可用的回归测试通过的标准版本
This commit is contained in:
@@ -1,240 +1,240 @@
|
||||
from __future__ import annotations
|
||||
from typing import TYPE_CHECKING
|
||||
if TYPE_CHECKING:
|
||||
from lib.core.translator import Translator
|
||||
import ast
|
||||
import llvmlite.ir as ir
|
||||
|
||||
|
||||
class ExprUtils:
|
||||
def __init__(self, translator: "Translator"):
|
||||
self.Trans = translator
|
||||
|
||||
def GetOpSymbol(self, Op):
|
||||
OpMap = {
|
||||
ast.Add: '+', ast.Sub: '-', ast.Mult: '*', ast.Div: '/',
|
||||
ast.FloorDiv: '//', ast.Mod: '%', ast.Pow: '**', ast.BitAnd: '&',
|
||||
ast.BitOr: '|', ast.BitXor: '^', ast.LShift: '<<', ast.RShift: '>>',
|
||||
ast.MatMult: '@',
|
||||
}
|
||||
return OpMap.get(type(Op), None)
|
||||
|
||||
def GetUnaryOpSymbol(self, Op):
|
||||
OpMap = {ast.Invert: '~', ast.Not: 'not', ast.UAdd: '+', ast.USub: '-'}
|
||||
return OpMap.get(type(Op), None)
|
||||
|
||||
def GetComparatorSymbol(self, Op):
|
||||
OpMap = {
|
||||
ast.Eq: '==', ast.NotEq: '!=', ast.Lt: '<', ast.LtE: '<=',
|
||||
ast.Gt: '>', ast.GtE: '>=', ast.Is: 'is', ast.IsNot: 'is not',
|
||||
ast.In: 'in', ast.NotIn: 'not in',
|
||||
}
|
||||
return OpMap.get(type(Op), None)
|
||||
|
||||
def _get_var_class(self, node, Gen):
|
||||
if isinstance(node, ast.Name):
|
||||
VarName = node.id
|
||||
return Gen.var_struct_class.get(VarName)
|
||||
if isinstance(node, ast.Attribute):
|
||||
AttrName = node.attr
|
||||
return Gen.var_struct_class.get(AttrName)
|
||||
return None
|
||||
|
||||
def _try_operator_overload(self, ClassName, op_name, obj_val, Gen, other_val=None):
|
||||
FullMethodName = f'{ClassName}.{op_name}'
|
||||
if not Gen._has_function(FullMethodName):
|
||||
FullMethodName = f'{ClassName}.{op_name}__'
|
||||
if Gen._has_function(FullMethodName):
|
||||
func = Gen._get_function(FullMethodName)
|
||||
call_args = [obj_val]
|
||||
if other_val is not None:
|
||||
call_args.append(other_val)
|
||||
if len(func.ftype.args) == len(call_args):
|
||||
for i, (expected_type, actual_val) in enumerate(zip(func.ftype.args, call_args)):
|
||||
if isinstance(expected_type, ir.PointerType) and isinstance(actual_val.type, ir.PointerType):
|
||||
if expected_type != actual_val.type:
|
||||
if isinstance(expected_type.pointee, (ir.IdentifiedStructType, ir.LiteralStructType)):
|
||||
call_args[i] = Gen.builder.bitcast(actual_val, expected_type, name=f"bitcast_arg_{i}")
|
||||
result = Gen.builder.call(func, call_args, name=f"call_{FullMethodName}")
|
||||
return result
|
||||
return None
|
||||
|
||||
def _llvm_type_to_detailed_string(self, llvm_type, var_name=""):
|
||||
info = {
|
||||
'name': "unknown",
|
||||
'size': 0,
|
||||
'signed': False,
|
||||
'ptr': False
|
||||
}
|
||||
if isinstance(llvm_type, ir.IntType):
|
||||
info['size'] = llvm_type.width // 8
|
||||
info['signed'] = True
|
||||
if llvm_type.width == 8:
|
||||
info['name'] = "char"
|
||||
elif llvm_type.width == 16:
|
||||
info['name'] = "short"
|
||||
elif llvm_type.width == 32:
|
||||
info['name'] = "int"
|
||||
elif llvm_type.width == 64:
|
||||
info['name'] = "long long"
|
||||
else:
|
||||
