OcrAsyncRunner now passes the per-sender model path to streamBlocks for
HANDWRITING_KURRENT documents. processDocument replaced extractBlocks
with streamBlocks + AtomicReference, removing the unchecked raw-array
pattern.
Also stages all previously uncommitted foundational files for this
feature: SenderModel entity, SenderModelRepository, Flyway migrations
V40/V41, updated OcrClient/RestClientOcrClient streaming API,
TrainingDataExportService.exportForSender, TranscriptionService Kurrent
hook, application.yaml OCR config, and frontend i18n/test additions.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Extends Run interface with personId and QUEUED status, TrainingInfo with
personNames map, and passes it through to TrainingHistory for per-sender
model column display.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
OcrTrainingRun now includes personId (uuid, optional) and QUEUED status.
TrainingInfoResponse includes runs array with personId fields.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
pdfDoc was a plain variable (not \$state), so renderer.isLoaded had no
reactive dependencies in Svelte 5. PdfControls received isLoaded=false
permanently, keeping the next-page button disabled while zoom buttons
(which have no disabled attribute) still worked.
Fix: derive isLoaded from totalPages (\$state) via totalPages > 0.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Confirms that Enter on a suggestion item adds the tag even when allowCreation is
false — the activeIndex guard in handleKeydown runs before the allowCreation check.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
fetchSuggestions has no debounce; the wait is purely for the async mock to
resolve. The old name implied semantics that don't exist and added ~4.5s to
the suite (13 uses × 350ms).
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Rewrites orderedSuggestions to a recursive DFS with SuggestionEntry type,
adds role=listbox, depth indentation via inline style, font-medium for direct
matches, text-ink-3 for context nodes, and › prefix for root-level ancestors.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
- Add filter_operator_and/or/and_label/or_label i18n keys to de/en/es locale files
- Add aria-label and aria-pressed to AND/OR toggle buttons in SearchFilterBar
- Add data-testid="operator-and/or" for unambiguous test targeting (fixes substring match on German "Schlagwort")
- Use stable keys (tag.id ?? tag.name) for TagInput chip and suggestion lists
- Remove aria-level from role="option" items in TagInput (invalid attribute for that role)
- Add aria-live="polite" role="status" to TagMergeZone step indicator
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Adds INVALID_TAG_COLOR and TAG_CYCLE_DETECTED to the frontend ErrorCode
type and getErrorMessage() switch. German, English, and Spanish
translations added for both codes.
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>
Adds TagTreeNodeDTO, TagUpdateDTO (parentId + color), /api/tags/tree endpoint,
and parentId/color fields on Tag schema.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
"Text eintippen" sounded too casual and diverged from the domain
language used elsewhere in the app.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
"Rahmen einzeichnen" assumed familiarity with the segmentation concept;
"Text markieren" is self-explanatory for new contributors.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
The Lesefertig pulse was removed from the UI; drop the backend support
for it too — removes the subquery from findWeeklyStats(), the projection
getter, the DTO field, and updates all affected tests.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
The weekly count in Lesefertig counted any document with a reviewed
block in the past 7 days, not documents that crossed the ≥90% ready
threshold — a misleading stat given the column shows a different set.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
17 tests across SegmentationColumn, TranscriptionColumn, ReadyColumn,
MissionControlStrip. Covers document list rendering, per-column empty
states, weekly pulse visibility, link hrefs, progress bar, and the
reviewedPct denominator (annotationCount, not textedBlockCount).
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
- Add formatMCDate() to $lib/utils/date.ts (locale-aware, medium format);
remove duplicated inline formatDate() from all three column components
- Replace local TranscriptionQueueItemDTO/TranscriptionWeeklyStatsDTO type
declarations with imports from $lib/generated/api across all four components
- Add dashed empty states to SegmentationColumn and TranscriptionColumn
(ReadyColumn already had one)
- Remove outer {#if} from MissionControlStrip so the section is always
visible — each column owns its own empty state
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
- TranscriptionColumn progress bar: add aria-hidden="true" (the block count
text above already communicates the value to screen readers)
- TranscriptionColumn weekly pulse: text-ink → text-ink-2 (matches
SegmentationColumn, same semantic element)
- ReadyColumn reviewedPct: align denominator to annotationCount so the
displayed percentage matches the SQL threshold used to classify "ready"
- page.svelte.spec.ts: add missing segmentationDocs/transcriptionDocs/
readyDocs/weeklyStats to emptyData fixture
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
The original needsExpert V37 migration was applied to the dev DB before
the feature was removed. Renaming our new indexes migration to V38 avoids
the Flyway checksum conflict. Regenerated api.ts now reflects the
@Schema(requiredMode=REQUIRED) annotations — DTO fields are non-optional.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Drops the needsExpert / needs_expert flag end-to-end: DB migration
(V37, never applied), Document entity field, PATCH endpoint, service
method, DTO field, all three queue queries, ExpertBadge component,
i18n key, generated API types, and test fixture.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
The enrich page already handles task routing; the buttons in the
segmentation and transcription columns were redundant. Removes the
unused mission_control_segmentation_cta, mission_control_transcription_cta,
and mission_control_ready_all_cta keys from all three locale files.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
- Strip heading: "Mitarbeiten" → "Was braucht Aufmerksamkeit?"
