combinedocnode.java

自己写的search engine, 有 boolean search, fuzzy search

package searchingEngine.dataPreprocessing.wordPosition;

import java.io.IOException;
import java.util.LinkedList;

import searchingEngine.dataPreprocessing.invertedFile.DocNode;
import searchingEngine.dataPreprocessing.invertedFile.TermNode;
import searchingEngine.utilites.dataConverter.RawConverter;

public class CombineDocNode {

	/**
	 * @param args
	 */
	public void combine(LinkedList<TermNode> host,LinkedList<TermNode> sub){
		int indexHost=0,indexSub = 0;
		int hostLen=host.size(), subLen = sub.size();
		while ((indexHost<hostLen)||(indexSub<subLen)){
			if (indexHost>=hostLen) {
				host.addAll(sub.subList(indexSub,subLen));
				break;
			}
			if (indexSub>=subLen) {
				break;
			}
			int checker = host.get(indexHost).compareTo(sub.get(indexSub));
			if ( checker ==0) {
				insert(host.get(indexHost).doc_list,sub.get(indexSub).doc_list);
				indexHost++;
				indexSub++;
			} else if ( checker <0) {
				indexHost++;
			} else if ( checker >0) {
				host.add(indexHost,sub.get(indexSub));
				indexSub++;
			}
		}
	}
		
	private void insert(LinkedList<DocNode> host,LinkedList<DocNode> sub){
		int indexHost=0,indexSub = 0;
		int hostLen=host.size(), subLen = sub.size();
		while ((indexHost<hostLen)||(indexSub<subLen)){
			if (indexHost>=hostLen) {
				host.addAll(sub.subList(indexSub,subLen));
				break;
			}
			if (indexSub>=subLen) {
				break;
			}
			int checker = host.get(indexHost).compareTo(sub.get(indexSub));
			if ( checker ==0) {
				insertInt(((DocNodeWpos)host.get(indexHost)).wpos_list,((DocNodeWpos)sub.get(indexSub)).wpos_list);
				indexHost++;
				indexSub++;
			} else if ( checker <0) {
				indexHost++;
			} else if ( checker >0) {
				host.add(indexHost,sub.get(indexSub));
				indexSub++;
			}
		}
	}
	
	private void insertInt(LinkedList<Integer> host,LinkedList<Integer> sub){
		int indexHost=0,indexSub = 0;
		int hostLen=host.size(), subLen = sub.size();
		while ((indexHost<hostLen)||(indexSub<subLen)){
			if (indexHost>=hostLen) {
				host.addAll(sub.subList(indexSub,subLen));
				break;
			}
			if (indexSub>subLen) {
				break;
			}
			int checker = host.get(indexHost).compareTo(sub.get(indexSub));
			if ( checker <0) {
				indexHost++;
			} else if ( checker >0) {
				host.add(indexHost,sub.get(indexSub));
				indexSub++;
			}
		}
	}
	
	public static void main(String[] args) throws IOException{
		/*
		LinkedList<TermNode> nodeA;
		LinkedList<TermNode> nodeB;
		CombineDocNode combine;
		for (int i=0;i<118;i++){
			nodeA = (LinkedList<TermNode>)RawConverter.loadObject("G:/X/X/combined"+i+".dat");
			nodeB = (LinkedList<TermNode>)RawConverter.loadObject("G:/X/X/combined"+(i+8)+".dat");
			combine = new CombineDocNode();
			combine.combine(nodeA,nodeB);
			RawConverter.saveListAsText(nodeA,"G:/X/X/X/combined"+i+".txt");
			RawConverter.saveObject(nodeA,"G:/X/X/X/combined"+i+".dat");
			i = i+15;
			combine = null;
			nodeA = null;
			nodeB = null;
			System.gc();
		}
		*/
		LinkedList<TermNode> nodeB;
		nodeB = (LinkedList<TermNode>)RawConverter.loadObject("combinedXX96.dat");
		RawConverter.saveListAsText(nodeB,"combinedXX96.txt");
	}

}