Thai Keyboard Corrector — TypeScript Edition

Fixes text typed with the wrong layout between Thai Kedmanee 🆚 English QWERTY.

Install

npm install thai-keyboard-corrector      # or -ts if name taken

Quick Use

import { correct } from "thai-keyboard-corrector";

// Clear cases (≥70% of one script)
console.log(correct("l;ylfu")); // สวัสดี
console.log(correct("ฟหกด")); // asdf
console.log(correct("hello world")); // ้ำสสน ไนพสก

// Unclear cases (<70% threshold, left unchanged)
console.log(correct("hello สวัสดี")); // hello สวัสดี (unchanged)
console.log(correct("สวัสดี hello")); // สวัสดี hello (unchanged)

Browser

<script src="https://cdn.jsdelivr.net/npm/thai-keyboard-corrector@latest/dist/thai-keyboard-corrector.js"></script>
<script>
  console.log(ThaiKeyboardCorrector.correct("l;ylfu")); // สวัสดี
</script>

API

| function | purpose | example | | ------------------- | ------------------------ | -------------------------- | | mapEngToThai(str) | Raw EN→TH character map | mapEngToThai("asdf") → "ฟหกด" | | mapThaiToEng(str) | Raw TH→EN character map | mapThaiToEng("ฟหกด") → "asdf" | | correct(str) | Smart auto-correction | correct("l;ylfu") → "สวัสดี" |

Note: mapThaiToEng() handles both base and shift layers, mapping Thai shifted characters to uppercase English.

How It Works

The correct() function uses a 70% threshold to determine layout:

• ≥70% Latin characters → Maps entire string to Thai
• ≥70% Thai characters → Maps entire string to English
• <70% of either script → Leaves unchanged (unclear ratio)

Keyboard Mapping

The library maps characters based on the Thai Kedmanee keyboard layout:

• Base layer: English QWERTY → Thai characters (e.g., q → ๆ, w → ไ)
• Shift layer: Thai shifted characters → Uppercase English (e.g., ฃ → W, ฅ → E)
• Numbers: Thai numerals → Arabic numerals (e.g., ๑ → 1, ๒ → 2)
• Symbols: Preserved as-is

Examples & Test Cases

Clear Cases (Always Swapped)

// 100% Latin → Thai
correct("l;ylfu")        // สวัสดี
correct("asdf")          // ฟหกด
correct("hello world")   // ้ำสสน ไนพสก
correct("qwerty")        // ๆไำพะั
correct("test123")       // ะำหะ123
correct("HELLO")         // ้ำสสน

// 100% Thai → English
correct("ฟหกด")          // asdf
correct("สวัสดี")        // l;ylfu
correct("ประเทศไทย")      // xitgmlwmp
correct("ครับ")          // คiy[
correct("๑๒๓๔๕")         // 12345
correct("ฃฅฆฑ")          // WERT (shifted characters)

Unclear Cases (Left Unchanged)

// Mixed content with unclear ratio
correct("hello สวัสดี")   // hello สวัสดี (50/50 split)
correct("สวัสดี hello")   // สวัสดี hello (50/50 split)
correct("hi สวัสดี")      // hi l;ylfu (25/75 split, Thai ≥70%)
correct("สวัสดี hi")      // l;ylfu hi (75/25 split, Thai ≥70%)

Edge Cases

// No script characters
correct("123456")        // 123456 (unchanged)
correct("!@#$%")         // !@#$% (unchanged)
correct("")              // "" (unchanged)
correct("   ")           // "   " (unchanged)

// Numbers and symbols are preserved
correct("test123")       // ะสสะ123
correct("hello!")        // ้ำสสน!

Use Cases

1. Chat Applications

// User types with wrong layout
const userInput = "l;ylfu";
const corrected = correct(userInput); // "สวัสดี"

2. Form Validation

// Check if input needs correction
const needsCorrection = correct(input) !== input;

3. Real-time Text Processing

// Process text as user types
const processedText = correct(rawInput);

4. Data Cleaning

// Clean mixed keyboard layouts in datasets
const cleaned = correct("hello สวัสดี"); // "hello สวัสดี" (preserved)

License

MIT