1
0
Fork 0

general changes in the RAW graph mapping: missing collectedfrom/hostedby causes records to be skipped; factored out most of the constants in ModelConstants class (dhp-schemas)

This commit is contained in:
Claudio Atzori 2020-05-06 13:20:02 +02:00
parent fdfecc9578
commit 17860d3ab6
7 changed files with 298 additions and 264 deletions

View File

@ -6,36 +6,83 @@ import eu.dnetlib.dhp.schema.oaf.Qualifier;
public class ModelConstants {
public static final String DNET_RESULT_TYPOLOGIES = "dnet:result_typologies";
public static final String DNET_PUBLICATION_RESOURCE = "dnet:publication_resource";
public static final String DNET_ACCESS_MODES = "dnet:access_modes";
public static final String DNET_LANGUAGES = "dnet:languages";
public static final String DNET_PID_TYPES = "dnet:pid_types";
public static final String DNET_DATA_CITE_DATE = "dnet:dataCite_date";
public static final String DNET_DATA_CITE_RESOURCE = "dnet:dataCite_resource";
public static final String DNET_PROVENANCE_ACTIONS = "dnet:provenanceActions";
public static final String SYSIMPORT_CROSSWALK_REPOSITORY = "sysimport:crosswalk:repository";
public static final String SYSIMPORT_CROSSWALK_ENTITYREGISTRY = "sysimport:crosswalk:entityregistry";
public static final String USER_CLAIM = "user:claim";
public static final String DATASET_RESULTTYPE_CLASSID = "dataset";
public static final String PUBLICATION_RESULTTYPE_CLASSID = "publication";
public static final String SOFTWARE_RESULTTYPE_CLASSID = "software";
public static final String ORP_RESULTTYPE_CLASSID = "other";
public static Qualifier PUBLICATION_DEFAULT_RESULTTYPE = new Qualifier();
public static Qualifier DATASET_DEFAULT_RESULTTYPE = new Qualifier();
public static Qualifier SOFTWARE_DEFAULT_RESULTTYPE = new Qualifier();
public static Qualifier ORP_DEFAULT_RESULTTYPE = new Qualifier();
public static final String RESULT_RESULT = "resultResult";
public static final String PUBLICATION_DATASET = "publicationDataset";
public static final String IS_RELATED_TO = "isRelatedTo";
public static final String SUPPLEMENT = "supplement";
public static final String IS_SUPPLEMENT_TO = "isSupplementTo";
public static final String IS_SUPPLEMENTED_BY = "isSupplementedBy";
public static final String PART = "part";
public static final String IS_PART_OF = "IsPartOf";
public static final String HAS_PARTS = "HasParts";
public static final String RELATIONSHIP = "relationship";
static {
PUBLICATION_DEFAULT_RESULTTYPE.setClassid(PUBLICATION_RESULTTYPE_CLASSID);
PUBLICATION_DEFAULT_RESULTTYPE.setClassname(PUBLICATION_RESULTTYPE_CLASSID);
PUBLICATION_DEFAULT_RESULTTYPE.setSchemeid(DNET_RESULT_TYPOLOGIES);
PUBLICATION_DEFAULT_RESULTTYPE.setSchemename(DNET_RESULT_TYPOLOGIES);
public static final String RESULT_PROJECT = "resultProject";
public static final String OUTCOME = "outcome";
public static final String IS_PRODUCED_BY = "isProducedBy";
public static final String PRODUCES = "produces";
DATASET_DEFAULT_RESULTTYPE.setClassid(DATASET_RESULTTYPE_CLASSID);
DATASET_DEFAULT_RESULTTYPE.setClassname(DATASET_RESULTTYPE_CLASSID);
DATASET_DEFAULT_RESULTTYPE.setSchemeid(DNET_RESULT_TYPOLOGIES);
DATASET_DEFAULT_RESULTTYPE.setSchemename(DNET_RESULT_TYPOLOGIES);
public static final String DATASOURCE_ORGANIZATION = "datasourceOrganization";
public static final String PROVISION = "provision";
public static final String IS_PROVIDED_BY = "isProvidedBy";
public static final String PROVIDES = "provides";
SOFTWARE_DEFAULT_RESULTTYPE.setClassid(SOFTWARE_RESULTTYPE_CLASSID);
SOFTWARE_DEFAULT_RESULTTYPE.setClassname(SOFTWARE_RESULTTYPE_CLASSID);
SOFTWARE_DEFAULT_RESULTTYPE.setSchemeid(DNET_RESULT_TYPOLOGIES);
SOFTWARE_DEFAULT_RESULTTYPE.setSchemename(DNET_RESULT_TYPOLOGIES);
public static final String PROJECT_ORGANIZATION = "projectOrganization";
public static final String PARTICIPATION = "participation";
public static final String HAS_PARTICIPANT = "hasParticipant";
public static final String IS_PARTICIPANT = "isParticipant";
ORP_DEFAULT_RESULTTYPE.setClassid(ORP_RESULTTYPE_CLASSID);
ORP_DEFAULT_RESULTTYPE.setClassname(ORP_RESULTTYPE_CLASSID);
ORP_DEFAULT_RESULTTYPE.setSchemeid(DNET_RESULT_TYPOLOGIES);
ORP_DEFAULT_RESULTTYPE.setSchemename(DNET_RESULT_TYPOLOGIES);
public static final Qualifier PUBLICATION_DEFAULT_RESULTTYPE = qualifier(
PUBLICATION_RESULTTYPE_CLASSID, PUBLICATION_RESULTTYPE_CLASSID,
DNET_RESULT_TYPOLOGIES, DNET_RESULT_TYPOLOGIES);
public static final Qualifier DATASET_DEFAULT_RESULTTYPE = qualifier(
DATASET_RESULTTYPE_CLASSID, DATASET_RESULTTYPE_CLASSID,
DNET_RESULT_TYPOLOGIES, DNET_RESULT_TYPOLOGIES);
public static final Qualifier SOFTWARE_DEFAULT_RESULTTYPE = qualifier(
SOFTWARE_RESULTTYPE_CLASSID, SOFTWARE_RESULTTYPE_CLASSID,
DNET_RESULT_TYPOLOGIES, DNET_RESULT_TYPOLOGIES);
public static final Qualifier ORP_DEFAULT_RESULTTYPE = qualifier(
ORP_RESULTTYPE_CLASSID, ORP_RESULTTYPE_CLASSID,
DNET_RESULT_TYPOLOGIES, DNET_RESULT_TYPOLOGIES);
public static final Qualifier REPOSITORY_PROVENANCE_ACTIONS = qualifier(
SYSIMPORT_CROSSWALK_REPOSITORY, SYSIMPORT_CROSSWALK_REPOSITORY,
DNET_PROVENANCE_ACTIONS, DNET_PROVENANCE_ACTIONS);
public static final Qualifier ENTITYREGISTRY_PROVENANCE_ACTION = qualifier(
SYSIMPORT_CROSSWALK_ENTITYREGISTRY, SYSIMPORT_CROSSWALK_ENTITYREGISTRY,
DNET_PROVENANCE_ACTIONS, DNET_PROVENANCE_ACTIONS);
private static Qualifier qualifier(
final String classid,
final String classname,
final String schemeid,
final String schemename) {
final Qualifier q = new Qualifier();
q.setClassid(classid);
q.setClassname(classname);
q.setSchemeid(schemeid);
q.setSchemename(schemename);
return q;
}
}

