refactor spam filter module

1 files changed, 70 insertions, 39 deletions

diff --git a/modules/m_spamfilter.sh b/modules/m_spamfilter.sh
index 4014847..f687cb4 100644
--- a/modules/m_spamfilter.sh
+++ b/modules/m_spamfilter.sh
@@ -18,16 +18,20 @@
 #  along with this program.  If not, see <http://www.gnu.org/licenses/>.  #
 #                                                                         #
 ###########################################################################
-#--------------------------------------------------------------------------
-## This module filters known spam using +q and +z modes.
-## It will relay any un-recognized message as: (nick said):.
-## Requires: $BECOME_OP_ON_JOIN non-zero and
-##           channel in $OP_CHANNELS in bot_settings.sh
-#--------------------------------------------------------------------------
+#-------------------------------------------------------------------------#
+## This module filters known spam using +q and +z modes.                 ##
+## It will relay any un-recognized message as: (nick said):.             ##
+## Requires: the 'ii' program to be installed                            ##
+## Requires: (in bot_settings.sh)                                        ##
+##   * $BECOME_OP_ON_JOIN non-zero                                       ##
+##   * channel in $OP_CHANNELS                                           ##
+##   * channel in $config_module_spamfilter_channels                     ##
+#-------------------------------------------------------------------------#
 
 
 readonly DEBUG=0
 readonly RELAY_NICK='a-user'
+readonly FILTER_CHANNELS="${config_module_spamfilter_channels}"
 readonly II_DIR=/home/pbot/irc/${config_server}
 readonly II_LOG_REGEX='[0-9]{4}-[0-9]{2}-[0-9]{2} [0-9]{2}:[0-9]{2} '
 readonly SPAMFILTER_CHANNELS="${config_module_spamfilter_channels}"
@@ -83,7 +87,7 @@ module_spamfilter_on_JOIN()
   local bot_nick=${server_nick_current}
   parse_hostmask_nick "$1" 'whojoined'
 
-DBG_JOIN ${whojoined} ${channel}
+DBG_JOIN "${whojoined}" "${channel}"
 
     if   [[ "${whojoined}"             == "${bot_nick}"  ]] && \
          [[ " ${SPAMFILTER_CHANNELS} " =~ " ${channel} " ]] && \
@@ -118,45 +122,73 @@ DBG_RAW_LINE "${raw_line}"
   local hostmask="${BASH_REMATCH[1]}"
   local target="${BASH_REMATCH[2]}"
   local query="${BASH_REMATCH[3]}"
+  local sender
   parse_hostmask_nick "${hostmask}" 'sender'
-  local log_line="<${sender}> ${query}"
-  local chat_log="$(sleep 1 ; tail -n 12 ${II_DIR}/${channel}/out)"
+  local was_handled
 
-DBG_CONDITIONS ${sender} ${target} ${log_line} ${chat_log}
+DBG_CRITERIA "${sender}" "${target}" "${query}"
 
-  # ignore chat that is present in the ii users's chat log
-  [[ " ${RELAY_NICK} ${INJECT_NICK} "         =~ " ${sender} "                ]] && return 0
-  [[ " ${config_module_spamfilter_channels} " =~ " ${target} "                ]] || return 0
-  [[ "${chat_log}"                            =~ ${II_LOG_REGEX}"${log_line}" ]] && return 0
+  # ignore internal messages and chat from registered users
+  if ! is_filtered_channel "${target}"            ||
+       is_internal_user    "${sender}"            ||
+       is_public_chat      "${sender}" "${query}"
+  then was_handled=0
+  else
 
-DBG_UNREGISTERED ${sender}
+DBG_UNREGISTERED "${sender}"
 
-  # this chat is from an unregistered user - relay it to the channel unless it is known spam
-  if is_spam "${query}"
-  then
+       # ignore chat that is known spam or otherwise nonsense
+       if is_spam "${query}"
+       then was_handled=1
 
 DBG_SPAM ${sender}
 
-  else
-    send_msg "${channel}" "(${sender} said): ${query}"
-    #echo "(${sender} said): ${query}" > ${II_DIR}/${channel}/in
+       # relay chat from an unregistered user to the channel
+       else
+         send_msg "${target}" "(${sender} said): ${query}"
+         #echo "(${sender} said): ${query}" > ${II_DIR}/${target}/in
+
+         was_handled=0
+       fi
   fi
 
-  # supress further handling of this message
-  return 1
+  # supress or allow further handling of this message
+  return ${was_handled}
 }
 
