Merge pull request #4928 from esbena/js/rewrite-multi-sanitization

Approved by asgerf
2026-04-30 11:15:13 +02:00 · 2021-01-18 05:11:42 -08:00
parent 545451e602 847687974f
commit fc2fe6cccb
4 changed files with 207 additions and 95 deletions
--- a/javascript/ql/test/query-tests/Security/CWE-116/IncompleteSanitization/IncompleteMultiCharacterSanitization.expected
+++ b/javascript/ql/test/query-tests/Security/CWE-116/IncompleteSanitization/IncompleteMultiCharacterSanitization.expected
@@ -1,33 +1,34 @@
-| tst-multi-character-sanitization.js:3:13:3:57 | content ... gi, "") | The replaced string may still contain a substring that starts matching at $@. | tst-multi-character-sanitization.js:3:30:3:49 | <.*cript.*\\/scrip.*> | <.*cript.*\\/scrip.*> |
-| tst-multi-character-sanitization.js:4:13:4:47 | content ... /g, "") | The replaced string may still contain a substring that starts matching at $@. | tst-multi-character-sanitization.js:4:30:4:40 |  on\\w+=".*" |  on\\w+=".*" |
-| tst-multi-character-sanitization.js:5:13:5:49 | content ... /g, "") | The replaced string may still contain a substring that starts matching at $@. | tst-multi-character-sanitization.js:5:30:5:42 |  on\\w+=\\'.*\\' |  on\\w+=\\'.*\\' |
-| tst-multi-character-sanitization.js:9:13:9:47 | content ... gi, "") | The replaced string may still contain a substring that starts matching at $@. | tst-multi-character-sanitization.js:9:30:9:39 | <.*cript.* | <.*cript.* |
-| tst-multi-character-sanitization.js:10:13:10:49 | content ... /g, "") | The replaced string may still contain a substring that starts matching at $@. | tst-multi-character-sanitization.js:10:30:10:42 | .on\\w+=.*".*" | .on\\w+=.*".*" |
-| tst-multi-character-sanitization.js:11:13:11:51 | content ... /g, "") | The replaced string may still contain a substring that starts matching at $@. | tst-multi-character-sanitization.js:11:30:11:44 | .on\\w+=.*\\'.*\\' | .on\\w+=.*\\'.*\\' |
-| tst-multi-character-sanitization.js:19:3:19:35 | respons ... pt, "") | The replaced string may still contain a substring that starts matching at $@. | tst-multi-character-sanitization.js:18:18:18:24 | <script | <script |
-| tst-multi-character-sanitization.js:25:10:25:40 | text.re ... /g, "") | The replaced string may still contain a substring that starts matching at $@. | tst-multi-character-sanitization.js:25:24:25:27 | <!-- | <!-- |
-| tst-multi-character-sanitization.js:38:8:38:30 | id.repl ... /g, "") | The replaced string may still contain a substring that starts matching at $@. | tst-multi-character-sanitization.js:38:20:38:23 | \\.\\. | \\.\\. |
-| tst-multi-character-sanitization.js:49:13:49:43 | req.url ... EL, "") | The replaced string may still contain a substring that starts matching at $@. | tst-multi-character-sanitization.js:48:21:48:31 | (\\/)?\\.\\.\\/ | (\\/)?\\.\\.\\/ |
-| tst-multi-character-sanitization.js:64:7:64:73 | x.repla ... /g, "") | The replaced string may still contain a substring that starts matching at $@. | tst-multi-character-sanitization.js:64:18:64:24 | <script | <script |
-| tst-multi-character-sanitization.js:66:7:66:56 | x.repla ... /g, "") | The replaced string may still contain a substring that starts matching at $@. | tst-multi-character-sanitization.js:66:18:66:49 | (\\/\|\\s)on\\w+=(\\'\|")?[^"]*(\\'\|")? | (\\/\|\\s)on\\w+=(\\'\|")?[^"]*(\\'\|")? |
-| tst-multi-character-sanitization.js:75:7:75:37 | x.repla ... gm, "") | The replaced string may still contain a substring that starts matching at $@. | tst-multi-character-sanitization.js:75:18:75:21 | <!-- | <!-- |
-| tst-multi-character-sanitization.js:77:7:77:36 | x.repla ... /g, "") | The replaced string may still contain a substring that starts matching at $@. | tst-multi-character-sanitization.js:77:18:77:29 | \\sng-[a-z-]+ | \\sng-[a-z-]+ |
-| tst-multi-character-sanitization.js:81:7:81:58 | x.repla ... /g, "") | The replaced string may still contain a substring that starts matching at $@. | tst-multi-character-sanitization.js:81:18:81:24 | <script | <script |
-| tst-multi-character-sanitization.js:81:7:81:58 | x.repla ... /g, "") | The replaced string may still contain a substring that starts matching at $@. | tst-multi-character-sanitization.js:81:36:81:39 | only | only |
