Category: Rust

バイトコードに変換するのはあくまでOpCode, Argであって、OpCodeと引数を元に処理するプログラム自体はバイトコードには変換しない。Instruction自体は別に持っておく。

use std::io::{BufReader, BufWriter, Read, Write};

#[derive(Debug, Clone, Copy)]
#[repr(u8)]
pub enum OpCode {
    LoadLiteral,
    Add,
}

impl From<u8> for OpCode {
    #[allow(non_upper_case_globals)]
    fn from(o: u8) -> Self {
      const LoadLiteral: u8 = OpCode::LoadLiteral as u8;
      const Add: u8 = OpCode::Add as u8;
  
      match o {
        LoadLiteral => OpCode::LoadLiteral,
        Add => OpCode::Add,
        _ => panic!("Opcode \"{:02X}\" unrecognized!", o),
      }
    }
  }

#[derive(Debug, Clone, Copy)]
#[repr(C)]
struct Instruction {
    op: OpCode,
    arg0: u8,
}

impl Instruction {
    fn new(op: OpCode, arg0: u8) -> Self {
        Self { op, arg0 }
    }

    fn serialize(
        &self,
        writer: &mut impl Write,
    ) -> Result<(), std::io::Error> {
        writer.write_all(&[self.op as u8, self.arg0])?;
        Ok(())
    }
}

fn write_program(file: &str) {
    let instructions = [
        Instruction::new(OpCode::LoadLiteral, 42),
        Instruction::new(OpCode::LoadLiteral, 36),
        Instruction::new(OpCode::Add, 0),
    ];

    if let Ok(writer) = std::fs::File::create(file) {
        let mut writer = BufWriter::new(writer);
        for instruction in &instructions {
            instruction.serialize(&mut writer).unwrap();
        }
        println!("Written {} instructions", instructions.len());
    }
}

fn main() {
    write_program("bytecode.bin");
}

fn interpret(instructions: &[Instruction]) -> Option<i64> {
    let mut stack = vec![];

    for instruction in instructions {
        match instruction.op {
          OpCode::LoadLiteral => {
            stack.push(instruction.arg0 as i64);
          }
          OpCode::Add => {
            let rhs = stack.pop().expect("Stack underflow");
            let lhs = stack.pop().expect("Stack underflow");
            stack.push(lhs + rhs);
          }
        }
      }
    stack.pop()
}

なるほど、データの流れは理解できました。

### デシリアライズ

use std::io::{BufReader, BufWriter, Read, Write};

#[derive(Debug, Clone, Copy)]
#[repr(u8)]
pub enum OpCode {
    LoadLiteral,
    Add,
}

impl From<u8> for OpCode {
    #[allow(non_upper_case_globals)]
    fn from(o: u8) -> Self {
      const LoadLiteral: u8 = OpCode::LoadLiteral as u8;
      const Add: u8 = OpCode::Add as u8;
  
      match o {
        LoadLiteral => OpCode::LoadLiteral,
        Add => OpCode::Add,
        _ => panic!("Opcode \"{:02X}\" unrecognized!", o),
      }
    }
  }

#[derive(Debug, Clone, Copy)]
#[repr(C)]
struct Instruction {
    op: OpCode,
    arg0: u8,
}

impl Instruction {
    fn new(op: OpCode, arg0: u8) -> Self {
        Self { op, arg0 }
    }

    fn serialize(
        &self,
        writer: &mut impl Write,
    ) -> Result<(), std::io::Error> {
        writer.write_all(&[self.op as u8, self.arg0])?;
        Ok(())
    }

    fn deserialize(
        reader: &mut impl Read,
    ) -> Result<Self, std::io::Error> {
        let mut buf = [0u8; 2];
        reader.read_exact(&mut buf)?;
        Ok(Self::new(buf[0].into(), buf[1]))
    }
}

fn write_program(file: &str) {
    let instructions = [
        Instruction::new(OpCode::LoadLiteral, 42),
        Instruction::new(OpCode::LoadLiteral, 36),
        Instruction::new(OpCode::Add, 0),
    ];

    if let Ok(writer) = std::fs::File::create(file) {
        let mut writer = BufWriter::new(writer);
        for instruction in &instructions {
            instruction.serialize(&mut writer).unwrap();
        }
        println!("Written {} instructions", instructions.len());
    }
}

fn read_program(file: &str) -> Option<Vec<Instruction>> {
    if let Ok(reader) = std::fs::File::open(file) {
        let mut reader = BufReader::new(reader);
        let mut instructions = vec![];
        while let Ok(inst) = Instruction::deserialize(&mut reader) {
            instructions.push(inst);
        }
        Some(instructions)
    } else {
        None
    }
}

fn main() {
    let mut args = std::env::args();
    args.next();
    match args.next().as_ref().map(|s| s as &str) {
        Some("w") => write_program("bytecode.bin"),
        Some("r") => {
            if let Some(instructions) = read_program("bytecode.bin") {
                let result = interpret(&instructions);
                println!("result: {result:?}");
            }
        }
        _ => println!("Please specify w or r as an argument"),
    }
    
