integrated changes from master

This commit is contained in:
Claudio Atzori 2020-04-28 08:55:28 +02:00
commit ac25f2d8d1
3 changed files with 925 additions and 913 deletions

View File

@ -1,4 +1,3 @@
package eu.dnetlib.dhp.oa.graph.raw;
import static eu.dnetlib.dhp.oa.graph.raw.common.OafMapperUtils.createOpenaireId;
@ -44,422 +43,426 @@ import org.dom4j.Node;
public abstract class AbstractMdRecordToOafMapper {
protected final Map<String, String> code2name;
protected static final Qualifier MAIN_TITLE_QUALIFIER = qualifier(
"main title", "main title",
"dnet:dataCite_title", "dnet:dataCite_title");
protected AbstractMdRecordToOafMapper(final Map<String, String> code2name) {
this.code2name = code2name;
}
public List<Oaf> processMdRecord(final String xml) {
try {
final Map<String, String> nsContext = new HashMap<>();
nsContext.put("dr", "http://www.driver-repository.eu/namespace/dr");
nsContext.put("dri", "http://www.driver-repository.eu/namespace/dri");
nsContext.put("oaf", "http://namespace.openaire.eu/oaf");
nsContext.put("oai", "http://www.openarchives.org/OAI/2.0/");
nsContext.put("prov", "http://www.openarchives.org/OAI/2.0/provenance");
nsContext.put("dc", "http://purl.org/dc/elements/1.1/");
nsContext.put("datacite", "http://datacite.org/schema/kernel-3");
DocumentFactory.getInstance().setXPathNamespaceURIs(nsContext);
final Document doc = DocumentHelper
.parseText(
xml
.replaceAll(
"http://datacite.org/schema/kernel-4", "http://datacite.org/schema/kernel-3"));
final String type = doc.valueOf("//dr:CobjCategory/@type");
final KeyValue collectedFrom = keyValue(
createOpenaireId(10, doc.valueOf("//oaf:collectedFrom/@id"), true),
doc.valueOf("//oaf:collectedFrom/@name"));
final KeyValue hostedBy = StringUtils.isBlank(doc.valueOf("//oaf:hostedBy/@id"))
? collectedFrom
: keyValue(
createOpenaireId(10, doc.valueOf("//oaf:hostedBy/@id"), true),
doc.valueOf("//oaf:hostedBy/@name"));
final DataInfo info = prepareDataInfo(doc);
final long lastUpdateTimestamp = new Date().getTime();
return createOafs(doc, type, collectedFrom, hostedBy, info, lastUpdateTimestamp);
} catch (final Exception e) {
throw new RuntimeException(e);
}
}
protected List<Oaf> createOafs(
final Document doc,
final String type,
final KeyValue collectedFrom,
final KeyValue hostedBy,
final DataInfo info,
final long lastUpdateTimestamp) {
final List<Oaf> oafs = new ArrayList<>();
switch (type.toLowerCase()) {
case "":
case "publication":
final Publication p = new Publication();
populateResultFields(p, doc, collectedFrom, hostedBy, info, lastUpdateTimestamp);
p.setResulttype(MigrationConstants.PUBLICATION_RESULTTYPE_QUALIFIER);
p.setJournal(prepareJournal(doc, info));
oafs.add(p);
break;
case "dataset":
final Dataset d = new Dataset();
populateResultFields(d, doc, collectedFrom, hostedBy, info, lastUpdateTimestamp);
d.setResulttype(MigrationConstants.DATASET_RESULTTYPE_QUALIFIER);
d.setStoragedate(prepareDatasetStorageDate(doc, info));
d.setDevice(prepareDatasetDevice(doc, info));
d.setSize(prepareDatasetSize(doc, info));
d.setVersion(prepareDatasetVersion(doc, info));
d.setLastmetadataupdate(prepareDatasetLastMetadataUpdate(doc, info));
d.setMetadataversionnumber(prepareDatasetMetadataVersionNumber(doc, info));
d.setGeolocation(prepareDatasetGeoLocations(doc, info));
oafs.add(d);
break;
case "software":
final Software s = new Software();
populateResultFields(s, doc, collectedFrom, hostedBy, info, lastUpdateTimestamp);
s.setResulttype(MigrationConstants.SOFTWARE_RESULTTYPE_QUALIFIER);
s.setDocumentationUrl(prepareSoftwareDocumentationUrls(doc, info));
s.setLicense(prepareSoftwareLicenses(doc, info));
s.setCodeRepositoryUrl(prepareSoftwareCodeRepositoryUrl(doc, info));
s.setProgrammingLanguage(prepareSoftwareProgrammingLanguage(doc, info));
oafs.add(s);
break;
case "otherresearchproducts":
default:
final OtherResearchProduct o = new OtherResearchProduct();
populateResultFields(o, doc, collectedFrom, hostedBy, info, lastUpdateTimestamp);
o.setResulttype(MigrationConstants.OTHER_RESULTTYPE_QUALIFIER);
o.setContactperson(prepareOtherResearchProductContactPersons(doc, info));
o.setContactgroup(prepareOtherResearchProductContactGroups(doc, info));
o.setTool(prepareOtherResearchProductTools(doc, info));
oafs.add(o);
break;
}
if (!oafs.isEmpty()) {
oafs.addAll(addProjectRels(doc, collectedFrom, info, lastUpdateTimestamp));
oafs.addAll(addOtherResultRels(doc, collectedFrom, info, lastUpdateTimestamp));
}
return oafs;
}
private List<Oaf> addProjectRels(
final Document doc,
final KeyValue collectedFrom,
final DataInfo info,
final long lastUpdateTimestamp) {
final List<Oaf> res = new ArrayList<>();
final String docId = createOpenaireId(50, doc.valueOf("//dri:objIdentifier"), false);
for (final Object o : doc.selectNodes("//oaf:projectid")) {
final String projectId = createOpenaireId(40, ((Node) o).getText(), true);
final Relation r1 = new Relation();
r1.setRelType("resultProject");
r1.setSubRelType("outcome");
r1.setRelClass("isProducedBy");
r1.setSource(docId);
r1.setTarget(projectId);
r1.setCollectedfrom(Arrays.asList(collectedFrom));
r1.setDataInfo(info);
r1.setLastupdatetimestamp(lastUpdateTimestamp);
res.add(r1);
final Relation r2 = new Relation();
r2.setRelType("resultProject");
r2.setSubRelType("outcome");
r2.setRelClass("produces");
r2.setSource(projectId);
r2.setTarget(docId);
r2.setCollectedfrom(Arrays.asList(collectedFrom));
r2.setDataInfo(info);
r2.setLastupdatetimestamp(lastUpdateTimestamp);
res.add(r2);
}
return res;
}
protected abstract List<Oaf> addOtherResultRels(
final Document doc,
final KeyValue collectedFrom,
final DataInfo info,
final long lastUpdateTimestamp);
private void populateResultFields(
final Result r,
final Document doc,
final KeyValue collectedFrom,
final KeyValue hostedBy,
final DataInfo info,
final long lastUpdateTimestamp) {
r.setDataInfo(info);
r.setLastupdatetimestamp(lastUpdateTimestamp);
r.setId(createOpenaireId(50, doc.valueOf("//dri:objIdentifier"), false));
r.setOriginalId(Arrays.asList(doc.valueOf("//dri:objIdentifier")));
r.setCollectedfrom(Arrays.asList(collectedFrom));
r
.setPid(
prepareListStructProps(
doc, "//oaf:identifier", "@identifierType", "dnet:pid_types", "dnet:pid_types", info));
r.setDateofcollection(doc.valueOf("//dr:dateOfCollection"));
r.setDateoftransformation(doc.valueOf("//dr:dateOfTransformation"));
r.setExtraInfo(new ArrayList<>()); // NOT PRESENT IN MDSTORES
r.setOaiprovenance(prepareOAIprovenance(doc));
r.setAuthor(prepareAuthors(doc, info));
r.setLanguage(prepareLanguages(doc));
r.setCountry(new ArrayList<>()); // NOT PRESENT IN MDSTORES
r.setSubject(prepareSubjects(doc, info));
r.setTitle(prepareTitles(doc, info));
r.setRelevantdate(prepareRelevantDates(doc, info));
r.setDescription(prepareDescriptions(doc, info));
r.setDateofacceptance(prepareField(doc, "//oaf:dateAccepted", info));
r.setPublisher(preparePublisher(doc, info));
r.setEmbargoenddate(prepareField(doc, "//oaf:embargoenddate", info));
r.setSource(prepareSources(doc, info));
r.setFulltext(new ArrayList<>()); // NOT PRESENT IN MDSTORES