-| tst-multi-character-sanitization.js:82:7:82:50 | x.repla ... /g, "") | The replaced string may still contain a substring that starts matching at $@. | tst-multi-character-sanitization.js:82:18:82:30 | <script async | <script async |
-| tst-multi-character-sanitization.js:83:7:83:63 | x.repla ... gi, "") | The replaced string may still contain a substring that starts matching at $@. | tst-multi-character-sanitization.js:83:18:83:21 | <!-- | <!-- |
-| tst-multi-character-sanitization.js:85:7:85:48 | x.repla ... /g, "") | The replaced string may still contain a substring that starts matching at $@. | tst-multi-character-sanitization.js:85:18:85:41 | \\x2E\\x2E\\x2F\\x2E\\x2E\\x2F | \\x2E\\x2E\\x2F\\x2E\\x2E\\x2F |
-| tst-multi-character-sanitization.js:87:7:87:47 | x.repla ... gi, "") | The replaced string may still contain a substring that starts matching at $@. | tst-multi-character-sanitization.js:87:18:87:24 | <script | <script |
-| tst-multi-character-sanitization.js:92:7:96:4 | x.repla ... ";\\n  }) | The replaced string may still contain a substring that starts matching at $@. | tst-multi-character-sanitization.js:92:18:92:24 | <script | <script |
-| tst-multi-character-sanitization.js:100:7:100:28 | x.repla ... /g, "") | The replaced string may still contain a substring that starts matching at $@. | tst-multi-character-sanitization.js:100:18:100:21 | \\.\\. | \\.\\. |
-| tst-multi-character-sanitization.js:101:7:101:30 | x.repla ... /g, "") | The replaced string may still contain a substring that starts matching at $@. | tst-multi-character-sanitization.js:101:18:101:23 | \\.\\.\\/ | \\.\\.\\/ |
-| tst-multi-character-sanitization.js:102:7:102:30 | x.repla ... /g, "") | The replaced string may still contain a substring that starts matching at $@. | tst-multi-character-sanitization.js:102:18:102:23 | \\/\\.\\. | \\/\\.\\. |
-| tst-multi-character-sanitization.js:104:7:104:58 | x.repla ... gi, "") | The replaced string may still contain a substring that starts matching at $@. | tst-multi-character-sanitization.js:104:18:104:24 | <script | <script |
-| tst-multi-character-sanitization.js:106:7:106:64 | x.repla ... gi, "") | The replaced string may still contain a substring that starts matching at $@. | tst-multi-character-sanitization.js:106:18:106:56 | <(script\|del)(?=[\\s>])[\\w\\W]*?<\\/\\1\\s*> | <(script\|del)(?=[\\s>])[\\w\\W]*?<\\/\\1\\s*> |
-| tst-multi-character-sanitization.js:107:7:107:62 | x.repla ... /g, "") | The replaced string may still contain a substring that starts matching at $@. | tst-multi-character-sanitization.js:107:18:107:55 | \\<script[\\s\\S]*?\\>[\\s\\S]*?\\<\\/script\\> | \\<script[\\s\\S]*?\\>[\\s\\S]*?\\<\\/script\\> |
-| tst-multi-character-sanitization.js:108:7:108:75 | x.repla ... gm, "") | The replaced string may still contain a substring that starts matching at $@. | tst-multi-character-sanitization.js:108:18:108:67 | <(script\|style\|title)[^<]+<\\/(script\|style\|title)> | <(script\|style\|title)[^<]+<\\/(script\|style\|title)> |
-| tst-multi-character-sanitization.js:109:7:109:58 | x.repla ... gi, "") | The replaced string may still contain a substring that starts matching at $@. | tst-multi-character-sanitization.js:109:18:109:24 | <script | <script |
-| tst-multi-character-sanitization.js:110:7:110:50 | x.repla ... gi, "") | The replaced string may still contain a substring that starts matching at $@. | tst-multi-character-sanitization.js:110:18:110:24 | <script | <script |
-| tst-multi-character-sanitization.js:111:7:111:32 | x.repla ... /g, "") | The replaced string may still contain a substring that starts matching at $@. | tst-multi-character-sanitization.js:111:20:111:23 | <!-- | <!-- |
-| tst-multi-character-sanitization.js:112:7:112:50 | x.repla ... /g, "") | The replaced string may still contain a substring that starts matching at $@. | tst-multi-character-sanitization.js:112:18:112:43 | require\\('\\.\\.\\/common'\\); | require\\('\\.\\.\\/common'\\); |
