brawer / osmdiffs / 42246208632

94%

push

github

feat(matchers): Tell brands by their website domains (#893)

Many OSM features link to their brand's website but lack
brand:wikidata. import_atp now writes alltheplaces.brand-domains.tsv,
the registrable domains that clearly belong to one brand in
AllThePlaces, and PoiMatcher uses it as a fallback: an OSM feature
without brand:wikidata whose website is on such a domain counts as
that brand, like an equal brand name. An ATP feature with only
operator:wikidata or network:wikidata also matches OSM features with
the same ID, but doesn't count for deletion coverage.

Adds URL normalization (Punycode, percent escapes, tracking
parameters, Public Suffix List domains), which the next step will
also use for matching URLs as contacts.

On Switzerland: 377 new matches and 9 moved to a closer feature, none
lost. On Japan: 408 new and 10 moved, none lost. See
docs/notes/2026-11-06-brand-domains.md.


Claude-Session: https://claude.ai/code/session_01Ttfb2MGwS3XXMu6tBsrXLX

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

676 of 681 new or added lines in 6 files covered. (99.27%)

15382 of 16247 relevant lines covered (94.68%)

4026.06 hits per line