fix(super-converter): bibliography/index/TOA field-code import fidelity (SD-3066)#3565
fix(super-converter): bibliography/index/TOA field-code import fidelity (SD-3066)#3565tupizz wants to merge 1 commit into
Conversation
…ty (SD-3066) Real Word documents surfaced several import defects in the block field-code family (BIBLIOGRAPHY, INDEX, XE, TOA). This addresses them and unifies the shared shape behind one set of helpers. Fixes: - Multi-run instruction aggregation joined fragments with an injected separator space, corrupting instructions Word splits across runs (e.g. `XE " Building Standard "`). Join verbatim; literal spacing is preserved. - Table of Authorities was dropped on import: the v2 importer had no `sd:tableOfAuthorities` handler, so the node was silently discarded. Register tableOfAuthoritiesImporter alongside index/bibliography. - A content control wrapping a block field imported as an inline `structuredContent` node; inside a block-only documentPartObject this threw "Invalid content for node type documentPartObject" and the editor failed to mount. Classify an SDT whose content is a block field as structuredContentBlock. - Bibliography neither captured nor replayed instructionTokens (unlike index/toa), so split BIBLIOGRAPHY instructions did not round-trip. Add the attribute and wire it through preprocessor, encode and decode. Refactors (DRY/KISS): - Extract buildBlockFieldNode (shared by the bibliography/index/toa preprocessors) and wrapParagraphsAsComplexField (shared by their translator decoders), mirroring the existing inline-field helpers. - Centralize BLOCK_FIELD_XML_NAMES so the paragraph importer and SDT classifier agree on which sd:* nodes are block content. Adds RED→GREEN unit tests for each fix and the new shared helpers. Full super-converter suite passes (2990). Linear: SD-3066 (parent of SD-3005)
|
The Status: PASS Here's what I checked and why it's clean: Real OOXML elements/attributes touched — all valid and used correctly:
Field codes / switches referenced in tests are all real Word fields and valid switches: The No non-existent OOXML elements/attributes, no missing required attributes (notably One caveat for transparency: I'd recommend re-running this with the |
![cubic-dev-ai[bot]](https://avatars.githubusercontent.com/in/1082092?s=60&v=4){kind=small}
There was a problem hiding this comment.
No issues found across 24 files
Tip: cubic could auto-approve low-risk PRs like this, if it thinks it's safe to merge. Learn more
Re-trigger cubic
1 participant
Note
📚 Stacked PRs — review bottom-up
documentPartObjectReview order: #3538 → #3565 → #3566. Each PR's base auto-retargets to
mainas the one below it merges.Summary
Real Word documents (a marked-citations TOA doc and a native Insert→Bibliography doc) surfaced several import defects in the block field-code family. This fixes them and unifies the shared shape behind one set of helpers (DRY/KISS).
Linear: SD-3066 (parent of SD-3005)
Fixes
XE " Building Standard "). Join verbatim.sd:tableOfAuthoritiesv2 importer existed, so the node was silently discarded. RegisteredtableOfAuthoritiesImporteralongside index/bibliography.structuredContent; inside a block-onlydocumentPartObjectthis threw "Invalid content for node type documentPartObject" and the editor failed to mount. Classify an SDT whose content is a block field asstructuredContentBlock.instructionTokens(unlike index/toa), so split BIBLIOGRAPHY instructions collapsed on export. Wired through preprocessor → extension attr → encode/decode.Refactors (DRY/KISS)
buildBlockFieldNode— shared by the bibliography/index/toa preprocessors.wrapParagraphsAsComplexField— shared by their translator decoders (mirrors the existing inline-field helpers).BLOCK_FIELD_XML_NAMES— one source of truth for the paragraph importer + SDT classifier.Test plan
pnpm --filter super-editor exec vitest run src/editors/v1/core/super-converter— 2990 pass