as the linux kernel mailing list is really noisy, I want to discard all mails which are send to my mailbox from the LKML, but which are not from me, to me or as answer to one of my mails. I do already do some filtering and redirect all patch-mails (including [PATCH at the beginning of the subject) to another inbox as the "normal" LKML mails. But it is still much too much.
How to do this with procmail?
What I have atm:
:0
* ^Sender:\ linux-kernel-owner#vger\.kernel\.org
* (PATCH|patch)
$MAILDIR/ml.kernel_org.linux-kernel.patches/
The real challenge here is to articulate how to reliably identify messages which are replies to something you wrote. Are you satisfied with excluding messages which are To: or Cc: yourself?
:0
* ^Sender: linux-kernel-owner#vger\.kernel\.org\>
* ! ^From: Your Self <you#example\.net>
* ! ^TO_you#example\.net\>
/dev/null
(obviously, edit the addresses to match what your mail client really puts there).
Or perhaps you have a vanity domain, in which case (properly constructed) replies will have an easily identifiable Message-Id of yours at the start of References:?
:0
* ^Sender: linux-kernel-owner#vger\.kernel\.org\>
* ! ^From: Your Self <you#example\.net>
* ! ^TO_you#example\.net\>
* ! ^References:[ ]*<[^<>#]*#yourdomain\.example\.net>
/dev/null
(the whitespace inside the square brackets should be a tab and a space).
Or you could expand that a bit to look for your domain anywhere in References:, to also include replies to replies to yourself, if you want that.
Or you could keep a local copy of all your outgoing message-id:s and look for them in References:, but that is already a significant endeavor which I will only point out as a possibility if you cannot use any of the above. (I do believe it has been hashed out in more detail before, perhaps on the Procmail mailing list.)
As an aside, I would change the "patch" rule to only examine the Subject: line. A match on "patch" in any other header is extremely likely to be a false positive. If you want to examine the body, you need extra flags, perhaps like this:
:0
* ^Sender: linux-kernel-owner#vger\.kernel\.org\>
{
:0
* ! B ?? \<patch\>
* ! ^Subject:(.*\<)?patch\>
{ } # empty "then", just so we can continue to "else"
:0E
$MAILDIR/ml.kernel_org.linux-kernel.patches/
# While we are inside these braces, let's continue with other LKML stuff
:0
* ! ^From: Your Self <you#example\.net>
* ! ^TO_you#example\.net
/dev/null
# Any additional LKML recipes? Add them here
# Anything which falls through here is regular LKML
:0
$MAILDIR/ml.kernel_org.linux-kernel/
}
(This can obviously be refactored in a number of different ways. Remember De Morgan's laws: NOT (A OR B)<=> NOT A AND NOT B.)
As a safety measure, you might want to look for messages which actually carry a patch as an attachment, rather than filter the discussion about such messages? That can also become quite complex, because there is a number of different ways to represent a patch as a MIME attachment (and some are also sent completely in-line, in a regular text/plain part amongst other text) but that isn't insurmountable, either, just significant drudgery.
Related
I would like to store e-mails sent by me to others using the "From" field to a folder called "/sent"
So, I use:
:0:
*^From.*user#outlook.com
$HOME/Mail/sent/.
And it used to work fine. However, now I am also forwarding my e-mail from the address:user#outlook.com, what is happending is that the e-mail envelope contains the header: Resent-From.*user#outlook.com so all forwarded e-mail is being saved to the sent folder.
Is it possible to have a double condition. That is something that says that if both Resent-From and From have *user#outlook.com, then it should go to the sent-folder. In other words, is it possible to use a AND or OR or Negation condition.
Update: The provided solution is correct. I was making an error in that I had neglected the ":". Thanks for both the solution and the patience. I also learnt a number of things and about a number of resources, for which also I am grateful,
Thanks!
Procmail by default does exactly what you ask. Multiple conditions will be ANDed together. The first one which fails causes the recipe to be abandoned; if they all succeed, the action is taken.
:0 # second colon removed - don't use lock on directories
* ^From:(.*\<)?user#outlook\.com
* ^Resent-From:(.*\<)?user#outlook\.com
$HOME/Mail/sent/.
Notice also how I modified your regex to tighten it up.
