KAFKA-10574: Fix infinite loop in Values::parseString

[C0urante]

Jira

The special byte sequence 0xEF, 0xBF, 0xBF, when parsed as a UTF-8 string, causes the StringCharacterIterator to return CharacterIterator.DONE from its next(), current(), and other similar methods. This caused an infinite loop in the Values class whenever that byte sequence was encountered.

The fix is pretty simple. To see if we're at the end of a string, we compare StandardCharacterIterator::getIndex to StandardCharacterIterator::getEndIndex, instead of comparing the last-read character to CharacterIterator.DONE (since that character may occur in strings that we're parsing).

I've added a unit test that replicates this bug (when the fix in the Values class is not present). I gave it a timeout of five seconds in case someone accidentally re-creates the infinite loop in order to save some time and potential confusion.

All existing unit tests for the Values class pass with this change.

Committer Checklist (excluded from commit message)

• Verify design and implementation
• Verify test coverage and CI build status
• Verify documentation (including upgrade notes)

[gharris1727]

LGTM!

[C0urante]

@rhauch @kkonstantine can one of you take a look?

[kkonstantine]

LGTM. The fixes makes sense to me but I'm not intimately familiar with this class.
@rhauch, any concerns with the new check?

@C0urante I'd like to ask for a comment to the test so it's not as cryptic as assuming what these bytes mean for the given encoding.

[kkonstantine]

fyi: jobs fail with checkstyle errors

[kkonstantine]

Need to fix checkstyle before merging.

[C0urante]

Thanks @kkonstantine, I've added a comment and addressed the Checkstyle issues.

[rhauch]

LGTM, pending a green build. Thanks, @C0urante.

[kkonstantine]

LGTM after the latest fixes

[C0urante]

Hi @rhauch @kkonstantine, this PR has been approved, all merge conflicts have been resolved, and every build is green. Would it be possible to merge sometime soon so that it can be backported to bug fix releases and avoid accruing merge conflicts in case the files touched here are modified in other PRs? Thanks!