Merge branch 'master' into solve_intermediate_variable_bug

Author: randolf-scholz

mypyc/codegen/emit.py

@@ -28,8 +28,7 @@
     RType,
     RUnion,
     int_rprimitive,
-    is_bit_rprimitive,
-    is_bool_rprimitive,
+    is_bool_or_bit_rprimitive,
     is_bytes_rprimitive,
     is_dict_rprimitive,
     is_fixed_width_rtype,
@@ -615,8 +614,7 @@ def emit_cast(
             or is_range_rprimitive(typ)
             or is_float_rprimitive(typ)
             or is_int_rprimitive(typ)
-            or is_bool_rprimitive(typ)
-            or is_bit_rprimitive(typ)
+            or is_bool_or_bit_rprimitive(typ)
             or is_fixed_width_rtype(typ)
         ):
             if declare_dest:
@@ -638,7 +636,7 @@ def emit_cast(
             elif is_int_rprimitive(typ) or is_fixed_width_rtype(typ):
                 # TODO: Range check for fixed-width types?
                 prefix = "PyLong"
-            elif is_bool_rprimitive(typ) or is_bit_rprimitive(typ):
+            elif is_bool_or_bit_rprimitive(typ):
                 prefix = "PyBool"
             else:
                 assert False, f"unexpected primitive type: {typ}"
@@ -889,7 +887,7 @@ def emit_unbox(
             self.emit_line("else {")
             self.emit_line(failure)
             self.emit_line("}")
-        elif is_bool_rprimitive(typ) or is_bit_rprimitive(typ):
+        elif is_bool_or_bit_rprimitive(typ):
             # Whether we are borrowing or not makes no difference.
             if declare_dest:
                 self.emit_line(f"char {dest};")
@@ -1015,7 +1013,7 @@ def emit_box(
         if is_int_rprimitive(typ) or is_short_int_rprimitive(typ):
             # Steal the existing reference if it exists.
             self.emit_line(f"{declaration}{dest} = CPyTagged_StealAsObject({src});")
-        elif is_bool_rprimitive(typ) or is_bit_rprimitive(typ):
+        elif is_bool_or_bit_rprimitive(typ):
             # N.B: bool is special cased to produce a borrowed value
             # after boxing, so we don't need to increment the refcount
             # when this comes directly from a Box op.

mypyc/ir/ops.py

@@ -42,8 +42,7 @@ class to enable the new behavior. Sometimes adding a new abstract
     cstring_rprimitive,
     float_rprimitive,
     int_rprimitive,
-    is_bit_rprimitive,
-    is_bool_rprimitive,
+    is_bool_or_bit_rprimitive,
     is_int_rprimitive,
     is_none_rprimitive,
     is_pointer_rprimitive,
@@ -1089,11 +1088,7 @@ def __init__(self, src: Value, line: int = -1) -> None:
         self.src = src
         self.type = object_rprimitive
         # When we box None and bool values, we produce a borrowed result
-        if (
-            is_none_rprimitive(self.src.type)
-            or is_bool_rprimitive(self.src.type)
-            or is_bit_rprimitive(self.src.type)
-        ):
+        if is_none_rprimitive(self.src.type) or is_bool_or_bit_rprimitive(self.src.type):
             self.is_borrowed = True
 
     def sources(self) -> list[Value]:

mypyc/ir/rtypes.py

@@ -582,6 +582,10 @@ def is_bit_rprimitive(rtype: RType) -> bool:
     return isinstance(rtype, RPrimitive) and rtype.name == "bit"
 
 
+def is_bool_or_bit_rprimitive(rtype: RType) -> bool:
+    return is_bool_rprimitive(rtype) or is_bit_rprimitive(rtype)
+
+
 def is_object_rprimitive(rtype: RType) -> bool:
     return isinstance(rtype, RPrimitive) and rtype.name == "builtins.object"
 

mypyc/irbuild/classdef.py

@@ -64,6 +64,7 @@
     handle_non_ext_method,
     load_type,
 )
+from mypyc.irbuild.prepare import GENERATOR_HELPER_NAME
 from mypyc.irbuild.util import dataclass_type, get_func_def, is_constant, is_dataclass_decorator
 from mypyc.primitives.dict_ops import dict_new_op, dict_set_item_op
 from mypyc.primitives.generic_ops import (
@@ -135,6 +136,14 @@ def transform_class_def(builder: IRBuilder, cdef: ClassDef) -> None:
         cls_builder = NonExtClassBuilder(builder, cdef)
 
