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

Example 1 with TopDocsCollector

use of org.apache.lucene.search.TopDocsCollector in project lucene-solr by apache.

the class ExpandComponent method process.

@SuppressWarnings("unchecked")
@Override
public void process(ResponseBuilder rb) throws IOException {
    if (!rb.doExpand) {
        return;
    }
    SolrQueryRequest req = rb.req;
    SolrParams params = req.getParams();
    String field = params.get(ExpandParams.EXPAND_FIELD);
    String hint = null;
    if (field == null) {
        List<Query> filters = rb.getFilters();
        if (filters != null) {
            for (Query q : filters) {
                if (q instanceof CollapsingQParserPlugin.CollapsingPostFilter) {
                    CollapsingQParserPlugin.CollapsingPostFilter cp = (CollapsingQParserPlugin.CollapsingPostFilter) q;
                    field = cp.getField();
                    hint = cp.hint;
                }
            }
        }
    }
    if (field == null) {
        throw new IOException("Expand field is null.");
    }
    String sortParam = params.get(ExpandParams.EXPAND_SORT);
    String[] fqs = params.getParams(ExpandParams.EXPAND_FQ);
    String qs = params.get(ExpandParams.EXPAND_Q);
    int limit = params.getInt(ExpandParams.EXPAND_ROWS, 5);
    Sort sort = null;
    if (sortParam != null) {
        sort = SortSpecParsing.parseSortSpec(sortParam, rb.req).getSort();
    }
    Query query;
    if (qs == null) {
        query = rb.getQuery();
    } else {
        try {
            QParser parser = QParser.getParser(qs, req);
            query = parser.getQuery();
        } catch (Exception e) {
            throw new IOException(e);
        }
    }
    List<Query> newFilters = new ArrayList<>();
    if (fqs == null) {
        List<Query> filters = rb.getFilters();
        if (filters != null) {
            for (Query q : filters) {
                if (!(q instanceof CollapsingQParserPlugin.CollapsingPostFilter)) {
                    newFilters.add(q);
                }
            }
        }
    } else {
        try {
            for (String fq : fqs) {
                if (fq != null && fq.trim().length() != 0 && !fq.equals("*:*")) {
                    QParser fqp = QParser.getParser(fq, req);
                    newFilters.add(fqp.getQuery());
                }
            }
        } catch (Exception e) {
            throw new IOException(e);
        }
    }
    SolrIndexSearcher searcher = req.getSearcher();
    LeafReader reader = searcher.getSlowAtomicReader();
    SchemaField schemaField = searcher.getSchema().getField(field);
    FieldType fieldType = schemaField.getType();
    SortedDocValues values = null;
    long nullValue = 0L;
    if (fieldType instanceof StrField) {
        //Get The Top Level SortedDocValues
        if (CollapsingQParserPlugin.HINT_TOP_FC.equals(hint)) {
            Map<String, UninvertingReader.Type> mapping = new HashMap();
            mapping.put(field, UninvertingReader.Type.SORTED);
            UninvertingReader uninvertingReader = new UninvertingReader(new ReaderWrapper(searcher.getSlowAtomicReader(), field), mapping);
            values = uninvertingReader.getSortedDocValues(field);
        } else {
            values = DocValues.getSorted(reader, field);
        }
    } else {
        //Get the nullValue for the numeric collapse field
        String defaultValue = searcher.getSchema().getField(field).getDefaultValue();
        final NumberType numType = fieldType.getNumberType();
        // we don't need to handle invalid 64-bit field types here.
        if (defaultValue != null) {
            if (numType == NumberType.INTEGER) {
                nullValue = Long.parseLong(defaultValue);
            } else if (numType == NumberType.FLOAT) {
                nullValue = Float.floatToIntBits(Float.parseFloat(defaultValue));
            }
        } else if (NumberType.FLOAT.equals(numType)) {
            // Integer case already handled by nullValue defaulting to 0
            nullValue = Float.floatToIntBits(0.0f);
        }
    }
    FixedBitSet groupBits = null;
    LongHashSet groupSet = null;
