iredmail-doc/html_bk/dovecot.imapsieve.html

399 lines
19 KiB
HTML

<!DOCTYPE html>
<html>
<head>
<meta http-equiv="Content-Type" content="text/html; charset=utf-8" />
<title>Auto learn spam/ham with Dovecot `imap_sieve` plugin</title>
<link rel="stylesheet" type="text/css" href="./css/markdown.css" />
</head>
<body>
<div id="navigation">
<a href="https://www.iredmail.org" target="_blank">
<img alt="iRedMail web site"
src="./images/logo-iredmail.png"
style="vertical-align: middle; height: 30px;"
/>&nbsp;
<span>iRedMail</span>
</a>
&nbsp;&nbsp;//&nbsp;&nbsp;<a href="./index.html">Document Index</a></div><h1 id="auto-learn-spamham-with-dovecot-imap_sieve-plugin">Auto learn spam/ham with Dovecot <code>imap_sieve</code> plugin</h1>
<div class="toc">
<ul>
<li><a href="#auto-learn-spamham-with-dovecot-imap_sieve-plugin">Auto learn spam/ham with Dovecot imap_sieve plugin</a><ul>
<li><a href="#summary">Summary</a></li>
<li><a href="#requirements">Requirements</a></li>
<li><a href="#enable-imap_sieve-plugin">Enable imap_sieve plugin</a></li>
<li><a href="#create-required-directories-and-files">Create required directories and files</a></li>
<li><a href="#setup-cron-job-to-scan-and-learn-spamham-messages">Setup cron job to scan and learn spam/ham messages</a></li>
<li><a href="#tests">Tests</a><ul>
<li><a href="#report-spam-move-email-from-inbox-to-junk">Report spam: Move email from Inbox to Junk</a></li>
<li><a href="#report-ham-move-email-from-junk-to-any-other-folder-except-trash">Report ham: Move email from Junk to any other folder (except Trash)</a></li>
<li><a href="#scan-reported-mails">Scan reported mails</a></li>
<li><a href="#check-detailed-bayes-learning-log-on-command-line">Check detailed bayes learning log on command line</a></li>
</ul>
</li>
<li><a href="#check-bayes-data">Check bayes data</a></li>
<li><a href="#see-also">See also</a></li>
<li><a href="#references">References</a></li>
</ul>
</li>
</ul>
</div>
<div class="admonition attention">
<p class="admonition-title">Attention</p>
<p>This feature is enabled by default if your iRedMail server was deployed
with our <a href="https://www.iredmail.org/easy.html">iRedMail Easy platform</a>.</p>
</div>
<div class="admonition warning">
<p class="admonition-title">Warning</p>
<p>The bayesian classifier can only score new messages after it already learn
200 known spams and 200 known hams.</p>
</div>
<h2 id="summary">Summary</h2>
<p>Dovecot offers plugin <code>imap_sieve</code> to run sieve script for spam/virus scanning,
it's useful to let end users report spam/ham messages within webmail or MUA,
then on server side we call SpamAssassin to learn the reported messages. The
more spams/hams end users reported, the more precisely SpamAssassin can catch
the spams.</p>
<p>This tutorial shows you how to enable Dovecot plugin <code>imap_sieve</code> and create
required shell/sieve scripts to learn spams automatically.</p>
<p>After setup, you can encourage end users to report spam messages by
moving/dragging spam to <code>Junk</code> folder. With more spams reported, your iRedMail
server can precisely catch more spams.</p>
<h2 id="requirements">Requirements</h2>
<ul>
<li>A working iRedMail server.</li>
<li>Dovecot version 2.2.24 or later. <code>imap_sieve</code> plugin is available in version
2.2.24 and later releases.<ul>
<li>CentOS 7 ships Dovecot-2.2.36</li>
<li>Debian 9 ships Dovecot-2.2.27</li>
<li>Ubuntu 16.04 ships Dovecot-2.2.22 (<strong>WARNING: Not qualified</strong>)</li>
<li>Ubuntu 18.04 ships Dovecot-2.2.33</li>
<li>OpenBSD 6.4 ships Dovecot-2.2.36</li>
<li>FreeBSD ships Dovecot-3.x in ports tree</li>
</ul>
</li>
</ul>
<h2 id="enable-imap_sieve-plugin">Enable <code>imap_sieve</code> plugin</h2>
<p>Please update Dovecot config file <code>/etc/dovecot/dovecot.conf</code> to:</p>
<ul>
<li>Enable new parameter <code>mail_attribute_dict</code> globally.</li>
<li>Enable new plugin <code>imap_sieve</code> in <code>protocol imap {}</code> section.</li>
<li>Add required settings for <code>imap_sieve</code> in <code>plugin {}</code> section.</li>
</ul>
<pre><code># Store METADATA information within user's HOME directory
mail_attribute_dict = file:%Lh/dovecot-attributes
protocol imap {
...
