2 # -*- coding: utf-8 -*-
3 #***************************************************************************
4 #* Copyright (C) 2003-2014 Polytechnique.org *
5 #* http://opensource.polytechnique.org/ *
7 #* This program is free software; you can redistribute it and/or modify *
8 #* it under the terms of the GNU General Public License as published by *
9 #* the Free Software Foundation; either version 2 of the License, or *
10 #* (at your option) any later version. *
12 #* This program is distributed in the hope that it will be useful, *
13 #* but WITHOUT ANY WARRANTY; without even the implied warranty of *
14 #* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the *
15 #* GNU General Public License for more details. *
17 #* You should have received a copy of the GNU General Public License *
18 #* along with this program; if not, write to the Free Software *
19 #* Foundation, Inc., *
20 #* 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA *
21 #***************************************************************************
24 Process as automatically as possible bounces from the newsletter
26 The goal is to extract the email adresses that actually bounced.
27 Bounces conforming to RFC 1894 will be automatically processed.
29 This script uses the X-Spam-Flag header to remove spam and heuristics
30 to detect out-of-office auto-replies and delivery status notifications.
32 All emails are saved in different mailboxes to make human post-processing easier.
42 #----------------------------------------------------------------------------#
45 """Applies a series of filters to each message in a mbox."""
47 def __init__(self
, mbox
):
49 self
.mbox
= mailbox
.mbox(self
.mbox_file
)
51 DirectBouncesFilter(),
56 DeliveryStatusNotificationFilter(),
60 def initialize_filters(self
):
61 for f
in self
.filters
: f
.initialize(self
.mbox_file
)
62 self
.start_time
= time
.clock()
64 def apply_filters(self
, message
):
65 return any(f
.process(message
) for f
in self
.filters
)
67 def finalize_filters(self
):
68 duration
= time
.clock() - self
.start_time
71 print('Processed the %d messages of %s in %.2fs' %
(len(self
.mbox
), self
.mbox_file
, duration
))
73 for f
in self
.filters
:
80 self
.initialize_filters()
81 for message
in self
.mbox
: self
.apply_filters(message
)
82 self
.finalize_filters()
87 #----------------------------------------------------------------------------#
90 """Defines an interface for filters."""
92 def initialize(self
, mbox_file
):
93 """Called by the processor before processing starts.
95 This is the place to open descriptors required during processing."""
98 def process(self
, message
):
99 """Called by the processor for each message that reaches this step.
101 Return true to stop processing, and false to go to the next filter."""
105 """Called by the processor after processing ends.
107 This is the place to display the results and close all descriptors."""
110 #----------------------------------------------------------------------------#
112 def findSubject(message
):
113 """Returns the subject of an email.Message as an unicode string."""
114 if message
['Subject'] is None:
117 # decode_header returns a list of (decoded_string, charset) pairs
118 decoded_seq
= email
.header
.decode_header(message
['Subject'])
119 decoded_seq
= [(subj
, enc
or 'utf-8') for subj
, enc
in decoded_seq
]
120 header
= email
.header
.make_header(decoded_seq
)
121 # Be Python 2 & 3 compatible
122 return unicode(header
) if sys
.version_info
< (3,) else str(header
)
125 _recipient_re
= re
.compile(r
'^rfc822; ?(.+)$', re
.I | re
.U
)
126 # Some MTA set the Final-Recipient with "LOCAL;" instead of "rfc822;"
127 _recipient_re2
= re
.compile(r
'^local; ?(.+)$', re
.I | re
.U
)
130 def findAddressInBounce(bounce
):
131 """Finds the faulty email address in a bounced email.
133 See RFC 1894 for more information.
134 Returns None or the email address."""
136 # Check that it is a bounce - a few MTA fail to set this correctly :(
137 if bounce
.get_content_type() != 'multipart/report':
138 print('! Not a valid bounce (expected multipart/report, found %s).' % bounce
.get_content_type())
140 # Extract the second component of the multipart/report
141 num_payloads
= len(bounce
.get_payload())
143 print('! Not a valid bounce (expected at least 2 parts, found %d).' % num_payloads
)
145 status
= bounce
.get_payload(1)
147 # If the second part is of type "message/rfc822" it is the undelivered message.
