refactor and parse identifiers and int literals

src/ast.rs

@@ -1,6 +1,6 @@
 use std::{fmt::Debug, rc::Rc};
 
-use crate::token::Token;
+use crate::token::{Token, TokenType};
 
 pub trait Node: Debug {}
 
@@ -50,14 +50,35 @@ impl Statement for Return {}
 impl Node for Return {}
 
 #[derive(Debug)]
-pub struct Identifier {
+pub struct ExpressionStatement {
+    // TODO: probably not needed
     pub token: Token,
+    pub expression: Rc<dyn Expression>,
+}
+
+impl Statement for ExpressionStatement {}
+
+impl Node for ExpressionStatement {}
+
+#[derive(Debug)]
+pub struct Identifier {
+    pub token_type: TokenType,
+    pub value: String,
 }
 
 impl Node for Identifier {}
 
 impl Expression for Identifier {}
 
+#[derive(Debug)]
+pub struct IntegerLiteral {
+    pub value: i64,
+}
+
+impl Node for IntegerLiteral {}
+
+impl Expression for IntegerLiteral {}
+
 #[derive(Debug)]
 pub struct DummyExpression {}
 

src/parser.rs

@@ -1,27 +1,44 @@
 use std::{fmt, rc::Rc};
 
 use crate::{
-    ast::{DummyExpression, Expression, Identifier, Let, Program, Statement},
+    ast::{
+        DummyExpression, Expression, ExpressionStatement, Identifier, IntegerLiteral, Let, Program,
+        Return, Statement,
+    },
     lexer::Lexer,
-    token::Token,
+    token::{Token, TokenType},
 };
 
 pub type Result<T> = std::result::Result<T, Error>;
 
+type PrefixParseFn = Box<dyn FnOnce(&mut Parser) -> Rc<dyn Expression>>;
+type InfixParseFn = Box<dyn FnOnce(&mut Parser, Rc<dyn Expression>) -> Rc<dyn Expression>>;
+
 #[derive(Debug)]
 pub enum Error {
-    UnexpectedToken {
-        expected: Token,
-        actual: Option<Token>,
+    UnexpectedTokenType {
+        expected: TokenType,
+        actual: Option<TokenType>,
     },
 }
 
+#[derive(PartialEq, PartialOrd)]
+enum Precedence {
+    Lowest = 0,
+    Equals,
+    LessGreater,
+    Sum,
+    Product,
+    Prefix,
+    Call,
+}
+
 impl fmt::Display for Error {
     fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
         write!(f, "parser error: ")?;
         use Error::*;
         match self {
-            UnexpectedToken { expected, actual } => write!(
+            UnexpectedTokenType { expected, actual } => write!(
                 f,
                 "expected token `{:?}`, actual token: `{:?}`",
                 expected, actual
@@ -78,7 +95,8 @@ impl Parser {
                 use Token::*;
                 match token {
                     Let => self.parse_let_statement(),
-                    t => None,
+                    Return => self.parse_return_statement(),
+                    _ => self.parse_expression_statement(),
                 }
             }
             None => unreachable!(),
@@ -86,69 +104,173 @@ impl Parser {
     }
 
     fn parse_let_statement(&mut self) -> Option<Rc<dyn Statement>> {
-        let token = self.cur_token.clone().unwrap();
-
-        if !self.expect_peek(&Token::Ident("".to_string())) {
+        let token = if let Some(token) = &self.cur_token {
+            token.clone()
+        } else {
             return None;
-        }
-
-        let name = Identifier {
-            token: self.cur_token.clone().unwrap(),
         };
 
-        if !self.expect_peek(&Token::Assign) {
+        let value = self.expect_peek_ident()?;
+
+        let name = Identifier {
+            token_type: TokenType::Let,
+            value,
+        };
+
+        if !self.expect_peek(TokenType::Assign) {
             return None;
         }
 
-        let value = self.parse_expression()?;
+        while !self.cur_token_is(TokenType::Semicolon) {
+            self.next();
+        }
+        let value = Rc::new(DummyExpression {});
 
         Some(Rc::new(Let { token, name, value }))
     }
 
-    fn parse_expression(&mut self) -> Option<Rc<dyn Expression>> {
-        while !self.cur_token_is(&Token::Semicolon) {
+    fn parse_return_statement(&mut self) -> Option<Rc<dyn Statement>> {
+        let token = self.cur_token.clone().unwrap();
+
+        while !self.cur_token_is(TokenType::Semicolon) {
             self.next();
         }
-        Some(Rc::new(DummyExpression {}))
+        let value = Rc::new(DummyExpression {});
+
+        Some(Rc::new(Return { token, value }))
     }
 
