class Factory:
def __init__(self,
username: str,
password: str,
timeout: int = 30,
executable_path: str = None):
self.username = username
self.password = password
self.timeout = timeout
self.executable_path = executable_path if executable_path else 'phantomjs'
def login(self, url):
self.driver = PhantomJS(executable_path=self.executable_path)
self.driver.get(url)
self.driver.find_element_by_id('userID').send_keys(self.username)
self.driver.find_element_by_id('password').send_keys(self.password)
self.driver.find_element_by_xpath("//input[@value='Log in']").click()
WebDriverWait(self.driver, self.timeout).until(
EC.text_to_be_present_in_element((By.XPATH, "//a[@alt='logout']"),
'Log Out'),
'%s failed to login within %d seconds timeout' %
(self.username, self.timeout))
def start(self, method: Callable):
method()
WebDriverWait(self.driver, self.timeout).until(
is_page_read(),
'Opertation failed to be completed before the %d seconds timeout' %
self.timeout)
def write_article(date, body):
url = 'http://gall.dcinside.com/mgallery/board/write/?id=nendoroid'
xpaths = {
'nick': "//input[@name='name']",
'pw': "//input[@name='password']",
'title': "//input[@name='subject']",
'body': "//div[@id='tx_canvas_source_holder']",
}
dcap = dict(dc.PHANTOMJS)
dcap["phantomjs.page.settings.userAgent"] = (
"Mozilla/5.0 (X11; Linux x86_64) AppleWebKit/53 "
"(KHTML, like Gecko) Chrome/16.0.82")
title_format = '[알림] 굿스마샵 {} 넨도 예약 마감 D-{}'
date = date.replace(hour=12)
today = datetime.now(
) #.replace(hour=0, minute=0, second=0, microsecond=0)
dday = date - today
if dday.days > 3 or dday.days < 0:
return
title = title_format.format(date.strftime('%m월%d일'), dday.days)
print(title)
driver = PhantomJS(desired_capabilities=dcap)
driver.get(url)
driver.find_element_by_xpath(xpaths['nick']).send_keys('넨갤봇')
driver.find_element_by_xpath(xpaths['pw']).send_keys('2648')
driver.find_element_by_xpath(xpaths['title']).send_keys(title)
WebDriverWait(driver, 2)
html = driver.find_element_by_xpath("//div[@id='tx_switchertoggle']")
bd = driver.find_element_by_xpath(xpaths['body'])
def make_body(date, body):
b = '<p>{} 정오에 예약이 마감됩니다.</p><br/><p>클릭하면 제품 페이지로 이동합니다.</p><br/><ul>'.format(
date.strftime('%m월 %d일'))
for name, link in body:
b += ('<li><a href="{}" target="_blank">{}</a></li>'.format(
link, name))
b += '</ul>'
return b
full_body = make_body(date, body)
print(full_body)
actions = webdriver.ActionChains(driver)
actions.move_to_element(html)
actions.click()
actions.move_to_element(bd)
actions.click()
actions.pause(1)
actions.send_keys(full_body)
actions.pause(1)
actions.perform()
submit = driver.find_element_by_xpath("//p[@class='btn_box_right']//input")
submit.click()
WebDriverWait(driver, 1)
#print(driver.get_log('browser'))
#print(driver.get_log('har'))
driver.save_screenshot('a.png')
driver.close()
print('done!')
def mainprocess(captchakey, saveloc):
user_agent = (
"Mozilla/5.0 (Macintosh; Intel Mac OS X 10_8_4) " +
"AppleWebKit/537.36 (KHTML, like Gecko) Chrome/29.0.1547.57 Safari/537.36"
)
dcap = dict(DesiredCapabilities.PHANTOMJS)
dcap["phantomjs.page.settings.userAgent"] = user_agent
driver = PhantomJS(desired_capabilities=dcap,
service_args=['--load-images=no'])
#driver = webdriver.PhantomJS()
driver.set_window_size(1600, 1200)
driver.implicitly_wait(10)
driver.get("https://club.pokemon.com/us/pokemon-trainer-club")
driver.find_element_by_id("user-account-signup").click()
delay = 5 # seconds
try:
myElem = WebDriverWait(driver, delay).until(
EC.presence_of_element_located((By.ID, 'id_dob')))
except TimeoutException:
logging.info("Loading took too much time!")
