策略设计模式
现在有一个需求: 给商场做一个收银软件. 营业员根据客户购买的产品的单价和数量, 向客户打印小票。
这个实现很简单. 一个类就可以搞定:
package com.designModel.chaper2_strategeModel.step1; import java.util.Scanner; /** * cash收银 * @author samsung * */ public class Cash { public String list = ""; public Double totalPrice = 0.00; public void buttonOK(){ Scanner scan = new Scanner(System.in); System.out.println("输入单价:"); String price = scan.nextLine(); scan = new Scanner(System.in); System.out.println("输入数量:"); String num = scan.nextLine(); scan = new Scanner(System.in); System.out.println("输入折扣:"); String zhekou = scan.nextLine(); Double xiaoji = Double.parseDouble(price) * Integer.parseInt(num) * Double.parseDouble(zhekou)/10; list += "单价:" + price +", 数量: "+ num +",折扣: " + zhekou + "\n"; totalPrice += xiaoji; } public static void main(String[] args) { Cash cash = new Cash(); boolean flag = true; while(flag) { cash.buttonOK(); if(cash.totalPrice > 10){ flag = false; } } System.out.println("========================"); System.out.println("清单:\n " + cash.list); System.out.println("总价: "+ cash.totalPrice); } }
但是: 用面向对象的角度思考. 这个类将前端输入和业务逻辑混合在一块了. 不利于维护, 扩展, 复用, 也不灵活.
假如: 现在商场搞活动, 所有商品打折, 7折,
过一段时间, 商场又搞活动, 所有商品打5折
国庆节, 商品满200减50.
如果按照上面的方式来写代码, 那么每次都要写一遍, 如何将其复用起来呢? 并且每次增加新的活动的时候, 又不会影响到原来的活动.
对了, 简单工厂设计模式, 我们之前刚刚学过的. 下面看看简单工厂设计模式的UML图:
首先, 有一个工厂类, 在这个工厂类里面, 根据类型, 依赖于不同的现金收费方式. 具体代码如下:
首先与一个抽象的现金收费方式类:
package com.designModel.chaper2_strategeModel.step2; /** * 现金收费类 * @author samsung * */ public abstract class CashFee { public abstract double acceptCash(double money); }
定义现金收费方式的实现类,分别是: 正常收费类, 折扣类, 返现类
package com.designModel.chaper2_strategeModel.step2; //正常收费类 public class NormalCashFee extends CashFee { @Override public double acceptCash(double money) { return money; } }
package com.designModel.chaper2_strategeModel.step2; public class DiscountCashFee extends CashFee { private double discount = 0.00; public DiscountCashFee(double discount){ this.discount = discount / 10; } @Override public double acceptCash(double money) { return this.discount * money; } public double getDiscount() { return discount; } public void setDiscount(double discount) { this.discount = discount; } }
package com.designModel.chaper2_strategeModel.step2; public class ReturnCashFee extends CashFee { //基础金额 private double baseCash; //返现金额 private double returnCash; public ReturnCashFee(double baseCash, double returnCash){ this.baseCash = baseCash; this.returnCash = returnCash; } @Override public double acceptCash(double money) { return money - Math.floor(money/300) * 50; } public double getBaseCash() { return baseCash; } public void setBaseCash(double baseCash) { this.baseCash = baseCash; } public double getReturnCash() { return returnCash; } public void setReturnCash(double returnCash) { this.returnCash = returnCash; } }
在定义一个工厂类, 用来产生各种现金收费方式
package com.designModel.chaper2_strategeModel.step2; import java.util.Scanner; public class CashFeeFactory { public static CashFee createCashFee(int type, double discount, double basePrice, double returnPrice){ CashFee cashFee = null; switch(type){ case 1: cashFee = new NormalCashFee(); break; case 2: cashFee = new DiscountCashFee(discount); break; case 3: cashFee = new ReturnCashFee(basePrice, returnPrice); break; } return cashFee; } public static void main(String[] args) { Scanner scan = new Scanner(System.in); System.out.println("输入单价:"); String price = scan.nextLine(); scan = new Scanner(System.in); System.out.println("输入数量:"); String num = scan.nextLine(); scan = new Scanner(System.in); System.out.println("输入折扣类型(1 无折扣 2打折 3满减):"); String zhekou = scan.nextLine(); double discount = 0.0d; double basePrice = 0; double returnPrice = 0; if("2".equals(zhekou)){ scan = new Scanner(System.in); System.out.println("输入折扣:"); discount = Double.parseDouble(scan.nextLine()); } if("3".equals(zhekou)){ scan = new Scanner(System.in); System.out.println("基础金额:"); basePrice = Double.parseDouble(scan.nextLine()); scan = new Scanner(System.in); System.out.println("返现金额:"); returnPrice = Double.parseDouble(scan.nextLine()); } Double xiaoji = Double.parseDouble(price) * Integer.parseInt(num) * Double.parseDouble(zhekou)/10; CashFee cashFee = CashFeeFactory.createCashFee(1, discount, basePrice, returnPrice); } }
完成了, 在main方法里, 我们模拟了客户端, 输入单价和数量,打印小票
-------------------------------------------------------------------------------------------------
思考:
使用简单工厂设计模式的优缺点:
优点:
1. 业务逻辑和前端展示相分离开了。业务逻辑的修改, 不影响前端代码
2. 每一个业务逻辑单独一个类, 修改或者添加一个类, 不会影响到其他的类.
