我想知道是否有一种通用方法可以将标准api与更复杂的模型结合使用？
我有一个与其他实体有一对一关系的实体类。我的服务包装器通过条件api进行数据库查询，从前端获取参数，找出分页，排序和过滤。
实体
@Entity
public class Person implements Serializable {
@Id
private Long id;
private String name;
private String givenName;
@Temporal(TemporalType.DATE)
private Date birthdate;
@OneToOne(cascade = CascadeType.ALL)
@JoinColumn(name = "INFORMATION_ID")
private Information information;
@OneToOne(cascade = CascadeType.ALL)
@JoinColumn(name = "ADDRESS_ID")
private Address address;
...
}
@Entity
public class Information implements Serializable {
@Id
private Long id;
private String detail;
...
}
@Entity
public class Address implements Serializable {
@Id
private Long id;
private String street;
private String city;
...
}
服务
@Stateless
public class PersonService {
@PersistenceContext(unitName = "ProblemGenericDatatableFilterPU")
private EntityManager em;
public List<Person> findAllPersons222(int first, int pageSize, String sortField, SortOrder sortOrder, Map<String, Object> filters) {
CriteriaBuilder builder = em.getCriteriaBuilder();
CriteriaQuery<Person> criteriaQuery = builder.createQuery(Person.class);
Root<Person> rootPerson = criteriaQuery.from(Person.class);
Join<Person, Information> joinPersonInformation = rootPerson.join(Person_.information);
Join<Person, Address> joinPersonAddress = rootPerson.join(Person_.address);
// select
criteriaQuery.select(rootPerson);
// filter
List<Predicate> allPredicates = new ArrayList<>();
for(Entry<String, Object> currentEntry : filters.entrySet()) {
Predicate currentPredicate;
if(currentEntry.getKey().startsWith("information_")) {
currentPredicate = builder.like(
builder.lower(joinPersonInformation.<String>get(currentEntry.getKey())),
builder.lower(builder.literal(String.valueOf(currentEntry.getValue())))
);
}
else if(currentEntry.getKey().startsWith("address_")) {
currentPredicate = builder.like(
builder.lower(joinPersonAddress.<String>get(currentEntry.getKey())),
builder.lower(builder.literal(String.valueOf(currentEntry.getValue())))
);
}
else {
currentPredicate = builder.like(
builder.lower(rootPerson.<String>get(currentEntry.getKey())),
builder.lower(builder.literal(String.valueOf(currentEntry.getValue())))
);
}
allPredicates.add(currentPredicate);
}
criteriaQuery.where(builder.and(allPredicates.toArray(new Predicate[0])));
// order
if(sortField != null && !sortField.isEmpty()) {
Order orderBy;
if(sortField.startsWith("information_")) {
orderBy = (sortOrder == SortOrder.DESCENDING
? builder.desc(joinPersonInformation.get(sortField))
: builder.asc(joinPersonInformation.get(sortField)));
}
else if(sortField.startsWith("address_")) {
orderBy = (sortOrder == SortOrder.DESCENDING
? builder.desc(joinPersonAddress.get(sortField))
: builder.asc(joinPersonAddress.get(sortField)));
}
else {
orderBy = (sortOrder == SortOrder.DESCENDING
? builder.desc(rootPerson.get(sortField))
: builder.asc(rootPerson.get(sortField)));
}
criteriaQuery.orderBy(orderBy);
}
Query query = em.createQuery(criteriaQuery);
// pagination
query.setFirstResult(first);
query.setMaxResults(pageSize);
return query.getResultList();
}
}
我需要区分过滤和排序的情况，具体取决于我访问该属性的root / join。另外，我需要在facelet中使用命名约定。除了排序之外， count -query也是如此。
现在我问自己是否有一些＆＃34;点符号＆＃34;或任何使案件可有可无的事情。在e。 G。本机SQL我会做一些事情，比如创建子查询并从内部投影中选择所有别名值（select * from (select person.name as name, address.street as street, ...) where name = ... and street like ...）。