    DocList docList = rb.getResults().docList;
    IntHashSet collapsedSet = new IntHashSet(docList.size() * 2);
    //Gather the groups for the current page of documents
    DocIterator idit = docList.iterator();
    int[] globalDocs = new int[docList.size()];
    int docsIndex = -1;
    while (idit.hasNext()) {
        globalDocs[++docsIndex] = idit.nextDoc();
    }
    Arrays.sort(globalDocs);
    Query groupQuery = null;
    /*
    * This code gathers the group information for the current page.
    */
    List<LeafReaderContext> contexts = searcher.getTopReaderContext().leaves();
    if (contexts.size() == 0) {
        //When no context is available we can skip the expanding
        return;
    }
    int currentContext = 0;
    int currentDocBase = contexts.get(currentContext).docBase;
    int nextDocBase = (currentContext + 1) < contexts.size() ? contexts.get(currentContext + 1).docBase : Integer.MAX_VALUE;
    IntObjectHashMap<BytesRef> ordBytes = null;
    if (values != null) {
        groupBits = new FixedBitSet(values.getValueCount());
        MultiDocValues.OrdinalMap ordinalMap = null;
        SortedDocValues[] sortedDocValues = null;
        LongValues segmentOrdinalMap = null;
        SortedDocValues currentValues = null;
        if (values instanceof MultiDocValues.MultiSortedDocValues) {
            ordinalMap = ((MultiDocValues.MultiSortedDocValues) values).mapping;
            sortedDocValues = ((MultiDocValues.MultiSortedDocValues) values).values;
            currentValues = sortedDocValues[currentContext];
            segmentOrdinalMap = ordinalMap.getGlobalOrds(currentContext);
        }
        int count = 0;
        ordBytes = new IntObjectHashMap<>();
        for (int i = 0; i < globalDocs.length; i++) {
            int globalDoc = globalDocs[i];
            while (globalDoc >= nextDocBase) {
                currentContext++;
                currentDocBase = contexts.get(currentContext).docBase;
                nextDocBase = (currentContext + 1) < contexts.size() ? contexts.get(currentContext + 1).docBase : Integer.MAX_VALUE;
                if (ordinalMap != null) {
                    currentValues = sortedDocValues[currentContext];
                    segmentOrdinalMap = ordinalMap.getGlobalOrds(currentContext);
                }
            }
            int contextDoc = globalDoc - currentDocBase;
            if (ordinalMap != null) {
                if (contextDoc > currentValues.docID()) {
                    currentValues.advance(contextDoc);
                }
                if (contextDoc == currentValues.docID()) {
                    int ord = currentValues.ordValue();
                    ++count;
                    BytesRef ref = currentValues.lookupOrd(ord);
                    ord = (int) segmentOrdinalMap.get(ord);
                    ordBytes.put(ord, BytesRef.deepCopyOf(ref));
                    groupBits.set(ord);
                    collapsedSet.add(globalDoc);
                }
            } else {
                if (globalDoc > values.docID()) {
                    values.advance(globalDoc);
                }
                if (globalDoc == values.docID()) {
                    int ord = values.ordValue();
                    ++count;
                    BytesRef ref = values.lookupOrd(ord);
                    ordBytes.put(ord, BytesRef.deepCopyOf(ref));
                    groupBits.set(ord);
                    collapsedSet.add(globalDoc);
                }
            }
        }
        if (count > 0 && count < 200) {
            try {
                groupQuery = getGroupQuery(field, count, ordBytes);
            } catch (Exception e) {
                throw new IOException(e);
            }
        }
    } else {
        groupSet = new LongHashSet(docList.size());
        NumericDocValues collapseValues = contexts.get(currentContext).reader().getNumericDocValues(field);
        int count = 0;
        for (int i = 0; i < globalDocs.length; i++) {
            int globalDoc = globalDocs[i];
            while (globalDoc >= nextDocBase) {
                currentContext++;
                currentDocBase = contexts.get(currentContext).docBase;
