def test_load_by_id():
ds = new_data_set("test-dataset")
run_id = ds.run_id
ds.mark_started()
ds.mark_completed()
loaded_ds = load_by_id(run_id)
assert ds.started is True
assert ds.pristine is False
assert ds.running is False
assert loaded_ds.completed is True
assert loaded_ds.exp_id == 1
ds = new_data_set("test-dataset-unfinished")
run_id = ds.run_id
loaded_ds = load_by_id(run_id)
assert ds.pristine is True
assert ds.running is False
assert ds.started is False
assert loaded_ds.completed is False
assert loaded_ds.exp_id == 1
# let's take a run number that is not in the temporary test database file
non_existing_run_id = run_id + 1
with pytest.raises(ValueError,
match=f"Run with run_id "
f"{non_existing_run_id} does not "
f"exist in the database"):
_ = load_by_id(non_existing_run_id)
def test_load_by_id():
ds = new_data_set("test-dataset")
run_id = ds.run_id
ds.mark_complete()
loaded_ds = load_by_id(run_id)
assert loaded_ds.completed is True
assert loaded_ds.exp_id == 1
ds = new_data_set("test-dataset-unfinished")
run_id = ds.run_id
loaded_ds = load_by_id(run_id)
assert loaded_ds.completed is False
assert loaded_ds.exp_id == 1
def test_experiment_info_in_dataset():
exp = new_experiment(name="for_loading", sample_name="no_sample")
ds = new_data_set("my_first_ds")
assert ds.exp_id == exp.exp_id
assert ds.exp_name == exp.name
assert ds.sample_name == exp.sample_name
def test_completed_timestamp_for_not_completed_dataset():
_ = new_experiment(name="for_loading", sample_name="no_sample")
ds = new_data_set("my_first_ds")
assert False is ds.completed
assert None is ds.completed_timestamp_raw
assert None is ds.completed_timestamp()
def test_run_timestamp():
_ = new_experiment(name="for_loading", sample_name="no_sample")
t_before_data_set = time.time()
ds = new_data_set("my_first_ds")
t_after_data_set = time.time()
actual_run_timestamp_raw = ds.run_timestamp_raw
assert t_before_data_set <= actual_run_timestamp_raw <= t_after_data_set
def new_data_set(self, name, specs: SPECS = None, values=None,
metadata=None) -> DataSet:
""" Create a new dataset in this experimetn
Args:
name: the name of the new dataset
specs: list of parameters to create this data_set with
values: the values to associate with the parameters
metadata: the values to associate with the dataset
"""
return new_data_set(name, self.exp_id, specs, values, metadata)
def test_load_by_counter():
exp = new_experiment(name="for_loading", sample_name="no_sample")
ds = new_data_set("my_first_ds")
loaded_ds = load_by_counter(exp.exp_id, 1)
assert loaded_ds.completed is False
ds.mark_complete()
loaded_ds = load_by_counter(exp.exp_id, 1)
assert loaded_ds.completed is True
def test_completed_timestamp_for_not_completed_dataset():
_ = new_experiment(name="for_loading", sample_name="no_sample")
ds = new_data_set("my_first_ds")
assert ds.pristine is True
assert ds.started is False
assert ds.running is False
assert ds.completed is False
assert ds.completed_timestamp_raw is None
assert ds.completed_timestamp() is None
def test_get_guids_from_run_spec_warns():
ds = new_data_set("test-dataset")
run_id = ds.run_id
ds.mark_started()
ds.mark_completed()
expected_guid = ds.guid
with pytest.warns(
expected_warning=QCoDeSDeprecationWarning, match="Unused part of private api"
):
loaded_guids = get_guids_from_run_spec(captured_run_id=run_id, conn=ds.conn)
assert len(loaded_guids) == 1
assert loaded_guids[0] == expected_guid
def test_completed_timestamp():
_ = new_experiment(name="for_loading", sample_name="no_sample")
ds = new_data_set("my_first_ds")
t_before_complete = time.time()
ds.mark_complete()
