Update agentic.py

agentic.py

@@ -16,6 +16,9 @@ from langchain_google_genai import ChatGoogleGenerativeAI
 from langfuse.callback import CallbackHandler
 
 import base64
+import json
+import time
+from pprint import pprint
 
 # import boto3
 
@@ -47,7 +50,7 @@ def get_assistant_model():
 
     if llm_provider == "mistral":
         assistant_model = ChatMistralAI(
-            model="mistral-small-latest",#"ministral-8b-latest",#
+            model="mistral-large-2411",#"mistral-small-latest",#"ministral-8b-latest",#
             temperature=0,
             max_retries=2,
             api_key=os.getenv("MISTRAL_API_KEY")
@@ -104,12 +107,6 @@ def get_video_handler_model():
 
     return video_handler_model
 
-# reviewer_model = ChatMistralAI(
-#     model="mistral-small-latest",
-#     temperature=0,
-#     max_retries=2,
-#     api_key=os.getenv("MISTRAL_API_KEY"),
-# )
 
 def download_youtube_content(url: str, output_path: Optional[str] = None) -> None:
     """
@@ -262,7 +259,7 @@ def vision_model_call(state: State) -> str:
         print(error_msg)
         return ""
     
-def video_handler_mode_call(state: State, video_url: str) -> str:
+def video_handler_model_call(state: State, video_url: str) -> str:
     """
         Video handler model that can analyze videos and answer questions about them.
 
@@ -322,28 +319,6 @@ def video_handler_mode_call(state: State, video_url: str) -> str:
     )
         
     return response.content + "\n\n"
-
-# def router(state: State)-> Literal["OK", "retry", "failed"]:
-#     """Determine the next step based on FINAL ANSWER"""
-
-#     print(f"MESSAGES : {state['messages'][-1].content}")
-#     print(f"STATE : {state['system_prompt']}")
-#     if True:
-#         return "OK"
-#     else:
-#         if state.status == "ERROR":
-#             if state.error_count >= 3:
-#                 return "failed"
-#         return "retry"
-
-# def reviewer_validation(state: State) -> Literal["OK", "NOK"]:
-#     print(f"MESSAGES : {state['messages'][-1].content}")
-#     print(f"STATE : {state}")
-#     return "OK"
-#     if True:
-#         return "OK"
-#     else:
-#         return "NOK"
     
 # Tools
 tools = [
@@ -351,11 +326,10 @@ tools = [
     # search_webpage,
     wikipedia_search,
     vision_model_call,
-    video_handler_mode_call
+    video_handler_model_call
 ]
 
 assistant_with_tools = assistant_model.bind_tools(tools, parallel_tool_calls=False)
-# reviewer_with_tools = reviewer_model.bind_tools(tools, parallel_tool_calls=False)
 
 
 def assistant(state: State)-> str:
@@ -382,22 +356,18 @@ def assistant(state: State)-> str:
         Returns:
             {vision_model_call.__annotations__['return']}
 
-    video_handler_mode_call:
-        {video_handler_mode_call.__doc__}
+    video_handler_model_call:
+        {video_handler_model_call.__doc__}
         Args:
-            {video_handler_mode_call.__annotations__}
+            {video_handler_model_call.__annotations__}
         Returns:
-            {video_handler_mode_call.__annotations__['return']}
+            {video_handler_model_call.__annotations__['return']}
 """
     
     with open("./prompt.txt", "r") as prompt_file:
         system_prompt = prompt_file.read()
 
-    file_prompt = f"If the question is about an image or a picture, and if you need to analyze a file, an image and so on... you must check the path {state['input_file']}"
-
-    video_handler_prompt = "If the question is about a video and if you need to analyze a video, you must use the video_handler_mode_call tool."
-
-    sys_msg = SystemMessage(content=system_prompt+file_prompt+video_handler_prompt+textual_description_of_tool)
+    sys_msg = SystemMessage(content=system_prompt+textual_description_of_tool)
 
     response = [assistant_with_tools.invoke([sys_msg] + state["messages"])]
 
@@ -408,29 +378,6 @@ def assistant(state: State)-> str:
         "answer": state.get("answer", "")
     }
 
-# def reviewer(state: State)-> str:
-
-#     with open("./prompt.txt", "r") as prompt_file:
-#         system_prompt = prompt_file.read()
-
-#     sys_msg = SystemMessage(content=f"""
-#     You are a powerful AI assistant reviewer.
-#     You must review the answer of the previous assistant.
-#     If you need you can correct the answer.
-#     You need to make sure the answer is formatted correctly.
-#     The response should not be sent if FINAL ANSWER: is not registered and does not respect the constraints of the initial prompt.
-                            
-#     Here is the prompt of the previous assistant :
-#     {system_prompt}
-
-#     Here is the question: {state['question']}
-#     """)
-
-#     response = [reviewer_with_tools.invoke([sys_msg] + state["messages"])]
-
-#     return {
-#         "messages": response,
-#     }
 
 def build_graph():
     builder = StateGraph(State)
@@ -448,25 +395,6 @@ def build_graph():
         tools_condition,
     )
     builder.add_edge("tools", "assistant")
-    # builder.add_edge("assistant", "reviewer")
-
-    # builder.add_conditional_edges(
-    #     "assistant",
-    #     router,
-    #     {
-    #         "OK": "reviewer",
-    #         "retry": "assistant",
-    #         "failed": END
-    #     }
-    # )
-    # builder.add_conditional_edges(
-    #     "reviewer",
-    #     reviewer_validation,
-    #     {
-    #         "OK": END,
-    #         "NOK": "assistant"
-    #     }
-    # )
 
     return builder.compile()
 
