单小程序文章删除

long-article-recommend-service/src/main/java/com/tzld/longarticle/recommend/server/mapper/aigc/AigcBaseMapper.java

@@ -70,4 +70,6 @@ public interface AigcBaseMapper {
     List<String> getOnlyMiniprogramArticle(Integer offset, Integer pageSize);
 
     List<ProduceTaskAtom> getProduceScoreGreaterThan(Long startTimeStamp, Long endTimeStamp);
+
+    void updatePublishContentSingleMiniProgram(List<String> publishContentIds);
 }

long-article-recommend-service/src/main/java/com/tzld/longarticle/recommend/server/repository/crawler/ArticleRepository.java

@@ -38,4 +38,6 @@ public interface ArticleRepository extends JpaRepository<Article, String> {
     List<Article> getByGhIdInAndUpdateTimeGreaterThanAndTypeEquals(List<String> ghIds, long l, String number);
 
     List<Article> getByTitleMd5(String titleMd5);
+
+    List<Article> getByPublishTimestampGreaterThan(Long publishTimestamp);
 }

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

@@ -778,7 +778,12 @@ public class ArticleAuditService {
             if (CollectionUtils.isEmpty(publishContentIds)) {
                 break;
             }
-            buildArticleAuditDelete(publishContentIds);
+            List<LongArticleAuditDelete> exists = longArticleAuditDeleteRepository.getByPublishContentIdIn(publishContentIds);
+            List<String> existsIds = exists.stream().map(LongArticleAuditDelete::getPublishContentId).collect(Collectors.toList());
+            publishContentIds.removeAll(existsIds);
+            if (CollectionUtils.isNotEmpty(publishContentIds)) {
+                buildArticleAuditDelete(publishContentIds);
+            }
             pageNum++;
         }
     }

long-article-recommend-service/src/main/resources/mapper/aigc/AigcBaseMapper.xml

@@ -15,13 +15,9 @@
         join publish_content_output output
           on content.id = output.publish_content_id and output.content_type = 3 AND output.select_status = 1
         where content.status = 2
-          and content.publish_account_id in
+        and (content.publish_account_id, output.output) in
         <foreach collection="list" item="item" open="(" close=")" separator=",">
-            #{item.publishAccountId}
-        </foreach>
-          and output.output in
-        <foreach collection="list" item="item" open="(" close=")" separator=",">
-            #{item.title}
+            (#{item.publishAccountId}, #{item.title})
         </foreach>
     </select>
 
@@ -370,4 +366,11 @@
         order by record.audit_timestamp desc
     </select>
 
+    <update id="updatePublishContentSingleMiniProgram">
+        update publish_content set gzh_only_miniprogram_flag = 1 where id in
+        <foreach collection="publishContentIds" item="item" open="(" close=")" separator=",">
+            #{item}
+        </foreach>
+    </update>
+
 </mapper>

long-article-recommend-service/src/test/java/com/tzld/longarticle/recommend/server/RecommendTest.java

@@ -5,11 +5,16 @@ import com.alibaba.fastjson.JSONArray;
 import com.alibaba.fastjson.JSONObject;
 import com.google.common.collect.Lists;
 import com.tzld.longarticle.recommend.server.common.enums.recommend.ArticleTypeEnum;
+import com.tzld.longarticle.recommend.server.mapper.aigc.AigcBaseMapper;
 import com.tzld.longarticle.recommend.server.mapper.crawler.CrawlerBaseMapper;
+import com.tzld.longarticle.recommend.server.model.dto.PublishContentDTO;
+import com.tzld.longarticle.recommend.server.model.entity.aigc.PublishAccount;
 import com.tzld.longarticle.recommend.server.model.entity.crawler.AccountAvgInfo;
 import com.tzld.longarticle.recommend.server.model.entity.crawler.Article;
 import com.tzld.longarticle.recommend.server.model.entity.crawler.ArticleDetailInfo;
 import com.tzld.longarticle.recommend.server.model.entity.crawler.PublishSortLog;
+import com.tzld.longarticle.recommend.server.model.param.PublishContentParam;
+import com.tzld.longarticle.recommend.server.repository.aigc.PublishAccountRepository;
 import com.tzld.longarticle.recommend.server.repository.crawler.AccountAvgInfoRepository;
 import com.tzld.longarticle.recommend.server.repository.crawler.ArticleDetailInfoRepository;
 import com.tzld.longarticle.recommend.server.repository.crawler.ArticleRepository;
@@ -26,6 +31,7 @@ import org.apache.poi.ss.usermodel.Workbook;
 import org.apache.poi.xssf.usermodel.XSSFWorkbook;
 import org.junit.jupiter.api.Test;
 import org.springframework.beans.BeanUtils;
+import org.springframework.beans.factory.annotation.Autowired;
 import org.springframework.boot.test.context.SpringBootTest;
 
 import javax.annotation.Resource;
@@ -55,6 +61,10 @@ public class RecommendTest {
     private CrawlerBaseMapper crawlerBaseMapper;
     @Resource
     private PublishSortLogRepository publishSortLogRepository;
+    @Autowired
+    private PublishAccountRepository publishAccountRepository;
+    @Autowired
+    private AigcBaseMapper aigcBaseMapper;
 
