Fixed: Only parse HTML when descending or showing links

mnot · Dec 26, 2023 · dcb01fa · dcb01fa
1 parent 2946031
commit dcb01fa
Show file tree

Hide file tree

Showing 3 changed files with 13 additions and 5 deletions.
diff --git a/redbot/formatter/html.py b/redbot/formatter/html.py
@@ -149,9 +149,7 @@ def format_body_sample(self, resource: HttpResource) -> Markup:
         except (TypeError, LookupError):
             uni_sample = sample.decode("utf-8", "replace")
         safe_sample = escape(uni_sample)
-        if self.config.getboolean("content_links", False) and hasattr(
-            resource, "links"
-        ):
+        if self.config.getboolean("content_links", False):
             for _, link_set in list(resource.links.items()):
                 for link in link_set:
                     if len(link) > 8000:  # avoid processing inline assets through regex

diff --git a/redbot/formatter/templates/response_finish.html b/redbot/formatter/templates/response_finish.html
@@ -57,7 +57,7 @@
             </div>
             {% endif %}
 
-            {% if is_resource and resource.link_count > 0 %}
+            {% if is_resource and resource.descendable() %}
             <div class='option'>
                 {{ descend_link }}
             </div>

diff --git a/redbot/resource/__init__.py b/redbot/resource/__init__.py
@@ -58,7 +58,8 @@ def __init__(self, config: SectionProxy, descend: bool = False) -> None:
         self._link_parser = link_parse.HTMLLinkParser(
             self.response, [self.process_link]
         )
-        self.response_content_processors.append(self._link_parser.feed_bytes)
+        if self.descend or config.getboolean("content_links", False):
+            self.response_content_processors.append(self._link_parser.feed_bytes)
 
     def run_active_checks(self) -> None:
         """
@@ -71,6 +72,15 @@ def run_active_checks(self) -> None:
         else:
             self.finish_check()
 
+    def descendable(self) -> bool:
+        """
+        Return whether this resource can be descended.
+        """
+        return (
+            self.response.headers.parsed.get("content-type", [None])[0]
+            in self._link_parser.link_parseable_types
+        )
+
     def add_check(self, *resources: RedFetcher) -> None:
         "Remember a subordinate check on one or more HttpResource instance."
         # pylint: disable=cell-var-from-loop