-    fn expect_peek(&mut self, token: &Token) -> bool {
-        if self.peek_token.is_none() {
-            self.errors.push(Error::UnexpectedToken {
-                expected: token.clone(),
+    fn parse_expression_statement(&mut self) -> Option<Rc<dyn Statement>> {
+        let token = self.cur_token.clone()?;
+
+        let expression = self.parse_expression(Precedence::Lowest)?;
+
+        if self.peek_token_is(TokenType::Semicolon) {
+            self.next();
+        }
+
+        Some(Rc::new(ExpressionStatement { token, expression }))
+    }
+
+    fn parse_expression(&mut self, precedence: Precedence) -> Option<Rc<dyn Expression>> {
+        let prefix = self.prefix_parse_fn()?;
+
+        let left_exp = prefix(self);
+        Some(left_exp)
+    }
+
+    fn parse_identifier(&mut self, value: String) -> Rc<dyn Expression> {
+        Rc::new(Identifier {
+            token_type: TokenType::Ident,
+            value,
+        })
+    }
+
+    fn parse_integer_literal(&mut self, value: u64) -> Rc<dyn Expression> {
+        let value = if let Ok(value) = i64::try_from(value) {
+            value
+        } else {
+            // TODO: emit error
+            return Rc::new(DummyExpression {});
+        };
+        Rc::new(IntegerLiteral { value })
+    }
+
+    fn expect_peek(&mut self, token_type: TokenType) -> bool {
+        let peek_token = if let Some(token) = &self.peek_token {
+            token
+        } else {
+            self.errors.push(Error::UnexpectedTokenType {
+                expected: token_type,
                 actual: None,
             });
             return false;
-        }
+        };
 
-        let peek_token = self.peek_token.clone().unwrap();
-
-        if token.is_same_type(&peek_token) {
+        if token_type == peek_token.token_type() {
             self.next();
             return true;
         } else {
-            self.errors.push(Error::UnexpectedToken {
-                expected: token.clone(),
+            self.errors.push(Error::UnexpectedTokenType {
+                expected: token_type,
                 actual: None,
             });
             return false;
         }
     }
 
-    fn peek_token_is(&self, token: &Token) -> bool {
-        if self.peek_token.is_none() {
-            return false;
+    fn expect_peek_ident(&mut self) -> Option<String> {
+        let peek_token = self.peek_token.clone()?;
+        match peek_token {
+            Token::Ident(value) => {
+                self.next();
+                Some(value)
+            }
+            _ => None,
         }
-
-        self.peek_token.clone().unwrap().is_same_type(token)
     }
 
-    fn cur_token_is(&self, token: &Token) -> bool {
-        if self.cur_token.is_none() {
-            return false;
+    fn expect_cur_int_literal(&mut self) -> Option<u64> {
+        let peek_token = self.peek_token.clone()?;
+        match peek_token {
+            Token::Int(value) => {
+                self.next();
+                Some(value)
+            }
+            _ => None,
         }
+    }
 
-        self.cur_token.clone().unwrap().is_same_type(token)
+    fn peek_token_is(&self, token_type: TokenType) -> bool {
+        let peek_token = if let Some(token) = &self.peek_token {
+            token
+        } else {
+            return false;
+        };
+
+        peek_token.token_type() == token_type
+    }
+
+    fn cur_token_is(&self, token_type: TokenType) -> bool {
+        let cur_token = if let Some(token) = &self.cur_token {
+            token
+        } else {
+            return false;
+        };
+
+        cur_token.token_type() == token_type
+    }
+
+    fn prefix_parse_fn(&self) -> Option<PrefixParseFn> {
+        let token = if let Some(token) = &self.cur_token {
+            token
+        } else {
+            return None;
+        };
+
+        use Token::*;
+        match token {
+            Ident(value) => {
+                let value = value.clone();
+                Some(Box::new(move |parser| {
+                    Self::parse_identifier(parser, value)
+                }))
+            }
+            Int(value) => {
+                let value = *value;
+                Some(Box::new(move |parser| {
+                    Self::parse_integer_literal(parser, value)
+                }))
+            }
+            _ => None,
+        }
+    }
+
+    fn infix_parse_fn(token: &Token) -> InfixParseFn {
+        use Token::*;
+        match token {
+            _ => unimplemented!(),
+        }
     }
 }
 
