You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.

2246 lines
116 KiB

#!/usr/bin/env python2
# (Requires Python 2.x, not 3; search for "3.3+" in
# comment below to see how awkward forward-port would be)
"ImapFix v1.752 (c) 2013-22 Silas S. Brown. License: Apache 2"
# Put your configuration into,
# overriding these options:
hostname = "" # or host:port
username = "me"
password = "xxxxxxxx"
# (If you are likely to be editing configuration rules when
# others can see your screen, you might like to store the
# password in a separate file and do, for example,
# password = open(".imapfix-pass").read().strip()
# instead, so that it won't be shown on-screen when you're
# editing other things.)
# You can also use OAuth2 by setting password like this:
# password = ("command to generate oauth2 string", 3500)
# where 3500 is the number of seconds before script is called again.
# (You'll then have to sort out calls to or whatever,
# and if you registered a 'test app' it is likely that refresh
# tokens will not persist more than a week or so and will need
# periodic renewal in a browser; this won't be the case if
# you use credentials from a production application.)
login_retry = False # True = don't stop on login failure
# (useful if your network connection is not always on)
filtered_inbox = "in" # or =None if you don't want to do
# any moving from inbox to filtered_inbox (i.e. no rules)
# but just use maildirs_to_imap or maildir_to_copyself,
# e.g. because you're on an auxiliary machine (which does
# not have your spamprobe database etc) but still want to
# use maildir_to_copyself for mutt (see notes below).
# You can also use environment variables with override
# e.g. import os; filtered_inbox = os.getenv("ImapFolder","in")
# so it can be overridden with ImapFolder for --upload
# If you want to keep your filtered_inbox on local maildir
# instead of IMAP server, set it to a tuple with the first
# item being "maildir": ("maildir","path/to/maildir")
leave_note_in_inbox = True # for "new mail" indicators
change_message_id = False # set this to True for
# providers like Gmail that refuse to accept new
# versions of messages with the same Message-ID. Adds
# an extra digit to Message-ID (if present) when saving
# a changed version of the message to the same imap.
# This may result in a breakage of "In-Reply-To" (unless
# you edit the extra character out on the client side),
# but might be a necessary loss if you're using Gmail.
newmail_directory = None
# or newmail_directory = "/path/to/a/directory"
# - any new mail put into any folder by header_rules will
# result in a file being created in that local directory
# with the same name as the folder
max_size_of_first_part = None
# e.g. max_size_of_first_part = 48*1024
# any messages whose first part is longer than this will be
# converted into attachments. This is for sync'ing in
# bandwidth-limited situations using a device that knows to
# not fetch attachments but doesn't necessarily know to not
# fetch more than a certain amount of text (especially if
# it's only HTML).
image_size = None # or e.g. image_size = (320,240)
# - adds scaled-down versions of any image attachment that
# exceeds this size, for previewing on low bandwidth
# (this option requires the PIL library). The scaled-down
# versions are added only if the file is actually smaller.
# This option also tries to ensure that all images are set to
# an image/ rather than application/ Content-Type, which
# helps some mailers.
office_convert = None # or "html" or "pdf", to use
# LibreOffice/OpenOffice's 'soffice --convert-to' option
# to produce converted versions of office documents
# (beware, I have not checked soffice for vulnerabilities)
# - resulting HTML might not work on old WM phones as they
# don't support data: image URLs.
# This option also enables summary generation of
# Microsoft Calendar attachments.
pdf_convert = False # True = use
# a.pdf # makes a-html.html as.html (ignores extra html argument or -stdout)
# imgs unless -i
# -c complex , or -noframes ?
use_tnef = False # True = use "tnef" command on winmail.dat
header_rules = [
["regexp to check for in header",
"regexp to check for in header",
"regexp to check for in header"]),
["regexp to check for in header"]),
# etc; folder name None = delete the message;
# "inbox" = change to filtered_inbox;
# "spam" = change to spam_folder;
# start with a * if this folder does not need any
# notification in newmail_directory, e.g. "*boxname"
# For saving to local maildir instead of IMAP, set
# folder name to ("maildir","path/to/maildir")
# or to skip newmail_directory notification also,
# set it to ("*","maildir","path/to/maildir")
def extra_rules(message_as_string): return False
# you can override this to any function you want, which
# returns the name of a folder (with or without a *), or
# None to delete the message, or False = no decision,
# or ("maildir","path/to/maildir") or ("*","maildir"...)
catch_extraRules_errors = True
def handle_authenticated_message(subject,firstPart,attach):
return False
# - you can override this to any function you want, which
# does anything you want in response to messages that are
# SSL-authenticated as coming from yourself (see below).
# Returns the name of a folder (or maildir as above), or
# None to delete the message, or False = undecided (normal
# rules will apply, except spamprobe will be bypassed).
# If folder name starts with *, mail will be marked 'seen'.
# firstPart is a UTF-8 copy of the first part of the body
# (which will typically contain plain text even if you
# were using an HTML mailer); subject is also UTF-8 coded.
# attach is a dictionary of filename:contents.
# Note: although we do check for SSL authentication, you
# are advised not to give too much "power" to authenticated
# messages (e.g. don't let them run arbitrary shell commands)
# in case your configuration does end up with a loophole
# or a trusted IMAP server gets compromised. The expected
# use of handle_authenticated_message is to save it in a
# folder it specifies, or add something to a database, or
# trigger a preconfigured build script, etc.
trusted_domain = None # or e.g. "" specifying
# the domain of "our" network whose Received headers we
# can trust for SMTPS authentication (below). You may
# also set it to a list of strings, e.g.
# trusted_domain=["",""]
# in which case you might want to add IP addresses to the
# list if one of your machines fails to do reverse-DNS to
# the other one on its Received header (but domain names
# must also be provided on this list).
# Including a blank ("") entry causes "Received" headers
# that do not sperify a "from" to not interrupt the scan
# for trusted Received headers (some IMAP networks have
# one of those first, but use with caution).
smtps_auth = None # or e.g. " with esmtpsa (LOGIN:me)"
# - if a Received header generated by your trusted domain
# (and listed before any untrusted headers) contains this
# string, the message is considered authentically from you
# (this string should be generated by your network only if
# you really did authenticate over HTTPS). You may also
# set smtps_auth to a list of strings, any one of which is
# acceptable, e.g. smtps_auth=["esmtpsa (LOGIN:me)","esmtpsa (PLAIN:me)"]
# If any item starts "from " then the next word is assumed to
# be a user ID that is trusted if the domain reports it as a
# local delivery (no IP address).
# Note: if trusted_domain and smtps_auth is set, any message
# that does NOT contain any Received headers will be assumed
# to be generated from your own account and therefore treated
# as though it had been authenticated by smtps_auth (which
# might be useful for using --multinote with the real inbox).
# You should therefore ensure that your local network ALWAYS
# adds at least one Received header to incoming mail.
# If handle_authenticated_message needs to send other mail
# via SMTP, it can call
# send_mail(to,subject_u8,txt,attachment_filenames=[],copyself=True)
# if the following SMTP options are set:
smtp_fromHeader = "Example Name <>"
smtp_fromAddr = ""
smtp_host = "localhost"
smtp_user = ""
smtp_password = "" # or e.g. ("oauth2 cmd",3600)
smtp_delay = 60 # seconds between each message
# (These smtp_ settings are not currently used by anything
# except user-supplied handle_authenticated_message functions)
spamprobe_command = "spamprobe -H all" # (or = None)
spam_folder = "spam"
# you can also set this to a local maildir: ("maildir","path/to/spam")
poll_interval = 4*60
# Note: set poll_interval=False if you want just one run,
# or poll_interval="idle" to use imap's IDLE command (this
# requires the imaplib2 module; you can add to sys.path from
# if you have it in your home directory).
# imaplib2 writes debug information to the console unless
# run with python -O; to suppress this, do
# sed -e s/__debug__/False/g < > i && mv i
logout_before_sleep = False # suggest set to True if using
# a long poll_interval (not "idle"), or if
# filtered_inbox=None, or if the server can't take more
# than one connection at a time
midnight_command = None
# or, midnight_command = "system command to run at midnight"
# (useful if you don't have crontab access on the machine)
calendar_file = None # set to run 'calendar' command on it
# and send output lines as messages
postponed_foldercheck = False
# if True, check for folders named YYYY-MM-DD according to
# the current date, and, if any are found, move their mail
# into filtered_inbox, updating Date lines (but inserting
# the original date into the message body in case it's
# needed for attribution in quoted replies, unless the
# message is from --note or --multinote below). This is so
# you can postpone a message for handling later, simply by
# saving it into a folder named according to the date you
# want to see it, in YYYY-MM-DD format. The check is made
# after midnight for the new date, and on startup for all
# previous dates (in case imapfix hadn't been run every day).
# Additionally, messages that are SSL-authenticated as coming
# from yourself, and whose subject lines start with a date in
# YYYY-MM-DD format, will be saved into a folder of that name
# (with the date removed from the subject line to save screen space).
postponed_daynames = False
# If True, similar to postponed_foldercheck above, but checks
# for lower-case abbreviated month and weekday names
# (e.g. mon, tue, jan, feb - the current locale is used, so
# if you want non-English names then set a different locale).
# This check is performed after midnight, but not on startup
# because they're not absolute dates. To avoid
# false positives, any use of these at the start of
# self-written Subject lines must be followed by : if
# anything else is on the line, e.g. mon: things to do today.
# You can set postponed_foldercheck and postponed_daynames in
# any combination.
postponed_maildir = None # or "path/to/maildir", will
# result in the above postponed_ options checking
# subfolders of this maildir as well as the IMAP server,
# and authenticated messages for postponing being written
# to subfolders of this maildir instead of the server
quiet = True # False = print messages (including quota)
# If you set quiet = 2, will be quiet if and only if the
# standand output is not connected to a terminal
maildirs_to_imap = None # or path to a local directory of
# maildirs; messages will be moved to their corresponding
# folders on IMAP (renaming inbox to filtered_inbox
# and spam to spam_folder, and converting character sets)
maildir_colon = ':'
maildir_delete_emacs_backup_files = True # in case need to
# edit any (e.g. postponed notes) directly in maildir
imap_to_maildirs = None # or path to a local directory of
# maildirs; messages will be moved from IMAP folders to
# subfolders of these maildirs, converting character sets,
# except for filtered_inbox and spam_folder if these are
# on IMAP. If you're using the 'postponed' options, you
# should set postponed_maildir above to the same value as
# imap_to_maildirs.
imap_maildir_exceptions = [] # or list folders like Drafts
# that should not be moved to maildirs
sync_command = None # or command to run after every cycle,
# e.g. "mbsync -a" or "~/.local/bin/offlineimap" if you
# want to maintain maildirs + remote IMAP folders in same state
# (if poll_interval is "idle" this can wait for the next
# change on the _remote_ side before it runs)
maildir_to_copyself = None # or path to a maildir whose
# messages should be moved to imap's copyself (for example if
# mutt is run on the same machine and saves its messages
# locally, which is more responsive than uploading to imap,
# but you still want them to be uploaded to imap eventually)
copyself_delete_attachments = False # if True, attachments
# are DELETED when moving from maildir_to_copyself to the
# IMAP folder (but a record is still kept of what was
# attached, unlike the fcc_attach='no' setting in Mutt 1.5)
copyself_folder_name = "Sent Items"
# Used as a destination folder for maildir_to_copyself
# and/or copyself_delete_attachments, etc.
# (I call it "copyself" because I first used email on a
# local-area network in 1995 with Pegasus Mail for DOS,
# which saved "copies to self" in "copyself.pmi";
# nowadays "sent" or "sent items" seems more common for
# new setups.)
# If you want to keep this on a local maildir instead, you
# can set it to ("maildir","path/to/maildir")
# (in which case also setting maildir_to_copyself will
# result in messages simply being moved from one maildir to
# another, but with copyself_delete_attachments still applied)
copyself_alt_folder = None # or the name of an IMAP
# folder, any messages found (if folder exists) will be
# moved to copyself_folder_name, with their attachments
# deleted if copyself_delete_attachments is True. You can
# specify more than one folder by separating them with a
# comma. Might be useful if some of your IMAP programs
# insist on doing their own sent-mail filing to folders of
# their own choice rather than yours.
check_copyself_alt_folder_on_secondary_too = False
# (see secondary_imap below)
auto_delete_folder = None # or "Trash" if you want everything
# in it to be automatically deleted periodically; useful if
# your IMAP client insists on moving messages there when you
# wanted them deleted permanently; use with caution
archive_path = "oldmail"
archive_rules = [
# These rules are used when you run with --archive
# Each rule is: ("folder-name", max-age (days), spamprobe-action)
# If max-age is not None, older messages will be archived (so you can set it to 0 to archive all messages). Independently of this, spamprobe-action (if specified) will be run on all messages.