     for stmt in cdef.defs.body:
+        if (
+            isinstance(stmt, (FuncDef, Decorator, OverloadedFuncDef))
+            and stmt.name == GENERATOR_HELPER_NAME
+        ):
+            builder.error(
+                f'Method name "{stmt.name}" is reserved for mypyc internal use', stmt.line
+            )
+
         if isinstance(stmt, OverloadedFuncDef) and stmt.is_property:
             if isinstance(cls_builder, NonExtClassBuilder):
                 # properties with both getters and setters in non_extension

mypyc/irbuild/generator.py

@@ -50,6 +50,7 @@
     setup_func_for_recursive_call,
 )
 from mypyc.irbuild.nonlocalcontrol import ExceptNonlocalControl
+from mypyc.irbuild.prepare import GENERATOR_HELPER_NAME
 from mypyc.primitives.exc_ops import (
     error_catch_op,
     exc_matches_op,
@@ -236,11 +237,11 @@ def add_helper_to_generator_class(
     builder: IRBuilder, arg_regs: list[Register], blocks: list[BasicBlock], fn_info: FuncInfo
 ) -> FuncDecl:
     """Generates a helper method for a generator class, called by '__next__' and 'throw'."""
-    helper_fn_decl = fn_info.generator_class.ir.method_decls["__mypyc_generator_helper__"]
+    helper_fn_decl = fn_info.generator_class.ir.method_decls[GENERATOR_HELPER_NAME]
     helper_fn_ir = FuncIR(
         helper_fn_decl, arg_regs, blocks, fn_info.fitem.line, traceback_name=fn_info.fitem.name
     )
-    fn_info.generator_class.ir.methods["__mypyc_generator_helper__"] = helper_fn_ir
+    fn_info.generator_class.ir.methods[GENERATOR_HELPER_NAME] = helper_fn_ir
     builder.functions.append(helper_fn_ir)
     fn_info.env_class.env_user_function = helper_fn_ir
 

