Function parse_hex_char

src/string.rs:210–274

(chars: &mut T)

}

fn parse_hex_char<'a, T: CharProvider<'a>>(chars: &mut T) -> Result<char, ParseStringErrorKind> {
  let mut buf1 = [0u8; 4];
  let hex_value = read_four_hex_digits(chars, &mut buf1)?;

  // check if this is a high surrogate (0xD800-0xDBFF)
  let hex_char = if (0xD800..=0xDBFF).contains(&hex_value) {
    // high surrogate - must be followed by low surrogate (\uXXXX)
    if chars.move_next_char() != Some('\\') {
      return Err(ParseStringErrorKind::InvalidUnicodeEscapeSequence(format!(
        "{} (unpaired high surrogate)",
        hex_buf_to_str(&buf1)
      )));
    }

    if chars.move_next_char() != Some('u') {
      return Err(ParseStringErrorKind::InvalidUnicodeEscapeSequence(format!(
        "{} (unpaired high surrogate)",
        hex_buf_to_str(&buf1)
      )));
    }

    // parse the second \uXXXX
    let mut buf2 = [0u8; 4];
    let hex_value2 = read_four_hex_digits(chars, &mut buf2)?;

    // verify it's a low surrogate (0xDC00-0xDFFF)
    if !(0xDC00..=0xDFFF).contains(&hex_value2) {
      return Err(ParseStringErrorKind::InvalidUnicodeEscapeSequence(format!(
        "{} (high surrogate not followed by low surrogate)",
        hex_buf_to_str(&buf1)
      )));
    }

    // combine surrogate pair using RFC 8259 formula
    let code_point = ((hex_value - 0xD800) * 0x400) + (hex_value2 - 0xDC00) + 0x10000;

    match std::char::from_u32(code_point) {
      Some(c) => c,
      None => {
        return Err(ParseStringErrorKind::InvalidUnicodeEscapeSequence(format!(
          "{}\\u{} (invalid surrogate pair)",
          hex_buf_to_str(&buf1),
          hex_buf_to_str(&buf2)
        )));
      }
    }
  } else if (0xDC00..=0xDFFF).contains(&hex_value) {
    // low surrogate without high surrogate
    return Err(ParseStringErrorKind::InvalidUnicodeEscapeSequence(format!(
      "{} (unpaired low surrogate)",
      hex_buf_to_str(&buf1)
    )));
  } else {
    // normal unicode escape
    match std::char::from_u32(hex_value) {
      Some(hex_char) => hex_char,
      None => {
        return Err(ParseStringErrorKind::InvalidUnicodeEscapeSequence(hex_buf_to_str(