jackrabbit-users mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "H. Wilson" <wils...@randdss.com>
Subject Re: updating a node tree on import
Date Thu, 12 Aug 2010 19:04:34 GMT
  Well, since no-one seems to have gotten back to you, I will offer up 
our solution to a (very loosely) similar situation. At the least it may 
point you in the right direction, or help others who search this forum 
(as I did) before coming up with it. However, our scenario was using OCM 
with custom, and sometimes versioned, Objects. We needed an import 
process that would import the version history, but we only needed to 
deal with any one of the following scenarios at one time:

    * Update existing Object (which would take the incoming import and
      create new versions, in addition to the existing versions there)
    * Overwrite existing Object (delete the existing, and use only the
      imported version(s))
    * Do nothing (exactly that, go to the next Object).

We didn't need to take note of Objects disappearing. We used the 
following (quasi-)code to manage importing, where the user was able to 
pick how to deal with import from the above 3 bullets. (This is probably 
over complicated for your purposes).


    public enum ImportMode {
    	DEFAULT; //this is for use by our system only, and not something users can specify

    public synchronized void importRepository (String importFile, ImportMode importMode )
    	 * We used JAXB to marshal on export, and unmarshal on import
    	 * our OCM Objects with their version histories from an XML file.
    	OurRepository importedRepo; //custom class to manage our Objects with their histories.
    	Unmarshaller um = JAXBContext.newInstance (OurRepository.class).createUnmarshaller();
    	importedRepo = (OurRepository)um.unmarshal(new FileReader(importFile));
    	ImportMode modeForLooping;
    	//for each node/object in import file, in our case on a name basis
    	for ( String objName : importedRepo.getObjectList().keySet() ) {
    		if (node/object DOES NOT exist in this Jackrabbit repo)
    			modeForLooping = ImportMode.DEFAULT;//system deals with these
    		else (node/object DOES exist in this Jackrabbit repo )
    			modeForLooping = importMode; //if it exists, do what the user specified.
    		/* Retrieve this Objects history however you exported.
    		 * We used a custom class (ObjectHistory) to manage
    		 * exporting/importing the histories.
    		ObjectHistory objHistory = importedRepo.getObjectList().get(objName);

    		//OurObject is what we store in OCM
    		Iterator<OurObject>  sortedObjIter = objHistory.getAllVersions().values().iterator();
    		OurObject thisObj;
    		switch ( modeForLooping ) {
    			case DO_NOTHING:
    				continue;// go for next OurObject! well, you may want to log the event.
    			case OVERWRITE:
    				ocm.remove (objName);
    				//deliberately fall through, so after we removed old one, we can add the new one.
    			case DEFAULT: //also used if OurObject with objName did not exist already
    				thisObj = sortedObjIter.next();
    				 * remember we had to deal with histories:
    				 * historical versions have different paths,
    				 * which we want to reset to force updates.
    				thisObj.setPath ( OurUtilityClass.convertNameToPath (objName));
    				 * if path was /north/south/east/west, make sure
    				 * /north/south/east is in the OCM already.
    				insertFillers (thisObj.getPath()); //custom method
    				ocm.insert (thisObj)
    				ocm.checkin (thisObj.getPath());
    				//again deliberately fall through to deal with the version history of this Object
    			case UPDATE:
    				while (sortedObjIter.hasNext()) {
    					thisObj = sortedObjIter.next();
    					thisObj.setPath( OurUtilityClass.convertNameToPath(objName));
    					ocm.checkout ( thisObj.getPath());
    					ocm.update ( thisObj);
    					ocm.checkin (thisObj.getPath());


Good luck!

H. Wilson

On 08/11/2010 01:51 PM, ChadDavis wrote:
> Hello.  My application loads some meta/config data at bootstrap.  I
> import the data from an xml file.  Currently, I just remove the
> existing tree and replace it with the incoming xml data.  I would like
> to enhance this algorithm to be an intelligent update process.  I
> would like to be able to take note of nodes that are going to
> disappear, nodes that are new, as well as modifications to nodes.
> I'm working on my algorithm and researching the JCR API's that might
> help me out.
> If you have any suggestions on how to go about such a task, please let me know.

  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message