updated create and update embeddings

db2pc.py

@@ -1,16 +1,37 @@
+import os
+import pinecone
+
 from database import *
 import pandas as pd
 
 from sentence_transformers import SentenceTransformer
 
-database_url = "sqlite:///jlm.db"
+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(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(
+			{
+				value: embeddings[idx]
+			})
+	index.upsert(to_send)