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

Example 36 with FieldDoc

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

the class InternalTopHitsTests method createTestInstance.

@Override
protected InternalTopHits createTestInstance(String name, List<PipelineAggregator> pipelineAggregators, Map<String, Object> metaData) {
    int from = 0;
    int requestedSize = between(1, 40);
    int actualSize = between(0, requestedSize);
    float maxScore = Float.MIN_VALUE;
    ScoreDoc[] scoreDocs = new ScoreDoc[actualSize];
    SearchHit[] hits = new SearchHit[actualSize];
    Set<Integer> usedDocIds = new HashSet<>();
    for (int i = 0; i < actualSize; i++) {
        float score = randomFloat();
        maxScore = max(maxScore, score);
        int docId = randomValueOtherThanMany(usedDocIds::contains, () -> between(0, IndexWriter.MAX_DOCS));
        usedDocIds.add(docId);
        Map<String, SearchHitField> searchHitFields = new HashMap<>();
        if (testInstancesLookSortedByField) {
            Object[] fields = new Object[testInstancesSortFields.length];
            for (int f = 0; f < testInstancesSortFields.length; f++) {
                fields[f] = randomOfType(testInstancesSortFields[f].getType());
            }
            scoreDocs[i] = new FieldDoc(docId, score, fields);
        } else {
            scoreDocs[i] = new ScoreDoc(docId, score);
        }
        hits[i] = new SearchHit(docId, Integer.toString(i), new Text("test"), searchHitFields);
        hits[i].score(score);
    }
    int totalHits = between(actualSize, 500000);
    SearchHits searchHits = new SearchHits(hits, totalHits, maxScore);
    TopDocs topDocs;
    Arrays.sort(scoreDocs, scoreDocComparator());
    if (testInstancesLookSortedByField) {
        topDocs = new TopFieldDocs(totalHits, scoreDocs, testInstancesSortFields, maxScore);
    } else {
        topDocs = new TopDocs(totalHits, scoreDocs, maxScore);
    }
    return new InternalTopHits(name, from, requestedSize, topDocs, searchHits, pipelineAggregators, metaData);
}

Example 37 with FieldDoc

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

the class InternalTopHitsTests method assertReduced.

@Override
protected void assertReduced(InternalTopHits reduced, List<InternalTopHits> inputs) {
    SearchHits actualHits = reduced.getHits();
    List<Tuple<ScoreDoc, SearchHit>> allHits = new ArrayList<>();
    float maxScore = Float.MIN_VALUE;
    long totalHits = 0;
    for (int input = 0; input < inputs.size(); input++) {
        SearchHits internalHits = inputs.get(input).getHits();
        totalHits += internalHits.getTotalHits();
        maxScore = max(maxScore, internalHits.getMaxScore());
        for (int i = 0; i < internalHits.internalHits().length; i++) {
            ScoreDoc doc = inputs.get(input).getTopDocs().scoreDocs[i];
            if (testInstancesLookSortedByField) {
                doc = new FieldDoc(doc.doc, doc.score, ((FieldDoc) doc).fields, input);
            } else {
                doc = new ScoreDoc(doc.doc, doc.score, input);
            }
            allHits.add(new Tuple<>(doc, internalHits.internalHits()[i]));
        }
    }
    allHits.sort(comparing(Tuple::v1, scoreDocComparator()));
    SearchHit[] expectedHitsHits = new SearchHit[min(inputs.get(0).getSize(), allHits.size())];
    for (int i = 0; i < expectedHitsHits.length; i++) {
        expectedHitsHits[i] = allHits.get(i).v2();
    }
    SearchHits expectedHits = new SearchHits(expectedHitsHits, totalHits, maxScore);
    assertEqualsWithErrorMessageFromXContent(expectedHits, actualHits);
}

Example 38 with FieldDoc

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

the class TestNumericDocValuesUpdates method testBiasedMixOfRandomUpdates.

