Spring Cloud GCP

The Spring Cloud GCP Bill of Materials (BOM) contains the versions of all the dependencies it uses.

If you’re a Maven user, adding the following to your pom.xml file will allow you omit any Spring Cloud GCP dependency version numbers from your configuration. Instead, the version of the BOM you’re using determines the versions of the used dependencies.

Or, if you’re a Gradle user:

See the sections in the README for selecting an available version and Maven repository.

In the following sections, it will be assumed you are using the Spring Cloud GCP BOM and the dependency snippets will not contain versions.

Spring Cloud GCP offers starter dependencies through Maven to easily depend on different modules of the library. Each starter contains all the dependencies and transitive dependencies needed to begin using their corresponding Spring Cloud GCP module.

For example, if you wish to write a Spring application with Cloud Pub/Sub, you would include the spring-cloud-gcp-starter-pubsub dependency in your project. You do not need to include the underlying spring-cloud-gcp-pubsub dependency, because the starter dependency includes it.

A summary of these artifacts are provided below.

Spring Initializr is a tool which generates the scaffolding code for a new Spring Boot project. It handles the work of generating the Maven or Gradle build file so you do not have to manually add the dependencies yourself.

Spring Initializr offers three modules from Spring Cloud GCP that you can use to generate your project.

The easiest way to learn how to use Spring Cloud GCP is to consult the sample applications on Github. Spring Cloud GCP provides sample applications which demonstrate how to use every integration in the library. The table below highlights several samples of the most commonly used integrations in Spring Cloud GCP.

Each sample application demonstrates how to use Spring Cloud GCP libraries in context and how to setup the dependencies for the project. The applications are fully functional and can be deployed to Google Cloud Platform as well. If you are interested, you may consult guides for deploying an application to AppEngine and to Google Kubernetes Engine.

For a more hands-on approach, there are several guides and codelabs to help you get up to speed. These guides provide step-by-step instructions for building an application using Spring Cloud GCP.

Some examples include:

The full collection of Spring codelabs can be found on the Google Developer Codelabs page.

By default, the credentials provided by the Spring Cloud GCP Starter contain scopes for every service supported by Spring Cloud GCP.

The Spring Cloud GCP starter allows you to configure a custom scope list for the provided credentials. To do that, specify a comma-delimited list of Google OAuth2 scopes in the spring.cloud.gcp.credentials.scopes property.

spring.cloud.gcp.credentials.scopes is a comma-delimited list of Google OAuth2 scopes for Google Cloud Platform services that the credentials returned by the provided CredentialsProvider support.

You can also use DEFAULT_SCOPES placeholder as a scope to represent the starters default scopes, and append the additional scopes you need to add.

You can set the content-type of Google Cloud Storage files from their corresponding Resource objects:

Spring Boot starter for Google Cloud SQL registers a CloudSqlEnvironmentPostProcessor that provides a correctly formatted spring.datasource.url property to the environment based on the properties defined above. It also provides defaults for spring.datasource.username and spring.datasource.driver-class-name, which can be overridden. The starter also configures credentials for the JDBC connection based on the properties above.

The user properties and the properties provided by the CloudSqlEnvironmentPostProcessor are then used by Spring Boot to create the DataSource. You can select the type of connection pool (e.g., Tomcat, HikariCP, etc.) by adding their dependency to the classpath.

Using the created DataSource in conjunction with Spring JDBC provides you with a fully configured and operational JdbcTemplate object that you can use to interact with your SQL database. You can connect to your database with as little as a database and instance names.

Currently, Cloud SQL only supports IAM database authentication for PostgreSQL. It allows you to connect to the database using an IAM account, rather than a predefined database username and password. You will need to do the following to enable it:

This section describes options for enabling the integration, specifying the GCP project and credentials, and setting whether the APIs should connect to an emulator for local testing.

This section describes configuration options to customize the behavior of the application’s Pub/Sub publishers and subscribers. Subscriber settings can be either global or subscription-specific.

The Pub/Sub API uses the GRPC protocol to send API requests to the Pub/Sub service. This section describes configuration options for customizing the GRPC behavior.

Subscription-specific Configuration

If you are using Spring Boot Actuator, you can take advantage of the Cloud Pub/Sub health indicator called pubsub. The health indicator will verify whether Cloud Pub/Sub is up and accessible by your application. To enable it, all you need to do is add the Spring Boot Actuator to your project.

The pubsub indicator will then roll up to the overall application status visible at localhost:8080/actuator/health (use the management.endpoint.health.show-details property to view per-indicator details).

The health indicator validates the connection to Pub/Sub by pulling messages from a Pub/Sub subscription.

If no subscription has been specified via spring.cloud.gcp.pubsub.health.subscription, it will pull messages from a random subscription that is expected not to exist. It will signal "up" if it is able to connect to GCP Pub/Sub APIs, i.e. the pull results in a response of NOT_FOUND or PERMISSION_DENIED.

If a custom subscription has been specified, this health indicator will signal "up" if messages are successfully pulled and (optionally) acknowledged, or when a successful pull is performed but no messages are returned from Pub/Sub. Note that messages pulled from the subscription will not be acknowledged, unless you set the spring.cloud.gcp.pubsub.health.acknowledge-messages option to true. So, take care not to configure a subscription that has a business impact, or instead leave the custom subscription out completely.