@@ -476,15 +404,11 @@ if __name__ == "__main__":
 
     file_name = ""
 
-    # agent_graph.get_graph(xray=True).draw_mermaid_png(output_file_path="./graph.png", draw_method=MermaidDrawMethod.PYPPETEER)
-    question = "Where were the Vietnamese specimens described by Kuznetzov in Nedoshivina's 2010 paper eventually deposited? Just give me the city name without abbreviations."
-    # question = "Given this table defining * on the set S = {a, b, c, d, e}\n\n|*|a|b|c|d|e|\n|---|---|---|---|---|---|\n|a|a|b|c|b|d|\n|b|b|c|a|e|c|\n|c|c|a|b|b|a|\n|d|b|e|b|e|d|\n|e|d|b|a|d|c|\n\nprovide the subset of S involved in any possible counter-examples that prove * is not commutative. Provide your answer as a comma separated list of the elements in the set in alphabetical order."
-    # question = ".rewsna eht sa \"tfel\" drow eht fo etisoppo eht etirw ,ecnetnes siht dnatsrednu uoy fI"
-    # question = "On June 6, 2023, an article by Carolyn Collins Petersen was published in Universe Today. This article mentions a team that produced a paper about their observations, linked at the bottom of the article. Find this paper. Under what NASA award number was the work performed by R. G. Arendt supported by?"
-    # question = "In the video https://www.youtube.com/watch?v=L1vXCYZAYYM, what is the highest number of bird species to be on camera simultaneously?"
-    # question = "Review the chess position provided in the image. It is black's turn. Provide the correct next move for black which guarantees a win. Please provide your response in algebraic notation."
-    # question = "Can you describe the image ?"
-    # file_name = "images/cca530fc-4052-43b2-b130-b30968d8aa44.png"
+    question = "In the video https://www.youtube.com/watch?v=L1vXCYZAYYM, what is the highest number of bird species to be on camera simultaneously?"
+    # file_name = question_json.get("file_name", "")
+
+    print(f"QUESTION : {question}")
+    print(f"FILE: {file_name}")
 
     messages = [HumanMessage(content=f"Can you answer this question please ? {question}")]
 
@@ -505,6 +429,76 @@ if __name__ == "__main__":
     except:
         regex_result = re.search(r"\s*(?P<answer>.*)$", messages['messages'][-1].content)
         answer = regex_result.group("answer")
+    
+    print(answer)
+
+    # with open("./questions.json", "r") as file:
+    #     questions_json = json.loads(file.read())
+
+    # for question_json in questions_json:
+    #     question = question_json.get("question", "")
+    #     file_name = question_json.get("file_name", "")
+
+    #     print(f"QUESTION : {question}")
+    #     print(f"FILE: {file_name}")
+
+    #     messages = [HumanMessage(content=f"Can you answer this question please ? {question}")]
+
+    #     messages = agent_graph.invoke(
+    #         input={"messages": messages, "question": question, "input_file": file_name}, 
+    #         config={
+    #             "recursion_limit": 10,
+    #             "callbacks": [langfuse_handler]
+    #         }
+    #     )
+
+    #     for m in messages['messages']:
+    #         m.pretty_print()
+        
+    #     try:
+    #         regex_result = re.search(r"FINAL ANSWER:\s*(?P<answer>.*)$", messages['messages'][-1].content)
+    #         answer = regex_result.group("answer")
+    #     except:
+    #         regex_result = re.search(r"\s*(?P<answer>.*)$", messages['messages'][-1].content)
+    #         answer = regex_result.group("answer")
+
+    #     print(answer)
+    #     time.sleep(10)
+
+    # def test():
+    #     # System message
+    #     textual_description_of_tool=f"""
+    #     web_search:
+    #         {web_search.description}
+    #         Args:
+    #             {web_search.args_schema.__doc__}
+    #             {web_search.args_schema.__annotations__}
+    #         Returns:
+    #            response_format: {web_search.response_format}
+        
+    #     wikipedia_search:
+    #         {wikipedia_search.description}
+    #         Args:
+    #             {wikipedia_search.args_schema.__doc__}
+    #             {wikipedia_search.args_schema.__annotations__}
+    #         Returns:
+    #             response_format: {wikipedia_search.response_format}
+
+    #     vision_model_call:
+    #         {vision_model_call.__doc__}
+    #         Args:
+    #             {vision_model_call.__annotations__}
+    #         Returns:
+    #             {vision_model_call.__annotations__['return']}
+
+    #     video_handler_model_call:
+    #         {video_handler_model_call.__doc__}
+    #         Args:
+    #             {video_handler_model_call.__annotations__}
+    #         Returns:
+    #             {video_handler_model_call.__annotations__['return']}
+    # """
+    
+    #     pprint(textual_description_of_tool)
 
-    print(f"ANSWER : {answer}")
-    # from pprint import pprint
+    # pprint(web_search.__annotations__)