| 1 |
41977
|
sandro.lab
|
package eu.dnetlib.index.actors;
|
| 2 |
|
|
|
| 3 |
52694
|
alessia.ba
|
import java.util.Map;
|
| 4 |
41977
|
sandro.lab
|
import java.util.concurrent.ExecutorService;
|
| 5 |
|
|
import java.util.concurrent.Executors;
|
| 6 |
|
|
import java.util.function.Function;
|
| 7 |
57157
|
enrico.ott
|
import java.util.stream.Stream;
|
| 8 |
|
|
import java.util.stream.StreamSupport;
|
| 9 |
41977
|
sandro.lab
|
|
| 10 |
52694
|
alessia.ba
|
import com.google.common.collect.Maps;
|
| 11 |
57157
|
enrico.ott
|
|
| 12 |
|
|
import eu.dnetlib.clients.index.model.Any.ValueType;
|
| 13 |
41977
|
sandro.lab
|
import eu.dnetlib.clients.index.model.document.IndexDocument;
|
| 14 |
|
|
import eu.dnetlib.clients.index.utils.IndexFieldUtility;
|
| 15 |
|
|
import eu.dnetlib.clients.index.utils.ServiceTools;
|
| 16 |
52694
|
alessia.ba
|
import eu.dnetlib.cql.CqlTranslator;
|
| 17 |
41977
|
sandro.lab
|
import eu.dnetlib.enabling.tools.DnetStreamSupport;
|
| 18 |
|
|
import eu.dnetlib.index.IndexCollection;
|
| 19 |
|
|
import eu.dnetlib.index.IndexServerDAO;
|
| 20 |
|
|
import eu.dnetlib.index.IndexServerDAOMap;
|
| 21 |
|
|
import eu.dnetlib.index.feed.DocumentFeeder;
|
| 22 |
|
|
import eu.dnetlib.index.feed.DocumentMapperFactory;
|
| 23 |
|
|
import eu.dnetlib.index.feed.FeedMode;
|
| 24 |
|
|
import eu.dnetlib.index.feed.FeedResult;
|
| 25 |
52694
|
alessia.ba
|
import eu.dnetlib.index.utils.IndexDateUtility;
|
| 26 |
41977
|
sandro.lab
|
import eu.dnetlib.miscutils.datetime.DateUtils;
|
| 27 |
|
|
import eu.dnetlib.rmi.provision.IndexServiceException;
|
| 28 |
52694
|
alessia.ba
|
import eu.dnetlib.utils.MetadataReference;
|
| 29 |
41977
|
sandro.lab
|
import org.apache.commons.logging.Log;
|
| 30 |
|
|
import org.apache.commons.logging.LogFactory;
|
| 31 |
|
|
|
| 32 |
|
|
/**
|
| 33 |
|
|
* The Class IndexFeedActorImpl.
|
| 34 |
|
|
*/
|
| 35 |
|
|
public class IndexFeedActorImpl implements IndexFeedActor {
|
| 36 |
|
|
|
| 37 |
|
|
/**
|
| 38 |
|
|
* The Constant log.
|
| 39 |
|
|
*/
|
| 40 |
|
|
private static final Log log = LogFactory.getLog(IndexFeedActorImpl.class); // NOPMD by marko on 11/24/08 5:02 PM
|
| 41 |
|
|
|
| 42 |
|
|
/**
|
| 43 |
|
|
* The index server dao map.
|
| 44 |
|
|
*/
|
| 45 |
|
|
private final transient IndexServerDAOMap indexServerDAOMap;
|
| 46 |
|
|
|
| 47 |
|
|
/**
|
| 48 |
|
|
* The service tools.
|
| 49 |
|
|
*/
|
| 50 |
|
|
private final transient ServiceTools serviceTools;
|
| 51 |
|
|
/**
|
| 52 |
|
|
* Thread pool used for the feeding process.
|
| 53 |
|
|
*/
|
| 54 |
|
|
private final transient ExecutorService threadPool = Executors.newCachedThreadPool();
|
| 55 |
|
|
/**
|
| 56 |
|
|
* CqlTranslator.
|
| 57 |
|
|
*/
|
| 58 |
|
|
private CqlTranslator translator;
|
| 59 |
|
|
|
| 60 |
|
|
/**
|
| 61 |
|
|
* Instantiates a new index feed actor impl.
