BIRT Developer Survey

Feedback from users is one of the key inputs into the BIRT planning process. By filling out this survey, you provide the BIRT team with valuable information on how you use BIRT and how you would like to see it evolve.

Survey should take less than 3 minutes to fill out.
BIRT Developer Survey

Replacing the default BIRT XLS Emitter

If you are using the BIRT 3.7 runtime, the BIRT engine is now in one JAR. See the BIRT 3.7 New and Notable for more details.

While this is optimal for a lot of reasons, it does make it a little more difficult to replace the default BIRT emitters. We have seen this with many users wanting to use the Tribix emitters.

See this thread and this bug for an example of the issues. While replacing the emitter is achievable through the method described in the forum post, it is not very elegant. BIRT supports having two emitters that emit the same output format. When the engine processes a request for a specific output format the first emitter found by the runtime generally processes the request. This may not be desirable. To work around this a user can specify a specific emitter id to process the given output format. So if you have both the Tribix XLS and the default BIRT XLS emitter deployed you can specify which emitter to use to process the report. To illustrate how to do this, we will use the Tribix XLS emitter.

Deploy Tribix Emitter

First download the Tribix emitter and deploy it. If you are using the WebViewer with Version 2.6.2 or earlier, just copy:

org.uguess.birt.report.engine.emitter.xls_version.jar
org.uguess.birt.report.engine.common_version.jar

to the WebViewer/WEB-INF/platform/plugins directory.

If you are using BIRT 3.7 or later extract the following jars from the org.uguess.birt.report.engine.emitter.xls_version.jar.

xl-emitter.jar
lib/poi-3.5..jar
lib/commons-logging-version.jar
lib/commons-jexl-version.jar

Copy these jars to the WebViewer/WEB-INF/lib directory. Make sure they are no longer in the org.uguess.birt.report.engine.emitter.xls_version.jar. Finally, copy the modified org.uguess.birt.report.engine.emitter.xls_version.jar and the org.uguess.birt.report.engine.common_version.jar to the WebViewer/WEB-INF/lib folder.

Specifying Which Emitter To Use

The WebViewer provides a URL parameter, named __emitterid that can be used to specify which emitter to use for the xls format. For example:

This URL will use the default BIRT XLS emitter

http://localhost:8090/WebViewer/frameset?__report=test.rptdesign&sample=my+parameter&__emitterid=org.eclipse.birt.report.engine.emitter.prototype.excel&__asattachment=true&__format=xls

This URL will use the Tribix XLS emitter

http://localhost:8090/WebViewer/frameset?__report=test.rptdesign&sample=my+parameter&__emitterid=org.uguess.birt.report.engine.emitter.xls&__asattachment=true&__format=xls

This does not address the case when you are using the export button in the web viewer. In order to change which XLS emitter is used when the button is pressed, you will need to modify the
BirtExportReportDialog.js file in the WebViewer/ webcontent/birt/ajax/ui/dialog directory. Modify the __exportAction function to add the emitter id parameter to the request.




 __exportAction : function( )
 {
.
.
.
    else
   {
    action = action.replace( reg, "$1=false" );
   }
   if( format=="xls"){
    //action = action + "&__emitterid=org.eclipse.birt.report.engine.emitter.prototype.excel";
    action = action + "&__emitterid=org.uguess.birt.report.engine.emitter.xls"
   }
  
   formObj.action = action;
   formObj.method = "post";   
   formObj.submit( );
   
   return true;
  }




If you are using the Report Engine API, there is a render option available for setting the emitter id.




