Attachments and Java Beans

Up until this point, I must admit that I have been lazy. Even though most of the XPages I have created in the last two years have made extensive use of Java Beans, I have left the attachments to the XspDocument and the typical upload and download controls. I did not want to open that can of worms and just wanted to stick with what I know works. Well, that is dumb. It is fine for one or two minor applications that are never going to be used anyway, but when it comes down to it, I want it to be correct. Today, I started that adventure and, with all new things, a google search was performed for anything that could help me and point me in the correct direction. (Honestly, what did you old folks do without search engines? I’d be lost! Or at the very least spending 10 hours a day at the public library!) What I did not find was a document that contained upload and download information. Since I do not want to loose what I found today, I decided to write a quick post. Thank you to everyone that I am stealing from to write this…. 😛

The first thing that I noticed was that my concept was faulty. At least I think it was. I wanted to have a single file in my Java Bean that I could upload and download at will and access and save in my DAO layer. Of course I could be mistaken, but it does not seem to work that way. Uploading documents and downloading them again needs to be performed in two different actions, and in two different ways, and with different objects. Futhermore, I do not even offer both functions on the same page, though both are possible with the same bean.

First off, my test bean is very simple. If I were to extract an interface (just to get a quick look at what the bean contains, it would hold the following information

public interface IFileTest {

* This function contains the information to save the document. Normally, I do this in a seperate DAO layer Object.
* The example that I used had the majority of the information in one single class.
* I did not experiment as I wanted to keep everything as simple as possible. Such experiments are further on my to-do list.
public abstract void save() throws Exception;

* This function contains the logic to download the attachement. It is performed in a separate XPage containing only this function call.
public abstract void downloadAttachment() throws Exception;

* This function returns a string that points to the xpages with the download attachment function call.
public abstract String getDownloadURL();

* This function will read the parameters from the URL in order to initialize the data for the viewscoped bean.
* Normally with the Beans I create, this function will access the DAO and set the data in an object contained by this Bean.
* This Bean is a controller in the MVC design pattern.
public abstract void init();

* I just find this helpful.
public abstract String getUnid();

* This object is used ONLY in the upload core XPage control.
public abstract UploadedFile getFile();

public abstract void setFile(UploadedFile file);



As i said, this test is done with as simple a construct as possible.

After this was completed, I worked on uploading a document. It seemed the most logical starting point. My primary source for this was a StackOverflow question posted by David Leedy so, note that the following code is primarily coming from Mark Leusink, the accepted answerer of Mr. Leedy’s question. The first part is the most simple. I have a property in my Bean that is of type . I have a corresponding getter/setter pair. I use EL to link the core control for uploading data to the bean. The real magic happens in the save logic.

public void save() throws Exception{
    /* I use the openNTF Domino API (ODA) for nearly all of my
     * applications. If this was not the case, we would have to worry about
     * proper recycling. Keep in mind that this is also just a test. Normally
     * my routines have much cleaner error handling.
     * The following statement uses a utility that I built that helps me get
     * key objects. I am assuming here that you know how to get a handle on the current
     * document. 
    Document doc = ODASessionHelper.getCurrentDatabase().createDocument();
    doc.replaceItemValue(FIELD_FORM, FORM_NAME);
    // file is and instance of UploadedFIle. This is the Property that is bound to the core FileUpload control
    if(file != null){ 
      IUploadedFile fl = file.getUploadedFile();
      //File is the standard variant.
      File file = fl.getServerFile();
      String fileName = fl.getClientFileName();
      // this gave me ONLY the name of the file without any path information.
      System.out.println(String.format("clientFileName: '%s'", fileName)); 
      // on my system, this gave the character ";"
      System.out.println(String.format("seperator is '%s'", File.pathSeparator));
      // This gives you the location of the file that was uploaded by the control on the server.
      File realNameFile = new File(file.getAbsoluteFile() + File.pathSeparator + fileName);
      System.out.println(String.format("realFile name: '%s'", realNameFile.getAbsoluteFile()));
      boolean renamedFile = file.renameTo(realNameFile);
        //typical code to attach a file to a document.
        RichTextItem body = doc.createRichTextItem(FIELD_BODY);
        body.embedObject(EmbeddedObject.EMBED_ATTACHMENT, "", realNameFile.getAbsolutePath(), null);
      } else {
        throw new Exception("file could not be renamed");
       * Normally at this stage, I save the UNID so that I get that document again
       * to prevent a bunch of new documents being created.  This is just me being 
       * lazy and wanting to get a test out ASAP.
    } else {
      throw new NullPointerException("file was null");

