{DRAFT]feat: implement attachment extraction and sync pipeline

internal/applescript/extractor.go

@@ -4,6 +4,9 @@ import (
 	"context"
 	"embed"
 	"fmt"
+	"os"
+	"path/filepath"
+	"strings"
 
 	"go.uber.org/zap"
 
@@ -23,6 +26,11 @@ type NoteExtractor interface {
 	// If accounts is nil, all accounts are included.
 	// If folders is nil, all folders are included.
 	GetAllNotes(ctx context.Context, accounts []string, folders []string) ([]model.Note, error)
+
+	// ResolveAttachments locates attachment files in the Apple Notes media
+	// directory and populates the Data field for each attachment. Attachments
+	// larger than maxSizeMB are skipped.
+	ResolveAttachments(ctx context.Context, notes []model.Note, maxSizeMB int) error
 }
 
 // AppleScriptExtractor extracts notes from Apple Notes by executing
@@ -133,3 +141,128 @@ func toSet(items []string) map[string]bool {
 	}
 	return set
 }
+
+// notesMediaDir is the directory where Apple Notes stores attachment files.
+const notesMediaDir = "Library/Group Containers/group.com.apple.notes"
+
+// ResolveAttachments walks the Apple Notes media directory and populates
+// attachment Data fields by matching filenames. Attachments larger than
+// maxSizeMB are skipped.
+func (e *AppleScriptExtractor) ResolveAttachments(ctx context.Context, notes []model.Note, maxSizeMB int) error {
+	homeDir, err := os.UserHomeDir()
+	if err != nil {
+		return fmt.Errorf("getting home directory: %w", err)
+	}
+
+	mediaRoot := filepath.Join(homeDir, notesMediaDir)
+	if _, err := os.Stat(mediaRoot); os.IsNotExist(err) {
+		e.logger.Warn("Apple Notes media directory not found", zap.String("path", mediaRoot))
+		return nil
+	}
+
+	return e.resolveAttachmentsFromDir(ctx, mediaRoot, notes, maxSizeMB)
+}
+
+// resolveAttachmentsFromDir is the core implementation of ResolveAttachments,
+// separated to allow testing with a custom directory.
+func (e *AppleScriptExtractor) resolveAttachmentsFromDir(ctx context.Context, mediaRoot string, notes []model.Note, maxSizeMB int) error {
+	fileIndex, err := buildFileIndex(ctx, mediaRoot)
+	if err != nil {
+		return fmt.Errorf("indexing Apple Notes media: %w", err)
+	}
+
+	e.logger.Debug("built attachment file index", zap.Int("files", len(fileIndex)))
+
+	maxBytes := int64(maxSizeMB) * 1024 * 1024
+	resolved := 0
+
+	for i := range notes {
+		for j := range notes[i].Attachments {
+			att := &notes[i].Attachments[j]
+			if att.Name == "" {
+				continue
+			}
+
+			paths, ok := fileIndex[att.Name]
+			if !ok || len(paths) == 0 {
+				e.logger.Debug("attachment file not found in media directory",
+					zap.String("name", att.Name),
+					zap.String("note", notes[i].Name),
+				)
+				continue
+			}
+
+			// Use the first match. If multiple exist, prefer the one matching
+			// the content identifier if possible.
+			filePath := paths[0]
+			if att.ContentID != "" && len(paths) > 1 {
+				for _, p := range paths {
+					if strings.Contains(p, att.ContentID) {
+						filePath = p
+						break
+					}
+				}
+			}
+
+			info, err := os.Stat(filePath)
+			if err != nil {
+				e.logger.Debug("cannot stat attachment file", zap.String("path", filePath), zap.Error(err))
+				continue
+			}
+
+			if info.Size() > maxBytes {
+				e.logger.Debug("skipping oversized attachment",
+					zap.String("name", att.Name),
+					zap.Int64("size_bytes", info.Size()),
+					zap.Int("max_mb", maxSizeMB),
+				)
+				continue
+			}
+
+			data, err := os.ReadFile(filePath)
+			if err != nil {
+				e.logger.Warn("failed to read attachment file",
+					zap.String("path", filePath),
+					zap.Error(err),
+				)
+				continue
+			}
+
+			att.Data = data
+			resolved++
+		}
+	}
+
+	e.logger.Info("resolved attachments", zap.Int("count", resolved))
+	return nil
+}
+
+// buildFileIndex walks a directory tree and returns a map from filename to
+// all absolute paths where that filename exists.
+func buildFileIndex(ctx context.Context, root string) (map[string][]string, error) {
+	index := make(map[string][]string)
+
+	err := filepath.Walk(root, func(path string, info os.FileInfo, err error) error {
+		if err != nil {
+			return nil // Skip inaccessible files.
+		}
+
+		select {
+		case <-ctx.Done():
+			return fmt.Errorf("context cancelled: %w", ctx.Err())
+		default:
+		}
+
+		if info.IsDir() {
+			return nil
+		}
+
+		index[info.Name()] = append(index[info.Name()], path)
+		return nil
+	})
+	if err != nil {
+		return nil, fmt.Errorf("walking %q: %w", root, err)
+	}
+
+	return index, nil
+}

