issues with chinese character encoding, utf-16

Author: elliotvilhelm

domain_validation/constants.py

@@ -2,7 +2,7 @@
 from __future__ import unicode_literals
 WHOIS_PORT = 43
 WHOIS_RESPONSE_LEN_LIMIT = 10000
-WHOIS_PORTION_SIZE = 64
+WHOIS_PORTION_SIZE = 1024
 SERVERS = {
     "ac": "whois.nic.ac",
     "ae": "whois.aeda.net.ae",

domain_validation/parse_date.py

@@ -1,3 +1,5 @@
+# -*- coding: utf-8 -*-
+
 import re
 from datetime import datetime
 from domain_validation.constants import MONTHS

domain_validation/registrar.py

@@ -1,3 +1,5 @@
+# -*- coding: utf-8 -*-
+
 import re
 from domain_validation.constants import SERVER_NOT_FOUND, REGISTRAR_NOT_FOUND
 

domain_validation/test/__pycache__/__init__.cpython-37.pyc

@@ -17,26 +17,38 @@ def test_WHOIS_com2():
     assert str(whois.creation_date()) == "1995-01-18"
     assert whois.registrar() == "MarkMonitor Inc."
 
+
 def test_WHOIS_org():
     sleep(.1)
     whois = WHOIS("npr.org")
     assert str(whois.creation_date()) == "1993-12-13"
     assert whois.registrar() == "Network Solutions, LLC"
 
+
 def test_WHOIS_uk():
     sleep(.1)
     whois = WHOIS("wow.uk")
     assert str(whois.creation_date()) == "2014-06-11"
     assert whois.registrar() == "Planet Hippo Internet Ltd t/a EUKHOST [Tag = PLANETHIPPO]"
 
+
 def test_WHOIS_com_silo():
     sleep(.1)
     whois = WHOIS("1shivom.com")
     assert str(whois.creation_date()) == "2018-07-30"
     assert whois.registrar() == "NameSilo, LLC"
 
+
 def test_WHOIS_cn():
     sleep(.1)
     whois = WHOIS("yo.cn")
     assert str(whois.creation_date()) == '2003-03-17'
     assert whois.registrar() == '浙江贰贰网络有限公司'
+
+
+def test_WHOIS_utf_encoding():
+    sleep(.1)
+    whois = WHOIS("我爱你.中国")
+    assert str(whois.creation_date()) == '2004-01-06'
+    assert whois.registrar() == '北京新网互联科技有限公司'
+

domain_validation/test/__pycache__/test_whois_obj.cpython-37-PYTEST.pyc

@@ -1,3 +1,5 @@
+# -*- coding: utf-8 -*-
+
 from domain_validation.whois_client import query_whois
 from domain_validation.creation_date import get_domain_creation_date
 from domain_validation.registrar import get_registrar

domain_validation/test/test_whois_obj.py

@@ -1,4 +1,5 @@
-# -*- coding: utf-8 -*- 
+# -*- coding: utf-8 -*-
+
 import socket
 from domain_validation.constants import WHOIS_PORTION_SIZE, WHOIS_PORT, WHOIS_RESPONSE_LEN_LIMIT, SERVERS, SERVER_NOT_FOUND
 
@@ -13,7 +14,10 @@ def query_whois(domain):
         sock.send((domain + "\r\n").encode('utf-8'))
         whois_response = ''
         while len(whois_response) < WHOIS_RESPONSE_LEN_LIMIT:
-            response_portion = sock.recv(WHOIS_PORTION_SIZE).decode('utf-8')
+            try:
+                response_portion = sock.recv(WHOIS_PORTION_SIZE).decode('utf-8')
+            except Exception as e:
+                response_portion = sock.recv(WHOIS_PORTION_SIZE).decode('utf-16')
             if response_portion == '':
                 break
             whois_response += response_portion