forked from D-Net/dnet-hadoop
Merge branch 'master' into graph_db
This commit is contained in:
commit
3be0e5c2cd
|
|
@ -24,7 +24,9 @@ public abstract class AbstractMdRecordToOafMapper {
|
|||
private final boolean invisible;
|
||||
|
||||
protected static final String DATACITE_SCHEMA_KERNEL_4 = "http://datacite.org/schema/kernel-4";
|
||||
protected static final String DATACITE_SCHEMA_KERNEL_4_SLASH = "http://datacite.org/schema/kernel-4/";
|
||||
protected static final String DATACITE_SCHEMA_KERNEL_3 = "http://datacite.org/schema/kernel-3";
|
||||
protected static final String DATACITE_SCHEMA_KERNEL_3_SLASH = "http://datacite.org/schema/kernel-3/";
|
||||
protected static final Qualifier ORCID_PID_TYPE = qualifier(
|
||||
"ORCID", "Open Researcher and Contributor ID", DNET_PID_TYPES, DNET_PID_TYPES);
|
||||
protected static final Qualifier MAG_PID_TYPE = qualifier(
|
||||
|
|
@ -55,7 +57,11 @@ public abstract class AbstractMdRecordToOafMapper {
|
|||
DocumentFactory.getInstance().setXPathNamespaceURIs(nsContext);
|
||||
|
||||
final Document doc = DocumentHelper
|
||||
.parseText(xml.replaceAll(DATACITE_SCHEMA_KERNEL_4, DATACITE_SCHEMA_KERNEL_3));
|
||||
.parseText(
|
||||
xml
|
||||
.replaceAll(DATACITE_SCHEMA_KERNEL_4, DATACITE_SCHEMA_KERNEL_3)
|
||||
.replaceAll(DATACITE_SCHEMA_KERNEL_4_SLASH, DATACITE_SCHEMA_KERNEL_3)
|
||||
.replaceAll(DATACITE_SCHEMA_KERNEL_3_SLASH, DATACITE_SCHEMA_KERNEL_3));
|
||||
|
||||
final KeyValue collectedFrom = getProvenanceDatasource(
|
||||
doc, "//oaf:collectedFrom/@id", "//oaf:collectedFrom/@name");
|
||||
|
|
|
|||
|
|
@ -14,6 +14,8 @@ import org.apache.commons.lang3.StringUtils;
|
|||
import org.dom4j.Document;
|
||||
import org.dom4j.Node;
|
||||
|
||||
import com.google.common.collect.Lists;
|
||||
|
||||
import eu.dnetlib.dhp.common.PacePerson;
|
||||
import eu.dnetlib.dhp.common.VocabularyGroup;
|
||||
import eu.dnetlib.dhp.schema.oaf.Author;
|
||||
|
|
@ -364,7 +366,7 @@ public class OdfToOafMapper extends AbstractMdRecordToOafMapper {
|
|||
|
||||
@Override
|
||||
protected List<StructuredProperty> prepareResultPids(final Document doc, final DataInfo info) {
|
||||
final List<StructuredProperty> res = new ArrayList<>();
|
||||
final Set<StructuredProperty> res = new HashSet();
|
||||
res
|
||||
.addAll(
|
||||
prepareListStructPropsWithValidQualifier(
|
||||
|
|
@ -380,7 +382,7 @@ public class OdfToOafMapper extends AbstractMdRecordToOafMapper {
|
|||
doc,
|
||||
"//datacite:alternateIdentifier[@alternateIdentifierType != 'URL' and @alternateIdentifierType != 'landingPage']",
|
||||
"@alternateIdentifierType", DNET_PID_TYPES, info));
|
||||
return res;
|
||||
return Lists.newArrayList(res);
|
||||
}
|
||||
|
||||
}
|
||||
|
|
|
|||
|
|
@ -286,6 +286,29 @@ public class MappersTest {
|
|||
System.out.println("***************");
|
||||
}
|
||||
|
||||
@Test
|
||||
void testNakala() throws IOException {
|
||||
