Examples with HashedUTF8NGramTokenFactory org.apache.hyracks.storage.am.lsm.invertedindex.tokenizers.HashedUTF8NGramTokenFactory used on opensource projects

Example 1 with HashedUTF8NGramTokenFactory

use of org.apache.hyracks.storage.am.lsm.invertedindex.tokenizers.HashedUTF8NGramTokenFactory in project asterixdb by apache.

the class HashedGramTokensDescriptor method createEvaluatorFactory.

@Override
public IScalarEvaluatorFactory createEvaluatorFactory(final IScalarEvaluatorFactory[] args) {
    return new IScalarEvaluatorFactory() {

        private static final long serialVersionUID = 1L;

        @Override
        public IScalarEvaluator createScalarEvaluator(IHyracksTaskContext ctx) throws HyracksDataException {
            ITokenFactory tokenFactory = new HashedUTF8NGramTokenFactory();
            NGramUTF8StringBinaryTokenizer tokenizer = new NGramUTF8StringBinaryTokenizer(3, true, true, true, tokenFactory);
            return new GramTokensEvaluator(args, ctx, tokenizer, BuiltinType.AINT32);
        }
    };
}

Example 2 with HashedUTF8NGramTokenFactory

use of org.apache.hyracks.storage.am.lsm.invertedindex.tokenizers.HashedUTF8NGramTokenFactory in project asterixdb by apache.

the class LSMInvertedIndexTestUtils method createHashedNGramInvIndexTestContext.

public static LSMInvertedIndexTestContext createHashedNGramInvIndexTestContext(LSMInvertedIndexTestHarness harness, InvertedIndexType invIndexType) throws IOException, HyracksDataException {
    ISerializerDeserializer[] fieldSerdes = getHashedIndexFieldSerdes(invIndexType);
    ITokenFactory tokenFactory = new HashedUTF8NGramTokenFactory();
    IBinaryTokenizerFactory tokenizerFactory = new NGramUTF8StringBinaryTokenizerFactory(TEST_GRAM_LENGTH, true, true, false, tokenFactory);
    LSMInvertedIndexTestContext testCtx = LSMInvertedIndexTestContext.create(harness, fieldSerdes, fieldSerdes.length - 1, tokenizerFactory, invIndexType, null, null, null, null, null, null);
    return testCtx;
}

Example 3 with HashedUTF8NGramTokenFactory

use of org.apache.hyracks.storage.am.lsm.invertedindex.tokenizers.HashedUTF8NGramTokenFactory in project asterixdb by apache.

the class CountHashedGramTokensDescriptor method createEvaluatorFactory.

@Override
public IScalarEvaluatorFactory createEvaluatorFactory(final IScalarEvaluatorFactory[] args) throws AlgebricksException {
    return new IScalarEvaluatorFactory() {

        private static final long serialVersionUID = 1L;

        @Override
        public IScalarEvaluator createScalarEvaluator(IHyracksTaskContext ctx) throws HyracksDataException {
            ITokenFactory tokenFactory = new HashedUTF8NGramTokenFactory();
            NGramUTF8StringBinaryTokenizer tokenizer = new NGramUTF8StringBinaryTokenizer(3, true, false, true, tokenFactory);
            return new GramTokensEvaluator(args, ctx, tokenizer, BuiltinType.AINT32);
        }
    };
}