info['name'] = f"int{llvm_type.width}"
|
||||
elif isinstance(llvm_type, ir.DoubleType):
|
||||
info['size'] = 8
|
||||
info['signed'] = True
|
||||
info['name'] = "double"
|
||||
elif isinstance(llvm_type, ir.FloatType):
|
||||
info['size'] = 4
|
||||
info['signed'] = True
|
||||
info['name'] = "float"
|
||||
elif isinstance(llvm_type, ir.PointerType):
|
||||
info['size'] = 8
|
||||
info['signed'] = False
|
||||
info['ptr'] = True
|
||||
pointee_info = self._llvm_type_to_detailed_string(llvm_type.pointee)
|
||||
info['name'] = f"{pointee_info['name']}*"
|
||||
elif isinstance(llvm_type, (ir.LiteralStructType, ir.IdentifiedStructType)):
|
||||
size = 0
|
||||
for elem in llvm_type.elements:
|
||||
elem_info = self._llvm_type_to_detailed_string(elem)
|
||||
size += elem_info['size']
|
||||
info['size'] = size
|
||||
info['signed'] = False
|
||||
if isinstance(llvm_type, ir.IdentifiedStructType):
|
||||
info['name'] = llvm_type.name
|
||||
else:
|
||||
info['name'] = "struct"
|
||||
elif isinstance(llvm_type, ir.ArrayType):
|
||||
elem_info = self._llvm_type_to_detailed_string(llvm_type.element)
|
||||
info['size'] = elem_info['size'] * llvm_type.count
|
||||
info['signed'] = elem_info['signed']
|
||||
info['name'] = f"{elem_info['name']}[]"
|
||||
elif isinstance(llvm_type, ir.VoidType):
|
||||
info['size'] = 0
|
||||
info['signed'] = False
|
||||
info['name'] = "void"
|
||||
else:
|
||||
info['size'] = 0
|
||||
info['signed'] = False
|
||||
info['name'] = str(llvm_type)
|
||||
return info
|
||||
|
||||
def _infer_expr_llvm_type_full(self, node):
|
||||
Gen = self.Trans.LlvmGen
|
||||
if isinstance(node, ast.Constant):
|
||||
if isinstance(node.value, bool):
|
||||
return ir.IntType(32)
|
||||
elif isinstance(node.value, int):
|
||||
return ir.IntType(32)
|
||||
elif isinstance(node.value, float):
|
||||
return ir.DoubleType()
|
||||
elif isinstance(node.value, str):
|
||||
if getattr(node, 'kind', None) == 'u':
|
||||
return ir.PointerType(ir.IntType(16))
|
||||
return ir.PointerType(ir.IntType(8))
|
||||
return ir.IntType(32)
|
||||
elif isinstance(node, ast.Name):
|
||||
if node.id in Gen.variables:
|
||||
var_ptr = Gen.variables[node.id]
|
||||
if isinstance(var_ptr.type, ir.PointerType):
|
||||
return var_ptr.type.pointee
|
||||
return var_ptr.type
|
||||
if node.id in Gen._reg_values:
|
||||
return Gen._reg_values[node.id].type
|
||||
return ir.IntType(32)
|
||||
elif isinstance(node, ast.Attribute):
|
||||
attr_name = node.attr
|
||||
if attr_name in self.Trans.SymbolTable:
|
||||
AttrInfo = self.Trans.SymbolTable[attr_name]
|
||||
if getattr(AttrInfo, 'IsEnumMember', None):
|
||||
return ir.IntType(32)
|
||||
return ir.IntType(32)
|
||||
elif isinstance(node, ast.BinOp):
|
||||
left_type = self._infer_expr_llvm_type_full(node.left)
|
||||
right_type = self._infer_expr_llvm_type_full(node.right)
|
||||
if isinstance(left_type, (ir.FloatType, ir.DoubleType)) or isinstance(right_type, (ir.FloatType, ir.DoubleType)):
|
||||
return ir.DoubleType()
|
||||
if isinstance(left_type, ir.PointerType) or isinstance(right_type, ir.PointerType):
|
||||
return ir.PointerType(ir.IntType(8))
|
||||
