new implementation with one more serialization step
This commit is contained in:
parent
dd2e698a72
commit
42ad51577a
|
@ -0,0 +1,145 @@
|
|||
|
||||
package eu.dnetlib.dhp.countrypropagation;
|
||||
|
||||
import static eu.dnetlib.dhp.PropagationConstant.*;
|
||||
import static eu.dnetlib.dhp.common.SparkSessionSupport.runWithSparkSession;
|
||||
|
||||
import java.util.ArrayList;
|
||||
import java.util.HashSet;
|
||||
import java.util.List;
|
||||
import java.util.Optional;
|
||||
import java.util.stream.Collectors;
|
||||
|
||||
import org.apache.commons.io.IOUtils;
|
||||
import org.apache.hadoop.io.compress.GzipCodec;
|
||||
import org.apache.spark.SparkConf;
|
||||
import org.apache.spark.api.java.JavaPairRDD;
|
||||
import org.apache.spark.api.java.JavaRDD;
|
||||
import org.apache.spark.api.java.JavaSparkContext;
|
||||
import org.apache.spark.api.java.function.MapFunction;
|
||||
import org.apache.spark.sql.Dataset;
|
||||
import org.apache.spark.sql.Encoders;
|
||||
import org.apache.spark.sql.SaveMode;
|
||||
import org.apache.spark.sql.SparkSession;
|
||||
import org.slf4j.Logger;
|
||||
import org.slf4j.LoggerFactory;
|
||||
|
||||
import com.fasterxml.jackson.databind.ObjectMapper;
|
||||
|
||||
import eu.dnetlib.dhp.application.ArgumentApplicationParser;
|
||||
import eu.dnetlib.dhp.schema.oaf.Country;
|
||||
import eu.dnetlib.dhp.schema.oaf.KeyValue;
|
||||
import eu.dnetlib.dhp.schema.oaf.Result;
|
||||
import scala.Tuple2;
|
||||
|
||||
public class SparkCountryPropagationJob3 {
|
||||
|
||||
private static final Logger log = LoggerFactory.getLogger(SparkCountryPropagationJob3.class);
|
||||
|
||||
private static final ObjectMapper OBJECT_MAPPER = new ObjectMapper();
|
||||
|
||||
public static void main(String[] args) throws Exception {
|
||||
|
||||
String jsonConfiguration = IOUtils
|
||||
.toString(
|
||||
SparkCountryPropagationJob3.class
|
||||
.getResourceAsStream(
|
||||
"/eu/dnetlib/dhp/countrypropagation/input_countrypropagation_parameters.json"));
|
||||
|
||||
final ArgumentApplicationParser parser = new ArgumentApplicationParser(jsonConfiguration);
|
||||
|
||||
parser.parseArgument(args);
|
||||
|
||||
Boolean isSparkSessionManaged = isSparkSessionManaged(parser);
|
||||
log.info("isSparkSessionManaged: {}", isSparkSessionManaged);
|
||||
|
||||
String inputPath = parser.get("sourcePath");
|
||||
log.info("inputPath: {}", inputPath);
|
||||
|
||||
final String outputPath = parser.get("outputPath");
|
||||
log.info("outputPath: {}", outputPath);
|
||||
|
||||
final String resultClassName = parser.get("resultTableName");
|
||||
log.info("resultTableName: {}", resultClassName);
|
||||
|
||||
final Boolean saveGraph = Optional
|
||||
.ofNullable(parser.get("saveGraph"))
|
||||
.map(Boolean::valueOf)
|
||||
.orElse(Boolean.TRUE);
|
||||
log.info("saveGraph: {}", saveGraph);
|
||||
|
||||
Class<? extends Result> resultClazz = (Class<? extends Result>) Class.forName(resultClassName);
|
||||
|
||||
SparkConf conf = new SparkConf();
|
||||
|
||||
runWithSparkSession(
|
||||
conf,
|
||||
isSparkSessionManaged,
|
