Aggregation queries

Stay organized with collections Save and categorize content based on your preferences.

An aggregation query processes the data from multiple indexed entities to return a single summary value. Firestore in Datastore mode supports the count() aggregation query. Read this page to learn how to use the count() aggregation query.

count() aggregation

Use the count() aggregation to return the total number of indexed entities that match a given query. This simplifies your application code and costs less than counting by fetching each entity.

For example, this count() aggregation returns the total number of entities in a kind.

Java
import static com.google.cloud.datastore.aggregation.Aggregation.count;

import com.google.cloud.datastore.AggregationQuery;
import com.google.cloud.datastore.AggregationResult;
import com.google.cloud.datastore.Datastore;
import com.google.cloud.datastore.DatastoreOptions;
import com.google.cloud.datastore.Entity;
import com.google.cloud.datastore.EntityQuery;
import com.google.cloud.datastore.Key;
import com.google.cloud.datastore.Query;
import com.google.common.collect.Iterables;

public class CountAggregationOnKind {
  // Instantiates a client.
  private static final Datastore datastore = DatastoreOptions.getDefaultInstance().getService();

  // The kind for the new entity.
  private static final String kind = "Task";

  // Setting up Tasks in database
  private static void setUpTasks() {
    Key task1Key = datastore.newKeyFactory().setKind(kind).newKey("task1");
    Key task2Key = datastore.newKeyFactory().setKind(kind).newKey("task2");
    Key task3Key = datastore.newKeyFactory().setKind(kind).newKey("task3");

    // Save all the tasks.
    datastore.put(
        Entity.newBuilder(task1Key).set("done", true).build(),
        Entity.newBuilder(task2Key).set("done", false).build(),
        Entity.newBuilder(task3Key).set("done", true).build());
  }

  // Accessing aggregation result by the generated alias.
  private static void usageWithGeneratedAlias() {
    EntityQuery selectAllTasks = Query.newEntityQueryBuilder().setKind(kind).build();
    // Creating an aggregation query to get the count of all tasks.
    AggregationQuery allTasksCountQuery =
        Query.newAggregationQueryBuilder().over(selectAllTasks).addAggregation(count()).build();
    // Executing aggregation query.
    AggregationResult aggregationResult =
        Iterables.getOnlyElement(datastore.runAggregation(allTasksCountQuery));

    System.out.printf(
        "Total tasks (accessible from default alias) is %d",
        aggregationResult.get("property_1")); // 3
  }

  // Accessing aggregation result by the provided custom alias.
  private static void usageWithCustomAlias() {
    EntityQuery selectAllTasks = Query.newEntityQueryBuilder().setKind(kind).build();
    // Creating an aggregation query to get the count of all tasks.
    AggregationQuery allTasksCountQuery =
        Query.newAggregationQueryBuilder()
            .over(selectAllTasks)
            // passing 'total_count' as alias in the aggregation query.
            .addAggregation(count().as("total_count"))
            .build();
    // Executing aggregation query.
    AggregationResult aggregationResult =
        Iterables.getOnlyElement(datastore.runAggregation(allTasksCountQuery));

    System.out.printf("Total tasks count is %d", aggregationResult.get("total_count")); // 3
  }

  public static void invoke() {
    setUpTasks();
    usageWithGeneratedAlias();
    usageWithCustomAlias();
  }
}
GQL
AGGREGATE COUNT(*) AS total OVER ( SELECT * AS total FROM tasks )

GQL supports a simplified form of count() queries:

SELECT COUNT(*) AS total FROM tasks

This example uses an optional alias of `AS total`.

The simplified form supports only FROM and WHERE clauses. See the GQL reference for more information.

The count() aggregation takes into account any filters on the query and any limit clauses. For example, the following aggregation returns a count of the number of entities that match the given filters.

Java

import static com.google.cloud.datastore.aggregation.Aggregation.count;

import com.google.cloud.datastore.AggregationQuery;
import com.google.cloud.datastore.AggregationResult;
import com.google.cloud.datastore.Datastore;
import com.google.cloud.datastore.DatastoreOptions;
import com.google.cloud.datastore.Entity;
import com.google.cloud.datastore.EntityQuery;
import com.google.cloud.datastore.Key;
import com.google.cloud.datastore.Query;
import com.google.cloud.datastore.StructuredQuery.PropertyFilter;
import com.google.common.collect.Iterables;

public class CountAggregationWithPropertyFilter {

  public static void invoke() {
    // Instantiates a client.
    Datastore datastore = DatastoreOptions.getDefaultInstance().getService();

    // The kind for the new entity.
    String kind = "Task";

    Key task1Key = datastore.newKeyFactory().setKind(kind).newKey("task1");
    Key task2Key = datastore.newKeyFactory().setKind(kind).newKey("task2");
    Key task3Key = datastore.newKeyFactory().setKind(kind).newKey("task3");

    // Save all the tasks.
    datastore.put(
        Entity.newBuilder(task1Key).set("done", true).build(),
        Entity.newBuilder(task2Key).set("done", false).build(),
        Entity.newBuilder(task3Key).set("done", true).build());

