Compute average response time for given SPARQL query and endpoint

time_sparql.sh

#!/bin/bash
#
# Compute average response time for given SPARQL query and endpoint
#
# Note: For proper benchmarking use SPARQL Query Benchmarker instead.
# <https://sourceforge.net/p/sparql-query-bm/wiki/Introduction/>

set -e

die () {
  echo >&2 "$@"
  exit 1
}

usage () {
  echo "Computes average response time for given SPARQL query and endpoint.

Usage: $(basename "$0") [-vh] -q query -e endpoint [-n runs] [-p pause]
  [query]    = Path to file with a SPARQL query (required)
  [endpoint] = URL of a SPARQL endpoint (required)
  [runs]     = Number of times the query should be run (defaults to 10)
  [pause]    = Number of seconds to pause between queries (defaults to 1)
  -v         = Print verbose output
  -h         = Print this help message"
}

# Argument defaults
n=10
pause=1
verbose='false'

while getopts :q:e:n:p:vh opt
do
  case $opt in
    q) query="$OPTARG" ;;
    e) endpoint="$OPTARG" ;;
    n) n=$OPTARG ;;
    p) pause=$OPTARG ;;
    v) verbose='true' ;;
    h) usage; exit ;;
    :) die "Missing argument for -$OPTARG" ;; 
    \?) die "Illegal option: -$OPTARG" ;;
  esac
done

# Validate arguments
[ $query ] || die "Missing SPARQL query!"
[ $endpoint ] || die "Missing SPARQL endpoint!"
! [ -f "$query" ] && die "File $query doesn't exist!"
! [ -r "$query" ] && die "File $query cannot be read!"
echo $n | grep -E -q "^[0-9]+$" || die "Number of runs must be a number!"
echo $pause | grep -E -q "^[0-9]+(\.[0-9]+)?$" || die "Pause must be a number!"
[ $(curl -o /dev/null -s -I --write-out "%{http_code}" "$endpoint") -eq 200 ] || \
  die "SPARQL endpoint $endpoint is unreachable!"

rtimes=()
$verbose && printf "Response times:\n"
for (( i=1; i<=$n; i++ ))
do
  rtime=$(curl -s -G --data-urlencode query@"$query" "$endpoint" -o /dev/null -w "%{time_starttransfer}")
  rtimes+=($rtime)
  $verbose && echo "$rtime"
  sleep $pause
done

# Compute average and standard deviation
avg=$(IFS=+; echo "scale=2;(${rtimes[*]})/${#rtimes[@]}" | bc | awk '{printf "%.2f", $0}')
stdev=$(for rtime in ${rtimes[@]}; do echo "$rtime"; done | \
        awk '{sum+=$0; sumsq+=$0 * $0} END {print sqrt(sumsq/NR - (sum/NR)**2)}')
echo "Average response time: $avg"
echo "Standard deviation: $stdev"