pids: cannot publish draft because of existing DOI, missing from pids metadata

[slint]

For some reason, users can end up in a state where a draft doesn't have any DOI information in its Draft.pids dictionary field data in the DB, but there is a DOI row for the managed DOI of the draft. When the user tries to publish the record (either directly or by accepting a review request), they get a "The persistent identifier is already registered" validation error on the upload form.

We should make the PID components/manager a bit more flexible in such cases, so that if the PID already exists in the DB (in the correct state), there is no validation error, but instead the PID model creation step is skipped, and the existing PID is used.

---

To fix this one would have to run a script like this:

from invenio_rdm_records.proxies import current_rdm_records_service as service
from invenio_pidstore.models import PersistentIdentifier, PIDStatus
from invenio_db import db

RECID = "123456"

draft = service.draft_cls.pid.resolve(RECID, registered_only=False)
assert "doi" not in draft.pids, f"Draft already has a DOI: {draft.pids}"

doi = PersistentIdentifier.query.filter_by(pid_type="doi", object_uuid=draft.id, object_type="rec").one_or_none()
assert doi and doi.status == PIDStatus.NEW, f"DOI {doi.pid_value} doesn't have status NEW: {doi.status}"

# Delete the DOI
db.session.delete(doi)
db.session.commit()

[github-actions]

This issue was automatically marked as stale.