#! /bin/bash
# extract specified columns from hydrobal.out. Columns are 
# are enclosed in |. Excess columns are removed one at a time.
# columns 1, 2, 3, 7, and 11 cannot be retained

#echo "$*"
# check for input file name existence
if [[ $1 = "" ]]
then
    echo 'invocation: getcols [file-1 file-2] [Col-1 Col-2 ... Col-n]'
    echo 'File names are required'
    exit
fi
infile="$1"
shift

# check for output file name existence
if [[ $1 = "" ]]
then
    echo 'invocation: getcols [file-1 file-2] [Col-1 Col-2 ... Col-n]'
    echo 'File names are required'
    exit
fi
outfile="$1"
shift

# invocation: getcols [file-1 file-2] [Col-1 Col-2 ... Col-n]

# set up column and title correspondence
mincol=4
maxcol=18
coltitle[4]=`echo "4 - start day of hydrobalance period"`
coltitle[5]=`echo "5 - starting soil water content (mm)"`
coltitle[6]=`echo "6 - starting snow water content (mm)"`
coltitle[7]=`echo "7 - DO NOT USE THIS NUMBER text field (auto delete)"`
coltitle[8]=`echo "8 - end day of hydrobalance period"`
coltitle[9]=`echo "9 - ending soil water content (mm)"`
coltitle[10]=`echo "10 - ending snow water content (mm)"`
coltitle[11]=`echo "11 - DO NOT USE THIS NUMBER text field (auto delete)"`
coltitle[12]=`echo "12 - rainfall for period (mm)"`
coltitle[13]=`echo "13 - runoff for period (mm)"`
coltitle[14]=`echo "14 - evaporation for period (mm)"`
coltitle[15]=`echo "15 - transpiration for period (mm)"`
coltitle[16]=`echo "16 - drainage for period (mm)"`
coltitle[17]=`echo "17 - check for period (mm)"`
coltitle[18]=`echo "18 - fallow efficiency (delta swc / precip)"`

if [[ $1 = "" ]]
then
    echo 'invocation: getcols [file-1 file-2] [Col-1 Col-2 ... Col-n]'
    echo 'Valid Column Numbers are:'
    for num in $( seq $mincol $maxcol )
    do
        echo ${coltitle[num]}
    done
    exit
fi

# make sure column numbers are within range
# set up array of selected column numbers. Values in array index 0 to maxcolnum-1
maxcolnum=0
for colnum in $*
do
    if [[ ${colnum} -lt $mincol || ${colnum} -gt $maxcol || ${colnum} -eq 7 || ${colnum} -eq 11 ]]
    then
        echo 'invocation: getcols [file-1 file-2] [Col-1 Col-2 ... Col-n]'
        echo 'Valid Column Numbers are:'
        for num in $( seq $mincol $maxcol )
        do
            echo ${coltitle[num]}
        done
        exit
    fi
    selcolnum[maxcolnum]=$colnum
    let maxcolnum=maxcolnum+1
done
selcolnum[maxcolnum]=0

# make sure column numbers are in ascending order
# if they are not in order, the larger number will not allow retaining the smaller numbers that follow

# remove initial date field column
sed s/[^\|.]*\|/\|/ "${infile}" > "${outfile}"
cp "${outfile}" atemp
#echo "atemp updated - First column"
#read

# with column 1 removed, set colnum to 2
colnum=2
selindex=0
blocknum=1

while [ `grep -c \| atemp` -gt 0 ]
do
    if [[ ${colnum} -eq 1 || ${colnum} -eq 2 || ${colnum} -eq 3 || ${colnum} -eq 7 || ${colnum} -eq 11 ]]
    then
        # automatically deleted columns
        sed s/\|[^\|]*\|/\|/ atemp > "${outfile}"   # remove autodelete column
        #echo "${outfile} updated - autodelete column ${colnum}"
        #read
        let colnum=colnum+1
        cp "${outfile}" atemp
    fi

    # selected columns
    #echo "checking column $colnum to retain column ${selcolnum[selindex]} at selindex $selindex"
    if [[ ${colnum} -eq ${selcolnum[selindex]} ]]
    then
        sed s/\|/\ / atemp > "${outfile}"            # remove extra column separator
        let selindex=selindex+1
        #echo "${outfile} updated - column ${colnum} retained"
        #read
    else
        # count number of delimiters left in first line
        delimcnt=`sed -n '1p' < atemp | tr -cd \| | wc -c`
        #echo "delimcnt = $delimcnt"
        if [ $delimcnt -gt 1 ]
        then
            # remove column between | characters
            sed s/\|[^\|]*\|/\|/ atemp > "${outfile}"
        else
            # remove last column
            sed s/\|[^\|]*$// atemp > "${outfile}"
        fi
        #echo "${outfile} updated - column ${colnum} deleted"
        #read
    fi
    let colnum=colnum+1
    cp "${outfile}" atemp

    # reset counters to get next column group
    if [[ ${colnum} -gt ${maxcol} ]]
    then
        colnum=1
        selindex=0
        #echo "Block $blocknum Completed"
        let blocknum=blocknum+1
    fi
done

rm atemp