forked from D-Net/dnet-hadoop
Merge branch 'master' of https://code-repo.d4science.org/D-Net/dnet-hadoop
This commit is contained in:
commit
e81960335c
|
|
@ -10,6 +10,20 @@ 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.qualifier;
|
||||||
import static eu.dnetlib.dhp.oa.graph.raw.common.OafMapperUtils.structuredProperty;
|
import static eu.dnetlib.dhp.oa.graph.raw.common.OafMapperUtils.structuredProperty;
|
||||||
|
|
||||||
|
import java.util.ArrayList;
|
||||||
|
import java.util.Arrays;
|
||||||
|
import java.util.Date;
|
||||||
|
import java.util.HashMap;
|
||||||
|
import java.util.List;
|
||||||
|
import java.util.Map;
|
||||||
|
|
||||||
|
import org.apache.commons.lang3.StringUtils;
|
||||||
|
import org.dom4j.Document;
|
||||||
|
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.Author;
|
||||||
import eu.dnetlib.dhp.schema.oaf.DataInfo;
|
import eu.dnetlib.dhp.schema.oaf.DataInfo;
|
||||||
import eu.dnetlib.dhp.schema.oaf.Dataset;
|
import eu.dnetlib.dhp.schema.oaf.Dataset;
|
||||||
|
|
@ -27,452 +41,388 @@ import eu.dnetlib.dhp.schema.oaf.Relation;
|
||||||
import eu.dnetlib.dhp.schema.oaf.Result;
|
import eu.dnetlib.dhp.schema.oaf.Result;
|
||||||
import eu.dnetlib.dhp.schema.oaf.Software;
|
import eu.dnetlib.dhp.schema.oaf.Software;
|
||||||
import eu.dnetlib.dhp.schema.oaf.StructuredProperty;
|
import eu.dnetlib.dhp.schema.oaf.StructuredProperty;
|
||||||
import java.util.ArrayList;
|
|
||||||
import java.util.Arrays;
|
|
||||||
import java.util.Date;
|
|
||||||
import java.util.HashMap;
|
|
||||||
import java.util.List;
|
|
||||||
import java.util.Map;
|
|
||||||
import org.apache.commons.lang3.StringUtils;
|
|
||||||
import org.dom4j.Document;
|
|
||||||
import org.dom4j.DocumentFactory;
|
|
||||||
import org.dom4j.DocumentHelper;
|
|
||||||
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 =
|
protected static final Qualifier MAIN_TITLE_QUALIFIER =
|
||||||
qualifier("main title", "main title", "dnet:dataCite_title", "dnet:dataCite_title");
|
qualifier("main title", "main title", "dnet:dataCite_title", "dnet:dataCite_title");
|
||||||
|
|
||||||
protected static final Qualifier PUBLICATION_RESULTTYPE_QUALIFIER =
|
protected AbstractMdRecordToOafMapper(final Map<String, String> code2name) {
|
||||||
qualifier(
|
this.code2name = code2name;
|
||||||
"publication",
|
}
|
||||||
"publication",
|
|
||||||
"dnet:result_typologies",
|
public List<Oaf> processMdRecord(final String xml) {
|
||||||
"dnet:result_typologies");
|
try {
|
||||||
protected static final Qualifier DATASET_RESULTTYPE_QUALIFIER =
|
final Map<String, String> nsContext = new HashMap<>();
|
||||||
qualifier("dataset", "dataset", "dnet:result_typologies", "dnet:result_typologies");
|
nsContext.put("dr", "http://www.driver-repository.eu/namespace/dr");
|
||||||
protected static final Qualifier SOFTWARE_RESULTTYPE_QUALIFIER =
|
nsContext.put("dri", "http://www.driver-repository.eu/namespace/dri");
|
||||||
qualifier("software", "software", "dnet:result_typologies", "dnet:result_typologies");
|
nsContext.put("oaf", "http://namespace.openaire.eu/oaf");
|
||||||
protected static final Qualifier OTHER_RESULTTYPE_QUALIFIER =
|
nsContext.put("oai", "http://www.openarchives.org/OAI/2.0/");
|
||||||
qualifier("other", "other", "dnet:result_typologies", "dnet:result_typologies");
|
nsContext.put("prov", "http://www.openarchives.org/OAI/2.0/provenance");
|
||||||
protected static final Qualifier REPOSITORY_QUALIFIER =
|
nsContext.put("dc", "http://purl.org/dc/elements/1.1/");
|
||||||
qualifier(
|
nsContext.put("datacite", "http://datacite.org/schema/kernel-3");
|
||||||
"sysimport:crosswalk:repository",
|
DocumentFactory.getInstance().setXPathNamespaceURIs(nsContext);
|
||||||
"sysimport:crosswalk:repository",
|
|
||||||
"dnet:provenanceActions",
|
final Document doc =
|
||||||
"dnet:provenanceActions");
|
DocumentHelper.parseText(xml.replaceAll("http://datacite.org/schema/kernel-4", "http://datacite.org/schema/kernel-3"));
|
||||||
|
|
||||||
protected AbstractMdRecordToOafMapper(final Map<String, String> code2name) {
|
final String type = doc.valueOf("//dr:CobjCategory/@type");
|
||||||
this.code2name = code2name;
|
final KeyValue collectedFrom =
|
||||||
}
|
keyValue(createOpenaireId(10, doc.valueOf("//oaf:collectedFrom/@id"), true), doc.valueOf("//oaf:collectedFrom/@name"));
|
||||||
|
final KeyValue hostedBy =
