Revision c80938b2
almost 100% fixing #2950 (Allow grouping of lucene (hibernate search) results) - only max score is not yet available
| .gitattributes | ||
|---|---|---|
| 606 | 606 |
cdmlib-model/src/main/java/eu/etaxonomy/cdm/hibernate/search/DateTimeBridge.java -text |
| 607 | 607 |
cdmlib-model/src/main/java/eu/etaxonomy/cdm/hibernate/search/DefinedTermBaseClassBridge.java -text |
| 608 | 608 |
cdmlib-model/src/main/java/eu/etaxonomy/cdm/hibernate/search/DescriptionBaseClassBridge.java -text |
| 609 |
cdmlib-model/src/main/java/eu/etaxonomy/cdm/hibernate/search/GroupByTaxonClassBridge.java -text |
|
| 609 | 610 |
cdmlib-model/src/main/java/eu/etaxonomy/cdm/hibernate/search/IdFieldOptions.java -text |
| 610 | 611 |
cdmlib-model/src/main/java/eu/etaxonomy/cdm/hibernate/search/MultilanguageTextFieldBridge.java -text |
| 611 | 612 |
cdmlib-model/src/main/java/eu/etaxonomy/cdm/hibernate/search/NotNullAwareIdBridge.java -text |
| cdmlib-model/src/main/java/eu/etaxonomy/cdm/hibernate/search/AbstractClassBridge.java | ||
|---|---|---|
| 17 | 17 |
protected final static NotNullAwareIdBridge idFieldBridge = new NotNullAwareIdBridge(); |
| 18 | 18 |
|
| 19 | 19 |
public static LuceneOptions idFieldOptions = new IdFieldOptions(); |
| 20 |
|
|
| 20 | 21 |
public static LuceneOptions sortFieldOptions = new SortFieldOptions(); |
| 21 | 22 |
|
| 22 | 23 |
|
| cdmlib-model/src/main/java/eu/etaxonomy/cdm/hibernate/search/DescriptionBaseClassBridge.java | ||
|---|---|---|
| 50 | 50 |
*/ |
| 51 | 51 |
public void set(String name, Object entity, Document document, LuceneOptions luceneOptions) {
|
| 52 | 52 |
|
| 53 |
|
|
| 54 | 53 |
if (entity instanceof TaxonDescription) {
|
| 55 |
Taxon taxon = ((TaxonDescription) entity).getTaxon(); |
|
| 54 |
|
|
| 55 |
Taxon taxon = ((TaxonDescription)entity).getTaxon(); |
|
| 56 | 56 |
|
| 57 | 57 |
if (taxon != null) {
|
| 58 | 58 |
|
| ... | ... | |
| 86 | 86 |
} |
| 87 | 87 |
} |
| 88 | 88 |
|
| 89 |
|
|
| 89 | 90 |
} |
| cdmlib-model/src/main/java/eu/etaxonomy/cdm/hibernate/search/GroupByTaxonClassBridge.java | ||
|---|---|---|
| 1 |
package eu.etaxonomy.cdm.hibernate.search; |
|
| 2 |
|
|
| 3 |
import org.apache.lucene.document.Document; |
|
| 4 |
import org.hibernate.search.annotations.ClassBridge; |
|
| 5 |
import org.hibernate.search.bridge.LuceneOptions; |
|
| 6 |
|
|
| 7 |
import eu.etaxonomy.cdm.model.description.TaxonDescription; |
|
| 8 |
import eu.etaxonomy.cdm.model.taxon.Taxon; |
|
| 9 |
|
|
| 10 |
/** |
|
| 11 |
* The <code>GroupByTaxonClassBridge</code> adds the field |
|
| 12 |
* <code>groupby_taxon.id</code> to the lucene document which can be used to |
|
| 13 |
* group search results based on the taxon which is associated with the indexed |
|
| 14 |
* cdm entity. So any cdm class which is involved in querying for taxa must |
|
| 15 |
* used this class bridge, e.g.: |
|
| 16 |
* |
|
| 17 |
* <pre> |
|
| 18 |
@ClassBridge(impl=GroupByTaxonClassBridge.class)} |
|
| 19 |
</pre> |
|
| 20 |
* |
|
| 21 |
* or |
|
| 22 |
* <pre> |
|
| 23 |
@ClassBridges({
|
|
| 24 |
@ClassBridge(impl=GroupByTaxonClassBridge.class), |
|
| 25 |
@ClassBridge(impl=DescriptionBaseClassBridge.class), |
|
| 26 |
}) |
|
| 27 |
} |
|
| 28 |
* </pre> |
|
| 29 |
* |
|
| 30 |
* @author a.kohlbecker |
|
| 31 |
* @date Oct 4, 2012 |
|
| 32 |
* |
|
| 33 |
*/ |
|
| 34 |
public class GroupByTaxonClassBridge extends AbstractClassBridge {
|
|
| 35 |
