lt-core/src/parser.rs

295 lines
9.7 KiB
Rust
Raw Normal View History

2024-05-09 20:13:10 +02:00
use {
2024-05-12 18:58:39 +02:00
crate::{
expr::{LTExpr, Statement, TypeError, TypeTag},
lexer::{LTIRLexer, LTIRToken, LexError},
},
2024-05-12 18:56:10 +02:00
std::{
iter::Peekable,
2024-05-12 18:58:39 +02:00
sync::{Arc, RwLock},
2024-05-12 18:56:10 +02:00
},
2024-05-09 20:13:10 +02:00
};
#[derive(Clone, Debug)]
pub enum ParseError {
LexError(LexError),
UnexpectedClose,
UnexpectedEnd,
2024-05-12 18:58:39 +02:00
UnexpectedToken,
2024-05-09 20:13:10 +02:00
}
2024-05-11 00:00:20 +02:00
pub fn parse_expect<It>(
tokens: &mut Peekable<LTIRLexer<It>>,
2024-05-12 18:58:39 +02:00
expected_token: LTIRToken,
) -> Result<(), ParseError>
where
It: Iterator<Item = char>,
2024-05-11 00:00:20 +02:00
{
match tokens.next() {
2024-05-13 22:55:24 +02:00
Some((region, Ok(t))) => {
2024-05-11 00:00:20 +02:00
if t == expected_token {
Ok(())
} else {
Err(ParseError::UnexpectedToken)
}
2024-05-12 18:58:39 +02:00
}
2024-05-13 22:55:24 +02:00
Some((region, Err(err))) => Err(ParseError::LexError(err)),
2024-05-12 18:58:39 +02:00
None => Err(ParseError::UnexpectedEnd),
2024-05-11 00:00:20 +02:00
}
}
2024-05-12 18:58:39 +02:00
pub fn parse_symbol<It>(tokens: &mut Peekable<LTIRLexer<It>>) -> Result<String, ParseError>
where
It: Iterator<Item = char>,
2024-05-11 00:00:20 +02:00
{
match tokens.next() {
2024-05-13 22:55:24 +02:00
Some((region, Ok(LTIRToken::Symbol(name)))) => Ok(name),
Some((region, Ok(_))) => Err(ParseError::UnexpectedToken),
Some((region, Err(err))) => Err(ParseError::LexError(err)),
2024-05-11 00:00:20 +02:00
None => Err(ParseError::UnexpectedEnd),
}
}
2024-05-12 18:56:10 +02:00
pub fn parse_type_tag<It>(
typectx: &Arc<RwLock<laddertypes::dict::TypeDict>>,
2024-05-12 18:58:39 +02:00
tokens: &mut Peekable<LTIRLexer<It>>,
2024-05-12 18:56:10 +02:00
) -> Option<TypeTag>
2024-05-12 18:58:39 +02:00
where
It: Iterator<Item = char>,
2024-05-12 18:56:10 +02:00
{
2024-05-13 22:55:24 +02:00
if let Some((region, peektok)) = tokens.peek().clone() {
2024-05-12 18:56:10 +02:00
match peektok.clone() {
Ok(LTIRToken::AssignType(typeterm_str)) => {
tokens.next();
match typectx.write().unwrap().parse(typeterm_str.as_str()) {
2024-05-12 18:58:39 +02:00
Ok(typeterm) => Some(Ok(typeterm)),
Err(parse_error) => Some(Err(TypeError::ParseError(parse_error))),
2024-05-12 18:56:10 +02:00
}
}
2024-05-12 18:58:39 +02:00
_ => None,
2024-05-12 18:56:10 +02:00
}
} else {
None
}
}
2024-05-09 20:13:10 +02:00
pub fn parse_statement<It>(
2024-05-12 18:56:10 +02:00
typectx: &Arc<RwLock<laddertypes::dict::TypeDict>>,
2024-05-12 18:58:39 +02:00
tokens: &mut Peekable<LTIRLexer<It>>,
) -> Result<crate::expr::Statement, ParseError>
where
It: Iterator<Item = char>,
2024-05-09 20:13:10 +02:00
{
2024-05-13 22:55:24 +02:00
if let Some((region, peektok)) = tokens.peek() {
2024-05-11 00:00:20 +02:00
match peektok {
Ok(LTIRToken::Symbol(sym)) => {
match sym.as_str() {
"!" => {
tokens.next();
2024-05-11 18:07:58 +02:00
// todo accept address-expression instead of symbol
2024-05-11 00:00:20 +02:00
let name = parse_symbol(tokens)?;
2024-05-12 18:56:10 +02:00
let val_expr = parse_expr(typectx, tokens)?;
2024-05-11 00:00:20 +02:00
let _ = parse_expect(tokens, LTIRToken::StatementSep)?;