final String xml = IOUtils.toString(getClass().getResourceAsStream("odf_nakala.xml"));
|
||||
final List<Oaf> list = new OdfToOafMapper(vocs, false).processMdRecord(xml);
|
||||
|
||||
System.out.println("***************");
|
||||
System.out.println(new ObjectMapper().writeValueAsString(list));
|
||||
System.out.println("***************");
|
||||
|
||||
assertEquals(1, list.size());
|
||||
assertTrue(list.get(0) instanceof Dataset);
|
||||
|
||||
final Dataset d = (Dataset) list.get(0);
|
||||
|
||||
assertValidId(d.getId());
|
||||
assertValidId(d.getCollectedfrom().get(0).getKey());
|
||||
assertTrue(StringUtils.isNotBlank(d.getTitle().get(0).getValue()));
|
||||
assertEquals(1, d.getAuthor().size());
|
||||
assertEquals(0, d.getSubject().size());
|
||||
assertEquals(1, d.getInstance().size());
|
||||
assertEquals(1, d.getPid().size());
|
||||
}
|
||||
|
||||
private void assertValidId(final String id) {
|
||||
assertEquals(49, id.length());
|
||||
assertEquals('|', id.charAt(2));
|
||||
|
|
|
|||
|
|
@ -951,6 +951,7 @@ dnet:countries @=@ ZW @=@ ABW
|
|||
dnet:protocols @=@ oai @=@ OAI-PMH
|
||||
dnet:protocols @=@ oai @=@ OAI_PMH
|
||||
dnet:pid_types @=@ orcid @=@ ORCID12
|
||||
dnet:pid_types @=@ handle @=@ hdl
|
||||
dnet:review_levels @=@ 0000 @=@ UNKNOWN
|
||||
dnet:review_levels @=@ 0002 @=@ 80 大阪経大学会「Working Paper」
|
||||
dnet:review_levels @=@ 0002 @=@ AO
|
||||
|
|
|
|||
|
|
@ -1045,6 +1045,7 @@ dnet:pid_types @=@ dnet:pid_types @=@ pmid @=@ pmid
|
|||
dnet:pid_types @=@ dnet:pid_types @=@ urn @=@ urn
|
||||
dnet:pid_types @=@ dnet:pid_types @=@ who @=@ WHO Identifier
|
||||
dnet:pid_types @=@ dnet:pid_types @=@ drks @=@ DRKS Identifier
|
||||
dnet:pid_types @=@ dnet:pid_types @=@ handle @=@ Handle
|
||||
dnet:topic_types @=@ dnet:topic_types @=@ ENRICH/MISSING/SUBJECT/ACM @=@ An ACM classification term that can be associated to your publications
|
||||
dnet:topic_types @=@ dnet:topic_types @=@ ENRICH/MISSING/SUBJECT/ARXIV @=@ An ARXIV classification term that can be associated to your publications
|
||||
dnet:topic_types @=@ dnet:topic_types @=@ ENRICH/MISSING/SUBJECT/DDC @=@ A Dewey Decimal classification term (DDC) that can be associated to your publications
|
||||
|
|
|
|||
|
|
@ -0,0 +1,88 @@
|
|||
<?xml version="1.0" encoding="UTF-8"?>
|
||||
<record xmlns:dr="http://www.driver-repository.eu/namespace/dr"
|
||||
xmlns:oaf="http://namespace.openaire.eu/oaf" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance">
|
||||
<oai:header xmlns="http://namespace.openaire.eu/"
|
||||
xmlns:dc="http://purl.org/dc/elements/1.1/"
|
||||
xmlns:dri="http://www.driver-repository.eu/namespace/dri"
|
||||
xmlns:oai="http://www.openarchives.org/OAI/2.0/" xmlns:prov="http://www.openarchives.org/OAI/2.0/provenance">
|
||||
<dri:objIdentifier>r3f5b9831893::cca7367159bc3ff90cd2f75bf9dc21c4</dri:objIdentifier>
|
||||
<dri:recordIdentifier>oai:nakala.fr:hdl_11280_847e01df</dri:recordIdentifier>
