Examples with ScoreDoc org.apache.lucene.search.ScoreDoc used on opensource projects

Example 6 with ScoreDoc

use of org.apache.lucene.search.ScoreDoc in project elasticsearch by elastic.

the class CollapseTopFieldDocs method merge.

/**
     * Returns a new CollapseTopDocs, containing topN collapsed results across
     * the provided CollapseTopDocs, sorting by score. Each {@link CollapseTopFieldDocs} instance must be sorted.
     **/
public static CollapseTopFieldDocs merge(Sort sort, int start, int size, CollapseTopFieldDocs[] shardHits) throws IOException {
    String collapseField = shardHits[0].field;
    for (int i = 1; i < shardHits.length; i++) {
        if (collapseField.equals(shardHits[i].field) == false) {
            throw new IllegalArgumentException("collapse field differ across shards [" + collapseField + "] != [" + shardHits[i].field + "]");
        }
    }
    final PriorityQueue<ShardRef> queue = new MergeSortQueue(sort, shardHits);
    int totalHitCount = 0;
    int availHitCount = 0;
    float maxScore = Float.MIN_VALUE;
    for (int shardIDX = 0; shardIDX < shardHits.length; shardIDX++) {
        final CollapseTopFieldDocs shard = shardHits[shardIDX];
        // totalHits can be non-zero even if no hits were
        // collected, when searchAfter was used:
        totalHitCount += shard.totalHits;
        if (shard.scoreDocs != null && shard.scoreDocs.length > 0) {
            availHitCount += shard.scoreDocs.length;
            queue.add(new ShardRef(shardIDX));
            maxScore = Math.max(maxScore, shard.getMaxScore());
        }
    }
    if (availHitCount == 0) {
        maxScore = Float.NaN;
    }
    final ScoreDoc[] hits;
    final Object[] values;
    if (availHitCount <= start) {
        hits = new ScoreDoc[0];
        values = new Object[0];
    } else {
        List<ScoreDoc> hitList = new ArrayList<>();
        List<Object> collapseList = new ArrayList<>();
        int requestedResultWindow = start + size;
        int numIterOnHits = Math.min(availHitCount, requestedResultWindow);
        int hitUpto = 0;
        Set<Object> seen = new HashSet<>();
        while (hitUpto < numIterOnHits) {
            if (queue.size() == 0) {
                break;
            }
            ShardRef ref = queue.top();
            final ScoreDoc hit = shardHits[ref.shardIndex].scoreDocs[ref.hitIndex];
            final Object collapseValue = shardHits[ref.shardIndex].collapseValues[ref.hitIndex++];
            if (seen.contains(collapseValue)) {
                if (ref.hitIndex < shardHits[ref.shardIndex].scoreDocs.length) {
                    queue.updateTop();
                } else {
                    queue.pop();
                }
                continue;
            }
            seen.add(collapseValue);
            hit.shardIndex = ref.shardIndex;
            if (hitUpto >= start) {
                hitList.add(hit);
                collapseList.add(collapseValue);
            }
            hitUpto++;
            if (ref.hitIndex < shardHits[ref.shardIndex].scoreDocs.length) {
                // Not done with this these TopDocs yet:
                queue.updateTop();
            } else {
                queue.pop();
            }
        }
        hits = hitList.toArray(new ScoreDoc[0]);
        values = collapseList.toArray(new Object[0]);
    }
    return new CollapseTopFieldDocs(collapseField, totalHitCount, hits, sort.getSort(), values, maxScore);
}

Example 7 with ScoreDoc

use of org.apache.lucene.search.ScoreDoc in project elasticsearch by elastic.

the class SearchScrollQueryThenFetchAsyncAction method executeFetchPhase.

