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)
def update_project_assessments_by_kind(wikidb, wp10db, project,
extra_assessments, kind, old_ratings,
seen):
if kind not in (AssessmentKind.QUALITY, AssessmentKind.IMPORTANCE):
raise ValueError(
'Parameter "kind" was not one of QUALITY or IMPORTANCE')
logger.info('Updating project %s assessments for %s', kind,
project.p_project)
rating_to_category = update_project_categories_by_kind(
wikidb, wp10db, project, extra_assessments, kind)
n = 0
new_ratings = defaultdict(list)
for current_rating, (category, ranking) in rating_to_category.items():
logger.info('Fetching article list for %r' % category.decode('utf-8'))
current_rating = current_rating.encode('utf-8')
for page in logic_page.get_pages_by_category(wikidb, category):
# Talk pages are tagged, we want the NS of the article itself.
namespace = page.page_namespace - 1
if not logic_util.is_namespace_acceptable(namespace):
logger.debug('Skipping %s with namespace=%s', page.page_title,
namespace)
continue
article_ref = str(namespace).encode(
'utf-8') + b':' + page.page_title
seen.add(article_ref)
old_rating = old_ratings.get(article_ref)
old_rating_value = None
if old_rating:
rating = Rating(**attr.asdict(old_rating))
if kind == AssessmentKind.QUALITY:
old_rating_value = rating.r_quality
elif kind == AssessmentKind.IMPORTANCE:
old_rating_value = rating.r_importance
else:
rating = Rating(r_project=project.p_project,
r_namespace=namespace,
r_article=page.page_title,
r_score=0)
old_rating_value = NOT_A_CLASS.encode('utf-8')
if kind == AssessmentKind.QUALITY:
rating.r_quality = current_rating
rating.set_quality_timestamp_dt(page.cl_timestamp)
elif kind == AssessmentKind.IMPORTANCE:
rating.r_importance = current_rating
rating.set_importance_timestamp_dt(page.cl_timestamp)
new_ratings[article_ref].append((rating, kind, old_rating_value))
n += 1
if n >= MAX_ARTICLES_BEFORE_COMMIT:
wp10db.ping()
wp10db.commit()
logger.info('End, committing db')
wp10db.ping()
wp10db.commit()
return (new_ratings, rating_to_category)