我很感激任何建议。
答案 0 :(得分：3)
最后我有时间处理我的问题。我找到了一个不完美的解决方案，但对我有用。
当我搜索另一个问题时，我来到this article by Leonardo Shikida并找到了一个非常方便的Path<?> getPath(...)方法（我还深入研究了CriteriaAPI中的优秀继承关系：Path，Root，Join，来自，等）。考虑到这一点，我重新记住了我以前的问题，并想到了这种方法的更通用的方法。以下是我对此的看法：
首先，我创建了我需要的所有连接（即Root<?>和Join<? ?>），并将它们放在Map<String, From<?, ?>>中，其中String是属性所在的元素以点分表示法（命名约定和完整解决方案的缺点）查询，From是相应的来源。
使用Map我可以用或多或少的方式进行过滤和排序。
为了使其工作，前端需要使用完全相同的命名约定并相应地传递filters-Map（即使用p：列中的primefaces field属性的JSF）。
public List<Person> newFindAllPersons(int first, int pageSize, String sortField, SortOrder sortOrder, Map<String, Object> filters)
{
CriteriaBuilder builder = em.getCriteriaBuilder();
CriteriaQuery<Person> criteriaQuery = builder.createQuery(Person.class);
// setting up the required joins
Root<Person> rootPerson = criteriaQuery.from(Person.class);
Join<Person, Information> joinPersonInformation = rootPerson.join(Person_.information);
Join<Person, Address> joinPersonAddress = rootPerson.join(Person_.address);
Join<Address, Information> joinAddressInformation = joinPersonAddress.join(Address_.information);
// putting all joins into a map with a dot`ted name
Map<String, From<?, ?>> mapFieldToFrom = new HashMap<>();
mapFieldToFrom.put("person", rootPerson);
mapFieldToFrom.put("person.address", joinPersonAddress);
mapFieldToFrom.put("person.information", joinPersonInformation);
mapFieldToFrom.put("person.address.information", joinAddressInformation);
// select
criteriaQuery.select(rootPerson);
// filter
List<Predicate> allPredicates = new ArrayList<>();
for(Entry<String, Object> currentEntry : filters.entrySet())
{
Predicate currentPredicate = builder.like(
builder.lower(getStringPath(currentEntry.getKey(), mapFieldToFrom)),
builder.lower(builder.literal("%" + String.valueOf(currentEntry.getValue()) + "%"))
);
allPredicates.add(currentPredicate);
}
criteriaQuery.where(builder.and(allPredicates.toArray(new Predicate[0])));
// order
if(sortField != null && !sortField.isEmpty())
{
Path<?> actualPath = getStringPath(sortField, mapFieldToFrom);
Order orderBy = (sortOrder == SortOrder.DESCENDING
? builder.desc(actualPath)
: builder.asc(actualPath));
criteriaQuery.orderBy(orderBy);
}
Query query = em.createQuery(criteriaQuery);
// pagination
query.setFirstResult(first);
query.setMaxResults(pageSize);
return query.getResultList();
}
/**
* divides the given field at the last dot and takes <br>
* - the first part as the key in the map to retrieve the From<?, ?> <br>
* - the last part as the name of the column in the entity
*/
private Path<String> getStringPath(String field, Map<String, From<?, ?>> mapFieldToFrom)
{
if(!field.matches(".+\\..+"))
{
throw new IllegalArgumentException("field '" + field + "' needs to be a dotted path (i. e. customer.address.city.zipcode)");
}
String fromPart = field.substring(0, field.lastIndexOf('.'));
String fieldPart = field.substring(field.lastIndexOf('.') + 1);
From<?, ?> actualFrom = mapFieldToFrom.get(fromPart);
if(actualFrom == null)
{
throw new IllegalStateException("the given map does not contain a from or for the value '" + fromPart + "' or is null");
}
return actualFrom.get(fieldPart);
}
示例前端
<p:dataTable>
<!-- mapFieldToFrom.put("person", rootPerson); -->
<p:column field="person.name">
</p:column>
<!-- mapFieldToFrom.put("person.address", joinPersonAddress); -->
<p:column field="person.address.street">
</p:column>
</p:dataTable>