begin work on parser

1 files changed, 253 insertions, 0 deletions

diff --git a/src/lisp/parse.rs b/src/lisp/parse.rs
new file mode 100644
index 0000000..13a9eff
--- /dev/null
+++ b/src/lisp/parse.rs
@@ -0,0 +1,253 @@
+use crate::lisp::{
+    error::LispError,
+    lex::{Lexer, Span, Token},
+    number::LispNumber,
+    LispExpr,
+};
+
+pub struct Parser<'lex> {
+    lexer: Lexer<'lex>,
+    cur_token: Option<(Span, Token<'lex>)>,
+}
+
+// pub struct ParseError {
+//     pub span: Span,
+//     pub kind: ParseErrorKind,
+// }
+//
+// pub enum ParseErrorKind {
+//     InvalidLiteral,
+//     InvalidToken,
+//     LiteralParseError,
+//     MissingCloseParen,
+//     UnbalancedComma,
+//     UnexpectedEof,
+//     UnexpectedToken {
+//         expected: &'static str,
+//         found: &'static str,
+//     },
+//     UnmatchedParen,
+//     UnterminatedString,
+// }
+
+enum Group {
+    Backticks(i32),
+    CommaAt,
+    Quotes(u32),
+    Parens(Vec<LispExpr>),
+}
+
+impl<'lex> Parser<'lex> {
+    pub fn new(lexer: Lexer<'lex>) -> Self {
+        Self {
+            lexer,
+            cur_token: None,
+        }
+    }
+
+    pub fn parse_expr(&mut self) -> Result<LispExpr, LispError> {
+        let mut stack = Vec::new();
+        let mut total_backticks = 0;
+        loop {
+            let (span, token) = self.next()?;
+            let r: Result<LispExpr, LispError> = match token {
+                Token::LeftParen => {
+                    stack.push(Group::Parens(Vec::new()));
+                    continue;
+                }
+                Token::RightParen => {
+                    let group = stack.pop().ok_or_else(
+                        || LispError::ParseError, // unmatched paren here
+                    )?;
+                    match group {
+                        Group::Parens(v) => Ok(LispExpr::List(v)),
+                        _ => Err(LispError::ParseError),
+                    }
+                }
+                Token::Float(f) => f
+                    .parse::<f64>()
+                    .map(|n| LispExpr::Number(LispNumber::Float(n)))
+                    .map_err(|_| LispError::ParseError),
+                Token::Integer(i) => i
+                    .parse::<i64>()
+                    .map(|n| LispExpr::Number(LispNumber::Integer(n)))
+                    .map_err(|_| LispError::ParseError),
+                Token::String(s) => Ok(LispExpr::StringLit(s.into())),
+                Token::Name(n) => Ok(name_expr(n)),
+                Token::BackQuote => {
+                    total_backticks += 1;
+                    if let Some(&mut Group::Backticks(ref mut n)) = stack.last_mut() {
+                        *n += 1;
+                        continue;
+                    }
+                    stack.push(Group::Backticks(1));
+                    continue;
+                }
+                Token::Comma => {
+                    if total_backticks <= 0 {
+                        return Err(LispError::ParseError); // unbalanced comma
+                    }
+                    total_backticks -= 1;
+                    if let Some(&mut Group::Backticks(ref mut n)) = stack.last_mut() {
+                        *n -= 1;
+                        continue;
+                    }
+                    stack.push(Group::Backticks(-1));
+                    continue;
+                }
+                Token::CommaAt => {
+                    if total_backticks <= 0 {
+                        return Err(LispError::ParseError); // unbalanced comma
+                    }
+                    total_backticks -= 1;
+                    stack.push(Group::CommaAt);
+                    continue;
+                }
+                Token::Quote => {
+                    if let Some(&mut Group::Quotes(ref mut n)) = stack.last_mut() {
+                        *n += 1;
+                        continue;
+                    }
+                    stack.push(Group::Quotes(1));
+                    continue;
+                }
+                Token::End => {