}

fn interpret(instructions: &[Instruction]) -> Option<i64> {
    let mut stack = vec![];

    for instruction in instructions {
        match instruction.op {
          OpCode::LoadLiteral => {
            stack.push(instruction.arg0 as i64);
          }
          OpCode::Add => {
            let rhs = stack.pop().expect("Stack underflow");
            let lhs = stack.pop().expect("Stack underflow");
            stack.push(lhs + rhs);
          }
        }
      }
    stack.pop()
}

Finished `dev` profile [unoptimized + debuginfo] target(s) in 0.61s
Running `target/debug/ruscal r`
result: Some(78)

バイトコードを読み取って、スタックを実行している。

fn main() {
    let input = "123 456 world";
    println!("source: {}, parsed: {:?}", input, source(input));

    let input = "(car cdr) cdr";
    println!("source: {}, parsed: {:?}", input, source(input));

    let input = "()())))((()))";
    println!("source: {}, parsed: {:?}", input, source(input));
}

fn advance_char(input: &str) -> &str {
    let mut chars = input.chars();
    chars.next();
    chars.as_str()
}

fn peek_char(input: &str) -> Option<char> {
    input.chars().next()
}

fn source(mut input: &str) -> Vec<Token> {
    let mut tokens = vec![];
    while !input.is_empty() {
        input = if let (next_input, Some(token)) = token(input) {
            tokens.push(token);
            next_input
        } else {
            break;
        }
    }
    tokens
}

#[derive(Debug, PartialEq, Eq)]
enum Token {
    Ident,
    Number,
    LParen,
    RParen,
}

fn token(i: &str) -> (&str, Option<Token>) {
    if let (i, Some(ident_res)) = ident(whitespace(i)) {
        return (i, Some(ident_res));
    }

    if let (i, Some(number_res)) = number(whitespace(i)) {
        return (i, Some(number_res));
    }

    if let (i, Some(lparen_res)) = lparen(whitespace(i)) {
        return (i, Some(lparen_res));
    }

    if let (i, Some(rparen_res)) = rparen(whitespace(i)) {
        return (i, Some(rparen_res));
    }

    (whitespace(i), None)
}

fn whitespace(mut input: &str) -> &str {
    while matches!(input.chars().next(), Some(' ')) {
        input = advance_char(input);
    }
    input
}

fn ident(mut input: &str) -> (&str, Option<Token>) {
    if matches!(
        peek_char(input),
        Some(_x @ ('a'..='z' | 'A'..='Z'))
    ) {
        input = advance_char(input);
        while matches!(
            peek_char(input),
            Some(_x @ ('a'..='z' | 'A'..='Z' | '0'..='9'))
        ) {
            input = advance_char(input);
        }
        (input, Some(Token::Ident))
    } else {
        (input, None)
    }
}

fn number(mut input: &str) -> (&str, Option<Token>) {
    if matches!(
        peek_char(input),
        Some(_x @ ('-'|'+'|'.'|'0'..='9'))
    ) {
        input = advance_char(input);
        while matches!(
            peek_char(input),
            Some(_x @ ('.'|'0'..='9'))
        ) {
            input = advance_char(input);
        }
        (input, Some(Token::Number))
    } else {
        (input, None)
    }
}

fn lparen(mut input: &str) -> (&str, Option<Token>) {
    if matches!(peek_char(input), Some('(')) {
        input = advance_char(input);
        (input, Some(Token::LParen))
    } else {
        (input, None)
    }
}

fn rparen(mut input: &str) -> (&str, Option<Token>) {
    if matches!(peek_char(input), Some(')')) {
        input = advance_char(input);
        (input, Some(Token::RParen))
    } else {
        (input, None)
    }
}

#[cfg(test)]
mod test {
    use super::*;

    #[test]
    fn test_whitespace() {
        assert_eq!(whitespace("    "), "");
    }

    #[test]
    fn test_indent() {
        assert_eq!(ident("Adam"), "");
    }

    #[test]
    fn test_number() {
        assert_eq!(number("123.45 "), " ");
    }
}

