def process(self):
list_domains=self.db['new_domaines'].distinct('domaine')
fex=Faup()
for domain in list_domains:
url='http://'+str(domain)
fex.decode(url, False)
print (fex.get_tld()+','+fex.get_domain()+','+','.join(fex.get_subdomain().split('.')[::-1]).replace('www','')).replace(',,',',')
def __post_init__(self):
if self.domain is None:
f = Faup(
) # Example code at https://programtalk.com/python-examples-amp/pyfaup.faup.Faup/
f.decode(self.address.split("@")[-1])
self.top_level_domain = f.get_tld()
self.domain = f.get_domain()
self.subdomain = f.get_subdomain()
def process(self):
list_domains = self.db['new_domaines'].distinct('domaine')
fex = Faup()
for domain in list_domains:
url = 'http://' + str(domain)
fex.decode(url, False)
print(fex.get_tld() + ',' + fex.get_domain() + ',' +
','.join(fex.get_subdomain().split('.')[::-1]).replace(
'www', '')).replace(',,', ',')
def __post_init__(self):
f = Faup(
) # Example code at https://programtalk.com/python-examples-amp/pyfaup.faup.Faup/
f.decode(self.url)
self.scheme = f.get_scheme()
self.top_level_domain = f.get_tld()
self.domain = f.get_domain()
self.subdomain = f.get_subdomain()
self.path = f.get_resource_path()
#!/usr/bin/python
from pyfaup.faup import Faup
import sys
import codecs
import binascii
f = Faup()
file_urls=codecs.open(sys.argv[1],'r','ascii',errors='ignore')
urls=file_urls.readlines()
for url in urls:
url=url.replace('\n','')
#print("We decode the url: %s" % (url))
#if sys.version.split('.')[0].split('.')[0]=='3':
f.decode(bytes(url,'utf-8'), False)
#if sys.version.split('.')[0].split('.')[0]=='2':
# f.decode(bytes(url),False)
#data = f.get()
f.get_tld()
#f.get_domain()
#f.get_subdomain()
#print(f.get_tld())
#print(f.get_domain())
#print("URL TLD: %s" % (f.get_tld()))
#!/usr/bin/python
from pyfaup.faup import Faup
import sys
import codecs
import binascii
f = Faup()
file_urls=codecs.open(sys.argv[1],'r','ascii',errors='ignore')
urls=file_urls.readlines()
for url in urls:
url=url.replace('\n','')
print("URL:[%s]" % (url))
f.decode(url)
print("-----> Extracted TLD:%s" % f.get_tld())
