python - Flatten 函数检索错误的值

Question

我目前有一个 Python 脚本，它使用 flatten 函数来展平 JSON 对象，然后我将使用 json_normalize将扁平的 JSON 转换为 DataFrame。

import pandas as pd
from pandas import json_normalize
from collections.abc import MutableMapping as mm

def flatten(dictionary, p_key=None, parent_key=False, separator='.'):
    items = []
    if isinstance(dictionary, list): 
        for listval in dictionary:
            items.extend(flatten(listval).items())
            return dict(items)
    for key, value in dictionary.items():
        if parent_key:
            new_key = f"{str(p_key)}{separator}{key}"
        else:
            new_key = p_key if p_key else key
        if isinstance(value, mm):
            items.extend(flatten(
                dictionary=value,
                p_key=new_key,
                parent_key=True,
                separator=separator).items())
        elif isinstance(value, list):
            for k, v in enumerate(value):
                items.extend(flatten(
                    dictionary={str(k): v},
                    p_key=new_key,
                    parent_key=False,
                    separator=separator).items())
        else:
            items.append((new_key, value))
    return dict(items)
    

jfile =  [{ 
    "id" : 1,
    "labelId" : 169,  
    "indicators" : [
        62
    ], 
    "Wait" : 6, 
    "Levels" : [
        {
            "isActive" : "true", 
            "pressure" : 3, 
            "actions" : [
                {
                    "isActive" : "true", 
                    "description" : "Place"
                }
            ], 
            "users" : [
                5467, 
                5469, 
                5
            ]
        }, 
        {
            "isActive" : "true", 
            "pressure" : 2, 
            "actions" : [
                {
                    "isActive" : "true", 
                    "description" : "Test"
                }
            ], 
            "users" : [
                3253, 
                6903
            ]
        }
    ]
}]


flatdoc = [flatten(i) for i in jfile]

flatdoc = json_normalize(flatdoc)

print(flatdoc)

当前输出:

   id  labelId  indicators  Wait Levels.isActive  Levels.pressure Levels.actions.isActive Levels.actions.description  Levels.users
   1      169          62     6            true                2                    true                       Test          6903

由于某种原因，我目前得到的输出是从 JSON 对象中找到的最后一个键中检索最后一个值。我需要它做的是从字典中的第一个键中检索它，然后当该列表没有嵌套并且只包含数字/整数时，它应该简单地分解它，以便结果如下所示:
预期输出:

   id  labelId  indicators  Wait Levels.isActive  Levels.pressure Levels.actions.isActive Levels.actions.description  Levels.users
   1      169          62     6            true                3                    true                       Place          5467
   1      169          62     6            true                3                    true                       Place          5469
   1      169          62     6            true                3                    true                       Place             5

我将如何修改函数以便它找到正确的键以产生适当的值？

Answer 1

您的展平功能是错误的:使用给定的 jfile它只返回一行，而预期会返回 5 行。
这是一个正确的版本:

def flatten(data):
    if isinstance(data, (list, tuple)):
        newdata = []
        for elt in data:
            elt = flatten(elt)
            if isinstance(elt, list):
                newdata.extend(elt)
            else:
                newdata.append(elt)
        return newdata
    elif isinstance(data, mm):
        for k,v in data.items():
            v = flatten(v)
            if isinstance(v, list):
                newdata = [data.copy() for _ in v]
                for i, elt in enumerate(v):
                    newdata[i][k] = elt
                return flatten(newdata)
            data[k] = v
    return data

然后您可以直接执行以下操作:

flatdoc = pd.json_normalize(flatten(jfile))

获得:

   id  labelId  indicators  Wait Levels.isActive  Levels.pressure Levels.actions.isActive Levels.actions.description  Levels.users
0   1      169          62     6            true                3                    true                      Place          5467
1   1      169          62     6            true                3                    true                      Place          5469
2   1      169          62     6            true                3                    true                      Place             5
3   1      169          62     6            true                2                    true                       Test          3253
4   1      169          62     6            true                2                    true                       Test          6903