plonkus/src/parse.rs

288 lines
8.3 KiB
Rust
Raw Normal View History

2023-09-16 22:21:53 +02:00
use std::collections::{HashMap, HashSet};
use crate::ast::*;
use crate::tokeniser::Keyword;
use crate::tokeniser::Token;
fn parse_math(tokens: &Vec<Token>) -> ValueNode {
let mut prioritizing_order: HashMap<i64, HashSet<Token>> = HashMap::new();
prioritizing_order.insert(0, HashSet::from([Token::Plus, Token::Minus]));
let mut ordered_binops: Vec<Token> = Vec::new();
let mut ordered_binops_indexes: Vec<usize> = Vec::new();
if tokens.len() == 1 {
return match &tokens[0] {
Token::Number(n) => ValueNode::Literal(Literal::Int(n.parse::<i64>().unwrap())),
2023-09-17 16:27:06 +02:00
Token::Identifier(identifier) => ValueNode::Variable(identifier.to_owned()),
2023-09-16 22:21:53 +02:00
_ => todo!(),
};
}
for j in prioritizing_order.keys() {
let mut i = 0;
let current_order_binops = prioritizing_order.get(j).unwrap();
while i < tokens.len() {
for binop in current_order_binops {
if &tokens[i] == binop {
ordered_binops.push(binop.to_owned());
ordered_binops_indexes.push(i);
}
}
i += 1;
}
}
let binop = ordered_binops.pop().unwrap();
let binop_index = ordered_binops_indexes.pop().unwrap();
let mut a: Vec<Token> = Vec::new();
let mut b: Vec<Token> = Vec::new();
let mut i = 0;
while i < tokens.len() {
if i < binop_index {
a.push(tokens[i].clone());
}
if i > binop_index {
b.push(tokens[i].clone());
}
i += 1;
}
return match binop {
Token::Plus => ValueNode::Operator(Operator::Add(
Box::new(parse_math(&a)),
Box::new(parse_math(&b)),
)),
Token::Minus => ValueNode::Operator(Operator::Sub(
Box::new(parse_math(&a)),
Box::new(parse_math(&b)),
)),
_ => todo!(),
};
}
fn parse_basic(tokens: &Vec<Token>, i: &mut usize) -> Expression {
let literal = match &tokens[*i] {
Token::Identifier(identifier) => Expression::Variable(identifier.to_owned()),
Token::Number(number) => Expression::Literal(Literal::Int(number.parse::<i64>().unwrap())),
_ => todo!(),
};
*i += 1;
literal
}
fn parse_grouping(tokens: &Vec<Token>, i: &mut usize) -> Expression {
match &tokens[*i] {
Token::OpenParanthesis => {
*i += 1;
}
_ => todo!(),
};
let inner_expression = parse_expression(tokens, i);
match &tokens[*i] {
Token::CloseParanthesis => {}
_ => todo!(),
};
inner_expression
}
// TODO: replace parse_math with this recursive function
pub fn parse_expression(tokens: &Vec<Token>, i: &mut usize) -> Expression {
#[derive(Debug)]
enum OpExpr {
Expression(Expression),
BinOp(BinOp),
Unary(Unary),
}
let mut in_expressions_and_operators: Vec<OpExpr> = Vec::new();
loop {
in_expressions_and_operators.push(match &tokens[*i] {
Token::OpenParanthesis => OpExpr::Expression(parse_grouping(tokens, i)),
Token::Plus => {
*i += 1;
OpExpr::BinOp(BinOp::Plus)
}
Token::Minus => {
*i += 1;
OpExpr::BinOp(BinOp::Minus)
}
Token::Ampersand => {
*i += 1;
OpExpr::Unary(Unary::ValueFrom)
}
Token::Star => {
*i += 1;
OpExpr::Unary(Unary::PointerTo)
}
Token::Number(_) => OpExpr::Expression(parse_basic(tokens, i)),
_other => break,
});
}
dbg!(in_expressions_and_operators);
Expression::Variable("a".to_string())
}
fn parse_statements(tokens: &Vec<Token>, i: &mut usize) -> Vec<Statement> {
dbg!(&i);
2023-09-17 21:52:14 +02:00
let mut statements = Vec::new();
while *i < tokens.len() {
statements.push(match &tokens[*i] {
2023-09-17 21:52:14 +02:00
Token::OpenSquiglyBracket => {
*i += 1;
let statements = parse_statements(tokens, i);
2023-09-17 21:52:14 +02:00
Statement::Scope(statements)
2023-09-17 16:27:06 +02:00
}
2023-09-17 21:52:14 +02:00
Token::Keyword(Keyword::Let) => {
*i += 1;
let name = match tokens[*i].clone() {
