Alexey Kakunin

Workflow and Reports in Liferay 6

2010-08-04T15:53:06Z

Liferay 6 has 2 very nice features, probably not well documented yet. It is support for Workflows and Reports. Here I want to discuss some major details about these features, since there are no too much info about them.

Support for Workflows

How it works from developer side:

Liferay has set of general interfaces, implemnted working with workflows (look at the package com.liferay.portal.kernel.workflow)
It is possible to deploy some implementation of these interfaces as separate plugin.
By default, Liferay 6.0.4 ships with kaleo-web - plugin implemented workflow functionality with using own developed workflow engine Kaleo;
But it is possible to deploy jbpm3 based implementation (available in svn: http://svn.liferay.com/repos/public/plugins/trunk/webs/jbpm-web)
As well as I've started development of jBPM4 based implementation (http://www.emforge.net/web/jbpm4-portlet), but, since there was no interest from community side, as well as no strong requirements in this implementation in my current project this implementation currently placed on-hold;
Then somebody (any portlet) calls core Workflow methods in Liferay - these calls routed into deployed implementation (with using internal Liferay's Message Bus).

How it works from Administrator Side:

After deploying some implementation, it is possible to manage workflows in the control panel (for example deploy workflow definitions);
Set of workflows is individual for each portal instance;
As well, for each group it is possible to define own Workflow Configuration: mapping of Assets with Workflows, responsible for their processing;

Important thing - in Liferay each process instance (workflow instance) are connected to some Asset. It is not possible to just "create a task", but, you can create/edit some asset and it will "flow" by some workflow (according to asset type and workflow linked with this type in Workflow Configuration).

How it works from User Side:

Actually, how to work with workflows from user side quite well described in Juan's blog-post. I can just add - fi you like to experiment with workflows - simple start default Liferay 6.0.4 bundle with 7cogz-hook deployed. It deploy's kaleo workflow engine, as well as defined users with required roles - you can play with editing web-content by "Editor" users, and approve changes by "Publisher" user. It is good enough to see how it works (and very fast to start).

Support for Reports

Supports for reports is very simular to support for Workflow:

Liferay core has some interfaces for working with reports (looks at ReportEngine class);
It is possible to deploy some implementation;
In Incubation SVN you can find JasperReport's based implementation: svn.liferay.com/repos/public/plugins/incubation/webs/reporting-jasper-web/
By save way - all calls to core ReportEngine are routed to implementation with using MessageBus;
After deploying some implementation, you can use reports in your portlets, but
You may deploy "Report Console" portlet from svn incubation: svn.liferay.com/repos/public/plugins/incubation/portlets/reports-console-portlet/
It will add Reports Tab into your Control Panel - to allow admin to deploy reports and manage them, manage data sources and some other settings, as well as place "View Report" portlet on any page

I think these two features are very good big steps forward to make Liferay more powerful and more "Enterprise". Just again - big "thank you" to Liferay team for great job!

First project migrated to 6.0.3

2010-07-17T20:15:00Z

I already have several projects based on Liferay 6 branch, some of them based on 6.0.1, some on 6.0.2. Few days ago first GA version of Liferay 6 was released so, it's time to move project forward.

Today was first - based on 6.0.1. Here is some notes for developers, to know what expect from such migration:

Database upgraded without any problems. Looks like error in upgrade-script between 6.0.1 and 6.0.2 is fixed and now everything works as expected. Simple place new Liferay instead old one and connect to same DB;
Dependencies - minor changes required (jstl.jar renamed to jstl-api.jar);
Source code - some minor changes required because of changed API (expectable);
JSP pages: here was major change I had to do: in my project jsp pages was located in /WEB-INF/jsp, and Liferay 6.0.3 is not allowed to load resources under WEB-INF (ok, I thin it is configurable - but I will prefer to stay with standard configuration as much as possible). So, I had to move jsp pages to /jsp (and of course change all source-code). Hopefully it is very easy change.

So, migration from 6.0.1 to 6.0.3 works almost without any problems. Good job Liferay Team!

Simple Vaadin Portlet in Liferay Plugins SDK

2010-07-12T08:49:44Z

Vaadin is a new word in Liferay world, it looks very promised, but, as always, not so much documentation and tutorials. Some of tutorials too complex, some not covered some (even simple) questions.

Here I want to explain for Liferay developers already used Plugins SDK (and have experience in Liferay Portlets development) how to write simple Vaadin portlet and access Portlet and Liferay infrastructure (handle render request, access ThemeDisplay object).

Actually, you can skip everything and simple get sources from SVN: svn.emforge.net/svnroot/akakunin-experiments/sample-vaadin-portlet - it is really very simple.

Prerequirements

I've used latest public version of Liferay (6.0.2) and it's plugins SDK. Also I'm using Eclipse (not using Liferay IDE yet, since had no time to switch into it) - but it is not required for this tutorial - it is based on Liferay SDK, so, used ant for build and may be used with any IDE

Create portlet

Switch to sdk/portlets and call

#  ./create.sh sample-vaadin

Then remove all renerated jsp-pages, css, images and so on - everything was generated for sample portlet

Configure web.xml for working with Vaadin

Create web.xml (in my case it wasn't created by plugins-sdk) with setting to enable using VAADIN in your portlet:

<?xml version="1.0" encoding="UTF-8"?>
<web-app xmlns="http://java.sun.com/xml/ns/j2ee" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xsi:schemaLocation="http://java.sun.com/xml/ns/j2ee http://java.sun.com/xml/ns/j2ee/web-app_2_4.xsd" version="2.4">
  <context-param>
      <description>Vaadin production mode</description>
      <param-name>productionMode</param-name>
      <param-value>false</param-value>
  </context-param>
  <servlet>
      <servlet-name>Vaadin_portlet Application</servlet-name>
      <servlet-class>
      com.vaadin.terminal.gwt.server.ApplicationServlet</servlet-class>
      <init-param>
          <description>
          Vaadin application class to start</description>
          <param-name>application</param-name>
          <param-value>com.example.vaadin_portlet.Vaadin_portletApplication</param-value>
      </init-param>
  </servlet>
  <servlet-mapping>
      <servlet-name>Vaadin_portlet Application</servlet-name>
      <url-pattern>/Vaadin_portletApplicationServlet/*</url-pattern>
  </servlet-mapping>
  <servlet-mapping>
      <servlet-name>Vaadin_portlet Application</servlet-name>
      <url-pattern>/VAADIN/*</url-pattern>
  </servlet-mapping>
</web-app>

Add Vaadin into portlet dependencies

Liferay itself already has vaadin, you do not need to include it into your portlet lib. Simple add it into portal-dependency-jars in liferay-plugin-package.properties:

name=
module-group-id=liferay
module-incremental-version=1
tags=
short-description=
change-log=
page-url=http://www.liferay.com
author=Liferay, Inc.
licenses=LGPL
portal-dependency-jars=\
    vaadin.jar

