Python: How to pass non-ASCII file names to Popen on Windows? - windows

Question

When I run a subprocess with Python, everything id fine with ASCII params but it fails if a parameter is an unicode (cyrillic) string:
cmd = [ 'dir.exe', u'по-русски' ]
p = subprocess.Popen([ 'dir.exe', u'по-русски' ])
Error log:
Traceback (most recent call last):
File "process.py", line 48, in <module>
cyrillic()
File "process.py", line 45, in cyrillic
p = subprocess.Popen(cmd, shell=True, stdin=None, stdout=None, stderr=subprocess.PIPE)
File "C:\Python\27\Lib\subprocess.py", line 679, in __init__
errread, errwrite)
File "C:\Python\27\Lib\subprocess.py", line 870, in _execute_child
args = '{} /c "{}"'.format (comspec, args)
UnicodeEncodeError: 'ascii' codec can't encode characters in position 8-10: ordinal not in range(128)
I tried different executables - 7z.ex, ls.exe - popen fails before even running them.
But what if I encode unicode string to a specific encoding?
# it works because 1251 is kinda native encoding for my Windows
cmd = [ 'dir.exe', CYRILLIC_FILE_NAME.encode('windows-1251') ]
# fails because 1257 cannot be converted to 1251 without errors
cmd = [ 'dir.exe', BALTIC_FILE_NAME.encode('windows-1251') ]
# this may work but it's not a solution because...
cmd = [ 'dir.exe', BALTIC_FILE_NAME.encode('windows-1257') ]
The "bad" thing, I have different file names on my computer - baltic, cyrillic and many more. So it looks there is no general way to pass a non-ASCII file name to Popen on Windows?! Or could that be fixed still? (Without dirty hacks preferably.)
Windows 7, Python 2.7.3

Answer 1

If you use Python 3, it will pass the parameters properly as Unicode. Assuming your subprocess can load unicode parameters on the command line (Python 2 cannot), then it should work.
For example, this script, when run under Python 3, will display the Cyrillic characters.
import subprocess
subprocess.call(["powershell", "-c", "echo", "'по-русски'"])