added code to dump the relations between organizaiton and projects in the subset of entities relevant for EOSC

dump-schema/src/main/java/eu/dnetlib/dhp/eosc/model/Funder.java

@@ -1,28 +1,27 @@
+
 package eu.dnetlib.dhp.eosc.model;
 
 /**
  * @author miriam.baglioni
  * @Date 25/10/23
  */
-
+import com.github.imifou.jsonschema.module.addon.annotation.JsonSchema;
-
-        import com.github.imifou.jsonschema.module.addon.annotation.JsonSchema;
 
 /**
  * To store information about the funder funding the project related to the result. It extends
  * eu.dnetlib.dhp.schema.dump.oaf.Funder with the following parameter: - - private
  * eu.dnetdlib.dhp.schema.dump.oaf.graph.Fundings funding_stream to store the fundingstream
  */
-        public class Funder extends FunderShort {
+public class Funder extends FunderShort {
 
-        @JsonSchema(description = "Description of the funding stream")
+	@JsonSchema(description = "Description of the funding stream")
-        private Fundings funding_stream;
+	private Fundings funding_stream;
 
-        public Fundings getFunding_stream() {
+	public Fundings getFunding_stream() {
-        return funding_stream;
+		return funding_stream;
-        }
+	}
 
-        public void setFunding_stream(Fundings funding_stream) {
+	public void setFunding_stream(Fundings funding_stream) {
-        this.funding_stream = funding_stream;
+		this.funding_stream = funding_stream;
-        }
+	}
-        }
+}

dump-schema/src/main/java/eu/dnetlib/dhp/eosc/model/Fundings.java

@@ -1,12 +1,10 @@
+
 package eu.dnetlib.dhp.eosc.model;
 
 /**
  * @author miriam.baglioni
  * @Date 25/10/23
  */
-
-
-
 import java.io.Serializable;
 
 import com.github.imifou.jsonschema.module.addon.annotation.JsonSchema;
@@ -22,23 +20,23 @@ import com.github.imifou.jsonschema.module.addon.annotation.JsonSchema;
  */
 public class Fundings implements Serializable {
 
-    @JsonSchema(description = "Id of the funding stream")
+	@JsonSchema(description = "Id of the funding stream")
-    private String id;
+	private String id;
-    private String description;
+	private String description;
 
-    public String getId() {
+	public String getId() {
-        return id;
+		return id;
-    }
+	}
 
-    public void setId(String id) {
+	public void setId(String id) {
-        this.id = id;
+		this.id = id;
-    }
+	}
 
-    public String getDescription() {
+	public String getDescription() {
-        return description;
+		return description;
-    }
+	}
 
-    public void setDescription(String description) {
+	public void setDescription(String description) {
-        this.description = description;
+		this.description = description;
-    }
+	}
 }

dump-schema/src/main/java/eu/dnetlib/dhp/eosc/model/Granted.java

@@ -1,11 +1,10 @@
+
 package eu.dnetlib.dhp.eosc.model;
 
 /**
  * @author miriam.baglioni
  * @Date 25/10/23
  */
-
-
 import java.io.Serializable;
 
 import com.github.imifou.jsonschema.module.addon.annotation.JsonSchema;
@@ -16,51 +15,51 @@ import com.github.imifou.jsonschema.module.addon.annotation.JsonSchema;
  * funded amount by the funder
  */
 public class Granted implements Serializable {
-    @JsonSchema(description = "The currency of the granted amount (e.g. EUR)")
+	@JsonSchema(description = "The currency of the granted amount (e.g. EUR)")
-    private String currency;
+	private String currency;
 
-    @JsonSchema(description = "The total cost of the project")
+	@JsonSchema(description = "The total cost of the project")
-    private float totalcost;
+	private float totalcost;
 
-    @JsonSchema(description = "The funded amount")
+	@JsonSchema(description = "The funded amount")
-    private float fundedamount;
+	private float fundedamount;
 
-    public String getCurrency() {
+	public String getCurrency() {
-        return currency;
+		return currency;
-    }
+	}
 
-    public void setCurrency(String currency) {
+	public void setCurrency(String currency) {
-        this.currency = currency;
+		this.currency = currency;
-    }
+	}
 
-    public float getTotalcost() {
+	public float getTotalcost() {
-        return totalcost;
+		return totalcost;
-    }
+	}
 
-    public void setTotalcost(float totalcost) {
+	public void setTotalcost(float totalcost) {
-        this.totalcost = totalcost;
+		this.totalcost = totalcost;
-    }
+	}
 
-    public float getFundedamount() {
+	public float getFundedamount() {
-        return fundedamount;
+		return fundedamount;
-    }
+	}
 
-    public void setFundedamount(float fundedamount) {
+	public void setFundedamount(float fundedamount) {
-        this.fundedamount = fundedamount;
+		this.fundedamount = fundedamount;
-    }
+	}
 
-    public static Granted newInstance(String currency, float totalcost, float fundedamount) {
+	public static Granted newInstance(String currency, float totalcost, float fundedamount) {
-        Granted granted = new Granted();
+		Granted granted = new Granted();
-        granted.currency = currency;
+		granted.currency = currency;
-        granted.totalcost = totalcost;
+		granted.totalcost = totalcost;
-        granted.fundedamount = fundedamount;
+		granted.fundedamount = fundedamount;
-        return granted;
+		return granted;
-    }
+	}
 
