Function mergePrefixUtf8

lib/internal/encoding/util.js:42–70

* Merge prefix `chunk` with `data` and return new combined prefix. * For data.length < 3, fully consumes data and can return unfinished data, * otherwise returns a prefix with no unfinished bytes * @param {Uint8Array} data Uint8Array of potentially UTF-8 bytes * @param {Uint8Array} chunk Prefix

(data, chunk)

 *   so that the result has no unfinished UTF-8 codepoints. If data.length < 3: concat(chunk, data).
 */
function mergePrefixUtf8(data, chunk) {
  if (data.length === 0) return chunk;
  if (data.length < 3) {
    // No reason to bruteforce offsets, also it's possible this doesn't yet end the sequence
    const res = new Uint8Array(data.length + chunk.length);
    res.set(chunk);
    res.set(data, chunk.length);
    return res;
  }

  // Slice off a small portion of data into prefix chunk so we can decode them separately without extending array size
  const temp = new Uint8Array(chunk.length + 3); // We have 1-3 bytes and need 1-3 more bytes
  temp.set(chunk);
  temp.set(data.subarray(0, 3), chunk.length);

  // Stop at the first offset where unfinished bytes reaches 0 or fits into data
  // If that doesn't happen (data too short), just concat chunk and data completely (above)
  for (let i = 1; i <= 3; i++) {
    const unfinished = unfinishedBytesUtf8(temp, chunk.length + i); // 0-3
    if (unfinished <= i) {
      // Always reachable at 3, but we still need 'unfinished' value for it
      const add = i - unfinished; // 0-3
      return add > 0 ? temp.subarray(0, chunk.length + add) : chunk;
    }
  }

  // Unreachable
  return null;
}

module.exports = { unfinishedBytesUtf8, mergePrefixUtf8 };