blob: 8f8c5f3bf4a397335050a0975beb9308e09c1fcc [file] [log] [blame]
package edu.uci.ics.asterix.runtime.evaluators.functions;
import edu.uci.ics.asterix.common.functions.FunctionConstants;
import edu.uci.ics.asterix.om.functions.IFunctionDescriptor;
import edu.uci.ics.asterix.om.functions.IFunctionDescriptorFactory;
import edu.uci.ics.asterix.om.types.ATypeTag;
import edu.uci.ics.asterix.om.types.BuiltinType;
import edu.uci.ics.asterix.runtime.evaluators.base.AbstractScalarFunctionDynamicDescriptor;
import edu.uci.ics.asterix.runtime.evaluators.common.WordTokensEvaluator;
import edu.uci.ics.hyracks.algebricks.common.exceptions.AlgebricksException;
import edu.uci.ics.hyracks.algebricks.core.algebra.functions.FunctionIdentifier;
import edu.uci.ics.hyracks.algebricks.runtime.base.ICopyEvaluator;
import edu.uci.ics.hyracks.algebricks.runtime.base.ICopyEvaluatorFactory;
import edu.uci.ics.hyracks.dataflow.common.data.accessors.IDataOutputProvider;
import edu.uci.ics.hyracks.storage.am.invertedindex.tokenizers.DelimitedUTF8StringBinaryTokenizer;
import edu.uci.ics.hyracks.storage.am.invertedindex.tokenizers.IBinaryTokenizer;
import edu.uci.ics.hyracks.storage.am.invertedindex.tokenizers.ITokenFactory;
import edu.uci.ics.hyracks.storage.am.invertedindex.tokenizers.UTF8WordTokenFactory;
public class WordTokensDescriptor extends AbstractScalarFunctionDynamicDescriptor {
private static final long serialVersionUID = 1L;
private final static FunctionIdentifier FID = new FunctionIdentifier(FunctionConstants.ASTERIX_NS, "word-tokens", 1);
public static final IFunctionDescriptorFactory FACTORY = new IFunctionDescriptorFactory() {
public IFunctionDescriptor createFunctionDescriptor() {
return new WordTokensDescriptor();
}
};
@Override
public FunctionIdentifier getIdentifier() {
return FID;
}
@Override
public ICopyEvaluatorFactory createEvaluatorFactory(final ICopyEvaluatorFactory[] args) throws AlgebricksException {
return new ICopyEvaluatorFactory() {
private static final long serialVersionUID = 1L;
@Override
public ICopyEvaluator createEvaluator(IDataOutputProvider output) throws AlgebricksException {
ITokenFactory tokenFactory = new UTF8WordTokenFactory(ATypeTag.STRING.serialize(),
ATypeTag.INT32.serialize());
IBinaryTokenizer tokenizer = new DelimitedUTF8StringBinaryTokenizer(true, true, tokenFactory);
return new WordTokensEvaluator(args, output, tokenizer, BuiltinType.ASTRING);
}
};
}
}