학습결과 파일 베스트 에폭 제외 삭제 추가, 납품데이터 등록 비동기 수정

This commit is contained in:
2026-03-27 09:32:49 +09:00
parent 04eddfce54
commit 50c965cb79
14 changed files with 884 additions and 204 deletions

View File

@@ -1,12 +1,12 @@
package com.kamco.cd.training.dataset;
import com.kamco.cd.training.config.api.ApiResponseDto;
import com.kamco.cd.training.config.api.ApiResponseDto.ResponseObj;
import com.kamco.cd.training.dataset.dto.DatasetDto;
import com.kamco.cd.training.dataset.dto.DatasetDto.AddDeliveriesReq;
import com.kamco.cd.training.dataset.dto.DatasetObjDto;
import com.kamco.cd.training.dataset.dto.DatasetObjDto.DatasetClass;
import com.kamco.cd.training.dataset.dto.DatasetObjDto.DatasetStorage;
import com.kamco.cd.training.dataset.service.DatasetAsyncService;
import com.kamco.cd.training.dataset.service.DatasetService;
import com.kamco.cd.training.model.dto.FileDto.FoldersDto;
import com.kamco.cd.training.model.dto.FileDto.SrchFoldersDto;
@@ -37,6 +37,7 @@ import org.springframework.web.bind.annotation.*;
public class DatasetApiController {
private final DatasetService datasetService;
private final DatasetAsyncService datasetAsyncService;
@Operation(summary = "학습데이터 관리 목록 조회", description = "학습데이터 목록을 조회합니다.")
@ApiResponses(
@@ -262,7 +263,7 @@ public class DatasetApiController {
content =
@Content(
mediaType = "application/json",
schema = @Schema(implementation = Page.class))),
schema = @Schema(implementation = FoldersDto.class))),
@ApiResponse(responseCode = "404", description = "조회 오류", content = @Content),
@ApiResponse(responseCode = "500", description = "서버 오류", content = @Content)
})
@@ -281,12 +282,13 @@ public class DatasetApiController {
content =
@Content(
mediaType = "application/json",
schema = @Schema(implementation = Page.class))),
schema = @Schema(implementation = String.class))),
@ApiResponse(responseCode = "404", description = "조회 오류", content = @Content),
@ApiResponse(responseCode = "500", description = "서버 오류", content = @Content)
})
@PostMapping("/deliveries")
public ApiResponseDto<ResponseObj> insertDeliveriesDataset(@RequestBody AddDeliveriesReq req) {
return ApiResponseDto.createOK(datasetService.insertDeliveriesDataset(req));
public ApiResponseDto<String> insertDeliveriesDataset(@RequestBody AddDeliveriesReq req) {
datasetAsyncService.insertDeliveriesDatasetAsync(req);
return ApiResponseDto.createOK("ok");
}
}

