D-Net

{"type_source": "SVN", "goal": "package -U -T 4C source:jar", "url": "http://svn-public.driver.research-infrastructures.eu/driver/dnet40/modules/dnet-openaireplus-mapping-utils/trunk/", "deploy_repository": "dnet4-snapshots", "version": "4", "mail": "sandro.labruzzo@isti.cnr.it,michele.artini@isti.cnr.it, claudio.atzori@isti.cnr.it, alessia.bardi@isti.cnr.it", "deploy_repository_url": "http://maven.research-infrastructures.eu/nexus/content/repositories/dnet4-snapshots", "name": "dnet-openaireplus-mapping-utils"}

package eu.dnetlib.pace.clustering;

import org.junit.Before;

import org.junit.Test;

import eu.dnetlib.pace.AbstractProtoPaceTest;

import eu.dnetlib.pace.config.Config;

import eu.dnetlib.pace.config.Type;

import eu.dnetlib.pace.model.FieldListImpl;

import eu.dnetlib.pace.model.FieldValueImpl;

import eu.dnetlib.pace.model.MapDocument;

public class BlacklistAwareClusteringCombinerTest extends AbstractProtoPaceTest {

	private Config config;

	@Before

	public void setUp() {

		config = getResultFullConf();

	}

	@Test

	public void testCombine() {

		final MapDocument result =

				result(config, "A", "Dipping in Cygnus X-2 in a multi-wavelength campaign due to absorption of extended ADC emission", "2013");

		final FieldListImpl fl = new FieldListImpl();

		fl.add(new FieldValueImpl(Type.String, "desc", "hello world description pipeline"));

		result.getFieldMap().put("desc", fl);

		fl.clear();

		fl.add(new FieldValueImpl(Type.String, "title", "lorem ipsum cabalie qwerty"));

		final FieldListImpl field = (FieldListImpl) result.getFieldMap().get("title");

		field.add(fl);

		System.out.println(BlacklistAwareClusteringCombiner.filterAndCombine(result, config, config.blacklists()));

	}

}

package eu.dnetlib.pace.clustering;

import org.junit.Before;

import org.junit.Test;

import eu.dnetlib.pace.AbstractProtoPaceTest;

import eu.dnetlib.pace.config.Config;

import eu.dnetlib.pace.config.Type;

import eu.dnetlib.pace.model.FieldListImpl;

import eu.dnetlib.pace.model.FieldValueImpl;

import eu.dnetlib.pace.model.MapDocument;

public class ClusteringCombinerTest extends AbstractProtoPaceTest {

	private Config config;

	@Before

	public void setUp() {

		config = getResultFullConf();

	}

	@Test

	public void testCombine() {

		String title = "Dipping in Cygnus X-2 in a multi-wavelength campaign due to absorption of extended ADC emission";

		MapDocument result = result(config, "A", title, "2013");

		FieldListImpl fl = new FieldListImpl();

		fl.add(new FieldValueImpl(Type.String, "desc", "lorem ipsum cabalie qwerty"));

		result.getFieldMap().put("desc", fl);

		System.out.println(title);

		System.out.println(ClusteringCombiner.combine(result, config));

	}

}

package eu.dnetlib.pace;

import java.io.IOException;

import java.io.StringWriter;

import java.util.List;

import org.apache.commons.io.IOUtils;

import org.apache.commons.lang.RandomStringUtils;

import org.apache.commons.lang.StringUtils;

import com.google.gson.Gson;

import eu.dnetlib.data.mapreduce.util.OafTest;

import eu.dnetlib.data.proto.FieldTypeProtos.Qualifier;

import eu.dnetlib.data.proto.FieldTypeProtos.StructuredProperty;

import eu.dnetlib.data.proto.FieldTypeProtos.StructuredProperty.Builder;

import eu.dnetlib.data.proto.OafProtos.Oaf;

import eu.dnetlib.data.proto.OafProtos.OafEntity;

import eu.dnetlib.data.proto.OrganizationProtos.Organization;

import eu.dnetlib.data.proto.PersonProtos.Person;

import eu.dnetlib.data.proto.ResultProtos.Result;

import eu.dnetlib.pace.config.Config;

import eu.dnetlib.pace.config.DedupConfig;

import eu.dnetlib.pace.config.Type;

import eu.dnetlib.pace.model.Field;

import eu.dnetlib.pace.model.FieldValueImpl;

import eu.dnetlib.pace.model.MapDocument;

import eu.dnetlib.pace.model.ProtoDocumentBuilder;

import eu.dnetlib.pace.model.gt.GTAuthor;

import eu.dnetlib.pace.model.gt.GTAuthorMapper;

public abstract class AbstractProtoPaceTest extends OafTest {

	protected DedupConfig getResultFullConf() {

		return DedupConfig.load(readFromClasspath("/eu/dnetlib/pace/result.full.pace.conf"));

	}

	protected DedupConfig getResultSimpleConf() {

		return DedupConfig.load(readFromClasspath("/eu/dnetlib/pace/result.simple.pace.conf"));

	}

	protected DedupConfig getResultConf() {

		return DedupConfig.load(readFromClasspath("/eu/dnetlib/pace/result.pace.conf"));

	}

	protected DedupConfig getOrganizationSimpleConf() {

		return DedupConfig.load(readFromClasspath("/eu/dnetlib/pace/organization.pace.conf"));

	}

	protected DedupConfig getResultAuthorsConf() {

		return DedupConfig.load(readFromClasspath("/eu/dnetlib/pace/result.authors.pace.conf"));

	}

	protected DedupConfig getPersonConf() {

		return DedupConfig.load(readFromClasspath("/eu/dnetlib/pace/person.pace.conf"));

	}

	protected DedupConfig getResultProdConf() {

		return DedupConfig.load(readFromClasspath("/eu/dnetlib/pace/result.prod.pace.conf"));

	}

	protected MapDocument person(final Config conf, final String id, final Oaf oaf) {

		return ProtoDocumentBuilder.newInstance(id, oaf.getEntity(), conf.model());

	}

	protected Oaf getPersonGT(final String path) {

		return new GTAuthorMapper().map(getGTAuthor(path));

	}

	protected GTAuthor getGTAuthor(final String path) {

		final Gson gson = new Gson();

		final String json = readFromClasspath(path);

		final GTAuthor gta = gson.fromJson(json, GTAuthor.class);

		return gta;

	}

	private String readFromClasspath(final String filename) {

		final StringWriter sw = new StringWriter();

		try {

			IOUtils.copy(getClass().getResourceAsStream(filename), sw);

			return sw.toString();

		} catch (final IOException e) {

			throw new RuntimeException("cannot load resource from classpath: " + filename);

		}

	}

	protected MapDocument result(final Config config, final String id, final String title) {

		return result(config, id, title, null, null, null);

	}

	protected MapDocument result(final Config config, final String id, final String title, final String date) {

		return result(config, id, title, date, null, null);

	}

	protected MapDocument result(final Config config, final String id, final String title, final String date, final String pid) {

		return result(config, id, title, date, pid, null);

	}

	protected MapDocument result(final Config config, final String id, final String title, final String date, final String pid, final List<String> authors) {

		final Result.Metadata.Builder metadata = Result.Metadata.newBuilder();

		if (!StringUtils.isBlank(title)) {

			metadata.addTitle(getStruct(title, getQualifier("main title", "dnet:titles")));

			metadata.addTitle(getStruct(RandomStringUtils.randomAlphabetic(10), getQualifier("alternative title", "dnet:titles")));

		}

		if (!StringUtils.isBlank(date)) {

			metadata.setDateofacceptance(sf(date));

		}

		final OafEntity.Builder entity = oafEntity(id, eu.dnetlib.data.proto.TypeProtos.Type.result);

		final Result.Builder result = Result.newBuilder().setMetadata(metadata);

		if (authors != null) {

			for (final String author : authors) {

				result.addAuthor(person(author));

			}

		}

		entity.setResult(result);

		if (!StringUtils.isBlank(pid)) {

			entity.addPid(sp(pid, "doi"));

			entity.addPid(sp(RandomStringUtils.randomAlphabetic(10), "oai"));

		}

		final OafEntity build = entity.build();

		return ProtoDocumentBuilder.newInstance(id, build, config.model());

	}

	private Person.Builder person(final String author) {

		final Person.Builder person = Person.newBuilder();

		final eu.dnetlib.pace.model.Person p = new eu.dnetlib.pace.model.Person(author, false);

		final Person.Metadata.Builder metadata = Person.Metadata.newBuilder();

		if (p.isAccurate()) {

			metadata.setFirstname(sf(p.getNormalisedFirstName()));

			metadata.addSecondnames(sf(p.getNormalisedSurname()));

			metadata.setFullname(sf(p.getNormalisedFullname()));

		} else {

			metadata.setFullname(sf(p.getOriginal()));

		}

		return person.setMetadata(metadata);

	}

	private OafEntity.Builder oafEntity(final String id, final eu.dnetlib.data.proto.TypeProtos.Type type) {

		final OafEntity.Builder entity = OafEntity.newBuilder().setId(id).setType(type);

		return entity;

	}

	protected MapDocument organization(final Config config, final String id, final String legalName) {

		return organization(config, id, legalName, null);

	}

	protected MapDocument organization(final Config config, final String id, final String legalName, final String legalShortName) {

		final Organization.Metadata.Builder metadata = Organization.Metadata.newBuilder();

		if (legalName != null) {

			metadata.setLegalname(sf(legalName));