148 # Let's try to understand the text part
149 if status
.get_content_type() == 'message/rfc822':
150 text_bounce
= bounce
.get_payload(0)
151 if text_bounce
.get_content_type() == 'text/plain':
152 return findAddressInPlainBounce(text_bounce
, bounce
)
153 # If it's not a text message, let's continue to the next error message
155 if status
.get_content_type() != 'message/delivery-status':
156 print('! Not a valid bounce (expected message/delivery-status, found %s).' % status
.get_content_type())
158 # The per-message-fields don't matter here, get only the per-recipient-fields
159 num_payloads
= len(status
.get_payload())
161 print('! Not a valid bounce (expected at least 2 parts, found %d).' % num_payloads
)
163 content
= status
.get_payload(1)
164 if content
.get_content_type() != 'text/plain':
165 print('! Not a valid bounce (expected text/plain, found %s).' % content
.get_content_type())
167 # Extract the faulty email address
168 # Some MTA don't set Final-Recipient but use Remote-Recipient instead
169 if 'Final-Recipient' in content
:
170 final_recipient
= content
['Final-Recipient']
171 elif 'Remote-Recipient' in content
:
172 final_recipient
= content
['Remote-Recipient']
174 print('! Not a valid bounce (no Final-Recipient).')
176 recipient_match
= _recipient_re
.search(final_recipient
)
177 if recipient_match
is None:
178 # Be nice, test another regexp
179 recipient_match
= _recipient_re2
.search(final_recipient
)
180 if recipient_match
is None:
181 print('! Missing final recipient.')
183 email
= recipient_match
.group(1)
184 # Check the action field
185 if content
['Action'].lower().strip() != 'failed':
186 print('! Not a failed action (%s).' % content
['Action'])
189 status
= content
['Status']
190 diag_code
= content
['Diagnostic-Code']
192 # Permanent failure state
193 if int(status
[:1]) == 5:
196 # Mail forwarding loops, DNS errors and connection timeouts cause X-Postfix errors
197 if diag_code
is not None and diag_code
.startswith('X-Postfix'):
201 "insufficient system storage",
203 "requested action aborted: local error in processing",
206 if 'quota' in status
.lower():
208 if diag_code
is not None:
209 ldiag_code
= diag_code
.lower()
210 if any(hint
in ldiag_code
for hint
in failure_hints
):
213 print('! Not a permanent failure status (%s).' % status
)
214 if diag_code
is not None:
215 print('! Diagnostic code was: %s' % diag_code
)
219 def findAddressInWeirdDeliveryStatus(message
):
220 """Finds the faulty email address in the delivery-status part of an email
222 Unlikely to findAddressInBounce, the status does NOT follow RFC 1894, so
223 try to learn to get data nevertheless...
224 Returns None or the email address.
226 if message
.get_content_type() != 'message/delivery-status':
227 print('! Not a valid weird bounce (expected message/delivery-status, found %s).' % message
.get_content_type())
229 # The per-message-fields don't matter here, get only the per-recipient-fields
230 num_payloads
= len(message
.get_payload())
232 print('! Not a valid weird bounce (expected at least 2 parts, found %d).' % num_payloads
)
234 content
= message
.get_payload(1)
235 # The content may be missing, but interesting headers still present in the first payload...
237 content
= message
.get_payload(0)
238 if 'Action' not in content
:
239 print('! Not a valid weird bounce (unable to find content).')
241 elif content
.get_content_type() != 'text/plain':
242 print('! Not a valid weird bounce (expected text/plain, found %s).' % content
.get_content_type())
245 # Extract the faulty email address
246 if 'Final-Recipient' in content
:
247 recipient_match
= _recipient_re
.search(content
['Final-Recipient'])
248 if recipient_match
is None:
249 # Be nice, test another regexp
250 recipient_match
= _recipient_re2
.search(content
['Final-Recipient'])
251 if recipient_match
is None:
252 print('! Unknown final recipient in weird bounce.')
254 email
= recipient_match
.group(1)
255 elif 'Original-Recipient' in content
:
256 recipient
= content
['Original-Recipient']
257 recipient_match
= _recipient_re
.search(recipient
)
258 if recipient_match
is None:
259 # Be nice, test another regexp
260 recipient_match
= _recipient_re2
.search(recipient
)
261 if recipient_match
is None:
262 recipient_match
= re
.match(r
'<([^>]+@[^@>]+)>', recipient
)
263 if recipient_match
is None:
264 print('! Unknown original recipient in weird bounce.')
266 email
= recipient_match
.group(1)
268 print('! Missing recipient in weird bounce.')
