def process_zip_archive(file, id):
filename = secure_filename(file.filename)
time_stamp = str(int(round(time.time())))
file_save_directory = os.path.join(current_app.config['CFG_DATADIR'], str(id), time_stamp)
if not os.path.exists(file_save_directory):
os.makedirs(file_save_directory)
if '.oldhepdata' not in filename:
file_path = os.path.join(file_save_directory, filename)
file.save(file_path)
submission_path = os.path.join(file_save_directory, remove_file_extension(filename))
if 'yaml' in filename:
# we split the singular yaml file and create a submission directory
split_files(file_path, submission_path)
else:
# we are dealing with a zip, tar, etc. so we extract the contents
extract(filename, file_path, submission_path)
submission_found = find_file_in_directory(submission_path,
lambda x: x == "submission.yaml")
else:
file_path = os.path.join(file_save_directory, 'oldhepdata')
if not os.path.exists(file_path):
os.makedirs(file_path)
if filename.endswith('.txt'):
filename = filename.replace(".txt", "")
print('Saving file to {}'.format(os.path.join(file_path, filename)))
file.save(os.path.join(file_path, filename))
submission_path = os.path.join(file_save_directory, 'oldhepdata')
submission_found = False
if submission_found:
basepath, submission_file_path = submission_found
else:
result = check_and_convert_from_oldhepdata(submission_path, id,
time_stamp)
# Check for errors
if type(result) == dict:
return result
else:
basepath, submission_file_path = result
return process_submission_directory(basepath, submission_file_path, id)
def test_file_download_and_split(app, migrator, identifiers):
"""___test_file_download_and_split___"""
with app.app_context():
for test_id in identifiers:
temp_file = migrator.download_file(test_id["hepdata_id"])
assert temp_file is not None
split_files(
temp_file,
os.path.join(app.config['CFG_TMPDIR'], test_id["hepdata_id"]),
os.path.join(app.config['CFG_TMPDIR'],
test_id["hepdata_id"] + ".zip"))
assert (os.path.exists(
os.path.join(app.config['CFG_TMPDIR'], test_id["hepdata_id"])))
def process_zip_archive(file, id):
filename = secure_filename(file.filename)
time_stamp = str(int(round(time.time())))
file_save_directory = os.path.join(current_app.config['CFG_DATADIR'],
str(id), time_stamp)
if not os.path.exists(file_save_directory):
os.makedirs(file_save_directory)
if not filename.endswith('.oldhepdata'):
file_path = os.path.join(file_save_directory, filename)
file.save(file_path)
submission_path = os.path.join(file_save_directory,
remove_file_extension(filename))
if filename.endswith('.yaml'):
# we split the singular yaml file and create a submission directory
error, last_updated = split_files(file_path, submission_path)
if error:
return {
"Single YAML file splitter": [{
"level": "error",
"message": str(error)
}]
}
else:
# we are dealing with a zip, tar, etc. so we extract the contents
extract(filename, file_path, submission_path)
submission_found = find_file_in_directory(
submission_path, lambda x: x == "submission.yaml")
else:
file_path = os.path.join(file_save_directory, 'oldhepdata')
if not os.path.exists(file_path):
os.makedirs(file_path)
print('Saving file to {}'.format(os.path.join(file_path, filename)))
file.save(os.path.join(file_path, filename))
submission_path = os.path.join(file_save_directory, 'oldhepdata')
submission_found = False
if submission_found:
basepath, submission_file_path = submission_found
else:
result = check_and_convert_from_oldhepdata(submission_path, id,
time_stamp)
# Check for errors
if type(result) == dict:
return result
else:
basepath, submission_file_path = result
return process_submission_directory(basepath, submission_file_path, id)
def prepare_files_for_submission(self, inspire_id, force_retrieval=False):
"""
Either returns a file if it already exists, or downloads it and
splits it.
:param inspire_id:
:return: output location if succesful, None if not
"""
output_location = os.path.join(current_app.config['CFG_DATADIR'], inspire_id)
if not os.path.exists(output_location) or force_retrieval:
print('Downloading file for {0}'.format(inspire_id))
file_location = self.download_file(inspire_id)
if file_location:
output_location = os.path.join(current_app.config['CFG_DATADIR'], inspire_id)
split_files(file_location, output_location, '{0}.zip'.format(output_location))
else:
return None
else:
print('File for {0} already in system...no download required.'.format(inspire_id))
return output_location
def prepare_files_for_submission(self, inspire_id, force_retrieval=False):
"""
Either returns a file if it already exists, or downloads it and
splits it.
