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

-

This commit is contained in:
Miriam Baglioni 2020-04-30 12:45:28 +02:00
parent 2349bfd8b8
commit 4b0bd91012
7 changed files with 144 additions and 41 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

7
: Normal file
View File

@ -0,0 +1,7 @@
Merge remote-tracking branch 'upstream/master'
# Please enter a commit message to explain why this merge is necessary,
# especially if it merges an updated upstream into a topic branch.
#
# Lines starting with '#' will be ignored, and an empty message aborts
# the commit.

7
dhp-schemas/src/main/java/eu/dnetlib/dhp/schema/common/ModelSupport.java
View File

@ -1,11 +1,14 @@
package eu.dnetlib.dhp.schema.common;
import com.google.common.collect.Maps;
import eu.dnetlib.dhp.schema.oaf.*;
import java.util.Map;
import java.util.Optional;
import java.util.function.Function;
import com.google.common.collect.Maps;
import eu.dnetlib.dhp.schema.oaf.*;
/** Oaf model utility methods. */
public class ModelSupport {

85
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;
public class MigrateDbEntitiesApplication extends AbstractMigrationApplication
implements Closeable {
implements Closeable {
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 {
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);
@ -76,7 +77,7 @@ public class MigrateDbEntitiesApplication extends AbstractMigrationApplication
final boolean processClaims = parser.get("action") != null && parser.get("action").equalsIgnoreCase("claims");
try (final MigrateDbEntitiesApplication smdbe = new MigrateDbEntitiesApplication(hdfsPath, dbUrl, dbUser,
dbPassword)) {
dbPassword)) {
if (processClaims) {
log.info("Processing claims...");
smdbe.execute("queryClaims.sql", smdbe::processClaims);
@ -107,15 +108,15 @@ public class MigrateDbEntitiesApplication extends AbstractMigrationApplication
}
public MigrateDbEntitiesApplication(
final String hdfsPath, final String dbUrl, final String dbUser, final String dbPassword)
throws Exception {
final String hdfsPath, final String dbUrl, final String dbUser, final String dbPassword)
throws Exception {
super(hdfsPath);
this.dbClient = new DbClient(dbUrl, dbUser, dbPassword);
this.lastUpdateTimestamp = new Date().getTime();
}
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 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.setOriginalId(Arrays.asList(rs.getString("datasourceid")));
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.setDateofcollection(asString(rs.getDate("dateofcollection")));
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.setPolicies(new ArrayList<>()); // The sql query returns an empty array
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.setLastupdatetimestamp(lastUpdateTimestamp);
@ -195,7 +199,10 @@ public class MigrateDbEntitiesApplication extends AbstractMigrationApplication
p.setId(createOpenaireId(40, rs.getString("projectid"), true));
p.setOriginalId(Arrays.asList(rs.getString("projectid")));
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.setDateofcollection(asString(rs.getDate("dateofcollection")));
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.setEcsc39(field(Boolean.toString(rs.getBoolean("ecsc39")), info));
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.setSubjects(prepareListOfStructProps(rs.getArray("subjects"), 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.setOriginalId(Arrays.asList(rs.getString("organizationid")));
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.setDateofcollection(asString(rs.getDate("dateofcollection")));
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.setEcnonprofit(field(Boolean.toString(rs.getBoolean("ecnonprofit")), info));
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
.setEcinternationalorganizationeurinterests(field(Boolean.toString(rs.getBoolean("ecinternationalorganizationeurinterests")), info));
.setEcinternationalorganizationeurinterests(
field(Boolean.toString(rs.getBoolean("ecinternationalorganizationeurinterests")), info));
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.setEcsmevalidated(field(Boolean.toString(rs.getBoolean("ecsmevalidated")), 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 String orgId = createOpenaireId(20, rs.getString("organization"), 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();
r1.setRelType("datasourceOrganization");
@ -321,7 +334,8 @@ public class MigrateDbEntitiesApplication extends AbstractMigrationApplication
final DataInfo info = prepareDataInfo(rs);
final String orgId = createOpenaireId(20, rs.getString("resporganization"), 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();
r1.setRelType("projectOrganization");
@ -351,10 +365,12 @@ 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");
final DataInfo info = 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");
final List<KeyValue> collectedFrom = listKeyValues(
createOpenaireId(10, "infrastruct_::openaire", true), "OpenAIRE");
try {
@ -440,11 +456,15 @@ 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, MigrationConstants.ENTITYREGISTRY_PROVENANCE_ACTION, trust);
return dataInfo(
deletedbyinference, inferenceprovenance, inferred, false,
MigrationConstants.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;
}
@ -458,19 +478,23 @@ 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()) {
@ -489,8 +513,10 @@ public class MigrateDbEntitiesApplication extends AbstractMigrationApplication
final String[] arr = sj.split("@@@");
if (arr.length == 3) {
final String issn = StringUtils.isNotBlank(arr[0]) ? arr[0].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 eissn = StringUtils.isNotBlank(arr[1]) ? arr[1].trim() : null;
;
final String lissn = StringUtils.isNotBlank(arr[2]) ? arr[2].trim() : null;
;
if (issn != null || eissn != null || lissn != null) {
return journal(name, issn, eissn, eissn, null, null, null, null, null, null, null, info);
}
@ -504,5 +530,4 @@ public class MigrateDbEntitiesApplication extends AbstractMigrationApplication
super.close();
dbClient.close();
}
}

