Merge pull request #15 from mist-go/prepare-rust-compiler

Prepare rust compiler
This commit is contained in:
2026-05-01 17:52:17 +02:00
committed by GitHub
12 changed files with 102 additions and 1086 deletions
Generated
-10
View File
@@ -79,7 +79,6 @@ name = "mist"
version = "0.1.0" version = "0.1.0"
dependencies = [ dependencies = [
"parser", "parser",
"semantic",
"serde", "serde",
"serde_json", "serde_json",
] ]
@@ -155,15 +154,6 @@ dependencies = [
"proc-macro2", "proc-macro2",
] ]
[[package]]
name = "semantic"
version = "0.1.0"
dependencies = [
"parser",
"serde",
"serde_json",
]
[[package]] [[package]]
name = "serde" name = "serde"
version = "1.0.228" version = "1.0.228"
-1
View File
@@ -7,4 +7,3 @@ edition = "2024"
serde = { version = "1.0.228", features = ["serde_derive"] } serde = { version = "1.0.228", features = ["serde_derive"] }
serde_json = "1.0.149" serde_json = "1.0.149"
parser = { path = "./parser" } parser = { path = "./parser" }
semantic = { path = "./semantic" }
+2 -9
View File
@@ -64,7 +64,7 @@ pub enum Statement {
// let/const/var x = ... // let/const/var x = ...
VarDecl { VarDecl {
kind: VarKind, mutable: bool,
name: String, name: String,
init: Option<Expression>, init: Option<Expression>,
type_: Option<TypeExpr>, type_: Option<TypeExpr>,
@@ -90,7 +90,7 @@ pub enum Statement {
// for (...) stmt // for (...) stmt
For { For {
init: (VarKind, String, Option<Expression>), init: (bool, String, Option<Expression>),
condition: Option<Expression>, condition: Option<Expression>,
update: Option<Box<Statement>>, update: Option<Box<Statement>>,
body: Box<Statement>, body: Box<Statement>,
@@ -115,10 +115,3 @@ pub enum Expression {
postfixes: Vec<Postfix>, postfixes: Vec<Postfix>,
}, },
} }
#[derive(Debug, Clone, Serialize)]
pub enum VarKind {
Let,
Const,
Var,
}
+2 -9
View File
@@ -188,7 +188,7 @@ impl From<pest::iterators::Pair<'_, Rule>> for Statement {
let init = inner.next().map(Expression::from); let init = inner.next().map(Expression::from);
Statement::VarDecl { Statement::VarDecl {
kind: VarKind::Var, mutable: false,
name: name.as_str().to_string(), name: name.as_str().to_string(),
init, init,
type_, type_,
@@ -243,19 +243,12 @@ impl From<pest::iterators::Pair<'_, Rule>> for Statement {
Rule::var_decl => { Rule::var_decl => {
let mut it = p.into_inner(); let mut it = p.into_inner();
let kind = match it.next().unwrap().as_str() {
"let" => VarKind::Let,
"const" => VarKind::Const,
"var" => VarKind::Var,
_ => unreachable!(),
};
let name = it.next().unwrap().as_str().to_string(); let name = it.next().unwrap().as_str().to_string();
let init_expr = it let init_expr = it
.next() .next()
.map(|e| Expression::from(e.into_inner().next().unwrap())); .map(|e| Expression::from(e.into_inner().next().unwrap()));
(kind, name, init_expr) (false, name, init_expr)
} }
_ => unimplemented!( _ => unimplemented!(
"For loop init parsing not implemented yet: {:?}", "For loop init parsing not implemented yet: {:?}",
-9
View File
@@ -1,9 +0,0 @@
[package]
name = "semantic"
version = "0.1.0"
edition = "2024"
[dependencies]
parser = { path = "../parser" }
serde = { version = "1.0.228", features = ["serde_derive"] }
serde_json = "1.0.149"
-318
View File
@@ -1,318 +0,0 @@
use std::{
collections::HashMap,
sync::{
Arc,
atomic::{AtomicUsize, Ordering},
},
};
use crate::top_level::{
FunctionSymbol, JSONScope, StructSymbol, TopLevelSymbolScope, TypeSymbol, VarSymbol,
};
#[derive(Clone, Debug)]
pub enum TypeRef {
Struct(StructRef),
Function(FunctionRef),
Package(PackageRef),
Name(String),
}
#[derive(Clone, Debug)]
pub struct PackageRef {
pub name: String,
pub variables: HashMap<String, Arc<VarRef>>,
}
#[derive(Clone, Debug)]
pub struct VarRef {
pub export: bool,
pub var_type: Arc<TypeRef>,
pub name: String,
}
#[derive(Clone, Debug)]
pub struct FunctionRef {
pub export: bool,
pub name: String,
pub params: HashMap<String, Arc<VarRef>>,
pub return_type: Option<Arc<TypeRef>>,
}
#[derive(Clone, Debug)]
pub struct StructRef {
pub export: bool,
pub name: String,
pub fields: HashMap<String, Arc<VarRef>>,
}
#[derive(Debug)]
pub struct TopLevelHirScope {
pub variables: HashMap<String, Arc<VarRef>>,
pub var_idx: AtomicUsize,
}
impl TopLevelHirScope {
pub fn from_tlss(tlss: &TopLevelSymbolScope) -> Self {
