Three small refinements from Felix's review cycle 1:
- replaceState(page.url.pathname, page.state ?? {}) — defend against
first-navigation cases where page.state can be undefined.
- Extract the inline tick + requestAnimationFrame into a named
waitForPanelRender() helper; intent is now readable from onMount.
- Attach .catch() to the fire-and-forget scrollToCommentFromQuery
promise so any helper throw surfaces via console.error instead
of silently disappearing.
No behavior change on the happy path. All existing tests stay green.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Seeds a document, transcription block, and block comment via API,
then visits /documents/{id}?commentId=X&annotationId=Y and asserts
the page enters transcribe mode, the comment article becomes visible,
and the URL query params are stripped. Runs at 320px and 1440px so
the collapsed PDF strip clipping on mobile is caught. An axe-core
pass guards the new tabindex + focus-visible ring against a11y
regressions.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
After navHeight setup, call scrollToCommentFromQuery with the page
URL and callbacks into the component's local state (transcribeMode,
activeAnnotationId, flashAnnotationId) plus SvelteKit's replaceState
to strip the consumed query params.
afterTick awaits both Svelte's tick() and one requestAnimationFrame,
mirroring the existing handleAnnotationClick timing so the annotation
panel has rendered before scrollIntoView fires.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Notification deep-link scroll targets #comment-{id}. Add the id to
the article wrapper along with tabindex="-1" so scrollIntoView +
.focus({preventScroll:true}) can land screen-reader and keyboard
focus on the specific comment. A focus-visible ring appears only
for keyboard users so mouse clicks don't trigger a visible outline.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Pure function that reads commentId + annotationId from the page URL,
enters transcribe mode if needed, activates the block's annotation,
scrolls the target comment into view, focuses it for screen readers,
fires the existing annotation flash, and strips the params via the
injected callback.
All side effects go through callbacks so the helper is unit-testable
without mounting the page or a DOM (only scrollIntoView/focus are
called on the injected element). Eight tests cover both absent params,
happy path, transcribe-mode activation, missing DOM target, reduced
motion, flash trigger, and URL strip.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Only block comments are surfaced by the frontend now. The document-level
and annotation-level comment endpoints and service methods existed but
had no consumer. Remove them along with their repository queries and
test coverage so the surface area matches the actual feature set.
Shared edit, delete, and block reply endpoints stay. postBlockComment
now carries the authorName/mention/audit behaviors previously tested
through the dropped postComment method, so those behaviors remain
covered by the block-scoped test suite.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Previously issued block-comment notifications were stored with
annotation_id=NULL because CommentService.postBlockComment did not
populate DocumentComment.annotationId. Now that the code fix is in
place, existing rows need to be filled in so legacy notifications
can also carry the query param that the frontend deep-link flow
expects.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
postBlockComment now looks up the block via TranscriptionService and
sets comment.annotationId from block.getAnnotationId(). This closes
the upstream root cause of issue #276, where notifications for block
comments were stored with annotationId=null, breaking the notification
deep-link flow on the document detail page.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
