fixed collectedFrom ID

This commit is contained in:
Michele Artini 2020-04-20 11:09:27 +02:00
parent 25307965d2
commit 8ff7facfa3
4 changed files with 504 additions and 510 deletions

View File

@ -80,11 +80,11 @@ public abstract class AbstractMdRecordToOafMapper {
final String type = doc.valueOf("//dr:CobjCategory/@type");
final KeyValue collectedFrom =
keyValue(doc.valueOf("//oaf:collectedFrom/@id"), doc.valueOf("//oaf:collectedFrom/@name"));
keyValue(createOpenaireId(10, doc.valueOf("//oaf:collectedFrom/@id"), true), doc.valueOf("//oaf:collectedFrom/@name"));
final KeyValue hostedBy =
StringUtils.isBlank(doc.valueOf("//oaf:hostedBy/@id"))
? collectedFrom
: keyValue(doc.valueOf("//oaf:hostedBy/@id"), doc.valueOf("//oaf:hostedBy/@name"));
: keyValue(createOpenaireId(10, doc.valueOf("//oaf:hostedBy/@id"), true), doc.valueOf("//oaf:hostedBy/@name"));
final DataInfo info = prepareDataInfo(doc);
final long lastUpdateTimestamp = new Date().getTime();

View File

@ -1,11 +1,15 @@
package eu.dnetlib.dhp.oa.graph.raw;
import static eu.dnetlib.dhp.oa.graph.raw.common.OafMapperUtils.*;
import static eu.dnetlib.dhp.oa.graph.raw.common.OafMapperUtils.asString;
import static eu.dnetlib.dhp.oa.graph.raw.common.OafMapperUtils.createOpenaireId;
import static eu.dnetlib.dhp.oa.graph.raw.common.OafMapperUtils.dataInfo;
import static eu.dnetlib.dhp.oa.graph.raw.common.OafMapperUtils.field;
import static eu.dnetlib.dhp.oa.graph.raw.common.OafMapperUtils.journal;
import static eu.dnetlib.dhp.oa.graph.raw.common.OafMapperUtils.listFields;
import static eu.dnetlib.dhp.oa.graph.raw.common.OafMapperUtils.listKeyValues;
import static eu.dnetlib.dhp.oa.graph.raw.common.OafMapperUtils.qualifier;
import static eu.dnetlib.dhp.oa.graph.raw.common.OafMapperUtils.structuredProperty;
import eu.dnetlib.dhp.application.ArgumentApplicationParser;
import eu.dnetlib.dhp.oa.graph.raw.common.AbstractMigrationApplication;
import eu.dnetlib.dhp.oa.graph.raw.common.DbClient;
import eu.dnetlib.dhp.schema.oaf.*;
import java.io.Closeable;
import java.io.IOException;
import java.sql.Array;
@ -17,20 +21,38 @@ import java.util.Date;
import java.util.List;
import java.util.function.Consumer;
import java.util.function.Function;
import org.apache.commons.io.IOUtils;
import org.apache.commons.lang3.StringUtils;
import org.apache.commons.logging.Log;
import org.apache.commons.logging.LogFactory;
import eu.dnetlib.dhp.application.ArgumentApplicationParser;
import eu.dnetlib.dhp.oa.graph.raw.common.AbstractMigrationApplication;
import eu.dnetlib.dhp.oa.graph.raw.common.DbClient;
import eu.dnetlib.dhp.schema.oaf.Context;
import eu.dnetlib.dhp.schema.oaf.DataInfo;
import eu.dnetlib.dhp.schema.oaf.Dataset;
import eu.dnetlib.dhp.schema.oaf.Datasource;
import eu.dnetlib.dhp.schema.oaf.Field;
import eu.dnetlib.dhp.schema.oaf.Journal;
import eu.dnetlib.dhp.schema.oaf.KeyValue;
import eu.dnetlib.dhp.schema.oaf.Oaf;
import eu.dnetlib.dhp.schema.oaf.Organization;
import eu.dnetlib.dhp.schema.oaf.OtherResearchProduct;
import eu.dnetlib.dhp.schema.oaf.Project;
import eu.dnetlib.dhp.schema.oaf.Publication;
import eu.dnetlib.dhp.schema.oaf.Qualifier;
import eu.dnetlib.dhp.schema.oaf.Relation;
import eu.dnetlib.dhp.schema.oaf.Result;
import eu.dnetlib.dhp.schema.oaf.Software;
import eu.dnetlib.dhp.schema.oaf.StructuredProperty;
public class MigrateDbEntitiesApplication extends AbstractMigrationApplication
