From b78889a0ce27a79c7ab2d8da05b118ee4f1bcb36 Mon Sep 17 00:00:00 2001
From: Claudio Atzori <claudio.atzori@isti.cnr.it>
Date: Fri, 5 Aug 2022 09:11:37 +0200
Subject: [PATCH] WIP: cleaning of subjects

---
 .../dhp/common/vocabulary/Vocabulary.java     |  6 ++++
 .../common/vocabulary/VocabularyGroup.java    |  7 +++++
 .../dhp/oa/graph/clean/CleaningRuleMap.java   | 29 +++++++++++++++----
 3 files changed, 37 insertions(+), 5 deletions(-)

diff --git a/dhp-common/src/main/java/eu/dnetlib/dhp/common/vocabulary/Vocabulary.java b/dhp-common/src/main/java/eu/dnetlib/dhp/common/vocabulary/Vocabulary.java
index b3eb98d4f..24a30500d 100644
--- a/dhp-common/src/main/java/eu/dnetlib/dhp/common/vocabulary/Vocabulary.java
+++ b/dhp-common/src/main/java/eu/dnetlib/dhp/common/vocabulary/Vocabulary.java
@@ -83,4 +83,10 @@ public class Vocabulary implements Serializable {
 			.orElse(null);
 	}
 
+	public Qualifier lookup(String id) {
+		return Optional
+				.ofNullable(getSynonymAsQualifier(id))
+				.orElse(getTermAsQualifier(id));
+	}
+
 }
diff --git a/dhp-common/src/main/java/eu/dnetlib/dhp/common/vocabulary/VocabularyGroup.java b/dhp-common/src/main/java/eu/dnetlib/dhp/common/vocabulary/VocabularyGroup.java
index 1c129ff9c..8435b8bf3 100644
--- a/dhp-common/src/main/java/eu/dnetlib/dhp/common/vocabulary/VocabularyGroup.java
+++ b/dhp-common/src/main/java/eu/dnetlib/dhp/common/vocabulary/VocabularyGroup.java
@@ -81,6 +81,13 @@ public class VocabularyGroup implements Serializable {
 		vocs.put(id.toLowerCase(), new Vocabulary(id, name));
 	}
 
+	public Optional<Vocabulary> find(final String vocId) {
+		return Optional
+				.ofNullable(vocId)
+				.map(String::toLowerCase)
+				.map(vocs::get);
+	}
+
 	public void addTerm(final String vocId, final String id, final String name) {
 		if (vocabularyExists(vocId)) {
 			vocs.get(vocId.toLowerCase()).addTerm(id, name);
diff --git a/dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/dhp/oa/graph/clean/CleaningRuleMap.java b/dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/dhp/oa/graph/clean/CleaningRuleMap.java
index 6c156edb7..2a4183f1b 100644
--- a/dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/dhp/oa/graph/clean/CleaningRuleMap.java
+++ b/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 org.apache.commons.lang3.StringUtils;
 
 import eu.dnetlib.dhp.common.FunctionalInterfaceSupport.SerializableConsumer;
 import eu.dnetlib.dhp.common.vocabulary.VocabularyGroup;
 import eu.dnetlib.dhp.schema.common.ModelConstants;
-import eu.dnetlib.dhp.schema.oaf.AccessRight;
-import eu.dnetlib.dhp.schema.oaf.Country;
-import eu.dnetlib.dhp.schema.oaf.Qualifier;
-import eu.dnetlib.dhp.schema.oaf.Relation;
+
+import javax.jws.WebParam;
 
 public class CleaningRuleMap extends HashMap<Class<?>, SerializableConsumer<Object>> implements Serializable {
 
@@ -27,10 +27,29 @@ public class CleaningRuleMap extends HashMap<Class<?>, SerializableConsumer<Obje
 		mapping.put(AccessRight.class, o -> cleanQualifier(vocabularies, (AccessRight) o));
 		mapping.put(Country.class, o -> cleanCountry(vocabularies, (Country) o));
 		mapping.put(Relation.class, o -> cleanRelation(vocabularies, (Relation) o));
-
+		mapping.put(Subject.class, o -> cleanSubject(vocabularies, (Subject) o));
 		return mapping;
 	}
 
+	private static void cleanSubject(VocabularyGroup vocabularies, Subject s) {
+		// TODO cleaning based on different subject vocabs can be added here
+		cleanSubjectForVocabulary(ModelConstants.DNET_SUBJECT_FOS_CLASSID, vocabularies, s);
+	}
+
+	private static void cleanSubjectForVocabulary(String vocabularyId, VocabularyGroup vocabularies, Subject s) {
+		vocabularies.find(vocabularyId).ifPresent(vocabulary -> {
+			if (!ModelConstants.DNET_SUBJECT_KEYWORD.equalsIgnoreCase(s.getQualifier().getClassid())) {
+				return;
+			}
+			Qualifier newValue = vocabulary.lookup(s.getValue());
+			if (!s.getValue().equals(newValue.getClassid())) {
+				s.setValue(newValue.getClassid());
+				s.getQualifier().setClassid(vocabularyId);
+				s.getQualifier().setClassname(vocabulary.getName());
+			}
+		});
+	}
+
 	private static void cleanRelation(VocabularyGroup vocabularies, Relation r) {
 		if (vocabularies.vocabularyExists(ModelConstants.DNET_RELATION_SUBRELTYPE)) {
 			Qualifier newValue = vocabularies.lookup(ModelConstants.DNET_RELATION_SUBRELTYPE, r.getSubRelType());