mirror of https://github.com/python/cpython.git
105 lines
2.8 KiB
Python
105 lines
2.8 KiB
Python
# Copyright (C) 2001 Python Software Foundation
|
||
# Author: barry@zope.com (Barry Warsaw)
|
||
|
||
"""Miscellaneous utilities.
|
||
"""
|
||
|
||
import re
|
||
|
||
from rfc822 import unquote, quote, parseaddr
|
||
from rfc822 import dump_address_pair
|
||
from rfc822 import AddrlistClass as _AddrlistClass
|
||
from rfc822 import parsedate_tz, parsedate, mktime_tz, formatdate
|
||
|
||
from quopri import decodestring as _qdecode
|
||
import base64
|
||
|
||
# Intrapackage imports
|
||
from Encoders import _bencode, _qencode
|
||
|
||
COMMASPACE = ', '
|
||
UEMPTYSTRING = u''
|
||
|
||
|
||
|
||
# Helpers
|
||
|
||
def _identity(s):
|
||
return s
|
||
|
||
|
||
def _bdecode(s):
|
||
if not s:
|
||
return s
|
||
# We can't quite use base64.encodestring() since it tacks on a "courtesy
|
||
# newline". Blech!
|
||
if not s:
|
||
return s
|
||
hasnewline = (s[-1] == '\n')
|
||
value = base64.decodestring(s)
|
||
if not hasnewline and value[-1] == '\n':
|
||
return value[:-1]
|
||
return value
|
||
|
||
|
||
|
||
def getaddresses(fieldvalues):
|
||
"""Return a list of (REALNAME, EMAIL) for each fieldvalue."""
|
||
all = COMMASPACE.join(fieldvalues)
|
||
a = _AddrlistClass(all)
|
||
return a.getaddrlist()
|
||
|
||
|
||
|
||
ecre = re.compile(r'''
|
||
=\? # literal =?
|
||
(?P<charset>[^?]*?) # non-greedy up to the next ? is the charset
|
||
\? # literal ?
|
||
(?P<encoding>[qb]) # either a "q" or a "b", case insensitive
|
||
\? # literal ?
|
||
(?P<atom>.*?) # non-greedy up to the next ?= is the atom
|
||
\?= # literal ?=
|
||
''', re.VERBOSE | re.IGNORECASE)
|
||
|
||
|
||
def decode(s):
|
||
"""Return a decoded string according to RFC 2047, as a unicode string."""
|
||
rtn = []
|
||
parts = ecre.split(s, 1)
|
||
while parts:
|
||
# If there are less than 4 parts, it can't be encoded and we're done
|
||
if len(parts) < 5:
|
||
rtn.extend(parts)
|
||
break
|
||
# The first element is any non-encoded leading text
|
||
rtn.append(parts[0])
|
||
charset = parts[1]
|
||
encoding = parts[2]
|
||
atom = parts[3]
|
||
# The next chunk to decode should be in parts[4]
|
||
parts = ecre.split(parts[4])
|
||
# The encoding must be either `q' or `b', case-insensitive
|
||
if encoding.lower() == 'q':
|
||
func = _qdecode
|
||
elif encoding.lower() == 'b':
|
||
func = _bdecode
|
||
else:
|
||
func = _identity
|
||
# Decode and get the unicode in the charset
|
||
rtn.append(unicode(func(atom), charset))
|
||
# Now that we've decoded everything, we just need to join all the parts
|
||
# together into the final string.
|
||
return UEMPTYSTRING.join(rtn)
|
||
|
||
|
||
|
||
def encode(s, charset='iso-8859-1', encoding='q'):
|
||
"""Encode a string according to RFC 2047."""
|
||
if encoding.lower() == 'q':
|
||
estr = _qencode(s)
|
||
elif encoding.lower() == 'b':
|
||
estr = _bencode(s)
|
||
else:
|
||
raise ValueError, 'Illegal encoding code: ' + encoding
|
||
return '=?%s?%s?%s?=' % (charset.lower(), encoding.lower(), estr)
|