org.hibernate.search.backend.elasticsearch.impl.ElasticsearchHSQueryImpl Maven / Gradle / Ivy
Go to download
Show more of this group Show more artifacts with this name
Show all versions of hibernate-search-backend-elasticsearch Show documentation
Show all versions of hibernate-search-backend-elasticsearch Show documentation
Hibernate Search backend which has indexing operations forwarded to Elasticsearch
/*
* Hibernate Search, full-text search for your domain model
*
* License: GNU Lesser General Public License (LGPL), version 2.1 or later
* See the lgpl.txt file in the root directory or .
*/
package org.hibernate.search.backend.elasticsearch.impl;
import java.io.IOException;
import java.io.Serializable;
import java.util.ArrayList;
import java.util.Collections;
import java.util.HashMap;
import java.util.HashSet;
import java.util.List;
import java.util.Map;
import java.util.Set;
import java.util.regex.Pattern;
import org.apache.lucene.document.Document;
import org.apache.lucene.document.DoubleField;
import org.apache.lucene.document.Field.Store;
import org.apache.lucene.document.FloatField;
import org.apache.lucene.document.IntField;
import org.apache.lucene.document.LongField;
import org.apache.lucene.document.StringField;
import org.apache.lucene.search.Explanation;
import org.apache.lucene.search.Filter;
import org.apache.lucene.search.Query;
import org.apache.lucene.search.SortField;
import org.apache.lucene.search.TopDocs;
import org.hibernate.search.backend.elasticsearch.ProjectionConstants;
import org.hibernate.search.backend.elasticsearch.client.impl.JestClient;
import org.hibernate.search.backend.elasticsearch.logging.impl.Log;
import org.hibernate.search.bridge.FieldBridge;
import org.hibernate.search.bridge.TwoWayFieldBridge;
import org.hibernate.search.engine.integration.impl.ExtendedSearchIntegrator;
import org.hibernate.search.engine.metadata.impl.DocumentFieldMetadata;
import org.hibernate.search.engine.service.spi.ServiceReference;
import org.hibernate.search.engine.spi.DocumentBuilderIndexedEntity;
import org.hibernate.search.engine.spi.EntityIndexBinding;
import org.hibernate.search.exception.SearchException;
import org.hibernate.search.filter.FullTextFilter;
import org.hibernate.search.indexes.spi.IndexManager;
import org.hibernate.search.metadata.NumericFieldSettingsDescriptor.NumericEncodingType;
import org.hibernate.search.query.engine.impl.AbstractHSQuery;
import org.hibernate.search.query.engine.impl.EntityInfoImpl;
import org.hibernate.search.query.engine.impl.TimeoutManagerImpl;
import org.hibernate.search.query.engine.spi.DocumentExtractor;
import org.hibernate.search.query.engine.spi.EntityInfo;
import org.hibernate.search.query.engine.spi.FacetManager;
import org.hibernate.search.query.engine.spi.HSQuery;
import org.hibernate.search.spatial.Coordinates;
import org.hibernate.search.util.logging.impl.LoggerFactory;
import com.google.gson.JsonArray;
import com.google.gson.JsonElement;
import com.google.gson.JsonObject;
import com.google.gson.JsonPrimitive;
import io.searchbox.core.DocumentResult;
import io.searchbox.core.Explain;
import io.searchbox.core.Search;
import io.searchbox.core.SearchResult;
import io.searchbox.core.search.sort.Sort;
import io.searchbox.core.search.sort.Sort.Sorting;
/**
* Query implementation based on Elasticsearch.
