我们需要在我们的应用程序执行的每个查询中附加查询号。
EX：SELECT * FROM ... WHERE ... QUERYNO 123456 ;
OpenJPA支持查询提示，但仅针对特定实现的特定提示。
...
Query q = em.createQuery("select m from Magazine m where ... ");
q.setHint("openjpa.hint.OptimizeResultCount", new Integer(2));
q.setHint("openjpa.FetchPlan.ReadLockMode","WRITE");
List r = q.getResultList();
...
但根据JPA规范，openjpa“忽略了特定数据库无法处理的无效提示或提示。否则，无效提示将导致抛出ArgumentException。” 因此，将“QUERYNO”指定为提示似乎没有任何影响。
如何创建自定义查询提示以在运行时指定？
... 查询q = em.createQuery（“从杂志中选择m，其中......”）; q.setHint（“ com.me.CustomQueryNoHint ”，new Integer（2234））; 列表r = q.getResultList（）; ...
答案 0 :(得分：0)
指定“QUERYNO”作为提示似乎没有任何影响。
正确。根据您引用的文档，“QUERYNO”是无效提示，因此会被忽略。该文档有点令人困惑，但我相信如果可以解释为支持您观察到的行为。 ： - ）
如何创建自定义查询提示以在运行时指定？
这是一项更高的任务。我不相信OpenJPA旨在允许编写自定义查询提示。
我想到了更多关于你想要将特定字符串附加到每个SQL的实际问题，我认为在OpenJPA中做起来并不容易。也许您可以为JDBC驱动程序编写一个包装器，并将字符串附加到每个SQL？
答案 1 :(得分：0)
不是完整答案，只是一个指针......
<强> QueryCounter
public class QueryCounter {
private static long COUNTER = 0;
private static long next() {
return ++COUNTER;
}
private static String getHintValue() {
return "/*Query No. " + next() + " */";
}
public static void setQueryCount(Query query) {
/* EclipseLink */
//query.setHint(QueryHints.HINT, getHintValue());
query.setHint("eclipselink.sql.hint", getHintValue());
/* OpenJPA + Oracle */
//query.setHint("openjpa.hint.OracleSelectHint", getQueryHint());
/* OpenJPA + MySQL */
//query.setHint("openjpa.hint.MySQLSelectHin", getQueryHint());
}
}
<强>用法
Organization sun = new Organization("Sun");
em.persist(sun);
tx.commit();
Assert.assertNotNull(sun.getEntityId());
Query query = em.createQuery("SELECT org.entityId FROM Organization org WHERE org.entityId = " + sun.getEntityId());
QueryCounter.setQueryCount(query);
query.getResultList();
/*ServerSession does NOT log ReadObjectQuery??*/
query = em.createQuery("SELECT org FROM Organization org WHERE org.entityId = " + sun.getEntityId());
QueryCounter.setQueryCount(query);
query.getResultList();
query = em.createQuery("SELECT org.entityId FROM Organization org WHERE org.entityId = " + sun.getEntityId());
QueryCounter.setQueryCount(query);
query.getResultList();
<强>控制台
[EL Finest]: 2010-11-20 19:06:16.45--UnitOfWork(717879615)--Thread(Thread[main,5,main])--Execute query ReportQuery(referenceClass=Organization sql="SELECT entity_id FROM organization_tt WHERE (entity_id = ?)")
[EL Fine]: 2010-11-20 19:06:16.475--ServerSession(699542937)--Connection(1949550475)--Thread(Thread[main,5,main])--SELECT /*Query No. 1 */ entity_id FROM organization_tt WHERE (entity_id = ?)
bind => [1]
[EL Finest]: 2010-11-20 19:06:23.372--UnitOfWork(717879615)--Thread(Thread[main,5,main])--Execute query ReadObjectQuery(referenceClass=Organization sql="SELECT entity_id, name FROM organization_tt WHERE (entity_id = ?)")
[EL Finest]: 2010-11-20 19:06:35.916--UnitOfWork(717879615)--Thread(Thread[main,5,main])--Execute query ReportQuery(referenceClass=Organization sql="SELECT entity_id FROM organization_tt WHERE (entity_id = ?)")