internal/applescript/extractor_test.go

@@ -2,6 +2,8 @@ package applescript
 
 import (
 	"context"
+	"os"
+	"path/filepath"
 	"testing"
 
 	"github.com/stretchr/testify/assert"
@@ -76,8 +78,8 @@ func TestAppleScriptExtractor_GetAllNotes(t *testing.T) {
 	mockExec := new(MockCommandExecutor)
 	extractor := NewAppleScriptExtractor(mockExec, newTestLogger())
 
-	output := "id1|||FIELD|||Note 1|||FIELD|||<p>body1</p>|||FIELD|||Monday, January 1, 2026 at 10:00:00 AM|||FIELD|||Monday, January 1, 2026 at 10:00:00 AM|||FIELD|||iCloud|||FIELD|||Notes|||FIELD|||false|||FIELD|||false|||NOTE|||" +
-		"id2|||FIELD|||Note 2|||FIELD|||<p>body2</p>|||FIELD|||Tuesday, January 2, 2026 at 11:00:00 AM|||FIELD|||Tuesday, January 2, 2026 at 11:00:00 AM|||FIELD|||Gmail|||FIELD|||Work|||FIELD|||false|||FIELD|||false|||NOTE|||"
+	output := "id1|||FIELD|||Note 1|||FIELD|||<p>body1</p>|||FIELD|||Monday, January 1, 2026 at 10:00:00 AM|||FIELD|||Monday, January 1, 2026 at 10:00:00 AM|||FIELD|||iCloud|||FIELD|||Notes|||FIELD|||false|||FIELD|||false|||FIELD||||||NOTE|||" +
+		"id2|||FIELD|||Note 2|||FIELD|||<p>body2</p>|||FIELD|||Tuesday, January 2, 2026 at 11:00:00 AM|||FIELD|||Tuesday, January 2, 2026 at 11:00:00 AM|||FIELD|||Gmail|||FIELD|||Work|||FIELD|||false|||FIELD|||false|||FIELD||||||NOTE|||"
 
 	mockExec.On("Execute", mock.Anything, "osascript", mock.Anything).
 		Return(&shell.CommandResult{Stdout: output}, nil)
@@ -96,8 +98,8 @@ func TestAppleScriptExtractor_GetAllNotes_FilterByAccount(t *testing.T) {
 	mockExec := new(MockCommandExecutor)
 	extractor := NewAppleScriptExtractor(mockExec, newTestLogger())
 
