Update app.py

app.py

@@ -4,7 +4,7 @@ import requests
 import pandas as pd
 from typing import TypedDict, Annotated, Sequence
 import operator
-from langchain_core.messages import BaseMessage, HumanMessage
+from langchain_core.messages import BaseMessage, HumanMessage, SystemMessage
 from langchain.agents import AgentExecutor
 from langchain_experimental.tools import PythonREPLTool
 from langchain_community.tools.youtube.search import YouTubeSearchTool
@@ -21,10 +21,10 @@ class AgentState(TypedDict):
 
 # Agentin rakentajafunktio
 def create_langgraph_agent():
-    print("Initializing Advanced LangGraph Agent...")
+    print("Initializing Advanced LangGraph Agent…")
 
     # 1. System prompt GAIA-tyyliin
-    system_prompt = """
+    SYSTEM_PROMPT = """
 You are a general AI assistant. I will ask you a question. Report your thoughts, and finish your answer with the following template:
 FINAL ANSWER: [YOUR FINAL ANSWER].
 
@@ -33,28 +33,18 @@ If you are asked for a number, don't use comma to write your number neither use
 If you are asked for a string, don't use articles, neither abbreviations (e.g. for cities), and write the digits in plain text unless specified otherwise.
 If you are asked for a comma separated list, apply the above rules depending on whether the element to be put in the list is a number or a string.
 """
-    llm = ChatOpenAI(model="gpt-4o", temperature=0, system_message=system_prompt)
 
-    # 2. Perustyökalut: Tavily, PythonREPL, YouTube
+    # 2. LLM (ei system_message-parametria -> annetaan prompt SystemMessage-na)
+    llm = ChatOpenAI(model="gpt-4o", temperature=0)
+
+    # 3. Perustyökalut
     tools = [
         TavilySearchResults(max_results=3),
         PythonREPLTool(),
         YouTubeSearchTool(),
     ]
 
-    # 3. Valinnainen Playwright-selain toolkit
-    try:
-        from langchain_community.tools.playwright.utils import create_sync_playwright_browser
-        from langchain_community.agent_toolkits.playwright.toolkit import PlayWrightBrowserToolkit
-
-        sync_browser = create_sync_playwright_browser()
-        browser_toolkit = PlayWrightBrowserToolkit.from_browser(sync_browser=sync_browser)
-        tools.extend(browser_toolkit.get_tools())
-        print("Playwright tools loaded.")
-    except Exception as e:
-        print("Playwright not available, skipping browser tools:", e)
-
-    # 4. Valinnainen FileManagement toolkit
+    # 4. Valinnainen FileManagement toolkit (kevyt, yleensä saatavilla)
     try:
         from langchain_community.agent_toolkits.file_management.toolkit import FileManagementToolkit
         file_toolkit = FileManagementToolkit(root_dir=".")
@@ -63,20 +53,21 @@ If you are asked for a comma separated list, apply the above rules depending on
     except Exception as e:
         print("FileManagement toolkit unavailable:", e)
 
-    # Bind tools to LLM
+    # 5. Bind tools
     llm_with_tools = llm.bind_tools(tools)
     print("LLM and tools initialized.")
 
-    # 5. Agentin solmu (kutsuu kielimallia)
+    # 6. Agent-solmu (lisää system prompt joka kierroksella)
     def agent_node(state):
-        print("Calling agent node...")
-        return {"messages": [llm_with_tools.invoke(state["messages"])]}
+        print("Calling agent node…")
+        full_msgs = [SystemMessage(content=SYSTEM_PROMPT)] + list(state["messages"])
+        reply = llm_with_tools.invoke(full_msgs)
+        return {"messages": [reply]}
 
-    # 6. Työkalusolmu
+    # 7. Työkalusolmu
     tool_node = ToolNode(tools)
-    print("Graph nodes defined.")
 
