introduced configuration param outputFormat: HDFS | SOLR

This commit is contained in:
Claudio Atzori 2020-11-19 10:34:28 +01:00
parent 5218718e8b
commit 0374d34c3e
5 changed files with 122 additions and 74 deletions

View File

@ -1,14 +1,21 @@
package eu.dnetlib.dhp.oa.provision;
import com.lucidworks.spark.util.SolrSupport;
import eu.dnetlib.dhp.application.ArgumentApplicationParser;
import eu.dnetlib.dhp.oa.provision.model.SerializableSolrInputDocument;
import eu.dnetlib.dhp.oa.provision.utils.ISLookupClient;
import eu.dnetlib.dhp.oa.provision.utils.StreamingInputDocumentFactory;
import eu.dnetlib.dhp.utils.ISLookupClientFactory;
import eu.dnetlib.dhp.utils.saxon.SaxonTransformerFactory;
import eu.dnetlib.enabling.is.lookup.rmi.ISLookUpException;
import static eu.dnetlib.dhp.common.SparkSessionSupport.runWithSparkSession;
import java.io.IOException;
import java.io.StringReader;
import java.io.StringWriter;
import java.text.SimpleDateFormat;
import java.util.Date;
import java.util.Optional;
import javax.swing.text.html.Option;
import javax.xml.transform.Transformer;
import javax.xml.transform.TransformerException;
import javax.xml.transform.stream.StreamResult;
import javax.xml.transform.stream.StreamSource;
import org.apache.commons.io.IOUtils;
import org.apache.commons.lang3.StringUtils;
import org.apache.hadoop.io.Text;
@ -22,23 +29,24 @@ import org.apache.spark.sql.SparkSession;
import org.slf4j.Logger;
import org.slf4j.LoggerFactory;
import javax.xml.transform.Transformer;
import javax.xml.transform.TransformerException;
import javax.xml.transform.stream.StreamResult;
import javax.xml.transform.stream.StreamSource;
import java.io.IOException;
import java.io.StringReader;
import java.io.StringWriter;
import java.text.SimpleDateFormat;
import java.util.Date;
import java.util.Optional;
import com.lucidworks.spark.util.SolrSupport;
import static eu.dnetlib.dhp.common.SparkSessionSupport.runWithSparkSession;
import eu.dnetlib.dhp.application.ArgumentApplicationParser;
import eu.dnetlib.dhp.oa.provision.model.SerializableSolrInputDocument;
import eu.dnetlib.dhp.oa.provision.utils.ISLookupClient;
import eu.dnetlib.dhp.oa.provision.utils.StreamingInputDocumentFactory;
import eu.dnetlib.dhp.utils.ISLookupClientFactory;
import eu.dnetlib.dhp.utils.saxon.SaxonTransformerFactory;
import eu.dnetlib.enabling.is.lookup.rmi.ISLookUpException;
public class XmlIndexingJob {
private static final Logger log = LoggerFactory.getLogger(XmlIndexingJob.class);
public enum OutputFormat {
SOLR, HDFS
}
private static final Integer DEFAULT_BATCH_SIZE = 1000;
protected static final String DATE_FORMAT = "yyyy-MM-dd'T'hh:mm:ss'Z'";
@ -49,6 +57,8 @@ public class XmlIndexingJob {
private int batchSize;
private OutputFormat outputFormat;
private String outputPath;
private SparkSession spark;
@ -75,17 +85,28 @@ public class XmlIndexingJob {
final String format = parser.get("format");
log.info("format: {}", format);
final String outputPath = Optional.ofNullable(parser.get("outputPath"))
final String outputPath = Optional
.ofNullable(parser.get("outputPath"))
.map(StringUtils::trim)
.orElse(null);
log.info("outputPath: {}", outputPath);
final Integer batchSize = parser.getObjectMap().containsKey("batchSize")
? Integer.valueOf(parser.get("batchSize"))
: DEFAULT_BATCH_SIZE;
final Integer batchSize = Optional
.ofNullable(parser.get("batchSize"))
.map(Integer::valueOf)
.orElse(DEFAULT_BATCH_SIZE);
log.info("batchSize: {}", batchSize);
final OutputFormat outputFormat = Optional
.ofNullable(parser.get("outputFormat"))
.map(OutputFormat::valueOf)
.orElse(OutputFormat.SOLR);
log.info("outputFormat: {}", outputFormat);
final SparkConf conf = new SparkConf();
conf.registerKryoClasses(new Class[] { SerializableSolrInputDocument.class });
conf.registerKryoClasses(new Class[] {
SerializableSolrInputDocument.class
});
runWithSparkSession(
conf,
@ -94,15 +115,18 @@ public class XmlIndexingJob {
final String isLookupUrl = parser.get("isLookupUrl");
log.info("isLookupUrl: {}", isLookupUrl);
final ISLookupClient isLookup = new ISLookupClient(ISLookupClientFactory.getLookUpService(isLookupUrl));
new XmlIndexingJob(spark, inputPath, format, batchSize, outputPath).run(isLookup);
new XmlIndexingJob(spark, inputPath, format, batchSize, outputFormat, outputPath).run(isLookup);
});
}
public XmlIndexingJob(SparkSession spark, String inputPath, String format, Integer batchSize, String outputPath) {
public XmlIndexingJob(SparkSession spark, String inputPath, String format, Integer batchSize,
OutputFormat outputFormat,
String outputPath) {
this.spark = spark;
this.inputPath = inputPath;
this.format = format;
this.batchSize = batchSize;
this.outputFormat = outputFormat;
this.outputPath = outputPath;
}
@ -131,16 +155,22 @@ public class XmlIndexingJob {
.map(s -> toIndexRecord(SaxonTransformerFactory.newInstance(indexRecordXslt), s))
.map(s -> new StreamingInputDocumentFactory(version, dsId).parseDocument(s));
if (StringUtils.isNotBlank(outputPath)) {
spark.createDataset(
switch (outputFormat) {
case SOLR:
final String collection = ProvisionConstants.getCollectionName(format);
SolrSupport.indexDocs(zkHost, collection, batchSize, docs.rdd());
break;
case HDFS:
spark
.createDataset(
docs.map(s -> new SerializableSolrInputDocument(s)).rdd(),
Encoders.kryo(SerializableSolrInputDocument.class))
.write()
.mode(SaveMode.Overwrite)
.parquet(outputPath);
} else {
final String collection = ProvisionConstants.getCollectionName(format);
SolrSupport.indexDocs(zkHost, collection, batchSize, docs.rdd());
break;
default:
throw new IllegalArgumentException("invalid outputFormat: " + outputFormat);
}
}

