def verify(uri):
for path in walk(uri):
file_hash = os.path.basename(
os.path.dirname(path)) + os.path.basename(path)
new_file_hash = calculate_hash_for_file(path)
print(('Verification %s %s' %
(path, 'OK' if file_hash == new_file_hash else 'FAIL')))
def test_calculate_hash(self):
"""Test accurate hash generated from file"""
data = (
b'QlJBTlQgSVMgU1VQRVIgQVdFU09NRSBDT09MIEFORCBTSElULiBTTFVHUyBBUkUg'
b'Q1VURQ==')
expected_hash = '03439afa9d61f99f35d936b01ea8b4982ab247a0'
t = tempfile.NamedTemporaryFile(delete=False)
t.write(data)
t.close()
ret = utils.calculate_hash_for_file(t.name)
self.assertEqual(ret, expected_hash)
def test_calculate_hash(self):
"""Test accurate hash generated from file"""
data = (
b'QlJBTlQgSVMgU1VQRVIgQVdFU09NRSBDT09MIEFORCBTSElULiBTTFVHUyBBUkUg'
b'Q1VURQ=='
)
expected_hash = '03439afa9d61f99f35d936b01ea8b4982ab247a0'
t = tempfile.NamedTemporaryFile(delete=False)
t.write(data)
t.close()
ret = utils.calculate_hash_for_file(t.name)
self.assertEqual(ret, expected_hash)
def post(self, request, project_name):
project = get_object_or_404(Project, pk=project_name)
if ('path_id' not in request.POST and 'path' not in request.POST) or 'file' not in request.FILES:
return Response('Malformed request, needs to contain filename and file fields', status=400)
if 'path' in request.POST:
path = Path.objects.get_or_create(project, request.POST['path'])
else:
path_id = request.POST['path_id']
path = get_object_or_404(Path, project=project, pk=path_id)
file_obj = request.FILES['file']
filename = file_obj.name
#Write file to temporary location
with tempfile.NamedTemporaryFile(delete=False) as destination:
for chunk in file_obj.chunks():
destination.write(chunk)
destination.flush()
hash = calculate_hash_for_file(destination.name)
files_dir = '/tmp/damn/files'
if not os.path.exists(files_dir):
os.makedirs(files_dir)
new_path = os.path.join(files_dir, hash)
os.rename(destination.name, new_path)
analyzer = Analyzer()
try:
file_descr = analyzer.analyze_file(new_path)
except AnalyzerException as ae:
file_descr = FileDescription(file=FileId())
file_descr.file.hash = hash
file_ref = FileReference.objects.update_or_create(request.user, project, path, filename, file_descr)
return Response(serializers.FileReferenceSerializer(file_ref, context={'request': request}).data)
(path, 'OK' if file_hash == new_file_hash else 'FAIL')))
if __name__ == '__main__':
test_files_dir = os.path.join(os.path.dirname(__file__),
'../../../peragro-test-files')
for path in walk(test_files_dir):
file_hash, block_hashes = block_hashes_for_file(path)
block_hashes_to_file(file_hash, block_hashes,
'/tmp/files/' + hash_to_dir(file_hash))
blocks_for_file(path, '/tmp/blocks')
statistics('/tmp/files/')
for path in walk(test_files_dir):
file_hash = calculate_hash_for_file(path)
block_hashes = block_hashes_from_file('/tmp/files/' +
hash_to_dir(file_hash))
new_path = '/tmp/out/' + os.path.basename(path)
blocks_to_file('/tmp/blocks', block_hashes, new_path)
new_file_hash = calculate_hash_for_file(new_path)
print(('Verification %s' %
('OK' if file_hash == new_file_hash else 'FAIL')))
verify('/tmp/blocks')
block_hashes = [
'b499ca988473a4abc0461717cc8d4ce8753aa6d9', # exists
'b497ca988473a4abc0461717cc8d4ce8753aa6d9',
'b497ca988473a4abc0461717cc8d4ce8753aa6d9',
