X-Git-Url: http://git.polytechnique.org/?a=blobdiff_plain;f=bin%2Fnewsletter.bounces.processor.py;h=89333491eeb26576b48efad3e3bf9226a1c37df9;hb=HEAD;hp=485e855fa7888c3a9ee02da0bfc6b872748b0df8;hpb=4aad6c9c0d06bcaedf17ea3d09e3667e388750cd;p=platal.git diff --git a/bin/newsletter.bounces.processor.py b/bin/newsletter.bounces.processor.py index 485e855..8933349 100755 --- a/bin/newsletter.bounces.processor.py +++ b/bin/newsletter.bounces.processor.py @@ -1,7 +1,7 @@ #!/usr/bin/env python # -*- coding: utf-8 -*- #*************************************************************************** -#* Copyright (C) 2003-2013 Polytechnique.org * +#* Copyright (C) 2003-2014 Polytechnique.org * #* http://opensource.polytechnique.org/ * #* * #* This program is free software; you can redistribute it and/or modify * @@ -123,6 +123,8 @@ def findSubject(message): _recipient_re = re.compile(r'^rfc822; ?(.+)$', re.I | re.U) +# Some MTA set the Final-Recipient with "LOCAL;" instead of "rfc822;" +_recipient_re2 = re.compile(r'^local; ?(.+)$', re.I | re.U) def findAddressInBounce(bounce): @@ -141,6 +143,15 @@ def findAddressInBounce(bounce): print('! Not a valid bounce (expected at least 2 parts, found %d).' % num_payloads) return None status = bounce.get_payload(1) + + # If the second part is of type "message/rfc822" it is the undelivered message. + # Let's try to understand the text part + if status.get_content_type() == 'message/rfc822': + text_bounce = bounce.get_payload(0) + if text_bounce.get_content_type() == 'text/plain': + return findAddressInPlainBounce(text_bounce, bounce) + # If it's not a text message, let's continue to the next error message + if status.get_content_type() != 'message/delivery-status': print('! Not a valid bounce (expected message/delivery-status, found %s).' % status.get_content_type()) return None @@ -154,23 +165,235 @@ def findAddressInBounce(bounce): print('! Not a valid bounce (expected text/plain, found %s).' % content.get_content_type()) return None # Extract the faulty email address - recipient_match = _recipient_re.search(content['Final-Recipient']) - if recipient_match is None: - print('! Missing final recipient.') + # Some MTA don't set Final-Recipient but use Remote-Recipient instead + if 'Final-Recipient' in content: + final_recipient = content['Final-Recipient'] + elif 'Remote-Recipient' in content: + final_recipient = content['Remote-Recipient'] + else: + print('! Not a valid bounce (no Final-Recipient).') return None + recipient_match = _recipient_re.search(final_recipient) + if recipient_match is None: + # Be nice, test another regexp + recipient_match = _recipient_re2.search(final_recipient) + if recipient_match is None: + print('! Missing final recipient.') + return None email = recipient_match.group(1) # Check the action field + if content['Action'].lower().strip() != 'failed': + print('! Not a failed action (%s).' % content['Action']) + return None + + status = content['Status'] + diag_code = content['Diagnostic-Code'] + + # Permanent failure state + if int(status[:1]) == 5: + return email + + # Mail forwarding loops, DNS errors and connection timeouts cause X-Postfix errors + if diag_code is not None and diag_code.startswith('X-Postfix'): + return email + + failure_hints = [ + "insufficient system storage", + "mailbox full", + "mailbox recipient does not have a mailbox database", + "over quota", + "requested action aborted: local error in processing", + "user unknown", + ] + if 'quota' in status.lower(): + return email + if diag_code is not None: + ldiag_code = diag_code.lower() + if any(hint in ldiag_code for hint in failure_hints): + return email + + print('! Not a permanent failure status (%s).' % status) + if diag_code is not None: + print('! Diagnostic code was: %s' % diag_code) + return None + + +def findAddressInWeirdDeliveryStatus(message): + """Finds the faulty email address in the delivery-status part of an email + + Unlikely to findAddressInBounce, the status does NOT follow RFC 1894, so + try to learn to get data nevertheless... + Returns None or the email address. + """ + if message.get_content_type() != 'message/delivery-status': + print('! Not a valid weird bounce (expected message/delivery-status, found %s).' % message.get_content_type()) + return None + # The per-message-fields don't matter here, get only the per-recipient-fields + num_payloads = len(message.get_payload()) + if num_payloads < 2: + print('! Not a valid weird bounce (expected at least 2 parts, found %d).' % num_payloads) + return None + content = message.get_payload(1) + # The content may be missing, but interesting headers still present in the first payload... + if not content: + content = message.get_payload(0) + if 'Action' not in content: + print('! Not a valid weird bounce (unable to find content).') + return None + elif content.get_content_type() != 'text/plain': + print('! Not a valid weird bounce (expected text/plain, found %s).' % content.get_content_type()) + return None + + # Extract the faulty email address + if 'Final-Recipient' in