If you are using Spring Boot Actuator, you can take advantage of the Cloud Pub/Sub subscription health indicator called pubsub-subscriber. The subscription health indicator will verify whether Pub/Sub subscriptions are actively processing messages from the subscription’s backlog. To enable it, you need to add the Spring Boot Actuator to your project and the GCP Monitoring. Also you need to set the following properties spring.cloud.gcp.pubsub.health.lagThreshold, spring.cloud.gcp.pubsub.health.backlogThreshold.

The pubsub-subscriber indicator will then roll up to the overall application status visible at localhost:8080/actuator/health (use the management.endpoint.health.show-details property to view per-indicator details).

The health indicator validates a subscriber’s health by checking the subscription’s message backlog and the last processed message. A subscription’s backlog is retrieved using Google Cloud’s Monitoring Metrics. The metric used is the num_undelivered_messages for a subscription.

If a message has been recently processed in a reasonable time threshold, then the subscriber is healthy. If the backlog of messages for a subscription is big but the subscriber consumes messages then subscriber is still healthy. If there hasn’t been any processing of recent messages but the backlog increases, then the subscriber is unhealthy.

PubSubTemplate provides asynchronous methods to publish messages to a Google Cloud Pub/Sub topic. The publish() method takes in a topic name to post the message to, a payload of a generic type and, optionally, a map with the message headers. The topic name could either be a canonical topic name within the current project, or the fully-qualified name referring to a topic in a different project using the projects/<project_name>/topics/<topic_name> format.

Here is an example of how to publish a message to a Google Cloud Pub/Sub topic:

By default, the SimplePubSubMessageConverter is used to convert payloads of type byte[], ByteString, ByteBuffer, and String to Pub/Sub messages.

Google Cloud Pub/Sub allows many subscriptions to be associated to the same topic. PubSubTemplate allows you to listen to subscriptions via the subscribe() method. When listening to a subscription, messages will be pulled from Google Cloud Pub/Sub asynchronously and passed to a user provided message handler. The subscription name could either be a canonical subscription name within the current project, or the fully-qualified name referring to a subscription in a different project using the projects/<project_name>/subscriptions/<subscription_name> format.

Google Cloud Pub/Sub supports synchronous pulling of messages from a subscription. This is different from subscribing to a subscription, in the sense that subscribing is an asynchronous task.

For serialization and deserialization of POJOs using Jackson JSON, configure a PubSubMessageConverter bean, and the Spring Boot starter for GCP Pub/Sub will automatically wire it into the PubSubTemplate.

Assuming you have the following class defined:

You can serialize objects to JSON on publish automatically:

And that’s how you convert messages to objects on pull:

Please refer to our Pub/Sub JSON Payload Sample App as a reference for using this functionality.

PubSubAdmin implements a method to create topics:

Here is an example of how to create a Google Cloud Pub/Sub topic:

PubSubAdmin implements a method to delete topics:

Here is an example of how to delete a Google Cloud Pub/Sub topic:

PubSubAdmin implements a method to list topics:

Here is an example of how to list every Google Cloud Pub/Sub topic name in a project:

PubSubAdmin implements several methods to create subscriptions to existing topics:

The default value for ackDeadline is 10 seconds. If pushEndpoint isn’t specified, the subscription uses message pulling, instead. You can also pass a Subscription.Builder for full control over any options or features available in the client library.

Here is an example of how to create a Google Cloud Pub/Sub subscription:

PubSubAdmin implements a method to delete subscriptions:

Here is an example of how to delete a Google Cloud Pub/Sub subscription:

PubSubAdmin implements a method to list subscriptions:

Here is an example of how to list every subscription name in a project:

PubSubInboundChannelAdapter is the inbound channel adapter for GCP Pub/Sub that listens to a GCP Pub/Sub subscription for new messages. It converts new messages to an internal Spring Message and then sends it to the bound output channel.

Google Pub/Sub treats message payloads as byte arrays. So, by default, the inbound channel adapter will construct the Spring Message with byte[] as the payload. However, you can change the desired payload type by setting the payloadType property of the PubSubInboundChannelAdapter. The PubSubInboundChannelAdapter delegates the conversion to the desired payload type to the PubSubMessageConverter configured in the PubSubTemplate.

To use the inbound channel adapter, a PubSubInboundChannelAdapter must be provided and configured on the user application side.

In the example, we first specify the MessageChannel where the adapter is going to write incoming messages to. The MessageChannel implementation isn’t important here. Depending on your use case, you might want to use a MessageChannel other than PublishSubscribeChannel.

Then, we declare a PubSubInboundChannelAdapter bean. It requires the channel we just created and a SubscriberFactory, which creates Subscriber objects from the Google Cloud Java Client for Pub/Sub. The Spring Boot starter for GCP Pub/Sub provides a configured PubSubSubscriberOperations object.

While PubSubInboundChannelAdapter, through the underlying Asynchronous Pull Pub/Sub mechanism, provides the best performance for high-volume applications that receive a steady flow of messages, it can create load balancing anomalies due to message caching. This behavior is most obvious when publishing a large batch of small messages that take a long time to process individually. It manifests as one subscriber taking up most messages, even if multiple subscribers are available to take on the work. For a more detailed explanation of this scenario, see GCP Pub/Sub documentation.

In such a scenario, a PubSubMessageSource can help spread the load between different subscribers more evenly.

As with the Inbound Channel Adapter, the message source has a configurable acknowledgement mode, payload type, and header mapping.

