PYTHON-6080 Introduce Custom Callback logic in Configurable DNS

doc/changelog.rst

@@ -16,6 +16,19 @@ PyMongo 4.19 brings a number of changes including:
   interpreter remain daemon threads and shutdown behavior is unchanged.
   Note that because these threads are non-daemon, a subinterpreter may block
   on teardown until any in-flight monitor work completes.
+- Added the ``srv_host_validator`` keyword argument to
+  :class:`~pymongo.synchronous.mongo_client.MongoClient` and
+  :class:`~pymongo.asynchronous.mongo_client.AsyncMongoClient`, an alternative to
+  ``srvAllowedHostsSuffix`` for deployments whose acceptable SRV hosts cannot be
+  expressed as a single suffix. The callback is invoked once per SRV-returned
+  host and returns ``True`` to accept it. It is mutually exclusive with
+  ``srvAllowedHostsSuffix`` and, because it takes a callable, cannot be set in a
+  connection string. See the
+  :class:`~pymongo.synchronous.mongo_client.MongoClient` and
+  :class:`~pymongo.asynchronous.mongo_client.AsyncMongoClient` documentation for
+  security considerations.
+- ``srvAllowedHostsSuffix`` may now be set to a single label reserved for
+  private or special use, such as ``localhost``, ``test``, or ``internal``.
 
 Bug fixes
 .........
@@ -37,6 +50,8 @@ Bug fixes
   delimiters (`PYTHON-6040`_).
 - :class:`~pymongo.driver_info.DriverInfo` now raises :class:`ValueError` when
   any field contains the reserved ``|`` delimiter (`PYTHON-6040`_).
+- Fixed a bug in SRV polling where invalid hosts where topology would not be
+  updated if one returned host was invalid.
 
 .. _PYTHON-6074: https://jira.mongodb.org/browse/PYTHON-6074
 .. _PYTHON-6040: https://jira.mongodb.org/browse/PYTHON-6040

pymongo/_psl.py

@@ -21,14 +21,34 @@
 
 _PUBLIC_SUFFIXES: Optional[tuple[set[str], set[str], set[str]]] = None
 
+# Single labels that srvAllowedHostsSuffix may be set to. See the
+# srvAllowedHostsSuffix section of the Initial DNS Seedlist Discovery spec.
+SPECIAL_USE_LABELS = frozenset(
+    [
+        # RFC 6761 special use names.
+        "test",
+        "localhost",
+        "invalid",
+        "example",
+        # RFC 6762 multicast DNS.
+        "local",
+        # Reserved by ICANN for private use.
+        "internal",
+        # Not officially reserved by ICANN but commonly used privately.
+        "corp",
+        "home",
+        "mail",
+    ]
+)
+
 
 def _to_punycode(string: str) -> str:
     """Convert a string to Punycode."""
     try:
         return string.encode("idna").decode("ascii")
     except UnicodeError:
         # Not every input is a valid IDNA label (e.g. empty or over-long
-        # labels). Leave it as-is and let the caller reject it.
+        # labels). Return it unchanged -- SRV host will fail verification.
         return string
 
 

pymongo/_telemetry.py

@@ -746,6 +746,17 @@ def log_srv_monitor_failure(failure: Exception) -> None:
         _debug_log(_SDAM_LOGGER, message="SRV monitor check failed", failure=repr(failure))
 
 
+def log_srv_monitor_invalid_host(host: str, failure: Exception) -> None:
+    """Emit a log entry when the SRV monitor skips a host that failed verification."""
+    if _is_debug_enabled(_SDAM_LOGGER):
+        _debug_log(
+            _SDAM_LOGGER,
+            message="SRV monitor skipped invalid host",
+            serverHost=host,
+            failure=repr(failure),
+        )
+
+
 def log_command_retry(
     topology_id: Optional[ObjectId],
     command_name: str,

pymongo/asynchronous/mongo_client.py

@@ -470,6 +470,31 @@ def __init__(
                        srvAllowedHostsSuffix=".internal.example.com",
                    )
 
+          - `srv_host_validator`: (callable or None) A callback used in place of the
+            default parent-domain check for hosts returned by SRV DNS records. It is
+            called once per returned host with the lowercased hostname as its only
+            argument, and must return ``True`` to accept the host or ``False`` to
+            reject it. During initial DNS discovery, rejecting a host raises
+            :exc:`~pymongo.errors.ConfigurationError`, as does an exception raised by
+            the callback itself. Use this when the set of acceptable hosts cannot be
+            expressed as a single suffix::
+
+                def validator(host: str) -> bool:
+                    return host.endswith((".a.example.com", ".b.example.com"))
+
+                AsyncMongoClient(
+                    "mongodb+srv://cluster.example.com/",
+                    srv_host_validator=validator,
+                )
+
+            The callback must not block. It is mutually exclusive with
+            ``srvAllowedHostsSuffix``. Because this option is a callable, it can
+            only be passed in as a keyword argument, not through the connection string.
+
+            .. warning::
+
+               This option replaces the built-in DNS spoofing safeguards.
+               Please use with caution.
 
           | **Write Concern options:**
           | (Only set if passed. No default values.)
@@ -637,6 +662,12 @@ def __init__(
 
         .. seealso:: The MongoDB documentation on `connections <https://dochub.mongodb.org/core/connections>`_.
 
