fix parse errors related to char escapes and char ranges

Author: erik-krogh

javascript/extractor/src/com/semmle/js/parser/RegExpParser.java

@@ -35,6 +35,7 @@
 import com.semmle.js.ast.regexp.ZeroWidthPositiveLookahead;
 import com.semmle.js.ast.regexp.ZeroWidthPositiveLookbehind;
 import java.util.ArrayList;
+import java.util.Arrays;
 import java.util.List;
 
 /** A parser for ECMAScript 2018 regular expressions. */
@@ -499,7 +500,11 @@ private RegExpTerm parseCharacterClass() {
   private RegExpTerm parseCharacterClassElement() {
     SourceLocation loc = new SourceLocation(pos());
     RegExpTerm atom = this.parseCharacterClassAtom();
-    if (!this.lookahead("-]") && this.match("-"))
+    for (String c : Arrays.asList("d", "D", "s", "S", "w", "W")) {
+      if (this.lookahead("-\\" + c))
+        return atom;
+    }
+    if (!this.lookahead("-]") && this.match("-") && !(atom instanceof CharacterClassEscape))
       return this.finishTerm(new CharacterClassRange(loc, atom, this.parseCharacterClassAtom()));
     return atom;
   }

javascript/ql/test/library-tests/RegExp/CharacterRange/CharaterRange.expected

@@ -0,0 +1,10 @@
+range
+| tst.js:1:13:1:17 | [w-z] | tst.js:1:14:1:16 | w-z |
+| tst.js:6:13:6:19 | [\\n-\\r] | tst.js:6:14:6:18 | \\n-\\r |
+| tst.js:7:13:7:18 | [\\n-z] | tst.js:7:14:7:17 | \\n-z |
+escapeClass
+| tst.js:2:13:2:16 | [\\w] | tst.js:2:14:2:15 | \\w |
+| tst.js:3:13:3:18 | [\\w-z] | tst.js:3:14:3:15 | \\w |
+| tst.js:4:13:4:19 | [\\w-\\w] | tst.js:4:14:4:15 | \\w |
+| tst.js:4:13:4:19 | [\\w-\\w] | tst.js:4:17:4:18 | \\w |
+| tst.js:5:13:5:18 | [z-\\w] | tst.js:5:16:5:17 | \\w |

javascript/ql/test/library-tests/RegExp/CharacterRange/CharaterRange.ql

@@ -0,0 +1,9 @@
+import javascript
+
+query predicate range(RegExpCharacterClass cla, RegExpCharacterRange range) {
+  cla.getAChild() = range
+}
+
+query predicate escapeClass(RegExpCharacterClass cla, RegExpCharacterClassEscape escape) {
+  cla.getAChild() = escape
+}

javascript/ql/test/library-tests/RegExp/CharacterRange/tst.js

@@ -0,0 +1,7 @@
+var reg1 = /[w-z]/; // normal range w-z, matches: wxyz
+var reg2 = /[\w]/; // escape class, same as \w.
+var reg3 = /[\w-z]/; // escape class \w and "-" and "z", same as [a-zA-Z0-9\-z]
+var reg4 = /[\w-\w]/; // escape class \w (twice) and the char "-".
+var reg5 = /[z-\w]/; // same as reg3
+var reg6 = /[\n-\r]/; // from \n (code 10) to \r (code 13).
+var reg7 = /[\n-z]/; // from \n (code 10) to z (code 122).