0.0 概述
JPA 2 引入了criteria API
来构建查询语句中的where
条件。这些查询条件用CriteriaBuilder
来构建,构建的每个条件都是Predicate
对象。
Specification
类实现了该API。通过继承JpaSpecificationExecutor
接口,就可以使用Specification
进行查询。
1.0 适用范围
- 动态构建查询语句
- 支持所有的查询条件
- 支持子查询、连接查询、排序、分页
-
不支持自定义模型,无法做到将查询的结果封装为自定义的model,或类似
List<String>
的单列结果。只能返回dao定义的entity。
即,只支持select * from ...
1.1 JpaSpecificationExecutor
接口方法介绍
public interface JpaSpecificationExecutor<T> {
// 查询单个对象
Optional<T> findOne(@Nullable Specification<T> spec);
// 查询列表
List<T> findAll(@Nullable Specification<T> spec);
// 查询全部,分页;可以使用PageRequest#of(int, int, Sort)方法创建pageable对象实现分页+排序
Page<T> findAll(@Nullable Specification<T> spec, Pageable pageable);
// 查询全部,排序
List<T> findAll(@Nullable Specification<T> spec, Sort sort);
// 统计查询
long count(@Nullable Specification<T> spec);
}
1.2 Specification
类
Specification
类是一个用于构建动态查询条件的抽象类。
public interface Specification<T> {
Predicate toPredicate(Root<T> root, CriteriaQuery<?> query, CriteriaBuilder builder);
}
1.2.1 toPredicate
方法介绍
Specification
类通过实现抽象方法Predicate toPredicate(Root<T> root, CriteriaQuery<?> query, CriteriaBuilder criteriaBuilder)
来构建动态查询条件,查询条件最终会生成一个查询条件对象(Predicate
对象)
-
Root
类:用于获取需要查询的列- 如:
root.get("lastName")
- 如:
-
CriteriaQuery
类:用于自定义查询方式- 如:
query.distinct(true).select(...).where(...).groupBy(...).having(...).getRestriction()
。最后通过getRestriction()
方法,可以获得一个Predicate
对象 - 详见
- 如:
-
CriteriaBuilder
类:构造查询条件(即,用于构建Predicate
对象),内部封装了很多的查询条件- 如:
criteriaBuilder.and(...)
、criteriaBuilder.equal(...)
、criteriaBuilder.greaterThan)
、criteriaBuilder.greaterThanOrEqualTo)
等,返回的结果是一个Predicate
对象 - 详见
- 如:
1.2.2 继承关系
Root
类、CriteriaQuery
类、CriteriaBuilder
类都在javax.persistence.criteria
包下,其继承关系如下:
1.3 示例
1.3.1 简单示例
(1) 编写查询条件
@Test
public void testSimpleCbSpecification() {
Specification<RetPrdInfo> specification = (root, query, cb) -> {
// 等于
Predicate predicate1 = cb.equal(root.get("crToolRunMode"), "CLN");
// 不等于
Predicate predicate2 = cb.notEqual(root.get("prdSeqId"), "L000001");
// 大于
Predicate predicate3 = cb.greaterThan(root.get("prdQty"), 1);
// 大于等于
Predicate predicate4 = cb.greaterThanOrEqualTo(root.get("prdStdQty"), 1);
// 小于
Predicate predicate5 = cb.lessThan(root.get("prdQty"), 2);
// 小于等于
Predicate predicate6 = cb.lessThanOrEqualTo(root.get("prdStdQty"), 2);
// between
Timestamp startTime = Timestamp.valueOf(LocalDateTime.of(2021, 1, 1, 8, 0, 0));
Timestamp endTime = Timestamp.valueOf(LocalDateTime.of(2022, 1, 1, 8, 0, 0));
Predicate predicate7 = cb.between(root.get("evtTimestamp"), startTime, endTime);
// 非空
Predicate predicate8 = cb.isNotNull(root.get("prty"));
// 为空
Predicate predicate9 = cb.isNull(root.get("costCode"));
// 模糊查询
Predicate predicate10 = cb.like(root.get("tempPrdSeqId"), "L%");
// in
Predicate predicate11 = root.get("evtUsr").in("AA", "BB");
Predicate[] predicateArray = {predicate1, predicate2, predicate3, predicate4, predicate5,
predicate6, predicate7, predicate8, predicate9, predicate10, predicate11};
return query.where(predicateArray).getRestriction();
};
List<RetPrdInfo> retPrdInfoList = retPrdInfoDao.findAll(specification);
retPrdInfoList.stream().forEach(System.out::println);
}
生成的sql语句:
select 略
from ret_prd_info
where cr_tool_run_mode = ?