return ir.IntType(32)
|
||||
elif isinstance(node, ast.UnaryOp):
|
||||
return self._infer_expr_llvm_type_full(node.operand)
|
||||
elif isinstance(node, ast.Subscript):
|
||||
val_type = self._infer_expr_llvm_type_full(node.value)
|
||||
if isinstance(val_type, ir.PointerType):
|
||||
if isinstance(val_type.pointee, ir.IntType) and val_type.pointee.width == 8:
|
||||
return ir.IntType(8)
|
||||
if isinstance(val_type.pointee, ir.ArrayType):
|
||||
return ir.PointerType(val_type.pointee.element)
|
||||
return val_type.pointee
|
||||
if isinstance(val_type, ir.ArrayType):
|
||||
return val_type.element
|
||||
if isinstance(val_type, ir.IntType):
|
||||
return ir.IntType(8)
|
||||
return ir.IntType(32)
|
||||
elif isinstance(node, ast.Call):
|
||||
if isinstance(node.func, ast.Name):
|
||||
func_name = node.func.id
|
||||
if Gen._has_function(func_name):
|
||||
fn = Gen._get_function(func_name)
|
||||
return fn.function_type.return_type
|
||||
if func_name == 'len':
|
||||
return ir.IntType(32)
|
||||
if func_name == 'sizeof':
|
||||
return ir.IntType(32)
|
||||
if func_name == 'ord':
|
||||
return ir.IntType(32)
|
||||
if func_name == 'chr':
|
||||
return ir.IntType(8)
|
||||
if func_name == 'int':
|
||||
return ir.IntType(32)
|
||||
if func_name == 'float' or func_name == 'double':
|
||||
return ir.DoubleType()
|
||||
elif isinstance(node.func, ast.Attribute):
|
||||
if isinstance(node.func.value, ast.Name) and node.func.value.id == 'c':
|
||||
if node.func.attr == 'Deref':
|
||||
return ir.IntType(64)
|
||||
if isinstance(node.func.value, ast.Name) and node.func.value.id == 't':
|
||||
if node.func.attr == 'CChar':
|
||||
return ir.PointerType(ir.IntType(8))
|
||||
return ir.IntType(32)
|
||||
elif isinstance(node, ast.Compare):
|
||||
return ir.IntType(1)
|
||||
elif isinstance(node, ast.BoolOp):
|
||||
return ir.IntType(1)
|
||||
elif isinstance(node, ast.IfExp):
|
||||
return self._infer_expr_llvm_type_full(node.body)
|
||||
elif isinstance(node, ast.Attribute):
|
||||
return ir.PointerType(ir.IntType(8))
|
||||
return ir.IntType(32)
|
||||
|
||||
def _collect_names(self, node, bound, free=None):
|
||||
"""收集 AST 节点中的绑定变量和自由变量
|
||||
|
||||
Args:
|
||||
node: AST 节点
|
||||
bound: 集合，收集绑定变量（ast.Store 上下文）
|
||||
free: 集合，收集自由变量（ast.Load 上下文），可为 None
|
||||
"""
|
||||
if free is None:
|
||||
free = set()
|
||||
if isinstance(node, ast.Name):
|
||||
if isinstance(node.ctx, ast.Store):
|
||||
bound.add(node.id)
|
||||
elif isinstance(node.ctx, ast.Load):
|
||||
free.add(node.id)
|
||||
elif isinstance(node, (ast.Lambda, ast.FunctionDef, ast.AsyncFunctionDef)):
|
||||
# 这些节点有自己的作用域，不递归进入
|
||||
# 但需要收集参数名作为 bound
|
||||
if hasattr(node, 'args') and node.args:
|
||||
for arg in node.args.args:
|
||||
bound.add(arg.arg)
|
||||
elif getattr(node, '_fields', None):
|
||||
for field in node._fields:
|
||||
value = getattr(node, field, None)
|
||||
if isinstance(value, list):
|
||||
for item in value:
|
||||
if isinstance(item, (ast.stmt, ast.expr)):
|
||||
self._collect_names(item, bound, free)
|
||||
elif isinstance(value, (ast.stmt, ast.expr)):
|
||||
self._collect_names(value, bound, free)