r.setFormat(prepareFormats(doc, info));
r.setContributor(prepareContributors(doc, info));
r.setResourcetype(prepareResourceType(doc, info));
r.setCoverage(prepareCoverages(doc, info));
r.setContext(prepareContexts(doc, info));
r.setExternalReference(new ArrayList<>()); // NOT PRESENT IN MDSTORES
r.setInstance(prepareInstances(doc, info, collectedFrom, hostedBy));
}
private List<Context> prepareContexts(final Document doc, final DataInfo info) {
final List<Context> list = new ArrayList<>();
for (final Object o : doc.selectNodes("//oaf:concept")) {
final String cid = ((Node) o).valueOf("@id");
if (StringUtils.isNotBlank(cid)) {
final Context c = new Context();
c.setId(cid);
c.setDataInfo(Arrays.asList(info));
list.add(c);
}
}
return list;
}
protected abstract Qualifier prepareResourceType(Document doc, DataInfo info);
protected abstract List<Instance> prepareInstances(
Document doc, DataInfo info, KeyValue collectedfrom, KeyValue hostedby);
protected abstract List<Field<String>> prepareSources(Document doc, DataInfo info);
protected abstract List<StructuredProperty> prepareRelevantDates(Document doc, DataInfo info);
protected abstract List<Field<String>> prepareCoverages(Document doc, DataInfo info);
protected abstract List<Field<String>> prepareContributors(Document doc, DataInfo info);
protected abstract List<Field<String>> prepareFormats(Document doc, DataInfo info);
protected abstract Field<String> preparePublisher(Document doc, DataInfo info);
protected abstract List<Field<String>> prepareDescriptions(Document doc, DataInfo info);
protected abstract List<StructuredProperty> prepareTitles(Document doc, DataInfo info);
protected abstract List<StructuredProperty> prepareSubjects(Document doc, DataInfo info);
protected abstract Qualifier prepareLanguages(Document doc);
protected abstract List<Author> prepareAuthors(Document doc, DataInfo info);
protected abstract List<Field<String>> prepareOtherResearchProductTools(
Document doc, DataInfo info);
protected abstract List<Field<String>> prepareOtherResearchProductContactGroups(
Document doc, DataInfo info);
protected abstract List<Field<String>> prepareOtherResearchProductContactPersons(
Document doc, DataInfo info);
protected abstract Qualifier prepareSoftwareProgrammingLanguage(Document doc, DataInfo info);
protected abstract Field<String> prepareSoftwareCodeRepositoryUrl(Document doc, DataInfo info);
protected abstract List<StructuredProperty> prepareSoftwareLicenses(Document doc, DataInfo info);
protected abstract List<Field<String>> prepareSoftwareDocumentationUrls(
Document doc, DataInfo info);
protected abstract List<GeoLocation> prepareDatasetGeoLocations(Document doc, DataInfo info);
protected abstract Field<String> prepareDatasetMetadataVersionNumber(Document doc, DataInfo info);
protected abstract Field<String> prepareDatasetLastMetadataUpdate(Document doc, DataInfo info);
protected abstract Field<String> prepareDatasetVersion(Document doc, DataInfo info);
protected abstract Field<String> prepareDatasetSize(Document doc, DataInfo info);
protected abstract Field<String> prepareDatasetDevice(Document doc, DataInfo info);
protected abstract Field<String> prepareDatasetStorageDate(Document doc, DataInfo info);
private Journal prepareJournal(final Document doc, final DataInfo info) {
final Node n = doc.selectSingleNode("//oaf:journal");
if (n != null) {
final String name = n.getText();
final String issnPrinted = n.valueOf("@issn");
final String issnOnline = n.valueOf("@eissn");
final String issnLinking = n.valueOf("@lissn");
final String ep = n.valueOf("@ep");
final String iss = n.valueOf("@iss");
final String sp = n.valueOf("@sp");
final String vol = n.valueOf("@vol");
final String edition = n.valueOf("@edition");
if (StringUtils.isNotBlank(name)) {
return journal(
name,
issnPrinted,
issnOnline,
issnLinking,
ep,
iss,
sp,
vol,
edition,
null,
null,
info);
}
}
return null;
}
protected Qualifier prepareQualifier(
final Node node, final String xpath, final String schemeId, final String schemeName) {
final String classId = node.valueOf(xpath);
final String className = code2name.get(classId);
return qualifier(classId, className, schemeId, schemeName);
}
protected List<StructuredProperty> prepareListStructProps(
final Node node,
final String xpath,
final String xpathClassId,
final String schemeId,
final String schemeName,
final DataInfo info) {
final List<StructuredProperty> res = new ArrayList<>();
for (final Object o : node.selectNodes(xpath)) {
final Node n = (Node) o;
final String classId = n.valueOf(xpathClassId);
final String className = code2name.get(classId);
res.add(structuredProperty(n.getText(), classId, className, schemeId, schemeName, info));
}
return res;
}
protected List<StructuredProperty> prepareListStructProps(
final Node node, final String xpath, final Qualifier qualifier, final DataInfo info) {
final List<StructuredProperty> res = new ArrayList<>();
for (final Object o : node.selectNodes(xpath)) {
final Node n = (Node) o;
res.add(structuredProperty(n.getText(), qualifier, info));
}
return res;
}
protected List<StructuredProperty> prepareListStructProps(
final Node node, final String xpath, final DataInfo info) {
final List<StructuredProperty> res = new ArrayList<>();
for (final Object o : node.selectNodes(xpath)) {
final Node n = (Node) o;
res
.add(
structuredProperty(
n.getText(),
n.valueOf("@classid"),
n.valueOf("@classname"),
n.valueOf("@schemeid"),
n.valueOf("@schemename"),
info));
}
return res;
}
protected OAIProvenance prepareOAIprovenance(final Document doc) {
final Node n = doc.selectSingleNode("//*[local-name()='provenance']/*[local-name()='originDescription']");
if (n == null) {
return null;
}
final String identifier = n.valueOf("./*[local-name()='identifier']");
final String baseURL = n.valueOf("./*[local-name()='baseURL']");
;
final String metadataNamespace = n.valueOf("./*[local-name()='metadataNamespace']");
;
final boolean altered = n.valueOf("@altered").equalsIgnoreCase("true");
final String datestamp = n.valueOf("./*[local-name()='datestamp']");
;
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, MigrationConstants.REPOSITORY_PROVENANCE_ACTIONS, "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;
}
protected final Map<String, String> code2name;
protected static final Qualifier MAIN_TITLE_QUALIFIER =
qualifier("main title", "main title", "dnet:dataCite_title", "dnet:dataCite_title");
protected AbstractMdRecordToOafMapper(final Map<String, String> code2name) {
this.code2name = code2name;
}
public List<Oaf> processMdRecord(final String xml) {
try {
final Map<String, String> nsContext = new HashMap<>();
nsContext.put("dr", "http://www.driver-repository.eu/namespace/dr");
nsContext.put("dri", "http://www.driver-repository.eu/namespace/dri");
nsContext.put("oaf", "http://namespace.openaire.eu/oaf");
nsContext.put("oai", "http://www.openarchives.org/OAI/2.0/");
nsContext.put("prov", "http://www.openarchives.org/OAI/2.0/provenance");
nsContext.put("dc", "http://purl.org/dc/elements/1.1/");
