def test_convert_to_uploaded_file_happy_path(self):
workflow = Workflow.create_and_init()
wf_module = workflow.tabs.first().wf_modules.create(order=0,
slug="step-1",
module_id_name="x")
ipu = wf_module.in_progress_uploads.create()
minio.put_bytes(ipu.Bucket, ipu.get_upload_key(), b"1234567")
uploaded_file = ipu.convert_to_uploaded_file("test sheet.xlsx")
self.assertEqual(uploaded_file.uuid, str(ipu.id))
final_key = wf_module.uploaded_file_prefix + str(ipu.id) + ".xlsx"
# New file on S3 has the right bytes and metadata
self.assertEqual(
minio.get_object_with_data(minio.UserFilesBucket,
final_key)["Body"],
b"1234567",
)
self.assertEqual(
minio.client.head_object(Bucket=minio.UserFilesBucket,
Key=final_key)["ContentDisposition"],
"attachment; filename*=UTF-8''test%20sheet.xlsx",
)
# InProgressUpload is completed
self.assertEqual(ipu.is_completed, True)
ipu.refresh_from_db()
self.assertEqual(ipu.is_completed, True) # also on DB
# Uploaded file is deleted
self.assertFalse(
minio.exists(minio.UserFilesBucket, ipu.get_upload_key()))
def test_upload_empty_file(self):
md5sum = _base64_md5sum(b'')
url, headers = minio.presign_upload(Bucket, 'key', 't.csv', 0, md5sum)
http = urllib3.PoolManager()
response = http.request('PUT', url, body=b'', headers=headers)
self.assertEqual(response.status, 200)
self.assertEqual(
minio.get_object_with_data(Bucket, 'key')['Body'], b'')
def test_multipart_upload_by_presigned_requests(self, send_delta):
"""Test presign_upload_part _and_ complete_multipart_upload"""
# Integration-test: use `urllib3` to run presigned responses.
# See `test_minio` for canonical usage.
user = User.objects.create(username='******', email='*****@*****.**')
workflow = Workflow.create_and_init(owner=user)
uuid = str(uuidgen.uuid4())
key = f'wf-123/wfm-234/{uuid}.csv'
upload_id = minio.create_multipart_upload(minio.UserFilesBucket,
key, 'file.csv')
wf_module = workflow.tabs.first().wf_modules.create(
order=0,
module_id_name='x',
inprogress_file_upload_id=upload_id,
inprogress_file_upload_key=key,
inprogress_file_upload_last_accessed_at=timezone.now()
)
data = b'1234567' * 1024 * 1024 # 7MB => 5MB+2MB parts
data1 = data[:5*1024*1024]
data2 = data[5*1024*1024:]
md5sum1 = _base64_md5sum(data1)
md5sum2 = _base64_md5sum(data2)
response1 = self.run_handler(presign_upload_part, user=user,
workflow=workflow,
wfModuleId=wf_module.id, uploadId=upload_id,
partNumber=1, nBytes=len(data1),
base64Md5sum=md5sum1)
self.assertEqual(response1.error, '')
response2 = self.run_handler(presign_upload_part, user=user,
workflow=workflow,
wfModuleId=wf_module.id,
uploadId=upload_id, partNumber=2,
nBytes=len(data2), base64Md5sum=md5sum2)
self.assertEqual(response2.error, '')
http = urllib3.PoolManager()
s3response1 = http.request('PUT', response1.data['url'], body=data1,
headers=response1.data['headers'])
self.assertEqual(s3response1.status, 200)
s3response2 = http.request('PUT', response2.data['url'], body=data2,
headers=response2.data['headers'])
self.assertEqual(s3response2.status, 200)
etag1 = s3response1.headers['ETag'][1:-1] # un-wrap quotes
etag2 = s3response2.headers['ETag'][1:-1] # un-wrap quotes
send_delta.side_effect = async_noop
response3 = self.run_handler(complete_multipart_upload, user=user,
workflow=workflow,
wfModuleId=wf_module.id,
