Adds support for emojis and more IDNA URLs

Author: automationator

tests/test_url.py

@@ -42,6 +42,10 @@
     u'http://مثال.إختبار',
     u'http://例子.测试',
     u'http://उदाहरण.परीक्षा',
+    u'http://www.😉.com',
+    u'http://😉.com/😁',
+    u'http://উদাহরণ.বাংলা',
+    u'http://xn--d5b6ci4b4b3a.xn--54b7fta0cc',
     u'http://-.~_!$&\'()*+,;=:%40:80%2f::::::@example.com',
     u'http://1337.net',
     u'http://a.b-c.de',

validators/url.py

@@ -69,16 +69,16 @@
     r"(25[0-5]|(2[0-4]|1{0,1}[0-9]){0,1}[0-9])"
     r")\]|"
     # host name
-    r"(?:(?:[a-z\u00a1-\uffff0-9]-?)*[a-z\u00a1-\uffff0-9]+)"
+    r"(?:(?:(?:xn--)|[a-z\u00a1-\uffff\U00010000-\U0010ffff0-9]-?)*[a-z\u00a1-\uffff\U00010000-\U0010ffff0-9]+)"
     # domain name
-    r"(?:\.(?:[a-z\u00a1-\uffff0-9]-?)*[a-z\u00a1-\uffff0-9]+)*"
+    r"(?:\.(?:(?:xn--)|[a-z\u00a1-\uffff\U00010000-\U0010ffff0-9]-?)*[a-z\u00a1-\uffff\U00010000-\U0010ffff0-9]+)*"
     # TLD identifier
-    r"(?:\.(?:[a-z\u00a1-\uffff]{2,}))"
+    r"(?:\.(?:(?:xn--[a-z\u00a1-\uffff\U00010000-\U0010ffff0-9]{2,})|[a-z\u00a1-\uffff\U00010000-\U0010ffff]{2,}))"
     r")"
     # port number
     r"(?::\d{2,5})?"
     # resource path
-    u"(?:/[-a-z\u00a1-\uffff0-9._~%!$&'()*+,;=:@/]*)?"
+    u"(?:/[-a-z\u00a1-\uffff\U00010000-\U0010ffff0-9._~%!$&'()*+,;=:@/]*)?"
     # query string
     r"(?:\?\S*)?"
     # fragment