function parsing and stuff..

This commit is contained in:
janis 2025-10-13 23:37:06 +02:00
parent c7d5e4e6c0
commit a2632ca06e
Signed by: janis
SSH key fingerprint: SHA256:bB1qbbqmDXZNT0KKD5c2Dfjg53JGhj7B3CFcLIzSqq8
2 changed files with 146 additions and 85 deletions

View file

@ -6,7 +6,7 @@ use chumsky::{
extra::{self, SimpleState},
input::{IterInput, MapExtra},
pratt::{infix, left, postfix, prefix, right},
prelude::{Recursive, choice, just, recursive},
prelude::{Recursive, choice, just, none_of, recursive, via_parser},
recursive::{Direct, Indirect},
select, text,
};
@ -424,13 +424,9 @@ pub enum Visibility {
}
#[derive(Debug, Error)]
pub enum ParseError<'a> {
#[error("End of file.")]
EOF,
#[error("Unexpected token: {0:?}")]
UnexpectedToken(Token<'a>),
#[error("Not a type.")]
NotAType,
pub enum ParseError {
#[error("expected a valid if condition expression")]
IfCondition,
}
#[derive(Default, Debug)]
@ -594,11 +590,16 @@ where
this
}
pub fn expr(&self) -> impl Parser<'src, TokenInput<'src>, PlaceOrValue, ParserExtra> + Clone {
pub fn expr(
&self,
) -> impl Parser<'src, TokenInput<'src>, PlaceOrValue, ParserExtra> + Clone + use<'src, 'b>
{
self.expr.clone()
}
pub fn function(&self) -> impl Parser<'src, TokenInput<'src>, Index, ParserExtra> + Clone {
pub fn function(
&self,
) -> impl Parser<'src, TokenInput<'src>, Index, ParserExtra> + Clone + use<'src, 'b> {
self.function.clone()
}
@ -627,6 +628,8 @@ where
);
let constant = select! {
Token::True => (Intern::new(Value::Bool(true)), Intern::new(InnerType::Bool)),
Token::False => (Intern::new(Value::Bool(false)), Intern::new(InnerType::Bool)),
Token::FloatingConstant(lexeme)|
Token::DotFloatingConstant(lexeme)|
Token::FloatingExpConstant(lexeme)|
@ -732,9 +735,25 @@ where
})
}
fn into_value(idx: PlaceOrValue, e: &mut E) -> PlaceOrValue {
match idx {
PlaceOrValue::Place(index) => e.state().push(AstNode::PlaceToValue { expr: index }),
PlaceOrValue::Value(index) => index,
}
.as_value()
}
fn into_place(idx: PlaceOrValue, e: &mut E) -> PlaceOrValue {
match idx {
PlaceOrValue::Value(index) => e.state().push(AstNode::ValueToPlace { expr: index }),
PlaceOrValue::Place(index) => index,
}
.as_place()
}
fn create_expr(
&'_ self,
) -> Box<impl Parser<'src, TokenInput<'src>, PlaceOrValue, ParserExtra> + Clone + use<'src, 'b>>
) -> impl Parser<'src, TokenInput<'src>, PlaceOrValue, ParserExtra> + Clone + use<'src, 'b>
{
let assignment = choice((
just(Token::Equal),
@ -780,22 +799,6 @@ where
let r#as = just(Token::As).ignore_then(type_parser::<ParserExtra>());
fn into_value(idx: PlaceOrValue, e: &mut E) -> PlaceOrValue {
match idx {
PlaceOrValue::Place(index) => e.state().push(AstNode::PlaceToValue { expr: index }),
PlaceOrValue::Value(index) => index,
}
.as_value()
}
fn into_place(idx: PlaceOrValue, e: &mut E) -> PlaceOrValue {
match idx {
PlaceOrValue::Value(index) => e.state().push(AstNode::ValueToPlace { expr: index }),
PlaceOrValue::Place(index) => index,
}
.as_place()
}
// TODO: postfix: function call, field access, array subscript
let _expr = self.expr.clone();
@ -806,12 +809,12 @@ where
just(Token::CloseSquareBracket),
)
// subscript takes a value as the index
.map_with(into_value);
.map_with(Self::into_value);
let arguments = _expr
.clone()
// arguments take values
.map_with(into_value)
.map_with(Self::into_value)
.map(PlaceOrValue::index)
.separated_by(just(Token::Comma))
.allow_trailing()
@ -823,7 +826,7 @@ where
let assignment_expr = self.simple_expr().pratt((
postfix(100, subscript, |expr, index: PlaceOrValue, e: &mut E| {
let node = AstNode::Subscript {
expr: into_value(expr, e).index(),
expr: Self::into_value(expr, e).index(),
index: index.index(),
};
// subscript yields a place
@ -831,7 +834,7 @@ where
}),
postfix(100, arguments, |callee, arguments, e: &mut E| {
let node = AstNode::CallExpr {
callee: into_value(callee, e).index(),
callee: Self::into_value(callee, e).index(),
arguments,
};
// function call yields a value
@ -839,7 +842,7 @@ where
}),
postfix(100, field, |expr, field: &str, e: &mut E| {
let node = AstNode::FieldAccess {
expr: into_place(expr, e).index(),
expr: Self::into_place(expr, e).index(),
field: field.to_string(),
};
// field access yields a place
@ -847,15 +850,15 @@ where
}),
postfix(99, r#as, |expr, ty, e: &mut E| {