# Folder name can be ("maildir","path/to/maildir") if
# you also want to archive from local maildirs to mbox
("spam-confirmed", 30, "train-spam"),
("some-folder", 90, None),
("some-other-folder", None, "train-good"),
compression = "bz2" # or "gz" or None
archived_attachments_path = "archived-attachments" # or None
save_attachments_for_confirmed_spam_too = False
attachment_filename_maxlen = 30
train_spamprobe_nightly = False # if true, will also run
# spamprobe training commands on yesterday's messages in
# these folders every midnight, not just during --archive,
# to catch any message-filing done the previous day
additional_inbox = None
# If you're using an IMAP server that runs its own spam
# filtering, but you want to replace this with yours, you
# can set additional_inbox to their spam folder to treat
# their spam folder as an extra inbox to be reclassified.
# Make sure additional_inbox does not equal spam_folder.
# Alternatively, you can leave additional_inbox at None
# and set spam_folder to match their spam folder to have
# 'or'-logic spam detection, but then you'd have to make
# any manual reviews / saves to spam-confirmed are done
# on the timetable set by the IMAP server (typically
# every 30 days will suffice), plus it may help to keep
# spam-confirmed off-IMAP to avoid accidentally telling
# their system that these messages are non-spam for you.
additional_inbox_train_spam = False # setting this
# to True is another way to get the 'or'-logic if you'd
# rather keep all spam elsewhere e.g. spam_folder is a
# local maildir. header_rules and extra_rules still
# override this.
check_additional_inbox_on_secondary_too = False
# (see secondary_imap below)
additional_inbox_might_not_exist = False # set to True if
# the folder does not appear when it's empty, in which
# case folder not existing will not be treated as an error
forced_names = {} # you can set this to a dictionary
# mapping email address to From name, and whatever other
# From name is used will be replaced by the one specified;
# this is for when one of your contacts has their "From"
# name set to something confusing like their ISP's name
important_regexps = [
# List here any regular expressions that will result in
# the "Importance" flag of a message being set. If the
# list is empty then Importance is unchanged.
# The default rule here marks as 'important' any message
# that seems to include a phone number, as an aid to
# quickly finding these when processing email on a phone
# (if you'd rather call than type and would therefore
# like to find messages that contain phone numbers)
# Set secondary_imap_hostname if you also want to check
# some other IMAP server and treat its messages as being
# in the inbox of the main server (i.e. copied over and
# processed as normal). You can check this less often by
# setting secondary_imap_delay. (Will NOT stay logged in
# between checks.) It's also possible to set the first
# three of these options to lists, in order to check
# multiple secondary IMAP servers or multiple identities.
secondary_imap_hostname = "" # host or host:port
secondary_imap_username = "me"
secondary_imap_password = "xxxxxxxx"
secondary_imap_delay = 24 * 3600
report_secondary_login_failures = False # if True, put a
# message in filtered_inbox about any login failures for
# any of the secondary_imap boxes (default just ignores &
# tries again next time)
secondary_is_insecure = False # if True, the --copy option
# will remove all email addresses (but not names) when
# copying to secondary. This is for when the secondary
# IMAP server is easier to log in to than the primary, but
# is less secure. E.g. your old WM6.5 phone can't do the
# modern TLS version of IMAPS, and the only other option
# is a completely unencrypted connection to some throwaway
# account somewhere. Using --copy to copy over messages
# to read on that phone, but with secondary_is_insecure as
# True, will mean anyone who breaks into that throwaway
# account can see the messages but won't easily be able to
# reply to them for scamming purposes. It will of course
# mean you can't reply so easily as well, but the idea is
# that the throwaway account is only for READING messages
# on the mobile, not for actual replying or management.
first_secondary_is_copy_only = False # if True, the first
# server listed in secondary will be used ONLY for --copy
# and not for periodic checking.
secLimit = 99999 # max number of bytes checked for email
# addresses by secondary_is_insecure (to prevent holdups
# if you have multi-megabyte attachments)
insecure_login = [] # set to a list of host names (or host:port)
# not expected to support SSL. This speeds things up by
# trying the non-SSL login first on those hosts. See also
# secondary_is_insecure option above.
exit_if_imapfix_config_py_changes = False # if True, does
# what it says, on the assumption that a wrapper script
# will restart it (TODO: make it restart by itself?)
# - and if this is set to the special value "stamp", it
# will additionally try to update the timestamp of the
# config file before starting, so that any other instance
# (even on another machine in a cluster) will stop; this
# can be more reliable than exit_if_other_running below
# if running on a cluster.
failed_address_to_subject = True # try to rewrite delivery
# failure reports so that failed addresses are included in
# the Subject line and are therefore visible from a table
# of subjects without having to go in to the message.
# Not all delivery failure reports can be adjusted in this
# way; it depends on how the bouncing relay formats them.
imap_8bit = False # set to True if your IMAP server & clients
# will accept raw 8-bit UTF-8 strings in message bodies (but
# this is not universal, so make sure it re-encodes as MIME
# when forwarding messages to others). Raw UTF-8 messages
# (if supported) save a little bandwidth and allow characters
# to be more readily displayed in Mutt's "Edit Message" etc.
archive_8bit = False # similar if you want --archive as 8bit
exit_if_other_running = True # when run without options,
# try to detect if another no-options imapfix is running
# as the same user, and exit if so. Not guaranteed (for
# example, it can't protect against instances being run on
# different machines of a cluster), but might help to
# reduce the build-up of processes if there is a runaway
# situation with whatever you're using to start them.
# Multiple imapfix instances are sort-of OK but may lead
# to messages being processed in duplicate and/or load the
# IMAP server too much; on the other hand lock files etc
# can have the problem of 'stale' locks. At least a limit
# of 1 process per server in the cluster is better than no
# limit at all. On the other hand you might want to set
# this to False if you run different instances of imapfix
# with different configuration files as the same user and
# don't want to rename (or symlink) imapfix so that these
# look different in the process table.
# (exit_if_other_running needs the Unix 'ps' command.)
# The newer exit_if_imapfix_config_py_changes="stamp"
# option (above) is likely better if you don't mind
# changing the timestamp of your
alarm_delay = 0 # with some Unix networked filesystems it is
# possible for imapfix or one of its subprocesses to get
# "stuck" - if this happens, set alarm_delay to a number of
# seconds (preferably thousands) after which to terminate the
# process using the Unix "alarm clock" mechanism. The clock
# will be reset every half that number of seconds if imapfix
# is still functioning. Note that a long-running filter etc
# could also cause imapfix to become "stuck" this long.
# Command-line options
# --------------------
# Run with no options = process mail as normal
# Run with --once = as if poll_interval=False
# Run with --quicksearch (search string) to search both
# archive_path and the server (all folders), but "quick"
# in that it doesn't decode MIME attachments etc
# (TODO: implement more thorough search, with regexps, but
# it would have to download all messages from the server)
# TODO: this does not yet search local maildirs
# Run with --delete (folder name) to delete a folder
# --delete-secondary to do it on secondary_imap_hostname
# (if there's more than one secondary, the first is used)
# --create (folder name) to create an empty folder
# (e.g. "Sent", some Android apps e.g. K-9 Mail require the
# folder to already exist before they can save messages to it
# (if the folder has been deleted, to get the functionality
# back do Manage folders / Refresh folder list, and Account
# Settings / Folders / Sent folder - set it to Sent)
# --create-secondary to do this on secondary_imap_hostname
# Run with --backup to take a full backup of ALL folders to local .mbox files
# without separating off attachments. Use this for example if your account is
# about to be cancelled and you want to be able to restore things as-is after
# reinstatement or migration to a new service, Read/Answered flags preserved.
# Run with --copy (folder name) to copy a folder onto secondary_imap_hostname
# (updating any that already exists of that name: may delete
# messages on secondary_imap_hostname that aren't on the primary;
# 1st secondary is used if there's a list of many)
# Run with --note (subject) to put a note to yourself
# (taken from standard input) directly into filtered_inbox
# - useful from scripts etc (you can get the note without
# having to wait for it to go through SMTP and polling).
# (If filtered_inbox is None, --note uses real inbox)
# Run with --htmlnote to do the same but send as HTML.
# Run with --maybenote to do --note only if standard input
# has text (no mail left if your script printed nothing).
# Run with --multinote (files) to transfer a group of text
# files into filtered_inbox notes. The first line of each
# file is the subject line of the note. Files will be
# deleted after being successfully uploaded to IMAP.
# Directories are processed recursively but not deleted.
# Backup files (with names ending ~) are deleted without
# being uploaded to IMAP.
# If handle_authenticated_message has been redefined, this
# will be called as well when using --multinote. (If you
# are using this function on a machine that is different
# from the one that does your mail processing, you may
# want to define handle_authenticated_message to
# return "" which will result in the messages being placed
# in the real inbox for processing by the other machine.
# Alternatively you can use --multinote-inbox for the same effect.)
# Use --multinote-fname (files) to use each file's filename
# instead of its first line as the subject, or --multinote-inbox-fname.
# All note options assume that any non-ASCII characters in
# the input will be encoded as UTF-8.
# Run with --upload (files) to upload as attachments into
# filtered_inbox messages, e.g. for transfer to a mobile
# (likely more efficient than SMTP and bypasses filtering)
# - recurse- into directories if specified; won't delete
# files after uploading.
# End of options - non-developers can stop reading now :)
# -------------------------------------------------------
# -------
# If you want to compare this code to old versions, most old
# versions are being kept on SourceForge's E-GuideDog SVN repository
# use: svn co
# and on GitHub at
# and on GitLab at
# and on Bitbucket
# and at
# and in China: git clone
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# See the License for the specific language governing permissions and
# limitations under the License.
from imapfix_config import *
import email,email.utils,time,os,sys,re,base64,quopri,mailbox,traceback,mimetypes
if not sys.version_info[0]==2:
print ("ERROR: ImapFix is a Python 2 program and should be run with 'python2'.\nIt needs major revision for Python 3's version of the email library.\nTry compiling Python 2.7 in your home directory if it's no longer installed on your system.")
# In particular, Python 3.3+ revised the Message class into an EmailMessage class (with Message as a compatibility option), need to use as_bytes rather than as_string; set_payload available only in compatibility mode and works in Python3 Unicode-strings so we'd need to figure out how to handle other charsets including invalid coding.
