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

fixed test

This commit is contained in:
Sandro La Bruzzo 2020-04-30 11:38:58 +02:00
parent b8e95295e2
commit 1e06bbaee8
3 changed files with 56 additions and 58 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
dhp-workflows/dhp-doiboost/src/test/java/eu/dnetlib/doiboost/CrossrefMappingTest.scala
View File

@ -22,7 +22,7 @@ class CrossrefMappingTest {
@Test //@Test
def testRelSpark() :Unit = { def testRelSpark() :Unit = {
val conf: SparkConf = new SparkConf() val conf: SparkConf = new SparkConf()
conf.set("spark.serializer", "org.apache.spark.serializer.KryoSerializer") conf.set("spark.serializer", "org.apache.spark.serializer.KryoSerializer")

28
dhp-workflows/dhp-doiboost/src/test/java/eu/dnetlib/doiboost/DoiBoostTest.java
View File

@ -1,28 +0,0 @@
package eu.dnetlib.doiboost;
import org.junit.jupiter.api.Test;
import org.slf4j.Logger;
import org.slf4j.LoggerFactory;
import com.fasterxml.jackson.databind.ObjectMapper;
import com.fasterxml.jackson.databind.SerializationFeature;
import eu.dnetlib.doiboost.crossref.SparkMapDumpIntoOAF;
public class DoiBoostTest {
final ObjectMapper mapper = new ObjectMapper().enable(SerializationFeature.INDENT_OUTPUT);
Logger logger = LoggerFactory.getLogger(DoiBoostTest.class);
@Test
public void test() throws Exception {
// SparkDownloadContentFromCrossref.main(null);
// CrossrefImporter.main("-n file:///tmp -t file:///tmp/p.seq -ts 1586110000749".split("
// "));
SparkMapDumpIntoOAF
.main(
"-m local[*] -s file:///data/doiboost/crossref_dump.seq -t /data/doiboost".split(" "));
}
}

84
dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/dhp/oa/graph/raw/MigrateDbEntitiesApplication.java
View File

