print u'测试中文'

睺raceback (most recent call last):
File "C:\Users\__test.py", line 96, in <module>
sys.exit(main())
File "C:\Users\__test.py", line 84, in main
print u'娴嬭瘯涓枃'
IOError: [Errno 28] No space left on device

the unicode must start with a ascii char, then print can works
like this: "print u' 测试中文'"

What is the code page of your console (try: "chcp").

It looks like a duplicate of the issue bpo-11395 which has been fixed in
Python 3.2.1, but not backported to Python 2.7. Can you please try
your program with Python 3.2.1? And maybe also attach the whole test
script?

Victor: why do you think this is a duplicate of bpo-11395? That issue deals with long output strings, which doesn't seem to be the case here.

Victor: why do you think this is a duplicate of bpo-11395? That issue deals with long output strings, which doesn't seem to be the case here.

IOError('[Errno 28] No space left on device') was exactly the bpo-11395
error message. Without the full test script, i cannot be sure that
it's a duplicate. "C:\Users\__test.py, line 96": it looks like the
program has ~100 lines and so may write something else to the console.
There are buffers in TextIOWrapper, BufferedWriter and the Windows
console itself. By the way, Python maps the Windows error
ERROR_DISK_FULL (112) to ENOSPC (28).

It's maybe something completly different.

IOError('[Errno 28] No space left on device') was exactly the bpo-11395
error message.

I read that report differently:

IOError: [Errno 12] Not enough space

ENOMEM != ENOSPC

Closing this as out-of-date. The original bug was reported against 2.7 and Win8, both of which are either end-of-life now.

No follow up in 9 years and I'm quite certain that modern Pythons handle all manner of Unicode chars on the console.