1 <?xml version="1.0" encoding="UTF-8"?>
2 <!DOCTYPE book PUBLIC "-//OASIS//DTD DocBook XML V4.2//EN" "http://www.oasis-open.org/docbook/xml/4.2/docbookx.dtd">
5 <title>Exim's interfaces to mail filtering</title>
6 <titleabbrev>Exim filtering</titleabbrev>
8 <author><firstname>Philip</firstname><surname>Hazel</surname></author>
9 <authorinitials>PH</authorinitials>
10 <revhistory><revision>
11 <revnumber>4.65</revnumber>
12 <date>31 July 2006</date>
13 <authorinitials>PH</authorinitials>
14 </revision></revhistory>
15 <copyright><year>2006</year><holder>University of Cambridge</holder></copyright>
18 <title>Forwarding and filtering in Exim</title>
20 This document describes the user interfaces to Exim’s in-built mail filtering
21 facilities, and is copyright © University of Cambridge 2006. It
22 corresponds to Exim version 4.63.
25 <title>Introduction</title>
27 Most Unix mail transfer agents (programs that deliver mail) permit individual
28 users to specify automatic forwarding of their mail, usually by placing a list
29 of forwarding addresses in a file called <filename>.forward</filename> in their home
30 directories. Exim extends this facility by allowing the forwarding instructions
31 to be a set of rules rather than just a list of addresses, in effect providing
32 <quote><filename>.forward</filename> with conditions</quote>. Operating the set of rules is called
33 <emphasis>filtering</emphasis>, and the file that contains them is called a <emphasis>filter file</emphasis>.
36 Exim supports two different kinds of filter file. An <emphasis>Exim filter</emphasis> contains
37 instructions in a format that is unique to Exim. A <emphasis>Sieve filter</emphasis> contains
38 instructions in the Sieve format that is defined by RFC 3028. As this is a
39 standard format, Sieve filter files may already be familiar to some users.
40 Sieve files should also be portable between different environments. However,
41 the Exim filtering facility contains more features (such as variable
42 expansion), and better integration with the host environment (such as the use
43 of external processes and pipes).
46 The choice of which kind of filter to use can be left to the end-user, provided
47 that the system administrator has configured Exim appropriately for both kinds
48 of filter. However, if interoperability is important, Sieve is the only
52 The ability to use filtering or traditional forwarding has to be enabled by the
53 system administrator, and some of the individual facilities can be separately
54 enabled or disabled. A local document should be provided to describe exactly
55 what has been enabled. In the absence of this, consult your system
59 This document describes how to use a filter file and the format of its
60 contents. It is intended for use by end-users. Both Sieve filters and Exim
61 filters are covered. However, for Sieve filters, only issues that relate to the
62 Exim implementation are discussed, since Sieve itself is described elsewhere.
65 The contents of traditional <filename>.forward</filename> files are not described here. They
66 normally contain just a list of addresses, file names, or pipe commands,
67 separated by commas or newlines, but other types of item are also available.
68 The full details can be found in the chapter on the <command>redirect</command> router in the
69 Exim specification, which also describes how the system administrator can set
70 up and control the use of filtering.
74 <title>Filter operation</title>
76 It is important to realize that, in Exim, no deliveries are actually made while
77 a filter or traditional <filename>.forward</filename> file is being processed. Running a filter
78 or processing a traditional <filename>.forward</filename> file sets up future delivery
79 operations, but does not carry them out.
82 The result of filter or <filename>.forward</filename> file processing is a list of destinations
83 to which a message should be delivered. The deliveries themselves take place
84 later, along with all other deliveries for the message. This means that it is
85 not possible to test for successful deliveries while filtering. It also means
86 that any duplicate addresses that are generated are dropped, because Exim never
87 delivers the same message to the same address more than once.
90 <section id="SECTtesting">
91 <title>Testing a new filter file</title>
93 Filter files, especially the more complicated ones, should always be tested, as
94 it is easy to make mistakes. Exim provides a facility for preliminary testing
95 of a filter file before installing it. This tests the syntax of the file and
96 its basic operation, and can also be used with traditional <filename>.forward</filename> files.
99 Because a filter can do tests on the content of messages, a test message is
100 required. Suppose you have a new filter file called <filename>myfilter</filename> and a test
101 message in a file called <filename>test-message</filename>. Assuming that Exim is installed with
102 the conventional path name <filename>/usr/sbin/sendmail</filename> (some operating systems use
103 <filename>/usr/lib/sendmail</filename>), the following command can be used:
105 <literallayout class="monospaced">
106 /usr/sbin/sendmail -bf myfilter <test-message
109 The <option>-bf</option> option tells Exim that the following item on the command line is
110 the name of a filter file that is to be tested. There is also a <option>-bF</option> option,
111 which is similar, but which is used for testing system filter files, as opposed
112 to user filter files, and which is therefore of use only to the system
116 The test message is supplied on the standard input. If there are no
117 message-dependent tests in the filter, an empty file (<filename>/dev/null</filename>) can be
118 used. A supplied message must start with header lines or the <quote>From </quote> message
119 separator line that is found in many multi-message folder files. Note that
120 blank lines at the start terminate the header lines. A warning is given if no
121 header lines are read.
124 The result of running this command, provided no errors are detected in the
125 filter file, is a list of the actions that Exim would try to take if presented
126 with the message for real. For example, for an Exim filter, the output
128 <literallayout class="monospaced">
129 Deliver message to: gulliver@lilliput.fict.example
130 Save message to: /home/lemuel/mail/archive
133 means that one copy of the message would be sent to
134 <emphasis>gulliver@lilliput.fict.example</emphasis>, and another would be added to the file
135 <filename>/home/lemuel/mail/archive</filename>, if all went well.
138 The actions themselves are not attempted while testing a filter file in this
139 way; there is no check, for example, that any forwarding addresses are valid.
140 For an Exim filter, if you want to know why a particular action is being taken,
141 add the <option>-v</option> option to the command. This causes Exim to output the results of
142 any conditional tests and to indent its output according to the depth of
143 nesting of <command>if</command> commands. Further additional output from a filter test can be
144 generated by the <command>testprint</command> command, which is described below.
147 When Exim is outputting a list of the actions it would take, if any text
148 strings are included in the output, non-printing characters therein are
149 converted to escape sequences. In particular, if any text string contains a
150 newline character, this is shown as <quote>\n</quote> in the testing output.
153 When testing a filter in this way, Exim makes up an <quote>envelope</quote> for the
154 message. The recipient is by default the user running the command, and so is
155 the sender, but the command can be run with the <option>-f</option> option to supply a
156 different sender. For example,
158 <literallayout class="monospaced">
159 /usr/sbin/sendmail -bf myfilter \
160 -f islington@never.where <test-message
163 Alternatively, if the <option>-f</option> option is not used, but the first line of the
164 supplied message is a <quote>From </quote> separator from a message folder file (not the
165 same thing as a <emphasis>From:</emphasis> header line), the sender is taken from there. If
166 <option>-f</option> is present, the contents of any <quote>From </quote> line are ignored.
169 The <quote>return path</quote> is the same as the envelope sender, unless the message
170 contains a <emphasis>Return-path:</emphasis> header, in which case it is taken from there. You
171 need not worry about any of this unless you want to test out features of a
172 filter file that rely on the sender address or the return path.
175 It is possible to change the envelope recipient by specifying further options.
176 The <option>-bfd</option> option changes the domain of the recipient address, while the
177 <option>-bfl</option> option changes the <quote>local part</quote>, that is, the part before the @
178 sign. An adviser could make use of these to test someone else’s filter file.
181 The <option>-bfp</option> and <option>-bfs</option> options specify the prefix or suffix for the local
182 part. These are relevant only when support for multiple personal mailboxes is
183 implemented; see the description in section <xref linkend="SECTmbox"/> below.
187 <title>Installing a filter file</title>
189 A filter file is normally installed under the name <filename>.forward</filename> in your home
190 directory – it is distinguished from a conventional <filename>.forward</filename> file by its
191 first line (described below). However, the file name is configurable, and some
192 system administrators may choose to use some different name or location for
197 <title>Testing an installed filter file</title>
199 Testing a filter file before installation cannot find every potential problem;
200 for example, it does not actually run commands to which messages are piped.
201 Some <quote>live</quote> tests should therefore also be done once a filter is installed.
204 If at all possible, test your filter file by sending messages from some other
205 account. If you send a message to yourself from the filtered account, and
206 delivery fails, the error message will be sent back to the same account, which
207 may cause another delivery failure. It won’t cause an infinite sequence of such
208 messages, because delivery failure messages do not themselves generate further
209 messages. However, it does mean that the failure won’t be returned to you, and
210 also that the postmaster will have to investigate the stuck message.
213 If you have to test an Exim filter from the same account, a sensible precaution
214 is to include the line
216 <literallayout class="monospaced">
217 if error_message then finish endif
220 as the first filter command, at least while testing. This causes filtering to
221 be abandoned for a delivery failure message, and since no destinations are
222 generated, the message goes on to be delivered to the original address. Unless
223 there is a good reason for not doing so, it is recommended that the above test
224 be left in all Exim filter files. (This does not apply to Sieve files.)
228 <title>Details of filtering commands</title>
230 The filtering commands for Sieve and Exim filters are completely different in
231 syntax and semantics. The Sieve mechanism is defined in RFC 3028; in the next
232 chapter we describe how it is integrated into Exim. The subsequent chapter
233 covers Exim filtering commands in detail.
