Files
breakpilot-lehrer/admin-lehrer/components/ocr-pipeline/StepWordRecognition.tsx
Benjamin Admin b681ddb131 [split-required] Split 58 monoliths across Python, Go, TypeScript (Phases 1-3)
Phase 1 — Python (klausur-service): 5 monoliths → 36 files
- dsfa_corpus_ingestion.py (1,828 LOC → 5 files)
- cv_ocr_engines.py (2,102 LOC → 7 files)
- cv_layout.py (3,653 LOC → 10 files)
- vocab_worksheet_api.py (2,783 LOC → 8 files)
- grid_build_core.py (1,958 LOC → 6 files)

Phase 2 — Go (edu-search-service, school-service): 8 monoliths → 19 files
- staff_crawler.go (1,402 → 4), policy/store.go (1,168 → 3)
- policy_handlers.go (700 → 2), repository.go (684 → 2)
- search.go (592 → 2), ai_extraction_handlers.go (554 → 2)
- seed_data.go (591 → 2), grade_service.go (646 → 2)

Phase 3 — TypeScript (admin-lehrer): 45 monoliths → 220+ files
- sdk/types.ts (2,108 → 16 domain files)
- ai/rag/page.tsx (2,686 → 14 files)
- 22 page.tsx files split into _components/ + _hooks/
- 11 component files split into sub-components
- 10 SDK data catalogs added to loc-exceptions
- Deleted dead backup index_original.ts (4,899 LOC)

All original public APIs preserved via re-export facades.
Zero new errors: Python imports verified, Go builds clean,
TypeScript tsc --noEmit shows only pre-existing errors.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
2026-04-24 17:28:57 +02:00

460 lines
16 KiB
TypeScript
Raw Blame History

This file contains ambiguous Unicode characters
This file contains Unicode characters that might be confused with other characters. If you think that this is intentional, you can safely ignore this warning. Use the Escape button to reveal them.
'use client'
import { useCallback, useEffect, useRef, useState } from 'react'
import type { GridResult, GridCell, WordEntry, WordGroundTruth } from '@/app/(admin)/ai/ocr-kombi/types'
import { WordRecognitionOverview } from './WordRecognitionOverview'
import { WordRecognitionLabeling } from './WordRecognitionLabeling'
import { WordRecognitionControls } from './WordRecognitionControls'
const KLAUSUR_API = '/klausur-api'
interface StepWordRecognitionProps {
sessionId: string | null
onNext: () => void
goToStep: (step: number) => void
/** Skip _heal_row_gaps in cell grid (better overlay positioning) */
skipHealGaps?: boolean
}
export function StepWordRecognition({ sessionId, onNext, goToStep, skipHealGaps = false }: StepWordRecognitionProps) {
const [gridResult, setGridResult] = useState<GridResult | null>(null)
const [detecting, setDetecting] = useState(false)
const [error, setError] = useState<string | null>(null)
const [gtNotes, setGtNotes] = useState('')
const [gtSaved, setGtSaved] = useState(false)
// Step-through labeling state
const [activeIndex, setActiveIndex] = useState(0)
const [editedEntries, setEditedEntries] = useState<WordEntry[]>([])
const [editedCells, setEditedCells] = useState<GridCell[]>([])
const [mode, setMode] = useState<'overview' | 'labeling'>('overview')
const [ocrEngine, setOcrEngine] = useState<'auto' | 'tesseract' | 'rapid' | 'paddle'>('auto')
const [usedEngine, setUsedEngine] = useState<string>('')
const [pronunciation, setPronunciation] = useState<'british' | 'american'>('british')
const [gridMethod, setGridMethod] = useState<'v2' | 'words_first'>('v2')
// Streaming progress state
const [streamProgress, setStreamProgress] = useState<{ current: number; total: number } | null>(null)
const enRef = useRef<HTMLInputElement>(null)
const tableEndRef = useRef<HTMLDivElement>(null)
const isVocab = gridResult?.layout === 'vocab'
useEffect(() => {
if (!sessionId) return
// Always run fresh detection -- word-lookup is fast (~0.03s)
// and avoids stale cached results from previous pipeline versions.