271 # Check the action field
272 if content
['Action'].lower() != 'failed':
273 print('! Not a failed action (%s).' % content
['Action'])
276 status
= content
['Status']
277 diag_code
= content
['Diagnostic-Code']
279 # Permanent failure state
280 if status
and int(status
[:1]) == 5:
283 # Mail forwarding loops, DNS errors and connection timeouts cause X-Postfix errors
284 if diag_code
is not None and diag_code
.startswith('X-Postfix'):
288 "insufficient system storage",
290 "requested action aborted: local error in processing",
291 "sender address rejected",
294 if status
and 'quota' in status
.lower():
296 if diag_code
is not None:
297 ldiag_code
= diag_code
.lower()
298 if any(hint
in ldiag_code
for hint
in failure_hints
):
301 print('! Not a permanent failure status (%s).' % status
)
302 if diag_code
is not None:
303 print('! Diagnostic code was: %s' % diag_code
)
307 def findAddressInPlainBounce(bounce
, real_bounce
=None):
308 """Finds the faulty email address in a non-RFC-1894 bounced email
310 # real_bounce is the full email and bounce only the text/plain part, if email have several MIME parts
311 real_bounce
= real_bounce
or bounce
312 lower_from
= real_bounce
['From'].lower()
313 if 'mailer-daemon@' not in lower_from
and 'postmaster' not in lower_from
:
314 print('! Not a valid plain bounce (expected from MAILER-DAEMON or postmaster, found %s).' % bounce
['From'])
316 if bounce
.get_content_type() != 'text/plain':
317 print('! Not a valid plain bounce (expected text/plain, found %s).' % bounce
.get_content_type())
319 subject
= findSubject(real_bounce
).lower()
321 "delivery status notification (failure)",
323 "mail delivery failure",
324 "returned mail: see transcript for details",
325 "undeliverable message",
326 "undelivered mail returned to sender",
328 if subject
not in known_subjects
and not subject
.startswith('mail delivery failed'):
329 print('! Not a valid plain bounce (unknown subject: %s).' % subject
)
332 # Read the 15 first lines of content and find some relevant keywords to validate the bounce
333 lines
= bounce
.get_payload().splitlines()[:15]
335 # ALTOSPAM is a service which requires to click on a link when sending an email
336 # Don't consider the "554 5.0.0 Service unavailable" returned by ALTOSPAM as a failure
337 # but put this message in the dsn-temp mailbox so that it can be processed by hand.
338 if any("ALTOSPAM which is used by the person" in line
for line
in lines
):
339 print('! ALTOSPAM has been detected. Moving this message to the dsn-temp mbox')
343 # A message that you sent could not be delivered to one or more of its recipients.
344 # I'm afraid I wasn't able to deliver your message to the following addresses.
345 # The following message to <email@example.com> was undeliverable.
346 non_delivery_hints
= [
347 "could not be delivered to",
348 "Delivery to the following recipient failed permanently",
349 "I'm sorry to have to inform you that your message could not",
350 "I wasn't able to deliver your message",
351 "try to send your message again at a later time",
352 "> was undeliverable.",
353 "we were unable to deliver your message",
355 if not any(any(hint
in line
for hint
in non_delivery_hints
) for line
in lines
):
356 print('! Unknown mailer-daemon message, unable to find an hint for non-delivery in message:')
357 print('\n'.join(lines
))
361 # This is a permanent error; I've given up. Sorry it didn't work out.
362 # 5.1.0 - Unknown address error 550-'email@example.com... No such user'
363 permanent_error_hints
= [
364 "Delivery to the following recipient failed permanently",
365 "failed due to an unavailable mailbox",
366 "I'm sorry to have to inform you that your message could not",
367 "This is a permanent error",
368 "Unknown address error",
369 "unreachable for too long",
370 "550 Requested action not taken",
372 if not any(any(hint
in line
for hint
in permanent_error_hints
) for line
in lines
):
373 print('! Unknown mailer-daemon message, unable to find an hint for permanent error in message:')
374 print('\n'.join(lines
))
377 # Retrieve the first occurence of <email@example.com>
379 match
= re
.match(r
'.*?<([0-9a-zA-Z_.-]+@[0-9a-zA-Z_.-]+)>', line
)
381 match
= re
.match(r
'^\s*"?([0-9a-zA-Z_.-]+@[0-9a-zA-Z_.-]+)"?\s*$', line
)
382 if match
is not None:
383 email
= match
.group(1)
384 if email
.endswith('@polytechnique.org'):
385 # First valid mail is something like <info_newsletter@polytechnique.org>, so we missed the real one
389 print('! Unknown mailer-daemon message, unable to find email address:')
390 print('\n'.join(lines
))
393 #----------------------------------------------------------------------------#
395 class DirectBouncesFilter(MboxFilter
):
397 def initialize(self
, mbox_file
):
399 self
.bad_problems
= 0
401 self
.mbox_file
= '%s.bounced' % mbox_file
402 self
.mbox
= mailbox
.mbox(self
.mbox_file
)
405 def process(self
, message
):
406 if message
['X-Spam-Flag'] is None:
407 # During finalization, we will verifiy that all messages were processed
409 # Special case: ignore mailman notifications for the mailing-list
410 # on which the NL is forwarded
411 if message
['From'] == 'newsletter-externes-owner@polytechnique.org':
412 print('! Dropping a notification from mailman for newsletter-externes@polytechnique.org, this should be OK.')