    EntityQuery completedTasks =
        Query.newEntityQueryBuilder()
            .setKind(kind)
            .setFilter(PropertyFilter.eq("done", true))
            .build();
    EntityQuery remainingTasks =
        Query.newEntityQueryBuilder()
            .setKind(kind)
            .setFilter(PropertyFilter.eq("done", false))
            .build();
    // Creating an aggregation query to get the count of all completed tasks.
    AggregationQuery completedTasksCountQuery =
        Query.newAggregationQueryBuilder()
            .over(completedTasks)
            .addAggregation(count().as("total_completed_count"))
            .build();
    // Creating an aggregation query to get the count of all remaining tasks.
    AggregationQuery remainingTasksCountQuery =
        Query.newAggregationQueryBuilder()
            .over(remainingTasks)
            .addAggregation(count().as("total_remaining_count"))
            .build();

    // Executing aggregation query.
    AggregationResult completedTasksCountQueryResult =
        Iterables.getOnlyElement(datastore.runAggregation(completedTasksCountQuery));
    AggregationResult remainingTasksCountQueryResult =
        Iterables.getOnlyElement(datastore.runAggregation(remainingTasksCountQuery));

    System.out.printf(
        "Total completed tasks count is %d",
        completedTasksCountQueryResult.get("total_completed_count")); // 2
    System.out.printf(
        "Total remaining tasks count is %d",
        remainingTasksCountQueryResult.get("total_remaining_count")); // 1
  }
}
GQL
AGGREGATE COUNT(*) OVER ( SELECT * FROM tasks WHERE is_done = false AND tag = 'house')

GQL supports a simplified form of COUNT(*) queries:

SELECT COUNT(*) AS total
FROM tasks
WHERE is_done = false AND tag = 'house'

This example uses an optional alias of `AS total`.

The simplified form supports only FROM and WHERE clauses. See the GQL reference for more information.

This example shows how to count up to a certain value. You can use this to, for example, stop counting at a certain number and inform users that they exceeded that number.

Java

import static com.google.cloud.datastore.aggregation.Aggregation.count;

import com.google.cloud.datastore.AggregationQuery;
import com.google.cloud.datastore.AggregationResult;
import com.google.cloud.datastore.Datastore;
import com.google.cloud.datastore.DatastoreOptions;
import com.google.cloud.datastore.Entity;
import com.google.cloud.datastore.EntityQuery;
import com.google.cloud.datastore.Key;
import com.google.cloud.datastore.Query;
import com.google.common.collect.Iterables;

public class CountAggregationWithLimit {
  public static void invoke() {
    // Instantiates a client.
    Datastore datastore = DatastoreOptions.getDefaultInstance().getService();

    // The kind for the new entity.
    String kind = "Task";

    Key task1Key = datastore.newKeyFactory().setKind(kind).newKey("task1");
    Key task2Key = datastore.newKeyFactory().setKind(kind).newKey("task2");
    Key task3Key = datastore.newKeyFactory().setKind(kind).newKey("task3");

    // Save all the tasks.
    datastore.put(
        Entity.newBuilder(task1Key).set("done", true).build(),
        Entity.newBuilder(task2Key).set("done", false).build(),
        Entity.newBuilder(task3Key).set("done", true).build());

    EntityQuery selectAllTasks = Query.newEntityQueryBuilder().setKind(kind).setLimit(2).build();
    // Creating an aggregation query to get the count of all tasks.
    AggregationQuery allTasksCountQuery =
        Query.newAggregationQueryBuilder()
            .over(selectAllTasks)
            .addAggregation(count().as("at_least"))
            .build();
    // Executing aggregation query.
    AggregationResult limitQueryResult =
        Iterables.getOnlyElement(datastore.runAggregation(allTasksCountQuery));

    System.out.printf("We have at least %d tasks", limitQueryResult.get("at_least")); // 2
  }
}
GQL
AGGREGATE COUNT_UP_TO(1000) OVER ( SELECT * FROM tasks WHERE is_done = false)

GQL supports a simplified form of COUNT_UP_TO() queries:

SELECT COUNT_UP_TO(1000) AS total
FROM tasks
WHERE is_done = false AND tag = 'house'

This example uses an optional alias of `AS total`.

The simplified form supports only FROM and WHERE clauses. See the GQL reference for more information.

Limitations

  • The query you provide to the count() aggregation must meet the restrictions on queries.
  • Entities with array properties are not de-duplicated. Each array value that matches the query adds one to the count.
  • If a count() aggregation cannot resolve within 60 seconds, it returns a DEADLINE_EXCEEDED error. Performance depends on your index configuration and on the size of the dataset.

    If the operation cannot be completed within the 60 second deadline, a possible workaround is to use cursors to merge multiple aggregations.

  • The count() aggregation reads from index entries and counts only indexed properties.

  • Adding an OrderBy clause to the query limits the count to the entities where the sorting property exists.

  • In GQL, the simplified form of count() does not support ORDER BY, LIMIT, or OFFSET clauses.

Pricing

Pricing for count() depends on the number of index entries scanned during the operation. You are billed one entity read for up to 1,000 index entries matched. Subsequent index entries matched cost additional read units. There is a minimum cost of one read unit for every query. For pricing information, see Firestore in Datastore mode pricing.

What's next