fix(markdown_parser): prevent unclosed <u> tags from pairing via emphasis algorithm

[Jason-Shen2]

Summary

When two <u> underline start tags appear without closing </u> tags, they were incorrectly paired as an opener/closer pair through the emphasis algorithm, causing the literal <u> markers and surrounding text to be silently deleted.

Root Cause

<u> was tokenized as UnderlineStart with can_close set from right-flanking rules, allowing process_emphasis to match two <u> openers as a pair. Underline is only meant to be closed by an explicit </u> tag (handled by parse_underline), never by another <u>.

Fix

1. Set can_close = false for UnderlineStart delimiters so <u> is never treated as a closing delimiter by process_emphasis
2. Add a guard in can_open_for to reject UnderlineStart pairs (defense in depth)

Regression Tests

Added tests covering the cases from the issue:

• <u><u> round-trips as literal text (was: empty line, all text deleted)
• <u>a<u> preserves both <u> markers (was: a, both markers deleted)
• a <u>word<u> b does not delete text (was: a word b, markers deleted)

Fixes #12863

[cla-bot]

Thank you for your pull request and welcome to our community. We could not parse the GitHub identity of the following contributors: zhenxing.shen.
This is most likely caused by a git client misconfiguration; please make sure to:

1. check if your git client is configured with an email to sign commits git config --list | grep email
2. If not, set it up using git config --global user.email email@example.com
3. Make sure that the git commit email is configured in your GitHub account settings, see https://github.com/settings/emails

[oz-for-oss]

@Jason-Shen2

I'm starting a first review of this pull request.

You can view the conversation on Warp.

I reviewed this pull request and requested human review from: @kevinyang372.

Comment /oz-review on this pull request to retrigger a review (up to 3 times on the same pull request).

Powered by Oz

[oz-for-oss]

Overview

This PR updates the inline Markdown delimiter handling so <u> start tags can no longer act as emphasis closers, while preserving explicit underline closing through </u>. It also adds regression coverage for unclosed underline-start sequences that previously dropped literal text.

Concerns

• No blocking correctness, security, or spec-alignment concerns found in the attached diff.

Verdict

Found: 0 critical, 0 important, 0 suggestions

Approve

Comment /oz-review on this pull request to retrigger a review (up to 3 times on the same pull request).

Powered by Oz