[mypyc] Generate __getattr__ wrapper

mypyc/codegen/emitclass.py

@@ -39,6 +39,12 @@ def native_slot(cl: ClassIR, fn: FuncIR, emitter: Emitter) -> str:
     return f"{NATIVE_PREFIX}{fn.cname(emitter.names)}"
 
 
+def dunder_attr_slot(cl: ClassIR, fn: FuncIR, emitter: Emitter) -> str:
+    wrapper_fn = cl.get_method(fn.name + "__wrapper")
+    assert wrapper_fn
+    return f"{NATIVE_PREFIX}{wrapper_fn.cname(emitter.names)}"
+
+
 # We maintain a table from dunder function names to struct slots they
 # correspond to and functions that generate a wrapper (if necessary)
 # and return the function name to stick in the slot.
@@ -55,6 +61,7 @@ def native_slot(cl: ClassIR, fn: FuncIR, emitter: Emitter) -> str:
     "__iter__": ("tp_iter", native_slot),
     "__hash__": ("tp_hash", generate_hash_wrapper),
     "__get__": ("tp_descr_get", generate_get_wrapper),
+    "__getattr__": ("tp_getattro", dunder_attr_slot),
 }
 
 AS_MAPPING_SLOT_DEFS: SlotTable = {

mypyc/ir/ops.py

@@ -1221,6 +1221,7 @@ def __init__(
         var_arg_idx: int = -1,
         *,
         is_pure: bool = False,
+        returns_null: bool = False,
     ) -> None:
         self.error_kind = error_kind
         super().__init__(line)
@@ -1235,7 +1236,10 @@ def __init__(
         # and all the arguments are immutable. Pure functions support
         # additional optimizations. Pure functions never fail.
         self.is_pure = is_pure
-        if is_pure:
+        # The function might return a null value that does not indicate
+        # an error.
+        self.returns_null = returns_null
+        if is_pure or returns_null:
             assert error_kind == ERR_NEVER
 
     def sources(self) -> list[Value]:

mypyc/irbuild/builder.py

@@ -1241,6 +1241,7 @@ def enter_method(
         ret_type: RType,
         fn_info: FuncInfo | str = "",
         self_type: RType | None = None,
+        internal: bool = False,
     ) -> Iterator[None]:
         """Generate IR for a method.
 
@@ -1268,7 +1269,7 @@ def enter_method(
             sig = FuncSignature(args, ret_type)
             name = self.function_name_stack.pop()
             class_ir = self.class_ir_stack.pop()
-            decl = FuncDecl(name, class_ir.name, self.module_name, sig)
+            decl = FuncDecl(name, class_ir.name, self.module_name, sig, internal=internal)
             ir = FuncIR(decl, arg_regs, blocks)
             class_ir.methods[name] = ir
             class_ir.method_decls[name] = ir.decl

mypyc/irbuild/function.py

@@ -42,6 +42,7 @@
 )
 from mypyc.ir.ops import (
     BasicBlock,
+    ComparisonOp,
     GetAttr,
     Integer,
     LoadAddress,
@@ -81,7 +82,7 @@
     dict_new_op,
     exact_dict_set_item_op,
 )
-from mypyc.primitives.generic_ops import py_setattr_op
+from mypyc.primitives.generic_ops import generic_getattr, py_setattr_op
 from mypyc.primitives.misc_ops import register_function
 from mypyc.primitives.registry import builtin_names
 from mypyc.sametype import is_same_method_signature, is_same_type
@@ -364,6 +365,56 @@ def gen_func_ir(
     return (func_ir, func_reg)
 
