stubgen: Use _typeshed.Incomplete instead of typing.Any (#12449)

Since python/typeshed#7535, typeshed has a new type `_typeshed.Incomplete`,
intended to mark fields that are partly annotated, but need more manual
investigation. This applies to all types generated by stubgen, except
when explicitly annotated with `Any` in the implementation.

mypy/stubgen.py

@@ -566,6 +566,7 @@ def visit_mypy_file(self, o: MypyFile) -> None:
         self.defined_names = find_defined_names(o)
         self.referenced_names = find_referenced_names(o)
         known_imports = {
+            "_typeshed": ["Incomplete"],
             "typing": ["Any", "TypeVar"],
             "collections.abc": ["Generator"],
         }
@@ -689,14 +690,14 @@ def visit_func_def(self, o: FuncDef, is_abstract: bool = False,
             return_name = 'None'
             for expr, in_assignment in all_yield_expressions(o):
                 if expr.expr is not None and not self.is_none_expr(expr.expr):
-                    self.add_typing_import('Any')
-                    yield_name = 'Any'
+                    self.add_typing_import('Incomplete')
+                    yield_name = 'Incomplete'
                 if in_assignment:
-                    self.add_typing_import('Any')
-                    send_name = 'Any'
+                    self.add_typing_import('Incomplete')
+                    send_name = 'Incomplete'
             if has_return_statement(o):
-                self.add_typing_import('Any')
-                return_name = 'Any'
+                self.add_typing_import('Incomplete')
+                return_name = 'Incomplete'
             generator_name = self.typing_name('Generator')
             retname = f'{generator_name}[{yield_name}, {send_name}, {return_name}]'
         elif not has_return_statement(o) and not is_abstract:
@@ -954,18 +955,18 @@ def process_namedtuple(self, lvalue: NameExpr, rvalue: CallExpr) -> None:
             list_items = cast(List[StrExpr], rvalue.args[1].items)
             items = [item.value for item in list_items]
         else:
-            self.add('%s%s: Any' % (self._indent, lvalue.name))
-            self.import_tracker.require_name('Any')
+            self.add('%s%s: Incomplete' % (self._indent, lvalue.name))
+            self.import_tracker.require_name('Incomplete')
             return
         self.import_tracker.require_name('NamedTuple')
         self.add('{}class {}(NamedTuple):'.format(self._indent, lvalue.name))
         if len(items) == 0:
             self.add(' ...\n')
         else:
-            self.import_tracker.require_name('Any')
+            self.import_tracker.require_name('Incomplete')
             self.add('\n')
             for item in items:
-                self.add('{}    {}: Any\n'.format(self._indent, item))
+                self.add('{}    {}: Incomplete\n'.format(self._indent, item))
         self._state = CLASS
 
     def is_alias_expression(self, expr: Expression, top_level: bool = True) -> bool:
@@ -1220,11 +1221,11 @@ def get_str_type_of_node(self, rvalue: Expression,
             return 'bool'
         if can_infer_optional and \
                 isinstance(rvalue, NameExpr) and rvalue.name == 'None':
-            self.add_typing_import('Any')
-            return '{} | None'.format(self.typing_name('Any'))
+            self.add_typing_import('Incomplete')
+            return '{} | None'.format(self.typing_name('Incomplete'))
         if can_be_any:
-            self.add_typing_import('Any')
-            return self.typing_name('Any')
+            self.add_typing_import('Incomplete')
+            return self.typing_name('Incomplete')
         else:
             return ''