                nextDocBase = currentContext + 1 < contexts.size() ? contexts.get(currentContext + 1).docBase : Integer.MAX_VALUE;
                collapseValues = contexts.get(currentContext).reader().getNumericDocValues(field);
            }
            int contextDoc = globalDoc - currentDocBase;
            int valueDocID = collapseValues.docID();
            if (valueDocID < contextDoc) {
                valueDocID = collapseValues.advance(contextDoc);
            }
            long value;
            if (valueDocID == contextDoc) {
                value = collapseValues.longValue();
            } else {
                value = 0;
            }
            if (value != nullValue) {
                ++count;
                groupSet.add(value);
                collapsedSet.add(globalDoc);
            }
        }
        if (count > 0 && count < 200) {
            if (fieldType.isPointField()) {
                groupQuery = getPointGroupQuery(schemaField, count, groupSet);
            } else {
                groupQuery = getGroupQuery(field, fieldType, count, groupSet);
            }
        }
    }
    Collector collector;
    if (sort != null)
        sort = sort.rewrite(searcher);
    Collector groupExpandCollector = null;
    if (values != null) {
        //Get The Top Level SortedDocValues again so we can re-iterate:
        if (CollapsingQParserPlugin.HINT_TOP_FC.equals(hint)) {
            Map<String, UninvertingReader.Type> mapping = new HashMap();
            mapping.put(field, UninvertingReader.Type.SORTED);
            UninvertingReader uninvertingReader = new UninvertingReader(new ReaderWrapper(searcher.getSlowAtomicReader(), field), mapping);
            values = uninvertingReader.getSortedDocValues(field);
        } else {
            values = DocValues.getSorted(reader, field);
        }
        groupExpandCollector = new GroupExpandCollector(values, groupBits, collapsedSet, limit, sort);
    } else {
        groupExpandCollector = new NumericGroupExpandCollector(field, nullValue, groupSet, collapsedSet, limit, sort);
    }
    if (groupQuery != null) {
        //Limits the results to documents that are in the same group as the documents in the page.
        newFilters.add(groupQuery);
    }
    SolrIndexSearcher.ProcessedFilter pfilter = searcher.getProcessedFilter(null, newFilters);
    if (pfilter.postFilter != null) {
        pfilter.postFilter.setLastDelegate(groupExpandCollector);
        collector = pfilter.postFilter;
    } else {
        collector = groupExpandCollector;
    }
    if (pfilter.filter == null) {
        searcher.search(query, collector);
    } else {
        Query q = new BooleanQuery.Builder().add(query, Occur.MUST).add(pfilter.filter, Occur.FILTER).build();
        searcher.search(q, collector);
    }
    LongObjectMap<Collector> groups = ((GroupCollector) groupExpandCollector).getGroups();
    NamedList outMap = new SimpleOrderedMap();
    CharsRefBuilder charsRef = new CharsRefBuilder();
    for (LongObjectCursor<Collector> cursor : groups) {
        long groupValue = cursor.key;
        TopDocsCollector<?> topDocsCollector = TopDocsCollector.class.cast(cursor.value);
        TopDocs topDocs = topDocsCollector.topDocs();
        ScoreDoc[] scoreDocs = topDocs.scoreDocs;
        if (scoreDocs.length > 0) {
            int[] docs = new int[scoreDocs.length];
            float[] scores = new float[scoreDocs.length];
            for (int i = 0; i < docs.length; i++) {
                ScoreDoc scoreDoc = scoreDocs[i];
                docs[i] = scoreDoc.doc;
                scores[i] = scoreDoc.score;
            }
            DocSlice slice = new DocSlice(0, docs.length, docs, scores, topDocs.totalHits, topDocs.getMaxScore());
            if (fieldType instanceof StrField) {
                final BytesRef bytesRef = ordBytes.get((int) groupValue);
                fieldType.indexedToReadable(bytesRef, charsRef);
                String group = charsRef.toString();
                outMap.add(group, slice);
            } else {
                outMap.add(numericToString(fieldType, groupValue), slice);
            }
        }
    }
    rb.rsp.add("expanded", outMap);
}

