> There's actually a direct solution to this problem. Register a service
> using the interface org.apache.sling.servlets.post.NodeNameGenerator
> instead of a PostProcessor.
>
> Your NodeNameGenerator would look like this:
>
> @Component
> @Service
> public class MyNodeNameGenerator implements NodeNameGenerator {
> public String getNodeName(SlingHttpServletRequest request, String
> parentPath, boolean requirePrefix, NodeNameGenerator defaultNNG) {
> if (parentPath.equals("/parent")) {
> return "folderA/" + defaultNNG.getNodeName(request, parentPath,
> requirePrefix, defaultNNG);
> } else {
>
> return null;
> }
> }
> }
Sounds interesting. Is the folder autogenerated if not existant (which
can be the case)?
If yes, there is one caveat: The folder needs a special nodetype. Any
chance to deal with that?
>
> On 1/21/11 6:58 AM, Markus Joschko wrote:
>> Hi,
>> I face some difficulties in realizing the following case:
>>
>> I have a node to which I want to add children. The problem is, that
>> (depending on the content of the to be created node) it is not added
>> to the node directly but a folder is created to which the node is
>> added. Like that
>>
>> + parent
>> + folderA (dynamically created)
>> - childABC
>> + folderB (dynamically created)
>> -childBAC
>>
>> Because the folders might not be available, I post to parent/* and
>> registered a postprocessor to create the folder and move the just
>> created child to the corresponding folder.
>> That works fine. However the redirect url is still pointing to the
>> previous location (parent/childABC) instead of
>> parent/folderA/childABC.
>> And I can't find a way to modify the location of the response within
>> the postprocessor.
>>
>> So I wonder whether I am on the right track or should choose a
>> different approach. Alternatively I could create my own postoperation
>> but that seems to be cumbersome as extending from ModifyOperation
>> isn't easy because of the dependencies in the constructor and
>> AbstractCreateOperation is package private. So I would basically need
>> to copy code directly.
>>
>> Any thoughts on this?
>>
>> Thanks,
>> Markus
>
>