|
|
| 36 |
public static final String GROUPBY_TAXON_FIELD = "groupby_taxon.id"; |
|
| 37 |
|
|
| 38 |
public GroupByTaxonClassBridge() {
|
|
| 39 |
super(); |
|
| 40 |
} |
|
| 41 |
|
|
| 42 |
/** |
|
| 43 |
* @param entity |
|
| 44 |
* @return |
|
| 45 |
*/ |
|
| 46 |
protected Taxon getAssociatedTaxon(Object entity) {
|
|
| 47 |
|
|
| 48 |
if (entity instanceof TaxonDescription) {
|
|
| 49 |
return ((TaxonDescription) entity).getTaxon(); |
|
| 50 |
} |
|
| 51 |
if (entity instanceof Taxon) {
|
|
| 52 |
return (Taxon)entity; |
|
| 53 |
} |
|
| 54 |
|
|
| 55 |
throw new RuntimeException("CDM class " + entity.getClass() + " not yet supported");
|
|
| 56 |
} |
|
| 57 |
|
|
| 58 |
@Override |
|
| 59 |
public void set(String name, Object value, Document document, LuceneOptions luceneOptions) {
|
|
| 60 |
|
|
| 61 |
Taxon taxon = getAssociatedTaxon(value); |
|
| 62 |
if(taxon != null){
|
|
| 63 |
idFieldBridge.set(GROUPBY_TAXON_FIELD, taxon.getId(), document, idFieldOptions); |
|
| 64 |
} |
|
| 65 |
} |
|
| 66 |
|
|
| 67 |
} |
|
| cdmlib-model/src/main/java/eu/etaxonomy/cdm/model/description/DescriptionBase.java | ||
|---|---|---|
| 34 | 34 |
import org.hibernate.annotations.CascadeType; |
| 35 | 35 |
import org.hibernate.envers.Audited; |
| 36 | 36 |
import org.hibernate.search.annotations.ClassBridge; |
| 37 |
import org.hibernate.search.annotations.ClassBridges; |
|
| 37 | 38 |
import org.hibernate.search.annotations.ContainedIn; |
| 38 | 39 |
import org.hibernate.search.annotations.Field; |
| 39 | 40 |
|
| 40 | 41 |
import eu.etaxonomy.cdm.hibernate.search.DescriptionBaseClassBridge; |
| 42 |
import eu.etaxonomy.cdm.hibernate.search.GroupByTaxonClassBridge; |
|
| 41 | 43 |
import eu.etaxonomy.cdm.model.common.IdentifiableEntity; |
| 42 | 44 |
import eu.etaxonomy.cdm.model.name.NameRelationship; |
| 43 | 45 |
import eu.etaxonomy.cdm.model.name.TypeDesignationBase; |
| ... | ... | |
| 73 | 75 |
@Entity |
| 74 | 76 |
@Audited |
| 75 | 77 |
@Inheritance(strategy=InheritanceType.SINGLE_TABLE) |
| 76 |
@ClassBridge(impl=DescriptionBaseClassBridge.class) |
|
| 78 |
@ClassBridges({
|
|
| 79 |
@ClassBridge(impl=DescriptionBaseClassBridge.class), |
|
| 80 |
@ClassBridge(impl=GroupByTaxonClassBridge.class) |
|
| 81 |
}) |
|
| 77 | 82 |
public abstract class DescriptionBase<S extends IIdentifiableEntityCacheStrategy> extends IdentifiableEntity<S> {
|
| 78 | 83 |
private static final long serialVersionUID = 5504218413819040193L; |
| 79 | 84 |
private static final Logger logger = Logger.getLogger(DescriptionBase.class); |
| cdmlib-model/src/main/java/eu/etaxonomy/cdm/model/taxon/Taxon.java | ||
|---|---|---|
| 40 | 40 |
import org.hibernate.annotations.Cascade; |
| 41 | 41 |
import org.hibernate.annotations.CascadeType; |
| 42 | 42 |
import org.hibernate.envers.Audited; |
| 43 |
import org.hibernate.search.annotations.ClassBridge; |
|
| 43 | 44 |
import org.hibernate.search.annotations.ContainedIn; |
| 44 | 45 |
import org.hibernate.search.annotations.Indexed; |
| 45 | 46 |
import org.hibernate.search.annotations.IndexedEmbedded; |
| 46 | 47 |
import org.springframework.beans.factory.annotation.Configurable; |
| 47 | 48 |
import org.springframework.util.ReflectionUtils; |
| 48 | 49 |
|
| 50 |
import eu.etaxonomy.cdm.hibernate.search.GroupByTaxonClassBridge; |
|
| 49 | 51 |
import eu.etaxonomy.cdm.model.common.IRelated; |
| 50 | 52 |
import eu.etaxonomy.cdm.model.common.RelationshipBase; |