*
* @author Gunnar Morling
*/
public class ElasticsearchHSQueryImpl extends AbstractHSQuery {
private static final Log LOG = LoggerFactory.make( Log.class );
private static final Pattern DOT = Pattern.compile( "\\." );
/**
* ES default limit for (firstResult + maxResult)
*/
private static final int MAX_RESULT_WINDOW_SIZE = 10000;
private final String jsonQuery;
private Integer resultSize;
private IndexSearcher searcher;
private SearchResult searchResult;
public ElasticsearchHSQueryImpl(String jsonQuery, ExtendedSearchIntegrator extendedIntegrator) {
super( extendedIntegrator );
this.jsonQuery = jsonQuery;
}
@Override
public HSQuery luceneQuery(Query query) {
throw new UnsupportedOperationException( "Cannot use Lucene query with Elasticsearch" );
}
@Override
public FacetManager getFacetManager() {
// TODO implement
throw new UnsupportedOperationException( "Not yet implemented" );
}
@Override
public Query getLuceneQuery() {
throw new UnsupportedOperationException( "Cannot use Lucene query with Elasticsearch" );
}
@Override
public DocumentExtractor queryDocumentExtractor() {
return new ElasticsearchDocumentExtractor();
}
@Override
public int queryResultSize() {
if ( searchResult == null ) {
execute();
}
return resultSize;
}
@Override
public Explanation explain(int documentId) {
if ( searchResult == null ) {
execute();
}
JsonObject hit = searchResult.getJsonObject()
.get( "hits" )
.getAsJsonObject()
.get( "hits" )
.getAsJsonArray()
// TODO Is it right to use the document id that way? I am not quite clear about its semantics
.get( documentId )
.getAsJsonObject();
try ( ServiceReference client = getExtendedSearchIntegrator().getServiceManager().requestReference( JestClient.class ) ) {
Explain request = new Explain.Builder(
hit.get( "_index" ).getAsString(),
hit.get( "_type" ).getAsString(),
hit.get( "_id" ).getAsString(),
searcher.executedQuery
)
.build();
DocumentResult response = client.get().executeRequest( request );
JsonObject explanation = response.getJsonObject().get( "explanation" ).getAsJsonObject();
return convertExplanation( explanation );
}
}
private Explanation convertExplanation(JsonObject explanation) {
float value = explanation.get( "value" ).getAsFloat();
String description = explanation.get( "description" ).getAsString();
JsonElement explanationDetails = explanation.get( "details" );
List details;
if ( explanationDetails != null ) {
details = new ArrayList<>( explanationDetails.getAsJsonArray().size() );
for ( JsonElement detail : explanationDetails.getAsJsonArray() ) {
details.add( convertExplanation( detail.getAsJsonObject() ) );
}
}
else {
details = Collections.emptyList();
}
return Explanation.match( value, description, details );
}
@Override
public HSQuery setSpatialParameters(Coordinates center, String fieldName) {
// TODO implement
throw new UnsupportedOperationException( "Not yet implemented" );
}
@Override
public HSQuery filter(Filter filter) {
// TODO implement
throw new UnsupportedOperationException( "Not yet implemented" );
}
@Override
public FullTextFilter enableFullTextFilter(String name) {
// TODO implement
throw new UnsupportedOperationException( "Not yet implemented" );
}
@Override
public void disableFullTextFilter(String name) {
// TODO implement
throw new UnsupportedOperationException( "Not yet implemented" );
}
@Override
protected void clearCachedResults() {
resultSize = null;
}
@Override
protected TimeoutManagerImpl buildTimeoutManager() {
return new TimeoutManagerImpl(
jsonQuery,
timeoutExceptionFactory,
this.extendedIntegrator.getTimingSource()
);
}
@Override
public List queryEntityInfos() {
if ( searchResult == null ) {
execute();
}
List results = new ArrayList<>( searchResult.getTotal() );
JsonArray hits = searchResult.getJsonObject().get( "hits" ).getAsJsonObject().get( "hits" ).getAsJsonArray();
for ( JsonElement hit : hits ) {
EntityInfo entityInfo = searcher.convertQueryHit( hit.getAsJsonObject() );
if ( entityInfo != null ) {
results.add( entityInfo );
}
}
return results;
}
private void execute() {
searcher = new IndexSearcher();
searchResult = searcher.runSearch();
resultSize = searchResult.getTotal();
}
private Object getId(JsonObject hit, EntityIndexBinding binding) {
Document tmp = new Document();
tmp.add( new StringField( "id", DocumentIdHelper.getEntityId( hit.get( "_id" ).getAsString() ), Store.NO) );
Object id = binding.getDocumentBuilder().getIdBridge().get( "id", tmp );
return id;
}
/**
* Determines the affected indexes and runs the given query against them.
