diff-counts.py

#!/usr/bin/env python

import re
import sys

def parse(path):
  counts = {}
  sizes = {}
  for l in file(path):
    if ':' not in l:
      # skip header lines
      continue
    try:
      rank, count, size, clazz = re.split(r'\s+', l.strip())
    except:
      raise Exception("bad line: %s" % l)

    counts[clazz] = int(count)
    sizes[clazz] = int(size)
  return counts, sizes

def print_diff(before, after):
  all_keys = set(before.keys()).union(after.keys())
  diffs = []
  for clazz in all_keys:
    diffs.append((clazz, after.get(clazz, 0) - before.get(clazz, 0)))
  diffs.sort(key=lambda x: x[1])
  for clazz, diff in diffs:
    print "%s: %d -> %d (%s%d)" % (
      clazz, before.get(clazz, 0), after.get(clazz, 0),
      diff > 0 and '+' or '',
      diff)


before_counts, before_sizes = parse(sys.argv[1])
after_counts, after_sizes = parse(sys.argv[2])
print "count diffs:"
print "-" * 70
print_diff(before_counts, after_counts)
print "size diffs:"
print "-" * 70
print_diff(before_sizes, after_sizes)