public void testBiasedMixOfRandomUpdates() throws Exception {
    // 3 types of operations: add, updated, updateDV.
    // rather then randomizing equally, we'll pick (random) cutoffs so each test run is biased,
    // in terms of some ops happen more often then others
    final int ADD_CUTOFF = TestUtil.nextInt(random(), 1, 98);
    final int UPD_CUTOFF = TestUtil.nextInt(random(), ADD_CUTOFF + 1, 99);
    Directory dir = newDirectory();
    IndexWriterConfig conf = newIndexWriterConfig(new MockAnalyzer(random()));
    IndexWriter writer = new IndexWriter(dir, conf);
    final int numOperations = atLeast(1000);
    final Map<Integer, Long> expected = new HashMap<>(numOperations / 3);
    // start with at least one doc before any chance of updates
    final int numSeedDocs = atLeast(1);
    for (int i = 0; i < numSeedDocs; i++) {
        final long val = random().nextLong();
        expected.put(i, val);
        writer.addDocument(doc(i, val));
    }
    int numDocUpdates = 0;
    int numValueUpdates = 0;
    for (int i = 0; i < numOperations; i++) {
        final int op = TestUtil.nextInt(random(), 1, 100);
        final long val = random().nextLong();
        if (op <= ADD_CUTOFF) {
            final int id = expected.size();
            //System.out.println("TEST i=" + i + ": addDocument id=" + id + " val=" + val);
            expected.put(id, val);
            writer.addDocument(doc(id, val));
        } else {
            final int id = TestUtil.nextInt(random(), 0, expected.size() - 1);
            expected.put(id, val);
            if (op <= UPD_CUTOFF) {
                numDocUpdates++;
                //System.out.println("TEST i=" + i + ": updateDocument id=" + id + " val=" + val);
                writer.updateDocument(new Term("id", "doc-" + id), doc(id, val));
            } else {
                numValueUpdates++;
                //System.out.println("TEST i=" + i + ": updateDV id=" + id + " val=" + val);
                writer.updateNumericDocValue(new Term("id", "doc-" + id), "val", val);
            }
        }
    }
    writer.commit();
    final DirectoryReader reader = DirectoryReader.open(dir);
    final IndexSearcher searcher = new IndexSearcher(reader);
    // TODO: make more efficient if max numOperations is going to be increased much
    for (Map.Entry<Integer, Long> expect : expected.entrySet()) {
        String id = "doc-" + expect.getKey();
        TopFieldDocs td = searcher.search(new TermQuery(new Term("id", id)), 1, new Sort(new SortField("val", SortField.Type.LONG)));
        assertEquals(id + " missing?", 1, td.totalHits);
        assertEquals(id + " value", expect.getValue(), ((FieldDoc) td.scoreDocs[0]).fields[0]);
    }
    IOUtils.close(reader, writer, dir);
}

Example 39 with FieldDoc

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

the class TestNumericDocValuesUpdates method testMultipleUpdatesSameDoc.

public void testMultipleUpdatesSameDoc() throws Exception {
    Directory dir = newDirectory();
    IndexWriterConfig conf = newIndexWriterConfig(new MockAnalyzer(random()));
    // small number of docs, so use a tiny maxBufferedDocs
    conf.setMaxBufferedDocs(3);
    IndexWriter writer = new IndexWriter(dir, conf);
    writer.updateDocument(new Term("id", "doc-1"), doc(1, 1000000000L));
    writer.updateNumericDocValue(new Term("id", "doc-1"), "val", 1000001111L);
    writer.updateDocument(new Term("id", "doc-2"), doc(2, 2000000000L));
    writer.updateDocument(new Term("id", "doc-2"), doc(2, 2222222222L));
    writer.updateNumericDocValue(new Term("id", "doc-1"), "val", 1111111111L);
    writer.commit();
    final DirectoryReader reader = DirectoryReader.open(dir);
    final IndexSearcher searcher = new IndexSearcher(reader);
    TopFieldDocs td;
    td = searcher.search(new TermQuery(new Term("id", "doc-1")), 1, new Sort(new SortField("val", SortField.Type.LONG)));
    assertEquals("doc-1 missing?", 1, td.scoreDocs.length);
    assertEquals("doc-1 value", 1111111111L, ((FieldDoc) td.scoreDocs[0]).fields[0]);
    td = searcher.search(new TermQuery(new Term("id", "doc-2")), 1, new Sort(new SortField("val", SortField.Type.LONG)));
    assertEquals("doc-2 missing?", 1, td.scoreDocs.length);
    assertEquals("doc-2 value", 2222222222L, ((FieldDoc) td.scoreDocs[0]).fields[0]);
    IOUtils.close(reader, writer, dir);
}

Example 40 with FieldDoc

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

the class TestFieldCacheSortRandom method testRandomStringSort.