-    public static Granted newInstance(String currency, float fundedamount) {
+	public static Granted newInstance(String currency, float fundedamount) {
-        Granted granted = new Granted();
+		Granted granted = new Granted();
-        granted.currency = currency;
+		granted.currency = currency;
-        granted.fundedamount = fundedamount;
+		granted.fundedamount = fundedamount;
-        return granted;
+		return granted;
-    }
+	}
 }

dump-schema/src/main/java/eu/dnetlib/dhp/eosc/model/Organization.java

@@ -1,3 +1,4 @@
+
 package eu.dnetlib.dhp.eosc.model;
 
 import java.io.Serializable;
@@ -16,76 +17,76 @@ import com.github.imifou.jsonschema.module.addon.annotation.JsonSchema;
  * - private List<OrganizationPid> pid to store the list of pids for the organization
  */
 public class Organization implements Serializable {
-    private String legalshortname;
+	private String legalshortname;
-    private String legalname;
+	private String legalname;
-    private String websiteurl;
+	private String websiteurl;
 
-    @JsonSchema(description = "Alternative names that identify the organisation")
+	@JsonSchema(description = "Alternative names that identify the organisation")
-    private List<String> alternativenames;
+	private List<String> alternativenames;
 
-    @JsonSchema(description = "The organisation country")
+	@JsonSchema(description = "The organisation country")
-    private Country country;
+	private Country country;
 
-    @JsonSchema(description = "The OpenAIRE id for the organisation")
+	@JsonSchema(description = "The OpenAIRE id for the organisation")
-    private String id;
+	private String id;
 
-    @JsonSchema(description = "Persistent identifiers for the organisation i.e. isni 0000000090326370")
+	@JsonSchema(description = "Persistent identifiers for the organisation i.e. isni 0000000090326370")
-    private List<OrganizationPid> pid;
+	private List<OrganizationPid> pid;
 
-    public String getLegalshortname() {
+	public String getLegalshortname() {
-        return legalshortname;
+		return legalshortname;
-    }
+	}
 
-    public void setLegalshortname(String legalshortname) {
+	public void setLegalshortname(String legalshortname) {
-        this.legalshortname = legalshortname;
+		this.legalshortname = legalshortname;
-    }
+	}
 
-    public String getLegalname() {
+	public String getLegalname() {
-        return legalname;
+		return legalname;
-    }
+	}
 
-    public void setLegalname(String legalname) {
+	public void setLegalname(String legalname) {
-        this.legalname = legalname;
+		this.legalname = legalname;
-    }
+	}
 
-    public String getWebsiteurl() {
+	public String getWebsiteurl() {
-        return websiteurl;
+		return websiteurl;
-    }
+	}
 
-    public void setWebsiteurl(String websiteurl) {
+	public void setWebsiteurl(String websiteurl) {
-        this.websiteurl = websiteurl;
+		this.websiteurl = websiteurl;
-    }
+	}
 
-    public List<String> getAlternativenames() {
+	public List<String> getAlternativenames() {
-        return alternativenames;
+		return alternativenames;
-    }
+	}
 
-    public void setAlternativenames(List<String> alternativenames) {
+	public void setAlternativenames(List<String> alternativenames) {
-        this.alternativenames = alternativenames;
+		this.alternativenames = alternativenames;
-    }
+	}
 
-    public Country getCountry() {
+	public Country getCountry() {
-        return country;
+		return country;
-    }
+	}
 
-    public void setCountry(Country country) {
+	public void setCountry(Country country) {
-        this.country = country;
+		this.country = country;
-    }
+	}
 
-    public String getId() {
+	public String getId() {
-        return id;
+		return id;
-    }
+	}
 
-    public void setId(String id) {
+	public void setId(String id) {
-        this.id = id;
+		this.id = id;
-    }
+	}
 
-    public List<OrganizationPid> getPid() {
+	public List<OrganizationPid> getPid() {
-        return pid;
+		return pid;
-    }
+	}
 
-    public void setPid(List<OrganizationPid> pid) {
+	public void setPid(List<OrganizationPid> pid) {
-        this.pid = pid;
+		this.pid = pid;
-    }
+	}
 
 }

dump-schema/src/main/java/eu/dnetlib/dhp/eosc/model/OrganizationPid.java

@@ -33,7 +33,7 @@ public class OrganizationPid implements Serializable {
 		this.value = value;
 	}
 
-	public static OrganizationPid newInstance(String type, String value){
+	public static OrganizationPid newInstance(String type, String value) {
 		OrganizationPid op = new OrganizationPid();
 		op.type = type;
 		op.value = value;

dump-schema/src/main/java/eu/dnetlib/dhp/eosc/model/Programme.java

@@ -1,11 +1,10 @@
+
 package eu.dnetlib.dhp.eosc.model;
 
 /**
  * @author miriam.baglioni
  * @Date 25/10/23
  */
-
-
 import java.io.Serializable;
 
 import com.github.imifou.jsonschema.module.addon.annotation.JsonSchema;
@@ -15,32 +14,32 @@ import com.github.imifou.jsonschema.module.addon.annotation.JsonSchema;
  * to store the code of the programme - private String description to store the description of the programme
  */
 public class Programme implements Serializable {
-    @JsonSchema(description = "The code of the programme")
+	@JsonSchema(description = "The code of the programme")
-    private String code;
+	private String code;
 
-    @JsonSchema(description = "The description of the programme")
+	@JsonSchema(description = "The description of the programme")
-    private String description;
+	private String description;
 
-    public String getCode() {
+	public String getCode() {
-        return code;
+		return code;
-    }
+	}
 
-    public void setCode(String code) {
+	public void setCode(String code) {
-        this.code = code;
+		this.code = code;
-    }
+	}
 
