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

Merge branch 'beta' into 8232-mdstore-synch-improve

This commit is contained in:
Claudio Atzori 2023-02-22 09:57:07 +01:00
parent 8920932dd8 42b6b5d5ce
commit cd3a51a15f
25 changed files with 2334 additions and 1610 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

79
dhp-workflows/dhp-aggregation/src/main/java/eu/dnetlib/dhp/actionmanager/usagestats/SparkAtomicActionUsageJob.java
View File

@ -8,13 +8,13 @@ import java.io.Serializable;
import java.util.Arrays;
import java.util.List;
import java.util.Optional;
import eu.dnetlib.dhp.schema.oaf.*;
import org.apache.commons.io.IOUtils;
import org.apache.hadoop.io.Text;
import org.apache.hadoop.mapred.SequenceFileOutputFormat;
import org.apache.spark.SparkConf;
import org.apache.spark.api.java.function.MapFunction;
import org.apache.spark.api.java.function.MapGroupsFunction;
import org.apache.spark.sql.Dataset;
import org.apache.spark.sql.Encoders;
import org.apache.spark.sql.SaveMode;
@ -28,9 +28,6 @@ import eu.dnetlib.dhp.application.ArgumentApplicationParser;
import eu.dnetlib.dhp.common.HdfsSupport;
import eu.dnetlib.dhp.schema.action.AtomicAction;
import eu.dnetlib.dhp.schema.common.ModelConstants;
import eu.dnetlib.dhp.schema.oaf.DataInfo;
import eu.dnetlib.dhp.schema.oaf.Measure;
import eu.dnetlib.dhp.schema.oaf.Result;
import eu.dnetlib.dhp.schema.oaf.utils.OafMapperUtils;
import scala.Tuple2;
@ -76,16 +73,19 @@ public class SparkAtomicActionUsageJob implements Serializable {
isSparkSessionManaged,
spark -> {
removeOutputDir(spark, outputPath);
prepareResults(dbname, spark, workingPath);
prepareData(dbname, spark, workingPath + "/usageDb", "usage_stats", "result_id");
prepareData(dbname, spark, workingPath + "/projectDb", "project_stats", "id");
prepareData(dbname, spark, workingPath + "/datasourceDb", "datasource_stats", "repositor_id");
writeActionSet(spark, workingPath, outputPath);
});
}
public static void prepareResults(String db, SparkSession spark, String workingPath) {
private static void prepareData(String dbname, SparkSession spark, String workingPath, String tableName, String attribute_name) {
spark
.sql(
"Select result_id, downloads, views " +
"from " + db + ".usage_stats")
"Select " + attribute_name + " as id, sum(downloads) as downloads, sum(views) as views " +
"from " + dbname + "." + tableName +
"group by " + attribute_name)
.as(Encoders.bean(UsageStatsModel.class))
.write()
.mode(SaveMode.Overwrite)
@ -93,24 +93,18 @@ public class SparkAtomicActionUsageJob implements Serializable {
.json(workingPath);
}
public static void writeActionSet(SparkSession spark, String inputPath, String outputPath) {
readPath(spark, inputPath, UsageStatsModel.class)
.groupByKey((MapFunction<UsageStatsModel, String>) us -> us.getResult_id(), Encoders.STRING())
.mapGroups((MapGroupsFunction<String, UsageStatsModel, Result>) (k, it) -> {
UsageStatsModel first = it.next();
it.forEachRemaining(us -> {
first.setDownloads(first.getDownloads() + us.getDownloads());
first.setViews(first.getViews() + us.getViews());
});
Result res = new Result();
res.setId("50|" + k);
res.setMeasures(getMeasure(first.getDownloads(), first.getViews()));
return res;
}, Encoders.bean(Result.class))
getFinalIndicatorsResult(spark, inputPath+ "/usageDb").
toJavaRDD().
map(p -> new AtomicAction(p.getClass(),p))
.union(getFinalIndicatorsProject(spark, inputPath + "/projectDb")
.toJavaRDD()
.map(p -> new AtomicAction(p.getClass(), p))
.map(p -> new AtomicAction(p.getClass(), p )))
.union(getFinalIndicatorsDatasource(spark, inputPath + "/datasourceDb")
.toJavaRDD()
.map(p -> new AtomicAction(p.getClass(), p)))
.mapToPair(
aa -> new Tuple2<>(new Text(aa.getClazz().getCanonicalName()),
new Text(OBJECT_MAPPER.writeValueAsString(aa))))
@ -118,6 +112,41 @@ public class SparkAtomicActionUsageJob implements Serializable {
}
private static Dataset<Result> getFinalIndicatorsResult(SparkSession spark, String inputPath) {
return readPath(spark, inputPath, UsageStatsModel.class)
.map((MapFunction<UsageStatsModel, Result>) usm -> {
Result r = new Result();
r.setId("50|" + usm.getId());
r.setMeasures(getMeasure(usm.getDownloads(), usm.getViews()));
return r;
}, Encoders.bean(Result.class));
}
private static Dataset<Project> getFinalIndicatorsProject(SparkSession spark, String inputPath) {
return readPath(spark, inputPath, UsageStatsModel.class)
.map((MapFunction<UsageStatsModel, Project>) usm -> {
Project r = new Project();
r.setId("40|" + usm.getId());
r.setMeasures(getMeasure(usm.getDownloads(), usm.getViews()));
return r;
}, Encoders.bean(Project.class));
}
private static Dataset<Datasource> getFinalIndicatorsDatasource(SparkSession spark, String inputPath) {
return readPath(spark, inputPath, UsageStatsModel.class)
.map((MapFunction<UsageStatsModel, Datasource>) usm -> {
Datasource r = new Datasource();
r.setId("10|" + usm.getId());
r.setMeasures(getMeasure(usm.getDownloads(), usm.getViews()));
return r;
}, Encoders.bean(Datasource.class));
}
private static List<Measure> getMeasure(Long downloads, Long views) {
DataInfo dataInfo = OafMapperUtils
.dataInfo(

10
dhp-workflows/dhp-aggregation/src/main/java/eu/dnetlib/dhp/actionmanager/usagestats/UsageStatsModel.java
View File

@ -4,16 +4,16 @@ package eu.dnetlib.dhp.actionmanager.usagestats;
import java.io.Serializable;
public class UsageStatsModel implements Serializable {
private String result_id;
private String id;
private Long downloads;
private Long views;
public String getResult_id() {
return result_id;
public String getId() {
return id;
}
public void setResult_id(String result_id) {
this.result_id = result_id;
public void setId(String id) {
this.id = id;
}
public Long getDownloads() {

2
dhp-workflows/dhp-aggregation/src/main/resources/eu/dnetlib/dhp/actionmanager/usagestats/oozie_app/workflow.xml
View File

@ -89,7 +89,7 @@
<arg>--hive_metastore_uris</arg><arg>${hiveMetastoreUris}</arg>
<arg>--outputPath</arg><arg>${outputPath}</arg>
<arg>--usagestatsdb</arg><arg>${usagestatsdb}</arg>
<arg>--workingPath</arg><arg>${workingDir}/usageDb</arg>
<arg>--workingPath</arg><arg>${workingDir}</arg>
</spark>
<ok to="End"/>
<error to="Kill"/>

235
dhp-workflows/dhp-aggregation/src/test/java/eu/dnetlib/dhp/actionmanager/usagestats/SparkAtomicActionCountJobTest.java
View File

