def filter_ads(self, url):
if filter.match(url):
self.msg = 'Blacklisted url'
self.status = False
self.code = 808
return False
else:
return True
def filter_ads(self, url): if filter.match(url): self.msg = 'Blacklisted url' self.status = False self.code = 808 return False else: return True
def valid_url(self):
'''checking url format and validity'''
for k, v in self.parse_link(self.url).items():
if k is None:
continue
if v is None or v == "":
setattr(self, k, "")
else:
setattr(self, k, v)
try:
if self.scheme not in ACCEPTED_PROTOCOL:
self.msg = 'URL: Wrong protocol %s' % self.scheme
self.status = False
self.code = 804
return self.status
except Exception as e:
logging.warning("%s" %str(e))
pass
try:
if self.filetype in BAD_TYPES:
self.msg = 'URL: Invalid webpage type %s' % self.filetype
self.status = False
self.code = 806
return self.status
except Exception as e:
pass
try:
if self.domain in BAD_DOMAINS:
self.msg = 'URL: Bad domain %s' % self.domain
self.status = False
self.code = 807
return self.status
except Exception as e:
logging.warning("%s" %str(e))
pass
try:
if self.subdomain in BAD_DOMAINS:
self.msg = 'URL: Bad subdomain %s' % self.subdomain
self.status = False
self.code = 807
return self.status
except Exception as e:
logging.warning("%s" %str(e))
pass
try:
if self.path in BAD_PATHS:
self.msg = 'URL: Bad path %s' % self.path
self.status = False
self.code = 807
return self.status
except Exception as e:
logging.warning("%s" %str(e))
pass
if filter.match(self.url):
self.msg = 'URL: Blacklisted url'
self.status = False
self.code = 808
return self.status
return self