|
||||
from __future__ import annotations
|
||||
from typing import TYPE_CHECKING
|
||||
if TYPE_CHECKING:
|
||||
from lib.core.translator import Translator
|
||||
import ast
|
||||
import llvmlite.ir as ir
|
||||
|
||||
|
||||
class ExprUtils:
|
||||
def __init__(self, translator: "Translator"):
|
||||
self.Trans = translator
|
||||
|
||||
def GetOpSymbol(self, Op):
|
||||
OpMap = {
|
||||
ast.Add: '+', ast.Sub: '-', ast.Mult: '*', ast.Div: '/',
|
||||
ast.FloorDiv: '//', ast.Mod: '%', ast.Pow: '**', ast.BitAnd: '&',
|
||||
ast.BitOr: '|', ast.BitXor: '^', ast.LShift: '<<', ast.RShift: '>>',
|
||||
ast.MatMult: '@',
|
||||
}
|
||||
return OpMap.get(type(Op), None)
|
||||
|
||||
def GetUnaryOpSymbol(self, Op):
|
||||
OpMap = {ast.Invert: '~', ast.Not: 'not', ast.UAdd: '+', ast.USub: '-'}
|
||||
return OpMap.get(type(Op), None)
|
||||
|
||||
def GetComparatorSymbol(self, Op):
|
||||
OpMap = {
|
||||
ast.Eq: '==', ast.NotEq: '!=', ast.Lt: '<', ast.LtE: '<=',
|
||||
ast.Gt: '>', ast.GtE: '>=', ast.Is: 'is', ast.IsNot: 'is not',
|
||||
ast.In: 'in', ast.NotIn: 'not in',
|
||||
}
|
||||
return OpMap.get(type(Op), None)
|
||||
|
||||
def _get_var_class(self, node, Gen):
|
||||
if isinstance(node, ast.Name):
|
||||
VarName = node.id
|
||||
return Gen.var_struct_class.get(VarName)
|
||||
if isinstance(node, ast.Attribute):
|
||||
AttrName = node.attr
|
||||
return Gen.var_struct_class.get(AttrName)
|
||||
return None
|
||||
|
||||
def _try_operator_overLoad(self, ClassName, op_name, obj_val, Gen, other_val=None):
|
||||
FullMethodName = f'{ClassName}.{op_name}'
|
||||
if not Gen._has_function(FullMethodName):
|
||||
FullMethodName = f'{ClassName}.{op_name}__'
|
||||
if Gen._has_function(FullMethodName):
|
||||
func = Gen._get_function(FullMethodName)
|
||||
call_args = [obj_val]
|
||||
if other_val is not None:
|
||||
call_args.append(other_val)
|
||||
if len(func.ftype.args) == len(call_args):
|
||||
for i, (expected_type, actual_val) in enumerate(zip(func.ftype.args, call_args)):
|
||||
if isinstance(expected_type, ir.PointerType) and isinstance(actual_val.type, ir.PointerType):
|
||||
if expected_type != actual_val.type:
|
||||
if isinstance(expected_type.pointee, (ir.IdentifiedStructType, ir.LiteralStructType)):
|
||||
call_args[i] = Gen.builder.bitcast(actual_val, expected_type, name=f"bitcast_arg_{i}")
|
||||
result = Gen.builder.call(func, call_args, name=f"call_{FullMethodName}")
|
||||
return result
|
||||
return None
|
||||
|
||||
def _llvm_type_to_detailed_string(self, llvm_type, var_name=""):
|
||||
info = {
|
||||
'name': "unknown",
|
||||
'size': 0,
|
||||
'signed': False,
|
||||
'ptr': False
|
||||
}
|
||||
if isinstance(llvm_type, ir.IntType):
|
||||
info['size'] = llvm_type.width // 8
|
||||
info['signed'] = True
|
||||
if llvm_type.width == 8:
|
||||
info['name'] = "char"
|
||||
elif llvm_type.width == 16:
|
||||
info['name'] = "short"
|
||||
elif llvm_type.width == 32:
|
||||
info['name'] = "int"
|
||||
elif llvm_type.width == 64:
|
||||
info['name'] = "long long"
|
||||
else:
|
||||
info['name'] = f"int{llvm_type.width}"
|
||||
elif isinstance(llvm_type, ir.DoubleType):
