Python pandas：对多级列框架的列进行排序重新排序

我有以下数据:

from pandas import Timestamp

values = [['IDX100', 'field1', Timestamp('1999-02-01 05:00:00'), '101'],
       ['IDX100', 'field1', Timestamp('1999-02-02 05:00:00'), '102'],
       ['IDX100', 'field1', Timestamp('1999-02-03 05:00:00'), '103'],
       ['IDX200', 'field1', Timestamp('1999-02-01 05:00:00'), '601'],
       ['IDX200', 'field1', Timestamp('1999-02-02 05:00:00'), '602'],
       ['IDX200', 'field1', Timestamp('1999-02-03 05:00:00'), '603'],
       ['IDX100', 'field2', Timestamp('1999-02-01 05:00:00'), '201'],
       ['IDX100', 'field2', Timestamp('1999-02-02 05:00:00'), '202'],
       ['IDX100', 'field2', Timestamp('1999-02-03 05:00:00'), '203'],
       ['IDX200', 'field2', Timestamp('1999-02-01 05:00:00'), '701'],
       ['IDX200', 'field2', Timestamp('1999-02-02 05:00:00'), '702'],
       ['IDX200', 'field2', Timestamp('1999-02-03 05:00:00'), '703'],
       ['IDX100', 'field3', Timestamp('1999-02-01 05:00:00'), '301'],
       ['IDX100', 'field3', Timestamp('1999-02-02 05:00:00'), '302'],
       ['IDX100', 'field3', Timestamp('1999-02-03 05:00:00'), '303'],
       ['IDX200', 'field3', Timestamp('1999-02-01 05:00:00'), '801'],
       ['IDX200', 'field3', Timestamp('1999-02-02 05:00:00'), '802'],
       ['IDX200', 'field3', Timestamp('1999-02-03 05:00:00'), '803']]

df = pd.DataFrame(values, columns = ['identifier', 'code', 'date', 'value'])

在旋转我的框架后，我结束了以下:

df = df.pivot(index=['date'], columns=['identifier', 'code'], values=['value'])

                     value                                   
identifier          IDX100 IDX200 IDX100 IDX200 IDX100 IDX200
code                field1 field1 field2 field2 field3 field3
date                                                         
1999-02-01 05:00:00    101    601    201    701    301    801
1999-02-02 05:00:00    102    602    202    702    302    802
1999-02-03 05:00:00    103    603    203    703    303    803

但是，我希望输出看起来像这样:

identifier           IDX100                IDX200 
code                 field3 field2 field1  field3 field2 field1
date                                                         
1999-02-01 05:00:00    301    201    101   801    701    601
1999-02-02 05:00:00    302    202    102   802    702    602
1999-02-03 05:00:00    303    203    103   803    703    603

我可以通过这样做来接近这一点:

df = df.reindex(sorted(df.columns), axis=1)

但这将保持level 2列的顺序为field1、field2、field3.我想要的是能够以不同的方式排序这个级别.最好是根据我提供的名单来确定. 例如，我可能希望将其排序为field 3，field 2，field 1，或field 2，field 1，field 3.

有人能帮我吗？