fix(ChatExcel): ChatExcel OutParse Bug Fix (#588)

1.ChatExcel OutParse Bug Fix

pilot/out_parser/base.py

@@ -129,44 +129,55 @@ def __illegal_json_ends(self, s):
         return temp_json
 
     def __extract_json(self, s):
-        temp_json = self.__json_interception(s, True)
-        if not temp_json:
-            temp_json = self.__json_interception(s)
         try:
+            # Get the dual-mode analysis first and get the maximum result
+            temp_json_simple = self.__json_interception(s)
+            temp_json_array = self.__json_interception(s, True)
+            if len(temp_json_simple) > len(temp_json_array):
+                temp_json = temp_json_simple
+            else:
+                temp_json = temp_json_array
+
+            if not temp_json:
+                temp_json = self.__json_interception(s)
+
             temp_json = self.__illegal_json_ends(temp_json)
             return temp_json
         except Exception as e:
-            raise ValueError("Failed to find a valid json response！" + temp_json)
+            raise ValueError("Failed to find a valid json in LLM response！" + temp_json)
 
     def __json_interception(self, s, is_json_array: bool = False):
-        if is_json_array:
-            i = s.find("[")
-            if i < 0:
-                return None
-            count = 1
-            for j, c in enumerate(s[i + 1 :], start=i + 1):
-                if c == "]":
-                    count -= 1
-                elif c == "[":
-                    count += 1
-                if count == 0:
-                    break
-            assert count == 0
-            return s[i : j + 1]
-        else:
-            i = s.find("{")
-            if i < 0:
-                return None
-            count = 1
-            for j, c in enumerate(s[i + 1 :], start=i + 1):
-                if c == "}":
-                    count -= 1
-                elif c == "{":
-                    count += 1
-                if count == 0:
-                    break
-            assert count == 0
-            return s[i : j + 1]
+        try:
+            if is_json_array:
+                i = s.find("[")
+                if i < 0:
+                    return ""
+                count = 1
+                for j, c in enumerate(s[i + 1 :], start=i + 1):
+                    if c == "]":
+                        count -= 1
+                    elif c == "[":
+                        count += 1
+                    if count == 0:
+                        break
+                assert count == 0
+                return s[i : j + 1]
+            else:
+                i = s.find("{")
+                if i < 0:
+                    return ""
+                count = 1
+                for j, c in enumerate(s[i + 1 :], start=i + 1):
+                    if c == "}":
+                        count -= 1
+                    elif c == "{":
+                        count += 1
+                    if count == 0:
+                        break
+                assert count == 0
+                return s[i : j + 1]
+        except Exception as e:
+            return ""
 
     def parse_prompt_response(self, model_out_text) -> T:
         """

pilot/scene/chat_data/chat_excel/excel_analyze/out_parser.py

@@ -27,15 +27,18 @@ def __init__(self, sep: str, is_stream_out: bool):
     def parse_prompt_response(self, model_out_text):
         clean_str = super().parse_prompt_response(model_out_text)
         print("clean prompt response:", clean_str)
-        response = json.loads(clean_str)
-        for key in sorted(response):
-            if key.strip() == "sql":
-                sql = response[key]
-            if key.strip() == "thoughts":
-                thoughts = response[key]
-            if key.strip() == "display":
-                display = response[key]
-        return ExcelAnalyzeResponse(sql, thoughts, display)
+        try:
+            response = json.loads(clean_str)
+            for key in sorted(response):
+                if key.strip() == "sql":
+                    sql = response[key]
+                if key.strip() == "thoughts":
+                    thoughts = response[key]
+                if key.strip() == "display":
+                    display = response[key]
+            return ExcelAnalyzeResponse(sql, thoughts, display)
+        except Exception as e:
+            raise ValueError(f"LLM Response Can't Parser! \n{ model_out_text}")
 
     def parse_view_response(self, speak, data) -> str:
         ### tool out data to table view

pilot/scene/chat_data/chat_excel/excel_learning/out_parser.py

@@ -41,7 +41,7 @@ def parse_prompt_response(self, model_out_text):
             return model_out_text
 
     def parse_view_response(self, speak, data) -> str:
-        if data:
+        if data and not isinstance(data, str):
             ### tool out data to table view
             html_title = f"### **Data Summary**\n{data.desciption} "
             html_colunms = f"### **Data Structure**\n"