-	output := "id1|||FIELD|||Note 1|||FIELD|||<p>body1</p>|||FIELD|||Monday, January 1, 2026 at 10:00:00 AM|||FIELD|||Monday, January 1, 2026 at 10:00:00 AM|||FIELD|||iCloud|||FIELD|||Notes|||FIELD|||false|||FIELD|||false|||NOTE|||" +
-		"id2|||FIELD|||Note 2|||FIELD|||<p>body2</p>|||FIELD|||Tuesday, January 2, 2026 at 11:00:00 AM|||FIELD|||Tuesday, January 2, 2026 at 11:00:00 AM|||FIELD|||Gmail|||FIELD|||Work|||FIELD|||false|||FIELD|||false|||NOTE|||"
+	output := "id1|||FIELD|||Note 1|||FIELD|||<p>body1</p>|||FIELD|||Monday, January 1, 2026 at 10:00:00 AM|||FIELD|||Monday, January 1, 2026 at 10:00:00 AM|||FIELD|||iCloud|||FIELD|||Notes|||FIELD|||false|||FIELD|||false|||FIELD||||||NOTE|||" +
+		"id2|||FIELD|||Note 2|||FIELD|||<p>body2</p>|||FIELD|||Tuesday, January 2, 2026 at 11:00:00 AM|||FIELD|||Tuesday, January 2, 2026 at 11:00:00 AM|||FIELD|||Gmail|||FIELD|||Work|||FIELD|||false|||FIELD|||false|||FIELD||||||NOTE|||"
 
 	mockExec.On("Execute", mock.Anything, "osascript", mock.Anything).
 		Return(&shell.CommandResult{Stdout: output}, nil)
@@ -114,8 +116,8 @@ func TestAppleScriptExtractor_GetAllNotes_FilterByFolder(t *testing.T) {
 	mockExec := new(MockCommandExecutor)
 	extractor := NewAppleScriptExtractor(mockExec, newTestLogger())
 
-	output := "id1|||FIELD|||Note 1|||FIELD|||<p>body1</p>|||FIELD|||Monday, January 1, 2026 at 10:00:00 AM|||FIELD|||Monday, January 1, 2026 at 10:00:00 AM|||FIELD|||iCloud|||FIELD|||Notes|||FIELD|||false|||FIELD|||false|||NOTE|||" +
-		"id2|||FIELD|||Note 2|||FIELD|||<p>body2</p>|||FIELD|||Tuesday, January 2, 2026 at 11:00:00 AM|||FIELD|||Tuesday, January 2, 2026 at 11:00:00 AM|||FIELD|||iCloud|||FIELD|||Work|||FIELD|||false|||FIELD|||false|||NOTE|||"
+	output := "id1|||FIELD|||Note 1|||FIELD|||<p>body1</p>|||FIELD|||Monday, January 1, 2026 at 10:00:00 AM|||FIELD|||Monday, January 1, 2026 at 10:00:00 AM|||FIELD|||iCloud|||FIELD|||Notes|||FIELD|||false|||FIELD|||false|||FIELD||||||NOTE|||" +
+		"id2|||FIELD|||Note 2|||FIELD|||<p>body2</p>|||FIELD|||Tuesday, January 2, 2026 at 11:00:00 AM|||FIELD|||Tuesday, January 2, 2026 at 11:00:00 AM|||FIELD|||iCloud|||FIELD|||Work|||FIELD|||false|||FIELD|||false|||FIELD||||||NOTE|||"
 
