def test_get_csv_header(self):
temp_directory = self._get_temp_dir()
delimiter = ','
records = ['feature1,feature2', '1.0,aa']
expected_header = ['feature1', 'feature2']
self._write_records_to_csv(records, temp_directory, 'input_data_1.csv')
self._write_records_to_csv(records, temp_directory, 'input_data_2.csv')
data_location = os.path.join(temp_directory, 'input_data_*.csv')
header = stats_gen_lib.get_csv_header(data_location, delimiter)
self.assertEqual(header, expected_header)
def test_get_csv_header_different_headers(self):
temp_directory = self._get_temp_dir()
delimiter = ','
records_1 = ['feature1,feature2', '1.0,aa']
records_2 = ['feature1,feature2_different', '2.0,bb']
self._write_records_to_csv(records_1, temp_directory, 'input_data_1.csv')
self._write_records_to_csv(records_2, temp_directory, 'input_data_2.csv')
data_location = os.path.join(temp_directory, 'input_data_*.csv')
with self.assertRaisesRegexp(ValueError, 'Files have different headers.'):
_ = stats_gen_lib.get_csv_header(data_location, delimiter)
def test_get_csv_header_new_line(self):
temp_directory = self._get_temp_dir()
delimiter = ','
records = ['"\n","feature2"', '1.0,aa']
expected_header = ['\n', 'feature2']
self._write_records_to_csv(records,
temp_directory,
'input_data_1.csv.gz',
compression_type='gzip')
self._write_records_to_csv(records,
temp_directory,
'input_data_2.csv.gz',
compression_type='gzip')
data_location = os.path.join(temp_directory, 'input_data_*.csv.gz')
header = stats_gen_lib.get_csv_header(data_location, delimiter)
self.assertEqual(header, expected_header)
def validate_examples_in_csv(
data_location: Text,
stats_options: options.StatsOptions,
column_names: Optional[List[types.FeatureName]] = None,
delimiter: Text = ',',
output_path: Optional[Text] = None,
# TODO(b/131719250): Add option to output a sample of anomalous examples for
# each anomaly reason.
pipeline_options: Optional[PipelineOptions] = None,
) -> statistics_pb2.DatasetFeatureStatisticsList:
"""Validates examples in csv files.
Runs a Beam pipeline to detect anomalies on a per-example basis. If this
function detects anomalous examples, it generates summary statistics regarding
the set of examples that exhibit each anomaly.
This is a convenience function for users with data in CSV format.
Users with data in unsupported file/data formats, or users who wish
to create their own Beam pipelines need to use the 'IdentifyAnomalousExamples'
PTransform API directly instead.
Args:
data_location: The location of the input data files.
stats_options: `tfdv.StatsOptions` for generating data statistics. This must
contain a schema.
column_names: A list of column names to be treated as the CSV header. Order
must match the order in the input CSV files. If this argument is not
specified, we assume the first line in the input CSV files as the header.
Note that this option is valid only for 'csv' input file format.
delimiter: A one-character string used to separate fields in a CSV file.
output_path: The file path to output data statistics result to. If None, the
function uses a temporary directory. The output will be a TFRecord file
containing a single data statistics list proto, and can be read with the
'load_statistics' function.
If you run this function on Google Cloud, you must specify an
output_path. Specifying None may cause an error.
pipeline_options: Optional beam pipeline options. This allows users to
specify various beam pipeline execution parameters like pipeline runner
(DirectRunner or DataflowRunner), cloud dataflow service project id, etc.
See https://cloud.google.com/dataflow/pipelines/specifying-exec-params for
more details.
Returns:
A DatasetFeatureStatisticsList proto in which each dataset consists of the
set of examples that exhibit a particular anomaly.
Raises:
ValueError: If the specified stats_options does not include a schema.
"""
if stats_options.schema is None:
raise ValueError('The specified stats_options must include a schema.')
if output_path is None:
output_path = os.path.join(tempfile.mkdtemp(),
'anomaly_stats.tfrecord')
output_dir_path = os.path.dirname(output_path)
if not tf.io.gfile.exists(output_dir_path):
tf.io.gfile.makedirs(output_dir_path)
# If a header is not provided, assume the first line in a file
# to be the header.
skip_header_lines = 1 if column_names is None else 0
if column_names is None:
column_names = stats_gen_lib.get_csv_header(data_location, delimiter)
with beam.Pipeline(options=pipeline_options) as p:
_ = (
p
| 'ReadData' >> beam.io.textio.ReadFromText(
file_pattern=data_location,
skip_header_lines=skip_header_lines)
| 'DecodeData' >> csv_decoder.DecodeCSV(
column_names=column_names,
delimiter=delimiter,
schema=stats_options.schema,
infer_type_from_schema=stats_options.infer_type_from_schema,
desired_batch_size=1)
| 'DetectAnomalies' >>
validation_api.IdentifyAnomalousExamples(stats_options)
| 'GenerateSummaryStatistics' >>
stats_impl.GenerateSlicedStatisticsImpl(stats_options,
is_slicing_enabled=True)
