The power of calculated fields in ADFbc

Lately I saw a couple of posts on the OTN JDev & ADF forum where users tried to add redundant data into their data model and store it to the DB table. One common use case here is to have the result of a calculation as an attribute of a table.

In general you should be very careful when doing this. This is error prone and will you get into trouble almost every time. If you do add an attribute for such a calculation to a table in the DB, you have to think of the integrity of the data. Let’s look into the use case and the integrity problem.

Use Case

We have a table in the DB which holds start and end for multiple data types like integer, data and timestamp:

Selection_719

We use the different start and end attributes to calculate the difference between start and end.

We do have the option to add attributes to the table and calculate the difference using a trigger in the DB each time the data is inserted or updated. Problem here is that the user will see the result only after the insert or update is done. For web pages this isn’t a good design.

Another option is to add the fields but do the calculation in the business component layer in ADFbc and store them in the DB together with all other changes done to the data. The your see the calculation, but other applications won’t see them until you store the record.

Problem with storing redundant data in a DB table

Both options have one flaw. When you store the result of a calculation in the DB, what happens if someone, person or program, changes one of the attributes used in the calculation?

Assume STARTINT is set to 5, ENDINT is set to 10. The result of the calculation is 5. This result we store in an attribute in the DB table. Now a bad programmer who does not know about the calculation, changes the ENDINT to 15 and commits the change.

When the other program looks at the data again the data is inconsistent. Which of the values is correct? The result? The STARTINT value? The ENDINT value? Or is the calculation simply wrong?

In this simple use case it’s fairly easy to find the problem. In more complex use cases where other workflows depend on the numbers it’s not as easy.

This leads to the solution shown in this post: don’t store results of calculations in the DB if possible. Do the calculation when they are  needed.

There are cases where storing the result would be the better way to archive the whole use case, but this has to be decided on the use case and weighted against the complications. Most simple use cases don’t need to store the results and should not.

The remainder of this post we see how to implement such calculated fields using ADFbc.

Implementing calculated fields in ADFbc using Groovy

We start with creating a new Fusion Web Application and building the ‘ADF Business Components from a Table’. The sql script to create the table is

CREATE TABLE "HR"."CALCULATION"
 ( "ID" NUMBER(*,0) NOT NULL ENABLE,
 "STARTINT" NUMBER(*,0),
 "ENDINT" NUMBER(*,0),
 "STARTTIME" DATE,
 "ENDTIME" DATE,
 "STARTTIMESTAMP" TIMESTAMP (6),
 "ENDTIMESTAMP" TIMESTAMP (6),
 CONSTRAINT "CALCULATION_PK" PRIMARY KEY ("ID")
 );
REM INSERTING into CALCULATION
 SET DEFINE OFF;
 Insert into CALCULATION (ID,STARTINT,ENDINT,STARTTIME,ENDTIME,STARTTIMESTAMP,ENDTIMESTAMP) values ('1','1',null,to_timestamp('24-DEZ-15','DD-MON-RR HH.MI.SSXFF AM'),to_timestamp('26-DEZ-15','DD-MON-RR HH.MI.SSXFF AM'),null,null);
 Insert into CALCULATION (ID,STARTINT,ENDINT,STARTTIME,ENDTIME,STARTTIMESTAMP,ENDTIMESTAMP) values ('2','4','6',to_timestamp('31-DEZ-15','DD-MON-RR HH.MI.SSXFF AM'),to_timestamp('05-JAN-16','DD-MON-RR HH.MI.SSXFF AM'),null,null);

We use the HR DB schema to add the table, but it can be added to any schema you want. The CALCULATION table consists of some start and end values of different types to later show how to work with them. To work with the table we add two records resulting in the following data

Selection_720.jpg

I don’t show the steps to create the basic application from the wizards as the application is available via the link GitHub base application.

Once you downloaded and unzipped the workspace you should see the base application as it will be created by following the wizard.

Selection_721

The first step is to create a transient field in the Calculation EO to hold the result of the calculation of the difference of STARTINT and ENDINT. The difference here  is, that we store the result in the EO as transient attribute which is not stored into the DB.

The real work is shown in the third image above ‘edit expression…’. Here we enter a Groovy expression to calculate the difference between STARTINT and ENDINT as

if (Endint == null) 
  {return 0} 
else 
  {return Endint-Startint}

The Groovy expression uses the attribute names from the EO not the ones from the DB table. First we check if the Endint is given, if not we return 0. If there is an Endint we return the (Endint-Startint).

We then add notifications to the calculated attribute whenever the attributes Startint or Endint change to recalculate the Durationint attribute (lower half of the dialog). Next we set the AutoSubmit  property of the Startint and Endint attributes to true to make sure we get the new values when we calculate the result.

Finally we add the new calculated attribute to the VO. We can now test the application module using the application module tester:

We now add a index page to the View Controller project to add an UI to the application. We can just drag the CalculationView1 and drop is as an ADFForm with navigation and submit onto the page.

In the resulting form we set the Startint and Endint fields to autosubmit=’true’ to make sure the new values are submitted. As the Durationint field isn’t updateble we set it to read only.

Running the application will show you

The application in this state can be downloaded from GitHub (feature/calculated_int_field).

To show that this can be done with other data types we can use the other attributes of the table. As the way to do this is the same I spare to give detailed instructions. You can download the final application from GitHub (final).

All samples yre using the HR DB schema and table called CALCULATION. The needed SQL code to create the table and to insert data to the table is posted in here.

Advertisements

JDev 12.2.1: Remote Task Flows in Action

The new JDeveloper version 12.2.1 is just out and has a lot of new features to investigate. In this post we see how remote task flows work. Yes, they are finally here and they are working. At least if you install a patch available from support.oracle.com.
The downloadable version on JDev 12.2.1 has a small bug which prevents you from running remote task flows (refer to https://community.oracle.com/thread/3816032). Support and the dev team quickly delivered a patch for this. To get the patch, open a service request and ask for a patch for bug 22132843.

Let’s start. We need two applications to show how remote task flows are implmented. One is the remote task flow producer, one consumes the remote task flow. An application can be both, producer and consumer. For this sample we keep it simple and define one app as producer and one as consumer.

Producer Application
This application is really simple as it consists of only one page and one task flow which shows the departments and its employees of the HR DB schema.

Remote Task Flow Producer Application

Remote Task Flow Producer Application

The image above shows the running application stand alone. The single page has the header and a simple task flow beneath it to show the departments and their employees.


There are two properties to set in the task flow.
1) in must be remote invocable
2) the transaction must be isolated

Next we have to make the application aware that it should be a remote task flow producer. For this we edit the projects properties and select the ‘ADF Task Flow’ node.

Project Properties for Producer Application

Project Properties for Producer Application


Please note is the second checkbox selected which allows anonymous users to access the remote task flow. This should not be used in a production environment as this would allow anybody to access the task flow. The doc shows how to secure the access to a remote task flow (see link below).

These settings will add a special servlet and a servlet filter to the web.xml file of the application.

There are more things to consider which you find in the docs at How to Configure an Application to Render Remote Regions

That’s it for the simple producer application.

Consumer Application
The second application is simple too. Here we use a single page which again uses the HR DB schema to show the departments as an editable table in a panel splitter. On the right of this we show the remote task flow of the producer application.

Consumer Application

Consumer Application


In the image above the remote task flow isn’t visible as it is not added at the moment.
To make the remote task flow available we need to run the producer application. Here we have to be careful if we try this out using the embedded WebLogic Server. As only one application can be started in debug mode, we need to start the producer application as a normal application.
Run Producer Application

Run Producer Application

In the consumer application we set the project properties for the ADF Task Flow to allow it to consume remote task flows

Consumer Application Project Properties

Consumer Application Project Properties

Now we create a remote task flow connection. Open the resource palette and select to create a ‘Remote Region Producer…’ from the IDE connections.
Here we fill in the needed info like the path to the remote producer servlet which will get us the names of all remote task flows the application holds. To access the remote task flow we define the URL endpoint


The details about what to fill in are again from the doc.

