feat: implement region-based QR code scanning for inbox documents

paperless-backend/src/barcode/barcode-scanner.service.ts

@@ -2,6 +2,7 @@ import { Injectable, Logger, OnApplicationBootstrap } from '@nestjs/common';
 import { InjectRepository } from '@nestjs/typeorm';
 import { Repository } from 'typeorm';
 import * as fs from 'fs/promises';
+import sharp = require('sharp');
 import { PdfService } from '../preprocessing/pdf.service';
 import { QrCodeService } from '../preprocessing/qr-code.service';
 import {
@@ -76,6 +77,7 @@ export class BarcodeScannerService implements OnApplicationBootstrap {
 
     doc.QrCodes = qrCodes;
     doc.PageCount = pageCount;
+    doc.IsScanned = true;
     try {
       await this.documentRepo.save(doc);
     } catch (err: any) {
@@ -174,6 +176,61 @@ export class BarcodeScannerService implements OnApplicationBootstrap {
     }
   }
 
+  /**
+   * Rendert eine einzelne Seite bei hoher DPI, beschneidet den angegebenen
+   * Bereich (normalisierte Koordinaten 0..1) und scannt ihn nach QR-Codes.
+   * Neu gefundene QR-Codes werden in der DB persistiert.
+   */
+  async scanRegion(
+    doc: InboxDocument,
+    pdfPath: string,
+    page: number,
+    x: number,
+    y: number,
+    w: number,
+    h: number,
+  ): Promise<{ found: string[] }> {
+    let imagePath: string | null = null;
+    try {
+      imagePath = await this.pdfService.pdfPageToImage(pdfPath, page, 400);
+
+      const image = sharp(imagePath);
+      const { width: imgW, height: imgH } = await image.metadata();
+      if (!imgW || !imgH) return { found: [] };
+
+      const left   = Math.round(Math.max(0, x * imgW));
+      const top    = Math.round(Math.max(0, y * imgH));
+      const width  = Math.round(Math.min(imgW - left, w * imgW));
+      const height = Math.round(Math.min(imgH - top,  h * imgH));
+      if (width <= 0 || height <= 0) return { found: [] };
+
+      const cropped = await image.extract({ left, top, width, height }).png().toBuffer();
+      const qrResults = await this.qrCodeService.extractFromImage(cropped);
+      if (qrResults.length === 0) return { found: [] };
+
+      const existingKeys = new Set((doc.QrCodes ?? []).map((qr) => `${qr.page}:${qr.value}`));
+      const found: string[] = [];
+      let changed = false;
+
+      for (const qr of qrResults) {
+        found.push(qr.data);
+        const key = `${page}:${qr.data}`;
+        if (!existingKeys.has(key)) {
+          doc.QrCodes = [...(doc.QrCodes ?? []), { page, value: qr.data }];
+          changed = true;
+        }
+      }
+
+      if (changed) {
+        await this.documentRepo.save(doc);
+      }
+
+      return { found };
+    } finally {
+      if (imagePath) await this.pdfService.cleanup([imagePath]);
+    }
+  }
+
   /**
    * Rescannt alle Inbox-Dokumente — wird nach Änderungen an Eingangsdokumentarten aufgerufen.
    * Läuft sequenziell, um PDF-Rendering nicht zu überlasten. Fire-and-forget vom Caller.