Add portlet definition into portlet.xml & liferay-portlet.xml

Add into portlet.xml:

    <portlet>
        <portlet-name>sample-vaadin</portlet-name>
        <display-name>Sample Vaadin</display-name>
       
        <portlet-class>com.vaadin.terminal.gwt.server.ApplicationPortlet2</portlet-class>
        <init-param>
            <name>application</name>
            <value>sample.vaadin.SampleApplication</value>
        </init-param>
       
        <!--
          To enable displaying multiple Vaadin portlets on a page,
          they should all use the same widgetset. In that case, the
          widgetset can be configured on the portal level (parameter
          vaadin.widgetset) or here for each portlet.
        -->
        <!--
        <init-param>
            <name>widgetset</name>
            <value>com.vaadin.portal.gwt.PortalDefaultWidgetSet</value>
        </init-param>
        -->
       
        <!-- Supported portlet modes and content types. -->
        <supports>
            <mime-type>text/html</mime-type>
            <portlet-mode>view</portlet-mode>
            <!-- <portlet-mode>edit</portlet-mode> -->
            <!-- <portlet-mode>help</portlet-mode> -->
        </supports>
       
        <!-- Not always required but Liferay uses these. -->
        <portlet-info>
            <title>Sample Vaadin</title>
            <short-title>Sample Vaadin</short-title>
        </portlet-info>
    </portlet>

and liferay-portlet.xml:

    <portlet>
        <portlet-name>sample-vaadin</portlet-name>
        <ajaxable>false</ajaxable>
    </portlet>

Implement Application Class

You can get sources of class from svn: svn.emforge.net/svnroot/akakunin-experiments/sample-vaadin-portlet/docroot/WEB-INF/src/sample/vaadin/

Some important points in this class:

Get PortletApplicationContext2 and add PortletListener into it;
SamplePortletListener (PortletListener implementation) will allow you to access Portlet API (render, action, even, resource);
As result, you can use standard way to access (for example) ThemeDisplay object - and information about your current context.

What is all - take fun with Liferay and Vaadin!

Ubutu 10.04, mysql, Java, localhost = WTF?

2010-07-11T21:56:15Z

Today I killed 2 hours of my life for very studip problem.

I just got new notebook (IBM Thnkpad T400) with 4Gb, so, to be able to use whole memory, installed Ubuntu 10.04 64 bit. Then installed mysql-server, sun java, eclipse, some other staff and started to work. mysql nd sun java was installed via apt with standard settings.

Created DB, successfully connected via MySQL Admin and Query Browser, and, then, trying to connect from my app: failed

Exception in thread "main" com.mysql.jdbc.exceptions.jdbc4.CommunicationsException: Communications link failure
Last packet sent to the server was 0 ms ago.
    at sun.reflect.NativeConstructorAccessorImpl.newInstance0(Native Method)

Checking JDBC Url - everything is correct, checking username/password - same, I can connect to localhost DB with using this username/password with using standard mysql tools. But not from java via JDBC.

WTF???

After spending 2 hours, found this forum. After commenting bind-address in my.cnf and changing jdbc url to use my computer IP address (like 192.168.0.157) instead of localhost it finally started to connect.

Why? Why it is not working with localhost out-of-box? Why I should spent 2 hours for such stupid problems?

Problem resolved

Ok, today (after sleeping a little bit) I realized the source of problem. It is not ubuntu, java, or mysql. It is (of course) - I'm.

Yesterday I changed by /etc/hosts - so, localhost was defined incorrectly (by some reason). But, MySQL clients, in case localhost server used - not used TCP at all - then connected to mysql directly.

JDBC driver does not use it - it always (looks like) tried to use TCP and resolve hostname. It is why it connection worked in mysql tools, but did not worked in java.

Ubuntu == cool!

Java == cool!

I'm == dummy developer :)

PetClinic GWT application in less then 30 minutes

2010-05-20T23:30:03Z

Yesterday I've read announce of new version of Spring-Roo 1.1.0M1 and found there at least 2 very interesting features: GWT & GAE support.

So, today I spent some time to play with it - and here is short instructions how to make PetClinic application in less then 30 minutes (depends from your connection speed).

Prepequirements

You will need to have JDK 1.6 and Apache Maven installed. I hope it is not a problem and sure 90% readers of this blog already have them

Download and Install Spring-Roo

Simple download it from SpringSource Site and unzip into some folder (lets say it <spring-roo-path>)

Get clinic.roo file

Download clinic.roo file. It is script for roo, almost same as default clinic.roo (you can find initial version in <spring-roo-path>/samples/clinic.roo) but adopted for using GWT:

Commented enums since Roo-GWT is not supported them; In entities enums replaced by string;
Boolean primitive looks like also not supported - so, changed to java.lang.Boolean;
Calendar is not supported changed to Date;
Commented out Spring-MVC specific commands
Added gwt setup to generate gwt UI.

Run Roo to generate sources

Lets generate sources with using this roo-file:

# ../bin/roo.sh
# roo> script -file clinic.roo
# roo> exit

It will took some time, but as result you will get maven-based project with all required sources.

Run it

Quite easy:

# mvn gwt:run

It may also take some time, since it will make a build and download all required artifacts from maven repositories. But it will long only for first time - after all requirements will be downloaded into local maven repo, next builds will go much faster.

Try it

Copy provided link into clipboard and open it in your browser (it will require to install GWT plugin)

So, now you can play with it. Not all forms working correctly - ok, GWT support in Spring-Roo is not in stable state, but at least it provide some ideas how it will be implemented.

What we have as result

As result, in less then 30 minutes we have great sample of:

Fast application generation with using Spring-Roo. Samples are very clear and you can use them as reference material for Roo features. Of course - if you interested - it make sense to read more Roo documentation
Sample of project used Spring Framework, GWT, Hibernate and many more;

If you just started with one of these technologies, these code-samples may be good reference.

Even you already using them for a long time - I'm pretty sure you can find many interesting things you probably did not know about Spring, Hibernate or GWT

Of course created application is not something you can show to end-users. But - it may be good "start-point" and safe your time for doing some standard things for project initialization again and again.

Magic options to monitor remote glassfish on ubuntu

2010-05-05T23:15:34Z

I've tried to setup JMX connection to my remote glassfish installed on ubuntu several times, and every time I've got 'cannot connect' error from my jconsole or VisualVM

I've checked many tutorials in internet, in most case it should work out of box. Some people reported (like I'm) what they cannot connect to remote glassfish - but, without any answer and proper solution.

Today I finally resolved 2 magic options enabled for me remote monitoring of Glassfish under Ubuntu (with using jconsole or VisualVM). They are:

-Djava.rmi.server.hostname=xxx (use domain name you are using to access this server);
-Djava.net.preferIPv4Stack=true (in my cases by default Glassfish used IPv6 stack);

So, getting Glassfish with default settings (by default JMX conector is enabled), add this two JVM options via glassfish admin console and (of course if you have no firewall behind you and this server) after restarting glassfish it is possible to connect and monitor it with using both jconsole and VisualVM (as well as I hope any JMX client).