The default behavior is to return from the synchronous pull operation immediately if no messages are present. This can be overridden by using setBlockOnPull() method to wait for at least one message to arrive.

By default, PubSubMessageSource pulls from the subscription one message at a time. To pull a batch of messages on each request, use the setMaxFetchSize() method to set the batch size.

The @InboundChannelAdapter annotation above ensures that the configured MessageSource is polled for messages, which are then available for manipulation with any Spring Integration mechanism on the pubsubInputChannel message channel. For example, messages can be retrieved in a method annotated with @ServiceActivator, as seen below.

For additional flexibility, PubSubMessageSource attaches an AcknowledgeablePubSubMessage object to the GcpPubSubHeaders.ORIGINAL_MESSAGE message header. The object can be used for manually (n)acking the message.

PubSubMessageHandler is the outbound channel adapter for GCP Pub/Sub that listens for new messages on a Spring MessageChannel. It uses PubSubTemplate to post them to a GCP Pub/Sub topic.

To construct a Pub/Sub representation of the message, the outbound channel adapter needs to convert the Spring Message payload to a byte array representation expected by Pub/Sub. It delegates this conversion to the PubSubTemplate. To customize the conversion, you can specify a PubSubMessageConverter in the PubSubTemplate that should convert the Object payload and headers of the Spring Message to a PubsubMessage.

To use the outbound channel adapter, a PubSubMessageHandler bean must be provided and configured on the user application side.

The provided PubSubTemplate contains all the necessary configuration to publish messages to a GCP Pub/Sub topic.

PubSubMessageHandler publishes messages asynchronously by default. A publish timeout can be configured for synchronous publishing. If none is provided, the adapter waits indefinitely for a response.

It is possible to set user-defined callbacks for the publish() call in PubSubMessageHandler through the setSuccessCallback() and setFailureCallback() methods (either one or both may be set). These give access to the Pub/Sub publish message ID in case of success, or the root cause exception in case of error. Both callbacks include the original message as the second argument. The old setPublishCallback() method that only gave access to message ID or root cause exception is deprecated and will be removed in a future release.

To override the default topic you can use the GcpPubSubHeaders.TOPIC header.

It is also possible to set an SpEL expression for the topic with the setTopicExpression() or setTopicExpressionString() methods.

These channel adapters contain header mappers that allow you to map, or filter out, headers from Spring to Google Cloud Pub/Sub messages, and vice-versa. By default, the inbound channel adapter maps every header on the Google Cloud Pub/Sub messages to the Spring messages produced by the adapter. The outbound channel adapter maps every header from Spring messages into Google Cloud Pub/Sub ones, except the ones added by Spring and some special headers, like headers with key "id", "timestamp", "gcp_pubsub_acknowledgement", and "gcp_pubsub_ordering_key". In the process, the outbound mapper also converts the value of the headers into string.

Note that you can provide the GcpPubSubHeaders.ORDERING_KEY ("gcp_pubsub_ordering_key") header, which will be automatically mapped to PubsubMessage.orderingKey property, and excluded from the headers in the published message. Remember to set spring.cloud.gcp.pubsub.publisher.enable-message-ordering to true, if you are publishing messages with this header.

Each adapter declares a setHeaderMapper() method to let you further customize which headers you want to map from Spring to Google Cloud Pub/Sub, and vice-versa.

For example, to filter out headers "foo", "bar" and all headers starting with the prefix "prefix_", you can use setHeaderMapper() along with the PubSubHeaderMapper implementation provided by this module.

In the previous example, the "*" pattern means every header is mapped. However, because it comes last in the list, the previous patterns take precedence.

Available examples:

The Google Cloud Storage inbound channel adapter polls a Google Cloud Storage bucket for new files and sends each of them in a Message payload to the MessageChannel specified in the @InboundChannelAdapter annotation. The files are temporarily stored in a folder in the local file system.

Here is an example of how to configure a Google Cloud Storage inbound channel adapter.

The inbound streaming channel adapter is similar to the normal inbound channel adapter, except it does not require files to be stored in the file system.

Here is an example of how to configure a Google Cloud Storage inbound streaming channel adapter.

If you would like to process the files in your bucket in a specific order, you may pass in a Comparator<BlobInfo> to the constructor GcsStreamingMessageSource to sort the files being processed.

The outbound channel adapter allows files to be written to Google Cloud Storage. When it receives a Message containing a payload of type File, it writes that file to the Google Cloud Storage bucket specified in the adapter.

Here is an example of how to configure a Google Cloud Storage outbound channel adapter.

See the Spring Integration with Google Cloud Storage Sample Code.

By default, this binder will send messages to Cloud Pub/Sub asynchronously. If synchronous sending is preferred (for example, to allow propagating errors back to the sender), set spring.cloud.stream.gcp.pubsub.default.producer.sync property to true.

If automatic resource creation is turned ON and the topic corresponding to the destination name does not exist, it will be created.

For example, for the following configuration, a topic called myEvents would be created.

A PubSubInboundChannelAdapter will be configured for your consumer endpoint. You may adjust the ack mode of the consumer endpoint using the ack-mode property. The ack mode controls how messages will be acknowledged when they are successfully received. The three possible options are: AUTO (default), AUTO_ACK, and MANUAL. These options are described in detail in the Pub/Sub channel adapter documentation.

