Google Cloud BigQuery
The BigQuery connector provides Akka Stream sources and sinks to connect to Google Cloud BigQuery. BigQuery is a serverless data warehouse for storing and analyzing massive datasets. This connector is primarily intended for streaming data into and out of BigQuery tables and running SQL queries, although it also provides basic support for managing datasets and tables and flexible access to the BigQuery REST API.
[+] Show project info| Project Info: Alpakka Google Cloud BigQuery | |
|---|---|
| Artifact | com.lightbend.akka
akka-stream-alpakka-google-cloud-bigquery
9.0.1
|
| JDK versions | Eclipse Temurin JDK 11 Eclipse Temurin JDK 17 |
| Scala versions | 2.13.12 |
| JPMS module name | akka.stream.alpakka.google.cloud.bigquery |
| License | |
| Readiness level |
Since 2.0.2, 2020-07-31
|
| Home page | https://doc.akka.io/libraries/alpakka/current |
| API documentation | |
| Forums | |
| Release notes | GitHub releases |
| Issues | Github issues |
| Sources | https://github.com/akka/alpakka |
Alpakka Google Cloud BigQuery was added in Alpakka 2.0.2 in July 2020 and is marked as “API may change”. Please try it out and suggest improvements. PR #2548
Artifacts
The Akka dependencies are available from Akka’s library repository. To access them there, you need to configure the URL for this repository.
Additionally, add the dependencies as below.
- sbt
val AkkaVersion = "2.10.0" val AkkaHttpVersion = "10.7.0" libraryDependencies ++= Seq( "com.lightbend.akka" %% "akka-stream-alpakka-google-cloud-bigquery" % "9.0.1", "com.typesafe.akka" %% "akka-stream" % AkkaVersion, "com.typesafe.akka" %% "akka-http" % AkkaHttpVersion, "com.typesafe.akka" %% "akka-http-spray-json" % AkkaHttpVersion )- Maven
- Gradle
To use the Jackson JSON library for marshalling you must also add the Akka HTTP module for Jackson support.
- sbt
val AkkaHttpVersion = "10.7.0" libraryDependencies += "com.typesafe.akka" %% "akka-http-jackson" % AkkaHttpVersion- Maven
- Gradle
The table below shows direct dependencies of this module and the second tab shows all libraries that it depends on transitively.
- Direct dependencies
Organization Artifact Version com.fasterxml.jackson.core jackson-annotations 2.17.3 com.fasterxml.jackson.core jackson-core 2.17.3 com.fasterxml.jackson.core jackson-databind 2.17.3 com.lightbend.akka akka-stream-alpakka-google-common_2.13 9.0.1 com.typesafe.akka akka-http-spray-json_2.13 10.7.0 com.typesafe.akka akka-http_2.13 10.7.0 com.typesafe.akka akka-pki_2.13 2.10.0 com.typesafe.akka akka-stream_2.13 2.10.0 io.spray spray-json_2.13 1.3.6 org.scala-lang scala-library 2.13.12 - Dependency tree
Configuration
The BigQuery connector shares its basic configuration with all the Google connectors in Alpakka. Additional BigQuery-specific configuration settings can be found in its reference.conf.
Imports
All of the examples below assume the following imports are in scope.
