[mypyc] Refactor: add two list primitive ops (#17058)

Add ops for getting list size and a pointer to list item data.

This simplifies the IR generated in the main irbuild pass. Continue work
on mypyc/mypyc#854.

mypyc/irbuild/ll_builder.py

@@ -78,10 +78,8 @@
     int_op_to_id,
 )
 from mypyc.ir.rtypes import (
-    PyListObject,
     PyObject,
     PySetObject,
-    PyVarObject,
     RArray,
     RInstance,
     RPrimitive,
@@ -163,8 +161,14 @@
     ssize_t_to_int_op,
     uint8_overflow,
 )
-from mypyc.primitives.list_ops import list_build_op, list_extend_op, new_list_op
-from mypyc.primitives.misc_ops import bool_op, buf_init_item, fast_isinstance_op, none_object_op
+from mypyc.primitives.list_ops import list_build_op, list_extend_op, list_items, new_list_op
+from mypyc.primitives.misc_ops import (
+    bool_op,
+    buf_init_item,
+    fast_isinstance_op,
+    none_object_op,
+    var_object_size,
+)
 from mypyc.primitives.registry import (
     ERR_NEG_INT,
     CFunctionDescription,
@@ -1623,8 +1627,7 @@ def new_list_op(self, values: list[Value], line: int) -> Value:
         if not values:
             return result_list
         args = [self.coerce(item, object_rprimitive, line) for item in values]
-        ob_item_ptr = self.add(GetElementPtr(result_list, PyListObject, "ob_item", line))
-        ob_item_base = self.add(LoadMem(pointer_rprimitive, ob_item_ptr, line))
+        ob_item_base = self.add(PrimitiveOp([result_list], list_items, line))
         for i in range(len(values)):
             self.primitive_op(
                 buf_init_item, [ob_item_base, Integer(i, c_pyssize_t_rprimitive), args[i]], line
@@ -2165,9 +2168,7 @@ def builtin_len(self, val: Value, line: int, use_pyssize_t: bool = False) -> Val
         typ = val.type
         size_value = None
         if is_list_rprimitive(typ) or is_tuple_rprimitive(typ) or is_bytes_rprimitive(typ):
-            elem_address = self.add(GetElementPtr(val, PyVarObject, "ob_size"))
-            size_value = self.add(LoadMem(c_pyssize_t_rprimitive, elem_address))
-            self.add(KeepAlive([val]))
+            size_value = self.primitive_op(var_object_size, [val], line)
         elif is_set_rprimitive(typ):
             elem_address = self.add(GetElementPtr(val, PySetObject, "used"))
             size_value = self.add(LoadMem(c_pyssize_t_rprimitive, elem_address))

mypyc/lower/list_ops.py

@@ -1,8 +1,13 @@
 from __future__ import annotations
 
 from mypyc.common import PLATFORM_SIZE
-from mypyc.ir.ops import Integer, IntOp, SetMem, Value
-from mypyc.ir.rtypes import c_pyssize_t_rprimitive, object_rprimitive, pointer_rprimitive
+from mypyc.ir.ops import GetElementPtr, Integer, IntOp, LoadMem, SetMem, Value
+from mypyc.ir.rtypes import (
+    PyListObject,
+    c_pyssize_t_rprimitive,
+    object_rprimitive,
+    pointer_rprimitive,
+)
 from mypyc.irbuild.ll_builder import LowLevelIRBuilder
 from mypyc.lower.registry import lower_primitive_op
 
@@ -32,3 +37,9 @@ def buf_init_item(builder: LowLevelIRBuilder, args: list[Value], line: int) -> V
             )
         )
     return builder.add(SetMem(object_rprimitive, ptr, value, line))
+
+
+@lower_primitive_op("list_items")
+def list_items(builder: LowLevelIRBuilder, args: list[Value], line: int) -> Value:
+    ob_item_ptr = builder.add(GetElementPtr(args[0], PyListObject, "ob_item", line))
+    return builder.add(LoadMem(pointer_rprimitive, ob_item_ptr, line))

mypyc/lower/misc_ops.py

@@ -0,0 +1,12 @@
+from __future__ import annotations
+
+from mypyc.ir.ops import GetElementPtr, LoadMem, Value
+from mypyc.ir.rtypes import PyVarObject, c_pyssize_t_rprimitive
+from mypyc.irbuild.ll_builder import LowLevelIRBuilder
+from mypyc.lower.registry import lower_primitive_op
+
+
+@lower_primitive_op("var_object_size")
+def var_object_size(builder: LowLevelIRBuilder, args: list[Value], line: int) -> Value:
+    elem_address = builder.add(GetElementPtr(args[0], PyVarObject, "ob_size"))
+    return builder.add(LoadMem(c_pyssize_t_rprimitive, elem_address))

mypyc/lower/registry.py

@@ -23,5 +23,4 @@ def wrapper(f: LowerFunc) -> LowerFunc:
 
 
 # Import various modules that set up global state.
-import mypyc.lower.int_ops
-import mypyc.lower.list_ops  # noqa: F401
+from mypyc.lower import int_ops, list_ops, misc_ops  # noqa: F401

mypyc/primitives/list_ops.py

@@ -11,12 +11,14 @@
     int_rprimitive,
     list_rprimitive,
     object_rprimitive,
+    pointer_rprimitive,
     short_int_rprimitive,
 )
 from mypyc.primitives.registry import (
     ERR_NEG_INT,
     binary_op,
     custom_op,
+    custom_primitive_op,
     function_op,
     load_address_op,
     method_op,
@@ -60,6 +62,14 @@
     steals=True,
 )
 
+# Get pointer to list items (ob_item PyListObject field)
+list_items = custom_primitive_op(
+    name="list_items",
+    arg_types=[list_rprimitive],
+    return_type=pointer_rprimitive,
+    error_kind=ERR_NEVER,
+)
+
 # list[index] (for an integer index)
 list_get_item_op = method_op(
     name="__getitem__",

mypyc/primitives/misc_ops.py

@@ -257,3 +257,11 @@
     error_kind=ERR_NEVER,
     steals=[False, False, True],
 )
+
+# Get length of PyVarObject instance (e.g. list or tuple)
+var_object_size = custom_primitive_op(
+    name="var_object_size",
+    arg_types=[object_rprimitive],
+    return_type=c_pyssize_t_rprimitive,
+    error_kind=ERR_NEVER,
+)

mypyc/test-data/irbuild-any.test

@@ -106,7 +106,7 @@ def f2(a, n, l):
     r9, r10 :: bit
     r11 :: list
     r12 :: object
-    r13, r14 :: ptr
+    r13 :: ptr
 L0:
     r0 = box(int, n)
     r1 = PyObject_GetItem(a, r0)
@@ -121,10 +121,9 @@ L0:
     r10 = CPyList_SetItem(l, n, a)
     r11 = PyList_New(2)
     r12 = box(int, n)
-    r13 = get_element_ptr r11 ob_item :: PyListObject
-    r14 = load_mem r13 :: ptr*
-    buf_init_item r14, 0, a
-    buf_init_item r14, 1, r12
+    r13 = list_items r11
+    buf_init_item r13, 0, a
+    buf_init_item r13, 1, r12
     keep_alive r11
     return 1
 def f3(a, n):