Revision 50543
Added by Claudio Atzori about 6 years ago
| OpenaireIdResolver.java | ||
|---|---|---|
| 3 | 3 |
import java.io.StringReader; |
| 4 | 4 |
import java.util.List; |
| 5 | 5 |
|
| 6 |
import com.google.common.base.Function; |
|
| 7 |
import com.google.common.collect.Iterables; |
|
| 8 |
import com.google.common.collect.Lists; |
|
| 9 |
import eu.dnetlib.enabling.is.lookup.rmi.ISLookUpException; |
|
| 10 |
import eu.dnetlib.enabling.is.lookup.rmi.ISLookUpService; |
|
| 11 |
import eu.dnetlib.enabling.locators.UniqueServiceLocator; |
|
| 12 |
import eu.dnetlib.functionality.index.client.solr.SolrIndexClient; |
|
| 13 |
import eu.dnetlib.functionality.index.client.solr.SolrIndexClientFactory; |
|
| 6 | 14 |
import org.apache.commons.lang.StringUtils; |
| 7 | 15 |
import org.apache.commons.logging.Log; |
| 8 | 16 |
import org.apache.commons.logging.LogFactory; |
| 9 |
import org.apache.solr.client.solrj.SolrQuery; |
|
| 10 |
import org.apache.solr.client.solrj.impl.CloudSolrClient; |
|
| 11 | 17 |
import org.apache.solr.client.solrj.response.QueryResponse; |
| 12 | 18 |
import org.apache.solr.common.SolrDocument; |
| 13 | 19 |
import org.apache.solr.common.SolrDocumentList; |
| ... | ... | |
| 17 | 23 |
import org.dom4j.io.SAXReader; |
| 18 | 24 |
import org.springframework.beans.factory.annotation.Autowired; |
| 19 | 25 |
|
| 20 |
import com.google.common.base.Function; |
|
| 21 |
import com.google.common.collect.Iterables; |
|
| 22 |
import com.google.common.collect.Lists; |
|
| 23 |
|
|
| 24 |
import eu.dnetlib.enabling.is.lookup.rmi.ISLookUpException; |
|
| 25 |
import eu.dnetlib.enabling.is.lookup.rmi.ISLookUpService; |
|
| 26 |
import eu.dnetlib.enabling.locators.UniqueServiceLocator; |
|
| 27 |
|
|
| 28 | 26 |
/** |
| 29 | 27 |
* The goal of this task is to return the original identifiers of objects merged in a representative object by deduplication. Created by |
| 30 | 28 |
* alessia on 09/02/16. |
| 31 | 29 |
*/ |
| 32 | 30 |
public class OpenaireIdResolver {
|
| 33 | 31 |
|
| 34 |
public final static String SOLR_COLLECTION_POSTFIX = "-index-openaire"; |
|
| 35 | 32 |
private static final Log log = LogFactory.getLog(OpenaireIdResolver.class); |
| 36 | 33 |
private final static String RESULT_FIELD = "__result"; |
| 37 | 34 |
private final static String XPATH_TO_MERGED = "//*[local-name()='entity']/*//children/result/@objidentifier"; |
| 38 | 35 |
private final SAXReader saxReader = new SAXReader(); |
| 39 | 36 |
@Autowired |
| 40 | 37 |
private UniqueServiceLocator serviceLocator; |
| 38 |
@Autowired |
|
| 39 |
private SolrIndexClientFactory indexClientFactory; |
|
| 41 | 40 |
|
| 42 | 41 |
public List<String> resolveIdentifier(final String id) {
|
| 43 | 42 |
if (StringUtils.isBlank(id)) return Lists.newArrayList(); |
| ... | ... | |
| 47 | 46 |
} |
| 48 | 47 |
|
| 49 | 48 |
protected List<String> findOriginalIds(final String id) {
|
| 49 |
try(final SolrIndexClient client = (SolrIndexClient) indexClientFactory.getClient(getPublicIndexCollection())) {
|
|
| 50 | 50 |
|
| 51 |
final SolrQuery q = new SolrQuery("objidentifier:\"" + id + "\"");
|
|
| 52 |
|
|
| 53 |
try (final CloudSolrClient solrCore = new CloudSolrClient.Builder().withZkHost(getIndexEndpoint()).build()) {
|
|
| 54 |
solrCore.setDefaultCollection(getPublicIndexCollection() + SOLR_COLLECTION_POSTFIX); |
|
| 55 |
final QueryResponse response = solrCore.query(q); |
|
| 51 |
final String query = String.format("objidentifier:\"%s\"", id);
|
|
| 52 |
final QueryResponse response = client.query(query, 1); |
|
| 56 | 53 |
final SolrDocumentList results = response.getResults(); |
| 57 | 54 |
if (results.isEmpty()) {
|
| 58 |
log.debug("Query " + q + " returned 0 documents");
|
|
| 55 |
log.debug("Query " + query + " returned 0 documents");
|
|
| 59 | 56 |
return Lists.newArrayList(); |
| 60 | 57 |
} |
| 61 | 58 |
// my results contain the document with the given identifier |
| ... | ... | |
| 71 | 68 |
protected List<String> extractMergedIdentifiers(final SolrDocument doc) throws DocumentException {
|
| 72 | 69 |
final String xmlRecord = (String) doc.getFirstValue(RESULT_FIELD); |
| 73 | 70 |
final Document xmlDoc = this.saxReader.read(new StringReader(xmlRecord)); |
| 74 |
return Lists.newArrayList(Iterables.transform(xmlDoc.selectNodes(XPATH_TO_MERGED), new Function<Attribute, String>() {
|
|
| 75 |
|
|
| 76 |
@Override |
|
| 77 |
public String apply(final Attribute a) {
|
|
| 78 |
return a.getStringValue(); |
|
| 79 |
} |
|
| 80 |
})); |
|
| 71 |
return Lists.newArrayList(Iterables.transform(xmlDoc.selectNodes(XPATH_TO_MERGED), (Function<Attribute, String>) a -> a.getStringValue())); |
|
| 81 | 72 |
} |
| 82 | 73 |
|
| 83 | 74 |
protected String getIndexEndpoint() throws ISLookUpException {
|
Also available in: Unified diff
consolidated index modules and updated to solr 7.2.0