@ -51,7 +51,7 @@ import eu.dnetlib.dhp.schema.oaf.Software;
import eu.dnetlib.dhp.schema.oaf.StructuredProperty; import eu.dnetlib.dhp.schema.oaf.StructuredProperty;
public class MigrateDbEntitiesApplication extends AbstractMigrationApplication public class MigrateDbEntitiesApplication extends AbstractMigrationApplication
implements Closeable { implements Closeable {
private static final Log log = LogFactory.getLog(MigrateDbEntitiesApplication.class); private static final Log log = LogFactory.getLog(MigrateDbEntitiesApplication.class);
@ -61,9 +61,10 @@ public class MigrateDbEntitiesApplication extends AbstractMigrationApplication
public static void main(final String[] args) throws Exception { public static void main(final String[] args) throws Exception {
final ArgumentApplicationParser parser = new ArgumentApplicationParser( final ArgumentApplicationParser parser = new ArgumentApplicationParser(
IOUtils IOUtils
.toString(MigrateDbEntitiesApplication.class .toString(
.getResourceAsStream("/eu/dnetlib/dhp/oa/graph/migrate_db_entities_parameters.json"))); MigrateDbEntitiesApplication.class
.getResourceAsStream("/eu/dnetlib/dhp/oa/graph/migrate_db_entities_parameters.json")));
parser.parseArgument(args); parser.parseArgument(args);
@ -76,7 +77,7 @@ public class MigrateDbEntitiesApplication extends AbstractMigrationApplication
final boolean processClaims = parser.get("action") != null && parser.get("action").equalsIgnoreCase("claims"); final boolean processClaims = parser.get("action") != null && parser.get("action").equalsIgnoreCase("claims");
try (final MigrateDbEntitiesApplication smdbe = new MigrateDbEntitiesApplication(hdfsPath, dbUrl, dbUser, try (final MigrateDbEntitiesApplication smdbe = new MigrateDbEntitiesApplication(hdfsPath, dbUrl, dbUser,
dbPassword)) { dbPassword)) {
if (processClaims) { if (processClaims) {
log.info("Processing claims..."); log.info("Processing claims...");
smdbe.execute("queryClaims.sql", smdbe::processClaims); smdbe.execute("queryClaims.sql", smdbe::processClaims);
@ -107,15 +108,15 @@ public class MigrateDbEntitiesApplication extends AbstractMigrationApplication
} }
public MigrateDbEntitiesApplication( public MigrateDbEntitiesApplication(
final String hdfsPath, final String dbUrl, final String dbUser, final String dbPassword) final String hdfsPath, final String dbUrl, final String dbUser, final String dbPassword)
throws Exception { throws Exception {
super(hdfsPath); super(hdfsPath);
this.dbClient = new DbClient(dbUrl, dbUser, dbPassword); this.dbClient = new DbClient(dbUrl, dbUser, dbPassword);
this.lastUpdateTimestamp = new Date().getTime(); this.lastUpdateTimestamp = new Date().getTime();
} }
public void execute(final String sqlFile, final Function<ResultSet, List<Oaf>> producer) public void execute(final String sqlFile, final Function<ResultSet, List<Oaf>> producer)
throws Exception { throws Exception {
final String sql = IOUtils.toString(getClass().getResourceAsStream("/eu/dnetlib/dhp/oa/graph/sql/" + sqlFile)); final String sql = IOUtils.toString(getClass().getResourceAsStream("/eu/dnetlib/dhp/oa/graph/sql/" + sqlFile));
final Consumer<ResultSet> consumer = rs -> producer.apply(rs).forEach(oaf -> emitOaf(oaf)); final Consumer<ResultSet> consumer = rs -> producer.apply(rs).forEach(oaf -> emitOaf(oaf));
@ -134,7 +135,10 @@ public class MigrateDbEntitiesApplication extends AbstractMigrationApplication
ds.setId(createOpenaireId(10, rs.getString("datasourceid"), true)); ds.setId(createOpenaireId(10, rs.getString("datasourceid"), true));
ds.setOriginalId(Arrays.asList(rs.getString("datasourceid"))); ds.setOriginalId(Arrays.asList(rs.getString("datasourceid")));
ds ds
.setCollectedfrom(listKeyValues(createOpenaireId(10, rs.getString("collectedfromid"), true), rs.getString("collectedfromname"))); .setCollectedfrom(
listKeyValues(
createOpenaireId(10, rs.getString("collectedfromid"), true),
rs.getString("collectedfromname")));
ds.setPid(new ArrayList<>()); ds.setPid(new ArrayList<>());
ds.setDateofcollection(asString(rs.getDate("dateofcollection"))); ds.setDateofcollection(asString(rs.getDate("dateofcollection")));
ds.setDateoftransformation(null); // Value not returned by the SQL query ds.setDateoftransformation(null); // Value not returned by the SQL query
@ -175,7 +179,7 @@ public class MigrateDbEntitiesApplication extends AbstractMigrationApplication
ds.setCertificates(field(rs.getString("certificates"), info)); ds.setCertificates(field(rs.getString("certificates"), info));
ds.setPolicies(new ArrayList<>()); // The sql query returns an empty array ds.setPolicies(new ArrayList<>()); // The sql query returns an empty array
ds ds
.setJournal(prepareJournal(rs.getString("officialname"), rs.getString("journal"), info)); // Journal .setJournal(prepareJournal(rs.getString("officialname"), rs.getString("journal"), info)); // Journal
ds.setDataInfo(info); ds.setDataInfo(info);
ds.setLastupdatetimestamp(lastUpdateTimestamp); ds.setLastupdatetimestamp(lastUpdateTimestamp);
@ -195,7 +199,10 @@ public class MigrateDbEntitiesApplication extends AbstractMigrationApplication
