bfin/src/tokenizer.rs

#[derive(Debug)]
pub enum TokenType {
    Plus(u32),
    Minus(u32),
    MoveRight(u32),
    MoveLeft(u32),
    Output,
    Input,
    BracketOpen,
    BracketClose,
}

#[derive(Debug)]
pub struct Token {
    pub r#type: TokenType,
}

impl Token {
    fn new(r#type: TokenType) -> Self {
        Self { r#type }
    }
}

#[derive(Debug, Default)]
pub struct Tokenizer {
    index: u32,
}

impl Tokenizer {
    pub fn tokenize(&self, program: Vec<u8>) -> Vec<Token> {
        let mut tokens = vec![];

        let mut ip = 0;
        let mut icount = 0;
        while ip < program.len() {
            let Some(instruction) = program.get(ip) else {
                break;
            };

            let next_instruction = program.get(&ip + 1);

            ip += 1;

            match instruction {
                b'+' => {
                    icount += 1;
                    if let Some(next) = next_instruction
                        && *next == b'+'
                    {
                        continue;
                    } else {
                        tokens.push(Token::new(TokenType::Plus(icount)));
                        icount = 0;
                    }
                }
                b'-' => {
                    icount += 1;
                    if let Some(next) = next_instruction
                        && *next == b'-'
                    {
                        continue;
                    } else {
                        tokens.push(Token::new(TokenType::Minus(icount)));
                        icount = 0;
                    }
                }
                b'>' => {}
                b'<' => {}
                b'.' => {
                    tokens.push(Token::new(TokenType::Output));
                }
                b',' => {}
                b'[' => {}
                b']' => {}
                _ => {}
            }
        }

        tokens
    }
}