Formatting

This commit is contained in:
Michael Schimmel
2026-02-22 02:35:06 +01:00
parent 2123f1d279
commit 329b885c4b
25 changed files with 8053 additions and 6400 deletions
+400 -345
View File
@@ -1,345 +1,400 @@
use std::rc::Rc;
use crate::ast::lexer::{Lexer, Token, TokenKind};
use crate::ast::types::{Identity, NodeIdentity, Value, Keyword};
use crate::ast::nodes::{Node, UntypedKind, Symbol};
pub struct Parser<'a> {
lexer: Lexer<'a>,
current_token: Token,
}
impl<'a> Parser<'a> {
pub fn new(input: &'a str) -> Result<Self, String> {
let mut lexer = Lexer::new(input);
let current_token = lexer.next_token()?;
Ok(Self { lexer, current_token })
}
fn advance(&mut self) -> Result<Token, String> {
let prev = std::mem::replace(&mut self.current_token, self.lexer.next_token()?);
Ok(prev)
}
fn peek(&self) -> &TokenKind {
&self.current_token.kind
}
pub fn parse_expression(&mut self) -> Result<Node<UntypedKind>, String> {
let token_loc = self.current_token.location;
let identity = Rc::new(NodeIdentity { location: token_loc });
match self.peek() {
TokenKind::LeftParen => self.parse_list(),
TokenKind::LeftBracket => self.parse_vector_literal(),
TokenKind::LeftBrace => self.parse_record_literal(),
TokenKind::Quote => {
self.advance()?; // consume '
let expr = self.parse_expression()?;
Ok(Node {
identity: identity.clone(),
kind: UntypedKind::Call {
callee: Box::new(self.make_id_node("quote", identity.clone())),
args: Box::new(Node {
identity,
kind: UntypedKind::Tuple { elements: vec![expr] },
ty: (),
}),
},
ty: (),
})
}
TokenKind::Backtick => {
self.advance()?; // consume `
let expr = self.parse_expression()?;
Ok(Node {
identity,
kind: UntypedKind::Template(Box::new(expr)),
ty: (),
})
}
TokenKind::Tilde => {
self.advance()?; // consume ~
if *self.peek() == TokenKind::At {
self.advance()?; // consume @
let expr = self.parse_expression()?;
Ok(Node {
identity,
kind: UntypedKind::Splice(Box::new(expr)),
ty: (),
})
} else {
let expr = self.parse_expression()?;
Ok(Node {
identity,
kind: UntypedKind::Placeholder(Box::new(expr)),
ty: (),
})
}
}
_ => self.parse_atom(),
}
}
pub fn at_eof(&self) -> bool {
matches!(self.current_token.kind, TokenKind::EOF)
}
fn parse_atom(&mut self) -> Result<Node<UntypedKind>, String> {
let token = self.advance()?;
let identity = Rc::new(NodeIdentity { location: token.location });
let kind = match token.kind {
TokenKind::Integer(n) => UntypedKind::Constant(Value::Int(n)),
TokenKind::Float(n) => UntypedKind::Constant(Value::Float(n)),
TokenKind::String(s) => UntypedKind::Constant(Value::Text(s)),
TokenKind::Keyword(k) => UntypedKind::Constant(Value::Keyword(Keyword::intern(&k))),
TokenKind::Identifier(id) => {
match id.as_ref() {
"..." => UntypedKind::Nop,
_ => UntypedKind::Identifier(id.into()),
}
}
_ => return Err(format!("Unexpected token in atom: {:?} at {:?}", token.kind, token.location)),
};
Ok(Node { identity, kind, ty: () })
}
fn parse_list(&mut self) -> Result<Node<UntypedKind>, String> {
let start_loc = self.advance()?.location; // consume '('
let identity = Rc::new(NodeIdentity { location: start_loc });
if *self.peek() == TokenKind::RightParen {
return Err(format!("Empty list () is not a valid expression at {:?}", start_loc));
}
let head = self.parse_expression()?;
let result = if let UntypedKind::Identifier(ref sym) = head.kind {
match sym.name.as_ref() {
"if" => self.parse_if(identity),
"fn" => self.parse_fn(identity),
"def" => self.parse_def(identity),
"assign" => self.parse_assign(identity),
"do" => self.parse_do(identity),
"macro" => self.parse_macro_decl(identity),
