Add @disjoint_base decorator in the stdlib (#14599)

And fix some other new stubtest finds.

Author: JelleZijlstra

.github/workflows/stubtest_stdlib.yml

@@ -46,5 +46,8 @@ jobs:
           check-latest: true
       - name: Install dependencies
         run: pip install -r requirements-tests.txt
+      # Temporary to get @disjoint_base support; can remove once mypy 1.18 is released
+      - name: Install mypy from git
+        run: pip install git+https://github.com/python/mypy.git@116b92bae7b5dbf5e6bd36fd9b0c6804973e5554
       - name: Run stubtest
         run: python tests/stubtest_stdlib.py

stdlib/@tests/stubtest_allowlists/common.txt

@@ -4,15 +4,14 @@
 
 # Please keep sorted alphabetically
 
-collections\.ChainMap\.fromkeys  # https://github.com/python/mypy/issues/17023
-http.client.HTTPConnection.response_class  # the actual type at runtime is abc.ABCMeta
 importlib.abc.Loader.exec_module  # See Lib/importlib/_abc.py. Might be defined for backwards compatibility
 importlib.abc.MetaPathFinder.find_spec  # Not defined on the actual class, but expected to exist.
 importlib.abc.PathEntryFinder.find_spec  # Not defined on the actual class, but expected to exist.
 tkinter.simpledialog.[A-Z_]+
 tkinter.simpledialog.TclVersion
 tkinter.simpledialog.TkVersion
-tkinter.Text.count  # stubtest somehow thinks that index1 parameter has a default value, but it doesn't in any of the overloads
+builtins.tuple  # should have @disjoint_base but hits pyright issue
+tarfile.TarInfo.__slots__  # it's a big dictionary at runtime and the dictionary values are a bit long
 
 
 # ===============================================================
@@ -201,7 +200,6 @@ _markupbase.ParserBase.parse_marked_section
 
 _pydecimal.*  # See comments in file
 _typeshed.*  # Utility types for typeshed, doesn't exist at runtime
-argparse.ArgumentParser.__init__  # stubtest doesn't recognise the runtime default (a class) as being compatible with a callback protocol (the stub annotation)
 argparse.Namespace.__getattr__  # The whole point of this class is its attributes are dynamic
 
 # Runtime AST node runtime constructor behaviour is too loose.
@@ -218,8 +216,6 @@ argparse.Namespace.__setattr__  # should allow setting any attribute
 
 ast.ImportFrom.level  # None on the class, but never None on instances
 ast.NodeVisitor.visit_\w+  # Methods are discovered dynamically, see #3796
-_?asyncio.Future.__init__  # Usually initialized from c object
-asyncio.futures.Future.__init__  # Usually initialized from c object
 
 # Condition functions are exported in __init__
 asyncio.Condition.acquire
@@ -230,7 +226,6 @@ asyncio.locks.Condition.locked
 asyncio.locks.Condition.release
 
 builtins.memoryview.__contains__  # C type that implements __getitem__
-builtins.object.__init__  # default C signature is incorrect
 builtins.reveal_locals  # Builtins that type checkers pretends exist
 builtins.reveal_type  # Builtins that type checkers pretends exist
 builtins.type.__dict__  # read-only but not actually a property; stubtest thinks it's a mutable attribute.
@@ -260,8 +255,6 @@ configparser.SectionProxy.getint  # SectionProxy get functions are set in __init
 contextlib.AbstractAsyncContextManager.__class_getitem__
 contextlib.AbstractContextManager.__class_getitem__
 
-_?contextvars.Context.__init__  # C signature is broader than what is actually accepted
-
 copy.PyStringMap  # defined only in Jython
 
 # The Dialect properties are initialized as None in Dialect but their values are enforced in _Dialect
@@ -329,9 +322,6 @@ importlib.machinery.ExtensionFileLoader.get_filename
 inspect.Parameter.__init__
 inspect.Signature.__init__
 
-inspect.Parameter.empty  # set as private marker _empty
-inspect.Signature.empty  # set as private marker _empty
-
 logging.LogRecord.__setattr__  # doesn't exist, but makes things easy if we pretend it does
 
 # Iterable classes that don't define __iter__ at runtime (usually iterable via __getitem__)
@@ -377,11 +367,6 @@ multiprocessing.queues.JoinableQueue.__init__
 multiprocessing.queues.Queue.__init__
 multiprocessing.queues.SimpleQueue.__init__
 
-# alias for a class defined elsewhere,
-# mypy infers the variable has type `(*args) -> ForkingPickler`
-# but stubtest infers the runtime type as <class ForkingPickler>
-multiprocessing.reduction.AbstractReducer.ForkingPickler
-
 # These methods are dynamically created after object initialization,
 # copied from a wrapped lock object. Stubtest doesn't think they exist
 # because of that.
@@ -406,7 +391,6 @@ os.PathLike.__class_getitem__  # PathLike is a protocol; we don't expect all Pat
 
 pickle._Pickler\..*  # Best effort typing for undocumented internals
 pickle._Unpickler\..*  # Best effort typing for undocumented internals
-_?queue.SimpleQueue.__init__  # C signature is broader than what is actually accepted
 shutil.rmtree  # function with attributes, which we approximate with a callable protocol
 socketserver.BaseServer.get_request  # Not implemented, but expected to exist on subclasses.
 ssl.PROTOCOL_SSLv2  # Depends on the existence and flags of SSL

stdlib/@tests/stubtest_allowlists/darwin.txt

@@ -43,7 +43,6 @@ curses.LINES  # Initialized only after initscr call
 curses.has_key  # stubtest gets confused because this is both a module and a function in curses
 multiprocessing.popen_spawn_win32  # exists on Darwin but fails to import
 readline.append_history_file  # Only available if compiled with GNU readline, not editline
-select.kqueue.__init__  # default C signature is wrong
 select.poll  # Actually a function; we have a class so it can be used as a type
 
