文章历史表现仅查询最近180天内发布文章

long-article-recommend-service/src/main/java/com/tzld/longarticle/recommend/server/mapper/crawler/ArticleMapper.java

@@ -15,7 +15,7 @@ public interface ArticleMapper {
 
     List<Article> getWaitingFindArticle(Long timestamp);
 
-    List<Article> getByTitleMd5InAndTypeEqualsAndStatusEquals(List<String> titleMd5s, String type, Integer status);
+    List<Article> getByTitleMd5InAndTypeEqualsAndStatusEquals(List<String> titleMd5s, String type, Long hisArticleStartTime, Integer status);
 
     List<Article> getByGhIdInAndAppMsgIdInAndItemIndexAndTypeEqualsAndStatusEquals(
             Set<String> ghIdList, Set<String> appMsgIdList, Integer itemIndex, String type, Integer status);

long-article-recommend-service/src/main/java/com/tzld/longarticle/recommend/server/service/recommend/recall/RecallService.java

@@ -440,8 +440,10 @@ public class RecallService implements ApplicationContextAware {
         // 获取历史已发布文章
         List<Article> hisArticleList = new ArrayList<>();
         List<List<String>> titleMd5Partition = Lists.partition(new ArrayList<>(titleMd5List), 1000);
+        // 历史已发布表现时限
+        Long hisArticleStartTime = DateUtils.getBeforeDayStart(180);
         for (List<String> titleMd5s : titleMd5Partition) {
-            hisArticleList.addAll(articleMapper.getByTitleMd5InAndTypeEqualsAndStatusEquals(titleMd5s, type, 1));
+            hisArticleList.addAll(articleMapper.getByTitleMd5InAndTypeEqualsAndStatusEquals(titleMd5s, type, hisArticleStartTime, 1));
         }
         Map<String, Map<Integer, List<Article>>> map = hisArticleList.stream()
                 .collect(Collectors.groupingBy(Article::getTitle, Collectors.groupingBy(Article::getItemIndex)));

long-article-recommend-service/src/main/resources/mapper/crawler/ArticleMapper.xml

@@ -31,7 +31,7 @@
         <foreach collection="titleMd5s" item="item" separator="," open="(" close=")">
             #{item}
         </foreach>
-        and Type = #{type} and status = #{status}
+        and Type = #{type} and status = #{status} and publish_timestamp > #{hisArticleStartTime}
     </select>
 
     <select id="getByGhIdInAndAppMsgIdInAndItemIndexAndTypeEqualsAndStatusEquals"