上一期介绍了Spring Data JPA的各种查询方法，这一期来详细讲一下动态查询类，JpaSpecificationExecutor.
JpaSpecificationExecutor接口
JpaSpecificationExecutor提供了findOne、findAll、count方法的几个重载：
-
Optional<T> findOne(Specification<T> spec); 根据规范查找一个结果
-
List<T> findAll(Specification<T> spec); 根据规范查找所有结果
-
Page<T> findAll(Specification<T> spec, Pageable pageable); 根据规范和分页方式查找结果
-
List<T> findAll(Specification<T> spec, Sort sort); 根据规范和排序方式查找结果
-
long count(Specification<T> spec); 根据规范统计个数
上一期我们提到了PagingAndSortingRepository接口，这个接口提供了排序和分页查找的方法，然而它却不支持条件查询，JpaSpecificationExecutor接口解决了这个问题。
初见Specification类
Specification类用于确定规范，那么这个规范要如何确定呢？接下来我们细看一下这个类：
Specification需要提供一个泛型确定实体类，类中有一个toPredicate方法需要我们重载，而toPerdicate方法拥有3个参数：
-
root 用于获取类属性
-
criteriaQuery 添加条件(不常用)
-
criteriaBuilder 添加条件
举例：
@Test
@Transactional
public void TestFind() {
Specification<User> spec = new Specification<User>() {
@Override
public Predicate toPredicate(Root<User> root, CriteriaQuery<?> criteriaQuery, CriteriaBuilder criteriaBuilder) {
Path<Object> name = root.get("name");
return criteriaBuilder.equal(name, "冬马和纱");
}
};
List<User> userList = userDao.findAll(spec);
for (User user : userList) {
System.out.println(user);
}
}
这里笔者使用root的get方法获取一个类属性，注意返回值使用Path<Object>接收；接着使用criteriaBuilder添加equal条件：
-
criteriaBuilder.equal(name, "冬马和纱") 即 name = "冬马和纱"
返回结果：
Hibernate: select user0_.u_id as u_id1_1_, user0_.u_address_id as u_addres3_1_, user0_.u_name as u_name2_1_ from user user0_ where user0_.u_name=?
Hibernate: select address0_.a_id as a_id1_0_0_, address0_.a_name as a_name2_0_0_ from address address0_ where address0_.a_id=?
ID: 1, Name: 冬马和纱, Address: 北京
Process finished with exit code 0
当然，在Java 8及以上版本，我们可以使用Lambda表达式使代码更美观：
Specification<User> spec = (Specification<User>) (root, criteriaQuery, criteriaBuilder) -> {
Path<Object> name = root.get("name");
return criteriaBuilder.equal(name, "冬马和纱");
};
Specification的更多功能
Specification的条件
对于toPredicate方法中的criteriaBuilder参数，除了equal判等之外，还有更多强大的功能：
-
equal / notEqual 判等
-
like / notLike 模糊查询
-
gt / ge / lt / le 大于、大于等于、小于、小于等于
-
between 范围查询
需要注意的是，除了判等方法可以直接传递参数之外，其他方法都需要使用as指定数据类型，如：
@Test
@Transactional
public void TestFind() {
Specification<User> spec = (Specification<User>) (root, criteriaQuery, criteriaBuilder) -> {
Path<Object> name = root.get("name");
return criteriaBuilder.like(name.as(String.class), "冬马%");
};
List<User> userList = userDao.findAll(spec);
for (User user : userList) {
System.out.println(user);
}
}
返回结果：
Hibernate: select user0_.u_id as u_id1_1_, user0_.u_address_id as u_addres3_1_, user0_.u_name as u_name2_1_ from user user0_ where user0_.u_name like ?
Hibernate: select address0_.a_id as a_id1_0_0_, address0_.a_name as a_name2_0_0_ from address address0_ where address0_.a_id=?
ID: 1, Name: 冬马和纱, Address: 北京
Process finished with exit code 0
Specification多条件查询
toPredicate的criteriaBuilder参数提供了条件处理的方法：and / or / not.
