过滤垃圾文章：包含很多链接

zhima1234 · zhima1234 · commit 182595d65fb7 · 2016-11-27T17:50:26.000+08:00
diff --git a/src/logic/article.go b/src/logic/article.go
@@ -149,10 +149,15 @@ func (ArticleLogic) ParseArticle(ctx context.Context, articleUrl string, auto bo
 
 	// 自动抓取，内容长度不能少于 300 字
 	if auto && len(txt) < 300 {
-		logger.Infoln(articleUrl, "content is short")
+		logger.Errorln(articleUrl, "content is short")
 		return nil, errors.New("content is short")
 	}
 
+	if auto && strings.Count(txt, "http://") > 10 {
+		logger.Errorln(articleUrl, "content contains too many link!")
+		return nil, errors.New("content contains too many link")
+	}
+
 	pubDate := times.Format("Y-m-d H:i:s")
 	if rule.PubDate != "" {
 		pubDate = strings.TrimSpace(doc.Find(rule.PubDate).First().Text())