238 <chapter id="CHAPsievefilter">
239 <title>Sieve filter files</title>
241 The code for Sieve filtering in Exim was contributed by Michael Haardt, and
242 most of the content of this chapter is taken from the notes he provided. Since
243 Sieve is an extensible language, it is important to understand <quote>Sieve</quote> in
244 this context as <quote>the specific implementation of Sieve for Exim</quote>.
247 This chapter does not contain a description of Sieve, since that can be found
248 in RFC 3028, which should be read in conjunction with these notes.
251 The Exim Sieve implementation offers the core as defined by RFC 3028,
252 comparison tests, the <emphasis role="bold">copy</emphasis>, <emphasis role="bold">envelope</emphasis>, <emphasis role="bold">fileinto</emphasis>, and <emphasis role="bold">vacation</emphasis>
253 extensions, but not the <emphasis role="bold">reject</emphasis> extension. Exim does not support message
254 delivery notifications (MDNs), so adding it just to the Sieve filter (as
255 required for <emphasis role="bold">reject</emphasis>) makes little sense.
258 In order for Sieve to work properly in Exim, the system administrator needs to
259 make some adjustments to the Exim configuration. These are described in the
260 chapter on the <command>redirect</command> router in the full Exim specification.
263 <title>Recognition of Sieve filters</title>
265 A filter file is interpreted as a Sieve filter if its first line is
267 <literallayout class="monospaced">
271 This is what distinguishes it from a conventional <filename>.forward</filename> file or an Exim
276 <title>Saving to specified folders</title>
278 If the system administrator has set things up as suggested in the Exim
279 specification, and you use <command>keep</command> or <command>fileinto</command> to save a mail into a
280 folder, absolute files are stored where specified, relative files are stored
281 relative to <varname>$home</varname>, and <filename>inbox</filename> goes to the standard mailbox location.
285 <title>Strings containing header names</title>
287 RFC 3028 does not specify what happens if a string denoting a header field does
288 not contain a valid header name, for example, it contains a colon. This
289 implementation generates an error instead of ignoring the header field in order
290 to ease script debugging, which fits in with the common picture of Sieve.
294 <title>Exists test with empty list of headers</title>
296 The <emphasis role="bold">exists</emphasis> test succeeds only if all the specified headers exist. RFC 3028
297 does not explicitly specify what happens on an empty list of headers. This
298 implementation evaluates that condition as true, interpreting the RFC in a
303 <title>Header test with invalid MIME encoding in header</title>
305 Some MUAs process invalid base64 encoded data, generating junk. Others ignore
306 junk after seeing an equal sign in base64 encoded data. RFC 2047 does not
307 specify how to react in this case, other than stating that a client must not
308 forbid to process a message for that reason. RFC 2045 specifies that invalid
309 data should be ignored (apparently looking at end of line characters). It also
310 specifies that invalid data may lead to rejecting messages containing them (and
311 there it appears to talk about true encoding violations), which is a clear
312 contradiction to ignoring them.
315 RFC 3028 does not specify how to process incorrect MIME words. This
316 implementation treats them literally, as it does if the word is correct but its
317 character set cannot be converted to UTF-8.
321 <title>Address test for multiple addresses per header</title>
323 A header may contain multiple addresses. RFC 3028 does not explicitly specify
324 how to deal with them, but since the address test checks if anything matches
325 anything else, matching one address suffices to satisfy the condition. That
326 makes it impossible to test if a header contains a certain set of addresses and
327 no more, but it is more logical than letting the test fail if the header
328 contains an additional address besides the one the test checks for.
332 <title>Semantics of keep</title>
334 The <command>keep</command> command is equivalent to
336 <literallayout class="monospaced">
340 It saves the message and resets the implicit keep flag. It does not set the
341 implicit keep flag; there is no command to set it once it has been reset.
345 <title>Semantics of fileinto</title>
347 RFC 3028 does not specify whether <command>fileinto</command> should try to create a mail
348 folder if it does not exist. This implementation allows the sysadmin to
349 configure that aspect using the <command>appendfile</command> transport options
350 <option>create_directory</option>, <option>create_file</option>, and <option>file_must_exist</option>. See the
351 <command>appendfile</command> transport in the Exim specification for details.
355 <title>Semantics of redirect</title>
357 Sieve scripts are supposed to be interoperable between servers, so this
358 implementation does not allow mail to be redirected to unqualified addresses,
359 because the domain would depend on the system being used. On systems with
360 virtual mail domains, the default domain is probably not what the user expects
365 <title>String arguments</title>
367 There has been confusion if the string arguments to <command>require</command> are to be
368 matched case-sensitively or not. This implementation matches them with the
369 match type <command>:is</command> (default, see section 2.7.1 of the RFC) and the comparator
370 <command>i;ascii-casemap</command> (default, see section 2.7.3 of the RFC). The RFC defines
371 the command defaults clearly, so any different implementations violate RFC
372 3028. The same is valid for comparator names, also specified as strings.
376 <title>Number units</title>
378 There is a mistake in RFC 3028: the suffix G denotes gibi-, not tebibyte.
379 The mistake is obvious, because RFC 3028 specifies G to denote 2^30
380 (which is gibi, not tebi), and that is what this implementation uses as
381 the scaling factor for the suffix G.
385 <title>RFC compliance</title>
387 Exim requires the first line of a Sieve filter to be
389 <literallayout class="monospaced">
393 Of course the RFC does not specify that line. Do not expect examples to work
394 without adding it, though.
397 RFC 3028 requires the use of CRLF to terminate a line. The rationale was that
398 CRLF is universally used in network protocols to mark the end of the line. This
399 implementation does not embed Sieve in a network protocol, but uses Sieve
400 scripts as part of the Exim MTA. Since all parts of Exim use LF as the newline
401 character, this implementation does, too, by default, though the system
402 administrator may choose (at Exim compile time) to use CRLF instead.
405 Exim violates RFC 2822, section 3.6.8, by accepting 8-bit header names, so this
406 implementation repeats this violation to stay consistent with Exim. This is in
407 preparation for UTF-8 data.
410 Sieve scripts cannot contain NUL characters in strings, but mail headers could
411 contain MIME encoded NUL characters, which could never be matched by Sieve
412 scripts using exact comparisons. For that reason, this implementation extends
413 the Sieve quoted string syntax with \0 to describe a NUL character, violating
414 \0 being the same as 0 in RFC 3028. Even without using \0, the following tests
415 are all true in this implementation. Implementations that use C-style strings
416 will only evaluate the first test as true.
418 <literallayout class="monospaced">
419 Subject: =?iso-8859-1?q?abc=00def
421 header :contains "Subject" ["abc"]
422 header :contains "Subject" ["def"]
423 header :matches "Subject" ["abc?def"]
426 Note that by considering Sieve to be an MUA, RFC 2047 can be interpreted in a
427 way that NUL characters truncating strings is allowed for Sieve
428 implementations, although not recommended. It is further allowed to use encoded
429 NUL characters in headers, but that’s not recommended either. The above example
433 RFC 3028 states that if an implementation fails to convert a character set to
434 UTF-8, two strings cannot be equal if one contains octets greater than 127.
435 Assuming that all unknown character sets are one-byte character sets with the
436 lower 128 octets being US-ASCII is not sound, so this implementation violates
437 RFC 3028 and treats such MIME words literally. That way at least something
441 The folder specified by <command>fileinto</command> must not contain the character sequence
442 <quote>..</quote> to avoid security problems. RFC 3028 does not specify the syntax of
443 folders apart from <command>keep</command> being equivalent to
445 <literallayout class="monospaced">
449 This implementation uses <filename>inbox</filename> instead.
452 Sieve script errors currently cause messages to be silently filed into
453 <filename>inbox</filename>. RFC 3028 requires that the user is notified of that condition.
454 This may be implemented in the future by adding a header line to mails that
455 are filed into <filename>inbox</filename> due to an error in the filter.
460 <chapter id="CHAPeximfilter">
461 <title>Exim filter files</title>
463 This chapter contains a full description of the contents of Exim filter files.
466 <title>Format of Exim filter files</title>
468 Apart from leading white space, the first text in an Exim filter file must be
470 <literallayout class="monospaced">
474 This is what distinguishes it from a conventional <filename>.forward</filename> file or a Sieve
475 filter file. If the file does not have this initial line (or the equivalent for
476 a Sieve filter), it is treated as a conventional <filename>.forward</filename> file, both when
477 delivering mail and when using the <option>-bf</option> testing mechanism. The white space
478 in the line is optional, and any capitalization may be used. Further text on
479 the same line is treated as a comment. For example, you could have
481 <literallayout class="monospaced">
482 # Exim filter <<== do not edit or remove this line!
485 The remainder of the file is a sequence of filtering commands, which consist of
486 keywords and data values. For example, in the command
488 <literallayout class="monospaced">
489 deliver gulliver@lilliput.fict.example
492 the keyword is <literal>deliver</literal> and the data value is
493 <literal>gulliver@lilliput.fict.example</literal>. White space or line breaks separate the
494 components of a command, except in the case of conditions for the <command>if</command>
495 command, where round brackets (parentheses) also act as separators. Complete
496 commands are separated from each other by white space or line breaks; there are
497 no special terminators. Thus, several commands may appear on one line, or one
498 command may be spread over a number of lines.
