在Python

时间:2018-05-19 21:34:56

标签: python oop inheritance deep-copy

我正在尝试在同一个类中创建一个类的新对象,但是它不是创建一个全新的对象,而是简单地创建一个对我正在使用的同一个对象的新引用。

因此,如果我改变一个对象的值,它也会改变另一个对象的值 - 即使我可能应该有两个完全不同的对象。

使用copy.deepcopy()方法修复了引用的问题,但我猜它也应该以不同的方式工作。

我的代码在这个特定的实现中如何表现?是否有理由创建同一对象的浅表副本,即使代码可能应该创建它的新实例?

这是一个略微减少的代码段:

class Vec4():
    def __init__(self, x = 0, y = 0, z = 0, w = 0):
        self.values = [x,y,z,w]

    def __str__(self):
        return str(self.values[0]) + ' ' + str(self.values[1]) + ' ' + str(self.values[2]) + ' ' + str(self.values[3])

    def setValue(self, index, value):
        self.values[index] = value


    def scalar(self, vector):
        """returns the result of the scalar multiplication"""
        result = 0
        for u in range(4):
            result += self.values[u] * vector.values[u]
        return result




class Matrix4():
    def __init__(self, row1 = Vec4(), row2 = Vec4(), row3 = Vec4(), row4 = Vec4()):
        self.m_values = [row1,row2,row3,row4]
        self.trans_values = [Vec4(),Vec4(),Vec4(),Vec4()]
        self.set_transp_matrix()

    def __str__(self):
        return self.m_values[0].__str__() + '\n' + self.m_values[1].__str__() + '\n' + self.m_values[2].__str__() + '\n' + self.m_values[3].__str__()

    def setIdentity(self):

        identity = Matrix4(Vec4(1,0,0,0),
                       Vec4(0,1,0,0),
                       Vec4(0,0,1,0),
                       Vec4(0,0,0,1))
        for i in range(4):
            for j in range(4):
                self.m_values[i].values[j] = identity.m_values[i].values[j]

    def set_transp_matrix(self):
         for t in range(4):
            for s in range(4):
                self.trans_values[t].values[s] = self.m_values[s].values[t]

    def get_trans_matrix(self):
        return self.trans_values[0].__str__() + '\n' + self.trans_values[1].__str__() + '\n' + self.trans_values[2].__str__() + '\n' + self.trans_values[3].__str__()

    def mulM(self, m):

        print(self, "\n")
        matrixResult = Matrix4()
        print(matrixResult, "\n")
        for row in range(4):  # rows of self
            for element in range(4):
                value = self.m_values[row].scalar(m.trans_values[element])
                matrixResult.m_values[row].setValue(element, value)
        return matrixResult


class ScaleMatrix(Matrix4):

    def __init__(self, m_scale = Vec4(1,1,1), *args, **kwargs):
        super(ScaleMatrix, self).__init__(*args, **kwargs)
        self.m_scale = m_scale
        self.update()

    def getScale(self):
        """Returns the scale vector, only x, y and z are relevant"""
        return self.m_scale

    def setScale(self, v):
        """Sets the scale vector, only x, y and z are relevant"""
        self.m_scale = v
        self.update()

    def update(self):
        """Calculates the scale matrix"""

        self.setIdentity()

        for i in range(3):
            self.m_values[i].values[i] = self.getScale().values[i]

        return self


if __name__ == "__main__":
    #Simple Constructor and Print
    a = Vec4(1,2,3,4)
    b = Vec4(5,6,7,8)
    c = Vec4(9,10,11,12)
    d = Vec4(13,14,15,16)


    A = Matrix4(a, b, c, d)
    D = ScaleMatrix()
    D.setScale(Vec4(3, 4, 5, 1))

    print(D.mulM(A))

问题出在班级Matrix4,方法mulM(),其中matrixResult = Matrix4()应创建Matrix4()的全新实例(其中Vec4()的所有值都应该是0)而不是简单地复制self对象。 print的输出显示以下内容:

3 0 0 0
0 4 0 0
0 0 5 0
0 0 0 1 

3 0 0 0
0 4 0 0
0 0 5 0
0 0 0 1 

3 6 51 672
20 64 508 6688
45 140 1170 15340
13 40 334 4396

所以第二个矩阵不应该等于第一个矩阵。 但是,如果我创建了一个普通的Matrix4()对象,而不是在上面的代码段末尾扩展ScaleMatrix()的{​​{1}}对象,则不会出现此问题。

Python v.3.6.4

2 个答案:

答案 0 :(得分:2)

Python在定义函数时评估默认参数。定义时:

class Matrix4():
    def __init__(self, row1 = Vec4() ...

您创建了一个Vec4实例,每次调用此__init__方法时,此实例都将用作默认值。

第一次创建Matrix4实例时,__init__将会执行,此实例将通过名称row1

引用

然后你有:

    self.m_values = [row1,row2,row3,row4]

所以这个实例现在由self.m_values [0]引用。

稍后,在ScaleMatrix.update中,您更新此Vec4实例:

for i in range(3):
            self.m_values[i].values[i] = self.getScale().values[i]

下次您将在没有参数的情况下致电Matrix4.__init__,系统会使用默认值,这是您刚刚更新的Vec4

当您使用空列表作为默认参数时,您有类似的行为,请参阅“Least Astonishment” and the Mutable Default Argument

避免此问题的常用方法是避免将可变对象用作默认参数。你可以这样做:

class Matrix4():
    def __init__(self, row1 = None, row2 = None, row3 = None, row4 = None):
        if row1 is None:
            row1 = Vec4()
        if row2 is None:
            row2 = Vec4()
        if row3 is None:
            row3 = Vec4()
        if row4 is None:
            row4 = Vec4()
        self.m_values = [row1,row2,row3,row4]
        self.trans_values = [Vec4(),Vec4(),Vec4(),Vec4()]
        self.set_transp_matrix()

作为输出:

3 0 0 0
0 4 0 0
0 0 5 0
0 0 0 1 

0 0 0 0
0 0 0 0
0 0 0 0
0 0 0 0 

3 6 9 12
20 24 28 32
45 50 55 60
13 14 15 16

答案 1 :(得分:1)

结帐Python constructor and default value"Least Astonishment" and the Mutable Default Argument您的问题与上述相关问题中描述的问题基本相似。

此外,在声明一个类时,将其设为/bin$ realpath awk /usr/bin/gawk

的子类
object

最后,要解决您的特定问题,请删除初始值设定项的默认值。例如,做这样的事情:

class Vec4(object):
class Matrix4(object):

class Matrix4(): def __init__(self, row1=None, row2=None, row3=None, row4=None): if row1 is None: row1 = Vec4() if row2 is None: row2 = Vec4() if row3 is None: row3 = Vec4() if row4 is None: row4 = Vec4() self.m_values = [row1,row2,row3,row4] 参数ScaleMatrix的默认值相同:

m_scale