[Krina-Paghadar]

Issue resolved by this Pull Request:
Resolves #4990

marko parses a backslash escape (snake\_case, \$5, 1\.) as a Literal node between two RawText nodes. The Markdown backend made one text item per node and wrapped them in an inline group, and the exporters join the items of an inline group with a space. The text came out as snake _ case.

This change merges RawText/Literal runs that touch each other into one run (_merge_adjacent_text_runs) before the items are created. An escaped & stays literal text. Inline code and emphasis stay separate items.

Checklist:

• Documentation has been updated, if necessary.
• Examples have been added, if necessary.
• Tests have been added, if necessary.

Review evidence:

Before: snake\_case here exports as snake \_ case here; 1\. not a list exports as 1 . not a list. After: snake_case here, 1. not a list.

Checks: the new test test_backslash_escapes_stay_in_their_text_run fails without the change and passes with it. In tests/test_backend_markdown.py, 40 pass and 3 fail, and the same 3 fail on main (docling-core 2.104.0, see #4982). ruff check and ruff format pass.

[Copilot]

Copilot was unable to review this pull request because the user who requested the review has reached their quota limit.

An organization owner can choose the repository organization as the billing source for reviews requested by members with a paid Copilot license. Learn about Copilot code review billing.

[github-actions]

✅ DCO Check Passed

Thanks @Krina-Paghadar, all your commits are properly signed off. 🎉

[mergify]

Merge Protections

🔴 1 of 2 protections blocking · waiting on 👀 reviews

	Protection	Waiting on
🔴	Require two reviewer for test updates	👀 reviews
🟢	Enforce conventional commit	—

🔴 Require two reviewer for test updates

Waiting for

• #approved-reviews-by >= 2

This rule is failing.

When test data is updated, we require two reviewers

• #approved-reviews-by >= 2

---

Show 1 satisfied protection

🟢 Enforce conventional commit

Make sure that we follow https://www.conventionalcommits.org/en/v1.0.0/

• title ~= ^(fix|feat|docs|style|refactor|perf|test|build|ci|chore|revert)(?:\(.+\))?(!)?:

[github-actions]

AI triage (advisory)

Checked commit 48f1cf3de3. The analysis read the diff only. It did not run code from this pull request.

The PR fixes backslash escapes in the Markdown backend. It adds _merge_adjacent_text_runs, which merges consecutive RawText/Literal AST nodes into one node before items are created, so snake\_case exports as snake_case instead of snake _ case.

CI safety: no concern found. A maintainer must still approve CI.

3 related pull request(s) in the same code

PR	State	Signal	Assessment
#4935	open	changes the same code	related: PR 4935 fixes GFM task-list checkbox markers in md_backend.py and touches the same test file, but solves a different problem with a different code change.
#4848	open	changes the same code	related: PR 4848 rewrites HTML-block handling in md_backend.py and adds many new tests, but it targets a completely different problem and code path.
#4984	closed	also references issue #4982	related: PR 4984 also references issue #4982 and changes md_backend.py, but it fixes HTML-block restoration after docling-core escaping, not backslash escapes.

Reference data changes

Change	Files
text	4

• text tests/data/md/groundtruth/escaped_characters.md.json: item attributes changed
• text tests/data/md/groundtruth/inline_and_formatting.md.yaml: +91 / -91 lines
• text tests/data/md/groundtruth/mixed_without_h1.md.md: +1 / -1 lines
• text tests/data/md/groundtruth/mixed_without_h1.md.yaml: +2 / -2 lines

Assessment: expected. The fix merges previously split text nodes, so existing reference outputs that recorded the wrong spaced text (e.g. snake _ case) are updated to the correct merged text. The four changed reference files match what the new merge logic would produce.

_{The ai:* labels are advisory. They do not approve CI and they do not add tests:full. A new push runs this check again.}

[github-actions]

AI first review (advisory): no blocker found

This review covers the full diff at 57075d90dd. It read the code and did not run it.

The PR fixes a real bug: marko splits backslash escapes into separate Literal nodes, causing exporters to add spaces around them. The merge-before-walk approach is correct. The main thing to check is the & round-trip: a Literal('&') node is stored as & in the merged RawText so that the later unescape() call at md_backend.py:884 converts it back to &; the test at line 1333 confirms this path.

_{A maintainer makes the merge decision. This review never approves the pull request.}

[Krina-Paghadar]

Ready for review. The 3 failing Markdown tests also fail on main with docling-core 2.104.0 (see #4982). Happy to adjust anything.

[codecov]

Codecov Report

❌ Patch coverage is 90.90909% with 2 lines in your changes missing coverage. Please review.

Files with missing lines	Patch %	Lines
docling/backend/md_backend.py	90.90%	1 Missing and 1 partial ⚠️

📢 Thoughts on this report? Let us know!

[PeterStaar-IBM]

@Krina-Paghadar we needto fix the regression-tests

[Krina-Paghadar]

Thanks @PeterStaar-IBM. The stored groundtruth still held the old buggy output (2 . strange). I regenerated it with the locked docling-core 2.103.1 (DOCLING_GEN_TEST_DATA=1) and pushed it in 48f1cf3. The content changes are 2. strange and one ordered list item that now gets the marker 1.. The rest is renumbered group references. All Markdown tests pass locally with the locked version. Could you please re-run the CI?