Merge pull request #869 from CodaFi/re-gex-enter-the-gecko

Regex Literals Are Never Left Bound

Author: CodaFi

Sources/SwiftParser/Lexer.swift

@@ -783,14 +783,14 @@ extension Lexer.Cursor {
       }
 
       // Try lex a regex literal.
-      if let token = self.tryLexRegexLiteral(start) {
+      if let token = self.tryLexRegexLiteral(start, ContentStart) {
         return (token, [])
       }
       // Otherwise try lex a magic pound literal.
       return self.lexMagicPoundLiteral()
     case UInt8(ascii: "/"):
       // Try lex a regex literal.
-      if let token = self.tryLexRegexLiteral(start) {
+      if let token = self.tryLexRegexLiteral(start, ContentStart) {
         return (token, [])
       }
 
@@ -2027,8 +2027,13 @@ extension Lexer.Cursor {
 
 extension Lexer.Cursor {
   mutating func tryLexRegexLiteral(
-    _ TokStart: Lexer.Cursor
+    _ TokStart: Lexer.Cursor,
+    _ ContentStart: Lexer.Cursor
   ) -> RawTokenKind? {
+    guard !TokStart.isLeftBound(ContentStart) else {
+      return nil
+    }
+
     var Tmp = TokStart
     var poundCount = 0
     var parenCount = 0

Tests/SwiftParserTest/LexerTests.swift

@@ -352,11 +352,6 @@ public class LexerTests: XCTestCase {
         lexeme(.regexLiteral, "#/abc|#def/#"),
         lexeme(.eof, ""),
       ]),
-      ("#/abc|#def/", [
-        lexeme(.pound, "#"),
-        lexeme(.regexLiteral, "/abc|#def/"),
-        lexeme(.eof, ""),
-      ]),
       ("#/abc\n/#", [
         lexeme(.pound, "#"),
         lexeme(.unspacedBinaryOperator, "/"),
@@ -567,15 +562,61 @@ public class LexerTests: XCTestCase {
       """
       n /= 2 // foo
       """
-      let lexemes = data.withUTF8 { buf in
+      data.withUTF8 { buf in
         let lexemes = Lexer.lex(buf)
         AssertEqualTokens(lexemes, [
           lexeme(.identifier, "n ", trailing: 1),
           lexeme(.spacedBinaryOperator, "/= ", trailing: 1),
           lexeme(.integerLiteral, "2 ", trailing: 1),
           lexeme(.eof, "// foo", leading: 6),
         ])
-        return lexemes
+      }
+    }
+
+    do {
+      var data =
+      """
+      UIColor(white: 216.0/255.0, alpha: 44.0/255.0)
+      """
+      data.withUTF8 { buf in
+        let lexemes = Lexer.lex(buf)
+        AssertEqualTokens(lexemes, [
+          lexeme(.identifier, "UIColor"),
+          lexeme(.leftParen, "("),
+          lexeme(.identifier, "white"),
+          lexeme(.colon, ": ", trailing: 1),
+          lexeme(.floatingLiteral, "216.0"),
+          lexeme(.unspacedBinaryOperator, "/"),
+          lexeme(.floatingLiteral, "255.0"),
+          lexeme(.comma, ", ", trailing: 1),
+          lexeme(.identifier, "alpha"),
+          lexeme(.colon, ": ", trailing: 1),
+          lexeme(.floatingLiteral, "44.0"),
+          lexeme(.unspacedBinaryOperator, "/"),
+          lexeme(.floatingLiteral, "255.0"),
+          lexeme(.rightParen, ")"),
+          lexeme(.eof, ""),
+        ])
+      }
+    }
+
+    do {
+      var data =
+      """
+      #/abc|#def/
+      """
+      data.withUTF8 { buf in
+        let lexemes = Lexer.lex(buf)
+        AssertEqualTokens(lexemes, [
+          lexeme(.pound, "#"),
+          lexeme(.unspacedBinaryOperator, "/"),
+          lexeme(.identifier, "abc"),
+          lexeme(.unspacedBinaryOperator, "|"),
+          lexeme(.pound, "#"),
+          lexeme(.identifier, "def"),
+          lexeme(.postfixOperator, "/"),
+          lexeme(.eof, ""),
+        ])
       }
     }
   }