qwc

#!/bin/sh

usage()
{
cat << EOF
usage: $0 options FILENAME

Inspects a small subset of lines and overall file size to acheive a rough line
count, or if all lines are the same length (fixed-width, such as CSV) an exact
line count, without counting each individual line

OPTIONS:
   -h      Show this message
   -s      Sample size, number of lines to inspect (default: 15)
   -o      Offset, number of lines to ignore from beginning of file (defaut: 1)
              (Useful for CSV headers, script hash-bangs, etc)
   -l      No effect, useful for typing \`q!!\` after aborting a regular wc -l
   -v      Verbose
EOF
}

VERBOSE=0
OFFSET=1
SAMPLE=15

while getopts "vhlo:s:" OPTION; do
	case $OPTION in
	h)
		usage
		exit 1
		;;
	v)
		VERBOSE=1 ;;
	l)
		# Do nothing, for q!! shortcut
		;;
	o)
		OFFSET=$OPTARG ;;
	s)
		SAMPLE=$OPTARG ;;
	esac
done

shift $(($OPTIND - 1))
SAMPLE_FILE=$1

# the first line is usually different (script header, column definitions, etc)
SAMPLE_LINE_COUNT=$( head -$(($SAMPLE + $OFFSET )) "$SAMPLE_FILE" | wc -l )
if [ "$SAMPLE_LINE_COUNT" -lt $(( $SAMPLE + $OFFSET )) ]; then
	# echo "$0: Not enough lines for sampling: $SAMPLE + $OFFSET"
	# exit 1
	# Just give them regular answer, if they're willing to sample the entire file
	wc -l "$SAMPLE_FILE"
	exit
fi
AVG_LINE_SIZE=$(( $( head -$(($SAMPLE + $OFFSET )) "$SAMPLE_FILE" | tail -$SAMPLE | wc -c ) / $SAMPLE ))

TOTAL_FILE_SIZE=$( stat -c %s "$1" )

if [ "$VERBOSE" -eq 1 ]; then
	echo "Average line size: $AVG_LINE_SIZE"
	echo "Total file size: $TOTAL_FILE_SIZE"
fi

echo $(( $TOTAL_FILE_SIZE / $AVG_LINE_SIZE )) $SAMPLE_FILE