In the consumer application we now open the one page and drag the remote task flow from the ressource palette onto the page and drop it in the right hand splitter

Drop Remote Region in Consumer Application

Drop Remote Region in Consumer Application


This will give us the known image in design mode as if you use a normal region
Consumer Page

Consumer Page


We are ready to run the consumer application and get
Running Consumer Application

Running Consumer Application

Nice!

You can download the sample application from GitHub:
Consumer Application
Producer Application
Both application use the HR DB schema. Make sure to adjust the DB connection to point to your db server.

Export to Excel enhancements in JDeveloper 11.1.1.9.0 and JDeveloper 12.1.3

In the current JDeveloper version 11.1.1.9.0 and 12.1.3 the af:exportCollectionActionListener got enhanced by options to filter the data to export.

Enhanced options of exportCollectionListener

Enhanced options of exportCollectionListener


The option this blog talks about is the one marked, the FilterMethod. The ducumentation for 12.12 Exporting Data from Table, Tree, or Tree Table does not reveal too much about how to use this FilterMethod.
The sample we build in this blog entry shows how the FilterMethod can be used to filter the data to be exported to excel.
In older version of JDev you hadto use a trick to filter the data which was downloaded from a table see Validate Data before Export via af:exportCollectionActionListener or af:fileDownloadActionListener. The new property of the af:exportCollectionActionListener allows to filter the data without using the trick.
The sample just load the employees table from the HR DB schema and shows it in a table on the screen. In the toolbar we add a button which has the af:exportCollectionActionListener attached.
Running application

Running application


Below is the page code of the toolbar holdign the export button:

                            <f:facet name="toolbar">
                                <af:toolbar id="t2">
                                    <af:button text="Export to Excel" id="b1">
                                        <af:exportCollectionActionListener type="excelHTML" exportedId="t1" filename="emp.xsl" title="Export"
                                                                           filterMethod="#{ExportToExcelBean.exportCollectionFilter}"/>
                                    </af:button>
                                </af:toolbar>

The filterMethod of the af:exportCollectionActionListener points to a bean method exportCollectionFilter in a request scoped bean ExportToExcelBean. The method gets called for each cell of the table which gets exported.

    /**
     * This method gets called for each cell which is to be exported.
     * It can be used to filter data to be exported. In this case salary values > 6000 are not exported
     * @param uIComponent component of the cess which gets to be exported
     * @param exportContext context of the exported data (holds e.g. file name, character set...)
     * @param formatHandler format to be exported
     * @return true if cell value is exported, false if not
     */
    public Boolean exportCollectionFilter(UIComponent uIComponent, ExportContext exportContext, FormatHandler formatHandler) {
        if (exportContext.isFirstInRow()) {
            count++;
            _logger.info("Start a new Row " + count);
        }
        _logger.info("Export Collection UIComponent: " + uIComponent.getId());
        if (uIComponent instanceof RichOutputText) {
            RichOutputText rot = (RichOutputText) uIComponent;
            Object val = rot.getValue();
            String headerText = "";
            UIComponent component = rot.getParent();
            if (component instanceof RichColumn) {
                RichColumn col = (RichColumn) component;
                headerText = col.getHeaderText();
            }
            StringBuilder sb = new StringBuilder();
            sb.append("Name: ");
            sb.append(headerText);
            sb.append(" Value: ");
            sb.append(val);
            _logger.info(sb.toString());
            // check if the salary is greater than 6000
            if ("Salary".equals(headerText)) {
                if (((BigDecimal) val).intValue() > 6000) {
                    // if yes return false so that the value isn't exported
                    _logger.info("Skip Vals > 6000");
                    return false;
                }
            }
        }

        return true;
    }

The method gets the uiComponent which represents the current cell to be exported, the ExportContext and the FormatHandler for the export. The ExportContext hold information about the filename, title, the used character set and status information about the row and cells currently exported. The status can be used to find out is a new row just starts to be exported ro is a cell is part of a span of cells. In the sample we use this information to print a log message for each row exported.
The FormatHandler is used to generate the document to be exported and the data in it. I did not find a way to use my own handler and there is no documentation about how to use another handler, so we leaf this as is for the moment.
In the sample method we like to filter the employee data in a way, that salaries greater than 6000 are not exported to the resulting file. As the method is called for each cell, the first thing to find out is which cell currently used. In lines 15-29 we use the current UIComponent to find out which column we are in. In lines 31-37 we check the salary column. In case the salary value is greater than 6000 we return false as this will trigger that the cell value is not exported. If the salary is below or equal to 6000 we return true and the cell value is exported.
Below we see the result we get if we export the table without the filterMethod set:

Exported table without filter

Exported table without filter


and the result with the filter method set:
Exported table with filter

Exported table with filter

You can download the sample application which was build using JDeveloper 12.1.3 and the HR DB schema from GitHub.

dvt:treemap showing node detail in popup

This post describes how to implement an dvt:treemap which shows a af:popup when the user clicks on a detail node in the map.
The documentation of the dvt:treemap component tell us that the dvt:treemapnode supports the af:showPopupBehaviortag and reacts on the ‘click’ and ‘mouseHover’ events.
This is part of the solution and allows us to begin implementing the use case. We add an af:showPopupBehavior to the nodes we want to show detail information for.

After creating a default Fusion Web Application which uses the HR DB schema, we begin with creating the data model for the model project. For this small sample the departments and employees tables will be sufficient.


The views are named according to their usage to make it easier to understand the model. This is all we need for the model.

Let’s start with the UI which only consist of a single page. The page has a header part and a center part. In the center area we build the treemap by dragging the Departments from the data controls onto the page and dropping it as treemap. After that, in the dialog we specify the first level of the map to be the departmentId (which shows the department name as the label) and the for the second level we choose the employeeId (which shows the last name of the employee as label) from the employees. The whole process is shown in the gallery below.


The resulting treemap is very basic in it’s features, e.g. there is no legend as you see later.
In the next step we create an af:popup to show the nodes detail information. This process is outlined in the next gallery. We drag the popup component onto the page below the af:treemap component

One thing to take note of are the properties of the popup. First we set the content delivery to ‘lazyUncached’, which makes sure that the data is loaded every time the popup is opened. Otherwise we’ll see only the data from the first time the popup has been opened. Second change is to set the launcherVar to ‘source’. This is the variable name we later use to access the node data. Third change is to set the event context to ‘launcher’. This means that events delivered by the popup and its descendents are delivered in the context of the launch source.

The treemap for example, when an event is delivered ‘in context’ then the data for the node clicked is made ‘current’ before the event listener is called, so if getRowData() is called on the collectionModel in the event listener it will return the data of the node that triggered the event. This is exactly what we need.

Finally we add a popupFetchListener to the popup which we use to get the data from the current node to a variable in the bindings. In the sample this variable ‘nodeInfo’ is defined in the variable iterator of the page and an attribute binding ‘nodeInfo1’ is added. More info on this can be found here.


