Update app.py

app.py

@@ -494,25 +494,19 @@ class TextToSQLSystem:
 
         return formatted.strip()
 
-    # in class TextToSQLSystem:
+# 在 class TextToSQLSystem 內
 
     def _validate_and_fix_sql(self, question: str, raw_response: str) -> Tuple[Optional[str], str]:
         """
-        (V23 / 统一实体识别版)
-        一個全面、多層次的 SQL 驗證與生成引擎。
-        引入了全新的、统一的实体识别引擎，能够准确解析 "买家 Gap", "c0761n", 
-        "买家ID c0761n" 等多种复杂的实体提问模式。
+        (V26 / 統一實體識別 + 模組化意圖引擎版)
+        在模組化意圖偵測前，先運行一個帶優先級的統一實體識別引擎。
         """
         q_lower = question.lower()
-        
+    
         # ==============================================================================
-        #  第一層：高價值意圖識別與模板覆寫 (Intent Recognition & Templating)
+        #  第零層：統一實體識別引擎 (Unified Entity Recognition Engine)
         # ==============================================================================
-        
-        # --- **全新的统一实体识别引擎** ---
         entity_match_data = None
-        
-        # 定义多种识别模式，【优先级从高到低】
         entity_patterns = [
             # 模式1: 匹配 "类型 + ID" (e.g., "买家ID C0761N") - 最高优先级
             {'pattern': r"(買家|buyer)\s*(?:id|代號|代碼)\s*'\"?\b([A-Z]\d{4}[A-Z])\b'\"?", 'column': 'sd.BuyerID', 'type': '買家ID'},
@@ -533,67 +527,103 @@ class TextToSQLSystem:
         for p in entity_patterns:
             match = re.search(p['pattern'], question, re.IGNORECASE)
             if match:
+                # 根據正則表達式捕獲組的數量來決定取哪個組
                 entity_value = match.group(2) if len(match.groups()) > 1 else match.group(1)
                 entity_match_data = {
                     "type": p['type'], 
-                    "name": entity_value.strip().upper(),
+                    "name": entity_value.strip().upper(), # 標準化：移除前後空格並轉大寫
                     "column": p['column']
                 }
+                # 找到第一個匹配項（最高優先級）後就立刻停止
                 break
-
-        # --- 预先检测其他意图 ---
-        job_no_match = re.search(r"(?:工單|jobno)\s*'\"?([A-Z]{2,3}\d+)'\"?", question, re.IGNORECASE)
+                
+        # ==============================================================================
+        #  第一層：模組化意圖偵測 (Modular Intent Detection)
+        # ==============================================================================
+        
+        # --- 1. 初始化 SQL 組件 ---
+        intents = {}
+        sql_components = {
+            'select': [],
+            'from': "FROM JobTimeline AS jt",
+            'joins': [],
+            'where': ["jt.ReportAuthorization IS NOT NULL"],
+            'group_by': [],
+            'order_by': [],
+            'log_parts': []
+        }
+    
+        # --- 2. 運行一系列獨立的意圖偵測器 ---
         