               EXCELRenderOption options = new EXCELRenderOption();  
               options.setOutputFormat("xls");
               options.setOutputFileName("output/resample/ps.xls");
               options.setOption(IRenderOption.EMITTER_ID, "org.eclipse.birt.report.engine.emitter.prototype.excel");
               IRenderTask task = engine.createRenderTask(document);   
               task.setRenderOption(options);

or



               EXCELRenderOption options = new EXCELRenderOption();  
               options.setOutputFormat("xls");
               options.setOutputFileName("output/resample/ps.xls");
               options.setOption(IRenderOption.EMITTER_ID, "org.uguess.birt.report.engine.emitter.xls");
               IRenderTask task = engine.createRenderTask(document);   
               task.setRenderOption(options);

BIRT 3.7 Released

BIRT 3.7 is now available and with this release many improvements and new features are available. BIRT 3.7 now provides a new POJO based runtime for ease of deployment when using the BIRT Viewer or the Report Engine. Open Document Text (ODT), Open Document Presentation (ODP), and Open Document Spreadsheet (ODS) outputs are also available with the addition of three new emitters. BIRT now provides support for Hadoop through a new Open Data Access Driver (ODA) that allows the developer to build queries using Hive Query Language (HQL) using the designer GUI.



To read more about these and other new features for BIRT, see the BIRT 3.7 New and Notable.

BIRT: Using XLS as a data source

BIRT currently supports many data sources, including XML, Web Services, JDBC and flat files. This list can also be extended by implementing BIRT extension points. Using this method is called implementing an ODA(Open Data Access) extension.

An ODA allows the developer to add custom GUI elements to the BIRT designer to create and manipulate a data source. This method is described in a set of articles located on BIRT-Exchange. BIRT also supplies a scripted data source that allows a developer to write code in a simplified interface to retrieve data using Java or JavaScript. This post explains how a scripted data source can be implemented to retrieve XLS data using the Source Forge Java Excel API project. For more information on using this API see this blog post.

Adding the JAR to the BIRT Project

First download the Java Excel API and extract the jxl.jar file from the download. You can add this Jar to your BIRT project in several different ways. See this post for designer classpath options. One simple way to add the jar is to put the jar in your BIRT Project Resource folder and then add a reference to it in the report design. You can set the resource folder in the BIRT Designer Preferences (Window-Preferences-Report Design-Resource). The resource folder is also a web.xml setting in the deployed viewer.



In the above screen shot we have a resources folder under the BIRT reports project. Just import the jxl.jar into this folder. To create the reference to the jar in a report design, select the properties view for the report and click on resources. Click the add file button in the jar files section of the report resources and add the jar.



The jar file should now be accessible in the designer from script or expressions. When you deploy the report to the Viewer make sure to copy the jar to the resource folder of the deployed Viewer.

If this method of adding the jar is not desirable you can always create a folder in the project containing the lib and add the jar to the project classpath entry (Window-Preferences-Report Design-Classpath). This will allow the jar to be accessed at design time. When deploying the report you can then just copy the jar to the WEB-ONF/lib folder of the deployed viewer.

Implementing the Scripted Data Source/Set

BIRT supports implementing a scripted data source in Java or JavaScript. In this example we implement it using JavaScript. Although the code is written in JavaScript we can still call Java code by using the importPackage method. This example uses this approach.

Use the new data source wizard to add a scripted data source and use the new data set wizard to add a new scripted data set. In this example we are reading a three column xls file, so we define 3 columns in our dataset.





The final portion of this example requires that we add code to the script events to call the Java Excel API. Select the new data set and click on the script tab. From the drop down menu select the open event. Enter the following code.

importPackage( Packages.java.io);
importPackage( Packages.jxl );
try{
 var inputWorkbook = new File(params["XLSLocation"].value);
 myworkbook = Workbook.getWorkbook(inputWorkbook);
 sheet = myworkbook.getSheet(0);
 numrows = sheet.getRows();
 curr = 0;
}catch(e){
 curr=0;
 numrows=0;
}

This code opens an xls file pointed to by the report parameter XLSLocation. You can hard code the file but this example is a bit more dynamic. The code then opens the first sheet in the workbook and sets a global JavaScript variable for the number of rows and initializes a current row counter to zero.



Select the drop down menu and select the fetch event and enter the following code.

if( curr < numrows ){

var cell1 = sheet.getCell(0, curr);
row["col1"] = cell1.getContents();
var cell2 = sheet.getCell(1, curr);
row["col2"] = cell2.getContents();
var cell3 = sheet.getCell(2, curr);
row["col3"] = cell3.getContents();

curr++
return true;
}else{
return false;
}

The fetch event is fired once per row of data and stops processing when the event returns false. In this code we are just retrieving the first three columns of the xls sheet and assigning them to the dataset row columns.



