Fix [^] and [] patterns in Regex (#173)

* add failing adaptable regex test * fix adapting regex * add more failing regex * fix failing regex with [] and [^] * make only the parser pass but don't change the output regex
sebastienros · Jul 18, 2021 · 611b7da · 611b7da
1 parent a47cf50
commit 611b7da
Show file tree

Hide file tree

Showing 2 changed files with 44 additions and 40 deletions.
diff --git a/src/Esprima/Scanner.cs b/src/Esprima/Scanner.cs
@@ -1549,13 +1549,6 @@ public Token ScanTemplate()
                 tmp = Regex.Replace(tmp, "[\uD800-\uDBFF][\uDC00-\uDFFF]", astralSubstitute);
             }
 
-            // .NET doesn't support [^] which is equivalent to `[^.]`
-            // c.f. https://github.com/sebastienros/esprima-dotnet/issues/146
-            if (tmp.Contains("[^]"))
-            {
-                tmp = tmp.Replace("[^]", "[^.]");
-            }
-
             // First, detect invalid regular expressions.
             var options = ParseRegexOptions(flags);
 
@@ -1565,7 +1558,16 @@ public Token ScanTemplate()
             }
             catch
             {
-                ThrowUnexpectedToken(Messages.InvalidRegExp);
+                tmp = EscapeFailingRegex(tmp);
+
+                try
+                {
+                    new Regex(tmp, options);
+                }
+                catch
+                {
+                    ThrowUnexpectedToken(Messages.InvalidRegExp);
+                }
             }
 
             // Return a regular expression object for this pattern-flag pair, or
@@ -1574,31 +1576,22 @@ public Token ScanTemplate()
             try
             {
                 // Do we need to convert the expression to its .NET equivalent?
-                if (_adaptRegexp)
+                if (_adaptRegexp && options.HasFlag(RegexOptions.Multiline))
                 {
-                    var newPattern = pattern;
+                    // Replace all non-escaped $ occurences by \r?$
+                    // c.f. http://programmaticallyspeaking.com/regular-expression-multiline-mode-whats-a-newline.html
 
-                    if (options.HasFlag(RegexOptions.Multiline))
+                    int index = 0;
+                    var newPattern = pattern;
+                    while ((index = newPattern.IndexOf("$", index, StringComparison.Ordinal)) != -1)
                     {
-                        // Replace all non-escaped $ occurences by \r?$
-                        // c.f. http://programmaticallyspeaking.com/regular-expression-multiline-mode-whats-a-newline.html
-
-                        int index = 0;
-                        while ((index = newPattern.IndexOf("$", index, StringComparison.Ordinal)) != -1)
+                        if (index > 0 && newPattern[index - 1] != '\\')
                         {
-                            if (index > 0 && newPattern[index - 1] != '\\')
-                            {
-                                newPattern = newPattern.Substring(0, index) + @"\r?" + newPattern.Substring(index);
-                                index += 4;
-                            }
+                            newPattern = newPattern.Substring(0, index) + @"\r?" + newPattern.Substring(index);
+                            index += 4;
                         }
                     }
 
-                    if (newPattern.Contains("[^]"))
-                    {
-                        newPattern = newPattern.Replace("[^]", "[^.]");
-                    }
-
                     pattern = newPattern;
                 }
 
@@ -1610,6 +1603,26 @@ public Token ScanTemplate()
             }
         }
 
+        public string EscapeFailingRegex(string pattern)
+        {
+            // .NET 4.x doesn't support [^] which should match any character including newline
+            // c.f. https://github.com/sebastienros/esprima-dotnet/issues/146
+            if (pattern.Contains("[^]"))
+            {
+                pattern = pattern.Replace("[^]", @"[\s\S]");
+            }
+
+
+            // .NET doesn't support [] which should not match any characters (inverse of [^])
+            if (pattern.Contains("[]"))
+            {
+                // This is a temporary solution to make the parser pass. It is not a correct replacement as it will match the \0 char.
+                pattern = pattern.Replace("[]", @"[\0]");
+            }
+
+            return pattern;
+        }
+
         public Token ScanRegExpBody()
         {
             var ch = Source[Index];

diff --git a/test/Esprima.Tests/RegExpTests.cs b/test/Esprima.Tests/RegExpTests.cs
@@ -23,27 +23,18 @@ private static Regex CreateRegex(string code)
         }
 
         [Theory]
-        [InlineData(@"/[^]*? (:[rp] [el] a[\w -]+)[^]*/")]
+        [InlineData(@"/[^]*?(:[rp][el]a[\w-]+)[^]*/")]
         [InlineData(@"/[^]/")]
         [InlineData(@"/[^ ]/")]
+        [InlineData(@"/[]/")]
+        [InlineData(@"/[]*/")]
+        [InlineData(@"/[]a/")]
         public void ShouldParseRegularExpression(string regexp)
         {
             var parser = new JavaScriptParser(@"var O = " + regexp);
             var program = parser.ParseScript();
 
             Assert.NotNull(program);
-        }
-
-        [Theory]
-        [InlineData(@"/[^]*? (:[rp] [el] a[\w -]+)[^]*/")]
-        [InlineData(@"/[^]/")]
-        [InlineData(@"/[^ ]/")]
-        public void ShouldGetNonNullRegexFromScanner(string regexp)
-        {
-            var scanner = new Scanner("", new ParserOptions { AdaptRegexp = true });
-            var regex = scanner.TestRegExp(regexp, "");
-
-            Assert.NotNull(regex);
         }
     }
-}
+}