Add support for PEP 661 (sentinels)

mypy/cache.py

@@ -69,7 +69,7 @@
 from mypy_extensions import u8
 
 # High-level cache layout format
-CACHE_VERSION: Final = 8
+CACHE_VERSION: Final = 9
 
 # Type used internally to represent errors:
 #   (path, line, column, end_line, end_column, severity, message, code)
@@ -308,6 +308,7 @@ def read(cls, data: ReadBuffer) -> CacheMetaEx | None:
 LITERAL_BYTES: Final[Tag] = 5
 LITERAL_FLOAT: Final[Tag] = 6
 LITERAL_COMPLEX: Final[Tag] = 7
+LITERAL_SENTINEL: Final[Tag] = 8
 
 # Collections.
 LIST_GEN: Final[Tag] = 20

mypy/messages.py

@@ -2777,13 +2777,18 @@ def format_literal_value(typ: LiteralType) -> str:
                 modifier += "="
             items.append(f"{item_name!r}{modifier}: {format(item_type)}")
         return f"TypedDict({{{', '.join(items)}}})"
+    elif isinstance(typ, LiteralType) and typ.is_sentinel_literal():
+        return format_literal_value(typ)
     elif isinstance(typ, LiteralType):
         return f"Literal[{format_literal_value(typ)}]"
     elif isinstance(typ, UnionType):
         typ = get_proper_type(ignore_last_known_values(typ))
         if not isinstance(typ, UnionType):
             return format(typ)
         literal_items, union_items = separate_union_literals(typ)
+        sentinel_items = [item for item in literal_items if item.is_sentinel_literal()]
+        literal_items = [item for item in literal_items if not item.is_sentinel_literal()]
+        union_items = [*sentinel_items, *union_items]
 
         # Coalesce multiple Literal[] members. This also changes output order.
         # If there's just one Literal item, retain the original ordering.

mypy/nodes.py

@@ -1414,6 +1414,7 @@ def is_dynamic(self) -> bool:
     "from_module_getattr",
     "has_explicit_value",
     "allow_incompatible_override",
+    "is_sentinel",
 ]
 
 
@@ -1452,6 +1453,7 @@ class Var(SymbolNode):
         "allow_incompatible_override",
         "invalid_partial_type",
         "is_argument",
+        "is_sentinel",
     )
 
     __match_args__ = ("name", "type", "final_value")
@@ -1514,6 +1516,8 @@ def __init__(self, name: str, type: mypy.types.Type | None = None) -> None:
         self.invalid_partial_type = False
         # Is it a variable symbol for a function argument?
         self.is_argument = False
+        # Was this variable created by PEP 661 sentinel()/Sentinel() syntax?
+        self.is_sentinel = False
 
     @property
     def name(self) -> str:
@@ -1596,6 +1600,7 @@ def write(self, data: WriteBuffer) -> None:
                 self.from_module_getattr,
                 self.has_explicit_value,
                 self.allow_incompatible_override,
+                self.is_sentinel,
             ],
         )
         write_literal(data, self.final_value)
@@ -1633,7 +1638,8 @@ def read(cls, data: ReadBuffer) -> Var:
             v.from_module_getattr,
             v.has_explicit_value,
             v.allow_incompatible_override,
-        ) = read_flags(data, num_flags=19)
+            v.is_sentinel,
+        ) = read_flags(data, num_flags=20)
         tag = read_tag(data)
         if tag == LITERAL_COMPLEX:
             v.final_value = complex(read_float_bare(data), read_float_bare(data))

mypy/semanal.py

@@ -269,6 +269,7 @@
     OVERRIDE_DECORATOR_NAMES,
     PROTOCOL_NAMES,
     REVEAL_TYPE_NAMES,
+    SENTINEL_TYPE_NAMES,
     TPDICT_NAMES,
     TYPE_ALIAS_NAMES,
     TYPE_CHECK_ONLY_NAMES,
@@ -287,6 +288,7 @@
     ParamSpecType,
     PlaceholderType,
     ProperType,
+    SentinelValue,
     TrivialSyntheticTypeTranslator,
     TupleType,
     Type,
@@ -3357,9 +3359,17 @@ def visit_assignment_stmt(self, s: AssignmentStmt) -> None:
         # may be set to True while there were still placeholders due to forward refs.
         s.is_alias_def = False
 
