forked from D-Net/dnet-hadoop
integrated changes from master
This commit is contained in:
commit
ac25f2d8d1
|
|
@ -1,4 +1,3 @@
|
||||||
|
|
||||||
package eu.dnetlib.dhp.oa.graph.raw;
|
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.createOpenaireId;
|
||||||
|
|
@ -44,422 +43,426 @@ import org.dom4j.Node;
|
||||||
|
|
||||||
public abstract class AbstractMdRecordToOafMapper {
|
public abstract class AbstractMdRecordToOafMapper {
|
||||||
|
|
||||||
protected final Map<String, String> code2name;
|
protected final Map<String, String> code2name;
|
||||||
|
|
||||||
protected static final Qualifier MAIN_TITLE_QUALIFIER = qualifier(
|
protected static final Qualifier MAIN_TITLE_QUALIFIER =
|
||||||
"main title", "main title",
|
qualifier("main title", "main title", "dnet:dataCite_title", "dnet:dataCite_title");
|
||||||
"dnet:dataCite_title", "dnet:dataCite_title");
|
|
||||||
|
protected AbstractMdRecordToOafMapper(final Map<String, String> code2name) {
|
||||||
protected AbstractMdRecordToOafMapper(final Map<String, String> code2name) {
|
this.code2name = code2name;
|
||||||
this.code2name = code2name;
|
}
|
||||||
}
|
|
||||||
|
public List<Oaf> processMdRecord(final String xml) {
|
||||||
public List<Oaf> processMdRecord(final String xml) {
|
try {
|
||||||
try {
|
final Map<String, String> nsContext = new HashMap<>();
|
||||||
final Map<String, String> nsContext = new HashMap<>();
|
nsContext.put("dr", "http://www.driver-repository.eu/namespace/dr");
|
||||||
nsContext.put("dr", "http://www.driver-repository.eu/namespace/dr");
|
nsContext.put("dri", "http://www.driver-repository.eu/namespace/dri");
|
||||||
nsContext.put("dri", "http://www.driver-repository.eu/namespace/dri");
|
nsContext.put("oaf", "http://namespace.openaire.eu/oaf");
|
||||||
nsContext.put("oaf", "http://namespace.openaire.eu/oaf");
|
nsContext.put("oai", "http://www.openarchives.org/OAI/2.0/");
|
||||||
nsContext.put("oai", "http://www.openarchives.org/OAI/2.0/");
|
nsContext.put("prov", "http://www.openarchives.org/OAI/2.0/provenance");
|
||||||
nsContext.put("prov", "http://www.openarchives.org/OAI/2.0/provenance");
|
nsContext.put("dc", "http://purl.org/dc/elements/1.1/");
|
||||||
nsContext.put("dc", "http://purl.org/dc/elements/1.1/");
|
nsContext.put("datacite", "http://datacite.org/schema/kernel-3");
|
||||||
nsContext.put("datacite", "http://datacite.org/schema/kernel-3");
|
DocumentFactory.getInstance().setXPathNamespaceURIs(nsContext);
|
||||||
DocumentFactory.getInstance().setXPathNamespaceURIs(nsContext);
|
|
||||||
|
final Document doc =
|
||||||
final Document doc = DocumentHelper
|
DocumentHelper.parseText(
|
||||||
.parseText(
|
xml.replaceAll(
|
||||||
xml
|
"http://datacite.org/schema/kernel-4", "http://datacite.org/schema/kernel-3"));
|
||||||
.replaceAll(
|
|
||||||
"http://datacite.org/schema/kernel-4", "http://datacite.org/schema/kernel-3"));
|
final String type = doc.valueOf("//dr:CobjCategory/@type");
|
||||||
|
final KeyValue collectedFrom =
|
||||||
final String type = doc.valueOf("//dr:CobjCategory/@type");
|
keyValue(
|
||||||
final KeyValue collectedFrom = keyValue(
|
createOpenaireId(10, doc.valueOf("//oaf:collectedFrom/@id"), true),
|
||||||
createOpenaireId(10, doc.valueOf("//oaf:collectedFrom/@id"), true),
|
doc.valueOf("//oaf:collectedFrom/@name"));
|
||||||
doc.valueOf("//oaf:collectedFrom/@name"));
|
final KeyValue hostedBy =
|
||||||
final KeyValue hostedBy = StringUtils.isBlank(doc.valueOf("//oaf:hostedBy/@id"))
|
StringUtils.isBlank(doc.valueOf("//oaf:hostedBy/@id"))
|
||||||
? collectedFrom
|
? collectedFrom
|
||||||
: keyValue(
|
: keyValue(
|
||||||
createOpenaireId(10, doc.valueOf("//oaf:hostedBy/@id"), true),
|
createOpenaireId(10, doc.valueOf("//oaf:hostedBy/@id"), true),
|
||||||
doc.valueOf("//oaf:hostedBy/@name"));
|
doc.valueOf("//oaf:hostedBy/@name"));
|
||||||
|
|
||||||
final DataInfo info = prepareDataInfo(doc);
|
final DataInfo info = prepareDataInfo(doc);
|
||||||
final long lastUpdateTimestamp = new Date().getTime();
|
final long lastUpdateTimestamp = new Date().getTime();
|
||||||
|
|
||||||
return createOafs(doc, type, collectedFrom, hostedBy, info, lastUpdateTimestamp);
|
return createOafs(doc, type, collectedFrom, hostedBy, info, lastUpdateTimestamp);
|
||||||
} catch (final Exception e) {
|
} catch (final Exception e) {
|
||||||
throw new RuntimeException(e);
|
throw new RuntimeException(e);
|
||||||
}
|
}
|
||||||
}
|
}
|
||||||
|
|
||||||
protected List<Oaf> createOafs(
|
protected List<Oaf> createOafs(
|
||||||
final Document doc,
|
final Document doc,
|
||||||
final String type,
|
final String type,
|
||||||
final KeyValue collectedFrom,
|
final KeyValue collectedFrom,
|
||||||
final KeyValue hostedBy,
|
final KeyValue hostedBy,
|
||||||
final DataInfo info,
|
final DataInfo info,
|
||||||
final long lastUpdateTimestamp) {
|
final long lastUpdateTimestamp) {
|
||||||
|
|
||||||
final List<Oaf> oafs = new ArrayList<>();
|
final List<Oaf> oafs = new ArrayList<>();
|
||||||
|
|
||||||
switch (type.toLowerCase()) {
|
switch (type.toLowerCase()) {
|
||||||
case "":
|
case "":
|
||||||
case "publication":
|
case "publication":
|
||||||
final Publication p = new Publication();
|
final Publication p = new Publication();
|
||||||
populateResultFields(p, doc, collectedFrom, hostedBy, info, lastUpdateTimestamp);
|
populateResultFields(p, doc, collectedFrom, hostedBy, info, lastUpdateTimestamp);
|
||||||
p.setResulttype(MigrationConstants.PUBLICATION_RESULTTYPE_QUALIFIER);
|
p.setResulttype(MigrationConstants.PUBLICATION_RESULTTYPE_QUALIFIER);
|
||||||
p.setJournal(prepareJournal(doc, info));
|
p.setJournal(prepareJournal(doc, info));
|
||||||
oafs.add(p);
|
oafs.add(p);
|
||||||
break;
|
break;
|
||||||
case "dataset":
|
case "dataset":
|
||||||
final Dataset d = new Dataset();
|
final Dataset d = new Dataset();
|
||||||
populateResultFields(d, doc, collectedFrom, hostedBy, info, lastUpdateTimestamp);
|
populateResultFields(d, doc, collectedFrom, hostedBy, info, lastUpdateTimestamp);
|
||||||
d.setResulttype(MigrationConstants.DATASET_RESULTTYPE_QUALIFIER);
|
d.setResulttype(MigrationConstants.DATASET_RESULTTYPE_QUALIFIER);
|
||||||
d.setStoragedate(prepareDatasetStorageDate(doc, info));
|
d.setStoragedate(prepareDatasetStorageDate(doc, info));
|
||||||
d.setDevice(prepareDatasetDevice(doc, info));
|
d.setDevice(prepareDatasetDevice(doc, info));
|
||||||
d.setSize(prepareDatasetSize(doc, info));
|
d.setSize(prepareDatasetSize(doc, info));
|
||||||
d.setVersion(prepareDatasetVersion(doc, info));
|
d.setVersion(prepareDatasetVersion(doc, info));
|
||||||
d.setLastmetadataupdate(prepareDatasetLastMetadataUpdate(doc, info));
|
d.setLastmetadataupdate(prepareDatasetLastMetadataUpdate(doc, info));
|
||||||
d.setMetadataversionnumber(prepareDatasetMetadataVersionNumber(doc, info));
|
d.setMetadataversionnumber(prepareDatasetMetadataVersionNumber(doc, info));
|
||||||
d.setGeolocation(prepareDatasetGeoLocations(doc, info));
|
d.setGeolocation(prepareDatasetGeoLocations(doc, info));
|
||||||
oafs.add(d);
|
oafs.add(d);
|
||||||
break;
|
break;
|
||||||
case "software":
|
case "software":
|
||||||
final Software s = new Software();
|
final Software s = new Software();
|
||||||
populateResultFields(s, doc, collectedFrom, hostedBy, info, lastUpdateTimestamp);
|
populateResultFields(s, doc, collectedFrom, hostedBy, info, lastUpdateTimestamp);
|
||||||
s.setResulttype(MigrationConstants.SOFTWARE_RESULTTYPE_QUALIFIER);
|
s.setResulttype(MigrationConstants.SOFTWARE_RESULTTYPE_QUALIFIER);
|
||||||
s.setDocumentationUrl(prepareSoftwareDocumentationUrls(doc, info));
|
s.setDocumentationUrl(prepareSoftwareDocumentationUrls(doc, info));
|
||||||
s.setLicense(prepareSoftwareLicenses(doc, info));
|
s.setLicense(prepareSoftwareLicenses(doc, info));
|
||||||
s.setCodeRepositoryUrl(prepareSoftwareCodeRepositoryUrl(doc, info));
|
s.setCodeRepositoryUrl(prepareSoftwareCodeRepositoryUrl(doc, info));
|
||||||
s.setProgrammingLanguage(prepareSoftwareProgrammingLanguage(doc, info));
|
s.setProgrammingLanguage(prepareSoftwareProgrammingLanguage(doc, info));
|
||||||
oafs.add(s);
|
oafs.add(s);
|
||||||
break;
|
break;
|
||||||
case "otherresearchproducts":
|
case "otherresearchproducts":
|
||||||
default:
|
default:
|
||||||
final OtherResearchProduct o = new OtherResearchProduct();
|
final OtherResearchProduct o = new OtherResearchProduct();
|
||||||
populateResultFields(o, doc, collectedFrom, hostedBy, info, lastUpdateTimestamp);
|
populateResultFields(o, doc, collectedFrom, hostedBy, info, lastUpdateTimestamp);
|
||||||
o.setResulttype(MigrationConstants.OTHER_RESULTTYPE_QUALIFIER);
|
o.setResulttype(MigrationConstants.OTHER_RESULTTYPE_QUALIFIER);
|
||||||
o.setContactperson(prepareOtherResearchProductContactPersons(doc, info));
|
o.setContactperson(prepareOtherResearchProductContactPersons(doc, info));
|
||||||
o.setContactgroup(prepareOtherResearchProductContactGroups(doc, info));
|
o.setContactgroup(prepareOtherResearchProductContactGroups(doc, info));
|
||||||
o.setTool(prepareOtherResearchProductTools(doc, info));
|
