添加来源判断

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

@@ -9,6 +9,8 @@ import java.util.Map.Entry;
 import org.apache.logging.log4j.LogManager;
 import org.apache.logging.log4j.Logger;
 
+import com.zhiwei.common.config.GroupType;
+import com.zhiwei.crawler.proxy.ProxyFactory;
 import com.zhiwei.source_forward.bean.SourceForwardBean;
 import com.zhiwei.source_forward.bean.SourceForwardBean.Attribution;
 import com.zhiwei.source_forward.crawler.SourceForwardCrawler;
@@ -77,13 +79,13 @@ 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.toutiao.com/a6634320415839748621");
-//        List<SourceForwardBean> da = SourceForward.getSourceForward(urlList);
-//        for(SourceForwardBean sfb : da) {
-//            System.out.println(sfb.toString());
-//        }
+	    ProxyFactory.init("zookeeper://192.168.0.36:2181","local",GroupType.PROVIDER);
+        List<String> urlList = new ArrayList<>();
+        urlList.add("http://www.northnews.cn/2019/0419/3080909.shtml");
+        List<SourceForwardBean> da = SourceForward.getSourceForward(urlList);
+        for(SourceForwardBean sfb : da) {
+            System.out.println(sfb.toString());
+        }
     }
 	
     static class SourceForwardCrawlerThread extends Thread{

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

@@ -39,7 +39,7 @@ public class MatchSource {
 	/**
 	 * @Title: findURLs
 	 * @author hero
-	 * @Description: TODO(验证并匹配数据)
+	 * @Description: (验证并匹配数据)
 	 * @param @param
 	 *            s
 	 * @param @param
@@ -91,17 +91,24 @@ public class MatchSource {
 				}else{
 					source = html.split("source\":\"")[1].split("\",\"")[0];
 				}
+			}else if(url.contains("tech.china.com")){
+                //中华网科技
+			    source = document.select("#chan_newsInfo").text().split("来源：")[1];
+            }else if(url.contains("caijing.com.cn")){
+                //财经网产经
+                source = document.select("#source_baidu").text();
             }else{ 
 				//其他网站处理
 				source = mathchOtherSource(html, htmlBody, sourceList);
 			}
 			if(source!=null){
 				//验证来源
-				for (String sourceMatch : sourceList) {
-					if (source.contains(sourceMatch)) {
-						return sourceMatch;
-					}
-				}
+//				for (String sourceMatch : sourceList) {
+//					if (source.contains(sourceMatch)) {
+//						return sourceMatch;
+//					}
+//				}
+			    return source;
 			}
 		} catch (Exception e) {
 			e.printStackTrace();

src/main/resources/sourceList.txt

@@ -3052,3 +3052,4 @@ ZOL中关村在线
 走进中关村
 最高人民法院网
 最高人民检察院
+今日湖北
\ No newline at end of file