_ => self.parse_call(head, identity),
}
} else {
self.parse_call(head, identity)
};
self.expect(TokenKind::RightParen)?;
result
}
fn parse_if(&mut self, identity: Identity) -> Result<Node<UntypedKind>, String> {
let cond = Box::new(self.parse_expression()?);
let then_br = Box::new(self.parse_expression()?);
let mut else_br = None;
if *self.peek() != TokenKind::RightParen {
else_br = Some(Box::new(self.parse_expression()?));
}
Ok(Node {
identity,
kind: UntypedKind::If { cond, then_br, else_br },
ty: (),
})
}
fn parse_def(&mut self, identity: Identity) -> Result<Node<UntypedKind>, String> {
let name_node = self.parse_expression()?;
let name = match name_node.kind {
UntypedKind::Identifier(sym) => sym,
_ => return Err("Expected identifier for def name".to_string()),
};
let value = Box::new(self.parse_expression()?);
Ok(Node {
identity,
kind: UntypedKind::Def { name, value },
ty: (),
})
}
fn parse_assign(&mut self, identity: Identity) -> Result<Node<UntypedKind>, String> {
// (assign target value)
let target = Box::new(self.parse_expression()?);
let value = Box::new(self.parse_expression()?);
Ok(Node {
identity,
kind: UntypedKind::Assign { target, value },
ty: (),
})
}
fn parse_do(&mut self, identity: Identity) -> Result<Node<UntypedKind>, String> {
let mut exprs = Vec::new();
while *self.peek() != TokenKind::RightParen && *self.peek() != TokenKind::EOF {
exprs.push(self.parse_expression()?);
}
Ok(Node {
identity,
kind: UntypedKind::Block { exprs },
ty: (),
})
}
fn parse_fn(&mut self, identity: Identity) -> Result<Node<UntypedKind>, String> {
let params = Box::new(self.parse_param_vector()?);
let body = self.parse_expression()?;
Ok(Node {
identity,
kind: UntypedKind::Lambda {
params,
body: Rc::new(body),
},
ty: (),
})
}
fn parse_macro_decl(&mut self, identity: Identity) -> Result<Node<UntypedKind>, String> {
let name_node = self.parse_expression()?;
let name = match name_node.kind {
UntypedKind::Identifier(sym) => sym,
_ => return Err("Expected identifier for macro name".to_string()),
};
let params = Box::new(self.parse_param_vector()?);
let body = self.parse_expression()?;
Ok(Node {
identity,
kind: UntypedKind::MacroDecl { name, params, body: Box::new(body) },
ty: (),
})
}
fn parse_param_vector(&mut self) -> Result<Node<UntypedKind>, String> {
if *self.peek() != TokenKind::LeftBracket {
return Err(format!("Expected parameter vector [...] for fn, found {:?}", self.peek()));
}
let token = self.advance()?;
let identity = Rc::new(NodeIdentity { location: token.location });
let mut elements = Vec::new();
while *self.peek() != TokenKind::RightBracket {
let next_peek = self.peek();
match next_peek {
TokenKind::Identifier(name) => {
let name = name.clone();
let token = self.advance()?;
let p_identity = Rc::new(NodeIdentity { location: token.location });
elements.push(Node {
identity: p_identity,
kind: UntypedKind::Parameter(name.into()),
ty: (),
});
},
TokenKind::LeftBracket => {
elements.push(self.parse_param_vector()?);
},
_ => return Err(format!("Expected identifier or nested parameter vector, found {:?}", next_peek)),
}
}
self.expect(TokenKind::RightBracket)?;
Ok(Node {
identity,
kind: UntypedKind::Tuple { elements },
ty: (),
})
}
fn parse_call(&mut self, callee: Node<UntypedKind>, identity: Identity) -> Result<Node<UntypedKind>, String> {
let mut elements = Vec::new();
while *self.peek() != TokenKind::RightParen && *self.peek() != TokenKind::EOF {
elements.push(self.parse_expression()?);
}
// The arguments are wrapped in a Tuple node, reusing the call's identity/location.