The only issue that I have with the above code is that the new name of the attachment is a bit messed up. I do not know if it is because the operating system is windows, or if it is because of the domino version, but the attachment name is changed to “_<strangenumbers>tmp;realAttachmentName.txt” . This is because File.pathSeperator is a semicolon. I have a workaround for this in my download function, but a workaround is still only a workaround.

As I previously said, I did not find a post with both upload and download functionality explained. I did find an awesome article on openNTF regarding downloading attachments programmatically. So, here is a quick shout-out to Naveen Maurya who posted the XSnippet. In the example provided, an XPage was built which called a server-side JavaScript function which got a handle on the FacesContext and the server response to download all files in a zip file. I just edited this to be run in my Bean and not in JavaScript.

   * same disclaimer. I wanted to this quickly. Normally my error handling is 
   * significantly better. I just want the theory here.
  public void downloadAttachment() throws Exception{
    Database db = null;
    Document doc = null;
    OutputStream stream = null;
    ZipOutputStream out = null;
    BufferedInputStream in = null;
      if(StringHelper.isNullOrEmpty(getUnid())) throw new IllegalStateException("Unid is null");
      // again, I am using ODA, and this is just a way to get the current database.
      db = ODASessionHelper.getCurrentDatabase();
       * I normally do this in multiple steps.
       * 1. try to get the document with the UNID
       * 2. try to get the document with the noteID
      doc = db.getDocumentByUNID(getUnid());
        throw new IllegalStateException("body not located");
      } else {
        Item item = doc.getFirstItem(FIELD_BODY);
        if(!(item instanceof RichTextItem)){
          // I would assume that I would have to come up with a MIME variant as well.
          throw new IllegalStateException("item is not of type richtext");
        } else {
          // normally I ask if item is instanceof RichTextItem
          RichTextItem body = (RichTextItem)item;
          Vector objs = body.getEmbeddedObjects();
            throw new IllegalStateException("body has no objects to download");
          } else {
            ExternalContext extContext = FacesContext.getCurrentInstance().getExternalContext();
            // javax.servlet.http.HttpServletResponse;
            HttpServletResponse response = (HttpServletResponse)extContext.getResponse();
            response.setHeader("Cache-Control", "no-cache");
            response.setDateHeader("Expires", -1);
            response.setContentType("application/zip"); // change this for different types.
            // I gave a static name to my zip file, but the original code was dynamic
            response.setHeader("Content-Disposition", "attachment;");
            stream = response.getOutputStream();
            out = new ZipOutputStream(stream);
            for(EmbeddedObject att : objs){
              in = new BufferedInputStream(att.getInputStream());
              int length = in.available();
              byte[] data = new byte[length];
    , 0, length);
              String nm = att.getName();
               * This is my workaround for the file names. Although they are saved in the document
               * with the incorrect name, I could at least download them again with the proper name.
              ZipEntry entry = new ZipEntry(nm.contains(";") ? StringHelper.rightSubstring(nm, ";") : nm);
          // cleanup should be done properly.  this is a 'do as I say, not as I do' moment.....
    } catch(Exception e){
      // very nasty error handling....
      throw e;

In conclusion, I have a test XPage application with one form, and with two xpages. The one xpage allows saving attachments. It has the File Upload control available by the XPages core and a save button. The second XPage is only used for downloading the attachments. It holds no content, but gets the file to download via the HTTPServletResponse in the beforeRenderResponse XPage action. The UNID of the document is passed with the URL.

Although not implemented in an xpage, I also built the logic to open the URL in a new window using client side javascript:"#{javascript:FileTest.getDownloadURL()}", "_blank");

FileTest in the above example is the name of the bean as configured in the FacesConfig.xml file.

My next steps would be

  • to build a view with which I could display the file names and other typical information available for file downloads
  • export files without being compressed into a zip file
  • it goes without saying that I would have to refine the above functions and build in proper cleanup and error handling

Happy Programming!!