import os
from dotenv import load_dotenv
from bs4.filter import SoupStrainer
from langchain_community.document_loaders import WebBaseLoader
from langchain.text_splitter import RecursiveCharacterTextSplitter
from langchain_openai import ChatOpenAI, OpenAIEmbeddings
from langchain_community.vectorstores import Chroma
from langchain.load import dumps, loads

def get_retriever(link: str):
    loader = WebBaseLoader(
    web_path=(link,),
    bs_kwargs=dict(
            parse_only = SoupStrainer(
                class_ = ("post-content", "post-title", "post-header")
            )
        )
    )
    blog_docs = loader.load()

    text_splitter = RecursiveCharacterTextSplitter(
        chunk_size=300,
        chunk_overlap=50,
    )

    splits = text_splitter.split_documents(blog_docs) # type: ignore
    vectorstore = Chroma.from_documents(
        documents=splits, # type: ignore
        embedding=OpenAIEmbeddings(),
        persist_directory="chroma_db",
        collection_name="lilian_weng_agent_blog"
    )

    retrievar = vectorstore.as_retriever()  
    return retrievar


def get_llm():
    llm = ChatOpenAI(temperature=0)
    return llm