usage.java

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

package searchingEngine.VectorSpaceModel;

import java.util.*;

import java.io.*;

import searchingEngine.VectorSpaceModel.*;
import searchingEngine.booleanModel.*;
import searchingEngine.dataPreprocessing.invertedFile.*;
import searchingEngine.queryExpansion.AssociationMartix;
import searchingEngine.utilites.dataConverter.RawConverter;
import searchingEngine.utilites.sorting.Sort;


public class usage{
	
	public static void main(String argv[]) throws Exception{
		/*
		InvertedFile inv = new InvertedFile(InvertedFile.loadHashTable("invertedFile.dat"));
		String docNameList[] = (String[])RawConverter.loadObject("fileNameList.dat");

		VectorSpaceModel vsm = new VectorSpaceModel(inv,docNameList);
		String[] tester = {"zylfiqarI","aaR"};//,"|"};//, "and", "B"};
		long time = System.nanoTime();
		System.out.println(vsm.retrieveDocs(tester, true));
		System.out.print("time used:" + (System.nanoTime()-time));
		*/
	}
}



	/*
	public static void main(String argv[]) throws Exception{
		InvertedFile invc = new InvertedFile("G:/ir/wpos/X/combined0.dat");
		long startTime = System.nanoTime();
		invc.buildInvertedFile();
		System.out.println("inv builded - time used : " + (double)(System.nanoTime() - startTime)/(1000000000) + " seconds.");
		startTime = System.nanoTime();
		RawConverter.saveObject(invc.getTable(),"invertedFile.dat");
		System.out.println("inv saved - time used : " + (double)(System.nanoTime() - startTime)/(1000000000) + " seconds.");
		startTime = System.nanoTime();
		InvertedFile inv = new InvertedFile(InvertedFile.loadHashTable("invertedFile.dat"));
		System.out.println("inv file loaded - time used : " + (double)(System.nanoTime() - startTime)/(1000000000) + " seconds.");
		
		String docNameList[] = (String[])RawConverter.loadObject("fileNameList.dat");
		String inputA = "aA";
		String input[] = inputA.split(" ");
		System.out.println("input= " + input[0] + " " + input.length);
		System.out.println(inv.getTable().get("aA"));
		BooleanModel booleanModel = new BooleanModel(inv,docNameList);

		long start = System.nanoTime();
		LinkedList<String> retrievedList= booleanModel.retrieveDocs(input);
		long endtime = System.nanoTime() - start;
		System.out.print("Testing on retirval, time used: ");
		System.out.println((double)endtime/(1000000000) + " seconds ");
		System.out.print(retrievedList);
		*/
		//Sqlquery--------------------------
		/*
		System.out.println("SQL:");
		SqlQuery sqlQuery = new SqlQuery();
		long id = 3000;
		try {
		      ResultSet rs = sqlQuery.getRS("Select distinct term from post where fileid < " + id);
		      while (rs.next()) {
		        System.out.print(rs.getString("term")+ "\t");
		      }
		    }
		    
		    catch (SQLException sqle) {
		      sqlQuery.takeDown();
		      throw sqle;
		    }
		   sqlQuery.takeDown();
		    //*/
		//Stemming--------------------------
		/*
		System.out.println("\nStemming:");
		StemRevised stem = new StemRevised();
		stem.main(null);
		//*/
		
		
		//splite--------------------------
		/*
		System.out.println("\nSplite:");
		QueryExtractor getList = new QueryExtractor();
		getList.main(null);
		//*/
		//Save inverted file
		/*
		String filename = "invertedFile.dat";
		VectorSpace vsm = new VectorSpace();
		vsm.saveInvertedFile(filename);
		System.out.println("inverted file save as: " + filename);
		//*/
		 /*LinkedList<PostNode> sortPost;
		 long startTime;
		 long fileSize = 300000;
		 for (int i=0;i<=120;i++) {
			 sortPost = RawConverter.loadPostList("J:/WorkFolder/0506sem2/comp433/project/assign/post.txt",i*fileSize,fileSize);
			 System.out.println("sort - Start time: " + (startTime= System.nanoTime()) );
			 Collections.sort(sortPost,new PostNodeComparator());
			 System.out.println("Sorting time used : " + (System.nanoTime() - startTime));
			 RawConverter.saveListAsText(sortPost,"g:/ir/sortedPost"+i+".txt");
		 }
		 //*/
	/*
		int j = 30000;
		//RawConverter.readPartOfFile("J:/WorkFolder/0506sem2/comp433/project/assign/post.txt","ir/post"+i+".txt",i);
		//RawConverter.readPartOfFile("J:/WorkFolder/0506sem2/comp433/project/assign/file.txt","ir/file"+i+".txt",i);
		Object temp;
		//temp =RawConverter.loadPostList("ir/post"+j+".txt","ir/post0.obj");
		LinkedList<PostNode> temp2 = (LinkedList<PostNode>)RawConverter.loadObject("ir/post0.obj");
		System.out.println("JavaSorting start...");
		long startTime = System.nanoTime();
		Collections.sort(temp2,new PostNodeComparator());
		System.out.println("Sorting time used : " + (System.nanoTime() - startTime));
		String tempString;
		if (temp2.size()!=0) {
		BufferedWriter bw =  new BufferedWriter(new FileWriter("ir/testJavaSortedPost2.txt"));;
		for (int i=0;i<temp2.size();i++){
				tempString = temp2.get(i).toString();
				bw.write(tempString,0,tempString.length());
				bw.newLine();
				
		}
			bw.close();
		}
		System.out.println("file saved");
	/*	
	System.out.println("maxMemory:\t" + Runtime.getRuntime().maxMemory() + "B");
	System.out.println("freeMemory:\t" + Runtime.getRuntime().freeMemory() + "B");
	System.out.println("totalMemory:\t" + Runtime.getRuntime().totalMemory() + "B");
	//*/
				
	///* read inverted file from backing store.
	//InvertedFile inverted_file = InvertedFile.loadHashTable("invertedFile.dat");
	//*/
	/*inverted file building
	InvertedFile inverted_file = new InvertedFile();
	inverted_file.buildInvertedFile("G:/ir","sortedPost","txt",0,0);
	RawConverter.saveListAsText(inverted_file.getInvertedFile(), "invertedFile1.txt");
	System.out.println("Save As Text Done");
	RawConverter.saveObject(inverted_file.getInvertedFile(), "invertedFile1.dat");
	System.out.println("All Done");
	//*/
	//}

//}