stringarraydocumentcollection.java

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

package test.it.unimi.dsi.mg4j.document;

import it.unimi.dsi.fastutil.objects.Reference2ObjectMap;
import it.unimi.dsi.fastutil.objects.Reference2ObjectMaps;
import it.unimi.dsi.mg4j.document.AbstractDocumentCollection;
import it.unimi.dsi.mg4j.document.Document;
import it.unimi.dsi.mg4j.document.DocumentCollection;
import it.unimi.dsi.mg4j.document.DocumentFactory;
import it.unimi.dsi.mg4j.document.IdentityDocumentFactory;
import it.unimi.dsi.fastutil.io.FastByteArrayInputStream;

import java.io.IOException;
import java.io.InputStream;

import org.apache.commons.configuration.ConfigurationException;

/** A document collection explicitly defined by a sequence of strings (mainly useful for testing).
 * 
 * <p>Every string in the provided sequence is considered a document, 
 * and the stream returned by the collection is the UTF-8 encoding of the string. By default,
 * the collection uses an {@link IdentityDocumentFactory}, but you can also
 * {@linkplain #StringArrayDocumentCollection(DocumentFactory, String[]) specify your own factory}.
 */

public class StringArrayDocumentCollection extends AbstractDocumentCollection {
	final public String[] document;
	final DocumentFactory factory;

	public StringArrayDocumentCollection( final String... document ) throws ConfigurationException {
		factory = new IdentityDocumentFactory( new String[] { "encoding=UTF-8" } );
		this.document = document;
	}

	public StringArrayDocumentCollection( final DocumentFactory factory, final String... document ) {
		this.factory = factory;
		this.document = document;
	}

	public int size() {
		return document.length;
	}

	public Document document( final int index ) throws IOException {
		return factory.getDocument( stream( index ), metadata( index ) );
	}

	public InputStream stream( final int index ) throws IOException {
		return new FastByteArrayInputStream( document[ index ].getBytes( "UTF-8" ) );
	}

	@SuppressWarnings("unchecked")
	public Reference2ObjectMap<Enum<?>, Object> metadata( int index ) throws IOException {
		return Reference2ObjectMaps.EMPTY_MAP;
	}

	public DocumentCollection copy() {
		return this;
	}

	public DocumentFactory factory() {
		return factory;
	};
}