brawer / osmdiffs / 40199920122

94%

push

github

feat(matchers): match by phone number and email address (#850)

* feat(matchers): match by phone number and email address

PoiMatcher now also matches when the ATP feature and the OSM candidate
share a descriptive phone number or email address, and their main
categories are equal or a known pair (src/matchers/contact/
category_pairs.tsv). A brand:wikidata conflict still vetoes the match.
Beyond 150 m, a contact match also needs the street address to agree.

Restaurants, hotels, schools and fuel stations now get a matcher too,
and more categories (cafes, fast food, bars, schools, clinics, doctors,
fitness centres, museums, healthcare=*) get a match mask.

The phone and email parsers move from pipeline::atp::nondescript to
matchers::contact, so both sides are parsed alike. OSM numbers in
national format go through the phonenumber crate, in the ATP feature's
country. See docs/notes/2026-10-25-phone-and-email-matching.md.

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01P43kACSyJVwszya8Z3d4qQ

* fix(deps): allow atomic-polyfill in cargo deny advisories

Ignore RUSTSEC-2023-0089 (atomic-polyfill is unmaintained), which
phonenumber pulls in through postcard and heapless 0.7, only for
targets without native atomics.

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01P43kACSyJVwszya8Z3d4qQ

* fix(matchers): parse single OSM phone numbers without an iterator

parse_osm_phone now calls phones::parse_phone_number directly instead
of taking the first item of parse_phone_numbers. With build-mode none,
CodeQL resolved that `.next()` to unrelated iterators in tables::graph
and reported a cleartext-logging flow that does not exist.

Also fill in the planet match count of this change in the note.

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01P43kACSyJVwszya8Z3d... (continued)

450 of 466 new or added lines in 7 files covered. (96.57%)

11959 of 12755 relevant lines covered (93.76%)

1381.13 hits per line