+        sentinel_definition = self.is_sentinel_declaration(s)
+        if sentinel_definition and self.is_existing_final_lvalue(s):
+            sentinel_definition = False
+
         # OK, this is a regular assignment, perform the necessary analysis steps.
         s.is_final_def = self.unwrap_final(s)
+        if sentinel_definition:
+            s.is_final_def = True
         self.analyze_lvalues(s)
+        if sentinel_definition:
+            self.setup_sentinel_var(s)
         self.check_final_implicit_def(s)
         self.store_final_status(s)
         self.check_classvar(s)
@@ -3372,6 +3382,60 @@ def visit_assignment_stmt(self, s: AssignmentStmt) -> None:
         self.process__deletable__(s)
         self.process__slots__(s)
 
+    def is_sentinel_declaration(self, s: AssignmentStmt) -> bool:
+        """Does this assignment define a PEP 661 sentinel singleton?"""
+        if self.is_func_scope() or s.unanalyzed_type is not None:
+            return False
+        if len(s.lvalues) != 1 or not isinstance(s.lvalues[0], NameExpr):
+            return False
+        if not isinstance(s.rvalue, CallExpr):
+            return False
+        call = s.rvalue
+        if not isinstance(call.callee, RefExpr):
+            return False
+        if call.callee.fullname not in SENTINEL_TYPE_NAMES:
+            return False
+        if not call.args or call.arg_kinds[0] != ARG_POS or not isinstance(call.args[0], StrExpr):
+            return False
+        return True
+
+    def is_existing_final_lvalue(self, s: AssignmentStmt) -> bool:
+        if len(s.lvalues) != 1 or not isinstance(s.lvalues[0], NameExpr):
+            return False
+        name = s.lvalues[0].name
+        existing = self.current_symbol_table().get(name)
+        if existing is not None and is_final_node(existing.node):
+            return True
+        return self.is_alias_for_final_name(name)
+
+    def setup_sentinel_var(self, s: AssignmentStmt) -> None:
+        lvalue = s.lvalues[0]
+        assert isinstance(lvalue, NameExpr)
+        if not isinstance(lvalue.node, Var):
+            return
+        var = lvalue.node
+        var.is_sentinel = True
+        typ = self.sentinel_type_for_var(var, s.rvalue)
+        if typ is not None:
+            s.type = typ
+
+    def sentinel_type_for_var(self, var: Var, rvalue: Expression) -> Instance | None:
+        assert isinstance(rvalue, CallExpr)
+        callee = rvalue.callee
+        assert isinstance(callee, RefExpr)
+        typ = self.named_type_or_none(callee.fullname)
+        if typ is None:
+            return None
+        name = f"{self.type.name}.{var.name}" if self.type is not None else var.name
+        return typ.copy_modified(
+            last_known_value=LiteralType(
+                SentinelValue(var.fullname, name),
+                fallback=typ,
+                line=rvalue.line,
+                column=rvalue.column,
+            )
+        )
+
     def analyze_identity_global_assignment(self, s: AssignmentStmt) -> bool:
         """Special case 'X = X' in global scope.
 