-| tst-multi-character-sanitization.js:113:7:113:41 | x.repla ... /g, "") | The replaced string may still contain a substring that starts matching at $@. | tst-multi-character-sanitization.js:113:18:113:34 | \\.\\.\\/\\.\\.\\/lib\\/ | \\.\\.\\/\\.\\.\\/lib\\/ |
+| tst-multi-character-sanitization.js:3:13:3:57 | content ... gi, "") | This string may still contain $@, which may cause a HTML element injection vulnerability. | tst-multi-character-sanitization.js:3:30:3:30 | < | <cript |
+| tst-multi-character-sanitization.js:4:13:4:47 | content ... /g, "") | This string may still contain $@, which may cause a HTML attribute injection vulnerability. | tst-multi-character-sanitization.js:4:30:4:40 |  on\\w+=".*" | on |
+| tst-multi-character-sanitization.js:5:13:5:49 | content ... /g, "") | This string may still contain $@, which may cause a HTML attribute injection vulnerability. | tst-multi-character-sanitization.js:5:30:5:42 |  on\\w+=\\'.*\\' | on |
+| tst-multi-character-sanitization.js:9:13:9:47 | content ... gi, "") | This string may still contain $@, which may cause a HTML element injection vulnerability. | tst-multi-character-sanitization.js:9:30:9:30 | < | <cript |
+| tst-multi-character-sanitization.js:10:13:10:49 | content ... /g, "") | This string may still contain $@, which may cause a HTML attribute injection vulnerability. | tst-multi-character-sanitization.js:10:30:10:42 | .on\\w+=.*".*" | on |
+| tst-multi-character-sanitization.js:11:13:11:51 | content ... /g, "") | This string may still contain $@, which may cause a HTML attribute injection vulnerability. | tst-multi-character-sanitization.js:11:30:11:44 | .on\\w+=.*\\'.*\\' | on |
+| tst-multi-character-sanitization.js:19:3:19:35 | respons ... pt, "") | This string may still contain $@, which may cause a HTML element injection vulnerability. | tst-multi-character-sanitization.js:18:18:18:24 | <script | <script |
+| tst-multi-character-sanitization.js:25:10:25:40 | text.re ... /g, "") | This string may still contain $@, which may cause a HTML element injection vulnerability. | tst-multi-character-sanitization.js:25:24:25:27 | <!-- | <!-- |
+| tst-multi-character-sanitization.js:49:13:49:43 | req.url ... EL, "") | This string may still contain $@, which may cause a path injection vulnerability. | tst-multi-character-sanitization.js:48:22:48:23 | \\/ | /.. |
+| tst-multi-character-sanitization.js:49:13:49:43 | req.url ... EL, "") | This string may still contain $@, which may cause a path injection vulnerability. | tst-multi-character-sanitization.js:48:26:48:27 | \\. | ../ |
+| tst-multi-character-sanitization.js:64:7:64:73 | x.repla ... /g, "") | This string may still contain $@, which may cause a HTML element injection vulnerability. | tst-multi-character-sanitization.js:64:18:64:24 | <script | <script |
+| tst-multi-character-sanitization.js:66:7:66:56 | x.repla ... /g, "") | This string may still contain $@, which may cause a HTML attribute injection vulnerability. | tst-multi-character-sanitization.js:66:18:66:49 | (\\/\|\\s)on\\w+=(\\'\|")?[^"]*(\\'\|")? | on |
+| tst-multi-character-sanitization.js:75:7:75:37 | x.repla ... gm, "") | This string may still contain $@, which may cause a HTML element injection vulnerability. | tst-multi-character-sanitization.js:75:18:75:21 | <!-- | <!-- |
+| tst-multi-character-sanitization.js:76:7:76:35 | x.repla ... +/, "") | This string may still contain $@, which may cause a HTML attribute injection vulnerability. | tst-multi-character-sanitization.js:76:18:76:29 | \\sng-[a-z-]+ | ng- |
+| tst-multi-character-sanitization.js:77:7:77:36 | x.repla ... /g, "") | This string may still contain $@, which may cause a HTML attribute injection vulnerability. | tst-multi-character-sanitization.js:77:18:77:29 | \\sng-[a-z-]+ | ng- |
+| tst-multi-character-sanitization.js:81:7:81:58 | x.repla ... /g, "") | This string may still contain $@, which may cause a HTML attribute injection vulnerability. | tst-multi-character-sanitization.js:81:36:81:39 | only | on |
+| tst-multi-character-sanitization.js:81:7:81:58 | x.repla ... /g, "") | This string may still contain $@, which may cause a HTML element injection vulnerability. | tst-multi-character-sanitization.js:81:18:81:24 | <script | <script |
