2019-10-24 16:00:28 +02:00
|
|
|
package eu.dnetlib.dhp.graph;
|
|
|
|
|
2019-11-04 12:28:56 +01:00
|
|
|
import com.fasterxml.jackson.databind.ObjectMapper;
|
2019-10-25 18:10:30 +02:00
|
|
|
import eu.dnetlib.dhp.application.ArgumentApplicationParser;
|
|
|
|
import org.apache.commons.io.IOUtils;
|
2019-10-24 16:00:28 +02:00
|
|
|
import org.apache.hadoop.io.Text;
|
|
|
|
import org.apache.spark.api.java.JavaSparkContext;
|
2019-10-25 09:24:18 +02:00
|
|
|
import org.apache.spark.sql.Encoders;
|
2019-11-06 17:45:52 +01:00
|
|
|
import org.apache.spark.sql.SaveMode;
|
2019-10-24 16:00:28 +02:00
|
|
|
import org.apache.spark.sql.SparkSession;
|
|
|
|
|
|
|
|
public class SparkGraphImporterJob {
|
|
|
|
|
2019-10-25 18:10:30 +02:00
|
|
|
public static void main(String[] args) throws Exception {
|
2019-10-24 16:00:28 +02:00
|
|
|
|
2019-10-25 18:10:30 +02:00
|
|
|
final ArgumentApplicationParser parser = new ArgumentApplicationParser(IOUtils.toString(SparkGraphImporterJob.class.getResourceAsStream("/eu/dnetlib/dhp/graph/input_graph_parameters.json")));
|
|
|
|
parser.parseArgument(args);
|
2019-10-24 16:00:28 +02:00
|
|
|
final SparkSession spark = SparkSession
|
|
|
|
.builder()
|
2019-11-06 17:45:52 +01:00
|
|
|
.appName(SparkGraphImporterJob.class.getSimpleName())
|
2019-10-25 18:10:30 +02:00
|
|
|
.master(parser.get("master"))
|
2019-11-07 17:38:09 +01:00
|
|
|
.config("hive.metastore.uris", parser.get("hive_metastore_uris"))
|
2019-11-06 17:45:52 +01:00
|
|
|
.enableHiveSupport()
|
2019-10-24 16:00:28 +02:00
|
|
|
.getOrCreate();
|
2019-11-06 17:45:52 +01:00
|
|
|
|
2019-10-24 16:00:28 +02:00
|
|
|
final JavaSparkContext sc = new JavaSparkContext(spark.sparkContext());
|
2019-11-04 17:41:01 +01:00
|
|
|
final String inputPath = parser.get("sourcePath");
|
2019-11-07 17:38:09 +01:00
|
|
|
final String hiveDbName = parser.get("hive_db_name");
|
|
|
|
|
|
|
|
spark.sql(String.format("CREATE DATABASE IF NOT EXISTS %s", hiveDbName));
|
2019-10-30 12:19:03 +01:00
|
|
|
|
2019-10-25 18:10:30 +02:00
|
|
|
// Read the input file and convert it into RDD of serializable object
|
2019-11-04 17:41:01 +01:00
|
|
|
GraphMappingUtils.types.forEach((name, clazz) -> {
|
2020-02-13 16:53:27 +01:00
|
|
|
spark.createDataset(sc.sequenceFile(inputPath + "/" + name, Text.class, Text.class)
|
|
|
|
.map(s -> new ObjectMapper().readValue(s._2().toString(), clazz))
|
2019-11-06 17:45:52 +01:00
|
|
|
.rdd(), Encoders.bean(clazz))
|
2020-02-13 16:53:27 +01:00
|
|
|
.write()
|
|
|
|
.mode(SaveMode.Overwrite)
|
|
|
|
.saveAsTable(hiveDbName + "." + name);
|
2019-11-04 12:28:56 +01:00
|
|
|
});
|
2019-10-24 17:02:35 +02:00
|
|
|
|
2019-10-24 16:00:28 +02:00
|
|
|
}
|
|
|
|
}
|