415 # Additionnal checks, just to be sure
416 elif message
['From'] != 'MAILER-DAEMON@polytechnique.org (Mail Delivery System)' \
417 or message
['Subject'] != 'Undelivered Mail Returned to Sender':
418 print('! Not an usual direct bounce (From=%r, Subject=%r).' %
(message
['From'], message
['Subject']))
420 email
= findAddressInBounce(message
)
421 if email
is not None:
422 self
.emails
.append(email
)
423 self
.mbox
.add(message
)
426 print('! => No email found in direct bounce, this is really bad.')
427 self
.bad_problems
+= 1
431 print('Found %d messages with no X-Spam-Flag header.' % self
.seen
)
432 print('Found %d of them that are confirmed bounces.' %
len(self
.mbox
))
433 print('They were saved in %s.' % self
.mbox_file
)
434 if self
.bad_problems
:
435 print('Found %d of them that are invalid.' % self
.bad_problems
)
436 if self
.seen
!= len(self
.mbox
) + self
.bad_problems
:
437 print(' /!\ These numbers shoud be equal! We have a problem! /!\\')
439 print('Here is the list of email adresses for these bounces:')
441 for email
in self
.emails
:
446 #----------------------------------------------------------------------------#
448 class SpamFilter(MboxFilter
):
450 def initialize(self
, mbox_file
):
451 self
.mbox_file
= '%s.spam' % mbox_file
452 self
.mbox
= mailbox
.mbox(self
.mbox_file
)
455 def process(self
, message
):
456 if message
['X-Spam-Flag'] is not None \
457 and message
['X-Spam-Flag'].startswith('Yes, tests=bogofilter'):
458 self
.mbox
.add(message
)
463 print('Found %d spams. This is reliable.' %
len(self
.mbox
))
464 print('They were saved in %s.' % self
.mbox_file
)
465 print('You might check the contents of this mbox.')
468 #----------------------------------------------------------------------------#
470 class UnsureFilter(MboxFilter
):
472 def initialize(self
, mbox_file
):
473 self
.mbox_file
= '%s.unsure' % mbox_file
474 self
.mbox
= mailbox
.mbox(self
.mbox_file
)
477 def process(self
, message
):
478 if message
['X-Spam-Flag'] is not None \
479 and message
['X-Spam-Flag'].startswith('Unsure, tests=bogofilter'):
480 self
.mbox
.add(message
)
485 print('Found %d unclassified messages. Most of them should be spams.' %
len(self
.mbox
))
486 print('They were saved in %s.' % self
.mbox_file
)
487 print('You must check the contents of this mbox and feed the antispam.')
490 #----------------------------------------------------------------------------#
492 class CheckNonSpamFilter(MboxFilter
):
494 def initialize(self
, mbox_file
):
497 def process(self
, message
):
498 if message
['X-Spam-Flag'] is None \
499 or not message
['X-Spam-Flag'].startswith('No, tests=bogofilter'):
505 print('Encountered %d messages that were neither spam, nor unsure, nor non-spams.' % self
.seen
)
506 print('Please investigate.')
508 print('All messages were either spam, or unsure, or non-spams. Good.')
510 #----------------------------------------------------------------------------#
512 class OutOfOfficeFilter(MboxFilter
):
514 def initialize(self
, mbox_file
):
515 self
.mbox_file
= '%s.ooo' % mbox_file
516 self
.mbox
= mailbox
.mbox(self
.mbox_file
)
520 r
'^(AUTO: )?Out of (the )?office',
523 r
'^Automatic reply: ',
524 r
'automatique d\'absence
',
527 r'^En dehors du bureau
',
529 r'I am currently away
',
530 r'(am|
is) out
of (the
)?office
',
531 r'Notification d
\'absence
',
532 r'^Out of email reach
',
533 r'R
.{1,2}ponse
automatique( :)?
