添加部分自媒体号采集

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

@@ -129,6 +129,15 @@ public class MediaSelfSourceCrawler {
                 if(url.contains("wap.peopleapp.com/article")) {
                     return "https://app.peopleapp.com/WapApi/610/ArtInfoApi/getInfoUp?id=" + url.split("article")[1].split("/")[1];
                 }
+                if(url.contains("a.mp.uc.cn/") && url.contains("wm_cid=")) {
+                    return "https://ff.dayu.com/contents/" + url.split("wm_cid=")[1].split("&")[0] + "?biz_id=1002&_fetch_author=1";
+                }
+                if(url.contains("tznew.58.com")) {
+                    return "https://tznew.58.com/tznew/c/info-detail?infoid=" + url.split("infoid=")[1].split("&")[0];
+                }
+                if(url.contains("wap.peopleapp.com/article")) {
+                    return "https://app.peopleapp.com/WapApi/610/ArtInfoApi/getInfoUp?id=" + url.split("article")[1].split("/")[1];
+                }
                 return url;
             }
         } catch (Exception e) {

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

@@ -32,7 +32,7 @@ public class MediaSelfSource {
     public static void main(String[] args) {
         ProxyFactory.init("zookeeper://192.168.0.11:2181?backup=192.168.0.30:2181,192.168.0.35:2181", "local", GroupType.PROVIDER, 10000002L);
         List<String> urlList = new ArrayList<>();
-        urlList.add("http://app.myzaker.com/news/article.php?pk=5dbef675b15ec03075727503");
+        urlList.add("https://wap.peopleapp.com/article/rmh12074926/0");
         List<MediaSelfSourceBean> u = MediaSelfSource.getMediaSelfSource(urlList);
         for(MediaSelfSourceBean b : u) {
             System.out.println(b.toString());

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

@@ -371,7 +371,19 @@ public class MatchSource {
 						source =  "搜狐-" + source;
 					}
 				}
-			}else if(url.contains("a.mini.eastday.com")){  
+			}else if(url.contains("tznew.58.com")){ 
+                //58
+                source = JSONObject.parseObject(html).getJSONObject("result").getString("author");
+                if(source!=null && source.length()>1){
+                    source =  "58-" + source;
+                }
+            }else if(url.contains("c.m.163.com")){ 
+                //58
+                source = document.select("section.g-article.js-article > div.js-article-inner > div > b").text();
+                if(source!=null && source.length()>1){
+                    source =  "网易新闻-" + source;
+                }
+            }else if(url.contains("a.mini.eastday.com")){  
 				//处理东方头条网-自媒体号匹配
 				source = document.select("[class=\"share_cnt_p clearfix\"]").select("div.fl").select("i").get(1).text().trim();
 				if(source!=null && source.length()>1){
@@ -677,9 +689,21 @@ public class MatchSource {
                 if(source!=null && !source.equals("")) {
                     source = "传送门-" + source;
                 }
+            }else if (url.contains("a.mp.uc.cn")) {
+                JSONObject json = JSONObject.parseObject(html);
+                source = json.getJSONObject("data").getJSONObject("_author").getString("author_name");
+                if(source!=null && !source.equals("")) {
+                    source = "uc-" + source;
+                }
+            }else if (url.contains("kd.youth.cn")) {
+                source = document.select("body > div > div > div.rich_media_meta_list > a").text();
+                if(source!=null && !source.equals("")) {
+                    source = "中青在线-" + source;
+                }
             }
 			return source;
 		} catch (Exception e) {
+		    e.printStackTrace();
 			return null;
 		}
 	}