All posts by julienprog

Experienced developer in Java, J2EE and C#. I like to unit tests systematically code (Java,C#, Javascript, bash shell, etc..). I believe all part of a project should be automatically tested. Good experience of maven,ant,shell to achieve continuous integration.

Exit from script if a process failed


I often found myself doing the same code which stopped a shell script if the returned value of a command is different from 0. This function is often use to do end to end testing for application, creation of database and so one , often for dev op purpose. It helps me to detect problems faster.


So for example i want the script ./ to exit if the compilation failed (so for example if ant command failed).

This is the result expected :

/work/build.xml:286:  Compile failed; see the compiler error output for details.
Total time: 26 seconds
Error in the last command executed with code 1

This is the code i need to the :

Modify the

As said in this link you need to have on top of the file

. /work/ 


cd /work
check_or_terminate $ant_return_code

do as well :

chmod +x

Create a function in shell script

The contained the following code to detect the return code of a command :

# set variables 
# Purpose: exit from shell if the value input is different from 0
# Arguments:
#   $1 -> output of a command line example output of ant command
#   if 0 value -> we continue the script
#   if different from 0 -> stop the script
function check_or_terminate() 
    local return_code="$@"
    if [ $return_code != 0 ]
	echo "Error in the last command executed with code $return_code"
	exit 1;

When classes are not generated by eclipse


Recently I moved from maven to ant. Since the move I have many problems with classpath in Eclipse. Maven eclipse plugin solved lot of classpath issues for you. I was told the Ant eclipse plugin is unreliable so I am not using it. I am using Eclipse Oxygen 4.7.2 .


When running Junit test , i noticed the new modifications on classes under test were not debugged. I had to compile the code with ant which takes 1 minute . Furthermore I had to add the classes output manually in the project classpath.


Identifying why classes are not generated

If the project has exclamation mark it is very likely that eclipse cannot generate compiled classes.


The eclipse tab “problems” was really helpful to understand the root cause of the problem. In my case some paths in the Eclipse Java Classpath had an “unbound variable”. Because of this issue, Eclipse would not generate the classes.

The fix for my specific issue

I had to add a variable in Java classpath project. Click on “Add variable” in the libraries tab like here :


Click new in “Variable selection” in order to add the variable which is unbounded.

Further Troubleshooting

Clean project

If the problem persists after fixing the issue, try cleaning up the project.

More investigation possible

This following link may be useful if this article cannot help you. However contrary to the tip in this link, the tab “problems” was helpful for my issue. Look for logs in [root_eclipse_workspace]/.metadata/.log .

Create PL/SQL script handling errors


The aim of this article is to create a shell script calling a pl/sql script. These two scripts should handle errors an generate logs.


This article will present the shell script and then the pl sql script.

1. The shell script

This shell script will show how to manage inputs, display information and handle errors from pl sql script.

cd "$dir"

# Prints the usage
    printf "$0 -a input1 -b input2 \n"
    printf "Script called.\n"
    printf "\t-a input1\t: my first input\n";
    printf "\t-b input2\t: my second input\n";
    printf "\t-h\t\t: shows usage\n";
    printf "\n"
    exit 1;

# Processes the options on the command line

# user of the database

# pass of the database
# service can be found in tnsnames.ora

while getopts "a:b" name; do
      case $name in
      a) INPUT1=$INPUT1 ;;
      b) INPUT2=$INPUT2 ;;   
      h) usage ;;
      \?)  usage ;;

if [ -z "$INPUT1" ]; then
    printf "ERROR: input 1 is not specified\n"

if [ -z "$INPUT2" ]; then
    printf "ERROR: input 2 is not specified\n"

echo "call pl sql script"
sqlplus -S $USER/$PASS@$SERVICE @mypl_sql_script.sql

if [ $sql_return_code != 0 ]
echo "error in the script"
echo "Error code $sql_return_code"
exit 0;
echo "call pl sql script finished"


processInputs "$@"

2. The pl sql script

This is a simple pl sql where a rollback is done if any errors occur. Also a log script is generated. This script is called by the previous shell script.

spool log.txt
set serveroutput on
VARIABLE  v_return_code number;
    vrequest varchar2(1000);
    vrequestInsert varchar2(1000);

	vrequest := 'SELECT * FROM CUSTOMER_USER WHERE NAME= ''john''';
       vrequestInsert := 'Insert into GROUP_USER (ID,VERSION,TYPE,NAME) values (''11'',''0'',''blah'',''admin'')';

execute immediate vrequestInsert ;
        execute immediate vrequest;
execute immediate vrequestInsert ;


     when others then
        dbms_output.put_line('KO - ' || vrequest || ' - ' || sqlerrm);  

spool off

EXIT :v_return_code;

3. Some DDL Statements have auto commit after exceution

For some DDL statements the rollback mechanism does not work.