Ok(Statement::Assignment {
var_id: name,
2024-05-12 18:58:39 +02:00
val_expr,
2024-05-11 00:00:20 +02:00
})
}
"let" => {
tokens.next();
let name = parse_symbol(tokens)?;
2024-05-12 18:56:10 +02:00
let typ = parse_type_tag(typectx, tokens);
let _ = parse_expect(tokens, LTIRToken::AssignValue);
2024-05-12 18:56:10 +02:00
let val_expr = parse_expr(typectx, tokens)?;
let _ = parse_expect(tokens, LTIRToken::StatementSep)?;
2024-05-12 18:58:39 +02:00
Ok(Statement::LetAssign {
2024-05-12 18:56:10 +02:00
typ,
var_id: name,
2024-05-12 18:58:39 +02:00
val_expr,
})
}
2024-05-11 00:00:20 +02:00
"while" => {
tokens.next();
let _ = parse_expect(tokens, LTIRToken::ExprOpen)?;
2024-05-12 18:56:10 +02:00
let cond = parse_expr(typectx, tokens)?;
2024-05-11 00:00:20 +02:00
let _ = parse_expect(tokens, LTIRToken::ExprClose)?;
Ok(Statement::WhileLoop {
condition: cond,
2024-05-12 18:58:39 +02:00
body: parse_block(typectx, tokens)?,
2024-05-11 00:00:20 +02:00
})
}
"return" => {
tokens.next();
2024-05-12 18:56:10 +02:00
let expr = parse_expr(typectx, tokens)?;
2024-05-11 00:00:20 +02:00
let _ = parse_expect(tokens, LTIRToken::StatementSep)?;
2024-05-12 18:56:10 +02:00
Ok(Statement::Return(parse_expr(typectx, tokens)?))
2024-05-11 00:00:20 +02:00
}
_ => {
2024-05-12 18:56:10 +02:00
let expr = parse_expr(typectx, tokens)?;
2024-05-11 00:00:20 +02:00
let _ = parse_expect(tokens, LTIRToken::StatementSep)?;
Ok(Statement::Expr(expr))
}
}
}
Ok(_) => {
2024-05-12 18:56:10 +02:00
let expr = parse_expr(typectx, tokens)?;
2024-05-11 00:00:20 +02:00
let _ = parse_expect(tokens, LTIRToken::StatementSep)?;
Ok(Statement::Expr(expr))
2024-05-12 18:58:39 +02:00
}
Err(err) => Err(ParseError::LexError(err.clone())),
2024-05-11 00:00:20 +02:00
}
} else {
Err(ParseError::UnexpectedEnd)
}
2024-05-09 20:13:10 +02:00
}
pub fn parse_block<It>(
2024-05-12 18:56:10 +02:00
typectx: &Arc<RwLock<laddertypes::dict::TypeDict>>,
2024-05-12 18:58:39 +02:00
tokens: &mut Peekable<LTIRLexer<It>>,
) -> Result<Vec<Statement>, ParseError>
where
It: Iterator<Item = char>,
2024-05-09 20:13:10 +02:00
{
2024-05-11 00:00:20 +02:00
let _ = parse_expect(tokens, LTIRToken::BlockOpen)?;
let mut statements = Vec::new();
2024-05-13 22:55:24 +02:00
while let Some((region, peektok)) = tokens.peek() {
2024-05-11 00:00:20 +02:00
match peektok {
Ok(LTIRToken::BlockClose) => {
tokens.next();
2024-05-12 18:58:39 +02:00
return Ok(statements);
}
Ok(_) => {
statements.push(parse_statement(typectx, tokens)?);
}
Err(err) => {
return Err(ParseError::LexError(err.clone()));
2024-05-11 00:00:20 +02:00
}
}
}
2024-05-09 20:13:10 +02:00
Err(ParseError::UnexpectedEnd)
}
2024-05-11 00:00:20 +02:00
pub fn parse_atom<It>(
2024-05-12 18:58:39 +02:00
tokens: &mut Peekable<LTIRLexer<It>>,
) -> Result<crate::expr::LTExpr, ParseError>
where
It: Iterator<Item = char>,
2024-05-11 00:00:20 +02:00
{
match tokens.next() {
2024-05-13 22:55:24 +02:00
Some((region, Ok(LTIRToken::Symbol(sym)))) => Ok(LTExpr::symbol(sym.as_str())),
Some((region, Ok(LTIRToken::Char(c)))) => Ok(LTExpr::lit_uint(c as u64)),
Some((region, Ok(LTIRToken::Num(n)))) => Ok(LTExpr::lit_uint(n as u64)),
Some((region, Ok(_))) => Err(ParseError::UnexpectedToken),
Some((region, Err(err))) => Err(ParseError::LexError(err)),
