| 1 |
29686
|
sandro.lab
|
package eu.dnetlib.functionality.index.query;
|
| 2 |
|
|
|
| 3 |
|
|
import java.util.List;
|
| 4 |
|
|
import java.util.Map;
|
| 5 |
|
|
|
| 6 |
|
|
import org.apache.commons.logging.Log;
|
| 7 |
|
|
import org.apache.commons.logging.LogFactory;
|
| 8 |
|
|
import org.dom4j.Document;
|
| 9 |
|
|
import org.dom4j.Element;
|
| 10 |
|
|
import org.springframework.beans.factory.annotation.Autowired;
|
| 11 |
|
|
import org.springframework.beans.factory.annotation.Required;
|
| 12 |
|
|
|
| 13 |
|
|
import com.google.common.collect.BiMap;
|
| 14 |
|
|
import com.google.common.collect.HashBiMap;
|
| 15 |
|
|
import com.google.common.collect.Maps;
|
| 16 |
|
|
|
| 17 |
|
|
import eu.dnetlib.data.provision.index.rmi.IndexServiceException;
|
| 18 |
|
|
import eu.dnetlib.functionality.index.client.IndexClientException;
|
| 19 |
|
|
import eu.dnetlib.functionality.index.utils.IndexFieldUtility;
|
| 20 |
|
|
import eu.dnetlib.functionality.index.utils.MDFormatReader;
|
| 21 |
|
|
import eu.dnetlib.functionality.index.utils.MetadataReference;
|
| 22 |
|
|
import eu.dnetlib.functionality.index.utils.ServiceTools;
|
| 23 |
|
|
|
| 24 |
|
|
/**
|
| 25 |
|
|
* The Class BrowseAliases.
|
| 26 |
|
|
*/
|
| 27 |
|
|
public class BrowseAliases {
|
| 28 |
|
|
|
| 29 |
|
|
/**
|
| 30 |
|
|
* logger.
|
| 31 |
|
|
*/
|
| 32 |
|
|
private static final Log log = LogFactory.getLog(BrowseAliases.class); // NOPMD
|
| 33 |
|
|
|
| 34 |
|
|
/** The aliases. */
|
| 35 |
|
|
private Map<MetadataReference, BiMap<String, String>> aliases = Maps.newConcurrentMap();
|
| 36 |
|
|
|
| 37 |
|
|
private ServiceTools serviceTools;
|
| 38 |
|
|
|
| 39 |
|
|
@Autowired
|
| 40 |
33052
|
alessia.ba
|
private MDFormatReader mdFormatReader;
|
| 41 |
29686
|
sandro.lab
|
|
| 42 |
|
|
/**
|
| 43 |
|
|
* Initialize.
|
| 44 |
33052
|
alessia.ba
|
*
|
| 45 |
29686
|
sandro.lab
|
* @throws IndexServiceException
|
| 46 |
|
|
* the index service exception
|
| 47 |
|
|
*/
|
| 48 |
|
|
public void initialize() throws IndexClientException {
|
| 49 |
|
|
log.info("initializing browse aliases");
|
| 50 |
|
|
for (MetadataReference mdRef : getServiceTools().listMDRefs()) {
|
| 51 |
33052
|
alessia.ba
|
log.debug("inside foreach");
|
| 52 |
29686
|
sandro.lab
|
put(mdRef);
|
| 53 |
|
|
}
|
| 54 |
|
|
log.info("browse aliases initialization completed");
|
| 55 |
|
|
}
|
| 56 |
|
|
|
| 57 |
|
|
/**
|
| 58 |
|
|
* Put.
