#
# Look for duplicate loci in script and update the
# information for the first version based on all subsequent versions
# deleting the latter
#
AWK=/bin/gawk
SCRIPT="updateped.in"
PEDFILE="updateped.ped"

HELP=0
if [ $# = 0 ] 
then
  HELP=1
else
  for i in $*
  do
    if [ ! -f $i ]
    then
      HELP=1
      echo "ERROR: $i does not exist"
      echo
    fi
  done 
fi

if [ $HELP = 1 ]
then
	echo "Usage: updateped <script>"
        echo "Update sib-pair type pedigree file and script"
        echo "where multiple fields for the same locus."
        echo "Expects duplicate loci to be named <loc>_v2 <loc>_v3 etc."
        exit
fi

$AWK -v script=$SCRIPT -v pedfile=$PEDFILE '
#
# Parse the script
#
BEGIN { ndup=0
        nloc=0
        pos=5
        stoponerr=0
}
substr($1,1,1) ~ /[!#]/ {
        print $0 > script
}
$1=="set" && substr($2,1,3)=="loc" {
        nloc++
        name[nloc]=$3
        dup=index($3,"_v")
        loctyp[nloc]=substr($4,1,1)
        if (substr($4,1,3)=="mar" || substr($4,1,3)=="xma") {
          locpos[nloc]=pos+1 
          pos+=2
        }else{
          locpos[nloc]=pos+1
          pos++ 
        }
        if (dup) {
          ndup++
          duplist[ndup]=nloc
          mult[nloc]=substr($3,1,dup-1)
          print "Duplication: " mult[nloc] " <- " $3
        }else{
          mult[nloc]="UNIQUE"
          print $0 > script
        }
}
substr($1,1,3)=="rea" && (substr($2,1,3)=="ped" || substr($2,1,3)=="lin") {
        origped=$3
}
END   { print "read pedigree " pedfile > script
        print "run" > script
        print " " > script
         
        for(i=1;i<=nloc;i++) {
          if (mult[i]=="UNIQUE") {
            mult[i]=0
          }else{
            for(j=1;j<=nloc;j++) if (name[j]==mult[i]) {
              mult[i]=j
              break
            }
          }
        }

        if (ndup>0) {
          while(getline < origped) {
           if (substr($1,1,1) ~ /[!#]/) {
             print $0 > pedfile
           }else{
             printf "%s %s %s %s %s",$1,$2,$3,$4,$5 > pedfile
             for(i=1;i<=ndup;i++) {
               master=locpos[mult[duplist[i]]]
               dup=locpos[duplist[i]]
               if ($(master)=="x") {
                 if ($(dup)!="x") {
                   $(master)=$(dup)
                   if (loctyp[duplist[i]] ~ "[mx]") $(master+1)=$(dup+1)
                 }
               }else if ($(dup)!="x" && $(master)!=$(dup)) {
                 print "In " origped ", person " $1 "-" $2 \
                   " has inconsistent values for " name[mult[duplist[i]]]   
                 print "1st value =" $(master)
                 print "2nd value =" $(dup)
                 if (stoponerr) {
                   print "Bailing out!"
                   exit
                 }
               }
             }     
             for(i=1;i<=nloc;i++) {
               if (!mult[i]) {
                 pos=locpos[i]
                 if (loctyp[i] ~ "[mx]") {
                   printf " %s %s",$(pos),$(pos+1) > pedfile
                 }else{
                   printf " %s",$(pos) > pedfile
                 }
               }
             }
             printf "\n" > pedfile
           }
          }
        }else{
          while(getline < origped) print $0 > pedfile
        }
}' $* 
bester $PEDFILE > tmp$$
mv tmp$$ $PEDFILE