# pm-janslookup.rc -- run nslookup on variable INPUT
# $Id: pm-janslookup.rc,v 2.11 2004/08/20 16:54:29 jaalto Exp $
#
#   File id
#
#       .Copyright (C)   1997-2004 Jari Aalto
#       .Keywords:       procmail, nslookup, subroutine
#
#       This code is free software in terms of GNU Gen. pub. Lic. v2 or later
#       Refer to http://www.gnu.org/copyleft/gpl.html
#
#   Description
#
#       This subroutine runs `nslookup' on given INPUT address. This may be
#       an effective way to test if the address is known to Internet. You
#       could use this information to determine if some automated reply to
#       a address can be sent. The know truth is that you can't validate
#       whole email address
#
#               to_someone@foo.com
#
#       but you can validate "foo.com"; that's the closest you get.
#
#       [Warning: If you don't use cache feature...]
#
#       Do not however use this module to regularly check _all_ incoming
#       from addresses with this subroutine for possible bogus UBE
#       addresses, because calling nslookup
#
#       o   may be slow, building to connection and querying the results
#           may take several seconds. (some times, usually it's quote fast)
#       o   consumes quite a lot resources.
#
#       You can however check _some_ messages that are likely UBE to verify
#       your doubts.
#
#   Required settings
#
#       PMSRC must point to source directory of procmail code. This subroutine
#       will include
#
#       o   pm-javar.rc
#       o   pm-jaaddr.rc
#
#   Call arguments (variables to set before calling)
#
#       o   INPUT, the address (only strict domain part) which is checked.
#           Eg. "this.domain.com". See examples for more.
#           If this string contains "@" character, then additional
#           subroutine pm-jaaddr.rc is called to extraxt the domain name
#           INPUT = "John Doe <foo@site.com>" --> INPUT = "site.com"
#       o   JA_NSLOOKUP_CACHE, filename. If exists, cache is used and updated.
#       o   JA_NSLOOKUP_FORCE, if "yes", then cache is not used but a forced
#           nslookup is performed.
#       o   JA_NSLOOKUP_OPT, is currently empty, but you could see if you
#           you want to use "-querytype=MX". However this option may give
#           you response: "No mail exchanger (MX) records available",
#           which is flagged as nslookup failure.
#       o   JA_NSLOOKUP_SERVER, optional, the server to user for nslookup
#
#       If the cache file can be read:
#
#       o   Each entry has format "address.com ns-error". The error
#           indication is added to the line if the nslookup failed when
#           address was checked. Otherwise line contains "address.com".
#       o   The cache is always checked first. If there is no entry matching
#           the current address, only then is nslookup called and new entry
#           added to cache.
#
#   Return values
#
#       o   Variable ERROR will be set to "yes" if nslookup failed or
#           to "no" if nslookup succeeded. It can also contain "maybe" if
#           nslookup returned "No address (A) records available"
#       o   ERROR_MATCH contains one line lookup failure reason.
#
#       Following conditions trigger "maybe" and no "ns-error" is written
#       into the cache.
#
#       o   "No address (A) records available for xxx"
#
#   Usage example
#
#       If you are going to check some header field, like From:, please
#       explode the content with pm-jaaddr.rc first. Suppose you have
#       string:
#
#           "From: foo@ingrid.sps.mot.com (Yoshiaki foo)"
#
#       You have to derive the address from string and pass the site name:
#       Read From: field and address from it.
#
#           PMSRC       = $HOME/pm
#           RC_NSLOOKUP = $PMSRC/pm-janslookup.rc   # name the subroutine
#
#           :0
#           * MAYBE_UBE ?? yes
#           * ^From:\/.*
#           {
#               INPUT       = $MATCH
#               INCLUDERC   = $RC_NSLOOKUP  #   to nslookup
#
#               :0
#               * ERROR ?? yes
#               {
#                   #   Hm, nslookup failed, can't send anything back to this
#                   #   address
#               }
#           }
#
#       Second example, check if the address is reachable before sending reply
#
#           INPUT       = `$FORMAIL -rt -x To:`
#           INCLUDERC   = $RC_NSLOOKUP
#
#           :0
#           *    ERROR ?? no
#           {
#               # okay, at least site address seems to be reachable
#           }
#
#
#   Change Log (none)

