将数据框转换为嵌套的 json

Question

我在用

pd.read_sql_query() 

从数据库中获取数据，然后使用

to_json(orient='records') 

这是数据框：

(1)
  price_formula_id  premium  product_id  exchange  product_name  product_code   weight  
0            30064      0.0        c001       CME          2018            CL      0.3
1            30064      0.0        c002       CME          2018            CL      0.7

(2)
price_formula_id  premium  product_id  exchange  product_name  product_code   weight  
0            30064      NONE        c001       CME          2018            CL      0.3
1            30064      NONE        c002       CME          2018            CL      0.7

转换成这个阵型。

[{
    "price_formula_id": "30064",
    "premium": "0.0",
    "product_id": "c001",
    "exchange": "CME",
    "product_name": "2018",
    "product_code": "CL",
    "weight": "0.3"
},
{
    "price_formula_id": "30064",
    "premium": "0.0",
    "product_id": "c002",
    "exchange": "CME",
    "product_name": "2018",
    "product_code": "CL",
    "weight": "0.7"
}]

但我真正想要的应该是这样的：

 { 
   "price_formula_id": "30064",
   "premium": "0.0",
   "basket": 
    [
     {"product_id": "c001",
      "exchange": "CME",
      "product_name": "2018",
      "product_code": "CL",
      "weight": "0.3"
     },
     {
      "product_id": "c002",
      "exchange": "CME",
      "product_name": "2018",
      "product_code": "CL",
      "weight": "0.7"
     }
    ]
 }

我需要对相同的信息进行分组，并为其余信息设置一个新的索引“篮子”。我怎么能做到呢？非常感谢。

Answer 1

groupby与自定义函数一起使用，to_dict用于所有由 过滤的列difference，reset_index最后将其转换为to_json：

cols = df.columns.difference(['price_formula_id','premium'])
j = (df.groupby(['price_formula_id','premium'])[cols]
       .apply(lambda x: x.to_dict('r'))
       .reset_index(name='basket')
       .to_json(orient='records'))
print (j)

[{
    "price_formula_id": 30064,
    "premium": 0.0,
    "basket": [{
            "exchange": "CME",
            "product_code": "CL",
            "product_id": "c001",
            "product_name": 2018,
            "weight": 0.3
        },
        {
            "exchange": "CME",
            "product_code": "CL",
            "product_id": "c002",
            "product_name": 2018,
            "weight": 0.7
        }
    ]
}]