@ -8,6 +8,7 @@ import java.nio.file.Files;
import java.nio.file.Path;
import java.util.stream.Collectors;
import eu.dnetlib.dhp.schema.oaf.OafEntity;
import org.apache.commons.io.FileUtils;
import org.apache.hadoop.io.Text;
import org.apache.spark.SparkConf;
@ -68,24 +69,26 @@ public class SparkAtomicActionCountJobTest {
@Test
void testMatch() {
String usageScoresPath = getClass()
.getResource("/eu/dnetlib/dhp/actionmanager/usagestats/usagestatsdb")
.getResource("/eu/dnetlib/dhp/actionmanager/usagestats")
.getPath();
SparkAtomicActionUsageJob.writeActionSet(spark, usageScoresPath, workingDir.toString() + "/actionSet");
final JavaSparkContext sc = new JavaSparkContext(spark.sparkContext());
JavaRDD<Result> tmp = sc
JavaRDD<AtomicAction> tmp = sc
.sequenceFile(workingDir.toString() + "/actionSet", Text.class, Text.class)
.map(usm -> OBJECT_MAPPER.readValue(usm._2.getBytes(), AtomicAction.class))
.map(aa -> (Result) aa.getPayload());
.map(usm -> OBJECT_MAPPER.readValue(usm._2.getBytes(), AtomicAction.class));
//.map(aa -> (Result) aa.getPayload());
Assertions.assertEquals(9, tmp.count());
Assertions.assertEquals(9,tmp.filter(aa -> ((OafEntity) aa.getPayload()).getId().startsWith("50|")).count());
Assertions.assertEquals(9,tmp.filter(aa -> ((OafEntity) aa.getPayload()).getId().startsWith("10|")).count());
Assertions.assertEquals(9,tmp.filter(aa -> ((OafEntity) aa.getPayload()).getId().startsWith("40|")).count());
tmp.foreach(r -> Assertions.assertEquals(2, r.getMeasures().size()));
tmp.foreach(r -> Assertions.assertEquals(2, ((OafEntity)r.getPayload()).getMeasures().size()));
tmp
.foreach(
r -> r
r -> ((OafEntity)r.getPayload())
.getMeasures()
.stream()
.forEach(
@ -95,14 +98,14 @@ public class SparkAtomicActionCountJobTest {
.forEach(u -> Assertions.assertFalse(u.getDataInfo().getDeletedbyinference()))));
tmp
.foreach(
r -> r
r -> ((OafEntity)r.getPayload())
.getMeasures()
.stream()
.forEach(
m -> m.getUnit().stream().forEach(u -> Assertions.assertTrue(u.getDataInfo().getInferred()))));
tmp
.foreach(
r -> r
r -> ((OafEntity)r.getPayload())
.getMeasures()
.stream()
.forEach(
@ -113,7 +116,7 @@ public class SparkAtomicActionCountJobTest {
tmp
.foreach(
r -> r
r -> ((OafEntity)r.getPayload())
.getMeasures()
.stream()
.forEach(
@ -127,7 +130,7 @@ public class SparkAtomicActionCountJobTest {
u.getDataInfo().getProvenanceaction().getClassid()))));
tmp
.foreach(
r -> r
r -> ((OafEntity)r.getPayload())
.getMeasures()
.stream()
.forEach(
@ -142,7 +145,7 @@ public class SparkAtomicActionCountJobTest {
tmp
.foreach(
r -> r
r -> ((OafEntity)r.getPayload())
.getMeasures()
.stream()
.forEach(
@ -157,12 +160,13 @@ public class SparkAtomicActionCountJobTest {
Assertions
.assertEquals(
1, tmp.filter(r -> r.getId().equals("50|dedup_wf_001::53575dc69e9ace947e02d47ecd54a7a6")).count());
1, tmp.filter(r -> ((OafEntity)r.getPayload()).getId().equals("50|dedup_wf_001::53575dc69e9ace947e02d47ecd54a7a6")).count());
Assertions
.assertEquals(
"0",
tmp
.map(r -> ((OafEntity)r.getPayload()))
.filter(r -> r.getId().equals("50|dedup_wf_001::53575dc69e9ace947e02d47ecd54a7a6"))
.collect()
.get(0)
@ -178,6 +182,7 @@ public class SparkAtomicActionCountJobTest {
.assertEquals(
"5",
tmp
.map(r -> ((OafEntity)r.getPayload()))
.filter(r -> r.getId().equals("50|dedup_wf_001::53575dc69e9ace947e02d47ecd54a7a6"))
.collect()
.get(0)
@ -194,6 +199,7 @@ public class SparkAtomicActionCountJobTest {
.assertEquals(
"0",
tmp
.map(r -> ((OafEntity)r.getPayload()))
.filter(r -> r.getId().equals("50|doi_________::17eda2ff77407538fbe5d3d719b9d1c0"))
.collect()
.get(0)
@ -209,6 +215,7 @@ public class SparkAtomicActionCountJobTest {
.assertEquals(
"1",
tmp
.map(r -> ((OafEntity)r.getPayload()))
.filter(r -> r.getId().equals("50|doi_________::17eda2ff77407538fbe5d3d719b9d1c0"))
.collect()
.get(0)
@ -225,6 +232,7 @@ public class SparkAtomicActionCountJobTest {
.assertEquals(
"2",
tmp
.map(r -> ((OafEntity)r.getPayload()))
.filter(r -> r.getId().equals("50|doi_________::3085e4c6e051378ca6157fe7f0430c1f"))
.collect()
.get(0)
@ -240,6 +248,7 @@ public class SparkAtomicActionCountJobTest {
.assertEquals(
"6",
tmp
.map(r -> ((OafEntity)r.getPayload()))
.filter(r -> r.getId().equals("50|doi_________::3085e4c6e051378ca6157fe7f0430c1f"))
.collect()
.get(0)
@ -251,6 +260,206 @@ public class SparkAtomicActionCountJobTest {
.getUnit()
.get(0)
.getValue());
Assertions
.assertEquals(
"0",
tmp
.map(r -> ((OafEntity)r.getPayload()))
.filter(r -> r.getId().equals("40|f1__________::53575dc69e9ace947e02d47ecd54a7a6"))
.collect()
.get(0)
.getMeasures()
.stream()
.filter(m -> m.getId().equals("downloads"))
.collect(Collectors.toList())
.get(0)
.getUnit()
.get(0)
.getValue());
Assertions
.assertEquals(
"5",
tmp
.map(r -> ((OafEntity)r.getPayload()))
.filter(r -> r.getId().equals("40|f1__________::53575dc69e9ace947e02d47ecd54a7a6"))
.collect()
.get(0)
.getMeasures()
.stream()
.filter(m -> m.getId().equals("views"))
.collect(Collectors.toList())
.get(0)
.getUnit()
.get(0)
.getValue());
Assertions
.assertEquals(
"0",
tmp
.map(r -> ((OafEntity)r.getPayload()))
.filter(r -> r.getId().equals("40|f11_________::17eda2ff77407538fbe5d3d719b9d1c0"))
.collect()
.get(0)
.getMeasures()
.stream()
.filter(m -> m.getId().equals("downloads"))
.collect(Collectors.toList())
.get(0)
.getUnit()
.get(0)
.getValue());
Assertions
.assertEquals(
"1",
tmp
.map(r -> ((OafEntity)r.getPayload()))
.filter(r -> r.getId().equals("40|f11_________::17eda2ff77407538fbe5d3d719b9d1c0"))
.collect()
.get(0)
.getMeasures()
.stream()
.filter(m -> m.getId().equals("views"))
.collect(Collectors.toList())
.get(0)
.getUnit()
.get(0)
.getValue());
Assertions
.assertEquals(
"2",
tmp
.map(r -> ((OafEntity)r.getPayload()))
.filter(r -> r.getId().equals("40|f12_________::3085e4c6e051378ca6157fe7f0430c1f"))
.collect()
.get(0)
.getMeasures()
.stream()
.filter(m -> m.getId().equals("downloads"))
.collect(Collectors.toList())
.get(0)
.getUnit()
.get(0)
.getValue());
Assertions
.assertEquals(
"6",
tmp
.map(r -> ((OafEntity)r.getPayload()))
.filter(r -> r.getId().equals("40|f12_________::3085e4c6e051378ca6157fe7f0430c1f"))
.collect()
.get(0)
.getMeasures()
.stream()
.filter(m -> m.getId().equals("views"))
.collect(Collectors.toList())
.get(0)
.getUnit()
.get(0)
.getValue());
Assertions
.assertEquals(
"0",
tmp
.map(r -> ((OafEntity)r.getPayload()))
.filter(r -> r.getId().equals("10|d1__________::53575dc69e9ace947e02d47ecd54a7a6"))
.collect()
.get(0)
.getMeasures()
.stream()
.filter(m -> m.getId().equals("downloads"))
.collect(Collectors.toList())
.get(0)
.getUnit()
.get(0)
.getValue());
Assertions
.assertEquals(
"5",
tmp
.map(r -> ((OafEntity)r.getPayload()))
.filter(r -> r.getId().equals("10|d1__________::53575dc69e9ace947e02d47ecd54a7a6"))
.collect()
.get(0)
.getMeasures()
.stream()
.filter(m -> m.getId().equals("views"))
.collect(Collectors.toList())
.get(0)
.getUnit()
.get(0)
.getValue());
Assertions
.assertEquals(
"0",
tmp
.map(r -> ((OafEntity)r.getPayload()))
.filter(r -> r.getId().equals("10|d11_________::17eda2ff77407538fbe5d3d719b9d1c0"))
.collect()
.get(0)
.getMeasures()
.stream()
.filter(m -> m.getId().equals("downloads"))
.collect(Collectors.toList())
.get(0)
.getUnit()
.get(0)
.getValue());
Assertions
.assertEquals(
"1",
tmp
.map(r -> ((OafEntity)r.getPayload()))
.filter(r -> r.getId().equals("10|d11_________::17eda2ff77407538fbe5d3d719b9d1c0"))
.collect()
.get(0)
.getMeasures()
.stream()
.filter(m -> m.getId().equals("views"))
.collect(Collectors.toList())
.get(0)
.getUnit()
.get(0)
.getValue());
Assertions
.assertEquals(
"2",
tmp
.map(r -> ((OafEntity)r.getPayload()))
.filter(r -> r.getId().equals("10|d12_________::3085e4c6e051378ca6157fe7f0430c1f"))
.collect()
.get(0)
.getMeasures()
.stream()
.filter(m -> m.getId().equals("downloads"))
.collect(Collectors.toList())
.get(0)
.getUnit()
.get(0)
.getValue());
Assertions
.assertEquals(
"6",
tmp
.map(r -> ((OafEntity)r.getPayload()))
.filter(r -> r.getId().equals("10|d12_________::3085e4c6e051378ca6157fe7f0430c1f"))
.collect()
.get(0)
.getMeasures()
.stream()
.filter(m -> m.getId().equals("views"))
.collect(Collectors.toList())
.get(0)
.getUnit()
.get(0)
.getValue());
}
}

