refactoring Scanner

rust/src/compiler.rs

@@ -1,4 +1,4 @@
-use crate::scanner::{Scanner, TokenKind};
+use crate::scanner::Scanner;
 use color_eyre::eyre::Result;
 
 pub fn compile(source: &str) -> Result<()> {

rust/src/scanner.rs

@@ -4,6 +4,10 @@ pub struct Scanner<'a> {
     line: usize,
 }
 
+fn is_alpha(c: &char) -> bool {
+    c.is_ascii_alphabetic() || *c == '_'
+}
+
 impl<'a> Scanner<'a> {
     pub fn new(source: &'a str) -> Self {
         Self {
@@ -13,111 +17,75 @@ impl<'a> Scanner<'a> {
         }
     }
 
-    fn make_token(&self, kind: TokenKind) -> Token<'a> {
-        Token {
-            kind,
-            value: std::str::from_utf8(&self.source[..self.current]).unwrap(),
-            line: self.line,
-        }
-    }
-
-    fn error_token(&self, message: &'static str) -> Token<'a> {
-        Token {
-            kind: TokenKind::Error,
-            value: message,
-            line: self.line,
-        }
-    }
-
-    fn advance(&mut self) -> char {
+    fn advance(&mut self) -> Option<char> {
         self.current += 1;
-        self.source[self.current - 1] as char
-    }
-
-    fn is_match(&mut self, expected: char) -> bool {
-        if self.peek() != Some(expected) {
-            return false;
+        self.source.get(self.current - 1).map(|&x| x as char)
     }
 
+    fn advance_if<F: Fn(&char) -> bool>(&mut self, f: F) -> bool {
+        if matches!(self.peek(), Some(c) if f(&c)) {
             self.current += 1;
             true
+        } else {
+            false
+        }
     }
 
     fn skip_whitespace(&mut self) {
         loop {
-            match self.peek() {
-                Some(' ') | Some('\r') | Some('\t') => {
-                    self.advance();
-                    break;
+            while self.advance_if(|&c| c == ' ' || c == '\r' || c == '\t') {}
+
+            if self.peek() == Some('/') && self.peek_next() == Some('/') {
+                while self.advance_if(|&c| c != '\n') {}
             }
-                Some('\n') => {
+
+            if self.advance_if(|&c| c == '\n') {
                 self.line += 1;
-                    self.advance();
-                    break;
-                }
-                Some('/') => {
-                    if matches!(self.peek_next(), Some(c) if c == '/') {
-                        while matches!(self.peek(), Some(c) if c != '\n') {
-                            self.advance();
-                        }
             } else {
                 return;
             }
         }
-                _ => {
-                    return;
-                }
-            }
-        }
     }
 
     fn peek(&self) -> Option<char> {
         self.source.get(self.current).map(|&x| x as char)
     }
 
-    fn peek_next(&mut self) -> Option<char> {
+    fn peek_next(&self) -> Option<char> {
         self.source.get(self.current + 1).map(|&x| x as char)
     }
 
     fn string(&mut self) -> Token<'a> {
-        while matches!(self.peek(), Some(c) if c != '"') {
-            if self.peek() == Some('\n') {
+        loop {
+            match self.advance() {
+                Some('"') => break,
+                Some('\n') => {
                     self.line += 1;
                 }
-            self.advance();
+                Some(_) => {}
+                None => return Token::error(self, "Unterminated string."),
             }
-
-        if self.peek().is_none() {
-            // Maybe this should return an Err?
-            return self.error_token("Unterminated string.");
         }
 
-        self.advance();
-        self.make_token(TokenKind::String)
+        Token::new(self, TokenKind::String)
     }
 
     fn number(&mut self) -> Token<'a> {
-        while matches!(self.peek(), Some(c) if c.is_ascii_digit()) {
-            self.advance();
-        }
+        while self.advance_if(char::is_ascii_digit) {}
 
         if self.peek() == Some('.') && matches!(self.peek_next(), Some(c) if c.is_ascii_digit()) {
             self.advance();
 
-            while matches!(self.peek(), Some(c) if c.is_ascii_digit()) {
-                self.advance();
-            }
+            while self.advance_if(char::is_ascii_digit) {}
         }
 
-        self.make_token(TokenKind::Number)
+        Token::new(self, TokenKind::Number)
     }
 
     fn identifier(&mut self) -> Token<'a> {
-        while matches!(self.peek(), Some(c) if is_alpha(c) || c.is_ascii_digit()) {
-            self.advance();
-        }
+        while self.advance_if(|c| is_alpha(c) || c.is_ascii_digit()) {}
 
-        self.make_token(self.identifier_type())
+        Token::new(self, self.identifier_type())
     }
 
     fn identifier_type(&self) -> TokenKind {
@@ -154,66 +122,87 @@ impl<'a> Iterator for Scanner<'a> {
 
     fn next(&mut self) -> Option<Self::Item> {
         self.skip_whitespace();
-        self.peek()?;
 