- Scala
- Java
-
source
import akka.Done; import akka.NotUsed; import akka.http.javadsl.marshallers.jackson.Jackson; import akka.http.javadsl.marshalling.Marshaller; import akka.http.javadsl.model.HttpEntity; import akka.http.javadsl.model.RequestEntity; import akka.http.javadsl.unmarshalling.Unmarshaller; import akka.stream.alpakka.google.GoogleAttributes; import akka.stream.alpakka.google.GoogleSettings; import akka.stream.alpakka.googlecloud.bigquery.InsertAllRetryPolicy; import akka.stream.alpakka.googlecloud.bigquery.javadsl.BigQuery; import akka.stream.alpakka.googlecloud.bigquery.javadsl.jackson.BigQueryMarshallers; import akka.stream.alpakka.googlecloud.bigquery.model.Dataset; import akka.stream.alpakka.googlecloud.bigquery.model.Job; import akka.stream.alpakka.googlecloud.bigquery.model.JobReference; import akka.stream.alpakka.googlecloud.bigquery.model.JobState; import akka.stream.alpakka.googlecloud.bigquery.model.QueryResponse; import akka.stream.alpakka.googlecloud.bigquery.model.Table; import akka.stream.alpakka.googlecloud.bigquery.model.TableDataInsertAllRequest; import akka.stream.alpakka.googlecloud.bigquery.model.TableDataListResponse; import akka.stream.alpakka.googlecloud.bigquery.model.TableFieldSchema; import akka.stream.alpakka.googlecloud.bigquery.model.TableFieldSchemaMode; import akka.stream.alpakka.googlecloud.bigquery.model.TableFieldSchemaType; import akka.stream.alpakka.googlecloud.bigquery.model.TableListResponse; import akka.stream.alpakka.googlecloud.bigquery.model.TableSchema; import akka.stream.javadsl.Flow; import akka.stream.javadsl.Sink; import akka.stream.javadsl.Source; import com.fasterxml.jackson.annotation.JsonCreator; import com.fasterxml.jackson.annotation.JsonProperty; import com.fasterxml.jackson.databind.JsonNode; import com.fasterxml.jackson.databind.ObjectMapper; import com.fasterxml.jackson.databind.ObjectReader; import java.io.IOException; import java.util.ArrayList; import java.util.Collections; import java.util.List; import java.util.Optional; import java.util.OptionalInt; import java.util.OptionalLong; import java.util.concurrent.CompletableFuture; import java.util.concurrent.CompletionStage; import java.util.function.Function; import java.util.stream.Collectors;
Setup data classes
As a working example throughout this documentation, we will use the Person class to model the data in our BigQuery tables.
- Scala
- Java
-
source
ObjectMapper objectMapper = new ObjectMapper(); public class Person { private String name; private Integer age; private List<Address> addresses; private Boolean isHakker; @JsonCreator public Person(@JsonProperty("f") JsonNode fields) throws IOException { name = fields.get(0).get("v").textValue(); age = Integer.parseInt(fields.get(1).get("v").textValue()); addresses = new ArrayList<>(); ObjectReader addressReader = objectMapper.readerFor(Address.class); for (JsonNode node : fields.get(2).get("v")) { Address address = addressReader.readValue(node.get("v")); addresses.add(address); } isHakker = Boolean.parseBoolean(fields.get(3).get("v").textValue()); } public String getName() { return name; } public Integer getAge() { return age; } public List<Address> getAddresses() { return addresses; } public Boolean getIsHakker() { return isHakker; } } public class Address { private String street; private String city; private Integer postalCode; @JsonCreator public Address(@JsonProperty("f") JsonNode fields) { street = fields.get(0).get("v").textValue(); city = fields.get(1).get("v").textValue(); postalCode = Optional.of(fields.get(2).get("v").textValue()).map(Integer::parseInt).orElse(null); } public String getStreet() { return street; } public String getCity() { return city; } public Integer getPostalCode() { return postalCode; } } public class NameAddressesPair { private String name; private List<Address> addresses; @JsonCreator public NameAddressesPair(@JsonProperty("f") JsonNode fields) throws IOException { name = fields.get(0).get("v").textValue(); addresses = new ArrayList<>(); ObjectReader addressReader = objectMapper.readerFor(Address.class); for (JsonNode node : fields.get(1).get("v")) { Address address = addressReader.readValue(node.get("v")); addresses.add(address); } } }
To enable support for (un)marshalling User and Address as BigQuery table rows and query results we use Jackson’s @JsonCreator and @JsonProperty annotations. Note that a custom @JsonCreator constructor is necessary due to BigQuery’s unusual encoding of rows as “a series of JSON f,v objects for indicating fields and values” (reference documentation). In addition, we also define NameAddressesPair to model the result of the query in the next section.