View File

@ -10,6 +10,7 @@ import static eu.dnetlib.dhp.oa.graph.raw.common.OafMapperUtils.listFields;
import static eu.dnetlib.dhp.oa.graph.raw.common.OafMapperUtils.oaiIProvenance;
import static eu.dnetlib.dhp.oa.graph.raw.common.OafMapperUtils.qualifier;
import static eu.dnetlib.dhp.oa.graph.raw.common.OafMapperUtils.structuredProperty;
import static eu.dnetlib.dhp.schema.common.ModelConstants.*;
import java.util.ArrayList;
import java.util.Arrays;
@ -24,7 +25,6 @@ import org.dom4j.DocumentFactory;
import org.dom4j.DocumentHelper;
import org.dom4j.Node;
import eu.dnetlib.dhp.oa.graph.raw.common.MigrationConstants;
import eu.dnetlib.dhp.schema.oaf.Author;
import eu.dnetlib.dhp.schema.oaf.Context;
import eu.dnetlib.dhp.schema.oaf.DataInfo;
@ -48,6 +48,21 @@ public abstract class AbstractMdRecordToOafMapper {
protected final Map<String, String> code2name;
protected static final String DATACITE_SCHEMA_KERNEL_4 = "http://datacite.org/schema/kernel-4";
protected static final String DATACITE_SCHEMA_KERNEL_3 = "http://datacite.org/schema/kernel-3";
protected static final Map<String, String> nsContext = new HashMap<>();
static {
nsContext.put("dr", "http://www.driver-repository.eu/namespace/dr");
nsContext.put("dri", "http://www.driver-repository.eu/namespace/dri");
nsContext.put("oaf", "http://namespace.openaire.eu/oaf");
nsContext.put("oai", "http://www.openarchives.org/OAI/2.0/");
nsContext.put("prov", "http://www.openarchives.org/OAI/2.0/provenance");
nsContext.put("dc", "http://purl.org/dc/elements/1.1/");
nsContext.put("datacite", DATACITE_SCHEMA_KERNEL_3);
}
protected static final Qualifier MAIN_TITLE_QUALIFIER = qualifier(
"main title", "main title", "dnet:dataCite_title", "dnet:dataCite_title");
@ -57,31 +72,27 @@ public abstract class AbstractMdRecordToOafMapper {
public List<Oaf> processMdRecord(final String xml) {
try {
final Map<String, String> nsContext = new HashMap<>();
nsContext.put("dr", "http://www.driver-repository.eu/namespace/dr");
nsContext.put("dri", "http://www.driver-repository.eu/namespace/dri");
nsContext.put("oaf", "http://namespace.openaire.eu/oaf");
nsContext.put("oai", "http://www.openarchives.org/OAI/2.0/");
nsContext.put("prov", "http://www.openarchives.org/OAI/2.0/provenance");
nsContext.put("dc", "http://purl.org/dc/elements/1.1/");
nsContext.put("datacite", "http://datacite.org/schema/kernel-3");
DocumentFactory.getInstance().setXPathNamespaceURIs(nsContext);
final Document doc = DocumentHelper
.parseText(
xml
.replaceAll(
"http://datacite.org/schema/kernel-4", "http://datacite.org/schema/kernel-3"));
xml.replaceAll(DATACITE_SCHEMA_KERNEL_4, DATACITE_SCHEMA_KERNEL_3));
final String type = doc.valueOf("//dr:CobjCategory/@type");
final KeyValue collectedFrom = keyValue(
createOpenaireId(10, doc.valueOf("//oaf:collectedFrom/@id"), true),
doc.valueOf("//oaf:collectedFrom/@name"));
final KeyValue collectedFrom = getProvenanceDatasource(
doc, "//oaf:collectedFrom/@id", "//oaf:collectedFrom/@name");
if (collectedFrom == null) {
return null;
}
final KeyValue hostedBy = StringUtils.isBlank(doc.valueOf("//oaf:hostedBy/@id"))
? collectedFrom
: keyValue(
createOpenaireId(10, doc.valueOf("//oaf:hostedBy/@id"), true),
doc.valueOf("//oaf:hostedBy/@name"));
: getProvenanceDatasource(doc, "//oaf:hostedBy/@id", "//oaf:hostedBy/@name");
if (hostedBy == null) {
return null;
}
final DataInfo info = prepareDataInfo(doc);
final long lastUpdateTimestamp = new Date().getTime();
@ -92,6 +103,19 @@ public abstract class AbstractMdRecordToOafMapper {
}
}
private KeyValue getProvenanceDatasource(Document doc, String xpathId, String xpathName) {
final String dsId = doc.valueOf(xpathId);
final String dsName = doc.valueOf(xpathName);
if (StringUtils.isBlank(dsId) | StringUtils.isBlank(dsName)) {
return null;
}
return keyValue(
createOpenaireId(10, dsId, true),
dsName);
}
protected List<Oaf> createOafs(
final Document doc,
final String type,
@ -107,14 +131,14 @@ public abstract class AbstractMdRecordToOafMapper {
case "publication":