content: + recipient_match = _recipient_re.search(content['Final-Recipient']) + if recipient_match is None: + # Be nice, test another regexp + recipient_match = _recipient_re2.search(content['Final-Recipient']) + if recipient_match is None: + print('! Unknown final recipient in weird bounce.') + return None + email = recipient_match.group(1) + elif 'Original-Recipient' in content: + recipient = content['Original-Recipient'] + recipient_match = _recipient_re.search(recipient) + if recipient_match is None: + # Be nice, test another regexp + recipient_match = _recipient_re2.search(recipient) + if recipient_match is None: + recipient_match = re.match(r'<([^>]+@[^@>]+)>', recipient) + if recipient_match is None: + print('! Unknown original recipient in weird bounce.') + return None + email = recipient_match.group(1) + else: + print('! Missing recipient in weird bounce.') + return None + + # Check the action field if content['Action'].lower() != 'failed': print('! Not a failed action (%s).' % content['Action']) return None + + status = content['Status'] + diag_code = content['Diagnostic-Code'] + + # Permanent failure state + if status and int(status[:1]) == 5: + return email + # Mail forwarding loops, DNS errors and connection timeouts cause X-Postfix errors - # Otherwise, the first sub-field should indicate a permanent failure - postfix_error = content['Diagnostic-Code'] is not None \ - and content['Diagnostic-Code'].startswith('X-Postfix') - if not postfix_error and int(content['Status'][:1]) != 5: - print('! Not a permanent failure status (%s).' % content['Status']) + if diag_code is not None and diag_code.startswith('X-Postfix'): + return email + + failure_hints = [ + "insufficient system storage", + "mailbox full", + "requested action aborted: local error in processing", + "sender address rejected", + "user unknown", + ] + if status and 'quota' in status.lower(): + return email + if diag_code is not None: + ldiag_code = diag_code.lower() + if any(hint in ldiag_code for hint in failure_hints): + return email + + print('! Not a permanent failure status (%s).' % status) + if diag_code is not None: + print('! Diagnostic code was: %s' % diag_code) + return None + + +def findAddressInPlainBounce(bounce, real_bounce=None): + """Finds the faulty email address in a non-RFC-1894 bounced email + """ + # real_bounce is the full email and bounce only the text/plain part, if email have several MIME parts + real_bounce = real_bounce or bounce + lower_from = real_bounce['From'].lower() + if 'mailer-daemon@' not in lower_from and 'postmaster' not in lower_from: + print('! Not a valid plain bounce (expected from MAILER-DAEMON or postmaster, found %s).' % bounce['From']) + return None + if bounce.get_content_type() != 'text/plain': + print('! Not a valid plain bounce (expected text/plain, found %s).' % bounce.get_content_type()) + return None + subject = findSubject(real_bounce).lower() + known_subjects = [ + "delivery status notification (failure)", + "failure notice", + "mail delivery failure", + "returned mail: see transcript for details", + "undeliverable message", + "undelivered mail returned to sender", + ] + if subject not in known_subjects and not subject.startswith('mail delivery failed'): + print('! Not a valid plain bounce (unknown subject: %s).' % subject) + return None + + # Read the 15 first lines of content and find some relevant keywords to validate the bounce + lines = bounce.get_payload().splitlines()[:15] + + # ALTOSPAM is a service which requires to click on a link when sending an email + # Don't consider the "554 5.0.0 Service unavailable" returned by ALTOSPAM as a failure + # but put this message in the dsn-temp mailbox so that it can be processed by hand. + if any("ALTOSPAM which is used by the person" in line for line in lines): + print('! ALTOSPAM has been detected. Moving this message to the dsn-temp mbox') + return None + + # Match: + # A message that you sent could not be delivered to one or more of its recipients. + # I'm afraid I wasn't able to deliver your message to the following addresses. + # The following message to was undeliverable. + non_delivery_hints = [ + "could not be delivered to", + "Delivery to the following recipient failed permanently", + "I'm sorry to have to inform you that your message could not", + "I wasn't able to deliver your message", + "try to send your message again at a later time", + "User unknown in local recipient table", + "> was undeliverable.", + "we were unable to deliver your message", + ] + if not any(any(hint in line for hint in non_delivery_hints) for line in lines): + print('! Unknown mailer-daemon message, unable to find an hint for non-delivery in message:') + print('\n'.join(lines)) return None - return email + + # Match: + # This is a permanent error; I've given up. Sorry it didn't work out. + # 5.1.0 - Unknown address error 550-'email@example.com... No such user' + permanent_error_hints = [ + "Delivery to the following recipient failed permanently", + "failed due to an unavailable mailbox", + "following addresses had permanent fatal errors", + "I'm sorry to have to inform you that your message could not", + "The email account that you tried to reach does not exist", + "This is a permanent error", + "Unknown address error", + "unreachable for too long", + "550 Requested action not taken", + ] + if not any(any(hint in line for hint in permanent_error_hints) for line in lines): + print('! Unknown mailer-daemon message, unable to find an hint for permanent error in message:') + print('\n'.join(lines)) + return None + + # Retrieve the first occurence of + for line in lines: + match = re.match(r'.