來源:Python中文社群

ID:python-china

起步

透過內建方法 isinstance(object, classinfo) 可以判斷一個物件是否是某個類的實體。但你是否想過關於鴨子協議的物件是如何進行判斷的呢？ 比如 list 類的父類是繼 object 類的，但透過 isinstance([], typing.Iterable) 傳回的卻是真，難道 list 是可迭代的子類？

根據 PEP 3119 的描述中得知實體的檢查是允許多載的：

The primary mechanism proposed here is to allow overloading the built-in functions isinstance() and issubclass(). The overloading works as follows: The call isinstance(x, C) first checks whether C.__instancecheck__ exists, and if so, calls C.__instancecheck__(x) instead of its normal implementation.

這段話的意思是，當呼叫 isinstance(x, C) 進行檢測時，會優先檢查是否存在 C.instancecheck，如果存在則呼叫 C.instancecheck(x) ，傳回的結果便是實體檢測的結果，預設的判斷方式就沒有了。

這種方式有助於我們來檢查鴨子型別，我用程式碼測了一下。

class Sizeable(object):
    def __instancecheck__(cls, instance):
        print("__instancecheck__ call")
        return hasattr(instance, "__len__")
class B(object):
    pass
b = B()
print(isinstance(b, Sizeable)) # output:False

[abstract.c]
int
PyObject_IsInstance(PyObject *inst, PyObject *cls)
{
    _Py_IDENTIFIER(__instancecheck__);
    PyObject *checker;
    /* Quick test for an exact match */
    if (Py_TYPE(inst) == (PyTypeObject *)cls)
        return 1;
    ....
}

/* We know what type's __instancecheck__ does. */
    if (PyType_CheckExact(cls)) {
        return recursive_isinstance(inst, cls);
    }

[object.h]
#define PyType_CheckExact(op) (Py_TYPE(op) == &PyType_Type)

def recursive_isinstance(inst, cls):
    return pyType_IsSubtype(inst, cls)
def pyType_IsSubtype(a, b):
    for mro in a.__class__.__mro__:
        if mro is b:
            return True
    return False

if (PyTuple_Check(cls)) {
    ...
}

checker = _PyObject_LookupSpecial(cls, &PyId___instancecheck__);
if (checker != NULL) {
    res = PyObject_CallFunctionObjArgs(checker, inst, NULL);
    ok = PyObject_IsTrue(res);
    return ok;
}

[typeobject.c]
PyObject *
_PyObject_LookupSpecial(PyObject *self, _Py_Identifier *attrid)
{
    PyObject *res;
    res = _PyType_LookupId(Py_TYPE(self), attrid);
    // 有回呼的話處理回呼
    // ...
    return res;
}

def _isinstance(x, C):
    # 快速匹配
    if type(x) is C:
        return True
    # 如果是由元類 type 直接構造的類
    if type(C) is type:
        return C in x.__class__.__mro__
    # 如果第二個引數是元組, 則遞迴呼叫
    if type(C) is tuple:
        for item in C:
            r = _isinstance(x, item)
            if r:
                return r
    # 使用者自定義檢測規則
    if hasattr(C, "__instancecheck__"):
        return C.__instancecheck__(x)
    # 預設行為
    return C in x.__class__.__mro__

class MetaSizeable(type):
    def __instancecheck__(cls, instance):
        print("__instancecheck__ call")
        return hasattr(instance, "__len__")
class Sizeable(metaclass=MetaSizeable):
    pass
class B(object):
    pass
b = B()
print(isinstance(b, Sizeable))  # output: False
print(isinstance([], Sizeable)) # output: True