@@ -3536,6 +3600,8 @@ def is_type_ref(self, rv: Expression, bare: bool = False) -> bool:
             # Assignment color = Color['RED'] defines a variable, not an alias.
             return not rv.node.is_enum
         if isinstance(rv.node, Var):
+            if rv.node.is_sentinel:
+                return True
             return rv.node.fullname in NEVER_NAMES
 
         if isinstance(rv, NameExpr):
@@ -4717,6 +4783,7 @@ def store_declared_types(self, lvalue: Lvalue, typ: Type) -> None:
                     var.is_final
                     and isinstance(typ, Instance)
                     and typ.last_known_value
+                    and not isinstance(typ.last_known_value.value, SentinelValue)
                     and (not self.type or not self.type.is_enum)
                 ):
                     var.final_value = typ.last_known_value.value

mypy/test/testtypes.py

@@ -37,6 +37,7 @@
     NoneType,
     Overloaded,
     ProperType,
+    SentinelValue,
     TupleType,
     Type,
     TypeOfAny,
@@ -65,6 +66,15 @@ def setUp(self) -> None:
     def test_any(self) -> None:
         assert_equal(str(AnyType(TypeOfAny.special_form)), "Any")
 
+    def test_sentinel_literal_json_roundtrip(self) -> None:
+        literal = LiteralType(SentinelValue("__main__.MISSING", "MISSING"), self.fx.a)
+        assert_equal(str(literal), "MISSING")
+        data = literal.serialize()
+        assert isinstance(data, dict)
+        roundtrip = LiteralType.deserialize(data)
+        self.assertEqual(roundtrip.value, literal.value)
+        self.assertEqual(roundtrip.fallback.type_ref, self.fx.a.type.fullname)
+
     def test_simple_unbound_type(self) -> None:
         u = UnboundType("Foo")
         assert_equal(str(u), "Foo?")

mypy/typeanal.py

@@ -1000,6 +1000,16 @@ def analyze_unbound_type_without_type_info(
                 column=t.column,
             )
 
+        if isinstance(sym.node, Var) and sym.node.is_sentinel:
+            typ = get_proper_type(sym.node.type)
+            if isinstance(typ, Instance) and typ.last_known_value is not None:
+                return LiteralType(
+                    value=typ.last_known_value.value,
+                    fallback=typ.last_known_value.fallback,
+                    line=t.line,
+                    column=t.column,
+                )
+
         # None of the above options worked. We parse the args (if there are any)
         # to make sure there are no remaining semanal-only types, then give up.
         t = t.copy_modified(args=self.anal_array(t.args))

mypy/typeops.py

@@ -1025,7 +1025,7 @@ def is_singleton_identity_type(typ: ProperType) -> bool:
             or (typ.type.fullname in NOT_IMPLEMENTED_TYPE_NAMES)
         )
     if isinstance(typ, LiteralType):
-        return typ.is_enum_literal() or isinstance(typ.value, bool)
+        return typ.is_enum_literal() or typ.is_sentinel_literal() or isinstance(typ.value, bool)
     if isinstance(typ, TypeType) and isinstance(typ.item, Instance) and typ.item.type.is_final:
         return True
     if isinstance(typ, FunctionLike) and typ.is_type_obj() and typ.type_object().is_final:

mypy/types.py

@@ -9,6 +9,7 @@
     Any,
     ClassVar,
     Final,
+    NamedTuple,
     NewType,
     TypeAlias as _TypeAlias,
     TypeGuard,
@@ -33,6 +34,7 @@
     EXTRA_ATTRS,
     LIST_GEN,
     LITERAL_NONE,
+    LITERAL_SENTINEL,
     ReadBuffer,
     Tag,
     WriteBuffer,
@@ -64,6 +66,7 @@
 
 JsonDict: _TypeAlias = dict[str, Any]
 
+
 # The set of all valid expressions that can currently be contained
 # inside of a Literal[...].
 #
@@ -94,7 +97,12 @@
 #
 # Note: Float values are only used internally. They are not accepted within
 # Literal[...].
-LiteralValue: _TypeAlias = int | str | bool | float
+class SentinelValue(NamedTuple):
+    fullname: str
+    name: str
+
+
+LiteralValue: _TypeAlias = int | str | bool | float | SentinelValue
 
 
 TUPLE_NAMES: Final = ("builtins.tuple", "typing.Tuple")
@@ -109,6 +117,12 @@
     "typing_extensions.TypeVarTuple",
 )
 