3. 使用工厂类封装了业务逻辑类. 前端不需要知道到底每种业务逻辑具体怎么实现的. 只需要知道他的父类即可.
缺点:
1. 如果活动很频繁, 经常会搞各种各样的活动, 那么业务逻辑类就会有很多, 每一次都要增加一个类.
2. 每增加一个类都要修改工厂类. 修改会非常频繁
小结: 简单工厂设计模式虽然也能解决这个问题, 但这个模式只是解决对类的创建问题.
--------------------------------------------------------------------------
下面介绍策略设计模式:
策略(Strategy)设计模式:定义了算法家族, 分别将算法封装起来, 让他们之间可以相互替换,此模式让算法的变化不会影响到使用算法的客户。
封装变化点, 是面向对象的一种很重要的思维方式.
下面以商场促销打印收银小票为例, 来分析策略设计模式
1. 在策略设计模式中, 有各种算法: 所有的算法有一个算法父类: CashFee. 这个父类是一个抽象类. 里面有一个抽象方法acceptCash(double money)
2. 所有的算法类实现抽象算法父类, 并实现抽象方法
3. 重点: 有一个上下文对象. 这个对象封装了对算法类的调用
下面看代码:
1. 现金类算法, 以及现金的各种实现类算法
package com.designModel.chaper2_strategeModel.step3; /** * 现金收费类 * @author samsung * */ public abstract class CashFee { public abstract double acceptCash(double money); }
package com.designModel.chaper2_strategeModel.step3; //正常收费类 public class NormalCashFee extends CashFee { @Override public double acceptCash(double money) { return money; } }
package com.designModel.chaper2_strategeModel.step3; public class DiscountCashFee extends CashFee { private double discount = 0.00; public DiscountCashFee(double discount){ this.discount = discount / 10; } @Override public double acceptCash(double money) { return this.discount * money; } public double getDiscount() { return discount; } public void setDiscount(double discount) { this.discount = discount; } }
package com.designModel.chaper2_strategeModel.step3; public class ReturnCashFee extends CashFee { //基础金额 private double baseCash; //返现金额 private double returnCash; public ReturnCashFee(double baseCash, double returnCash){ this.baseCash = baseCash; this.returnCash = returnCash; } @Override public double acceptCash(double money) { return money - Math.floor(money/300) * 50; } public double getBaseCash() { return baseCash; } public void setBaseCash(double baseCash) { this.baseCash = baseCash; } public double getReturnCash() { return returnCash; } public void setReturnCash(double returnCash) { this.returnCash = returnCash; } }
最后. 来看看上下文类
package com.designModel.chaper2_strategeModel.step3; import java.util.Scanner; public class CashContext { private CashFee cashFee; public CashContext(CashFee cashFee) { this.cashFee = cashFee; } public static double getResult(double money){return cashFee.acceptCash(money); } public static void main(String[] args) { Scanner scan = new Scanner(System.in); System.out.println("输入单价:"); String price = scan.nextLine(); scan = new Scanner(System.in); System.out.println("输入数量:"); String num = scan.nextLine(); scan = new Scanner(System.in); System.out.println("输入折扣类型(1 无折扣 2打折 3满减):"); String type = scan.nextLine(); double discount = 0.0d; double basePrice = 0; double returnPrice = 0; if("2".equals(type)){ scan = new Scanner(System.in); System.out.println("输入折扣:"); discount = Double.parseDouble(scan.nextLine()); } if("3".equals(type)){ scan = new Scanner(System.in); System.out.println("基础金额:"); basePrice = Double.parseDouble(scan.nextLine()); scan = new Scanner(System.in); System.out.println("返现金额:"); returnPrice = Double.parseDouble(scan.nextLine()); } Double xianjin = Double.parseDouble(price) * Integer.parseInt(num); CashContext cc = null; switch(type){ case "1": cc = new CashContext(new NormalCashFee()); break; case "2": cc = new CashContext(new DiscountCashFee(discount)); break; case "3": cc = new CashContext(new ReturnCashFee(basePrice, returnPrice)); break; } cc.getResult(xianjin); } }
分析:
1. 业务逻辑和前端页面展示分开
2. 有一个context上下文类, 在其内部引用了CashFee类. 构造方法定义了具体的实现类.
