Examples with QueryShardContext org.opensearch.index.query.QueryShardContext used on opensource projects

Example 86 with QueryShardContext

use of org.opensearch.index.query.QueryShardContext in project OpenSearch by opensearch-project.

the class PhraseSuggester method innerExecute.

/*
     * More Ideas:
     *   - add ability to find whitespace problems -> we can build a poor mans decompounder with our index based on a automaton?
     *   - add ability to build different error models maybe based on a confusion matrix?
     *   - try to combine a token with its subsequent token to find / detect word splits (optional)
     *      - for this to work we need some way to defined the position length of a candidate
     *   - phonetic filters could be interesting here too for candidate selection
     */
@Override
public Suggestion<? extends Entry<? extends Option>> innerExecute(String name, PhraseSuggestionContext suggestion, IndexSearcher searcher, CharsRefBuilder spare) throws IOException {
    double realWordErrorLikelihood = suggestion.realworldErrorLikelihood();
    final PhraseSuggestion response = new PhraseSuggestion(name, suggestion.getSize());
    final IndexReader indexReader = searcher.getIndexReader();
    List<PhraseSuggestionContext.DirectCandidateGenerator> generators = suggestion.generators();
    final int numGenerators = generators.size();
    final List<CandidateGenerator> gens = new ArrayList<>(generators.size());
    for (int i = 0; i < numGenerators; i++) {
        PhraseSuggestionContext.DirectCandidateGenerator generator = generators.get(i);
        DirectSpellChecker directSpellChecker = generator.createDirectSpellChecker();
        Terms terms = MultiTerms.getTerms(indexReader, generator.field());
        if (terms != null) {
            gens.add(new DirectCandidateGenerator(directSpellChecker, generator.field(), generator.suggestMode(), indexReader, realWordErrorLikelihood, generator.size(), generator.preFilter(), generator.postFilter(), terms));
        }
    }
    final String suggestField = suggestion.getField();
    final Terms suggestTerms = MultiTerms.getTerms(indexReader, suggestField);
    if (gens.size() > 0 && suggestTerms != null) {
        final NoisyChannelSpellChecker checker = new NoisyChannelSpellChecker(realWordErrorLikelihood, suggestion.getRequireUnigram(), suggestion.getTokenLimit());
        final BytesRef separator = suggestion.separator();
        WordScorer wordScorer = suggestion.model().newScorer(indexReader, suggestTerms, suggestField, realWordErrorLikelihood, separator);
        Result checkerResult;
        try (TokenStream stream = tokenStream(suggestion.getAnalyzer(), suggestion.getText(), spare, suggestion.getField())) {
            checkerResult = checker.getCorrections(stream, new MultiCandidateGeneratorWrapper(suggestion.getShardSize(), gens.toArray(new CandidateGenerator[gens.size()])), suggestion.maxErrors(), suggestion.getShardSize(), wordScorer, suggestion.confidence(), suggestion.gramSize());
        }
        PhraseSuggestion.Entry resultEntry = buildResultEntry(suggestion, spare, checkerResult.cutoffScore);
        response.addTerm(resultEntry);
        final BytesRefBuilder byteSpare = new BytesRefBuilder();
        final TemplateScript.Factory scriptFactory = suggestion.getCollateQueryScript();
        final boolean collatePrune = (scriptFactory != null) && suggestion.collatePrune();
        for (int i = 0; i < checkerResult.corrections.length; i++) {
            Correction correction = checkerResult.corrections[i];
            spare.copyUTF8Bytes(correction.join(SEPARATOR, byteSpare, null, null));
            boolean collateMatch = true;
            if (scriptFactory != null) {
                // Checks if the template query collateScript yields any documents
                // from the index for a correction, collateMatch is updated
                final Map<String, Object> vars = suggestion.getCollateScriptParams();
                vars.put(SUGGESTION_TEMPLATE_VAR_NAME, spare.toString());
                QueryShardContext shardContext = suggestion.getShardContext();
                final String querySource = scriptFactory.newInstance(vars).execute();
                try (XContentParser parser = XContentFactory.xContent(querySource).createParser(shardContext.getXContentRegistry(), LoggingDeprecationHandler.INSTANCE, querySource)) {
                    QueryBuilder innerQueryBuilder = AbstractQueryBuilder.parseInnerQueryBuilder(parser);
                    final ParsedQuery parsedQuery = shardContext.toQuery(innerQueryBuilder);
                    collateMatch = Lucene.exists(searcher, parsedQuery.query());
                }
            }
            if (!collateMatch && !collatePrune) {
                continue;
            }
            Text phrase = new Text(spare.toString());
            Text highlighted = null;
            if (suggestion.getPreTag() != null) {
                spare.copyUTF8Bytes(correction.join(SEPARATOR, byteSpare, suggestion.getPreTag(), suggestion.getPostTag()));
                highlighted = new Text(spare.toString());
            }
            if (collatePrune) {
                resultEntry.addOption(new PhraseSuggestion.Entry.Option(phrase, highlighted, (float) (correction.score), collateMatch));
            } else {
                resultEntry.addOption(new PhraseSuggestion.Entry.Option(phrase, highlighted, (float) (correction.score)));
            }
        }
    } else {
        response.addTerm(buildResultEntry(suggestion, spare, Double.MIN_VALUE));
    }
    return response;
}

Example 87 with QueryShardContext

use of org.opensearch.index.query.QueryShardContext in project OpenSearch by opensearch-project.

the class HighlightBuilderTests method testBuildSearchContextHighlight.

