From 3612b88e2e30f83fe8b1034de1c41ca3557bc7ee Mon Sep 17 00:00:00 2001
From: Neil Booth <kyuupichan@gmail.com>
Date: Sun, 16 Jul 2017 13:58:02 +0900
Subject: [PATCH] Permit underscores in hostnames.

Add tests.
---
 lib/util.py            |  4 +++-
 tests/lib/test_util.py | 27 +++++++++++++++++++++++++++
 2 files changed, 30 insertions(+), 1 deletion(-)

diff --git a/lib/util.py b/lib/util.py
index 2c09127..40e5869 100644
--- a/lib/util.py
+++ b/lib/util.py
@@ -259,7 +259,9 @@ def address_string(address):
     return fmt.format(host, port)
 
 # See http://stackoverflow.com/questions/2532053/validate-a-hostname-string
-SEGMENT_REGEX = re.compile("(?!-)[A-Z\d-]{1,63}(?<!-)$", re.IGNORECASE)
+# Note underscores are valid in domain names, but strictly invalid in host
+# names.  We ignore that distinction.
+SEGMENT_REGEX = re.compile("(?!-)[A-Z_\d-]{1,63}(?<!-)$", re.IGNORECASE)
 def is_valid_hostname(hostname):
     if len(hostname) > 255:
         return False
diff --git a/tests/lib/test_util.py b/tests/lib/test_util.py
index 4142679..b55e6b6 100644
--- a/tests/lib/test_util.py
+++ b/tests/lib/test_util.py
@@ -56,3 +56,30 @@ def test_increment_byte_string():
     assert util.increment_byte_string(b'1') == b'2'
     assert util.increment_byte_string(b'\x01\x01') == b'\x01\x02'
     assert util.increment_byte_string(b'\xff\xff') is None
+
+def test_is_valid_hostname():
+    is_valid_hostname = util.is_valid_hostname
+    assert not is_valid_hostname('')
+    assert is_valid_hostname('a')
+    assert is_valid_hostname('_')
+    # Hyphens
+    assert not is_valid_hostname('-b')
+    assert not is_valid_hostname('a.-b')
+    assert is_valid_hostname('a-b')
+    assert not is_valid_hostname('b-')
+    assert not is_valid_hostname('b-.c')
+    # Dots
+    assert is_valid_hostname('a.')
+    assert is_valid_hostname('foo1.Foo')
+    assert not is_valid_hostname('foo1..Foo')
+    assert is_valid_hostname('12Foo.Bar.Bax_')
+    assert is_valid_hostname('12Foo.Bar.Baz_12')
+    # 63 octets in part
+    assert is_valid_hostname('a.abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMN'
+                             'OPQRSTUVWXYZ0123456789_.bar')
+    # Over 63 octets in part
+    assert not is_valid_hostname('a.abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMN'
+                                 'OPQRSTUVWXYZ0123456789_1.bar')
+    len255 = ('a' * 62 + '.') * 4 + 'abc'
+    assert is_valid_hostname(len255)
+    assert not is_valid_hostname(len255 + 'd')