let mut scope = Self {
variables: HashMap::new(),
var_idx: AtomicUsize::new(0),
};
for (_, symbol) in &tlss.functions {
scope.function_ref(tlss, symbol);
}
for (_, symbol) in &tlss.structs {
scope.struct_ref(tlss, symbol);
}
for (_, symbol) in &tlss.imports {
scope.variables.insert(
symbol.package_name.clone(),
Arc::new(VarRef {
export: false,
name: symbol.package_name.clone(),
var_type: Arc::new(TypeRef::Package(PackageRef::from_tlss(symbol))),
}),
);
}
scope
}
pub fn function_ref(&mut self, tlss: &TopLevelSymbolScope, symbol: &FunctionSymbol) {
if self.variables.get(&symbol.name).is_none() {
let name = self.get_name(symbol.export);
if let Some(_) = tlss.functions.get(&symbol.name) {
let rf = FunctionRef {
export: symbol.export,
name: name.clone(),
params: symbol
.params
.iter()
.map(|(name, v)| (name.clone(), self.var_ref(tlss, v)))
.collect(),
return_type: symbol
.return_type
.clone()
.map(|rt| self.type_ref(tlss, &rt)),
};
self.variables.insert(
symbol.name.clone(),
Arc::new(VarRef {
export: false,
name: name.clone(),
var_type: Arc::new(TypeRef::Function(rf)),
}),
);
} else {
unimplemented!()
}
}
}
pub fn struct_ref(
&mut self,
tlss: &TopLevelSymbolScope,
symbol: &StructSymbol,
) -> Arc<TypeRef> {
if let Some(rf) = self.variables.get(&symbol.name) {
rf.var_type.clone()
} else {
let name = self.get_name(symbol.export);
let rf = Arc::new(TypeRef::Struct(StructRef {
export: symbol.export,
name: name.clone(),
fields: symbol
.fields
.iter()
.map(|(name, v)| (name.clone(), self.var_ref(tlss, v)))
.collect(),
}));
self.variables.insert(
symbol.name.clone(),
Arc::new(VarRef {
export: symbol.export,
name: name.clone(),
var_type: rf.clone(),
}),
);
rf
}
}
pub fn var_ref(&mut self, tlss: &TopLevelSymbolScope, symbol: &VarSymbol) -> Arc<VarRef> {
Arc::new(VarRef {
export: symbol.export,
var_type: self.type_ref(tlss, &symbol.var_type),
name: self.get_name(symbol.export),
})
}
pub fn type_ref(&mut self, tlss: &TopLevelSymbolScope, symbol: &TypeSymbol) -> Arc<TypeRef> {
if let Some(var_ref) = self.variables.get(&symbol.0) {
var_ref.var_type.clone()
} else {
if let Some(tlss_rf) = tlss.structs.get(&symbol.0) {
self.struct_ref(tlss, tlss_rf)
} else {
let var_ref = Arc::new(VarRef {
export: false,
name: symbol.0.clone(),
var_type: Arc::new(TypeRef::Name(symbol.0.clone())),
});
self.variables.insert(symbol.0.clone(), var_ref.clone());
var_ref.var_type.clone()
}
}
}
pub fn get_reference(&self, name: &String) -> Option<Arc<VarRef>> {
self.variables.get(name).cloned()
}
pub fn next_var_idx(&self) -> usize {
self.var_idx.fetch_add(1, Ordering::Relaxed)
}
pub fn get_name(&self, export: bool) -> String {
format!("{}{}", if export { 'V' } else { 'v' }, self.next_var_idx())
}
}
impl TypeRef {
pub fn get_name(&self) -> String {
match self {
TypeRef::Function(f) => f.name.clone(),
TypeRef::Struct(s) => s.name.clone(),
TypeRef::Package(p) => p.name.clone(),
TypeRef::Name(n) => n.clone(),
}
}
}
impl PackageRef {
pub fn from_tlss(json_scope: &JSONScope) -> Self {
let mut scope = Self {
name: json_scope.package_name.clone(),
variables: HashMap::new(),
};
for (token_name, symbol) in &json_scope.functions {
scope.function_ref(json_scope, token_name, symbol);
}
for (_, symbol) in &json_scope.structs {
scope.struct_ref(json_scope, symbol);
}
scope
}
pub fn function_ref(
&mut self,
json_scope: &JSONScope,
token_name: &String,
symbol: &FunctionSymbol,
) {
if self.variables.get(&symbol.name).is_none() {
let name = symbol.name.clone();
if let Some(_) = json_scope.functions.get(token_name) {
let rf = FunctionRef {
export: symbol.export,
name: name.clone(),
params: symbol
.params
.iter()
.map(|(name, v)| (name.clone(), self.var_ref(json_scope, v)))
.collect(),
return_type: symbol
.return_type
.clone()
.map(|rt| self.type_ref(json_scope, &rt)),
};
self.variables.insert(
token_name.clone(),
Arc::new(VarRef {
export: false,
name: name.clone(),
var_type: Arc::new(TypeRef::Function(rf)),
}),
);
} else {
unimplemented!()
}
}
}
pub fn struct_ref(&mut self, json_scope: &JSONScope, symbol: &StructSymbol) -> Arc<TypeRef> {
if let Some(rf) = self.variables.get(&symbol.name) {
rf.var_type.clone()
} else {
let name = symbol.name.clone();
let rf = Arc::new(TypeRef::Struct(StructRef {
export: symbol.export,
name: name.clone(),