|
||||||
public List<Oaf> processMdRecord(final String xml) {
|
StringUtils.isBlank(doc.valueOf("//oaf:hostedBy/@id"))
|
||||||
try {
|
? collectedFrom
|
||||||
final Map<String, String> nsContext = new HashMap<>();
|
: keyValue(createOpenaireId(10, doc.valueOf("//oaf:hostedBy/@id"), true), doc.valueOf("//oaf:hostedBy/@name"));
|
||||||
nsContext.put("dr", "http://www.driver-repository.eu/namespace/dr");
|
|
||||||
nsContext.put("dri", "http://www.driver-repository.eu/namespace/dri");
|
final DataInfo info = prepareDataInfo(doc);
|
||||||
nsContext.put("oaf", "http://namespace.openaire.eu/oaf");
|
final long lastUpdateTimestamp = new Date().getTime();
|
||||||
nsContext.put("oai", "http://www.openarchives.org/OAI/2.0/");
|
|
||||||
nsContext.put("prov", "http://www.openarchives.org/OAI/2.0/provenance");
|
return createOafs(doc, type, collectedFrom, hostedBy, info, lastUpdateTimestamp);
|
||||||
nsContext.put("dc", "http://purl.org/dc/elements/1.1/");
|
} catch (final Exception e) {
|
||||||
nsContext.put("datacite", "http://datacite.org/schema/kernel-3");
|
throw new RuntimeException(e);
|
||||||
DocumentFactory.getInstance().setXPathNamespaceURIs(nsContext);
|
}
|
||||||
|
}
|
||||||
final Document doc =
|
|
||||||
DocumentHelper.parseText(
|
protected List<Oaf> createOafs(
|
||||||
xml.replaceAll(
|
final Document doc,
|
||||||
"http://datacite.org/schema/kernel-4",
|
final String type,
|
||||||
"http://datacite.org/schema/kernel-3"));
|
final KeyValue collectedFrom,
|
||||||
|
final KeyValue hostedBy,
|
||||||
final String type = doc.valueOf("//dr:CobjCategory/@type");
|
final DataInfo info,
|
||||||
final KeyValue collectedFrom =
|
final long lastUpdateTimestamp) {
|
||||||
keyValue(
|
|
||||||
createOpenaireId(10, doc.valueOf("//oaf:collectedFrom/@id"), true),
|
final List<Oaf> oafs = new ArrayList<>();
|
||||||
doc.valueOf("//oaf:collectedFrom/@name"));
|
|
||||||
final KeyValue hostedBy =
|
switch (type.toLowerCase()) {
|
||||||
StringUtils.isBlank(doc.valueOf("//oaf:hostedBy/@id"))
|
case "":
|
||||||
? collectedFrom
|
case "publication":
|
||||||
: keyValue(
|
final Publication p = new Publication();
|
||||||
createOpenaireId(10, doc.valueOf("//oaf:hostedBy/@id"), true),
|
populateResultFields(p, doc, collectedFrom, hostedBy, info, lastUpdateTimestamp);
|
||||||
doc.valueOf("//oaf:hostedBy/@name"));
|
p.setResulttype(MigrationConstants.PUBLICATION_RESULTTYPE_QUALIFIER);
|
||||||
|
p.setJournal(prepareJournal(doc, info));
|
||||||
final DataInfo info = prepareDataInfo(doc);
|
oafs.add(p);
|
||||||
final long lastUpdateTimestamp = new Date().getTime();
|
break;
|
||||||
|
case "dataset":
|
||||||
return createOafs(doc, type, collectedFrom, hostedBy, info, lastUpdateTimestamp);
|
final Dataset d = new Dataset();
|
||||||
} catch (final Exception e) {
|
populateResultFields(d, doc, collectedFrom, hostedBy, info, lastUpdateTimestamp);
|
||||||
throw new RuntimeException(e);
|
d.setResulttype(MigrationConstants.DATASET_RESULTTYPE_QUALIFIER);
|
||||||
}
|
d.setStoragedate(prepareDatasetStorageDate(doc, info));
|
||||||
}
|
d.setDevice(prepareDatasetDevice(doc, info));
|
||||||
|
d.setSize(prepareDatasetSize(doc, info));
|
||||||
protected List<Oaf> createOafs(
|
d.setVersion(prepareDatasetVersion(doc, info));
|
||||||
final Document doc,
|
d.setLastmetadataupdate(prepareDatasetLastMetadataUpdate(doc, info));
|
||||||
final String type,
|
d.setMetadataversionnumber(prepareDatasetMetadataVersionNumber(doc, info));
|
||||||
final KeyValue collectedFrom,
|
d.setGeolocation(prepareDatasetGeoLocations(doc, info));
|
||||||
final KeyValue hostedBy,
|
oafs.add(d);
|
||||||
final DataInfo info,
|
break;
|
||||||
final long lastUpdateTimestamp) {
|
case "software":
|
||||||
|
final Software s = new Software();
|
||||||
final List<Oaf> oafs = new ArrayList<>();
|
populateResultFields(s, doc, collectedFrom, hostedBy, info, lastUpdateTimestamp);
|
||||||
|
s.setResulttype(MigrationConstants.SOFTWARE_RESULTTYPE_QUALIFIER);
|
||||||
switch (type.toLowerCase()) {
|
s.setDocumentationUrl(prepareSoftwareDocumentationUrls(doc, info));
|
||||||
case "":
|
s.setLicense(prepareSoftwareLicenses(doc, info));
|
||||||
case "publication":
|
s.setCodeRepositoryUrl(prepareSoftwareCodeRepositoryUrl(doc, info));
|
||||||
final Publication p = new Publication();
|
s.setProgrammingLanguage(prepareSoftwareProgrammingLanguage(doc, info));
