enrico.ottonello
/
dnet-hadoop
forked from D-Net/dnet-hadoop
1
0
Fork 0
Code Issues Pull Requests Releases Wiki Activity

validation attributes in resultProject relations

This commit is contained in:
Michele Artini 2020-12-09 11:18:26 +01:00
parent 75bf708351
commit 370a5e650b
4 changed files with 88 additions and 93 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

168
dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/dhp/oa/graph/raw/AbstractMdRecordToOafMapper.java
View File

@ -70,10 +70,8 @@ public abstract class AbstractMdRecordToOafMapper {
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(
"MAGIdentifier", "Microsoft Academic Graph Identifier", DNET_PID_TYPES, DNET_PID_TYPES);
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("MAGIdentifier", "Microsoft Academic Graph Identifier", DNET_PID_TYPES, DNET_PID_TYPES);
protected static final Map<String, String> nsContext = new HashMap<>();
@ -87,8 +85,7 @@ public abstract class AbstractMdRecordToOafMapper {
nsContext.put("datacite", DATACITE_SCHEMA_KERNEL_3);
}
protected static final Qualifier MAIN_TITLE_QUALIFIER = qualifier(
"main title", "main title", "dnet:dataCite_title", "dnet:dataCite_title");
protected static final Qualifier MAIN_TITLE_QUALIFIER = qualifier("main title", "main title", "dnet:dataCite_title", "dnet:dataCite_title");
protected AbstractMdRecordToOafMapper(final VocabularyGroup vocs, final boolean invisible) {
this.vocs = vocs;
@ -100,26 +97,20 @@ public abstract class AbstractMdRecordToOafMapper {
DocumentFactory.getInstance().setXPathNamespaceURIs(nsContext);
final Document doc = DocumentHelper
.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));
.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");
final KeyValue collectedFrom = getProvenanceDatasource(doc, "//oaf:collectedFrom/@id", "//oaf:collectedFrom/@name");
if (collectedFrom == null) {
return null;
}
if (collectedFrom == null) { return null; }
final KeyValue hostedBy = StringUtils.isBlank(doc.valueOf("//oaf:hostedBy/@id"))
? collectedFrom
: getProvenanceDatasource(doc, "//oaf:hostedBy/@id", "//oaf:hostedBy/@name");
if (hostedBy == null) {
return null;
}
if (hostedBy == null) { return null; }
final DataInfo info = prepareDataInfo(doc, invisible);
final long lastUpdateTimestamp = new Date().getTime();
@ -161,9 +152,7 @@ public abstract class AbstractMdRecordToOafMapper {
final String dsId = doc.valueOf(xpathId);
final String dsName = doc.valueOf(xpathName);
if (StringUtils.isBlank(dsId) | StringUtils.isBlank(dsName)) {
return null;
}
if (StringUtils.isBlank(dsId) | StringUtils.isBlank(dsName)) { return null; }
return keyValue(createOpenaireId(10, dsId, true), dsName);
}
@ -179,47 +168,47 @@ public abstract class AbstractMdRecordToOafMapper {
final List<Oaf> oafs = new ArrayList<>();
switch (type.toLowerCase()) {
case "publication":
final Publication p = new Publication();
populateResultFields(p, doc, instances, collectedFrom, info, lastUpdateTimestamp);
p.setResulttype(PUBLICATION_DEFAULT_RESULTTYPE);
p.setJournal(prepareJournal(doc, info));
oafs.add(p);
break;
case "dataset":
final Dataset d = new Dataset();
populateResultFields(d, doc, instances, collectedFrom, info, lastUpdateTimestamp);
d.setResulttype(DATASET_DEFAULT_RESULTTYPE);
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, instances, collectedFrom, info, lastUpdateTimestamp);
s.setResulttype(SOFTWARE_DEFAULT_RESULTTYPE);
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 "":
case "otherresearchproducts":
default:
final OtherResearchProduct o = new OtherResearchProduct();
populateResultFields(o, doc, instances, collectedFrom, info, lastUpdateTimestamp);
o.setResulttype(ORP_DEFAULT_RESULTTYPE);
o.setContactperson(prepareOtherResearchProductContactPersons(doc, info));
o.setContactgroup(prepareOtherResearchProductContactGroups(doc, info));
o.setTool(prepareOtherResearchProductTools(doc, info));
oafs.add(o);
break;
case "publication":
final Publication p = new Publication();
populateResultFields(p, doc, instances, collectedFrom, info, lastUpdateTimestamp);
p.setResulttype(PUBLICATION_DEFAULT_RESULTTYPE);
p.setJournal(prepareJournal(doc, info));
oafs.add(p);
break;
case "dataset":
final Dataset d = new Dataset();
populateResultFields(d, doc, instances, collectedFrom, info, lastUpdateTimestamp);
d.setResulttype(DATASET_DEFAULT_RESULTTYPE);
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, instances, collectedFrom, info, lastUpdateTimestamp);
s.setResulttype(SOFTWARE_DEFAULT_RESULTTYPE);
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 "":
case "otherresearchproducts":
