有没有办法用normalizer保持递归子顺序?

我正在从服务器检索已排序的注释数组。每条评论都有 comment.children 属性,它也是其他注释的排序数组。这些可以嵌套 n 很深。例如:

const nestedComments = [
  {
    id: 1,
    body: "parent comment",
    children: [
        {
          id: 4,
          body: 'child comment',
          children: [
            {
              id: 7,
              body: 'grandchild comment #1',
              children: [],
            },
            {
              id: 6,
              body: 'grandchild comment #2',
              children: [],
            },
            {
              id: 5,
              body: 'grandchild comment #3',
              children: [],
            },
          ]
        },
        {
          id: 3,
          body: 'second child comment',
          children: []
        }
    ],
  },
  {
    id: 8,
    body: 'parent comment #2',
    children: []
  },
];

然后我使用normalizer库对其进行规范化,如下所示:

const commentSchema = new schema.Entity('comments');
const commentListSchema = new schema.Array(commentSchema);
commentSchema.define({children: commentListSchema});
const normalizedComments = normalize(nestedComments, commentListSchema);

结果与预期差不多:

{
  entities: {
    // All of the comments ordered their id's, this is fine and what I want
  },
  results: [1,8] // understandable but not what I want
}

因此,它保留根注释的顺序,但对嵌套的子注释不做任何操作。有没有办法让每一组兄弟姐妹都有自己的 results 阵列?像这样的东西:

results: [[1,8], [4,3], [7,6,5]]; 

或者也许有更好的方法来保存这些信息,我也很高兴听到这些。