Remove regex compilation from hotpath

rozukke · Aug 9, 2024 · cf75e4c · cf75e4c
1 parent f73e690
commit cf75e4c
Showing 1 changed file with 22 additions and 18 deletions.
diff --git a/src/lexer/mod.rs b/src/lexer/mod.rs
@@ -1,3 +1,4 @@
+use lazy_static::lazy_static;
 use regex::Regex;
 
 use crate::lexer::cursor::Cursor;
@@ -43,6 +44,25 @@ pub fn tokenize(input: &str) -> impl Iterator<Item = Token> + '_ {
         }
     })
 }
+lazy_static! {
+    // Order is important since some patterns are subpatterns of others.
+    // Do NOT rearrange without a good hard think.
+    static ref PATTERNS: Vec<(TokenKind, Regex)> = vec![
+        (TokenKind::Junk, Regex::new(r"^[,\s]+").unwrap()),
+        (
+            TokenKind::Lit(LiteralKind::Hex),
+            Regex::new(r"^(0x|x)[0-9a-fA-F]+\b").unwrap(),
+        ),
+        (
+            TokenKind::Lit(LiteralKind::Dec),
+            Regex::new(r"^#[0-9]+\b").unwrap(),
+        ),
+        (TokenKind::Reg, Regex::new(r"^[rR][0-8]\b").unwrap()),
+        (TokenKind::Ident, Regex::new(r"^[a-zA-Z_]\w*\b").unwrap()),
+        (TokenKind::Comment, Regex::new(r"^;[^\n]*").unwrap()),
+        (TokenKind::Direc, Regex::new(r"^\.[a-zA-Z_]*\b").unwrap()),
+    ];
+}
 
 impl Cursor<'_> {
     pub fn advance_token(&mut self) -> Token {
@@ -53,26 +73,10 @@ impl Cursor<'_> {
             };
         }
 
-        let patterns = vec![
-            (TokenKind::Junk, Regex::new(r"^[,\s]+").unwrap()),
-            (
-                TokenKind::Lit(LiteralKind::Hex),
-                Regex::new(r"^(0x|x)[0-9a-fA-F]+\b").unwrap(),
-            ),
-            (
-                TokenKind::Lit(LiteralKind::Dec),
-                Regex::new(r"^#[0-9]+\b").unwrap(),
-            ),
-            (TokenKind::Ident, Regex::new(r"^[a-zA-Z_]\w*\b").unwrap()),
-            (TokenKind::Reg, Regex::new(r"^[rR][0-8]\b").unwrap()),
-            (TokenKind::Comment, Regex::new(r"^;[^\n]*").unwrap()),
-            (TokenKind::Direc, Regex::new(r"^\.[a-zA-Z_]*\b").unwrap()),
-        ];
-
-        for (kind, re) in patterns {
+        for (kind, re) in PATTERNS.iter() {
             if let Some(tok) = re.find(self.at_curr_pt()) {
                 let token = Token {
-                    kind,
+                    kind: *kind,
                     span: Span::new(Idx(self.curr_pt() as u32), tok.len() as u16),
                 };
                 self.advance(tok.len());