# need to add bang line, for bash !
# gopher server won't serve this file in its original state
#!/usr/pkg/bin/bash

# for extracting individual postings as single files
# from input stream of bboard archive searches
# captured with tee (or otherwise)
# files are written to boardname directories under the working directory

# debug=0;   # uncomment for debug information echos

# these didn't help [me]; couldn't get caseless matching to work with 'expr' matching
#shopt -s nocaseglob #required for caseless matching
#shopt -s nocasematch
   
tempfile="/tmp/${LOGNAME}_tmp"	#write to /tmp
[ -f $tempfile ] && rm $tempfile

theend='\(^<.*-0[0-9]>\)'	#defines end of post
thestart=TACKER.*		#defines beginning of post
thestart2=Tacker.*		#defines beginning of post
thesubject="SUBJECT: "
thesubject2="Subject: "

# boolean functions, to provide status of the current line, whether at beginning or end
atstart () {
   expr "$theline" : ".*$thestart" >/dev/null \
   || expr "$theline" : ".*$thestart2" >/dev/null
   }
atend () {
   expr "$theline" : "$theend" >/dev/null
   }

while ( atstart || read theline )
   do	# read a new line if we're not already at the start of a new post
#   echo + $theline
   until atstart	# until we find the start of a new post
      do read theline
      if [ $? -ne 0 ]; then echo "Done!"; exit 0; fi	# exit at EOF
      [ $debug ] && echo - $theline
      done
   theline=`expr "$theline" : ".*\($thestart\)"`	# trim leading dots from $thestart line
#   echo $theline

# now we have a start; look for end
# if we find another start before we find an end, discard this post, and restart
# this condition occurs with duplicated posts (which don't have an end delimiter)

# enter with the startline ....
#   if [ -f $tempfile ]; then rm $tempfile; fi
#   echo $theline >> $tempfile
   echo $theline >$tempfile
   read theline
   subject=${theline#$thesubject} || subject=${theline#$thesubject2}
   [ $debug ] && echo "S: $subject"
# now we don't enter the loop with atstart condition/line
# and we must test for discovery of atstart before finding atend
   until ( atend || atstart )
      do \
      echo $theline >> $tempfile
      read theline 
      if [ $? -ne 0 ]; then echo "Done!"; exit 0; fi	# exit at EOF
      [ $debug ] && echo . $theline
      done
# now we have either the end of the post, or the start of a new one
   if atstart
      then rm "$tempfile"
         echo "!restart $subject"
#         subject=""
      elif atend
         then	# only now do we have the basis of a destination file dir/name
# all this substring manipulation could probably be improved upon:
         destfile=`expr "$theline" : $theend`
         destfile=${destfile#\<}
         destfile=${destfile%\>}
         echo "$destfile" >> $tempfile; echo >> $tempfile
         destdir=`expr "$destfile" : '\(.*\.\)'`
         destdir=${destdir%\.}
         if [ ! -d "${destdir}" ]; then mkdir "${destdir}"; fi
         destfile=${destfile#${destdir}.}
         destfile="${destdir}/${destfile}_${subject}"
         echo "$destfile"
# there can be multiple posts to common filename as derived
         cat $tempfile >> "${destfile}"	
      fi
   done
exit 0