rcut

#!/usr/bin/env bash

# use mawk by default
# will be changed to gawk if -g option is used
cmd='mawk'

# similar to grep's -F option
# treats the input field separator(IFS) literally
F=''

# initialize input and output field separators as single space
# which is same default as awk
d=' '
o=' '

# - means print all fields
# this default helps to format input with value of output field separator(OFS)
f='-'

# allow negative indexing if set
n=''

# complement fields if set
c=''

# suppress lines not matching IFS
s=''

# flag variable to check if output field separator was set by the user or not
ofs_flag='false'

# first : makes it silent error reporting mode
# d f o options require arguments if used
# n c g s F options do not accept arguments
while getopts ":d:f:o:ncgsF" opt; do
    case $opt in
        d)
            d="$OPTARG"
            ;;
        f)
            f="$OPTARG"
            ;;
        o)
            o="$OPTARG"
            ofs_flag='true'
            ;;
        n)
            n=':'
            ;;
        c)
            c='1'
            ;;
        g)
            cmd='gawk'
            ;;
        s)
            s='1'
            ;;
        F)
            F='1'
            ;;
        \?)
            echo "Invalid option: -$OPTARG" >&2
            exit 1
            ;;
        :)
            echo "Option -$OPTARG requires an argument." >&2
            exit 1
            ;;
    esac
done
# Discard the options and sentinel --
shift "$((OPTIND-1))"

# argument to -f option cannot be empty
# it can have only digits comma and hyphen characters
# if -n is active, colon is also allowed
if [[ ! "$f" =~ ^[${n}0-9,-]+$ ]]; then
    echo "Field number can only use integer values" >&2
    exit 1
fi

# if -o option isn't used, two cases where OFS will set to same value as IFS
# 1) if -F option is used
# 2) if IFS is a single character (partially mimics cut command behavior)
# \ character needs to be doubled since it is a metacharacter in awk strings
if [[ "$ofs_flag" == 'false' && ( ${#d} == 1 || "$F" == '1' ) ]] ; then
    o="${d//\\/\\\\}"
fi

# if -F option is used, change IFS such that it matches literally
if [[ "$F" == '1' ]] ; then
    d="$(echo "$d" | "$cmd" '{gsub(/[^^\\]/, "[&]");
                              gsub(/\\/, "\\x5c\\x5c");
                              gsub(/\^/, "\\x5c\\x5e")} 1')"
fi

# corner case: when -o is \ character (needed for gawk)
# since -o is used, this isn't covered by earlier \ doubling
if [[ "$o" == '\' ]] ; then
    o='\\'
fi

"$cmd" -F "$d" -v fields="$f" -v OFS="$o" -v neg="$n" -v complement="$c" -v suppress="$s" '
    BEGIN{
        fn = split(fields, fields_arr, /,/)
    }

    {
        if(suppress && NF<=1)
            next

        sep = ""

        for(i=1; i<=fn; i++){
            if(neg){
                rn = split(fields_arr[i], range, /:/)
                if(range[1] ~ /^-/) range[1] += NF + 1
                if(range[2] ~ /^-/) range[2] += NF + 1
            }
            else
                rn = split(fields_arr[i], range, /-/)

            start = range[1] <= 0 ? 1: range[1]
            if(start > NF) start = NF

            if(rn == 1)
                end = start
            else if(range[2] == "" || range[2] > NF)
                end = NF
            else if(range[2] <= 0)
                end = 1
            else
                end = range[2]

            for(j=start; j<=end; j++)
                if(complement)
                    ignore_fields[j]
                else {
                    printf "%s%s", sep, $j
                    sep = OFS
                }
        }

        if(complement){
            for(k=1; k<=NF; k++)
                if(!(k in ignore_fields)){
                    printf "%s%s", sep, $k
                    sep = OFS
                }
            delete ignore_fields
        }

        print ""
    }' "$@"