从树数据中查找根节点

Question

I have the following class:

class Category(object):
    def __init__(self, *args, **kwargs):
        self.id = kwargs.get('id')
        self.name = kwargs.get('name')
        self.parent_id = kwargs.get('parent_id', None)

    def get_top_parent_category(self, obj_list):
        # This algorithm is using extensive resource 
        category = self
        while category.parent_id:
            def filter_func(obj):
                return obj.id == category.parent_id
            parent = list(filter(filter_func, obj_list))

            category = parent[0]
        return category

    @classmethod
    def all(cls):
        url = BASE_URL + '/api/v1/categories/'
        response = requests.get(url, headers=headers, verify=False)
        if not response.status_code == 200:
            raise Exception('Error')
        categories = response.json()
        _temp_categories = []
        for _i in categories['results']:
            _temp_categories.append(cls(**_i))
        return _temp_categories

I am getting all categories by:

all_categories = Category.all()

Now I need to find root node of any Category provided.

category = Category(**category_data)
category.get_top_parent_category(all_categories)

I get the desired results but i feel there might be some better way to find the root node using Graph Theory

What can be the better approach solving this problem?

Answer 1

If you need to do any more tree-related processing on it, you might want to link the Category objects to each other, instead of the indirection via the parent identifier.

But in the code you post the main problem is these repeated calls where you have to scan the entire object list:

parent = list(filter(filter_func, obj_list))

If you replace this with a dictionary, your performance will be a lot better, since the lookup time for a single parent will be ~ constant time

eg just for example

parent_map = dict([(c.id, c) for c in obj_list])

(obviously don't do this inside the get_top_parent_category() method, since it's just as expensive)

Then looking up the parent of a category can be done with a simple:

parent = parent_map[parent.id]

The same loop you have right now would be an order of magnitude faster this way.