- BulkSelectionBar component: sticky bottom bar shown only when canWrite
and selection is non-empty. Buttons meet WCAG 44px touch targets and
iOS safe-area inset is honoured.
- Bar mounted on /documents and /enrich.
- Alle X editieren button on /documents replaces the selection with
every UUID matching the active filter (via /api/documents/ids) and
jumps to /documents/bulk-edit.
Refs #225
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Rewrites BulkDocumentEditLayout to match the spec exactly:
- Fixed viewport layout (same as DocumentEditLayout) filling viewport below nav
- Split panel visible in all states (N=0/1/≥2) — was fullscreen dark drop zone
- N=0: centered drop-zone-box in left panel; shared form visible but greyed out
- N≥1: real PDF preview via URL.createObjectURL (no server upload required)
- N≥2: FileSwitcherStrip at bottom of left panel; count pill + discard in topbar
- FileEntry gains previewUrl; blob URLs created on add, revoked on remove/destroy
- save() checks response.ok and marks failed files with status: 'error'
- BulkDropZone redesigned: spec-accurate box with circular mint icon, serif title
- FileSwitcherStrip: number badges, arrows, keyboard nav via data-chip-id selector
- ScopeCard, UploadSaveBar: hardcoded German replaced with Paraglide i18n keys
- +page.svelte simplified to bare component render (layout is self-contained)
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Replaces the single-file form-action flow with BulkDocumentEditLayout,
enabling multi-file drag-and-drop upload with local preview, per-file
title editing, and shared metadata. Server load function unchanged.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
triggerSearch (local state, filter change) and buildPageHref (server data,
page nav) were each iterating over the same ~10 filter params. Any new
filter would have had to land in two places. buildSearchParams is now the
single source of truth for which params the /documents URL understands;
both callers just pass their snapshot and an optional targetPage. (#316)
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Frontend side of the /documents pagination work. The page.server.ts load
reads ?page= from the URL, forwards page+size=50 to the backend, and
exposes the new totalElements/pageNumber/pageSize/totalPages fields on
`data`. +page.svelte renders a <Pagination> component below the result
list; buildPageHref preserves every filter param and only updates page.
The existing triggerSearch debounce flow intentionally drops `page`
when any filter changes, so filter edits reset to page 0 automatically.
<Pagination> uses plain <a href> links (not goto) so SvelteKit's default
scroll restoration scrolls new pages to the top — the expected senior-UX
behaviour. Decorative chevrons wrapped in aria-hidden spans, 44px touch
targets, focus-visible ring, stacks vertically under 640px. The control
hides itself when totalPages ≤ 1.
Test coverage: 9 cases on Pagination (label, aria-current, prev/next
enable/disable, makeHref invocation, decorative chevron, touch target),
plus a filter-reset assertion on +page.svelte (page 5 → edit q →
goto URL must drop page=). Adds i18n keys in de/en/es. Manual edit to
api.ts pending a post-merge npm run generate:api against a rebuilt
dev backend. (#315)
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
- BackButton now accepts a `class` prop (default 'mb-4') so callers can
override spacing; resolves hardcoded margin in flex-row topbar snippets
- documents/[id]/edit and enrich/[id] pass class="" to suppress the margin
- Replace weak className unit test with class-prop behaviour tests
- Add [data-hydrated] comment in E2E spec explaining what emits the attribute
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
All 7 in-scope back navigation links converted to use history.back().
Admin panel mobile chevron converted inline (icon-only, different
visual pattern). Cancel buttons left as static <a> links.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Comments only render inside TranscriptionEditView, so a deep-link
into a document with existing reviewed transcriptions landed the
user in read mode with no comment element in the DOM — the scroll
target silently missed.
scrollToCommentFromQuery now takes a setPanelMode callback and calls
it with 'edit' whenever both query params are present. The page's
own transcribe-mode $effect checks a skipInitialPanelMode flag the
deep-link flow sets, so its default-panel-mode logic doesn't race
against the explicit override.
Two new helper tests pin the contract: panel mode is forced to
'edit' both when transcribe mode is off (entering fresh) and when
it is already on (same-page notification click).