final Publication p = new Publication();
populateResultFields(p, doc, collectedFrom, hostedBy, info, lastUpdateTimestamp);
p.setResulttype(MigrationConstants.PUBLICATION_RESULTTYPE_QUALIFIER);
p.setResulttype(PUBLICATION_DEFAULT_RESULTTYPE);
p.setJournal(prepareJournal(doc, info));
oafs.add(p);
break;
case "dataset":
final Dataset d = new Dataset();
populateResultFields(d, doc, collectedFrom, hostedBy, info, lastUpdateTimestamp);
d.setResulttype(MigrationConstants.DATASET_RESULTTYPE_QUALIFIER);
d.setResulttype(PUBLICATION_DEFAULT_RESULTTYPE);
d.setStoragedate(prepareDatasetStorageDate(doc, info));
d.setDevice(prepareDatasetDevice(doc, info));
d.setSize(prepareDatasetSize(doc, info));
@ -127,7 +151,7 @@ public abstract class AbstractMdRecordToOafMapper {
case "software":
final Software s = new Software();
populateResultFields(s, doc, collectedFrom, hostedBy, info, lastUpdateTimestamp);
s.setResulttype(MigrationConstants.SOFTWARE_RESULTTYPE_QUALIFIER);
s.setResulttype(SOFTWARE_DEFAULT_RESULTTYPE);
s.setDocumentationUrl(prepareSoftwareDocumentationUrls(doc, info));
s.setLicense(prepareSoftwareLicenses(doc, info));
s.setCodeRepositoryUrl(prepareSoftwareCodeRepositoryUrl(doc, info));
@ -138,7 +162,7 @@ public abstract class AbstractMdRecordToOafMapper {
default:
final OtherResearchProduct o = new OtherResearchProduct();
populateResultFields(o, doc, collectedFrom, hostedBy, info, lastUpdateTimestamp);
o.setResulttype(MigrationConstants.OTHER_RESULTTYPE_QUALIFIER);
o.setResulttype(ORP_DEFAULT_RESULTTYPE);
o.setContactperson(prepareOtherResearchProductContactPersons(doc, info));
o.setContactgroup(prepareOtherResearchProductContactGroups(doc, info));
o.setTool(prepareOtherResearchProductTools(doc, info));
@ -171,33 +195,36 @@ public abstract class AbstractMdRecordToOafMapper {
if (StringUtils.isNotBlank(originalId)) {
final String projectId = createOpenaireId(40, originalId, true);
final Relation r1 = new Relation();
r1.setRelType("resultProject");
r1.setSubRelType("outcome");
r1.setRelClass("isProducedBy");
r1.setSource(docId);
r1.setTarget(projectId);
r1.setCollectedfrom(Arrays.asList(collectedFrom));
r1.setDataInfo(info);
r1.setLastupdatetimestamp(lastUpdateTimestamp);
res.add(r1);
final Relation r2 = new Relation();
r2.setRelType("resultProject");
r2.setSubRelType("outcome");
r2.setRelClass("produces");
r2.setSource(projectId);
r2.setTarget(docId);
r2.setCollectedfrom(Arrays.asList(collectedFrom));
r2.setDataInfo(info);
r2.setLastupdatetimestamp(lastUpdateTimestamp);
res.add(r2);
res
.add(
getRelation(
docId, projectId, RESULT_PROJECT, OUTCOME, IS_PRODUCED_BY, collectedFrom, info,
lastUpdateTimestamp));
res
.add(
getRelation(
projectId, docId, RESULT_PROJECT, OUTCOME, PRODUCES, collectedFrom, info,
lastUpdateTimestamp));
}
}
return res;
}
protected Relation getRelation(String source, String target, String relType, String subRelType, String relClass,
KeyValue collectedFrom, DataInfo info, long lastUpdateTimestamp) {
final Relation rel = new Relation();
rel.setRelType(relType);
rel.setSubRelType(subRelType);
rel.setRelClass(relClass);
rel.setSource(source);
rel.setTarget(target);
rel.setCollectedfrom(Arrays.asList(collectedFrom));
rel.setDataInfo(info);
rel.setLastupdatetimestamp(lastUpdateTimestamp);
return rel;
}
protected abstract List<Oaf> addOtherResultRels(
final Document doc,
final KeyValue collectedFrom,
@ -423,7 +450,7 @@ public abstract class AbstractMdRecordToOafMapper {
if (n == null) {
return dataInfo(
false, null, false, false, MigrationConstants.REPOSITORY_PROVENANCE_ACTIONS, "0.9");
false, null, false, false, REPOSITORY_PROVENANCE_ACTIONS, "0.9");
}
final String paClassId = n.valueOf("./oaf:provenanceaction/@classid");

View File

@ -95,6 +95,7 @@ public class GenerateEntitiesApplication {
.sequenceFile(sp, Text.class, Text.class)
.map(k -> new Tuple2<>(k._1().toString(), k._2().toString()))
.map(k -> convertToListOaf(k._1(), k._2(), code2name))
.filter(Objects::nonNull)
.flatMap(list -> list.iterator()));
}

