From fcca31bc4f6c4b1393c72ec2e17b1bea2adc2a8e Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Bj=C3=B6rn=20P=C3=B6ttker?= <bjoern@poettker-holtrup.de>
Date: Mon, 4 May 2026 15:45:19 +0200
Subject: [PATCH] perf: optimize PDF to image conversion by using a single
 Ghostscript execution and improve cleanup logic

---
 .../src/preprocessing/pdf.service.ts          | 42 ++++++++++++++-----
 1 file changed, 31 insertions(+), 11 deletions(-)

diff --git a/paperless-backend/src/preprocessing/pdf.service.ts b/paperless-backend/src/preprocessing/pdf.service.ts
index a95b2de..3da643d 100644
--- a/paperless-backend/src/preprocessing/pdf.service.ts
+++ b/paperless-backend/src/preprocessing/pdf.service.ts
@@ -37,16 +37,37 @@ export class PdfService {
 
   /**
    * Konvertiert alle Seiten einer PDF in Bilder.
+   * Verwendet einen einzigen Ghostscript-Aufruf mit %d-Platzhalter für alle Seiten.
    */
   async pdfToImages(pdfPath: string, dpi = 200): Promise<string[]> {
     const pageCount = await this.getPageCount(pdfPath);
-    const images: string[] = [];
+    if (pageCount === 0) return [];
 
+    const tmpDir = await fs.mkdtemp(path.join(os.tmpdir(), 'pdf-'));
+    const outputPattern = path.join(tmpDir, 'page-%d.png');
+
+    await execFileAsync('gs', [
+      '-dNOPAUSE',
+      '-dBATCH',
+      '-dSAFER',
+      '-sDEVICE=png16m',
+      `-r${dpi}`,
+      `-sOutputFile=${outputPattern}`,
+      pdfPath,
+    ]);
+
+    const images: string[] = [];
     for (let i = 1; i <= pageCount; i++) {
-      const imgPath = await this.pdfPageToImage(pdfPath, i, dpi);
-      images.push(imgPath);
+      const imgPath = path.join(tmpDir, `page-${i}.png`);
+      try {
+        await fs.access(imgPath);
+        images.push(imgPath);
+      } catch {
+        this.logger.warn(`Ghostscript hat Seite ${i} nicht erstellt: ${imgPath}`);
+      }
     }
 
+    this.logger.debug(`PDF konvertiert: ${images.length}/${pageCount} Seite(n) in ${tmpDir}`);
     return images;
   }
 
@@ -82,17 +103,16 @@ export class PdfService {
   }
 
   /**
-   * Räumt temporäre Bilder auf.
+   * Räumt temporäre Bilder und ihre Verzeichnisse auf.
    */
   async cleanup(imagePaths: string[]): Promise<void> {
+    const dirs = new Set<string>();
     for (const imgPath of imagePaths) {
-      try {
-        await fs.unlink(imgPath);
-        const dir = path.dirname(imgPath);
-        await fs.rmdir(dir).catch(() => {});
-      } catch {
-        // Ignorieren
-      }
+      try { await fs.unlink(imgPath); } catch {}
+      dirs.add(path.dirname(imgPath));
+    }
+    for (const dir of dirs) {
+      await fs.rmdir(dir).catch(() => {});
     }
   }
 }