A Java library that provides the tools you need to authenticate with, and use the Qubole Data Service API.
The SDK is available in The Central Repository. To use the SDK, add the following dependency to your Java application.
<dependency>
<groupId>com.qubole.qds-sdk-java</groupId>
<artifactId>qds-sdk-java</artifactId>
<version>0.8.0</version>
</dependency>
NOTE: see the bullet below regarding Jersery 2.0
In your application initialization code, allocate a QdsClient object:
QdsConfiguration configuration = new DefaultQdsConfiguration(YOUR_API_KEY);
QdsClient client = QdsClientFactory.newClient(configuration);
YOUR_API_KEY referenced in above code is a placeholder for Qubole API Token. Please refer to following documentation for further details on how to obtain them: http://docs.qubole.com/en/latest/rest-api/api_overview.html#authentication
Then, make api calls as needed. E.g.
Future<CommandResponse> hiveCommandResponseFuture = client
.command()
.hive()
.query("show tables;")
.invoke();
CommandResponse hiveCommandResponse = hiveCommandResponseFuture.get();
...
Alternatively, you can use Jersey's callback mechanism. E.g.
InvocationCallback<CommandResponse> callback = new InvocationCallback<CommandResponse>()
{
@Override
public void completed(CommandResponse clusterItems)
{
// ...
}
@Override
public void failed(Throwable throwable)
{
// ...
}
};
client.command()
.hive()
.query("show tables;")
.withCallback(callback)
.invoke();
...
As part of your application's shutdown, close the client:
client.close();
Important: when you submit a command/query, it can take time for it to execute. You cannot get the result until it is ready. A utility is provided that polls the command and waits for the results: ResultLatch. You can use it to block in the foreground or using a callback.
Blocking:
ResultLatch latch = new ResultLatch(client, queryId);
ResultValue = latch.awaitResult();
With callback:
ResultLatch.Callback callback = new ResultLatch.Callback()
{
@Override
public void result(String queryId, ResultValue resultValue)
{
// use results
}
@Override
public void error(String queryId, Exception e)
{
// handle error
}
};
ResultLatch latch = new ResultLatch(client, queryId);
latch.callback(callback);
Some Qubole APIs write large result sets to S3. If you would like to stream those results, use ResultStreamer. E.g.
ResultStreamer streamer = new ResultStreamer(client); // save this until the end of your application
...
Future<ResultValue> results = client.command().results(id)...invoke();
Reader in = streamer.getResults(results.get());
As part of your application's shutdown, close the client:
streamer.close();
Some of the APIs support paging. These APIs have the "forPage" method. E.g.
// return page 2 using 3 per page
client.command().history().forPage(2, 3).invoke();
Using the QdsClient, you can access any of the Qubole APIs:
API | Example |
---|---|
Reports | client.report().allCommandsReport().start_date(...).end_date(...).limit(...).invoke(); |
Scheduler | client.scheduler().list().invoke(); |
DbTaps | client.dbTaps().list().invoke(); |
Hive Metadata | client.hiveMetadata().getTableProperties("table").invoke(); |
Cluster | client.cluster().list().invoke(); |
Command | client.command().history().invoke(); |
The SDK uses Jersey 2.0. Some widely used open source libraries such as Dropwizard are incompatible with Jersey 2.0. To workaround this incompatiblity, you can build the SDK using the Maven shade plugin which will hide the SDK's usage of Jersey 2.0. To build a shaded version of the SDK, follow these steps:
- Download the SDK. Either:
- Clone the project:
git clone [email protected]:qubole/qds-sdk-java.git
- Or download one of the releases from https://github.com/qubole/qds-sdk-java/releases
- cd to the directory
mvn -P shaded install