tricu

io-utf8.ts (2097B)

---

 // Strict stateless UTF-8. No BOM stripping, replacement or normalization.
 // Source carries the copied <=3-byte suffix between requests.
 export class UTF8Error extends Error {}
 export function decodeUTF8(bytes: readonly number[], final: boolean): { text: string; carry: number[] } {
   if (bytes.length > 4096 || bytes.some(b => !Number.isInteger(b) || b < 0 || b > 255)) throw new UTF8Error('invalid byte chunk');
   const chars: string[] = [];
   const bad = (offset: number, why: string): never => { throw new UTF8Error(`${why} at byte ${offset}`); };
   for (let offset = 0; offset < bytes.length;) {
     const lead = bytes[offset]!;
     if (lead <= 0x7f) { chars.push(String.fromCodePoint(lead)); offset++; continue; }
     let count: number, low = 0x80, high = 0xbf, code: number;
     if (lead >= 0xc2 && lead <= 0xdf) { count = 2; code = lead & 0x1f; }
     else if (lead >= 0xe0 && lead <= 0xef) {
       count = 3; code = lead & 0x0f; if (lead === 0xe0) low = 0xa0; if (lead === 0xed) high = 0x9f;
     } else if (lead >= 0xf0 && lead <= 0xf4) {
       count = 4; code = lead & 7; if (lead === 0xf0) low = 0x90; if (lead === 0xf4) high = 0x8f;
     } else return bad(offset, 'invalid UTF-8 leading byte');
     const available = Math.min(count - 1, bytes.length - offset - 1);
     for (let i = 1; i <= available; i++) {
       const b = bytes[offset + i]!;
       if (b < (i === 1 ? low : 0x80) || b > (i === 1 ? high : 0xbf)) return bad(offset, 'invalid UTF-8 sequence');
       code = code * 64 + (b & 63);
     }
     if (available < count - 1) {
       if (final) return bad(offset, 'incomplete UTF-8 sequence');
       return { text: chars.join(''), carry: bytes.slice(offset) };
     }
     chars.push(String.fromCodePoint(code)); offset += count;
   }
   return { text: chars.join(''), carry: [] };
 }
 export function encodeUTF8(text: string): number[] {
   let count = 0;
   for (const char of text) {
     const scalar = char.codePointAt(0)!;
     if (++count > 1024 || scalar >= 0xd800 && scalar <= 0xdfff) throw new UTF8Error('invalid scalar chunk');
   }
   return [...new TextEncoder().encode(text)];
 }