View File

@@ -0,0 +1,134 @@
package com.kamco.cd.training.dataset.service;
import com.kamco.cd.training.common.enums.LearnDataRegister;
import com.kamco.cd.training.dataset.dto.DatasetDto.AddDeliveriesReq;
import com.kamco.cd.training.dataset.dto.DatasetDto.DatasetMngRegDto;
import com.kamco.cd.training.postgres.core.DatasetCoreService;
import java.util.UUID;
import lombok.RequiredArgsConstructor;
import lombok.extern.log4j.Log4j2;
import org.springframework.scheduling.annotation.Async;
import org.springframework.stereotype.Service;
@Service
@Log4j2
@RequiredArgsConstructor
public class DatasetAsyncService {
private final DatasetService datasetService;
private final DatasetCoreService datasetCoreService;
private static final String LOG_PREFIX = "[납품 데이터셋]";
/**
* 납품 데이터셋 등록 비동기 업로드 처리 1) 데이터셋 구조/파일 검증 2) UID 생성 및 마스터 데이터 저장 3) 상태를 UPLOADING으로 변경 4) 실제
* 데이터(train/val/test) 등록 5) 완료 시 COMPLETED 상태로 변경 6) 실패 시 상태를 UPLOAD_FAILED로 변경 후 데이터 정리(삭제)
*
* @param req
*/
@Async("datasetExecutor")
public void insertDeliveriesDatasetAsync(AddDeliveriesReq req) {
long startTime = System.currentTimeMillis();
log.info("{} 업로드 시작 ==========", LOG_PREFIX);
log.info(
"{} filePath={}, targetYyyy={}, compareYyyy={}, roundNo={}",
LOG_PREFIX,
req.getFilePath(),
req.getTargetYyyy(),
req.getCompareYyyy(),
req.getRoundNo());
Long datasetUid = null;
try {
// ===== 1. 폴더/파일 검증 =====
long validateStart = System.currentTimeMillis();
// 폴더 구조 검증
DatasetService.validateTrainValTestDirs(req.getFilePath());
// 파일 개수 검증
DatasetService.validateDirFileCount(req.getFilePath());
log.info("{} 데이터셋 검증 완료. ({} ms)", LOG_PREFIX, System.currentTimeMillis() - validateStart);
// ===== 2. UID 생성 =====
String uid = UUID.randomUUID().toString().replace("-", "").toUpperCase();
log.info("{} 생성된 UID: {}", LOG_PREFIX, uid);
// ===== 3. 마스터 데이터 생성 =====
String title = req.getTitle();
if (title == null || title.isBlank()) {
Integer compareYyyy = req.getCompareYyyy();
Integer targetYyyy = req.getTargetYyyy();
if (compareYyyy != null && targetYyyy != null) {
title = compareYyyy + "-" + targetYyyy;
} else {
title = null;
}
}
DatasetMngRegDto datasetMngRegDto = new DatasetMngRegDto();
datasetMngRegDto.setUid(uid);
datasetMngRegDto.setDataType("DELIVER");
datasetMngRegDto.setCompareYyyy(req.getCompareYyyy() == null ? 0 : req.getCompareYyyy());
datasetMngRegDto.setTargetYyyy(req.getTargetYyyy() == null ? 0 : req.getTargetYyyy());
datasetMngRegDto.setRoundNo(req.getRoundNo());
datasetMngRegDto.setTitle(title);
datasetMngRegDto.setMemo(req.getMemo());
datasetMngRegDto.setDatasetPath(req.getFilePath());
// 마스터 저장
datasetUid = datasetCoreService.insertDatasetMngData(datasetMngRegDto);
log.info("{} 마스터 저장 완료. datasetUid={}", LOG_PREFIX, datasetUid);
// ===== 4. 상태 변경 (업로드중) =====
datasetCoreService.updateDatasetUploadStatus(datasetUid, LearnDataRegister.UPLOADING);
log.info("{} 상태 변경 → UPLOADING. datasetUid={}", LOG_PREFIX, datasetUid);
// ===== 5. 데이터 등록 =====
long insertStart = System.currentTimeMillis();
// 납품 데이터 obj 등록
datasetService.insertDeliveriesDataset(req, datasetUid);
log.info(
"{} 데이터 등록 완료. datasetUid={}, 소요시간={} ms",
LOG_PREFIX,
datasetUid,
System.currentTimeMillis() - insertStart);
// ===== 6. 상태 변경 (완료) =====
datasetCoreService.updateDatasetUploadStatus(datasetUid, LearnDataRegister.COMPLETED);
log.info("{} 상태 변경 → COMPLETED. datasetUid={}", LOG_PREFIX, datasetUid);
log.info(
"{} 업로드 완료. 총 소요시간={} ms ==========", LOG_PREFIX, System.currentTimeMillis() - startTime);
} catch (Exception e) {
log.error(
"{} 업로드 실패. datasetUid={}, filePath={}", LOG_PREFIX, datasetUid, req.getFilePath(), e);
if (datasetUid != null) {
try {
// ===== 실패 처리 =====
datasetCoreService.updateDatasetUploadStatus(datasetUid, LearnDataRegister.UPLOAD_FAILED);
log.error("{} 상태 변경 → 업로드 실패. datasetUid={}", LOG_PREFIX, datasetUid);
// 실패 시 데이터 정리
datasetCoreService.deleteAllDatasetObj(datasetUid);
log.error("{} 데이터 정리 완료. datasetUid={}", LOG_PREFIX, datasetUid);
} catch (Exception ex) {
log.error("{} 실패 후 정리 작업 중 오류. datasetUid={}", LOG_PREFIX, datasetUid, ex);
}
}
}
}
}

