add script to convert Online-IDE PDFs into plaintext

1 files changed, 39 insertions, 0 deletions

diff --git a/scripts/oideprint2txt b/scripts/oideprint2txt
new file mode 100755
index 00000000..7f7f6d2f
--- /dev/null
+++ b/scripts/oideprint2txt
@@ -0,0 +1,39 @@
+#!/usr/bin/env python3
+
+import argparse
+import json
+import sys
+import csv
+import subprocess
+
+parser = argparse.ArgumentParser()
+parser.add_argument('input')
+parser.add_argument('-t', default='json', choices=['csv', 'json'])
+
+args = parser.parse_args()
+
+data = subprocess.check_output(['pdftotext', args.input, '-'], text=True).splitlines()
+
+users = []
+
+user = {}
+i = 0
+while i < len(data):
+    if not data[i].startswith('URL'):
+        i = i + 1
+        continue
+
+    user = {
+        'Name': (data[i+1].split(': ')[1]).strip(),
+        'Klasse': (data[i+2].split(': ')[1]).strip(),
+        'Benutzername': (data[i+3].split(': ')[1]).strip(),
+        'Passwort': (data[i+4].split(': ')[1]).strip(),
+    }
+    users.append(user)
+    i = i + 5
+
+if args.t == 'json':
+    json.dump(users, sys.stdout)
+elif args.t == 'csv':
+    w = csv.writer(sys.stdout)
+    w.writerows([user.values() for user in users])