tools.py

import os
from dotenv import load_dotenv
from langgraph.graph import START, StateGraph, MessagesState
from langgraph.prebuilt import tools_condition
from langgraph.prebuilt import ToolNode
from langchain_google_genai import ChatGoogleGenerativeAI
from langchain_huggingface import ChatHuggingFace, HuggingFaceEndpoint, HuggingFaceEmbeddings
from langchain_community.tools.tavily_search import TavilySearchResults
from langchain_community.document_loaders import WikipediaLoader
from langchain_community.document_loaders import ArxivLoader
from langchain_community.vectorstores import SupabaseVectorStore
from langchain_core.messages import SystemMessage, HumanMessage
from langchain_core.tools import tool
from langchain.tools.retriever import create_retriever_tool

load_dotenv()

@tool
def multiply(a: int, b: int) -> int:
    """Multiply two numbers.
    Args:
        a: first int
        b: second int
    """
    return a * b

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

@tool
def subtract(a: int, b: int) -> int:
    """Subtract two numbers.
    
    Args:
        a: first int
        b: second int
    """
    return a - b

@tool
def divide(a: int, b: int) -> int:
    """Divide two numbers.
    
    Args:
        a: first int
        b: second int
    """
    if b == 0:
        raise ValueError("Cannot divide by zero.")
    return a / b

@tool
def modulus(a: int, b: int) -> int:
    """Get the modulus of two numbers.
    
    Args:
        a: first int
        b: second int
    """
    return a % b

@tool
def wiki_search(query: str) -> str:
    """Search Wikipedia for a query and return maximum two results.
    
    Args:
        query: The search query."""
    search_docs = WikipediaLoader(query=query, load_max_docs=2).load()
    formatted_search_docs = "\n\n---\n\n".join(
        [
            f'<Document source="{doc.metadata["source"]}" page="{doc.metadata.get("page", "")}"/>\n{doc.page_content}\n</Document>'
            for doc in search_docs
        ])
    return {"wiki_results": formatted_search_docs}

@tool
def web_search(query: str) -> str:
    """Search Tavily for a query and return maximum three results.
    
    Args:
        query: The search query."""
    search_docs = TavilySearchResults(max_results=3).invoke(query=query)
    formatted_search_docs = "\n\n---\n\n".join(
        [
            f'<Document source="{doc.metadata["source"]}" page="{doc.metadata.get("page", "")}"/>\n{doc.page_content}\n</Document>'
            for doc in search_docs
        ])
    return {"web_results": formatted_search_docs}

@tool
def arvix_search(query: str) -> str:
    """Search Arxiv for a query and return maximum three results.
    
    Args:
        query: The search query."""
    search_docs = ArxivLoader(query=query, load_max_docs=3).load()
    formatted_search_docs = "\n\n---\n\n".join(
        [
            f'<Document source="{doc.metadata["source"]}" page="{doc.metadata.get("page", "")}"/>\n{doc.page_content[:1000]}\n</Document>'
            for doc in search_docs
        ])
    return {"arvix_results": formatted_search_docs}



system_prompt="You are a helpful assistant tasked with answering questions using a set of tools. Now, 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, do not use comma to write your number neither use units such as $ or percent sign unless specified otherwise. If you are asked for a string, do not 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.Your answer should only start with 'FINAL ANSWER: ', then follows with the answer."
sys_msg = SystemMessage(content=system_prompt)

import pandas as pd
import ast
import chromadb
from chromadb.utils import embedding_functions

# Step 1: Read the CSV file
csv_file_path = 'embeddings.csv'
df = pd.read_csv(csv_file_path)

# Convert the embeddings from string to list
embeddings = df['embedding'].apply(ast.literal_eval).tolist()

# Convert the metadata from string to dictionary
metadata = df['metadata'].apply(ast.literal_eval).tolist()

# Create unique IDs for each embedding
ids = [str(i) for i in range(len(embeddings))]

# Step 2: Initialize ChromaDB client and create a collection
client = chromadb.Client()
collection = client.create_collection(name="my_collection")

# Step 3: Add embeddings and metadata to the collection
for embedding, meta, id in zip(embeddings, metadata, ids):
    collection.add(
        embeddings=[embedding],
        metadatas=[meta],  # Ensure metadata is a dictionary
        ids=[id]
    )

# Define a function to perform a similarity search
def as_retriever():
    def retriever(query):
        # Assuming `embeddings` is an instance of HuggingFaceEmbeddings
        query_embedding = embeddings.embed_query(query)
        results = collection.query(
            query_embeddings=[query_embedding],
            n_results=1  # Number of nearest neighbors to retrieve
        )
        return results
    return retriever

# Create the retriever tool
create_retriever_tool = {
    "retriever": as_retriever(),
    "name": "Question Search",
    "description": "A tool to retrieve similar questions from a vector store.",
}


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

# Build graph function
def build_graph(provider: str = "huggingface"):
    """Build the graph"""
    # Load environment variables from .env file
    if provider == "google":
        # Google Gemini
        llm = ChatGoogleGenerativeAI(model="gemini-2.0-flash", temperature=0)
    elif provider == "huggingface":
        # HuggingFace Endpoint
        llm = ChatHuggingFace(
            llm=HuggingFaceEndpoint(
                endpoint_url="https://api-inference.huggingface.co/models/HuggingFaceTB/SmolLM2-1.7B-Instruct",
                huggingfacehub_api_token=os.getenv("HF_INFERENCE_ENDPOINT")  # Ensure you have this in your .env file
            )
        )
    else:
        raise ValueError("Invalid provider. Choose 'google' or 'huggingface'.")

    # Bind tools to LLM
    llm_with_tools = llm.bind_tools(tools)

    # Node
    def assistant(state: MessagesState):
        """Assistant node"""
        return {"messages": [llm_with_tools.invoke(state["messages"])]}
    from typing import Dict, List, Any

    from langchain_huggingface import HuggingFaceEmbeddings

    # Initialize the embedding model
    embeddings_model = HuggingFaceEmbeddings(model_name="sentence-transformers/all-mpnet-base-v2")

    def retriever(state: Dict[str, Any]) -> Dict[str, List[HumanMessage]]:
        """Retriever node using ChromaDB for similarity search."""
        # Extract the query from the state
        query = state["messages"][0].content

        # Generate the query embedding using the embedding model
        query_embedding = embeddings_model.embed_query(query)

        # Perform similarity search using ChromaDB
        results = collection.query(
            query_embeddings=[query_embedding],
            n_results=1  # Retrieve the most similar question
        )

        # Extract the similar question content from the results
        similar_question_content = results['documents'][0][0]  # Adjust based on actual structure

        # Create an example message with the similar question
        example_msg = HumanMessage(
            content=f"Here I provide a similar question and answer for reference: \n\n{similar_question_content}",
        )

        # Return the updated state with the example message
        return {"messages": [sys_msg] + state["messages"] + [example_msg]}



    builder = StateGraph(MessagesState)
    builder.add_node("retriever", retriever)
    builder.add_node("assistant", assistant)
    builder.add_node("tools", ToolNode(tools))
    builder.add_edge(START, "retriever")
    builder.add_edge("retriever", "assistant")
    builder.add_conditional_edges(
        "assistant",
        tools_condition,
    )
    builder.add_edge("tools", "assistant")

    # Compile graph
    return builder.compile()