Move ParseTree to ast::Ast

This commit is contained in:
mars 2022-03-25 16:16:02 -06:00
parent 3172d7d460
commit 62109e84ad
3 changed files with 769 additions and 766 deletions

759
src/parse/ast.rs Normal file
View File

@ -0,0 +1,759 @@
// Copyright (c) 2022 Marceline Cramer
// SPDX-License-Identifier: GPL-3.0-or-later
use super::lexer::{Lexer, Token};
#[derive(Debug)]
pub struct Ast<'a> {
pub defs: Vec<Definition<'a>>,
}
impl<'a> Ast<'a> {
pub fn build(lexer: &mut Lexer<'a>) -> Self {
let mut defs = Vec::new();
let mut associated_struct = None;
while let Some(tok) = lexer.next() {
if let Some(struct_name) = associated_struct {
let mut mutable = false;
if tok == Token::Mut {
mutable = true;
lexer.eat_expect(Token::Function);
} else if tok != Token::Function {
lexer.panic_message("Expected fn or mut");
}
defs.push(Definition::build_function(
mutable,
Some(struct_name),
lexer,
));
associated_struct = None;
} else {
match tok {
Token::Struct => defs.push(Definition::build_structure(lexer)),
Token::Function => defs.push(Definition::build_function(false, None, lexer)),
Token::Interface => defs.push(Definition::build_interface(lexer)),
Token::Impl => defs.push(Definition::build_impl(lexer)),
Token::Identifier => associated_struct = Some(lexer.slice()),
_ => lexer.panic_message("Expected function, struct, impl block, or interface"),
}
}
}
Self { defs }
}
}
#[derive(Debug)]
pub enum Definition<'a> {
Struct {
name: &'a str,
members: Vec<StructMember<'a>>,
},
Function {
associated_struct: Option<&'a str>,
implementation: FnImpl<'a>,
},
Interface {
name: &'a str,
functions: Vec<FnDef<'a>>,
},
Impl {
structure: &'a str,
interface: Option<&'a str>,
functions: Vec<FnImpl<'a>>,
},
}
impl<'a> Definition<'a> {
pub fn build_structure(lexer: &mut Lexer<'a>) -> Self {
let name = lexer.eat_expect_id();
lexer.eat_expect(Token::BraceOpen);
let mut members = Vec::new();
loop {
match lexer.eat_id() {
Ok(member_type) => {
let member_name = lexer.eat_expect_id();
members.push(StructMember {
type_name: member_type,
name: member_name,
});
}
Err(Some(Token::Comma)) => {}
Err(Some(Token::BraceClose)) => break,
_ => lexer.panic_message("Expected comma or closing brace"),
}
}
Self::Struct { name, members }
}
pub fn build_function(
mutable: bool,
associated_struct: Option<&'a str>,
lexer: &mut Lexer<'a>,
) -> Self {
let implementation = FnImpl::build(mutable, lexer);
Self::Function {
associated_struct,
implementation,
}
}
pub fn build_interface(lexer: &mut Lexer<'a>) -> Self {
let name = lexer.eat_expect_id();
lexer.eat_expect(Token::BraceOpen);
let mut functions = Vec::new();
loop {
match lexer.next().unwrap() {
Token::Function => functions.push(FnDef::eat_with_semicolon(false, lexer)),
Token::Mut => {
lexer.eat_expect(Token::Function);
functions.push(FnDef::eat_with_semicolon(true, lexer));
}
Token::BraceClose => break,
_ => lexer.panic_message("Expected function definition or closing bracket"),
}
}
Self::Interface { name, functions }
}
pub fn build_impl(lexer: &mut Lexer<'a>) -> Self {
let structure = lexer.eat_expect_id();
let interface = match lexer.next().unwrap() {
Token::Identifier => {
let interface = lexer.slice();
lexer.eat_expect(Token::BraceOpen);
Some(interface)
}
Token::BraceOpen => None,
_ => lexer.panic_message("Expected interface name or opening brace"),
};
let mut functions = Vec::new();
loop {
match lexer.next().unwrap() {
Token::BraceClose => break,
Token::Function => functions.push(FnImpl::build(false, lexer)),
Token::Mut => {
lexer.eat_expect(Token::Function);
functions.push(FnImpl::build(true, lexer));
}
_ => lexer.panic_message("Expected function implementation or closing brace"),
}
}
Self::Impl {
structure,
interface,
functions,
}
}
}
#[derive(Debug)]
pub struct FnImpl<'a> {
pub def: FnDef<'a>,
pub body: BranchBody<'a>,
}
impl<'a> FnImpl<'a> {
pub fn build(mutable: bool, lexer: &mut Lexer<'a>) -> Self {
let def = FnDef::eat_with_brace(mutable, lexer);
let body = BranchBody::build(lexer);
Self { def, body }
}
}
#[derive(Debug)]
