Merge branch 'source-forward-chen' of…

Merge branch 'source-forward-chen' of http://git.zhiweidata.top/zhangzhiwei/source_forward.git into source-forward-chen

pom.xml

@@ -3,7 +3,7 @@
     <modelVersion>4.0.0</modelVersion>
     <groupId>com.zhiwei</groupId>
     <artifactId>source-forward</artifactId>
-    <version>0.2.6-SNAPSHOT</version>
+    <version>0.2.7-SNAPSHOT</version>
     <name>source-forward</name>
     <description>验证网媒的转发关系及链接的有效性（转发验证微信及自媒体匹配率不高）</description>
 
@@ -24,18 +24,18 @@
       <dependency>
             <groupId>com.alibaba</groupId>
             <artifactId>fastjson</artifactId>
-            <version>1.2.62</version>
+            <version>1.2.71</version>
             <scope>provided</scope>
         </dependency>
         <dependency>
             <groupId>com.zhiwei.tools</groupId>
             <artifactId>zhiwei-tools</artifactId>
-            <version>0.1.9-SNAPSHOT</version>
+            <version>0.2.0-SNAPSHOT</version>
         </dependency>
         <dependency>
         	<groupId>com.zhiwei.crawler</groupId>
         	<artifactId>crawler-core</artifactId>
-        	<version>0.6.6.3-SNAPSHOT</version>
+        	<version>0.6.6.8-SNAPSHOT</version>
         </dependency>
     </dependencies>
 

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

@@ -80,7 +80,7 @@ public class SourceForward {
 	public static void main(String[] args) {
         ProxyInit.initProxy();
         List<String> urlList = new ArrayList<>();
-        urlList.add("http://software.it168.com/a2019/0621/6005/000006005693.shtml");
+        urlList.add("http://www.wangjiaozixun.com/html/zx20/2020/0730/1396388.html");
         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

@@ -39,6 +39,7 @@ public class MatchSource {
 			+ "|(\\d{4}\\D+\\d{0,2}\\D+\\d{0,2}\\D+\\d{0,2}\\D+\\d{0,2})"
 			+ "|(\\d{4}\\D+\\d{0,2}\\D+\\d{0,2}\\D+\\d{0,2})"
 			+ "|(\\d{4}\\D+\\d{0,2}\\D+\\d{0,2})"
+			+ "(\\d+[^\\d]{1,2})+\\d+"
 			;
 	/**
 	 * @Title: findURLs
@@ -60,9 +61,13 @@ public class MatchSource {
 			/***特定网站单独处理**/
 			if(url.contains("thepaper.cn")){  
 				//单独处理澎湃数据
+				if(StringUtils.isNotBlank(document.select("div.name").text())){
+					source = document.select("div.name").text();
+				}else{
 					source = document.select("div.news_about").select("p").select("span").text().replaceAll(".*来源：", "");
 					if(StringUtils.isNotBlank(source)) {
-					source = document.select("div.news_about").text().replaceAll(" \\d{4}.*|.*/", "");
+						source = document.select("div.news_about").text().replaceAll("\\d{4}.*|.*/", "");
+					}
 				}
 			}else if(url.contains("sports.eastday.com")){
 				//单独处理东方体育网
@@ -314,6 +319,11 @@ public class MatchSource {
                 source = document.select("div.tpl_header_author").text();
             }else if(url.contains("china.prcfe.com")) {
                 source = html.split("\"")[1];
+            }else if(url.contains("wangjiaozixun.com")) {
+				source = document.select("p.em_media").text();
+				if(StringUtils.isNotBlank(source)){
+					source = source.replaceAll(".*来源：|）", "");
+				}
 			}
 			
             if(Objects.nonNull(source) && source.length() != 0) {