		}

		if (legalShortName != null) {

			metadata.setLegalshortname(sf(legalShortName));

		}

		final OafEntity.Builder entity = oafEntity(id, eu.dnetlib.data.proto.TypeProtos.Type.result);

		entity.setOrganization(Organization.newBuilder().setMetadata(metadata));

		return ProtoDocumentBuilder.newInstance(id, entity.build(), config.model());

	}

	private StructuredProperty sp(final String pid, final String type) {

		final Builder pidSp =

				StructuredProperty.newBuilder().setValue(pid)

						.setQualifier(Qualifier.newBuilder().setClassid(type).setClassname(type).setSchemeid("dnet:pid_types").setSchemename("dnet:pid_types"));

		return pidSp.build();

	}

	protected Field title(final String s) {

		return new FieldValueImpl(Type.String, "title", s);

	}

	protected static StructuredProperty.Builder getStruct(final String value, final Qualifier.Builder qualifier) {

		return StructuredProperty.newBuilder().setValue(value).setQualifier(qualifier);

	}

	/*

	 * protected static StringField.Builder sf(final String s) { return StringField.newBuilder().setValue(s); }

	 * 

	 * protected static Qualifier.Builder getQualifier(final String classname, final String schemename) { return

	 * Qualifier.newBuilder().setClassid(classname).setClassname(classname).setSchemeid(schemename).setSchemename(schemename); }

	 */

}

package eu.dnetlib.pace.model;

import static org.junit.Assert.assertFalse;

import static org.junit.Assert.assertTrue;

import org.junit.Test;

import com.google.common.collect.Iterables;

import com.google.common.collect.Sets;

import com.google.common.collect.Sets.SetView;

import eu.dnetlib.pace.AbstractProtoPaceTest;

import eu.dnetlib.pace.config.Config;

public class ProtoDocumentBuilderTest extends AbstractProtoPaceTest {

	@Test

	public void test_serialise1() {

		final String id = "12345";

		final Config config = getResultFullConf();

		final MapDocument document = ProtoDocumentBuilder.newInstance(id, getResult(id), config.model());

		assertFalse(document.fieldNames().isEmpty());

		assertFalse(Iterables.isEmpty(document.fields()));

		System.out.println("original:\n" + document);

		final String stringDoc = MapDocumentSerializer.toString(document);

		System.out.println("srialization:\n" + stringDoc);

		final MapDocument decoded = MapDocumentSerializer.decode(stringDoc.getBytes());

		final SetView<String> diff = Sets.difference(document.fieldNames(), decoded.fieldNames());

		assertTrue(diff.isEmpty());

		System.out.println("decoded:\n" + decoded);

	}

	@Test

	public void test_serialise2() {

		final String id = "12345";

		final String path = "/eu/dnetlib/pace/model/gt.author.manghi1.json";

		final Config config = getPersonConf();

		final MapDocument document = ProtoDocumentBuilder.newInstance(id, getPersonGT(path).getEntity(), config.model());

		assertFalse(document.fieldNames().isEmpty());

		assertFalse(Iterables.isEmpty(document.fields()));

		System.out.println("original:\n" + document);

		final String stringDoc = MapDocumentSerializer.toString(document);

		System.out.println("srialization:\n" + stringDoc);

		final MapDocument decoded = MapDocumentSerializer.decode(stringDoc.getBytes());

		final SetView<String> diff = Sets.difference(document.fieldNames(), decoded.fieldNames());

		assertTrue(diff.isEmpty());

		System.out.println("decoded:\n" + decoded);

	}

}

package eu.dnetlib.pace.model.gt;

import static org.junit.Assert.assertTrue;

import java.util.Set;

import org.junit.Test;

import com.google.common.collect.Sets;

import eu.dnetlib.pace.model.gt.Author;

import eu.dnetlib.pace.model.gt.Authors;

public class AuthorTest {

	@Test

	public void test() {

		final Set<Author> s1 = getAuthors(3);

		final Set<Author> s2 = getAuthors(3);

		final Set<Author> i = Sets.intersection(s1, s2);

		System.out.println(i);

		assertTrue(i.size() == 3);

	}

	@Test

	public void test1() {

		final Authors a1 = new Authors(a("1", "Wang, M."));

		final Authors a2 = new Authors(a("1", "Wang, M."));

		final Set<Author> i = Sets.intersection(a1, a2);

		assertTrue(i.size() == 1);

	}

	private Set<Author> getAuthors(final int n) {

		final Set<Author> s = Sets.newHashSet();

		for (int i = 0; i < n; i++) {

			s.add(a(i + "", "name" + i));

		}

		return s;

	}

	private Author a(final String id, final String fullname) {

		final Author a = new Author();

		a.setId(id);

		a.setFullname(fullname);

		return a;

	}

}

package eu.dnetlib.pace.distance;

import static org.junit.Assert.assertTrue;

import java.util.List;

import org.junit.Test;

import com.google.common.collect.Lists;

import eu.dnetlib.pace.AbstractProtoPaceTest;

import eu.dnetlib.pace.config.Config;

import eu.dnetlib.pace.model.MapDocument;

public class DetectorTest extends AbstractProtoPaceTest {

	@Test

	public void testDistanceResultSimple() {

		final Config config = getResultSimpleConf();

		final MapDocument resA = result(config, "A", "Recent results from CDF");

		final MapDocument resB = result(config, "B", "Recent results from CDF");

		final double d = new PaceDocumentDistance().between(resA, resB, config);

		System.out.println(String.format(" d ---> %s", d));

		assertTrue(d == 1.0);

	}

	@Test

	public void testDistanceResultSimpleMissingDates() {

		final Config config = getResultSimpleConf();

		final MapDocument resA = result(config, "A", "Recent results from BES");

		final MapDocument resB = result(config, "A", "Recent results from CES");

		final double d = new PaceDocumentDistance().between(resA, resB, config);

		System.out.println(String.format(" d ---> %s", d));

		assertTrue(d > 0.97);

	}

	@Test

	public void testDistanceResultInvalidDate() {

		final Config config = getResultConf();

		final MapDocument resA = result(config, "A", "title title title 6BESR", "2013-01-05");

		final MapDocument resB = result(config, "B", "title title title 6BESR", "qwerty");

		final double d = new PaceDocumentDistance().between(resA, resB, config);

		System.out.println(String.format(" d ---> %s", d));

		assertTrue(d == 1.0);

	}

	@Test

	public void testDistanceResultMissingOneDate() {

		final Config config = getResultConf();

		final MapDocument resA = result(config, "A", "title title title 6BESR", null);

		final MapDocument resB = result(config, "B", "title title title 6CLER", "2012-02");

		final double d = new PaceDocumentDistance().between(resA, resB, config);

		System.out.println(String.format(" d ---> %s", d));

		assertTrue((d > 0.9) && (d < 1.0));

	}

	@Test

	public void testDistanceResult() {

		final Config config = getResultConf();

		final MapDocument resA = result(config, "A", "title title title BES", "");

		final MapDocument resB = result(config, "B", "title title title CLEO");

		final double d = new PaceDocumentDistance().between(resA, resB, config);

		System.out.println(String.format(" d ---> %s", d));

		assertTrue((d > 0.9) && (d < 1.0));

	}

	@Test

	public void testDistanceResultMissingTwoDate() {

		final Config config = getResultConf();

		final MapDocument resA = result(config, "A", "title title title 6BESR");

		final MapDocument resB = result(config, "B", "title title title 6CLER");

		final double d = new PaceDocumentDistance().between(resA, resB, config);

		System.out.println(String.format(" d ---> %s", d));

		assertTrue((d > 0.9) && (d < 1.0));

	}

	@Test

	public void testDistanceOrganizationIgnoreMissing() {

		final Config config = getOrganizationSimpleConf();

		final MapDocument orgA = organization(config, "A", "CONSIGLIO NAZIONALE DELLE RICERCHE");

		final MapDocument orgB = organization(config, "B", "CONSIGLIO NAZIONALE DELLE RICERCHE", "CNR");

		final double d = new PaceDocumentDistance().between(orgA, orgB, config);

		System.out.println(String.format(" d ---> %s", d));

		assertTrue(d == 1.0);

	}

	@Test

	public void testDistanceResultCase1() {

		final Config config = getResultConf();

		final MapDocument resA = result(config, "A", "Search the Standard Model Higgs boson", "2003");

		final MapDocument resB = result(config, "B", "Search for the Standard Model Higgs Boson", "2003");

		final double d = new PaceDocumentDistance().between(resA, resB, config);

		System.out.println(String.format(" d ---> %s", d));

		assertTrue((d > 0.9) && (d < 1.0));

	}

	@Test

	public void testDistanceResultCaseDoiMatch1() {

		final Config config = getResultConf();

		final MapDocument resA = result(config, "A", "Search the Standard Model Higgs boson", "2003", "http://dx.doi.org/10.1594/PANGAEA.726855");

		final MapDocument resB = result(config, "B", "Search the Standard Model Higgs Boson", "2003", "10.1594/PANGAEA.726855");

		final double d = new PaceDocumentDistance().between(resA, resB, config);

		System.out.println(String.format(" d ---> %s", d));

		assertTrue("exact DOIs will produce an exact match", d == 1.0);

