Document processing in Liferay portal

Question

I've been using Liferay a lot for past 2 years, but I have never needed any extensive document management.

Now I have a portlet where users upload documents (MS office OLE2 documents, ODS documents, PDF etc.) and I have to persist them with all metadata available.

I know how would I do that without using Liferay, I'd probably use Apache solr with Apache Tika ( UpdateRichDocuments and ExtractingRequestHandler ) or Apache Jackrabbit that are using Apache Tika under the hood (org.apache.jackrabbit.extractor.*).

The problem is, that If I look at the trunk of Liferay, there are some key classes :

Hooks (JCRHook, FileSystemHook, CMISHook, s3Hook) that are employed from within DLLocalServiceImpl kinda directly

Another alternative is using DLAppLocalServiceImpl that is employing DLRepositoryLocalServiceImpl and the files are persisted into repository also via Hooks, but a lot of additional stuff is done in there.

1. There is not jackrabbit-text-extractors library in Liferay, so I suppose If I wanted metadata to be extracted from PDF, DOCs, ODS documents, I would have very hard times... because the DL service layer doesn't accept additional properties

   1. I think I'd have to avoid using DL services and JCR hook and access Jackrabbit directly... But I would loose the compatibility and possibility migrate my repository etc.

Could please anybody collaborate on this one please ? Thank you

Answer 1

SOLR for indexing, Jackrabbit for document storage. Managing Liferay Document Library in code is fairly easy, just look at the DL*LocalServiceUtil classes, namely DLFolderLocalServiceUtil and DLFileLocalServiceUtil . By default Liferay just creates a matching folder/file structure on the hard drive (with names changed) so you'd only need to write code or use Jackrabbit if you wanted more than this since Liferay allows up/download and viewing out of the box via the control panel and various portlets.

I haven't used JackRabbit with Liferay but once configured everything should be managed under the covers and you shouldn't need to worry about it on the front end.

When you say "with all metadata available" I'm not sure what is retained, but aside from renaming the file so that it can be tracked there shouldn't be any other changes. It should be quick and easy to test by uploading a file of each type and checking the entries in the LIFERAY/data/document_library directory and subdirectories. Again this would be different if Jackrabbit is used.

Answer 2

those two services DLLocalServiceImpl and DLAppLocalServiceImpl both are and will, I suppose, important. The former one if for direct access to repository. Notice that when adding a file via this service you need to persist corresponding DlFileEntry into database and than reference that addFile(...., fileEntryId, ...).

The latter service is doing additional stuff for you, mainly asset management and workflow.

Regarding your use case, I would avoid using document library, because no metadata can go down into the JCR repository. Actually only metadata/custom properties that you could store would be custom properties AKA Expando feature of Liferay portal.

Best way for you seem to be implement your own jackrabbit hook to store data into repository and let Liferay document library use that repository.

Answer 3

You need to always use DLAppServiceUtil ( as Liferay instructs specifically ). Here is my working code that saves a file to the CMS:

public static void saveFileToCMS(ActionRequest aReq, long groupId, String fileName, File filenameWithPath) {
    try {
        ServiceContext serviceContext = ServiceContextFactory.getInstance(
                Group.class.getName(), aReq);

        // prevents duplicate entries based on unique title name
        Random rand = new Random();
        Integer suffix = new Integer(rand.nextInt(10000));

        DLAppServiceUtil.addFileEntry(groupId, 0, fileName, "application/vnd.ms-excel",
                fileName + suffix.toString(), "description goes here", "changelogname",
                filenameWithPath, serviceContext);

        //log.info("Successfully added the new file");

    } catch (PortalException pe) {
        log.error("Portal Exception occurred while saving file to CMS");
        pe.printStackTrace();
    } catch (SystemException e) {
        log.error("System Exception occurred while saving file to CMS");
        e.printStackTrace();
    }
}

Answer 4

Think Edgar is correct. If you check the current trunk via http://svn.liferay.com/repos/public/portal/trunk/portal-service/src/com/liferay/documentlibrary/service/DLLocalService.java (login as guest and no password), you will no longer find the class DLFolderLocalServiceUtil. We are using the existing DLFolderLocalServiceUtil class as well. Thanks for the heads up. We will refactor our code so when 6.1 comes around we can still use the DocumentLibrary services.