private void executeFetchPhase() throws Exception {
    sortedShardDocs = searchPhaseController.sortDocs(true, queryResults);
    if (sortedShardDocs.length == 0) {
        finishHim(searchPhaseController.reducedQueryPhase(queryResults.asList()));
        return;
    }
    final IntArrayList[] docIdsToLoad = searchPhaseController.fillDocIdsToLoad(queryResults.length(), sortedShardDocs);
    SearchPhaseController.ReducedQueryPhase reducedQueryPhase = searchPhaseController.reducedQueryPhase(queryResults.asList());
    final ScoreDoc[] lastEmittedDocPerShard = searchPhaseController.getLastEmittedDocPerShard(reducedQueryPhase, sortedShardDocs, queryResults.length());
    final AtomicInteger counter = new AtomicInteger(docIdsToLoad.length);
    for (int i = 0; i < docIdsToLoad.length; i++) {
        final int index = i;
        final IntArrayList docIds = docIdsToLoad[index];
        if (docIds != null) {
            final QuerySearchResult querySearchResult = queryResults.get(index);
            ScoreDoc lastEmittedDoc = lastEmittedDocPerShard[index];
            ShardFetchRequest shardFetchRequest = new ShardFetchRequest(querySearchResult.id(), docIds, lastEmittedDoc);
            DiscoveryNode node = nodes.get(querySearchResult.shardTarget().getNodeId());
            searchTransportService.sendExecuteFetchScroll(node, shardFetchRequest, task, new ActionListener<FetchSearchResult>() {

                @Override
                public void onResponse(FetchSearchResult result) {
                    result.shardTarget(querySearchResult.shardTarget());
                    fetchResults.set(index, result);
                    if (counter.decrementAndGet() == 0) {
                        finishHim(reducedQueryPhase);
                    }
                }

                @Override
                public void onFailure(Exception t) {
                    if (logger.isDebugEnabled()) {
                        logger.debug("Failed to execute fetch phase", t);
                    }
                    successfulOps.decrementAndGet();
                    if (counter.decrementAndGet() == 0) {
                        finishHim(reducedQueryPhase);
                    }
                }
            });
        } else {
            // the counter is set to the total size of docIdsToLoad which can have null values so we have to count them down too
            if (counter.decrementAndGet() == 0) {
                finishHim(reducedQueryPhase);
            }
        }
    }
}

Example 8 with ScoreDoc

use of org.apache.lucene.search.ScoreDoc in project elasticsearch by elastic.

the class SearchPhaseController method fillDocIdsToLoad.

/**
     * Builds an array, with potential null elements, with docs to load.
     */
public IntArrayList[] fillDocIdsToLoad(int numShards, ScoreDoc[] shardDocs) {
    IntArrayList[] docIdsToLoad = new IntArrayList[numShards];
    for (ScoreDoc shardDoc : shardDocs) {
        IntArrayList shardDocIdsToLoad = docIdsToLoad[shardDoc.shardIndex];
        if (shardDocIdsToLoad == null) {
            shardDocIdsToLoad = docIdsToLoad[shardDoc.shardIndex] = new IntArrayList();
        }
        shardDocIdsToLoad.add(shardDoc.doc);
    }
    return docIdsToLoad;
}

Example 9 with ScoreDoc

use of org.apache.lucene.search.ScoreDoc in project elasticsearch by elastic.

the class SearchPhaseController method getHits.