fields: symbol
.fields
.iter()
.map(|(name, v)| (name.clone(), self.var_ref(json_scope, v)))
.collect(),
}));
self.variables.insert(
symbol.name.clone(),
Arc::new(VarRef {
export: symbol.export,
name: name.clone(),
var_type: rf.clone(),
}),
);
rf
}
}
pub fn var_ref(&mut self, json_scope: &JSONScope, symbol: &VarSymbol) -> Arc<VarRef> {
Arc::new(VarRef {
export: symbol.export,
var_type: self.type_ref(json_scope, &symbol.var_type),
name: symbol.name.clone(),
})
}
pub fn type_ref(&mut self, json_scope: &JSONScope, symbol: &TypeSymbol) -> Arc<TypeRef> {
if let Some(var_ref) = self.variables.get(&symbol.0) {
var_ref.var_type.clone()
} else {
if let Some(json_scope_rf) = json_scope.structs.get(&symbol.0) {
self.struct_ref(json_scope, json_scope_rf)
} else {
let var_ref = Arc::new(VarRef {
export: false,
name: symbol.0.clone(),
var_type: Arc::new(TypeRef::Name(symbol.0.clone())),
});
self.variables.insert(symbol.0.clone(), var_ref.clone());
var_ref.var_type.clone()
}
}
}
pub fn get_reference(&self, name: &String) -> Option<Arc<VarRef>> {
self.variables.get(name).cloned()
}
}
-61
View File
@@ -1,61 +0,0 @@
use std::{collections::HashMap, sync::Arc};
use parser::ast::{ParamList, TypeExpr};
use crate::{
hir::VarRef,
scope::{LocalScope, Scope},
};
pub mod hir;
pub mod scope;
pub mod top_level;
pub fn walk_ast(top_scope: Arc<Scope>, tl: &mut Vec<parser::ast::TopLevel>) {
for tl in tl {
match tl {
parser::ast::TopLevel::Import(_) => {}
parser::ast::TopLevel::FunctionDecl {
params, body, name, ..
} => {
let rf = top_scope.get_reference(name).unwrap();
if name != "main" {
*name = rf.name.clone();
}
let scope = LocalScope::new(top_scope.clone());
scope.with_params(params);
scope.with_block(body);
}
parser::ast::TopLevel::StructDecl { name, fields, .. } => {
let rf = top_scope.get_reference(name).unwrap();
*name = rf.name.clone();
match &*rf.var_type {
hir::TypeRef::Struct(s) => walk_param_list(&s.fields, fields),
_ => unimplemented!(),
}
}
}
}
}
pub fn walk_param_list(fields: &HashMap<String, Arc<VarRef>>, param_list: &mut ParamList) {
let old_param_list = param_list.clone();
param_list.0.clear();
for (name, param) in fields {
param_list.0.insert(
param.name.clone(),
(
old_param_list.0.get(name).map(|a| a.0).unwrap_or_default(),
TypeExpr::Identifier(param.var_type.get_name()),
),
);
}
}
-294
View File
@@ -1,294 +0,0 @@
use std::{
collections::HashMap,
sync::{Arc, Mutex},
};
use parser::ast::{self, ParamList, Postfix, Statement};
use crate::{
hir::{TopLevelHirScope, TypeRef, VarRef},
top_level::TopLevelSymbolScope,
};
#[derive(Debug)]
pub enum Scope {
TopLevel(TopLevelHirScope),
Local(LocalScope),
}
impl Scope {
pub fn from_top(
path: &std::path::PathBuf,
top_level: &Vec<parser::ast::TopLevel>,
) -> Arc<Self> {
let tl = TopLevelSymbolScope::from(path, top_level);
Arc::new(Self::TopLevel(TopLevelHirScope::from_tlss(&tl)))
}
pub fn get_reference(&self, name: &String) -> Option<Arc<VarRef>> {
match self {
Scope::TopLevel(tl) => tl.get_reference(name),
Scope::Local(l) => l.get_reference(name),
}
}
pub fn next_var_idx(&self) -> usize {
match self {
Scope::TopLevel(tl) => tl.next_var_idx(),
Scope::Local(l) => l.parent.next_var_idx(),
}
}
pub fn get_name(&self, export: bool) -> String {
format!("{}{}", if export { 'V' } else { 'v' }, self.next_var_idx())
}
}
#[derive(Debug)]
pub struct LocalScope {
pub parent: Arc<Scope>,
pub variables: Mutex<HashMap<String, Arc<VarRef>>>,
}
impl LocalScope {
pub fn new(parent: Arc<Scope>) -> Arc<Self> {
Arc::new(LocalScope {
parent,
variables: Mutex::new(HashMap::new()),
})
}
pub fn get_reference(&self, name: &String) -> Option<Arc<VarRef>> {
self.variables
.lock()
.unwrap()
.get(name)
.cloned()
.or_else(|| self.parent.get_reference(name))
}
pub fn get_type_reference(&self, name: &String) -> Option<Arc<VarRef>> {
match self.get_reference(name) {
Some(v) => Some(v),
None => {
let var_ref = Arc::new(VarRef {
export: false,
name: name.clone(),
var_type: Arc::new(TypeRef::Name(name.clone())),
});
self.variables
.lock()
.unwrap()
.insert(name.clone(), var_ref.clone());
Some(var_ref)
}
}
}
pub fn with_statement(self: &Arc<Self>, statement: &mut Statement) {
match statement {
Statement::Block(b) => self.clone().with_block(b),
Statement::VarDecl {
name, init, type_, ..
