D-Net
/
dnet-hadoop
14
5
Fork 11
Code Issues 14 Pull Requests 13 Projects 2 Releases 2 Wiki 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

12
dhp-workflows/dhp-aggregation/src/main/java/eu/dnetlib/dhp/actionmanager/createunresolvedentities/PrepareFOSSparkJob.java
View File

@ -65,8 +65,9 @@ public class PrepareFOSSparkJob implements Serializable {
private static void distributeFOSdois(SparkSession spark, String sourcePath, String outputPath) { private static void distributeFOSdois(SparkSession spark, String sourcePath, String outputPath) {
Dataset<FOSDataModel> fosDataset = readPath(spark, sourcePath, FOSDataModel.class); Dataset<FOSDataModel> fosDataset = readPath(spark, sourcePath, FOSDataModel.class);
fosDataset.groupByKey((MapFunction<FOSDataModel,String>)v->v.getDoi(), Encoders.STRING()) fosDataset
.mapGroups((MapGroupsFunction<String, FOSDataModel, Result>)(k,it)->{ .groupByKey((MapFunction<FOSDataModel, String>) v -> v.getDoi(), Encoders.STRING())
.mapGroups((MapGroupsFunction<String, FOSDataModel, Result>) (k, it) -> {
Result r = new Result(); Result r = new Result();
FOSDataModel first = it.next(); FOSDataModel first = it.next();
r.setId(DHPUtils.generateUnresolvedIdentifier(first.getDoi(), DOI)); r.setId(DHPUtils.generateUnresolvedIdentifier(first.getDoi(), DOI));
@ -75,7 +76,7 @@ public class PrepareFOSSparkJob implements Serializable {
HashSet<String> level3 = new HashSet<>(); HashSet<String> level3 = new HashSet<>();
addLevels(level1, level2, level3, first); addLevels(level1, level2, level3, first);
it.forEachRemaining(v -> addLevels(level1, level2, level3, v)); it.forEachRemaining(v -> addLevels(level1, level2, level3, v));
List<StructuredProperty>sbjs = new ArrayList<>(); List<StructuredProperty> sbjs = new ArrayList<>();
level1.forEach(l -> sbjs.add(getSubject(l, FOS_CLASS_ID, FOS_CLASS_NAME))); 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))); 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))); level3.forEach(l -> sbjs.add(getSubject(l, FOS_CLASS_ID, FOS_CLASS_NAME)));
@ -88,12 +89,11 @@ public class PrepareFOSSparkJob implements Serializable {
.json(outputPath + "/fos"); .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()); level1.add(first.getLevel1());
level2.add(first.getLevel2()); level2.add(first.getLevel2());
level3.add(first.getLevel3()); level3.add(first.getLevel3());
} }
} }

34
dhp-workflows/dhp-aggregation/src/test/java/eu/dnetlib/dhp/actionmanager/createunresolvedentities/ProduceTest.java
View File

