def test_get_most_recent_move_from_api(self, patched_site):
patched_site.logevents.side_effect = lambda *args, **kwargs: self.le_multi
move_data = logic_page.get_move_data(self.wp10db, self.wikidb, 0,
b'Some Moved Article',
datetime(1970, 1, 1))
self.assertIsNotNone(move_data)
self.assertEqual(self.expected_ns, move_data['dest_ns'])
self.assertEqual(self.expected_title.encode('utf-8'),
move_data['dest_title'])
self.assertEqual(self.expected_dt, move_data['timestamp_dt'])
def test_get_single_move_too_old_from_api(self, patched_site):
patched_site.logevents.side_effect = lambda *args, **kwargs: self.le_return
move_data = logic_page.get_move_data(self.wp10db, self.wikidb, 0,
b'Some Moved Article',
datetime(2014, 1, 1))
self.assertIsNone(move_data)
def test_no_redirect_no_move(self, unused_patched_site):
move_data = logic_page.get_move_data(self.wp10db, self.wikidb, 0,
b'Some Moved Article',
datetime(1970, 1, 1))
self.assertIsNone(move_data)
def process_unseen_articles(wikidb, wp10db, project, old_ratings, seen):
denom = len(old_ratings.keys())
ratio = len(seen) / denom if denom != 0 else 'NaN'
logger.debug('Looking for unseen articles, ratio was: %s', ratio)
in_seen = 0
skipped = 0
processed = 0
n = 0
for ref, old_rating in old_ratings.items():
if ref in seen:
in_seen += 1
continue
# By default, we evaluate both assessment kinds.
kind = AssessmentKind.BOTH
if old_rating.r_quality == NOT_A_CLASS or old_rating.r_quality is None:
# The quality rating is not set, so just evaluate importance
kind = AssessmentKind.IMPORTANCE
if (old_rating.r_importance == NOT_A_CLASS
or old_rating.r_importance is None):
# The importance rating is also not set, so don't do anything.
skipped += 1
continue
logger.debug('Processing unseen article %s', ref.decode('utf-8'))
processed += 1
ns, title = ref.decode('utf-8').split(':', 1)
ns = int(ns.encode('utf-8'))
title = title.encode('utf-8')
move_data = logic_page.get_move_data(wp10db, wikidb, ns, title,
project.timestamp_dt)
if move_data is not None:
logic_page.update_page_moved(wp10db, project, ns, title,
move_data['dest_ns'],
move_data['dest_title'],
move_data['timestamp_dt'])
# Mark this article as having NOT_A_CLASS for it's quality or importance.
# This probably means the article was deleted, but could in fact mean that
# we just failed to find its move data. Either way, the new article would
# have already been picked up by the assessment updater, assuming it was
# tagged correctly.
rating = Rating(r_project=project.p_project,
r_namespace=ns,
r_article=title,
r_score=0)
if kind in (AssessmentKind.QUALITY, AssessmentKind.BOTH):
rating.quality = NOT_A_CLASS.encode('utf-8')
if move_data:
rating.set_quality_timestamp_dt(move_data['timestamp_dt'])
else:
rating.r_quality_timestamp = GLOBAL_TIMESTAMP_WIKI
if kind in (AssessmentKind.IMPORTANCE, AssessmentKind.BOTH):
rating.importance = NOT_A_CLASS.encode('utf-8')
if move_data:
rating.set_importance_timestamp_dt(move_data['timestamp_dt'])
else:
rating.r_importance_timestamp = GLOBAL_TIMESTAMP_WIKI
logic_rating.insert_or_update(wp10db, rating, kind)
if kind in (AssessmentKind.QUALITY, AssessmentKind.BOTH):
logic_rating.add_log_for_rating(wp10db, rating,
AssessmentKind.QUALITY,
old_rating.r_quality)
if kind in (AssessmentKind.IMPORTANCE, AssessmentKind.BOTH):
logic_rating.add_log_for_rating(wp10db, rating,
AssessmentKind.IMPORTANCE,
old_rating.r_importance)
n += 1
if n >= MAX_ARTICLES_BEFORE_COMMIT:
wp10db.ping()
wp10db.commit()
logger.info('End, committing db')
wp10db.ping()
wp10db.commit()
logger.debug('SEEN REPORT:\nin seen: %s\nskipped: %s\nprocessed: %s',
in_seen, skipped, processed)