Mercurial > p > roundup > code

annotate roundup/anypy/email_.py @ 4979:f1a2bd1dea77

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
issue2550877: Writing headers with the email module will use continuation_ws = ' ' now for python 2.5 and 2.6 when importing anypy.email_.
author Bernhard Reiter <bernhard@intevation.de>
date Mon, 16 Mar 2015 16:16:02 +0100
parents c426cb251bc7
children 28a72f125a14
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
4575
c426cb251bc7 Be more tolerant when parsing RFC2047 encoded mail headers.
Ralf Schlatterbeck <rsc@runtux.com>
parents: 4447
diff changeset
1 import re
c426cb251bc7 Be more tolerant when parsing RFC2047 encoded mail headers.
Ralf Schlatterbeck <rsc@runtux.com>
parents: 4447
diff changeset
2 import binascii
4979
f1a2bd1dea77 issue2550877: Writing headers with the email module will use continuation_ws = ' ' now for python 2.5 and 2.6 when importing anypy.email_.
Bernhard Reiter <bernhard@intevation.de>
parents: 4575
diff changeset
3 import email
4575
c426cb251bc7 Be more tolerant when parsing RFC2047 encoded mail headers.
Ralf Schlatterbeck <rsc@runtux.com>
parents: 4447
diff changeset
4 from email import quoprimime, base64mime
c426cb251bc7 Be more tolerant when parsing RFC2047 encoded mail headers.
Ralf Schlatterbeck <rsc@runtux.com>
parents: 4447
diff changeset
5
4979
f1a2bd1dea77 issue2550877: Writing headers with the email module will use continuation_ws = ' ' now for python 2.5 and 2.6 when importing anypy.email_.
Bernhard Reiter <bernhard@intevation.de>
parents: 4575
diff changeset
6 ## please import this file if you are using the email module
f1a2bd1dea77 issue2550877: Writing headers with the email module will use continuation_ws = ' ' now for python 2.5 and 2.6 when importing anypy.email_.
Bernhard Reiter <bernhard@intevation.de>
parents: 4575
diff changeset
7 #
f1a2bd1dea77 issue2550877: Writing headers with the email module will use continuation_ws = ' ' now for python 2.5 and 2.6 when importing anypy.email_.
Bernhard Reiter <bernhard@intevation.de>
parents: 4575
diff changeset
8 # a "monkey patch" to unify the behaviour of python 2.5 2.6 2.7
f1a2bd1dea77 issue2550877: Writing headers with the email module will use continuation_ws = ' ' now for python 2.5 and 2.6 when importing anypy.email_.
Bernhard Reiter <bernhard@intevation.de>
parents: 4575
diff changeset
9 # when generating header files, see http://bugs.python.org/issue1974
f1a2bd1dea77 issue2550877: Writing headers with the email module will use continuation_ws = ' ' now for python 2.5 and 2.6 when importing anypy.email_.
Bernhard Reiter <bernhard@intevation.de>
parents: 4575
diff changeset
10 # and https://hg.python.org/cpython/rev/5deb27042e5a/
f1a2bd1dea77 issue2550877: Writing headers with the email module will use continuation_ws = ' ' now for python 2.5 and 2.6 when importing anypy.email_.
Bernhard Reiter <bernhard@intevation.de>
parents: 4575
diff changeset
11 # can go away once the minimum requirement is python 2.7
f1a2bd1dea77 issue2550877: Writing headers with the email module will use continuation_ws = ' ' now for python 2.5 and 2.6 when importing anypy.email_.
Bernhard Reiter <bernhard@intevation.de>
parents: 4575
diff changeset
12 _oldheaderinit = email.Header.Header.__init__
f1a2bd1dea77 issue2550877: Writing headers with the email module will use continuation_ws = ' ' now for python 2.5 and 2.6 when importing anypy.email_.
Bernhard Reiter <bernhard@intevation.de>
parents: 4575
diff changeset
13 def _unifiedheaderinit(self, *args, **kw):
f1a2bd1dea77 issue2550877: Writing headers with the email module will use continuation_ws = ' ' now for python 2.5 and 2.6 when importing anypy.email_.
