fix type parsing bugs

This commit is contained in:
2024-10-23 22:10:48 -04:00
parent 8fbb4780df
commit ae7600ba81
4 changed files with 62 additions and 34 deletions

View File

@@ -317,6 +317,8 @@ where
}, },
ParseTree::FunctionCall(ident, args) => { ParseTree::FunctionCall(ident, args) => {
let obj = self.get_object_mut(&ident)?; let obj = self.get_object_mut(&ident)?;
let globals = Self::obj_globals(obj);
let locals = Self::obj_locals(obj);
let v = Self::eval(obj)?; let v = Self::eval(obj)?;
match v { match v {
@@ -325,7 +327,7 @@ where
.map(|x| Object::variable(x, self.globals.clone(), self.locals.clone())) .map(|x| Object::variable(x, self.globals.clone(), self.locals.clone()))
.collect(); .collect();
f.call(Self::obj_globals(obj), Self::obj_locals(obj), args) f.call(globals, locals, args)
}, },
_ => Err(RuntimeError::FunctionUndefined(ident.clone())) _ => Err(RuntimeError::FunctionUndefined(ident.clone()))
} }

View File

@@ -18,7 +18,7 @@ impl Display for FunctionType {
#[derive(Clone, Debug, PartialEq)] #[derive(Clone, Debug, PartialEq)]
pub struct Function { pub struct Function {
name: Option<String>, pub(crate) name: Option<String>,
t: FunctionType, t: FunctionType,
arg_names: Vec<String>, arg_names: Vec<String>,
body: Box<ParseTree>, body: Box<ParseTree>,