 //    @Test
 //    void recall() {
@@ -528,9 +538,9 @@ public class RecommendTest {
 
     @Test
     public void account() {
-        List<String> ghIds = Arrays.asList("gh_d7fa1998b4e1","gh_52100b6803fb","gh_8d7fc54d5026");
-        List<String> accountNames = Arrays.asList("生活超读","灵读生活","生活情感读");
-        List<Integer> fans = Arrays.asList(85759,103083,79214);
+        List<String> ghIds = Arrays.asList("gh_d7fa1998b4e1", "gh_52100b6803fb", "gh_8d7fc54d5026");
+        List<String> accountNames = Arrays.asList("生活超读", "灵读生活", "生活情感读");
+        List<Integer> fans = Arrays.asList(85759, 103083, 79214);
         List<Article> articleList = articleRepository.getByGhIdInAndPublishTimestampLessThanAndTypeEquals(
                 Arrays.asList("gh_02f5bca5b5d9"), 1729353600L, ArticleTypeEnum.QUNFA.getVal());
         for (int i = 0; i < ghIds.size(); i++) {
@@ -572,7 +582,7 @@ public class RecommendTest {
         List<Article> articleList = articleRepository.getByGhIdInAndUpdateTimeGreaterThanAndTypeEquals(ghIds, 1725638400L, "9");
         articleList = articleList.stream().filter(o -> o.getItemIndex() == 1).collect(Collectors.toList());
         Map<String, Map<String, Article>> articleMap = articleList.stream().collect(Collectors.groupingBy(Article::getGhId, Collectors.toMap(
-                o -> DateUtils.timestampToYMDStr(o.getUpdateTime(),"yyyyMMdd"), o -> o,
+                o -> DateUtils.timestampToYMDStr(o.getUpdateTime(), "yyyyMMdd"), o -> o,
                 (existing, replacement) -> replacement)));
         List<AccountAvgInfo> accountAvgInfoList = accountAvgInfoRepository.getAllByGhIdIn(new HashSet<>(ghIds));
         Map<String, Map<String, AccountAvgInfo>> accountAvgInfoMap = accountAvgInfoList.stream()
@@ -616,17 +626,17 @@ public class RecommendTest {
                 title = publishSortLog.getTitle();
                 scoreArr[i] = Double.parseDouble(publishSortLog.getScore());
                 JSONObject scoreMap = JSONObject.parseObject(publishSortLog.getScoreMap());
-                HisFissionFansRateRateStrategyArr[i] =  Double.parseDouble(String.format("%.3f", Optional.of(scoreMap.getDoubleValue("HisFissionFansRateRateStrategy")).orElse(0.0)));
-                HisFissionAvgReadRateRateStrategyArr[i] =  Double.parseDouble(String.format("%.3f", Optional.of(scoreMap.getDoubleValue("HisFissionAvgReadRateRateStrategy")).orElse(0.0)));
-                PublishTimesStrategyArr[i] =  Double.parseDouble(String.format("%.3f", Optional.of(scoreMap.getDoubleValue("PublishTimesStrategy")).orElse(0.0)));
-                ViewCountRateCorrelationStrategyArr[i] =  Double.parseDouble(String.format("%.3f", Optional.of(scoreMap.getDoubleValue("ViewCountRateCorrelationStrategy")).orElse(0.0)));
-                HisFissionAvgReadSumRateStrategyArr[i] =  Double.parseDouble(String.format("%.3f", Optional.of(scoreMap.getDoubleValue("HisFissionAvgReadSumRateStrategy")).orElse(0.0)));
-                HisFissionAvgReadRateCorrelationRateStrategyArr[i] =  Double.parseDouble(String.format("%.3f", Optional.of(scoreMap.getDoubleValue("HisFissionAvgReadRateCorrelationRateStrategy")).orElse(0.0)));
-                HisFissionFansSumRateStrategyArr[i] =  Double.parseDouble(String.format("%.3f", Optional.of(scoreMap.getDoubleValue("HisFissionFansSumRateStrategy")).orElse(0.0)));
-                SimilarityStrategyArr[i] =  Double.parseDouble(String.format("%.3f", Optional.of(scoreMap.getDoubleValue("SimilarityStrategy")).orElse(0.0)));
-                ViewCountStrategyArr[i] =  Double.parseDouble(String.format("%.3f", Optional.of(scoreMap.getDoubleValue("ViewCountStrategy")).orElse(0.0)));
-                ViewCountRateStrategyArr[i] =  Double.parseDouble(String.format("%.3f", Optional.of(scoreMap.getDoubleValue("ViewCountRateStrategy")).orElse(0.0)));
-                HisFissionDeWeightAvgReadSumRateStrategyArr[i] =  Double.parseDouble(String.format("%.3f", Optional.of(scoreMap.getDoubleValue("HisFissionDeWeightAvgReadSumRateStrategy")).orElse(0.0)));
+                HisFissionFansRateRateStrategyArr[i] = Double.parseDouble(String.format("%.3f", Optional.of(scoreMap.getDoubleValue("HisFissionFansRateRateStrategy")).orElse(0.0)));
+                HisFissionAvgReadRateRateStrategyArr[i] = Double.parseDouble(String.format("%.3f", Optional.of(scoreMap.getDoubleValue("HisFissionAvgReadRateRateStrategy")).orElse(0.0)));
+                PublishTimesStrategyArr[i] = Double.parseDouble(String.format("%.3f", Optional.of(scoreMap.getDoubleValue("PublishTimesStrategy")).orElse(0.0)));
+                ViewCountRateCorrelationStrategyArr[i] = Double.parseDouble(String.format("%.3f", Optional.of(scoreMap.getDoubleValue("ViewCountRateCorrelationStrategy")).orElse(0.0)));
+                HisFissionAvgReadSumRateStrategyArr[i] = Double.parseDouble(String.format("%.3f", Optional.of(scoreMap.getDoubleValue("HisFissionAvgReadSumRateStrategy")).orElse(0.0)));
+                HisFissionAvgReadRateCorrelationRateStrategyArr[i] = Double.parseDouble(String.format("%.3f", Optional.of(scoreMap.getDoubleValue("HisFissionAvgReadRateCorrelationRateStrategy")).orElse(0.0)));
+                HisFissionFansSumRateStrategyArr[i] = Double.parseDouble(String.format("%.3f", Optional.of(scoreMap.getDoubleValue("HisFissionFansSumRateStrategy")).orElse(0.0)));
+                SimilarityStrategyArr[i] = Double.parseDouble(String.format("%.3f", Optional.of(scoreMap.getDoubleValue("SimilarityStrategy")).orElse(0.0)));
+                ViewCountStrategyArr[i] = Double.parseDouble(String.format("%.3f", Optional.of(scoreMap.getDoubleValue("ViewCountStrategy")).orElse(0.0)));
+                ViewCountRateStrategyArr[i] = Double.parseDouble(String.format("%.3f", Optional.of(scoreMap.getDoubleValue("ViewCountRateStrategy")).orElse(0.0)));
+                HisFissionDeWeightAvgReadSumRateStrategyArr[i] = Double.parseDouble(String.format("%.3f", Optional.of(scoreMap.getDoubleValue("HisFissionDeWeightAvgReadSumRateStrategy")).orElse(0.0)));
                 Map<String, AccountAvgInfo> map = accountAvgInfoMap.get(article.getGhId());
                 if (Objects.nonNull(map)) {
                     List<String> avgMapDateList = new ArrayList<>(map.keySet());
@@ -688,7 +698,6 @@ public class RecommendTest {
     }
 