The code below shows the popupFetchListener:

package de.hahn.blog.treemappopup.view.beans;

import javax.el.ELContext;
import javax.el.ExpressionFactory;

import javax.faces.application.Application;
import javax.faces.context.FacesContext;

import oracle.adf.model.BindingContext;
import oracle.adf.share.logging.ADFLogger;
import oracle.adf.view.rich.event.PopupFetchEvent;

import oracle.binding.AttributeBinding;
import oracle.binding.BindingContainer;


/**
 * Treemap handler bean
 * @author Timo Hahn
 */
public class TreemapBean {
    private static ADFLogger logger = ADFLogger.createADFLogger(TreemapBean.class);

    public TreemapBean() {
    }

    /**
     * listen to popup fetch.
     * @param popupFetchEvent event triggerd the fetch
     */
    public void fetchListener(PopupFetchEvent popupFetchEvent) {
        // retrieve node information 
        String lastName = (String) getValueFromExpression("#{source.currentRowData.lastName}");
        Integer id = (Integer) getValueFromExpression("#{source.currentRowData.EmployeeId}");
        //build info string
        String res = lastName + " id: " + id;
        logger.info("Information: " + res);
        // get the binding container
        BindingContainer bindings = BindingContext.getCurrent().getCurrentBindingsEntry();

        // get an ADF attributevalue from the ADF page definitions
        AttributeBinding attr = (AttributeBinding) bindings.getControlBinding("nodeInfo1");
        //set the value to it
        attr.setInputValue(res);
    }

    // get a value as object from an expression
    private Object getValueFromExpression(String name) {
        FacesContext facesCtx = FacesContext.getCurrentInstance();
        Application app = facesCtx.getApplication();
        ExpressionFactory elFactory = app.getExpressionFactory();
        ELContext elContext = facesCtx.getELContext();
        Object obj = elFactory.createValueExpression(elContext, name, Object.class).getValue(elContext);
        return obj;
    }
}

Finally we have to design the popup to show the node info from the attribute binding ‘nodeInfo1’. The popup uses a dialog with an af:outputText like

Show the node info in the popup

Show the node info in the popup


and set an af:showPopupBehavior to the node showing the employees

Running the finished application brings up the treemap, not pretty but enough to see this use case working. If we click on an employee node we see the popup with the last name of the employee and the employee id, the primary key of the selected row in the employees iterator.

You can download the sample application which was build using JDeveloper 12.1.3 and the HR DB schema from GitHub.

Pitfalls when using libraries of newer version than shipped with JDeveloper or WebLogic Server

A question on JDeveloper & ADF OTN forum cought my attention. A user wanted to use a method of the Apache Commons-IO library named FileUtils.getTempDiretory() but got an error when he tried to use code completion or when he tried to compile the code. The problem was that the compiler (or code completion) did not pick up the right java class from the library even as it was installed in the project a library.
As the original code used belongs to one of my samples I was interested in finding a reason for this behavior as I could see no obvious reason for this behavior.

An inspection of a provided test case quickly revealed the problem and a solution was found too. This blog is about the problem and the solution to it. Lets start with building a test case:


The test case had a model project which used a couple of libraries which we add too to make this sample as close as possible to the test case.
 Model Project Properties

Model Project Properties


There is no code whatsoever used in the model project just the libraries are defined!

To make use of the FileUtils.getTempDiretory() method we have to first download the Apache Commons-IO in a version higher then 2.0. The current version is 2.4 which you get from the given link. Once you unzip the zip (or tar.gz) to a directory of your choice we create a new library for JDeveloper (Tools->Manage Libraries…)

We add This new library to the view controller project


Next is to create a java bean where we try to use the FileUtils.getTempDiretory() method

Here we see the problem mentioned in the OTN question. The FileUtils.getTempDiretory() does not show up at all. The JavaDoc of the Apache Commons-IO 2.4 package shows that the method is available since version 2.0
JavaDoc of FileUtils Class

JavaDoc of FileUtils Class


If we try to compile the code we get an compilation error as seen in the last image.

What is the problem?
Well, it looks like there is another version of the Apache Commons-IO library already loaded in the classpath which gets loaded first. Once a library or class is loaded, another version of the same class will not overwrite the existing one.
First thing we can try is to move the new commons-io library to the top of the list of libraries.
In the test case presented here, this doesn’t work. We still get the same error. So there are libraries loaded before the view controller project libraries come to play.
Remember we added some libraries to the model project even as there is no code in the project at all?
Because the view controller project has dependency defined to the model project when we create an Fusion Web Application by default, libraries of the model project are loaded before the view controller projects.
We have can solve the problem in multiple ways:
1. remove the dependency to the model project. This is not recommended as it would mean that we have to build the model project ourselves if we have to change something in the model and want to run the application.
2. find the library which loads the FindUtils class and see if we can remove it (not all libraries are needed).
3. add the new Apache Commons-IO library to the model project and move it up front. This should load the newer version of the FindUtils class before any other.

Solution 1 isn’t really one. Solution 2 is possible and I’ll blog about it later. For this blog we use solution 3.

Solution
All we have to do is to add the Apache Commons-IO 2.4 library to the model project and move to the top of the list.

Model Project Properties with Commons-IO

Model Project Properties with Commons-IO


If we now rebuild the workspace we see that to error is gone
No Compilation Error

No Compilation Error


The code completion still shows the method red underlined. This is a bug in JDeveloper which doesn’t pick up the right library. Anyway, the compiler will use the right library and we can compile the application.

Now we add another method to the FileBean which returns the path to the temporary directory. This we use in a page index.jsf to show it on the ui.

<?xml version='1.0' encoding='UTF-8'?>
<!DOCTYPE html>
<f:view xmlns:f="http://java.sun.com/jsf/core" xmlns:af="http://xmlns.oracle.com/adf/faces/rich">
  <af:document title="index.jsf" id="d1">
    <af:form id="f1">
      <af:panelGridLayout id="pgl1">
        <af:gridRow height="50px" id="gr2">
          <af:gridCell width="100%" halign="stretch" valign="stretch" id="gc1">
            <!-- Header -->
            <af:outputText value="Preferred Package Test" id="ot1" inlineStyle="font-size:x-large;"/>
          </af:gridCell>
        </af:gridRow>
        <af:gridRow height="100%" id="gr1">
          <af:gridCell width="100%" halign="stretch" valign="stretch" id="gc2">
            <!-- Content -->
            <af:outputText value="Tempdir path = #{FileBean.tempDir}" id="ot2"/>
          </af:gridCell>
        </af:gridRow>
      </af:panelGridLayout>
    </af:form>
  </af:document>
</f:view>

When we run the application we get an exception

    at weblogic.work.ExecuteThread.execute(ExecuteThread.java:311)
    at weblogic.work.ExecuteThread.run(ExecuteThread.java:263)
Caused by: java.lang.NoSuchMethodError: org.apache.commons.io.FileUtils.getTempDirectoryPath()Ljava/lang/String;
    at de.hahn.blog.preferredpackages.view.beans.FileBean.getTempDir(FileBean.java:16)
    at sun.reflect.NativeMethodAccessorImpl.invoke0(Native Method)
    at sun.reflect.NativeMethodAccessorImpl.invoke(NativeMethodAccessorImpl.java:57)
    at sun.reflect.DelegatingMethodAccessorImpl.invoke(DelegatingMethodAccessorImpl.java:43)
    at java.lang.reflect.Method.invoke(Method.java:606)

Why’s that?
The application compiled without an error and we still get a NoSuchMethodError. The reason for this is that when we start the WebLogic Server the older version of the Apache Commons-IO jar is loaded first, blocking loading of the newer version we need to get to the FileUtils.getTempDirectoryPath() method.
To allow the server to load our newer version of the jar we need to change a descriptor named weblogic-application.xml which is specific for WebLogic Server. For other servers there exist other descriptors allowing the same.
In this descriptor we add a preferred package for the org.apache.commons.io package. Open the weblogic-appliaction.xml descriptor and select the ‘Class Loading…’ node.

Application Descriptors: weblogic-application.xml

Application Descriptors: weblogic-application.xml


Here we enter the package name org.apache.commons.io to the ‘Application Preferred Libraries’ section.

to get this result in the source view of the descriptor:

    <prefer-application-packages>
        <package-name>org.apache.commons.io</package-name>
    </prefer-application-packages>

After restarting the application the index.jsf page show up OK

Running Test Page

Running Test Page

You can download the sample application which was build using JDeveloper 12.1.3 from GitHub.

Master/Detail using af:tree for master and af:form for detail

A question on OTN JDeveloper and ADF forum about a sample for a master/detail navigation using a tree as master and a form as detail cought my attention. I quickly setup a sample I like to share in this post.