Finished `dev` profile [unoptimized + debuginfo] target(s) in 0.58s
Running `target/debug/ruscal`
source: 123 456 world, parsed: [Number, Number, Ident]
source: (car cdr) cdr, parsed: [LParen, Ident, Ident, RParen, Ident]
source: ()())))((())), parsed: [LParen, RParen, LParen, RParen, RParen, RParen, RParen, LParen, LParen, LParen, RParen, RParen, RParen]

fn main() {
    let input = "123 world";
    println!("source: {}, parsed: {:?}", input, source(input));

    let input = "Hello world";
    println!("source: {}, parsed: {:?}", input, source(input));

    let input = "      world";
    println!("source: {}, parsed: {:?}", input, source(input));
}

fn source(mut input: &str) -> Vec<Token> {
    let mut tokens = vec![];
    while !input.is_empty() {
        input = if let (next_input, Some(token)) = token(input) {
            tokens.push(token);
            next_input
        } else {
            break;
        }
    }
    tokens
}

#[derive(Debug, PartialEq, Eq)]
enum Token {
    Ident,
    Number,
}

fn token(i: &str) -> (&str, Option<Token>) {
    if let (i, Some(ident_res)) = ident(whitespace(i)) {
        return (i, Some(ident_res));
    }

    if let (i, Some(number_res)) = number(whitespace(i)) {
        return (i, Some(number_res));
    }
    (i, None)
}

fn whitespace(mut input: &str) -> &str {
    while matches!(input.chars().next(), Some(' ')) {
        let mut chars = input.chars();
        chars.next();
        input = chars.as_str();
    }
    input
}

fn ident(mut input: &str) -> (&str, Option<Token>) {
    if matches!(
        input.chars().next(),
        Some(_x @ ('a'..='z' | 'A'..='Z'))
    ) {
        while matches!(
            input.chars().next(),
            Some(_x @ ('a'..='z' | 'A'..='Z' | '0'..='9'))
        ) {
            let mut chars = input.chars();
            chars.next();
            input = chars.as_str();
        }
        (input, Some(Token::Ident))
    } else {
        (input, None)
    }
}

fn number(mut input: &str) -> (&str, Option<Token>) {
    if matches!(
        input.chars().next(),
        Some(_x @ ('-'|'+'|'.'|'0'..='9'))
    ) {
        while matches!(
            input.chars().next(),
            Some(_x @ ('.'|'0'..='9'))
        ) {
            let mut chars = input.chars();
            chars.next();
            input = chars.as_str();
        }
        (input, Some(Token::Number))
    } else {
        (input, None)
    }
}

#[cfg(test)]
mod test {
    use super::*;

    #[test]
    fn test_whitespace() {
        assert_eq!(whitespace("    "), "");
    }

    #[test]
    fn test_indent() {
        assert_eq!(ident("Adam"), "");
    }

    #[test]
    fn test_number() {
        assert_eq!(number("123.45 "), " ");
    }
}

Finished `dev` profile [unoptimized + debuginfo] target(s) in 2.04s
Running `target/debug/ruscal`
source: 123 world, parsed: [Number, Ident]
source: Hello world, parsed: [Ident, Ident]
source: world, parsed: [Ident]

一文字ずつ処理したい場合
matches!の場合は、Someと比較する

fn main() {
    let mut str = "hello world!";
    while matches!(str.chars().next(), Some(_x @ ('a'..='z' | 'A'..='Z'))) {
        let mut chars = str.chars();
        chars.next();
        str = chars.as_str();
        println!("{:?}", str);
    }
}

Running `target/debug/sample`
“ello world!”
“llo world!”
“lo world!”
“o world!”
” world!”

これを応用して、文字列をパースする

fn main() {
    let source = "123 world";
    println!(
        "source: {}, parsed: {:?}",
        source,
        ident(whitespace(number(source)))
    );
}

fn whitespace(mut input: &str) -> &str {
    while matches!(input.chars().next(), Some(' ')) {
        let mut chars = input.chars();
        chars.next();
        input = chars.as_str();
    }
    input
}

fn ident(mut input: &str) -> &str {
    if matches!(
        input.chars().next(),
        Some(_x @ ('a'..='z' | 'A'..='Z'))
    ) {
        while matches!(
            input.chars().next(),
            Some(_x @ ('a'..='z' | 'A'..='Z' | '0'..='9'))
        ) {
            let mut chars = input.chars();
            chars.next();
            input = chars.as_str();
        }
    }
    input
}

fn number(mut input: &str) -> &str {
    if matches!(
        input.chars().next(),
        Some(_x @ ('-'|'+'|'.'|'0'..='9'))
    ) {
        while matches!(
            input.chars().next(),
            Some(_x @ ('.'|'0'..='9'))
        ) {
            let mut chars = input.chars();
            chars.next();
            input = chars.as_str();
        }
    }
    input
}

