JS: Extract surrogate pairs as one constant node

asger-semmle · asger-semmle · commit 68d23bcf8c52 · 2019-11-15T09:27:19.000Z
diff --git a/javascript/extractor/src/com/semmle/js/parser/RegExpParser.java b/javascript/extractor/src/com/semmle/js/parser/RegExpParser.java
@@ -489,6 +489,11 @@ private RegExpTerm parseCharacterClassAtom() {
       if (this.match("b")) return this.finishTerm(new ControlEscape(loc, "\b", 8, "\\b"));
       return this.finishTerm(this.parseAtomEscape(loc, true));
     }
-    return this.finishTerm(new Constant(loc, String.valueOf(c)));
+    String value = String.valueOf(c);
+    // Extract a surrogate pair as a single constant.
+    if (Character.isHighSurrogate(c) && Character.isLowSurrogate(peekChar(true))) {
+      value += this.nextChar();
+    }
+    return this.finishTerm(new Constant(loc, value));
   }
 }
diff --git a/javascript/ql/test/query-tests/RegExp/DuplicateCharacterInCharacterClass/tst.js b/javascript/ql/test/query-tests/RegExp/DuplicateCharacterInCharacterClass/tst.js
@@ -8,4 +8,6 @@
 /[aaa]/;
 /[\x0a\x0a]/;
 /[\u000a\n]/;
-/[\u{ff}]/;
+/[\u{ff}]/;
+/[\u{12340}-\u{12345}]/u; // OK
+new RegExp("[\u{12340}-\u{12345}]", "u"); // OK

Original file line number	Diff line number	Diff line change
`@@ -489,6 +489,11 @@ private RegExpTerm parseCharacterClassAtom() {`
`489`	`489`	`if (this.match("b")) return this.finishTerm(new ControlEscape(loc, "\b", 8, "\\b"));`
`490`	`490`	`return this.finishTerm(this.parseAtomEscape(loc, true));`
`491`	`491`	`}`
`492`		`- return this.finishTerm(new Constant(loc, String.valueOf(c)));`
	`492`	`+ String value = String.valueOf(c);`
	`493`	`+ // Extract a surrogate pair as a single constant.`
	`494`	`+ if (Character.isHighSurrogate(c) && Character.isLowSurrogate(peekChar(true))) {`
	`495`	`+ value += this.nextChar();`
	`496`	`+ }`
	`497`	`+ return this.finishTerm(new Constant(loc, value));`
`493`	`498`	`}`
`494`	`499`	`}`