uploadId=upload_id, etags=[etag1, etag2])
self.assertResponse(response3, data={'uuid': uuid})
self.assertEqual(
minio.get_object_with_data(minio.UserFilesBucket, key)['Body'],
data
)
def _external_module_get_html_bytes(id_name: str,
version: str) -> Optional[bytes]:
prefix = '%s/%s/' % (id_name, version)
all_keys = minio.list_file_keys(minio.ExternalModulesBucket, prefix)
try:
html_key = next(k for k in all_keys if k.endswith('.html'))
except StopIteration:
return None # there is no HTML file
return minio.get_object_with_data(minio.ExternalModulesBucket,
html_key)['Body']
def test_assume_role_to_write_multipart(self):
client = self._assume_role_session_client_with_write_access(Bucket, "key")
from boto3.s3.transfer import TransferConfig
data = b"1234567" * 1024 * 1024 # 7MB => 5MB+2MB parts
client.upload_fileobj(
io.BytesIO(data),
Bucket,
"key",
Config=TransferConfig(multipart_threshold=5 * 1024 * 1024),
)
self.assertEqual(minio.get_object_with_data(Bucket, "key")["Body"], data)
def test_upload_by_presigned_request(self):
data = b'1234567'
md5sum = _base64_md5sum(data)
url, headers = minio.presign_upload(Bucket, 'key', 'file.csv',
len(data), md5sum)
http = urllib3.PoolManager()
response = http.request('PUT', url, body=data, headers=headers)
self.assertEqual(response.status, 200)
self.assertEqual(
minio.get_object_with_data(Bucket, 'key')['Body'], data)
head = minio.client.head_object(Bucket=Bucket, Key='key')
self.assertEqual(head['ContentDisposition'],
"attachment; filename*=UTF-8''file.csv")
def test_finish_upload_happy_path(self, send_delta):
user = User.objects.create(username="******", email="*****@*****.**")
workflow = Workflow.create_and_init(owner=user)
wf_module = workflow.tabs.first().wf_modules.create(
order=0, slug="step-1", module_id_name="x"
)
in_progress_upload = wf_module.in_progress_uploads.create(
id="147a9f5d-5b3e-41c3-a968-a84a5a9d587f"
)
key = in_progress_upload.get_upload_key()
minio.put_bytes(in_progress_upload.Bucket, key, b"1234567")
send_delta.side_effect = async_noop
response = self.run_handler(
finish_upload,
user=user,
workflow=workflow,
wfModuleId=wf_module.id,
key=key,
filename="test sheet.csv",
)
self.assertResponse(
response, data={"uuid": "147a9f5d-5b3e-41c3-a968-a84a5a9d587f"}
)
# The uploaded file is deleted
self.assertFalse(minio.exists(in_progress_upload.Bucket, key))
# A new upload is created
uploaded_file = wf_module.uploaded_files.first()
self.assertEqual(uploaded_file.name, "test sheet.csv")
self.assertEqual(uploaded_file.size, 7)
self.assertEqual(uploaded_file.uuid, "147a9f5d-5b3e-41c3-a968-a84a5a9d587f")
self.assertEqual(uploaded_file.bucket, in_progress_upload.Bucket)
final_key = f"wf-{workflow.id}/wfm-{wf_module.id}/147a9f5d-5b3e-41c3-a968-a84a5a9d587f.csv"
self.assertEqual(uploaded_file.key, final_key)
# The file has the right bytes and metadata
self.assertEqual(
minio.get_object_with_data(minio.UserFilesBucket, final_key)["Body"],
b"1234567",
)
self.assertEqual(
minio.client.head_object(Bucket=minio.UserFilesBucket, Key=final_key)[
"ContentDisposition"
],
"attachment; filename*=UTF-8''test%20sheet.csv",
)
# wf_module is updated
send_delta.assert_called()
def test_wf_module_duplicate_copy_uploaded_file(self):
workflow = Workflow.create_and_init()
tab = workflow.tabs.first()
wf_module = tab.wf_modules.create(
