Raible's Wiki

Raible Designs
Wiki Home
News
Recent Changes

AppFuse

Homepage
  - Korean
  - Chinese
  - Italian
  - Japanese

QuickStart Guide
  - Chinese
  - French
  - German
  - Italian
  - Korean
  - Portuguese
  - Spanish
  - Japanese

User Guide
  - Korean
  - Chinese

Tutorials
  - Chinese
  - German
  - Italian
  - Korean
  - Portuguese
  - Spanish

FAQ
  - Korean

Latest Downloads

Other Applications

Struts Resume
Security Example
Struts Menu

Set your name in
UserPreferences


Referenced by
Articles
Articles_de
CreateManager_de
LeftMenu




JSPWiki v2.2.33

[RSS]


Hide Menu

CreateDAO_de


This is version 8. It is not the current version, and thus it cannot be edited.
[Back to current version]   [Restore this version]


Part I: Neue DAOs und Objekte in AppFuse anlegen - Eine Anleitung zum Erstellen von Java Objekten (welche Tabellen darstellen) und Java Klassen, um diese Objekte persistent in einer Datenbank abzulegen.

Über dieses Tutorial

Dieses Tutorial erläutert, wie man eine neue Tabelle in der Datenbank anlegt und wie man Java Code erzeugt, um auf diese Tabelle zuzugreifen.

Zuerst werden wir ein Objekt erstellen und anschliessend weitere Klassen, um dieses Objekt in eine Datenbank zu speichern, wieder auszulesen und zu löschen. In der Java Sprache nennt sich dieses Objekt ein Plain Old Java Object (auch bekannt als POJO). Dieses Objekt stellt hauptsächlich eine Datenbanktabelle dar. Die anderen Klassen sind folgende:

  • Ein Datenzugriffsobjekt (Data Access Object (auch bekannt als DAO), ein Interface Interface und eine Hibernate Implementierung
  • Eine JUnit Klasse, um zu testen, ob unser DAO-Objekt richtig funktioniert.

Bemerkung: Falls man MySQL verwendet und Transaktionen genutzt werden sollen (was normalerweise der Fall ist), muss man InnoDB Tabellen nutzen. Um dies zu tun, muss man folgende Zeilen zur MySQL Konfigurationsdatei hinzufügen: (/etc/my.cnf oder c:\Windows\my.ini). Die 2. Zeile (Für UTF-8 character set) benötigt man für MySQL Versionen ab 4.1.7+.
[mysqld]
default-table-type=innodb
default-character-set=utf8

Wenn man PostgreSQL nutzt und komische Fehlermeldungen über batch processing erhält, kann man versuchen dies abzustellen indem man die Zeile <prop key="hibernate.jdbc.batch_size">0</prop> zur Datei src/dao/**/hibernate/applicationContext-hibernate.xml hinzufügt.

AppFuse nutzt Hibernate als Standard Persistenz Schicht. Hibernate ist ein Object/Relational (O/R) Framework, welches es einem ermöglicht, Java Objekte und Datenbanktabellen zu verknüpfen. Das Framework erlaubt es einem, sehr einfach CRUD (Create, Retrieve, Update, Delete) Methoden auf seine Objekte anzuwenden.

Man kann auch iBATIS als Persistenz Framework verwenden. Weitere Informationen, um iBATIS in AppFuse zu installieren, findet man in der README.txt in extras/ibatis. Falls man IBatis über Hibernate nutzen will, gehe ich davon aus, dass man gute Gründe dafür hat und sich mit dem Framework auskennt. Ich nehme auch an, dass man sich selbst vorstellen kann, wie man dieses Tutorial in Zusammenhang mit IBatis nutzen sollte. ;-)

Font Konventionen (in Arbeit)

Zeichenketten die über die Kommandozeile ausgeführt werden sollen, sehen folgendermasen aus: ant test-all.
Verweise auf Dateien, Verzeichnisse und Pakete, welche im Sourceverzeichnis existieren: build.xml.
und Vorschläge, wie man Dinge in der "realen Welt" durchführt, sind blau und schräg gestellt.

Beginnen wir mit dem Erstellen eines neuen Objekts, der DAO Klasse und der Tests in Appfuse Projekt Struktur.