@@ -201,10 +323,73 @@ mod tests {
         }
     }
 
+    #[test]
+    fn return_statements() {
+        let source = "return 5;\
+                     return 10;\
+                     return 838383;\
+                     "
+        .to_string();
+
+        let lexer = Lexer::new(source);
+
+        let mut parser = Parser::new(lexer);
+
+        let program = parser.parse().unwrap();
+        check_parser_errors(parser);
+
+        assert_eq!(program.statements.len(), 3);
+
+        for stmt in program.statements {
+            assert_eq!(
+                format!("{stmt:?}"),
+                "Return { token: Return, value: DummyExpression }"
+            )
+        }
+    }
+
+    #[test]
+    fn identifier_expression() {
+        let source = "foobar;".to_owned();
+
+        let lexer = Lexer::new(source);
+
+        let mut parser = Parser::new(lexer);
+
+        let program = parser.parse().unwrap();
+        check_parser_errors(parser);
+
+        let expected_identifiers = vec!["foobar"];
+        let mut statements_iter = program.statements.iter();
+        for tt in expected_identifiers {
+            let statement = statements_iter.next().unwrap();
+            test_identifier_expression(statement.clone(), tt);
+        }
+    }
+
+    #[test]
+    fn integer_literal_expression() {
+        let source = "6;".to_owned();
+
+        let lexer = Lexer::new(source);
+
+        let mut parser = Parser::new(lexer);
+
+        let program = parser.parse().unwrap();
+        check_parser_errors(parser);
+
+        let expected_integers = vec![6];
+        let mut statements_iter = program.statements.iter();
+        for tt in expected_integers {
+            let statement = statements_iter.next().unwrap();
+            test_integer_literal_expression(statement.clone(), tt);
+        }
+    }
+
     fn test_let_statement(stmt: Rc<dyn Statement>, name: &str) {
         assert_eq!(
             format!("{stmt:?}"),
-            format!("Let {{ token: Let, name: Identifier {{ token: Ident(\"{name}\") }}, value: DummyExpression }}"),
+            format!("Let {{ token: Let, name: Identifier {{ token_type: Let, value: \"{name}\" }}, value: DummyExpression }}"),
         );
     }
 
@@ -221,4 +406,22 @@ mod tests {
 
         panic!("{err}");
     }
+
+    fn test_identifier_expression(stmt: Rc<dyn Statement>, name: &str) {
+        assert_eq!(
+            format!("{stmt:?}"),
+            format!(
+                "ExpressionStatement {{ token: Ident(\"{name}\"), expression: Identifier {{ token_type: Ident, value: \"{name}\" }} }}"
+            ),
+        );
+    }
+
+    fn test_integer_literal_expression(stmt: Rc<dyn Statement>, num: i64) {
+        assert_eq!(
+            format!("{stmt:?}"),
+            format!(
+                "ExpressionStatement {{ token: Int({num}), expression: IntegerLiteral {{ value: {num} }} }}"
+            ),
+        );
+    }
 }

src/token.rs

@@ -4,7 +4,44 @@ pub enum Token {
     EOF,
 
     Ident(String),
-    Int(i64),
+    Int(u64),
+
+    Assign,
+    Plus,
+    Minus,
+    Bang,
+    Asterisk,
+    Slash,
+
+    Lt,
+    Gt,
+    Eq,
+    NotEq,
+
+    Comma,
+    Semicolon,
+
+    Lparen,
+    Rparen,
+    Lbrace,
+    Rbrace,
+
+    Function,
+    Let,
+    True,
+    False,
+    If,
+    Else,
+    Return,
+}
+
+#[derive(Debug, PartialEq, Eq)]
+pub enum TokenType {
+    Illegal,
+    EOF,
+
+    Ident,
+    Int,
 
     Assign,
     Plus,
@@ -36,24 +73,35 @@ pub enum Token {
 }
 
 impl Token {
-    pub fn is_same_type(&self, other: &Token) -> bool {
-        use Token::*;
+    pub fn token_type(&self) -> TokenType {
         match self {
-            Ident(_) => {
-                if let Ident(_) = other {
-                    true
-                } else {
-                    false
-                }
-            }
-            Int(_) => {
-                if let Int(_) = other {
-                    true
-                } else {
-                    false
-                }
-            }
-            tok => tok == other,
+            Token::Illegal => TokenType::Illegal,
+            Token::EOF => TokenType::EOF,
+            Token::Ident(_) => TokenType::Ident,
+            Token::Int(_) => TokenType::Int,
+            Token::Assign => TokenType::Assign,
+            Token::Plus => TokenType::Plus,
+            Token::Minus => TokenType::Minus,
+            Token::Bang => TokenType::Bang,
+            Token::Asterisk => TokenType::Asterisk,
+            Token::Slash => TokenType::Slash,
+            Token::Lt => TokenType::Lt,
+            Token::Gt => TokenType::Gt,
+            Token::Eq => TokenType::Eq,
+            Token::NotEq => TokenType::NotEq,
+            Token::Comma => TokenType::Comma,
+            Token::Semicolon => TokenType::Semicolon,
+            Token::Lparen => TokenType::Lparen,
+            Token::Rparen => TokenType::Rparen,
+            Token::Lbrace => TokenType::Lbrace,
+            Token::Rbrace => TokenType::Rbrace,
+            Token::Function => TokenType::Function,
+            Token::Let => TokenType::Let,
+            Token::True => TokenType::True,
+            Token::False => TokenType::False,
+            Token::If => TokenType::If,
+            Token::Else => TokenType::Else,
+            Token::Return => TokenType::Return,
         }
     }