update runtime

Dockerfile

@@ -13,12 +13,12 @@ RUN pip install --no-cache-dir -r requirements.txt
 COPY . .
 
 # Install agents 
-# RUN pip install -e pandas_expression_generator
+RUN pip install -e pandas_expression_generator
 
 # # Make the script executable
-# RUN chmod +x start.sh
+RUN chmod +x start.sh
 
 # Start both processes
-# CMD ["./start.sh"]
+CMD ["./start.sh"]
 
 

pandas_expression_generator/src/pandas_expression_generator/pandas_expression_generator_function.py

@@ -81,7 +81,7 @@ async def pandas_expression_generator_function(
 
 
     # Implement your function logic here
-    async def _response_fn(input_query: FunctionInput) -> str:
+    async def _response_fn(input_query: str) -> str:
 
         # Create LLM 
         llm_ = await builder.get_llm(config.llm, wrapper_type=LLMFrameworkEnum.LANGCHAIN)
@@ -96,12 +96,7 @@ async def pandas_expression_generator_function(
         {df_meta}
         
         Generate python expression to solve the query:
-        current_step: {input_query.current_step}
-
-        You can reformulate(extend or reduce) the current step based on the following information 
-        original_user_query: {input_query.original_user_query}
-        past_step: {input_query.thought}
-        next_step: {input_query.next_step}
+        {input_query}
 
         You must generate list of python expressions output format must as follows:
 
@@ -132,8 +127,20 @@ async def pandas_expression_generator_function(
             logger.info("[Following Should just be structured list of string]")
             logger.info(ai_message)
             structured_text_str = extract_bracket_content(ai_message.content)
-            output_message = str(expression_executor(structured_text_str, df))[:250]
+            output_message = str(expression_executor(structured_text_str, df))[:800]
             #
+            prompt_ = f'''You are part of big system. Your task is to respond the user in conversational way.
+        
+                This was the query of the end users:
+                {input_query}
+
+                The following the context data that you can use to respond to the user.
+                {output_message}
+
+                You may or may not use the context data to respond to the end user. Your output must be in markdown. Do not fence your output.
+                '''
+            ai_message: AIMessage = await llm_.ainvoke(prompt_)
+            return str(ai_message.content)
             # Loop break 
         except Exception as e: 
             logger.error(str(e))