fixed bug where MarkdownHeaderSplitter's split result missed the first direct parent's header in the metadata and added lark to pyproject.toml

[MechaCritter]

Related Issues

• Fixes Bug: MarkdownHeaderSplitter's split result misses the direct parents header in the parent_headers` key in the metadata of the chunk #11041

Proposed Changes:

The error that occured in the issue #11041 happens when a parent header has its own content chunk before the first
child. In that path the code clears active_parents, so the first child loses its
ancestor while later siblings inherit it correctly.

The fix is to derive parent_headers from the current header_stack instead of
carrying a separate mutable “active parents” list that gets out of sync after contentful
headers.

How did you test it?

Tested locally using:

hatch run test:unit    # 13 failed, 4926 passed, 4 skipped, 356 deselected, 10 warnings
hatch run test:type    # Success: no issues found in 334 source files
hatch run fmt    #All checks passed!

Notes for the reviewer

You can copy the haystack/components/preprocessors/markdown_header_splitter.py file and rerun the code snippet provided in issue #11041. The bug does not occur anymore afterward.

Checklist

• I have read the contributors guidelines and the code of conduct.
• I have updated the related issue with new insights and changes.
• I have added unit tests and updated the docstrings.
• I've used one of the conventional commit types for my PR title: fix:, feat:, build:, chore:, ci:, docs:, style:, refactor:, perf:, test: and added ! in case the PR includes breaking changes.
• I have documented my code.
• I have added a release note file, following the contributors guidelines.
• I have run pre-commit hooks and fixed any issue.

[CLAassistant]

All committers have signed the CLA.

[vercel]

@MechaCritter is attempting to deploy a commit to the deepset Team on Vercel.

A member of the Team first needs to authorize it.

[sjrl]

@MechaCritter thanks for opening a PR! I see you are missing tests for the change. Could you add a test based on your example showing how your changes fix the issue.

[MechaCritter]

@MechaCritter thanks for opening a PR! I see you are missing tests for the change. Could you add a test based on your example showing how your changes fix the issue.

I added the corresponding test. Here are the corresponding log files as proof that my fix works:

Success run on bugfix branch:
bugfix_branch.log

Failed run on current main branch:
main_branch.log

I ran all other tests again for the markdown_header_splitter (test_markdown_header_splitter.py). Everything passed, so no break changes were introduced.
all_test_bugbranch.log

[sjrl]

@MechaCritter a few more minor comments but otherwise looks good! The logic is much easier to understand the previous as well!

[sjrl]

Thanks!

[vercel]

The latest updates on your projects. Learn more about Vercel for GitHub.

1 Skipped Deployment

Project	Deployment	Actions	Updated (UTC)
haystack-docs	Ignored		Apr 9, 2026 1:58pm

[MechaCritter]

I'm actually a bit confused: the CI pipeline failed a mypy test, which has nothing to do with my changes. @sjrl could you help me out?

[sjrl]

Hey @MechaCritter yeah it was related to a new release of sentence transformers. Fix incoming here #11066 once merged we can update your branch

[coveralls]

Coverage Report for CI Build 24196334786

Coverage decreased (-0.002%) to 92.841%

Details

• Coverage decreased (-0.002%) from the base build.
• Patch coverage: No coverable lines changed in this PR.
• 6 coverage regressions across 1 file.

Uncovered Changes

No uncovered changes found.

Coverage Regressions

6 previously-covered lines in 1 file lost coverage.

File	Lines Losing Coverage	Coverage
components/preprocessors/markdown_header_splitter.py	6	96.3%

---

Coverage Stats

Relevant Lines:	17154
Covered Lines:	15926
Line Coverage:	92.84%
Coverage Strength:	0.93 hits per line

---

💛 - Coveralls