Files
TransPyC/lib/core/Handles/HandlesAugAssign.py
T
TermiNexus 6fc9e19e24 feat(vast): 新增自研 Viper/Python 语法解析器取代标准库 ast
TPC 侧取缔对 Python 标准库 ast 模块的依赖,改用 lib/vast 包:
- Nodes: 与 CPython 3.12 ast 完全同构的 131 个节点类(含抽象基类继承)
- Tokenizer: 词法分析(缩进/字符串/f-string/数字/运算符/软关键字)
- ExprParser: 表达式递归下降解析(优先级/lambda/推导式/f-string/切片)
- Parser: 语句解析(def/class/if/for/try/with/match/import/注解)
- Utils/Unparser: walk/dump/unparse/copy_location/literal_eval 等辅助
- 软关键字 fn/fun/func/function 并入解析器, 废除 ExpandSoftKeywords 预处理
- lib/_bootstrap 将 lib 目录加入 sys.path, import vast as ast 零修改替换
- 52 个文件 import ast 替换为 import vast as ast; 94 文件差分测试 0 失败
2026-08-17 09:36:38 +08:00

243 lines
15 KiB
Python
Raw Blame History

This file contains ambiguous Unicode characters
This file contains Unicode characters that might be confused with other characters. If you think that this is intentional, you can safely ignore this warning. Use the Escape button to reveal them.
from __future__ import annotations
from typing import TYPE_CHECKING
import vast as ast
import sys
import llvmlite.ir as ir
if TYPE_CHECKING:
from lib.core.translator import Translator
from lib.core.VLogger import get_logger as _vlog
from lib.core.Handles.HandlesBase import BaseHandle
class AugAssignHandle(BaseHandle):
_AUGOP_MAP: dict[type, str] = {
ast.Add: '__iadd__', ast.Sub: '__isub__', ast.Mult: '__imul__',
ast.Div: '__itruediv__', ast.FloorDiv: '__ifloordiv__', ast.Mod: '__imod__', ast.Pow: '__ipow__',
}
_AUGOP_FALLBACK: dict[type, str] = {
ast.Add: '__add__', ast.Sub: '__sub__', ast.Mult: '__mul__',
ast.Div: '__truediv__', ast.FloorDiv: '__floordiv__', ast.Mod: '__mod__', ast.Pow: '__pow__',
}
def _resolve_global_var_type(self, Gen: "Translator.LlvmGen", VarName: str, fallback: ir.Type) -> ir.Type:
"""解析 global 变量的 LLVM 类型:优先用模块级 AnnAssign 的注解类型。
问题场景:函数内 `_g_alloc_count += 1`(global 声明),模块级注解为
`_g_alloc_count: t.CSizeT`(i64)。若不查模块级注解类型,此处会用 RHS 字面量
类型(i32)自动建 i32 全局,与模块内其他读取/写入(i64)类型不匹配,
产生 "Operands must be the same type, got (i64, i32)" 类错误。
模块级 AnnAssign 已通过 _EmitGlobalAnnAssignLlvm 创建 LLVM 全局,
但注解无初始值(如 `_g_alloc_max: t.CSizeT = 0`)时 initializer 为空、
Gen.module.globals 中无该符号,故需在此兜底解析注解类型。
"""
module_ann: dict = getattr(Gen, '_module_ann_types', None)
if module_ann and VarName in module_ann:
ann_type: ir.Type = module_ann[VarName]
if ann_type is not None:
return ann_type
if isinstance(fallback, ir.IntType):
return fallback
return ir.IntType(64)
def _is_aug_unsigned(self, Node: ast.AugAssign, OldVal: ir.Value | None, Value: ir.Value) -> bool:
Gen: "Translator.LlvmGen" = self.Trans.LlvmGen
if isinstance(Node.target, ast.Name):
if Gen._check_node_unsigned(Node.target):
return True
elif isinstance(Node.target, ast.Attribute):
if Gen._check_node_unsigned(Node.target):
return True
elif isinstance(Node.target, ast.Subscript):
if Gen._check_node_unsigned(Node.target):
return True
if Node.value is not None and Gen._check_node_unsigned(Node.value):
return True
return False
def _HandleAugAssignLlvm(self, Node: ast.AugAssign) -> None:
Gen: "Translator.LlvmGen" = self.Trans.LlvmGen
if isinstance(Node.target, ast.Name):
VarName: str = Node.target.id
if (VarName not in Gen.variables
and VarName not in Gen.global_vars
and VarName not in Gen._reg_values
and VarName not in Gen._direct_values):
src_info: str = Gen._get_node_info()
_vlog().error(f"变量 '{VarName}' 未声明,不能用于增量赋值(需用 global 声明或先声明局部变量){src_info}")
sys.exit(1)
if VarName in Gen.var_const_flags:
src_info: str = Gen._get_node_info()
_vlog().error(f"不能对 const 变量 '{VarName}' 增量赋值{src_info}")
sys.exit(1)
ClassName: str | None = Gen.var_struct_class.get(VarName)
iop_name: str | None = self._AUGOP_MAP.get(type(Node.op))
if ClassName and iop_name:
iFuncName: str = f'{ClassName}.{iop_name}'
FuncName: str = f'{ClassName}.{self._AUGOP_FALLBACK.get(type(Node.op), "")}'