mypyc/irbuild/ll_builder.py

@@ -93,8 +93,7 @@
     dict_rprimitive,
     float_rprimitive,
     int_rprimitive,
-    is_bit_rprimitive,
-    is_bool_rprimitive,
+    is_bool_or_bit_rprimitive,
     is_bytes_rprimitive,
     is_dict_rprimitive,
     is_fixed_width_rtype,
@@ -175,7 +174,12 @@
     unary_ops,
 )
 from mypyc.primitives.set_ops import new_set_op
-from mypyc.primitives.str_ops import str_check_if_true, str_ssize_t_size_op, unicode_compare
+from mypyc.primitives.str_ops import (
+    str_check_if_true,
+    str_eq,
+    str_ssize_t_size_op,
+    unicode_compare,
+)
 from mypyc.primitives.tuple_ops import list_tuple_op, new_tuple_op, new_tuple_with_length_op
 from mypyc.rt_subtype import is_runtime_subtype
 from mypyc.sametype import is_same_type
@@ -376,16 +380,12 @@ def coerce(
             ):
                 # Equivalent types
                 return src
-            elif (is_bool_rprimitive(src_type) or is_bit_rprimitive(src_type)) and is_tagged(
-                target_type
-            ):
+            elif is_bool_or_bit_rprimitive(src_type) and is_tagged(target_type):
                 shifted = self.int_op(
                     bool_rprimitive, src, Integer(1, bool_rprimitive), IntOp.LEFT_SHIFT
                 )
                 return self.add(Extend(shifted, target_type, signed=False))
-            elif (
-                is_bool_rprimitive(src_type) or is_bit_rprimitive(src_type)
-            ) and is_fixed_width_rtype(target_type):
+            elif is_bool_or_bit_rprimitive(src_type) and is_fixed_width_rtype(target_type):
                 return self.add(Extend(src, target_type, signed=False))
             elif isinstance(src, Integer) and is_float_rprimitive(target_type):
                 if is_tagged(src_type):
@@ -1336,7 +1336,11 @@ def binary_op(self, lreg: Value, rreg: Value, op: str, line: int) -> Value:
             return self.compare_strings(lreg, rreg, op, line)
         if is_bytes_rprimitive(ltype) and is_bytes_rprimitive(rtype) and op in ("==", "!="):
             return self.compare_bytes(lreg, rreg, op, line)
-        if is_bool_rprimitive(ltype) and is_bool_rprimitive(rtype) and op in BOOL_BINARY_OPS:
+        if (
+            is_bool_or_bit_rprimitive(ltype)
+            and is_bool_or_bit_rprimitive(rtype)
+            and op in BOOL_BINARY_OPS
+        ):
             if op in ComparisonOp.signed_ops:
                 return self.bool_comparison_op(lreg, rreg, op, line)
             else:
@@ -1350,7 +1354,7 @@ def binary_op(self, lreg: Value, rreg: Value, op: str, line: int) -> Value:
                     op_id = int_op_to_id[op]
                 else:
                     op_id = IntOp.DIV
-                if is_bool_rprimitive(rtype) or is_bit_rprimitive(rtype):
+                if is_bool_or_bit_rprimitive(rtype):
                     rreg = self.coerce(rreg, ltype, line)
                     rtype = ltype
                 if is_fixed_width_rtype(rtype) or is_tagged(rtype):
@@ -1362,7 +1366,7 @@ def binary_op(self, lreg: Value, rreg: Value, op: str, line: int) -> Value:
             elif op in ComparisonOp.signed_ops:
                 if is_int_rprimitive(rtype):
                     rreg = self.coerce_int_to_fixed_width(rreg, ltype, line)
-                elif is_bool_rprimitive(rtype) or is_bit_rprimitive(rtype):
+                elif is_bool_or_bit_rprimitive(rtype):
                     rreg = self.coerce(rreg, ltype, line)
                 op_id = ComparisonOp.signed_ops[op]
                 if is_fixed_width_rtype(rreg.type):
@@ -1382,13 +1386,13 @@ def binary_op(self, lreg: Value, rreg: Value, op: str, line: int) -> Value:
                     )
                 if is_tagged(ltype):
                     return self.fixed_width_int_op(rtype, lreg, rreg, op_id, line)
-                if is_bool_rprimitive(ltype) or is_bit_rprimitive(ltype):
+                if is_bool_or_bit_rprimitive(ltype):
                     lreg = self.coerce(lreg, rtype, line)
                     return self.fixed_width_int_op(rtype, lreg, rreg, op_id, line)
             elif op in ComparisonOp.signed_ops:
                 if is_int_rprimitive(ltype):
                     lreg = self.coerce_int_to_fixed_width(lreg, rtype, line)
-                elif is_bool_rprimitive(ltype) or is_bit_rprimitive(ltype):
+                elif is_bool_or_bit_rprimitive(ltype):
                     lreg = self.coerce(lreg, rtype, line)
                 op_id = ComparisonOp.signed_ops[op]
                 if isinstance(lreg, Integer):
@@ -1471,6 +1475,11 @@ def check_tagged_short_int(self, val: Value, line: int, negated: bool = False) -
 
     def compare_strings(self, lhs: Value, rhs: Value, op: str, line: int) -> Value:
         """Compare two strings"""
+        if op == "==":
+            return self.primitive_op(str_eq, [lhs, rhs], line)
+        elif op == "!=":
+            eq = self.primitive_op(str_eq, [lhs, rhs], line)
+            return self.add(ComparisonOp(eq, self.false(), ComparisonOp.EQ, line))
         compare_result = self.call_c(unicode_compare, [lhs, rhs], line)
         error_constant = Integer(-1, c_int_rprimitive, line)
         compare_error_check = self.add(
@@ -1534,7 +1543,7 @@ def compare_tuples(self, lhs: Value, rhs: Value, op: str, line: int = -1) -> Val
             compare = self.binary_op(lhs_item, rhs_item, op, line)
             # Cast to bool if necessary since most types uses comparison returning a object type
             # See generic_ops.py for more information
-            if not (is_bool_rprimitive(compare.type) or is_bit_rprimitive(compare.type)):
+            if not is_bool_or_bit_rprimitive(compare.type):
                 compare = self.primitive_op(bool_op, [compare], line)
             if i < len(lhs.type.types) - 1:
                 branch = Branch(compare, early_stop, check_blocks[i + 1], Branch.BOOL)
@@ -1553,7 +1562,7 @@ def compare_tuples(self, lhs: Value, rhs: Value, op: str, line: int = -1) -> Val
 