Securing Remote EJB on Glassfish 2

2010-05-04T12:51:27Z

Today I needed to write simplest remote EJB and command-line client for it and protect this EJB - request authentication. Glassfish2 was used as application server, but, I suppose same will work (except probably required libraries) by same way with Glassfish 3.

Write EJB

@Stateless
public class TestSecurityBean implements TestSecurityBeanRemote, TestSecurityBeanLocal {
    private static Logger log = Logger.getLogger(TestSecurityBean.class.getName());
    @Resource
    EJBContext ejbContext;
   
    public String sayRemoteHello() {
        log.info("EJBContext: " + ejbContext);
        Principal principal = ejbContext.getCallerPrincipal();
        log.info("Caller Principal: " + principal);
       
        return "Hello " + principal.getName() + "!";
    }
   
    // Add business logic below. (Right-click in editor and choose
    // "Insert Code > Add Business Method")
 
}

Say glassfish to request authentication for calling this EJB (in META-INF/sun-ejb-jar.xml):

<sun-ejb-jar>
    <enterprise-beans>
        <ejb>
            <ejb-name>TestSecurityBean</ejb-name>
            <ior-security-config>
                <transport-config>
                    <integrity>supported</integrity>
                    <confidentiality>supported</confidentiality>
                    <establish-trust-in-target>supported</establish-trust-in-target>
                    <establish-trust-in-client>supported</establish-trust-in-client>
                </transport-config>
                <as-context>
                    <auth-method>username_password</auth-method>
                    <realm>default</realm>
                    <required>true</required>
                </as-context>
                <sas-context>
                    <caller-propagation>supported</caller-propagation>
                </sas-context>
            </ior-security-config>
        </ejb>
    </enterprise-beans>
</sun-ejb-jar>

Write client:

public class Client {
    private static Logger log = Logger.getLogger(Client.class.getName());
    /**
     * @param args the command line arguments
     */
    public static void main(String[] args) throws NamingException {
        TestSecurityBeanRemote testSecurityService = getRemoteService();
        String result = testSecurityService.sayRemoteHello();
        log.info("Result:" + result);
    }
    public static TestSecurityBeanRemote getRemoteService() throws NamingException {
        TestSecurityBeanRemote testSecurityService = null;
        String host = "localhost";
        String port = "3700";
        Properties props = new Properties();
        props.setProperty("java.naming.factory.initial", "com.sun.enterprise.naming.SerialInitContextFactory");
        props.setProperty("java.naming.factory.url.pkgs", "com.sun.enterprise.naming");
        props.setProperty("java.naming.factory.state", "com.sun.corba.ee.impl.presentation.rmi.JNDIStateFactoryImpl");
        props.setProperty("org.omg.CORBA.ORBInitialHost", host);
        props.setProperty("org.omg.CORBA.ORBInitialPort", port);
        // standard way to put authentication info into remote ejb call
        // it is not working with Glassfish - own way for authentication should be used
        // props.setProperty(Context.SECURITY_PRINCIPAL, "user1");
        // props.setProperty(Context.SECURITY_CREDENTIALS, "password1");
        ProgrammaticLogin programmaticLogin = new ProgrammaticLogin();
        programmaticLogin.login("user1", "password1");
        Context amInitial = null;
        amInitial = new InitialContext(props);
        testSecurityService = (TestSecurityBeanRemote) amInitial.lookup(TestSecurityBeanRemote.class.getName());
        return testSecurityService;
    }
}

Write auth.conf

default { com.sun.enterprise.security.auth.login.ClientPasswordLoginModule required debug=false; };

Add required libraries from glassfish into project dependencies:

$GF_PATH/lib/appserv-rt.jar
$GF_PATH/lib/appserv-admin.jar
$GF_PATH/lib/javaee.jar

Run client with option to use this file with option -Djava.security.auth.login.config=./auth.conf

Add user1 into file realm:

#./asadmin create-file-user user1

Now you should be able to call your remote EJB with using specified user and password

Liferay 5.2.3 Amazon EC2 Image (AMI) for EU region

2010-02-03T01:41:19Z

In one of previous blog post: www.emforge.net/web/akakunin/blogs/-/blogs/liferay-5-2-3-amazon-ec2-image-ami I've placed information about Liferay AMI.

Only problem - that AMI was only for US region and I received request to make same to be run in EU.

So - here it is:

AMI: ami-0bddf67f

To run it you can use command like:

$ ec2-run-instances --region eu-west-1 ami-0bddf67f

Everything else same as for US region image

Spring 3 + JPA application on Glassfish 2.1

2010-01-24T22:58:07Z

Today I've spent quite a lot of time trying to run one Spring 3.0 based application used JPA (Hibernate) for working with DB on Glassfish.

In development we sing tomcat and everything works well, but, on test-server I have glassfish installed, and - application did not work on it:

Only persistence units with transaction type JTA can be used as a container managed entity manager. 
Please verify your application.

Solution was found on Spring Forum: http://forum.springsource.org/showthread.php?t=58527

Short fix (since I do not need JTA transactions) - change web.xml version to 2.4

This thead also has set of settings need to be done in the project to enable usage of JTA

Install jBPM 4.3 into existed JBoss 5.1

2010-01-19T18:37:12Z

jBPM package provided ant-based installed with bunch of targets. Main target demo.setup.jboss supposed to do everything for you:

Download and install jBoss server (5.0 by default);
Download and install eclipse;
Install jBPM service and demo applications into jBoss;
Install GPD into Eclipse;

But, in my case I do not need it - I've already have JBoss 5.1 installed and want to use it. As well, as I've already have Eclipse, no needs to install another one.

So, here is small set of instructions how to run new jBPM 4.3 into existed JBoss 5.1 installation.

In examples bellow I suppose JBoss installed in some <path-to-jboss> and jbpm unzipped into some <path-to-jbpm>

Change configuration files

To perform installation you need to change a little bit <path-to-jbpm>/install/build.xml:

  <property name="jboss.version" value="5.1.0.GA" />
  <property name="jboss.parent.dir" value="${jbpm.home}" />
  <property name="jboss.home" value="<path-to-jboss>" />

So, we will tell jBPM istall scripts to use your JBoss server

Change DB configuration

By default, jBPM configured to use HSQL, what is OK in case we are using JBoss server. But, jBPM supposed own HSQL server to be run. Do you really need it? In my case I've prefer to use same way as used in server for DefaultDS. So, we need to edit <path-to-jbpm>/install/jdbc/hsqldb.properties:

jdbc.driver=org.hsqldb.jdbcDriver
jdbc.url=jdbc:hsqldb:<path-to-jboss>/server/default/data/hypersonic/jbpmDB
jdbc.username=sa
jdbc.password=