+| tst-multi-character-sanitization.js:83:7:83:63 | x.repla ... gi, "") | This string may still contain $@, which may cause a HTML element injection vulnerability. | tst-multi-character-sanitization.js:83:18:83:21 | <!-- | <!-- |
+| tst-multi-character-sanitization.js:85:7:85:48 | x.repla ... /g, "") | This string may still contain $@, which may cause a path injection vulnerability. | tst-multi-character-sanitization.js:85:18:85:21 | \\x2E | ../ |
+| tst-multi-character-sanitization.js:87:7:87:47 | x.repla ... gi, "") | This string may still contain $@, which may cause a HTML element injection vulnerability. | tst-multi-character-sanitization.js:87:18:87:24 | <script | <script |
+| tst-multi-character-sanitization.js:92:7:96:4 | x.repla ... ";\\n  }) | This string may still contain $@, which may cause a HTML element injection vulnerability. | tst-multi-character-sanitization.js:92:18:92:24 | <script | <script |
+| tst-multi-character-sanitization.js:101:7:101:30 | x.repla ... /g, "") | This string may still contain $@, which may cause a path injection vulnerability. | tst-multi-character-sanitization.js:101:18:101:19 | \\. | ../ |
+| tst-multi-character-sanitization.js:102:7:102:30 | x.repla ... /g, "") | This string may still contain $@, which may cause a path injection vulnerability. | tst-multi-character-sanitization.js:102:18:102:19 | \\/ | /.. |
+| tst-multi-character-sanitization.js:104:7:104:58 | x.repla ... gi, "") | This string may still contain $@, which may cause a HTML element injection vulnerability. | tst-multi-character-sanitization.js:104:18:104:24 | <script | <script |
+| tst-multi-character-sanitization.js:106:7:106:64 | x.repla ... gi, "") | This string may still contain $@, which may cause a HTML element injection vulnerability. | tst-multi-character-sanitization.js:106:18:106:18 | < | <script |
+| tst-multi-character-sanitization.js:107:7:107:62 | x.repla ... /g, "") | This string may still contain $@, which may cause a HTML element injection vulnerability. | tst-multi-character-sanitization.js:107:18:107:19 | \\< | <script |
+| tst-multi-character-sanitization.js:108:7:108:75 | x.repla ... gm, "") | This string may still contain $@, which may cause a HTML element injection vulnerability. | tst-multi-character-sanitization.js:108:18:108:18 | < | <script |
+| tst-multi-character-sanitization.js:109:7:109:58 | x.repla ... gi, "") | This string may still contain $@, which may cause a HTML element injection vulnerability. | tst-multi-character-sanitization.js:109:18:109:24 | <script | <script |
+| tst-multi-character-sanitization.js:110:7:110:50 | x.repla ... gi, "") | This string may still contain $@, which may cause a HTML element injection vulnerability. | tst-multi-character-sanitization.js:110:18:110:24 | <script | <script |
+| tst-multi-character-sanitization.js:111:7:111:32 | x.repla ... /g, "") | This string may still contain $@, which may cause a HTML element injection vulnerability. | tst-multi-character-sanitization.js:111:20:111:23 | <!-- | <!-- |
+| tst-multi-character-sanitization.js:126:7:129:34 | x\\n    . ... //, "") | This string may still contain $@, which may cause a path injection vulnerability. | tst-multi-character-sanitization.js:129:21:129:22 | \\/ | /.. |
+| tst-multi-character-sanitization.js:135:2:135:44 | content ... gi, "") | This string may still contain $@, which may cause a HTML element injection vulnerability. | tst-multi-character-sanitization.js:135:19:135:25 | <script | <script |
+| tst-multi-character-sanitization.js:136:2:136:46 | content ... gi, "") | This string may still contain $@, which may cause a HTML element injection vulnerability. | tst-multi-character-sanitization.js:136:19:136:19 | < | <script |
+| tst-multi-character-sanitization.js:138:2:138:48 | content ... gi, "") | This string may still contain $@, which may cause a HTML element injection vulnerability. | tst-multi-character-sanitization.js:138:21:138:21 | < | <script |
--- a/javascript/ql/test/query-tests/Security/CWE-116/IncompleteSanitization/tst-multi-character-sanitization.js
+++ b/javascript/ql/test/query-tests/Security/CWE-116/IncompleteSanitization/tst-multi-character-sanitization.js
@@ -35,7 +35,7 @@

