Merge branch 'source-forward-chen' of

http://git.zhiweidata.top/zhangzhiwei/source_forward.git into
source-forward-chen

Conflicts:
	src/main/java/com/zhiwei/source_forward/run/MediaSelfSource.java

src/main/java/com/zhiwei/source_forward/crawler/UrlLiveCrawler.java

@@ -82,7 +82,7 @@ public class UrlLiveCrawler {
                                 callBack(callback, attr, 1,String.valueOf(rs.code()));
                             }
                         } else {
-                            callBack(callback, attr, 1,String.valueOf(rs.code()));
+                            callBack(callback, attr, 1,"未访问成功");
                         }
                     } catch (Exception e) {
                         logger.error(" 数据是否删除 采集出错 {} ",e);

src/main/java/com/zhiwei/source_forward/run/MediaSelfSource.java

@@ -26,6 +26,7 @@ public class MediaSelfSource {
         ProxyFactory.init("zookeeper://192.168.0.36:2181","local",GroupType.PROVIDER);
         List<String> urlList = new ArrayList<>();
         urlList.add("http://yugang.blogchina.com/713055888.html");
+        urlList.add("https://item.btime.com/m_9bf5d805a257ddc87");
         List<MediaSelfSourceBean> u = MediaSelfSource.getMediaSelfSource(urlList);
         for(MediaSelfSourceBean b : u) {
             System.out.println(b.toString());

src/main/java/com/zhiwei/source_forward/run/SourceForward.java

@@ -81,7 +81,7 @@ public class SourceForward {
 	public static void main(String[] args) {
 	    ProxyFactory.init("zookeeper://192.168.0.36:2181","local",GroupType.PROVIDER);
         List<String> urlList = new ArrayList<>();
-        urlList.add("https://www.jiemian.com/article/2782869.html");
+        urlList.add("http://industry.caijing.com.cn/20190423/4582310.shtml");
         List<SourceForwardBean> da = SourceForward.getSourceForward(urlList);
         for(SourceForwardBean sfb : da) {
             System.out.println(sfb.toString());

src/main/java/com/zhiwei/source_forward/util/MatchSource.java

@@ -3,7 +3,6 @@ package com.zhiwei.source_forward.util;
 import java.util.List;
 import java.util.Objects;
 
-import org.checkerframework.checker.units.qual.s;
 import org.jsoup.Jsoup;
 import org.jsoup.nodes.Document;
 
@@ -101,10 +100,7 @@ public class MatchSource {
             }else if(url.contains("caijing.com.cn")){
                 //财经网产经
                 source = document.select("#source_baidu").text();
-            }
-			
-			
-            else if(url.contains("news.eastday.com")){   
+            }else if(url.contains("news.eastday.com")){   
 				//单独处理东方网
 				source = document.select("div#sectionleft").select("div").select("p").select("a").text();
 			}else if(url.contains("ny.chinacenn.com")){   
@@ -217,7 +213,7 @@ public class MatchSource {
 				source = document.select("div.tip.fl").select("a").text();
 			}else if(url.contains("finance.jrj.com.cn")){
 				//单独处理金融界
-				source = document.select("p.inftop").select("span").select("a").text().replaceAll("价值.*| ", "");
+				source = document.select("p.inftop").select("span").get(1).select("a").text().replaceAll("价值.*| ", "");
 			}else if(url.contains("tech.china.com.cn")){
 				//单独处理中国网
 				source = document.select("span.fl.time2").select("a").text();
@@ -298,8 +294,12 @@ public class MatchSource {
 				source = document.select("span.label_nr").text();
 			}else if(url.contains("jiemian.com")){
 				//单独处理界面新闻
-				source = document.select("div.article-info").select("span").text().replaceAll(".*来源：| 字体[\\w\\W]*", "");
-			}
+//				source = document.select("div.article-info").select("span").text().replaceAll(".*来源：| 字体[\\w\\W]*", "");
+			    return "界面新闻";
+			}else if(url.contains("finance.youth.cn")){
+                //单独处理中国青年网
+			    source = document.select("span#source_baidu").text().replaceAll("来源：|作者.*", "");
+            }
 			
             if(Objects.nonNull(source) && source.length() != 0) {
             	return source;
@@ -339,9 +339,11 @@ public class MatchSource {
 			/***特定网站单独处理**/
 			if(url.contains("toutiao.com")){  
 				//今日头条帐号匹配
-				if(html.contains("name: '")){
+				if(html.contains("name: '") && html.contains("mediaInfo")){
 					source = html.split("mediaInfo:")[1].split("name: '")[1].split("',")[0].trim();
-				}else if(html.contains("screen_name:")){
+				}else if(html.contains("name: '") && html.contains("ugcInfo")){
+                    source = html.split("ugcInfo:")[1].split("name: '")[1].split("',")[0].trim();
+                }else if(html.contains("screen_name:")){
 					source = html.split("screen_name:'")[1].split("',")[0].trim();
 				}
 				if(source!=null && source.length()>1){
@@ -422,6 +424,9 @@ public class MatchSource {
 			}else if(url.contains("item.btime.com")){
 				//北京时间
 				source = document.select("a.author").text();
+				if(Objects.isNull(source) || source.length() < 1){
+                    source = document.select("div.content-info > span.col.cite").text();
+                }
 				if(source!=null && source.length()>1){
 					source = "北京时间-" + source;
 				}