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

Example 6 with FieldDoc

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

the class TestNumericDocValuesUpdates method testUpdateSegmentWithNoDocValues2.

@Test
public void testUpdateSegmentWithNoDocValues2() throws Exception {
    Directory dir = newDirectory();
    IndexWriterConfig conf = newIndexWriterConfig(new MockAnalyzer(random()));
    // prevent merges, otherwise by the time updates are applied
    // (writer.close()), the segments might have merged and that update becomes
    // legit.
    conf.setMergePolicy(NoMergePolicy.INSTANCE);
    IndexWriter writer = new IndexWriter(dir, conf);
    // first segment with NDV
    Document doc = new Document();
    doc.add(new StringField("id", "doc0", Store.NO));
    doc.add(new NumericDocValuesField("ndv", 3));
    writer.addDocument(doc);
    doc = new Document();
    // document without 'ndv' field
    doc.add(new StringField("id", "doc4", Store.NO));
    writer.addDocument(doc);
    writer.commit();
    // second segment with no NDV, but another dv field "foo"
    doc = new Document();
    doc.add(new StringField("id", "doc1", Store.NO));
    doc.add(new NumericDocValuesField("foo", 3));
    writer.addDocument(doc);
    doc = new Document();
    // document that isn't updated
    doc.add(new StringField("id", "doc2", Store.NO));
    writer.addDocument(doc);
    writer.commit();
    // update document in the first segment - should not affect docsWithField of
    // the document without NDV field
    writer.updateNumericDocValue(new Term("id", "doc0"), "ndv", 5L);
    // update document in the second segment - field should be added and we should
    // be able to handle the other document correctly (e.g. no NPE)
    writer.updateNumericDocValue(new Term("id", "doc1"), "ndv", 5L);
    writer.close();
    DirectoryReader reader = DirectoryReader.open(dir);
    for (LeafReaderContext context : reader.leaves()) {
        LeafReader r = context.reader();
        NumericDocValues ndv = r.getNumericDocValues("ndv");
        assertEquals(0, ndv.nextDoc());
        assertEquals(5L, ndv.longValue());
        assertTrue(ndv.nextDoc() > 1);
    }
    reader.close();
    TestUtil.checkIndex(dir);
    conf = newIndexWriterConfig(new MockAnalyzer(random()));
    writer = new IndexWriter(dir, conf);
    writer.forceMerge(1);
    writer.close();
    reader = DirectoryReader.open(dir);
    LeafReader ar = getOnlyLeafReader(reader);
    assertEquals(DocValuesType.NUMERIC, ar.getFieldInfos().fieldInfo("foo").getDocValuesType());
    IndexSearcher searcher = new IndexSearcher(reader);
    TopFieldDocs td;
    // doc0
    td = searcher.search(new TermQuery(new Term("id", "doc0")), 1, new Sort(new SortField("ndv", SortField.Type.LONG)));
    assertEquals(5L, ((FieldDoc) td.scoreDocs[0]).fields[0]);
    // doc1
    td = searcher.search(new TermQuery(new Term("id", "doc1")), 1, new Sort(new SortField("ndv", SortField.Type.LONG), new SortField("foo", SortField.Type.LONG)));
    assertEquals(5L, ((FieldDoc) td.scoreDocs[0]).fields[0]);
    assertEquals(3L, ((FieldDoc) td.scoreDocs[0]).fields[1]);
    // doc2
    td = searcher.search(new TermQuery(new Term("id", "doc2")), 1, new Sort(new SortField("ndv", SortField.Type.LONG)));
    assertEquals(0L, ((FieldDoc) td.scoreDocs[0]).fields[0]);
    // doc4
    td = searcher.search(new TermQuery(new Term("id", "doc4")), 1, new Sort(new SortField("ndv", SortField.Type.LONG)));
    assertEquals(0L, ((FieldDoc) td.scoreDocs[0]).fields[0]);
    reader.close();
    dir.close();
}

Example 7 with FieldDoc

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

the class LatLonPoint method nearest.