implements Closeable {
private static final Qualifier ENTITYREGISTRY_PROVENANCE_ACTION =
qualifier(
"sysimport:crosswalk:entityregistry",
"sysimport:crosswalk:entityregistry",
"dnet:provenance_actions",
"dnet:provenance_actions");
qualifier("sysimport:crosswalk:entityregistry", "sysimport:crosswalk:entityregistry", "dnet:provenance_actions", "dnet:provenance_actions");
private static final Log log = LogFactory.getLog(MigrateDbEntitiesApplication.class);
@ -41,9 +63,8 @@ public class MigrateDbEntitiesApplication extends AbstractMigrationApplication
public static void main(final String[] args) throws Exception {
final ArgumentApplicationParser parser =
new ArgumentApplicationParser(
IOUtils.toString(
MigrateDbEntitiesApplication.class.getResourceAsStream(
"/eu/dnetlib/dhp/oa/graph/migrate_db_entities_parameters.json")));
IOUtils.toString(MigrateDbEntitiesApplication.class
.getResourceAsStream("/eu/dnetlib/dhp/oa/graph/migrate_db_entities_parameters.json")));
parser.parseArgument(args);
@ -72,8 +93,7 @@ public class MigrateDbEntitiesApplication extends AbstractMigrationApplication
smdbe.execute("queryOrganizations.sql", smdbe::processOrganization);
log.info("Processing relations ds <-> orgs ...");
smdbe.execute(
"queryDatasourceOrganization.sql", smdbe::processDatasourceOrganization);
smdbe.execute("queryDatasourceOrganization.sql", smdbe::processDatasourceOrganization);
log.info("Processing projects <-> orgs ...");
smdbe.execute("queryProjectOrganization.sql", smdbe::processProjectOrganization);
@ -99,8 +119,7 @@ public class MigrateDbEntitiesApplication extends AbstractMigrationApplication
public void execute(final String sqlFile, final Function<ResultSet, List<Oaf>> producer)
throws Exception {
final String sql =
IOUtils.toString(
getClass().getResourceAsStream("/eu/dnetlib/dhp/oa/graph/sql/" + sqlFile));
IOUtils.toString(getClass().getResourceAsStream("/eu/dnetlib/dhp/oa/graph/sql/" + sqlFile));
final Consumer<ResultSet> consumer = rs -> producer.apply(rs).forEach(oaf -> emitOaf(oaf));
@ -117,17 +136,14 @@ public class MigrateDbEntitiesApplication extends AbstractMigrationApplication
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.setCollectedfrom(listKeyValues(createOpenaireId(10, rs.getString("collectedfromid"), true), rs.getString("collectedfromname")));
ds.setPid(new ArrayList<>());
ds.setDateofcollection(asString(rs.getDate("dateofcollection")));
ds.setDateoftransformation(null); // Value not returned by the SQL query
ds.setExtraInfo(new ArrayList<>()); // Values not present in the DB
ds.setOaiprovenance(null); // Values not present in the DB
ds.setDatasourcetype(prepareQualifierSplitting(rs.getString("datasourcetype")));
ds.setOpenairecompatibility(
prepareQualifierSplitting(rs.getString("openairecompatibility")));
ds.setOpenairecompatibility(prepareQualifierSplitting(rs.getString("openairecompatibility")));
ds.setOfficialname(field(rs.getString("officialname"), info));
ds.setEnglishname(field(rs.getString("englishname"), info));
ds.setWebsiteurl(field(rs.getString("websiteurl"), info));
@ -160,11 +176,7 @@ public class MigrateDbEntitiesApplication extends AbstractMigrationApplication
ds.setPidsystems(field(rs.getString("pidsystems"), info));
ds.setCertificates(field(rs.getString("certificates"), info));
ds.setPolicies(new ArrayList<>()); // The sql query returns an empty array