private void testRandomStringSort(SortField.Type type) throws Exception {
    Random random = new Random(random().nextLong());
    final int NUM_DOCS = atLeast(100);
    final Directory dir = newDirectory();
    final RandomIndexWriter writer = new RandomIndexWriter(random, dir);
    final boolean allowDups = random.nextBoolean();
    final Set<String> seen = new HashSet<>();
    final int maxLength = TestUtil.nextInt(random, 5, 100);
    if (VERBOSE) {
        System.out.println("TEST: NUM_DOCS=" + NUM_DOCS + " maxLength=" + maxLength + " allowDups=" + allowDups);
    }
    int numDocs = 0;
    final List<BytesRef> docValues = new ArrayList<>();
    // TODO: deletions
    while (numDocs < NUM_DOCS) {
        final Document doc = new Document();
        // 10% of the time, the document is missing the value:
        final BytesRef br;
        if (random().nextInt(10) != 7) {
            final String s;
            if (random.nextBoolean()) {
                s = TestUtil.randomSimpleString(random, maxLength);
            } else {
                s = TestUtil.randomUnicodeString(random, maxLength);
            }
            if (!allowDups) {
                if (seen.contains(s)) {
                    continue;
                }
                seen.add(s);
            }
            if (VERBOSE) {
                System.out.println("  " + numDocs + ": s=" + s);
            }
            doc.add(new StringField("stringdv", s, Field.Store.NO));
            docValues.add(new BytesRef(s));
        } else {
            br = null;
            if (VERBOSE) {
                System.out.println("  " + numDocs + ": <missing>");
            }
            docValues.add(null);
        }
        doc.add(new IntPoint("id", numDocs));
        doc.add(new StoredField("id", numDocs));
        writer.addDocument(doc);
        numDocs++;
        if (random.nextInt(40) == 17) {
            // force flush
            writer.getReader().close();
        }
    }
    Map<String, UninvertingReader.Type> mapping = new HashMap<>();
    mapping.put("stringdv", Type.SORTED);
    mapping.put("id", Type.INTEGER_POINT);
    final IndexReader r = UninvertingReader.wrap(writer.getReader(), mapping);
    writer.close();
    if (VERBOSE) {
        System.out.println("  reader=" + r);
    }
    final IndexSearcher s = newSearcher(r, false);
    final int ITERS = atLeast(100);
    for (int iter = 0; iter < ITERS; iter++) {
        final boolean reverse = random.nextBoolean();
        final TopFieldDocs hits;
        final SortField sf;
        final boolean sortMissingLast;
        final boolean missingIsNull;
        sf = new SortField("stringdv", type, reverse);
        sortMissingLast = random().nextBoolean();
        missingIsNull = true;
        if (sortMissingLast) {
            sf.setMissingValue(SortField.STRING_LAST);
        }
        final Sort sort;
        if (random.nextBoolean()) {
            sort = new Sort(sf);
        } else {
            sort = new Sort(sf, SortField.FIELD_DOC);
        }
        final int hitCount = TestUtil.nextInt(random, 1, r.maxDoc() + 20);
        final RandomQuery f = new RandomQuery(random.nextLong(), random.nextFloat(), docValues);
        int queryType = random.nextInt(2);
        if (queryType == 0) {
            hits = s.search(new ConstantScoreQuery(f), hitCount, sort, random.nextBoolean(), random.nextBoolean());
        } else {
            hits = s.search(f, hitCount, sort, random.nextBoolean(), random.nextBoolean());
        }
        if (VERBOSE) {
            System.out.println("\nTEST: iter=" + iter + " " + hits.totalHits + " hits; topN=" + hitCount + "; reverse=" + reverse + "; sortMissingLast=" + sortMissingLast + " sort=" + sort);
        }
        // Compute expected results:
        Collections.sort(f.matchValues, new Comparator<BytesRef>() {

            @Override
            public int compare(BytesRef a, BytesRef b) {
                if (a == null) {
                    if (b == null) {
                        return 0;
                    }
                    if (sortMissingLast) {
                        return 1;
                    } else {
                        return -1;
                    }
                } else if (b == null) {
                    if (sortMissingLast) {
                        return -1;
                    } else {
                        return 1;
                    }
                } else {
                    return a.compareTo(b);
                }
            }
        });
        if (reverse) {
            Collections.reverse(f.matchValues);
        }
        final List<BytesRef> expected = f.matchValues;
        if (VERBOSE) {
            System.out.println("  expected:");
            for (int idx = 0; idx < expected.size(); idx++) {
                BytesRef br = expected.get(idx);
                if (br == null && missingIsNull == false) {
                    br = new BytesRef();
                }
                System.out.println("    " + idx + ": " + (br == null ? "<missing>" : br.utf8ToString()));
                if (idx == hitCount - 1) {
                    break;
                }
            }
        }
        if (VERBOSE) {
            System.out.println("  actual:");
            for (int hitIDX = 0; hitIDX < hits.scoreDocs.length; hitIDX++) {
                final FieldDoc fd = (FieldDoc) hits.scoreDocs[hitIDX];
                BytesRef br = (BytesRef) fd.fields[0];
                System.out.println("    " + hitIDX + ": " + (br == null ? "<missing>" : br.utf8ToString()) + " id=" + s.doc(fd.doc).get("id"));
            }
        }
        for (int hitIDX = 0; hitIDX < hits.scoreDocs.length; hitIDX++) {
            final FieldDoc fd = (FieldDoc) hits.scoreDocs[hitIDX];
            BytesRef br = expected.get(hitIDX);
            if (br == null && missingIsNull == false) {
                br = new BytesRef();
            }
            // Normally, the old codecs (that don't support
            // docsWithField via doc values) will always return
            // an empty BytesRef for the missing case; however,
            // if all docs in a given segment were missing, in
            // that case it will return null!  So we must map
            // null here, too:
            BytesRef br2 = (BytesRef) fd.fields[0];
            if (br2 == null && missingIsNull == false) {
                br2 = new BytesRef();
            }
            assertEquals(br, br2);
        }
    }
    r.close();
    dir.close();
}