Fix feed validation output

repub/rss.py

@@ -1,5 +1,7 @@
-from datetime import datetime
-from time import mktime
+import re
+from calendar import timegm
+from datetime import UTC, datetime
+from email.utils import format_datetime
 
 import lxml.etree as ET
 import lxml.html
@@ -93,20 +95,54 @@ def serialize(root):
 
 def date_format(d):
     if d:
-        return d.strftime("%a, %d %b %Y %H:%M:%S %z")
+        return format_datetime(d.astimezone(UTC))
 
 
 def to_datetime(struct_time):
     if struct_time:
-        return datetime.fromtimestamp(mktime(struct_time))
+        return datetime.fromtimestamp(timegm(struct_time), tz=UTC)
 
 
 def normalize_date(struct_time):
     return date_format(to_datetime(struct_time))
 
 
+HTML_ATTRIBUTE_DENYLIST = frozenset({"contenteditable", "mode", "querystring"})
+
+
+def parse_html_fragment(raw_html):
+    if raw_html.strip() == "":
+        return None
+    return lxml.html.fragment_fromstring(raw_html, create_parent=True)
+
+
+def sanitize_html(raw_html: str) -> str:
+    fragment = parse_html_fragment(raw_html)
+    if fragment is None:
+        return raw_html
+    for el in fragment.iter():
+        for attr in HTML_ATTRIBUTE_DENYLIST:
+            el.attrib.pop(attr, None)
+    return (fragment.text or "") + "".join(
+        lxml.html.tostring(child, encoding="unicode") for child in fragment
+    )
+
+
+def plain_text_summary(raw_html: str | None, max_length: int = 4000) -> str | None:
+    if raw_html is None:
+        return None
+    fragment = parse_html_fragment(raw_html)
+    text = raw_html if fragment is None else fragment.text_content()
+    normalized = re.sub(r"\s+", " ", text).strip()
+    if normalized == "":
+        return None
+    return normalized[:max_length]
+
+
 def munge_cdata_html(raw_html, replace_link_fn) -> str:
-    html = lxml.html.fromstring(raw_html)
+    html = parse_html_fragment(raw_html)
+    if html is None:
+        return raw_html
     for el, attr, link, pos in html.iterlinks():
         if attr == "srcset":
             # these are a messy special case
@@ -133,4 +169,9 @@ def munge_cdata_html(raw_html, replace_link_fn) -> str:
             else:
                 new = cur[:pos] + new_link + cur[pos + len(link) :]
             el.set(attr, new)
-    return lxml.html.tostring(html, encoding="utf-8", pretty_print=True).decode("utf-8")
+    for el in html.iter():
+        for attr in HTML_ATTRIBUTE_DENYLIST:
+            el.attrib.pop(attr, None)
+    return (html.text or "") + "".join(
+        lxml.html.tostring(child, encoding="unicode") for child in html
+    )