Dead simple benchmark for testing sparkey. Running: $ python <amount-of-entries> <amount-of-random-reads> (or use pypy, speed guaranteed)

sparkeybench.py

from __future__ import print_function

import os
import sys
import sparkey
import tempfile
import timeit

from random import randint


def main(hashfile, logfile, size, lookups):
    writer = sparkey.HashWriter(hashfile, logfile)
    for i in xrange(size):
        writer.put('key_%d' % i, 'value_%d' % i)
    writer.close()

    reader = sparkey.HashReader(hashfile, logfile)
    for _ in xrange(lookups):
        reader['key_%d' % randint(0, size - 1)]
    reader.close()


if __name__ == '__main__':
    def _timed(size, lookups):
        try:
            _, hash = tempfile.mkstemp(dir='.')
            _, log = tempfile.mkstemp(dir='.')

            print('writing', hash, log)
            main(hash, log, size, lookups)
        finally:
            os.remove(hash)
            os.remove(log)
    print(timeit.repeat('_timed(%s)' % ','.join(sys.argv[1:]),
                        setup='from __main__ import _timed',
                        number=10, repeat=3))