Improve command-injection example and provide a fixed version.

javascript/ql/src/Security/CWE-078/CommandInjection.inc.qhelp

@@ -2,43 +2,52 @@
   "-//Semmle//qhelp//EN"
   "qhelp.dtd">
 <qhelp>
+
 <overview>
-<p>Code that passes user input directly to
-<code>require('child_process').exec</code>, or some other library
-routine that executes a command, allows the user to execute malicious
-code.</p>
-
+<p>Code that passes untrusted user input directly to
+<code>child_process.exec</code> or similar APIs that execute commands by
+spawning a shell allows the user to execute malicious code.</p>
 </overview>
+
 <recommendation>
+<p>If possible, use hard-coded string literals to specify the command to run.
+Instead of interpreting the user input directly as a shell command, examine the
+user input and then choose among hard-coded string literals.</p>
 
-<p>If possible, use hard-coded string literals to specify the command to run
-or library to load. Instead of passing the user input directly to the
-process or library function, examine the user input and then choose
-among hard-coded string literals.</p>
-
-<p>If the applicable libraries or commands cannot be determined at
-compile time, then add code to verify that the user input string is
-safe before using it.</p>
+<p>If the applicable libraries or commands cannot be determined until runtime,
+then add code to verify that the user input string is safe before using it.</p>
 
+<p>If possible, prefer APIs that run the commands directly rather than via a
+shell, and that accept command arguments as an array of strings rather than a
+single concatenated string. This is both safer and more portable.</p>
 </recommendation>
-<example>
 
-<p>The following example shows code that takes a shell script that can be changed
-maliciously by a user, and passes it straight to <code>child_process.exec</code>
-without examining it first.</p>
+<example>
+<p>The following example shows code that extracts a filename from an HTTP query
+parameter that may contain untrusted data, and then embeds it into a shell
+command to count its lines without examining it first.</p>
 
 <sample src="examples/command-injection.js" />
 
-</example>
-<references>
+<p>A malicious user can exploit this code to execute arbitrary shell commands by
+passing a filename like <code>foo.txt; rm -rf .</code>, which will first count
+the lines in <code>foo.txt</code> and then delete all files in the current
+directory.</p>
 
+<p>To avoid this potentially catastrophic loophole, use an API like
+<code>child_process.execFileSync</code> that does not spawn a shell by
+default:</p>
+
+<sample src="examples/command-injection_fixed.js" />
+</example>
+
+<references>
 <li>
 OWASP:
 <a href="https://www.owasp.org/index.php/Command_Injection">Command Injection</a>.
 </li>
-
 <!--  LocalWords:  CWE untrusted unsanitized Runtime
  -->
-
 </references>
+
 </qhelp>

javascript/ql/src/Security/CWE-078/examples/command-injection.js

@@ -1,9 +1,10 @@
 var cp = require("child_process"),
     http = require('http'),
+    path = require('path'),
     url = require('url');
 
 var server = http.createServer(function(req, res) {
-    let cmd = url.parse(req.url, true).query.path;
+    let file = path.basename(url.parse(req.url, true).query.path);
 
-    cp.exec(cmd); // BAD
+    cp.execSync(`wc -l ${file}`); // BAD
 });

javascript/ql/src/Security/CWE-078/examples/command-injection_fixed.js

@@ -0,0 +1,10 @@
+var cp = require("child_process"),
+    http = require('http'),
+    path = require('path'),
+    url = require('url');
+
+var server = http.createServer(function(req, res) {
+    let file = path.basename(url.parse(req.url, true).query.path);
+
+    cp.execFileSync('wc', ['-l', file]); // GOOD
+});