[sakshamarora1]

closes: #370

1. Renamed cds_ref -> cdsrn
2. Added component to mint cdsrn pid_type

Steps:

1. Merge this PR
2. Run the following scripts (It will automatically mint pids for the existing ones as it uses the service layer.

# Script to update cds_ref to cdsrn and mint cdsrn
from invenio_access.permissions import system_identity
from invenio_pidstore.errors import PIDAlreadyExists
from invenio_db import db
from invenio_drafts_resources.services.records.errors import DraftNotCreatedError
from invenio_rdm_records.proxies import current_rdm_records_service as records_service

from cds_rdm.components import MintAlternateIdentifierComponent

alternate_identifier_component = MintAlternateIdentifierComponent(records_service)

# Drafts (also, works for new versions)
hits = records_service._search(
    identity=system_identity,
    action="scan",
    params={"q": "metadata.identifiers.scheme:cds_ref", "allversions": True, "include_deleted": True},
    search_preference=None,
    record_cls=records_service.draft_cls,
    search_opts=records_service.config.search_drafts,
    permission_action="read_draft",
).scan()
for hit in hits:
    draft = records_service.read_draft(system_identity, hit['id'])
    for identifier in draft.data["metadata"].get("identifiers", []):
        if identifier.get("scheme") == "cds_ref":
            identifier["scheme"] = "cdsrn"
    updated_draft = records_service.update_draft(system_identity, draft.id, draft.data)
    print(f"INFO: Draft <{updated_draft.id}> updated with cds_ref to cdsrn.")

# Records, run after deploying changes: 1. cds_ref to cdsrn, 2. service component to mint cdsrn
hits = records_service.scan(system_identity, params={"q": "metadata.identifiers.scheme:cds_ref", "allversions": True, "include_deleted": True})
for hit in hits:
    record = records_service.read(
        system_identity, hit["id"], include_deleted=True
    )
    try:
        draft = records_service.read_draft(system_identity, record.id)
        # Step 1: Update cds_ref to cdsrn in the record via low-level API
        print(f"INFO: Record <{record.id}> has an existing draft <{draft.id}>! Updating record via low-level API.")
        # Update the record directly without affecting the draft
        for identifier in record._record["metadata"].get("identifiers", []):
            if identifier.get("scheme") == "cds_ref":
                identifier["scheme"] = "cdsrn"
        # Save the record changes and reindex
        record._record.commit()
        db.session.commit()
        records_service.indexer.index(record._record)
        # Run service component to mint cdsrn
        alternate_identifier_component.publish(system_identity, draft=draft, record=record)
        print(f"INFO: Record <{record.id}> updated with cds_ref to cdsrn and cdsrn minted.")

        # Step 2: Update cds_ref to cdsrn in the draft and save draft
        draft = records_service.edit(system_identity, draft.id)
        draft._record.fork_version_id = record._record.revision_id
        for identifier in draft.data["metadata"].get("identifiers", []):
            if identifier.get("scheme") == "cds_ref":
                identifier["scheme"] = "cdsrn"
        updated_draft = records_service.update_draft(
            system_identity, draft.id, draft.data
        )
        print(f"INFO: Draft <{updated_draft.id}> updated with cds_ref to cdsrn.")
    except DraftNotCreatedError:
        # If the draft didn't exist, we simply edit and publish the record
        draft = records_service.edit(system_identity, record.id)
        for identifier in draft.data["metadata"].get("identifiers", []):
            if identifier.get("scheme") == "cds_ref":
                identifier["scheme"] = "cdsrn"
        updated_draft = records_service.update_draft(
            system_identity, draft.id, draft.data
        )
        record = records_service.publish(system_identity, updated_draft.id)
        print(f"INFO: Record <{record.id}> updated with cds_ref to cdsrn and cdsrn minted.")

[kpsherva]

not sure if this will cover the case:

1. user adds report number + another identifier, let's say legacy recid
2. report number has a duplicate -> throws a validation error.

legacy recid is not remembered in the form, is it? all the alt identifiers will be wiped out if I am reading this correctly.

[sakshamarora1]

I have updated the code, so this should not happen anymore, added a test case too.

[kpsherva]

do we need this utitlity ? can we use the .create directly
@ntarocco side question: in this case service (via component) access the data layer directly. Are we OK with that or should we find another way to manage it?

[ntarocco]

I don't think that there are much alternatives, because we don't have a data layer exposing all the needed APIs. Most of our data layer objects are DB models that allows direct access, they are not wrapped exposing rich APIs.
Do you have other ideas?

[ntarocco]

Thanks for adding all these tests, super good.
Can you please only change the identity, and always use the normal user and not system or admin whenever possible?

[kpsherva]

check how long the scan should be

[kpsherva]

on the draft lets mint with "reserved" status

[palkerecsenyi]

Looks good! I appreciate the nicely commented tests :)

[palkerecsenyi]

Maybe we could extract this into a shared method to reduce code duplication? But not really a huge deal since it's only used twice.

[kpsherva]

didn't we say to have CDS report number as a name in the end? @ntarocco

[ntarocco]

I would say so, what was the outcome to this question in your interviews? (pinging @zzacharo too)

[zzacharo]

From LHCb I remember that in their system the column name was "CDS reportnumber" but in their team they were calling it actually CDS reference number...

[kpsherva]

to be scheduled for next deployment

[ntarocco]

Great, thanks for the fix above, in the next PR, can you please add something like (pseudocode):

assert pid.pid_type not in [recid, doi, oai]

It is just a safety check!

[sakshamarora1]

Added!