Permit Mastodon's weird HTML through

1 files changed, 22 insertions, 2 deletions

diff --git a/core/html.py b/core/html.py
index e63dda3..fd41a50 100644
--- a/core/html.py
+++ b/core/html.py
@@ -1,11 +1,31 @@
 import bleach
+from bleach.linkifier import LinkifyFilter
 from django.utils.safestring import mark_safe
 
 
+def allow_a(tag: str, name: str, value: str):
+    if name in ["href", "title", "class"]:
+        return True
+    elif name == "rel":
+        # Only allow rel attributes with a small subset of values
+        # (we're defending against, for example, rel=me)
+        rel_values = value.split()
+        if all(v in ["nofollow", "noopener", "noreferrer", "tag"] for v in rel_values):
+            return True
+    return False
+
+
 def sanitize_post(post_html: str) -> str:
     """
     Only allows a, br, p and span tags, and class attributes.
     """
-    return mark_safe(
-        bleach.clean(post_html, tags=["a", "br", "p", "span"], attributes=["class"])
+    cleaner = bleach.Cleaner(
+        tags=["a", "br", "p", "span"],
+        attributes={  # type:ignore
+            "a": allow_a,
+            "p": ["class"],
+            "span": ["class"],
+        },
+        filters=[LinkifyFilter],
     )
+    return mark_safe(cleaner.clean(post_html))