Bound the author-email match to avoid quadratic parsing on hostile feeds - #580
Open
arpitjain099 wants to merge 1 commit into
Open
Bound the author-email match to avoid quadratic parsing on hostile feeds#580arpitjain099 wants to merge 1 commit into
arpitjain099 wants to merge 1 commit into
Conversation
The author-email pattern's domain part backtracks quadratically on long, almost-matching strings, and feeds are untrusted input. A crafted <author> or dc:creator value around 32 KB makes feedparser.parse() burn several seconds of CPU, and it scales O(N^2), so a larger value is worse. Guard the length before running the search: values longer than a name plus an RFC 5321 sized address are not real author strings, so skip the match for them. Normal author and email parsing is unchanged. Signed-off-by: Arpit Jain <arpitjain099@gmail.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Feeds are untrusted input, so any regex run over feed-controlled text needs to stay linear. The author-email pattern in
mixin.pydoes not: its domain alternative(([a-zA-Z0-9-]+\.)+)overlaps the trailing label group, so on a long, almost-matching value it backtracks quadratically._sync_author_detailrunsemail_pattern.search(author)directly on the text from an<author>ordc:creatorelement, with no length bound.I reproduced it on the current main. A single item whose author is
"a@" + "a."*N + "!"gives, measuring CPU time on one core:Doubling the input roughly quadruples the time, so a ~64 KB author field is around 19 s. That is enough for one crafted feed to tie up an ingester.
The fix is a length guard: only feed
authorto the pattern when it is short enough to plausibly hold an address. RFC 5321 caps an address at 254 octets, and I left generous room for a display name alongside it, so no real author string is affected. Longer values just skip the email match and keep their name verbatim. After the change the 32 KB case parses in a few milliseconds, and I confirmed the usual forms (Name (me@example.com), a bareme@example.com, andName <me@example.com>) still split out the email exactly as before.Added
tests/test_author_email_bound.pycovering the normal extraction plus an oversized author that must parse without pulling an email out of it. I kept the timing out of the assertions to avoid a flaky test and put the numbers here instead. Full suite passes (4300 passed, 8 skipped) and mypy is clean. There is a Security changelog fragment as well.