o.setTool(prepareOtherResearchProductTools(doc, info));
|
||||||
oafs.add(o);
|
oafs.add(o);
|
||||||
break;
|
break;
|
||||||
}
|
}
|
||||||
|
|
||||||
if (!oafs.isEmpty()) {
|
if (!oafs.isEmpty()) {
|
||||||
oafs.addAll(addProjectRels(doc, collectedFrom, info, lastUpdateTimestamp));
|
oafs.addAll(addProjectRels(doc, collectedFrom, info, lastUpdateTimestamp));
|
||||||
oafs.addAll(addOtherResultRels(doc, collectedFrom, info, lastUpdateTimestamp));
|
oafs.addAll(addOtherResultRels(doc, collectedFrom, info, lastUpdateTimestamp));
|
||||||
}
|
}
|
||||||
|
|
||||||
return oafs;
|
return oafs;
|
||||||
}
|
}
|
||||||
|
|
||||||
private List<Oaf> addProjectRels(
|
private List<Oaf> addProjectRels(
|
||||||
final Document doc,
|
final Document doc,
|
||||||
final KeyValue collectedFrom,
|
final KeyValue collectedFrom,
|
||||||
final DataInfo info,
|
final DataInfo info,
|
||||||
final long lastUpdateTimestamp) {
|
final long lastUpdateTimestamp) {
|
||||||
|
|
||||||
final List<Oaf> res = new ArrayList<>();
|
final List<Oaf> res = new ArrayList<>();
|
||||||
|
|
||||||
final String docId = createOpenaireId(50, doc.valueOf("//dri:objIdentifier"), false);
|
final String docId = createOpenaireId(50, doc.valueOf("//dri:objIdentifier"), false);
|
||||||
|
|
||||||
for (final Object o : doc.selectNodes("//oaf:projectid")) {
|
for (final Object o : doc.selectNodes("//oaf:projectid")) {
|
||||||
final String projectId = createOpenaireId(40, ((Node) o).getText(), true);
|
|
||||||
|
final String originalId = ((Node) o).getText();
|
||||||
final Relation r1 = new Relation();
|
|
||||||
r1.setRelType("resultProject");
|
if (StringUtils.isNotBlank(originalId)) {
|
||||||
r1.setSubRelType("outcome");
|
final String projectId = createOpenaireId(40, originalId, true);
|
||||||
r1.setRelClass("isProducedBy");
|
|
||||||
r1.setSource(docId);
|
final Relation r1 = new Relation();
|
||||||
r1.setTarget(projectId);
|
r1.setRelType("resultProject");
|
||||||
r1.setCollectedfrom(Arrays.asList(collectedFrom));
|
r1.setSubRelType("outcome");
|
||||||
r1.setDataInfo(info);
|
r1.setRelClass("isProducedBy");
|
||||||
r1.setLastupdatetimestamp(lastUpdateTimestamp);
|
r1.setSource(docId);
|
||||||
res.add(r1);
|
r1.setTarget(projectId);
|
||||||
|
r1.setCollectedfrom(Arrays.asList(collectedFrom));
|
||||||
final Relation r2 = new Relation();
|
r1.setDataInfo(info);
|
||||||
r2.setRelType("resultProject");
|
r1.setLastupdatetimestamp(lastUpdateTimestamp);
|
||||||
r2.setSubRelType("outcome");
|
res.add(r1);
|
||||||
r2.setRelClass("produces");
|
|
||||||
r2.setSource(projectId);
|
final Relation r2 = new Relation();
|
||||||
r2.setTarget(docId);
|
r2.setRelType("resultProject");
|
||||||
r2.setCollectedfrom(Arrays.asList(collectedFrom));
|
r2.setSubRelType("outcome");
|
||||||
r2.setDataInfo(info);
|
r2.setRelClass("produces");
|
||||||
r2.setLastupdatetimestamp(lastUpdateTimestamp);
|
r2.setSource(projectId);
|
||||||
res.add(r2);
|
r2.setTarget(docId);
|
||||||
}
|
r2.setCollectedfrom(Arrays.asList(collectedFrom));
|
||||||
|
r2.setDataInfo(info);
|
||||||
return res;
|
r2.setLastupdatetimestamp(lastUpdateTimestamp);
|
||||||
}
|
res.add(r2);
|
||||||
|
}
|
||||||
protected abstract List<Oaf> addOtherResultRels(
|
}
|
||||||
final Document doc,
|
|
||||||
final KeyValue collectedFrom,
|
return res;
|
||||||
final DataInfo info,
|
}
|
||||||
final long lastUpdateTimestamp);
|
|
||||||
|
protected abstract List<Oaf> addOtherResultRels(
|
||||||
private void populateResultFields(
|
final Document doc,
|
||||||
final Result r,
|
final KeyValue collectedFrom,
|
||||||
final Document doc,
|
final DataInfo info,
|
||||||
final KeyValue collectedFrom,
|
final long lastUpdateTimestamp);
|
||||||
final KeyValue hostedBy,
|
|
||||||
final DataInfo info,
|
private void populateResultFields(
|
||||||
final long lastUpdateTimestamp) {
|
final Result r,
|
||||||
r.setDataInfo(info);
|
final Document doc,
|
||||||
r.setLastupdatetimestamp(lastUpdateTimestamp);
|
final KeyValue collectedFrom,
|
||||||
r.setId(createOpenaireId(50, doc.valueOf("//dri:objIdentifier"), false));
|
final KeyValue hostedBy,
|
||||||
r.setOriginalId(Arrays.asList(doc.valueOf("//dri:objIdentifier")));
|
final DataInfo info,
|
||||||
r.setCollectedfrom(Arrays.asList(collectedFrom));
|
final long lastUpdateTimestamp) {
|
||||||
r
|
r.setDataInfo(info);
|
||||||
.setPid(
|
r.setLastupdatetimestamp(lastUpdateTimestamp);
|
||||||
prepareListStructProps(
|
r.setId(createOpenaireId(50, doc.valueOf("//dri:objIdentifier"), false));
|
||||||
doc, "//oaf:identifier", "@identifierType", "dnet:pid_types", "dnet:pid_types", info));
|
r.setOriginalId(Arrays.asList(doc.valueOf("//dri:objIdentifier")));
|
||||||
r.setDateofcollection(doc.valueOf("//dr:dateOfCollection"));
|
r.setCollectedfrom(Arrays.asList(collectedFrom));
|
||||||
r.setDateoftransformation(doc.valueOf("//dr:dateOfTransformation"));
|
r.setPid(
|
||||||
r.setExtraInfo(new ArrayList<>()); // NOT PRESENT IN MDSTORES
|
prepareListStructProps(
|
||||||
r.setOaiprovenance(prepareOAIprovenance(doc));
|
doc, "//oaf:identifier", "@identifierType", "dnet:pid_types", "dnet:pid_types", info));
|
||||||
r.setAuthor(prepareAuthors(doc, info));
|
r.setDateofcollection(doc.valueOf("//dr:dateOfCollection"));
|
||||||
r.setLanguage(prepareLanguages(doc));
|
r.setDateoftransformation(doc.valueOf("//dr:dateOfTransformation"));
|
||||||
r.setCountry(new ArrayList<>()); // NOT PRESENT IN MDSTORES
|
r.setExtraInfo(new ArrayList<>()); // NOT PRESENT IN MDSTORES
|
||||||
r.setSubject(prepareSubjects(doc, info));
|
r.setOaiprovenance(prepareOAIprovenance(doc));
|
||||||
r.setTitle(prepareTitles(doc, info));
|
r.setAuthor(prepareAuthors(doc, info));
|
||||||
r.setRelevantdate(prepareRelevantDates(doc, info));
|
r.setLanguage(prepareLanguages(doc));
|
||||||
r.setDescription(prepareDescriptions(doc, info));
|
r.setCountry(new ArrayList<>()); // NOT PRESENT IN MDSTORES
|
||||||
r.setDateofacceptance(prepareField(doc, "//oaf:dateAccepted", info));
|
r.setSubject(prepareSubjects(doc, info));
|
||||||
r.setPublisher(preparePublisher(doc, info));
|
r.setTitle(prepareTitles(doc, info));
|
||||||
r.setEmbargoenddate(prepareField(doc, "//oaf:embargoenddate", info));
|
r.setRelevantdate(prepareRelevantDates(doc, info));
|
||||||
r.setSource(prepareSources(doc, info));
|
r.setDescription(prepareDescriptions(doc, info));
|
||||||
r.setFulltext(new ArrayList<>()); // NOT PRESENT IN MDSTORES
|
r.setDateofacceptance(prepareField(doc, "//oaf:dateAccepted", info));
|
||||||
r.setFormat(prepareFormats(doc, info));
|
r.setPublisher(preparePublisher(doc, info));
|
||||||
r.setContributor(prepareContributors(doc, info));
|
r.setEmbargoenddate(prepareField(doc, "//oaf:embargoenddate", info));
|
||||||
r.setResourcetype(prepareResourceType(doc, info));
|
r.setSource(prepareSources(doc, info));
|
||||||
r.setCoverage(prepareCoverages(doc, info));
|
r.setFulltext(new ArrayList<>()); // NOT PRESENT IN MDSTORES
|
||||||
r.setContext(prepareContexts(doc, info));
|
r.setFormat(prepareFormats(doc, info));
|
||||||
r.setExternalReference(new ArrayList<>()); // NOT PRESENT IN MDSTORES
|
r.setContributor(prepareContributors(doc, info));
|
||||||
r.setInstance(prepareInstances(doc, info, collectedFrom, hostedBy));
|
r.setResourcetype(prepareResourceType(doc, info));
|
||||||
}
|
r.setCoverage(prepareCoverages(doc, info));
|
||||||
|
r.setContext(prepareContexts(doc, info));
|
||||||
private List<Context> prepareContexts(final Document doc, final DataInfo info) {
|
r.setExternalReference(new ArrayList<>()); // NOT PRESENT IN MDSTORES
|
||||||
final List<Context> list = new ArrayList<>();
|
r.setInstance(prepareInstances(doc, info, collectedFrom, hostedBy));
|
||||||
for (final Object o : doc.selectNodes("//oaf:concept")) {
|
}
|
||||||
final String cid = ((Node) o).valueOf("@id");
|
|
||||||
if (StringUtils.isNotBlank(cid)) {
|
private List<Context> prepareContexts(final Document doc, final DataInfo info) {
|
||||||
final Context c = new Context();
|
final List<Context> list = new ArrayList<>();
|
||||||
c.setId(cid);
|
for (final Object o : doc.selectNodes("//oaf:concept")) {
|
||||||
c.setDataInfo(Arrays.asList(info));
|
final String cid = ((Node) o).valueOf("@id");
|
||||||
list.add(c);
|
if (StringUtils.isNotBlank(cid)) {
|
||||||
}
|
final Context c = new Context();
|
||||||
}
|
c.setId(cid);
|
||||||
return list;
|
c.setDataInfo(Arrays.asList(info));
|
||||||
}
|
list.add(c);
|
||||||
|
}
|
||||||
protected abstract Qualifier prepareResourceType(Document doc, DataInfo info);
|
}
|
||||||
|
return list;
|
||||||
protected abstract List<Instance> prepareInstances(
|
}
|
||||||
Document doc, DataInfo info, KeyValue collectedfrom, KeyValue hostedby);
|
|
||||||
|
protected abstract Qualifier prepareResourceType(Document doc, DataInfo info);
|
||||||
protected abstract List<Field<String>> prepareSources(Document doc, DataInfo info);
|
|
||||||
|
protected abstract List<Instance> prepareInstances(
|
||||||
protected abstract List<StructuredProperty> prepareRelevantDates(Document doc, DataInfo info);
|
Document doc, DataInfo info, KeyValue collectedfrom, KeyValue hostedby);