Example 2 with TopDocsCollector

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

the class QueryPhase method execute.

/**
     * In a package-private method so that it can be tested without having to
     * wire everything (mapperService, etc.)
     * @return whether the rescoring phase should be executed
     */
static boolean execute(SearchContext searchContext, final IndexSearcher searcher) throws QueryPhaseExecutionException {
    QuerySearchResult queryResult = searchContext.queryResult();
    queryResult.searchTimedOut(false);
    final boolean doProfile = searchContext.getProfilers() != null;
    final SearchType searchType = searchContext.searchType();
    boolean rescore = false;
    try {
        queryResult.from(searchContext.from());
        queryResult.size(searchContext.size());
        Query query = searchContext.query();
        final int totalNumDocs = searcher.getIndexReader().numDocs();
        int numDocs = Math.min(searchContext.from() + searchContext.size(), totalNumDocs);
        Collector collector;
        Callable<TopDocs> topDocsCallable;
        DocValueFormat[] sortValueFormats = new DocValueFormat[0];
        // already rewritten
        assert query == searcher.rewrite(query);
        if (searchContext.size() == 0) {
            // no matter what the value of from is
            final TotalHitCountCollector totalHitCountCollector = new TotalHitCountCollector();
            collector = totalHitCountCollector;
            if (searchContext.getProfilers() != null) {
                collector = new InternalProfileCollector(collector, CollectorResult.REASON_SEARCH_COUNT, Collections.emptyList());
            }
            topDocsCallable = new Callable<TopDocs>() {

                @Override
                public TopDocs call() throws Exception {
                    return new TopDocs(totalHitCountCollector.getTotalHits(), Lucene.EMPTY_SCORE_DOCS, 0);
                }
            };
        } else {
            // Perhaps have a dedicated scroll phase?
            final ScrollContext scrollContext = searchContext.scrollContext();
            assert (scrollContext != null) == (searchContext.request().scroll() != null);
            final Collector topDocsCollector;
            ScoreDoc after = null;
            if (searchContext.request().scroll() != null) {
                numDocs = Math.min(searchContext.size(), totalNumDocs);
                after = scrollContext.lastEmittedDoc;
                if (returnsDocsInOrder(query, searchContext.sort())) {
                    if (scrollContext.totalHits == -1) {
                        // first round
                        assert scrollContext.lastEmittedDoc == null;
                    // there is not much that we can optimize here since we want to collect all
                    // documents in order to get the total number of hits
                    } else {
                        // skip to the desired doc and stop collecting after ${size} matches
                        if (scrollContext.lastEmittedDoc != null) {
                            BooleanQuery bq = new BooleanQuery.Builder().add(query, BooleanClause.Occur.MUST).add(new MinDocQuery(after.doc + 1), BooleanClause.Occur.FILTER).build();
                            query = bq;
                        }
                        searchContext.terminateAfter(numDocs);
                    }
                }
            } else {
                after = searchContext.searchAfter();
            }
            if (totalNumDocs == 0) {
                // top collectors don't like a size of 0
                numDocs = 1;
            }
            assert numDocs > 0;
            if (searchContext.collapse() == null) {
                if (searchContext.sort() != null) {
                    SortAndFormats sf = searchContext.sort();
                    topDocsCollector = TopFieldCollector.create(sf.sort, numDocs, (FieldDoc) after, true, searchContext.trackScores(), searchContext.trackScores());
                    sortValueFormats = sf.formats;
                } else {
                    rescore = !searchContext.rescore().isEmpty();
                    for (RescoreSearchContext rescoreContext : searchContext.rescore()) {
                        numDocs = Math.max(rescoreContext.window(), numDocs);
                    }
                    topDocsCollector = TopScoreDocCollector.create(numDocs, after);
                }
            } else {
                Sort sort = Sort.RELEVANCE;
                if (searchContext.sort() != null) {
                    sort = searchContext.sort().sort;
                }
                CollapseContext collapse = searchContext.collapse();
                topDocsCollector = collapse.createTopDocs(sort, numDocs, searchContext.trackScores());
                if (searchContext.sort() == null) {
                    sortValueFormats = new DocValueFormat[] { DocValueFormat.RAW };
                } else {
                    sortValueFormats = searchContext.sort().formats;
                }
            }
            collector = topDocsCollector;
            if (doProfile) {
                collector = new InternalProfileCollector(collector, CollectorResult.REASON_SEARCH_TOP_HITS, Collections.emptyList());
            }
            topDocsCallable = new Callable<TopDocs>() {