|
||||
<dri:dateOfCollection>2020-08-01T00:16:24.742Z</dri:dateOfCollection>
|
||||
<oaf:datasourceprefix>r3f5b9831893</oaf:datasourceprefix>
|
||||
<identifier xmlns="http://www.openarchives.org/OAI/2.0/">oai:nakala.fr:hdl_11280_847e01df</identifier>
|
||||
<datestamp xmlns="http://www.openarchives.org/OAI/2.0/">2020-06-08T01:01:38Z</datestamp>
|
||||
<setSpec xmlns="http://www.openarchives.org/OAI/2.0/">hdl_11280_2b09fc10</setSpec>
|
||||
<setSpec xmlns="http://www.openarchives.org/OAI/2.0/">hdl_11280_c1bc48d0</setSpec>
|
||||
<setSpec xmlns="http://www.openarchives.org/OAI/2.0/">hdl_11280_57c8db3a</setSpec>
|
||||
<dr:dateOfTransformation>2020-08-01T00:31:35.625Z</dr:dateOfTransformation>
|
||||
</oai:header>
|
||||
<metadata>
|
||||
<datacite:resource xmlns="http://www.openarchives.org/OAI/2.0/"
|
||||
xmlns:datacite="http://datacite.org/schema/kernel-4"
|
||||
xmlns:dc="http://purl.org/dc/elements/1.1/"
|
||||
xmlns:dri="http://www.driver-repository.eu/namespace/dri"
|
||||
xmlns:oai="http://www.openarchives.org/OAI/2.0/"
|
||||
xmlns:prov="http://www.openarchives.org/OAI/2.0/provenance" xsi:schemaLocation="http://datacite.org/schema/kernel-4 https://schema.datacite.org/meta/kernel-4/metadata.xsd">
|
||||
<datacite:alternateIdentifier identifierType="URL" xmlns:datacite="http://datacite.org/schema/kernel-4/">277</datacite:alternateIdentifier>
|
||||
<datacite:identifier identifierType="Handle" xmlns:datacite="http://datacite.org/schema/kernel-4/">http://hdl.handle.net/11280/847e01df</datacite:identifier>
|
||||
<alternateIdentifiers>
|
||||
<alternateIdentifier alternateIdentifierType="URL">http://hdl.handle.net/http://hdl.handle.net/11280/847e01df</alternateIdentifier>
|
||||
</alternateIdentifiers>
|
||||
<datacite:alternateIdentifier identifierType="URL" xmlns:datacite="http://datacite.org/schema/kernel-4/">http://nakala.fr/data/11280/847e01df</datacite:alternateIdentifier>
|
||||
<datacite:creators xmlns:datacite="http://datacite.org/schema/kernel-4/">
|
||||
<datacite:creator>
|
||||
<datacite:creatorName>DHAAP</datacite:creatorName>
|
||||
</datacite:creator>
|
||||
</datacite:creators>
|
||||
<datacite:titles xmlns:datacite="http://datacite.org/schema/kernel-4/">
|
||||
<datacite:title>CVP_Notice277-1 place du Docteur Antoine Béclère _PHO02.jpg</datacite:title>
|
||||
</datacite:titles>
|
||||
<datacite:descriptions xmlns:datacite="http://datacite.org/schema/kernel-4/">
|
||||
<datacite:description descriptionType="Abstract">Hôpital Saint-Antoine. Fragment de dalle funéraire trouvée en décembre 1932. Paris (XIIème arr.). Photographie d'Albert Citerne (1876-1970). Plaque de verre, 1932. Département Histoire de l'Architecture et Archéologie de Paris.</datacite:description>
|
||||
<datacite:description descriptionType="Abstract">Nfa_1146</datacite:description>
|
||||
