jqueguiner / num2words2 / 25218394358 / 4

95%

main: 95%

push

github

feat(en_aero): ICAO/aviation English digit-by-digit reading (#114)

Implements savoirfairelinux/num2words#478 — a new lang='en_AERO'
that reads numbers digit-by-digit using ICAO Annex 10 vol II
respellings (3→tree, 4→fower, 5→fife, 7→seven, 9→niner). Goal is
unambiguous transmission over noisy radio links, matching how pilots
and ATC actually enunciate numerals on voice channels.

  num2words(5739, lang='en_AERO')        # 'fife seven tree niner'
  num2words(1971, lang='en_AERO', to='year')  # 'one niner seven one'
  num2words(127.5, lang='en_AERO')       # 'one two seven decimal fife'
  num2words(-42, lang='en_AERO')         # 'minus fower two'

Implementation:
  - num2words2/lang_EN_AERO.py — Num2Word_EN_AERO inherits from
    Num2Word_EN, overrides to_cardinal/to_year for digit-by-digit
    reading, and delegates ordinal / fraction / currency / cheque
    to a sibling Num2Word_EN instance (ICAO doesn't standardise
    those, and otherwise the digit-by-digit cardinal would leak
    into the ordinal/fraction builders and produce broken output
    like "one treeth" instead of "one third").
  - Registered in CONVERTER_CLASSES under 'en_AERO'; resolves
    case-insensitively and tolerates dash/underscore separators
    via the existing dispatcher fallback ('en-aero', 'en_aero',
    'en-AERO' all map to the same converter).

Decimal mark reads as "decimal" (ICAO standard).
Negative numbers prefix with "minus".

Tests: tests/test_478_en_aero.py — 13 cases covering the digit
table, multi-digit composition, year format, decimals, negatives,
string/dash/underscore lookup variants, and the delegation path
for fractions (where 'en_AERO' falls back to plain English so the
fraction reads "one third" not "one treeth").

Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

8010 of 8796 branches covered (91.06%)

Branch coverage included in aggregate %.

42260 of 44010 relevant lines covered (96.02%)

0.96 hits per line