litvinovg/vitro
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

VIVO-742 Fix an issue with SearchInputDocument.addField()

Browse source 
Behavior should be that if a field with the same name already exists on the document, then the new values are added, and the new boost is factored in.
This commit is contained in:
Jim Blake 2014-04-24 16:49:58 -04:00
parent 630fc3c1f0
commit b642f30391
4 changed files with 50 additions and 41 deletions
Download patch file Download diff file Expand all files Collapse all files

32
webapp/src/edu/cornell/mannlib/vitro/webapp/modules/searchEngine/SearchInputDocument.java
View file

@ -22,33 +22,41 @@ public interface SearchInputDocument {
void addField(SearchInputField field); void addField(SearchInputField field);
/** /**
* Create a field with this name and values, and put it into the document. If * Create a field with this name and values, and put it into the document.
* a field with this name already exists in the document, it will be *
* replaced. * If a field with this name already exists in the document, these values
* will be added to the existing values on the field.
*/ */
void addField(String name, Object... values); void addField(String name, Object... values);
/** /**
* Create a field with this name and values, and put it into the document. If * Create a field with this name and values, and put it into the document.
* a field with this name already exists in the document, it will be *
* replaced. * If a field with this name already exists in the document, these values
* will be added to the existing values on the field.
*/ */
void addField(String name, Collection<Object> values); void addField(String name, Collection<Object> values);
/** /**
* Create a field with this name, boost level and values, and put it into * Create a field with this name, boost level and values, and put it into
* the document. If a field with this name already exists in the document, * the document.
* it will be replaced. *
* If a field with this name already exists in the document,
* these values will be added to the existing values on the field, and the
* existing boost will be multipled by this boost.
*/ */
void addField(String name, float boost, Object... values); void addField(String name, float boost, Object... values);
/** /**
* Create a field with this name, boost level and values, and put it into * Create a field with this name, boost level and values, and put it into
* the document. If a field with this name already exists in the document, * the document.
* it will be replaced. *
* If a field with this name already exists in the document,
* these values will be added to the existing values on the field, and the
* existing boost will be multipled by this boost.
*/ */
void addField(String name, float boost, Collection<Object> values); void addField(String name, float boost, Collection<Object> values);
/** /**
* Set a boost level for the document as a whole. * Set a boost level for the document as a whole.
*/ */

8
webapp/src/edu/cornell/mannlib/vitro/webapp/search/documentBuilding/ContextNodeFields.java
View file

@ -62,13 +62,7 @@ public class ContextNodeFields implements DocumentModifier{
} }
/* get text from the context nodes and add the to ALLTEXT */ /* get text from the context nodes and add the to ALLTEXT */
StringBuffer values = executeQueryForValues(individual, queries); StringBuffer values = executeQueryForValues(individual, queries);
doc.addField(VitroSearchTermNames.ALLTEXT, values);
SearchInputField field = doc.getField(VitroSearchTermNames.ALLTEXT);
if( field == null ){
doc.addField(VitroSearchTermNames.ALLTEXT, values);
}else{
field.addValues(values, field.getBoost());
}
} }
/** /**

42
webapp/src/edu/cornell/mannlib/vitro/webapp/search/documentBuilding/IndividualToSearchDocument.java
View file

