[ZA] Investigate why Andrew Ntwampe Morwamoche's profile is showing appearances from Mr Morwamoche of a different party from many years ago

[JenMysoc]

From PA:

Mr Andrew Ntwampe Morwamoche is a member of the PAC (candidate for the 2014 election). The appearances on his profile reflect that of a Mr Morwamoche from the ANC who seemed to have been a MP years ago. Is there a way to delete the appearances from this profile. See link below.

http://www.pa.org.za/person/morwamoche-andrew-ntwampe/

[mhl]

This is a name resolution error; generally we delete the erroneously assigned speeches and add an exception to the code to stop them be reassigned, as in this issue:

• remove misattributed committee appearances for Mr Madoda Sambatha and Ms Beryl Belinda Miller #1718

In this case, Mr Andrew Ntwampe Morwamoche from PAC doesn't appear to hold a position that would allow him to speak in parliament, so the same could be done in this case by a developer.

(We're resisting getting caught up in fixing the underlying issues in the name resolution until something like #1748 is done.)

[duncanparkes]

Misattributed speeches removed with

from pombola.core.models import Person as CorePerson
from speeches.models import Speaker, Speech
CorePerson.objects.get(slug='morwamoche-andrew-ntwampe').__dict__
person = Speaker.objects.get(person__popit_id='core_person:8277')
Speech.objects.filter(speaker=person).count()
[x.text for x in Speech.objects.filter(speaker=person)]
Speech.objects.filter(speaker=person).delete()

in the manage.py shell.

[duncanparkes]

It looks like Morwamoche Andtrew Ntwampe is still getting things attributed to him despite us blacklisting his popit ID mysociety/za-hansard@14c0f13

[mhl]

The committee appearance is on date datetime.date(1999, 9, 21) committee ID 87 and meeting ID 3195, so we investigated this by putting in a breakpoint in the if options['import_to_sayit'] block and single-stepping. self.get_person in za_hansard/importers/import_json.py is called with display_name Morwamoche and party ANC. That returns a speaker with speaker.person.popit_id of core_person:8277 even though that is in the popit_id_blacklist.

What's happening is that this the blacklist is only used to reject results from self.resolver.get_person - if a speaker's not found there, it still goes on to try to find a unique speaker with that name from the Speaker model at this line: https://github.com/mysociety/za-hansard/blob/32664ad25f6b38da7a898ce9b0bde4cd552b95af/za_hansard/importers/import_base.py#L49

There's a fix here:

mysociety/za-hansard#69

... which can then fix the misassignment on the live site by running:

./manage.py za_hansard_pmg_api_scraper --committee=87 --meeting=3195 --scrape --save-json --delete-existing --import-to-sayit --commit

[duncanparkes]

I've deployed the site with the new za-hansard, and reparsed the meeting as @mhl suggested.