我想了解如何使用Criteria API而不是JPQL来应用某些条件,特别是如果Criteria可以通过Join层次结构以与JPQL相同的方式递归地从Joins获取子实体。
更新
Tiny和Chris的评论促使我首先明确了我想要实现的目标:
我的示例有4个实体,如下图所示。 Enitems与Ensources有很多关系。 Entopics与Enitems有OneToMany关系。 Entopics与SegmentsNew有OneToMany关系。
我正在构建一个搜索页面,用户可以通过在搜索条件中输入尽可能多的内容来查找所选项目。在下面的例子中搜索"公司法"应归还公司法部分的所有项目(即使没有输入任何其他内容)。
我的实体:
Enitems:
@Entity
@Table(name = "enitem")
private static final long serialVersionUID = 1L;
@Id
@GeneratedValue(strategy = GenerationType.IDENTITY)
@Basic(optional = false)
@Column(name = "itemid")
private Integer itemid;
@Size(max = 500)
@Column(name = "itemname")
private String itemname;
@Column(name = "daterec")
@Temporal(TemporalType.DATE)
private Date daterec;
@Lob
@Size(max = 65535)
@Column(name = "itemdetails")
private String itemdetails;
private String enteredby;
@OneToMany(mappedBy = "items")
private Collection<Endatamaster> endatamasterCollection;
@JoinColumn(name = "topicid", referencedColumnName = "topicid")
@ManyToOne
private Entopic topics;
@JoinColumn(name = "sourceid", referencedColumnName = "sourceid")
@ManyToOne
private Ensource source;
Entopics:
public class Entopic implements Serializable {
private static final long serialVersionUID = 1L;
@Id
@GeneratedValue(strategy = GenerationType.IDENTITY)
@Basic(optional = false)
@Column(name = "topicid")
private Integer topicid;
@Size(max = 500)
@Column(name = "topicname")
private String topicname;
@Size(max = 500)
@Column(name = "description")
private String description;
@Column(name = "locksec")
private boolean locksec;
@JoinColumn(name = "segmentid", referencedColumnName = "SEGMENTID")
@ManyToOne
private Segmentnew segments;
@JoinColumn(name = "marketid", referencedColumnName = "marketid")
@ManyToOne
private Enmarkets markets;
@OneToMany(mappedBy = "topics")
private Collection<Enitem> enitemCollection;
Ensource:
@Entity
@Table(name = "ensource")
@NamedQueries({
@NamedQuery(name = "Ensource.findAll", query = "SELECT e FROM Ensource e")})
public class Ensource implements Serializable {
private static final long serialVersionUID = 1L;
@Id
@GeneratedValue(strategy = GenerationType.IDENTITY)
@Basic(optional = false)
@Column(name = "sourceid")
private Integer sourceid;
@Size(max = 500)
@Column(name = "sourcename")
private String sourcename;
@Size(max = 500)
@Column(name = "description")
private String description;
@JoinColumn(name = "typeid", referencedColumnName = "typeid")
@ManyToOne
private Enitemtype entype;
Segmentsnew:
public class Segmentnew implements Serializable {
private static final long serialVersionUID = 1L;
@Id
@Basic(optional = false)
@NotNull
@Column(name = "SEGMENTID")
private Integer segmentid;
@Size(max = 255)
@Column(name = "SEGMENTNAME")
private String segmentname;
@OneToMany(mappedBy = "segments")
private Collection<Entopic> entopicCollection;
@JoinColumn(name = "sectorId", referencedColumnName = "SECTORID")
@ManyToOne
private Sectorsnew sectors;
所需的JPQL字符串表示形式为:
Select e FROM Enitems e WHERE e.topics.topicid = :topicid
AND e.source.sourceid = :sourceid; AND e.topics.segments.segmentid = :segmentid
搜索页面使用JSf Primefaces:
<p:panelGrid columns="2">
<p:outputLabel value="Sectors: "/>
<p:selectOneMenu value="#{sectorBean.sectorid}"
filter="true">
<f:selectItem itemValue="0" itemLabel="NULL"/>
<f:selectItems value="#{sectorBean.secList}"
var="sect"
itemLabel="#{sect.sectorname}"
itemValue="#{sect.sectorid}"/>
<f:ajax listener="#{segmentsBean.segFromSec}" render="segs"/>
</p:selectOneMenu>
<p:outputLabel value="Segments: "/>
<p:panel id="segs">
<p:selectOneMenu value="#{queryBean.segmentid}"
rendered="#{not empty segmentsBean.menuNormList}">
<f:selectItems value="#{segmentsBean.menuNormList}"
var="segs"
itemLabel="#{segs.segmentname}"
itemValue="#{segs.segmentid}"/>
</p:selectOneMenu>
</p:panel>
<p:outputLabel value="Topics: "/>
<p:selectOneMenu value="#{queryBean.topicid}" filter="true">
<f:selectItem itemValue="" itemLabel="NULL"/>
<f:selectItems value="#{clienTopicBean.publicTopMenu}"
var="pubs"
itemLabel="#{pubs.topicname}"