View File

@ -10,6 +10,7 @@ import static eu.dnetlib.dhp.oa.graph.raw.common.OafMapperUtils.listFields;
import static eu.dnetlib.dhp.oa.graph.raw.common.OafMapperUtils.listKeyValues;
import static eu.dnetlib.dhp.oa.graph.raw.common.OafMapperUtils.qualifier;
import static eu.dnetlib.dhp.oa.graph.raw.common.OafMapperUtils.structuredProperty;
import static eu.dnetlib.dhp.schema.common.ModelConstants.*;
import java.io.Closeable;
import java.io.IOException;
@ -31,7 +32,6 @@ import org.apache.commons.logging.LogFactory;
import eu.dnetlib.dhp.application.ArgumentApplicationParser;
import eu.dnetlib.dhp.oa.graph.raw.common.AbstractMigrationApplication;
import eu.dnetlib.dhp.oa.graph.raw.common.DbClient;
import eu.dnetlib.dhp.oa.graph.raw.common.MigrationConstants;
import eu.dnetlib.dhp.schema.oaf.Context;
import eu.dnetlib.dhp.schema.oaf.DataInfo;
import eu.dnetlib.dhp.schema.oaf.Dataset;
@ -55,6 +55,9 @@ public class MigrateDbEntitiesApplication extends AbstractMigrationApplication
private static final Log log = LogFactory.getLog(MigrateDbEntitiesApplication.class);
public static final String SOURCE_TYPE = "source_type";
public static final String TARGET_TYPE = "target_type";
private final DbClient dbClient;
private final long lastUpdateTimestamp;
@ -62,7 +65,8 @@ public class MigrateDbEntitiesApplication extends AbstractMigrationApplication
public static void main(final String[] args) throws Exception {
final ArgumentApplicationParser parser = new ArgumentApplicationParser(
IOUtils
.toString(MigrateDbEntitiesApplication.class
.toString(
MigrateDbEntitiesApplication.class
.getResourceAsStream("/eu/dnetlib/dhp/oa/graph/migrate_db_entities_parameters.json")));
parser.parseArgument(args);
@ -134,7 +138,10 @@ public class MigrateDbEntitiesApplication extends AbstractMigrationApplication
ds.setId(createOpenaireId(10, rs.getString("datasourceid"), true));
ds.setOriginalId(Arrays.asList(rs.getString("datasourceid")));
ds
.setCollectedfrom(listKeyValues(createOpenaireId(10, rs.getString("collectedfromid"), true), rs.getString("collectedfromname")));
.setCollectedfrom(
listKeyValues(
createOpenaireId(10, rs.getString("collectedfromid"), true),
rs.getString("collectedfromname")));
ds.setPid(new ArrayList<>());
ds.setDateofcollection(asString(rs.getDate("dateofcollection")));
ds.setDateoftransformation(null); // Value not returned by the SQL query
@ -195,7 +202,10 @@ public class MigrateDbEntitiesApplication extends AbstractMigrationApplication
p.setId(createOpenaireId(40, rs.getString("projectid"), true));
p.setOriginalId(Arrays.asList(rs.getString("projectid")));
p
.setCollectedfrom(listKeyValues(createOpenaireId(10, rs.getString("collectedfromid"), true), rs.getString("collectedfromname")));
.setCollectedfrom(
listKeyValues(
createOpenaireId(10, rs.getString("collectedfromid"), true),
rs.getString("collectedfromname")));
p.setPid(new ArrayList<>());
p.setDateofcollection(asString(rs.getDate("dateofcollection")));
p.setDateoftransformation(asString(rs.getDate("dateoftransformation")));
@ -249,7 +259,10 @@ public class MigrateDbEntitiesApplication extends AbstractMigrationApplication
o.setId(createOpenaireId(20, rs.getString("organizationid"), true));
o.setOriginalId(Arrays.asList(rs.getString("organizationid")));
o
.setCollectedfrom(listKeyValues(createOpenaireId(10, rs.getString("collectedfromid"), true), rs.getString("collectedfromname")));
.setCollectedfrom(
listKeyValues(
createOpenaireId(10, rs.getString("collectedfromid"), true),
rs.getString("collectedfromname")));
o.setPid(new ArrayList<>());
o.setDateofcollection(asString(rs.getDate("dateofcollection")));
o.setDateoftransformation(asString(rs.getDate("dateoftransformation")));
@ -267,9 +280,11 @@ public class MigrateDbEntitiesApplication extends AbstractMigrationApplication
.setEcresearchorganization(field(Boolean.toString(rs.getBoolean("ecresearchorganization")), info));
o.setEchighereducation(field(Boolean.toString(rs.getBoolean("echighereducation")), info));
o
.setEcinternationalorganizationeurinterests(field(Boolean.toString(rs.getBoolean("ecinternationalorganizationeurinterests")), info));
