經過JAVA獲取優酷、土豆、酷六、6間房等視頻

經過JAVA獲取優酷、土豆、酷六、6間房、56視頻,如今不少社會網站都有這個功能,用戶輸入優酷、土豆、酷六、6間房、56視頻地址後,能找到對應的視頻及視頻的縮略圖,有些社區網站還能獲取到視頻的時長。
好比:新浪微博就有這個功能,當用戶輸入優酷、土豆、酷六、6間房、56視頻網址後,就能獲取到相應的視頻地址及視頻的縮略圖。

代碼以下:
import org.jsoup.Jsoup;
import org.jsoup.nodes.Document;
import org.jsoup.nodes.Element;
import org.jsoup.select.Elements;

/**
* 視頻工具類
* @author sunlightcs
* 2011-4-6
* http://hi.juziku.com/sunlightcs/
*/
public class VideoUtil {
	
	/**
	 * 獲取視頻信息
	 * @param url
	 * @return
	 */
	public static Video getVideoInfo(String url){
		Video video = new Video();
		
		if(url.indexOf("v.youku.com")!=-1){
			try {
				video = getYouKuVideo(url);
			} catch (Exception e) {
				video = null;
			}
		}else if(url.indexOf("tudou.com")!=-1){
			try {
				video = getTudouVideo(url);
			} catch (Exception e) {
				video = null;
			}
		}else if(url.indexOf("v.ku6.com")!=-1){
			try {
				video = getKu6Video(url);
			} catch (Exception e) {
				video = null;
			}
		}else if(url.indexOf("6.cn")!=-1){
			try {
				video = get6Video(url);
			} catch (Exception e) {
				video = null;
			}
		}else if(url.indexOf("56.com")!=-1){
			try {
				video = get56Video(url);
			} catch (Exception e) {
				video = null;
			}
		}
		
		return video;
	}
	
	
	/**
	 * 獲取優酷視頻
	 * @param url  視頻URL
	 */
	public static Video getYouKuVideo(String url) throws Exception{
		Document doc = getURLContent(url);
		
		/**
		 *獲取視頻縮略圖 
		 */
		String pic = getElementAttrById(doc, "s_sina", "href");
		int local = pic.indexOf("pic=");
		pic = pic.substring(local+4);
		
		/**
		 * 獲取視頻地址
		 */		
		String flash = getElementAttrById(doc, "link2", "value");
		
		/**
		 * 獲取視頻時間
		 */	
		String time = getElementAttrById(doc, "download", "href");
		String []arrays = time.split("\\|");
		time = arrays[4];
		
		Video video = new Video();
		video.setPic(pic);
		video.setFlash(flash);
		video.setTime(time);
		
		return video;
	}
	
	
	/**
	 * 獲取土豆視頻
	 * @param url  視頻URL
	 */
	public static Video getTudouVideo(String url) throws Exception{
		Document doc = getURLContent(url);
		String content = doc.html();
		int beginLocal = content.indexOf("<script>document.domain");
		int endLocal = content.indexOf("</script>");
		content = content.substring(beginLocal, endLocal);
		
		/**
		 * 獲取視頻地址
		 */	
		String flash = getScriptVarByName("iid_code", content);
		flash = "http://www.tudou.com/v/" + flash + "/v.swf";
		
		/**
		 *獲取視頻縮略圖 
		 */
		String pic = getScriptVarByName("thumbnail", content);
		
		/**
		 * 獲取視頻時間
		 */	
		String time = getScriptVarByName("time", content);

		Video video = new Video();
		video.setPic(pic);
		video.setFlash(flash);
		video.setTime(time);
		
		return video;
	}
	
	
	/**
	 * 獲取酷6視頻
	 * @param url  視頻URL
	 */
	public static Video getKu6Video(String url) throws Exception{
		Document doc = getURLContent(url);
		
		/**
		 * 獲取視頻地址
		 */
		Element flashEt = doc.getElementById("outSideSwfCode");
		String flash = flashEt.attr("value");
		
		/**
		 * 獲取視頻縮略圖
		 */
		Element picEt = doc.getElementById("plVideosList");
		String time = null;
		String pic = null;
		if(picEt!=null){
			Elements pics = picEt.getElementsByTag("img");
			pic = pics.get(0).attr("src");
			
			/**
			 * 獲取視頻時長
			 */
			Element timeEt = picEt.select("span.review>cite").first(); 
			time = timeEt.text();
		}else{
			pic = doc.getElementsByClass("s_pic").first().text();
		}
		
		Video video = new Video();
		video.setPic(pic);
		video.setFlash(flash);
		video.setTime(time);
		
		return video;
		
	}
	
	
	/**
	 * 獲取6間房視頻
	 * @param url  視頻URL
	 */
	public static Video get6Video(String url) throws Exception{
		Document doc = getURLContent(url);
		