|
||||||
|
|
||||||
protected abstract List<Field<String>> prepareCoverages(Document doc, DataInfo info);
|
protected abstract List<Field<String>> prepareSources(Document doc, DataInfo info);
|
||||||
|
|
||||||
protected abstract List<Field<String>> prepareContributors(Document doc, DataInfo info);
|
protected abstract List<StructuredProperty> prepareRelevantDates(Document doc, DataInfo info);
|
||||||
|
|
||||||
protected abstract List<Field<String>> prepareFormats(Document doc, DataInfo info);
|
protected abstract List<Field<String>> prepareCoverages(Document doc, DataInfo info);
|
||||||
|
|
||||||
protected abstract Field<String> preparePublisher(Document doc, DataInfo info);
|
protected abstract List<Field<String>> prepareContributors(Document doc, DataInfo info);
|
||||||
|
|
||||||
protected abstract List<Field<String>> prepareDescriptions(Document doc, DataInfo info);
|
protected abstract List<Field<String>> prepareFormats(Document doc, DataInfo info);
|
||||||
|
|
||||||
protected abstract List<StructuredProperty> prepareTitles(Document doc, DataInfo info);
|
protected abstract Field<String> preparePublisher(Document doc, DataInfo info);
|
||||||
|
|
||||||
protected abstract List<StructuredProperty> prepareSubjects(Document doc, DataInfo info);
|
protected abstract List<Field<String>> prepareDescriptions(Document doc, DataInfo info);
|
||||||
|
|
||||||
protected abstract Qualifier prepareLanguages(Document doc);
|
protected abstract List<StructuredProperty> prepareTitles(Document doc, DataInfo info);
|
||||||
|
|
||||||
protected abstract List<Author> prepareAuthors(Document doc, DataInfo info);
|
protected abstract List<StructuredProperty> prepareSubjects(Document doc, DataInfo info);
|
||||||
|
|
||||||
protected abstract List<Field<String>> prepareOtherResearchProductTools(
|
protected abstract Qualifier prepareLanguages(Document doc);
|
||||||
Document doc, DataInfo info);
|
|
||||||
|
protected abstract List<Author> prepareAuthors(Document doc, DataInfo info);
|
||||||
protected abstract List<Field<String>> prepareOtherResearchProductContactGroups(
|
|
||||||
Document doc, DataInfo info);
|
protected abstract List<Field<String>> prepareOtherResearchProductTools(
|
||||||
|
Document doc, DataInfo info);
|
||||||
protected abstract List<Field<String>> prepareOtherResearchProductContactPersons(
|
|
||||||
Document doc, DataInfo info);
|
protected abstract List<Field<String>> prepareOtherResearchProductContactGroups(
|
||||||
|
Document doc, DataInfo info);
|
||||||
protected abstract Qualifier prepareSoftwareProgrammingLanguage(Document doc, DataInfo info);
|
|
||||||
|
protected abstract List<Field<String>> prepareOtherResearchProductContactPersons(
|
||||||
protected abstract Field<String> prepareSoftwareCodeRepositoryUrl(Document doc, DataInfo info);
|
Document doc, DataInfo info);
|
||||||
|
|
||||||
protected abstract List<StructuredProperty> prepareSoftwareLicenses(Document doc, DataInfo info);
|
protected abstract Qualifier prepareSoftwareProgrammingLanguage(Document doc, DataInfo info);
|
||||||
|
|
||||||
protected abstract List<Field<String>> prepareSoftwareDocumentationUrls(
|
protected abstract Field<String> prepareSoftwareCodeRepositoryUrl(Document doc, DataInfo info);
|
||||||
Document doc, DataInfo info);
|
|
||||||
|
protected abstract List<StructuredProperty> prepareSoftwareLicenses(Document doc, DataInfo info);
|
||||||
protected abstract List<GeoLocation> prepareDatasetGeoLocations(Document doc, DataInfo info);
|
|
||||||
|
protected abstract List<Field<String>> prepareSoftwareDocumentationUrls(
|
||||||
protected abstract Field<String> prepareDatasetMetadataVersionNumber(Document doc, DataInfo info);
|
Document doc, DataInfo info);
|
||||||
|
|
||||||
protected abstract Field<String> prepareDatasetLastMetadataUpdate(Document doc, DataInfo info);
|
protected abstract List<GeoLocation> prepareDatasetGeoLocations(Document doc, DataInfo info);
|
||||||
|
|
||||||
protected abstract Field<String> prepareDatasetVersion(Document doc, DataInfo info);
|
protected abstract Field<String> prepareDatasetMetadataVersionNumber(Document doc, DataInfo info);
|
||||||
|
|
||||||
protected abstract Field<String> prepareDatasetSize(Document doc, DataInfo info);
|
protected abstract Field<String> prepareDatasetLastMetadataUpdate(Document doc, DataInfo info);
|
||||||
|
|
||||||
protected abstract Field<String> prepareDatasetDevice(Document doc, DataInfo info);
|
protected abstract Field<String> prepareDatasetVersion(Document doc, DataInfo info);
|
||||||
|
|
||||||
protected abstract Field<String> prepareDatasetStorageDate(Document doc, DataInfo info);
|
protected abstract Field<String> prepareDatasetSize(Document doc, DataInfo info);
|
||||||
|
|
||||||
private Journal prepareJournal(final Document doc, final DataInfo info) {
|
protected abstract Field<String> prepareDatasetDevice(Document doc, DataInfo info);
|
||||||
final Node n = doc.selectSingleNode("//oaf:journal");
|
|
||||||
if (n != null) {
|
protected abstract Field<String> prepareDatasetStorageDate(Document doc, DataInfo info);
|
||||||
final String name = n.getText();
|
|
||||||
final String issnPrinted = n.valueOf("@issn");
|
private Journal prepareJournal(final Document doc, final DataInfo info) {
|
||||||
final String issnOnline = n.valueOf("@eissn");
|
final Node n = doc.selectSingleNode("//oaf:journal");
|
||||||
final String issnLinking = n.valueOf("@lissn");
|
if (n != null) {
|
||||||
final String ep = n.valueOf("@ep");
|
final String name = n.getText();
|
||||||
final String iss = n.valueOf("@iss");
|
final String issnPrinted = n.valueOf("@issn");
|
||||||
final String sp = n.valueOf("@sp");
|
final String issnOnline = n.valueOf("@eissn");
|
||||||
final String vol = n.valueOf("@vol");
|
final String issnLinking = n.valueOf("@lissn");
|
||||||
final String edition = n.valueOf("@edition");
|
final String ep = n.valueOf("@ep");
|
||||||
if (StringUtils.isNotBlank(name)) {
|
final String iss = n.valueOf("@iss");
|
||||||
return journal(
|
final String sp = n.valueOf("@sp");
|
||||||
name,
|
final String vol = n.valueOf("@vol");
|
||||||
issnPrinted,
|
final String edition = n.valueOf("@edition");
|
||||||
issnOnline,
|
if (StringUtils.isNotBlank(name)) {
|
||||||
issnLinking,
|
return journal(
|
||||||
ep,
|
name,
|
||||||
iss,
|
issnPrinted,
|
||||||
sp,
|
issnOnline,
|
||||||
vol,
|
issnLinking,
|
||||||
edition,
|
ep,
|
||||||
null,
|
iss,
|
||||||
null,
|
sp,
|
||||||
info);
|
vol,
|
||||||
}
|
edition,
|
||||||
}
|
null,
|
||||||
return null;
|
null,
|
||||||
}
|
info);
|
||||||
|
}
|
||||||
protected Qualifier prepareQualifier(
|
}
|
||||||
final Node node, final String xpath, final String schemeId, final String schemeName) {
|
return null;
|
||||||
final String classId = node.valueOf(xpath);
|
}
|
||||||
final String className = code2name.get(classId);
|
|
||||||
return qualifier(classId, className, schemeId, schemeName);
|
protected Qualifier prepareQualifier(
|
||||||
}
|
final Node node, final String xpath, final String schemeId, final String schemeName) {
|
||||||
|
final String classId = node.valueOf(xpath);
|
||||||
protected List<StructuredProperty> prepareListStructProps(
|
final String className = code2name.get(classId);
|
||||||
final Node node,
|
return qualifier(classId, className, schemeId, schemeName);
|
||||||
final String xpath,
|
}
|
||||||
final String xpathClassId,
|
|
||||||
final String schemeId,
|
protected List<StructuredProperty> prepareListStructProps(
|
||||||
final String schemeName,
|
final Node node,
|
||||||
final DataInfo info) {
|
final String xpath,
|
||||||
final List<StructuredProperty> res = new ArrayList<>();
|
final String xpathClassId,
|
||||||
for (final Object o : node.selectNodes(xpath)) {
|
final String schemeId,
|
||||||
final Node n = (Node) o;
|
final String schemeName,
|
||||||
final String classId = n.valueOf(xpathClassId);
|
final DataInfo info) {
|
||||||
final String className = code2name.get(classId);
|
final List<StructuredProperty> res = new ArrayList<>();
|
||||||
res.add(structuredProperty(n.getText(), classId, className, schemeId, schemeName, info));
|
for (final Object o : node.selectNodes(xpath)) {
|
||||||
}
|
final Node n = (Node) o;
|
||||||
return res;
|
final String classId = n.valueOf(xpathClassId);
|
||||||
}
|
final String className = code2name.get(classId);
|
||||||
|
res.add(structuredProperty(n.getText(), classId, className, schemeId, schemeName, info));
|
||||||
protected List<StructuredProperty> prepareListStructProps(
|
}
|
||||||
final Node node, final String xpath, final Qualifier qualifier, final DataInfo info) {
|
return res;
|
||||||
final List<StructuredProperty> res = new ArrayList<>();
|
}
|
||||||
for (final Object o : node.selectNodes(xpath)) {
|
|
||||||
final Node n = (Node) o;
|
protected List<StructuredProperty> prepareListStructProps(
|
||||||
res.add(structuredProperty(n.getText(), qualifier, info));
|
final Node node, final String xpath, final Qualifier qualifier, final DataInfo info) {
|
||||||
}
|
final List<StructuredProperty> res = new ArrayList<>();
