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

Example 16 with SortField

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

the class CollapsingTopDocsCollectorTests method testCollapseDouble.

public void testCollapseDouble() throws Exception {
    CollapsingDocValuesProducer producer = new CollapsingDocValuesProducer<Double>() {

        @Override
        public Double randomGroup(int maxGroup) {
            return new Double(randomIntBetween(0, maxGroup - 1));
        }

        @Override
        public void add(Document doc, Double value, boolean multivalued) {
            if (multivalued) {
                doc.add(new SortedNumericDocValuesField("field", NumericUtils.doubleToSortableLong(value)));
            } else {
                doc.add(new NumericDocValuesField("field", Double.doubleToLongBits(value)));
            }
        }

        @Override
        public SortField sortField(boolean multivalued) {
            if (multivalued) {
                return new SortedNumericSortField("field", SortField.Type.DOUBLE);
            } else {
                return new SortField("field", SortField.Type.DOUBLE);
            }
        }
    };
    assertSearchCollapse(producer, true);
}

Example 17 with SortField

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

the class CollapsingTopDocsCollectorTests method assertSearchCollapse.

private <T extends Comparable> void assertSearchCollapse(CollapsingDocValuesProducer<T> dvProducers, boolean numeric, boolean multivalued) throws IOException {
    final int numDocs = randomIntBetween(1000, 2000);
    int maxGroup = randomIntBetween(2, 500);
    final Directory dir = newDirectory();
    final RandomIndexWriter w = new RandomIndexWriter(random(), dir);
    Set<T> values = new HashSet<>();
    int totalHits = 0;
    for (int i = 0; i < numDocs; i++) {
        final T value = dvProducers.randomGroup(maxGroup);
        values.add(value);
        Document doc = new Document();
        dvProducers.add(doc, value, multivalued);
        doc.add(new NumericDocValuesField("sort1", randomIntBetween(0, 10)));
        doc.add(new NumericDocValuesField("sort2", randomLong()));
        w.addDocument(doc);
        totalHits++;
    }
    List<T> valueList = new ArrayList<>(values);
    Collections.sort(valueList);
    final IndexReader reader = w.getReader();
    final IndexSearcher searcher = newSearcher(reader);
    final SortField collapseField = dvProducers.sortField(multivalued);
    final SortField sort1 = new SortField("sort1", SortField.Type.INT);
    final SortField sort2 = new SortField("sort2", SortField.Type.LONG);
    Sort sort = new Sort(sort1, sort2, collapseField);
    int expectedNumGroups = values.size();
    final CollapsingTopDocsCollector collapsingCollector;
    if (numeric) {
        collapsingCollector = CollapsingTopDocsCollector.createNumeric(collapseField.getField(), sort, expectedNumGroups, false);
    } else {
        collapsingCollector = CollapsingTopDocsCollector.createKeyword(collapseField.getField(), sort, expectedNumGroups, false);
    }
    TopFieldCollector topFieldCollector = TopFieldCollector.create(sort, totalHits, true, false, false);
    searcher.search(new MatchAllDocsQuery(), collapsingCollector);
    searcher.search(new MatchAllDocsQuery(), topFieldCollector);
    CollapseTopFieldDocs collapseTopFieldDocs = collapsingCollector.getTopDocs();
    TopFieldDocs topDocs = topFieldCollector.topDocs();
    assertEquals(collapseField.getField(), collapseTopFieldDocs.field);
    assertEquals(expectedNumGroups, collapseTopFieldDocs.scoreDocs.length);
    assertEquals(totalHits, collapseTopFieldDocs.totalHits);
    assertEquals(totalHits, topDocs.scoreDocs.length);
    assertEquals(totalHits, topDocs.totalHits);
    Set<Object> seen = new HashSet<>();
    // collapse field is the last sort
    int collapseIndex = sort.getSort().length - 1;
    int topDocsIndex = 0;
    for (int i = 0; i < expectedNumGroups; i++) {
