Add AST parser skeleton

3 files changed, 221 insertions, 0 deletions

diff --git a/compiler/src/ast/error.rs b/compiler/src/ast/error.rs
new file mode 100644
index 0000000..11f552d
--- /dev/null
+++ b/compiler/src/ast/error.rs
@@ -0,0 +1,12 @@
+use std::{error, fmt};
+
+#[derive(Debug)]
+pub enum Error {}
+
+impl fmt::Display for Error {
+    fn fmt(&self, _f: &mut fmt::Formatter) -> fmt::Result {
+        todo!()
+    }
+}
+
+impl error::Error for Error {}
diff --git a/compiler/src/ast/mod.rs b/compiler/src/ast/mod.rs
new file mode 100644
index 0000000..4587ea7
--- /dev/null
+++ b/compiler/src/ast/mod.rs
@@ -0,0 +1,27 @@
+mod error;
+mod parser;
+
+use std::rc::Rc;
+
+use crate::span::Spanned;
+pub use error::Error;
+pub use parser::Parser;
+
+#[derive(Clone, Debug)]
+pub enum Atom {
+    Float(f64),
+    Integer(i64),
+    String(Rc<str>),
+    Symbol(Rc<str>),
+    Bool(bool),
+    Nil,
+}
+
+#[derive(Clone, Debug)]
+pub enum Expr {
+    Atom(Atom),
+    List(Vec<Spanned<Expr>>),
+}
+
+#[derive(Clone, Debug)]
+pub struct Program(pub Vec<Spanned<Expr>>);
diff --git a/compiler/src/ast/parser.rs b/compiler/src/ast/parser.rs
new file mode 100644
index 0000000..171ecfe
--- /dev/null
+++ b/compiler/src/ast/parser.rs
@@ -0,0 +1,182 @@
+use std::{iter::Peekable, result};
+
+use crate::{
+    ast::{Atom, Error, Expr, Program},
+    lexer::Token,
+    span::{Pos, Span, Spanned},
+};
+
+type Result<T> = result::Result<T, Spanned<Error>>;
+
+fn parse_number(number: &str) -> Atom {
+    let is_float = number.bytes().any(|b| matches!(b, b'.' | b'e' | b'E'));
+
+    if is_float {
+        match number.parse() {
+            Ok(ok) => Atom::Float(ok),
+            Err(err) => todo!("invalid float literal {number}: {err}"),
+        }
+    } else {
+        match number.parse() {
+            Ok(ok) => Atom::Integer(ok),
+            Err(err) => todo!("invalid integer literal {number}: {err}"),
+        }
+    }
+}
+
+fn parse_string(string: &str) -> Atom {
+    let mut result = String::new();
+    let mut is_escape = false;
+
+    for ch in string.chars() {
+        if !is_escape {
+            match ch {
+                '\\' => is_escape = true,
+                _ => result.push(ch),
+            }
+        } else {
+            match ch {
+                '"' => result.push('"'),
+                'n' => result.push('\n'),
+                '\\' => result.push('\\'),
+                '\n' => {}
+                _ => todo!("unexpected escape char {ch:?}"),
+            }
+            is_escape = false;
+        }
+    }
+
+    if is_escape {
+        todo!("unclosed string");
+    }
+
+    Atom::String(result.into())
+}
+
+fn parse_symbol(symbol: &str) -> Atom {
+    match symbol {
+        "true" => Atom::Bool(true),
+        "false" => Atom::Bool(false),
+        "nil" => Atom::Nil,
+        _ => Atom::Symbol(symbol.into()),
+    }
+}
+
+pub struct Parser<'a, I>
+where
+    I: Iterator<Item = Spanned<Token<'a>>>,
+{
+    tokens: Peekable<I>,
+    cursor: Pos,
+}
+
+impl<'a, I> Parser<'a, I>
+where
+    I: Iterator<Item = Spanned<Token<'a>>>,
+{
+    pub fn new(tokens: I) -> Self {
+        Self {
+            tokens: tokens.peekable(),
+            cursor: Pos::new(1, 0, 0),
+        }
+    }
+
+    fn peek(&mut self) -> Option<Spanned<Token<'a>>> {
+        self.tokens.peek().copied()
+    }
+
+    fn consume(&mut self) -> Option<Spanned<Token<'a>>> {
+        self.tokens.next().inspect(|s| self.cursor = s.span.end)
+    }
+
+    fn parse_expr(&mut self) -> Result<Spanned<Expr>> {
+        let Spanned { inner: token, span } = match self.peek() {
+            Some(spanned) => spanned,
+            None => todo!("unexpected eof"),
+        };
+
+        let expr = match token {
+            Token::LeftPar => {
+                self.consume();
+                let list = self.parse_list()?;
+                let expr = if !list.is_empty() {
+                    Expr::List(list)
+                } else {
+                    Expr::Atom(Atom::Nil)
+                };
+
+                Spanned::new(expr, Span::new(span.start, self.cursor))
+            }
+            Token::RightPar => todo!("unexpected par"),
+            Token::Quote => {
+                self.consume();
+                let quote = Spanned::new(
+                    Expr::Atom(Atom::Symbol("quote".into())),
+                    Span::new(span.start, self.cursor),
+                );
+                let expr = self.parse_expr()?;
+
+                Spanned::new(
+                    Expr::List(vec![quote, expr]),
+                    Span::new(span.start, self.cursor),
+                )
+            }
+            Token::Number(number) => {
+                self.consume();
+
+                Spanned::new(
+                    Expr::Atom(parse_number(number)),
+                    Span::new(span.start, self.cursor),
+                )
+            }
+            Token::String(string) => {
+                self.consume();
+
+                Spanned::new(
+                    Expr::Atom(parse_string(string)),
+                    Span::new(span.start, self.cursor),
+                )
+            }
+            Token::UnclosedString(string) => {
+                self.consume();
+                todo!("unclosed string {string:?}")
+            }
+            Token::Symbol(symbol) => {
+                self.consume();
+
+                Spanned::new(
+                    Expr::Atom(parse_symbol(symbol)),
+                    Span::new(span.start, self.cursor),
+                )
+            }
+        };
+
+        Ok(expr)
+    }
+
+    fn parse_list(&mut self) -> Result<Vec<Spanned<Expr>>> {
+        let mut list = Vec::new();
+
+        while let Some(Spanned { inner: token, .. }) = self.peek() {
+            match token {
+                Token::RightPar => {
+                    self.consume();
+                    return Ok(list);
+                }
+                _ => list.push(self.parse_expr()?),
+            }
+        }
+
+        todo!("unclosed par")
+    }
+
+    pub fn parse(mut self) -> Result<Program> {
+        let mut program = Vec::new();
+
+        while self.peek().is_some() {
+            program.push(self.parse_expr()?)
+        }
+
+        Ok(Program(program))
+    }
+}