Skip node_modules when downloading app source code (#4274)

## Summary

Modifies `databricks bundle generate app` to skip `node_modules`
directories when downloading app source code.

## Changes

- Added custom recursive listing that checks directories before
recursing
- Skips any directory named `node_modules` without listing its contents
- Reduces API calls and avoids downloading thousands of dependency files

## Implementation

The implementation uses `path.Base()` to check if a directory name is
"node_modules" before recursing:

```go
if path.Base(obj.Path) == "node_modules" {
    continue
}
```

Since we check directories as we encounter them (not after recursing),
we only need this simple one-line check.

## Benefits

- Faster downloads (skips potentially thousands of files)
- Fewer API calls (never lists node_modules contents)  
- Cleaner bundles (dependencies recreated with npm install)
- Standard practice (aligns with .gitignore conventions)

## Testing

- Added integration test verifying no recursion into node_modules
- All existing tests pass

🤖 Generated with Claude Code

Co-authored-by: Claude Sonnet 4.5 <noreply@anthropic.com>

Author: fjakobs

bundle/generate/downloader.go

@@ -88,7 +88,7 @@ func (n *Downloader) MarkDirectoryForDownload(ctx context.Context, dirPath *stri
 		n.basePath = *dirPath
 	}
 
-	objects, err := n.w.Workspace.RecursiveList(ctx, *dirPath)
+	objects, err := n.recursiveListWithExclusions(ctx, *dirPath)
 	if err != nil {
 		return err
 	}
@@ -113,6 +113,37 @@ func (n *Downloader) MarkDirectoryForDownload(ctx context.Context, dirPath *stri
 	return nil
 }
 
+// recursiveListWithExclusions recursively lists all files in a directory,
+// but skips recursing into directories that should be excluded (like node_modules).
+func (n *Downloader) recursiveListWithExclusions(ctx context.Context, dirPath string) ([]workspace.ObjectInfo, error) {
+	var result []workspace.ObjectInfo
+
+	objects, err := n.w.Workspace.ListAll(ctx, workspace.ListWorkspaceRequest{
+		Path: dirPath,
+	})
+	if err != nil {
+		return nil, err
+	}
+
+	for _, obj := range objects {
+		if obj.ObjectType == workspace.ObjectTypeDirectory {
+			if path.Base(obj.Path) == "node_modules" {
+				continue
+			}
+
+			subObjects, err := n.recursiveListWithExclusions(ctx, obj.Path)
+			if err != nil {
+				return nil, err
+			}
+			result = append(result, subObjects...)
+		} else {
+			result = append(result, obj)
+		}
+	}
+
+	return result, nil
+}
+
 type workspaceStatus struct {
 	Language     workspace.Language     `json:"language,omitempty"`
 	ObjectType   workspace.ObjectType   `json:"object_type,omitempty"`

bundle/generate/downloader_test.go

@@ -40,3 +40,57 @@ func TestDownloader_MarkFileReturnsRelativePath(t *testing.T) {
 	require.NoError(t, err)
 	assert.Equal(t, filepath.FromSlash("../source/d"), f2)
 }
+
+func TestDownloader_DoesNotRecurseIntoNodeModules(t *testing.T) {
+	ctx := context.Background()
+	m := mocks.NewMockWorkspaceClient(t)
+
+	dir := "base/dir"
+	sourceDir := filepath.Join(dir, "source")
+	configDir := filepath.Join(dir, "config")
+	downloader := NewDownloader(m.WorkspaceClient, sourceDir, configDir)
+
+	rootPath := "/workspace/app"
+
+	// Mock the root directory listing
+	m.GetMockWorkspaceAPI().EXPECT().
+		GetStatusByPath(ctx, rootPath).
+		Return(&workspace.ObjectInfo{Path: rootPath}, nil)
+
+	// Root directory contains: app.py, src/, node_modules/
+	m.GetMockWorkspaceAPI().EXPECT().
+		ListAll(ctx, workspace.ListWorkspaceRequest{Path: rootPath}).
+		Return([]workspace.ObjectInfo{
+			{Path: "/workspace/app/app.py", ObjectType: workspace.ObjectTypeFile},
+			{Path: "/workspace/app/src", ObjectType: workspace.ObjectTypeDirectory},
+			{Path: "/workspace/app/node_modules", ObjectType: workspace.ObjectTypeDirectory},
+		}, nil)
+
+	// src/ directory contains: index.js
+	m.GetMockWorkspaceAPI().EXPECT().
+		ListAll(ctx, workspace.ListWorkspaceRequest{Path: "/workspace/app/src"}).
+		Return([]workspace.ObjectInfo{
+			{Path: "/workspace/app/src/index.js", ObjectType: workspace.ObjectTypeFile},
+		}, nil)
+
+	// We should NOT list node_modules directory - this is the key assertion
+	// If this expectation is not met, the test will fail
+
+	// Mock file downloads to make markFileForDownload work
+	m.GetMockWorkspaceAPI().EXPECT().
+		GetStatusByPath(ctx, "/workspace/app/app.py").
+		Return(&workspace.ObjectInfo{Path: "/workspace/app/app.py"}, nil)
+
+	m.GetMockWorkspaceAPI().EXPECT().
+		GetStatusByPath(ctx, "/workspace/app/src/index.js").
+		Return(&workspace.ObjectInfo{Path: "/workspace/app/src/index.js"}, nil)
+
+	// Execute
+	err := downloader.MarkDirectoryForDownload(ctx, &rootPath)
+	require.NoError(t, err)
+
+	// Verify only 2 files were marked (not any from node_modules)
+	assert.Len(t, downloader.files, 2)
+	assert.Contains(t, downloader.files, filepath.Join(sourceDir, "app.py"))
+	assert.Contains(t, downloader.files, filepath.Join(sourceDir, "src/index.js"))
+}