Why do UTF-8 characters become corrupted when decoding network chunks separately?

Network chunk boundaries do not align with character boundaries. A multi-byte character can be split across two chunks. Use a streaming decoder such as StringDecoder or TextDecoder with streaming enabled, then flush it at end of input. Test by splitting a non-ASCII character at every byte position and comparing the final text. Do not call each Buffer’s toString independently and concatenate damaged results. The decoder must also have a policy for malformed byte sequences. Correct decoding does not establish message boundaries; a line or framed protocol still needs its own incremental parser. Minimal example:

import assert from 'node:assert/strict';
import { StringDecoder } from 'node:string_decoder';
const bytes = Buffer.from('€', 'utf8');
for (let split = 1; split < bytes.length; split++) {
  const decoder = new StringDecoder('utf8');
  const text = decoder.write(bytes.subarray(0, split))
    + decoder.write(bytes.subarray(split)) + decoder.end();
  assert.equal(text, '€');
}