mail_plugins = ... imap_sieve
}
plugin {
sieve_plugins = sieve_imapsieve sieve_extprograms
imapsieve_url = sieve://127.0.0.1:4190
# From elsewhere to Junk folder
imapsieve_mailbox1_name = Junk
imapsieve_mailbox1_causes = COPY APPEND
imapsieve_mailbox1_before = file:/var/vmail/sieve/report_spam.sieve
# From Junk folder to elsewhere
imapsieve_mailbox2_name = *
imapsieve_mailbox2_from = Junk
imapsieve_mailbox2_causes = COPY
imapsieve_mailbox2_before = file:/var/vmail/sieve/report_ham.sieve
sieve_pipe_bin_dir = /etc/dovecot/sieve/pipe
sieve_global_extensions = +vnd.dovecot.pipe +vnd.dovecot.environment
}
</code></pre>
<h2 id="create-required-directories-and-files">Create required directories and files</h2>
<p>We will create few directories and files used by <code>imap_sieve</code> plugin:</p>
<ul>
<li>Directories:<ul>
<li><code>/etc/dovecot/sieve/pipe</code>: used to store script called by <code>imap_sieve</code> plugin.</li>
<li><code>/var/vmail/imapsieve_copy</code>: used to store reported spam/ham emails.</li>
</ul>
</li>
<li>Files:<ul>
<li><code>/var/vmail/sieve/report_spam.sieve</code>: used to save a copy of reported spam.</li>
<li><code>/var/vmail/sieve/report_ham.sieve</code>: used to save a copy of reported ham.</li>
</ul>
</li>
<li>Shell script:<ul>
<li><code>/etc/dovecot/sieve/pipe/imapsieve_copy</code></li>
</ul>
</li>
</ul>
<p>Create directories:</p>
<pre><code>mkdir -p /etc/dovecot/sieve/pipe
mkdir -p /var/vmail/imapsieve_copy
chown vmail:vmail /var/vmail/imapsieve_copy
chmod 0700 /var/vmail/imapsieve_copy
</code></pre>
<p>Create file <code>/var/vmail/sieve/report_spam.sieve</code> with content below:</p>
<pre><code>require [&quot;vnd.dovecot.pipe&quot;, &quot;copy&quot;, &quot;imapsieve&quot;, &quot;environment&quot;, &quot;variables&quot;];
if environment :matches &quot;imap.user&quot; &quot;*&quot; {
set &quot;username&quot; &quot;${1}&quot;;
}
pipe :copy &quot;imapsieve_copy&quot; [ &quot;${username}&quot;, &quot;spam&quot; ];
</code></pre>
<p>Create file <code>/var/vmail/sieve/report_ham.sieve</code> with content below:</p>
<pre><code>require [&quot;vnd.dovecot.pipe&quot;, &quot;copy&quot;, &quot;imapsieve&quot;, &quot;environment&quot;, &quot;variables&quot;];
if environment :matches &quot;imap.mailbox&quot; &quot;*&quot; {
set &quot;mailbox&quot; &quot;${1}&quot;;
}
if string &quot;${mailbox}&quot; &quot;Trash&quot; {
stop;
}
if environment :matches &quot;imap.user&quot; &quot;*&quot; {
set &quot;username&quot; &quot;${1}&quot;;
}
pipe :copy &quot;imapsieve_copy&quot; [ &quot;${username}&quot;, &quot;ham&quot; ];
</code></pre>
<p>Create file <code>/etc/dovecot/sieve/pipe/imapsieve_copy</code> with content below:</p>