With automatic resource creation turned ON for a consumer, the library creates a topic and/or a subscription if they do not exist. The topic name becomes the same as the destination name, and the subscription name follows these rules (in order of precedence):

For example, with this configuration:

Only an anonymous subscription named anonymous.myEvents.a6d83782-c5a3-4861-ac38-e6e2af15a7be is created and later cleaned up.

In another example, with the following configuration:

These resources will be created:

You may customize channel routing by defining a ConsumerEndpointCustomizer in your autoconfiguration. This is useful if you want to customize the default configurations provided by the Pub/Sub Spring Cloud Stream Binder.

The example below demonstrates how to use a ConsumerEndpointCustomizer to override the default error channel configured by the binder.

A Cloud Logging appender is available using com/google/cloud/spring/logging/logback-appender.xml. This appender builds a Cloud Logging log entry from a JUL or Logback log entry, adds a trace ID to it and sends it to Cloud Logging.

STACKDRIVER_LOG_NAME and STACKDRIVER_LOG_FLUSH_LEVEL environment variables can be used to customize the STACKDRIVER appender.

Your configuration may then look like this:

If you want to have more control over the log output, you can further configure the appender. The following properties are available:

If you would like to send logs asynchronously to Cloud Logging, you can use the AsyncAppender.

Your configuration may then look like this:

For Logback, a com/google/cloud/spring/logging/logback-json-appender.xml file is made available for import to make it easier to configure the JSON Logback appender.

Your configuration may then look something like this:

If your application is running on Google Kubernetes Engine, Google Compute Engine or Google App Engine Flexible, your console logging is automatically saved to Google Cloud Logging. Therefore, you can just include com/google/cloud/spring/logging/logback-json-appender.xml in your logging configuration, which logs JSON entries to the console. The trace id will be set correctly.

If you want to have more control over the log output, you can further configure the appender. The following properties are available:

This is an example of such an Logback configuration:

You can use the Spring Boot Starter for Spring Data Spanner to autoconfigure Google Cloud Spanner in your Spring application. It contains all the necessary setup that makes it easy to authenticate with your Google Cloud project. The following configuration options are available:

Spring Data Repositories can be configured via the @EnableSpannerRepositories annotation on your main @Configuration class. With our Spring Boot Starter for Spring Data Cloud Spanner, @EnableSpannerRepositories is automatically added. It is not required to add it to any other class, unless there is a need to override finer grain configuration parameters provided by @EnableSpannerRepositories.

Our Spring Boot autoconfiguration creates the following beans available in the Spring application context:

Simple constructors are supported on POJOs. The constructor arguments can be a subset of the persistent properties. Every constructor argument needs to have the same name and type as a persistent property on the entity and the constructor should set the property from the given argument. Arguments that are not directly set to properties are not supported.

The @Table annotation can provide the name of the Cloud Spanner table that stores instances of the annotated class, one per row. This annotation is optional, and if not given, the name of the table is inferred from the class name with the first character uncapitalized.

For a simple table, you may only have a primary key consisting of a single column. Even in that case, the @PrimaryKey annotation is required. @PrimaryKey identifies the one or more ID properties corresponding to the primary key.

Spanner has first class support for composite primary keys of multiple columns. You have to annotate all of your POJO’s fields that the primary key consists of with @PrimaryKey as below:

The keyOrder parameter of @PrimaryKey identifies the properties corresponding to the primary key columns in order, starting with 1 and increasing consecutively. Order is important and must reflect the order defined in the Cloud Spanner schema. In our example the DDL to create the table and its primary key is as follows:

Spanner does not have automatic ID generation. For most use-cases, sequential IDs should be used with caution to avoid creating data hotspots in the system. Read Spanner Primary Keys documentation for a better understanding of primary keys and recommended practices.

All accessible properties on POJOs are automatically recognized as a Cloud Spanner column. Column naming is generated by the PropertyNameFieldNamingStrategy by default defined on the SpannerMappingContext bean. The @Column annotation optionally provides a different column name than that of the property and some other settings:

If an object of type B is embedded as a property of A, then the columns of B will be saved in the same Cloud Spanner table as those of A.

If B has primary key columns, those columns will be included in the primary key of A. B can also have embedded properties. Embedding allows reuse of columns between multiple entities, and can be useful for implementing parent-child situations, because Cloud Spanner requires child tables to include the key columns of their parents.

For example:

Entities of B can be stored in a table defined as:

Note that the following restrictions apply when you use embedded objects:

Spring Data Cloud Spanner supports parent-child relationships using the Cloud Spanner parent-child interleaved table mechanism. Cloud Spanner interleaved tables enforce the one-to-many relationship and provide efficient queries and operations on entities of a single domain parent entity. These relationships can be up to 7 levels deep. Cloud Spanner also provides automatic cascading delete or enforces the deletion of child entities before parents.

While one-to-one and many-to-many relationships can be implemented in Cloud Spanner and Spring Data Cloud Spanner using constructs of interleaved parent-child tables, only the parent-child relationship is natively supported.

For example, the following Java entities:

These classes can correspond to an existing pair of interleaved tables. The @Interleaved annotation may be applied to Collection properties and the inner type is resolved as the child entity type. The schema needed to create them can also be generated using the SpannerSchemaUtils and run by using the SpannerDatabaseAdminTemplate:

The createStrings list contains table schema statements using column names and types compatible with the provided Java type and any resolved child relationship types contained within based on the configured custom converters.