| 51 | 53 |
import eu.etaxonomy.cdm.model.description.TaxonDescription; |
| ... | ... | |
| 82 | 84 |
@Indexed(index = "eu.etaxonomy.cdm.model.taxon.TaxonBase") |
| 83 | 85 |
@Audited |
| 84 | 86 |
@Configurable |
| 87 |
@ClassBridge(impl=GroupByTaxonClassBridge.class) |
|
| 85 | 88 |
public class Taxon extends TaxonBase<IIdentifiableEntityCacheStrategy<Taxon>> implements IRelated<RelationshipBase>, Cloneable{
|
| 86 | 89 |
private static final long serialVersionUID = -584946869762749006L; |
| 87 | 90 |
private static final Logger logger = Logger.getLogger(Taxon.class); |
| cdmlib-services/src/main/java/eu/etaxonomy/cdm/api/service/TaxonServiceImpl.java | ||
|---|---|---|
| 1217 | 1217 |
Map<CdmBaseType, String> idFieldMap = new HashMap<CdmBaseType, String>(); |
| 1218 | 1218 |
idFieldMap.put(CdmBaseType.DESCRIPTION_ELEMENT, "inDescription.taxon.id"); |
| 1219 | 1219 |
|
| 1220 |
// --- initialize taxa, thighlight matches ....
|
|
| 1220 |
// --- initialize taxa, highlight matches .... |
|
| 1221 | 1221 |
ISearchResultBuilder searchResultBuilder = new SearchResultBuilder(luceneSearch, luceneSearch.getQuery()); |
| 1222 | 1222 |
@SuppressWarnings("rawtypes")
|
| 1223 | 1223 |
List<SearchResult<TaxonBase>> searchResults = searchResultBuilder.createResultSet( |
| cdmlib-services/src/main/java/eu/etaxonomy/cdm/api/service/search/LuceneSearch.java | ||
|---|---|---|
| 42 | 42 |
import org.hibernate.search.reader.ReaderProvider; |
| 43 | 43 |
import org.hibernate.search.store.DirectoryProvider; |
| 44 | 44 |
|
| 45 |
import eu.etaxonomy.cdm.hibernate.search.GroupByTaxonClassBridge; |
|
| 45 | 46 |
import eu.etaxonomy.cdm.model.common.CdmBase; |
| 46 | 47 |
import eu.etaxonomy.cdm.model.description.DescriptionElementBase; |
| 47 | 48 |
import eu.etaxonomy.cdm.model.description.TextData; |
| ... | ... | |
| 56 | 57 |
*/ |
| 57 | 58 |
public class LuceneSearch {
|
| 58 | 59 |
|
| 59 |
private static final String GROUP_BY_FIELD = "id";
|
|
| 60 |
private static final String GROUP_BY_FIELD = GroupByTaxonClassBridge.GROUPBY_TAXON_FIELD;
|
|
| 60 | 61 |
|
| 61 | 62 |
public static final Logger logger = Logger.getLogger(LuceneSearch.class); |
| 62 | 63 |
|
| ... | ... | |
| 112 | 113 |
|
| 113 | 114 |
protected String[] highlightFields = new String[0]; |
| 114 | 115 |
|
| 116 |
private int maxDocsPerGroup = 10; |
|
| 117 |
|
|
| 118 |
|
|
| 119 |
public int getMaxDocsPerGroup() {
|
|
| 120 |
return maxDocsPerGroup; |
|
| 121 |
} |
|
| 122 |
|
|
| 123 |
public void setMaxDocsPerGroup(int maxDocsPerGroup) {
|
|
| 124 |
this.maxDocsPerGroup = maxDocsPerGroup; |
|
| 125 |
} |
|
| 115 | 126 |
|
| 116 | 127 |
/** |
| 117 | 128 |
* @param session |
| ... | ... | |
| 238 | 249 |
} |
| 239 | 250 |
|
| 240 | 251 |
Query fullQuery = expandQuery(); |
| 241 |
|
|
| 242 | 252 |
logger.info("final query: " + fullQuery.toString());
|
| 243 | 253 |
|
| 244 | 254 |
int offset = pageNumber * pageSize; |
| 245 | 255 |
int limit = (pageNumber + 1) * pageSize - 1 ; |
| 246 |
|
|
| 247 | 256 |
logger.debug("start: " + offset + "; limit:" + limit);
|
| 248 | 257 |
|
| 249 |
// TopDocs topDocs = null; |
|
| 250 |
|
|
| 251 |
// sort must be non null default: Sort.RELEVANCE |
|
| 252 | 258 |
Sort groupSort = null; |
| 253 | 259 |
Sort withinGroupSort = Sort.RELEVANCE; |
| 254 | 260 |
if(sortFields != null && sortFields.length > 0){
|
| 255 | 261 |
Sort sort = new Sort(sortFields); |