501 If the character # follows a separator anywhere in a command, everything from
502 # up to the next newline is ignored. This provides a way of including comments
507 <title>Data values in filter commands</title>
509 There are two ways in which a data value can be input:
514 If the text contains no white space, it can be typed verbatim. However, if it
515 is part of a condition, it must also be free of round brackets (parentheses),
516 as these are used for grouping in conditions.
521 Otherwise, text must be enclosed in double quotation marks. In this case, the
522 character \ (backslash) is treated as an <quote>escape character</quote> within the
523 string, causing the following character or characters to be treated specially:
526 <literal>\n</literal> is replaced by a newline
527 <literal>\r</literal> is replaced by a carriage return
528 <literal>\t</literal> is replaced by a tab
533 Backslash followed by up to three octal digits is replaced by the character
534 specified by those digits, and <literal>\x</literal> followed by up to two hexadecimal digits
535 is treated similarly. Backslash followed by any other character is replaced by
536 the second character, so that in particular, <literal>\"</literal> becomes <literal>"</literal> and <literal>\\</literal>
537 becomes <literal>\</literal>. A data item enclosed in double quotes can be continued onto the
538 next line by ending the first line with a backslash. Any leading white space at
539 the start of the continuation line is ignored.
542 In addition to the escape character processing that occurs when strings are
543 enclosed in quotes, most data values are also subject to <emphasis>string expansion</emphasis>
544 (as described in the next section), in which case the characters <literal>$</literal> and
545 <literal>\</literal> are also significant. This means that if a single backslash is actually
546 required in such a string, and the string is also quoted, <literal>\\\\</literal> has to be
550 The maximum permitted length of a data string, before expansion, is 1024
554 <section id="SECTfilterstringexpansion">
555 <title>String expansion</title>
557 Most data values are expanded before use. Expansion consists of replacing
558 substrings beginning with <literal>$</literal> with other text. The full expansion facilities
559 available in Exim are extensive. If you want to know everything that Exim can
560 do with strings, you should consult the chapter on string expansion in the Exim
564 In filter files, by far the most common use of string expansion is the
565 substitution of the contents of a variable. For example, the substring
567 <literallayout class="monospaced">
571 is replaced by the address to which replies to the message should be sent. If
572 such a variable name is followed by a letter or digit or underscore, it must be
573 enclosed in curly brackets (braces), for example,
575 <literallayout class="monospaced">
579 If a <literal>$</literal> character is actually required in an expanded string, it must be
580 escaped with a backslash, and because backslash is also an escape character in
581 quoted input strings, it must be doubled in that case. The following two
582 examples illustrate two different ways of testing for a <literal>$</literal> character in a
585 <literallayout class="monospaced">
586 if $message_body contains \$ then ...
587 if $message_body contains "\\$" then ...
590 You can prevent part of a string from being expanded by enclosing it between
591 two occurrences of <literal>\N</literal>. For example,
593 <literallayout class="monospaced">
594 if $message_body contains \N$$$$\N then ...
597 tests for a run of four dollar characters.
601 <title>Some useful general variables</title>
603 A complete list of the available variables is given in the Exim documentation.
604 This shortened list contains the ones that are most likely to be useful in
605 personal filter files:
608 <varname>$body_linecount</varname>: The number of lines in the body of the message.
611 <varname>$body_zerocount</varname>: The number of binary zero characters in the body of the
615 <varname>$home</varname>: In conventional configurations, this variable normally contains the
616 user’s home directory. The system administrator can, however, change this.
619 <varname>$local_part</varname>: The part of the email address that precedes the @ sign –
620 normally the user’s login name. If support for multiple personal mailboxes is
621 enabled (see section <xref linkend="SECTmbox"/> below) and a prefix or suffix for the local
622 part was recognized, it is removed from the string in this variable.
625 <varname>$local_part_prefix</varname>: If support for multiple personal mailboxes is enabled
626 (see section <xref linkend="SECTmbox"/> below), and a local part prefix was recognized,
627 this variable contains the prefix. Otherwise it contains an empty string.
630 <varname>$local_part_suffix</varname>: If support for multiple personal mailboxes is enabled
631 (see section <xref linkend="SECTmbox"/> below), and a local part suffix was recognized,
632 this variable contains the suffix. Otherwise it contains an empty string.
635 <varname>$message_body</varname>: The initial portion of the body of the message. By default,
636 up to 500 characters are read into this variable, but the system administrator
637 can configure this to some other value. Newlines in the body are converted into
641 <varname>$message_body_end</varname>: The final portion of the body of the message, formatted
642 and limited in the same way as <varname>$message_body</varname>.
645 <varname>$message_body_size</varname>: The size of the body of the message, in bytes.
648 <varname>$message_exim_id</varname>: The message’s local identification string, which is unique
649 for each message handled by a single host.
652 <varname>$message_headers</varname>: The header lines of the message, concatenated into a
653 single string, with newline characters between them.
656 <varname>$message_size</varname>: The size of the entire message, in bytes.
659 <varname>$original_local_part</varname>: When an address that arrived with the message is
660 being processed, this contains the same value as the variable <varname>$local_part</varname>.
661 However, if an address generated by an alias, forward, or filter file is being
662 processed, this variable contains the local part of the original address.
665 <varname>$reply_address</varname>: The contents of the <emphasis>Reply-to:</emphasis> header, if the message
666 has one; otherwise the contents of the <emphasis>From:</emphasis> header. It is the address to
667 which normal replies to the message should be sent.
670 <varname>$return_path</varname>: The return path – that is, the sender field that will be
671 transmitted as part of the message’s envelope if the message is sent to another
672 host. This is the address to which delivery errors are sent. In many cases,
673 this variable has the same value as <varname>$sender_address</varname>, but if, for example,
674 an incoming message to a mailing list has been expanded, <varname>$return_path</varname> may
675 have been changed to contain the address of the list maintainer.
678 <varname>$sender_address</varname>: The sender address that was received in the envelope of
679 the message. This is not necessarily the same as the contents of the <emphasis>From:</emphasis>
680 or <emphasis>Sender:</emphasis> header lines. For delivery error messages (<quote>bounce messages</quote>)
681 there is no sender address, and this variable is empty.
684 <varname>$tod_full</varname>: A full version of the time and date, for example: Wed, 18 Oct
685 1995 09:51:40 +0100. The timezone is always given as a numerical offset from
689 <varname>$tod_log</varname>: The time and date in the format used for writing Exim’s log files,
690 without the timezone, for example: 1995-10-12 15:32:29.
693 <varname>$tod_zone</varname>: The local timezone offset, for example: +0100.
696 <section id="SECTheadervariables">
697 <title>Header variables</title>
699 There is a special set of expansion variables containing the header lines of
700 the message being processed. These variables have names beginning with
701 <varname>$header_</varname> followed by the name of the header line, terminated by a colon.
704 <literallayout class="monospaced">
709 The whole item, including the terminating colon, is replaced by the contents of
710 the message header line. If there is more than one header line with the same
711 name, their contents are concatenated. For header lines whose data consists of
712 a list of addresses (for example, <emphasis>From:</emphasis> and <emphasis>To:</emphasis>), a comma and newline
713 is inserted between each set of data. For all other header lines, just a
717 Leading and trailing white space is removed from header line data, and if there
718 are any MIME <quote>words</quote> that are encoded as defined by RFC 2047 (because they
719 contain non-ASCII characters), they are decoded and translated, if possible, to
720 a local character set. Translation is attempted only on operating systems that
721 have the <function>iconv()</function> function. This makes the header line look the same as it
722 would when displayed by an MUA. The default character set is ISO-8859-1, but
723 this can be changed by means of the <command>headers</command> command (see below).
726 If you want to see the actual characters that make up a header line, you can
727 specify <varname>$rheader_</varname> instead of <varname>$header_</varname>. This inserts the <quote>raw</quote>
728 header line, unmodified.
731 There is also an intermediate form, requested by <varname>$bheader_</varname>, which removes
732 leading and trailing space and decodes MIME <quote>words</quote>, but does not do any
733 character translation. If an attempt to decode what looks superficially like a
734 MIME <quote>word</quote> fails, the raw string is returned. If decoding produces a binary
735 zero character, it is replaced by a question mark.
738 The capitalization of the name following <varname>$header_</varname> is not significant.
739 Because any printing character except colon may appear in the name of a
740 message’s header (this is a requirement of RFC 2822, the document that
741 describes the format of a mail message) curly brackets must <emphasis>not</emphasis> be used in
742 this case, as they will be taken as part of the header name. Two shortcuts are
743 allowed in naming header variables:
748 The initiating <varname>$header_</varname>, <varname>$rheader_</varname>, or <varname>$bheader_</varname> can be
749 abbreviated to <varname>$h_</varname>, <varname>$rh_</varname>, or <varname>$bh_</varname>, respectively.
754 The terminating colon can be omitted if the next character is white space. The
755 white space character is retained in the expanded string. However, this is not
756 recommended, because it makes it easy to forget the colon when it really is
762 If the message does not contain a header of the given name, an empty string is
763 substituted. Thus it is important to spell the names of headers correctly. Do
764 not use <varname>$header_Reply_to</varname> when you really mean <varname>$header_Reply-to</varname>.
768 <title>User variables</title>
770 There are ten user variables with names <varname>$n0</varname> – <varname>$n9</varname> that can be
771 incremented by the <command>add</command> command (see section <xref linkend="SECTadd"/>). These can be
772 used for <quote>scoring</quote> messages in various ways. If Exim is configured to run a
773 <quote>system filter</quote> on every message, the values left in these variables are
774 copied into the variables <varname>$sn0</varname> – <varname>$sn9</varname> at the end of the system filter,
775 thus making them available to users’ filter files. How these values are used is
776 entirely up to the individual installation.