<pre><code>#!/usr/bin/env bash
# Author: Zhang Huangbin &lt;zhb@iredmail.org&gt;
# Purpose: Read full email message from stdin, and save to a local file.
# Usage: bash imapsieve_copy &lt;email&gt; &lt;spam|ham&gt; &lt;output_base_dir&gt;
export USER=&quot;$1&quot;
export MSG_TYPE=&quot;$2&quot;
export OUTPUT_BASE_DIR=&quot;/var/vmail/imapsieve_copy&quot;
export OUTPUT_DIR=&quot;${OUTPUT_BASE_DIR}/${MSG_TYPE}&quot;
export FILE=&quot;${OUTPUT_DIR}/${USER}-$(date +%Y%m%d%H%M%S)-${RANDOM}${RANDOM}.eml&quot;
export OWNER=&quot;vmail&quot;
export GROUP=&quot;vmail&quot;
for dir in &quot;${OUTPUT_BASE_DIR}&quot; &quot;${OUTPUT_DIR}&quot;; do
if [[ ! -d ${dir} ]]; then
mkdir -p ${dir}
chown ${OWNER}:${GROUP} ${dir}
chmod 0700 ${dir}
fi
done
cat &gt; ${FILE} &lt; /dev/stdin
# Logging
#export LOG='logger -p local5.info -t imapsieve_copy'
#[[ $? == 0 ]] &amp;&amp; ${LOG} &quot;Copied one ${MSG_TYPE} email reported by ${USER}: ${FILE}&quot;
</code></pre>
<p>Set correct file owner and permissions:</p>
<pre><code>chown vmail:vmail /var/vmail/sieve/report_spam.sieve \
/var/vmail/sieve/report_ham.sieve \
/etc/dovecot/sieve/pipe/imapsieve_copy
chmod 0700 /var/vmail/sieve/report_spam.sieve \
/var/vmail/sieve/report_ham.sieve \
/etc/dovecot/sieve/pipe/imapsieve_copy
</code></pre>
<p>Restart Dovecot service to enable this plugin.</p>
<pre><code>service dovecot restart
</code></pre>
<h2 id="setup-cron-job-to-scan-and-learn-spamham-messages">Setup cron job to scan and learn spam/ham messages</h2>
<p>Dovecot can now save a copy of reported spam/ham automatically, we still need
a shell script to call SpamAssassin to actually learn spam/ham periodly.</p>
<p>Create script <code>/etc/dovecot/sieve/scan_reported_mails.sh</code> with content below,
it's used to call <code>sa-learn</code> command to learn reported spam/ham emails:</p>
<div class="admonition attention">
<p class="admonition-title">Attention</p>
<p>If you're running FreeBSD or OpenBSD, please change the Amavisd daemon
user name in variable <code>AMAVISD_USER</code> below.</p>
</div>
<pre><code>#!/usr/bin/env bash
# Author: Zhang Huangbin &lt;zhb@iredmail.org&gt;
# Purpose: Copy spam/ham to another directory and call sa-learn to learn.
# Paths to find program.
export PATH=&quot;/bin:/usr/bin:/usr/local/bin:$PATH&quot;
export OWNER=&quot;vmail&quot;
export GROUP=&quot;vmail&quot;