.setEcinternationalorganizationeurinterests(
field(Boolean.toString(rs.getBoolean("ecinternationalorganizationeurinterests")), info));
o
.setEcinternationalorganization(field(Boolean.toString(rs.getBoolean("ecinternationalorganization")), info));
.setEcinternationalorganization(
field(Boolean.toString(rs.getBoolean("ecinternationalorganization")), info));
o.setEcenterprise(field(Boolean.toString(rs.getBoolean("ecenterprise")), info));
o.setEcsmevalidated(field(Boolean.toString(rs.getBoolean("ecsmevalidated")), info));
o.setEcnutscode(field(Boolean.toString(rs.getBoolean("ecnutscode")), info));
@ -288,12 +303,13 @@ public class MigrateDbEntitiesApplication extends AbstractMigrationApplication
final DataInfo info = prepareDataInfo(rs);
final String orgId = createOpenaireId(20, rs.getString("organization"), true);
final String dsId = createOpenaireId(10, rs.getString("datasource"), true);
final List<KeyValue> collectedFrom = listKeyValues(createOpenaireId(10, rs.getString("collectedfromid"), true), rs.getString("collectedfromname"));
final List<KeyValue> collectedFrom = listKeyValues(
createOpenaireId(10, rs.getString("collectedfromid"), true), rs.getString("collectedfromname"));
final Relation r1 = new Relation();
r1.setRelType("datasourceOrganization");
r1.setSubRelType("provision");
r1.setRelClass("isProvidedBy");
r1.setRelType(DATASOURCE_ORGANIZATION);
r1.setSubRelType(PROVISION);
r1.setRelClass(IS_PROVIDED_BY);
r1.setSource(dsId);
r1.setTarget(orgId);
r1.setCollectedfrom(collectedFrom);
@ -301,9 +317,9 @@ public class MigrateDbEntitiesApplication extends AbstractMigrationApplication
r1.setLastupdatetimestamp(lastUpdateTimestamp);
final Relation r2 = new Relation();
r2.setRelType("datasourceOrganization");
r2.setSubRelType("provision");
r2.setRelClass("provides");
r2.setRelType(DATASOURCE_ORGANIZATION);
r2.setSubRelType(PROVISION);
r2.setRelClass(PROVIDES);
r2.setSource(orgId);
r2.setTarget(dsId);
r2.setCollectedfrom(collectedFrom);
@ -321,12 +337,13 @@ public class MigrateDbEntitiesApplication extends AbstractMigrationApplication
final DataInfo info = prepareDataInfo(rs);
final String orgId = createOpenaireId(20, rs.getString("resporganization"), true);
final String projectId = createOpenaireId(40, rs.getString("project"), true);
final List<KeyValue> collectedFrom = listKeyValues(createOpenaireId(10, rs.getString("collectedfromid"), true), rs.getString("collectedfromname"));
final List<KeyValue> collectedFrom = listKeyValues(
createOpenaireId(10, rs.getString("collectedfromid"), true), rs.getString("collectedfromname"));
final Relation r1 = new Relation();
r1.setRelType("projectOrganization");
r1.setSubRelType("participation");
r1.setRelClass("hasParticipant");
r1.setRelType(PROJECT_ORGANIZATION);
r1.setSubRelType(PARTICIPATION);
r1.setRelClass(HAS_PARTICIPANT);
r1.setSource(projectId);
r1.setTarget(orgId);
r1.setCollectedfrom(collectedFrom);
@ -334,9 +351,9 @@ public class MigrateDbEntitiesApplication extends AbstractMigrationApplication
r1.setLastupdatetimestamp(lastUpdateTimestamp);
final Relation r2 = new Relation();
r2.setRelType("projectOrganization");
r2.setSubRelType("participation");
r2.setRelClass("isParticipant");
r2.setRelType(PROJECT_ORGANIZATION);
r2.setSubRelType(PARTICIPATION);
r2.setRelClass(IS_PARTICIPANT);
r2.setSource(orgId);
r2.setTarget(projectId);
r2.setCollectedfrom(collectedFrom);
@ -351,28 +368,30 @@ public class MigrateDbEntitiesApplication extends AbstractMigrationApplication
public List<Oaf> processClaims(final ResultSet rs) {
final DataInfo info =
dataInfo(false, null, false, false, qualifier("user:claim", "user:claim", "dnet:provenanceActions", "dnet:provenanceActions"), "0.9");
final DataInfo info = dataInfo(
false, null, false, false,
qualifier(USER_CLAIM, USER_CLAIM, DNET_PROVENANCE_ACTIONS, DNET_PROVENANCE_ACTIONS), "0.9");
final List<KeyValue> collectedFrom = listKeyValues(createOpenaireId(10, "infrastruct_::openaire", true), "OpenAIRE");
final List<KeyValue> collectedFrom = listKeyValues(
createOpenaireId(10, "infrastruct_::openaire", true), "OpenAIRE");
try {
if (rs.getString("source_type").equals("context")) {
