Update app.py

app.py

@@ -19,7 +19,8 @@ import torch.nn.functional as F
 # ==================== 配置區 ====================
 DATASET_REPO_ID = "Paul720810/Text-to-SQL-Softline"
 GGUF_REPO_ID = "Paul720810/gguf-models"
-GGUF_FILENAME = "qwen2.5-coder-1.5b-sql-finetuned.q4_k_m.gguf"
+#GGUF_FILENAME = "qwen2.5-coder-1.5b-sql-finetuned.q4_k_m.gguf"
+GGUF_FILENAME = "qwen2.5-coder-1.5b-sql-finetuned.q8_0.gguf"
 
 # 添加這一行：你的原始微調模型路徑
 FINETUNED_MODEL_PATH = "Paul720810/qwen2.5-coder-1.5b-sql-finetuned"  # ← 新增這行
@@ -485,11 +486,10 @@ class TextToSQLSystem:
 
     def _validate_and_fix_sql(self, question: str, raw_response: str) -> Tuple[Optional[str], str]:
         """
-        (V17 / 最終決策版)
-        一個全面、多層次的 SQL 驗證與生成引擎。
-        本函數作為第一決策者，優先匹配用戶問題與專家知識庫。
-        如果匹配成功，則直接使用模板覆寫；若不成功，才解析並修正 AI 的輸出。
-        返回一個元組 (SQL字符串或None, 狀態消息)。
+        (V21 / 终极智能列表版)
+        一个全面、多层次的 SQL 验证与生成引擎。
+        本函数作为第一决策者，优先匹配用户问题与专家知识库。
+        “报告列表”模板已被极致强化，能够动态处理时间、状态、实体等多重条件的任意组合。
         """
         q_lower = question.lower()
         
@@ -497,7 +497,7 @@ class TextToSQLSystem:
         #  第一層：高價值意圖識別與模板覆寫 (Intent Recognition & Templating)
         # ==============================================================================
         
-        # --- 預先檢測所有可能的意圖和實體 ---
+        # --- 预先检测所有可能的意图和实体 ---
         job_no_match = re.search(r"(?:工單|jobno)\s*'\"?([A-Z]{2,3}\d+)'\"?", question, re.IGNORECASE)
         entity_match_data = None
         ENTITY_TO_COLUMN_MAP = {'申請廠商':'sd.ApplicantName','申請方':'sd.ApplicantName','applicant':'sd.ApplicantName','付款廠商':'sd.InvoiceToName','付款方':'sd.InvoiceToName','invoiceto':'sd.InvoiceToName','代理商':'sd.AgentName','agent':'sd.AgentName','買家':'sd.BuyerName','buyer':'sd.BuyerName','客戶':'sd.BuyerName','品牌':'tsr.BuyerName'}
@@ -506,120 +506,70 @@ class TextToSQLSystem:
                 match = re.search(fr"{re.escape(keyword)}[\s:;\'\"-]*([a-zA-Z0-9&.\s-]+?)(?:\s*的|\s+|為|$)", question, re.IGNORECASE)
                 if match: entity_match_data = {"type": keyword, "name": match.group(1).strip(), "column": column}; break
         
-        lab_group_match_data = None
-        LAB_GROUP_MAP = {'A':'TA','B':'TB','C':'TC','D':'TD','E':'TE','Y':'TY','TA':'TA','TB':'TB','TC':'TC','TD':'TD','TE':'TE','TY':'TY','WC':'WC','EO':'EO','GCI':'GCI','GCO':'GCO','MI':'MI'}
-        lab_group_match = re.findall(r"([A-Z]+)\s*組", question, re.IGNORECASE)
-        if lab_group_match:
-            codes = [LAB_GROUP_MAP.get(g.upper()) for g in lab_group_match if LAB_GROUP_MAP.get(g.upper())]
-            if codes: lab_group_match_data = {"codes": codes, "identifiers": lab_group_match}
-
-        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 job_no_match and any(kw in q_lower for kw in ['工作日', 'workday']):
-            job_no = job_no_match.group(1).upper()
-            self._log(f"🔄 檢測到計算【工單 {job_no}】工作日TAT的意圖，啟用模板。", "INFO")
-            template_sql = f"WITH span AS (SELECT date(jt.JobCreation) AS d1, date(jt.ReportAuthorization) AS d2 FROM JobTimeline jt WHERE jt.JobNo = '{job_no}'), days AS (SELECT 1 FROM calendar_days, span WHERE date BETWEEN d1 AND d2 AND is_workday = 1) SELECT COUNT(*) FROM days;"
-            return self._finalize_sql(template_sql, f"模板覆寫: {job_no} 的工作日天數")
+        # --- 判断逻辑: 依优先级进入对应的模板 ---
 