# That's on top of the usual 'make sure all our code works whether or not type("")==type(u"")' issue.
from email import encoders
from cStringIO import StringIO
if poll_interval=="idle":
import imaplib2 as imaplib
assert not logout_before_sleep, "Can't logout_before_sleep when poll_interval==\"idle\""
else: import imaplib
assert not secondary_imap_delay=="idle", "'idle' polling not implemented for secondary"
if filtered_inbox==None: spamprobe_command = None
if compression=="bz2":
import bz2
compression_ext = ".bz2"
elif compression=="gz":
import gzip
compression_ext = ".gz"
else: compression_ext = ""
if image_size:
from PIL import Image
import imghdr
import commands
def debug(*args):
if not quiet:
print (reduce((lambda a,b:a+str(b)), args, ""))
def check_ok(r):
"Checks that the return value of an IMAP call 'r' is OK, raises exception if not"
typ, data = r
if not typ=='OK': raise Exception(typ+' '+repr(data))
def spamprobe_rules(message,already_spam=False):
if already_spam: # additional_inbox_train_spam
run_spamprobe("train-spam",message) # might help our own classifier to spot other features that weren't what the IMAP server's classifier picked up on (as long as we train-good if it wasn't)
return spam_folder # (if JUST doing this, rename additional_inbox_train_spam to additional_inbox_direct_to_spam or something)
elif run_spamprobe("train",message).startswith("SPAM"): return spam_folder # "train" = classify + maybe update database if wasn't already confident ("receive" = always update database)
else: return filtered_inbox
def run_spamprobe(action,message):
if not spamprobe_command: return ""
cIn, cOut = os.popen2(spamprobe_command+" "+action)
cIn.write(message); cIn.close()
def spamprobe_cleanup():
if not spamprobe_command: return
debug("spamprobe cleanup")
os.system(spamprobe_command+" cleanup")
def process_header_rules(header):
for box,matchList in header_rules:
for headerLine in header.split("\n"):
for m in matchList:
except StopIteration: continue
if box:
if box[0]=='*': box=box[1:] # just save it without indication
elif newmail_directory:
if type(box)==tuple: bStr=re.sub(".*/","",box[1])
else: bStr = box
return rename_folder(box)
return False
def myAsString(msg):
message = msg.as_string()
if not "\r" in message: message=message.replace("\n","\r\n") # in case it came in from Maildir and the library didn't add the CRs back in
if not "\r\n\r\n" in message or ("\n\n" in message and message.index("\n\n")<message.index("\r\n\r\n")):
# oops, broken library?
a,b = message.split("\r\n\r\n",1)
message = re.sub('\r*\n','\r\n',a)+"\r\n\r\n"+b
# Bug in python2.7/email/ CRLF + whitespace sometimes added by _split_ascii after semicolons or commas, but if this occurs inside quoted-printable strings it'll break the display in some versions of alpine & mutt (RFC 2822 says CRLF + whitespace is interpreted as whitespace, and that's not allowed inside ?=..?= sections)
a,b = message.split("\r\n\r\n",1)
message = re.sub(header_charset_regex,lambda x:re.sub(r"\s_","_",re.sub("\r\n\s","",,a,flags=re.DOTALL)+"\r\n\r\n"+b
return message
imapfix_name = sys.argv[0]
if os.sep in imapfix_name: imapfix_name=imapfix_name[imapfix_name.rindex(os.sep)+1:]
if not imapfix_name: imapfix_name = ""
try: _a,_b = imapfix_name.split('.',1)
except: _a,_b = imapfix_name,'0'
imapfix_From_line = "%s <%s@%s>" % (imapfix_name,_a,_b) # previously just imapfix_name, but K-9 Mail Version 5.8 started to require this to be formatted with an @ or it would just say "Unknown Sender"
del _a,_b
def postponed_match(subject):
if postponed_foldercheck:
m = re.match(isoDate,subject)
if m and >= isoToday(): return m.end() # TODO: Y10K (lexicographic comparison)
if postponed_daynames:
m = re.match(weekMonth,subject)
if m: return m.end()
def authenticated_wrapper(subject,firstPart,attach={}):
subject = subject.replace('\n',' ').replace('\r','') # in case the original header is corrupt (c.f. globalise_charsets)
mLen = postponed_match(subject)
if mLen:
newSubj = re.sub("^:","",subject[mLen:]).lstrip() # take the date itself out of the subject line before putting the message in that folder: it could take up valuable screen real-estate in summaries on large-print or mobile devices, and just duplicates information that can be found in the folder name (or in the Date: field after it's put back into filtered_inbox)
if not newSubj: # oops, subject contained only the date and nothing else: take from 1st line of text instead
newSubj = firstPart.lstrip().split('\n')[0]
if len(newSubj) > 60: newSubj = newSubj[:57] + "..." # TODO: configurable abbreviation length?
if postponed_maildir: box=('maildir',postponed_maildir+os.sep+subject[:mLen])
else: box=subject[:mLen] # don't resolve weekday/month names to a date here, because user might actually rely on the "doesn't check for past days on startup" behaviour to postpone to after a trip or something
return box, newSubj
try: r=handle_authenticated_message(subject,firstPart,attach)
if not catch_extraRules_errors: raise # TODO: document it's also catch_authMsg_errors, or have another variable for that
o = StringIO() ; traceback.print_exc(None,o)
save_to(filtered_inbox,"From: "+imapfix_From_line+"\r\nSubject: imapfix_config exception in handle_authenticated_message, treating it as return False\r\nDate: %s\r\n\r\n%s\n" % (email.utils.formatdate(localtime=True),o.getvalue()))
return r, None
def yield_all_messages(searchQuery=None,since=None):
"Generator giving (message ID, flags, message) for each message in the current folder of 'imap', without setting the 'seen' flag as a side effect. Optional searchQuery limits to a text search."
if searchQuery: typ, data =, 'TEXT', '"'+searchQuery.replace("\\","\\\\").replace('"',r'\"')+'"')
elif since: typ, data =, 'SINCE', since)
else: typ, data =, 'ALL')
if not typ=='OK': raise Exception(typ)
bodyPeek_works = True # will set to False if it doesn't
dList = data[0].split(); count = 1
lastTime = time.time()
for msgID in dList:
if not quiet and time.time() > lastTime + 2:
sys.stdout.write(" %d%%\r" % (100*count/len(dList))) ; sys.stdout.flush()
lastTime = time.time()
count += 1
typ, data = imap.fetch(msgID, '(FLAGS)')
if not typ=='OK': continue
flags = data[0]
if '\\Deleted' in flags: continue # we don't mark messages deleted until they're processed; if imapfix was interrupted in the middle of a run, then don't process this message a second time
if '(' in flags: flags=flags[flags.rindex('('):flags.index(')')+1] # so it's suitable for below
if bodyPeek_works:
typ, data = imap.fetch(msgID, '(BODY.PEEK[])')
if not typ=='OK': bodyPeek_works = False
if not bodyPeek_works: # fall back to older RFC822:
typ, data = imap.fetch(msgID, '(RFC822)')
if not "seen" in flags.lower(): # fetching RFC822 will set 'seen' flag, so we'll need to clear it again
try:, 'FLAGS', flags)
except:, 'FLAGS', "()") # gmail can give a \Recent flag but not accept setting it
if not typ=='OK': continue
yield msgID, flags, data[0][1] # data[0][0] is e.g. '1 (RFC822 {1015}'
def rewrite_deliveryfail(msg):
if not failed_address_to_subject: return
subj = msg.get("Subject","")
if not subj.lower().startswith("mail delivery") or not msg.get("From","").lower().startswith("mail delivery"): return
fr = msg.get("X-Failed-Recipients","")
if not fr: return
del msg['Subject'] ; msg['Subject']=fr+' '+subj
return True
for k in forced_names.keys():
if not k==k.lower():
del forced_names[k]
def forced_from(msg):
def f(fr):
if fr.lower() in forced_names:
del msg['From']
msg["From"] = forced_names[fr.lower()] + ' <'+fr.lower()+'>'
return True
fr = msg.get("From","").strip()
if f(fr): return True
if fr.endswith('>') and '<' in fr and f(fr[fr.rindex('<')+1:-1]): return True
# TODO: any other formats to check?
def quote_display_name_if_needed(msg):
# 'From: Testing @ ZOE COVID Study <...>'
# gave 'Unknown Sender' in K9, please quote it
def ensureQuoted(dnMatch):
displayName =
if re.match("^[A-Za-z0-9 .]*$",displayName):
# should be OK to leave unchanged if it didn't use @ etc
return displayName
else: return '"'+displayName+'"'
f = msg.get("From","").strip()
f2 = re.sub(r'^[^<"]*(?= <[^>]*>$)',ensureQuoted,f)
if not f==f2:
del msg['From'] ; msg['From'] = f2 ; return True
def body_text(msg):
"Returns a representation of the message's body text (all parts), for rules"
if msg.is_multipart(): return "\n".join(body_text(p) for p in msg.get_payload())
if not msg.get_content_type().startswith("text/"): return ""
return msg.get_payload(decode=True).strip()
def rewrite_importance(msg):
if not important_regexps: return
changed = False
for h in ['Priority', # e.g. Urgent
'X-Priority', # 1 is highest
'X-MSMail-Priority', # e.g. High
'Importance']: # e.g. High; clients that recognise the previous 3 tend to recognise this one as well, and other clients recognise only this one, so we'll clear all but set only this one
if h in msg:
del msg[h] ; changed = True
bTxt = body_text(msg)
for r in important_regexps:
msg['Importance'] = 'High'; changed=True; break
return changed
def process_imap_inbox():
doneSomething = True
while doneSomething:
doneSomething = False # if we do something on 1st loop, we'll loop again before handing control back to the delay or IMAP-event wait. This is to catch the case where new mail comes in WHILE we are processing the last batch of mail (e.g. another imapfix is running with --multinote and some of the processing calls send_mail() with a callSMTP_time delay: we don't want the rest of the notes to be delayed 29 minutes for the next IMAP-wait to time out)
for is_additional in [False,True]:
if is_additional:
if not additional_inbox: break
if additional_inbox_might_not_exist:
typ, data =
if not typ=='OK': break
else: check_ok(
else: check_ok( # the inbox
imapMsgid = None ; newMail = False
for msgID,flags,message in yield_all_messages():
if not is_additional and leave_note_in_inbox and imap==saveImap and isImapfixNote(message):
if imapMsgid: # somehow ended up with 2, delete one, '+FLAGS', '\\Deleted')
imapMsgid = msgID ; continue
doneSomething = True
msg = email.message_from_string(message)
box = changed = bypass_spamprobe = False ; seenFlag=""
if authenticates(msg):
# do auth'd-msgs processing before any convert-to-attachment etc
debug("Message authenticates")
bypass_spamprobe = True
box,newSubj = authenticated_wrapper(re.sub(header_charset_regex,header_to_u8,msg.get("Subject",""),flags=re.DOTALL),getFirstPart(msg).lstrip(),get_attachments(msg))
if newSubj: # for postponed_foldercheck
del msg["Subject"]
msg["Subject"] = utf8_to_header(newSubj)
changed = True
if box and box[0]=='*':
box=box[1:] ; seenFlag="\\Seen"
if not box==None:
# globalise charsets BEFORE the filtering rules
# (especially if they've been developed based on
# post-charset-conversion saved messages) - but
# no point doing preview images or Office
# conversions for spam or to-delete messages
changed = globalise_charsets(msg,imap_8bit) or changed
changed = remove_blank_inline_parts(msg) or changed
changed = rewrite_deliveryfail(msg) or changed
changed = forced_from(msg) or changed
changed = quote_display_name_if_needed(msg) or changed
changed = rewrite_importance(msg) or changed
