From 2f7d06680882b340565631fbfd408cbaa9c0b3e6 Mon Sep 17 00:00:00 2001
From: Bo-Chun Louis Chen <louistw@uab.edu>
Date: Wed, 27 Mar 2024 11:15:09 -0500
Subject: [PATCH] Add data migration script for created and origin column
---
migrations/20240327_add_created_table.py | 50 ++++++++++++++++++++++++
1 file changed, 50 insertions(+)
create mode 100644 migrations/20240327_add_created_table.py
diff --git a/migrations/20240327_add_created_table.py b/migrations/20240327_add_created_table.py
new file mode 100644
index 0000000..f9b6fa0
--- /dev/null
+++ b/migrations/20240327_add_created_table.py
@@ -0,0 +1,50 @@
+import csv
+import dataset
+from datetime import datetime
+import rabbit_config as rcfg
+
+
+def upgrade():
+ db_path = rcfg.db_path
+ db_name = "user_reg.db"
+ csv_filename = "./migrations/account-registration-history.csv"
+
+ db = dataset.connect(f"sqlite:///{db_path}/{db_name}")
+ table = db["users"]
+
+ # Get all username and last_update
+ usernames = db.query("SELECT username, last_update FROM users")
+ users = {}
+ for u in usernames:
+ users[u["username"]] = dict(u)
+
+ # Get created from csv file
+ with open(csv_filename) as f:
+ reader = csv.reader(f, delimiter=",")
+ # Skip header
+ next(reader)
+ for row in reader:
+ if row[0] in users:
+ users[row[0]]["created"] = datetime.fromisoformat(row[1][:-3])
+ users[row[0]]["origin"] = row[2]
+
+ # Add copy created column from last_update if they are not in csv file
+ for username in users:
+ if "created" not in users[username]:
+ users[username]["created"] = datetime.fromisoformat(
+ users[username]["last_update"]
+ )
+ users[username]["origin"] = "last_update"
+ # No need to update this column back to the db
+ del users[username]["last_update"]
+
+ # Create new columns
+ table.create_column("created", db.types.datetime)
+ table.create_column("origin", db.types.text)
+
+ # Update database in chunk
+ table.update_many(list(users.values()), ["username"], ensure=True)
+
+
+if __name__ == "__main__":
+ upgrade()
--
GitLab