cm_perl -- A Perl Plugin for Claws Mail
DESCRIPTION
This plugin provides an extended filtering engine for the email client Claws Mail. It allows for the use of full Perl power in email filters.
QUICK START
To get started, you can use the matcherrc2perlfilter.pl script in the tools-directory to translate your old filtering rules to Perl. Simply execute the script and follow the instructions. (note that with recent versions of Claws Mail, this script might not work due to upstream syntax changes. This will get updated in the future. Send me an email if you have problems getting started).
However, you might want to consider reading the rest of this manual and rewriting your rules if you choose to use the plugin, since the Perl code produced by this script is not exactly pretty.
USAGE
The Perl plugin expects a Perl script file called perl_filter in Claws
Mail' config directory (usually $HOME/.claws-mail -- try `claws-mail
--config-dir' if you're unsure). If that file doesn't exist on
plugin start, an empty one is created. This file, which doesn't need to
start with a sha-bang (!#/bin/perl), holds the Perl instructions for your email
filters. To encourage some good manners, the code is executed in a use
strict;
environment.
Both Claws Mail' filtering conditions and actions are mapped to Perl functions with corresponding names, wherever this is possible.
FUNCTION LISTING
For a detailed function description, see section "FUNCTION DESCRIPTIONS", below.
- Standard Filtering Conditions
-
all, marked, unread, deleted, new, replied, forwarded, locked, ignore_thread, colorlabel, match, matchcase, regexp, S<regexpcase, test, size_greater, size_smaller, size_equal, score_greater, score_lower, score_equal, age_greater, age_lower, partial, tagged
- Standard Filtering Actions
-
mark, unmark, dele, mark_as_unread, mark_as_read, lock, unlock, move, copy, color, execute, hide, set_score, change_score, stop, forward, forward_as_attachment, redirect, set_tag, unset_tag, clear_tags
- Fun stuff
-
header, body, filepath, extract_addresses, move_to_trash, abort, addr_in_addressbook, from_in_addressbook, get_attribute_value, SA_is_spam, exit, manual, make_sure_folder_exists, filter_log, filter_log_verbosity, make_sure_tag_exists
FUNCTION DESCRIPTIONS
In general, after the filtering invoked by the Perl script, the mail is passed on to Claws' internal filtering engine, unless a final rule was hit. Final rules stop not only the Perl filtering script at the point of their occurence, but also prevent processing that email by Claws' internal filtering engine (this might sound confusing, but you are already familiar with that concept from standard filters: After an email was e.g. moved, the following rules don't apply anymore).
Also, be careful with the way you quote. In particular, remember that the @-sign has a special meaning in Perl, and gets interpolated inside double quotes. See "Quote and Quote-like Operators" in perlop to learn more about quoting and interpolation.
Standard Filtering Conditions
- all
-
Returns a true value. Available for completness only.
- marked
-
Returns a true value if the marked flag is set, false otherwise.
- unread
-
Returns a true value if the unread flag is set, false otherwise.
- deleted
-
Returns a true value if the deleted flag is set, false otherwise.
- new
-
Returns a true value if the new flag is set, false otherwise.
- replied
-
Returns a true value if the replied flag is set, false otherwise.
- forwarded
-
Returns a true value if the forwarded flag is set, false otherwise.
- locked
-
Returns a true value if the locked flag is set, false otherwise.
- ignore_thread
-
Returns a true value if the "Ignore Thread" flag is set, false otherwise.
- colorlabel COLOR
- colorlabel
-
Returns a true value if message has the color COLOR. COLOR can be either a numeric value between 0 and 7 (with colors corresponding to the internal filtering engine), or the english color name as it is introduced in the filtering dialog (that is, one of: none, orange, red, pink, sky blue, blue, green or brown, while upper and lower case letters make no difference). If COLOR is omitted, 0 (none) is assumed.
- size_greater SIZE
-
Returns a true value if message size is greater than SIZE, false otherwise.
- size_smaller SIZE
-
Returns a true value if message size is smaller than SIZE, false otherwise.