pub struct FnDef<'a> {
pub mutable: bool,
pub name: &'a str,
pub signature: FnSig<'a>,
}
impl<'a> FnDef<'a> {
pub fn build(mutable: bool, lexer: &mut Lexer<'a>) -> (Self, Token) {
let name = lexer.eat_expect_id();
let (signature, tail) = FnSig::build(lexer);
(
Self {
mutable,
name,
signature,
},
tail,
)
}
pub fn eat_with_semicolon(mutable: bool, lexer: &mut Lexer<'a>) -> Self {
let (fn_def, tail) = Self::build(mutable, lexer);
if tail != Token::Semicolon {
lexer.panic_message("Expected semicolon");
}
fn_def
}
pub fn eat_with_brace(mutable: bool, lexer: &mut Lexer<'a>) -> Self {
let (fn_def, tail) = Self::build(mutable, lexer);
if tail != Token::BraceOpen {
lexer.panic_message("Expected opening brace");
}
fn_def
}
}
#[derive(Debug)]
pub struct StructMember<'a> {
pub type_name: &'a str,
pub name: &'a str,
}
#[derive(Debug)]
pub struct FnSig<'a> {
pub args: Vec<FnArg<'a>>,
pub return_type: Option<&'a str>,
}
impl<'a> FnSig<'a> {
pub fn build(lexer: &mut Lexer<'a>) -> (Self, Token) {
lexer.eat_expect(Token::ParenOpen);
let mut args = Vec::new();
loop {
match lexer.eat_id() {
Ok(type_name) => match lexer.next().unwrap() {
Token::Identifier => args.push(FnArg {
type_name,
name: Some(lexer.slice()),
}),
Token::Comma => args.push(FnArg {
type_name,
name: None,
}),
Token::ParenClose => {
args.push(FnArg {
type_name,
name: None,
});
break;
}
_ => lexer.panic_message("Unexpected token"),
},
Err(Some(Token::Comma)) => {}
Err(Some(Token::ParenClose)) => break,
_ => lexer.panic_message("Expected comma, type, or closing parentheses"),
}
}
let mut next = lexer.next().unwrap();
let mut return_type = None;
if next == Token::Identifier {
return_type = Some(lexer.slice());
next = lexer.next().unwrap();
}
(Self { args, return_type }, next)
}
}
#[derive(Debug)]
pub struct FnArg<'a> {
pub type_name: &'a str,
pub name: Option<&'a str>,
}
#[derive(Debug)]
pub struct BranchBody<'a> {
pub statements: Vec<Statement<'a>>,
pub tail_expr: Option<Box<Expr<'a>>>,
}
impl<'a> BranchBody<'a> {
pub fn build(lexer: &mut Lexer<'a>) -> Self {
let mut statements = Vec::new();
let mut next = None;
let tail_expr = loop {
let tok = if let Some(tok) = next {
next = None;
tok
} else {
lexer.next().unwrap()
};
match tok {
Token::Let => {
let (statement, next) = Statement::build_let(lexer);
statements.push(statement);
match next {
Token::Semicolon => {}
Token::BraceClose => break None,
_ => lexer.panic_message("Unexpected token"),
}
}
Token::If => {
let (stmt, tail) = IfStmt::build(lexer);
if tail == Token::BraceClose {
if let Some(_) = stmt.else_body {
let expr = stmt.to_expr();
break Some(Expr::If(expr));
} else {
statements.push(Statement::If(stmt));
break None;
}
} else {
statements.push(Statement::If(stmt));
next = Some(tail);
}
}
Token::While => {
let test_expr = match Expr::build(lexer) {
(Some(test_expr), Token::BraceOpen) => test_expr,
(Some(_), _) => lexer.panic_message("Expected opening brace"),
_ => lexer.panic_message("Expected test expression"),
};
let loop_body = BranchBody::build(lexer);
statements.push(Statement::While {
test_expr,
loop_body,
});
}
Token::BraceClose => break None,
_ => match Expr::build_start(tok, lexer) {
(None, Token::Semicolon | Token::BraceClose) => {}
(Some(expr), Token::Semicolon) => statements.push(Statement::Expr(expr)),
(Some(expr), Token::BraceClose) => break Some(expr),
_ => lexer.panic_message("Unexpected token"),
},
}
};
Self {
statements,
tail_expr: tail_expr.map(|x| Box::new(x)),
}
}
}
type ExprPair<'a> = Box<(Expr<'a>, Expr<'a>)>;
#[derive(Debug)]
pub enum Expr<'a> {
BinaryOp(BinaryOp, ExprPair<'a>),
UnaryOp(UnaryOp, Box<Expr<'a>>),
Literal(Literal<'a>),
Local(&'a str),
Member(&'a str, Box<Expr<'a>>),
SelfMember(&'a str),
Group(Box<Expr<'a>>),
FnCall(Box<Expr<'a>>, Vec<Expr<'a>>),
If(IfExpr<'a>),
}
impl<'a> Expr<'a> {
pub fn build_start(tok: Token, lexer: &mut Lexer<'a>) -> (Option<Self>, Token) {
let (unary_op, tok) = if let Some(op) = UnaryOp::from_token(tok) {
(Some(op), lexer.next().unwrap())
} else {
(None, tok)
};
let lhs = match tok {
