committed 04:59AM - 16 Apr 26 UTC
- Fix case-sensitivity bug in filter_by_name by adding
processor=fuzz_utils.de…fault_process to rapidfuzz call;
lowercase queries like 'darmstadt' now match correctly
- Switch scorer from token_set_ratio to token_sort_ratio,
fixing word-order variations (e.g. 'Koeln Bonn' -> 'Köln/Bonn');
zero regressions across 1281 stations, 149 improvements
- Lower default match threshold from 0.9 to 0.8 to handle
single-character typos and common shorthands while keeping
100% precision
- Expose name_threshold in StationsRequest and ValuesRequest
models, CLI --name-threshold option (stations + values),
and fix stale 0.9 defaults in stripes endpoints
- Update tests, docs and CHANGELOG
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>