主持人采集

src/main/java/com/zhiwei/searchhotcrawler/bean/HotSearchCache.java

@@ -108,6 +108,11 @@ public class HotSearchCache {
      */
     private double riseSpeed;
 
+    /**
+     * 主持人
+     */
+    private String downtext;
+
     public HotSearchCache(String url, String name, String topicLead, Integer highestCount, Integer lastCount, Boolean hot,
                            Date startTime, Date endTime, Integer highestRank, Integer lastRank, String type, Integer duration){
         this.id = name + "_" + type;

src/main/java/com/zhiwei/searchhotcrawler/crawler/WeiboHotSearchCrawler.java

@@ -258,6 +258,12 @@ public class WeiboHotSearchCrawler {
 						document.put("readCount", TipsUtils.getHotCount(read));
 						document.put("discussCount", TipsUtils.getHotCount(discussCount));
 						document.put("pictureUrl",pictureUrl);
+						if (readJson.getJSONObject("head_data").containsKey("downtext")){
+							String downtext = readJson.getJSONObject("head_data").getString("downtext");
+							if(!"".equals(downtext)) {
+								document.put("downtext",downtext.replaceAll("主持人：",""));
+							}
+						}
 					}
 				}
 				return document;

src/main/java/com/zhiwei/searchhotcrawler/dao/HotSearchCacheDAO.java

@@ -197,6 +197,9 @@ public class HotSearchCacheDAO  {
                        if (nowDoc.containsKey("pictureUrl")) {
                            nowDoc.put("pictureUrl",nowDoc.getString("pictureUrl"));
                        }
+                       if (nowDoc.containsKey("downtext")) {
+                           nowDoc.put("downtext",nowDoc.getString("downtext"));
+                       }
                    }
                    collection.insertOne(nowDoc);
                }
@@ -342,6 +345,9 @@ public class HotSearchCacheDAO  {
             if(document.containsKey("pictureUrl")){
                 nowDoc.put("pictureUrl",document.getString("pictureUrl"));
             }
+            if (Objects.isNull(nowDoc.get("downtext")) && document.containsKey("downtext")){
+                nowDoc.put("downtext",document.getString("downtext"));
+            }
             collection.replaceOne(query, nowDoc);
         }
     }