@pcampus/thai-romanization

Rule-based Thai → Latin romanization library. No per-word hardcoded overrides in the core pipeline — segmentation uses a morpheme trie, orthography rules, and lookup tables for loanwords and provinces.

📖 คู่มือการใช้งานฉบับเต็ม: USAGE.md

Install

npm install @pcampus/thai-romanization

Quick start

import { romanize, romanizeSentence } from '@pcampus/thai-romanization';

romanize('สวัสดี');           // 'sawatdi'
romanize('กษัตริย์');         // 'kasatri'
romanize('มกราคม');           // 'mokkarakhom'

romanizeSentence('สวัสดี ครับ'); // 'sawatdi khrap'

API

| Function | Description | |----------|-------------| | romanize(text, options?) | Romanize a single word or compound | | romanizeSentence(text, options?) | Romanize text, preserving spaces and punctuation | | romanizeDetailed(text, options?) | Full pipeline debug output | | romanizeBatch(words[], options?) | Romanize multiple words | | reverseLookup(latin) | Latin → Thai candidates (vocabulary, morphemes, loanwords, provinces) | | segment(text) | Trie-based word segmentation | | splitIntoSyllables(text) | Syllable-level split | | getRomanizationUnits(text) | Internal romanization units |

Options

romanize('สวัสดี', { separator: '-', tone: 'number' });
// separator: '' | '-' | ' '  (default: '')
// tone: 'omit' | 'diacritic' | 'number'  (default: 'omit')

Vocabulary packs

The trie dictionary is split into layered packs:

import { coreVocab, geoVocab, calendarVocab, mergeVocabulary } from '@pcampus/thai-romanization';

const custom = mergeVocabulary(coreVocab, geoVocab);

Optional subpath imports:

import { geoVocab } from '@pcampus/thai-romanization/packs/geo';
import { calendarVocab } from '@pcampus/thai-romanization/packs/calendar';
import { coreVocab } from '@pcampus/thai-romanization/packs/core';

CLI

npx thai-romanize "สวัสดี"
npx thai-romanize --sentence "สวัสดี ครับ"
npx thai-romanize --reverse software
npx thai-romanize --separator - --tone number "มกราคม"

Pipeline

romanize(text, options?)
  ├─ loanword?     → LOANWORD_MAP
  ├─ province?     → PROVINCE_ROMAN
  └─ segment + rules
       ├─ applySyllableBoundaries()
       ├─ Trie longest-match (vocabulary packs)
       ├─ monthRules (-าคม → khom/akhom)
       └─ applyTone()

Development

npm test          # vitest
npm run build     # tsup → dist/
npm run playground  # Next.js demo UI

Publish

npm login
npm publish

License

MIT