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

[EOSC TAG] added test for galaxy for title and description criterias

This commit is contained in:
Miriam Baglioni 2022-04-22 18:35:03 +02:00
parent dfbd2bcbea
commit 88562c0930
1 changed files with 80 additions and 0 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

80
dhp-workflows/dhp-enrichment/src/test/java/eu/dnetlib/dhp/bulktag/EOSCTagJobTest.java
View File

@ -166,4 +166,84 @@ public class EOSCTagJobTest {
}
@Test
void galaxyUpdatesTest() throws Exception {
spark.read().textFile(getClass().getResource("/eu/dnetlib/dhp/eosctag/galaxy/software").getPath())
.map((MapFunction<String, Software>) value -> OBJECT_MAPPER.readValue(value, Software.class), Encoders.bean(Software.class))
.write()
.option("compression","gzip")
.json(workingDir.toString() + "/input/software");
spark.read().textFile(getClass().getResource("/eu/dnetlib/dhp/eosctag/galaxy/dataset").getPath())
.map((MapFunction<String, Dataset>) value -> OBJECT_MAPPER.readValue(value, Dataset.class), Encoders.bean(Dataset.class))
.write()
.option("compression","gzip")
.json(workingDir.toString() + "/input/dataset");
spark.read().textFile(getClass().getResource("/eu/dnetlib/dhp/eosctag/galaxy/otherresearchproduct").getPath())
.map((MapFunction<String, OtherResearchProduct>) value -> OBJECT_MAPPER.readValue(value, OtherResearchProduct.class), Encoders.bean(OtherResearchProduct.class))
.write()
.option("compression","gzip")
.json(workingDir.toString() + "/input/otherresearchproduct");
SparkEoscTag
.main(
new String[] {
"-isSparkSessionManaged", Boolean.FALSE.toString(),
"-sourcePath",
workingDir.toString() + "/input",
"-workingPath", workingDir.toString() + "/working"
});
final JavaSparkContext sc = JavaSparkContext.fromSparkContext(spark.sparkContext());
JavaRDD<Software> tmp = sc
.textFile(workingDir.toString() + "/input/software")
.map(item -> OBJECT_MAPPER.readValue(item, Software.class));
Assertions.assertEquals(10, tmp.count());
Assertions.assertEquals(2, tmp.filter(s -> s.getSubject().stream().anyMatch(sbj -> sbj.getValue().equals("EOSC::Galaxy Workflow"))).count());
Assertions.assertEquals(2, tmp.filter(sw -> sw.getId().equals("50|od______1582::4132f5ec9496f0d6adc7b00a50a56ff4")).collect()
.get(0).getSubject().size());
Assertions.assertTrue(tmp.filter(sw -> sw.getId().equals("50|od______1582::4132f5ec9496f0d6adc7b00a50a56ff4")).collect()
.get(0).getSubject().stream().anyMatch(s -> s.getValue().equals("EOSC::Galaxy Workflow")));
Assertions.assertEquals(6, tmp.filter(sw -> sw.getId().equals("50|od______1582::501b25d420f808c8eddcd9b16e917f11")).collect()
.get(0).getSubject().size());
Assertions.assertTrue(tmp.filter(sw -> sw.getId().equals("50|od______1582::501b25d420f808c8eddcd9b16e917f11")).collect()
.get(0).getSubject().stream().anyMatch(s -> s.getValue().equals("EOSC::Galaxy Workflow")));
Assertions.assertEquals(8, tmp.filter(sw -> sw.getId().equals("50|od______1582::581621232a561b7e8b4952b18b8b0e56")).collect()
.get(0).getSubject().size());
Assertions.assertFalse(tmp.filter(sw -> sw.getId().equals("50|od______1582::581621232a561b7e8b4952b18b8b0e56")).collect()
.get(0).getSubject().stream().anyMatch(s -> s.getValue().equals("EOSC::Galaxy Workflow")));
JavaRDD<OtherResearchProduct> orp = sc.textFile(workingDir.toString() + "/input/otherresearchproduct").map(item -> OBJECT_MAPPER.readValue(item, OtherResearchProduct.class));
Assertions.assertEquals(10, orp.count());
Assertions.assertEquals(2, orp.filter(s -> s.getSubject().stream().anyMatch(sbj -> sbj.getValue().equals("EOSC::Galaxy Workflow"))).count());
Assertions.assertEquals(3, orp.filter(sw -> sw.getId().equals("50|od______2017::0750a4d0782265873d669520f5e33c07")).collect()
.get(0).getSubject().size());
Assertions.assertTrue(orp.filter(sw -> sw.getId().equals("50|od______2017::0750a4d0782265873d669520f5e33c07")).collect()
.get(0).getSubject().stream().anyMatch(s -> s.getValue().equals("EOSC::Galaxy Workflow")));
Assertions.assertEquals(2, orp.filter(sw -> sw.getId().equals("50|od______2017::1bd97baef19dbd2db3203b112bb83bc5")).collect()
.get(0).getSubject().size());
Assertions.assertFalse(orp.filter(sw -> sw.getId().equals("50|od______2017::1bd97baef19dbd2db3203b112bb83bc5")).collect()
.get(0).getSubject().stream().anyMatch(s -> s.getValue().equals("EOSC::Galaxy Workflow")));
Assertions.assertEquals(3, orp.filter(sw -> sw.getId().equals("50|od______2017::1e400f1747487fd15998735c41a55c72")).collect()
.get(0).getSubject().size());
Assertions.assertTrue(orp.filter(sw -> sw.getId().equals("50|od______2017::1e400f1747487fd15998735c41a55c72")).collect()
.get(0).getSubject().stream().anyMatch(s -> s.getValue().equals("EOSC::Galaxy Workflow")));
}
}