def make_new_members_request(self):
log('Doing the request with cursor {}'.format(self.cursor.group(1)))
payload = {
'gid': self.group_id,
'order': 'date',
'view': 'list',
'limit': self.basket_size,
'sectiontype': 'recently_joined',
'memberstabexp': 1,
'cursor': self.cursor.group(1),
'start': 15,
'dpr': 1,
'__user': self.user_id,
'__a': 1,
'__dyn': self.dyn_rand(),
'__req': 1,
'__be': 1,
'__pc': 'PHASED%3ADEFAULT',
'__rev': self.next_req_params['__rev'],
'__spin_r': self.next_req_params['__spin_r'],
'__spin_b': 'trunk',
'__spin_t': self.next_req_params['__spin_t']
}
start_time = time.time()
self.current_result = self.sess.get(self.members_url, params=payload)
log('Doing new request:')
exec_time = round(time.time() - start_time, 2)
self.requests += 1
log('~~~~URL~~~~')
log(self.current_result.url)
log_html(self.current_result, 'members_last_req')
log('Request number {} finished with status code {} in {} seconds'.
format(self.requests, self.current_result.status_code, exec_time))
time.sleep(2)
def get_current_cursor(self):
'''
If no match is found, cursor is set no None
:return:
'''
log('Retrieving cursor from response ..')
self.cursor = re.search(
r'sectiontype=recently_joined&memberstabexp=1&cursor=(.*?)&',
self.current_result.text)
if self.cursor is None:
log('No cursor found in current response!')
log_html(self.current_result, 'cursor_error')
raise NoCursor('No cursor found in current response!')
log('Current cursor is now {}'.format(self.cursor.group(1)))
def __init__(self, group_id, user_id, basket_size, sess):
self.group_id = group_id
self.user_id = user_id
self.basket_size = basket_size
self.sess = sess
self.cursor = ''
self.members = set()
self.cache_file_path = os.path.join(self.cache_path,
'{}.pkl'.format(self.group_id))
# set current result to members page
self.current_result = self.sess.get(
'https://www.facebook.com/groups/{}/members/'.format(group_id))
log_html(self.current_result, 'group_members')
log('Retrieving members from initial group page..')
self.gather_members()
self.get_current_cursor()
self.next_req_params = self.get_next_req_params()
def login(self):
log('Getting to login page...')
login_url = "https://www.facebook.com/"
self.current_result = self.sess.get(login_url)
# we need to parse login form for constructing the post data
page_soup = BeautifulSoup(self.current_result.text, 'html.parser')
login_form = page_soup.find_all('form', id='login_form')[0]
send_login_form_to = login_form["action"]
post_data = {}
for input_field in login_form.find_all('input'):
if input_field.has_attr("name"):
name = input_field["name"]
value = input_field["value"] if input_field.has_attr(
"value") else ''
post_data[name] = value
post_data['email'] = self.user['username']
post_data['pass'] = self.user['password']
log('Sending login request to {}'.format(send_login_form_to))
self.current_result = self.sess.post(send_login_form_to, post_data)
log_html(self.current_result, 'login')
def get_about_section(self, section):
if section not in self.profile_sections:
self.profile_sections[section] = self.sess.get(
self.secured_url, params={'section': section})
log_html(self.profile_sections[section], section)
return self.profile_sections[section]