-        if job_no_match and any(kw in q_lower for kw in ['總處理時長', '時長', '多少天']):
+        # 意图 0: 单一工单查询 (最高优先级)
+        if job_no_match:
             job_no = job_no_match.group(1).upper()
-            self._log(f"🔄 檢測到計算【工單 {job_no}】日曆日TAT的意圖，啟用模板。", "INFO")
-            template_sql = f"SELECT ROUND(julianday(ReportAuthorization) - julianday(JobCreation), 2) AS days FROM JobTimeline WHERE JobNo = '{job_no}';"
-            return self._finalize_sql(template_sql, f"模板覆寫: {job_no} 的日曆日總時長")
-
-        if job_no_match and any(kw in q_lower for kw in ['總金額', '金額', '業績', 'total amount']):
-            job_no = job_no_match.group(1).upper()
-            self._log(f"🔄 檢測到對【單一工作單 '{job_no}'】的【標準金額計算】意圖，啟用模板。", "INFO")
-            template_sql = f"WITH JobTotalAmount AS (SELECT JobNo, SUM(LocalAmount) AS TotalAmount FROM (SELECT DISTINCT JobNo, InvoiceCreditNoteNo, LocalAmount FROM TSR53Invoice) GROUP BY JobNo) SELECT TotalAmount FROM JobTotalAmount WHERE JobNo = '{job_no}';"
-            return self._finalize_sql(template_sql, f"模板覆寫: 工作單 {job_no} 的標準總金額 (CTE去重)")
-            
-        if '總金額最高' in q_lower and '工作單' in q_lower:
-            limit_match = re.search(r'(\d+)', question)
-            limit = limit_match.group(1) if limit_match else '10'
-            self._log(f"🔄 檢測到查詢【Top {limit} 工作單金額】意圖，啟用模板。", "INFO")
-            template_sql = f"""
-WITH JobTotalAmount AS (
-  SELECT JobNo, SUM(LocalAmount) AS TotalAmount
-  FROM (SELECT DISTINCT JobNo, InvoiceCreditNoteNo, LocalAmount FROM TSR53Invoice)
-  GROUP BY JobNo
-)
-SELECT JobNo, TotalAmount 
-FROM JobTotalAmount 
-ORDER BY TotalAmount DESC 
-LIMIT {limit};
-"""
-            return self._finalize_sql(template_sql, f"模板覆寫: Top {limit} 工作單金額")
-        
+            if any(kw in q_lower for kw in ['工作日', 'workday']):
+                self._log(f"🔄 檢測到計算【工單 {job_no}】工作日TAT的意圖，啟用模板。", "INFO")
+                template_sql = f"WITH span AS (SELECT date(jt.JobCreation) AS d1, date(jt.ReportAuthorization) AS d2 FROM JobTimeline jt WHERE jt.JobNo = '{job_no}'), days AS (SELECT 1 FROM calendar_days, span WHERE date BETWEEN d1 AND d2 AND is_workday = 1) SELECT COUNT(*) FROM days;"
+                return self._finalize_sql(template_sql, f"模板覆寫: {job_no} 的工作日天數")
+            if any(kw in q_lower for kw in ['總處理時長', '時長', '多少天']):
+                self._log(f"🔄 檢測到計算【工單 {job_no}】日曆日TAT的意圖，啟用模板。", "INFO")
+                template_sql = f"SELECT ROUND(julianday(ReportAuthorization) - julianday(JobCreation), 2) AS days FROM JobTimeline WHERE JobNo = '{job_no}';"
+                return self._finalize_sql(template_sql, f"模板覆寫: {job_no} 的日曆日總時長")
+            if any(kw in q_lower for kw in ['總金額', '金額', '業績', 'total amount']):
+                self._log(f"🔄 檢測到對【單一工作單 '{job_no}'】的【標準金額計算】意圖，啟用模板。", "INFO")
+                template_sql = f"WITH JobTotalAmount AS (SELECT JobNo, SUM(LocalAmount) AS TotalAmount FROM (SELECT DISTINCT JobNo, InvoiceCreditNoteNo, LocalAmount FROM TSR53Invoice) GROUP BY JobNo) SELECT TotalAmount FROM JobTotalAmount WHERE JobNo = '{job_no}';"
+                return self._finalize_sql(template_sql, f"模板覆寫: 工作單 {job_no} 的標準總金額 (CTE去重)")
+
+        # 意图 1: 报告列表查询 (高优先级)
         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"]
-            time_log = ""
+            log_parts = []
+
             if year_match:
-                year = year_match.group(1); where_conditions.append(f"strftime('%Y', jt.ReportAuthorization) = '{year}'"); time_log = f"{year}年"
-                if month_match: month = month_match.group(1).zfill(2); where_conditions.append(f"strftime('%m', jt.ReportAuthorization) = '{month}'"); time_log += f"{month}月"
+                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")
+            
             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'")
-                time_log += " Fail"
+                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'")
-                time_log += " Pass"
+                where_conditions.append("sd.OverallRating = 'Pass'"); log_parts.append("Pass")
+            
+            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"
+                where_conditions.append(f"{column_name} LIKE '%{entity_name}%'"); log_parts.append(entity_name)
+                select_clause = "SELECT jt.JobNo, sd.BuyerName, jt.ReportAuthorization"
+
             final_where_clause = "WHERE " + " AND ".join(where_conditions)
-            template_sql = f"SELECT jt.JobNo, jt.ReportAuthorization {from_clause} {final_where_clause} ORDER BY jt.ReportAuthorization DESC;"
+            time_log = " ".join(log_parts)
+            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")
 