		/**
		 * 獲取視頻縮略圖
		 */
		Element picEt = doc.getElementsByClass("summary").first();
		String pic = picEt.getElementsByTag("img").first().attr("src");
		
		/**
		 * 獲取視頻時長
		 */
		String time = getVideoTime(doc, url, "watchUserVideo");
		if(time==null){
			time = getVideoTime(doc, url, "watchRelVideo");
		}
		
		/**
		 * 獲取視頻地址
		 */
		Element flashEt = doc.getElementById("video-share-code");
		doc = Jsoup.parse(flashEt.attr("value"));  
		String flash = doc.select("embed").attr("src");
		
		Video video = new Video();
		video.setPic(pic);
		video.setFlash(flash);
		video.setTime(time);
		
		return video;
	}
	
	
	/**
	 * 獲取56視頻
	 * @param url  視頻URL
	 */
	public static Video get56Video(String url) throws Exception{
		Document doc = getURLContent(url);
		String content = doc.html();
		
		/**
		 * 獲取視頻縮略圖
		 */
		int begin = content.indexOf("\"img\":\"");
		content = content.substring(begin+7, begin+200);
		int end = content.indexOf("\"};");
		String pic = content.substring(0, end).trim();
		pic = pic.replaceAll("\\\\", "");		
		
		/**
		 * 獲取視頻地址
		 */
		String flash = "http://player.56.com" + url.substring(url.lastIndexOf("/"), url.lastIndexOf(".html")) + ".swf";
		
		Video video = new Video();
		video.setPic(pic);
		video.setFlash(flash);
		
		return video;
	}

	/**
	 * 獲取6間房視頻時長    
	 */
	private static String getVideoTime(Document doc, String url, String id) {
		String time = null;
		
		Element timeEt = doc.getElementById(id); 
		Elements links = timeEt.select("dt > a");
		
		
		for (Element link : links) {
		  String linkHref = link.attr("href");
		  if(linkHref.equalsIgnoreCase(url)){
			  time = link.parent().getElementsByTag("em").first().text();
			  break;
		  }
		}
		return time;
	}
	
			
	/**
	 * 獲取script某個變量的值
	 * @param name  變量名稱
	 * @return   返回獲取的值 
	 */
	private static String getScriptVarByName(String name, String content){
		String script = content;
		
		int begin = script.indexOf(name);
		
		script = script.substring(begin+name.length()+2);
		
		int end = script.indexOf(",");
		
		script = script.substring(0,end);
		
		String result=script.replaceAll("'", "");
		result = result.trim();
		
		return result;
	}
	
	
	/**
	 * 根據HTML的ID鍵及屬於名,獲取屬於值
	 * @param id  HTML的ID鍵
	 * @param attrName  屬於名
	 * @return  返回屬性值
	 */
	private static String getElementAttrById(Document doc, String id, String attrName)throws Exception{
		Element et = doc.getElementById(id);
		String attrValue = et.attr(attrName);
		
		return attrValue;
	}
	
	
	
	/**
	 * 獲取網頁的內容
	 */
	private static Document getURLContent(String url) throws Exception{
		Document doc = Jsoup.connect(url)
		  .data("query", "Java")
		  .userAgent("Mozilla")
		  .cookie("auth", "token")
		  .timeout(6000)
		  .post();
		return doc;
	}
	
	
	public static void main(String[] args) {
		//String url = "http://v.youku.com/v_show/id_XMjU0MjI2NzY0.html";
		//String url = "http://www.tudou.com/programs/view/pVploWOtCQM/";
		//String url = "http://v.ku6.com/special/show_4024167/9t7p64bisV2A31Hz.html";
		//String url = "http://v.ku6.com/show/BpP5LeyVwvikbT1F.html";
		//String url = "http://6.cn/watch/14757577.html";
		String url = "http://www.56.com/u64/v_NTkzMDEzMTc.html";
		Video video = getVideoInfo(url);
		System.out.println("視頻縮略圖:"+video.getPic());
		System.out.println("視頻地址:"+video.getFlash());
		System.out.println("視頻時長:"+video.getTime());
	}
}
 
視頻封裝類
/**
* 視頻封裝
* @author sunlightcs
* 2011-4-6
* http://hi.juziku.com/sunlightcs/
*/
public class Video {
	private String flash;
	private String pic;
	private String time;
	public String getFlash() {
		return flash;
	}
	public void setFlash(String flash) {
		this.flash = flash;
	}
	public String getPic() {
		return pic;
	}
	public void setPic(String pic) {
		this.pic = pic;
	}
	public String getTime() {
		return time;
	}
	public void setTime(String time) {
		this.time = time;
	}
}
其中要用到一個JAR包,連接地址以下: http://jsoup.org/packages/jsoup-1.5.2.jar
相關文章
相關標籤/搜索