1
dhp-workflows/dhp-propagation/src/main/java/eu/dnetlib/dhp/resulttoorganizationfrominstrepo/SparkResultToOrganizationFromIstRepoJob2.java
View File

@ -99,6 +99,7 @@ public class SparkResultToOrganizationFromIstRepoJob2 {
String outputPath,
Class<? extends Result> resultClazz,
String resultType,
Boolean saveGraph) {
final JavaSparkContext sc = new JavaSparkContext(spark.sparkContext());

6
dhp-workflows/dhp-propagation/src/main/resources/eu/dnetlib/dhp/resulttocommunityfromsemrel/input_communitytoresult_parameters.json
View File

@ -6,6 +6,12 @@
"paramDescription": "the path of the sequencial file to read",
"paramRequired": true
},
{
"paramName":"sg",
"paramLongName":"saveGraph",
"paramDescription": "true if the new version of the graph must be saved",
"paramRequired": false
},
{
"paramName":"h",
"paramLongName":"hive_metastore_uris",

76
dhp-workflows/dhp-propagation/src/main/resources/eu/dnetlib/dhp/resulttocommunityfromsemrel/oozie_app/workflow.xml
View File

@ -12,6 +12,10 @@
<name>isLookupUrl</name>
<description>the isLookup service endpoint</description>
</property>
<property>
<name>outputPath</name>
<description>the output path</description>
</property>
</parameters>
<start to="reset-outputpath"/>
@ -21,22 +25,72 @@
</kill>
<action name="reset-outputpath">
<fs>
<delete path='${workingDir}/resulttocommunitysemrel_propagation'/>
<delete path='${outputPath}/relation'/>
<delete path='${outputPath}/datasource'/>
<delete path='${outputPath}/organization'/>
<delete path='${outputPath}/project'/>
<delete path='${outputPath}/publication'/>
<delete path='${outputPath}/dataset'/>
<delete path='${outputPath}/software'/>
<delete path='${outputPath}/otherresearchproduct'/>
<delete path='${workingDir}/preparedInfo'/>
</fs>
<ok to="copy_relation"/>
<ok to="copy_entities"/>
<error to="Kill"/>
</action>
<fork name="copy_entities">
<path start="copy_relation"/>
<path start="copy_organization"/>
<path start="copy_projects"/>
<path start="copy_datasources"/>
</fork>
<action name="copy_relation">
<distcp xmlns="uri:oozie:distcp-action:0.2">
<job-tracker>${jobTracker}</job-tracker>
<name-node>${nameNode}</name-node>
<arg>${nameNode}/${sourcePath}/relation</arg>
<arg>${nameNode}/${workingDir}/projecttoresult_propagation/relation</arg>
<arg>${nameNode}/${outputPath}/relation</arg>
</distcp>
<ok to="fork_prepare_assoc_step1"/>
<ok to="copy_wait"/>
<error to="Kill"/>
</action>
<action name="copy_organization">
<distcp xmlns="uri:oozie:distcp-action:0.2">
<job-tracker>${jobTracker}</job-tracker>
<name-node>${nameNode}</name-node>
<arg>${nameNode}/${sourcePath}/organization</arg>
<arg>${nameNode}/${outputPath}/organization</arg>
</distcp>
<ok to="copy_wait"/>
<error to="Kill"/>
</action>
<action name="copy_projects">
<distcp xmlns="uri:oozie:distcp-action:0.2">
<job-tracker>${jobTracker}</job-tracker>
<name-node>${nameNode}</name-node>
<arg>${nameNode}/${sourcePath}/project</arg>
<arg>${nameNode}/${outputPath}/project</arg>
</distcp>
<ok to="copy_wait"/>
<error to="Kill"/>
</action>
<action name="copy_datasources">
<distcp xmlns="uri:oozie:distcp-action:0.2">
