Question

Is there a Hive equivalent to Pig’s PigRunner class that makes it easy to run HQL scripts from within a Java program?

Answer 1

The Spring for Apache Hadoop framework has a Hive integration, having a look at
the source code might give you some idea how to run hql scripts from code.

On the other hand you can also check the Hive sources as well (especially CliSessionState and CliDriver)
to see how Hive shell picks up a hql file (i.e: hive -f file.q) .

Based on these such a raw implementation can do the job:

import java.io.PrintStream;
import org.apache.hadoop.hive.cli.CliDriver;
import org.apache.hadoop.hive.cli.CliSessionState;
import org.apache.hadoop.hive.common.LogUtils;
import org.apache.hadoop.hive.conf.HiveConf;
import org.apache.hadoop.hive.ql.session.SessionState;

public class RunHQLScript {

    private static class MyCliSessionState extends CliSessionState {
        public MyCliSessionState(HiveConf conf, String host, int port) {
            super(conf);
            this.host = host;
            this.port = port;
        }
    }

    public static void main(String[] args) throws Exception {

        LogUtils.initHiveLog4j();
        CliSessionState ss = new MyCliSessionState(new HiveConf(SessionState.class),
                "localhost", 10000);

        ss.in = System.in;
        ss.out = new PrintStream(System.out, true, "UTF-8");
        ss.err = new PrintStream(System.err, true, "UTF-8");
        ss.fileName = "file.q";  //HQL file

        SessionState.start(ss);
        ss.connect();
        CliDriver cli = new CliDriver();
        int processFile = cli.processFile(ss.fileName);
        System.out.println("return code: " +processFile);
        ss.close();
    }
}

Note that a running Thrift service (on port 10000 by default) is needed to be able to execute the script.