let args_node = Node {
identity: identity.clone(),
kind: UntypedKind::Tuple { elements },
ty: (),
};
Ok(Node {
identity,
kind: UntypedKind::Call { callee: Box::new(callee), args: Box::new(args_node) },
ty: (),
})
}
fn parse_vector_literal(&mut self) -> Result<Node<UntypedKind>, String> {
let token = self.advance()?;
let mut elements = Vec::new();
while *self.peek() != TokenKind::RightBracket && *self.peek() != TokenKind::EOF {
let expr = self.parse_expression()?;
elements.push(expr);
}
self.expect(TokenKind::RightBracket)?;
Ok(Node {
identity: Rc::new(NodeIdentity { location: token.location }),
kind: UntypedKind::Tuple { elements },
ty: (),
})
}
fn parse_record_literal(&mut self) -> Result<Node<UntypedKind>, String> {
let token = self.advance()?;
let mut fields = Vec::new();
while *self.peek() != TokenKind::RightBrace {
if *self.peek() == TokenKind::EOF {
return Err("Unexpected EOF in record literal".to_string());
}
let key_node = self.parse_expression()?;
// We check for keyword kind here (syntactically) to avoid ambiguity, but
// strictly we could allow any expression and check at runtime.
// Delphi enforces keywords. We can do minimal check here.
match &key_node.kind {
UntypedKind::Constant(Value::Keyword(_)) => {},
_ => return Err("Record keys must be keywords (syntactically)".to_string()),
}
if *self.peek() == TokenKind::RightBrace {
return Err("Record literal must have even number of forms".to_string());
}
let val_node = self.parse_expression()?;
fields.push((key_node, val_node));
}
self.expect(TokenKind::RightBrace)?;
Ok(Node {
identity: Rc::new(NodeIdentity { location: token.location }),
kind: UntypedKind::Record { fields },
ty: (),
})
}
fn expect(&mut self, kind: TokenKind) -> Result<(), String> {
let token = self.advance()?;
if token.kind == kind {
Ok(())
} else {
Err(format!("Expected {:?}, but found {:?} at {:?}", kind, token.kind, token.location))
}
}
fn make_id_node(&self, name: &str, identity: Identity) -> Node<UntypedKind> {
Node {
identity,
kind: UntypedKind::Identifier(Symbol::from(name)),
ty: (),
}
}
}
use crate::ast::lexer::{Lexer, Token, TokenKind};
use crate::ast::nodes::{Node, Symbol, UntypedKind};
use crate::ast::types::{Identity, Keyword, NodeIdentity, Value};
use std::rc::Rc;
pub struct Parser<'a> {
lexer: Lexer<'a>,
current_token: Token,
}
impl<'a> Parser<'a> {
pub fn new(input: &'a str) -> Result<Self, String> {
let mut lexer = Lexer::new(input);
let current_token = lexer.next_token()?;
Ok(Self {
lexer,
current_token,
})
}
fn advance(&mut self) -> Result<Token, String> {
let prev = std::mem::replace(&mut self.current_token, self.lexer.next_token()?);
Ok(prev)
}
fn peek(&self) -> &TokenKind {
&self.current_token.kind
}
pub fn parse_expression(&mut self) -> Result<Node<UntypedKind>, String> {
let token_loc = self.current_token.location;
let identity = Rc::new(NodeIdentity {
location: token_loc,
});
match self.peek() {
TokenKind::LeftParen => self.parse_list(),
TokenKind::LeftBracket => self.parse_vector_literal(),
TokenKind::LeftBrace => self.parse_record_literal(),
TokenKind::Quote => {
self.advance()?; // consume '
let expr = self.parse_expression()?;
Ok(Node {
identity: identity.clone(),
kind: UntypedKind::Call {
callee: Box::new(self.make_id_node("quote", identity.clone())),
args: Box::new(Node {
identity,
kind: UntypedKind::Tuple {
elements: vec![expr],
},
ty: (),
}),
},
ty: (),
})
}
TokenKind::Backtick => {
self.advance()?; // consume `
let expr = self.parse_expression()?;
Ok(Node {
identity,
kind: UntypedKind::Template(Box::new(expr)),
ty: (),
})
}
TokenKind::Tilde => {
self.advance()?; // consume ~
if *self.peek() == TokenKind::At {
self.advance()?; // consume @