nsContext.put("datacite", "http://datacite.org/schema/kernel-3");
DocumentFactory.getInstance().setXPathNamespaceURIs(nsContext);
final Document doc =
DocumentHelper.parseText(
xml.replaceAll(
"http://datacite.org/schema/kernel-4", "http://datacite.org/schema/kernel-3"));
final String type = doc.valueOf("//dr:CobjCategory/@type");
final KeyValue collectedFrom =
keyValue(
createOpenaireId(10, doc.valueOf("//oaf:collectedFrom/@id"), true),
doc.valueOf("//oaf:collectedFrom/@name"));
final KeyValue hostedBy =
StringUtils.isBlank(doc.valueOf("//oaf:hostedBy/@id"))
? collectedFrom
: keyValue(
createOpenaireId(10, doc.valueOf("//oaf:hostedBy/@id"), true),
doc.valueOf("//oaf:hostedBy/@name"));
final DataInfo info = prepareDataInfo(doc);
final long lastUpdateTimestamp = new Date().getTime();
return createOafs(doc, type, collectedFrom, hostedBy, info, lastUpdateTimestamp);
} catch (final Exception e) {
throw new RuntimeException(e);
}
}
protected List<Oaf> createOafs(
final Document doc,
final String type,
final KeyValue collectedFrom,
final KeyValue hostedBy,
final DataInfo info,
final long lastUpdateTimestamp) {
final List<Oaf> oafs = new ArrayList<>();
switch (type.toLowerCase()) {
case "":
case "publication":
final Publication p = new Publication();
populateResultFields(p, doc, collectedFrom, hostedBy, info, lastUpdateTimestamp);
p.setResulttype(MigrationConstants.PUBLICATION_RESULTTYPE_QUALIFIER);
p.setJournal(prepareJournal(doc, info));
oafs.add(p);
break;
case "dataset":
final Dataset d = new Dataset();
populateResultFields(d, doc, collectedFrom, hostedBy, info, lastUpdateTimestamp);
d.setResulttype(MigrationConstants.DATASET_RESULTTYPE_QUALIFIER);
d.setStoragedate(prepareDatasetStorageDate(doc, info));
d.setDevice(prepareDatasetDevice(doc, info));
d.setSize(prepareDatasetSize(doc, info));
d.setVersion(prepareDatasetVersion(doc, info));
d.setLastmetadataupdate(prepareDatasetLastMetadataUpdate(doc, info));
d.setMetadataversionnumber(prepareDatasetMetadataVersionNumber(doc, info));
d.setGeolocation(prepareDatasetGeoLocations(doc, info));
oafs.add(d);
break;
case "software":
final Software s = new Software();
populateResultFields(s, doc, collectedFrom, hostedBy, info, lastUpdateTimestamp);
s.setResulttype(MigrationConstants.SOFTWARE_RESULTTYPE_QUALIFIER);
s.setDocumentationUrl(prepareSoftwareDocumentationUrls(doc, info));
s.setLicense(prepareSoftwareLicenses(doc, info));
s.setCodeRepositoryUrl(prepareSoftwareCodeRepositoryUrl(doc, info));
s.setProgrammingLanguage(prepareSoftwareProgrammingLanguage(doc, info));
oafs.add(s);
break;
case "otherresearchproducts":
default:
final OtherResearchProduct o = new OtherResearchProduct();
populateResultFields(o, doc, collectedFrom, hostedBy, info, lastUpdateTimestamp);
o.setResulttype(MigrationConstants.OTHER_RESULTTYPE_QUALIFIER);
o.setContactperson(prepareOtherResearchProductContactPersons(doc, info));
o.setContactgroup(prepareOtherResearchProductContactGroups(doc, info));
o.setTool(prepareOtherResearchProductTools(doc, info));
oafs.add(o);
break;
}
if (!oafs.isEmpty()) {
oafs.addAll(addProjectRels(doc, collectedFrom, info, lastUpdateTimestamp));
oafs.addAll(addOtherResultRels(doc, collectedFrom, info, lastUpdateTimestamp));
}
return oafs;
}
private List<Oaf> addProjectRels(
final Document doc,
final KeyValue collectedFrom,
final DataInfo info,
final long lastUpdateTimestamp) {
final List<Oaf> res = new ArrayList<>();
final String docId = createOpenaireId(50, doc.valueOf("//dri:objIdentifier"), false);
for (final Object o : doc.selectNodes("//oaf:projectid")) {
final String originalId = ((Node) o).getText();
if (StringUtils.isNotBlank(originalId)) {
final String projectId = createOpenaireId(40, originalId, true);
final Relation r1 = new Relation();
r1.setRelType("resultProject");
r1.setSubRelType("outcome");
r1.setRelClass("isProducedBy");
r1.setSource(docId);
r1.setTarget(projectId);
r1.setCollectedfrom(Arrays.asList(collectedFrom));
r1.setDataInfo(info);
r1.setLastupdatetimestamp(lastUpdateTimestamp);
res.add(r1);
final Relation r2 = new Relation();
r2.setRelType("resultProject");
r2.setSubRelType("outcome");
r2.setRelClass("produces");
r2.setSource(projectId);
r2.setTarget(docId);
r2.setCollectedfrom(Arrays.asList(collectedFrom));
r2.setDataInfo(info);
r2.setLastupdatetimestamp(lastUpdateTimestamp);
res.add(r2);
}
}
return res;
}
protected abstract List<Oaf> addOtherResultRels(
final Document doc,
final KeyValue collectedFrom,
final DataInfo info,
final long lastUpdateTimestamp);
private void populateResultFields(
final Result r,
final Document doc,
final KeyValue collectedFrom,
final KeyValue hostedBy,
final DataInfo info,
final long lastUpdateTimestamp) {
r.setDataInfo(info);
r.setLastupdatetimestamp(lastUpdateTimestamp);
r.setId(createOpenaireId(50, doc.valueOf("//dri:objIdentifier"), false));
r.setOriginalId(Arrays.asList(doc.valueOf("//dri:objIdentifier")));
r.setCollectedfrom(Arrays.asList(collectedFrom));
r.setPid(
prepareListStructProps(
doc, "//oaf:identifier", "@identifierType", "dnet:pid_types", "dnet:pid_types", info));
r.setDateofcollection(doc.valueOf("//dr:dateOfCollection"));
r.setDateoftransformation(doc.valueOf("//dr:dateOfTransformation"));
r.setExtraInfo(new ArrayList<>()); // NOT PRESENT IN MDSTORES
r.setOaiprovenance(prepareOAIprovenance(doc));
r.setAuthor(prepareAuthors(doc, info));
r.setLanguage(prepareLanguages(doc));
r.setCountry(new ArrayList<>()); // NOT PRESENT IN MDSTORES
r.setSubject(prepareSubjects(doc, info));
r.setTitle(prepareTitles(doc, info));
r.setRelevantdate(prepareRelevantDates(doc, info));
r.setDescription(prepareDescriptions(doc, info));
r.setDateofacceptance(prepareField(doc, "//oaf:dateAccepted", info));
r.setPublisher(preparePublisher(doc, info));
r.setEmbargoenddate(prepareField(doc, "//oaf:embargoenddate", info));
r.setSource(prepareSources(doc, info));
r.setFulltext(new ArrayList<>()); // NOT PRESENT IN MDSTORES
r.setFormat(prepareFormats(doc, info));
r.setContributor(prepareContributors(doc, info));
r.setResourcetype(prepareResourceType(doc, info));
r.setCoverage(prepareCoverages(doc, info));
r.setContext(prepareContexts(doc, info));
r.setExternalReference(new ArrayList<>()); // NOT PRESENT IN MDSTORES
r.setInstance(prepareInstances(doc, info, collectedFrom, hostedBy));
}
private List<Context> prepareContexts(final Document doc, final DataInfo info) {
final List<Context> list = new ArrayList<>();
for (final Object o : doc.selectNodes("//oaf:concept")) {
final String cid = ((Node) o).valueOf("@id");
if (StringUtils.isNotBlank(cid)) {
final Context c = new Context();
c.setId(cid);
c.setDataInfo(Arrays.asList(info));
list.add(c);
}
}
return list;
}
protected abstract Qualifier prepareResourceType(Document doc, DataInfo info);
protected abstract List<Instance> prepareInstances(
Document doc, DataInfo info, KeyValue collectedfrom, KeyValue hostedby);
protected abstract List<Field<String>> prepareSources(Document doc, DataInfo info);
protected abstract List<StructuredProperty> prepareRelevantDates(Document doc, DataInfo info);
