SOLR-1726: add deep paging support

git-svn-id: https://svn.apache.org/repos/asf/lucene/dev/trunk@1239181 13f79535-47bb-0310-9956-ffa450edef68
This commit is contained in:
Grant Ingersoll 2012-02-01 16:01:34 +00:00
parent 5d2de95e35
commit b912f6da98
7 changed files with 117 additions and 6 deletions

View File

@ -199,6 +199,8 @@ New Features
* SOLR-3069: Ability to add openSearcher=false to not open a searcher when doing
a hard commit. commitWithin now only invokes a softCommit. (yonik)
* SOLR-1726: Added deep paging support to search (sort by score only) which should use less memory when paging deeply into results
by keeping the priority queue small. (Manojkumar Rangasamy Kannadasan, gsingers)
Optimizations
----------------------

View File

@ -128,7 +128,8 @@ public class QueryComponent extends SearchComponent
rb.setQuery( q );
rb.setSortSpec( parser.getSort(true) );
rb.setQparser(parser);
rb.setScoreDoc(parser.getPaging());
String[] fqs = req.getParams().getParams(CommonParams.FQ);
if (fqs!=null && fqs.length!=0) {
List<Query> filters = rb.getFilters();

View File

@ -18,6 +18,7 @@
package org.apache.solr.handler.component;
import org.apache.lucene.search.Query;
import org.apache.lucene.search.ScoreDoc;
import org.apache.lucene.search.TopDocs;
import org.apache.lucene.search.grouping.SearchGroup;
import org.apache.lucene.search.grouping.TopGroups;
@ -69,6 +70,9 @@ public class ResponseBuilder
private List<Query> filters = null;
private SortSpec sortSpec = null;
private GroupingSpecification groupingSpec;
//used for handling deep paging
private ScoreDoc scoreDoc;
private DocListAndSet results = null;
private NamedList<Object> debugInfo = null;
@ -377,7 +381,8 @@ public class ResponseBuilder
.setOffset(getSortSpec().getOffset())
.setLen(getSortSpec().getCount())
.setFlags(getFieldFlags())
.setNeedDocSet(isNeedDocSet());
.setNeedDocSet(isNeedDocSet())
.setScoreDoc(getScoreDoc()); //Issue 1726
return cmd;
}
@ -390,4 +395,14 @@ public class ResponseBuilder
rsp.getResponseHeader().add("partialResults", Boolean.TRUE);
}
}
public ScoreDoc getScoreDoc()
{
return scoreDoc;
}
public void setScoreDoc(ScoreDoc scoreDoc)
{
this.scoreDoc = scoreDoc;
}
}

View File

@ -18,6 +18,7 @@ package org.apache.solr.search;
import org.apache.lucene.queryparser.classic.ParseException;
import org.apache.lucene.search.Query;
import org.apache.lucene.search.ScoreDoc; //Issue 1726
import org.apache.lucene.search.Sort;
import org.apache.solr.common.params.CommonParams;
import org.apache.solr.common.params.MapSolrParams;
@ -209,7 +210,31 @@ public abstract class QParser {
return nestedParser;
}
/**
* use common params to look up pageScore and pageDoc in global params
* @return the ScoreDoc
*/
public ScoreDoc getPaging() throws ParseException
{
String pageScoreS = null;
String pageDocS = null;
pageScoreS = params.get(CommonParams.PAGESCORE);
pageDocS = params.get(CommonParams.PAGEDOC);
if (pageScoreS == null || pageDocS == null)
return null;
int pageDoc = pageDocS != null ? Integer.parseInt(pageDocS) : -1;
float pageScore = pageScoreS != null ? new Float(pageScoreS) : -1;
if(pageDoc != -1 && pageScore != -1){
return new ScoreDoc(pageDoc, pageScore);
}
else {
return null;
}
}
/**
* @param useGlobalParams look up sort, start, rows in global params if not in local params
* @return the sort specification