/**
 * test that build() outputs a {@link SearchHighlightContext} that is has similar parameters
 * than what we have in the random {@link HighlightBuilder}
 */
public void testBuildSearchContextHighlight() throws IOException {
    Settings indexSettings = Settings.builder().put(IndexMetadata.SETTING_VERSION_CREATED, Version.CURRENT).build();
    Index index = new Index(randomAlphaOfLengthBetween(1, 10), "_na_");
    IndexSettings idxSettings = IndexSettingsModule.newIndexSettings(index, indexSettings);
    // shard context will only need indicesQueriesRegistry for building Query objects nested in highlighter
    QueryShardContext mockShardContext = new QueryShardContext(0, idxSettings, BigArrays.NON_RECYCLING_INSTANCE, null, null, null, null, null, xContentRegistry(), namedWriteableRegistry, null, null, System::currentTimeMillis, null, null, () -> true, null) {

        @Override
        public MappedFieldType fieldMapper(String name) {
            TextFieldMapper.Builder builder = new TextFieldMapper.Builder(name, createDefaultIndexAnalyzers());
            return builder.build(new Mapper.BuilderContext(idxSettings.getSettings(), new ContentPath(1))).fieldType();
        }
    };
    mockShardContext.setMapUnmappedFieldAsString(true);
    for (int runs = 0; runs < NUMBER_OF_TESTBUILDERS; runs++) {
        HighlightBuilder highlightBuilder = randomHighlighterBuilder();
        highlightBuilder = Rewriteable.rewrite(highlightBuilder, mockShardContext);
        SearchHighlightContext highlight = highlightBuilder.build(mockShardContext);
        for (SearchHighlightContext.Field field : highlight.fields()) {
            String encoder = highlightBuilder.encoder() != null ? highlightBuilder.encoder() : HighlightBuilder.DEFAULT_ENCODER;
            assertEquals(encoder, field.fieldOptions().encoder());
            final Field fieldBuilder = getFieldBuilderByName(highlightBuilder, field.field());
            assertNotNull("expected a highlight builder for field " + field.field(), fieldBuilder);
            FieldOptions fieldOptions = field.fieldOptions();
            BiConsumer<Function<AbstractHighlighterBuilder<?>, Object>, Function<FieldOptions, Object>> checkSame = mergeBeforeChek(highlightBuilder, fieldBuilder, fieldOptions);
            checkSame.accept(AbstractHighlighterBuilder::boundaryChars, FieldOptions::boundaryChars);
            checkSame.accept(AbstractHighlighterBuilder::boundaryScannerType, FieldOptions::boundaryScannerType);
            checkSame.accept(AbstractHighlighterBuilder::boundaryMaxScan, FieldOptions::boundaryMaxScan);
            checkSame.accept(AbstractHighlighterBuilder::fragmentSize, FieldOptions::fragmentCharSize);
            checkSame.accept(AbstractHighlighterBuilder::fragmenter, FieldOptions::fragmenter);
            checkSame.accept(AbstractHighlighterBuilder::requireFieldMatch, FieldOptions::requireFieldMatch);
            checkSame.accept(AbstractHighlighterBuilder::noMatchSize, FieldOptions::noMatchSize);
            checkSame.accept(AbstractHighlighterBuilder::numOfFragments, FieldOptions::numberOfFragments);
            checkSame.accept(AbstractHighlighterBuilder::phraseLimit, FieldOptions::phraseLimit);
            checkSame.accept(AbstractHighlighterBuilder::highlighterType, FieldOptions::highlighterType);
            checkSame.accept(AbstractHighlighterBuilder::highlightFilter, FieldOptions::highlightFilter);
            checkSame.accept(AbstractHighlighterBuilder::preTags, FieldOptions::preTags);
            checkSame.accept(AbstractHighlighterBuilder::postTags, FieldOptions::postTags);
            checkSame.accept(AbstractHighlighterBuilder::options, FieldOptions::options);
            checkSame.accept(AbstractHighlighterBuilder::order, op -> op.scoreOrdered() ? Order.SCORE : Order.NONE);
            assertEquals(fieldBuilder.fragmentOffset, fieldOptions.fragmentOffset());
            if (fieldBuilder.matchedFields != null) {
                String[] copy = Arrays.copyOf(fieldBuilder.matchedFields, fieldBuilder.matchedFields.length);
                Arrays.sort(copy);
                assertArrayEquals(copy, new TreeSet<>(fieldOptions.matchedFields()).toArray(new String[fieldOptions.matchedFields().size()]));
            } else {
                assertNull(fieldOptions.matchedFields());
            }
            Query expectedValue = null;
            if (fieldBuilder.highlightQuery != null) {
                expectedValue = Rewriteable.rewrite(fieldBuilder.highlightQuery, mockShardContext).toQuery(mockShardContext);
            } else if (highlightBuilder.highlightQuery != null) {
                expectedValue = Rewriteable.rewrite(highlightBuilder.highlightQuery, mockShardContext).toQuery(mockShardContext);
            }
            assertEquals(expectedValue, fieldOptions.highlightQuery());
        }
    }
}