||||
spark -> {
|
||||
|
||||
execPropagation(
|
||||
spark,
|
||||
inputPath,
|
||||
outputPath,
|
||||
resultClazz,
|
||||
saveGraph);
|
||||
});
|
||||
}
|
||||
|
||||
private static <R extends Result> void execPropagation(
|
||||
SparkSession spark,
|
||||
String inputPath,
|
||||
String outputPath,
|
||||
Class<R> resultClazz,
|
||||
boolean saveGraph) {
|
||||
|
||||
if (saveGraph) {
|
||||
// updateResultTable(spark, potentialUpdates, inputPath, resultClazz, outputPath);
|
||||
log.info("Reading Graph table from: {}", inputPath);
|
||||
final JavaSparkContext sc = new JavaSparkContext(spark.sparkContext());
|
||||
JavaPairRDD<String, R> results = sc
|
||||
.textFile(inputPath)
|
||||
.map(r -> OBJECT_MAPPER.readValue(r, resultClazz))
|
||||
.mapToPair(r -> new Tuple2<>(r.getId(), r));
|
||||
|
||||
JavaPairRDD<String, R> tmp = results.reduceByKey((r1, r2) -> {
|
||||
if (r1 == null) {
|
||||
return r2;
|
||||
}
|
||||
if (r2 == null) {
|
||||
return r1;
|
||||
}
|
||||
if (Optional.ofNullable(r1.getCollectedfrom()).isPresent()) {
|
||||
r1.setCountry(getUnionCountries(r1.getCountry(), r2.getCountry()));
|
||||
return r1;
|
||||
}
|
||||
if (Optional.ofNullable(r2.getCollectedfrom()).isPresent()) {
|
||||
r2.setCountry(getUnionCountries(r1.getCountry(), r2.getCountry()));
|
||||
return r2;
|
||||
}
|
||||
r1.setCountry(getUnionCountries(r1.getCountry(), r2.getCountry()));
|
||||
return r1;
|
||||
});
|
||||
|
||||
tmp
|
||||
.map(c -> c._2())
|
||||
.map(r -> OBJECT_MAPPER.writeValueAsString(r))
|
||||
.saveAsTextFile(outputPath, GzipCodec.class);
|
||||
}
|
||||
}
|
||||
|
||||
private static List<Country> getUnionCountries(List<Country> country, List<Country> country1) {
|
||||
HashSet<String> countries = country
|
||||
.stream()
|
||||
.map(c -> c.getClassid())
|
||||
.collect(Collectors.toCollection(HashSet::new));
|
||||
country
|
||||
.addAll(
|
||||
country1
|
||||
.stream()
|
||||
.filter(c -> !(countries.contains(c.getClassid())))
|
||||
.collect(Collectors.toList()));
|
||||
return country;
|
||||
}
|
||||
|
||||
}
|
|
@ -0,0 +1,32 @@
|
|||
[
|
||||
{
|
||||
"paramName":"s",
|
||||
"paramLongName":"sourcePath",
|
||||
"paramDescription": "the path of the sequencial file to read",
|
||||
"paramRequired": true
|
||||
},
|
||||
{
|
||||
"paramName":"h",
|
||||
"paramLongName":"hive_metastore_uris",
|
||||
"paramDescription": "the hive metastore uris",
|
||||
"paramRequired": true
|
||||
},
|
||||
{
|
||||
"paramName":"tn",
|
||||
"paramLongName":"resultTableName",
|
||||
"paramDescription": "the name of the result table we are currently working on",
|
||||
"paramRequired": true
|
||||
},
|
||||
{
|
||||
"paramName": "p",
|
||||
"paramLongName": "preparedInfoPath",
|
||||
"paramDescription": "the path where prepared info have been stored",
|
||||
"paramRequired": true
|
||||
},
|
||||
{
|
||||
"paramName": "ssm",
|
||||
"paramLongName": "isSparkSessionManaged",
|
||||
"paramDescription": "true if the spark session is managed, false otherwise",
|
||||
"paramRequired": false
|
||||
}
|
||||
]
|
Loading…
Reference in New Issue