Staging
v0.8.1
v0.8.1
https://github.com/python/cpython
Tip revision: 74f4bd53e03ded8408bcc2de67cf0f5a4ac5b1a1 authored by Barry Warsaw on 23 February 2012, 15:59:38 UTC
Bump some more copyright years (as per PEP 101), since this is the first
Bump some more copyright years (as per PEP 101), since this is the first
Tip revision: 74f4bd5
utf_16.py
""" Python 'utf-16' Codec
Written by Marc-Andre Lemburg (mal@lemburg.com).
(c) Copyright CNRI, All Rights Reserved. NO WARRANTY.
"""
import codecs, sys
### Codec APIs
encode = codecs.utf_16_encode
def decode(input, errors='strict'):
return codecs.utf_16_decode(input, errors, True)
class IncrementalEncoder(codecs.IncrementalEncoder):
def __init__(self, errors='strict'):
codecs.IncrementalEncoder.__init__(self, errors)
self.encoder = None
def encode(self, input, final=False):
if self.encoder is None:
result = codecs.utf_16_encode(input, self.errors)[0]
if sys.byteorder == 'little':
self.encoder = codecs.utf_16_le_encode
else:
self.encoder = codecs.utf_16_be_encode
return result
return self.encoder(input, self.errors)[0]
def reset(self):
codecs.IncrementalEncoder.reset(self)
self.encoder = None
def getstate(self):
# state info we return to the caller:
# 0: stream is in natural order for this platform
# 2: endianness hasn't been determined yet
# (we're never writing in unnatural order)
return (2 if self.encoder is None else 0)
def setstate(self, state):
if state:
self.encoder = None
else:
if sys.byteorder == 'little':
self.encoder = codecs.utf_16_le_encode
else:
self.encoder = codecs.utf_16_be_encode
class IncrementalDecoder(codecs.BufferedIncrementalDecoder):
def __init__(self, errors='strict'):
codecs.BufferedIncrementalDecoder.__init__(self, errors)
self.decoder = None
def _buffer_decode(self, input, errors, final):
if self.decoder is None:
(output, consumed, byteorder) = \
codecs.utf_16_ex_decode(input, errors, 0, final)
if byteorder == -1:
self.decoder = codecs.utf_16_le_decode
elif byteorder == 1:
self.decoder = codecs.utf_16_be_decode
elif consumed >= 2:
raise UnicodeError("UTF-16 stream does not start with BOM")
return (output, consumed)
return self.decoder(input, self.errors, final)
def reset(self):
codecs.BufferedIncrementalDecoder.reset(self)
self.decoder = None
class StreamWriter(codecs.StreamWriter):
def __init__(self, stream, errors='strict'):
codecs.StreamWriter.__init__(self, stream, errors)
self.encoder = None
def reset(self):
codecs.StreamWriter.reset(self)
self.encoder = None
def encode(self, input, errors='strict'):
if self.encoder is None:
result = codecs.utf_16_encode(input, errors)
if sys.byteorder == 'little':
self.encoder = codecs.utf_16_le_encode
else:
self.encoder = codecs.utf_16_be_encode
return result
else:
return self.encoder(input, errors)
class StreamReader(codecs.StreamReader):
def reset(self):
codecs.StreamReader.reset(self)
try:
del self.decode
except AttributeError:
pass
def decode(self, input, errors='strict'):
(object, consumed, byteorder) = \
codecs.utf_16_ex_decode(input, errors, 0, False)
if byteorder == -1:
self.decode = codecs.utf_16_le_decode
elif byteorder == 1:
self.decode = codecs.utf_16_be_decode
elif consumed>=2:
raise UnicodeError,"UTF-16 stream does not start with BOM"
return (object, consumed)
### encodings module API
def getregentry():
return codecs.CodecInfo(
name='utf-16',
encode=encode,
decode=decode,
incrementalencoder=IncrementalEncoder,
incrementaldecoder=IncrementalDecoder,
streamreader=StreamReader,
streamwriter=StreamWriter,
)