package com.skyline.wo.service;
import org.osm.jsoup.Jsoup;
import org.osm.jsoup.nodes.Document;
import org.osm.jsoup.nodes.Element;
import org.osm.jsoup.select.Elements;
import com.skyline.wo.model.Video;
import com.skyline.wo.type.VideoConstants;
/**
* 视频网址视频信息提取工具类 支持以下网址：优酷、土豆、酷6、6间房、新浪、搜狐、56(我乐)、凤凰视频 版本更新说明： 1、2011-09-12 创建
* 2、2011-09-13 新加对凤凰视频的支持，链接地址不在支持的列表中时返回原链接地址以及链接的页面标题
*
* @author Guihua Liu
* @version v1.0
* @QQ 459041012
* @email 850823422@139.com
* @Website http://liuguihua.com
*/
public class VideoUtil {
/**
* 获取视频信息
*
* @param url
* @return
* @throws Exception
*/
public static Video getVideoInfo(String url) throws Exception {
Video video = null;
if (url.indexOf(VideoConstants.VIDEO_DOMAIN_YOUKU) != -1) {
try {
video = getYouKuVideo(url);
} catch (Exception e) {
e.printStackTrace();
video = null;
}
}
else if(url.indexOf(VideoConstants.VIDEO_DOMAIN_TUDOU_PLAYLIST)!=-1){
try {
video = getTudouPlayListVideo(url);
} catch (Exception e) {
video = null;
}
}else if (url.indexOf(VideoConstants.VIDEO_DOMAIN_TUDOU) != -1) {
try {
video = getTudouVideo(url);
} catch (Exception e) {
video = null;
}
} else if (url.indexOf(VideoConstants.VIDEO_DOMAIN_KU6) != -1) {
try {
video = getKu6Video(url);
} catch (Exception e) {
video = null;
}
} else if (url.indexOf(VideoConstants.VIDEO_DOMAIN_CN6) != -1) {
try {
video = get6Video(url);
} catch (Exception e) {
video = null;
}
} else if (url.indexOf(VideoConstants.VIDEO_DOMAIN_WOLE) != -1) {
try {
video = get56Video(url);
} catch (Exception e) {
video = null;
}
} else if (url.indexOf(VideoConstants.VIDEO_DOMAIN_SINA) != -1) {
try {
video = getSinaVideo(url);
} catch (Exception e) {
video = null;
}
} else if (url.indexOf(VideoConstants.VIDEO_DOMAIN_SOHU) != -1) {
try {
video = getSohuVideo(url);
} catch (Exception e) {
video = null;
}
} else if (url.indexOf(VideoConstants.VIDEO_DOMAIN_IFENG) != -1) {
try {
video = getIfengVideo(url);
} catch (Exception e) {
video = null;
}
} else if (url.indexOf(VideoConstants.VIDEO_DOMAIN_YINYUETAI) != -1) {
try {
video = getYinYueTaiVideo(url);
} catch (Exception e) {
video = null;
}
} else {
// 链接地址不在支持的列表中时返回原链接地址以及链接的页面标题
Document doc = getURLContent(url);
video = new Video();
video.setTitle(doc.title());
video.setPageUrl(url);
}
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);
/**
* 视频标题
*/
String title = doc.title().split("-")[0].trim();
/**
* 获取视频缩略图
*/
String pic = getElementAttrById(doc, "s_sina", "href");
int local = pic.indexOf("pic=");
pic = pic.substring(local + 4);
/**
* 获取视频地址
*/
String flash = getElementAttrById(doc, "link2", "value");
/**
* 获取视频网页代码
*/
String htmlCode = getElementAttrById(doc, "link3", "value");
/**
* 获取视频时间
*/
String time = getElementAttrById(doc, "download", "_href");
String[] arrays = time.split("\\|");
time = arrays[4];
/**
* 获取视频简介
*/
String summary = "";
try {
Element el = doc.getElementById("long");
summary = el.select(".item").get(0).html();
} catch (Exception e) {
}
Video video = new Video();
video.setTitle(title);
video.setThumbnail(pic);
video.setFlashUrl(flash);
video.setTime(time);
video.setSource("优酷视频");
video.setPageUrl(url);
video.setSummary(summary);
video.setHtmlCode(htmlCode);
return video;
}
/**
* 获取土豆视频
* 土豆的播放列表
* @author burningcl
*
* @param url
* 视频URL
*/
public static Video getTudouPlayListVideo(String url) throws Exception {
Document doc = getURLContent(url);
String content = doc.html();
int beginLocal = content.indexOf("<script>var");
int endLocal = content.indexOf("</script>");
content = content.substring(beginLocal, endLocal);
String[] playListInfo=url.split("playlist")[1].split("i");
String id="";
if(playListInfo.length==1){
//URL中不带有播放的文件id
id = getScriptVarByName("defaultIid", content).trim();
content=content.split(id)[3];
//System.out.println(id);
}else{
//URL带有播放的文件id
id = playListInfo[1].split("\\.")[0];
content=content.split(id)[1];
//System.out.println(id);
}
/**
* 视频标题
*/
String title = "";
try {
//title:"
title = content.split("title:\"")[1].split("\"")[0].trim();
