sabeel
/
dnet-hadoop
forked from D-Net/dnet-hadoop
1
0
Fork 0
Code Pull Requests Activity

[FOS]fixed doi value in test

This commit is contained in:
Miriam Baglioni 2021-12-22 23:10:16 +01:00
parent 6116fc5d40
commit 10579c0dd0
2 changed files with 146 additions and 154 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

40
dhp-workflows/dhp-aggregation/src/main/java/eu/dnetlib/dhp/actionmanager/createunresolvedentities/PrepareFOSSparkJob.java
View File

@ -65,35 +65,35 @@ public class PrepareFOSSparkJob implements Serializable {
private static void distributeFOSdois(SparkSession spark, String sourcePath, String outputPath) {
Dataset<FOSDataModel> fosDataset = readPath(spark, sourcePath, FOSDataModel.class);
fosDataset.groupByKey((MapFunction<FOSDataModel,String>)v->v.getDoi(), Encoders.STRING())
.mapGroups((MapGroupsFunction<String, FOSDataModel, Result>)(k,it)->{
Result r = new Result();
FOSDataModel first = it.next();
r.setId(DHPUtils.generateUnresolvedIdentifier(first.getDoi(), DOI));
HashSet<String> level1 = new HashSet<>();
HashSet<String> level2 = new HashSet<>();
HashSet<String> level3 = new HashSet<>();
addLevels(level1, level2, level3, first);
it.forEachRemaining(v -> addLevels(level1, level2, level3, v));
List<StructuredProperty>sbjs = new ArrayList<>();
level1.forEach(l -> sbjs.add(getSubject(l, FOS_CLASS_ID, FOS_CLASS_NAME)));
level2.forEach(l -> sbjs.add(getSubject(l, FOS_CLASS_ID, FOS_CLASS_NAME)));
level3.forEach(l -> sbjs.add(getSubject(l, FOS_CLASS_ID, FOS_CLASS_NAME)));
r.setSubject(sbjs);
return r;
}, Encoders.bean(Result.class))
fosDataset
.groupByKey((MapFunction<FOSDataModel, String>) v -> v.getDoi(), Encoders.STRING())
.mapGroups((MapGroupsFunction<String, FOSDataModel, Result>) (k, it) -> {
Result r = new Result();
FOSDataModel first = it.next();
r.setId(DHPUtils.generateUnresolvedIdentifier(first.getDoi(), DOI));
HashSet<String> level1 = new HashSet<>();
HashSet<String> level2 = new HashSet<>();
HashSet<String> level3 = new HashSet<>();
addLevels(level1, level2, level3, first);
it.forEachRemaining(v -> addLevels(level1, level2, level3, v));
List<StructuredProperty> sbjs = new ArrayList<>();
level1.forEach(l -> sbjs.add(getSubject(l, FOS_CLASS_ID, FOS_CLASS_NAME)));
level2.forEach(l -> sbjs.add(getSubject(l, FOS_CLASS_ID, FOS_CLASS_NAME)));
level3.forEach(l -> sbjs.add(getSubject(l, FOS_CLASS_ID, FOS_CLASS_NAME)));
r.setSubject(sbjs);
return r;
}, Encoders.bean(Result.class))
.write()
.mode(SaveMode.Overwrite)
.option("compression", "gzip")
.json(outputPath + "/fos");
}
private static void addLevels(HashSet<String> level1, HashSet<String> level2, HashSet<String> level3, FOSDataModel first) {
private static void addLevels(HashSet<String> level1, HashSet<String> level2, HashSet<String> level3,
FOSDataModel first) {
level1.add(first.getLevel1());
level2.add(first.getLevel2());
level3.add(first.getLevel3());
}
}

258
dhp-workflows/dhp-aggregation/src/test/java/eu/dnetlib/dhp/actionmanager/createunresolvedentities/ProduceTest.java
View File

@ -67,29 +67,29 @@ public class ProduceTest {
}
@Test
void produceTestSubjects()throws Exception{
void produceTestSubjects() throws Exception {
JavaRDD<Result> tmp = getResultJavaRDD();
List<StructuredProperty> sbjs = tmp
.filter(row -> row.getSubject()!= null && row.getSubject().size()>0)