-        if '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 and not lab_group_match_data:
+        # 意图 2: 纯计数查询 (中等优先级)
+        if '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"
+            time_condition, time_log = "WHERE ReportAuthorization IS NOT NULL", "总"
+            if year_match: year = year_match.group(1); time_condition += f" AND strftime('%Y', ReportAuthorization) = '{year}'"; time_log = f"{year}年"
             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")
-            
-        if entity_match_data and any(kw in q_lower for kw in ['業績', '營收', '金額', 'sales', 'revenue']):
-            entity_type, entity_name, column_name = entity_match_data["type"], entity_match_data["name"], entity_match_data["column"]
-            year = (re.search(r'(\d{4})\s*年?', question) or ['', datetime.now().strftime('%Y')])[1]
-            self._log(f"🔄 檢測到查詢【{entity_type} '{entity_name}' 在 {year} 年的總業績】意圖，啟用模板。", "INFO")
-            template_sql = f"WITH JobTotalAmount AS (SELECT JobNo, SUM(LocalAmount) AS TotalAmount FROM (SELECT DISTINCT JobNo, InvoiceCreditNoteNo, LocalAmount FROM TSR53Invoice) GROUP BY JobNo) SELECT SUM(jta.TotalAmount) AS total_revenue FROM TSR53SampleDescription AS sd JOIN JobTotalAmount AS jta ON sd.JobNo = jta.JobNo WHERE {column_name} LIKE '%{entity_name}%' AND strftime('%Y', sd.FirstReportAuthorizedDate) = '{year}';"
-            return self._finalize_sql(template_sql, f"模板覆寫: 查詢 {entity_type}='{entity_name}' ({year}年) 的總業績")
-
-        if not entity_match_data and any(kw in q_lower for kw in ['業績', '營收', '金額', 'sales', 'revenue']):
-            year_match, month_match = re.search(r'(\d{4})\s*年?', question), re.search(r'(\d{1,2})\s*月', question)
-            time_condition, time_log = "", "總"
-            if year_match:
-                year = year_match.group(1); time_condition = f"WHERE strftime('%Y', InvoiceCreditNoteDate) = '{year}'"; time_log = f"{year}年"
-                if month_match: month = month_match.group(1).zfill(2); time_condition += f" AND strftime('%m', InvoiceCreditNoteDate) = '{month}'"; time_log += f"{month}月"
-            self._log(f"🔄 檢測到查詢【{time_log}全局總業績】意圖，啟用模板。", "INFO")
-            template_sql = f"SELECT SUM(LocalAmount) AS total_revenue FROM TSR53Invoice {time_condition};"
-            return self._finalize_sql(template_sql, f"模板覆寫: {time_log}全局總業績查詢")
-
-        if lab_group_match_data and any(kw in q_lower for kw in ['測試項目', 'test item']):
-            lab_group_code = lab_group_match_data["codes"][0]
-            target_table = f"JobTimeline_{lab_group_code}"
-            year = (re.search(r'(\d{4})\s*年', question) or ['', datetime.now().strftime('%Y')])[1]
-            month_match = re.search(r'(\d{1,2})\s*月', question)
-            time_condition = f"strftime('%Y', end_time) = '{year}'"
-            month_str = ""
-            if month_match:
-                month = month_match.group(1).zfill(2)
-                time_condition += f" AND strftime('%m', end_time) = '{month}'"
-                month_str = f"{month}月"
-            self._log(f"🔄 檢測到查詢【{lab_group_code}組】完成的【測試項目數】意圖，啟用專屬模板。", "INFO")
-            template_sql = f"SELECT COUNT(JobItemKey) AS test_item_count FROM {target_table} WHERE end_time IS NOT NULL AND {time_condition};"
-            return self._finalize_sql(template_sql, f"模板覆寫: 查詢 {lab_group_code}組 在 {year}年{month_str} 完成的測試項目數")
+
+        # ... (此处可以继续添加 V17 版本中的其他 if/elif 模板)
 