View File

@@ -0,0 +1,152 @@
package com.kamco.cd.training.dataset.service;
import com.fasterxml.jackson.databind.JsonNode;
import com.fasterxml.jackson.databind.ObjectMapper;
import com.fasterxml.jackson.databind.node.ArrayNode;
import com.fasterxml.jackson.databind.node.ObjectNode;
import com.kamco.cd.training.dataset.dto.DatasetObjDto.DatasetObjRegDto;
import com.kamco.cd.training.postgres.core.DatasetCoreService;
import java.nio.file.Paths;
import java.util.List;
import java.util.Map;
import lombok.RequiredArgsConstructor;
import lombok.extern.log4j.Log4j2;
import org.springframework.stereotype.Service;
import org.springframework.transaction.annotation.Transactional;
@Service
@Log4j2
@RequiredArgsConstructor
public class DatasetBatchService {
private final DatasetCoreService datasetCoreService;
private final ObjectMapper mapper;
/**
* 배치 단위 데이터 저장
*
* <p>- 전달받은 데이터 목록을 순회하며 개별 insert 처리 - batch 단위로 트랜잭션 관리
*/
@Transactional
public void saveBatch(List<Map<String, Object>> batch, Long datasetUid, String type) {
for (Map<String, Object> map : batch) {
insertTrainTestData(map, datasetUid, type);
}
}
/**
* 단일 데이터 처리 및 insert DTO 생성
*
* <p>처리 흐름: 1) 경로/JSON 데이터 추출 2) 파일명에서 연도 및 도엽번호 파싱 3) label JSON → feature 단위 분리 4) feature별 DTO
* 생성 후 DB insert
*/
private void insertTrainTestData(Map<String, Object> map, Long datasetUid, String subDir) {
String comparePath = (String) map.get("input1");
String targetPath = (String) map.get("input2");
String labelPath = (String) map.get("label");
String geojsonPath = (String) map.get("geojson_path");
Object labelJson = map.get("label-json");
// JSON 파싱
JsonNode json = parseJson(labelJson);
// 파일명 파싱
String fileName = Paths.get(comparePath).getFileName().toString();
String[] fileNameStr = fileName.split("_");
if (fileNameStr.length < 4) {
log.error("파일명 파싱 실패: {}", fileName);
throw new IllegalArgumentException("잘못된 파일명 형식: " + fileName);
}
int compareYyyy = parseInt(fileNameStr[1], "compareYyyy", fileName);
int targetYyyy = parseInt(fileNameStr[2], "targetYyyy", fileName);
String mapSheetNum = fileNameStr[3];
// JSON 유효성 체크
JsonNode featuresNode = json.path("features");
if (featuresNode.isMissingNode() || !featuresNode.isArray() || featuresNode.isEmpty()) {
return; // skip
}
ObjectNode base = mapper.createObjectNode();
base.put("type", "FeatureCollection");
for (JsonNode feature : featuresNode) {
JsonNode prop = feature.path("properties");
String compareClassCd = prop.path("before").asText(null);
String targetClassCd = prop.path("after").asText(null);
ArrayNode arr = mapper.createArrayNode();
arr.add(feature);
ObjectNode root = base.deepCopy();
root.set("features", arr);
DatasetObjRegDto objRegDto =
DatasetObjRegDto.builder()
.datasetUid(datasetUid)
.compareYyyy(compareYyyy)
.compareClassCd(compareClassCd)
.targetYyyy(targetYyyy)
.targetClassCd(targetClassCd)
.comparePath(comparePath)
.targetPath(targetPath)
.labelPath(labelPath)
.mapSheetNum(mapSheetNum)
.geojson(root)
.geojsonPath(geojsonPath)
.fileName(fileName)
.build();
// 데이터 타입별 insert
insertByType(subDir, objRegDto);
}
}
/** 데이터 타입별 insert 처리 - type 값에 따라 대상 테이블 분기 - 잘못된 타입 입력 시 예외 발생 */
private void insertByType(String type, DatasetObjRegDto dto) {
switch (type) {
case "train" -> datasetCoreService.insertDatasetObj(dto);
case "val" -> datasetCoreService.insertDatasetValObj(dto);
case "test" -> datasetCoreService.insertDatasetTestObj(dto);
default -> throw new IllegalArgumentException("잘못된 타입: " + type);
}
}
/**
* label_json → JsonNode 변환
*
* <p>- JsonNode면 그대로 사용 - 문자열이면 파싱 수행 - 실패 시 로그 후 예외 발생
*/
private JsonNode parseJson(Object labelJson) {
try {
if (labelJson instanceof JsonNode jn) {
return jn;
}
return mapper.readTree(labelJson.toString());
} catch (Exception e) {
log.error("label_json parse error: {}", labelJson, e);
throw new RuntimeException("label_json parse error", e);
}
}
/**
* 문자열 → 정수 변환
*
* <p>- 파싱 실패 시 어떤 필드/파일에서 발생했는지 로그 기록 - 잘못된 데이터는 즉시 예외 처리
*/
private int parseInt(String value, String field, String fileName) {
try {
return Integer.parseInt(value);
} catch (NumberFormatException e) {
log.error("{} 파싱 실패. fileName={}, value={}", field, fileName, value);
throw new IllegalArgumentException(field + " 파싱 실패: " + fileName);
}
}
}

