Simple commandline program to construct ASCII histograms

histogram.sh

set -beEo pipefail

DEFAULTNUMBINS=10
DEFAULTMAXWIDTH=75

show_help() {
	cat << EOF
Usage: ${0##*/} [OPTIONS]
Takes a set of numbers and plots an ASCII histogram. The input must be fed in
through stdin, and must consist of one number on each line.
Options:
	-n|--numbins: the number of equally sized bins to use; defaults to $DEFAULTNUMBINS;
		conflicts with binsize
	-s|--binsize: the size of the bin to use; conflicts with numbins
	-m|--minbin: the minimum edge of the first bin; defaults to the minimum
		element
	-w|--maxwidth: maximum length of the histogram bars in characters; defaults
		to $DEFAULTMAXWIDTH
EOF
}

POSARGS=""  # Positional arguments
while [ $# -gt 0 ]
do
	case "$1" in
		-h|--help)
			show_help
			exit 0
			;;
		-n|--numbins)
			numbins=$2
			shift 2
			;;
		-s|--binsize)
			binsize=$2
			shift 2
			;;
		-m|--minbin)
			minbin=$2
			shift 2
			;;
		-w|--maxwidth)
			maxwidth=$2
			shift 2
			;;
		-*|--*)
			echo "Unsupported flag error: $1" >&2
			show_help >&2
			exit 1
			;;
		*)
			POSARGS="$POSARGS $1"  # Preserve positional arguments
			shift
	esac
done
eval set -- "$POSARGS"  # Restore positional arguments to expected indices

if [[ ! -z $1 ]]
then
	show_help
	exit 1
fi

if [[ ! -z $binsize ]] && [[ ! -z $numbins ]]
then
	# Cannot specify both
	show_help
	exit 1
fi

# Pipe stdin to temporary file
tempfile=$(mktemp)
cat - | sort -n > $tempfile

maxval=$(tail -n 1 $tempfile)

if [[ -z $minbin ]]
then
	minval=$(head -n 1 $tempfile)
else
	minval=$minbin
fi

if [[ -z $maxwidth ]]
then
	maxwidth=$DEFAULTMAXWIDTH
fi

if [[ -z $binsize ]]
then
	# Figure out the bin size
	if [[ -z $numbins ]]
	then
		numbins=$DEFAULTNUMBINS
	fi
	
	binsize=$(echo "(${maxval}-(${minval}))/${numbins}" | bc -l)
else
	numbins=$(echo "(${maxval}-(${minval}))/${binsize}" | bc)
fi

cat $tempfile | awk -v binsize=$binsize -v numbins=$numbins -v minval=$minval -v maxwidth=$maxwidth '
	BEGIN{
		# First, figure out how many digits of precision we need to display
		maxval = minval + (numbins * binsize);
		minabs = minval > 0 ? minval : minval * -1;
		maxabs = maxval > 0 ? maxval : maxval * -1;
		bestabs = maxval > minval ? maxval : minval;
		leftdigits = int(log(bestabs) / log(10)) + 1;
		rightdigits = int(-1 * log(binsize) / log(10)) + 1;
		rightdigits += 1;  # Extra padding for precision
		totallength = leftdigits + 1 + rightdigits;
	}
	{
		# Count how many entries are in each bin
		binindex = int(($1 - minval) / binsize);
		counts[binindex] += 1;
	}
	END{
		# Figure out the maximum count to determine the mapping from count to bar length
		maxcount = 0;
		for (bin in counts) {
			if (counts[bin] > maxcount) {
				maxcount = counts[bin];
			}
		}
		
		widthfactor = maxwidth / maxcount;
		
		# For each bin, print out the starting point and the bar
		for (i = 0; i <= numbins; i++) {
			binstart = minval + (i * binsize);
			if (minval < 0) {
				if (binstart < 0) {
					printf "-";
				} else {
					printf " ";
				}
			}
			binstart *= binstart < 0 ? -1 : 1;
			printf "%*.*f<|", totallength, rightdigits, binstart;
			for (j = 0; j < (widthfactor * counts[i]); j++) {
				printf "*";
			}
			printf "\n";
		}
	}
'

rm $tempfile