The final event to implement is the close event. Select close from the drop down box and enter the following code.

if( typeof myworkbook != 'undefined'  ){
myworkbook.close();
}

This code closes the workbook.



You should now be able to preview the dataset.



Remember to add the parameter that defines the location of the xls file. Once the dataset is complete you can construct your BIRT report as usual. This attached example produces the following report.



This example is available at BIRT-Exchange.

BIRT - Reducing large queries and custom navigation controls

Often times when writing a report that requires large amounts of data, the query times become burdensome. This will lead to dissatisfaction from customers of your report. To get around this issue certain companies like Actuate implement progressive viewing in their products that start returning formatted data before the report generation process is complete. This allows users of the report to start getting data long before the total report is completed.

Another option is to reduce the large queries into smaller queries and add navigation controls to your report to get the next smaller set of data. For example, if you have a customer listing report, the first query may return all customers whose name starts with the letter A. A set of navigation controls could have a next button that returns customers whose name starts with the letter B, etc. This approach is very simple in BIRT and this post details one way of doing it.

As an example we can use the classic models sample database that comes with BIRT. It contains a customers table. To return all customers starting with A we could use the query select * from customers where customername like ‘A%’. We want the report to be dynamic, so we will need script to modify the query on the fly. We can begin by just adding a new datasource that uses the classic models database. Next we create a dataset with the following query:

Select * from customers

The next step is to add a hidden report parameter that we use in script to add a where clause to the query. In this example the parameter is named “like”.



The parameter is hidden and has a default value of A. We can now use this parameter in the beforeOpen script of the data set we created earlier. This script should look like the following:

this.queryText = this.queryText + " where customername like '" + params["like"].value + "%'";

This script will change the query every time the report runs and will append the where clause with the hidden report parameter. The default value for the parameter is A which should list all customers starting with A. Before adding navigation controls to the report we need to create and calculate some global variables that our controls will need. Ideally we will add a previous and a next button to the report. In addition it would also be nice to have a list A-Z that the user could click on to go directly to a set of customers. So at the minimum when the report runs we need to get the parameter for the report and calculate what the previous and next customer names should start with. For example if the parameter is C, previous should be B and next should be D. This can be done with a simple beforeFactory script like:




switch( params['like'].value ){
case 'A':
prev='Z';
next='B';
break;

case 'B':
prev='A';
next='C';
break;

case 'C':
prev='B';
next='D';
break;
.
.
.
}



The beforeFactory script is used because it runs before data is retrieved and executes only once. The BIRT initialize script can be executed multiple times so it is not ideal for this script. The var keyword is also not used with the variables prev and next which will make them visible to downstream events. You could also use the reportContext.setGlobalVariable method to do store the variables as well.

After the table is added to the report you can then add a Text element that contains a script tag to implement your prev and next buttons.



Remember to set the type to HTML. The script tag contains a client side script that just reloads the report with a new value for the like parameter. This function is called nextData and takes the new like value as a parameter. The text element also contains two buttons that call the nextData function with the prev and next server side javascript variables we calculated in the beforeFactory. Notice this is done with the VALUE-OF tag. This tag allows us to combine server side scripting with client side scripting.

<script>
function nextData( a ){
var cl = new String(location.href);
var targetURL = new String();
var tst = cl.indexOf("&like");
if(  tst == -1 ){
 cl = cl + "&like=" + a;
}else{
 var ch = cl.indexOf("&like=");
 cl = cl.substring(0, ch) + "&like=" + a;
    
}
window.location = cl;
}
</script>
<input type="button" value="Previous"
onclick="nextData('<VALUE-OF>prev;</VALUE-OF>')" >
<input type="button" value="Next"
onclick="nextData('<VALUE-OF>next;</VALUE-OF>')" >

The last thing we add to the Text element is an Alpha index for all letters of the alphabet.

<br>
<a href="#" onclick="nextData('A');return false">A</href>
<a href="#" onclick="nextData('B');return false">B</href>
<a href="#" onclick="nextData('C');return false">C</href>
<a href="#" onclick="nextData('D');return false">D</href>
<a href="#" onclick="nextData('E');return false">E</href>
.
.

