def mixed_rss_with_report(self, user_id, id):
c.rss_user = meta.find(model.User, user_id)
log.debug("c.rss_user: %s" % c.rss_user)
feed_data = meta.find(model.Feed, id)
log.debug("feed_data.id %s" % feed_data.id)
import feed
fetch_result = feed_data.fetch()
feed = h.DefaultFeed(
title=feed_data.title,
link=feed_data.link,
description="TESTING",
language=feed_data.language,
)
c.base_url = config['base_url']
log.debug('c.base_url: %s' % c.base_url)
guesser = Guesser(feed_data, c.rss_user, config)
settings = c.rss_user.get_bayes_feed_setting(feed_data.id)
meta.Session.add(settings)
delta = h.timedelta_from_string(settings.summarize_at)
log.debug("delta %s" % delta)
if not settings.report_offset:
entries = feed_data.get_entries().order_by(model.FeedEntry.id).all()
log.warn("no report_offset available, read %s entries" % len(entries))
else:
entries = feed_data.get_entries().filter(model.FeedEntry.id >= settings.report_offset).order_by(model.FeedEntry.id).all()
from rssmonster.lib.reporter import Reporter
reporter = Reporter(None, None, delta, 30)
for entry in entries:
reporter.add_item(entry, guesser.is_spam(entry))
for entry_box in reporter.entry_queue:
#log.debug("entry_box: %s" % entry_box)
if entry_box['type'] == 'ham':
c.entry = entry_box['entry']
c.entry.is_spam=guesser.is_spam(c.entry)
feed.add_item(title=c.entry.title,
link=c.entry.link,
description=render('bayes/rss_summary.mako'),
unique_id=c.entry.uid,
pubdate=c.entry.updated)
elif entry_box['type'] == 'spam':
add_spam_report(feed, entry_box['entries'])
settings.report_offset = reporter.offset_id()
log.debug("settings.report_offset: %s" % settings.report_offset)
log.debug("holding back: %s" % len(reporter.spam_entries))
meta.Session.commit()
return feed.writeString('utf-8')
class TestReporterQueued(TestController):
def setUp(self):
self.reporter = Reporter(None, None, timedelta(minutes=10), max_entries=5)
def test_ham_only(self):
self.assertEqual(None, self.reporter.offset_id())
self.reporter.add_item(create_mock({'updated':datetime(2009,11,14, 12,12), 'id':1}), False)
self.assertEqual(1, self.reporter.offset_id())
self.reporter.add_item(create_mock({'updated':datetime(2009,11,14, 12,12), 'id':2}), False)
self.assertEqual(1, self.reporter.offset_id())
self.reporter.add_item(create_mock({'updated':datetime(2009,11,14, 12,12), 'id':3}), False)
self.assertEqual(1, self.reporter.offset_id())
self.reporter.add_item(create_mock({'updated':datetime(2009,11,14, 12,12), 'id':4}), False)
self.assertEqual(1, self.reporter.offset_id())
self.reporter.add_item(create_mock({'updated':datetime(2009,11,14, 12,12), 'id':5}), False)
self.assertEqual(1, self.reporter.offset_id())
self.reporter.add_item(create_mock({'updated':datetime(2009,11,14, 12,12), 'id':6}), False)
self.assertEqual(2, self.reporter.offset_id())
self.reporter.add_item(create_mock({'updated':datetime(2009,11,14, 12,12), 'id':7}), False)
self.assertEqual(3, self.reporter.offset_id())
self.reporter.add_item(create_mock({'updated':datetime(2009,11,14, 12,12), 'id':8}), False)
self.assertEqual(4, self.reporter.offset_id())
self.reporter.add_item(create_mock({'updated':datetime(2009,11,14, 12,12), 'id':9}), False)
self.assertEqual(5, self.reporter.offset_id())
self.reporter.add_item(create_mock({'updated':datetime(2009,11,14, 12,12), 'id':10}), False)
self.assertEqual(6, self.reporter.offset_id())
self.reporter.add_item(create_mock({'updated':datetime(2009,11,14, 12,12), 'id':11}), False)
self.assertEqual(7, self.reporter.offset_id())
def test_with_spam(self):
self.assertEqual(None, self.reporter.offset_id())
self.reporter.add_item(create_mock({'updated':datetime(2009,11,14, 12,01), 'id':1}), True)
self.assertEqual(1, self.reporter.offset_id())
self.assertEqual(0, len(self.reporter.entry_queue))
self.assertEqual(1, self.reporter.spam_entries[0].id)
self.reporter.add_item(create_mock({'updated':datetime(2009,11,14, 12,02), 'id':2}), False)
self.assertEqual(1, self.reporter.offset_id())
self.assertEqual(1, len(self.reporter.entry_queue))
self.assertEqual('ham', self.reporter.entry_queue[0]['type'])
self.assertEqual(2, self.reporter.entry_queue[0]['entry'].id)
self.reporter.add_item(create_mock({'updated':datetime(2009,11,14, 12,03), 'id':3}), True)
self.assertEqual(1, self.reporter.offset_id())
self.assertEqual(1, len(self.reporter.entry_queue))
self.assertEqual(1, self.reporter.spam_entries[0].id)
self.assertEqual(3, self.reporter.spam_entries[1].id)
self.reporter.add_item(create_mock({'updated':datetime(2009,11,14, 12,4), 'id':4}), False)
self.assertEqual(1, self.reporter.offset_id())
