IMAPClient concepts

Message Identifiers

In the IMAP protocol, messages are identified using an integer. These message ids are specific to a given folder.

There are two types of message identifiers in the IMAP protocol.

One type is the message sequence number where the messages in a folder are numbered from 1 to N where N is the number of messages in the folder. These numbers don’t persist between sessions and may be reassigned after some operations such as an expunge.

A more convenient approach is Unique Identifiers (UIDs). Unique Identifiers are integers assigned to each message by the IMAP server that will persist across sessions. They do not change when folders are expunged. Almost all IMAP servers support UIDs.

Each call to the IMAP server can use either message sequence numbers or UIDs in the command arguments and return values. The client specifies to the server which type of identifier should be used. You can set whether IMAPClient should use UIDs or message sequence number via the use_uid argument passed when an IMAPClient instance is created and the use_uid attribute. The use_uid attribute can be used to change the message id type between calls to the server. IMAPClient uses UIDs by default.

Any method that accepts message ids takes either a sequence containing message ids (eg. [1,2,3]), or a single message id integer, or a string representing sets and ranges of messages as supported by the IMAP protocol (e.g. '50-65', '2:*' or '2,4:7,9,12:*').

Message Flags

An IMAP server keeps zero or more flags for each message. These indicate certain properties of the message or can be used by IMAP clients to keep track of data related to a message.

The IMAPClient package has constants for a number of commmonly used flags:

DELETED = br'\Deleted'
SEEN = br'\Seen'
ANSWERED = br'\Answered'
FLAGGED = br'\Flagged'
DRAFT = br'\Draft'
RECENT = br'\Recent'         # This flag is read-only

Any method that accepts message flags takes either a sequence containing message flags (eg. [DELETED, 'foo', 'Bar']) or a single message flag (eg. 'Foo').

Folder Name Encoding

Any method that takes a folder name will accept a standard string or a unicode string. Unicode strings will be transparently encoded using modified UTF-7 as specified by RFC 3501#section-5.1.3. This allows for arbitrary unicode characters (eg. non-English characters) to be used in folder names.

The ampersand character (“&”) has special meaning in IMAP folder names. IMAPClient automatically escapes and unescapes this character so that the caller doesn’t have to.

Automatic folder name encoding and decoding can be enabled or disabled with the folder_encode attribute. It defaults to True.

If folder_encode is True, all folder names returned by IMAPClient are always returned as unicode strings. If folder_encode is False, folder names are returned as str (Python 2) or bytes (Python 3).

TLS/SSL

IMAPClient uses sensible TLS parameter defaults for encrypted connections and also allows for a high level of control of TLS parameters if required. To provide a consistent API and capabilities across Python versions the backports.ssl library is used instead of the standard library ssl package. backports.ssl provides an API that aims to mimic the Python 3.4 ssl package so it should be familiar to developers that have used the ssl package in recent versions of Python.

TLS parameters are controlled by passing a backports.ssl.SSLContext when creating an IMAPClient instance. When ssl=True is used without passing a SSLContext, a default context is used. The default context avoids the use of known insecure ciphers and SSL protocol versions, with certificate verification and hostname verification turned on. The default context will use system installed certificate authority trust chains, if available.

IMAPClient.tls.create_default_context() returns IMAPClient’s default context. When constructing a custom context it is usually best to start with the default context and modify it to suit your needs.

The following example shows how to to disable certification verification and certificate host name checks if required.

# Copyright (c) 2015, Menno Smits
# Released subject to the New BSD License
# Please see http://en.wikipedia.org/wiki/BSD_licenses
#
# Establish an encrypted connection to a server without checking its
# certificate. This setup is insecure, DO NOT USE to connect to servers
# over the Internet.

import ssl

from imapclient import IMAPClient

HOST = 'imap.host.com'
USERNAME = 'someuser'
PASSWORD = 'secret'

ssl_context = ssl.create_default_context()

# don't check if certificate hostname doesn't match target hostname
ssl_context.check_hostname = False

# don't check if the certificate is trusted by a certificate authority
ssl_context.verify_mode = ssl.CERT_NONE

with IMAPClient(HOST, ssl_context=ssl_context) as server:
    server.login(USERNAME, PASSWORD)
    # ...do something...

The next example shows how to create a context that will use custom CA certificate. This is required to perform verification of a self-signed certificate used by the IMAP server.

# Copyright (c) 2015, Menno Smits
# Released subject to the New BSD License
# Please see http://en.wikipedia.org/wiki/BSD_licenses
#
# Establish a secure connection to a server that does not have a certificate
# signed by a trusted authority.

import ssl

from imapclient import IMAPClient

HOST = 'imap.host.com'
USERNAME = 'someuser'
PASSWORD = 'secret'

ssl_context = ssl.create_default_context(cafile="/path/to/cacert.pem")

with IMAPClient(HOST, ssl_context=ssl_context) as server:
    server.login(USERNAME, PASSWORD)
    # ...do something...

The above examples show some of the most common TLS parameter customisations but there are many other tweaks are possible. Consult the Python 3 ssl package documentation for further options.

Old pyOpenSSL Versions

IMAPClient’s TLS functionality will not behaviour correctly if an out-of-date version of pyOpenSSL is used. On some systems (particularly OS X) the system installed version of pyOpenSSL will take precedence over any user installed version. Use of virtualenvs is strongly encouraged to work around this.

IMAPClient checks the installed pyOpenSSL version at import time and will fail early if an old pyOpenSSL version is found.

Using gevent with IMAPClient

Some extra monkey patching is required so that the gevent package can work with pyOpenSSL (used by IMAPClient for TLS support). The gevent_openssl package performs this patching. Please use gevent_openssl 1.2 or later.

Here’s an example of how gevent_openssl can be used with IMAPClient:

from gevent import monkey; monkey.patch_all()
import gevent_openssl; gevent_openssl.monkey_patch()

import imapclient

client = imapclient.IMAPClient(...)
...

Logging

IMAPClient logs debug lines using the standard Python logging module. Its logger is imapclient.*.

A simple way to display log messages is to setup logging:

import logging

logging.basicConfig(
    format='%(asctime)s - %(levelname)s: %(message)s',
    level=logging.DEBUG
)

For advanced usage please refer to the Python documentation.