|
||||
info['size'] = 8
|
||||
info['signed'] = True
|
||||
info['name'] = "double"
|
||||
elif isinstance(llvm_type, ir.FloatType):
|
||||
info['size'] = 4
|
||||
info['signed'] = True
|
||||
info['name'] = "float"
|
||||
elif isinstance(llvm_type, ir.PointerType):
|
||||
info['size'] = 8
|
||||
info['signed'] = False
|
||||
info['ptr'] = True
|
||||
pointee_info = self._llvm_type_to_detailed_string(llvm_type.pointee)
|
||||
info['name'] = f"{pointee_info['name']}*"
|
||||
elif isinstance(llvm_type, (ir.LiteralStructType, ir.IdentifiedStructType)):
|
||||
size = 0
|
||||
for elem in llvm_type.elements:
|
||||
elem_info = self._llvm_type_to_detailed_string(elem)
|
||||
size += elem_info['size']
|
||||
info['size'] = size
|
||||
info['signed'] = False
|
||||
if isinstance(llvm_type, ir.IdentifiedStructType):
|
||||
info['name'] = llvm_type.name
|
||||
else:
|
||||
info['name'] = "struct"
|
||||
elif isinstance(llvm_type, ir.ArrayType):
|
||||
elem_info = self._llvm_type_to_detailed_string(llvm_type.element)
|
||||
info['size'] = elem_info['size'] * llvm_type.count
|
||||
info['signed'] = elem_info['signed']
|
||||
info['name'] = f"{elem_info['name']}[]"
|
||||
elif isinstance(llvm_type, ir.VoidType):
|
||||
info['size'] = 0
|
||||
info['signed'] = False
|
||||
info['name'] = "void"
|
||||
else:
|
||||
info['size'] = 0
|
||||
info['signed'] = False
|
||||
info['name'] = str(llvm_type)
|
||||
return info
|
||||
|
||||
def _infer_expr_llvm_type_full(self, node):
|
||||
Gen = self.Trans.LlvmGen
|
||||
if isinstance(node, ast.Constant):
|
||||
if isinstance(node.value, bool):
|
||||
return ir.IntType(32)
|
||||
elif isinstance(node.value, int):
|
||||
return ir.IntType(32)
|
||||
elif isinstance(node.value, float):
|
||||
return ir.DoubleType()
|
||||
elif isinstance(node.value, str):
|
||||
if getattr(node, 'kind', None) == 'u':
|
||||
return ir.PointerType(ir.IntType(16))
|
||||
return ir.PointerType(ir.IntType(8))
|
||||
return ir.IntType(32)
|
||||
elif isinstance(node, ast.Name):
|
||||
if node.id in Gen.variables:
|
||||
var_ptr = Gen.variables[node.id]
|
||||
if isinstance(var_ptr.type, ir.PointerType):
|
||||
return var_ptr.type.pointee
|
||||
return var_ptr.type
|
||||
if node.id in Gen._reg_values:
|
||||
return Gen._reg_values[node.id].type
|
||||
return ir.IntType(32)
|
||||
elif isinstance(node, ast.Attribute):
|
||||
attr_name = node.attr
|
||||
if attr_name in self.Trans.SymbolTable:
|
||||
AttrInfo = self.Trans.SymbolTable[attr_name]
|
||||
if getattr(AttrInfo, 'IsEnumMember', None):
|
||||
return ir.IntType(32)
|
||||
return ir.IntType(32)
|
||||
elif isinstance(node, ast.BinOp):
|
||||
left_type = self._infer_expr_llvm_type_full(node.left)
|
||||
right_type = self._infer_expr_llvm_type_full(node.right)
|
||||
if isinstance(left_type, (ir.FloatType, ir.DoubleType)) or isinstance(right_type, (ir.FloatType, ir.DoubleType)):
|
||||
return ir.DoubleType()
|
||||
if isinstance(left_type, ir.PointerType) or isinstance(right_type, ir.PointerType):
|
||||
return ir.PointerType(ir.IntType(8))
|
||||
return ir.IntType(32)
|
||||
elif isinstance(node, ast.UnaryOp):