self.assertEqual(2, len(self.reporter.entry_queue))
self.assertEqual('ham', self.reporter.entry_queue[1]['type'])
self.assertEqual(2, self.reporter.entry_queue[0]['entry'].id)
self.assertEqual(4, self.reporter.entry_queue[1]['entry'].id)
self.reporter.add_item(create_mock({'updated':datetime(2009,11,14, 12,5), 'id':5}), False)
self.assertEqual(1, self.reporter.offset_id())
self.assertEqual(2, self.reporter.entry_queue[0]['entry'].id)
self.assertEqual(4, self.reporter.entry_queue[1]['entry'].id)
self.assertEqual(5, self.reporter.entry_queue[2]['entry'].id)
self.reporter.add_item(create_mock({'updated':datetime(2009,11,14, 12,6), 'id':6}), False)
self.assertEqual(1, self.reporter.offset_id())
self.assertEqual(2, self.reporter.entry_queue[0]['entry'].id)
self.assertEqual(4, self.reporter.entry_queue[1]['entry'].id)
self.assertEqual(5, self.reporter.entry_queue[2]['entry'].id)
self.assertEqual(6, self.reporter.entry_queue[3]['entry'].id)
self.reporter.add_item(create_mock({'updated':datetime(2009,11,14, 12,7), 'id':7}), False)
self.assertEqual(1, self.reporter.offset_id())
self.assertEqual(2, self.reporter.entry_queue[0]['entry'].id)
self.assertEqual(4, self.reporter.entry_queue[1]['entry'].id)
self.assertEqual(5, self.reporter.entry_queue[2]['entry'].id)
self.assertEqual(6, self.reporter.entry_queue[3]['entry'].id)
self.assertEqual(7, self.reporter.entry_queue[4]['entry'].id)
self.reporter.add_item(create_mock({'updated':datetime(2009,11,14, 12,8), 'id':8}), False)
self.assertEqual(1, self.reporter.offset_id())
self.assertEqual(4, self.reporter.entry_queue[0]['entry'].id)
self.assertEqual(5, self.reporter.entry_queue[1]['entry'].id)
self.assertEqual(6, self.reporter.entry_queue[2]['entry'].id)
self.assertEqual(7, self.reporter.entry_queue[3]['entry'].id)
self.assertEqual(8, self.reporter.entry_queue[4]['entry'].id)
self.assertEqual(datetime(2009,11,14, 12,01),
self.reporter.last_report)
#next will trigger report
self.reporter.add_item(create_mock({'updated':datetime(2009,11,14, 12,12), 'id':9}), False)
self.assertEqual(1, self.reporter.offset_id())
self.assertEqual(6, self.reporter.entry_queue[0]['entry'].id)
self.assertEqual(7, self.reporter.entry_queue[1]['entry'].id)
self.assertEqual(8, self.reporter.entry_queue[2]['entry'].id)
self.assertEqual('spam', self.reporter.entry_queue[3]['type'])
self.assertEqual(1, self.reporter.entry_queue[3]['entries'][0].id)
self.assertEqual(3, self.reporter.entry_queue[3]['entries'][1].id)
self.assertEqual(9, self.reporter.entry_queue[4]['entry'].id)
self.reporter.add_item(create_mock({'updated':datetime(2009,11,14, 12,8), 'id':10}), False)
self.assertEqual(1, self.reporter.offset_id())
self.assertEqual('spam', self.reporter.entry_queue[2]['type'])
self.reporter.add_item(create_mock({'updated':datetime(2009,11,14, 12,8), 'id':11}), False)
self.assertEqual(1, self.reporter.offset_id())
self.assertEqual('spam', self.reporter.entry_queue[1]['type'])
self.reporter.add_item(create_mock({'updated':datetime(2009,11,14, 12,8), 'id':12}), False)
self.assertEqual(1, self.reporter.offset_id())
self.assertEqual('spam', self.reporter.entry_queue[0]['type'])
# this popped the spam report entries offset is > 1
self.reporter.add_item(create_mock({'updated':datetime(2009,11,14, 12,8), 'id':13}), False)
self.assertEqual('ham', self.reporter.entry_queue[1]['type'])
self.assertEqual(9, self.reporter.entry_queue[0]['entry'].id)
log.debug("offset_queue: %s" % self.reporter.offset_queue[0].id)
self.assertEqual(9, self.reporter.offset_id())
def test_spam_only(self):
self.reporter.add_item(create_mock({'updated':datetime(2009,11,14, 12,11), 'id':10}), True)
self.reporter.add_item(create_mock({'updated':datetime(2009,11,14, 12,12), 'id':11}), True)
self.reporter.add_item(create_mock({'updated':datetime(2009,11,14, 12,22), 'id':20}), True)
self.reporter.add_item(create_mock({'updated':datetime(2009,11,14, 12,33), 'id':30}), True)
self.reporter.add_item(create_mock({'updated':datetime(2009,11,14, 12,44), 'id':40}), True)
self.reporter.add_item(create_mock({'updated':datetime(2009,11,14, 12,55), 'id':50}), True)
self.reporter.add_item(create_mock({'updated':datetime(2009,11,14, 13,06), 'id':60}), True)
self.assertEqual(10, self.reporter.entry_queue[0]['entries'][0].id)
self.assertEqual(11, self.reporter.entry_queue[0]['entries'][1].id)
self.assertEqual(20, self.reporter.entry_queue[1]['entries'][0].id)
self.assertEqual(30, self.reporter.entry_queue[2]['entries'][0].id)
self.assertEqual(40, self.reporter.entry_queue[3]['entries'][0].id)
self.assertEqual(50, self.reporter.entry_queue[4]['entries'][0].id)