Calcolo di k-map per un set di dati

La mappa K è molto simile all'anonimità k, tranne per il fatto che presuppone che l'aggressore molto probabilmente non sappia chi è presente nel set di dati. Utilizza la mappatura k se il tuo set di dati è relativamente piccolo o se il livello di impegno necessario per generalizzare gli attributi sarebbe troppo elevato.

Come per l'anonimizzazione k, anche la mappatura k richiede di determinare quali colonne del database sono quasi-identificatori. In questo modo, stai indicando quali dati un aggressore utilizzerà molto probabilmente per identificare nuovamente i soggetti. Inoltre, il calcolo di un valore della mappa k richiede un set di dati di reidentificazione: una tabella più grande con cui confrontare le righe del set di dati originale.

Questo argomento mostra come calcolare i valori di k-map per un set di dati utilizzando Sensitive Data Protection. Per ulteriori informazioni sulla mappa k o sull'analisi del rischio in generale, consulta l'argomento sui concetti di analisi del rischio prima di continuare.

Prima di iniziare

Prima di continuare, assicurati di aver eseguito quanto segue:

  1. Accedi al tuo Account Google.
  2. Nella console Google Cloud, nella pagina del selettore di progetti, seleziona o crea un progetto Google Cloud.
  3. Vai al selettore dei progetti
  4. Verifica che la fatturazione sia attivata per il tuo progetto Google Cloud. Scopri come verificare che la fatturazione sia attivata per il tuo progetto.
  5. Attiva Sensitive Data Protection.
  6. Abilita Sensitive Data Protection

  7. Seleziona un set di dati BigQuery da analizzare. La funzionalità Protezione dei dati sensibili stima la metrica mappa k eseguendo la scansione di una tabella BigQuery.
  8. Determina i tipi di set di dati che vuoi utilizzare per modellare il set di dati di attacco. Per ulteriori informazioni, consulta la pagina di riferimento dell'oggetto KMapEstimationConfig e i termini e le tecniche di analisi del rischio.

Calcolare le stime della mappa k

Puoi stimare i valori della mappa k utilizzando Sensitive Data Protection, che utilizza un modello statistico per stimare un set di dati di reidentificazione. Ciò è in contrasto con gli altri metodi di analisi del rischio, in cui il set di dati sugli attacchi è esplicitamente conosciuto. A seconda del tipo di dati, Sensitive Data Protection utilizza set di dati disponibili pubblicamente (ad esempio, dal censimento degli Stati Uniti) o un modello statistico personalizzato (ad esempio, una o più tabelle BigQuery specificate) oppure esegue l'estrapolazione dalla distribuzione dei valori nel set di dati di input. Per ulteriori informazioni, consulta la pagina di riferimento per l'oggetto KMapEstimationConfig.

Per calcolare una stima della mappa k utilizzando Sensitive Data Protection, configura prima il job di rischio. Componi una richiesta alla risorsa projects.dlpJobs, dove PROJECT_ID indica l'identificatore del progetto:

https://dlp.googleapis.com/v2/projects/PROJECT_ID/dlpJobs

La richiesta contiene un oggetto RiskAnalysisJobConfig composto da quanto segue:

  • Un oggetto PrivacyMetric. Qui specifichi che vuoi calcolare la mappa k specificando un oggetto KMapEstimationConfig contenente quanto segue:

    • quasiIds[]: obbligatorio. Campi (oggetti TaggedField) considerati quasi identificatori da scansionare e utilizzare per calcolare k-map. Non è possibile assegnare lo stesso tag a due colonne. Può trattarsi di uno dei seguenti:

      • Un infoType: in questo modo, la funzionalità Protezione dei dati sensibili utilizza il set di dati pubblico pertinente come modello statistico della popolazione, inclusi i codici postali, i codici regionali, le età e i generi degli Stati Uniti.
      • Un infoType personalizzato: un tag personalizzato in cui indichi una tabella ausiliaria (un oggetto AuxiliaryTable) contenente informazioni statistiche sui possibili valori di questa colonna.
      • Il tag inferred: se non è indicato alcun tag semantico, specifica inferred. Sensitive Data Protection deducono il modello statistico dalla distribuzione dei valori nei dati di input.
    • regionCode: un codice regione ISO 3166-1 alpha-2 da utilizzare nella creazione di modelli statistici per Sensitive Data Protection. Questo valore è obbligatorio se nessuna colonna è contrassegnata con un infoType specifico per regione (ad esempio un codice postale statunitense) o un codice regione.

    • auxiliaryTables[]: tabelle ausiliarie (oggetti AuxiliaryTable) da utilizzare nell'analisi. Ogni tag personalizzato utilizzato per taggare una colonna di quasi identificatori (da quasiIds[]) deve apparire in esattamente una colonna di una tabella ausiliaria.

  • Un oggetto BigQueryTable. Specifica la tabella BigQuery da analizzare includendo quanto segue:

    • projectId: l'ID del progetto contenente la tabella.
    • datasetId: l'ID set di dati della tabella.
    • tableId: il nome della tabella.
  • Un insieme di uno o più Action oggetti, che rappresentano le azioni da eseguire, nell'ordine specificato, al completamento del job. Ogni oggetto Action può contenere una delle seguenti azioni:

Esempi di codice

Di seguito è riportato codice campione in diversi linguaggi che mostra come utilizzare Sensitive Data Protection per calcolare un valore della mappa k.