DDL statements do commit (ALTER/CREATE etc) and this will happen even if something failed. If you’re running EXECUTE IMMEDIATE like dynamic statement that runs a DDL, this will also commit your transaction.

I would add that DROP as well does auto commit after execute immediate.

SSH connection from Jenkins


How to access remotely to a machine from jenkins without storing password ?

– We were using maven ssh plugin to remotely access a machine. However this solution was not satisfying because the password of the username was stored inside a file. We should not expose the password in a file.


The solution chosen is using the Credentials Jenkins plugin

Configure the job to pass the SSH private-key using an SSH agent

This allows the user to securely provide access to remote resources (using SSH authentication) without exposing the SSH private-key material to the job.

1. Create key private and key public :

In a linux machine , generate a key pair-value pulic and private , with the following command

It generates, in the folder .ssh, two files : id_rsa et
o Id_rsa is the private key , it should be added on Jenkins.
o is the public key, it should be added on the distant server to be recognized by Jenkins.

2. Add the public key on the target distant server

We put the public key in the folder .ssh of the home of the user with wich we will access. It is added in the files called authorized_keys of the hidden folder “.ssh”

3. Add the private key in Jenkins

Go to manage Credentials and follow steps from here to add a username which can access the distant machine :

Select “SSH Username with private key”

4. Modify The build to allow SSH Agent

Choose the username you will like to use to access to the distant machine.


Following this, it is possible to access remotely with ssh from Jenkins with no login/password. It applied only to the machine which has username with public key in .ssh/authorized_keys folder.


Ssh Credentials Management :

CSRF prevention for Wicket


This article will show you how to fix a CSRF vulnerability on a website built with Wicket framework. For simplicity this article will focus on fixing a CSRF vulnerability on one form only rather than all forms.

I already talked briefly about CSRF prevention here :

First of all i will talk about the CSRF prevention of the Wicket framework and then about the solution i chose based on this article :

I would suggest to use Spring security to perform a global solution from the start :

Is Wicket built-in CSRF prevention efficient ?

Wicket has some built-in listener to detect CSRF and cancelled request :

This solution checked if the origin of the request and the requested URL have the same domain. If it is not true then it is considered as a CSRF attack.
Few problems with this solution:

  • Legitimate request from the same domain can be considered as a CSRF attack. For example the request url was like : and referer/origin .It seems because the two url are not identical it was considered as an attack.
  • To overcome this problem I allowed all the requested URL coming from the host domain to bypass the CSRF prevention :

    I don’t like this solution because we desactivate CSRF prevention for the website domain. All POST/Ajax request are authorised from the website domain.
    Second problem is that i need to put the name of the domain in configuration file.

  • What if the sender does not have Origin/Referrer in the POST request? It happens some legitimate client do not have it. It is considered a CSRF attack in that case.

  • For all these reasons i gave up on the Wicket built-in CSRF prevention technique .

    Token CSRF prevention solution

    I started doing the classic token solution to prevent CSRF . I followed this principle :
    Except in my case i applied anti CSRF measure only to one form. This solution can be improved with the previous article to make the solution more generic to all forms.
    This solution is decomposed like this :

  • Modification of the wicket form to add a anti CSRF token
  • Added a hidden field to the form to identify the form exactly.
  • Add the same anti CSRF token in HttpRequest
  • Add a Filter in the application to filter every request of the form
  • 1. Modification of the wicket form to add a anti CSRF token

    I modified the wicket form to add hidden input “csrftoken”. The input hidden was created in Java in the same time as the token CSRF in HttpRequest. Therefore i skipped the class LoadSalt from the previous article.

    I modified a Wicket form based on this link :

    public class SecureForm<T> extends Form<T> { 
    private static final String TOKEN_NAME = "SECURE_FORM_TOKEN";
            private static final String TOKEN_CSRF_SESSION = "tokenCSRF";
            private String token;
            protected void onBeforeRender() {
                this.token = UUID.randomUUID().toString();
            public void onComponentTagBody(MarkupStream markupStream, ComponentTag openTag) {
                // render the hidden field
                if (isRootForm()) {
                    AppendingStringBuffer buffer = new AppendingStringBuffer(
                            "<div style=\"display:none\"><input type=\"hidden\" name=\"");
                            .append("\" id=\"")
                            .append("\" value=\"")
                            .append("\" /></div>");
                // CRF prevention applied to only one form "SpecificForm"
                if(panelConf.isSpecificForm())) {
                       String identifyForm = "identifyForm";
                       AppendingStringBuffer buffer = new AppendingStringBuffer(
                                "<div style=\"display:none\"><input type=\"hidden\" name=\"");
                                .append("\" id=\"")
                                .append("\" value=\"")
                                .append("\" /></div>");
                HttpServletRequest request = ((HttpServletRequest) getRequest().getContainerRequest());
                <strong>request.getSession().setAttribute(TOKEN_CSRF_SESSION, this.token);</strong>
                // do the rest of the processing
                super.onComponentTagBody(markupStream, openTag);
  • As you can in the previous code I have added a hidden field “identifyForm” to the form to identify it uniquely. It will be used later on in the filter class in order to apply CSRF prevention only to this form.
  • I have added the anti CSRF token in HttpRequest at the following line
    request.getSession().setAttribute(TOKEN_CSRF_SESSION, this.token);