Frank Nimphius wrote an article How-to select multiple parent table rows and synchronize a detail table with the combined resul about this which is quite complex as it not only shows how the synchronization is done but adds CRUD operation to the tree too.
My sample is for beginners and only shows how to do the synchronization of the tree navigation and the detail from. If you need more functionality please refer to Frank’s article.

Problem description
At first the use case sounds easy. However, using a tree has it’s pitfalls. The tree navigation doesn’t synchronize the child node iterators with the node selection. This means that when we select a child node that the current row of the child iterator is not set to the selected row. This is shown in the gallery below:


As we see, in the second picture the form show the wrong data for the selected node. The form shows the first child row of the selected parent node.

Page layout
the sample uses a panel splitter which holds the af:tree on the left and the detail af:form on the right panel. This is shown in the gallery below:


The af:form has a partialTrigger set which listens to updates of the af:tree. The af:tree was build by dragging the Departments from the data control onto the left panel of the splitter. This created the markup
Default selectionListener of af:tree

Default selectionListener of af:tree


with the default selectionListener marked in blue. This is where we need to make the change.

Solution
To make the use case work, we have to use a custom selectionListener for the af:tree. The gallery below shows how a request scoped bean and the selectionListener is created:


The last image shows that the bean is automatically registered in the task flow (in adfc-config.xml in this case).
Now that we have created the selectionListener method in the bean, we can code it like

    /**
     * Custom managed bean method that takes a SelectEvent input argument
     * to generically set the current row corresponding to the selected row
     * in the tree.
     *
     * This method is not generic as it uses the normal binding.iterator.model.makecurrent the ui component uses.
     * The child iterator must be known too to select the child not in the chile view.
     *
     * @param selectionEvent object passed in by ADF Faces when configuring
     * this method to become the selection listener
     */
    public void onTreeNodeSelection(SelectionEvent selectionEvent) {
        //original selection listener set by ADF
        //#{bindings.Departments.treeModel.makeCurrent}
        String adfSelectionListener = "#{bindings.Departments.treeModel.makeCurrent}";
        //make sure the default selection listener functionality is
        //preserved. you don't need to do this for multi select trees
        //as the ADF binding only supports single current row selection

        /* START PRESERVER DEFAULT ADF SELECT BEHAVIOR */
        FacesContext fctx = FacesContext.getCurrentInstance();
        Application application = fctx.getApplication();
        ELContext elCtx = fctx.getELContext();
        ExpressionFactory exprFactory = application.getExpressionFactory();

        MethodExpression me = null;
        me = exprFactory.createMethodExpression(elCtx, adfSelectionListener, Object.class, new Class[] { SelectionEvent.class });
        me.invoke(elCtx, new Object[] { selectionEvent });

        /* END PRESERVER DEFAULT ADF SELECT BEHAVIOR */
        RichTree tree = (RichTree)selectionEvent.getSource();
        TreeModel model = (TreeModel)tree.getValue();

        //get selected nodes
        RowKeySet rowKeySet = selectionEvent.getAddedSet();
        Iterator<Object> rksIterator = (Iterator<Object>)rowKeySet.iterator();
        //for single select configurations,this only is called once
        while (rksIterator.hasNext()) {
            List<Object> key = (List<Object>)rksIterator.next();
            JUCtrlHierBinding treeBinding = null;
            CollectionModel collectionModel = (CollectionModel)tree.getValue();
            treeBinding = (JUCtrlHierBinding)collectionModel.getWrappedData();
            JUCtrlHierNodeBinding nodeBinding = null;
            nodeBinding = treeBinding.findNodeByKeyPath(key);
            Row rw = nodeBinding.getRow();
            //print first row attribute. Note that in a tree you have to
            //determine the node type if you want to select node attributes
            //by name and not index
            String rowType = rw.getStructureDef().getDefName();

            // check the node type as we don'T have to do anything is the parent node is selected
            if (rowType.equalsIgnoreCase("DepartmentsView")) {
                logger.info("This row is a department: " + rw.getAttribute("DepartmentId"));
            } else if (rowType.equalsIgnoreCase("EmployeesView")) {
                // for the child node we search the row which was selected in the tree
                logger.info("This row is an employee: " + rw.getAttribute("EmployeeId"));
                Object attribute = rw.getAttribute("EmployeeId");
                // make the selected row the current row in the child iterator
                DCBindingContainer dcBindings = (DCBindingContainer)BindingContext.getCurrent().getCurrentBindingsEntry();
                DCIteratorBinding iterBind = (DCIteratorBinding)dcBindings.get("EmployeesOfDepartmentsIterator");
                iterBind.setCurrentRowWithKeyValue(attribute.toString());
            } else {
                // tif you end here your tree has more then two node types
                logger.info("Huh????");
            }
            // ... do more useful stuff here
        }
    }

Line 15 holds the original selectionListener expression from the af:tree. This we need to select the master node and row.
Lines 16-30 preserve the original function by executing the expression.
lines 31-38 check if there are selected nodes in the tree. Here we have to remember that a selected node in a tree isn’t just a rowKey the selectionListener always returns a set of nodes. If the tree is set to single selection the set contains exactly one node.
lines 39-49 get the data model of the tree and getting the node data (or row) and it’s type. This type we use to distinguish between master and detail selection.
Lines 51-65 do exactly this. In case of a master node (Departments) there is nothing to do as the default iterator does all the work. In case of a detail node (Employees) we have to do the magic.
Lines 54-61 Here we get the PK of the child row from the node and set the current row in the child iterator to this row (line 61).

That’s it. when we now run the application we see that the detail form synchronizes with the selected node in the tree.

The sample needs the HR DB schema. You can download the code for the sample, which was build using JDeveloper 11.1.1.9.0, from GitHub. Please note that if you run the sample in your environment, that you have to change the DB connection to the HR DB schema according to your environment.

Initialize an execute af:quickQuery default criteria on page load

Recently a question on the OTN JDeveloper & ADF Space caught my interest. The question was how to initialize an af:quickQuery component with a parameter passed to a task flow on load of a page.
At first I thought that this would be a simple case of setting a property (InitialQueryOverwritten=true) as mentioned by Frank Nimphius in his article How-to query af:quickQuery on page load ?, but after a short test it turned out, that this setting only executes the query but can’t be used to initialize the criteria.

This blog is about a solution to this problem. The question can be divided into two smaller problems to solve. The first is to pass a parameter to a bounded task flow and use the passed parameter in the bounded task flow. The second problem is to initialize a default query attribute of a af:quickQuery component and execute the query.

Let’s have a look at the running application.

Start Page

Start Page


On the start page the user can enter a parameter, which is used as input parameter in the second page, which holds a region (as bounded task flow) with the quick query component. Clicking on the ‘Go Query’ button passes the entered parameter to a pageFlowScope variable. and navigates to the second page.
Start Page Page with initialized af:quickQuery

Start Page Page with initialized af:quickQuery


As we see, the passed parameter is visible in the quick query component and the table shows the corresponding data in the table.

The first problem mentioned isn’t really one as the solution the well documented. So passing a parameter from an af:inputText to a bounded task flow will only showed briefly here. The button on the start page uses a af:setPropertyListener to set the parameter to a pageFlowScope variable. On the second page the parameter is passed as input parameter to the bounded task flow which assembles the af:quickQuery.


The images above showing the navigation between the two pages and the region (QuickQuery.jsf) which holds the af:quickQuery.

First Try
The first method I tried to initialize the af:quickQuery was to overwrite the QueryListener of the af:quickQuery component to set the parameter to the default search attribute. The already mentioned property InitialQueryOverwritten=true would then execute the query with the parameter set. This should show the right result in the table. As it turned out, if the property InitialQueryOverwritten is set to true, the QueryListener is not called on load of the page. No change to set the parameter which is passed to the bounded task flow.

Second Try
For the next try I used a method activity in the bounded task flow and tried to set the parameter from this method. This will not work as the component is not present when the method is called as default activity in the task flow. You can set the parameter to the view object and filter the data after it, however, the overwritten property InitialQueryOverwritten then executed the default query again, this time without the parameter. If you set the property to false, you see the data, but the parameter is not set in the af:inputText component.