View File

@@ -4,6 +4,7 @@ import com.fasterxml.jackson.databind.JsonNode;
import com.fasterxml.jackson.databind.ObjectMapper;
import com.fasterxml.jackson.databind.node.ArrayNode;
import com.fasterxml.jackson.databind.node.ObjectNode;
import com.kamco.cd.training.common.enums.LearnDataRegister;
import com.kamco.cd.training.common.enums.LearnDataType;
import com.kamco.cd.training.common.exception.CustomApiException;
import com.kamco.cd.training.common.service.FormatStorage;
@@ -22,8 +23,6 @@ import com.kamco.cd.training.dataset.dto.DatasetObjDto.SearchReq;
import com.kamco.cd.training.model.dto.FileDto.FoldersDto;
import com.kamco.cd.training.model.dto.FileDto.SrchFoldersDto;
import com.kamco.cd.training.postgres.core.DatasetCoreService;
import jakarta.persistence.EntityManager;
import jakarta.persistence.PersistenceContext;
import jakarta.validation.Valid;
import java.io.File;
import java.io.IOException;
@@ -45,7 +44,6 @@ import java.util.stream.Collectors;
import java.util.stream.Stream;
import lombok.RequiredArgsConstructor;
import lombok.extern.slf4j.Slf4j;
import org.springframework.beans.factory.annotation.Value;
import org.springframework.core.io.InputStreamResource;
import org.springframework.core.io.Resource;
import org.springframework.data.domain.Page;
@@ -62,11 +60,7 @@ import org.springframework.transaction.annotation.Transactional;
public class DatasetService {
private final DatasetCoreService datasetCoreService;
@PersistenceContext private EntityManager em;
private final ObjectMapper mapper = new ObjectMapper();
@Value("${file.dataset-dir}")
private String datasetDir;
private final DatasetBatchService datasetBatchService;
private static final List<String> LABEL_DIRS = List.of("label-json", "label", "input1", "input2");
private static final List<String> REQUIRED_DIRS = Arrays.asList("train", "val", "test");
@@ -143,7 +137,6 @@ public class DatasetService {
return datasetCoreService.searchDatasetObjectList(searchReq);
}
@Transactional
public UUID deleteDatasetObjByUuid(UUID uuid) {
return datasetCoreService.deleteDatasetObjByUuid(uuid);
}
@@ -241,7 +234,7 @@ public class DatasetService {
return new ResponseObj(ApiResponseCode.INTERNAL_SERVER_ERROR, e.getMessage());
}
datasetCoreService.updateDatasetUploadStatus(datasetUid);
datasetCoreService.updateDatasetUploadStatus(datasetUid, LearnDataRegister.COMPLETED);
return new ResponseObj(ApiResponseCode.OK, "업로드 성공하였습니다.");
}
@@ -616,146 +609,65 @@ public class DatasetService {
* 납품 데이터 등록
*
* @param req 폴더경로, 메모
* @return 성공/실패 여부
* @return 성공/실패 여부0
*/
public ResponseObj insertDeliveriesDataset(AddDeliveriesReq req) {
public void insertDeliveriesDataset(AddDeliveriesReq req, Long datasetUid) {
long startTime = System.currentTimeMillis();
log.info("========== 납품 데이터셋 업로드 시작 ==========");
log.info("filePath: {}", req.getFilePath());
DatasetMngRegDto datasetMngRegDto = new DatasetMngRegDto();
String uid = UUID.randomUUID().toString().replace("-", "").toUpperCase();
datasetMngRegDto.setUid(uid);
datasetMngRegDto.setDataType("DELIVER");
datasetMngRegDto.setCompareYyyy(req.getCompareYyyy() == null ? 0 : req.getCompareYyyy());
datasetMngRegDto.setTargetYyyy(req.getTargetYyyy() == null ? 0 : req.getTargetYyyy());
datasetMngRegDto.setRoundNo(req.getRoundNo() == null ? null : req.getRoundNo());
datasetMngRegDto.setTitle(req.getTitle() == null ? null : req.getTitle());
datasetMngRegDto.setMemo(req.getMemo() == null ? null : req.getMemo());
datasetMngRegDto.setDatasetPath(req.getFilePath());
// 마스터 저장 (트랜잭션 내부)
Long datasetUid = datasetCoreService.insertDatasetMngData(datasetMngRegDto);
log.info("납품 Dataset 마스터 저장 완료. datasetUid: {}", datasetUid);
// 검증
