def handle(self, *args, **options):
date_string = datetime.datetime.now().strftime("%d_%m_%Y-%H_%M_%S")
default_filename = f"backup_{date_string}.tar"
full_backup = options.get('full_backup')
username = options.get('user')
location = options.get('location') or default_filename
try:
user = User.objects.get(username=username)
except User.DoesNotExist:
logger.error(f"Username {username} not found.")
return
# consider the case when user provides directory location i.e.
# ./manage.py backup /backup/papermerge/
if os.path.isdir(location):
# in case location is a non existing directory path e.g. "blah/"
# os.path.isdir will return False
backup_location = os.path.join(location, default_filename)
else:
backup_location = location
try:
with open(backup_location, 'wb') as backup_file:
backup_documents(backup_file=backup_file,
user=user,
full_backup=full_backup)
except IsADirectoryError:
logger.error(
"Provided location is a directory which does not exist."
"Please create it first and try again.")
def test_backup_single_document(self):
document_path = os.path.join(
BASE_DIR, "data", "berlin.pdf"
)
doc = Document.create_document(
user=self.testcase_user,
title='berlin.pdf',
size=os.path.getsize(document_path),
lang='deu',
file_name='berlin.pdf',
parent_id=None,
page_count=3
)
default_storage.copy_doc(
src=document_path,
dst=doc.path.url(),
)
with io.BytesIO() as memoryfile:
backup_documents(memoryfile, self.testcase_user)
memoryfile.seek(0)
self.assertTrue(
_can_restore(memoryfile),
'generated backup.tar is not valid'
)
memoryfile.seek(0)
backup_file = tarfile.open(fileobj=memoryfile, mode='r')
backup_json = backup_file.extractfile('backup.json')
backup_info = json.loads(backup_json.read())
self.assertIsNotNone(
backup_info.get('documents'),
'backup.json did not have a key "documents"'
)
self.assertIs(
len(backup_info.get('documents')), 1,
'backup.json key documents had more or less than one entry'
)
self.assertIs(
len(backup_file.getnames()),
2,
'backup.tar had more or less than 2 entries'
)
self.assertTrue(
'berlin.pdf' in backup_file.getnames(),
'berlin.pdf was not in the backup.tar'
)
def test_backup_document_hierachy(self):
folder_1 = Folder.objects.create(title='1',
parent=None,
user=self.testcase_user)
folder_2 = Folder.objects.create(title='2',
parent=folder_1,
user=self.testcase_user)
folder_3 = Folder.objects.create(title='3',
parent=folder_1,
user=self.testcase_user)
Folder.objects.create(title='4', parent=None, user=self.testcase_user)
document_path = os.path.join(BASE_DIR, "data", "berlin.pdf")
doc_1 = Document.create_document(user=self.testcase_user,
title='berlin.pdf',
size=os.path.getsize(document_path),
lang='deu',
file_name='berlin.pdf',
parent_id=folder_2.id,
page_count=3)
default_storage.copy_doc(
src=document_path,
dst=doc_1.path.url(),
)
doc_2 = Document.create_document(user=self.testcase_user,
title='berlin.pdf',
size=os.path.getsize(document_path),
lang='deu',
file_name='berlin.pdf',
parent_id=folder_3.id,
page_count=3)
default_storage.copy_doc(
src=document_path,
dst=doc_2.path.url(),
)
with io.BytesIO() as memoryfile:
backup_documents(memoryfile, self.testcase_user)
memoryfile.seek(0)
self.assertTrue(_can_restore(memoryfile),
'generated backup.tar is not valid')
memoryfile.seek(0)
backup_file = tarfile.open(fileobj=memoryfile, mode='r')
backup_json = backup_file.extractfile('backup.json')
backup_info = json.loads(backup_json.read())
self.assertIsNotNone(backup_info.get('documents'),
'backup.json did not have a key "documents"')
self.assertIs(
len(backup_info.get('documents')), 2,
'backup.json key documents had more or less than two entry')
self.assertIs(len(backup_file.getnames()), 3,
'backup.tar had more or less than 2 entries')
self.assertTrue(
f"1/2/berlin.pdf__{doc_1.id}" in backup_file.getnames(),
'berlin.pdf was not in the backup.tar at folder 1/2/')
self.assertTrue(
f"1/3/berlin.pdf__{doc_2.id}" in backup_file.getnames(),
'berlin.pdf was not in the backup.tar at folder 1/3/')
self.assertFalse(
'4' in backup_file.getnames(),
'Folder 4 was in backup.tar but should have been ignored')