if iFuncName in Gen.functions or FuncName in Gen.functions:
OldVal: ir.Value | None = self.Trans.ExprHandler.HandleExprLlvm(ast.Name(id=VarName, ctx=ast.Load()))
Value: ir.Value | None = self.HandleExprLlvm(Node.value)
if OldVal and Value:
result: ir.Value | None
if iFuncName in Gen.functions:
result = self.Trans.ExprHandler._try_operator_overLoad(ClassName, iop_name, OldVal, Gen, Value)
else:
result = self.Trans.ExprHandler._try_operator_overLoad(ClassName, self._AUGOP_FALLBACK[type(Node.op)], OldVal, Gen, Value)
if result is not None:
if VarName in Gen.variables and Gen.variables[VarName] is not None:
VarPtr: ir.Value = Gen.variables[VarName]
if VarPtr.type.pointee == result.type:
Gen._store(result, VarPtr)
elif isinstance(VarPtr.type.pointee, ir.PointerType) and isinstance(result.type, ir.PointerType):
CastedVal: ir.Value = Gen.builder.bitcast(result, VarPtr.type.pointee, name=f"cast_{VarName}")
Gen._store(CastedVal, VarPtr)
else:
NewVar: ir.AllocaInstr = Gen._allocaEntry(result.type, name=VarName)
Gen._store(result, NewVar)
Gen.variables[VarName] = NewVar
else:
Gen._reg_values[VarName] = result
Gen.variables[VarName] = None
return
Value: ir.Value | None = self.HandleExprLlvm(Node.value)
if not Value:
return
is_unsigned: bool = self._is_aug_unsigned(Node, None, Value)
if isinstance(Node.target, ast.Name):
VarName: str = Node.target.id
Op: str = self.Trans.ExprHandler.GetOpSymbol(Node.op)
if VarName in Gen.global_vars:
# 治本修复:global 声明的变量若未被模块级 AnnAssign 创建为 LLVM 全局
# (注解无初始值的全局变量),此处自动创建零初始化全局变量
GVar: ir.GlobalVariable
if VarName not in Gen.module.globals:
_gv_type: ir.Type = self._resolve_global_var_type(Gen, VarName, Value.type)
GVar = ir.GlobalVariable(Gen.module, _gv_type, name=VarName)
GVar.initializer = ir.Constant(_gv_type, 0)
else:
GVar = Gen.module.globals[VarName]
Gen.variables[VarName] = GVar
OldVal: ir.Value = Gen._load(GVar, name=VarName)
NewVal: ir.Value = Gen.emit_binary_op(Op, OldVal, Value, is_unsigned=is_unsigned)
Gen._store(NewVal, GVar)
return
if VarName in Gen._reg_values:
OldVal: ir.Value = Gen._reg_values[VarName]
var: ir.AllocaInstr = Gen._allocaEntry(OldVal.type, name=VarName)
Gen._store(OldVal, var)
Gen.variables[VarName] = var
del Gen._reg_values[VarName]
if VarName in Gen._direct_values:
OldVal: ir.Value = Gen._direct_values.pop(VarName)
var: ir.AllocaInstr = Gen._allocaEntry(OldVal.type, name=VarName)
Gen._store(OldVal, var)
Gen.variables[VarName] = var
if Gen.builder.block.is_terminated:
return
if VarName in Gen.variables and Gen.variables[VarName] is not None:
OldVal: ir.Value = Gen._load(Gen.variables[VarName], name=VarName)
store_ptr: ir.Value = Gen.variables[VarName]
if isinstance(OldVal.type, ir.PointerType) and isinstance(Value.type, ir.IntType):
if Value.type.width != 64:
Value = Gen.builder.zext(Value, ir.IntType(64), name="zext_ptr_offset")
NewVal: ir.Value
if Op == '-' or Op == '+':
if Op == '-':
NegValue: ir.Value = Gen.builder.neg(Value, name="neg_ptr_offset")
NewVal = Gen.builder.gep(OldVal, [NegValue], name="ptr_sub")
else:
NewVal = Gen.builder.gep(OldVal, [Value], name="ptr_add")
else:
ptr_val: ir.Value = OldVal
OldVal = Gen._load(OldVal, name="deref_ptr_for_op")
orig_type: ir.Type = OldVal.type
if isinstance(OldVal.type, ir.IntType) and isinstance(Value.type, ir.IntType):
if OldVal.type.width < Value.type.width:
OldVal = Gen.builder.zext(OldVal, Value.type, name="zext_left")
elif OldVal.type.width > Value.type.width:
Value = Gen.builder.zext(Value, OldVal.type, name="zext_right")
NewVal = Gen.emit_binary_op(Op, OldVal, Value, is_unsigned=is_unsigned)
if isinstance(NewVal.type, ir.IntType) and isinstance(orig_type, ir.IntType):
if NewVal.type.width > orig_type.width:
NewVal = Gen.builder.trunc(NewVal, orig_type, name="trunc_result")
store_ptr = ptr_val