runAutoDetection()
// eslint-disable-next-line react-hooks/exhaustive-deps
}, [sessionId])
const applyGridResult = (data: GridResult) => {
setGridResult(data)
setUsedEngine(data.ocr_engine || '')
if (data.layout === 'vocab' && data.entries) {
initEntries(data.entries)
}
if (data.cells) {
setEditedCells(data.cells.map(c => ({ ...c, status: c.status || 'pending' })))
}
}
const initEntries = (entries: WordEntry[]) => {
setEditedEntries(entries.map(e => ({ ...e, status: e.status || 'pending' })))
setActiveIndex(0)
}
const runAutoDetection = useCallback(async (engine?: string) => {
if (!sessionId) return
const eng = engine || ocrEngine
setDetecting(true)
setError(null)
setStreamProgress(null)
setEditedCells([])
setEditedEntries([])
setGridResult(null)
try {
// PP-OCRv5 forces words_first on the backend, so align frontend accordingly
const effectiveGridMethod = eng === 'paddle' ? 'words_first' : gridMethod
const useStream = effectiveGridMethod === 'v2'
// Retry once if initial request fails (e.g. after container restart,
// session cache may not be warm yet when navigating via wizard)
let res: Response | null = null
for (let attempt = 0; attempt < 2; attempt++) {
res = await fetch(
`${KLAUSUR_API}/api/v1/ocr-pipeline/sessions/${sessionId}/words?stream=${useStream ? 'true' : 'false'}&engine=${eng}&pronunciation=${pronunciation}${skipHealGaps ? '&skip_heal_gaps=true' : ''}&grid_method=${effectiveGridMethod}`,
{ method: 'POST' },
)
if (res.ok) break
if (attempt === 0 && (res.status === 400 || res.status === 404)) {
// Wait briefly for cache to warm up, then retry
await new Promise(r => setTimeout(r, 2000))
continue
}
break
}
if (!res || !res.ok) {
const err = await res?.json().catch(() => ({ detail: res?.statusText })) || { detail: 'Worterkennung fehlgeschlagen' }
throw new Error(err.detail || 'Worterkennung fehlgeschlagen')
}
// words_first / pp-ocrv5 returns plain JSON (no streaming)
if (!useStream) {
const data = await res.json() as GridResult
applyGridResult(data)
return
}
const reader = res.body!.getReader()
const decoder = new TextDecoder()
let buffer = ''
let streamLayout: string | null = null
let streamColumnsUsed: GridResult['columns_used'] = []
let streamGridShape: GridResult['grid_shape'] | null = null
let streamCells: GridCell[] = []
while (true) {
const { done, value } = await reader.read()
if (done) break
buffer += decoder.decode(value, { stream: true })
// Parse SSE events (separated by \n\n)
while (buffer.includes('\n\n')) {
const idx = buffer.indexOf('\n\n')
const chunk = buffer.slice(0, idx).trim()
buffer = buffer.slice(idx + 2)
if (!chunk.startsWith('data: ')) continue
const dataStr = chunk.slice(6) // strip "data: "
let event: any
try {
event = JSON.parse(dataStr)
} catch {
continue
}
if (event.type === 'meta') {
streamLayout = event.layout || 'generic'
streamGridShape = event.grid_shape || null
// Show partial grid result so UI renders structure
setGridResult(prev => ({
...prev,
layout: event.layout || 'generic',
grid_shape: event.grid_shape,
columns_used: [],
cells: [],
summary: { total_cells: event.grid_shape?.total_cells || 0, non_empty_cells: 0, low_confidence: 0 },
duration_seconds: 0,
ocr_engine: '',
} as GridResult))
}
if (event.type === 'columns') {
streamColumnsUsed = event.columns_used || []
setGridResult(prev => prev ? { ...prev, columns_used: streamColumnsUsed } : prev)
}
if (event.type === 'cell') {
const cell: GridCell = { ...event.cell, status: 'pending' }
streamCells = [...streamCells, cell]
setEditedCells(streamCells)
setStreamProgress(event.progress)
// Auto-scroll table to bottom
setTimeout(() => tableEndRef.current?.scrollIntoView({ behavior: 'smooth', block: 'nearest' }), 16)
}
if (event.type === 'complete') {
// Build final GridResult
const finalResult: GridResult = {
cells: streamCells,
grid_shape: streamGridShape || { rows: 0, cols: 0, total_cells: streamCells.length },
columns_used: streamColumnsUsed,