Example 88 with QueryShardContext

use of org.opensearch.index.query.QueryShardContext in project OpenSearch by opensearch-project.

the class ValuesSourceConfigTests method testUnmappedKeyword.

public void testUnmappedKeyword() throws Exception {
    IndexService indexService = createIndex("index", Settings.EMPTY, "type");
    client().prepareIndex("index").setId("1").setSource().setRefreshPolicy(WriteRequest.RefreshPolicy.IMMEDIATE).get();
    try (Engine.Searcher searcher = indexService.getShard(0).acquireSearcher("test")) {
        QueryShardContext context = indexService.newQueryShardContext(0, searcher, () -> 42L, null);
        ValuesSourceConfig config = ValuesSourceConfig.resolve(context, ValueType.STRING, "bytes", null, null, null, null, CoreValuesSourceType.BYTES);
        ValuesSource.Bytes valuesSource = (ValuesSource.Bytes) config.getValuesSource();
        assertNotNull(valuesSource);
        assertFalse(config.hasValues());
        config = ValuesSourceConfig.resolve(context, ValueType.STRING, "bytes", null, "abc", null, null, CoreValuesSourceType.BYTES);
        valuesSource = (ValuesSource.Bytes) config.getValuesSource();
        LeafReaderContext ctx = searcher.getIndexReader().leaves().get(0);
        SortedBinaryDocValues values = valuesSource.bytesValues(ctx);
        assertTrue(values.advanceExact(0));
        assertEquals(1, values.docValueCount());
        assertEquals(new BytesRef("abc"), values.nextValue());
    }
}

Example 89 with QueryShardContext

use of org.opensearch.index.query.QueryShardContext in project OpenSearch by opensearch-project.

the class ValuesSourceConfigTests method testFieldAlias.

public void testFieldAlias() throws Exception {
    IndexService indexService = createIndex("index", Settings.EMPTY, "type", "field", "type=keyword", "alias", "type=alias,path=field");
    client().prepareIndex("index").setId("1").setSource("field", "value").setRefreshPolicy(WriteRequest.RefreshPolicy.IMMEDIATE).get();
    try (Engine.Searcher searcher = indexService.getShard(0).acquireSearcher("test")) {
        QueryShardContext context = indexService.newQueryShardContext(0, searcher, () -> 42L, null);
        ValuesSourceConfig config = ValuesSourceConfig.resolve(context, ValueType.STRING, "alias", null, null, null, null, CoreValuesSourceType.BYTES);
        ValuesSource.Bytes valuesSource = (ValuesSource.Bytes) config.getValuesSource();
        LeafReaderContext ctx = searcher.getIndexReader().leaves().get(0);
        SortedBinaryDocValues values = valuesSource.bytesValues(ctx);
        assertTrue(values.advanceExact(0));
        assertEquals(1, values.docValueCount());
        assertEquals(new BytesRef("value"), values.nextValue());
    }
}

Example 90 with QueryShardContext

use of org.opensearch.index.query.QueryShardContext in project OpenSearch by opensearch-project.

the class ValuesSourceConfigTests method testUnmappedBoolean.

public void testUnmappedBoolean() throws Exception {
    IndexService indexService = createIndex("index", Settings.EMPTY, "type");
    client().prepareIndex("index").setId("1").setSource().setRefreshPolicy(WriteRequest.RefreshPolicy.IMMEDIATE).get();
    try (Engine.Searcher searcher = indexService.getShard(0).acquireSearcher("test")) {
        QueryShardContext context = indexService.newQueryShardContext(0, searcher, () -> 42L, null);
        ValuesSourceConfig config = ValuesSourceConfig.resolve(context, ValueType.BOOLEAN, "bool", null, null, null, null, CoreValuesSourceType.BYTES);
        ValuesSource.Numeric valuesSource = (ValuesSource.Numeric) config.getValuesSource();
        assertNotNull(valuesSource);
        assertFalse(config.hasValues());
        config = ValuesSourceConfig.resolve(context, ValueType.BOOLEAN, "bool", null, true, null, null, CoreValuesSourceType.BYTES);
        valuesSource = (ValuesSource.Numeric) config.getValuesSource();
        LeafReaderContext ctx = searcher.getIndexReader().leaves().get(0);
        SortedNumericDocValues values = valuesSource.longValues(ctx);
        assertTrue(values.advanceExact(0));
        assertEquals(1, values.docValueCount());
        assertEquals(1, values.nextValue());
    }
}