Finished up to and including chapter 16

vm/Cargo.toml

@@ -0,0 +1,17 @@
+[package]
+name = "rlox2-vm"
+version = "0.1.0"
+edition = "2021"
+
+
+[dependencies.rlox2-frontend]
+path = "../frontend"
+
+
+[dependencies]
+itertools = "0.10.5"
+lazy_static = "1.4.0"
+num-derive = "0.3.3"
+num-traits = "0.2.15"
+regex = "1.7.1"
+thiserror = "1.0.38"

vm/src/chunk.rs

@@ -0,0 +1,68 @@
+use crate::debug::ChunkDebugInfo;
+use crate::{Opcode, Value};
+
+#[derive(Debug, Default)]
+pub struct Chunk {
+    code: Vec<u8>,
+
+    constants: Vec<Value>,
+
+    debug_info: Option<ChunkDebugInfo>,
+}
+
+impl Chunk {
+    pub fn new() -> Self {
+        Chunk {
+            code: Vec::new(),
+            constants: Vec::new(),
+            debug_info: Some(ChunkDebugInfo::new()),
+        }
+    }
+
+    pub fn code(&self) -> &[u8] {
+        &self.code
+    }
+
+    pub fn debug_info(&self) -> Option<&ChunkDebugInfo> {
+        self.debug_info.as_ref()
+    }
+
+    pub fn debug_info_mut(&mut self) -> Option<&mut ChunkDebugInfo> {
+        self.debug_info.as_mut()
+    }
+
+    pub fn write_byte<T>(&mut self, byte: T, line: u32)
+    where
+        T: TryInto<u8> + std::fmt::Debug,
+        <T as std::convert::TryInto<u8>>::Error: std::fmt::Debug,
+    {
+        let byte = byte.try_into().unwrap();
+
+        self.code.push(byte);
+
+        if let Some(ref mut debug_info) = self.debug_info {
+            debug_info.write_line(line, self.code.len() - 1);
+        }
+    }
+
+    pub fn write_bytes(&mut self, bytes: &[u8], line: u32) {
+        for byte in bytes {
+            self.write_byte(*byte, line);
+        }
+    }
+
+    pub fn write_opcode(&mut self, opcode: Opcode, line: u32) {
+        let byte = opcode as u8;
+        self.write_byte(byte, line);
+    }
+
+    pub fn add_constant(&mut self, value: Value) -> usize {
+        self.constants.push(value);
+
+        self.constants.len() - 1
+    }
+
+    pub fn get_constant(&self, offset: usize) -> &Value {
+        &self.constants[offset]
+    }
+}

vm/src/compile.rs

@@ -0,0 +1,6 @@
+use itertools::Itertools;
+use rlox2_frontend::lexer::Token;
+
+pub fn compile(tokens: Vec<Token>) {
+    println!("{}", tokens.iter().map(|token| token.to_string()).join(" "));
+}

vm/src/debug.rs

@@ -0,0 +1,76 @@
+#[derive(Debug)]
+struct LineInfo {
+    start_offset: usize,
+    line: u32,
+}
+
+#[derive(Debug, Default)]
+pub struct ChunkDebugInfo {
+    line_infos: Vec<LineInfo>,
+}
+
+impl ChunkDebugInfo {
+    pub fn new() -> Self {
+        ChunkDebugInfo { line_infos: Vec::new() }
+    }
+
+    pub fn write_line(&mut self, line: u32, offset: usize) {
+        let line_info = || LineInfo {
+            start_offset: offset,
+            line,
+        };
+
+        if self.line_infos.is_empty() {
+            self.line_infos.push(line_info())
+        }
+
+        match line.cmp(&self.line_infos.last().unwrap().line) {
+            // assert increasing line numbers
+            std::cmp::Ordering::Less => panic!(
+                "Tried to write_byte for line {line}, but line {} is already written",
+                self.line_infos.last().unwrap().line
+            ),
+            // line same as for last byte -> do nothing
+            std::cmp::Ordering::Equal => {}
+            // line greater than for last byte -> insert new line marker at current position
+            std::cmp::Ordering::Greater => self.line_infos.push(line_info()),
+        }
+    }
+
+    pub fn line_number(&self, offset: usize) -> u32 {
+        /* if self.line_infos.len() > 2 && offset < self.line_infos[1].start_offset {
+            assert!(offset >= self.line_infos[0].start_offset);
+
+            return self.line_infos[0].line;
+        }
+
+        let mut low = 0; // will be max { line_offset : line_offset < offset }
+        let mut high = self.line_infos.len() - 1; // will be min { line_offset : line_offset >= offset }
+
+        // bisect line_numbers
+        while high > low + 1 {
+            // as high >= low + 2, never high or low
+            let mid = (low + high) / 2;
+
+            match offset.cmp(&self.line_infos[mid].start_offset) {
+                std::cmp::Ordering::Less => high = mid,
+                std::cmp::Ordering::Equal => return self.line_infos[mid].line,
+                std::cmp::Ordering::Greater => low = mid,
+            }
+        }
+
+        self.line_infos[high].line */
+
+        // special case: all start_offsets are <= than offset:
+        // need to manually return the last line number as slice::partition_point doesn't like that
+        if self.line_infos.last().unwrap().start_offset <= offset {
+            return self.line_infos.last().unwrap().line;
+        }
+
+        let idx = self
+            .line_infos
+            .partition_point(|line_info| line_info.start_offset < offset);
+
+        self.line_infos[idx].line
+    }
+}