3. 这样操作的问题: 客户端依然需要switch判断.
----------------------------------------------
策略设计模式和简单工厂设计模式相结合的方案:
其他都不变: 变化的是CashContext类
package com.designModel.chaper2_strategeModel.step4; import java.util.Scanner; public class CashContext { private CashFee cashFee; public CashContext(int type, double discount, double basePrice, double returnPrice) { switch(type){ case 1: cashFee = new NormalCashFee(); break; case 2: cashFee = new DiscountCashFee(discount); break; case 3: cashFee = new ReturnCashFee(basePrice, returnPrice); break; } } public static double getResult(double money){ CashFee cashFee = null; return cashFee.acceptCash(money); } public static void main(String[] args) { Scanner scan = new Scanner(System.in); System.out.println("输入单价:"); String price = scan.nextLine(); scan = new Scanner(System.in); System.out.println("输入数量:"); String num = scan.nextLine(); scan = new Scanner(System.in); System.out.println("输入折扣类型(1 无折扣 2打折 3满减):"); String zhekou = scan.nextLine(); double discount = 0.0d; double basePrice = 0; double returnPrice = 0; if("2".equals(zhekou)){ scan = new Scanner(System.in); System.out.println("输入折扣:"); discount = Double.parseDouble(scan.nextLine()); } if("3".equals(zhekou)){ scan = new Scanner(System.in); System.out.println("基础金额:"); basePrice = Double.parseDouble(scan.nextLine()); scan = new Scanner(System.in); System.out.println("返现金额:"); returnPrice = Double.parseDouble(scan.nextLine()); } Double xianjin = Double.parseDouble(price) * Integer.parseInt(num); CashContext cc = new CashContext(1, discount, basePrice, returnPrice); cc.getResult(xianjin); } }
我们将前端的switch转移到了CashContext的内部. 这样,前端只需要传递给我, 类型信息就可以了.
下面来看看: 简单工厂设计模式 和 策略模式+简单工厂设计模式的区别:
对于客户端而言: 简单工厂设计模式, 客户端需要知道两个类, 而简单工厂+策略设计模式, 客户端只需要知道一个类, 减低了耦合性.
---------------------------------------------------------------------------------------
使用策略模式的优点:
1. 策略模式定义了一系列算法的方法, 从概念上来看, 这些算法完成的都是相同的工作, 只是实现不同,他可以以相同的方式调用算法, 减少了各种算法类与使用算法类之间的耦合。
2. 策略模式的另一个优点: 简化了单元测试, 因为每个算法都有自己的类, 可以通过自己的接口单独测试.
3. 每个算法都是独立的, 修改其中一个, 不会影响其他的算法.
4. 在策略模式中, 选择所用具体实现的职责有客户端对象承担, 并转给策略模式的Context对象。这本身并没有解除客户端需要选择判断的压力。 而策略模式和简单工厂设计模式结合后, 选择具体实现的职责也交给了Context来承担, 最大化的减轻了客户端的职责。这比期初的策略模式好用,但是依然不够完美。因为他在context里还是用到了switch,这样人很不爽。如何才能不用switch呢? 反射。
---------------------------------------------------------------------------------------
思考,
书上说: 为什么要用策略设计模式呢?
因为商场打折的算法很多, 每次增加一个促销活动, 可能就会有一个新的算法. 这时候, 如果用简单工厂设计模式, 需要增加一个算法类, 同时修改工厂类. 这是, 最好使用策略设计模式.