Go

Per scoprire come installare e utilizzare la libreria client per Sensitive Data Protection, consulta Librerie client di Sensitive Data Protection.

Per autenticarti a Sensitive Data Protection, configura le Credenziali predefinite dell'applicazione. Per ulteriori informazioni, consulta Configurare l'autenticazione per un ambiente di sviluppo locale.

import (
	"context"
	"fmt"
	"io"
	"strings"
	"time"

	dlp "cloud.google.com/go/dlp/apiv2"
	"cloud.google.com/go/dlp/apiv2/dlppb"
	"cloud.google.com/go/pubsub"
	"github.com/golang/protobuf/ptypes/empty"
)

// riskKMap runs K Map on the given data.
func riskKMap(w io.Writer, projectID, dataProject, pubSubTopic, pubSubSub, datasetID, tableID, region string, columnNames ...string) error {
	// projectID := "my-project-id"
	// dataProject := "bigquery-public-data"
	// pubSubTopic := "dlp-risk-sample-topic"
	// pubSubSub := "dlp-risk-sample-sub"
	// datasetID := "san_francisco"
	// tableID := "bikeshare_trips"
	// region := "US"
	// columnNames := "zip_code"
	ctx := context.Background()
	client, err := dlp.NewClient(ctx)
	if err != nil {
		return fmt.Errorf("dlp.NewClient: %w", err)
	}

	// Create a PubSub Client used to listen for when the inspect job finishes.
	pubsubClient, err := pubsub.NewClient(ctx, projectID)
	if err != nil {
		return err
	}
	defer pubsubClient.Close()

	// Create a PubSub subscription we can use to listen for messages.
	// Create the Topic if it doesn't exist.
	t := pubsubClient.Topic(pubSubTopic)
	topicExists, err := t.Exists(ctx)
	if err != nil {
		return err
	}
	if !topicExists {
		if t, err = pubsubClient.CreateTopic(ctx, pubSubTopic); err != nil {
			return err
		}
	}

	// Create the Subscription if it doesn't exist.
	s := pubsubClient.Subscription(pubSubSub)
	subExists, err := s.Exists(ctx)
	if err != nil {
		return err
	}
	if !subExists {
		if s, err = pubsubClient.CreateSubscription(ctx, pubSubSub, pubsub.SubscriptionConfig{Topic: t}); err != nil {
			return err
		}
	}

	// topic is the PubSub topic string where messages should be sent.
	topic := "projects/" + projectID + "/topics/" + pubSubTopic

	// Build the QuasiID slice.
	var q []*dlppb.PrivacyMetric_KMapEstimationConfig_TaggedField
	for _, c := range columnNames {
		q = append(q, &dlppb.PrivacyMetric_KMapEstimationConfig_TaggedField{
			Field: &dlppb.FieldId{
				Name: c,
			},
			Tag: &dlppb.PrivacyMetric_KMapEstimationConfig_TaggedField_Inferred{
				Inferred: &empty.Empty{},
			},
		})
	}

	// Create a configured request.
	req := &dlppb.CreateDlpJobRequest{
		Parent: fmt.Sprintf("projects/%s/locations/global", projectID),
		Job: &dlppb.CreateDlpJobRequest_RiskJob{
			RiskJob: &dlppb.RiskAnalysisJobConfig{
				// PrivacyMetric configures what to compute.
				PrivacyMetric: &dlppb.PrivacyMetric{
					Type: &dlppb.PrivacyMetric_KMapEstimationConfig_{
						KMapEstimationConfig: &dlppb.PrivacyMetric_KMapEstimationConfig{
							QuasiIds:   q,
							RegionCode: region,
						},
					},
				},
				// SourceTable describes where to find the data.
				SourceTable: &dlppb.BigQueryTable{
					ProjectId: dataProject,
					DatasetId: datasetID,
					TableId:   tableID,
				},
				// Send a message to PubSub using Actions.
				Actions: []*dlppb.Action{
					{
						Action: &dlppb.Action_PubSub{
							PubSub: &dlppb.Action_PublishToPubSub{
								Topic: topic,
							},
						},
					},
				},
			},
		},
	}
	// Create the risk job.
	j, err := client.CreateDlpJob(ctx, req)
	if err != nil {
		return fmt.Errorf("CreateDlpJob: %w", err)
	}
	fmt.Fprintf(w, "Created job: %v\n", j.GetName())
	// Wait for the risk job to finish by waiting for a PubSub message.
	// This only waits for 10 minutes. For long jobs, consider using a truly
	// asynchronous execution model such as Cloud Functions.
	ctx, cancel := context.WithTimeout(ctx, 10*time.Minute)
	defer cancel()
	err = s.Receive(ctx, func(ctx context.Context, msg *pubsub.Message) {
		// If this is the wrong job, do not process the result.

