def create_project_response(requests, size, overlap, result_type, encoding_size=128, uses_blocking=False):
"""
Create a project with the given size, overlap and result_type.
Tests that use one of these projects will get a dict like the following:
{
"project_id": "ID",
"upload-mode": "BINARY" | "JSON",
"size": [size 1, size 2],
"encoding-size": int number of bytes in each encoding e.g. 128,
"overlap": float between 0 and 1,
"result_token": "TOKEN",
"upload_tokens": [TOKENS, ...],
"dp_1": <JSON RESPONSE TO DATA UPLOAD>
"dp_2": <JSON RESPONSE TO DATA UPLOAD>
}
"""
project, dp_responses = create_project_upload_fake_data(
requests, size, overlap=overlap, result_type=result_type, encoding_size=encoding_size)
project.update({
'size': size,
'encoding-size': encoding_size,
'upload-mode': 'JSON',
'uses_blocking': uses_blocking,
'overlap': overlap,
'dp_responses': dp_responses
})
return project
def _create_data_linkage_run(requests, result_type_number_parties):
result_type, number_parties = result_type_number_parties
project, _ = create_project_upload_fake_data(requests,
[100] * number_parties,
overlap=0.5,
result_type=result_type)
run_id = post_run(requests, project, 1.0)
return project, run_id
def test_project_json_data_upload_with_too_small_encoded_size(
requests, result_type_number_parties):
result_type, number_parties = result_type_number_parties
new_project_data, _ = create_project_upload_fake_data(
requests,
[500] * number_parties,
overlap=0.8,
result_type=result_type,
encoding_size=4
)
with pytest.raises(AssertionError):
run_id = post_run(requests, new_project_data, 0.9)
get_run_result(requests, new_project_data, run_id, wait=True)
def test_project_json_data_upload_with_invalid_encoded_size(
requests, result_type_number_parties):
result_type, number_parties = result_type_number_parties
new_project_data, _ = create_project_upload_fake_data(
requests,
[500] * number_parties,
overlap=0.8,
result_type=result_type,
encoding_size=20 # not multiple of 8
)
with pytest.raises(AssertionError):
run_id = post_run(requests, new_project_data, 0.9)
get_run_result(requests,
new_project_data,
run_id,
wait=True,
timeout=240)
def test_project_json_data_upload_with_various_encoded_sizes(
requests,
encoding_size, result_type_number_parties):
result_type, number_parties = result_type_number_parties
new_project_data, _ = create_project_upload_fake_data(
requests,
[500] * number_parties,
overlap=0.8,
result_type=result_type,
encoding_size=encoding_size
)
run_id = post_run(requests, new_project_data, 0.9)
result = get_run_result(requests, new_project_data, run_id, wait=True)
if result_type == 'groups':
assert 'groups' in result
# This is a pretty bad bound, but we're not testing the
# accuracy.
assert len(result['groups']) >= 400
def test_project_json_data_upload_with_too_large_encoded_size(
requests, result_type_number_parties):
result_type, number_parties = result_type_number_parties
new_project_data, _ = create_project_upload_fake_data(
requests, [500] * number_parties,
overlap=0.8,
result_type=result_type,
encoding_size=4096)
# Just initializing it before the loop.
project_description = {'error': False}
rep = 0
max_rep = 10
while not project_description['error'] and rep < max_rep:
rep += 1
time.sleep(2)
project_description = requests.get(
url + '/projects/{}'.format(new_project_data['project_id']),
headers={
'Authorization': new_project_data['result_token']
}).json()
assert 'error' in project_description
assert project_description['error']