定义:
提供一个创建一系列相关或相互依赖对象的接口,而无需指定他们具体的类。
结构:(书中图,侵删)
这个图相对来说有一点点复杂,其实就是在工厂方法模式的基础上做了一些扩展,工厂方法模式只用于生成一种产品(把上图ProductB相关的都去掉就是了),而抽象工厂模式可用于生产多种产品。
加上例子吧,假设生产海贼的手办(路飞和娜美)。
一个抽象工厂抽象接口(包含生成所有类型产品的方法,即生成路飞和娜美的方法)
若干个具体工厂(各种生成产品的不同实现的工厂,理论上,同一个具体工厂底下生成的都是同一个系列的产品。类似于A工厂生成两年前的,B工厂生成两年后的,生成出来的都是同一个人物)
若干个抽象的产品接口(这里就是路飞和娜美两个)
每个抽象的产品接口下有若干个具体的产品类(路飞下有(两年前路飞、两年后路飞);娜美下有(两年前娜美,两年后娜美))
根据上例照着原格式再来画张图,便于理解:(把client去掉了,懒得画)
实例:
鉴于书中的例子相当的常见,所以决定延用书中的例子。
就是更换数据库的例子。
假设系统中有员工、部门两个类。
然后系统需要使用mysql和oracle两个数据库。
为了代码简洁,不分什么dao层之类的,直接把调用数据库的方法写在实体里。
员工抽象类:
package designpattern.abstractfactory; public abstract class Employee {
private String name; abstract void insert(Employee employee); public String getName() {
return name;
} public void setName(String name) {
this.name = name;
} @Override
public String toString() {
return "Employee [name=" + name + "]";
} }
oracle员工类:
package designpattern.abstractfactory; public class OracleEmployee extends Employee { @Override
void insert(Employee employee) {
System.out.println("往oracle数据库插入一条Employee员工数据:" + employee);
} }
mysql员工类:
package designpattern.abstractfactory; public class MysqlEmployee extends Employee {
@Override
public void insert(Employee employee) {
System.out.println("往mysql数据库插入一条Employee员工数据:" + employee);
} }
部门抽象类:
package designpattern.abstractfactory; public abstract class Department {
String name; abstract void insert(Department department); public String getName() {
return name;
} public void setName(String name) {
this.name = name;
} @Override
public String toString() {
return "Department [name=" + name + "]";
} }
oracle部门类:
package designpattern.abstractfactory; public class OracleDepartment extends Department { @Override
void insert(Department department) {
System.out.println("往oracle数据库插入一条Department部门数据:" + department);
} }
mysql部门类:
package designpattern.abstractfactory; public class MysqlDepartment extends Department { @Override
void insert(Department department) {
System.out.println("往mysql数据库插入一条Department部门数据:"+department);
} }
抽象工厂类:
package designpattern.abstractfactory; public interface Factory {
Employee createEmployee(); Department createDepartment();
}
mysql工厂类:
package designpattern.abstractfactory; public class MysqlFactory implements Factory { @Override
public Employee createEmployee() {
return new MysqlEmployee();
} @Override
public Department createDepartment() {
return new MysqlDepartment();
} }
oracle工厂类:
package designpattern.abstractfactory; public class OracleFactory implements Factory { @Override
public Employee createEmployee() {
return new OracleEmployee();
} @Override
public Department createDepartment() {
return new OracleDepartment();
} }
客户端:
package designpattern.abstractfactory; public class Client {
public static void main(String[] args) {
Factory factory = new MysqlFactory();
// Factory factory=new OracleFactory(); Employee employee = factory.createEmployee();
employee.setName("张三");
employee.insert(employee); Department department = factory.createDepartment();
department.setName("技术部");
department.insert(department); }
}
结果输出:
往mysql数据库插入一条Employee员工数据:Employee [name=张三]
往mysql数据库插入一条Department部门数据:Department [name=技术部]
这个设计模式很好的解除了客户端与实例创建过程的耦合,通过抽象出接口的方式,使客户端只需要和接口打交道。
同时也使得切换数据库变得容易,只需要修改初始化的语句即可。
这同样也是这个模式的不足之处,意味着所有需要用到数据库连接的地方都要写上这句初始化语句,使得修改的工作量变得很大。
接下来就一步一步优化它:
首先,使用哪个数据库的判断是在客户端,我们需要把这个判断转移,使用简单工厂模式,将判断转移至简单工厂:
简单工厂:
package designpattern.abstractfactory; public class SimpleFactory {
static String db = "mysql";
//static String db="oracle"; static Employee createEmployee() {
switch (db) {
case "mysql":
return new MysqlEmployee();
case "oracle":
return new OracleEmployee();
default:
return null;
}
} static Department createDepartment() {
