def is404(self, source):
if Bundle().getString(31) in source:
Logging(4, Bundle().getString(22), Bundle().getString(30))
return True
else:
#Logging(1,Bundle().getString(22) ,Bundle().getString(32))
return False
def nextLink(self):
link = __http__(Bundle().getString(19) + "127.0.0.2", "", False,
False).GET()
if len(link) > 75:
Logging(1, Bundle().getString(27), link[:75])
else:
Logging(1, Bundle().getString(27), link)
return link
def __execute__(self, command):
Logging(1, Bundle().getString(28), str(command))
try:
self.chrome.execute_script(str(command))
except JavascriptException:
self.__execute__(command)
raise Bundle().getString(48)
except TypeError:
pass
except NoSuchWindowException:
print("CLOSED JS")
def getCapcha(self):
url = self.Stacked_extraction(Bundle().getString(43))
try:
url = url[0]
if Bundle().getString(44) in url or Bundle().getString(45) in url:
return Murphy().noCapcha(url)
except KeyError:
pass
except IndexError:
pass
except ValueError:
pass
def Stacked_extraction(self, __target__):
try:
if self.is404(self.source):
return ""
if self.isBlocked(self.source) == False:
content = html.fromstring(self.source)
return content.xpath(__target__)
else:
return ""
except etree.ParseError:
raise Bundle().getString(47)
except etree.XPathEvalError:
raise Bundle().getString(47) + '\r\n' + __target__
def __new_ip__(self):
try:
with Controller.from_port(port=9051) as controller:
controller.authenticate(password='******')
socks.setdefaultproxy(proxy_type=socks.PROXY_TYPE_SOCKS5,
addr="127.0.0.1",
port=9050)
controller.signal(Signal.NEWNYM)
return True
except Exception as ex:
Logging(3,
Bundle().getString(57),
Bundle().getString(58) + '\n' + str(ex))
return False
def Navigate(self):
try:
__tor__().__new_ip__()
Anonimity().__get_ip__()
self.chrome.get(self.address)
except exceptions.MaxRetryError:
Bundle().getString(51)
time.sleep(10)
self.Navigate()
Bundle().getString(52)
except WebDriverException:
raise Bundle().getString(53)
except NoSuchWindowException:
print("CLOSED")
def __init__(self, selector):
self.delimiter = ":"
self.delimiter2 = '~'
self.newline = "\n"
self.path = os.getcwd() + "/Spider/Extractor/Selectors/" + str(
selector) + ".cort"
if os.path.isfile(self.path) == False:
Logging(2, Bundle().getString(60), Bundle().getString(61))
exit()
print(self.path)
self.stack = fs().Read(self.path)
self.cluster = {'Keys': [], 'Values': []}
self.__xpath__ = None
self.end = False
def Read(self, path):
if os.path.isfile(path):
with open(path, "r", encoding="utf-8") as reader:
return reader.read()
else:
Logging("Error", 'read', Bundle().getString(13))
return ""
def Start(self, source):
if source != "" and self.isCaptcha(source) == False:
self.source = source
Bundle().getString(49)
else:
self.source = source
self.setCaptcha()
def __init__(self, address, data, PROXY_ROTATION, AGENT_ROTATION):
if address != "":
Logging(1, Bundle().getString(22), address)
self.address = address
self.data = data
self.PROXY_ROTATION = PROXY_ROTATION
self.AGENT_ROTATION = AGENT_ROTATION
self.OUTPUT = ''
def lxmlExtract(self, xpath: str):
if xpath != None and self.chrome.Source() != None:
try:
content = html.fromstring(self.chrome.Source())
found = content.xpath(xpath)
for item in found:
try:
if item != None:
if item != "":
return item
else:
return 1
except TypeError:
return 1
except etree.ParseError:
raise Bundle().getString(47)
except etree.XPathEvalError:
raise Bundle().getString(47) + '\r\n' + xpath
def isBlocked(self, source):
if Bundle().getString(33) in source:
Logging(4, Bundle().getString(22), Bundle().getString(34))
Logging(4, Bundle().getString(22), Bundle().getString(35))
Tor().__new_ip__()
Clock().waitM(.5)
return True
else:
Logging(1, Bundle().getString(22), Bundle().getString(32))
return False
def noCapcha(self, url):
__IMG__ = __http__(url, "", True, True).getImage()
captcha = pytesseract.image_to_string(
Image.open(__IMG__), config='--psm 7').strip().replace(" ", "")
Logging(2, Bundle().getString(58), captcha)
return captcha
def getImage(self):
try:
with req.Session() as __session__:
if self.PROXY_ROTATION:
__tor__ = Tor()
__tor__.__new_ip__()
__session__.proxies = ProxyFactory().__tor_gen__()
if self.AGENT_ROTATION:
__session__.headers = HeadersFactory().__aws_random__()
self.OUTPUT = __session__.get(self.address, stream=True).raw
except exceptions.URLRequired:
raise Bundle().getString("1")
except exceptions.URLRequired:
raise Bundle().getString("2")
except exceptions.MissingSchema:
raise Bundle().getString("3")
except exceptions.InvalidSchema:
raise Bundle().getString("4")
except exceptions.TooManyRedirects:
raise Bundle().getString("5")
except exceptions.ConnectTimeout:
raise Bundle().getString("6")
except exceptions.Timeout:
raise Bundle().getString("7")
except exceptions.HTTPError:
raise Bundle().getString("8")
except exceptions.SSLError:
raise Bundle().getString("9")
except exceptions.ProxyError:
raise Bundle().getString("10")
except exceptions.ConnectionError:
raise Bundle().getString("11")
finally:
if self.OUTPUT is not None and self.OUTPUT != "":
return self.OUTPUT
else:
Logging(1, Bundle().getString(26), Bundle().getString(25))
return ""
def setCaptcha(self,captcha,_by):
if captcha is not None and len(captcha) > 3:
Clock().waitS(5)
self.element.Input(Bundle().getString(46),captcha,_by)
def isCaptcha(self, source):
if Bundle().getString(36) in source:
return True
else:
return False
