def test_uses_library_driver_backend(self):
self._my_library.config.services.search = None
# switch to sqlite.
self._my_library.database.driver = 'sqlite'
search = Search(self._my_library)
self.assertIsInstance(search.backend, SQLiteSearchBackend)
# switch to postgres.
self._my_library.database.driver = 'postgres'
search = Search(self._my_library)
self.assertIsInstance(search.backend, PostgreSQLSearchBackend)
def test_raises_missing_backend_exception_if_config_contains_invalid_backend(
self):
# services.search
try:
Search(self._my_library)
except Exception as exc:
self.assertIn('Missing backend', str(exc))
def test_uses_backend_from_config(self, fake_init):
# Disable backend initialization to reduce amount of mocks.
fake_init.return_value = None
self._my_library.config.services.search = 'whoosh'
search = Search(self._my_library)
self.assertIsInstance(search.backend, WhooshSearchBackend)
def test_uses_default_backend_if_library_database_search_is_not_implemented(
self, fake_init):
# Disable backend initialization to reduce amount of mocks.
fake_init.return_value = None
self._my_library.config.services.search = None
with patch.object(self._my_library.database, 'driver', 'mysql'):
search = Search(self._my_library)
self.assertIsInstance(search.backend, WhooshSearchBackend)
def test_indexes_library_datasets(self):
ds1 = MagicMock(spec=Dataset)
ds2 = MagicMock(spec=Dataset)
ds3 = MagicMock(spec=Dataset)
self._my_library.datasets = [ds1, ds2, ds3]
fake_backend = MagicMock(spec=SQLiteSearchBackend)
fake_backend.dataset_index = Mock()
fake_backend.partition_index = Mock()
fake_backend.identifier_index = Mock()
search = Search(self._my_library, backend=fake_backend)
search.index_library_datasets()
self.assertEqual(len(fake_backend.dataset_index.index_one.mock_calls),
3)
def test_postgres_query(self):
self._my_library.config.services.search = None
# switch to postgres.
self._my_library.database.driver = 'postgres'
search = Search(self._my_library)
self.assertIsInstance(search.backend, PostgreSQLSearchBackend)
be = search.backend
pg = be._get_dataset_index()
q = pg._make_query_from_terms(
'source healthindicators.gov diabetes asthma')
print str(q[0])
print q[1]
def test_feeds_tick_function_with_indexed_dataset(self):
# prepare mocks
fake_backend = MagicMock(spec=SQLiteSearchBackend)
fake_backend.dataset_index = Mock()
fake_backend.partition_index = Mock()
fake_backend.identifier_index = Mock()
tick_f = Mock()
fake_library = MagicMock(spec=Library)
fake_dataset = MagicMock(spec=Dataset)
fake_library.datasets = [fake_dataset]
# run
search = Search(fake_library, backend=fake_backend)
search.index_library_datasets(tick_f=tick_f)
# test
tick_f.assert_called_once_with('datasets: 1 partitions: 0')
def __init__(self, config=None, search=None, echo=None, read_only=False):
from sqlalchemy.exc import OperationalError
from ambry.orm.exc import DatabaseMissingError
if config:
self._config = config
else:
self._config = get_runconfig()
self.logger = logger
self.read_only = read_only # allow optimizations that assume we aren't building bundles.
self._echo = echo
self._fs = LibraryFilesystem(config)
self._db = Database(self._fs.database_dsn, echo=echo)
self._account_password = self.config.accounts.password
self._warehouse = None # Will be populated in the warehouse property.
try:
self._db.open()
except OperationalError as e:
raise DatabaseMissingError(
"Failed to open database '{}': {} ".format(self._db.dsn, e))
self.processes = None # Number of multiprocessing proccors. Default to all of them
if search:
self._search = Search(self, search)
else:
self._search = None
def search(self):
if not self._search:
self._search = Search(self)
return self._search