package eu.dnetlib.resolver; import com.googlecode.protobuf.format.JsonFormat; import eu.dnetlib.data.collector.plugins.datacite.DataciteParser; import eu.dnetlib.data.proto.dli.ScholixObjectProtos; import eu.dnetlib.data.transform.xml.AbstractDNetXsltFunctions; import eu.dnetlib.dli.DLIUtils; import eu.dnetlib.dli.resolver.*; import eu.dnetlib.dli.resolver.model.CompletionStatus; import eu.dnetlib.dli.resolver.model.DLIObjectProvenance; import eu.dnetlib.dli.resolver.model.DLIResolvedObject; import eu.dnetlib.dli.resolver.model.ObjectProvisionMode; import eu.dnetlib.dli.resolver.model.serializer.DLIResolverSerializer; import eu.dnetlib.pid.resolver.AbstractPIDResolver; import eu.dnetlib.resolver.parser.DMFResolverParser; import eu.dnetlib.resolver.parser.PMFResolverParser; import eu.dnetlib.resolver.parser.ScholixResolverParser; import org.apache.commons.io.IOUtils; import org.apache.commons.lang3.StringUtils; import org.apache.commons.logging.Log; import org.apache.commons.logging.LogFactory; import org.junit.Assert; import org.junit.Ignore; import org.junit.Test; import org.springframework.beans.factory.annotation.Autowired; import java.io.IOException; import java.io.InputStream; import java.net.URISyntaxException; import java.net.URL; import java.time.LocalDateTime; import java.util.Arrays; /** * Created by sandro on 9/12/16. */ //@RunWith(SpringJUnit4ClassRunner.class) //@ContextConfiguration(classes = { ConfigurationTestConfig.class}) @Ignore public class DLIParserTest extends DLIAbstractTest{ private static final Log log = LogFactory.getLog(DLIParserTest.class); @Autowired private DLIResolverSerializer serializer; @Test public void testDMFParser() throws IOException { final InputStream resourceAsStream = getClass().getResourceAsStream("/eu/dnetlib/dli/transform/huge.xml"); String str = IOUtils.toString(resourceAsStream); DMFResolverParser parser = new DMFResolverParser(); DLIResolvedObject object = parser.parseObject(str); Assert.assertNotNull(object); Assert.assertNotNull(object.getTitles()); Assert.assertNotNull(object.getDescription()); Assert.assertNotNull(object.getType()); System.out.println(object); // object.setTitles(Arrays.asList("TITOLO 1", "TITOLO 2")); // Assert.assertEquals(CompletionStatus.complete.toString(), object.getCompletionStatus()); // DLIResolvedObject nullObject = parser.parseObject(IOUtils.toString(getClass().getResourceAsStream("/eu/dnetlib/dli/parser/InputRecordScholix.xml"))); // Assert.assertNull(nullObject); // nullObject = parser.parseObject(IOUtils.toString(getClass().getResourceAsStream("/eu/dnetlib/dli/parser/InputPMFRecord.xml"))); // Assert.assertNull(nullObject); } @Test public void testDOI(){ final String url="https://ssl.fao.org/glisapi/v1/pgrfas?doi=5gg0w"; DLIUtils.isValidDoi(url); } @Test public void testResolvedDate() throws IOException { final InputStream resourceAsStream = getClass().getResourceAsStream("/eu/dnetlib/dli/parser/InputPMF.xml"); String str = IOUtils.toString(resourceAsStream); PMFResolverParser parser = new PMFResolverParser(); DLIResolvedObject object = parser.parseObject(str); Assert.assertNotNull(object.getResolvedDate()); } @Test public void testDataciteParser() throws Exception { final InputStream resourceAsStream = getClass().getResourceAsStream("/eu/dnetlib/dli/parser/InputRecordOpenaire.xml"); String str = IOUtils.toString(resourceAsStream); DMFResolverParser parser = new DMFResolverParser(); DLIResolvedObject object = parser.parseObject(str); System.out.println(object.getType()); object.getRelations().forEach(it -> System.out.println(it.getTargetPID())); Assert.assertTrue(object.getCompletionStatus().equals("complete")); System.out.println("object = " + object); final String replacingXML = serializer.serializeReplacingXML(str, object); log.debug(replacingXML); } @Test