[EL Fine]: 2010-11-20 19:06:35.92--ServerSession(699542937)--Connection(1949550475)--Thread(Thread[main,5,main])--SELECT /*Query No. 3 */ entity_id FROM organization_tt WHERE (entity_id = ?)
bind => [1]
OpenJPA有类似的概念1.8.7. Database-Specific Hints。 看看这些特定提示是否可以解决您的目的。
@Eddie，看看这可以帮到你............
<强> CustomLogFactory
public class MyLogFactory extends org.apache.openjpa.lib.log.LogFactoryImpl {
/* copied from LogFactoryImpl.NEWLINE */
private static final String NEWLINE = J2DoPrivHelper.getLineSeparator();
private boolean sqlLogger;
@Override
public Log getLog(String channel) {
if("openjpa.jdbc.SQL".equals(channel)) { // OR equalsIgnoreCase(channel) ???
sqlLogger = true;
}
return super.getLog(channel);
}
@Override
protected LogImpl newLogImpl() {
if(sqlLogger) {
sqlLogger = false; /* once an SQL Logger is created, we dont't need any more instances */
return new LogImpl() {
private long sqlCounter = 0;
@Override
protected String formatMessage(short level, String message, Throwable t) {
if(isSQLString(message)) {
StringBuffer formattedMessage = new StringBuffer(super.formatMessage(level, message, t));
StringBuffer queryNo = new StringBuffer();
queryNo.append(" [Query # ").append(++sqlCounter).append("]").append(NEWLINE);
formattedMessage.delete(formattedMessage.length() - NEWLINE.length(), formattedMessage.length());
formattedMessage.append(queryNo);
return formattedMessage.toString();
}
return super.formatMessage(level, message, t);
}
/* just a sample implementation; checks whether message contains the word "executing"
* more concrete implementation should check the message for SELECT, UPDATE, INSERT INTO, ALTER.... clauses */
private boolean isSQLString(String message) {
if(message.contains("executing")) {
return true;
}
return false;
}
};
}
return super.newLogImpl();
}
}
<强> peristence.xml
<property name="openjpa.Log" value="org.opensource.logger.MyLogFactory(DefaultLevel=WARN, Runtime=INFO, Tool=INFO, SQL=TRACE)"/>
<强>测试
EntityManager em = Persistence.createEntityManagerFactory("default").createEntityManager();
EntityTransaction tx = em.getTransaction();
tx.begin();
Person person = new Person();
person.setName("Bond-OO7");
person.setAge(22);
em.persist(person);
tx.commit();
em.close();
<强>控制台
............
2084 default TRACE [main] openjpa.jdbc.SQL - <t 346613126, conn 1551158018> executing prepstmnt 556472773 SELECT SEQUENCE_OWNER AS SEQUENCE_SCHEMA, SEQUENCE_NAME FROM ALL_SEQUENCES [Query # 1]
2136 default TRACE [main] openjpa.jdbc.SQL - <t 346613126, conn 1551158018> [52 ms] spent
2305 default TRACE [main] openjpa.jdbc.SQL - <t 346613126, conn 2026561073> executing prepstmnt 6637010 INSERT INTO Person (id, age, name) VALUES (?, ?, ?) [params=?, ?, ?] [Query # 2]
2306 default TRACE [main] openjpa.jdbc.SQL - <t 346613126, conn 2026561073> [1 ms] spent
............
答案 2 :(得分：0)
您是否考虑过编写映射到实体返回类型的本机查询，而不是使用JPQL并尝试强制使用特定于供应商的内容？
em.createNativeQuery(YOUR_DB2_NATIVE_SQL_QUERY_STRING, Magazine.class)
由于您的本机查询需要选择与实体类中的映射列匹配的列值，因此它将更适合您，但它应该可以工作。使用本机SQL时，查询引擎不得解析和解释特定于供应商的SQL SQL，因此最后的DB2特定子句应该在运行时传递给底层数据库。