implement functions for DOI creation and metadata upload

[mbjones]

Once we have DataCite metadata available from #311, we need to be able to mint a new DOI for that item and register it with EZID. The steps involved will likely include:

• API function to create a new DOI-labeled resource
• reserve that DOI in EZID and register it in the identifiers table in vegbank (status=reserved)
• generate metadata for that resource (create functions to generate DataCite metadata #311)
• update the EZID record with DataCite metadata
• make the DOI public and indexable on EZID

Any time the core metadata for a vegbank-published item changes, we may need to either 1) update the DOI metadata with EZID, or 2) issue a new DOI with that metadata that is versioned as a replacement for the exissting DOI, assuming that we want it to reflect a versioned change in the resource. Need to discuss the situtations in VegBank that represent mutable versus versioned DOI metadata changes.

The EZID group provides a loose API script that can be used to interact with EZID, but it it would be best if it were refactored into a standalone python package that could be imported into vegbank. Some of the above functions could be members of an ezid python package rather than part of the vegbank2 API package per se, TBD during design discussions.

For tooling, see:

• https://github.com/CDLUC3/ezid-client-tools
• https://ezid.cdlib.org/doc/apidoc.html#python-command-line-tool

[mbjones]

See also: datacite python package: https://datacite.readthedocs.io/en/latest/

@rushirajnenuji @datadavev do you know if EZID supports the datacite API for submissions?

[datadavev]

EZID can generate DOIs registered with Crossref or DataCite. When generating for DataCite a complete DataCite metadata document in XML format may be included. So EZID uses the DataCite API for submissions, but does not expose the EZID api like a proxy or reflection of DataCite. Updates to metadata are generally pretty quick, depending on the service load.