View File

@ -1358,7 +1358,12 @@ public class SolrIndexSearcher extends IndexSearcher implements Closeable,SolrIn
} else {
TopDocsCollector topCollector;
if (cmd.getSort() == null) {
topCollector = TopScoreDocCollector.create(len, true);
if(cmd.getScoreDoc() != null) {
topCollector = TopScoreDocCollector.create(len, cmd.getScoreDoc(), true); //create the Collector with InOrderPagingCollector
} else {
topCollector = TopScoreDocCollector.create(len, true);
}
} else {
topCollector = TopFieldCollector.create(weightSort(cmd.getSort()), len, false, needScores, needScores, true);
}
@ -1382,7 +1387,6 @@ public class SolrIndexSearcher extends IndexSearcher implements Closeable,SolrIn
TopDocs topDocs = topCollector.topDocs(0, len);
maxScore = totalHits>0 ? topDocs.getMaxScore() : 0.0f;
nDocsReturned = topDocs.scoreDocs.length;
ids = new int[nDocsReturned];
scores = (cmd.getFlags()&GET_SCORES)!=0 ? new float[nDocsReturned] : null;
for (int i=0; i<nDocsReturned; i++) {
@ -1392,7 +1396,6 @@ public class SolrIndexSearcher extends IndexSearcher implements Closeable,SolrIn
}
}
int sliceLen = Math.min(lastDocRequested,nDocsReturned);
if (sliceLen < 0) sliceLen=0;
qr.setDocList(new DocSlice(0,sliceLen,ids,scores,totalHits,maxScore));
@ -2011,6 +2014,18 @@ public class SolrIndexSearcher extends IndexSearcher implements Closeable,SolrIn
private int supersetMaxDoc;
private int flags;
private long timeAllowed = -1;
//Issue 1726 start
private ScoreDoc scoreDoc;
public ScoreDoc getScoreDoc()
{
return scoreDoc;
}
public void setScoreDoc(ScoreDoc scoreDoc)
{
this.scoreDoc = scoreDoc;
}
//Issue 1726 end
// public List<Grouping.Command> groupCommands;

View File

@ -33,6 +33,7 @@ import org.apache.lucene.document.Field;
import org.apache.lucene.index.IndexWriter;
import org.apache.lucene.index.IndexableField;
import org.apache.lucene.index.LogMergePolicy;
import org.apache.lucene.util.English;
import org.apache.solr.common.SolrException;
import org.apache.solr.common.params.CommonParams;
import org.apache.solr.common.params.MapSolrParams;
@ -761,6 +762,50 @@ public class BasicFunctionalityTest extends SolrTestCaseJ4 {
}
@Test
public void testDeepPaging() throws Exception {
for (int i = 0; i < 1000; i++){
assertU(adoc("id", String.valueOf(i), "foo_t", English.intToEnglish(i)));
}
assertU(commit());
SolrQueryRequest goldReq = null;
try {
goldReq = req("q", "foo_t:one", "rows", "50", "fl", "docid, score");
SolrQueryResponse gold = h.queryAndResponse("standard", goldReq);
ResultContext response = (ResultContext) gold.getValues().get("response");
assertQ("page: " + 0 + " failed",
req("q", "foo_t:one", "rows", "10", CommonParams.QT, "standard", "fl", "[docid], score"),
"*[count(//doc)=10]");
//ugh, what a painful way to get the document
DocIterator iterator = response.docs.subset(9, 1).iterator();
int lastDoc = iterator.nextDoc();
float lastScore = iterator.score();
for (int i = 1; i < 5; i++){
//page through some results
DocList subset = response.docs.subset(i * 10, 1);
iterator = subset.iterator();
int compareDoc = iterator.nextDoc();
float compareScore = iterator.score();
assertQ("page: " + i + " failed",
req("q", "foo_t:one", CommonParams.QT, "standard", "fl", "[docid], score",
"start", String.valueOf(i * 10), "rows", "1", //only get one doc, and then compare it to gold
CommonParams.PAGEDOC, String.valueOf(lastDoc), CommonParams.PAGESCORE, String.valueOf(lastScore)),
"*[count(//doc)=1]",
"//float[@name='score'][.='" + compareScore + "']",
"//int[@name='[docid]'][.='" + compareDoc + "']"
);
lastScore = compareScore;
lastDoc = compareDoc;
}
} finally {
if (goldReq != null ) {
goldReq.close();
}
}
}
// /** this doesn't work, but if it did, this is how we'd test it. */
// public void testOverwriteFalse() {

View File

@ -46,6 +46,14 @@ public interface CommonParams {
/** number of documents to return starting at "start" */
public static final String ROWS ="rows";
//Issue 1726 start
/** score of the last document of the previous page */
public static final String PAGESCORE ="pageScore";
/** docid of the last document of the previous page */
public static final String PAGEDOC ="pageDoc";
//Issue 1726 end
/** stylesheet to apply to XML results */
public static final String XSL ="xsl";