Hibernate多对多

///Hibernate:///多对多关联是Hibernate中一种比较特殊的关联，它需要借助中间表来完成多对多信息的保存。多对多关联只有双向关联。对于一个订单（Order）可以包含多种产品（Products），而对于每种产品可以存在于多个订单之中，是一个典型的多对多关系。本次演示用到的两个POJO类如下：public class Order{private Integer id;private String name;private String num;private Set<Product> products = new HashSet<Product>();//......}public class Product{private Integer id;private String name;private Set<Order> orders = new HashSet<Order>();//......}它们对应的映射文件分别为:Order.hbm.xml<hibernate-mapping package="org.louis.domain"><class name="Order" table="TEST_ORDER"><id name="id" column="ID"><generator class="native"></generator></id><property name="name" column="NAME"></property><property name="num" column="NUM"></property><set name="products" table="TEST_ORDER_PRODUCT" cascade="save-update" fetch="join"><key column="ORDER_ID"></key><many-to-many class="Product" column="PRODUCT_ID"></many-to-many></set></class></hibernate-mapping>   Product.hbm.xml<hibernate-mapping package="org.louis.domain"><class name="Product" table="TEST_PRODUCT"><id name="id" column="ID"><generator class="native"></generator></id><property name="name" column="NAME"></property><set name="orders" table="TEST_ORDER_PRODUCT" cascade="save-update" fetch="join"><key column="PRODUCT_ID"></key><many-to-many class="Order" column="ORDER_ID"></many-to-many></set></class></hibernate-mapping>  我们注意到，在两个文件中都有<set>节点，同时都指定了其table属性，这是两个对象的的中间表，在多对多关联中，很多操作都是通过的中间表来完成的；<set>元素的子元素<key>指定了中间表中的外键，它对应于当前文件的对象的主键，如Order.hbm.xml中，对应于Order的主键；<many-to-many>指定了关联的另一方和中间表的对应于另一方主键的外键。下面来做一些测试。a、插入数据public void insert(){Session session = HibernateSessionFactory.getSessionFactory().getCurrentSession();session.beginTransaction();Product p1 = new Product();p1.setName("p1");Product p2 = new Product();p2.setName("p2");Order order = (Order)session.load(Order.class, 3);//      order.getProducts().add(p1);//      order.getProducts().add(p2);p1.getOrders().add(order);p2.getOrders().add(order);session.save(p1);session.save(p2);session.getTransaction().commit();}查看Hibernate在后台执行的SQL语句如下：Hibernate: selectorder0_.ID as ID6_1_,order0_.NAME as NAME6_1_,order0_.NUM as NUM6_1_,order0_.MEMBER_ID as MEMBER4_6_1_,products1_.ORDER_ID as ORDER1_3_,product2_.ID as PRODUCT2_3_,product2_.ID as ID8_0_,product2_.NAME as NAME8_0_ fromTEST_ORDER order0_ left outer joinTEST_ORDER_PRODUCT products1_ on order0_.ID=products1_.ORDER_ID left outer joinTEST_PRODUCT product2_ on products1_.PRODUCT_ID=product2_.ID whereorder0_.ID=?Hibernate: insert intoTEST_PRODUCT(NAME) values(?)Hibernate: insert intoTEST_PRODUCT(NAME) values(?)Hibernate: insert intoTEST_ORDER_PRODUCT(PRODUCT_ID, ORDER_ID) values(?, ?)Hibernate: insert intoTEST_ORDER_PRODUCT(PRODUCT_ID, ORDER_ID) values(?, ?)第一条是更加ID加载Order对象，同时将其关联对象也加载了。接着插入两个Product，然后用两条SQL语句向中间表里面插入反映关联关系的数据。如果将Java代码修改为下面这样：public void insert(){Session session = HibernateSessionFactory.getSessionFactory().getCurrentSession();session.beginTransaction();Product p1 = new Product();p1.setName("p1");Product p2 = new Product();p2.setName("p2");Order order = (Order)session.load(Order.class, 3);order.getProducts().add(p1);order.getProducts().add(p2);session.save(order);session.getTransaction().commit();}我们发现Hibernate执行的SQL和上面的一样，由此看来，在这种多对多情况下，无论让谁来维护关系都是可以的，因为默认情况下inverse="false";我们再来看当设置Product.hbm.xml中的<set>节点属性inverse="true"， 这表示对于两个对象的关联关系的维护有Order来做。我们使用下面的代码来测试一下：public void insert(){Session session = HibernateSessionFactory.getSessionFactory().getCurrentSession();session.beginTransaction();Product p1 = new Product();p1.setName("p1");Product p2 = new Product();p2.setName("p2");Order order = (Order)session.load(Order.class, 3);//      order.getProducts().add(p1);//      order.getProducts().add(p2);p1.getOrders().add(order);p2.getOrders().add(order);session.save(p1);session.save(p2);session.getTransaction().commit();}在这段代码中，明显可以看出关联关系的维护交给了Product来做了，这个Product.hbm.xml中配置的方式相矛盾，因此必然会在保存数据的时候出现问题。