Work with JSON data

PostgreSQL interface note: The JSON data type is not supported in PostgreSQL-dialect databases. Use the JSONB data type for PostgreSQL-dialect databases.

This page describes how to work with JSON using Spanner.

The JSON data type is a semi-structured data type used for holding JSON (JavaScript Object Notation) data. The specifications for the JSON format are described in RFC 7159.

JSON is useful to supplement a relational schema for data that is sparse or has a loosely-defined or changing structure. However, the query optimizer relies on the relational model to efficiently filter, join, aggregate, and sort at scale. Queries over JSON will have fewer built-in optimizations and fewer affordances to inspect and tune performance.

Specifications

Spanner JSON type stores a normalized representation of the input JSON document.

JSON can be nested to a maximum of 80 levels.
Whitespace is not preserved.
Comments are not supported. Transactions or queries with comments will fail.
Members of a JSON object are sorted lexicographically.
JSON array elements have their order preserved.
If a JSON object has duplicate keys, only the first one is preserved.
Primitive types (string, boolean, number, and null) have their type and value preserved.
- String type values are preserved exactly.
- Number type values are preserved, but may have their textual representation changed as a result of the normalization process. For example, an input number of 10000 may have a normalized representation of 1e+4. Number value preservation semantics are as follows:
  - Signed integers in the range of [INT64_MIN, INT64_MAX] are preserved.
  - Unsigned integers in the range of [0, UINT64_MAX] are preserved.
  - Double values that can be roundtripped from string to double to string without precision loss are preserved. If a double value cannot round trip in this manner, the transaction or query fails.
    - For example, SELECT JSON '2.2412421353246235436' fails.
    - A functional workaround is PARSE_JSON('2.2412421353246235436', wide_number_mode=>'round'), which returns JSON '2.2412421353246237'.

The maximum permitted size of the normalized document is 2621440 of UTF-8 characters.

Nullability

JSON null values are treated as SQL non-NULL.

For example:

SELECT (JSON '{"a":null}').a IS NULL; -- Returns FALSE
SELECT (JSON '{"a":null}').b IS NULL; -- Returns TRUE

SELECT JSON_QUERY(JSON '{"a":null}', "$.a"); -- Returns a JSON 'null'
SELECT JSON_QUERY(JSON '{"a":null}', "$.b"); -- Returns a SQL NULL

Encoding

JSON documents must be encoded in UTF-8. Transactions or queries with JSON documents encoded in other formats return an error.

Create a table with JSON columns

A JSON column can be added to a table when the table is created. JSON type values can be nullable.

CREATE TABLE Venues (
  VenueId   INT64 NOT NULL,
  VenueName  STRING(1024),
  VenueAddress STRING(1024),
  VenueFeatures JSON,
  DateOpened  DATE,
) PRIMARY KEY(VenueId);

Add and remove JSON columns from existing tables

A JSON column can also be added to and dropped from existing tables.

ALTER TABLE Venues ADD COLUMN VenueDetails JSON;
ALTER TABLE Venues DROP COLUMN VenueDetails;

The following sample shows how to add a JSON column called VenueDetails to the Venues table using the Spanner client libraries.

C++

void AddJsonColumn(google::cloud::spanner_admin::DatabaseAdminClient client,
                   std::string const& project_id,
                   std::string const& instance_id,
                   std::string const& database_id) {
  google::cloud::spanner::Database database(project_id, instance_id,
                                            database_id);
  auto metadata = client
                      .UpdateDatabaseDdl(database.FullName(), {R"""(
                        ALTER TABLE Venues ADD COLUMN VenueDetails JSON)"""})
                      .get();
  if (!metadata) throw std::move(metadata).status();
  std::cout << "`Venues` table altered, new DDL:\n" << metadata->DebugString();
}

C#


using Google.Cloud.Spanner.Data;
using System;
using System.Threading.Tasks;

public class AddJsonColumnAsyncSample
{
    public async Task AddJsonColumnAsync(string projectId, string instanceId, string databaseId)
    {
        string connectionString = $"Data Source=projects/{projectId}/instances/{instanceId}/databases/{databaseId}";
        string alterStatement = "ALTER TABLE Venues ADD COLUMN VenueDetails JSON";

        using var connection = new SpannerConnection(connectionString);
        using var updateCmd = connection.CreateDdlCommand(alterStatement);
        await updateCmd.ExecuteNonQueryAsync();
        Console.WriteLine("Added the VenueDetails column.");
    }
}

Go


import (
	"context"
	"fmt"
	"io"
	"regexp"

	database "cloud.google.com/go/spanner/admin/database/apiv1"
	adminpb "google.golang.org/genproto/googleapis/spanner/admin/database/v1"
)

// addJsonColumn creates a column in the database of type JSON
func addJsonColumn(w io.Writer, db string) error {
	// db = `projects/<project>/instances/<instance-id>/database/<database-id>`
	matches := regexp.MustCompile("^(.*)/databases/(.*)$").FindStringSubmatch(db)
	if matches == nil || len(matches) != 3 {
		return fmt.Errorf("addJsonColumn: invalid database id %s", db)
	}

	ctx := context.Background()
	adminClient, err := database.NewDatabaseAdminClient(ctx)
	if err != nil {
		return err
	}
	defer adminClient.Close()

	op, err := adminClient.UpdateDatabaseDdl(ctx, &adminpb.UpdateDatabaseDdlRequest{
		Database: db,
		Statements: []string{
			"ALTER TABLE Venues ADD COLUMN VenueDetails JSON",
		},
	})
	if err != nil {
		return err
	}
	if err := op.Wait(ctx); err != nil {
		return err
	}
	fmt.Fprintf(w, "Added VenueDetails column\n")
	return nil
}