780 <title>Current directory</title>
782 The contents of your filter file should not make any assumptions about the
783 current directory. It is best to use absolute paths for file names; you can
784 normally make use of the <varname>$home</varname> variable to refer to your home directory. The
785 <command>save</command> command automatically inserts <varname>$home</varname> at the start of non-absolute
789 <section id="SECTsigdel">
790 <title>Significant deliveries</title>
792 When in the course of delivery a message is processed by a filter file, what
793 happens next, that is, after the filter file has been processed, depends on
794 whether or not the filter sets up any <emphasis>significant deliveries</emphasis>. If at least
795 one significant delivery is set up, the filter is considered to have handled
796 the entire delivery arrangements for the current address, and no further
797 processing of the address takes place. If, however, no significant deliveries
798 are set up, Exim continues processing the current address as if there were no
799 filter file, and typically sets up a delivery of a copy of the message into a
800 local mailbox. In particular, this happens in the special case of a filter file
801 containing only comments.
804 The delivery commands <command>deliver</command>, <command>save</command>, and <command>pipe</command> are by default
805 significant. However, if such a command is preceded by the word <quote>unseen</quote>, its
806 delivery is not considered to be significant. In contrast, other commands such
807 as <command>mail</command> and <command>vacation</command> do not set up significant deliveries unless
808 preceded by the word <quote>seen</quote>. The following example commands set up
809 significant deliveries:
811 <literallayout class="monospaced">
812 deliver jack@beanstalk.example
813 pipe $home/bin/mymailscript
814 seen mail subject "message discarded"
818 The following example commands do not set up significant deliveries:
820 <literallayout class="monospaced">
821 unseen deliver jack@beanstalk.example
822 unseen pipe $home/bin/mymailscript
823 mail subject "message discarded"
828 <title>Filter commands</title>
830 The filter commands that are described in subsequent sections are listed
831 below, with the section in which they are described in brackets:
833 <informaltable frame="none">
834 <tgroup cols="2" colsep="0" rowsep="0">
835 <colspec colwidth="100pt" align="left"/>
836 <colspec colwidth="300pt" align="left"/>
839 <entry><command>add</command></entry>
840 <entry> increment a user variable (section <xref linkend="SECTadd"/>)</entry>
843 <entry><command>deliver</command></entry>
844 <entry> deliver to an email address (section <xref linkend="SECTdeliver"/>)</entry>
847 <entry><command>fail</command></entry>
848 <entry> force delivery failure (sysadmin use) (section <xref linkend="SECTfail"/>)</entry>
851 <entry><command>finish</command></entry>
852 <entry> end processing (section <xref linkend="SECTfinish"/>)</entry>
855 <entry><command>freeze</command></entry>
856 <entry> freeze message (sysadmin use) (section <xref linkend="SECTfreeze"/>)</entry>
859 <entry><command>headers</command></entry>
860 <entry> set the header character set (section <xref linkend="SECTheaders"/>)</entry>
863 <entry><command>if</command></entry>
864 <entry> test condition(s) (section <xref linkend="SECTif"/>)</entry>
867 <entry><command>logfile</command></entry>
868 <entry> define log file (section <xref linkend="SECTlog"/>)</entry>
871 <entry><command>logwrite</command></entry>
872 <entry> write to log file (section <xref linkend="SECTlog"/>)</entry>
875 <entry><command>mail</command></entry>
876 <entry> send a reply message (section <xref linkend="SECTmail"/>)</entry>
879 <entry><command>pipe</command></entry>
880 <entry> pipe to a command (section <xref linkend="SECTpipe"/>)</entry>
883 <entry><command>save</command></entry>
884 <entry> save to a file (section <xref linkend="SECTsave"/>)</entry>
887 <entry><command>testprint</command></entry>
888 <entry> print while testing (section <xref linkend="SECTtestprint"/>)</entry>
891 <entry><command>vacation</command></entry>
892 <entry> tailored form of <command>mail</command> (section <xref linkend="SECTmail"/>)</entry>
898 The <command>headers</command> command has additional parameters that can be used only in a
899 system filter. The <command>fail</command> and <command>freeze</command> commands are available only when
900 Exim’s filtering facilities are being used as a system filter, and are
901 therefore usable only by the system administrator and not by ordinary users.
902 They are mentioned only briefly in this document; for more information, see the
903 main Exim specification.
906 <section id="SECTadd">
907 <title>The add command</title>
909 <literal> add </literal><<emphasis>number</emphasis>><literal> to </literal><<emphasis>user variable</emphasis>>
910 <literal>e.g. add 2 to n3</literal>
913 There are 10 user variables of this type, with names <varname>$n0</varname> – <varname>$n9</varname>. Their
914 values can be obtained by the normal expansion syntax (for example <varname>$n3</varname>) in
915 other commands. At the start of filtering, these variables all contain zero.
916 Both arguments of the <command>add</command> command are expanded before use, making it
917 possible to add variables to each other. Subtraction can be obtained by adding
921 <section id="SECTdeliver">
922 <title>The deliver command</title>
924 <literal> deliver</literal> <<emphasis>mail address</emphasis>>
925 <literal>e.g. deliver "Dr Livingstone <David@somewhere.africa.example>"</literal>
928 This command provides a forwarding operation. The delivery that it sets up is
929 significant unless the command is preceded by <quote>unseen</quote> (see section
930 <xref linkend="SECTsigdel"/>). The message is sent on to the given address, exactly as
931 happens if the address had appeared in a traditional <filename>.forward</filename> file. If you
932 want to deliver the message to a number of different addresses, you can use
933 more than one <command>deliver</command> command (each one may have only one address).
934 However, duplicate addresses are discarded.
937 To deliver a copy of the message to your normal mailbox, your login name can be
938 given as the address. Once an address has been processed by the filtering
939 mechanism, an identical generated address will not be so processed again, so
940 doing this does not cause a loop.
943 However, if you have a mail alias, you should <emphasis>not</emphasis> refer to it here. For
944 example, if the mail address <emphasis>L.Gulliver</emphasis> is aliased to <emphasis>lg303</emphasis> then all
945 references in Gulliver’s <filename>.forward</filename> file should be to <emphasis>lg303</emphasis>. A reference
946 to the alias will not work for messages that are addressed to that alias,
947 since, like <filename>.forward</filename> file processing, aliasing is performed only once on an
948 address, in order to avoid looping.
951 Following the new address, an optional second address, preceded by
952 <quote>errors_to</quote> may appear. This changes the address to which delivery errors on
953 the forwarded message will be sent. Instead of going to the message’s original
954 sender, they go to this new address. For ordinary users, the only value that is
955 permitted for this address is the user whose filter file is being processed.
956 For example, the user <emphasis>lg303</emphasis> whose mailbox is in the domain
957 <emphasis>lilliput.example</emphasis> could have a filter file that contains
959 <literallayout class="monospaced">
960 deliver jon@elsewhere.example errors_to lg303@lilliput.example
963 Clearly, using this feature makes sense only in situations where not all
964 messages are being forwarded. In particular, bounce messages must not be
965 forwarded in this way, as this is likely to create a mail loop if something
969 <section id="SECTsave">
970 <title>The save command</title>
972 <literal> save </literal><<emphasis>file name</emphasis>>
973 <literal>e.g. save $home/mail/bookfolder</literal>
976 This command specifies that a copy of the message is to be appended to the
977 given file (that is, the file is to be used as a mail folder). The delivery
978 that <command>save</command> sets up is significant unless the command is preceded by
979 <quote>unseen</quote> (see section <xref linkend="SECTsigdel"/>).
982 More than one <command>save</command> command may be obeyed; each one causes a copy of the
983 message to be written to its argument file, provided they are different
984 (duplicate <command>save</command> commands are ignored).
987 If the file name does not start with a / character, the contents of the
988 <varname>$home</varname> variable are prepended, unless it is empty, <phrase revisionflag="changed">or the system
989 administrator has disabled this feature.</phrase> In conventional configurations, this
990 variable is normally set in a user filter to the user’s home directory, but the
991 system administrator may set it to some other path. In some configurations,
992 <varname>$home</varname> may be unset, <phrase revisionflag="changed">or prepending may be disabled,</phrase> in which case a
993 non-absolute path name may be generated. Such configurations convert this to an
994 absolute path when the delivery takes place. In a system filter, <varname>$home</varname> is
998 The user must of course have permission to write to the file, and the writing
999 of the file takes place in a process that is running as the user, under the
1000 user’s primary group. Any secondary groups to which the user may belong are not
1001 normally taken into account, though the system administrator can configure Exim
1002 to set them up. In addition, the ability to use this command at all is
1003 controlled by the system administrator – it may be forbidden on some systems.
1006 An optional mode value may be given after the file name. The value for the mode
1007 is interpreted as an octal number, even if it does not begin with a zero. For
1010 <literallayout class="monospaced">
1011 save /some/folder 640
1014 This makes it possible for users to override the system-wide mode setting for
1015 file deliveries, which is normally 600. If an existing file does not have the
1016 correct mode, it is changed.
1019 An alternative form of delivery may be enabled on your system, in which each
1020 message is delivered into a new file in a given directory. If this is the case,
1021 this functionality can be requested by giving the directory name terminated by
1022 a slash after the <command>save</command> command, for example
1024 <literallayout class="monospaced">
1025 save separated/messages/
1028 There are several different formats for such deliveries; check with your system
1029 administrator or local documentation to find out which (if any) are available
1030 on your system. If this functionality is not enabled, the use of a path name
1031 ending in a slash causes an error.