ds.setJournal(
prepareJournal(
rs.getString("officialname"),
rs.getString("journal"),
info)); // Journal
ds.setJournal(prepareJournal(rs.getString("officialname"), rs.getString("journal"), info)); // Journal
ds.setDataInfo(info);
ds.setLastupdatetimestamp(lastUpdateTimestamp);
@ -183,9 +195,7 @@ public class MigrateDbEntitiesApplication extends AbstractMigrationApplication
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.setCollectedfrom(listKeyValues(createOpenaireId(10, rs.getString("collectedfromid"), true), rs.getString("collectedfromname")));
p.setPid(new ArrayList<>());
p.setDateofcollection(asString(rs.getDate("dateofcollection")));
p.setDateoftransformation(asString(rs.getDate("dateoftransformation")));
@ -201,8 +211,7 @@ public class MigrateDbEntitiesApplication extends AbstractMigrationApplication
p.setKeywords(field(rs.getString("keywords"), info));
p.setDuration(field(Integer.toString(rs.getInt("duration")), info));
p.setEcsc39(field(Boolean.toString(rs.getBoolean("ecsc39")), info));
p.setOamandatepublications(
field(Boolean.toString(rs.getBoolean("oamandatepublications")), info));
p.setOamandatepublications(field(Boolean.toString(rs.getBoolean("oamandatepublications")), info));
p.setEcarticle29_3(field(Boolean.toString(rs.getBoolean("ecarticle29_3")), info));
p.setSubjects(prepareListOfStructProps(rs.getArray("subjects"), info));
p.setFundingtree(prepareListFields(rs.getArray("fundingtree"), info));
@ -238,9 +247,7 @@ public class MigrateDbEntitiesApplication extends AbstractMigrationApplication
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.setCollectedfrom(listKeyValues(createOpenaireId(10, rs.getString("collectedfromid"), true), rs.getString("collectedfromname")));
o.setPid(new ArrayList<>());
o.setDateofcollection(asString(rs.getDate("dateofcollection")));
o.setDateoftransformation(asString(rs.getDate("dateoftransformation")));
@ -254,17 +261,10 @@ public class MigrateDbEntitiesApplication extends AbstractMigrationApplication
o.setEclegalbody(field(Boolean.toString(rs.getBoolean("eclegalbody")), info));
o.setEclegalperson(field(Boolean.toString(rs.getBoolean("eclegalperson")), info));
o.setEcnonprofit(field(Boolean.toString(rs.getBoolean("ecnonprofit")), info));
o.setEcresearchorganization(
field(Boolean.toString(rs.getBoolean("ecresearchorganization")), info));
o.setEchighereducation(
field(Boolean.toString(rs.getBoolean("echighereducation")), info));
o.setEcinternationalorganizationeurinterests(
field(
Boolean.toString(
rs.getBoolean("ecinternationalorganizationeurinterests")),
info));
o.setEcinternationalorganization(
field(Boolean.toString(rs.getBoolean("ecinternationalorganization")), info));
o.setEcresearchorganization(field(Boolean.toString(rs.getBoolean("ecresearchorganization")), info));
o.setEchighereducation(field(Boolean.toString(rs.getBoolean("echighereducation")), info));
o.setEcinternationalorganizationeurinterests(field(Boolean.toString(rs.getBoolean("ecinternationalorganizationeurinterests")), info));
o.setEcinternationalorganization(field(Boolean.toString(rs.getBoolean("ecinternationalorganization")), info));
o.setEcenterprise(field(Boolean.toString(rs.getBoolean("ecenterprise")), info));
o.setEcsmevalidated(field(Boolean.toString(rs.getBoolean("ecsmevalidated")), info));
o.setEcnutscode(field(Boolean.toString(rs.getBoolean("ecnutscode")), info));