		if msg.Attributes["DlpJobName"] != j.GetName() {
			msg.Nack()
			return
		}
		msg.Ack()
		time.Sleep(500 * time.Millisecond)
		j, err := client.GetDlpJob(ctx, &dlppb.GetDlpJobRequest{
			Name: j.GetName(),
		})

		if err != nil {
			fmt.Fprintf(w, "GetDlpJob: %v", err)
			return
		}
		h := j.GetRiskDetails().GetKMapEstimationResult().GetKMapEstimationHistogram()
		for i, b := range h {
			fmt.Fprintf(w, "Histogram bucket %v\n", i)
			fmt.Fprintf(w, "  Anonymity range: [%v,%v]\n", b.GetMaxAnonymity(), b.GetMaxAnonymity())
			fmt.Fprintf(w, "  %v unique values total\n", b.GetBucketSize())
			for _, v := range b.GetBucketValues() {
				var qvs []string
				for _, qv := range v.GetQuasiIdsValues() {
					qvs = append(qvs, qv.String())
				}
				fmt.Fprintf(w, "    QuasiID values: %s\n", strings.Join(qvs, ", "))
				fmt.Fprintf(w, "    Estimated anonymity: %v\n", v.GetEstimatedAnonymity())
			}
		}
		// Stop listening for more messages.
		cancel()
	})
	if err != nil {
		return fmt.Errorf("Recieve: %w", err)
	}
	return nil
}

Java

Per scoprire come installare e utilizzare la libreria client per Sensitive Data Protection, consulta Librerie client di Sensitive Data Protection.

Per autenticarti a Sensitive Data Protection, configura le Credenziali predefinite dell'applicazione. Per ulteriori informazioni, consulta Configurare l'autenticazione per un ambiente di sviluppo locale.


import com.google.api.core.SettableApiFuture;
import com.google.cloud.dlp.v2.DlpServiceClient;
import com.google.cloud.pubsub.v1.AckReplyConsumer;
import com.google.cloud.pubsub.v1.MessageReceiver;
import com.google.cloud.pubsub.v1.Subscriber;
import com.google.privacy.dlp.v2.Action;
import com.google.privacy.dlp.v2.Action.PublishToPubSub;
import com.google.privacy.dlp.v2.AnalyzeDataSourceRiskDetails.KMapEstimationResult;
import com.google.privacy.dlp.v2.AnalyzeDataSourceRiskDetails.KMapEstimationResult.KMapEstimationHistogramBucket;
import com.google.privacy.dlp.v2.AnalyzeDataSourceRiskDetails.KMapEstimationResult.KMapEstimationQuasiIdValues;
import com.google.privacy.dlp.v2.BigQueryTable;
import com.google.privacy.dlp.v2.CreateDlpJobRequest;
import com.google.privacy.dlp.v2.DlpJob;
import com.google.privacy.dlp.v2.FieldId;
import com.google.privacy.dlp.v2.GetDlpJobRequest;
import com.google.privacy.dlp.v2.InfoType;
import com.google.privacy.dlp.v2.LocationName;
import com.google.privacy.dlp.v2.PrivacyMetric;
import com.google.privacy.dlp.v2.PrivacyMetric.KMapEstimationConfig;
import com.google.privacy.dlp.v2.PrivacyMetric.KMapEstimationConfig.TaggedField;
import com.google.privacy.dlp.v2.RiskAnalysisJobConfig;
import com.google.pubsub.v1.ProjectSubscriptionName;
import com.google.pubsub.v1.ProjectTopicName;
import com.google.pubsub.v1.PubsubMessage;
import java.io.IOException;
import java.util.ArrayList;
import java.util.Arrays;
import java.util.List;
import java.util.concurrent.ExecutionException;
import java.util.concurrent.TimeUnit;
import java.util.concurrent.TimeoutException;
import java.util.stream.Collectors;

@SuppressWarnings("checkstyle:AbbreviationAsWordInName")
class RiskAnalysisKMap {

  public static void main(String[] args) throws Exception {
    // TODO(developer): Replace these variables before running the sample.
    String projectId = "your-project-id";
    String datasetId = "your-bigquery-dataset-id";
    String tableId = "your-bigquery-table-id";
    String topicId = "pub-sub-topic";
    String subscriptionId = "pub-sub-subscription";
    calculateKMap(projectId, datasetId, tableId, topicId, subscriptionId);
  }

  public static void calculateKMap(
      String projectId, String datasetId, String tableId, String topicId, String subscriptionId)
      throws ExecutionException, InterruptedException, IOException {
    // Initialize client that will be used to send requests. This client only needs to be created
    // once, and can be reused for multiple requests. After completing all of your requests, call
    // the "close" method on the client to safely clean up any remaining background resources.
    try (DlpServiceClient dlpServiceClient = DlpServiceClient.create()) {
      // Specify the BigQuery table to analyze
      BigQueryTable bigQueryTable =
          BigQueryTable.newBuilder()
              .setProjectId(projectId)
              .setDatasetId(datasetId)
              .setTableId(tableId)
              .build();

      // These values represent the column names of quasi-identifiers to analyze
      List<String> quasiIds = Arrays.asList("Age", "Gender");

      // These values represent the info types corresponding to the quasi-identifiers above
      List<String> infoTypeNames = Arrays.asList("AGE", "GENDER");

      // Tag each of the quasiId column names with its corresponding infoType
      List<InfoType> infoTypes =
          infoTypeNames.stream()
              .map(it -> InfoType.newBuilder().setName(it).build())
              .collect(Collectors.toList());

      if (quasiIds.size() != infoTypes.size()) {
        throw new IllegalArgumentException("The numbers of quasi-IDs and infoTypes must be equal!");
      }

      List<TaggedField> taggedFields = new ArrayList<TaggedField>();
      for (int i = 0; i < quasiIds.size(); i++) {
        TaggedField taggedField =
            TaggedField.newBuilder()
                .setField(FieldId.newBuilder().setName(quasiIds.get(i)).build())
                .setInfoType(infoTypes.get(i))
                .build();
        taggedFields.add(taggedField);
      }