+                    let any_paren = stack.iter().any(|group| match *group {
+                        Group::Parens(_) => true,
+                        _ => false,
+                    });
+
+                    if any_paren {
+                        Err(LispError::ParseError) // unbalanced paren
+                    } else {
+                        Err(LispError::ParseError) // unexpected eof
+                    }
+                }
+            };
+            let mut v = r?;
+            loop {
+                match stack.last_mut() {
+                    None => return Ok(v),
+                    Some(&mut Group::Parens(ref mut values)) => {
+                        values.push(v);
+                        break;
+                    }
+                    _ => (),
+                }
+
+                let group = stack.pop().unwrap();
+
+                match group {
+                    Group::Backticks(n) if n > 0 => {
+                        total_backticks -= n;
+                        v = v.quasiquote(n as u32);
+                    }
+                    Group::Backticks(n) if n < 0 => {
+                        total_backticks -= n;
+                        v = v.comma((-n) as u32);
+                    }
+                    Group::CommaAt => {
+                        total_backticks += 1;
+                        v = v.comma_at(1);
+                    }
+                    Group::Quotes(n) => v = v.quote(n),
+                    _ => (),
+                }
+            }
+        }
+    }
+
+    fn next(&mut self) -> Result<(Span, Token<'lex>), LispError> {
+        let r = self.peek()?;
+        self.cur_token = None;
+        Ok(r)
+    }
+
+    fn peek(&mut self) -> Result<(Span, Token<'lex>), LispError> {
+        if let Some(tok) = self.cur_token {
+            Ok(tok)
+        } else {
+            let tok = self.lexer.next_token()?;
+            self.cur_token = Some(tok);
+            Ok(tok)
+        }
+    }
+
+    pub fn parse_single_expr(&mut self) -> Result<LispExpr, LispError> {
+        let expr = self.parse_expr()?;
+
+        match self.next()? {
+            (_, Token::End) => Ok(expr),
+            _ => Err(LispError::ParseError), // too many tokens
+        }
+    }
+
+    pub fn parse_exprs(&mut self) -> Result<Vec<LispExpr>, LispError> {
+        let mut res = Vec::new();
+        loop {
+            match self.peek()? {
+                (_, Token::End) => break,
+                _ => res.push(self.parse_expr()?),
+            }
+        }
+        Ok(res)
+    }
+}
+
+fn name_expr(input: &str) -> LispExpr {
+    match input {
+        "#t" => LispExpr::BoolLit(true),
+        "#f" => LispExpr::BoolLit(false),
+        _ => LispExpr::Ident(input.into()),
+    }
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+    fn parse(input: &str) -> Result<LispExpr, LispError> {
+        let mut parser = Parser::new(Lexer::new(input, 0));
+
+        parser.parse_single_expr()
+    }
+
+    #[test]
+    fn parse_lisp_expr() {
+        assert_eq!(
+            parse("1.5").unwrap(),
+            LispExpr::Number(LispNumber::Float(1.5))
+        );
+
+        assert_eq!(
+            parse(r#""hello""#).unwrap(),
+            LispExpr::StringLit(r#""hello""#.into())
+        );
+
+        assert_eq!(parse("foo").unwrap(), LispExpr::Ident("foo".into()));
+
+        let items = (1..=5)
+            .map(LispNumber::Integer)
+            .map(LispExpr::Number)
+            .collect::<Vec<_>>();
+        assert_eq!(parse("(1 2 3 4 5)").unwrap(), LispExpr::List(items));
+
+        let foo = LispExpr::Ident("foo".into());
+        let bar = LispExpr::Comma(Box::new(LispExpr::Ident("bar".into())), 1);
+        assert_eq!(
+            parse("`(foo ,bar)").unwrap(),
+            LispExpr::Quasiquote(Box::new(LispExpr::List(vec![foo, bar])), 1)
+        )
+    }
+
+    #[should_panic]
+    #[test]
+    fn unbalanced_comma() {
+        parse("`(foo ,,bar)").unwrap();
+    }
+
+    #[should_panic]
+    #[test]
+    fn unbalanced_paren() {
+        parse("(foo").unwrap();
+    }
+}