1 # This Python file uses the following encoding: utf-8
3 # The software in this package is distributed under the GNU General
4 # Public License version 2 (with a special exception described below).
6 # A copy of GNU General Public License (GPL) is included in this distribution,
7 # in the file COPYING.GPL.
9 # As a special exception, if other files instantiate templates or use macros
10 # or inline functions from this file, or you compile this file and link it
11 # with other works to produce a work based on this file, this file
12 # does not by itself cause the resulting work to be covered
13 # by the GNU General Public License.
15 # However the source code for this file must still be made available
16 # in accordance with section (3) of the GNU General Public License.
18 # This exception does not invalidate any other reasons why a work based
19 # on this file might be covered by the GNU General Public License.
21 # Copyright (c) 2016-2018 Intra2net AG <info@intra2net.com>
26 ------------------------------------------------------
27 Utilities for dealing with email
29 .. seealso:: :py:mod:`pyi2ncommon.mail_validator`,
30 :py:mod:`pyi2ncommon.imap_mailbox`
32 Copyright: Intra2net AG
36 ------------------------------------------------------
40 from base64 import b64decode
41 from email.utils import parsedate_to_datetime
42 from email.parser import BytesParser
43 from email import policy
45 # outsourced source, import required for compatiblity
46 from .imap_mailbox import ImapMailbox # pylint: disable=unused-import
47 from .mail_validator import * # pylint: disable=unused-import
48 from .sysmisc import replace_file_regex
50 log = logging.getLogger('pyi2ncommon.mail_utils')
53 def prep_email_header(email_file, value, regex=None, criterion="envelopeto"):
55 Replace value in a provided email file.
57 :param str email_file: file to use for the replacement
58 :param str value: value to replace the first matched group with
59 :param regex: regular expression to use when replacing a header value
60 :type regex: str or None
61 :param str criterion: criterion to use for replacement, one
62 of 'envelopeto' or 'received'
63 :raises: :py:class:`ValueError` if the choice of criterion is invalid
65 ..todo:: In some cases this function is reusing arnied wrapper's cnf
66 value preparation but for email headers.
68 if criterion == "envelopeto":
69 logging.debug("Updating test emails' EnvelopeTo header")
70 replace_file_regex(email_file, value, regex=regex)
71 elif criterion == "received":
72 logging.debug("Updating test emails' Received header")
73 with open(email_file, "r") as file_handle:
74 email_text = file_handle.read()
75 email_text = re.sub(regex, value, email_text)
76 email_text = re.sub(regex, value, email_text)
77 with open(email_file, "w") as file_handle:
78 file_handle.write(email_text)
80 raise ValueError("Invalid header preparation criterion '%s'"
84 def parse_mail_file(file_name, headers_only=True, attachment_filenames=False,
85 raise_on_defect=False, new_message_type=False):
87 Parse given email file (e.g. a banned message).
89 This is basically a `email.parser.BytesParser().parse(...)` with given
90 `headers_only` and policy selection, that can also handle BSMTP. As an
91 extra bonus, you can just request headers plus the names of attached files.
93 Removes the SMTP envelope surrounding the email if present. Only left-over
94 might be a line with a '.' at end of non-multipart messages if
95 `headers_only` is False.
97 :param str file_name: path to the file that contains the email text
98 :param bool headers_only: whether to parse only the email headers; set this
99 to False, e.g. if you want to check for
100 attachments using message.walk()
101 :param bool attachment_filenames: if you just want headers and names of
102 attached files, set `headers_only` and
104 :param bool raise_on_defect: whether to raise an error if email parser
105 encounters a defect (email policy `strict`) or
106 just add the defect to message's `defect`
108 :param bool new_message_type: whether to return the older
109 :py:class:`email.message.Message` (policy
110 `compat32`, our default), or the newer
111 :py:class:`email.message.EmailMessage` type
112 (policy `default`). Big difference!
113 :returns: either msg or 2-tuple `(msg, filenames)` if requested per arg
114 `attachment_filenames`
115 :rtype: :py:class:`email.message.Message` or
116 (:py:class:`email.message.Message`, (str)) or
117 one of these two with :py:class:`email.message.EmailMessage`
123 mail_policy = policy.default
125 mail_policy = policy.compat32
127 mail_policy += policy.strict
129 with open(file_name, 'rb') as read_handle:
130 line = read_handle.readline()
131 if line.startswith(b'EHLO'):
132 # there is a smtp header. skip to its end
133 while line.strip() != b'DATA':
134 line = read_handle.readline()
135 # the rest is the email plus a trailing '.' (ignored by parser if
138 read_handle.seek(0) # forget we read the first line already
139 start_pos = read_handle.tell()
140 msg = BytesParser(policy=mail_policy).parse(read_handle,
141 headersonly=headers_only)
143 if not attachment_filenames:
146 # otherwise need to parse complete message to get attachment file names
148 with open(file_name, 'rb') as read_handle:
149 read_handle.seek(start_pos)
150 full_msg = BytesParser(policy=mail_policy).parse(read_handle,
154 filenames = [get_filename(part) for part in full_msg.walk()]
155 return msg, tuple(filename for filename in filenames
156 if filename is not None)
159 def parse_mail_date(message):
