def main():
files = sys.argv[1:]
cache_path = 'geo_counts.csv'
if os.path.exists(cache_path):
counts = load_cache(cache_path)
else:
counts = count_files(files, keepers, criteria)
write_counts(counts, cache_path)
plot_error_vs_time(counts, dates)
plot_country_error_vs_time(counts,dates)
plot_mom_vs_time(counts,dates)
basic_stats(counts)
def main():
parser = SquidArgumentParser()
parser.add_argument('--nprocs', default=10)
args = parser.parse_args()
logger.info(pprint.pformat(args.__dict__))
keepers = ['date', 'language', 'project', 'site', 'country', 'na']
criteria = [
lambda r : r.old_init_request(),
lambda r : r.site() == 'M',
lambda r : r.datetime() > args.start,
lambda r : r.datetime() < args.end,
]
counts = count_files(args.squid_files,
keepers,
criteria,
count_event=1000,
limit=args.max_lines,
nproc=15,
fname='country_counts_incremental.csv')
write_counts(counts, 'country_counts.csv')