    This token has also been added in the hidden field SECURE_FORM_TOKEN

  • 2. Add a Filter in the application to filter every request of the form

    My second task was to add filter similar to ValidateSalt of dzone link. Because the CSRF protection is on one form only, i have to check that a request was sent for the specific form before checking csrf token.

    public class ValidateSalt implements Filter  {
    	private static final Logger LOG = LoggerFactory.getLogger(ValidateSalt.class.getName());
    	private static final String TOKEN_NAME = "SECURE_FORM_TOKEN";
    	private static final String TOKEN_CSRF_SESSION = "tokenCSRF";
        public void doFilter(ServletRequest request, ServletResponse response, FilterChain chain)
            throws IOException, ServletException {
            // Assume its HTTP
            HttpServletRequest httpReq = (HttpServletRequest) request;
            // Get the salt sent with the request
            String sessionTokenCSRF = (String) httpReq.getSession().getAttribute(TOKEN_CSRF_SESSION);
            LOG.debug("tokenCSRF : " + sessionTokenCSRF);
            String token =(String) httpReq.getParameter(TOKEN_NAME);
            LOG.debug("token : " + token);
            String identifyForm= (String) httpReq.getParameter("identifyForm");
            LOG.debug("identifyForm: " + identifyForm);
            if(identifyForm!= null) {
    	        if (sessionTokenCSRF != null &&
    	        		token != null && token.equals(sessionTokenCSRF)){
    	            chain.doFilter(request, response);
    	        } else {
    	        	LOG.error("Potential CSRF detected!! Inform a scary sysadmin ASAP.");
    	            // Otherwise we throw an exception aborting the request flow
    	            HttpServletResponse httpResponse = (HttpServletResponse) response;
            } else {
            	chain.doFilter(request, response);
        public void init(FilterConfig filterConfig) throws ServletException {
        public void destroy() {

    In this example, first i check if the request contain the identity of the form. Then i checked for the CSRF token.

    I have also modified web.xml to add the filter :

      <!-- Filter Example 1 -->
      <!-- end Filter Example 1 -->

    How to test for CSRF vulnerability ?

    First of all authenticate into you webapp, then execute a form of your webapp within a html file. The html file contains a copy of the form you want to test. The execution of the html file should fail if there is a CSRF prevention. Example of html file :

    <body onload='document.CSRF.submit()'>
    <meta charset="utf-8"/>
    <form action='' method='POST' name='CSRF'>
    	<input type='hidden' name='id' value=''>
    	<input type='hidden' name='identifyForm' value='identifyForm'>
    	<input type='hidden' name='variable1' value='0'>	
    	<input type='hidden' name='variable2' value='1'>	
    	<input type="buttonSearch" value="Submit">

    I put a link to owasp testing here :

    Create Unit Test for Drools Ruleflow


    Drools is a Business Rules Management System (BRMS) solution. For more information :

    The aim of this article will provide information how to realize a Unit test of drools ruleflow using Junit.

    Create a ruleflow

    This link will show you how to create a ruleflow with Drools :

    Let’s say the ruleflow name is “my_basic_ruleflow.rf”. The processName is my_basic_ruleflow.

    Create your Unit Test

    This is a simple Junit Test to launch the processName “my_basic_ruleflow”.
    It will launch the ruleflow and executes all part of the ruleflow.

    import org.drools.SystemEventListenerFactory;
    import org.drools.agent.KnowledgeAgent;
    import org.drools.agent.KnowledgeAgentConfiguration;
    import org.drools.agent.KnowledgeAgentFactory;
    import org.drools.definition.KnowledgePackage;
    import org.drools.logger.KnowledgeRuntimeLogger;
    import org.drools.logger.KnowledgeRuntimeLoggerFactory;
    import org.drools.runtime.StatefulKnowledgeSession;
    public class TestRuleflow {
        public void TestSimpleRuleflow() {
            try {
                final StatefulKnowledgeSession session = kagent.getKnowledgeBase().newStatefulKnowledgeSession();
                String processName = "my_basic_ruleflow";
         } catch (Exception e) {
                LOG.fatal("Exception : " + e);

