v0.9.3: add cleanup command

cmd/photoscli/main.go

@@ -8,6 +8,7 @@ import (
 	"encoding/xml"
 	"fmt"
 	"io"
+	"io/fs"
 	"os"
 	"path/filepath"
 	"sort"
@@ -94,6 +95,8 @@ func run(args []string, stdout, stderr io.Writer, bridge photos.Bridge) int {
 		return cmdVerify(args[1:], stdout, stderr)
 	case "manifest":
 		return cmdManifest(args[1:], stdout, stderr)
+	case "cleanup":
+		return cmdCleanup(args[1:], stdout, stderr)
 	case "retry-failed":
 		return cmdRetryFailed(args[1:], stdout, stderr, bridge)
 	case "failures":
@@ -140,6 +143,7 @@ USAGE
   photoscli diff --album-id <id-or-title> --out <dir> [--manifest jsonl|sqlite]
   photoscli verify --out <dir> [--manifest jsonl|sqlite]
   photoscli manifest repair --out <dir> [--manifest jsonl|sqlite] [--checksum sha256] [--dry-run]
+  photoscli cleanup --out <dir> [--manifest jsonl|sqlite] [--dry-run]
   photoscli retry-failed --out <dir>
   photoscli retry-failed --out <dir> --clear-on-success
   photoscli failures list --out <dir>
@@ -188,6 +192,9 @@ COMMANDS
   manifest repair --out <dir> [--manifest jsonl|sqlite] [--checksum sha256] [--dry-run]
       Fill missing manifest size/checksum metadata from files that exist on disk.
 
+  cleanup --out <dir> [--manifest jsonl|sqlite] [--dry-run]
+      Remove files not referenced by the manifest. Use --dry-run first.
+
   retry-failed --out <dir>
       Retry assets previously written to failures.jsonl.
 
@@ -2389,6 +2396,75 @@ func verifySidecar(stdout io.Writer, outDir, id, checkPath string, strict bool)
 	return 0
 }
 
+func cmdCleanup(args []string, stdout, stderr io.Writer) int {
+	outDir := flagVal(args, "--out")
+	if outDir == "" {
+		fmt.Fprintln(stderr, "error: --out is required")
+		return exitErr
+	}
+	mf, err := manifest.ParseFormat(flagValWithDefault(args, "--manifest", "jsonl"))
+	if err != nil {
+		fmt.Fprintf(stderr, "error: %v\n", err)
+		return exitErr
+	}
+	entries, err := loadManifestEntries(outDir, mf)
+	if err != nil {
+		fmt.Fprintf(stderr, "error: %v\n", err)
+		return exitErr
+	}
+	keep := map[string]bool{
+		"downloads.jsonl": true,
+		"downloads.db":    true,
+		"export.log":      true,
+		"failures.jsonl":  true,
+	}
+	for _, e := range entries {
+		checkPath := e.Path
+		if checkPath == "" {
+			checkPath = e.Filename
+		}
+		if checkPath == "" {
+			continue
+		}
+		clean := filepath.Clean(checkPath)
+		if clean == "." || strings.HasPrefix(clean, "..") || filepath.IsAbs(clean) {
+			continue
+		}
+		keep[clean] = true
+		keep[filepath.Clean(sidecarPath(clean))] = true
+		keep[filepath.Clean(jsonSidecarPath(clean))] = true
+	}
+	dryRun := hasFlag(args, "--dry-run")
+	removed := 0
+	skipped := 0
+	_ = filepath.WalkDir(outDir, func(path string, d fs.DirEntry, _ error) error {
+		if path == outDir {
+			return nil
+		}
+		rel, _ := filepath.Rel(outDir, path)
+		rel = filepath.Clean(rel)
+		if d.IsDir() {
+			if rel == ".photoscli" || strings.HasPrefix(rel, ".photoscli"+string(os.PathSeparator)) {
+				return filepath.SkipDir
+			}
+			return nil
+		}
+		if keep[rel] {
+			return nil
+		}
+		fmt.Fprintf(stdout, "%s\torphan\n", rel)
+		removed++
+		if !dryRun {
+			if err := removeFunc(path); err != nil {
+				skipped++
+			}
+		}
+		return nil
+	})
+	fmt.Fprintf(stdout, "removed\t%d\nskipped\t%d\n", removed, skipped)
+	return exitOK
+}
+
 func cmdManifest(args []string, stdout, stderr io.Writer) int {
 	if len(args) < 1 || args[0] != "repair" {
 		fmt.Fprintln(stderr, "error: expected manifest repair")