+SENTINEL_TYPE_NAMES: Final = (
+    "builtins.sentinel",
+    "typing_extensions.sentinel",
+    "typing_extensions.Sentinel",
+)
+
 TYPED_NAMEDTUPLE_NAMES: Final = ("typing.NamedTuple", "typing_extensions.NamedTuple")
 
 # Supported names of TypedDict type constructors.
@@ -3226,11 +3240,15 @@ def __init__(
     #       almost no test cases where we would redundantly compute
     #       `can_be_false`/`can_be_true`.
     def can_be_false_default(self) -> bool:
+        if isinstance(self.value, SentinelValue):
+            return False
         if self.fallback.type.is_enum:
             return self.fallback.can_be_false
         return not self.value
 
     def can_be_true_default(self) -> bool:
+        if isinstance(self.value, SentinelValue):
+            return True
         if self.fallback.type.is_enum:
             return self.fallback.can_be_true
         return bool(self.value)
@@ -3251,13 +3269,19 @@ def __eq__(self, other: object) -> bool:
     def is_enum_literal(self) -> bool:
         return self.fallback.type.is_enum
 
+    def is_sentinel_literal(self) -> bool:
+        return isinstance(self.value, SentinelValue)
+
     def value_repr(self) -> str:
         """Returns the string representation of the underlying type.
 
         This function is almost equivalent to running `repr(self.value)`,
         except it includes some additional logic to correctly handle cases
         where the value is a string, byte string, a unicode string, or an enum.
         """
+        if isinstance(self.value, SentinelValue):
+            return self.value.name
+
         raw = repr(self.value)
         fallback_name = self.fallback.type.fullname
 
@@ -3276,29 +3300,41 @@ def value_repr(self) -> str:
             return raw
 
     def serialize(self) -> JsonDict | str:
-        return {
-            ".class": "LiteralType",
-            "value": self.value,
-            "fallback": self.fallback.serialize(),
-        }
+        value: LiteralValue | JsonDict = self.value
+        if isinstance(value, SentinelValue):
+            value = {".class": "SentinelValue", "fullname": value.fullname, "name": value.name}
+        return {".class": "LiteralType", "value": value, "fallback": self.fallback.serialize()}
 
     @classmethod
     def deserialize(cls, data: JsonDict) -> LiteralType:
         assert data[".class"] == "LiteralType"
-        return LiteralType(value=data["value"], fallback=Instance.deserialize(data["fallback"]))
+        value = data["value"]
+        if isinstance(value, dict):
+            assert value[".class"] == "SentinelValue"
+            value = SentinelValue(value["fullname"], value["name"])
+        return LiteralType(value=value, fallback=Instance.deserialize(data["fallback"]))
 
     def write(self, data: WriteBuffer) -> None:
         write_tag(data, LITERAL_TYPE)
         self.fallback.write(data)
-        write_literal(data, self.value)
+        if isinstance(self.value, SentinelValue):
+            write_tag(data, LITERAL_SENTINEL)
+            write_str(data, self.value.fullname)
+            write_str(data, self.value.name)
+        else:
+            write_literal(data, self.value)
         write_tag(data, END_TAG)
 
     @classmethod
     def read(cls, data: ReadBuffer) -> LiteralType:
         assert read_tag(data) == INSTANCE
         fallback = Instance.read(data)
         tag = read_tag(data)
-        ret = LiteralType(read_literal(data, tag), fallback)
+        if tag == LITERAL_SENTINEL:
+            value = SentinelValue(read_str(data), read_str(data))
+        else:
+            value = read_literal(data, tag)
+        ret = LiteralType(value, fallback)
         assert read_tag(data) == END_TAG
         return ret
 
@@ -3958,6 +3994,8 @@ def visit_raw_expression_type(self, t: RawExpressionType, /) -> str:
         return repr(t.literal_value)
 
     def visit_literal_type(self, t: LiteralType, /) -> str:
+        if isinstance(t.value, SentinelValue):
+            return t.value_repr()
         return f"Literal[{t.value_repr()}]"
 
     def visit_union_type(self, t: UnionType, /) -> str: