Version bump main: 9.55.0 -> 9.56.0, react: 9.0.4 -> 9.0.5#3393
Merged
Conversation
…path valid (#3358) When formatInsertPointWithContentModel persists the Content Model group path during DOM to Model conversion, a FormatContainer that falls back to a plain paragraph would be removed from the model, leaving the persisted path invalid. Add a skipFormatContainerFallbackCheck option that keeps the FormatContainer in that scenario so the formatting callback can still rely on the path. Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
When handleBlockGroupChildren finishes processing a block group whose list node stack is rooted at the parent being left, clear the entire node stack instead of only its inner levels so a sibling block group does not incorrectly reuse the invalidated list element. Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
#3362) When setSelection applies a range selection, a leading or trailing SelectionMarker that has a real (non-marker) selected segment next to it in the same paragraph carries no selection information and can be removed. Track, per paragraph, whether any non-marker segment was selected and which boundary markers were kept; after the segment loop, drop those boundary markers if the paragraph contained real selected content. This is scoped to the same paragraph on purpose: a marker at a paragraph edge must be kept to distinguish "selection starts at the beginning of this line" from "selection starts at the end of the previous line". Update the #3341 reconcileSelection snapshots accordingly: the preserveMarker guard still hands setSelection two live markers (so the selected segment keeps isSelected), and setSelection now drops those redundant boundary markers. Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Reconcile IMG src and data-* attribute changes in Content Model cache Previously any attribute change other than id on editor content would set mutation type to "unknown" and invalidate the whole cached Content Model. Recognize src and data-* changes on an indexed IMG element and update the related ContentModelImage (src property or dataset) in place, so the cache stays valid. Add a new "attribute" mutation type carrying the element and attribute name, and a reconcileImageAttribute method on DomIndexer. @ Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Adds two new public APIs to roosterjs-content-model-markdown and a "Paste as Markdown" demo button that uses them to decide whether to convert clipboard content with convertMarkdownToContentModel or fall back to the regular paste flow.
#3365) @ Treat a backslash-escaped ASCII punctuation character (e.g. "\*") as a literal character so it is not parsed as a formatting marker. @ Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
* Add MarkdownPastePlugin to convert pasted markdown content Add a MarkdownPastePlugin in the roosterjs-content-model-markdown package that handles the BeforePaste event and, when the clipboard content can be interpreted as markdown, converts the plain text into a Content Model and pastes it as rich content. - Introduce the 'asMarkdown' PasteType so callers can explicitly request a markdown paste; the demo's "Paste as Markdown" button now routes through it. - Add MarkdownPasteOptions with an autoConversion flag (default false) so the plugin can optionally auto-convert any markdown-looking paste. - Map the new PasteType to Normal in the legacy editor adapter. - Wire the plugin and its autoConversion option into the demo site's plugins side pane.
This change makes the converter prefer the pasteType already present on the new event, only falling back to the mapping from the old event when the new event doesn't have one
Adds a new project skill, draft-pr, at .claude/skills/draft-pr/SKILL.md. The skill walks through creating a pull request for the current branch: it inspects the diff against master, generates a concise imperative title and a structured description (a ## Summary plus a ## How to test section), and opens the PR with the gh CLI. It follows the same authoring conventions as the existing version-bump skill and enforces repo-specific rules from AGENTS.md (PRs target master, tests use yarn test:fast).
…rd uppercase (#3372) * new tests * Only capitalize the beginning of a word * Add button to demo * Normalize line endings to LF for capitalization changes Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> --------- Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Bumps [dompurify](https://github.com/cure53/DOMPurify) from 3.4.0 to 3.4.9. - [Release notes](https://github.com/cure53/DOMPurify/releases) - [Commits](cure53/DOMPurify@3.4.0...3.4.9) --- updated-dependencies: - dependency-name: dompurify dependency-version: 3.4.9 dependency-type: direct:development ... Signed-off-by: dependabot[bot] <support@github.com> Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com> Co-authored-by: Jiuqing Song <jisong@microsoft.com>
Bumps [dompurify](https://github.com/cure53/DOMPurify) from 3.4.9 to 3.4.11. - [Release notes](https://github.com/cure53/DOMPurify/releases) - [Commits](cure53/DOMPurify@3.4.9...3.4.11) --- updated-dependencies: - dependency-name: dompurify dependency-version: 3.4.11 dependency-type: direct:development ... Signed-off-by: dependabot[bot] <support@github.com> Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>
Bumps [@babel/core](https://github.com/babel/babel/tree/HEAD/packages/babel-core) from 7.11.1 to 7.29.7. - [Release notes](https://github.com/babel/babel/releases) - [Changelog](https://github.com/babel/babel/blob/main/CHANGELOG.md) - [Commits](https://github.com/babel/babel/commits/v7.29.7/packages/babel-core) --- updated-dependencies: - dependency-name: "@babel/core" dependency-version: 7.29.7 dependency-type: indirect ... Signed-off-by: dependabot[bot] <support@github.com> Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com> Co-authored-by: Jiuqing Song <jisong@microsoft.com>
Fixes the editor jumping/scrolling when an image is selected for editing. In createImageWrapper.ts, attaching the shadow root replaces the rendered light-DOM image with a taller edit wrapper (border + resize/rotate/crop handles), which grows the surrounding line box and causes the browser to scroll the selection back into view. This pins the shadow host span to the original image's footprint — captured from image.offsetWidth/offsetHeight before the shadow root takes over — by setting display: inline-block, fixed width/height, vertical-align: bottom, and overflow: visible. The handles and wrapper still overflow the host visually, but the line box no longer grows, so the editor stays put. Pinning is skipped when the image has no measurable footprint (offsetWidth/offsetHeight of 0).
* Re-enable and stabilize previously-disabled tests Several test suites were disabled (xit/xdescribe) due to flakiness or environment dependence. Re-enable them and fix the underlying issues so they pass deterministically regardless of spec order: - selectionConverterTest: spy on the deep createRange module did not intercept the barrel import; exercise the real createRange with a document-attached image instead. - updateRotateHandleTest: the spy targeted rotateHandle while the code reads rotateCenter; point it at the right element and feed each "hidden" case the intended edge distance (and correct an impossible expected handle value). - applyChangeTest: replace brittle exact PNG data-URL comparisons with IHDR-dimension checks (encoder-independent), and create a fresh contentModelImage per test to remove order-dependent editingInfo state. - tableEditorTest: give the editor a deterministic size so the table is never flush against the scroll-container edge, which otherwise made inserter visibility depend on leftover document.body layout. - tableResizerTest: recreate the content model table per test (onDragging mutates it in place) and gate verifyCellRectChange on growth only so an incidental cross-axis border-box shift does not flip the assertion. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * fix test * improve * improve * fix --------- Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
…ion (#3379) * Fix normalizeSingleSegment to preserve leading space for word separation - Search backward for the nearest non-empty text segment to determine whether leading spaces should be preserved or removed - Skip empty text segments during normalization to avoid corrupting context - Replace nbsp with regular space for leading space preservation since nbsp prevents word wrapping - Add unit tests for normalizeSingleSegment Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> * fix comment --------- Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
…3378) Cover previously-untested code paths: - ContextMenuPluginBase, TouchPlugin (new test files) - CustomReplacePlugin guard branches and replacement callback - TableEditPlugin onMouseMove/ensureTableRects/defaultTableSelector - TableEditor editing lifecycle, public methods, getOnMouseOut, RTL - EditorAdapter API surface (new describe block) - buildRangeEx and insertNode utils (new test files) Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
* Retain segments with undeletable links during model pruning Keep segments that have link.format.undeletable set to true in pruneUnselectedModel, even when they are not selected. This prevents undeletable link segments from being lost during selection-based pruning. Add unit tests covering the new behavior. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> * improve --------- Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
…oConversion option (#3381)
) * Pin aria-live announcer to top:0; left:0 to avoid scroll overflow createAriaLiveElement sets position:absolute on the screen-reader announcer container but leaves top/left unset. CSS resolves them to the element's static position (its place in normal flow), which sits after every other child of the announcer's eventual container. When that container has a non-trivial content height (typical app layouts), the absolute element contributes to the documentElement's scrollable overflow region and pushes out a small but visible scrollbar even though nothing is laid out beyond the viewport. Standard visually-hidden patterns anchor the element to a known viewport corner. Add top:0; left:0 so the absolute element sits at the top-left of its containing block and never extends the scroll area. The element remains invisible (clip-path:inset(100%) + 1x1 size + overflow:hidden) and aria-live continues to announce textContent changes. Repro: mount the editor in any host page where document body has a height >= viewport. A 1px-tall scrollbar appears even with no editor content overflowing. * Update announce tests for new top:0; left:0 in announcer style Existing tests use toEqual to assert the exact shape of the announcer div's style object. Adding top/left in createAriaLiveElement breaks those assertions; add the same two properties to expected style blocks to match. --------- Co-authored-by: Bryan Valverde U <bvalverde@microsoft.com>
* Add ariaLabel to TextField for accessibility * Refactor label handling in InputDialogItem component
* Add support for HTML data element in Content Model Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> * Exclude dataValue from EmptySegmentFormat to avoid leaking into segment formats Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> * Refactor data element support to use a segment decorator Replace the flat dataValue segment-format approach with a dedicated ContentModelData segment decorator, mirroring how the link and code decorators work. This isolates the data value from ContentModelSegmentFormat and removes the EmptySegmentFormat pollution that caused the paste E2E failures. - Add ContentModelData decorator and ContentModelDataFormat types - Add dataValueFormatHandler and 'data' format category - Wire data into DomToModel/ModelToDom contexts and stackFormat - dataProcessor now uses knownElementProcessor + addData decorator - Handle data in handleSegmentDecorator, cloneModel and mergeTextSegments - Update and add tests Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> * Add createText test for data decorator parameter Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> * Add mergeNode tests for DATA element optimization Cover the addition of 'DATA' to mergeNode OptimizeTags: adjacent <data> elements with the same value attribute merge, while different values do not. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> * Add end-to-end tests for DATA element round-trip Cover a normal <data value> round-trip and a special case where the <data> element contains an inner <b>. Verifies that after Model => DOM rewrite the per-segment <data> wrappers are merged back into a single <data> element by mergeNode. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> --------- Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
|
vinguyen12
marked this pull request as ready for review
July 1, 2026 17:53
BryanValverdeU
previously approved these changes
Jul 1, 2026
BryanValverdeU
previously approved these changes
Jul 1, 2026
Collaborator
|
We added a bunch of new types, so this should be 9.56.0 |
vinguyen12
changed the title
Contributor
Author
I changed it to 9.56.0 |
JiuqingSong
approved these changes
Jul 1, 2026
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
7 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Version bump
legacyAdapter unchanged (8.66.1). Patch bump per request.
Changes included