	}

	@Test

	public void testDistanceResultCaseDoiMatch2() {

		final Config config = getResultConf();

		final MapDocument resA = result(config, "A", "Conference proceedings on X. Appendix", "2003", "http://dx.doi.org/10.1594/PANGAEA.726855");

		final MapDocument resB = result(config, "B", "Search the Standard Model Higgs Boson", "2005", "doi:10.1594/PANGAEA.726855");

		final double d = new PaceDocumentDistance().between(resA, resB, config);

		System.out.println(String.format(" d ---> %s", d));

		assertTrue("exact DOIs will produce an exact match, regardless of different titles or publication years", d == 1.0);

	}

	@Test

	public void testDistanceResultCaseDoiMatch3() {

		final Config config = getResultConf();

		final MapDocument resA = result(config, "A", "Conference proceedings on X. Appendix", "2003", "10.1016/j.jmb.2010.12.024");

		final MapDocument resB = result(config, "B", "Conference proceedings on X. Appendix", "2003");

		final double d = new PaceDocumentDistance().between(resA, resB, config);

		System.out.println(String.format(" d ---> %s", d));

		assertTrue("a missing DOI will casue the comparsion to continue with the following conditions", d == 1.0);

	}

	@Test

	public void testDistanceResultCaseDoiMatch4() {

		final Config config = getResultConf();

		final MapDocument resA = result(config, "A", "Conference proceedings on X. Appendix", "2003", "10.1016/j.jmb.2010.12.024");

		final MapDocument resB = result(config, "B", "Conference proceedings on X. Appendix", "2005");

		final double d = new PaceDocumentDistance().between(resA, resB, config);

		System.out.println(String.format(" d ---> %s", d));

		assertTrue("a missing DOI, comparsion continues with the following conditions, different publication years will drop the score to 0", d == 0.0);

	}

	@Test

	public void testDistanceResultCaseDoiMatch5() {

		final Config config = getResultConf();

		final MapDocument resA = result(config, "A", "Search for the Standard Model Higgs Boson", "2003", "10.1016/j.jmb.2010.12.020");

		final MapDocument resB = result(config, "B", "Search the Standard Model Higgs Boson", "2003");

		final double d = new PaceDocumentDistance().between(resA, resB, config);

		System.out.println(String.format(" d ---> %s", d));

		assertTrue("a missing DOI, comparsion continues with the following conditions", (d > 0.9) && (d < 1.0));

	}

	@Test

	public void testDistanceResultCaseDoiMatch6() {

		final Config config = getResultConf();

		final MapDocument resA = result(config, "A", "Conference proceedings on X. Appendix", "2003", "10.1016/j.jmb.2010.12.024");

		final MapDocument resB = result(config, "B", "Conference proceedings on X. Appendix", "2003", "anotherDifferentDOI");

		final double d = new PaceDocumentDistance().between(resA, resB, config);

		System.out.println(String.format(" d ---> %s", d));

		assertTrue("different DOIs will drop the score to 0, regardless of the other fields", d == 0.0);

	}

	// http://dx.doi.org/10.1594/PANGAEA.726855 doi:10.1594/PANGAEA.726855

	@Test

	public void testDistanceResultCaseAuthor1() {

		final Config config = getResultAuthorsConf();

		final List<String> authorsA = Lists.newArrayList("a", "b", "c", "d");

		final List<String> authorsB = Lists.newArrayList("a", "b", "c");

		final MapDocument resA = result(config, "A", "Search the Standard Model Higgs Boson", "2003", null, authorsA);

		final MapDocument resB = result(config, "B", "Search the Standard Model Higgs Boson", "2003", null, authorsB);

		final double d = new PaceDocumentDistance().between(resA, resB, config);

		System.out.println(String.format(" d ---> %s", d));

		assertTrue(d == 0.0);

	}

	@Test

	public void testDistanceResultCaseAuthor2() {

		final Config config = getResultAuthorsConf();

		final List<String> authorsA = Lists.newArrayList("a", "b", "c");

		final List<String> authorsB = Lists.newArrayList("a", "b", "c");

		final MapDocument resA = result(config, "A", "Search the Standard Model Higgs Boson", "2003", null, authorsA);

		final MapDocument resB = result(config, "B", "Search the Standard Model Higgs Boson", "2003", null, authorsB);

		final double d = new PaceDocumentDistance().between(resA, resB, config);

		System.out.println(String.format(" d ---> %s", d));

		assertTrue(d == 1.0);

	}

	@Test

	public void testDistanceResultCaseAuthor3() {

		final Config config = getResultAuthorsConf();

		final List<String> authorsA = Lists.newArrayList("Bardi, A.", "Manghi, P.", "Artini, M.");

		final List<String> authorsB = Lists.newArrayList("Bardi Alessia", "Manghi Paolo", "Artini Michele");

		final MapDocument resA = result(config, "A", "Search the Standard Model Higgs Boson", "2003", null, authorsA);

		final MapDocument resB = result(config, "B", "Search the Standard Model Higgs Boson", "2003", null, authorsB);

		final double d = new PaceDocumentDistance().between(resA, resB, config);

		System.out.println(String.format(" d ---> %s", d));

		assertTrue((d > 0.9) && (d < 1.0));

	}

	@Test

	public void testDistanceResultCaseAuthor4() {

		final Config config = getResultAuthorsConf();

		final List<String> authorsA = Lists.newArrayList("Bardi, Alessia", "Manghi, Paolo", "Artini, Michele", "a");

		final List<String> authorsB = Lists.newArrayList("Bardi Alessia", "Manghi Paolo", "Artini Michele");

		final MapDocument resA = result(config, "A", "Search the Standard Model Higgs Boson", "2003", null, authorsA);

		final MapDocument resB = result(config, "B", "Search the Standard Model Higgs Boson", "2003", null, authorsB);

		final double d = new PaceDocumentDistance().between(resA, resB, config);

		System.out.println(String.format(" d ---> %s", d));

		// assertTrue(d == 0.0);

	}

	@Test

	public void testDistanceResultFullConf() {

		final Config config = getResultFullConf();

		final List<String> authorsA = Lists.newArrayList("Nagarajan Pranesh", "Guy Vautier", "Punyanganie de Silva");

		final List<String> authorsB = Lists.newArrayList("Pranesh Nagarajan", "Vautier Guy", "de Silva Punyanganie");

		final MapDocument resA =

				result(config, "A", "Presentations of perforated colonic pathology in patients with polymyalgia rheumatica: two case reports", "2010",

						"10.1186/1752-1947-4-299", authorsA);

		final MapDocument resB =

				result(config, "B", "Presentations of perforated colonic pathology in patients with polymyalgia rheumatica: two case reports", "2010", null,

						authorsB);

		final double d = new PaceDocumentDistance().between(resA, resB, config);

		System.out.println(String.format(" d ---> %s", d));

		// assertTrue(d == 0.0);

	}

	@Test

	public void testDistanceProdConf1() {

		final Config config = getResultProdConf();

		final MapDocument resA =

				result(config,

						"A",

						" Analysis of Transfer Embryo-Derived de-duplication");

		final MapDocument resB =

				result(config,

						"B",

						" Analysis of Transfer Embryo Derived deduplication");

		final double d = new PaceDocumentDistance().between(resA, resB, config);

		System.out.println(String.format(" d ---> %s", d));

		// assertTrue(d == 0.0);

	}

	@Test

	public void testDistanceProdConf2() {

		final Config config = getResultProdConf();

		final MapDocument resA =

				result(config,

						"A",

						"qwerty aaabbbbbbbb bbb ccc ddddd");

		final MapDocument resB =

				result(config,

						"B",

						"qwert aaabbbbbbbb bbb ccc ddddd");

		final double d = new PaceDocumentDistance().between(resA, resB, config);

		System.out.println(String.format(" d ---> %s", d));

		// assertTrue(d == 0.0);

	}

	@Test

	public void testDistancePersonConf1() {

		final Config config = getPersonConf();

		final MapDocument p1 = person(config, "p1_id", getPersonGT("/eu/dnetlib/pace/model/gt.author.manghi1.fo.json"));

		final MapDocument p2 = person(config, "p2_id", getPersonGT("/eu/dnetlib/pace/model/gt.author.manghi2.fo.json"));

		final double d = new PaceDocumentDistance().between(p1, p2, config);

		System.out.println(String.format(" d ---> %s", d));

		// assertTrue(d == 0.0);

	}

}

package eu.dnetlib.data.mapreduce.util;

import static org.junit.Assert.assertFalse;

import static org.junit.Assert.assertNotNull;

import java.util.List;

import org.junit.Test;

import eu.dnetlib.data.proto.KindProtos.Kind;

import eu.dnetlib.miscutils.functional.xml.IndentXmlString;

public class OafDecoderTest {

	@Test

	public void testAsXml() {

		final OafDecoder decoder = OafTest.embed(OafTest.getResult("50|id_1"), Kind.entity);

		assertNotNull(decoder);

		assertNotNull(decoder.asXml());

		System.out.println(IndentXmlString.apply(decoder.asXml()));

	}

	@Test

	public void testGetFieldValues() {

		final OafDecoder decoder = OafTest.embed(OafTest.getResult("50|id_1"), Kind.entity);

		final String path = "result/metadata/title/value";

		final List<String> titles = decoder.decodeEntity().getFieldValues(path);

		assertNotNull(titles);

		assertFalse(titles.isEmpty());

