src/allmydata/windows/fixups.py

   1
   2 done = False
   3
   4 def initialize():
   5     global done
   6     import sys
   7     if sys.platform != "win32" or done:
   8         return True
   9     done = True
  10
  11     import codecs, re
  12     from ctypes import WINFUNCTYPE, windll, CFUNCTYPE, cdll, POINTER, byref, \
  13         c_wchar_p, c_char_p, c_void_p, c_int, c_size_t
  14     from allmydata.util import log
  15     from allmydata.util.encodingutil import canonical_encoding
  16
  17     # Work around <http://bugs.python.org/issue6058>.
  18     codecs.register(lambda name: name == 'cp65001' and codecs.lookup('utf-8') or None)
  19
  20     # Make Unicode console output work independently of the current code page.
  21     # This also fixes <http://bugs.python.org/issue1602>.
  22     # Credit to Michael Kaplan <http://blogs.msdn.com/b/michkap/archive/2008/03/18/8306597.aspx>
  23     # and TZOmegaTZIOY
  24     # <http://stackoverflow.com/questions/878972/windows-cmd-encoding-change-causes-python-crash/1432462#1432462>.
  25     try:
  26         STDOUT_FILENO = 1
  27         STDERR_FILENO = 2
  28         real_stdout = hasattr(sys.stdout, 'fileno') and sys.stdout.fileno() == STDOUT_FILENO
  29         real_stderr = hasattr(sys.stderr, 'fileno') and sys.stderr.fileno() == STDERR_FILENO
  30
  31         def force_utf8(stream, name):
  32             if hasattr(stream, 'encoding') and canonical_encoding(stream.encoding) != 'utf-8':
  33                 log.msg("%s (%r) had encoding %r, but we're going to write UTF-8 to it" %
  34                         (name, stream, stream.encoding), level=log.CURIOUS)
  35             stream.encoding = 'utf-8'
  36
  37         if not real_stdout:
  38             force_utf8(sys.stdout, "sys.stdout")
  39
  40         if not real_stderr:
  41             force_utf8(sys.stderr, "sys.stderr")
  42
  43         if real_stdout or real_stderr:
  44             # FILE * _fdopen(int fd, const char *mode);
  45             # #define _IOLBF 0x0040
  46             # int setvbuf(FILE *stream, char *buffer, int mode, size_t size);
  47             # #define _O_U8TEXT 0x40000
  48             # int _setmode(int fd, int mode);
  49             # int fputws(const wchar_t *ws, FILE *stream);
  50             # int fflush(FILE *stream);
  51
  52             c_runtime = cdll.msvcrt
  53             NULL = None
  54             _fdopen = CFUNCTYPE(c_void_p, c_int, c_char_p)(("_fdopen", c_runtime))
  55             _IOLBF = 0x0040
  56             setvbuf = CFUNCTYPE(c_int, c_void_p, c_char_p, c_int, c_size_t)(("setvbuf", c_runtime))
  57             _O_U8TEXT = 0x40000
  58             _setmode = CFUNCTYPE(c_int, c_int, c_int)(("_setmode", c_runtime))
  59             fputws = CFUNCTYPE(c_int, c_wchar_p, c_void_p)(("fputws", c_runtime));
  60             fflush = CFUNCTYPE(c_int, c_void_p)(("fflush", c_runtime));
  61
  62             buffer_chars = 1024
  63
  64             class UnicodeOutput:
  65                 def __init__(self, fileno, name):
  66                     self._stream = _fdopen(fileno, "w")
  67                     assert self._stream is not NULL
  68
  69                     # Deep magic. MSVCRT supports writing wide-oriented output to stdout/stderr
  70                     # to the console using the Unicode APIs, but it does the conversion in the
  71                     # stdio buffer, so you need that buffer to be as large as the maximum amount
  72                     # you're going to write in a single call (in bytes, not characters).
  73                     setvbuf(self._stream, NULL, _IOLBF, buffer_chars*4 + 100)
  74                     _setmode(fileno, _O_U8TEXT)
  75
  76                     self._fileno = fileno
  77                     self.closed = False
  78                     self.softspace = False
  79                     self.mode = 'w'
  80                     self.encoding = 'utf-8'
  81                     self.name = name
  82
  83                 def isatty(self):
  84                     return False
  85                 def close(self):
  86                     self.closed = True
  87                     self.flush()
  88                 def fileno(self):
  89                     return self._fileno
  90                 def flush(self):
  91                     fflush(self._stream)
  92
  93                 def write(self, text):
  94                     if not isinstance(text, unicode):
  95                         text = str(text).decode('utf-8')
  96                     for i in xrange(0, len(text), buffer_chars):
  97                         fputws(text[i:(i+buffer_chars)], self._stream)
  98                         fflush(self._stream)
  99
 100                 def writelines(self, lines):
 101                     for line in lines:
 102                         self.write(line)
 103
 104             if real_stdout:
 105                 sys.stdout = UnicodeOutput(STDOUT_FILENO, '<Unicode stdout>')
 106
 107             if real_stderr:
 108                 sys.stderr = UnicodeOutput(STDERR_FILENO, '<Unicode stderr>')
 109     except Exception, e:
 110         log.msg("exception %r while fixing up sys.stdout and sys.stderr" % (e,), level=log.WEIRD)
 111
 112     # Unmangle command-line arguments.
 113     GetCommandLineW = WINFUNCTYPE(c_wchar_p)(("GetCommandLineW", windll.kernel32))
 114     CommandLineToArgvW = WINFUNCTYPE(POINTER(c_wchar_p), c_wchar_p, POINTER(c_int)) \
 115                             (("CommandLineToArgvW", windll.shell32))
 116
 117     argc = c_int(0)
 118     argv_unicode = CommandLineToArgvW(GetCommandLineW(), byref(argc))
 119
 120     def unmangle(s):
 121         return re.sub(ur'\x7f[0-9a-fA-F]*\;', lambda m: unichr(int(m.group(0)[1:-1], 16)), s)
 122
 123     try:
 124         sys.argv = [unmangle(argv_unicode[i]).encode('utf-8') for i in xrange(1, argc.value)]
 125     except Exception, e:
 126         print >>sys.stderr, "%s:  could not unmangle Unicode arguments" % (sys.argv[0],)
 127         print >>sys.stderr, [argv_unicode[i] for i in xrange(1, argc.value)]
 128         raise
 129
 130     if sys.argv[0].endswith('.pyscript'):
 131         sys.argv[0] = sys.argv[0][:-9]