Bernhard Reiter <bernhard@intevation.de>
parents: 4575
diff changeset
14 # override continuation_ws
f1a2bd1dea77 issue2550877: Writing headers with the email module will use continuation_ws = ' ' now for python 2.5 and 2.6 when importing anypy.email_.
Bernhard Reiter <bernhard@intevation.de>
parents: 4575
diff changeset
15 kw['continuation_ws'] = ' '
f1a2bd1dea77 issue2550877: Writing headers with the email module will use continuation_ws = ' ' now for python 2.5 and 2.6 when importing anypy.email_.
Bernhard Reiter <bernhard@intevation.de>
parents: 4575
diff changeset
16 _oldheaderinit(self, *args, **kw)
f1a2bd1dea77 issue2550877: Writing headers with the email module will use continuation_ws = ' ' now for python 2.5 and 2.6 when importing anypy.email_.
Bernhard Reiter <bernhard@intevation.de>
parents: 4575
diff changeset
17 email.Header.Header.__dict__['__init__'] = _unifiedheaderinit
f1a2bd1dea77 issue2550877: Writing headers with the email module will use continuation_ws = ' ' now for python 2.5 and 2.6 when importing anypy.email_.
Bernhard Reiter <bernhard@intevation.de>
parents: 4575
diff changeset
18 ##
f1a2bd1dea77 issue2550877: Writing headers with the email module will use continuation_ws = ' ' now for python 2.5 and 2.6 when importing anypy.email_.
Bernhard Reiter <bernhard@intevation.de>
parents: 4575
diff changeset
19
4447
9d37875416c3 python2.3 compatibility fixes
Ralf Schlatterbeck <schlatterbeck@users.sourceforge.net>
parents:
diff changeset
20 try:
9d37875416c3 python2.3 compatibility fixes
Ralf Schlatterbeck <schlatterbeck@users.sourceforge.net>
parents:
diff changeset
21 # Python 2.5+
9d37875416c3 python2.3 compatibility fixes
Ralf Schlatterbeck <schlatterbeck@users.sourceforge.net>
parents:
diff changeset
22 from email.parser import FeedParser
9d37875416c3 python2.3 compatibility fixes
Ralf Schlatterbeck <schlatterbeck@users.sourceforge.net>
parents:
diff changeset
23 except ImportError:
9d37875416c3 python2.3 compatibility fixes
Ralf Schlatterbeck <schlatterbeck@users.sourceforge.net>
parents:
diff changeset
24 # Python 2.4
9d37875416c3 python2.3 compatibility fixes
Ralf Schlatterbeck <schlatterbeck@users.sourceforge.net>
parents:
diff changeset
25 try :
9d37875416c3 python2.3 compatibility fixes
Ralf Schlatterbeck <schlatterbeck@users.sourceforge.net>
parents:
diff changeset
26 from email.Parser import FeedParser
9d37875416c3 python2.3 compatibility fixes
Ralf Schlatterbeck <schlatterbeck@users.sourceforge.net>
parents:
diff changeset
27 except ImportError:
9d37875416c3 python2.3 compatibility fixes
Ralf Schlatterbeck <schlatterbeck@users.sourceforge.net>
parents:
diff changeset
28 from email.Parser import Parser
9d37875416c3 python2.3 compatibility fixes
Ralf Schlatterbeck <schlatterbeck@users.sourceforge.net>
parents:
diff changeset
29 class FeedParser:
9d37875416c3 python2.3 compatibility fixes
Ralf Schlatterbeck <schlatterbeck@users.sourceforge.net>
parents:
diff changeset
30 def __init__(self):
9d37875416c3 python2.3 compatibility fixes
Ralf Schlatterbeck <schlatterbeck@users.sourceforge.net>
parents:
diff changeset
31 self.content = []
9d37875416c3 python2.3 compatibility fixes
Ralf Schlatterbeck <schlatterbeck@users.sourceforge.net>
parents:
diff changeset
32
9d37875416c3 python2.3 compatibility fixes
Ralf Schlatterbeck <schlatterbeck@users.sourceforge.net>
parents:
diff changeset
33 def feed(self, s):
9d37875416c3 python2.3 compatibility fixes
Ralf Schlatterbeck <schlatterbeck@users.sourceforge.net>
parents:
diff changeset
34 self.content.append(s)
9d37875416c3 python2.3 compatibility fixes
Ralf Schlatterbeck <schlatterbeck@users.sourceforge.net>
parents:
diff changeset
35
9d37875416c3 python2.3 compatibility fixes
Ralf Schlatterbeck <schlatterbeck@users.sourceforge.net>
parents:
diff changeset
36 def close(self):
9d37875416c3 python2.3 compatibility fixes
Ralf Schlatterbeck <schlatterbeck@users.sourceforge.net>
parents:
diff changeset
37 p = Parser()
9d37875416c3 python2.3 compatibility fixes
Ralf Schlatterbeck <schlatterbeck@users.sourceforge.net>
parents:
diff changeset
38 return p.parsestr(''.join(self.content))
4575
c426cb251bc7 Be more tolerant when parsing RFC2047 encoded mail headers.
