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

EOSC IF #230

Merged
claudio.atzori merged 6 commits from tagEosc into beta 2022-07-25 14:14:54 +02:00
Conversation 0 Commits 6 Files Changed 3 +27 -57
2 changed files with 27 additions and 57 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files
Showing only changes of commit 438abdf96f - Show all commits

82
dhp-workflows/dhp-enrichment/src/main/java/eu/dnetlib/dhp/bulktag/SparkEoscTag.java
View File

@ -28,28 +28,7 @@ import eu.dnetlib.dhp.schema.oaf.utils.OafMapperUtils;
public class SparkEoscTag {
private static final Logger log = LoggerFactory.getLogger(SparkEoscTag.class);
public static final ObjectMapper OBJECT_MAPPER = new ObjectMapper();
public static final Qualifier EOSC_QUALIFIER = OafMapperUtils
.qualifier(
"EOSC",
"European Open Science Cloud",
ModelConstants.DNET_SUBJECT_TYPOLOGIES, ModelConstants.DNET_SUBJECT_TYPOLOGIES);
public static final DataInfo EOSC_DATAINFO = OafMapperUtils
.dataInfo(
false, "propagation", true, false,
OafMapperUtils
.qualifier(
"propagation:subject", "Inferred by OpenAIRE",
ModelConstants.DNET_PROVENANCE_ACTIONS, ModelConstants.DNET_PROVENANCE_ACTIONS),
"0.9");
public final static StructuredProperty EOSC_NOTEBOOK = OafMapperUtils
.structuredProperty(
"EOSC::Jupyter Notebook", EOSC_QUALIFIER, EOSC_DATAINFO);
public final static StructuredProperty EOSC_GALAXY = OafMapperUtils
.structuredProperty(
"EOSC::Galaxy Workflow", EOSC_QUALIFIER, EOSC_DATAINFO);
public final static StructuredProperty EOSC_TWITTER = OafMapperUtils
.structuredProperty(
"EOSC::Twitter Data", EOSC_QUALIFIER, EOSC_DATAINFO);
public static void main(String[] args) throws Exception {
String jsonConfiguration = IOUtils
@ -84,29 +63,30 @@ public class SparkEoscTag {
});
}
public static EoscIfGuidelines newInstance(String code, String label, String url, String semantics){
EoscIfGuidelines eig = new EoscIfGuidelines();
eig.setCode( code);
eig.setLabel(label);
eig.setUrl(url);
eig.setSemanticRelation(semantics);
return eig;
}
private static void execEoscTag(SparkSession spark, String inputPath, String workingPath) {
readPath(spark, inputPath + "/software", Software.class)
.map((MapFunction<Software, Software>) s -> {
List<StructuredProperty> sbject;
if (!Optional.ofNullable(s.getSubject()).isPresent())
s.setSubject(new ArrayList<>());
sbject = s.getSubject();
if (containsCriteriaNotebook(s)) {
sbject.add(EOSC_NOTEBOOK);
if (sbject.stream().anyMatch(sb -> sb.getValue().equals("EOSC Jupyter Notebook"))) {
sbject = sbject.stream().map(sb -> {
if (sb.getValue().equals("EOSC Jupyter Notebook")) {
return null;
}
return sb;
}).filter(Objects::nonNull).collect(Collectors.toList());
s.setSubject(sbject);
}
if (!Optional.ofNullable(s.getEoscifguidelines()).isPresent())
s.setEoscifguidelines(new ArrayList<>());
s.getEoscifguidelines().add(newInstance("EOSC::Jupyter Notebook", "EOSC::Jupyter Notebook", "", "compliesWith"));
}
if (containsCriteriaGalaxy(s)) {
sbject.add(EOSC_GALAXY);
if (!Optional.ofNullable(s.getEoscifguidelines()).isPresent())
s.setEoscifguidelines(new ArrayList<>());
s.getEoscifguidelines().add(newInstance("EOSC::Galaxy Workflow", "EOSC::Galaxy Workflow", "", "compliesWith"));
}
return s;
}, Encoders.bean(Software.class))
@ -124,14 +104,14 @@ public class SparkEoscTag {
readPath(spark, inputPath + "/otherresearchproduct", OtherResearchProduct.class)
.map((MapFunction<OtherResearchProduct, OtherResearchProduct>) orp -> {
List<StructuredProperty> sbject;
if (!Optional.ofNullable(orp.getSubject()).isPresent())
orp.setSubject(new ArrayList<>());
sbject = orp.getSubject();
if (!Optional.ofNullable(orp.getEoscifguidelines()).isPresent())
orp.setEoscifguidelines(new ArrayList<>());
if (containsCriteriaGalaxy(orp)) {
sbject.add(EOSC_GALAXY);
orp.getEoscifguidelines().add(newInstance("EOSC::Galaxy Workflow", "EOSC::Galaxy Workflow", "", "compliesWith"));
}
if (containscriteriaTwitter(orp)) {
sbject.add(EOSC_TWITTER);
orp.getEoscifguidelines().add(newInstance("EOSC::Twitter Data", "EOSC::Twitter Data", "", "compliesWith"));
}
return orp;
}, Encoders.bean(OtherResearchProduct.class))
@ -149,11 +129,10 @@ public class SparkEoscTag {
readPath(spark, inputPath + "/dataset", Dataset.class)
.map((MapFunction<Dataset, Dataset>) d -> {
List<StructuredProperty> sbject;
if (!Optional.ofNullable(d.getSubject()).isPresent())
d.setSubject(new ArrayList<>());
sbject = d.getSubject();
if (!Optional.ofNullable(d.getEoscifguidelines()).isPresent())
d.setEoscifguidelines(new ArrayList<>());
if (containscriteriaTwitter(d)) {
sbject.add(EOSC_TWITTER);
d.getEoscifguidelines().add(newInstance("EOSC::Twitter Data", "EOSC::Twitter Data", "", "compliesWith"));
}
return d;
}, Encoders.bean(Dataset.class))
@ -212,13 +191,6 @@ public class SparkEoscTag {
return false;
}
private static Set<String> getSubjects(List<StructuredProperty> s) {
Set<String> subjects = new HashSet<>();
s.stream().forEach(sbj -> subjects.addAll(Arrays.asList(sbj.getValue().toLowerCase().split(" "))));
s.stream().forEach(sbj -> subjects.add(sbj.getValue().toLowerCase()));
return subjects;
}
private static Set<String> getWordsSP(List<StructuredProperty> elem) {
Set<String> words = new HashSet<>();
Optional
@ -242,9 +214,7 @@ public class SparkEoscTag {
t -> words
.addAll(
Arrays.asList(t.getValue().toLowerCase().replaceAll("[^a-zA-Z ]", "").split(" ")))));
// elem
// .forEach(
// t -> words.addAll(Arrays.asList(t.getValue().toLowerCase().replaceAll("[^a-zA-Z ]", "").split(" "))));
return words;
}

2
pom.xml
View File

@ -801,7 +801,7 @@
<mockito-core.version>3.3.3</mockito-core.version>
<mongodb.driver.version>3.4.2</mongodb.driver.version>
<vtd.version>[2.12,3.0)</vtd.version>
<dhp-schemas.version>[2.12.0]</dhp-schemas.version>
<dhp-schemas.version>[2.12.2-SNAPSHOT]</dhp-schemas.version>
<dnet-actionmanager-api.version>[4.0.3]</dnet-actionmanager-api.version>
<dnet-actionmanager-common.version>[6.0.5]</dnet-actionmanager-common.version>
<dnet-openaire-broker-common.version>[3.1.6]</dnet-openaire-broker-common.version>