vm/src/disassemble.rs

@@ -0,0 +1,61 @@
+use num_traits::FromPrimitive;
+
+use crate::Chunk;
+use crate::Opcode;
+
+impl Chunk {
+    pub fn disassemble(&self, name: &str) {
+        println!("==== begin {name} ====\n");
+
+        let mut offset = 0;
+
+        while offset < self.code().len() {
+            offset = self.disassemble_instruction(offset)
+        }
+
+        println!("\n====  end {name}  ====");
+    }
+
+    pub fn disassemble_instruction(&self, offset: usize) -> usize {
+        use Opcode::*;
+
+        let mut offset = offset;
+
+        print!("{offset:04} ");
+
+        if let Some(debug_info) = self.debug_info() {
+            if offset == 0 || debug_info.line_number(offset) != debug_info.line_number(offset - 1) {
+                let line = debug_info.line_number(offset);
+                print!("{line:4} ");
+            } else {
+                print!("   | ");
+            };
+        }
+
+        let opcode: Opcode = FromPrimitive::from_u8(self.code()[offset]).unwrap();
+        offset += 1;
+
+        print!("{opcode:<16} ");
+
+        match opcode {
+            LoadConst => {
+                let constant_idx = self.code()[offset];
+                let value = self.get_constant(constant_idx as usize);
+                print!("{constant_idx:4} '{value}'");
+                offset += 1;
+            }
+            LoadConstLong => {
+                let bytes = &self.code()[offset..offset + 3];
+                let constant_idx = u32::from_le_bytes([bytes[0], bytes[1], bytes[2], 0]);
+                let value = self.get_constant(constant_idx as usize);
+                print!("{constant_idx:4} '{value}'");
+                offset += 3;
+            }
+            Add | Subtract | Multiply | Divide | Negate | Return => {}
+        }
+
+        println!();
+
+        offset
+    }
+}

vm/src/error.rs

@@ -0,0 +1,58 @@
+use itertools::Itertools;
+use rlox2_frontend::lexer::LexerError;
+use thiserror::Error;
+
+use crate::{Opcode, Value};
+
+#[derive(Error, Debug)]
+pub enum CompileError {}
+
+#[derive(Error, Debug)]
+pub enum RuntimeError {
+    #[error("Opcopde {opcode} had invalid operand {operand}")]
+    UnaryInvalidOperand { opcode: Opcode, operand: Value },
+    #[error("Opcopde {opcode} had invalid operands {left} and {right}")]
+    BinaryInvalidOperand { opcode: Opcode, left: Value, right: Value },
+    #[error("Division by zero")]
+    DivisionByZero,
+}
+
+#[derive(Error, Debug)]
+pub enum InterpretError {
+    #[error("{0}", format_multiple_errors(inner))]
+    LexerError { inner: Vec<LexerError> },
+    #[error("{inner}")]
+    CompileError { inner: CompileError },
+    #[error("{inner}")]
+    RuntimeError { inner: RuntimeError },
+    #[error("Called exit() with exit code {exit_code}")]
+    Exit { exit_code: i32 },
+}
+
+impl From<Vec<LexerError>> for InterpretError {
+    fn from(lexer_errs: Vec<LexerError>) -> Self {
+        InterpretError::LexerError { inner: lexer_errs }
+    }
+}
+
+impl From<CompileError> for InterpretError {
+    fn from(compile_err: CompileError) -> Self {
+        InterpretError::CompileError { inner: compile_err }
+    }
+}
+
+impl From<RuntimeError> for InterpretError {
+    fn from(runtime_err: RuntimeError) -> Self {
+        InterpretError::RuntimeError { inner: runtime_err }
+    }
+}
+
+fn format_multiple_errors(errs: &Vec<impl std::error::Error>) -> String {
+    let msg = if errs.len() == 1 {
+        errs[0].to_string()
+    } else {
+        errs.iter().map(|err| err.to_string()).join("\n")
+    };
+
+    msg
+}