# The Amavisd daemon user.
# Note: on OpenBSD, it's &quot;_vscan&quot;. On FreeBSD, it's &quot;vscan&quot;.
export AMAVISD_USER='amavis'
export AMAVISD_USER_HOMEDIR=&quot;$(eval echo ~${AMAVISD_USER})&quot;
# Kernel name, in upper cases.
export KERNEL_NAME=&quot;$(uname -s | tr '[a-z]' '[A-Z]')&quot;
# A temporary lock file. should be removed after successfully examed messages.
export LOCK_FILE='/tmp/scan_reported_mails.lock'
# Logging to syslog with 'logger' command.
export LOG='logger -p local5.info -t scan_reported_mails'
# `sa-learn` command, with optional arguments.
export SA_LEARN=&quot;sa-learn -u ${AMAVISD_USER} --dbpath ${AMAVISD_USER_HOMEDIR}/.spamassassin&quot;
# Spool directory.
# Must be owned by vmail:vmail.
export SPOOL_DIR='/var/vmail/imapsieve_copy'
# Directories which store spam and ham emails.
# These 2 should be created while setup Dovecot antispam plugin.
export SPOOL_SPAM_DIR=&quot;${SPOOL_DIR}/spam&quot;
export SPOOL_HAM_DIR=&quot;${SPOOL_DIR}/ham&quot;
# Directory used to store emails we're going to process.
# We will copy new spam/ham messages to these directories, scan them, then
# remove them.
export SPOOL_LEARN_SPAM_DIR=&quot;${SPOOL_DIR}/processing/spam&quot;
export SPOOL_LEARN_HAM_DIR=&quot;${SPOOL_DIR}/processing/ham&quot;
if [ -e ${LOCK_FILE} ]; then
find $(dirname ${LOCK_FILE}) -maxdepth 1 -ctime 1 &quot;$(basename ${LOCK_FILE})&quot; &gt;/dev/null 2&gt;&amp;1
if [ X&quot;$?&quot; == X'0' ]; then
rm -f ${LOCK_FILE} &gt;/dev/null 2&gt;&amp;1
else
${LOG} &quot;Lock file exists (${LOCK_FILE}), abort.&quot;
exit
fi
fi
for dir in &quot;${SPOOL_DIR}&quot; &quot;${SPOOL_LEARN_SPAM_DIR}&quot; &quot;${SPOOL_LEARN_HAM_DIR}&quot;; do
if [[ ! -d ${dir} ]]; then
mkdir -p ${dir}
fi
chown ${OWNER}:${GROUP} ${dir}
chmod 0700 ${dir}
done
# If there're a lot files, direct `mv` command may fail with error like
# `argument list too long`, so we need `find` in this case.
if [[ X&quot;${KERNEL_NAME}&quot; == X'OPENBSD' ]] || [[ X&quot;${KERNEL_NAME}&quot; == X'FREEBSD' ]]; then
[[ -d ${SPOOL_SPAM_DIR} ]] &amp;&amp; find ${SPOOL_SPAM_DIR} -name '*.eml' -exec mv {} ${SPOOL_LEARN_SPAM_DIR}/ \;
[[ -d ${SPOOL_HAM_DIR} ]] &amp;&amp; find ${SPOOL_HAM_DIR} -name '*.eml' -exec mv {} ${SPOOL_LEARN_HAM_DIR}/ \;
else
[[ -d ${SPOOL_SPAM_DIR} ]] &amp;&amp; find ${SPOOL_SPAM_DIR} -name '*.eml' -exec mv -t ${SPOOL_LEARN_SPAM_DIR}/ {} +
[[ -d ${SPOOL_HAM_DIR} ]] &amp;&amp; find ${SPOOL_HAM_DIR} -name '*.eml' -exec mv -t ${SPOOL_LEARN_HAM_DIR}/ {} +
