From: Marek Stępniowski <marek@stepniowski.com>
Date: Mon, 22 Sep 2008 15:39:28 +0000 (+0200)
Subject: Added script to remove duplicated fragments.
X-Git-Url: https://git.mdrn.pl/wolnelektury.git/commitdiff_plain/4c93aaa7c94e4475ac6efbfeefbd747f1467b6ff

Added script to remove duplicated fragments.
---

diff --git a/remove_duplicates.py b/remove_duplicates.py
new file mode 100755
index 000000000..e72512ba4
--- /dev/null
+++ b/remove_duplicates.py
@@ -0,0 +1,31 @@
+#!/usr/bin/env python
+import sys
+
+from django.core.management import setup_environ
+from wolnelektury import settings
+try:
+    set
+except AttributeError:
+    from set import Set as set
+
+setup_environ(settings)
+
+from catalogue import models
+
+fragment_identifiers = set()
+
+print
+print 'Before: %d fragments' % models.Fragment.objects.count()
+print
+
+for fragment in models.Fragment.objects.all():
+    if (fragment.book_id, fragment.anchor) in fragment_identifiers:
+        fragment.delete()
+        sys.stderr.write('X')
+    else:
+        fragment_identifiers.add((fragment.book_id, fragment.anchor))
+        sys.stderr.write('.')
+
+print
+print 'After: %d fragments' % models.Fragment.objects.count()
+print
\ No newline at end of file