if max_size_of_first_part and size_of_first_part(msg) > max_size_of_first_part: msg,changed = turn_into_attachment(msg),True
if changed: message = myAsString(msg)
changed0 = changed # for change_message_id
if box==False:
header = message[:message.find("\r\n\r\n")]
box = process_header_rules(header)
if box==False:
try: box = rename_folder(extra_rules(message))
if not catch_extraRules_errors: raise
o = StringIO() ; traceback.print_exc(None,o)
save_to(filtered_inbox,"From: "+imapfix_From_line+"\r\nSubject: imapfix_config exception in extra_rules (message has been saved to '%s')\r\nDate: %s\r\n\r\n%s\n" % (filtered_inbox,email.utils.formatdate(localtime=True),o.getvalue()))
box = filtered_inbox
if box==False:
if bypass_spamprobe: box = filtered_inbox
else: box = spamprobe_rules(message,is_additional and additional_inbox_train_spam)
if box:
if not box==spam_folder:
changed2 = False
if use_tnef: changed2 = add_tnef(msg) or changed2
if image_size: changed2 = add_previews(msg) or changed2
if office_convert: changed2 = add_office(msg) or changed2
if pdf_convert: changed2 = add_pdf(msg) or changed2
changed = changed2 or changed # don't need to alter changed0 here, since if the only change is adding parts then change_message_id does not need to take effect (at least, not with Gmail January 2022)
if changed2: message = myAsString(msg)
if seenFlag: copyWorked = False # unless we can get copy_to to set the Seen flag on the copy, which means the IMAP server must have an extension that causes COPY to return the new ID unless we want to search for it
elif type(box)==tuple: copyWorked = False # e.g. imap to maildir
elif not changed and saveImap == imap and (not imap_to_maildirs or box in imap_maildir_exceptions):
debug("Copying unchanged message to ",box)
copyWorked = copy_to(box, msgID)
if not copyWorked: debug("... failed; falling back to re-upload")
else: copyWorked = False
if not copyWorked:
debug("Saving message to ",box)
save_to(box, message, seenFlag, changed0 and saveImap==imap)
if box==filtered_inbox: newMail = True
else: debug("Deleting message"), '+FLAGS', '\\Deleted')
if not is_additional and leave_note_in_inbox and imap==saveImap:
if newMail:
if imapMsgid: # delete the old one, '+FLAGS', '\\Deleted')
save_to("", imapfixNote())
elif imapMsgid:
# un-"seen" it (in case the IMAP server treats our fetching it as "seen"); TODO what if the client really read it but didn't delete?, '-FLAGS', '\\Seen')
if (not quiet) and imap==saveImap and not type(filtered_inbox)==tuple: debug("Quota ",repr(imap.getquotaroot(filtered_inbox)[1])) # RFC 2087 "All mailboxes that share the same named quota root share the resource limits of the quota root" - so if the IMAP server has been set up in a typical way with just one limit, this command should print the current and max values for that shared limit. (STORAGE = size in Kb, MESSAGE = number)
def authenticates(msg):
if not trusted_domain or not smtps_auth: return
rxd = msg.get_all("Received",[])
if not rxd: return True # no Received headers = put in by imapfix on another machine
for rx in rxd:
rx = re.sub(r'\s+',' ',rx)
m=re.match(r"from ([^ ]+)( \([^)]*\))* by ([^ ]+)( .*)",rx) # RFC 2821 section 4.4 extended a bit (not everybody follows this, but trusting the machines on our network implies trusting they'll follow at least this regex and cannot be tricked into not doing so)
if not m:
# we can't process this one, but should it break the chain?
if type(trusted_domain)==list and "" in trusted_domain:
# option is set: non-"from" rx header doesn't interrupt scan
else: break # if in doubt, don't trust this message
claimed_from, tcp_from_and_other, receiving_machine, other = m.groups()
# The identity of receiving_machine is trusted by induction: if all (0 or more) previous headers were trusted to correctly report both themselves and the next hop, and if said 'next hop' reports all identified trusted networks, then we trust the claim of the receiving machine on this header. (This implies we always trust the claim of the first such header if any trusted_domain is set.)
if type(trusted_domain)==list:
if not any(receiving_machine.endswith(t) for t in trusted_domain if t): break # (not on any of our networks: this normally means there will have been 0 previous headers and this message has been introduced by some other method, so don't trust it)
elif not receiving_machine.endswith(trusted_domain): break # (as above)
# We are trusting all machines so far never to incorrectly report an authentication. So if THIS machine reports an authentication (wherever the message came from on previous hop) then we pass it.
if not tcp_from_and_other: tcp_from_and_other = ""
if not other: other = ""
check_for_auth_info = tcp_from_and_other + other # smtps_auth has been known to occur in EITHER of these places (Postfix puts it after the TCP info; Exim puts it after the receiver in the "with" section)
if type(smtps_auth)==list:
if any((s in check_for_auth_info or len(s.split())==2 and s.split()[0]=="from" and s.split()[1]==claimed_from and not tcp_from_and_other) for s in smtps_auth): return True
elif smtps_auth in check_for_auth_info or len(smtps_auth.split())==2 and smtps_auth.split()[0]=="from" and smtps_auth.split()[1]==claimed_from and not tcp_from_and_other: return True
# Now check next-older hop. If it is not on any of our trusted networks, then stop trusting, no matter what is claimed in further Received headers.
# We CANNOT trust claimed_from: we MUST check the first item in tcp_from_and_other. If any server on your network fails to put a reverse-DNS in here then the IP address of the next hop must be included in the trusted_domain list.
m = re.match(r" \(([^)\[]*)\[([^)\]]*)\][^)]*\)",tcp_from_and_other)
if not m: break
reverse_dns, ip = m.groups()
reverse_dns = reverse_dns.strip()
if reverse_dns.endswith('.'): reverse_dns=reverse_dns[:-1] # gmail bug
if type(trusted_domain)==list:
if not any((reverse_dns.endswith(t) or ip==t) for t in trusted_domain if t): break
elif not reverse_dns.endswith(trusted_domain): break
def imapfixNote(): return "From: "+imapfix_From_line+"\r\nSubject: Folder "+repr(filtered_inbox)+" has new mail\r\nDate: "+email.utils.formatdate(localtime=True)+"\r\n\r\n \n" # make sure there's at least one space in the message, for some clients that don't like empty body
# (and don't put a date in the Subject line: the message's date is usually displayed anyway, and screen space might be in short supply)
def isImapfixNote(msg): return ("From: "+imapfix_name) in msg and ("Subject: Folder "+repr(filtered_inbox)+" has new mail") in msg
def get_maildir(dirName,create=True):
if maildir_delete_emacs_backup_files:
for c in ['cur','new']:
try: l=os.listdir(dirName+os.sep+c)
except: l = []
for f in l:
if f.endswith('~'): tryRm(dirName+os.sep+c+os.sep+f)
return mailbox.Maildir(dirName,None,create)
def archive(foldername, mboxpath, age, spamprobe_action):
if not age==None:
if spamprobe_action: extra = ", spamprobe="+spamprobe_action
else: extra = ""
toDbg="Archiving from "+str(foldername)+" to "+str(mboxpath)+extra+"..."
suf = "" ; sc = 0
while os.path.exists(mboxpath+suf+compression_ext):
sc += 1 ; suf = "."+str(sc)
while sc:
if sc==1: suf = ""
else: suf = "."+str(sc-1)
suf2 = "."+str(sc)
sc -= 1
mbox = mailbox.mbox(mboxpath) # will compress below
toDbg = "Processing "+str(foldername)+"..."
mbox = None
is_maildir = type(foldername)==tuple and foldername[0]=='maildir'
if is_maildir:
try: maildir = get_maildir(foldername[1])
except: return # couldn't select that folder
def generator():
for m in maildir.iteritems():
k,v = m
yield k,v.get_flags(),myAsString(v)
toDel = []
make_sure_logged_in() ; debug(toDbg)
typ, data =
if not typ=='OK': return # couldn't select folder
generator = yield_all_messages
for msgID,flags,message in generator():
if spamprobe_action:
# TODO: combine multiple messages first?
run_spamprobe(spamprobe_action, message)
if age==None: continue
if type(message)==type(''):
msg = email.message_from_string(message)
else: # maildir returns it as a Message etc
msg,message = message,myAsString(message)
try: t = email.utils.mktime_tz(email.utils.parsedate_tz(msg['Date']))
except: t = time.time() # undated message or invalid date (sometimes happens in spam for example)
if t >= time.time() - age: continue
if not message.startswith("From "): # needed for Unix mbox, otherwise mbox lib fills it in with MAILER-DAEMON
f = []
if 'From' in msg:
fr = msg['From']
if '<' in fr and '>' in fr[fr.index('<'):]:
fr = fr[fr.index('<')+1:fr.rindex('>')]
if not fr: fr = 'unknown'
f.append(time.asctime(time.gmtime(t))) # must be THIS format for at least some versions of mutt to parse the message
message="From "+' '.join(f)+"\r\n"+message
msg = email.message_from_string(message)
globalise_charsets(msg,archive_8bit) # in case wasn't done on receive (this also makes sure UTF-8 is quopri if it would be shorter, which can make archives easier to search)
if archived_attachments_path and (save_attachments_for_confirmed_spam_too or not spamprobe_action or not "spam" in spamprobe_action): save_attachments_separately(msg)
mbox.add(msg) # TODO: .set_flags A=answered R=read ? (or doesn't it matter so much for old-message archiving; flags aren't always right anyway as you might have answered some using another method)
if is_maildir: toDel.append(msgID)
else:, '+FLAGS', '\\Deleted')
if mbox:
if not os.stat(mboxpath).st_size: tryRm(mboxpath) # ended up with an empty file - delete it
if archive_8bit:
d1 = open(mboxpath,'rb').read()
d2 = quopri_to_u8_8bitOnly(d1) # TODO: the ENTIRE mbox in one operation? what if there's some non-MIME messages in there that just happen to contain URLs with =(hex digits) in them?
if not d2==d1: open(mboxpath,'wb').write(d2)
if compression:
open_compressed(mboxpath,'wb').write(open(mboxpath,'rb').read()) # will write a .bz2 etc
# don't do this until got here without error:
if is_maildir: # (don't del during iteritems)
for msgID in toDel: del maildir[msgID]
else: check_ok(imap.expunge())
def fix_archives_written_by_imapfix_v1_308():
for f in listdir(archive_path):
f = archive_path+os.sep+f
if f.endswith(compression_ext): f2 = open_compressed(f[:-len(compression_ext)],'r')
else: f2 = open(f)
r = [] ; changed = False
for l in f2:
if l.startswith("From "):
l2 = l.replace('<','')
try: t = email.utils.mktime_tz(email.utils.parsedate_tz(' '.join(l2.split()[-6:])))
except: t = None
if t:
l2 = ' '.join(l2.split()[:-6])+' '+time.asctime(time.gmtime(t))+'\r\n'
r.append(l2) ; changed = True ; continue
if changed:
print ("Fixing "+f)
if f.endswith(compression_ext): open_compressed(f[:-len(compression_ext)],'wb').write(''.join(r))
else: open(f,'wb').write(''.join(r))
else: print ("No need to fix "+f)
def get_attachments(msg):
if msg.is_multipart():
d = {}
for i in msg.get_payload():
return d
try: fname = msg.get_filename()
except: fname="illegal-filename"
if not fname: return {}
data = msg.get_payload(None,True)
if data: return {fname:data}
else: return {}
def remove_blank_inline_parts(msg):
# some mailers send HTML-only messages with completely blank text/plain alternatives. In this case it's best to remove the blank alternative so that a client told to prefer text can display the HTML. TODO: document that we do this?