@ -284,8 +284,7 @@ public class MigrateDbEntitiesApplication extends AbstractMigrationApplication
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"));
listKeyValues(createOpenaireId(10, rs.getString("collectedfromid"), true), rs.getString("collectedfromname"));
final Relation r1 = new Relation();
r1.setRelType("datasourceOrganization");
@ -319,8 +318,7 @@ public class MigrateDbEntitiesApplication extends AbstractMigrationApplication
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"));
listKeyValues(createOpenaireId(10, rs.getString("collectedfromid"), true), rs.getString("collectedfromname"));
final Relation r1 = new Relation();
r1.setRelType("projectOrganization");
@ -351,17 +349,9 @@ public class MigrateDbEntitiesApplication extends AbstractMigrationApplication
public List<Oaf> processClaims(final ResultSet rs) {
final DataInfo info =
dataInfo(
false,
null,
false,
false,
qualifier(
"user:claim",
"user:claim",
"dnet:provenanceActions",
"dnet:provenanceActions"),
"0.9");
dataInfo(false, null, false, false, qualifier("user:claim", "user:claim", "dnet:provenanceActions", "dnet:provenanceActions"), "0.9");
final List<KeyValue> collectedFrom = listKeyValues(createOpenaireId(10, "infrastruct_::openaire", true), "OpenAIRE");
try {
@ -381,32 +371,35 @@ public class MigrateDbEntitiesApplication extends AbstractMigrationApplication
r.setLastupdatetimestamp(lastUpdateTimestamp);
r.setContext(prepareContext(rs.getString("source_id"), info));
r.setDataInfo(info);
r.setCollectedfrom(collectedFrom);
return Arrays.asList(r);
} else {
final String sourceId =
createOpenaireId(
rs.getString("source_type"), rs.getString("source_id"), false);
createOpenaireId(rs.getString("source_type"), rs.getString("source_id"), false);
final String targetId =
createOpenaireId(
rs.getString("target_type"), rs.getString("target_id"), false);
createOpenaireId(rs.getString("target_type"), rs.getString("target_id"), false);
final Relation r1 = new Relation();
final Relation r2 = new Relation();
if (rs.getString("source_type").equals("project")) {
r1.setCollectedFrom(collectedFrom);
r1.setRelType("resultProject");
r1.setSubRelType("outcome");
r1.setRelClass("produces");
r2.setCollectedFrom(collectedFrom);
r2.setRelType("resultProject");
r2.setSubRelType("outcome");
r2.setRelClass("isProducedBy");
} else {
r1.setCollectedFrom(collectedFrom);
r1.setRelType("resultResult");
r1.setSubRelType("relationship");
r1.setRelClass("isRelatedTo");
r2.setCollectedFrom(collectedFrom);
r2.setRelType("resultResult");
r2.setSubRelType("relationship");
r2.setRelClass("isRelatedTo");
@ -442,19 +435,11 @@ public class MigrateDbEntitiesApplication extends AbstractMigrationApplication
final String inferenceprovenance = rs.getString("inferenceprovenance");
final Boolean inferred = rs.getBoolean("inferred");
final String trust = rs.getString("trust");
return dataInfo(
deletedbyinference,
inferenceprovenance,
inferred,
false,
ENTITYREGISTRY_PROVENANCE_ACTION,
trust);
return dataInfo(deletedbyinference, inferenceprovenance, inferred, false, ENTITYREGISTRY_PROVENANCE_ACTION, trust);
}
private Qualifier prepareQualifierSplitting(final String s) {
if (StringUtils.isBlank(s)) {
return null;
}
if (StringUtils.isBlank(s)) { return null; }
final String[] arr = s.split("@@@");
return arr.length == 4 ? qualifier(arr[0], arr[1], arr[2], arr[3]) : null;
}