 
+def generate_getattr_wrapper(builder: IRBuilder, cdef: ClassDef, getattr: FuncDef) -> None:
+    """
+    Generate a wrapper function for __getattr__ that can be put into the tp_getattro slot.
+    The wrapper takes one argument besides self which is the attribute name.
+    It first checks if the name matches any of the attributes of this class.
+    If it does, it returns that attribute. If none match, it calls __getattr__.
+
+    __getattr__ is not supported in classes that allow interpreted subclasses because the
+    tp_getattro slot is inherited by subclasses and if the subclass overrides __getattr__,
+    the override would be ignored in our wrapper. TODO: To support this, the wrapper would
+    have to check type of self and if it's not the compiled class, resolve "__getattr__" against
+    the type at runtime and call the returned method, like _Py_slot_tp_getattr_hook in cpython.
+
+    __getattr__ is not supported in classes which inherit from non-native classes because those
+    have __dict__ which currently has some strange interactions when class attributes and
+    variables are assigned through __dict__ vs. through regular attribute access. Allowing
+    __getattr__ on top of that could be problematic.
+    """
+    name = getattr.name + "__wrapper"
+    ir = builder.mapper.type_to_ir[cdef.info]
+    line = getattr.line
+
+    error_base = f'"__getattr__" not supported in class "{cdef.name}" because '
+    if ir.allow_interpreted_subclasses:
+        builder.error(error_base + "it allows interpreted subclasses", line)
+    if ir.inherits_python:
+        builder.error(error_base + "it inherits from a non-native class", line)
+
+    with builder.enter_method(ir, name, object_rprimitive, internal=True):
+        attr_arg = builder.add_argument("attr", object_rprimitive)
+        generic_getattr_result = builder.call_c(generic_getattr, [builder.self(), attr_arg], line)
+
+        return_generic, call_getattr = BasicBlock(), BasicBlock()
+        null = Integer(0, object_rprimitive, line)
+        got_generic = builder.add(
+            ComparisonOp(generic_getattr_result, null, ComparisonOp.NEQ, line)
+        )
+        builder.add_bool_branch(got_generic, return_generic, call_getattr)
+
+        builder.activate_block(return_generic)
+        builder.add(Return(generic_getattr_result, line))
+
+        builder.activate_block(call_getattr)
+        # No attribute matched so call user-provided __getattr__.
+        getattr_result = builder.gen_method_call(
+            builder.self(), getattr.name, [attr_arg], object_rprimitive, line
+        )
+        builder.add(Return(getattr_result, line))
+
+
 def handle_ext_method(builder: IRBuilder, cdef: ClassDef, fdef: FuncDef) -> None:
     # Perform the function of visit_method for methods inside extension classes.
     name = fdef.name
@@ -430,6 +481,9 @@ def handle_ext_method(builder: IRBuilder, cdef: ClassDef, fdef: FuncDef) -> None
         class_ir.glue_methods[(class_ir, name)] = f
         builder.functions.append(f)
 
+    if fdef.name == "__getattr__":
+        generate_getattr_wrapper(builder, cdef, fdef)
+
 
 def handle_non_ext_method(
     builder: IRBuilder, non_ext: NonExtClassInfo, cdef: ClassDef, fdef: FuncDef

mypyc/irbuild/ll_builder.py

@@ -2048,6 +2048,7 @@ def call_c(
                 line,
                 var_arg_idx,
                 is_pure=desc.is_pure,
+                returns_null=desc.returns_null,
             )
         )
         if desc.is_borrowed:
@@ -2131,6 +2132,7 @@ def primitive_op(
                 desc.extra_int_constants,
                 desc.priority,
                 is_pure=desc.is_pure,
+                returns_null=False,
             )
             return self.call_c(c_desc, args, line, result_type=result_type)
 

mypyc/lib-rt/CPy.h

@@ -949,6 +949,10 @@ PyObject *CPy_GetANext(PyObject *aiter);
 void CPy_SetTypeAliasTypeComputeFunction(PyObject *alias, PyObject *compute_value);
 void CPyTrace_LogEvent(const char *location, const char *line, const char *op, const char *details);
 
+static inline PyObject *CPyObject_GenericGetAttr(PyObject *self, PyObject *name) {
+    return _PyObject_GenericGetAttrWithDict(self, name, NULL, 1);
+}
+
 #if CPY_3_11_FEATURES
 PyObject *CPy_GetName(PyObject *obj);
 #endif

mypyc/primitives/generic_ops.py

@@ -401,3 +401,12 @@
     c_function_name="CPy_GetName",
     error_kind=ERR_MAGIC,
 )
+
+# look-up name in tp_dict but don't raise AttributeError on failure
+generic_getattr = custom_op(
+    arg_types=[object_rprimitive, object_rprimitive],
+    return_type=object_rprimitive,
+    c_function_name="CPyObject_GenericGetAttr",
+    error_kind=ERR_NEVER,
+    returns_null=True,
+)

mypyc/primitives/registry.py

@@ -61,6 +61,7 @@ class CFunctionDescription(NamedTuple):
     extra_int_constants: list[tuple[int, RType]]
     priority: int
     is_pure: bool
+    returns_null: bool
 
 
 # A description for C load operations including LoadGlobal and LoadAddress
@@ -253,6 +254,7 @@ def custom_op(
     is_borrowed: bool = False,
     *,
     is_pure: bool = False,
+    returns_null: bool = False,
 ) -> CFunctionDescription:
     """Create a one-off CallC op that can't be automatically generated from the AST.
 
