lukeye/doc-parser-server
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

fix

Browse Source
This commit is contained in:
luke 2025-06-21 22:02:55 +08:00
parent 9670477f51
commit b82f44543e
10 changed files with 93 additions and 17 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

7
src/main/java/com/knowledge/base/application/service/DocAppService.java
View File

@ -41,4 +41,11 @@ public interface DocAppService {
* 保存上传记录
*/
void saveOSUplodRecord(OSRecordDTO dto);
/**
* 将同路径下上传的历史同名文件，全部抛弃
* @param localRelaFilePath
* @param latestMillis --- 最新版本文件的上传时间（毫秒）
*/
int markOldSearchableRecordsAsDiscarded(String localRelaFilePath, long latestMillis);
}

7
src/main/java/com/knowledge/base/application/service/DocAppServiceImpl.java
View File

@ -3,6 +3,7 @@ package com.knowledge.base.application.service;
import cn.hutool.core.collection.CollectionUtil;
import com.google.common.collect.Lists;
import com.knowledge.base.application.exceptions.AppException;
import com.knowledge.base.domain.common.enums.SearchableStatusEnum;
import com.knowledge.base.domain.common.model.PageResult;
import com.knowledge.base.domain.doc.model.OSRecordDO;
import com.knowledge.base.domain.doc.service.iface.FileDomainService;
@ -119,4 +120,10 @@ public class DocAppServiceImpl implements DocAppService{
throw new AppException("保存对象存储记录失败！", e);
}
}
@Override
public int markOldSearchableRecordsAsDiscarded(String localRelaFilePath, long latestMillis) {
return fileDomainService.updateOldSearchableRecordsByRelaPathAndTime(localRelaFilePath, latestMillis,
SearchableStatusEnum.SUCCESS.code, SearchableStatusEnum.DISCARDED.code);
}
}

1
src/main/java/com/knowledge/base/domain/common/enums/SearchableStatusEnum.java
View File

@ -8,6 +8,7 @@ public enum SearchableStatusEnum {
INIT(0, "未开始进行导入"),
SUCCESS(1, "成功"),
FAILED(2, "导入失败"),
DISCARDED(3, "已过期，改文件不允许被检索；同人当天多次上次同名文件会进入该状态"),
PROHIBITED(99, "用户禁止文件被检索"),
;

15
src/main/java/com/knowledge/base/domain/doc/repository/iface/OSRecordRepository.java
View File

@ -28,10 +28,21 @@ public interface OSRecordRepository {
/**
* 更新上传记录中文档的可搜索状态
* @param localRelaFilePath
* @param searchableStatus
* @param oldSearchableStatus --- 为空时，表示所有数据都需要更新
* @param newSearchableStatus
* @return
*/
boolean updateSearchableStatusByRelaPath(String localRelaFilePath, int searchableStatus);
boolean updateSearchableStatusByRelaPath(String localRelaFilePath, Integer oldSearchableStatus, int newSearchableStatus);
/**
* 跟新latestMills时间戳之前的记录可搜索状态为searchableStatus
* @param localRelaFilePath
* @param oldSearchableStatus --- 为空时，表示所有数据都需要更新
* @param newSearchableStatus
* @param latestMills
* @return
*/
int updateOldSearchableRecordsByRelaPathAndTime(String localRelaFilePath, long latestMills, Integer oldSearchableStatus, int newSearchableStatus);
/**
* 批量获取

15
src/main/java/com/knowledge/base/domain/doc/service/iface/FileDomainService.java
View File

@ -35,10 +35,21 @@ public interface FileDomainService {
/**
* 更新上传记录中文档的可搜索状态
* @param localRelaFilePath
* @param searchableStatus
* @param oldSearchableStatus
* @param newSearchableStatus
* @return
*/
boolean updateSearchableStatusByRelaPath(String localRelaFilePath, int searchableStatus);
boolean updateSearchableStatusByRelaPath(String localRelaFilePath, Integer oldSearchableStatus, int newSearchableStatus);
/**
* 跟新latestMills时间戳之前的记录可搜索状态为searchableStatus
* @param localRelaFilePath
* @param oldSearchableStatus --- 为空时，表示所有数据都需要更新
* @param newSearchableStatus
* @param latestMills
* @return
*/
int updateOldSearchableRecordsByRelaPathAndTime(String localRelaFilePath, long latestMills, Integer oldSearchableStatus, int newSearchableStatus);
/**
* 批量获取上传记录

2
src/main/java/com/knowledge/base/domain/doc/service/impl/AbstractBaseFileImporter.java
View File

@ -147,7 +147,7 @@ public abstract class AbstractBaseFileImporter implements DocumentImporter {
// === Step 6: 写入 Elasticsearch 并更新 导入状态&缓存 ===
boolean saved = esGateway.saveDoc(fileEsModel);
logger.info("导入结束: relativePah: {}, status: {}", localRelaFilePath, saved);
fileDomainService.updateSearchableStatusByRelaPath(localRelaFilePath, saved ? SearchableStatusEnum.SUCCESS.code : SearchableStatusEnum.FAILED.code);
fileDomainService.updateSearchableStatusByRelaPath(localRelaFilePath, SearchableStatusEnum.INIT.code, saved ? SearchableStatusEnum.SUCCESS.code : SearchableStatusEnum.FAILED.code);
String metaJson = transToMetaJson(BeanUtil.beanToMap(fileEsModel, false, true));
fileCacheService.cacheMeta(localRelaFilePath, metaJson, ConstantConfig.FILE_META_CACHE_EXPIRED_MINUTES);