    The function init() will set properties and load drl and ruleflows. This function will also scan for files and directories with a specific interval.

       public final void init() {
            SystemEventListenerFactory.setSystemEventListener(new Log4jSystemEventListener());
            try {
                final ResourceChangeScannerConfiguration scannerConf = ResourceFactory.getResourceChangeScannerService().newResourceChangeScannerConfiguration();
                scannerConf.setProperty("drools.resource.scanner.interval", scanInterval);
                final KnowledgeAgentConfiguration agentconf = KnowledgeAgentFactory.newKnowledgeAgentConfiguration();
                agentconf.setProperty("drools.agent.scanDirectories", "true");
                agentconf.setProperty("drools.agent.newInstance", "true");
                agentconf.setProperty("drools.agent.scanResources", "true");
                agentconf.setProperty("drools.agent.monitorChangeSetEvents", "true");
                kagent = KnowledgeAgentFactory.newKnowledgeAgent("my_appli", aconf);
                String configurationName = "file_rules.xml";
            } catch (Exception e) {

    The configuration file_rules.xml contains all the drl,functions and ruleflows. This configuration example is used only for our unit tests.


    In production environement we use a PKG ressource. Essentially it is an archive containing all files (DRl,RF,function,etc…).


    Session ID in the URL : is it a vulnerability ?


    I noticed at authentication of several JAVA web applications, the Session ID attached at the url like this…

    When first authenticated, the website reveals in the URL a sensitive information “the session ID”. This is a security risk according to OWASP reference.
    Indeed if an attacker get the session ID it can lead to the vulnerability of session fixation.

    Scanner like Acunetix will detect is as a security risk too:

    Why the session id is passed to the URL ?

    According to this post, it is by design of JavaEE:

    This isn’t a bug, it’s by design. When a new session is created, the server isn’t sure if the client supports cookies or not, and so it generates a cookie as well as the jsessionid on the URL. When the client comes back the second time, and presents the cookie, the server knows the jsessionid isn’t necessary, and drops it for the rest of the session. If the client comes back with no cookie, then the server needs to continue to use jsessionid rewriting.
    You may not explicitly use cookies, but you do implicitly have a session, and the container needs to track that session.

    What is a Session Fixation ?

    Session Fixation is an attack that permits an attacker to hijack a valid user session.

    This vulnerability is part of Top 10 2013-A2-Broken Authentication and Session Management

    How to test for Session Fixation ?

    What is a Session ?

    HTTP protocol and Web Servers are stateless, what it means is that for web server every request is a new request to process and they can’t identify if it’s coming from client that has been sending request previously.

    But sometimes in web applications, we should know who the client is and process the request accordingly. For example, a shopping cart application should know who is sending the request to add an item and in which cart the item has to be added or who is sending checkout request so that it can charge the amount to correct client.

    This is a link to Servlet specification to understand better the session scope in Java :

    SRV.7.3 Session Scope
    HttpSession objects must be scoped at the application (or servlet context) level. The underlying mechanism, such as the cookie used to establish the session, can be the same for different contexts, but the object referenced, including the attributes in that object, must never be shared between contexts by the container.


    Validate Session Id on server side
    For sometimes i thought that cookies or hidden input fields is the solution against the “session ID” in the url. According to the link below it is a limited solution. Even if it is hard to copy paste cookies and hidden fields it is still possible to retrieve the Session ID information with special tools on unencrypted website.

    Depending of the website it is possible that the sessionId on the URL is not a security risk. If the web application is well designed and the session ID is validated on server side, this is not a problem :

    The best practice , in all case is to validate on the server side.
    Other solutions are possible :

    Use Filter to get rid of Session ID in URL

    Here is an example of using a Filter to get rid of session Id in the URL :

    I believe this solution is not necessary if good validation of session ID is done on the server Side.

    Spring Security Framework

    Probably the best solution for JavaEE is to use well tested framework like spring security framework. It is also a default security against other common vulnerabilities.
    The authentication form using spring security provide high security against common vulnerabilities :

    The SecurityConfig will:

    Require authentication to every URL in your application

    Generate a login form for you

    Allow the user with the Username user and the Password password to authenticate with form based authentication

    Allow the user to logout

    CSRF attack prevention

    Session Fixation protection

    Security Header integration

    HTTP Strict Transport Security for secure requests

    X-Content-Type-Options integration

    Cache Control (can be overridden later by your application to allow caching of your static resources)

    X-XSS-Protection integration

    X-Frame-Options integration to help prevent Clickjacking

    Integrate with the following Servlet API methods




    HttpServletRequest.html#login(java.lang.String, java.lang.String)