Function parseUtf8

src/machine.cpp:809–845

}

object parseUtf8(Thread* t, AbstractStream& s, unsigned length)
{
  GcByteArray* value = makeByteArray(t, length + 1);
  unsigned vi = 0;
  for (unsigned si = 0; si < length; ++si) {
    unsigned a = s.read1();
    if (a & 0x80) {
      if (a & 0x20) {
        // 3 bytes
        return parseUtf8NonAscii(t, s, value, vi, si, a, NoByte);
      } else {
        // 2 bytes
        unsigned b = s.read1();

        if (a == 0xC0 and b == 0x80) {
          ++si;
          assertT(t, si < length);
          value->body()[vi++] = 0;
        } else {
          return parseUtf8NonAscii(t, s, value, vi, si, a, b);
        }
      }
    } else {
      value->body()[vi++] = a;
    }
  }

  if (vi < length) {
    PROTECT(t, value);

    GcByteArray* v = makeByteArray(t, vi + 1);
    memcpy(v->body().begin(), value->body().begin(), vi);
    value = v;
  }

  return value;
}

GcByteArray* makeByteArray(Thread* t, Stream& s, unsigned length)
{