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
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>
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>
When sort=SENDER, documents group under the sender's display name card.
When sort=RECEIVER, a document appears under each receiver's card
(with multi-receiver duplication). Falls back to i18n labels for unknown
sender/receiver. Passes sort prop from /documents page to DocumentList.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
- DashboardResumeStrip: text-[10px] → text-xs on collaborator initials (WCAG 1.4.4)
- documents/+page.server.ts: use !result.response.ok per CLAUDE.md; keep narrow try/catch for network-level failures only
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
- ContributorStack: text-xs for WCAG 1.4.4 (was text-[10px]), safeColor()
validation to block CSS injection via actor.color, role="img" aria-label
on empty placeholder, {#each} keyed by index
- ContributorStack spec: update empty-state assertion to getByRole('img')
- DocumentRow spec: add stopPropagation regression test for tag click
- documents/page.svelte.spec.ts: new — debounce, URL building, initial state
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
- New documents/+page.svelte wires SearchFilterBar + DocumentList with
URL-driven navigation (goto + SvelteURLSearchParams)
- Reset button in SearchFilterBar now navigates to /documents
- Rename documents/+page.server.spec.ts → page.server.spec.ts to avoid
SvelteKit route-file conflict on the + prefix
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Covers: button present, confirm dialog opens, form submitted on confirm,
form not submitted on cancel.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Extract DocumentEditLayout shared component for the PDF+form split-panel
UI, replacing the old scrolling layout on /documents/[id]/edit with the
same fixed-panel structure used by /enrich/[id]. Removes TranscriptionSection
and FileSectionEdit from the edit page; file upload/replace is now handled
by the shared layout. Delete SaveBar and FileSectionEdit as dead code.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Backend:
- TagRepository: add findDescendantIdsByName() recursive CTE query
- TagService: add expandTagNamesToDescendantIdSets() for document search
Frontend:
- TagInput: accept Tag[] (id, name, color, parentId) instead of string[]
- Chips show color dot via var(--c-tag-{color}) when tag has color
- Suggestions grouped hierarchically: children indented under their parents
- Update DescriptionSection, edit/new pages, SearchFilterBar, +page.svelte
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Move blob URL lifecycle management into a reusable createFileLoader()
hook that owns revoke-before-create and revoke-on-destroy. Replace
identical inline logic in documents/[id] and enrich/[id] with the hook.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Calling loadFile a second time previously leaked the previous object URL.
Add URL.revokeObjectURL(fileUrl) before creating a new one and in
onDestroy so all URLs are freed. Revoke behavior will be covered by the
useFileLoader hook tests in the next commit.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
When a document has manually drawn annotation boxes, the user can now
enable "Nur annotierte Bereiche" in the OCR trigger panel. The engine
skips layout detection entirely and runs recognition only within the
pre-drawn bounding boxes, preserving manual transcription blocks.
- Python: adds OcrRegion model, extend OcrRequest/OcrBlock; guided
branch in /ocr/stream groups by page and crops each region
- Engines: add extract_region_text() to both Kraken and Surya
- Java: adds OcrBlockResult.annotationId, OcrClient.OcrRegion,
TriggerOcrDTO.useExistingAnnotations; OcrAsyncRunner dispatches to
upsertGuidedBlock when annotationId is present; OcrService threads
the flag through to runSingleDocument
- TranscriptionService: adds upsertGuidedBlock (creates, updates OCR,
or preserves MANUAL blocks)
- Frontend: guided OCR toggle in OcrTrigger shown when blocks exist;
skips destructive-replace confirmation in guided mode
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
The progress message already says "Seite 3 von 7 wird analysiert…"
so the separate "3 / 7" counter was redundant. Remove the
OcrProgressBar from the page and inline only the skipped-pages
warning.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Replace inline translateOcrProgress with the extracted module. Add
OcrProgressBar below the spinner during OCR. Parse page numbers from
ANALYZING_PAGE progress codes and feed them to the bar. On Done, fill
bar to 100% briefly before clearing the overlay.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
OCR engines are CPU-bound and were blocking Uvicorn's single async
event loop, making /health unresponsive during processing. This caused
new OCR requests to fail silently (health check failure → no DB record
→ UI shows NONE). Wrap engine calls in asyncio.to_thread() to keep the
event loop free. Also surface OCR trigger errors in the frontend
instead of silently resetting the spinner.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Prevents users from drawing annotations that would be cleared when
the OCR job finishes. transcribeMode is set to false for the PDF
viewer while ocrRunning is true.