	}

}

package eu.dnetlib.data.mapreduce.util;

import org.junit.Test;

public class OafRowKeyDecoderTest {

	@Test

	public void test() {

		String id1 = "50|acnbad______::0a454baf9c61e63d42fb83ab549f8062";

		OafRowKeyDecoder d = OafRowKeyDecoder.decode(id1);

		System.out.println(d.getId());

	}

}

package eu.dnetlib.data.mapreduce.util;

import static org.junit.Assert.assertEquals;

import static org.junit.Assert.assertNotNull;

import org.junit.Before;

import org.junit.Test;

import com.google.protobuf.Descriptors.FieldDescriptor;

import eu.dnetlib.data.proto.OafProtos.OafRel;

import eu.dnetlib.data.proto.PersonResultProtos.PersonResult.Authorship;

import eu.dnetlib.data.proto.RelTypeProtos.RelType;

import eu.dnetlib.data.proto.RelTypeProtos.SubRelType;

public class OafRelDecoderTest {

	private OafRel oafRel;

	@Before

	public void setUp() {

		oafRel = OafTest.getPersonResult("ID_1", "ID_2", "1", "isAuthor");

	}

	@Test

	public void testSetClass() {

		OafRelDecoder d1 = OafRelDecoder.decode(oafRel);

		assertNotNull(d1);

		assertEquals("isAuthor", d1.getRelClass());

		OafRelDecoder d2 = OafRelDecoder.decode(d1.setClassId("hasAuthor").build());

		assertEquals("hasAuthor", d2.getRelClass());

		assertEquals("hasAuthor", d2.getRelMetadata().getSemantics().getClassid());

		assertEquals("hasAuthor", d2.getRelMetadata().getSemantics().getClassname());

		FieldDescriptor fd = Authorship.getDescriptor().findFieldByName("ranking");

		assertEquals(d1.getSubRel().getField(fd), d2.getSubRel().getField(fd));

	}

	@Test

	public void testGetCF() {

		assertEquals("personResult_authorship_isAuthorOf", OafRelDecoder.getCFQ(RelType.personResult, SubRelType.authorship, Authorship.RelName.isAuthorOf));

		assertEquals("personResult_authorship_isAuthorOf", OafRelDecoder.getCFQ(RelType.personResult, SubRelType.authorship, "isAuthorOf"));

	}

}

package eu.dnetlib.data.transform;

import static org.junit.Assert.assertFalse;

import static org.junit.Assert.assertNotNull;

import java.io.IOException;

import java.io.StringWriter;

import org.apache.commons.codec.binary.Base64;

import org.apache.commons.io.IOUtils;

import org.apache.commons.logging.Log;

import org.apache.commons.logging.LogFactory;

import org.apache.solr.common.SolrInputDocument;

import org.apache.solr.common.SolrInputField;

import org.dom4j.DocumentException;

import org.junit.Before;

import org.junit.Test;

import com.google.protobuf.InvalidProtocolBufferException;

import com.googlecode.protobuf.format.JsonFormat;

import eu.dnetlib.data.mapreduce.util.OafTest;

import eu.dnetlib.data.proto.KindProtos.Kind;

import eu.dnetlib.data.proto.OafProtos.Oaf;

import eu.dnetlib.data.proto.OafProtos.OafEntity;

import eu.dnetlib.functionality.index.solr.feed.InputDocumentFactory;

public class SolrProtoMapperTest {

	private static final Log log = LogFactory.getLog(SolrProtoMapper.class); // NOPMD by marko on 11/24/08 5:02 PM

	private String fields;

	@Before

	public void setUp() throws IOException {

		final StringWriter sw = new StringWriter();

		IOUtils.copy(getClass().getResourceAsStream("fields.xml"), sw);

		fields = sw.toString();

		assertNotNull(fields);

		assertFalse(fields.isEmpty());

		log.info(fields);

	}

	@Test

	public void testProto2SolrDocument() throws DocumentException, InvalidProtocolBufferException {

		final SolrProtoMapper mapper = new SolrProtoMapper(fields);

		assertNotNull(mapper);

		final OafEntity.Builder entity = OafTest.getResultBuilder("01");

		entity.addChildren(OafTest.getResultBuilder("01_children"));

		final Oaf oaf = OafTest.embed(entity.build(), Kind.entity).getOaf();

		assertNotNull(oaf.getEntity().getChildrenList());

		assertFalse(oaf.getEntity().getChildrenList().isEmpty());

		log.info("byte[] size: " + oaf.toByteArray().length);

		log.info("json size:   " + JsonFormat.printToString(oaf).length());

		log.info("base64 size: " + Base64.encodeBase64String(oaf.toByteArray()).length());

		final byte[] decodeBase64 = Base64.decodeBase64(Base64.encodeBase64String(oaf.toByteArray()));

		log.info("decoded: " + JsonFormat.printToString(Oaf.parseFrom(decodeBase64)));

		final SolrInputDocument doc = mapper.map(oaf, InputDocumentFactory.getParsedDateField("2015-02-15"), "asd", "action-set");

		assertNotNull(doc);

		for (final SolrInputField f : doc.values()) {

			log.info(f);

		}

	}

}

{ 

	"wf" : { 

        "threshold" : "0.99", 

        "run" : "001", 

        "entityType" : "organization", 

        "orderField" : "title", 

        "queueMaxSize" : "2000",

        "groupMaxSize" : "10",

        "slidingWindowSize" : "200",

        "rootBuilder" : [ "organization" ],

        "includeChildren" : "true" 

    },

	"pace" : {		

  		"conditions" : [ ],		

		"model" : [

			{ "name" : "legalname", "algo" : "JaroWinkler", "type" : "String", "weight" : "0.6", "ignoreMissing" : "false", "path" : "organization/metadata/legalname/value" },

			{ "name" : "legalshortname", "algo" : "JaroWinkler", "type" : "String", "weight" : "0.4", "ignoreMissing" : "true", "path" : "organization/metadata/legalshortname/value" }