To do NOT, add a ! in front of the condition. To do OR, you can negate both conditions, and take the action in the "else" part (de Morgan's law).
:0
* ! First condition
* ! Other condition
{ } # do nothing
:0E # else, meaning at least one was true
action
Of course, if both conditions are regular expressions, you can simply use the regex or operator.
:0
* First condition|Other condition
action
newbie asking first question :)
I'm running a mail server (Ubuntu/Postfix/Dovecot) with SpamAssassin. Most of the known spam is flagged (RBLs, and obvious UCE) except for this particular malspam in attached zip files like "order_info_654321.zip", "paymet_document_123456.zip", and so on, when it doesn't fit any other SA rules. I'd like to procure a rule which drops the matching offenders into oblivion.
After fiddling with regex101.com, I've come up with an expression that matches these patterns exclusively:
/\w+[_][0-9]{6}.zip$/img
Question is... How to format it all, get it to work, and where to put it? So far, I edited /etc/spamassassin/local.cf, added this to the bottom, and restarted:
mimeheader TROJAN_ATTACHED Content-Type =~ /\w+[_][0-9]{6}.zip$/img
describe ZIP_ATTACHED email contains a zip trojan attachment
score TROJAN_ATTACHED 99.
But it doesn't seem to do the magic. Where else can I look for this?
Thank you all,
Keijo.-
You have a wrong regex. You do not need a $ char at the end, because filename strings are not necessarily at the end of the Content-Type header. Instead, you can use a word boundary \b anchor. In my rules, I have the following, and it perfectly works:
mimeheader MIME_FAIL Content-Type =~ /\.(ade|adp|bat|chm|cmd|com|cpl|exe|hta|ins|isp|jse|lib|lnk|mde|msc|msp|mst|pif|scr|sct|shb|sys|vb|vbe|vbs|vxd|wsc|wsf|wsh|reg)\b/i
describe MIME_FAIL Blacklisted file extension detected
score MIME_FAIL 5
First up, SA doesn't drop e-mails by default, but it can score them so high on spam content that they don't show up to anyone's inbox. Second, the "ingredients" I started with were incorrect, plus messed up with SA ability to function at all.
This actually did the trick when added into/etc/spamassassin/local.cf:
full TROJAN_ZIPUNDS /\w*[_][\d]{1,6}\.zip/img
score TROJAN_ZIPUNDS 99
describe TROJAN_ZIPUNDS RM zip attached trojan underscore
Even though these spammers altered from zip to rar, to underscores to dashes, different filenames, and so on, creating rules to counter them became simple after succeeding with the first one. Here's what I added too:
full TROJAN_RARDASH /\w*[-][\d]{1,6}\.rar/img
score TROJAN_RARDASH 99
describe TROJAN_RARDASH RM rar attached trojan dash
Also, as first described, I needed to specifically block certain zip file names which soon morphed to rar and dashes, so, morphing the regex and appending as a rule triad to spamassassin's local.cf (and restarting) is currently holding, until next spam wave :-)
Finally, this is a very very blunt workaround, so anyone with expertise on the subject is more than welcome to chime in.
You are using the wrong mime header to check for the filename. Use this instead:
mimeheader TROJAN_ATTACHED Content-Disposition =~ /\w+[_][0-9]{6}.zip/img
Also make sure you have the MimeHeader plugin loaded.
loadplugin Mail::SpamAssassin::Plugin::MIMEHeader
I'm trying to use procmail to send emails to a PHP script so the script will check a MySQL database and edit the subject line based on the sender email. I believe I've got a working procmail to do this:
:0:
* ^To:.*#barrett.com
! '/usr/local/bin/php-5.2 -f $HOME/ticket/emailcustcheck.php'
However, I'm not sure exactly how procmail executes the command. How does the email get passed to the PHP script, and therefore, how do I refer to it inside the script?
The correct syntax for piping to a script is
:0 # no lock file
* ^To:.*#barrett\.com
| /usr/local/bin/php-5.2 -f $HOME/ticket/emailcustcheck.php # no quotes, use pipe
The ! action would attempt to forward to an email address, but of course, the long quoted string with the path to your PHP interpreter is not a valid email address.
