全文获取修改为晋豪获取

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

@@ -8,8 +8,7 @@ import org.jsoup.nodes.Document;
 import org.slf4j.Logger;
 import org.slf4j.LoggerFactory;
 
-import com.zhiwei.source_forward.content.ContentExtractor;
-import com.zhiwei.source_forward.content.News;
+import com.kohlschutter.boilerpipe.extractors.ArticleExtractor;
 import com.zhiwei.tools.tools.ZhiWeiTools;
 
 /**
@@ -103,12 +102,20 @@ public class MatchContent {
 		/** 正文抽取,目的是避免正文中存在相应匹配规则 **/
 	    String content = null;
         try {
-			News news = ContentExtractor.getNewsByHtml(html);
-			content = TreateData.filterSpecialCharacter(news.getContent());
+          content = ArticleExtractor.getInstance().getText(html);
         } catch (Exception e) {
             logger.error("正文抽取失败,获取全文文本:",e);
             content = document.text();
         }
+	    
+//		String content = null;
+//		try {
+//			News news = ContentExtractor.getNewsByHtml(html);
+//			content = TreateData.filterSpecialCharacter(news.getContent());
+//		} catch (Exception e) {
+//			logger.error("正文抽取失败,获取全文文本:",e);
+//			content = document.text();
+//		}
 		return content;
 	}
 }