 // CVE-2019-10767
 (function(id) {
-  id = id.replace(/\.\./g, ""); // NOT OK
+  id = id.replace(/\.\./g, ""); // OK (can not contain '..' afterwards)
  return id;
 });
 (function(id) {
@@ -73,13 +73,13 @@
  x = x.replace(/<li><\/li>/gi, ""); // OK

  x = x.replace(/<!--(.*?)-->/gm, ""); // NOT OK
-  x = x.replace(/\sng-[a-z-]+/, ""); // OK (single ng-attribute, should be flagged by some other query!)
+  x = x.replace(/\sng-[a-z-]+/, ""); // NOT OK
  x = x.replace(/\sng-[a-z-]+/g, ""); // NOT OK (ng-attributes)

  x = x.replace(/(<!--\[CDATA\[|\]\]-->)/g, "\n"); // OK: not a sanitizer

-  x = x.replace(/<script.+desktop\-only.+<\/script>/g, ""); // OK, but still flagged [INCONSISTENCY]
-  x = x.replace(/<script async.+?<\/script>/g, ""); // OK, but still flagged [INCONSISTENCY]
+  x = x.replace(/<script.+desktop\-only.+<\/script>/g, ""); // OK [INCONSISTENCY]
+  x = x.replace(/<script async.+?<\/script>/g, ""); // OK
  x = x.replace(/<!--[\s\S]*?-->|<\?(?:php)?[\s\S]*?\?>/gi, ""); // NOT OK

  x = x.replace(/\x2E\x2E\x2F\x2E\x2E\x2F/g, ""); // NOT OK (matches "../../")
@@ -97,7 +97,7 @@

  x = x.replace(/<\/?([a-z][a-z0-9]*)\b[^>]*>/gi, ""); // NOT OK [INCONSISTENCY]

-  x = x.replace(/\.\./g, ""); // NOT OK
+  x = x.replace(/\.\./g, ""); // OK
  x = x.replace(/\.\.\//g, ""); // NOT OK
  x = x.replace(/\/\.\./g, ""); // NOT OK

@@ -109,8 +109,31 @@
  x = x.replace(/<script[^>]*>([\s\S]*?)<\/script>/gi, ""); // NOT OK
  x = x.replace(/<script[\s\S]*?<\/script>/gi, ""); // NOT OK
  x = x.replace(/ ?<!-- ?/g, ""); // NOT OK
-  x = x.replace(/require\('\.\.\/common'\);/g, ""); // OK [INCONSISTENCY] permit alphanum-suffix after the dots?
-  x = x.replace(/\.\.\/\.\.\/lib\//g, ""); // OK [INCONSISTENCY] permit alphanum-suffix after the dots?
+  x = x.replace(/require\('\.\.\/common'\);/g, ""); // OK
+  x = x.replace(/\.\.\/\.\.\/lib\//g, ""); // OK
+
+  while (x.indexOf(".") !== -1) {
+    x = x
+      .replace(/^\.\//, "")
+      .replace(/\/\.\//, "/")
+      .replace(/[^\/]*\/\.\.\//, ""); // OK
+  }
+
+  x = x.replace(/([^.\s]+\.)+/, ""); // OK
+
+  x = x.replace(/<!\-\-DEVEL[\d\D]*?DEVEL\-\->/g, ""); // OK
+
+  x = x
+    .replace(/^\.\//, "")
+    .replace(/\/\.\//, "/")
+    .replace(/[^\/]*\/\.\.\//, ""); // NOT OK

  return x;
 });
+
+(function (content) {
+	content.replace(/<script.*\/script>/gi, ""); // NOT OK
+	content.replace(/<(script).*\/script>/gi, ""); // NOT OK
+	content.replace(/.+<(script).*\/script>/gi, ""); // OK
+	content.replace(/.*<(script).*\/script>/gi, ""); // NOT OK
+});