 	mockExec.On("Execute", mock.Anything, "osascript", mock.Anything).
 		Return(&shell.CommandResult{Stdout: output}, nil)
@@ -195,3 +197,111 @@ func TestToSet(t *testing.T) {
 	assert.True(t, s["b"])
 	assert.False(t, s["c"])
 }
+
+func TestBuildFileIndex(t *testing.T) {
+	dir := t.TempDir()
+
+	// Create some files in subdirectories.
+	subDir := filepath.Join(dir, "sub1", "sub2")
+	require.NoError(t, os.MkdirAll(subDir, 0755))
+	require.NoError(t, os.WriteFile(filepath.Join(dir, "photo.jpg"), []byte("img1"), 0644))
+	require.NoError(t, os.WriteFile(filepath.Join(subDir, "photo.jpg"), []byte("img2"), 0644))
+	require.NoError(t, os.WriteFile(filepath.Join(subDir, "doc.pdf"), []byte("pdf"), 0644))
+
+	index, err := buildFileIndex(context.Background(), dir)
+	require.NoError(t, err)
+
+	assert.Len(t, index["photo.jpg"], 2)
+	assert.Len(t, index["doc.pdf"], 1)
+	assert.Empty(t, index["missing.txt"])
+}
+
+func TestBuildFileIndex_EmptyDir(t *testing.T) {
+	dir := t.TempDir()
+
+	index, err := buildFileIndex(context.Background(), dir)
+	require.NoError(t, err)
+	assert.Empty(t, index)
+}
+
+func TestResolveAttachmentsFromDir(t *testing.T) {
+	// Create a fake Notes media directory.
+	dir := t.TempDir()
+	subDir := filepath.Join(dir, "uuid-123")
+	require.NoError(t, os.MkdirAll(subDir, 0755))
+	require.NoError(t, os.WriteFile(filepath.Join(subDir, "photo.jpg"), []byte("fake-image"), 0644))
+	require.NoError(t, os.WriteFile(filepath.Join(subDir, "huge.bin"), make([]byte, 2*1024*1024), 0644))
+
+	mockExec := new(MockCommandExecutor)
+	extractor := NewAppleScriptExtractor(mockExec, newTestLogger())
+
+	notes := []model.Note{
+		{
+			Name: "Test Note",
+			Attachments: []model.Attachment{
+				{Name: "photo.jpg", ContentID: "cid-1"},
+				{Name: "huge.bin", ContentID: "cid-2"},
+				{Name: "missing.png", ContentID: "cid-3"},
+			},
+		},
+	}
+
+	// Use 1 MB max to test size filtering.
+	err := extractor.resolveAttachmentsFromDir(context.Background(), dir, notes, 1)
+	require.NoError(t, err)
+
+	// photo.jpg should be resolved (< 1 MB).
+	assert.Equal(t, []byte("fake-image"), notes[0].Attachments[0].Data)
+	// huge.bin should be skipped (> 1 MB).
+	assert.Nil(t, notes[0].Attachments[1].Data)
+	// missing.png should be nil.
+	assert.Nil(t, notes[0].Attachments[2].Data)
+}
+
+func TestResolveAttachmentsFromDir_ContentIDMatch(t *testing.T) {
+	// Create two files with the same name in different dirs.
+	dir := t.TempDir()
+	dir1 := filepath.Join(dir, "uuid-AAA")
+	dir2 := filepath.Join(dir, "uuid-BBB")
+	require.NoError(t, os.MkdirAll(dir1, 0755))
+	require.NoError(t, os.MkdirAll(dir2, 0755))
+	require.NoError(t, os.WriteFile(filepath.Join(dir1, "image.png"), []byte("wrong"), 0644))
+	require.NoError(t, os.WriteFile(filepath.Join(dir2, "image.png"), []byte("correct"), 0644))
+
+	mockExec := new(MockCommandExecutor)
+	extractor := NewAppleScriptExtractor(mockExec, newTestLogger())
+
+	notes := []model.Note{
+		{
+			Name: "Note",
+			Attachments: []model.Attachment{
+				{Name: "image.png", ContentID: "BBB"},
+			},
+		},
+	}
+
+	err := extractor.resolveAttachmentsFromDir(context.Background(), dir, notes, 50)
+	require.NoError(t, err)
+
+	// Should match the one containing "BBB" in its path.
+	assert.Equal(t, []byte("correct"), notes[0].Attachments[0].Data)
+}
+
+func TestResolveAttachmentsFromDir_EmptyName(t *testing.T) {
+	dir := t.TempDir()
+	mockExec := new(MockCommandExecutor)
+	extractor := NewAppleScriptExtractor(mockExec, newTestLogger())
+
+	notes := []model.Note{
+		{
+			Name: "Note",
+			Attachments: []model.Attachment{
+				{Name: "", ContentID: "cid"},
+			},
+		},
+	}
+
+	err := extractor.resolveAttachmentsFromDir(context.Background(), dir, notes, 50)
+	require.NoError(t, err)
+	assert.Nil(t, notes[0].Attachments[0].Data)
+}