Add example import configuration for DSpace xoai format

[alexklbuckley]

Sponsored-by: Auckland University of Technology, New Zealand

[alexklbuckley]

We are using a custom xsl file to harvest DSpace XOAI format into VuFind Solr.

As there is no current xsl file for XOAI we would like to upstream it so other libraries can use it.

I've removed custom elements of the xsl - I have left in the mapping of dc.rights to the edition Solr field. Please do let us know if we should remove that from this upstream file. Or if there are any other changes you would like us to make.

[demiankatz]

Thanks, @alexklbuckley! See below for a few comments. I also think it would be helpful to either add the matching properties file, or else put a comment in one of the existing DSpace-related properties files raising awareness of the existence of this alternate XSLT.

It also looks to me like this XSLT probably won't work correctly with a collection of records, though maybe I'm mistaken. Are you doing all your harvesting and indexing one record at a time? If so, you might want to consider revising so that you can do batches -- the performance is tremendously improved that way.

[demiankatz]

Is "advisor" the only possible value here that could apply to author2?

[alexklbuckley]

Hi @demiankatz I'm trying to find the answer to this. On the AUT DSpace instance it does look like 'advisor' is the only possible value here. I am checking for a wider DSpace context if that is the case.

[alexklbuckley]

Thanks, @alexklbuckley! See below for a few comments. I also think it would be helpful to either add the matching properties file, or else put a comment in one of the existing DSpace-related properties files raising awareness of the existence of this alternate XSLT.

It also looks to me like this XSLT probably won't work correctly with a collection of records, though maybe I'm mistaken. Are you doing all your harvesting and indexing one record at a time? If so, you might want to consider revising so that you can do batches -- the performance is tremendously improved that way.

Thanks @demiankatz I have pushed a follow-up addressing all but three of your points. Which I will be following up on in the next few days.

I am checking about if 'advisor' is the only possible value for author2.

I will also add a comment to one of the existing DSpace-related properties files raising awareness of the existence of this alternative XSLT.

You're right, the harvesting is being done one record at a time. Thank you for noting that recommendation to do it in batches! Is there any documentation that you could please point me towards for making that change?

[demiankatz]

Thanks for the progress, @alexklbuckley! Regarding your question about supporting multiple records, you might find it helpful to look at #2034 for reference. This was the pull request where we added multi-record support to all of the existing XSLT files. The key is to create a top-level template (matching "/") that can handle a collection tag or a single record, and then creating a more carefully-scoped template to transform a single record. The main key is avoiding global selectors like // since that can be problematic.

I'm far from an XSLT expert, so I apologize if my advice is a bit unclear (or if I'm using the wrong language). But I was able to muddle through and create #2034 by imitating the patterns of better XSLT programmers than myself, and perhaps you can do the same. If you run into trouble, I'll help if I can! :-)

[alexklbuckley]

Thanks for the progress, @alexklbuckley! Regarding your question about supporting multiple records, you might find it helpful to look at #2034 for reference. This was the pull request where we added multi-record support to all of the existing XSLT files. The key is to create a top-level template (matching "/") that can handle a collection tag or a single record, and then creating a more carefully-scoped template to transform a single record. The main key is avoiding global selectors like // since that can be problematic.

I'm far from an XSLT expert, so I apologize if my advice is a bit unclear (or if I'm using the wrong language). But I was able to muddle through and create #2034 by imitating the patterns of better XSLT programmers than myself, and perhaps you can do the same. If you run into trouble, I'll help if I can! :-)

Thanks so much @demiankatz !