JS: whitelist delimiter unwrapping for js/incomplete-sanitization

Author: Esben Sparre Andreasen

javascript/ql/src/Security/CWE-116/IncompleteSanitization.ql

@@ -101,6 +101,34 @@ predicate allBackslashesEscaped(DataFlow::Node nd) {
   allBackslashesEscaped(nd.getAPredecessor())
 }
 
+/**
+ * Holds if `repl` looks like a call to "String.prototype.replace" that deliberately removes the first occurrence of `str`.
+ */
+predicate removesFirstOccurence(DataFlow::MethodCallNode repl, string str) {
+  repl.getMethodName() = "replace" and
+  repl.getArgument(0).getStringValue() = str and
+  repl.getArgument(1).getStringValue() = ""
+}
+
+/**
+ * Holds if `leftUnwrap` and `rightUnwrap` unwraps a string from a pair of surrounding delimiters.
+ */
+predicate isDelimiterUnwrapper(
+  DataFlow::MethodCallNode leftUnwrap, DataFlow::MethodCallNode rightUnwrap
+) {
+  exists(string left, string right |
+    left = "[" and right = "]"
+    or
+    left = "{" and right = "}"
+    or
+    left = "(" and right = ")"
+  |
+    removesFirstOccurence(leftUnwrap, left) and
+    removesFirstOccurence(rightUnwrap, right) and
+    leftUnwrap.getAMethodCall() = rightUnwrap
+  )
+}
+
 from MethodCallExpr repl, Expr old, string msg
 where
   repl.getMethodName() = "replace" and
@@ -122,7 +150,10 @@ where
       )
     ) and
     // don't flag replace operations in a loop
-    not DataFlow::valueNode(repl.getReceiver()) = DataFlow::valueNode(repl).getASuccessor+()
+    not DataFlow::valueNode(repl.getReceiver()) = DataFlow::valueNode(repl).getASuccessor+() and
+    // dont' flag unwrapper
+    not isDelimiterUnwrapper(repl.flow(), _) and
+    not isDelimiterUnwrapper(_, repl.flow())
     or
     exists(RegExpLiteral rel |
       isBackslashEscape(repl, rel) and

javascript/ql/test/query-tests/Security/CWE-116/IncompleteSanitization/IncompleteSanitization.expected

@@ -15,18 +15,12 @@
 | tst.js:61:10:61:18 | s.replace | This replaces only the first occurrence of "'" + "". |
 | tst.js:65:10:65:18 | s.replace | This replaces only the first occurrence of "'". |
 | tst.js:69:10:69:18 | s.replace | This replaces only the first occurrence of "'" + "". |
-| tst.js:130:2:130:10 | s.replace | This replaces only the first occurrence of '['. |
-| tst.js:130:2:130:27 | s.repla ... replace | This replaces only the first occurrence of ']'. |
-| tst.js:132:2:132:10 | s.replace | This replaces only the first occurrence of '{'. |
-| tst.js:132:2:132:27 | s.repla ... replace | This replaces only the first occurrence of '}'. |
 | tst.js:133:2:133:10 | s.replace | This replaces only the first occurrence of '<'. |
 | tst.js:133:2:133:27 | s.repla ... replace | This replaces only the first occurrence of '>'. |
 | tst.js:135:2:135:10 | s.replace | This replaces only the first occurrence of '['. |
 | tst.js:135:2:135:30 | s.repla ... replace | This replaces only the first occurrence of ']'. |
 | tst.js:136:2:136:10 | s.replace | This replaces only the first occurrence of '{'. |
 | tst.js:136:2:136:30 | s.repla ... replace | This replaces only the first occurrence of '}'. |
-| tst.js:138:6:138:14 | s.replace | This replaces only the first occurrence of '['. |
-| tst.js:139:6:139:14 | s.replace | This replaces only the first occurrence of ']'. |
 | tst.js:140:2:140:10 | s.replace | This replaces only the first occurrence of /{/. |
 | tst.js:140:2:140:27 | s.repla ... replace | This replaces only the first occurrence of /}/. |
 | tst.js:141:2:141:10 | s.replace | This replaces only the first occurrence of ']'. |