-    public String getDescription() {
+	public String getDescription() {
-        return description;
+		return description;
-    }
+	}
 
-    public void setDescription(String description) {
+	public void setDescription(String description) {
-        this.description = description;
+		this.description = description;
-    }
+	}
 
-    public static Programme newInstance(String code, String description) {
+	public static Programme newInstance(String code, String description) {
-        Programme p = new Programme();
+		Programme p = new Programme();
-        p.code = code;
+		p.code = code;
-        p.description = description;
+		p.description = description;
-        return p;
+		return p;
-    }
+	}
 }

dump-schema/src/main/java/eu/dnetlib/dhp/eosc/model/Project.java

@@ -1,17 +1,15 @@
+
 package eu.dnetlib.dhp.eosc.model;
 
 /**
  * @author miriam.baglioni
  * @Date 25/10/23
  */
-
-
 import java.io.Serializable;
 import java.util.List;
 
 import com.github.imifou.jsonschema.module.addon.annotation.JsonSchema;
 
-
 /**
  * This is the class representing the Project in the model used for the dumps of the whole graph. At the moment the dump
  * of the Projects differs from the other dumps because we do not create relations between Funders (Organization) and
@@ -43,164 +41,162 @@ import com.github.imifou.jsonschema.module.addon.annotation.JsonSchema;
  */
 
 public class Project implements Serializable {
-    private String id;
+	private String id;
 
-    private String websiteurl;
+	private String websiteurl;
-    private String code;
+	private String code;
-    private String acronym;
+	private String acronym;
-    private String title;
+	private String title;
 
-    private String startdate;
+	private String startdate;
 
-    private String enddate;
+	private String enddate;
 
-    private String callidentifier;
+	private String callidentifier;
 
-    private String keywords;
+	private String keywords;
 
-    private boolean openaccessmandateforpublications;
+	private boolean openaccessmandateforpublications;
 
-    private boolean openaccessmandatefordataset;
+	private boolean openaccessmandatefordataset;
-    private List<String> subject;
+	private List<String> subject;
 
-    @JsonSchema(description = "Funding information for the project")
+	@JsonSchema(description = "Funding information for the project")
-    private List<Funder> funding;
+	private List<Funder> funding;
 
-    private String summary;
+	private String summary;
 
-    @JsonSchema(description = "The money granted to the project")
+	@JsonSchema(description = "The money granted to the project")
-    private Granted granted;
+	private Granted granted;
 
-    @JsonSchema(description = "The h2020 programme funding the project")
+	@JsonSchema(description = "The h2020 programme funding the project")
-    private List<Programme> h2020programme;
+	private List<Programme> h2020programme;
 
+	public String getId() {
+		return id;
+	}
 
+	public void setId(String id) {
+		this.id = id;
+	}
 
-    public String getId() {
+	public String getWebsiteurl() {
-        return id;
+		return websiteurl;
-    }
+	}
 
-    public void setId(String id) {
+	public void setWebsiteurl(String websiteurl) {
-        this.id = id;
+		this.websiteurl = websiteurl;
-    }
+	}
 
-    public String getWebsiteurl() {
+	public String getCode() {
-        return websiteurl;
+		return code;
-    }
+	}
 
-    public void setWebsiteurl(String websiteurl) {
+	public void setCode(String code) {
-        this.websiteurl = websiteurl;
+		this.code = code;
-    }
+	}
 
-    public String getCode() {
+	public String getAcronym() {
-        return code;
+		return acronym;
-    }
+	}
 
-    public void setCode(String code) {
+	public void setAcronym(String acronym) {
-        this.code = code;
+		this.acronym = acronym;
-    }
+	}
 
-    public String getAcronym() {
+	public String getTitle() {
-        return acronym;
+		return title;
-    }
+	}
 
-    public void setAcronym(String acronym) {
+	public void setTitle(String title) {
-        this.acronym = acronym;
+		this.title = title;
-    }
+	}
 
-    public String getTitle() {
+	public String getStartdate() {
-        return title;
+		return startdate;
-    }
+	}
 
-    public void setTitle(String title) {
+	public void setStartdate(String startdate) {
-        this.title = title;
+		this.startdate = startdate;
-    }
+	}
 
-    public String getStartdate() {
+	public String getEnddate() {
-        return startdate;
+		return enddate;
-    }
+	}
 
-    public void setStartdate(String startdate) {
+	public void setEnddate(String enddate) {
-        this.startdate = startdate;
+		this.enddate = enddate;
-    }
+	}
 
-    public String getEnddate() {
+	public String getCallidentifier() {
-        return enddate;
+		return callidentifier;
-    }
+	}
 
-    public void setEnddate(String enddate) {
+	public void setCallidentifier(String callidentifier) {
-        this.enddate = enddate;
+		this.callidentifier = callidentifier;
-    }
+	}
 
-    public String getCallidentifier() {
+	public String getKeywords() {
-        return callidentifier;
+		return keywords;
-    }
+	}
 
-    public void setCallidentifier(String callidentifier) {
+	public void setKeywords(String keywords) {
-        this.callidentifier = callidentifier;
+		this.keywords = keywords;
-    }
+	}
 
-    public String getKeywords() {
+	public boolean isOpenaccessmandateforpublications() {
-        return keywords;
+		return openaccessmandateforpublications;
-    }
+	}
 
-    public void setKeywords(String keywords) {
+	public void setOpenaccessmandateforpublications(boolean openaccessmandateforpublications) {
-        this.keywords = keywords;
+		this.openaccessmandateforpublications = openaccessmandateforpublications;
-    }
+	}
 