p.setId(createOpenaireId(40, rs.getString("projectid"), true)); p.setId(createOpenaireId(40, rs.getString("projectid"), true));
p.setOriginalId(Arrays.asList(rs.getString("projectid"))); p.setOriginalId(Arrays.asList(rs.getString("projectid")));
p p
.setCollectedfrom(listKeyValues(createOpenaireId(10, rs.getString("collectedfromid"), true), rs.getString("collectedfromname"))); .setCollectedfrom(
listKeyValues(
createOpenaireId(10, rs.getString("collectedfromid"), true),
rs.getString("collectedfromname")));
p.setPid(new ArrayList<>()); p.setPid(new ArrayList<>());
p.setDateofcollection(asString(rs.getDate("dateofcollection"))); p.setDateofcollection(asString(rs.getDate("dateofcollection")));
p.setDateoftransformation(asString(rs.getDate("dateoftransformation"))); p.setDateoftransformation(asString(rs.getDate("dateoftransformation")));
@ -212,7 +219,7 @@ public class MigrateDbEntitiesApplication extends AbstractMigrationApplication
p.setDuration(field(Integer.toString(rs.getInt("duration")), info)); p.setDuration(field(Integer.toString(rs.getInt("duration")), info));
p.setEcsc39(field(Boolean.toString(rs.getBoolean("ecsc39")), info)); p.setEcsc39(field(Boolean.toString(rs.getBoolean("ecsc39")), info));
p p
.setOamandatepublications(field(Boolean.toString(rs.getBoolean("oamandatepublications")), info)); .setOamandatepublications(field(Boolean.toString(rs.getBoolean("oamandatepublications")), info));
p.setEcarticle29_3(field(Boolean.toString(rs.getBoolean("ecarticle29_3")), info)); p.setEcarticle29_3(field(Boolean.toString(rs.getBoolean("ecarticle29_3")), info));
p.setSubjects(prepareListOfStructProps(rs.getArray("subjects"), info)); p.setSubjects(prepareListOfStructProps(rs.getArray("subjects"), info));
p.setFundingtree(prepareListFields(rs.getArray("fundingtree"), info)); p.setFundingtree(prepareListFields(rs.getArray("fundingtree"), info));
@ -249,7 +256,10 @@ public class MigrateDbEntitiesApplication extends AbstractMigrationApplication
o.setId(createOpenaireId(20, rs.getString("organizationid"), true)); o.setId(createOpenaireId(20, rs.getString("organizationid"), true));
o.setOriginalId(Arrays.asList(rs.getString("organizationid"))); o.setOriginalId(Arrays.asList(rs.getString("organizationid")));
o o
.setCollectedfrom(listKeyValues(createOpenaireId(10, rs.getString("collectedfromid"), true), rs.getString("collectedfromname"))); .setCollectedfrom(
listKeyValues(
createOpenaireId(10, rs.getString("collectedfromid"), true),
rs.getString("collectedfromname")));
o.setPid(new ArrayList<>()); o.setPid(new ArrayList<>());
o.setDateofcollection(asString(rs.getDate("dateofcollection"))); o.setDateofcollection(asString(rs.getDate("dateofcollection")));
o.setDateoftransformation(asString(rs.getDate("dateoftransformation"))); o.setDateoftransformation(asString(rs.getDate("dateoftransformation")));
@ -264,12 +274,14 @@ public class MigrateDbEntitiesApplication extends AbstractMigrationApplication
o.setEclegalperson(field(Boolean.toString(rs.getBoolean("eclegalperson")), info)); o.setEclegalperson(field(Boolean.toString(rs.getBoolean("eclegalperson")), info));
o.setEcnonprofit(field(Boolean.toString(rs.getBoolean("ecnonprofit")), info)); o.setEcnonprofit(field(Boolean.toString(rs.getBoolean("ecnonprofit")), info));
o o
.setEcresearchorganization(field(Boolean.toString(rs.getBoolean("ecresearchorganization")), info)); .setEcresearchorganization(field(Boolean.toString(rs.getBoolean("ecresearchorganization")), info));
o.setEchighereducation(field(Boolean.toString(rs.getBoolean("echighereducation")), info)); o.setEchighereducation(field(Boolean.toString(rs.getBoolean("echighereducation")), info));
o o
.setEcinternationalorganizationeurinterests(field(Boolean.toString(rs.getBoolean("ecinternationalorganizationeurinterests")), info)); .setEcinternationalorganizationeurinterests(
field(Boolean.toString(rs.getBoolean("ecinternationalorganizationeurinterests")), info));
o o
.setEcinternationalorganization(field(Boolean.toString(rs.getBoolean("ecinternationalorganization")), info)); .setEcinternationalorganization(
field(Boolean.toString(rs.getBoolean("ecinternationalorganization")), info));
o.setEcenterprise(field(Boolean.toString(rs.getBoolean("ecenterprise")), info)); o.setEcenterprise(field(Boolean.toString(rs.getBoolean("ecenterprise")), info));
o.setEcsmevalidated(field(Boolean.toString(rs.getBoolean("ecsmevalidated")), info)); o.setEcsmevalidated(field(Boolean.toString(rs.getBoolean("ecsmevalidated")), info));
o.setEcnutscode(field(Boolean.toString(rs.getBoolean("ecnutscode")), info)); o.setEcnutscode(field(Boolean.toString(rs.getBoolean("ecnutscode")), info));
@ -288,7 +300,8 @@ public class MigrateDbEntitiesApplication extends AbstractMigrationApplication
final DataInfo info = prepareDataInfo(rs); final DataInfo info = prepareDataInfo(rs);
final String orgId = createOpenaireId(20, rs.getString("organization"), true); final String orgId = createOpenaireId(20, rs.getString("organization"), true);