validateTrainValTestDirs(req.getFilePath());
validateDirFileCount(req.getFilePath());
// 처리
processType(req.getFilePath(), datasetUid, "train");
processType(req.getFilePath(), datasetUid, "val");
processType(req.getFilePath(), datasetUid, "test");
datasetCoreService.updateDatasetUploadStatus(datasetUid);
log.info("========== 전체 완료. 총 소요시간: {} ms ==========", System.currentTimeMillis() - startTime);
return new ResponseObj(ApiResponseCode.OK, "업로드 성공하였습니다.");
}
/**
* 납품 데이터 등록 처리
*
* @param path
* @param datasetUid
* @param type
*/
private void processType(String path, Long datasetUid, String type) {
long start = System.currentTimeMillis();
log.info("[{}] 데이터 처리 시작", type.toUpperCase());
log.info("[납품 데이터 등록 처리][{}] 시작", type.toUpperCase());
List<Map<String, Object>> list = getUnzipDatasetFiles(path, type);
log.info("[{}] 파일 개수: {}", type.toUpperCase(), list.size());
int batchSize = 1000;
int total = list.size();
int processed = 0;
int count = 0;
for (int i = 0; i < total; i += batchSize) {
for (Map<String, Object> map : list) {
insertTrainTestData(map, datasetUid, type);
count++;
List<Map<String, Object>> batch = list.subList(i, Math.min(i + batchSize, total));
if (count % 1000 == 0 || count == list.size()) {
log.info("[{}] 진행건수: {}", type.toUpperCase(), count);
try {
log.info("[납품 데이터 등록 처리][{}] batch 시작: {} ~ {}", type, i, i + batch.size());
datasetBatchService.saveBatch(batch, datasetUid, type);
processed += batch.size();
} catch (Exception e) {
log.error("batch 실패 row 데이터: {}", batch);
log.error(
"[납품 데이터 등록 처리][{}] batch 실패. range: {} ~ {}, datasetUid={}",
type,
i,
i + batch.size(),
datasetUid,
e);
throw e;
}
}
log.info(
"[{}] 완료. 총 {}건, 소요시간: {} ms",
type.toUpperCase(),
count,
"[납품 데이터 등록 처리][{}] 완료. 총 {}건, 소요시간: {} ms",
type,
total,
System.currentTimeMillis() - start);
}
@Transactional
public void insertTrainTestData(Map<String, Object> map, Long datasetUid, String subDir) {
String comparePath = (String) map.get("input1");
String targetPath = (String) map.get("input2");
String labelPath = (String) map.get("label");
String geojsonPath = (String) map.get("geojson_path");
Object labelJson = map.get("label-json");
JsonNode json;
try {
if (labelJson instanceof JsonNode jn) {
json = jn;
} else {
json = mapper.readTree(labelJson.toString());
}
} catch (Exception e) {
throw new RuntimeException("label_json parse error", e);
}
String fileName = Paths.get(comparePath).getFileName().toString();
String[] fileNameStr = fileName.split("_");
String compareYyyy = fileNameStr[1];
String targetYyyy = fileNameStr[2];
String mapSheetNum = fileNameStr[3];
if (json != null && json.path("features") != null && !json.path("features").isEmpty()) {
for (JsonNode feature : json.path("features")) {
JsonNode prop = feature.path("properties");
String compareClassCd = prop.path("before").asText(null);
String targetClassCd = prop.path("after").asText(null);
ObjectNode root = mapper.createObjectNode();
root.put("type", "FeatureCollection");
ArrayNode features = mapper.createArrayNode();
features.add(feature);
root.set("features", features);
DatasetObjRegDto objRegDto =
DatasetObjRegDto.builder()
.datasetUid(datasetUid)
.compareYyyy(Integer.parseInt(compareYyyy))
.compareClassCd(compareClassCd)
.targetYyyy(Integer.parseInt(targetYyyy))
.targetClassCd(targetClassCd)
.comparePath(comparePath)
.targetPath(targetPath)
.labelPath(labelPath)
.mapSheetNum(mapSheetNum)
.geojson(root)
.geojsonPath(geojsonPath)
.fileName(fileName)
.build();
// insert
if (subDir.equals("train")) {
datasetCoreService.insertDatasetObj(objRegDto);
} else if (subDir.equals("val")) {
datasetCoreService.insertDatasetValObj(objRegDto);
} else {
datasetCoreService.insertDatasetTestObj(objRegDto);
}
}
}
}
}