1034 <section id="SECTpipe">
1035 <title>The pipe command</title>
1037 <literal> pipe </literal><<emphasis>command</emphasis>>
1038 <literal>e.g. pipe "$home/bin/countmail $sender_address"</literal>
1041 This command specifies that the message is to be delivered to the specified
1042 command using a pipe. The delivery that it sets up is significant unless the
1043 command is preceded by <quote>unseen</quote> (see section <xref linkend="SECTsigdel"/>). Remember,
1044 however, that no deliveries are done while the filter is being processed. All
1045 deliveries happen later on. Therefore, the result of running the pipe is not
1046 available to the filter.
1049 When the deliveries are done, a separate process is run, and a copy of the
1050 message is passed on its standard input. The process runs as the user, under
1051 the user’s primary group. Any secondary groups to which the user may belong are
1052 not normally taken into account, though the system administrator can configure
1053 Exim to set them up. More than one <command>pipe</command> command may appear; each one causes
1054 a copy of the message to be written to its argument pipe, provided they are
1055 different (duplicate <command>pipe</command> commands are ignored).
1058 When the time comes to transport the message, the command supplied to <command>pipe</command>
1059 is split up by Exim into a command name and a number of arguments. These are
1060 delimited by white space except for arguments enclosed in double quotes, in
1061 which case backslash is interpreted as an escape, or in single quotes, in which
1062 case no escaping is recognized. Note that as the whole command is normally
1063 supplied in double quotes, a second level of quoting is required for internal
1064 double quotes. For example:
1066 <literallayout class="monospaced">
1067 pipe "$home/myscript \"size is $message_size\""
1070 String expansion is performed on the separate components after the line has
1071 been split up, and the command is then run directly by Exim; it is not run
1072 under a shell. Therefore, substitution cannot change the number of arguments,
1073 nor can quotes, backslashes or other shell metacharacters in variables cause
1077 Documentation for some programs that are normally run via this kind of pipe
1078 often suggest that the command should start with
1080 <literallayout class="monospaced">
1084 This is a shell command, and should <emphasis>not</emphasis> be present in Exim filter files,
1085 since it does not normally run the command under a shell.
1088 However, there is an option that the administrator can set to cause a shell to
1089 be used. In this case, the entire command is expanded as a single string and
1090 passed to the shell for interpretation. It is recommended that this be avoided
1091 if at all possible, since it can lead to problems when inserted variables
1092 contain shell metacharacters.
1095 The default PATH set up for the command is determined by the system
1096 administrator, usually containing at least <filename>/bin</filename> and <filename>/usr/bin</filename> so that
1097 common commands are available without having to specify an absolute file name.
1098 However, it is possible for the system administrator to restrict the pipe
1099 facility so that the command name must not contain any / characters, and must
1100 be found in one of the directories in the configured PATH. It is also possible
1101 for the system administrator to lock out the use of the <command>pipe</command> command
1105 When the command is run, a number of environment variables are set up. The
1106 complete list for pipe deliveries may be found in the Exim reference manual.
1107 Those that may be useful for pipe deliveries from user filter files are:
1110 <literal>DOMAIN </literal> the domain of the address
1111 <literal>HOME </literal> your home directory
1112 <literal>LOCAL_PART </literal> see below
1113 <literal>LOCAL_PART_PREFIX </literal> see below
1114 <literal>LOCAL_PART_SUFFIX </literal> see below
1115 <literal>LOGNAME </literal> your login name
1116 <literal>MESSAGE_ID </literal> the unique id of the message
1117 <literal>PATH </literal> the command search path
1118 <literal>RECIPIENT </literal> the complete recipient address
1119 <literal>SENDER </literal> the sender of the message
1120 <literal>SHELL </literal> <literal>/bin/sh</literal>
1121 <literal>USER </literal> see below
1124 LOCAL_PART, LOGNAME, and USER are all set to the same value, namely, your login
1125 id. LOCAL_PART_PREFIX and LOCAL_PART_SUFFIX may be set if Exim is configured to
1126 recognize prefixes or suffixes in the local parts of addresses. For example, a
1127 message addressed to <emphasis>pat-suf2@domain.example</emphasis> may cause the filter for user
1128 <emphasis>pat</emphasis> to be run. If this sets up a pipe delivery, LOCAL_PART_SUFFIX is
1129 <literal>-suf2</literal> when the pipe command runs. The system administrator has to configure
1130 Exim specially for this feature to be available.
1133 If you run a command that is a shell script, be very careful in your use of
1134 data from the incoming message in the commands in your script. RFC 2822 is very
1135 generous in the characters that are permitted to appear in mail addresses, and
1136 in particular, an address may begin with a vertical bar or a slash. For this
1137 reason you should always use quotes round any arguments that involve data from
1138 the message, like this:
1140 <literallayout class="monospaced">
1141 /some/command '$SENDER'
1144 so that inserted shell meta-characters do not cause unwanted effects.
1147 Remember that, as was explained earlier, the pipe command is not run at the
1148 time the filter file is interpreted. The filter just defines what deliveries
1149 are required for one particular addressee of a message. The deliveries
1150 themselves happen later, once Exim has decided everything that needs to be done
1154 A consequence of this is that you cannot inspect the return code from the pipe
1155 command from within the filter. Nevertheless, the code returned by the command
1156 is important, because Exim uses it to decide whether the delivery has succeeded
1160 The command should return a zero completion code if all has gone well. Most
1161 non-zero codes are treated by Exim as indicating a failure of the pipe. This is
1162 treated as a delivery failure, causing the message to be returned to its
1163 sender. However, there are some completion codes that are treated as temporary
1164 errors. The message remains on Exim’s spool disk, and the delivery is tried
1165 again later, though it will ultimately time out if the delivery failures go on
1166 too long. The completion codes to which this applies can be specified by the
1167 system administrator; the default values are 73 and 75.
1170 The pipe command should not normally write anything to its standard output or
1171 standard error file descriptors. If it does, whatever is written is normally
1172 returned to the sender of the message as a delivery error, though this action
1173 can be varied by the system administrator.
1176 <section id="SECTmail">
1177 <title>Mail commands</title>
1179 There are two commands that cause the creation of a new mail message, neither
1180 of which count as a significant delivery unless the command is preceded by the
1181 word <quote>seen</quote> (see section <xref linkend="SECTsigdel"/>). This is a powerful facility, but
1182 it should be used with care, because of the danger of creating infinite
1183 sequences of messages. The system administrator can forbid the use of these
1184 commands altogether.
1187 To help prevent runaway message sequences, these commands have no effect when
1188 the incoming message is a bounce (delivery error) message, and messages sent by
1189 this means are treated as if they were reporting delivery errors. Thus, they
1190 should never themselves cause a bounce message to be returned. The basic
1191 mail-sending command is
1194 <literal>mail [to </literal><<emphasis>address-list</emphasis>><literal>]</literal>
1195 <literal> [cc </literal><<emphasis>address-list</emphasis>><literal>]</literal>
1196 <literal> [bcc </literal><<emphasis>address-list</emphasis>><literal>]</literal>
1197 <literal> [from </literal><<emphasis>address</emphasis>><literal>]</literal>
1198 <literal> [reply_to </literal><<emphasis>address</emphasis>><literal>]</literal>
1199 <literal> [subject </literal><<emphasis>text</emphasis>><literal>]</literal>
1200 <literal> [extra_headers </literal><<emphasis>text</emphasis>><literal>]</literal>
1201 <literal> [text </literal><<emphasis>text</emphasis>><literal>]</literal>
1202 <literal> [[expand] file </literal><<emphasis>filename</emphasis>><literal>]</literal>
1203 <literal> [return message]</literal>
1204 <literal> [log </literal><<emphasis>log file name</emphasis>><literal>]</literal>
1205 <literal> [once </literal><<emphasis>note file name</emphasis>><literal>]</literal>
1206 <literal> [once_repeat </literal><<emphasis>time interval</emphasis>><literal>]</literal>
1208 <literal>e.g. mail text "Your message about $h_subject: has been received"</literal>
1211 Each <<emphasis>address-list</emphasis>> can contain a number of addresses, separated by commas,
1212 in the format of a <emphasis>To:</emphasis> or <emphasis>Cc:</emphasis> header line. In fact, the text you supply
1213 here is copied exactly into the appropriate header line. It may contain
1214 additional information as well as email addresses. For example:
1216 <literallayout class="monospaced">
1217 mail to "Julius Caesar <jc@rome.example>, \
1218 <ma@rome.example> (Mark A.)"
1221 Similarly, the texts supplied for <option>from</option> and <option>reply_to</option> are copied into
1222 their respective header lines.
1225 As a convenience for use in one common case, there is also a command called
1226 <command>vacation</command>. It behaves in the same way as <command>mail</command>, except that the defaults
1227 for the <option>subject</option>, <option>file</option>, <option>log</option>, <option>once</option>, and <option>once_repeat</option> options
1230 <literallayout class="monospaced">
1231 subject "On vacation"
1232 expand file .vacation.msg
1238 respectively. These are the same file names and repeat period used by the
1239 traditional Unix <command>vacation</command> command. The defaults can be overridden by
1240 explicit settings, but if a file name is given its contents are expanded only
1241 if explicitly requested.