if not msg.is_multipart(): return
ll = msg.get_payload()
i = 0 ; changed = False
while i < len(ll):
if (not 'Content-Type' in ll[i] or ll[i]["Content-Type"].startswith("text/")) and not ll[i].get_payload(decode=True).strip() and len(ll)>1:
del ll[i] ; changed = True
else: i += 1
return changed
def save_attachments_separately(msg):
def save_attachment_separately(msg):
try: fname = msg.get_filename()
except: return
if not fname: return
if fname.startswith("imapfix-preview"): return
if not type(fname)==type(u""):
try: fname = fname.decode('utf-8')
except: pass
try: fname=fname.encode("unicode-escape").replace(r"\u","_").replace("/"," ").replace("\\"," ")
except: fname="illegal-filename-B"
fname = re.sub(r'\s+',' ',fname)
if '.' in fname: fext = fname[fname.rindex('.'):]
else: fext = ""
fname = fname.replace(os.sep,'.')
if len(fext) > attachment_filename_maxlen:
fname = fname[:attachment_filename_maxlen]
if fext: fname = fname[:-len(fext)]
fname = fname[:attachment_filename_maxlen-len(fext)] + fext
data = msg.get_payload(None,True)
if not data: return
try: os.mkdir(archived_attachments_path)
except: pass # OK if exists
f2 = archived_attachments_path+os.sep+fname
suffix = "" ; sufCount = 0
while os.path.exists(f2+suffix+compression_ext):
if open_compressed(f2+suffix,'rb').read() == data:
# it's a duplicate - keep what we already have
sufCount += 1 ; suffix = "."+str(sufCount)
def delete_attachments(msg):
def delete_attachment(msg):
if msg.get_filename(): msg.set_payload("")
def nightly_train(foldername, spamprobe_action):
if type(foldername)==tuple:
try: maildir = get_maildir(foldername[1],False)
except: return
timeYesterday = time.time()-24*3600
for msgID,msg in maildir.iteritems():
try: t = email.utils.mktime_tz(email.utils.parsedate_tz(msg['Date']))
except: t = 0
if t >= timeYesterday: run_spamprobe(spamprobe_action, myAsString(msg)) # TODO: combine multiple messages first?
typ, data =
if not typ=='OK': return # couldn't select that folder
for msgID,flags,message in yield_all_messages(since='-'.join(email.utils.formatdate(time.time()-24*3600,localtime=True).split()[1:4])): run_spamprobe(spamprobe_action, message) # TODO: combine multiple messages first?
def open_compressed(fname,mode):
if compression=="bz2":
return bz2.BZ2File(fname+".bz2",mode)
# (compresslevel default is 9)
elif compression=="gz":
# (again, compresslevel default is 9)
elif compression: raise Exception("Unrecognised compression type") # essential, as archive() assumes it can delete the original file after calling open_compressed
return open(fname,mode)
def copy_to(mailbox, message_id):
"Try to save an unmodified message to a folder, using the imap COPY command instead of re-uploading the message"
assert imap == saveImap, "Can't call copy_to if saveImap != imap"
typ, data = imap.fetch(message_id, "(UID)")
if not typ=='OK': return False
uid = data[0]
if '(' in uid:
if uid.startswith("UID "): uid=uid[4:]
try: return imap.uid('COPY', uid, mailbox)[0]=='OK'
except: return False
already_created = set()
def maybe_create(mailbox):
if mailbox and not mailbox in already_created:
assert not type(mailbox)==tuple, "maybe_create should not be called with a maildir mailbox"
saveImap.create(mailbox) # error if exists OK
def save_to(mailbox, message_as_string, flags="", mayNeedNewMsgID=True):
"Saves message to a mailbox on the saveImap connection, creating the mailbox if necessary"
if type(mailbox)==tuple and mailbox[0]=='maildir': return save_to_maildir(mailbox[1],message_as_string,flags)
elif imap_to_maildirs and not mailbox in imap_maildir_exceptions: return save_to_maildir(imap_to_maildirs+os.sep+mailbox,message_as_string,flags) # might as well skip the intermediate step of putting it in an IMAP folder to be picked up by imap_to_maildirs on next cycle
make_sure_logged_in() ; maybe_create(mailbox)
msg = email.message_from_string(message_as_string)
if 'Date' in msg: imap_timestamp = email.utils.mktime_tz(email.utils.parsedate_tz(msg['Date'])) # We'd better not set the IMAP timestamp to anything other than the Date line. IMAP timestamps are sometimes used for ordering messages by Received (e.g. Mutt, Alpine, Windows Mobile 6), but not always (e.g. Android 4 can sort only by Date); they're sometimes displayed (e.g. WM6) but sometimes not (e.g. Alpine), and some IMAP servers have sometimes been known to set them to Date anyway, so we can't rely on a different value (e.g. for postponed_foldercheck) always working.
else: imap_timestamp = time.time() # undated message ? (TODO: could parse Received lines)
if change_message_id and mayNeedNewMsgID and 'Message-ID' in msg: message_as_string = message_as_string.replace("Message-ID: <","Message-ID: <2",1) # .muttrc editor default adds '1', so let's add a '2'
if imap_8bit and"(?i)Content-Transfer-Encoding: quoted-printable",message_as_string): message_as_string = quopri_to_u8_8bitOnly(message_as_string) # TODO: check the "Content-Transfer-Encoding" is in the header or one of the MIME parts (don't do it if it's a non-MIME ascii-only message that happens to contain that text and perhaps some URLs with =(hex digits) in them)
check_ok(saveImap.append(mailbox, flags, imaplib.Time2Internaldate(imap_timestamp), message_as_string))
def save_to_maildir(maildir_path, message_as_string, flags=""):
mailbox.Maildir.colon = maildir_colon
m = mailbox.Maildir(maildir_path,None)
if"(?i)Content-Transfer-Encoding: quoted-printable",message_as_string): message_as_string = quopri_to_u8_8bitOnly(message_as_string)
msg = mailbox.MaildirMessage(email.message_from_string(message_as_string.replace("\r\n","\n")))
def rename_folder(folder):
if not isinstance(folder,str): return folder
if folder.lower()=="inbox":
return filtered_inbox
elif folder.lower()=="spam": return spam_folder
else: return folder
def listdir(d): return sorted(os.listdir(d))
def do_maildirs_to_imap():
mailbox.Maildir.colon = maildir_colon
for d in listdir(maildirs_to_imap):
d2 = maildirs_to_imap+os.sep+d
if not os.path.exists(d2+os.sep+"cur"):
continue # not a maildir
to = rename_folder(d)
debug("Moving messages from maildir ",d," to imap ",to)
m = get_maildir(d2)
for k,msg in m.items():
del m[k]
def clean_empty_maildir(d2):
try: mailbox.Maildir(d2,None).clean()
except: pass
newcurtmp = ["new","cur","tmp"]
if not any(listdir(d2+os.sep+ntc) for ntc in newcurtmp):
# folder is now empty: remove it
for nct in newcurtmp: os.rmdir(d2+os.sep+nct)
try: os.rmdir(d2)
except: pass
def imap_flags_from_maildir_msg(msg): return " ".join(" ".join({'S':r'\Seen','D':r'\Deleted','R':r'\Answered','F':r'\Flagged'}.get(flag,"") for flag in msg.get_flags()).split())
def maildir_flags_from_imap(flags):
r = ""
if r"\Deleted" in flags: r += 'D'
if r"\Flagged" in flags: r += 'F'
if r"\Answered" in flags: r += 'R'
if r"\Seen" in flags: r += 'S'
return r
def do_maildir_to_copyself():
mailbox.Maildir.colon = maildir_colon
m = get_maildir(maildir_to_copyself)
said = False
for k,msg in m.items():
if not said:
debug("Moving messages from ",maildir_to_copyself," to imap ",copyself_folder_name)
said = True
if copyself_delete_attachments:
del m[k]
assert not copyself_folder_name == ('maildir', maildir_to_copyself), "this can lead to loops"
assert not (maildirs_to_imap and imap_to_maildirs), "Setting both maildirs_to_imap and imap_to_maildirs at the same time is not supported"
def do_copyself_to_copyself():
for folder in copyself_alt_folder.split(","):
if folder==copyself_folder_name:
debug("Cannot specify copyself_folder_name in copyself_alt_folder: skipping ",folder)
typ, data =
if not typ=='OK': continue # skip non-selectable folder, without output: it may appear later if the relevant MUA is used
said = False
for msgID,flags,message in yield_all_messages():
if not said: # don't say until we know there's at least some messages in the folder
debug("Moving messages from ",folder," to ",copyself_folder_name)
said = True
msg = email.message_from_string(message)
charsets_changed = globalise_charsets(msg,imap_8bit)
if copyself_delete_attachments:
delete_attachments(msg) # apparently deleting an attachment does not affect mayNeedNewMsgID, at least Gmail Jan 2022: use only charsets_changed for that
save_to(copyself_folder_name,myAsString(msg),"\\Seen",charsets_changed) # can't use copy_to even if msg hasn't changed, because we can't get copy_to to set the Seen flag on the copy (unless an IMAP extension that can help us do that is installed), '+FLAGS', '\\Deleted')
if said: check_ok(imap.expunge())
def do_auto_delete():
for folder in auto_delete_folder.split(","):
typ, data =
if not typ=='OK':
debug("Skipping non-selectable folder ",folder)
said = False
for msgID,flags,message in yield_all_messages():
if not said:
debug("Deleting messages from ",folder)
said = True, '+FLAGS', '\\Deleted')
if said: check_ok(imap.expunge())
header_charset_regex = r'=\?(.*?)\?(.*?)\?(.*?)\?=' # RFC 2047
def header_to_u8(match):
charset =
if charset in ['gb2312','gbk']: charset='gb18030'
encoding =
text =
if encoding.upper()=='Q': text = quopri.decodestring(text,header=True)
else: text = base64.decodestring(text)
text = text.decode(charset)
debug("Bad header line: exception decoding ",repr(text)," in ",charset,", leaving unchanged")
return text.encode('utf-8')
def globalise_header_charset(match):
# return # no changes needed
# - actually, do it anyway, because some UTF8 headers might be
# quopri-encoded when they're only ASCII, etc, and "normalising"
# these might help the writing of filtering rules.
# Also, we want to de-"stylise" Unicode 1D400..1D6A3
# so it works better with speech synths and rules.
hu8 = header_to_u8(match)
if hu8 == return hu8 # something went wrong; at least don't double-encode it
return utf8_to_header(destylise_u8_header(hu8))
def destylise_u8_header(u8):
def sub(match):
aNum = ((ord(m[2])-0x90)*(0xc0-0x80)+ord(m[3])-0x80) % 52
if aNum>=26: return chr(ord('a')+aNum-26)
else: return chr(ord('A')+aNum)
u8 = re.sub("\xf0\x9d[\x90-\x99][\x80-\xbf]",sub,re.sub("\xf0\x9d\x9a[\x80-\xa3]",sub,u8)) # U+1D400..U+1D6A3 (works even if Python is narrow build)
if"\xef[\xbc\xbd]",u8): u8=re.sub(u"[\uFF01-\uFF5E]",lambda m:chr(ord('!')),u8.decode('utf-8')).encode('utf-8') # full-width ASCII
return u8
def utf8_to_header(u8):
if not ('=?' in u8 or"[^ -~]",u8)): return u8 # ASCII and no encoding needed
ret = "B?"+base64.encodestring(u8).replace("\n","")
qp = "Q?"+re.sub("=?\n","",quopri.encodestring(u8,header=True)).replace('?','=3F') # must have header=True for alpine (although mutt and Outlook etc may work either way, especially if the with-spaces version is not wrapped, but alpine fails to decode the quopri if any space is present)
if len(qp) <= len(ret): ret = qp
return "=?UTF-8?"+ret+"?="
import email.mime.multipart,email.mime.message,email.mime.text,email.mime.image,email.charset,email.mime.base
def turn_into_attachment(message,covering_text=None,attach_raw=False):
if covering_text==None: covering_text = "Large message converted to attachment" # by imapfix, but best not mention this as it might bias the filters?
m2 = email.mime.multipart.MIMEMultipart()
for k,v in message.items():
if not k.lower() in ['content-length','content-type','content-transfer-encoding','lines','mime-version','content-disposition']: m2[k]=v
if attach_raw: m2.attach(message)
else: m2.attach(email.mime.message.MIMEMessage(message))
return m2
def size_of_first_part(message):
if message.is_multipart():
for i in message.get_payload():
return size_of_first_part(i)
return 0
return len(message.get_payload())
def getFirstPart(message):
if message.is_multipart():
for i in message.get_payload():
return getFirstPart(i)
return ""
try: pl = message.get_payload(decode=True)
except: return ""
cs = message.get_content_charset(None)
if cs in [None,'us-ascii','utf-8']: return pl
if cs in ['gb2312','gbk']: cs = 'gb18030'
return pl.decode(cs).encode('utf-8')
def globalise_charsets(message,will_use_8bit=False,force_change=False):
"""'Globalises' the character sets of all parts of
email.message.Message object 'message'.