*?<([0-9a-zA-Z_.-]+@[0-9a-zA-Z_.-]+)>', line) + if match is None: + match = re.match(r'^\s*"?([0-9a-zA-Z_.-]+@[0-9a-zA-Z_.-]+)"?\s*$', line) + if match is not None: + email = match.group(1) + if email.endswith('@polytechnique.org'): + # First valid mail is something like , so we missed the real one + break + return email + + print('! Unknown mailer-daemon message, unable to find email address:') + print('\n'.join(lines)) + return None #----------------------------------------------------------------------------# @@ -190,14 +413,14 @@ class DirectBouncesFilter(MboxFilter): self.seen += 1 # Special case: ignore mailman notifications for the mailing-list # on which the NL is forwarded - if message['From'] == 'newsletter-externes-bounces@polytechnique.org': + if message['From'] == 'newsletter-externes-owner@polytechnique.org': print('! Dropping a notification from mailman for newsletter-externes@polytechnique.org, this should be OK.') self.seen -= 1 return True # Additionnal checks, just to be sure elif message['From'] != 'MAILER-DAEMON@polytechnique.org (Mail Delivery System)' \ or message['Subject'] != 'Undelivered Mail Returned to Sender': - print('! Not an usual direct bounce (From="%s", Subject="%s").' % (message['From'], message['Subject'])) + print('! Not an usual direct bounce (From=%r, Subject=%r).' % (message['From'], message['Subject'])) else: email = findAddressInBounce(message) if email is not None: @@ -300,15 +523,20 @@ class OutOfOfficeFilter(MboxFilter): subject_re = [ r'^Absen(t|ce)', r'^(AUTO: )?Out of (the )?office', + r'^Auto( ?): ', + r'^AutoRe( ?):', r'^Automatic reply: ', r'automatique d\'absence', r'AutoReply', r'(est|is) absent', + r'^En dehors du bureau', r'I am out of town', r'I am currently away', - r'is out of (the )?office', + r'(am|is) out of (the )?office', r'Notification d\'absence', + r'^Out of email reach', r'R.{1,2}ponse automatique( :)?', # There may be encoding error of e acute + r'^Respuesta de Estoy ausente:', ] self.subject_regexes = [re.compile(sre, re.I | re.U) for sre in subject_re] @@ -317,6 +545,21 @@ class OutOfOfficeFilter(MboxFilter): if subject is not None and any(regex.search(subject) for regex in self.subject_regexes): self.mbox.add(message) return True + + # Some systems reply with "Re: ". Be smart here! + if subject is not None and subject.startswith('Re: '): + # Delivered-To: Autoresponder + if 'Autoresponder' in message.get_all('Delivered-To'): + self.mbox.add(message) + return True + # Parse content if it is simple enough + if message.get_content_type() == 'text/plain': + firstline = message.get_payload().splitlines()[0].lower() + if (' absent du bureau ' in firstline + or ' away from my office ' in firstline): + self.mbox.add(message) + return True + return False def finalize(self): @@ -339,15 +582,45 @@ class DeliveryStatusNotificationFilter(MboxFilter): self.mbox_temp.clear() def process(self, message): - if message.get_content_type() == 'multipart/report': - email = findAddressInBounce(message) + # Don't modify message variable for "self.mbox.add(message)" + report_message = message + # Find real report inside attachment + if message.get_content_type() == 'multipart/mixed': + # Some MTA confuse multipart/mixed with multipart/report + # Let's try to find a report! + if len(message.get_payload()) >= 2: + try_status = message.get_payload(1) + if try_status.get_content_type() == 'message/delivery-status': + # The world would be a nice place if delivery-status were + # formatted as expected... + email = findAddressInWeirdDeliveryStatus(try_status) + if email is not None: + self.emails.append(email) + self.mbox.add(message) + return True + try_status = None + report_message = message.get_payload(0) + + # Process report if its type is correct + if report_message.get_content_type() == 'multipart/report': + email = findAddressInBounce(report_message) if email is not None: self.emails.append(email) self.mbox.add(message) - return True else: print("! => Moved to temporary DSN mailbox") self.mbox_temp.add(message) + return True + + # Detect ill-formatted reports, sent as plain text email + if report_message.get_content_type() == 'text/plain' and ( + 'MAILER-DAEMON@' in message.get('From', '').upper() or + 'mail delivery failure' == message.get('Subject', '').lower() + ): + email = findAddressInPlainBounce(report_message) + if email is not None: + self.emails.append(email) + self.mbox.add(message) return True return False