Update app.py

app.py

@@ -596,11 +596,12 @@ SELECT
         if m:
             y1, y2 = m.group(1), m.group(2)
             return (
-                "SELECT strftime('%Y-%m', completed_time) AS month, "
-                f"SUM(CASE WHEN strftime('%Y', completed_time)='{y1}' THEN 1 ELSE 0 END) AS count_{y1}, "
-                f"SUM(CASE WHEN strftime('%Y', completed_time)='{y2}' THEN 1 ELSE 0 END) AS count_{y2} "
-                "FROM jobtimeline "
-                f"WHERE strftime('%Y', completed_time) IN ('{y1}','{y2}') "
+                "SELECT strftime('%Y-%m', jt.ReportAuthorization) AS month, "
+                f"COUNT(DISTINCT CASE WHEN strftime('%Y', jt.ReportAuthorization)='{y1}' THEN jt.JobNo END) AS count_{y1}, "
+                f"COUNT(DISTINCT CASE WHEN strftime('%Y', jt.ReportAuthorization)='{y2}' THEN jt.JobNo END) AS count_{y2} "
+                "FROM JobTimeline AS jt "
+                "WHERE jt.ReportAuthorization IS NOT NULL "
+                f"AND strftime('%Y', jt.ReportAuthorization) IN ('{y1}','{y2}') "
                 "GROUP BY month ORDER BY month;"
             )
 
@@ -609,33 +610,46 @@ SELECT
         if m:
             year = m.group(1)
             return (
-                "SELECT strftime('%Y-%m', completed_time) AS month, COUNT(*) AS count "
-                "FROM jobtimeline "
-                f"WHERE strftime('%Y', completed_time)='{year}' "
+                "SELECT strftime('%Y-%m', jt.ReportAuthorization) AS month, COUNT(DISTINCT jt.JobNo) AS count "
+                "FROM JobTimeline AS jt "
+                "WHERE jt.ReportAuthorization IS NOT NULL "
+                f"AND strftime('%Y', jt.ReportAuthorization)='{year}' "
                 "GROUP BY month ORDER BY month;"
             )
 
         # 評級分布（Pass/Fail）
         if ("評級" in q) or ("pass" in q_lower) or ("fail" in q_lower):
-            return "SELECT rating, COUNT(*) AS count FROM tsr53sampledescription GROUP BY rating;"
+            return (
+                "SELECT sd.OverallRating AS rating, COUNT(*) AS count "
+                "FROM TSR53SampleDescription AS sd "
+                "GROUP BY sd.OverallRating;"
+            )
 
         # 金額最高 Top N（預設 10）
         m = re.search(r"金額.*?(?:最高|前|top)\s*(\d+)?", q_lower)
         if m:
             n = m.group(1) or "10"
-            return f"SELECT * FROM tsr53invoice ORDER BY amount DESC LIMIT {n};"
+            return f"SELECT iv.* FROM TSR53Invoice AS iv ORDER BY iv.LocalAmount DESC LIMIT {n};"
 
         # 客戶工作單數量最多 Top N
         m = re.search(r"客戶.*?(?:最多|top|前)\s*(\d+)?", q_lower)
         if m:
             n = m.group(1) or "10"
-            return f"SELECT applicant, COUNT(*) AS count FROM tsr53sampledescription GROUP BY applicant ORDER BY count DESC LIMIT {n};"
+            return (
+                f"SELECT sd.ApplicantName AS applicant, COUNT(DISTINCT jt.JobNo) AS count "
+                "FROM JobTimeline AS jt "
+                "JOIN TSR53SampleDescription AS sd ON jt.JobNo = sd.JobNo "
+                "WHERE jt.ReportAuthorization IS NOT NULL "
+                "GROUP BY sd.ApplicantName ORDER BY count DESC "
+                f"LIMIT {n};"
+            )
 
