sabeel
/
dnet-hadoop
forked from D-Net/dnet-hadoop
1
0
Fork 0
Code Pull Requests Activity

Merge branch 'master' of https://code-repo.d4science.org/D-Net/dnet-hadoop

This commit is contained in:
Claudio Atzori 2020-03-04 14:54:46 +01:00
parent d7137e566e 7a2a466161
commit 5474e8ac9f
5 changed files with 32 additions and 28 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

20
dhp-workflows/dhp-aggregation/src/main/java/eu/dnetlib/dhp/migration/step1/MigrateDbEntitiesApplication.java
View File

@ -117,7 +117,7 @@ public class MigrateDbEntitiesApplication extends AbstractMigrationApplication i
final Datasource ds = new Datasource();
ds.setId(createOpenaireId(10, rs.getString("datasourceid")));
ds.setId(createOpenaireId(10, rs.getString("datasourceid"), true));
ds.setOriginalId(Arrays.asList(rs.getString("datasourceid")));
ds.setCollectedfrom(listKeyValues(rs.getString("collectedfromid"), rs.getString("collectedfromname")));
ds.setPid(new ArrayList<>());
@ -224,7 +224,7 @@ public class MigrateDbEntitiesApplication extends AbstractMigrationApplication i
final Project p = new Project();
p.setId(createOpenaireId(40, rs.getString("projectid")));
p.setId(createOpenaireId(40, rs.getString("projectid"), true));
p.setOriginalId(Arrays.asList(rs.getString("projectid")));
p.setCollectedfrom(listKeyValues(rs.getString("collectedfromid"), rs.getString("collectedfromname")));
p.setPid(new ArrayList<>());
@ -314,7 +314,7 @@ public class MigrateDbEntitiesApplication extends AbstractMigrationApplication i
final Organization o = new Organization();
o.setId(createOpenaireId(20, rs.getString("organizationid")));
o.setId(createOpenaireId(20, rs.getString("organizationid"), true));
o.setOriginalId(Arrays.asList(rs.getString("organizationid")));
o.setCollectedfrom(listKeyValues(rs.getString("collectedfromid"), rs.getString("collectedfromname")));
o.setPid(new ArrayList<>());
@ -378,8 +378,8 @@ public class MigrateDbEntitiesApplication extends AbstractMigrationApplication i
try {
final DataInfo info = prepareDataInfo(rs);
final String orgId = createOpenaireId(20, rs.getString("organization"));
final String dsId = createOpenaireId(10, rs.getString("datasource"));
final String orgId = createOpenaireId(20, rs.getString("organization"), true);
final String dsId = createOpenaireId(10, rs.getString("datasource"), true);
final List<KeyValue> collectedFrom = listKeyValues(rs.getString("collectedfromid"), rs.getString("collectedfromname"));
final Relation r1 = new Relation();
@ -427,8 +427,8 @@ public class MigrateDbEntitiesApplication extends AbstractMigrationApplication i
try {
final DataInfo info = prepareDataInfo(rs);
final String orgId = createOpenaireId(20, rs.getString("resporganization"));
final String projectId = createOpenaireId(40, rs.getString("project"));
final String orgId = createOpenaireId(20, rs.getString("resporganization"), true);
final String projectId = createOpenaireId(40, rs.getString("project"), true);
final List<KeyValue> collectedFrom = listKeyValues(rs.getString("collectedfromid"), rs.getString("collectedfromname"));
final Relation r1 = new Relation();
@ -493,14 +493,14 @@ public class MigrateDbEntitiesApplication extends AbstractMigrationApplication i
} else {
r = new Publication();
}
r.setId(createOpenaireId(50, rs.getString("target_id")));
r.setId(createOpenaireId(50, rs.getString("target_id"), false));
r.setLastupdatetimestamp(lastUpdateTimestamp);
r.setContext(prepareContext(rs.getString("source_id"), info));
r.setDataInfo(info);
emitOaf(r);
} else {
final String sourceId = createOpenaireId(rs.getString("source_type"), rs.getString("source_id"));
final String targetId = createOpenaireId(rs.getString("target_type"), rs.getString("target_id"));
final String sourceId = createOpenaireId(rs.getString("source_type"), rs.getString("source_id"), false);
final String targetId = createOpenaireId(rs.getString("target_type"), rs.getString("target_id"), false);
final Relation r1 = new Relation();
final Relation r2 = new Relation();

