Bash script to output a fixed number of random lines from a file.

pluck.sh

#!/bin/sh

#
# pluck - Pluck some random lines from a file.
#
# This simple tool chooses a fixed number of random lines from a file and
# outputs them to stdout (in order of their line number).
#

function usage {
	echo "usage:\t`basename $0` filename [k=100]"
}

#
# Samples from the set of numbers 0..n, with replacement, k times. Requires 2
# arguments:
#  n The total number of numbers.
#  k The number of samples to take.
#
function sample {
	for (( COUNTER=1; COUNTER<=$2; COUNTER+=1 )); do
		echo `expr $RANDOM \% $1`
	done | sort -b -n
}

if [ -z "$1" ]; then
	usage
	exit 1
fi

file=$1
n=`cat $file | wc -l`
k=100
if [ -n "$2" ]; then
	k="$2"
fi

sed = $file | sed "N;s/\n/ /" | grep -P "^(`sample $n $k | tr '\n' '|'`x) " | cut -d ' ' -f 2-