Ralf Schlatterbeck <rsc@runtux.com>
parents: 4447
diff changeset
39
c426cb251bc7 Be more tolerant when parsing RFC2047 encoded mail headers.
Ralf Schlatterbeck <rsc@runtux.com>
parents: 4447
diff changeset
40 # Match encoded-word strings in the form =?charset?q?Hello_World?=
c426cb251bc7 Be more tolerant when parsing RFC2047 encoded mail headers.
Ralf Schlatterbeck <rsc@runtux.com>
parents: 4447
diff changeset
41 ecre = re.compile(r'''
c426cb251bc7 Be more tolerant when parsing RFC2047 encoded mail headers.
Ralf Schlatterbeck <rsc@runtux.com>
parents: 4447
diff changeset
42 =\? # literal =?
c426cb251bc7 Be more tolerant when parsing RFC2047 encoded mail headers.
Ralf Schlatterbeck <rsc@runtux.com>
parents: 4447
diff changeset
43 (?P<charset>[^?]*?) # non-greedy up to the next ? is the charset
c426cb251bc7 Be more tolerant when parsing RFC2047 encoded mail headers.
Ralf Schlatterbeck <rsc@runtux.com>
parents: 4447
diff changeset
44 \? # literal ?
c426cb251bc7 Be more tolerant when parsing RFC2047 encoded mail headers.
Ralf Schlatterbeck <rsc@runtux.com>
parents: 4447
diff changeset
45 (?P<encoding>[qb]) # either a "q" or a "b", case insensitive
c426cb251bc7 Be more tolerant when parsing RFC2047 encoded mail headers.
Ralf Schlatterbeck <rsc@runtux.com>
parents: 4447
diff changeset
46 \? # literal ?
c426cb251bc7 Be more tolerant when parsing RFC2047 encoded mail headers.
Ralf Schlatterbeck <rsc@runtux.com>
parents: 4447
diff changeset
47 (?P<encoded>.*?) # non-greedy up to the next ?= is the encoded string
c426cb251bc7 Be more tolerant when parsing RFC2047 encoded mail headers.
Ralf Schlatterbeck <rsc@runtux.com>
parents: 4447
diff changeset
48 \?= # literal ?=
c426cb251bc7 Be more tolerant when parsing RFC2047 encoded mail headers.
Ralf Schlatterbeck <rsc@runtux.com>
parents: 4447
diff changeset
49 ''', re.VERBOSE | re.IGNORECASE | re.MULTILINE)
c426cb251bc7 Be more tolerant when parsing RFC2047 encoded mail headers.
Ralf Schlatterbeck <rsc@runtux.com>
parents: 4447
diff changeset
50
c426cb251bc7 Be more tolerant when parsing RFC2047 encoded mail headers.
Ralf Schlatterbeck <rsc@runtux.com>
parents: 4447
diff changeset
51
c426cb251bc7 Be more tolerant when parsing RFC2047 encoded mail headers.
Ralf Schlatterbeck <rsc@runtux.com>
parents: 4447
diff changeset
52 # Fixed header parser, see my proposed patch and discussions:
c426cb251bc7 Be more tolerant when parsing RFC2047 encoded mail headers.
Ralf Schlatterbeck <rsc@runtux.com>
parents: 4447
diff changeset
53 # http://bugs.python.org/issue1079 "decode_header does not follow RFC 2047"
c426cb251bc7 Be more tolerant when parsing RFC2047 encoded mail headers.