-        Some(match self.advance() {
-            c if is_alpha(c) => self.identifier(),
+        self.advance()
+            .map(|c| match c {
+                c if is_alpha(&c) => self.identifier(),
                 c if c.is_ascii_digit() => self.number(),
-            '(' => self.make_token(TokenKind::LeftParen),
-            ')' => self.make_token(TokenKind::RightParen),
-            '{' => self.make_token(TokenKind::LeftBrace),
-            '}' => self.make_token(TokenKind::RightBrace),
-            ';' => self.make_token(TokenKind::Semicolon),
-            ',' => self.make_token(TokenKind::Comma),
-            '.' => self.make_token(TokenKind::Dot),
-            '-' => self.make_token(TokenKind::Minus),
-            '+' => self.make_token(TokenKind::Plus),
-            '/' => self.make_token(TokenKind::Slash),
-            '*' => self.make_token(TokenKind::Star),
+                '(' => Token::new(self, TokenKind::LeftParen),
+                ')' => Token::new(self, TokenKind::RightParen),
+                '{' => Token::new(self, TokenKind::LeftBrace),
+                '}' => Token::new(self, TokenKind::RightBrace),
+                ';' => Token::new(self, TokenKind::Semicolon),
+                ',' => Token::new(self, TokenKind::Comma),
+                '.' => Token::new(self, TokenKind::Dot),
+                '-' => Token::new(self, TokenKind::Minus),
+                '+' => Token::new(self, TokenKind::Plus),
+                '/' => Token::new(self, TokenKind::Slash),
+                '*' => Token::new(self, TokenKind::Star),
                 '!' => {
-                if self.is_match('=') {
-                    self.make_token(TokenKind::BangEqual)
+                    if self.advance_if(|&c| c == '=') {
+                        Token::new(self, TokenKind::BangEqual)
                     } else {
-                    self.make_token(TokenKind::Bang)
+                        Token::new(self, TokenKind::Bang)
                     }
                 }
                 '=' => {
-                if self.is_match('=') {
-                    self.make_token(TokenKind::EqualEqual)
+                    if self.advance_if(|&c| c == '=') {
+                        Token::new(self, TokenKind::EqualEqual)
                     } else {
-                    self.make_token(TokenKind::Equal)
+                        Token::new(self, TokenKind::Equal)
                     }
                 }
                 '<' => {
-                if self.is_match('=') {
-                    self.make_token(TokenKind::LessEqual)
+                    if self.advance_if(|&c| c == '=') {
+                        Token::new(self, TokenKind::LessEqual)
                     } else {
-                    self.make_token(TokenKind::Less)
+                        Token::new(self, TokenKind::Less)
                     }
                 }
                 '>' => {
-                if self.is_match('=') {
-                    self.make_token(TokenKind::GreaterEqual)
+                    if self.advance_if(|&c| c == '=') {
+                        Token::new(self, TokenKind::GreaterEqual)
                     } else {
-                    self.make_token(TokenKind::Greater)
+                        Token::new(self, TokenKind::Greater)
                     }
                 }
                 '"' => self.string(),
-            _ => self.error_token("Unexpected character."),
+                _ => Token::error(self, "Unexpected character."),
+            })
+            .or({
+                Some(Token {
+                    kind: TokenKind::Eof,
+                    value: "",
+                    line: self.line,
+                })
             })
     }
 }
 
-fn is_alpha(c: char) -> bool {
-    c.is_ascii_alphabetic() || c == '_'
-}
-
 pub struct Token<'a> {
     pub kind: TokenKind,
     pub value: &'a str,
     pub line: usize,
 }
 
+impl<'a> Token<'a> {
+    fn new(scanner: &Scanner<'a>, kind: TokenKind) -> Self {
+        Token {
+            kind,
+            value: std::str::from_utf8(&scanner.source[..scanner.current]).unwrap(),
+            line: scanner.line,
+        }
+    }
+
+    fn error(scanner: &Scanner, message: &'static str) -> Self {
+        Token {
+            kind: TokenKind::Error,
+            value: message,
+            line: scanner.line,
+        }
+    }
+}
+
 #[derive(Debug, PartialEq)]
 pub enum TokenKind {
     // Single-character tokens.
@@ -293,7 +282,7 @@ mod tests {
         assert_source_scan("fa", Some(TokenKind::Identifier));
 
         assert_source_scan("@", Some(TokenKind::Error));
-        assert_source_scan("", None);
+        assert_source_scan("", Some(TokenKind::Eof));
     }
 
     #[test]
@@ -301,14 +290,14 @@ mod tests {
         let mut scanner = Scanner::new("()");
         assert_scan(&mut scanner, Some(TokenKind::LeftParen));
         assert_scan(&mut scanner, Some(TokenKind::RightParen));
-        assert_scan(&mut scanner, None);
+        assert_scan(&mut scanner, Some(TokenKind::Eof));
     }
 
     #[test]
     fn test_whitespace() {
         assert_source_scan(" foo", Some(TokenKind::Identifier));
         assert_source_scan("\tfoo", Some(TokenKind::Identifier));
-        assert_source_scan("//   \n", None);
+        assert_source_scan("//   \n", Some(TokenKind::Eof));
     }
 
     fn assert_source_scan(source: &str, kind: Option<TokenKind>) {