#[cfg(test)]
mod test {
    use super::*;

    #[test]
    fn test_whitespace() {
        assert_eq!(whitespace("    "), "");
    }

    #[test]
    fn test_indent() {
        assert_eq!(ident("Adam"), "");
    }

    #[test]
    fn test_number() {
        assert_eq!(number("123.45 "), " ");
    }
}

関数を値として持っておくのだが変数と同じように、/で始まり、かつ{ } の場合、{ }の中身をoperationとして評価している

use std::{
    collections::HashMap,
    io::{BufRead, BufReader},
};

#[derive(Debug, Clone, PartialEq, Eq)]
enum Value {
    Num(i32),
    Op(String),
    Sym(String),
    Block(Vec<Value>),
    Native(NativeOp),
}

impl Value {
    fn as_num(&self) -> i32 {
        match self {
        Self::Num(val) => *val,
        _ => panic!("Value is not a number"),
        }
    }

    fn to_block(self) -> Vec<Value> {
        match self {
            Self::Block(val) => val,
            _ => panic!("Value is not a block"),
        }
    }

    fn as_sym(&self) -> &str {
        if let Self::Sym(sym) = self {
            sym
        } else {
            panic!("Value is not a symbol");
        }
    }

    fn to_string(&self) -> String {
        match self {
            Self::Num(i) => i.to_string(),
            Self::Op(ref s) | Self::Sym(ref s) => s.clone(),
            Self::Block(_) => "<Block>".to_string(),
            Self::Native(_) => "<Native>".to_string(),
        }
    }
}

#[derive(Clone)]
struct NativeOp(fn(&mut Vm));

impl PartialEq for NativeOp {
    fn eq(&self, other: &NativeOp) -> bool {
        self.0 as *const fn() == other.0 as *const fn()
    }
}

impl Eq for NativeOp {}

impl std::fmt::Debug for NativeOp {
    fn fmt(
        &self,
        f: &mut std::fmt::Formatter<'_>,
    ) -> std::fmt::Result {
        write!(f, "<NativeOp>")
    }
}

struct Vm {
    stack: Vec<Value>,
    vars: HashMap<String, Value>,
    blocks: Vec<Vec<Value>>,
}

impl Vm {
  fn new() -> Self {
    let functions: [(&str, fn(&mut Vm)); 12] = [
      ("+", add),
      ("-", sub),
      ("*", mul),
      ("/", div),
      ("<", lt),
      ("if", op_if),
      ("def", op_def),
      ("puts", puts),
      ("pop", pop),
      ("dup", dup),
      ("exch", exch),
      ("index", index),
    ];
    Self {
      stack: vec![],
      vars: functions
        .into_iter()
        .map(|(name, fun)| {
          (name.to_owned(), Value::Native(NativeOp(fun)))
        })
        .collect(),
      blocks: vec![],
    }
  }
}

fn main() {
  if let Some(f) = std::env::args()
    .nth(1)
    .and_then(|f| std::fs::File::open(f).ok())
  {
    parse_batch(BufReader::new(f));
  } else {
    parse_interactive();
  }
}

fn parse_batch(source: impl BufRead) -> Vec<Value> {
  let mut vm = Vm::new();
  for line in source.lines().flatten() {
    for word in line.split(" ") {
      parse_word(word, &mut vm);
    }
  }
  vm.stack
}

fn parse_interactive() {
  let mut vm = Vm::new();
  for line in std::io::stdin().lines().flatten() {
    for word in line.split(" ") {
      parse_word(word, &mut vm);
    }
    println!("stack: {:?}", vm.stack);
  }
}

fn parse_word(word: &str, vm: &mut Vm) {
  if word.is_empty() {
    return;
  }
  if word == "{" {
    vm.blocks.push(vec![]);
  } else if word == "}" {
    let top_block =
      vm.blocks.pop().expect("Block stack underflow!");
    eval(Value::Block(top_block), vm);
  } else {
    let code = if let Ok(num) = word.parse::<i32>() {
      Value::Num(num)
    } else if word.starts_with("/") {
      Value::Sym(word[1..].to_string())
    } else {
      Value::Op(word.to_string())
    };
    eval(code, vm);
  }
}