The ON DELETE CASCADE clause indicates that Cloud Spanner will delete all Albums of a singer if the Singer is deleted. The alternative is ON DELETE NO ACTION, where a Singer cannot be deleted until all of its Albums have already been deleted. When using SpannerSchemaUtils to generate the schema strings, the spring.cloud.gcp.spanner.createInterleavedTableDdlOnDeleteCascade boolean setting determines if these schema are generated as ON DELETE CASCADE for true and ON DELETE NO ACTION for false.

Cloud Spanner restricts these relationships to 7 child layers. A table may have multiple child tables.

On updating or inserting an object to Cloud Spanner, all of its referenced children objects are also updated or inserted in the same request, respectively. On read, all of the interleaved child rows are also all read.

Spring Data Cloud Spanner natively supports the following types for regular fields but also utilizes custom converters (detailed in following sections) and dozens of pre-defined Spring Data custom converters to handle other common Java types.

Natively supported types:

Spanner supports JSON type for columns. JSON columns are mapped to custom POJOs annotated with @Column(spannerType = TypeCode.JSON). Read, write and query with custom SQL query are supported for JSON annotated fields.

Spanner supports ARRAY types for columns. ARRAY columns are mapped to List fields in POJOs.

Example:

The types inside the lists can be any singular property type.

Cloud Spanner queries can construct STRUCT values that appear as columns in the result. Cloud Spanner requires STRUCT values appear in ARRAYs at the root level: SELECT ARRAY(SELECT STRUCT(1 as val1, 2 as val2)) as pair FROM Users.

Spring Data Cloud Spanner will attempt to read the column STRUCT values into a property that is an Iterable of an entity type compatible with the schema of the column STRUCT value.

For the previous array-select example, the following property can be mapped with the constructed ARRAY<STRUCT> column: List<TwoInts> pair; where the TwoInts type is defined:

Custom converters can be used to extend the type support for user defined types.

For example:

We would like to have a field of type Person on our Trade POJO:

Where Person is a simple class:

We have to define the two converters:

That will be configured in our @Configuration file:

Note that ConverterAwareMappingSpannerEntityProcessor takes a list of Converters for write and read operations to support multiple user-defined types. When there are duplicate Converters for a user-defined class in the list, it chooses the first matching item in the lists. This means that, for a user-defined class U, write operations use the first Converter<U, …​> from the write Converters and read operations use the first Converter<…​, U> from the read Converters.

If a Converter<Struct, A> is provided, then properties of type List<A> can be used in your entity types.

Cloud Spanner has SQL support for running read-only queries. All the query related methods start with query on SpannerTemplate. By using SpannerTemplate, you can run SQL queries that map to POJOs:

Spanner exposes a Read API for reading single row or multiple rows in a table or in a secondary index.

Using SpannerTemplate you can run reads, as the following example shows:

Main benefit of reads over queries is reading multiple rows of a certain pattern of keys is much easier using the features of the KeySet class.

The write methods of SpannerOperations accept a POJO and writes all of its properties to Spanner. The corresponding Spanner table and entity metadata is obtained from the given object’s actual type.

If a POJO was retrieved from Spanner and its primary key properties values were changed and then written or updated, the operation will occur as if against a row with the new primary key values. The row with the original primary key values will not be affected.

DML statements can be run by using SpannerOperations.executeDmlStatement. Inserts, updates, and deletions can affect any number of rows and entities.

You can run partitioned DML updates by using the executePartitionedDmlStatement method. Partitioned DML queries have performance benefits but also have restrictions and cannot be used inside transactions.

SpannerOperations provides methods to run java.util.Function objects within a single transaction while making available the read and write methods from SpannerOperations.

SpannerTemplate supports DML Statements. DML statements can also be run in transactions by using performReadWriteTransaction or by using the @Transactional annotation.

CrudRepository methods work as expected, with one thing Spanner specific: the save and saveAll methods work as update-or-insert.

You can also use PagingAndSortingRepository with Spanner Spring Data. The sorting and pageable findAll methods available from this interface operate on the current state of the Spanner database. As a result, beware that the state of the database (and the results) might change when moving page to page.

The SpannerRepository extends the PagingAndSortingRepository, but adds the read-only and the read-write transaction functionality provided by Spanner. These transactions work very similarly to those of SpannerOperations, but is specific to the repository’s domain type and provides repository functions instead of template functions.

For example, this is a read-only transaction:

When creating custom repositories for your own domain types and query methods, you can extend SpannerRepository to access Cloud Spanner-specific features as well as all features from PagingAndSortingRepository and CrudRepository.

In the example above, the query methods in TradeRepository are generated based on the name of the methods, using the Spring Data Query creation naming convention.

List<Trade> findByAction(String action) would translate to a SELECT * FROM trades WHERE action = ?.

The function List<Trade> findTop3DistinctByActionAndSymbolIgnoreCaseOrTraderIdOrderBySymbolDesc(String action, String symbol, String traderId); will be translated as the equivalent of this SQL query:

The following filter options are supported:

Note that the phrase SymbolIgnoreCase is translated to LOWER(SYMBOL) = LOWER(?) indicating a non-case-sensitive matching. The IgnoreCase phrase may only be appended to fields that correspond to columns of type STRING or BYTES. The Spring Data "AllIgnoreCase" phrase appended at the end of the method name is not supported.

The Like or NotLike naming conventions:

The param symbolFragment can contain wildcard characters for string matching such as _ and %.

