Add black box test for AIP reingest

[replaceafill]

This implements the AIP reingest scenarios from https://docs.google.com/document/d/1KGTsuc9NV5yaq-lUY7CVK3XY33C21zS3Ykr-N-6gpEg/edit#.

Connects to archivematica/Issues#561

[ross-spencer]

Nice simple test again. I've left two comments that you can choose to address for the two different queries. I wasn't sure about the fsentry method of retrieving a list of deleted files...

No reason not to merge once you're happy.

[ross-spencer]

Is this Xpath? Very cool. It seems like this query has reuse potential, I wonder if it can be extracted to a helper module somewhere?

def retrieve_filesec_by_use(mets_path, use):
   filesec_query = "'mets:fileSec/mets:fileGrp[@USE="{}"]/mets:file'".format(use)
   return reingest_mets.tree.findall(
               filesec_query,
               namespaces=context.mets_nsmap,,
      )

[replaceafill]

Great idea! I used this more generic approach so we can retrieve all the files as well (something we were already doing in the Create AIP tests and I updated those calls to use this):

def get_filesec_files(tree, use=None, nsmap={}):
    use_query = ""
    # an empty use parameter will retrieve all the files in the fileSec
    if use:
        use_query = '="{}"'.format(use)
    return tree.findall(
        "mets:fileSec/mets:fileGrp[@USE{}]/mets:file".format(use_query),
        namespaces=nsmap,
)

Let me know what you think.

[ross-spencer]

Could the deleted_files list be constructed the same way? fsentry for fsentry in initial_mets.all_files() if fsentry.use == "deleted"? If so, we should consider the xpath query above.

[replaceafill]

I tried this and it doesn't seem possible. Maybe the metsrw bases its file logic in the structMap and the deleted files are not listed there after the reingest.

[ross-spencer]

As the tests start to grow, it could be useful to have values like deleted, original, normalization in a controlled dictionary. But perhaps it isn't needed here? What do you think?

In my head, I'm also considering if that's something that could be found from metsrw? Or is that too circular a dependency when it comes to testing? Probably...

[replaceafill]

I agree. I started doing something like that in the transfer micro-services PR. Although the purpose of that is different I agree that the metsrw could help us abstract these constants.