protected abstract List<Field<String>> prepareCoverages(Document doc, DataInfo info);
protected abstract List<Field<String>> prepareContributors(Document doc, DataInfo info);
protected abstract List<Field<String>> prepareFormats(Document doc, DataInfo info);
protected abstract Field<String> preparePublisher(Document doc, DataInfo info);
protected abstract List<Field<String>> prepareDescriptions(Document doc, DataInfo info);
protected abstract List<StructuredProperty> prepareTitles(Document doc, DataInfo info);
protected abstract List<StructuredProperty> prepareSubjects(Document doc, DataInfo info);
protected abstract Qualifier prepareLanguages(Document doc);
protected abstract List<Author> prepareAuthors(Document doc, DataInfo info);
protected abstract List<Field<String>> prepareOtherResearchProductTools(
Document doc, DataInfo info);
protected abstract List<Field<String>> prepareOtherResearchProductContactGroups(
Document doc, DataInfo info);
protected abstract List<Field<String>> prepareOtherResearchProductContactPersons(
Document doc, DataInfo info);
protected abstract Qualifier prepareSoftwareProgrammingLanguage(Document doc, DataInfo info);
protected abstract Field<String> prepareSoftwareCodeRepositoryUrl(Document doc, DataInfo info);
protected abstract List<StructuredProperty> prepareSoftwareLicenses(Document doc, DataInfo info);
protected abstract List<Field<String>> prepareSoftwareDocumentationUrls(
Document doc, DataInfo info);
protected abstract List<GeoLocation> prepareDatasetGeoLocations(Document doc, DataInfo info);
protected abstract Field<String> prepareDatasetMetadataVersionNumber(Document doc, DataInfo info);
protected abstract Field<String> prepareDatasetLastMetadataUpdate(Document doc, DataInfo info);
protected abstract Field<String> prepareDatasetVersion(Document doc, DataInfo info);
protected abstract Field<String> prepareDatasetSize(Document doc, DataInfo info);
protected abstract Field<String> prepareDatasetDevice(Document doc, DataInfo info);
protected abstract Field<String> prepareDatasetStorageDate(Document doc, DataInfo info);
private Journal prepareJournal(final Document doc, final DataInfo info) {
final Node n = doc.selectSingleNode("//oaf:journal");
if (n != null) {
final String name = n.getText();
final String issnPrinted = n.valueOf("@issn");
final String issnOnline = n.valueOf("@eissn");
final String issnLinking = n.valueOf("@lissn");
final String ep = n.valueOf("@ep");
final String iss = n.valueOf("@iss");
final String sp = n.valueOf("@sp");
final String vol = n.valueOf("@vol");
final String edition = n.valueOf("@edition");
if (StringUtils.isNotBlank(name)) {
return journal(
name,
issnPrinted,
issnOnline,
issnLinking,
ep,
iss,
sp,
vol,
edition,
null,
null,
info);
}
}
return null;
}
protected Qualifier prepareQualifier(
final Node node, final String xpath, final String schemeId, final String schemeName) {
final String classId = node.valueOf(xpath);
final String className = code2name.get(classId);
return qualifier(classId, className, schemeId, schemeName);
}
protected List<StructuredProperty> prepareListStructProps(
final Node node,
final String xpath,
final String xpathClassId,
final String schemeId,
final String schemeName,
final DataInfo info) {
final List<StructuredProperty> res = new ArrayList<>();
for (final Object o : node.selectNodes(xpath)) {
final Node n = (Node) o;
final String classId = n.valueOf(xpathClassId);
final String className = code2name.get(classId);
res.add(structuredProperty(n.getText(), classId, className, schemeId, schemeName, info));
}
return res;
}
protected List<StructuredProperty> prepareListStructProps(
final Node node, final String xpath, final Qualifier qualifier, final DataInfo info) {
final List<StructuredProperty> res = new ArrayList<>();
for (final Object o : node.selectNodes(xpath)) {
final Node n = (Node) o;
res.add(structuredProperty(n.getText(), qualifier, info));
}
return res;
}
protected List<StructuredProperty> prepareListStructProps(
final Node node, final String xpath, final DataInfo info) {
final List<StructuredProperty> res = new ArrayList<>();
for (final Object o : node.selectNodes(xpath)) {
final Node n = (Node) o;
res.add(
structuredProperty(
n.getText(),
n.valueOf("@classid"),
n.valueOf("@classname"),
n.valueOf("@schemeid"),
n.valueOf("@schemename"),
info));
}
return res;
}
protected OAIProvenance prepareOAIprovenance(final Document doc) {
final Node n =
doc.selectSingleNode("//*[local-name()='provenance']/*[local-name()='originDescription']");
if (n == null) {
return null;
}
final String identifier = n.valueOf("./*[local-name()='identifier']");
final String baseURL = n.valueOf("./*[local-name()='baseURL']");
;
final String metadataNamespace = n.valueOf("./*[local-name()='metadataNamespace']");
;
final boolean altered = n.valueOf("@altered").equalsIgnoreCase("true");
final String datestamp = n.valueOf("./*[local-name()='datestamp']");
;
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, MigrationConstants.REPOSITORY_PROVENANCE_ACTIONS, "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;
}
}

View File

@ -1,256 +1,267 @@
package eu.dnetlib.dhp.oa.graph.raw;
import static eu.dnetlib.dhp.oa.graph.raw.common.OafMapperUtils.createOpenaireId;
import static eu.dnetlib.dhp.oa.graph.raw.common.OafMapperUtils.field;
import 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.Arrays;
import java.util.List;
import java.util.Map;
import org.apache.commons.lang3.StringUtils;
import org.dom4j.Document;
import org.dom4j.Node;
public class OafToOafMapper extends AbstractMdRecordToOafMapper {
public OafToOafMapper(final Map<String, String> code2name) {
super(code2name);
}
public OafToOafMapper(final Map<String, String> code2name) {
super(code2name);
}
@Override
protected List<Author> prepareAuthors(final Document doc, final DataInfo info) {
final List<Author> res = new ArrayList<>();
int pos = 1;
for (final Object o : doc.selectNodes("//dc:creator")) {
final Node n = (Node) o;
final Author author = new Author();
author.setFullname(n.getText());
author.setRank(pos++);
final PacePerson p = new PacePerson(n.getText(), false);
if (p.isAccurate()) {
author.setName(p.getNormalisedFirstName());
author.setSurname(p.getNormalisedSurname());
}
res.add(author);
}
return res;
}
@Override
protected List<Author> prepareAuthors(final Document doc, final DataInfo info) {
final List<Author> res = new ArrayList<>();
int pos = 1;
for (final Object o : doc.selectNodes("//dc:creator")) {
final Node n = (Node) o;
final Author author = new Author();
author.setFullname(n.getText());
author.setRank(pos++);
final PacePerson p = new PacePerson(n.getText(), false);
if (p.isAccurate()) {
author.setName(p.getNormalisedFirstName());
author.setSurname(p.getNormalisedSurname());
}
res.add(author);
}
return res;
}
@Override
protected Qualifier prepareLanguages(final Document doc) {
return prepareQualifier(doc, "//dc:language", "dnet:languages", "dnet:languages");
}
@Override
protected Qualifier prepareLanguages(final Document doc) {
return prepareQualifier(doc, "//dc:language", "dnet:languages", "dnet:languages");