+        .. versionchanged:: 4.19
+           Added the ``srv_host_validator`` keyword argument.
+
+        .. versionchanged:: 4.18
+           Added the ``srvAllowedHostsSuffix`` URI and keyword argument.
+
         .. versionchanged:: 4.17
            Added the ``max_adaptive_retries`` and ``enable_overload_retargeting`` URI and keyword arguments.
 
@@ -823,6 +854,7 @@ def __init__(
         srv_service_name = keyword_opts.get("srvservicename")
         srv_max_hosts = keyword_opts.get("srvmaxhosts")
         srv_allowed_hosts_suffix = keyword_opts.get("srvallowedhostssuffix")
+        srv_host_validator = keyword_opts.get("srv_host_validator")
         if len([h for h in self._host if "/" in h]) > 1:
             raise ConfigurationError("host must not contain multiple MongoDB URIs")
         for entity in self._host:
@@ -915,7 +947,11 @@ def __init__(
         )
 
         self._init_based_on_options(
-            self._seeds, srv_max_hosts, srv_service_name, srv_allowed_hosts_suffix
+            self._seeds,
+            srv_max_hosts,
+            srv_service_name,
+            srv_allowed_hosts_suffix,
+            srv_host_validator,
         )
 
         self._opened = False
@@ -935,6 +971,7 @@ async def _resolve_srv(self) -> None:
         srv_service_name = keyword_opts.get("srvservicename")
         srv_max_hosts = keyword_opts.get("srvmaxhosts")
         srv_allowed_hosts_suffix = keyword_opts.get("srvallowedhostssuffix")
+        srv_host_validator = keyword_opts.get("srv_host_validator")
         for entity in self._host:
             # A hostname can only include a-z, 0-9, '-' and '.'. If we find a '/'
             # it must be a URI,
@@ -956,6 +993,7 @@ async def _resolve_srv(self) -> None:
                     srv_service_name=srv_service_name,
                     srv_max_hosts=srv_max_hosts,
                     srv_allowed_hosts_suffix=srv_allowed_hosts_suffix,
+                    srv_host_validator=srv_host_validator,
                 )
                 seeds.update(res["nodelist"])
                 opts = res["options"]
@@ -1000,7 +1038,11 @@ async def _resolve_srv(self) -> None:
             )
 
             self._init_based_on_options(
-                seeds, srv_max_hosts, srv_service_name, srv_allowed_hosts_suffix
+                seeds,
+                srv_max_hosts,
+                srv_service_name,
+                srv_allowed_hosts_suffix,
+                srv_host_validator,
             )
 
     def _init_based_on_options(
@@ -1009,7 +1051,17 @@ def _init_based_on_options(
         srv_max_hosts: Any,
         srv_service_name: Any,
         srv_allowed_hosts_suffix: Any,
+        srv_host_validator: Any = None,
     ) -> None:
+        if srv_host_validator is not None:
+            if srv_allowed_hosts_suffix is not None:
+                raise ConfigurationError(
+                    "Cannot specify both srv_host_validator and srvAllowedHostsSuffix"
+                )
+            if not self._resolve_srv_info["is_srv"]:
+                raise ConfigurationError(
+                    "The srv_host_validator option is only allowed with 'mongodb+srv://' URIs"
+                )
         self._event_listeners = self._options.pool_options._event_listeners
         self._topology_settings = TopologySettings(
             seeds=seeds,
@@ -1028,6 +1080,7 @@ def _init_based_on_options(
             srv_service_name=srv_service_name,
             srv_max_hosts=srv_max_hosts,
             srv_allowed_hosts_suffix=srv_allowed_hosts_suffix,
+            srv_host_validator=srv_host_validator,
             server_monitoring_mode=self._options.server_monitoring_mode,
             topology_id=self._topology_settings._topology_id if self._topology_settings else None,
         )

pymongo/asynchronous/monitor.py

@@ -368,6 +368,7 @@ async def _get_seedlist(self) -> Optional[list[tuple[str, Any]]]:
                 self._settings.pool_options.connect_timeout,
                 self._settings.srv_service_name,
                 srv_allowed_hosts_suffix=self._settings.srv_allowed_hosts_suffix,
+                srv_host_validator=self._settings.srv_host_validator,
             )
             seedlist, ttl = await resolver.get_hosts_and_min_ttl()
             if len(seedlist) == 0:

pymongo/asynchronous/settings.py

@@ -18,7 +18,7 @@
 
 import threading
 from collections.abc import Collection
-from typing import Optional
+from typing import Callable, Optional
 
 from bson.objectid import ObjectId
 from pymongo import common
@@ -51,6 +51,7 @@ def __init__(
         srv_service_name: str = common.SRV_SERVICE_NAME,
         srv_max_hosts: int = 0,
         srv_allowed_hosts_suffix: Optional[str] = None,
+        srv_host_validator: Optional[Callable[[str], bool]] = None,
         server_monitoring_mode: str = common.SERVER_MONITORING_MODE,
         topology_id: Optional[ObjectId] = None,
     ):
@@ -77,6 +78,7 @@ def __init__(
             srv_service_name,
             srv_max_hosts,
             srv_allowed_hosts_suffix,
+            srv_host_validator,
             server_monitoring_mode,
             topology_id,
         )