def test_by_name():
s = fake.session()
instance = fake.instance()
project = tc.project.by_name(s, instance, "proj")
assert project.name == "proj"
assert project.description == "Mastering Project"
def test_by_name():
s = fake.session()
instance = fake.instance()
dataset = tc.dataset.by_name(s, instance, "dataset 1 name")
assert dataset.name == "dataset 1 name"
assert dataset.description == "dataset 1 description"
assert dataset.key_attribute_names == ("tamr_id",)
def test_from_resource_id():
s = fake.session()
instance = fake.instance()
dataset = tc.dataset.from_resource_id(s, instance, "1")
assert dataset.name == "dataset 1 name"
assert dataset.description == "dataset 1 description"
assert dataset.key_attribute_names == ("tamr_id", )
def test_by_resource_id():
s = fake.session()
instance = fake.instance()
resource_id = "2020-08-17_21-32-10-961"
tc.backup.by_resource_id(session=s,
instance=instance,
resource_id=resource_id)
def test_by_resource_id_categorization():
s = fake.session()
instance = fake.instance()
project = tc.project.by_resource_id(s, instance, "2")
assert isinstance(project, tc.CategorizationProject)
assert project.name == "Party Categorization"
assert project.description == "Categorizes organization at the Party/Domestic level"
def test_create_no_primary_key():
s = fake.session()
instance = fake.instance()
df = pd.DataFrame(_records_with_keys_json_2)
with pytest.raises(tc.primary_key.NotFound):
tc.dataframe.create(s, instance, df, name="df_dataset")
def test_by_resource_id_mastering():
s = fake.session()
instance = fake.instance()
project = tc.project.by_resource_id(s, instance, "1")
assert isinstance(project, tc.MasteringProject)
assert project.name == "proj"
assert project.description == "Mastering Project"
def test_manual_labels():
s = fake.session()
instance = fake.instance()
project = fake.categorization_project()
tc.categorization.project.manual_labels(session=s,
instance=instance,
project=project)
def test_from_project_dataset_not_found():
s = fake.session()
instance = fake.instance()
project = fake.mastering_project()
url = tc.URL(path="projects/1/unifiedDataset")
responses.add(responses.GET, str(url), status=404)
with pytest.raises(tc.dataset.unified.NotFound):
tc.dataset.unified.from_project(s, instance, project)
def test_create_handle_record_failure():
s = fake.session()
instance = fake.instance()
df = pd.DataFrame(_records_with_keys_json_2)
with pytest.raises(tc.dataframe.CreationFailure):
tc.dataframe.create(s,
instance,
df,
name="df_dataset",
primary_key_name="primary_key")
def test_create_dataset_already_exists():
s = fake.session()
instance = fake.instance()
with pytest.raises(tc.dataset.AlreadyExists):
tc.dataset.create(
s,
instance,
name="new dataset",
key_attribute_names=("primary_key",),
description="a new dataset",
)
def test_get_all_filter():
s = fake.session()
instance = fake.instance()
all_projects = tc.project.get_all(
s, instance, filter="description==Categorization Project")
assert len(all_projects) == 1
project = all_projects[0]
assert isinstance(project, tc.CategorizationProject)
assert project.name == "project 2"
assert project.description == "Categorization Project"
def test_create_project_already_exists():
s = fake.session()
instance = fake.instance()
with pytest.raises(tc.project.AlreadyExists):
tc.project._create(
s,
instance,
name="New Mastering Project",
project_type="DEDUP",
description="A Mastering Project",
)
def test_from_project():
s = fake.session()
instance = fake.instance()
project = fake.mastering_project()
dataset_json = utils.load_json("dataset.json")
url = tc.URL(path="projects/1/unifiedDataset")
responses.add(responses.GET, str(url), json=dataset_json)
unified_dataset = tc.dataset.unified.from_project(s, instance, project)
assert unified_dataset.name == "dataset 1 name"
assert unified_dataset.description == "dataset 1 description"
assert unified_dataset.key_attribute_names == ("tamr_id", )
def test_create():
s = fake.session()
instance = fake.instance()
df = pd.DataFrame(_records_with_keys_json_2)
dataset = tc.dataframe.create(s,
instance,
df,
name="df_dataset",
primary_key_name="primary_key")
assert dataset.name == "df_dataset"
assert dataset.key_attribute_names == ("primary_key", )
def test_create_infer_primary_key_from_index():
s = fake.session()
instance = fake.instance()
df = pd.DataFrame(
_records_json_2,
index=[record["primary_key"] for record in _records_with_keys_json_2],
)
df.index.name = "primary_key"
dataset = tc.dataframe.create(s, instance, df, name="df_dataset")
assert dataset.name == "df_dataset"
assert dataset.key_attribute_names == ("primary_key", )
def test_create():
s = fake.session()
instance = fake.instance()
project = tc.schema_mapping.project.create(
s,
instance,
name="New Schema Mapping Project",
description="A Schema Mapping Project",
)
assert isinstance(project, tc.SchemaMappingProject)
assert project.name == "New Schema Mapping Project"
assert project.description == "A Schema Mapping Project"
def test_create():
s = fake.session()