-        # --- 判断逻辑: 依优先级进入对应的模板 ---
-        if any(kw in q_lower for kw in ['報告號碼', '報告清單', '列出報告', 'report number', 'list of reports']):
-            year_match = re.search(r'(\d{4})\s*年?', question)
-            month_match = re.search(r'(\d{1,2})\s*月', question)
-            from_clause = "FROM JobTimeline AS jt"
-            select_clause = "SELECT jt.JobNo, jt.ReportAuthorization"
-            where_conditions = ["jt.ReportAuthorization IS NOT NULL"]
-            log_parts = []
-
-            if year_match: year = year_match.group(1); where_conditions.append(f"strftime('%Y', jt.ReportAuthorization) = '{year}'"); log_parts.append(f"{year}年")
-            if month_match: month = month_match.group(1).zfill(2); where_conditions.append(f"strftime('%m', jt.ReportAuthorization) = '{month}'"); log_parts.append(f"{month}月")
+        # 偵測器 2.1: 核心動作意圖 (計數 vs. 列表)
+        if any(kw in q_lower for kw in ['幾份', '多少', '數量', '總數', 'how many', 'count']):
+            intents['action'] = 'count'
+            sql_components['select'].append("COUNT(DISTINCT jt.JobNo) AS report_count")
+            sql_components['log_parts'].append("報告總數")
+        elif any(kw in q_lower for kw in ['報告號碼', '報告清單', '列出報告', 'report number', 'list of reports']):
+            intents['action'] = 'list'
+            sql_components['select'].append("jt.JobNo, jt.ReportAuthorization")
+            sql_components['order_by'].append("jt.ReportAuthorization DESC")
+            sql_components['log_parts'].append("報告列表")
+    
+        # 偵測器 2.2: 時間意圖
+        year_match = re.search(r'(\d{4})\s*年?', question)
+        month_match = re.search(r'(\d{1,2})\s*月', question)
+        if year_match:
+            year = year_match.group(1)
+            sql_components['where'].append(f"strftime('%Y', jt.ReportAuthorization) = '{year}'")
+            sql_components['log_parts'].append(f"{year}年")
+        if month_match:
+            month = month_match.group(1).zfill(2)
+            sql_components['where'].append(f"strftime('%m', jt.ReportAuthorization) = '{month}'")
+            sql_components['log_parts'].append(f"{month}月")
+    
+        # 偵測器 2.3: 實體意圖 (現在使用預處理的結果)
+        if entity_match_data:
+            # 確保 JOIN 只添加一次
+            if "TSR53SampleDescription" not in " ".join(sql_components['joins']):
+                 sql_components['joins'].append("JOIN TSR53SampleDescription AS sd ON jt.JobNo = sd.JobNo")
             
-            if 'fail' in q_lower or '失敗' in q_lower:
-                if "JOIN TSR53SampleDescription" not in from_clause: from_clause = "FROM JobTimeline AS jt JOIN TSR53SampleDescription AS sd ON jt.JobNo = sd.JobNo"
-                where_conditions.append("sd.OverallRating = 'Fail'"); log_parts.append("Fail")
-            elif 'pass' in q_lower or '通過' in q_lower:
-                if "JOIN TSR53SampleDescription" not in from_clause: from_clause = "FROM JobTimeline AS jt JOIN TSR53SampleDescription AS sd ON jt.JobNo = sd.JobNo"
-                where_conditions.append("sd.OverallRating = 'Pass'"); log_parts.append("Pass")
+            entity_name, column_name = entity_match_data["name"], entity_match_data["column"]
+            match_operator = "=" if column_name.endswith("ID") else "LIKE"
+            # 對於 LIKE 查詢，在值兩邊加上 %
+            entity_value = f"'%{entity_name}%'" if match_operator == "LIKE" else f"'{entity_name}'"
+            sql_components['where'].append(f"{column_name} {match_operator} {entity_value}")
+            sql_components['log_parts'].append(entity_match_data["type"] + ":" + entity_name)
             