View File

@ -1,11 +1,12 @@
package eu.dnetlib.dhp.oa.provision.model;
import org.apache.solr.common.SolrInputDocument;
import org.apache.solr.common.SolrInputField;
package eu.dnetlib.dhp.oa.provision.model;
import java.util.HashMap;
import java.util.Map;
import org.apache.solr.common.SolrInputDocument;
import org.apache.solr.common.SolrInputField;
/**
* Wrapper class needed to make the SolrInputDocument compatible with the Kryo serialization mechanism.
*/

View File

@ -24,7 +24,13 @@
"paramRequired": false
},
{
"paramName": "o",
"paramName": "of",
"paramLongName": "outputFormat",
"paramDescription": "decides the job output format, SOLR | HDFS",
"paramRequired": false
},
{
"paramName": "op",
"paramLongName": "outputPath",
"paramDescription": "path on hdfs activating an alternative output for the SolrInputDocuments",
"paramRequired": false

View File

@ -42,6 +42,7 @@
<value>*:*</value>
<description>query used in the deleted by query operation</description>
</property>
<property>
<name>sparkDriverMemoryForJoining</name>
<description>memory for driver process</description>
@ -638,7 +639,8 @@
<arg>--isLookupUrl</arg><arg>${isLookupUrl}</arg>
<arg>--format</arg><arg>${format}</arg>
<arg>--batchSize</arg><arg>${batchSize}</arg>
<arg>--outputPath</arg><arg>${outputPath}</arg>
<arg>--outputFormat</arg><arg>${outputFormat}</arg>
<arg>--outputPath</arg><arg>${workingDir}/solr_documents</arg>
</spark>
<ok to="commit_solr_collection"/>
<error to="Kill"/>

