Function decodeTextWithErrors

packages/kaos/src/internal.ts:135–179

(
  data: Buffer,
  encoding: BufferEncoding,
  errors: 'strict' | 'replace' | 'ignore' = 'strict',
  ignoreBOM: boolean = false,
)

 * @internal
 */
export function decodeTextWithErrors(
  data: Buffer,
  encoding: BufferEncoding,
  errors: 'strict' | 'replace' | 'ignore' = 'strict',
  ignoreBOM: boolean = false,
): string {
  // Map Node's BufferEncoding names to Web TextDecoder labels where the two
  // diverge. Only UTF-family encodings participate in the strict/replace/
  // ignore dance; the others are lossless and use Buffer.toString directly.
  let webLabel: string | undefined;
  // eslint-disable-next-line typescript-eslint/switch-exhaustiveness-check
  switch (encoding) {
    case 'utf-8':
    case 'utf8':
      webLabel = 'utf-8';
      break;
    case 'utf16le':
    case 'ucs2':
    case 'ucs-2':
      webLabel = 'utf-16le';
      break;
    default:
      webLabel = undefined;
  }

  if (webLabel === undefined) {
    // Non-UTF encodings (hex/base64/latin1/binary/ascii) are lossless byte↔
    // character mappings; `errors` is meaningless for them. Return raw.
    return data.toString(encoding);
  }

  if (errors === 'strict') {
    return new TextDecoder(webLabel, { fatal: true, ignoreBOM }).decode(data);
  }

  // 'ignore' must skip invalid input bytes/code units, not delete every
  // replacement character in the decoded output. A file can contain a valid
  // U+FFFD, and Python preserves it under errors="ignore".
  if (errors === 'ignore') {
    return webLabel === 'utf-8' ? decodeUtf8Ignore(data) : decodeUtf16LeIgnore(data);
  }

  // 'replace' → substitute each invalid sequence with U+FFFD (default).
  return new TextDecoder(webLabel, { fatal: false, ignoreBOM }).decode(data);
}

/**
 * Convert a glob pattern segment (e.g. "*.txt", "file?.log") into a RegExp.