      // The k-map distribution region can be specified by any ISO-3166-1 region code.
      String regionCode = "US";

      // Configure the privacy metric for the job
      KMapEstimationConfig kmapConfig =
          KMapEstimationConfig.newBuilder()
              .addAllQuasiIds(taggedFields)
              .setRegionCode(regionCode)
              .build();
      PrivacyMetric privacyMetric =
          PrivacyMetric.newBuilder().setKMapEstimationConfig(kmapConfig).build();

      // Create action to publish job status notifications over Google Cloud Pub/Sub
      ProjectTopicName topicName = ProjectTopicName.of(projectId, topicId);
      PublishToPubSub publishToPubSub =
          PublishToPubSub.newBuilder().setTopic(topicName.toString()).build();
      Action action = Action.newBuilder().setPubSub(publishToPubSub).build();

      // Configure the risk analysis job to perform
      RiskAnalysisJobConfig riskAnalysisJobConfig =
          RiskAnalysisJobConfig.newBuilder()
              .setSourceTable(bigQueryTable)
              .setPrivacyMetric(privacyMetric)
              .addActions(action)
              .build();

      // Build the request to be sent by the client
      CreateDlpJobRequest createDlpJobRequest =
          CreateDlpJobRequest.newBuilder()
              .setParent(LocationName.of(projectId, "global").toString())
              .setRiskJob(riskAnalysisJobConfig)
              .build();

      // Send the request to the API using the client
      DlpJob dlpJob = dlpServiceClient.createDlpJob(createDlpJobRequest);

      // Set up a Pub/Sub subscriber to listen on the job completion status
      final SettableApiFuture<Boolean> done = SettableApiFuture.create();

      ProjectSubscriptionName subscriptionName =
          ProjectSubscriptionName.of(projectId, subscriptionId);

      MessageReceiver messageHandler =
          (PubsubMessage pubsubMessage, AckReplyConsumer ackReplyConsumer) -> {
            handleMessage(dlpJob, done, pubsubMessage, ackReplyConsumer);
          };
      Subscriber subscriber = Subscriber.newBuilder(subscriptionName, messageHandler).build();
      subscriber.startAsync();

      // Wait for job completion semi-synchronously
      // For long jobs, consider using a truly asynchronous execution model such as Cloud Functions
      try {
        done.get(15, TimeUnit.MINUTES);
      } catch (TimeoutException e) {
        System.out.println("Job was not completed after 15 minutes.");
        return;
      } finally {
        subscriber.stopAsync();
        subscriber.awaitTerminated();
      }

      // Build a request to get the completed job
      GetDlpJobRequest getDlpJobRequest =
          GetDlpJobRequest.newBuilder().setName(dlpJob.getName()).build();

      // Retrieve completed job status
      DlpJob completedJob = dlpServiceClient.getDlpJob(getDlpJobRequest);
      System.out.println("Job status: " + completedJob.getState());
      System.out.println("Job name: " + dlpJob.getName());

      // Get the result and parse through and process the information
      KMapEstimationResult kmapResult = completedJob.getRiskDetails().getKMapEstimationResult();

      for (KMapEstimationHistogramBucket result : kmapResult.getKMapEstimationHistogramList()) {
        System.out.printf(
            "\tAnonymity range: [%d, %d]\n", result.getMinAnonymity(), result.getMaxAnonymity());
        System.out.printf("\tSize: %d\n", result.getBucketSize());

        for (KMapEstimationQuasiIdValues valueBucket : result.getBucketValuesList()) {
          List<String> quasiIdValues =
              valueBucket.getQuasiIdsValuesList().stream()
                  .map(
                      value -> {
                        String s = value.toString();
                        return s.substring(s.indexOf(':') + 1).trim();
                      })
                  .collect(Collectors.toList());

          System.out.printf("\tValues: {%s}\n", String.join(", ", quasiIdValues));
          System.out.printf(
              "\tEstimated k-map anonymity: %d\n", valueBucket.getEstimatedAnonymity());
        }
      }
    }
  }

  // handleMessage injects the job and settableFuture into the message reciever interface
  private static void handleMessage(
      DlpJob job,
      SettableApiFuture<Boolean> done,
      PubsubMessage pubsubMessage,
      AckReplyConsumer ackReplyConsumer) {
    String messageAttribute = pubsubMessage.getAttributesMap().get("DlpJobName");
    if (job.getName().equals(messageAttribute)) {
      done.set(true);
      ackReplyConsumer.ack();
    } else {
      ackReplyConsumer.nack();
    }
  }
}

Node.js

Per scoprire come installare e utilizzare la libreria client per Sensitive Data Protection, consulta Librerie client di Sensitive Data Protection.

Per autenticarti a Sensitive Data Protection, configura le Credenziali predefinite dell'applicazione. Per ulteriori informazioni, consulta Configurare l'autenticazione per un ambiente di sviluppo locale.