}
@Override
protected List<StructuredProperty> prepareSubjects(final Document doc, final DataInfo info) {
return prepareListStructProps(doc, "//dc:subject", info);
}
@Override
protected List<StructuredProperty> prepareSubjects(final Document doc, final DataInfo info) {
return prepareListStructProps(doc, "//dc:subject", info);
}
@Override
protected List<StructuredProperty> prepareTitles(final Document doc, final DataInfo info) {
return prepareListStructProps(doc, "//dc:title", MAIN_TITLE_QUALIFIER, info);
}
@Override
protected List<StructuredProperty> prepareTitles(final Document doc, final DataInfo info) {
return prepareListStructProps(doc, "//dc:title", MAIN_TITLE_QUALIFIER, info);
}
@Override
protected List<Field<String>> prepareDescriptions(final Document doc, final DataInfo info) {
return prepareListFields(doc, "//dc:description", info);
}
@Override
protected List<Field<String>> prepareDescriptions(final Document doc, final DataInfo info) {
return prepareListFields(doc, "//dc:description", info);
}
@Override
protected Field<String> preparePublisher(final Document doc, final DataInfo info) {
return prepareField(doc, "//dc:publisher", info);
}
@Override
protected Field<String> preparePublisher(final Document doc, final DataInfo info) {
return prepareField(doc, "//dc:publisher", info);
}
@Override
protected List<Field<String>> prepareFormats(final Document doc, final DataInfo info) {
return prepareListFields(doc, "//dc:format", info);
}
@Override
protected List<Field<String>> prepareFormats(final Document doc, final DataInfo info) {
return prepareListFields(doc, "//dc:format", info);
}
@Override
protected List<Field<String>> prepareContributors(final Document doc, final DataInfo info) {
return prepareListFields(doc, "//dc:contributor", info);
}
@Override
protected List<Field<String>> prepareContributors(final Document doc, final DataInfo info) {
return prepareListFields(doc, "//dc:contributor", info);
}
@Override
protected List<Field<String>> prepareCoverages(final Document doc, final DataInfo info) {
return prepareListFields(doc, "//dc:coverage", info);
}
@Override
protected List<Field<String>> prepareCoverages(final Document doc, final DataInfo info) {
return prepareListFields(doc, "//dc:coverage", info);
}
@Override
protected List<Instance> prepareInstances(
final Document doc,
final DataInfo info,
final KeyValue collectedfrom,
final KeyValue hostedby) {
final List<Instance> res = new ArrayList<>();
for (final Object o : doc.selectNodes("//dc:identifier")) {
final String url = ((Node) o).getText().trim();
if (url.startsWith("http")) {
final Instance instance = new Instance();
instance.setUrl(Arrays.asList(url));
instance
.setInstancetype(
prepareQualifier(
doc,
"//dr:CobjCategory",
"dnet:publication_resource",
"dnet:publication_resource"));
instance.setCollectedfrom(collectedfrom);
instance.setHostedby(hostedby);
instance.setDateofacceptance(field(doc.valueOf("//oaf:dateAccepted"), info));
instance.setDistributionlocation(doc.valueOf("//oaf:distributionlocation"));
instance
.setAccessright(
prepareQualifier(doc, "//oaf:accessrights", "dnet:access_modes", "dnet:access_modes"));
instance.setLicense(field(doc.valueOf("//oaf:license"), info));
instance.setRefereed(field(doc.valueOf("//oaf:refereed"), info));
instance
.setProcessingchargeamount(
field(doc.valueOf("//oaf:processingchargeamount"), info));
instance
.setProcessingchargecurrency(
field(doc.valueOf("//oaf:processingchargeamount/@currency"), info));
res.add(instance);
}
}
return res;
}
@Override
protected List<Instance> prepareInstances(
final Document doc,
final DataInfo info,
final KeyValue collectedfrom,
final KeyValue hostedby) {
final List<Instance> res = new ArrayList<>();
for (final Object o : doc.selectNodes("//dc:identifier")) {
final String url = ((Node) o).getText().trim();
if (url.startsWith("http")) {
final Instance instance = new Instance();
instance.setUrl(Arrays.asList(url));
instance.setInstancetype(
prepareQualifier(
doc,
"//dr:CobjCategory",
"dnet:publication_resource",
"dnet:publication_resource"));
instance.setCollectedfrom(collectedfrom);
instance.setHostedby(hostedby);
instance.setDateofacceptance(field(doc.valueOf("//oaf:dateAccepted"), info));
instance.setDistributionlocation(doc.valueOf("//oaf:distributionlocation"));
instance.setAccessright(
prepareQualifier(doc, "//oaf:accessrights", "dnet:access_modes", "dnet:access_modes"));
instance.setLicense(field(doc.valueOf("//oaf:license"), info));
instance.setRefereed(field(doc.valueOf("//oaf:refereed"), info));
instance.setProcessingchargeamount(
field(doc.valueOf("//oaf:processingchargeamount"), info));
instance.setProcessingchargecurrency(
field(doc.valueOf("//oaf:processingchargeamount/@currency"), info));
res.add(instance);
}
}
return res;
}
@Override
protected List<Field<String>> prepareSources(final Document doc, final DataInfo info) {
return prepareListFields(doc, "//dc:source", info);
}
@Override
protected List<Field<String>> prepareSources(final Document doc, final DataInfo info) {
return prepareListFields(doc, "//dc:source", info);
}
@Override
protected List<StructuredProperty> prepareRelevantDates(final Document doc, final DataInfo info) {
return new ArrayList<>(); // NOT PRESENT IN OAF
}
@Override
protected List<StructuredProperty> prepareRelevantDates(final Document doc, final DataInfo info) {
return new ArrayList<>(); // NOT PRESENT IN OAF
}
// SOFTWARES
// SOFTWARES
@Override
protected Qualifier prepareSoftwareProgrammingLanguage(final Document doc, final DataInfo info) {
return null; // NOT PRESENT IN OAF
}
@Override
protected Qualifier prepareSoftwareProgrammingLanguage(final Document doc, final DataInfo info) {
return null; // NOT PRESENT IN OAF
}
@Override
protected Field<String> prepareSoftwareCodeRepositoryUrl(
final Document doc, final DataInfo info) {
return null; // NOT PRESENT IN OAF
}
@Override
protected Field<String> prepareSoftwareCodeRepositoryUrl(
final Document doc, final DataInfo info) {
return null; // NOT PRESENT IN OAF
}
@Override
protected List<StructuredProperty> prepareSoftwareLicenses(
final Document doc, final DataInfo info) {
return new ArrayList<>(); // NOT PRESENT IN OAF
}
@Override
protected List<StructuredProperty> prepareSoftwareLicenses(
final Document doc, final DataInfo info) {
return new ArrayList<>(); // NOT PRESENT IN OAF
}
@Override
protected List<Field<String>> prepareSoftwareDocumentationUrls(
final Document doc, final DataInfo info) {
return new ArrayList<>(); // NOT PRESENT IN OAF
}
@Override
protected List<Field<String>> prepareSoftwareDocumentationUrls(
final Document doc, final DataInfo info) {
return new ArrayList<>(); // NOT PRESENT IN OAF
}
// DATASETS
@Override
protected List<GeoLocation> prepareDatasetGeoLocations(final Document doc, final DataInfo info) {
return new ArrayList<>(); // NOT PRESENT IN OAF
}
// DATASETS
@Override
protected List<GeoLocation> prepareDatasetGeoLocations(final Document doc, final DataInfo info) {
return new ArrayList<>(); // NOT PRESENT IN OAF
}
@Override