', # There may be encoding error of e acute
534 r'^Respuesta de Estoy ausente
:',
536 self.subject_regexes = [re.compile(sre, re.I | re.U) for sre in subject_re]
538 def process(self, message):
539 subject = findSubject(message)
540 if subject is not None and any(regex.search(subject) for regex in self.subject_regexes):
541 self.mbox.add(message)
544 # Some systems reply with "Re: ". Be smart here!
545 if subject is not None and subject.startswith('Re
: '):
546 # Delivered-To: Autoresponder
547 if 'Autoresponder
' in message.get_all('Delivered
-To
'):
548 self.mbox.add(message)
550 # Parse content if it is simple enough
551 if message.get_content_type() == 'text
/plain
':
552 firstline = message.get_payload().splitlines()[0].lower()
553 if (' absent du bureau
' in firstline
554 or ' away
from my office
' in firstline):
555 self.mbox.add(message)
561 print('Found %d
"out of office". This
is generally reliable
.' % len(self.mbox))
562 print('They were saved
in %s
.' % self.mbox_file)
563 print('You may check the contents of this mbox
.')
566 #----------------------------------------------------------------------------#
568 class DeliveryStatusNotificationFilter(MboxFilter):
570 def initialize(self, mbox_file):
572 self.mbox_file = '%s
.dsn
' % mbox_file
573 self.mbox = mailbox.mbox(self.mbox_file)
575 self.mbox_temp_file = '%s
.dsn
-temp
' % mbox_file
576 self.mbox_temp = mailbox.mbox(self.mbox_temp_file)
577 self.mbox_temp.clear()
579 def process(self, message):
580 # Don't modify message variable
for "self.mbox.add(message)"
581 report_message
= message
582 # Find real report inside attachment
583 if message
.get_content_type() == 'multipart/mixed':
584 # Some MTA confuse multipart/mixed with multipart/report
585 # Let's try to find a report!
586 if len(message
.get_payload()) >= 2:
587 try_status
= message
.get_payload(1)
588 if try_status
.get_content_type() == 'message/delivery-status':
589 # The world would be a nice place if delivery-status were
590 # formatted as expected...
591 email
= findAddressInWeirdDeliveryStatus(try_status
)
592 if email
is not None:
593 self
.emails
.append(email
)
594 self
.mbox
.add(message
)
597 report_message
= message
.get_payload(0)
599 # Process report if its type is correct
600 if report_message
.get_content_type() == 'multipart/report':
601 email
= findAddressInBounce(report_message
)
602 if email
is not None:
603 self
.emails
.append(email
)
604 self
.mbox
.add(message
)
606 print("! => Moved to temporary DSN mailbox")
607 self
.mbox_temp
.add(message
)
610 # Detect ill-formatted reports, sent as plain text email
611 if report_message
.get_content_type() == 'text/plain' and (
612 'MAILER-DAEMON@' in message
.get('From', '').upper() or
613 'mail delivery failure' == message
.get('Subject', '').lower()
615 email
= findAddressInPlainBounce(report_message
)
616 if email
is not None:
617 self
.emails
.append(email
)
618 self
.mbox
.add(message
)
623 print('Found %d delivery status notifications. This is generally reliable.' %
len(self
.mbox
))
624 print('They were saved in %s.' % self
.mbox_file
)
626 print('Here is the list of email adresses for these bounces:')
628 for email
in self
.emails
:
632 print('Found %d temporary and invalid delivery status notifications.' %
len(self
.mbox_temp
))
633 print('They were saved in %s.' % self
.mbox_temp_file
)
634 self
.mbox_temp
.close()
636 #----------------------------------------------------------------------------#
638 class CatchAllFilter(MboxFilter
):
640 def initialize(self
, mbox_file
):
641 self
.mbox_file
= '%s.catchall' % mbox_file
642 self
.mbox
= mailbox
.mbox(self
.mbox_file
)
645 def process(self
, message
):
646 self
.mbox
.add(message
)
650 if len(self
.mbox
) > 0:
651 print('%d messages reached the catchall.' %
len(self
.mbox
))
652 print('They were saved in %s.' % self
.mbox_file
)
653 print('You must process the contents of this mbox manually.')
656 print('No messages reached the catchall. Nice.')
658 os
.unlink(self
.mbox_file
)
660 #----------------------------------------------------------------------------#
662 if __name__
== '__main__':
664 if len(sys
.argv
) != 2:
665 print('Usage: %s mbox' % sys
.argv
[0])
668 if not os
.path
.exists(sys
.argv
[1]):
669 print('No such file: %s' % sys
.argv
[1])
672 processor
= MboxProcessor(sys
.argv
[1])