@ -470,22 +455,19 @@ public class MigrateDbEntitiesApplication extends AbstractMigrationApplication
}
private StructuredProperty prepareStructProp(final String s, final DataInfo dataInfo) {
if (StringUtils.isBlank(s)) {
return null;
}
if (StringUtils.isBlank(s)) { return null; }
final String[] parts = s.split("###");
if (parts.length == 2) {
final String value = parts[0];
final String[] arr = parts[1].split("@@@");
if (arr.length == 4) {
return structuredProperty(value, arr[0], arr[1], arr[2], arr[3], dataInfo);
}
if (arr.length == 4) { return structuredProperty(value, arr[0], arr[1], arr[2], arr[3], dataInfo); }
}
return null;
}
private List<StructuredProperty> prepareListOfStructProps(
final Array array, final DataInfo dataInfo) throws SQLException {
final Array array,
final DataInfo dataInfo) throws SQLException {
final List<StructuredProperty> res = new ArrayList<>();
if (array != null) {
for (final String s : (String[]) array.getArray()) {
@ -504,14 +486,10 @@ public class MigrateDbEntitiesApplication extends AbstractMigrationApplication
final String[] arr = sj.split("@@@");
if (arr.length == 3) {
final String issn = StringUtils.isNotBlank(arr[0]) ? arr[0] : null;
final String eissn = StringUtils.isNotBlank(arr[1]) ? arr[1] : null;
;
final String lissn = StringUtils.isNotBlank(arr[2]) ? arr[2] : null;
;
final String eissn = StringUtils.isNotBlank(arr[1]) ? arr[1] : null;;
final String lissn = StringUtils.isNotBlank(arr[2]) ? arr[2] : null;;
if (issn != null || eissn != null || lissn != null) {
return journal(
name, issn, eissn, eissn, null, null, null, null, null, null, null,
info);
return journal(name, issn, eissn, eissn, null, null, null, null, null, null, null, info);
}
}
}

View File

@ -51,6 +51,7 @@ public class MappersTest {
final Relation r2 = (Relation) list.get(2);
assertValidId(p.getId());
assertValidId(p.getCollectedfrom().get(0).getKey());
assertTrue(StringUtils.isNotBlank(p.getTitle().get(0).getValue()));
assertTrue(p.getAuthor().size() > 0);
assertTrue(p.getSubject().size() > 0);
@ -61,6 +62,8 @@ public class MappersTest {
assertValidId(r1.getTarget());
assertValidId(r2.getSource());
assertValidId(r2.getTarget());
assertValidId(r1.getCollectedFrom().get(0).getKey());
assertValidId(r2.getCollectedFrom().get(0).getKey());
assertNotNull(r1.getDataInfo());
assertNotNull(r2.getDataInfo());
assertNotNull(r1.getDataInfo().getTrust());
@ -92,6 +95,7 @@ public class MappersTest {
final Relation r2 = (Relation) list.get(2);
assertValidId(d.getId());
assertValidId(d.getCollectedfrom().get(0).getKey());
assertTrue(StringUtils.isNotBlank(d.getTitle().get(0).getValue()));
assertTrue(d.getAuthor().size() > 0);
assertTrue(d.getSubject().size() > 0);
@ -124,6 +128,7 @@ public class MappersTest {
final Software s = (Software) list.get(0);
assertValidId(s.getId());
assertValidId(s.getCollectedfrom().get(0).getKey());
assertTrue(StringUtils.isNotBlank(s.getTitle().get(0).getValue()));
assertTrue(s.getAuthor().size() > 0);
assertTrue(s.getSubject().size() > 0);

View File

@ -29,6 +29,7 @@ import eu.dnetlib.dhp.schema.oaf.Oaf;
import eu.dnetlib.dhp.schema.oaf.Organization;
import eu.dnetlib.dhp.schema.oaf.Project;
import eu.dnetlib.dhp.schema.oaf.Relation;
import eu.dnetlib.dhp.schema.oaf.Result;
@ExtendWith(MockitoExtension.class)
public class MigrateDbEntitiesApplicationTest {
@ -53,12 +54,12 @@ public class MigrateDbEntitiesApplicationTest {