Token::Identifier => Self::Local(lexer.slice()),
Token::Dot => Self::SelfMember(lexer.eat_expect_id()),
Token::ParenOpen => {
let (inner, next) = Self::build(lexer);
if next != Token::ParenClose {
lexer.panic_message("Expected closing parentheses");
}
Self::Group(Box::new(inner.unwrap()))
}
Token::If => Self::If(IfExpr::build(lexer)),
Token::BinaryInteger => Self::Literal(Literal::BinaryInteger(lexer.slice())),
Token::OctalInteger => Self::Literal(Literal::OctalInteger(lexer.slice())),
Token::HexInteger => Self::Literal(Literal::HexInteger(lexer.slice())),
Token::DecimalInteger => Self::Literal(Literal::DecimalInteger(lexer.slice())),
Token::DecimalFloat => Self::Literal(Literal::DecimalFloat(lexer.slice())),
Token::StringLiteral => Self::Literal(Literal::String(lexer.slice())),
Token::True => Self::Literal(Literal::Boolean(true)),
Token::False => Self::Literal(Literal::Boolean(false)),
Token::Semicolon | Token::BraceClose | Token::ParenClose => return (None, tok),
_ => lexer.panic_message("Unexpected token"),
};
let mut lhs = if let Some(op) = unary_op {
Self::UnaryOp(op, Box::new(lhs))
} else {
lhs
};
loop {
let tok = lexer.next().unwrap();
match tok {
Token::Dot => {
let member = lexer.eat_expect_id();
let expr = Self::Member(member, Box::new(lhs));
lhs = expr;
}
Token::ParenOpen => lhs = Self::FnCall(Box::new(lhs), Self::eat_args(lexer)),
_ => {
if let Some(op) = BinaryOp::from_token(tok) {
match Self::build(lexer) {
(Some(rhs), tail) => {
break (Some(Self::BinaryOp(op, Box::new((lhs, rhs)))), tail)
}
_ => lexer.panic_message("Expected right-hand expression"),
}
} else {
break (Some(lhs), tok);
}
}
}
}
}
pub fn build(lexer: &mut Lexer<'a>) -> (Option<Self>, Token) {
Self::build_start(lexer.next().unwrap(), lexer)
}
pub fn eat_args(lexer: &mut Lexer<'a>) -> Vec<Self> {
let mut args = Vec::new();
loop {
match Self::build(lexer) {
(Some(arg), Token::Comma) => args.push(arg),
(Some(arg), Token::ParenClose) => {
args.push(arg);
break;
}
(None, Token::ParenClose) => break,
_ => lexer.panic_message("Unexpected token"),
}
}
args
}
}
#[derive(Debug)]
pub enum BinaryOp {
Add,
Sub,
Mul,
Div,
Assign,
BoolOr,
BoolAnd,
BitOr,
BitAnd,
BitXor,
Less,
LessEq,
Greater,
GreaterEq,
Eq,
Neq,
}
impl BinaryOp {
pub fn from_token(tok: Token) -> Option<Self> {
use BinaryOp::*;
use Token::*;
match tok {
OpAdd => Some(Add),
OpSub => Some(Sub),
OpMul => Some(Mul),
OpDiv => Some(Div),
OpAssign => Some(Assign),
OpBoolOr => Some(BoolOr),
OpBoolAnd => Some(BoolAnd),
OpBitOr => Some(BitOr),
OpBitAnd => Some(BitAnd),
OpBitXor => Some(BitXor),
OpLess => Some(Less),
OpLessEq => Some(LessEq),
OpGreater => Some(Greater),
OpGreaterEq => Some(GreaterEq),
OpEq => Some(Eq),
OpNeq => Some(Neq),
_ => None,
}
}
}
#[derive(Debug)]
pub enum UnaryOp {
BoolNot,
BitNot,
Negate,
}
impl UnaryOp {
pub fn from_token(tok: Token) -> Option<Self> {
use Token::*;
use UnaryOp::*;
match tok {
OpBoolNot => Some(BoolNot),
OpBitNot => Some(BitNot),
OpSub => Some(Negate),
_ => None,
}
}
}
#[derive(Debug)]
pub enum Literal<'a> {
BinaryInteger(&'a str),
OctalInteger(&'a str),
HexInteger(&'a str),
DecimalInteger(&'a str),
DecimalFloat(&'a str),
String(&'a str),
Boolean(bool),
}
#[derive(Debug)]
pub enum Statement<'a> {
Expr(Expr<'a>),
If(IfStmt<'a>),
While {
test_expr: Expr<'a>,
loop_body: BranchBody<'a>,
},
Let {
var: &'a str,
mutable: bool,
expr: Expr<'a>,
},
}
impl<'a> Statement<'a> {
pub fn eat_assign(lexer: &mut Lexer<'a>) -> (Expr<'a>, Token) {
lexer.eat_expect(Token::OpAssign);
let (expr, tail) = Expr::build(lexer);
(expr.unwrap(), tail)
}
pub fn build_let(lexer: &mut Lexer<'a>) -> (Self, Token) {
let mut mutable = false;
let var = match lexer.next().unwrap() {
Token::Identifier => lexer.slice(),
Token::Mut => {
mutable = true;
lexer.eat_expect_id()
}
_ => lexer.panic_message("Unexpected token"),
};
let (expr, tail) = Self::eat_assign(lexer);
(Self::Let { var, mutable, expr }, tail)
}
}
#[derive(Debug)]
pub struct IfExpr<'a> {
pub test_expr: Box<Expr<'a>>,
pub then_body: BranchBody<'a>,