161 Parse the 'Date' header of the given message.
163 Shortcut for :py:func:`email.utils.parsedate_to_datetime`.
165 This is no longer necessary for newer
166 :py:class:`email.message.EmailMessage` since the `Date` Header is
167 automatically parsed to a :py:class:`email.headerregistry.DateHeader`.
169 :param message: Email message
170 :type message: :py:class:`email.message.Message`
171 :returns: datetime from Email "Date" header or None if header not present
172 :rtype: :py:class:`datetime.datetime` or None
174 date_str = message.get('Date', '')
177 return parsedate_to_datetime(date_str)
180 def get_user_mail_files(user, mailbox='INBOX'):
182 Iterate over mails in given folder of given user; yields file names.
184 Works on local cyrus file system, not on imap server.
186 :param str user: Name of user whose mailbox is analyzed
187 :param str mailbox: name of mailbox to use, INBOX (default) for base
188 folder; name is modified using :py:func:`cyrus_escape`
189 :returns: nothing; but yields full path to messages on disc
191 # base folder of user mail
192 folder = os.path.join('/datastore', 'imap-mails', 'user', user)
194 # adapt paths like "INBOX/sub/dir" to "sub/dir"
195 subdirs = mailbox.split('/')
196 if subdirs[0].upper() == 'INBOX':
197 subdirs = subdirs[1:]
198 folder = os.path.join(folder,
199 *(cyrus_escape(subdir) for subdir in subdirs))
201 for filename in os.listdir(folder):
202 if not re.match(r'\d+\.', filename):
204 full_path = os.path.join(folder, filename)
208 def get_user_mail(user, mailbox='INBOX', **kwargs):
210 Iterate over mails in given folder of given user; yields parsed mails.
212 :param str user: see :py:func:`get_user_mail_files`
213 :param str mailbox: see :py:func:`get_user_mail_files`
214 :param dict kwargs: all other args are forwarded to
215 :py:func:`parse_mail_file`
216 :returns: nothing; but yields 2-tuples (path, email_msg) where first is the
217 full path to the message on disc, and the latter is the outcome
218 of :py:func:`parse_mail_file` for that file
220 for full_path in get_user_mail_files(user, mailbox):
221 yield full_path, parse_mail_file(full_path, **kwargs)
224 def get_message_text(filename, fallback_encoding='iso8859-1',
225 include_all_text=False):
227 Extract message text as string from email message.
229 Intended as complementary addition to get_user_mail, e.g. ::
231 for filename, msg in get_user_mail(user):
232 # rough filtering based on headers
233 if msg['Subject'] != 'Expected Subject':
235 # get message text for closer inspection
236 text = get_message_text(filename)
237 if 'Expected Text' not in text:
241 Finds the first part in message that is of type `text/plain` and decodes it
242 using encoding specified in mail or otherwise fallback encoding. If none
243 found takes first part of type `text/*`, or otherwise just the first part.
245 If include_all_text is True, all `text/*` parts are included, with `text/plain`
248 :param str filename: complete path of message file in filesystem
249 :param str fallback_encoding: Encoding of email text if none is specified
251 :param bool include_all_text: include all `text/*` parts in returned text
252 :returns: text(s) of message
253 :rtype: [str] if include_all_text else str
256 msg = parse_mail_file(filename, headers_only=False)
257 for part in msg.walk():
258 if part.get_content_type() != 'text/plain':
260 encoding = part.get_content_charset(fallback_encoding)
261 result.append(part.get_payload(decode=True).decode(encoding))
263 if result and not include_all_text:
266 # no text/plain found. Try only "text/":
267 for part in msg.walk():
268 cont_type = part.get_content_type()
269 if cont_type.startswith('text/') and cont_type != 'text/plain':
270 encoding = part.get_content_charset(fallback_encoding)
271 result.append(part.get_payload(decode=True).decode(encoding))
274 if not include_all_text:
278 # no "text/" found. Just take first part
279 while msg.is_multipart():
280 msg = msg.get_payload(0)
282 encoding = msg.get_content_charset(fallback_encoding)
284 return [msg.get_payload(decode=True).decode(encoding), ]
285 return msg.get_payload(decode=True).decode(encoding)
288 def cyrus_escape(user_or_folder, keep_path=False, regex=False):
