添加验证来源中的渠道验证

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

@@ -33,10 +33,14 @@ public class SourceForwardPageProcessor implements PageProcessor {
 		String source = null;
 		String channel = "新闻";
 		try {
-			if(page.getStatusCode()!=404){
-				List<Node> nodeList = page.getHtml().getDocument().head().childNodes();
-				source = TreateData.matchSource(page.getUrl().get(),page.getHtml().toString(), sourceList);
-				channel = TreateData.matchChannel(nodeList);
+			
+			channel = verifyChannel(page.getUrl().get());
+			if(channel==null){
+				if(page.getStatusCode()!=404){
+					List<Node> nodeList = page.getHtml().getDocument().head().childNodes();
+					source = TreateData.matchSource(page.getUrl().get(),page.getHtml().toString(), sourceList);
+					channel = TreateData.matchChannel(nodeList);
+				}
 			}
 		} catch (Exception e) {
 			source = null;
@@ -49,6 +53,46 @@ public class SourceForwardPageProcessor implements PageProcessor {
 		
 		page.putField("data", data);
 	}
-
+	
+	
+	
+	/**
+	 * @Title: verifyChannel 
+	 * @author hero 
+	 * @Description: 根据链接验证文章频道
+	 * @param @param url
+	 * @param @return 设定文件 
+	 * @return String 返回类型
+	 */
+	private static String verifyChannel(String url){
+		String channel = null;
+		if(url.contains("news.")){
+			channel = "新闻";
+		}else if(url.contains("finance.") || url.contains("business.")
+				|| url.contains("money.")){
+			channel = "财经";
+		}else if(url.contains("tech.") || url.contains("it.")){
+			channel = "科技";
+		}else if(url.contains("sports.")){
+			channel = "体育";
+		}else if(url.contains("ent.") || url.contains("yule.")){
+			channel = "娱乐";
+		}else if(url.contains("auto.")){
+			channel = "汽车";
+		}else if(url.contains("fashion.")){
+			channel = "时尚";
+		}else if(url.contains("learning.") || url.contains("edu.")){
+			channel = "教育";
+		}else if(url.contains("baobao.")){
+			channel = "母婴";
+		}else if(url.contains("house.") ||url.contains("leju.")
+				|| url.contains("focus.")){
+			channel = "房产";
+		}else if(url.contains("games.")){
+			channel = "游戏";
+		}
+		
+		return channel;
+	}
 	
 }