# .................................................... &initialising ...


dummy = "
========================================================================
pm-janslookup.rc: init: INPUT = $INPUT"

:0
* !  WSPC ?? [ ]
{
    INCLUDERC = $PMSRC/pm-javar.rc
}

#   If user gave string that had Email, explode it automatically
#   INPUT = "John doe <foo@site.com>" --> "site.com"

:0
* INPUT ?? @
{
    INCLUDERC   = $PMSRC/pm-jaaddr.rc   # - explode INPUT string
    INPUT       = $SITE                 # - the address is only fed
}

# ...................................................... &input-vars ...
#   o   INPUT contains the address to check
#   o   JA_NSLOOKUP_CACHE is optional

#       Set to empty "" if you don't wan't to use cache in some particular
#       cases

JA_NSLOOKUP_CACHE   = ${JA_NSLOOKUP_CACHE:-"$HOME/.nslookup.cache"}

#   other options

JA_NSLOOKUP_FORCE   = ${JA_NSLOOKUP_FORCE:-"no"}
JA_NSLOOKUP_OPT     = ${JA_NSLOOKUP_OPT:-"-querytype=MX"}
JA_NSLOOKUP_SERVER  = ${JA_NSLOOKUP_SERVER:-""}

NSLOOKUP            = ${NSLOOKUP:-"nslookup"} # Add `-silent' in Linux
TOUCH               = ${TOUCH:-"touch"}       # binary to create cache file

# ..................................................... &output-vars ...
#   output variables

ERROR = "yes"
ERROR_MATCH

# ........................................................... &do-it ...
#   ErrorWord is not really ment for user configurable, but
#   as you can see, it is coded that way in case you would like to
#   use some other error word.

ErrorWord = ${ErrorWord:-"nlookup-error"}
dummy     = "pm-janslookup.rc: do the work"

#   The string must have one dot somewhere, otherwise it is not a domain

:0
*$ INPUT ?? $NSPC+\.$NSPC+
{
    topDomain   = "xxx-dummy-regexp"
    cache       = "no"
    entry       = "no"

    :0
    * INPUT ?? ()\/[^.]+\.[^.]+^^
    {
        topDomain = $MATCH                      # xxx.foo.com --> foo.com
    }

    #   If variable is not empty, suppose that user wants to use cache

    :0
    * ! JA_NSLOOKUP_CACHE ?? ^^^^
    *   JA_NSLOOKUP_FORCE ?? no
    {
        cache = "yes"

        :0 hwic                         # Create file if it doesn't exist.
        * ! ? $IS_EXIST $JA_NSLOOKUP_CACHE
        | $TOUCH $JA_NSLOOKUP_CACHE
    }

    dummy = "pm-janslookup.rc: Test if cache is in use"

    :0
    *    cache ?? yes
    *$ ? $IS_READABLE $JA_NSLOOKUP_CACHE
    {
        #   Convert regexp to "safe" format, escapeping regexp
        #   metacharacters. See pm-tips.txt

        regexp = "$\INPUT"

        :0                              # kill leading "()"
        * regexp ?? ^^\(\)\/.*
        {
            regexp = "$MATCH"
        }

        line = `$GREP "$regexp" $JA_NSLOOKUP_CACHE`

        :0                              # was this address in cache?
        *$ line ?? $INPUT
        {
            entry = "yes"
        }

        dummy = "pm-janslookup.rc: Test if cache line does _not_ include $ErrorWord"

        :0
        *    entry ?? yes
        *$ ! line  ?? $ErrorWord
        {
            ERROR = "no"
        }
    }

    #   Cache is not used OR there was no line in the cache

    dummy = "pm-janslookup.rc: Test if we call nslookup"

    :0
    * entry ?? no
    {
        stat = `$NSLOOKUP $JA_NSLOOKUP_OPT "$INPUT" $JA_NSLOOKUP_SERVER`