Refs #226
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Backend sends progress codes (PREPARING, LOADING, ANALYZING,
CREATING_BLOCKS:N, DONE:N, ERROR) via OcrJob.progressMessage.
Frontend translates them via Paraglide (de/en/es) and displays
below the spinner.
- V27 migration: adds progress_message column to ocr_jobs
- OcrAsyncRunner updates progress at each phase
- Poll interval reduced to 2s for snappier updates
Refs #226
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Single-document OCR now creates an OcrJobDocument row so
GET /api/documents/{id}/ocr-status can find running jobs.
OcrAsyncRunner updates the job document status (RUNNING → DONE/FAILED).
Frontend checks OCR status when entering transcription mode —
if a job is running, resumes polling and shows the spinner.
Refs #226
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
OcrService → OcrAsyncRunner was circular. Fixed by moving all OCR
processing logic (processDocument, clearExistingBlocks, createBlocks)
into OcrAsyncRunner. OcrService is now a thin entry point that
validates, creates the job, and dispatches to OcrAsyncRunner.
Architecture:
- OcrService: validates document, checks health, creates OcrJob, delegates
- OcrAsyncRunner: @Async processDocument + runSingleDocument + runBatch
- OcrBatchService: creates job + job documents, delegates to OcrAsyncRunner
- No circular dependencies
Single-document OCR is now async (returns jobId immediately).
Frontend polls GET /api/ocr/jobs/{jobId} every 3s until DONE/FAILED.
816 backend tests pass, 687 frontend tests pass.
Refs #226
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
- triggerOcr captures jobId from POST response and shows OcrProgress
- OcrProgress rendered in the transcription panel when ocrJobId is set
- handleOcrDone reloads blocks and annotations when OCR completes
- checkOcrStatus called when entering transcription mode — resumes
progress display if a job is already running for this document
Refs #226
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
- OcrTrigger component rendered in the transcription empty state when
the document has a file and user has write permission
- Review checkmark toggle on each TranscriptionBlock (turquoise when
reviewed, muted outline when not). Calls PUT .../review to toggle.
- TranscriptionBlockData type: added source + reviewed fields
- +page.svelte: triggerOcr() and reviewToggle() functions wired up
- Paraglide translations (de/en/es) for review toggle + reviewed count
All 687 frontend tests pass.
Refs #226, #230
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Add displayName and personType to all Person mock objects in
component and page tests. Update assertions from reversed
"lastName, firstName" format to forward-order displayName.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
- Add displayName default method to PersonSummaryDTO
- Update native SQL queries to include title, person_type columns
- Add getInitials() utility to personFormat.ts
- Update abbreviateName/abbreviateCompact for nullable firstName
- Replace firstName+lastName concatenation with displayName in all
person-displaying components and server load files
- Regenerate API types with displayName on Person and PersonSummaryDTO
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
PDF viewer collapses to 70px on mobile in read mode, expandable to
50vh. Toggle button with chevron. Paragraph tap auto-expands strip.
Mode toggle abbreviates to "Bearb." on small screens.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Paragraph click flashes the PDF annotation outline (1.5s fade).
Annotation click highlights the paragraph with a background flash.
Both directions scroll the target into view.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Adds TranscriptionPanelHeader and TranscriptionReadView to the
document detail page. Default mode is 'read' when blocks exist,
'edit' otherwise. Annotations dimmed in read mode.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Comment text:
- Body and quote bumped from text-sm (14px) to text-base (16px)
to visually match the font-sans author name at text-sm
Annotation reload on delete:
- Add annotationReloadKey prop through DocumentViewer → PdfViewer
- Increment key after block delete in +page.svelte
- PdfViewer reloads annotations when key changes
- Annotation rectangle disappears immediately, not just after refresh
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Pass activeAnnotationId to TranscriptionEditView. An $effect watches
it and sets activeBlockId to the block matching the annotation,
activating its turquoise focus border.
