量 (包含 Pass/Fail 等狀態) ---
+        ENTITY_TO_COLUMN_MAP = {
+            '買家': 'T1.BuyerName', 'buyer': 'T1.BuyerName', '客戶': 'T1.BuyerName',
+            '申請廠商': 'T1.ApplicantName', '申請方': 'T1.ApplicantName', 'applicant': 'T1.ApplicantName',
+            '付款廠商': 'T1.InvoiceToName', 'invoiceto': 'T1.InvoiceToName',
+            '代理商': 'T1.AgentName', 'agent': 'T1.AgentName',
+        }
+        entity_keywords_pattern = '|'.join(ENTITY_TO_COLUMN_MAP.keys())
+        dynamic_pattern = fr"({entity_keywords_pattern})\s*'\"?([a-zA-Z0-9\s&.-]+)'\"?"
+        entity_match = re.search(dynamic_pattern, question, re.IGNORECASE)
+
+        # --- 意圖 3: 計算平均處理時長 (TAT) ---
+        is_tat_query = any(k in q_lower for k in ['平均', 'average']) and any(k in q_lower for k in ['時間', '時長', '多久', '天', 'tat', 'turnaround'])
+
+        # --- 判斷邏輯: 依優先級進入對應的模板 ---
         if top_n_match:
-            limit = top_n_match.group(1) or '10' # 如果沒寫 N，預設為 10
+            limit = top_n_match.group(1) or '10'
             entity_keyword = top_n_match.group(2).lower()
             
-            # 建立實體關鍵詞到欄位的映射
-            ENTITY_MAP = {
-                '買家': 'T1.BuyerName', 'buyer': 'T1.BuyerName', '客戶': 'T1.BuyerName',
-                '申請廠商': 'T1.ApplicantName', '申請方': 'T1.ApplicantName', 'applicant': 'T1.ApplicantName',
-                '付款廠商': 'T1.InvoiceToName', '付款方': 'T1.InvoiceToName', 'invoiceto': 'T1.InvoiceToName',
-            }
-            column_name = next((v for k, v in ENTITY_MAP.items() if k in entity_keyword), None)
+            # 從實體映射中找到對應的欄位
+            column_name = next((v for k, v in ENTITY_TO_COLUMN_MAP.items() if k in entity_keyword), None)
 
             if column_name:
                 self._log(f"🔄 檢測到【Top {limit} {entity_keyword} 營收】查詢意圖，啟用模板。", "INFO")
@@ -459,46 +477,33 @@ ORDER BY total_revenue DESC
 LIMIT {limit};
 """
                 fixes_applied.append(f"模板覆寫: Top {limit} {entity_keyword} 營收查詢")
-
-        # --- 意圖 2: 查詢特定實體的報告數量 (包含 Pass/Fail 等狀態) ---
-        # 匹配 "買家 ABC 有幾份 Fail 的報告", "申請方 XYZ 的 Pass 報告數量"
-        elif '報告' in q_lower and ('數量' in q_lower or '幾份' in q_lower or 'count' in q_lower):
-            ENTITY_TO_COLUMN_MAP = {
-                '買家': 'T1.BuyerName', 'buyer': 'T1.BuyerName', '客戶': 'T1.BuyerName',
-                '申請廠商': 'T1.ApplicantName', '申請方': 'T1.ApplicantName', 'applicant': 'T1.ApplicantName',
-                '付款廠商': 'T1.InvoiceToName', 'invoiceto': 'T1.InvoiceToName',
-                '代理商': 'T1.AgentName', 'agent': 'T1.AgentName',
-            }
-            entity_keywords_pattern = '|'.join(ENTITY_TO_COLUMN_MAP.keys())
-            dynamic_pattern = fr"({entity_keywords_pattern})\s*'\"?([a-zA-Z0-9\s&.-]+)'\"?"
-            entity_match = re.search(dynamic_pattern, question, re.IGNORECASE)
-            
-            # 必須匹配到實體，且模型生成了錯誤 SQL (作為觸發器)
-            if entity_match and ('tsr53reportauthorization' in fixed_sql.lower() or 'testresult' in fixed_sql.lower()):
-                entity_type = entity_match.group(1).lower()
-                entity_name = entity_match.group(2).strip()
-                column_name = ENTITY_TO_COLUMN_MAP.get(entity_type)
-                
-                # 確定報告狀態 (Fail/Pass)
-                status = "'Fail'"
-                if 'pass' in q_lower or '通過' in q_lower:
-                    status = "'Pass'"
-                
-                self._log(f"🔄 檢測到查詢【{entity_type} '{entity_name}' 的 {status} 報告數】意圖，啟用模板。", "INFO")
-                fixed_sql = f"""
+        
+        elif entity_match and any(kw in q_lower for kw in ['份數', '數量', 'count', '幾份']):
+            entity_type = entity_match.group(1).lower()
+            entity_name = entity_match.group(2).strip()
+            column_name = ENTITY_TO_COLUMN_MAP.get(entity_type)
+            
+            year_match = re.search(r'(\d{4})\s*年?', question)
+            year = year_match.group(1) if year_match else '2024'
+            
+            status_condition = ""
+            if 'fail' in q_lower or '失敗' in q_lower:
+                status_condition = "AND T1.OverallRating = 'Fail'"
+            elif 'pass' in q_lower or '通過' in q_lower:
+                status_condition = "AND T1.OverallRating = 'Pass'"
+            
+            self._log(f"🔄 檢測到查詢【{entity_type} '{entity_name}' 在 {year} 年的報告數】意圖，啟用模板。", "INFO")
+            fixed_sql = f"""
 SELECT COUNT(T1.JobNo) AS report_count
 FROM TSR53SampleDescription AS T1
 JOIN JobTimeline AS T2 ON T1.JobNo = T2.JobNo
 WHERE {column_name} = '{entity_name}'
-  AND T1.OverallRating = {status}
-  AND strftime('%Y', T2.ReportAuthorization) = '2024';
+  AND strftime('%Y', T2.ReportAuthorization) = '{year}'
+  {status_condition};
 """
-                fixes_applied.append(f"模板覆寫: 查詢 {entity_type}='{entity_name}' 的 {status} 報告數")
+            fixes_applied.append(f"模板覆寫: 查詢 {entity_type}='{entity_name}' ({year}年) 的報告數")
 
