This repository has been archived on 2026-09-15. You can view files and clone it. You cannot open issues or pull requests or push a commit.
Files
OxideJS/crates/oxide_compiler/src/compiler.rs
T
2026-06-06 09:52:00 +08:00

539 lines
20 KiB
Rust

use crate::module::CompiledModule;
use crate::opcode::{self, OpCode};
use crate::symbol_table::SymbolTable;
pub use crate::hash::structural_hash;
pub use crate::module::Constant;
pub use oxide_parser::{BinaryOperator, Expression, Statement, UnaryOperator};
pub struct Compiler;
fn is_int_literal(value: f64) -> bool {
value.fract() == 0.0 && value >= i32::MIN as f64 && value <= i32::MAX as f64
}
struct CompileCtx {
bytecode: Vec<opcode::Instr>,
constants: Vec<Constant>,
next_reg: u8,
max_regs: u8,
#[allow(dead_code)]
symbols: SymbolTable,
}
impl CompileCtx {
fn new() -> Self {
Self {
bytecode: Vec::new(),
constants: Vec::new(),
next_reg: 0,
max_regs: 0,
symbols: SymbolTable::new(),
}
}
fn emit(&mut self, instr: opcode::Instr) {
self.bytecode.push(instr);
}
fn alloc_reg(&mut self) -> u8 {
let r = self.next_reg;
self.next_reg += 1;
if self.next_reg > self.max_regs {
self.max_regs = self.next_reg;
}
r
}
fn reset_regs(&mut self) {
self.next_reg = 0;
}
fn add_constant(&mut self, c: Constant) -> u16 {
if let Some(idx) = self.constants.iter().position(|x| x == &c) {
return idx as u16;
}
let idx = self.constants.len();
self.constants.push(c);
idx as u16
}
#[allow(dead_code)]
fn push_scope(&mut self) {
self.symbols.push_scope();
}
#[allow(dead_code)]
fn pop_scope(&mut self) {
self.symbols.pop_scope();
}
#[allow(dead_code)]
fn declare(&mut self, name: &str, reg: u8) -> Result<(), String> {
self.symbols.declare(name, reg)
}
#[allow(dead_code)]
fn lookup(&self, name: &str) -> Result<u8, String> {
self.symbols.lookup(name)
}
#[allow(dead_code)]
fn lookup_or_global(&mut self, name: &str) -> u8 {
let reg = self.alloc_reg();
self.symbols.lookup_or_global(name, reg)
}
#[allow(dead_code)]
fn init_var(&mut self, name: &str) {
self.symbols.init_var(name);
}
}
impl Compiler {
pub fn new() -> Self {
Self
}
pub fn compile(&self, program: &oxide_parser::Program) -> Result<CompiledModule, String> {
let mut ctx = CompileCtx::new();
for stmt in &program.body {
self.count_statement(stmt, &mut ctx);
}
ctx.max_regs = ctx.max_regs.max(1);
ctx.reset_regs();
let mut last_result = 0u8;
for stmt in &program.body {
if let Some(r) = self.emit_statement(stmt, &mut ctx)? {
last_result = r;
}
}
if last_result != 0 {
ctx.emit(opcode::encode(OpCode::LOAD_VAR, 0, last_result, 0));
}
ctx.emit(opcode::encode(OpCode::HALT, 0, 0, 0));
Ok(CompiledModule {
bytecode: ctx.bytecode,
constants: ctx.constants,
n_registers: ctx.max_regs,
})
}
}
impl Default for Compiler {
fn default() -> Self {
Self::new()
}
}
// ── Pass 1: Counter ──
impl Compiler {
fn count_statement(&self, stmt: &Statement, ctx: &mut CompileCtx) {
match stmt {
Statement::ExpressionStatement(es) => {
self.count_expression(&es.expression, ctx);
}
Statement::VariableDeclaration(decl) => {
for d in &decl.declarations {
if let Some(init) = &d.init {
self.count_expression(init, ctx);
}
}
}
Statement::ReturnStatement(ret) => {
if let Some(arg) = &ret.argument {
self.count_expression(arg, ctx);