- size_equal SIZE
-
Returns a true value if message size is equal to SIZE, false otherwise.
- score_greater SCORE
-
Returns a true value if message score is greater than SCORE, false otherwise.
- score_lower SCORE
-
Returns a true value if message score is lower than SCORE, false otherwise.
- score_equal SCORE
-
Returns a true value if message score is equal to SCORE, false otherwise.
- age_greater AGE
-
Returns a true value if message age is greater than AGE, false otherwise.
- age_lower AGE
-
Returns a true value if message age is lower than AGE, false otherwise.
- partial
-
Returns a true value if message has only partially been downloaded, false otherwise.
- tagged
-
Returns a true value if the messages has one or more tags.
- test
-
Corresponds the 'test' internal filtering rule. In particular, it accepts the same symbols, namely:
- %%
-
%
- %s
-
Subject
- %f
-
From
- %t
-
To
- %c
-
Cc
- %d
-
Date
- %i
-
Message-ID
- %n
-
Newsgroups
- %r
-
References
- %F
-
Filename -- should not be modified
- match WHERE WHAT
- matchcase WHERE WHAT
- regexp WHERE WHAT
- regexpcase WHERE WHAT
-
The matching functions have a special syntax. The first argument is either any of to_or_cc, body_part, headers_part, headers_cont, message, to, from, subject, cc, newsgroups, inreplyto, references, or tag (those strings may or may not be quoted), the pattern matching works on that area. If it is any other string (which must then be quoted), this string is taken to be the name of a header field.
The second argument is the string to look for. For match, matchcase, regexp and regexpcase we have case sensitive normal matching, case insensitive normal matching, case sensitive regular expression matching and case insensitive regular expression pattern matching, respectively.
The functions return true if the pattern was found, false otherwise.
Just as with the built-in filtering engine, the message body is searched and provided as is - no character-set analysis is done. Likewise, no HTML-tags are stripped. It should be possible to use external modules or programs for these tasks though. If you're doing that, drop me a message with your experiences.
With Perl having its strenghts in pattern matching, using Perl's builtin operators are usually a better option than using these functions.
Standard Filtering Actions
The actions return a true value upon success, and 'undef' when an error occured. Final message rules are indicated. (See above for a sketch what a final rule is)
- mark
-
Mark the message.
- unmark
-
Unmark the message.
- dele
-
Delete the message. Note the name change of Claws Mail' "delete" to "dele". This is because "delete" is one of Perl's builtin commands which cannot be redefined (if it can, tell me how).
This is a final rule.
- mark_as_read
-
Mark the message as read
- mark_as_unread
-
Mark the message as unread
- lock
-
Lock the message
- unlock
-
Remove the message lock
- move DESTINATION
-
Move the message to folder DESTINATION. The folder notation is the same that Claws Mail uses. You can copy & paste from the move dialog of the normal filtering, until you get a feeling for the notation.
This is a final rule.
- copy DESTINATION
-
Copy the message to folder DESTINATION. The folder notation is the same that Claws Mail uses. You can copy & paste from the move dialog of the normal filtering, until you get a feeling for the notation.
- execute COMMAND
-
This is the same as the test - rule from section "Standard Filtering Conditions" execpt that it always returns a true value.
- hide
-
Hide the message
- set_score SCORE
-
Set message score to SCORE
- change_score SCORE
-
Change message score by SCORE
- stop
-
Stop Perl script at this point. Note that this is not a final rule, meaning that the email gets passed on to the internal filtering engine. See "abort" below if you don't want that.
- forward ACCOUNT, EMAIL
-
Forward the message to email address EMAIL, using the account ID ACCOUNT as sender account. So far, you have to create a rule in the normal filtering engine to find out that number.
- forward_as_attachment, ACCOUNT EMAIL
-
Forward the message to email address EMAIL in an attachment, using the account ID ACCOUNT as sender account. So far, you have to create a rule in the normal filtering engine to find out that number.
- redirect ACCOUNT, EMAIL
-
Redirect the message to EMAIL, using the account ID ACCOUNT as sender account. So far, you have to create a rule in the normal filtering engine to find out that number.
