chiark - git - mdw - exim-config/blob - spam.m4

   1 ### -*-m4-*-
   2 ###
   3 ### Spam filtering for distorted.org.uk Exim configuration
   4 ###
   5 ### (c) 2012 Mark Wooding
   6 ###
   7
   8 ###----- Licensing notice ---------------------------------------------------
   9 ###
  10 ### This program is free software; you can redistribute it and/or modify
  11 ### it under the terms of the GNU General Public License as published by
  12 ### the Free Software Foundation; either version 2 of the License, or
  13 ### (at your option) any later version.
  14 ###
  15 ### This program is distributed in the hope that it will be useful,
  16 ### but WITHOUT ANY WARRANTY; without even the implied warranty of
  17 ### MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  18 ### GNU General Public License for more details.
  19 ###
  20 ### You should have received a copy of the GNU General Public License
  21 ### along with this program; if not, write to the Free Software Foundation,
  22 ### Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA.
  23
  24 DIVERT(null)
  25 ###--------------------------------------------------------------------------
  26 ### Spam filtering.
  27
  28 ## The Exim documentation tells lies.
  29 ##
  30 ## : *${run{*<_command_>* *<_args_>*}{*<_string1_>*}{*<_string2_>*}}*
  31 ## :     The command and its arguments are first expanded separately, [...]
  32 ##
  33 ## They aren't.  The whole command-and-args are expanded together, and then
  34 ## split at unquoted spaces.  This unpleasant hack sorts out the mess.
  35 m4_define(<:SHQUOTE:>, <:"${rxquote:$1}":>)
  36
  37 SECTION(global, policy)m4_dnl
  38 spamd_address = CONF_spamd_address CONF_spamd_port
  39
  40 SECTION(routers, allspam)m4_dnl
  41 ## If we're verifying an address and the recipient has a `~/.mail/spam-limit'
  42 ## file, then look up the recipient and sender addresses to find a plausible
  43 ## limit and insert it into the `address_data' where the RCPT ACL can find
  44 ## it.  This router always declines, so it doesn't affect the overall outcome
  45 ## of the verification.
  46 fetch_spam_limit:
  47         driver = redirect
  48         data = :unknown:
  49         verify_only = true
  50         local_part_suffix = CONF_user_suffix_list
  51         local_part_suffix_optional = true
  52         check_local_user
  53         address_data = \
  54                 ${if def:address_data {$address_data}{}} \
  55                 ${if and {{!eq{$acl_c_mode}{submission}} \
  56                           {exists {CONF_userconf_dir/spam-limit}}} \
  57                      {${lookup {$local_part_prefix\
  58                                         $local_part\
  59                                         $local_part_suffix\
  60                                         @$domain/\
  61                                         $sender_address} \
  62                                nwildlsearch {CONF_userconf_dir/spam-limit} \
  63                                {${if match{$value}{\N^-?[0-9]+$\N} \
  64                                      {spam_limit=$value} \
  65                                      {}}} \
  66                                {}}} \
  67                      {}} \
  68                 ${if and {{!eq{$acl_c_mode}{submission}} \
  69                           {exists {CONF_userconf_dir/spam-limit.userv}}} \
  70                      {${run {timeout 5s -- \
  71                                 userv SHQUOTE($local_part) exim-spam-limit \
  72                                         SHQUOTE($sender_address) \
  73                                         SHQUOTE($local_part_prefix) \
  74                                         SHQUOTE($local_part) \
  75                                         SHQUOTE($local_part_suffix) \
  76                                         SHQUOTE(@$domain)} \
  77                             {${if match{$value}{\N^-?[0-9]+$\N} \
  78                                   {spam_limit=$value} \
  79                                   {}}} \
  80                             {}}} \
  81                      {}}
  82
  83 SECTION(acl, rcpt-hooks)m4_dnl
  84         ## Do per-recipient spam-filter processing.
  85         require  acl = rcpt_spam
  86
  87 SECTION(acl, misc)m4_dnl
  88 skip_spam_check:
  89
  90         ## If the client is trusted, or this is a new submission, don't
  91         ## bother with any of this.  We will have verified the sender
  92         ## fairly aggressively before granting this level of trust.
  93         accept   hosts = +trusted
  94         accept   condition = ${if eq{$acl_c_mode}{submission}}
  95
  96         ## Otherwise we should check.
  97         deny
  98
  99 rcpt_spam:
 100
 101         ## See if we should do this check.
 102         accept   acl = skip_spam_check
 103
 104         ## Always accept mail to `postmaster'.  Currently this is not
 105         ## negotiable; maybe a tweak can be added to `domains.conf' if
 106         ## necessary.
 107         accept   local_parts = postmaster
 108
 109         ## Collect the user's spam threshold from the `address_data'
 110         ## variable, where it was left by the `fetch_spam_limit' router
 111         ## during recipient verification.  (This just saves duplicating this
 112         ## enormous expression.)
