fix(layout-adapter): preserve nested block content-control ancestry (#3745)#3753
fix(layout-adapter): preserve nested block content-control ancestry (#3745)#3753caio-pizzol wants to merge 6 commits into
Conversation
Stamp an ordered outer-to-inner SDT container chain (attrs.sdtContainers) on every emitted block - paragraphs, tables, images, and drawings - so a nested block content control keeps its outer control's identity. attrs.sdt stays each block's nearest control. Data-model step of the nested-SDT fix; resolved layout and the painter consume the chain in follow-ups. Refs #3745
Codecov Report✅ All modified and coverable lines are covered by tests. 📢 Thoughts on this report? Let us know! |
![chatgpt-codex-connector[bot]](https://avatars.githubusercontent.com/in/1144995?s=60&v=4){kind=small}
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: bec8863d8c
ℹ️ About Codex in GitHub
Codex has been enabled to automatically review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
When you sign up for Codex through ChatGPT, Codex can also answer questions or update the PR, like "@codex address that feedback".
| if (nearest && target.attrs.sdt == null) target.attrs.sdt = nearest; | ||
| if (chain.length > 0) target.attrs.sdtContainers = chain as SdtMetadata[]; |
There was a problem hiding this comment.
When this stamps the outer-to-inner chain only into attrs.sdtContainers, the production render path still ignores it: resolveFragmentSdtContainerKey derives the painter's sdtContainerKey only from attrs.sdt / attrs.containerSdt, and rg sdtContainers shows no runtime consumer outside these new tests. For a nested block SDT, the inner paragraph now has attrs.sdt set to the inner control, so the outer control is still lost for resolved-layout boundaries and DomPainter chrome even though the chain was preserved on the FlowBlock.
Useful? React with 👍 / 👎.
3 participants
Nested block content controls (a
w:sdtwhose content holds another blockw:sdt) were flattened during layout conversion: the inner control's content was dropped, and the outer control's identity was lost on whatever survived. This is the first step of #3752, which fixes it end to end.This step records the real structure.
handleStructuredContentBlockNodenow threads an ordered outer-to-inner container chain and stamps it (attrs.sdtContainers) on every emitted block - paragraphs, tables, images, and drawings - whileattrs.sdtstays each block's nearest control. The inner content is recovered and the outer/inner relationship is preserved in the data.Rendering catches up in the linked follow-ups. Resolved layout (FU2) carries the key chain - including image/drawing fragments, which
resolveFragmentSdtContainerKeyskips today - and busts paint reuse when it changes; that keeps a control's fragments in one run. The painter (FU3) then applies container chrome to image/drawing fragments and draws depth-aware boxes so the outer box wraps the inner. Until both land, a drawing inside a control can still split its box. Editor lock enforcement is unchanged and stays on the PM tree.Scope: data model and content recovery only - no nested chrome yet. The matching documentSection nested-SDT drop is tracked in #3752 and lands with the shared-traversal cleanup.
Builds on and incorporates @tom-material's fix from #3746, which identified the failing path. Supersedes #3746. Refs #3745.
Verified: imported a real Word file with a directly-nested block SDT through the importer and toFlowBlocks; the inner subtree is recovered and every block carries its outer-to-inner chain (inner blocks:
[outer, inner]). Opened as draft pending CI.