let node = AstNode::ExplicitCast {
expr: into_value(expr, e).index(),
expr: Self::into_value(expr, e).index(),
ty,
};
e.state().push(node).as_value()
}),
prefix(95, prefixes, |op, expr, e: &mut E| {
let node = match op {
Token::Bang => AstNode::Not(into_value(expr, e).index()),
Token::Minus => AstNode::Negate(into_value(expr, e).index()),
Token::Bang => AstNode::Not(Self::into_value(expr, e).index()),
Token::Minus => AstNode::Negate(Self::into_value(expr, e).index()),
Token::Star => {
return e
.state()
@ -863,15 +866,15 @@ where
.as_place();
}
Token::Ampersand => AstNode::AddressOf {
expr: into_place(expr, e).index(),
expr: Self::into_place(expr, e).index(),
},
_ => unreachable!(),
};
e.state().push(node).as_value()
}),
infix(left(90), multiplicative, |left, op, right, e: &mut E| {
let left = into_value(left, e).index();
let right = into_value(right, e).index();
let left = Self::into_value(left, e).index();
let right = Self::into_value(right, e).index();
let node = match op {
Token::Star => AstNode::Multiply { left, right },
Token::Slash => AstNode::Divide { left, right },
@ -881,8 +884,8 @@ where
e.state().push(node).as_value()
}),
infix(left(80), additive, |left, op, right, e: &mut E| {
let left = into_value(left, e).index();
let right = into_value(right, e).index();
let left = Self::into_value(left, e).index();
let right = Self::into_value(right, e).index();
let node = match op {
Token::Plus => AstNode::Add { left, right },
Token::Minus => AstNode::Subtract { left, right },
@ -891,8 +894,8 @@ where
e.state().push(node).as_value()
}),
infix(left(70), shift, |left, op, right, e: &mut E| {
let left = into_value(left, e).index();
let right = into_value(right, e).index();
let left = Self::into_value(left, e).index();
let right = Self::into_value(right, e).index();
let node = match op {
Token::LessLess => AstNode::ShiftLeft { left, right },
Token::GreaterGreater => AstNode::ShiftRight { left, right },
@ -901,8 +904,8 @@ where
e.state().push(node).as_value()
}),
infix(left(60), relational, |left, op, right, e: &mut E| {
let left = into_value(left, e).index();
let right = into_value(right, e).index();
let left = Self::into_value(left, e).index();
let right = Self::into_value(right, e).index();
let node = match op {
Token::Less => AstNode::Less { left, right },
Token::LessEqual => AstNode::LessEq { left, right },
@ -913,8 +916,8 @@ where
e.state().push(node).as_value()
}),
infix(left(50), equality, |left, op, right, e: &mut E| {
let left = into_value(left, e).index();
let right = into_value(right, e).index();
let left = Self::into_value(left, e).index();
let right = Self::into_value(right, e).index();
let node = match op {
Token::EqualEqual => AstNode::Eq { left, right },
Token::BangEqual => AstNode::NotEq { left, right },
@ -923,42 +926,42 @@ where
e.state().push(node).as_value()
}),
infix(left(40), and, |left, _op, right, e: &mut E| {
let left = into_value(left, e).index();
let right = into_value(right, e).index();
let left = Self::into_value(left, e).index();
let right = Self::into_value(right, e).index();
let node = AstNode::BitAnd { left, right };
e.state().push(node).as_value()
}),
infix(left(30), xor, |left, _op, right, e: &mut E| {
let left = into_value(left, e).index();
let right = into_value(right, e).index();
let left = Self::into_value(left, e).index();
let right = Self::into_value(right, e).index();
let node = AstNode::BitXor { left, right };
e.state().push(node).as_value()
}),
infix(left(20), or, |left, _op, right, e: &mut E| {
let left = into_value(left, e).index();
let right = into_value(right, e).index();
let left = Self::into_value(left, e).index();
let right = Self::into_value(right, e).index();
let node = AstNode::BitOr { left, right };
e.state().push(node).as_value()
}),
infix(left(10), logical_and, |left, _op, right, e: &mut E| {
let left = into_value(left, e).index();
let right = into_value(right, e).index();
let left = Self::into_value(left, e).index();
let right = Self::into_value(right, e).index();
let node = AstNode::LogicalAnd { left, right };
e.state().push(node).as_value()
}),
infix(left(5), logical_or, |left, _op, right, e: &mut E| {
let left = into_value(left, e).index();
let right = into_value(right, e).index();