		],

		"blacklists" : { } 		

	}

}

{"id":"30|dedup_wf_001::e78059705c3885a10440c8021afbdd4a","author":{"frequency":6,"id":"30|od______2294::9897283f935d7c2f4c11cebf65ae3098","fullname":"Manghi, Paolo","firstname":"Paolo","secondnames":"Manghi"},"merged":[{"id":"30|od______2294::9897283f935d7c2f4c11cebf65ae3098","fullname":"Manghi, Paolo","firstname":"Paolo","secondnames":"Manghi"},{"id":"30|doaj10829873::6cff63f7eafbb51fd6a1c268f4f0227f","fullname":"Manghi, Paolo","firstname":"Paolo","secondnames":"Manghi"},{"id":"30|od______2367::9897283f935d7c2f4c11cebf65ae3098","fullname":"Manghi, Paolo","firstname":"Paolo","secondnames":"Manghi"},{"id":"30|doaj10829873::f67c45f3bd4e4e4c4942a373799e0457","fullname":"Manghi, Paolo","firstname":"Paolo","secondnames":"Manghi"},{"id":"30|doaj10829873::39a29e7c15e04cdef8aecaa062b4f000","fullname":"Manghi, Paolo","firstname":"Paolo","secondnames":"Manghi"},{"id":"30|doaj10829873::f9b4de41343d956271d88f93ba68c31d","fullname":"Manghi, Paolo","firstname":"Paolo","secondnames":"Manghi"}],"coAuthors":[{"anchorId":"30|dedup_wf_001::73aac4e0ca21747def81773cdf242152","id":"30|od______2367::069aab4b3defa55d5fb573838e54ed10","fullname":"Werf, Titia","firstname":"Titia","secondnames":"Werf"},{"anchorId":"30|dedup_wf_001::cd493693f2e04159d419cfbea6042944","id":"30|doaj10829873::0acf7cdc45040bd02482b94760af9db5","fullname":"Smith, Tim","firstname":"Tim","secondnames":"Smith"},{"anchorId":"30|dedup_wf_001::4867c41ae4006a89a4e00fa5b2ace526","id":"30|od______2294::717184cb6436add31e3ce0ba86ba2476","fullname":"Katifori, Akrivi","firstname":"Akrivi","secondnames":"Katifori"},{"anchorId":"30|dedup_wf_001::d17df318b8ab8b17ac2b016767e183b4","id":"30|od______2367::d24f772460583eb514d0465f77f15030","fullname":"Rettberg, Najla","firstname":"Najla","secondnames":"Rettberg"},{"anchorId":"30|dedup_wf_001::ce0c9316f1d270fe7c021ac010378189","id":"30|od______2367::ef2914f870d83e32764f2e0db122d0c4","fullname":"Houssos, Nikos","firstname":"Nikos","secondnames":"Houssos"},{"anchorId":"30|dedup_wf_001::ce0c9316f1d270fe7c021ac010378189","id":"30|od______2294::ef2914f870d83e32764f2e0db122d0c4","fullname":"Houssos, Nikos","firstname":"Nikos","secondnames":"Houssos"},{"anchorId":"30|dedup_wf_001::271f098792b54069a4d77b4fee16b354","id":"30|od______2294::217bb36de851aacdeae92250959e5af9","fullname":"Schmidt, Birgit","firstname":"Birgit","secondnames":"Schmidt"},{"anchorId":"30|dedup_wf_001::19651c25c06f8ddbf8fb582efa51bb10","id":"30|od______2367::c48a8036130fc6f9a73348ba3be40041","fullname":"Biagini, Federico","firstname":"Federico","secondnames":"Biagini"},{"anchorId":"30|dedup_wf_001::49e6185992465c4473a19d74aaf0c774","id":"30|od______2367::44afefb40235eadaa9424f735dc569ad","fullname":"Schirrwagen, Jochen","firstname":"Jochen","secondnames":"Schirrwagen"},{"anchorId":"30|dedup_wf_001::21bff9fc229ffae537509689a3eb2bd1","id":"30|doaj10829873::269dbf736b973d298160217072f127bd","fullname":"Candela, Leonardo","firstname":"Leonardo","secondnames":"Candela"},{"anchorId":"30|dedup_wf_001::84b80c9fc38b022ae4e4a5c25ffa7999","id":"30|od______2367::a93eb6a6cbfce5bcdf86cb743f788077","fullname":"Castelli, Donatella","firstname":"Donatella","secondnames":"Castelli"},{"anchorId":"30|dedup_wf_001::731b5090d3f9c6dac58ad770b1a29e13","id":"30|doaj10829873::c6cc48d8aa310a1d2c87f5df7c5576bb","fullname":"Bardi, Alessia","firstname":"Alessia","secondnames":"Bardi"},{"anchorId":"30|dedup_wf_001::5a835806fb58a8199640d0b54de6b70a","id":"30|od______2367::9c743b25e78b06608d4214ba18f92690","fullname":"Bolikowski, Lukasz","firstname":"Lukasz","secondnames":"Bolikowski"},{"anchorId":"30|dedup_wf_001::d71c73319eda487f9b56e860e2c1794f","id":"30|od______2367::c11bdd64efcd19316ef10c1967392565","fullname":"Peters, Dale","firstname":"Dale","secondnames":"Peters"},{"anchorId":"30|dedup_wf_001::2c561a4cbd689f4729324744fb046995","id":"30|doaj10829873::80724b7b737d6c51b3733eaf9da4b605","fullname":"Mikulicic, Marko","firstname":"Marko","secondnames":"Mikulicic"},{"anchorId":"30|dedup_wf_001::cd493693f2e04159d419cfbea6042944","id":"30|od______2367::b340050a287b2601d330547261fe82bc","fullname":"Smith, Tim","firstname":"Tim","secondnames":"Smith"},{"anchorId":"30|dedup_wf_001::49e6185992465c4473a19d74aaf0c774","id":"30|od______2294::44afefb40235eadaa9424f735dc569ad","fullname":"Schirrwagen, Jochen","firstname":"Jochen","secondnames":"Schirrwagen"},{"anchorId":"30|dedup_wf_001::f41b3d7b95be894f9bd7a30b08f50313","id":"30|od______2294::45f6127ede0d24d16a5007ceff0a2d4b","fullname":"Horstmann, Wolfram","firstname":"Wolfram","secondnames":"Horstmann"},{"anchorId":"30|dedup_wf_001::75d72e914bd4d88674bd2a99a01c9f19","id":"30|od______2367::08fc98b1a1e2e2de97e3ecdd1b8174d0","fullname":"Pagano, Pasquale","firstname":"Pasquale","secondnames":"Pagano"},{"anchorId":"30|dedup_wf_001::30530ec3554a58bf9eaa0c497b9c83e5","id":"30|od______2367::c68257b7b29d6c1b372856f371e3ba24","fullname":"Manola, Natalia","firstname":"Natalia","secondnames":"Manola"},{"anchorId":"30|dedup_wf_001::21bff9fc229ffae537509689a3eb2bd1","id":"30|od______2367::8fd116733ee055e8bee96e422b4142ca","fullname":"Candela, Leonardo","firstname":"Leonardo","secondnames":"Candela"},{"anchorId":"30|dedup_wf_001::4867c41ae4006a89a4e00fa5b2ace526","id":"30|od______2367::717184cb6436add31e3ce0ba86ba2476","fullname":"Katifori, Akrivi","firstname":"Akrivi","secondnames":"Katifori"},{"anchorId":"30|dedup_wf_001::2c561a4cbd689f4729324744fb046995","id":"30|od______2367::fe0e336ba013e433486e92b8e30435c1","fullname":"Mikulicic, Marko","firstname":"Marko","secondnames":"Mikulicic"},{"anchorId":"30|dedup_wf_001::21bff9fc229ffae537509689a3eb2bd1","id":"30|od______2294::8fd116733ee055e8bee96e422b4142ca","fullname":"Candela, Leonardo","firstname":"Leonardo","secondnames":"Candela"},{"anchorId":"30|dedup_wf_001::ef5fbc9697bc7bf23f9cedc92f855466","id":"30|od______2367::7aab6e55da61a2f412b9764e3554154a","fullname":"Assante, Massimiliano","firstname":"Massimiliano","secondnames":"Assante"},{"anchorId":"30|dedup_wf_001::49e6185992465c4473a19d74aaf0c774","id":"30|doaj10829873::6bb80887219a5ea64915d84d373e3eb5","fullname":"Schirrwagen, Jochen","firstname":"Jochen","secondnames":"Schirrwagen"},{"anchorId":"30|dedup_wf_001::ca57380f88fb4920d67e024c0af0c715","id":"30|od______2294::2ca703acc4dd66064135c1afc5e976b8","fullname":"Horst, Marek","firstname":"Marek","secondnames":"Horst"},{"anchorId":"30|dedup_wf_001::2317e62318581e89976523d2553977e2","id":"30|od______2367::545b78ed508339ab728beee05f312db1","fullname":"Zoppi, Franco","firstname":"Franco","secondnames":"Zoppi"},{"anchorId":"30|dedup_wf_001::152c124f2f79b57d16500a3b8a63b93c","id":"30|od______2367::ee0556ce5974761280b8be0812a586b3","fullname":"Bolikowski, Å<U+0081>ukasz","firstname":"Å<U+0081>ukasz","secondnames":"Bolikowski"},{"anchorId":"30|dedup_wf_001::84b80c9fc38b022ae4e4a5c25ffa7999","id":"30|doaj10829873::a3c59fe161823ef7fadbd1bb216c34bb","fullname":"Castelli, Donatella","firstname":"Donatella","secondnames":"Castelli"},{"anchorId":"30|dedup_wf_001::731b5090d3f9c6dac58ad770b1a29e13","id":"30|od______2367::3fbd0b3533bd0e6089c1fe68b115f772","fullname":"Bardi, Alessia","firstname":"Alessia","secondnames":"Bardi"},{"anchorId":"30|dedup_wf_001::75d72e914bd4d88674bd2a99a01c9f19","id":"30|doaj10829873::9e7037bbf8b59564442aa12b6b424747","fullname":"Pagano, Pasquale","firstname":"Pasquale","secondnames":"Pagano"},{"anchorId":"30|dedup_wf_001::eacced7c9132e343948bf51698a1c927","id":"30|od______2367::22272e3e38161be671abad3f27c68e36","fullname":"Mikulic, Marko","firstname":"Marko","secondnames":"Mikulic"},{"anchorId":"30|dedup_wf_001::30530ec3554a58bf9eaa0c497b9c83e5","id":"30|od______2294::c68257b7b29d6c1b372856f371e3ba24","fullname":"Manola, Natalia","firstname":"Natalia","secondnames":"Manola"},{"anchorId":"30|dedup_wf_001::f41b3d7b95be894f9bd7a30b08f50313","id":"30|od______2367::45f6127ede0d24d16a5007ceff0a2d4b","fullname":"Horstmann, Wolfram","firstname":"Wolfram","secondnames":"Horstmann"},{"anchorId":"30|dedup_wf_001::ca57380f88fb4920d67e024c0af0c715","id":"30|od______2367::2ca703acc4dd66064135c1afc5e976b8","fullname":"Horst, Marek","firstname":"Marek","secondnames":"Horst"},{"anchorId":"30|dedup_wf_001::7b1fe51141418a62b56f9c3bd5bccaa7","id":"30|od______2367::2638f44b18089930e67bbd5633f9b103","fullname":"Debole, Franca","firstname":"Franca","secondnames":"Debole"},{"anchorId":"30|dedup_wf_001::d17df318b8ab8b17ac2b016767e183b4","id":"30|od______2294::d24f772460583eb514d0465f77f15030","fullname":"Rettberg, Najla","firstname":"Najla","secondnames":"Rettberg"},{"anchorId":"30|dedup_wf_001::75d72e914bd4d88674bd2a99a01c9f19","id":"30|doaj10829873::b3795485a89b87aabeff16b81f2ff24d","fullname":"Pagano, Pasquale","firstname":"Pasquale","secondnames":"Pagano"},{"anchorId":"30|dedup_wf_001::d71c73319eda487f9b56e860e2c1794f","id":"30|od______2294::c11bdd64efcd19316ef10c1967392565","fullname":"Peters, Dale","firstname":"Dale","secondnames":"Peters"},{"anchorId":"30|dedup_wf_001::9bfaee15b25e210ab56dfa6f57c6417e","id":"30|od______2367::81d934153eb100552ef4b398e243dd45","fullname":"La Bruzzo, Sandro","firstname":"Sandro","secondnames":"La Bruzzo"},{"anchorId":"30|dedup_wf_001::5a835806fb58a8199640d0b54de6b70a","id":"30|doaj10829873::2abbf092ab99aaaf0a52969141fc0070","fullname":"Bolikowski, Lukasz","firstname":"Lukasz","secondnames":"Bolikowski"},{"anchorId":"30|dedup_wf_001::cd493693f2e04159d419cfbea6042944","id":"30|od______2294::b340050a287b2601d330547261fe82bc","fullname":"Smith, Tim","firstname":"Tim","secondnames":"Smith"},{"anchorId":"30|dedup_wf_001::2d413333206599ee864738c521dbe5b3","id":"30|od______2367::39c67e010b85a6a79e54e94b2a789221","fullname":"Savino, Pasquale","firstname":"Pasquale","secondnames":"Savino"},{"anchorId":"30|dedup_wf_001::21bff9fc229ffae537509689a3eb2bd1","id":"30|doaj10829873::515b938f185f85e1a91849337310c4b7","fullname":"Candela, Leonardo","firstname":"Leonardo","secondnames":"Candela"},{"anchorId":"30|dedup_wf_001::c4d99fedf82781b5cae2225cf8049426","id":"30|od______2294::e931b1326351ef5b3bb2b89f3fc4244c","fullname":"Kobos, Mateusz","firstname":"Mateusz","secondnames":"Kobos"},{"anchorId":"30|dedup_wf_001::30530ec3554a58bf9eaa0c497b9c83e5","id":"30|doaj10829873::d4df586c008bca6df713f84070d3e1c3","fullname":"Manola, Natalia","firstname":"Natalia","secondnames":"Manola"},{"anchorId":"30|dedup_wf_001::152c124f2f79b57d16500a3b8a63b93c","id":"30|od______2294::029cfdbbd0fcb888f4efbfe1dde81721","fullname":"Bolikowski, Łukasz","firstname":"Łukasz","secondnames":"Bolikowski"},{"anchorId":"30|dedup_wf_001::7b9bd8e96778de90c911e0fcc8821bcc","id":"30|od______2367::c96e9db5145834320c37332ef96802f6","fullname":"Artini, Michele","firstname":"Michele","secondnames":"Artini"},{"anchorId":"30|dedup_wf_001::5a835806fb58a8199640d0b54de6b70a","id":"30|od______2294::9c743b25e78b06608d4214ba18f92690","fullname":"Bolikowski, Lukasz","firstname":"Lukasz","secondnames":"Bolikowski"},{"anchorId":"30|dedup_wf_001::84b80c9fc38b022ae4e4a5c25ffa7999","id":"30|doaj10829873::347ee2414e420e4aec5395146956d8c0","fullname":"Castelli, Donatella","firstname":"Donatella","secondnames":"Castelli"},{"anchorId":"30|dedup_wf_001::5850f94defb717ab4e7be63e6d4d2d12","id":"30|od______2367::de34cde81e9e864703f58d615ea0cf69","fullname":"Siebinga, Sjoerd","firstname":"Sjoerd","secondnames":"Siebinga"},{"anchorId":"30|dedup_wf_001::271f098792b54069a4d77b4fee16b354","id":"30|od______2367::217bb36de851aacdeae92250959e5af9","fullname":"Schmidt, Birgit","firstname":"Birgit","secondnames":"Schmidt"},{"anchorId":"30|dedup_wf_001::ca57ee88d3a6d5e2908b09c5aae07f97","id":"30|od______2367::ddd745ebf411904db3cd353539f6e079","fullname":"Mieldijk, Mario","firstname":"Mario","secondnames":"Mieldijk"},{"anchorId":"30|dedup_wf_001::c4d99fedf82781b5cae2225cf8049426","id":"30|od______2367::e931b1326351ef5b3bb2b89f3fc4244c","fullname":"Kobos, Mateusz","firstname":"Mateusz","secondnames":"Kobos"}],"anchor":true}