default:
final OtherResearchProduct o = new OtherResearchProduct();
populateResultFields(o, doc, instances, collectedFrom, info, lastUpdateTimestamp);
o.setResulttype(ORP_DEFAULT_RESULTTYPE);
o.setContactperson(prepareOtherResearchProductContactPersons(doc, info));
o.setContactgroup(prepareOtherResearchProductContactGroups(doc, info));
o.setTool(prepareOtherResearchProductTools(doc, info));
oafs.add(o);
break;
}
if (!oafs.isEmpty()) {
@ -244,25 +233,37 @@ public abstract class AbstractMdRecordToOafMapper {
final String originalId = ((Node) o).getText();
final String validationdDate = ((Node) o).valueOf("@validationDate");
if (StringUtils.isNotBlank(originalId)) {
final String projectId = createOpenaireId(40, originalId, true);
res
.add(
getRelation(
docId, projectId, RESULT_PROJECT, OUTCOME, IS_PRODUCED_BY, collectedFrom, info,
lastUpdateTimestamp));
.add(getRelationWithValidationDate(docId, projectId, RESULT_PROJECT, OUTCOME, IS_PRODUCED_BY, collectedFrom, info, lastUpdateTimestamp, validationdDate));
res
.add(
getRelation(
projectId, docId, RESULT_PROJECT, OUTCOME, PRODUCES, collectedFrom, info,
lastUpdateTimestamp));
.add(getRelationWithValidationDate(projectId, docId, RESULT_PROJECT, OUTCOME, PRODUCES, collectedFrom, info, lastUpdateTimestamp, validationdDate));
}
}
return res;
}
protected Relation getRelationWithValidationDate(final String source,
final String target,
final String relType,
final String subRelType,
final String relClass,
final KeyValue collectedFrom,
final DataInfo info,
final long lastUpdateTimestamp,
final String validationDate) {
final Relation r = getRelation(source, target, relType, subRelType, relClass, collectedFrom, info, lastUpdateTimestamp);
r.setValidated(StringUtils.isNotBlank(validationDate));
r.setValidationDate(StringUtils.isNotBlank(validationDate) ? validationDate : null);
return r;
}
protected Relation getRelation(final String source,
final String target,
final String relType,
@ -455,9 +456,7 @@ public abstract class AbstractMdRecordToOafMapper {
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);
}
if (StringUtils.isNotBlank(name)) { return journal(name, issnPrinted, issnOnline, issnLinking, ep, iss, sp, vol, edition, null, null, info); }
}
return null;
}
@ -466,9 +465,7 @@ public abstract class AbstractMdRecordToOafMapper {
final Node n = doc.selectSingleNode("//*[local-name()='provenance']/*[local-name()='originDescription']");
if (n != null) {
final String id = n.valueOf("./*[local-name()='identifier']");
if (StringUtils.isNotBlank(id)) {
return id;
}
if (StringUtils.isNotBlank(id)) { return id; }
}
return doc.valueOf("//*[local-name()='header']/*[local-name()='identifier']");
@ -537,10 +534,7 @@ public abstract class AbstractMdRecordToOafMapper {
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));
.add(structuredProperty(n.getText(), n.valueOf("@classid"), n.valueOf("@classname"), n.valueOf("@schemeid"), n.valueOf("@schemename"), info));
}
return res;
}
@ -548,9 +542,7 @@ public abstract class AbstractMdRecordToOafMapper {
protected OAIProvenance prepareOAIprovenance(final Document doc) {
final Node n = doc.selectSingleNode("//*[local-name()='provenance']/*[local-name()='originDescription']");
if (n == null) {
return null;
}
if (n == null) { return null; }
final String identifier = n.valueOf("./*[local-name()='identifier']");
final String baseURL = n.valueOf("./*[local-name()='baseURL']");
@ -565,9 +557,7 @@ public abstract class AbstractMdRecordToOafMapper {
protected DataInfo prepareDataInfo(final Document doc, final boolean invisible) {
final Node n = doc.selectSingleNode("//oaf:datainfo");
if (n == null) {
return dataInfo(false, null, false, invisible, REPOSITORY_PROVENANCE_ACTIONS, "0.9");
}
if (n == null) { return dataInfo(false, null, false, invisible, REPOSITORY_PROVENANCE_ACTIONS, "0.9"); }
final String paClassId = n.valueOf("./oaf:provenanceaction/@classid");
final String paClassName = n.valueOf("./oaf:provenanceaction/@classname");
@ -579,9 +569,7 @@ public abstract class AbstractMdRecordToOafMapper {
final Boolean inferred = Boolean.parseBoolean(n.valueOf("./oaf:inferred"));
final String trust = n.valueOf("./oaf:trust");
return dataInfo(
deletedbyinference, inferenceprovenance, inferred, invisible,
qualifier(paClassId, paClassName, paSchemeId, paSchemeName), trust);
return dataInfo(deletedbyinference, inferenceprovenance, inferred, invisible, qualifier(paClassId, paClassName, paSchemeId, paSchemeName), trust);
}
protected Field<String> prepareField(final Node node, final String xpath, final DataInfo info) {