Only us-ascii and utf-8 charsets are 'global'.
Also tries to use quoted-printable rather than
base64 if doing so won't increase the length.
Returns True if any changes were made.
(Use force_change if you need to ensure encoding
options are set correctly in case you want to do
setPayload yourself to change the text.)"""
changed = False
for line in ["From","To","Cc","Subject","Reply-To"]:
if not line in message: continue
l = message[line]
l = l.replace(" ?utf-8?q?"," =?utf-8?q?") # bug observed in mail generated by "SOGoMail 2.3.23"
l2 = re.sub(header_charset_regex,globalise_header_charset,l,flags=re.DOTALL).replace('\n',' ').replace('\r','') # the \n and \r replacements are in case the original header is corrupt
if l==l2: continue
# debug("Changing ",line," from ",l," to ",l2)
# message["X-Imapfix-Old-"+line] = l
del message[line]
message[line] = l2
changed = True
if message.is_multipart():
for i in message.get_payload():
if globalise_charsets(i,will_use_8bit,force_change): changed = True
return changed
cType = message.get_content_type()
is_html = cType and cType.startswith("text/html")
specified_charset = message.get_content_charset(None)
while specified_charset and specified_charset.startswith("charset="):
# bug in some mailers: charset=charset=
specified_charset = specified_charset[len("charset="):]
force_change = True
is_unspecified = cType and cType.startswith("text/") and not specified_charset
try: p0 = message.get_payload(decode=True) # in most cases we need it (TODO: in a few small cases we don't, but low-priority as the entire message has probably been loaded into RAM already)
message['X-ImapFix-Globalise-Charset-Decode-Error'] = str(sys.exc_info()[1])
return True
if specified_charset=='us-ascii' and'[\x80-\xff]',p0): # mislabelled ASCII
force_change = is_unspecified = True
specified_charset = None
changed = True
if is_unspecified:
if p0.startswith('\xfe\xff'): specified_charset = 'utf-16be'
elif p0.startswith('\xff\xfe'): specified_charset = 'utf-16le'
elif p0.startswith('\xef\xbb\xbf'): specified_charset = 'utf-8'
import chardet
d = chardet.detect(p0)
threshold = 0.99
if d['confidence'] >= threshold:
specified_charset = d['encoding']
message['X-ImapFix-Chardet']=d['encoding']+"; confidence="+str(d['confidence'])+" does not meet threshold "+str(threshold)+" so not setting charset"
changed = True
except: pass # no chardet or sthg
if not force_change: # should we SET force_change? :
if 'Content-Transfer-Encoding' in message and not message['Content-Transfer-Encoding']=='quoted-printable': force_change = True # if it's base64, always see if we can change it
elif specified_charset=='utf-8':
if p0.startswith('\xef\xbb\xbf'): force_change = True # we'll want to remove that "BOM"
elif will_use_8bit:
if"[^\x00-\x80]",message.get_payload(decode=True)): force_change = True
except: pass # we would full back to return anyway on the 'problems decoding this message' below
if specified_charset:
if not force_change and specified_charset in ['us-ascii','utf-8'] and not is_html: return changed # no further conversion required
if specified_charset in ['gb2312','gbk']: specified_charset = 'gb18030'
try: p = p0.decode(specified_charset)
message['X-ImapFix-Globalise-Charset-Error'] = str(sys.exc_info()[1])
return True
if is_html:
q = '[\'"]' # could use either " or ' quotes
p = re.sub(r'(?i)<meta\s+http[_-]equiv='+q+r'?content-type'+q+r'?\s+content='+q+'[^\'"]*'+q+r'>','',p) # better remove charset meta tags after we changed the charset (TODO: what if they conflict with the message header anyway?)
p = re.sub(r'(?i)<meta\s+content='+q+'[^\'"]*'+q+r'\s+http[_-]equiv='+q+r'?content-type'+q+r'?>','',p) # some authoring tools emit the attributes in THIS order
if p.startswith(u"\ufeff"): p=p[1:]
specified_charset = 'utf-8'
p = p.encode(specified_charset)
else: # no specified_charset and we couldn't detect
p = p0
specified_charset = 'x-unknown' # contrary to the documentation, at least some versions of the library add 'us-ascii' if set to None
if not force_change and p==p0: return changed # didn't fix meta tags or change charset so don't need to re-encode
return setPayload(message,p,specified_charset,will_use_8bit)
def setPayload(message,p,charset,will_use_8bit=False):
if 'Content-Transfer-Encoding' in message:
isQP = (message['Content-Transfer-Encoding']=='quoted-printable')
del message['Content-Transfer-Encoding']
else: isQP = False
if not isQP: # not already decided on quoted-printable; SHOULD we?
b64len = len(base64.encodestring(p))
pp = quopri.encodestring(p)
isQP = (len(pp) <= b64len or (will_use_8bit and len(quopri_to_u8_8bitOnly(pp)) <= b64len)) # use Quoted-Printable rather than Base64 for UTF-8 if the original was quopri or if doing so is shorter (besides anything else it's easier to search emails without tools that way)
for i in [1,2]: # before AND after, just in case
if isQP: email.charset.add_charset(charset,email.charset.SHORTEST,email.charset.QP,charset)
else: email.charset.add_charset(charset,email.charset.SHORTEST,email.charset.BASE64,charset)
if i==2: return True
def quopri_to_u8_8bitOnly(s): # used by imap_8bit and archive_8bit (off by default). Must ensure any header blocks in s are not touched! (or imap server may substitute Xs etc)
if'[\x80-\xff]',s): return s # message already contains 8-bit stuff: we probably shouldn't redo this in case of coincidental UTF-8 quoted-printable in a binary attachment (TODO: what if such an attachment happens to lack any bytes with the high bit set)
avoid = set()
def avoidAdd(m):
re.sub(header_charset_regex,avoidAdd,s,flags=re.DOTALL) # don't want to interfere with MIME 'charset' etc in subject lines
re.sub(r'h(=\r?\n)?t(=\r?\n)?t(=\r?\n)?p(=\r?\n)?(s(=\r?\n)?)?:(=\r?\n)?/(=\r?\n)?/(=\r?\n)?([^ "<>^'+"`'"+'](=\r?\n)?)+',avoidAdd,s) # and don't want to interfere with =hex in URLs (the '=' of which should have been double-encoded anyway, but sometimes isn't)
def maybeDecode(m):
for start,end in avoid: # TODO: improve efficiency when dealing with a very large archive with many headers?
if start <= m.start() < end: return
return quopri.decodestring(
return re.sub(r"(=(([C][2-9A-F]|[D][0-9A-F])|E0=[AB][0-9A-F]|(E[1-9A-CEF]|F0=[9AB][0-9A-F]|F4=8[0-F])(=\r?\n)?=[89AB][0-9A-F]|ED=[89][0-9A-F]|F[1-3]=[89AB][0-9A-F]=[89AB][0-9A-F])(=\r?\n)?=[89AB][0-9A-F])+",maybeDecode,s) # decodes only 8-bit utf-8 characters from the quoted-printable string, leaving alone any 7-bit characters that are encoded as quoted-printable ("^From" etc)
def walk_msg(message,partFunc,*args):
if message.is_multipart():
changed = False
global to_attach # for add_preview: it needs to add previews to the container immediately above, not necessarily the top-level container (as we might have a multipart/related within a multipart/alternative or something)
try: to_attach
except: to_attach = None # not called from add_preview
o,to_attach = to_attach,[]
for i in message.get_payload():
changed = walk_msg(i,partFunc,*args) or changed
if to_attach and message.get("Content-Type","").startswith("multipart/alternative"): # can happen with Exchange clients even if only one attachment
ct = message["Content-Type"].replace("multipart/alternative","multipart/mixed",1)
del message["Content-Type"]
message["Content-Type"] = ct
for i in to_attach: message.attach(i)
to_attach = o ; return changed
else: return partFunc(message,*args)
def add_previews(message):
global to_attach ; to_attach = None
accum = [1]
return walk_msg(message,add_preview,accum)
def add_preview(message,accum):
if to_attach == None: return False # TODO? (non-multipart message sent with a single image and nothing else)
if not 'Content-Type' in message or message["Content-Type"].startswith("text/"): return False
payload = message.get_payload(decode=True)
except: return False # not an image, or corrupt
changed = False
if message["Content-Type"].startswith("application/"):
try: what = imghdr.what(None,payload)
except: what = None
if what:
del message["Content-Type"]
changed = True
if 274 in exif:
if exif[274]==3:
elif exif[274]==6:
elif exif[274]==8:
except: pass # no EXIF data, or can't de-rotate
try: img.thumbnail(image_size,Image.ANTIALIAS)
except: return changed # probably a JPEG-variant decoder not available
s1 = StringIO();,'JPEG');s1=s1.getvalue()
except: s1 = None
s2 = StringIO();,'PNG'); s2=s2.getvalue()
except: s2 = None
if s1==None and s2==None: return changed
elif s1==None: s,ext = s2,"png"
elif s2==None or len(s2) > len(s1): s,ext = s1,"jpg"
else: s,ext = s2,"png"
if len(s) > len(payload): return changed # we failed to actually compress the image
if ext=="png": subtype = "png"
else: subtype = "jpeg"
i = email.mime.image.MIMEImage(s,_subtype=subtype) # _subtype defaults to auto-detect but it can fail
if not i['Content-Type'].lower().startswith('image/'): i['Content-Type']='image/'+i['Content-Type'] # depends on version of PIL?
