在Java中建模选择性继承特征?

时间:2016-07-19 02:20:49

标签: java

我不确定我是否在问正确的问题。如果我错了,请更正。

我有3个模型继承模型,其中一些子类不应该继承父属性。

我们以下面的类为例:

abstract public class Root{
    // Has some attributes that must be inherited by it's sub classes
    List<Child> children;
}

abstract public class Child extends Root {
    // Has the root's attributes, plus additional attributes pertaining to the child class, including the position.
    private int position;
}

abstract public class Leaf extends Child {
    // this class inherits all attributes of child class but it does not have any children
    // How would I exclude the children attribute here?
}

我正在考虑使用接口但是它不允许我为类保存可变数据,就像抽象类一样。例如,我想保存与Child和Leaf类相关的特定变量,例如子项在其父项列表中的位置。

我也不能这样做:

abstract public class Leaf {
    // This child does not have a parent, but then again, it's not considered a child either because it comes before the Root who has children.
}

abstract public class Root extends Leaf {
    List<Child> children;
}

abstract public class Child extends Root {
    private int position;
}

再次,我将如何对此进行建模?

2 个答案:

答案 0 :(得分:0)

你的问题很混乱,因为它在现实和编程方面使用了父/子关系而没有说明你的意思。但是,我想我对你的要求有一个大致的了解。

我相信你要找的是复合图案
(参见:https://en.wikipedia.org/wiki/Composite_pattern

在以下实现中,ChildChildless已成为Composite模式的一部分,并通过添加Root类从Person抽象出一个额外的图层。< / p>

所以你会有

abstract public class Root {
    //attributes that everything has
}

abstract public class Person extends Root {
    protected int position; //All children have parents
    //other attributes
}

abstract public class Child extends Person {
    protected List<Person> children; //But only Child has more children
    //other attributes
}

abstract public class Childless extends Person{
    //attributes
}

Person是“组件”,现在有position字段,因为每个孩子在父母列表中都有一个职位。

Child是“复合”,现在有children字段,因为每个非贫瘠的孩子都可以有更多的孩子。

Childless是“叶子”,因为它不再有更多的孩子。

另外,请注意,子类无法访问private字段,因此请改用protected

答案 1 :(得分:0)

简单地说明没有LeafChild的子类,移动字段,可以正常工作

abstract public class Node {
    private int position; //Every Node can have a position
}

public class InnerNode extends Node { //Renamed for clarity
    List<Node> children; //Children can be InnerNodes or Leafs 
}

abstract public class LeafNode extends Node {
    //As this is no longer an extension of InnerNode, it won't have the children field.
}

这个模型让IMO更有意义。每个LeafNode和InnerNode都被视为节点,但只有InnerNode有子节点。

如果需要,您甚至可以向Node添加一个抽象的getter来返回子项,并让LeafNode返回null或空列表:

abstract public class Node {
    private int position; //Every Node can have a position

    public abstract List<Node> getChildren();
}

public class InnerNode extends Node { //Renamed for clarity
    List<Node> children; //Children can be InnerNodes or Leafs 

    public List<Node> getChildren(){
        return new ArrayList<>(children);
    }
}

abstract public class LeafNode extends Node {
    //As this is no longer an extension of InnerNode, it won't have the children field.

    public List<Node> getChildren() {
        return new ArrayList<>(); //Empty
    }
}

这大致完成了你打算做的事情 - 只有一些节点(InnerNodes)实际上会有子节点,但你可以尝试访问任意Node实例的子节点而不进行转换。现在可以进行以下操作:

Node n = ...
List<Node> children = n.getChildren();