class TwitterBot:
__favorites_regex = '<a href="http://twitter.com/[a-zA-Z0-9_]+" title="[a-zA-Z0-9\-_ .]+">[a-zA-Z0-9_]+</a>'
__following_regex = '<a href="http://twitter.com/[a-zA-Z0-9_]+" rel="contact"><img alt="[a-zA-Z0-9\-_ .]+" class'
__fav_complete_name_regex = 'title="[a-zA-Z0-9\-_ .]+">'
__fav_complete_name_prefix = 'title="'
__fav_complete_name_sufix = '">'
__foll_complete_name_regex = 'img alt="[a-zA-Z0-9\-_ .]+" class'
__foll_complete_name_prefix = 'img alt="'
__foll_complete_name_sufix = '" class'
__url_regex = 'href="http://twitter.com/[a-zA-Z0-9_]+" '
__url_prefix = 'href="'
__url_sufix = '" '
def __init__(self):
pass
def initialize(self,
proxies_per_proto={},
user=None,
passw=None,
debug=False):
print 'INIT: TwitterBot'
self.__br = Browser()
self.__br.set_proxies(proxies_per_proto)
self.__br.set_debug_http(debug)
self.__debug = debug
self.__sandman = SandMan('TwitterBot')
self.__ngd = NGD()
self.__ngd.set_proxies(proxies_per_proto)
self.__lock = Lock()
try:
# sign in
self.__br.open("http://twitter.com/")
self.__br.select_form(nr=1)
self.__br['session[username_or_email]'] = user
self.__br['session[password]'] = passw
resp = self.__br.submit()
time.sleep(0.2)
except Exception, e:
if self.__debug:
traceback.print_exc(file=sys.stdout)
print str(e)
print 'EXCEPTION on TwitterBot, possibly bad user/password or https login don\' work behind a proxy.'
class SearchEngineBot:
def __init__(self):
pass
def initialize(self,
proxies_per_proto={},
user=None,
passw=None,
debug=False):
print 'INIT: SearchEngineBot'
self.__br = Browser()
self.__br.set_proxies(proxies_per_proto)
self.__br.set_debug_http(debug)
self.__ngd = NGD(proxies_per_proto)
self.__harvest_command = EmailHarvestingCommand()
self.__harvest_command.set_only_complete_names(False)
self.__sandman = SandMan('SearchEngineBot')
# no sign in
def set_proxies_per_proto(self, proxies):
self.__proxies = proxies
try:
self.__ngd.set_proxies(proxies)
except:
print 'EXCEPTION on SeachEngineBot, possibly bad user/password or https login don\' work behind a proxy.'
if len(proxies) == 0:
proxy = None
else:
proxy = tuple(proxies['http'].split(':'))
proxy = (proxy[0], int(proxy[1]))
self.__proxy = proxy
def set_sleep_secs(self, secs):
self.__sandman.set_sleep_secs(secs)
def set_sleep_module(self, iterations):
self.__sandman.set_sleep_module(iterations)
def set_sleep_failure(self, secs):
self.__sandman.set_sleep_failure(secs)
def set_sleep_random_flag(self, bool):
self.__sandman.set_sleep_random_flag(bool)
def self_email(self, email, name):
if name.lower().startswith(email.split('@')[0].lower()):
return True
if len(name.split(' ')) == 1 and name.lower() == email.split(
'@')[0].lower():
return True
if len(name.split(' ')) == 2 and '.'.join(
name.split(' ')).lower() == email.split('@')[0].lower():
return True
if len(name.split(' ')) == 2 and '_'.join(
name.split(' ')).lower() == email.split('@')[0].lower():
return True
if len(name.split(' ')) == 2 and name.split(
' ')[0].lower() == email.split('@')[0].lower():
return True
if len(name.split(' ')) == 2 and (
name.split(' ')[0][0] +
name.split(' ')[1]).lower() == email.split('@')[0].lower():
return True
return False
def name_to_emails(self, (aliases, graph)):
self.__harvest_command.set_only_complete_names(False)
return self.__name_to_emails(aliases, 'all_mails')