|
| 62 |
|
|
*
|
| 63 |
|
|
* @param indexServerDAOMap the index server dao map
|
| 64 |
|
|
* @param serviceTools the service tools
|
| 65 |
|
|
*/
|
| 66 |
|
|
public IndexFeedActorImpl(final IndexServerDAOMap indexServerDAOMap, final ServiceTools serviceTools, final CqlTranslator translator) {
|
| 67 |
|
|
super();
|
| 68 |
|
|
this.indexServerDAOMap = indexServerDAOMap;
|
| 69 |
|
|
this.serviceTools = serviceTools;
|
| 70 |
|
|
this.translator = translator;
|
| 71 |
|
|
}
|
| 72 |
|
|
|
| 73 |
|
|
/**
|
| 74 |
|
|
* {@inheritDoc}
|
| 75 |
|
|
*/
|
| 76 |
|
|
@Override
|
| 77 |
|
|
public void feedIndex(final String dsId,
|
| 78 |
|
|
final FeedMode feedMode,
|
| 79 |
|
|
final Iterable<String> docIterator,
|
| 80 |
|
|
final ResultsetKeepAliveCallback startCallback,
|
| 81 |
|
|
final BlackboardActorCallback endCallback,
|
| 82 |
|
|
final String backendId, final boolean emptyResult) {
|
| 83 |
|
|
IndexCollection indexCollection = null;
|
| 84 |
|
|
try {
|
| 85 |
|
|
final MetadataReference mdref = serviceTools.getMetadataRef(dsId);
|
| 86 |
|
|
|
| 87 |
|
|
final IndexServerDAO serverDAO = indexServerDAOMap.getIndexServerDAO(backendId);
|
| 88 |
|
|
final DocumentMapperFactory docMapperFactory = serverDAO.getDocumentMapperFactory();
|
| 89 |
50565
|
alessia.ba
|
final String version = IndexDateUtility.getParsedDateField(DateUtils.now_ISO8601());
|
| 90 |
57157
|
enrico.ott
|
Map<String, ValueType> fields = serverDAO.getSchema(mdref);
|
| 91 |
|
|
final Function<String, IndexDocument> docMapper = docMapperFactory.getRecordMapper(fields, mdref, dsId, version, emptyResult);
|
| 92 |
|
|
|
| 93 |
|
|
Stream<IndexDocument> docStream = DnetStreamSupport.generateStreamFromIterator(docIterator.iterator()).map(docMapper);
|
| 94 |
|
|
|
| 95 |
41977
|
sandro.lab
|
indexCollection = serverDAO.getIndexCollection(mdref);
|
| 96 |
57157
|
enrico.ott
|
final FeedResult res = threadPool.submit(new DocumentFeeder(indexCollection, docStream)).get();
|
| 97 |
|
|
|
| 98 |
41977
|
sandro.lab
|
cleanMarkedDocuments(indexCollection, dsId);
|
| 99 |
|
|
if (feedMode.equals(FeedMode.REFRESH)) {
|
| 100 |
50565
|
alessia.ba
|
deleteByVersion(indexCollection, dsId, version);
|
| 101 |
|
|
}
|
| 102 |
41977
|
sandro.lab
|
|
| 103 |
|
|
indexCollection.commit();
|
| 104 |
|
|
indexCollection.shutdown();
|
| 105 |
|
|
|
| 106 |
|
|
log.info("FeedResult: " + res.setTimeElapsed(System.currentTimeMillis()));
|
| 107 |
52694
|
alessia.ba
|
endCallback.setJobDone(buildParams(res));
|
| 108 |
41977
|
sandro.lab
|
} catch (final Exception e) {
|
| 109 |
|
|
endCallback.setJobFailed(e);
|
| 110 |
|
|
log.error("feed index job failed", e);
|
| 111 |
|
|
} finally {
|
| 112 |
|
|
if (indexCollection != null) {
|
| 113 |
|
|
indexCollection.shutdown();
|
| 114 |
|
|
}
|
| 115 |
|
|
}
|
| 116 |
|
|
|
| 117 |
|
|
}
|
| 118 |
|
|
|
| 119 |
52694
|
alessia.ba
|
private Map<String, String> buildParams(final FeedResult feedResult) {
|
| 120 |
|
|
Map<String, String> params = Maps.newHashMap();
|
| 121 |
|
|
params.put("added", String.valueOf(feedResult.getAdded()));
|
| 122 |
|
|
params.put("skipped", String.valueOf(feedResult.getSkipped()));
|
| 123 |
|
|
params.put("marked", String.valueOf(feedResult.getMarked()));
|
| 124 |
|
|
params.put("time", String.valueOf(feedResult.getTime()));
|
| 125 |
|
|
return params;
|
| 126 |
|
|
}
|
| 127 |
|
|
|
| 128 |
41977
|
sandro.lab
|
/**
|
| 129 |
|
|
* mdFormatVersion.
|
| 130 |
|
|
*
|
| 131 |
|
|
* @param indexCollection the server dao
|
| 132 |
|
|
* @param dsId the ds id
|
| 133 |
|
|
* @throws IndexServiceException the index service exception
|
| 134 |
|
|
*/
|
| 135 |
44419
|
claudio.at
|
private void deleteByVersion(final IndexCollection indexCollection, final String dsId, final String version) throws IndexServiceException {
|
| 136 |
44824
|
sandro.lab
|
final String query = String.format("%s:[* TO \"%s\"}", IndexFieldUtility.DS_VERSION, version);
|
| 137 |
|
|
indexCollection.deleteByQuery(query, dsId);
|
| 138 |
41977
|
sandro.lab
|
}
|
| 139 |
|
|
|
| 140 |
|
|
/**
|
| 141 |
|
|
* method delete documents where IndexMap.DELETE_DOCUMENT field is true
|
| 142 |
|
|
*
|
| 143 |
|
|
* @param indexCollection the server dao
|
| 144 |
|
|
* @param dsId the ds id
|
| 145 |
|
|
* @throws IndexServiceException the index service exception
|
| 146 |
|
|
*/
|
| 147 |
|
|
public void cleanMarkedDocuments(final IndexCollection indexCollection, final String dsId) throws IndexServiceException {
|
| 148 |
44420
|
claudio.at
|
final String query = IndexFieldUtility.DELETE_DOCUMENT + ":true ";
|
| 149 |
|
|
indexCollection.deleteByQuery(query, dsId);
|
| 150 |
41977
|
sandro.lab
|
}
|
| 151 |
|
|
|
| 152 |
|
|
}
|