// Import the Google Cloud client libraries
const DLP = require('@google-cloud/dlp');
const {PubSub} = require('@google-cloud/pubsub');

// Instantiates clients
const dlp = new DLP.DlpServiceClient();
const pubsub = new PubSub();

// The project ID to run the API call under
// const projectId = 'my-project';

// The project ID the table is stored under
// This may or (for public datasets) may not equal the calling project ID
// const tableProjectId = 'my-project';

// The ID of the dataset to inspect, e.g. 'my_dataset'
// const datasetId = 'my_dataset';

// The ID of the table to inspect, e.g. 'my_table'
// const tableId = 'my_table';

// The name of the Pub/Sub topic to notify once the job completes
// TODO(developer): create a Pub/Sub topic to use for this
// const topicId = 'MY-PUBSUB-TOPIC'

// The name of the Pub/Sub subscription to use when listening for job
// completion notifications
// TODO(developer): create a Pub/Sub subscription to use for this
// const subscriptionId = 'MY-PUBSUB-SUBSCRIPTION'

// The ISO 3166-1 region code that the data is representative of
// Can be omitted if using a region-specific infoType (such as US_ZIP_5)
// const regionCode = 'USA';

// A set of columns that form a composite key ('quasi-identifiers'), and
// optionally their reidentification distributions
// const quasiIds = [{ field: { name: 'age' }, infoType: { name: 'AGE' }}];
async function kMapEstimationAnalysis() {
  const sourceTable = {
    projectId: tableProjectId,
    datasetId: datasetId,
    tableId: tableId,
  };

  // Construct request for creating a risk analysis job
  const request = {
    parent: `projects/${projectId}/locations/global`,
    riskJob: {
      privacyMetric: {
        kMapEstimationConfig: {
          quasiIds: quasiIds,
          regionCode: regionCode,
        },
      },
      sourceTable: sourceTable,
      actions: [
        {
          pubSub: {
            topic: `projects/${projectId}/topics/${topicId}`,
          },
        },
      ],
    },
  };
  // Create helper function for unpacking values
  const getValue = obj => obj[Object.keys(obj)[0]];

  // Run risk analysis job
  const [topicResponse] = await pubsub.topic(topicId).get();
  const subscription = await topicResponse.subscription(subscriptionId);
  const [jobsResponse] = await dlp.createDlpJob(request);
  const jobName = jobsResponse.name;
  console.log(`Job created. Job name: ${jobName}`);

  // Watch the Pub/Sub topic until the DLP job finishes
  await new Promise((resolve, reject) => {
    const messageHandler = message => {
      if (message.attributes && message.attributes.DlpJobName === jobName) {
        message.ack();
        subscription.removeListener('message', messageHandler);
        subscription.removeListener('error', errorHandler);
        resolve(jobName);
      } else {
        message.nack();
      }
    };

    const errorHandler = err => {
      subscription.removeListener('message', messageHandler);
      subscription.removeListener('error', errorHandler);
      reject(err);
    };

    subscription.on('message', messageHandler);
    subscription.on('error', errorHandler);
  });
  setTimeout(() => {
    console.log(' Waiting for DLP job to fully complete');
  }, 500);
  const [job] = await dlp.getDlpJob({name: jobName});

  const histogramBuckets =
    job.riskDetails.kMapEstimationResult.kMapEstimationHistogram;

  histogramBuckets.forEach((histogramBucket, histogramBucketIdx) => {
    console.log(`Bucket ${histogramBucketIdx}:`);
    console.log(
      `  Anonymity range: [${histogramBucket.minAnonymity}, ${histogramBucket.maxAnonymity}]`
    );
    console.log(`  Size: ${histogramBucket.bucketSize}`);
    histogramBucket.bucketValues.forEach(valueBucket => {
      const values = valueBucket.quasiIdsValues.map(value => getValue(value));
      console.log(`    Values: ${values.join(' ')}`);
      console.log(
        `    Estimated k-map anonymity: ${valueBucket.estimatedAnonymity}`
      );
    });
  });
}

await kMapEstimationAnalysis();

PHP

Per scoprire come installare e utilizzare la libreria client per Sensitive Data Protection, consulta Librerie client di Sensitive Data Protection.

Per autenticarti a Sensitive Data Protection, configura le Credenziali predefinite dell'applicazione. Per ulteriori informazioni, consulta Configurare l'autenticazione per un ambiente di sviluppo locale.

use Exception;
use Google\Cloud\Dlp\V2\Action;
use Google\Cloud\Dlp\V2\Action\PublishToPubSub;
use Google\Cloud\Dlp\V2\BigQueryTable;
use Google\Cloud\Dlp\V2\Client\DlpServiceClient;
use Google\Cloud\Dlp\V2\CreateDlpJobRequest;
use Google\Cloud\Dlp\V2\DlpJob\JobState;
use Google\Cloud\Dlp\V2\FieldId;
use Google\Cloud\Dlp\V2\GetDlpJobRequest;
use Google\Cloud\Dlp\V2\InfoType;
use Google\Cloud\Dlp\V2\PrivacyMetric;
use Google\Cloud\Dlp\V2\PrivacyMetric\KMapEstimationConfig;
use Google\Cloud\Dlp\V2\PrivacyMetric\KMapEstimationConfig\TaggedField;
use Google\Cloud\Dlp\V2\RiskAnalysisJobConfig;
use Google\Cloud\PubSub\PubSubClient;