let expr = self.parse_expression()?;
Ok(Node {
identity,
kind: UntypedKind::Splice(Box::new(expr)),
ty: (),
})
} else {
let expr = self.parse_expression()?;
Ok(Node {
identity,
kind: UntypedKind::Placeholder(Box::new(expr)),
ty: (),
})
}
}
_ => self.parse_atom(),
}
}
pub fn at_eof(&self) -> bool {
matches!(self.current_token.kind, TokenKind::EOF)
}
fn parse_atom(&mut self) -> Result<Node<UntypedKind>, String> {
let token = self.advance()?;
let identity = Rc::new(NodeIdentity {
location: token.location,
});
let kind = match token.kind {
TokenKind::Integer(n) => UntypedKind::Constant(Value::Int(n)),
TokenKind::Float(n) => UntypedKind::Constant(Value::Float(n)),
TokenKind::String(s) => UntypedKind::Constant(Value::Text(s)),
TokenKind::Keyword(k) => UntypedKind::Constant(Value::Keyword(Keyword::intern(&k))),
TokenKind::Identifier(id) => match id.as_ref() {
"..." => UntypedKind::Nop,
_ => UntypedKind::Identifier(id.into()),
},
_ => {
return Err(format!(
"Unexpected token in atom: {:?} at {:?}",
token.kind, token.location
));
}
};
Ok(Node {
identity,
kind,
ty: (),
})
}
fn parse_list(&mut self) -> Result<Node<UntypedKind>, String> {
let start_loc = self.advance()?.location; // consume '('
let identity = Rc::new(NodeIdentity {
location: start_loc,
});
if *self.peek() == TokenKind::RightParen {
return Err(format!(
"Empty list () is not a valid expression at {:?}",
start_loc
));
}
let head = self.parse_expression()?;
let result = if let UntypedKind::Identifier(ref sym) = head.kind {
match sym.name.as_ref() {
"if" => self.parse_if(identity),
"fn" => self.parse_fn(identity),
"def" => self.parse_def(identity),
"assign" => self.parse_assign(identity),
"do" => self.parse_do(identity),
"macro" => self.parse_macro_decl(identity),
_ => self.parse_call(head, identity),
}
} else {
self.parse_call(head, identity)
};
self.expect(TokenKind::RightParen)?;
result
}
fn parse_if(&mut self, identity: Identity) -> Result<Node<UntypedKind>, String> {
let cond = Box::new(self.parse_expression()?);
let then_br = Box::new(self.parse_expression()?);
let mut else_br = None;
if *self.peek() != TokenKind::RightParen {
else_br = Some(Box::new(self.parse_expression()?));
}
Ok(Node {
identity,
kind: UntypedKind::If {
cond,
then_br,
else_br,
},
ty: (),
})
}
fn parse_def(&mut self, identity: Identity) -> Result<Node<UntypedKind>, String> {
let name_node = self.parse_expression()?;
let name = match name_node.kind {
UntypedKind::Identifier(sym) => sym,
_ => return Err("Expected identifier for def name".to_string()),
};
let value = Box::new(self.parse_expression()?);
Ok(Node {
identity,
kind: UntypedKind::Def { name, value },
ty: (),
})
}
fn parse_assign(&mut self, identity: Identity) -> Result<Node<UntypedKind>, String> {
// (assign target value)
let target = Box::new(self.parse_expression()?);
let value = Box::new(self.parse_expression()?);
Ok(Node {
identity,
kind: UntypedKind::Assign { target, value },
ty: (),
})
}
fn parse_do(&mut self, identity: Identity) -> Result<Node<UntypedKind>, String> {
let mut exprs = Vec::new();
while *self.peek() != TokenKind::RightParen && *self.peek() != TokenKind::EOF {
exprs.push(self.parse_expression()?);
}
Ok(Node {
identity,
kind: UntypedKind::Block { exprs },
ty: (),
})
}
fn parse_fn(&mut self, identity: Identity) -> Result<Node<UntypedKind>, String> {
let params = Box::new(self.parse_param_vector()?);
let body = self.parse_expression()?;
Ok(Node {
identity,
kind: UntypedKind::Lambda {
params,
body: Rc::new(body),
},
ty: (),
})
}
fn parse_macro_decl(&mut self, identity: Identity) -> Result<Node<UntypedKind>, String> {
let name_node = self.parse_expression()?;
let name = match name_node.kind {
UntypedKind::Identifier(sym) => sym,
