concatenate.java

MG4J (Managing Gigabytes for Java) is a free full-text search engine for large document collections

package it.unimi.dsi.mg4j.tool;

/*		 
 * MG4J: Managing Gigabytes for Java
 *
 * Copyright (C) 2005-2007 Sebastiano Vigna 
 *
 *  This library is free software; you can redistribute it and/or modify it
 *  under the terms of the GNU Lesser General Public License as published by the Free
 *  Software Foundation; either version 2.1 of the License, or (at your option)
 *  any later version.
 *
 *  This library is distributed in the hope that it will be useful, but
 *  WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY
 *  or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU Lesser General Public License
 *  for more details.
 *
 *  You should have received a copy of the GNU Lesser General Public License
 *  along with this program; if not, write to the Free Software
 *  Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA.
 *
 */

import it.unimi.dsi.fastutil.ints.IntIterator;
import it.unimi.dsi.mg4j.index.Index;
import it.unimi.dsi.mg4j.index.IndexIterator;
import it.unimi.dsi.mg4j.index.CompressionFlags.Coding;
import it.unimi.dsi.mg4j.index.CompressionFlags.Component;
import it.unimi.dsi.io.OutputBitStream;

import java.io.Closeable;
import java.io.IOException;
import java.lang.reflect.InvocationTargetException;
import java.net.URISyntaxException;
import java.util.Map;

import org.apache.commons.configuration.ConfigurationException;

import com.martiansoftware.jsap.JSAPException;

/** Concatenates several indices.
 * 
 * <p>This implementation of {@link it.unimi.dsi.mg4j.tool.Combine} concatenates
 * the involved indices: document 0 of the first index is document 0 of the 
 * final collection, but document 0 of the second index is numbered after
 * the number of documents in the first index, and so on. The resulting
 * index is exactly what you would obtain by concatenating the document
 * sequences at the origin of each index.
 * 
 * <p>Note that this class can be used also with a single index, making it possible to recompress easily
 * an index using different compression flags.
 * 
 * @author Sebastiano Vigna
 * @since 1.0
 * 
 */

final public class Concatenate extends Combine {
	
	public Concatenate( final String outputBasename,
			final String[] inputBasename,
			final boolean metadataOnly,
			final int bufferSize,
			final Map<Component,Coding> writerFlags,
			final boolean interleaved,
			final boolean skips,
			final int quantum,
			final int height,
			final int skipBufferSize,
			final long logInterval ) throws IOException, ConfigurationException, URISyntaxException, ClassNotFoundException, SecurityException, InstantiationException, IllegalAccessException, InvocationTargetException, NoSuchMethodException {
		super( outputBasename, inputBasename, metadataOnly, bufferSize, writerFlags, interleaved, skips, quantum, height, skipBufferSize, logInterval );
	}

	protected int combineNumberOfDocuments() {
		int n = 0;
		for( int i = 0; i < numIndices; i++ ) n += index[ i ].numberOfDocuments;
		return n;
	}
	
	
	protected int combineSizes() throws IOException {
		int currDoc = 0, maxDocSize = 0;
		for( int i = 0; i < numIndices; i++ ) {
			final IntIterator sizes = sizes( i );
			int s = 0;
			int j = index[ i ].numberOfDocuments;
			while( j-- != 0 ) {
				s = ( size[ currDoc++ ] += sizes.nextInt() );
				if ( s > maxDocSize ) maxDocSize = s;
			}
			if ( sizes instanceof Closeable ) ((Closeable)sizes).close();
		}
		return maxDocSize;
	}

	protected int combine( final int numUsedIndices ) throws IOException {
		int currIndex, numPrevDocs = 0, currDoc, count, totalFrequency;
		OutputBitStream obs;
		Index i;
		IndexIterator ii;

		// We gather the frequencies from the subindices and just add up.
		totalFrequency = 0;
		for( int k = numUsedIndices; k-- != 0; ) 
			totalFrequency += ( frequency[ usedIndex[ k ] ] = indexIterator[ usedIndex[ k ] ].frequency() );
		
		indexWriter.newInvertedList();
		indexWriter.writeFrequency( totalFrequency );

		for( int k = currIndex = 0; k < numUsedIndices; k++ ) { // We can just concatenated posting lists.
			
			// We must update the number of previously seen documents, possibly adding those in skipped indices.
			while( currIndex < usedIndex[ k ] ) numPrevDocs += index[ currIndex++ ].numberOfDocuments;
			
			i = index[ currIndex ];
			ii = indexIterator[ currIndex ];

			for( int j = frequency[ currIndex ]; j-- != 0; ) {
				obs = indexWriter.newDocumentRecord();
				currDoc = ii.nextDocument() + numPrevDocs;
				indexWriter.writeDocumentPointer( obs, currDoc );
				
				if ( i.hasPayloads ) indexWriter.writePayload( obs, ii.payload() );
				
				if ( i.hasCounts ) {
					count = ii.count();
					if ( hasCounts ) indexWriter.writePositionCount( obs, count );
					if ( i.hasPositions && hasPositions ) indexWriter.writeDocumentPositions( obs, ii.positionArray(), 0, count, size != null ? size[ currDoc ] : -1 );
				}		
			}
		}
		
		return totalFrequency;
	}
	
	public static void main( String arg[] ) throws ConfigurationException, SecurityException, JSAPException, IOException, URISyntaxException, ClassNotFoundException, InstantiationException, IllegalAccessException, InvocationTargetException, NoSuchMethodException {
		Combine.main( arg, Concatenate.class );
	}

}