@ -2,6 +2,16 @@
/* $This file is distributed under the terms of the license in /doc/license.txt$ */ /* $This file is distributed under the terms of the license in /doc/license.txt$ */
package edu.cornell.mannlib.vitro.webapp.search.documentBuilding; package edu.cornell.mannlib.vitro.webapp.search.documentBuilding;
import static edu.cornell.mannlib.vitro.webapp.search.VitroSearchTermNames.ALLTEXT;
import static edu.cornell.mannlib.vitro.webapp.search.VitroSearchTermNames.ALLTEXTUNSTEMMED;
import static edu.cornell.mannlib.vitro.webapp.search.VitroSearchTermNames.CLASSGROUP_URI;
import static edu.cornell.mannlib.vitro.webapp.search.VitroSearchTermNames.DOCID;
import static edu.cornell.mannlib.vitro.webapp.search.VitroSearchTermNames.INDEXEDTIME;
import static edu.cornell.mannlib.vitro.webapp.search.VitroSearchTermNames.MOST_SPECIFIC_TYPE_URIS;
import static edu.cornell.mannlib.vitro.webapp.search.VitroSearchTermNames.NAME_LOWERCASE_SINGLE_VALUED;
import static edu.cornell.mannlib.vitro.webapp.search.VitroSearchTermNames.NAME_RAW;
import static edu.cornell.mannlib.vitro.webapp.search.VitroSearchTermNames.RDFTYPE;
import static edu.cornell.mannlib.vitro.webapp.search.VitroSearchTermNames.URI;
import java.util.ArrayList; import java.util.ArrayList;
import java.util.HashMap; import java.util.HashMap;
@ -27,14 +37,11 @@ import edu.cornell.mannlib.vitro.webapp.beans.VClass;
import edu.cornell.mannlib.vitro.webapp.modules.searchEngine.SearchInputDocument; import edu.cornell.mannlib.vitro.webapp.modules.searchEngine.SearchInputDocument;
import edu.cornell.mannlib.vitro.webapp.modules.searchEngine.SearchResultDocument; import edu.cornell.mannlib.vitro.webapp.modules.searchEngine.SearchResultDocument;
import edu.cornell.mannlib.vitro.webapp.search.IndexingException; import edu.cornell.mannlib.vitro.webapp.search.IndexingException;
import edu.cornell.mannlib.vitro.webapp.search.VitroSearchTermNames;
public class IndividualToSearchDocument { public class IndividualToSearchDocument {
public static final Log log = LogFactory.getLog(IndividualToSearchDocument.class.getName()); public static final Log log = LogFactory.getLog(IndividualToSearchDocument.class.getName());
public static VitroSearchTermNames term = new VitroSearchTermNames();
protected final String label = "http://www.w3.org/2000/01/rdf-schema#label"; protected final String label = "http://www.w3.org/2000/01/rdf-schema#label";
protected List<DocumentModifier> documentModifiers = new ArrayList<DocumentModifier>(); protected List<DocumentModifier> documentModifiers = new ArrayList<DocumentModifier>();
@ -46,7 +53,6 @@ public class IndividualToSearchDocument {
this.documentModifiers = docModifiers; this.documentModifiers = docModifiers;
} }
@SuppressWarnings("static-access")
public SearchInputDocument translate(Individual ind) throws IndexingException{ public SearchInputDocument translate(Individual ind) throws IndexingException{
try{ try{
String excludeMsg = checkExcludes( ind ); String excludeMsg = checkExcludes( ind );
@ -58,10 +64,10 @@ public class IndividualToSearchDocument {
SearchInputDocument doc = ApplicationUtils.instance().getSearchEngine().createInputDocument(); SearchInputDocument doc = ApplicationUtils.instance().getSearchEngine().createInputDocument();
//DocID //DocID
doc.addField(term.DOCID, getIdForUri( ind.getURI() ) ); doc.addField(DOCID, getIdForUri( ind.getURI() ) );
//vitro id //vitro id
doc.addField(term.URI, ind.getURI()); doc.addField(URI, ind.getURI());
log.debug(ind.getURI() + " init boost: " + doc.getDocumentBoost()); log.debug(ind.getURI() + " init boost: " + doc.getDocumentBoost());
//get label from ind //get label from ind
@ -80,7 +86,7 @@ public class IndividualToSearchDocument {
addObjectPropertyText(ind, doc, objectNames, addUri); addObjectPropertyText(ind, doc, objectNames, addUri);
//time of index in msec past epoch //time of index in msec past epoch
doc.addField(term.INDEXEDTIME, (Object) new DateTime().getMillis() ); doc.addField(INDEXEDTIME, (Object) new DateTime().getMillis() );
addAllText( ind, doc, classPublicNames, objectNames ); addAllText( ind, doc, classPublicNames, objectNames );
@ -140,7 +146,7 @@ public class IndividualToSearchDocument {
if( log.isDebugEnabled()){ if( log.isDebugEnabled()){
long delta = System.currentTimeMillis() - start; long delta = System.currentTimeMillis() - start;
synchronized(docModClassToTime){ synchronized(docModClassToTime){
Class clz = modifier.getClass(); Class<?> clz = modifier.getClass();
if( docModClassToTime.containsKey( clz.getName() )){ if( docModClassToTime.containsKey( clz.getName() )){
