Spaces:
Running
Running
Commit ·
f2a62c1
1
Parent(s): 3040a98
migrated research papers to supabase
Browse files- migrate.py +51 -0
migrate.py
ADDED
|
@@ -0,0 +1,51 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
import psycopg2
|
| 2 |
+
from psycopg2.extras import RealDictCursor
|
| 3 |
+
from dotenv import load_dotenv
|
| 4 |
+
import os
|
| 5 |
+
|
| 6 |
+
load_dotenv(dotenv_path='E:/ClearVoice/.env', override=True)
|
| 7 |
+
|
| 8 |
+
|
| 9 |
+
local_conn = psycopg2.connect(os.getenv('DATABASE_URL'), cursor_factory=RealDictCursor)
|
| 10 |
+
local_cur = local_conn.cursor()
|
| 11 |
+
|
| 12 |
+
supa_conn = psycopg2.connect(os.getenv('SUPABASE_DATABASE_URL'))
|
| 13 |
+
supa_cur = supa_conn.cursor()
|
| 14 |
+
|
| 15 |
+
print("Fetching papers from local DB...")
|
| 16 |
+
local_cur.execute("SELECT pmid, title, abstract, authors, year, journal, study_type, embedding FROM studies")
|
| 17 |
+
papers = local_cur.fetchall()
|
| 18 |
+
print(f"Found {len(papers)} papers. Migrating...")
|
| 19 |
+
|
| 20 |
+
for i, paper in enumerate(papers):
|
| 21 |
+
try:
|
| 22 |
+
supa_cur.execute("""
|
| 23 |
+
INSERT INTO studies (pmid, title, abstract, authors, year, journal, study_type, embedding)
|
| 24 |
+
VALUES (%s, %s, %s, %s, %s, %s, %s, %s)
|
| 25 |
+
ON CONFLICT (pmid) DO NOTHING
|
| 26 |
+
""", (
|
| 27 |
+
paper['pmid'],
|
| 28 |
+
paper['title'],
|
| 29 |
+
paper['abstract'],
|
| 30 |
+
paper['authors'],
|
| 31 |
+
paper['year'],
|
| 32 |
+
paper['journal'],
|
| 33 |
+
paper['study_type'],
|
| 34 |
+
paper['embedding']
|
| 35 |
+
))
|
| 36 |
+
|
| 37 |
+
if (i + 1) % 100 == 0:
|
| 38 |
+
supa_conn.commit()
|
| 39 |
+
print(f" Migrated {i + 1} papers...")
|
| 40 |
+
|
| 41 |
+
except Exception as e:
|
| 42 |
+
print(f"Error on {paper['pmid']}: {e}")
|
| 43 |
+
supa_conn.rollback()
|
| 44 |
+
continue
|
| 45 |
+
|
| 46 |
+
supa_conn.commit()
|
| 47 |
+
local_cur.close()
|
| 48 |
+
local_conn.close()
|
| 49 |
+
supa_cur.close()
|
| 50 |
+
supa_conn.close()
|
| 51 |
+
print(f"Migration complete. {len(papers)} papers migrated to Supabase.")
|