Function UTF8Encode

cpp/src/arrow/util/utf8_internal.h:324–343

}

static inline uint8_t* UTF8Encode(uint8_t* str, uint32_t codepoint) {
  if (codepoint < 0x80) {
    *str++ = codepoint;
  } else if (codepoint < 0x800) {
    *str++ = 0xC0 + (codepoint >> 6);
    *str++ = 0x80 + (codepoint & 0x3F);
  } else if (codepoint < 0x10000) {
    *str++ = 0xE0 + (codepoint >> 12);
    *str++ = 0x80 + ((codepoint >> 6) & 0x3F);
    *str++ = 0x80 + (codepoint & 0x3F);
  } else {
    // Assume proper codepoints are always passed
    assert(codepoint < kMaxUnicodeCodepoint);
    *str++ = 0xF0 + (codepoint >> 18);
    *str++ = 0x80 + ((codepoint >> 12) & 0x3F);
    *str++ = 0x80 + ((codepoint >> 6) & 0x3F);
    *str++ = 0x80 + (codepoint & 0x3F);
  }
  return str;
}

static inline bool UTF8Decode(const uint8_t** data, uint32_t* codepoint) {
  const uint8_t* str = *data;