track flow through string replace calls that just replace single chars

Author: erik-krogh

javascript/ql/lib/semmle/javascript/security/performance/PolynomialReDoSCustomizations.qll

@@ -95,15 +95,28 @@ module PolynomialReDoS {
       this.(StringReplaceCall).isGlobal() and
       // not lone char classes - they don't remove any repeated pattern.
       not exists(RegExpTerm root | root = this.(StringReplaceCall).getRegExp().getRoot() |
-        root instanceof RegExpCharacterClass
-        or
-        root instanceof RegExpCharacterClassEscape
+        isCharClassLike(root)
       )
       or
       this.(DataFlow::MethodCallNode).getMethodName() = StringOps::substringMethodName()
     }
   }
 
+  /**
+   * Holds if `term` matches a set of strings of length 1.
+   */
+  predicate isCharClassLike(RegExpTerm term) {
+    term instanceof RegExpCharacterClass
+    or
+    term instanceof RegExpCharacterClassEscape
+    or
+    term.(RegExpConstant).getValue().length() = 1
+    or
+    exists(RegExpAlt alt | term = alt |
+      forall(RegExpTerm choice | choice = alt.getAlternative() | isCharClassLike(choice))
+    )
+  }
+
   /**
    * An check on the length of a string, seen as a sanitizer guard.
    */

javascript/ql/test/query-tests/Performance/ReDoS/PolynomialBackTracking.expected

@@ -113,6 +113,8 @@
 | polynomial-redos.js:116:21:116:28 | [\\d\\D]*? | Strings starting with '/*' and with many repetitions of 'a/*' can start matching anywhere after the start of the preceeding \\/\\*[\\d\\D]*?\\*\\/ |
 | polynomial-redos.js:118:17:118:23 | (#\\d+)+ | Strings with many repetitions of '9' can start matching anywhere after the start of the preceeding \\d+ |
 | polynomial-redos.js:124:33:124:35 | \\s+ | Strings with many repetitions of ' ' can start matching anywhere after the start of the preceeding \\s+$ |
+| polynomial-redos.js:130:21:130:22 | c+ | Strings starting with 'c' and with many repetitions of 'c' can start matching anywhere after the start of the preceeding cc+D |
+| polynomial-redos.js:133:22:133:23 | f+ | Strings starting with 'f' and with many repetitions of 'f' can start matching anywhere after the start of the preceeding ff+G |
 | regexplib/address.js:27:3:27:5 | \\s* | Strings with many repetitions of ' ' can start matching anywhere after the start of the preceeding (\\s*\\(?0\\d{4}\\)?(\\s*\|-)\\d{3}(\\s*\|-)\\d{3}\\s*) |
 | regexplib/address.js:27:48:27:50 | \\s* | Strings with many repetitions of ' ' can start matching anywhere after the start of the preceeding (\\s*\\(?0\\d{3}\\)?(\\s*\|-)\\d{3}(\\s*\|-)\\d{4}\\s*) |
 | regexplib/address.js:27:93:27:95 | \\s* | Strings with many repetitions of ' ' can start matching anywhere after the start of the preceeding (\\s*(7\|8)(\\d{7}\|\\d{3}(\\-\|\\s{1})\\d{4})\\s*) |