elem = driver.find_element_by_name("dob")
driver.execute_script(
"var input = document.createElement('input'); input.type='text'; input.setAttribute('name', 'dob'); arguments[0].parentNode.replaceChild(input, arguments[0])",
elem)
randdate = datetime.date(randint(1975, 1990), randint(1, 12),
randint(1, 28))
elem = driver.find_element_by_name("dob")
elem.send_keys(datetime.datetime.strftime(randdate, '%Y-%m-%d'))
driver.save_screenshot('bday4.png')
time.sleep(1)
delay = 5 # seconds
try:
myElem = WebDriverWait(driver, delay).until(
EC.presence_of_element_located((
By.XPATH,
'//*[@id="sign-up-theme"]/section/div/div/div[1]/form/div[2]/div/div/label'
)))
except TimeoutException:
logging.info("Loading took too much time!")
myElem.click()
time.sleep(1)
delay = 5 # seconds
try:
myElem = WebDriverWait(driver, delay).until(
EC.presence_of_element_located((
By.XPATH,
'//*[@id="sign-up-theme"]/section/div/div/div[1]/form/div[2]/div/div/div/div/ul/li[168]'
)))
except TimeoutException:
logging.info("Loading took too much time!")
myElem.click()
driver.find_element_by_xpath(
'//*[@id="sign-up-theme"]/section/div/div/div[1]/form/input[2]').click(
)
randomemail = ''.join(
random.choice(string.ascii_uppercase + string.digits)
for _ in range(10))
randompass = ''.join(
random.SystemRandom().choice(string.ascii_lowercase +
string.ascii_uppercase + string.digits +
string.punctuation) for _ in range(10))
driver2 = PhantomJS(desired_capabilities=dcap,
service_args=['--load-images=no'])
driver2.set_window_size(1600, 1200)
driver2.get("https://temp-mail.org/en/option/change/")
driver2.find_element_by_name("mail").send_keys(randomemail)
doms = Select(driver2.find_element_by_xpath('//*[@id="domain"]'))
x = doms.options
domain = str(x[0].text)
time.sleep(3)
driver2.find_element_by_id('postbut').click()
driver2.get("https://temp-mail.org/en/option/refresh/")
driver.save_screenshot('screenshot.png')
driver.find_element_by_id('id_username').send_keys(randomemail)
driver.find_element_by_id('check-availability-username').click()
delay = 5 # seconds
try:
myElem = WebDriverWait(driver, delay).until(
EC.presence_of_element_located(
(By.XPATH, '//*[@id="username-suggestion"]/div/h3')))
if driver.find_element_by_xpath(
'//*[@id="username-suggestion"]/div').find_element_by_tag_name(
'h3').text == 'Username is invalid.':
logging.info('Username already in use exiting')
exit()
except TimeoutException:
logging.info("Loading took too much time!")
driver.find_element_by_id('id_password').send_keys(randompass)
driver.find_element_by_id('id_confirm_password').send_keys(randompass)
driver.find_element_by_id('id_email').send_keys(randomemail + domain)
driver.find_element_by_id('id_confirm_email').send_keys(randomemail +
domain)
driver.find_element_by_id('id_screen_name').send_keys(randomemail)
driver.find_element_by_id('check-availability-screen-name').click()
delay = 5 # seconds
try:
myElem = WebDriverWait(driver, delay).until(
EC.presence_of_element_located((
By.XPATH,
'//*[@id="user-signup-create-account-form"]/fieldset[1]/div/div/div[1]/div/div[11]/div/div/h3'
)))
if driver.find_element_by_xpath(
'//*[@id="user-signup-create-account-form"]/fieldset[1]/div/div/div[1]/div/div[11]/div/div'
).find_element_by_tag_name(
'h3'
).text == 'This screen name already exists. Please try the following:':
logging.info('Screen Name already in use exiting')
exit()
except TimeoutException:
logging.info("Loading took too much time!")