pub else_body: BranchBody<'a>,
}
impl<'a> IfExpr<'a> {
pub fn build(lexer: &mut Lexer<'a>) -> Self {
let test_expr = Box::new(Self::eat_test_expr(lexer));
let then_body = BranchBody::build(lexer);
lexer.eat_expect(Token::Else);
let else_body = match lexer.next().unwrap() {
Token::BraceOpen => BranchBody::build(lexer),
Token::If => BranchBody {
statements: Vec::new(),
tail_expr: Some(Box::new(Expr::If(IfExpr::build(lexer)))),
},
_ => lexer.panic_message("Expected if or opening brace"),
};
Self {
test_expr,
then_body,
else_body,
}
}
pub fn eat_test_expr(lexer: &mut Lexer<'a>) -> Expr<'a> {
match Expr::build(lexer) {
(Some(test_expr), Token::BraceOpen) => test_expr,
(Some(_), _) => lexer.panic_message("Expected opening brace"),
_ => lexer.panic_message("Expected test expression"),
}
}
}
#[derive(Debug)]
pub struct IfStmt<'a> {
pub test_expr: Expr<'a>,
pub then_body: BranchBody<'a>,
pub else_body: Option<BranchBody<'a>>,
}
impl<'a> IfStmt<'a> {
pub fn build(lexer: &mut Lexer<'a>) -> (Self, Token) {
let test_expr = IfExpr::eat_test_expr(lexer);
let then_body = BranchBody::build(lexer);
let (else_body, tail) = match lexer.next().unwrap() {
Token::Else => {
lexer.eat_expect(Token::BraceOpen);
(Some(BranchBody::build(lexer)), lexer.next().unwrap())
}
next => (None, next),
};
(
Self {
test_expr,
then_body,
else_body,
},
tail,
)
}
pub fn to_expr(self) -> IfExpr<'a> {
let Self {
test_expr,
then_body,
else_body,
} = self;
IfExpr {
test_expr: Box::new(test_expr),
then_body,
else_body: else_body.unwrap(),
}
}
}
#[cfg(test)]
mod tests {
use super::*;
fn parse(source: &str) {
let mut lex = Lexer::new(source);
let parse_tree = Ast::build(&mut lex);
println!("{:#?}", parse_tree);
}
#[test]
fn function() {
parse(include_str!("../test/function.fae"));
}
#[test]
fn tail_expression() {
parse("fn add(i32 x, i32 y) i32 { x + y }");
}
#[test]
fn structure() {
parse(include_str!("../test/structure.fae"));
}
#[test]
fn interface() {
parse(include_str!("../test/interface.fae"));
}
#[test]
fn example() {
parse(include_str!("../test/example.fae"));
}
#[test]
fn clock() {
parse(include_str!("../test/clock.fae"));
}
mod expr {
use super::*;
fn parse_expr(source: &str) {
println!("Source: {}", source);
let mut lex = Lexer::new(source);
let expr = Expr::build(&mut lex);
println!("{:#?}", expr);
assert_eq!(expr.1, Token::Semicolon);
}
#[test]
fn int_literals() {
parse_expr("0b1 + 0x2 - 0o3 * 0x4;");
}
#[test]
fn float_literals() {
parse_expr("0.1 + .1 * 1.;");
}
#[test]
fn string_literals() {
parse_expr("\"String contents\";");
}
#[test]
fn locals() {
parse_expr("local1 + local2 - local3;");
}
#[test]
fn nesting() {
parse_expr("(1 + 2) * (3 / (4 + 5));")
}
#[test]
fn unary_ops() {
parse_expr("not false + -(2 + 3) * -1 / ~0xff;");
}
#[test]
fn members() {
parse_expr(".member1 * .member2;");
}
#[test]
fn if_expr() {
parse_expr("if true { 1 } else { 0 };");
}
#[test]
fn else_if_expr() {
parse_expr("if a { 1 } else if b { 2 } else { c };");
}
#[test]
fn func_call() {
parse_expr("func_call(1 + 2, 3, 4 * 5);");
}
#[test]
fn method() {
parse_expr(".method();");
}
#[test]
fn submembers() {
parse_expr(".member.submember1 / .member.submember2.submember3;");
}
}
}

View File

@ -1,762 +1,6 @@
// Copyright (c) 2022 Marceline Cramer
// SPDX-License-Identifier: GPL-3.0-or-later
pub mod ast;
pub mod lexer;
pub mod pest;
use lexer::{Lexer, Token};
#[derive(Debug)]
pub struct ParseTree<'a> {
pub defs: Vec<Definition<'a>>,
}
impl<'a> ParseTree<'a> {
pub fn build(lexer: &mut Lexer<'a>) -> Self {
let mut defs = Vec::new();
let mut associated_struct = None;
while let Some(tok) = lexer.next() {
if let Some(struct_name) = associated_struct {
let mut mutable = false;
if tok == Token::Mut {
mutable = true;
lexer.eat_expect(Token::Function);
} else if tok != Token::Function {
lexer.panic_message("Expected fn or mut");
}
defs.push(Definition::build_function(
mutable,
Some(struct_name),
lexer,
));
associated_struct = None;
} else {
match tok {
Token::Struct => defs.push(Definition::build_structure(lexer)),
