How to change the ordering of sql execution in hibernate - java

I am trying to model bidirectional parent-child design, with ordered children.
When removing child (eg. child #2 of 3 children) from the parent, hibernate's generated sql led to unique constraint violation because "update" (sibling) is being executed before the "delete" (target).
The RDBMS that I use (H2) does not support deferred constraint. What are the options I have besides the following ?
remove the unique constraint from schema
explicitly manage the sorting myself, rather than depend on hibernate
Is there any way to make hibernate generate sql with 'delete' precedes 'update' ?
Some old discussion found in the forum:
DELETE then INSERT in collection - Order of executed SQL
DB Schema:
CREATE TABLE IF NOT EXISTS Sequences (
ID BIGINT NOT NULL AUTO_INCREMENT,
Name LONGVARCHAR,
Type LONGVARCHAR,
Sequence LONGVARCHAR,
ParentId BIGINT DEFAULT NULL,
Index INT,
CONSTRAINT pk_SequenceId PRIMARY KEY (ID),
CONSTRAINT uc_Sequences UNIQUE (ParentId, Index),
CONSTRAINT fk_Sequences
FOREIGN KEY (ParentId)
REFERENCES Sequences(ID)
ON UPDATE CASCADE
ON DELETE CASCADE
);
Class:
public class Sequence {
protected Long ID;
protected String name;
protected String type;
protected String sequence;
protected Sequence parentSequence;
protected List<Sequence> childSequences = new ArrayList<Sequence>();
}
HBM Mapping:
<hibernate-mapping>
<class name="Sequence" table="Sequences">
<id name="ID" column="ID" type="long">
<generator class="native"/>
</id>
<property name="name" type="string" column="Name"/>
<property name="type" type="string" column="Type"/>
<property name="sequence" type="string" column="Sequence"/>
<many-to-one name="parentSequence" column="parentId" cascade="save-update" insert="false" update="false" class="Sequence" />
<list name="childSequences" inverse="false" lazy="true" cascade="all-delete-orphan">
<key column="parentId" not-null="true"/>
<list-index column="Index" base="0"/>
<one-to-many class="Sequence"/>
</list>
</class>

Hibernate does not execute the HQL (or SQL) statements directly but in the moment of a commit() or flush() it re-orders the SQL statement with the goal to do them in the most effective way. But it can happen the re-order from Hibernate is wrong and for example causes constraint violations, as in your case.
The solution is to introduce an intermediate flush(). flush() forces the re-order and send the SQL statements, but it does not commit.
In your case you can modify your code like (as a sketch):
transaction = session.beginTransaction();
session.delete(obj);
session.flush(); /* newly introduced */
session.update(...);
transaction.commit();
If the problem should be in a cascade delete or some delete executed by Hibernate without your control, then you have to take the control over the delete and update process and do the operations explicitely in your code instead of relying on the automatisms of Hibernate.

Related

How to create NO primary key table with Hibernate?

My goal is to create a no PK table in the database (I will never use it for ORM). I only need to create it, that's all.
I know it is possible to create a no PK table with NHibernate from this stack overflow question Why is NHibernate creating a table without primary key? But I need it to work in Java not .Net
Here is what works. It has PK, but I don't need it
<?xml version="1.0" encoding="UTF-8"?>
<hibernate-mapping>
<class entity-name="testtablenopi">
<id name="id"
type="long"
column="ID">
<generator class="sequence"/>
</id>
<property name='xarxint' column='xarxint' type='int' precision='11' scale='0'/>
<property name='xarxbig_decimal' column='xarxbig_decimal' type='big_decimal' precision='17' scale='2'/>
</class>
When I remove <id/> it doesn't work .
I tried replacing class with bag it didn't work.
How can I create a database table without a PI in hibernate?
It's not possible. Hibernate and all ORMs I know require a PK to be defined for an object.
You'll need to use a plain old JDBC insert statement.

