Implement support for migrating RDF serializations other than turtle

[cjcolvar]

What does this Pull Request do?

The upgrade-utils has an option for source rdf language (source-rdf) but in reality only turtle was supported. This PR adds support for other rdf serializations and tests with ntriples.

Jira Ticket

https://fedora-repository.atlassian.net/browse/FCREPO-3989

What's new?

• Added getSrcRdfExt() convenience method to Config
• In F47ToF5UpgradeManager:
  • Replaced TURTLE_EXTENSION with new method
  • Removed TURTLE_EXTENSION contsant
  • Replaced hard-coded Lang.TTL references to getSrcRdfLang()
• DRY'd up F5ToF6UpgradeManager by using new method
• Added tests and test fixtures for ntriples

How should this be tested?

Try exporting from fedora 4 or 5 in ntriples (or another rdf serialization) and running the upgrade utils.

Interested parties

@whikloj

[whikloj]

Thanks for this @cjcolvar.

I'm having a little trouble with my 4.7.5 export, it seems to be lacking some information. But while I figure that out, could I get you to add a better test of the -r, --source-rdf argument. Currently if you pass a bad one (I was trying NT and NTRIPLES to begin) we just default to text/turtle and move on.

fcrepo-upgrade-utils/src/main/java/org/fcrepo/upgrade/utils/UpgradeUtilDriver.java

Lines 142 to 144 in 9242276

	if (cmd.hasOption("source-rdf")) {
	config.setSrcRdfLang(RDFLanguages.contentTypeToLang(cmd.getOptionValue("source-rdf")));
	}

Something like

if (cmd.hasOption("source-rdf")) {
    final Lang lang = RDFLanguages.contentTypeToLang(cmd.getOptionValue("source-rdf"));
    if (lang == null) {
        printHelpAndExit(format("invalid RDF content-type (%s) provided", cmd.getOptionValue("source-rdf")),
            configOptions);
    }
    config.setSrcRdfLang(lang);
}

[whikloj]

@cjcolvar I also figured out part of my issue, which might be that there are more spots assuming text/turtle in this codebase.
https://fedora-repository.atlassian.net/browse/FCREPO-3992

I'm good with this work, it does was it says. If you could add the rdf type check so people don't specify something and just get turtle then I'll merge.

But just be aware that the Fedora 4 to 5 "upgrade" might look like it is working but doesn't. It then fails to allow you to upgrade from Fedora 5 to 6.

[cjcolvar]

Thanks @whikloj for the quick review! I added the check you suggested.

FWIW I think my upgrade to 6 worked correctly with these changes, but maybe my dataset is different from yours and wasn't hitting the issue you are seeing.

[whikloj]

👍