@@ -274,6 +276,7 @@ def custom_op(
         extra_int_constants,
         0,
         is_pure=is_pure,
+        returns_null=returns_null,
     )
 
 

mypyc/test-data/irbuild-classes.test

@@ -2088,3 +2088,128 @@ class NonNative:
 @mypyc_attr(free_list_len=1, allow_interpreted_subclasses=True)  # E: "free_list_len" can't be used in a class that allows interpreted subclasses
 class InterpSub:
     pass
+
+[case testUnsupportedGetAttr]
+from mypy_extensions import mypyc_attr
+
+@mypyc_attr(allow_interpreted_subclasses=True)
+class AllowsInterpreted:
+    def __getattr__(self, attr: str) -> object:  # E: "__getattr__" not supported in class "AllowsInterpreted" because it allows interpreted subclasses
+        return 0
+
+class InheritsInterpreted(dict):
+    def __getattr__(self, attr: str) -> object:  # E: "__getattr__" not supported in class "InheritsInterpreted" because it inherits from a non-native class
+        return 0
+
+@mypyc_attr(native_class=False)
+class NonNative:
+    pass
+
+class InheritsNonNative(NonNative):
+    def __getattr__(self, attr: str) -> object:  # E: "__getattr__" not supported in class "InheritsNonNative" because it inherits from a non-native class
+        return 0
+
+[case testGetAttr]
+from typing import ClassVar
+
+class GetAttr:
+    class_var = "x"
+    class_var_annotated: ClassVar[int] = 99
+
+    def __init__(self, regular_attr: int):
+        self.regular_attr = regular_attr
+
+    def __getattr__(self, attr: str) -> object:
+        return attr
+
+    def method(self) -> int:
+        return 0
+
+def test_getattr() -> list[object]:
+    i = GetAttr(42)
+    one = i.one
+    two = i.regular_attr
+    three = i.class_var
+    four = i.class_var_annotated
+    five = i.method()
+    return [one, two, three, four, five]
+
+[typing fixtures/typing-full.pyi]
+[out]
+def GetAttr.__init__(self, regular_attr):
+    self :: __main__.GetAttr
+    regular_attr :: int
+L0:
+    self.regular_attr = regular_attr
+    return 1
+def GetAttr.__getattr__(self, attr):
+    self :: __main__.GetAttr
+    attr :: str
+L0:
+    return attr
+def GetAttr.__getattr____wrapper(__mypyc_self__, attr):
+    __mypyc_self__ :: __main__.GetAttr
+    attr, r0 :: object
+    r1 :: bit
+    r2 :: str
+    r3 :: object
+L0:
+    r0 = CPyObject_GenericGetAttr(__mypyc_self__, attr)
+    r1 = r0 != 0
+    if r1 goto L1 else goto L2 :: bool
+L1:
+    return r0
+L2:
+    r2 = cast(str, attr)
+    r3 = __mypyc_self__.__getattr__(r2)
+    return r3
+def GetAttr.method(self):
+    self :: __main__.GetAttr
+L0:
+    return 0
+def GetAttr.__mypyc_defaults_setup(__mypyc_self__):
+    __mypyc_self__ :: __main__.GetAttr
+    r0 :: str
+L0:
+    r0 = 'x'
+    __mypyc_self__.class_var = r0
+    return 1
+def test_getattr():
+    r0, i :: __main__.GetAttr
+    r1 :: str
+    r2, one :: object
+    r3, two :: int
+    r4, three, r5 :: str
+    r6 :: object
+    r7, four, r8, five :: int
+    r9 :: list
+    r10, r11, r12 :: object
+    r13 :: ptr
+L0:
+    r0 = GetAttr(84)
+    i = r0
+    r1 = 'one'
+    r2 = CPyObject_GetAttr(i, r1)
+    one = r2
+    r3 = i.regular_attr
+    two = r3
+    r4 = i.class_var
+    three = r4
+    r5 = 'class_var_annotated'
+    r6 = CPyObject_GetAttr(i, r5)
+    r7 = unbox(int, r6)
+    four = r7
+    r8 = i.method()
+    five = r8
+    r9 = PyList_New(5)
+    r10 = box(int, two)
+    r11 = box(int, four)
+    r12 = box(int, five)
+    r13 = list_items r9
+    buf_init_item r13, 0, one
+    buf_init_item r13, 1, r10
+    buf_init_item r13, 2, three
+    buf_init_item r13, 3, r11
+    buf_init_item r13, 4, r12
+    keep_alive r9
+    return r9