def test_help(capsys):
with pytest.raises(SystemExit) as error:
run_create_mining_cache(["--help"])
stdout, stderr = capsys.readouterr()
assert error.value.code == 0
assert stdout.startswith("usage:")
assert stderr == ""
def test_missing_sqlite_db():
with pytest.raises(FileNotFoundError, match="^No database found"):
run_create_mining_cache([
"--data-and-models-dir",
"/some/path",
"--db-type",
"sqlite",
"--db-url",
"fake$?#",
], )
def test_send_through(
monkeypatch,
tmpdir,
db_type,
db_url,
target_table_name,
n_processes_per_model,
restrict_to_models,
):
# Monkey-patching
data_dir = "/data_dir"
df_model_library = pd.DataFrame(
columns=["entity_type", "entity_type_name", "model_id", "model_path"],
data=[
[
"CELL_COMPARTMENT",
"CELLULAR_COMPONENT",
"path/to/model_1",
f"{data_dir}/path/to/model_1",
],
[
"CELL_TYPE",
"CELL_TYPE",
"path/to/model_2",
f"{data_dir}/path/to/model_2",
],
[
"CHEMICAL", "CHEBI", "path/to/model_3",
f"{data_dir}/path/to/model_3"
],
],
)
fake_load_ee_models_library = Mock()
fake_load_ee_models_library.return_value = df_model_library
fake_sqlalchemy = Mock()
fake_create_mining_cache = Mock()
monkeypatch.setattr("bluesearch.entrypoint.create_mining_cache.sqlalchemy",
fake_sqlalchemy)
monkeypatch.setattr(
"bluesearch.entrypoint.create_mining_cache.load_ee_models_library",
fake_load_ee_models_library,
)
monkeypatch.setattr("bluesearch.database.CreateMiningCache",
fake_create_mining_cache)
monkeypatch.setattr("bluesearch.database.CreateMiningCache",
fake_create_mining_cache)
monkeypatch.setattr(
"bluesearch.entrypoint.create_mining_cache.getpass.getpass",
lambda _: "fake_password",
)
# Create temporary sqlite database
if db_type == "sqlite":
db_url = pathlib.Path(tmpdir) / "my.db"
db_url.touch()
# Construct arguments
argv = [
"--data-and-models-dir=/some/fake/path",
f"--db-type={db_type}",
f"--db-url={db_url}",
f"--target-table-name={target_table_name}",
f"--n-processes-per-model={n_processes_per_model}",
f"--restrict-to-models={restrict_to_models}",
]
# Call entrypoint method
# import pdb; pdb.set_trace()
run_create_mining_cache(argv)
# Checks
# Check that CreateMiningCache(...) was called once and get its arguments
fake_create_mining_cache.assert_called_once()
args, kwargs = fake_create_mining_cache.call_args
# Construct the restricted model library data frame
selected_models = restrict_to_models.split(",")
df_model_library_selected = df_model_library[
df_model_library["model_id"].isin(selected_models).tolist()]
# Check the args/kwargs
assert kwargs["database_engine"] == fake_sqlalchemy.create_engine()
assert isinstance(kwargs["ee_models_library"], pd.DataFrame)
assert len(df_model_library_selected) > 0
assert kwargs["ee_models_library"].equals(df_model_library_selected)
assert kwargs["target_table_name"] == target_table_name
assert kwargs["workers_per_model"] == n_processes_per_model
# Check that CreateMiningCache.construct() was called
fake_create_mining_cache().construct.assert_called_once()
def test_send_through(
monkeypatch,
tmpdir,
db_type,
db_url,
target_table_name,
n_processes_per_model,
restrict_to_etypes,
entity_types,
spacy_model_path,
):
# Monkey-patching
fake_sqlalchemy = Mock()
fake_create_mining_cache = Mock()
monkeypatch.setattr("bluesearch.entrypoint.create_mining_cache.sqlalchemy",
fake_sqlalchemy)
monkeypatch.setattr("bluesearch.database.CreateMiningCache",
fake_create_mining_cache)
monkeypatch.setattr("bluesearch.database.CreateMiningCache",
fake_create_mining_cache)
monkeypatch.setattr(
"bluesearch.entrypoint.create_mining_cache.getpass.getpass",
lambda _: "fake_password",
)
# Create temporary sqlite database
if db_type == "sqlite":
db_url = pathlib.Path(tmpdir) / "my.db"
db_url.touch()
# Construct arguments
argv = [
f"--data-and-models-dir={spacy_model_path}",
f"--db-type={db_type}",
f"--db-url={db_url}",
f"--target-table-name={target_table_name}",
f"--n-processes-per-model={n_processes_per_model}",
f"--restrict-to-etypes={restrict_to_etypes}",
]
# Call entrypoint method
# import pdb; pdb.set_trace()
run_create_mining_cache(argv)
# Checks
# Check that CreateMiningCache(...) was called once and get its arguments
fake_create_mining_cache.assert_called_once()
args, kwargs = fake_create_mining_cache.call_args
# Construct the restricted etypes
available_models = set(restrict_to_etypes.split(",")) & set(entity_types)
# Check the args/kwargs
assert kwargs["database_engine"] == fake_sqlalchemy.create_engine()
assert isinstance(kwargs["ee_models_paths"], dict)
assert len(kwargs["ee_models_paths"]) == len(available_models)
assert kwargs["target_table_name"] == target_table_name
assert kwargs["workers_per_model"] == n_processes_per_model
# Check that CreateMiningCache.construct() was called
fake_create_mining_cache().construct.assert_called_once()