2023-07-06 10:28:53 +02:00
|
|
|
|
2018-10-02 10:37:54 +02:00
|
|
|
package eu.dnetlib.pace.clustering;
|
|
|
|
|
|
|
|
import java.util.Collection;
|
|
|
|
import java.util.List;
|
|
|
|
import java.util.Map;
|
|
|
|
|
2023-07-06 10:28:53 +02:00
|
|
|
import org.apache.commons.lang3.StringUtils;
|
|
|
|
|
2018-10-02 10:37:54 +02:00
|
|
|
import com.google.common.collect.Lists;
|
|
|
|
import com.google.common.collect.Sets;
|
2023-07-06 10:28:53 +02:00
|
|
|
|
2019-10-08 14:53:52 +02:00
|
|
|
import eu.dnetlib.pace.config.Config;
|
2018-10-02 10:37:54 +02:00
|
|
|
import eu.dnetlib.pace.model.Field;
|
|
|
|
|
2018-10-24 12:09:41 +02:00
|
|
|
@ClusteringClass("lowercase")
|
2018-10-02 10:37:54 +02:00
|
|
|
public class LowercaseClustering extends AbstractClusteringFunction {
|
|
|
|
|
|
|
|
public LowercaseClustering(final Map<String, Integer> params) {
|
|
|
|
super(params);
|
|
|
|
}
|
|
|
|
|
|
|
|
@Override
|
2019-10-08 14:53:52 +02:00
|
|
|
public Collection<String> apply(Config conf, List<Field> fields) {
|
2018-10-02 10:37:54 +02:00
|
|
|
Collection<String> c = Sets.newLinkedHashSet();
|
2023-07-06 10:28:53 +02:00
|
|
|
for (Field f : fields) {
|
2019-10-08 14:53:52 +02:00
|
|
|
c.addAll(doApply(conf, f.stringValue()));
|
2018-10-02 10:37:54 +02:00
|
|
|
}
|
|
|
|
return c;
|
|
|
|
}
|
|
|
|
|
|
|
|
@Override
|
2019-10-08 14:53:52 +02:00
|
|
|
protected Collection<String> doApply(final Config conf, final String s) {
|
2023-07-06 10:28:53 +02:00
|
|
|
if (StringUtils.isBlank(s)) {
|
2018-10-02 10:37:54 +02:00
|
|
|
return Lists.newArrayList();
|
|
|
|
}
|
|
|
|
return Lists.newArrayList(s.toLowerCase().trim());
|
|
|
|
}
|
|
|
|
}
|