<datacite:description descriptionType="Abstract">Hôpital Saint-Antoine. Fragment de dalle funéraire trouvée en décembre 1932. Paris (XIIème arr.). Photographie d'Albert Citerne (1876-1970). Plaque de verre, 1932. Département Histoire de l'Architecture et Archéologie de Paris.</datacite:description>
|
||||
</datacite:descriptions>
|
||||
<datacite:publisher xmlns:datacite="http://datacite.org/schema/kernel-4/">Nakala by Huma-Num</datacite:publisher>
|
||||
<datacite:contributors xmlns:datacite="http://datacite.org/schema/kernel-4/">
|
||||
<datacite:contributor contributorType="Other">
|
||||
<datacite:contributorName>DHAAP, Pôle Archéologique</datacite:contributorName>
|
||||
</datacite:contributor>
|
||||
</datacite:contributors>
|
||||
<datacite:dates xmlns:datacite="http://datacite.org/schema/kernel-4/">
|
||||
<datacite:date dateType="Created">1932</datacite:date>
|
||||
</datacite:dates>
|
||||
<datacite:resourceType resourceTypeGeneral="Image" xmlns:datacite="http://datacite.org/schema/kernel-4/">StillImage</datacite:resourceType>
|
||||
<datacite:rightsList xmlns:datacite="http://datacite.org/schema/kernel-4/">
|
||||
<datacite:rights rightsURI="info:eu-repo/semantics/openAccess"/>
|
||||
</datacite:rightsList>
|
||||
</datacite:resource>
|
||||
<oaf:identifier identifierType="handle">http://hdl.handle.net/11280/847e01df</oaf:identifier>
|
||||
<dr:CobjCategory type="dataset">0025</dr:CobjCategory>
|
||||
<oaf:dateAccepted/>
|
||||
<oaf:accessrights>OPEN</oaf:accessrights>
|
||||
<oaf:language>und</oaf:language>
|
||||
<oaf:hostedBy id="re3data_____::r3d100012102" name="NAKALA"/>
|
||||
<oaf:collectedFrom id="re3data_____::r3d100012102" name="NAKALA"/>
|
||||
</metadata>
|
||||
<about xmlns:dc="http://purl.org/dc/elements/1.1/"
|
||||
xmlns:dri="http://www.driver-repository.eu/namespace/dri"
|
||||
xmlns:oai="http://www.openarchives.org/OAI/2.0/" xmlns:prov="http://www.openarchives.org/OAI/2.0/provenance">
|
||||
<provenance xmlns="http://www.openarchives.org/OAI/2.0/provenance" xsi:schemaLocation="http://www.openarchives.org/OAI/2.0/provenance http://www.openarchives.org/OAI/2.0/provenance.xsd">
|
||||
<originDescription altered="true" harvestDate="2020-08-01T00:16:24.742Z">
|
||||
<baseURL>https%3A%2F%2Fwww.nakala.fr%2Foai_oa%2F11280%2F8892ab4b</baseURL>
|
||||
<identifier>oai:nakala.fr:hdl_11280_847e01df</identifier>
|
||||
<datestamp>2020-06-08T01:01:38Z</datestamp>
|
||||
<metadataNamespace/>
|
||||
</originDescription>
|
||||
</provenance>
|
||||
<oaf:datainfo>
|
||||
<oaf:inferred>false</oaf:inferred>
|
||||
<oaf:deletedbyinference>false</oaf:deletedbyinference>
|
||||
<oaf:trust>0.9</oaf:trust>
|
||||
<oaf:inferenceprovenance/>
|
||||
<oaf:provenanceaction classid="sysimport:crosswalk:datasetarchive"
|
||||
classname="sysimport:crosswalk:datasetarchive"
|
||||
schemeid="dnet:provenanceActions" schemename="dnet:provenanceActions"/>
|
||||
</oaf:datainfo>
|
||||
</about>
|
||||
</record>
|
||||
Loading…
Reference in New Issue