在 Pandas 数据框中读取非结构化词典

Question

我正在尝试从我从 json 文件中读取的字典集合中创建一个 Pandas 数据框。 字典如下——

d1 = {"DisplayName": "Test_drive", "permissions": {"read": True, "read_acp": True, "write": True, "write_acp": True}}
d2= {"DisplayName": "Log delivery","URI": "http://test_drive.com/Logs", "permissions": {"read": False, "read_acp": True, "write": True, "write_acp": False}}

我正在尝试将这些放入熊猫数据框中。 当我尝试在如下所示的数据框中读取它们时 -

df = pd.DataFrame(d) **or** df = pd.DataFrame.from_dict(d)

它产生这个 -

                DisplayName  permissions
read       Test_drive         True
read_acp   Test_drive         True
write      Test_drive         True
write_acp  Test_drive         True

或阅读如下 -

df1 = pd.DataFrame(d).Transpose()

它产生这个 -

                         read          read_acp             write         write_acp
DisplayName  Test_drive  Test_drive  Test_drive  Test_drive
permissions              True              True              True              True

我正在尝试阅读这些词典并将它们加入一个数据框 -

**DisplayName**              **read**          **read_acp**             **write**         **write_acp**         URI
Test_drive         True              True              True              True         NA
Log delivery            False             True              True             False         http://test_drive.com/Logs

有没有pytonic方法来做到这一点？

Answer 1

通过附加创建数据框，然后使用枢轴将其重塑为您需要的结构

df = pd.DataFrame.from_dict(d1).append(pd.DataFrame.from_dict(d2))
df.reset_index().pivot(index='DisplayName', columns='index', values='permissions')

包含 URI

>>> df.reset_index().pivot(index='DisplayName', columns='index', values=['permissions', 'URI'])
             permissions                                                  URI                                                                                    
index               read read_acp write write_acp                        read                    read_acp                       write                   write_acp
DisplayName                                                                                                                                                      
Log delivery       False     True  True     False  http://test_drive.com/Logs  http://test_drive.com/Logs  http://test_drive.com/Logs  http://test_drive.com/Logs
Test_drive          True     True  True      True                         NaN                         NaN                         NaN                         NaN

Answer 2

import pandas as pd

# Input Data
d1 = {"DisplayName": "Test_drive", "permissions": {"read": True, "read_acp": True, "write": True, "write_acp": True}}
d2= {"DisplayName": "Log delivery","URI": "http://test_drive.com/Logs", "permissions": {"read": False, "read_acp": True, "write": True, "write_acp": False}}

# Convert to DataFrame
dicts = [d1, d2]
df_rows = [pd.DataFrame(d) for d in dicts]
df = pd.concat(df_rows, axis=0).reset_index(drop=False)

# Reshape As Desired
tp1 = df.pivot(index='DisplayName', columns='index', values='permissions')
answer = tp1.merge(df[['DisplayName', 'URI']].drop_duplicates(), 
                   how='left', 
                   left_index=True, 
                   right_on='DisplayName').set_index('DisplayName')

输出：

>>> answer
               read  read_acp  write  write_acp                         URI
DisplayName                                                                
Log delivery  False      True   True      False  http://test_drive.com/Logs
Test_drive     True      True   True       True                         NaN

Answer 3

感谢Vishnudev和Max Power的帮助。 我认为以下答案为我提供了我试图获得的确切数据框。

d1 = {"DisplayName": "Test_drive", "permissions": {"read": True, "read_acp": True, "write": True, "write_acp": True}}
d2= {"DisplayName": "Log delivery","URI": "http://test_drive.com/Logs", "permissions": {"read": False, "read_acp": True, "write": True, "write_acp": False}}
df = pd.concat([pd.Series(d1),pd.Series(d2)], axis=1).transpose()
df = pd.concat([df.drop(['permissions'], axis=1),df['permissions'].apply(pd.Series)],axis=1)

**DisplayName                         URI   read  read_acp  write  write_acp**
0    Test_drive                         NaN   True      True   True       True
1  Log delivery  http://test_drive.com/Logs  False      True   True      False