Stop validating Zone IDs before normalization (#45)

Author: sethmlarson

src/rfc3986/misc.py

@@ -58,6 +58,7 @@
              abnf_regexp.PORT_RE))
 
 
+HOST_MATCHER = re.compile('^' + abnf_regexp.HOST_RE + '$')
 IPv4_MATCHER = re.compile('^' + abnf_regexp.IPv4_RE + '$')
 IPv6_MATCHER = re.compile(r'^\[' + abnf_regexp.IPv6_ADDRZ_RFC4007_RE + r'\]$')
 

src/rfc3986/uri.py

@@ -171,12 +171,6 @@ def authority_info(self):
             # valid bytes, it is an InvalidAuthority.
             raise exc.InvalidAuthority(self.authority.encode(self.encoding))
 
-        if (host and misc.IPv6_MATCHER.match(host) and not
-                misc.IPv6_NO_RFC4007_MATCHER.match(host)):
-            # If it's an IPv6 address that has RFC 4007 IPv6
-            # Zone IDs then it's invalid.
-            raise exc.InvalidAuthority(self.authority.encode(self.encoding))
-
         return matches
 
     @property

src/rfc3986/validators.py

@@ -304,6 +304,24 @@ def authority_is_valid(authority, host=None, require=False):
         bool
     """
     validated = is_valid(authority, misc.SUBAUTHORITY_MATCHER, require)
+    if validated and host is not None:
+        return host_is_valid(host, require)
+    return validated
+
+
+def host_is_valid(host, require=False):
+    """Determine if the host string is valid.
+
+    :param str host:
+        The host to validate.
+    :param bool require:
+        (optional) Specify if host must not be None.
+    :returns:
+        ``True`` if valid, ``False`` otherwise
+    :rtype:
+        bool
+    """
+    validated = is_valid(host, misc.HOST_MATCHER, require)
     if validated and host is not None and misc.IPv4_MATCHER.match(host):
         return valid_ipv4_host_address(host)
     elif validated and host is not None and misc.IPv6_MATCHER.match(host):
@@ -397,7 +415,9 @@ def subauthority_component_is_valid(uri, component):
 
     # If we can parse the authority into sub-components and we're not
     # validating the port, we can assume it's valid.
-    if component != 'port':
+    if component == 'host':
+        return host_is_valid(subauthority_dict['host'], require=True)
+    elif component != 'port':
         return True
 
     try:

tests/conftest.py

@@ -32,9 +32,7 @@
     '[FF02::3::5]',  # IPv6 can only have one ::
     '[FADF:01]',  # Not properly compacted (missing a :)
     '[FADF:01%en0]',  # Not properly compacted (missing a :), Invalid ZoneID
-    '[FADF::01%en0]',  # ZoneID is per RFC 4007
-    '[FADF::01%]',  # Invalid ZoneID separator and no ZoneID
-    '[FADF::01%25]',  # Missing ZoneID in RFC 6974, is 25 in RFC 4007
+    '[FADF::01%]',  # Empty Zone ID
     'localhost:80:80:80',  # Too many ports
     '256.256.256.256',  # Invalid IPv4 Address
     SNOWMAN.decode('utf-8')

tests/test_builder.py

@@ -67,16 +67,22 @@ def test_add_credentials_requires_username():
         builder.URIBuilder().add_credentials(None, None)
 
 
-@pytest.mark.parametrize('hostname', [
-    'google.com',
-    'GOOGLE.COM',
-    'gOOgLe.COM',
-    'goOgLE.com',
-])
-def test_add_host(hostname):
+@pytest.mark.parametrize(
+    ['hostname', 'expected_hostname'],
+    [
+    ('google.com', 'google.com'),
+    ('GOOGLE.COM', 'google.com'),
+    ('gOOgLe.COM', 'google.com'),
+    ('goOgLE.com', 'google.com'),
+    ('[::ff%etH0]', '[::ff%25etH0]'),
+    ('[::ff%25etH0]', '[::ff%25etH0]'),
+    ('[::FF%etH0]', '[::ff%25etH0]'),
+    ]
+)
+def test_add_host(hostname, expected_hostname):
     """Verify we normalize hostnames in add_host."""
     uribuilder = builder.URIBuilder().add_host(hostname)
-    assert uribuilder.host == 'google.com'
+    assert uribuilder.host == expected_hostname
 
 
 @pytest.mark.parametrize('port', [

tests/test_normalizers.py

@@ -67,10 +67,17 @@ def test_hostname_normalization():
             URIReference(None, 'example.com', None, None, None))
 
 
-def test_authority_normalization():
+@pytest.mark.parametrize(
+    ['authority', 'expected_authority'],
+    [
+    ('user%[email protected]', 'user%[email protected]'),
+    ('[::1%eth0]', '[::1%25eth0]')
+    ]
+)
+def test_authority_normalization(authority, expected_authority):
     uri = URIReference(
-        None, 'user%[email protected]', None, None, None).normalize()
-    assert uri.authority == 'user%[email protected]'
+        None, authority, None, None, None).normalize()
+    assert uri.authority == expected_authority
 
 
 def test_fragment_normalization():

tests/test_validators.py

@@ -235,3 +235,21 @@ def test_invalid_uri_with_invalid_path(invalid_uri):
         validators.Validator().check_validity_of(
             'host', 'path',
         ).validate(uri)
+
+
+def test_validating_rfc_4007_ipv6_zone_ids():
+    """Verify that RFC 4007 IPv6 Zone IDs are invalid
+    host/authority but after normalization are valid
+    """
+    uri = rfc3986.uri_reference("http://[::1%eth0]")
+    with pytest.raises(exceptions.InvalidComponentsError):
+        validators.Validator().check_validity_of(
+            'host'
+        ).validate(uri)
+
+    uri = uri.normalize()
+    assert uri.host == '[::1%25eth0]'
+
+    validators.Validator().check_validity_of(
+        'host'
+    ).validate(uri)