运行这段代码，查看后台HIbernate使用的SQL语句为：Hibernate: selectorder0_.ID as ID6_1_,order0_.NAME as NAME6_1_,order0_.NUM as NUM6_1_,order0_.MEMBER_ID as MEMBER4_6_1_,products1_.ORDER_ID as ORDER1_3_,product2_.ID as PRODUCT2_3_,product2_.ID as ID8_0_,product2_.NAME as NAME8_0_ fromTEST_ORDER order0_ left outer joinTEST_ORDER_PRODUCT products1_ on order0_.ID=products1_.ORDER_ID left outer joinTEST_PRODUCT product2_ on products1_.PRODUCT_ID=product2_.ID whereorder0_.ID=?Hibernate: insert intoTEST_PRODUCT(NAME) values(?)Hibernate: insert intoTEST_PRODUCT(NAME) values(?)从这段SQL代码可以看出，只是进行了Product数据的插入，并没有对反映Order和Product关系的中间表进行数据插入，看来配置文件中的设定由Order来维护关联关系起了一定的作用（为什么是一定的作用而不是完全的呢？这是因为还没有进行必要的测试，只是在这里反映出Product已经不能维护关联关系了）。下面我们将Java代码改动一下如下：public void insert(){Session session = HibernateSessionFactory.getSessionFactory().getCurrentSession();session.beginTransaction();Product p1 = new Product();p1.setName("p1");Product p2 = new Product();p2.setName("p2");Order order = (Order)session.load(Order.class, 3);order.getProducts().add(p1);order.getProducts().add(p2);session.save(order);session.getTransaction().commit();}再次运行，查看后台SQL语句如下：Hibernate: selectorder0_.ID as ID6_1_,order0_.NAME as NAME6_1_,order0_.NUM as NUM6_1_,order0_.MEMBER_ID as MEMBER4_6_1_,products1_.ORDER_ID as ORDER1_3_,product2_.ID as PRODUCT2_3_,product2_.ID as ID8_0_,product2_.NAME as NAME8_0_ fromTEST_ORDER order0_ left outer joinTEST_ORDER_PRODUCT products1_ on order0_.ID=products1_.ORDER_ID left outer joinTEST_PRODUCT product2_ on products1_.PRODUCT_ID=product2_.ID whereorder0_.ID=?Hibernate: insert intoTEST_PRODUCT(NAME) values(?)Hibernate: insert intoTEST_PRODUCT(NAME) values(?)Hibernate: insert intoTEST_ORDER_PRODUCT(ORDER_ID, PRODUCT_ID) values(?, ?)Hibernate: insert intoTEST_ORDER_PRODUCT(ORDER_ID, PRODUCT_ID) values(?, ?)显然，和没有使用任何inverse="true"时的情况一摸一样，是正确的，也就是说现在是由Order来维护两个关联对象的关联关系的。对于多对多关联来说，可以在任何一端来设置为主控方，也可都不设，那么两端都可以维护关联关系（并且，我们发现和设定一方为主控方在性能上没有什么变化）。如果两端都设定为inverse="true"的话呢？我们来做个测试，将两个映射文件的<set>节点的inverse属性都设置为true。对于Order来说，是让Product作为主控方；对于Product来说让Order作为主控方。先用下面代码测试：public void insert(){Session session = HibernateSessionFactory.getSessionFactory().getCurrentSession();session.beginTransaction();Product p1 = new Product();p1.setName("p1");Product p2 = new Product();p2.setName("p2");Order order = (Order)session.load(Order.class, 3);//  order.getProducts().add(p1);//  order.getProducts().add(p2);p1.getOrders().add(order);p2.getOrders().add(order);session.save(p1);session.save(p2);session.getTransaction().commit();}查看后台SQL语句如下：Hibernate: selectorder0_.ID as ID6_1_,order0_.NAME as NAME6_1_,order0_.NUM as NUM6_1_,order0_.MEMBER_ID as MEMBER4_6_1_,products1_.ORDER_ID as ORDER1_3_,product2_.ID as PRODUCT2_3_,product2_.ID as ID8_0_,product2_.NAME as NAME8_0_ fromTEST_ORDER order0_ left outer joinTEST_ORDER_PRODUCT products1_ on order0_.ID=products1_.ORDER_ID left outer joinTEST_PRODUCT product2_ on products1_.PRODUCT_ID=product2_.ID whereorder0_.ID=?Hibernate: insert intoTEST_PRODUCT(NAME) values(?)Hibernate: insert intoTEST_PRODUCT(NAME) values(?)看来没有对中间表进行数据插入。使用下面代码来测试：public void insert(){Session session = HibernateSessionFactory.getSessionFactory().getCurrentSession();session.beginTransaction();Product p1 = new Product();p1.setName("p1");Product p2 = new Product();p2.setName("p2");Order order = (Order)session.load(Order.class, 3);order.getProducts().add(p1);order.getProducts().add(p2);session.save(order);session.getTransaction().commit();}查看后台的SQL语句如下：Hibernate: selectorder0_.ID as ID6_1_,order0_.NAME as NAME6_1_,order0_.NUM as NUM6_1_,order0_.MEMBER_ID as MEMBER4_6_1_,products1_.ORDER_ID as ORDER1_3_,product2_.ID as PRODUCT2_3_,product2_.ID as ID8_0_,product2_.NAME as NAME8_0_ fromTEST_ORDER order0_ left outer joinTEST_ORDER_PRODUCT products1_ on order0_.ID=products1_.ORDER_ID left outer joinTEST_PRODUCT product2_ on products1_.PRODUCT_ID=product2_.ID whereorder0_.ID=?Hibernate: insert intoTEST_PRODUCT(NAME) values(?)Hibernate: insert intoTEST_PRODUCT(NAME) values(?)比较两次的结果，发现一摸一样。看来如果两端都设置为主控方，是不能对中间表进行数据保存的。b、数据加载数据加载就不再演示了，也就是默认采用左外连接的方式加载数据，能够将Order和它的关联对象Product都加载，当然是通过中间表；对于Product来说也是这样。c、删除数据删除数据也不再演示，无论删除Order还是Product，都是简单的删除自己，而不会删除关联的对象，这是因为设定了cascade="save-update"