def test_RC_pagegenerator_result(self):
import logging
lgr = logging.getLogger('socketIO_client')
lgr.setLevel(logging.DEBUG)
ch = logging.StreamHandler()
ch.setLevel(logging.DEBUG)
lgr.addHandler(ch)
site = self.get_site()
pagegenerator = pagegenerators.LiveRCPageGenerator(site,
total=self.length)
entries = list(pagegenerator)
self.assertEqual(len(entries), self.length)
testentry = entries[0]
self.assertEqual(testentry.site, site)
self.assertTrue(hasattr(testentry, '_rcinfo'))
rcinfo = testentry._rcinfo
self.assertEqual(rcinfo['server_name'], site.hostname())
self.assertEqual(rcinfo['wiki'], site.dbName())
for key in [
"id", "type", "namespace", "title", "comment", "timestamp",
"user", "bot"
]:
self.assertIn(key, rcinfo.keys())
def main(*args):
"""
Procesa los parámetros desde la línea de comandos
If args is an empty list, sys.argv is used.
@param args: command line arguments
@type args: list of unicode
"""
opts = {}
local_args = pywikibot.handle_args(args)
for arg in local_args:
if arg.startswith('-gf:'):
opts['gf'] = float(arg[4:])
elif arg.startswith('-dm:'):
opts['dm'] = float(arg[4:])
elif arg.startswith('-wiki:'):
opts['wiki'] = arg[6:]
elif arg.startswith('--debug'):
opts['debug'] = True
site = pywikibot.Site()
if 'wiki' in opts and opts['wiki'] != 'eswiki':
lang = opts['wiki'][0:2]
family = opts['wiki'][2:]
site = pywikibot.Site(lang, family)
bot = SeroBOT(pagegenerators.LiveRCPageGenerator(site), site=site, **opts)
bot.run()
def run(self):
global MIN_SIZE
self.generator = []
log('Starting live bot')
filter_gen = lambda gen: [p for p in gen if self.page_filter(p)]
if self.use_stream:
live_gen = pagegenerators.LiveRCPageGenerator(self.site)
live_gen = filter_gen(live_gen)
else:
from IRCRCListener import irc_rc_listener
live_gen = (p for p in irc_rc_listener(self.site, filter_gen))
pending_checks = []
reconnect_index = 500
try:
for page in live_gen:
rcinfo = page._rcinfo
#log('Adding page:' + page.title())
# TODO: remove rolledback edits from generator
log('Page in buffer: {}'.format(len(pending_checks)))
pending_checks.append(
(page, rcinfo['revision']['new'], rcinfo['revision']['old']
or 0))
if len(pending_checks
) < self.rcthreshold or not self.uploads_ready():
continue # move to next edit if not enough edits accomulated
# handle uploads or send new changes to process
if len(self.uploads) > 0:
pywikibot.output('reporting uploads')
self.report_uploads() # report checked edits
print('reported')
self.uploads = []
reconnect_index -= 1
if reconnect_index == 0:
pywikibot.output('Reconnect after many uploads')
self._init_server()
reconnect_index = 500
else:
pywikibot.output('checking pending')
self.generator = pending_checks
log('checking pending')
self.process_changes()
pending_checks = []
except KeyboardInterrupt:
pywikibot.output('handling uploaded changes')
while not self.uploads_ready():
continue
# handle uploads or send new changes to process
if len(self.uploads) > 0:
self.report_uploads() # report checked edits
def main(*args):
"""
Procesa los parámetros desde la línea de comandos
If args is an empty list, sys.argv is used.
@param args: command line arguments
@type args: list of unicode
"""
opts = {}
local_args = pywikibot.handle_args(args)
for arg in local_args:
if arg.startswith('-gf:'):
opts['gf'] = float(arg[4:])
elif arg.startswith('-dm:'):
opts['dm'] = float(arg[4:])
elif arg.startswith('--debug'):
opts['debug'] = True
bot = SeroBOT(pagegenerators.LiveRCPageGenerator(),
site=pywikibot.Site(), **opts)
bot.run()
prediccion = goodfaith.get('score').get('prediction')
probabilidad = goodfaith.get('score').get('probability').get('true')
d_prediccion = damaging.get('score').get('prediction')
d_probabilidad = damaging.get('score').get('probability').get('true')
retornar[rev] = {'buena_fe': prediccion, 'prob': probabilidad, 'danina': d_prediccion, 'prob_d': d_probabilidad}
return retornar
def score(rev):
return scores([rev]).get(rev)
Site = pywikibot.Site()
if Site.logged_in() == False:
Site.login()
for page in pagegenerators.LiveRCPageGenerator(site=Site):
if page._rcinfo.get('type') == 'edit' and page._rcinfo.get('namespace') in [0, 100] and page._rcinfo.get('user') != Site.username():
revision = page._rcinfo.get('revision')
ores = score(str(revision.get('new')))
print ('>' if ores.get('prob') < 0.095 else ' ', revision.get('new'), ores.get('prob'), ores.get('prob_d'), page.title(), page._rcinfo.get('user'))
if ores.get('prob') < 0.095 or ores.get('prob_d') > 0.97 or (ores.get('prob') < 0.13 and ores.get('prob_d') > 0.95):
old = page.text
page.text = page.getOldVersion(revision.get('old'))
pywikibot.showDiff(old, page.text)
try:
#page.save(u'BOT - Reversión de página ([[:mw:ORES|ORES]]: '+ str(ores.get('prob'))+')')
token = pywikibot.data.api.Request(site=Site, parameters={'action': 'query', 'meta':'tokens', 'type': 'rollback'}).submit()['query']['tokens']['rollbacktoken']
parameters={'action': 'rollback',
'title': page.title(),
logging.info(template)
logging.info(summary)
logging.info(human_readable)
if not check_exceptions(page.text):
page.save(summary)
except pywikibot.exceptions.NoPage:
logging.error("No Page, pass")
pass
except:
logging.error(sys.exc_info())
pass
if __name__ == "__main__":
# Global Settings
verbose = 1
logging.basicConfig(format='%(asctime)s - %(message)s', level=logging.INFO)
copyright_query = "https://copyvios.toolforge.org/api.json?version=1&action=search&project=" + project + "&lang=" + language + "&title="
human_query = "https://copyvios.toolforge.org/?lang=" + language + "&project=" + project + "&title={0}&oldid=&action=search&use_engine=1&use_links=1&turnitin=0"
# Generators
# generator = pagegenerators.TextfilePageGenerator(filename='TestPages.txt') # Offline testing
generator = pagegenerators.LiveRCPageGenerator(wiki) # Live RC
logging.info("Telif Bot starting up...")
for page in generator:
main()
