D-Net
/
dnet-hadoop
14
5
Fork 11
Code Issues 14 Pull Requests 12 Projects 2 Releases 2 Wiki Activity

Aggiornare 'dhp-workflows/dhp-aggregation/src/test/java/eu/dnetlib/dhp/actionmanager/webcrawl/CreateASTest.java' 

Removed test for other datasource
This commit is contained in:
Miriam Baglioni 2024-04-22 11:01:46 +02:00
parent c987beb723
commit a8e72a7238
1 changed files with 1 additions and 161 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

162
dhp-workflows/dhp-aggregation/src/test/java/eu/dnetlib/dhp/actionmanager/webcrawl/CreateASTest.java
View File

@ -344,166 +344,6 @@ public class CreateASTest {
} }
@Test
void testRelationsDataInfo() throws Exception {
String inputPath = getClass()
.getResource(
"/eu/dnetlib/dhp/actionmanager/opencitations/COCI")
.getPath();
CreateActionSetSparkJob
.main(
new String[] {
"-isSparkSessionManaged",
Boolean.FALSE.toString(),
"-inputPath",
inputPath,
"-outputPath",
workingDir.toString() + "/actionSet4"
});
final JavaSparkContext sc = new JavaSparkContext(spark.sparkContext());
JavaRDD<Relation> tmp = sc
.sequenceFile(workingDir.toString() + "/actionSet4", Text.class, Text.class)
.map(value -> OBJECT_MAPPER.readValue(value._2().toString(), AtomicAction.class))
.map(aa -> ((Relation) aa.getPayload()));
tmp.foreach(r -> {
assertEquals(false, r.getDataInfo().getInferred());
assertEquals(false, r.getDataInfo().getDeletedbyinference());
assertEquals("0.91", r.getDataInfo().getTrust());
assertEquals(
CreateActionSetSparkJob.OPENCITATIONS_CLASSID, r.getDataInfo().getProvenanceaction().getClassid());
assertEquals(
CreateActionSetSparkJob.OPENCITATIONS_CLASSNAME, r.getDataInfo().getProvenanceaction().getClassname());
assertEquals(ModelConstants.DNET_PROVENANCE_ACTIONS, r.getDataInfo().getProvenanceaction().getSchemeid());
assertEquals(ModelConstants.DNET_PROVENANCE_ACTIONS, r.getDataInfo().getProvenanceaction().getSchemename());
});
}
@Test
void testRelationsSemantics() throws Exception {
String inputPath = getClass()
.getResource(
"/eu/dnetlib/dhp/actionmanager/opencitations/COCI")
.getPath();
CreateActionSetSparkJob
.main(
new String[] {
"-isSparkSessionManaged",
Boolean.FALSE.toString(),
"-inputPath",
inputPath,
"-outputPath",
workingDir.toString() + "/actionSet5"
});
final JavaSparkContext sc = new JavaSparkContext(spark.sparkContext());
JavaRDD<Relation> tmp = sc
.sequenceFile(workingDir.toString() + "/actionSet5", Text.class, Text.class)
.map(value -> OBJECT_MAPPER.readValue(value._2().toString(), AtomicAction.class))
.map(aa -> ((Relation) aa.getPayload()));
tmp.foreach(r -> {
assertEquals("citation", r.getSubRelType());
assertEquals("resultResult", r.getRelType());
});
assertEquals(23, tmp.filter(r -> r.getRelClass().equals("Cites")).count());
assertEquals(0, tmp.filter(r -> r.getRelClass().equals("IsCitedBy")).count());
}
@Test
void testRelationsSourceTargetPrefix() throws Exception {
String inputPath = getClass()
.getResource(
"/eu/dnetlib/dhp/actionmanager/opencitations/COCI")
.getPath();
CreateActionSetSparkJob
.main(
new String[] {
"-isSparkSessionManaged",
Boolean.FALSE.toString(),
"-inputPath",
inputPath,
"-outputPath",
workingDir.toString() + "/actionSet6"
});
final JavaSparkContext sc = new JavaSparkContext(spark.sparkContext());
JavaRDD<Relation> tmp = sc
.sequenceFile(workingDir.toString() + "/actionSet6", Text.class, Text.class)
.map(value -> OBJECT_MAPPER.readValue(value._2().toString(), AtomicAction.class))
.map(aa -> ((Relation) aa.getPayload()));
tmp.foreach(r -> {
assertEquals("50|doi_________::", r.getSource().substring(0, 17));
assertEquals("50|doi_________::", r.getTarget().substring(0, 17));
});
}
@Test
void testRelationsSourceTargetCouple() throws Exception {
final String doi1 = "50|doi_________::"
+ IdentifierFactory.md5(CleaningFunctions.normalizePidValue("doi", "10.1007/s10854-015-3684-x"));
final String doi2 = "50|doi_________::"
+ IdentifierFactory.md5(CleaningFunctions.normalizePidValue("doi", "10.1111/j.1551-2916.2008.02408.x"));
final String doi3 = "50|doi_________::"
+ IdentifierFactory.md5(CleaningFunctions.normalizePidValue("doi", "10.1007/s10854-014-2114-9"));
final String doi4 = "50|doi_________::"
+ IdentifierFactory.md5(CleaningFunctions.normalizePidValue("doi", "10.1016/j.ceramint.2013.09.069"));
final String doi5 = "50|doi_________::"
+ IdentifierFactory.md5(CleaningFunctions.normalizePidValue("doi", "10.1007/s10854-009-9913-4"));
final String doi6 = "50|doi_________::"
+ IdentifierFactory.md5(CleaningFunctions.normalizePidValue("doi", "10.1016/0038-1098(72)90370-5"));
String inputPath = getClass()
.getResource(
"/eu/dnetlib/dhp/actionmanager/opencitations/COCI")
.getPath();
CreateActionSetSparkJob
.main(
new String[] {
"-isSparkSessionManaged",
Boolean.FALSE.toString(),
"-inputPath",
inputPath,
"-outputPath",
workingDir.toString() + "/actionSet7"
});
final JavaSparkContext sc = new JavaSparkContext(spark.sparkContext());
JavaRDD<Relation> tmp = sc
.sequenceFile(workingDir.toString() + "/actionSet7", Text.class, Text.class)
.map(value -> OBJECT_MAPPER.readValue(value._2().toString(), AtomicAction.class))
.map(aa -> ((Relation) aa.getPayload()));
JavaRDD<Relation> check = tmp.filter(r -> r.getSource().equals(doi1) || r.getTarget().equals(doi1));
assertEquals(5, check.count());
// check.foreach(r -> {
// if (r.getSource().equals(doi2) || r.getSource().equals(doi3) || r.getSource().equals(doi4) ||
// r.getSource().equals(doi5) || r.getSource().equals(doi6)) {
// assertEquals(ModelConstants.IS_CITED_BY, r.getRelClass());
// assertEquals(doi1, r.getTarget());
// }
// });
assertEquals(5, check.filter(r -> r.getSource().equals(doi1)).count());
check.filter(r -> r.getSource().equals(doi1)).foreach(r -> assertEquals(ModelConstants.CITES, r.getRelClass()));
}
} }