9
dhp-workflows/dhp-aggregation/src/test/resources/eu/dnetlib/dhp/actionmanager/usagestats/datasourceDb Normal file
View File

@ -0,0 +1,9 @@
{"id":"d1__________::53575dc69e9ace947e02d47ecd54a7a6","downloads":0,"views":5}
{"id":"d11_________::17eda2ff77407538fbe5d3d719b9d1c0","downloads":0,"views":1}
{"id":"d11_________::1d4dc08605fd0a2be1105d30c63bfea1","downloads":1,"views":3}
{"id":"d11_________::2e3527822854ca9816f6dfea5bff61a8","downloads":1,"views":1}
{"id":"d12_________::3085e4c6e051378ca6157fe7f0430c1f","downloads":2,"views":6}
{"id":"d12_________::33f710e6dd30cc5e67e35b371ddc33cf","downloads":0,"views":1}
{"id":"d12_________::39738ebf10654732dd3a7af9f24655f8","downloads":1,"views":3}
{"id":"d13_________::3c3b65f07c1a06c7894397eda1d11bbf","downloads":1,"views":10}
{"id":"d13_________::4938a71a884dd481d329657aa543b850","downloads":0,"views":3}

12
dhp-workflows/dhp-aggregation/src/test/resources/eu/dnetlib/dhp/actionmanager/usagestats/datasourceDb_old Normal file
View File

@ -0,0 +1,12 @@
{"id":"d1__________::53575dc69e9ace947e02d47ecd54a7a6","downloads":0,"views":4}
{"id":"d1__________::53575dc69e9ace947e02d47ecd54a7a6","downloads":0,"views":1}
{"id":"d11_________::17eda2ff77407538fbe5d3d719b9d1c0","downloads":0,"views":1}
{"id":"d11_________::1d4dc08605fd0a2be1105d30c63bfea1","downloads":1,"views":3}
{"id":"d11_________::2e3527822854ca9816f6dfea5bff61a8","downloads":1,"views":1}
{"id":"d12_________::3085e4c6e051378ca6157fe7f0430c1f","downloads":2,"views":3}
{"id":"d12_________::3085e4c6e051378ca6157fe7f0430c1f","downloads":0,"views":3}
{"id":"d12_________::33f710e6dd30cc5e67e35b371ddc33cf","downloads":0,"views":1}
{"id":"d12_________::39738ebf10654732dd3a7af9f24655f8","downloads":1,"views":3}
{"id":"d13_________::3c3b65f07c1a06c7894397eda1d11bbf","downloads":1,"views":8}
{"id":"d13_________::3c3b65f07c1a06c7894397eda1d11bbf","downloads":0,"views":2}
{"id":"d13_________::4938a71a884dd481d329657aa543b850","downloads":0,"views":3}

9
dhp-workflows/dhp-aggregation/src/test/resources/eu/dnetlib/dhp/actionmanager/usagestats/projectDb Normal file
View File

@ -0,0 +1,9 @@
{"id":"f1__________::53575dc69e9ace947e02d47ecd54a7a6","downloads":0,"views":5}
{"id":"f11_________::17eda2ff77407538fbe5d3d719b9d1c0","downloads":0,"views":1}
{"id":"f11_________::1d4dc08605fd0a2be1105d30c63bfea1","downloads":1,"views":3}
{"id":"f11_________::2e3527822854ca9816f6dfea5bff61a8","downloads":1,"views":1}
{"id":"f12_________::3085e4c6e051378ca6157fe7f0430c1f","downloads":2,"views":6}
{"id":"f12_________::33f710e6dd30cc5e67e35b371ddc33cf","downloads":0,"views":1}
{"id":"f12_________::39738ebf10654732dd3a7af9f24655f8","downloads":1,"views":3}
{"id":"f13_________::3c3b65f07c1a06c7894397eda1d11bbf","downloads":1,"views":10}
{"id":"f13_________::4938a71a884dd481d329657aa543b850","downloads":0,"views":3}

12
dhp-workflows/dhp-aggregation/src/test/resources/eu/dnetlib/dhp/actionmanager/usagestats/projectDb_old Normal file
View File

@ -0,0 +1,12 @@
{"id":"f1__________::53575dc69e9ace947e02d47ecd54a7a6","downloads":0,"views":4}
{"id":"f1__________::53575dc69e9ace947e02d47ecd54a7a6","downloads":0,"views":1}
{"id":"f11_________::17eda2ff77407538fbe5d3d719b9d1c0","downloads":0,"views":1}
{"id":"f11_________::1d4dc08605fd0a2be1105d30c63bfea1","downloads":1,"views":3}
{"id":"f11_________::2e3527822854ca9816f6dfea5bff61a8","downloads":1,"views":1}
{"id":"f12_________::3085e4c6e051378ca6157fe7f0430c1f","downloads":2,"views":3}
{"id":"f12_________::3085e4c6e051378ca6157fe7f0430c1f","downloads":0,"views":3}
{"id":"f12_________::33f710e6dd30cc5e67e35b371ddc33cf","downloads":0,"views":1}
{"id":"f12_________::39738ebf10654732dd3a7af9f24655f8","downloads":1,"views":3}
{"id":"f13_________::3c3b65f07c1a06c7894397eda1d11bbf","downloads":1,"views":8}
{"id":"f13_________::3c3b65f07c1a06c7894397eda1d11bbf","downloads":0,"views":2}
{"id":"f13_________::4938a71a884dd481d329657aa543b850","downloads":0,"views":3}

9
dhp-workflows/dhp-aggregation/src/test/resources/eu/dnetlib/dhp/actionmanager/usagestats/usageDb Normal file
View File

@ -0,0 +1,9 @@
{"id":"dedup_wf_001::53575dc69e9ace947e02d47ecd54a7a6","downloads":0,"views":5}
{"id":"doi_________::17eda2ff77407538fbe5d3d719b9d1c0","downloads":0,"views":1}
{"id":"doi_________::1d4dc08605fd0a2be1105d30c63bfea1","downloads":1,"views":3}
{"id":"doi_________::2e3527822854ca9816f6dfea5bff61a8","downloads":1,"views":1}
{"id":"doi_________::3085e4c6e051378ca6157fe7f0430c1f","downloads":2,"views":6}
{"id":"doi_________::33f710e6dd30cc5e67e35b371ddc33cf","downloads":0,"views":1}
{"id":"doi_________::39738ebf10654732dd3a7af9f24655f8","downloads":1,"views":3}
{"id":"doi_________::3c3b65f07c1a06c7894397eda1d11bbf","downloads":1,"views":10}
{"id":"doi_________::4938a71a884dd481d329657aa543b850","downloads":0,"views":3}

12
dhp-workflows/dhp-aggregation/src/test/resources/eu/dnetlib/dhp/actionmanager/usagestats/usageDb_old Normal file
View File

@ -0,0 +1,12 @@
{"id":"dedup_wf_001::53575dc69e9ace947e02d47ecd54a7a6","downloads":0,"views":4}
{"id":"dedup_wf_001::53575dc69e9ace947e02d47ecd54a7a6","downloads":0,"views":1}
{"id":"doi_________::17eda2ff77407538fbe5d3d719b9d1c0","downloads":0,"views":1}
{"id":"doi_________::1d4dc08605fd0a2be1105d30c63bfea1","downloads":1,"views":3}
{"id":"doi_________::2e3527822854ca9816f6dfea5bff61a8","downloads":1,"views":1}
{"id":"doi_________::3085e4c6e051378ca6157fe7f0430c1f","downloads":2,"views":3}
{"id":"doi_________::3085e4c6e051378ca6157fe7f0430c1f","downloads":0,"views":3}
{"id":"doi_________::33f710e6dd30cc5e67e35b371ddc33cf","downloads":0,"views":1}
{"id":"doi_________::39738ebf10654732dd3a7af9f24655f8","downloads":1,"views":3}
{"id":"doi_________::3c3b65f07c1a06c7894397eda1d11bbf","downloads":1,"views":8}
{"id":"doi_________::3c3b65f07c1a06c7894397eda1d11bbf","downloads":0,"views":2}
{"id":"doi_________::4938a71a884dd481d329657aa543b850","downloads":0,"views":3}

