def test_cache_no_columns():
items_to_cache = set()
cache = Cache(items_to_cache)
data = [{'bla': 1, 'xyz': 2}, {'bla': 3, 'xyz': 4}]
# Should not be cached
cache.add('a', data)
a_bla = cache.retrieve('a.bla')
assert a_bla == []
def test_cache():
items_to_cache = set((('a', 'bla'), ('a', 'foo'), ('b', 'foo')))
cache = Cache(items_to_cache)
data = [{'bla': 1, 'xyz': 2}, {'bla': 3, 'xyz': 4}]
# Should be cached
cache.add('a', data)
# Should not be cached
cache.add('b', data)
a_bla = cache.retrieve('a.bla')
assert a_bla == [1, 3]
b_bla = cache.retrieve('b.bla')
assert b_bla == []
def _run_helper(self, sequence: Sequence[str],
deps: AbstractSet[Tuple[str, str]], seed: int,
num_rows: int) -> None:
cache = Cache(deps)
with DB(self.args.dsn) as dbconn:
rand_gen = Random(seed=seed)
for table_name in sequence:
table = self.tables[table_name]
rows_to_gen = Executor._get_num_rows_to_gen(
rand_gen, num_rows, table.scaler)
logger.info(
f'Generating {rows_to_gen} rows (seed {seed}) for table { table_name }'
)
data = BaseObject.sample_from_source(rand_gen, rows_to_gen,
table.schema, cache)
dbconn.ingest_table(table_name, table.schema, data)
cache.add(table_name, data)