.flatMap(row -> row.getSubject().iterator())
.collect();
.filter(row -> row.getSubject() != null && row.getSubject().size() > 0)
.flatMap(row -> row.getSubject().iterator())
.collect();
sbjs.forEach(sbj -> Assertions.assertEquals("FOS", sbj.getQualifier().getClassid()));
sbjs
.forEach(
sbj -> Assertions
.assertEquals(
"Fields of Science and Technology classification", sbj.getQualifier().getClassname()));
.forEach(
sbj -> Assertions
.assertEquals(
"Fields of Science and Technology classification", sbj.getQualifier().getClassname()));
sbjs
.forEach(
sbj -> Assertions
.assertEquals(ModelConstants.DNET_SUBJECT_TYPOLOGIES, sbj.getQualifier().getSchemeid()));
.forEach(
sbj -> Assertions
.assertEquals(ModelConstants.DNET_SUBJECT_TYPOLOGIES, sbj.getQualifier().getSchemeid()));
sbjs
.forEach(
sbj -> Assertions
.assertEquals(ModelConstants.DNET_SUBJECT_TYPOLOGIES, sbj.getQualifier().getSchemename()));
.forEach(
sbj -> Assertions
.assertEquals(ModelConstants.DNET_SUBJECT_TYPOLOGIES, sbj.getQualifier().getSchemename()));
sbjs.forEach(sbj -> Assertions.assertEquals(false, sbj.getDataInfo().getDeletedbyinference()));
sbjs.forEach(sbj -> Assertions.assertEquals(true, sbj.getDataInfo().getInferred()));
@ -97,38 +97,36 @@ public class ProduceTest {
sbjs.forEach(sbj -> Assertions.assertEquals("", sbj.getDataInfo().getTrust()));
sbjs.forEach(sbj -> Assertions.assertEquals("update", sbj.getDataInfo().getInferenceprovenance()));
sbjs
.forEach(
sbj -> Assertions.assertEquals("subject:fos", sbj.getDataInfo().getProvenanceaction().getClassid()));
.forEach(
sbj -> Assertions.assertEquals("subject:fos", sbj.getDataInfo().getProvenanceaction().getClassid()));
sbjs
.forEach(
sbj -> Assertions
.assertEquals("Inferred by OpenAIRE", sbj.getDataInfo().getProvenanceaction().getClassname()));
.forEach(
sbj -> Assertions
.assertEquals("Inferred by OpenAIRE", sbj.getDataInfo().getProvenanceaction().getClassname()));
sbjs
.forEach(
sbj -> Assertions
.assertEquals(
ModelConstants.DNET_PROVENANCE_ACTIONS, sbj.getDataInfo().getProvenanceaction().getSchemeid()));
.forEach(
sbj -> Assertions
.assertEquals(
ModelConstants.DNET_PROVENANCE_ACTIONS, sbj.getDataInfo().getProvenanceaction().getSchemeid()));
sbjs
.forEach(
sbj -> Assertions
.assertEquals(
ModelConstants.DNET_PROVENANCE_ACTIONS,
sbj.getDataInfo().getProvenanceaction().getSchemename()));
.forEach(
sbj -> Assertions
.assertEquals(
ModelConstants.DNET_PROVENANCE_ACTIONS,
sbj.getDataInfo().getProvenanceaction().getSchemename()));
}
@Test
void produceTestMeasuress()throws Exception{
void produceTestMeasuress() throws Exception {
JavaRDD<Result> tmp = getResultJavaRDD();
List<KeyValue> mes = tmp
.filter(row -> row.getInstance()!= null && row.getInstance().size()>0)
.flatMap(row -> row.getInstance().iterator())
.flatMap(i->i.getMeasures().iterator())
.flatMap(m ->m.getUnit().iterator())
.collect();
.filter(row -> row.getInstance() != null && row.getInstance().size() > 0)
.flatMap(row -> row.getInstance().iterator())
.flatMap(i -> i.getMeasures().iterator())
.flatMap(m -> m.getUnit().iterator())
.collect();
mes.forEach(sbj -> Assertions.assertEquals(false, sbj.getDataInfo().getDeletedbyinference()));
mes.forEach(sbj -> Assertions.assertEquals(true, sbj.getDataInfo().getInferred()));
@ -136,119 +134,118 @@ public class ProduceTest {
mes.forEach(sbj -> Assertions.assertEquals("", sbj.getDataInfo().getTrust()));