 
 ## helpers ##
 
-is_spam()
+is_internal_user() # (sender)
+{
+  local sender=$1
+
+  [[ " ${RELAY_NICK} ${INJECT_NICK} " =~ " ${sender} " ]]
+}
+
+is_filtered_channel() # (target)
+{
+  local target=$1
+
+  [[ " ${FILTER_CHANNELS} " =~ " ${target} " ]]
+}
+
+is_public_chat() # (sender , query)
+{
+  local sender=$1
+  local query=$2
+  local log_line="<${sender}> ${query}"
+  local chat_log="$(sleep 2 ; tail -n 12 ${II_DIR}/${target}/out 2> /dev/null ;)"
+
+  [[ "${chat_log}" =~ ${II_LOG_REGEX}"${log_line}" ]]
+}
+
+is_spam() # (chat_msg)
 {
   local needle=$1
   local haystack=("${SPAM[@]}")
   local straw
 
-  [ -z "$(echo ${needle})" ] && return 0
-
   for straw in "${haystack[@]}"; do [[ ${needle} = *"${straw}"* ]] && return 0 ; done ;
 
   return 1
@@ -165,22 +197,21 @@ is_spam()
 
 ## DEBUG ##
 
-DBG_JOIN() { (( $DEBUG )) || return ; local whojoined=$1 ; local channel=$2 ; echo "[SPAMFILTER]: whojoined=${whojoined} channel=${channel}" ; }
-DBG_RELAY_USER() { (( $DEBUG )) && echo "[SPAMFILTER]: launched ii relay user: '${RELAY_NICK}'" ; }
-DBG_SET_MODE() { (( $DEBUG )) && echo "[SPAMFILTER]: set mode +qz" ; }
-DBG_RAW_LINE() { (( $DEBUG )) && echo -e "[SPAMFILTER]: \n\traw_line='$1'" ; }
-DBG_CONDITIONS()
+DBG_JOIN()         { (( ${DEBUG} )) || return ; local whojoined=$1 ; local channel=$2 ; echo "[SPAMFILTER]: whojoined=${whojoined} channel=${channel}" ; }
+DBG_RELAY_USER()   { (( ${DEBUG} )) && echo "[SPAMFILTER]: launched ii relay user: '${RELAY_NICK}'" ; }
+DBG_SET_MODE()     { (( ${DEBUG} )) && echo "[SPAMFILTER]: set mode +qz" ; }
+DBG_RAW_LINE()     { (( ${DEBUG} )) && echo -e "[SPAMFILTER]: incoming raw_line=\n\t'$1'" ; }
+DBG_CRITERIA()
 {
-  (( $DEBUG )) || return
+  (( ${DEBUG} )) || return
 
   local sender=$1
   local target=$2
-  local log_line=$3
-  local chat_log=$4
+  local query=$3
 
-  echo -n "[SPAMFILTER]: sender='${sender}'" ; [[ " ${RELAY_NICK} ${INJECT_NICK} "         =~ " ${sender} "                ]] && echo -n " => from RELAY_NICK - returning"      ; echo
-  echo -n "[SPAMFILTER]: target='${target}'" ; [[ " ${config_module_spamfilter_channels} " =~ " ${target} "                ]] || echo -n " => wrong channel - returning"        ; echo
-  echo -n "[SPAMFILTER]: query='${query}'"   ; [[ "${chat_log}"                            =~ ${II_LOG_REGEX}"${log_line}" ]] && echo -n " => from registered user - returning" ; echo
+  echo -n "[SPAMFILTER]: target='${target}'" ; ! is_filtered_channel "${target}"            && echo -n " => wrong channel - returning"        ; echo
+  echo -n "[SPAMFILTER]: sender='${sender}'" ;   is_internal_user    "${sender}"            && echo -n " => from internal user - returning"   ; echo
+  echo -n "[SPAMFILTER]: query='${query}'"   ;   is_public_chat      "${sender}" "${query}" && echo -n " => from registered user - returning" ; echo
 }
-DBG_UNREGISTERED() { (( $DEBUG )) || return ; local sender=$1 ; echo "[SPAMFILTER]: unregistered user sender=${sender}" ; }
-DBG_SPAM() { (( $DEBUG )) || return ; local sender=$1 ; echo "[SPAMFILTER]: !!!triggered!!! spambot=${sender}" ; }
+DBG_UNREGISTERED() { (( ${DEBUG} )) || return ; local sender=$1 ; echo "[SPAMFILTER]: unregistered user sender=${sender}" ; }
+DBG_SPAM()         { (( ${DEBUG} )) || return ; local sender=$1 ; echo "[SPAMFILTER]: !!!triggered!!! spambot=${sender}" ; }