parser.add_argument("--no-processing",
action='store_true',
help="Do not process the aggregated files",
default=False)
parser.add_argument(
'--sample',
nargs='?',
const=10,
type=int,
help="Include a sample only from each book. Default size 10.")
args = vars(parser.parse_args())
if args['list_books']:
print_books_list()
exit(0)
if args["books"] is not None and args["books"].lower() != "all":
args["books"] = [s.strip() for s in args["books"].split(",")]
for b in args["books"]: # Check if books exist
if b not in dict_to_long:
print(
"Unknown Book: %s, please use -l to list available books."
% b)
exit(-1)
else:
args["books"] = list(dict_to_long.keys())
print_banner()
aggregated_files = []
aggregate(args["books"], args["sample"])
args['no_processing'] and exit(0)
with Timer('process(aggregated_files)'):
process(aggregated_files)