*/
private class IndexSearcher {
private final Search search;
private final Map> entityTypesByName;
private final String executedQuery;
private IndexSearcher() {
entityTypesByName = new HashMap<>();
String idFieldName = null;
JsonArray typeFilters = new JsonArray();
Set indexNames = new HashSet<>();
for ( Class queriedEntityType : getQueriedEntityTypes() ) {
entityTypesByName.put( queriedEntityType.getName(), queriedEntityType );
EntityIndexBinding binding = extendedIntegrator.getIndexBinding( queriedEntityType );
IndexManager[] indexManagers = binding.getIndexManagers();
for ( IndexManager indexManager : indexManagers ) {
if ( !( indexManager instanceof ElasticsearchIndexManager ) ) {
throw LOG.cannotRunEsQueryTargetingEntityIndexedWithNonEsIndexManager(
queriedEntityType,
jsonQuery
);
}
// TODO will this be a problem when querying multiple entity types, with one using a field as id
// field and the other not; is that possible?
idFieldName = binding.getDocumentBuilder().getIdentifierName();
ElasticsearchIndexManager esIndexManager = (ElasticsearchIndexManager) indexManager;
indexNames.add( esIndexManager.getActualIndexName() );
}
typeFilters.add( getEntityTypeFilter( queriedEntityType ) );
}
// Query filters; always a type filter, possibly a tenant id filter;
// TODO feed in user-provided filters
JsonObject effectiveFilter = getEffectiveFilter( typeFilters );
// TODO can we avoid the forth and back between GSON and String?
executedQuery = "{ \"query\" : { \"filtered\" : { " + jsonQuery.substring( 1, jsonQuery.length() - 1 ) + ", \"filter\" : " + effectiveFilter.toString() + " } } }";
Search.Builder search = new Search.Builder( executedQuery );
search.addIndex( indexNames );
search.setParameter( "from", firstResult );
// If the user has given a value, take it as is, let ES itself complain if it's too high; if no value is
// given, I take as much as possible, as by default only 10 rows would be returned
search.setParameter( "size", maxResults != null ? maxResults : MAX_RESULT_WINDOW_SIZE - firstResult );
// TODO: embedded fields
if ( sort != null ) {
for ( SortField sortField : sort.getSort() ) {
String sortFieldName = sortField.getField().equals( idFieldName ) ? "_uid" : sortField.getField();
search.addSort( new Sort( sortFieldName, sortField.getReverse() ? Sorting.DESC : Sorting.ASC ) );
}
}
this.search = search.build();
}
private JsonObject getEffectiveFilter(JsonArray typeFilters) {
JsonArray filters = new JsonArray();
JsonObject tenantFilter = getTenantIdFilter();
if ( tenantFilter != null ) {
filters.add( tenantFilter );
}
// wrap type filters into should if there is more than one
JsonObject effectiveTypeFilter = new JsonObject();
if ( typeFilters.size() == 1 ) {
effectiveTypeFilter = typeFilters.get( 0 ).getAsJsonObject();
}
else {
JsonObject should = new JsonObject();
should.add( "should", typeFilters );
effectiveTypeFilter = new JsonObject();
effectiveTypeFilter.add( "bool", should );
}
filters.add( effectiveTypeFilter );
// wrap filters into must if there is more than one
JsonObject effectiveFilter = new JsonObject();
if ( filters.size() == 1 ) {
effectiveFilter = filters.get( 0 ).getAsJsonObject();
}