i['Content-Disposition']='attachment; filename=imapfix-preview'+str(accum[0])+'.'+ext # needed for some clients to show it
to_attach.append(i) ; accum[0] += 1
return True
def add_office(message):
global to_attach ; to_attach = None
accum = [1]
return walk_msg(message,add_office0,accum)
def filename_ext(message):
if not 'Content-Disposition' in message: return False
fn = str(message['Content-Disposition'])
if not 'filename' in fn: return False
fn = re.sub(header_charset_regex,header_to_u8,fn[fn.index('filename'):].replace("filename=","",1).replace('"',""))
if not '.' in fn: return False
ext = fn[fn.rindex('.'):].lower()
if ';' in ext: ext=ext[:ext.index(';')]
return fn,ext
def add_office0(message,accum):
if to_attach == None: return False # TODO? (non-multipart message sent with a single document and nothing else)
if "Content-Type" in message and (str(message['Content-Type']).startswith("text/calendar") or str(message['Content-Type']).startswith("application/ics")):
cal = ["Calendar file:"]
for l in re.sub("LANGUAGE=[a-zA-Z-]*:","",message.get_payload(decode=True).replace("\r\n","\n").replace("\n "," ").replace("\nX-MICROSOFT-","\n").replace(";ROLE=REQ-PARTICIPANT","").replace(";PARTSTAT=NEEDS-ACTION","").replace(";RSVP=TRUE","").replace(";VALUE=DATE","").replace("mailto:","")).split("\n"):
if any(l.startswith(f) for f in ["TZID:","ORGANIZER","ATTENDEE","SUMMARY","DTSTART:2","DTEND:2","LOCATIONDISPLAYNAME","LOCATIONSTREET","LOCATIONCITY"]) and ':' in l and l[l.index(':')+1:].replace(r'\n','').strip(): cal.append(" ".join(l.replace(";","; ").replace(":",": ").split()))
if len(cal)==1: return False # no details found
b = email.mime.base.MIMEBase("text","plain")
to_attach.append(b) ; return True
fn = filename_ext(message)
if fn==False: return False
fn,ext = fn
if not ext in ".doc .docx .rtf .odt .xls .xlsx .ods .ppt .odp".split(): return False
debug("Getting payload")
payload = message.get_payload(decode=True)
infile = "tmpdoc-%d%s" % (os.getpid(),ext)
outfile = "tmpdoc-%d.%s"%(os.getpid(),office_convert)
debug("Running soffice to get ",office_convert)
if os.system("soffice --convert-to %s %s" % (office_convert, infile)): # conversion error, or soffice not found
debug("soffice run returned failure")
clean_tmpdoc() ; return False
mimeType = {"html":"text/html; charset=utf-8","pdf":"application/pdf"}.get(office_convert,"application/binary")
mT,subT = mimeType.split("/")
b = email.mime.base.MIMEBase(mT,subT)
if not office_convert=="html": encoders.encode_base64(b)
if office_convert in ["html"] and not (max_size_of_first_part and len(b.get_payload) > max_size_of_first_part): pass # show it inline
else: b['Content-Disposition']='attachment; filename=imapfix-preview'+str(accum[0])+'.'+office_convert
to_attach.append(b) ; accum[0] += 1
return True
def add_pdf(message):
global to_attach ; to_attach = None
accum = [1]
return walk_msg(message,add_pdf0,accum)
def add_pdf0(message,accum):
if to_attach == None: return False # TODO? (non-multipart message sent with a single pdf and nothing else)
fn = filename_ext(message)
if fn==False: return False
fn,ext = fn
if not ext==".pdf": return False
debug("Getting payload")
payload = message.get_payload(decode=True)
infile = "tmpdoc-%d.pdf" % (os.getpid(),)
outfile = "tmpdoc-%ds.html" % (os.getpid(),)
debug("Running pdftohtml")
if os.system("pdftohtml -i -enc UTF-8 %s" % (infile,)): # (-i = ignore images, TODO: allow images and also attach them? may need temporary directory. Also what if libreoffice creates images in its html in add_office0?) (Do not use -s for single page, it results in only the last PDF page being output, it does not mean combine all to a single HTML page, which happens anyway)
# conversion error
debug("pdftohtml run returned failure")
clean_tmpdoc() ; return False
b = email.mime.base.MIMEBase("text","html; charset=utf-8")
if not (max_size_of_first_part and len(b.get_payload) > max_size_of_first_part): pass # show it inline
else: b['Content-Disposition']='attachment; filename=imapfix-preview'+str(accum[0])+'.html'
to_attach.append(b) ; accum[0] += 1
return True
def clean_tmpdoc():
# soffice sometimes leaves images also, so check all
prefix = re.compile("tmpdoc-%d[^0-9]" % (os.getpid(),))
for f in os.listdir('.'):
if re.match(prefix,f): tryRm(f)
def add_tnef(message):
global to_attach ; to_attach = None
accum = [1]
return walk_msg(message,add_tnef0,accum)
def add_tnef0(message,accum):
if to_attach == None: return False # TODO? (non-multipart message sent with winmail.dat and nothing else?)
if not 'Content-Disposition' in message: return False
fn = str(message['Content-Disposition'])
if not 'winmail.dat' in fn.lower(): return False
ret = False
debug("Getting winmail.dat")
payload = message.get_payload(decode=True)
debug("Running tnef on winmail.dat")
outdir = "tnefout-%d" % (os.getpid(),)
os.popen("tnef -C "+outdir+" --number-backups --unix-paths --save-body --ignore-checksum --ignore-encode --ignore-cruft","wb").write(payload)
for f in listdir(outdir):
debug("Adding ",repr(f))
origF,f = f,outdir+os.sep+f
b = getMimeBase(f)
b['Content-Disposition']='attachment; filename='+origF
to_attach.append(b) ; accum[0] += 1 ; ret = True
try: os.rmdir(outdir)
except: debug("Could not remove ",outdir)
if ret: message.set_payload("") # no point keeping the winmail.dat itself if we successfully got its contents out
return ret
def getMimeBase(f):
mimeType = mimetypes.guess_type(f)[0]
if not mimeType: mimeType = "application/binary"
mT,subT = mimeType.split("/")
return email.mime.base.MIMEBase(mT,subT)
def folderList(pattern="*"):
typ,data = imap.list(pattern=pattern)
if not typ=='OK': return []
return [re.sub('.*"." ','',i).replace('"','') for i in data if i and not r"\Noselect" in i]
isoDate = "[1-9][0-9][0-9][0-9]-[0-1][0-9]-[0-3][0-9]"
if postponed_daynames:
# use the current locale's idea of the abbreviated names
weekdays = [time.strftime("%a",time.gmtime(time.mktime((2001,1,i,0,0,0,0,0,0)))).lower() for i in range(1,8)]
months = [time.strftime("%b",time.gmtime(time.mktime((2001,i,1,0,0,0,0,0,0)))).lower() for i in range(1,13)]
weekMonth = "("+"|".join(weekdays+months)+")(?=$|:)"
def isoToday(): return "%04d-%02d-%02d" % time.localtime()[:3]
def check_todays_dayname():
if postponed_daynames:
do_postponed_foldercheck(time.strftime("%a").lower()) # weekday
if time.localtime()[2]==1: # 1st of the month
do_postponed_foldercheck(time.strftime("%b").lower()) # month
def do_postponed_foldercheck(dayToCheck="today"):
today = isoToday()
if dayToCheck=="old": # called only on startup (and only if postponed_foldercheck)
if postponed_daynames and time.localtime()[3:5] < (0,30):
# If we restarted less than half an hour after midnight, check dayname folders on startup (because it might not have happened if we were in the middle of a 29-minute imap idle when the server rebooted)
# and regardless of what time of day we are doing this (re)start, check for dated folders of today and older
if postponed_maildir: mList=os.listdir(postponed_maildir)
else: mList = []
for f in folderList()+mList:
if re.match(isoDate+'$',f) and f <= today: # TODO: Y10K (lexicographic comparison)
elif dayToCheck=="today": # called at midnight rollover (if postponed_foldercheck or postponed_daynames)
if postponed_foldercheck: dayToCheck = today
else: return
if postponed_maildir:
try: maildir = get_maildir(postponed_maildir+os.sep+dayToCheck,False) # don't create if not exist
except: maildir = None
if maildir:
said = False ; toDel = []
for msgID,msg in maildir.iteritems():
if not said:
debug("Moving messages from maildir ",postponed_maildir+os.sep+dayToCheck," to ",filtered_inbox)
said = True
for msgID in toDel: del maildir[msgID]
f = folderList(dayToCheck)
if not len(f)==1: return # no folder of that name
folder = f[0] ; ; said = False
for msgID,flags,message in yield_all_messages():
if not said:
debug("Moving messages from ",folder," to ",filtered_inbox)
said = True
msg = email.message_from_string(message)
save_to(filtered_inbox,myAsString(msg)), '+FLAGS', '\\Deleted')
if said: check_ok(imap.expunge())
check_ok( ; do_delete(folder)
def reDate(msg):
theFrom = msg.get('From','').replace('"','').strip()
if theFrom==imapfix_name: # pre v1.5
del msg['From'] ; msg['From'] = imapfix_From_line # K-9 5.8+
old_date = msg.get("Date","")
if old_date:
if theFrom==imapfix_From_line: pass # no need to add old date if it's a --note or --multinote
elif authenticates(msg) and 'To' in msg and (username in msg['To'] or getAddr(msg['To'])==getAddr(msg.get('From',''))): pass # probably no need to add old date if it's a message from yourself to yourself (similar to --note/--multinote)
else: walk_msg(msg,addOldDateFunc(old_date))
del msg['Date']
msg['Date'] = email.utils.formatdate(localtime=True)
def getAddr(a): return ''.join(a.split()[-1:]).replace('<','').replace('>','')
def addOldDateFunc(old_date):
def addOldDate(message):
newPara = ""
if 'Content-Type' in message:
if not message["Content-Type"].startswith("text/"): return False
if message["Content-Type"].startswith("text/html"): newPara = "<p>" # TODO: might end up being before the HTML tag; depends which browser you use
if 'Content-Disposition' in message and message['Content-Disposition'].startswith('attachment'): return False
changed = globalise_charsets(message, imap_8bit) # just in case
dateIntro = imapfix_name+": original Date: "
if message.get_payload(decode=True).startswith(dateIntro): return changed
if not changed: globalise_charsets(message, imap_8bit, True) # ensure set up for:
return setPayload(message,dateIntro + old_date + newPara + "\n\n" + message.get_payload(decode=True),'utf-8') # Fixed in v1.498: postponing a message previously caused quopri to be decoded but still declared, invalidating URLs that had = followed by hex code in them, breaking links on commercial mailing-list trackers like MailChimp
return addOldDate
def do_calendar():
for l in os.popen("calendar -A 0 -f \"%s\"" % calendar_file):
if not l.strip(): continue
subj = l.split()[1:]
if not subj: continue
if re.match("^[0-9*]*$",subj[0]): subj=subj[1:] # month day txt
subj = ' '.join(l.split()[2:])
if len(subj)>60: subj=subj[:57]+"..."