1244 <emphasis role="bold">Warning</emphasis>: The <command>vacation</command> command should always be used conditionally,
1245 subject to at least the <command>personal</command> condition (see section <xref linkend="SECTpersonal"/>
1246 below) so as not to send automatic replies to non-personal messages from
1247 mailing lists or elsewhere. Sending an automatic response to a mailing list or
1248 a mailing list manager is an Internet Sin.
1251 For both commands, the key/value argument pairs can appear in any order. At
1252 least one of <option>text</option> or <option>file</option> must appear (except with <command>vacation</command>, where
1253 there is a default for <option>file</option>); if both are present, the text string appears
1254 first in the message. If <option>expand</option> precedes <option>file</option>, each line of the file is
1255 subject to string expansion before it is included in the message.
1258 Several lines of text can be supplied to <option>text</option> by including the escape
1259 sequence <quote>\n</quote> in the string wherever a newline is required. If the command is
1260 output during filter file testing, newlines in the text are shown as <quote>\n</quote>.
1263 Note that the keyword for creating a <emphasis>Reply-To:</emphasis> header is <option>reply_to</option>,
1264 because Exim keywords may contain underscores, but not hyphens. If the <option>from</option>
1265 keyword is present and the given address does not match the user who owns the
1266 forward file, Exim normally adds a <emphasis>Sender:</emphasis> header to the message, though it
1267 can be configured not to do this.
1270 The <option>extra_headers</option> keyword allows you to add custom header lines to the
1271 message. The text supplied must be one or more syntactically valid RFC 2822
1272 header lines. You can use <quote>\n</quote> within quoted text to specify newlines between
1273 headers, and also to define continued header lines. For example:
1275 <literallayout class="monospaced">
1276 extra_headers "h1: first\nh2: second\n continued\nh3: third"
1279 No newline should appear at the end of the final header line.
1282 If no <option>to</option> argument appears, the message is sent to the address in the
1283 <varname>$reply_address</varname> variable (see section <xref linkend="SECTfilterstringexpansion"/> above).
1284 An <emphasis>In-Reply-To:</emphasis> header is automatically included in the created message,
1285 giving a reference to the message identification of the incoming message.
1288 If <option>return message</option> is specified, the incoming message that caused the filter
1289 file to be run is added to the end of the message, subject to a maximum size
1293 If a log file is specified, a line is added to it for each message sent.
1296 If a <option>once</option> file is specified, it is used to hold a database for remembering
1297 who has received a message, and no more than one message is ever sent to any
1298 particular address, unless <option>once_repeat</option> is set. This specifies a time
1299 interval after which another copy of the message is sent. The interval is
1300 specified as a sequence of numbers, each followed by the initial letter of one
1301 of <quote>seconds</quote>, <quote>minutes</quote>, <quote>hours</quote>, <quote>days</quote>, or <quote>weeks</quote>. For example,
1303 <literallayout class="monospaced">
1307 causes a new message to be sent if at least 5 days and 4 hours have elapsed
1308 since the last one was sent. There must be no white space in a time interval.
1311 Commonly, the file name specified for <option>once</option> is used as the base name for
1312 direct-access (DBM) file operations. There are a number of different DBM
1313 libraries in existence. Some operating systems provide one as a default, but
1314 even in this case a different one may have been used when building Exim. With
1315 some DBM libraries, specifying <option>once</option> results in two files being created,
1316 with the suffixes <filename>.dir</filename> and <filename>.pag</filename> being added to the given name. With
1317 some others a single file with the suffix <filename>.db</filename> is used, or the name is used
1321 Using a DBM file for implementing the <option>once</option> feature means that the file
1322 grows as large as necessary. This is not usually a problem, but some system
1323 administrators want to put a limit on it. The facility can be configured not to
1324 use a DBM file, but instead, to use a regular file with a maximum size. The
1325 data in such a file is searched sequentially, and if the file fills up, the
1326 oldest entry is deleted to make way for a new one. This means that some
1327 correspondents may receive a second copy of the message after an unpredictable
1328 interval. Consult your local information to see if your system is configured
1332 More than one <command>mail</command> or <command>vacation</command> command may be obeyed in a single filter
1333 run; they are all honoured, even when they are to the same recipient.
1336 <section id="SECTlog">
1337 <title>Logging commands</title>
1339 A log can be kept of actions taken by a filter file. This facility is normally
1340 available in conventional configurations, but there are some situations where
1341 it might not be. Also, the system administrator may choose to disable it. Check
1342 your local information if in doubt.
1345 Logging takes place while the filter file is being interpreted. It does not
1346 queue up for later like the delivery commands. The reason for this is so that a
1347 log file need be opened only once for several write operations. There are two
1348 commands, neither of which constitutes a significant delivery. The first
1349 defines a file to which logging output is subsequently written:
1352 <literal> logfile </literal><<emphasis>file name</emphasis>>
1353 <literal>e.g. logfile $home/filter.log</literal>
1356 The file name must be fully qualified. You can use <varname>$home</varname>, as in this
1357 example, to refer to your home directory. The file name may optionally be
1358 followed by a mode for the file, which is used if the file has to be created.
1361 <literallayout class="monospaced">
1362 logfile $home/filter.log 0644
1365 The number is interpreted as octal, even if it does not begin with a zero.
1366 The default for the mode is 600. It is suggested that the <command>logfile</command> command
1367 normally appear as the first command in a filter file. Once a log file has
1368 been obeyed, the <command>logwrite</command> command can be used to write to it:
1371 <literal> logwrite "</literal><<emphasis>some text string</emphasis>><literal>"</literal>
1372 <literal>e.g. logwrite "$tod_log $message_id processed"</literal>
1375 It is possible to have more than one <command>logfile</command> command, to specify writing to
1376 different log files in different circumstances. Writing takes place at the end
1377 of the file, and a newline character is added to the end of each string if
1378 there isn’t one already there. Newlines can be put in the middle of the string
1379 by using the <quote>\n</quote> escape sequence. Lines from simultaneous deliveries may get
1380 interleaved in the file, as there is no interlocking, so you should plan your
1381 logging with this in mind. However, data should not get lost.
1384 <section id="SECTfinish">
1385 <title>The finish command</title>
1387 The command <command>finish</command>, which has no arguments, causes Exim to stop
1388 interpreting the filter file. This is not a significant action unless preceded
1389 by <quote>seen</quote>. A filter file containing only <quote>seen finish</quote> is a black hole.
1392 <section id="SECTtestprint">
1393 <title>The testprint command</title>
1395 It is sometimes helpful to be able to print out the values of variables when
1396 testing filter files. The command
1399 <literal> testprint </literal><<emphasis>text</emphasis>>
1400 <literal>e.g. testprint "home=$home reply_address=$reply_address"</literal>
1403 does nothing when mail is being delivered. However, when the filtering code is
1404 being tested by means of the <option>-bf</option> option (see section <xref linkend="SECTtesting"/>
1405 above), the value of the string is written to the standard output.
1408 <section id="SECTfail">
1409 <title>The fail command</title>
1411 When Exim’s filtering facilities are being used as a system filter, the
1412 <command>fail</command> command is available, to force delivery failure. Because this command
1413 is normally usable only by the system administrator, and not enabled for use by
1414 ordinary users, it is described in more detail in the main Exim specification
1415 rather than in this document.
1418 <section id="SECTfreeze">
1419 <title>The freeze command</title>
1421 When Exim’s filtering facilities are being used as a system filter, the
1422 <command>freeze</command> command is available, to freeze a message on the queue. Because this
1423 command is normally usable only by the system administrator, and not enabled
1424 for use by ordinary users, it is described in more detail in the main Exim
1425 specification rather than in this document.
1428 <section id="SECTheaders">
1429 <title>The headers command</title>
1431 The <command>headers</command> command can be used to change the target character set that is
1432 used when translating the contents of encoded header lines for insertion by the
1433 <varname>$header_</varname> mechanism (see section <xref linkend="SECTheadervariables"/> above). The
1434 default can be set in the Exim configuration; if not specified, ISO-8859-1 is
1435 used. The only currently supported format for the <command>headers</command> command in user
1436 filters is as in this example:
1438 <literallayout class="monospaced">
1439 headers charset "UTF-8"
1442 That is, <command>headers</command> is followed by the word <quote>charset</quote> and then the name of a
1443 character set. This particular example would be useful if you wanted to compare
1444 the contents of a header to a UTF-8 string.
1447 In system filter files, the <command>headers</command> command can be used to add or remove
1448 header lines from the message. These features are described in the main Exim
1452 <section id="SECTif">
1453 <title>Obeying commands conditionally</title>
1455 Most of the power of filtering comes from the ability to test conditions and
1456 obey different commands depending on the outcome. The <command>if</command> command is used to
1457 specify conditional execution, and its general form is
1460 <literal>if </literal><<emphasis>condition</emphasis>>
1461 <literal>then </literal><<emphasis>commands</emphasis>>
1462 <literal>elif </literal><<emphasis>condition</emphasis>>
1463 <literal>then </literal><<emphasis>commands</emphasis>>
1464 <literal>else </literal><<emphasis>commands</emphasis>>
1465 <literal>endif</literal>
1468 There may be any number of <command>elif</command> and <command>then</command> sections (including none) and
1469 the <command>else</command> section is also optional. Any number of commands, including nested
1470 <command>if</command> commands, may appear in any of the <<emphasis>commands</emphasis>> sections.