2023-09-17 21:52:14 +02:00
Token::Identifier(identifier) => identifier,
_ => todo!(),
};
*i += 1;
match tokens[*i].clone() {
2023-09-17 21:52:14 +02:00
Token::Equals => {}
_ => todo!(),
};
2023-09-16 22:21:53 +02:00
*i += 1;
2023-09-16 22:21:53 +02:00
2023-09-17 21:52:14 +02:00
let mut declaration_tokens: Vec<Token> = Vec::new();
while tokens[*i] != Token::Semi {
declaration_tokens.push(tokens[*i].clone());
*i += 1;
2023-09-16 22:21:53 +02:00
}
2023-09-17 21:52:14 +02:00
Statement::VariableDeclaration(name, parse_math(&declaration_tokens))
2023-09-16 22:21:53 +02:00
}
2023-09-17 21:52:14 +02:00
Token::Keyword(Keyword::Assign) => {
*i += 1;
let name = match tokens[*i].clone() {
2023-09-17 21:52:14 +02:00
Token::Identifier(identifier) => identifier,
_ => todo!(),
};
2023-09-16 22:21:53 +02:00
*i += 1;
match tokens[*i].clone() {
2023-09-17 21:52:14 +02:00
Token::Equals => {}
_ => todo!(),
};
2023-09-16 22:21:53 +02:00
*i += 1;
2023-09-16 22:21:53 +02:00
2023-09-17 21:52:14 +02:00
let mut declaration_tokens: Vec<Token> = Vec::new();
while tokens[*i] != Token::Semi {
declaration_tokens.push(tokens[*i].clone());
*i += 1;
2023-09-16 22:21:53 +02:00
}
2023-09-17 21:52:14 +02:00
Statement::VariableAssignment(name, parse_math(&declaration_tokens))
2023-09-16 22:21:53 +02:00
}
2023-09-17 21:52:14 +02:00
Token::Identifier(identifier) => {
*i += 1;
2023-09-17 21:52:14 +02:00
let mut declaration_tokens: Vec<Vec<Token>> = Vec::new();
*i += 1;
while tokens[*i] != Token::CloseParanthesis {
2023-09-17 21:52:14 +02:00
let mut argument_tokens: Vec<Token> = Vec::new();
while tokens[*i] != Token::Comma {
argument_tokens.push(tokens[*i].clone());
*i += 1;
2023-09-17 21:52:14 +02:00
}
declaration_tokens.push(argument_tokens);
*i += 1;
2023-09-17 21:52:14 +02:00
}
2023-09-16 22:21:53 +02:00
*i += 1;
2023-09-17 21:52:14 +02:00
let mut arguments: Vec<ValueNode> = Vec::new();
2023-09-16 22:21:53 +02:00
2023-09-17 21:52:14 +02:00
for token_vec in &declaration_tokens {
if token_vec == &Vec::new() {
continue;
}
arguments.push(parse_math(token_vec));
}
2023-09-16 22:21:53 +02:00
2023-09-17 21:52:14 +02:00
Statement::FunctionCall(identifier.to_owned(), arguments)
}
Token::CloseSquiglyBracket => {
*i += 1;
return statements;
2023-09-17 21:52:14 +02:00
}
Token::Semi => {
*i += 1;
2023-09-17 21:52:14 +02:00
continue;
}
_other => {
dbg!(_other);
dbg!(i);
2023-09-17 21:52:14 +02:00
todo!()
}
});
}
*i += 1;
statements
}
pub fn parse_arguments(tokens: &Vec<Token>, i: &mut usize) -> Vec<String> {
let mut argument_variables = Vec::new();
while tokens[*i] != Token::OpenSquiglyBracket {
match &tokens[*i] {
Token::Identifier(identifier) => argument_variables.push(identifier.to_owned()),
Token::OpenSquiglyBracket => {}
_other => todo!(),
}
*i += 1;
}
argument_variables
2023-09-16 22:21:53 +02:00
}
pub fn ast(tokens: Vec<Token>) -> Vec<Function> {
let mut functions: Vec<Function> = Vec::new();
let mut i = 0;
while i < tokens.len() {
2023-09-17 21:52:14 +02:00
match &tokens[i] {
Token::Keyword(Keyword::Function) => {
i += 1;
let name = match &tokens[i] {
Token::Identifier(identifier) => identifier,
2023-09-17 21:52:14 +02:00
_ => todo!(),
}
.to_owned();
i += 1;
let arguments = parse_arguments(&tokens, &mut i);
i += 1;
let statements = parse_statements(&tokens, &mut i);
let statement = Statement::Scope(statements);
functions.push(Function::new(name, arguments, statement));
2023-09-16 22:21:53 +02:00
}
2023-09-17 21:52:14 +02:00
_other => {
dbg!(&functions);
dbg!(i);
dbg!(_other);
todo!();
2023-09-16 22:21:53 +02:00
}
};
i += 1;
}
2023-09-17 21:52:14 +02:00
functions
2023-09-16 22:21:53 +02:00
}