Token::Function => defs.push(Definition::build_function(false, None, lexer)),
Token::Interface => defs.push(Definition::build_interface(lexer)),
Token::Impl => defs.push(Definition::build_impl(lexer)),
Token::Identifier => associated_struct = Some(lexer.slice()),
_ => lexer.panic_message("Expected function, struct, impl block, or interface"),
}
}
}
Self { defs }
}
}
#[derive(Debug)]
pub enum Definition<'a> {
Struct {
name: &'a str,
members: Vec<StructMember<'a>>,
},
Function {
associated_struct: Option<&'a str>,
implementation: FnImpl<'a>,
},
Interface {
name: &'a str,
functions: Vec<FnDef<'a>>,
},
Impl {
structure: &'a str,
interface: Option<&'a str>,
functions: Vec<FnImpl<'a>>,
},
}
impl<'a> Definition<'a> {
pub fn build_structure(lexer: &mut Lexer<'a>) -> Self {
let name = lexer.eat_expect_id();
lexer.eat_expect(Token::BraceOpen);
let mut members = Vec::new();
loop {
match lexer.eat_id() {
Ok(member_type) => {
let member_name = lexer.eat_expect_id();
members.push(StructMember {
type_name: member_type,
name: member_name,
});
}
Err(Some(Token::Comma)) => {}
Err(Some(Token::BraceClose)) => break,
_ => lexer.panic_message("Expected comma or closing brace"),
}
}
Self::Struct { name, members }
}
pub fn build_function(
mutable: bool,
associated_struct: Option<&'a str>,
lexer: &mut Lexer<'a>,
) -> Self {
let implementation = FnImpl::build(mutable, lexer);
Self::Function {
associated_struct,
implementation,
}
}
pub fn build_interface(lexer: &mut Lexer<'a>) -> Self {
let name = lexer.eat_expect_id();
lexer.eat_expect(Token::BraceOpen);
let mut functions = Vec::new();
loop {
match lexer.next().unwrap() {
Token::Function => functions.push(FnDef::eat_with_semicolon(false, lexer)),
Token::Mut => {
lexer.eat_expect(Token::Function);
functions.push(FnDef::eat_with_semicolon(true, lexer));
}
Token::BraceClose => break,
_ => lexer.panic_message("Expected function definition or closing bracket"),
}
}
Self::Interface { name, functions }
}
pub fn build_impl(lexer: &mut Lexer<'a>) -> Self {
let structure = lexer.eat_expect_id();
let interface = match lexer.next().unwrap() {
Token::Identifier => {
let interface = lexer.slice();
lexer.eat_expect(Token::BraceOpen);
Some(interface)
}
Token::BraceOpen => None,
_ => lexer.panic_message("Expected interface name or opening brace"),
};
let mut functions = Vec::new();
loop {
match lexer.next().unwrap() {
Token::BraceClose => break,
Token::Function => functions.push(FnImpl::build(false, lexer)),
Token::Mut => {
lexer.eat_expect(Token::Function);
functions.push(FnImpl::build(true, lexer));
}
_ => lexer.panic_message("Expected function implementation or closing brace"),
}
}
Self::Impl {
structure,
interface,
functions,
}
}
}
#[derive(Debug)]
pub struct FnImpl<'a> {
pub def: FnDef<'a>,
pub body: BranchBody<'a>,
}
impl<'a> FnImpl<'a> {
pub fn build(mutable: bool, lexer: &mut Lexer<'a>) -> Self {
let def = FnDef::eat_with_brace(mutable, lexer);
let body = BranchBody::build(lexer);
Self { def, body }
}
}
#[derive(Debug)]
pub struct FnDef<'a> {
pub mutable: bool,
pub name: &'a str,
pub signature: FnSig<'a>,
}
impl<'a> FnDef<'a> {
pub fn build(mutable: bool, lexer: &mut Lexer<'a>) -> (Self, Token) {
let name = lexer.eat_expect_id();
let (signature, tail) = FnSig::build(lexer);
(
Self {
mutable,
name,
signature,
},
tail,
)
}
pub fn eat_with_semicolon(mutable: bool, lexer: &mut Lexer<'a>) -> Self {
let (fn_def, tail) = Self::build(mutable, lexer);
if tail != Token::Semicolon {
lexer.panic_message("Expected semicolon");
}
fn_def
}
pub fn eat_with_brace(mutable: bool, lexer: &mut Lexer<'a>) -> Self {
let (fn_def, tail) = Self::build(mutable, lexer);
if tail != Token::BraceOpen {
lexer.panic_message("Expected opening brace");
}
fn_def
}
}
#[derive(Debug)]
pub struct StructMember<'a> {
pub type_name: &'a str,
pub name: &'a str,
}
#[derive(Debug)]
pub struct FnSig<'a> {
pub args: Vec<FnArg<'a>>,
pub return_type: Option<&'a str>,
}
impl<'a> FnSig<'a> {
pub fn build(lexer: &mut Lexer<'a>) -> (Self, Token) {
