fix: preserve validated embedded raster icons in SVG

AppImage/scripts/local_app_icons.py

@@ -166,7 +166,7 @@ def _svg(body):
         raise IconError('invalid SVG') from exc
     ns = 'http://www.w3.org/2000/svg'
     tags = {'svg', 'g', 'defs', 'path', 'rect', 'circle', 'ellipse', 'line', 'polyline',
-            'polygon', 'linearGradient', 'radialGradient', 'stop', 'clipPath', 'title', 'desc'}
+            'polygon', 'linearGradient', 'radialGradient', 'stop', 'clipPath', 'title', 'desc', 'image'}
     attrs = {'viewBox', 'width', 'height', 'x', 'y', 'x1', 'x2', 'y1', 'y2', 'cx', 'cy',
              'r', 'rx', 'ry', 'd', 'points', 'fill', 'stroke', 'stroke-width', 'opacity',
              'fill-opacity', 'stroke-opacity', 'fill-rule', 'clip-rule', 'stroke-linecap',
@@ -175,20 +175,40 @@ def _svg(body):
     if original.tag not in ('svg', '{' + ns + '}svg'):
         raise IconError('invalid SVG root')
     count = 0
+    images = 0
     def clean(node, depth=0):
-        nonlocal count
+        nonlocal count, images
         count += 1
         if depth > 32 or count > 2048:
             raise IconError('SVG complexity limit')
         name = node.tag.removeprefix('{' + ns + '}')
         if name not in tags:
             return None
+        embedded = None
+        if name == 'image':
+            refs = {value for key, value in node.attrib.items()
+                    if key in ('href', '{http://www.w3.org/1999/xlink}href')}
+            if len(refs) != 1:
+                return None
+            match = re.fullmatch(r'data:(image/(?:png|jpeg|gif|webp|x-icon|vnd\.microsoft\.icon));base64,([A-Za-z0-9+/=]+)', refs.pop())
+            if not match:
+                return None  # No external resource or recursive SVG data URI.
+            images += 1
+            if images > 1:
+                raise IconError('SVG embedded image limit')
+            try:
+                raster = base64.b64decode(match[2], validate=True)
+            except ValueError as exc:
+                raise IconError('invalid embedded image') from exc
+            raster, _mime = normalize(raster, match[1])
+            embedded = 'data:image/png;base64,' + base64.b64encode(raster).decode()
         out = xml.Element('{' + ns + '}' + name)
         for key, value in node.attrib.items():
             if key not in attrs or len(value) > 32768:
                 continue
             # Only local fragment paint/clip references; no CSS, href, events,
-            # foreign namespaces, entities, animation, images, filters or use.
+            # foreign namespaces, entities, animation, filters or use.
+            # A single validated raster data URI is installed separately below.
             if re.search(r'url\s*\(', value, re.I):
                 if not re.fullmatch(r'url\(#[A-Za-z_][\w.-]*\)', value):
                     continue
@@ -198,6 +218,8 @@ def _svg(body):
                 if not re.fullmatch(r'\d+(?:\.\d+)?(?:px)?', value) or not 0 < float(value.removesuffix('px')) <= MAX_DIMENSION:
                     continue
             out.set(key, value)
+        if embedded:
+            out.set('href', embedded)
         if name in ('title', 'desc'):
             out.text = (node.text or '')[:512]
         for child in node:
@@ -206,6 +228,13 @@ def _svg(body):
                 out.append(cleaned)
         return out
     result = clean(original)
+    def has_drawing(node):
+        name = node.tag.removeprefix('{' + ns + '}')
+        if name in ('defs', 'clipPath'):
+            return False
+        return name in {'path', 'rect', 'circle', 'ellipse', 'line', 'polyline', 'polygon', 'image'} or any(has_drawing(child) for child in node)
+    if not has_drawing(result):
+        raise IconError('SVG contains no supported drawing')
     xml.register_namespace('', ns)
     output = xml.tostring(result, encoding='utf-8')
     if len(output) > MAX_BYTES:
@@ -230,7 +259,9 @@ def normalize(body, mime):
                 raise IconError('image dimensions too large')
             picture.load()
             out = io.BytesIO()
-            picture.convert('RGBA').save(out, format='PNG')
+            converted = picture.convert('RGBA')
+            converted.info.clear()
+            converted.save(out, format='PNG')
         result = out.getvalue()
         if len(result) > MAX_BYTES:
             raise IconError('normalized icon too large')

AppImage/scripts/tests/test_local_app_icons.py

@@ -98,6 +98,42 @@ class Icons(unittest.TestCase):
             with self.assertRaises(icons.IconError):
                 icons.normalize(data, 'image/svg+xml')
 
+    def test_svg_embedded_raster_is_validated_and_preserved_without_external_fetch(self):
+        from xml.etree import ElementTree as xml
+        embedded = base64.b64encode(picture()).decode()
+        raw = f'<svg width="88" height="88" viewBox="0 0 88 88"><image x="4" y="4" width="80" height="80" href="data:image/png;base64,{embedded}"/></svg>'.encode()
+        with patch.object(icons, '_open') as network:
+            data, mime = icons.normalize(raw, 'image/svg+xml')
+            network.assert_not_called()
+        image = xml.fromstring(data).find('{http://www.w3.org/2000/svg}image')
+        self.assertIsNotNone(image, 'sanitizing must not report an empty SVG as a found icon')
+        self.assertEqual(image.get('x'), '4')
+        self.assertEqual(image.get('width'), '80')
+        decoded = base64.b64decode(image.get('href').split(',', 1)[1], validate=True)
+        self.assertEqual(icons.normalize(decoded, 'image/png')[0], decoded)
+        self.assertEqual(icons.normalize(data, mime), (data, mime))
+
+    def test_raster_normalization_discards_source_metadata(self):
+        source = io.BytesIO()
+        Image.new('RGB', (2, 2), 'red').save(source, format='PNG', icc_profile=b'fixture-private-metadata')
+        data, _mime = icons.normalize(source.getvalue(), 'image/png')
+        with Image.open(io.BytesIO(data)) as result:
+            self.assertNotIn('icc_profile', result.info)
+
+    def test_svg_empty_invalid_or_recursive_embedded_images_are_rejected(self):
+        nested = base64.b64encode(b'<svg onload="bad()"/>').decode()
+        for raw in [b'<svg/>', b'<svg><script>bad()</script></svg>',
+                    b'<svg><image href="https://outside.test/icon"/></svg>',
+                    b'<svg><image href="data:image/png;base64,bm90IGFuIGltYWdl"/></svg>',
+                    f'<svg><image href="data:image/svg+xml;base64,{nested}"/></svg>'.encode()]:
+            with self.assertRaises(icons.IconError), patch.object(icons, '_open') as network:
+                icons.normalize(raw, 'image/svg+xml')
+            network.assert_not_called()
+        embedded = base64.b64encode(picture()).decode()
+        image = f'<image href="data:image/png;base64,{embedded}"/>'
+        with self.assertRaises(icons.IconError):
+            icons.normalize(('<svg>' + image * 2 + '</svg>').encode(), 'image/svg+xml')
+
     def test_explicit_lookup_derives_address_rechecks_and_never_writes_apps(self):
         with patch.object(icons, '_run', side_effect=self.resolve) as run, patch.object(icons, '_worker', return_value=(self.png, 'image/png')) as worker:
             result = icons.discover(101, self.payload)