Introduce Funder Data to the core

[ajnyga]

This issue is for introducing Funder Data to the application core. This metadata has been handled before mainly with the Funding Plugin.

The new feature will introduce a manager component for handling Funder Data. It will use ROR's as data source like we use already with affiliations. This is a change from the Funding Plugin which uses the Open Funder Registry. However, it will be soon deprecated, see https://ror.readme.io/docs/funder-registry

• Workflow Metadata settings, allow editors to enable Funder Data field
• Eloquent class for Funders
• API endpoint for Funders
• UI-library Vue components
  • Funder Manager component for listing Funders
  • Funder Form component for adding/editing Funders. Separate FieldFunder and FieldFunderGrant vue fields.
• Submission Wizard allows adding Funders
• Editorial dashboard allows editing/adding Funders
• Landing page shows Funders
• Grant ID validation based on Zenodo API (BUT maybe this should be a validation plugin and not in the core. It maybe does not make sense that the core has EU specific code and this would be easy to solve with a plugin with validation hooks already existing OR maybe add this validation to the existin Zenodo plugin?)

ojs: pkp/ojs#5378
pkp-lib: #12397
ui-library: pkp/ui-library#813

crossref-ojs: pkp/crossref-ojs#88
jatsTemplate: pkp/jatsTemplate#102
pflPlugin: pkp/pflPlugin#65
zenodo: pkp/zenodo#5

Things to consider

• Review of Manager component and funder form UI
• Review of publication landing page layout for Funder Data
• Support for Funder Data in Native import and export
• Support for Funder Data in exports like JATS, ONIX, Crossref, Datacite
• Support for Funder Data in plugins that are dependent on Funding Plugin, like pfl
• Migration of Funding plugin data to the new Funder Data model

Migration considerations

Migrating the Funding Plugin data to the new core Funder Data is the most complex thing to solve if we want to support migration from the Funding plugin. Some considerations in the ROR documentation see https://ror.readme.io/docs/funder-registry.

I can see two approaches:
- In the migration try to map the existing DOIs to RORs. This can be tricky. See comments below, there might a fairly easy way of doing this using a csv file
- Do not map DOIs to RORs. Instead allow the database to store either a DOI or an ROR as the identifier and keep the DOIs from the plugin data as they are. This is probably the ideal way to go.

[kaitlinnewson]

Hi @ajnyga, I wanted to note that for the migration approaches, if you do look into mapping the existing DOIs to RORs, they can be retrieved via the ROR API - here's an example from the Zenodo plugin for 3.6: https://github.com/pkp/zenodo/blob/0e96285182f9117267cd843435254ccd11b39d83/filter/ZenodoJsonFilter.php#L561

[ajnyga]

Thanks! Yes I wrote an initial migration script, will add the whole initial pr here soon.

What I am thinking is that if we would have this in the upgrade script, it would mean hundreds of api calls in a short time with maybe a high probability of errors and broken upgrades. The other option is to use a datadump for the ROR data that they have available, but those are very large files. Of course since we do not need the names, just the identifiers, we could make the file smaller https://ror.readme.io/docs/funder-registry#get-a-csv-file-of-all-funder-id-to-ror-id-mappings

[ajnyga]

ok so by removing all the unnecessary columns and just leaving three columns: ROR, funder identifiers and preferred funder identifier and the removing the organisations that do not have a funder identifier at all, we end up with a 800kb file which I think could work very well in the migration. (you can also see that there are at least two errors in the data right in the beginning :)

ror-to-doi-mapping-funders-only.csv

The thing is that I do not think we have a source for this data that we can fetch during the upgrade so we would have to use this self-made file and that means that any changes coming to the data afterwards are not taken into account.

One option could be that I maintain the file in the Funding Plugin repository and it is fetched from there.

edit: oh realized that while converting from DOI to ROR we do not actually need the preferred column, so the file would be even smaller.