Hibernate (hbm) - refreshing set with where clause after merging the container object

I'm working with Hibernate 4.3.5, Java 1.6 and Spring 4.0.3.
I've mapped the entities through hbm, and I want my application works with logical deletion.
So, in each mapped entity, I've added a property named 'deleted', which indicates if an entity is deleted or not.
Because I don't want to load the deleted entities (the ones having true the deleted property), I've used the where clause in the mapped classes, so I only get the entities aren't logically deleted.
And also, I've added the same where clause to every one-to-many relationship.
In one particular case, I've got a Report entity that has a one-to-many relationship with the Document entity.
So, when I mark a Document as deleted, and I save the Report entity (with merge), I expect than the Report entity doesn't keep the Document marked as deleted. But this doesn't occur.
This is the hbm for the Report entity:
<hibernate-mapping>
<class
name="es.entities.Report"
table="reports"
dynamic-insert="false"
dynamic-update="false"
where="deleted = 0">
<id name="id">
<generator class="identity"/>
</id>
<property name="title"></property>
<property name="deleted"></property>
<set
name="documents"
table="documents"
cascade="all"
lazy="false"
where="deleted=0">
<key column="id_report"/>
<one-to-many class="es.entities.Document"/>
</set>
</class>
</hibernate-mapping>
Here it is the hbm for the Document entity:
<hibernate-mapping>
<class
name="es.entities.Document"
table="documents"
dynamic-insert="false"
dynamic-update="false"
where="deleted = 0">
<id name="id">
<generator class="identity"/>
</id>
<property name="name"></property>
<property name="type"></property>
<property name="size"></property>
<property name="deleted"></property>
</class>
</hibernate-mapping>
I use a Service (ReportService) to open a Spring transaction. The method is:
#Autowired
private ReportDao reportDao;
#Transactional
public Report save(Report report) {
this.reportDao.save(report);
}
And this is the DAO (ReportDao) method I use to save the Report entity:
public Report save(Report report) {
return (Report) this.currentSession().merge(report);
}
I put an example:
The parameter I send to the service contains a Report object, with two Document objects, one of them deleted and the other not.
The DAO method returns the same information, but I'd like this method returns only the documents are not deleted.
Note: if I use another method with another transaction, I obtain the report only with the document is not deleted, but I'd like to do this in the same transaction.
Can anybody help me or show me an alternate to this? It is possible to use other Session method than merge?
Thanks a lot.
Merge method create a copy from the passed entity object and return it. Try re-fetching the report entity post merge.

