Repository navigation
[ISIN] Hyphen-separated input tolerance via community extension #180
Description
Follow-up to #177 (ISIN capability, paxman/capabilities/ISIN/).
Status quo: hyphenated input such as US-037833100-5 yields MISSING by design — the recognition pattern (paxman/capabilities/ISIN/grammar/isin_recognition.py) accepts only compact and single-space-grouped bodies, with zero code-level tolerance for hyphens. Documented as DEFER in docs/user/capabilities/isin.md.
Ask: add an opt-in community grammar (registered via register_grammar, enabled through the existing extra_grammars contract field) that recognizes hyphen-grouped ISIN mentions and normalizes them to the same compact canonical form. Core grammar stays hyphen-free.
Acceptance sketch: canonicalize('US-037833100-5', contract_with_extra_grammar) → SUCCESS US0378331005; default contract still MISSING; unit + integration coverage mirroring the space-grouped rows in tests/integration/test_isin_pipeline.py.
Activity
azaharizaman commentedon Sep 30, 2026
Triage 2026-09-30: keeping open, labeled v1.0.0 + enhancement + followup (follow-up to PR #177). Plain English: ISINs written with hyphens (US-037833100-5) are currently declined; this adds an opt-in add-on that reads hyphenated forms, with default behavior unchanged. Groups with #183 (same hyphen pattern for LEI) but kept separate per capability. No consolidation needed.