Final Try: Working solution
The working solution uses a trick which is kind of lazy initializing the component. For this we bind a property of the component to a bean and overwrite the getter method for the property. In the getter we check a private variable of the bean if the component has been called already or not. In case the getter has already been called we just return the value for the property. In case the getter method is called the first time we initialize the component before returning the value of the property.

Let’s look at the af:quickQuery in the region:

                        <af:quickQuery label="Search" searchDesc="#{viewScope.QuickQueryBean.dummy}" id="qryId1"
                                       value="#{bindings.ImplicitViewCriteriaQuery.quickQueryDescriptor}"
                                       model="#{bindings.ImplicitViewCriteriaQuery.queryModel}"
                                       queryListener="#{bindings.ImplicitViewCriteriaQuery.processQuery}" binding="#{viewScope.QuickQueryBean.quickQuery}">
                            <f:facet name="end">
                                <af:commandLink text="Advanced" rendered="false" id="cl1"/>
                            </f:facet>
                        </af:quickQuery>

Two things to note are
1. the component is bound to the viewScope bean QuickQueryBean
2. the searchDesc property is bound to the same QuickQueryBean bean
The component is bound to the bean as a convenience to get the query descriptor easily in the initialization method. To make this save we use a ComponentReference to store the component.

    private ComponentReference quickQuery;
    /**
     * setter for component to ComponentReference
     * @param quickQuery the component
     */
    public void setQuickQuery(RichQuickQuery quickQuery) {
        this.quickQuery = ComponentReference.newUIComponentReference(quickQuery);
    }

    /**
     * getter for the component from the component reference
     * @return
     */
    public RichQuickQuery getQuickQuery() {
        if (quickQuery != null) {
            return (RichQuickQuery) quickQuery.getComponent();
        }
        return null;
    }

For more information about this technique see Rules and Best Practices for JSF Component Binding in ADF

The lazy initialization is done by binding the searchDesc property to the QuickQueryBean. The trick is that the component has to call the getter for this property to get it’s value. In the getter in the bean

    /**
     * getter for a string value names dummy in EL
     * @return value of the dummy property
     */
    public String getDummy() {
        if (needInit) {
            needInit = false;
            initQuickQuery();
        }
        return "Search";
    }

we check a local variable ‘needInit’ which is set to true when the bean is created each time the page gets loaded. As the bean is in viewScope it guarantees that the bean is created each time the page is loaded and stays active until the page is visible.
The real work is done in the initQuickQuery() method:

    /**
     * Initialize the quickQuery component if a parameter tpCityName is found in the pageFlowScope. Once this is done, the pageFlowScope
     * variable tpCityName is set to null or removed.
     */
    public void initQuickQuery() {
        // get the PageFlowScope Params
        AdfFacesContext adfFacesCtx = AdfFacesContext.getCurrentInstance();
        Map<String, Object> scopePageFlowScopeVar = adfFacesCtx.getPageFlowScope();
        String paramCity = (String) scopePageFlowScopeVar.get("tpCityName");
        if (paramCity != null && !paramCity.isEmpty()) {
            // get query descriptor (the components value property)
            FilterableQueryDescriptor queryDescriptor = (FilterableQueryDescriptor) getQuickQuery().getValue();
            // get the current selected criterion (which should set in the ImplicitViewCriteriaQuery in hte pageDef
            AttributeCriterion attributeCriterion = queryDescriptor.getCurrentCriterion();
            // get the attribute name and check if it'S 'City'
            AttributeDescriptor attribute = attributeCriterion.getAttribute();
            String name = attribute.getName();
            // only set parameter if hte attribute matches the parameter
            if ("City".equalsIgnoreCase(name)) {
                attributeCriterion.setValue(paramCity);
                // remove value to allow new one in component
                scopePageFlowScopeVar.put("tpCityName", null);
                // set the parameter to the attributeCriterion 
                QueryModel model = getQuickQuery().getModel();
                model.setCurrentDescriptor(queryDescriptor);
                // create a queryEvent and invoke it
                QueryEvent qe = new QueryEvent(getQuickQuery(), queryDescriptor);
                invokeMethodExpression("#{bindings.ImplicitViewCriteriaQuery.processQuery}", Object.class, QueryEvent.class, qe);
            }
        }
    }

In this method we check if a parameter named ‘tpCityName’ is present in the pageFlowScope (lines 8-10). If yes the next check is if the current selected criterion the for the selected parameter, in this case the ‘City’ (lines 11-19) . Only if this test is positive the value from the parameter is set to the criterion (line 20), the pageFlowScope variable ‘tpCityName’ is removed and the new criterion is set back to the query model (lines 21-25). Finally to execute the af:quickQuery we create a new QueryEvent and invoke it via an EL (lines 26 -28).
The solution does not need to set the InitialQueryOverwritten property to true to run. The query is fired after setting the attribute via the QueryEvent. Here is an image of the af:quickQuery binding

Definition of the ImpliciteViewCriteriaQuery

Definition of the ImpliciteViewCriteriaQuery

The sample needs the HR DB schema. You can download the code for the sample, which was build using JDeveloper 12.1.3, from GitHub. Please note that if you run the sample in your environment, that you have to change the DB connection to the HR DB schema according to your environment.

How-to filter ADF bound tables by date range (JDeveloper 12.1.x)

Based on an older article from Frank Nimphius How-to filter ADF bound tables by date range JDeveloper 11.1.1.4 I got a interesting question on the OTN JDeveloper & ADF forum why the solution provided in the article does not work in JDev 12c.

The solution from Frank’s article is designed for JDev 11.1.1.4.0. Today’s version of JDev is 12.1.3 where the solution does not seem to work. Migrating the source of the article and running it under JDev 12.1.3 indeed shows, that filtering the employees records for a date range does not work at all. Setting dates into the filter and hitting enter to activate the filter does not filter the data in the table.
The reason for this was easily found by debugging the code. Set a breakpoint into the query listener which is setup in the table

<af:table value="#{bindings.allEmployees.collectionModel}" var="row" 
  rows="#{bindings.allEmployees.rangeSize}"
  emptyText="#{bindings.allEmployees.viewable ? 'No data to display.' : 'Access Denied.'}"
  fetchSize="#{bindings.allEmployees.rangeSize}" rowBandingInterval="0"
  filterModel="#{bindings.allEmployeesQuery.queryDescriptor}" filterVisible="true" 
  varStatus="vs" selectedRowKeys="#{bindings.allEmployees.collectionModel.selectedRow}"
  selectionListener="#{bindings.allEmployees.collectionModel.makeCurrent}" 
  rowSelection="single" id="t1" styleClass="AFStretchWidth"  partialTriggers="::cb1"
  queryListener="#{EmployeeQueryBean.onEmployeeQuery}">

As you can see it’s pointing to a bean method ‘onEmplyoeeQuery’. A look into this method reveals that the method FilterableQueryDescriptor.getFilterCriteria() has been deprecated.

        FilterableQueryDescriptor fqd = (FilterableQueryDescriptor) queryEvent.getDescriptor();
        Map map = fqd.getFilterCriteria();

Instead of the deprecated method you should use the method FilterableQueryDescriptor.getFilterConjunctionCriterion() which now holds the map of parameters.

        FilterableQueryDescriptor fqd = (FilterableQueryDescriptor) queryEvent.getDescriptor();
        ConjunctionCriterion cc = fqd.getFilterConjunctionCriterion();
        Map<String, Criterion> criterionMap = cc.getCriterionMap();

When you set a breakpoint in this method and step through the code you see that the values entered into the filter fields in the UI are not visible in the map as Frank describes in his article.

Criterion Map and old FilterCriteria Map

Criterion Map and old FilterCriteria Map


As you can see there are no map entries for the made up variables ‘HireStartRange’ and ‘HireEndRange’. This is the reason the filter by date range does not work. There are simply not dates to filter the rows.