}
}
Statement::IfStatement(ifs) => {
self.count_expression(&ifs.test, ctx);
self.count_statement(&ifs.consequent, ctx);
if let Some(alt) = &ifs.alternate {
self.count_statement(alt, ctx);
}
}
Statement::WhileStatement(wh) => {
self.count_expression(&wh.test, ctx);
self.count_statement(&wh.body, ctx);
}
Statement::ForStatement(fr) => {
if let Some(init) = &fr.init {
if let Some(expr) = init.as_expression() {
self.count_expression(expr, ctx);
}
}
if let Some(test) = &fr.test {
self.count_expression(test, ctx);
}
if let Some(update) = &fr.update {
self.count_expression(update, ctx);
}
self.count_statement(&fr.body, ctx);
}
Statement::BlockStatement(block) => {
for s in &block.body {
self.count_statement(s, ctx);
}
}
_ => {}
}
}
fn count_expression(&self, expr: &Expression, ctx: &mut CompileCtx) {
match expr {
Expression::BinaryExpression(bin) => {
self.count_expression(&bin.left, ctx);
self.count_expression(&bin.right, ctx);
ctx.alloc_reg();
}
Expression::UnaryExpression(un) => {
self.count_expression(&un.argument, ctx);
ctx.alloc_reg();
}
Expression::CallExpression(call) => {
self.count_expression(&call.callee, ctx);
ctx.alloc_reg();
}
Expression::AssignmentExpression(assign) => {
if let oxide_parser::AssignmentTarget::StaticMemberExpression(member) = &assign.left
{
self.count_expression(&member.object, ctx);
self.count_expression(&assign.right, ctx);
ctx.alloc_reg();
} else if let oxide_parser::AssignmentTarget::ComputedMemberExpression(member) =
&assign.left
{
self.count_expression(&member.object, ctx);
self.count_expression(&member.expression, ctx);
self.count_expression(&assign.right, ctx);
ctx.alloc_reg();
} else {
self.count_expression(&assign.right, ctx);
ctx.alloc_reg();
}
}
Expression::ConditionalExpression(cond) => {
self.count_expression(&cond.test, ctx);
self.count_expression(&cond.consequent, ctx);
self.count_expression(&cond.alternate, ctx);
ctx.alloc_reg();
}
Expression::SequenceExpression(seq) => {
for e in &seq.expressions {
self.count_expression(e, ctx);
}
}
Expression::LogicalExpression(log) => {
self.count_expression(&log.left, ctx);
self.count_expression(&log.right, ctx);
ctx.alloc_reg();
}
Expression::ObjectExpression(obj) => {
ctx.alloc_reg();
for prop in &obj.properties {
if let oxide_parser::ObjectPropertyKind::ObjectProperty(p) = prop {
ctx.alloc_reg();
self.count_expression(&p.value, ctx);
}
}
}
Expression::ArrayExpression(arr) => {
ctx.alloc_reg();
for elem in &arr.elements {
if let Some(e) = elem.as_expression() {
self.count_expression(e, ctx);
ctx.alloc_reg();
}
}
}
Expression::StaticMemberExpression(member) => {
self.count_expression(&member.object, ctx);
ctx.alloc_reg();
ctx.alloc_reg();
}
Expression::ComputedMemberExpression(member) => {
self.count_expression(&member.object, ctx);
self.count_expression(&member.expression, ctx);
ctx.alloc_reg();
}
Expression::ParenthesizedExpression(p) => {
self.count_expression(&p.expression, ctx);
}
_ => {
ctx.alloc_reg();
}
}
}
}
// ── Pass 2: Emitter ──
impl Compiler {
fn emit_statement(&self, stmt: &Statement, ctx: &mut CompileCtx) -> Result<Option<u8>, String> {
match stmt {
Statement::ExpressionStatement(es) => {
Ok(Some(self.emit_expression(&es.expression, ctx)?))