protected Field<String> prepareDatasetMetadataVersionNumber(
final Document doc, final DataInfo info) {
return null; // NOT PRESENT IN OAF
}
@Override
protected Field<String> prepareDatasetMetadataVersionNumber(
final Document doc, final DataInfo info) {
return null; // NOT PRESENT IN OAF
}
@Override
protected Field<String> prepareDatasetLastMetadataUpdate(
final Document doc, final DataInfo info) {
return null; // NOT PRESENT IN OAF
}
@Override
protected Field<String> prepareDatasetLastMetadataUpdate(
final Document doc, final DataInfo info) {
return null; // NOT PRESENT IN OAF
}
@Override
protected Field<String> prepareDatasetVersion(final Document doc, final DataInfo info) {
return null; // NOT PRESENT IN OAF
}
@Override
protected Field<String> prepareDatasetVersion(final Document doc, final DataInfo info) {
return null; // NOT PRESENT IN OAF
}
@Override
protected Field<String> prepareDatasetSize(final Document doc, final DataInfo info) {
return null; // NOT PRESENT IN OAF
}
@Override
protected Field<String> prepareDatasetSize(final Document doc, final DataInfo info) {
return null; // NOT PRESENT IN OAF
}
@Override
protected Field<String> prepareDatasetDevice(final Document doc, final DataInfo info) {
return null; // NOT PRESENT IN OAF
}
@Override
protected Field<String> prepareDatasetDevice(final Document doc, final DataInfo info) {
return null; // NOT PRESENT IN OAF
}
@Override
protected Field<String> prepareDatasetStorageDate(final Document doc, final DataInfo info) {
return null; // NOT PRESENT IN OAF
}
@Override
protected Field<String> prepareDatasetStorageDate(final Document doc, final DataInfo info) {
return null; // NOT PRESENT IN OAF
}
// OTHER PRODUCTS
// OTHER PRODUCTS
@Override
protected List<Field<String>> prepareOtherResearchProductTools(
final Document doc, final DataInfo info) {
return new ArrayList<>(); // NOT PRESENT IN OAF
}
@Override
protected List<Field<String>> prepareOtherResearchProductTools(
final Document doc, final DataInfo info) {
return new ArrayList<>(); // NOT PRESENT IN OAF
}
@Override
protected List<Field<String>> prepareOtherResearchProductContactGroups(
final Document doc, final DataInfo info) {
return new ArrayList<>(); // NOT PRESENT IN OAF
}
@Override
protected List<Field<String>> prepareOtherResearchProductContactGroups(
final Document doc, final DataInfo info) {
return new ArrayList<>(); // NOT PRESENT IN OAF
}
@Override
protected List<Field<String>> prepareOtherResearchProductContactPersons(
final Document doc, final DataInfo info) {
return new ArrayList<>(); // NOT PRESENT IN OAF
}
@Override
protected List<Field<String>> prepareOtherResearchProductContactPersons(
final Document doc, final DataInfo info) {
return new ArrayList<>(); // NOT PRESENT IN OAF
}
@Override
protected List<Oaf> addOtherResultRels(
final Document doc,
final KeyValue collectedFrom,
final DataInfo info,
final long lastUpdateTimestamp) {
final String docId = createOpenaireId(50, doc.valueOf("//dri:objIdentifier"), false);
@Override
protected List<Oaf> addOtherResultRels(
final Document doc,
final KeyValue collectedFrom,
final DataInfo info,
final long lastUpdateTimestamp) {
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']")) {
final String otherId = createOpenaireId(50, ((Node) o).getText(), false);
for (final Object o : doc.selectNodes("//*[local-name()='relatedDataset']")) {
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 String originalId = ((Node) o).getText();
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;
}
if (StringUtils.isNotBlank(originalId)) {
@Override
protected Qualifier prepareResourceType(final Document doc, final DataInfo info) {
return null; // NOT PRESENT IN OAF
}
final String otherId = createOpenaireId(50, originalId, false);
final Relation r1 = new Relation();
r1.setRelType("resultResult");
r1.setSubRelType("publicationDataset");
r1.setRelClass("isRelatedTo");
r1.setSource(docId);
r1.setTarget(otherId);
r1.setCollectedfrom(Arrays.asList(collectedFrom));
r1.setDataInfo(info);
r1.setLastupdatetimestamp(lastUpdateTimestamp);
res.add(r1);
final Relation r2 = new Relation();
r2.setRelType("resultResult");
r2.setSubRelType("publicationDataset");
r2.setRelClass("isRelatedTo");
r2.setSource(otherId);
r2.setTarget(docId);
r2.setCollectedfrom(Arrays.asList(collectedFrom));
r2.setDataInfo(info);
r2.setLastupdatetimestamp(lastUpdateTimestamp);
res.add(r2);
}
}
return res;
}
@Override
protected Qualifier prepareResourceType(final Document doc, final DataInfo info) {
return null; // NOT PRESENT IN OAF
}
}

View File

@ -1,4 +1,3 @@
package eu.dnetlib.dhp.oa.graph.raw;
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 OdfToOafMapper(final Map<String, String> code2name) {
super(code2name);
}
public OdfToOafMapper(final Map<String, String> code2name) {
super(code2name);
}
@Override
protected List<StructuredProperty> prepareTitles(final Document doc, final DataInfo info) {
return prepareListStructProps(doc, "//datacite:title", MAIN_TITLE_QUALIFIER, info);
}
@Override
protected List<StructuredProperty> prepareTitles(final Document doc, final DataInfo info) {
return prepareListStructProps(doc, "//datacite:title", MAIN_TITLE_QUALIFIER, info);
}
@Override
protected List<Author> prepareAuthors(final Document doc, final DataInfo info) {
final List<Author> res = new ArrayList<>();
int pos = 1;
for (final Object o : doc.selectNodes("//datacite:creator")) {
final Node n = (Node) o;
final Author author = new Author();
author.setFullname(n.valueOf("./datacite:creatorName"));
author.setName(n.valueOf("./datacite:givenName"));
author.setSurname(n.valueOf("./datacite:familyName"));
author.setAffiliation(prepareListFields(doc, "./datacite:affiliation", info));
author.setPid(preparePids(doc, info));
author.setRank(pos++);
res.add(author);
}
return res;
}
@Override
protected List<Author> prepareAuthors(final Document doc, final DataInfo info) {
final List<Author> res = new ArrayList<>();
int pos = 1;
for (final Object o : doc.selectNodes("//datacite:creator")) {
final Node n = (Node) o;
final Author author = new Author();
author.setFullname(n.valueOf("./datacite:creatorName"));
author.setName(n.valueOf("./datacite:givenName"));
author.setSurname(n.valueOf("./datacite:familyName"));
author.setAffiliation(prepareListFields(doc, "./datacite:affiliation", info));
author.setPid(preparePids(doc, info));
author.setRank(pos++);
res.add(author);
}
return res;
}
private List<StructuredProperty> preparePids(final Document doc, final DataInfo info) {
final List<StructuredProperty> res = new ArrayList<>();
for (final Object o : doc.selectNodes("./datacite:nameIdentifier")) {
res
.add(
structuredProperty(
((Node) o).getText(),
prepareQualifier(
(Node) o, "./@nameIdentifierScheme", "dnet:pid_types", "dnet:pid_types"),
info));
}
return res;
}
private List<StructuredProperty> preparePids(final Document doc, final DataInfo info) {
final List<StructuredProperty> res = new ArrayList<>();
for (final Object o : doc.selectNodes("./datacite:nameIdentifier")) {
res.add(
structuredProperty(
((Node) o).getText(),
prepareQualifier(
