alessia.bardi
/
dnet-hadoop
forked from D-Net/dnet-hadoop
1
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

applying sonarLint suggestions

This commit is contained in:
Claudio Atzori 2021-06-23 09:53:29 +02:00
parent 2dd5449c13
commit 5edcc6832a
1 changed files with 22 additions and 26 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

48
dhp-workflows/dhp-aggregation/src/main/java/eu/dnetlib/dhp/actionmanager/project/PrepareProgramme.java
View File

@ -144,22 +144,7 @@ public class PrepareProgramme {
JavaRDD<CSVProgramme> h2020Programmes = programme
.toJavaRDD()
.mapToPair(csvProgramme -> new Tuple2<>(csvProgramme.getCode(), csvProgramme))
.reduceByKey((a, b) -> {
if (!a.getLanguage().equals("en")) {
if (b.getLanguage().equalsIgnoreCase("en")) {
a.setTitle(b.getTitle());
a.setLanguage(b.getLanguage());
}
}
if (StringUtils.isEmpty(a.getShortTitle())) {
if (!StringUtils.isEmpty(b.getShortTitle())) {
a.setShortTitle(b.getShortTitle());
}
}
return a;
})
.reduceByKey(PrepareProgramme::groupProgrammeByCode)
.map(p -> {
CSVProgramme csvProgramme = p._2();
String programmeTitle = csvProgramme.getTitle().trim();
@ -176,20 +161,31 @@ public class PrepareProgramme {
return csvProgramme;
});
// prepareClassification(h2020Programmes);
JavaSparkContext jsc = new JavaSparkContext(spark.sparkContext());
final JavaSparkContext jsc = JavaSparkContext.fromSparkContext(spark.sparkContext());
JavaRDD<CSVProgramme> rdd = jsc.parallelize(prepareClassification(h2020Programmes), 1);
rdd
.map(csvProgramme -> {
String tmp = OBJECT_MAPPER.writeValueAsString(csvProgramme);
return tmp;
})
.map(OBJECT_MAPPER::writeValueAsString)
.saveAsTextFile(outputPath);
}
private static CSVProgramme groupProgrammeByCode(CSVProgramme a, CSVProgramme b) {
if (!a.getLanguage().equals("en")) {
if (b.getLanguage().equalsIgnoreCase("en")) {
a.setTitle(b.getTitle());
a.setLanguage(b.getLanguage());
}
}
if (StringUtils.isEmpty(a.getShortTitle())) {
if (!StringUtils.isEmpty(b.getShortTitle())) {
a.setShortTitle(b.getShortTitle());
}
}
return a;
}
private static List<CSVProgramme> prepareClassification(JavaRDD<CSVProgramme> h2020Programmes) {
Object[] codedescription = h2020Programmes
.map(
@ -240,10 +236,10 @@ public class PrepareProgramme {
if (!ent.contains("Euratom")) {
String parent;
String tmp_key = tmp[0] + ".";
String tmpKey = tmp[0] + ".";
for (int i = 1; i < tmp.length - 1; i++) {
tmp_key += tmp[i] + ".";
parent = map.get(tmp_key)._1().toLowerCase().trim();
tmpKey += tmp[i] + ".";
parent = map.get(tmpKey)._1().toLowerCase().trim();
if (parent.contains("|")) {
parent = parent.substring(parent.lastIndexOf("|") + 1).trim();
}