添加根据链接采集全文

src/main/java/com/zhiwei/wechat/entity/WechatAricle.java

@@ -40,6 +40,12 @@ public class WechatAricle {
 	
 	private String isFirst; //是否为头条文章
 	
+	private String biz; //微信公众号唯一标识
+	
+	private String wxId; //微信公众号id
+	
+	private String user_name;//微信公众号初始id
+	
 	public String getIsFirst() {
 		return isFirst;
 	}
@@ -103,6 +109,24 @@ public class WechatAricle {
 	}
 	
 	
+	public String getBiz() {
+		return biz;
+	}
+	public String getWxId() {
+		return wxId;
+	}
+	public String getUser_name() {
+		return user_name;
+	}
+	public void setBiz(String biz) {
+		this.biz = biz;
+	}
+	public void setWxId(String wxId) {
+		this.wxId = wxId;
+	}
+	public void setUser_name(String user_name) {
+		this.user_name = user_name;
+	}
 	public WechatAricle(){}
 	
 	public WechatAricle(String id,String title,String source,String content

src/main/java/com/zhiwei/wechat/search/WechatAritcleSearch.java

@@ -231,7 +231,7 @@ public class WechatAritcleSearch {
 							title = ZhiWeiTools.SBC2DBC(title);
 							content = ZhiWeiTools.SBC2DBC(content);
 							wechat = new WechatAricle(link, title, source, content, date, readNum, 0, openid, "unknow");
-							wechat = getContentAndSource(link, proxyHolder, wechat);
+							wechat = getWechatAricleInfo(link, proxyHolder, wechat);
 							result.add(wechat);
 						} catch (Exception e) {
 							logger.debug("解析数据出现错误:{}", e.getMessage());
@@ -267,37 +267,54 @@ public class WechatAritcleSearch {
 	 * @return
 	 * @throws IOException
 	 */
-	private static WechatAricle getContentAndSource(String url, ProxyHolder proxy,WechatAricle wechatAricle){
+	private static WechatAricle getWechatAricleInfo(String url, ProxyHolder proxy,WechatAricle wechatAricle){
 		try {
-			String htmlBody = httpBoot.syncCall(RequestUtils.wrapGet(url), proxy.getProxy()).body().string();
-			if(htmlBody!=null){
-				Document document = Jsoup.parse(htmlBody);
-				String content = null;
-				String source = null;
-				String text = null;
-				if(htmlBody.contains("js_article")){
+			String contentHtml = httpBoot.syncCall(RequestUtils.wrapGet(url), proxy.getProxy()).body().string();
+			String content = null;
+			String time = null;
+			String source = null;
+			String biz = null;
+			String title = null;
+			String user_name = null;
+			String wxId = null;
+			if(contentHtml!=null){
+				Document document = Jsoup.parse(contentHtml);
+				title = document.select("title").text();
+				wxId = document.select("p.profile_meta").get(0).select("span.profile_meta_value").text();
+				
+				if(contentHtml.contains("js_article")){
 					content = document.select("div#js_article").text();
-				}else if(htmlBody.contains("js_share_content")){
+				}else if(contentHtml.contains("js_share_content")){
 					content = document.select("div#js_share_content").text();
 				}
-				if(htmlBody.contains("content_tpl")){
-					text = document.select("script#content_tpl").html();
-					text = Jsoup.parse(text).text();
+				if(contentHtml.contains("content_tpl")){
+					String text = document.select("script#content_tpl").html();
+					content = Jsoup.parse(text).text();
 				}
-				content = content+text;
-				if(htmlBody.contains("js_name")){
-					source = document.select("a#js_name").text().trim();
-				}else if(htmlBody.contains("account_nickname")){
-					source = document.select("div.account_nickname").text().trim();
-				}
-//				System.out.println(source+"=========="+content);
-				if(content!=null && content.length()>50){
-					wechatAricle.setContent(content);
-				}
-				if(source!=null && content.length()>0){
-					wechatAricle.setSource(source);
+				
+				if(contentHtml.contains("d.nick_name = ")){
+					time = contentHtml.split("d.ct = \"")[1].split("\";")[0];
+					source = contentHtml.split("d.nick_name = \"")[1].split("\";")[0];
+					biz = contentHtml.split("d.biz = \"")[1].split("\"")[0];
+					user_name = contentHtml.split("d.user_name = \"")[1].split("\"")[0];
+				}else if(contentHtml.contains("var nickname = ")){
+					time = contentHtml.split("var ct = \"")[1].split("\";")[0];
+					source = contentHtml.split("var nickname = \"")[1].split("\";")[0];
+					biz = contentHtml.split("var appuin = \"\"||\"")[1].split("\"")[0];
+					user_name = contentHtml.split("var user_name = \"")[1].split("\"")[0];
 				}
 			}
+			if(wechatAricle == null) {
+				wechatAricle = new WechatAricle();
+				wechatAricle.setTitle(title);
+				wechatAricle.setTime(new Date(Long.valueOf(time)*1000));
+				wechatAricle.setSource(source);
+			}
+			wechatAricle.setBiz(biz);
+			wechatAricle.setContent(content);
+			wechatAricle.setWxId(wxId);
+			wechatAricle.setUser_name(user_name);
+			
 		} catch (Exception e) {
 			e.printStackTrace();
 			return wechatAricle;
@@ -385,7 +402,7 @@ public class WechatAritcleSearch {
 							title = ZhiWeiTools.SBC2DBC(title);
 							content = ZhiWeiTools.SBC2DBC(content);
 							wechat = new WechatAricle(link, title, source, content, date, readNum, 0, openid, "unknow");
-							wechat = getContentAndSource(link, proxyHolder, wechat);
+							wechat = getWechatAricleInfo(link, proxyHolder, wechat);
 							result.add(wechat);
 						} catch (Exception e) {
 							logger.debug("解析数据出现错误:{}", e.getMessage());
@@ -432,7 +449,6 @@ public class WechatAritcleSearch {
 		    
 		try {
 			htmlBody = httpBoot.syncCall(RequestUtils.wrapGet(url), proxyHolder).body().string();
-			System.out.println(htmlBody);
 			if (htmlBody != null) {
 				JSONObject json = JSONObject.parseObject(htmlBody);
 				openId = json.getString("openid");