如何使基于`unicode`的逻辑兼容python2-3

Question

我想让这段代码与python2-3兼容：

def normalize_text(text, ignore_characters=''):
    if type(ignore_characters) not in [list, str, unicode]:
        ignore_characters = ''
    if type(ignore_characters) == str:
        ignore_characters = ignore_characters.decode('utf-8')
    if type(ignore_characters) == list:
        new_ignore_characters = []
        for item in ignore_characters:
            if type(item) == str:
                new_ignore_characters.append(item.decode('utf-8'))
            elif type(item) == unicode:
                new_ignore_characters.append(item)
        ignore_characters = new_ignore_characters

    if type(text) == str:
        text = text.decode('utf-8')

在Python3中的str类型上没有unicode或decode。什么是使这段代码与python2-3兼容的最佳解决方案？

Answer 1

我强烈推荐使用库来编写与Python2/3兼容的代码。

另外，用isinstance()代替type()来检查类型。在多重继承的情况下，type()将不起作用：

from six import text_type, binary_type

if isinstance(ignore_characters, binary_type):
   # do something with bytes
elif isinstance(ignore_characters, text_type):
   # do something with unicode.

# Python 2 
>>> import six
>>> six.binary_type, six.text_type
(<type 'str'>, <type 'unicode'>)

# Python 3
>>> import six
>>> six.binary_type, six.text_type
(<class 'bytes'>, <class 'str'>)

另一种方法是根据使用sys.version_info获得的Python版本编写自己的别名，以获得兼容性

库中的compat.py文件就是一个很好的例子：

_ver = sys.version_info

#: Python 2.x?
is_py2 = (_ver[0] == 2)

#: Python 3.x?
is_py3 = (_ver[0] == 3)

if is_py2:
    builtin_str = str
    bytes = str
    str = unicode
    basestring = basestring
    numeric_types = (int, long, float)
    integer_types = (int, long)

elif is_py3:
    builtin_str = str
    str = str
    bytes = bytes
    basestring = (str, bytes)
    numeric_types = (int, float)
    integer_types = (int,)

现在，您可以从此文件导入这些函数，而不是直接使用内置函数。