def test_delete_remove_leaked_stored_objects_and_uploaded_files(self):
workflow = Workflow.create_and_init()
# If the user deletes a workflow, all data associated with that
# workflow should disappear. Postgres handles DB objects; but Django's
# ORM doesn't do a great job with StoredObjects and UploadedFiles.
#
# This test isn't about minutae. It's just: if the user deletes a
# Workflow, make sure all data gets deleted.
#
# TODO fix all other bugs that leak data.
wf_module = workflow.tabs.first().wf_modules.create(
order=0, slug="step-1", module_id_name="x"
)
# Add StoredObject ... and leak it
wf_module.store_fetched_table(pd.DataFrame({"A": [1, 2]}))
stored_object_key = wf_module.stored_objects.first().key
wf_module.stored_objects.all()._raw_delete("default") # skip S3-delete
# Add UploadedFile, missing a DB entry. (Even if we fix all bugs that
# leak an S3 object after deleting a DB entry [and 2019-06-03 there are
# still more] we'll still need to handle missing DB entries from legacy
# code.)
uploaded_file_key = f"{wf_module.uploaded_file_prefix}{uuid.uuid4()}.csv"
minio.put_bytes(minio.UserFilesBucket, uploaded_file_key, b"A\nb")
workflow.delete()
self.assertFalse(minio.exists(minio.StoredObjectsBucket, stored_object_key))
self.assertFalse(minio.exists(minio.UserFilesBucket, uploaded_file_key))
def test_abort_upload(self):
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'
wf_module = workflow.tabs.first().wf_modules.create(
order=0,
module_id_name='x',
inprogress_file_upload_id=None,
inprogress_file_upload_key=key,
inprogress_file_upload_last_accessed_at=timezone.now(),
)
# let's pretend the user has uploaded at least partial data.
minio.put_bytes(
minio.UserFilesBucket,
key,
b'1234567',
ContentDisposition="attachment; filename*=UTF-8''file.csv",
)
response = self.run_handler(abort_upload, user=user,
workflow=workflow, wfModuleId=wf_module.id,
key=key)
self.assertResponse(response, data=None)
wf_module.refresh_from_db()
self.assertIsNone(wf_module.inprogress_file_upload_id)
self.assertIsNone(wf_module.inprogress_file_upload_key)
self.assertIsNone(wf_module.inprogress_file_upload_last_accessed_at)
# Ensure the file is deleted from S3
self.assertFalse(minio.exists(minio.UserFilesBucket, key))
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_delete_wfmodule(self):
result = ProcessResult(pandas.DataFrame({"a": [1]}))
self.wf_module.cache_render_result(self.delta.id, result)
parquet_key = self.wf_module.cached_render_result.parquet_key
self.wf_module.delete()
self.assertFalse(minio.exists(minio.CachedRenderResultsBucket, parquet_key))
def test_delete_deletes_from_s3(self):
minio.put_bytes(minio.StoredObjectsBucket, "test.dat", b"abcd")
workflow = Workflow.create_and_init()
wf_module = workflow.tabs.first().wf_modules.create(order=0, slug="step-1")
so = wf_module.stored_objects.create(
size=4, bucket=minio.StoredObjectsBucket, key="test.dat", hash="123"
)
so.delete()
self.assertFalse(minio.exists(minio.StoredObjectsBucket, "test.dat"))
def test_delete_wf_module_deletes_from_s3(self):
minio.put_bytes(minio.StoredObjectsBucket, 'test.dat', b'abcd')
workflow = Workflow.objects.create()
tab = workflow.tabs.create(position=0)
wf_module = tab.wf_modules.create(order=0)
wf_module.stored_objects.create(size=4, bucket=minio.StoredObjectsBucket,
key='test.dat', hash='123')
wf_module.delete()
self.assertFalse(minio.exists(minio.StoredObjectsBucket, 'test.dat'))
def test_delete_s3_data_leaked_file(self):
# Delete a file with our UUID but without an UploadedFile.
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()
key = wf_module.uploaded_file_prefix + str(ipu.id) + ".xlsx"
minio.put_bytes(minio.UserFilesBucket, key, b"1234567")
ipu.delete_s3_data()
self.assertFalse(minio.exists(minio.UserFilesBucket, key))
def test_set_to_empty(self):
result = ProcessResult(pandas.DataFrame({"a": [1]}))
self.wf_module.cache_render_result(self.delta.id, result)
parquet_key = self.wf_module.cached_render_result.parquet_key
db_wf_module = WfModule.objects.get(id=self.wf_module.id)
db_wf_module.clear_cached_render_result()
self.assertIsNone(db_wf_module.cached_render_result)
db_wf_module.refresh_from_db()
self.assertIsNone(db_wf_module.cached_render_result)
self.assertFalse(minio.exists(minio.CachedRenderResultsBucket, parquet_key))
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_abort_upload_happy_path_after_complete(self):
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")
response = self.run_handler(
abort_upload, user=user, workflow=workflow, wfModuleId=wf_module.id, key=key
)
self.assertResponse(response, data=None)
wf_module.refresh_from_db()
self.assertFalse(minio.exists(in_progress_upload.Bucket, key))
def test_delete_s3_data_ignore_non_leaked_file(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()
key = wf_module.uploaded_file_prefix + str(ipu.id) + ".xlsx"
minio.put_bytes(minio.UserFilesBucket, key, b"1234567")
wf_module.uploaded_files.create(
name="text.xlsx",
size=7,
uuid=str(self.id),
bucket=minio.UserFilesBucket,
key=key,
)
ipu.delete_s3_data()
self.assertFalse(minio.exists(minio.UserFilesBucket, key))
def test_abort(self):
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()
key = upload.get_upload_key()
minio.put_bytes(upload.Bucket, key, b"1234567")
response = self.client.delete(
f"/api/v1/workflows/{workflow.id}/steps/step-123/uploads/{upload.id}",
HTTP_AUTHORIZATION="Bearer abc123",
)
self.assertEqual(response.status_code, 200)
self.assertEqual(json.loads(response.content), {})
self.assertFalse(minio.exists(upload.Bucket, key)) # file was deleted
upload.refresh_from_db()
self.assertTrue(upload.is_completed)
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 move_uploaded_file(workflow, wf_module, uploaded_file):
"""
Move files from /uuid.ext to /wf-1/wfm-2/uuid.ext.
This helps delete leaked files and find problem files.
"""
from server import minio
bucket = uploaded_file.bucket
old_key = uploaded_file.key
if '/' in old_key:
return
new_key = f'wf-{workflow.id}/wfm-{wf_module.id}/{old_key}'
logger.info(f'Move %s/%s to %s/%s', bucket, old_key, bucket, new_key)
try:
minio.copy(bucket, new_key, f'{bucket}/{old_key}')
minio.remove(bucket, old_key)
except minio.error.NoSuchKey:
# old_key is missing. Two possibilities:
#
# 1. We're re-running this script after it failed once with
# atomic=True (which used to be set, by accident); the move already
# succeeded but the DB doesn't know it. In that case, continue
# because this error actually means, "all is well."
# 2. The file didn't exist to begin with. In that case, write a blank
# file in its stead. That way the user will remark, "hey, Workbench
# ate my file!" instead of undefined behavior (which is worse).
# https://www.pivotaltracker.com/story/show/163336822
if minio.exists(bucket, new_key):
pass # "all is well"
else:
# write an empty file
minio.put_bytes(bucket, new_key, b'')
uploaded_file.size = 0
uploaded_file.save(update_fields=['size'])
uploaded_file.key = new_key
uploaded_file.save(update_fields=['key'])
