csa

#! /bin/bash

# (C) 2008 Chr. Junghans
# junghans@mpip-mainz.mpg.de

#version 0.1   ,04.07.08 -- initial version
#version 0.1.1 ,10.07.08 -- fixed bug with empty lines
#version 0.1.2 ,01.08.08 -- fixes bug with empty lines
#version 0.1.3 ,25.09.08 -- fixes bug with short name of del

#########Defaults
files=""
commandline=""
print="no"
emptyline='/^[[:space:]]*$/d'

showhelp() {
 cat << eoh
Christoph's simple analysis
The best of perl,sed and awk to make simple analysis to a file or stdin

${0##*/} [OPTIONS] COMMAND FILES

COMMANDS:

print(p) NUMBERS print columns (use -- to perfom  filename begin with 0-9)
del(d)   RANGE   delete a range of lines (e.g. 42 or 1,12 or /BEGIN/ or /BEGIN/,/END/
delc(c)          Delete comments (same as "del /^#/")
dele(e)          Delete empty lines (same as "csa m \$0")
bgrep(b)  P1 P2  Grep what is between P1 and P2 (in every line)

SIMPLE ANALYSIS (still commands)
sum(s) NUMBERS  sum of the columns
avg(a) NUMBERS  average (with standard error)
ravg(r) NUMBER running average (with standard error)
mod(m) FORMULAR modify (e.g \$1+\$2,  0.5*\$2)
help             show this help

OPTIONS:
-f, --file      Give a file (maybe multiple)
-r, --noremove  Don't remove empty lines
-p, --print     Also print the commandline
-v, --version   Prints version
-h, --help      Show this help message
eoh
}

shopt -s extglob

while [ "${1#-}" != "$1" ]; do
  if [ "${1#--}" = "$1" ] && [ -n "${1:2}" ]; then
    if [ "${1#-[r]}" != "${1}" ]; then
       set -- "${1:0:2}" "${1:2}" "${@:2}"
    else
       set -- "${1:0:2}" "-${1:2}" "${@:2}"
    fi
 fi
 case $1 in
  -f | --file )
   if [ -r "$2" ]; then
      files="$files $2"
   else
      echo Error: file \"$2\" not readable >&2
      exit 1
   fi
   shift 2;;
  -p | --print )
   print="yes"
   shift;;
  -r | --noremove )
   emptyline=''
   shift;;
  -v | --version)
   echo ${0##*/}, $(sed -ne 's/^#\(version.*\) -- .*$/\1/p' $0 | sed -n '$p') by C. Junghans
   exit 0;;
  -h | --help)
   showhelp
   exit 0;;
  *)
   echo "Unknown option '$1'" >&2
   exit 1;;
 esac
done

if [ -z "$1" ]; then
   echo No command given ! >&2
   echo Help with ${0##*/} help >&2
   exit 1
fi

echo "#This is $(${0##*/} -v)"

case $1 in
   p|print)
      program="awk"
      progstring="/^#/{print \$0;next}{print "
      if [[ "$2" = +([0-9]) ]]; then
         progstring="$progstring \$$2"
         shift 2
      else
         echo Expected a number after \"print\" >&2
         exit 1
      fi
      while (! [[ "$1" = "--" ]] ) && [[ "$1" = +([0-9]) ]] ; do
         progstring="$progstring,\$$1"
         shift
      done
      progstring="$progstring}"
      [[ "$1" = "--" ]] && shift
      ;;
   d|del)
      program="sed"
      if [ -n "$2" ]; then
         progstring="${2}d"
         shift 2
      else
         echo Expected a pattern after \"del\" >&2
         exit 1
      fi
      ;;
   c|delc)
      program="sed"
      progstring="/^#/d"
      shift ;;
   e|dele)
      #this is done anyway ;-)
      program="tee"
      progstring=""
      shift ;;
   b|bgrep)
      program="perl -ne"
      if [ -n "$2" ] && [ -n "$3" ]; then
         progstring="print \"\$1\n\" if /$2(.*?)$3/"
         shift 3
      else
         echo Expected two pattern after \"bgrep\" >&2
         exit 1
      fi
      ;;
   s|sum)
      program="awk"
      if [[ "$2" = +([0-9]) ]]; then
         progstring="(\$$2!=\"\"){sum[$2]+=\$$2;c[$2]++}END{print c[$2],sum[$2]}"
         shift 2
      else
         echo Expected a number after \"ravg\" >&2
         exit 1
      fi
      echo "#Output format: # sum"
      ;;
   a|avg)
      program="awk"
      if [[ "$2" = +([0-9]) ]]; then
         progstring="(\$$2!=\"\"){sum[$2]+=\$$2;sum2[$2]+=\$$2*\$$2;c[$2]++}END{print c[$2],sum[$2]/c[$2],sum2[$2]/c[$2]}"
         shift 2
      else
         echo Expected a number after \"ravg\" >&2
         exit 1
      fi
      echo "#Output format: # avg se"
      ;;
   r|ravg)
      program="awk"
      if [[ "$2" = +([0-9]) ]]; then
         progstring="(\$$2!=\"\"){sum[$2]+=\$$2;sum2[$2]+=\$$2*\$$2;c[$2]++;print c,sum[$2]/c[$2],sum2[$2]/c[$2]}"
         shift 2
      else
         echo Expected a number after \"ravg\" >&2
         exit 1
      fi
      echo "#Output format: # ravg rse"
      ;;
   m|mod)
      program="awk"
      if [ -n "$2" ]; then
         progstring="/^#/{print \$0;next}{print $2}"
         shift 2
      else
         echo Expected a something after \"mod\" >&2
         exit 1
      fi
      ;;
   help)
      showhelp ;;
esac

#rest should be files !
while [ -n "$1" ]; do
   if [ -r "$1" ]; then
      files="$files $1"
      shift
   else
      echo Error: file \"$1\" not readable >&2
      exit 1
   fi
done

if [ "$print" = "yes" ]; then
  echo \#COMMAND: $program \"$progstring\"
fi

if [ -n "$files" ]; then
  cat $files | $program "$progstring" | sed "$emptyline"
else
  while read -r; do
    data="${data}${REPLY}\n"
  done
  #remove last \n 
  data=${data%\\n}
  echo -e $data | $program "$progstring" | sed "$emptyline"
fi