t_after_complete = time.time()
actual_completed_timestamp_raw = ds.completed_timestamp_raw
assert t_before_complete \
<= actual_completed_timestamp_raw \
<= t_after_complete
def test_get_run_attributes() -> None:
name = "test-dataset"
ds = new_data_set(name)
ds.mark_started()
ds.mark_completed()
ds.add_metadata("foo", "bar")
loaded_raw_attrs = get_raw_run_attributes(ds.conn, ds.guid)
assert loaded_raw_attrs is not None
assert loaded_raw_attrs["run_id"] == ds.run_id
assert loaded_raw_attrs["counter"] == ds.counter
assert loaded_raw_attrs["captured_counter"] == ds.captured_counter
assert loaded_raw_attrs["captured_run_id"] == ds.captured_run_id
assert loaded_raw_attrs["captured_run_id"] == ds.captured_run_id
assert loaded_raw_attrs["experiment"] == get_experiment_attributes_by_exp_id(
ds.conn, ds.exp_id
)
assert loaded_raw_attrs["experiment"]["exp_id"] == ds.exp_id
assert loaded_raw_attrs["experiment"]["name"] == ds.exp_name
assert loaded_raw_attrs["experiment"]["sample_name"] == ds.sample_name
assert loaded_raw_attrs["name"] == name
assert loaded_raw_attrs["run_timestamp"] == ds.run_timestamp_raw
assert loaded_raw_attrs["completed_timestamp"] == ds.completed_timestamp_raw
assert loaded_raw_attrs["parent_dataset_links"] == "[]"
assert "interdependencies" in loaded_raw_attrs["run_description"]
assert loaded_raw_attrs["snapshot"] is None
assert loaded_raw_attrs["metadata"] == {"foo": "bar"}
loaded_attrs = get_run_attributes(ds.conn, ds.guid)
assert loaded_attrs is not None
assert loaded_attrs["run_id"] == ds.run_id
assert loaded_attrs["counter"] == ds.counter
assert loaded_attrs["captured_counter"] == ds.captured_counter
assert loaded_attrs["captured_run_id"] == ds.captured_run_id
assert loaded_attrs["captured_run_id"] == ds.captured_run_id
assert loaded_attrs["experiment"] == get_experiment_attributes_by_exp_id(
ds.conn, ds.exp_id
)
assert loaded_attrs["experiment"]["exp_id"] == ds.exp_id
assert loaded_attrs["name"] == name
assert loaded_attrs["run_timestamp"] == raw_time_to_str_time(ds.run_timestamp_raw)
assert loaded_attrs["completed_timestamp"] == raw_time_to_str_time(
ds.completed_timestamp_raw
)
assert loaded_attrs["parent_dataset_links"] == []
assert isinstance(loaded_attrs["run_description"], RunDescriber)
assert loaded_attrs["snapshot"] is None
assert loaded_attrs["metadata"] == {"foo": "bar"}
def new_data_set(self, name: str,
specs: Optional[SPECS] = None,
values: Optional[VALUES] = None,
metadata: Optional[Any] = None) -> DataSet:
"""
Create a new dataset in this experiment
Args:
name: the name of the new dataset
specs: list of parameters (as ParamSpecs) to create this data_set
with
values: the values to associate with the parameters
metadata: the metadata to associate with the dataset
"""
return new_data_set(name, self.exp_id, specs, values, metadata,
conn=self.conn)
def test_run_timestamp_with_default_format():
_ = new_experiment(name="for_loading", sample_name="no_sample")
t_before_data_set = time.time()
ds = new_data_set("my_first_ds")
t_after_data_set = time.time()
# Note that here we also test the default format of `run_timestamp`
actual_run_timestamp_raw = time.mktime(
time.strptime(ds.run_timestamp(), "%Y-%m-%d %H:%M:%S"))
# Note that because the default format precision is 1 second, we add this
# second to the right side of the comparison
t_before_data_set_secs = floor(t_before_data_set)
t_after_data_set_secs = floor(t_after_data_set)
assert t_before_data_set_secs \
<= actual_run_timestamp_raw \
<= t_after_data_set_secs + 1
def dataset(experiment):
dataset = new_data_set("test-dataset")
yield dataset
dataset.conn.close()