fi
# Try to delete empty directory, if failed, that means we have some messages to
# scan.
rmdir ${SPOOL_LEARN_SPAM_DIR} &amp;&gt;/dev/null
if [[ X&quot;$?&quot; != X'0' ]]; then
output=&quot;$(${SA_LEARN} --spam ${SPOOL_LEARN_SPAM_DIR})&quot;
rm -rf ${SPOOL_LEARN_SPAM_DIR} &amp;&gt;/dev/null
${LOG} '[SPAM]' ${output}
fi
rmdir ${SPOOL_LEARN_HAM_DIR} &amp;&gt;/dev/null
if [[ X&quot;$?&quot; != X'0' ]]; then
output=&quot;$(${SA_LEARN} --ham ${SPOOL_LEARN_HAM_DIR})&quot;
rm -rf ${SPOOL_LEARN_HAM_DIR} &amp;&gt;/dev/null
${LOG} '[CLEAN]' ${output}
fi
rm -f ${LOCK_FILE} &amp;&gt;/dev/null
</code></pre>
<p>Run command <code>crontab -e -u root</code> to setup cron job for root user, scan emails
every 10 minutes:</p>
<pre><code># iRedMail: Scan reported mails.
*/10 * * * * /bin/bash /etc/dovecot/sieve/scan_reported_mails.sh
</code></pre>
<h2 id="tests">Tests</h2>
<h3 id="report-spam-move-email-from-inbox-to-junk">Report spam: Move email from Inbox to Junk</h3>
<div class="admonition attention">
<p class="admonition-title">Attention</p>
<p>If you're running Roundcube webmail, you can enable its plugin <code>markasjunk</code>
to help move spam to Junk folder with one click.</p>
</div>
<ul>
<li>Login to webmail or any IMAP client like Outlook/Thunderbird.</li>
<li>Move an email from <code>Inbox</code> folder to <code>Junk</code> folder.</li>
</ul>
<p>In Dovecot log file <code>/var/log/dovecot/imap.log</code> (or <code>dovecot.log</code> if you didn't
configure syslog daemon to separate log content), you should see log lines like
below:</p>
<pre><code>Jan 31 21:10:42 c7 dovecot: imap(&lt;email&gt;): sieve: pipe action: piped message to program `imapsieve_copy'
Jan 31 21:10:42 c7 dovecot: imap(&lt;email&gt;): sieve: left message in mailbox 'Junk'
Jan 31 21:10:42 c7 dovecot: imap(&lt;email&gt;): expunge: box=INBOX, uid=7, msgid=, size=7805, from=&lt;email&gt;, subject=&lt;subject&gt;
</code></pre>
<p>In the meantime, you should see an email in <code>/var/vmail/imapsieve_copy/spam/</code>,
file name in <code>&lt;email&gt;-&lt;timestamp&gt;-&lt;random_number&gt;.eml</code> format.</p>
<h3 id="report-ham-move-email-from-junk-to-any-other-folder-except-trash">Report ham: Move email from Junk to any other folder (except <code>Trash</code>)</h3>
<p>If you found a clean email in <code>Junk</code> folder, just move it from <code>Junk</code> to any
other folder except <code>Trash</code>.</p>
<p>In Dovecot log file <code>/var/log/dovecot/imap.log</code> (or <code>dovecot.log</code>), you should
see log lines like below:</p>
<pre><code>Jan 31 21:15:51 c7 dovecot: imap(&lt;email&gt;): sieve: pipe action: piped message to program `imapsieve_copy'
Jan 31 21:15:51 c7 dovecot: imap(&lt;email&gt;): sieve: left message in mailbox 'INBOX'
Jan 31 21:15:51 c7 dovecot: imap(&lt;email&gt;): expunge: box=Junk, uid=7, msgid=, size=7805, from=&lt;email&gt;, subject=&lt;subject&gt;
</code></pre>
<p>In the meantime, you should see an email in <code>/var/vmail/imapsieve_copy/ham/</code>,
file name in <code>&lt;email&gt;-&lt;timestamp&gt;-&lt;random_number&gt;.eml</code> format.</p>
<h3 id="scan-reported-mails">Scan reported mails</h3>
<p>It's ok to run the script manually to scan reported mails:</p>
<pre><code>bash /etc/dovecot/sieve/scan_reported_mails.sh
</code></pre>
<p>If it scanned messages, it will log a message in <code>/var/log/syslog</code> or
<code>/var/log/messages</code> like this:</p>