/**
   * Finds the {@code n} nearest indexed points to the provided point, according to Haversine distance.
   * <p>
   * This is functionally equivalent to running {@link MatchAllDocsQuery} with a {@link LatLonDocValuesField#newDistanceSort},
   * but is far more efficient since it takes advantage of properties the indexed BKD tree.  Currently this
   * only works with {@link Lucene60PointsFormat} (used by the default codec).  Multi-valued fields are
   * currently not de-duplicated, so if a document had multiple instances of the specified field that
   * make it into the top n, that document will appear more than once.
   * <p>
   * Documents are ordered by ascending distance from the location. The value returned in {@link FieldDoc} for
   * the hits contains a Double instance with the distance in meters.
   * 
   * @param searcher IndexSearcher to find nearest points from.
   * @param field field name. must not be null.
   * @param latitude latitude at the center: must be within standard +/-90 coordinate bounds.
   * @param longitude longitude at the center: must be within standard +/-180 coordinate bounds.
   * @param n the number of nearest neighbors to retrieve.
   * @return TopFieldDocs containing documents ordered by distance, where the field value for each {@link FieldDoc} is the distance in meters
   * @throws IllegalArgumentException if the underlying PointValues is not a {@code Lucene60PointsReader} (this is a current limitation), or
   *         if {@code field} or {@code searcher} is null, or if {@code latitude}, {@code longitude} or {@code n} are out-of-bounds
   * @throws IOException if an IOException occurs while finding the points.
   */
// TODO: what about multi-valued documents? what happens?
public static TopFieldDocs nearest(IndexSearcher searcher, String field, double latitude, double longitude, int n) throws IOException {
    GeoUtils.checkLatitude(latitude);
    GeoUtils.checkLongitude(longitude);
    if (n < 1) {
        throw new IllegalArgumentException("n must be at least 1; got " + n);
    }
    if (field == null) {
        throw new IllegalArgumentException("field must not be null");
    }
    if (searcher == null) {
        throw new IllegalArgumentException("searcher must not be null");
    }
    List<BKDReader> readers = new ArrayList<>();
    List<Integer> docBases = new ArrayList<>();
    List<Bits> liveDocs = new ArrayList<>();
    int totalHits = 0;
    for (LeafReaderContext leaf : searcher.getIndexReader().leaves()) {
        PointValues points = leaf.reader().getPointValues(field);
        if (points != null) {
            if (points instanceof BKDReader == false) {
                throw new IllegalArgumentException("can only run on Lucene60PointsReader points implementation, but got " + points);
            }
            totalHits += points.getDocCount();
            BKDReader reader = (BKDReader) points;
            if (reader != null) {
                readers.add(reader);
                docBases.add(leaf.docBase);
                liveDocs.add(leaf.reader().getLiveDocs());
            }
        }
    }
    NearestNeighbor.NearestHit[] hits = NearestNeighbor.nearest(latitude, longitude, readers, liveDocs, docBases, n);
    // Convert to TopFieldDocs:
    ScoreDoc[] scoreDocs = new ScoreDoc[hits.length];
    for (int i = 0; i < hits.length; i++) {
        NearestNeighbor.NearestHit hit = hits[i];
        scoreDocs[i] = new FieldDoc(hit.docID, 0.0f, new Object[] { Double.valueOf(hit.distanceMeters) });
    }
    return new TopFieldDocs(totalHits, scoreDocs, null, 0.0f);
}

Example 8 with FieldDoc

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

the class AnalyzingInfixSuggester method createResults.