Ralf Schlatterbeck <rsc@runtux.com>
parents: 4447
diff changeset
54 # http://bugs.python.org/issue1467619 "Header.decode_header eats up spaces"
c426cb251bc7 Be more tolerant when parsing RFC2047 encoded mail headers.
Ralf Schlatterbeck <rsc@runtux.com>
parents: 4447
diff changeset
55 # This implements the decode_header specific parts of my proposed patch
c426cb251bc7 Be more tolerant when parsing RFC2047 encoded mail headers.
Ralf Schlatterbeck <rsc@runtux.com>
parents: 4447
diff changeset
56 # backported to python2.X
c426cb251bc7 Be more tolerant when parsing RFC2047 encoded mail headers.
Ralf Schlatterbeck <rsc@runtux.com>
parents: 4447
diff changeset
57 def decode_header(header):
c426cb251bc7 Be more tolerant when parsing RFC2047 encoded mail headers.
Ralf Schlatterbeck <rsc@runtux.com>
parents: 4447
diff changeset
58 """Decode a message header value without converting charset.
c426cb251bc7 Be more tolerant when parsing RFC2047 encoded mail headers.
Ralf Schlatterbeck <rsc@runtux.com>
parents: 4447
diff changeset
59
c426cb251bc7 Be more tolerant when parsing RFC2047 encoded mail headers.
Ralf Schlatterbeck <rsc@runtux.com>
parents: 4447
diff changeset
60 Returns a list of (string, charset) pairs containing each of the decoded
c426cb251bc7 Be more tolerant when parsing RFC2047 encoded mail headers.
Ralf Schlatterbeck <rsc@runtux.com>
parents: 4447
diff changeset
61 parts of the header. Charset is None for non-encoded parts of the header,
c426cb251bc7 Be more tolerant when parsing RFC2047 encoded mail headers.
Ralf Schlatterbeck <rsc@runtux.com>
parents: 4447
diff changeset
62 otherwise a lower-case string containing the name of the character set
c426cb251bc7 Be more tolerant when parsing RFC2047 encoded mail headers.
Ralf Schlatterbeck <rsc@runtux.com>
parents: 4447
diff changeset
63 specified in the encoded string.
c426cb251bc7 Be more tolerant when parsing RFC2047 encoded mail headers.
Ralf Schlatterbeck <rsc@runtux.com>
parents: 4447
diff changeset
64
c426cb251bc7 Be more tolerant when parsing RFC2047 encoded mail headers.
Ralf Schlatterbeck <rsc@runtux.com>
parents: 4447
diff changeset
65 header may be a string that may or may not contain RFC2047 encoded words,
c426cb251bc7 Be more tolerant when parsing RFC2047 encoded mail headers.
Ralf Schlatterbeck <rsc@runtux.com>
parents: 4447
diff changeset
66 or it may be a Header object.
c426cb251bc7 Be more tolerant when parsing RFC2047 encoded mail headers.
Ralf Schlatterbeck <rsc@runtux.com>
parents: 4447
diff changeset
67
c426cb251bc7 Be more tolerant when parsing RFC2047 encoded mail headers.
Ralf Schlatterbeck <rsc@runtux.com>
parents: 4447
diff changeset
68 An email.errors.HeaderParseError may be raised when certain decoding error
c426cb251bc7 Be more tolerant when parsing RFC2047 encoded mail headers.
Ralf Schlatterbeck <rsc@runtux.com>
parents: 4447
diff changeset
69 occurs (e.g. a base64 decoding exception).
c426cb251bc7 Be more tolerant when parsing RFC2047 encoded mail headers.
Ralf Schlatterbeck <rsc@runtux.com>
parents: 4447
diff changeset
70 """
c426cb251bc7 Be more tolerant when parsing RFC2047 encoded mail headers.
Ralf Schlatterbeck <rsc@runtux.com>
parents: 4447
diff changeset
71 # If it is a Header object, we can just return the encoded chunks.
c426cb251bc7 Be more tolerant when parsing RFC2047 encoded mail headers.
Ralf Schlatterbeck <rsc@runtux.com>
parents: 4447
diff changeset
72 if hasattr(header, '_chunks'):
c426cb251bc7 Be more tolerant when parsing RFC2047 encoded mail headers.
