1#!/usr/bin/env python
2
3import os, re, subprocess, sys
4
5result_re = re.compile(r'^\s*(\d+) good, (\d+\.\d+)s$', re.M)
6
7if len(sys.argv) > 1:
8    parser_exe = sys.argv[1]
9else:
10    parser_exe = ('dist/build/aeson-benchmark-aeson-parse/' +
11                  'aeson-benchmark-aeson-parse')
12
13def run(count, filename):
14    print '    %s :: %s times' % (filename, count)
15    p = subprocess.Popen([parser_exe, '65536', str(count), filename],
16                         stdout=subprocess.PIPE)
17    output = p.stdout.read()
18    p.wait()
19    m = result_re.search(output)
20    if not m:
21        print >> sys.stderr, 'run gave confusing output!?'
22        sys.stderr.write(output)
23        return
24    else:
25        #sys.stdout.write(output)
26        pass
27    good, elapsed = m.groups()
28    good, elapsed = int(good), float(elapsed)
29    st = os.stat(filename)
30    parses_per_second = good / elapsed
31    mb_per_second = st.st_size * parses_per_second / 1048576
32    print ('      %.3f seconds, %d parses/sec, %.3f MB/sec' %
33           (elapsed, parses_per_second, mb_per_second))
34    return parses_per_second, mb_per_second, st.st_size, elapsed
35
36def runtimes(count, filename, times=1):
37    for i in xrange(times):
38        yield run(count, filename)
39
40info = '''
41json-data/twitter1.json   60000
42json-data/twitter10.json  13000
43json-data/twitter20.json   7500
44json-data/twitter50.json   2500
45json-data/twitter100.json  1000
46json-data/jp10.json        4000
47json-data/jp50.json        1200
48json-data/jp100.json        700
49'''
50
51for i in info.strip().splitlines():
52    name, count = i.split()
53    best = sorted(runtimes(int(count), name, times=3), reverse=True)[0]
54    parses_per_second, mb_per_second, size, elapsed = best
55    print ('%.1f KB: %d msg\\/sec (%.1f MB\\/sec)' %
56           (size / 1024.0, int(round(parses_per_second)), mb_per_second))
57