如何获取嵌套字典列表中所有键的路径

Question

我想获取列表中嵌套字典的所有键的路径。例如，如果我的字典如下所示

{
"persons": [{
    "id": "f4d322fa8f552",
    "address": {
        "building": "710",
        "coord": "[123, 465]",
        "street": "Avenue Road",
        "zipcode": "12345"
    },
    "cuisine": "Chinese",
    "grades": [{
        "date": "2013-03-03T00:00:00.000Z",
        "grade": "B",
        "score": {
          "x": 3,
          "y": 2
        }
    }, {
        "date": "2012-11-23T00:00:00.000Z",
        "grade": "C",
        "score": {
          "x": 1,
          "y": 22
        }
    }],
    "name": "Shash"
}]
}

我想获得 path = [['persons'], ['persons','id'],['persons','address'],['persons','address','building']...]直到最后一个键的路径。

我尝试遍历整个字典来附加路径变量。试图从打印 python 嵌套字典的所有值的完整键路径中获得一些灵感，但我无法获取列表内的路径。

还有其他可能的方法来实现这一点吗？

Answer 1

您可以递归地描述数据结构，这是一种使用队列q与递归的方法。但很难判断这是否是您正在寻找的内容，因为它显示了列表索引，但可以很容易地排除它们：

def get_paths(d):
    q = [(d, [])]
    while q:
        n, p = q.pop(0)
        yield p
        if isinstance(n, dict):
            for k, v in n.items():
                q.append((v, p+[k]))
        elif isinstance(n, list):
            for i, v in enumerate(n):
                q.append((v, p+[i]))   # Change to q.append((v, p)) to remove index

In []:
list(get_paths(d))

Out[]:
[[],
 ['persons'],
 ['persons', 0],
 ['persons', 0, 'id'],
 ['persons', 0, 'address'],
 ['persons', 0, 'cuisine'],
 ['persons', 0, 'grades'],
 ['persons', 0, 'name'],
 ['persons', 0, 'address', 'building'],
 ['persons', 0, 'address', 'coord'],
 ['persons', 0, 'address', 'street'],
 ['persons', 0, 'address', 'zipcode'],
 ['persons', 0, 'grades', 0],
 ['persons', 0, 'grades', 1],
 ['persons', 0, 'grades', 0, 'date'],
 ['persons', 0, 'grades', 0, 'grade'],
 ['persons', 0, 'grades', 0, 'score'],
 ['persons', 0, 'grades', 1, 'date'],
 ['persons', 0, 'grades', 1, 'grade'],
 ['persons', 0, 'grades', 1, 'score'],
 ['persons', 0, 'grades', 0, 'score', 'x'],
 ['persons', 0, 'grades', 0, 'score', 'y'],
 ['persons', 0, 'grades', 1, 'score', 'x'],
 ['persons', 0, 'grades', 1, 'score', 'y'],