#!/bin/sh
## Get info on one or more specific rules
## Usage: sa-rule-info [-s|-v[v]] [-l|-L] RULE [RULE...]
##   -s    Short output; show just the score(s); no description or regex
##   -l    Enable local configurations to trump upstream rule info (default)
##   -L    Ignore local configurations (check defaults + updates)
##   -v    Turns on verbose, which reveals alternate language entries
##   -vv   Also shows ALL rule hits in descending priority, with filenames
##   RULE  Rule to check.  Tip:  you can use a basic regex (like egrep)
## sa-rule-info v0.4, Copyright (C) 2008 by Adam Katz <scriptsATkhopiscom>
## Licensed under the GNU AGPL v3+ (relicensing available upon request).

unset short
type egrep >/dev/null 2>&1 || egrep() { grep "$@"; } # failover for egrep
no_comm() { cat; } # default to showing comments
hide_fname() { sed 's/^[^:]*://g'; } # default: hide filenames. override w/ -vv
DO_LOCAL=true
while [ -n "$1" ]; do
  case $1 in
    -v|-vv )
      [ "x$1" = "x-vv${1#-vv}" ] && very_verbose=true # -vv
      lang=true # parse languages
      unset short
      ;;
    -s )
      short=true
      unset lang very_verbose
      ;;
    -l )
      DO_LOCAL=true
      ;;
    -L )
      unset DO_LOCAL
      ;;
    -- ) 
      shift
      break
      ;;
    -?*) # --help
      sed -e '/^## /!d' -e 's///' "`which $0 2>/dev/null || echo $0`"
      [ "x$1" = x-h ]; exit $? # exit true if passed -h, else exit false
      ;;
    -|/dev/stdin ) shift; set `cat` ${1:+"$@"}; break ;;
    * ) break ;;
  esac
  shift
done
[ -n "$very_verbose" ] && hide_fname() { cat; } # -vv
# strip comments and redundant score prefix, so "score FOO 3 # hi" -> "FOO 3"
[ -n "$short" ] && \
  no_comm() { sed -e 's/[ 	]*#.*//g' -e 's/^.*score[ 	][ 	]*//'; }


# find most recently modified of the given dirs or else failover to /dev/null
latest_dir() { ls -td "$@" 2>/dev/null |head -n1 |grep . || echo '/dev/null'; }

s="[	 ()]"	# match whitespace, aka "[\t ]" and almost like "\s"

update_dir=`latest_dir /var/lib/spamassassin*/*/ /var/db/spamassassin*/*/`
share_dir=`latest_dir /usr/local/share/spamassassin*/ /usr/share/spamassassin*/`
conf_dir=`latest_dir /usr/local/etc/mail/spamassassin*/ /etc/spamassassin*/ \
                     /etc/mail/spamassassin*/`
local_cf="${conf_dir%/}/local.cf"
user_dir=`latest_dir "$HOME/.spamassassin"`
# note, this ignores the possibility of spamd --virtual-config-dir

TMP=`tempfile 2>/dev/null || mktemp /tmp/tmp$$XX 2>/dev/null || echo /tmp/tmp$$`
TMP2=`tempfile 2>/dev/null || mktemp /tmp/tmp$$XX 2>/dev/null || echo /tmp/tp$$`
trap "rm -f $TMP $TMP2" 0 1 2 5 15 # clean up temp files when we exit

for RULE in "$@"; do

  if [ "$RULE" != "$1" ]; then echo "--"; fi # delimiter between RULEs

  [ "$RULE" = "^${RULE#?}" ] && b="$s" RULE="${RULE#?}" || b=''
  [ "$RULE" = "${RULE%?}$" ] && a="$s" RULE="${RULE%?}" || a=''
  # dot should match non-whitespace, force capitals
  RULE=`echo $RULE |sed "s/\./[^ .	]/g" |tr '[a-z]' '[A-Z]'`
  if [ "${#RULE}" -lt 4 ]; then
    echo "Please use a query longer than '$RULE'" >&2
    continue
  fi

  # find files in all those dirs that are smaller than 2mb, search for RULE
  # (local.cf gets priority over other things in $conf_dir)
  find ${DO_LOCAL:+"$user_dir" "$local_cf" "$conf_dir"} \
    "$update_dir" "$share_dir" -size -4096 -print0 2>/dev/null \
    |xargs -0 egrep "^[^#]*$b$RULE$a" 2>/dev/null \
    |grep -v "^Binary file" |hide_fname >$TMP

  for setting in `sed -e "s/^$s*//" -e "s/$s*[^ .	]*$RULE$a.*//" -e "s/$s/./g" $TMP`; do
    egrep "^$setting" $TMP2 >/dev/null && continue # skip token if found already
    [ -z "$lang" ] && [ "$setting" != "${setting#lang}" ] && continue # no langs
    [ -n "$short" ] && [ "$setting" = "${setting#score}" ] && continue # short?
    echo "$setting" >>$TMP2 # note that we have found this token
    egrep "^$s*$setting" $TMP |head -n1 |no_comm
  done # end settings parsing

  [ -s $TMP ] && y=y || echo "No SpamAssassin rules found for query '$RULE'" >&2
  printf '' >$TMP

done # end loop of given rules

[ -n "$y" ]; exit $? # exits 0 if we had one hit, 1 if no queries matched.