analyze_crashes_winver.py

#!/usr/bin/env python
""" Detect Windows versions in crash reports

Runs on *pub-crashdata.csv.gz files from:

    https://crash-analysis.mozilla.com/crash_analysis
"""
from __future__ import division, print_function

import sys
import gzip
import csv
import collections

# Simplified from:
# http://msdn.microsoft.com/en-us/library/windows/desktop/ms724832(v=vs.85).aspx
NT_VERSIONS = {
   '6.3': '8',
   '6.2': '8',
   '6.1': '7',
   '6.0': 'Vista',
   '5.2': 'XP',
   '5.1': 'XP',
   '5.0': '2000'}

ORDERED_VERSIONS = ['8',
                    '7',
                    'Vista',
                    'XP',
                    '2000']


def read_file(fname):
    with gzip.open(fname, 'rt') as fobj:
        reader = csv.DictReader(fobj, dialect=csv.excel_tab)
        rows = list(reader)
    return rows


def windows_version(row):
    if not row['os_name'] == 'Windows NT':
        return None
    maj_min = '.'.join(row['os_version'].split('.', 2)[:2])
    return NT_VERSIONS.get(maj_min)


def version_percents(rows):
    versions = [windows_version(row) for row in rows]
    good_versions = [v for v in versions if not v is None]
    scalar = 100. / len(good_versions)
    counts = collections.Counter(good_versions)
    return dict((key, value * scalar) for key, value in counts.items())


def main():
    for fname in sys.argv[1:]:
        rows = read_file(fname)
        pcts = version_percents(rows)
        print(fname, ':')
        for ver in ORDERED_VERSIONS:
            print('    {0:<6}: {1:8.2f}'.format(ver, pcts.get(ver, 0)))


if __name__ == '__main__':
    main()