LONG as primary key in Hibernate mapping to MySQL

I am trying to implement persistence of some Java objects via Hibernate mapping to a MySQL table. When I commit I get a message saying 'Batch update returned unexpected row count from update [0]; actual row count: 0; expected: 1'.
My hypothesis is that the problem is caused from having a long-field in my Java POJO that I want to use as my primary key in the MySQL table. Since I was not able to use datatype LONG as my primary key in MySQL table (ERROR 1170: BLOB/TEXT column 'id' used in key specification without a key length) I concluded from some googling and this post that BIGINT would be the suitable mapping for long. However it is not updating.
My test POJO Personis very simple. It has 3 fields: id (long), firstname (String), lastname (String) with setters and getters, etc.
I do the hibernate mapping in xml (person.hbm.xml) that essentially looks like (minus headings):
<hibernate-mapping>
<class name="hibernatetest.Person" table="hibernatetest">
<id name="id" type="long" column="id" >
<generator class="native"/>
</id>
<property name="firstname">
<column name="firstname" />
</property>
<property name="lastname">
<column name="lastname"/>
</property>
</class>
</hibernate-mapping>
My actual java code snippet that is supposed to save or update the record is simple:
Transaction tr = session.beginTransaction();
Person person = new Person(1,"John","Doe");
session.saveOrUpdate(person);
tr.commit();
And here's that thing, this all works just fine if I change the type of id to an int (Integer) in the Person object and in the MySQL table. However, I do not have that option for the actual objects that I want to persist so the question is; what am I doing wrong or what should I do to get it to work? Thanks.
ADDING Stacktrace:
Hibernate: update hibernatetest set firstname=?, lastname=? where id=?
org.hibernate.StaleStateException: Batch update returned unexpected row count from update [0]; actual row count: 0; expected: 1
at org.hibernate.jdbc.Expectations$BasicExpectation.checkBatched(Expectations.java:81)
at org.hibernate.jdbc.Expectations$BasicExpectation.verifyOutcome(Expectations.java:73)
at org.hibernate.engine.jdbc.batch.internal.NonBatchingBatch.addToBatch(NonBatchingBatch.java:57)
at org.hibernate.persister.entity.AbstractEntityPersister.update(AbstractEntityPersister.java:3006)
at org.hibernate.persister.entity.AbstractEntityPersister.updateOrInsert(AbstractEntityPersister.java:2908)
at org.hibernate.persister.entity.AbstractEntityPersister.update(AbstractEntityPersister.java:3237)
at org.hibernate.action.internal.EntityUpdateAction.execute(EntityUpdateAction.java:113)
at org.hibernate.engine.spi.ActionQueue.execute(ActionQueue.java:273)
at org.hibernate.engine.spi.ActionQueue.executeActions(ActionQueue.java:265)
at org.hibernate.engine.spi.ActionQueue.executeActions(ActionQueue.java:187)
at org.hibernate.event.internal.AbstractFlushingEventListener.performExecutions(AbstractFlushingEventListener.java:337)
at org.hibernate.event.internal.DefaultFlushEventListener.onFlush(DefaultFlushEventListener.java:50)
at org.hibernate.internal.SessionImpl.flush(SessionImpl.java:1082)
at org.hibernate.internal.SessionImpl.managedFlush(SessionImpl.java:317)
at org.hibernate.engine.transaction.internal.jdbc.JdbcTransaction.beforeTransactionCommit(JdbcTransaction.java:101)
at org.hibernate.engine.transaction.spi.AbstractTransactionImpl.commit(AbstractTransactionImpl.java:175)
at com.hibernate.test.TestMain.main(TestMain.java:38)
nested transactions not supported
UPDATE:
OK, I have finally worked it out. I changed the hibernate generator class from 'native' to 'assigned' and now it works as expected. So now the hibernate mapping looks like:
<hibernate-mapping>
<class name="hibernatetest.Person" table="hibernatetest">
<id name="id" type="long" column="id" >
<generator class="assigned"/>
</id>
<property name="firstname">
<column name="firstname" />
</property>
<property name="lastname">
<column name="lastname"/>
</property>
</class>
</hibernate-mapping>
Must admit I did not know the meaning of that parameter (copied from somewhere) and had no idea it could cause this much headache. Found this explanation which was quite useful.
Apparently I do not have enough credentials to answer my own questions so I guess that it will remain open or if someone provides an empty answer, I will accept it. Thanks.
When you use the saveOrUpdate() method hibernate fires the insert query if the id of the object is null and update if it is any other value. I can see the code,
Person person = new Person(1,"John","Doe"); setting the id to 1 and calling the saveOrUpdate() method. I am assuming there are no entries for the id 1 and hence the error is thrown.
To make it work, you need to make the below changes.
Change the Type of id in person to Long from long(The wrapper class so that it can support null).
Write the constructor new Person("John","Doe"); and save that object.
It is not a good Idea to keep the <generator class="assigned"/> for the transactional data. Instead you should be sticking to the native as you were trying first.
I feel this is a cleaner way to solve your initial problem, even though you have found an alternate solution.

Hibernate, Set keys, CompositeUserType and SQL Procedures

I am having some trouble with returning a non-empty Set into an object using Hibernate and a custom CompositeUserType key.
I have a set of tables and views (simplified here):
create table lang (lang_id,lang_cd);
create table article (art_id,...);
create table article_lang (art_id, lang_id,title,...);
create view article_lang_vw (select * from article join article_lang on art_id);
create table authors(user_id,...);
create table article_authors(art_id,lang_id,user_id);
And database functions:
create or replace procedure addarticle(title,art_id,lang_id) ...
create or replace procedure updatearticle(title,art_id,lang_id)..
create or replace procedure delarticle(art_id,lang_id)..
create or replace procedure addarticleauthor(user_id,art_id,lang_id)...
create or replace procedure delarticleauthor(user_id,art_id,lang_id)...
So to accomplish this mapping using those functions I had to implement CompositeUserType so now I have Java classes like this:
class ProcedureGenerator implements PostInsertIdentityGenerator ...
class Language { int lang_id }
class ArticleLangPKType implements CompositeUserType { //implemented methods }
class ArticleLangPK { int art_id; Language l; }
class Article { ArticleLangPK id; String title; Set<Author> authors; }
class Author { int user_id; String name; }
I want to have a List or Set of Authors. But cannot figure out how to map this part in the *.hbm.xml files. It currently looks something like this:
<class name="Author" mutable="false">
<id name="user_id"/>
<property name="name"/>
</class>
<class name="Article">
<id name="id" type="ArticleLangPKType">
<column name="art_id"/>
<column name="lang_id"/>
<generator class="ProcedureGenerator"/>
</id>
<property name="title"/>
<set name="authors" table="article_authors">
<key> <!-- <key type="ArticleLangPKType"> -->
<column name="art_id"/>
<column name="lang_id"/>
</key>
<many-to-many class="Author" table="article_authors" unique="true"/>
<!-- addauthor, delauthor sql here some how -->
</set>
<sql-insert callable="true">{call addarticle(?,?,?)}</sql-insert>
<sql-update callable="true">{call updatearticle(?,?,?)}</sql-update>
<sql-delete callable="true">{call adddelete(?,?)}</sql-delete>
</class>
But when I run this session.load(Article.class, pk) on an article I know has authors I get a Set size of zero. Otherwise I have no problems inserting, updating, deleting using Hibernate, but now I am stumped. This seems to me to indicate a problem with my ArticleLangPKType.
Any ideas what to do to complete this? Why is my Set always size 0? How would I save the author using the Article's Set and the SQL procedures as provided? Is Hibernate right for me? Do I need a break to see this clearly?
Thanks in advance!
Nevermind I did need a long break. My ArticleLangPK did not override hashCode and Equals correctly. Now just to figure out how to call those other two stored procedures correctly.

