def _do_ip_test(self, address):
return ip.is_IP(address)
def _do_ip_test(self, address):
return ip.is_IP(address)
def find_authors(self, page):
"""
Retrieve main authors of given page.
@note: userPut() sets current_page therefore we cannot use it.
@param page: Page object to retrieve main authors
@type page: pywikibot.Page
@return: yield tuple of user name and edit quantity
@rtype: generator
"""
percent = 0
if page.namespace() == pywikibot.site.Namespace.MAIN:
url = ('https://tools.wmflabs.org/wikihistory/dewiki/'
'getauthors.php?page_id={0}'.format(page.pageid))
try:
r = fetch(url)
except requests.exceptions.ConnectionError:
pywikibot.exception()
else:
if r.status not in (200, ):
pywikibot.warning('wikihistory request status is %d'
% r.status)
else:
pattern = r'>(?P<author>.+?)</a>\s\((?P<percent>\d{1,3})&'
for main, main_cnt in re.findall(pattern,
r.decode('utf-8')):
main_cnt = int(main_cnt)
percent += main_cnt
if ' weitere' in main:
break
yield main, main_cnt
if percent > 50:
break
if percent != 0:
return
# A timeout occured or not main namespace, calculate it yourself
pywikibot.output('No wikihistory data available for %s.\n'
'Retrieving revisions.' % page)
cnt = Counter()
for rev in page.revisions():
if is_IP(rev.user):
continue
if rev.minor:
cnt[rev.user] += 0.2
else:
cnt[rev.user] += 1
s = sum(cnt.values())
s2 = sum(i ** 2 for i in cnt.values())
n = float(max(len(cnt), 1))
x_ = s / n
# avg + stdabw
limit = max(3, (s2 / n - x_ ** 2) ** 0.5 * 1.5 + x_)
for main, main_cnt in cnt.most_common(7):
if main_cnt < limit:
break
yield main, main_cnt * 100 / s