if (rs.getString(SOURCE_TYPE).equals("context")) {
final Result r;
if (rs.getString("target_type").equals("dataset")) {
if (rs.getString(TARGET_TYPE).equals("dataset")) {
r = new Dataset();
r.setResulttype(MigrationConstants.DATASET_RESULTTYPE_QUALIFIER);
} else if (rs.getString("target_type").equals("software")) {
r.setResulttype(DATASET_DEFAULT_RESULTTYPE);
} else if (rs.getString(TARGET_TYPE).equals("software")) {
r = new Software();
r.setResulttype(MigrationConstants.SOFTWARE_RESULTTYPE_QUALIFIER);
} else if (rs.getString("target_type").equals("other")) {
r.setResulttype(SOFTWARE_DEFAULT_RESULTTYPE);
} else if (rs.getString(TARGET_TYPE).equals("other")) {
r = new OtherResearchProduct();
r.setResulttype(MigrationConstants.OTHER_RESULTTYPE_QUALIFIER);
r.setResulttype(ORP_DEFAULT_RESULTTYPE);
} else {
r = new Publication();
r.setResulttype(MigrationConstants.PUBLICATION_RESULTTYPE_QUALIFIER);
r.setResulttype(PUBLICATION_DEFAULT_RESULTTYPE);
}
r.setId(createOpenaireId(50, rs.getString("target_id"), false));
r.setLastupdatetimestamp(lastUpdateTimestamp);
@ -382,32 +401,32 @@ public class MigrateDbEntitiesApplication extends AbstractMigrationApplication
return Arrays.asList(r);
} else {
final String sourceId = createOpenaireId(rs.getString("source_type"), rs.getString("source_id"), false);
final String targetId = createOpenaireId(rs.getString("target_type"), rs.getString("target_id"), false);
final String sourceId = createOpenaireId(rs.getString(SOURCE_TYPE), rs.getString("source_id"), false);
final String targetId = createOpenaireId(rs.getString(TARGET_TYPE), rs.getString("target_id"), false);
final Relation r1 = new Relation();
final Relation r2 = new Relation();
if (rs.getString("source_type").equals("project")) {
if (rs.getString(SOURCE_TYPE).equals("project")) {
r1.setCollectedfrom(collectedFrom);
r1.setRelType("resultProject");
r1.setSubRelType("outcome");
r1.setRelClass("produces");
r1.setRelType(RESULT_PROJECT);
r1.setSubRelType(OUTCOME);
r1.setRelClass(PRODUCES);
r2.setCollectedfrom(collectedFrom);
r2.setRelType("resultProject");
r2.setSubRelType("outcome");
r2.setRelClass("isProducedBy");
r2.setRelType(RESULT_PROJECT);
r2.setSubRelType(OUTCOME);
r2.setRelClass(IS_PRODUCED_BY);
} else {
r1.setCollectedfrom(collectedFrom);
r1.setRelType("resultResult");
r1.setSubRelType("relationship");
r1.setRelClass("isRelatedTo");
r1.setRelType(RESULT_RESULT);
r1.setSubRelType(RELATIONSHIP);
r1.setRelClass(IS_RELATED_TO);
r2.setCollectedfrom(collectedFrom);
r2.setRelType("resultResult");
r2.setSubRelType("relationship");
r2.setRelClass("isRelatedTo");
r2.setRelType(RESULT_RESULT);
r2.setSubRelType(RELATIONSHIP);
r2.setRelClass(IS_RELATED_TO);
}
r1.setSource(sourceId);
@ -440,11 +459,14 @@ public class MigrateDbEntitiesApplication extends AbstractMigrationApplication
final String inferenceprovenance = rs.getString("inferenceprovenance");
final Boolean inferred = rs.getBoolean("inferred");
final String trust = rs.getString("trust");
return dataInfo(deletedbyinference, inferenceprovenance, inferred, false, MigrationConstants.ENTITYREGISTRY_PROVENANCE_ACTION, trust);
return dataInfo(
deletedbyinference, inferenceprovenance, inferred, false, ENTITYREGISTRY_PROVENANCE_ACTION, trust);
}
private Qualifier prepareQualifierSplitting(final String s) {
if (StringUtils.isBlank(s)) { return null; }
if (StringUtils.isBlank(s)) {
return null;
}
final String[] arr = s.split("@@@");
return arr.length == 4 ? qualifier(arr[0], arr[1], arr[2], arr[3]) : null;
}
@ -458,12 +480,16 @@ public class MigrateDbEntitiesApplication extends AbstractMigrationApplication
}
private StructuredProperty prepareStructProp(final String s, final DataInfo dataInfo) {
if (StringUtils.isBlank(s)) { return null; }
if (StringUtils.isBlank(s)) {
return null;
}
final String[] parts = s.split("###");
if (parts.length == 2) {
final String value = parts[0];
final String[] arr = parts[1].split("@@@");
if (arr.length == 4) { return structuredProperty(value, arr[0], arr[1], arr[2], arr[3], dataInfo); }
if (arr.length == 4) {
return structuredProperty(value, arr[0], arr[1], arr[2], arr[3], dataInfo);
}
}
return null;
}

