Problem: PREMIS files are not validated

[sallain]

Is your feature request related to a problem? Please describe.

Whether generated by Enduro (through a child workflow) or included in a SIP, PREMIS XML files should be validated before the package is sent to preservation. Archivematica/a3m can parse a PREMIS file to add the file's events to the AIP METS, which happens quite late in the AM/a3m workflow - ensuring that the PREMIS file is valid will hopefully avoid errors at this late point.

Describe the solution you'd like

Add a new activity to validate the premis.xml file against the PREMIS v3 schema before sending to AM/a3m, ensuring that it's well-formed and valid.

PREMIS files generated by Enduro child workflows should always be validated. A PREMIS file included in a transfer may have been validated in advance, so it might not be necessary to validate these. A reasonable approach might be to validate any PREMIS file in the SIP's metadata directory, regardless of origin, as this is the file that will be picked up by Archivematica/a3m.

Describe alternatives you've considered

None

Additional context

• There is a related task to generate a PREMIS file based on child workflow tasks - preprocessing-sfa#19
• Information about this activity should be visible in the Enduro UI, as described in Problem: child workflow tasks aren't visible in the UI #949
• I've added a hand-tooled, valid, multi-object sample PREMIS file in the comments

[sallain]

premis-annotated-multi.zip

[sallain]

Note: I've only listed validating against the schema as a first iteration. Other checks might include:

• Ensuring that each object listed in the PREMIS file exists in the transfer
• Ensuring that identifiers (for objects, events, and agents) are unique, and replacing them with unique IDs if needed
• Ensuring that there are no extra agents
• and possibly others!

[fiver-watson]

Note additionally that this is something that will be used repeatedly for any Enduro user performing custom ingest activities that might generate PREMIS, and/or anyone submitting their own PREMIS files with a SIP. For this reason, ideally this will be implemented as a reusable temporal activity, rather than a client-specific child workflow.

[fiver-watson]

@mcantelon also, as discussed in the meeting today:

Let's make this a general "Validate XML" task for its first pass, that can accept both a file to validate and a schema file to use for the validation as inputs.

[mcantelon]

PR for CR: artefactual-sdps/temporal-activities#21

[jraddaoui]

There are some comments about this issue in artefactual-sdps/preprocessing-sfa#22 (comment).

[sallain]

PREMIS file looks good and it's being properly parsed into the METS. I think we can finally put this issue to bed!