-            if entity_match_data:
-                entity_name, column_name = entity_match_data["name"], entity_match_data["column"]
-                if "JOIN TSR53SampleDescription" not in from_clause: from_clause = "FROM JobTimeline AS jt JOIN TSR53SampleDescription AS sd ON jt.JobNo = sd.JobNo"
-                match_operator = "=" if column_name.endswith("ID") else "LIKE"
-                entity_value = f"'{entity_name}'" if match_operator == "=" else f"'%{entity_name}%'"
-                where_conditions.append(f"{column_name} {match_operator} {entity_value}")
-                log_parts.append(entity_name)
-                select_clause = "SELECT jt.JobNo, sd.BuyerName, jt.ReportAuthorization"
-
-            final_where_clause = "WHERE " + " AND ".join(where_conditions)
-            time_log = " ".join(log_parts) if log_parts else "全部"
-            self._log(f"🔄 檢測到查詢【{time_log} 報告列表】意圖，啟用智能模板。", "INFO")
-            template_sql = f"{select_clause} {from_clause} {final_where_clause} ORDER BY jt.ReportAuthorization DESC;"
-            return self._finalize_sql(template_sql, f"模板覆寫: {time_log} 報告列表查詢")
-
-        # ... (此处可以继续添加 V17 版本中的其他所有 if/elif 模板)
-        elif '報告' in q_lower and any(kw in q_lower for kw in ['幾份', '多少', '數量', '總數']) and not entity_match_data:
-            year_match = re.search(r'(\d{4})\s*年?', question)
-            time_condition, time_log = "", "總"
-            if year_match:
-                year = year_match.group(1)
-                time_condition = f"WHERE ReportAuthorization IS NOT NULL AND strftime('%Y', ReportAuthorization) = '{year}'"
-                time_log = f"{year}年"
-            else:
-                time_condition = "WHERE ReportAuthorization IS NOT NULL"
-            self._log(f"🔄 檢測到查詢【{time_log}全局報告總數】意圖，啟用模板。", "INFO")
-            template_sql = f"SELECT COUNT(DISTINCT JobNo) AS report_count FROM JobTimeline {time_condition};"
-            return self._finalize_sql(template_sql, f"模板覆寫: {time_log}全局報告總數查詢")
-
+            # 如果是列表查詢，我們可以顯示更多實體相關的資訊
+            if intents.get('action') == 'list':
+                sql_components['select'].append("sd.BuyerName")
+    
+        # 偵測器 2.4: 評級意圖 (Pass/Fail)
+        if 'fail' in q_lower or '失敗' in q_lower:
+            if "TSR53SampleDescription" not in " ".join(sql_components['joins']):
+                 sql_components['joins'].append("JOIN TSR53SampleDescription AS sd ON jt.JobNo = sd.JobNo")
+            sql_components['where'].append("sd.OverallRating = 'Fail'")
+            sql_components['log_parts'].append("Fail")
+        elif 'pass' in q_lower or '通過' in q_lower:
+            if "TSR53SampleDescription" not in " ".join(sql_components['joins']):
+                 sql_components['joins'].append("JOIN TSR53SampleDescription AS sd ON jt.JobNo = sd.JobNo")
+            sql_components['where'].append("sd.OverallRating = 'Pass'")
+            sql_components['log_parts'].append("Pass")
+            
+        # --- 3. ���斷是否觸發了模板，並組合 SQL ---
+        
+        if 'action' in intents:
+            final_select = "SELECT " + ", ".join(sorted(list(set(sql_components['select']))))
+            from_clause = sql_components['from']
+            joins_clause = " ".join(sql_components['joins'])
+            where_clause = "WHERE " + " AND ".join(sql_components['where'])
+            orderby_clause = "ORDER BY " + ", ".join(sql_components['order_by']) if sql_components['order_by'] else ""
+            
+            template_sql = f"{final_select} {from_clause} {joins_clause} {where_clause} {orderby_clause};"
+            
+            query_log = " ".join(sql_components['log_parts'])
+            self._log(f"🔄 偵測到組合意圖【{query_log}】，啟用動態模板。", "INFO")
+            return self._finalize_sql(template_sql, f"模板覆寫: {query_log} 查詢")
+    
         # ==============================================================================
-        #  第二层：常规修正流程 (Fallback Corrections)
+        #  第二层：AI 生成修正流程 (Fallback)
         # ==============================================================================
         self._log("未觸發任何模板，嘗試解析並修正 AI 輸出...", "INFO")
         
@@ -601,7 +631,7 @@ class TextToSQLSystem:
         if not parsed_sql:
             self._log(f"❌ 未能從模型回應中解析出任何 SQL。原始回應: {raw_response}", "ERROR")
             return None, f"無法解析SQL。原始回應:\n{raw_response}"
-
+    
         self._log(f"📊 解析出的原始 SQL: {parsed_sql}", "DEBUG")
         
         fixed_sql = " " + parsed_sql.strip() + " "