- set_tag TAG
-
Apply tag TAG. TAG must exist.
- unset_tag TAG
-
Unset tag TAG.
-
Clear all tags.
Fun stuff
Functions
- header ARG
- header
-
If ARG is not given, returns a list of all header field names of the mail.
If ARG is given, returns 'undef' if the header field ARG does not exist in the email. Otherwise, it returns
- in scalar context
-
The value of the header field ARG.
- in list context
-
A list of all available header field values. This is useful if a header field occurs more than once in an email (eg the Received - header).
The header field "References" forms a special case. In a scalar context, it returns the first reference. In a list context, it returns a list of all references.
- body
-
Returns the email body in a scalar.
- filepath
-
Returns the file and path of the email that is currently filtered (corresponds to the %F arguemnt in the 'test' rule).
- extract_addresses
-
Extracts email addresses from a string and gives back a list of addresses found. Currently an email address is found using the regular expression '[-.+\w]+\@[-.+\w]+'. This will not find all valid email addresses. Feel free to send me a better regexp.
- move_to_trash
-
Move the email message to default trash folder.
This is a final rule.
- abort
-
Stop Perl script at this point.
In contrast to 'stop', this is a final rule.
- addr_in_addressbook EMAIL, ADDRESSBOOK
- addr_in_addressbook EMAIL
-
Returns a true value if the email address EMAIL is in the addressbook with the name ADDRESSBOOK. If ADDRESSBOOK is not given, returns true if the email address is in any addressbook.
- from_in_addressbook ADDRESSBOOK
- from_in_addressbook
-
Checks if the email address found in the From-header is in addressbook ADDRESSBOOK (or any, if omitted). It is implemented as
my ($from) = extract_addresses(header("from")); return 0 unless $from; return addr_in_addressbook($from,@_);
so the same restrictions as to extract_addresses apply.
- get_attribute_value EMAIL, ATTRIBUTE, ADDRESSBOOK
- get_attribute_value EMAIL, ATTRIBUTE
-
Looks through the addressbook ADDRESSBOOK (or all addressbooks, if omitted) for a contact with the an email address EMAIL. If found, the function checks if this contact has a user attribute with name ATTRIBUTE. It returns the value of this attribute, or an empty string if it was not found. As usual, 'undef' is returned if an error occured.
- SA_is_spam
-
Is an alias to
not test 'spamc -c < %F > /dev/null'
- exit
-
Has been redefined to be an alias to 'stop'. You shouldn't use Perl's own 'exit' command, since it would exit Claws Mail.
- manual
-
Returns a true value if the filter script was invoked manually, that is, via the Tools menu.
- make_sure_folder_exists IDENTIFIER
-
Returns a true value if the folder with id IDENTIFIER (e.g. #mh/Mail/foo/bar) exists or could be created.
- make_sure_tag_exists TAG
-
Returns a true value if the tag TAG exists or could be created.
- filter_log SECTION, TEXT
- filter_log TEXT
-
Writes TEXT to the filter logfile. SECTION can be any of
"LOG_MANUAL"
"LOG_MATCH"
"LOG_ACTION"
If the SECTION is omitted, "LOG_MANUAL" is assumed.
- filter_log_verbosity VERBOSITY
- filter_log_verbosity
-
Changes the filter log verbosity for the current mail. VERBOSITY must be any of
0
-
Be silent
1
-
Log MANUAL type
2
-
Log Action type
3
-
Log MATCH type
For the meaning of those numbers, read section "LOGGING". If VERBOSITY is omitted, the filter logfile verbosity is not changed.
This function returns the filter_log_verbosity number before the change (if any).
Variables
- $permanent
-
This scalar keeps its value between filtered mail messages. On plugin start, it is initialized to the empty string.