If you need locking (i.e. no two instances of this PHP script are allowed to run at the same time), you need to name a lock file; Procmail cannot infer a lock file name here, so the lock action you had would only produce an error message anyway. If you are uncertain, adding a named lock file is the safer bet, but if you don't have concurrency issues (such as, the script needs to write to a database while no other process is using the database) it should not be necessary, and could potentially slow down processing.
The condition regex also looks somewhat imprecise, but I can only speculate that you might want to trigger on Cc mail as well as direct To:. Look up the ^TO_ macro in the documentation if so.
The script gets the message as its standard input; it should probably read all input lines to an array, or split into two arrays so that everything before the first empty line goes into the "headers" array and the rest goes into the "body" array. Or perhaps PHP has some class which can read an email message into an object from standard input.
:0 wf
* ^To:.*#barrett\.com
| /usr/local/bin/php-5.2 -f $HOME/ticket/emailcustcheck.php
The f tells procmail that you are going to filter the message ie change it.
The w Wait for the filter or program to finish and check its exitcode.
If you want to work only on the body of the message you must add the flag b
If you want to work only on the header of the message you must add the flag h
We receive regular automated build messages from Jenkins build servers at work.
It'd be nice to ferret these away into a label, skipping the inbox.
Using a filter is of course the right choice.
The desired identifier is the string [RELEASE] at the beginning of a subject line.
Attempting to specify any of the following regexes causes emails with the string release in any case anywhere in the subject line to be matched:
\[RELEASE\]*
^\[RELEASE\]
^\[RELEASE\]*
^\[RELEASE\].*
From what I've read subsequently, Gmail doesn't have standard regex support, and from experimentation it seems, as with google search, special characters are simply ignored.
I'm therefore looking for a search parameter which can be used, maybe something like atstart:mystring in keeping with their has:, in: notations.
Is there a way to force the match only if it occurs at the start of the line, and only in the case where square brackets are included?
Sincere thanks.
Regex is not on the list of search features, and it was on (more or less, as Better message search functionality (i.e. Wildcard and partial word search)) the list of pre-canned feature requests, so the answer is "you cannot do this via the Gmail web UI" :-(
There are no current Labs features which offer this. SIEVE filters would be another way to do this, that too was not supported, there seems to no longer be any definitive statement on SIEVE support in the Gmail help.
Updated for link rot The pre-canned list of feature requests was, er canned, the original is on archive.org dated 2012, now you just get redirected to a dumbed down page telling you how to give feedback. Lack of SIEVE support was covered in answer 78761 Does Gmail support all IMAP features?, since some time in 2015 that answer silently redirects to the answer about IMAP client configuration, archive.org has a copy dated 2014.
With the current search facility brackets of any form () {} [] are used for grouping, they have no observable effect if there's just one term within. Using (aaa|bbb) and [aaa|bbb] are equivalent and will both find words aaa or bbb. Most other punctuation characters, including \, are treated as a space or a word-separator, + - : and " do have special meaning though, see the help.
As of 2016, only the form "{term1 term2}" is documented for this, and is equivalent to the search "term1 OR term2".
You can do regex searches on your mailbox (within limits) programmatically via Google docs: http://www.labnol.org/internet/advanced-gmail-search/21623/ has source showing how it can be done (copy the document, then Tools > Script Editor to get the complete source).
You could also do this via IMAP as described here:
Python IMAP search for partial subject
and script something to move messages to different folder. The IMAP SEARCH verb only supports substrings, not regex (Gmail search is further limited to complete words, not substrings), further processing of the matches to apply a regex would be needed.