_ => return Err("Expected identifier for macro name".to_string()),
};
let params = Box::new(self.parse_param_vector()?);
let body = self.parse_expression()?;
Ok(Node {
identity,
kind: UntypedKind::MacroDecl {
name,
params,
body: Box::new(body),
},
ty: (),
})
}
fn parse_param_vector(&mut self) -> Result<Node<UntypedKind>, String> {
if *self.peek() != TokenKind::LeftBracket {
return Err(format!(
"Expected parameter vector [...] for fn, found {:?}",
self.peek()
));
}
let token = self.advance()?;
let identity = Rc::new(NodeIdentity {
location: token.location,
});
let mut elements = Vec::new();
while *self.peek() != TokenKind::RightBracket {
let next_peek = self.peek();
match next_peek {
TokenKind::Identifier(name) => {
let name = name.clone();
let token = self.advance()?;
let p_identity = Rc::new(NodeIdentity {
location: token.location,
});
elements.push(Node {
identity: p_identity,
kind: UntypedKind::Parameter(name.into()),
ty: (),
});
}
TokenKind::LeftBracket => {
elements.push(self.parse_param_vector()?);
}
_ => {
return Err(format!(
"Expected identifier or nested parameter vector, found {:?}",
next_peek
));
}
}
}
self.expect(TokenKind::RightBracket)?;
Ok(Node {
identity,
kind: UntypedKind::Tuple { elements },
ty: (),
})
}
fn parse_call(
&mut self,
callee: Node<UntypedKind>,
identity: Identity,
) -> Result<Node<UntypedKind>, String> {
let mut elements = Vec::new();
while *self.peek() != TokenKind::RightParen && *self.peek() != TokenKind::EOF {
elements.push(self.parse_expression()?);
}
// The arguments are wrapped in a Tuple node, reusing the call's identity/location.
let args_node = Node {
identity: identity.clone(),
kind: UntypedKind::Tuple { elements },
ty: (),
};
Ok(Node {
identity,
kind: UntypedKind::Call {
callee: Box::new(callee),
args: Box::new(args_node),
},
ty: (),
})
}
fn parse_vector_literal(&mut self) -> Result<Node<UntypedKind>, String> {
let token = self.advance()?;
let mut elements = Vec::new();
while *self.peek() != TokenKind::RightBracket && *self.peek() != TokenKind::EOF {
let expr = self.parse_expression()?;
elements.push(expr);
}
self.expect(TokenKind::RightBracket)?;
Ok(Node {
identity: Rc::new(NodeIdentity {
location: token.location,
}),
kind: UntypedKind::Tuple { elements },
ty: (),
})
}
fn parse_record_literal(&mut self) -> Result<Node<UntypedKind>, String> {
let token = self.advance()?;
let mut fields = Vec::new();
while *self.peek() != TokenKind::RightBrace {
if *self.peek() == TokenKind::EOF {
return Err("Unexpected EOF in record literal".to_string());
}
let key_node = self.parse_expression()?;
// We check for keyword kind here (syntactically) to avoid ambiguity, but
// strictly we could allow any expression and check at runtime.
// Delphi enforces keywords. We can do minimal check here.
match &key_node.kind {
UntypedKind::Constant(Value::Keyword(_)) => {}
_ => return Err("Record keys must be keywords (syntactically)".to_string()),
}
if *self.peek() == TokenKind::RightBrace {
return Err("Record literal must have even number of forms".to_string());
}
let val_node = self.parse_expression()?;
fields.push((key_node, val_node));
}
self.expect(TokenKind::RightBrace)?;
Ok(Node {
identity: Rc::new(NodeIdentity {
location: token.location,
}),
kind: UntypedKind::Record { fields },
ty: (),
})
}
fn expect(&mut self, kind: TokenKind) -> Result<(), String> {
let token = self.advance()?;
if token.kind == kind {
Ok(())
} else {
Err(format!(
"Expected {:?}, but found {:?} at {:?}",
kind, token.kind, token.location
))
}
}
fn make_id_node(&self, name: &str, identity: Identity) -> Node<UntypedKind> {
Node {
identity,
kind: UntypedKind::Identifier(Symbol::from(name)),
ty: (),
}
}
}