fix(latex): fully unwrap deeply nested formatting macros

[Smeet23]

Summary

Fixes #3207

Two related bugs when formatting macros are nested inside each other:

Bug 1 — Colour name leaks into extracted text

_nodes_to_text fell through to its generic else-branch for \textcolor, which concatenates all arguments including the colour name. For example:

\section{\textcolor{blue}{\textbf{\textsc{[SEP]}}}}

Previously produced heading text "blue [SEP]" instead of "[SEP]".

Bug 2 — Inline paragraph broken into fragments

\textsc, \textsf, \textrm, \textnormal, \mbox, \textcolor, and \colorbox are listed in MACROS_STRUCTURAL. When encountered mid-sentence, _process_macro_node_inline flushed the text buffer and created a new doc node, splitting a single sentence into separate paragraphs.

Changes

• handlers/macros.py — Add explicit MACROS_TEXT_STYLE and textcolor/colorbox branches in _process_macro_node_inline before the MACROS_STRUCTURAL flush path, so they are accumulated inline (same pattern as MACROS_TEXT_FORMATTING).

• utils/text.py — Add matching branches in _nodes_to_text so that MACROS_TEXT_STYLE macros extract only their text argument, and textcolor/colorbox skip the colour argument and recurse into the text-content argument only.

Test plan

• New test test_latex_nested_formatting_macros covering all five patterns (textsc inline, textcolor+textbf inline, deep three-level nesting, textbf wrapping textsc, heading with nested textcolor)
• All 66 existing latex backend tests pass
• No raw LaTeX commands (\textbf, \textsc, \textcolor, …) appear in converted output

[github-actions]

✅ DCO Check Passed

Thanks @Smeet23, all your commits are properly signed off. 🎉

[mergify]

Merge Protections

Your pull request matches the following merge protections and will not be merged until they are valid.

🟢 Enforce conventional commit

Wonderful, this rule succeeded.

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)(?:\(.+\))?(!)?:

[codecov]

Codecov Report

❌ Patch coverage is 85.00000% with 9 lines in your changes missing coverage. Please review.

Files with missing lines	Patch %	Lines
docling/backend/latex/utils/text.py	81.25%	9 Missing ⚠️

📢 Thoughts on this report? Let us know!

[PeterStaar-IBM]

@Smeet23 Can you fix the merge conflicts?

[Smeet23]

Hi @PeterStaar-IBM — merge conflicts have been resolved. Here is what happened and what was done:

Conflict source: tests/test_backend_latex.py had a conflict between the two \vspace/\hspace leak tests that were merged into main (from another PR) and our new test_latex_nested_formatting_macros test. Both additions were to the end of the same file, so git flagged it as a conflict.

Resolution: Rebased onto the latest main and kept all three tests — test_vspace_argument_does_not_leak, test_hspace_argument_does_not_leak, and test_latex_nested_formatting_macros. Also moved a stray import re (which was inside the test function body) to the top-level imports where it belongs.

The branch is now clean and up to date with main. Ready for re-review when you have a moment.

[PeterStaar-IBM]

@Smeet23 Please run uv run pre-commit run --all-files and fix all errors you encounter, we are close to get this merged!

[Smeet23]

Done @PeterStaar-IBM! Pre-commit is now clean:

• Ruff formatter — Passed
• Ruff linter — Fixed the one remaining C901 complexity error in _nodes_to_text by extracting the macro-handling logic into a _macro_node_to_text helper method (complexity dropped from 31 → within limit)
• MyPy — The only remaining mypy error is the pre-existing playwright.sync_api import in html_backend.py, unrelated to this PR

Ready for merge!

[PeterStaar-IBM]

@Smeet23 Can you redo the DCO? Then we merge this.

[Smeet23]

Hi @cau-git and @PeterStaar-IBM — all CI checks are now passing and the DCO sign-off has been fixed across all commits. Could you please take a final look and merge if everything looks good? Happy to address any remaining concerns. Thanks!