codefilter.java

一个Mp3在线搜索器

/*
 * To change this template, choose Tools | Templates
 * and open the template in the editor.
 */

package biz.tbuy.huliqing.jloading.ext.mp3;

import biz.tbuy.huliqing.share.StringFilter;
import java.io.IOException;
import java.net.URLEncoder;

/**
 *
 * @author huliqing
 */
public class CodeFilter {
    
    public MusicModel filter(String htmlCode, String encode) {
        // 分析第一页面数据
        String url = parseURL(htmlCode);
        String name = parseName(htmlCode);
        String size = parseSize(htmlCode);

        // 分析出最踪的MP3地址
        try {
            if (encode != null) {
                url = url.replaceAll(name, URLEncoder.encode(name, encode));
            } else {
                url = url.replaceAll(name, URLEncoder.encode(name, "UTF-8"));
            }
            String code = StringFilter.getHtmlCode(url);
            url = getMp3Address(code);
        } catch (IOException e) {
            System.out.println("无法访问URL");
        }
        return new MusicModel(name, url, size);
    }
    
    public MusicModel filter(String htmlCode) {
        return filter(htmlCode, null);
    }

    private String parseURL(String str) {
        int start = str.indexOf("href=") + 6;
        String temp = str.substring(start);
        return temp.substring(0, temp.indexOf("\""));
    }

    private String parseName(String str) {
        int start = str.indexOf("blank\">") + 7;
        int end = str.indexOf("</a>");
        String name = str.substring(start, end);
        return StringFilter.clearHtml(name);
    }

    private String parseSize(String str) {
        //System.out.println("====" + str);
        String[] arr = str.split("<td");
        String temp = arr[8];
        String size = temp.substring(1, temp.indexOf("</td")); // 从1开始 <td>
        //System.out.println("size======" + size);
        return size;
    }

    private String getMp3Address(String code) {
        int start = code.indexOf("<body");
        if (start != -1) {
            code = code.substring(start);
            start = code.indexOf("href=") + 6;
            code = code.substring(start);
            code = code.substring(0, code.indexOf("\""));
        }
        return code;
    }
}