        FieldDoc fieldDoc = null;
        for (; topDocsIndex < totalHits; topDocsIndex++) {
            fieldDoc = (FieldDoc) topDocs.scoreDocs[topDocsIndex];
            if (seen.contains(fieldDoc.fields[collapseIndex]) == false) {
                break;
            }
        }
        FieldDoc collapseFieldDoc = (FieldDoc) collapseTopFieldDocs.scoreDocs[i];
        assertNotNull(fieldDoc);
        assertEquals(collapseFieldDoc.doc, fieldDoc.doc);
        assertArrayEquals(collapseFieldDoc.fields, fieldDoc.fields);
        seen.add(fieldDoc.fields[fieldDoc.fields.length - 1]);
    }
    for (; topDocsIndex < totalHits; topDocsIndex++) {
        FieldDoc fieldDoc = (FieldDoc) topDocs.scoreDocs[topDocsIndex];
        assertTrue(seen.contains(fieldDoc.fields[collapseIndex]));
    }
    // check merge
    final IndexReaderContext ctx = searcher.getTopReaderContext();
    final SegmentSearcher[] subSearchers;
    final int[] docStarts;
    if (ctx instanceof LeafReaderContext) {
        subSearchers = new SegmentSearcher[1];
        docStarts = new int[1];
        subSearchers[0] = new SegmentSearcher((LeafReaderContext) ctx, ctx);
        docStarts[0] = 0;
    } else {
        final CompositeReaderContext compCTX = (CompositeReaderContext) ctx;
        final int size = compCTX.leaves().size();
        subSearchers = new SegmentSearcher[size];
        docStarts = new int[size];
        int docBase = 0;
        for (int searcherIDX = 0; searcherIDX < subSearchers.length; searcherIDX++) {
            final LeafReaderContext leave = compCTX.leaves().get(searcherIDX);
            subSearchers[searcherIDX] = new SegmentSearcher(leave, compCTX);
            docStarts[searcherIDX] = docBase;
            docBase += leave.reader().maxDoc();
        }
    }
    final CollapseTopFieldDocs[] shardHits = new CollapseTopFieldDocs[subSearchers.length];
    final Weight weight = searcher.createNormalizedWeight(new MatchAllDocsQuery(), false);
    for (int shardIDX = 0; shardIDX < subSearchers.length; shardIDX++) {
        final SegmentSearcher subSearcher = subSearchers[shardIDX];
        final CollapsingTopDocsCollector c;
        if (numeric) {
            c = CollapsingTopDocsCollector.createNumeric(collapseField.getField(), sort, expectedNumGroups, false);
        } else {
            c = CollapsingTopDocsCollector.createKeyword(collapseField.getField(), sort, expectedNumGroups, false);
        }
        subSearcher.search(weight, c);
        shardHits[shardIDX] = c.getTopDocs();
    }
    CollapseTopFieldDocs mergedFieldDocs = CollapseTopFieldDocs.merge(sort, 0, expectedNumGroups, shardHits);
    assertTopDocsEquals(mergedFieldDocs, collapseTopFieldDocs);
    w.close();
    reader.close();
    dir.close();
}

Example 18 with SortField

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

the class CollapsingTopDocsCollectorTests method testCollapseInt.

public void testCollapseInt() throws Exception {
    CollapsingDocValuesProducer producer = new CollapsingDocValuesProducer<Integer>() {

        @Override
        public Integer randomGroup(int maxGroup) {
            return randomIntBetween(0, maxGroup - 1);
        }

        @Override
        public void add(Document doc, Integer value, boolean multivalued) {
            if (multivalued) {
                doc.add(new SortedNumericDocValuesField("field", value));
            } else {
                doc.add(new NumericDocValuesField("field", value));
            }
        }

        @Override
        public SortField sortField(boolean multivalued) {
            if (multivalued) {
                return new SortedNumericSortField("field", SortField.Type.INT);
            } else {
                return new SortField("field", SortField.Type.INT);
            }
        }
    };
    assertSearchCollapse(producer, true);
}

Example 19 with SortField

use of org.apache.lucene.search.SortField in project OpenGrok by OpenGrok.

the class SearchHelper method prepareExec.

