Change missinginforeason to unspecifiedEntityDetails or unspecifiedPersonDetails

[StephenAbbott]

In version 0.2 of BODS, the missinginforeason field from version 0.1 was renamed.

The Register v2 code is still using missinginforeason rather than unspecifiedEntityDetails for entities or unspecifiedPersonDetails for people https://github.com/search?q=repo%3Aopenownership%2Fregister-v2%20missinginforeason&type=code

I suspect this is why the unknown_person.unknown_reason is not working in the Register v2 prototype.

Hopefully making this change will allow Register v2 to generate the person statements for unknown persons, and to assign the relevant text to explain why the person is unknown: https://github.com/openownership/register/blob/bb03cbc850c478b95276c8c21fc8ebd524b8d4e4/app/service_objects/bods_mapper.rb#L189 https://github.com/openownership/register/blob/bb03cbc850c478b95276c8c21fc8ebd524b8d4e4/spec/shared_contexts/bods.rb#L482

[tiredpixel]

I don't see any references to missingInfoReason on current register master branch (7df086eff6e072a3294b97b2f9dc330c7905a4a3). In addition, the only references to missingInfoReason I can find cross-repository is in register-sources-bods, where it is used for PepStatusDetails type. Despite the rename missingInfoReason to unspecifiedEntityDetails and unspecifiedPersonDetails in BODS 0.2, this does not appear to affect PEP status, so the current reference to missingInfoReason would seem to be correct.

Additionally, unspecifiedEntityDetails and unspecifiedPersonDetails appear in register-sources-bods, register-transformer-psc, register-transformer-sk, register-transformer-dk. register is using a version of register-sources-bods which has this code.

@StephenAbbott, is this issue still current, or has it already been fixed?

[StephenAbbott]

Hmmm. This is one where we maybe need a little more checking/conversation before I can say whether it has been fixed or not.

There is no sign of the unspecifiedEntityDetails and unspecifiedPersonDetails fields in the BODS fields appearing on Datasette for the full Open Ownership Register dataset https://bods-data-datasette.openownership.org/register

If the issue had been fixed and changes implemented in the BODS mappers, then I would expect that we'd be seeing the data coming through.

Might it be the case that we've made changes but still need to do a full import to generate this data, @spacesnottabs?

[spacesnottabs]

I think this is missing the correct implementation in the PSC transformer:

• https://github.com/openownership/register-transformer-psc/blob/main/lib/register_transformer_psc/bods_mapping/person_statement.rb#L61 This must have been missed when rewriting the mapper before the BODS v0.2 launch. Once this is fixed, reimporting the affected records which don't have the unknown person details will be necessary.

[tiredpixel]

Noting that no test cases were able to be found in the imported data. To verify this, the original PSC bulk data from 2023-11-01 was checked. No entity or person could be identified with unspecified details, however.

Additionally, the PSC bulk data link is currently not working, resulting in timeouts: https://download.companieshouse.gov.uk/en_pscdata.html This has been confirmed across multiple IPs and people.

Further work on this ticket is paused pending identification of test cases—i.e. entities and people for whom these unspecified details can be found in the raw data. Then, work will be needed to ensure those details get import into Elasticsearch raw indexes, and subsequently also to implement the missing parts to transform into BODS as noted above. Until these test cases are identified, however, nothing more can be done.

[StephenAbbott]

On hold pending @StephenAbbott being able to find examples

[maria-jofre]

An issue regarding 'unknownPerson' in the register has been identified based on an external enquiry. The case involves the BODS JSON for Novo Nordisk, where there is an “Unknown Person(s)” personStatement with id 5845298339200587547-unknown, listed as a UBO of Novo Nordisk. However, in the downloadable register, the 06-07 version does not include this entity.

Additionally, there are no other types of persons listed across all three registers in the OO register except for 'knownPerson' in the downloadable bulk data:

Kadie suspects that this issue is connected to the ongoing investigation and kindly suggested adding it here, as it might be the kind of example needed to test the fix.