12
dhp-workflows/dhp-aggregation/src/test/resources/eu/dnetlib/dhp/actionmanager/usagestats/usagestatsdb
View File

@ -1,12 +0,0 @@
{"result_id":"dedup_wf_001::53575dc69e9ace947e02d47ecd54a7a6","downloads":0,"views":4}
{"result_id":"dedup_wf_001::53575dc69e9ace947e02d47ecd54a7a6","downloads":0,"views":1}
{"result_id":"doi_________::17eda2ff77407538fbe5d3d719b9d1c0","downloads":0,"views":1}
{"result_id":"doi_________::1d4dc08605fd0a2be1105d30c63bfea1","downloads":1,"views":3}
{"result_id":"doi_________::2e3527822854ca9816f6dfea5bff61a8","downloads":1,"views":1}
{"result_id":"doi_________::3085e4c6e051378ca6157fe7f0430c1f","downloads":2,"views":3}
{"result_id":"doi_________::3085e4c6e051378ca6157fe7f0430c1f","downloads":0,"views":3}
{"result_id":"doi_________::33f710e6dd30cc5e67e35b371ddc33cf","downloads":0,"views":1}
{"result_id":"doi_________::39738ebf10654732dd3a7af9f24655f8","downloads":1,"views":3}
{"result_id":"doi_________::3c3b65f07c1a06c7894397eda1d11bbf","downloads":1,"views":8}
{"result_id":"doi_________::3c3b65f07c1a06c7894397eda1d11bbf","downloads":0,"views":2}
{"result_id":"doi_________::4938a71a884dd481d329657aa543b850","downloads":0,"views":3}

18
dhp-workflows/dhp-broker-events/src/main/java/eu/dnetlib/dhp/broker/oa/util/SubscriptionUtils.java
View File

@ -37,12 +37,24 @@ public class SubscriptionUtils {
}
public static boolean verifyDateRange(final long date, final String min, final String max) {
long from = 0;
long to = Long.MAX_VALUE;
try {
return date >= DateUtils.parseDate(min, "yyyy-MM-dd").getTime()
&& date < DateUtils.parseDate(max, "yyyy-MM-dd").getTime() + ONE_DAY;
from = min != null ? DateUtils.parseDate(min, "yyyy-MM-dd").getTime() : 0;
} catch (final ParseException e) {
return false;
from = 0;
}
try {
to = max != null ? DateUtils.parseDate(max, "yyyy-MM-dd").getTime() + ONE_DAY : Long.MAX_VALUE;
} catch (final ParseException e) {
to = Long.MAX_VALUE;
}
return date >= from && date < to;
}
public static boolean verifyExact(final String s1, final String s2) {

12
dhp-workflows/dhp-broker-events/src/test/java/eu/dnetlib/dhp/broker/oa/util/SubscriptionUtilsTest.java
View File

@ -41,6 +41,18 @@ public class SubscriptionUtilsTest {
assertTrue(SubscriptionUtils.verifyDateRange(date, "2010-01-01", "2011-01-01"));
assertFalse(SubscriptionUtils.verifyDateRange(date, "2020-01-01", "2021-01-01"));
assertTrue(SubscriptionUtils.verifyDateRange(date, "2010-01-01", "NULL"));
assertTrue(SubscriptionUtils.verifyDateRange(date, "2010-01-01", null));
assertTrue(SubscriptionUtils.verifyDateRange(date, "NULL", "2011-01-01"));
assertTrue(SubscriptionUtils.verifyDateRange(date, null, "2011-01-01"));
assertTrue(SubscriptionUtils.verifyDateRange(date, "NULL", "NULL"));
assertTrue(SubscriptionUtils.verifyDateRange(date, null, null));
assertFalse(SubscriptionUtils.verifyDateRange(date, "2020-01-01", null));
assertFalse(SubscriptionUtils.verifyDateRange(date, "2020-01-01", "NULL"));
assertFalse(SubscriptionUtils.verifyDateRange(date, null, "2005-01-01"));
assertFalse(SubscriptionUtils.verifyDateRange(date, "NULL", "2005-01-01"));
}
@Test

69
dhp-workflows/dhp-doiboost/src/main/scala/eu/dnetlib/doiboost/crossref/Crossref2Oaf.scala
View File

@ -370,10 +370,50 @@ case object Crossref2Oaf {
case dataset: Dataset => convertDataset(dataset)
}
val doisReference: List[String] = for {
JObject(reference_json) <- json \ "reference"
JField("DOI", JString(doi_json)) <- reference_json
} yield doi_json
if (doisReference != null && doisReference.nonEmpty) {
val citation_relations: List[Relation] = generateCitationRelations(doisReference, result)
resultList = resultList ::: citation_relations
}
resultList = resultList ::: List(result)
resultList
}
private def createCiteRelation(source: Result, targetPid: String, targetPidType: String): List[Relation] = {
val targetId = IdentifierFactory.idFromPid("50", targetPidType, targetPid, true)
val from = new Relation
from.setSource(source.getId)
from.setTarget(targetId)
from.setRelType(ModelConstants.RESULT_RESULT)
from.setRelClass(ModelConstants.CITES)
from.setSubRelType(ModelConstants.CITATION)
from.setCollectedfrom(source.getCollectedfrom)
from.setDataInfo(source.getDataInfo)
from.setLastupdatetimestamp(source.getLastupdatetimestamp)
val to = new Relation
to.setTarget(source.getId)
to.setSource(targetId)
to.setRelType(ModelConstants.RESULT_RESULT)
to.setRelClass(ModelConstants.IS_CITED_BY)
to.setSubRelType(ModelConstants.CITATION)
to.setCollectedfrom(source.getCollectedfrom)
to.setDataInfo(source.getDataInfo)
to.setLastupdatetimestamp(source.getLastupdatetimestamp)
List(from, to)
}
def generateCitationRelations(dois: List[String], result: Result): List[Relation] = {
dois.flatMap(d => createCiteRelation(result, d, "doi"))
}
def mappingFunderToRelations(
funders: List[mappingFunder],
sourceId: String,
@ -446,6 +486,7 @@ case object Crossref2Oaf {
case "10.13039/501100000781" =>
generateSimpleRelationFromAward(funder, "corda_______", extractECAward)
generateSimpleRelationFromAward(funder, "corda__h2020", extractECAward)
generateSimpleRelationFromAward(funder, "corda_____he", extractECAward)
case "10.13039/100000001" => generateSimpleRelationFromAward(funder, "nsf_________", a => a)
case "10.13039/501100001665" => generateSimpleRelationFromAward(funder, "anr_________", a => a)
case "10.13039/501100002341" => generateSimpleRelationFromAward(funder, "aka_________", a => a)
@ -487,6 +528,34 @@ case object Crossref2Oaf {
val targetId = getProjectId("wt__________", "1e5e62235d094afd01cd56e65112fc63")
queue += generateRelation(sourceId, targetId, ModelConstants.IS_PRODUCED_BY)
queue += generateRelation(targetId, sourceId, ModelConstants.PRODUCES)
//ASAP
case "10.13039/100018231" => generateSimpleRelationFromAward(funder, "asap________", a => a)
//CHIST-ERA
case "10.13039/501100001942" =>
val targetId = getProjectId("chistera____", "1e5e62235d094afd01cd56e65112fc63")
queue += generateRelation(sourceId, targetId, ModelConstants.IS_PRODUCED_BY)
queue += generateRelation(targetId, sourceId, ModelConstants.PRODUCES)
//HE
case "10.13039/100018693" | "10.13039/100018694" | "10.13039/100019188" | "10.13039/100019180" |
"10.13039/100018695" | "10.13039/100019185" | "10.13039/100019186" | "10.13039/100019187" =>
generateSimpleRelationFromAward(funder, "corda_____he", extractECAward)
//FCT
case "10.13039/501100001871" =>
generateSimpleRelationFromAward(funder, "fct_________", extractECAward)
//NHMRC
case "10.13039/501100000925" =>
generateSimpleRelationFromAward(funder, "mhmrc_______", extractECAward)
//NIH
case "10.13039/100000002" =>
generateSimpleRelationFromAward(funder, "nih_________", extractECAward)
//NWO
case "10.13039/501100003246" =>
generateSimpleRelationFromAward(funder, "nwo_________", extractECAward)
//UKRI
case "10.13039/100014013" | "10.13039/501100000267" | "10.13039/501100000268" | "10.13039/501100000269" |
"10.13039/501100000266" | "10.13039/501100006041" | "10.13039/501100000265" | "10.13039/501100000270" |
"10.13039/501100013589" | "10.13039/501100000271" =>
generateSimpleRelationFromAward(funder, "nwo_________", extractECAward)
case _ => logger.debug("no match for " + funder.DOI.get)

62
dhp-workflows/dhp-doiboost/src/test/resources/eu/dnetlib/doiboost/crossref/publication_license_embargo.json
View File

