Content Filtering: Limited file protocol to just anchor hrefs

Updated allow list/purifier system to only allow file protocol use on
anchor hrefs to avoid potential security concerns with, after export,
content being auto loaded via interactive elements like
embeds/objects/videos etc...

Updated tests to cover.
Thanks to Gurmandeep Deol at Seneca Polytechnic for reporting.

app/Util/HtmlContentFilter.php

@@ -2,6 +2,7 @@
 
 namespace BookStack\Util;
 
+use BookStack\Util\HtmlPurifier\ConfiguredHtmlPurifier;
 use DOMAttr;
 use DOMElement;
 use DOMNodeList;

app/Util/HtmlPurifier/ConfiguredHtmlPurifier.php

@@ -1,13 +1,15 @@
 <?php
 
-namespace BookStack\Util;
+namespace BookStack\Util\HtmlPurifier;
 
 use BookStack\App\AppVersion;
+use BookStack\Util\HtmlPurifier\Filters\UriLimitFileProtocolToAnchors;
 use HTMLPurifier;
 use HTMLPurifier_Config;
 use HTMLPurifier_DefinitionCache_Serializer;
 use HTMLPurifier_HTML5Config;
 use HTMLPurifier_HTMLDefinition;
+use HTMLPurifier_URIDefinition;
 
 /**
  * Provides a configured HTML Purifier instance.
@@ -33,7 +35,13 @@ class ConfiguredHtmlPurifier
 
         $htmlDef = $config->getDefinition('HTML', true, true);
         if ($htmlDef instanceof HTMLPurifier_HTMLDefinition) {
-            $this->configureDefinition($htmlDef);
+            $this->configureHtmlDefinition($htmlDef);
+        }
+
+        /** @var \HTMLPurifier_URIDefinition $uriDef */
+        $uriDef = $config->getDefinition('URI', true, true);
+        if ($uriDef instanceof HTMLPurifier_URIDefinition) {
+            $this->configureUriDefinition($uriDef);
         }
 
         $this->purifier = new HTMLPurifier($config);
@@ -91,7 +99,7 @@ class ConfiguredHtmlPurifier
          // $config->set('Cache.DefinitionImpl', null); // Disable cache during testing
     }
 
-    public function configureDefinition(HTMLPurifier_HTMLDefinition $definition): void
+    protected function configureHtmlDefinition(HTMLPurifier_HTMLDefinition $definition): void
     {
         // Allow the object element
         $definition->addElement(
@@ -151,6 +159,11 @@ class ConfiguredHtmlPurifier
         $definition->addAttribute('a', 'data-mention-user-id', 'Number');
     }
 
+    protected function configureUriDefinition(HTMLPurifier_URIDefinition $definition): void
+    {
+        $definition->registerFilter(new UriLimitFileProtocolToAnchors());
+    }
+
     public function purify(string $html): string
     {
         return $this->purifier->purify($html);

app/Util/HtmlPurifier/Filters/UriLimitFileProtocolToAnchors.php

@@ -0,0 +1,55 @@
+<?php
+
+namespace BookStack\Util\HtmlPurifier\Filters;
+
+use HTMLPurifier_Config;
+use HTMLPurifier_Context;
+use HTMLPurifier_URI;
+use HTMLPurifier_URIFilter;
+
+/**
+ * Limits file:// URIs to only be used on anchor tags href attributes.
+ * This prevents use on iframes/embeds/images where they can be used to load external
+ * content on the network, triggering calls which may include NTLM auth hashes when in
+ * certain windows based environments.
+ */
+class UriLimitFileProtocolToAnchors extends HTMLPurifier_URIFilter
+{
+    /**
+     * @type string
+     */
+    public $name = 'LimitFileProtocolToAnchors';
+
+    /**
+     * @type bool
+     */
+    public $always_load = true;
+
+    /**
+     * @param HTMLPurifier_URI $uri
+     * @param HTMLPurifier_Config $config
+     * @param HTMLPurifier_Context $context
+     * @return bool
+     */
+    public function filter(&$uri, $config, $context)
+    {
+        // Ensure we're only filtering file:// URIs'
+        if ($uri->scheme !== 'file') {
+            return true;
+        }
+
+        $token = $context->get('CurrentToken', true);
+        $attr = $context->get('CurrentAttr', true);
+
+        // Only allow if used on hrefs on anchor tags
+        $isAnchor = $token && $token->name === 'a';
+        $isHref = $attr === 'href';
+        if ($isAnchor && $isHref) {
+            return true;
+        }
+
+        return false;
+    }
+}
+
+// vim: et sw=4 sts=4

tests/Entity/PageContentFilteringTest.php

@@ -464,6 +464,11 @@ HTML;
             '<div style="background:#FF0000;left:0;color:#00FFEE;">Hello!</div>' => '<div style="background:#FF0000;color:#00FFEE;">Hello!</div>',
             '<div style="color:#00FFEE;">Hello!<style>testinghello!</style></div>' => '<div style="color:#00FFEE;">Hello!</div>',
             '<div drawio-diagram="5332" another-attr="cat">Hello!</div>' => '<div drawio-diagram="5332">Hello!</div>',
+            '<iframe src="file://link/to/file" id="bkmrk-file-iframe"></iframe>' => '<iframe id="bkmrk-file-iframe"></iframe>',
+            '<embed src="file://link/to/file" id="bkmrk-file-embed"></embed>' => '<embed id="bkmrk-file-embed">',
+            '<object data="file://link/to/file" id="bkmrk-file-object"></object>' => '<object id="bkmrk-file-object">',
+            '<div id="bkmrk-file-img"><img src="file://link/to/file" alt="My local image"></div>' => '<div id="bkmrk-file-img"></div>',
+            '<div id="bkmrk-file-img"><img srcset="file://link/to/file" alt="My local image"></div>' => '<div id="bkmrk-file-img"></div>',
         ];
 
         config()->set('app.content_filtering', 'a');
@@ -476,6 +481,7 @@ HTML;
             $resp = $this->get($page->getUrl());
 
             $resp->assertSee($expected, false);
+            $resp->assertDontSee($input, false);
         }
     }
 
@@ -484,6 +490,7 @@ HTML;
         $testCasesExpectedByInput = [
             '<p><a href="https://example.com" target="_blank">New tab linkydoodle</a></p>',
             '<p><a href="https://example.com/user/1" data-mention-user-id="5">@mentionusertext</a></p>',
+            '<p><a href="file://link/to/file">Link to file</a></p>',
             '<details><summary>Hello</summary><p>Mydetailshere</p></details>',
         ];