Update agentic.py

agentic.py

@@ -51,7 +51,7 @@ class State(TypedDict):
     video_node_result: AnyMessage
     audio_node_result: AnyMessage
     code_node_result: AnyMessage
-    next: str
+    next_node: str
 
 ########################
 
@@ -306,7 +306,7 @@ You do not handle code
 3. Report your thought process in detail, explaining your reasoning step-by-step.
 
 Here is the question {state['question']}
-Now provide your response immediately without any preamble.
+Now provide your response immediately without any preamble in text but not in markdown.
     """
 
     state["thinking_node_result"] = state.get("thinking_node_result", "")
@@ -352,7 +352,7 @@ You do not handle audio
 Here is the question : {state['question']}
 Here is the code : {code}
 
-Now provide your response immediately without any preamble.
+Now provide your response immediately without any preamble in text but not in markdown.
     """
 
     sys_msg = SystemMessage(content=prompt)
@@ -410,7 +410,7 @@ wikipedia_search:
         {wikipedia_search.response_format}
 
 Here is the question {state['question']}
-Now provide your response immediately without any preamble.
+Now provide your response immediately without any preamble in text but not in markdown.
     """
 
     state["web_wiki_search_node_result"] = state.get("web_wiki_search_node_result", "")
@@ -451,7 +451,7 @@ You do not handle code
 4. Report your thought process in detail, explaining your reasoning step-by-step.
 
 Here is the question {state['question']}
-Now provide your response immediately without any preamble.
+Now provide your response immediately without any preamble in text but not in markdown.
     """
 
     image_base64 = ""
@@ -540,7 +540,7 @@ You do not handle code.
 6. Do not make up information that is not visible or inferable from the video.
 
 Here is the question {state['question']}
-Now provide your response immediately without any preamble.
+Now provide your response immediately without any preamble in text but not in markdown.
     """
 
     if re.search(r'youtube\.com', state["question"]):
@@ -623,7 +623,7 @@ You do not handle code.
 6. Do not make up information that is not audible or inferable from the audio.
 
 Here is the question {state['question']}
-Now provide your response immediately without any preamble.
+Now provide your response immediately without any preamble in text but not in markdown.
     """
 
     downloaded_audio = state["downloaded_file"]
@@ -691,16 +691,18 @@ FINAL ANSWER: [YOUR FINAL ANSWER]
     - Write the string without articles (a, an, the).
     - Don't answer a full sentence when a short version is enough.
     - Do not use abbreviations (e.g., for cities).
-    - Write digits in plain text (e.g., "one" instead of "1") unless specified otherwise.
+    - Write digits in text but (e.g., "one" instead of "1") unless specified otherwise.
+    - Start the first word with a capital letter.
 - If asked for a comma-separated list:
     - Apply the above rules for numbers and strings to each element in the list.
+    - And take care of having a space after each comma.
 
 ## Constraints
 - You must not answer if the constraints above are not respected.
 - Your final answer should be provided in the format: FINAL ANSWER: [YOUR FINAL ANSWER]
 - Your final answer should be a number, a string, or a comma-separated list of numbers and/or strings, following the specified formatting rules.
 
-Now provide your response immediately without any preamble.
+Now provide your response immediately without any preamble in text but not in markdown.
     """
 
     nodes_response = [HumanMessage(content="Here are the results of the previous nodes")]
@@ -760,11 +762,11 @@ Here are the nodes you can choose:
 Here is the question : {state['question']}
 Here is the file : {state.get("input_file", "no file to handle")}
 
-Now provide your response immediately respecting this format: 'next node: \'the node name you choose\''.
+Now provide your response immediately respecting this format in lower case: 'next node: \'the node name you choose\''.
     """
 
     downloaded = ""
-    # If there's an input file, download it directlynext node:
+    # If there's an input file, download it directly:
     if state.get("input_file", None):
         downloaded = download_input_file(state.get("task_id"))
 
@@ -774,15 +776,18 @@ Now provide your response immediately respecting this format: 'next node: \'the
 
     entry_node_response[-1].pretty_print()
 
-    regex_result = re.search(r'.*next\snode:\s+(?P<next_node>[a-z_]*)$', entry_node_response[-1].content)
+    regex_result = re.search(r'.*next\s*node:\s*(?P<next_node>[a-z_]*).*$', entry_node_response[-1].content, re.IGNORECASE)
 
     next_node = "END"
     if regex_result:
         # Extract the node name and remove any quotes around it
-        next_node = regex_result.group("next_node").strip().strip('\'"')
+        next_node = regex_result.group("next_node")
+        next_node = next_node.lower()
+
+    print(f"Next node to invoke: {next_node}")
 
     return {
-        "next": next_node,
+        "next_node": next_node,
         "downloaded_file": downloaded
     }
 
@@ -806,7 +811,7 @@ def build_graph():
     # Conditional routing from entry_node to specialized nodes
     builder.add_conditional_edges(
         "entry_node",
-        lambda state: state["next"],
+        lambda state: state["next_node"],
         {
             "web_wiki_search_node": "web_wiki_search_node",
             "vision_node": "vision_node",
@@ -839,40 +844,57 @@ if __name__ == "__main__":
     
     # print(vision_node.__doc__)
 
-    input =  {
-        "task_id": "cca530fc-4052-43b2-b130-b30968d8aa44",
-        "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.",
-        "Level": "1",
-        "file_name": "cca530fc-4052-43b2-b130-b30968d8aa44.png"
-    }
+    with open("./responses.json", "r") as responses:
+        json_responses = json.loads(responses.read())
 
-    question = input.get("question", "No question found")
-    file_name = input.get("file_name", "")
-    task_id = input.get("task_id", "")
+    # input =  {
+    #     "task_id": "cca530fc-4052-43b2-b130-b30968d8aa44",
+    #     "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.",
+    #     "Level": "1",
+    #     "file_name": "cca530fc-4052-43b2-b130-b30968d8aa44.png"
+    # }
 
-    print(f"QUESTION : {question}")
-    print(f"FILE: {file_name}")
+    with open("questions.json", "r") as questions:
+        json_questions = json.loads(questions.read())
 
-    user_prompt = [HumanMessage(content=f"Can you answer the question please ?")]
+    for input in json_questions:
 
-    user_input = {"messages": user_prompt, "question": question, "input_file": file_name, "task_id": task_id}
+        question = input.get("question", "No question found")
+        file_name = input.get("file_name", "")
+        task_id = input.get("task_id", "")
 
-    messages = agent_graph.invoke(
-        input=user_input,
-        config={
-            "recursion_limit": 10,
-            # "callbacks": [langfuse_handler]
-        }
-    )
+        print(f"QUESTION : {question}")
+        print(f"FILE: {file_name}")
+
+        user_prompt = [HumanMessage(content="Can you answer the question please ?")]
+
+        user_input = {"messages": user_prompt, "question": question, "input_file": file_name, "task_id": task_id}
+
+        messages = agent_graph.invoke(
+            input=user_input,
+            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)
+        if answer == json_responses.get(task_id, ""):
+            print("The answer is correct !")
+        else:
+            print("The answer is incorrect !")
+            print(f"Expected: {json_responses.get(task_id, '')}")
+            print(f"Got: {answer}")
+
+        time.sleep(10)
 
-    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)