@felix-tech/singlish-js

Zero-dependency TypeScript library that converts romanised Sinhala (Singlish) to Sinhala Unicode. Ships as dual CJS + ESM with full TypeScript declarations. < 1 KB gzipped.

Installation

npm install @felix-tech/singlish-js
# or
pnpm add @felix-tech/singlish-js
# or
yarn add @felix-tech/singlish-js

Quick start

import { transliterate } from "@felix-tech/singlish-js";

transliterate("mama"); // → "මම"
transliterate("mama gamata yanawa"); // → "මම ගමට යනව"
transliterate("keels paan"); // → "කීල්ස් පාන්"
transliterate("shrii"); // → "ශ්‍රී"
transliterate("sathYa"); // → "සත්‍ය"

API

transliterate(input: string): string

Converts a Singlish string to Sinhala Unicode. Non-Latin characters (digits, punctuation, spaces) pass through unchanged. The function never throws.

transliterate("mama gamata yanawa 123!"); // → "මම ගමට යනව 123!"
transliterate(""); // → ""

Transliteration reference

The golden rule — inherent vowel

Single a after a consonant is the inherent vowel — it produces a bare consonant with no modifier. To write the long ā (ා) you must type aa.

ma   → ම
maa  → මා
mama → මම

Vowels

Independent (word-initial / standalone)

| Input | Sinhala | | ------------------ | ------- | | a | අ | | aa | ආ | | A | ඇ | | Aa | ඈ | | ae | ඈ | | i | ඉ | | ii / ie / ee | ඊ | | u | උ | | uu / oo | ඌ | | e | එ | | ea / ei | ඒ | | o | ඔ | | oe | ඕ | | ai | ඓ | | au | ඖ |

Dependent modifiers (after a consonant)

Same input triggers the corresponding pilla instead of an independent vowel.

ka  → ක
kaa → කා
kA  → කැ
kAa → කෑ
ki  → කි
kii → කී
ku  → කු
ke  → කෙ
ko  → කො
koe → කෝ
kea → කේ

Base consonants

| Input | Sinhala | | Input | Sinhala | | | ----- | ------- | --- | --------- | ------- | --- | | k | ක | ka | n | න | na | | g | ග | ga | p | ප | pa | | j | ජ | ja | b | බ | ba | | t | ට | ṭa | m | ම | ma | | d | ඩ | ḍa | y | ය | ya | | sh | ශ | śa | r | ර | ra | | ch | ච | ca | l | ල | la | | th | ත | ta | v / w | ව | va | | dh | ද | da | s | ස | sa | | h | හ | ha | f | ෆ | fa | | kh | ඛ | kha | bh | භ | bha | | gh | ඝ | gha | ph | ඵ | pha | | q | ඣ | jha | | | |

Special / aspirated consonants (case-sensitive)

Uppercase letters invoke the aspirated or special form.

| Input | Sinhala | vs lowercase | | ----- | ------- | ------------ | | Sh | ෂ | sh → ශ | | Ch | ඡ | ch → ච | | Th | ථ | th → ත | | Dh | ධ | dh → ද | | K | ඛ | k → ක | | G | ඝ | g → ග | | T | ඨ | t → ට | | D | ඪ | d → ඩ | | P | ඵ | p → ප | | B | ඹ | b → බ | | N | ණ | n → න | | L | ළ | l → ල | | GN | ඥ | jña | | KN | ඤ | ña |

transliterate("Shri"); // → "ෂ්‍රි"
transliterate("shri"); // → "ශ්‍රි"
transliterate("aBa"); // → "අඹ"
transliterate("amba"); // → "අම්බ" (phonetic: a + m + ba)

Sanyaka (prenasalised) consonants

| Input | Sinhala | | ------ | ------- | | nng | ඟ | | nnd | ඬ | | nndh | ඳ |

Conjunct consonants

Conjuncts use two distinct triggers:

• Rakaransaya — lowercase r after a consonant: krama, shrii
• Yansaya — uppercase Y after a consonant: sathYa, aachaarYa

Lowercase y is always plain ය — no conjunct binding.

| Input | Sinhala | | ----------- | ------- | | krama | ක්‍රම | | shrii | ශ්‍රී | | sathYa | සත්‍ය | | aachaarYa | ආචාර්‍ය | | sathya | සත්ය |

Standalone glyphs via backslash sequences

| Input | Sinhala | | ----- | ------- | | \n | ං | | \h | ඃ | | \N | ඞ | | \R | ඍ |

transliterate("sitha\\n"); // → "සීතං"

Unicode output

All output codepoints are in the Sinhala Unicode block (U+0D80–U+0DFF), with the exception of U+200D (Zero Width Joiner) used to form conjunct ligatures.

TypeScript

The package ships .d.ts declarations. No @types/ package needed.

import type { transliterate } from "@felix-tech/singlish-js"; // fully typed

Use cases

• Live input fields — wire transliterate to an oninput handler to render Sinhala as the user types romanised text.
• IME / keyboard apps — use as the core conversion engine behind a mobile or browser-based Sinhala input method.
• Content pipelines — bulk-convert legacy romanised Sinhala datasets to proper Unicode.
• CLI tooling — generate Sinhala strings in i18n files or database seeds without a system IME.
• Search preprocessing — expand romanised queries to Unicode before querying Sinhala document indices.

License

MIT