mes.forEach(sbj -> Assertions.assertEquals("update", sbj.getDataInfo().getInferenceprovenance()));
mes
.forEach(
sbj -> Assertions.assertEquals("measure:bip", sbj.getDataInfo().getProvenanceaction().getClassid()));
.forEach(
sbj -> Assertions.assertEquals("measure:bip", sbj.getDataInfo().getProvenanceaction().getClassid()));
mes
.forEach(
sbj -> Assertions
.assertEquals("Inferred by OpenAIRE", sbj.getDataInfo().getProvenanceaction().getClassname()));
.forEach(
sbj -> Assertions
.assertEquals("Inferred by OpenAIRE", sbj.getDataInfo().getProvenanceaction().getClassname()));
mes
.forEach(
sbj -> Assertions
.assertEquals(
ModelConstants.DNET_PROVENANCE_ACTIONS, sbj.getDataInfo().getProvenanceaction().getSchemeid()));
.forEach(
sbj -> Assertions
.assertEquals(
ModelConstants.DNET_PROVENANCE_ACTIONS, sbj.getDataInfo().getProvenanceaction().getSchemeid()));
mes
.forEach(
sbj -> Assertions
.assertEquals(
ModelConstants.DNET_PROVENANCE_ACTIONS,
sbj.getDataInfo().getProvenanceaction().getSchemename()));
.forEach(
sbj -> Assertions
.assertEquals(
ModelConstants.DNET_PROVENANCE_ACTIONS,
sbj.getDataInfo().getProvenanceaction().getSchemename()));
}
@Test
void produceTest6Subjects() throws Exception{
void produceTest6Subjects() throws Exception {
final String doi = "unresolved::10.3390/s18072310::doi";
JavaRDD<Result> tmp = getResultJavaRDD();
Assertions
.assertEquals(
6, tmp
.filter(row -> row.getId().equals(doi))
.collect()
.get(0)
.getSubject()
.size());
.assertEquals(
6, tmp
.filter(row -> row.getId().equals(doi))
.collect()
.get(0)
.getSubject()
.size());
List<StructuredProperty> sbjs = tmp
.filter(row -> row.getId().equals(doi))
.flatMap(row -> row.getSubject().iterator())
.collect();
.filter(row -> row.getId().equals(doi))
.flatMap(row -> row.getSubject().iterator())
.collect();
Assertions
.assertEquals(
true, sbjs.stream().anyMatch(sbj -> sbj.getValue().equals("04 agricultural and veterinary sciences")));
.assertEquals(
true, sbjs.stream().anyMatch(sbj -> sbj.getValue().equals("04 agricultural and veterinary sciences")));
Assertions
.assertEquals(
true, sbjs.stream().anyMatch(sbj -> sbj.getValue().equals("0404 agricultural biotechnology")));
.assertEquals(
true, sbjs.stream().anyMatch(sbj -> sbj.getValue().equals("0404 agricultural biotechnology")));
Assertions.assertEquals(true, sbjs.stream().anyMatch(sbj -> sbj.getValue().equals("040502 food science")));
Assertions
.assertEquals(true, sbjs.stream().anyMatch(sbj -> sbj.getValue().equals("03 medical and health sciences")));
.assertEquals(true, sbjs.stream().anyMatch(sbj -> sbj.getValue().equals("03 medical and health sciences")));
Assertions.assertEquals(true, sbjs.stream().anyMatch(sbj -> sbj.getValue().equals("0303 health sciences")));
Assertions
.assertEquals(true, sbjs.stream().anyMatch(sbj -> sbj.getValue().equals("030309 nutrition & dietetics")));
.assertEquals(true, sbjs.stream().anyMatch(sbj -> sbj.getValue().equals("030309 nutrition & dietetics")));
}
@Test
void produceTest3Measures()throws Exception{
void produceTest3Measures() throws Exception {
final String doi = "unresolved::10.3390/s18072310::doi";
JavaRDD<Result> tmp = getResultJavaRDD();
Assertions
.assertEquals(
3, tmp
.filter(row -> row.getId().equals(doi))
.collect()
.get(0)
.getInstance()
.get(0)
.getMeasures()
.size());
.assertEquals(