javascript/ql/test/query-tests/Performance/ReDoS/PolynomialReDoS.expected

@@ -165,6 +165,16 @@ nodes
 | polynomial-redos.js:123:13:123:20 | replaced |
 | polynomial-redos.js:124:12:124:17 | result |
 | polynomial-redos.js:124:12:124:17 | result |
+| polynomial-redos.js:129:6:129:42 | modified |
+| polynomial-redos.js:129:17:129:23 | tainted |
+| polynomial-redos.js:129:17:129:42 | tainted ... g, "b") |
+| polynomial-redos.js:130:2:130:9 | modified |
+| polynomial-redos.js:130:2:130:9 | modified |
+| polynomial-redos.js:132:6:132:50 | modified2 |
+| polynomial-redos.js:132:18:132:24 | tainted |
+| polynomial-redos.js:132:18:132:50 | tainted ... g, "e") |
+| polynomial-redos.js:133:2:133:10 | modified2 |
+| polynomial-redos.js:133:2:133:10 | modified2 |
 edges
 | lib/closure.js:3:21:3:21 | x | lib/closure.js:4:16:4:16 | x |
 | lib/closure.js:3:21:3:21 | x | lib/closure.js:4:16:4:16 | x |
@@ -317,6 +327,8 @@ edges
 | polynomial-redos.js:5:6:5:32 | tainted | polynomial-redos.js:118:2:118:8 | tainted |
 | polynomial-redos.js:5:6:5:32 | tainted | polynomial-redos.js:118:2:118:8 | tainted |
 | polynomial-redos.js:5:6:5:32 | tainted | polynomial-redos.js:121:18:121:24 | tainted |
+| polynomial-redos.js:5:6:5:32 | tainted | polynomial-redos.js:129:17:129:23 | tainted |
+| polynomial-redos.js:5:6:5:32 | tainted | polynomial-redos.js:132:18:132:24 | tainted |
 | polynomial-redos.js:5:16:5:32 | req.query.tainted | polynomial-redos.js:5:6:5:32 | tainted |
 | polynomial-redos.js:5:16:5:32 | req.query.tainted | polynomial-redos.js:5:6:5:32 | tainted |
 | polynomial-redos.js:68:18:68:24 | req.url | polynomial-redos.js:68:18:68:24 | req.url |
@@ -327,6 +339,14 @@ edges
 | polynomial-redos.js:123:3:123:20 | result | polynomial-redos.js:124:12:124:17 | result |
 | polynomial-redos.js:123:3:123:20 | result | polynomial-redos.js:124:12:124:17 | result |
 | polynomial-redos.js:123:13:123:20 | replaced | polynomial-redos.js:123:3:123:20 | result |
+| polynomial-redos.js:129:6:129:42 | modified | polynomial-redos.js:130:2:130:9 | modified |
+| polynomial-redos.js:129:6:129:42 | modified | polynomial-redos.js:130:2:130:9 | modified |
+| polynomial-redos.js:129:17:129:23 | tainted | polynomial-redos.js:129:17:129:42 | tainted ... g, "b") |
+| polynomial-redos.js:129:17:129:42 | tainted ... g, "b") | polynomial-redos.js:129:6:129:42 | modified |
+| polynomial-redos.js:132:6:132:50 | modified2 | polynomial-redos.js:133:2:133:10 | modified2 |
+| polynomial-redos.js:132:6:132:50 | modified2 | polynomial-redos.js:133:2:133:10 | modified2 |
+| polynomial-redos.js:132:18:132:24 | tainted | polynomial-redos.js:132:18:132:50 | tainted ... g, "e") |
+| polynomial-redos.js:132:18:132:50 | tainted ... g, "e") | polynomial-redos.js:132:6:132:50 | modified2 |
 #select
 | lib/closure.js:4:5:4:17 | /u*o/.test(x) | lib/closure.js:3:21:3:21 | x | lib/closure.js:4:16:4:16 | x | This $@ that depends on $@ may run slow on strings with many repetitions of 'u'. | lib/closure.js:4:6:4:7 | u* | regular expression | lib/closure.js:3:21:3:21 | x | library input |
 | lib/lib.js:4:2:4:18 | regexp.test(name) | lib/lib.js:3:28:3:31 | name | lib/lib.js:4:14:4:17 | name | This $@ that depends on $@ may run slow on strings with many repetitions of 'a'. | lib/lib.js:1:15:1:16 | a* | regular expression | lib/lib.js:3:28:3:31 | name | library input |
@@ -409,3 +429,5 @@ edges
 | polynomial-redos.js:116:2:116:35 | tainted ... \\*\\//g) | polynomial-redos.js:5:16:5:32 | req.query.tainted | polynomial-redos.js:116:2:116:8 | tainted | This $@ that depends on $@ may run slow on strings starting with '/*' and with many repetitions of 'a/*'. | polynomial-redos.js:116:21:116:28 | [\\d\\D]*? | regular expression | polynomial-redos.js:5:16:5:32 | req.query.tainted | a user-provided value |
 | polynomial-redos.js:118:2:118:25 | tainted ... \\d+)+/) | polynomial-redos.js:5:16:5:32 | req.query.tainted | polynomial-redos.js:118:2:118:8 | tainted | This $@ that depends on $@ may run slow on strings with many repetitions of '9'. | polynomial-redos.js:118:17:118:23 | (#\\d+)+ | regular expression | polynomial-redos.js:5:16:5:32 | req.query.tainted | a user-provided value |
 | polynomial-redos.js:124:12:124:43 | result. ... /g, '') | polynomial-redos.js:5:16:5:32 | req.query.tainted | polynomial-redos.js:124:12:124:17 | result | This $@ that depends on $@ may run slow on strings with many repetitions of ' '. | polynomial-redos.js:124:33:124:35 | \\s+ | regular expression | polynomial-redos.js:5:16:5:32 | req.query.tainted | a user-provided value |
+| polynomial-redos.js:130:2:130:31 | modifie ... g, "b") | polynomial-redos.js:5:16:5:32 | req.query.tainted | polynomial-redos.js:130:2:130:9 | modified | This $@ that depends on $@ may run slow on strings starting with 'c' and with many repetitions of 'c'. | polynomial-redos.js:130:21:130:22 | c+ | regular expression | polynomial-redos.js:5:16:5:32 | req.query.tainted | a user-provided value |
+| polynomial-redos.js:133:2:133:32 | modifie ... g, "b") | polynomial-redos.js:5:16:5:32 | req.query.tainted | polynomial-redos.js:133:2:133:10 | modified2 | This $@ that depends on $@ may run slow on strings starting with 'f' and with many repetitions of 'f'. | polynomial-redos.js:133:22:133:23 | f+ | regular expression | polynomial-redos.js:5:16:5:32 | req.query.tainted | a user-provided value |

javascript/ql/test/query-tests/Performance/ReDoS/polynomial-redos.js

@@ -125,4 +125,10 @@ app.use(function(req, res) {
 	})();
 
 	tainted.match(/(https?:\/\/[^\s]+)/gm); // OK
+
+	var modified = tainted.replace(/a/g, "b");
+	modified.replace(/cc+D/g, "b"); // NOT OK
+	
+	var modified2 = tainted.replace(/a|b|c|\d/g, "e");
+	modified2.replace(/ff+G/g, "b"); // NOT OK
 });