Clean up Duplicate DOIs

[CarsonDavis]

Overview

Prior to the DOI Ingest Logic refactor, if the DOI sync was run multiple times, it was possible for duplicate DOI drafts to be created in the system. These will show up as unpublished update drafts. We want to remove all these unnecessary DOIs before running any new DOI update code.

Notes

We don't want to delete any work in progress by curators, so we need to check with them that we don't run the script on DOIs that they are legitimately working on and are unpublished.

Acceptance Criteria

• script that loops through each of the campaigns and finds and deletes any associated unpublished DOI drafts
• don't delete any in-progress curator work

[CarsonDavis]

Here is some dirty example code

aces_uuid = str(Campaign.objects.get(short_name='ACES').uuid)
matching = []
for doi in Change.objects.of_type(DOI).all():
    if aces_uuid in doi.update['campaigns']:
        matching.append(doi)
len(matching)

for doi in matching:
    print(doi.update['concept_id'], doi.status)

# 6 published 
# 4 unpublished

# delete the unpublished ones
# if the status is specifically 0, then that means that no curator has done any work on the doi
for doi in matching:
    if doi.status == 0: 
        doi.delete()

[CarsonDavis]

from data_models.models import DOI, Campaign
from api_app.models import Change
created_dois_drafts = Change.objects.of_type(DOI).filter(status=0)
published_campaign_uuids = [str(uuid) for uuid in Campaign.objects.values_list("uuid",flat=True)]
dois_to_delete = []
for doi in created_dois_drafts:
    if any(uuid in doi.update.get('campaigns', []) for uuid in published_campaign_uuids):
        dois_to_delete.append(doi)
len(dois_to_delete)