/**
   * Create the results based on the search hits.
   * Can be overridden by subclass to add particular behavior (e.g. weight transformation).
   * Note that there is no prefix toke (the {@code prefixToken} argument will
   * be null) whenever the final token in the incoming request was in fact finished
   * (had trailing characters, such as white-space).
   *
   * @throws IOException If there are problems reading fields from the underlying Lucene index.
   */
protected List<LookupResult> createResults(IndexSearcher searcher, TopFieldDocs hits, int num, CharSequence charSequence, boolean doHighlight, Set<String> matchedTokens, String prefixToken) throws IOException {
    List<LeafReaderContext> leaves = searcher.getIndexReader().leaves();
    List<LookupResult> results = new ArrayList<>();
    for (int i = 0; i < hits.scoreDocs.length; i++) {
        FieldDoc fd = (FieldDoc) hits.scoreDocs[i];
        BinaryDocValues textDV = MultiDocValues.getBinaryValues(searcher.getIndexReader(), TEXT_FIELD_NAME);
        textDV.advance(fd.doc);
        BytesRef term = textDV.binaryValue();
        String text = term.utf8ToString();
        long score = (Long) fd.fields[0];
        // This will just be null if app didn't pass payloads to build():
        // TODO: maybe just stored fields?  they compress...
        BinaryDocValues payloadsDV = MultiDocValues.getBinaryValues(searcher.getIndexReader(), "payloads");
        BytesRef payload;
        if (payloadsDV != null) {
            if (payloadsDV.advance(fd.doc) == fd.doc) {
                payload = BytesRef.deepCopyOf(payloadsDV.binaryValue());
            } else {
                payload = new BytesRef(BytesRef.EMPTY_BYTES);
            }
        } else {
            payload = null;
        }
        // Must look up sorted-set by segment:
        int segment = ReaderUtil.subIndex(fd.doc, leaves);
        SortedSetDocValues contextsDV = leaves.get(segment).reader().getSortedSetDocValues(CONTEXTS_FIELD_NAME);
        Set<BytesRef> contexts;
        if (contextsDV != null) {
            contexts = new HashSet<BytesRef>();
            int targetDocID = fd.doc - leaves.get(segment).docBase;
            if (contextsDV.advance(targetDocID) == targetDocID) {
                long ord;
                while ((ord = contextsDV.nextOrd()) != SortedSetDocValues.NO_MORE_ORDS) {
                    BytesRef context = BytesRef.deepCopyOf(contextsDV.lookupOrd(ord));
                    contexts.add(context);
                }
            }
        } else {
            contexts = null;
        }
        LookupResult result;
        if (doHighlight) {
            result = new LookupResult(text, highlight(text, matchedTokens, prefixToken), score, payload, contexts);
        } else {
            result = new LookupResult(text, score, payload, contexts);
        }
        results.add(result);
    }
    return results;
}

Example 9 with FieldDoc

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

the class TestDemoExpressions method testDollarVariable.

/** Uses variables with $ */
public void testDollarVariable() throws Exception {
    Expression expr = JavascriptCompiler.compile("$0+$score");
    SimpleBindings bindings = new SimpleBindings();
    bindings.add(new SortField("$0", SortField.Type.SCORE));
    bindings.add(new SortField("$score", SortField.Type.SCORE));
    Sort sort = new Sort(expr.getSortField(bindings, true));
    Query query = new TermQuery(new Term("body", "contents"));
    TopFieldDocs td = searcher.search(query, 3, sort, true, true);
    for (int i = 0; i < 3; i++) {
        FieldDoc d = (FieldDoc) td.scoreDocs[i];
        float expected = 2 * d.score;
        float actual = ((Double) d.fields[0]).floatValue();
        assertEquals(expected, actual, CheckHits.explainToleranceDelta(expected, actual));
    }
}

Example 10 with FieldDoc

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

the class TestDemoExpressions method testStaticExtendedVariableExample.

public void testStaticExtendedVariableExample() throws Exception {
    Expression popularity = JavascriptCompiler.compile("doc[\"popularity\"].value");
    SimpleBindings bindings = new SimpleBindings();
    bindings.add("doc['popularity'].value", DoubleValuesSource.fromIntField("popularity"));
    Sort sort = new Sort(popularity.getSortField(bindings, true));
    TopFieldDocs td = searcher.search(new MatchAllDocsQuery(), 3, sort);
    FieldDoc d = (FieldDoc) td.scoreDocs[0];
    assertEquals(20D, (Double) d.fields[0], 1E-4);
    d = (FieldDoc) td.scoreDocs[1];
    assertEquals(5D, (Double) d.fields[0], 1E-4);
    d = (FieldDoc) td.scoreDocs[2];
    assertEquals(2D, (Double) d.fields[0], 1E-4);
}