As result jbpmDB will be created in same folder as default localDB

Of course, if you want to use some other DB - like MySQL or PostgreSQL you will need to make your changes accordenly

Create DB Schema

Simple run in your <path-to-jbpm>/install

# ant create.jbpm.schema

Install jBPM

Now to install jBPM do:

# ant install.jbpm.into.jboss

Install Example Data

If you like - you can add example data (users and processes) into DB (JBoss server should be stopped during this actions):

# ant load.example.identities

In my cases it is reported 14 SQL statements were executed, but it is not created required users (I do not know why), so, I had to do it by hand:

Run HSQL Database Manager:

# ant hsqldb.databasemanager

Connect it to my DB at jdbc:hsqldb: <path-to-jboss>/server/default/data/hypersonic/jbpmDB

Run SQL statements from <path-to-jbpm>/install/src/demo/example.identities.sql

Run Server, Validate everything working

Now, you can run your server: jbpm-console will be available at http://localhost:8080/jbpm-console and you can login as one of the users: alex, mike, peter, mary and password 'password'

As result you have:

JbpmDS datasource configured, pointed to HSQL DB;
JbpmService configured;
jbpm-console & jbpmeditor web-applications deployed

As well as some other staff required for proper jBPM working

SVN Authentication by using username/passwords in Liferay DB (MySQL)

2010-01-19T12:45:19Z

Last days I've worked on SVN repositories setup on emforge.net and of course I've wanted to use same username/passwords for user authentication in SVN as used to login into EmForge Portal.

After googling a little bit I found what I need: mod-auth-mysql - module for Apache allowed to perform authentication with using data in MySQL DB.

Starting links I've used here:

dev.e-taxonomy.eu/trac/wiki/ApacheMySQLAuthentication
www.howtoforge.com/mod_auth_mysql_apache2_debian
http://maff.ailoo.net/2009/03/authenticate-apache-against-redmine-with-authmysql/

So, to setup svn authentication (ok, any authentication in apache server) used same usernames/password as in Liferay Portal you need:

Setup SVN Repository

Of course you need to create svn repository and add it's configuration into dav_svn.conf. I'm avoiding here these topics - hope it is well-known how to do.

Also you need to setup and enable mod-auth-mysql module for apache server. It depends from your OS used, so, also not covered here.

Then in your svn-location in dav_svn.conf for specific repository you need to setup authentication against mysql DB

  Require valid-user
  # disable password-file based authentication
  #AuthUserFile /vol/svn/passwd
  AuthUserfile /dev/null
  # authenticate against mysql DB
  AuthBasicAuthoritative Off
  AuthMySQL On
  AuthMySQL_Authoritative      on
  AuthMySQL_Host <mysql host>
  AuthMySQL_User <mysql user name>
  AuthMySQL_Password <mysql user password>
  AuthMySQL_DB <portal db in mysql server>
  AuthMySQL_Password_Table user_
  AuthMySQL_Username_Field screenName
  AuthMySQL_Password_Field password_
  AuthMySQL_Encryption_Types SHA1Sum
  AuthMySQL_Group_Table user_
  AuthMySQL_Group_Field active_
  Require group         1

This configuration setup connection to Liferay DB (you need use own host, db, user and password to connect), specified which table and fields to use for authentication.

Also, it is filtered to allow authentication only for active users.

Change configuration in Liferay

Unfortunatelly, it is not enough to setup only SVN settings. Even both - Liferay and mod-auth-mysql - used SHA1 for password encoding, Liferay used base64 to digester encoded passwords, mod-auth-mysql used hex.

Hopefully, liferay also can use it - you need simple add option into your portal-ext.properties:

passwords.digest.encoding=hex

Convert passwords

OK, for new Liferay installation it will work, but, in case you already have Liferay with many users, you need convert password from be digested by base64 to hex.

Here is a code snippet you can use for such conversion (you can add it into StartUpAction for example - but, please, backup you DB before it will be run - if by some reason this code will run twice - data in DB become invalid):

    private void convertPasswords() throws Exception {
        int count = UserLocalServiceUtil.getUsersCount();
        List<User> users = UserLocalServiceUtil.getUsers(0, count);
        Date now = new Date();
       
        for (User user : users) {
            if (user.getPasswordEncrypted()) {
                String oldPassword = user.getPassword();
               
                byte[] decodedPassword = com.liferay.portal.kernel.util.Base64.decode(oldPassword);
                String newPassword = String.valueOf(org.apache.commons.codec.binary.Hex.encodeHex(decodedPassword));
               
                UserLocalServiceUtil.updatePasswordManually(user.getUserId(),
                                                            newPassword,
                                                            true, false,
                                                            now);
            }
        }
    }

Liferay 5.2.3 Amazon EC2 Image (AMI)

2010-02-03T01:40:11Z

After I've done some performance testing for Liferay on Amazon (actually, I've simple tested our EmForge.net) - I received some requests to provide AMI with Liferay to play with.

So - here it is:

AMI ID: ami-b88c61d1
AMI ID (EU Region) : ami-0bddf67f
Architecture: i386

What's inside (in short)

Ubuntu 9.10 (Latest public for a moment with latest updates) - actually this AMI is based on official Ubuntu AMI;
Sun JDK 1.6;
Tomcat 6.0.20 as application system;
Apache 2.2.12 as front-end server (conneced to tomcat via mod_jk);
Liferay 5.2.3 (latest for a moment);
MySQL 5.1.37;

How to Use

Since it is based on standard Ubuntu AMI - I'm highly recommend to read Ubuntu EC2 Starters Guide - everything described there correct for our AMI. In general - if you have everything setup - you may run instance with command:

ec2-run-instances ami-b88c61d1 -k ec2-keypair

After it - wait some time - check your running instances - after instance started Liferay will be available at the default http port 80 on running server.

What is inside (a little bit more detailed)

Tomcat was installed from Ubuntu repository - only few settings was changed to be able to run Liferay (disabled tomcat security, increased memory, changed list of folder used as catalina libs). Liferay itself deployed into tomcat and available at /var/lib/tomcat6/webapps/ROOT

I've removed 7Coz hook and theme - so, actually you will have default liferay installation with some additional portlets (Chat, Mail, WOL)

You can login into Liferay as admin wih default username/password (test@liferay.com/test) - Please change it!

Liferay is configured to use local MySQL as DB and /var/lib/liferay as liferay home (for hot-deploy, lucene indexes and rest Liferay files)

Connection to DB done via JNDI datasource, so, connection settings you can find in /etc/tomcat6/context.xml

MySQL configured to store data in same instance memory - so, afer instance will be stopped - all data will lost. If you want to use this AMI in production - you need to change MySQL configuration to store data in ESB or to perform backups (for example via replications).

MySQL root password is changeme - please change it!

I've did not configured smtp server on the instance. So, sending emails will not work - please configure your Liferay to use some external smtp server or install/configure postfix

Feedback

Feel free to use this AMI - I hope it will be helpful for you. I will be happy to get any comments/feedack - you can leave comments for this post (required register on the site).