Inhaltsverzeichnis

  • [1] Ein neues Objekt erzeugen und XDoclet tags hinzufügen
  • [2] Eine neue Datenbanktabelle mit Hilfe von Ant aus dem Objekt erzeugen
  • [3] Eine neuen DaoTest erzeugen, um JUnit Tests auf der DAO Klasse auszuführen
  • [4] Eine neue DAO Klasse erzuegen, um CRUD Methoden auf dem Objekt auszuführen
  • [5] Spring für das Person Objekt und die PersonDao umkonfigurieren
  • [6] Den DaoTest ausführen

Ein neues Objekt erzeugen und XDoclet tags hinzufügen [#1]

The first thing we need to do is create an object to persist. Let's create a simple "Person" object (in the src/dao/**/model directory) that has an id, a firstName and a lastName (as properties).

NOTE: Copying the Java code in these tutorials doesn't work in Firefox. A workaround is to CTRL+Click (Command+Click on OS X) the code block and then copy it.


package org.appfuse.model;

public class Person extends BaseObject {
    private Long id;
    private String firstName;
    private String lastName;

    /*
     Generate your getters and setters using your favorite IDE: 
     In Eclipse:
     Right-click -> Source -> Generate Getters and Setters
    */
}

This class should extend BaseObject, which has 3 abstract methods: (equals(), hashCode() and toString()) that you will need to implement in the Person class. The first two are required by Hibernate. The easiest way to do this is using Commonclipse. More information on using this tool can be found on Lee Grey's site. Another Eclipse Plugin you can use is Commons4E. I haven't used it, so I can't comment on its functionality.

If you're using IntelliJ IDEA, you can generate equals() and hashCode(), but not toString(). There is a ToStringPlugin that works reasonably well.
NOTE: If installing these plugins doesn't work for you, you can find all of these methods in the Person.java object that's used to test AppGen. Just look in extras/appgen/test/dao/org/appfuse/model/Person.java and copy and paste the methods from that class.

Now that we have this POJO created, we need to add XDoclet tags to generate the Hibernate mapping file. This mapping file is used by Hibernate to map objects → tables and properties (variables) → columns.

First of all, we add a @hibernate.class tag that tells Hibernate what table this object relates to:


/**
 * @hibernate.class table="person"
 */
public class Person extends BaseObject {

We also have to add a primary key mapping or XDoclet will puke when generating the mapping file. Note that all @hibernate.* tags should be placed in the getters' Javadocs of your POJOs.


    /**
     @return Returns the id.
     * @hibernate.id column="id"
     *  generator-class="increment" unsaved-value="null"
     */

    public Long getId() {
        return this.id;
    }

I'm using generator-class="increment" instead of generate-class="native" because I found some issues when using "native" on other databases. If you only plan on using MySQL, I recommend you use the "native" value. This tutorial uses increment.

Create a new database table from the object using Ant [#2]

At this point, you can create the person table by running ant setup-db. This task creates the Person.hbm.xml file and creates a database table called "person". From the ant console, you can see the table schema the Hibernate creates for you:
[schemaexport] create table person (
[schemaexport]    id bigint not null,
[schemaexport]    primary key (id)
[schemaexport] );

If you want to look at the Person.hbm.xml file that Hibernate generates for you, look in the build/dao/gen/**/model directory. Here's the contents of Person.hbm.xml (so far):


<?xml version="1.0"?>

<!DOCTYPE hibernate-mapping PUBLIC
    "-//Hibernate/Hibernate Mapping DTD 2.0//EN" 
    "http://hibernate.sourceforge.net/hibernate-mapping-2.0.dtd">

<hibernate-mapping>
    <class
        name="org.appfuse.model.Person"
        table="person"
        dynamic-update="false"
        dynamic-insert="false"
    >

        <id
            name="id"
            column="id"
            type="java.lang.Long"
            unsaved-value="null"
        >
            <generator class="increment">
            </generator>
        </id>

        <!--
            To add non XDoclet property mappings, create a file named


                hibernate-properties-Person.xml
            containing the additional properties and place it in your merge dir.
        -->

    </class>

</hibernate-mapping>

Now we'll add additional @hibernate.property tags for our other columns (first_name, last_name):


    /**
     * @hibernate.property column="first_name" length="50"
     */
    public String getFirstName() {
        return this.firstName;
    }

    /**
     * @hibernate.property column="last_name" length="50"
     */
    public String getLastName() {
        return this.lastName;
    }

In this example, the only reason for adding the column attribute is because the column name is different from our property name. If they're the same, you don't need to specify the column attribute. See the @hibernate.property reference for other attributes you can specify for this tag.

Run ant setup-db again to get the additional columns added to your table.

[schemaexport] create table person (
[schemaexport]    id bigint not null,
[schemaexport]    first_name varchar(50),
[schemaexport]    last_name varchar(50),
[schemaexport]    primary key (id)
[schemaexport] );

If you want to change the size of your columns, modify the length attribute in your @hibernate.property tag. If you want to make it a required field (NOT NULL), add not-null="true".

