Java: Calling openStream on URLs created from remote source can lead to local file disclosure.

Author: p-

java/ql/src/Security/CWE/CWE-036/OpenStream.java

@@ -0,0 +1,8 @@
+public class TestServlet extends HttpServlet {
+	protected void doGet(HttpServletRequest request, HttpServletResponse response)
+	throws ServletException, IOException {
+		// BAD: a URL from a remote source is opened with URL#openStream()
+        URL url = new URL(request.getParameter("url"));
+        InputStream inputStream = new URL(url).openStream();
+	}
+}

java/ql/src/Security/CWE/CWE-036/OpenStream.qhelp

@@ -0,0 +1,39 @@
+<!DOCTYPE qhelp PUBLIC
+  "-//Semmle//qhelp//EN"
+  "qhelp.dtd">
+<qhelp>
+<overview>
+<p>Calling <code>openStream</code> on URLs created from remote source can lead to local file disclosure.</p>
+
+<p>If <code>openStream</code> is called on a <code>java.net.URL</code>, that was created from a remote source
+an attacker can try to pass absolute URLs starting with <code>file://</code> or <code>jar://</code> to access
+local resources in addition to remote ones.</p>
+
+</overview>
+<recommendation>
+
+<p>When you construct a URL using <code>java.net.URL</code> from a remote source, make sure
+to not call openStream on it. Instead fetch the URL with a HTTP Client to access its content.
+Also validate that the URL uses the correct protocol and host combination.</p>
+
+</recommendation>
+<example>
+
+<p>The following example shows an URL that is constructed from a request parameter. Afterwards <code>openStream</code>
+is called on the URL, potentially leading to a local file access.</p>
+
+<sample src="OpenStream.java" />
+
+</example>
+<references>
+
+<li>Java Platform, Standard Edition 11, API Specification: 
+<a href="https://docs.oracle.com/en/java/javase/11/docs/api/java.base/java/net/URL.html">
+Class URL</a>.</li>
+<li>
+
+<!--  LocalWords:  CWE
+ -->
+
+</references>
+</qhelp>

java/ql/src/Security/CWE/CWE-036/OpenStream.ql

@@ -0,0 +1,50 @@
+import java
+import semmle.code.java.dataflow.TaintTracking
+import semmle.code.java.dataflow.FlowSources
+import DataFlow::PathGraph
+
+class URLConstructor extends ClassInstanceExpr {
+  URLConstructor() { this.getConstructor().getDeclaringType().getQualifiedName() = "java.net.URL" }
+
+  Expr stringArg() {
+    // Query only in URL's that were constructed by calling the single parameter string constructor.
+    if
+      this.getConstructor().getNumberOfParameters() = 1 and
+      this.getConstructor().getParameter(0).getType().getName() = "String"
+    then result = this.getArgument(0)
+    else none()
+  }
+}
+
+class URLOpenStreamMethod extends Method {
+  URLOpenStreamMethod() {
+    this.getDeclaringType().getQualifiedName() = "java.net.URL" and
+    this.getName() = "openStream"
+  }
+}
+
+class RemoteURLToOpenStreamFlowConfig extends TaintTracking::Configuration {
+  RemoteURLToOpenStreamFlowConfig() { this = "OpenStream::RemoteURLToOpenStreamFlowConfig" }
+
+  override predicate isSource(DataFlow::Node source) { source instanceof RemoteFlowSource }
+
+  override predicate isSink(DataFlow::Node sink) {
+    exists(MethodAccess m |
+      sink.asExpr() = m.getQualifier() and m.getMethod() instanceof URLOpenStreamMethod
+    )
+  }
+
+  override predicate isAdditionalTaintStep(DataFlow::Node node1, DataFlow::Node node2) {
+    exists(URLConstructor u |
+      node1.asExpr() = u.stringArg() and
+      node2.asExpr() = u
+    )
+  }
+}
+
+from DataFlow::PathNode source, DataFlow::PathNode sink, MethodAccess call
+where
+  sink.getNode().asExpr() = call.getQualifier() and
+  any(RemoteURLToOpenStreamFlowConfig c).hasFlowPath(source, sink)
+select call, source, sink,
+  "URL on which openStream is called may have been constructed from remote source"