| 256 | 262 |
groupSort = new Sort(sortFields); |
| 257 |
// topDocs = getSearcher().search(fullQuery, null, limit, sort); |
|
| 258 | 263 |
} else {
|
| 259 | 264 |
groupSort = Sort.RELEVANCE; // == SortField.FIELD_SCORE !! |
| 260 |
// topDocs = getSearcher().search(fullQuery, null, limit); |
|
| 261 | 265 |
} |
| 266 |
|
|
| 262 | 267 |
FirstPassGroupingCollector groupingCollector_1 = new FirstPassGroupingCollector(GROUP_BY_FIELD, withinGroupSort, limit); |
| 263 | 268 |
getSearcher().search(fullQuery, groupingCollector_1); |
| 264 | 269 |
|
| ... | ... | |
| 272 | 277 |
boolean getMaxScores = true; |
| 273 | 278 |
boolean fillFields = true; |
| 274 | 279 |
AllGroupsCollector c3 = new AllGroupsCollector(GROUP_BY_FIELD); |
| 275 |
SecondPassGroupingCollector c2 = new SecondPassGroupingCollector(GROUP_BY_FIELD, topGroups, groupSort, withinGroupSort, limit, getScores, getMaxScores, fillFields);
|
|
| 280 |
SecondPassGroupingCollector c2 = new SecondPassGroupingCollector(GROUP_BY_FIELD, topGroups, groupSort, withinGroupSort, maxDocsPerGroup , getScores, getMaxScores, fillFields);
|
|
| 276 | 281 |
getSearcher().search(fullQuery, MultiCollector.wrap(c2, c3)); |
| 277 | 282 |
|
| 278 | 283 |
TopGroups groupsResult = c2.getTopGroups(offset); |
| cdmlib-services/src/test/java/eu/etaxonomy/cdm/api/service/TaxonServiceSearchTest.java | ||
|---|---|---|
| 275 | 275 |
Pager<SearchResult<TaxonBase>> pager; |
| 276 | 276 |
|
| 277 | 277 |
pager = taxonService.findByDescriptionElementFullText(CommonTaxonName.class, "Rot*", null, null, null, false, null, null, null, null); |
| 278 |
Assert.assertEquals("Expecting 1024 entities when searching for Rot*", Integer.valueOf(1024), pager.getCount());
|
|
| 278 |
Assert.assertEquals("Expecting all 1024 entities grouped into one SearchResult item when searching for Rot*", 1, pager.getCount().intValue());
|
|
| 279 | 279 |
} |
| 280 | 280 |
|
| 281 | 281 |
/** |
| ... | ... | |
| 356 | 356 |
pager = taxonService.findByDescriptionElementFullText(CommonTaxonName.class, "Rot*", null, null, null, highlightFragments, pageSize, null, null, null); |
| 357 | 357 |
Assert.assertEquals("All matches should be grouped in one page", 1, pager.getPagesAvailable().intValue());
|
| 358 | 358 |
Map<String, String[]> highlightMap = pager.getRecords().get(0).getFieldHighlightMap(); |
| 359 |
Assert.assertEquals("expecting " + numOfItems+ " highlighted fragments of field 'name'", numOfItems, highlightMap.get("name").length);
|
|
| 359 |
// maxDocsPerGroup is defined in LuceneSearch and defaults to 10 |
|
| 360 |
int maxDocsPerGroup = 10; |
|
| 361 |
Assert.assertEquals("expecting 10 highlighted fragments of field 'name'", maxDocsPerGroup, highlightMap.get("name").length);
|
|
| 360 | 362 |
|
| 361 | 363 |
} |
| 362 | 364 |
|
| cdmlib-services/src/test/resources/log4j.properties | ||
|---|---|---|
| 35 | 35 |
#log4j.logger.eu.etaxonomy.cdm.model.taxon.TaxonNodeByNameComparator=trace |
| 36 | 36 |
# |
| 37 | 37 |
log4j.logger.eu.etaxonomy.cdm.test.function = info |
| 38 |
log4j.logger.eu.etaxonomy.cdm.test.integration = DEBUG
|
|
| 38 |
log4j.logger.eu.etaxonomy.cdm.test.integration = info
|
|
| 39 | 39 |
log4j.logger.eu.etaxonomy.cdm.api.application = warn |
| 40 | 40 |
log4j.logger.eu.etaxonomy.cdm.api.service = debug |
| 41 | 41 |
#log4j.logger.eu.etaxonomy.cdm.database.VocabularyStoreImpl = warn |
Also available in: Unified diff