|
||||||
populateResultFields(p, doc, collectedFrom, hostedBy, info, lastUpdateTimestamp);
|
oafs.add(s);
|
||||||
p.setResulttype(PUBLICATION_RESULTTYPE_QUALIFIER);
|
break;
|
||||||
p.setJournal(prepareJournal(doc, info));
|
case "otherresearchproducts":
|
||||||
oafs.add(p);
|
default:
|
||||||
break;
|
final OtherResearchProduct o = new OtherResearchProduct();
|
||||||
case "dataset":
|
populateResultFields(o, doc, collectedFrom, hostedBy, info, lastUpdateTimestamp);
|
||||||
final Dataset d = new Dataset();
|
o.setResulttype(MigrationConstants.OTHER_RESULTTYPE_QUALIFIER);
|
||||||
populateResultFields(d, doc, collectedFrom, hostedBy, info, lastUpdateTimestamp);
|
o.setContactperson(prepareOtherResearchProductContactPersons(doc, info));
|
||||||
d.setResulttype(DATASET_RESULTTYPE_QUALIFIER);
|
o.setContactgroup(prepareOtherResearchProductContactGroups(doc, info));
|
||||||
d.setStoragedate(prepareDatasetStorageDate(doc, info));
|
o.setTool(prepareOtherResearchProductTools(doc, info));
|
||||||
d.setDevice(prepareDatasetDevice(doc, info));
|
oafs.add(o);
|
||||||
d.setSize(prepareDatasetSize(doc, info));
|
break;
|
||||||
d.setVersion(prepareDatasetVersion(doc, info));
|
}
|
||||||
d.setLastmetadataupdate(prepareDatasetLastMetadataUpdate(doc, info));
|
|
||||||
d.setMetadataversionnumber(prepareDatasetMetadataVersionNumber(doc, info));
|
if (!oafs.isEmpty()) {
|
||||||
d.setGeolocation(prepareDatasetGeoLocations(doc, info));
|
oafs.addAll(addProjectRels(doc, collectedFrom, info, lastUpdateTimestamp));
|
||||||
oafs.add(d);
|
oafs.addAll(addOtherResultRels(doc, collectedFrom, info, lastUpdateTimestamp));
|
||||||
break;
|
}
|
||||||
case "software":
|
|
||||||
final Software s = new Software();
|
return oafs;
|
||||||
populateResultFields(s, doc, collectedFrom, hostedBy, info, lastUpdateTimestamp);
|
}
|
||||||
s.setResulttype(SOFTWARE_RESULTTYPE_QUALIFIER);
|
|
||||||
s.setDocumentationUrl(prepareSoftwareDocumentationUrls(doc, info));
|
private List<Oaf> addProjectRels(
|
||||||
s.setLicense(prepareSoftwareLicenses(doc, info));
|
final Document doc,
|
||||||
s.setCodeRepositoryUrl(prepareSoftwareCodeRepositoryUrl(doc, info));
|
final KeyValue collectedFrom,
|
||||||
s.setProgrammingLanguage(prepareSoftwareProgrammingLanguage(doc, info));
|
final DataInfo info,
|
||||||
oafs.add(s);
|
final long lastUpdateTimestamp) {
|
||||||
break;
|
|
||||||
case "otherresearchproducts":
|
final List<Oaf> res = new ArrayList<>();
|
||||||
default:
|
|
||||||
final OtherResearchProduct o = new OtherResearchProduct();
|
final String docId = createOpenaireId(50, doc.valueOf("//dri:objIdentifier"), false);
|
||||||
populateResultFields(o, doc, collectedFrom, hostedBy, info, lastUpdateTimestamp);
|
|
||||||
o.setResulttype(OTHER_RESULTTYPE_QUALIFIER);
|
for (final Object o : doc.selectNodes("//oaf:projectid")) {
|
||||||
o.setContactperson(prepareOtherResearchProductContactPersons(doc, info));
|
final String projectId = createOpenaireId(40, ((Node) o).getText(), true);
|
||||||
o.setContactgroup(prepareOtherResearchProductContactGroups(doc, info));
|
|
||||||
o.setTool(prepareOtherResearchProductTools(doc, info));
|
final Relation r1 = new Relation();
|
||||||
oafs.add(o);
|
r1.setRelType("resultProject");
|
||||||
break;
|
r1.setSubRelType("outcome");
|
||||||
}
|
r1.setRelClass("isProducedBy");
|
||||||
|
r1.setSource(docId);
|
||||||
if (!oafs.isEmpty()) {
|
r1.setTarget(projectId);
|
||||||
oafs.addAll(addProjectRels(doc, collectedFrom, info, lastUpdateTimestamp));
|
r1.setCollectedfrom(Arrays.asList(collectedFrom));
|
||||||
oafs.addAll(addOtherResultRels(doc, collectedFrom, info, lastUpdateTimestamp));
|
r1.setDataInfo(info);
|
||||||
}
|
r1.setLastupdatetimestamp(lastUpdateTimestamp);
|
||||||
|
res.add(r1);
|
||||||
return oafs;
|
|
||||||
}
|
final Relation r2 = new Relation();
|
||||||
|
r2.setRelType("resultProject");
|
||||||
private List<Oaf> addProjectRels(
|
r2.setSubRelType("outcome");
|
||||||
final Document doc,
|
r2.setRelClass("produces");
|
||||||
final KeyValue collectedFrom,
|
r2.setSource(projectId);
|
||||||
final DataInfo info,
|
r2.setTarget(docId);
|
||||||
final long lastUpdateTimestamp) {
|
r2.setCollectedfrom(Arrays.asList(collectedFrom));
|
||||||
|
r2.setDataInfo(info);
|
||||||
final List<Oaf> res = new ArrayList<>();
|
r2.setLastupdatetimestamp(lastUpdateTimestamp);
