我有一个csv文件,其中一些作为x; y; z格式的列。我正在使用熊猫读取此数据,进行一些预处理并使用熊猫的to_json / to_dict方法转换为json对象列表。在转换这些特殊列时,该列的json对象应采用{x:{y:{z:value}}}的格式。可能会有不同的列,例如x:y:z和x:y:a,这2列必须合并在一起,形成格式为{x:{y:{z:value1, a:value2}}}
CSV:
Id,Name,X;Y;Z,X;Y;A,X;B;Z
101,Adam,1,2,3
102,John,4,5,6
103,Sara,7,8,9
输出:
[
{
"Id":101,
"Name":"Adam",
"X":{
"Y":{
"Z":1,
"A":2
},
"B":{
"Z":3
}
}
},
{
"Id":102,
"Name":"John",
"X":{
"Y":{
"Z":4,
"A":5
},
"B":{
"Z":6
}
}
},
{
"Id":103,
"Name":"Sara",
"X":{
"Y":{
"Z":7,
"A":8
},
"B":{
"Z":9
}
}
}
]
答案 0 :(得分:1)
我发现使用pandas将数据作为dict转储会更容易,然后使用递归函数来遍历键,并且在遇到包含;
的键的情况下,我用此分隔符将键拆分为递归地创建嵌套的字典。当我到达分割键中的最后一个元素时,我会用原始值更新键,然后从dict中删除原始键。
import pandas as pd
from io import StringIO
import json
def split_key_to_nested_dict(original_dict, original_key, nested_dict, nested_keys):
if nested_keys[0] not in nested_dict:
nested_dict[nested_keys[0]] = {}
if len(nested_keys) == 1:
nested_dict[nested_keys[0]] = original_dict[original_key]
del original_dict[original_key]
else:
split_key_to_nested_dict(original_dict, original_key, nested_dict[nested_keys[0]], nested_keys[1:])
csv_data = StringIO("""Id,Name,X;Y;Z,X;Y;A,X;B;Z
101,Adam,1,2,3
102,John,4,5,6
103,Sara,7,8,9""")
df = pd.DataFrame.from_csv(csv_data)
df.insert(0, df.index.name, df.index)
dict_data = df.to_dict('records')
for data in dict_data:
keys = list(data.keys())
for key in keys:
if ';' in key:
nested_keys = key.split(';')
split_key_to_nested_dict(data, key, data, nested_keys)
print(json.dumps(dict_data))
输出
[{"Id": 101, "Name": "Adam", "X": {"Y": {"Z": 1, "A": 2}, "B": {"Z": 3}}}, {"Id": 102, "Name": "John", "X": {"Y": {"Z": 4, "A": 5}, "B": {"Z": 6}}}, {"Id": 103, "Name": "Sara", "X": {"Y": {"Z": 7, "A": 8}, "B": {"Z": 9}}}]
格式化输出
[
{
"Id": 101,
"Name": "Adam",
"X": {
"Y": {
"Z": 1,
"A": 2
},
"B": {
"Z": 3
}
}
},
{
"Id": 102,
"Name": "John",
"X": {
"Y": {
"Z": 4,
"A": 5
},
"B": {
"Z": 6
}
}
},
{
"Id": 103,
"Name": "Sara",
"X": {
"Y": {
"Z": 7,
"A": 8
},
"B": {
"Z": 9
}
}
}
]