gistfile1.py

# from anyall.org/util.py

#########  Make UTF-8 hurt less

# My rant about pre-py3k encoding handling
# They like to say, always use unicode internally, then decode/encode at I/O boundaries
# That's good once you've accomplished it, but it's impractical without the following shims
# Since Python has inconsistent policies for what encoding an arbitrary stream will be.

def fix_stdio(encoding='utf8', errors='strict', buffering=0):
  """ forces utf8 at I/O boundaries, since it's ascii by default when using
  pipes .. ugh ..  Never call this multple times in the same process; horrible
  things sometimes seem to happen."""
  en,er,bu=encoding,errors,buffering
  sys.stdout = codecs.open('/dev/stdout', 'w', encoding=en, errors=er, buffering=bu)
  sys.stdout = ShutUpAboutBrokenPipe(sys.stdout)
  sys.stdin  = codecs.open('/dev/stdin',  'r', encoding=en, errors=er, buffering=bu)
  sys.stderr = codecs.open('/dev/stderr', 'w', encoding=en, errors=er, buffering=0)

def unicodify(s, encoding='utf8', *args):
  """ because {str,unicode}.{encode,decode} is anti-polymorphic, but sometimes
  you can't control which you have. """
  if isinstance(s,unicode): return s
  return s.decode(encoding, *args)

def stringify(s, encoding='utf8', *args):
  if isinstance(s,str): return s
  return s.encode(encoding, *args)

class ShutUpAboutBrokenPipe:
  """i like to press ctrl-c; why is python yelling at me?"""
  def __init__(self, fp):
    self.fp = fp
  def write(self,*a,**k):
    try:
      self.fp.write(*a,**k)
    except IOError, e:
      if e.errno == 32:  # broken pipe
        sys.exit(0)
      raise e