/**
 * GetURLs2.java
 **/

import java.io.*;
import java.net.*;
import java.util.*;


public class GetURLs2 {
    /** タグのリーダ */
    ReadTag2 reader;

    public GetURLs2(URL theURL) throws IOException {
	reader = new ReadTag2(theURL);
    }

    public GetURLs2(String theURL) throws MalformedURLException, IOException {
	reader = new ReadTag2(theURL);
    }

    /** 抽出対象となるタグ */
    public final static String[] wantTags = { "<a href", "<A HREF","<A href", "<a HREF"};

    public ArrayList getURLs() throws IOException {
	ArrayList al = new ArrayList();
	String tag;
	while ((tag = reader.nextTag()) != null) {
	    for (int i = 0; i < wantTags.length; i++) {
		if (tag.startsWith(wantTags[i]))
		    al.add(tag);
	    }
	}
	return al;
    }



   /** 抽出対象となるtopicタグ */
    public final static String[] wantTopicTags = { "<B>","<I>","<META","<STRONG>","<TITLE>","<b>","<i>","<meta","<strong>","<title>","<H1","<h1","<H2","<h2","<H3","<h3","<H4","<h4","<H5","<h5"};

    public ArrayList getTags() throws IOException {
	ArrayList TopicTags = new ArrayList();
	String tag;

	try{
	    while ((tag = reader.nextTag()) != null) {
		StringBuffer EndTag = new StringBuffer();
		EndTag.append("</");
		for (int i = 0; i < wantTopicTags.length; i++) {

		    if (tag.startsWith(wantTopicTags[i])){
			if(i==2 || i==7){//"<META" or "<meta"の時
			    TopicTags.add(tag);
			    break;
			}

			String tmp_EndTag0 = wantTopicTags[i];
			int length = tmp_EndTag0.length();
			String tmp_EndTag = tmp_EndTag0.substring(1,length);
			EndTag.append(tmp_EndTag);
			String str_EndTag = EndTag.toString();
			String TopicTag = reader.readTopicTag(str_EndTag);
			StringBuffer topictag = new StringBuffer(tag);
			topictag.append(TopicTag);
			topictag.toString();
			//System.out.println("topictag:"+topictag);

			TopicTags.add(topictag);
			break;
		    } //end if
		} // end for

	    } // end while
	}catch(IOException e){
	    System.out.println("getTags Error:"+e);
	}

	return TopicTags;
    }


    public void close() throws IOException {
	if (reader != null)
	    reader.close();
    }


    /*************************************************************/
    public static Maps todo(Maps MAP)
	throws MalformedURLException, IOException {

	int i=-1;
	int RSsize = MAP.RootSetSize;
	System.out.print("\nRootSet");;
	while(RSsize>i+1){
	    try{
		i++;
		Object objurl = MAP.BaseSet.get(i);
		String theURL = objurl.toString();
		System.out.print("["+i+"]");

		if(theURL.startsWith("http://www.nias.affrc.go.jp/")){
		    System.out.println("continue");
		    continue;
		}
		GetURLs2 gu2 = new GetURLs2(theURL);
		ArrayList urls = new ArrayList(gu2.getURLs());
		/*取ってきたアンカーを表示
		System.out.println("***************");
		Tools.Output(urls);
		System.out.println("***************");
		*/

		// 親URLの修正
		String tmp_theURL = Change.Change_theURL(theURL);
		int size = urls.size();
		for(int j=0;j<size;j++){
		    Object obj = urls.get(j);
		    String tmp_url1 = obj.toString();

		    /***<A HREF="???.html">から[???.html]を抜きだす***/
		    int e_index = tmp_url1.indexOf("\"",9);
		    String tmp_url2 = tmp_url1.substring(9,e_index);
		    /********** 抜きだし完了 **********/

		    if(Check.EndCheck(tmp_url2)==-1) //拡張子等判定
			continue;