The Contains and NotContains naming conventions:

The param symbolFragment is a regular expression that is checked for occurrences.

The In and NotIn keywords must be used with Iterable corresponding parameters.

Delete queries are also supported. For example, query methods such as deleteByAction or removeByAction delete entities found by findByAction. The delete operation happens in a single transaction.

Delete queries can have the following return types: * An integer type that is the number of entities deleted * A collection of entities that were deleted * void

The example above for List<Trade> fetchByActionNamedQuery(String action) does not match the Spring Data Query creation naming convention, so we have to map a parametrized Spanner SQL query to it.

The SQL query for the method can be mapped to repository methods in one of two ways:

The names of the tags of the SQL correspond to the @Param annotated names of the method parameters.

Interleaved properties are loaded eagerly, unless they are annotated with @Interleaved(lazy = true).

Custom SQL query methods can accept a single Sort or Pageable parameter that is applied on top of the specified custom query. It is the recommended way to control the sort order of the results, which is not guaranteed by the ORDER BY clause in the SQL query. This is due to the fact that the user-provided query is used as a sub-query, and Cloud Spanner doesn’t preserve order in subquery results.

You might want to use ORDER BY with LIMIT to obtain the top records, according to a specified order. However, to ensure the correct sort order of the final result set, sort options have to be passed in with a Pageable.

This can be used:

The results would be sorted by "id" in ascending order.

Your query method can also return non-entity types:

DML statements can also be run by query methods, but the only possible return value is a long representing the number of affected rows. The dmlStatement boolean setting must be set on @Query to indicate that the query method is run as a DML statement.

Spring Data Spanner supports projections. You can define projection interfaces based on domain types and add query methods that return them in your repository:

Projections can be provided by name-convention-based query methods as well as by custom SQL queries. If using custom SQL queries, you can further restrict the columns retrieved from Spanner to just those required by the projection to improve performance.

Properties of projection types defined using SpEL use the fixed name target for the underlying domain object. As a result accessing underlying properties take the form target.<property-name>.

Java java.util.Optional can be used to indicate the potential absence of a return value.

Alternatively, query methods can return the result without a wrapper. In that case the absence of a query result is indicated by returning null. Repository methods returning collections are guaranteed never to return null but rather the corresponding empty collection.

When running with Spring Boot, repositories can be exposed as REST services by simply adding this dependency to your pom file:

If you prefer to configure parameters (such as path), you can use @RepositoryRestResource annotation:

For example, you can retrieve all Trade objects in the repository by using curl http://<server>:<port>/trades, or any specific trade via curl http://<server>:<port>/trades/<trader_id>,<trade_id>.

The separator between your primary key components, id and trader_id in this case, is a comma by default, but can be configured to any string not found in your key values by extending the SpannerKeyIdConverter class:

You can also write trades using curl -XPOST -H"Content-Type: application/json" -d@test.json http://<server>:<port>/trades/ where the file test.json holds the JSON representation of a Trade object.

If you are using Spring Boot Actuator, you can take advantage of the Cloud Spanner health indicator called spanner. The health indicator will verify whether Cloud Spanner is up and accessible by your application. To enable it, all you need to do is add the Spring Boot Actuator to your project.

The spanner indicator will then roll up to the overall application status visible at localhost:8080/actuator/health (use the management.endpoint.health.show-details property to view per-indicator details).

The health indicator validates the connection to Spanner by executing a query. A query to validate can be configured via spring.cloud.gcp.spanner.health.query property.

You can use the Spring Boot Starter for Spring Data Datastore to autoconfigure Google Cloud Datastore in your Spring application. It contains all the necessary setup that makes it easy to authenticate with your Google Cloud project. The following configuration options are available:

Spring Data Repositories can be configured via the @EnableDatastoreRepositories annotation on your main @Configuration class. With our Spring Boot Starter for Spring Data Cloud Datastore, @EnableDatastoreRepositories is automatically added. It is not required to add it to any other class, unless there is a need to override finer grain configuration parameters provided by @EnableDatastoreRepositories.

Our Spring Boot autoconfiguration creates the following beans available in the Spring application context:

This Spring Boot autoconfiguration can also configure and start a local Datastore Emulator server if enabled by property.

It is useful for integration testing, but not for production.

When enabled, the spring.cloud.gcp.datastore.host property will be ignored and the Datastore autoconfiguration itself will be forced to connect to the autoconfigured local emulator instance.

It will create an instance of LocalDatastoreHelper as a bean that stores the DatastoreOptions to get the Datastore client connection to the emulator for convenience and lower level API for local access. The emulator will be properly stopped after the Spring application context shutdown.

Simple constructors are supported on POJOs. The constructor arguments can be a subset of the persistent properties. Every constructor argument needs to have the same name and type as a persistent property on the entity and the constructor should set the property from the given argument. Arguments that are not directly set to properties are not supported.

The @Entity annotation can provide the name of the Cloud Datastore kind that stores instances of the annotated class, one per row.

@Id identifies the property corresponding to the ID value.

You must annotate one of your POJO’s fields as the ID value, because every entity in Cloud Datastore requires a single ID value:

Datastore can automatically allocate integer ID values. If a POJO instance with a Long ID property is written to Cloud Datastore with null as the ID value, then Spring Data Cloud Datastore will obtain a newly allocated ID value from Cloud Datastore and set that in the POJO for saving. Because primitive long ID properties cannot be null and default to 0, keys will not be allocated.

