holy_lisp_archive/src/parser.rs

use std::{collections::HashMap, fmt};

#[cfg(test)]
mod tests {
	use crate::{Parser, PreLispValue, Tokenizer, TokenizerError};

	fn quick_parse(source: &str) -> Result<Vec<PreLispValue>, TokenizerError> {
		let mut tokenizer = Tokenizer::new(String::from(source));
		let tokens = tokenizer.tokenize()?;
		let mut parser = Parser::new(tokens);
		Ok(parser.parse())
	}

	fn var(name: &str) -> PreLispValue {
		PreLispValue::Var(String::from(name))
	}
	fn string(s: &str) -> PreLispValue {
		PreLispValue::String(String::from(s))
	}
	fn list(v: Vec<PreLispValue>, quoted: bool) -> PreLispValue {
		PreLispValue::List(v, quoted)
	}
	fn int(n: i32) -> PreLispValue {
		PreLispValue::Integer(n)
	}
	fn float(n: f32) -> PreLispValue {
		PreLispValue::Float(n)
	}

	#[test]
	fn test_hello_world() -> Result<(), TokenizerError> {
		assert_eq!(
			vec![list(vec![var("print"), string("Hello, World!")], false)],
			quick_parse(
				"
				(print \"Hello, World!\")
				"
			)?
		);
		Ok(())
	}

	#[test]
	fn test_math() -> Result<(), TokenizerError> {
		assert_eq!(
			vec![list(
				vec![
					var("print"),
					list(
						vec![
							var("add"),
							list(
								vec![
									var("div"),
									float(4.5),
									list(vec![var("sub"), float(0.5), float(0.2)], false)
								],
								false
							),
							list(vec![var("mul"), int(21), float(0.05)], false)
						],
						false
					)
				],
				false
			)],
			quick_parse(
				"
				(print 
					(add 
						(div
							4.5
							(sub 0.5 0.2)
						)
						(mul 21 0.05)
					)
				)
			"
			)?
		);
		Ok(())
	}
}

use crate::Token;

#[derive(Debug, Clone, PartialEq)]
pub enum PreLispValue {
	Nil,
	True,
	False,
	String(String),
	Var(String),
	Integer(i32),
	Float(f32),
	List(Vec<PreLispValue>, bool), // Values, Quoted
	LispFunction(String, Vec<String>, Vec<PreLispValue>),
}

impl fmt::Display for PreLispValue {
	fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
		match self {
			PreLispValue::Nil => write!(f, "nil"),
			PreLispValue::True => write!(f, "true"),
			PreLispValue::False => write!(f, "false"),
			PreLispValue::String(str) => write!(f, "{}", str),
			PreLispValue::LispFunction(name, _, _) => write!(f, "<function {}>", name),
			PreLispValue::Integer(num) => write!(f, "{}", num),
			PreLispValue::Float(num) => write!(f, "{}", num),
			_ => todo!(),
		}
	}
}

impl From<&str> for PreLispValue {
	fn from(value: &str) -> Self {
		PreLispValue::String(value.to_string())
	}
}
impl From<String> for PreLispValue {
	fn from(value: String) -> Self {
		PreLispValue::String(value)
	}
}
impl From<i32> for PreLispValue {
	fn from(value: i32) -> Self {
		PreLispValue::Integer(value)
	}
}
impl From<f32> for PreLispValue {
	fn from(value: f32) -> Self {
		PreLispValue::Float(value)
	}
}
impl From<Vec<PreLispValue>> for PreLispValue {
	fn from(value: Vec<PreLispValue>) -> Self {
		PreLispValue::List(value, false)
	}
}

// TODO: Handle failure
fn read_exp(open_paren_idx: usize, tokens: &Vec<Token>) -> Option<(Vec<Token>, usize)> /* Option<tokens, close_paren_idx> */
{
	let mut tkns = Vec::new();
	let mut depth = 0;

	for i in open_paren_idx..tokens.len() {
		match &tokens[i] {
			Token::OpenParen => {
				if depth != 0 {
					tkns.push(Token::OpenParen);
				}
				depth += 1;
			}
			Token::CloseParen => {
				depth -= 1;
				if depth == 0 {
					return Some((tkns, i));
				} else {
					tkns.push(Token::CloseParen)
				}
			}
			token => {
				tkns.push(token.clone());
			}
		}
	}