order=0,
module_id_name='upload',
)
uuid = str(uuidgen.uuid4())
key = f'{wf_module.uploaded_file_prefix}{uuid}.csv'
minio.put_bytes(minio.UserFilesBucket, key, b'1234567')
# Write the uuid to the old module -- we'll check the new module points
# to a valid file
wf_module.params = {'file': uuid, 'has_header': True}
wf_module.save(update_fields=['params'])
uploaded_file = wf_module.uploaded_files.create(
name='t.csv',
uuid=uuid,
bucket=minio.UserFilesBucket,
key=key,
size=7,
)
workflow2 = Workflow.create_and_init()
tab2 = workflow2.tabs.first()
wf_module2 = wf_module.duplicate(tab2)
uploaded_file2 = wf_module2.uploaded_files.first()
self.assertIsNotNone(uploaded_file2)
# New file gets same uuid -- because it's the same file and we don't
# want to edit params during copy
self.assertEqual(uploaded_file2.uuid, uuid)
self.assertEqual(wf_module2.params['file'], uuid)
self.assertTrue(
# The new file should be in a different path
uploaded_file2.key.startswith(wf_module2.uploaded_file_prefix)
)
self.assertEqual(uploaded_file2.name, 't.csv')
self.assertEqual(uploaded_file2.size, 7)
self.assertEqual(uploaded_file2.created_at, uploaded_file.created_at)
self.assertEqual(
minio.get_object_with_data(uploaded_file2.bucket,
uploaded_file2.key)['Body'],
b'1234567'
)
def test_complete_happy_path(self, queue_render, send_delta):
send_delta.return_value = async_noop()
queue_render.return_value = async_noop()
_init_module("x")
workflow = Workflow.create_and_init()
wf_module = workflow.tabs.first().wf_modules.create(
order=0,
slug="step-123",
module_id_name="x",
file_upload_api_token="abc123")
upload = wf_module.in_progress_uploads.create()
uuid = str(upload.id)
key = upload.get_upload_key()
minio.put_bytes(upload.Bucket, key, b"1234567")
response = self.client.post(
f"/api/v1/workflows/{workflow.id}/steps/step-123/uploads/{upload.id}",
{"filename": "test.csv"},
content_type="application/json",
HTTP_AUTHORIZATION="Bearer abc123",
)
self.assertEqual(response.status_code, 200)
# Upload and its S3 data were deleted
self.assertFalse(minio.exists(upload.Bucket, key))
upload.refresh_from_db()
self.assertTrue(upload.is_completed)
# Final upload was created
uploaded_file = wf_module.uploaded_files.first()
self.assertEqual(uploaded_file.key,
f"wf-{workflow.id}/wfm-{wf_module.id}/{uuid}.csv")
self.assertEqual(
minio.get_object_with_data(minio.UserFilesBucket,
uploaded_file.key)["Body"],
b"1234567",
)
self.assertEqual(uploaded_file.name, "test.csv")
# Return value includes uuid
data = json.loads(response.content)
self.assertEqual(data["uuid"], uuid)
self.assertEqual(data["name"], "test.csv")
self.assertEqual(data["size"], 7)
# Send deltas
send_delta.assert_called()
queue_render.assert_called()
def test_wf_module_duplicate_copy_uploaded_file(self):
workflow = Workflow.create_and_init()
tab = workflow.tabs.first()
wf_module = tab.wf_modules.create(
order=0, slug="step-1", module_id_name="upload"
)
uuid = str(uuidgen.uuid4())
key = f"{wf_module.uploaded_file_prefix}{uuid}.csv"
minio.put_bytes(minio.UserFilesBucket, key, b"1234567")
# Write the uuid to the old module -- we'll check the new module points
# to a valid file
wf_module.params = {"file": uuid, "has_header": True}
wf_module.save(update_fields=["params"])
uploaded_file = wf_module.uploaded_files.create(