1473 Conditions can be combined by using the words <command>and</command> and <command>or</command>, and round
1474 brackets (parentheses) can be used to specify how several conditions are to
1475 combine. Without brackets, <command>and</command> is more binding than <command>or</command>. For example:
1477 <literallayout class="monospaced">
1479 $h_subject: contains "Make money" or
1480 $h_precedence: is "junk" or
1481 ($h_sender: matches ^\\d{8}@ and not personal) or
1482 $message_body contains "this is not spam"
1488 A condition can be preceded by <command>not</command> to negate it, and there are also some
1489 negative forms of condition that are more English-like.
1493 <title>String testing conditions</title>
1495 There are a number of conditions that operate on text strings, using the words
1496 <quote>begins</quote>, <quote>ends</quote>, <quote>is</quote>, <quote>contains</quote> and <quote>matches</quote>. If you want to
1497 apply the same test to more than one header line, you can easily concatenate
1498 them into a single string for testing, as in this example:
1500 <literallayout class="monospaced">
1501 if "$h_to:, $h_cc:" contains me@domain.example then ...
1504 If a string-testing condition name is written in lower case, the testing
1505 of letters is done without regard to case; if it is written in upper case
1506 (for example, <quote>CONTAINS</quote>), the case of letters is taken into account.
1509 <literal> </literal><<emphasis>text1</emphasis>><literal> begins </literal><<emphasis>text2</emphasis>>
1510 <literal> </literal><<emphasis>text1</emphasis>><literal> does not begin </literal><<emphasis>text2</emphasis>>
1511 <literal>e.g. $header_from: begins "Friend@"</literal>
1514 A <quote>begins</quote> test checks for the presence of the second string at the start of
1515 the first, both strings having been expanded.
1518 <literal> </literal><<emphasis>text1</emphasis>><literal> ends </literal><<emphasis>text2</emphasis>>
1519 <literal> </literal><<emphasis>text1</emphasis>><literal> does not end </literal><<emphasis>text2</emphasis>>
1520 <literal>e.g. $header_from: ends "public.com.example"</literal>
1523 An <quote>ends</quote> test checks for the presence of the second string at the end of
1524 the first, both strings having been expanded.
1527 <literal> </literal><<emphasis>text1</emphasis>><literal> is </literal><<emphasis>text2</emphasis>>
1528 <literal> </literal><<emphasis>text1</emphasis>><literal> is not </literal><<emphasis>text2</emphasis>>
1529 <literal>e.g. $local_part_suffix is "-foo"</literal>
1532 An <quote>is</quote> test does an exact match between the strings, having first expanded
1536 <literal> </literal><<emphasis>text1</emphasis>><literal> contains </literal><<emphasis>text2</emphasis>>
1537 <literal> </literal><<emphasis>text1</emphasis>><literal> does not contain </literal><<emphasis>text2</emphasis>>
1538 <literal>e.g. $header_subject: contains "evolution"</literal>
1541 A <quote>contains</quote> test does a partial string match, having expanded both strings.
1544 <literal> </literal><<emphasis>text1</emphasis>><literal> matches </literal><<emphasis>text2</emphasis>>
1545 <literal> </literal><<emphasis>text1</emphasis>><literal> does not match </literal><<emphasis>text2</emphasis>>
1546 <literal>e.g. $sender_address matches "(bill|john)@"</literal>
1549 For a <quote>matches</quote> test, after expansion of both strings, the second one is
1550 interpreted as a regular expression. Exim uses the PCRE regular expression
1551 library, which provides regular expressions that are compatible with Perl.
1554 The match succeeds if the regular expression matches any part of the first
1555 string. If you want a regular expression to match only at the start or end of
1556 the subject string, you must encode that requirement explicitly, using the
1557 <literal>^</literal> or <literal>$</literal> metacharacters. The above example, which is not so constrained,
1558 matches all these addresses:
1560 <literallayout class="monospaced">
1563 spoonbill@example.com
1564 littlejohn@example.com
1567 To match only the first two, you could use this:
1569 <literallayout class="monospaced">
1570 if $sender_address matches "^(bill|john)@" then ...
1573 Care must be taken if you need a backslash in a regular expression, because
1574 backslashes are interpreted as escape characters both by the string expansion
1575 code and by Exim’s normal processing of strings in quotes. For example, if you
1576 want to test the sender address for a domain ending in <emphasis>.com</emphasis> the regular
1579 <literallayout class="monospaced">
1583 The backslash and dollar sign in that expression have to be escaped when used
1584 in a filter command, as otherwise they would be interpreted by the expansion
1585 code. Thus, what you actually write is
1587 <literallayout class="monospaced">
1588 if $sender_address matches \\.com\$
1591 An alternative way of handling this is to make use of the <literal>\N</literal> expansion
1592 flag for suppressing expansion:
1594 <literallayout class="monospaced">
1595 if $sender_address matches \N\.com$\N
1598 Everything between the two occurrences of <literal>\N</literal> is copied without change by
1599 the string expander (and in fact you do not need the final one, because it is
1600 at the end of the string). If the regular expression is given in quotes
1601 (mandatory only if it contains white space) you have to write either
1603 <literallayout class="monospaced">
1604 if $sender_address matches "\\\\.com\\$"
1609 <literallayout class="monospaced">
1610 if $sender_address matches "\\N\\.com$\\N"
1613 If the regular expression contains bracketed sub-expressions, numeric
1614 variable substitutions such as <varname>$1</varname> can be used in the subsequent actions
1615 after a successful match. If the match fails, the values of the numeric
1616 variables remain unchanged. Previous values are not restored after <command>endif</command>.
1617 In other words, only one set of values is ever available. If the condition
1618 contains several sub-conditions connected by <command>and</command> or <command>or</command>, it is the
1619 strings extracted from the last successful match that are available in
1620 subsequent actions. Numeric variables from any one sub-condition are also
1621 available for use in subsequent sub-conditions, because string expansion of a
1622 condition occurs just before it is tested.
1626 <title>Numeric testing conditions</title>
1628 The following conditions are available for performing numerical tests:
1631 <literal> </literal><<emphasis>number1</emphasis>><literal> is above </literal><<emphasis>number2</emphasis>>
1632 <literal> </literal><<emphasis>number1</emphasis>><literal> is not above </literal><<emphasis>number2</emphasis>>
1633 <literal> </literal><<emphasis>number1</emphasis>><literal> is below </literal><<emphasis>number2</emphasis>>
1634 <literal> </literal><<emphasis>number1</emphasis>><literal> is not below </literal><<emphasis>number2</emphasis>>
1635 <literal>e.g. $message_size is not above 10k</literal>
1638 The <<emphasis>number</emphasis>> arguments must expand to strings of digits, optionally
1639 followed by one of the letters K or M (upper case or lower case) which cause
1640 multiplication by 1024 and 1024x1024 respectively.
1644 <title>Testing for significant deliveries</title>
1646 You can use the <command>delivered</command> condition to test whether or not any previously
1647 obeyed filter commands have set up a significant delivery. For example:
1649 <literallayout class="monospaced">
1650 if not delivered then save mail/anomalous endif
1653 <quote>Delivered</quote> is perhaps a poor choice of name for this condition, because the
1654 message has not actually been delivered; rather, a delivery has been set up for
1659 <title>Testing for error messages</title>
1661 The condition <command>error_message</command> is true if the incoming message is a bounce
1662 (mail delivery error) message. Putting the command
1664 <literallayout class="monospaced">
1665 if error_message then finish endif
1668 at the head of your filter file is a useful insurance against things going
1669 wrong in such a way that you cannot receive delivery error reports. <emphasis role="bold">Note</emphasis>:
1670 <command>error_message</command> is a condition, not an expansion variable, and therefore is
1671 not preceded by <literal>$</literal>.
1675 <title>Testing a list of addresses</title>
1677 There is a facility for looping through a list of addresses and applying a
1678 condition to each of them. It takes the form
1681 <literal>foranyaddress </literal><<emphasis>string</emphasis>><literal> (</literal><<emphasis>condition</emphasis>><literal>)</literal>
1684 where <<emphasis>string</emphasis>> is interpreted as a list of RFC 2822 addresses, as in a
1685 typical header line, and <<emphasis>condition</emphasis>> is any valid filter condition or
1686 combination of conditions. The <quote>group</quote> syntax that is defined for certain
1687 header lines that contain addresses is supported.
1690 The parentheses surrounding the condition are mandatory, to delimit it from
1691 possible further sub-conditions of the enclosing <command>if</command> command. Within the
1692 condition, the expansion variable <varname>$thisaddress</varname> is set to the non-comment
1693 portion of each of the addresses in the string in turn. For example, if the
1696 <literallayout class="monospaced">
1697 B.Simpson <bart@sfld.example>, lisa@sfld.example (his sister)
1700 then <varname>$thisaddress</varname> would take on the values <literal>bart@sfld.example</literal> and
1701 <literal>lisa@sfld.example</literal> in turn.
1704 If there are no valid addresses in the list, the whole condition is false. If
1705 the internal condition is true for any one address, the overall condition is
1706 true and the loop ends. If the internal condition is false for all addresses in
1707 the list, the overall condition is false. This example tests for the presence
1708 of an eight-digit local part in any address in a <emphasis>To:</emphasis> header:
1710 <literallayout class="monospaced">
1711 if foranyaddress $h_to: ( $thisaddress matches ^\\d{8}@ ) then ...