View File

@ -1,10 +1,10 @@
package eu.dnetlib.dhp.oa.provision;
import eu.dnetlib.dhp.oa.provision.model.SerializableSolrInputDocument;
import eu.dnetlib.dhp.oa.provision.utils.ISLookupClient;
import eu.dnetlib.enabling.is.lookup.rmi.ISLookUpException;
import eu.dnetlib.enabling.is.lookup.rmi.ISLookUpService;
import java.io.IOException;
import java.io.StringReader;
import java.net.URI;
import org.apache.commons.io.IOUtils;
import org.apache.hadoop.io.Text;
import org.apache.solr.client.solrj.SolrQuery;
@ -25,9 +25,10 @@ import org.mockito.Mock;
import org.mockito.Mockito;
import org.mockito.junit.jupiter.MockitoExtension;
import java.io.IOException;
import java.io.StringReader;
import java.net.URI;
import eu.dnetlib.dhp.oa.provision.model.SerializableSolrInputDocument;
import eu.dnetlib.dhp.oa.provision.utils.ISLookupClient;
import eu.dnetlib.enabling.is.lookup.rmi.ISLookUpException;
import eu.dnetlib.enabling.is.lookup.rmi.ISLookUpService;
@ExtendWith(MockitoExtension.class)
public class XmlIndexingJobTest extends SolrTest {
@ -65,7 +66,9 @@ public class XmlIndexingJobTest extends SolrTest {
SparkConf conf = new SparkConf();
conf.setAppName(XmlIndexingJobTest.class.getSimpleName());
conf.registerKryoClasses(new Class[] { SerializableSolrInputDocument.class });
conf.registerKryoClasses(new Class[] {
SerializableSolrInputDocument.class
});
conf.setMaster("local[1]");
conf.set("spark.driver.host", "localhost");
@ -95,13 +98,16 @@ public class XmlIndexingJobTest extends SolrTest {
.sequenceFile(inputPath, Text.class, Text.class)
.count();
new XmlIndexingJob(spark, inputPath, FORMAT, batchSize, null).run(isLookupClient);
new XmlIndexingJob(spark, inputPath, FORMAT, batchSize, XmlIndexingJob.OutputFormat.SOLR, null)
.run(isLookupClient);
Assertions.assertEquals(0, miniCluster.getSolrClient().commit().getStatus());
QueryResponse rsp = miniCluster.getSolrClient().query(new SolrQuery().add(CommonParams.Q, "*:*"));
Assertions.assertEquals(nRecord, rsp.getResults().getNumFound(),
Assertions
.assertEquals(
nRecord, rsp.getResults().getNumFound(),
"the number of indexed records should be equal to the number of input records");
}
@ -118,13 +124,16 @@ public class XmlIndexingJobTest extends SolrTest {
long xmlIdUnique = xmlRecords
.map(t -> t._2().toString())
.map(s -> new SAXReader().read(new StringReader(s)).valueOf(ID_XPATH))
.distinct().count();
.distinct()
.count();
Assertions.assertEquals(nRecord, xmlIdUnique, "IDs should be unique among input records");
final String outputPath = workingDir.resolve("outputPath").toAbsolutePath().toString();
new XmlIndexingJob(spark, inputPath, FORMAT, batchSize, outputPath).run(isLookupClient);
new XmlIndexingJob(spark, inputPath, FORMAT, batchSize, XmlIndexingJob.OutputFormat.HDFS, outputPath)
.run(isLookupClient);
final Dataset<SerializableSolrInputDocument> solrDocs = spark.read()
final Dataset<SerializableSolrInputDocument> solrDocs = spark
.read()
.load(outputPath)
.as(Encoders.kryo(SerializableSolrInputDocument.class));
long docIdUnique = solrDocs.map((MapFunction<SerializableSolrInputDocument, String>) doc -> {