@ -94,7 +94,6 @@
"family": "Stein",
"sequence": "first",
"affiliation": [
]
},
{
@ -104,7 +103,6 @@
"family": "Velzen",
"sequence": "additional",
"affiliation": [
]
},
{
@ -114,7 +112,6 @@
"family": "Kowalski",
"sequence": "additional",
"affiliation": [
]
},
{
@ -122,7 +119,6 @@
"family": "Franckowiak",
"sequence": "additional",
"affiliation": [
]
},
{
@ -132,7 +128,6 @@
"family": "Gezari",
"sequence": "additional",
"affiliation": [
]
},
{
@ -142,7 +137,6 @@
"family": "Miller-Jones",
"sequence": "additional",
"affiliation": [
]
},
{
@ -150,7 +144,6 @@
"family": "Frederick",
"sequence": "additional",
"affiliation": [
]
},
{
@ -160,7 +153,6 @@
"family": "Sfaradi",
"sequence": "additional",
"affiliation": [
]
},
{
@ -168,7 +160,6 @@
"family": "Bietenholz",
"sequence": "additional",
"affiliation": [
]
},
{
@ -178,7 +169,6 @@
"family": "Horesh",
"sequence": "additional",
"affiliation": [
]
},
{
@ -186,7 +176,6 @@
"family": "Fender",
"sequence": "additional",
"affiliation": [
]
},
{
@ -196,7 +185,6 @@
"family": "Garrappa",
"sequence": "additional",
"affiliation": [
]
},
{
@ -206,7 +194,6 @@
"family": "Ahumada",
"sequence": "additional",
"affiliation": [
]
},
{
@ -214,7 +201,6 @@
"family": "Andreoni",
"sequence": "additional",
"affiliation": [
]
},
{
@ -222,7 +208,6 @@
"family": "Belicki",
"sequence": "additional",
"affiliation": [
]
},
{
@ -232,7 +217,6 @@
"family": "Bellm",
"sequence": "additional",
"affiliation": [
]
},
{
@ -240,7 +224,6 @@
"family": "Böttcher",
"sequence": "additional",
"affiliation": [
]
},
{
@ -248,7 +231,6 @@
"family": "Brinnel",
"sequence": "additional",
"affiliation": [
]
},
{
@ -256,7 +238,6 @@
"family": "Burruss",
"sequence": "additional",
"affiliation": [
]
},
{
@ -266,7 +247,6 @@
"family": "Cenko",
"sequence": "additional",
"affiliation": [
]
},
{
@ -276,7 +256,6 @@
"family": "Coughlin",
"sequence": "additional",
"affiliation": [
]
},
{
@ -286,7 +265,6 @@
"family": "Cunningham",
"sequence": "additional",
"affiliation": [
]
},
{
@ -294,7 +272,6 @@
"family": "Drake",
"sequence": "additional",
"affiliation": [
]
},
{
@ -302,7 +279,6 @@
"family": "Farrar",
"sequence": "additional",
"affiliation": [
]
},
{
@ -310,7 +286,6 @@
"family": "Feeney",
"sequence": "additional",
"affiliation": [
]
},
{
@ -318,7 +293,6 @@
"family": "Foley",
"sequence": "additional",
"affiliation": [
]
},
{
@ -328,7 +302,6 @@
"family": "Gal-Yam",
"sequence": "additional",
"affiliation": [
]
},
{
@ -336,7 +309,6 @@
"family": "Golkhou",
"sequence": "additional",
"affiliation": [
]
},
{
@ -346,7 +318,6 @@
"family": "Goobar",
"sequence": "additional",
"affiliation": [
]
},
{
@ -356,7 +327,6 @@
"family": "Graham",
"sequence": "additional",
"affiliation": [
]
},
{
@ -364,7 +334,6 @@
"family": "Hammerstein",
"sequence": "additional",
"affiliation": [
]
},
{
@ -374,7 +343,6 @@
"family": "Helou",
"sequence": "additional",
"affiliation": [
]
},
{
@ -384,7 +352,6 @@
"family": "Hung",
"sequence": "additional",
"affiliation": [
]
},
{
@ -392,7 +359,6 @@
"family": "Kasliwal",
"sequence": "additional",
"affiliation": [
]
},
{
@ -402,7 +368,6 @@
"family": "Kilpatrick",
"sequence": "additional",
"affiliation": [
]
},
{
@ -412,7 +377,6 @@
"family": "Kong",
"sequence": "additional",
"affiliation": [
]
},
{
@ -422,7 +386,6 @@
"family": "Kupfer",
"sequence": "additional",
"affiliation": [
]
},
{
@ -432,7 +395,6 @@
"family": "Laher",
"sequence": "additional",
"affiliation": [
]
},
{
@ -442,7 +404,6 @@
"family": "Mahabal",
"sequence": "additional",
"affiliation": [
]
},
{
@ -452,7 +413,6 @@
"family": "Masci",
"sequence": "additional",
"affiliation": [
]
},
{
@ -462,7 +422,6 @@
"family": "Necker",
"sequence": "additional",
"affiliation": [
]
},
{
@ -472,7 +431,6 @@
"family": "Nordin",
"sequence": "additional",
"affiliation": [
]
},
{
@ -480,7 +438,6 @@
"family": "Perley",
"sequence": "additional",
"affiliation": [
]
},
{
@ -490,7 +447,6 @@
"family": "Rigault",
"sequence": "additional",
"affiliation": [
]
},
{
@ -500,7 +456,6 @@
"family": "Reusch",
"sequence": "additional",
"affiliation": [
]
},
{
@ -508,7 +463,6 @@
"family": "Rodriguez",
"sequence": "additional",
"affiliation": [
]
},
{
@ -518,7 +472,6 @@
"family": "Rojas-Bravo",
"sequence": "additional",
"affiliation": [
]
},
{
@ -528,7 +481,6 @@
"family": "Rusholme",
"sequence": "additional",
"affiliation": [
]
},
{
@ -538,7 +490,6 @@
"family": "Shupe",
"sequence": "additional",
"affiliation": [
]
},
{
@ -548,7 +499,6 @@
"family": "Singer",
"sequence": "additional",
"affiliation": [
]
},
{
@ -558,7 +508,6 @@
"family": "Sollerman",
"sequence": "additional",
"affiliation": [
]
},
{
@ -566,7 +515,6 @@
"family": "Soumagnac",
"sequence": "additional",
"affiliation": [
]
},
{
@ -574,7 +522,6 @@
"family": "Stern",
"sequence": "additional",
"affiliation": [
]
},
{
@ -582,7 +529,6 @@
"family": "Taggart",
"sequence": "additional",
"affiliation": [
]
},
{
@ -590,7 +536,6 @@
"family": "van Santen",
"sequence": "additional",
"affiliation": [
]
},
{
@ -598,7 +543,6 @@
"family": "Ward",
"sequence": "additional",
"affiliation": [
]
},
{
@ -606,7 +550,6 @@
"family": "Woudt",
"sequence": "additional",
"affiliation": [
]
},
{
@ -616,7 +559,6 @@
"family": "Yao",
"sequence": "additional",
"affiliation": [
]
}
],
@ -1412,7 +1354,6 @@
"Nature Astronomy"
],
"original-title": [
],
"language": "en",
"link": [
@ -1448,10 +1389,8 @@
},
"score": 1.0,
"subtitle": [
],
"short-title": [
],
"issued": {
"date-parts": [
@ -1480,7 +1419,6 @@
"URL": "http://dx.doi.org/10.1038/s41550-020-01295-8",
"relation": {
"cites": [
]
},
"ISSN": [

46
dhp-workflows/dhp-doiboost/src/test/scala/eu/dnetlib/dhp/doiboost/crossref/CrossrefMappingTest.scala
View File

@ -1,9 +1,14 @@
package eu.dnetlib.dhp.doiboost.crossref
import eu.dnetlib.dhp.schema.common.ModelConstants
import eu.dnetlib.dhp.schema.oaf._
import eu.dnetlib.dhp.utils.DHPUtils
import eu.dnetlib.doiboost.crossref.Crossref2Oaf
import org.codehaus.jackson.map.{ObjectMapper, SerializationConfig}
import org.json4s
import org.json4s.JsonAST.{JField, JObject, JString}
import org.json4s.{DefaultFormats, JValue}
import org.json4s.jackson.JsonMethods
import org.junit.jupiter.api.Assertions._
import org.junit.jupiter.api.Test
import org.slf4j.{Logger, LoggerFactory}
@ -109,6 +114,47 @@ class CrossrefMappingTest {
}
private def parseJson(input: String): JValue = {
implicit lazy val formats: DefaultFormats.type = org.json4s.DefaultFormats
lazy val json: json4s.JValue = JsonMethods.parse(input)
json
}
@Test
def testCitationRelations(): Unit = {
val json = Source
.fromInputStream(getClass.getResourceAsStream("/eu/dnetlib/doiboost/crossref/publication_license_embargo.json"))
.mkString
assertNotNull(json)
assertFalse(json.isEmpty)
val result: List[Oaf] = Crossref2Oaf.convert(json)
assertTrue(result.nonEmpty)
val j = parseJson(json)
val doisReference: List[String] = for {
JObject(reference_json) <- j \ "reference"
JField("DOI", JString(doi_json)) <- reference_json
} yield doi_json
val relationList: List[Relation] = result
.filter(s => s.isInstanceOf[Relation])
.map(r => r.asInstanceOf[Relation])
.filter(r => r.getSubRelType.equalsIgnoreCase(ModelConstants.CITATION))
assertNotNull(relationList)
assertFalse(relationList.isEmpty)
assertEquals(doisReference.size * 2, relationList.size)
mapper.getSerializationConfig.enable(SerializationConfig.Feature.INDENT_OUTPUT)
relationList.foreach(p => println(mapper.writeValueAsString(p)))
}
@Test
def testEmptyTitle(): Unit = {
val json = Source