All accessible properties on POJOs are automatically recognized as a Cloud Datastore field. Field naming is generated by the PropertyNameFieldNamingStrategy by default defined on the DatastoreMappingContext bean. The @Field annotation optionally provides a different field name than that of the property.

Spring Data Cloud Datastore supports the following types for regular fields and elements of collections:

In addition, all types that can be converted to the ones listed in the table by org.springframework.core.convert.support.DefaultConversionService are supported.

Custom converters can be used extending the type support for user defined types.

For example:

We would like to have a field of type Album on our Singer POJO and want it to be stored as a string property:

Where Album is a simple class:

We have to define the two converters:

That will be configured in our @Configuration file:

Arrays and collections (types that implement java.util.Collection) of supported types are supported. They are stored as com.google.cloud.datastore.ListValue. Elements are converted to Cloud Datastore supported types individually. byte[] is an exception, it is converted to com.google.cloud.datastore.Blob.

Users can provide converters from List<?> to the custom collection type. Only read converter is necessary, the Collection API is used on the write side to convert a collection to the internal list type.

Collection converters need to implement the org.springframework.core.convert.converter.Converter interface.

Example:

Let’s improve the Singer class from the previous example. Instead of a field of type Album, we would like to have a field of type Set<Album>:

We have to define a read converter only:

And add it to the list of custom converters:

Java entity types related by inheritance can be stored in the same Kind. When reading and querying entities using DatastoreRepository or DatastoreTemplate with a superclass as the type parameter, you can receive instances of subclasses if you annotate the superclass and its subclasses with DiscriminatorField and DiscriminatorValue:

Instances of all 3 types are stored in the pets Kind. Because a single Kind is used, all classes in the hierarchy must share the same ID property and no two instances of any type in the hierarchy can share the same ID value.

Entity rows in Cloud Datastore store their respective types' DiscriminatorValue in a field specified by the root superclass’s DiscriminatorField (pet_type in this case). Reads and queries using a given type parameter will match each entity with its specific type. For example, reading a List<Pet> will produce a list containing instances of all 3 types. However, reading a List<Dog> will produce a list containing only Dog and Pug instances. You can include the pet_type discrimination field in your Java entities, but its type must be convertible to a collection or array of String. Any value set in the discrimination field will be overwritten upon write to Cloud Datastore.

Fields whose types are also annotated with @Entity are converted to EntityValue and stored inside the parent entity.

Here is an example of Cloud Datastore entity containing an embedded entity in JSON:

This corresponds to a simple pair of Java entities:

Child entities are not stored in their own kind. They are stored in their entirety in the child field of the parents kind.

Multiple levels of embedded entities are supported.

Example:

Entities can hold embedded entities that are their own type. We can store trees in Cloud Datastore using this feature:

Parent-child relationships are supported via the @Descendants annotation.

Unlike embedded children, descendants are fully-formed entities residing in their own kinds. The parent entity does not have an extra field to hold the descendant entities. Instead, the relationship is captured in the descendants' keys, which refer to their parent entities:

For example, an instance of a GQL key-literal representation for Item would also contain the parent ShoppingOrder ID value:

The GQL key-literal representation for the parent ShoppingOrder would be:

The Cloud Datastore entities exist separately in their own kinds.

The ShoppingOrder:

The two items inside that order:

The parent-child relationship structure of objects is stored in Cloud Datastore using Datastore’s ancestor relationships. Because the relationships are defined by the Ancestor mechanism, there is no extra column needed in either the parent or child entity to store this relationship. The relationship link is part of the descendant entity’s key value. These relationships can be many levels deep.

Properties holding child entities must be collection-like, but they can be any of the supported inter-convertible collection-like types that are supported for regular properties such as List, arrays, Set, etc…​ Child items must have Key as their ID type because Cloud Datastore stores the ancestor relationship link inside the keys of the children.

Reading or saving an entity automatically causes all subsequent levels of children under that entity to be read or saved, respectively. If a new child is created and added to a property annotated @Descendants and the key property is left null, then a new key will be allocated for that child. The ordering of the retrieved children may not be the same as the ordering in the original property that was saved.

Child entities cannot be moved from the property of one parent to that of another unless the child’s key property is set to null or a value that contains the new parent as an ancestor. Since Cloud Datastore entity keys can have multiple parents, it is possible that a child entity appears in the property of multiple parent entities. Because entity keys are immutable in Cloud Datastore, to change the key of a child you must delete the existing one and re-save it with the new key.

General relationships can be stored using the @Reference annotation.

@Reference relationships are between fully-formed entities residing in their own kinds. The relationship between ShoppingOrder and Item entities are stored as a Key field inside ShoppingOrder, which are resolved to the underlying Java entity type by Spring Data Cloud Datastore:

Reference properties can either be singular or collection-like. These properties correspond to actual columns in the entity and Cloud Datastore Kind that hold the key values of the referenced entities. The referenced entities are full-fledged entities of other Kinds.

Similar to the @Descendants relationships, reading or writing an entity will recursively read or write all of the referenced entities at all levels. If referenced entities have null ID values, then they will be saved as new entities and will have ID values allocated by Cloud Datastore. There are no requirements for relationships between the key of an entity and the keys that entity holds as references. The order of collection-like reference properties is not preserved when reading back from Cloud Datastore.

In addition to retrieving entities by their IDs, you can also submit queries.