View File

@@ -112,13 +112,16 @@ impl<'a, I: Iterator<Item = Result<Token, TokenizeError>>> Parser<'a, I> {
} }
} }
pub fn add_global(self, k: String, v: Type) -> Self { pub fn add_global(self, k: String, v: Type) -> Self {
self.globals.insert(k, v); self.globals.insert(k, v);
self self
} }
pub fn add_globals<Items: Iterator<Item = (String, Type)>>(self, items: Items) -> Self { pub fn add_globals<Items: Iterator<Item = (String, Type)>>(self, items: Items) -> Self {
items.for_each(|(name, t)| _ = self.globals.insert(name, t)); items.for_each(|(name, t)| {
self.globals.insert(name, t);
});
self self
} }
@@ -133,7 +136,9 @@ impl<'a, I: Iterator<Item = Result<Token, TokenizeError>>> Parser<'a, I> {
} }
pub fn add_locals<Items: Iterator<Item = (String, Type)>>(mut self, items: Items) -> Self { pub fn add_locals<Items: Iterator<Item = (String, Type)>>(mut self, items: Items) -> Self {
items.for_each(|(name, t)| _ = self.locals.insert(name, t)); items.for_each(|(name, t)| {
self.locals.insert(name, t);
});
self self
} }
@@ -369,22 +374,27 @@ impl<'a, I: Iterator<Item = Result<Token, TokenizeError>>> Parser<'a, I> {
.locals(locals).parse()?))) .locals(locals).parse()?)))
} }
fn parse_function_declaration(tokens: &mut Peekable<I>, arg_count: usize) -> Result<(FunctionType, Vec<String>), ParseError> { fn parse_function_declaration(
tokens: &mut Peekable<I>,
arg_count: usize) -> Result<(FunctionType, Vec<String>), ParseError>
{
let args: Vec<(Type, String)> = (0..arg_count) let args: Vec<(Type, String)> = (0..arg_count)
.map(|_| Self::parse_function_declaration_parameter(tokens)) .map(|_| Self::parse_function_declaration_parameter(tokens))
.collect::<Result<_, _>>()?; .collect::<Result<_, _>>()?;
let (types, names): (Vec<_>, Vec<_>) = args.into_iter().unzip(); let (types, names): (Vec<_>, Vec<_>) = args.into_iter().unzip();
let mut ret = Type::Any; let ret = if tokens.next_if(|x| matches!(x, Ok(Token::Operator(Op::Arrow)))).is_some() {
Self::parse_type(tokens)?
if tokens.next_if(|x| matches!(x, Ok(Token::Operator(Op::Arrow)))).is_some() { } else {
ret = Self::parse_type(tokens)?; Type::Any
} };
Ok((FunctionType(Box::new(ret), types), names)) Ok((FunctionType(Box::new(ret), types), names))
} }
fn parse_function_declaration_parameter(mut tokens: &mut Peekable<I>) -> Result<(Type, String), ParseError> { fn parse_function_declaration_parameter(
mut tokens: &mut Peekable<I>) -> Result<(Type, String), ParseError>
{
match tokens.next() { match tokens.next() {
// untyped variable // untyped variable
Some(Ok(Token::Identifier(x))) => Ok((Type::Any, x)), Some(Ok(Token::Identifier(x))) => Ok((Type::Any, x)),
@@ -412,7 +422,7 @@ impl<'a, I: Iterator<Item = Result<Token, TokenizeError>>> Parser<'a, I> {
let mut ret = Type::Any; let mut ret = Type::Any;
// this is annoying // this is annoying
// inside of the next_if closure, we already can know that its an error // inside the next_if closure, we already can know that its an error
// and return it, but we cannot return out of a closure // and return it, but we cannot return out of a closure
if let Some(t) = tokens.next_if(|x| matches!(x, Ok(Token::Operator(Op::Arrow)))) if let Some(t) = tokens.next_if(|x| matches!(x, Ok(Token::Operator(Op::Arrow))))
{ {
@@ -433,20 +443,16 @@ impl<'a, I: Iterator<Item = Result<Token, TokenizeError>>> Parser<'a, I> {
} }
} }
fn parse_type(tokens: &mut Peekable<I>) -> Result<Type, ParseError> { // for some dumbass reason,
// this is the only code that breaks if it doesn't take an impl Iterator instead of simply I ...
fn parse_type(tokens: &mut Peekable<impl Iterator<Item = Result<Token, TokenizeError>>>) -> Result<Type, ParseError> {
match tokens.next() { match tokens.next() {
Some(Ok(Token::Type(t))) => Ok(t), Some(Ok(Token::Type(t))) => Ok(t),
Some(Ok(Token::Operator(Op::FunctionDefine(n)))) => {
let args: Vec<Type> = (0..n)
.map(|_| Self::parse_type(tokens))
.collect::<Result<_, ParseError>>()?;
let rett = Self::parse_type(tokens)?;
Ok(Type::Function(FunctionType(Box::new(rett), args.clone())))
},
Some(Ok(Token::Operator(Op::OpenArray))) => { Some(Ok(Token::Operator(Op::OpenArray))) => {
let mut depth = 1; let mut depth = 1;
// take tokens until we reach the end of this array
// if we don't collect them here it causes rust to overflow computing the types
let array_tokens = tokens.by_ref().take_while(|t| match t { let array_tokens = tokens.by_ref().take_while(|t| match t {
Ok(Token::Operator(Op::OpenArray)) => { Ok(Token::Operator(Op::OpenArray)) => {
depth += 1; depth += 1;
@@ -459,18 +465,25 @@ impl<'a, I: Iterator<Item = Result<Token, TokenizeError>>> Parser<'a, I> {
_ => true, _ => true,
}).collect::<Result<Vec<_>, TokenizeError>>().map_err(|e| ParseError::TokenizeError(e))?; }).collect::<Result<Vec<_>, TokenizeError>>().map_err(|e| ParseError::TokenizeError(e))?;
if array_tokens.len() == 0 { // ... thanks to this conversion here. The compiler complains that the types don't
return Ok(Type::Array(Box::new(Type::Any))); // match. there is code elsewhere in this codebase that looks exactly like this and
} // still simply uses &mut Peekable<I> as the type. I don't understand why this code
// is special, but we have to do horribleness for it to work.
let mut array_tokens = array_tokens
.into_iter()
.map(|t| Ok(t))
.collect::<Vec<Result<Token, TokenizeError>>>()
.into_iter()
.peekable();
let t = Self::parse_type(tokens)?; let t = match Self::parse_type(&mut array_tokens) {
let _ = match tokens.next() { Ok(t) => t,
Some(Ok(Token::Operator(Op::CloseArray))) => (), Err(ParseError::UnexpectedEndInput) => Type::Any,
_ => return Err(ParseError::UnmatchedArrayClose), Err(e) => return Err(e),
}; };
Ok(Type::Array(Box::new(t))) Ok(Type::Array(Box::new(t)))
} },
Some(Ok(t)) => Err(ParseError::UnwantedToken(t.clone())), Some(Ok(t)) => Err(ParseError::UnwantedToken(t.clone())),
Some(Err(e)) => Err(ParseError::TokenizeError(e)), Some(Err(e)) => Err(ParseError::TokenizeError(e)),
None => Err(ParseError::UnexpectedEndInput), None => Err(ParseError::UnexpectedEndInput),

View File

@@ -419,15 +419,28 @@ impl<R: BufRead> std::iter::Iterator for Tokenizer<R> {
#[cfg(test)] #[cfg(test)]
mod tests { mod tests {
use std::str::FromStr; use std::str::FromStr;
use crate::parser::Parser;
use super::*; use super::*;
#[test] #[test]
fn a() { fn tokenizer() {
let program = ":. map : f .? x [Any] -> [Any]"; let program = ": length ?. x [] -> Int ?? x + 1 length tail x 0 length [ 1 2 3 ]";
let tokens: Vec<Token> = Tokenizer::from_str(program).unwrap().collect::<Result<_, _>>().unwrap(); let tokens: Vec<Token> = Tokenizer::from_str(program).unwrap().collect::<Result<_, _>>().unwrap();
println!("{tokens:?}"); println!("{tokens:#?}");
}
#[test]
fn a() {
let program = ": length ?. x [] -> Int ?? x + 1 length tail x 0 length [ 1 2 3 ]";
let mut tokenizer = Tokenizer::from_str(program).unwrap().peekable();
let mut globals = HashMap::new();
let mut parser = Parser::new(&mut tokenizer, &mut globals);
let tree = parser.next();
println!("{tree:#?}");
} }
} }