-        # --- 意圖 3: 計算平均處理時長 (Turnaround Time, TAT) ---
-        # 匹配 "平均處理時間", "LabIn 到 LabOut 平均多久", "TAT"
-        elif any(k in q_lower for k in ['平均', 'average']) and any(k in q_lower for k in ['時間', '時長', '多久', '天', 'tat', 'turnaround']):
-            # 預設計算從 LabIn 到 ReportAuthorization 的總時長
+        elif is_tat_query:
             start_col, end_col = 'T2.LabIn', 'T2.ReportAuthorization'
             log_msg = "總流程平均時長 (天)"
             if 'labin' in q_lower and 'labout' in q_lower:
@@ -506,7 +511,6 @@ WHERE {column_name} = '{entity_name}'
                 log_msg = "實驗室平均處理時長 (天)"
             
             self._log(f"🔄 檢測到【{log_msg}】查詢意圖，啟用模板。", "INFO")
-            # SQLite 中，JULIANDAY 用於精確計算天數差
             fixed_sql = f"""
 SELECT AVG(JULIANDAY({end_col}) - JULIANDAY({start_col})) AS average_tat_days
 FROM JobTimeline AS T2
@@ -516,10 +520,11 @@ WHERE {start_col} IS NOT NULL AND {end_col} IS NOT NULL AND {end_col} > {start_c
 
         # 如果沒有任何模板被觸發，則進入常規修正流程
         if not fixes_applied:
+            self._log("未觸發任何模板，執行常規修正流程...", "DEBUG")
+            
             # ==============================================================================
             #  第二層：常規修正流程 (Fallback Corrections)
             # ==============================================================================
-            self._log("未觸發任何模板，執行常規修正流程...", "DEBUG")
             
             # 步驟 2.1: SQL 方言修正
             dialect_corrections = {
@@ -536,21 +541,26 @@ WHERE {start_col} IS NOT NULL AND {end_col} IS NOT NULL AND {end_col} > {start_c
                 # --- 常見幻覺表 ---
                 'TSR53ReportAuthorization': 'TSR53SampleDescription',
                 'TSR53TestResult': 'TSR53SampleDescription',
+                'JobInvoice': 'TSR53Invoice',
+                'JobInvoiceAuthorization': 'TSR53Invoice',
                 'Customer': 'TSR53SampleDescription', 'Customers': 'TSR53SampleDescription',
                 'Invoice': 'TSR53Invoice', 'Invoices': 'TSR53Invoice',
                 'Job': 'JobTimeline', 'Jobs': 'JobsInProgress',
                 
                 # --- 常見幻覺或錯誤欄位 ---
+                'AuthorizationDate': 'ReportAuthorization',
                 'ReportAuthorizationDate': 'ReportAuthorization',
+                'LegalAuthorization': 'OverallRating',
                 'TestResult': 'OverallRating', 'Rating': 'OverallRating',
-                'CustomerName': 'BuyerName', # 優先使用 BuyerName 作為通用客戶名
+                'CustomerName': 'BuyerName',
                 'InvoiceTo': 'InvoiceToName',
                 'Applicant': 'ApplicantName',
                 'Agent': 'AgentName',
                 'JobNumber': 'JobNo',
+                'ReportNo': 'JobNo',
                 'CreationDate': 'JobCreation', 'CreateDate': 'JobCreation',
                 'CompletedDate': 'ReportAuthorization',
-                'Amount': 'LocalAmount', # 優先使用 LocalAmount 作為金額
+                'Amount': 'LocalAmount',
                 'Price': 'LocalAmount',
                 'Lab': 'LabGroup'
             }
@@ -562,7 +572,6 @@ WHERE {start_col} IS NOT NULL AND {end_col} IS NOT NULL AND {end_col} > {start_c
 
             # 步驟 2.3: 基礎邏輯意圖修正
             if any(kw in q_lower for kw in ['幾份', '多少', 'how many', 'count', '數量']) and 'select ' in fixed_sql.lower() and 'count' not in fixed_sql.lower():
-                # 僅在不是分組查詢時替換，避免破壞 GROUP BY
                 if 'group by' not in fixed_sql.lower():
                     fixed_sql = re.sub(r'SELECT\s+.*?FROM', 'SELECT COUNT(*) FROM', fixed_sql, count=1, flags=re.IGNORECASE)
                     fixes_applied.append("修正邏輯: 補全 COUNT(*)")