Ralf Schlatterbeck <rsc@runtux.com>
parents: 4447
diff changeset
73 return [(_charset._encode(string, str(charset)), str(charset))
c426cb251bc7 Be more tolerant when parsing RFC2047 encoded mail headers.
Ralf Schlatterbeck <rsc@runtux.com>
parents: 4447
diff changeset
74 for string, charset in header._chunks]
c426cb251bc7 Be more tolerant when parsing RFC2047 encoded mail headers.
Ralf Schlatterbeck <rsc@runtux.com>
parents: 4447
diff changeset
75 # If no encoding, just return the header with no charset.
c426cb251bc7 Be more tolerant when parsing RFC2047 encoded mail headers.
Ralf Schlatterbeck <rsc@runtux.com>
parents: 4447
diff changeset
76 if not ecre.search(header):
c426cb251bc7 Be more tolerant when parsing RFC2047 encoded mail headers.
Ralf Schlatterbeck <rsc@runtux.com>
parents: 4447
diff changeset
77 return [(header, None)]
c426cb251bc7 Be more tolerant when parsing RFC2047 encoded mail headers.
Ralf Schlatterbeck <rsc@runtux.com>
parents: 4447
diff changeset
78 # First step is to parse all the encoded parts into triplets of the form
c426cb251bc7 Be more tolerant when parsing RFC2047 encoded mail headers.
Ralf Schlatterbeck <rsc@runtux.com>
parents: 4447
diff changeset
79 # (encoded_string, encoding, charset). For unencoded strings, the last
c426cb251bc7 Be more tolerant when parsing RFC2047 encoded mail headers.
Ralf Schlatterbeck <rsc@runtux.com>
parents: 4447
diff changeset
80 # two parts will be None.
c426cb251bc7 Be more tolerant when parsing RFC2047 encoded mail headers.
Ralf Schlatterbeck <rsc@runtux.com>
parents: 4447
diff changeset
81 words = []
c426cb251bc7 Be more tolerant when parsing RFC2047 encoded mail headers.
Ralf Schlatterbeck <rsc@runtux.com>
parents: 4447
diff changeset
82 for line in header.splitlines():
c426cb251bc7 Be more tolerant when parsing RFC2047 encoded mail headers.
Ralf Schlatterbeck <rsc@runtux.com>
parents: 4447
diff changeset
83 parts = ecre.split(line)
c426cb251bc7 Be more tolerant when parsing RFC2047 encoded mail headers.
Ralf Schlatterbeck <rsc@runtux.com>
parents: 4447
diff changeset
84 first = True
c426cb251bc7 Be more tolerant when parsing RFC2047 encoded mail headers.
Ralf Schlatterbeck <rsc@runtux.com>
parents: 4447
diff changeset
85 while parts:
c426cb251bc7 Be more tolerant when parsing RFC2047 encoded mail headers.
Ralf Schlatterbeck <rsc@runtux.com>
parents: 4447
diff changeset
86 unencoded = parts.pop(0)
c426cb251bc7 Be more tolerant when parsing RFC2047 encoded mail headers.
Ralf Schlatterbeck <rsc@runtux.com>
parents: 4447
diff changeset
87 if first:
c426cb251bc7 Be more tolerant when parsing RFC2047 encoded mail headers.
Ralf Schlatterbeck <rsc@runtux.com>
parents: 4447
diff changeset
88 unencoded = unencoded.lstrip()
c426cb251bc7 Be more tolerant when parsing RFC2047 encoded mail headers.
Ralf Schlatterbeck <rsc@runtux.com>
parents: 4447
diff changeset
89 first = False
c426cb251bc7 Be more tolerant when parsing RFC2047 encoded mail headers.
Ralf Schlatterbeck <rsc@runtux.com>
parents: 4447
diff changeset
90 if unencoded:
c426cb251bc7 Be more tolerant when parsing RFC2047 encoded mail headers.
Ralf Schlatterbeck <rsc@runtux.com>
parents: 4447
diff changeset
91 words.append((unencoded, None, None))
c426cb251bc7 Be more tolerant when parsing RFC2047 encoded mail headers.
