json.dump unicode 来自 Python 2 中的 unicode？ - 类型错误

Question

追溯

    dump({'foo':{'bar': 5}}, getwriter('utf8')(f), ensure_ascii=False, encoding='utf8')
  File "/usr/lib/pypy/lib-python/2.7/json/__init__.py", line 196, in dump
    fp.write(chunk)
  File "/usr/lib/pypy/lib-python/2.7/codecs.py", line 370, in write
    self.stream.write(data)
TypeError: unicode argument expected, got 'str'

代码

from __future__ import unicode_literals

from codecs import getwriter
from io import open
from os import path
from tempfile import gettempdir
from json import dump

with open(path.join(gettempdir(), 'bar'), 'wt', encoding='utf8') as f:
    dump({'foo':{'bar': 5}}, getwriter('utf8')(f), ensure_ascii=False,
         encoding='utf8')

(run with ideone)

Answer 1

这很奇怪，将导入更改为：

from io import open

收件人：

from codecs import open

成功了。尽管https://stackoverflow.com/a/844443 说第一个是新 最佳实践：\