Create a new DaoTest to run JUnit tests on your DAO [#3]

NOTE: AppFuse versions 1.6.1+ contain include an AppGen tool that can be used to generate all the classes for the rest of these tutorials. However, it's best that you go through these tutorials before using this tool - then you'll know what code it's generating.

Now we'll create a DaoTest to test that our DAO works. "Wait a minute," you say, "we haven't created a DAO!" You are correct. However, I've found that Test-Driven Development breeds higher quality software. For years, I thought write your test before your class was hogwash. It just seemed stupid. Then I tried it and I found that it works great. The only reason I do all this test-driven stuff now is because I've found it rapidly speeds up the process of software development.

To start, create a PersonDaoTest.java class in the test/dao/**/dao directory. This class should extend BaseDaoTestCase, a subclass of Spring's AbstractDependencyInjectionSpringContextTests which already exists in this package. This parent class is used to load Spring's ApplicationContext (since Spring binds the layers together), and for (optionally) loading a .properties file (ResourceBundle) that has the same name as your *Test.class. In this example, if you put a PersonDaoTest.properties file in the same directory as PersonDaoTest.java, this file's properties will be available via an "rb" variable.

I usually copy (open → save as) an existing test (i.e. UserDaoTest.java) and find/replace [Uu]ser with [Pp]erson, or whatever the name of my object is.


package org.appfuse.dao;

import org.appfuse.model.Person;
import org.springframework.dao.DataAccessException;

public class PersonDaoTest extends BaseDaoTestCase {
    
    private Person person = null;
    private PersonDao dao = null;

    public void setPersonDao(PersonDao dao) {
        this.dao = dao;
    }
}

The code you see above is what we need for a basic Spring integration test that initializes and destroys our PersonDao. Spring will use autowiring byType to call the setPersonDao() method and set the "personDao" bean as a dependency of this class.

Now we need to actually test that the CRUD (create, retrieve, update, delete) methods work in our DAO. To do this we created methods that begin with "test" (all lower case). As long as these methods are public, have a void return type and take no arguments, they will be called by our <junit> task in our Ant build.xml file. Here's some simple tests for testing CRUD. An important thing to remember is that each method (also known as a test), should be autonomous. Add the following methods to your PersonDaoTest.java file:


    public void testGetPerson() throws Exception {
        person = new Person();
        person.setFirstName("Matt");
        person.setLastName("Raible");

        dao.savePerson(person);
        assertNotNull(person.getId());

        person = dao.getPerson(person.getId());
        assertEquals(person.getFirstName()"Matt");
    }

    public void testSavePerson() throws Exception {
        person = dao.getPerson(new Long(1));
        person.setFirstName("Matt");

        person.setLastName("Last Name Updated");

        dao.savePerson(person);

        if (log.isDebugEnabled()) {
            log.debug("updated Person: " + person);
        }

        assertEquals(person.getLastName()"Last Name Updated");
    }

    public void testAddAndRemovePerson() throws Exception {
        person = new Person();
        person.setFirstName("Bill");
        person.setLastName("Joy");

        dao.savePerson(person);

        assertEquals(person.getFirstName()"Bill");
        assertNotNull(person.getId());

        if (log.isDebugEnabled()) {
            log.debug("removing person...");
        }

        dao.removePerson(person.getId());

        try {
            person = dao.getPerson(person.getId());
            fail("Person found in database");
        catch (DataAccessException dae) {
            log.debug("Expected exception: " + dae.getMessage());
            assertNotNull(dae);
        }
    }

In the testGetPerson method, we're creating a person and then calling a get. I usually enter a record in the database that I can always rely on. Since DBUnit is used to populate the database with test data before our tests are run, you can simply add the new table/record to the metadata/sql/sample-data.xml file:

<table name='person'>
    <column>id</column>
    <column>first_name</column>
    <column>last_name</column>
    <row>
      <value>1</value>
      <value>Matt</value>
      <value>Raible</value>
    </row>
</table>
This way, you can eliminate the "create new" functionality in the testGetPerson method. If you'd rather add this record directly into the database (via SQL or a GUI), you can rebuild your sample-data.xml file using ant db-export and then cp db-export.xml metadata/sql/sample-data.xml.