Hibernate Reference column in table to indicate how to unmarshall an attribute in a different column in the same table

I have an entity that I want to persist through Hibernate (3.2)
The EntityBean has a column that indicates how another value of the entity bean should be unmarshalled:
<class name="ServiceAttributeValue" table="service_attribute_value">
<cache usage="nonstrict-read-write"/>
<id name="id" column="id" type="int-long">
<generator class="native"/>
</id>
<property name="serviceAttribute" type="service-attribute" column="service_attribute" not-null="true" />
<!-- order is important here -->
<property name="value" type="attribute-value" not-null="true">
<column name="service_attribute" />
<column name="id_value"/>
<column name="enum_value"/>
<column name="string_value"/>
<column name="int_value"/>
<column name="boolean_value"/>
<column name="double_value"/>
</property>
</class>
The "service_attribute" column indicates which of the columns for the "value" property to look at when it unmarshalls the value and, more importantly, exactly what Type the value should be, for example the class of the Enum if the enum_value is to be read, or the type of Bean if the the id_value is to be read.
The value property uses a custom CompositeUserType to do the unmarshalling and within this I wish to reference the service_attribute column (although not write to it), however when I try to do this I get the following error:
org.hibernate.MappingException: Repeated column in mapping for entity: com.precurse.apps.rank.model.service.ServiceAttributeValue column: service_attribute (should be mapped with insert="false" update="false")
However within the definition of the composite property these xml attributes are not defined (only within a normal property).
Does anyone know of a way of overcoming this, or if there is a better solution to this propblem.
If you need any more information please let me know,
Cheers
Simon
I had a similar problem and changing the case of one column solved the problem. Could give a try!
e.g., one column could be service_attribute other Service_Attribute.
You can try this. Instead of mapping both values as property on the same table, map one of the property using join to itself and keep the other property as the way it is. This case you will be able to access the same property in both places. Just remember to name the property as different name.
<join table="service_attribute_value">
<key column = "id" />
<property name="serviceAttribute" type="service-attribute" column="service_attribute" not-null="true" />
</join>
<!-- order is important here -->
<property name="value" type="attribute-value" not-null="true">
<column name="service_attribute" />
<column name="id_value"/>
<column name="enum_value"/>
<column name="string_value"/>
<column name="int_value"/>
<column name="boolean_value"/>
<column name="double_value"/>
</property>
based on your description, it seems like what you want to do is creating different subclasses based on the service_attribute. Instead of trying to achieve repeated column mapping which is not allow in hibernate, you can take a look hibernate inheritance mapping.
I Think I found a solution albeit not a very elegant one.
in the
public Object nullSafeGet(ResultSet rs, String[] names, SessionImplementor session, Object owner)
throws HibernateException, SQLException {
method of the CompositeUserType the "owner" argument passed to the method contains the id of the object who's service_attribute I want to access.
Annoyingly the actual serviceAttribute of the owner is not accessable or has not been set at this stage (I played around with the ordering of the elements in the hbm.xml config, in case this was an ordering thing, but unfortunatly still no joy), so I can't simply access it.
Anyway the id of the owner object is set, so I then used the session argument to run a HQL query based on the id to access the serviceAttribute which I then used to correctly unmarshall the value property.
The drawback of this solution is that it requires a HQL query as an overhead to the unmarshalling process, although its within the same session, its still not optimal.
If anyone has any ideas for a better solution I'd be very grateful.
Cheers

Categories

Resources