-    public boolean isOpenaccessmandateforpublications() {
+	public boolean isOpenaccessmandatefordataset() {
-        return openaccessmandateforpublications;
+		return openaccessmandatefordataset;
-    }
+	}
 
-    public void setOpenaccessmandateforpublications(boolean openaccessmandateforpublications) {
+	public void setOpenaccessmandatefordataset(boolean openaccessmandatefordataset) {
-        this.openaccessmandateforpublications = openaccessmandateforpublications;
+		this.openaccessmandatefordataset = openaccessmandatefordataset;
-    }
+	}
 
-    public boolean isOpenaccessmandatefordataset() {
+	public List<String> getSubject() {
-        return openaccessmandatefordataset;
+		return subject;
-    }
+	}
 
-    public void setOpenaccessmandatefordataset(boolean openaccessmandatefordataset) {
+	public void setSubject(List<String> subject) {
-        this.openaccessmandatefordataset = openaccessmandatefordataset;
+		this.subject = subject;
-    }
+	}
 
-    public List<String> getSubject() {
+	public List<Funder> getFunding() {
-        return subject;
+		return funding;
-    }
+	}
 
-    public void setSubject(List<String> subject) {
+	public void setFunding(List<Funder> funding) {
-        this.subject = subject;
+		this.funding = funding;
-    }
+	}
 
-    public List<Funder> getFunding() {
+	public String getSummary() {
-        return funding;
+		return summary;
-    }
+	}
 
-    public void setFunding(List<Funder> funding) {
+	public void setSummary(String summary) {
-        this.funding = funding;
+		this.summary = summary;
-    }
+	}
 
-    public String getSummary() {
+	public Granted getGranted() {
-        return summary;
+		return granted;
-    }
+	}
 
-    public void setSummary(String summary) {
+	public void setGranted(Granted granted) {
-        this.summary = summary;
+		this.granted = granted;
-    }
+	}
 
-    public Granted getGranted() {
+	public List<Programme> getH2020programme() {
-        return granted;
+		return h2020programme;
-    }
+	}
 
-    public void setGranted(Granted granted) {
+	public void setH2020programme(List<Programme> h2020programme) {
-        this.granted = granted;
+		this.h2020programme = h2020programme;
-    }
+	}
-
-    public List<Programme> getH2020programme() {
-        return h2020programme;
-    }
-
-    public void setH2020programme(List<Programme> h2020programme) {
-        this.h2020programme = h2020programme;
-    }
 }

dump/src/main/java/eu/dnetlib/dhp/oa/graph/dump/eosc/ExtendEoscResultWithOrganization.java

@@ -6,7 +6,6 @@ import static eu.dnetlib.dhp.common.SparkSessionSupport.runWithSparkSession;
 import java.io.Serializable;
 import java.util.*;
 
-import eu.dnetlib.dhp.eosc.model.Affiliation;
 import org.apache.commons.io.IOUtils;
 import org.apache.spark.SparkConf;
 import org.apache.spark.api.java.function.FilterFunction;
@@ -20,6 +19,7 @@ import org.slf4j.Logger;
 import org.slf4j.LoggerFactory;
 
 import eu.dnetlib.dhp.application.ArgumentApplicationParser;
+import eu.dnetlib.dhp.eosc.model.Affiliation;
 import eu.dnetlib.dhp.eosc.model.OrganizationPid;
 import eu.dnetlib.dhp.eosc.model.Result;
 import eu.dnetlib.dhp.schema.common.ModelConstants;

dump/src/main/java/eu/dnetlib/dhp/oa/graph/dump/eosc/ExtendEoscResultWithOrganizationStep2.java

@@ -7,8 +7,6 @@ import java.io.Serializable;
 import java.util.*;
 import java.util.stream.Collectors;
 
-import eu.dnetlib.dhp.eosc.model.Affiliation;
-import eu.dnetlib.dhp.eosc.model.Country;
 import org.apache.commons.io.IOUtils;
 import org.apache.spark.SparkConf;
 import org.apache.spark.api.java.function.FilterFunction;
@@ -22,6 +20,8 @@ import org.slf4j.Logger;
 import org.slf4j.LoggerFactory;
 
 import eu.dnetlib.dhp.application.ArgumentApplicationParser;
+import eu.dnetlib.dhp.eosc.model.Affiliation;
+import eu.dnetlib.dhp.eosc.model.Country;
 import eu.dnetlib.dhp.eosc.model.OrganizationPid;
 import eu.dnetlib.dhp.eosc.model.Result;
 import eu.dnetlib.dhp.schema.common.ModelConstants;
@@ -70,7 +70,7 @@ public class ExtendEoscResultWithOrganizationStep2 implements Serializable {
 			isSparkSessionManaged,
 			spark -> {
 				Utils.removeOutputDir(spark, workingPath + "publicationextendedaffiliation");
-				addOrganizations(spark, inputPath, workingPath , outputPath);
+				addOrganizations(spark, inputPath, workingPath, outputPath);
 			});
 	}
 
