nx_phrase_match.nx

nx_phrase_match.nx -- case-insensitive substring matcher. Tiny utility used by the Tier-2 element extractors (nx_outfit_extract / nx_location_extract / nx_domain_classify) when scanning beat text for known phrases. Why dedicated and not nx_str_find: the canonical nx_string_ops search is case-sensitive (matches C strstr semantics) and pulls in a longer import chain. This file is a small, self-contained case-INSENSITIVE substring detector that handles ASCII upper / lower equivalence inline. Limitations (deliberate): - ASCII only. Multi-byte UTF-8 case folding is NOT done; CJK phrases are matched byte-exact (they have no case anyway). - Word-boundary aware via optional require_boundary flag: if true, match only when neither end is adjacent to an ASCII letter / digit (prevents "apron" matching inside "apronk"). Per the bounded-loop cardinal: outer + inner loops both ride explicit JPL Rule 2 BUDGETS. nx_safety_envelope: intended_use: "Case-insensitive substring search inside a byte range, used by element extractors." sil_target: SIL2 asil_target: QM dal_target: DAL C iec_62304_class: NONE evidence: [no_floating_point, bounded_outer_loop_jpl_rule_2, bounded_inner_loop_jpl_rule_2, ascii_only_case_folding_documented] hazard_register: [bug-tape-utf8-case-fold-not-supported, bug-tape-boundary-check-permissive-on-punct] residual_risk: "ASCII-only fold; CJK phrases match byte-exact. Boundary check treats non- alphanumeric (incl. punctuation) as a boundary -- intended." verdict: NOT_YET_EVALUATED