|
||||||
return res;
|
for (final Object o : node.selectNodes(xpath)) {
|
||||||
}
|
final Node n = (Node) o;
|
||||||
|
res.add(structuredProperty(n.getText(), qualifier, info));
|
||||||
protected List<StructuredProperty> prepareListStructProps(
|
}
|
||||||
final Node node, final String xpath, final DataInfo info) {
|
return res;
|
||||||
final List<StructuredProperty> res = new ArrayList<>();
|
}
|
||||||
for (final Object o : node.selectNodes(xpath)) {
|
|
||||||
final Node n = (Node) o;
|
protected List<StructuredProperty> prepareListStructProps(
|
||||||
res
|
final Node node, final String xpath, final DataInfo info) {
|
||||||
.add(
|
final List<StructuredProperty> res = new ArrayList<>();
|
||||||
structuredProperty(
|
for (final Object o : node.selectNodes(xpath)) {
|
||||||
n.getText(),
|
final Node n = (Node) o;
|
||||||
n.valueOf("@classid"),
|
res.add(
|
||||||
n.valueOf("@classname"),
|
structuredProperty(
|
||||||
n.valueOf("@schemeid"),
|
n.getText(),
|
||||||
n.valueOf("@schemename"),
|
n.valueOf("@classid"),
|
||||||
info));
|
n.valueOf("@classname"),
|
||||||
}
|
n.valueOf("@schemeid"),
|
||||||
return res;
|
n.valueOf("@schemename"),
|
||||||
}
|
info));
|
||||||
|
}
|
||||||
protected OAIProvenance prepareOAIprovenance(final Document doc) {
|
return res;
|
||||||
final Node n = doc.selectSingleNode("//*[local-name()='provenance']/*[local-name()='originDescription']");
|
}
|
||||||
|
|
||||||
if (n == null) {
|
protected OAIProvenance prepareOAIprovenance(final Document doc) {
|
||||||
return null;
|
final Node n =
|
||||||
}
|
doc.selectSingleNode("//*[local-name()='provenance']/*[local-name()='originDescription']");
|
||||||
|
|
||||||
final String identifier = n.valueOf("./*[local-name()='identifier']");
|
if (n == null) {
|
||||||
final String baseURL = n.valueOf("./*[local-name()='baseURL']");
|
return null;
|
||||||
;
|
}
|
||||||
final String metadataNamespace = n.valueOf("./*[local-name()='metadataNamespace']");
|
|
||||||
;
|
final String identifier = n.valueOf("./*[local-name()='identifier']");
|
||||||
final boolean altered = n.valueOf("@altered").equalsIgnoreCase("true");
|
final String baseURL = n.valueOf("./*[local-name()='baseURL']");
|
||||||
final String datestamp = n.valueOf("./*[local-name()='datestamp']");
|
;
|
||||||
;
|
final String metadataNamespace = n.valueOf("./*[local-name()='metadataNamespace']");
|
||||||
final String harvestDate = n.valueOf("@harvestDate");
|
;
|
||||||
;
|
final boolean altered = n.valueOf("@altered").equalsIgnoreCase("true");
|
||||||
|
final String datestamp = n.valueOf("./*[local-name()='datestamp']");
|
||||||
return oaiIProvenance(identifier, baseURL, metadataNamespace, altered, datestamp, harvestDate);
|
;
|
||||||
}
|
final String harvestDate = n.valueOf("@harvestDate");
|
||||||
|
;
|
||||||
protected DataInfo prepareDataInfo(final Document doc) {
|
|
||||||
final Node n = doc.selectSingleNode("//oaf:datainfo");
|
return oaiIProvenance(identifier, baseURL, metadataNamespace, altered, datestamp, harvestDate);
|
||||||
|
}
|
||||||
if (n == null) {
|
|
||||||
return dataInfo(
|
protected DataInfo prepareDataInfo(final Document doc) {
|
||||||
false, null, false, false, MigrationConstants.REPOSITORY_PROVENANCE_ACTIONS, "0.9");
|
final Node n = doc.selectSingleNode("//oaf:datainfo");
|
||||||
}
|
|
||||||
|
if (n == null) {
|
||||||
final String paClassId = n.valueOf("./oaf:provenanceaction/@classid");
|
return dataInfo(
|
||||||
final String paClassName = n.valueOf("./oaf:provenanceaction/@classname");
|
false, null, false, false, MigrationConstants.REPOSITORY_PROVENANCE_ACTIONS, "0.9");
|
||||||
final String paSchemeId = n.valueOf("./oaf:provenanceaction/@schemeid");
|
}
|
||||||
final String paSchemeName = n.valueOf("./oaf:provenanceaction/@schemename");
|
|
||||||
|
final String paClassId = n.valueOf("./oaf:provenanceaction/@classid");
|
||||||
final boolean deletedbyinference = Boolean.parseBoolean(n.valueOf("./oaf:deletedbyinference"));
|
final String paClassName = n.valueOf("./oaf:provenanceaction/@classname");
|
||||||
final String inferenceprovenance = n.valueOf("./oaf:inferenceprovenance");
|
final String paSchemeId = n.valueOf("./oaf:provenanceaction/@schemeid");
|
||||||
final Boolean inferred = Boolean.parseBoolean(n.valueOf("./oaf:inferred"));
|
final String paSchemeName = n.valueOf("./oaf:provenanceaction/@schemename");
|
||||||
final String trust = n.valueOf("./oaf:trust");
|
|
||||||
|
final boolean deletedbyinference = Boolean.parseBoolean(n.valueOf("./oaf:deletedbyinference"));
|
||||||
return dataInfo(
|
final String inferenceprovenance = n.valueOf("./oaf:inferenceprovenance");
|
||||||
deletedbyinference,
|
final Boolean inferred = Boolean.parseBoolean(n.valueOf("./oaf:inferred"));
|
||||||
inferenceprovenance,
|
final String trust = n.valueOf("./oaf:trust");
|
||||||
inferred,
|
|
||||||
false,
|
return dataInfo(
|
||||||
qualifier(paClassId, paClassName, paSchemeId, paSchemeName),
|
deletedbyinference,
|
||||||
trust);
|
inferenceprovenance,
|
||||||
}
|
inferred,
|
||||||
|
false,
|
||||||
protected Field<String> prepareField(final Node node, final String xpath, final DataInfo info) {
|
qualifier(paClassId, paClassName, paSchemeId, paSchemeName),
|
||||||
return field(node.valueOf(xpath), info);
|
trust);
|
||||||
}
|
}
|
||||||
|
|
||||||
protected List<Field<String>> prepareListFields(
|
protected Field<String> prepareField(final Node node, final String xpath, final DataInfo info) {
|
||||||
final Node node, final String xpath, final DataInfo info) {
|
return field(node.valueOf(xpath), info);
|
||||||
return listFields(info, prepareListString(node, xpath));
|
}
|
||||||
}
|
|
||||||
|
protected List<Field<String>> prepareListFields(
|
||||||
protected List<String> prepareListString(final Node node, final String xpath) {
|
final Node node, final String xpath, final DataInfo info) {
|
||||||
final List<String> res = new ArrayList<>();
|
return listFields(info, prepareListString(node, xpath));
|
||||||
for (final Object o : node.selectNodes(xpath)) {
|
}
|
||||||
final String s = ((Node) o).getText().trim();
|
|
||||||
if (StringUtils.isNotBlank(s)) {
|
protected List<String> prepareListString(final Node node, final String xpath) {
|
||||||
res.add(s);
|
final List<String> res = new ArrayList<>();
|
||||||
}
|
for (final Object o : node.selectNodes(xpath)) {
|
||||||
}
|
final String s = ((Node) o).getText().trim();
|
||||||
return res;
|
if (StringUtils.isNotBlank(s)) {
|
||||||
}
|
res.add(s);
|
||||||
|
}
|
||||||
|
}
|
||||||
|
return res;
|
||||||
|
}
|
||||||
}
|
}
|
||||||
|
|
|
||||||
|
|
@ -1,256 +1,267 @@
|
||||||
|
|
||||||
package eu.dnetlib.dhp.oa.graph.raw;
|
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.createOpenaireId;
|
||||||
import static eu.dnetlib.dhp.oa.graph.raw.common.OafMapperUtils.field;
|
import static eu.dnetlib.dhp.oa.graph.raw.common.OafMapperUtils.field;
|
||||||
|
|
||||||
import eu.dnetlib.dhp.oa.graph.raw.common.PacePerson;
|
import eu.dnetlib.dhp.oa.graph.raw.common.PacePerson;
|
||||||
import eu.dnetlib.dhp.schema.oaf.*;
|
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 java.util.ArrayList;
|
import java.util.ArrayList;
|
||||||
import java.util.Arrays;
|
import java.util.Arrays;
|
||||||
import java.util.List;
|
import java.util.List;
|
||||||
import java.util.Map;
|
import java.util.Map;
|
||||||
|
import org.apache.commons.lang3.StringUtils;
|
||||||
import org.dom4j.Document;
|
import org.dom4j.Document;
|
||||||
import org.dom4j.Node;
|
import org.dom4j.Node;
|
||||||
|
|
||||||
public class OafToOafMapper extends AbstractMdRecordToOafMapper {
|
public class OafToOafMapper extends AbstractMdRecordToOafMapper {
|
||||||
|
|
||||||
public OafToOafMapper(final Map<String, String> code2name) {
|
public OafToOafMapper(final Map<String, String> code2name) {
|
||||||
super(code2name);
|
super(code2name);
|
||||||
}
|
}
|
||||||
|
|
||||||
@Override
|
@Override
|
||||||
protected List<Author> prepareAuthors(final Document doc, final DataInfo info) {
|
protected List<Author> prepareAuthors(final Document doc, final DataInfo info) {
|
||||||
final List<Author> res = new ArrayList<>();
|
final List<Author> res = new ArrayList<>();
|
||||||
int pos = 1;
|
int pos = 1;
|
||||||
for (final Object o : doc.selectNodes("//dc:creator")) {
|
for (final Object o : doc.selectNodes("//dc:creator")) {
|
||||||
final Node n = (Node) o;
|
final Node n = (Node) o;
|
||||||
final Author author = new Author();
|
final Author author = new Author();
|
||||||
author.setFullname(n.getText());
|
author.setFullname(n.getText());
|
||||||
author.setRank(pos++);
|
author.setRank(pos++);
|
||||||
final PacePerson p = new PacePerson(n.getText(), false);
|
final PacePerson p = new PacePerson(n.getText(), false);
|
||||||
if (p.isAccurate()) {
|
if (p.isAccurate()) {
|
||||||
author.setName(p.getNormalisedFirstName());
|
author.setName(p.getNormalisedFirstName());
|
||||||
author.setSurname(p.getNormalisedSurname());
|
author.setSurname(p.getNormalisedSurname());
|
||||||
}
|
}
|
||||||
res.add(author);
|