vm/src/lib.rs

@@ -0,0 +1,17 @@
+mod chunk;
+mod compile;
+mod debug;
+mod disassemble;
+mod error;
+mod opcode;
+mod run;
+mod value;
+mod vm;
+
+pub use chunk::Chunk;
+pub use compile::compile;
+pub use error::InterpretError;
+pub use opcode::Opcode;
+pub use run::{run, run_repl};
+pub use value::Value;
+pub use vm::VM;

vm/src/opcode.rs

@@ -0,0 +1,35 @@
+use std::fmt::Display;
+
+use itertools::Itertools;
+use lazy_static::lazy_static;
+use num_derive::FromPrimitive;
+use regex::Regex;
+
+lazy_static! {
+    static ref CAMEL_CASE_REGEX: Regex = Regex::new(r"[A-Z][a-z]*").unwrap();
+}
+
+#[repr(u8)]
+#[derive(Debug, FromPrimitive, Clone, Copy)]
+#[allow(non_camel_case_types)] // hack so Constant_Long will be printed as CONSTANT_LONG rather than CONSTANTLONG
+pub enum Opcode {
+    LoadConst,
+    LoadConstLong,
+    Add,
+    Subtract,
+    Multiply,
+    Divide,
+    Negate,
+    Return,
+}
+
+impl Display for Opcode {
+    fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
+        let name = CAMEL_CASE_REGEX
+            .captures_iter(&format!("{self:?}"))
+            .map(|cap| cap[0].to_ascii_uppercase())
+            .join("_");
+
+        f.pad(&name)
+    }
+}

vm/src/run.rs

@@ -0,0 +1,97 @@
+use std::io::Write;
+
+use rlox2_frontend::lexer::{scan_tokens, Token};
+
+use crate::{compile, InterpretError, VM};
+
+/* pub fn vm_main() {
+    let mut chunk = Chunk::new();
+
+    let constant1 = chunk.add_constant(Value::Number(1.2));
+    let constant2 = chunk.add_constant(Value::Number(3.4));
+    let constant3 = chunk.add_constant(Value::Number(5.6));
+
+    chunk.write_opcode(Opcode::LoadConst, 123);
+    chunk.write_byte(constant1, 123);
+
+    chunk.write_opcode(Opcode::LoadConst, 123);
+    chunk.write_byte(constant2, 123);
+
+    chunk.write_opcode(Opcode::Add, 123);
+
+    chunk.write_opcode(Opcode::LoadConst, 124);
+    chunk.write_byte(constant3, 124);
+
+    chunk.write_opcode(Opcode::Divide, 124);
+    chunk.write_opcode(Opcode::Negate, 124);
+
+    chunk.write_opcode(Opcode::Return, 125);
+
+    println!();
+    chunk.disassemble("test chunk");
+    println!();
+
+    let mut vm = VM::new();
+
+    vm.interpret(&chunk).unwrap();
+} */
+
+pub fn run_repl(vm: &mut VM) {
+    let stdin = std::io::stdin();
+
+    loop {
+        let mut input_buf = String::new();
+
+        print!("> ");
+        std::io::stdout().flush().unwrap();
+
+        'inner: loop {
+            stdin.read_line(&mut input_buf).unwrap_or_else(|err| {
+                eprintln!("Could not read from stdin: {err}");
+                std::process::exit(66);
+            });
+
+            let num_open_braces = (input_buf.matches('{').count() as i64) - (input_buf.matches('}').count() as i64);
+            let num_open_parens = (input_buf.matches('(').count() as i64) - (input_buf.matches(')').count() as i64);
+            let num_open_brackets = (input_buf.matches('[').count() as i64) - (input_buf.matches(']').count() as i64);
+
+            // all braces/parens/brackets closed => break
+            if num_open_braces == 0 && num_open_parens == 0 && num_open_brackets == 0 {
+                break 'inner;
+            }
+
+            // any braces/parens/brackets more closing than opening => break (will be parse error)
+            if num_open_braces < 0 || num_open_parens < 0 || num_open_brackets < 0 {
+                break 'inner;
+            }
+
+            print!("< ");
+
+            // let indentation = "  ".repeat((num_open_braces + num_open_brackets + num_open_parens) as usize);
+
+            // print!("{indentation}");
+
+            std::io::stdout().flush().unwrap();
+        }
+
+        let input_buf = input_buf.trim();
+
+        if input_buf.is_empty() || input_buf == "exit" || input_buf == "quit" {
+            std::process::exit(0);
+        }
+
+        match run(input_buf, vm) {
+            Ok(()) => {}
+            Err(InterpretError::Exit { exit_code }) => std::process::exit(exit_code),
+            Err(err) => eprintln!("{err}"),
+        }
+    }
+}
+
+pub fn run(source: &str, _vm: &mut VM) -> Result<(), InterpretError> {
+    let tokens: Vec<Token> = scan_tokens(source)?;
+
+    compile(tokens);
+
+    Ok(())
+}

