replaceallurl.java

java写的搜索引擎网络爬虫 java写的搜索引擎网络爬虫 java写的搜索引擎网络爬虫

/*
 * File name : ReplaceAllUrl.java
 * Create Time : 2006-5-23 13:13:07
 * Author : 
 * Descript : 
 * Version : 
 */
package com.snoics.reptile.parse;

import java.util.List;
import java.util.Map;

import com.snoics.base.util.UtilTool;
import com.snoics.reptile.file.CreateBinFile;
import com.snoics.reptile.file.ICreateFile;
import com.snoics.reptile.link.createUrl.BuildUrl;
import com.snoics.reptile.link.createUrl.CreateDownloadUrl;
import com.snoics.reptile.link.createUrl.CreateForbidUrl;
import com.snoics.reptile.link.createUrl.CreateRangeUrl;
import com.snoics.reptile.link.createUrl.CreateUnDownloadUrl;
import com.snoics.reptile.link.createUrl.IBuildUrl;
import com.snoics.reptile.link.createUrl.ICreateUrl;
import com.snoics.reptile.regex.url.IMakeUpUrl;

public class ReplaceAllUrl implements IReplaceAllUrl{
	private IBuildUrl buildUrl=new BuildUrl();
	
	/**
	 * 替换forbid部分的URL
	 * @param makeUpUrl
	 * @param htmlString
	 * @return String
	 */ 
	public String replaceForbidUrl(IMakeUpUrl makeUpUrl,String htmlString){
		IReplaceUrl replaceUrl=new ReplaceUrl();
		List urlList=makeUpUrl.getForbidUrl();
		ICreateUrl createUrl=new CreateForbidUrl();
		createUrl.setParentUrl(makeUpUrl.getParentUrl());
		createUrl.setAllUrl(urlList);
		createUrl.create();
		
		Map urlMap=createUrl.getResolvedUrlMap();
		if((urlMap!=null)&&(urlMap.size()>0)){
			Object[][] objects=UtilTool.getMapKeyValue(urlMap);
			int length=objects.length;
			for(int i=0;i<length;i++){
				String url=(String)objects[i][0];
				String theReplaceUrl=(String)objects[i][1];
				replaceUrl.setHtmlString(htmlString);
				replaceUrl.setUrl(url);
				replaceUrl.setReplaceUrl(theReplaceUrl);
				replaceUrl.setUrlRegexList(makeUpUrl.getUrlRegexList());
				htmlString=replaceUrl.replace();
			}
		}
		return htmlString;
	}
	
	/**
	 * 替换undownload部分的URL
	 * @param makeUpUrl
	 * @param htmlString
	 * @return String
	 */
	public String replaceUnDownloadUrl(IMakeUpUrl makeUpUrl,String htmlString){
		IReplaceUrl replaceUrl=new ReplaceUrl();
		List urlList=makeUpUrl.getUnDownloadUrl();
		ICreateUrl createUrl=new CreateUnDownloadUrl();
		createUrl.setParentUrl(makeUpUrl.getParentUrl());
		createUrl.setAllUrl(urlList);
		createUrl.create();
		
		Map urlMap=createUrl.getResolvedUrlMap();
		if((urlMap!=null)&&(urlMap.size()>0)){
			Object[][] objects=UtilTool.getMapKeyValue(urlMap);
			int length=objects.length;
			for(int i=0;i<length;i++){
				String url=(String)objects[i][0];
				String theReplaceUrl=(String)objects[i][1];
				replaceUrl.setHtmlString(htmlString);
				replaceUrl.setUrl(url);
				replaceUrl.setReplaceUrl(theReplaceUrl);
				replaceUrl.setUrlRegexList(makeUpUrl.getUrlRegexList());
				htmlString=replaceUrl.replace();
			}
		}
		return htmlString;
	}
	
	/**
	 * 替换Range部分的URL
	 * @param makeUpUrl
	 * @param htmlString
	 * @return String
	 */
	public String replaceRangeUrl(IMakeUpUrl makeUpUrl,String htmlString){
		IReplaceUrl replaceUrl=new ReplaceUrl();
		List urlList=makeUpUrl.getRangeUrl();
		ICreateUrl createUrl=new CreateRangeUrl();
		createUrl.setParentUrl(makeUpUrl.getParentUrl());
		createUrl.setAllUrl(urlList);
		createUrl.create();
		
		Map urlMap=createUrl.getRelativizeWithParentUrlMap();
		if((urlMap!=null)&&(urlMap.size()>0)){
			Object[][] objects=UtilTool.getMapKeyValue(urlMap);
			int length=objects.length;
			for(int i=0;i<length;i++){
				String url=(String)objects[i][0];
				String theReplaceUrl=(String)objects[i][1];
				replaceUrl.setHtmlString(htmlString);
				replaceUrl.setUrl(url);
				replaceUrl.setReplaceUrl(theReplaceUrl);
				replaceUrl.setUrlRegexList(makeUpUrl.getUrlRegexList());
				htmlString=replaceUrl.replace();
			}
		}
		return htmlString;
	}
	
	/**
	 * 替换download部分的URL
	 * @param makeUpUrl
	 * @param htmlString
	 * @return String
	 */
	public String replaceDownloadUrl(IMakeUpUrl makeUpUrl,String htmlString){
		ICreateFile createBinFile=new CreateBinFile();
		IReplaceUrl replaceUrl=new ReplaceUrl();
		List urlList=makeUpUrl.getDownloadUrl();
		ICreateUrl createUrl=new CreateDownloadUrl();
		createUrl.setParentUrl(makeUpUrl.getParentUrl());
		createUrl.setAllUrl(urlList);
		createUrl.create();
		
		Map urlMap=createUrl.getRelativizeWithParentUrlMap();
		
		if((urlMap!=null)&&(urlMap.size()>0)){
			Object[][] objects=UtilTool.getMapKeyValue(urlMap);
			int length=objects.length;
			for(int i=0;i<length;i++){
				String url=(String)objects[i][0];
				String theReplaceUrl=(String)objects[i][1];
				replaceUrl.setHtmlString(htmlString);
				replaceUrl.setUrl(url);
				replaceUrl.setReplaceUrl(theReplaceUrl);
				replaceUrl.setUrlRegexList(makeUpUrl.getUrlRegexList());
				htmlString=replaceUrl.replace();
				
				createBinFile.setResolvedFile(buildUrl.buildLocalBinFileResolvedName(url,makeUpUrl.getParentUrl()));
				createBinFile.setRelativizeUrl(buildUrl.buildRelativizeUrl(url,makeUpUrl.getParentUrl()));
				createBinFile.setResolvedUrl(buildUrl.buildResolvedUrl(url,makeUpUrl.getParentUrl()));
				
				createBinFile.create();
			}
		}
		return htmlString;
	}
}