 # Some of these exist on non-windows, but they are useless and this is not intended

stdlib/@tests/stubtest_allowlists/py310.txt

@@ -55,6 +55,8 @@ asyncio.locks.Lock.__init__
 asyncio.locks.Semaphore.__init__
 asyncio.queues.Queue.__init__
 
+_random.Random.__init__  # Issues with __new__/__init__ correspondence
+
 bdb.Breakpoint.clearBreakpoints  # Exists at runtime, but missing from stubs
 
 
@@ -100,7 +102,6 @@ typing_extensions.Sentinel.__call__
 # <= 3.11
 # =======
 
-_?bz2.BZ2Decompressor.__init__  # function does not accept parameters but C signature is set
 enum.Enum._generate_next_value_
 importlib.abc.Finder.find_module
 urllib.request.HTTPPasswordMgrWithPriorAuth.__init__  # Args are passed as is to super, so super args are specified

stdlib/@tests/stubtest_allowlists/py311.txt

@@ -68,7 +68,6 @@ importlib.metadata._meta.SimplePath.__truediv__  # Runtime definition of protoco
 # <= 3.11
 # =======
 
-_?bz2.BZ2Decompressor.__init__  # function does not accept parameters but C signature is set
 enum.Enum._generate_next_value_
 importlib.abc.Finder.find_module
 urllib.request.HTTPPasswordMgrWithPriorAuth.__init__  # Args are passed as is to super, so super args are specified

stdlib/@tests/stubtest_allowlists/py39.txt

@@ -54,7 +54,6 @@ typing_extensions.Sentinel.__call__
 # <= 3.11
 # =======
 
-_?bz2.BZ2Decompressor.__init__  # function does not accept parameters but C signature is set
 enum.Enum._generate_next_value_
 importlib.abc.Finder.find_module
 urllib.request.HTTPPasswordMgrWithPriorAuth.__init__  # Args are passed as is to super, so super args are specified

stdlib/@tests/stubtest_allowlists/win32.txt

@@ -2,10 +2,6 @@
 # TODO: Allowlist entries that should be fixed
 # ============================================
 
-# alias for a class defined elsewhere,
-# mypy infers the variable has type `(*args) -> DupHandle` but stubtest infers the runtime type as <class DupHandle>
-multiprocessing.reduction.AbstractReducer.DupHandle
-
 # Exists at runtime, but missing from stubs
 _winapi.CreateFileMapping
 _winapi.MapViewOfFile

stdlib/_asyncio.pyi

@@ -4,12 +4,13 @@ from collections.abc import Awaitable, Callable, Coroutine, Generator
 from contextvars import Context
 from types import FrameType, GenericAlias
 from typing import Any, Literal, TextIO, TypeVar
-from typing_extensions import Self, TypeAlias
+from typing_extensions import Self, TypeAlias, disjoint_base
 
 _T = TypeVar("_T")
 _T_co = TypeVar("_T_co", covariant=True)
 _TaskYieldType: TypeAlias = Future[object] | None
 
+@disjoint_base
 class Future(Awaitable[_T]):
     _state: str
     @property
@@ -49,6 +50,7 @@ else:
 # While this is true in general, here it's sort-of okay to have a covariant subclass,
 # since the only reason why `asyncio.Future` is invariant is the `set_result()` method,
 # and `asyncio.Task.set_result()` always raises.
+@disjoint_base
 class Task(Future[_T_co]):  # type: ignore[type-var]  # pyright: ignore[reportInvalidTypeArguments]
     if sys.version_info >= (3, 12):
         def __init__(

stdlib/_csv.pyi

@@ -3,7 +3,7 @@ import sys
 from _typeshed import SupportsWrite
 from collections.abc import Iterable
 from typing import Any, Final, Literal, type_check_only
-from typing_extensions import Self, TypeAlias
+from typing_extensions import Self, TypeAlias, disjoint_base
 
 __version__: Final[str]
 
@@ -24,6 +24,7 @@ class Error(Exception): ...
 
 _DialectLike: TypeAlias = str | Dialect | csv.Dialect | type[Dialect | csv.Dialect]
 
+@disjoint_base
 class Dialect:
     delimiter: str
     quotechar: str | None
@@ -48,6 +49,7 @@ class Dialect:
 
 if sys.version_info >= (3, 10):
     # This class calls itself _csv.reader.
+    @disjoint_base
     class Reader:
         @property
         def dialect(self) -> Dialect: ...
@@ -56,6 +58,7 @@ if sys.version_info >= (3, 10):
         def __next__(self) -> list[str]: ...
 
     # This class calls itself _csv.writer.
+    @disjoint_base
     class Writer:
         @property
         def dialect(self) -> Dialect: ...

stdlib/_hashlib.pyi

@@ -3,7 +3,7 @@ from _typeshed import ReadableBuffer
 from collections.abc import Callable
 from types import ModuleType
 from typing import AnyStr, Protocol, final, overload, type_check_only
-from typing_extensions import Self, TypeAlias
+from typing_extensions import Self, TypeAlias, disjoint_base
 
 _DigestMod: TypeAlias = str | Callable[[], _HashObject] | ModuleType | None
 
@@ -22,6 +22,7 @@ class _HashObject(Protocol):
     def hexdigest(self) -> str: ...
     def update(self, obj: ReadableBuffer, /) -> None: ...
 
+@disjoint_base
 class HASH:
     @property
     def digest_size(self) -> int: ...