favela-android / test / utf8.test.mjs

// The bytes↔text seam of the RN IPC bridge, tested off-device.
//
// This is where a fully-booted node looked dead: bare-kit hands up a Uint8Array, `.toString()` on it
// returns "123,34,116,..." (Array.prototype.toString), and the message parser silently saw garbage
// forever. These tests pin both directions and the chunk-boundary case that a naive fix would miss.
import test from 'brittle'
import { encodeUtf8, createUtf8Decoder } from '../lib/utf8.js'

test('round-trips the node’s own log line, em dash included', (t) => {
  const line = JSON.stringify({ t: 'log', msg: 'no DID paired — read-only until pairing' }) + '\n'
  const decode = createUtf8Decoder()
  t.is(decode(encodeUtf8(line)), line)
})

test('decodes what bare-kit actually pushes: a plain Uint8Array', (t) => {
  const decode = createUtf8Decoder()
  const bytes = new Uint8Array(encodeUtf8('{"t":"ready"}\n'))
  t.is(decode(bytes), '{"t":"ready"}\n')
  t.not(decode(bytes), String(bytes), 'must not be the comma-joined byte values')
})

test('a multi-byte character split across two chunks is not corrupted', (t) => {
  const line = 'appview ready — 100%\n'
  const full = encodeUtf8(line)
  const decode = createUtf8Decoder()

  // Cut mid-em-dash: the 3-byte sequence starts at `cut - 1`.
  const cut = full.indexOf(0xe2) + 1
  t.ok(cut > 0, 'the fixture really does contain a multi-byte character')

  const first = decode(full.subarray(0, cut))
  const second = decode(full.subarray(cut))
  t.is(first + second, line)
})

test('holds back an incomplete trailing sequence rather than emitting a replacement char', (t) => {
  const decode = createUtf8Decoder()
  const full = encodeUtf8('— done')
  t.is(decode(full.subarray(0, 1)), '', 'a lone lead byte decodes to nothing yet')
  t.is(decode(full.subarray(1)), '— done')
})

test('encodes 4-byte code points (surrogate pairs)', (t) => {
  const decode = createUtf8Decoder()
  t.is(decode(encodeUtf8('🌱')), '🌱')
})