使用嵌套数组将 Dataframe 转换为 Json

Question

我有一个数据框df，如下所示

   ID               Aisle            Residence        HomePhone        CellPhone
   ------------------------------------------------------------------------------
0  1245,3214        A1, A2, A3, A4   Home             NaN              888888888
1  5674             B2,B3            Cell             777777777        999999999

预期结果：

{
 "0":{
    {
      "column": "ID",
      "values": [
        "1245",
        "3214"
      ]
    },
    {
      "column": "Aisle",
      "values": [
        "A1",
        "A2",
        "A3",
        "A4"
      ]
    },
    {
      "column": "Residence",
      "values": [
        "Home"
      ]
    },
    {
      "column": "HomePhone",
      "values": []
    },
    {
      "column": "CellPhone",
      "values": [
        "888888888"
      ]
    }
   },
"1":{
    {
      "column": "ID",
      "values": [
        "5674"
      ]
    },
    {
      "column": "Aisle",
      "values": [
        "B2",
        "B3"
      ]
    },
    {
      "column": "Residence",
      "values": [
        "Cell"
      ]
    },
    {
      "column": "HomePhone",
      "values": [
        "777777777"
      ]
    },
    {
      "column": "CellPhone",
      "values": [
        "999999999"
      ]
    }
   },

我有 2 行，分别是 0 和 1，而 json 下的每一行都有信息。 所以基本上，我想添加属性并将它们分配给列名的值，例如"column":"Aisle","values":["A1","B1,...]"

另外，我有一个限制，即列名总是会更改（ID、Aisle、Residence...等）并且列数会有所不同，因此在从 DF 转换为 JSON 时我无法对列进行硬编码。

Answer 1

我认为这是最接近您正在寻找的东西。使用DataFrame.to_json：

df2=df.copy()
df2[['ID','Aisle']]=df2[['ID','Aisle']].apply(lambda x: x.replace(' ','').str.split(','))
print(df2)

---

             ID                Aisle Residence    HomePhone  CellPhone
0  [1245, 3214]  [A1,  A2,  A3,  A4]      Home          NaN  888888888
1        [5674]             [B2, B3]      Cell  777777777.0  999999999

---

df2.T.to_json()

输出：

'{"0":{"ID":["1245","3214"],"Aisle":["A1"," A2"," A3"," A4"],"Residence":"Home","HomePhone":null,"CellPhone":888888888},"1":{"ID":["5674"],"Aisle":["B2","B3"],"Residence":"Cell","HomePhone":777777777.0,"CellPhone":999999999}}'

---

你也可以试试这个：

def split_func(x):
    try:
        return x.replace(' ','').str.split(',') 
    except:
        return x
df2=df2.apply(split_func)

但请记住，所有 str 类型的单元格都将转换为列表

df2.T.to_json()

---

'{"0":{"ID":["1245","3214"],"Aisle":["A1","A2","A3","A4"],"Residence":["Home"],"HomePhone":null,"CellPhone":888888888},"1":{"ID":["5674"],"Aisle":["B2","B3"],"Residence":["Cell"],"HomePhone":777777777.0,"CellPhone":999999999}}'