View File

@ -3,6 +3,7 @@ package eu.dnetlib.dhp.oa.graph.raw;
import static eu.dnetlib.dhp.oa.graph.raw.common.OafMapperUtils.createOpenaireId;
import static eu.dnetlib.dhp.oa.graph.raw.common.OafMapperUtils.field;
import static eu.dnetlib.dhp.schema.common.ModelConstants.*;
import java.util.ArrayList;
import java.util.Arrays;
@ -13,17 +14,10 @@ import org.apache.commons.lang3.StringUtils;
import org.dom4j.Document;
import org.dom4j.Node;
import com.google.common.collect.Lists;
import eu.dnetlib.dhp.oa.graph.raw.common.PacePerson;
import eu.dnetlib.dhp.schema.oaf.Author;
import eu.dnetlib.dhp.schema.oaf.DataInfo;
import eu.dnetlib.dhp.schema.oaf.Field;
import eu.dnetlib.dhp.schema.oaf.GeoLocation;
import eu.dnetlib.dhp.schema.oaf.Instance;
import eu.dnetlib.dhp.schema.oaf.KeyValue;
import eu.dnetlib.dhp.schema.oaf.Oaf;
import eu.dnetlib.dhp.schema.oaf.Qualifier;
import eu.dnetlib.dhp.schema.oaf.Relation;
import eu.dnetlib.dhp.schema.oaf.StructuredProperty;
import eu.dnetlib.dhp.schema.oaf.*;
public class OafToOafMapper extends AbstractMdRecordToOafMapper {
@ -52,7 +46,7 @@ public class OafToOafMapper extends AbstractMdRecordToOafMapper {
@Override
protected Qualifier prepareLanguages(final Document doc) {
return prepareQualifier(doc, "//dc:language", "dnet:languages", "dnet:languages");
return prepareQualifier(doc, "//dc:language", DNET_LANGUAGES, DNET_LANGUAGES);
}
@Override
@ -96,26 +90,22 @@ public class OafToOafMapper extends AbstractMdRecordToOafMapper {
final DataInfo info,
final KeyValue collectedfrom,
final KeyValue hostedby) {
final List<Instance> res = new ArrayList<>();
for (final Object o : doc.selectNodes("//dc:identifier")) {
final String url = ((Node) o).getText().trim();
if (url.startsWith("http")) {
final Instance instance = new Instance();
instance.setUrl(Arrays.asList(url));
instance
.setInstancetype(
prepareQualifier(
doc,
"//dr:CobjCategory",
"dnet:publication_resource",
"dnet:publication_resource"));
DNET_PUBLICATION_RESOURCE,
DNET_PUBLICATION_RESOURCE));
instance.setCollectedfrom(collectedfrom);
instance.setHostedby(hostedby);
instance.setDateofacceptance(field(doc.valueOf("//oaf:dateAccepted"), info));
instance.setDistributionlocation(doc.valueOf("//oaf:distributionlocation"));
instance
.setAccessright(
prepareQualifier(doc, "//oaf:accessrights", "dnet:access_modes", "dnet:access_modes"));
prepareQualifier(doc, "//oaf:accessrights", DNET_ACCESS_MODES, DNET_ACCESS_MODES));
instance.setLicense(field(doc.valueOf("//oaf:license"), info));
instance.setRefereed(field(doc.valueOf("//oaf:refereed"), info));
instance
@ -124,10 +114,14 @@ public class OafToOafMapper extends AbstractMdRecordToOafMapper {
instance
.setProcessingchargecurrency(
field(doc.valueOf("//oaf:processingchargeamount/@currency"), info));
res.add(instance);
for (final Object o : doc.selectNodes("//dc:identifier")) {
final String url = ((Node) o).getText().trim();
if (url.startsWith("http")) {
instance.setUrl(Arrays.asList(url));
}
}
return res;
return Lists.newArrayList(instance);
}
@Override
@ -241,27 +235,16 @@ public class OafToOafMapper extends AbstractMdRecordToOafMapper {
final String otherId = createOpenaireId(50, originalId, false);
final Relation r1 = new Relation();
r1.setRelType("resultResult");
r1.setSubRelType("publicationDataset");
r1.setRelClass("isRelatedTo");
r1.setSource(docId);
r1.setTarget(otherId);
r1.setCollectedfrom(Arrays.asList(collectedFrom));
r1.setDataInfo(info);
r1.setLastupdatetimestamp(lastUpdateTimestamp);
res.add(r1);
final Relation r2 = new Relation();
r2.setRelType("resultResult");
r2.setSubRelType("publicationDataset");
r2.setRelClass("isRelatedTo");
r2.setSource(otherId);
r2.setTarget(docId);
r2.setCollectedfrom(Arrays.asList(collectedFrom));
r2.setDataInfo(info);
r2.setLastupdatetimestamp(lastUpdateTimestamp);
res.add(r2);
res
.add(
getRelation(
docId, otherId, RESULT_RESULT, PUBLICATION_DATASET, IS_RELATED_TO, collectedFrom, info,
lastUpdateTimestamp));
res
.add(
getRelation(
otherId, docId, RESULT_RESULT, PUBLICATION_DATASET, IS_RELATED_TO, collectedFrom, info,
lastUpdateTimestamp));
}
}
return res;