Ralf Schlatterbeck <rsc@runtux.com>
parents: 4447
diff changeset
92 if parts:
c426cb251bc7 Be more tolerant when parsing RFC2047 encoded mail headers.
Ralf Schlatterbeck <rsc@runtux.com>
parents: 4447
diff changeset
93 charset = parts.pop(0).lower()
c426cb251bc7 Be more tolerant when parsing RFC2047 encoded mail headers.
Ralf Schlatterbeck <rsc@runtux.com>
parents: 4447
diff changeset
94 encoding = parts.pop(0).lower()
c426cb251bc7 Be more tolerant when parsing RFC2047 encoded mail headers.
Ralf Schlatterbeck <rsc@runtux.com>
parents: 4447
diff changeset
95 encoded = parts.pop(0)
c426cb251bc7 Be more tolerant when parsing RFC2047 encoded mail headers.
Ralf Schlatterbeck <rsc@runtux.com>
parents: 4447
diff changeset
96 words.append((encoded, encoding, charset))
c426cb251bc7 Be more tolerant when parsing RFC2047 encoded mail headers.
Ralf Schlatterbeck <rsc@runtux.com>
parents: 4447
diff changeset
97 # Now loop over words and remove words that consist of whitespace
c426cb251bc7 Be more tolerant when parsing RFC2047 encoded mail headers.
Ralf Schlatterbeck <rsc@runtux.com>
parents: 4447
diff changeset
98 # between two encoded strings.
c426cb251bc7 Be more tolerant when parsing RFC2047 encoded mail headers.
Ralf Schlatterbeck <rsc@runtux.com>
parents: 4447
diff changeset
99 import sys
c426cb251bc7 Be more tolerant when parsing RFC2047 encoded mail headers.
Ralf Schlatterbeck <rsc@runtux.com>
parents: 4447
diff changeset
100 droplist = []
c426cb251bc7 Be more tolerant when parsing RFC2047 encoded mail headers.
Ralf Schlatterbeck <rsc@runtux.com>
parents: 4447
diff changeset
101 for n, w in enumerate(words):
c426cb251bc7 Be more tolerant when parsing RFC2047 encoded mail headers.
Ralf Schlatterbeck <rsc@runtux.com>
parents: 4447
diff changeset
102 if n>1 and w[1] and words[n-2][1] and words[n-1][0].isspace():
c426cb251bc7 Be more tolerant when parsing RFC2047 encoded mail headers.
Ralf Schlatterbeck <rsc@runtux.com>
parents: 4447
diff changeset
103 droplist.append(n-1)
c426cb251bc7 Be more tolerant when parsing RFC2047 encoded mail headers.
Ralf Schlatterbeck <rsc@runtux.com>
parents: 4447
diff changeset
104 for d in reversed(droplist):
c426cb251bc7 Be more tolerant when parsing RFC2047 encoded mail headers.
Ralf Schlatterbeck <rsc@runtux.com>
parents: 4447
diff changeset
105 del words[d]
c426cb251bc7 Be more tolerant when parsing RFC2047 encoded mail headers.
Ralf Schlatterbeck <rsc@runtux.com>
parents: 4447
diff changeset
106
c426cb251bc7 Be more tolerant when parsing RFC2047 encoded mail headers.
Ralf Schlatterbeck <rsc@runtux.com>
parents: 4447
diff changeset
107 # The next step is to decode each encoded word by applying the reverse
c426cb251bc7 Be more tolerant when parsing RFC2047 encoded mail headers.
Ralf Schlatterbeck <rsc@runtux.com>
parents: 4447
diff changeset
108 # base64 or quopri transformation. decoded_words is now a list of the
c426cb251bc7 Be more tolerant when parsing RFC2047 encoded mail headers.
Ralf Schlatterbeck <rsc@runtux.com>
parents: 4447
diff changeset
109 # form (decoded_word, charset).
c426cb251bc7 Be more tolerant when parsing RFC2047 encoded mail headers.
Ralf Schlatterbeck <rsc@runtux.com>
parents: 4447
diff changeset
110 decoded_words = []
c426cb251bc7 Be more tolerant when parsing RFC2047 encoded mail headers.
Ralf Schlatterbeck <rsc@runtux.com>
parents: 4447
diff changeset
111 for encoded_string, encoding, charset in words:
c426cb251bc7 Be more tolerant when parsing RFC2047 encoded mail headers.