I’m not sure if this is a bug or a change in behavior which was made for a reason. Anyway, you can’t just simply add values to the map anymore.

The solution to fix the problem is simple. As you can’t store additional values in the criterion map, you have to store the values entered by the user somewhere else. A valid storage area is the variables iterator each pagedef holds.
In one of my other blogs Creating Variables and Attribute Bindings to Store Values Temporarily in the PageDef I showed how to add temporary variables in this iterator.

Create two new variables inside the variable iterator of type oracle.jbo.domain.Date, name them ‘startDate’ and ‘endDate’. Then create attribute bindings for them.
The final touch is to wire the new variables up in the HireDate filter for start range and end range:

                                    <af:column sortProperty="HireDate" filterable="true" sortable="true"
                                               headerText="#{bindings.allEmployees.hints.HireDate.label}" id="c1" width="277">
                                        <f:facet name="filter">
                                            <af:panelGroupLayout id="pgl2" layout="horizontal">
                                                <af:panelLabelAndMessage label="From: " id="plam1">
                                                    <af:inputDate id="id2" value="#{bindings.startDate1.inputValue}" clientComponent="false">
                                                        <af:convertDateTime pattern="#{bindings.allEmployees.hints.HireDate.format}"/>
                                                        <f:validator binding="#{bindings.HireDate.validator}"/>
                                                    </af:inputDate>
                                                </af:panelLabelAndMessage>
                                                <af:spacer width="5" height="5" id="s1"/>
                                                <af:panelLabelAndMessage label="To:" id="plam2">
                                                    <af:inputDate id="id3" value="#{bindings.endDate1.inputValue}" required="false" clientComponent="false">
                                                        <f:validator binding="#{bindings.HireDate.validator}"/>
                                                        <af:convertDateTime pattern="#{bindings.allEmployees.hints.HireDate.format}"/>
                                                    </af:inputDate>
                                                </af:panelLabelAndMessage>
                                            </af:panelGroupLayout>
                                        </f:facet>
                                        <af:inputDate value="#{row.bindings.HireDate.inputValue}" label="#{bindings.allEmployees.hints.HireDate.label}"
                                                      required="#{bindings.allEmployees.hints.HireDate.mandatory}"
                                                      shortDesc="#{bindings.allEmployees.hints.HireDate.tooltip}" id="id1" styleClass="AFStretchWidth">
                                            <f:validator binding="#{row.bindings.HireDate.validator}"/>
                                            <af:convertDateTime pattern="#{bindings.allEmployees.hints.HireDate.format}"/>
                                        </af:inputDate>
                                    </af:column>

The code above shows the new column for the HireDate and the new storage location for the startDateRange as ‘value=”#{bindings.startDate1.inputValue}”‘ and EndDateRange as ‘value=”#{bindings.endDate1.inputValue}”‘. Next we change the bean method which reads the filter values and calls the query:

    public void onEmployeeQuery(QueryEvent queryEvent) {
        //default EL string created when dragging the table
        //to the JSF page
        //#{bindings.allEmployeesQuery.processQuery}

        BindingContext bctx = BindingContext.getCurrent();
        DCBindingContainer bindings = (DCBindingContainer) bctx.getCurrentBindingsEntry();

        //access the method bindings to set the bind variables on the ViewCriteria
        OperationBinding rangeStartOperationBinding = bindings.getOperationBinding("setHireDateRangeStart");
        OperationBinding rangeEndOperationBinding = bindings.getOperationBinding("setHireDateRangeEnd");

        // get the start date and end date from the temporary valiables
        AttributeBinding attr = (AttributeBinding) bindings.getControlBinding("startDate1");
        oracle.jbo.domain.Date sd = (oracle.jbo.domain.Date) attr.getInputValue();
        attr = (AttributeBinding) bindings.getControlBinding("endDate1");
        oracle.jbo.domain.Date ed = (oracle.jbo.domain.Date) attr.getInputValue();

        //set the start and end date of the range to search
        rangeStartOperationBinding.getParamsMap().put("value", sd);
        rangeEndOperationBinding.getParamsMap().put("value", ed);

        //set bind variable on the business service
        rangeStartOperationBinding.execute();
        rangeEndOperationBinding.execute();

        invokeMethodExpression("#{bindings.allEmployeesQuery.processQuery}", Object.class, QueryEvent.class, queryEvent);
    }

In line 14-17 you see that we read the values from the newly created attribute bindings for the temporary variables. After removing the unnecessary parts of the code, which tried to read the values from the map, the rest of the code remains as is.

Here is an image of the now working filter by date range

Filter Table by Date Range

Filter Table by Date Range

Please note that if you run the sample in your environment, that you have to change the DB connection to the HR DB schema according to your environment. You can download the changed code for the sample from GitHub

Change Application Configuration at Run Time through a Properties File (Part 2)

In this second part of the mini series we look into the problem left over from part 1.

We left the task to change the location and name of the property file we read when a configuration property is needed by the application. You my ask why we need to change the path at all. The answer is that most administrators won’t allow you to copy a file to any location on a server. They normally don’t allow access to a production server at all. You can ask them to put put the configuration file to a location of their choice and then configure this path during deployment of the application. This is exactly what we do in this blog.

In the first part we finished the sample application which read a property file from a location we can set via a context parameter in web.xml. The question now is, how to change this parameter during deployment of the application. The answer to this is to use a Deployment Plan.

Typically, deployment plans are created by developers along with the associated application files, then distributed to the administrator or another deployer, who updates the plan for a particular environment (such as staging, testing, or production). The deployment plan is stored outside of an application archive or exploded archive directory. We store the initial plan in the ViewControllers src/META-INF folder as BlogReadConfigFile_Plan.xml.

<?xml version='1.0' encoding='UTF-8'?>
<deployment-plan xmlns="http://xmlns.oracle.com/weblogic/deployment-plan" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"
                 xsi:schemaLocation="http://xmlns.oracle.com/weblogic/deployment-plan http://xmlns.oracle.com/weblogic/deployment-plan/1.0/deployment-plan.xsd">
    <application-name>BlogReadConfigFile</application-name>
    <variable-definition>
        <variable>
            <name>ResourceFileLocation</name>
            <value>/tmp/readconfigfile.properties</value>
        </variable>
    </variable-definition>
    <module-override>
        <module-name>BlogReadConfigFile_BRCFViewController_webapp.war</module-name>
        <module-type>war</module-type>
        <module-descriptor external="false">
            <root-element>web-app</root-element>
            <uri>WEB-INF/web.xml</uri>
            <variable-assignment>
                <name>ResourceFileLocation</name>
                <xpath>/web-app/context-param/[param-name="de.hahn.blog.readconfigfile.FILENAME"]/param-value</xpath>
                <operation>replace</operation>
            </variable-assignment>
        </module-descriptor>
    </module-override>
</deployment-plan>

There are three notable parts in the plan. The first is the application name which is the same as you set under application properties in the deployment descriptor


The second section is the variable-definition section. Here we define variables which we later can use in the other parts of the descriptor as reference. Without a variable definition we can’t change a thing in the plan.
We name our variable ‘ResourceFileLocation’ and set the value to any appropriate location we like. This location don’t have to exist on the target server. We change it later anyway.
The third part is the module-overwrite where we specify which part of the of any descriptor, which is part of the deployment, we like to change.
It’s essential to name all parts exactly as they are named in the descriptors in your application.

The module name is the name of the war file we build, the type is war, as we build a war artifact. The root element describes which section of the deployed application we want to change and the URI exactly where the descriptor is located relative from its root.
The interesting part is the variable-assignment where we specify the variable name defined in the variable-definition section and which element of the defined module we want to change. This is done with a XPath expression:

<xpath>/web-app/context-param/[param-name="de.hahn.blog.readconfigfile.FILENAME"]/param-value</xpath>

which describes the location of the context parameter in web.xml with the name “de.hahn.blog.readconfigfile.FILENAME” and set it’s value to the variable value.
The operation tells what we want to do. As we want to replace the value we choose REPLACE as operation.