@@ -156,85 +156,89 @@ public class ExtendEoscResultWithOrganizationStep2 implements Serializable {
 			.json(workingPath + "publicationextendedaffiliation");
 
 		relations
-				.joinWith(organizations, relations.col("source").equalTo(organizations.col("id")))
+			.joinWith(organizations, relations.col("source").equalTo(organizations.col("id")))
-				.map((MapFunction<Tuple2<Relation, Organization>, eu.dnetlib.dhp.eosc.model.Organization>) t2 -> mapOrganization(t2._2()),Encoders.bean(eu.dnetlib.dhp.eosc.model.Organization.class))
+			.map(
-				.filter(Objects::nonNull)
+				(MapFunction<Tuple2<Relation, Organization>, eu.dnetlib.dhp.eosc.model.Organization>) t2 -> mapOrganization(
-				.write()
+					t2._2()),
-				.mode(SaveMode.Overwrite)
+				Encoders.bean(eu.dnetlib.dhp.eosc.model.Organization.class))
-				.option("compression","gzip")
+			.filter(Objects::nonNull)
-				.json(outputPath + "organization");
+			.write()
+			.mode(SaveMode.Overwrite)
+			.option("compression", "gzip")
+			.json(outputPath + "organization");
 
 		relations
-				.joinWith(organizations, relations.col("source").equalTo(organizations.col("id")))
+			.joinWith(organizations, relations.col("source").equalTo(organizations.col("id")))
-				.map((MapFunction<Tuple2<Relation, Organization>, eu.dnetlib.dhp.eosc.model.Relation>) t2 -> eu.dnetlib.dhp.eosc.model.Relation.newInstance(t2._1().getSource(), t2._1().getTarget()), Encoders.bean(eu.dnetlib.dhp.eosc.model.Relation.class) )
+			.map(
-				.write()
+				(MapFunction<Tuple2<Relation, Organization>, eu.dnetlib.dhp.eosc.model.Relation>) t2 -> eu.dnetlib.dhp.eosc.model.Relation
-				.mode(SaveMode.Overwrite)
+					.newInstance(t2._1().getSource(), t2._1().getTarget()),
-				.option("compression","gzip")
+				Encoders.bean(eu.dnetlib.dhp.eosc.model.Relation.class))
-				.json(outputPath + "resultOrganization");
+			.write()
+			.mode(SaveMode.Overwrite)
+			.option("compression", "gzip")
+			.json(outputPath + "resultOrganization");
 
 	}
 
-	private static eu.dnetlib.dhp.eosc.model.Organization mapOrganization(Organization org){
+	private static eu.dnetlib.dhp.eosc.model.Organization mapOrganization(Organization org) {
 
-			if (Boolean.TRUE.equals(org.getDataInfo().getDeletedbyinference()))
+		if (Boolean.TRUE.equals(org.getDataInfo().getDeletedbyinference()))
-				return null;
+			return null;
-			if (!Optional.ofNullable(org.getLegalname()).isPresent()
+		if (!Optional.ofNullable(org.getLegalname()).isPresent()
-					&& !Optional.ofNullable(org.getLegalshortname()).isPresent())
+			&& !Optional.ofNullable(org.getLegalshortname()).isPresent())
-				return null;
+			return null;
 
 		eu.dnetlib.dhp.eosc.model.Organization organization = new eu.dnetlib.dhp.eosc.model.Organization();
 
-			Optional
+		Optional
-					.ofNullable(org.getLegalshortname())
+			.ofNullable(org.getLegalshortname())
-					.ifPresent(value -> organization.setLegalshortname(value.getValue()));
+			.ifPresent(value -> organization.setLegalshortname(value.getValue()));
 
-			Optional
+		Optional
-					.ofNullable(org.getLegalname())
+			.ofNullable(org.getLegalname())
-					.ifPresent(value -> organization.setLegalname(value.getValue()));
+			.ifPresent(value -> organization.setLegalname(value.getValue()));
 
-			Optional
+		Optional
-					.ofNullable(org.getWebsiteurl())
+			.ofNullable(org.getWebsiteurl())
-					.ifPresent(value -> organization.setWebsiteurl(value.getValue()));
+			.ifPresent(value -> organization.setWebsiteurl(value.getValue()));
 
-			Optional
+		Optional
-					.ofNullable(org.getAlternativeNames())
+			.ofNullable(org.getAlternativeNames())
-					.ifPresent(
+			.ifPresent(
-							value -> organization
+				value -> organization
-									.setAlternativenames(
+					.setAlternativenames(
-											value
+						value
-													.stream()
+							.stream()
-													.map(v -> v.getValue())
+							.map(v -> v.getValue())
-													.collect(Collectors.toList())));
+							.collect(Collectors.toList())));
 
-			Optional
+		Optional
-					.ofNullable(org.getCountry())
+			.ofNullable(org.getCountry())
-					.ifPresent(
+			.ifPresent(
-							value -> {
+				value -> {
-								if (!value.getClassid().equals(UNKNOWN)) {
+					if (!value.getClassid().equals(UNKNOWN)) {
-									organization
+						organization
-											.setCountry(
+							.setCountry(
-													Country.newInstance(value.getClassid(), value.getClassname()));
+								Country.newInstance(value.getClassid(), value.getClassname()));
-								}
+					}
 
-							});
+				});
 
-			Optional
+		Optional
-					.ofNullable(org.getId())
+			.ofNullable(org.getId())
-					.ifPresent(value -> organization.setId(value));
+			.ifPresent(value -> organization.setId(value));
 
-			Optional
+		Optional
-					.ofNullable(org.getPid())
+			.ofNullable(org.getPid())
-					.ifPresent(
+			.ifPresent(
-							value -> organization
+				value -> organization
-									.setPid(
+					.setPid(
-											value
+						value
-													.stream()
+							.stream()
-													.map(p -> OrganizationPid.newInstance(p.getQualifier().getClassid(), p.getValue()))
+							.map(p -> OrganizationPid.newInstance(p.getQualifier().getClassid(), p.getValue()))
-													.collect(Collectors.toList())));
+							.collect(Collectors.toList())));
-
-			return organization;
-		}
 