|
||||
return self._infer_expr_llvm_type_full(node.operand)
|
||||
elif isinstance(node, ast.Subscript):
|
||||
val_type = self._infer_expr_llvm_type_full(node.value)
|
||||
if isinstance(val_type, ir.PointerType):
|
||||
if isinstance(val_type.pointee, ir.IntType) and val_type.pointee.width == 8:
|
||||
return ir.IntType(8)
|
||||
if isinstance(val_type.pointee, ir.ArrayType):
|
||||
return ir.PointerType(val_type.pointee.element)
|
||||
return val_type.pointee
|
||||
if isinstance(val_type, ir.ArrayType):
|
||||
return val_type.element
|
||||
if isinstance(val_type, ir.IntType):
|
||||
return ir.IntType(8)
|
||||
return ir.IntType(32)
|
||||
elif isinstance(node, ast.Call):
|
||||
if isinstance(node.func, ast.Name):
|
||||
func_name = node.func.id
|
||||
if Gen._has_function(func_name):
|
||||
fn = Gen._get_function(func_name)
|
||||
return fn.function_type.return_type
|
||||
if func_name == 'len':
|
||||
return ir.IntType(32)
|
||||
if func_name == 'sizeof':
|
||||
return ir.IntType(32)
|
||||
if func_name == 'ord':
|
||||
return ir.IntType(32)
|
||||
if func_name == 'chr':
|
||||
return ir.IntType(8)
|
||||
if func_name == 'int':
|
||||
return ir.IntType(32)
|
||||
if func_name == 'float' or func_name == 'double':
|
||||
return ir.DoubleType()
|
||||
elif isinstance(node.func, ast.Attribute):
|
||||
if isinstance(node.func.value, ast.Name) and node.func.value.id == 'c':
|
||||
if node.func.attr == 'Deref':
|
||||
return ir.IntType(64)
|
||||
if isinstance(node.func.value, ast.Name) and node.func.value.id == 't':
|
||||
if node.func.attr == 'CChar':
|
||||
return ir.PointerType(ir.IntType(8))
|
||||
return ir.IntType(32)
|
||||
elif isinstance(node, ast.Compare):
|
||||
return ir.IntType(1)
|
||||
elif isinstance(node, ast.BoolOp):
|
||||
return ir.IntType(1)
|
||||
elif isinstance(node, ast.IfExp):
|
||||
return self._infer_expr_llvm_type_full(node.body)
|
||||
elif isinstance(node, ast.Attribute):
|
||||
return ir.PointerType(ir.IntType(8))
|
||||
return ir.IntType(32)
|
||||
|
||||
def _collect_names(self, node, bound, free=None):
|
||||
"""收集 AST 节点中的绑定变量和自由变量
|
||||
|
||||
Args:
|
||||
node: AST 节点
|
||||
bound: 集合，收集绑定变量（ast.Store 上下文）
|
||||
free: 集合，收集自由变量（ast.Load 上下文），可为 None
|
||||
"""
|
||||
if free is None:
|
||||
free = set()
|
||||
if isinstance(node, ast.Name):
|
||||
if isinstance(node.ctx, ast.Store):
|
||||
bound.add(node.id)
|
||||
elif isinstance(node.ctx, ast.Load):
|
||||
free.add(node.id)
|
||||
elif isinstance(node, (ast.Lambda, ast.FunctionDef, ast.AsyncFunctionDef)):
|
||||
# 这些节点有自己的作用域，不递归进入
|
||||
# 但需要收集参数名作为 bound
|
||||
if hasattr(node, 'args') and node.args:
|
||||
for arg in node.args.args:
|
||||
bound.add(arg.arg)
|
||||
elif getattr(node, '_fields', None):
|
||||
for field in node._fields:
|
||||
value = getattr(node, field, None)
|
||||
if isinstance(value, list):
|
||||
for item in value:
|
||||
if isinstance(item, (ast.stmt, ast.expr)):
|
||||
self._collect_names(item, bound, free)
|
||||
elif isinstance(value, (ast.stmt, ast.expr)):
|
||||
self._collect_names(value, bound, free)
|
||||
|
||||
Reference in New Issue
Block a user