} => {
if let Some(init) = init {
let init_type = self.get_type_from_expr(init);
let var_type = type_
.as_mut()
.map(|t| self.get_type_from_type_expr(t))
.or_else(|| Some(init_type))
.unwrap()
.unwrap();
let var_name = name.clone();
*name = self.parent.get_name(false);
self.variables.lock().unwrap().insert(
var_name,
Arc::new(VarRef {
export: false,
name: name.clone(),
var_type,
}),
);
}
}
Statement::Expression(e) => {
self.get_type_from_expr(e);
}
Statement::If {
condition,
then_branch,
else_branch,
} => {
self.get_type_from_expr(condition);
self.clone().with_statement(then_branch);
if let Some(else_branch) = else_branch {
self.clone().with_statement(else_branch);
}
}
Statement::Return(expr) => {
if let Some(expr) = expr {
self.get_type_from_expr(expr);
}
}
_ => {}
}
}
pub fn with_block(self: &Arc<Self>, block: &mut parser::ast::Block) {
for statement in &mut block.0 {
self.with_statement(statement);
}
}
pub fn walk_postfixes(
self: &Arc<Self>,
initial: &mut Box<ast::Expression>,
postfixes: &mut Vec<Postfix>,
) -> Option<Arc<TypeRef>> {
let mut current_type = self.get_type_from_expr(initial)?;
for postfix in postfixes {
match postfix {
Postfix::FieldAccess(id) => match &*current_type {
TypeRef::Struct(s) => {
let field = s.fields.get(id)?;
*id = field.name.clone();
current_type = field.var_type.clone();
}
TypeRef::Package(p) => {
let var_ref = p.variables.get(id)?;
*id = var_ref.name.clone();
current_type = var_ref.var_type.clone();
}
_ => unimplemented!(),
},
Postfix::Call(args) => match &*current_type {
TypeRef::Function(s) => {
for arg in args {
self.get_type_from_expr(arg);
}
current_type = s.return_type.clone()?;
}
_ => unimplemented!(),
},
Postfix::StructCall(fields) => match &*current_type {
TypeRef::Struct(s) => {
let mut old_fields = fields.clone();
fields.clear();
for (name, mut expr) in old_fields.drain() {
self.get_type_from_expr(&mut expr);
fields.insert(s.fields.get(&name).unwrap().name.clone(), expr);
}
}
_ => unimplemented!(),
},
Postfix::Binary(op, right) => match op {
parser::ast::BinaryOp::Equal
| parser::ast::BinaryOp::NotEqual
| parser::ast::BinaryOp::GreaterThan
| parser::ast::BinaryOp::LessThan
| parser::ast::BinaryOp::GreaterThanOrEqual
| parser::ast::BinaryOp::LessThanOrEqual => {
self.get_type_from_expr(right)?;
current_type = Arc::new(TypeRef::Name("bool".to_string()));
}
parser::ast::BinaryOp::Plus
| parser::ast::BinaryOp::Minus
| parser::ast::BinaryOp::Multiply
| parser::ast::BinaryOp::Divide
| parser::ast::BinaryOp::Modulo => {
self.get_type_from_expr(right)?;
}
},
_ => unimplemented!(),
}
}
Some(current_type)
}
pub fn get_type_from_type_expr(
self: &Arc<Self>,
expr: &mut ast::TypeExpr,
) -> Option<Arc<TypeRef>> {
match expr {
ast::TypeExpr::Identifier(id) => {
if let Some(v) = self.get_type_reference(id).map(|r| r.var_type.clone()) {
*id = v.get_name();
Some(v)
} else {
None
}
}
}
}
pub fn get_type_from_expr(
self: &Arc<Self>,
expr: &mut ast::Expression,
) -> Option<Arc<TypeRef>> {
match expr {
ast::Expression::IntLiteral(_) => self
.get_type_reference(&"int".to_string())
.map(|r| r.var_type.clone()),
ast::Expression::FloatLiteral(_) => self
.get_type_reference(&"float".to_string())
.map(|r| r.var_type.clone()),
ast::Expression::BoolLiteral(_) => self
.get_type_reference(&"bool".to_string())
.map(|r| r.var_type.clone()),
ast::Expression::StringLiteral(_) => self
.get_type_reference(&"string".to_string())
.map(|r| r.var_type.clone()),
ast::Expression::Identifier(id) => {
let rf = self.get_reference(id)?;
*id = rf.name.clone();
Some(rf.var_type.clone())
}
ast::Expression::Postfix { initial, postfixes } => {
self.walk_postfixes(initial, postfixes)
}
}
}
pub fn with_params(self: &Arc<Self>, param_list: &mut ParamList) {
let mut new_param_list = HashMap::new();
for (param_name, (export, type_expr)) in param_list.0.drain() {
match &type_expr {
parser::ast::TypeExpr::Identifier(id) => {
let new_name = self.parent.get_name(export);
self.variables.lock().unwrap().insert(
param_name.clone(),
Arc::new(VarRef {
export: export,
var_type: self
.parent
.get_reference(id)
.map(|r| r.var_type.clone())
.unwrap(),
name: new_name.clone(),
}),
);
new_param_list.insert(new_name, (export, type_expr));
}
}
}
param_list.0 = new_param_list;
}
}
-174
View File
@@ -1,174 +0,0 @@
use std::collections::HashMap;
use std::fs;
use std::path::PathBuf;
use parser::ast::TopLevel;
use parser::ast::{ParamList, TypeExpr};
use serde::{Deserialize, Serialize};
#[derive(Clone, Debug, Serialize, Deserialize)]
