Files
breakpilot-lehrer/admin-lehrer/components/ocr-overlay/useSlideWordPositions.ts
Benjamin Admin 0ee92e7210
Some checks failed
CI / go-lint (push) Has been skipped
CI / python-lint (push) Has been skipped
CI / nodejs-lint (push) Has been skipped
CI / test-go-school (push) Successful in 37s
CI / test-go-edu-search (push) Successful in 32s
CI / test-python-klausur (push) Failing after 2m10s
CI / test-python-agent-core (push) Successful in 19s
CI / test-nodejs-website (push) Successful in 20s
feat: OCR word_boxes fuer pixelgenaue Overlay-Positionierung
Backend: _ocr_cell_crop speichert jetzt word_boxes mit exakten
Tesseract/RapidOCR Wort-Koordinaten (left, top, width, height)
im Cell-Ergebnis. Absolute Bildkoordinaten, bereits zurueckgemappt.

Frontend: Slide-Hook nutzt word_boxes direkt wenn vorhanden —
jedes Wort wird exakt an seiner OCR-Position platziert. Kein
Pixel-Scanning noetig. Fallback auf alten Slide wenn keine Boxes.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
2026-03-11 19:39:49 +01:00

201 lines
6.1 KiB
TypeScript

import { useEffect, useState } from 'react'
import type { GridCell } from '@/app/(admin)/ai/ocr-overlay/types'
export interface WordPosition {
xPct: number
wPct: number
text: string
fontRatio: number
}
/**
* "Slide from left" positioning using OCR word bounding boxes.
*
* If the backend provides `word_boxes` (exact per-word coordinates from
* Tesseract/RapidOCR), we place each word directly at its OCR position.
* This gives pixel-accurate overlay without any heuristic pixel scanning.
*
* Fallback: if no word_boxes, slide tokens across dark-pixel projection
* (original slide algorithm).
*
* Font size: fontRatio = 1.0 for all (matches fallback rendering).
*/
export function useSlideWordPositions(
imageUrl: string,
cells: GridCell[],
active: boolean,
rotation: 0 | 180 = 0,
): Map<string, WordPosition[]> {
const [result, setResult] = useState<Map<string, WordPosition[]>>(new Map())
useEffect(() => {
if (!active || cells.length === 0 || !imageUrl) return
const img = new Image()
img.crossOrigin = 'anonymous'
img.onload = () => {
const imgW = img.naturalWidth
const imgH = img.naturalHeight
// Check if we can use word_boxes (fast path — no canvas needed)
const hasWordBoxes = cells.some(c => c.word_boxes && c.word_boxes.length > 0)
if (hasWordBoxes) {
// --- FAST PATH: use OCR word bounding boxes directly ---
const positions = new Map<string, WordPosition[]>()
for (const cell of cells) {
if (!cell.bbox_pct || !cell.text) continue
const boxes = cell.word_boxes
if (!boxes || boxes.length === 0) continue
const wordPos: WordPosition[] = boxes
.filter(wb => wb.text.trim())
.map(wb => ({
xPct: (wb.left / imgW) * 100,
wPct: (wb.width / imgW) * 100,
text: wb.text,
fontRatio: 1.0,
}))
if (wordPos.length > 0) {
positions.set(cell.cell_id, wordPos)
}
}
setResult(positions)
return
}
// --- SLOW PATH: pixel-projection slide (fallback if no word_boxes) ---
const canvas = document.createElement('canvas')
canvas.width = imgW
canvas.height = imgH
const ctx = canvas.getContext('2d')
if (!ctx) return
if (rotation === 180) {
ctx.translate(imgW, imgH)
ctx.rotate(Math.PI)
ctx.drawImage(img, 0, 0)
ctx.setTransform(1, 0, 0, 1, 0, 0)
} else {
ctx.drawImage(img, 0, 0)
}
const refFontSize = 40
const fontFam = "'Liberation Sans', Arial, sans-serif"
ctx.font = `${refFontSize}px ${fontFam}`
const cellHeights = cells
.filter(c => c.bbox_pct && c.bbox_pct.h > 0)
.map(c => Math.round(c.bbox_pct.h / 100 * imgH))
.sort((a, b) => a - b)
const medianCh = cellHeights.length > 0
? cellHeights[Math.floor(cellHeights.length / 2)]
: 30
const renderedFontImgPx = medianCh * 0.7
const measureScale = renderedFontImgPx / refFontSize
const spaceWidthPx = Math.max(2, Math.round(ctx.measureText(' ').width * measureScale))
const positions = new Map<string, WordPosition[]>()
for (const cell of cells) {
if (!cell.bbox_pct || !cell.text) continue
let cx: number, cy: number
const cw = Math.round(cell.bbox_pct.w / 100 * imgW)
const ch = Math.round(cell.bbox_pct.h / 100 * imgH)
if (rotation === 180) {
cx = Math.round((100 - cell.bbox_pct.x - cell.bbox_pct.w) / 100 * imgW)
cy = Math.round((100 - cell.bbox_pct.y - cell.bbox_pct.h) / 100 * imgH)
} else {
cx = Math.round(cell.bbox_pct.x / 100 * imgW)
cy = Math.round(cell.bbox_pct.y / 100 * imgH)
}
if (cw <= 0 || ch <= 0) continue
if (cx < 0) cx = 0
if (cy < 0) cy = 0
if (cx + cw > imgW || cy + ch > imgH) continue
const imageData = ctx.getImageData(cx, cy, cw, ch)
const proj = new Float32Array(cw)
for (let y = 0; y < ch; y++) {
for (let x = 0; x < cw; x++) {
const idx = (y * cw + x) * 4
const lum = 0.299 * imageData.data[idx] + 0.587 * imageData.data[idx + 1] + 0.114 * imageData.data[idx + 2]
if (lum < 128) proj[x]++
}
}
const threshold = Math.max(1, ch * 0.03)
const ink = new Uint8Array(cw)
for (let x = 0; x < cw; x++) {
ink[x] = proj[x] >= threshold ? 1 : 0
}
if (rotation === 180) {
ink.reverse()
}
const tokens = cell.text.split(/\s+/).filter(Boolean)
if (tokens.length === 0) continue
const tokenWidthsPx = tokens.map(t =>
Math.max(4, Math.round(ctx.measureText(t).width * measureScale))
)
const wordPos: WordPosition[] = []
let cursor = 0
for (let ti = 0; ti < tokens.length; ti++) {
const tokenW = tokenWidthsPx[ti]
const coverageNeeded = Math.max(1, Math.round(tokenW * 0.15))
let bestX = cursor
const searchLimit = Math.max(cursor, cw - tokenW)
for (let x = cursor; x <= searchLimit; x++) {
let inkCount = 0
const spanEnd = Math.min(x + tokenW, cw)
for (let dx = 0; dx < spanEnd - x; dx++) {
inkCount += ink[x + dx]
}
if (inkCount >= coverageNeeded) {
bestX = x
break
}
if (x > cursor + cw * 0.3 && ti > 0) {
bestX = cursor
break
}
}
if (bestX + tokenW > cw) {
bestX = Math.max(0, cw - tokenW)
}
wordPos.push({
xPct: cell.bbox_pct.x + (bestX / cw) * cell.bbox_pct.w,
wPct: (tokenW / cw) * cell.bbox_pct.w,
text: tokens[ti],
fontRatio: 1.0,
})
cursor = bestX + tokenW + spaceWidthPx
}
if (wordPos.length > 0) {
positions.set(cell.cell_id, wordPos)
}
}
setResult(positions)
}
img.src = imageUrl
}, [active, cells, imageUrl, rotation])
return result
}