仿微博抓取视频网站，支持优酷/土豆/酷6/六间房/56网/音悦台/搜狐/奇艺的视频发布的java实现
视频分享效果为在站点内输入一视频网站某一视频链接，分享后即可类似微博或qq空间中那样直接点击播放，此代码赞只支持上述视频网站，6间房的视频有点问题，后面将继续扩展抓取其他视频网站。
分享后效果图：
实现代码如下：
package com;
/**
* 视频实体
* @author hanfei
*
*/
public class Video {
private String title;
private String desc;
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;
}
public void setTitle(String title) {
this.title = title;
}
public String getTitle() {
return title;
}
public void setDesc(String desc) {
this.desc = desc;
}
public String getDesc() {
return desc;
}
}
视频工具类：
package com;
/**
* 视频抓取工具类
* @author hanfei
*
*/
public class VideoUtil {
/**
*
* 获取视频信息
*
* @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) {
System.out.println(e);
video = null;
}
} else if (url.indexOf("tudou.com") != -1) {
try {
video = getTudouVideo(url);
} catch (Exception e) {
System.out.println("视频为空。。。。。");
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;
}
}
else if (url.indexOf("qiyi.com") != -1) {
try {
System.out.println("执行奇异网");
video = getQiyiVideo(url);
} catch (Exception e) {
System.out.println(e);
video = null;
}
}
return video;
}
/**
*
* 获取优酷视频
*
* @param url
* 视频URL
*/
public static Video getYouKuVideo(String url) throws Exception {
Document doc = getURLContent(url);
/**
*
* 获取视频缩略图
*/
String pic = getElementAttrById(doc, "s_sina", "href");
int local = pic.indexOf("pic=");
System.out.println(pic);
pic = pic.substring(local + 4);
/**
*
* 获取视频地址
*/
String flash = getElementAttrById(doc, "link2", "value");
/**
*
* 获取视频时间
*/
String time = getElementAttrById(doc, "download", "_href");
String[] arrays = time.split("\\|");
System.out.println("time" + time);
System.out.println(arrays.length);
time = arrays[4];
Video video = new Video();
video.setPic(pic);
video.setFlash(flash);
video.setTime(time);
return video;
}
/**
*
* 获取土豆视频
*
* @param url
* 视频URL
*/
public static Video getTudouVideo(String url) throws Exception {
// http://www.tudou.com/v/102387574/v.swf
Document doc = getURLContent(url);
String content = doc.html();
if (url.endsWith(".html")) {
System.out.println("执行htmol");
int beginLocal = content.indexOf("<script>var");
int endLocal = content.indexOf("</script>");
content = content.substring(beginLocal, endLocal);
// System.out.println(content);
//
// ScriptEngineManager mgr = new ScriptEngineManager();
// ScriptEngine engine = mgr.getEngineByName("JavaScript");
// engine.eval(content);
// Invocable inv = (Invocable) engine;
// inv.
/**
*
* 获取视频地址
*/
String flash = getScriptVarByName("defaultIid", content);
flash = "http://js.tudouui.com/bin/player2/opn_22.swf?iid=" + flash;
System.out.print("flashURL" + flash);
/**
*
* 获取视频缩略图
*/
String pic = getScriptVarByName("thumbnail", content);
/**
*
* 获取视频时间
*/
String time = getScriptVarByName("time", content);
Video video = new Video();
video.setPic(pic);
video.setFlash(flash);
video.setTime(time);
return video;
} else {
int beginLocal = content.indexOf("<script>document.domain");
int endLocal = content.indexOf("</script>");
content = content.substring(beginLocal, endLocal);
/**
*
* 获取视频地址
*/
String flash = getScriptVarByName("iid", content);
flash = "http://js.tudouui.com/bin/player2/opn_22.swf?iid=" + flash;
/**
*
* 获取视频缩略图
*/
String pic = getScriptVarByName("thumbnail", content);
/**
*
* 获取视频时间
*/
String time = getScriptVarByName("time", content);
Video video = new Video();
video.setPic(pic);
video.setFlash(flash);
video.setTime(time);
return video;
}
}
/**
*
* 获取酷6视频
*
* @param url
* 视频URL
*/
public static Video getKu6Video(String url) throws Exception {
Document doc = getURLContent(url);
/**
*
* 获取视频地址
*/
Element flashEt = doc.getElementById("outSideSwfCode");
String flash = flashEt.attr("value");
/**
*
* 获取视频缩略图
