Let hostname() accept a bare single-label name with the RFC 1034 dot

hostname("yu", rfc_1034=True) passes today through the maybe_simple
fast path, since _simple_hostname_regex has no problem with a lone
alphanumeric label. hostname("yu.", rfc_1034=True) doesn't, because
that regex has no notion of a trailing dot at all, and the fallback to
domain() requires at least a second label before the TLD. So the exact
same name is accepted or rejected purely based on whether it carries
the dot RFC 1034 is supposed to permit, which is backwards from what
that flag promises.

Added a small wrapper around the simple-hostname match that strips a
lone trailing dot first when rfc_1034 is set, mirroring what domain()
already does for its own regex. Left everything else about the simple
path untouched, so this doesn't change hostname() for the vast
majority of callers who never pass rfc_1034.

Fixes GH-442.

Author: afonsojanu

src/validators/hostname.py

@@ -29,6 +29,20 @@ def _simple_hostname_regex():
     return re.compile(r"^(?!-)[a-z0-9](?:[a-z0-9-]{0,59}[a-z0-9])?(?<!-)$", re.IGNORECASE)
 
 
+def _simple_hostname_match(value: str, rfc_1034: bool):
+    """Match value against the simple hostname regex.
+
+    rfc_1034 promises an optional trailing dot is allowed, but the simple
+    regex above has no notion of one, so a bare single-label name like
+    "yu" would pass without the dot and fail with it. Strip a lone
+    trailing dot first when that flag is set, same as domain() already
+    does for its own regex, so the two paths agree.
+    """
+    if rfc_1034 and value.endswith(".") and len(value) > 1:
+        value = value[:-1]
+    return _simple_hostname_regex().match(value)
+
+
 def _port_validator(value: str):
     """Returns host segment if port is valid."""
     if value.count("]:") == 1:
@@ -115,14 +129,14 @@ def hostname(
 
     if may_have_port and (host_seg := _port_validator(value)):
         return (
-            (_simple_hostname_regex().match(host_seg) if maybe_simple else False)
+            (_simple_hostname_match(host_seg, rfc_1034) if maybe_simple else False)
             or domain(host_seg, consider_tld=consider_tld, rfc_1034=rfc_1034, rfc_2782=rfc_2782)
             or (False if skip_ipv4_addr else ipv4(host_seg, cidr=False, private=private))
             or (False if skip_ipv6_addr else ipv6(host_seg, cidr=False))
         )
 
     return (
-        (_simple_hostname_regex().match(value) if maybe_simple else False)
+        (_simple_hostname_match(value, rfc_1034) if maybe_simple else False)
         or domain(value, consider_tld=consider_tld, rfc_1034=rfc_1034, rfc_2782=rfc_2782)
         or (False if skip_ipv4_addr else ipv4(value, cidr=False, private=private))
         or (False if skip_ipv6_addr else ipv6(value, cidr=False))

tests/test_hostname.py

@@ -30,6 +30,9 @@
         ("[dead:beef:0:0:0:0000:42:1]:5731", False, False),
         ("[0:0:0:0:0:ffff:1.2.3.4]:80", False, False),
         ("[0:a:b:c:d:e:f::]:53", False, False),
+        # bare single-label name with the RFC 1034 trailing dot, GH-442
+        ("yu.", True, False),
+        ("yu.:443", True, False),
     ],
 )
 def test_returns_true_on_valid_hostname(value: str, rfc_1034: bool, rfc_2782: bool):
@@ -60,6 +63,10 @@ def test_returns_true_on_valid_hostname(value: str, rfc_1034: bool, rfc_2782: bo
         ("[dead:beef:0:-:0:-:42:1]:5731", False, False),
         ("[0:0:0:0:0:ffff:1.2.3.4]:-65538", False, False),
         ("[0:&:b:c:@:e:f:::9999", False, False),
+        # bad (trailing dot only allowed when rfc_1034 is requested)
+        ("yu.", False, False),
+        # bad (a lone dot has no label to strip down to)
+        (".", True, False),
     ],
 )
 def test_returns_failed_validation_on_invalid_hostname(value: str, rfc_1034: bool, rfc_2782: bool):