and prd_seq_id <> ?
and prd_qty > 1
and prd_std_qty >= 1
and prd_qty < 2
and prd_qty <= 2
and (evt_timestamp between ? and ?)
and (prty is not null)
and (cost_code is null)
and (temp_prd_seq_id like ?)
and (retprdinfo0_.evt_usr in (?, ?));
(2) 查询条件的连接
1️⃣ 写法一
@Test
public void testConjunctionSpecification1() {
Specification<RetPrdInfo> specification = (root, query, cb) -> {
Predicate predicate1 = cb.greaterThan(root.get("prdQty"), 1);
Predicate predicate2 = cb.lessThan(root.get("prdQty"), 2);
Predicate orPredicate1 = cb.or(predicate1, predicate2); // 使用or连接两个条件
Predicate predicate3 = cb.greaterThanOrEqualTo(root.get("prdStdQty"), 1);
Predicate predicate4 = cb.lessThanOrEqualTo(root.get("prdStdQty"), 2);
Predicate orPredicate2 = cb.or(predicate3, predicate4); // 使用or连接两个条件
Predicate andPredicate = cb.and(orPredicate1, orPredicate2);
return query.where(andPredicate).getRestriction(); // 使用and连接两个条件
};
List<RetPrdInfo> retPrdInfoList = retPrdInfoDao.findAll(specification);
retPrdInfoList.stream().forEach(System.out::println);
}
生成的sql语句:
select 略
from ret_prd_info
where prd_qty > 1 and prd_qty < 2
or prd_std_qty >= 1 and prd_std_qty <= 2
where 后面如果有and、or的条件,则or自动会把左右的查询条件分开,即先执行and,再执行or。
原因就是:and的执行优先级最高。 关系型运算符优先级高到低为:not and or
2️⃣ 写法二
@Test
public void testConjunctionSpecification2() {
Specification<RetPrdInfo> specification = (root, query, cb) -> {
Predicate predicate1 = cb.greaterThan(root.get("prdQty"), 1);
Predicate predicate2 = cb.lessThan(root.get("prdQty"), 2);
Predicate[] orArray1 = {predicate1, predicate2};
Predicate orPredicate1 = cb.and(orArray1); // 用or连接数组中的每个条件
Predicate predicate3 = cb.greaterThanOrEqualTo(root.get("prdStdQty"), 1);
Predicate predicate4 = cb.lessThanOrEqualTo(root.get("prdStdQty"), 2);
Predicate[] orArray2 = {predicate3, predicate4};
Predicate orPredicate2 = cb.or(orArray2); // 用or连接数组中的每个条件
Predicate[] andArray = {orPredicate1, orPredicate2};
Predicate andPredicate = cb.and(andArray); // 用and连接数组中的每个条件
return query.where(andPredicate).getRestriction();
};
List<RetPrdInfo> retPrdInfoList = retPrdInfoDao.findAll(specification);
retPrdInfoList.stream().forEach(System.out::println);
}
3️⃣ 写法三
@Test{
public void testConjunctionSpecification3() {
Specification<RetPrdInfo> specification = (root, query, cb) -> {
Predicate predicate1 = cb.greaterThan(root.get("prdQty"), 1);
Predicate predicate2 = cb.lessThan(root.get("prdQty"), 2);
Predicate[] predicateArray = {predicate1, predicate2};
return query.where(predicateArray).getRestriction(); // where中放入predicate数组,使用and连接数组中的每个条件
};
List<RetPrdInfo> retPrdInfoList = retPrdInfoDao.findAll(specification);
retPrdInfoList.stream().forEach(System.out::println);
}}
生成的sql:
select 略
from ret_prd_info
where prd_qty > 1
and prd_qty < 2
(3) 对查询结果排序
@Test
public void testOrder() {
Specification<Customer> spec = (root, query, cb) -> cb.equal(root.get("crToolRunMode"), "CLN");
Sort sort = Sort.by(Sort.Direction.DESC, "id");