To make the setup complete we have to specify the BlogReadConfigFile_Plan.xml in the application properties

EAR Deployment Profile Properties

EAR Deployment Profile Properties


For more info about how to use deployment plans refer to the documentation at How to Use Deployment Plans

We can now deploy the application as usual and run the application from within JDeveloper. To show how it’s done when you deploy the application to a stand alone server we first create the ear file, then start the integrated WLS in JDevloper and open the admin console to deploy the application

Deploy the Application

Deploy the Application


which will create the ear file as we see in the log window

[06:33:59 PM] ----  Deployment started.  ----
[06:33:59 PM] Target platform is  (Weblogic 12.x).
[06:33:59 PM] Running dependency analysis...
[06:33:59 PM] Building...
[06:33:59 PM] Deploying 2 profiles...
[06:33:59 PM] Wrote Web Application Module to /data/development/ENTW_12.1.3.0.0/QT/BlogReadConfigFile/BRCFViewController/deploy/BlogReadConfigFile_BRCFViewController_webapp.war
[06:33:59 PM] Wrote Enterprise Application Module to /data/development/ENTW_12.1.3.0.0/QT/BlogReadConfigFile/deploy/BlogReadConfigFile.ear
[06:33:59 PM] Elapsed time for deployment:  1 second
[06:33:59 PM] ----  Deployment finished.  ----

Next step is to open the admin console and to deploy the ear file


If we run the application we see that the application tries to read the properties file in the log window
Running Application

Running Application


We now want to change the properties file the application is using. For this we change the location and name of the properties file we configured in the web.xml file and change the content of the properties file:

The UI has not changed, however, after you click the refresh Properties button and look into the log output you see
Application Tries to Read Properties from New Location

Application Tries to Read Properties from New Location


Please note that the location of the properties file has changed. If we had the file at the position defined in the deployment plan, the application would have read the properties from there.

Handling images/files in ADF (Part 5)

I received a couple of questions regarding the handling of the images directly after upload for the sample application done in part 1-4.

    Part 1 gives an overview of the sample application I’m going to build and how to set it up
    Part 2 shows how to upload a file, store it and download it back to the client
    Part 3 implements two techniques to show the data (image) on the user interface
    Part 4 backport of the sample to JDeveloper 11gR1
    Part 5 implements a technique to show the uploaded file right after upload without the need to commit first

The sample application finished in part 3 (part 4 is a backport to JDev 11gR1 only) has one minor glitch: it doesn’t display an uploaded image directly to the user after uploading it. The user has to commit the data after insert or update of an image before the image becomes visible. Users like to see the newly uploaded image before committing the row. This allows the users to cancel the change or select and upload another image. In this 5th part of the series we implement this.

Before we start to implement let’s talk about how to implement this enhancement. Why isn’t it possible to upload the image data into the blob and then just show the image from the blob via the servlet (see part 3)?
The problem is that the BlobDomain uses a stream to read the data uploaded from the user. This stream can only be read after the BlobDomain is saved, meaning after the commit.

The solution we implement in this part stores the uploaded data (inserted or updated) in a temporary file on the server. Then the server uses the image data from the temporary file to visualize the data. This sounds easy enough, however there is some house keeping to do to make it work.

First we have have to find a place (folder) where we can store the uploaded data until it’s stored in the db or the operation is canceled. Then we need to distinguish which data to show from the servlet (file or blob). Finally we have to clean up the temporary file when we are done.

Lets dive into the implementation. We start from the application at the end of part 3. As the current JDeveloper version is 12.1.3 we do the implementation in this version. The first task is to migrate the old application to 12.1.3. This is done automatically when opening the old work space in JDev 12.1.3 by answering the ‘OK’ to the migration popup. Nothing need to be done here. However, when you download  the work space you’ll notice some clean up I did, like changing the old af:commandButton to the new af:button.

One thing to notice is that the Apache Commons-IO version is updated to 2.4. This update made one other change necessary in the weblogic-application.xml file.

  <prefer-application-packages>
    <package-name>org.apache.commons.io</package-name>
  </prefer-application-packages>

This entry allows the application to use the included commons-io jar to be loaded before the already available commons-io jar, of an older version, in WebLogic server 12.1.3.

Here are the steps we take to implement the tasks:
1) Save the uploaded data to a temporary file as well as to the blob. This is done for convenience. It’S possible to store the data first in the temporary file and only copy it to the BlobDomain when the user commits the changes.
We implement a new java class UploadBlob which holds the BlogDomain and the path to the temporary file. This class also allows to test if a temporary is available.

package de.hahn.blog.uldl.view.types;

import oracle.jbo.domain.BlobDomain;

/**
 * This type class holds the BlogDomain and a path to a temporary file holding the uploaded image data
 */
public class UploadBlob {
    /**
     * Holds the uploaded data
     */
    BlobDomain dataBlob;

    /**
     * Path to the temporary file if availabe
     */
    String tempFile;

    /**
     * C'tor.
     */
    public UploadBlob() {
        super();
        tempFile = null;
        dataBlob = null;
    }

    /**
     * Gets the status of the temporary file
     *
     * @return true if a temporary file is available, false otherwise
     */
    public Boolean getTempFileAvailabe() {
        return (tempFile != null ? Boolean.TRUE : Boolean.FALSE);
    }

    /**
     * @param inageBlob
     */
    public void setInageBlob(BlobDomain dataBlob) {
        this.dataBlob = dataBlob;
    }

    /**
     * Gets the BlobDomain holding the uploaded data
     * @return
     */
    public BlobDomain getDataBlob() {
        return dataBlob;
    }

    /**
     * Sete the path to the temporary file holding the uploaded data
     * @param tempFile path to the temporary file
     */
    public void setTempFile(String tempFile) {
        this.tempFile = tempFile;
    }

    /**
     * Getter for path to temp file holding the data of the uploaded data
     * @return path to the temporary file holding the uploaded data
     */
    public String getTempFile() {
        return tempFile;
    }
}

2) Use this class in the ImageBean.java class where the uploaded data is read. This happens in the valueChangeListener uploadFileValueChangeEvent(ValueChangeEvent valueChangeEvent).

    /**
     * @param valueChangeEvent
     */
    public void uploadFileValueChangeEvent(ValueChangeEvent valueChangeEvent) {
        // The event give access to an Uploade dFile which contains data about the file and its content
        UploadedFile file = (UploadedFile) valueChangeEvent.getNewValue();
        // Get the original file name
        String fileName = file.getFilename();
        // get the mime type
        String contentType = ContentTypes.get(fileName);
        // get the current roew from the ImagesView2Iterator via the binding
        DCBindingContainer lBindingContainer = (DCBindingContainer) BindingContext.getCurrent().getCurrentBindingsEntry();
        DCIteratorBinding lBinding = lBindingContainer.findIteratorBinding("ImagesView2Iterator");
        Row newRow = lBinding.getCurrentRow();
        // set the file name
        newRow.setAttribute("ImageName", fileName);
        // create the BlobDomain and set it into the row
        UploadBlob blob = createBlobDomain(file, Boolean.TRUE);
        newRow.setAttribute("ImageData", blob.getDataBlob());
        // set the mime type
        newRow.setAttribute("ContentType", contentType);
        String tmp = (blob.getTempFileAvailabe() ? blob.getTempFile() : null);
        setTemporaryFileVar(tmp);
        UIComponent ui = (UIComponent) valueChangeEvent.getSource();
        // PPR refresh a jsf component
        ui = ui.getParent();
        AdfFacesContext.getCurrentInstance().addPartialTarget(ui);

