eryksun added the comment:
> It would be interesting to know under what circumstances these
> functions can fail.
The CRT _put[w]ch and _get[w]ch[e] functions will fail when called in a process
that doesn't have a console. (Except get[w]ch may succeed if it follows
unget[w]ch.) This is the case when running under pythonw.exe if the process
hasn't first attached to a console via AllocConsole() or AttachConsole(). It
also applies to python.exe when run as a detached process (i.e. the creation
flag DETACHED_PROCESS) or after having detached the console via FreeConsole().
Note that even though the docs for get[w]ch [1] and get[w]che [2] state that
"[t]here is no error return", these functions actually do return an error value
of [W]EOF. This has been the case since at least back to Windows NT. Maybe
Steve Dower can shed light on why this is undocumented.
Here's an example, tested on 64-bit Windows 10 in Python 3.5. This example
calls the console I/O functions using both msvcrt and ctypes. Since there's no
attached console, an error is expected, except when calling ungetwch followed
by getwch.
import os
import sys
import subprocess
cmds_msvcrt = [
"import msvcrt; msvcrt.ungetwch('a'); msvcrt.getwch()",
"import msvcrt; msvcrt.ungetwch('a'); msvcrt.ungetwch('a')",
"import msvcrt; msvcrt.getwch()",
"import msvcrt; msvcrt.putwch('a')",
]
csetup = "import sys,ctypes; ucrt = ctypes.cdll.ucrtbase; "
cmds_ctypes = [
csetup + "ucrt._ungetwch(ord('a')); sys.exit(ucrt._getwch())",
csetup + "ucrt._ungetwch(ord('a')); sys.exit(ucrt._ungetwch(ord('a')))",
csetup + "sys.exit(ucrt._getwch())",
csetup + "sys.exit(ucrt._putwch(ord('a')))",
]
def test(cmds):
pythonw = os.path.join(sys.prefix, 'pythonw.exe')
return [subprocess.call([pythonw, '-c', cmd]) for cmd in cmds]
>>> test(cmds_msvcrt)
[0, 1, 0, 0]
>>> test(cmds_ctypes)
[97, 65535, 65535, 65535]
65535 is WEOF.
[1]: https://msdn.microsoft.com/en-us/library/078sfkak
[2]: https://msdn.microsoft.com/en-us/library/kswce429
----------
nosy: +eryksun
_______________________________________
Python tracker <rep...@bugs.python.org>
<http://bugs.python.org/issue25386>
_______________________________________
_______________________________________________
Python-bugs-list mailing list
Unsubscribe:
https://mail.python.org/mailman/options/python-bugs-list/archive%40mail-archive.com