For completeness, one last workaround is: Gmail supports plus addressing, if you can change the destination address to youraddress+jenkinsrelease#gmail.com it will still be sent to your mailbox where you can filter by recipient address. Make sure to filter using the full email address to:youraddress+jenkinsrelease#gmail.com. This is of course more or less the same thing as setting up a dedicated Gmail address for this purpose :-)
Using Google Apps Script, you can use this function to filter email threads by a given regex:
function processInboxEmailSubjects() {
var threads = GmailApp.getInboxThreads();
for (var i = 0; i < threads.length; i++) {
var subject = threads[i].getFirstMessageSubject();
const regex = /^\[RELEASE\]/; //change this to whatever regex you want, this one should cover OP's scenario
let isAtLeast40 = regex.test(subject)
if (isAtLeast40) {
Logger.log(subject);
// Now do what you want to do with the email thread. For example, skip inbox and add an already existing label, like so:
threads[i].moveToArchive().addLabel("customLabel")
}
}
}
As far as I know, unfortunately there isn't a way to trigger this with every new incoming email, so you have to create a time trigger like so (feel free to change it to whatever interval you think best):
function createTrigger(){ //you only need to run this once, then the trigger executes the function every hour in perpetuity
ScriptApp.newTrigger('processInboxEmailSubjects').timeBased().everyHours(1).create();
}
The only option I have found to do this is find some exact wording and put that under the "Has the words" option. Its not the best option, but it works.
I was wondering how to do this myself; it seems Gmail has since silently implemented this feature. I created the following filter:
Matches: subject:([test])
Do this: Skip Inbox
And then I sent a message with the subject
[test] foo
And the message was archived! So it seems all that is necessary is to create a filter for the subject prefix you wish to handle.
I have a peculiar problem. I have an email group that pipes emails to a message board. The word wrap of the emails varies. In yahoo, the messages tend to fill the entire container on the message board. But in all other mail clients, only part of the container width is filled, because the original mail was wrapped. I want all of the email messages to fill the entire width of the container. I've thought of two possible solutions: CSS, or a Regex that eliminates line breaks. Because I am only a garage mechanic (at these sorts of things), I simply cannot get the job done. Any help out there?
Here is a link that shows the issue: http://seanwilson.org/forum/index.php?t=msg&th=1729&start=0&S=171399e41f2c10c4357dd9b217caaa3f
(compare the message of "sean" with that of "rob." One fills the container, the other not).
Can any of you suggest how to get all the mail to fill the container?
You gave too little information - what programming language are you using - PHP/Javascript/anything different?
I think you only need to replace \n, \r and \r\n with whitespace. PHP code for that:
$nowrap = str_replace('\r\n', ' ', $nowrap);
$nowrap = str_replace('\r', ' ', $nowrap);
$nowrap = str_replace('\n', ' ', $nowrap);
You can do that analogically in other languages (for JS see string.replace method: http://www.tizag.com/javascriptT/javascript-string-replace.php).
Depending on the situation (people always seem to add 2 linebreaks between paragraphs), you could say the problem is: replace all newlines not directly preceded or followed by a newline with a space.
//just to be sure, remove \r's
$string = str_replace("\r",'',$string);
$string = preg_replace('/(?<!\n)\n(?!\n)/',' ',$string);
While allowing \r's:
$string = preg_replace('/(?<!\r|\n)\r?\n(?!\r|\n)/',' ',$string);
Edit: nevermind: do not use: while people tend to write their email text in paragraphs, you will break their signature / signoff with this regex. One could fiddle around with a minimum linelength before deeming it 'breakable' (i chose 63), but fiddly it will be:
$string = preg_replace('/([^\r\n]{63,})\r?\n(?!\r|\n)/','$1 ',$string);
The problem is: there are no assurance the linebreak wasn't intended. With a fiddleable line-length you could base it on average users, but the question is: what do they mind more: the differences between breaking & non-breaking paragraphs, or the breaking of their signatures?
Thanks for getting back so quickly!
The discussion board uses php (and also CSS). The only trouble is that I am somewhat limited in my ability to tinker with its programing. If I am to do this at my current level of skilty, I have only one of two options.
using a preg-replace in php. The discussion board allows us to do this from a control panel. So If I could do it with one preg-replace statement, it should work.
Would Wrikken's solution work if I do not remove \r's? Because that seems to be spot on. (could the \r's be added to the preg-replace?)
I had hoped the solution could come through a css property of some sort. I guess that isn't possible.
Thanks so much for your help!
[NOTE: thanks so much for your help! The solution worked!!! I changed the number to 53 or so. It needed to be a little smaller. I don't care that a rare, long signature lines may lose its carriage return. That's a small price to pay for a full message box! You easily saved me several days of learning something that was bound to be moderately frustrating, Thanks so much for that quick fix. I am joyous at the help I received here.]