mygpo/data/management/commands/merge-duplicates.py

   1 from django.core.management.base import BaseCommand
   2 from mygpo.api.models import Podcast, Episode
   3 from mygpo.api.sanitizing import rewrite_podcasts, rewrite_episodes
   4
   5 class Command(BaseCommand):
   6
   7     def handle(self, *args, **options):
   8
   9         if len(args) == 0:
  10             podcast_urls = []
  11             for p in Podcast.objects.only('id', 'url'):
  12                 if Podcast.objects.filter(url__exact=p.url).exclude(id=p.id).exists():
  13                     print 'found podcast url %s' % p.url
  14                     podcast_urls.append(p.url)
  15
  16             episode_ids = []
  17             for e in Episode.objects.only('id', 'url'):
  18                 if Episode.objects.filter(url__exact=e.url, podcast=e.podcast).exclude(id=e.id).exists():
  19                     print 'found episode url %s' % e.url
  20                     episode_ids.append(e.id)
  21
  22         else:
  23             pocast_urls = args
  24
  25         for url in podcast_urls:
  26             p = Podcast.objects.filter(url=url).order_by('id')[0]
  27             for p2 in Podcast.objects.filter(url=url).exclude(id=p.id).order_by('id'):
  28                 print 'merging podcast %s with %s' % (p.id, p2.id)
  29                 rewrite_podcasts(p2, p)
  30                 p2.delete()
  31
  32         for e_id in episode_ids:
  33             e = Episode.objects.get(id=e_id)
  34             for e2 in Episode.objects.filter(url=e.url, podcast=e.podcast).exlcude(id=e.id).order_by('id'):
  35                 print 'merging episode %s with %s' % (e.id, e2.id)
  36                 rewrite_episodes(e2, e)
  37                 e2.delete()
  38