D-Net
/
dnet-hadoop
14
5
Fork 11
Code Issues 14 Pull Requests 12 Projects 2 Releases 2 Wiki Activity

rel_project_validation #69

Merged
claudio.atzori merged 3 commits from rel_project_validation into master 2020-12-09 19:01:21 +01:00
Conversation 0 Commits 3 Files Changed 4 +92 -61
1 changed files with 92 additions and 61 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files
Showing only changes of commit 5f21a356fd - Show all commits

153
dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/dhp/oa/graph/raw/AbstractMdRecordToOafMapper.java
View File

@ -70,8 +70,10 @@ 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_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 = "http://datacite.org/schema/kernel-3";
protected static final String DATACITE_SCHEMA_KERNEL_3_SLASH = "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 ORCID_PID_TYPE = qualifier(
protected static final Qualifier MAG_PID_TYPE = qualifier("MAGIdentifier", "Microsoft Academic Graph Identifier", DNET_PID_TYPES, DNET_PID_TYPES); "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<>(); protected static final Map<String, String> nsContext = new HashMap<>();
@ -85,7 +87,8 @@ public abstract class AbstractMdRecordToOafMapper {
nsContext.put("datacite", DATACITE_SCHEMA_KERNEL_3); 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) { protected AbstractMdRecordToOafMapper(final VocabularyGroup vocs, final boolean invisible) {
this.vocs = vocs; this.vocs = vocs;
@ -97,20 +100,26 @@ public abstract class AbstractMdRecordToOafMapper {
DocumentFactory.getInstance().setXPathNamespaceURIs(nsContext); DocumentFactory.getInstance().setXPathNamespaceURIs(nsContext);
final Document doc = DocumentHelper final Document doc = DocumentHelper
.parseText(xml .parseText(
.replaceAll(DATACITE_SCHEMA_KERNEL_4, DATACITE_SCHEMA_KERNEL_3) xml
.replaceAll(DATACITE_SCHEMA_KERNEL_4_SLASH, DATACITE_SCHEMA_KERNEL_3) .replaceAll(DATACITE_SCHEMA_KERNEL_4, DATACITE_SCHEMA_KERNEL_3)
.replaceAll(DATACITE_SCHEMA_KERNEL_3_SLASH, 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")) final KeyValue hostedBy = StringUtils.isBlank(doc.valueOf("//oaf:hostedBy/@id"))
? collectedFrom ? collectedFrom
: getProvenanceDatasource(doc, "//oaf:hostedBy/@id", "//oaf:hostedBy/@name"); : 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 DataInfo info = prepareDataInfo(doc, invisible);
final long lastUpdateTimestamp = new Date().getTime(); final long lastUpdateTimestamp = new Date().getTime();
@ -152,7 +161,9 @@ public abstract class AbstractMdRecordToOafMapper {
final String dsId = doc.valueOf(xpathId); final String dsId = doc.valueOf(xpathId);
final String dsName = doc.valueOf(xpathName); 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); return keyValue(createOpenaireId(10, dsId, true), dsName);
} }
@ -168,47 +179,47 @@ public abstract class AbstractMdRecordToOafMapper {
final List<Oaf> oafs = new ArrayList<>(); final List<Oaf> oafs = new ArrayList<>();
switch (type.toLowerCase()) { switch (type.toLowerCase()) {
case "publication": case "publication":
final Publication p = new Publication(); final Publication p = new Publication();
populateResultFields(p, doc, instances, collectedFrom, info, lastUpdateTimestamp); populateResultFields(p, doc, instances, collectedFrom, info, lastUpdateTimestamp);
p.setResulttype(PUBLICATION_DEFAULT_RESULTTYPE); p.setResulttype(PUBLICATION_DEFAULT_RESULTTYPE);
p.setJournal(prepareJournal(doc, info)); p.setJournal(prepareJournal(doc, info));
oafs.add(p); oafs.add(p);
break; break;
case "dataset": case "dataset":
final Dataset d = new Dataset(); final Dataset d = new Dataset();
