Update app.py

app.py

@@ -14,6 +14,17 @@ from youtube_tool import youtube_transcript_tool, youtube_transcript_snippet_too
 from multiple_tools import round_to_two_decimals_tool, text_inverter_tool, google_web_search_tool, wikipedia_search_tool
 from agent import smart_agent
 from llama_index.llms.openai import OpenAI
+#-----------------------------------------------------------------
+from langgraph_impl.my_agent import create_langchain_agent
+#from langchain_community.llms import OpenAI
+from langchain.tools import Tool
+from langgraph_impl.my_agent import create_langchain_agent
+
+from langgraph_impl.wikipedia_tool import wiki_search
+from langgraph_impl.youtube_tool import get_youtube_transcript
+from langgraph_impl.audio_transcriber import transcribe_audio
+from langgraph_impl.image_chess_solver import solve_chess_image
+from langgraph_impl.file_parser import parse_file_and_summarize
 
 # (Keep Constants as is)
 # --- Constants ---
@@ -22,71 +33,50 @@ DEFAULT_API_URL = "https://agents-course-unit4-scoring.hf.space"
 # --- Basic Agent Definition ---
 
 # ----- THIS IS WERE YOU CAN BUILD WHAT YOU WANT ------
+# class BasicAgent:
+#     def __init__(self):
+#         print("BasicAgent initialized. . . .")
+#         # self.agent = smart_agent()
+#         # self.api_key = os.getenv("HF_TOKEN")
+#         OpenAI_key = os.getenv("OPEN_AI_TOKEN")
+#         self.llm = OpenAI(model="gpt-4o-mini", temperature=0.2, api_key=OpenAI_key)
+#         self.agent = AgentWorkflow.from_tools_or_functions(
+#             [wikipedia_search_tool, youtube_transcript_tool, youtube_transcript_snippet_tool, round_to_two_decimals_tool, text_inverter_tool, google_web_search_tool],
+#             llm=self.llm,
+#         )
+
+#     async def run(self, question: str) -> str:
+#         print(f"Agent received question (first 50 chars): {question[:50]}...")
+#         # answer = await self.agent.run(question)
+#         answer = await self.agent.run(
+#             f"{question}\n\nIf you have enough information, respond with a concise final answer.",
+#             max_iterations=50
+#         )
+#         if hasattr(answer, "output"):
+#             print(f"Agent returning answer: {answer}")
+#             return str(answer.output)
+#         else:
+#             print(f"Agent returning answer: {answer}")
+#             return str(answer)
+        
+#     def __call__(self, question: str) -> str:
+#         return asyncio.run(self.run(question))
+
+
 class BasicAgent:
     def __init__(self):
-        print("BasicAgent initialized. . . .")
-        # self.agent = smart_agent()
-        # self.api_key = os.getenv("HF_TOKEN")
-        # self.llm = HuggingFaceInferenceAPI(
-        #     model_name="deepseek-ai/DeepSeek-R1-0528",
-        #     token=self.api_key,
-        #     provider="auto",
-        #     max_iterations=10
-        # )
-        # self.agent = AgentWorkflow.from_tools_or_functions(
-        #     [wikipedia_search_tool, youtube_transcript_tool, youtube_transcript_snippet_tool, 
-        #      round_to_two_decimals_tool, text_inverter_tool, google_web_search_tool],
-        #     llm=self.llm
-        # )
-        OpenAI_key = os.getenv("OPEN_AI_TOKEN")
-        agent_kwargs = {
-            "prefix": (
-                "You are a helpful AI assistant completing GAIA benchmark tasks.\n"
-                "You MUST use the tools provided to answer the user's question. Do not answer from your own knowledge.\n"
-                "Carefully analyze the question to determine the most appropriate tool to use.\n"
-                "Here are guidelines for using the tools:\n"
-                "- Use 'wikipedia_search' to find factual information about topics, events, people, etc. (e.g., 'Use wikipedia_search to find the population of France').\n"
-                "- Use 'youtube_transcript' to extract transcripts from YouTube videos when the question requires understanding the video content. (e.g., 'Use youtube_transcript to summarize the key points of this video').\n"
-                "- Use 'audio_transcriber' to transcribe uploaded audio files. (e.g., 'Use audio_transcriber to get the text from this audio recording').\n"
-                "- Use 'chess_image_solver' to analyze and solve chess puzzles from images. (e.g., 'Use chess_image_solver to determine the best move in this chess position').\n"
-                "- Use 'file_parser' to parse and analyze data from Excel or CSV files. (e.g., 'Use file_parser to calculate the average sales from this data').\n"
-                "- Use 'vegetable_classifier_2022' to classify a list of food items and extract only the vegetables. (e.g., 'Use vegetable_classifier_2022 to get a list of the vegetables in this grocery list').\n"
-                "- Use 'excel_food_sales_sum' to extract total food sales from excel files. (e.g., 'Use excel_food_sales_sum to calculate the total food sales').\n"
-                "Do NOT guess or make up answers. If a tool cannot provide the answer, truthfully respond that you were unable to find the information.\n"
-            ),
-            "suffix": (
-                "Use the tools to research or calculate the answer.\n"
-                "If a tool fails, explain the reason for the failure instead of hallucinating an answer.\n"
-                "Provide concise and direct answers as requested in the questions. Do not add extra information unless explicitly asked for.\n"
-                "For example, if asked for a number, return only the number. If asked for a list, return only the list.\n"
-            ),
-            "max_iterations": 100,
-            "verbose": false
-        }
-        self.llm = OpenAI(model="gpt-4o-mini", temperature=0.2, api_key=OpenAI_key)
-        self.agent = AgentWorkflow.from_tools_or_functions(
-            [wikipedia_search_tool, youtube_transcript_tool, youtube_transcript_snippet_tool, round_to_two_decimals_tool, text_inverter_tool, google_web_search_tool],
-            llm=self.llm,
-        )
+        print("Initializing LangChain Agent...")
+        self.agent = create_langchain_agent()
 
-    async def run(self, question: str) -> str:
-        print(f"Agent received question (first 50 chars): {question[:50]}...")
-        # answer = await self.agent.run(question)
-        answer = await self.agent.run(
-            f"{question}\n\nIf you have enough information, respond with a concise final answer.",
-            max_iterations=50
-        )
-        if hasattr(answer, "output"):
-            print(f"Agent returning answer: {answer}")
-            return str(answer.output)
-        else:
-            print(f"Agent returning answer: {answer}")
-            return str(answer)
-        
     def __call__(self, question: str) -> str:
-        return asyncio.run(self.run(question))
-        #return self.run(question)
-        #return self.agent.run(question)
+        try:
+            result = self.agent.invoke({"input": question})
+            if isinstance(result, dict) and "output" in result:
+                return result["output"]
+            return result
+        except Exception as e:
+            return f"Agent error: {e}"
+
 
 
 def run_and_submit_all( profile: gr.OAuthProfile | None):