-    # 7. Graafin määritys
+    # 8. Rakenna graafi
     graph = StateGraph(AgentState)
     graph.add_node("agent", agent_node)
     graph.add_node("tools", tool_node)
@@ -84,8 +75,7 @@ If you are asked for a comma separated list, apply the above rules depending on
     graph.add_conditional_edges("agent", tools_condition)
     graph.add_edge("tools", "agent")
 
-    # 8. Graafin kääntäminen
-    app = graph.compile()  # rekursion raja määritellään invoke-kutsussa
+    app = graph.compile()
     print("LangGraph agent compiled and ready.")
     return app
 
@@ -110,15 +100,16 @@ def run_agent(agent_executor, question: str) -> str:
     print(f"Agent returning answer: {final_answer}")
     return str(final_answer)
 
-# Evaluaation ajaminen
+# Evaluaation ajaminen ja tulosten lähetys
+
 def run_and_submit_all(profile: gr.OAuthProfile | None):
     space_id = os.getenv("SPACE_ID")
     if not profile:
-        return "Please Login to Hugging Face with the button.", None
+        return "Please login to Hugging Face.", None
     username = f"{profile.username}"
 
     if not os.getenv("TAVILY_API_KEY") or not os.getenv("OPENAI_API_KEY"):
-        return "One or more API keys (TAVILY_API_KEY, OPENAI_API_KEY) are not set.", None
+        return "API keys missing (TAVILY_API_KEY, OPENAI_API_KEY)", None
 
     try:
         agent_executor = create_langgraph_agent()
@@ -126,7 +117,8 @@ def run_and_submit_all(profile: gr.OAuthProfile | None):
         return f"Error initializing agent: {e}", None
 
     agent_code = f"https://huggingface.co/spaces/{space_id}/tree/main"
-    questions_url = f"https://agents-course-unit4-scoring.hf.space/questions"
+    questions_url = "https://agents-course-unit4-scoring.hf.space/questions"
+
     try:
         response = requests.get(questions_url, timeout=20)
         response.raise_for_status()
@@ -141,8 +133,13 @@ def run_and_submit_all(profile: gr.OAuthProfile | None):
             submitted_answer = run_agent(agent_executor, question_text)
             answers_payload.append({"task_id": task_id, "submitted_answer": submitted_answer})
 
-    submission_data = {"username": username.strip(), "agent_code": agent_code, "answers": answers_payload}
-    submit_url = f"https://agents-course-unit4-scoring.hf.space/submit"
+    submission_data = {
+        "username": username.strip(),
+        "agent_code": agent_code,
+        "answers": answers_payload,
+    }
+    submit_url = "https://agents-course-unit4-scoring.hf.space/submit"
+
     try:
         response = requests.post(submit_url, json=submission_data, timeout=240)
         response.raise_for_status()
@@ -152,7 +149,7 @@ def run_and_submit_all(profile: gr.OAuthProfile | None):
             f"User: {result_data.get('username')}\n"
             f"Overall Score: {result_data.get('score', 'N/A')}% "
             f"({result_data.get('correct_count', '?')}/{result_data.get('total_attempted', '?')} correct)\n"
-            f"Message: {result_data.get('message', 'No message received.') }"
+            f"Message: {result_data.get('message', 'No message received.')}"
         )
         return final_status, pd.DataFrame(answers_payload)
     except Exception as e:
@@ -160,12 +157,13 @@ def run_and_submit_all(profile: gr.OAuthProfile | None):
 
 # Gradio-käyttöliittymä
 with gr.Blocks() as demo:
-    gr.Markdown("# Agent Evaluation Runner (Advanced Tools - Corrected)")
+    gr.Markdown("# Agent Evaluation Runner (GAIA Prompt)")
     gr.LoginButton()
     run_button = gr.Button("Run Evaluation & Submit All Answers")
     status_output = gr.Textbox(label="Run Status / Submission Result", lines=5, interactive=False)
     results_table = gr.DataFrame(label="Questions and Agent Answers", wrap=True)
+
     run_button.click(fn=run_and_submit_all, outputs=[status_output, results_table])
 
 if __name__ == "__main__":
-    demo.launch()
+    demo.launch()