290 Convert names of users or mailbox folders to cyrus format.
292 quite a hack, just does the following hard-coded replacements:
295 * / --> . (except if keep_path is True)
296 * "u --> &APw- , "o --> &APY- , "a --> &AOQ-
297 (if need more: this is modified utf-7)
298 * inbox --> (the empty string)
300 Would like to use a general modified utf-7-encoder/decoder but python has
301 none builtin (see https://bugs.python.org/issue5305) and an extra lib like
302 https://bitbucket.org/mjs0/imapclient/ would be overkill. After all, we
303 control the input to this function via params and this is enough umlaut-
306 :param str user_or_folder: name of the user or folder string to escape
307 :param bool keep_path: do not replace '/' with '.' so can still use result
309 :param bool regex: result is used in grep or other regex, so ^, . and & are
310 escaped again with a backslash
311 :returns: escaped user or folder string
314 .. seealso:: :py:func:`cyrus_unescape`
316 temp = user_or_folder.replace('.', '^') \
317 .replace('ü', '&APw-').replace('ä', '&AOQ-') \
318 .replace('ö', '&APY-') \
319 .replace('inbox', '').replace('INBOX', '').replace('Inbox', '')
321 temp = temp.replace('/', '.')
323 return temp.replace('^', r'\^').replace('&', r'\&') \
324 .replace('.', r'\.').replace('$', r'\$')
328 def cyrus_unescape(user_or_folder):
330 Undo effects of :py:func:`cyrus_escape` (but not all of them).
332 :param str user_or_folder: name of the user or folder string to unescape
333 :returns: unescaped user or folder string
336 if user_or_folder == '':
338 return user_or_folder.replace('.', '/')\
339 .replace(r'\^', '.').replace('^', '.')
342 def get_filename(message, failobj=None, do_unwrap=True):
344 Get filename of a message part, even if it is base64-encoded.
346 For attachments with base64-encoded file name, the
347 :py:func:`email.message.Message.get_filename()` does not work. This
348 function tries that first and if it fails tries to interprete the
349 Content-Disposition of the message part. If all fails, returns `failobj`.
351 Only for ascii filenames: also unwraps file names if they are line-wrapped.
352 But note that this may remove too much whitespace from the filename if
353 line-wrapping happened in the same position as the filename's whitespace.
354 To get unwrapped version, set param `do_unwrap` to `False`.
356 See also: https://en.wikipedia.org/wiki/MIME#Encoded-Word
358 :param message: message part, e.g. from
359 :py:meth:`email.message.Message.walk`
360 :type message: :py:class:`email.message.Message` or
361 :py:class:`email.message.EmailMessage`
362 :param failobj: object to return in case of failure (defaults to None)
363 :param bool do_unwrap: undo line-break inserted by mail-creator; may remove
364 whitespace from file name; only applies to ascii
366 :returns: either a string or failobj
368 # try the old way and unwrap
369 filename = message.get_filename(failobj)
371 if isinstance(filename, bytes) and not filename.startswith(b'=?') \
372 and not filename.endswith(b'?='):
373 filename = filename.decode('utf8')
375 if isinstance(filename, str):
377 return re.sub('[\\r\\n]+', '', filename)
380 if 'Content-Disposition' not in message:
383 # try parsing content-disposition. e.g.:
384 # attachment; filename="2018年度公开课计划表.xlsx" -->
385 # '=?utf-8?b?YXR0YWNobWVudDsgZmlsZW5hbWU9IjIwMTjlubTluqY=?=\r\n =?utf-8?b?'
386 # '5YWs5byA6K++6K6h5YiS6KGoLnhsc3gi?='
388 # This may be a re-implementation of email.utils.collapse_rfc2231_value()
389 # as mentioned in email.message.EmailMessage.get_param()
391 # The form is: "=?charset?encoding?encoded text?="
392 SPLIT_REGEX = '\r?\n *' # should be CRNL but some files miss the \r
393 ENCODED_WORD_REGEX = r'\s*=\?([^?]+)\?([^?]+)\?(.*)\?=\s*$'
394 LINE_REGEX = r'attachment\s*;\s*filename=(")?(.+)\1\s*$'
396 for word in re.split(SPLIT_REGEX, message['Content-Disposition']):
397 match = re.match(ENCODED_WORD_REGEX, word)
400 charset, encoding, data = match.groups()
401 if encoding.lower() == 'b':
402 temp = b64decode(data)
403 elif encoding.lower() == 'q':
404 raise NotImplementedError('use quopri.decodestring, handle _')
406 raise ValueError('not allowed according to wikipedia: "{}"'
408 decoded.append(temp.decode(charset))
409 decoded = u''.join(decoded)
411 match = re.match(LINE_REGEX, decoded)
413 return match.groups()[1]