         # 昨天完成多少
         if "昨天" in q:
             return (
-                "SELECT COUNT(*) AS count FROM jobtimeline "
-                "WHERE date(completed_time)=date('now','-1 day');"
+                "SELECT COUNT(DISTINCT jt.JobNo) AS count FROM JobTimeline AS jt "
+                "WHERE jt.ReportAuthorization IS NOT NULL "
+                "AND date(jt.ReportAuthorization)=date('now','-1 day');"
             )
 
         return None
@@ -651,6 +665,28 @@ SELECT
             self._log(f"最終整理 SQL 失敗: {e}", "ERROR")
             return (sql_text or ""), status
 
+    def _regenerate_sql_strict(self, question: str) -> Optional[str]:
+        """當模型輸出非 SQL 或無法解析時，使用嚴格限制的提示詞重生一次。"""
+        try:
+            rel = self._identify_relevant_tables(question)
+            schema_str = self._format_relevant_schema(rel)
+            strict_prompt = (
+                "You are a SQLite SQL generator.\n"
+                + "Given the schema below and the question, output ONE valid SQL query only.\n\n"
+                + "SCHEMA:\n" + schema_str + "\n\n"
+                + "QUESTION:\n" + (question or "").strip() + "\n\n"
+                + "Return only the final SQL query in a fenced code block (```sql ... ```). "
+                + "The SQL must start with SELECT and end with a semicolon. No explanation."
+            )
+            raw = self.huggingface_api_call(strict_prompt)
+            sql = parse_sql_from_response(raw)
+            if sql:
+                self._log("🔁 嚴格模式重生成功。")
+            return sql
+        except Exception as e:
+            self._log(f"嚴格模式重生失敗: {e}", "ERROR")
+            return None
+
     def _validate_and_fix_sql(self, question: str, raw_response: str) -> Tuple[Optional[str], str]:
         """
         (V29 / 穩健正則 + 智能計數) 多層次 SQL 生成：
@@ -741,7 +777,8 @@ SELECT
             entity_name, column_name = entity_match_data['name'], entity_match_data['column']
             match_op = '=' if column_name.endswith('ID') else 'LIKE'
             entity_val = f"'%{entity_name}%'" if match_op == 'LIKE' else f"'{entity_name}'"
-            sql['where'].append(f"{column_name} {match_op} {entity_val}")
+            collate = " COLLATE NOCASE" if match_op == 'LIKE' else ""
+            sql['where'].append(f"{column_name} {match_op} {entity_val}{collate}")
             sql['log_parts'].append(entity_match_data['type'] + ":" + entity_name)
             if intents.get('action') == 'list':
                 sql['select'].append("sd.BuyerName")
@@ -794,6 +831,9 @@ SELECT
                 salvage_sql = 'SELECT ' + resp
                 parsed_sql = parse_sql_from_response(salvage_sql) or salvage_sql
 
+        if not parsed_sql:
+            self._log("模型輸出非 SQL，啟用嚴格模式重生一次…")
+            parsed_sql = self._regenerate_sql_strict(q)
         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}"
@@ -842,45 +882,45 @@ SELECT
         return self._finalize_sql(fixed_sql, status)
 