+		return organization;
 	}
 
-
+}

dump/src/main/java/eu/dnetlib/dhp/oa/graph/dump/eosc/SparkDumpOrganizationProject.java

@@ -6,8 +6,12 @@ import static eu.dnetlib.dhp.common.SparkSessionSupport.runWithSparkSession;
 import java.io.Serializable;
 import java.util.Optional;
 
+import eu.dnetlib.dhp.eosc.model.Organization;
+import eu.dnetlib.dhp.eosc.model.Project;
+import eu.dnetlib.dhp.schema.common.ModelConstants;
 import org.apache.commons.io.IOUtils;
 import org.apache.spark.SparkConf;
+import org.apache.spark.api.java.function.FilterFunction;
 import org.apache.spark.api.java.function.MapFunction;
 import org.apache.spark.sql.Dataset;
 import org.apache.spark.sql.Encoders;
@@ -21,19 +25,20 @@ import eu.dnetlib.dhp.eosc.model.Provenance;
 import eu.dnetlib.dhp.eosc.model.RelType;
 import eu.dnetlib.dhp.schema.oaf.DataInfo;
 import eu.dnetlib.dhp.schema.oaf.Relation;
+import scala.Tuple2;
 
 /**
  * @author miriam.baglioni
  * @Date 12/01/23
  */
-public class SparkDumpRelation implements Serializable {
+public class SparkDumpOrganizationProject implements Serializable {
 
-	private static final Logger log = LoggerFactory.getLogger(SparkDumpRelation.class);
+	private static final Logger log = LoggerFactory.getLogger(SparkDumpOrganizationProject.class);
 
 	public static void main(String[] args) throws Exception {
 		String jsonConfiguration = IOUtils
 			.toString(
-				SparkDumpRelation.class
+				SparkDumpOrganizationProject.class
 					.getResourceAsStream(
 						"/eu/dnetlib/dhp/oa/graph/dump/input_relationdump_parameters.json"));
 
@@ -66,49 +71,21 @@ public class SparkDumpRelation implements Serializable {
 	}
 
 	private static void dumpRelation(SparkSession spark, String inputPath, String outputPath) {
-		Dataset<Relation> relations = Utils.readPath(spark, inputPath, Relation.class);
+		Dataset<Organization> organization = Utils.readPath(spark, outputPath + "organization", Organization.class);
-		relations
+		Dataset<Project> project = Utils.readPath(spark, outputPath + "project", Project.class);
 
-			.map((MapFunction<Relation, eu.dnetlib.dhp.eosc.model.Relation>) relation -> {
+		Dataset<Relation> relation = Utils.readPath(spark, inputPath + "/relation", Relation.class)
-				eu.dnetlib.dhp.eosc.model.Relation relNew = new eu.dnetlib.dhp.eosc.model.Relation();
+				.filter((FilterFunction<Relation>) r-> !r.getDataInfo().getDeletedbyinference() && r.getRelClass().equalsIgnoreCase(ModelConstants.IS_PARTICIPANT));
-				relNew
-					.setSource(
 
-						relation.getSource());
+		Dataset<Relation> eoscOrgs = relation.joinWith(organization, relation.col("source").equalTo(organization.col("id")))
+				.map((MapFunction<Tuple2<Relation, Organization>, Relation>) t2 -> t2._1(), Encoders.bean(Relation.class));
 
-				relNew
+		eoscOrgs.joinWith(project, eoscOrgs.col("target").equalTo(project.col("id")))
-					.setTarget(
+				.map((MapFunction<Tuple2<Relation, Project>, eu.dnetlib.dhp.eosc.model.Relation>) t2-> eu.dnetlib.dhp.eosc.model.Relation.newInstance(t2._1().getSource(), t2._1().getTarget()), Encoders.bean(eu.dnetlib.dhp.eosc.model.Relation.class))
-
+				.write()
-						relation.getTarget());
+				.mode(SaveMode.Overwrite)
-
+				.option("compression","gzip")
-				relNew
+				.json(outputPath + "organizationProject");
-					.setReltype(
-						RelType
-							.newInstance(
-								relation.getRelClass(),
-								relation.getSubRelType()));
-
-				Optional<DataInfo> odInfo = Optional.ofNullable(relation.getDataInfo());
-				if (odInfo.isPresent()) {
-					DataInfo dInfo = odInfo.get();
-					if (Optional.ofNullable(dInfo.getProvenanceaction()).isPresent() &&
-						Optional.ofNullable(dInfo.getProvenanceaction().getClassname()).isPresent()) {
-						relNew
-							.setProvenance(
-								Provenance
-									.newInstance(
-										dInfo.getProvenanceaction().getClassname(),
-										dInfo.getTrust()));
-					}
-				}
-
-				return relNew;
-
-			}, Encoders.bean(eu.dnetlib.dhp.eosc.model.Relation.class))
-			.write()
-			.option("compression", "gzip")
-			.mode(SaveMode.Append)
-			.json(outputPath);
 
 	}
 

dump/src/main/java/eu/dnetlib/dhp/oa/graph/dump/eosc/SparkUpdateProjectInfo.java

@@ -10,9 +10,6 @@ import java.util.List;
 import java.util.Optional;
 import java.util.stream.Collectors;
 