}
Statement::VariableDeclaration(decl) => {
let mut r = None;
for d in &decl.declarations {
if let Some(init) = &d.init {
r = Some(self.emit_expression(init, ctx)?);
}
}
Ok(r)
}
Statement::ReturnStatement(ret) => {
match &ret.argument {
Some(expr) => {
let r = self.emit_expression(expr, ctx)?;
ctx.emit(opcode::encode(OpCode::RETURN, r, 0, 0));
}
None => {
ctx.emit(opcode::encode(OpCode::RETURN, 0, 0, 0));
}
}
Ok(None)
}
Statement::BlockStatement(block) => {
let mut r = None;
for s in &block.body {
if let Some(rr) = self.emit_statement(s, ctx)? {
r = Some(rr);
}
}
Ok(r)
}
_ => Ok(None),
}
}
fn emit_expression(&self, expr: &Expression, ctx: &mut CompileCtx) -> Result<u8, String> {
match expr {
Expression::NumericLiteral(n) => {
let idx = if is_int_literal(n.value) {
ctx.add_constant(Constant::Int(n.value as i32))
} else {
ctx.add_constant(Constant::Number(n.value))
};
let r = ctx.alloc_reg();
ctx.emit(opcode::encode(
OpCode::LOAD_CONST,
r,
(idx & 0xFF) as u8,
((idx >> 8) & 0xFF) as u8,
));
Ok(r)
}
Expression::StringLiteral(s) => {
let idx = ctx.add_constant(Constant::String(s.value.to_string()));
let r = ctx.alloc_reg();
ctx.emit(opcode::encode(
OpCode::LOAD_CONST,
r,
(idx & 0xFF) as u8,
((idx >> 8) & 0xFF) as u8,
));
Ok(r)
}
Expression::BooleanLiteral(b) => {
let idx = ctx.add_constant(Constant::Boolean(b.value));
let r = ctx.alloc_reg();
ctx.emit(opcode::encode(
OpCode::LOAD_CONST,
r,
(idx & 0xFF) as u8,
((idx >> 8) & 0xFF) as u8,
));
Ok(r)
}
Expression::NullLiteral(_) => {
let idx = ctx.add_constant(Constant::Null);
let r = ctx.alloc_reg();
ctx.emit(opcode::encode(
OpCode::LOAD_CONST,
r,
(idx & 0xFF) as u8,
((idx >> 8) & 0xFF) as u8,
));
Ok(r)
}
Expression::BinaryExpression(bin) => {
let left = self.emit_expression(&bin.left, ctx)?;
let right = self.emit_expression(&bin.right, ctx)?;
let op = match bin.operator {
BinaryOperator::Addition => OpCode::ADD,
BinaryOperator::Subtraction => OpCode::SUB,
BinaryOperator::Multiplication => OpCode::MUL,
BinaryOperator::Division => OpCode::DIV,
BinaryOperator::Remainder => OpCode::MOD,
BinaryOperator::Equality => OpCode::EQ,
BinaryOperator::Inequality => OpCode::NEQ,
BinaryOperator::LessThan => OpCode::LT,
BinaryOperator::GreaterThan => OpCode::GT,
BinaryOperator::LessEqualThan => OpCode::LTE,
BinaryOperator::GreaterEqualThan => OpCode::GTE,
BinaryOperator::In => OpCode::IN,
_ => return Err(format!("unsupported binary operator: {:?}", bin.operator)),
};
let r = ctx.alloc_reg();
ctx.emit(opcode::encode(op, r, left, right));
Ok(r)
}
Expression::UnaryExpression(un) => {
let arg = self.emit_expression(&un.argument, ctx)?;
match un.operator {
UnaryOperator::UnaryNegation => {
let r = ctx.alloc_reg();
ctx.emit(opcode::encode(OpCode::NEG, r, arg, 0));
Ok(r)
}
UnaryOperator::Typeof => {
let r = ctx.alloc_reg();
ctx.emit(opcode::encode(OpCode::TYPEOF, r, arg, 0));
Ok(r)
}
UnaryOperator::Void => {
let r = ctx.alloc_reg();
ctx.emit(opcode::encode(OpCode::VOID, r, arg, 0));
Ok(r)
}
_ => Err(format!("unsupported unary operator: {:?}", un.operator)),
}
}
Expression::StaticMemberExpression(member) => {
let obj_reg = self.emit_expression(&member.object, ctx)?;
let prop_name = member.property.name.as_str();
let idx = ctx.add_constant(Constant::String(prop_name.to_string()));
let key_reg = ctx.alloc_reg();