{"id":"30|dedup_wf_001::9044d2ebdd6112d84e5ff279ae3d0a26","author":{"frequency":2,"id":"30|od______1457::9897283f935d7c2f4c11cebf65ae3098","fullname":"Manghi, Paolo","firstname":"Paolo","secondnames":"Manghi"},"merged":[{"id":"30|od______1457::9897283f935d7c2f4c11cebf65ae3098","fullname":"Manghi, Paolo","firstname":"Paolo","secondnames":"Manghi"},{"score":1.0,"id":"30|od_______307::9897283f935d7c2f4c11cebf65ae3098","fullname":"Manghi, Paolo","firstname":"Paolo","secondnames":"Manghi"}],"coAuthors":[{"anchorId":"30|dedup_wf_001::179d5e120c57d760e8847dedfa7bfb9b","id":"30|od_______307::e0c601b1613eec62abb99e2591168fba","fullname":"Rodrigues, Eloy","firstname":"Eloy","secondnames":"Rodrigues"},{"anchorId":"30|dedup_wf_001::951cb66881f278b74cb366add7f0ce0d","id":"30|od_______307::ef2914f870d83e32764f2e0db122d0c4","fullname":"Houssos, Nikos","firstname":"Nikos","secondnames":"Houssos"},{"anchorId":"30|dedup_wf_001::958d36ee987d35f7d0a4f74a7ad10680","id":"30|od______1457::980daed35d7008f6f4de325f235fde65","fullname":"Príncipe, Pedro","firstname":"Pedro","secondnames":"Príncipe"},{"anchorId":"30|dedup_wf_001::4fe9f7dd4a96f983dcc8f17436238ba6","id":"30|od_______307::ae2f5173f1f1745e8521107ffd9e37ae","fullname":"Dvořák, Jan","firstname":"Jan","secondnames":"Dvořák"},{"anchorId":"30|dedup_wf_001::d09e46b26192d335be7aef7ddcbc9c0c","id":"30|od_______307::56bbbdcfbdda05ec257fabc0c7a596cc","fullname":"Elbæk, Mikael Karstensen","firstname":"Mikael Karstensen","secondnames":"Elbæk"},{"anchorId":"30|dedup_wf_001::d09e46b26192d335be7aef7ddcbc9c0c","id":"30|od______1457::56bbbdcfbdda05ec257fabc0c7a596cc","fullname":"Elbæk, Mikael Karstensen","firstname":"Mikael Karstensen","secondnames":"Elbæk"},{"anchorId":"30|dedup_wf_001::958d36ee987d35f7d0a4f74a7ad10680","id":"30|od_______307::980daed35d7008f6f4de325f235fde65","fullname":"Príncipe, Pedro","firstname":"Pedro","secondnames":"Príncipe"},{"anchorId":"30|dedup_wf_001::29611ace913f53e57c9201010b24434b","id":"30|od______1457::99d2bbd6eb21de2603f2d355875fe7b2","fullname":"Jörg, Brigitte","firstname":"Brigitte","secondnames":"Jörg"},{"anchorId":"30|dedup_wf_001::951cb66881f278b74cb366add7f0ce0d","id":"30|od______1457::ef2914f870d83e32764f2e0db122d0c4","fullname":"Houssos, Nikos","firstname":"Nikos","secondnames":"Houssos"},{"anchorId":"30|dedup_wf_001::29611ace913f53e57c9201010b24434b","id":"30|od_______307::99d2bbd6eb21de2603f2d355875fe7b2","fullname":"Jörg, Brigitte","firstname":"Brigitte","secondnames":"Jörg"},{"anchorId":"30|dedup_wf_001::179d5e120c57d760e8847dedfa7bfb9b","id":"30|od______1457::e0c601b1613eec62abb99e2591168fba","fullname":"Rodrigues, Eloy","firstname":"Eloy","secondnames":"Rodrigues"},{"anchorId":"30|dedup_wf_001::4fe9f7dd4a96f983dcc8f17436238ba6","id":"30|od______1457::ae2f5173f1f1745e8521107ffd9e37ae","fullname":"Dvořák, Jan","firstname":"Jan","secondnames":"Dvořák"}],"anchor":true}

{

  "id": "30|dedup_wf_001::e78059705c3885a10440c8021afbdd4a",

  "author": {

    "frequency": 6,

    "id": "30|od______2294::9897283f935d7c2f4c11cebf65ae3098",

    "fullname": "Manghi, Paolo",

    "firstname": "Paolo",

    "secondnames": "Manghi"

  },

  "merged": [

    {

      "id": "30|od______2294::9897283f935d7c2f4c11cebf65ae3098",

      "fullname": "Manghi, Paolo",

      "firstname": "Paolo",

      "secondnames": "Manghi"

    },

    {

      "id": "30|doaj10829873::6cff63f7eafbb51fd6a1c268f4f0227f",

      "fullname": "Manghi, Paolo",

      "firstname": "Paolo",

      "secondnames": "Manghi"

    },

    {

      "id": "30|od______2367::9897283f935d7c2f4c11cebf65ae3098",

      "fullname": "Manghi, Paolo",

      "firstname": "Paolo",

      "secondnames": "Manghi"

    },

    {

      "id": "30|doaj10829873::f67c45f3bd4e4e4c4942a373799e0457",

      "fullname": "Manghi, Paolo",

      "firstname": "Paolo",

      "secondnames": "Manghi"

    },

    {

      "id": "30|doaj10829873::39a29e7c15e04cdef8aecaa062b4f000",

      "fullname": "Manghi, Paolo",

      "firstname": "Paolo",

      "secondnames": "Manghi"

    },

    {

      "id": "30|doaj10829873::f9b4de41343d956271d88f93ba68c31d",

      "fullname": "Manghi, Paolo",

      "firstname": "Paolo",

      "secondnames": "Manghi"