(Node) o, "./@nameIdentifierScheme", "dnet:pid_types", "dnet:pid_types"),
info));
}
return res;
}
@Override
protected List<Instance> prepareInstances(
final Document doc,
final DataInfo info,
final KeyValue collectedfrom,
final KeyValue hostedby) {
@Override
protected List<Instance> prepareInstances(
final Document doc,
final DataInfo info,
final KeyValue collectedfrom,
final KeyValue hostedby) {
final Instance instance = new Instance();
instance.setUrl(new ArrayList<>());
instance
.setInstancetype(
prepareQualifier(
doc, "//dr:CobjCategory", "dnet:publication_resource", "dnet:publication_resource"));
instance.setCollectedfrom(collectedfrom);
instance.setHostedby(hostedby);
instance.setDateofacceptance(field(doc.valueOf("//oaf:dateAccepted"), info));
instance.setDistributionlocation(doc.valueOf("//oaf:distributionlocation"));
instance
.setAccessright(
prepareQualifier(doc, "//oaf:accessrights", "dnet:access_modes", "dnet:access_modes"));
instance.setLicense(field(doc.valueOf("//oaf:license"), info));
instance.setRefereed(field(doc.valueOf("//oaf:refereed"), info));
instance.setProcessingchargeamount(field(doc.valueOf("//oaf:processingchargeamount"), info));
instance
.setProcessingchargecurrency(
field(doc.valueOf("//oaf:processingchargeamount/@currency"), info));
final Instance instance = new Instance();
instance.setUrl(new ArrayList<>());
instance.setInstancetype(
prepareQualifier(
doc, "//dr:CobjCategory", "dnet:publication_resource", "dnet:publication_resource"));
instance.setCollectedfrom(collectedfrom);
instance.setHostedby(hostedby);
instance.setDateofacceptance(field(doc.valueOf("//oaf:dateAccepted"), info));
instance.setDistributionlocation(doc.valueOf("//oaf:distributionlocation"));
instance.setAccessright(
prepareQualifier(doc, "//oaf:accessrights", "dnet:access_modes", "dnet:access_modes"));
instance.setLicense(field(doc.valueOf("//oaf:license"), info));
instance.setRefereed(field(doc.valueOf("//oaf:refereed"), info));
instance.setProcessingchargeamount(field(doc.valueOf("//oaf:processingchargeamount"), info));
instance.setProcessingchargecurrency(
field(doc.valueOf("//oaf:processingchargeamount/@currency"), info));
for (final Object o : 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:alternateIdentifier[@alternateIdentifierType='DOI']")) {
instance.getUrl().add("http://dx.doi.org/" + ((Node) o).getText().trim());
}
for (final Object o : doc.selectNodes("//datacite:identifier[@identifierType='DOI']")) {
instance.getUrl().add("http://dx.doi.org/" + ((Node) o).getText().trim());
}
return Arrays.asList(instance);
}
for (final Object o :
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:alternateIdentifier[@alternateIdentifierType='DOI']")) {
instance.getUrl().add("http://dx.doi.org/" + ((Node) o).getText().trim());
}
for (final Object o : doc.selectNodes("//datacite:identifier[@identifierType='DOI']")) {
instance.getUrl().add("http://dx.doi.org/" + ((Node) o).getText().trim());
}
return Arrays.asList(instance);
}
@Override
protected List<Field<String>> prepareSources(final Document doc, final DataInfo info) {
return new ArrayList<>(); // Not present in ODF ???
}
@Override
protected List<Field<String>> prepareSources(final Document doc, final DataInfo info) {
return new ArrayList<>(); // Not present in ODF ???
}
@Override
protected List<StructuredProperty> prepareRelevantDates(final Document doc, final DataInfo info) {
final List<StructuredProperty> res = new ArrayList<>();
for (final Object o : doc.selectNodes("//datacite:date")) {
final String dateType = ((Node) o).valueOf("@dateType");
if (StringUtils.isBlank(dateType)
&& !dateType.equalsIgnoreCase("Accepted")
&& !dateType.equalsIgnoreCase("Issued")
&& !dateType.equalsIgnoreCase("Updated")
&& !dateType.equalsIgnoreCase("Available")) {
res
.add(
structuredProperty(
((Node) o).getText(),
"UNKNOWN",
"UNKNOWN",
"dnet:dataCite_date",
"dnet:dataCite_date",
info));
}
}
return res;
}
@Override
protected List<StructuredProperty> prepareRelevantDates(final Document doc, final DataInfo info) {
final List<StructuredProperty> res = new ArrayList<>();
for (final Object o : doc.selectNodes("//datacite:date")) {
final String dateType = ((Node) o).valueOf("@dateType");
if (StringUtils.isBlank(dateType)
&& !dateType.equalsIgnoreCase("Accepted")
&& !dateType.equalsIgnoreCase("Issued")
&& !dateType.equalsIgnoreCase("Updated")
&& !dateType.equalsIgnoreCase("Available")) {
res.add(
structuredProperty(
((Node) o).getText(),
"UNKNOWN",
"UNKNOWN",
"dnet:dataCite_date",
"dnet:dataCite_date",
info));
}
}
return res;
}
@Override
protected List<Field<String>> prepareCoverages(final Document doc, final DataInfo info) {
return new ArrayList<>(); // Not present in ODF ???
}
@Override
protected List<Field<String>> prepareCoverages(final Document doc, final DataInfo info) {
return new ArrayList<>(); // Not present in ODF ???
}
@Override
protected List<Field<String>> prepareContributors(final Document doc, final DataInfo info) {
return prepareListFields(doc, "//datacite:contributorName", info);
}
@Override
protected List<Field<String>> prepareContributors(final Document doc, final DataInfo info) {
return prepareListFields(doc, "//datacite:contributorName", info);
}
@Override
protected List<Field<String>> prepareFormats(final Document doc, final DataInfo info) {
return prepareListFields(doc, "//datacite:format", info);
}
@Override
protected List<Field<String>> prepareFormats(final Document doc, final DataInfo info) {
return prepareListFields(doc, "//datacite:format", info);
}
@Override
protected Field<String> preparePublisher(final Document doc, final DataInfo info) {
return prepareField(doc, "//datacite:publisher", info);
}
@Override
protected Field<String> preparePublisher(final Document doc, final DataInfo info) {
return prepareField(doc, "//datacite:publisher", info);
}
@Override
protected List<Field<String>> prepareDescriptions(final Document doc, final DataInfo info) {
return prepareListFields(doc, "//datacite:description[@descriptionType='Abstract']", info);
}
@Override
protected List<Field<String>> prepareDescriptions(final Document doc, final DataInfo info) {
return prepareListFields(doc, "//datacite:description[@descriptionType='Abstract']", info);
}
@Override
protected List<StructuredProperty> prepareSubjects(final Document doc, final DataInfo info) {
return prepareListStructProps(doc, "//datacite:subject", info);
}
@Override
protected List<StructuredProperty> prepareSubjects(final Document doc, final DataInfo info) {
return prepareListStructProps(doc, "//datacite:subject", info);
}
@Override
protected Qualifier prepareLanguages(final Document doc) {
return prepareQualifier(doc, "//datacite:language", "dnet:languages", "dnet:languages");
}
@Override
protected Qualifier prepareLanguages(final Document doc) {
return prepareQualifier(doc, "//datacite:language", "dnet:languages", "dnet:languages");
}
@Override
protected List<Field<String>> prepareOtherResearchProductTools(
final Document doc, final DataInfo info) {
return new ArrayList<>(); // Not present in ODF ???