Run a query
You can run a SQL query and stream the unmarshalled results with the BigQuery.<Out>query method. To create the unmarshaller, use the BigQueryMarshallers.<Out>queryResponseUnmarshaller method.
- Scala
- Java
-
source
String sqlQuery = String.format("SELECT name, addresses FROM %s.%s WHERE age >= 100", datasetId, tableId); Unmarshaller<HttpEntity, QueryResponse<NameAddressesPair>> queryResponseUnmarshaller = BigQueryMarshallers.queryResponseUnmarshaller(NameAddressesPair.class); Source<NameAddressesPair, CompletionStage<QueryResponse<NameAddressesPair>>> centenarians = BigQuery.query(sqlQuery, false, false, queryResponseUnmarshaller);
Notice that the source materializes a CompletionStage<QueryResponse<NameAddressesTuple>> which can be used to retrieve metadata related to the query. For example, you can use a dry run to estimate the number of bytes that will be read by a query.
- Scala
- Java
-
source
Source<NameAddressesPair, CompletionStage<QueryResponse<NameAddressesPair>>> centenariansDryRun = BigQuery.query(sqlQuery, false, false, queryResponseUnmarshaller); CompletionStage<Long> bytesProcessed = centenariansDryRun .to(Sink.ignore()) .run(system) .thenApply(r -> r.getTotalBytesProcessed().getAsLong());
Finally, you can also stream all of the rows in a table without the expense of running a query with the BigQuery.<Out>listTableData method.
- Scala
- Java
-
source
Unmarshaller<HttpEntity, TableDataListResponse<Person>> tableDataListUnmarshaller = BigQueryMarshallers.tableDataListResponseUnmarshaller(Person.class); Source<Person, CompletionStage<TableDataListResponse<Person>>> everyone = BigQuery.listTableData( datasetId, tableId, OptionalLong.empty(), OptionalInt.empty(), Collections.emptyList(), tableDataListUnmarshaller);
Load data into BigQuery
The BigQuery connector enables loading data into tables via real-time streaming inserts or batch loading. For an overview of these strategies see the BigQuery documentation.
The BigQuery.<In>insertAll method creates a sink that accepts batches of List<In> (for example created via the batch operator) and streams them directly into a table. To enable/disable BigQuery’s best-effort deduplication feature use the appropriate InsertAllRetryPolicy.
- Scala
- Java
-
source
Marshaller<TableDataInsertAllRequest<Person>, RequestEntity> tableDataInsertAllMarshaller = BigQueryMarshallers.tableDataInsertAllRequestMarshaller(); Sink<List<Person>, NotUsed> peopleInsertSink = BigQuery.insertAll( datasetId, tableId, InsertAllRetryPolicy.withDeduplication(), Optional.empty(), tableDataInsertAllMarshaller);
As a cost-saving alternative to streaming inserts, you can also add data to a table via asynchronous load jobs. The BigQuery.<In>insertAllAsync method creates a flow that starts a series of batch load jobs. By default, a new load job is created every minute to attempt to emulate near-real-time streaming inserts, although there is no guarantee when the job will actually run. The frequency with which new load jobs are created is controlled by the alpakka.google.bigquery.load-job-per-table-quota configuration setting.
Pending the resolution of Google BigQuery issue 176002651, the BigQuery.insertAllAsync API may not work as expected.
As a workaround, you can use the config setting akka.http.parsing.conflicting-content-type-header-processing-mode = first with Akka HTTP v10.2.4 or later.
- Scala
- Java
-
source
Flow<Person, Job, NotUsed> peopleLoadFlow = BigQuery.insertAllAsync(datasetId, tableId, Jackson.marshaller());
To check the status of the load jobs use the BigQuery.getJob method.