driver.find_element_by_id('id_terms').click()
logging.info("Starting captcha solve")
# AutoCapcha
autocaptcha(captchakey, driver)
driver.find_element_by_xpath(
'//*[@id="user-signup-create-account-form"]/fieldset[2]/div/input'
).click()
driver.close()
logging.info("Waiting on Email")
for z in range(0, 10):
try:
delay = 60 # seconds
try:
myElem = WebDriverWait(driver2, delay).until(
EC.presence_of_element_located(
(By.XPATH, '//*[@id="mails"]/tbody')))
except TimeoutException:
if z == 9:
logging.info("Can't Find the email sorries :(")
exit()
else:
logging.info("No Email Yet " + str(z + 1) + "/10")
driver2.refresh()
except:
continue
myElem.click()
time.sleep(5)
elems = driver2.find_elements_by_tag_name('a')
time.sleep(5)
for elem in elems:
test = str(elem.get_attribute("href"))
if "https://club.pokemon.com/us/pokemon-trainer-club/activated/" in test:
actions = ActionChains(driver2)
actions.move_to_element(elem).perform()
elem.click()
break
time.sleep(10)
driver2.quit()
logging.info("Account created saving details")
with open(os.path.join(saveloc, 'accounts.txt'), "a") as myfile:
myfile.write("\n" + randomemail + ' - ' + randompass)
logging.info(randomemail + " account created")
class Vassar_Downloader(object):
err_log_name = 'ERROR_LOG.txt'
download_log = "temp.txt"
raw_file_folder = 'Raw files -- %s' % str(time.localtime()[0:3])
def __init__(self,
urn=None,
pwd=None,
init_dict=[res for res in permutation(lowercase, 2)]):
if not os.path.exists(self.raw_file_folder):
os.mkdir(self.raw_file_folder)
os.chdir(self.raw_file_folder)
if not urn:
urn = raw_input('Enter the username: '******''
else:
print 'Username: '******'\n'
if not pwd:
pwd = getpass.getpass('Enter the password: '******'cls')
with open(self.err_log_name, 'w') as f:
f.write('Starting crawling from %s...\n' % time.ctime())
self.url = 'https://aisapps.vassar.edu/directory/internal_auth_studir_srch.php'
print 'Initializing...\n'
self.browser = PhantomJS()
self.i = 0
self.urn = urn
self.pwd = pwd
self.DICT = init_dict
self.searchReq = 'https://aisapps.vassar.edu/directory/internal_auth_studirctry_rslts.php'
self.init()
def init(self):
self.if_continue()
self.cluster()
def if_continue(self):
self.finished_list = []
if os.path.exists(self.download_log):
with open(self.download_log, 'r') as f:
data = f.read()
self.finished_list = data.split(',')
self.finished_list.pop(-2)
self.waiting_list = filter(
lambda x: x not in self.finished_list, self.DICT)
self.status = "Continue from the last dumping... %d / %d\n" % (
len(self.DICT) - len(self.waiting_list), len(self.DICT))
print self.status
else:
self.waiting_list = self.DICT
def login(self):
print 'Connecting to "%s"...' % self.url
self.browser.get(self.url)
self.status = '\nTrying to login as %s...' % self.urn
print self.status
username = self.browser.find_element_by_name("username")
password = self.browser.find_element_by_name("password")
username.send_keys(self.urn)
password.send_keys(self.pwd)
login_attempt = self.browser.find_element_by_xpath(
"//*[@type='submit']")
login_attempt.submit()
return self.browser.page_source
def request(self, data, method='POST', error_time=0):
filename = 'source - %s.txt' % (str(data.values()))
#print 'downloading %s...'%filename
try:
resp = self.browser.request(method, self.searchReq, data=data)
with open(filename, 'w') as f:
#f.write(clean_content(resp.content))
f.write(resp.content)
temp = data['lastname']
self.finished_list.append(temp)
self.waiting_list.remove(temp)
self.status = '%d / %d done...' % (
self.total - len(self.waiting_list), self.total)
print self.status
with open(self.download_log, 'a') as f:
f.write('%s,' % temp)
except Exception as e:
error_time += 1
if error_time <= 3:
self.logger('Error! %s failed %d times.' %
(str(data), error_time))
self.request(data, error_time=error_time)
else:
self.logger('<<Error! %s not downloaded!>>' % (str(data)))
def cluster(self):
s = self.login()
if 'http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd' not in s:
self.status = '\nAuthentification failed.\n'
print self.status
self.logger('Authentification failed.')
return None
else:
self.status = '\nAuthentification succeeded.\n\nDownloading data...\n\n'
print self.status
self.total = len(self.DICT)
threads = [
MyThread(self.request, ({
'lastname': ln
}, 'POST'), name=ln) for ln in self.waiting_list
]
for t in threads:
t.start()
for t in threads:
t.join()
self.status = '\nClearing cache...\n'
print self.status
os.remove(self.download_log)
print 'Done...\n'
def logger(self, msg):
with open(self.err_log_name, 'a') as f:
f.write('%s --- %s\n' % (msg, time.ctime()))