}
@Override
protected List<Field<String>> prepareOtherResearchProductTools(
final Document doc, final DataInfo info) {
return new ArrayList<>(); // Not present in ODF ???
}
@Override
protected List<Field<String>> prepareOtherResearchProductContactGroups(
final Document doc, final DataInfo info) {
return prepareListFields(
doc,
"//datacite:contributor[@contributorType='ContactGroup']/datacite:contributorName",
info);
}
@Override
protected List<Field<String>> prepareOtherResearchProductContactGroups(
final Document doc, final DataInfo info) {
return prepareListFields(
doc,
"//datacite:contributor[@contributorType='ContactGroup']/datacite:contributorName",
info);
}
@Override
protected List<Field<String>> prepareOtherResearchProductContactPersons(
final Document doc, final DataInfo info) {
return prepareListFields(
doc,
"//datacite:contributor[@contributorType='ContactPerson']/datacite:contributorName",
info);
}
@Override
protected List<Field<String>> prepareOtherResearchProductContactPersons(
final Document doc, final DataInfo info) {
return prepareListFields(
doc,
"//datacite:contributor[@contributorType='ContactPerson']/datacite:contributorName",
info);
}
@Override
protected Qualifier prepareSoftwareProgrammingLanguage(final Document doc, final DataInfo info) {
return prepareQualifier(
doc, "//datacite:format", "dnet:programming_languages", "dnet:programming_languages");
}
@Override
protected Qualifier prepareSoftwareProgrammingLanguage(final Document doc, final DataInfo info) {
return prepareQualifier(
doc, "//datacite:format", "dnet:programming_languages", "dnet:programming_languages");
}
@Override
protected Field<String> prepareSoftwareCodeRepositoryUrl(
final Document doc, final DataInfo info) {
return null; // Not present in ODF ???
}
@Override
protected Field<String> prepareSoftwareCodeRepositoryUrl(
final Document doc, final DataInfo info) {
return null; // Not present in ODF ???
}
@Override
protected List<StructuredProperty> prepareSoftwareLicenses(
final Document doc, final DataInfo info) {
return new ArrayList<>(); // Not present in ODF ???
}
@Override
protected List<StructuredProperty> prepareSoftwareLicenses(
final Document doc, final DataInfo info) {
return new ArrayList<>(); // Not present in ODF ???
}
@Override
protected List<Field<String>> prepareSoftwareDocumentationUrls(
final Document doc, final DataInfo info) {
return prepareListFields(
doc,
"//datacite:relatedIdentifier[@relatedIdentifierType='URL' and @relationType='IsDocumentedBy']",
info);
}
@Override
protected List<Field<String>> prepareSoftwareDocumentationUrls(
final Document doc, final DataInfo info) {
return prepareListFields(
doc,
"//datacite:relatedIdentifier[@relatedIdentifierType='URL' and @relationType='IsDocumentedBy']",
info);
}
// DATASETS
// DATASETS
@Override
protected List<GeoLocation> prepareDatasetGeoLocations(final Document doc, final DataInfo info) {
final List<GeoLocation> res = new ArrayList<>();
@Override
protected List<GeoLocation> prepareDatasetGeoLocations(final Document doc, final DataInfo info) {
final List<GeoLocation> res = new ArrayList<>();
for (final Object o : doc.selectNodes("//datacite:geoLocation")) {
final GeoLocation loc = new GeoLocation();
loc.setBox(((Node) o).valueOf("./datacite:geoLocationBox"));
loc.setPlace(((Node) o).valueOf("./datacite:geoLocationPlace"));
loc.setPoint(((Node) o).valueOf("./datacite:geoLocationPoint"));
res.add(loc);
}
return res;
}
for (final Object o : doc.selectNodes("//datacite:geoLocation")) {
final GeoLocation loc = new GeoLocation();
loc.setBox(((Node) o).valueOf("./datacite:geoLocationBox"));
loc.setPlace(((Node) o).valueOf("./datacite:geoLocationPlace"));
loc.setPoint(((Node) o).valueOf("./datacite:geoLocationPoint"));
res.add(loc);
}
return res;
}
@Override
protected Field<String> prepareDatasetMetadataVersionNumber(
final Document doc, final DataInfo info) {
return null; // Not present in ODF ???
}
@Override
protected Field<String> prepareDatasetMetadataVersionNumber(
final Document doc, final DataInfo info) {
return null; // Not present in ODF ???
}
@Override
protected Field<String> prepareDatasetLastMetadataUpdate(
final Document doc, final DataInfo info) {
return prepareField(doc, "//datacite:date[@dateType='Updated']", info);
}
@Override
protected Field<String> prepareDatasetLastMetadataUpdate(
final Document doc, final DataInfo info) {
return prepareField(doc, "//datacite:date[@dateType='Updated']", info);
}
@Override
protected Field<String> prepareDatasetVersion(final Document doc, final DataInfo info) {
return prepareField(doc, "//datacite:version", info);
}
@Override
protected Field<String> prepareDatasetVersion(final Document doc, final DataInfo info) {
return prepareField(doc, "//datacite:version", info);
}
@Override
protected Field<String> prepareDatasetSize(final Document doc, final DataInfo info) {
return prepareField(doc, "//datacite:size", info);
}
@Override
protected Field<String> prepareDatasetSize(final Document doc, final DataInfo info) {
return prepareField(doc, "//datacite:size", info);
}
@Override
protected Field<String> prepareDatasetDevice(final Document doc, final DataInfo info) {
return null; // Not present in ODF ???
}
@Override
protected Field<String> prepareDatasetDevice(final Document doc, final DataInfo info) {
return null; // Not present in ODF ???
}
@Override
protected Field<String> prepareDatasetStorageDate(final Document doc, final DataInfo info) {
return prepareField(doc, "//datacite:date[@dateType='Issued']", info);
}
@Override
protected Field<String> prepareDatasetStorageDate(final Document doc, final DataInfo info) {
return prepareField(doc, "//datacite:date[@dateType='Issued']", info);
}
@Override
protected List<Oaf> addOtherResultRels(
final Document doc,
final KeyValue collectedFrom,
final DataInfo info,
final long lastUpdateTimestamp) {
@Override
protected List<Oaf> addOtherResultRels(
final Document doc,
final KeyValue collectedFrom,
final DataInfo info,
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']")) {
final String otherId = createOpenaireId(50, ((Node) o).getText(), false);
final String type = ((Node) o).valueOf("@relationType");
for (final Object o :
doc.selectNodes("//datacite:relatedIdentifier[@relatedIdentifierType='OPENAIRE']")) {
if (type.equals("IsSupplementTo")) {
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;
}
final String originalId = ((Node) o).getText();
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;
}
if (StringUtils.isNotBlank(originalId)) {
final String otherId = createOpenaireId(50, originalId, false);
final String type = ((Node) o).valueOf("@relationType");
@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");
}
if (type.equals("IsSupplementTo")) {
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(
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");
}
}