查重

community-service/src/main/java/com/tanpu/community/service/ThemeTextCheckService.java

@@ -5,6 +5,7 @@ import com.tanpu.common.constant.BizStatus;
 import com.tanpu.common.exception.BizException;
 import com.tanpu.community.dao.entity.community.ThemeCheckDuplicateEntity;
 import com.tanpu.community.dao.mapper.community.ThemeCheckDuplicateMapper;
+import com.tanpu.community.util.TimeUtils;
 import org.apache.commons.lang3.StringUtils;
 import org.springframework.stereotype.Service;
 
@@ -24,22 +25,22 @@ public class ThemeTextCheckService {
     private final String regex = "[，。！？；;：, ]";
 
     // 插入
-    public void insert(String content, String themeId, String userId,Integer themeType) {
+    public void insert(String content, String themeId, String userId, Integer themeType) {
         themeCheckDuplicateMapper.deleteByThemeId(themeId);
         content = content.replaceAll(regex, "。");
         String[] split = content.split("。");
         for (int i = 0; i < split.length; i++) {
             String trim = StringUtils.trim(split[i]);
-            insert(themeId, trim, getMD5(trim), i + 1, split.length,userId);
+            insert(themeId, trim, getMD5(trim), i + 1, split.length, userId);
         }
     }
 
     // 删除
-    public void deleteByThemeId(String themeId){
+    public void deleteByThemeId(String themeId) {
         themeCheckDuplicateMapper.deleteByThemeId(themeId);
     }
 
-    private void insert(String themeId, String partition, String partitionHash, Integer num, Integer total,String userId) {
+    private void insert(String themeId, String partition, String partitionHash, Integer num, Integer total, String userId) {
 
         ThemeCheckDuplicateEntity build = ThemeCheckDuplicateEntity.builder()
                 .themeId(themeId)
@@ -53,7 +54,8 @@ public class ThemeTextCheckService {
     }
 
     public boolean checkDuplicate(String content) {
-        if (content.length()<15){
+        // 文字数小于50不查重
+        if (content.length() < 50) {
             return false;
         }
         content = content.replaceAll(regex, "。");
@@ -63,11 +65,14 @@ public class ThemeTextCheckService {
             String trim = StringUtils.trim(split[i]);
             list.add(getMD5(trim));
         }
+        // 一年以内
         LambdaQueryWrapper<ThemeCheckDuplicateEntity> w = new LambdaQueryWrapper<ThemeCheckDuplicateEntity>().in(ThemeCheckDuplicateEntity::getPartitionHash, list)
                 .eq(ThemeCheckDuplicateEntity::getDeleteTag, BizStatus.DeleteTag.tag_init)
+                .gt(ThemeCheckDuplicateEntity::getCreateTime, TimeUtils.getDaysBefore(360))
                 .groupBy(ThemeCheckDuplicateEntity::getThemeId);
         Integer check = themeCheckDuplicateMapper.check(w);
-        if (check!=null && check*10>=split.length*8){
+        // 重复率大于80%
+        if (check != null && check * 10 >= split.length * 8) {
             return true;
         }
         return false;
@@ -89,7 +94,7 @@ public class ThemeTextCheckService {
     }
 
     // 初始化
-    public void insertInit(String content, String themeId,String userId, LocalDateTime createTime) {
+    public void insertInit(String content, String themeId, String userId, LocalDateTime createTime) {
         themeCheckDuplicateMapper.deleteByThemeId(themeId);
 
         content = content.replaceAll(regex, "。");
@@ -101,7 +106,7 @@ public class ThemeTextCheckService {
                     .themeId(themeId)
                     .partitionText(trim)
                     .partitionHash(getMD5(split[i]))
-                    .partitionNum(i+1)
+                    .partitionNum(i + 1)
                     .totalParts(split.length)
                     .userId(userId)
                     .createTime(createTime)