LOGGING
To keep track of what has been done to the mails while filtering, the plugin supports logging. Three verbosity levels are recognized:
0
-
logging disabled
1
-
log only manual messages, that is, messages introduced by the
filter_log
command in filter scripts 2
-
log manual messages and filter actions
3
-
log manual messages, filter actions and filter matches
The messages are logged in Claws Mail' log window. The default log level
is 2. Log level 3 is not recommended, because the matcher functions log a
message if they succeeded, and thus, if you have negative checks, you'll get
confusing entries. If you want to keep track of matching, do it manually, using
filter_log
, or do it by temporary enabling matcher logging using
filter_log_verbosity
.
The first time you unload this plugin (or shut down Claws Mail), a section called [PerlPlugin] will be created in Claws Mail' configuration file clawsrc, containing one variable:
* filter_log_verbosity
If you want to change the default behaviour, you can edit this line. Make sure Claws Mail is not running while you do this.
It will be possible to access these setting via the GUI, as soon as I find the time to write a corresponding GTK plugin, or somebody else is interested in contributing that.
EXAMPLE
This section lists a small example of a Perl script file. I'm sure you get the idea..
#-8<----------------------------------------------------
# -*- perl -*-
# local functions
# Learn ham messages, and move them to specified folder. This is
# useful for making sure a bayes filter sees ham as well.
sub learn_and_move {
execute('put command to learn ham here');
move(@_);
}
# Two-stage spam filter. Every email that scores higher than 15
# on SpamAssassin gets moved into the default trash folder.
# All mails lower than that, but higher than SpamAssassin's
# 'required_hits' go into #mh/mail/Spam.
sub spamcheck {
my $surely_spam = 15;
my $filepath = filepath;
my $spamc = `spamc -c < $filepath`;
my ($value,$threshold) = ($spamc =~ m|([-.,0-9]+)/([-.,0-9]+)|);
if($value >= $surely_spam) {
mark_as_read;
move_to_trash;
}
if($value >= $threshold) {mark_as_read; move '#mh/mail/Spam';}
}
# Perl script execution starts here.
# Some specific sorting
learn_and_move '#mh/mail/MailLists/Claws Mail/user'
if
matchcase('sender','claws-mail-users-admin@lists.sourceforge.net');
learn_and_move '#mh/mail/MailLists/Sylpheed'
if matchcase('list-id','sylpheed.good-day.net');
# Implement imcomming folders using addressbook
# attributes. Target folders for specific email addresses are
# stored directly in the addressbook. This way, if an email
# address changes, we only have to update the addressbook, not
# the filter rules! Besides that, we can greatly unclutter the
# filter script.
# get the email address in the from header
my $fromheader = header "from";
my ($from) = extract_addresses $fromheader;
# check if this email address has an associated attribute
# called "incomming_folder". If if has, the value of this
# attribute is supposed to be the target folder.
my $value = get_attribute_value $from, "incomming_folder";
learn_and_move($value) if $value;
# An example of a whitelist: If the from-address is in my
# "office" addressbook, move the mail to folder #mh/mail/office
learn_and_move '#mh/mail/office' if
from_in_addressbook("office");
# If the from-address is in any other addressbook, move the
# mail to folder #mh/mail/inbox/known
learn_and_move '#mh/mail/inbox/known' if from_in_addressbook;
# Feed the remaining mails through SpamAssassin.
spamcheck;
# mails that make it to the end of the script are passed on to
# the internal filtering engine. If the internal rules don't say
# otherwise, the mails end up in the default inbox.
#-8<----------------------------------------------------
BUGS
Do not use this plugin together with other filtering plugins, especially the Spamassassin and ClamAV plugins. They are registered on the same hook and the order in which the plugins are executed is not guaranteed.
The filter script is not (yet) updated automatically when a folder gets renamed. The same applies for folder names in addressbook user attributes.
This plugin has only be tested with POP3 accounts. If you have experiences with IMAP or newsgroup accounts, drop me a message.
Warning during compile time:
*** Warning: Linking the shared library perl_plugin.la against the *** static library /usr/lib/perl5/5.8.3/i586-linux-thread-multi/auto/DynaLoader/DynaLoader.a is not portable!
Ideas to solve this one are welcome :-)