Validating and troubleshooting feeds

A feed that a browser renders can still be rejected by readers. Validation in the build is the only way to know before subscribers do.

The checks worth automating

# 1. Well-formedness: catches unescaped ampersands and stray markup
xmllint --noout feed.xml

# 2. Structural validation against a schema
xmllint --noout --schema rss-2.0.xsd feed.xml

# 3. Feed-specific rules the W3C validator enforces (dates, guids, absolute URLs)
#    Note that the hosted validator expects a publicly reachable feed URL.

# 4. Practical assertions the validator cannot know about
python - <<'PY'
import sys, feedparser
from datetime import datetime, timezone

f = feedparser.parse("feed.xml")
assert f.version, "not a recognised feed format"
assert f.feed.get("title"), "channel is missing a title"
assert f.feed.get("link"), "channel is missing a link"

newest = None
for e in f.entries:
    assert e.get("link", "").startswith(("http://", "https://")), "relative link"
    assert e.get("id") or e.get("guid"), "item without an identifier"
    if not e.get("published_parsed"):
        print("warning: item without a date:", e.get("title"))
    else:
        d = datetime(*e.published_parsed[:6], tzinfo=timezone.utc)
        assert d <= datetime.now(timezone.utc), "date in the future: " + str(d)
        newest = max(newest or d, d)

assert newest, "feed has no dated items"
print("feed looks publishable; newest item", newest.isoformat())
PY

# Validate in the build, so a broken feed never reaches subscribers
import subprocess, sys

def build_and_check(path: str) -> None:
    xml = render_feed()
    open(path, "w", encoding="utf-8").write(xml)
    r = subprocess.run(["xmllint", "--noout", path], capture_output=True, text=True)
    if r.returncode != 0:
        print(r.stderr, file=sys.stderr)
        raise SystemExit("feed is not well-formed XML")

Dates and identifiers, the two chronic problems

from datetime import datetime, timezone

# RSS 2.0 wants RFC 822 with a numeric offset and an English day and month name
def rfc822(dt: datetime) -> str:
    if dt.tzinfo is None:
        raise ValueError("a datetime without a timezone is a bug, not a default")
    return dt.astimezone(timezone.utc).strftime("%a, %d %b %Y %H:%M:%S +0000")

# Atom wants RFC 3339 and requires a timezone
def rfc3339(dt: datetime) -> str:
    if dt.tzinfo is None:
        raise ValueError("a datetime without a timezone is a bug, not a default")
    return dt.astimezone(timezone.utc).isoformat().replace("+00:00", "Z")

# A stable guid: derived from data that never changes, never from the title
def guid(release_id: str) -> str:
    return "release-" + release_id

• Always use a timezone-aware datetime. A naive datetime serialised as a local time shifts by an hour twice a year and by a day across regions.
• In RSS 2.0, use a numeric offset such as +0100. The colon form is ISO 8601 and is rejected by strict parsers.
• Do not use the connection between the guid and the URL unless the URL is genuinely permanent, which requires isPermaLink="true" to be honest.
• If you must change a guid, accept that every existing subscriber will see the affected items as new. Warn them in a post rather than doing it silently.

FAQ

Related

Generating a valid feed Podcast feeds and the Apple and Podcasting 2.0 tags

Last refreshed 2026-09-18.