# TODO(b/112014711) Implement a custom sink to write the stats proto.
| 'WriteStatsOutput' >> beam.io.WriteToTFRecord(
output_path,
shard_name_template='',
coder=beam.coders.ProtoCoder(
statistics_pb2.DatasetFeatureStatisticsList)))
return stats_gen_lib.load_statistics(output_path)
def test_get_csv_header_empty_file(self):
empty_file = os.path.join(self._get_temp_dir(), 'empty.csv')
open(empty_file, 'w+').close()
delimiter = ','
with self.assertRaisesRegexp(ValueError, 'Found empty file.*'):
_ = stats_gen_lib.get_csv_header(empty_file, delimiter)
def test_get_csv_header_no_file(self):
data_location = os.path.join(self._get_temp_dir(), 'fileA.csv')
delimiter = ','
with self.assertRaisesRegexp(ValueError, 'No file found.*'):
_ = stats_gen_lib.get_csv_header(data_location, delimiter)
def validate_examples_in_csv(
data_location: Text,
stats_options: options.StatsOptions,
column_names: Optional[List[types.FeatureName]] = None,
delimiter: Text = ',',
output_path: Optional[Text] = None,
pipeline_options: Optional[PipelineOptions] = None,
num_sampled_examples=0,
) -> Union[statistics_pb2.DatasetFeatureStatisticsList, Tuple[
statistics_pb2.DatasetFeatureStatisticsList, Mapping[str,
pd.DataFrame]]]:
"""Validates examples in csv files.
Runs a Beam pipeline to detect anomalies on a per-example basis. If this
function detects anomalous examples, it generates summary statistics regarding
the set of examples that exhibit each anomaly.
This is a convenience function for users with data in CSV format.
Users with data in unsupported file/data formats, or users who wish
to create their own Beam pipelines need to use the 'IdentifyAnomalousExamples'
PTransform API directly instead.
Args:
data_location: The location of the input data files.
stats_options: `tfdv.StatsOptions` for generating data statistics. This must
contain a schema.
column_names: A list of column names to be treated as the CSV header. Order
must match the order in the input CSV files. If this argument is not
specified, we assume the first line in the input CSV files as the header.
Note that this option is valid only for 'csv' input file format.
delimiter: A one-character string used to separate fields in a CSV file.
output_path: The file path to output data statistics result to. If None, the
function uses a temporary directory. The output will be a TFRecord file
containing a single data statistics list proto, and can be read with the
'load_statistics' function. If you run this function on Google Cloud, you
must specify an output_path. Specifying None may cause an error.
pipeline_options: Optional beam pipeline options. This allows users to
specify various beam pipeline execution parameters like pipeline runner
(DirectRunner or DataflowRunner), cloud dataflow service project id, etc.
See https://cloud.google.com/dataflow/pipelines/specifying-exec-params for
more details.
num_sampled_examples: If set, returns up to this many examples of each
anomaly type as a map from anomaly reason string to pd.DataFrame.
Returns:
If num_sampled_examples is zero, returns a single
DatasetFeatureStatisticsList proto in which each dataset consists of the
set of examples that exhibit a particular anomaly. If
num_sampled_examples is nonzero, returns the same statistics
proto as well as a mapping from anomaly to a pd.DataFrame of CSV rows
exhibiting that anomaly.
Raises:
ValueError: If the specified stats_options does not include a schema.
"""
if stats_options.schema is None:
raise ValueError('The specified stats_options must include a schema.')
if output_path is None:
output_path = os.path.join(tempfile.mkdtemp(),
'anomaly_stats.tfrecord')
output_dir_path = os.path.dirname(output_path)
if not tf.io.gfile.exists(output_dir_path):
tf.io.gfile.makedirs(output_dir_path)
if num_sampled_examples:
sample_materializer = io_util.Materializer(output_dir_path)
# If a header is not provided, assume the first line in a file
# to be the header.
skip_header_lines = 1 if column_names is None else 0
if column_names is None:
column_names = stats_gen_lib.get_csv_header(data_location, delimiter)
with beam.Pipeline(options=pipeline_options) as p:
anomalous_examples = (
p
| 'ReadData' >> beam.io.textio.ReadFromText(
file_pattern=data_location,
skip_header_lines=skip_header_lines)
| 'DecodeData' >> csv_decoder.DecodeCSV(
column_names=column_names,
delimiter=delimiter,
schema=stats_options.schema
if stats_options.infer_type_from_schema else None,
desired_batch_size=1)
| 'DetectAnomalies' >>
validation_api.IdentifyAnomalousExamples(stats_options))
_ = (anomalous_examples
| 'GenerateSummaryStatistics' >>
stats_impl.GenerateSlicedStatisticsImpl(stats_options,
is_slicing_enabled=True)
| 'WriteStatsOutput' >>
stats_api.WriteStatisticsToTFRecord(output_path))
if num_sampled_examples:
_ = (anomalous_examples
| 'Sample' >>
beam.combiners.Sample.FixedSizePerKey(num_sampled_examples)
| 'ToPandas' >> beam.FlatMap(_encode_pandas_and_key)
| 'WriteSamples' >> sample_materializer.writer())
if num_sampled_examples:
samples_per_reason_acc = collections.defaultdict(list)
for reason, pandas_dataframe in sample_materializer.reader():
samples_per_reason_acc[reason].append(pandas_dataframe)
samples_per_reason = {}
for reason, dataframes in samples_per_reason_acc.items():
samples_per_reason[reason] = pd.concat(dataframes)
sample_materializer.cleanup()
return stats_util.load_statistics(output_path), samples_per_reason
return stats_util.load_statistics(output_path)
