From 6d835aefac4c954e77ea137809ada213562eac11 Mon Sep 17 00:00:00 2001
From: sysops <sysops@LenovoT14Gen5>
Date: Fri, 8 May 2026 22:19:20 +0200
Subject: [PATCH] fix(PROJ-35): OCR Tempdir auf storage_dir umleiten

Mit systemd ProtectSystem=strict ist /tmp fuer den Service read-only.
ocr.SetTempDir(storage_path/ocr-tmp) nutzt einen RW-Pfad innerhalb der
ohnehin freigegebenen ReadWritePaths.
---
 cmd/archivmail/cmd_ocr_reprocess.go |  2 ++
 cmd/archivmail/main.go              |  3 +++
 internal/ocr/ocr.go                 | 23 +++++++++++++++++++++--
 3 files changed, 26 insertions(+), 2 deletions(-)

diff --git a/cmd/archivmail/cmd_ocr_reprocess.go b/cmd/archivmail/cmd_ocr_reprocess.go
index 4e0d72f..f0c102b 100644
--- a/cmd/archivmail/cmd_ocr_reprocess.go
+++ b/cmd/archivmail/cmd_ocr_reprocess.go
@@ -75,6 +75,8 @@ func runOCRReprocess(args []string) {
 		logger.Warn("ocr tools not on PATH — install tesseract-ocr + poppler-utils",
 			"pdftotext", ts.HasPdftotext, "tesseract", ts.HasTesseract, "pdftoppm", ts.HasPdftoppm)
 	}
+	// Keep scratch space inside the storage dir (matches the daemon's setup).
+	ocr.SetTempDir(cfg.Storage.StorePath + "/ocr-tmp")
 
 	ctx := context.Background()
 
diff --git a/cmd/archivmail/main.go b/cmd/archivmail/main.go
index fd7f9d9..203e260 100644
--- a/cmd/archivmail/main.go
+++ b/cmd/archivmail/main.go
@@ -174,6 +174,9 @@ func main() {
 	// PROJ-35: OCR-Worker — extracts text from PDF/image attachments and feeds
 	// it back into the per-tenant Manticore index. Non-blocking submit so the
 	// mail intake pipeline is never delayed.
+	// systemd's ProtectSystem=strict restricts /tmp; route OCR scratch space
+	// into the storage-dir which is guaranteed-RW.
+	ocr.SetTempDir(cfg.Storage.StorePath + "/ocr-tmp")
 	ocrWorker := ocr.NewWorker(mailStore, idxMgr, ocr.Options{
 		Workers:   2,
 		QueueSize: 1000,
diff --git a/internal/ocr/ocr.go b/internal/ocr/ocr.go
index 047ef53..81a085a 100644
--- a/internal/ocr/ocr.go
+++ b/internal/ocr/ocr.go
@@ -33,6 +33,25 @@ var (
 	ErrUnavailable = errors.New("ocr: tesseract or pdftotext not available")
 )
 
+// tempDirRoot may be overridden by callers (e.g. when the systemd unit
+// restricts /tmp). Empty string means "use the OS default" via os.MkdirTemp.
+var tempDirRoot = ""
+
+// SetTempDir overrides the parent directory used for temporary OCR work.
+// When dir is empty, the OS default ($TMPDIR or /tmp) is used.
+// Safe to call once at startup; not safe for concurrent reconfiguration.
+func SetTempDir(dir string) {
+	tempDirRoot = dir
+}
+
+func mkTempDir() (string, error) {
+	if tempDirRoot != "" {
+		_ = os.MkdirAll(tempDirRoot, 0o755)
+		return os.MkdirTemp(tempDirRoot, "archivmail-ocr-*")
+	}
+	return os.MkdirTemp("", "archivmail-ocr-*")
+}
+
 // IsAvailable reports whether at least one OCR tool is on PATH.
 // pdftotext alone enables text extraction from native PDFs;
 // tesseract enables OCR for images and scanned PDFs.
@@ -128,7 +147,7 @@ func classify(contentType, filename string) fileKind {
 // ── PDF extraction (pdftotext → tesseract fallback) ───────────────────────
 
 func extractPDF(ctx context.Context, data []byte, langs []string) (string, error) {
-	tmp, err := os.MkdirTemp("", "archivmail-ocr-*")
+	tmp, err := mkTempDir()
 	if err != nil {
 		return "", fmt.Errorf("ocr: tempdir: %w", err)
 	}
@@ -214,7 +233,7 @@ func extractImage(ctx context.Context, data []byte, filename string, langs []str
 	if _, err := exec.LookPath("tesseract"); err != nil {
 		return "", ErrUnavailable
 	}
-	tmp, err := os.MkdirTemp("", "archivmail-ocr-*")
+	tmp, err := mkTempDir()
 	if err != nil {
 		return "", fmt.Errorf("ocr: tempdir: %w", err)
 	}