res.add(author);
|
||||||
}
|
}
|
||||||
return res;
|
return res;
|
||||||
}
|
}
|
||||||
|
|
||||||
@Override
|
@Override
|
||||||
protected Qualifier prepareLanguages(final Document doc) {
|
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
|
@Override
|
||||||
protected List<StructuredProperty> prepareSubjects(final Document doc, final DataInfo info) {
|
protected List<StructuredProperty> prepareSubjects(final Document doc, final DataInfo info) {
|
||||||
return prepareListStructProps(doc, "//dc:subject", info);
|
return prepareListStructProps(doc, "//dc:subject", info);
|
||||||
}
|
}
|
||||||
|
|
||||||
@Override
|
@Override
|
||||||
protected List<StructuredProperty> prepareTitles(final Document doc, final DataInfo info) {
|
protected List<StructuredProperty> prepareTitles(final Document doc, final DataInfo info) {
|
||||||
return prepareListStructProps(doc, "//dc:title", MAIN_TITLE_QUALIFIER, info);
|
return prepareListStructProps(doc, "//dc:title", MAIN_TITLE_QUALIFIER, info);
|
||||||
}
|
}
|
||||||
|
|
||||||
@Override
|
@Override
|
||||||
protected List<Field<String>> prepareDescriptions(final Document doc, final DataInfo info) {
|
protected List<Field<String>> prepareDescriptions(final Document doc, final DataInfo info) {
|
||||||
return prepareListFields(doc, "//dc:description", info);
|
return prepareListFields(doc, "//dc:description", info);
|
||||||
}
|
}
|
||||||
|
|
||||||
@Override
|
@Override
|
||||||
protected Field<String> preparePublisher(final Document doc, final DataInfo info) {
|
protected Field<String> preparePublisher(final Document doc, final DataInfo info) {
|
||||||
return prepareField(doc, "//dc:publisher", info);
|
return prepareField(doc, "//dc:publisher", info);
|
||||||
}
|
}
|
||||||
|
|
||||||
@Override
|
@Override
|
||||||
protected List<Field<String>> prepareFormats(final Document doc, final DataInfo info) {
|
protected List<Field<String>> prepareFormats(final Document doc, final DataInfo info) {
|
||||||
return prepareListFields(doc, "//dc:format", info);
|
return prepareListFields(doc, "//dc:format", info);
|
||||||
}
|
}
|
||||||
|
|
||||||
@Override
|
@Override
|
||||||
protected List<Field<String>> prepareContributors(final Document doc, final DataInfo info) {
|
protected List<Field<String>> prepareContributors(final Document doc, final DataInfo info) {
|
||||||
return prepareListFields(doc, "//dc:contributor", info);
|
return prepareListFields(doc, "//dc:contributor", info);
|
||||||
}
|
}
|
||||||
|
|
||||||
@Override
|
@Override
|
||||||
protected List<Field<String>> prepareCoverages(final Document doc, final DataInfo info) {
|
protected List<Field<String>> prepareCoverages(final Document doc, final DataInfo info) {
|
||||||
return prepareListFields(doc, "//dc:coverage", info);
|
return prepareListFields(doc, "//dc:coverage", info);
|
||||||
}
|
}
|
||||||
|
|
||||||
@Override
|
@Override
|
||||||
protected List<Instance> prepareInstances(
|
protected List<Instance> prepareInstances(
|
||||||
final Document doc,
|
final Document doc,
|
||||||
final DataInfo info,
|
final DataInfo info,
|
||||||
final KeyValue collectedfrom,
|
final KeyValue collectedfrom,
|
||||||
final KeyValue hostedby) {
|
final KeyValue hostedby) {
|
||||||
final List<Instance> res = new ArrayList<>();
|
final List<Instance> res = new ArrayList<>();
|
||||||
for (final Object o : doc.selectNodes("//dc:identifier")) {
|
for (final Object o : doc.selectNodes("//dc:identifier")) {
|
||||||
final String url = ((Node) o).getText().trim();
|
final String url = ((Node) o).getText().trim();
|
||||||
if (url.startsWith("http")) {
|
if (url.startsWith("http")) {
|
||||||
final Instance instance = new Instance();
|
final Instance instance = new Instance();
|
||||||
instance.setUrl(Arrays.asList(url));
|
instance.setUrl(Arrays.asList(url));
|
||||||
instance
|
instance.setInstancetype(
|
||||||
.setInstancetype(
|
prepareQualifier(
|
||||||
prepareQualifier(
|
doc,
|
||||||
doc,
|
"//dr:CobjCategory",
|
||||||
"//dr:CobjCategory",
|
"dnet:publication_resource",
|
||||||
"dnet:publication_resource",
|
"dnet:publication_resource"));
|
||||||
"dnet:publication_resource"));
|
instance.setCollectedfrom(collectedfrom);
|
||||||
instance.setCollectedfrom(collectedfrom);
|
instance.setHostedby(hostedby);
|
||||||
instance.setHostedby(hostedby);
|
instance.setDateofacceptance(field(doc.valueOf("//oaf:dateAccepted"), info));
|
||||||
instance.setDateofacceptance(field(doc.valueOf("//oaf:dateAccepted"), info));
|
instance.setDistributionlocation(doc.valueOf("//oaf:distributionlocation"));
|
||||||
instance.setDistributionlocation(doc.valueOf("//oaf:distributionlocation"));
|
instance.setAccessright(
|
||||||
instance
|
prepareQualifier(doc, "//oaf:accessrights", "dnet:access_modes", "dnet:access_modes"));
|
||||||
.setAccessright(
|
instance.setLicense(field(doc.valueOf("//oaf:license"), info));
|
||||||
prepareQualifier(doc, "//oaf:accessrights", "dnet:access_modes", "dnet:access_modes"));
|
instance.setRefereed(field(doc.valueOf("//oaf:refereed"), info));
|
||||||
instance.setLicense(field(doc.valueOf("//oaf:license"), info));
|
instance.setProcessingchargeamount(
|
||||||
instance.setRefereed(field(doc.valueOf("//oaf:refereed"), info));
|
field(doc.valueOf("//oaf:processingchargeamount"), info));
|
||||||
instance
|
instance.setProcessingchargecurrency(
|
||||||
.setProcessingchargeamount(
|
field(doc.valueOf("//oaf:processingchargeamount/@currency"), info));
|
||||||
field(doc.valueOf("//oaf:processingchargeamount"), info));
|
res.add(instance);
|
||||||
instance
|
}
|
||||||
.setProcessingchargecurrency(
|
}
|
||||||
field(doc.valueOf("//oaf:processingchargeamount/@currency"), info));
|
return res;
|
||||||
res.add(instance);
|
}
|
||||||
}
|
|
||||||
}
|
|
||||||
return res;
|
|
||||||
}
|
|
||||||
|
|
||||||
@Override
|
@Override
|
||||||
protected List<Field<String>> prepareSources(final Document doc, final DataInfo info) {
|
protected List<Field<String>> prepareSources(final Document doc, final DataInfo info) {
|
||||||
return prepareListFields(doc, "//dc:source", info);
|
return prepareListFields(doc, "//dc:source", info);
|
||||||
}
|
}
|
||||||
|
|
||||||
@Override
|
@Override
|
||||||
protected List<StructuredProperty> prepareRelevantDates(final Document doc, final DataInfo info) {
|
protected List<StructuredProperty> prepareRelevantDates(final Document doc, final DataInfo info) {
|
||||||
return new ArrayList<>(); // NOT PRESENT IN OAF
|
return new ArrayList<>(); // NOT PRESENT IN OAF
|
||||||
}
|
}
|
||||||
|
|
||||||
// SOFTWARES
|
// SOFTWARES
|
||||||
|
|
||||||
@Override
|
@Override
|
||||||
protected Qualifier prepareSoftwareProgrammingLanguage(final Document doc, final DataInfo info) {
|
protected Qualifier prepareSoftwareProgrammingLanguage(final Document doc, final DataInfo info) {
|
||||||
return null; // NOT PRESENT IN OAF
|
return null; // NOT PRESENT IN OAF
|
||||||
}
|
}
|
||||||
|
|
||||||
@Override
|
@Override
|
||||||
protected Field<String> prepareSoftwareCodeRepositoryUrl(
|
protected Field<String> prepareSoftwareCodeRepositoryUrl(
|
||||||
final Document doc, final DataInfo info) {
|
final Document doc, final DataInfo info) {
|
||||||
return null; // NOT PRESENT IN OAF
|
return null; // NOT PRESENT IN OAF
|
||||||
}
|
}
|
||||||
|
|
||||||
@Override
|
@Override
|
||||||
protected List<StructuredProperty> prepareSoftwareLicenses(
|
protected List<StructuredProperty> prepareSoftwareLicenses(
|
||||||
final Document doc, final DataInfo info) {
|
final Document doc, final DataInfo info) {
|
||||||
return new ArrayList<>(); // NOT PRESENT IN OAF
|
return new ArrayList<>(); // NOT PRESENT IN OAF
|
||||||
}
|
}
|
||||||
|
|
||||||
@Override
|
@Override
|
||||||
protected List<Field<String>> prepareSoftwareDocumentationUrls(
|
protected List<Field<String>> prepareSoftwareDocumentationUrls(
|
||||||
final Document doc, final DataInfo info) {
|
final Document doc, final DataInfo info) {
|
||||||
return new ArrayList<>(); // NOT PRESENT IN OAF
|
return new ArrayList<>(); // NOT PRESENT IN OAF
|
||||||
}
|
}
|
||||||
|
|
||||||
// DATASETS
|
// DATASETS
|
||||||
@Override
|
@Override
|
||||||
protected List<GeoLocation> prepareDatasetGeoLocations(final Document doc, final DataInfo info) {
|
protected List<GeoLocation> prepareDatasetGeoLocations(final Document doc, final DataInfo info) {
|
||||||
return new ArrayList<>(); // NOT PRESENT IN OAF
|
return new ArrayList<>(); // NOT PRESENT IN OAF
|
||||||
}
|
}
|
||||||
|
|
||||||
@Override
|
@Override
|
||||||
protected Field<String> prepareDatasetMetadataVersionNumber(
|
protected Field<String> prepareDatasetMetadataVersionNumber(
|
||||||
final Document doc, final DataInfo info) {
|
final Document doc, final DataInfo info) {
|
||||||
return null; // NOT PRESENT IN OAF
|
return null; // NOT PRESENT IN OAF
|
||||||
}
|
}
|
||||||
|
|
||||||
@Override
|
@Override
|
||||||
protected Field<String> prepareDatasetLastMetadataUpdate(
|
protected Field<String> prepareDatasetLastMetadataUpdate(
|
||||||
final Document doc, final DataInfo info) {
|
final Document doc, final DataInfo info) {
