agent.py

from langgraph.graph.message import add_messages
from langchain_core.messages import AnyMessage, HumanMessage, AIMessage, SystemMessage
from langgraph.prebuilt import ToolNode
from langgraph.graph import START, StateGraph
from langgraph.prebuilt import tools_condition
from langchain_huggingface import HuggingFaceEndpoint, ChatHuggingFace
from tools import extract_text, describe_image
from langchain_community.tools import DuckDuckGoSearchRun
from langchain_openai import ChatOpenAI
from typing import TypedDict, Annotated, Optional


class AgentState(TypedDict):
    messages: Annotated[list[AnyMessage], add_messages]


class BasicAgent():
    def __init__(self, llm):
        chat = ChatHuggingFace(llm=llm, verbose=True)
        
        search_tool = DuckDuckGoSearchRun()
        vision_llm = ChatOpenAI(model="gpt-4o")
        self.tools = [extract_text, describe_image, search_tool]
        self.chat_with_tools = chat.bind_tools(self.tools)
        self._initialize_graph()
        print("BasicAgent initialized.")

    
    def _initialize_graph(self):
        builder = StateGraph(AgentState)

        # Define nodes
        builder.add_node("assistant", self.assistant)
        builder.add_node("tools", ToolNode(self.tools))

        # Define edges
        builder.add_edge(START, "assistant")
        builder.add_conditional_edges("assistant",tools_condition)
        builder.add_edge("tools", "assistant")

        # Compile the graph
        self.agent = builder.compile()

    
    def __call__(self, question: str) -> str:
        print(f"Agent received question (first 50 chars): {question[:50]}...")
        messages=[HumanMessage(content=question)]
        response = self.agent.invoke({"messages":messages})
        answer = response['messages'][-1].content
        print(f"Agent returning answer: {answer}")
        return answer

    
    def assistant(self, state: AgentState):
        textual_description_of_tool="""
        extract_text(img_path: str) -> str:
            Extract text from an image file using a multimodal model.
        
            Args:
                img_path: A url pointing to an image (e.g., PNG, JPEG).
        
            Returns:
                A single string containing the concatenated text extracted from each image.
        
        search_tool(query: str) -> str:
            Search the web using the DuckDuckGoSearchRun to perform a search query and return a summarized textual result.

            Args:
                query: A string representing the search query.
        
            Returns:
                A single string containing the search result or summary.

        describe_image(img_path: str, query: str) -> str:
            Generate a detailed description of an image using a multimodal model.
            This function reads a image from an url, encodes it, and sends it to a 
            vision-capable language model to obtain a comprehensive, natural language 
            description of the image's content, including its objects, actions, and context,
            following a specific query.
            
            Args:
                img_path: A url pointing to an image (e.g., PNG, JPEG).
                query: Information to extract from the image
                
            Returns:
                A single string containing a detailed, human-readable description of the image.                
        """        
        sys_msg = SystemMessage(content=f"""
        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]. 
        YOUR FINAL ANSWER should be a number OR as few words as possible OR a comma separated list of numbers and/or strings. If you are asked for a number, don't use comma to write your number neither use units such as $ or percent sign unless specified otherwise. 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 of whether the element to be put in the list is a number or a string.
        You have access to the following tools:\n{textual_description_of_tool}\n""")

        return {
            "messages": [self.chat_with_tools.invoke([sys_msg] + state["messages"])],            
        }