NewsBlur-viq/apps/rss_feeds/management/commands/merge_feeds.py

46 lines
2.3 KiB
Python
Raw Normal View History

2010-06-27 22:40:22 -04:00
from django.core.management.base import BaseCommand
from apps.rss_feeds.models import merge_feeds, MStory
2010-06-27 22:40:22 -04:00
from optparse import make_option
from django.db import connection
2010-06-27 22:40:22 -04:00
class Command(BaseCommand):
option_list = BaseCommand.option_list + (
make_option("-f", "--feed", dest="feed", default=None),
make_option("-V", "--verbose", dest="verbose", action="store_true"),
)
def handle(self, *args, **options):
cursor = connection.cursor()
cursor.execute("""SELECT DISTINCT f.id AS original_id, f2.id AS duplicate_id,
f.feed_address AS original_feed_address,
f2.feed_address AS duplicate_feed_address,
f.feed_title AS original_feed_title,
f2.feed_title AS duplicate_feed_title,
f.feed_link AS original_feed_link,
f2.feed_link AS duplicate_feed_link,
f2.feed_tagline AS original_feed_tagline,
f.feed_tagline AS duplicate_feed_tagline
FROM feeds f, feeds f2
WHERE f2.id > f.id
AND f.feed_tagline = f2.feed_tagline
AND f.feed_link = f2.feed_link
AND f.feed_title = f2.feed_title
ORDER BY original_id ASC;""")
feed_fields = ('original_id', 'duplicate_id', 'original_feed_address', 'duplicate_feed_address')
for feeds_values in cursor.fetchall():
feeds = dict(zip(feed_fields, feeds_values))
original_stories = MStory.objects(story_feed_id=int(feeds['original_id'])).only('story_guid')[10:13]
print original_stories
original_story_ids = [story.id for story in original_stories]
print original_story_ids
duplicate_stories = MStory.objects(story_feed_id=int(feeds['duplicate_id']), story_guid__in=original_story_ids)
print duplicate_stories
if duplicate_stories.count() != original_stories.count():
print "Skipping: %s" % (feeds)
continue
else:
print "Merging: %s" % feeds
# merge_feeds(feeds['original_id'], feeds['duplicate_id'])