trace_histogram.py

import collections
import json
import gzip
import sys

TOPN = 15

def print_hist(hist, evt_count, top=TOPN):
  for key, count in sorted(hist.iteritems(), key=lambda x:x[1], reverse=True):
    print '    %-32s  %6d %.2f %%' % (key, count, 100.0 * count / evt_count)
    top -= 1
    if top == 0:
      break


def main():
  with gzip.GzipFile(sys.argv[1]) as gzf:
    trace = json.load(gzf)
  evt_count = len(trace['traceEvents'])
  hist_cat = collections.defaultdict(int)
  hist_evt = collections.defaultdict(int)
  for evt in trace['traceEvents']:
    hist_cat[evt['cat']] += 1
    key = '%-12s - %-60s' % (evt['cat'], evt['name'])
    hist_evt[key] += 1
  print 'Total trace events: %d' % evt_count
  print '\nHistogram by category (top %d)' % TOPN
  print_hist(hist_cat, evt_count)

  print '\nHistogram by event (top %d)' % TOPN
  print_hist(hist_evt, evt_count)


if __name__ == '__main__':
  main()