|
||||||
return null; // NOT PRESENT IN OAF
|
return null; // NOT PRESENT IN OAF
|
||||||
}
|
}
|
||||||
|
|
||||||
@Override
|
@Override
|
||||||
protected Field<String> prepareDatasetVersion(final Document doc, final DataInfo info) {
|
protected Field<String> prepareDatasetVersion(final Document doc, final DataInfo info) {
|
||||||
return null; // NOT PRESENT IN OAF
|
return null; // NOT PRESENT IN OAF
|
||||||
}
|
}
|
||||||
|
|
||||||
@Override
|
@Override
|
||||||
protected Field<String> prepareDatasetSize(final Document doc, final DataInfo info) {
|
protected Field<String> prepareDatasetSize(final Document doc, final DataInfo info) {
|
||||||
return null; // NOT PRESENT IN OAF
|
return null; // NOT PRESENT IN OAF
|
||||||
}
|
}
|
||||||
|
|
||||||
@Override
|
@Override
|
||||||
protected Field<String> prepareDatasetDevice(final Document doc, final DataInfo info) {
|
protected Field<String> prepareDatasetDevice(final Document doc, final DataInfo info) {
|
||||||
return null; // NOT PRESENT IN OAF
|
return null; // NOT PRESENT IN OAF
|
||||||
}
|
}
|
||||||
|
|
||||||
@Override
|
@Override
|
||||||
protected Field<String> prepareDatasetStorageDate(final Document doc, final DataInfo info) {
|
protected Field<String> prepareDatasetStorageDate(final Document doc, final DataInfo info) {
|
||||||
return null; // NOT PRESENT IN OAF
|
return null; // NOT PRESENT IN OAF
|
||||||
}
|
}
|
||||||
|
|
||||||
// OTHER PRODUCTS
|
// OTHER PRODUCTS
|
||||||
|
|
||||||
@Override
|
@Override
|
||||||
protected List<Field<String>> prepareOtherResearchProductTools(
|
protected List<Field<String>> prepareOtherResearchProductTools(
|
||||||
final Document doc, final DataInfo info) {
|
final Document doc, final DataInfo info) {
|
||||||
return new ArrayList<>(); // NOT PRESENT IN OAF
|
return new ArrayList<>(); // NOT PRESENT IN OAF
|
||||||
}
|
}
|
||||||
|
|
||||||
@Override
|
@Override
|
||||||
protected List<Field<String>> prepareOtherResearchProductContactGroups(
|
protected List<Field<String>> prepareOtherResearchProductContactGroups(
|
||||||
final Document doc, final DataInfo info) {
|
final Document doc, final DataInfo info) {
|
||||||
return new ArrayList<>(); // NOT PRESENT IN OAF
|
return new ArrayList<>(); // NOT PRESENT IN OAF
|
||||||
}
|
}
|
||||||
|
|
||||||
@Override
|
@Override
|
||||||
protected List<Field<String>> prepareOtherResearchProductContactPersons(
|
protected List<Field<String>> prepareOtherResearchProductContactPersons(
|
||||||
final Document doc, final DataInfo info) {
|
final Document doc, final DataInfo info) {
|
||||||
return new ArrayList<>(); // NOT PRESENT IN OAF
|
return new ArrayList<>(); // NOT PRESENT IN OAF
|
||||||
}
|
}
|
||||||
|
|
||||||
@Override
|
@Override
|
||||||
protected List<Oaf> addOtherResultRels(
|
protected List<Oaf> addOtherResultRels(
|
||||||
final Document doc,
|
final Document doc,
|
||||||
final KeyValue collectedFrom,
|
final KeyValue collectedFrom,
|
||||||
final DataInfo info,
|
final DataInfo info,
|
||||||
final long lastUpdateTimestamp) {
|
final long lastUpdateTimestamp) {
|
||||||
final String docId = createOpenaireId(50, doc.valueOf("//dri:objIdentifier"), false);
|
final String docId = createOpenaireId(50, doc.valueOf("//dri:objIdentifier"), false);
|
||||||
|
|
||||||
final List<Oaf> res = new ArrayList<>();
|
final List<Oaf> res = new ArrayList<>();
|
||||||
|
|
||||||
for (final Object o : doc.selectNodes("//*[local-name()='relatedDataset']")) {
|
for (final Object o : doc.selectNodes("//*[local-name()='relatedDataset']")) {
|
||||||
final String otherId = createOpenaireId(50, ((Node) o).getText(), false);
|
|
||||||
|
|
||||||
final Relation r1 = new Relation();
|
final String originalId = ((Node) o).getText();
|
||||||
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();
|
if (StringUtils.isNotBlank(originalId)) {
|
||||||
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);
|
|
||||||
}
|
|
||||||
return res;
|
|
||||||
}
|
|
||||||
|
|
||||||
@Override
|
final String otherId = createOpenaireId(50, originalId, false);
|
||||||
protected Qualifier prepareResourceType(final Document doc, final DataInfo info) {
|
|
||||||
return null; // NOT PRESENT IN OAF
|
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);
|
||||||
|
}
|
||||||
|
}
|
||||||
|
return res;
|
||||||
|
}
|
||||||
|
|
||||||
|
@Override
|
||||||
|
protected Qualifier prepareResourceType(final Document doc, final DataInfo info) {
|
||||||
|
return null; // NOT PRESENT IN OAF
|
||||||
|
}
|
||||||
}
|
}
|
||||||
|
|
|
||||||
|
|
@ -1,4 +1,3 @@
|
||||||
|
|
||||||
package eu.dnetlib.dhp.oa.graph.raw;
|
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.createOpenaireId;
|
||||||
|
|
@ -25,329 +24,328 @@ import org.dom4j.Node;
|
||||||
|
|
||||||
public class OdfToOafMapper extends AbstractMdRecordToOafMapper {
|
public class OdfToOafMapper extends AbstractMdRecordToOafMapper {
|
||||||
|
|
||||||
public OdfToOafMapper(final Map<String, String> code2name) {
|
public OdfToOafMapper(final Map<String, String> code2name) {
|
||||||
super(code2name);
|
super(code2name);
|
||||||
}
|
}
|
||||||
|
|
||||||
@Override
|
@Override
|
||||||
protected List<StructuredProperty> prepareTitles(final Document doc, final DataInfo info) {
|
protected List<StructuredProperty> prepareTitles(final Document doc, final DataInfo info) {
|
||||||
return prepareListStructProps(doc, "//datacite:title", MAIN_TITLE_QUALIFIER, info);
|
return prepareListStructProps(doc, "//datacite:title", MAIN_TITLE_QUALIFIER, info);
|
||||||
}
|
}
|
||||||
|
|
||||||
@Override
|
@Override
|
||||||
protected List<Author> prepareAuthors(final Document doc, final DataInfo info) {
|
protected List<Author> prepareAuthors(final Document doc, final DataInfo info) {
|
||||||
final List<Author> res = new ArrayList<>();
|
final List<Author> res = new ArrayList<>();
|
||||||
int pos = 1;
|
int pos = 1;
|
||||||
for (final Object o : doc.selectNodes("//datacite:creator")) {
|
for (final Object o : doc.selectNodes("//datacite:creator")) {
|
||||||
final Node n = (Node) o;
|
final Node n = (Node) o;
|
||||||
final Author author = new Author();
|
final Author author = new Author();
|
||||||
author.setFullname(n.valueOf("./datacite:creatorName"));
|
author.setFullname(n.valueOf("./datacite:creatorName"));
|
||||||
author.setName(n.valueOf("./datacite:givenName"));
|
author.setName(n.valueOf("./datacite:givenName"));
|
||||||
author.setSurname(n.valueOf("./datacite:familyName"));
|
author.setSurname(n.valueOf("./datacite:familyName"));
|
||||||
author.setAffiliation(prepareListFields(doc, "./datacite:affiliation", info));
|
author.setAffiliation(prepareListFields(doc, "./datacite:affiliation", info));
|
||||||
author.setPid(preparePids(doc, info));
|
author.setPid(preparePids(doc, info));
|
||||||
author.setRank(pos++);
|
author.setRank(pos++);
|
||||||
res.add(author);
|
res.add(author);
|
||||||
}
|
}
|
||||||
return res;
|
return res;
|
||||||
}
|
}
|
||||||
|
|
||||||
private List<StructuredProperty> preparePids(final Document doc, final DataInfo info) {
|
private List<StructuredProperty> preparePids(final Document doc, final DataInfo info) {
|
||||||
final List<StructuredProperty> res = new ArrayList<>();
|
final List<StructuredProperty> res = new ArrayList<>();
|
||||||
for (final Object o : doc.selectNodes("./datacite:nameIdentifier")) {
|
for (final Object o : doc.selectNodes("./datacite:nameIdentifier")) {
|
||||||
res
|
res.add(
|
||||||
.add(
|
structuredProperty(
|
||||||
structuredProperty(
|
((Node) o).getText(),
|
||||||
((Node) o).getText(),
|
prepareQualifier(
|
||||||
prepareQualifier(
|
(Node) o, "./@nameIdentifierScheme", "dnet:pid_types", "dnet:pid_types"),
|
||||||
(Node) o, "./@nameIdentifierScheme", "dnet:pid_types", "dnet:pid_types"),
|
info));
|
||||||
info));
|
}
|
||||||
}
|
return res;
|
||||||
return res;
|
}
|
||||||
}
|
|
||||||
|
|
||||||
@Override
|
@Override
|
||||||
protected List<Instance> prepareInstances(
|
protected List<Instance> prepareInstances(
|
||||||
final Document doc,
|
final Document doc,
|
||||||
final DataInfo info,
|
final DataInfo info,
|
||||||
final KeyValue collectedfrom,
|
final KeyValue collectedfrom,
|
||||||
final KeyValue hostedby) {
|
final KeyValue hostedby) {
|
||||||
|
|
||||||
final Instance instance = new Instance();
|
final Instance instance = new Instance();
|
||||||
instance.setUrl(new ArrayList<>());
|
instance.setUrl(new ArrayList<>());
|
||||||
instance
|
instance.setInstancetype(
|
||||||
.setInstancetype(
|
prepareQualifier(
|
||||||
prepareQualifier(
|
doc, "//dr:CobjCategory", "dnet:publication_resource", "dnet:publication_resource"));
|
||||||
doc, "//dr:CobjCategory", "dnet:publication_resource", "dnet:publication_resource"));
|
instance.setCollectedfrom(collectedfrom);
|
||||||
instance.setCollectedfrom(collectedfrom);
|
instance.setHostedby(hostedby);
|
||||||
instance.setHostedby(hostedby);
|
instance.setDateofacceptance(field(doc.valueOf("//oaf:dateAccepted"), info));
|
||||||
instance.setDateofacceptance(field(doc.valueOf("//oaf:dateAccepted"), info));
|
