X-Git-Url: https://git.mdrn.pl/wolnelektury.git/blobdiff_plain/1a5d9a134e88915b5db447924f532fd63aaeb45e..e8390b9f10338ff5ced700f5dbd87ba30a3566bd:/scripts/remove_duplicates.py diff --git a/scripts/remove_duplicates.py b/scripts/remove_duplicates.py new file mode 100755 index 000000000..e72512ba4 --- /dev/null +++ b/scripts/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