/**
 * Computes the k-map risk estimation of a column set in a Google BigQuery table.
 *
 * @param string   $callingProjectId  The project ID to run the API call under
 * @param string   $dataProjectId     The project ID containing the target Datastore
 * @param string   $topicId           The name of the Pub/Sub topic to notify once the job completes
 * @param string   $subscriptionId    The name of the Pub/Sub subscription to use when listening for job
 * @param string   $datasetId         The ID of the dataset to inspect
 * @param string   $tableId           The ID of the table to inspect
 * @param string   $regionCode        The ISO 3166-1 region code that the data is representative of
 * @param string[] $quasiIdNames      Array columns that form a composite key (quasi-identifiers)
 * @param string[] $infoTypes         Array of infoTypes corresponding to the chosen quasi-identifiers
 */
function k_map(
    string $callingProjectId,
    string $dataProjectId,
    string $topicId,
    string $subscriptionId,
    string $datasetId,
    string $tableId,
    string $regionCode,
    array $quasiIdNames,
    array $infoTypes
): void {
    // Instantiate a client.
    $dlp = new DlpServiceClient();
    $pubsub = new PubSubClient();
    $topic = $pubsub->topic($topicId);

    // Verify input
    if (count($infoTypes) != count($quasiIdNames)) {
        throw new Exception('Number of infoTypes and number of quasi-identifiers must be equal!');
    }

    // Map infoTypes to quasi-ids
    $quasiIdObjects = array_map(function ($quasiId, $infoType) {
        $quasiIdField = (new FieldId())
            ->setName($quasiId);

        $quasiIdType = (new InfoType())
            ->setName($infoType);

        $quasiIdObject = (new TaggedField())
            ->setInfoType($quasiIdType)
            ->setField($quasiIdField);

        return $quasiIdObject;
    }, $quasiIdNames, $infoTypes);

    // Construct analysis config
    $statsConfig = (new KMapEstimationConfig())
        ->setQuasiIds($quasiIdObjects)
        ->setRegionCode($regionCode);

    $privacyMetric = (new PrivacyMetric())
        ->setKMapEstimationConfig($statsConfig);

    // Construct items to be analyzed
    $bigqueryTable = (new BigQueryTable())
        ->setProjectId($dataProjectId)
        ->setDatasetId($datasetId)
        ->setTableId($tableId);

    // Construct the action to run when job completes
    $pubSubAction = (new PublishToPubSub())
        ->setTopic($topic->name());

    $action = (new Action())
        ->setPubSub($pubSubAction);

    // Construct risk analysis job config to run
    $riskJob = (new RiskAnalysisJobConfig())
        ->setPrivacyMetric($privacyMetric)
        ->setSourceTable($bigqueryTable)
        ->setActions([$action]);

    // Listen for job notifications via an existing topic/subscription.
    $subscription = $topic->subscription($subscriptionId);

    // Submit request
    $parent = "projects/$callingProjectId/locations/global";
    $createDlpJobRequest = (new CreateDlpJobRequest())
        ->setParent($parent)
        ->setRiskJob($riskJob);
    $job = $dlp->createDlpJob($createDlpJobRequest);

    // Poll Pub/Sub using exponential backoff until job finishes
    // Consider using an asynchronous execution model such as Cloud Functions
    $attempt = 1;
    $startTime = time();
    do {
        foreach ($subscription->pull() as $message) {
            if (
                isset($message->attributes()['DlpJobName']) &&
                $message->attributes()['DlpJobName'] === $job->getName()
            ) {
                $subscription->acknowledge($message);
                // Get the updated job. Loop to avoid race condition with DLP API.
                do {
                    $getDlpJobRequest = (new GetDlpJobRequest())
                        ->setName($job->getName());
                    $job = $dlp->getDlpJob($getDlpJobRequest);
                } while ($job->getState() == JobState::RUNNING);
                break 2; // break from parent do while
            }
        }
        print('Waiting for job to complete' . PHP_EOL);
        // Exponential backoff with max delay of 60 seconds
        sleep(min(60, pow(2, ++$attempt)));
    } while (time() - $startTime < 600); // 10 minute timeout

    // Print finding counts
    printf('Job %s status: %s' . PHP_EOL, $job->getName(), JobState::name($job->getState()));
    switch ($job->getState()) {
        case JobState::DONE:
            $histBuckets = $job->getRiskDetails()->getKMapEstimationResult()->getKMapEstimationHistogram();

            foreach ($histBuckets as $bucketIndex => $histBucket) {
                // Print bucket stats
                printf('Bucket %s:' . PHP_EOL, $bucketIndex);
                printf(
                    '  Anonymity range: [%s, %s]' . PHP_EOL,
                    $histBucket->getMinAnonymity(),
                    $histBucket->getMaxAnonymity()
                );
                printf('  Size: %s' . PHP_EOL, $histBucket->getBucketSize());

                // Print bucket values
                foreach ($histBucket->getBucketValues() as $percent => $valueBucket) {
                    printf(
                        '  Estimated k-map anonymity: %s' . PHP_EOL,
                        $valueBucket->getEstimatedAnonymity()
                    );

                    // Pretty-print quasi-ID values
                    print('  Values: ' . PHP_EOL);
                    foreach ($valueBucket->getQuasiIdsValues() as $index => $value) {
                        print('    ' . $value->serializeToJsonString() . PHP_EOL);
                    }
                }
            }
            break;
        case JobState::FAILED:
            printf('Job %s had errors:' . PHP_EOL, $job->getName());
            $errors = $job->getErrors();
            foreach ($errors as $error) {
                var_dump($error->getDetails());
            }
            break;
        case JobState::PENDING:
            print('Job has not completed. Consider a longer timeout or an asynchronous execution model' . PHP_EOL);
            break;
        default:
            print('Unexpected job state. Most likely, the job is either running or has not yet started.');
    }
}

Python

Per scoprire come installare e utilizzare la libreria client per Sensitive Data Protection, consulta Librerie client di Sensitive Data Protection.