-import eu.dnetlib.dhp.eosc.model.*;
-import eu.dnetlib.dhp.schema.oaf.Field;
-import eu.dnetlib.dhp.schema.oaf.Project;
 import org.apache.commons.io.IOUtils;
 import org.apache.spark.SparkConf;
 import org.apache.spark.api.java.function.FlatMapFunction;
@@ -31,7 +28,10 @@ import org.slf4j.LoggerFactory;
 import com.fasterxml.jackson.databind.ObjectMapper;
 
 import eu.dnetlib.dhp.application.ArgumentApplicationParser;
+import eu.dnetlib.dhp.eosc.model.*;
 import eu.dnetlib.dhp.oa.graph.dump.Constants;
+import eu.dnetlib.dhp.schema.oaf.Field;
+import eu.dnetlib.dhp.schema.oaf.Project;
 import scala.Array;
 import scala.Tuple2;
 
@@ -110,68 +110,86 @@ public class SparkUpdateProjectInfo implements Serializable {
 
 		Dataset<Project> project = Utils.readPath(spark, inputPath + "/project", Project.class);
 
-		Dataset<String> projectIds = result.joinWith(resultProject, result.col("id").equalTo(resultProject.col("resultId")))
+		Dataset<String> projectIds = result
-				.flatMap((FlatMapFunction<Tuple2<Result, ResultProject>, String>) t2 -> t2._2().getProjectsList()
+			.joinWith(resultProject, result.col("id").equalTo(resultProject.col("resultId")))
-						.stream().map(p -> p.getId()).collect(Collectors.toList()).iterator(), Encoders.STRING())
+			.flatMap(
-				.distinct();
+				(FlatMapFunction<Tuple2<Result, ResultProject>, String>) t2 -> t2
+					._2()
+					.getProjectsList()
+					.stream()
+					.map(p -> p.getId())
+					.collect(Collectors.toList())
+					.iterator(),
+				Encoders.STRING())
+			.distinct();
 
-		projectIds.joinWith(project, projectIds.col("value").equalTo(project.col("id")))
+		projectIds
-				.map((MapFunction<Tuple2<String, Project>, eu.dnetlib.dhp.eosc.model.Project>)t2->mapProject(t2._2()), Encoders.bean(eu.dnetlib.dhp.eosc.model.Project.class) )
+			.joinWith(project, projectIds.col("value").equalTo(project.col("id")))
-				.write()
+			.map(
-				.mode(SaveMode.Overwrite)
+				(MapFunction<Tuple2<String, Project>, eu.dnetlib.dhp.eosc.model.Project>) t2 -> mapProject(t2._2()),
-				.option("compression","gzip")
+				Encoders.bean(eu.dnetlib.dhp.eosc.model.Project.class))
-				.json(outputPath + "project");
+			.write()
+			.mode(SaveMode.Overwrite)
+			.option("compression", "gzip")
+			.json(outputPath + "project");
 
-		resultProject.flatMap((FlatMapFunction<ResultProject, Relation>) rp ->
+		resultProject
-			rp.getProjectsList().stream().map(p -> Relation.newInstance(rp.getResultId(), p.getId()))
+			.flatMap(
-					.collect(Collectors.toList()).iterator(), Encoders.bean(Relation.class))
+				(FlatMapFunction<ResultProject, Relation>) rp -> rp
-				.write()
+					.getProjectsList()
-				.mode(SaveMode.Overwrite)
+					.stream()
-				.option("compression","gzip")
+					.map(p -> Relation.newInstance(rp.getResultId(), p.getId()))
-				.json(outputPath + "resultProject");
+					.collect(Collectors.toList())
+					.iterator(),
+				Encoders.bean(Relation.class))
+			.write()
+			.mode(SaveMode.Overwrite)
+			.option("compression", "gzip")
+			.json(outputPath + "resultProject");
 	}
 
-	private static eu.dnetlib.dhp.eosc.model.Project mapProject(eu.dnetlib.dhp.schema.oaf.Project p) throws DocumentException {
+	private static eu.dnetlib.dhp.eosc.model.Project mapProject(eu.dnetlib.dhp.schema.oaf.Project p)
+		throws DocumentException {
 		if (Boolean.TRUE.equals(p.getDataInfo().getDeletedbyinference()))
 			return null;
 
 		eu.dnetlib.dhp.eosc.model.Project project = new eu.dnetlib.dhp.eosc.model.Project();
 
 		Optional
-				.ofNullable(p.getId())
+			.ofNullable(p.getId())
-				.ifPresent(id -> project.setId(id));
+			.ifPresent(id -> project.setId(id));
 
 		Optional
-				.ofNullable(p.getWebsiteurl())
+			.ofNullable(p.getWebsiteurl())
-				.ifPresent(w -> project.setWebsiteurl(w.getValue()));
+			.ifPresent(w -> project.setWebsiteurl(w.getValue()));
 
 		Optional
-				.ofNullable(p.getCode())
+			.ofNullable(p.getCode())
-				.ifPresent(code -> project.setCode(code.getValue()));
+			.ifPresent(code -> project.setCode(code.getValue()));
 
 		Optional
-				.ofNullable(p.getAcronym())
+			.ofNullable(p.getAcronym())
-				.ifPresent(acronynim -> project.setAcronym(acronynim.getValue()));
+			.ifPresent(acronynim -> project.setAcronym(acronynim.getValue()));
 
 		Optional
-				.ofNullable(p.getTitle())
+			.ofNullable(p.getTitle())
-				.ifPresent(title -> project.setTitle(title.getValue()));
+			.ifPresent(title -> project.setTitle(title.getValue()));
 
 		Optional
-				.ofNullable(p.getStartdate())
+			.ofNullable(p.getStartdate())
-				.ifPresent(sdate -> project.setStartdate(sdate.getValue()));
+			.ifPresent(sdate -> project.setStartdate(sdate.getValue()));
 