	None
}

fn parse_exp(tokens: Vec<Token>, quoted: bool) -> PreLispValue {
	let mut values = Vec::new();

	if tokens.len() < 1 {
		return PreLispValue::List(values, false);
	}

	let mut quote_next_list = false;
	let mut unquote_next_list = false;

	let mut i = 0;
	while i < tokens.len() {
		let tkn = &tokens[i];
		match tkn {
			Token::Quote => {
				quote_next_list = true;
			}
			Token::Unquote => {
				unquote_next_list = true;
			}
			Token::OpenParen => {
				let (tkns, close_paren_idx) = read_exp(i, &tokens).unwrap();
				values.push(parse_exp(
					tkns,
					if unquote_next_list {
						false
					} else {
						quote_next_list || quoted
					},
				));
				i = close_paren_idx;
				quote_next_list = false;
				unquote_next_list = false;
			}
			Token::Identifier(name) => match name.as_str() {
				"nil" => values.push(PreLispValue::Nil),
				"true" => values.push(PreLispValue::True),
				"false" => values.push(PreLispValue::False),
				_ => values.push(PreLispValue::Var(name.to_owned())),
			},
			Token::Index(map, idx) => {
				values.push(PreLispValue::List(
					vec![
						PreLispValue::Var(String::from("get")),
						PreLispValue::Var(map.to_owned()),
						PreLispValue::String(idx.to_owned()),
					],
					false,
				));
			}
			Token::Integer(num) => {
				values.push(PreLispValue::Integer(*num));
			}
			Token::Float(num) => {
				values.push(PreLispValue::Float(*num));
			}
			Token::String(str) => {
				values.push(PreLispValue::String(str.to_owned()));
			}
			Token::CloseParen => {
				panic!("Unexpected closing parenthesis");
			}
		}
		i += 1;
	}

	PreLispValue::List(values, quoted)
}

pub struct Parser {
	tokens: Vec<Token>,
	macros: HashMap<String, fn(Vec<PreLispValue>) -> Vec<PreLispValue>>,
}

fn add_macro(
	macros: &mut HashMap<String, fn(Vec<PreLispValue>) -> Vec<PreLispValue>>,
	name: impl Into<String>,
	mac: fn(Vec<PreLispValue>) -> Vec<PreLispValue>,
) {
	macros.insert(name.into(), mac);
}

impl Parser {
	pub fn new(tokens: Vec<Token>) -> Parser {
		let mut macros = HashMap::new();

		add_macro(&mut macros, "set", |x| {
			if x.len() != 3
				|| (!matches!(x[1], PreLispValue::Var(_))
					&& !matches!(x[1], PreLispValue::String(_)))
			{
				return x;
			}

			let mut list = x.clone();

			if let PreLispValue::Var(name) = &list[1] {
				list[1] = PreLispValue::String(name.to_owned())
			}

			list
		});

		add_macro(&mut macros, "fn", |x| {
			if x.len() < 3 || x.len() > 4 {
				return x;
			}

			let mut name: String = String::new();
			let mut args: &Vec<PreLispValue> = &Vec::new();
			let mut list: &Vec<PreLispValue> = &Vec::new();

			if x.len() == 3 {
				if !matches!(x[1], PreLispValue::List(_, _))
					|| !matches!(x[2], PreLispValue::List(_, _))
				{
					return x;
				}
				name = String::from("Anonymous");
				if let PreLispValue::List(l, _) = &x[1] {
					args = l;
				}
				if let PreLispValue::List(l, _) = &x[2] {
					list = l;
				}
			} else if x.len() == 4 {
				if !matches!(x[1], PreLispValue::Var(_))
					|| !matches!(x[2], PreLispValue::List(_, _))
					|| !matches!(x[3], PreLispValue::List(_, _))
				{
					return x;
				}
				if let PreLispValue::Var(n) = &x[1] {
					name = n.to_owned();
				}
				if let PreLispValue::List(l, _) = &x[2] {
					args = l;
				}
				if let PreLispValue::List(l, _) = &x[3] {
					list = l;
				}
			} else {
				return x;
			}

			let mut arg_names = Vec::new();
			for arg in args {
				if let PreLispValue::Var(n) = arg {
					arg_names.push(n.to_owned());
				}
			}

			let func = PreLispValue::LispFunction(name.clone(), arg_names, list.to_owned());

			if x.len() == 4 {
				return vec![PreLispValue::Var(String::from("set")), name.into(), func];
			}

			return vec![PreLispValue::Var(String::from("echo")), func];
		});

		Parser { tokens, macros }
	}

	fn macro_check(&mut self, input: Vec<PreLispValue>, quoted: bool) -> Vec<PreLispValue> {
		if quoted || input.len() < 1 {
			return input;
		}

		let mut output = input.clone();

		for i in 0..output.len() {
			if let PreLispValue::List(elems, quoted) = &output[i] {
				output[i] = PreLispValue::List(self.macro_check(elems.to_vec(), *quoted), *quoted);
			}
		}

		if let PreLispValue::Var(name) = &output[0] {
			if self.macros.contains_key(name) {
				return self.macros.get(name).unwrap()(output);
			}
		}

		output
	}

	pub fn parse(&mut self) -> Vec<PreLispValue> {
		let mut values = Vec::new();

		let mut i = 0;
		while i < self.tokens.len() {
			match &self.tokens[i] {
				Token::OpenParen => {
					let (tkns, close_paren_idx) = read_exp(i, &self.tokens).unwrap();
					values.push(parse_exp(tkns, false));
					i = close_paren_idx;
				}
				tkn => {
					// TODO: Include more error data
					panic!("Unexpected token {:?}", tkn);
				}
			}
			i += 1;
		}

		self.macro_check(values, false)
	}
}