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

Merge pull request 'force orginalId for claimed records' (#124) from forceOrginalId_claims into master 

Reviewed-on: D-Net/dnet-hadoop#124
This commit is contained in:
Miriam Baglioni 2021-07-19 17:41:58 +02:00
parent 9913b6073c 5e5f65a3c3
commit 13cf444f85
4 changed files with 34 additions and 6 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

24
dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/dhp/oa/graph/raw/AbstractMdRecordToOafMapper.java
View File

@ -71,6 +71,8 @@ public abstract class AbstractMdRecordToOafMapper {
private final boolean shouldHashId; private final boolean shouldHashId;
private final boolean forceOriginalId;
protected static final String DATACITE_SCHEMA_KERNEL_4 = "http://datacite.org/schema/kernel-4"; 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_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";
@ -98,11 +100,20 @@ public abstract class AbstractMdRecordToOafMapper {
nsContext.put("datacite", DATACITE_SCHEMA_KERNEL_3); nsContext.put("datacite", DATACITE_SCHEMA_KERNEL_3);
} }
protected AbstractMdRecordToOafMapper(final VocabularyGroup vocs, final boolean invisible,
final boolean shouldHashId, final boolean forceOriginalId) {
this.vocs = vocs;
this.invisible = invisible;
this.shouldHashId = shouldHashId;
this.forceOriginalId = forceOriginalId;
}
protected AbstractMdRecordToOafMapper(final VocabularyGroup vocs, final boolean invisible, protected AbstractMdRecordToOafMapper(final VocabularyGroup vocs, final boolean invisible,
final boolean shouldHashId) { final boolean shouldHashId) {
this.vocs = vocs; this.vocs = vocs;
this.invisible = invisible; this.invisible = invisible;
this.shouldHashId = shouldHashId; this.shouldHashId = shouldHashId;
this.forceOriginalId = false;
} }
public List<Oaf> processMdRecord(final String xml) { public List<Oaf> processMdRecord(final String xml) {
@ -190,10 +201,15 @@ public abstract class AbstractMdRecordToOafMapper {
final long lastUpdateTimestamp) { final long lastUpdateTimestamp) {
final OafEntity entity = createEntity(doc, type, instances, collectedFrom, info, lastUpdateTimestamp); final OafEntity entity = createEntity(doc, type, instances, collectedFrom, info, lastUpdateTimestamp);
final String id = IdentifierFactory.createIdentifier(entity, shouldHashId);
if (!id.equals(entity.getId())) { if (!forceOriginalId) {
entity.getOriginalId().add(entity.getId()); final String id = IdentifierFactory.createIdentifier(entity, shouldHashId);
entity.setId(id); if (!id.equals(entity.getId())) {
final Set<String> originalId = Sets.newHashSet(entity.getOriginalId());
originalId.add(entity.getId());
entity.setOriginalId(Lists.newArrayList(originalId));
entity.setId(id);
}
} }
final List<Oaf> oafs = Lists.newArrayList(entity); final List<Oaf> oafs = Lists.newArrayList(entity);

6
dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/dhp/oa/graph/raw/GenerateEntitiesApplication.java
View File

@ -163,11 +163,13 @@ public class GenerateEntitiesApplication {
switch (type.toLowerCase()) { switch (type.toLowerCase()) {
case "oaf-store-cleaned": case "oaf-store-cleaned":
case "oaf-store-claim":
return new OafToOafMapper(vocs, false, shouldHashId).processMdRecord(s); return new OafToOafMapper(vocs, false, shouldHashId).processMdRecord(s);
case "oaf-store-claim":
return new OafToOafMapper(vocs, false, shouldHashId, true).processMdRecord(s);
case "odf-store-cleaned": case "odf-store-cleaned":
case "odf-store-claim":
return new OdfToOafMapper(vocs, false, shouldHashId).processMdRecord(s); return new OdfToOafMapper(vocs, false, shouldHashId).processMdRecord(s);
case "odf-store-claim":
return new OdfToOafMapper(vocs, false, shouldHashId, true).processMdRecord(s);
case "oaf-store-intersection": case "oaf-store-intersection":
return new OafToOafMapper(vocs, true, shouldHashId).processMdRecord(s); return new OafToOafMapper(vocs, true, shouldHashId).processMdRecord(s);
case "odf-store-intersection": case "odf-store-intersection":

5
dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/dhp/oa/graph/raw/OafToOafMapper.java
View File

@ -27,6 +27,11 @@ import eu.dnetlib.dhp.schema.oaf.utils.ModelHardLimits;
public class OafToOafMapper extends AbstractMdRecordToOafMapper { public class OafToOafMapper extends AbstractMdRecordToOafMapper {
public OafToOafMapper(final VocabularyGroup vocs, final boolean invisible, final boolean shouldHashId,
final boolean forceOrginalId) {
super(vocs, invisible, shouldHashId, forceOrginalId);
}
public OafToOafMapper(final VocabularyGroup vocs, final boolean invisible, final boolean shouldHashId) { public OafToOafMapper(final VocabularyGroup vocs, final boolean invisible, final boolean shouldHashId) {
super(vocs, invisible, shouldHashId); super(vocs, invisible, shouldHashId);
} }

5
dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/dhp/oa/graph/raw/OdfToOafMapper.java
View File

@ -22,6 +22,11 @@ public class OdfToOafMapper extends AbstractMdRecordToOafMapper {
public static final String HTTP_DX_DOI_PREIFX = "http://dx.doi.org/"; public static final String HTTP_DX_DOI_PREIFX = "http://dx.doi.org/";
public OdfToOafMapper(final VocabularyGroup vocs, final boolean invisible, final boolean shouldHashId,
final boolean forceOrginalId) {
super(vocs, invisible, shouldHashId, forceOrginalId);
}
public OdfToOafMapper(final VocabularyGroup vocs, final boolean invisible, final boolean shouldHashId) { public OdfToOafMapper(final VocabularyGroup vocs, final boolean invisible, final boolean shouldHashId) {
super(vocs, invisible, shouldHashId); super(vocs, invisible, shouldHashId);
} }