brawer / osmdiffs / 43025836778

94%

push

github

feat(wikidata): Count OSM features and names per brand (#928)

* feat(wikidata): Count OSM features and names per brand

collect_wikidata_ids already visits every ATP and OSM feature with a
brand:wikidata, operator:wikidata or network:wikidata tag. It now also
writes two files for the spider stats:

- osm-brands.tsv: OSM features per Wikidata ID, country and main
  category (such as shop=supermarket). A feature that names the same
  ID twice counts once.
- brand-names.tsv: the three most common brand, operator or network
  names per Wikidata ID, country and key, with how often OSM and ATP
  use each. Names count only where the Wikidata tag has a single ID.

The country of an OSM feature comes from its centroid, so features
with a Wikidata tag now get their geometry decoded.

Part of #924.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Ttfb2LhDRExioXkdHbxLno

* refactor(wikidata): Keep one display name per brand

The names are only for telling brands apart in the spider stats, so
brand-names.tsv now has one name per Wikidata ID and key, not three
per country: the most common name in Latin script, or else the most
common one. ATP places no longer need their country looked up.
osm-brands.tsv keeps its countries, so the spider stats can compare a
per-country spider with OSM in that country.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Ttfb2LhDRExioXkdHbxLno

---------

Co-authored-by: Claude <noreply@anthropic.com>

279 of 283 new or added lines in 2 files covered. (98.59%)

17721 of 18864 relevant lines covered (93.94%)

3502.83 hits per line