Sanitize: element-aware URI scheme check so data: can be disallowed without losing data:image

src/Sanitize.php

@@ -677,7 +677,7 @@ public function replace_urls(DOMDocument $document, string $tag, $attributes)
                         $value = $this->registry->call(Misc::class, 'absolutize_url', [$element->getAttribute($attribute), $this->base]);
                         if ($value !== false) {
                             // Block disallowed URI protocols (e.g. javascript:), otherwise force HTTPS where applicable
-                            $value = ($this->disallowed_uri_schemes !== [] && !$this->is_allowed_scheme($value))
+                            $value = ($this->disallowed_uri_schemes !== [] && !$this->is_allowed_uri($tag, $attribute, $value))
                                 ? 'unsafe:' . $value
                                 : $this->https_url($value);
                             $element->setAttribute($attribute, $value);
@@ -779,6 +779,36 @@ public function is_allowed_scheme(string $uri): bool
         return !in_array($scheme, $this->disallowed_uri_schemes, true);
     }
 
+    /**
+     * Elements/attributes on which a `data:` URI carrying an image is still allowed
+     * even when the `data` scheme is otherwise disallowed. Such URIs are inert
+     * (an `<img>` never executes script), whereas `data:` on navigational or framing
+     * attributes (`<a href>`, `<iframe src>`, ...) may load attacker-controlled HTML.
+     * @var array<string, list<string>>
+     */
+    private const DATA_IMAGE_URI_ALLOWED = [
+        'image' => ['src'],
+        'img' => ['src'],
+        'video' => ['poster'],
+    ];
+
+    /**
+     * Like {@see is_allowed_scheme()}, but aware of the element and attribute the URI sits on,
+     * so that `data:image/...` can stay allowed on image-bearing attributes while `data:` is
+     * blocked elsewhere (e.g. `<a href="data:text/html,...">`, `<iframe src="data:...">`).
+     */
+    public function is_allowed_uri(string $tag, string $attribute, string $uri): bool
+    {
+        $pos = strpos($uri, ':');
+        if ($pos !== false && strtolower(substr($uri, 0, $pos)) === 'data'
+            && in_array('data', $this->disallowed_uri_schemes, true)
+            && in_array($attribute, self::DATA_IMAGE_URI_ALLOWED[strtolower($tag)] ?? [], true)
+            && preg_match('%^data:image/[a-z0-9.+-]+[;,]%i', $uri) === 1) {
+            return true;
+        }
+        return $this->is_allowed_scheme($uri);
+    }
+
     /**
      * Block disallowed URI schemes (protocols) on elements carrying an href attribute.
      *

tests/Unit/SanitizeTest.php

@@ -304,6 +304,48 @@ public static function disallowedUriSchemesProvider(): iterable
             '<svg xmlns:xlink="http://www.w3.org/1999/xlink"><a xlink:href="https://example.com/page">x</a></svg>',
             ['javascript'],
         ];
+
+        yield 'data scheme in href' => [
+            '<a href="data:text/html,<script>alert(1)</script>">Click me</a>',
+            '<a href="unsafe:data:text/html,%3Cscript%3Ealert(1)%3C/script%3E">Click me</a>',
+            ['javascript', 'data'],
+        ];
+
+        yield 'data scheme in iframe src' => [
+            '<iframe src="data:text/html,x"></iframe>',
+            '<iframe src="unsafe:data:text/html,x" sandbox="allow-scripts allow-same-origin"></iframe>',
+            ['javascript', 'data'],
+        ];
+
+        yield 'data scheme case insensitive' => [
+            '<a href="DATA:text/html,x">Click me</a>',
+            '<a href="unsafe:data:text/html,x">Click me</a>',
+            ['javascript', 'data'],
+        ];
+
+        yield 'data image kept on img src' => [
+            '<img src="data:image/png;base64,iVBORw0KGgo=">',
+            '<img src="data:image/png;base64,iVBORw0KGgo=">',
+            ['javascript', 'data'],
+        ];
+
+        yield 'data non-image blocked on img src' => [
+            '<img src="data:text/html,x">',
+            '<img src="unsafe:data:text/html,x">',
+            ['javascript', 'data'],
+        ];
+
+        yield 'data image blocked on href' => [
+            '<a href="data:image/png;base64,iVBORw0KGgo=">Click me</a>',
+            '<a href="unsafe:data:image/png;base64,iVBORw0KGgo=">Click me</a>',
+            ['javascript', 'data'],
+        ];
+
+        yield 'data scheme unaffected when not disallowed' => [
+            '<a href="data:text/html,x">Click me</a>',
+            '<a href="data:text/html,x">Click me</a>',
+            ['javascript'],
+        ];
     }
 
     /**