layout: streamLayout || 'generic',
image_width: 0,
image_height: 0,
duration_seconds: event.duration_seconds || 0,
ocr_engine: event.ocr_engine || '',
summary: event.summary || {},
}
// If vocab: apply post-processed entries from complete event
if (event.vocab_entries) {
finalResult.entries = event.vocab_entries
finalResult.vocab_entries = event.vocab_entries
finalResult.entry_count = event.vocab_entries.length
}
applyGridResult(finalResult)
setUsedEngine(event.ocr_engine || '')
setStreamProgress(null)
}
}
}
} catch (e) {
setError(e instanceof Error ? e.message : 'Unbekannter Fehler')
} finally {
setDetecting(false)
}
// eslint-disable-next-line react-hooks/exhaustive-deps
}, [sessionId, ocrEngine, pronunciation, gridMethod])
const handleGroundTruth = useCallback(async (isCorrect: boolean) => {
if (!sessionId) return
const gt: WordGroundTruth = {
is_correct: isCorrect,
corrected_entries: isCorrect ? undefined : (isVocab ? editedEntries : undefined),
notes: gtNotes || undefined,
}
try {
await fetch(`${KLAUSUR_API}/api/v1/ocr-pipeline/sessions/${sessionId}/ground-truth/words`, {
method: 'POST',
headers: { 'Content-Type': 'application/json' },
body: JSON.stringify(gt),
})
setGtSaved(true)
} catch (e) {
console.error('Ground truth save failed:', e)
}
}, [sessionId, gtNotes, editedEntries, isVocab])
// Generic mode: update cell text
const updateCell = (cellId: string, value: string) => {
setEditedCells(prev => prev.map(c =>
c.cell_id === cellId ? { ...c, text: value, status: 'edited' as const } : c
))
}
// Helper: get unique row indices from cells
const getUniqueRowCount = () => {
if (!editedCells.length) return 0
return new Set(editedCells.map(c => c.row_index)).size
}
// Helper: get cells for a given row index (by position in sorted unique rows)
const getRowCells = (rowPosition: number) => {
const uniqueRows = [...new Set(editedCells.map(c => c.row_index))].sort((a, b) => a - b)
const rowIdx = uniqueRows[rowPosition]
return editedCells.filter(c => c.row_index === rowIdx)
}
// Step-through: confirm current row (always cell-based)
const confirmEntry = () => {
const rowCells = getRowCells(activeIndex)
const cellIds = new Set(rowCells.map(c => c.cell_id))
setEditedCells(prev => prev.map(c =>
cellIds.has(c.cell_id) ? { ...c, status: c.status === 'edited' ? 'edited' : 'confirmed' } : c
))
const maxIdx = getUniqueRowCount() - 1
if (activeIndex < maxIdx) {
setActiveIndex(activeIndex + 1)
}
}
// Step-through: skip current row
const skipEntry = () => {
const rowCells = getRowCells(activeIndex)
const cellIds = new Set(rowCells.map(c => c.cell_id))
setEditedCells(prev => prev.map(c =>
cellIds.has(c.cell_id) ? { ...c, status: 'skipped' as const } : c
))
const maxIdx = getUniqueRowCount() - 1
if (activeIndex < maxIdx) {
setActiveIndex(activeIndex + 1)
}
}
// Focus english input when active entry changes in labeling mode
useEffect(() => {
if (mode === 'labeling' && enRef.current) {
enRef.current.focus()
}
}, [activeIndex, mode])
// Keyboard shortcuts in labeling mode
useEffect(() => {
if (mode !== 'labeling') return
const handler = (e: KeyboardEvent) => {
if (e.key === 'Enter' && !e.shiftKey) {
e.preventDefault()
confirmEntry()
} else if (e.key === 'ArrowDown' && e.ctrlKey) {
e.preventDefault()
skipEntry()
} else if (e.key === 'ArrowUp' && e.ctrlKey) {
e.preventDefault()
if (activeIndex > 0) setActiveIndex(activeIndex - 1)
}
}
window.addEventListener('keydown', handler)
return () => window.removeEventListener('keydown', handler)
// eslint-disable-next-line react-hooks/exhaustive-deps
}, [mode, activeIndex, editedEntries, editedCells])
if (!sessionId) {
return (
<div className="flex flex-col items-center justify-center py-16 text-center">
<div className="text-5xl mb-4">🔤</div>
<h3 className="text-lg font-medium text-gray-700 dark:text-gray-300 mb-2">
Schritt 5: Worterkennung
</h3>
<p className="text-gray-500 dark:text-gray-400 max-w-md">
Bitte zuerst Schritte 1-4 abschliessen.