itemValue="#{pubs.topicid}"/>
</p:selectOneMenu>
<p:outputLabel value="Type: "/>
<p:selectOneMenu value="#{typeBean.typeid}" filter="true">
<f:selectItem itemValue="" itemLabel="NULL"/>
<f:selectItems value="#{typeBean.menuList}"
var="type"
itemLabel="#{type.typename}"
itemValue="#{type.typeid}"/>
<f:ajax listener="#{sourceBean.sourceFromType}" render="src"/>
</p:selectOneMenu>
<p:outputLabel value="Sources: "/>
<p:panel id="src">
<p:selectOneMenu value="#{queryBean.sourceid}"
rendered="#{not empty sourceBean.sourceListNorm}">
<f:selectItems value="#{sourceBean.sourceListNorm}"
var="srcs"
itemLabel="#{srcs.sourcename}"
itemValue="#{srcs.sourceid}"/>
</p:selectOneMenu>
</p:panel>
</p:panelGrid>
这就是我尝试使用CAPI的原因:
public List<Enitem> superQ(Integer topicid, Integer sourceid,
Integer segmentid) {
em.clear();
CriteriaBuilder cb = em.getCriteriaBuilder();
CriteriaQuery cq = cb.createQuery(Enitem.class);
Root<Enitem> rt = cq.from(Enitem.class);
cq.select(rt);
cq.distinct(true);
List<Predicate> criteria = new ArrayList<Predicate>();
Predicate whereClause = cb.conjunction();
// if () {
Path<Entopic> topJoin = rt.get("topics");
if (topicid != 0) {
ParameterExpression<Integer> p
= cb.parameter(Integer.class, "topicid");
whereClause = cb.and(whereClause, cb.equal(topJoin.get("topicid"), p));
}
if (segmentid != 0) {
ParameterExpression<Integer> p
= cb.parameter(Integer.class, "segmentid");
whereClause = cb.and(whereClause, cb.equal(topJoin.get("segments").get("segmentid"), p));
}
//}
if (sourceid != 0) {
ParameterExpression<Integer> p
= cb.parameter(Integer.class, "sourceid");
whereClause = cb.and(whereClause, cb.equal(rt.get("source").get("sourceid"), p));
}
// if(whereClause.getExpressions().isEmpty()) {
// throw new RuntimeException("no criteria");
// }
cq.where(whereClause);
TypedQuery q = em.createQuery(cq);
if (topicid != 0) {
q.setParameter("topicid", topicid);
}
if (segmentid != 0) {
q.setParameter("segmentid", segmentid);
}
if (sourceid != 0) {
q.setParameter("sourceid", sourceid);
}
return q.getResultList();
}
重要的是陈述if(entityName!= 0)而不是if(entityName!= null),这是我之前做过的,这导致应用程序要求用户填充所有参数。这可能是因为null的整数值实际上是数字零
生成的SQL:
SELECT DISTINCT t1.itemid, t1.daterec, t1.ENTEREDBY, t1.itemdetails, t1.itemname,
t1.sourceid, t1.topicid FROM enitem t1 LEFT OUTER JOIN entopic t0 ON (t0.topicid
= t1.topicid) LEFT OUTER JOIN ensource t2 ON (t2.sourceid = t1.sourceid) WHERE
(((t0.topicid = ?) AND (t2.sourceid = ?)) AND (t0.segmentid = ?))
应用程序动态运行,因为用户只需要在搜索页面中输入任何单个值,并返回与该值对应的列表。 我现在遇到的问题是,如果我进行第二次查询,即使我已经在方法开始时清除了EntityManager,也会返回相同的结果。因此,应用程序仅在重新启动时才有效。我是否需要参考实体?
答案 0 :(得分:1)
使用From和join两者为您提供可以强制转换为根接口的对象,允许连接被链接 尝试类似:
public List<Enitem> superQ(Integer topicid, Integer sourceid,
Integer segmentid) {
CriteriaBuilder cb = em.getCriteriaBuilder();
CriteriaQuery cq = cb.createQuery(Enitem.class);
Root<Enitem> rt = cq.from(Enitem.class);
cq.select(rt);
cq.distinct(true);
List<Predicate> criteria = new ArrayList<Predicate>();
Predicate whereClause = cb.conjunction();
if ((topicid != null)||(segmentid != null)){
Path<Entopic> topJoin =rt.get("topics");
if(topicid != null) {
ParameterExpression<Integer> p =
cb.parameter(Integer.class, "topicid");
whereClause = cb.and(whereClause, cb.equal(topJoin.get("topicid"), p));
}
if(segmentid != null) {
ParameterExpression<Integer> p =
cb.parameter(Integer.class, "segmentid");
whereClause = cb.and(whereClause, cb.equal(topJoin.get("segments").get("segmentid"), p));
}
}
if(sourceid != null) {
ParameterExpression<Integer> p =
cb.parameter(Integer.class, "sourceid");
whereClause = cb.and(whereClause, cb.equal(rt.get("source").get("sourceid"), p));
}
if(whereClause.getExpressions().isEmpty()) {
throw new RuntimeException("no criteria");
}
cq.where(whereClause);
}
除非指定了需要连接的参数,否则这将产生没有任何连接的东西,并且将使用内部连接作为提供的JPQL。
答案 1 :(得分:0)
所以事实证明JPA的逻辑是正确的,但问题在于Web框架,这意味着我必须相应地调整我的JPA。 JSF不接受'null'作为搜索页面中整数的定义,只接受数值或只接受“”。更改if语句:
if(entity != null);
到
if(entity != 0);
导致应用程序按预期运行。这解决了这个问题的问题。感谢Chris和Tiny的帮助