def read_dataset_config(dataset_config_filename):
"""Returns a DeepVariantDatasetConfig proto read from the dataset config file.
Args:
dataset_config_filename: String. Path to the dataset config pbtxt file.
Returns:
A DeepVariantDatasetConfig proto from the dataset_config file.
Raises:
ValueError: if the dataset config doesn't have the necessary information.
"""
with tf.gfile.GFile(dataset_config_filename) as f:
dataset_config = text_format.Parse(
f.read(), deepvariant_pb2.DeepVariantDatasetConfig())
if not dataset_config.name:
raise ValueError('dataset_config needs to have a name')
if not dataset_config.tfrecord_path:
raise ValueError('The dataset in the config {} does not have a '
'tfrecord_path.'.format(dataset_config_filename))
# redacted
# of num_examples.
if not dataset_config.num_examples:
raise ValueError('The dataset in the config {} does not have a '
'num_examples.'.format(dataset_config_filename))
return dataset_config
def _test_dataset_config(filename, **kwargs):
"""Creates a DeepVariantDatasetConfig(**kwargs) and writes it to filename."""
dataset_config_pbtext_filename = test_utils.test_tmpfile(filename)
dataset_config = deepvariant_pb2.DeepVariantDatasetConfig(**kwargs)
data_providers.write_dataset_config_to_pbtxt(
dataset_config, dataset_config_pbtext_filename)
return dataset_config_pbtext_filename
def read_dataset_config(dataset_config_pbtxt_filename):
with tf.gfile.GFile(dataset_config_pbtxt_filename) as f:
return text_format.Parse(
f.read(), deepvariant_pb2.DeepVariantDatasetConfig())