很好理解，不多做解释。
举例：
@Test
@Transactional
public void TestFind() {
Specification<User> spec = (Specification<User>) (root, criteriaQuery, criteriaBuilder) -> {
Path<Object> id = root.get("id");
Path<Object> address = root.get("address");
Predicate p1 = criteriaBuilder.ge(id.as(Integer.class), 2);
Predicate p2 = criteriaBuilder.equal(address, 10001);
return criteriaBuilder.and(p1, p2);
};
List<User> userList = userDao.findAll(spec);
for (User user : userList) {
System.out.println(user);
}
}
返回结果：
Hibernate: select user0_.u_id as u_id1_1_, user0_.u_address_id as u_addres3_1_, user0_.u_name as u_name2_1_ from user user0_ where cast(user0_.u_id as signed)>=2 and user0_.u_address_id=10001
Hibernate: select address0_.a_id as a_id1_0_0_, address0_.a_name as a_name2_0_0_ from address address0_ where address0_.a_id=?
ID: 3, Name: 鹿目圆香, Address: 北京
ID: 5, Name: 雪之下雪乃, Address: 北京
Process finished with exit code 0
排序和分页
在旧版本中，Sort和Pageable可以通过构造方法设定参数，然而在2.2版本之后，构造方法不再被支持。
使用Sort
新版本中使用Sort.by设定参数，第一个参数表示排序方向，第二个参数表示排序对象。
举例：
@Test
@Transactional
public void TestFind() {
Specification<User> spec = (Specification<User>) (root, criteriaQuery, criteriaBuilder) -> {
Path<Object> address = root.get("address");
return criteriaBuilder.equal(address, 10001);
};
List<User> userList = userDao.findAll(spec, Sort.by(Sort.Direction.DESC, "id"));
for (User user : userList) {
System.out.println(user);
}
}
返回结果：
Hibernate: select user0_.u_id as u_id1_1_, user0_.u_address_id as u_addres3_1_, user0_.u_name as u_name2_1_ from user user0_ where user0_.u_address_id=10001 order by user0_.u_id desc
Hibernate: select address0_.a_id as a_id1_0_0_, address0_.a_name as a_name2_0_0_ from address address0_ where address0_.a_id=?
ID: 5, Name: 雪之下雪乃, Address: 北京
ID: 3, Name: 鹿目圆香, Address: 北京
ID: 1, Name: 冬马和纱, Address: 北京
Process finished with exit code 0
如果要进行多条件排序查询，可以使用Sort.and(Sort)，本文不再演示。
使用PageRequest
PageRequest.of()方法返回一个Pageable对象，需要提供两个参数，第一个参数表示开始页(从第零页开始)，第二个参数表示每页的数据量。
findAll(Spec, Pageable)方法返回一个Page对象，这个对象有以下常用方法：
-
getTotalElements 返回元素总数
-
getTotalPages 返回总页数
-
getContent 以List形式返回当前页面的元素
举例：
@Test
@Transactional
public void TestFind() {
Specification<User> spec = (Specification<User>) (root, criteriaQuery, criteriaBuilder) -> {
Path<Object> id = root.get("id");
return criteriaBuilder.gt(id.as(Integer.class), 1);
};
Page<User> userPage = userDao.findAll(spec, PageRequest.of(1, 2));
List<User> userList = userPage.getContent();
for (User user : userList) {
System.out.println(user);
}
}
返回结果：
Hibernate: select user0_.u_id as u_id1_1_, user0_.u_address_id as u_addres3_1_, user0_.u_name as u_name2_1_ from user user0_ where cast(user0_.u_id as signed)>1 limit ?, ?
Hibernate: select address0_.a_id as a_id1_0_0_, address0_.a_name as a_name2_0_0_ from address address0_ where address0_.a_id=?
Hibernate: select address0_.a_id as a_id1_0_0_, address0_.a_name as a_name2_0_0_ from address address0_ where address0_.a_id=?
Hibernate: select count(user0_.u_id) as col_0_0_ from user user0_ where cast(user0_.u_id as signed)>1
ID: 5, Name: 雪之下雪乃, Address: 北京
ID: 10, Name: 由比滨结衣, Address: 深圳
Process finished with exit code 0
