From 15725c024f9cbebaeeeb9599d7a9956afa707b34 Mon Sep 17 00:00:00 2001
From: Stephen D <webmaster@scd31.com>
Date: Sun, 9 Mar 2025 11:40:11 -0400
Subject: [PATCH] parser
---
src/expr.rs | 8 ++
src/main.rs | 15 +++-
src/parser.rs | 194 +++++++++++++++++++++++++++++++++++++++++++++++++
src/scanner.rs | 11 +--
src/token.rs | 5 +-
5 files changed, 221 insertions(+), 12 deletions(-)
create mode 100644 src/expr.rs
create mode 100644 src/parser.rs
diff --git a/src/expr.rs b/src/expr.rs
new file mode 100644
index 0000000..a8a8a8d
--- /dev/null
+++ b/src/expr.rs
@@ -0,0 +1,8 @@
+#[derive(Debug)]
+pub enum Expr {
+ Lambda(Box<Expr>),
+ Abs(Box<Expr>),
+ App(Box<Expr>, Box<Expr>),
+ Var(usize),
+ DefinitionName(String),
+}
diff --git a/src/main.rs b/src/main.rs
index 900fb56..4de6c3e 100644
--- a/src/main.rs
+++ b/src/main.rs
@@ -1,5 +1,8 @@
+use parser::Parser;
use scanner::Scanner;
+mod expr;
+mod parser;
mod scanner;
mod token;
@@ -14,7 +17,15 @@ $true = λtf.t
",
);
- for t in scanner.scan_tokens().unwrap() {
- dbg!(t.token_type);
+ let tokens = scanner.scan_tokens().unwrap();
+
+ for t in &tokens {
+ println!("{} {:?}", t.line, t.token_type);
}
+
+ let mut parser = Parser::new(tokens);
+ let (funcs, main) = parser.parse().unwrap();
+
+ dbg!(funcs);
+ dbg!(main);
}
diff --git a/src/parser.rs b/src/parser.rs
new file mode 100644
index 0000000..8dd05f8
--- /dev/null
+++ b/src/parser.rs
@@ -0,0 +1,194 @@
+use std::collections::HashMap;
+
+use crate::{
+ expr::Expr,
+ token::{Token, TokenType},
+};
+
+pub struct Parser {
+ tokens: Vec<Token>,
+ current: usize,
+ env: Vec<char>,
+}
+
+impl Parser {
+ pub fn new(tokens: Vec<Token>) -> Self {
+ Self {
+ tokens,
+ current: 0,
+ env: vec![],
+ }
+ }
+
+ pub fn parse(&mut self) -> Option<(HashMap<String, Expr>, Expr)> {
+ let mut definitions = HashMap::new();
+ let mut main_expr = None;
+ while !self.is_at_end() {
+ let t = self.peek().token_type.clone();
+
+ match t {
+ TokenType::DefinitionName(name) => {
+ self.advance(); // consume definition name
+
+ // function expr
+
+ if self.peek().token_type != TokenType::Equals {
+ self.print_error("Expected '=' after definition name");
+
+ return None;
+ }
+
+ self.advance(); // consume equals
+
+ let definition = self.expr()?;
+
+ if definitions.contains_key(&name) {
+ self.print_error(&format!("Duplicate definition for ${}", name));
+ return None;
+ }
+
+ definitions.insert(name, definition);
+ }
+
+ // main expr
+ _ => {
+ let expr = self.expr()?;
+
+ if main_expr.is_some() {
+ self.print_error("Duplicate primary expression");
+ return None;
+ }
+
+ main_expr = Some(expr);
+ }
+ }
+ }
+
+ match main_expr {
+ Some(expr) => Some((definitions, expr)),
+ None => {
+ self.print_error("Missing primary expression");
+
+ None
+ }
+ }
+ }
+
+ fn expr(&mut self) -> Option<Expr> {
+ match &self.peek().token_type {
+ TokenType::Lambda => self.abstraction(),
+
+ TokenType::DefinitionName(name) => {
+ let name = name.clone();
+ self.advance();
+
+ Some(Expr::DefinitionName(name))
+ }
+
+ TokenType::LeftParen => self.application(),
+
+ TokenType::Variable(c) => {
+ let c = *c;
+ self.advance();
+
+ Some(Expr::Var(self.find_variable(c)?))