lexer.eat_expect(Token::ParenOpen);
let mut args = Vec::new();
loop {
match lexer.eat_id() {
Ok(type_name) => match lexer.next().unwrap() {
Token::Identifier => args.push(FnArg {
type_name,
name: Some(lexer.slice()),
}),
Token::Comma => args.push(FnArg {
type_name,
name: None,
}),
Token::ParenClose => {
args.push(FnArg {
type_name,
name: None,
});
break;
}
_ => lexer.panic_message("Unexpected token"),
},
Err(Some(Token::Comma)) => {}
Err(Some(Token::ParenClose)) => break,
_ => lexer.panic_message("Expected comma, type, or closing parentheses"),
}
}
let mut next = lexer.next().unwrap();
let mut return_type = None;
if next == Token::Identifier {
return_type = Some(lexer.slice());
next = lexer.next().unwrap();
}
(Self { args, return_type }, next)
}
}
#[derive(Debug)]
pub struct FnArg<'a> {
pub type_name: &'a str,
pub name: Option<&'a str>,
}
#[derive(Debug)]
pub struct BranchBody<'a> {
pub statements: Vec<Statement<'a>>,
pub tail_expr: Option<Box<Expr<'a>>>,
}
impl<'a> BranchBody<'a> {
pub fn build(lexer: &mut Lexer<'a>) -> Self {
let mut statements = Vec::new();
let mut next = None;
let tail_expr = loop {
let tok = if let Some(tok) = next {
next = None;
tok
} else {
lexer.next().unwrap()
};
match tok {
Token::Let => {
let (statement, next) = Statement::build_let(lexer);
statements.push(statement);
match next {
Token::Semicolon => {}
Token::BraceClose => break None,
_ => lexer.panic_message("Unexpected token"),
}
}
Token::If => {
let (stmt, tail) = IfStmt::build(lexer);
if tail == Token::BraceClose {
if let Some(_) = stmt.else_body {
let expr = stmt.to_expr();
break Some(Expr::If(expr));
} else {
statements.push(Statement::If(stmt));
break None;
}
} else {
statements.push(Statement::If(stmt));
next = Some(tail);
}
}
Token::While => {
let test_expr = match Expr::build(lexer) {
(Some(test_expr), Token::BraceOpen) => test_expr,
(Some(_), _) => lexer.panic_message("Expected opening brace"),
_ => lexer.panic_message("Expected test expression"),
};
let loop_body = BranchBody::build(lexer);
statements.push(Statement::While {
test_expr,
loop_body,
});
}
Token::BraceClose => break None,
_ => match Expr::build_start(tok, lexer) {
(None, Token::Semicolon | Token::BraceClose) => {}
(Some(expr), Token::Semicolon) => statements.push(Statement::Expr(expr)),
(Some(expr), Token::BraceClose) => break Some(expr),
_ => lexer.panic_message("Unexpected token"),
},
}
};
Self {
statements,
tail_expr: tail_expr.map(|x| Box::new(x)),
}
}
}
type ExprPair<'a> = Box<(Expr<'a>, Expr<'a>)>;
#[derive(Debug)]
pub enum Expr<'a> {
BinaryOp(BinaryOp, ExprPair<'a>),
UnaryOp(UnaryOp, Box<Expr<'a>>),
Literal(Literal<'a>),
Local(&'a str),
Member(&'a str, Box<Expr<'a>>),
SelfMember(&'a str),
Group(Box<Expr<'a>>),
FnCall(Box<Expr<'a>>, Vec<Expr<'a>>),
If(IfExpr<'a>),
}
impl<'a> Expr<'a> {
pub fn build_start(tok: Token, lexer: &mut Lexer<'a>) -> (Option<Self>, Token) {
let (unary_op, tok) = if let Some(op) = UnaryOp::from_token(tok) {
(Some(op), lexer.next().unwrap())
} else {
(None, tok)
};
let lhs = match tok {
Token::Identifier => Self::Local(lexer.slice()),
Token::Dot => Self::SelfMember(lexer.eat_expect_id()),
Token::ParenOpen => {
let (inner, next) = Self::build(lexer);
if next != Token::ParenClose {
lexer.panic_message("Expected closing parentheses");
}
Self::Group(Box::new(inner.unwrap()))
}
Token::If => Self::If(IfExpr::build(lexer)),
Token::BinaryInteger => Self::Literal(Literal::BinaryInteger(lexer.slice())),
Token::OctalInteger => Self::Literal(Literal::OctalInteger(lexer.slice())),
Token::HexInteger => Self::Literal(Literal::HexInteger(lexer.slice())),
Token::DecimalInteger => Self::Literal(Literal::DecimalInteger(lexer.slice())),
Token::DecimalFloat => Self::Literal(Literal::DecimalFloat(lexer.slice())),
Token::StringLiteral => Self::Literal(Literal::String(lexer.slice())),
Token::True => Self::Literal(Literal::Boolean(true)),
Token::False => Self::Literal(Literal::Boolean(false)),
Token::Semicolon | Token::BraceClose | Token::ParenClose => return (None, tok),
