forked from D-Net/dnet-hadoop
101 lines
3.4 KiB
Java
101 lines
3.4 KiB
Java
|
|
package eu.dnetlib.dhp.oa.dedup;
|
|
|
|
import java.io.IOException;
|
|
import java.io.Serializable;
|
|
import java.io.StringReader;
|
|
import java.util.ArrayList;
|
|
import java.util.List;
|
|
|
|
import org.apache.spark.SparkConf;
|
|
import org.apache.spark.sql.Dataset;
|
|
import org.apache.spark.sql.SaveMode;
|
|
import org.apache.spark.sql.SparkSession;
|
|
import org.dom4j.Document;
|
|
import org.dom4j.DocumentException;
|
|
import org.dom4j.Element;
|
|
import org.dom4j.io.SAXReader;
|
|
|
|
import com.fasterxml.jackson.databind.DeserializationFeature;
|
|
import com.fasterxml.jackson.databind.ObjectMapper;
|
|
|
|
import eu.dnetlib.dhp.application.ArgumentApplicationParser;
|
|
import eu.dnetlib.dhp.common.HdfsSupport;
|
|
import eu.dnetlib.dhp.schema.oaf.*;
|
|
import eu.dnetlib.enabling.is.lookup.rmi.ISLookUpException;
|
|
import eu.dnetlib.enabling.is.lookup.rmi.ISLookUpService;
|
|
import eu.dnetlib.pace.config.DedupConfig;
|
|
|
|
abstract class AbstractSparkAction implements Serializable {
|
|
|
|
protected static final int NUM_PARTITIONS = 1000;
|
|
protected static final int NUM_CONNECTIONS = 20;
|
|
|
|
protected static final ObjectMapper OBJECT_MAPPER = new ObjectMapper()
|
|
.configure(DeserializationFeature.FAIL_ON_UNKNOWN_PROPERTIES, false);
|
|
|
|
public ArgumentApplicationParser parser; // parameters for the spark action
|
|
public SparkSession spark; // the spark session
|
|
|
|
public AbstractSparkAction(ArgumentApplicationParser parser, SparkSession spark) {
|
|
|
|
this.parser = parser;
|
|
this.spark = spark;
|
|
}
|
|
|
|
public List<DedupConfig> getConfigurations(ISLookUpService isLookUpService, String orchestrator)
|
|
throws ISLookUpException, DocumentException, IOException {
|
|
|
|
final String xquery = String.format("/RESOURCE_PROFILE[.//DEDUPLICATION/ACTION_SET/@id = '%s']", orchestrator);
|
|
|
|
String orchestratorProfile = isLookUpService.getResourceProfileByQuery(xquery);
|
|
|
|
final Document doc = new SAXReader().read(new StringReader(orchestratorProfile));
|
|
|
|
final String actionSetId = doc.valueOf("//DEDUPLICATION/ACTION_SET/@id");
|
|
|
|
final List<DedupConfig> configurations = new ArrayList<>();
|
|
|
|
for (final Object o : doc.selectNodes("//SCAN_SEQUENCE/SCAN")) {
|
|
configurations.add(loadConfig(isLookUpService, actionSetId, o));
|
|
}
|
|
|
|
return configurations;
|
|
}
|
|
|
|
private DedupConfig loadConfig(
|
|
final ISLookUpService isLookUpService, final String actionSetId, final Object o)
|
|
throws ISLookUpException, IOException {
|
|
final Element s = (Element) o;
|
|
final String configProfileId = s.attributeValue("id");
|
|
final String conf = isLookUpService
|
|
.getResourceProfileByQuery(
|
|
String
|
|
.format(
|
|
"for $x in /RESOURCE_PROFILE[.//RESOURCE_IDENTIFIER/@value = '%s'] return $x//DEDUPLICATION/text()",
|
|
configProfileId));
|
|
|
|
DedupConfig dedupConfig = new ObjectMapper().readValue(conf, DedupConfig.class);
|
|
dedupConfig.getPace().initModel();
|
|
dedupConfig.getPace().initTranslationMap();
|
|
dedupConfig.getWf().setConfigurationId(actionSetId);
|
|
|
|
return dedupConfig;
|
|
}
|
|
|
|
abstract void run(ISLookUpService isLookUpService)
|
|
throws DocumentException, IOException, ISLookUpException;
|
|
|
|
protected static SparkSession getSparkSession(SparkConf conf) {
|
|
return SparkSession.builder().config(conf).getOrCreate();
|
|
}
|
|
|
|
protected static <T> void save(Dataset<T> dataset, String outPath, SaveMode mode) {
|
|
dataset.write().option("compression", "gzip").mode(mode).json(outPath);
|
|
}
|
|
|
|
protected static void removeOutputDir(SparkSession spark, String path) {
|
|
HdfsSupport.remove(path, spark.sparkContext().hadoopConfiguration());
|
|
}
|
|
}
|