Use mount metadata for path lookups

2026-07-19 23:17:38 +02:00
parent 4b96db068a
commit ce1f9290b8
2 changed files with 158 additions and 85 deletions
@@ -2,6 +2,7 @@ package main

 import (
 	"context"
+	"encoding/json"
 	"flag"
 	"fmt"
 	"os"
@@ -35,6 +36,23 @@ type fileNode struct {
 	data         []byte
 }

+type staticFileNode struct {
+	name    string
+	data    []byte
+	modTime time.Time
+}
+
+type mountMetadata struct {
+	Items map[string]mountedItemMetadata `json:"items"`
+}
+
+type mountedItemMetadata struct {
+	FriendlyPath string `json:"friendlyPath"`
+	ByIDPath     string `json:"byIDPath"`
+}
+
+const mountMetadataFileName = ".docspell-cli-metadata.json"
+
 func runMountCommand(args []string, cfg config, password string) error {
 	fsFlags := flag.NewFlagSet("mount", flag.ExitOnError)
 	query := fsFlags.String("query", cfg.MountQuery, "Docspell item query to expose")
@@ -127,6 +145,10 @@ func runGetLocalPathCommand(args []string, cfg config, password string) error {
 		return err
 	}
 	id := itemIDFromURLOrID(flags.Arg(0))
+	if path, err := friendlyLocalPathForItemID(mountPoint, id); err == nil && path != "" {
+		fmt.Println(path)
+		return nil
+	}
 	matches, err := filepath.Glob(filepath.Join(mountPoint, "by-id", idShard(id), id+".*"))
 	if err != nil {
 		return err
@@ -134,44 +156,30 @@ func runGetLocalPathCommand(args []string, cfg config, password string) error {
 	if len(matches) == 0 {
 		return fmt.Errorf("entry %s not found under mounted by-id tree", id)
 	}
-	path := matches[0]
-	if friendlyPath, err := friendlyLocalPath(mountPoint, path); err == nil && friendlyPath != "" {
-		path = friendlyPath
-	}
-	fmt.Println(path)
+	fmt.Println(matches[0])
 	return nil
 }

-func friendlyLocalPath(mountPoint, byIDPath string) (string, error) {
-	byIDInfo, err := os.Stat(byIDPath)
+func friendlyLocalPathForItemID(mountPoint, id string) (string, error) {
+	metadata, err := readMountMetadata(mountPoint)
 	if err != nil {
 		return "", err
 	}
-	byIDRoot := filepath.Join(mountPoint, "by-id")
-	var found string
-	err = filepath.Walk(mountPoint, func(path string, info os.FileInfo, err error) error {
-		if err != nil || info == nil {
-			return nil
+	item, ok := metadata.Items[id]
+	if !ok || item.FriendlyPath == "" {
+		return "", nil
 	}
-		if path == byIDRoot || strings.HasPrefix(path, byIDRoot+string(os.PathSeparator)) {
-			if info.IsDir() {
-				return filepath.SkipDir
+	return filepath.Join(mountPoint, filepath.FromSlash(item.FriendlyPath)), nil
 }
-			return nil
-		}
-		if info.IsDir() {
-			return nil
-		}
-		if os.SameFile(byIDInfo, info) {
-			found = path
-			return filepath.SkipAll
-		}
-		return nil
-	})
+
+func readMountMetadata(mountPoint string) (mountMetadata, error) {
+	var metadata mountMetadata
+	data, err := os.ReadFile(filepath.Join(mountPoint, mountMetadataFileName))
 	if err != nil {
-		return "", err
+		return metadata, err
 	}
-	return found, nil
+	err = json.Unmarshal(data, &metadata)
+	return metadata, err
 }

 func runGetURLCommand(args []string, cfg config, password string) error {
@@ -300,34 +308,25 @@ func itemIDFromLocalPath(path string, cfg config) (string, error) {
 	if len(parts) >= 3 && parts[0] == "by-id" {
 		return strings.TrimSuffix(parts[2], filepath.Ext(parts[2])), nil
 	}
-	info, err := os.Stat(abs)
+	metadata, err := readMountMetadata(mountPoint)
 	if err != nil {
 		return "", err
 	}
-	var found string
-	err = filepath.Walk(filepath.Join(mountPoint, "by-id"), func(p string, fi os.FileInfo, err error) error {
-		if err != nil || fi == nil || fi.IsDir() {
-			return nil
+	rel = filepath.ToSlash(filepath.Clean(rel))
+	for id, item := range metadata.Items {
+		if filepath.ToSlash(filepath.Clean(filepath.FromSlash(item.FriendlyPath))) == rel {
+			return id, nil
 		}
-		if os.SameFile(info, fi) {
-			found = strings.TrimSuffix(filepath.Base(p), filepath.Ext(p))
-			return filepath.SkipAll
 	}
-		return nil
-	})
-	if err != nil {
-		return "", err
-	}
-	if found == "" {
-		return "", fmt.Errorf("could not map mounted path to item id; try a by-id path")
-	}
-	return found, nil
+	return "", fmt.Errorf("could not map mounted path to item id")
 }