                @Override
                public TopDocs call() throws Exception {
                    final TopDocs topDocs;
                    if (topDocsCollector instanceof TopDocsCollector) {
                        topDocs = ((TopDocsCollector<?>) topDocsCollector).topDocs();
                    } else if (topDocsCollector instanceof CollapsingTopDocsCollector) {
                        topDocs = ((CollapsingTopDocsCollector) topDocsCollector).getTopDocs();
                    } else {
                        throw new IllegalStateException("Unknown top docs collector " + topDocsCollector.getClass().getName());
                    }
                    if (scrollContext != null) {
                        if (scrollContext.totalHits == -1) {
                            // first round
                            scrollContext.totalHits = topDocs.totalHits;
                            scrollContext.maxScore = topDocs.getMaxScore();
                        } else {
                            // subsequent round: the total number of hits and
                            // the maximum score were computed on the first round
                            topDocs.totalHits = scrollContext.totalHits;
                            topDocs.setMaxScore(scrollContext.maxScore);
                        }
                        if (searchContext.request().numberOfShards() == 1) {
                            // if we fetch the document in the same roundtrip, we already know the last emitted doc
                            if (topDocs.scoreDocs.length > 0) {
                                // set the last emitted doc
                                scrollContext.lastEmittedDoc = topDocs.scoreDocs[topDocs.scoreDocs.length - 1];
                            }
                        }
                    }
                    return topDocs;
                }
            };
        }
        final boolean terminateAfterSet = searchContext.terminateAfter() != SearchContext.DEFAULT_TERMINATE_AFTER;
        if (terminateAfterSet) {
            final Collector child = collector;
            // throws Lucene.EarlyTerminationException when given count is reached
            collector = Lucene.wrapCountBasedEarlyTerminatingCollector(collector, searchContext.terminateAfter());
            if (doProfile) {
                collector = new InternalProfileCollector(collector, CollectorResult.REASON_SEARCH_TERMINATE_AFTER_COUNT, Collections.singletonList((InternalProfileCollector) child));
            }
        }
        if (searchContext.parsedPostFilter() != null) {
            final Collector child = collector;
            // this will only get applied to the actual search collector and not
            // to any scoped collectors, also, it will only be applied to the main collector
            // since that is where the filter should only work
            final Weight filterWeight = searcher.createNormalizedWeight(searchContext.parsedPostFilter().query(), false);
            collector = new FilteredCollector(collector, filterWeight);
            if (doProfile) {
                collector = new InternalProfileCollector(collector, CollectorResult.REASON_SEARCH_POST_FILTER, Collections.singletonList((InternalProfileCollector) child));
            }
        }
        // plug in additional collectors, like aggregations
        final List<Collector> subCollectors = new ArrayList<>();
        subCollectors.add(collector);
        subCollectors.addAll(searchContext.queryCollectors().values());
        collector = MultiCollector.wrap(subCollectors);
        if (doProfile && collector instanceof InternalProfileCollector == false) {
            // When there is a single collector to wrap, MultiCollector returns it
            // directly, so only wrap in the case that there are several sub collectors
            final List<InternalProfileCollector> children = new AbstractList<InternalProfileCollector>() {

                @Override
                public InternalProfileCollector get(int index) {
                    return (InternalProfileCollector) subCollectors.get(index);
                }

                @Override
                public int size() {
                    return subCollectors.size();
                }
            };
            collector = new InternalProfileCollector(collector, CollectorResult.REASON_SEARCH_MULTI, children);
        }
        // apply the minimum score after multi collector so we filter aggs as well
        if (searchContext.minimumScore() != null) {
            final Collector child = collector;
            collector = new MinimumScoreCollector(collector, searchContext.minimumScore());
            if (doProfile) {
                collector = new InternalProfileCollector(collector, CollectorResult.REASON_SEARCH_MIN_SCORE, Collections.singletonList((InternalProfileCollector) child));
            }
        }
        if (collector.getClass() == TotalHitCountCollector.class) {
            // instead of using a collector
            while (true) {
                // a constant_score query
                if (query instanceof ConstantScoreQuery) {
                    query = ((ConstantScoreQuery) query).getQuery();
                } else {
                    break;
                }
            }
            if (query.getClass() == MatchAllDocsQuery.class) {
                collector = null;
                topDocsCallable = new Callable<TopDocs>() {

                    @Override
                    public TopDocs call() throws Exception {
                        int count = searcher.getIndexReader().numDocs();
                        return new TopDocs(count, Lucene.EMPTY_SCORE_DOCS, 0);
                    }
                };
            } else if (query.getClass() == TermQuery.class && searcher.getIndexReader().hasDeletions() == false) {
                final Term term = ((TermQuery) query).getTerm();
                collector = null;
                topDocsCallable = new Callable<TopDocs>() {

