search

ualbertalib / fcrepo4-oaiprovider

Fedora 4 OAI Provider implementation
3 stars 1 forks source link

Set-based selective requests not working #49

Closed anayram closed 4 years ago

anayram commented 5 years ago

When trying to get records limited by collection, Set constructs are not working. Selective harvesting does work with other parameters like from or until, so the problem seems to be limited to Set.

If LAC tries to harvest only theses and dissertations in DublinCore format, they will get many other records that are not part of the Theses and Dissertations collection.

Because requests with metadataPrefix=oai_etdms output only theses and dissertations, I can't be sure if this bug is exclusive to requests with metadataPrefix=oai_dc.

To reproduce: https://era.library.ualberta.ca/oai?verb=ListRecords&metadataPrefix=oai_dc&Set=db9a4e71-f809-4385-a274-048f28eb6814:f42f3da6-00c3-4581-b785-63725c33c7ce or http://gillingham2.library.ualberta.ca:8080/fedora/rest/oai?verb=ListRecords&metadataPrefix=oai_dc&Set=db9a4e71-f809-4385-a274-048f28eb6814:f42f3da6-00c3-4581-b785-63725c33c7ce

piyapongch commented 4 years ago

@anayram @weiweishi @pbinkley @jefferya: I have looked at the OAI provider source codes. It seems to be correct. It queries filter on predicate http://terms.library.ualberta.ca/identifiers/hasCollectionId for the set parameter. If we have not changed the predicate for hasCollectionId, the result should be correct.

I will schedule the Fedora re-indexing when it is possible. Then, we can try it again.

anayram commented 4 years ago

@piyapongch thank you. Seems like set-based harvesting is not working yet.

If LAC is not collecting DC metadata anymore (only ETD-MS) then we don't need to fix this just now for their harvesting since ETD-MS only outputs theses. @weiweishi do you know if this has been confirmed with LAC or is is there a way we can contact them to check whether or not they need DC?

Still this might be something to fix later.

weiweishi commented 4 years ago

Thanks @anayram I have not heard back from LAC yet but will contact them to confirm re:DC metadata. @ConnorSheremeta is going to compare output from Fedora vs output from post-fedora OAISys. So I am tagging him so that he's aware.

sfarnel commented 4 years ago

Thanks @weiweishi It would be great if Connor and Mariana can work together on the comparison of data

jefferya commented 4 years ago

@anayram In the original example, is set (with lowercase s) the correct property? The original example uses an uppercase - sorry for not catching this in July.

https://era.library.ualberta.ca/oai?verb=ListIdentifiers&metadataPrefix=oai_dc&set=5a69bd79-86e9-4208-82d9-917c95873ed5:384e12b4-4427-43fc-9689-33d8eef979db

anayram commented 4 years ago

@jefferya yes! Thank you this is great news. Sorry for the confusion. Closing this ticket.