2 new tests (RED/GREEN):
- activates block matching activeAnnotationId (turquoise border)
- no block activated when activeAnnotationId is null
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Mobile layout (< 768px):
- Split view stacks vertically: PDF top (min 40vh), blocks below
- Blocks panel gets border-top instead of border-left
- PDF remains interactive for drawing in stacked mode
Scroll-sync (block → PDF):
- Clicking a block sets activeAnnotationId
- PdfViewer effect watches activeAnnotationId, navigates to the
annotation's page if different from current, then scrolls the
annotation element into view (double-rAF for async render timing)
- Works across pages: block on page 3 navigates PDF to page 3
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Own comments:
- Click the text to open inline edit (textarea replaces text)
- Enter saves, Escape cancels
- Small trash icon always visible in bottom-right corner
- Hover on text shows cursor-text + subtle bg highlight
Other people's comments: read-only, no edit/delete affordances.
Re-add currentUserId prop chain for ownership check.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Add blockNumbers prop through AnnotationLayer → PdfViewer → DocumentViewer.
Each turquoise annotation rectangle now shows a numbered badge (top-left,
matching the block card number in the right panel).
Block numbers are derived from sorted transcriptionBlocks, mapped by
annotationId, creating a visual link between PDF regions and block cards.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
When clicking a turquoise annotation on the PDF:
- If not in transcribe mode, enters it and loads blocks
- Waits for DOM render, then scrolls to the corresponding block card
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
The yellow annotation+comment system is now redundant. Transcription
blocks handle the same use case (mark region → discuss) but better,
because they also produce a transcription.
Removed:
- annotateMode state and all wiring through page/topbar/viewer/pdfviewer
- Annotate/Stop annotate buttons from DocumentTopBar
- AnnotateHintStrip import and rendering
- AnnotationSidePanel from document detail page
- canAnnotate prop from DocumentTopBar
- Color picker from PdfViewer
- Comment count badges and loadCommentCounts from PdfViewer
- Delete button from AnnotationLayer (blocks own annotation lifecycle)
- dimColor prop from AnnotationLayer
Simplified:
- AnnotationLayer: only canDraw + color + onDraw + onAnnotationClick
- PdfViewer: only draws in transcribeMode with turquoise
- Clicking annotation in transcribe mode scrolls to corresponding block
- canComment derived from canWrite (no longer needs canAnnotate)
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
TranscriptionBlock.svelte:
- "Kommentieren" button opens expandable comment thread per block
- Text selection in textarea captured as quoted text (> "...") prefix
- Quote hint "Text markieren für Zitat" shown when block is active/focused
- Comment thread uses existing CommentThread with blockId prop
CommentThread.svelte:
- Add blockId prop for block-level comments URL routing
- Add quotedText prop — pre-fills comment input with markdown blockquote
- commentsBase now supports 3 URL patterns: document, annotation, block
TranscriptionEditView.svelte:
- Pass canComment + currentUserId through to block components
3 new frontend tests:
- Kommentieren button present
- Quote hint shown when active
- Quote hint hidden when inactive
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
- AnnotationLayer: add dimColor prop — annotations matching dim color
render at 30% opacity with pointer-events disabled (300ms transition)
- PdfViewer: add transcribeMode prop, derived drawingEnabled/drawColor;
in transcribe mode draws with turquoise (#00C7B1), routes draw events
to onTranscriptionDraw callback instead of annotation endpoint
- DocumentViewer: pass through transcribeMode + onTranscriptionDraw
- Document detail page: createBlockFromDraw() POSTs to transcription
blocks API on draw completion, adds created block to list
- Mode-based dimming: yellow annotations dim in transcribe mode,
turquoise annotations dim in annotate mode
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
