WIP: cleaning of subjects

dhp-common/src/main/java/eu/dnetlib/dhp/common/vocabulary/Vocabulary.java

@@ -85,8 +85,8 @@ public class Vocabulary implements Serializable {
 
 	public Qualifier lookup(String id) {
 		return Optional
-				.ofNullable(getSynonymAsQualifier(id))
-				.orElse(getTermAsQualifier(id));
+			.ofNullable(getSynonymAsQualifier(id))
+			.orElse(getTermAsQualifier(id));
 	}
 
 }

dhp-common/src/main/java/eu/dnetlib/dhp/common/vocabulary/VocabularyGroup.java

@@ -83,9 +83,9 @@ public class VocabularyGroup implements Serializable {
 
 	public Optional<Vocabulary> find(final String vocId) {
 		return Optional
-				.ofNullable(vocId)
-				.map(String::toLowerCase)
-				.map(vocs::get);
+			.ofNullable(vocId)
+			.map(String::toLowerCase)
+			.map(vocs::get);
 	}
 
 	public void addTerm(final String vocId, final String id, final String name) {

dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/dhp/oa/graph/clean/CleaningRuleMap.java

@@ -4,15 +4,15 @@ package eu.dnetlib.dhp.oa.graph.clean;
 import java.io.Serializable;
 import java.util.HashMap;
 
-import eu.dnetlib.dhp.common.vocabulary.Vocabulary;
-import eu.dnetlib.dhp.schema.oaf.*;
+import javax.jws.WebParam;
+
 import org.apache.commons.lang3.StringUtils;
 
 import eu.dnetlib.dhp.common.FunctionalInterfaceSupport.SerializableConsumer;
+import eu.dnetlib.dhp.common.vocabulary.Vocabulary;
 import eu.dnetlib.dhp.common.vocabulary.VocabularyGroup;
 import eu.dnetlib.dhp.schema.common.ModelConstants;
-
-import javax.jws.WebParam;
+import eu.dnetlib.dhp.schema.oaf.*;
 
 public class CleaningRuleMap extends HashMap<Class<?>, SerializableConsumer<Object>> implements Serializable {
 

dhp-workflows/dhp-graph-mapper/src/test/java/eu/dnetlib/dhp/oa/graph/clean/GraphCleaningFunctionsTest.java

@@ -251,6 +251,17 @@ public class GraphCleaningFunctionsTest {
 							pid.getQualifier().getClassname()));
 		});
 
+		assertNotNull(p_cleaned.getSubject());
+
+		List<Subject> fos_subjects = p_cleaned
+			.getSubject()
+			.stream()
+			.filter(s -> ModelConstants.DNET_SUBJECT_FOS_CLASSID.equals(s.getQualifier().getClassid()))
+			.collect(Collectors.toList());
+
+		assertNotNull(fos_subjects);
+		assertEquals(2, fos_subjects.size());
+
 		// TODO add more assertions to verity the cleaned values
 		System.out.println(MAPPER.writeValueAsString(p_cleaned));
 	}

dhp-workflows/dhp-graph-mapper/src/test/resources/eu/dnetlib/dhp/oa/graph/clean/result.json

@@ -743,12 +743,12 @@
         "trust": "0.9"
       },
       "qualifier": {
-        "classid": "",
-        "classname": "",
-        "schemeid": "",
-        "schemename": ""
+        "classid": "keyword",
+        "classname": "keyword",
+        "schemeid": "dnet:subject_classification_typologies",
+        "schemename": "dnet:subject_classification_typologies"
       },
-      "value": "infrared detectors"
+      "value": "FOS: Mathematics"
     },
     {
       "dataInfo": {
@@ -765,12 +765,12 @@
         "trust": "0.9"
       },
       "qualifier": {
-        "classid": "",
-        "classname": "",
-        "schemeid": "",
-        "schemename": ""
+        "classid": "keyword",
+        "classname": "keyword",
+        "schemeid": "dnet:subject_classification_typologies",
+        "schemename": "dnet:subject_classification_typologies"
       },
-      "value": "lens antennas"
+      "value": "FOS: Computer and information sciences"
     },
     {
       "dataInfo": {

dhp-workflows/dhp-graph-mapper/src/test/resources/eu/dnetlib/dhp/oa/graph/clean/synonyms.txt

@@ -1243,4 +1243,6 @@ dnet:relation_relClass @=@ IsSupplementTo @=@ isSupplementTo
 dnet:relation_relClass @=@ IsSupplementedBy @=@ isSupplementedBy
 dnet:relation_relClass @=@ IsRelatedTo @=@ isRelatedTo
 dnet:relation_subRelType @=@ relationship @=@ publicationDataset
-dnet:provenanceActions @=@ iis @=@ erroneous label to be cleaned
+dnet:provenanceActions @=@ iis @=@ erroneous label to be cleaned
+FOS @=@ 0101 mathematics @=@ FOS: Mathematics
+FOS @=@ 0102 computer and information sciences @=@ FOS: Computer and information sciences

dhp-workflows/dhp-graph-mapper/src/test/resources/eu/dnetlib/dhp/oa/graph/clean/terms.txt

@@ -1117,4 +1117,7 @@ dnet:relation_subRelType @=@ dnet:relation_subRelType @=@ relationship @=@ relat
 dnet:relation_subRelType @=@ dnet:relation_subRelType @=@ review @=@ review
 dnet:relation_subRelType @=@ dnet:relation_subRelType @=@ similarity @=@ similarity
 dnet:relation_subRelType @=@ dnet:relation_subRelType @=@ supplement @=@ supplement
-dnet:relation_subRelType @=@ dnet:relation_subRelType @=@ version @=@ version
+dnet:relation_subRelType @=@ dnet:relation_subRelType @=@ version @=@ version
+FOS @=@ Fields of Science and Technology classification @=@ 0101 mathematics @=@ 0101 mathematics
+FOS @=@ Fields of Science and Technology classification @=@ 0102 computer and information sciences @=@ 0102 computer and information sciences
+FOS @=@ Fields of Science and Technology classification @=@ 0103 physical sciences @=@ 0103 physical sciences