agent.py

import os
from dotenv import load_dotenv
from langgraph.graph import START, StateGraph, MessagesState, END
from langgraph.prebuilt import tools_condition, ToolNode
from langchain_google_genai import ChatGoogleGenerativeAI
from langchain_groq import ChatGroq
from langchain_huggingface import ChatHuggingFace, HuggingFaceEndpoint
from langchain_community.tools.tavily_search import TavilySearchResults
from langchain_community.document_loaders import WikipediaLoader, ArxivLoader
from langchain_core.messages import SystemMessage, HumanMessage, AIMessage
from langchain_core.tools import tool
from langchain_groq import ChatGroq
from supabase import create_client
from langchain_huggingface import HuggingFaceEmbeddings
from langchain_community.vectorstores import SupabaseVectorStore
from langchain_openai import ChatOpenAI
from langchain_core.documents import Document
import json

load_dotenv()

# ------------------- TOOL DEFINITIONS -------------------
@tool
def multiply(a: int, b: int) -> int:
    """Multiply two numbers."""
    return a * b

@tool
def add(a: int, b: int) -> int:
    """Add two numbers."""
    return a + b

@tool
def subtract(a: int, b: int) -> int:
    """Subtract b from a."""
    return a - b

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

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

@tool
def wiki_search(query: str) -> str:
    """Search Wikipedia for a query (max 2 results)."""
    docs = WikipediaLoader(query=query, load_max_docs=2).load()
    return "\n\n".join([doc.page_content for doc in docs])

@tool
def web_search(query: str) -> str:
    """Search the web using Tavily (max 3 results)."""
    results = TavilySearchResults(max_results=3).invoke(query)
    texts = [doc.get("content", "") or doc.get("text", "") for doc in results if isinstance(doc, dict)]
    return "\n\n".join(texts)

@tool
def arvix_search(query: str) -> str:
    """Search Arxiv for academic papers (max 3 results, truncated to 1000 characters each)."""
    docs = ArxivLoader(query=query, load_max_docs=3).load()
    return "\n\n".join([doc.page_content[:1000] for doc in docs])

@tool
def read_excel_file(path: str) -> str:
    """Read an Excel file and return the first few rows of each sheet as text."""
    import pandas as pd
    try:
        xls = pd.ExcelFile(path)
        content = ""
        for sheet in xls.sheet_names:
            df = xls.parse(sheet)
            content += f"Sheet: {sheet}\n"
            content += df.head(5).to_string(index=False) + "\n\n"
        return content.strip()
    except Exception as e:
        return f"Error reading Excel file: {str(e)}"



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

# ------------------- SYSTEM PROMPT -------------------
system_prompt_path = "system_prompt.txt"
if os.path.exists(system_prompt_path):
    with open(system_prompt_path, "r", encoding="utf-8") as f:
        system_prompt = f.read()
else:
    system_prompt = (
        "You are an intelligent AI agent who can solve math, science, factual, and research-based problems. "
        "You can use tools like Wikipedia, Web search, or Arxiv when needed. Always give precise and helpful answers."
    )
sys_msg = SystemMessage(content=system_prompt)

# ------------------- GRAPH CONSTRUCTION -------------------
def build_graph(provider: str = "groq"):
    if provider == "google":
        llm = ChatGoogleGenerativeAI(model="gemini-2.0-flash", temperature=0)
    elif provider == "groq":
        groq_key = os.getenv("GROQ_API_KEY")
        if not groq_key:
            raise ValueError("GROQ_API_KEY is not set.")
        llm = ChatGroq(model="qwen-qwq-32b", temperature=0, api_key=groq_key)
    elif provider == "huggingface":
        llm = ChatHuggingFace(
            llm=HuggingFaceEndpoint(
                url="https://api-inference.huggingface.co/models/Meta-DeepLearning/llama-2-7b-chat-hf",
                temperature=0
            )
        )
    elif provider == "openai":
        openai_key = os.getenv("OPENAI_API_KEY")
        if not openai_key:
            raise ValueError("OPENAI_API_KEY is not set.")
        llm = ChatOpenAI(model="gpt-3.5-turbo", temperature=0, api_key=openai_key)
    else:
        raise ValueError("Invalid provider")

    llm_with_tools = llm.bind_tools(tools)

    def assistant(state: MessagesState):
        return {"messages": [sys_msg] + [llm_with_tools.invoke(state["messages"])]}

    SUPABASE_URL = os.getenv("SUPABASE_URL")
    SUPABASE_KEY = os.getenv("SUPABASE_SERVICE_KEY")
    supabase = create_client(SUPABASE_URL, SUPABASE_KEY)

    embedding_model = HuggingFaceEmbeddings(model_name="sentence-transformers/all-mpnet-base-v2")
    vectorstore = SupabaseVectorStore(
        client=supabase,
        embedding=embedding_model,
        table_name="QA_db"
    )
    retriever = vectorstore.as_retriever(search_kwargs={"k": 1})


    # ✅ 替换 similarity_search_by_vector_with_relevance_scores 方法，直接调用 supabase.rpc
    original_fn = vectorstore.similarity_search_by_vector_with_relevance_scores

    # ✅ 覆盖 vectorstore 的方法
    def patched_fn(embedding, k=4, filter=None, **kwargs):
        response = supabase.rpc(
            "match_documents",
            {
                "query_embedding": embedding,
                "match_count": k
            }
        ).execute()

        documents = []
        for r in response.data:
            metadata = r["metadata"]
            if isinstance(metadata, str):
                try:
                    metadata = json.loads(metadata)
                except Exception:
                    metadata = {}
            doc = Document(
                page_content=r["content"],
                metadata=metadata
            )
            documents.append((doc, r["similarity"]))
        return documents

    # ✅ 覆盖 vectorstore 的方法
    vectorstore.similarity_search_by_vector_with_relevance_scores = patched_fn

    def qa_retriever_node(state: MessagesState):
        user_question = state["messages"][-1].content
        docs = retriever.invoke(user_question)
        if docs:
            return {
                "messages": state["messages"] + [AIMessage(content=docs[0].page_content)],
                "__condition__": "complete"
            }
        return {"messages": state["messages"], "__condition__": "default"}

    builder = StateGraph(MessagesState)
    builder.add_node("retriever", qa_retriever_node)
    builder.add_node("assistant", assistant)
    builder.add_node("tools", ToolNode(tools))

    builder.add_edge(START, "retriever")
    builder.add_conditional_edges("retriever", {
        "default": lambda x: "assistant",
        "complete": lambda x: END,
    })
    builder.add_conditional_edges("assistant", tools_condition)
    builder.add_edge("tools", "assistant")

    return builder.compile()

# ------------------- LOCAL TEST -------------------
if __name__ == "__main__":
    question = "When was a picture of St. Thomas Aquinas first added to the Wikipedia page on the Principle of double effect?"
    graph = build_graph(provider="openai")
    messages = graph.invoke({"messages": [HumanMessage(content=question)]})
    print("=== AI Agent Response ===")
    for m in messages["messages"]:
        m.pretty_print()