populateResultFields(d, doc, instances, collectedFrom, info, lastUpdateTimestamp); populateResultFields(d, doc, instances, collectedFrom, info, lastUpdateTimestamp);
d.setResulttype(DATASET_DEFAULT_RESULTTYPE); d.setResulttype(DATASET_DEFAULT_RESULTTYPE);
d.setStoragedate(prepareDatasetStorageDate(doc, info)); d.setStoragedate(prepareDatasetStorageDate(doc, info));
d.setDevice(prepareDatasetDevice(doc, info)); d.setDevice(prepareDatasetDevice(doc, info));
d.setSize(prepareDatasetSize(doc, info)); d.setSize(prepareDatasetSize(doc, info));
d.setVersion(prepareDatasetVersion(doc, info)); d.setVersion(prepareDatasetVersion(doc, info));
d.setLastmetadataupdate(prepareDatasetLastMetadataUpdate(doc, info)); d.setLastmetadataupdate(prepareDatasetLastMetadataUpdate(doc, info));
d.setMetadataversionnumber(prepareDatasetMetadataVersionNumber(doc, info)); d.setMetadataversionnumber(prepareDatasetMetadataVersionNumber(doc, info));
d.setGeolocation(prepareDatasetGeoLocations(doc, info)); d.setGeolocation(prepareDatasetGeoLocations(doc, info));
oafs.add(d); oafs.add(d);
break; break;
case "software": case "software":
final Software s = new Software(); final Software s = new Software();
populateResultFields(s, doc, instances, collectedFrom, info, lastUpdateTimestamp); populateResultFields(s, doc, instances, collectedFrom, info, lastUpdateTimestamp);
s.setResulttype(SOFTWARE_DEFAULT_RESULTTYPE); s.setResulttype(SOFTWARE_DEFAULT_RESULTTYPE);
s.setDocumentationUrl(prepareSoftwareDocumentationUrls(doc, info)); s.setDocumentationUrl(prepareSoftwareDocumentationUrls(doc, info));
s.setLicense(prepareSoftwareLicenses(doc, info)); s.setLicense(prepareSoftwareLicenses(doc, info));
s.setCodeRepositoryUrl(prepareSoftwareCodeRepositoryUrl(doc, info)); s.setCodeRepositoryUrl(prepareSoftwareCodeRepositoryUrl(doc, info));
s.setProgrammingLanguage(prepareSoftwareProgrammingLanguage(doc, info)); s.setProgrammingLanguage(prepareSoftwareProgrammingLanguage(doc, info));
oafs.add(s); oafs.add(s);
break; break;
case "": case "":
case "otherresearchproducts": case "otherresearchproducts":
default: default:
final OtherResearchProduct o = new OtherResearchProduct(); final OtherResearchProduct o = new OtherResearchProduct();
populateResultFields(o, doc, instances, collectedFrom, info, lastUpdateTimestamp); populateResultFields(o, doc, instances, collectedFrom, info, lastUpdateTimestamp);
o.setResulttype(ORP_DEFAULT_RESULTTYPE); o.setResulttype(ORP_DEFAULT_RESULTTYPE);
o.setContactperson(prepareOtherResearchProductContactPersons(doc, info)); o.setContactperson(prepareOtherResearchProductContactPersons(doc, info));
o.setContactgroup(prepareOtherResearchProductContactGroups(doc, info)); o.setContactgroup(prepareOtherResearchProductContactGroups(doc, info));
o.setTool(prepareOtherResearchProductTools(doc, info)); o.setTool(prepareOtherResearchProductTools(doc, info));
oafs.add(o); oafs.add(o);
break; break;
} }
if (!oafs.isEmpty()) { if (!oafs.isEmpty()) {
@ -239,9 +250,15 @@ public abstract class AbstractMdRecordToOafMapper {
final String projectId = createOpenaireId(40, originalId, true); final String projectId = createOpenaireId(40, originalId, true);
res res
.add(getRelationWithValidationDate(docId, projectId, RESULT_PROJECT, OUTCOME, IS_PRODUCED_BY, collectedFrom, info, lastUpdateTimestamp, validationdDate)); .add(
getRelationWithValidationDate(
docId, projectId, RESULT_PROJECT, OUTCOME, IS_PRODUCED_BY, collectedFrom, info,
lastUpdateTimestamp, validationdDate));
res res
.add(getRelationWithValidationDate(projectId, docId, RESULT_PROJECT, OUTCOME, PRODUCES, collectedFrom, info, lastUpdateTimestamp, validationdDate)); .add(