_ => lexer.panic_message("Unexpected token"),
};
let mut lhs = if let Some(op) = unary_op {
Self::UnaryOp(op, Box::new(lhs))
} else {
lhs
};
loop {
let tok = lexer.next().unwrap();
match tok {
Token::Dot => {
let member = lexer.eat_expect_id();
let expr = Self::Member(member, Box::new(lhs));
lhs = expr;
}
Token::ParenOpen => lhs = Self::FnCall(Box::new(lhs), Self::eat_args(lexer)),
_ => {
if let Some(op) = BinaryOp::from_token(tok) {
match Self::build(lexer) {
(Some(rhs), tail) => {
break (Some(Self::BinaryOp(op, Box::new((lhs, rhs)))), tail)
}
_ => lexer.panic_message("Expected right-hand expression"),
}
} else {
break (Some(lhs), tok);
}
}
}
}
}
pub fn build(lexer: &mut Lexer<'a>) -> (Option<Self>, Token) {
Self::build_start(lexer.next().unwrap(), lexer)
}
pub fn eat_args(lexer: &mut Lexer<'a>) -> Vec<Self> {
let mut args = Vec::new();
loop {
match Self::build(lexer) {
(Some(arg), Token::Comma) => args.push(arg),
(Some(arg), Token::ParenClose) => {
args.push(arg);
break;
}
(None, Token::ParenClose) => break,
_ => lexer.panic_message("Unexpected token"),
}
}
args
}
}
#[derive(Debug)]
pub enum BinaryOp {
Add,
Sub,
Mul,
Div,
Assign,
BoolOr,
BoolAnd,
BitOr,
BitAnd,
BitXor,
Less,
LessEq,
Greater,
GreaterEq,
Eq,
Neq,
}
impl BinaryOp {
pub fn from_token(tok: Token) -> Option<Self> {
use BinaryOp::*;
use Token::*;
match tok {
OpAdd => Some(Add),
OpSub => Some(Sub),
OpMul => Some(Mul),
OpDiv => Some(Div),
OpAssign => Some(Assign),
OpBoolOr => Some(BoolOr),
OpBoolAnd => Some(BoolAnd),
OpBitOr => Some(BitOr),
OpBitAnd => Some(BitAnd),
OpBitXor => Some(BitXor),
OpLess => Some(Less),
OpLessEq => Some(LessEq),
OpGreater => Some(Greater),
OpGreaterEq => Some(GreaterEq),
OpEq => Some(Eq),
OpNeq => Some(Neq),
_ => None,
}
}
}
#[derive(Debug)]
pub enum UnaryOp {
BoolNot,
BitNot,
Negate,
}
impl UnaryOp {
pub fn from_token(tok: Token) -> Option<Self> {
use Token::*;
use UnaryOp::*;
match tok {
OpBoolNot => Some(BoolNot),
OpBitNot => Some(BitNot),
OpSub => Some(Negate),
_ => None,
}
}
}
#[derive(Debug)]
pub enum Literal<'a> {
BinaryInteger(&'a str),
OctalInteger(&'a str),
HexInteger(&'a str),
DecimalInteger(&'a str),
DecimalFloat(&'a str),
String(&'a str),
Boolean(bool),
}
#[derive(Debug)]
pub enum Statement<'a> {
Expr(Expr<'a>),
If(IfStmt<'a>),
While {
test_expr: Expr<'a>,
loop_body: BranchBody<'a>,
},
Let {
var: &'a str,
mutable: bool,
expr: Expr<'a>,
},
}
impl<'a> Statement<'a> {
pub fn eat_assign(lexer: &mut Lexer<'a>) -> (Expr<'a>, Token) {
lexer.eat_expect(Token::OpAssign);
let (expr, tail) = Expr::build(lexer);
(expr.unwrap(), tail)
}
pub fn build_let(lexer: &mut Lexer<'a>) -> (Self, Token) {
let mut mutable = false;
let var = match lexer.next().unwrap() {
Token::Identifier => lexer.slice(),
Token::Mut => {
mutable = true;
lexer.eat_expect_id()
}
_ => lexer.panic_message("Unexpected token"),
};
let (expr, tail) = Self::eat_assign(lexer);
(Self::Let { var, mutable, expr }, tail)
}
}
#[derive(Debug)]
pub struct IfExpr<'a> {
pub test_expr: Box<Expr<'a>>,
pub then_body: BranchBody<'a>,
pub else_body: BranchBody<'a>,
}
impl<'a> IfExpr<'a> {
pub fn build(lexer: &mut Lexer<'a>) -> Self {
let test_expr = Box::new(Self::eat_test_expr(lexer));
let then_body = BranchBody::build(lexer);
lexer.eat_expect(Token::Else);
let else_body = match lexer.next().unwrap() {
Token::BraceOpen => BranchBody::build(lexer),
Token::If => BranchBody {
statements: Vec::new(),
tail_expr: Some(Box::new(Expr::If(IfExpr::build(lexer)))),
},
_ => lexer.panic_message("Expected if or opening brace"),
};
Self {
test_expr,
then_body,
else_body,
}
}
pub fn eat_test_expr(lexer: &mut Lexer<'a>) -> Expr<'a> {
match Expr::build(lexer) {
(Some(test_expr), Token::BraceOpen) => test_expr,
(Some(_), _) => lexer.panic_message("Expected opening brace"),
_ => lexer.panic_message("Expected test expression"),
}
}
}
#[derive(Debug)]
pub struct IfStmt<'a> {
pub test_expr: Expr<'a>,
pub then_body: BranchBody<'a>,
pub else_body: Option<BranchBody<'a>>,
}
impl<'a> IfStmt<'a> {
