import os import pinecone from database import * import pandas as pd from sentence_transformers import SentenceTransformer from tqdm import tqdm database_url = "sqlite:///jlm.db" engine, Session = init_db_stuff(database_url) PINECONE_KEY = os.getenv("PINECONE_API_DEFAULT") pinecone.init(api_key=PINECONE_KEY, environment="us-west1-gcp") index = pinecone.Index("movies") model = SentenceTransformer("paraphrase-multilingual-MiniLM-L12-v2") batch_size = 32 df = pd.read_sql("Select * from movies", engine) df["combined_text"] = ( df["title"] + ": " + df["overview"].fillna("") + " - " + df["tagline"].fillna("") + " Genres:- " + df["genres"].fillna("") ) print(f'Length of Combined Text: {len(df["combined_text"].tolist())}') for x in tqdm(range(0, len(df), batch_size)): to_send = [] trakt_ids = df["trakt_id"][x : x + batch_size].tolist() sentences = df["combined_text"][x : x + batch_size].tolist() embeddings = model.encode(sentences) for idx, value in enumerate(trakt_ids): to_send.append((str(value), embeddings[idx].tolist())) index.upsert(to_send)