9
dhp-workflows/dhp-graph-mapper/src/test/java/eu/dnetlib/dhp/oa/graph/raw/MappersTest.java
View File

@ -141,7 +141,10 @@ public class MappersTest {
assertTrue(StringUtils.isNotBlank(r2.getRelClass()));
assertTrue(StringUtils.isNotBlank(r1.getRelType()));
assertTrue(StringUtils.isNotBlank(r2.getRelType()));
assertTrue(r1.getValidated());
assertTrue(r2.getValidated());
assertEquals(r1.getValidationDate(), "2020-01-01");
assertEquals(r2.getValidationDate(), "2020-01-01");
// System.out.println(new ObjectMapper().writeValueAsString(p));
// System.out.println(new ObjectMapper().writeValueAsString(r1));
// System.out.println(new ObjectMapper().writeValueAsString(r2));
@ -246,6 +249,10 @@ public class MappersTest {
assertTrue(StringUtils.isNotBlank(r2.getRelClass()));
assertTrue(StringUtils.isNotBlank(r1.getRelType()));
assertTrue(StringUtils.isNotBlank(r2.getRelType()));
assertTrue(r1.getValidated());
assertTrue(r2.getValidated());
assertEquals(r1.getValidationDate(), "2020-01-01");
assertEquals(r2.getValidationDate(), "2020-01-01");
}
@Test

2
dhp-workflows/dhp-graph-mapper/src/test/resources/eu/dnetlib/dhp/oa/graph/raw/oaf_record.xml
View File

@ -51,7 +51,7 @@
<!--<dr:CobjCategory type="publication">0001</dr:CobjCategory>-->
<dr:CobjCategory>0001</dr:CobjCategory>
<oaf:dateAccepted>2017-01-01</oaf:dateAccepted>
<oaf:projectid>corda_______::226852</oaf:projectid>
<oaf:projectid validationDate="2020-01-01">corda_______::226852</oaf:projectid>
<oaf:accessrights>OPEN</oaf:accessrights>
<oaf:hostedBy id="openaire____::issn226852" name="One Ecosystem"/>
<oaf:collectedFrom

2
dhp-workflows/dhp-graph-mapper/src/test/resources/eu/dnetlib/dhp/oa/graph/raw/odf_dataset.xml
View File

@ -89,7 +89,7 @@
<oaf:language>und</oaf:language>
<oaf:concept id="https://zenodo.org/communities/epfl"/>
<oaf:hostedBy id="re3data_____::r3d100010468" name="Zenodo"/>
<oaf:projectid>corda_______::226852</oaf:projectid>
<oaf:projectid validationDate="2020-01-01">corda_______::226852</oaf:projectid>
<oaf:collectedFrom id="re3data_____::r3d100010468" name="Zenodo"/>
<oaf:refereed>0001</oaf:refereed>s
</metadata>