Implement canonical names for c-ares

src/gevent/resolver/ares.py

@@ -90,6 +90,9 @@ class Resolver(AbstractResolver):
       ``getaddrinfo``; unknown flags are ignored. System-specific flags
       such as ``AI_V4MAPPED_CFG`` are not supported.
 
+    - ``getaddrinfo`` may return canonical names even without the ``AI_CANONNAME``
+      being set.
+
     .. caution::
 
         This module is considered extremely experimental on PyPy, and
@@ -102,6 +105,10 @@ class Resolver(AbstractResolver):
        resolved <https://github.com/c-ares/c-ares/issues/196>`_ or even
        sent to the DNS server.
 
+    .. versionchanged:: NEXT
+       ``getaddrinfo`` is now implemented using the native c-ares function
+       from c-ares 1.16 or newer.
+
     .. _c-ares: http://c-ares.haxx.se
     """
 
@@ -186,8 +193,6 @@ class Resolver(AbstractResolver):
         """
         Returns a list ``(family, socktype, proto, canonname, sockaddr)``
 
-        TODO: Implement canonical names.
-
         :raises gaierror: If no results are found.
         """
         # pylint:disable=too-many-locals,too-many-branches

src/gevent/resolver/cares.pyx

@@ -535,8 +535,11 @@ cdef class channel:
                               int timeouts,
                               cares.ares_addrinfo* result):
         cdef cares.ares_addrinfo_node* nodes
+        cdef cares.ares_addrinfo_cname* cnames
         cdef sockaddr_in* sadr4
         cdef sockaddr_in6* sadr6
+        cdef object canonname = ''
+
         cdef channel channel
         cdef object callback
         # INET6_ADDRSTRLEN is 46, but we can't use that named constant
@@ -559,6 +562,19 @@ cdef class channel:
             if status != cares.ARES_SUCCESS:
                 callback(Result(None, gaierror(status, strerror(status))))
                 return
+            if result.cnames:
+                # These tend to come in pairs:
+                #
+                # alias: www.gevent.org name: python-gevent.readthedocs.org
+                # alias: python-gevent.readthedocs.org name: readthedocs.io
+                #
+                # The standard library returns the last name so we do too.
+
+                cnames = result.cnames
+                while cnames:
+                    canonname = _as_str(cnames.name)
+                    cnames = cnames.next
+
             nodes = result.nodes
             while nodes:
                 if nodes.ai_family == AF_INET:
@@ -584,7 +600,7 @@ cdef class channel:
                     nodes.ai_family,
                     nodes.ai_socktype,
                     nodes.ai_protocol,
-                    '',
+                    canonname,
                     sockaddr,
                 ))
                 nodes = nodes.ai_next

src/gevent/tests/test__socket_dns.py

@@ -330,16 +330,25 @@ class TestCase(greentest.TestCase):
         # On some systems, the hostname can get caps
         return (result[0].lower(), [], ips)
 
-    def _normalize_result_getaddrinfo(self, result):
-        if not RESOLVER_NOT_SYSTEM:
+    IGNORE_CANONICAL_NAME = RESOLVER_ARES # It tends to return them even when not asked for
+    if not RESOLVER_NOT_SYSTEM:
+        def _normalize_result_getaddrinfo(self, result):
+            return result
+    else:
+        def _normalize_result_getaddrinfo(self, result):
+            # On Python 3, the builtin resolver can return SOCK_RAW results, but
+            # c-ares doesn't do that. So we remove those if we find them.
+            if hasattr(socket, 'SOCK_RAW') and isinstance(result, list):
+                result = [x for x in result if x[1] != socket.SOCK_RAW]
+            if self.IGNORE_CANONICAL_NAME:
+                result = [
+                    (family, kind, proto, '', addr)
+                    for family, kind, proto, _, addr
+                    in result
+                ]
+            if isinstance(result, list):
+                result.sort()
             return result
-        # On Python 3, the builtin resolver can return SOCK_RAW results, but
-        # c-ares doesn't do that. So we remove those if we find them.
-        if hasattr(socket, 'SOCK_RAW') and isinstance(result, list):
-            result = [x for x in result if x[1] != socket.SOCK_RAW]
-        if isinstance(result, list):
-            result.sort()
-        return result
 
     def _normalize_result_getnameinfo(self, result):
         return result
@@ -583,6 +592,22 @@ class TestGeventOrg(TestCase):
                 result = ('readthedocs.io', ) + result[1:]
             return result
 
+    def test_AI_CANONNAME(self):
+        self.IGNORE_CANONICAL_NAME = False
+        result = self._test('getaddrinfo',
+                            # host
+                            TestGeventOrg.HOSTNAME,
+                            # port
+                            None,
+                            # family
+                            socket.AF_INET,
+                            # type
+                            0,
+                            # proto
+                            0,
+                            # flags
+                            socket.AI_CANONNAME)
+        self.assertEqual(result[0][3], 'readthedocs.io')
 
 add(TestGeventOrg, TestGeventOrg.HOSTNAME)