|
| 59 |
33052
|
alessia.ba
|
*
|
| 60 |
29686
|
sandro.lab
|
* @param mdRef
|
| 61 |
|
|
* the metadata refeference
|
| 62 |
|
|
*/
|
| 63 |
|
|
public void put(final MetadataReference mdRef) {
|
| 64 |
|
|
final Document fields = mdFormatReader.getFields(mdRef);
|
| 65 |
|
|
if (fields != null) {
|
| 66 |
|
|
aliases.put(mdRef, extractBrowsingAliases(fields));
|
| 67 |
|
|
} else {
|
| 68 |
|
|
// log.info("couldn't find any");
|
| 69 |
|
|
BiMap<String, String> m = HashBiMap.create();
|
| 70 |
|
|
aliases.put(mdRef, m);
|
| 71 |
|
|
}
|
| 72 |
|
|
}
|
| 73 |
|
|
|
| 74 |
|
|
/**
|
| 75 |
|
|
* Gets the.
|
| 76 |
33052
|
alessia.ba
|
*
|
| 77 |
29686
|
sandro.lab
|
* @param mdRef
|
| 78 |
|
|
* the md ref
|
| 79 |
|
|
* @return browsing aliases for given mdRef.
|
| 80 |
|
|
* @throws IndexClientException
|
| 81 |
|
|
*/
|
| 82 |
|
|
public BiMap<String, String> get(final MetadataReference mdRef) throws IndexClientException {
|
| 83 |
|
|
if ((aliases == null) || (aliases.size() == 0)) {
|
| 84 |
|
|
initialize();
|
| 85 |
|
|
}
|
| 86 |
|
|
return aliases.get(mdRef);
|
| 87 |
|
|
}
|
| 88 |
|
|
|
| 89 |
|
|
/**
|
| 90 |
|
|
* Method extract aliases field names from the given fields.
|
| 91 |
33052
|
alessia.ba
|
*
|
| 92 |
29686
|
sandro.lab
|
* @param fields
|
| 93 |
|
|
* the fields
|
| 94 |
33052
|
alessia.ba
|
* @return aliases map. Keys are "normal" field names, values are names of the non-tokenized version of the field
|
| 95 |
29686
|
sandro.lab
|
*/
|
| 96 |
|
|
protected BiMap<String, String> extractBrowsingAliases(final Document fields) {
|
| 97 |
|
|
// default tokenizer splits field names, this would cause to
|
| 98 |
|
|
// have too many browsing results, so we use an untokenized
|
| 99 |
|
|
// alias in place of it.
|
| 100 |
|
|
|
| 101 |
|
|
final BiMap<String, String> aliases = HashBiMap.create();
|
| 102 |
|
|
|
| 103 |
|
|
@SuppressWarnings("unchecked")
|
| 104 |
|
|
final List<Element> fieldList = fields.getRootElement().selectNodes(IndexFieldUtility.XPATH_BROWSING_ALIAS_FOR);
|
| 105 |
|
|
for (final Element e : fieldList) {
|
| 106 |
|
|
final String name = e.attribute(IndexFieldUtility.FIELD_BROWSING_ALIAS_FOR).getValue().toLowerCase();
|
| 107 |
|
|
final String alias = e.attribute(IndexFieldUtility.FIELD_NAME).getValue().toLowerCase();
|
| 108 |
|
|
aliases.put(name, alias);
|
| 109 |
|
|
}
|
| 110 |
|
|
|
| 111 |
|
|
if (aliases.isEmpty()) {
|
| 112 |
|
|
log.warn("couldn'f find alias fields for browsing");
|
| 113 |
|
|
}
|
| 114 |
|
|
return aliases;
|
| 115 |
|
|
}
|
| 116 |
|
|
|
| 117 |
|
|
/**
|
| 118 |
|
|
* @return the serviceTools
|
| 119 |
|
|
*/
|
| 120 |
|
|
public ServiceTools getServiceTools() {
|
| 121 |
|
|
return serviceTools;
|
| 122 |
|
|
}
|
| 123 |
|
|
|
| 124 |
|
|
/**
|
| 125 |
|
|
* @param serviceTools
|
| 126 |
|
|
* the serviceTools to set
|
| 127 |
|
|
*/
|
| 128 |
|
|
@Required
|
| 129 |
|
|
public void setServiceTools(final ServiceTools serviceTools) {
|
| 130 |
|
|
this.serviceTools = serviceTools;
|
| 131 |
|
|
}
|
| 132 |
|
|
|
| 133 |
|
|
}
|