final String dsId = createOpenaireId(10, rs.getString("datasource"), true); final String dsId = createOpenaireId(10, rs.getString("datasource"), true);
final List<KeyValue> collectedFrom = listKeyValues(createOpenaireId(10, rs.getString("collectedfromid"), true), rs.getString("collectedfromname")); final List<KeyValue> collectedFrom = listKeyValues(
createOpenaireId(10, rs.getString("collectedfromid"), true), rs.getString("collectedfromname"));
final Relation r1 = new Relation(); final Relation r1 = new Relation();
r1.setRelType("datasourceOrganization"); r1.setRelType("datasourceOrganization");
@ -321,7 +334,8 @@ public class MigrateDbEntitiesApplication extends AbstractMigrationApplication
final DataInfo info = prepareDataInfo(rs); final DataInfo info = prepareDataInfo(rs);
final String orgId = createOpenaireId(20, rs.getString("resporganization"), true); final String orgId = createOpenaireId(20, rs.getString("resporganization"), true);
final String projectId = createOpenaireId(40, rs.getString("project"), true); final String projectId = createOpenaireId(40, rs.getString("project"), true);
final List<KeyValue> collectedFrom = listKeyValues(createOpenaireId(10, rs.getString("collectedfromid"), true), rs.getString("collectedfromname")); final List<KeyValue> collectedFrom = listKeyValues(
createOpenaireId(10, rs.getString("collectedfromid"), true), rs.getString("collectedfromname"));
final Relation r1 = new Relation(); final Relation r1 = new Relation();
r1.setRelType("projectOrganization"); r1.setRelType("projectOrganization");
@ -351,10 +365,12 @@ public class MigrateDbEntitiesApplication extends AbstractMigrationApplication
public List<Oaf> processClaims(final ResultSet rs) { public List<Oaf> processClaims(final ResultSet rs) {
final DataInfo info = final DataInfo info = dataInfo(
dataInfo(false, null, false, false, qualifier("user:claim", "user:claim", "dnet:provenanceActions", "dnet:provenanceActions"), "0.9"); 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"); final List<KeyValue> collectedFrom = listKeyValues(
createOpenaireId(10, "infrastruct_::openaire", true), "OpenAIRE");
try { try {
@ -440,11 +456,15 @@ public class MigrateDbEntitiesApplication extends AbstractMigrationApplication
final String inferenceprovenance = rs.getString("inferenceprovenance"); final String inferenceprovenance = rs.getString("inferenceprovenance");
final Boolean inferred = rs.getBoolean("inferred"); final Boolean inferred = rs.getBoolean("inferred");
final String trust = rs.getString("trust"); final String trust = rs.getString("trust");
return dataInfo(deletedbyinference, inferenceprovenance, inferred, false, MigrationConstants.ENTITYREGISTRY_PROVENANCE_ACTION, trust); return dataInfo(
deletedbyinference, inferenceprovenance, inferred, false,
MigrationConstants.ENTITYREGISTRY_PROVENANCE_ACTION, trust);
} }
private Qualifier prepareQualifierSplitting(final String s) { private Qualifier prepareQualifierSplitting(final String s) {
if (StringUtils.isBlank(s)) { return null; } if (StringUtils.isBlank(s)) {
return null;
}
final String[] arr = s.split("@@@"); final String[] arr = s.split("@@@");
return arr.length == 4 ? qualifier(arr[0], arr[1], arr[2], arr[3]) : null; return arr.length == 4 ? qualifier(arr[0], arr[1], arr[2], arr[3]) : null;
} }
@ -458,19 +478,23 @@ public class MigrateDbEntitiesApplication extends AbstractMigrationApplication
} }
private StructuredProperty prepareStructProp(final String s, final DataInfo dataInfo) { 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("###"); final String[] parts = s.split("###");
if (parts.length == 2) { if (parts.length == 2) {
final String value = parts[0]; final String value = parts[0];
final String[] arr = parts[1].split("@@@"); 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; return null;
} }
private List<StructuredProperty> prepareListOfStructProps( private List<StructuredProperty> prepareListOfStructProps(
final Array array, final Array array,
final DataInfo dataInfo) throws SQLException { final DataInfo dataInfo) throws SQLException {
final List<StructuredProperty> res = new ArrayList<>(); final List<StructuredProperty> res = new ArrayList<>();
if (array != null) { if (array != null) {
for (final String s : (String[]) array.getArray()) { for (final String s : (String[]) array.getArray()) {
@ -489,8 +513,10 @@ public class MigrateDbEntitiesApplication extends AbstractMigrationApplication
final String[] arr = sj.split("@@@"); final String[] arr = sj.split("@@@");
if (arr.length == 3) { if (arr.length == 3) {
final String issn = StringUtils.isNotBlank(arr[0]) ? arr[0].trim() : null; final String issn = StringUtils.isNotBlank(arr[0]) ? arr[0].trim() : null;
final String eissn = StringUtils.isNotBlank(arr[1]) ? arr[1].trim() : null;; final String eissn = StringUtils.isNotBlank(arr[1]) ? arr[1].trim() : null;
final String lissn = StringUtils.isNotBlank(arr[2]) ? arr[2].trim() : null;; ;
final String lissn = StringUtils.isNotBlank(arr[2]) ? arr[2].trim() : null;
;
if (issn != null || eissn != null || lissn != 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);
} }