def __call__(self, user_record):
total = 0
start = get_date(self.start, user_record)
end = get_date(self.end, user_record)
if not start or not end:
user_record[self.id] = None
return user_record
for i , contrib in enumerate(user_record['_contribs']):
if (contrib['ns'] not in self.ns or
contrib['timestamp'] < start or
contrib['timestamp'] > end):
continue
logging.debug('processing user %s contrib %d/%d' %
(user_record['username'], i, len(user_record['_contribs'])))
curr_rev = user_record['_rev_cache'][contrib['revid']]
end_rev = self.get_last_revision(user_record, contrib['title'], start, end)
try:
curr_content = curr_rev['*']
except (KeyError, TypeError):
logging.warning('curr rev does not contain the content field: %s', curr_rev)
curr_content = ''
try:
end_content = end_rev['*']
except (KeyError, TypeError):
logging.warning('end rev does not contain the content field: %s', end_rev)
end_content = ''
diff_with_end = get_diff(curr_content, end_content)
overlap_ranges = self.get_overlaps(contrib['diff'], diff_with_end)
contrib_total = reduce(lambda cum, (start,end) : cum + end - start, overlap_ranges, 0)
total += contrib_total
user_record[self.id] = total
return user_record
def __call__(self, user_record):
count = 0
start = get_date(self.start, user_record)
end = get_date(self.end, user_record)
if not (start and end):
# this means that this is not a valid date range for this user
user_record[self.id] = 'NA'
return user_record
for contrib in user_record['_contribs']:
if contrib['timestamp'] < start:
continue
if contrib['timestamp'] > end:
continue
if contrib['ns'] not in self.ns:
continue
count += 1
user_record[self.id] = count
return user_record
def __call__(self, user_record):
pages = []
start = get_date(self.start, user_record)
end = get_date(self.end, user_record)
if not (start and end):
# this means that this is not a valid date range for this user
user_record[self.id] = 'NA'
return user_record
for contrib in user_record['_contribs']:
if contrib['timestamp'] < start:
continue
if contrib['timestamp'] > end:
continue
if contrib['ns'] not in self.ns:
continue
pages.append(contrib['pageid'])
user_record[self.id] = len(set(pages))
return user_record