|
||||||
|
res.add(r2);
|
||||||
final String docId = createOpenaireId(50, doc.valueOf("//dri:objIdentifier"), false);
|
}
|
||||||
|
|
||||||
for (final Object o : doc.selectNodes("//oaf:projectid")) {
|
return res;
|
||||||
final String projectId = createOpenaireId(40, ((Node) o).getText(), true);
|
}
|
||||||
|
|
||||||
final Relation r1 = new Relation();
|
protected abstract List<Oaf> addOtherResultRels(
|
||||||
r1.setRelType("resultProject");
|
final Document doc,
|
||||||
r1.setSubRelType("outcome");
|
final KeyValue collectedFrom,
|
||||||
r1.setRelClass("isProducedBy");
|
final DataInfo info,
|
||||||
r1.setSource(docId);
|
final long lastUpdateTimestamp);
|
||||||
r1.setTarget(projectId);
|
|
||||||
r1.setCollectedfrom(Arrays.asList(collectedFrom));
|
private void populateResultFields(
|
||||||
r1.setDataInfo(info);
|
final Result r,
|
||||||
r1.setLastupdatetimestamp(lastUpdateTimestamp);
|
final Document doc,
|
||||||
res.add(r1);
|
final KeyValue collectedFrom,
|
||||||
|
final KeyValue hostedBy,
|
||||||
final Relation r2 = new Relation();
|
final DataInfo info,
|
||||||
r2.setRelType("resultProject");
|
final long lastUpdateTimestamp) {
|
||||||
r2.setSubRelType("outcome");
|
r.setDataInfo(info);
|
||||||
r2.setRelClass("produces");
|
r.setLastupdatetimestamp(lastUpdateTimestamp);
|
||||||
r2.setSource(projectId);
|
r.setId(createOpenaireId(50, doc.valueOf("//dri:objIdentifier"), false));
|
||||||
r2.setTarget(docId);
|
r.setOriginalId(Arrays.asList(doc.valueOf("//dri:objIdentifier")));
|
||||||
r2.setCollectedfrom(Arrays.asList(collectedFrom));
|
r.setCollectedfrom(Arrays.asList(collectedFrom));
|
||||||
r2.setDataInfo(info);
|
r.setPid(prepareListStructProps(doc, "//oaf:identifier", "@identifierType", "dnet:pid_types", "dnet:pid_types", info));
|
||||||
r2.setLastupdatetimestamp(lastUpdateTimestamp);
|
r.setDateofcollection(doc.valueOf("//dr:dateOfCollection"));
|
||||||
res.add(r2);
|
r.setDateoftransformation(doc.valueOf("//dr:dateOfTransformation"));
|
||||||
}
|
r.setExtraInfo(new ArrayList<>()); // NOT PRESENT IN MDSTORES
|
||||||
|
r.setOaiprovenance(prepareOAIprovenance(doc));
|
||||||
return res;
|
r.setAuthor(prepareAuthors(doc, info));
|
||||||
}
|
r.setLanguage(prepareLanguages(doc));
|
||||||
|
r.setCountry(new ArrayList<>()); // NOT PRESENT IN MDSTORES
|
||||||
protected abstract List<Oaf> addOtherResultRels(
|
r.setSubject(prepareSubjects(doc, info));
|
||||||
final Document doc,
|
r.setTitle(prepareTitles(doc, info));
|
||||||
final KeyValue collectedFrom,
|
r.setRelevantdate(prepareRelevantDates(doc, info));
|
||||||
final DataInfo info,
|
r.setDescription(prepareDescriptions(doc, info));
|
||||||
final long lastUpdateTimestamp);
|
r.setDateofacceptance(prepareField(doc, "//oaf:dateAccepted", info));
|
||||||
|
r.setPublisher(preparePublisher(doc, info));
|
||||||
private void populateResultFields(
|
r.setEmbargoenddate(prepareField(doc, "//oaf:embargoenddate", info));
|
||||||
final Result r,
|
r.setSource(prepareSources(doc, info));
|
||||||
final Document doc,
|
r.setFulltext(new ArrayList<>()); // NOT PRESENT IN MDSTORES
|
||||||
final KeyValue collectedFrom,
|
r.setFormat(prepareFormats(doc, info));
|
||||||
final KeyValue hostedBy,
|
r.setContributor(prepareContributors(doc, info));
|
||||||
final DataInfo info,
|
r.setResourcetype(prepareResourceType(doc, info));
|
||||||
final long lastUpdateTimestamp) {
|
r.setCoverage(prepareCoverages(doc, info));
|
||||||
r.setDataInfo(info);
|
r.setContext(new ArrayList<>()); // NOT PRESENT IN MDSTORES
|
||||||
r.setLastupdatetimestamp(lastUpdateTimestamp);
|
r.setExternalReference(new ArrayList<>()); // NOT PRESENT IN MDSTORES
|
||||||
r.setId(createOpenaireId(50, doc.valueOf("//dri:objIdentifier"), false));
|
r.setInstance(prepareInstances(doc, info, collectedFrom, hostedBy));
|
||||||
r.setOriginalId(Arrays.asList(doc.valueOf("//dri:objIdentifier")));
|
}
|
||||||
r.setCollectedfrom(Arrays.asList(collectedFrom));
|
|
||||||
r.setPid(
|
protected abstract Qualifier prepareResourceType(Document doc, DataInfo info);
|
||||||
prepareListStructProps(
|
|
||||||
doc,
|
protected abstract List<Instance> prepareInstances(
|
||||||
"//oaf:identifier",
|
Document doc,
|
||||||
"@identifierType",
|
DataInfo info,
|
||||||
"dnet:pid_types",
|
KeyValue collectedfrom,
|
||||||
"dnet:pid_types",
|
KeyValue hostedby);
|
||||||
info));