- Scala
- Java
-
source
Function<List<JobReference>, CompletionStage<Boolean>> checkIfJobsDone = jobReferences -> { GoogleSettings settings = GoogleSettings.create(system); CompletionStage<Boolean> allAreDone = CompletableFuture.completedFuture(true); for (JobReference jobReference : jobReferences) { CompletionStage<Job> job = BigQuery.getJob(jobReference.getJobId().get(), Optional.empty(), settings, system); CompletionStage<Boolean> jobIsDone = job.thenApply( j -> j.getStatus().map(s -> s.getState().equals(JobState.done())).orElse(false)); allAreDone = allAreDone.thenCombine(jobIsDone, (a, b) -> a & b); } return allAreDone; }; CompletionStage<List<Job>> jobs = Source.from(people).via(peopleLoadFlow).runWith(Sink.<Job>seq(), system); CompletionStage<List<JobReference>> jobReferences = jobs.thenApply( js -> js.stream().map(j -> j.getJobReference().get()).collect(Collectors.toList())); CompletionStage<Boolean> isDone = jobReferences.thenCompose(checkIfJobsDone);
Managing datasets and tables
The BigQuery connector provides methods for basic management of datasets and tables.
- Scala
- Java
-
source
GoogleSettings settings = GoogleSettings.create(system); Source<Dataset, NotUsed> allDatasets = BigQuery.listDatasets(OptionalInt.empty(), Optional.empty(), Collections.emptyMap()); CompletionStage<Dataset> existingDataset = BigQuery.getDataset(datasetId, settings, system); CompletionStage<Dataset> newDataset = BigQuery.createDataset("newDatasetId", settings, system); CompletionStage<Done> datasetDeleted = BigQuery.deleteDataset(datasetId, false, settings, system); Source<Table, CompletionStage<TableListResponse>> allTablesInDataset = BigQuery.listTables(datasetId, OptionalInt.empty()); CompletionStage<Table> existingTable = BigQuery.getTable(datasetId, tableId, settings, system); CompletionStage<Done> tableDeleted = BigQuery.deleteTable(datasetId, tableId, settings, system);
Creating a table requires a little more work to specify the schema.
- Scala
- Java
-
source
TableSchema personSchema = TableSchema.create( TableFieldSchema.create("name", TableFieldSchemaType.string(), Optional.empty()), TableFieldSchema.create("age", TableFieldSchemaType.integer(), Optional.empty()), TableFieldSchema.create( "addresses", TableFieldSchemaType.record(), Optional.of(TableFieldSchemaMode.repeated()), TableFieldSchema.create("street", TableFieldSchemaType.string(), Optional.empty()), TableFieldSchema.create("city", TableFieldSchemaType.string(), Optional.empty()), TableFieldSchema.create( "postalCode", TableFieldSchemaType.integer(), Optional.of(TableFieldSchemaMode.nullable()))), TableFieldSchema.create("isHakker", TableFieldSchemaType.bool(), Optional.empty())); CompletionStage<Table> newTable = BigQuery.createTable(datasetId, "newTableId", personSchema, settings, system);
Apply custom settings to a part of the stream
In certain situations it may be desirable to modify the GoogleSettings applied to a part of the stream, for example to change the project ID or use different RetrySettings.
- Scala
- Java
-
source
GoogleSettings defaultSettings = GoogleSettings.create(system); GoogleSettings customSettings = defaultSettings.withProjectId("myOtherProjectId"); BigQuery.query(sqlQuery, false, false, queryResponseUnmarshaller) .withAttributes(GoogleAttributes.settings(customSettings));
Make raw API requests
If you would like to interact with the BigQuery REST API beyond what the BigQuery connector supports, you can make authenticated raw requests via the BigQuery.singleRequest and BigQuery.<Out>paginatedRequest methods.