pub struct TypeSymbol(pub String);
#[derive(Clone, Debug, Serialize, Deserialize)]
pub struct VarSymbol {
pub export: bool,
pub var_type: TypeSymbol,
pub name: String,
}
#[derive(Clone, Debug, Serialize, Deserialize)]
pub struct FunctionSymbol {
pub export: bool,
pub name: String,
pub params: HashMap<String, VarSymbol>,
pub return_type: Option<TypeSymbol>,
}
#[derive(Clone, Debug, Serialize, Deserialize)]
pub struct StructSymbol {
pub export: bool,
pub name: String,
pub fields: HashMap<String, VarSymbol>,
pub methods: HashMap<String, FunctionSymbol>,
}
impl TypeSymbol {
pub fn from_ast(expr: TypeExpr) -> Self {
TypeSymbol(match expr {
TypeExpr::Identifier(i) => i.to_string(),
})
}
}
impl FunctionSymbol {
pub fn from_ast(
export: bool,
name: String,
params: ParamList,
return_type: Option<TypeExpr>,
) -> Self {
Self {
export: export,
name: name.clone(),
params: params
.0
.iter()
.map(|(name, (export, v))| {
(
name.clone(),
VarSymbol {
export: *export,
name: name.clone(),
var_type: TypeSymbol(match v {
TypeExpr::Identifier(i) => i.to_string(),
}),
},
)
})
.collect(),
return_type: return_type.map(TypeSymbol::from_ast),
}
}
}
impl StructSymbol {
pub fn from_ast(export: bool, name: String, fields: ParamList) -> Self {
Self {
export,
name,
fields: fields
.0
.iter()
.map(|(name, (export, v))| {
(
name.clone(),
VarSymbol {
export: *export,
name: name.clone(),
var_type: TypeSymbol(match v {
TypeExpr::Identifier(i) => i.to_string(),
}),
},
)
})
.collect(),
// TODO - parse struct methods
methods: HashMap::new(),
}
}
}
#[derive(Clone, Debug, Serialize, Deserialize)]
pub struct JSONScope {
pub structs: HashMap<String, StructSymbol>,
pub functions: HashMap<String, FunctionSymbol>,
pub package_name: String,
}
type JsonScopeMap = HashMap<String, JSONScope>;
#[derive(Clone, Debug)]
pub struct TopLevelSymbolScope {
pub imports: HashMap<String, JSONScope>,
pub structs: HashMap<String, StructSymbol>,
pub functions: HashMap<String, FunctionSymbol>,
}
impl TopLevelSymbolScope {
pub fn from(path: &PathBuf, top_level: &Vec<TopLevel>) -> Self {
let mut scope = TopLevelSymbolScope {
imports: HashMap::new(),
structs: HashMap::new(),
functions: HashMap::new(),
};
for top in top_level {
match top {
TopLevel::Import(package) => {
let json = fs::read(path.join("mist.map.json")).unwrap();
let json_scope_map: JsonScopeMap = serde_json::from_slice(&json).unwrap();
let json_scope = json_scope_map
.get(&package.replace("\"", ""))
.unwrap()
.clone();
scope
.imports
.insert(json_scope.package_name.clone(), json_scope);
}
TopLevel::FunctionDecl {
export,
name,
params,
return_type,
..
} => {
scope.functions.insert(
name.clone(),
FunctionSymbol::from_ast(
*export,
name.clone(),
params.clone(),
return_type.clone(),
),
);
}
TopLevel::StructDecl {
export,
name,
fields,
} => {
scope.structs.insert(
name.clone(),
StructSymbol::from_ast(*export, name.clone(), fields.clone()),
);
}
}
}
scope
}
}
+94 -186
View File
@@ -1,11 +1,11 @@
use parser::ast::{BinaryOp, Block, Expression, Postfix, Statement, TopLevel, TypeExpr, VarKind}; use parser::ast::{BinaryOp, Block, Expression, Postfix, Statement, TopLevel, TypeExpr};
pub struct GoCodegen { pub struct RustCodegen {
output: String, output: String,
indent: usize, indent: usize,
} }
impl GoCodegen { impl RustCodegen {
pub fn new() -> Self { pub fn new() -> Self {
Self { Self {
output: String::new(), output: String::new(),
@@ -17,10 +17,6 @@ impl GoCodegen {
" ".repeat(self.indent) " ".repeat(self.indent)
} }
fn add_indented(&mut self, s: &str) {
self.output.push_str(&format!("{}{}", self.indent_str(), s));
}
fn add(&mut self, s: &str) { fn add(&mut self, s: &str) {
self.output.push_str(s); self.output.push_str(s);
} }
@@ -31,8 +27,7 @@ impl GoCodegen {
} }
fn add_indentedln(&mut self, s: &str) { fn add_indentedln(&mut self, s: &str) {
self.add_indented(s); self.add(&format!("{}{}\n", self.indent_str(), s));
self.add("\n");
} }
pub fn generate(&mut self, toplevels: &[TopLevel]) -> String { pub fn generate(&mut self, toplevels: &[TopLevel]) -> String {
@@ -45,29 +40,29 @@ impl GoCodegen {
fn generate_toplevel(&mut self, tl: &TopLevel) { fn generate_toplevel(&mut self, tl: &TopLevel) {
match tl { match tl {
TopLevel::Import(path) => { TopLevel::Import(path) => {
let import_path = path.replace("\"", ""); let path = path.replace("\"", "");
if import_path.starts_with("./") || import_path.starts_with("/") { self.addln(&format!("use {};", path));
self.addln(&format!("import \"{}\"", import_path));
