def task_ner(file_ids_list=None):
data_to_process = boilerplate.get_process_data(file_ids_list)
processed_file_ids = list()
for processed_file_id, contents in process_ner(data_to_process):
processed_file_ids.append(
boilerplate.add_processed_file(boilerplate.NER_PREFIX,
processed_file_id,
contents,
extension='txt'))
return processed_file_ids
def process_task(file_ids_list=None):
files_to_process = boilerplate.list_files(recursive=True,
prefix=boilerplate.UPLOAD_PREFIX)
if file_ids_list:
files_to_process = [
boilerplate.UPLOAD_PREFIX + file_id for file_id in file_ids_list
if (boilerplate.UPLOAD_PREFIX + file_id) in files_to_process
]
data_to_process = {
file_id[len(boilerplate.UPLOAD_PREFIX):]: boilerplate.get_file(file_id)
for file_id in files_to_process
}
processed_file_ids = list()
for processed_file_id, contents in process_data(data_to_process):
processed_file_ids.append(
boilerplate.add_processed_file(processed_file_id, contents))
return processed_file_ids
def process_task(file_ids_list=None):
files_to_process = boilerplate.list_files(recursive=True,
prefix=boilerplate.UPLOAD_PREFIX)
if file_ids_list:
files_to_process = [
boilerplate.UPLOAD_PREFIX + file_id for file_id in file_ids_list
if (boilerplate.UPLOAD_PREFIX + file_id) in files_to_process
]
data_to_process = {
file_id[len(boilerplate.UPLOAD_PREFIX):]: boilerplate.get_file(file_id)
for file_id in files_to_process
}
for filename, file_contents in data_to_process.items():
with open(join(TOMITA_PATH_IN, filename), 'wb') as f:
f.write(file_contents)
i = inotify.adapters.Inotify()
i.add_watch(TOMITA_PATH_OUT)
processed_file_ids = set()
for (_, type_names, path, out_filename) in i.event_gen(yield_nones=False):
print("PATH=[{}] FILENAME=[{}] EVENT_TYPES={}".format(
path, out_filename, type_names))
if not out_filename.startswith('.') and \
out_filename.endswith('.xml') and \
'IN_CLOSE_WRITE' in type_names:
full_filename = join(path, out_filename)
with open(full_filename, 'rb') as f:
contents = BytesIO(f.read())
contents_length = getsize(full_filename)
print(contents)
generated_filename = boilerplate.add_processed_file(
None, contents, "xml", contents_length)
processed_file_ids.add(generated_filename)
if len(processed_file_ids) >= len(set(data_to_process.keys())):
break
return list(processed_file_ids)
