When a tag that isn't compatible with Telegram closes, don't force all other markup to reset (but still allow handle_endtag(None) to close all markup).

Author: nmlorg

metabot/util/html.py

@@ -105,6 +105,9 @@ def handle_starttag(self, tag, attrs):  # pylint: disable=too-many-branches,too-
             self.__pieces.append(f'<{tag} {attribute_name}="{escape(attribute_value)}">')
 
     def handle_endtag(self, tag):
+        if tag and tag not in self.__stack:
+            return
+
         while self.__stack:
             lasttag = self.__stack.pop()
             self.__pieces.append(f'</{lasttag}>')

metabot/util/test_html.py

@@ -149,3 +149,10 @@ def test_MessageEntity_cpp_quirks():  # pylint: disable=invalid-name
     assert html.sanitize(
         '<code class="language-python">code</code>') == '<code class="language-python">code</code>'
     assert html.sanitize('<code class="other">code</code>') == '<code>code</code>'
+
+
+def test_wellformed_close():
+    """Verify well-formed but Telegram-incompatible markup does not close other tags."""
+
+    assert html.sanitize(
+        'hi <code>there <span>friend</span> still') == 'hi <code>there friend still</code>'