</p>
</div>
)
}
const columnsUsed = gridResult?.columns_used || []
const gridShape = gridResult?.grid_shape
// Counts for labeling progress (always cell-based)
const confirmedRowIds = new Set(
editedCells.filter(c => c.status === 'confirmed' || c.status === 'edited').map(c => c.row_index)
)
const confirmedCount = confirmedRowIds.size
const totalCount = getUniqueRowCount()
// Group cells by row (shared between overview and labeling)
const cellsByRow: Map<number, GridCell[]> = new Map()
for (const cell of editedCells) {
const existing = cellsByRow.get(cell.row_index) || []
existing.push(cell)
cellsByRow.set(cell.row_index, existing)
}
const sortedRowIndices = [...cellsByRow.keys()].sort((a, b) => a - b)
return (
<div className="space-y-4">
{/* Loading with streaming progress */}
{detecting && (
<div className="space-y-1">
<div className="flex items-center gap-2 text-teal-600 dark:text-teal-400 text-sm">
<div className="animate-spin w-4 h-4 border-2 border-teal-500 border-t-transparent rounded-full" />
{streamProgress
? `Zelle ${streamProgress.current}/${streamProgress.total} erkannt...`
: 'Worterkennung startet...'}
</div>
{streamProgress && streamProgress.total > 0 && (
<div className="w-full bg-gray-200 dark:bg-gray-700 rounded-full h-1.5">
<div
className="bg-teal-500 h-1.5 rounded-full transition-all duration-150"
style={{ width: `${(streamProgress.current / streamProgress.total) * 100}%` }}
/>
</div>
)}
</div>
)}
{/* Layout badge + Mode toggle */}
{gridResult && (
<div className="flex items-center gap-2">
{/* Layout badge */}
<span className={`px-2 py-0.5 rounded text-[10px] uppercase font-semibold ${
isVocab
? 'bg-indigo-100 dark:bg-indigo-900/30 text-indigo-700 dark:text-indigo-300'
: 'bg-gray-100 dark:bg-gray-700 text-gray-600 dark:text-gray-400'
}`}>
{isVocab ? 'Vokabel-Layout' : 'Generisch'}
</span>
{gridShape && (
<span className="text-[10px] text-gray-400">
{gridShape.rows}×{gridShape.cols} = {gridShape.total_cells} Zellen
</span>
)}
<div className="flex-1" />
<button
onClick={() => setMode('overview')}
className={`px-3 py-1.5 text-xs rounded-lg font-medium transition-colors ${
mode === 'overview'
? 'bg-teal-600 text-white'
: 'bg-gray-100 dark:bg-gray-700 text-gray-600 dark:text-gray-300 hover:bg-gray-200 dark:hover:bg-gray-600'
}`}
>
Uebersicht
</button>
<button
onClick={() => setMode('labeling')}
className={`px-3 py-1.5 text-xs rounded-lg font-medium transition-colors ${
mode === 'labeling'
? 'bg-teal-600 text-white'
: 'bg-gray-100 dark:bg-gray-700 text-gray-600 dark:text-gray-300 hover:bg-gray-200 dark:hover:bg-gray-600'
}`}
>
Labeling ({confirmedCount}/{totalCount})
</button>
</div>
)}
{/* Overview mode */}
{mode === 'overview' && (
<WordRecognitionOverview
sessionId={sessionId}
gridResult={gridResult}
detecting={detecting}
editedCells={editedCells}
activeIndex={activeIndex}
setActiveIndex={setActiveIndex}
setMode={setMode}
tableEndRef={tableEndRef}
/>
)}
{/* Labeling mode */}
{mode === 'labeling' && (
<WordRecognitionLabeling
sessionId={sessionId}
gridResult={gridResult}
editedCells={editedCells}
activeIndex={activeIndex}
setActiveIndex={setActiveIndex}
columnsUsed={columnsUsed}
getUniqueRowCount={getUniqueRowCount}
getRowCells={getRowCells}
updateCell={updateCell}
confirmEntry={confirmEntry}
skipEntry={skipEntry}
enRef={enRef}
cellsByRow={cellsByRow}
sortedRowIndices={sortedRowIndices}
/>
)}
{/* Controls */}
{gridResult && (
<WordRecognitionControls
gridResult={gridResult}
isVocab={!!isVocab}
detecting={detecting}
usedEngine={usedEngine}
ocrEngine={ocrEngine}
setOcrEngine={setOcrEngine}
pronunciation={pronunciation}
setPronunciation={setPronunciation}
gridMethod={gridMethod}
setGridMethod={setGridMethod}
gtNotes={gtNotes}
setGtNotes={setGtNotes}
gtSaved={gtSaved}
runAutoDetection={runAutoDetection}
handleGroundTruth={handleGroundTruth}
goToStep={goToStep}
onNext={onNext}
/>
)}
{error && (
<div className="p-3 bg-red-50 dark:bg-red-900/20 text-red-600 dark:text-red-400 rounded-lg text-sm">
{error}
</div>
)}
</div>
)
}