From 08f39e0d22a444edcefafb18f512217887ea268d Mon Sep 17 00:00:00 2001
From: eclipse <eclipse@unterdemradar.de>
Date: Tue, 6 May 2025 12:19:12 +0200
Subject: [PATCH] this script corrects wrongly formatted dates in the DB table
 "Werk"; needs only to be run once, and only if there are bad legacy dates in
 the DB

---
 utils/format_dates.py | 74 +++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 74 insertions(+)
 create mode 100755 utils/format_dates.py

diff --git a/utils/format_dates.py b/utils/format_dates.py
new file mode 100755
index 0000000..1510108
--- /dev/null
+++ b/utils/format_dates.py
@@ -0,0 +1,74 @@
+#!/usr/bin/env python3
+# -*- coding: utf8 -*-
+
+from flask import Flask
+from dotenv import load_dotenv
+import os
+from flask_sqlalchemy import SQLAlchemy
+from sqlalchemy import select
+import re
+
+# initialize db exactly like in the main app
+app = Flask(__name__)
+load_dotenv()
+app.config.from_prefixed_env()
+if os.getenv("SQLALCHEMY_DATABASE_DIALECT") == "sqlite":
+  app.config['SQLALCHEMY_DATABASE_URI'] = "sqlite:///" + os.path.abspath(app.root_path + "/..") + "/" + os.getenv("SQLALCHEMY_DATABASE_SQLITE_FILENAME")
+else:
+  exit("no SQLite database URI given; exiting")
+db = SQLAlchemy()
+db.init_app(app)
+
+with app.app_context():
+  db.reflect()
+
+  class Werk(db.Model):
+    __table__ = db.Model.metadata.tables['Werk']
+
+  # iterate over ids
+  for id in range(1, 54):
+    werk = db.session.get(Werk, id)
+    if werk is None:
+      print(f"ID {id}: Werk does not exist") 
+      continue
+
+    # case: None/NULL
+    if werk.Erscheinungsdatum is None:
+      print(f"ID {id}: date is None; was normalized to '{werk.Erscheinungsdatum}'")
+      continue
+
+    # case: empty string
+    if werk.Erscheinungsdatum == "":
+      werk.Erscheinungsdatum = None
+      print(f"ID {id}: date is empty string; was normalized to '{werk.Erscheinungsdatum}'")
+      continue
+
+    # case: YYYY
+    if len(werk.Erscheinungsdatum) == 4:
+      print(f"ID {id}: date matches format YYYY; was normalized to '{werk.Erscheinungsdatum}'")
+      continue
+
+    # case: YYYY-MM-DD
+    result = re.search(r"^(\d\d\d\d)-(\d\d)-(\d\d)$", werk.Erscheinungsdatum)
+    if result is not None:
+      print(f"ID {id}: date matches format YYYY-MM-DD; was normalized to '{werk.Erscheinungsdatum}'")
+      continue
+
+    # case: DD.MM.YYYY
+    result = re.search(r"^(\d\d)\.(\d\d)\.(\d\d\d\d)$", werk.Erscheinungsdatum)
+    if result is not None:
+      werk.Erscheinungsdatum = result.group(3) + "-" + result.group(2) + "-" + result.group(1)
+      print(f"ID {id}: date matches format DD.MM.YYYY; was normalized to '{werk.Erscheinungsdatum}'")
+      continue
+
+    # case: DD.M.YYYY
+    result = re.search(r"^(\d\d)\.(\d)\.(\d\d\d\d)$", werk.Erscheinungsdatum)
+    if result is not None:
+      werk.Erscheinungsdatum = result.group(3) + "-0" + result.group(2) + "-" + result.group(1)
+      print(f"ID {id}: date matches format DD.M.YYYY; was normalized to '{werk.Erscheinungsdatum}'")
+      continue
+
+    print(f"ID {id}: date did not match any format")
+
+  db.session.commit()
+