forzagreen / n2words / 20760301385

99%

main: 97%

Build Type

push

github

Committed by web-flow

Commit Message

perf: add segment-based ScaleLanguage for 10x faster conversion (#205)

Replace GreedyScaleLanguage with a new segment-based architecture that uses
O(1) property lookups instead of O(n) array scans, achieving ~10x faster
number-to-words conversion.

## New Base Classes

- **ScaleLanguage**: Segment-based decomposition for most Western languages
  - O(1) vocabulary lookups using direct property access
  - Implicit one omission flags: omitOneBeforeHundred, omitOneBeforeThousand, omitOneBeforeScale
  - thousandWord property for languages that separate thousand from scale words

- **CompoundScaleLanguage**: Long scale with "thousand + previous" pattern
  - Pattern: million (10^6), thousand million (10^9), billion (10^12)
  - Used by: French, Portuguese, Spanish

- **MyriadLanguage**: 4-digit 万-based grouping for East Asian languages
  - Groups by 10,000 instead of 1,000
  - Used by: Japanese, Korean, Simplified Chinese, Traditional Chinese

## Languages Migrated (24 total)

ScaleLanguage: Amharic, Amharic Latin, Azerbaijani, Danish, Dutch, English,
Filipino, Finnish, German, Greek, Hausa, Indonesian, Malay, Norwegian Bokmål,
Swedish, Turkish

CompoundScaleLanguage: French, French Belgium, Portuguese, Spanish

MyriadLanguage: Japanese, Korean, Simplified Chinese, Traditional Chinese

## Removed

- GreedyScaleLanguage class (superseded by ScaleLanguage)
- TurkicLanguage class (merged into ScaleLanguage with omission flags)

## Other Changes

- Add groupByFours and placeValuesFour utilities for myriad mode
- Update add-language.js scaffolding to default to ScaleLanguage
- Store benchmark results in ~/.n2words-bench/ for persistence
- Support comma-separated language scopes in commitlint

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>

Run Details

1456 of 1535 branches covered (94.85%)

Branch coverage included in aggregate %.

3323 of 3354 new or added lines in 25 files covered. (99.08%)

2 existing lines in 2 files now uncovered.

9599 of 9679 relevant lines covered (99.17%)

131.54 hits per line