def test_get_samples(test_client, login_as_admin, virtual_dataset):
"""
Dataset API: Test get dataset samples
"""
# 1. should cache data
uri = (
f"/datasource/samples?datasource_id={virtual_dataset.id}&datasource_type=table"
)
# feeds data
test_client.post(uri)
# get from cache
rv = test_client.post(uri)
rv_data = json.loads(rv.data)
assert rv.status_code == 200
assert len(rv_data["result"]["data"]) == 10
assert QueryCacheManager.has(
rv_data["result"]["cache_key"],
region=CacheRegion.DATA,
)
assert rv_data["result"]["is_cached"]
# 2. should read through cache data
uri2 = f"/datasource/samples?datasource_id={virtual_dataset.id}&datasource_type=table&force=true"
# feeds data
test_client.post(uri2)
# force query
rv2 = test_client.post(uri2)
rv_data2 = json.loads(rv2.data)
assert rv2.status_code == 200
assert len(rv_data2["result"]["data"]) == 10
assert QueryCacheManager.has(
rv_data2["result"]["cache_key"],
region=CacheRegion.DATA,
)
assert not rv_data2["result"]["is_cached"]
# 3. data precision
assert "colnames" in rv_data2["result"]
assert "coltypes" in rv_data2["result"]
assert "data" in rv_data2["result"]
eager_samples = virtual_dataset.database.get_df(
f"select * from ({virtual_dataset.sql}) as tbl"
f' limit {app.config["SAMPLES_ROW_LIMIT"]}')
# the col3 is Decimal
eager_samples["col3"] = eager_samples["col3"].apply(float)
eager_samples = eager_samples.to_dict(orient="records")
assert eager_samples == rv_data2["result"]["data"]
def test_get_samples_on_physical_dataset(test_client, login_as_admin, physical_dataset):
uri = (
f"/datasource/samples?datasource_id={physical_dataset.id}&datasource_type=table"
)
rv = test_client.post(uri)
assert rv.status_code == 200
assert QueryCacheManager.has(
rv.json["result"]["cache_key"], region=CacheRegion.DATA
)
assert len(rv.json["result"]["data"]) == 10