114 lines
3.5 KiB
Java
114 lines
3.5 KiB
Java
package eu.dnetlib.data.utility.cleaner;
|
|
|
|
import java.util.HashMap;
|
|
import java.util.Map;
|
|
import java.util.Set;
|
|
|
|
import org.apache.commons.logging.Log;
|
|
import org.apache.commons.logging.LogFactory;
|
|
|
|
import com.google.common.base.Joiner;
|
|
import com.google.common.collect.Maps;
|
|
import com.google.common.collect.Sets;
|
|
|
|
import eu.dnetlib.data.utility.cleaner.rmi.CleanerException;
|
|
import eu.dnetlib.enabling.is.lookup.rmi.ISLookUpService;
|
|
|
|
/**
|
|
* @author michele
|
|
*
|
|
* Vocabulary rules must be declared in a CleanerDS profile, for each vocabulary must be present the relative VocabularyDS profile:
|
|
*
|
|
* <RULE xpath="..." vocabularies="VOC1" /> <RULE xpath="..." vocabularies="VOC1, VOC2, VOC3" />
|
|
*/
|
|
|
|
public class VocabularyRule extends XPATHCleaningRule {
|
|
|
|
private Set<String> vocabularies;
|
|
|
|
private static final Log log = LogFactory.getLog(VocabularyRule.class); // NOPMD by marko on 11/24/08 5:02 PM
|
|
|
|
private Map<String, String> synonyms = Maps.newHashMap();
|
|
private Set<String> validTerms = Sets.newHashSet();
|
|
|
|
public VocabularyRule(final Set<String> vocabularies, final ISLookUpService lookup) throws CleanerException {
|
|
this.vocabularies = vocabularies;
|
|
|
|
loadSynonymsAndTerms(lookup);
|
|
}
|
|
|
|
@Override
|
|
protected String calculateNewValue(final String oldValue) throws CleanerException {
|
|
log.debug("calculating new value for: " + oldValue);
|
|
|
|
if (synonyms.isEmpty()) {
|
|
log.warn("Vocabulary terms is void, vocabularies: " + this.vocabularies);
|
|
}
|
|
|
|
String newValue = null;
|
|
|
|
if (synonyms.containsKey(oldValue.toLowerCase())) {
|
|
newValue = synonyms.get(oldValue.toLowerCase());
|
|
}
|
|
|
|
if (newValue == null) {
|
|
log.debug("Synonym " + oldValue + " not found in vocabulary");
|
|
return oldValue;
|
|
}
|
|
|
|
return newValue;
|
|
}
|
|
|
|
private void loadSynonymsAndTerms(final ISLookUpService lookup) throws CleanerException {
|
|
|
|
for (final String vocabulary : vocabularies) {
|
|
try {
|
|
final String query = "for $x in collection('/db/DRIVER/VocabularyDSResources/VocabularyDSResourceType')"
|
|
+ "//RESOURCE_PROFILE[.//VOCABULARY_NAME/@code='" + vocabulary + "']//TERM return "
|
|
+ "( concat($x/@code,'|-:-|', $x/@code), concat($x/@english_name,'|-:-|', $x/@code), concat($x/@native_name,'|-:-|', $x/@code), "
|
|
+ "for $y in $x//SYNONYM return concat($y/@term,'|-:-|', $x/@code) )";
|
|
|
|
for (final String s : lookup.quickSearchProfile(query)) {
|
|
log.debug("SYNONYM : " + s);
|
|
final String[] arr = s.split("\\|-:-\\|");
|
|
if (arr[0] == null || arr[0].isEmpty()) {
|
|
continue;
|
|
}
|
|
synonyms.put(arr[0].toLowerCase(), arr[1]);
|
|
validTerms.add(arr[1].toLowerCase());
|
|
}
|
|
|
|
log.info("VOCABULARY " + vocabulary.trim() + " - terms size " + synonyms.size());
|
|
} catch (final Exception e) {
|
|
throw new CleanerException("Error obtaining vocabulary " + vocabulary, e);
|
|
}
|
|
}
|
|
|
|
}
|
|
|
|
@Override
|
|
protected Map<String, String> verifyValue(final String value) throws CleanerException {
|
|
if (synonyms.isEmpty()) {
|
|
log.warn("Vocabulary terms is void, vocabularies: " + this.vocabularies);
|
|
}
|
|
|
|
if (validTerms.contains(value.toLowerCase())) { return null; }
|
|
|
|
final Map<String, String> error = new HashMap<String, String>();
|
|
error.put("term", value);
|
|
error.put("vocabularies", this.vocabularies.toString().replaceAll("\\[", "").replaceAll("\\]", ""));
|
|
error.put("xpath", this.getXpath());
|
|
return error;
|
|
}
|
|
|
|
public Map<String, String> getVocabularyTerms() {
|
|
return synonyms;
|
|
}
|
|
|
|
@Override
|
|
public String toString() {
|
|
return "VOCABULARIES: [" + Joiner.on(", ").join(vocabularies) + "]";
|
|
}
|
|
|
|
}
|