Java


import com.google.cloud.spanner.Spanner;
import com.google.cloud.spanner.SpannerOptions;
import com.google.cloud.spanner.admin.database.v1.DatabaseAdminClient;
import com.google.common.collect.ImmutableList;
import com.google.spanner.admin.database.v1.DatabaseName;
import java.util.concurrent.ExecutionException;

class AddJsonColumnSample {

  static void addJsonColumn() throws InterruptedException, ExecutionException {
    // TODO(developer): Replace these variables before running the sample.
    String projectId = "my-project";
    String instanceId = "my-instance";
    String databaseId = "my-database";

    addJsonColumn(projectId, instanceId, databaseId);
  }

  static void addJsonColumn(String projectId, String instanceId, String databaseId)
      throws InterruptedException, ExecutionException {
    try (Spanner spanner =
        SpannerOptions.newBuilder()
            .setProjectId(projectId)
            .build()
            .getService();
        DatabaseAdminClient databaseAdminClient = spanner.createDatabaseAdminClient()) {
      // Wait for the operation to finish.
      // This will throw an ExecutionException if the operation fails.
      databaseAdminClient.updateDatabaseDdlAsync(
          DatabaseName.of(projectId, instanceId, databaseId),
          ImmutableList.of("ALTER TABLE Venues ADD COLUMN VenueDetails JSON")).get();
      System.out.printf("Successfully added column `VenueDetails`%n");
    }
  }
}

Node.js


/**
 * TODO(developer): Uncomment the following lines before running the sample.
 */
// const projectId = 'my-project-id';
// const instanceId = 'my-instance';
// const databaseId = 'my-database';

// Imports the Google Cloud client library
const {Spanner} = require('@google-cloud/spanner');

// creates a client
const spanner = new Spanner({
  projectId: projectId,
});

const databaseAdminClient = spanner.getDatabaseAdminClient();

const request = ['ALTER TABLE Venues ADD COLUMN VenueDetails JSON'];

// Alter existing table to add a column.
const [operation] = await databaseAdminClient.updateDatabaseDdl({
  database: databaseAdminClient.databasePath(
    projectId,
    instanceId,
    databaseId
  ),
  statements: request,
});

console.log(`Waiting for operation on ${databaseId} to complete...`);

await operation.promise();

console.log(
  `Added VenueDetails column to Venues table in database ${databaseId}.`
);

PHP

use Google\Cloud\Spanner\Admin\Database\V1\Client\DatabaseAdminClient;
use Google\Cloud\Spanner\Admin\Database\V1\UpdateDatabaseDdlRequest;

/**
 * Adds a JSON column to a table.
 * Example:
 * ```
 * add_json_column($projectId, $instanceId, $databaseId);
 * ```
 *
 * @param string $projectId The Google Cloud project ID.
 * @param string $instanceId The Spanner instance ID.
 * @param string $databaseId The Spanner database ID.
 */
function add_json_column(string $projectId, string $instanceId, string $databaseId): void
{
    $databaseAdminClient = new DatabaseAdminClient();
    $databaseName = DatabaseAdminClient::databaseName($projectId, $instanceId, $databaseId);

    $request = new UpdateDatabaseDdlRequest([
        'database' => $databaseName,
        'statements' => ['ALTER TABLE Venues ADD COLUMN VenueDetails JSON']
    ]);

    $operation = $databaseAdminClient->updateDatabaseDdl($request);

    print('Waiting for operation to complete...' . PHP_EOL);
    $operation->pollUntilComplete();

    printf('Added VenueDetails as a JSON column in Venues table' . PHP_EOL);
}

Python

def add_json_column(instance_id, database_id):
    """Adds a new JSON column to the Venues table in the example database."""
    # instance_id = "your-spanner-instance"
    # database_id = "your-spanner-db-id"

    from google.cloud.spanner_admin_database_v1.types import \
        spanner_database_admin

    spanner_client = spanner.Client()
    database_admin_api = spanner_client.database_admin_api

    request = spanner_database_admin.UpdateDatabaseDdlRequest(
        database=database_admin_api.database_path(
            spanner_client.project, instance_id, database_id
        ),
        statements=["ALTER TABLE Venues ADD COLUMN VenueDetails JSON"],
    )

    operation = database_admin_api.update_database_ddl(request)

    print("Waiting for operation to complete...")
    operation.result(OPERATION_TIMEOUT_SECONDS)

    print(
        'Altered table "Venues" on database {} on instance {}.'.format(
            database_id, instance_id
        )
    )

Ruby

# project_id  = "Your Google Cloud project ID"
# instance_id = "Your Spanner instance ID"
# database_id = "Your Spanner database ID"

require "google/cloud/spanner"
require "google/cloud/spanner/admin/database"

db_admin_client = Google::Cloud::Spanner::Admin::Database.database_admin

database_path = db_admin_client.database_path project: project_id,
                                              instance: instance_id,
                                              database: database_id

statements = ["ALTER TABLE Venues ADD COLUMN VenueDetails JSON"]
job = db_admin_client.update_database_ddl database: database_path,
                                          statements: statements
job.wait_until_done!

puts "Added VenueDetails column to Venues table in database #{database_id}"