public void testScholixParser() throws IOException, URISyntaxException { final InputStream resourceAsStream = getClass().getResourceAsStream("/eu/dnetlib/dli/parser/ieda.xml"); String str = IOUtils.toString(resourceAsStream); ScholixResolverParser parser = new ScholixResolverParser(); DLIResolvedObject object = parser.parseObject(str); final ScholixObjectProtos.Scholix.Builder scholix = ScholixObjectProtos.Scholix.newBuilder(); scholix.addLinkproviderBuilder() .setName(object.getDatasourceProvenance().get(0).getDatasource()) .addIdentifiersBuilder() .setIdentifier(object.getDatasourceProvenance().get(0).getDatasourceId()) .setSchema("dnetIdentifier"); scholix.setRelationship(ScholixObjectProtos.ScholixRelationship.newBuilder() .setName(object.getRelations().get(0).getRelationSemantics()) .setInverse(object.getRelations().get(0).getInverseRelation()) .setSchema("datacite") .build()); final CrossRefParserJSON parserJSON = new CrossRefParserJSON(); DLIResolvedObject dli_source = parserJSON.parseRecord(IOUtils.toString(new URL("https://api.crossref.org/works/10.1002/2016JG003630").openStream())); DLIResolvedObject dli_target = DataciteResolver.parseResponse(AbstractPIDResolver.requestURL("http://search.datacite.org/api?wt=json&q=doi:10.6073/pasta/6038bd9745347610df8a51c92ca0e452")); final ScholixObjectProtos.ScholixResource source = generateResource(dli_source); final ScholixObjectProtos.ScholixResource target = generateResource(dli_target); scholix.setSource(source); scholix.setTarget(target); scholix.setPublicationDate(LocalDateTime.now().toString()); System.out.println(JsonFormat.printToString(scholix.build())); System.out.println(generateIdentifier(dli_source, dli_target)); } private String generateIdentifier(final DLIResolvedObject source, DLIResolvedObject target) { return AbstractDNetXsltFunctions.md5(String.format("%s::%s", source.getPid().toLowerCase().trim(), target.getPid().toLowerCase().trim())); } private ScholixObjectProtos.ScholixResource generateResource(DLIResolvedObject result) { final ScholixObjectProtos.ScholixResource.Builder builder = ScholixObjectProtos.ScholixResource.newBuilder(); if (result.getDatasourceProvenance() != null) result.getDatasourceProvenance().forEach( objectProvenance -> { builder.addCollectedFrom(ScholixObjectProtos.ScholixCollectedFrom.newBuilder() .setProvisionMode(((DLIObjectProvenance) objectProvenance).getProvisionMode()) .setCompletionStatus(((DLIObjectProvenance) objectProvenance).getCompletionStatus()) .setProvider(ScholixObjectProtos.ScholixEntityId.newBuilder() .setName(objectProvenance.getDatasource()) .addIdentifiers(ScholixObjectProtos.ScholixIdentifier.newBuilder().setIdentifier(objectProvenance.getDatasourceId()) .setSchema("dnetIdentifier").build()) .build())); if (StringUtils.isNotEmpty(((DLIObjectProvenance) objectProvenance).getPublisher())) { builder.addPublisher(ScholixObjectProtos.ScholixEntityId.newBuilder() .setName(((DLIObjectProvenance) objectProvenance).getPublisher()) .build()); } }); builder.addIdentifier(ScholixObjectProtos.ScholixIdentifier.newBuilder(). setIdentifier(result.getPid()) .setSchema(result.getPidType()) .build()); builder.setObjectType(result.getType().toString()); if (result.getTitles() != null && result.getTitles().size() > 0) builder.setTitle(result.getTitles().get(0)); if (result.getAuthors() != null) result.getAuthors().forEach(author -> builder.addCreator( ScholixObjectProtos.ScholixEntityId.newBuilder() .setName(author) .build())); if (StringUtils.isNotBlank(result.getDate())) { builder.setPublicationDate(result.getDate()); } String tp = null; switch (result.getType()) { case dataset: tp = "60"; break; case unknown: tp = "70"; break; case publication: tp = "50"; break; } builder.setDnetIdentifier(tp + "|dnet________::" + result.getIdentifier()); return