View File

@ -4,6 +4,7 @@ package eu.dnetlib.dhp.oa.graph.raw;
import static eu.dnetlib.dhp.oa.graph.raw.common.OafMapperUtils.createOpenaireId;
import static eu.dnetlib.dhp.oa.graph.raw.common.OafMapperUtils.field;
import static eu.dnetlib.dhp.oa.graph.raw.common.OafMapperUtils.structuredProperty;
import static eu.dnetlib.dhp.schema.common.ModelConstants.*;
import java.util.ArrayList;
import java.util.Arrays;
@ -14,6 +15,7 @@ import org.apache.commons.lang3.StringUtils;
import org.dom4j.Document;
import org.dom4j.Node;
import eu.dnetlib.dhp.schema.common.ModelConstants;
import eu.dnetlib.dhp.schema.oaf.Author;
import eu.dnetlib.dhp.schema.oaf.DataInfo;
import eu.dnetlib.dhp.schema.oaf.Field;
@ -27,6 +29,8 @@ import eu.dnetlib.dhp.schema.oaf.StructuredProperty;
public class OdfToOafMapper extends AbstractMdRecordToOafMapper {
public static final String HTTP_DX_DOI_PREIFX = "http://dx.doi.org/";
public OdfToOafMapper(final Map<String, String> code2name) {
super(code2name);
}
@ -62,7 +66,7 @@ public class OdfToOafMapper extends AbstractMdRecordToOafMapper {
structuredProperty(
((Node) o).getText(),
prepareQualifier(
(Node) o, "./@nameIdentifierScheme", "dnet:pid_types", "dnet:pid_types"),
(Node) o, "./@nameIdentifierScheme", DNET_PID_TYPES, DNET_PID_TYPES),
info));
}
return res;
@ -80,14 +84,14 @@ public class OdfToOafMapper extends AbstractMdRecordToOafMapper {
instance
.setInstancetype(
prepareQualifier(
doc, "//dr:CobjCategory", "dnet:publication_resource", "dnet:publication_resource"));
doc, "//dr:CobjCategory", DNET_PUBLICATION_RESOURCE, DNET_PUBLICATION_RESOURCE));
instance.setCollectedfrom(collectedfrom);
instance.setHostedby(hostedby);
instance.setDateofacceptance(field(doc.valueOf("//oaf:dateAccepted"), info));
instance.setDistributionlocation(doc.valueOf("//oaf:distributionlocation"));
instance
.setAccessright(
prepareQualifier(doc, "//oaf:accessrights", "dnet:access_modes", "dnet:access_modes"));
prepareQualifier(doc, "//oaf:accessrights", DNET_ACCESS_MODES, DNET_ACCESS_MODES));
instance.setLicense(field(doc.valueOf("//oaf:license"), info));
instance.setRefereed(field(doc.valueOf("//oaf:refereed"), info));
instance.setProcessingchargeamount(field(doc.valueOf("//oaf:processingchargeamount"), info));
@ -102,10 +106,10 @@ public class OdfToOafMapper extends AbstractMdRecordToOafMapper {
instance.getUrl().add(((Node) o).getText().trim());
}
for (final Object o : doc.selectNodes("//datacite:alternateIdentifier[@alternateIdentifierType='DOI']")) {
instance.getUrl().add("http://dx.doi.org/" + ((Node) o).getText().trim());
instance.getUrl().add(HTTP_DX_DOI_PREIFX + ((Node) o).getText().trim());
}
for (final Object o : doc.selectNodes("//datacite:identifier[@identifierType='DOI']")) {
instance.getUrl().add("http://dx.doi.org/" + ((Node) o).getText().trim());
instance.getUrl().add(HTTP_DX_DOI_PREIFX + ((Node) o).getText().trim());
}
return Arrays.asList(instance);
}
@ -131,8 +135,8 @@ public class OdfToOafMapper extends AbstractMdRecordToOafMapper {
((Node) o).getText(),
"UNKNOWN",
"UNKNOWN",
"dnet:dataCite_date",
"dnet:dataCite_date",
DNET_DATA_CITE_DATE,
DNET_DATA_CITE_DATE,
info));
}
}
@ -171,7 +175,7 @@ public class OdfToOafMapper extends AbstractMdRecordToOafMapper {
@Override
protected Qualifier prepareLanguages(final Document doc) {
return prepareQualifier(doc, "//datacite:language", "dnet:languages", "dnet:languages");
return prepareQualifier(doc, "//datacite:language", DNET_LANGUAGES, DNET_LANGUAGES);
}
@Override
@ -292,36 +296,29 @@ public class OdfToOafMapper extends AbstractMdRecordToOafMapper {
final String otherId = createOpenaireId(50, originalId, false);
final String type = ((Node) o).valueOf("@relationType");
if (type.equals("IsSupplementTo")) {
if (type.equalsIgnoreCase("IsSupplementTo")) {
res
.add(
prepareOtherResultRel(
collectedFrom,
info,
lastUpdateTimestamp,
docId,
otherId,
"supplement",
"isSupplementTo"));
getRelation(
docId, otherId, RESULT_RESULT, SUPPLEMENT, IS_SUPPLEMENT_TO, collectedFrom, info,
lastUpdateTimestamp));
res
.add(
prepareOtherResultRel(
collectedFrom,
info,
lastUpdateTimestamp,
otherId,
docId,
"supplement",
"isSupplementedBy"));
getRelation(
otherId, docId, RESULT_RESULT, SUPPLEMENT, IS_SUPPLEMENTED_BY, collectedFrom, info,
lastUpdateTimestamp));
} else if (type.equals("IsPartOf")) {
res
.add(
prepareOtherResultRel(
collectedFrom, info, lastUpdateTimestamp, docId, otherId, "part", "IsPartOf"));
getRelation(
docId, otherId, RESULT_RESULT, PART, IS_PART_OF, collectedFrom, info,
lastUpdateTimestamp));
res
.add(
prepareOtherResultRel(
collectedFrom, info, lastUpdateTimestamp, otherId, docId, "part", "HasParts"));
getRelation(
otherId, docId, RESULT_RESULT, PART, HAS_PARTS, collectedFrom, info,
lastUpdateTimestamp));
} else {
}
}
@ -329,32 +326,12 @@ public class OdfToOafMapper extends AbstractMdRecordToOafMapper {
return res;
}
private Relation prepareOtherResultRel(
final KeyValue collectedFrom,
final DataInfo info,
final long lastUpdateTimestamp,
final String source,
final String target,
final String subRelType,
final String relClass) {
final Relation r = new Relation();
r.setRelType("resultResult");
r.setSubRelType(subRelType);
r.setRelClass(relClass);
r.setSource(source);
r.setTarget(target);
r.setCollectedfrom(Arrays.asList(collectedFrom));
r.setDataInfo(info);
r.setLastupdatetimestamp(lastUpdateTimestamp);
return r;
}
@Override
protected Qualifier prepareResourceType(final Document doc, final DataInfo info) {
return prepareQualifier(
doc,
"//*[local-name() = 'resource']//*[local-name() = 'resourceType']",
"dnet:dataCite_resource",
"dnet:dataCite_resource");
DNET_DATA_CITE_RESOURCE,
DNET_DATA_CITE_RESOURCE);
}
}

View File

@ -1,27 +0,0 @@
package eu.dnetlib.dhp.oa.graph.raw.common;
import static eu.dnetlib.dhp.oa.graph.raw.common.OafMapperUtils.qualifier;
import eu.dnetlib.dhp.schema.oaf.Qualifier;
public class MigrationConstants {
public static final Qualifier PUBLICATION_RESULTTYPE_QUALIFIER = qualifier(
"publication", "publication", "dnet:result_typologies", "dnet:result_typologies");
public static final Qualifier DATASET_RESULTTYPE_QUALIFIER = qualifier(
"dataset", "dataset",
"dnet:result_typologies", "dnet:result_typologies");
public static final Qualifier SOFTWARE_RESULTTYPE_QUALIFIER = qualifier(
"software", "software",
"dnet:result_typologies", "dnet:result_typologies");
public static final Qualifier OTHER_RESULTTYPE_QUALIFIER = qualifier(
"other", "other",
"dnet:result_typologies", "dnet:result_typologies");
public static final Qualifier REPOSITORY_PROVENANCE_ACTIONS = qualifier(
"sysimport:crosswalk:repository", "sysimport:crosswalk:repository",
"dnet:provenanceActions", "dnet:provenanceActions");
public static final Qualifier ENTITYREGISTRY_PROVENANCE_ACTION = qualifier(
"sysimport:crosswalk:entityregistry", "sysimport:crosswalk:entityregistry",
"dnet:provenanceActions", "dnet:provenanceActions");
}