|
|
||||||
r.setDateofcollection(doc.valueOf("//dr:dateOfCollection"));
|
protected abstract List<Field<String>> prepareSources(Document doc, DataInfo info);
|
||||||
r.setDateoftransformation(doc.valueOf("//dr:dateOfTransformation"));
|
|
||||||
r.setExtraInfo(new ArrayList<>()); // NOT PRESENT IN MDSTORES
|
protected abstract List<StructuredProperty> prepareRelevantDates(Document doc, DataInfo info);
|
||||||
r.setOaiprovenance(prepareOAIprovenance(doc));
|
|
||||||
r.setAuthor(prepareAuthors(doc, info));
|
protected abstract List<Field<String>> prepareCoverages(Document doc, DataInfo info);
|
||||||
r.setLanguage(prepareLanguages(doc));
|
|
||||||
r.setCountry(new ArrayList<>()); // NOT PRESENT IN MDSTORES
|
protected abstract List<Field<String>> prepareContributors(Document doc, DataInfo info);
|
||||||
r.setSubject(prepareSubjects(doc, info));
|
|
||||||
r.setTitle(prepareTitles(doc, info));
|
protected abstract List<Field<String>> prepareFormats(Document doc, DataInfo info);
|
||||||
r.setRelevantdate(prepareRelevantDates(doc, info));
|
|
||||||
r.setDescription(prepareDescriptions(doc, info));
|
protected abstract Field<String> preparePublisher(Document doc, DataInfo info);
|
||||||
r.setDateofacceptance(prepareField(doc, "//oaf:dateAccepted", info));
|
|
||||||
r.setPublisher(preparePublisher(doc, info));
|
protected abstract List<Field<String>> prepareDescriptions(Document doc, DataInfo info);
|
||||||
r.setEmbargoenddate(prepareField(doc, "//oaf:embargoenddate", info));
|
|
||||||
r.setSource(prepareSources(doc, info));
|
protected abstract List<StructuredProperty> prepareTitles(Document doc, DataInfo info);
|
||||||
r.setFulltext(new ArrayList<>()); // NOT PRESENT IN MDSTORES
|
|
||||||
r.setFormat(prepareFormats(doc, info));
|
protected abstract List<StructuredProperty> prepareSubjects(Document doc, DataInfo info);
|
||||||
r.setContributor(prepareContributors(doc, info));
|
|
||||||
r.setResourcetype(prepareResourceType(doc, info));
|
protected abstract Qualifier prepareLanguages(Document doc);
|
||||||
r.setCoverage(prepareCoverages(doc, info));
|
|
||||||
r.setContext(new ArrayList<>()); // NOT PRESENT IN MDSTORES
|
protected abstract List<Author> prepareAuthors(Document doc, DataInfo info);
|
||||||
r.setExternalReference(new ArrayList<>()); // NOT PRESENT IN MDSTORES
|
|
||||||
r.setInstance(prepareInstances(doc, info, collectedFrom, hostedBy));
|
protected abstract List<Field<String>> prepareOtherResearchProductTools(
|
||||||
}
|
Document doc,
|
||||||
|
DataInfo info);
|
||||||
protected abstract Qualifier prepareResourceType(Document doc, DataInfo info);
|
|
||||||
|
protected abstract List<Field<String>> prepareOtherResearchProductContactGroups(
|
||||||
protected abstract List<Instance> prepareInstances(
|
Document doc,
|
||||||
Document doc, DataInfo info, KeyValue collectedfrom, KeyValue hostedby);
|
DataInfo info);
|
||||||
|
|
||||||
protected abstract List<Field<String>> prepareSources(Document doc, DataInfo info);
|
protected abstract List<Field<String>> prepareOtherResearchProductContactPersons(
|
||||||
|
Document doc,
|
||||||
protected abstract List<StructuredProperty> prepareRelevantDates(Document doc, DataInfo info);
|
DataInfo info);
|
||||||
|
|
||||||
protected abstract List<Field<String>> prepareCoverages(Document doc, DataInfo info);
|
protected abstract Qualifier prepareSoftwareProgrammingLanguage(Document doc, DataInfo info);
|
||||||
|
|
||||||
protected abstract List<Field<String>> prepareContributors(Document doc, DataInfo info);
|
protected abstract Field<String> prepareSoftwareCodeRepositoryUrl(Document doc, DataInfo info);
|
||||||
|
|
||||||
protected abstract List<Field<String>> prepareFormats(Document doc, DataInfo info);
|
protected abstract List<StructuredProperty> prepareSoftwareLicenses(
|
||||||
|
Document doc,
|
||||||
protected abstract Field<String> preparePublisher(Document doc, DataInfo info);
|
DataInfo info);
|
||||||
|
|
||||||
protected abstract List<Field<String>> prepareDescriptions(Document doc, DataInfo info);
|
protected abstract List<Field<String>> prepareSoftwareDocumentationUrls(
|
||||||
|
Document doc,
|
||||||
protected abstract List<StructuredProperty> prepareTitles(Document doc, DataInfo info);
|
DataInfo info);
|
||||||
|
|
||||||
protected abstract List<StructuredProperty> prepareSubjects(Document doc, DataInfo info);