} catch (Exception e) {
}
/**
* 获取视频地址
*/
String flash = "";
flash = content.split("icode:\"")[1].split("\"")[0];
flash = "http://www.tudou.com/v/" + flash.trim() + "/v.swf";
/**
* 获取视频缩略图
*/
String pic = "";
pic = content.split("pic:\"")[1].split("\"")[0].trim();
//pic = pic.trim();
/**
* 视频简介
*/
String summary = "";
try {
summary = doc.select("meta[name=Description]").attr("content");
} catch (Exception e) {
}
/**
* 获取视频时间
*/
String time = "";
try {
time = content.split("time:\"")[1].split("\"")[0].trim();
} catch (Exception e) {
}
Video video = new Video();
video.setTitle(title);
video.setThumbnail(pic);
video.setFlashUrl(flash);
video.setTime(time);
video.setSource("土豆视频");
video.setPageUrl(url);
video.setSummary(summary);
video.setHtmlCode(getHtmlCode(flash));
return video;
}
/**
* 获取土豆视频
*
* @param url
* 视频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);
/**
* 视频标题
*/
String title = "";
try {
title = doc.title().split("_")[0].trim();
} catch (Exception e) {
}
/**
* 获取视频地址
*/
String flash = "";
flash = getScriptVarByName("iid_code", content);
flash = "http://www.tudou.com/v/" + flash.split("=")[1].trim() + "/v.swf";
/**
* 获取视频缩略图
*/
String pic = "";
pic = getScriptVarByName("thumbnail", content);
pic = pic.split("=")[1].trim();
/**
* 视频简介
*/
String summary = "";
try {
summary = doc.select("meta[name=Description]").attr("content");
} catch (Exception e) {
}
/**
* 获取视频时间
*/
String time = "";
try {
time = getScriptVarByName("time", content);
} catch (Exception e) {
}
Video video = new Video();
video.setTitle(title);
video.setThumbnail(pic);
video.setFlashUrl(flash);
video.setTime(time);
video.setSource("土豆视频");
video.setPageUrl(url);
video.setSummary(summary);
video.setHtmlCode(getHtmlCode(flash));
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);
/**
* 获取视频标题
*/
String title = doc.title().split(" ")[0].trim();
/**
* 获取视频地址
*/
String flash = getElementAttrById(doc, "outSideSwfCode", "value");
/**
* 获取视频地址
*/
String htmlCode = getElementAttrById(doc, "outSideHtmlCode", "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();
}
/**
* 视频简介
*/
String summary = doc.select("meta[name=Description]").attr("content");
Video video = new Video();
video.setTitle(title);
video.setThumbnail(pic);
video.setFlashUrl(flash);
video.setTime(time);
video.setSource("酷6视频");
video.setPageUrl(url);
video.setSummary(summary);
video.setHtmlCode(htmlCode);
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);
/**
* 视频标题
*/
String title = doc.title().split("-")[0].trim();
/**
* 获取视频缩略图
*/
Element picEt = doc.getElementsByClass("summary").first();
String pic = picEt.getElementsByTag("img").first().attr("src");
/**
* 视频简介
*/
String summary = doc.select("meta[name=Description]").attr("content");
/**
* 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.setTitle(title);
video.setThumbnail(pic);
video.setFlashUrl(flash);
video.setSource("6间房");
video.setPageUrl(url);
video.setSummary(summary);
video.setHtmlCode(getHtmlCode(flash));
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();
/**
* 视频标题
*/
String title = doc.getElementById("VideoTitle").select("h1").text();
/**
* 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";
/**
* 视频简介
*/
String summary = doc.select("meta[name=Description]").attr("content");
Video video = new Video();
video.setTitle(title);
video.setThumbnail(pic);
video.setFlashUrl(flash);
video.setSource("56视频");
video.setPageUrl(url);
video.setSummary(summary);
video.setHtmlCode(getHtmlCode(flash));
return video;
}
/**
* 获取新浪视频
*
* @param url
* @return
* @throws Exception
*/
public static Video getSinaVideo(String url) throws Exception {
Document doc = getURLContent(url);
/**
* 视频标题
*/
String title = doc.getElementById("videoTitle").text();
/**
* 视频简介
*/
String summary = doc.getElementById("videoContent").text();
String content = doc.html();
int beginLocal = content.indexOf("document.domain");
int endLocal = content.indexOf("</script>");
