D-Net
/
dnet-hadoop
14
5
Fork 11
Code Issues 14 Pull Requests 13 Projects 2 Releases 2 Wiki Activity

cleaning workflow to include cleaning of default values

This commit is contained in:
Claudio Atzori 2020-06-18 19:37:25 +02:00
parent 52f62d5d8c
commit d0ac7514b2
5 changed files with 114 additions and 16 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

77
dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/dhp/oa/graph/clean/CleanGraphSparkJob.java
View File

@ -3,9 +3,13 @@ package eu.dnetlib.dhp.oa.graph.clean;
import static eu.dnetlib.dhp.common.SparkSessionSupport.runWithSparkSession;
import java.io.BufferedInputStream;
import java.util.Objects;
import java.util.Optional;
import java.util.stream.Collectors;
import org.apache.commons.io.IOUtils;
import org.apache.commons.lang3.StringUtils;
import org.apache.spark.SparkConf;
import org.apache.spark.api.java.function.MapFunction;
import org.apache.spark.sql.Dataset;
@ -19,7 +23,9 @@ import com.fasterxml.jackson.databind.ObjectMapper;
import eu.dnetlib.dhp.application.ArgumentApplicationParser;
import eu.dnetlib.dhp.common.HdfsSupport;
import eu.dnetlib.dhp.oa.graph.raw.common.OafMapperUtils;
import eu.dnetlib.dhp.oa.graph.raw.common.VocabularyGroup;
import eu.dnetlib.dhp.schema.common.ModelConstants;
import eu.dnetlib.dhp.schema.oaf.*;
import eu.dnetlib.dhp.utils.ISLookupClientFactory;
import eu.dnetlib.enabling.is.lookup.rmi.ISLookUpService;
@ -84,12 +90,83 @@ public class CleanGraphSparkJob {
readTableFromPath(spark, inputPath, clazz)
.map((MapFunction<T, T>) value -> OafCleaner.apply(value, mapping), Encoders.bean(clazz))
.map((MapFunction<T, T>) value -> fixDefaults(value), Encoders.bean(clazz))
.write()
.mode(SaveMode.Overwrite)
.option("compression", "gzip")
.json(outputPath);
}
private static <T extends Oaf> T fixDefaults(T value) {
if (value instanceof Datasource) {
// nothing to clean here
} else if (value instanceof Project) {
// nothing to clean here
} else if (value instanceof Organization) {
Organization o = (Organization) value;
if (Objects.isNull(o.getCountry()) || StringUtils.isBlank(o.getCountry().getClassid())) {
o.setCountry(qualifier("UNKNOWN", "Unknown", ModelConstants.DNET_COUNTRY_TYPE));
}
} else if (value instanceof Relation) {
// nothing to clean here
} else if (value instanceof Result) {
Result r = (Result) value;
if (Objects.isNull(r.getLanguage()) || StringUtils.isBlank(r.getLanguage().getClassid())) {
r
.setLanguage(
qualifier("und", "Undetermined", ModelConstants.DNET_LANGUAGES));
}
if (Objects.nonNull(r.getSubject())) {
r
.setSubject(
r
.getSubject()
.stream()
.filter(Objects::nonNull)
.filter(sp -> StringUtils.isNotBlank(sp.getValue()))
.filter(sp -> Objects.nonNull(sp.getQualifier()))
.filter(sp -> StringUtils.isNotBlank(sp.getQualifier().getClassid()))
.collect(Collectors.toList()));
}
if (Objects.isNull(r.getResourcetype()) || StringUtils.isBlank(r.getResourcetype().getClassid())) {
r
.setResourcetype(
qualifier("UNKNOWN", "Unknown", ModelConstants.DNET_DATA_CITE_RESOURCE));
}
if (Objects.isNull(r.getBestaccessright()) || StringUtils.isBlank(r.getBestaccessright().getClassid())) {
r
.setBestaccessright(
qualifier("UNKNOWN", "not available", ModelConstants.DNET_ACCESS_MODES));
}
if (Objects.nonNull(r.getInstance())) {
for (Instance i : r.getInstance()) {
if (Objects.isNull(i.getAccessright()) || StringUtils.isBlank(i.getAccessright().getClassid())) {
i.setAccessright(qualifier("UNKNOWN", "not available", ModelConstants.DNET_ACCESS_MODES));
}
}
}
if (value instanceof Publication) {
} else if (value instanceof eu.dnetlib.dhp.schema.oaf.Dataset) {
} else if (value instanceof OtherResearchProduct) {
} else if (value instanceof Software) {
}
}
return value;
}
private static Qualifier qualifier(String classid, String classname, String scheme) {
return OafMapperUtils
.qualifier(
classid, classname, scheme, scheme);
}
private static <T extends Oaf> Dataset<T> readTableFromPath(
SparkSession spark, String inputEntityPath, Class<T> clazz) {

34
dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/dhp/oa/graph/clean/CleaningRuleMap.java
View File