 		Optional
-				.ofNullable(p.getEnddate())
+			.ofNullable(p.getEnddate())
-				.ifPresent(edate -> project.setEnddate(edate.getValue()));
+			.ifPresent(edate -> project.setEnddate(edate.getValue()));
 
 		Optional
-				.ofNullable(p.getCallidentifier())
+			.ofNullable(p.getCallidentifier())
-				.ifPresent(cide -> project.setCallidentifier(cide.getValue()));
+			.ifPresent(cide -> project.setCallidentifier(cide.getValue()));
 
 		Optional
-				.ofNullable(p.getKeywords())
+			.ofNullable(p.getKeywords())
-				.ifPresent(key -> project.setKeywords(key.getValue()));
+			.ifPresent(key -> project.setKeywords(key.getValue()));
 
 		Optional<Field<String>> omandate = Optional.ofNullable(p.getOamandatepublications());
 		Optional<Field<String>> oecsc39 = Optional.ofNullable(p.getEcsc39());
@@ -191,19 +209,19 @@ public class SparkUpdateProjectInfo implements Serializable {
 		project.setOpenaccessmandatefordataset(false);
 
 		Optional
-				.ofNullable(p.getEcarticle29_3())
+			.ofNullable(p.getEcarticle29_3())
-				.ifPresent(oamandate -> project.setOpenaccessmandatefordataset(oamandate.getValue().equals("true")));
+			.ifPresent(oamandate -> project.setOpenaccessmandatefordataset(oamandate.getValue().equals("true")));
 
 		project
-				.setSubject(
+			.setSubject(
-						Optional
+				Optional
-								.ofNullable(p.getSubjects())
+					.ofNullable(p.getSubjects())
-								.map(subjs -> subjs.stream().map(s -> s.getValue()).collect(Collectors.toList()))
+					.map(subjs -> subjs.stream().map(s -> s.getValue()).collect(Collectors.toList()))
-								.orElse(new ArrayList<>()));
+					.orElse(new ArrayList<>()));
 
 		Optional
-				.ofNullable(p.getSummary())
+			.ofNullable(p.getSummary())
-				.ifPresent(summary -> project.setSummary(summary.getValue()));
+			.ifPresent(summary -> project.setSummary(summary.getValue()));
 
 		Optional<Float> ofundedamount = Optional.ofNullable(p.getFundedamount());
 		Optional<Field<String>> ocurrency = Optional.ofNullable(p.getCurrency());
@@ -213,8 +231,8 @@ public class SparkUpdateProjectInfo implements Serializable {
 			if (ofundedamount.isPresent()) {
 				if (ototalcost.isPresent()) {
 					project
-							.setGranted(
+						.setGranted(
-									Granted.newInstance(ocurrency.get().getValue(), ototalcost.get(), ofundedamount.get()));
+							Granted.newInstance(ocurrency.get().getValue(), ototalcost.get(), ofundedamount.get()));
 				} else {
 					project.setGranted(Granted.newInstance(ocurrency.get().getValue(), ofundedamount.get()));
 				}
@@ -222,21 +240,21 @@ public class SparkUpdateProjectInfo implements Serializable {
 		}
 
 		project
-				.setH2020programme(
+			.setH2020programme(
-						Optional
+				Optional
-								.ofNullable(p.getH2020classification())
+					.ofNullable(p.getH2020classification())
-								.map(
+					.map(
-										classification -> classification
+						classification -> classification
-												.stream()
+							.stream()
-												.map(
+							.map(
-														c -> Programme
+								c -> Programme
-																.newInstance(
+									.newInstance(
-																		c.getH2020Programme().getCode(), c.getH2020Programme().getDescription()))
+										c.getH2020Programme().getCode(), c.getH2020Programme().getDescription()))
-												.collect(Collectors.toList()))
+							.collect(Collectors.toList()))
-								.orElse(new ArrayList<>()));
+					.orElse(new ArrayList<>()));
 
 		Optional<List<Field<String>>> ofundTree = Optional
-				.ofNullable(p.getFundingtree());
+			.ofNullable(p.getFundingtree());
 		List<Funder> funList = new ArrayList<>();
 		if (ofundTree.isPresent()) {
 			for (Field<String> fundingtree : ofundTree.get()) {

dump/src/main/resources/eu/dnetlib/dhp/oa/graph/dump/eosc_dump_organizationprojectrelations.json

@@ -0,0 +1,26 @@
+
+[
+
+
+  {
+    "paramName":"s",
+    "paramLongName":"sourcePath",
+    "paramDescription": "the name node",
+    "paramRequired": true
+  },
+  {
+    "paramName": "out",
+    "paramLongName": "outputPath",
+    "paramDescription": "the path used to store temporary output files",
+    "paramRequired": true
+  },
+  {
+    "paramName": "issm",
+    "paramLongName": "isSparkSessionManaged",
+    "paramDescription": "the path used to store temporary output files",
+    "paramRequired": false
+  }
+]
+
+
+

dump/src/test/java/eu/dnetlib/dhp/oa/graph/dump/eosc/SelectEoscResultTest.java

@@ -24,8 +24,8 @@ import org.slf4j.LoggerFactory;
 
 import com.fasterxml.jackson.databind.ObjectMapper;
 
-import eu.dnetlib.dhp.eosc.model.Indicator;
 import eu.dnetlib.dhp.eosc.model.Affiliation;
+import eu.dnetlib.dhp.eosc.model.Indicator;
 import eu.dnetlib.dhp.eosc.model.Result;
 import eu.dnetlib.dhp.schema.action.AtomicAction;
 import scala.Tuple2;