def publish_input_data(self, expected_input_id: str,
metadata: InputMetadata,
input_data_stream: BinaryIO) -> None:
input_file_path = self.input_file(expected_input_id)
if os.path.exists(input_file_path):
input_data_stream.close()
return
file_hash = hashlib.sha256()
fd, temp_file_path = tempfile.mkstemp(dir=self.temp_data_dir)
try:
with os.fdopen(fd, 'wb') as f:
bytes_read = 0
while True:
data = input_data_stream.read(READ_BUFFER_SIZE)
bytes_read += len(data)
log.debug(f'{bytes_read} bytes of input read')
if not data:
break
f.write(data)
file_hash.update(data)
input_id = file_hash.hexdigest()
if input_id != expected_input_id:
raise IncorrectInputIDException()
os.rename(temp_file_path, input_file_path)
if metadata.has_all_args():
self._store_input_id(metadata, input_id)
return
except Exception:
os.remove(temp_file_path)
raise
def check_input_data(self, input_id: str, metadata: InputMetadata) -> bool:
if self._input_file_exists(input_id):
if metadata.has_all_args():
# The reason to do this is that, if there's a blob that
# changed timestamp but not hash (because of a `touch`, for
# instance), the timestamp check will always return false and
# the tarball will be constructed all the times on the client
# side. It happened.
self._store_input_id(metadata, input_id)
return True
else:
return False