 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 != "" {
@@ -343,7 +342,8 @@ func buildMountTree(ctx context.Context, client *DocspellClient, items []SearchI
 			modTime = time.UnixMilli(date)
 			yearMonth = modTime.Format("2006/01")
 		}
-		dir := ensureDir(root, folderName, strings.Split(yearMonth, "/")...)
+		dirParts := append([]string{folderName}, strings.Split(yearMonth, "/")...)
+		dir := ensureDir(root, dirParts[0], dirParts[1:]...)
 		base := sanitizeName(item.Name)
 		if base == "" {
 			base = item.ID
@@ -366,6 +366,7 @@ func buildMountTree(ctx context.Context, client *DocspellClient, items []SearchI
 			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)...))

 			byIDName := item.ID + ext
 			if len(item.Attachments) > 1 {
@@ -375,12 +376,35 @@ func buildMountTree(ctx context.Context, client *DocspellClient, items []SearchI
 			}
 			shard := idShard(item.ID)
 			shardDir := ensureDir(byID, shard)
-			shardDir.children[uniqueName(shardDir.children, sanitizeName(byIDName))] = node
+			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}
+	}
 	return root
 }

+func sanitizePathParts(parts []string) []string {
+	sanitized := make([]string, len(parts))
+	for i, part := range parts {
+		part = sanitizeName(part)
+		if part == "" {
+			part = "_"
+		}
+		sanitized[i] = part
+	}
+	return sanitized
+}
+
 func idShard(id string) string {
 	id = sanitizeName(id)
 	if len(id) >= 2 {
@@ -491,6 +515,28 @@ func (f *fileNode) ReadAll(ctx context.Context) ([]byte, error) {
 	return f.load(ctx)
 }

+func (f *staticFileNode) Attr(ctx context.Context, a *fuse.Attr) error {
+	a.Mode = 0444
+	a.Size = uint64(len(f.data))
+	a.Mtime = f.modTime
+	a.Ctime = f.modTime
+	return nil
+}
+
+func (f *staticFileNode) Open(ctx context.Context, req *fuse.OpenRequest, resp *fuse.OpenResponse) error {
+	resp.Flags |= fuse.OpenDirectIO
+	return nil
+}
+
+func (f *staticFileNode) Read(ctx context.Context, req *fuse.ReadRequest, resp *fuse.ReadResponse) error {
+	fuseutil.HandleRead(req, resp, f.data)
+	return nil
+}
+
+func (f *staticFileNode) ReadAll(ctx context.Context) ([]byte, error) {
+	return f.data, nil
+}
+
 func (f *fileNode) load(ctx context.Context) ([]byte, error) {
 	if f.data != nil {
 		return f.data, nil
@@ -2,6 +2,7 @@ package main

 import (
 	"context"
+	"encoding/json"
 	"os"
 	"path/filepath"
 	"testing"
@@ -59,6 +60,28 @@ func TestItemIDFromLocalByIDPath(t *testing.T) {
 	}
 }

+func TestItemIDFromLocalFriendlyPathUsesMetadata(t *testing.T) {
+	mountPath := t.TempDir()
+	metadata := mountMetadata{Items: map[string]mountedItemMetadata{
+		"item-1": {FriendlyPath: "Bills/2024/07/Invoice.pdf"},
+	}}
+	data, err := json.Marshal(metadata)
+	if err != nil {
+		t.Fatal(err)
+	}
+	if err := os.WriteFile(filepath.Join(mountPath, mountMetadataFileName), data, 0644); err != nil {
+		t.Fatal(err)
+	}
+
+	got, err := itemIDFromLocalPath(filepath.Join(mountPath, "Bills", "2024", "07", "Invoice.pdf"), config{MountPath: mountPath})
+	if err != nil {
+		t.Fatal(err)
+	}
+	if got != "item-1" {
+		t.Fatalf("expected item-1, got %q", got)
+	}
+}
+
 func TestBuildMountTreeDeduplicatesAttachmentNames(t *testing.T) {
 	items := []SearchItem{{
 		ID:   "item-1",
@@ -121,50 +144,54 @@ func TestBuildMountTreeAddsByIDDirectory(t *testing.T) {
 	}
 }