instance = fake.instance()
project = tc.categorization.project.create(
s,
instance,
name="New Categorization Project",
description="A Categorization Project",
)
assert isinstance(project, tc.CategorizationProject)
assert project.name == "New Categorization Project"
assert project.description == "A Categorization Project"
def test_get_all_filter():
s = fake.session()
instance = fake.instance()
all_datasets = tc.dataset.get_all(
s, instance, filter="description==dataset 2 description"
)
assert len(all_datasets) == 1
dataset = all_datasets[0]
assert dataset.name == "dataset 2 name"
assert dataset.description == "dataset 2 description"
assert dataset.key_attribute_names == ("tamr_id",)
def test_create():
s = fake.session()
instance = fake.instance()
dataset = tc.dataset.create(
s,
instance,
name="new dataset",
key_attribute_names=("primary_key",),
description="a new dataset",
)
assert dataset.name == "new dataset"
assert dataset.description == "a new dataset"
assert dataset.key_attribute_names == ("primary_key",)
def test_by_resource_id():
s = fake.session()
instance = fake.instance()
url = tc.URL(path="operations/1")
operation_json = utils.load_json("operation_succeeded.json")
responses.add(responses.GET, str(url), json=operation_json)
resource_id = "1"
op = tc.operation.by_resource_id(s, instance, resource_id)
assert op.url == url
assert op.type == operation_json["type"]
assert op.description == operation_json["description"]
assert op.status == operation_json["status"]
assert tc.operation.succeeded(op)
def test_operation_from_response():
s = fake.session()
instance = fake.instance()
url = tc.URL(path="operations/1")
operation_json = utils.load_json("operation_succeeded.json")
responses.add(responses.GET, str(url), json=operation_json)
r = s.get(str(url))
op = tc.operation._from_response(instance, r)
assert op.url == url
assert op.type == operation_json["type"]
assert op.description == operation_json["description"]
assert op.status == operation_json["status"]
assert tc.operation.succeeded(op)
def test_get_all():
s = fake.session()
instance = fake.instance()
all_projects = tc.project.get_all(s, instance)
assert len(all_projects) == 2
project_1 = all_projects[0]
assert isinstance(project_1, tc.MasteringProject)
assert project_1.name == "project 1"
assert project_1.description == "Mastering Project"
project_2 = all_projects[1]
assert isinstance(project_2, tc.CategorizationProject)
assert project_2.name == "project 2"
assert project_2.description == "Categorization Project"
def test_get_all():
s = fake.session()
instance = fake.instance()
all_datasets = tc.dataset.get_all(s, instance)
assert len(all_datasets) == 2
dataset_1 = all_datasets[0]
assert dataset_1.name == "dataset 1 name"
assert dataset_1.description == "dataset 1 description"
assert dataset_1.key_attribute_names == ("tamr_id",)
dataset_2 = all_datasets[1]
assert dataset_2.name == "dataset 2 name"
assert dataset_2.description == "dataset 2 description"
assert dataset_2.key_attribute_names == ("tamr_id",)
def test_from_json_unrecognized_project_type():
instance = fake.instance()
url = tc.URL("project/1", instance)
data: tc._types.JsonDict = {
"id": "unify://unified-data/v1/projects/1",
"name": "project 1",
"description": "A project of unknown type",
"type": "UNKNOWN",
"unifiedDatasetName": "",
"relativeId": "projects/1",
"externalId": "58bdbe72-3c08-427d-97bd-45b16d92c79c",
}
project = tc.project._from_json(url, data)
assert isinstance(project, tc.UnknownProject)
assert project.name == "project 1"
assert project.description == "A project of unknown type"
def test_operation_from_response_noop():
s = fake.session()
instance = fake.instance()
url = tc.URL(path="operations/2")
responses.add(responses.GET, str(url), status=204)
url_dummy = tc.URL(path="operations/-1")
responses.add(responses.GET, str(url_dummy), status=404)
r = s.get(str(url))
op2 = tc.operation._from_response(instance, r)
assert op2.url is not None
assert op2.type == "NOOP"
assert op2.description is not None
assert op2.status is not None
assert op2.status["state"] == "SUCCEEDED"
assert tc.operation.succeeded(op2)
op2w = tc.operation.wait(s, op2)
assert tc.operation.succeeded(op2w)
with pytest.raises(tc.operation.NotFound):
tc.operation.poll(s, op2w)
def test_from_json_unrecognized_project_type():
instance = fake.instance()
url = tc.URL("project/1", instance)
data: tc._types.JsonDict = {"type": "NOT_A_PROJECT_TYPE"}
with pytest.raises(ValueError):
tc.project._from_json(url, data)
def test_by_name_project_ambiguous():
s = fake.session()
instance = fake.instance()
with pytest.raises(tc.project.Ambiguous):
tc.project.by_name(s, instance, "ambiguous project")
def test_by_name_project_not_found():
s = fake.session()
instance = fake.instance()
with pytest.raises(tc.project.NotFound):
tc.project.by_name(s, instance, "missing project")
def test_by_resource_id_not_found():
s = fake.session()
instance = fake.instance()
with pytest.raises(tc.project.NotFound):
tc.project.by_resource_id(s, instance, "1")