def get_run_profiles(run_id: str, dataset_name: str = "default", client=None):
"""
Retrieve all whylogs DatasetProfile for a given run and a given dataset name.
:param client: :py:class:`mlflow.tracking.MlflowClient`
:rtype: :py:class:`typing.List[whylogs.DatasetProfile]`
:param run_id: the run id
:param dataset_name: the dataset name within a run. If not set, use the default value "default"
"""
import mlflow
import tempfile
import shutil
from whylogs import DatasetProfile
if client is None:
client = mlflow.tracking.MlflowClient()
artifacts = client.list_artifacts(run_id,
path=f"{_WHYLOGS_PATH}/{dataset_name}")
if len(artifacts) == 1 and not artifacts[0].is_dir:
tmp_dir = tempfile.mkdtemp()
output_file = client.download_artifacts(run_id, artifacts[0].path,
tmp_dir)
try:
with open(output_file, "rb") as f:
return list(DatasetProfile.parse_delimited(f.read()))
finally:
shutil.rmtree(tmp_dir)
else:
return []
def test_write_template_path():
data_time = time.from_utc_ms(9999)
session_time = time.from_utc_ms(88888)
path_template = "$name-$session_timestamp-$dataset_timestamp-$session_id"
writer_config = WriterConfig("local", ["protobuf", "flat"], "output", path_template, "dataset-profile-$name")
writer = writer_from_config(writer_config)
dp = DatasetProfile("name", data_time, session_time, session_id="session")
assert writer.path_suffix(dp) == "name-88888-9999-session"
assert writer.file_name(dp, ".txt") == "dataset-profile-name.txt"