stubgen: Fix generated dataclass __init__ signature (python#16906)

Fixes python#16811

stubgen was swallowing default values for `__init__` methods generated
by the dataclass plugin making their signature incorrect. This is
because the plugin does not include the argument's initializer in the
generated signature. I changed it to include a dummy ellipsis so that
stubgen can generate correct code.
I also fixed arguments added by the dataclass plugin with the invalid
names `*` and `**` to have the valid and unique names `*generated_args`
and `**generated_kwargs` (with extra underscores to make them unique if
necessary). This removes the need for the hack to special case them in
stubgen and is less confusing for someone looking at them in a stub
file.

Author: hamdanal

mypy/plugins/dataclasses.py

@@ -24,6 +24,7 @@
     Context,
     DataclassTransformSpec,
     Decorator,
+    EllipsisExpr,
     Expression,
     FuncDef,
     FuncItem,
@@ -149,13 +150,13 @@ def to_argument(
         return Argument(
             variable=self.to_var(current_info),
             type_annotation=self.expand_type(current_info),
-            initializer=None,
+            initializer=EllipsisExpr() if self.has_default else None,  # Only used by stubgen
             kind=arg_kind,
         )
 
     def expand_type(self, current_info: TypeInfo) -> Type | None:
         if self.type is not None and self.info.self_type is not None:
-            # In general, it is not safe to call `expand_type()` during semantic analyzis,
+            # In general, it is not safe to call `expand_type()` during semantic analysis,
             # however this plugin is called very late, so all types should be fully ready.
             # Also, it is tricky to avoid eager expansion of Self types here (e.g. because
             # we serialize attributes).
@@ -269,11 +270,17 @@ def transform(self) -> bool:
                     if arg.kind == ARG_POS:
                         arg.kind = ARG_OPT
 
-                nameless_var = Var("")
+                existing_args_names = {arg.variable.name for arg in args}
+                gen_args_name = "generated_args"
+                while gen_args_name in existing_args_names:
+                    gen_args_name += "_"
+                gen_kwargs_name = "generated_kwargs"
+                while gen_kwargs_name in existing_args_names:
+                    gen_kwargs_name += "_"
                 args = [
-                    Argument(nameless_var, AnyType(TypeOfAny.explicit), None, ARG_STAR),
+                    Argument(Var(gen_args_name), AnyType(TypeOfAny.explicit), None, ARG_STAR),
                     *args,
-                    Argument(nameless_var, AnyType(TypeOfAny.explicit), None, ARG_STAR2),
+                    Argument(Var(gen_kwargs_name), AnyType(TypeOfAny.explicit), None, ARG_STAR2),
                 ]
 
             add_method_to_class(

mypy/stubgen.py

@@ -537,17 +537,6 @@ def _get_func_args(self, o: FuncDef, ctx: FunctionContext) -> list[ArgSig]:
             if new_args is not None:
                 args = new_args
 
-        is_dataclass_generated = (
-            self.analyzed and self.processing_dataclass and o.info.names[o.name].plugin_generated
-        )
-        if o.name == "__init__" and is_dataclass_generated and "**" in [a.name for a in args]:
-            # The dataclass plugin generates invalid nameless "*" and "**" arguments
-            new_name = "".join(a.name.strip("*") for a in args)
-            for arg in args:
-                if arg.name == "*":
-                    arg.name = f"*{new_name}_"  # this name is guaranteed to be unique
-                elif arg.name == "**":
-                    arg.name = f"**{new_name}__"  # same here
         return args
 
     def _get_func_return(self, o: FuncDef, ctx: FunctionContext) -> str | None:

test-data/unit/check-dataclasses.test

@@ -1610,10 +1610,16 @@ B: Any
 @dataclass
 class A(B):
     a: int
+@dataclass
+class C(B):
+    generated_args: int
+    generated_kwargs: int
 
 A(a=1, b=2)
 A(1)
 A(a="foo")  # E: Argument "a" to "A" has incompatible type "str"; expected "int"
+C(generated_args="foo", generated_kwargs="bar")  # E: Argument "generated_args" to "C" has incompatible type "str"; expected "int" \
+                                                 # E: Argument "generated_kwargs" to "C" has incompatible type "str"; expected "int"
 [builtins fixtures/dataclasses.pyi]
 
 [case testDataclassesCallableFrozen]

test-data/unit/stubgen.test

@@ -4083,20 +4083,21 @@ class W: ...
 class V: ...
 
 [case testDataclass_semanal]
-from dataclasses import dataclass, InitVar
+from dataclasses import InitVar, dataclass, field
 from typing import ClassVar
 
 @dataclass
 class X:
     a: int
-    b: str = "hello"
-    c: ClassVar
-    d: ClassVar = 200
+    b: InitVar[str]
+    c: str = "hello"
+    d: ClassVar
+    e: ClassVar = 200
     f: list[int] = field(init=False, default_factory=list)
     g: int = field(default=2, kw_only=True)
     h: int = 1
-    i: InitVar[str]
-    j: InitVar = 100
+    i: InitVar = 100
+    j: list[int] = field(default_factory=list)
     non_field = None
 
 @dataclass(init=False, repr=False, frozen=True)
@@ -4109,23 +4110,24 @@ from typing import ClassVar
 @dataclass
 class X:
     a: int
-    b: str = ...
-    c: ClassVar
-    d: ClassVar = ...
+    b: InitVar[str]
+    c: str = ...
+    d: ClassVar
+    e: ClassVar = ...
     f: list[int] = ...
     g: int = ...
     h: int = ...
-    i: InitVar[str]
-    j: InitVar = ...
+    i: InitVar = ...
+    j: list[int] = ...
     non_field = ...
-    def __init__(self, a, b, f, g, h, i, j) -> None: ...
+    def __init__(self, a, b, c=..., *, g=..., h=..., i=..., j=...) -> None: ...
 
 @dataclass(init=False, repr=False, frozen=True)
 class Y: ...
 
 [case testDataclassWithKwOnlyField_semanal]
 # flags: --python-version=3.10
-from dataclasses import dataclass, InitVar, KW_ONLY
+from dataclasses import dataclass, field, InitVar, KW_ONLY
 from typing import ClassVar
 
 @dataclass
@@ -4162,7 +4164,7 @@ class X:
     i: InitVar[str]
     j: InitVar = ...
     non_field = ...
-    def __init__(self, a, b, f, g, *, h, i, j) -> None: ...
+    def __init__(self, a, b=..., *, g=..., h=..., i, j=...) -> None: ...
 
 @dataclass(init=False, repr=False, frozen=True)
 class Y: ...
@@ -4193,14 +4195,29 @@ import missing
 class X(missing.Base):
     a: int
 
+@dataclass
+class Y(missing.Base):
+    generated_args: str
+    generated_args_: str
+    generated_kwargs: float
+    generated_kwargs_: float
+
 [out]
 import missing
 from dataclasses import dataclass
 
 @dataclass
 class X(missing.Base):
     a: int
-    def __init__(self, *selfa_, a, **selfa__) -> None: ...
+    def __init__(self, *generated_args, a, **generated_kwargs) -> None: ...
+
+@dataclass
+class Y(missing.Base):
+    generated_args: str
+    generated_args_: str
+    generated_kwargs: float
+    generated_kwargs_: float
+    def __init__(self, *generated_args__, generated_args, generated_args_, generated_kwargs, generated_kwargs_, **generated_kwargs__) -> None: ...
 
 [case testAlwaysUsePEP604Union]
 import typing