def init_writer(mr_schema):
"""
init writer
"""
print("Init writer ...")
mr_writer = FileWriter(args.mindrecord_file,
args.mindrecord_partitions)
# set the header size
if args.mindrecord_header_size_by_bit != 24:
header_size = 1 << args.mindrecord_header_size_by_bit
mr_writer.set_header_size(header_size)
# set the page size
if args.mindrecord_page_size_by_bit != 25:
page_size = 1 << args.mindrecord_page_size_by_bit
mr_writer.set_page_size(page_size)
# create the schema
mr_writer.add_schema(mr_schema, "mindrecord_graph_schema")
# open file and set header
mr_writer.open_and_set_header()
return mr_writer
try:
mindrecord_schema = mr_api.mindrecord_schema
except AttributeError:
raise RuntimeError("mindrecord_schema is not defined in mr_api.py.")
# create the schema
writer.add_schema(mindrecord_schema, "mindrecord_schema")
# add the index
try:
index_fields = mr_api.mindrecord_index_fields
writer.add_index(index_fields)
except AttributeError:
print("Default index fields: all simple fields are indexes.")
writer.open_and_set_header()
task_list = list(range(num_tasks))
# set number of workers
num_workers = args.mindrecord_workers
if num_tasks < 1:
num_tasks = 1
if num_workers > num_tasks:
num_workers = num_tasks
if os.name == 'nt':
for window_task_id in task_list:
_exec_task(window_task_id, False)