vm/src/value.rs

@@ -0,0 +1,22 @@
+use std::fmt::Display;
+
+#[derive(Debug, Clone, PartialEq)]
+pub enum Value {
+    Nil,
+    Number(f64),
+}
+
+impl Default for Value {
+    fn default() -> Self {
+        Value::Nil
+    }
+}
+
+impl Display for Value {
+    fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
+        match self {
+            Value::Nil => write!(f, "nil"),
+            Value::Number(num) => write!(f, "{num}"),
+        }
+    }
+}

vm/src/vm.rs

@@ -0,0 +1,225 @@
+use std::ptr;
+
+use num_traits::FromPrimitive;
+
+use crate::error::RuntimeError;
+use crate::{Chunk, Opcode};
+use crate::{InterpretError, Value};
+
+/*====================================================================================================================*/
+
+const STACK_MAX: usize = 256;
+const DEBUG_TRACE_EXECUTION: bool = true;
+
+/*====================================================================================================================*/
+
+#[derive(Debug)]
+pub struct VM {
+    chunk_ptr: *const Chunk,
+
+    ip: *const u8,
+
+    stack: [Value; STACK_MAX],
+    stack_top: *mut Value,
+}
+
+/* macro_rules! debug_println {
+    ($($arg:tt)*) => {
+        if cfg!(debug_assertions) {
+            println!($($arg)*);
+        }
+    };
+} */
+
+impl VM {
+    pub fn new() -> Self {
+        const NIL: Value = Value::Nil;
+        let stack = [NIL; STACK_MAX];
+
+        let mut vm = VM {
+            chunk_ptr: ptr::null(),
+            ip: ptr::null(),
+            stack,
+            stack_top: ptr::null_mut(),
+        };
+
+        vm.stack_top = vm.stack.as_mut_ptr();
+
+        vm
+    }
+
+    unsafe fn chunk(&self) -> &Chunk {
+        &*self.chunk_ptr
+    }
+
+    unsafe fn offset(&self) -> usize {
+        let offset = self.ip.offset_from(self.chunk().code().as_ptr());
+
+        debug_assert!(offset >= 0);
+
+        offset as usize
+    }
+
+    unsafe fn stack_len(&self) -> usize {
+        let offset = self.stack_top.offset_from(self.stack.as_ptr());
+
+        debug_assert!(offset >= 0);
+
+        offset as usize
+    }
+
+    unsafe fn push_value(&mut self, value: Value) {
+        // ptr::write(self.stack_top, value);
+        debug_assert!(self.stack_len() < STACK_MAX);
+        let old_value = std::mem::replace(&mut *self.stack_top, value);
+        debug_assert_eq!(old_value, Value::Nil);
+        self.stack_top = self.stack_top.add(1);
+    }
+
+    unsafe fn pop_value(&mut self) -> Value {
+        assert!(self.stack_len() < STACK_MAX);
+
+        self.stack_top = self.stack_top.sub(1);
+        // ptr::read(self.stack_top)
+        std::mem::take(&mut *self.stack_top)
+    }
+
+    pub fn interpret(&mut self, chunk: &Chunk) -> Result<(), InterpretError> {
+        self.chunk_ptr = chunk;
+        self.ip = chunk.code().as_ptr();
+
+        unsafe { self.run()? }
+
+        Ok(())
+    }
+
+    unsafe fn read_byte(&mut self) -> u8 {
+        debug_assert!(self.offset() < self.chunk().code().len());
+
+        let byte = *self.ip;
+        self.ip = self.ip.add(1);
+        byte
+    }
+
+    unsafe fn read_constant(&mut self) -> &Value {
+        let constant_idx = self.read_byte() as usize;
+
+        self.chunk().get_constant(constant_idx)
+    }
+
+    unsafe fn read_constant_long(&mut self) -> &Value {
+        let bytes = [self.read_byte(), self.read_byte(), self.read_byte(), 0];
+        let constant_idx = u32::from_le_bytes(bytes) as usize;
+
+        self.chunk().get_constant(constant_idx)
+    }
+
+    unsafe fn run(&mut self) -> Result<(), RuntimeError> {
+        loop {
+            if DEBUG_TRACE_EXECUTION {
+                println!();
+                self.print_stack();
+                self.chunk().disassemble_instruction(self.offset());
+                println!();
+            }
+
+            let opcode: Opcode = FromPrimitive::from_u8(self.read_byte()).unwrap();
+
+            match opcode {