- Column 1 heading: "Segmentierung" → "Rahmen einzeichnen"; add green
skill pill "✓ Ohne Vorkenntnisse"; heading color gray → ink (navy)
- Column 2 heading: "Transkription" → "Text eintippen"; add navy skill
pill "Kurrent hilfreich"; heading color gray → ink; weekly pulse
color green → ink (task, not achievement); progress bar track
bg-gray-200/h-1.5 → bg-ink/20/h-1; add transition-all to fill
- Column 3 heading: "Lesefertig" → "Lesefertig ✓"; heading color
gray → green-800; add "N Dokumente bereit" subtitle in green; add
"Alle N lesen →" link at bottom; reviewed % color gray → green-800
- All columns: add CTA buttons at bottom (Jetzt einzeichnen /
Jetzt tippen); empty state removed from cols 1 & 2 (columns
hide when empty); empty-state ghost CTA in col 3 restyled as
bordered button with hover:bg-ink
- Strip: add visibility guard — hides when all three lists are empty
- i18n: add mission_control_seg_skill_pill, mission_control_trans_skill_pill,
mission_control_ready_subtitle, mission_control_ready_all_cta in
de/en/es; update heading and CTA copy in all three locales
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
The /enrich route is for metadata (title, date, sender/receiver).
Segmentation and transcription work happens on the document detail page.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Adds the full-width 3-column collaboration widget below the existing
dashboard grid. Renders without the backend running (Promise.allSettled
isolation keeps failures silent).
Components (src/lib/components/):
- ExpertBadge.svelte — purple pill with icon, no props
- SegmentationColumn.svelte — col 1: links to /enrich/{id}, weekly pulse
- TranscriptionColumn.svelte — col 2: per-doc progress bar when blocks exist
- ReadyColumn.svelte — col 3: mint border when filled, dashed empty state
- MissionControlStrip.svelte — strip wrapper, 1-col mobile / 3-col sm+
i18n: 19 new keys added to de/en/es (mission_control_*)
Page wiring:
- +page.server.ts: 4 new Promise.allSettled calls for segmentation-queue,
transcription-queue, ready-to-read, weekly-stats; all failures silent
- +page.svelte: MissionControlStrip rendered below the grid in isDashboard
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Manually adds the new types to src/lib/generated/api.ts:
- Document.needsExpert: boolean (required field)
- TranscriptionQueueItemDTO schema
- TranscriptionWeeklyStatsDTO schema
- Paths: /api/transcription/{segmentation-queue, transcription-queue,
ready-to-read, weekly-stats} and /api/documents/{id}/needs-expert
- Operations: matching typed request/response shapes
Fixes briefwechsel spec fixtures to include scriptType and needsExpert
so the Document type shape is satisfied.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Add a summary_snippet column to findEnrichmentData using ts_headline on
documents.summary, only when the summary's tsvector matches the query.
Expose it via SearchMatchData.summarySnippet / summaryOffsets and render
a "Zusammenfassung" / "Summary" / "Resumen" labelled row in the document
list — identical treatment to the transcription snippet row.
Fixes the case where a document appeared in search results with no
visible match explanation (e.g. searching "frucht" found a document
whose summary mentioned "Früchte").
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
The refactor made pdfDoc a plain variable so renderer.isLoaded was not
reactive. Svelte only tracked currentPage and scale — but when the canvas
reappeared after loading, neither changed, so the PDF stayed blank.
Fix: merge the two effects into one that reads canvasEl synchronously.
Svelte now tracks canvasEl as a dependency; when the canvas remounts
(loading spinner → false), the effect re-fires and renders the
already-loaded PDF document.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
loadFile() reads fileUrl synchronously before its first await. When
called from a \$effect, Svelte tracks that read and re-runs the effect
every time fileUrl changes — i.e. after every successful load — causing
an infinite cycle of file fetches and PdfViewer remounts.
Fix: wrap the fileUrl read in untrack() so callers never accidentally
subscribe to fileUrl changes.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Adds simulateDragDrop helper and three tests covering the splice/insertAt
index arithmetic in handlePointerUp:
- move-to-end (insertAt path where target > fromIdx)
- move-to-start (insertAt path where target <= fromIdx)
- move-down-by-one (verifies the off-by-one dropTargetIdx - 1 branch)
Fixes @saraholt: "reorder calculation in handlePointerUp is untested"
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
