聽 聽 聽2.鍐欑▼搴忓垎鏋愶紝棣栧厛鏍规嵁url鑾峰彇div鏂囨湰锛屽啀鏉ユ牴鎹枃鏈鍐呭杩涜鍒嗘瀽

聽

	/**
	 * 鏍规嵁jsoup鏂规硶鑾峰彇htmlContent
         * 鍔犲叆绠€鍗曠殑鏃堕棿璁板綍
	 * @throws IOException 
	 */
	public static String getContentByJsoup(String url){
		String content="";
		try {
			System.out.println("time=====start");
			Date startdate=new Date();
			Document doc=Jsoup.connect(url)
			.data("jquery", "java")
			.userAgent("Mozilla")
			.cookie("auth", "token")
			.timeout(50000)
			.get();
			Date enddate=new Date();
			Long time=enddate.getTime()-startdate.getTime();
			System.out.println("浣跨敤Jsoup鑰楁椂=="+time);
			System.out.println("time=====end");
			content=doc.toString();//鑾峰彇iteye缃戠珯鐨勬簮鐮乭tml鍐呭
			System.out.println(doc.title());//鑾峰彇iteye缃戠珯鐨勬爣棰?
		} catch (IOException e) {
			e.printStackTrace();
		}
		System.out.println(content); 
		return content;
	}

聽

聽 聽 聽3. 鏍规嵁鏁翠釜姣忔棩璧勮鎵€鍦ㄧ殑div灞傦紝鑾峰彇閭ｆ鍐呭锛堢簿纭幏鍙栵級

聽

        /**
	 * 浣跨敤jsoup鏉ュ鏂囨。鍒嗘瀽
         * 鑾峰彇鐩爣鍐呭鎵€鍦ㄧ殑鐩爣灞?
         * 杩欎釜鐩爣灞傚彲浠ユ槸div锛宼able锛宼r绛夌瓑
	 */
	public static String getDivContentByJsoup(String content){
		String divContent="";
		Document doc=Jsoup.parse(content);
		Elements divs=doc.getElementsByClass("main_left");
		divContent=divs.toString();
	      //System.out.println("div==="+divContent);
		return divContent;
	}

聽 聽聽

聽 聽 聽4.鏍规嵁鑾峰彇鐨勭洰鏍囧眰寰楀埌浣犳墍瑕佺殑鍐呭锛坱itle锛寀rl鍦板潃...绛夌瓑锛?/strong>

/**
	 * 浣跨敤jsoup鍒嗘瀽divContent
	 * 1.鑾峰彇閾炬帴 2.鑾峰彇url鍦板潃锛堢粷瀵硅矾寰勶級
	 */
	public static void getLinksByJsoup(String divContent){
		String abs="http://www.iteye.com/";
		Document doc=Jsoup.parse(divContent,abs);
		Elements linkStrs=doc.getElementsByTag("li");
		System.out.println("閾炬帴==="+linkStrs.size());
		for(Element linkStr:linkStrs){
		    String url=linkStr.getElementsByTag("a").attr("abs:href");
		    String title=linkStr.getElementsByTag("a").text();
		    System.out.println("鏍囬:"+title+" url:"+url);
		}
	}

/**
	 * @method 娴嬭瘯鑾峰彇鍐呭绋嬪簭
	 */
	public static void main(String[] args) throws IOException {
		
		/**
		 * 鎵ц鍒嗘瀽绋嬪簭
		 */
		String url="http://www.iteye.com/";
		String HtmlContent=getContentByJsoup(url);
		String divContent=getDivContentByJsoup(HtmlContent);
		getLinksByJsoup(divContent);
	}