Liferay on Amazon EC2 Performance testing results

2009-12-07T14:45:52Z

Today, during our weekly meeting Marek told, what some of his friends reported him about performance problems, they met with Liferay - in some cases project was failed exactly because it was not possible to resolve performance these problems. But - they did not remember exact version of Liferay - it was quite long time ago.

So, we decided to make performance testing of our current emforge.net installation - run on the small Amazon EC2 instance (cheapest one) and create User Story for it

Initially we planned to write tests on JMeter, but after some investigation I found it is much simpler to use LoadStorm - free account allowed to test up to 50 users - what was quite enough for us - I'm not expecting bigger traffic in nearest time.

I will not explain here how to create test-case in LoadStrom - just will say - it is quite easy - after spending about 15 minutes for registration and test-case creation I've prepared test visited 15 pages covered most tools used in EmForge: Wiki, Blog, Forum, Activities, User Account, Tasks and so on.

System Configuration is:

Amazon EC2 small instance
Ubuntu 9.10
Java: Sun JDK 1.6
Liferay 5.2.3 with mostly default settings
Server: Tomcat 6.0.20
DB: MySQL - db stored on EBS

Quite standard configuration without any special settings done for performance by hand comparing default liferay configuration - so, we can say it is default liferay configuration.

Here is results:

	Response (average)	Requests	RPS (average)	RPS (peak)	Throughput (average)	Throughput (peak)	Total Transfer
HTML	0.167	1,674	0.8	1.5	15	29	31 MB
Other	0.077	2,092	1.0	1.8	144	269	303 MB
Total	0.117	3,766	1.7	3.3	158	298	334 MB

So, average page generation time is 0.167 seconds, static resources average time is 0.077 seconds - what is good enough I hope.

Here is some charts:

This chart shows how number of users and requests per second is growing. Maximum is about 50 users and 3 request per second

Second chart display response time

Average time is flat - that means system is not loading 100% and number of users is not affected on response time.

There are one peak of maximum time (5 secs) - but it is because at this time Garbage Collector was run.

Results

OK, it is hard to say 50 users is a serious testing, but from another side Liferay run on smallest Amazon instance without any additional settings - and, hand this amount of users without any problem.

I hope this information may be helpful for anybody, interested in liferay performance. Any comments as always welcomed!

jBPM vs Intalio in 2 words

2009-12-04T20:16:04Z

OK, not sure will able to compare them in 2 words, but will try to do it in short:

From my point of view, jBPM is great tool for developers. If you need embedded workflow solution, with easy to access to java-code, spring beans, sql, task forms from your process definition (I'm not taling about testing processes with JUnit tests!) - jBPM is defenetly right solution for you.

Intalio - is not so easy to use as embeddable solution, processes operated only (in general) with web-service, so, even for any simple operations in java (or sql) you will need to define Web-Services. But it is much more powerful (I think in all context - starting from more poewrful designer and finishing with server).

So, both are goos, but what is better - depends form your needs.

Very Promised Liferay 5.3 RoadMap

2009-12-04T11:56:03Z

Just reviewed Liferay 5.3 RoadMap and wanted to write about most expected features - but, looks like most of them are VERY expected. It should be really huge step forward.

But still- from this list I found some quite important (at least for me):

Possibility to use JPA - I think it is first - but I hope not last step for close integration with J2EE application server. In one of my previous post I pointed to weak reusing of functionality provided by J2EE server (in case then Liferay run not in Tomcat or Jetty). Looks like it should be changed in future
Maven2 - if it will be possible to use maven-based projects for Liferay development (including ServiceBuilder functionality) - it will be great!
Allow Plugins to publish JSON and SOAP web services - also good improvement - I do not like to use Ext Environment, but - exoporting services via JSON or SOAP was possible only in Ext Environment (even I do not want to change basic Liferay functionality). Hope it will work well in 5.3 even in Plugins
Integration with Gmail & Google Docs - I think it may be very helpful in some projects;
jBPM support and ability to switch to Intalio - I never used jBPM - but going to use soon. Liferay already had had some kind of integration - but not directly (via Mule) and as I understand from forums - very painful for configuration. I think it is important to have some workflow functionality in Portal out-of-box without additional configurations;
JasperReports & BIRT support - another task I planned to touch in some nearest feature - if it will be ready to use out-of-box - it should save me a lot of time!

So, let's wait for 2010 (planned release day) - I hope to get good New Year present from Liferay team :)

Autologin from custom portlet

2009-12-03T22:59:07Z

Liferay wiki has page explained how to customize login page.

But let's say you need to autologin during some user-actions in some of your portlets? And, it happens, you are working in plugin environment - so, cannot access some very helpful, but internal Liferay classes (like LoginUtil)?

here is small sample how to login - it simple redirects to the login action in standard login portlet. It looks like more a hack - but it works!

public class LoginPortlet extends JSPPortlet {
    /** process login action
     *
     * @param actionRequest
     * @param actionResponse
     * @throws IOException
     * @throws PortletException
     */
    public void login(ActionRequest actionRequest,
                      ActionResponse actionResponse) throws IOException, PortletException {
        String login = ParamUtil.getString(actionRequest, "login");
        String password = ParamUtil.getString(actionRequest, "password");
       
        // go o the login url
        ThemeDisplay td = (ThemeDisplay) actionRequest.getAttribute(WebKeys.THEME_DISPLAY);
       
        String loginUrl = td.getURLPortal() + "/web/guest/home?p_p_id=58&" +
                          "p_p_lifecycle=1&" +
                          "p_p_state=normal&" +
                          "p_p_mode=view&" +
                          "saveLastPath=0&" +
                          "_58_struts_action=%2Flogin%2Flogin" +
                          "&login=" + login +
                          "&password=" + password;
        actionResponse.sendRedirect(loginUrl);
    }
}

Liferay with JBoss Cache

2009-11-27T20:55:32Z

Last days I'm working on Liferay based project with JBossAS 5.1 used as server.

Since during redeploy I've met strange problems from ehcache used as second-level cache (in general it works well, but looks like during redeploy it cannot find some classes/methods from Hibernate) I've decided to try to use JBoss Cache as second-level cache for hibernate.

OK, first try is failed - I did not managed to make it working - but I hope I will able soon - any comments or suggestions will be very helpful!

Until now - to make Liferay to use JBoss Cache - I changed portal.properties (in my case, since I'm working in ext-environment I've edited portal-ext.properties) to contains:

hibernate.cache.provider_class=
hibernate.cache.region.factory_class=org.hibernate.cache.jbc2.JndiMultiplexedJBossCacheRegionFactory
hibernate.cache.region.jbc2.cachefactory=java:CacheManager
hibernate.cache.region.jbc2.cfg.entity=mvcc-entity
hibernate.cache.region.jbc2.cfg.query=local-query
hibernate.cache.region_prefix=tempdb

Don't have clear ideas about all these configurations - just got it from Hibernate + JBoss Cache tutorial.