|
protected abstract List<GeoLocation> prepareDatasetGeoLocations(Document doc, DataInfo info);
|
||||||
|
|
||||||
protected abstract Qualifier prepareLanguages(Document doc);
|
protected abstract Field<String> prepareDatasetMetadataVersionNumber(
|
||||||
|
Document doc,
|
||||||
protected abstract List<Author> prepareAuthors(Document doc, DataInfo info);
|
DataInfo info);
|
||||||
|
|
||||||
protected abstract List<Field<String>> prepareOtherResearchProductTools(
|
protected abstract Field<String> prepareDatasetLastMetadataUpdate(Document doc, DataInfo info);
|
||||||
Document doc, DataInfo info);
|
|
||||||
|
protected abstract Field<String> prepareDatasetVersion(Document doc, DataInfo info);
|
||||||
protected abstract List<Field<String>> prepareOtherResearchProductContactGroups(
|
|
||||||
Document doc, DataInfo info);
|
protected abstract Field<String> prepareDatasetSize(Document doc, DataInfo info);
|
||||||
|
|
||||||
protected abstract List<Field<String>> prepareOtherResearchProductContactPersons(
|
protected abstract Field<String> prepareDatasetDevice(Document doc, DataInfo info);
|
||||||
Document doc, DataInfo info);
|
|
||||||
|
protected abstract Field<String> prepareDatasetStorageDate(Document doc, DataInfo info);
|
||||||
protected abstract Qualifier prepareSoftwareProgrammingLanguage(Document doc, DataInfo info);
|
|
||||||
|
private Journal prepareJournal(final Document doc, final DataInfo info) {
|
||||||
protected abstract Field<String> prepareSoftwareCodeRepositoryUrl(Document doc, DataInfo info);
|
final Node n = doc.selectSingleNode("//oaf:journal");
|
||||||
|
if (n != null) {
|
||||||
protected abstract List<StructuredProperty> prepareSoftwareLicenses(
|
final String name = n.getText();
|
||||||
Document doc, DataInfo info);
|
final String issnPrinted = n.valueOf("@issn");
|
||||||
|
final String issnOnline = n.valueOf("@eissn");
|
||||||
protected abstract List<Field<String>> prepareSoftwareDocumentationUrls(
|
final String issnLinking = n.valueOf("@lissn");
|
||||||
Document doc, DataInfo info);
|
final String ep = n.valueOf("@ep");
|
||||||
|
final String iss = n.valueOf("@iss");
|
||||||
protected abstract List<GeoLocation> prepareDatasetGeoLocations(Document doc, DataInfo info);
|
final String sp = n.valueOf("@sp");
|
||||||
|
final String vol = n.valueOf("@vol");
|
||||||
protected abstract Field<String> prepareDatasetMetadataVersionNumber(
|
final String edition = n.valueOf("@edition");
|
||||||
Document doc, DataInfo info);
|
if (StringUtils.isNotBlank(name)) { return journal(name, issnPrinted, issnOnline, issnLinking, ep, iss, sp, vol, edition, null, null, info); }
|
||||||
|
}
|
||||||
protected abstract Field<String> prepareDatasetLastMetadataUpdate(Document doc, DataInfo info);
|
return null;
|
||||||
|
}
|
||||||
protected abstract Field<String> prepareDatasetVersion(Document doc, DataInfo info);
|
|
||||||
|
protected Qualifier prepareQualifier(
|
||||||
protected abstract Field<String> prepareDatasetSize(Document doc, DataInfo info);
|
final Node node,
|
||||||
|
final String xpath,
|
||||||
protected abstract Field<String> prepareDatasetDevice(Document doc, DataInfo info);
|
final String schemeId,
|
||||||
|
final String schemeName) {
|
||||||
protected abstract Field<String> prepareDatasetStorageDate(Document doc, DataInfo info);
|
final String classId = node.valueOf(xpath);
|
||||||
|
final String className = code2name.get(classId);
|
||||||
private Journal prepareJournal(final Document doc, final DataInfo info) {
|
return qualifier(classId, className, schemeId, schemeName);
|
||||||
final Node n = doc.selectSingleNode("//oaf:journal");
|
}
|
||||||
if (n != null) {
|
|
||||||
final String name = n.getText();
|
protected List<StructuredProperty> prepareListStructProps(
|
||||||
final String issnPrinted = n.valueOf("@issn");
|
final Node node,
|
||||||
final String issnOnline = n.valueOf("@eissn");
|
final String xpath,
|
||||||
final String issnLinking = n.valueOf("@lissn");
|
final String xpathClassId,
|
||||||
final String ep = n.valueOf("@ep");
|
final String schemeId,
|
||||||
final String iss = n.valueOf("@iss");
|
final String schemeName,
|
||||||
final String sp = n.valueOf("@sp");
|
final DataInfo info) {
|
||||||
final String vol = n.valueOf("@vol");
|
final List<StructuredProperty> res = new ArrayList<>();
|
||||||
final String edition = n.valueOf("@edition");
