这是unicode还是str？

Question

I have a dictionary that I need to share from Python 3 to Python 2. I dumped it as JSON in P3 and then loaded in Python2. Finally I (thought I) converted to P2 string using yaml.load .

Now there is something that seems wrong to me. I know that there are some unicode string(s) that are breaking a package. I tried to find them using

[key  for key in my_dict if type(key) != str]
[] 

Why then

'C%2B%2B' in my_dict
True

and

type('C%2B%2B')
unicode

is not caught? Why does the list comprehension think it is a string?

[type(key)  for key in my_dict if key == 'C%2B%2B']
[str]

Finally, is there a way I can convert it to str so that it will not break my code?

EDIT: ipython2 seems to have no consistent behavior on my computer:

vs

Answer 1

You must keep control over your data before encoding it to JSON. In Python 3, make sure all of the strings are unicode objects (ie of type str ), and then write them to a JSON file using eg UTF-8 encoding. Then decode that file using the same codec in Python 2. The resulting string types must be unicode . The code snippets you are showing do not help much, it is unclear in which version of Python you executed each of the snippets shown (the type('C%2B%2B') == unicode case suggests that you are using Python 2 with from __future__ import unicode_literals , the last snippet, however, is likely created from within Python 3). You must understand that type str in Python 2 is fundamentally different from str in Python 3.