32
dhp-workflows/dhp-monitor-update/pom.xml Normal file
View File

@ -0,0 +1,32 @@
<?xml version="1.0" encoding="UTF-8"?>
<project xmlns="http://maven.apache.org/POM/4.0.0" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xsi:schemaLocation="http://maven.apache.org/POM/4.0.0 http://maven.apache.org/xsd/maven-4.0.0.xsd">
<parent>
<artifactId>dhp-workflows</artifactId>
<groupId>eu.dnetlib.dhp</groupId>
<version>1.2.4-SNAPSHOT</version>
</parent>
<modelVersion>4.0.0</modelVersion>
<artifactId>dhp-monitor-update</artifactId>
<dependencies>
<dependency>
<groupId>org.apache.spark</groupId>
<artifactId>spark-core_2.11</artifactId>
</dependency>
<dependency>
<groupId>org.apache.spark</groupId>
<artifactId>spark-sql_2.11</artifactId>
</dependency>
</dependencies>
<build>
<plugins>
<plugin>
<groupId>pl.project13.maven</groupId>
<artifactId>git-commit-id-plugin</artifactId>
<version>2.1.11</version>
<configuration>
<failOnNoGitDirectory>false</failOnNoGitDirectory>
</configuration>
</plugin>
</plugins>
</build>
</project>

34
dhp-workflows/dhp-monitor-update/src/main/resources/eu/dnetlib/dhp/oa/graph/monitor/oozie_app/config-default.xml Normal file
View File

@ -0,0 +1,34 @@
<configuration>
<property>
<name>jobTracker</name>
<value>${jobTracker}</value>
</property>
<property>
<name>nameNode</name>
<value>${nameNode}</value>
</property>
<property>
<name>oozie.use.system.libpath</name>
<value>true</value>
</property>
<property>
<name>oozie.action.sharelib.for.spark</name>
<value>spark2</value>
</property>
<property>
<name>hive_metastore_uris</name>
<value>thrift://iis-cdh5-test-m3.ocean.icm.edu.pl:9083</value>
</property>
<property>
<name>hive_jdbc_url</name>
<value>jdbc:hive2://iis-cdh5-test-m3.ocean.icm.edu.pl:10000/;UseNativeQuery=1;?spark.executor.memory=19166291558;spark.yarn.executor.memoryOverhead=3225;spark.driver.memory=11596411699;spark.yarn.driver.memoryOverhead=1228</value>
</property>
<property>
<name>oozie.wf.workflow.notification.url</name>
<value>{serviceUrl}/v1/oozieNotification/jobUpdate?jobId=$jobId%26status=$status</value>
</property>
<property>
<name>stats_tool_api_url</name>
<value>${stats_tool_api_url}</value>
</property>
</configuration>

18
dhp-workflows/dhp-monitor-update/src/main/resources/eu/dnetlib/dhp/oa/graph/monitor/oozie_app/monitor.sh Normal file
View File

@ -0,0 +1,18 @@
export PYTHON_EGG_CACHE=/home/$(whoami)/.python-eggs
export link_folder=/tmp/impala-shell-python-egg-cache-$(whoami)
if ! [ -L $link_folder ]
then
rm -Rf "$link_folder"
ln -sfn ${PYTHON_EGG_CACHE}${link_folder} ${link_folder}
fi
export SOURCE=$1
export TARGET=$2
export SCRIPT_PATH=$3
echo "Getting file from " $3
hdfs dfs -copyToLocal $3
echo "Updating monitor database"
cat createMonitorDB.sql | sed s/SOURCE/$1/g | sed s/TARGET/$2/g1 | impala-shell -f -
echo "Impala shell finished"

146
dhp-workflows/dhp-monitor-update/src/main/resources/eu/dnetlib/dhp/oa/graph/monitor/oozie_app/scripts/createMonitorDB.sql Normal file
View File