name="t.csv", uuid=uuid, bucket=minio.UserFilesBucket, key=key, size=7
)
workflow2 = Workflow.create_and_init()
tab2 = workflow2.tabs.first()
wf_module2 = wf_module.duplicate_into_new_workflow(tab2)
uploaded_file2 = wf_module2.uploaded_files.first()
self.assertIsNotNone(uploaded_file2)
# New file gets same uuid -- because it's the same file and we don't
# want to edit params during copy
self.assertEqual(uploaded_file2.uuid, uuid)
self.assertEqual(wf_module2.params["file"], uuid)
self.assertTrue(
# The new file should be in a different path
uploaded_file2.key.startswith(wf_module2.uploaded_file_prefix)
)
self.assertEqual(uploaded_file2.name, "t.csv")
self.assertEqual(uploaded_file2.size, 7)
self.assertEqual(uploaded_file2.created_at, uploaded_file.created_at)
self.assertEqual(
minio.get_object_with_data(uploaded_file2.bucket, uploaded_file2.key)[
"Body"
],
b"1234567",
)
def test_multipart_upload_by_presigned_requests(self):
upload_id = minio.create_multipart_upload(Bucket, 'key', 'file.csv')
data = b'1234567' * 1024 * 1024 # 7MB => 5MB+2MB parts
data1 = data[:5 * 1024 * 1024]
data2 = data[5 * 1024 * 1024:]
md5sum1 = _base64_md5sum(data1)
md5sum2 = _base64_md5sum(data2)
url1, headers1 = minio.presign_upload_part(Bucket, 'key', upload_id, 1,
len(data1), md5sum1)
url2, headers2 = minio.presign_upload_part(Bucket, 'key', upload_id, 2,
len(data2), md5sum2)
http = urllib3.PoolManager()
response1 = http.request('PUT', url1, body=data1, headers=headers1)
self.assertEqual(response1.status, 200)
etag1 = response1.headers['ETag'][1:-1] # un-wrap quotes
response2 = http.request('PUT', url2, body=data2, headers=headers2)
self.assertEqual(response2.status, 200)
etag2 = response2.headers['ETag'][1:-1] # un-wrap quotes
minio.complete_multipart_upload(Bucket, 'key', upload_id,
[etag1, etag2])
self.assertEqual(
minio.get_object_with_data(Bucket, 'key')['Body'], data)
def test_integration_happy_path(self):
workflow = Workflow.create_and_init()
wf_module = workflow.tabs.first().wf_modules.create(order=0,
slug="step-1",
module_id_name="x")
ipu = wf_module.in_progress_uploads.create()
updated_at1 = ipu.updated_at
time.sleep(0.000001) # so updated_at changes
params = ipu.generate_upload_parameters()
ipu.refresh_from_db() # ensure we wrote updated_at
updated_at2 = ipu.updated_at
self.assertGreater(updated_at2, updated_at1)
# Upload using a separate S3 client
# Import _after_ we've imported minio -- so server.minio's monkey-patch
# takes effect.
import boto3
credentials = params["credentials"]
session = boto3.session.Session(
aws_access_key_id=credentials["accessKeyId"],
aws_secret_access_key=credentials["secretAccessKey"],
aws_session_token=credentials["sessionToken"],
region_name=params["region"],
)
client = session.client("s3", endpoint_url=params["endpoint"])
client.put_object(Bucket=ipu.Bucket,
Key=ipu.get_upload_key(),
Body=b"1234567")
# Complete the upload
uploaded_file = ipu.convert_to_uploaded_file("test.csv")
self.assertEqual(
minio.get_object_with_data(uploaded_file.bucket,
uploaded_file.key)["Body"],
b"1234567",
)
def test_assume_role_to_write(self):
client = self._assume_role_session_client_with_write_access(Bucket, "key")
data = b"1234567"
client.upload_fileobj(io.BytesIO(data), Bucket, "key")
self.assertEqual(minio.get_object_with_data(Bucket, "key")["Body"], data)