*/
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");
/**
*
* 获取视频时长
*/
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;
}
/**
*
* 获取6间房视频
*
* @param url
* 视频URL
*/
public static Video get6Video(String url) throws Exception {
Document doc = getURLContent(url);
/**
*
* 获取视频缩略图
*/
Element picEt = doc.getElementsByClass("summary").first();
String pic = picEt.getElementsByTag("img").first().attr("src");
/**
*
* 获取视频时长
*/
String time = getVideoTime(doc, url, "watchUserVideo");
if (time == null) {
time = getVideoTime(doc, url, "watchRelVideo");
}
/**
*
* 获取视频地址
*/
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;
}
/**
*
* 获取56视频
*
* @param url
* 视频URL
*/
public static Video get56Video(String url) throws Exception {
Document doc = getURLContent(url);
String content = doc.html();
/**
*
* 获取视频缩略图
*/
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("\\\\", "");
/**
*
* 获取视频地址
*/
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;
}
public static Video getQiyiVideo(String url) throws Exception {
Document doc = getURLContent(url);
String content = doc.html();
System.out.println("执行htmol");
int beginLocal = content.indexOf("<script");
int endLocal = content.indexOf("</script>");
content = content.substring(beginLocal, endLocal);
System.out.println(content);
String infoString = content.substring(content.indexOf("{"),
content.length());
System.out.println(infoString);
JSONObject infoOBJ = JSONObject.fromObject(infoString);
String videoId = infoOBJ.getString("videoId");
String title = infoOBJ.getString("title");
// http://www.qiyipic.com/thumb/20111013/v134588.jpg
String tvId = infoOBJ.getString("tvId");
String vurl = infoOBJ.getString("url");
vurl = vurl.replace("http://www.", "");
vurl = vurl.replace("http://", "");
// vurl=vurl.replace(":", "");
vurl = vurl.replace("/", ".");
vurl = vurl.replace(".", ",");
String[] urldata = vurl.split(",");
System.out.println(vurl);
String image = "http://www.qiyipic.com/thumb/" + urldata[3] + "/v"
+ tvId + ".jpg";
/**
*
* 获取视频地址
*/
String flashUrl = "http://player.video.qiyi.com/" + videoId + "/0/600/"
+ urldata[0] + "/" + urldata[3] + "/" + urldata[4] + ".swf";
Video video = new Video();
video.setPic(image);
video.setFlash(flashUrl);
video.setTime("77");
return video;
}
/**
*
* 获取6间房视频时长
*/
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;
}
/**
*
* 获取script某个变量的值
*
* @param name
* 变量名称
*
* @return 返回获取的值
*/
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;
}
/**
*
* 根据HTML的ID键及属于名，获取属于值
*
* @param id
* HTML的ID键
*
* @param attrName
* 属于名
*
* @return 返回属性值
*/
private static String getElementAttrById(Document doc, String id,
String attrName) throws Exception {
Element et = doc.getElementById(id);
String attrValue = et.attr(attrName);
return attrValue;
}
/**
*
* 获取网页的内容
*/
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_XMzExNzA1MDA4.html";
// String url = "http://www.tudou.com/programs/view/CHKM-8D5MmE/";
String url = "http://v.youku.com/v_playlist/f16343935o1p0.html";
// String url = "http://v.ku6.com/show/BpP5LeyVwvikbT1F.html";
// String url = "http://www.tudou.com/playlist/p/l10485826.html";
// String url = "http://www.56.com/u64/v_NTkzMDEzMTc.html";
// String url="http://www.tudou.com/playlist/p/l10485826i60175370.html";
url = "http://www.qiyi.com/dianshiju/20111008/23e2566d0ae56486.html";
// String[] s="yule,qiyi.com.20111011.c1837626ef2280a5.html".split(",");
// String a=s[1];
Video video = getVideoInfo(url);
if (video == null) {
System.out.println(url);
} else
{
System.out.println("视频缩略图：" + video.getPic());
System.out.println("视频地址：" + video.getFlash());
System.out.println("视频时长：" + video.getTime());
}
System.out.print(url.endsWith(".html"));
}
}