|
for (final Object o : node.selectNodes(xpath)) {
|
||||||
if (StringUtils.isNotBlank(name)) {
|
final Node n = (Node) o;
|
||||||
return journal(
|
final String classId = n.valueOf(xpathClassId);
|
||||||
name,
|
final String className = code2name.get(classId);
|
||||||
issnPrinted,
|
res.add(structuredProperty(n.getText(), classId, className, schemeId, schemeName, info));
|
||||||
issnOnline,
|
}
|
||||||
issnLinking,
|
return res;
|
||||||
ep,
|
}
|
||||||
iss,
|
|
||||||
sp,
|
protected List<StructuredProperty> prepareListStructProps(
|
||||||
vol,
|
final Node node,
|
||||||
edition,
|
final String xpath,
|
||||||
null,
|
final Qualifier qualifier,
|
||||||
null,
|
final DataInfo info) {
|
||||||
info);
|
final List<StructuredProperty> res = new ArrayList<>();
|
||||||
}
|
for (final Object o : node.selectNodes(xpath)) {
|
||||||
}
|
final Node n = (Node) o;
|
||||||
return null;
|
res.add(structuredProperty(n.getText(), qualifier, info));
|
||||||
}
|
}
|
||||||
|
return res;
|
||||||
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 DataInfo info) {
|
||||||
|
final List<StructuredProperty> res = new ArrayList<>();
|
||||||
protected List<StructuredProperty> prepareListStructProps(
|
for (final Object o : node.selectNodes(xpath)) {
|
||||||
final Node node,
|
final Node n = (Node) o;
|
||||||
final String xpath,
|
res.add(structuredProperty(n.getText(), n.valueOf("@classid"), n.valueOf("@classname"), n.valueOf("@schemeid"), n.valueOf("@schemename"), info));
|
||||||
final String xpathClassId,
|
}
|
||||||
final String schemeId,
|
return res;
|
||||||
final String schemeName,
|
}
|
||||||
final DataInfo info) {
|
|
||||||
final List<StructuredProperty> res = new ArrayList<>();
|
protected OAIProvenance prepareOAIprovenance(final Document doc) {
|
||||||
for (final Object o : node.selectNodes(xpath)) {
|
final Node n =
|
||||||
final Node n = (Node) o;
|
doc.selectSingleNode("//*[local-name()='provenance']/*[local-name()='originDescription']");
|
||||||
final String classId = n.valueOf(xpathClassId);
|
|
||||||
final String className = code2name.get(classId);
|
if (n == null) { return null; }
|
||||||
res.add(
|
|
||||||
structuredProperty(
|
final String identifier = n.valueOf("./*[local-name()='identifier']");
|
||||||
n.getText(), classId, className, schemeId, schemeName, info));
|
final String baseURL = n.valueOf("./*[local-name()='baseURL']");;
|
||||||
}
|
final String metadataNamespace = n.valueOf("./*[local-name()='metadataNamespace']");;
|
||||||
return res;
|
final boolean altered = n.valueOf("@altered").equalsIgnoreCase("true");
|
||||||
}
|
final String datestamp = n.valueOf("./*[local-name()='datestamp']");;
|
||||||
|
final String harvestDate = n.valueOf("@harvestDate");;
|
||||||
protected List<StructuredProperty> prepareListStructProps(
|
|
||||||
final Node node, final String xpath, final Qualifier qualifier, final DataInfo info) {
|
return oaiIProvenance(identifier, baseURL, metadataNamespace, altered, datestamp, harvestDate);
|
||||||
final List<StructuredProperty> res = new ArrayList<>();
|
}
|
||||||
for (final Object o : node.selectNodes(xpath)) {
|
|
||||||
final Node n = (Node) o;
|
protected DataInfo prepareDataInfo(final Document doc) {
|
||||||
res.add(structuredProperty(n.getText(), qualifier, info));
|
final Node n = doc.selectSingleNode("//oaf:datainfo");
|
||||||
}
|
|
||||||
return res;
|
if (n == null) { return dataInfo(false, null, false, false, MigrationConstants.REPOSITORY_PROVENANCE_ACTIONS, "0.9"); }
|
||||||
}
|
|
||||||
|
final String paClassId = n.valueOf("./oaf:provenanceaction/@classid");
|
||||||
protected List<StructuredProperty> prepareListStructProps(
|
final String paClassName = n.valueOf("./oaf:provenanceaction/@classname");
|
||||||
final Node node, final String xpath, final DataInfo info) {
|
final String paSchemeId = n.valueOf("./oaf:provenanceaction/@schemeid");
|
||||||
final List<StructuredProperty> res = new ArrayList<>();
|
final String paSchemeName = n.valueOf("./oaf:provenanceaction/@schemename");
|
||||||
for (final Object o : node.selectNodes(xpath)) {
|
|
||||||
final Node n = (Node) o;
|
final boolean deletedbyinference =
|
||||||
res.add(
|
Boolean.parseBoolean(n.valueOf("./oaf:deletedbyinference"));
|
||||||
structuredProperty(
|
final String inferenceprovenance = n.valueOf("./oaf:inferenceprovenance");