3, tmp
.filter(row -> row.getId().equals(doi))
.collect()
.get(0)
.getInstance()
.get(0)
.getMeasures()
.size());
List<Measure> measures = tmp
.filter(row -> row.getId().equals(doi))
.flatMap(row -> row.getInstance().iterator())
.flatMap(inst -> inst.getMeasures().iterator())
.collect();
.filter(row -> row.getId().equals(doi))
.flatMap(row -> row.getInstance().iterator())
.flatMap(inst -> inst.getMeasures().iterator())
.collect();
Assertions
.assertEquals(
"7.5597134689e-09", measures
.stream()
.filter(mes -> mes.getId().equals("influence"))
.collect(Collectors.toList())
.get(0)
.getUnit()
.get(0)
.getValue());
.assertEquals(
"7.5597134689e-09", measures
.stream()
.filter(mes -> mes.getId().equals("influence"))
.collect(Collectors.toList())
.get(0)
.getUnit()
.get(0)
.getValue());
Assertions
.assertEquals(
"4.903880192", measures
.stream()
.filter(mes -> mes.getId().equals("popularity_alt"))
.collect(Collectors.toList())
.get(0)
.getUnit()
.get(0)
.getValue());
.assertEquals(
"4.903880192", measures
.stream()
.filter(mes -> mes.getId().equals("popularity_alt"))
.collect(Collectors.toList())
.get(0)
.getUnit()
.get(0)
.getValue());
Assertions
.assertEquals(
"1.17977512835e-08", measures
.stream()
.filter(mes -> mes.getId().equals("popularity"))
.collect(Collectors.toList())
.get(0)
.getUnit()
.get(0)
.getValue());
.assertEquals(
"1.17977512835e-08", measures
.stream()
.filter(mes -> mes.getId().equals("popularity"))
.collect(Collectors.toList())
.get(0)
.getUnit()
.get(0)
.getValue());
}
@Test
void produceTestSomeNumbers() throws Exception {
@ -316,43 +313,38 @@ public class ProduceTest {
.map(item -> OBJECT_MAPPER.readValue(item, Result.class));
}
@Test
void prepareTest5Subjects()throws Exception{
final String doi = "unresolved::10.3390/s18072310::doi";
void prepareTest5Subjects() throws Exception {
final String doi = "unresolved::10.1063/5.0032658::doi";
JavaRDD<Result> tmp = getResultJavaRDD();
Assertions.assertEquals(1, tmp.filter(row -> row.getId().equals(doi)).count());
Assertions
.assertEquals(
5, tmp
.filter(row -> row.getId().equals(doi))
.collect()
.get(0)
.getSubject()
.size());
.assertEquals(
5, tmp
.filter(row -> row.getId().equals(doi))
.collect()
.get(0)
.getSubject()
.size());
List<StructuredProperty> sbjs = tmp
.filter(row -> row.getId().equals(doi))
.flatMap(row -> row.getSubject().iterator())
.collect();
.filter(row -> row.getId().equals(doi))
.flatMap(row -> row.getSubject().iterator())
.collect();
Assertions
.assertEquals(
true, sbjs.stream().anyMatch(sbj -> sbj.getValue().equals("01 natural sciences")));
.assertEquals(
true, sbjs.stream().anyMatch(sbj -> sbj.getValue().equals("01 natural sciences")));
Assertions.assertEquals(true, sbjs.stream().anyMatch(sbj -> sbj.getValue().equals("0103 physical sciences")));
Assertions
.assertEquals(true, sbjs.stream().anyMatch(sbj -> sbj.getValue().equals("010304 chemical physics")));
.assertEquals(true, sbjs.stream().anyMatch(sbj -> sbj.getValue().equals("010304 chemical physics")));
Assertions.assertEquals(true, sbjs.stream().anyMatch(sbj -> sbj.getValue().equals("0104 chemical sciences")));
Assertions
.assertEquals(true, sbjs.stream().anyMatch(sbj -> sbj.getValue().equals("010402 general chemistry")));
.assertEquals(true, sbjs.stream().anyMatch(sbj -> sbj.getValue().equals("010402 general chemistry")));
}