        #       If nslookup succeeds, at least HP-UX 9/10, ULTRIX and SUN
        #       return some of these choices upon success:
        #
        #           Non-authoritative answer:
        #           Name:    uta.fi
        #
        #       Or
        #
        #           Non-authoritative answer:
        #           geocities.com   preference = 0, mail exchanger = ...
        #
        #       Or
        #
        #           Name:    foo.com
        #           Address:  209.54.94.60
        #
        #       Or
        #
        #           Authoritative answers can be found from:
        #           foo.com
        #           origin = xx.foo.com
        #
        #       Or
        #           Non-authoritative answer:
        #           foo.com mail exchanger = 0 mx1.foo.com.
        #
        #       Or ^($)$topDomain
        #
        #       This needs some explaining. When procmail captures
        #       the `nslookup' command it does it wrong for some unknown reason
        #       for some mail hosts. The captured output is:
        #
        #           procmail: Assigning "stat=Name Server:  zeus.tele.nokia.fi
        #           Address:  131.228.134.50
        #
        #           umd.umich.edu
        #           origin = tiamat.umd.umich.edu
        #           mail addr = hostmaster.tiamat.umd.umich.edu
        #           serial = 970727
        #           refresh = 28800 (8 hours)
        #           retry   = 14400 (4 hours)
        #           expire  = 2419200 (28 days)
        #           minimum ttl = 14400 (4 hours)"
        #
        #       notice? The line "Authoritative answers can be found from:"
        #       is not there while it is if I run the same command is
        #       run from command line.

        # |Authoritative answers can be found from:($).*

        :0
        * stat ?? ()\/No +address.*available.*
        {
            #   *** No address information is available for ...
            ERROR_MATCH = $MATCH
            ERROR       = "yes"
        }

        :0 E
        *$ stat ?? ()\/answer:($)(Name:)? *$INPUT\
                   |(Name: *$INPUT)|from:($) +$topDomain\
                   |($) *$topDomain
        {
            ERROR = "no"
        }

        # ............................................. Linux/Cygwin ...
        # Win32/Cygwin nslookup returns this answer

        # prompt> nslookup -querytype=MX cs2.tpu.fi
        #
        # Non-authoritative answer:
        # Server:  nsX.koti.tpo.fi
        # Address:  212.63.10.250
        #
        # cs2.tpu.fi      MX preference = 0, mail exchanger = xxx.tpu.fi
        # cs2.tpu.fi      MX preference = 10, mail exchanger = yyy.tpu.fi
        # cs2.tpu.fi      MX preference = 20, mail exchanger = zzz.funet.fi
        #
        # fi      nameserver = HYDRA.HELSINKI.fi
        # fi      nameserver = NS-SE.ELISA.NET
        # fi      nameserver = PRIFI.FICORA.fi
        # fi      nameserver = NS1-FIN.GLOBAL.SONERA.fi
        # fi      nameserver = T.NS.VERIO.NET
        # fi      nameserver = NS.UU.NET
        # xxx.tpu.fi      internet address = 193.167.70.45
        # yyy.tpu.fi      internet address = 193.167.70.20
        # mail.funet.fi   internet address = 193.166.0.98
        # mail.funet.fi   internet address = 193.166.0.97
        # HYDRA.HELSINKI.fi       internet address = 128.214.4.29
        # PRIFI.FICORA.fi internet address = 193.229.4.44

        dummy="Win32/Cygwin nslookup check"

        :0
        *$ stat ?? internet address = .*$topDomain
        {
            ERROR = "no"
        }

        # .................................................. message ...

        msg = ""

        :0
        * ERROR ?? yes
        {
            msg = $ErrorWord
        }

        :0
        * ERROR ?? maybe
        {
            msg = "maybe"
        }

        # .............................................. write cache ...

        dummy = "pm-janslookup.rc: Test if '$INPUT' should be added to cache"

        :0 hwic:
        * cache ?? yes
        | echo "$INPUT $msg" >> $JA_NSLOOKUP_CACHE

    }
}

:0 E
{
    dummy = "pm-janslookup.rc: $JA_MSG_ERROR invalid INPUT"
}

dummy = "pm-janslookup.rc: end: input: $INPUT error: $ERROR"

# end of file pm-janslokup.rc