Need help with XML export using Dublin Core

CarolineHayden

Hi all,

I really need figuring out why my XML looks so darn terrible. I apologize that the sample is long, but I'm hoping it provides some clues as to what I'm doing wrong. I also have a copy of my export map attached.

Basically, I'm trying to put together an XML export using Dublin Core. We ultimately want the metadata to be harvested, but our current goal is to just export the metadata records from Collective Access and import them into another system. The main issues that I am noticing are repeated empty XML tags even when the export map has an "omitIfEmpty" rule for that element, metadata elements like an object's dimensions not showing up in the XML, though it's there in the Collective Access record, and metadata elements not displaying in the template specified in the export map. Also, there are some instances in the XML of ![CDATA[]] appearing in some tags. I don't know why that is. Any thoughts are greatly appreciated, but I am truly stuck. We are using Providence 1.7.5.

CarolineHayden

I apologize that my files are zipped, but that was the only way I could submit this post with examples for you. Thanks to Seth for the solution for that!

Jonathan

Hi Caroline, CDATA is a wrapper that encloses strings of text for XML. It's generally a good thing because it instructs the XML reader to treat anything wrapped inside CDATA as text, rather than as XML. If you don't have CDATA, and your textual data happens to contain special characters like ampersands, or characters that XML interprets as markup, it will invalidate your XML and throw errors. CDATA basically just means "this is just text, don't try to parse it."

However, if you really don't want CDATA in your export, you can add a Setting called "stripCDATA", and set the Setting Value to 1.

Your omitIfEmpty instances look fine to me at a glance - it's a little hard to see why they wouldn't be working without access to your system and data. I would suggest double checking your spelling and your element_codes before doing anything else. I did notice at least one instance where dimensions is spelled "dimentions." Things like that, as well is incomplete or incorrect element codes could be part of the problem.

I hope this helps!

CarolineHayden

Thanks Jonathan! The CDATA explanation is really helpful; now I'm not so worried about that.

I do have a habit of typing faster than I think, so hopefully typos are the crux of the issue. Thanks for taking the time to review my work and providing feedback.

I did also notice as I was creating the map that Dublin Core is just not robust enough for my liking, and I ended up mapping many different elements to the same Dublin Core term originally. When I get better at this, I plan on incorporating other metadata schemas like MODS.

CarolineHayden

I have some updates: The XML output is looking much better thanks to some trial-and-error testing. What I've realized is that the syntax for many of these elements are not consistent with documentation around the web. Oh well, now I know.

I have a new question about mapping and pulling relationship type codes and relationship type names from related entities. In my newest version of my export map, I have mapped some related entities to the dc:contributor field, with a restrictToRelationshipTypes option. I tried mapping an XML attribute to that element, @dc:AgentClass, to show the specific relationship type of those entities (what I could call a contributor role). In the XML, that mapped attribute (dc:AgentClass) is showing, but not pulling the data.

I attached another .zip file, and the fields I'm talking about are highlighted in the export map. I guess my concern is that it seems like the relationship_typecode was mapped correctly, since it's showing up in the export, but why wouldn't it pull the data? Did I miss something?

CarolineHayden

Update: I got the related entity relationship type names to work!! I had the context for my dc:contributor field mapping set to ca_entities; I needed to set the context to ca_objects_x_entities. Then, I could use the relationship_typename source for the contributor's role. Now, onto the dimensions!