switch (db) {
case "mysql":
return new MysqlDepartment();
case "oracle":
return new OracleDepartment();
default:
return null;
}
} }
客户端:
package designpattern.abstractfactory; public class Client2 {
public static void main(String[] args) {
Employee employee = SimpleFactory.createEmployee();
employee.setName("张三");
employee.insert(employee); Department department = SimpleFactory.createDepartment();
department.setName("技术部");
department.insert(department); }
}
然后,如果再增加一个数据库,需要在所有的方法里增加switch的case,这也是很麻烦的事情,这里需要用到反射来解决这个问题:
反射版简单工厂:
package designpattern.abstractfactory; public class ReflectSimpleFactory {
static String db = "Mysql";
// static String db="Oracle"; static String path = "designpattern.abstractfactory";// 包路径 static Employee createEmployee() {
try {
Class<Employee> employee = (Class<Employee>) Class.forName(path + "." + db + "Employee");
return employee.newInstance();
} catch (ClassNotFoundException e) {
e.printStackTrace();
} catch (InstantiationException e) {
e.printStackTrace();
} catch (IllegalAccessException e) {
e.printStackTrace();
}
return null;
} static Department createDepartment() {
try {
Class<Department> department = (Class<Department>) Class.forName(path + "." + db + "Department");
return department.newInstance();
} catch (ClassNotFoundException e) {
e.printStackTrace();
} catch (InstantiationException e) {
e.printStackTrace();
} catch (IllegalAccessException e) {
e.printStackTrace();
}
return null;
} }
客户端:
package designpattern.abstractfactory; public class Client3 {
public static void main(String[] args) {
Employee employee = ReflectSimpleFactory.createEmployee();
employee.setName("张三");
employee.insert(employee); Department department = ReflectSimpleFactory.createDepartment();
department.setName("技术部");
department.insert(department); }
}
通过反射,将程序由编译时改为运行时,彻底取代了switch语句。
现在,还剩最后一个问题,决定使用什么数据库的字符串还是写在代码中,修改之后还需要重新编译,这里只需要把字符串改到配置文件中即可:
简单工厂:
package designpattern.abstractfactory; import java.io.IOException;
import java.io.InputStream;
import java.util.Properties; public class ReflectSimpleFactory2 { static String path = "designpattern.abstractfactory";// 包路径 static Employee createEmployee() {
try {
Class<Employee> employee = (Class<Employee>) Class.forName(path + "." + getDBName() + "Employee");
return employee.newInstance();
} catch (ClassNotFoundException e) {
e.printStackTrace();
} catch (InstantiationException e) {
e.printStackTrace();
} catch (IllegalAccessException e) {
e.printStackTrace();
}
return null;
} static Department createDepartment() {
try {
Class<Department> department = (Class<Department>) Class.forName(path + "." + getDBName() + "Department");
return department.newInstance();
} catch (ClassNotFoundException e) {
e.printStackTrace();
} catch (InstantiationException e) {
e.printStackTrace();
} catch (IllegalAccessException e) {
e.printStackTrace();
}
return null;
} private static String getDBName() {
String dbName = null;
try {
InputStream in = ReflectSimpleFactory2.class.getResourceAsStream("db.properties");
Properties pro = new Properties();
pro.load(in);
in.close();
dbName = pro.getProperty("db");
} catch (IOException e) {
e.printStackTrace();
} return dbName; }
}
配置文件:
db=Mysql
#db=Oracle
客户端:
package designpattern.abstractfactory; public class Client4 {
public static void main(String[] args) {
Employee employee = ReflectSimpleFactory2.createEmployee();
employee.setName("张三");
employee.insert(employee); Department department = ReflectSimpleFactory2.createDepartment();
department.setName("技术部");
department.insert(department); }
}
大功告成!
总结:
抽象工厂设计模式和其他的工厂类设计模式一样,就是将客户端与具体的对象创建过程分离。
只不过这里所涉及到的不再是一种类,而是多种类,结构相对复杂。
同时也像上文说的一样,存在一些不足,可以具体情况具体分析,应该如何使用。