def parse_args():
parser = SquidArgumentParser(
description="Process a collection of squid logs and write certain extracted metrics to file"
)
parser.add_argument(
"providers",
metavar="PROVIDER_IDENTIFIER",
nargs="*",
default=DEFAULT_PROVIDERS,
help="list of provider identifiers used in squid log file names",
)
parser.add_argument(
"--name_format",
dest="name_format",
type=str,
default="%s.log-%.gz.counts",
help="a printf style format string which is formatted with the tuple: (provider_name, date_representation",
)
parser.set_defaults(datadir="/home/erosen/src/dashboard/mobile/zero_counts")
args = parser.parse_args()
# custom logic for which files to grab
prov_files = {}
for prov in args.providers:
args.basename = prov
logging.info("args prior to ge_files: %s", pprint.pformat(args.__dict__))
prov_files[prov] = SquidArgumentParser.get_files(args)
setattr(args, "squid_files", prov_files)
logging.info(pprint.pformat(args.__dict__))
return args
def parse_args():
parser = SquidArgumentParser(
description="Process a collection of squid logs and write certain extracted metrics to file",
formatter_class=argparse.ArgumentDefaultsHelpFormatter,
)
parser.add_argument(
"providers",
metavar="PROVIDER_IDENTIFIER",
nargs="*",
default=DEFAULT_PROVIDERS,
help="list of provider identifiers used in squid log file names",
)
parser.add_argument(
"--name_format",
dest="name_format",
type=str,
default="%s.tab.log-%.gz",
help="a printf style format string which is formatted with the tuple: (provider_name, date_representation",
)
parser.set_defaults(datadir="/a/squid/archive/zero")
args = parser.parse_args()
# custom logic for which files to grab
prov_files = {}
for prov in args.providers:
basename = "zero-%s" % prov
logger.debug("basename: %s", basename)
prov_files[prov] = get_files(args.start, args.end, args.datadir, basename)
setattr(args, "squid_files", prov_files)
logger.info(pprint.pformat(args.__dict__))
return args
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')