2024-05-12 18:58:39 +02:00
None => Err(ParseError::UnexpectedEnd),
2024-05-11 00:00:20 +02:00
}
}
2024-05-09 20:13:10 +02:00
pub fn parse_expr<It>(
2024-05-12 18:56:10 +02:00
typectx: &Arc<RwLock<laddertypes::dict::TypeDict>>,
2024-05-12 18:58:39 +02:00
tokens: &mut Peekable<LTIRLexer<It>>,
) -> Result<crate::expr::LTExpr, ParseError>
where
It: Iterator<Item = char>,
2024-05-09 20:13:10 +02:00
{
let mut children = Vec::new();
2024-05-13 22:55:24 +02:00
while let Some((region, tok)) = tokens.peek() {
2024-05-11 00:00:20 +02:00
match tok {
Ok(LTIRToken::Lambda) => {
if children.len() == 0 {
tokens.next();
let mut args = Vec::new();
2024-05-13 22:55:24 +02:00
while let Some((region, Ok(LTIRToken::Symbol(_)))) = tokens.peek() {
2024-05-12 18:58:39 +02:00
args.push((parse_symbol(tokens)?, parse_type_tag(typectx, tokens)));
}
2024-05-12 18:56:10 +02:00
let _ = parse_expect(tokens, LTIRToken::LambdaBody);
let body = parse_expr(typectx, tokens)?;
2024-05-11 00:00:20 +02:00
2024-05-12 18:58:39 +02:00
return Ok(LTExpr::Abstraction {
args,
2024-05-12 18:58:39 +02:00
body: Box::new(body),
2024-05-11 00:00:20 +02:00
});
} else {
return Err(ParseError::UnexpectedToken);
}
2024-05-09 20:13:10 +02:00
}
2024-05-11 00:00:20 +02:00
Ok(LTIRToken::ExprOpen) => {
tokens.next();
2024-05-13 22:55:24 +02:00
while let Some((region, peektok)) = tokens.peek() {
2024-05-11 00:00:20 +02:00
match peektok {
Ok(LTIRToken::ExprClose) => {
tokens.next();
break;
}
_ => {}
}
2024-05-12 18:56:10 +02:00
children.push(parse_expr(typectx, tokens)?);
2024-05-11 00:00:20 +02:00
}
2024-05-12 18:58:39 +02:00
}
Ok(LTIRToken::ExprClose) => {
break;
}
2024-05-12 18:56:10 +02:00
Ok(LTIRToken::BlockOpen) => {
2024-05-12 18:58:39 +02:00
children.push(LTExpr::block(parse_block(typectx, tokens)?));
2024-05-12 18:56:10 +02:00
}
2024-05-12 18:58:39 +02:00
Ok(LTIRToken::BlockClose) => {
break;
}
Ok(LTIRToken::StatementSep) => {
break;
}
Ok(LTIRToken::Symbol(name)) => match name.as_str() {
"if" => {
tokens.next();
let _ = parse_expect(tokens, LTIRToken::ExprOpen)?;
let cond = parse_expr(typectx, tokens)?;
let _ = parse_expect(tokens, LTIRToken::ExprClose)?;
let if_expr = LTExpr::block(parse_block(typectx, tokens)?);
let mut else_expr = LTExpr::block(vec![]);
2024-05-11 00:00:20 +02:00
2024-05-13 22:55:24 +02:00
if let Some((region, peektok)) = tokens.peek() {
2024-05-12 18:58:39 +02:00
if let Ok(LTIRToken::Symbol(name)) = peektok {
if name == "else" {
tokens.next();
else_expr = parse_expr(typectx, tokens)?;
2024-05-11 00:00:20 +02:00
}
}
}
2024-05-12 18:58:39 +02:00
children.push(LTExpr::Branch {
condition: Box::new(cond),
if_expr: Box::new(if_expr),
else_expr: Box::new(else_expr),
});
}
name => {
children.push(parse_atom(tokens)?);
2024-05-11 00:00:20 +02:00
}
2024-05-12 18:58:39 +02:00
},
Ok(atom) => {
children.push(parse_atom(tokens)?);
}
Err(err) => {
return Err(ParseError::LexError(err.clone()));
2024-05-11 00:00:20 +02:00
}
2024-05-09 20:13:10 +02:00
}
}
if children.len() > 0 {
let head = children.remove(0);
Ok(LTExpr::Application {
2024-05-12 18:56:10 +02:00
typ: None,
2024-05-09 20:13:10 +02:00
head: Box::new(head),
2024-05-12 18:58:39 +02:00
body: children,
2024-05-09 20:13:10 +02:00
})
} else {
Err(ParseError::UnexpectedEnd)
}
}