builder.build(); } @Test public void testScholixCompleteParser() throws IOException { final InputStream resourceAsStream = getClass().getResourceAsStream("/eu/dnetlib/dli/parser/InputRecordScholix.xml"); String str = IOUtils.toString(resourceAsStream); ScholixResolverParser parser = new ScholixResolverParser(); DLIResolvedObject object = parser.parseObject(str); log.debug("object = " + object); Assert.assertNotNull(object); log.debug("object.getCompletionStatus() = " + object.getCompletionStatus()); // Assert.assertTrue(object.getCompletionStatus().equals(CompletionStatus.complete.toString())); // // Assert.assertNotNull(object.getDate()); // Assert.assertNotNull(object.getTitles()); // Assert.assertTrue(object.getTitles().size() == 1); // // Assert.assertNotNull(object.getDatasourceProvenance()); // Assert.assertTrue(object.getDatasourceProvenance().size() == 1); // Assert.assertNotNull(object.getDatasourceProvenance().get(0).getPublisher()); // Assert.assertNotNull(object.getDatasourceProvenance().get(0).getPublisherId()); // // // Assert.assertNotNull(object.getRelations()); // // Assert.assertTrue(object.getRelations().size() == 1); // // Assert.assertNotNull(object.getRelations().get(0).getTargetPID().getId()); // Assert.assertNotNull(object.getRelations().get(0).getTargetPID().getType()); // // // DLIResolvedObject nullObject = parser.parseObject(IOUtils.toString(getClass().getResourceAsStream("/eu/dnetlib/dli/parser/InputRecordDMF.xml"))); // Assert.assertNull(nullObject); // nullObject = parser.parseObject(IOUtils.toString(getClass().getResourceAsStream("/eu/dnetlib/dli/parser/InputPMFRecord.xml"))); // Assert.assertNull(nullObject); } @Test public void testPMFParser() throws Exception { final InputStream resource = getClass().getResourceAsStream("/eu/dnetlib/dli/parser/InputPMF.xml"); String xml = IOUtils.toString(resource); PMFResolverParser parser = new PMFResolverParser(); final DLIResolvedObject resolvedObject = parser.parseObject(xml); Assert.assertNotNull(resolvedObject); Assert.assertNotNull(resolvedObject.getPidType()); log.debug("resolvedObject = " + resolvedObject); Assert.assertNotNull(resolvedObject.getRelations()); resolvedObject.getRelations().forEach(it -> Assert.assertNotNull(it)); DLIResolvedObject nullObject = parser.parseObject(IOUtils.toString(getClass().getResourceAsStream("/eu/dnetlib/dli/parser/InputRecordScholix.xml"))); Assert.assertNull(nullObject); nullObject = parser.parseObject(IOUtils.toString(getClass().getResourceAsStream("/eu/dnetlib/dli/parser/InputRecordDMF.xml"))); Assert.assertNull(nullObject); } @Test public void testOpenAIREParser() throws Exception { final InputStream resource = getClass().getResourceAsStream("/eu/dnetlib/dli/parser/InputRecordOpenaire.xml"); final String record = IOUtils.toString(resource); final OpenAireParser parser = new OpenAireParser(); final DLIResolvedObject object = parser.parseObject(record); object.setPid("od_______908::ff0270cae25acd9367ef26a53ee5e4a3"); object.setPidType("openaire"); final DLIObjectProvenance provenance = new DLIObjectProvenance(); provenance.setCompletionStatus(CompletionStatus.complete.toString()); provenance.setDatasourceId("dli::r3d100010778"); provenance.setDatasource("NCBI Nucleotide"); provenance.setProvisionMode(ObjectProvisionMode.resolved.toString()); object.setDatasourceProvenance(Arrays.asList(provenance)); log.debug(object); } @Test public void testProteinParser() throws IOException { final NCBINResolver resolver = new NCBINResolver(); resolver.setNCBINParser(new NCBINParser()); final DLIResolvedObject p02768 = resolver.resolve("U36380", "protein"); log.debug(p02768.getDescription()); Assert.assertNotNull(p02768); Assert.assertNotNull(p02768.getDescription()); log.debug("p02768 = " + p02768); } }