List<Customer> list = customerDao.findAll(spec, sort);
}
(4) 分页查询
@Test
public void testPagging() {
Specification<Customer> spec = (root, query, cb) -> cb.equal(root.get("crToolRunMode"), "CLN");
Pageable pageable = PageRequest.of(0, 20);
// 可以使用`Pageable pageable = PageRequest.of(0, 20, Sort.by("evtTimestamp"));`实现分页+排序
Page<Customer> page = customerDao.findAll(spec, pageable);
List<Customer> list = page.getContent(); // 该页的内容
long totalElements = page.getTotalElements(); // 全表的记录数目
}
1.3.2 多条件查询示例(and、equal、like、sort、page)
每一个Predicate就是一个条件
public void queryByCondition(FbpbisdatI inTrx, FbpbisdatO outTrx) {
FbpbisdatIA iary = inTrx.getIary().get(0);
Specification<BisData> specification = (root, query, cb) -> {
List<Predicate> predicateList = new ArrayList<>();
if (StringUtils.isNotBlank(iary.getDataCate())) {
predicateList.add(cb.equal(root.get("dataCate"), iary.getDataCate()));
}
if (StringUtils.isNotBlank(iary.getDataSeqId())) {
predicateList.add(cb.like(root.get("dataSeqId").as(String.class), "%" + ia.getMdlId() + "%"));
// `"%" + ia.getMdlId() + "%"` 可以改写为 `Util.liktStr(ia.getmdlId())`
}
// 略
Predicate[] p = new Predicate[predicateList.size()];
query.where(predicateList.toArray(p));
return query.getRestriction();
};
Sort sort = Sort.by("dataCate", "dataId", "dataExt");
List<BisData> bisDataList;
// 判断是否需要分页
if (Util.shouldPage(inTrx)) {
Page<BisData> bisDataPage = bisDataDao.findAll(specification, PageRequest.of(inTrx.getPageNum() - 1, inTrx.getPageSize(), sort));
// `PageRequest.of(inTrx.getPageNum() - 1, inTrx.getPageSize(), sort)`
// 可以改写为 `Util.getPageable(inTrx, sort)`
bisDataList = bisDataPage.getContent();
long tblCnt = bisDataPage.getTotalElements();
outTrx.setTblCnt(tblCnt);
} else {
bisDataList = bisDataDao.findAll(specification, sort);
}
List<FbpbisdatOA> oary = CloneUtil.cloneList(bisDataList, FbpbisdatOA::new);
outTrx.setOary(oary);
}
1.3.3 子查询示例
public void testSubSpecification() {
Specification<RetPrdInfo> specification = (root, query, cb) -> {
// 1).父查询条件
Predicate parentPredicate = cb.equal(root.get("crToolRunMode"), "CLN");
// 2).子查询
Subquery<RetPrdAbn> subQuery = query.subquery(RetPrdAbn.class);
Root<RetPrdAbn> subRoot = subQuery.from(RetPrdAbn.class);
subQuery = subQuery.select(subRoot.get("prdSeqIdFk"));
// 子查询的cb和父查询的cb相同
Predicate subPredicate1 = cb.equal(subRoot.get("lotIdFk"), "LW0001");
Predicate subPredicate2 = cb.equal(subRoot.get("lotSpltIdFk"), "00");
subQuery.where(cb.and(subPredicate1, subPredicate2));
// 3).将父查询条件和子查询合并
Predicate parentAndSubConjunctPredicate = root.get("prdSeqId").in(subQuery);
return query.where(parentPredicate, parentAndSubConjunctPredicate).getRestriction();
};
List<RetPrdInfo> retPrdInfoList = retPrdInfoDao.findAll(specification);
retPrdInfoList.stream().forEach(System.out::println);
}
生成的sql语句:
select
略
from
ret_prd_info
where
cr_tool_run_mode = ?
and (prd_seq_id in
(
select prd_seq_id_fk
from ret_prd_abn
where lot_id_fk = ? and retprdabn1_.lot_splt_id_fk = ?
)
)