|
||||||
n.getText(),
|
final Boolean inferred = Boolean.parseBoolean(n.valueOf("./oaf:inferred"));
|
||||||
n.valueOf("@classid"),
|
final String trust = n.valueOf("./oaf:trust");
|
||||||
n.valueOf("@classname"),
|
|
||||||
n.valueOf("@schemeid"),
|
return dataInfo(deletedbyinference, inferenceprovenance, inferred, false, qualifier(paClassId, paClassName, paSchemeId, paSchemeName), trust);
|
||||||
n.valueOf("@schemename"),
|
}
|
||||||
info));
|
|
||||||
}
|
protected Field<String> prepareField(final Node node, final String xpath, final DataInfo info) {
|
||||||
return res;
|
return field(node.valueOf(xpath), info);
|
||||||
}
|
}
|
||||||
|
|
||||||
protected OAIProvenance prepareOAIprovenance(final Document doc) {
|
protected List<Field<String>> prepareListFields(
|
||||||
final Node n =
|
final Node node,
|
||||||
doc.selectSingleNode(
|
final String xpath,
|
||||||
"//*[local-name()='provenance']/*[local-name()='originDescription']");
|
final DataInfo info) {
|
||||||
|
return listFields(info, prepareListString(node, xpath));
|
||||||
if (n == null) {
|
}
|
||||||
return null;
|
|
||||||
}
|
protected List<String> prepareListString(final Node node, final String xpath) {
|
||||||
|
final List<String> res = new ArrayList<>();
|
||||||
final String identifier = n.valueOf("./*[local-name()='identifier']");
|
for (final Object o : node.selectNodes(xpath)) {
|
||||||
final String baseURL = n.valueOf("./*[local-name()='baseURL']");
|
final String s = ((Node) o).getText().trim();
|
||||||
;
|
if (StringUtils.isNotBlank(s)) {
|
||||||
final String metadataNamespace = n.valueOf("./*[local-name()='metadataNamespace']");
|
res.add(s);
|
||||||
;
|
}
|
||||||
final boolean altered = n.valueOf("@altered").equalsIgnoreCase("true");
|
}
|
||||||
final String datestamp = n.valueOf("./*[local-name()='datestamp']");
|
return res;
|
||||||
;
|
}
|
||||||
final String harvestDate = n.valueOf("@harvestDate");
|
|
||||||
;
|
|
||||||
|
|
||||||
return oaiIProvenance(
|
|
||||||
identifier, baseURL, metadataNamespace, altered, datestamp, harvestDate);
|
|
||||||
}
|
|
||||||
|
|
||||||
protected DataInfo prepareDataInfo(final Document doc) {
|
|
||||||
final Node n = doc.selectSingleNode("//oaf:datainfo");
|
|
||||||
|
|
||||||
if (n == null) {
|
|
||||||
return dataInfo(false, null, false, false, REPOSITORY_QUALIFIER, "0.9");
|
|
||||||
}
|
|
||||||
|
|
||||||
final String paClassId = n.valueOf("./oaf:provenanceaction/@classid");
|
|
||||||
final String paClassName = n.valueOf("./oaf:provenanceaction/@classname");
|
|
||||||
final String paSchemeId = n.valueOf("./oaf:provenanceaction/@schemeid");
|
|
||||||
final String paSchemeName = n.valueOf("./oaf:provenanceaction/@schemename");
|
|
||||||
|
|
||||||
final boolean deletedbyinference =
|
|
||||||
Boolean.parseBoolean(n.valueOf("./oaf:deletedbyinference"));
|
|
||||||
final String inferenceprovenance = n.valueOf("./oaf:inferenceprovenance");
|
|
||||||
final Boolean inferred = Boolean.parseBoolean(n.valueOf("./oaf:inferred"));
|
|
||||||
final String trust = n.valueOf("./oaf:trust");
|
|
||||||
|
|
||||||
return dataInfo(
|
|
||||||
deletedbyinference,
|
|
||||||
inferenceprovenance,
|
|
||||||
inferred,
|
|
||||||
false,
|
|
||||||
qualifier(paClassId, paClassName, paSchemeId, paSchemeName),
|
|
||||||
trust);
|
|
||||||
}
|
|
||||||
|
|
||||||
protected Field<String> prepareField(final Node node, final String xpath, final DataInfo info) {
|
|
||||||
return field(node.valueOf(xpath), info);
|
|
||||||
}
|
|
||||||
|
|
||||||
protected List<Field<String>> prepareListFields(
|
|
||||||
final Node node, final String xpath, final DataInfo info) {
|
|
||||||
return listFields(info, prepareListString(node, xpath));
|
|
||||||
}
|
|
||||||
|
|
||||||
protected List<String> prepareListString(final Node node, final String xpath) {
|
|
||||||
final List<String> res = new ArrayList<>();
|
|
||||||
for (final Object o : node.selectNodes(xpath)) {
|
|
||||||
final String s = ((Node) o).getText().trim();
|
|
||||||
if (StringUtils.isNotBlank(s)) {
|
|
||||||
res.add(s);
|
|
||||||
}
|
|
||||||
}
|
|
||||||
return res;
|
|
||||||
}
|
|
||||||
}
|
}
|
||||||
|
|
|
||||||
File diff suppressed because it is too large
Load Diff
|
|
@ -0,0 +1,22 @@
|
||||||
|
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");
|
||||||
|
|
||||||
|
}
|
||||||
Loading…
Reference in New Issue