Per autenticarti a Sensitive Data Protection, configura le Credenziali predefinite dell'applicazione. Per ulteriori informazioni, consulta Configurare l'autenticazione per un ambiente di sviluppo locale.

import concurrent.futures
from typing import List

import google.cloud.dlp
from google.cloud.dlp_v2 import types
import google.cloud.pubsub


def k_map_estimate_analysis(
    project: str,
    table_project_id: str,
    dataset_id: str,
    table_id: str,
    topic_id: str,
    subscription_id: str,
    quasi_ids: List[str],
    info_types: List[str],
    region_code: str = "US",
    timeout: int = 300,
) -> None:
    """Uses the Data Loss Prevention API to compute the k-map risk estimation
        of a column set in a Google BigQuery table.
    Args:
        project: The Google Cloud project id to use as a parent resource.
        table_project_id: The Google Cloud project id where the BigQuery table
            is stored.
        dataset_id: The id of the dataset to inspect.
        table_id: The id of the table to inspect.
        topic_id: The name of the Pub/Sub topic to notify once the job
            completes.
        subscription_id: The name of the Pub/Sub subscription to use when
            listening for job completion notifications.
        quasi_ids: A set of columns that form a composite key and optionally
            their re-identification distributions.
        info_types: Type of information of the quasi_id in order to provide a
            statistical model of population.
        region_code: The ISO 3166-1 region code that the data is representative
            of. Can be omitted if using a region-specific infoType (such as
            US_ZIP_5)
        timeout: The number of seconds to wait for a response from the API.

    Returns:
        None; the response from the API is printed to the terminal.
    """

    # Create helper function for unpacking values
    def get_values(obj: types.Value) -> int:
        return int(obj.integer_value)

    # Instantiate a client.
    dlp = google.cloud.dlp_v2.DlpServiceClient()

    # Convert the project id into full resource ids.
    topic = google.cloud.pubsub.PublisherClient.topic_path(project, topic_id)
    parent = f"projects/{project}/locations/global"

    # Location info of the BigQuery table.
    source_table = {
        "project_id": table_project_id,
        "dataset_id": dataset_id,
        "table_id": table_id,
    }

    # Check that numbers of quasi-ids and info types are equal
    if len(quasi_ids) != len(info_types):
        raise ValueError(
            """Number of infoTypes and number of quasi-identifiers
                            must be equal!"""
        )

    # Convert quasi id list to Protobuf type
    def map_fields(quasi_id: str, info_type: str) -> dict:
        return {"field": {"name": quasi_id}, "info_type": {"name": info_type}}

    quasi_ids = map(map_fields, quasi_ids, info_types)

    # Tell the API where to send a notification when the job is complete.
    actions = [{"pub_sub": {"topic": topic}}]

    # Configure risk analysis job
    # Give the name of the numeric column to compute risk metrics for
    risk_job = {
        "privacy_metric": {
            "k_map_estimation_config": {
                "quasi_ids": quasi_ids,
                "region_code": region_code,
            }
        },
        "source_table": source_table,
        "actions": actions,
    }

    # Call API to start risk analysis job
    operation = dlp.create_dlp_job(request={"parent": parent, "risk_job": risk_job})

    def callback(message: google.cloud.pubsub_v1.subscriber.message.Message) -> None:
        if message.attributes["DlpJobName"] == operation.name:
            # This is the message we're looking for, so acknowledge it.
            message.ack()

            # Now that the job is done, fetch the results and print them.
            job = dlp.get_dlp_job(request={"name": operation.name})
            print(f"Job name: {job.name}")
            histogram_buckets = (
                job.risk_details.k_map_estimation_result.k_map_estimation_histogram
            )
            # Print bucket stats
            for i, bucket in enumerate(histogram_buckets):
                print(f"Bucket {i}:")
                print(
                    "   Anonymity range: [{}, {}]".format(
                        bucket.min_anonymity, bucket.max_anonymity
                    )
                )
                print(f"   Size: {bucket.bucket_size}")
                for value_bucket in bucket.bucket_values:
                    print(
                        "   Values: {}".format(
                            map(get_values, value_bucket.quasi_ids_values)
                        )
                    )
                    print(
                        "   Estimated k-map anonymity: {}".format(
                            value_bucket.estimated_anonymity
                        )
                    )
            subscription.set_result(None)
        else:
            # This is not the message we're looking for.
            message.drop()

    # Create a Pub/Sub client and find the subscription. The subscription is
    # expected to already be listening to the topic.
    subscriber = google.cloud.pubsub.SubscriberClient()
    subscription_path = subscriber.subscription_path(project, subscription_id)
    subscription = subscriber.subscribe(subscription_path, callback)

    try:
        subscription.result(timeout=timeout)
    except concurrent.futures.TimeoutError:
        print(
            "No event received before the timeout. Please verify that the "
            "subscription provided is subscribed to the topic provided."
        )
        subscription.close()

C#

Per scoprire come installare e utilizzare la libreria client per Sensitive Data Protection, consulta Librerie client di Sensitive Data Protection.

