cal-ua.py

# -*- coding: utf-8 -*-
import argparse
import csv
import json
import httpagentparser
import os

oss = {}
browsers = {}


def calc(filename):
    fh = open(filename)
    rows = []
    for line in fh.readlines():
        info = httpagentparser.detect(json.loads(line.split('\t')[-1])['ua'])
        if 'browser' not in info:
            print('invalid agent: {}'.format(info))
            continue
        versions = browsers.setdefault(info['browser']['name'], {})
        versions.setdefault(info['browser']['version'], 0)
        browsers[info['browser']['name']][info['browser']['version']] += 1
        try:
            oss.setdefault(info['os']['name'], 0)
            oss[info['os']['name']] += 1
        except KeyError:
            pass
        rows.append([
            info.get('browser', '') and info.get('browser').get('name', ''),
            info.get('browser', '') and info.get('browser').get('version', ''),
            info.get('os', '') and info.get('os').get('name', ''),
            info.get('dist', '') and info.get('dist').get('name', ''),
            info.get('dist', '') and info.get('dist').get('version', ''),
        ])
    return rows


if __name__ == '__main__':
    parser = argparse.ArgumentParser()
    parser.add_argument('path')
    parser.add_argument('export')
    args = parser.parse_args()
    with open(args.export, 'w') as fh:
        writer = csv.writer(fh)
        writer.writerow([
            'browserName', 'browserVersion', 'OS', 'distName', 'distVersion'
        ])
    for filename in os.listdir(args.path):
        print('parsing {}...'.format(filename))
        with open(args.export, 'a+') as fh:
            writer = csv.writer(fh)
            rows = calc(os.path.join(args.path, filename))
            writer.writerows(rows)
    print('=' * 20)
    print('Browsers')
    print('=' * 20)
    for name, versions in browsers.items():
        print('## {}'.format(name))
        for v, count in versions.items():
            print(v, count)
    print('=' * 20)
    print('OSs')
    print('=' * 20)
    for name, count in oss.items():
        print('{}: {}'.format(name, count))