These methods, respectively, allow querying for:

Using DatastoreTemplate you can find entities by id. For example:

Cloud Datastore uses key-based reads with strong consistency, but queries with eventual consistency. In the example above the first two reads utilize keys, while the third is run by using a query based on the corresponding Kind of Trader.

The write methods of DatastoreOperations accept a POJO and writes all of its properties to Datastore. The required Datastore kind and entity metadata is obtained from the given object’s actual type.

If a POJO was retrieved from Datastore and its ID value was changed and then written or updated, the operation will occur as if against a row with the new ID value. The entity with the original ID value will not be affected.

The save method behaves as update-or-insert.

Read and write transactions are provided by DatastoreOperations via the performTransaction method:

The performTransaction method accepts a Function that is provided an instance of a DatastoreOperations object. The final returned value and type of the function is determined by the user. You can use this object just as you would a regular DatastoreOperations with an exception:

Because of Cloud Datastore’s consistency guarantees, there are limitations to the operations and relationships among entities used inside transactions.

You can work with Maps of type Map<String, ?> instead of with entity objects by directly reading and writing them to and from Cloud Datastore.

The map keys are used as field names for a Datastore entity and map values are converted to Datastore supported types. Only simple types are supported (i.e. collections are not supported). Converters for custom value types can be added (see Custom types section).

Example:

In the example above the query methods in TradeRepository are generated based on the name of the methods using the Spring Data Query creation naming convention.

Cloud Datastore only supports filter components joined by AND, and the following operations:

After writing a custom repository interface specifying just the signatures of these methods, implementations are generated for you and can be used with an auto-wired instance of the repository. Because of Cloud Datastore’s requirement that explicitly selected fields must all appear in a composite index together, find name-based query methods are run as SELECT *.

Delete queries are also supported. For example, query methods such as deleteByAction or removeByAction delete entities found by findByAction. Delete queries are run as separate read and delete operations instead of as a single transaction because Cloud Datastore cannot query in transactions unless ancestors for queries are specified. As a result, removeBy and deleteBy name-convention query methods cannot be used inside transactions via either performInTransaction or @Transactional annotation.

Delete queries can have the following return types:

Methods can have org.springframework.data.domain.Pageable parameter to control pagination and sorting, or org.springframework.data.domain.Sort parameter to control sorting only. See Spring Data documentation for details.

For returning multiple items in a repository method, we support Java collections as well as org.springframework.data.domain.Page and org.springframework.data.domain.Slice. If a method’s return type is org.springframework.data.domain.Page, the returned object will include current page, total number of results and total number of pages.

Java java.util.Optional can be used to indicate the potential absence of a return value.

Alternatively, query methods can return the result without a wrapper. In that case the absence of a query result is indicated by returning null. Repository methods returning collections are guaranteed never to return null but rather the corresponding empty collection.

Query by Example is an alternative querying technique. It enables dynamic query generation based on a user-provided object. See Spring Data Documentation for details.

Custom GQL queries can be mapped to repository methods in one of two ways:

These transactions work very similarly to those of DatastoreOperations, but is specific to the repository’s domain type and provides repository functions instead of template functions.

For example, this is a read-write transaction:

Spring Data Cloud Datastore supports projections. You can define projection interfaces based on domain types and add query methods that return them in your repository:

Projections can be provided by name-convention-based query methods as well as by custom GQL queries. If using custom GQL queries, you can further restrict the fields retrieved from Cloud Datastore to just those required by the projection. However, custom select statements (those not using SELECT *) require composite indexes containing the selected fields.

Properties of projection types defined using SpEL use the fixed name target for the underlying domain object. As a result, accessing underlying properties take the form target.<property-name>.

When running with Spring Boot, repositories can be exposed as REST services by simply adding this dependency to your pom file:

If you prefer to configure parameters (such as path), you can use @RepositoryRestResource annotation:

For example, you can retrieve all Trade objects in the repository by using curl http://<server>:<port>/trades, or any specific trade via curl http://<server>:<port>/trades/<trader_id>.

You can also write trades using curl -XPOST -H"Content-Type: application/json" -d@test.json http://<server>:<port>/trades/ where the file test.json holds the JSON representation of a Trade object.

To delete trades, you can use curl -XDELETE http://<server>:<port>/trades/<trader_id>

If you are using Spring Boot Actuator, you can take advantage of the Cloud Datastore health indicator called datastore. The health indicator will verify whether Cloud Datastore is up and accessible by your application. To enable it, all you need to do is add the Spring Boot Actuator to your project.

The Spring Boot starter for Google Cloud Firestore provides the following configuration options:

You may use the following field types when defining your persistent entities or when binding query parameters:

Spring Data Repositories can be configured via the @EnableReactiveFirestoreRepositories annotation on your main @Configuration class. With our Spring Boot Starter for Spring Data Cloud Firestore, @EnableReactiveFirestoreRepositories is automatically added. It is not required to add it to any other class, unless there is a need to override finer grain configuration parameters provided by @EnableReactiveFirestoreRepositories.

Our Spring Boot autoconfiguration creates the following beans available in the Spring application context:

Spring Data Cloud Firestore supports embedded properties of custom types and lists. Given a custom POJO definition, you can have properties of this type or lists of this type in your entities. They are stored as embedded documents (or arrays, correspondingly) in the Cloud Firestore.

Example:

This feature requires a bean of SpannerTransactionManager, which is provided when using spring-cloud-gcp-starter-data-firestore.