Merge pull request 'advConstraintsInBeta' (#288) from advConstraintsInBeta into master

Reviewed-on: D-Net/dnet-hadoop#288

dhp-common/src/main/java/eu/dnetlib/dhp/common/MakeTarArchive.java

@@ -58,9 +58,12 @@ public class MakeTarArchive implements Serializable {
 		makeTArArchive(fileSystem, inputPath, outputPath, gBperSplit, rename);
 
 	}
-	public static void makeTArArchive(FileSystem fileSystem, String inputPath, String outputPath, int gBperSplit) throws IOException{
+
+	public static void makeTArArchive(FileSystem fileSystem, String inputPath, String outputPath, int gBperSplit)
+		throws IOException {
 		makeTArArchive(fileSystem, inputPath, outputPath, gBperSplit, false);
 	}
+
 	public static void makeTArArchive(FileSystem fileSystem, String inputPath, String outputPath, int gBperSplit,
 		boolean rename)
 		throws IOException {

dhp-workflows/dhp-enrichment/src/main/java/eu/dnetlib/dhp/bulktag/community/Community.java

@@ -24,7 +24,8 @@ public class Community implements Serializable {
 	public boolean isValid() {
 		return !getSubjects().isEmpty()
 			|| !getProviders().isEmpty()
-			|| !getZenodoCommunities().isEmpty();
+			|| !getZenodoCommunities().isEmpty()
+			|| getConstraints().getCriteria() != null;
 	}
 
 	public String getId() {

dhp-workflows/dhp-enrichment/src/main/java/eu/dnetlib/dhp/bulktag/community/CommunityConfigurationFactory.java

@@ -85,9 +85,23 @@ public class CommunityConfigurationFactory {
 		c.setSubjects(parseSubjects(node));
 		c.setProviders(parseDatasources(node));
 		c.setZenodoCommunities(parseZenodoCommunities(node));
+		c.setConstraints(parseConstrains(node));
 		return c;
 	}
 
+	private static SelectionConstraints parseConstrains(Node node) {
+		Node advConstsNode = node.selectSingleNode("./advancedConstraints");
+		if (advConstsNode == null || StringUtils.isBlank(StringUtils.trim(advConstsNode.getText()))) {
+			return new SelectionConstraints();
+		}
+		SelectionConstraints selectionConstraints = new Gson()
+			.fromJson(advConstsNode.getText(), SelectionConstraints.class);
+
+		selectionConstraints.setSelection(resolver);
+		log.info("number of selection constraints set " + selectionConstraints.getCriteria().size());
+		return selectionConstraints;
+	}
+
 	private static List<String> parseSubjects(final Node node) {
 
 		final List<String> subjects = Lists.newArrayList();

dhp-workflows/dhp-enrichment/src/main/java/eu/dnetlib/dhp/bulktag/community/ResultTagger.java

@@ -139,7 +139,7 @@ public class ResultTagger implements Serializable {
 			.getSelectionConstraintsMap()
 			.keySet()
 			.forEach(communityId -> {
-				if (conf.getSelectionConstraintsMap().get(communityId) != null &&
+				if (conf.getSelectionConstraintsMap().get(communityId).getCriteria() != null &&
 					conf
 						.getSelectionConstraintsMap()
 						.get(communityId)

dhp-workflows/dhp-enrichment/src/main/java/eu/dnetlib/dhp/bulktag/criteria/StartsWithIgnoreCaseVerb.java

@@ -0,0 +1,34 @@
+
+package eu.dnetlib.dhp.bulktag.criteria;
+
+import java.io.Serializable;
+
+/**
+ * @author miriam.baglioni
+ * @Date 06/04/23
+ */
+
+@VerbClass("starts_with_caseinsensitive")
+public class StartsWithIgnoreCaseVerb implements Selection, Serializable {
+	private String param;
+
+	public StartsWithIgnoreCaseVerb() {
+	}
+
+	public StartsWithIgnoreCaseVerb(final String param) {
+		this.param = param;
+	}
+
+	@Override
+	public boolean apply(String value) {
+		return value.toLowerCase().startsWith(param.toLowerCase());
+	}
+
+	public String getParam() {
+		return param;
+	}
+
+	public void setParam(String param) {
+		this.param = param;
+	}
+}

dhp-workflows/dhp-enrichment/src/main/java/eu/dnetlib/dhp/bulktag/criteria/StartsWithVerb.java