else {
JsonObject must = new JsonObject();
must.add( "must", filters );
effectiveFilter = new JsonObject();
effectiveFilter.add( "bool", must );
}
return effectiveFilter;
}
private JsonObject getEntityTypeFilter(Class queriedEntityType) {
JsonObject value = new JsonObject();
value.addProperty( "value", queriedEntityType.getName() );
JsonObject type = new JsonObject();
type.add( "type", value );
return type;
}
private JsonObject getTenantIdFilter() {
if ( tenantId == null ) {
return null;
}
JsonObject value = new JsonObject();
value.addProperty( DocumentBuilderIndexedEntity.TENANT_ID_FIELDNAME, tenantId );
JsonObject tenantFilter = new JsonObject();
tenantFilter.add( "term", value );
return tenantFilter;
}
private Iterable> getQueriedEntityTypes() {
if ( indexedTargetedEntities == null || indexedTargetedEntities.isEmpty() ) {
return extendedIntegrator.getIndexBindings().keySet();
}
else {
return indexedTargetedEntities;
}
}
SearchResult runSearch() {
try ( ServiceReference client = getExtendedSearchIntegrator().getServiceManager().requestReference( JestClient.class ) ) {
return client.get().executeRequest( search );
}
}
EntityInfo convertQueryHit(JsonObject hit) {
String type = hit.get( "_type" ).getAsString();
Class clazz = entityTypesByName.get( type );
if ( clazz == null ) {
LOG.warnf( "Found unknown type in Elasticsearch index: " + type );
return null;
}
EntityIndexBinding binding = extendedIntegrator.getIndexBinding( clazz );
Object id = getId( hit, binding );
Object[] projections = null;
List indexesOfThis = null;
if ( projectedFields != null ) {
projections = new Object[projectedFields.length];
indexesOfThis = new ArrayList<>();
int i = 0;
for ( String field : projectedFields ) {
switch ( field ) {
case ProjectionConstants.SOURCE:
projections[i] = hit.getAsJsonObject().get( "_source" ).toString();
break;
case ProjectionConstants.DOCUMENT:
throw new IllegalArgumentException( "Projection of Lucene document not supported with Elasticsearch backend" );
case DOCUMENT_ID:
throw new IllegalArgumentException( "Projection of Lucene document id not supported with Elasticsearch backend" );
case ProjectionConstants.EXPLANATION:
throw new UnsupportedOperationException( "Not yet implemented" );
case ProjectionConstants.ID:
projections[i] = id;
break;
case ProjectionConstants.OBJECT_CLASS:
projections[i] = clazz;
break;
case ProjectionConstants.SCORE:
projections[i] = hit.getAsJsonObject().get( "_score" ).getAsFloat();
break;
case ProjectionConstants.SPATIAL_DISTANCE:
throw new UnsupportedOperationException( "Not yet implemented" );
case ProjectionConstants.THIS:
indexesOfThis.add( i );
break;
default:
projections[i] = getFieldValue( binding, hit, field );
}
i++;
}
}
EntityInfoImpl entityInfo = new EntityInfoImpl( clazz, binding.getDocumentBuilder().getIdentifierName(), (Serializable) id, projections );
if ( indexesOfThis != null ) {
entityInfo.getIndexesOfThis().addAll( indexesOfThis );
}
return entityInfo;
}
/**
* Returns the value of the given field as retrieved from the ES result and converted using the corresponding
* field bridge. In case this bridge is not a 2-way bridge, the unconverted value will be returned.