:param inspire_id:
:return: output location if successful, None if not
"""
output_location = os.path.join(current_app.config["CFG_DATADIR"],
inspire_id)
last_updated = datetime.now()
download = not os.path.exists(output_location) or (
get_file_in_directory(output_location, 'yaml') is None)
if download or force_retrieval:
print("Downloading file for {0}".format(inspire_id))
file_location = self.download_file(inspire_id)
if file_location:
output_location = os.path.join(
current_app.config["CFG_DATADIR"], inspire_id)
error, last_updated = split_files(
file_location, output_location,
"{0}.zip".format(output_location))
# remove temporary download file after processing
try:
os.remove(file_location)
except:
log.info('Unable to remove {0}'.format(file_location))
else:
output_location = None
else:
print("File for {0} already in system...no download required.".
format(inspire_id))
return output_location, last_updated
def process_zip_archive(file, id):
filename = secure_filename(file.filename)
time_stamp = str(int(round(time.time())))
file_save_directory = os.path.join(current_app.config['CFG_DATADIR'],
str(id), time_stamp)
if not os.path.exists(file_save_directory):
os.makedirs(file_save_directory)
if not filename.endswith('.oldhepdata'):
file_path = os.path.join(file_save_directory, filename)
print('Saving file to {}'.format(file_path))
file.save(file_path)
submission_path = os.path.join(file_save_directory,
remove_file_extension(filename))
submission_temp_path = tempfile.mkdtemp(
dir=current_app.config["CFG_TMPDIR"])
if filename.endswith('.yaml'):
# we split the singular yaml file and create a submission directory
error, last_updated = split_files(file_path, submission_temp_path)
if error:
return {
"Single YAML file splitter": [{
"level": "error",
"message": str(error)
}]
}
else:
# we are dealing with a zip, tar, etc. so we extract the contents
if not extract(file_path, submission_temp_path):
return {
"Archive file extractor": [{
"level":
"error",
"message":
"{} is not a valid zip or tar archive file.".format(
file_path)
}]
}
if not os.path.exists(submission_path):
os.makedirs(submission_path)
# Move files from submission_temp_path to submission_path (try to avoid problems with EOS disk).
if current_app.config.get('PRODUCTION_MODE',
False): # production instance at CERN
copy_command = ['xrdcp', '-N', '-f']
copy_submission_path = submission_path.replace(
current_app.config['CFG_DATADIR'],
current_app.config['EOS_DATADIR'])
else: # local instance
copy_command = ['cp']
copy_submission_path = submission_path
print('Copying with: {} -r {} {}'.format(' '.join(copy_command),
submission_temp_path + '/.',
copy_submission_path))
subprocess.check_output(
copy_command +
['-r', submission_temp_path + '/.', copy_submission_path])
rmtree(submission_temp_path, ignore_errors=True
) # can uncomment when this is definitely working
submission_found = find_file_in_directory(
submission_path, lambda x: x == "submission.yaml")
else:
file_path = os.path.join(file_save_directory, 'oldhepdata')
if not os.path.exists(file_path):
os.makedirs(file_path)
print('Saving file to {}'.format(os.path.join(file_path, filename)))
file.save(os.path.join(file_path, filename))
submission_found = False
if submission_found:
basepath, submission_file_path = submission_found
else:
result = check_and_convert_from_oldhepdata(file_path, id, time_stamp)
# Check for errors
if type(result) == dict:
return result
else:
basepath, submission_file_path = result
return process_submission_directory(basepath, submission_file_path, id)
def process_zip_archive(file_path, id, old_submission_schema=False,
old_data_schema=False):
(file_save_directory, filename) = os.path.split(file_path)
if not filename.endswith('.oldhepdata'):
file_save_directory = os.path.dirname(file_path)
submission_path = os.path.join(file_save_directory, remove_file_extension(filename))
submission_temp_path = tempfile.mkdtemp(dir=current_app.config["CFG_TMPDIR"])
if filename.endswith('.yaml.gz'):
print('Extracting: {} to {}'.format(file_path, file_path[:-3]))
if not extract(file_path, file_path[:-3]):
return {
"Archive file extractor": [{
"level": "error", "message": "{} is not a valid .gz file.".format(file_path)
}]
}
return process_zip_archive(file_path[:-3], id,
old_submission_schema=old_submission_schema,
old_data_schema=False)
elif filename.endswith('.yaml'):
# we split the singular yaml file and create a submission directory
error, last_updated = split_files(file_path, submission_temp_path)
if error:
return {
"Single YAML file splitter": [{
"level": "error",
"message": str(error)
}]
}
else:
# we are dealing with a zip, tar, etc. so we extract the contents
try:
unzipped_path = extract(file_path, submission_temp_path)
except Exception as e:
unzipped_path = None
if not unzipped_path:
return {
"Archive file extractor": [{
"level": "error", "message": "{} is not a valid zip or tar archive file.".format(file_path)
}]
}
copy_errors = move_files(submission_temp_path, submission_path)
if copy_errors:
return copy_errors
submission_found = find_file_in_directory(submission_path, lambda x: x == "submission.yaml")
if not submission_found:
return {
"Archive file extractor": [{
"level": "error", "message": "No submission.yaml file has been found in the archive."
}]
}
basepath, submission_file_path = submission_found
else:
file_dir = os.path.dirname(file_save_directory)
time_stamp = os.path.split(file_dir)[1]
result = check_and_convert_from_oldhepdata(os.path.dirname(file_save_directory), id, time_stamp)
# Check for errors
if type(result) == dict:
return result
else:
basepath, submission_file_path = result
old_data_schema = True
return process_submission_directory(basepath, submission_file_path, id,
old_data_schema=old_data_schema,
old_submission_schema=old_submission_schema)