Long time = docModClassToTime.get(clz.getName() ); Long time = docModClassToTime.get(clz.getName() );
docModClassToTime.put(clz.getName(), time + delta); docModClassToTime.put(clz.getName(), time + delta);
@ -198,8 +204,8 @@ public class IndividualToSearchDocument {
String alltext = allTextValue.toString(); String alltext = allTextValue.toString();
doc.addField(term.ALLTEXT, alltext); doc.addField(ALLTEXT, alltext);
doc.addField(term.ALLTEXTUNSTEMMED, alltext); doc.addField(ALLTEXTUNSTEMMED, alltext);
} }
@ -247,8 +253,6 @@ public class IndividualToSearchDocument {
* @throws SkipIndividualException * @throws SkipIndividualException
*/ */
protected void addClasses(Individual ind, SearchInputDocument doc, StringBuffer classPublicNames) throws SkipIndividualException{ protected void addClasses(Individual ind, SearchInputDocument doc, StringBuffer classPublicNames) throws SkipIndividualException{
ArrayList<String> superClassNames = null;
List<VClass> vclasses = ind.getVClasses(false); List<VClass> vclasses = ind.getVClasses(false);
if( vclasses == null || vclasses.isEmpty() ){ if( vclasses == null || vclasses.isEmpty() ){
throw new SkipIndividualException("Not indexing because individual has no classes"); throw new SkipIndividualException("Not indexing because individual has no classes");
@ -265,7 +269,7 @@ public class IndividualToSearchDocument {
doc.setDocumentBoost(doc.getDocumentBoost() + clz.getSearchBoost()); doc.setDocumentBoost(doc.getDocumentBoost() + clz.getSearchBoost());
} }
doc.addField(term.RDFTYPE, clz.getURI()); doc.addField(RDFTYPE, clz.getURI());
if(clz.getName() != null){ if(clz.getName() != null){
classPublicNames.append(" "); classPublicNames.append(" ");
@ -274,7 +278,7 @@ public class IndividualToSearchDocument {
//Add the Classgroup URI to a field //Add the Classgroup URI to a field
if(clz.getGroupURI() != null){ if(clz.getGroupURI() != null){
doc.addField(term.CLASSGROUP_URI,clz.getGroupURI()); doc.addField(CLASSGROUP_URI,clz.getGroupURI());
} }
} }
} }
@ -285,7 +289,7 @@ public class IndividualToSearchDocument {
if( mstURIs != null ){ if( mstURIs != null ){
for( String typeURI : mstURIs ){ for( String typeURI : mstURIs ){
if( typeURI != null && ! typeURI.trim().isEmpty() ) if( typeURI != null && ! typeURI.trim().isEmpty() )
doc.addField(term.MOST_SPECIFIC_TYPE_URIS, typeURI); doc.addField(MOST_SPECIFIC_TYPE_URIS, typeURI);
} }
} }
} }
@ -299,8 +303,8 @@ public class IndividualToSearchDocument {
value = ind.getLocalName(); value = ind.getLocalName();
} }
doc.addField(term.NAME_RAW, value); doc.addField(NAME_RAW, value);
doc.addField(term.NAME_LOWERCASE_SINGLE_VALUED,value); doc.addField(NAME_LOWERCASE_SINGLE_VALUED,value);
// NAME_RAW will be copied by the search engine into the following fields: // NAME_RAW will be copied by the search engine into the following fields:
// NAME_LOWERCASE, NAME_UNSTEMMED, NAME_STEMMED, NAME_PHONETIC, AC_NAME_UNTOKENIZED, AC_NAME_STEMMED // NAME_LOWERCASE, NAME_UNSTEMMED, NAME_STEMMED, NAME_PHONETIC, AC_NAME_UNTOKENIZED, AC_NAME_STEMMED
@ -319,7 +323,7 @@ public class IndividualToSearchDocument {
} }
public String getQueryForId(String uri ){ public String getQueryForId(String uri ){
return term.DOCID + ':' + getIdForUri(uri); return DOCID + ':' + getIdForUri(uri);
} }
public Individual unTranslate(Object result) { public Individual unTranslate(Object result) {
@ -327,7 +331,7 @@ public class IndividualToSearchDocument {
if( result instanceof SearchResultDocument){ if( result instanceof SearchResultDocument){
SearchResultDocument hit = (SearchResultDocument) result; SearchResultDocument hit = (SearchResultDocument) result;
String uri= (String) hit.getFirstValue(term.URI); String uri= (String) hit.getFirstValue(URI);
ent = new IndividualImpl(); ent = new IndividualImpl();
ent.setURI(uri); ent.setURI(uri);

9
webapp/src/edu/cornell/mannlib/vitro/webapp/searchengine/base/BaseSearchInputDocument.java
View file

@ -39,10 +39,13 @@ public class BaseSearchInputDocument implements SearchInputDocument {
@Override @Override
public void addField(String name, float boost, Collection<Object> values) { public void addField(String name, float boost, Collection<Object> values) {
BaseSearchInputField field = new BaseSearchInputField(name); SearchInputField field = fieldMap.get(name);
field.setBoost(boost); if (field == null) {
field = new BaseSearchInputField(name);
fieldMap.put(name, field);
}
field.addValues(values); field.addValues(values);
fieldMap.put(name, field); field.setBoost(boost * field.getBoost());
} }
@Override @Override