+ }
+
+ _ => {
+ self.print_error(
+ "Unexpected token. Expected one of 'λ', '/', '\\', '$', '(', or a variable",
+ );
+ None
+ }
+ }
+ }
+
+ fn abstraction(&mut self) -> Option<Expr> {
+ self.advance(); // consume lambda
+
+ let mut abs_count = 0;
+ while let TokenType::Variable(v) = self.peek().token_type {
+ self.env.push(v);
+ abs_count += 1;
+ self.advance();
+ }
+
+ if self.advance().token_type != TokenType::Period {
+ self.print_error("Expected '.' after abstraction variable definitions");
+ return None;
+ }
+
+ let mut expr = self.expr()?;
+
+ for _ in 0..abs_count {
+ self.env.pop().unwrap();
+ expr = Expr::Lambda(Box::new(expr));
+ }
+
+ Some(expr)
+ }
+
+ fn application(&mut self) -> Option<Expr> {
+ self.advance(); // consume left paren
+
+ let left = self.expr()?;
+ let right = self.expr()?;
+
+ dbg!(&left, &right);
+ dbg!(&self.peek().token_type);
+
+ if self.peek().token_type != TokenType::RightParen {
+ self.print_error("Expected ')' after application");
+ return None;
+ }
+
+ self.advance(); // consume right paren
+
+ Some(Expr::App(Box::new(left), Box::new(right)))
+ }
+
+ fn find_variable(&self, name: char) -> Option<usize> {
+ let ret = self
+ .env
+ .iter()
+ .enumerate()
+ .find(|(_, x)| **x == name)
+ .map(|(i, _)| i);
+
+ match ret {
+ Some(x) => Some(x),
+ None => {
+ self.print_error(&format!("Unknown variable '{}'.", name));
+
+ None
+ }
+ }
+ }
+
+ fn peek(&self) -> &Token {
+ &self.tokens[self.current]
+ }
+
+ fn peek_next(&self) -> &TokenType {
+ if self.current + 1 == self.tokens.len() {
+ &TokenType::NoToken
+ } else {
+ &self.tokens[self.current + 1].token_type
+ }
+ }
+
+ fn advance(&mut self) -> &Token {
+ let t = &self.tokens[self.current];
+ self.current += 1;
+
+ t
+ }
+
+ fn is_at_end(&self) -> bool {
+ self.tokens.len() == self.current
+ }
+
+ fn print_error(&self, err: &str) {
+ eprintln!("Error on line [{}]: {err}", self.peek().line);
+ }
+}
diff --git a/src/scanner.rs b/src/scanner.rs
index d5fced3..20fc7b3 100644
--- a/src/scanner.rs
+++ b/src/scanner.rs
@@ -26,9 +26,6 @@ impl Scanner {
self.scan_token()?;
}
- self.tokens
- .push(Token::new(TokenType::Eof, "".to_string(), self.line));
-
Ok(self.tokens)
}
@@ -45,7 +42,7 @@ impl Scanner {
'\\' | '/' | 'λ' => Lambda,
_ if c.is_alphabetic() => Variable(c),
'$' => {
- self.function()?;
+ self.definition()?;
NoToken
}
@@ -59,7 +56,7 @@ impl Scanner {
}
_ => {
- self.print_error(&format!("Unexpected character `{}`.", c));
+ self.print_error(&format!("Unexpected character '{}'.", c));
return Err(());
}
@@ -70,7 +67,7 @@ impl Scanner {
Ok(())
}
- fn function(&mut self) -> Result<(), ()> {
+ fn definition(&mut self) -> Result<(), ()> {
while self.peek().is_alphanumeric() {
self.advance();
}
@@ -81,7 +78,7 @@ impl Scanner {
return Err(());
}
- self.add_token(TokenType::Function(
+ self.add_token(TokenType::DefinitionName(
self.chars[(self.start + 1)..self.current].iter().collect(),
));
diff --git a/src/token.rs b/src/token.rs
index b02bbbf..8e2cd83 100644
--- a/src/token.rs
+++ b/src/token.rs
@@ -1,12 +1,11 @@
-#[derive(PartialEq, Eq, Debug)]
+#[derive(Clone, PartialEq, Eq, Debug)]
pub enum TokenType {
Variable(char),
- Function(String),
+ DefinitionName(String),
Lambda,
Period,
Equals,
- Eof,
LeftParen,
RightParen,
--
GitLab