@ -0,0 +1,146 @@
DROP TABLE IF EXISTS TARGET.result_new;
create table TARGET.result_new as
select distinct * from (
select * from SOURCE.result r where exists (select 1 from SOURCE.result_organization ro where ro.id=r.id and ro.organization in (
-- 'openorgs____::b8b8ca674452579f3f593d9f5e557483', -- University College Cork
-- 'openorgs____::38d7097854736583dde879d12dacafca' -- Brown University
'openorgs____::57784c9e047e826fefdb1ef816120d92' --Arts et Métiers ParisTech
) )) foo;
COMPUTE STATS TARGET.result_new;
INSERT INTO TARGET.result select * from TARGET.result_new;
COMPUTE STATS TARGET.result;
INSERT INTO TARGET.result_citations select * from TARGET.result_citations orig where exists (select 1 from TARGET.result_new r where r.id=orig.id);
COMPUTE STATS TARGET.result_citations;
INSERT INTO TARGET.result_references_oc select * from TARGET.result_references_oc orig where exists (select 1 from TARGET.result_new r where r.id=orig.id);
COMPUTE STATS TARGET.result_references_oc;
INSERT INTO TARGET.result_citations_oc select * from TARGET.result_citations_oc orig where exists (select 1 from TARGET.result_new r where r.id=orig.id);
COMPUTE STATS TARGET.result_citations_oc;
INSERT INTO TARGET.result_classifications select * from TARGET.result_classifications orig where exists (select 1 from TARGET.result_new r where r.id=orig.id);
COMPUTE STATS TARGET.result_classifications;
INSERT INTO TARGET.result_apc select * from TARGET.result_apc orig where exists (select 1 from TARGET.result_new r where r.id=orig.id);
COMPUTE STATS TARGET.result_apc;
INSERT INTO TARGET.result_concepts select * from TARGET.result_concepts orig where exists (select 1 from TARGET.result_new r where r.id=orig.id);
COMPUTE STATS TARGET.result_concepts;
INSERT INTO TARGET.result_datasources select * from TARGET.result_datasources orig where exists (select 1 from TARGET.result_new r where r.id=orig.id);
COMPUTE STATS TARGET.result_datasources;
INSERT INTO TARGET.result_fundercount select * from TARGET.result_fundercount orig where exists (select 1 from TARGET.result_new r where r.id=orig.id);
COMPUTE STATS TARGET.result_fundercount;
INSERT INTO TARGET.result_gold select * from TARGET.result_gold orig where exists (select 1 from TARGET.result_new r where r.id=orig.id);
COMPUTE STATS TARGET.result_gold;
INSERT INTO TARGET.result_greenoa select * from TARGET.result_greenoa orig where exists (select 1 from TARGET.result_new r where r.id=orig.id);
COMPUTE STATS TARGET.result_greenoa;
INSERT INTO TARGET.result_languages select * from TARGET.result_languages orig where exists (select 1 from TARGET.result_new r where r.id=orig.id);
COMPUTE STATS TARGET.result_languages;
INSERT INTO TARGET.result_licenses select * from TARGET.result_licenses orig where exists (select 1 from TARGET.result_new r where r.id=orig.id);
COMPUTE STATS TARGET.result_licenses;
INSERT INTO TARGET.result_oids select * from TARGET.result_oids orig where exists (select 1 from TARGET.result_new r where r.id=orig.id);
COMPUTE STATS TARGET.result_oids;
INSERT INTO TARGET.result_organization select * from TARGET.result_organization orig where exists (select 1 from TARGET.result_new r where r.id=orig.id);
COMPUTE STATS TARGET.result_organization;
INSERT INTO TARGET.result_peerreviewed select * from TARGET.result_peerreviewed orig where exists (select 1 from TARGET.result_new r where r.id=orig.id);
COMPUTE STATS TARGET.result_peerreviewed;
INSERT INTO TARGET.result_pids select * from TARGET.result_pids orig where exists (select 1 from TARGET.result_new r where r.id=orig.id);
COMPUTE STATS TARGET.result_pids;
INSERT INTO TARGET.result_projectcount select * from TARGET.result_projectcount orig where exists (select 1 from TARGET.result_new r where r.id=orig.id);
COMPUTE STATS TARGET.result_projectcount;
INSERT INTO TARGET.result_projects select * from TARGET.result_projects orig where exists (select 1 from TARGET.result_new r where r.id=orig.id);
COMPUTE STATS TARGET.result_projects;
INSERT INTO TARGET.result_refereed select * from TARGET.result_refereed orig where exists (select 1 from TARGET.result_new r where r.id=orig.id);
COMPUTE STATS TARGET.result_refereed;
INSERT INTO TARGET.result_sources select * from TARGET.result_sources orig where exists (select 1 from TARGET.result_new r where r.id=orig.id);
COMPUTE STATS TARGET.result_sources;
INSERT INTO TARGET.result_topics select * from TARGET.result_topics orig where exists (select 1 from TARGET.result_new r where r.id=orig.id);
COMPUTE STATS TARGET.result_topics;
INSERT INTO TARGET.result_fos select * from TARGET.result_fos orig where exists (select 1 from TARGET.result_new r where r.id=orig.id);
COMPUTE STATS TARGET.result_fos;
create view TARGET.foo1 as select * from TARGET.result_result rr where rr.source in (select id from TARGET.result_new);
create view TARGET.foo2 as select * from TARGET.result_result rr where rr.target in (select id from TARGET.result_new);
INSERT INTO TARGET.result_result select distinct * from (select * from TARGET.foo1 union all select * from TARGET.foo2) foufou;
drop view TARGET.foo1;
drop view TARGET.foo2;
COMPUTE STATS TARGET.result_result;
-- indicators
-- Sprint 1 ----
INSERT INTO TARGET.indi_pub_green_oa select * from TARGET.indi_pub_green_oa orig where exists (select 1 from TARGET.result_new r where r.id=orig.id);
COMPUTE STATS TARGET.indi_pub_green_oa;
INSERT INTO TARGET.indi_pub_grey_lit select * from TARGET.indi_pub_grey_lit orig where exists (select 1 from TARGET.result_new r where r.id=orig.id);
COMPUTE STATS TARGET.indi_pub_grey_lit;
INSERT INTO TARGET.indi_pub_doi_from_crossref select * from TARGET.indi_pub_doi_from_crossref orig where exists (select 1 from TARGET.result_new r where r.id=orig.id);
COMPUTE STATS TARGET.indi_pub_doi_from_crossref;
-- Sprint 2 ----
INSERT INTO TARGET.indi_result_has_cc_licence select * from TARGET.indi_result_has_cc_licence orig where exists (select 1 from TARGET.result_new r where r.id=orig.id);
COMPUTE STATS TARGET.indi_result_has_cc_licence;
INSERT INTO TARGET.indi_result_has_cc_licence_url select * from TARGET.indi_result_has_cc_licence_url orig where exists (select 1 from TARGET.result_new r where r.id=orig.id);
COMPUTE STATS TARGET.indi_result_has_cc_licence_url;
INSERT INTO TARGET.indi_pub_has_abstract select * from TARGET.indi_pub_has_abstract orig where exists (select 1 from TARGET.result_new r where r.id=orig.id);
COMPUTE STATS TARGET.indi_pub_has_abstract;
INSERT INTO TARGET.indi_result_with_orcid select * from TARGET.indi_result_with_orcid orig where exists (select 1 from TARGET.result_new r where r.id=orig.id);
COMPUTE STATS TARGET.indi_result_with_orcid;
---- Sprint 3 ----
INSERT INTO TARGET.indi_funded_result_with_fundref select * from TARGET.indi_funded_result_with_fundref orig where exists (select 1 from TARGET.result_new r where r.id=orig.id);
COMPUTE STATS TARGET.indi_funded_result_with_fundref;
---- Sprint 4 ----
INSERT INTO TARGET.indi_pub_diamond select * from TARGET.indi_pub_diamond orig where exists (select 1 from TARGET.result_new r where r.id=orig.id);
COMPUTE STATS TARGET.indi_pub_diamond;
INSERT INTO TARGET.indi_pub_in_transformative select * from TARGET.indi_pub_in_transformative orig where exists (select 1 from TARGET.result_new r where r.id=orig.id);
COMPUTE STATS TARGET.indi_pub_in_transformative;
INSERT INTO TARGET.indi_pub_closed_other_open select * from TARGET.indi_pub_closed_other_open orig where exists (select 1 from TARGET.result_new r where r.id=orig.id);
COMPUTE STATS TARGET.indi_pub_closed_other_open;
---- Sprint 5 ----
INSERT INTO TARGET.indi_result_no_of_copies select * from TARGET.indi_result_no_of_copies orig where exists (select 1 from TARGET.result_new r where r.id=orig.id);
COMPUTE STATS TARGET.indi_result_no_of_copies;
---- Sprint 6 ----
INSERT INTO TARGET.indi_pub_hybrid_oa_with_cc select * from TARGET.indi_pub_hybrid_oa_with_cc orig where exists (select 1 from TARGET.result_new r where r.id=orig.id);
COMPUTE STATS TARGET.indi_pub_hybrid_oa_with_cc;
INSERT INTO TARGET.indi_pub_downloads select * from TARGET.indi_pub_downloads orig where exists (select 1 from TARGET.result_new r where r.id=orig.result_id);
COMPUTE STATS TARGET.indi_pub_downloads;
INSERT INTO TARGET.indi_pub_downloads_datasource select * from TARGET.indi_pub_downloads_datasource orig where exists (select 1 from TARGET.result_new r where r.id=orig.result_id);
COMPUTE STATS TARGET.indi_pub_downloads_datasource;
INSERT INTO TARGET.indi_pub_downloads_year select * from TARGET.indi_pub_downloads_year orig where exists (select 1 from TARGET.result_new r where r.id=orig.result_id);
COMPUTE STATS TARGET.indi_pub_downloads_year;
INSERT INTO TARGET.indi_pub_downloads_datasource_year select * from TARGET.indi_pub_downloads_datasource_year orig where exists (select 1 from TARGET.result_new r where r.id=orig.result_id);
COMPUTE STATS TARGET.indi_pub_downloads_datasource_year;
---- Sprint 7 ----
INSERT INTO TARGET.indi_pub_gold_oa select * from TARGET.indi_pub_gold_oa orig where exists (select 1 from TARGET.result_new r where r.id=orig.id);
COMPUTE STATS TARGET.indi_pub_gold_oa;
INSERT INTO TARGET.indi_pub_hybrid select * from TARGET.indi_pub_hybrid orig where exists (select 1 from TARGET.result_new r where r.id=orig.id);
COMPUTE STATS TARGET.indi_pub_hybrid;
INSERT INTO TARGET.indi_pub_has_preprint select * from TARGET.indi_pub_has_preprint orig where exists (select 1 from TARGET.result_new r where r.id=orig.id);
COMPUTE STATS TARGET.indi_pub_has_preprint;
INSERT INTO TARGET.indi_pub_in_subscribed select * from TARGET.indi_pub_in_subscribed orig where exists (select 1 from TARGET.result_new r where r.id=orig.id);
COMPUTE STATS TARGET.indi_pub_in_subscribed;
INSERT INTO TARGET.indi_result_with_pid select * from TARGET.indi_result_with_pid orig where exists (select 1 from TARGET.result_new r where r.id=orig.id);
COMPUTE STATS TARGET.indi_result_with_pid;
--create table TARGET.indi_datasets_gold_oa stored as parquet as select * from SOURCE.indi_datasets_gold_oa orig where exists (select 1 from TARGET.result r where r.id=orig.id);
--compute stats TARGET.indi_datasets_gold_oa;
--create table TARGET.indi_software_gold_oa stored as parquet as select * from SOURCE.indi_software_gold_oa orig where exists (select 1 from TARGET.result r where r.id=orig.id);
--compute stats TARGET.indi_software_gold_oa;
DROP TABLE TARGET.result_new;

77
dhp-workflows/dhp-monitor-update/src/main/resources/eu/dnetlib/dhp/oa/graph/monitor/oozie_app/workflow.xml Normal file
View File