+                Opcode::LoadConst => {
+                    let value = self.read_constant().clone();
+                    println!("Constant: {value}");
+                    self.push_value(value);
+                }
+                Opcode::LoadConstLong => {
+                    let value = self.read_constant_long().clone();
+                    println!("LongConstant: {value}");
+                    self.push_value(value);
+                }
+
+                Opcode::Add => {
+                    let right = self.pop_value();
+                    let left = self.pop_value();
+                    match (left, right) {
+                        (Value::Number(left), Value::Number(right)) => {
+                            self.push_value(Value::Number(left + right));
+                        }
+                        // (Value::String(left), Value::String(right)) => todo!(),
+                        (left, right) => return Err(RuntimeError::BinaryInvalidOperand { opcode, left, right }),
+                    }
+                }
+                Opcode::Subtract => {
+                    let right = self.pop_value();
+                    let left = self.pop_value();
+                    match (left, right) {
+                        (Value::Number(left), Value::Number(right)) => {
+                            self.push_value(Value::Number(left - right));
+                        }
+                        (left, right) => return Err(RuntimeError::BinaryInvalidOperand { opcode, left, right }),
+                    }
+                }
+                Opcode::Multiply => {
+                    let right = self.pop_value();
+                    let left = self.pop_value();
+                    match (left, right) {
+                        (Value::Number(left), Value::Number(right)) => {
+                            self.push_value(Value::Number(left * right));
+                        }
+                        (left, right) => return Err(RuntimeError::BinaryInvalidOperand { opcode, left, right }),
+                    }
+                }
+                Opcode::Divide => {
+                    let right = self.pop_value();
+                    let left = self.pop_value();
+                    match (left, right) {
+                        (Value::Number(left), Value::Number(right)) => {
+                            if right == 0.0 {
+                                return Err(RuntimeError::DivisionByZero);
+                            }
+                            self.push_value(Value::Number(left / right));
+                        }
+                        (left, right) => return Err(RuntimeError::BinaryInvalidOperand { opcode, left, right }),
+                    }
+                }
+                Opcode::Negate => {
+                    let value = self.pop_value();
+                    println!("Negate: {value}");
+                    if let Value::Number(num) = value {
+                        self.push_value(Value::Number(-num));
+                    } else {
+                        return Err(RuntimeError::UnaryInvalidOperand { opcode, operand: value });
+                    }
+                }
+                Opcode::Return => {
+                    let value = self.pop_value();
+                    debug_assert_eq!(self.stack_len(), 0);
+                    println!("Return: {value}");
+                    return Ok(());
+                }
+            }
+        }
+    }
+
+    unsafe fn print_stack(&self) {
+        /* let s = self.stack[0..self.stack_len()]
+            .iter()
+            .map(|value| value.to_string())
+            .join(", ");
+
+        print!("{s}"); */
+
+        print!("Stack:    ");
+
+        for value in self.stack[0..self.stack_len()].iter() {
+            print!("[ {value} ]");
+        }
+
+        println!()
+    }
+}
+
+impl Default for VM {
+    fn default() -> Self {
+        VM::new()
+    }
+}