<pre><code>Jan 31 04:51:34 mail scan_reported_mails: [CLEAN] Learned tokens from 1 message(s) (1 message(s) examined)
Jan 31 05:03:16 mail scan_reported_mails: [SPAM] Learned tokens from 1 message(s) (1 message(s) examined)
</code></pre>
<h3 id="check-detailed-bayes-learning-log-on-command-line">Check detailed bayes learning log on command line</h3>
<p>You can either <a href="./debug.amavisd.html">turn on debug mode in Amavisd and SpamAssassin</a>
to check how bayes learning works in SpamAssassin, or run <code>sa-learn</code> manually
to check it with a sample email.</p>
<p>To check on command line, please upload/save a sample email to
<code>/opt/sample.eml</code>, then run <code>sa-learn</code> as root user:</p>
<pre><code># su -s /bin/bash - amavis -c &quot;spamassassin -D bayes &lt; /opt/sample.eml&quot;
May 21 05:27:08.244 [32241] dbg: bayes: learner_new self=Mail::SpamAssassin::Plugin::Bayes=HASH(0x2fe8cb8), bayes_store_module=Mail::SpamAssassin::BayesStore::DBM
May 21 05:27:08.264 [32241] dbg: bayes: using username: amavis
May 21 05:27:08.264 [32241] dbg: bayes: learner_new: got store=Mail::SpamAssassin::BayesStore::DBM=HASH(0x387a1c8)
M
...
</code></pre>
<h2 id="check-bayes-data">Check bayes data</h2>
<p>Run <code>sa-learn</code> as Amavisd daemon user with <code>--dump</code> argument will show the bayes data like below:</p>
<pre><code># su -s /bin/bash amavis -c &quot;sa-learn --dump magic&quot;
0.000 0 3 0 non-token data: bayes db version
0.000 0 3778575 0 non-token data: nspam
0.000 0 6326326 0 non-token data: nham
0.000 0 539978 0 non-token data: ntokens
0.000 0 1558372204 0 non-token data: oldest atime
0.000 0 1558415857 0 non-token data: newest atime
0.000 0 0 0 non-token data: last journal sync atime
0.000 0 1558415403 0 non-token data: last expiry atime
0.000 0 43200 0 non-token data: last expire atime delta
0.000 0 59325 0 non-token data: last expire reduction count
</code></pre>
<ul>
<li><code>nspam</code> means number of learnt spams.</li>
<li><code>nham</code> means number of learnt ham/clean emails.</li>
</ul>
<h2 id="see-also">See also</h2>
<ul>
<li><a href="./store.spamassassin.bayes.in.sql.html">Store SpamAssassin bayes in SQL</a></li>
</ul>
<h2 id="references">References</h2>
<ul>
<li>
<p><a href="https://wiki.dovecot.org/HowTo/AntispamWithSieve">Dovecot wiki: Antispam with Sieve</a></p>
<p>You may notice a difference between current tutorial and Dovecot wiki
tutorial: our setup saves reported mails and scan it later with <code>sa-learn</code>
by cron job, but setup in Dovecot wiki calls <code>sa-learn</code> directly. We make
this change due to performance issue: when user moves a message to <code>Junk</code>
folder, webmail will wait for <code>sa-learn</code> to finish the scan then return
responsive, but if user moves a log messages at the same time, webmail will
hang and user have to wait there. This is not good user experience.</p>
</li>
</ul><div class="footer">
<p style="text-align: center; color: grey;">All documents are available in <a href="https://github.com/iredmail/docs/">GitHub repository</a>, and published under <a href="http://creativecommons.org/licenses/by-nd/3.0/us/" target="_blank">Creative Commons</a> license. You can <a href="https://github.com/iredmail/docs/archive/master.zip">download the latest version</a> for offline reading. If you found something wrong, please do <a href="https://www.iredmail.org/contact.html">contact us</a> to fix it.</p>
</div></body></html>