private SearchHits getHits(ReducedQueryPhase reducedQueryPhase, boolean ignoreFrom, ScoreDoc[] sortedDocs, AtomicArray<? extends QuerySearchResultProvider> fetchResultsArr) {
    List<? extends AtomicArray.Entry<? extends QuerySearchResultProvider>> fetchResults = fetchResultsArr.asList();
    boolean sorted = false;
    int sortScoreIndex = -1;
    if (reducedQueryPhase.oneResult.topDocs() instanceof TopFieldDocs) {
        TopFieldDocs fieldDocs = (TopFieldDocs) reducedQueryPhase.oneResult.queryResult().topDocs();
        if (fieldDocs instanceof CollapseTopFieldDocs && fieldDocs.fields.length == 1 && fieldDocs.fields[0].getType() == SortField.Type.SCORE) {
            sorted = false;
        } else {
            sorted = true;
            for (int i = 0; i < fieldDocs.fields.length; i++) {
                if (fieldDocs.fields[i].getType() == SortField.Type.SCORE) {
                    sortScoreIndex = i;
                }
            }
        }
    }
    // clean the fetch counter
    for (AtomicArray.Entry<? extends QuerySearchResultProvider> entry : fetchResults) {
        entry.value.fetchResult().initCounter();
    }
    int from = ignoreFrom ? 0 : reducedQueryPhase.oneResult.queryResult().from();
    int numSearchHits = (int) Math.min(reducedQueryPhase.fetchHits - from, reducedQueryPhase.oneResult.size());
    // with collapsing we can have more fetch hits than sorted docs
    numSearchHits = Math.min(sortedDocs.length, numSearchHits);
    // merge hits
    List<SearchHit> hits = new ArrayList<>();
    if (!fetchResults.isEmpty()) {
        for (int i = 0; i < numSearchHits; i++) {
            ScoreDoc shardDoc = sortedDocs[i];
            QuerySearchResultProvider fetchResultProvider = fetchResultsArr.get(shardDoc.shardIndex);
            if (fetchResultProvider == null) {
                continue;
            }
            FetchSearchResult fetchResult = fetchResultProvider.fetchResult();
            int index = fetchResult.counterGetAndIncrement();
            if (index < fetchResult.hits().internalHits().length) {
                SearchHit searchHit = fetchResult.hits().internalHits()[index];
                searchHit.score(shardDoc.score);
                searchHit.shard(fetchResult.shardTarget());
                if (sorted) {
                    FieldDoc fieldDoc = (FieldDoc) shardDoc;
                    searchHit.sortValues(fieldDoc.fields, reducedQueryPhase.oneResult.sortValueFormats());
                    if (sortScoreIndex != -1) {
                        searchHit.score(((Number) fieldDoc.fields[sortScoreIndex]).floatValue());
                    }
                }
                hits.add(searchHit);
            }
        }
    }
    return new SearchHits(hits.toArray(new SearchHit[hits.size()]), reducedQueryPhase.totalHits, reducedQueryPhase.maxScore);
}

Example 10 with ScoreDoc

use of org.apache.lucene.search.ScoreDoc in project elasticsearch by elastic.

the class TermVectorsUnitTests method writeEmptyTermVector.

private void writeEmptyTermVector(TermVectorsResponse outResponse) throws IOException {
    Directory dir = newDirectory();
    IndexWriterConfig conf = new IndexWriterConfig(new StandardAnalyzer());
    conf.setOpenMode(OpenMode.CREATE);
    IndexWriter writer = new IndexWriter(dir, conf);
    FieldType type = new FieldType(TextField.TYPE_STORED);
    type.setStoreTermVectorOffsets(true);
    type.setStoreTermVectorPayloads(false);
    type.setStoreTermVectorPositions(true);
    type.setStoreTermVectors(true);
    type.freeze();
    Document d = new Document();
    d.add(new Field("id", "abc", StringField.TYPE_STORED));
    writer.updateDocument(new Term("id", "abc"), d);
    writer.commit();
    writer.close();
    DirectoryReader dr = DirectoryReader.open(dir);
    IndexSearcher s = new IndexSearcher(dr);
    TopDocs search = s.search(new TermQuery(new Term("id", "abc")), 1);
    ScoreDoc[] scoreDocs = search.scoreDocs;
    int doc = scoreDocs[0].doc;
    Fields fields = dr.getTermVectors(doc);
    EnumSet<Flag> flags = EnumSet.of(Flag.Positions, Flag.Offsets);
    outResponse.setFields(fields, null, flags, fields);
    outResponse.setExists(true);
    dr.close();
    dir.close();
}