save_to(filtered_inbox,"From: "+imapfix_From_line.replace(imapfix_name,"calendar",1)+"\r\nSubject: "+subj+"\r\nDate: "+email.utils.formatdate(localtime=True)+"\r\n\r\n"+l+"\n")
setAlarmAt = 0
def checkAlarmDelay():
global setAlarmAt
if time.time() > setAlarmAt:
import signal
setAlarmAt = time.time() + alarm_delay/2
def mainloop():
newDay = oldDay = time.localtime()[:3] # for midnight
done_spamprobe_cleanup_today = False
secondary_imap_due = 0
if exit_if_imapfix_config_py_changes:
if exit_if_imapfix_config_py_changes=="stamp":
try: os.utime("",None) # open with "a" doesn't always update timestamp
except: pass
mtime = os.stat("").st_mtime
if postponed_foldercheck or postponed_daynames: do_postponed_foldercheck("old")
while True:
if alarm_delay: checkAlarmDelay()
if maildirs_to_imap: do_maildirs_to_imap()
if maildir_to_copyself: do_maildir_to_copyself()
if copyself_alt_folder: do_copyself_to_copyself()
if auto_delete_folder: do_auto_delete()
if imap_to_maildirs: do_imap_to_maildirs()
global filtered_inbox
if filtered_inbox:
if time.time() > secondary_imap_due and secondary_imap_hostname:
fiO = filtered_inbox
if not filtered_inbox:
filtered_inbox="" # rather than None
filtered_inbox = fiO
secondary_imap_due = time.time() + secondary_imap_delay
if logout_before_sleep: make_sure_logged_out()
if sync_command: os.system(sync_command) # might make a separate login, hence after logout_before_sleep
if not poll_interval: break # --once
if not done_spamprobe_cleanup_today:
done_spamprobe_cleanup_today = True
if poll_interval=="idle":
make_sure_logged_in() ;
debug("Waiting for IMAP event") ; imap.idle()
# Can take a timeout parameter, default 29 mins. TODO: allow shorter timeouts for clients behind NAT boxes or otherwise needing more keepalive? IDLE can still be useful in these circumstances if the server's 'announce interval' is very short but we don't want across-network polling to be so short, e.g. slow link (however you probably don't want to be running imapfix over slow/wobbly links - it's better to run it on a well-connected server)
debug("Sleeping for ",poll_interval," seconds")
# TODO catch imap connection errors and re-open? or just put this whole process in a loop
newDay = time.localtime()[:3]
if not oldDay==newDay:
if midnight_command: os.system(midnight_command)
if calendar_file: do_calendar()
if postponed_foldercheck or postponed_daynames:
if train_spamprobe_nightly: do_nightly_train()
done_spamprobe_cleanup_today = False
if exit_if_imapfix_config_py_changes and not near_equal(mtime,os.stat("").st_mtime): break
finally: make_sure_logged_out()
def near_equal(time1,time2):
# allow small difference due to filesystem quirks
# (e.g. if the server caches file times at a higher
# resolution than the underlying filesystem, which can
# be the case in some Linux+NetWare setups)
if time1 > time2: time2,time1 = time1,time2
return (time2-time1) <= 5
if secondary_imap_hostname:
assert type(secondary_imap_hostname) == type(secondary_imap_username) == type(secondary_imap_password) and type(secondary_imap_hostname) in [str,list], "Invalid combination of types in secondary_imap settings"
if type(secondary_imap_hostname) == str:
secondary_imap_hostname = [secondary_imap_hostname]
secondary_imap_username = [secondary_imap_username]
secondary_imap_password = [secondary_imap_password]
else: assert len(secondary_imap_hostname) == len(secondary_imap_username) == len(secondary_imap_password), "secondary_imap lists have differing lengths"
def get_logged_in_imap(host,user,pwd,insecureFirst=False):
debug("Logging in to ",host)
if host in insecure_login: # expect no SSL
order = [imaplib.IMAP4, imaplib.IMAP4_SSL]
else: order = [imaplib.IMAP4_SSL,imaplib.IMAP4]
for Class in order:
if len(host.split(':'))==2:
host,port = host.split(':')
imap = Class(host, int(port))
else: imap = Class(host)
if type(pwd)==tuple: # OAuth2
cmd,secs = pwd
if oauth2_string_cache.setdefault(cmd,(None,0))[1] < time.time():
debug("Generating OAuth2 access string")
access_string = commands.getoutput(cmd).strip()
try: access_string = base64.decodestring(access_string)
except: pass # maybe it wasn't base64
oauth2_string_cache[cmd] = (access_string,time.time()+secs)
debug("Using OAuth2 access string") # if hangs here, refresh token might have been invalidated (e.g. in GMail 'testing-only apps' the refresh tokens are short-lived) and the server is delaying it after repeated use
check_ok(imap.authenticate('XOAUTH2', lambda _:oauth2_string_cache[cmd][0]))
else: check_ok(imap.login(user,pwd))
debug("Logged in")
return imap
except: pass
raise Exception("Could not log in to "+host)
oauth2_string_cache = {}
def process_secondary_imap():
global imap ; first=True
for sih,siu,sip in zip(secondary_imap_hostname, secondary_imap_username, secondary_imap_password):
if first_secondary_is_copy_only and first:
first = False ; continue
try: imap = get_logged_in_imap(sih,siu,sip)
msg = "Could not log in as %s to secondary IMAP %s: skipping it this time" % (siu,sih)
if report_secondary_login_failures:
imap = saveImap # for make_sure_logged_in
save_to(filtered_inbox,"From: "+imapfix_From_line+"\r\nSubject: imapfix_config secondary_imap problem or server down\r\nDate: %s\r\n\r\n%s\n" % (email.utils.formatdate(localtime=True),msg))
imap = None
if imap:
global additional_inbox
oAI = additional_inbox
if not check_additional_inbox_on_secondary_too:
additional_inbox = None
additional_inbox = oAI
if check_copyself_alt_folder_on_secondary_too:
imap = saveImap
def do_archive():
try: os.mkdir(archive_path)
except: pass # no error if exists
for foldername,age,action in archive_rules:
if not age==None: age = age*24*3600
if type(foldername)==tuple: bStr=foldername[1]
else: bStr = foldername
if os.sep in bStr:
assert bStr, "shouldn't end mbox with /"
archive(foldername, archive_path+os.sep+bStr, age, action)
def do_nightly_train():
for foldername,age,action in archive_rules:
nightly_train(foldername, action)
def yield_folders():
"iterates through folders in imap, selecting each one as it goes"
for foldername in imap.list()[1]:
if '"/"' in foldername: foldername=foldername[foldername.index('"/"')+3:].lstrip()
if foldername.startswith('"') and foldername.endswith('"'): foldername=foldername[1:-1] # TODO: check if any other unquoting is needed
typ, data =
if not typ=='OK': continue
yield foldername
def do_note(subject,ctype="text/plain",maybe=0):
subject = subject.strip()
if not subject: subject = "Note to self (via imapfix)"
if isatty(sys.stdin):
sys.stderr.write("Type the note, then EOF\n")
body =
if maybe and not body.strip(): return
if not body: body = " " # make sure there's at least one space in the message, for some clients that don't like empty body
if filtered_inbox==None: saveTo = ""
else: saveTo = filtered_inbox
save_to(saveTo,"From: "+imapfix_From_line+"\r\nSubject: "+utf8_to_header(subject)+"\r\nDate: "+email.utils.formatdate(localtime=True)+"\r\nMIME-Version: 1.0\r\nContent-type: "+ctype+"; charset=utf-8\r\n\r\n"+from_mangle(body)+"\n")
def from_mangle(body): return re.sub('(?<![^\n])From ','>From ',body) # (Not actually necessary for IMAP, but might be useful if the message is later processed by something that expects a Unix mailbox. Could MIME-encode instead, but not so convenient for editing.)
def upload(filelist):
for f in filelist:
if os.path.isdir(f): upload([(f+os.sep+g) for g in listdir(f)])
elif not os.path.isfile(f): debug("Ignoring non-file non-directory ",f)
elif not f.endswith('~') and do_upload(open(f,"rb").read(),os.stat(f).st_mtime,f): debug("Uploaded ",f)
def multinote(filelist,to_real_inbox,use_filename=False):
if not filtered_inbox: to_real_inbox = True
for f in filelist:
if os.path.isdir(f):
multinote([(f+os.sep+g) for g in listdir(f)],to_real_inbox,use_filename)
if not os.path.isfile(f):
debug("Ignoring non-file non-directory ",f)
if not f.endswith('~'):
if use_filename:
subj = f
if os.sep in subj: subj=subj[subj.rindex(os.sep)+1:]
else: subj = None
if do_multinote(open(f).read(),os.stat(f).st_mtime,to_real_inbox,subj): debug("Uploaded ",f)
def tryRm(f):
try: os.remove(f)
except: pass
def do_upload(data,theDate,fname):
b = getMimeBase(fname)
b['Content-Disposition']='attachment; filename='+(os.sep+fname)[(os.sep+fname).rindex(os.sep)+1:]
message = turn_into_attachment(b,"Attached "+fname,True)
message["From"] = imapfix_From_line
message["Subject"] = fname
message["Date"] = email.utils.formatdate(theDate,localtime=True)
return True
def do_multinote(body,theDate,to_real_inbox,subject):
body = re.sub("\r\n?","\n",body.strip())
if not body and not subject:
debug("Not creating message from blank file")
return False
if not subject: subject,body = (body+"\n").split("\n",1)
if to_real_inbox: box = ""
box,newSubj = authenticated_wrapper(subject,body)
if newSubj: subject = newSubj
if box==False: box=filtered_inbox
if not box==None: save_to(box,"From: "+imapfix_From_line+"\r\nSubject: "+utf8_to_header(subject)+"\r\nDate: "+email.utils.formatdate(theDate,localtime=True)+"\r\nMIME-Version: 1.0\r\nContent-type: text/plain; charset=utf-8\r\n\r\n"+from_mangle(body)+"\n")
return True
def isatty(f): return hasattr(f,"isatty") and f.isatty()
if quiet==2: quiet = not isatty(sys.stdout)
def do_delete(foldername):
foldername = foldername.strip()
if not foldername:
print ("No folder name specified")
print ("Deleting folder "+repr(foldername))
typ, data = imap.delete(foldername)
if not typ=='OK': # some folders can't be deleted, so just log
print("Ignoring failed folder delete: "+str(typ)+' '+repr(data))
def do_create(foldername):
foldername = foldername.strip()
if not foldername:
print ("No folder name specified")
print ("Creating folder "+repr(foldername))
def secondary_security(message_as_string):
oms = message_as_string
if imap_8bit and"(?i)Content-Transfer-Encoding: quoted-printable",message_as_string): message_as_string = quopri_to_u8_8bitOnly(message_as_string) # because saveTo will do this (same TODO as there) so normalise for comparing across servers
if secondary_is_insecure: message_as_string = re.sub(addr_regex,"",message_as_string[:secLimit])+message_as_string[secLimit:] # NB the substitute email MUST fit into the original pattern: in particular the top-level domain must be between 2 and 5 letters, otherwise reapplying this function will get different text (e.g. ".removeded" if using @email.removed)
if not message_as_string == oms: message_as_string = re.sub("\r\nContent-Length: [0-9]+\r\n","\r\n",message_as_string) # secondary IMAP will likely fix or delete this anyway
return message_as_string
def ccnl(c): return c+r'+(?:=\r?\n'+c+'*)?' # character class + maybe newline: for speed we limit the number of line breaks that occur in each ccnl to one
def cnl(c): return c+r'(?:=\r?\n)?' # char + opt newline
addr_regex = re.compile("".join([
# TODO: need to make this faster. (Or can we be more selective about which parts of the message get it i.e. not images etc)
'(?:'+cnl('[a-zA-Z]')+'){2,5}(?![a-zA-Z])'])) # TODO: this deals with the header easily, and with SOME quoted-printable-in-long-HTML-line situations, but should also rm from Base64 in body (if quoting) (+ email at very end of msg w. no trailing \n)
def do_backup():
for foldername in folderList():
fname = foldername.replace("/","-").replace('"','')+"-backup.mbox"
if os.path.exists(fname): # must create new
mbox = mailbox.mbox(fname)
print ("Backing up "+foldername+" to "+fname)
for msgID,flags,message in yield_all_messages():
msg = email.message_from_string(message)
if not message.startswith("From "): # as above
if 'From' in msg:
fr = msg['From']
if '<' in fr and '>' in fr[fr.index('<'):]: fr = fr[fr.index('<')+1:fr.rindex('>')]
if not fr: fr = 'unknown'
message="From "+fr+"\r\n"+message
msg = email.message_from_string(message)
newFlags = ""
if "\\seen" in flags.lower() and not "old" in flags.lower(): newFlags += "R"
if "\\answered