This allows the user to directly select a letter to use in the like clause. The output for the report looks like the following.


The report also handles the situation when no data is returned.



The no data label is placed in a grid below the table. The grid is bound to the table so it uses the same dataset. The dataset contains an aggregate computed column that counts the number of rows. This computed column is then used in the grid and the table visibility property.



For example the grid visibility is shown below.



The expression is:
(row["myrowcount"] > 0)
The expression returns true when rows exist. This will then hide the grid and label. The opposite condition is used on the table.
(row["myrowcount"] < 1)
This report is available at Birt-Exchange.

BIRT Roadshow coming to a city near you.

Actuate is continuing the very popular BIRT Road Shows that feature a free day of training on BIRT technologies. If you are interested, the dates and cities for April and May are listed below. Slots fill up fast so be sure to register quickly. More information is available here.

April
Seattle, WA Tuesday | 4/5
Minneapolis, MN Wednesday | 4/6
Indianapolis, IN Thursday | 4/7
Philadelphia, PA Wednesday | 4/13
May
Denver, CO Tuesday | 5/3
Toronto, ON Wednesday | 5/4
Princeton, NJ Tuesday | 5/10
Orlando, FL Wednesday | 5/11
Dallas, TX Thursday | 5/12
London, UK Thursday | 5/19
Louisville, KY Tuesday | 5/24
Washington, DC Wednesday | 5/25

BIRT 2.6.2 Enhancements

BIRT 2.6.2 is now available and is a maintenance release with a primary concern on bug fixes. This generally means not many new features are introduced. That said a couple of bug fixes merit special attention. The full list of available fixes can be found here.

BIRT Pagination and Page Scripts

BIRT 2.5 introduced page variables and page scripts. These features added the ability to repaginate based on things like a group change. So in one report output you could add a page variable that showed the page count and current group page number within a group. The full details for this feature are available in the BIRT 2.5 new and notable.

While using this feature is very nice, a particular bug caused the scripts not to fire when using a page break interval of 0. The page break interval is one of the many ways in BIRT to introduce a page break in the output. It essentially informs the engine to process a certain number of rows and then insert a page break. If the interval for a table is set to 0, the engine inserts page breaks when all the master page available real-estate is used. In BIRT 2.6.2 this bug is now fixed, allowing the designer to enter a page break interval of 0 in conjunction with page scripts.



The above example can be downloaded from BIRT-Exchange.

To see some of the pagination bugs that were addressed in 2.6.2 see this bugzilla query.

URL security bug

A security issue with the example Web Viewer was identified which posed a cross-site scripting vulnerability. The issue has to do with the __report URL parameter. This parameter specifies the report for the viewer to run. In previous versions of BIRT a fully qualified URL could be used to specify the report, which allowed the report to reside on a completely different server. Since BIRT reports can contain JavaScript which runs server side this poses a scripting risk. The full details for the bug are located here.

BIRT 2.6.2 fixes this issue by adding a web.xml entry that allows the developer to control what value can be used in the __report URL parameter. The entry is named URL_REPORT_PATH_POLICY and has three valid values (all, domain, and none). If the all value is specified no checks on the URL parameter are performed. This value should only be used when access to the viewer is restricted to a known user base. When the none option is used, only local reports to the viewer context for the server are allowed. When the domain option is used the report can contain a fully qualified URL as long as the domain for the server is the same.



If you cannot upgrade the viewer, the patch details are available in the bugzilla entry.

Formatter object

BIRT 2.6.2 now offers a formatter object in the expression builder to format dates, strings and numbers within a given expression. This function is detailed in the following post.

Radar Chart

The Radar Chart in BIRT was implemented in BIRT 2.6 to function as an example on how to implement your own chart types. The source for this chart type is available in two of the BIRT plug-ins (org.eclipse.birt.chart.examples and org.eclipse.birt.chart.examples.core). In BIRT 2.6.2 this chart type has been improved and it now supports rendering multiple series on the same radial within a radar chart.



The GUI for building the radar chart has also been improved to support a more optimal layout for configuring portions of the chart.