pub fn build(lexer: &mut Lexer<'a>) -> (Self, Token) {
let test_expr = IfExpr::eat_test_expr(lexer);
let then_body = BranchBody::build(lexer);
let (else_body, tail) = match lexer.next().unwrap() {
Token::Else => {
lexer.eat_expect(Token::BraceOpen);
(Some(BranchBody::build(lexer)), lexer.next().unwrap())
}
next => (None, next),
};
(
Self {
test_expr,
then_body,
else_body,
},
tail,
)
}
pub fn to_expr(self) -> IfExpr<'a> {
let Self {
test_expr,
then_body,
else_body,
} = self;
IfExpr {
test_expr: Box::new(test_expr),
then_body,
else_body: else_body.unwrap(),
}
}
}
#[cfg(test)]
mod tests {
use super::*;
fn parse(source: &str) {
let mut lex = Lexer::new(source);
let parse_tree = ParseTree::build(&mut lex);
println!("{:#?}", parse_tree);
}
#[test]
fn function() {
parse(include_str!("../test/function.fae"));
}
#[test]
fn tail_expression() {
parse("fn add(i32 x, i32 y) i32 { x + y }");
}
#[test]
fn structure() {
parse(include_str!("../test/structure.fae"));
}
#[test]
fn interface() {
parse(include_str!("../test/interface.fae"));
}
#[test]
fn example() {
parse(include_str!("../test/example.fae"));
}
#[test]
fn clock() {
parse(include_str!("../test/clock.fae"));
}
mod expr {
use super::*;
fn parse_expr(source: &str) {
println!("Source: {}", source);
let mut lex = Lexer::new(source);
let expr = Expr::build(&mut lex);
println!("{:#?}", expr);
assert_eq!(expr.1, Token::Semicolon);
}
#[test]
fn int_literals() {
parse_expr("0b1 + 0x2 - 0o3 * 0x4;");
}
#[test]
fn float_literals() {
parse_expr("0.1 + .1 * 1.;");
}
#[test]
fn string_literals() {
parse_expr("\"String contents\";");
}
#[test]
fn locals() {
parse_expr("local1 + local2 - local3;");
}
#[test]
fn nesting() {
parse_expr("(1 + 2) * (3 / (4 + 5));")
}
#[test]
fn unary_ops() {
parse_expr("not false + -(2 + 3) * -1 / ~0xff;");
}
#[test]
fn members() {
parse_expr(".member1 * .member2;");
}
#[test]
fn if_expr() {
parse_expr("if true { 1 } else { 0 };");
}
#[test]
fn else_if_expr() {
parse_expr("if a { 1 } else if b { 2 } else { c };");
}
#[test]
fn func_call() {
parse_expr("func_call(1 + 2, 3, 4 * 5);");
}
#[test]
fn method() {
parse_expr(".method();");
}
#[test]
fn submembers() {
parse_expr(".member.submember1 / .member.submember2.submember3;");
}
}
}

View File

@ -1,7 +1,7 @@
// Copyright (c) 2022 Marceline Cramer
// SPDX-License-Identifier: GPL-3.0-or-later
use crate::parse;
use crate::parse::ast;
use std::collections::HashMap;
type SymbolMap<T> = HashMap<String, T>;
@ -31,11 +31,11 @@ pub struct SymbolTable {
}
impl SymbolTable {
pub fn from_parse_tree<'a>(pt: &parse::ParseTree<'a>) -> Self {
pub fn from_parse_tree<'a>(pt: &ast::Ast<'a>) -> Self {
let mut structures = SymbolMap::new();
for def in pt.defs.iter() {
if let parse::Definition::Struct { name, members } = def {
if let ast::Definition::Struct { name, members } = def {
structures.insert(
name.to_string(),
StructDef {
@ -56,7 +56,7 @@ impl SymbolTable {
let mut interfaces = SymbolMap::new();
for def in pt.defs.iter() {
if let parse::Definition::Interface { name, functions } = def {
if let ast::Definition::Interface { name, functions } = def {
interfaces.insert(
name.to_string(),
IfaceDef {
@ -69,7 +69,7 @@ impl SymbolTable {
let mut functions = SymbolMap::new();
for def in pt.defs.iter() {
if let parse::Definition::Function {
if let ast::Definition::Function {
associated_struct,
implementation,
} = def
@ -121,8 +121,8 @@ pub struct FnDef {
pub return_type: Option<String>,
}
impl<'a> From<&parse::FnDef<'a>> for FnDef {
fn from(def: &parse::FnDef<'a>) -> Self {
impl<'a> From<&ast::FnDef<'a>> for FnDef {
fn from(def: &ast::FnDef<'a>) -> Self {
Self {
mutable: def.mutable,
name: def.name.to_string(),
@ -151,8 +151,8 @@ mod tests {
use super::*;
fn from_source(source: &str) -> SymbolTable {
let mut lex = parse::lexer::Lexer::new(source);
let pt = parse::ParseTree::build(&mut lex);
let mut lex = crate::parse::lexer::Lexer::new(source);
let pt = ast::Ast::build(&mut lex);
let symtab = SymbolTable::from_parse_tree(&pt);
println!("{:#?}", symtab);
symtab