getRelationWithValidationDate(
projectId, docId, RESULT_PROJECT, OUTCOME, PRODUCES, collectedFrom, info,
lastUpdateTimestamp, validationdDate));
} }
} }
@ -258,7 +275,8 @@ public abstract class AbstractMdRecordToOafMapper {
final long lastUpdateTimestamp, final long lastUpdateTimestamp,
final String validationDate) { final String validationDate) {
final Relation r = getRelation(source, target, relType, subRelType, relClass, collectedFrom, info, lastUpdateTimestamp); final Relation r = getRelation(
source, target, relType, subRelType, relClass, collectedFrom, info, lastUpdateTimestamp);
r.setValidated(StringUtils.isNotBlank(validationDate)); r.setValidated(StringUtils.isNotBlank(validationDate));
r.setValidationDate(StringUtils.isNotBlank(validationDate) ? validationDate : null); r.setValidationDate(StringUtils.isNotBlank(validationDate) ? validationDate : null);
return r; return r;
@ -456,7 +474,9 @@ public abstract class AbstractMdRecordToOafMapper {
final String sp = n.valueOf("@sp"); final String sp = n.valueOf("@sp");
final String vol = n.valueOf("@vol"); final String vol = n.valueOf("@vol");
final String edition = n.valueOf("@edition"); 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; return null;
} }
@ -465,7 +485,9 @@ public abstract class AbstractMdRecordToOafMapper {
final Node n = doc.selectSingleNode("//*[local-name()='provenance']/*[local-name()='originDescription']"); final Node n = doc.selectSingleNode("//*[local-name()='provenance']/*[local-name()='originDescription']");
if (n != null) { if (n != null) {
final String id = n.valueOf("./*[local-name()='identifier']"); 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']"); return doc.valueOf("//*[local-name()='header']/*[local-name()='identifier']");
@ -534,7 +556,10 @@ public abstract class AbstractMdRecordToOafMapper {
for (final Object o : node.selectNodes(xpath)) { for (final Object o : node.selectNodes(xpath)) {
final Node n = (Node) o; final Node n = (Node) o;
res 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; return res;
} }
@ -542,7 +567,9 @@ public abstract class AbstractMdRecordToOafMapper {
protected OAIProvenance prepareOAIprovenance(final Document doc) { protected OAIProvenance prepareOAIprovenance(final Document doc) {
final Node n = doc.selectSingleNode("//*[local-name()='provenance']/*[local-name()='originDescription']"); 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 identifier = n.valueOf("./*[local-name()='identifier']");
final String baseURL = n.valueOf("./*[local-name()='baseURL']"); final String baseURL = n.valueOf("./*[local-name()='baseURL']");
@ -557,7 +584,9 @@ public abstract class AbstractMdRecordToOafMapper {
protected DataInfo prepareDataInfo(final Document doc, final boolean invisible) { protected DataInfo prepareDataInfo(final Document doc, final boolean invisible) {
final Node n = doc.selectSingleNode("//oaf:datainfo"); 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 paClassId = n.valueOf("./oaf:provenanceaction/@classid");
final String paClassName = n.valueOf("./oaf:provenanceaction/@classname"); final String paClassName = n.valueOf("./oaf:provenanceaction/@classname");
@ -569,7 +598,9 @@ public abstract class AbstractMdRecordToOafMapper {
final Boolean inferred = Boolean.parseBoolean(n.valueOf("./oaf:inferred")); final Boolean inferred = Boolean.parseBoolean(n.valueOf("./oaf:inferred"));
final String trust = n.valueOf("./oaf:trust"); 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) { protected Field<String> prepareField(final Node node, final String xpath, final DataInfo info) {