content = content.substring(beginLocal + 2, endLocal);
/**
* 视频缩略图
*/
String pic = getScriptVarByName("pic", content);
/**
* flash地址
*/
String flash = getScriptVarByName("swfOutsideUrl", content);
Video video = new Video();
video.setTitle(title);
video.setThumbnail(pic);
video.setFlashUrl(flash);
video.setSource("新浪视频");
video.setPageUrl(url);
video.setSummary(summary);
video.setHtmlCode(getHtmlCode(flash));
return video;
}
/**
* 获取搜狐视频
*
* @param url
* @return
* @throws Exception
*/
public static Video getSohuVideo(String url) throws Exception {
Document doc = getURLContent(url);
/**
* 视频标题
*/
String title = doc.title().split("-")[0].trim();
/**
* 视频简介
*/
String summary = doc.select(".vIntro.clear > p").text();
/**
* 视频缩略图
*/
String thumbnail = doc.getElementById("thumbnail").attr("src");
String videoId = thumbnail.split("_")[2];
/**
* 视频FLASH地址
*/
String flash = "http://share.vrs.sohu.com/" + videoId + "/v.swf&autoplay=false";
Video video = new Video();
video.setTitle(title);
video.setThumbnail(thumbnail);
video.setFlashUrl(flash);
video.setSource("搜狐视频");
video.setPageUrl(url);
video.setSummary(summary);
video.setHtmlCode(getHtmlCode(flash));
return video;
}
/**
* 获取凤凰视频
*
* @param url
* @return
* @throws Exception
*/
public static Video getIfengVideo(String url) throws Exception {
Document doc = getURLContent(url);
String content = doc.html();
int beginLocal = content.indexOf("videoinfo={");
int endLocal = content.indexOf("</script>");
content = content.substring(beginLocal, endLocal);
content = content.replaceAll("\"", "").replaceAll("\n", "").trim();
content = content.substring(11, content.length() - 2);
String[] params = content.split(",");
/**
* 视频标题
*/
String title = doc.select(".playtitle > h1").text();
/**
* 视频简介
*/
String summary = doc.getElementById("full_des").text();
summary = summary.replaceAll("<<收起", "");
/**
* 视频FLASH地址
*/
String videoId = url.substring(url.lastIndexOf("/"), url.lastIndexOf(".shtml"));
String flash = "http://v.ifeng.com/include/exterior.swf?guid=" + videoId + "&AutoPlay=false";
/**
* 视频缩略图
*/
String thumbnail = params[5].substring(4, params[5].length());
/**
* 视频时长
*/
String time = params[3].substring(9, params[3].length());
long ss = Long.parseLong(time);
long hour = ss / 3600;
long minute = ss % 3600 / 60;
long second = ss % 60;
Video video = new Video();
video.setTitle(title);
video.setThumbnail(thumbnail);
video.setFlashUrl(flash);
video.setTime(hour + ":" + minute + ":" + second);
video.setSource("凤凰视频");
video.setPageUrl(url);
video.setSummary(summary);
video.setHtmlCode(getHtmlCode(flash));
return video;
}
/**
* 获取音悦台MV
*
* @param url
* @return
* @throws Exception
*/
public static Video getYinYueTaiVideo(String url) throws Exception {
Document doc = getURLContent(url);
/**
* 获取视频地址
*/
String flash = getElementAttrById(doc, "flashCode", "value");
/**
* 视频标题
*/
String title = doc.getElementById("videoTitle").html() + " - " + doc.getElementById("videoArtistName").html();
/**
* 视频内容
*/
String summary = doc.getElementById("videoContent").select("span").first().html();
/**
* 视频缩略图
*/
String thumbnail = doc.select("meta[property=og:image]").attr("content");
Video video = new Video();
video.setTitle(title);
video.setThumbnail(thumbnail);
video.setFlashUrl(flash);
video.setSource("音悦台MV");
video.setPageUrl(url);
video.setSummary(summary);
video.setHtmlCode(getHtmlCode(flash));
return video;
}
/**
* 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("'", "");
return result.trim();
}
/**
* 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;
}
/**
* 根据FLASH地址生成页面代码
*
* @param flashUrl
* @return
* @throws Exception
*/
private static String getHtmlCode(String flashUrl) throws Exception {
return "<embed src=\""
+ flashUrl
+ "\" allowFullScreen=\"true\" quality=\"high\" width=\"480\" height=\"400\" align=\"middle\" allowScriptAccess=\"always\" type=\"application/x-shockwave-flash\"></embed>";
}
/**
* 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(5000).get();
return doc;
}
}