His settings said to use cache from JBoss service CacheManager (got from JNDI) - what is nice - it is possible to configure cache to work (for example) clustering on server-level, and use this cache in different applications.

Also - note:

JBoss Cache and CacheManager service is included (by default) only into all configuration, so, you need run JBoss in this configuration and deploy Liferay into it
If you are using ant-script to clean/deploy ext-environment - please be sure it is not removed libraries from <jboss>/server/all/lib folder - if removed - please return them back

Why it is failed?

So, as I told - in my case it is not working. During starting Liferay I've got exception:

Caused by: org.hibernate.cache.CacheException: unsupported access type [read-write]
    at org.hibernate.cache.jbc2.entity.EntityRegionImpl.buildAccessStrategy(EntityRegionImpl.java:66)
    at org.hibernate.impl.SessionFactoryImpl.<init>(SessionFactoryImpl.java:256)
    at org.hibernate.cfg.Configuration.buildSessionFactory(Configuration.java:1327)

I've checked sources - yes, this class supports only AccessType.READ_ONLY & AccessType.TRANSACTIONAL. There is TODO to add support for READ_WRITE - but - looks like it is not implemented yet.

As I checked - for all classes in liferay mappings - read-write strategy is used for class-cache, so, it is failed to use JBoss Cache...

How to fix?

I do not have clear idea for now. I see few steps:

Understand why read-write is not supported - and is it difficult to add such support?
use wrapper - to convert requested read-write access-type into read-only - but - it need to be investigated - probably such change of strategy will produce ome problems during Liferay working - or big performance degradation.

OK, at least for now it is not top priority for me - so, leave it for some future time.

Ajax based File-Upload for PHP 5.2 under Win32 system

2009-11-21T17:15:01Z

If you will look into Google for words 'file-upload, ajax, php' you will find a lot of links to the different articles described how to do it. But there were several problems:

For versions less then 5.2 you needed to patch PHP
For version 5.2 you need to use Upload Progress PHP Extension. It was possible to use this extension without any patching for PHP, but... It doesn't worked for Win32 due to this problem
Also, on the uploadprogress extension developer blog a lot of reports about problems with this extension under windows.

We had a task to implement uploading of several files with PHP and displaying progress-bar on windows... As result - this file contains solution for this task. I will just describe major ideas:

Ajax-based upload itself implemented with using iframe (standard solution for this task);
demo.html contains all client code - it is also used prototype java-script framework - big friend for web-developers;
receiving the file is implemented in uploader.php;
For getting upload-progress information we are using Ajax.PeriodicalUpdater Prototype function on client-side and monitor.php - on server-side. Monitor.php simple get information from uploadprogress extension;
To make uploadprogress extension working under windows we had to fix it a little bit. Source code and result dll included in separate attachments here. I hope this fixes will be included by extension developer and correct version will be included into php destribution by default:
1. Removed PHP_EXPORTS define in project settings;
2. Use fopen instead of (todo). We are not very familiar with PHP extensions development, but in our case this function caused exceptino then we tried to use opened file with fprintf and fget;
3. Delete target file before renaming;
4. Cast time_t to the integers;
It is important to set No-Cache options in monitor.php - we got this code working under FireFox but it defently did not work under IE (both 6.0 and 7.0) until we did not included this option.

Ajax Based File-Upload for Java

2009-11-21T16:58:36Z

Today everything in the web is about AJAX. If you want to have a fashionable site with user-friendly interface, you have to use it. HTML hasn't been designed to be a user interface, so we have to provide some tricks and hacks to make it look like real GUI. A visual response is one of the things that HTML lacks, especially for the file uploading. Googling around i was lead to the article "AJAX file upload progress for Java using commons fileupload and prototype" by carson The last comment about the new version of file upload listener looked very promising and i decided to check it out.

Server side code

Main idea: Use commons-fileupload-1.2 ProgressListener class to monitor the upload of files dynamically.

The first thing you need is Apache Commons FileUpload, go there and grab the new binaries for version 1.2. Add it to your project. You will also need to update commons-io library to version 1.3

If you're using Maven, just add or change dependencies if they already exist in your pom.xml:

    <dependency>
      <groupId>commons-io</groupId>
      <artifactId>commons-io</artifactId>
      <version>1.3</version>
    </dependency>   
    <dependency>
      <groupId>commons-fileupload</groupId>
      <artifactId>commons-fileupload</artifactId>
      <version>1.2</version>
    </dependency>

Create a servlet that will handle file uploads:

import java.io.IOException;
import java.io.PrintWriter;
import java.util.Date;
import java.util.Iterator;
import java.util.List;
import javax.servlet.ServletException;
import javax.servlet.http.HttpServletRequest;
import javax.servlet.http.HttpServletResponse;
import javax.servlet.http.HttpSession;
import org.apache.commons.fileupload.FileItem;
import org.apache.commons.fileupload.FileItemFactory;
import org.apache.commons.fileupload.FileUploadException;
import org.apache.commons.fileupload.disk.DiskFileItemFactory;
import org.apache.commons.fileupload.servlet.ServletFileUpload;
import org.apache.commons.io.FilenameUtils;
import org.apache.commons.lang.StringUtils;
public class TestUpload extends javax.servlet.http.HttpServlet implements javax.servlet.Servlet {
    public TestUpload() {
        super();
    }      
    protected void doGet(HttpServletRequest request, HttpServletResponse response) throws ServletException, IOException {
    }      
    protected void doPost(HttpServletRequest request, HttpServletResponse response) throws ServletException, IOException {
    }                  
}

Add it to servlet mapping (web.xml), don't forget to change package name:

<servlet>
    <description>Servlet for test file uploads</description>
    <display-name>TestUpload</display-name>
    <servlet-name>TestUpload</servlet-name>      
    <!-- TODO: add here fully qualified TestUpload class name! -->
    <servlet-class>ru.emdev.EmForge.web.TestUpload</servlet-class>
</servlet>
<servlet-mapping>
    <servlet-name>TestUpload</servlet-name>
    <url-pattern>/testupload</url-pattern>
</servlet-mapping>

Create a class that implements org.apache.commons.fileupload.ProgressListener

public class UploadListener implements org.apache.commons.fileupload.ProgressListener {
    private volatile long bytesRead;
    private volatile long contentLength;
    private volatile long item;  
    public UploadListener() {
        bytesRead = 0;
        contentLength = 0;
        item = 0;
    }
    public void update(long i_bytesRead, long i_contentLength, int i_item) {
        bytesRead = i_bytesRead;
        contentLength = i_contentLength;
        item = i_item;
    }
    public long getBytesRead() {
        return bytesRead;
    }
    public long getContentLength() {
        return contentLength;
    }
    public long getItem() {
        return item;
    }
}

Add this piece of code into doPost method of the TestUpload servlet:

    protected void doPost(HttpServletRequest request, HttpServletResponse response) throws ServletException, IOException {
        response.setContentType("text/html");
       