let left = Self::into_value(left, e).index();
let right = Self::into_value(right, e).index();
let node = AstNode::LogicalOr { left, right };
e.state().push(node).as_value()
}),
infix(right(1), assignment, |left, op, right, e: &mut E| {
let dest = into_place(left, e).index();
let right = into_value(right, e).index();
let dest = Self::into_place(left, e).index();
let right = Self::into_value(right, e).index();
let node = if op == Token::Equal {
AstNode::Assignment { dest, expr: right }
} else {
let left = into_value(left, e).index();
let left = Self::into_value(left, e).index();
let right = match op {
Token::PlusEqual => e.state().push(AstNode::Add { left, right }),
Token::MinusEqual => e.state().push(AstNode::Subtract { left, right }),
@ -980,17 +983,43 @@ where
}),
));
let else_expr = just(Token::Else).ignore_then(_expr.clone());
let expr = choice((self.if_else_expr(), assignment_expr)).labelled("expression");
let if_expr = just(Token::If)
expr
}
fn if_else_expr(
&self,
) -> impl Parser<'src, TokenInput<'src>, PlaceOrValue, ParserExtra> + Clone + use<'src, 'b>
{
let else_expr = just(Token::Else).ignore_then(self.expr());
just(Token::If)
.ignore_then(
_expr
.clone()
.map_with(into_value)
self.expr()
.map_with(Self::into_value)
.map(PlaceOrValue::index)
.delimited_by(just(Token::OpenParens), just(Token::CloseParens)),
.delimited_by(just(Token::OpenParens), just(Token::CloseParens))
.recover_with(via_parser({
// parenthised recovery:
recursive(|recovery| {
just(Token::OpenParens)
.ignored()
.ignore_then(choice((
none_of([Token::CloseParens, Token::OpenParens]).map(|_| ()),
recovery.map(|_| ()),
)))
.repeated()
.then_ignore(just(Token::CloseParens))
})
.map_with(|_, e: &mut E| {
e.state().push(AstNode::Error {
err: Box::new(ParseError::IfCondition),
})
})
})),
)
.then(_expr.clone())
.then(self.expr())
.then(else_expr.or_not())
.map_with(|((condition, then), or), e: &mut E| {
// TODO: determine placeness from branches
@ -1000,11 +1029,7 @@ where
r#else: or.map(PlaceOrValue::index),
};
e.state().push(node).as_value()
});
let expr = choice((if_expr, assignment_expr)).labelled("expression");
Box::new(expr)
})
}
fn global_decl(&self) -> impl Parser<'src, TokenInput<'src>, Index, ParserExtra> + Clone {
@ -1071,7 +1096,7 @@ mod pretty;
mod tests {
use chumsky::{Parser, extra::SimpleState};
use crate::{Ast, AstNode, new_token_input, pretty, type_parser};
use crate::{Ast, AstNode, ParserCtx, ParserExtra, new_token_input, pretty, type_parser};
#[test]
fn print_ast_node_size() {
@ -1131,24 +1156,51 @@ mod tests {
);
}
fn print_ast<'src, T>(
tokens: crate::TokenInput<'src>,
parser: impl Parser<'src, crate::TokenInput<'src>, T, ParserExtra>,
) {
let mut state = SimpleState(Ast::new());
let out = parser.parse_with_state(tokens, &mut state);
let ast = state.0;
let mut pretty = pretty::PrettyPrint::new();
pretty.print(&ast);
}
#[test]
fn parse_exprs() {
let ctx = crate::ParserCtx::new();
let print_ast = |tokens| {
let mut state = SimpleState(Ast::new());
let out = ctx.function().parse_with_state(tokens, &mut state).unwrap();
let ast = state.0;
let mut pretty = pretty::PrettyPrint::new();
pretty.print(&ast);
};
print_ast(new_token_input(
r#"
print_ast(new_token_input("true"), ctx.expr());
}
#[test]
fn parse_fns() {
let ctx = crate::ParserCtx::new();
print_ast(
new_token_input(
r#"
fn my_function(a: i32, b: *const u8) -> i32 {
x = a + 1;
x
}
"#,
));
),
ctx.function(),
);
}
#[test]
fn parse_if_expr() {
let ctx = crate::ParserCtx::new();
print_ast(
new_token_input(
r#"
if (true) 1 / 2 else true
"#,
),
ctx.if_else_expr(),
);
}
}

View file

@ -277,13 +277,22 @@ impl PrettyPrint {
self.with_indent([*condition].into_iter(), |this, idx| {
this.stuff(ast, idx);
});
self.push_line("THEN".to_string());
if let Some(r#else) = r#else {
self.stuff(ast, *then);
self.with_indent(core::iter::once(*then), |this, idx| {
this.stuff(ast, idx);
});
*self.indents.last_mut().unwrap() = Indent::End;
self.stuff(ast, *r#else);
self.push_line("ELSE".to_string());
self.with_indent(core::iter::once(*r#else), |this, idx| {
this.stuff(ast, idx);
});
} else {
*self.indents.last_mut().unwrap() = Indent::End;
self.stuff(ast, *then);
self.with_indent(core::iter::once(*then), |this, idx| {
this.stuff(ast, idx);
});
}
self.indents.pop();
}