instance.setDistributionlocation(doc.valueOf("//oaf:distributionlocation"));
|
||||||
instance.setDistributionlocation(doc.valueOf("//oaf:distributionlocation"));
|
instance.setAccessright(
|
||||||
instance
|
prepareQualifier(doc, "//oaf:accessrights", "dnet:access_modes", "dnet:access_modes"));
|
||||||
.setAccessright(
|
instance.setLicense(field(doc.valueOf("//oaf:license"), info));
|
||||||
prepareQualifier(doc, "//oaf:accessrights", "dnet:access_modes", "dnet:access_modes"));
|
instance.setRefereed(field(doc.valueOf("//oaf:refereed"), info));
|
||||||
instance.setLicense(field(doc.valueOf("//oaf:license"), info));
|
instance.setProcessingchargeamount(field(doc.valueOf("//oaf:processingchargeamount"), info));
|
||||||
instance.setRefereed(field(doc.valueOf("//oaf:refereed"), info));
|
instance.setProcessingchargecurrency(
|
||||||
instance.setProcessingchargeamount(field(doc.valueOf("//oaf:processingchargeamount"), info));
|
field(doc.valueOf("//oaf:processingchargeamount/@currency"), info));
|
||||||
instance
|
|
||||||
.setProcessingchargecurrency(
|
|
||||||
field(doc.valueOf("//oaf:processingchargeamount/@currency"), info));
|
|
||||||
|
|
||||||
for (final Object o : doc.selectNodes("//datacite:alternateIdentifier[@alternateIdentifierType='URL']")) {
|
for (final Object o :
|
||||||
instance.getUrl().add(((Node) o).getText().trim());
|
doc.selectNodes("//datacite:alternateIdentifier[@alternateIdentifierType='URL']")) {
|
||||||
}
|
instance.getUrl().add(((Node) o).getText().trim());
|
||||||
for (final Object o : doc.selectNodes("//datacite:identifier[@identifierType='URL']")) {
|
}
|
||||||
instance.getUrl().add(((Node) o).getText().trim());
|
for (final Object o : doc.selectNodes("//datacite:identifier[@identifierType='URL']")) {
|
||||||
}
|
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());
|
for (final Object o :
|
||||||
}
|
doc.selectNodes("//datacite:alternateIdentifier[@alternateIdentifierType='DOI']")) {
|
||||||
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.org/" + ((Node) o).getText().trim());
|
}
|
||||||
}
|
for (final Object o : doc.selectNodes("//datacite:identifier[@identifierType='DOI']")) {
|
||||||
return Arrays.asList(instance);
|
instance.getUrl().add("http://dx.doi.org/" + ((Node) o).getText().trim());
|
||||||
}
|
}
|
||||||
|
return Arrays.asList(instance);
|
||||||
|
}
|
||||||
|
|
||||||
@Override
|
@Override
|
||||||
protected List<Field<String>> prepareSources(final Document doc, final DataInfo info) {
|
protected List<Field<String>> prepareSources(final Document doc, final DataInfo info) {
|
||||||
return new ArrayList<>(); // Not present in ODF ???
|
return new ArrayList<>(); // Not present in ODF ???
|
||||||
}
|
}
|
||||||
|
|
||||||
@Override
|
@Override
|
||||||
protected List<StructuredProperty> prepareRelevantDates(final Document doc, final DataInfo info) {
|
protected List<StructuredProperty> prepareRelevantDates(final Document doc, final DataInfo info) {
|
||||||
final List<StructuredProperty> res = new ArrayList<>();
|
final List<StructuredProperty> res = new ArrayList<>();
|
||||||
for (final Object o : doc.selectNodes("//datacite:date")) {
|
for (final Object o : doc.selectNodes("//datacite:date")) {
|
||||||
final String dateType = ((Node) o).valueOf("@dateType");
|
final String dateType = ((Node) o).valueOf("@dateType");
|
||||||
if (StringUtils.isBlank(dateType)
|
if (StringUtils.isBlank(dateType)
|
||||||
&& !dateType.equalsIgnoreCase("Accepted")
|
&& !dateType.equalsIgnoreCase("Accepted")
|
||||||
&& !dateType.equalsIgnoreCase("Issued")
|
&& !dateType.equalsIgnoreCase("Issued")
|
||||||
&& !dateType.equalsIgnoreCase("Updated")
|
&& !dateType.equalsIgnoreCase("Updated")
|
||||||
&& !dateType.equalsIgnoreCase("Available")) {
|
&& !dateType.equalsIgnoreCase("Available")) {
|
||||||
res
|
res.add(
|
||||||
.add(
|
structuredProperty(
|
||||||
structuredProperty(
|
((Node) o).getText(),
|
||||||
((Node) o).getText(),
|
"UNKNOWN",
|
||||||
"UNKNOWN",
|
"UNKNOWN",
|
||||||
"UNKNOWN",
|
"dnet:dataCite_date",
|
||||||
"dnet:dataCite_date",
|
"dnet:dataCite_date",
|
||||||
"dnet:dataCite_date",
|
info));
|
||||||
info));
|
}
|
||||||
}
|
}
|
||||||
}
|
return res;
|
||||||
return res;
|
}
|
||||||
}
|
|
||||||
|
|
||||||
@Override
|
@Override
|
||||||
protected List<Field<String>> prepareCoverages(final Document doc, final DataInfo info) {
|
protected List<Field<String>> prepareCoverages(final Document doc, final DataInfo info) {
|
||||||
return new ArrayList<>(); // Not present in ODF ???
|
return new ArrayList<>(); // Not present in ODF ???
|
||||||
}
|
}
|
||||||
|
|
||||||
@Override
|
@Override
|
||||||
protected List<Field<String>> prepareContributors(final Document doc, final DataInfo info) {
|
protected List<Field<String>> prepareContributors(final Document doc, final DataInfo info) {
|
||||||
return prepareListFields(doc, "//datacite:contributorName", info);
|
return prepareListFields(doc, "//datacite:contributorName", info);
|
||||||
}
|
}
|
||||||
|
|
||||||
@Override
|
@Override
|
||||||
protected List<Field<String>> prepareFormats(final Document doc, final DataInfo info) {
|
protected List<Field<String>> prepareFormats(final Document doc, final DataInfo info) {
|
||||||
return prepareListFields(doc, "//datacite:format", info);
|
return prepareListFields(doc, "//datacite:format", info);
|
||||||
}
|
}
|
||||||
|
|
||||||
@Override
|
@Override
|
||||||
protected Field<String> preparePublisher(final Document doc, final DataInfo info) {
|
protected Field<String> preparePublisher(final Document doc, final DataInfo info) {
|
||||||
return prepareField(doc, "//datacite:publisher", info);
|
return prepareField(doc, "//datacite:publisher", info);
|
||||||
}
|
}
|
||||||
|
|
||||||
@Override
|
@Override
|
||||||
protected List<Field<String>> prepareDescriptions(final Document doc, final DataInfo info) {
|
protected List<Field<String>> prepareDescriptions(final Document doc, final DataInfo info) {
|
||||||
return prepareListFields(doc, "//datacite:description[@descriptionType='Abstract']", info);
|
return prepareListFields(doc, "//datacite:description[@descriptionType='Abstract']", info);
|
||||||
}
|
}
|
||||||
|
|
||||||
@Override
|
@Override
|
||||||
protected List<StructuredProperty> prepareSubjects(final Document doc, final DataInfo info) {
|
protected List<StructuredProperty> prepareSubjects(final Document doc, final DataInfo info) {
|
||||||
return prepareListStructProps(doc, "//datacite:subject", info);
|
return prepareListStructProps(doc, "//datacite:subject", info);
|
||||||
}
|
}
|
||||||
|
|
||||||
@Override
|
@Override
|
||||||
protected Qualifier prepareLanguages(final Document doc) {
|
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
|
@Override
|
||||||
protected List<Field<String>> prepareOtherResearchProductTools(
|
protected List<Field<String>> prepareOtherResearchProductTools(
|
||||||
final Document doc, final DataInfo info) {
|
final Document doc, final DataInfo info) {
|
||||||
return new ArrayList<>(); // Not present in ODF ???
|
return new ArrayList<>(); // Not present in ODF ???
|
||||||
}
|
}
|
||||||
|
|
||||||
@Override
|
@Override
|
||||||
protected List<Field<String>> prepareOtherResearchProductContactGroups(
|
protected List<Field<String>> prepareOtherResearchProductContactGroups(
|
||||||
final Document doc, final DataInfo info) {
|
final Document doc, final DataInfo info) {
|
||||||
return prepareListFields(
|
return prepareListFields(
|
||||||
doc,
|
doc,
|
||||||
"//datacite:contributor[@contributorType='ContactGroup']/datacite:contributorName",
|
"//datacite:contributor[@contributorType='ContactGroup']/datacite:contributorName",
|
||||||
info);
|
info);
|
||||||
}
|
}
|
||||||
|
|
||||||
@Override
|
@Override
|
||||||
protected List<Field<String>> prepareOtherResearchProductContactPersons(
|
protected List<Field<String>> prepareOtherResearchProductContactPersons(
|
||||||
final Document doc, final DataInfo info) {
|
final Document doc, final DataInfo info) {
|
||||||
return prepareListFields(
|
return prepareListFields(
|
||||||
doc,
|
doc,
|
||||||
"//datacite:contributor[@contributorType='ContactPerson']/datacite:contributorName",
|
"//datacite:contributor[@contributorType='ContactPerson']/datacite:contributorName",
|
||||||
info);
|
info);
|
||||||
}
|
}
|
||||||
|
|
||||||
@Override
|
@Override
|
||||||
protected Qualifier prepareSoftwareProgrammingLanguage(final Document doc, final DataInfo info) {
|
protected Qualifier prepareSoftwareProgrammingLanguage(final Document doc, final DataInfo info) {
|
||||||
return prepareQualifier(
|
return prepareQualifier(
|
||||||
doc, "//datacite:format", "dnet:programming_languages", "dnet:programming_languages");
|
doc, "//datacite:format", "dnet:programming_languages", "dnet:programming_languages");
|
||||||
}
|
}
|
||||||
|
|
||||||
@Override
|
@Override
|
||||||
protected Field<String> prepareSoftwareCodeRepositoryUrl(
|
protected Field<String> prepareSoftwareCodeRepositoryUrl(
|
||||||
final Document doc, final DataInfo info) {
|
final Document doc, final DataInfo info) {
|
||||||
return null; // Not present in ODF ???