<job-tracker>${jobTracker}</job-tracker>
<name-node>${nameNode}</name-node>
<arg>${nameNode}/${sourcePath}/datasource</arg>
<arg>${nameNode}/${outputPath}/datasource</arg>
</distcp>
<ok to="copy_wait"/>
<error to="Kill"/>
</action>
<join name="copy_wait" to="fork_prepare_assoc_step1"/>
<fork name="fork_prepare_assoc_step1">
<path start="join_prepare_publication"/>
<path start="join_prepare_dataset"/>
@ -67,7 +121,7 @@
<arg>--resultTableName</arg><arg>eu.dnetlib.dhp.schema.oaf.Publication</arg>
<arg>--outputPath</arg><arg>${workingDir}/preparedInfo/targetCommunityAssoc</arg>
<arg>--allowedsemrels</arg><arg>${allowedsemrels}</arg>
<arg>--isLookupUrl</arg><arg>${isLookupUrl}</arg>
<arg>--isLookupUrl</arg><arg>${isLookUpUrl}</arg>
</spark>
<ok to="wait"/>
<error to="Kill"/>
@ -215,7 +269,8 @@
<arg>--sourcePath</arg><arg>${sourcePath}/publication</arg>
<arg>--hive_metastore_uris</arg><arg>${hive_metastore_uris}</arg>
<arg>--resultTableName</arg><arg>eu.dnetlib.dhp.schema.oaf.Publication</arg>
<arg>--outputPath</arg><arg>${workingDir}/publication</arg>
<arg>--outputPath</arg><arg>${outputPath}/publication</arg>
<arg>--saveGrap</arg><arg>${saveGraph</arg>
</spark>
<ok to="wait2"/>
<error to="Kill"/>
@ -242,7 +297,8 @@
<arg>--sourcePath</arg><arg>${sourcePath}/dataset</arg>
<arg>--hive_metastore_uris</arg><arg>${hive_metastore_uris}</arg>
<arg>--resultTableName</arg><arg>eu.dnetlib.dhp.schema.oaf.Dataset</arg>
<arg>--outputPath</arg><arg>${workingDir}/dataset</arg>
<arg>--outputPath</arg><arg>${outputPath}/dataset</arg>
<arg>--saveGrap</arg><arg>${saveGraph</arg>
</spark>
<ok to="wait2"/>
<error to="Kill"/>
@ -269,7 +325,8 @@
<arg>--sourcePath</arg><arg>${sourcePath}/otherresearchproduct</arg>
<arg>--hive_metastore_uris</arg><arg>${hive_metastore_uris}</arg>
<arg>--resultTableName</arg><arg>eu.dnetlib.dhp.schema.oaf.OtherResearchProduct</arg>
<arg>--outputPath</arg><arg>${workingDir}/otherresearchproduct</arg>
<arg>--outputPath</arg><arg>${outputPath}/otherresearchproduct</arg>
<arg>--saveGrap</arg><arg>${saveGraph</arg>
</spark>
<ok to="wait2"/>
<error to="Kill"/>
@ -296,7 +353,8 @@
<arg>--sourcePath</arg><arg>${sourcePath}/software</arg>
<arg>--hive_metastore_uris</arg><arg>${hive_metastore_uris}</arg>
<arg>--resultTableName</arg><arg>eu.dnetlib.dhp.schema.oaf.Software</arg>
<arg>--outputPath</arg><arg>${workingDir}/software</arg>
<arg>--outputPath</arg><arg>${outputPath}/software</arg>
<arg>--saveGrap</arg><arg>${saveGraph</arg>
</spark>
<ok to="wait2"/>
<error to="Kill"/>

3
dhp-workflows/dhp-propagation/src/test/java/eu/dnetlib/dhp/resulttoorganizationfrominstrepo/Result2OrganizationJobTest.java
View File

@ -88,6 +88,7 @@ public class Result2OrganizationJobTest {
"",
"-resultTableName",
"eu.dnetlib.dhp.schema.oaf.Software",
"-saveGraph",
"true",
"-outputPath",
@ -138,6 +139,7 @@ public class Result2OrganizationJobTest {
"",
"-resultTableName",
"eu.dnetlib.dhp.schema.oaf.Software",
"-saveGraph",
"true",
"-outputPath",
@ -211,6 +213,7 @@ public class Result2OrganizationJobTest {
"",
"-resultTableName",
"eu.dnetlib.dhp.schema.oaf.Software",
"-saveGraph",
"true",
"-outputPath",