        // create file upload factory and upload servlet
        FileItemFactory factory = new DiskFileItemFactory();
        ServletFileUpload upload = new ServletFileUpload(factory);
       
        // set file upload progress listener
        UploadListener listener = new UploadListener();
        HttpSession session = request.getSession();
        session.setAttribute("LISTENER", listener);
        // upload servlet allows to set upload listener
        upload.setProgressListener(listener);
               
        List items = null;
        FileItem fileItem = null;
        String filename = null;
        try {
            // iterate over all uploaded files
            items = upload.parseRequest(request);
            for (Iterator i = items.iterator(); i.hasNext();) {
                fileItem = (FileItem) i.next();
                if (!fileItem.isFormField()) {
                    if(fileItem.getSize()>0) {
                        // code that handle uploaded fileItem
                        // don't forget to delete uploaded files after you done with them! Use fileItem.delete();
                    }
                }
            }
            // indicate that the upload was successfull
            response.getWriter().write("upload successful");
        } catch (FileUploadException e) {
            response.getWriter().write(e.getMessage());
        } catch (Exception e) {
            response.getWriter().write(e.getMessage());
        } finally {       
            session.removeAttribute("LISTENER");
        }       
    }

This part will handle the file upload process. When request for file comes to doUpload function, we create and attach UploadListener to ServletFileUpload object. It allows us to monitor the upload progress.

And now the doGet method:

protected void doGet(HttpServletRequest request, HttpServletResponse response) throws ServletException, IOException {
    PrintWriter out = response.getWriter();
               
    HttpSession session = null;
    UploadListener listener = null;
    long contentLength = 0;
               
    if ( ((session = request.getSession()) == null) ||
         ((listener = (UploadListener)session.getAttribute("LISTENER")) == null) ||                   
         ((contentLength = listener.getContentLength())<1)) {
        out.write("");
        out.close();
        return;                         
    }
              
    response.setContentType("text/html");
    long percentComplite = ((100*listener.getBytesRead()) / contentLength);        
    out.print(percentComplite);
    out.close();           
}

In this method we can poll our upload listener for the progress of upload. In our case we want to find out how much of file has been uploaded, in percents. We just get listener that has been binded to our session, poll it for the number of uploaded bytes and calculate the result.

Client part of code

Main idea: use IFRAME as a target for uploading files; as the result, the page won't be reloaded. Use AJAX to periodically poll progress of upload and draw progress bar.

Create simple HTML page.

<HTML>
<HEAD>
</HEAD>
<BODY>
</BODY>
</HTML>

Add inside <BODY></BODY> tags iframe and form below it, exactly in that order.

<iframe id="trgID" name="uploadTrg" height="0" width="0" frameborder="0" scrolling="yes"></iframe>             
<form id="myFrm" enctype="multipart/form-data" method="post" target="uploadTrg" action="testupload" onsubmit="submitPostUsingAjax();">
    <input type="file" name="txtFile0" id="txtFile0"></input><br>
    <input type="submit" id="submitID1" name="submit" value="Upload"></input>
</form>

Here we specify the form "action" that will send file data to our servlet, and the form "target" for the iframe name. Using of form attributes method="post" and enctype="multipart/form-data" is required too. As the result, when we click on the "Upload" button, the request will be sent to "TestUpload" servlet, the response will be written to iframe, and the code in javascript function submitPostUsingAjax() will be executed.

Add hidden div for updating and the progress bar (just a table) below the form:

<div id="uploadStatusDiv" style="width:0px; height:0px; visibility:hidden;"></div>
<table id="progressBar" width="0px" style="border:1px; width:0px; background-color: red; color:black; visibility:hidden; ">
<tr><td></td></tr>
</table>

Now about monitoring part. I have used the prototype because it's a simple, clean, one-file solution for AJAX, DHTML manipulation and even more. Code also looks much more clean with it.

Download latest prototype from it's offical site. When writing this article i used version 1.5.0

Place prototype.js to the same place where your HMTL page is located and include it to the top inside <HEAD></HEAD> tags:

<script type="text/javascript" src="prototype.js">
</script>

Then add the following code:

<script type="text/javascript">
    // global object that contains Ajax.PeriodicalUpdater
    var updater; 
 
    // main function for upload monitoring
    function submitPostUsingAjax() {               
        try {
            // get upload status
            updater = new Ajax.PeriodicalUpdater('uploadStatusDiv','testupload', {
                asynchronous:true,
                frequency:1,
                method:'get',
                onSuccess:function(request) {
                    if (request.responseText.length > 1) {        
                        $('progressBar').style.visibility = 'visible';
                        $('progressBar').style.width = request.responseText + '%';
                    }
                }                                                                    
            });
        } catch(e) {
            alert('submitPostUsingAjax() failed, reason: ' + e);
        } finally {
        }                                      
        return false;
    }                                                                      
       
    // executes function after page loaded
    function addLoadEvent(func) {
        var oldonload = window.onload;      
        if (typeof window.onload != 'function') {
            window.onload = func;
        } else {
            window.onload = function() {
                oldonload();
                func();
            }
        }
    }                                      
       
    // this function will be executed after "target" IFRAME content changed
    function handleUploadFinished() {
        // stop updater manually
        if (typeof updater != 'undefined') {
            updater.stop();
            updater = null;
        }                                                                  
    }
       
    // observe IFrame object for "load" event to stop AJAX updater
    function observeFormSubmit() {
        // add event to observe upload target
        Event.observe('trgID', 'load', handleUploadFinished);
    }                              
      
    // add event handler only after page loaded
    addLoadEvent(observeFormSubmit);       
</script>

The function submitPostUsingAjax() creates Ajax.PeriodicalUpdater that will poll our servlet every second and retrives the progress of file upload in percent. The parameter "uploadStatusDiv" Ajax.PeriodicalUpdater is needed only to receive the update events.

The function addLoadEvent() is needed for attaching the listener to IFRAME object after it is created. This function was taken from Top 10 custom JavaScript functions of all time.

In the handleUploadFinished() function AJAX updater will be stopped when the upload is finished or interrupted.

After the page is loaded, IFRAME object will monitor the 'onload' event. When a user clicks on the "Upload" button, Ajax.PeriodicalUpdater will periodically poll our servlet for upload status and accordingly set the width of the progress bar. After the upload has been finished, successfully or not, IFRAME content is changed, and we will stop AJAX updater manually.

That's it, enjoy.

Implementing Captcha with JSF

2009-11-24T21:13:12Z

In registration form, used in EmForge we needed Captcha-implementation to save it from robots. I looked into internet -but did not found any ready solution for JSF-based project - so, here is one.

Captcha Generation

For generating Captcha we used JCaptcha library. It is quite powerfull library with many features - and one of them - generation and validation of captcha images.

Adding JCaptcha into Project.

