查重bug fix

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

@@ -18,6 +18,7 @@ import java.util.List;
 @Slf4j
 public class ThemeTextCheckService {
 
+    public static final int SENTENCE_MIN_LENGTH = 5;
     @Resource
     private ThemeCheckDuplicateMapper themeCheckDuplicateMapper;
 
@@ -37,8 +38,9 @@ public class ThemeTextCheckService {
             String[] split = content.split("。");
             for (int i = 0; i < split.length; i++) {
                 String trim = StringUtils.trim(split[i]);
+                if (trim.length() > SENTENCE_MIN_LENGTH) {
                     insert(themeId, trim, getHash(trim), i + 1, split.length, userId, themeType);
-
+                }
             }
         } catch (Exception e) {
             log.error("文本查重insert失败,themeId:" + themeId);
@@ -75,8 +77,15 @@ public class ThemeTextCheckService {
         List<Integer> list = new ArrayList<>();
         for (int i = 0; i < split.length; i++) {
             String trim = StringUtils.trim(split[i]);
+            if (trim.length() > SENTENCE_MIN_LENGTH) {
                 list.add(getHash(trim));
             }
+        }
+
+        if (list.size() <= 2) {
+            return false;
+        }
+
         // 一年以内
         LambdaQueryWrapper<ThemeCheckDuplicateEntity> w = new LambdaQueryWrapper<ThemeCheckDuplicateEntity>().in(ThemeCheckDuplicateEntity::getPartitionHash, list)
                 .eq(ThemeCheckDuplicateEntity::getDeleteTag, BizStatus.DeleteTag.tag_init)
@@ -84,7 +93,7 @@ public class ThemeTextCheckService {
                 .groupBy(ThemeCheckDuplicateEntity::getThemeId);
         Integer check = themeCheckDuplicateMapper.check(w);
         // 重复率大于80%
-        if (check != null && check * 10 >= split.length * 8) {
+        if (check != null && check >= list.size() * 0.8) {
             return true;
         }
         return false;
@@ -114,7 +123,7 @@ public class ThemeTextCheckService {
         String[] split = content.split("。");
         for (int i = 0; i < split.length; i++) {
             String trim = StringUtils.trim(split[i]);
-
+            if (trim.length() > SENTENCE_MIN_LENGTH) {
                 ThemeCheckDuplicateEntity build = ThemeCheckDuplicateEntity.builder()
                         .themeId(themeId)
                         .partitionText(trim)
@@ -129,6 +138,8 @@ public class ThemeTextCheckService {
                 themeCheckDuplicateMapper.insert(build);
             }
 
+        }
+
 
     }