    }

  ],

  "coAuthors": [

    {

      "anchorId": "30|dedup_wf_001::179d5e120c57d760e8847dedfa7bfb9b",

      "id": "30|od_______307::e0c601b1613eec62abb99e2591168fba",

      "fullname": "Rodrigues, Eloy",

      "firstname": "Eloy",

      "secondnames": "Rodrigues"

    },

    {

      "anchorId": "30|dedup_wf_001::73aac4e0ca21747def81773cdf242152",

      "id": "30|od______2367::069aab4b3defa55d5fb573838e54ed10",

      "fullname": "Werf, Titia",

      "firstname": "Titia",

      "secondnames": "Werf"

    },

    {

      "anchorId": "30|dedup_wf_001::cd493693f2e04159d419cfbea6042944",

      "id": "30|doaj10829873::0acf7cdc45040bd02482b94760af9db5",

      "fullname": "Smith, Tim",

      "firstname": "Tim",

      "secondnames": "Smith"

    },

    {

      "anchorId": "30|dedup_wf_001::4867c41ae4006a89a4e00fa5b2ace526",

      "id": "30|od______2294::717184cb6436add31e3ce0ba86ba2476",

      "fullname": "Katifori, Akrivi",

      "firstname": "Akrivi",

      "secondnames": "Katifori"

    },

    {

      "anchorId": "30|dedup_wf_001::d17df318b8ab8b17ac2b016767e183b4",

      "id": "30|od______2367::d24f772460583eb514d0465f77f15030",

      "fullname": "Rettberg, Najla",

      "firstname": "Najla",

      "secondnames": "Rettberg"

    },

    {

      "anchorId": "30|dedup_wf_001::951cb66881f278b74cb366add7f0ce0d",

      "id": "30|od______2367::ef2914f870d83e32764f2e0db122d0c4",

      "fullname": "Houssos, Nikos",

      "firstname": "Nikos",

      "secondnames": "Houssos"

    },

    {

      "anchorId": "30|dedup_wf_001::ce0c9316f1d270fe7c021ac010378189",

      "id": "30|od______2294::ef2914f870d83e32764f2e0db122d0c4",

      "fullname": "Houssos, Nikos",

      "firstname": "Nikos",

      "secondnames": "Houssos"

    },

    {

      "anchorId": "30|dedup_wf_001::271f098792b54069a4d77b4fee16b354",

      "id": "30|od______2294::217bb36de851aacdeae92250959e5af9",

      "fullname": "Schmidt, Birgit",

      "firstname": "Birgit",

      "secondnames": "Schmidt"

    },

    {

      "anchorId": "30|dedup_wf_001::19651c25c06f8ddbf8fb582efa51bb10",

      "id": "30|od______2367::c48a8036130fc6f9a73348ba3be40041",

      "fullname": "Biagini, Federico",

      "firstname": "Federico",

      "secondnames": "Biagini"

    },

    {

      "anchorId": "30|dedup_wf_001::49e6185992465c4473a19d74aaf0c774",

      "id": "30|od______2367::44afefb40235eadaa9424f735dc569ad",

      "fullname": "Schirrwagen, Jochen",

      "firstname": "Jochen",

      "secondnames": "Schirrwagen"

    },

    {

      "anchorId": "30|dedup_wf_001::21bff9fc229ffae537509689a3eb2bd1",

      "id": "30|doaj10829873::269dbf736b973d298160217072f127bd",

      "fullname": "Candela, Leonardo",

      "firstname": "Leonardo",

      "secondnames": "Candela"

    },

    {

      "anchorId": "30|dedup_wf_001::84b80c9fc38b022ae4e4a5c25ffa7999",

      "id": "30|od______2367::a93eb6a6cbfce5bcdf86cb743f788077",

      "fullname": "Castelli, Donatella",

      "firstname": "Donatella",

      "secondnames": "Castelli"

    },

    {

      "anchorId": "30|dedup_wf_001::731b5090d3f9c6dac58ad770b1a29e13",

      "id": "30|doaj10829873::c6cc48d8aa310a1d2c87f5df7c5576bb",

      "fullname": "Bardi, Alessia",

      "firstname": "Alessia",

      "secondnames": "Bardi"

    },

    {

      "anchorId": "30|dedup_wf_001::5a835806fb58a8199640d0b54de6b70a",

      "id": "30|od______2367::9c743b25e78b06608d4214ba18f92690",

      "fullname": "Bolikowski, Lukasz",

      "firstname": "Lukasz",

      "secondnames": "Bolikowski"

    },

    {

      "anchorId": "30|dedup_wf_001::d71c73319eda487f9b56e860e2c1794f",

      "id": "30|od______2367::c11bdd64efcd19316ef10c1967392565",

      "fullname": "Peters, Dale",

      "firstname": "Dale",

      "secondnames": "Peters"

    },

    {

      "anchorId": "30|dedup_wf_001::2c561a4cbd689f4729324744fb046995",

      "id": "30|doaj10829873::80724b7b737d6c51b3733eaf9da4b605",

      "fullname": "Mikulicic, Marko",

      "firstname": "Marko",

      "secondnames": "Mikulicic"

    },

    {

      "anchorId": "30|dedup_wf_001::cd493693f2e04159d419cfbea6042944",

      "id": "30|od______2367::b340050a287b2601d330547261fe82bc",

      "fullname": "Smith, Tim",

      "firstname": "Tim",

      "secondnames": "Smith"

    },

    {

      "anchorId": "30|dedup_wf_001::49e6185992465c4473a19d74aaf0c774",

      "id": "30|od______2294::44afefb40235eadaa9424f735dc569ad",

      "fullname": "Schirrwagen, Jochen",

      "firstname": "Jochen",

      "secondnames": "Schirrwagen"

    },

    {

      "anchorId": "30|dedup_wf_001::f41b3d7b95be894f9bd7a30b08f50313",

      "id": "30|od______2294::45f6127ede0d24d16a5007ceff0a2d4b",

      "fullname": "Horstmann, Wolfram",

      "firstname": "Wolfram",

      "secondnames": "Horstmann"

    },

    {

      "anchorId": "30|dedup_wf_001::75d72e914bd4d88674bd2a99a01c9f19",

      "id": "30|od______2367::08fc98b1a1e2e2de97e3ecdd1b8174d0",

      "fullname": "Pagano, Pasquale",

      "firstname": "Pasquale",

      "secondnames": "Pagano"

    },

    {

      "anchorId": "30|dedup_wf_001::30530ec3554a58bf9eaa0c497b9c83e5",

      "id": "30|od______2367::c68257b7b29d6c1b372856f371e3ba24",

      "fullname": "Manola, Natalia",

      "firstname": "Natalia",

      "secondnames": "Manola"

    },

    {

      "anchorId": "30|dedup_wf_001::21bff9fc229ffae537509689a3eb2bd1",

      "id": "30|od______2367::8fd116733ee055e8bee96e422b4142ca",

      "fullname": "Candela, Leonardo",

      "firstname": "Leonardo",

      "secondnames": "Candela"

    },

    {

      "anchorId": "30|dedup_wf_001::4867c41ae4006a89a4e00fa5b2ace526",

      "id": "30|od______2367::717184cb6436add31e3ce0ba86ba2476",

      "fullname": "Katifori, Akrivi",

      "firstname": "Akrivi",

      "secondnames": "Katifori"

    },

    {

      "anchorId": "30|dedup_wf_001::2c561a4cbd689f4729324744fb046995",

      "id": "30|od______2367::fe0e336ba013e433486e92b8e30435c1",

      "fullname": "Mikulicic, Marko",

      "firstname": "Marko",

      "secondnames": "Mikulicic"

    },

    {

      "anchorId": "30|dedup_wf_001::21bff9fc229ffae537509689a3eb2bd1",

      "id": "30|od______2294::8fd116733ee055e8bee96e422b4142ca",

      "fullname": "Candela, Leonardo",

      "firstname": "Leonardo",

      "secondnames": "Candela"

    },

    {

      "anchorId": "30|dedup_wf_001::ef5fbc9697bc7bf23f9cedc92f855466",

      "id": "30|od______2367::7aab6e55da61a2f412b9764e3554154a",

      "fullname": "Assante, Massimiliano",

      "firstname": "Massimiliano",

      "secondnames": "Assante"

    },

    {

      "anchorId": "30|dedup_wf_001::49e6185992465c4473a19d74aaf0c774",

      "id": "30|doaj10829873::6bb80887219a5ea64915d84d373e3eb5",

      "fullname": "Schirrwagen, Jochen",

      "firstname": "Jochen",

      "secondnames": "Schirrwagen"

    },

    {

      "anchorId": "30|dedup_wf_001::ca57380f88fb4920d67e024c0af0c715",

      "id": "30|od______2294::2ca703acc4dd66064135c1afc5e976b8",

      "fullname": "Horst, Marek",

      "firstname": "Marek",

      "secondnames": "Horst"

    },

    {

      "anchorId": "30|dedup_wf_001::2317e62318581e89976523d2553977e2",

      "id": "30|od______2367::545b78ed508339ab728beee05f312db1",

      "fullname": "Zoppi, Franco",

      "firstname": "Franco",

      "secondnames": "Zoppi"

    },

    {

      "anchorId": "30|dedup_wf_001::152c124f2f79b57d16500a3b8a63b93c",

      "id": "30|od______2367::ee0556ce5974761280b8be0812a586b3",

      "fullname": "Bolikowski, Å<U+0081>ukasz",

      "firstname": "Å<U+0081>ukasz",

      "secondnames": "Bolikowski"