     def _generate_fallback_sql(self, prompt: str) -> str:
-            """當模型不可用時的備用 SQL 生成"""
-            prompt_lower = (prompt or "").lower()
-            # 統計類：優先使用 JobTimeline.ReportAuthorization，避免不存在的 completed_time 欄位
-            if ("統計" in prompt) or ("數量" in prompt) or ("多少" in prompt) or ("count" in prompt_lower):
-                if ("月" in prompt) or ("per month" in prompt_lower) or ("monthly" in prompt_lower):
-                    return (
-                        "SELECT strftime('%Y-%m', jt.ReportAuthorization) AS month, "
-                        "COUNT(DISTINCT jt.JobNo) AS count "
-                        "FROM JobTimeline AS jt "
-                        "WHERE jt.ReportAuthorization IS NOT NULL "
-                        "GROUP BY month ORDER BY month;"
-                    )
-                elif ("客戶" in prompt) or ("buyer" in prompt_lower) or ("applicant" in prompt_lower):
-                    return (
-                        "SELECT sd.ApplicantName AS applicant, COUNT(DISTINCT jt.JobNo) AS count "
-                        "FROM JobTimeline AS jt "
-                        "JOIN TSR53SampleDescription AS sd ON jt.JobNo = sd.JobNo "
-                        "WHERE jt.ReportAuthorization IS NOT NULL "
-                        "GROUP BY sd.ApplicantName ORDER BY count DESC;"
-                    )
-                else:
-                    return (
-                        "SELECT COUNT(DISTINCT jt.JobNo) AS total_count "
-                        "FROM JobTimeline AS jt "
-                        "WHERE jt.ReportAuthorization IS NOT NULL;"
-                    )
-            # 金額彙總
-            if ("金額" in prompt) or ("總額" in prompt) or ("amount" in prompt_lower) or ("sum" in prompt_lower):
-                return "SELECT SUM(LocalAmount) AS total_amount FROM TSR53Invoice;"
-            # 評級分布
-            if ("評級" in prompt) or ("rating" in prompt_lower) or ("pass" in prompt_lower) or ("fail" in prompt_lower):
-                return "SELECT OverallRating AS rating, COUNT(*) AS count FROM TSR53SampleDescription GROUP BY OverallRating;"
-            # 通用後備：最近 10 筆報告
-            return (
-                "SELECT jt.JobNo, jt.ReportAuthorization "
-                "FROM JobTimeline AS jt "
-                "WHERE jt.ReportAuthorization IS NOT NULL "
-                "ORDER BY jt.ReportAuthorization DESC LIMIT 10;"
-            )
+        """當模型不可用時的備用 SQL 生成"""
+        prompt_lower = (prompt or "").lower()
+        # 統計類：優先使用 JobTimeline.ReportAuthorization，避免不存在的 completed_time 欄位
+        if ("統計" in prompt) or ("數量" in prompt) or ("多少" in prompt) or ("count" in prompt_lower):
+            if ("月" in prompt) or ("per month" in prompt_lower) or ("monthly" in prompt_lower):
+                return (
+                    "SELECT strftime('%Y-%m', jt.ReportAuthorization) AS month, "
+                    "COUNT(DISTINCT jt.JobNo) AS count "
+                    "FROM JobTimeline AS jt "
+                    "WHERE jt.ReportAuthorization IS NOT NULL "
+                    "GROUP BY month ORDER BY month;"
+                )
+            elif ("客戶" in prompt) or ("buyer" in prompt_lower) or ("applicant" in prompt_lower):
+                return (
+                    "SELECT sd.ApplicantName AS applicant, COUNT(DISTINCT jt.JobNo) AS count "
+                    "FROM JobTimeline AS jt "
+                    "JOIN TSR53SampleDescription AS sd ON jt.JobNo = sd.JobNo "
+                    "WHERE jt.ReportAuthorization IS NOT NULL "
+                    "GROUP BY sd.ApplicantName ORDER BY count DESC;"
+                )
+            else:
+                return (
+                    "SELECT COUNT(DISTINCT jt.JobNo) AS total_count "
+                    "FROM JobTimeline AS jt "
+                    "WHERE jt.ReportAuthorization IS NOT NULL;"
+                )
+        # 金額彙總
+        if ("金額" in prompt) or ("總額" in prompt) or ("amount" in prompt_lower) or ("sum" in prompt_lower):
+            return "SELECT SUM(LocalAmount) AS total_amount FROM TSR53Invoice;"
+        # 評級分布
+        if ("評級" in prompt) or ("rating" in prompt_lower) or ("pass" in prompt_lower) or ("fail" in prompt_lower):
+            return "SELECT OverallRating AS rating, COUNT(*) AS count FROM TSR53SampleDescription GROUP BY OverallRating;"
+        # 通用後備：最近 10 筆報告
+        return (
+            "SELECT jt.JobNo, jt.ReportAuthorization "
+            "FROM JobTimeline AS jt "
+            "WHERE jt.ReportAuthorization IS NOT NULL "
+            "ORDER BY jt.ReportAuthorization DESC LIMIT 10;"
+        )
 
     def process_question(self, question: str) -> Tuple[str, str]:
         """處理使用者問題"""