1714 When the overall condition is true, the value of <varname>$thisaddress</varname> in the
1715 commands that follow <command>then</command> is the last value it took on inside the loop. At
1716 the end of the <command>if</command> command, the value of <varname>$thisaddress</varname> is reset to what it
1717 was before. It is best to avoid the use of multiple occurrences of
1718 <command>foranyaddress</command>, nested or otherwise, in a single <command>if</command> command, if the
1719 value of <varname>$thisaddress</varname> is to be used afterwards, because it isn’t always
1720 clear what the value will be. Nested <command>if</command> commands should be used instead.
1723 Header lines can be joined together if a check is to be applied to more than
1724 one of them. For example:
1726 <literallayout class="monospaced">
1727 if foranyaddress $h_to:,$h_cc: ....
1730 This scans through the addresses in both the <emphasis>To:</emphasis> and the <emphasis>Cc:</emphasis> headers.
1733 <section id="SECTpersonal">
1734 <title>Testing for personal mail</title>
1736 A common requirement is to distinguish between incoming personal mail and mail
1737 from a mailing list, or from a robot or other automatic process (for example, a
1738 bounce message). In particular, this test is normally required for <quote>vacation
1742 The <command>personal</command> condition checks that the message is not a bounce message and
1743 that the current user’s email address appears in the <emphasis>To:</emphasis> header. It also
1744 checks that the sender is not the current user or one of a number of common
1745 daemons, and that there are no header lines starting <emphasis>List-</emphasis> in the message.
1746 Finally, it checks the content of the <emphasis>Precedence:</emphasis> header line, if there is
1750 You should always use the <command>personal</command> condition when generating automatic
1751 responses. This example shows the use of <command>personal</command> in a filter file that is
1752 sending out vacation messages:
1754 <literallayout class="monospaced">
1756 mail to $reply_address
1757 subject "I am on holiday"
1758 file $home/vacation/message
1759 once $home/vacation/once
1764 It is tempting, when writing commands like the above, to quote the original
1765 subject in the reply. For example:
1767 <literallayout class="monospaced">
1768 subject "Re: $h_subject:"
1771 There is a danger in doing this, however. It may allow a third party to
1772 subscribe you to an opt-in mailing list, provided that the list accepts bounce
1773 messages as subscription confirmations. (Messages sent from filters are always
1774 sent as bounce messages.) Well-managed lists require a non-bounce message to
1775 confirm a subscription, so the danger is relatively small.
1778 If prefixes or suffixes are in use for local parts – something which depends
1779 on the configuration of Exim (see section <xref linkend="SECTmbox"/> below) – the tests
1780 for the current user are done with the full address (including the prefix and
1781 suffix, if any) as well as with the prefix and suffix removed. If the system is
1782 configured to rewrite local parts of mail addresses, for example, to rewrite
1783 <literal>dag46</literal> as <literal>Dirk.Gently</literal>, the rewritten form of the address is also used in
1788 <title>Alias addresses for the personal condition</title>
1790 It is quite common for people who have mail accounts on a number of different
1791 systems to forward all their mail to one system, and in this case a check for
1792 personal mail should test all their various mail addresses. To allow for this,
1793 the <command>personal</command> condition keyword can be followed by
1796 <literal>alias </literal><<emphasis>address</emphasis>>
1799 any number of times, for example:
1801 <literallayout class="monospaced">
1802 if personal alias smith@else.where.example
1803 alias jones@other.place.example
1807 The alias addresses are treated as alternatives to the current user’s email
1808 address when testing the contents of header lines.
1812 <title>Details of the personal condition</title>
1814 The basic <command>personal</command> test is roughly equivalent to the following:
1816 <literallayout class="monospaced">
1817 not error_message and
1818 $message_headers does not contain "\nList-Id:" and
1819 $message_headers does not contain "\nList-Help:" and
1820 $message_headers does not contain "\nList-Subscribe:" and
1821 $message_headers does not contain "\nList-Unsubscribe:" and
1822 $message_headers does not contain "\nList-Post:" and
1823 $message_headers does not contain "\nList-Owner:" and
1824 $message_headers does not contain "\nList-Archive:" and
1826 "${if def h_auto-submitted:{present}{absent}}" is "absent" or
1827 $header_auto-submitted: is "no"
1829 $header_precedence: does not contain "bulk" and
1830 $header_precedence: does not contain "list" and
1831 $header_precedence: does not contain "junk" and
1832 foranyaddress $header_to:
1833 ( $thisaddress contains "$local_part$domain" ) and
1834 not foranyaddress $header_from:
1836 $thisaddress contains "$local_part@$domain" or
1837 $thisaddress contains "server@" or
1838 $thisaddress contains "daemon@" or
1839 $thisaddress contains "root@" or
1840 $thisaddress contains "listserv@" or
1841 $thisaddress contains "majordomo@" or
1842 $thisaddress contains "-request@" or
1843 $thisaddress matches "^owner-[^@]+@"
1847 The variable <varname>$local_part</varname> contains the local part of the mail address of
1848 the user whose filter file is being run – it is normally your login id. The
1849 <varname>$domain</varname> variable contains the mail domain. As explained above, if aliases
1850 or rewriting are defined, or if prefixes or suffixes are in use, the tests for
1851 the current user are also done with alternative addresses.
1855 <title>Testing delivery status</title>
1857 There are two conditions that are intended mainly for use in system filter
1858 files, but which are available in users’ filter files as well. The condition
1859 <command>first_delivery</command> is true if this is the first process that is attempting to
1860 deliver the message, and false otherwise. This indicator is not reset until the
1861 first delivery process successfully terminates; if there is a crash or a power
1862 failure (for example), the next delivery attempt is also a <quote>first delivery</quote>.
1865 In a user filter file <command>first_delivery</command> will be false if there was previously
1866 an error in the filter, or if a delivery for the user failed owing to, for
1867 example, a quota error, or if forwarding to a remote address was deferred for
1871 The condition <command>manually_thawed</command> is true if the message was <quote>frozen</quote> for
1872 some reason, and was subsequently released by the system administrator. It is
1873 unlikely to be of use in users’ filter files.
1876 <section id="SECTmbox">
1877 <title>Multiple personal mailboxes</title>
1879 The system administrator can configure Exim so that users can set up variants
1880 on their email addresses and handle them separately. Consult your system
1881 administrator or local documentation to see if this facility is enabled on your
1882 system, and if so, what the details are.
1885 The facility involves the use of a prefix or a suffix on an email address. For
1886 example, all mail addressed to <emphasis>lg303-</emphasis><<emphasis>something</emphasis>> would be the property
1887 of user <emphasis>lg303</emphasis>, who could determine how it was to be handled, depending on
1888 the value of <<emphasis>something</emphasis>>.
1891 There are two possible ways in which this can be set up. The first possibility
1892 is the use of multiple <filename>.forward</filename> files. In this case, mail to <emphasis>lg303-foo</emphasis>,
1893 for example, is handled by looking for a file called <filename>.forward-foo</filename> in
1894 <emphasis>lg303</emphasis>’s home directory. If such a file does not exist, delivery fails
1895 and the message is returned to its sender.
1898 The alternative approach is to pass all messages through a single <filename>.forward</filename>
1899 file, which must be a filter file so that it can distinguish between the
1900 different cases by referencing the variables <varname>$local_part_prefix</varname> or
1901 <varname>$local_part_suffix</varname>, as in the final example in section <xref linkend="SECTex"/> below.
1904 It is possible to configure Exim to support both schemes at once. In this case,
1905 a specific <filename>.forward-foo</filename> file is first sought; if it is not found, the basic
1906 <filename>.forward</filename> file is used.
1909 The <command>personal</command> test (see section <xref linkend="SECTpersonal"/>) includes prefixes and
1910 suffixes in its checking.
1914 <title>Ignoring delivery errors</title>
1916 As was explained above, filtering just sets up addresses for delivery – no
1917 deliveries are actually done while a filter file is active. If any of the
1918 generated addresses subsequently suffers a delivery failure, an error message
1919 is generated in the normal way. However, if a filter command that sets up a
1920 delivery is preceded by the word <quote>noerror</quote>, errors for that delivery,
1921 and any deliveries consequent on it (that is, from alias, forwarding, or
1922 filter files it invokes) are ignored.
1925 <section id="SECTex">
1926 <title>Examples of Exim filter commands</title>
1930 <literallayout class="monospaced">
1932 deliver baggins@rivendell.middle-earth.example
1935 Vacation handling using traditional means, assuming that the <filename>.vacation.msg</filename>
1936 and other files have been set up in your home directory:
1938 <literallayout class="monospaced">
1940 unseen pipe "/usr/ucb/vacation \"$local_part\""
1943 Vacation handling inside Exim, having first created a file called
1944 <filename>.vacation.msg</filename> in your home directory:
1946 <literallayout class="monospaced">
1948 if personal then vacation endif
1951 File some messages by subject:
1953 <literallayout class="monospaced">
1955 if $header_subject: contains "empire" or
1956 $header_subject: contains "foundation"
1962 Save all non-urgent messages by weekday:
1964 <literallayout class="monospaced">
1966 if $header_subject: does not contain "urgent" and
1967 $tod_full matches "^(...),"
1973 Throw away all mail from one site, except from postmaster:
1975 <literallayout class="monospaced">
1977 if $reply_address contains "@spam.site.example" and
1978 $reply_address does not contain "postmaster@"
1984 Handle multiple personal mailboxes:
1986 <literallayout class="monospaced">
1988 if $local_part_suffix is "-foo"
1991 elif $local_part_suffix is "-bar"
git://git.exim.org / exim-website.git / blob