14
src/main/java/com/knowledge/base/domain/doc/service/impl/FileDomainServiceImpl.java
View File

@ -72,15 +72,17 @@ public class FileDomainServiceImpl implements FileDomainService {
return Optional.empty();
}
Optional<OSRecord> osRecordOpt = osRecordRepository.getLatestRecordByRelaPath(localRelaFilePath);
if(osRecordOpt.isEmpty()) {
return Optional.empty();
}
return Optional.of(BeanConvertUtil.convert(osRecordOpt.get(), OSRecordDO.class));
return osRecordOpt.map(osRecord -> BeanConvertUtil.convert(osRecord, OSRecordDO.class));
}
@Override
public boolean updateSearchableStatusByRelaPath(String localRelaFilePath, int searchableStatus) {
return osRecordRepository.updateSearchableStatusByRelaPath(localRelaFilePath, searchableStatus);
public boolean updateSearchableStatusByRelaPath(String localRelaFilePath, Integer oldSearchableStatus, int newSearchableStatus) {
return osRecordRepository.updateSearchableStatusByRelaPath(localRelaFilePath, oldSearchableStatus, newSearchableStatus);
}
@Override
public int updateOldSearchableRecordsByRelaPathAndTime(String localRelaFilePath, long latestMills, Integer oldSearchableStatus, int newSearchableStatus) {
return osRecordRepository.updateOldSearchableRecordsByRelaPathAndTime(localRelaFilePath, latestMills, oldSearchableStatus, newSearchableStatus);
}
@Override

3
src/main/java/com/knowledge/base/infrastructure/north/controller/FileWriteController.java
View File

@ -229,6 +229,9 @@ public class FileWriteController {
Path targetPath = Paths.get(localDir, originFileNameWithSuffix);
Files.write(targetPath, fileBytes);
// 根据localRelaFilePath删除可搜索的文件历史记录（仅保存最近的一条初始状态的数据）
docAppService.markOldSearchableRecordsAsDiscarded(localRelaFilePath, DateUtil.toMillis(now));
// 直接导入ES
Map<String, Object> extInfo = Map.of(
DocMetaPropEnum.UPLOADER.code, uploader,

39
src/main/java/com/knowledge/base/infrastructure/repository/persistence/doc/OSRecordRepositoryImpl.java
View File

@ -14,7 +14,9 @@ import com.knowledge.base.infrastructure.repository.mapper.doc.OSRecordMapper;
import lombok.RequiredArgsConstructor;
import org.springframework.stereotype.Repository;
import java.util.Date;
import java.util.List;
import java.util.Objects;
import java.util.Optional;
@Repository
@ -44,22 +46,49 @@ public class OSRecordRepositoryImpl implements OSRecordRepository {
@Override
public Optional<OSRecord> getLatestRecordByRelaPath(String localRelaFilePath) {
OSRecord fileLatestRecord = osRecordMapper.selectOne(
List<OSRecord> records = osRecordMapper.selectList(
new QueryWrapper<OSRecord>()
.eq("local_rela_file_path", localRelaFilePath)
.orderByDesc("upload_time"));
return Optional.ofNullable(fileLatestRecord);
.orderByDesc("upload_time")
.last("LIMIT 1")
);
return records.stream().findFirst();
}
@Override
public boolean updateSearchableStatusByRelaPath(String localRelaFilePath, int searchableStatus) {
public boolean updateSearchableStatusByRelaPath(String localRelaFilePath, Integer oldSearchableStatus, int newSearchableStatus) {
UpdateWrapper<OSRecord> updateWrapper = new UpdateWrapper<>();
if(Objects.nonNull(oldSearchableStatus)) {
updateWrapper.eq("local_rela_file_path", localRelaFilePath)
.set("searchable_status", searchableStatus);
.eq("searchable_status", oldSearchableStatus)
.set("searchable_status", newSearchableStatus);
} else {
updateWrapper.eq("local_rela_file_path", localRelaFilePath)
.set("searchable_status", newSearchableStatus);
}
int rows = osRecordMapper.update(null, updateWrapper);
return rows > 0;
}
@Override
public int updateOldSearchableRecordsByRelaPathAndTime(String localRelaFilePath, long latestMills, Integer oldSearchableStatus, int newSearchableStatus) {
UpdateWrapper<OSRecord> updateWrapper = new UpdateWrapper<>();
if(oldSearchableStatus != null) {
updateWrapper.eq("local_rela_file_path", localRelaFilePath)
.eq("searchable_status", oldSearchableStatus) // 筛选旧状态
.lt("upload_time", new Date(latestMills)) // upload_time 要小于当前 millis
.set("searchable_status", newSearchableStatus); // 设置为新状态
} else {
updateWrapper.eq("local_rela_file_path", localRelaFilePath)
.lt("upload_time", new Date(latestMills)) // upload_time 要小于当前 millis
.set("searchable_status", newSearchableStatus); // 设置为新状态
}
return osRecordMapper.update(null, updateWrapper);
}
@Override
public List<OSRecord> batchQueryOSRecord(List<Long> recordIds) {

5
src/main/java/com/knowledge/base/infrastructure/util/DateUtil.java
View File

@ -9,6 +9,11 @@ import java.time.ZoneId;
*/
public class DateUtil {
/**
* 转毫秒
* @param time
* @return
*/
public static long toMillis(LocalDateTime time) {
return time.atZone(ZoneId.of("Asia/Shanghai")).toInstant().toEpochMilli();
}