Per autenticarti a Sensitive Data Protection, configura le Credenziali predefinite dell'applicazione. Per ulteriori informazioni, consulta Configurare l'autenticazione per un ambiente di sviluppo locale.


using Google.Api.Gax.ResourceNames;
using Google.Cloud.Dlp.V2;
using Google.Cloud.PubSub.V1;
using Newtonsoft.Json;
using System;
using System.Collections.Generic;
using System.Linq;
using System.Threading;
using System.Threading.Tasks;
using static Google.Cloud.Dlp.V2.Action.Types;
using static Google.Cloud.Dlp.V2.PrivacyMetric.Types;
using static Google.Cloud.Dlp.V2.PrivacyMetric.Types.KMapEstimationConfig.Types;

public class RiskAnalysisCreateKMap
{
    public static object KMap(
        string callingProjectId,
        string tableProjectId,
        string datasetId,
        string tableId,
        string topicId,
        string subscriptionId,
        IEnumerable<FieldId> quasiIds,
        IEnumerable<InfoType> infoTypes,
        string regionCode)
    {
        var dlp = DlpServiceClient.Create();

        // Construct + submit the job
        var kmapEstimationConfig = new KMapEstimationConfig
        {
            QuasiIds =
                {
                    quasiIds.Zip(
                        infoTypes,
                        (Field, InfoType) => new TaggedField
                        {
                            Field = Field,
                            InfoType = InfoType
                        }
                    )
                },
            RegionCode = regionCode
        };

        var config = new RiskAnalysisJobConfig()
        {
            PrivacyMetric = new PrivacyMetric
            {
                KMapEstimationConfig = kmapEstimationConfig
            },
            SourceTable = new BigQueryTable
            {
                ProjectId = tableProjectId,
                DatasetId = datasetId,
                TableId = tableId
            },
            Actions =
            {
                new Google.Cloud.Dlp.V2.Action
                {
                    PubSub = new PublishToPubSub
                    {
                        Topic = $"projects/{callingProjectId}/topics/{topicId}"
                    }
                }
            }
        };

        var submittedJob = dlp.CreateDlpJob(
            new CreateDlpJobRequest
            {
                ParentAsProjectName = new ProjectName(callingProjectId),
                RiskJob = config
            });

        // Listen to pub/sub for the job
        var subscriptionName = new SubscriptionName(
            callingProjectId,
            subscriptionId);
        var subscriber = SubscriberClient.CreateAsync(
            subscriptionName).Result;

        // SimpleSubscriber runs your message handle function on multiple
        // threads to maximize throughput.
        var done = new ManualResetEventSlim(false);
        subscriber.StartAsync((PubsubMessage message, CancellationToken cancel) =>
        {
            if (message.Attributes["DlpJobName"] == submittedJob.Name)
            {
                Thread.Sleep(500); // Wait for DLP API results to become consistent
                done.Set();
                return Task.FromResult(SubscriberClient.Reply.Ack);
            }
            else
            {
                return Task.FromResult(SubscriberClient.Reply.Nack);
            }
        });

        done.Wait(TimeSpan.FromMinutes(10)); // 10 minute timeout; may not work for large jobs
        subscriber.StopAsync(CancellationToken.None).Wait();

        // Process results
        var resultJob = dlp.GetDlpJob(new GetDlpJobRequest
        {
            DlpJobName = DlpJobName.Parse(submittedJob.Name)
        });

        var result = resultJob.RiskDetails.KMapEstimationResult;

        for (var histogramIdx = 0; histogramIdx < result.KMapEstimationHistogram.Count; histogramIdx++)
        {
            var histogramValue = result.KMapEstimationHistogram[histogramIdx];
            Console.WriteLine($"Bucket {histogramIdx}");
            Console.WriteLine($"  Anonymity range: [{histogramValue.MinAnonymity}, {histogramValue.MaxAnonymity}].");
            Console.WriteLine($"  Size: {histogramValue.BucketSize}");

            foreach (var datapoint in histogramValue.BucketValues)
            {
                // 'UnpackValue(x)' is a prettier version of 'x.toString()'
                Console.WriteLine($"    Values: [{String.Join(',', datapoint.QuasiIdsValues.Select(x => UnpackValue(x)))}]");
                Console.WriteLine($"    Estimated k-map anonymity: {datapoint.EstimatedAnonymity}");
            }
        }

        return 0;
    }

    public static string UnpackValue(Value protoValue)
    {
        var jsonValue = JsonConvert.DeserializeObject<Dictionary<string, object>>(protoValue.ToString());
        return jsonValue.Values.ElementAt(0).ToString();
    }
}

Visualizzazione dei risultati dei job k-map

Per recuperare i risultati del job di analisi del rischio delle mappe k utilizzando l'API REST, invia la seguente richiesta GET alla risorsa projects.dlpJobs. Sostituisci PROJECT_ID con l'ID del tuo progetto e JOB_ID con l'identificatore del job per cui vuoi ottenere i risultati. L'ID job è stato restituito quando hai avviato il job e può essere recuperato anche elencando tutti i job.

GET https://dlp.googleapis.com/v2/projects/PROJECT_ID/dlpJobs/JOB_ID

La richiesta restituisce un oggetto JSON contenente un'istanza del job. I risultati dell'analisi si trovano all'interno della chiave "riskDetails", in un oggetto AnalyzeDataSourceRiskDetails. Per ulteriori informazioni, consulta il riferimento all'API per la risorsa DlpJob.

Passaggi successivi

  • Scopri come calcolare il valore di k per un set di dati.
  • Scopri come calcolare il valore l-diversity per un set di dati.
  • Scopri come calcolare il valore δ-presence per un set di dati.