} else {
self.addln(&format!("import \"{}\"", import_path));
}
self.addln("");
} }
TopLevel::StructDecl { TopLevel::StructDecl {
export, export,
name, name,
fields, fields,
} => { } => {
let name = if *export { name } else { name }; let vis = if *export { "pub " } else { "" };
self.addln(&format!("type {} struct {{", name));
self.addln(&format!("{}struct {} {{", vis, name));
self.indent += 1; self.indent += 1;
for (field_name, (_, ty)) in &fields.0 { for (field_name, (_, ty)) in &fields.0 {
let go_ty = self.translate_type(ty); let ty = self.translate_type(ty);
self.addln(&format!("{} {}", field_name, go_ty)); self.add_indentedln(&format!("pub {}: {},", field_name, ty));
} }
self.indent -= 1; self.indent -= 1;
self.addln("}\n"); self.addln("}\n");
} }
TopLevel::FunctionDecl { TopLevel::FunctionDecl {
export, export,
name, name,
@@ -75,31 +70,26 @@ impl GoCodegen {
return_type, return_type,
body, body,
} => { } => {
let name = if *export { let vis = if *export { "pub " } else { "" };
format!("{}", name)
} else {
name.clone()
};
let params_str = params let params_str = params
.0 .0
.iter() .iter()
.map(|(n, (_, t))| format!("{} {}", n, self.translate_type(t))) .map(|(n, (_, t))| format!("{}: {}", n, self.translate_type(t)))
.collect::<Vec<_>>() .collect::<Vec<_>>()
.join(", "); .join(", ");
let ret_ty = return_type let ret = return_type
.as_ref() .as_ref()
.map(|t| self.translate_type(t)) .map(|t| format!(" -> {}", self.translate_type(t)))
.unwrap_or_else(|| "".to_string()); .unwrap_or_default();
self.addln(&format!("{}fn {}({}){} {{", vis, name, params_str, ret));
if ret_ty.is_empty() {
self.addln(&format!("func {}({}) {{", name, params_str));
} else {
self.addln(&format!("func {}({}) {} {{", name, params_str, ret_ty));
}
self.indent += 1; self.indent += 1;
self.generate_block(body); self.generate_block(body);
self.indent -= 1; self.indent -= 1;
self.addln("}\n"); self.addln("}\n");
} }
} }
@@ -108,13 +98,11 @@ impl GoCodegen {
fn translate_type(&self, ty: &TypeExpr) -> String { fn translate_type(&self, ty: &TypeExpr) -> String {
match ty { match ty {
TypeExpr::Identifier(name) => match name.as_str() { TypeExpr::Identifier(name) => match name.as_str() {
"int" => "int".to_string(), "int" => "i32".into(),
"float" | "float64" => "float64".to_string(), "float" | "float64" => "f64".into(),
"float32" => "float32".to_string(), "float32" => "f32".into(),
"bool" => "bool".to_string(), "bool" => "bool".into(),
"string" => "string".to_string(), "string" => "String".into(),
"byte" => "byte".to_string(),
"rune" => "rune".to_string(),
_ => name.clone(), _ => name.clone(),
}, },
} }
@@ -131,112 +119,92 @@ impl GoCodegen {
Statement::Expression(expr) => { Statement::Expression(expr) => {
self.add_indentedln(&format!("{};", self.generate_expression(expr))); self.add_indentedln(&format!("{};", self.generate_expression(expr)));
} }
Statement::Block(block) => { Statement::Block(block) => {
self.add_indentedln("{"); self.add_indentedln("{");
self.indent += 1; self.indent += 1;
self.generate_block(block); self.generate_block(block);
self.indent -= 1; self.indent -= 1;
self.add_indentedln("}\n"); self.add_indentedln("}");
} }
Statement::VarDecl { Statement::VarDecl {
kind, mutable,
name, name,
init, init,
type_, type_,
} => { } => {
let go_kind = match kind { let mutability = if *mutable { "mut " } else { "" };
VarKind::Let | VarKind::Const => "var",
VarKind::Var => "var", let ty = type_
}; .as_ref()
let init_expr = init .map(|t| format!(": {}", self.translate_type(t)))
.unwrap_or_default();
let init = init
.as_ref() .as_ref()
.map(|e| format!(" = {}", self.generate_expression(e))) .map(|e| format!(" = {}", self.generate_expression(e)))
.unwrap_or_else(|| "".to_string());
let type_expr = type_
.clone()
.map(|t| self.translate_type(&t))
.unwrap_or_default(); .unwrap_or_default();
self.add_indentedln(&format!(
"{} {} {}{};\n", self.add_indentedln(&format!("let {}{}{}{};", mutability, name, ty, init));
go_kind, name, type_expr, init_expr
));
} }
Statement::VarAssign { target, value } => { Statement::VarAssign { target, value } => {
self.add_indentedln(&format!( self.add_indentedln(&format!(
"{} = {};\n", "{} = {};",
self.generate_expression(target), self.generate_expression(target),
self.generate_expression(value) self.generate_expression(value)
)); ));
} }
Statement::If { Statement::If {
condition, condition,
then_branch, then_branch,
else_branch, else_branch,
} => { } => {
