tools.py

import os
from dotenv import load_dotenv
from langchain.chat_models import ChatOpenAI
from langchain.document_loaders import PyPDFLoader
from langchain.text_splitter import CharacterTextSplitter
from langchain.vectorstores import FAISS
from langchain.embeddings import HuggingFaceEmbeddings
import serpapi

load_dotenv()

os.environ["HF_HOME"] = "/tmp/huggingface"
os.environ["TRANSFORMERS_CACHE"] = "/tmp/huggingface/transformers"
os.environ["HF_DATASETS_CACHE"] = "/tmp/huggingface/datasets"


# LLM (Groq + LLaMA3)
llm = ChatOpenAI(
    model="llama3-8b-8192",
    openai_api_base="https://api.groq.com/openai/v1",
    openai_api_key=os.environ["GROQ_API_KEY"]
)

# Embeddings (HuggingFace)
embeddings = HuggingFaceEmbeddings(model_name="all-MiniLM-L6-v2")

# Load PDFs and create FAISS vectorstore
def load_vectorstore(pdf_dir="pdfs/"):
    docs = []
    for file in os.listdir(pdf_dir):
        if file.endswith(".pdf"):
            loader = PyPDFLoader(os.path.join(pdf_dir, file))
            docs.extend(loader.load())
    splitter = CharacterTextSplitter(chunk_size=1000, chunk_overlap=200)
    chunks = splitter.split_documents(docs)
    return FAISS.from_documents(chunks, embedding=embeddings)

# Custom Web Search tool using SerpAPI
def search_tool(query: str):
    client = serpapi.Client(api_key=os.getenv("SERPAPI_API_KEY"))
    search = client.search({
        "engine": "google",
        "q": query,
    })
    results = dict(search)
    return results["organic_results"][0]["snippet"]  # Return the snippet or any part of the result