    }

Instead of reading the data into the BlobDomain a changed method createBlobDomain is called (line 18). the method now returns an instance of the new class UploadBlob. Below is the code of the new method:

    private UploadBlob createBlobDomain(UploadedFile file, Boolean createTempFile) {
        // init the internal variables
        InputStream in = null;
        OutputStream outTmp = null;
        UploadBlob blobDomain = null;
        OutputStream out = null;
        File tempfile = null;
        logger.info("Starting to create UploadBlog from data...");
        try {
            logger.info("... create BlobDomain...");
            blobDomain = new UploadBlob();
            // Get the input stream representing the data from the client
            in = file.getInputStream();
            // if a temporary file should be created , we do this first as we can't get
            // data data back from the blob until we commit the row. in the next step we
            // write the upload data to a temp file and then copy it into the blob
            if (createTempFile) {
                logger.info("... Creating temporary file...");
                File tempdir = FileUtils.getTempDirectory();
                String ext = FilenameUtils.getExtension(file.getFilename());
                if (!ext.isEmpty()) {
                    ext = "." + ext;
                }
                logger.info("... set extension to " + ext + "...");
                tempfile = File.createTempFile("upl", ext, tempdir);
                logger.info("... " + tempfile.getAbsolutePath() + "...");
                // set path to temporary file
                blobDomain.setTempFile(tempfile.getAbsolutePath());
                FileOutputStream fileOutputStream = FileUtils.openOutputStream(tempfile);
                logger.info("... copy data to temporary file...");
                IOUtils.copy(in, fileOutputStream);
                in = FileUtils.openInputStream(tempfile);
                logger.info("... set inputstream for blog to temporary file...");
            }
            // create the BlobDomain datatype to store the data in the db
            blobDomain.setInageBlob(new BlobDomain());
            // get the outputStream for hte BlobDomain
            out = blobDomain.getDataBlob().getBinaryOutputStream();
            // copy the input stream into the output stream
            logger.info("... copy data to BlobDomain ...");
            /*
             * IOUtils is a class from the Apache Commons IO Package (http://www.apache.org/)
             * Here version 2.0.1 is used
             * please download it directly from http://projects.apache.org/projects/commons_io.html
             */
            IOUtils.copy(in, out);
            logger.info("... Finished OK");
        } catch (Exception e) {
            logger.severe("Error!", e);
            if (tempfile != null) {
                // delete temp file on exception but don'T throw one if there is another exception
                logger.info("Deleted temporary file " + tempfile.getAbsolutePath());
                FileUtils.deleteQuietly(tempfile);
            }
        }
        // return the filled BlobDomain
        return blobDomain;
    }

Depending on the new boolean parameter passed to the method a temporary file is created and the uploaded data is first saved to the temporary file. After that the data is copied from the temporary file into the BlobDomain. At this point the path to the temporary file is saved in the new class for later reference. In case of an exception the temporary file is removed.
Finally in line 22 and 23 of the value change listener we check if a temporary file was generated and we set the path to it to a pageDef variable (see Creating Variables and Attribute Bindings to Store Values Temporarily in the PageDef). For this we use the code below.

    /**
     * Set the temporary file name into a page variable for later use
     * @param name
     */
    private void setTemporaryFileVar(String name) {
        // set pathto temporary file to page variable
        BindingContainer bindings = BindingContext.getCurrent().getCurrentBindingsEntry();
        // get an ADF attributevalue from the ADF page definitions
        AttributeBinding attr = (AttributeBinding) bindings.getControlBinding("TemporaryFile1");
        if (attr != null) {
            attr.setInputValue(name);
        }
    }

The variable is used in the af:image component in the editImage.jsff fragment

                   <af:image source="/render_image?id=#{bindings.ImageId.inputValue}&tmp=#{bindings.TemporaryFile1.inputValue}" id="i1"
                              shortDesc="#{bindings.ImageName.hints.tooltip}" inlineStyle="width:200px;" partialTriggers="cb3" visible="true"/>

here the path to the temporary file is passed to the servlet as second parameter ‘tmp’. In lines 24-27 of the value change listener we send a ppr to the parent component of the af:image to show the now uploaded image.

Another thing to do is to cleanup after the user either cancel or commit the changes. This is done in the cancel_action() or the commit_action() in the ImageBean. Here we call the deleteTemporaryFile() method which checks the existence of a temporary file and deletes it.

    /**
     * delete the temporary file if is present
     */
    public void deleteTemporaryFile() {
        String tempfile = getTemporaryFileVar();
        removeTemporaryFile(tempfile);
        setTemporaryFileVar(null);
    }

3) The final part of the implementation is done in the servlet which is used to get the data back to the client. This is simple as we read the second parameter passed to the servlet. If it’s not empty we always read the image data from the temporary file. If the parameter is empty the servlet gets the data by reading the row from the DB and read the data from the blob. Here are the relevant parts from the servlet:

    public void doGet(HttpServletRequest request, HttpServletResponse response) throws ServletException, IOException {
        StringBuilder sb = new StringBuilder(100);
        String appModuleName = "de.hahn.blog.uldl.model.facade.ULDLAppModule";

        sb.append("ImageServlet ").append(appModuleName);

        try {
            // get parameter from request
            Map paramMap = request.getParameterMap();
            oracle.jbo.domain.Number id = null;
            String tmporaryFilePath = "";
            if (paramMap.containsKey("id")) {
                String[] pVal = (String[]) paramMap.get("id");
                id = new oracle.jbo.domain.Number(pVal[0]);
                sb.append(" id=").append(pVal[0]);
            }
            // check if we find a temporary file name. In this case we allways use this!
            if (paramMap.containsKey("tmp")) {
                String[] pVal = (String[]) paramMap.get("tmp");
                tmporaryFilePath = pVal[0];
                sb.append(" tmp=").append(pVal[0]);
            }

            OutputStream outputStream = response.getOutputStream();
            InputStream inputStream = null;
            BlobDomain image = null;
            String mimeType = null;
            // no temporary file path given, read everything from DB
            if (tmporaryFilePath.isEmpty()) {
                // get method action from pagedef
                BindingContext bindingContext = BindingContext.getCurrent();
                DCBindingContainer amx = bindingContext.findBindingContainer("de_hahn_blog_uldl_view_image_dummyPageDef");
                JUCtrlActionBinding lBinding = (JUCtrlActionBinding) amx.findCtrlBinding("getImageById");
                // set parameter
                lBinding.getParamsMap().put("aId", id);
                // execute method
                lBinding.invoke();
                // get result
                Object obj = lBinding.getResult();
                ImageAccessViewRow imageRow = (ImageAccessViewRow) obj;

                // Check if a row has been found
                if (imageRow != null) {
                    // Get the blob data
                    image = imageRow.getImageData();
                    mimeType = imageRow.getContentType();
                    // if no image data can be found and no temporary file is present then return and do nothing
                    if (image == null) {
                        mLogger.info("No data found !!! (id = " + id + ")");
                        return;
                    }
                    inputStream = image.getInputStream();
                } else {
                    mLogger.warning("No row found to get image from !!! (id = " + id + ")");
                    return;
                }
                sb.append(" ").append(mimeType).append(" ...");
                mLogger.info(sb.toString());
            } else {
                // read everything from temporary file path
                mimeType = ContentTypes.get(tmporaryFilePath);
                File file = FileUtils.getFile(tmporaryFilePath);
                FileInputStream fileInputStream = FileUtils.openInputStream(file);
                inputStream = fileInputStream;
            }

            // Set the content-type. Only images are taken into account
            response.setContentType(mimeType + "; charset=utf8");
            IOUtils.copy(inputStream, outputStream);
            if (tmporaryFilePath.isEmpty()) {
                // cloase the blob to release the recources
                image.closeInputStream();
            }
            inputStream.close();
            // flush the outout stream
            outputStream.flush();
        } catch (Exception e) {
            mLogger.log(Level.WARNING, "Fehler bei der Ausführung: " + e.getMessage(), e);
        } finally {
            mLogger.info("...done!");
        }
    }

The gallery below shows the new work flow.

The work space for part 5 can be downloaded from the ADF EMG Sample side BlogUploadDownload_12.1.3V4.zip.
Or if you are in GIT you can get the work space from GitHub BlogUploadDownload_12.1.3V4