self.add_indented(&format!("if {} ", self.generate_expression(condition))); self.add_indentedln(&format!("if {} {{", self.generate_expression(condition)));
self.indent += 1;
self.generate_statement(then_branch); self.generate_statement(then_branch);
self.indent -= 1;
self.add_indentedln("}");
if let Some(else_br) = else_branch { if let Some(else_br) = else_branch {
self.add_indented("else "); self.add_indentedln("else {");
self.indent += 1;
self.generate_statement(else_br); self.generate_statement(else_br);
self.indent -= 1;
self.add_indentedln("}");
} }
} }
Statement::While { condition, body } => { Statement::While { condition, body } => {
self.add_indented(&format!("for {} ", self.generate_expression(condition))); self.add_indentedln(&format!("while {} {{", self.generate_expression(condition)));
self.indent += 1;
self.generate_statement(body); self.generate_statement(body);
self.indent -= 1;
self.add_indentedln("}");
} }
Statement::For {
init,
condition,
update,
body,
} => {
let (kind, init_name, init_val) = init;
let init_expr = init_val
.as_ref()
.map(|e| format!(" = {}", self.generate_expression(e)))
.unwrap_or_else(|| "".to_string());
let init_str = format!("{} {}{}", self.var_kind_to_go(kind), init_name, init_expr);
let cond_str = condition Statement::For { .. } => {
.as_ref() // Rust doesn't support C-style for loops
.map(|e| self.generate_expression(e)) self.add_indentedln("// TODO: transform into iterator-based loop");
.unwrap_or_else(|| "true".to_string());
let update_str = update
.as_ref()
.map(|s| self.generate_expression(&self.stmt_to_expr(s)))
.unwrap_or_else(|| "".to_string());
self.add_indented(&format!("for {}; {}; {} ", init_str, cond_str, update_str));
self.generate_statement(body);
} }
Statement::Return(expr) => { Statement::Return(expr) => {
let ret_val = expr let val = expr
.as_ref() .as_ref()
.map(|e| self.generate_expression(e)) .map(|e| self.generate_expression(e))
.unwrap_or_else(|| "".to_string()); .unwrap_or_default();
self.add_indentedln(&format!("return {};\n", ret_val));
}
Statement::Break => {
self.add_indentedln("break;\n");
}
Statement::Continue => {
self.add_indentedln("continue;\n");
}
}
}
fn stmt_to_expr(&self, stmt: &Statement) -> Expression { self.add_indentedln(&format!("return {};", val));
match stmt { }
Statement::Expression(e) => e.clone(),
_ => Expression::Identifier(String::new()),
}
}
fn var_kind_to_go(&self, kind: &VarKind) -> String { Statement::Break => self.add_indentedln("break;"),
match kind { Statement::Continue => self.add_indentedln("continue;"),
VarKind::Let | VarKind::Const => "var".to_string(),
VarKind::Var => "var".to_string(),
} }
} }
@@ -246,7 +214,8 @@ impl GoCodegen {
Expression::IntLiteral(n) => n.to_string(), Expression::IntLiteral(n) => n.to_string(),
Expression::FloatLiteral(n) => n.to_string(), Expression::FloatLiteral(n) => n.to_string(),
Expression::BoolLiteral(b) => b.to_string(), Expression::BoolLiteral(b) => b.to_string(),
Expression::StringLiteral(s) => format!("\"{}\"", s), Expression::StringLiteral(s) => format!("\"{}\".to_string()", s),
Expression::Postfix { initial, postfixes } => { Expression::Postfix { initial, postfixes } => {
let base = self.generate_expression(initial); let base = self.generate_expression(initial);
self.apply_postfixes(&base, postfixes) self.apply_postfixes(&base, postfixes)
@@ -256,28 +225,35 @@ impl GoCodegen {
fn apply_postfixes(&self, base: &str, postfixes: &[Postfix]) -> String { fn apply_postfixes(&self, base: &str, postfixes: &[Postfix]) -> String {
let mut result = base.to_string(); let mut result = base.to_string();
for postfix in postfixes { for postfix in postfixes {
result = match postfix { result = match postfix {
Postfix::FieldAccess(field) => format!("{}.{}", result, field), Postfix::FieldAccess(field) => format!("{}.{}", result, field),
Postfix::Call(args) => { Postfix::Call(args) => {
let args_str = args let args = args
.iter() .iter()
.map(|a| self.generate_expression(a)) .map(|a| self.generate_expression(a))
.collect::<Vec<_>>() .collect::<Vec<_>>()
.join(", "); .join(", ");
format!("{}({})", result, args_str) format!("{}({})", result, args)
} }
Postfix::StructCall(fields) => { Postfix::StructCall(fields) => {
let args_str = fields let fields = fields
.iter() .iter()
.map(|(a, b)| format!("{a}: {}", self.generate_expression(b))) .map(|(k, v)| format!("{}: {}", k, self.generate_expression(v)))
.collect::<Vec<_>>() .collect::<Vec<_>>()
.join(", "); .join(", ");
format!("{}{{{}}}", result, args_str) format!("{} {{ {} }}", result, fields)
