Adding New Files To Works With Files Through Bulkrax Causes Issues With Existing Filesets and Display in Viewer

[KevinJonesMeta]

Descriptive summary

When using a Bulkrax importer to add new files to Works that already have files, the new files are added, but the existing filesets on the work are declared to be removed and the Removed page icon and thumbnail appears in the Universal Viewer. The existing file continues to display in the UV as well.

Example importer on Staging: https://staging.oregondigital.org/importers/280
Example affected work on Staging: https://staging.oregondigital.org/concern/documents/0r9673755

A small number of affected works are on Production but are actively being fixed by Ryan Wick.

Steps To Reproduce:

1. Export set of existing works from Oregon Digital via Bulkrax.
2. Remove unneeded existing metadata from exporter csv.
3. Add new filenames to files column in csv.
4. Zip edited csv and new files per standard Bulkrax importer procedure.
5. Create and run importer with created zip of csv and files.
6. Removed icons and thumbnails show up in Works and Files section and UV; new fileset is added to work.

Expected behavior

New files can be added through Bulkrax to existing works that already have one or more files without impacting the files already present on a work.

Related work

Link to related tickets or prior related work here.

Accessibility Concerns

Adding files to works with existing files will be an increasingly common procedure as more accessible versions of files are created for inclusion in Oregon Digital. These can be added to existing works through the form without issue (so far) but is potentially more efficient through Bulkrax if this issue is resolved.

[CGillen]

@KevinJonesMeta It looks like this is the default behavior that Bulkrax is moving to. If you want to keep the file, Bulkrax wants you to upload it again in the metadata (I think, haven't tried).
I can easily change this behavior, but are we anticipating using Bulkrax to ever replace files instead of just append?

[KevinJonesMeta]

@CGillen We will pursue that conversation at our next Metadeities meeting. Having to pull original files and re-upload original files every time we add files is probably not tenable because the error rate of exporter/importer round trip is too high and needs too much developer time to fix? We'll see. The time to wait for exporters is a secondary factor, but I don't know how much that will matter. I haven't included files in my exporters except to make sure the functionality works.

[KevinJonesMeta]

Also, we have a use case for the current functionality with existing images if they get alt text and need to replace the existing fileset so the alt text will display/travel when the file is downloaded. So we need the existing configuration for at least one image use case, and we need the opposite for adding files for documents when we add accessible versions of the files.

[CGillen]

hrmmm... two different paths based on resource type is pretty sketchy to me & could easily pile-up exceptions and smelly code (yeah, that's a real technical term).
Let me see if I got this right and maybe we can explore an alternative.

If the work being imported is not an image: The existing behavior is desired. The new file is added as a new fileset and the previous fileset(s) is changed to the "removed" image. The "removed" image is in the UV along with the old PDF content (I assume this is something that needs to change). The "removed" image is still in the works/files section.

If the work being imported is an image: The old fileset(s) are untouched. A new fileset(s) is added for each file, regardless of other filesets attached to work (ie. don't clobber or do any fancy stuffs).

If this is (roughly) the case, lets see what we can do for generalized option for these. "Import and replace" vs "Import and append" maybe? I'm not really well versed in Bulkrax, so I don't know what's possible or what amount of effort it would take, but that would be my guess for a solution 😬

[KevinJonesMeta]

@CGillen Some clarifications:

If the work being imported is not an image: the existing behavior is NOT desired. We need to be able to add files to existing works with files so we can get the original pdf and the accessible version (docx; epub; accessible pdf) on the work and both be available to download and the original continuing to show up in the viewer.

If the work being imported is an image: the newly imported image needs to be the one available for download and show up in the viewer because it is the one with alternative text that needs to show up in Works and Files, the UV information panel, and travel with the file when it is downloaded. I think this is existing behavior but I don't know how reliable the original file is removed/replaced by the new file in all the places it needs to be.

A generalized option makes sense to me from a code standpoint. I don't know what to do about the very different workflows because they are tied to the accessibility solutions we have.