@@ -0,0 +1,34 @@
+
+package eu.dnetlib.dhp.bulktag.criteria;
+
+import java.io.Serializable;
+
+/**
+ * @author miriam.baglioni
+ * @Date 06/04/23
+ */
+
+@VerbClass("starts_with")
+public class StartsWithVerb implements Selection, Serializable {
+	private String param;
+
+	public StartsWithVerb() {
+	}
+
+	public StartsWithVerb(final String param) {
+		this.param = param;
+	}
+
+	@Override
+	public boolean apply(String value) {
+		return value.startsWith(param);
+	}
+
+	public String getParam() {
+		return param;
+	}
+
+	public void setParam(String param) {
+		this.param = param;
+	}
+}

dhp-workflows/dhp-enrichment/src/test/java/eu/dnetlib/dhp/bulktag/BulkTagJobTest.java

@@ -39,7 +39,10 @@ public class BulkTagJobTest {
 		+ "  \"title\" : \"$['title'][*]['value']\","
 		+ "  \"orcid\" : \"$['author'][*]['pid'][*][?(@['key']=='ORCID')]['value']\","
 		+ "  \"contributor\" : \"$['contributor'][*]['value']\","
-		+ "  \"description\" : \"$['description'][*]['value']\"}";
+		+ "  \"description\" : \"$['description'][*]['value']\", "
+		+ " \"subject\" :\"$['subject'][*]['value']\" , " +
+		"\"fos\" : \"$['subject'][?(@['qualifier']['classid']=='FOS')].value\"" +
+		"} ";
 
 	private static SparkSession spark;
 
@@ -763,7 +766,7 @@ public class BulkTagJobTest {
 		org.apache.spark.sql.Dataset<Row> idExplodeCommunity = spark.sql(query);
 
 		idExplodeCommunity.show(false);
-		Assertions.assertEquals(3, idExplodeCommunity.count());
+		Assertions.assertEquals(4, idExplodeCommunity.count());
 
 		Assertions
 			.assertEquals(

dhp-workflows/dhp-enrichment/src/test/resources/eu/dnetlib/dhp/bulktag/communityconfiguration/tagging_conf.xml

@@ -844,6 +844,88 @@
     <organizations/>
 </community>
     <community id="dariah">
+        <advancedConstraints>
+                        {
+                        "criteria": [
+                        {
+                        "constraint": [
+                        {
+                        "verb": "equals_caseinsensitive",
+                        "field": "subject",
+                        "value": "North America"
+                        },
+                        {
+                        "verb": "contains",
+                        "field": "fos",
+                        "value": "05"
+                        }
+                        ]
+                        },
+                        {
+                        "constraint": [
+                        {
+                        "verb": "equals_caseinsensitive",
+                        "field": "subject",
+                        "value": "North America"
+                        },
+                        {
+                        "verb": "contains",
+                        "field": "fos",
+                        "value": "06"
+                        }
+                        ]
+                        },
+                        {
+                        "constraint": [
+                        {
+                        "verb": "equals_caseinsensitive",
+                        "field": "subject",
+                        "value": "Mexico"
+                        },
+                        {
+                        "verb": "equals_caseinsensitive",
+                        "field": "subject",
+                        "value": "United States"
+                        },
+                        {
+                        "verb": "equals_caseinsensitive",
+                        "field": "subject",
+                        "value": "Canada"
+                        },
+                        {
+                        "verb": "contains",
+                        "field": "fos",
+                        "value": "05"
+                        }
+                        ]
+                        },
+                        {
+                        "constraint": [
+                        {
+                        "verb": "equals_caseinsensitive",
+                        "field": "subject",
+                        "value": "Mexico"
+                        },
+                        {
+                        "verb": "equals_caseinsensitive",
+                        "field": "subject",
+                        "value": "United States"
+                        },
+                        {
+                        "verb": "equals_caseinsensitive",
+                        "field": "subject",
+                        "value": "Canada"
+                        },
+                        {
+                        "verb": "contains",
+                        "field": "fos",
+                        "value": "06"
+                        }
+                        ]
+                        }
+                        ]
+                        }
+        </advancedConstraints>
     <subjects/>
     <datasources>
         <datasource>