360网页采集

src/main/java/com/zhiwei/media_data_crawler/crawler/SoCrawlerParse.java

@@ -215,7 +215,7 @@ public class SoCrawlerParse extends HttpClientTemplateOK {
 					String time = dataJson.getJSONObject("data").getJSONArray("question").getJSONObject(0).getString("create_time");
 					String source = dataJson.getJSONObject("data").getJSONArray("question").getJSONObject(0).getJSONObject("user").getString("uname");
 					String user_id = dataJson.getJSONObject("data").getJSONArray("question").getJSONObject(0).getJSONObject("user").getString("user_id");
-					String link = "https://www.toutiao.com/a"+dataJson.getJSONObject("data").getJSONArray("question").getJSONObject(0).getString("qid")+"/";
+					String link = "https://www.toutiao.com/a"+dataJson.getJSONObject("data").getJSONArray("question").getJSONObject(0).getString("qid");
 					return new NewsData(link, title, source, time, content, "头条问答", word, user_id);
 				}else if(htmlBody.contains("var BASE_DATA = ")){
 					
@@ -226,7 +226,7 @@ public class SoCrawlerParse extends HttpClientTemplateOK {
 					String time = dataJson.getJSONObject("articleInfo").getJSONObject("subInfo").getString("time");
 					String source = dataJson.getJSONObject("mediaInfo").getString("name");
 					String user_id = dataJson.getJSONObject("mediaInfo").getString("uid");
-					String link = "https://www.toutiao.com/a"+dataJson.getJSONObject("articleInfo").getString("groupId")+"/";
+					String link = "https://www.toutiao.com/a"+dataJson.getJSONObject("articleInfo").getString("groupId");
 					return new NewsData(link, title, source, time, content, "今日头条", word, user_id);
 				}else if(htmlBody.contains("404错误页")){
 					logger.info("{}：：：数据有问题,该文章已被删除}", url);