                    @Override
                    public TopDocs call() throws Exception {
                        int count = 0;
                        for (LeafReaderContext context : searcher.getIndexReader().leaves()) {
                            count += context.reader().docFreq(term);
                        }
                        return new TopDocs(count, Lucene.EMPTY_SCORE_DOCS, 0);
                    }
                };
            }
        }
        final boolean timeoutSet = searchContext.timeout() != null && !searchContext.timeout().equals(SearchService.NO_TIMEOUT);
        if (timeoutSet && collector != null) {
            // collector might be null if no collection is actually needed
            final Collector child = collector;
            // TODO: change to use our own counter that uses the scheduler in ThreadPool
            // throws TimeLimitingCollector.TimeExceededException when timeout has reached
            collector = Lucene.wrapTimeLimitingCollector(collector, searchContext.timeEstimateCounter(), searchContext.timeout().millis());
            if (doProfile) {
                collector = new InternalProfileCollector(collector, CollectorResult.REASON_SEARCH_TIMEOUT, Collections.singletonList((InternalProfileCollector) child));
            }
        }
        if (collector != null) {
            final Collector child = collector;
            collector = new CancellableCollector(searchContext.getTask()::isCancelled, searchContext.lowLevelCancellation(), collector);
            if (doProfile) {
                collector = new InternalProfileCollector(collector, CollectorResult.REASON_SEARCH_CANCELLED, Collections.singletonList((InternalProfileCollector) child));
            }
        }
        try {
            if (collector != null) {
                if (doProfile) {
                    searchContext.getProfilers().getCurrentQueryProfiler().setCollector((InternalProfileCollector) collector);
                }
                searcher.search(query, collector);
            }
        } catch (TimeLimitingCollector.TimeExceededException e) {
            assert timeoutSet : "TimeExceededException thrown even though timeout wasn't set";
            queryResult.searchTimedOut(true);
        } catch (Lucene.EarlyTerminationException e) {
            assert terminateAfterSet : "EarlyTerminationException thrown even though terminateAfter wasn't set";
            queryResult.terminatedEarly(true);
        } finally {
            searchContext.clearReleasables(SearchContext.Lifetime.COLLECTION);
        }
        if (terminateAfterSet && queryResult.terminatedEarly() == null) {
            queryResult.terminatedEarly(false);
        }
        queryResult.topDocs(topDocsCallable.call(), sortValueFormats);
        if (searchContext.getProfilers() != null) {
            ProfileShardResult shardResults = SearchProfileShardResults.buildShardResults(searchContext.getProfilers());
            searchContext.queryResult().profileResults(shardResults);
        }
        return rescore;
    } catch (Exception e) {
        throw new QueryPhaseExecutionException(searchContext, "Failed to execute main query", e);
    }
}

Example 3 with TopDocsCollector

use of org.apache.lucene.search.TopDocsCollector in project lucene-solr by apache.

the class TopGroupsCollector method getTopGroups.

/**
   * Get the TopGroups recorded by this collector
   * @param withinGroupOffset the offset within each group to start collecting documents
   */
public TopGroups<T> getTopGroups(int withinGroupOffset) {
    @SuppressWarnings({ "unchecked", "rawtypes" }) final GroupDocs<T>[] groupDocsResult = (GroupDocs<T>[]) new GroupDocs[groups.size()];
    int groupIDX = 0;
    float maxScore = Float.MIN_VALUE;
    for (SearchGroup<T> group : groups) {
        TopDocsCollector<?> collector = (TopDocsCollector<?>) groupReducer.getCollector(group.groupValue);
        final TopDocs topDocs = collector.topDocs(withinGroupOffset, maxDocsPerGroup);
        groupDocsResult[groupIDX++] = new GroupDocs<>(Float.NaN, topDocs.getMaxScore(), topDocs.totalHits, topDocs.scoreDocs, group.groupValue, group.sortValues);
        maxScore = Math.max(maxScore, topDocs.getMaxScore());
    }
    return new TopGroups<>(groupSort.getSort(), withinGroupSort.getSort(), totalHitCount, totalGroupedHitCount, groupDocsResult, maxScore);
}