|
return null; // Not present in ODF ???
|
||||||
}
|
}
|
||||||
|
|
||||||
@Override
|
@Override
|
||||||
protected List<StructuredProperty> prepareSoftwareLicenses(
|
protected List<StructuredProperty> prepareSoftwareLicenses(
|
||||||
final Document doc, final DataInfo info) {
|
final Document doc, final DataInfo info) {
|
||||||
return new ArrayList<>(); // Not present in ODF ???
|
return new ArrayList<>(); // Not present in ODF ???
|
||||||
}
|
}
|
||||||
|
|
||||||
@Override
|
@Override
|
||||||
protected List<Field<String>> prepareSoftwareDocumentationUrls(
|
protected List<Field<String>> prepareSoftwareDocumentationUrls(
|
||||||
final Document doc, final DataInfo info) {
|
final Document doc, final DataInfo info) {
|
||||||
return prepareListFields(
|
return prepareListFields(
|
||||||
doc,
|
doc,
|
||||||
"//datacite:relatedIdentifier[@relatedIdentifierType='URL' and @relationType='IsDocumentedBy']",
|
"//datacite:relatedIdentifier[@relatedIdentifierType='URL' and @relationType='IsDocumentedBy']",
|
||||||
info);
|
info);
|
||||||
}
|
}
|
||||||
|
|
||||||
// DATASETS
|
// DATASETS
|
||||||
|
|
||||||
@Override
|
@Override
|
||||||
protected List<GeoLocation> prepareDatasetGeoLocations(final Document doc, final DataInfo info) {
|
protected List<GeoLocation> prepareDatasetGeoLocations(final Document doc, final DataInfo info) {
|
||||||
final List<GeoLocation> res = new ArrayList<>();
|
final List<GeoLocation> res = new ArrayList<>();
|
||||||
|
|
||||||
for (final Object o : doc.selectNodes("//datacite:geoLocation")) {
|
for (final Object o : doc.selectNodes("//datacite:geoLocation")) {
|
||||||
final GeoLocation loc = new GeoLocation();
|
final GeoLocation loc = new GeoLocation();
|
||||||
loc.setBox(((Node) o).valueOf("./datacite:geoLocationBox"));
|
loc.setBox(((Node) o).valueOf("./datacite:geoLocationBox"));
|
||||||
loc.setPlace(((Node) o).valueOf("./datacite:geoLocationPlace"));
|
loc.setPlace(((Node) o).valueOf("./datacite:geoLocationPlace"));
|
||||||
loc.setPoint(((Node) o).valueOf("./datacite:geoLocationPoint"));
|
loc.setPoint(((Node) o).valueOf("./datacite:geoLocationPoint"));
|
||||||
res.add(loc);
|
res.add(loc);
|
||||||
}
|
}
|
||||||
return res;
|
return res;
|
||||||
}
|
}
|
||||||
|
|
||||||
@Override
|
@Override
|
||||||
protected Field<String> prepareDatasetMetadataVersionNumber(
|
protected Field<String> prepareDatasetMetadataVersionNumber(
|
||||||
final Document doc, final DataInfo info) {
|
final Document doc, final DataInfo info) {
|
||||||
return null; // Not present in ODF ???
|
return null; // Not present in ODF ???
|
||||||
}
|
}
|
||||||
|
|
||||||
@Override
|
@Override
|
||||||
protected Field<String> prepareDatasetLastMetadataUpdate(
|
protected Field<String> prepareDatasetLastMetadataUpdate(
|
||||||
final Document doc, final DataInfo info) {
|
final Document doc, final DataInfo info) {
|
||||||
return prepareField(doc, "//datacite:date[@dateType='Updated']", info);
|
return prepareField(doc, "//datacite:date[@dateType='Updated']", info);
|
||||||
}
|
}
|
||||||
|
|
||||||
@Override
|
@Override
|
||||||
protected Field<String> prepareDatasetVersion(final Document doc, final DataInfo info) {
|
protected Field<String> prepareDatasetVersion(final Document doc, final DataInfo info) {
|
||||||
return prepareField(doc, "//datacite:version", info);
|
return prepareField(doc, "//datacite:version", info);
|
||||||
}
|
}
|
||||||
|
|
||||||
@Override
|
@Override
|
||||||
protected Field<String> prepareDatasetSize(final Document doc, final DataInfo info) {
|
protected Field<String> prepareDatasetSize(final Document doc, final DataInfo info) {
|
||||||
return prepareField(doc, "//datacite:size", info);
|
return prepareField(doc, "//datacite:size", info);
|
||||||
}
|
}
|
||||||
|
|
||||||
@Override
|
@Override
|
||||||
protected Field<String> prepareDatasetDevice(final Document doc, final DataInfo info) {
|
protected Field<String> prepareDatasetDevice(final Document doc, final DataInfo info) {
|
||||||
return null; // Not present in ODF ???
|
return null; // Not present in ODF ???
|
||||||
}
|
}
|
||||||
|
|
||||||
@Override
|
@Override
|
||||||
protected Field<String> prepareDatasetStorageDate(final Document doc, final DataInfo info) {
|
protected Field<String> prepareDatasetStorageDate(final Document doc, final DataInfo info) {
|
||||||
return prepareField(doc, "//datacite:date[@dateType='Issued']", info);
|
return prepareField(doc, "//datacite:date[@dateType='Issued']", info);
|
||||||
}
|
}
|
||||||
|
|
||||||
@Override
|
@Override
|
||||||
protected List<Oaf> addOtherResultRels(
|
protected List<Oaf> addOtherResultRels(
|
||||||
final Document doc,
|
final Document doc,
|
||||||
final KeyValue collectedFrom,
|
final KeyValue collectedFrom,
|
||||||
final DataInfo info,
|
final DataInfo info,
|
||||||
final long lastUpdateTimestamp) {
|
final long lastUpdateTimestamp) {
|
||||||
|
|
||||||
final String docId = createOpenaireId(50, doc.valueOf("//dri:objIdentifier"), false);
|
final String docId = createOpenaireId(50, doc.valueOf("//dri:objIdentifier"), false);
|
||||||
|
|
||||||
final List<Oaf> res = new ArrayList<>();
|
final List<Oaf> res = new ArrayList<>();
|
||||||
|
|
||||||
for (final Object o : doc.selectNodes("//datacite:relatedIdentifier[@relatedIdentifierType='OPENAIRE']")) {
|
for (final Object o :
|
||||||
final String otherId = createOpenaireId(50, ((Node) o).getText(), false);
|
doc.selectNodes("//datacite:relatedIdentifier[@relatedIdentifierType='OPENAIRE']")) {
|
||||||
final String type = ((Node) o).valueOf("@relationType");
|
|
||||||
|
|
||||||
if (type.equals("IsSupplementTo")) {
|
final String originalId = ((Node) o).getText();
|
||||||
res
|
|
||||||
.add(
|
|
||||||
prepareOtherResultRel(
|
|
||||||
collectedFrom,
|
|
||||||
info,
|
|
||||||
lastUpdateTimestamp,
|
|
||||||
docId,
|
|
||||||
otherId,
|
|
||||||
"supplement",
|
|
||||||
"isSupplementTo"));
|
|
||||||
res
|
|
||||||
.add(
|
|
||||||
prepareOtherResultRel(
|
|
||||||
collectedFrom,
|
|
||||||
info,
|
|
||||||
lastUpdateTimestamp,
|
|
||||||
otherId,
|
|
||||||
docId,
|
|
||||||
"supplement",
|
|
||||||
"isSupplementedBy"));
|
|
||||||
} else if (type.equals("IsPartOf")) {
|
|
||||||
res
|
|
||||||
.add(
|
|
||||||
prepareOtherResultRel(
|
|
||||||
collectedFrom, info, lastUpdateTimestamp, docId, otherId, "part", "IsPartOf"));
|
|
||||||
res
|
|
||||||
.add(
|
|
||||||
prepareOtherResultRel(
|
|
||||||
collectedFrom, info, lastUpdateTimestamp, otherId, docId, "part", "HasParts"));
|
|
||||||
} else {
|
|
||||||
}
|
|
||||||
}
|
|
||||||
return res;
|
|
||||||
}
|
|
||||||
|
|
||||||
private Relation prepareOtherResultRel(
|
if (StringUtils.isNotBlank(originalId)) {
|
||||||
final KeyValue collectedFrom,
|
final String otherId = createOpenaireId(50, originalId, false);
|
||||||
final DataInfo info,
|
final String type = ((Node) o).valueOf("@relationType");
|
||||||
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
|
if (type.equals("IsSupplementTo")) {
|
||||||
protected Qualifier prepareResourceType(final Document doc, final DataInfo info) {
|
res.add(
|
||||||
return prepareQualifier(
|
prepareOtherResultRel(
|
||||||
doc,
|
collectedFrom,
|
||||||
"//*[local-name() = 'resource']//*[local-name() = 'resourceType']",
|
info,
|
||||||
"dnet:dataCite_resource",
|
lastUpdateTimestamp,
|
||||||
"dnet:dataCite_resource");
|
docId,
|
||||||
}
|
otherId,
|
||||||
|
"supplement",
|
||||||
|
"isSupplementTo"));
|
||||||
|
res.add(
|
||||||
|
prepareOtherResultRel(
|
||||||
|
collectedFrom,
|
||||||
|
info,
|
||||||
|
lastUpdateTimestamp,
|
||||||
|
otherId,
|
||||||
|
docId,
|
||||||
|
"supplement",
|
||||||
|
"isSupplementedBy"));
|
||||||
|
} else if (type.equals("IsPartOf")) {
|
||||||
|
res.add(
|
||||||
|
prepareOtherResultRel(
|
||||||
|
collectedFrom, info, lastUpdateTimestamp, docId, otherId, "part", "IsPartOf"));
|
||||||
|
res.add(
|
||||||
|
prepareOtherResultRel(
|
||||||
|
collectedFrom, info, lastUpdateTimestamp, otherId, docId, "part", "HasParts"));
|
||||||
|
} else {
|
||||||
|
}
|
||||||
|
}
|
||||||
|
}
|
||||||
|
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");
|
||||||
|
}
|
||||||
}
|
}
|
||||||
|
|
|
||||||
Loading…
Reference in New Issue