mailfilter/filter-rules.py

BLOCK_EMAIL = [
    'chirhart@amazon.com',
    'ron.krogel@citrix.com',
    'replieswelcome@duo.com',
    'webinars@duo.com',
    'diversity@oregonstate.edu',
    'jramiro@pagerduty.com',
    'info@snowflake.com',
    'lisa@duo.com',
    'cortana@microsoft.com'
]

BLOCK_KEYWORDS = [
    'email preferences',
    'end these email updates',
    'manage your preferences',
    'modify your preferences',
    'opt-out',
    'opt out',
    'prefer not to receive',
    'prefer not to see',
    'Samsung SDS America',
    'sidekickopen',
    'subscription preferences',
    'survey',
    'unsubscribe',
    'want to receive',
    'webinar',
    'whitepaper',
    'wish to receive',
]

BLOCK_DOMAINS = [
    'customeriomail.com',
    'mailgun.net',
    'norstar.net'
]

BLOCK_ANNOYING = [
    'charitable fund drive',
    'election reminder',
    'food drive',
    'food share',
    'general election',
    'giving tuesday',
    'hardship leave donations needed',
    'register to vote'
    'special election',
    'voter registration'
]

ALLOW = [
    'oregonstate.edu',
    'github.com',
    'duo.com',
    'sns.amazonaws.com',
    'opsgenie.net',
    'notify@teamdynamixapp.com',
    'newsbites@email.sans.org',
    'noreply@box.com',
    'noreply@email.teams.microsoft.com',
    'no-reply@sharepointonline.com',
    'slalom.com',
    'govdelivery.com',
    'linkoregon.org',
    'megan@pdxwit.org',
    'busyconf.com',
    'support@githubsupport.com',
    'microsoft.com'
]

normalized = {}
folder_cache = {}

def filter_message(self, message):
    # normalize message attributes
    normalized_to = [x.address.lower() for x in message.to]
    normalized_from = message.sender.address.lower()

    # filter unactionable coresys emails
    if 'coresys@lists.oregonstate.edu' in normalized_from:
        unactionable = [x.lower() for x in [
            'DWPRODRAW Verification',
            'BFPDB Reimbursement Job Monitor - OK',
            'DSDB Important Job Monitor - OK'
        ]]
        if is_in_message(unactionable, message):
            self._log_result(message, 'moving to unactionable')
            move_message(message, 'unactionable')
            return

    # filter alerts-sig
    if (normalized_from in ['mcc-b11-stor1@oregonstate.edu',
                            'mcc-b12-stor1@oregonstate.edu',
                            'ousclus@oregonstate.edu',
                            'isilon@storage.sig.oregonstate.edu']
            or 'alarm.DatastoreDiskUsageAlarm' in message.subject):
        self._log_result(message, 'moving to alerts-sig')
        move_message(message, 'alerts-sig')
        return

    # filter conference spam
    if 'brocks+conf@onid.oregonstate.edu' in normalized_to:
        self._log_result(message, 'deleting conference spam')
        message.delete()
        return

    # filter backup-nightly
    if 'backup-nightly@lists.oregonstate.edu' in message.to:
        self._log_result(message, 'moving to backup-nightly')
        move_message(message, 'backup-nightly')
        return

    # delete HP alert spam
    if normalized_from == 'alerts@alerts.mail.hpe.com':
        self._log_result(message, 'deleting HP alert spam')
        message.delete()
        return

    # delete other spam
    if is_in_message(normalized['BLOCK_ANNOYING'], message):
        self._log_result(message, 'junking spam containing annoying content')
        move_message(message, 'Junk Email')
        return

    # keep messages from allowed emails and domains
    for good in ALLOW:
        if good in normalized_from and normalized_from not in BLOCK_EMAIL:
            self._log_result(message,
                             f"keeping message from allowed sender {good}")
            return

    # junk messages from blocked senders
    if normalized_from in BLOCK_EMAIL:
        self._log_result(message, 'junking spam from blocked sender')
        move_message(message, 'Junk Email')
        return

    # junk messages with blocked keywords
    if is_in_message(normalized['BLOCK_KEYWORDS'], message):
        self._log_result(message, 'junking spam containing blocked keyword')
        move_message(message, 'Junk Email')

        return

    # process message headers into a sane data structure
    headers = []
    for header in message.message_headers:
        h = {}
        h[header['name']] = header['value'].lower()
        headers.append(h)

    # junk messages from blocked domains
    is_spam = False
    for domain in BLOCK_DOMAINS:
        if domain in normalized_from:
            is_spam = True
            break
        if search_headers(domain, headers):
            is_spam = True
            break
    if is_spam:
        self._log_result(message, 'junking spam from blocked domain')
        move_message(message, 'Junk Email')
        return

    # junk known spam headers
    if (get_header('X-Spam-Flag' == 'YES', headers)
            or int(get_header('X-MS-Exchange-Organization-SCL', headers)) >= 5):
        self._log_result(message, 'junking spam with known header')
        move_message(message, 'Junk Email')
        return

    # KEEP MESSAGE
    self._log_result(message, 'keeping message, passed all filter checks')

def is_in_message(list_, message):
    """search a message for a list of strings

    Returns True if any string in the list is found in the message
    """
    is_found = False
    message_body = message.body.lower()
    message_subject = message.subject.lower()

    for term in list_:
        if term in message_subject:
            is_found = True
            break
        if term in message_body:
            is_found = True
            break
    return is_found

def move_message(message, folder):
    if folder not in folder_cache:
        return False
    message.move(folder_cache[folder])

def init_filters(self):
    # hack to copy a dict from parent object into local object
    for k,v in self._folders.items():
        folder_cache[k] = v

    normalized['BLOCK_KEYWORDS'] = [x.lower() for x in BLOCK_KEYWORDS]
    normalized['BLOCK_ANNOYING'] = [x.lower() for x in BLOCK_ANNOYING]

def get_header(header_key, headers):
    vals = []
    for header in headers:
        if header_key in header:
            for val in header.values():
                vals.append(val)
    if len(vals) > 1:
        return vals
    elif len(vals) == 1:
        return vals[0]
    else:
        return False

def search_headers(search, headers):
    for header in headers:
        for val in header.values():
            if search in val:
                return True
    return False