Add export command and shared item tree

Introduces a pure, FUSE-independent item tree (buildItemTree) shared by the mount and export commands. The new export subcommand downloads Docspell search results into a local directory, mirroring the mount's friendly folder/year/month hierarchy without the by-id directory or metadata JSON file, for archival or migration. - itemtree.go: pure item tree builder shared by mount and export - mount.go: buildMountTree now converts the pure tree into FUSE nodes, with a shared fileNode per attachment across friendly and by-id paths - export.go: export command with concurrent downloads, --dry-run, --skip-existing, --concurrency, --query, --limit, --auth-header - main.go: wires up the export subcommand and its flags - export_test.go: tests for plan collection, writes, skip-existing, failures, modtime - README.md: documents the export command
2026-07-21 23:35:03 +02:00
1 parent c4a2e9d085
commit 5e88eaea97
@@ -337,76 +337,63 @@ func itemIDFromLocalPath(path string, cfg config) (string, error) {
 	return "", fmt.Errorf("could not map mounted path to item id")
 }

+// buildMountTree converts the pure item tree produced by buildItemTree into the
+// FUSE node tree served by the mount command. A single *fileNode is created per
+// *itemFile (shared between the friendly hierarchy and by-id lookups) so that a
+// downloaded attachment is cached once regardless of which path reads it.
 func buildMountTree(ctx context.Context, client *DocspellClient, items []SearchItem) *dirNode {
-	root := &dirNode{name: "", children: map[string]fs.Node{}}
-	byID := &dirNode{name: "by-id", children: map[string]fs.Node{}}
-	root.children["by-id"] = byID
-	metadata := mountMetadata{Items: map[string]mountedItemMetadata{}}
-	metadataModTime := time.Now()
-	for _, item := range items {
-		folderName := "No Folder"
-		if item.Folder != nil && item.Folder.Name != "" {
-			folderName = item.Folder.Name
-		}
-		date := item.Date
-		if date == 0 {
-			date = item.ItemDate
-		}
-		yearMonth := "No Date"
-		modTime := time.Now()
-		if date > 0 {
-			modTime = time.UnixMilli(date)
-			yearMonth = modTime.Format("2006/01")
-		}
-		dirParts := append([]string{folderName}, strings.Split(yearMonth, "/")...)
-		dir := ensureDir(root, dirParts[0], dirParts[1:]...)
-		base := sanitizeName(item.Name)
-		if base == "" {
-			base = item.ID
-		}
-		for i, att := range item.Attachments {
-			ext := filepath.Ext(att.Name)
-			if ext == "" {
-				ext = ".pdf"
-			}
-			name := base + ext
-			if len(item.Attachments) > 1 {
-				name = fmt.Sprintf("%s - %02d%s", base, i+1, ext)
-			}
-			size := att.Size
-			if size <= 0 {
-				if headSize, err := client.AttachmentSize(ctx, att.ID); err == nil && headSize > 0 {
-					size = headSize
-				}
-			}
-			name = uniqueName(dir.children, sanitizeName(name))
-			node := &fileNode{name: name, attachmentID: att.ID, size: uint64(size), knownSize: size > 0, modTime: modTime, client: client}
-			dir.children[name] = node
-			friendlyRelPath := filepath.ToSlash(filepath.Join(append(sanitizePathParts(dirParts), name)...))
+	root, metadata := buildItemTree(items)
+	fuseRoot := convertItemDir(root, ctx, client, map[*itemFile]*fileNode{})