    },

    {

      "anchorId": "30|dedup_wf_001::84b80c9fc38b022ae4e4a5c25ffa7999",

      "id": "30|doaj10829873::a3c59fe161823ef7fadbd1bb216c34bb",

      "fullname": "Castelli, Donatella",

      "firstname": "Donatella",

      "secondnames": "Castelli"

    },

    {

      "anchorId": "30|dedup_wf_001::731b5090d3f9c6dac58ad770b1a29e13",

      "id": "30|od______2367::3fbd0b3533bd0e6089c1fe68b115f772",

      "fullname": "Bardi, Alessia",

      "firstname": "Alessia",

      "secondnames": "Bardi"

    },

    {

      "anchorId": "30|dedup_wf_001::75d72e914bd4d88674bd2a99a01c9f19",

      "id": "30|doaj10829873::9e7037bbf8b59564442aa12b6b424747",

      "fullname": "Pagano, Pasquale",

      "firstname": "Pasquale",

      "secondnames": "Pagano"

    },

    {

      "anchorId": "30|dedup_wf_001::eacced7c9132e343948bf51698a1c927",

      "id": "30|od______2367::22272e3e38161be671abad3f27c68e36",

      "fullname": "Mikulic, Marko",

      "firstname": "Marko",

      "secondnames": "Mikulic"

    },

    {

      "anchorId": "30|dedup_wf_001::30530ec3554a58bf9eaa0c497b9c83e5",

      "id": "30|od______2294::c68257b7b29d6c1b372856f371e3ba24",

      "fullname": "Manola, Natalia",

      "firstname": "Natalia",

      "secondnames": "Manola"

    },

    {

      "anchorId": "30|dedup_wf_001::f41b3d7b95be894f9bd7a30b08f50313",

      "id": "30|od______2367::45f6127ede0d24d16a5007ceff0a2d4b",

      "fullname": "Horstmann, Wolfram",

      "firstname": "Wolfram",

      "secondnames": "Horstmann"

    },

    {

      "anchorId": "30|dedup_wf_001::ca57380f88fb4920d67e024c0af0c715",

      "id": "30|od______2367::2ca703acc4dd66064135c1afc5e976b8",

      "fullname": "Horst, Marek",

      "firstname": "Marek",

      "secondnames": "Horst"

    },

    {

      "anchorId": "30|dedup_wf_001::7b1fe51141418a62b56f9c3bd5bccaa7",

      "id": "30|od______2367::2638f44b18089930e67bbd5633f9b103",

      "fullname": "Debole, Franca",

      "firstname": "Franca",

      "secondnames": "Debole"

    },

    {

      "anchorId": "30|dedup_wf_001::d17df318b8ab8b17ac2b016767e183b4",

      "id": "30|od______2294::d24f772460583eb514d0465f77f15030",

      "fullname": "Rettberg, Najla",

      "firstname": "Najla",

      "secondnames": "Rettberg"

    },

    {

      "anchorId": "30|dedup_wf_001::75d72e914bd4d88674bd2a99a01c9f19",

      "id": "30|doaj10829873::b3795485a89b87aabeff16b81f2ff24d",

      "fullname": "Pagano, Pasquale",

      "firstname": "Pasquale",

      "secondnames": "Pagano"

    },

    {

      "anchorId": "30|dedup_wf_001::d71c73319eda487f9b56e860e2c1794f",

      "id": "30|od______2294::c11bdd64efcd19316ef10c1967392565",

      "fullname": "Peters, Dale",

      "firstname": "Dale",

      "secondnames": "Peters"

    },

    {

      "anchorId": "30|dedup_wf_001::9bfaee15b25e210ab56dfa6f57c6417e",

      "id": "30|od______2367::81d934153eb100552ef4b398e243dd45",

      "fullname": "La Bruzzo, Sandro",

      "firstname": "Sandro",

      "secondnames": "La Bruzzo"

    },

    {

      "anchorId": "30|dedup_wf_001::5a835806fb58a8199640d0b54de6b70a",

      "id": "30|doaj10829873::2abbf092ab99aaaf0a52969141fc0070",

      "fullname": "Bolikowski, Lukasz",

      "firstname": "Lukasz",

      "secondnames": "Bolikowski"

    },

    {

      "anchorId": "30|dedup_wf_001::cd493693f2e04159d419cfbea6042944",

      "id": "30|od______2294::b340050a287b2601d330547261fe82bc",

      "fullname": "Smith, Tim",

      "firstname": "Tim",

      "secondnames": "Smith"

    },

    {

      "anchorId": "30|dedup_wf_001::2d413333206599ee864738c521dbe5b3",

      "id": "30|od______2367::39c67e010b85a6a79e54e94b2a789221",

      "fullname": "Savino, Pasquale",

      "firstname": "Pasquale",

      "secondnames": "Savino"

    },

    {

      "anchorId": "30|dedup_wf_001::21bff9fc229ffae537509689a3eb2bd1",

      "id": "30|doaj10829873::515b938f185f85e1a91849337310c4b7",

      "fullname": "Candela, Leonardo",

      "firstname": "Leonardo",

      "secondnames": "Candela"

    },

    {

      "anchorId": "30|dedup_wf_001::c4d99fedf82781b5cae2225cf8049426",

      "id": "30|od______2294::e931b1326351ef5b3bb2b89f3fc4244c",

      "fullname": "Kobos, Mateusz",

      "firstname": "Mateusz",

      "secondnames": "Kobos"

    },

    {

      "anchorId": "30|dedup_wf_001::30530ec3554a58bf9eaa0c497b9c83e5",

      "id": "30|doaj10829873::d4df586c008bca6df713f84070d3e1c3",

      "fullname": "Manola, Natalia",

      "firstname": "Natalia",

      "secondnames": "Manola"

    },

    {

      "anchorId": "30|dedup_wf_001::152c124f2f79b57d16500a3b8a63b93c",

      "id": "30|od______2294::029cfdbbd0fcb888f4efbfe1dde81721",

      "fullname": "Bolikowski, Łukasz",

      "firstname": "Łukasz",

      "secondnames": "Bolikowski"

    },

    {

      "anchorId": "30|dedup_wf_001::7b9bd8e96778de90c911e0fcc8821bcc",

      "id": "30|od______2367::c96e9db5145834320c37332ef96802f6",

      "fullname": "Artini, Michele",

      "firstname": "Michele",

      "secondnames": "Artini"

    },

    {

      "anchorId": "30|dedup_wf_001::5a835806fb58a8199640d0b54de6b70a",

      "id": "30|od______2294::9c743b25e78b06608d4214ba18f92690",

      "fullname": "Bolikowski, Lukasz",

      "firstname": "Lukasz",

      "secondnames": "Bolikowski"

    },

    {

      "anchorId": "30|dedup_wf_001::84b80c9fc38b022ae4e4a5c25ffa7999",

      "id": "30|doaj10829873::347ee2414e420e4aec5395146956d8c0",

      "fullname": "Castelli, Donatella",

      "firstname": "Donatella",

      "secondnames": "Castelli"

    },

    {

      "anchorId": "30|dedup_wf_001::5850f94defb717ab4e7be63e6d4d2d12",

      "id": "30|od______2367::de34cde81e9e864703f58d615ea0cf69",

      "fullname": "Siebinga, Sjoerd",

      "firstname": "Sjoerd",

      "secondnames": "Siebinga"

    },

    {

      "anchorId": "30|dedup_wf_001::271f098792b54069a4d77b4fee16b354",

      "id": "30|od______2367::217bb36de851aacdeae92250959e5af9",

      "fullname": "Schmidt, Birgit",

      "firstname": "Birgit",

      "secondnames": "Schmidt"

    },

    {

      "anchorId": "30|dedup_wf_001::ca57ee88d3a6d5e2908b09c5aae07f97",

      "id": "30|od______2367::ddd745ebf411904db3cd353539f6e079",

      "fullname": "Mieldijk, Mario",

      "firstname": "Mario",

      "secondnames": "Mieldijk"

    },

    {

      "anchorId": "30|dedup_wf_001::c4d99fedf82781b5cae2225cf8049426",

      "id": "30|od______2367::e931b1326351ef5b3bb2b89f3fc4244c",

      "fullname": "Kobos, Mateusz",

      "firstname": "Mateusz",

      "secondnames": "Kobos"

    }

  ],

  "anchor": true

}

{

  "id": "30|dedup_wf_001::9044d2ebdd6112d84e5ff279ae3d0a26",

  "author": {

    "frequency": 2,

    "id": "30|od______1457::9897283f935d7c2f4c11cebf65ae3098",

    "fullname": "Manghi, Paolo",

    "firstname": "Paolo",

    "secondnames": "Manghi"

  },

  "merged": [

    {

      "id": "30|od______1457::9897283f935d7c2f4c11cebf65ae3098",

      "fullname": "Manghi, Paolo",

      "firstname": "Paolo",

      "secondnames": "Manghi"

    },

    {

      "score": 1,

      "id": "30|od_______307::9897283f935d7c2f4c11cebf65ae3098",

      "fullname": "Manghi, Paolo",

      "firstname": "Paolo",

      "secondnames": "Manghi"

    }

  ],

  "coAuthors": [

    {

      "anchorId": "30|dedup_wf_001::179d5e120c57d760e8847dedfa7bfb9b",