-func TestFriendlyLocalPathPrefersNonByIDPath(t *testing.T) {
-	mountPath := t.TempDir()
-	friendlyDir := filepath.Join(mountPath, "Bills", "2024", "07")
-	byIDDir := filepath.Join(mountPath, "by-id", "it")
-	if err := os.MkdirAll(friendlyDir, 0755); err != nil {
+func TestBuildMountTreeAddsMetadataLookup(t *testing.T) {
+	items := []SearchItem{{
+		ID:   "item-1",
+		Name: "Invoice/July",
+		Date: 1720396800000,
+		Folder: &DocspellEntity{
+			Name: "Bills/Private",
+		},
+		Attachments: []ItemAttachment{{ID: "att-1", Name: "invoice.pdf", Size: 123}},
+	}}
+
+	root := buildMountTree(context.Background(), &DocspellClient{}, items)
+	metadataNode, ok := root.children[mountMetadataFileName].(*staticFileNode)
+	if !ok {
+		t.Fatalf("expected metadata file, got %#v", root.children[mountMetadataFileName])
+	}
+	var metadata mountMetadata
+	if err := json.Unmarshal(metadataNode.data, &metadata); err != nil {
 		t.Fatal(err)
 	}
-	if err := os.MkdirAll(byIDDir, 0755); err != nil {
-		t.Fatal(err)
+	got := metadata.Items["item-1"]
+	if got.FriendlyPath != "Bills_Private/2024/07/Invoice_July.pdf" {
+		t.Fatalf("expected friendly metadata path, got %q", got.FriendlyPath)
 	}
-	friendlyPath := filepath.Join(friendlyDir, "Invoice.pdf")
-	byIDPath := filepath.Join(byIDDir, "item-1.pdf")
-	if err := os.WriteFile(friendlyPath, []byte("pdf"), 0644); err != nil {
-		t.Fatal(err)
+	if got.ByIDPath != "by-id/it/item-1.pdf" {
+		t.Fatalf("expected by-id metadata path, got %q", got.ByIDPath)
 	}
-	if err := os.Link(friendlyPath, byIDPath); err != nil {
-		t.Fatal(err)
 }

-	got, err := friendlyLocalPath(mountPath, byIDPath)
+func TestFriendlyLocalPathForItemIDUsesMetadata(t *testing.T) {
+	mountPath := t.TempDir()
+	metadata := mountMetadata{Items: map[string]mountedItemMetadata{
+		"item-1": {FriendlyPath: "Bills/2024/07/Invoice.pdf"},
+	}}
+	data, err := json.Marshal(metadata)
 	if err != nil {
 		t.Fatal(err)
 	}
-	if got != friendlyPath {
-		t.Fatalf("expected friendly path %q, got %q", friendlyPath, got)
-	}
-}
-
-func TestFriendlyLocalPathReturnsEmptyWhenNoFriendlyPathExists(t *testing.T) {
-	mountPath := t.TempDir()
-	byIDDir := filepath.Join(mountPath, "by-id", "it")
-	if err := os.MkdirAll(byIDDir, 0755); err != nil {
-		t.Fatal(err)
-	}
-	byIDPath := filepath.Join(byIDDir, "item-1.pdf")
-	if err := os.WriteFile(byIDPath, []byte("pdf"), 0644); err != nil {
+	if err := os.WriteFile(filepath.Join(mountPath, mountMetadataFileName), data, 0644); err != nil {
 		t.Fatal(err)
 	}

-	got, err := friendlyLocalPath(mountPath, byIDPath)
+	got, err := friendlyLocalPathForItemID(mountPath, "item-1")
 	if err != nil {
 		t.Fatal(err)
 	}
-	if got != "" {
-		t.Fatalf("expected no friendly path, got %q", got)
+	want := filepath.Join(mountPath, "Bills", "2024", "07", "Invoice.pdf")
+	if got != want {
+		t.Fatalf("expected %q, got %q", want, got)
 	}
 }