Since both: EmForge and JCaptcha are Maven based projects and I found jcaptcha in the public maven repositories it was quite easy to add jcaptcha into project - just add it into projects dependency in pom.xml

<dependency>
    <groupId>com.octo.captcha</groupId>
    <artifactId>jcaptcha-all</artifactId>
    <version>1.0-RC6</version>
</dependency>

If your project is not maven-based - just download jcaptcha jar and add into your WEB-INF/libs folder.

Including Captcha into JSF Page

For rendering Captcha in JSF page we used paint2D control from rich-faces:

<h:form id="registerForm">
    <h:panelGrid columns="2" styleClass="properties-table">
    ...
    <!-- CaptchaImage -->
    <rich:paint2D id="captcha"
                  width="#{registerController.captchaWidth}"
                  height="#{registerController.captchaHeight}"
                  format="jpeg"
                  paint="#{registerController.paintCaptcha}"
                  data="#{registerController.randomString}"/>
           
    <h:outputText value="Text on Image:"/>
        <h:panelGroup>
            <h:inputText id="secureText"
                         value="#{registerController.secureText}"
                         size="40"
                         required="true">
                <emforge:captchaValidator/>
            </h:inputText>                           
            <h:message for="secureText" errorClass="errorMessage"/>
        </h:panelGroup>                         
        <h:commandLink action="#{registerController.submit}">
          <h:outputText value="Submit"/>
        </h:commandLink>
     </h:panelGrid>
</h:form>

In RegisterController we need to:

define id, used for storing generated text;
Add ImageCaptchaService;
Add Method, generated image
And accessor to secure text, entered by user
And method to generate random string:

(only part of class is here - whole code you can see here: RegisterController.java)

public class RegisterController extends BaseControllerImpl implements InitializingBean {
    public static final String CAPTCHA_ID = "EmForge";
    private ImageCaptchaService    m_imageCaptchaService;
   
    private String             m_secureText;
    public void setImageCaptchaService(ImageCaptchaService i_imageCaptchaService) {
        m_imageCaptchaService = i_imageCaptchaService;
    }
   
    public String getSecureText() {
        return m_secureText;
    }
    public void setSecureText(String i_secureText) {
        m_secureText = i_secureText;
    }
   
    protected BufferedImage generateCaptcha() {
        try {
            return m_imageCaptchaService.getImageChallengeForID(CAPTCHA_ID);
        } catch (Exception ex) {
            logger.error("Cannot generate captcha image", ex);
            return null;
        }
    }
   
    public int getCaptchaWidth() {
        return 200;
    }
   
    public int getCaptchaHeight() {
        return 100;
    }
   
    /** Paints Captcha Image
     *
     * @param g2d
     * @param obj
     */
    public void paintCaptcha(Graphics2D g2d, Object obj) {
        BufferedImage secureImage = generateCaptcha();
       
        try {
            g2d.setClip(0,0,secureImage.getWidth(), secureImage.getHeight());
            g2d.drawImage(secureImage, 0,0, null);
        } catch (Exception ex) {
            logger.error("Cannot generate captcha image", ex);
        }
    }
    /** Generates Random Text for displaying on the image */
    public String getRandomString() {
        String str=new  String("QAa0bcLdUK2eHfJgTP8XhiFj61DOklNm9nBoI5pGqYVrs3CtSuMZvwWx4yE7zR");
         StringBuffer sb=new StringBuffer();
         Random r = new Random();
         int te=0;
         for(int i=1;i<=6;i++){
             te=r.nextInt(62);
             sb.append(str.charAt(te));
         }
        
         return sb.toString();
    }

Since JCaptcha supports integration with spring-framework, ImageCaptchaService may be defined as spring bean and passed int registerController as param (part of beans configuration):

<beans>
    ...
    <bean id="imageCaptchaService" class="com.octo.captcha.service.image.DefaultManageableImageCaptchaService" scope="session"/>
       
    <bean id="registerController" class="ru.emdev.EmForge.security.RegisterController" scope="request">
        ...       
       
        <property name="imageCaptchaService" ref="imageCaptchaService"/>
    </bean>
    ...
</beans>

Now, you will able to generate and render the captcha image, as well as enter the text.

Validating Captcha

For validating Captcha we implemented special validator - CaptchaValidator

(assigned to secureText input):

package ru.emdev.EmForge.security.web.validator;
import javax.faces.application.FacesMessage;
import javax.faces.component.UIComponent;
import javax.faces.context.FacesContext;
import javax.faces.validator.Validator;
import javax.faces.validator.ValidatorException;
import javax.servlet.http.HttpServletRequest;
import org.apache.commons.lang.StringUtils;
import org.springframework.web.context.WebApplicationContext;
import org.springframework.web.context.support.WebApplicationContextUtils;
import ru.emdev.EmForge.security.RegisterController;
import com.octo.captcha.service.image.ImageCaptchaService;
/**
 * Validates inputed name for a project
 *
 * @author mchirkov
 */
public class CaptchaValidator implements Validator {
    public static final char[]     INVALID_CHARS                     = new char[] {'*', '@', '/', '\\'};
    public static final int        MAX_NAME_LENGTH                 = 50;
    public static final String     VALUEREQUIRED_ERROR_MESSAGE     = "Value is required.";
    public static final String     TEXTNOTNATCH_ERROR_MESSAGE         = "Text not match text on image";
   
    public void validate(FacesContext context, UIComponent component,
            Object value) throws ValidatorException {
        String val = StringUtils.strip((String)value);
        if (StringUtils.isEmpty(val)) {
            throw new ValidatorException(
                new FacesMessage(VALUEREQUIRED_ERROR_MESSAGE, VALUEREQUIRED_ERROR_MESSAGE)
            );
        }
       
        HttpServletRequest request = (HttpServletRequest)context.getExternalContext().getRequest();
        WebApplicationContext appContext =
            WebApplicationContextUtils.getWebApplicationContext(request.getSession().getServletContext());
        assert appContext != null;
       
        ImageCaptchaService imageCaptchaService = (ImageCaptchaService)appContext.getBean("imageCaptchaService");       
       
        if (!imageCaptchaService.validateResponseForID(RegisterController.CAPTCHA_ID, val)) {
            throw new ValidatorException(
                new FacesMessage(TEXTNOTNATCH_ERROR_MESSAGE, TEXTNOTNATCH_ERROR_MESSAGE));
        }
    }
}

ImageCaptchaService, stored inside itself the text, generated in image, so, here, we simple get this bean from Spring Context, and validate it

So, now we only need to declare validator in faces-config.xml:

    <validator>
        <validator-id>captchaValidator</validator-id>
        <validator-class>ru.emdev.EmForge.security.web.validator.CaptchaValidator</validator-class>
    </validator>

and define validator tag in our facelets taglib (emforge.taglib.xml)

    <tag>
        <tag-name>captchaValidator</tag-name>
        <validator>
            <validator-id>captchaValidator</validator-id>
        </validator>   
    </tag>

What is all. Enjoy!