/**
     * Create the searcher to use wrt. to currently set parameters and the given
     * projects. Does not produce any {@link #redirect} link. It also does
     * nothing if {@link #redirect} or {@link #errorMsg} have a
     * none-{@code null} value.
     * <p>
     * Parameters which should be populated/set at this time: <ul>
     * <li>{@link #builder}</li> <li>{@link #dataRoot}</li>
     * <li>{@link #order} (falls back to relevance if unset)</li>
     * <li>{@link #parallel} (default: false)</li> </ul> Populates/sets: <ul>
     * <li>{@link #query}</li> <li>{@link #searcher}</li> <li>{@link #sort}</li>
     * <li>{@link #projects}</li> <li>{@link #errorMsg} if an error occurs</li>
     * </ul>
     *
     * @param projects project to use query. If empty, a no-project setup
     * is assumed (i.e. DATA_ROOT/index will be used instead of possible
     * multiple DATA_ROOT/$project/index).
     * @return this instance
     */
public SearchHelper prepareExec(SortedSet<String> projects) {
    if (redirect != null || errorMsg != null) {
        return this;
    }
    // the Query created by the QueryBuilder
    try {
        indexDir = new File(dataRoot, IndexDatabase.INDEX_DIR);
        query = builder.build();
        if (projects == null) {
            errorMsg = "No project selected!";
            return this;
        }
        this.projects = projects;
        if (projects.isEmpty()) {
            // no project setup
            FSDirectory dir = FSDirectory.open(indexDir.toPath());
            searcher = new IndexSearcher(DirectoryReader.open(dir));
            closeOnDestroy = true;
        } else {
            // We use MultiReader even for single project. This should
            // not matter given that MultiReader is just a cheap wrapper
            // around set of IndexReader objects.
            closeOnDestroy = false;
            MultiReader multireader = RuntimeEnvironment.getInstance().getMultiReader(projects, searcherList);
            if (multireader != null) {
                searcher = new IndexSearcher(multireader);
            } else {
                errorMsg = "Failed to initialize search. Check the index.";
            }
        }
        // Most probably they are not reused. SearcherLifetimeManager might help here.
        switch(order) {
            case LASTMODIFIED:
                sort = new Sort(new SortField(QueryBuilder.DATE, SortField.Type.STRING, true));
                break;
            case BY_PATH:
                sort = new Sort(new SortField(QueryBuilder.FULLPATH, SortField.Type.STRING));
                break;
            default:
                sort = Sort.RELEVANCE;
                break;
        }
        checker = new DirectSpellChecker();
    } catch (ParseException e) {
        errorMsg = PARSE_ERROR_MSG + e.getMessage();
    } catch (FileNotFoundException e) {
        //          errorMsg = "Index database(s) not found: " + e.getMessage();
        errorMsg = "Index database(s) not found.";
    } catch (IOException e) {
        errorMsg = e.getMessage();
    }
    return this;
}

Example 20 with SortField

use of org.apache.lucene.search.SortField in project neo4j-mobile-android by neo4j-contrib.

the class QueryContext method sort.

/**
     * Returns a QueryContext with sorting added to it.
     *
     * @param key The key to sort on.
     * @param additionalKeys Any additional keys to sort on.
     * @return A QueryContext with sorting added to it.
     */
public QueryContext sort(String key, String... additionalKeys) {
    SortField firstSortField = new SortField(key, SortField.STRING);
    if (additionalKeys.length == 0) {
        return sort(new Sort(firstSortField));
    }
    SortField[] sortFields = new SortField[1 + additionalKeys.length];
    sortFields[0] = firstSortField;
    for (int i = 0; i < additionalKeys.length; i++) {
        sortFields[1 + i] = new SortField(additionalKeys[i], SortField.STRING);
    }
    return sort(new Sort(sortFields));
}