else:
if isinstance(OldVal.type, ir.IntType) and isinstance(Value.type, ir.IntType):
if OldVal.type.width < Value.type.width:
OldVal = Gen.builder.zext(OldVal, Value.type, name="zext_left")
elif OldVal.type.width > Value.type.width:
Value = Gen.builder.zext(Value, OldVal.type, name="zext_right")
elif isinstance(OldVal.type, ir.PointerType) and isinstance(Value.type, ir.IntType):
if Op in ('+', '-'):
if Value.type.width < 64:
Value = Gen.builder.zext(Value, ir.IntType(64), name="zext_ptr_offset_aug")
NewVal: ir.Value
if Op == '-':
NegValue: ir.Value = Gen.builder.neg(Value, name="neg_ptr_offset_aug")
NewVal = Gen.builder.gep(OldVal, [NegValue], name="ptr_sub_aug")
else:
NewVal = Gen.builder.gep(OldVal, [Value], name="ptr_add_aug")
Gen._store(NewVal, store_ptr)
return
NewVal: ir.Value = Gen.emit_binary_op(Op, OldVal, Value, is_unsigned=is_unsigned)
if isinstance(store_ptr.type, ir.PointerType):
TargetType: ir.Type = store_ptr.type.pointee
if NewVal.type != TargetType:
Coerced: ir.Value | None = Gen._coerce_value(NewVal, TargetType)
if Coerced is not None:
NewVal = Coerced
Gen._store(NewVal, store_ptr)
elif VarName == 'pos':
var: ir.AllocaInstr = Gen._allocaEntry(Value.type, name=VarName)
Gen.variables[VarName] = var
OldVal: ir.Value = Gen._load(var, name=VarName)
NewVal: ir.Value = Gen.emit_binary_op(Op, OldVal, Value, is_unsigned=is_unsigned)
Gen._store(NewVal, var)
elif isinstance(Node.target, ast.Attribute):
AttrName: str = Node.target.attr
OldVal: ir.Value | None = self.Trans.ExprHandler.HandleExprLlvm(Node.target)
if OldVal:
def _deref_if_ptr(val: ir.Value) -> ir.Value:
if isinstance(val.type, ir.PointerType):
pointee: ir.Type = val.type.pointee
if isinstance(pointee, (ir.IntType, ir.FloatType, ir.DoubleType)):
return Gen._load(val, name="deref_aug")
return val
OldVal = _deref_if_ptr(OldVal)
Op: str = self.Trans.ExprHandler.GetOpSymbol(Node.op)
if isinstance(OldVal.type, ir.IntType) and isinstance(Value.type, ir.IntType):
if OldVal.type.width < Value.type.width:
OldVal = Gen.builder.zext(OldVal, Value.type, name="zext_left_attr")
elif OldVal.type.width > Value.type.width:
Value = Gen.builder.zext(Value, OldVal.type, name="zext_right_attr")
NewVal: ir.Value = Gen.emit_binary_op(Op, OldVal, Value, is_unsigned=is_unsigned)
self.Trans.AssignHandler._HandleAttributeStoreLlvm(Node.target, NewVal)
elif isinstance(Node.target, ast.Subscript):
SubTarget: ast.Subscript = Node.target
ElemPtr: ir.Value | None = self.Trans.ExprAttrHandle.HandleSubscriptPtrLlvm(SubTarget)
if ElemPtr:
if isinstance(ElemPtr.type, ir.PointerType):
OldVal: ir.Value = Gen._load(ElemPtr, name="old_subscript_val")
Op: str = self.Trans.ExprHandler.GetOpSymbol(Node.op)
if isinstance(OldVal.type, ir.IntType) and isinstance(Value.type, ir.IntType):
if OldVal.type.width < Value.type.width:
OldVal = Gen.builder.zext(OldVal, Value.type, name="zext_left_sub")
elif OldVal.type.width > Value.type.width:
Value = Gen.builder.zext(Value, OldVal.type, name="zext_right_sub")
elif isinstance(OldVal.type, ir.PointerType) and isinstance(Value.type, ir.IntType):
if Op in ('+', '-'):
if Value.type.width < 64:
Value = Gen.builder.zext(Value, ir.IntType(64), name="zext_ptr_offset_sub")
NewVal: ir.Value
if Op == '-':
NegValue: ir.Value = Gen.builder.neg(Value, name="neg_ptr_offset_sub")
NewVal = Gen.builder.gep(OldVal, [NegValue], name="ptr_sub_sub")
else:
NewVal = Gen.builder.gep(OldVal, [Value], name="ptr_add_sub")
Gen._store(NewVal, ElemPtr)
return
NewVal: ir.Value = Gen.emit_binary_op(Op, OldVal, Value, is_unsigned=is_unsigned)
if isinstance(NewVal.type, ir.IntType) and isinstance(ElemPtr.type.pointee, ir.IntType):
if NewVal.type.width > ElemPtr.type.pointee.width:
NewVal = Gen.builder.trunc(NewVal, ElemPtr.type.pointee, name="trunc_sub_result")
Gen._store(NewVal, ElemPtr)