@ -67,12 +67,12 @@ public class ProduceTest {
} }
@Test @Test
void produceTestSubjects()throws Exception{ void produceTestSubjects() throws Exception {
JavaRDD<Result> tmp = getResultJavaRDD(); JavaRDD<Result> tmp = getResultJavaRDD();
List<StructuredProperty> sbjs = tmp List<StructuredProperty> sbjs = tmp
.filter(row -> row.getSubject()!= null && row.getSubject().size()>0) .filter(row -> row.getSubject() != null && row.getSubject().size() > 0)
.flatMap(row -> row.getSubject().iterator()) .flatMap(row -> row.getSubject().iterator())
.collect(); .collect();
@ -117,19 +117,17 @@ public class ProduceTest {
} }
@Test @Test
void produceTestMeasuress()throws Exception{ void produceTestMeasuress() throws Exception {
JavaRDD<Result> tmp = getResultJavaRDD(); JavaRDD<Result> tmp = getResultJavaRDD();
List<KeyValue> mes = tmp List<KeyValue> mes = tmp
.filter(row -> row.getInstance()!= null && row.getInstance().size()>0) .filter(row -> row.getInstance() != null && row.getInstance().size() > 0)
.flatMap(row -> row.getInstance().iterator()) .flatMap(row -> row.getInstance().iterator())
.flatMap(i->i.getMeasures().iterator()) .flatMap(i -> i.getMeasures().iterator())
.flatMap(m ->m.getUnit().iterator()) .flatMap(m -> m.getUnit().iterator())
.collect(); .collect();
mes.forEach(sbj -> Assertions.assertEquals(false, sbj.getDataInfo().getDeletedbyinference())); mes.forEach(sbj -> Assertions.assertEquals(false, sbj.getDataInfo().getDeletedbyinference()));
mes.forEach(sbj -> Assertions.assertEquals(true, sbj.getDataInfo().getInferred())); mes.forEach(sbj -> Assertions.assertEquals(true, sbj.getDataInfo().getInferred()));
mes.forEach(sbj -> Assertions.assertEquals(false, sbj.getDataInfo().getInvisible())); mes.forEach(sbj -> Assertions.assertEquals(false, sbj.getDataInfo().getInvisible()));
@ -154,8 +152,9 @@ public class ProduceTest {
ModelConstants.DNET_PROVENANCE_ACTIONS, ModelConstants.DNET_PROVENANCE_ACTIONS,
sbj.getDataInfo().getProvenanceaction().getSchemename())); sbj.getDataInfo().getProvenanceaction().getSchemename()));
} }
@Test @Test
void produceTest6Subjects() throws Exception{ void produceTest6Subjects() throws Exception {
final String doi = "unresolved::10.3390/s18072310::doi"; final String doi = "unresolved::10.3390/s18072310::doi";
JavaRDD<Result> tmp = getResultJavaRDD(); JavaRDD<Result> tmp = getResultJavaRDD();
@ -174,7 +173,6 @@ public class ProduceTest {
.flatMap(row -> row.getSubject().iterator()) .flatMap(row -> row.getSubject().iterator())
.collect(); .collect();
Assertions Assertions
.assertEquals( .assertEquals(
true, sbjs.stream().anyMatch(sbj -> sbj.getValue().equals("04 agricultural and veterinary sciences"))); true, sbjs.stream().anyMatch(sbj -> sbj.getValue().equals("04 agricultural and veterinary sciences")));
@ -190,11 +188,10 @@ public class ProduceTest {
Assertions 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 @Test
void produceTest3Measures()throws Exception{ void produceTest3Measures() throws Exception {
final String doi = "unresolved::10.3390/s18072310::doi"; final String doi = "unresolved::10.3390/s18072310::doi";
JavaRDD<Result> tmp = getResultJavaRDD(); JavaRDD<Result> tmp = getResultJavaRDD();
@ -209,7 +206,6 @@ public class ProduceTest {
.getMeasures() .getMeasures()
.size()); .size());
List<Measure> measures = tmp List<Measure> measures = tmp
.filter(row -> row.getId().equals(doi)) .filter(row -> row.getId().equals(doi))
.flatMap(row -> row.getInstance().iterator()) .flatMap(row -> row.getInstance().iterator())
@ -249,6 +245,7 @@ public class ProduceTest {
.getValue()); .getValue());
} }
@Test @Test
void produceTestSomeNumbers() throws Exception { void produceTestSomeNumbers() throws Exception {
@ -316,10 +313,10 @@ public class ProduceTest {
.map(item -> OBJECT_MAPPER.readValue(item, Result.class)); .map(item -> OBJECT_MAPPER.readValue(item, Result.class));
} }
@Test @Test
void prepareTest5Subjects()throws Exception{ void prepareTest5Subjects() throws Exception {
final String doi = "unresolved::10.3390/s18072310::doi"; final String doi = "unresolved::10.1063/5.0032658::doi";
JavaRDD<Result> tmp = getResultJavaRDD(); JavaRDD<Result> tmp = getResultJavaRDD();
Assertions.assertEquals(1, tmp.filter(row -> row.getId().equals(doi)).count()); Assertions.assertEquals(1, tmp.filter(row -> row.getId().equals(doi)).count());
@ -333,16 +330,11 @@ public class ProduceTest {
.getSubject() .getSubject()
.size()); .size());
List<StructuredProperty> sbjs = tmp List<StructuredProperty> sbjs = tmp
.filter(row -> row.getId().equals(doi)) .filter(row -> row.getId().equals(doi))
.flatMap(row -> row.getSubject().iterator()) .flatMap(row -> row.getSubject().iterator())
.collect(); .collect();
Assertions Assertions
.assertEquals( .assertEquals(
true, sbjs.stream().anyMatch(sbj -> sbj.getValue().equals("01 natural sciences"))); true, sbjs.stream().anyMatch(sbj -> sbj.getValue().equals("01 natural sciences")));