From 5b3ccc0747f9c0c1c5a2af719a5ce6387952cc8c Mon Sep 17 00:00:00 2001
From: navanchauhan <navanchauhan@gmail.com>
Date: Sat, 21 May 2022 10:18:26 -0600
Subject: adding db to pinecone

---
 db2pc.py | 16 ++++++++++++++++
 1 file changed, 16 insertions(+)
 create mode 100644 db2pc.py

(limited to 'db2pc.py')

diff --git a/db2pc.py b/db2pc.py
new file mode 100644
index 0000000..f0d6ba9
--- /dev/null
+++ b/db2pc.py
@@ -0,0 +1,16 @@
+from database import *
+import pandas as pd
+
+from sentence_transformers import SentenceTransformer
+
+database_url = "sqlite:///jlm.db"
+
+engine, Session = init_db_stuff(database_url)
+
+model = SentenceTransformer("paraphrase-multilingual-MiniLM-L12-v2")
+
+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()))
+
-- 
cgit v1.2.3