def test_create_project_handles_503(requests_mock):
requests_mock.post('http://testing-es-url/api/v1/projects',
text='',
status_code=503)
with pytest.raises(ServiceError):
rest_client.project_create('http://testing-es-url', {'id': 'schema'},
'restype', 'myname', 'mynote')
def test_create_project_default_data(requests_mock):
requests_mock.post('http://testing-es-url/api/v1/projects', json={'status': 'ok'}, status_code=201)
rest_client.project_create('http://testing-es-url', {'id': 'schema'}, 'restype', 'myname')
posted_data = requests_mock.last_request.json()
assert all(expected_field in posted_data for expected_field in {'schema', 'result_type', 'number_parties', 'name', 'notes'})
assert 'created by clkhash' in posted_data['notes']
assert posted_data['number_parties'] == 2
def test_create_project_handles_400(requests_mock):
requests_mock.post('http://testing-es-url/api/v1/projects',
json={
'title': 'not ok',
'status': 400
},
status_code=400)
with pytest.raises(ServiceError):
rest_client.project_create('http://testing-es-url', {'id': 'schema'},
'restype', 'myname', 'mynote')
def create_project(type, schema, server, name, output, verbose):
"""Create a new project on an entity matching server.
See entity matching service documentation for details on mapping type and schema
Returns authentication details for the created project.
"""
if verbose:
log("Entity Matching Server: {}".format(server))
if schema is not None:
schema_json = json.load(schema)
# Validate the schema
clkhash.schema.validate_schema_dict(schema_json)
else:
raise ValueError(
"Schema must be provided when creating new linkage project")
name = name if name is not None else ''
# Creating new project
try:
project_creation_reply = project_create(server, schema_json, type,
name)
except ServiceError as e:
log("Unexpected response - {}".format(e.status_code))
log(e.text)
raise SystemExit
else:
log("Project created")
json.dump(project_creation_reply, output)
def run_experiments(config):
server = config['server']
rest_client.server_get_status(server)
results = {'experiments': []}
for experiment in config['experiments']:
try:
threshold = experiment['threshold']
logger.info('running experiment: {}'.format(experiment))
size_a, size_b = get_exp_sizes(experiment)
# create project
credentials = rest_client.project_create(
server, config['schema'], 'mapping',
"benchy_{}".format(experiment))
try:
# upload clks
upload_binary_clks(config, size_a, size_b, credentials)
# create run
run = rest_client.run_create(
server, credentials['project_id'],
credentials['result_token'], threshold,
"{}_{}".format(experiment, threshold))
# wait for result
run_id = run['run_id']
logger.info(f'waiting for run {run_id} to finish')
status = rest_client.wait_for_run(server,
credentials['project_id'],
run['run_id'],
credentials['result_token'],
timeout=config['timeout'])
if status['state'] != 'completed':
raise RuntimeError(
'run did not finish!\n{}'.format(status))
logger.info('experiment successful. Evaluating results now...')
mapping = rest_client.run_get_result_text(
server, credentials['project_id'], run['run_id'],
credentials['result_token'])
mapping = json.loads(mapping)['mapping']
mapping = {int(k): int(v) for k, v in mapping.items()}
tt = score_mapping(mapping,
*load_truth(config, size_a, size_b))
result = compose_result(status, tt, experiment,
(size_a, size_b), threshold)
results['experiments'].append(result)
logger.info('cleaning up...')
delete_resources(config, credentials, run)
except Exception as e:
delete_resources(config, credentials, run)
raise e
except Exception as e:
e_trace = format_exc()
logger.warning("experiment '{}' failed: {}".format(
experiment, e_trace))
results['experiments'].append({
'name': experiment,
'status': 'ERROR',
'description': e_trace
})
return results
def _create_project(self,
schema=None,
result_type='permutations',
name='',
notes='',
parties=2):
if schema is None:
schema = json.load(open(SIMPLE_SCHEMA_PATH, 'rt'))
return rest_client.project_create(self.url, schema, result_type, name,
notes, parties)
def _create_project(self,
schema=None,
result_type='permutations',
name='',
notes='',
parties=2):
if schema is None:
schema = json.load(open(SIMPLE_SCHEMA_PATH, 'rt'))
try:
response = rest_client.project_create(self.url, schema,
result_type, name, notes,
parties)
self._created_projects.append(response)
return response
except ServiceError:
raise