@ -4,10 +4,13 @@ package eu.dnetlib.dhp.oa.graph.clean;
import java.io.Serializable;
import java.util.HashMap;
import org.apache.commons.lang3.StringUtils;
import eu.dnetlib.dhp.common.FunctionalInterfaceSupport.SerializableConsumer;
import eu.dnetlib.dhp.oa.graph.raw.common.VocabularyGroup;
import eu.dnetlib.dhp.schema.common.ModelConstants;
import eu.dnetlib.dhp.schema.oaf.Country;
import eu.dnetlib.dhp.schema.oaf.Qualifier;
import eu.dnetlib.dhp.schema.oaf.StructuredProperty;
public class CleaningRuleMap extends HashMap<Class, SerializableConsumer<Object>> implements Serializable {
@ -18,23 +21,24 @@ public class CleaningRuleMap extends HashMap<Class, SerializableConsumer<Object>
*/
public static CleaningRuleMap create(VocabularyGroup vocabularies) {
CleaningRuleMap mapping = new CleaningRuleMap();
mapping.put(Qualifier.class, o -> {
Qualifier q = (Qualifier) o;
if (vocabularies.vocabularyExists(q.getSchemeid())) {
Qualifier newValue = vocabularies.lookup(q.getSchemeid(), q.getClassid());
q.setClassid(newValue.getClassid());
q.setClassname(newValue.getClassname());
mapping.put(Qualifier.class, o -> cleanQualifier(vocabularies, (Qualifier) o));
mapping.put(Country.class, o -> {
final Country c = (Country) o;
if (StringUtils.isBlank(c.getSchemeid())) {
c.setSchemeid(ModelConstants.DNET_COUNTRY_TYPE);
c.setSchemename(ModelConstants.DNET_COUNTRY_TYPE);
}
});
mapping.put(StructuredProperty.class, o -> {
StructuredProperty sp = (StructuredProperty) o;
// TODO implement a policy
/*
* if (StringUtils.isBlank(sp.getValue())) { sp.setValue(null); sp.setQualifier(null); sp.setDataInfo(null);
* }
*/
cleanQualifier(vocabularies, c);
});
return mapping;
}
private static <Q extends Qualifier> void cleanQualifier(VocabularyGroup vocabularies, Q q) {
if (vocabularies.vocabularyExists(q.getSchemeid())) {
Qualifier newValue = vocabularies.lookup(q.getSchemeid(), q.getClassid());
q.setClassid(newValue.getClassid());
q.setClassname(newValue.getClassname());
}
}
}

6
dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/dhp/oa/graph/raw/common/VocabularyGroup.java
View File

@ -122,7 +122,11 @@ public class VocabularyGroup implements Serializable {
}
public boolean vocabularyExists(final String vocId) {
return vocs.containsKey(vocId.toLowerCase());
return Optional
.ofNullable(vocId)
.map(String::toLowerCase)
.map(id -> vocs.containsKey(id))
.orElse(false);
}
private void addSynonyms(final String vocId, final String termId, final String syn) {

7
dhp-workflows/dhp-graph-mapper/src/test/java/eu/dnetlib/dhp/oa/graph/clean/CleaningFunctionTest.java
View File

@ -21,6 +21,7 @@ import com.fasterxml.jackson.databind.ObjectMapper;
import eu.dnetlib.dhp.oa.graph.raw.common.VocabularyGroup;
import eu.dnetlib.dhp.schema.oaf.Publication;
import eu.dnetlib.dhp.schema.oaf.Qualifier;
import eu.dnetlib.dhp.schema.oaf.Result;
import eu.dnetlib.enabling.is.lookup.rmi.ISLookUpException;
import eu.dnetlib.enabling.is.lookup.rmi.ISLookUpService;
@ -56,6 +57,9 @@ public class CleaningFunctionTest {
String json = IOUtils.toString(getClass().getResourceAsStream("/eu/dnetlib/dhp/oa/graph/clean/result.json"));
Publication p_in = MAPPER.readValue(json, Publication.class);
assertTrue(p_in instanceof Result);
assertTrue(p_in instanceof Publication);
Publication p_out = OafCleaner.apply(p_in, mapping);
assertNotNull(p_out);
@ -63,6 +67,9 @@ public class CleaningFunctionTest {
assertEquals("und", p_out.getLanguage().getClassid());
assertEquals("Undetermined", p_out.getLanguage().getClassname());
assertEquals("DE", p_out.getCountry().get(0).getClassid());
assertEquals("Germany", p_out.getCountry().get(0).getClassname());
assertEquals("0018", p_out.getInstance().get(0).getInstancetype().getClassid());
assertEquals("Annotation", p_out.getInstance().get(0).getInstancetype().getClassname());

6
dhp-workflows/dhp-graph-mapper/src/test/resources/eu/dnetlib/dhp/oa/graph/clean/result.json
View File

@ -202,6 +202,12 @@
"contributor": [
],
"country": [
{
"classid": "DE",
"classname": "DE",
"schemeid": "dnet:countries",
"schemename": "dnet:countries"
}
],
"coverage": [
],