-			byIDName := item.ID + ext
-			if len(item.Attachments) > 1 {
-				if i > 0 {
-					byIDName = fmt.Sprintf("%s-%d%s", item.ID, i+1, ext)
-				}
-			}
-			shard := idShard(item.ID)
-			shardDir := ensureDir(byID, shard)
-			byIDName = uniqueName(shardDir.children, sanitizeName(byIDName))
-			shardDir.children[byIDName] = node
-			if i == 0 {
-				metadata.Items[item.ID] = mountedItemMetadata{
-					FriendlyPath: friendlyRelPath,
-					ByIDPath:     filepath.ToSlash(filepath.Join("by-id", shard, byIDName)),
-				}
-			}
-		}
-	}
 	metadataData, err := json.MarshalIndent(metadata, "", "  ")
 	if err == nil {
-		root.children[mountMetadataFileName] = &staticFileNode{name: mountMetadataFileName, data: append(metadataData, '\n'), modTime: metadataModTime}
+		fuseRoot.children[mountMetadataFileName] = &staticFileNode{
+			name:    mountMetadataFileName,
+			data:    append(metadataData, '\n'),
+			modTime: time.Now(),
+		}
 	}
-	return root
+	return fuseRoot
+}
+
+// convertItemDir recursively converts a pure itemDir into a FUSE dirNode. The
+// fileCache ensures that the same *itemFile (referenced from both the friendly
+// hierarchy and the by-id tree) maps to a single shared *fileNode.
+func convertItemDir(dir *itemDir, ctx context.Context, client *DocspellClient, fileCache map[*itemFile]*fileNode) *dirNode {
+	node := &dirNode{name: dir.name, children: map[string]fs.Node{}}
+	for name, child := range dir.children {
+		switch c := child.(type) {
+		case *itemDir:
+			node.children[name] = convertItemDir(c, ctx, client, fileCache)
+		case *itemFile:
+			node.children[name] = convertItemFile(c, ctx, client, fileCache)
+		}
+	}
+	return node
+}
+
+func convertItemFile(file *itemFile, ctx context.Context, client *DocspellClient, fileCache map[*itemFile]*fileNode) *fileNode {
+	if cached, ok := fileCache[file]; ok {
+		return cached
+	}
+	size := file.size
+	knownSize := size > 0
+	if size <= 0 {
+		if headSize, err := client.AttachmentSize(ctx, file.attachmentID); err == nil && headSize > 0 {
+			size = headSize
+			knownSize = true
+		}
+	}
+	node := &fileNode{
+		name:         file.name,
+		attachmentID: file.attachmentID,
+		size:         uint64(size),
+		knownSize:    knownSize,
+		modTime:      file.modTime,
+		client:       client,
+	}
+	fileCache[file] = node
+	return node
 }

 func sanitizePathParts(parts []string) []string {
@@ -432,24 +419,6 @@ func idShard(id string) string {
 	return "__"
 }

-func ensureDir(root *dirNode, first string, rest ...string) *dirNode {
-	cur := root
-	parts := append([]string{first}, rest...)
-	for _, part := range parts {
-		part = sanitizeName(part)
-		if part == "" {
-			part = "_"
-		}
-		next, ok := cur.children[part].(*dirNode)
-		if !ok {
-			next = &dirNode{name: part, children: map[string]fs.Node{}}
-			cur.children[part] = next
-		}
-		cur = next
-	}
-	return cur
-}
-
 func sanitizeName(s string) string {
 	s = strings.TrimSpace(s)
 	s = strings.ReplaceAll(s, "/", "_")
@@ -457,20 +426,6 @@ func sanitizeName(s string) string {
 	return s
 }

-func uniqueName(children map[string]fs.Node, name string) string {
-	if _, ok := children[name]; !ok {
-		return name
-	}
-	ext := filepath.Ext(name)
-	base := strings.TrimSuffix(name, ext)
-	for i := 2; ; i++ {
-		candidate := fmt.Sprintf("%s (%d)%s", base, i, ext)
-		if _, ok := children[candidate]; !ok {
-			return candidate
-		}
-	}
-}
-
 func (f *docspellFS) Root() (fs.Node, error) { return f.root, nil }

 func (d *dirNode) Attr(ctx context.Context, a *fuse.Attr) error {