		    /***[if] 「<a href="http:」で始まらない場合、親URLに子リンクをつなげる***/
		    String href = tmp_url1.substring(0,14);
		    href = href.toLowerCase();
		    if(!href.equals("<a href=\"http:")){
			int cd =0;
			int home=0;
			StringBuffer cd_url = new StringBuffer(tmp_url2);

			try{// 子リンクの修正
			    while(tmp_url2.startsWith(".") || tmp_url2.startsWith("/")){
				if(tmp_url2.startsWith("./")){
				    cd_url.delete(0,2);
				    tmp_url2 = new String(cd_url);
				}else if(tmp_url2.startsWith("../")){
				    cd_url.delete(0,3);
				    tmp_url2 = new String(cd_url);
				    cd ++;
				}else if(tmp_url2.startsWith("/")){
				    cd_url.delete(0,1);
				    tmp_url2 = new String(cd_url);
				    home=1;
				    break;
				}else{
				    System.out.println("cd_else:" + tmp_url2);
				    cd = -1;
				    break;
				}
			    }
			}catch(Exception e){
			    System.out.println("cd_error:" + e);
			    cd =-1;
			}

			StringBuffer tmp_theURL1;
			if(cd==-1)
			    continue;
			if(cd>0 || home>0){// 親URLのディレクトリ修正
			    String cd_theURL = Change.cd_theURL(tmp_theURL,cd,home);
			    tmp_theURL1 = new StringBuffer(cd_theURL);
			}else{
			    tmp_theURL1 = new StringBuffer(tmp_theURL);
			}

			tmp_theURL1.append(tmp_url2);
			String tmp_theURL2 = tmp_theURL1.toString();

			if(Check.LoopCheck(MAP,tmp_theURL2,theURL)==0){
			    MAP.LHMap.put(tmp_theURL2,theURL);
			    if(Check.LoopCheck_BS(MAP,tmp_theURL2)==0)
				MAP.BaseSet.add(tmp_theURL2);
			}else{//同じkeyがある場合→同じ親かチェック
			    if(Check.LoopCheck_Oya(MAP,tmp_theURL2,theURL)==0){
				MAP.DiffOya.add(theURL);
				MAP.DiffOya.add(tmp_theURL2);
			    }
			}
		    } // end if{

		    /***[else] 「<a href="http:」で始まる場合、抜き出したリンクをそのまま追加***/
		    else{
			try{
			    if(Check.LoopCheck(MAP,tmp_url2,theURL)==0){
				MAP.LHMap.put(tmp_url2,theURL);
				if(Check.LoopCheck_BS(MAP,tmp_url2)==0)
				    MAP.BaseSet.add(tmp_url2);
			    }else{
				if(Check.LoopCheck_Oya(MAP,tmp_url2,theURL)==0){
				    MAP.DiffOya.add(theURL);
				    MAP.DiffOya.add(tmp_url2);
				}
			    }
			}catch(Exception e){
			    System.out.println("GetURLs Error:" + e);
			    break;
			}
		    }
		} // end for()
	    }catch(Exception e){
		System.out.print(" Not anchor ");
	    }
	} //end while
	return(MAP);
    }

    /******topictag*************************************/
    public static Maps SearchTopicTags2(Maps MAP)
	throws MalformedURLException, IOException {
	System.out.println("\nSearchTopicTags2");
	try{
	    for(int i=0;i<(MAP.BaseSet.size());i++){
		try{
		    System.out.print(i+" ");
		    String str_i = String.valueOf(i);
		    MAP.TopicTags.add(str_i);

		    Object URL = MAP.BaseSet.get(i);
		    System.out.println(URL);
		    String URLURL = URL.toString();

		    GetURLs2 gu2 = new GetURLs2(URL.toString());
		    ArrayList tmp_TT = gu2.getTags();
		    Iterator urliterator = tmp_TT.iterator();

		    while (urliterator.hasNext()) {
			// System.out.println(urliterator.next());
			MAP.TopicTags.add(urliterator.next());
		    }
		}catch(Exception e){
		    System.out.println("SearchTopicTags2 for() Error:"+e);
		}
		MAP.TopicTags.add("0");
	    }
	}catch(Exception e){
	    System.out.println("SearchTopicTags2 Error:"+e);
	}
	MAP.TopicTags.add("00");
	System.out.println("end");
	return MAP;
    }

}