fn eval(code: Value, vm: &mut Vm) {
  if let Some(top_block) = vm.blocks.last_mut() {
    top_block.push(code);
    return;
  }
  if let Value::Op(ref op) = code {
    let val = vm
      .vars
      .get(op)
      .expect(&format!("{op:?} is not a defined operation"))
      .clone();
    match val {
      Value::Block(block) => {
        for code in block {
          eval(code, vm);
        }
      }
      Value::Native(op) => op.0(vm),
      _ => vm.stack.push(val),
    }
  } else {
    vm.stack.push(code.clone());
  }
}

macro_rules! impl_op {
    {$name:ident, $op:tt} => {
        fn $name(vm: &mut Vm) {
            let rhs = vm.stack.pop().unwrap().as_num();
            let lhs = vm.stack.pop().unwrap().as_num();
            vm.stack.push(Value::Num((lhs $op rhs) as i32));
        }
    }
}

impl_op!(add, +);
impl_op!(sub, -);
impl_op!(mul, *);
impl_op!(div, /);
impl_op!(lt, <);

fn op_if(vm: &mut Vm) {
    let false_branch = vm.stack.pop().unwrap().to_block();
    let true_branch = vm.stack.pop().unwrap().to_block();
    let cond = vm.stack.pop().unwrap().to_block();

    for code in cond {
        eval(code, vm);
    }

    let cond_result = vm.stack.pop().unwrap().as_num();

    if cond_result != 0 {
        for code in true_branch {
            eval(code, vm);
        }
    } else {
        for code in false_branch {
            eval(code, vm);
        }
    }
}

fn op_def(vm: &mut Vm) {
    let value = vm.stack.pop().unwrap();
    eval(value, vm);
    let value = vm.stack.pop().unwrap();
    let sym = vm.stack.pop().unwrap().as_sym().to_string();

    vm.vars.insert(sym, value);
}

fn puts(vm: &mut Vm){
    let value = vm.stack.pop().unwrap();
    println!("{}", value.to_string());
}

fn pop(vm: &mut Vm) {
    vm.stack.pop().unwrap();
}

fn dup(vm: &mut Vm) {
    let value = vm.stack.last().unwrap();
    vm.stack.push(value.clone());
}

fn exch(vm: &mut Vm) {
    let last = vm.stack.pop().unwrap();
    let second = vm.stack.pop().unwrap();
    vm.stack.push(last);
    vm.stack.push(second);
}

fn index(vm: &mut Vm) {
    let index = vm.stack.pop().unwrap().as_num() as usize;
    let value = vm.stack[vm.stack.len() - index - 1].clone();
    vm.stack.push(value);
}

#[cfg(test)]
mod test {
    use super::{Value::*, *};
    use std::io::Cursor;

    fn parse(input: &str) -> Vec<Value> {
        parse_batch(Cursor::new(input))
    }

    #[test]
    fn test_group(){
        assert_eq!(
            parse("1 2 + { 3 4 }"),
            vec![Num(3), Block(vec![Num(3), Num(4)])]
        );
    }

    #[test]
    fn test_if_false(){
        assert_eq!(
            parse("{ 1 -1 + } { 100 } { -100 } if"),
            vec![Num(-100)]
        );
    }

    #[test]
    fn test_if_true() {
        assert_eq!(
            parse("{ 1 1 + } { 100 } { -100 } if"),
            vec![Num(100)]
        );
    }

    #[test]
    fn test_var() {
        assert_eq!(
            parse("/x 10 def /y 20 def x y *"),
            vec![Num(200)]
        )
    }

    #[test]
    fn test_var_if() {
        assert_eq!(
            parse("/x 10 def /y 20 def { x y < } { x } { y } if"),
            vec![Num(10)]
        );
    }

    #[test]
    fn test_multiline() {
        assert_eq!(
            parse(
                r#"
                
            /x 10 def
            /y 20 def

            { x y < }
            { x }
            { y }
            if
            "#
            ),
            vec![Num(10)]
        );
    }

    #[test]
    fn test_function() {
        assert_eq!(
            parse(
                r#"
            /double { 2 * } def
            10 double"#
            ),
            vec![Num(20)]
        );
    }
}

Finished `dev` profile [unoptimized + debuginfo] target(s) in 0.48s
Running `target/debug/stackmachine`
/double { 2 * } def
stack: []
10 double puts
20
stack: []

ルールとしてわかりやすいように定義していくのね
考え方としては非常に参考になるものがあります