Ralf Schlatterbeck <rsc@runtux.com>
parents: 4447
diff changeset
112 if encoding is None:
c426cb251bc7 Be more tolerant when parsing RFC2047 encoded mail headers.
Ralf Schlatterbeck <rsc@runtux.com>
parents: 4447
diff changeset
113 # This is an unencoded word.
c426cb251bc7 Be more tolerant when parsing RFC2047 encoded mail headers.
Ralf Schlatterbeck <rsc@runtux.com>
parents: 4447
diff changeset
114 decoded_words.append((encoded_string, charset))
c426cb251bc7 Be more tolerant when parsing RFC2047 encoded mail headers.
Ralf Schlatterbeck <rsc@runtux.com>
parents: 4447
diff changeset
115 elif encoding == 'q':
c426cb251bc7 Be more tolerant when parsing RFC2047 encoded mail headers.
Ralf Schlatterbeck <rsc@runtux.com>
parents: 4447
diff changeset
116 word = quoprimime.header_decode(encoded_string)
c426cb251bc7 Be more tolerant when parsing RFC2047 encoded mail headers.
Ralf Schlatterbeck <rsc@runtux.com>
parents: 4447
diff changeset
117 decoded_words.append((word, charset))
c426cb251bc7 Be more tolerant when parsing RFC2047 encoded mail headers.
Ralf Schlatterbeck <rsc@runtux.com>
parents: 4447
diff changeset
118 elif encoding == 'b':
c426cb251bc7 Be more tolerant when parsing RFC2047 encoded mail headers.
Ralf Schlatterbeck <rsc@runtux.com>
parents: 4447
diff changeset
119 paderr = len(encoded_string) % 4 # Postel's law: add missing padding
c426cb251bc7 Be more tolerant when parsing RFC2047 encoded mail headers.
Ralf Schlatterbeck <rsc@runtux.com>
parents: 4447
diff changeset
120 if paderr:
c426cb251bc7 Be more tolerant when parsing RFC2047 encoded mail headers.
Ralf Schlatterbeck <rsc@runtux.com>
parents: 4447
diff changeset
121 encoded_string += '==='[:4 - paderr]
c426cb251bc7 Be more tolerant when parsing RFC2047 encoded mail headers.
Ralf Schlatterbeck <rsc@runtux.com>
parents: 4447
diff changeset
122 try:
c426cb251bc7 Be more tolerant when parsing RFC2047 encoded mail headers.
Ralf Schlatterbeck <rsc@runtux.com>
parents: 4447
diff changeset
123 word = base64mime.decode(encoded_string)
c426cb251bc7 Be more tolerant when parsing RFC2047 encoded mail headers.
Ralf Schlatterbeck <rsc@runtux.com>
parents: 4447
diff changeset
124 except binascii.Error:
c426cb251bc7 Be more tolerant when parsing RFC2047 encoded mail headers.
Ralf Schlatterbeck <rsc@runtux.com>
parents: 4447
diff changeset
125 raise HeaderParseError('Base64 decoding error')
c426cb251bc7 Be more tolerant when parsing RFC2047 encoded mail headers.
Ralf Schlatterbeck <rsc@runtux.com>
parents: 4447
diff changeset
126 else:
c426cb251bc7 Be more tolerant when parsing RFC2047 encoded mail headers.
Ralf Schlatterbeck <rsc@runtux.com>
parents: 4447
diff changeset
127 decoded_words.append((word, charset))
c426cb251bc7 Be more tolerant when parsing RFC2047 encoded mail headers.
Ralf Schlatterbeck <rsc@runtux.com>
parents: 4447
diff changeset
128 else:
c426cb251bc7 Be more tolerant when parsing RFC2047 encoded mail headers.
Ralf Schlatterbeck <rsc@runtux.com>
parents: 4447
diff changeset
129 raise AssertionError('Unexpected encoding: ' + encoding)
c426cb251bc7 Be more tolerant when parsing RFC2047 encoded mail headers.
Ralf Schlatterbeck <rsc@runtux.com>
parents: 4447
diff changeset
130 # Now convert all words to bytes and collapse consecutive runs of
c426cb251bc7 Be more tolerant when parsing RFC2047 encoded mail headers.