6
dhp-workflows/dhp-aggregation/src/main/java/eu/dnetlib/dhp/migration/step2/AbstractMdRecordToOafMapper.java
View File

@ -153,10 +153,10 @@ public abstract class AbstractMdRecordToOafMapper {
final List<Oaf> res = new ArrayList<>();
final String docId = createOpenaireId(50, doc.valueOf("//dri:objIdentifier"));
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());
final String projectId = createOpenaireId(40, ((Node) o).getText(), true);
final Relation r1 = new Relation();
r1.setRelType("resultProject");
@ -197,7 +197,7 @@ public abstract class AbstractMdRecordToOafMapper {
final long lastUpdateTimestamp) {
r.setDataInfo(info);
r.setLastupdatetimestamp(lastUpdateTimestamp);
r.setId(createOpenaireId(50, doc.valueOf("//dri:objIdentifier")));
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));

4
dhp-workflows/dhp-aggregation/src/main/java/eu/dnetlib/dhp/migration/step2/OafToOafMapper.java
View File

@ -202,12 +202,12 @@ public class OafToOafMapper extends AbstractMdRecordToOafMapper {
final KeyValue collectedFrom,
final DataInfo info,
final long lastUpdateTimestamp) {
final String docId = createOpenaireId(50, doc.valueOf("//dri:objIdentifier"));
final String docId = createOpenaireId(50, doc.valueOf("//dri:objIdentifier"), false);
final List<Oaf> res = new ArrayList<>();
for (final Object o : doc.selectNodes("//*[local-name()='relatedDataset']")) {
final String otherId = createOpenaireId(50, ((Node) o).getText());
final String otherId = createOpenaireId(50, ((Node) o).getText(), false);
final Relation r1 = new Relation();
r1.setRelType("resultResult");

6
dhp-workflows/dhp-aggregation/src/main/java/eu/dnetlib/dhp/migration/step2/OdfToOafMapper.java
View File

@ -219,12 +219,12 @@ public class OdfToOafMapper extends AbstractMdRecordToOafMapper {
@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"));
final String docId = createOpenaireId(50, doc.valueOf("//dri:objIdentifier"), false);
final List<Oaf> res = new ArrayList<>();
for (final Object o : doc.selectNodes("//*[local-name() = 'resource']//*[local-name()='relatedIdentifier' and ./@relatedIdentifierType='OPENAIRE']")) {
final String otherId = createOpenaireId(50, ((Node) o).getText());
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");
if (type.equals("IsSupplementTo")) {

24
dhp-workflows/dhp-aggregation/src/main/java/eu/dnetlib/dhp/migration/utils/OafMapperUtils.java
View File

@ -163,24 +163,28 @@ public class OafMapperUtils {
return d;
}
public static String createOpenaireId(final int prefix, final String originalId) {
final String nsPrefix = StringUtils.substringBefore(originalId, "::");
final String rest = StringUtils.substringAfter(originalId, "::");
return String.format("%s|%s::%s", prefix, nsPrefix, DHPUtils.md5(rest));
public static String createOpenaireId(final int prefix, final String originalId, final boolean to_md5) {
if (to_md5) {
final String nsPrefix = StringUtils.substringBefore(originalId, "::");
final String rest = StringUtils.substringAfter(originalId, "::");
return String.format("%s|%s::%s", prefix, nsPrefix, DHPUtils.md5(rest));
} else {
return String.format("%s|%s", prefix, originalId);
}
}
public static String createOpenaireId(final String type, final String originalId) {
public static String createOpenaireId(final String type, final String originalId, final boolean to_md5) {
switch (type) {
case "datasource":
return createOpenaireId(10, originalId);
return createOpenaireId(10, originalId, to_md5);
case "organization":
return createOpenaireId(20, originalId);
return createOpenaireId(20, originalId, to_md5);
case "person":
return createOpenaireId(30, originalId);
return createOpenaireId(30, originalId, to_md5);
case "project":
return createOpenaireId(40, originalId);
return createOpenaireId(40, originalId, to_md5);
default:
return createOpenaireId(50, originalId);
return createOpenaireId(50, originalId, to_md5);
}
}