ctx.emit(opcode::encode(
OpCode::LOAD_CONST,
key_reg,
(idx & 0xFF) as u8,
((idx >> 8) & 0xFF) as u8,
));
let result_reg = ctx.alloc_reg();
ctx.emit(opcode::encode(
OpCode::IC_GET_PROP,
result_reg,
obj_reg,
key_reg,
));
ctx.emit(0);
Ok(result_reg)
}
Expression::ComputedMemberExpression(member) => {
let obj_reg = self.emit_expression(&member.object, ctx)?;
let key_reg = self.emit_expression(&member.expression, ctx)?;
let r = ctx.alloc_reg();
ctx.emit(opcode::encode(
OpCode::GET_PROP_DYNAMIC,
obj_reg,
key_reg,
r,
));
Ok(r)
}
Expression::ObjectExpression(obj) => {
let obj_reg = ctx.alloc_reg();
ctx.emit(opcode::encode(OpCode::NEW_OBJECT, obj_reg, 0, 0));
for prop in &obj.properties {
let oxide_parser::ObjectPropertyKind::ObjectProperty(p) = prop else {
return Err("spread properties not yet supported".into());
};
let prop_name = match &p.key {
oxide_parser::PropertyKey::StaticIdentifier(ident) => {
ident.name.as_str().to_string()
}
oxide_parser::PropertyKey::StringLiteral(s) => s.value.to_string(),
_ => return Err("unsupported object property key type".into()),
};
let idx = ctx.add_constant(Constant::String(prop_name.to_string()));
let key_reg = ctx.alloc_reg();
ctx.emit(opcode::encode(
OpCode::LOAD_CONST,
key_reg,
(idx & 0xFF) as u8,
((idx >> 8) & 0xFF) as u8,
));
let val_reg = self.emit_expression(&p.value, ctx)?;
ctx.emit(opcode::encode(OpCode::SET_PROP, obj_reg, val_reg, key_reg));
}
Ok(obj_reg)
}
Expression::ArrayExpression(arr) => {
let arr_reg = ctx.alloc_reg();
let n = arr.elements.len() as u16;
ctx.emit(opcode::encode(
OpCode::NEW_ARRAY,
arr_reg,
(n & 0xFF) as u8,
((n >> 8) & 0xFF) as u8,
));
for (i, elem) in arr.elements.iter().enumerate() {
let Some(e) = elem.as_expression() else {
return Err("spread not supported".into());
};
let val_reg = self.emit_expression(e, ctx)?;
let idx_reg = ctx.alloc_reg();
let idx = ctx.add_constant(Constant::Number(i as f64));
ctx.emit(opcode::encode(
OpCode::LOAD_CONST,
idx_reg,
(idx & 0xFF) as u8,
((idx >> 8) & 0xFF) as u8,
));
ctx.emit(opcode::encode(OpCode::SET_ELEM, arr_reg, idx_reg, val_reg));
}
Ok(arr_reg)
}
Expression::AssignmentExpression(assign) => {
if let oxide_parser::AssignmentTarget::StaticMemberExpression(member) = &assign.left
{
let obj_reg = self.emit_expression(&member.object, ctx)?;
let val_reg = self.emit_expression(&assign.right, ctx)?;
let prop_name = member.property.name.as_str();
let idx = ctx.add_constant(Constant::String(prop_name.to_string()));
let key_reg = ctx.alloc_reg();
ctx.emit(opcode::encode(
OpCode::LOAD_CONST,
key_reg,
(idx & 0xFF) as u8,
((idx >> 8) & 0xFF) as u8,
));
ctx.emit(opcode::encode(
OpCode::IC_SET_PROP,
obj_reg,
val_reg,
key_reg,
));
ctx.emit(0);
Ok(val_reg)
} else if let oxide_parser::AssignmentTarget::ComputedMemberExpression(member) =
&assign.left
{
let obj_reg = self.emit_expression(&member.object, ctx)?;
let key_reg = self.emit_expression(&member.expression, ctx)?;
let val_reg = self.emit_expression(&assign.right, ctx)?;
ctx.emit(opcode::encode(
OpCode::SET_PROP_DYNAMIC,
obj_reg,
key_reg,
val_reg,
));
Ok(val_reg)
} else {
Err("assignment target not supported".into())
}
}
Expression::ParenthesizedExpression(p) => self.emit_expression(&p.expression, ctx),
_ => Err(format!("unsupported expression type: {:?}", expr)),
}
}
}