tensorflow::Status ValidateFeatureStatisticsWithSerializedInputs()

in tensorflow_data_validation/anomalies/feature_statistics_validator.cc [174:266]
83 lines of code
15 McCabe index (conditional complexity)

tensorflow::Status ValidateFeatureStatisticsWithSerializedInputs(
    const string& feature_statistics_proto_string,
    const string& schema_proto_string, const string& environment,
    const string& previous_span_statistics_proto_string,
    const string& serving_statistics_proto_string,
    const string& previous_version_statistics_proto_string,
    const string& features_needed_string,
    const string& validation_config_string, const bool enable_diff_regions,
    string* anomalies_proto_string) {
  tensorflow::metadata::v0::Schema schema;
  if (!schema.ParseFromString(schema_proto_string)) {
    return tensorflow::errors::InvalidArgument("Failed to parse Schema proto.");
  }

  tensorflow::metadata::v0::DatasetFeatureStatistics feature_statistics;
  if (!feature_statistics.ParseFromString(feature_statistics_proto_string)) {
    return tensorflow::errors::InvalidArgument(
        "Failed to parse DatasetFeatureStatistics proto.");
  }

  absl::optional<tensorflow::metadata::v0::DatasetFeatureStatistics>
      previous_span_statistics = tensorflow::gtl::nullopt;
  if (!previous_span_statistics_proto_string.empty()) {
    tensorflow::metadata::v0::DatasetFeatureStatistics tmp_stats;
    if (!tmp_stats.ParseFromString(previous_span_statistics_proto_string)) {
      return tensorflow::errors::InvalidArgument(
          "Failed to parse DatasetFeatureStatistics proto.");
    }
    previous_span_statistics = tmp_stats;
  }

  absl::optional<tensorflow::metadata::v0::DatasetFeatureStatistics>
      serving_statistics = tensorflow::gtl::nullopt;
  if (!serving_statistics_proto_string.empty()) {
    tensorflow::metadata::v0::DatasetFeatureStatistics tmp_stats;
    if (!tmp_stats.ParseFromString(serving_statistics_proto_string)) {
      return tensorflow::errors::InvalidArgument(
          "Failed to parse DatasetFeatureStatistics proto.");
    }
    serving_statistics = tmp_stats;
  }

  absl::optional<tensorflow::metadata::v0::DatasetFeatureStatistics>
      previous_version_statistics = tensorflow::gtl::nullopt;
  if (!previous_version_statistics_proto_string.empty()) {
    tensorflow::metadata::v0::DatasetFeatureStatistics tmp_stats;
    if (!tmp_stats.ParseFromString(previous_version_statistics_proto_string)) {
      return tensorflow::errors::InvalidArgument(
          "Failed to parse DatasetFeatureStatistics proto.");
    }
    previous_version_statistics = tmp_stats;
  }

  absl::optional<string> may_be_environment =
      tensorflow::gtl::nullopt;
  if (!environment.empty()) {
    may_be_environment = environment;
  }

  absl::optional<FeaturesNeeded> features_needed = gtl::nullopt;
  if (!features_needed_string.empty()) {
    FeaturesNeededProto parsed_proto;
    if (!parsed_proto.ParseFromString(features_needed_string)) {
      return tensorflow::errors::InvalidArgument(
          "Failed to parse FeaturesNeeded");
    }

    FeaturesNeeded parsed_feature_needed;
    TF_RETURN_IF_ERROR(
        FromFeaturesNeededProto(parsed_proto, &parsed_feature_needed));
    if (!parsed_feature_needed.empty()) {
      features_needed = parsed_feature_needed;
    }
  }

  data_validation::ValidationConfig validation_config;
  if (!validation_config.ParseFromString(validation_config_string)) {
    return tensorflow::errors::InvalidArgument(
        "Failed to parse ValidationConfig");
  }

  tensorflow::metadata::v0::Anomalies anomalies;
  TF_RETURN_IF_ERROR(ValidateFeatureStatistics(
      feature_statistics, schema, may_be_environment, previous_span_statistics,
      serving_statistics, previous_version_statistics, features_needed,
      validation_config, enable_diff_regions, &anomalies));

  if (!anomalies.SerializeToString(anomalies_proto_string)) {
    return tensorflow::errors::Internal(
        "Could not serialize Anomalies output proto to string.");
  }
  return tensorflow::Status::OK();
}