疑问: 使用策略设计模式. 每次增加一个算法, 也要增加一个算法类, 而且要修改CashContext类. 并没有得到简化呀
应该是我修行不够, 还没能够理解, 忘懂得大神指点.
-------------------------------------------------------------------
后续---使用反射加载类, 这里的修改主要是对Context中构造方法的修改.
package com.designModel.chaper2_strategeModel.step5; import java.lang.reflect.Constructor; import java.lang.reflect.InvocationTargetException; import java.util.HashMap; import java.util.Map; import java.util.Scanner; public class CashContext { private CashFee cashFee = null; static Map<String, String> map = new HashMap<String, String>(); static{ map.put("1", "com.designModel.chaper2_strategeModel.step5.NormalCashFee"); map.put("2", "com.designModel.chaper2_strategeModel.step5.DiscountCashFee"); map.put("3", "com.designModel.chaper2_strategeModel.step5.ReturnCashFee"); } public CashContext(String type, double discount, double basePrice, double returnPrice) throws ClassNotFoundException, NoSuchMethodException, SecurityException { /** * 使用反射加载策略算法 */ try { String className = map.get(type); System.out.println(className); Class cla = Class.forName(className); //Class cla = Class.forName("com.designModel.chaper2_strategeModel.step5.NormalCashFee"); Constructor cons = cla.getConstructor(double.class,double.class,double.class); cashFee = (CashFee) cons.newInstance(discount, basePrice, returnPrice); } catch (Exception e) { e.printStackTrace(); } /*switch(type){ case 1: cashFee = new NormalCashFee(); break; case 2: cashFee = new DiscountCashFee(discount); break; case 3: cashFee = new ReturnCashFee(basePrice, returnPrice); break; }*/ } public double getResult(double money){ return cashFee.acceptCash(money); } public static void main(String[] args) { Scanner scan = new Scanner(System.in); System.out.println("输入单价:"); String price = scan.nextLine(); scan = new Scanner(System.in); System.out.println("输入数量:"); String num = scan.nextLine(); scan = new Scanner(System.in); System.out.println("输入折扣类型(1:无折扣 2:打折 3:满减):"); String zhekou = scan.nextLine(); double discount = 0.0d; double basePrice = 0; double returnPrice = 0; if("2".equals(zhekou)){ scan = new Scanner(System.in); System.out.println("输入折扣:"); discount = Double.parseDouble(scan.nextLine()); } if("3".equals(zhekou)){ scan = new Scanner(System.in); System.out.println("基础金额:"); basePrice = Double.parseDouble(scan.nextLine()); scan = new Scanner(System.in); System.out.println("返现金额:"); returnPrice = Double.parseDouble(scan.nextLine()); } Double xianjin = Double.parseDouble(price) * Integer.parseInt(num); CashContext cc; try { cc = new CashContext(zhekou, discount, basePrice, returnPrice); double result = cc.getResult(xianjin); System.out.println("result:"+result); } catch (Exception e) { e.printStackTrace(); } } }
这样做的好处很明显, 不用每次增加一个类, 添加一个switch分支啦. 并且消除了switch分支.
今天有回顾了一下策略模式, 似乎有了一点点更加深入的理解.