*/
private Object getFieldValue(EntityIndexBinding binding, JsonObject hit, String projectedField) {
DocumentFieldMetadata field = FieldHelper.getFieldMetadata( binding, projectedField );
if ( field == null ) {
throw new IllegalArgumentException( "Unknown field " + projectedField + " for entity "
+ binding.getDocumentBuilder().getMetadata().getType().getName() );
}
JsonElement value;
if ( field.isId() ) {
value = hit.get( "_id" );
}
else {
value = getFieldValue( hit.get( "_source" ).getAsJsonObject(), projectedField );
}
if ( value == null || value.isJsonNull() ) {
return null;
}
return convertFieldValue( binding, field, value );
}
private Object convertFieldValue(EntityIndexBinding binding, DocumentFieldMetadata field, JsonElement value) {
FieldBridge fieldBridge = field.getFieldBridge();
if ( FieldHelper.isBoolean( binding, field.getName() ) ) {
return value.getAsBoolean();
}
else if ( fieldBridge instanceof TwoWayFieldBridge ) {
Document tmp = new Document();
if ( FieldHelper.isNumeric( field ) ) {
NumericEncodingType numericEncodingType = FieldHelper.getNumericEncodingType( binding, field );
switch( numericEncodingType ) {
case INTEGER:
tmp.add( new IntField( field.getName(), value.getAsInt(), Store.NO ) );
break;
case LONG:
tmp.add( new LongField( field.getName(), value.getAsLong(), Store.NO ) );
break;
case FLOAT:
tmp.add( new FloatField( field.getName(), value.getAsFloat(), Store.NO ) );
break;
case DOUBLE:
tmp.add( new DoubleField( field.getName(), value.getAsDouble(), Store.NO ) );
break;
default:
throw new SearchException( "Unexpected numeric field type: " + binding.getDocumentBuilder().getMetadata().getType() + " "
+ field.getName() );
}
}
else {
tmp.add( new StringField( field.getName(), value.getAsString(), Store.NO ) );
}
return ( (TwoWayFieldBridge) fieldBridge ).get( field.getName(), tmp );
}
// Should only be the case for custom bridges
else {
JsonPrimitive primitive = value.getAsJsonPrimitive();
if ( primitive.isBoolean() ) {
return primitive.getAsBoolean();
}
else if ( primitive.isNumber() ) {
// TODO this will expose a Gson-specific Number implementation; Can we somehow return an Integer,
// Long... etc. instead?
return primitive.getAsNumber();
}
else if ( primitive.isString() ) {
return primitive.getAsString();
}
else {
// TODO Better raise an exception?
return primitive.toString();
}
}
}
private JsonElement getFieldValue(JsonObject parent, String projectedField) {
String field = projectedField;
if ( projectedField.contains( "." ) ) {
String[] parts = DOT.split( projectedField );
field = parts[parts.length - 1];
for ( int i = 0; i < parts.length - 1; i++ ) {
JsonElement newParent = parent.get( parts[i] );
if ( newParent == null ) {
return null;
}
parent = newParent.getAsJsonObject();
}
}
return parent.getAsJsonObject().get( field );
}
}
// TODO: Investigate scrolling API:
// https://www.elastic.co/guide/en/elasticsearch/reference/current/search-request-scroll.html
private class ElasticsearchDocumentExtractor implements DocumentExtractor {
private final IndexSearcher searcher;
private List results;
private ElasticsearchDocumentExtractor() {
searcher = new IndexSearcher();
}
@Override
public EntityInfo extract(int index) throws IOException {
if ( results == null ) {
runSearch();
}
return results.get( index );
}
@Override
public int getFirstIndex() {
return 0;
}
@Override
public int getMaxIndex() {
if ( results == null ) {
runSearch();
}
return results.size() - 1;
}
@Override
public void close() {
}
@Override
public TopDocs getTopDocs() {
throw new UnsupportedOperationException( "TopDocs not available with Elasticsearch backend" );
}
private void runSearch() {
SearchResult searchResult = searcher.runSearch();
JsonArray hits = searchResult.getJsonObject().get( "hits" ).getAsJsonObject().get( "hits" ).getAsJsonArray();
results = new ArrayList<>( searchResult.getTotal() );
for ( JsonElement hit : hits ) {
EntityInfo converted = searcher.convertQueryHit( hit.getAsJsonObject() );
if ( converted != null ) {
results.add( converted );
}
}
}
}
}