 
-
     @Test
     void getScoreFromLogFile() {
 
@@ -811,4 +820,43 @@ public class RecommendTest {
         }
     }
 
+    @Test
+    public void checkTest() {
+        List<Article> articleList = articleRepository.getByPublishTimestampGreaterThan(1732982400L);
+        List<Article> singleArticleList = new ArrayList<>();
+        for (Article article : articleList) {
+            if (Objects.nonNull(article.getRootSourceIdList())) {
+                try {
+                    List<String> rootSourceIdList = JSONArray.parseArray(article.getRootSourceIdList(), String.class);
+                    if (rootSourceIdList.size() == 1) {
+                        singleArticleList.add(article);
+                    }
+                } catch (Exception ignore) {
+                }
+            }
+        }
+        List<String> ghIds = singleArticleList.stream().map(Article::getGhId).distinct().collect(Collectors.toList());
+        List<PublishAccount> publishAccountList = publishAccountRepository.getAllByGhIdIn(ghIds);
+        log.info("newSortStrategyData publishAccountList finish");
+        Map<String, PublishAccount> publishAccountMap = publishAccountList.stream().collect(Collectors.toMap(PublishAccount::getGhId, o -> o));
+        // 获取发布内容
+        List<PublishContentParam> publishContentParamList = singleArticleList.stream().map(article -> {
+            PublishContentParam item = new PublishContentParam();
+            item.setTitle(article.getTitle());
+            PublishAccount account = publishAccountMap.get(article.getGhId());
+            if (Objects.nonNull(account)) {
+                item.setPublishAccountId(account.getId());
+                return item;
+            }
+            return null;
+        }).filter(Objects::nonNull).collect(Collectors.toList());
+        List<PublishContentDTO> publishContents = new ArrayList<>();
+        for (List<PublishContentParam> partitions : Lists.partition(publishContentParamList, 100)) {
+            publishContents.addAll(aigcBaseMapper.getPublishContentByTitle(partitions));
+        }
+        List<String> publishContentIds = publishContents.stream().map(PublishContentDTO::getId).collect(Collectors.toList());
+        for (List<String> partition : Lists.partition(publishContentIds, 500)) {
+            aigcBaseMapper.updatePublishContentSingleMiniProgram(partition);
+        }
+    }
 }