In the above example, you can see that we're calling person.set*(value) to populate our object before saving it. This is easy in this example, but it could get quite cumbersome if we're persisting an object with 10 required fields (not-null="true"). This is why I created the ResourceBundle in the BaseDaoTestCase. Simply create a PersonDaoTest.properties file in the same directory as PersonDaoTest.java and define your property values inside it:

I tend to just hard-code test values into Java code - but the .properties file is an option that works great for large objects.
firstName=Matt
lastName=Raible
Then, rather than calling person.set* to populate your objects, you can use the BaseDaoTestCase.populate(java.lang.Object) method:


person = new Person();
person = (Personpopulate(person);

At this point, the PersonDaoTest class won't compile yet because there is no PersonDao.class in our classpath, we need to create it. PersonDAO.java is an interface, and PersonDAOHibernate.java is the Hibernate implementation of that interface. Let's go ahead and create those.

Create a new DAO to perform CRUD on the object [#4]

First off, create a PersonDao.java interface in the src/dao/**/dao directory and specify the basic CRUD methods for any implementation classes. I've eliminated the JavaDocs in the class below for display purposes.


package org.appfuse.dao;

import org.appfuse.model.Person;

public interface PersonDao extends Dao {
    public Person getPerson(Long personId);
    public void savePerson(Person person);
    public void removePerson(Long personId);
}

Notice in the class above there are no exceptions on the method signatures. This is due to the power of Spring and how it wraps Exceptions with RuntimeExceptions. At this point, you should be able to compile all the source in src/dao and test/dao using ant compile-dao. However, if you try to run ant test-dao -Dtestcase=PersonDao, you will get an error: No bean named 'personDao' is defined. This is an error message from Spring - indicating that we need to specify a bean named personDAO in applicationContext-hibernate.xml. Before we do that, we need to create the PersonDao implementation class.

The ant task for running dao tests is called test-dao. If you pass in a testcase parameter (using -Dtestcase=name), it will look for **/*${testcase}* - allowing us to pass in Person, PersonDao, or PersonDaoTest - all of which will execute the PersonDaoTest class.

Let's start by creating a PersonDaoHibernate class that implements the methods in PersonDao and uses Hibernate to get/save/delete the Person object. To do this, create a new class in src/dao/**/dao/hibernate and name it PersonDAOHibernate.java. It should extend BaseDaoHibernate and implement PersonDAO. Javadocs eliminated for brevity.


package org.appfuse.dao.hibernate;

import org.appfuse.model.Person;
import org.appfuse.dao.PersonDao;
import org.springframework.orm.ObjectRetrievalFailureException;

public class PersonDaoHibernate extends BaseDaoHibernate implements PersonDao {

    public Person getPerson(Long id) {
        Person person = (PersongetHibernateTemplate().get(Person.class, id);

        if (person == null) {
            throw new ObjectRetrievalFailureException(Person.class, id);   
        }

        return person;
    }

    public void savePerson(Person person) {
        getHibernateTemplate().saveOrUpdate(person);
    }

    public void removePerson(Long id) {
        // object must be loaded before it can be deleted
        getHibernateTemplate().delete(getPerson(id));
    }
}

Now, if you try to run ant test-dao -Dtestcase=PersonDao, you will get the same error. We need to configure Spring so it knows that PersonDaoHibernate is the implementation of PersonDAO, and we also need to tell it about the Person object.

Configure Spring for the Person object and PersonDao [#5]

First, we need to tell Spring where the Hibernate mapping file is located. To do this, open src/dao/**/dao/hibernate/applicationContext-hibernate.xml and add "Person.hbm.xml" to the following code block.


<property name="mappingResources"
    <list> 
        <value>org/appfuse/model/Person.hbm.xml</value> 
        <value>org/appfuse/model/Role.hbm.xml</value> 
        <value>org/appfuse/model/User.hbm.xml</value> 
        <value>org/appfuse/model/UserCookie.hbm.xml</value>  
    </list> 
</property> 

Now we need to add some XML to this file to bind PersonDaoHibernate to PersonDao. To do this, add the following at the bottom of the file:


<!-- PersonDao: Hibernate implementation --> 
<bean id="personDao" class="org.appfuse.dao.hibernate.PersonDaoHibernate"
    <property name="sessionFactory" ref="sessionFactory"/>
</bean> 

You could also use autowire="byName" to the <bean> and get rid of the "sessionFactory" property. Personally, I like having the dependencies of my objects documented (in XML).

Run the DaoTest [#6]

Save all your edited files and try running ant test-dao -Dtestcase=PersonDao one more time.

Yeah Baby, Yeah: BUILD SUCCESSFUL
Total time: 9 seconds


Next Up: Part II: Creating new Managers - A HowTo for creating Business Facades, which are similar to Session Facades, but don't use EJBs. These facades are used to provide communication from the front-end to the DAO layer.



Go to top   More info...   Attach file...
This particular version was published on 06-Nov-2006 13:52:49 MST by PeterSchneider-Manzell.