#! /bin/csh -f
# Last edited on 2025-05-01 18:46:28 by stolfi

# Reads a list of "dex" features with counts (as from uniq -c).
# Sorts them by number of connected X/e components.

/bin/cat \
  | gawk \
      ' /./ { \
          c=$2; gsub(/[Xe][Xe]*/,"@",c);gsub(/[^@]/,"",c);c=length(c); \
          r=(match($2,/[D]/)>0); \
          s=(match($2,/[X]/)>0); \
          t=(match($2,/[e]/)>0); \
          printf "%02d %d %d %d %s\n",c,r,s,t,$0; \
        } \
      ' \
  | sort -b -k1 -k2 -k3 -k4 -k5nr \
  | gawk \
      ' /./ { \
          if(($1!=c)||($2!=r)||($3!=s)||($4!=t)) \
            { if(n>0) { printf "#%7d  TOTAL c=%s r=%s s=%s t=%s\n", n,c,r,s,t; } \
              c=$1; r=$2; s=$3; t=$4; n=0; \
            } \
          print substr($0,9); \
          n+=$5; \
        } \
        END { \
          if(n>0) { printf "#%7d  TOTAL c=%s r=%s s=%s t=%s\n", n,c,r,s,t; }  \
        } \
      '