def _copy_input_files(self, writer):
for file in self._input_files:
dest_file = "/tmp/submission-src"
put_file(container=writer, src=file, dest=dest_file)
if hasattr(file, "content_type"):
mimetype = file.content_type
else:
with file.open("rb") as f:
mimetype = get_file_mimetype(f)
if mimetype.lower() == "application/zip":
# Unzip the file in the container rather than in the python
# process. With resource limits this should provide some
# protection against zip bombs etc.
writer.exec_run(
f"unzip {dest_file} -d /input/ -x '__MACOSX/*'")
# Remove a duplicated directory
input_files = (writer.exec_run(
"ls -1 /input/").output.decode().splitlines())
if (len(input_files) == 1 and not writer.exec_run(
f"ls -d /input/{input_files[0]}/").exit_code):
writer.exec_run(
f'/bin/sh -c "mv /input/{input_files[0]}/* /input/ '
f'&& rm -r /input/{input_files[0]}/"')
elif mimetype.lower() == "application/json":
writer.exec_run(f"mv {dest_file} /input/predictions.json")
else:
# Not a zip file, so must be a csv
writer.exec_run(f"mv {dest_file} /input/submission.csv")
def _copy_input_files(self, writer):
for file in self._input_files:
dest_file = "/tmp/submission-src"
put_file(container=writer, src=file, dest=dest_file)
with file.open("rb") as f:
mimetype = get_file_mimetype(f)
if mimetype.lower() == "application/zip":
# Unzip the file in the container rather than in the python
# process. With resource limits this should provide some
# protection against zip bombs etc.
writer.exec_run(
f"unzip {dest_file} -d /input/ -x '__MACOSX/*'")
self.__was_unzipped = True
else:
# Not a zip file, so must be a csv
writer.exec_run(f"mv {dest_file} /input/submission.csv")