final Datasource ds = (Datasource) list.get(0);
assertValidId(ds.getId());
assertValidId(ds.getCollectedfrom().get(0).getKey());
assertEquals(ds.getOfficialname().getValue(), getValueAsString("officialname", fields));
assertEquals(ds.getEnglishname().getValue(), getValueAsString("englishname", fields));
assertEquals(ds.getContactemail().getValue(), getValueAsString("contactemail", fields));
assertEquals(ds.getWebsiteurl().getValue(), getValueAsString("websiteurl", fields));
assertEquals(ds.getNamespaceprefix().getValue(), getValueAsString("namespaceprefix", fields));
assertEquals(ds.getCollectedfrom().get(0).getKey(), getValueAsString("collectedfromid", fields));
assertEquals(ds.getCollectedfrom().get(0).getValue(), getValueAsString("collectedfromname", fields));
}
@ -72,10 +73,11 @@ public class MigrateDbEntitiesApplicationTest {
final Project p = (Project) list.get(0);
assertValidId(p.getId());
assertValidId(p.getCollectedfrom().get(0).getKey());
assertEquals(p.getAcronym().getValue(), getValueAsString("acronym", fields));
assertEquals(p.getTitle().getValue(), getValueAsString("title", fields));
assertEquals(p.getCollectedfrom().get(0).getKey(), getValueAsString("collectedfromid", fields));
assertEquals(p.getCollectedfrom().get(0).getValue(), getValueAsString("collectedfromname", fields));
}
@Test
@ -90,6 +92,7 @@ public class MigrateDbEntitiesApplicationTest {
final Organization o = (Organization) list.get(0);
assertValidId(o.getId());
assertValidId(o.getCollectedfrom().get(0).getKey());
assertEquals(o.getLegalshortname().getValue(), getValueAsString("legalshortname", fields));
assertEquals(o.getLegalname().getValue(), getValueAsString("legalname", fields));
assertEquals(o.getWebsiteurl().getValue(), getValueAsString("websiteurl", fields));
@ -97,7 +100,6 @@ public class MigrateDbEntitiesApplicationTest {
assertEquals(o.getCountry().getClassname(), getValueAsString("country", fields).split("@@@")[1]);
assertEquals(o.getCountry().getSchemeid(), getValueAsString("country", fields).split("@@@")[2]);
assertEquals(o.getCountry().getSchemename(), getValueAsString("country", fields).split("@@@")[3]);
assertEquals(o.getCollectedfrom().get(0).getKey(), getValueAsString("collectedfromid", fields));
assertEquals(o.getCollectedfrom().get(0).getValue(), getValueAsString("collectedfromname", fields));
}
@ -133,6 +135,8 @@ public class MigrateDbEntitiesApplicationTest {
assertValidId(r2.getSource());
assertEquals(r1.getSource(), r2.getTarget());
assertEquals(r2.getSource(), r1.getTarget());
assertValidId(r1.getCollectedFrom().get(0).getKey());
assertValidId(r2.getCollectedFrom().get(0).getKey());
}
@Test
@ -142,8 +146,12 @@ public class MigrateDbEntitiesApplicationTest {
final List<Oaf> list = app.processClaims(rs);
assertEquals(1, list.size());
assertTrue(list.get(0) instanceof Result);
final Result r = (Result) list.get(0);
verifyMocks(fields);
assertValidId(r.getCollectedfrom().get(0).getKey());
}
@Test
@ -176,6 +184,9 @@ public class MigrateDbEntitiesApplicationTest {
assertTrue(StringUtils.isNotBlank(r1.getRelType()));
assertTrue(StringUtils.isNotBlank(r2.getRelType()));
assertValidId(r1.getCollectedFrom().get(0).getKey());
assertValidId(r2.getCollectedFrom().get(0).getKey());
// System.out.println(new ObjectMapper().writeValueAsString(r1));
// System.out.println(new ObjectMapper().writeValueAsString(r2));
}