} }
Postfix::Index(idx) => format!("{}[{}]", result, self.generate_expression(idx)),
Postfix::Index(idx) => {
format!("{}[{}]", result, self.generate_expression(idx))
}
Postfix::Binary(op, rhs) => { Postfix::Binary(op, rhs) => {
let op_str = match op { let op = match op {
BinaryOp::Plus => "+", BinaryOp::Plus => "+",
BinaryOp::Minus => "-", BinaryOp::Minus => "-",
BinaryOp::Multiply => "*", BinaryOp::Multiply => "*",
@@ -290,86 +266,18 @@ impl GoCodegen {
BinaryOp::LessThanOrEqual => "<=", BinaryOp::LessThanOrEqual => "<=",
BinaryOp::GreaterThanOrEqual => ">=", BinaryOp::GreaterThanOrEqual => ">=",
}; };
format!("{} {} {}", result, op_str, self.generate_expression(rhs))
format!("{} {} {}", result, op, self.generate_expression(rhs))
} }
}; };
} }
result result
} }
} }
impl Default for GoCodegen { impl Default for RustCodegen {
fn default() -> Self { fn default() -> Self {
Self::new() Self::new()
} }
} }
#[cfg(test)]
mod tests {
use super::*;
use parser::ast::{Block, Expression, ParamList, Statement, TopLevel, TypeExpr};
use std::collections::HashMap;
#[test]
fn test_int_literal() {
let cg = GoCodegen::new();
let expr = Expression::IntLiteral(42);
let result = cg.generate_expression(&expr);
assert_eq!(result, "42");
}
#[test]
fn test_string_literal() {
let cg = GoCodegen::new();
let expr = Expression::StringLiteral("hello".to_string());
let result = cg.generate_expression(&expr);
assert_eq!(result, "\"hello\"");
}
#[test]
fn test_function_decl() {
let mut cg = GoCodegen::new();
let toplevel = TopLevel::FunctionDecl {
export: true,
name: "main".to_string(),
params: ParamList(HashMap::new()),
return_type: None,
body: Block(vec![Statement::Return(None)]),
};
cg.generate_toplevel(&toplevel);
let output = cg.output.clone();
assert!(output.contains("func main()"));
}
#[test]
fn test_struct_decl() {
let mut cg = GoCodegen::new();
let mut fields = HashMap::new();
fields.insert(
"x".to_string(),
(true, TypeExpr::Identifier("int".to_string())),
);
let toplevel = TopLevel::StructDecl {
export: true,
name: "Point".to_string(),
fields: ParamList(fields),
};
cg.generate_toplevel(&toplevel);
let output = cg.output.clone();
assert!(output.contains("type Point struct"));
assert!(output.contains("x int"));
}
#[test]
fn test_if_statement() {
let mut cg = GoCodegen::new();
let stmt = Statement::If {
condition: Expression::Identifier("x".to_string()),
then_branch: Box::new(Statement::Return(Some(Expression::IntLiteral(1)))),
else_branch: None,
};
cg.generate_statement(&stmt);
let output = cg.output.clone();
assert!(output.contains("if x"));
}
}
+4 -13
View File
@@ -30,14 +30,6 @@ pub fn build() {
let file_name = entry.file_name(); let file_name = entry.file_name();
let file_name = file_name.to_string_lossy(); let file_name = file_name.to_string_lossy();
let script = if file_name.ends_with(".ms") {
true
} else if file_name.ends_with(".mist") {
false
} else {
continue;
};
// 3. read entry file // 3. read entry file
let source = match fs::read_to_string(&entry_path) { let source = match fs::read_to_string(&entry_path) {
Ok(s) => s, Ok(s) => s,
@@ -49,7 +41,7 @@ pub fn build() {
let parser_result = parser::parse(&source).map_err(|e| e.to_string()); let parser_result = parser::parse(&source).map_err(|e| e.to_string());
let mut ast = match parser_result { let ast = match parser_result {
Ok(ast) => ast, Ok(ast) => ast,
Err(e) => { Err(e) => {
eprintln!("error: parse failed\n{}", e); eprintln!("error: parse failed\n{}", e);
@@ -57,9 +49,9 @@ pub fn build() {
} }
}; };
semantic::walk_ast(semantic::scope::Scope::from_top(&root, &ast), &mut ast); // semantic::walk_ast(semantic::scope::Scope::from_top(&root, &ast), &mut ast);
let mut gc = crate::codegen::GoCodegen::new(); let mut gc = crate::codegen::RustCodegen::new();
let output = gc.generate(&ast); let output = gc.generate(&ast);
if let Err(e) = fs::create_dir_all(&out_dir) { if let Err(e) = fs::create_dir_all(&out_dir) {
@@ -67,8 +59,7 @@ pub fn build() {
process::exit(1); process::exit(1);
} }
let out_file = let out_file = out_dir.join(file_name.replace(".mist", ".rs"));
out_dir.join(file_name.replace(if script { ".ms" } else { ".mist" }, ".go"));
if let Err(e) = fs::write(&out_file, output) { if let Err(e) = fs::write(&out_file, output) {
eprintln!("error: failed to write output\n {}", e); eprintln!("error: failed to write output\n {}", e);
-2
View File
@@ -1,5 +1,3 @@
package main;
public struct Idk { public struct Idk {
int a, int a,
int b, int b,