agent.py

"""LangGraph Agent – retry 5s, 30s, 60s; senza Supabase"""

import os
import time
from dotenv import load_dotenv
from langgraph.graph import START, StateGraph, MessagesState
from langgraph.prebuilt import ToolNode, tools_condition

# LLM providers
from langchain_google_genai import ChatGoogleGenerativeAI
from langchain_groq import ChatGroq
from langchain_huggingface import ChatHuggingFace, HuggingFaceEndpoint

# Tools & loaders
from langchain_community.tools.tavily_search import TavilySearchResults
from langchain_community.document_loaders import WikipediaLoader, ArxivLoader
from langchain_core.messages import SystemMessage, HumanMessage
from langchain_core.tools import tool

load_dotenv()

# --------------------------------------------------------------------------- #
#  TOOLS                                                                      #
# --------------------------------------------------------------------------- #
@tool
def multiply(a: int, b: int) -> int:
    """Multiply two integers and return the product."""
    return a * b

@tool
def add(a: int, b: int) -> int:
    """Add two integers and return the sum."""
    return a + b

@tool
def subtract(a: int, b: int) -> int:
    """Subtract the second integer from the first and return the difference."""
    return a - b

@tool
def divide(a: int, b: int) -> float:
    """Divide a by b and return the quotient (error if b == 0)."""
    if b == 0:
        raise ValueError("Cannot divide by zero.")
    return a / b

@tool
def modulus(a: int, b: int) -> int:
    """Return the remainder of the division of a by b."""
    return a % b

@tool
def wiki_search(query: str) -> str:
    """Search Wikipedia (max 2 docs) and return formatted content."""
    docs = WikipediaLoader(query=query, load_max_docs=2).load()
    return "\n\n---\n\n".join(
        f'<Document source="{d.metadata["source"]}" page="{d.metadata.get("page","")}"/>\n'
        f"{d.page_content}\n</Document>"
        for d in docs
    )

@tool
def web_search(query: str) -> str:
    """Perform a web search with Tavily (max 3 docs) and return formatted content."""
    docs = TavilySearchResults(max_results=3).invoke(query=query)
    return "\n\n---\n\n".join(
        f'<Document source="{d.metadata["source"]}" page="{d.metadata.get("page","")}"/>\n'
        f"{d.page_content}\n</Document>"
        for d in docs
    )

@tool
def arxiv_search(query: str) -> str:
    """Search ArXiv (max 3 docs) and return first 1000 characters per paper."""
    docs = ArxivLoader(query=query, load_max_docs=3).load()
    return "\n\n---\n\n".join(
        f'<Document source="{d.metadata["source"]}" page="{d.metadata.get("page","")}"/>\n'
        f"{d.page_content[:1000]}\n</Document>"
        for d in docs
    )

# --------------------------------------------------------------------------- #
#  System prompt                                                              #
# --------------------------------------------------------------------------- #
with open("system_prompt.txt", "r", encoding="utf-8") as f:
    system_prompt = f.read()
sys_msg = SystemMessage(content=system_prompt)

tools = [
    multiply, add, subtract, divide, modulus,
    wiki_search, web_search, arxiv_search,
]

# --------------------------------------------------------------------------- #
#  Retry parameters                                                           #
# --------------------------------------------------------------------------- #
RETRY_DELAYS = [0, 5, 30, 60]   # 4 tentativi complessivi
MAX_ATTEMPTS = len(RETRY_DELAYS)

# --------------------------------------------------------------------------- #
#  Build LangGraph                                                            #
# --------------------------------------------------------------------------- #
def build_graph(provider: str = "groq"):
    """Return a LangGraph graph with custom retry logic."""

    # ----------- LLM selection -------------------------------------------- #
    if provider == "google":
        llm_selected = ChatGoogleGenerativeAI(model="gemini-2.0-flash", temperature=0)

    elif provider == "groq":
        llm_selected = ChatGroq(
            model="qwen-qwq-32b",
            temperature=0,
            max_retries=0,  # gestiamo noi i retry
        )

    elif provider == "huggingface":
        llm_selected = ChatHuggingFace(
            llm=HuggingFaceEndpoint(
                url="https://api-inference.huggingface.co/models/Meta-DeepLearning/llama-2-7b-chat-hf",
                temperature=0,
            )
        )
    else:
        raise ValueError("Invalid provider. Choose 'google', 'groq' or 'huggingface'.")

    llm_with_tools = llm_selected.bind_tools(tools)

    # ---------------- Retry wrapper -------------------------------------- #
    def invoke_with_retry(messages):
        last_err = None
        for attempt, delay in enumerate(RETRY_DELAYS):
            if delay:
                print(f"[Retry {attempt}/{MAX_ATTEMPTS-1}] waiting {delay}s")
                time.sleep(delay)
            try:
                return llm_with_tools.invoke(messages)
            except Exception as e:
                err_text = str(e)
                if ("503" in err_text or "Service Unavailable" in err_text) and attempt < MAX_ATTEMPTS - 1:
                    last_err = e
                    continue  # retry
                raise  # altro errore o tentativi finiti
        raise last_err or RuntimeError("Unknown error during LLM invocation")

    # ---------------- Nodes ---------------------------------------------- #
    def assistant(state: MessagesState):
        messages = [sys_msg] + state["messages"]
        return {"messages": [invoke_with_retry(messages)]}

    # ---------------- Graph ---------------------------------------------- #
    builder = StateGraph(MessagesState)
    builder.add_node("assistant", assistant)
    builder.add_node("tools", ToolNode(tools))

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

    return builder.compile()


# --------------------------------------------------------------------------- #
#  Stand-alone test                                                           #
# --------------------------------------------------------------------------- #
if __name__ == "__main__":
    graph = build_graph(provider="groq")
    question = (
        "When was a picture of St. Thomas Aquinas first added to the Wikipedia "
        "page on the Principle of double effect?"
    )
    msgs = [HumanMessage(content=question)]
    result = graph.invoke({"messages": msgs})
    for m in result["messages"]:
        m.pretty_print()