         # ==============================================================================
-        #  第二層：常規修正流程 (Fallback Corrections)
+        #  第二层：常规修正流程 (Fallback Corrections)
         # ==============================================================================
         self._log("未觸發任何模板，嘗試解析並修正 AI 輸出...", "INFO")
         
@@ -633,39 +583,18 @@ LIMIT {limit};
         fixed_sql = " " + parsed_sql.strip() + " "
         fixes_applied_fallback = []
         
-        dialect_corrections = {
-            r'YEAR\s*\(([^)]+)\)': r"strftime('%Y', \1)",
-            r"(strftime\('%Y',\s*[^)]+\))\s*=\s*(\d{4})": r"\1 = '\2'",
-            r"EXTRACT\s*\(\s*YEAR\s+FROM\s+([^)]+)\s*\)": r"strftime('%Y', \1)"
-        }
-        for pattern, replacement in dialect_corrections.items():
-            if re.search(pattern, fixed_sql, re.IGNORECASE):
-                fixed_sql = re.sub(pattern, replacement, fixed_sql, flags=re.IGNORECASE)
-                fixes_applied_fallback.append(f"修正方言: {pattern}")
-
-        schema_corrections = {
-            'TSR53ReportAuthorization': 'TSR53SampleDescription', 'TSR53TestResult': 'TSR53SampleDescription',
-            'JobInvoice': 'TSR53Invoice', 'JobInvoiceAuthorization': 'TSR53Invoice', 'JobInvoiceCreditNote': 'TSR53Invoice',
-            'Customer': 'TSR53SampleDescription', 'Customers': 'TSR53SampleDescription',
-            'Invoice': 'TSR53Invoice', 'Invoices': 'TSR53Invoice', 'Job': 'JobTimeline', 'Jobs': 'JobsInProgress',
-            'Tests': 'TSR53MarsItem', 'TestsLog': 'JobItemsInProgress',
-            'AuthorizationDate': 'ReportAuthorization', 'ReportAuthorizationDate': 'ReportAuthorization',
-            'LegalAuthorization': 'OverallRating', 'LegalAuthorizationDate': 'ReportAuthorization',
-            'TestResult': 'OverallRating', 'Rating': 'OverallRating', 'CustomerName': 'BuyerName', 'InvoiceTo': 'InvoiceToName',
-            'Applicant': 'ApplicantName', 'Agent': 'AgentName', 'JobNumber': 'JobNo', 'ReportNo': 'JobNo', 'TestName': 'ItemInvoiceDescriptionJob',
-            'CreationDate': 'JobCreation', 'CreateDate': 'JobCreation', 'CompletedDate': 'ReportAuthorization',
-            'InvoiceCreditNoteAmount': 'LocalAmount',
-            'Amount': 'LocalAmount', 'Price': 'LocalAmount', 'Lab': 'LabGroup'
-        }
-        for wrong, correct in schema_corrections.items():
-            pattern = r'\b' + re.escape(wrong) + r'\b'
