[BE]: Enable ruff SLOT checks (#146276)

This enables a check that which a class which only inherits from immutable classes like str, tuple, and NamedTuple, also defined `__slots__` so they don't allocate memory unnecessarily. This also ensure contributors think about how they define their classes with subclass NamedTuples and str, of which we have many in our codebase

Pull Request resolved: https://github.com/pytorch/pytorch/pull/146276
Approved by: https://github.com/aorenste

.github/scripts/pytest_caching_utils.py

@@ -30,6 +30,8 @@ UNZIPPED_CACHES = "unzipped-caches"
 # Since the pr identifier can be based on include user defined text (like a branch name)
 # we hash it to sanitize the input and avoid corner cases
 class PRIdentifier(str):
+    __slots__ = ()
+
     def __new__(cls, value: str) -> "PRIdentifier":
         md5 = hashlib.md5(value.encode("utf-8")).hexdigest()
         return super().__new__(cls, md5)

functorch/dim/reference.py

@@ -139,6 +139,8 @@ def seq(a, b):
 
 
 class isin:
+    __slots__ = ()
+
     def __contains__(self, item):
         for x in self:
             if seq(item, x):
@@ -153,11 +155,11 @@ class isin:
 
 
 class llist(isin, list):
-    pass
+    __slots__ = ()
 
 
 class ltuple(isin, tuple):
-    pass
+    __slots__ = ()
 
 
 empty_dict = {}

pyproject.toml

@@ -148,6 +148,7 @@ select = [
     "RUF019", # unnecessary-key-check
     "RUF024", # from keys mutable
     "RUF026", # default factory kwarg
+    "SLOT",
     "TCH",
     "TRY002", # ban vanilla raise (todo fix NOQAs)
     "TRY203",

test/dynamo/test_functions.py

@@ -4448,7 +4448,7 @@ class DefaultsTests(torch._dynamo.test_case.TestCase):
         self.assertEqual(fn(inputs, x), opt_fn(inputs, x))
 
     def test_udf_tuple(self):
-        class MyTuple(tuple):
+        class MyTuple(tuple):  # noqa: SLOT001
             def len_mulitply_2(self):
                 return len(self) * 2
 
@@ -4475,7 +4475,7 @@ class DefaultsTests(torch._dynamo.test_case.TestCase):
         self.assertTrue(res_tup.checked)
 
     def test_udf_tuple_reconstruction(self):
-        class MyTuple(tuple):
+        class MyTuple(tuple):  # noqa: SLOT001
             pass
 
         def fn(x, klass):

test/torch_np/numpy_tests/core/test_indexing.py

@@ -413,7 +413,7 @@ class TestIndexing(TestCase):
 
         # A tuple subclass should also be an nd-index
         class TupleSubclass(tuple):
-            pass
+            __slots__ = ()
 
         index = ([1], [1])
         index = TupleSubclass(index)

torch/_dynamo/polyfills/pytree.py

@@ -105,6 +105,8 @@ if python_pytree._cxx_pytree_dynamo_traceable:
     __all__ += ["tree_leaves"]
 
     class _Asterisk(str):
+        __slots__ = ()
+
         def __new__(cls) -> Self:
             return super().__new__(cls, "*")
 

torch/_export/serde/union.py

@@ -5,6 +5,7 @@ from dataclasses import fields
 
 
 class _UnionTag(str):
+    __slots__ = ("_cls",)
     _cls: Hashable
 
     @staticmethod

torch/_utils.py

@@ -696,6 +696,8 @@ def render_call(fn, args, kwargs):
 class KeyErrorMessage(str):
     r"""str subclass that returns itself in repr"""
 
+    __slots__ = ()
+
     def __repr__(self):
         return self
 

torch/ao/quantization/fx/_equalize.py

@@ -266,6 +266,8 @@ class EqualizationQConfig(
                                     weight=_WeightEqualizationObserver.with_args(dtype=torch.qint8))
     """
 
+    __slots__ = ()
+
     def __new__(cls, input_activation=torch.nn.Identity, weight=torch.nn.Identity):
         if isinstance(input_activation, nn.Module) or isinstance(weight, nn.Module):
             raise ValueError(

torch/ao/quantization/qconfig.py

@@ -102,6 +102,8 @@ class QConfig(namedtuple("QConfig", ["activation", "weight"])):
 
     """
 
+    __slots__ = ()
+
     def __new__(cls, activation, weight):
         # catch common mistakes
         if isinstance(activation, nn.Module) or isinstance(weight, nn.Module):
@@ -133,6 +135,8 @@ class QConfigDynamic(namedtuple("QConfigDynamic", ["activation", "weight"])):
       my_qconfig = QConfigDynamic(weight=default_observer.with_args(dtype=torch.qint8))
     """
 
+    __slots__ = ()
+
     def __new__(cls, activation=torch.nn.Identity, weight=torch.nn.Identity):
         # catch common mistakes
         if isinstance(weight, nn.Module):

torch/distributed/distributed_c10d.py

@@ -231,7 +231,8 @@ def supports_complex(reduceOp: ReduceOp) -> bool:
     return reduceOp not in denyList
 
 
-class Backend(str):
+# TODO refactor into enum/strenum
+class Backend(str):  # noqa: SLOT000
     """
     An enum-like class for backends.
 

torch/fx/passes/infra/pass_base.py

@@ -18,6 +18,8 @@ class PassResult(namedtuple("PassResult", ["graph_module", "modified"])):
         modified: A flag for if the pass has modified the graph module
     """
 
+    __slots__ = ()
+
     def __new__(cls, graph_module, modified):
         return super().__new__(cls, graph_module, modified)
 

torch/nn/modules/module.py

@@ -40,6 +40,8 @@ T = TypeVar("T", bound="Module")
 class _IncompatibleKeys(
     namedtuple("IncompatibleKeys", ["missing_keys", "unexpected_keys"]),
 ):
+    __slots__ = ()
+
     def __repr__(self):
         if not self.missing_keys and not self.unexpected_keys:
             return "<All keys matched successfully>"

torch/testing/_internal/common_dtype.py

@@ -17,6 +17,8 @@ def _validate_dtypes(*dtypes):
 
 # class for tuples corresponding to a PyTorch dispatch macro
 class _dispatch_dtypes(tuple):
+    __slots__ = ()
+
     def __add__(self, other):
         assert isinstance(other, tuple)
         return _dispatch_dtypes(tuple.__add__(self, other))

torch/testing/_internal/jit_metaprogramming_utils.py

@@ -33,7 +33,7 @@ def unpack_variables(args):
         return args
 
 class dont_convert(tuple):
-    pass
+    __slots__ = ()
 
 non_differentiable = collections.namedtuple('non_differentiable', ['tensor'])
 

torch/torch_version.py

@@ -26,6 +26,8 @@ class TorchVersion(str):
             TorchVersion('1.10.0a') > '1.2.1'
     """
 
+    __slots__ = ()
+
     # fully qualified type names here to appease mypy
     def _convert_to_version(self, inp: Any) -> Any:
         if isinstance(inp, Version):