Ralf Schlatterbeck <rsc@runtux.com>
parents: 4447
diff changeset
131 # similarly encoded words.
c426cb251bc7 Be more tolerant when parsing RFC2047 encoded mail headers.
Ralf Schlatterbeck <rsc@runtux.com>
parents: 4447
diff changeset
132 collapsed = []
c426cb251bc7 Be more tolerant when parsing RFC2047 encoded mail headers.
Ralf Schlatterbeck <rsc@runtux.com>
parents: 4447
diff changeset
133 last_word = last_charset = None
c426cb251bc7 Be more tolerant when parsing RFC2047 encoded mail headers.
Ralf Schlatterbeck <rsc@runtux.com>
parents: 4447
diff changeset
134 for word, charset in decoded_words:
c426cb251bc7 Be more tolerant when parsing RFC2047 encoded mail headers.
Ralf Schlatterbeck <rsc@runtux.com>
parents: 4447
diff changeset
135 if isinstance(word, str):
c426cb251bc7 Be more tolerant when parsing RFC2047 encoded mail headers.
Ralf Schlatterbeck <rsc@runtux.com>
parents: 4447
diff changeset
136 pass
c426cb251bc7 Be more tolerant when parsing RFC2047 encoded mail headers.
Ralf Schlatterbeck <rsc@runtux.com>
parents: 4447
diff changeset
137 if last_word is None:
c426cb251bc7 Be more tolerant when parsing RFC2047 encoded mail headers.
Ralf Schlatterbeck <rsc@runtux.com>
parents: 4447
diff changeset
138 last_word = word
c426cb251bc7 Be more tolerant when parsing RFC2047 encoded mail headers.
Ralf Schlatterbeck <rsc@runtux.com>
parents: 4447
diff changeset
139 last_charset = charset
c426cb251bc7 Be more tolerant when parsing RFC2047 encoded mail headers.
Ralf Schlatterbeck <rsc@runtux.com>
parents: 4447
diff changeset
140 elif charset != last_charset:
c426cb251bc7 Be more tolerant when parsing RFC2047 encoded mail headers.
Ralf Schlatterbeck <rsc@runtux.com>
parents: 4447
diff changeset
141 collapsed.append((last_word, last_charset))
c426cb251bc7 Be more tolerant when parsing RFC2047 encoded mail headers.
Ralf Schlatterbeck <rsc@runtux.com>
parents: 4447
diff changeset
142 last_word = word
c426cb251bc7 Be more tolerant when parsing RFC2047 encoded mail headers.
Ralf Schlatterbeck <rsc@runtux.com>
parents: 4447
diff changeset
143 last_charset = charset
c426cb251bc7 Be more tolerant when parsing RFC2047 encoded mail headers.
Ralf Schlatterbeck <rsc@runtux.com>
parents: 4447
diff changeset
144 elif last_charset is None:
c426cb251bc7 Be more tolerant when parsing RFC2047 encoded mail headers.
Ralf Schlatterbeck <rsc@runtux.com>
parents: 4447
diff changeset
145 last_word += BSPACE + word
c426cb251bc7 Be more tolerant when parsing RFC2047 encoded mail headers.
Ralf Schlatterbeck <rsc@runtux.com>
parents: 4447
diff changeset
146 else:
c426cb251bc7 Be more tolerant when parsing RFC2047 encoded mail headers.
Ralf Schlatterbeck <rsc@runtux.com>
parents: 4447
diff changeset
147 last_word += word
c426cb251bc7 Be more tolerant when parsing RFC2047 encoded mail headers.
Ralf Schlatterbeck <rsc@runtux.com>
parents: 4447
diff changeset
148 collapsed.append((last_word, last_charset))
c426cb251bc7 Be more tolerant when parsing RFC2047 encoded mail headers.
Ralf Schlatterbeck <rsc@runtux.com>
parents: 4447
diff changeset
149 return collapsed
c426cb251bc7 Be more tolerant when parsing RFC2047 encoded mail headers.
Ralf Schlatterbeck <rsc@runtux.com>
parents: 4447
diff changeset
150
Roundup Issue Tracker: http://roundup-tracker.org/