-----------------------------------------------------------------------
下面来一个最终版的全部代码:
package com.designModel.chaper2_strategeModel.step5; /** * 现金收费类 * @author samsung * */ public abstract class CashFee { private double discount; private double basePrice; private double returnPrice; public CashFee(){} public CashFee(double discount, double basePrice, double returnPrice){ this.discount = discount; this.basePrice = basePrice; this.returnPrice = returnPrice; } public abstract double acceptCash(double money); public double getDiscount() { return discount; } public void setDiscount(double discount) { this.discount = discount; } public double getBasePrice() { return basePrice; } public void setBasePrice(double basePrice) { this.basePrice = basePrice; } public double getReturnPrice() { return returnPrice; } public void setReturnPrice(double returnPrice) { this.returnPrice = returnPrice; } }
package com.designModel.chaper2_strategeModel.step5; //正常收费类 public class NormalCashFee extends CashFee { public NormalCashFee(double discount, double basePrice, double returnPrice){ }; public NormalCashFee(){} @Override public double acceptCash(double money) { return money; } }
package com.designModel.chaper2_strategeModel.step5; public class ReturnCashFee extends CashFee { /** * @param baseCash 基础金额 * @param returnCash 返现金额 */ public ReturnCashFee(double discount, double basePrice, double returnPrice){ super(0, basePrice, returnPrice); } @Override public double acceptCash(double money) { return money - Math.floor(money/super.getBasePrice()) * super.getReturnPrice(); } }
package com.designModel.chaper2_strategeModel.step5; public class DiscountCashFee extends CashFee { public DiscountCashFee(){} public DiscountCashFee(double discount, double basePrice, double returnPrice){ super(discount/10, 0, 0); } @Override public double acceptCash(double money) { return super.getDiscount() * money; } }
package com.designModel.chaper2_strategeModel.step5; import java.lang.reflect.Constructor; import java.lang.reflect.InvocationTargetException; import java.util.HashMap; import java.util.Map; import java.util.Scanner; public class CashContext { private CashFee cashFee = null; static Map<String, String> map = new HashMap<String, String>(); static{ map.put("1", "com.designModel.chaper2_strategeModel.step5.NormalCashFee"); map.put("2", "com.designModel.chaper2_strategeModel.step5.DiscountCashFee"); map.put("3", "com.designModel.chaper2_strategeModel.step5.ReturnCashFee"); } public CashContext(String type, double discount, double basePrice, double returnPrice) throws ClassNotFoundException, NoSuchMethodException, SecurityException { /** * 使用反射加载策略算法 */ try { String className = map.get(type); System.out.println(className); Class cla = Class.forName(className); //Class cla = Class.forName("com.designModel.chaper2_strategeModel.step5.NormalCashFee"); Constructor cons = cla.getConstructor(double.class,double.class,double.class); cashFee = (CashFee) cons.newInstance(discount, basePrice, returnPrice); } catch (Exception e) { e.printStackTrace(); } /*switch(type){ case 1: cashFee = new NormalCashFee(); break; case 2: cashFee = new DiscountCashFee(discount); break; case 3: cashFee = new ReturnCashFee(basePrice, returnPrice); break; }*/ } public double getResult(double money){ return cashFee.acceptCash(money); } public static void main(String[] args) { Scanner scan = new Scanner(System.in); System.out.println("输入单价:"); String price = scan.nextLine(); scan = new Scanner(System.in); System.out.println("输入数量:"); String num = scan.nextLine(); scan = new Scanner(System.in); System.out.println("输入折扣类型(1:无折扣 2:打折 3:满减):"); String zhekou = scan.nextLine(); double discount = 0.0d; double basePrice = 0; double returnPrice = 0; if("2".equals(zhekou)){ scan = new Scanner(System.in); System.out.println("输入折扣:"); discount = Double.parseDouble(scan.nextLine()); } if("3".equals(zhekou)){ scan = new Scanner(System.in); System.out.println("基础金额:"); basePrice = Double.parseDouble(scan.nextLine()); scan = new Scanner(System.in); System.out.println("返现金额:"); returnPrice = Double.parseDouble(scan.nextLine()); } Double xianjin = Double.parseDouble(price) * Integer.parseInt(num); CashContext cc; try { cc = new CashContext(zhekou, discount, basePrice, returnPrice); double result = cc.getResult(xianjin); System.out.println("result:"+result); } catch (Exception e) { e.printStackTrace(); } } }
亲测代码可用, 一下是运行结果: 分为三种
第一种: 正常收费
第二种: 打折, 打5折
第三种: 满减, 满300减50
------------------------------------------------------
问题:
使用反射, 依然会有问题, 从代码就可以看出. 为了能够使用反射调用通用的构造函数, 所以在CashFee中定义了一个带有三个参数的构造方法
public CashFee(double discount, double basePrice, double returnPrice){ this.discount = discount; this.basePrice = basePrice; this.returnPrice = returnPrice; }
然后在3个子类中, 调用这个构造方法
public NormalCashFee(double discount, double basePrice, double returnPrice){ };
public DiscountCashFee(double discount, double basePrice, double returnPrice){ super(discount/10, 0, 0); }
public ReturnCashFee(double discount, double basePrice, double returnPrice){ super(0, basePrice, returnPrice); }
可以明显看出, NormalCashFee类中的这个带三个参数的构造方法, 其内部根本没有任何操作.
而DiscountCashFee只想传递第一个参数.
而ReturnCashFee只想传递第二个和第三个参数.
可是为了能够使用反射. 他们必须传递本身所不必须的参数. 这一点感觉还是不好, 应该优化. 可是如何优化呢? 尚未可知. 有知道的朋友, 请给我留言.