@ -0,0 +1,77 @@
<workflow-app name="Monitor DB" xmlns="uri:oozie:workflow:0.5">
<parameters>
<property>
<name>stats_db_name</name>
<description>the target stats database name</description>
</property>
<property>
<name>stats_db_shadow_name</name>
<description>the name of the shadow schema</description>
</property>
<property>
<name>monitor_db_name</name>
<description>the target monitor db name</description>
</property>
<property>
<name>monitor_db_shadow_name</name>
<description>the name of the shadow monitor db</description>
</property>
<property>
<name>stats_tool_api_url</name>
<description>The url of the API of the stats tool. Is used to trigger the cache update.</description>
</property>
<property>
<name>hive_metastore_uris</name>
<description>hive server metastore URIs</description>
</property>
<property>
<name>hive_jdbc_url</name>
<description>hive server jdbc url</description>
</property>
<property>
<name>hive_timeout</name>
<description>the time period, in seconds, after which Hive fails a transaction if a Hive client has not sent a hearbeat. The default value is 300 seconds.</description>
</property>
<property>
<name>context_api_url</name>
<description>the base url of the context api (https://services.openaire.eu/openaire)</description>
</property>
</parameters>
<global>
<job-tracker>${jobTracker}</job-tracker>
<name-node>${nameNode}</name-node>
<configuration>
<property>
<name>hive.metastore.uris</name>
<value>${hive_metastore_uris}</value>
</property>
<property>
<name>hive.txn.timeout</name>
<value>${hive_timeout}</value>
</property>
</configuration>
</global>
<start to="Step1-createMonitorDB"/>
<kill name="Kill">
<message>Action failed, error message[${wf:errorMessage(wf:lastErrorNode())}]</message>
</kill>
<action name="Step1-createMonitorDB">
<shell xmlns="uri:oozie:shell-action:0.1">
<job-tracker>${jobTracker}</job-tracker>
<name-node>${nameNode}</name-node>
<exec>monitor.sh</exec>
<argument>${stats_db_name}</argument>
<argument>${monitor_db_name}</argument>
<argument>${wf:appPath()}/scripts/createMonitorDB.sql</argument>
<file>monitor.sh</file>
</shell>
<ok to="End"/>
<error to="Kill"/>
</action>
<end name="End"/>
</workflow-app>

7
dhp-workflows/dhp-stats-update/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step15_5.sql
View File

@ -30,11 +30,16 @@ from rcount
group by rcount.pid;
create view ${stats_db_name}.rndexpenditure as select * from stats_ext.rndexpediture;
create view ${stats_db_name}.rndgdpexpenditure as select * from stats_ext.rndgdpexpenditure;
create view ${stats_db_name}.doctoratestudents as select * from stats_ext.doctoratestudents;
create view ${stats_db_name}.totalresearchers as select * from stats_ext.totalresearchers;
create view ${stats_db_name}.totalresearchersft as select * from stats_ext.totalresearchersft;
create view ${stats_db_name}.hrrst as select * from stats_ext.hrrst;
create table ${stats_db_name}.result_instance stored as parquet as
select distinct r.*
from (
select substr(r.id, 4) as id, inst.accessright.classname as accessright, substr(inst.collectedfrom.key, 4) as collectedfrom,
select substr(r.id, 4) as id, inst.accessright.classname as accessright, inst.accessright.openaccessroute as accessright_uw, substr(inst.collectedfrom.key, 4) as collectedfrom,
substr(inst.hostedby.key, 4) as hostedby, inst.dateofacceptance.value as dateofacceptance, inst.license.value as license, p.qualifier.classname as pidtype, p.value as pid
from ${openaire_db_name}.result r lateral view explode(r.instance) instances as inst lateral view explode(inst.pid) pids as p) r
join ${stats_db_name}.result res on res.id=r.id;

15
dhp-workflows/dhp-stats-update/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step20-createMonitorDB.sql
View File

@ -10,6 +10,11 @@ create view if not exists TARGET.creation_date as select * from SOURCE.creation_
create view if not exists TARGET.funder as select * from SOURCE.funder;
create view if not exists TARGET.fundref as select * from SOURCE.fundref;
create view if not exists TARGET.rndexpenditure as select * from SOURCE.rndexpediture;
create view if not exists TARGET.rndgdpexpenditure as select * from SOURCE.rndgdpexpenditure;
create view if not exists TARGET.doctoratestudents as select * from SOURCE.doctoratestudents;
create view if not exists TARGET.totalresearchers as select * from SOURCE.totalresearchers;
create view if not exists TARGET.totalresearchersft as select * from SOURCE.totalresearchersft;
create view if not exists TARGET.hrrst as select * from SOURCE.hrrst;
create table TARGET.result stored as parquet as
select distinct * from (
@ -50,7 +55,15 @@ create table TARGET.result stored as parquet as
'openorgs____::1698a2eb1885ef8adb5a4a969e745ad3', -- École des Ponts ParisTech
'openorgs____::e15adb13c4dadd49de4d35c39b5da93a', -- Nanyang Technological University
'openorgs____::4b34103bde246228fcd837f5f1bf4212', -- Autonomous University of Barcelona
'openorgs____::72ec75fcfc4e0df1a76dc4c49007fceb' -- McMaster University
'openorgs____::72ec75fcfc4e0df1a76dc4c49007fceb', -- McMaster University
'openorgs____::51c7fc556e46381734a25a6fbc3fd398', -- University of Modena and Reggio Emilia
'openorgs____::235d7f9ad18ecd7e6dc62ea4990cb9db', -- Bilkent University
'openorgs____::31f2fa9e05b49d4cf40a19c3fed8eb06', -- Saints Cyril and Methodius University of Skopje
'openorgs____::db7686f30f22cbe73a4fde872ce812a6', -- University of Milan
'openorgs____::b8b8ca674452579f3f593d9f5e557483', -- University College Cork
'openorgs____::38d7097854736583dde879d12dacafca', -- Brown University
'openorgs____::57784c9e047e826fefdb1ef816120d92', --Arts et Métiers ParisTech
'openorgs____::2530baca8a15936ba2e3297f2bce2e7e' -- University of Cape Town
))) foo;
compute stats TARGET.result;

35
dhp-workflows/dhp-stats-update/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/workflow.xml
View File

@ -70,7 +70,40 @@
</configuration>
</global>
<start to="Step1"/>
<start to="resume_from"/>
<decision name="resume_from">
<switch>
<case to="Step1">${wf:conf('resumeFrom') eq 'Step1'}</case>
<case to="Step2">${wf:conf('resumeFrom') eq 'Step2'}</case>
<case to="Step3">${wf:conf('resumeFrom') eq 'Step3'}</case>
<case to="Step4">${wf:conf('resumeFrom') eq 'Step4'}</case>
<case to="Step5">${wf:conf('resumeFrom') eq 'Step5'}</case>
<case to="Step6">${wf:conf('resumeFrom') eq 'Step6'}</case>
<case to="Step7">${wf:conf('resumeFrom') eq 'Step7'}</case>
<case to="Step8">${wf:conf('resumeFrom') eq 'Step8'}</case>
<case to="Step9">${wf:conf('resumeFrom') eq 'Step9'}</case>
<case to="Step10">${wf:conf('resumeFrom') eq 'Step10'}</case>
<case to="Step11">${wf:conf('resumeFrom') eq 'Step11'}</case>
<case to="Step12">${wf:conf('resumeFrom') eq 'Step12'}</case>
<case to="Step13">${wf:conf('resumeFrom') eq 'Step13'}</case>
<case to="Step14">${wf:conf('resumeFrom') eq 'Step14'}</case>
<case to="Step15">${wf:conf('resumeFrom') eq 'Step15'}</case>
<case to="Step15_5">${wf:conf('resumeFrom') eq 'Step15_5'}</case>
<case to="Contexts">${wf:conf('resumeFrom') eq 'Contexts'}</case>
<case to="Step16-createIndicatorsTables">${wf:conf('resumeFrom') eq 'Step16-createIndicatorsTables'}</case>
<case to="Step16_1-definitions">${wf:conf('resumeFrom') eq 'Step16_1-definitions'}</case>
<case to="Step16_5">${wf:conf('resumeFrom') eq 'Step16_5'}</case>
<case to="Step19-finalize">${wf:conf('resumeFrom') eq 'Step19-finalize'}</case>
<case to="step20-createMonitorDB">${wf:conf('resumeFrom') eq 'step20-createMonitorDB'}</case>
<case to="step21-createObservatoryDB-pre">${wf:conf('resumeFrom') eq 'step21-createObservatoryDB-pre'}</case>
<case to="step21-createObservatoryDB">${wf:conf('resumeFrom') eq 'step21-createObservatoryDB'}</case>
<case to="step21-createObservatoryDB-post">${wf:conf('resumeFrom') eq 'step21-createObservatoryDB-post'}</case>
<case to="Step22">${wf:conf('resumeFrom') eq 'Step22'}</case>
<default to="Step1"/>
</switch>
</decision>
<kill name="Kill">
<message>Action failed, error message[${wf:errorMessage(wf:lastErrorNode())}]</message>

2
pom.xml
View File

@ -807,7 +807,7 @@
<mockito-core.version>3.3.3</mockito-core.version>
<mongodb.driver.version>3.4.2</mongodb.driver.version>
<vtd.version>[2.12,3.0)</vtd.version>
<dhp-schemas.version>[3.15.0]</dhp-schemas.version>
<dhp-schemas.version>[3.15.1-SNAPSHOT]</dhp-schemas.version>
<dnet-actionmanager-api.version>[4.0.3]</dnet-actionmanager-api.version>
<dnet-actionmanager-common.version>[6.0.5]</dnet-actionmanager-common.version>
<dnet-openaire-broker-common.version>[3.1.6]</dnet-openaire-broker-common.version>