Generate PDF using Managed Bean with custom HTTP headers

Question

Background
Generate a report in various formats (e.g., PDF, delimited, Excel, HTML, etc.) using JDeveloper 11g Release 2 (11.1.2.3.0) upon clicking an af:commandButton. See also the StackOverflow version of this question:
http://stackoverflow.com/q/13654625/59087
Problem
HTTP headers are being sent twice: once by the framework and once by a bean.
Source Code
The source code includes:
- Button Action
- Managed Bean
- Task Flow
Button Action
The button action:
<af:commandButton text="Report" id="submitReport" action="Execute" />
Managed Bean
The Managed Bean is fairly complex. The code to `responseComplete` is getting called, however it does not seem to be called sufficiently early to prevent the application framework from writing the HTTP headers.
HTTP Response Header Override
* Sets the HTTP headers required to indicate to the browser that the
* report is to be downloaded (rather than displayed in the current
* window).
protected void setDownloadHeaders() {
HttpServletResponse response = getServletResponse();
response.setHeader( "Content-Description", getContentDescription() );
response.setHeader( "Content-Disposition", "attachment, filename="
+ getFilename() );
response.setHeader( "Content-Type", getContentType() );
response.setHeader( "Content-Transfer-Encoding",
getContentTransferEncoding() );
Issue Response Complete
The bean indirectly tells the framework that the response is handled (by the bean):
getFacesContext().responseComplete();
Bean Run and Configure
public void run() {
try {
Report report = getReport();
configure(report.getParameters());
report.run();
} catch (Exception e) {
e.printStackTrace();
private void configure(Parameters p) {
p.put(ReportImpl.SYSTEM_REPORT_PROTOCOL, "http");
p.put(ReportImpl.SYSTEM_REPORT_HOST, "localhost");
p.put(ReportImpl.SYSTEM_REPORT_PORT, "7002");
p.put(ReportImpl.SYSTEM_REPORT_PATH, "/reports/rwservlet");
p.put(Parameters.PARAM_REPORT_FORMAT, "pdf");
p.put("report_cmdkey", getReportName());
p.put("report_ORACLE_1", getReportDestinationType());
p.put("report_ORACLE_2", getReportDestinationFormat());
Task Flow
The Task Flow calls Execute, which refers to the bean's `run()` method:
entry -> main -> Execute -> ReportBeanRun
Where:
<method-call id="ReportBeanRun">
<description>Executes a report</description>
<display-name>Execute Report</display-name>
<method>#{reportBean.run}</method>
<outcome>
<fixed-outcome>success</fixed-outcome>
</outcome>
</method-call>
The bean is assigned to the `request` scope, with a few managed properties:
<control-flow-rule id="__3">
<from-activity-id>main</from-activity-id>
<control-flow-case id="ExecuteReport">
<from-outcome>Execute</from-outcome>
<to-activity-id>ReportBeanRun</to-activity-id>
</control-flow-case>
</control-flow-rule>
<managed-bean id="ReportBean">
<description>Executes a report</description>
<display-name>ReportBean</display-name>
<managed-bean-scope>request</managed-bean-scope>
</managed-bean>
The `<fixed-outcome>success</fixed-outcome>` strikes me as incorrect -- I don't want the method call to return to another task.
Restrictions
The report server receives requests from the web server exclusively. The report server URL cannot be used by browsers to download directly, for security reasons.
Error Messages
The error message that is generated:
Duplicate headers received from server
Error 349 (net::ERR_RESPONSE_HEADERS_MULTIPLE_CONTENT_DISPOSITION): Multiple distinct Content-Disposition headers received. This is disallowed to protect against HTTP response splitting attacks.Nevertheless, the report is being generated. Preventing the framework from writing the HTTP headers would resolve this issue.
Question
How can you set the HTTP headers in ADF while using a Task Flow to generate a PDF by calling a managed bean?
Ideas
Some additional ideas:
- Override the Page Lifecycle Phase Listener (`ADFPhaseListener` + `PageLifecycle`)
- Develop a custom Servlet on the web server
Related Links
- http://www.oracle.com/technetwork/middleware/bi-publisher/adf-bip-ucm-integration-179699.pdf
- http://www.slideshare.net/lucbors/reports-no-notes#btnNext
- http://www.techartifact.com/blogs/2012/03/calling-oracle-report-from-adf-applications.html?goback=%2Egde_4212375_member_102062735
- http://docs.oracle.com/cd/E29049_01/web.1112/e16182/adf_lifecycle.htm#CIABEJFB
Thank you!

Answer 1

The problem was that the HTTP headers were in fact being written twice:
1. The report server was returning HTTP response headers.
2. The bean was including its own HTTP response headers (as shown in the question).
3. The bean was copying the entire contents of the report server response, including the headers, into the output stream.
Firefox ignored the duplicate header errors, but Google Chrome did not.