import os
from dotenv import load_dotenv
from langchain_openai import OpenAIEmbeddings
import pandas as pd
from supabase import Client, create_client

import nest_asyncio
nest_asyncio.apply()

from news_documents import NewsLoader
from add_vectordb import GetVectorStore

load_dotenv("../.env")
supabase_url = os.environ.get("SUPABASE_URL")
supabase_key = os.environ.get("SUPABASE_KEY")
document_table = "documents2"
supabase: Client = create_client(supabase_url, supabase_key)

embeddings = OpenAIEmbeddings()

vector_store = GetVectorStore(embeddings, supabase, document_table)

response = supabase.table("systex_website_data").select("title", "date", "url", "search_kw", "category", "related_kw", "official_website_source").execute()
url_list = [data['url'] for data in response.data]
supabase_data_list = response.data

loader = NewsLoader(url_list, supabase_data_list)
loader.requests_per_second = 2
_, documents, document_metadatas = loader.aload()
vector_store.insert(documents, document_metadatas)