     def translate_instance_contains(self, inst: Value, item: Value, op: str, line: int) -> Value:
         res = self.gen_method_call(inst, "__contains__", [item], None, line)
-        if not is_bool_rprimitive(res.type):
+        if not is_bool_or_bit_rprimitive(res.type):
             res = self.primitive_op(bool_op, [res], line)
         if op == "not in":
             res = self.bool_bitwise_op(res, Integer(1, rtype=bool_rprimitive), "^", line)
@@ -1580,7 +1589,7 @@ def unary_not(self, value: Value, line: int) -> Value:
 
     def unary_op(self, value: Value, expr_op: str, line: int) -> Value:
         typ = value.type
-        if is_bool_rprimitive(typ) or is_bit_rprimitive(typ):
+        if is_bool_or_bit_rprimitive(typ):
             if expr_op == "not":
                 return self.unary_not(value, line)
             if expr_op == "+":
@@ -1738,7 +1747,7 @@ def bool_value(self, value: Value) -> Value:
 
         The result type can be bit_rprimitive or bool_rprimitive.
         """
-        if is_bool_rprimitive(value.type) or is_bit_rprimitive(value.type):
+        if is_bool_or_bit_rprimitive(value.type):
             result = value
         elif is_runtime_subtype(value.type, int_rprimitive):
             zero = Integer(0, short_int_rprimitive)

mypyc/irbuild/prepare.py

@@ -71,6 +71,8 @@
 from mypyc.options import CompilerOptions
 from mypyc.sametype import is_same_type
 
+GENERATOR_HELPER_NAME = "__mypyc_generator_helper__"
+
 
 def build_type_map(
     mapper: Mapper,
@@ -229,7 +231,7 @@ def create_generator_class_if_needed(
 
         # The implementation of most generator functionality is behind this magic method.
         helper_fn_decl = FuncDecl(
-            "__mypyc_generator_helper__", name, module_name, helper_sig, internal=True
+            GENERATOR_HELPER_NAME, name, module_name, helper_sig, internal=True
         )
         cir.method_decls[helper_fn_decl.name] = helper_fn_decl
 

mypyc/irbuild/statement.py

@@ -90,6 +90,7 @@
     FinallyNonlocalControl,
     TryFinallyNonlocalControl,
 )
+from mypyc.irbuild.prepare import GENERATOR_HELPER_NAME
 from mypyc.irbuild.targets import (
     AssignmentTarget,
     AssignmentTargetAttr,
@@ -933,7 +934,7 @@ def emit_yield_from_or_await(
     to_yield_reg = Register(object_rprimitive)
     received_reg = Register(object_rprimitive)
 
-    helper_method = "__mypyc_generator_helper__"
+    helper_method = GENERATOR_HELPER_NAME
     if (
         isinstance(val, (Call, MethodCall))
         and isinstance(val.type, RInstance)

mypyc/lib-rt/CPy.h

@@ -726,6 +726,7 @@ static inline char CPyDict_CheckSize(PyObject *dict, CPyTagged size) {
 #define RIGHTSTRIP 1
 #define BOTHSTRIP  2
 
+char CPyStr_Equal(PyObject *str1, PyObject *str2);
 PyObject *CPyStr_Build(Py_ssize_t len, ...);
 PyObject *CPyStr_GetItem(PyObject *str, CPyTagged index);
 CPyTagged CPyStr_Find(PyObject *str, PyObject *substr, CPyTagged start, int direction);

mypyc/lib-rt/str_ops.c

@@ -64,6 +64,22 @@ make_bloom_mask(int kind, const void* ptr, Py_ssize_t len)
 #undef BLOOM_UPDATE
 }
 
+// Adapted from CPython 3.13.1 (_PyUnicode_Equal)
+char CPyStr_Equal(PyObject *str1, PyObject *str2) {
+    if (str1 == str2) {
+        return 1;
+    }
+    Py_ssize_t len = PyUnicode_GET_LENGTH(str1);
+    if (PyUnicode_GET_LENGTH(str2) != len)
+        return 0;
+    int kind = PyUnicode_KIND(str1);
+    if (PyUnicode_KIND(str2) != kind)
+        return 0;
+    const void *data1 = PyUnicode_DATA(str1);
+    const void *data2 = PyUnicode_DATA(str2);
+    return memcmp(data1, data2, len * kind) == 0;
+}
+
 PyObject *CPyStr_GetItem(PyObject *str, CPyTagged index) {
     if (PyUnicode_READY(str) != -1) {
         if (CPyTagged_CheckShort(index)) {