 113         warn     set acl_m_this_spam_limit = \
 114                         ${sg {${extract {spam_limit} \
 115                                         {${if def:address_data \
 116                                               {$address_data}{}}} \
 117                                         {$value}{nil}}} \
 118                              {^(|.*\\D.*)\$}{CONF_spam_max}}
 119
 120         ## If there's a spam limit already established, and it's different
 121         ## from this user's limit, then the sender will have to try this user
 122         ## again later.
 123         defer   !hosts = +trusted
 124                  message = "You'd better try this one later"
 125                  condition = ${if def:acl_m_spam_limit {true}{false}}
 126                  condition = ${if ={$acl_m_spam_limit} \
 127                                    {$acl_m_this_spam_limit} \
 128                                   {false}{true}}
 129
 130         ## There's no limit set yet, or the user's limit is the same as the
 131         ## existing one, or the client's local and we're not checking for
 132         ## spam anyway.  Whichever way, it's safe to set it now.
 133         warn     set acl_m_spam_limit = $acl_m_this_spam_limit
 134
 135         ## All done.
 136         accept
 137
 138 SECTION(acl, data-spam)m4_dnl
 139         ## Do spam checking.
 140         require  acl = data_spam
 141
 142 SECTION(acl, misc)m4_dnl
 143 data_spam:
 144
 145         ## See if we should do this check.
 146         accept   acl = skip_spam_check
 147
 148         ## Check header validity.
 149         require  verify = header_syntax
 150
 151         ## Check the message for spam, comparing to the configured limit.
 152         deny     spam = exim:true
 153                  message = Tinned meat product detected ($spam_score)
 154                  condition = ${if >{$spam_score_int}{$acl_m_spam_limit} \
 155                                   {true}{false}}
 156
 157         ## Insert headers from the spam check now that we've decided to
 158         ## accept the message.
 159         warn
 160
 161                  ## Convert the limit (currently 10x fixed point) into a
 162                  ## decimal for presentation.
 163                  set acl_m_spam_limit_presentation = \
 164                         ${sg{$acl_m_spam_limit}{\N(\d)$\N}{.\$1}}
 165
 166                  ## Convert the report into something less obnoxious.  Plain
 167                  ## old SpamAssassin has an `X-Spam-Status' header which
 168                  ## lists the matched rules and provides some other basic
 169                  ## information.  Try to extract something similar from the
 170                  ## report.
 171                  ##
 172                  ## This is rather fiddly.
 173
 174                  ## Firstly, escape angle brackets, because we'll be using
 175                  ## them for our own purposes.
 176                  set acl_m_spam_tests = ${sg{$spam_report}{([!<>])}{!\$1}}
 177
 178                  ## Trim off the blurb paragraph and the preview.  The rest
 179                  ## should be fairly well behaved.  Wrap double angle-
 180                  ## brackets around the remainder; these can't appear in the
 181                  ## body because we escaped them all earlier.
 182                  set acl_m_spam_tests = \
 183                         ${sg{$acl_m_spam_tests} \
 184                             {\N^(?s).*\n Content analysis details:(.*)$\N} \
 185                             {<<\$1>>}}
 186
 187                  ## Extract the information about the matching rules and
 188                  ## their scores.  Leave `<<...>>' around everything else.
 189                  set acl_m_spam_tests = \
 190                         ${sg{$acl_m_spam_tests} \
 191                             {\N(?s)\n\s*(-?[\d.]+)\s+([-\w]+)\s\N} \
 192                             {>>\$2:\$1,<<}}
 193
 194                  ## Strip everything still in `<<...>>' pairs, including any
 195                  ## escaped characters inside.
 196                  set acl_m_spam_tests = \
 197                         ${sg{$acl_m_spam_tests}{\N(?s)<<([^!>]+|!.)*>>\N}{}}
 198
 199                  ## Trim off a trailing comma.
 200                  set acl_m_spam_tests = ${sg{$acl_m_spam_tests}{,\s*\$}{}}
 201
 202                  ## Undo the escaping.
 203                  set acl_m_spam_tests = ${sg{$acl_m_spam_tests}{!(.)}{\$1}}
 204
 205                  ## Insert the headers.
 206                  add_header = X-SpamAssassin-Score: \
 207                         $spam_score/$acl_m_spam_limit_presentation \
 208                         ($spam_bar)
 209                  add_header = X-SpamAssassin-Status: \
 210                         score=$spam_score, \
 211                         limit=$acl_m_spam_limit_presentation, \n\t\
 212                         tests=$acl_m_spam_tests
 213
 214         ## We're good.
 215         accept
 216
 217 DIVERT(null)
 218 ###----- That's all, folks --------------------------------------------------