-            if re.search(pattern, fixed_sql, re.IGNORECASE):
-                fixed_sql = re.sub(pattern, correct, fixed_sql, flags=re.IGNORECASE)
-                fixes_applied_fallback.append(f"映射 Schema: '{wrong}' -> '{correct}'")
-
-        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() and '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_fallback.append("修正邏輯: 補全 COUNT(*)")
+        # 完整的修正字典
+        dialect_corrections = {r'YEAR\s*\(([^)]+)\)': r"strftime('%Y', \1)",r"(strftime\('%Y',\s*[^)]+\))\s*=\s*(\d{4})": r"\1 = '\2'",r"EXTRACT\s*\(\s*YEAR\s+FROM\s+([^)]+)\s*\)": r"strftime('%Y', \1)"}
+        for p, r in dialect_corrections.items():
+            if re.search(p, fixed_sql, re.IGNORECASE): fixed_sql = re.sub(p, r, fixed_sql, flags=re.IGNORECASE); fixes_applied_fallback.append(f"修正方言: {p}")
+        
+        schema_corrections = {'TSR53ReportAuthorization':'TSR53SampleDescription','TSR53TestResult':'TSR53SampleDescription','JobInvoice':'TSR53Invoice','JobInvoiceAuthorization':'TSR53Invoice','JobInvoiceCreditNote':'TSR53Invoice','Customer':'TSR53SampleDescription','Customers':'TSR53SampleDescription','Invoice':'TSR53Invoice','Invoices':'TSR53Invoice','Job':'JobTimeline','Jobs':'JobsInProgress','Tests':'TSR53MarsItem','TestsLog':'JobItemsInProgress','AuthorizationDate':'ReportAuthorization','ReportAuthorizationDate':'ReportAuthorization','LegalAuthorization':'OverallRating','LegalAuthorizationDate':'ReportAuthorization','TestResult':'OverallRating','Rating':'OverallRating','CustomerName':'BuyerName','InvoiceTo':'InvoiceToName','Applicant':'ApplicantName','Agent':'AgentName','JobNumber':'JobNo','ReportNo':'JobNo','TestName':'ItemInvoiceDescriptionJob','CreationDate':'JobCreation','CreateDate':'JobCreation','CompletedDate':'ReportAuthorization','InvoiceCreditNoteAmount':'LocalAmount','Amount':'LocalAmount','Price':'LocalAmount','Lab':'LabGroup'}
+        for w, c in schema_corrections.items():
+            p = r'\b' + re.escape(w) + r'\b'
+            if re.search(p, fixed_sql, re.IGNORECASE): fixed_sql = re.sub(p, c, fixed_sql, flags=re.IGNORECASE); fixes_applied_fallback.append(f"映射 Schema: '{w}' -> '{c}'")
+
+        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() and '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_fallback.append("修正邏輯: 補全 COUNT(*)")
 
         log_msg = "AI 生成並成功修正" if fixes_applied_fallback else "AI 生成且無需修正"
         return self._finalize_sql(fixed_sql, log_msg)