什么,你的EasyExcel導(dǎo)出一萬條數(shù)據(jù)就OOM了?
前言
前段時(shí)間在做一個(gè)導(dǎo)出的功能,本以為是平平無奇的一個(gè)功能。就用公司內(nèi)部的一個(gè)導(dǎo)出工具類三下五除二就寫完了,做法是直接查全量數(shù)據(jù),然后直接往Excel里寫。一開始沒多少數(shù)據(jù)也沒什么問題,但是當(dāng)數(shù)據(jù)量逐漸多了起來后,達(dá)到一萬多條,導(dǎo)出的時(shí)候就會(huì)報(bào)OOM。然后我就換成了阿里開源的EasyExcel,但是導(dǎo)出的時(shí)候也不太穩(wěn)定,偶爾也會(huì)OOM。所以應(yīng)該是數(shù)據(jù)量太大了,在寫入的時(shí)候把內(nèi)存占滿了。然后我就放棄了查全量數(shù)據(jù)一次性寫入Excel的做法,采用分頁查詢,分批次寫入Excel的方式,果然不會(huì)出現(xiàn)OOM了。
雖然這種方式不會(huì)出現(xiàn)OOM,但是每次導(dǎo)出都寫一遍重復(fù)的代碼著實(shí)有點(diǎn)麻煩,所以結(jié)合自己平時(shí)的使用場景,封裝了一個(gè)EasyExcel的導(dǎo)出工具類,這樣只要在分頁查詢的基礎(chǔ)上寫少量的代碼,就可以實(shí)現(xiàn)分批次寫入Excel,簡化代碼的編寫并且解決OOM的問題。
實(shí)現(xiàn)
@Slf4j
public abstract class EasyExcelExport<T, S> {
/**
* EasyExcel導(dǎo)出Excel表格,每個(gè)sheet默認(rèn)最大10萬條數(shù)據(jù)
*
* @param fileName excel文件前綴名
* @param sheetName 表頁名
*/
public void easyExcelBatchExport(String fileName, String sheetName, HttpServletResponse response) {
this.easyExcelBatchExport(fileName, sheetName, 100000, response);
}
/**
* 分批次導(dǎo)出excel數(shù)據(jù)
*
* @param fileName excel文件前綴名
* @param sheetSize 每個(gè)sheet的數(shù)據(jù)量,默認(rèn)10萬,excel有限制不能大于1048576
* @param sheetName 表頁名
*/
public void easyExcelBatchExport(String fileName, String sheetName, Integer sheetSize, HttpServletResponse response) {
fileName = fileName + LocalDateTime.now().format(DateTimeFormatter.ofPattern("yyyyMMddHHmmss")) + ".xlsx";
int currentSheet = 1; // 當(dāng)前處于第幾個(gè)sheet
int totalLine = 0; // 總共寫入的條數(shù)
int currentBatch = 1; // 當(dāng)前寫入excel的批次(第幾頁)
int lineNum = 1; // 行號,當(dāng)前寫入的是第幾條數(shù)據(jù)
long startTime = System.currentTimeMillis();
try {
response.setCharacterEncoding("utf-8");
// 告訴瀏覽器用什么軟件可以打開此文件
response.setHeader("content-Type", "application/vnd.ms-excel");
// 下載文件的默認(rèn)名稱
response.setHeader("Content-Disposition", "attachment;filename=" + URLEncoder.encode(fileName, "utf-8"));
ExcelWriter excelWriter = EasyExcel.write(response.getOutputStream(), (Class<T>) ((ParameterizedType) getClass().getGenericSuperclass()).getActualTypeArguments()[0]).build();
WriteSheet sheet = EasyExcel.writerSheet(sheetName).build();
while (true) {
// 獲取數(shù)據(jù),然后currentBatch+1,下次調(diào)用就會(huì)獲取新的數(shù)據(jù)
List<S> sourceDataList = getData(currentBatch);
currentBatch++;
List<T> exportEntityList = new ArrayList<>();
if (CollUtil.isNotEmpty(sourceDataList)) {
totalLine += sourceDataList.size();
log.info("EasyExcel開始寫入第{}批數(shù)據(jù),當(dāng)前批次數(shù)據(jù)大小為{}", currentBatch - 1, sourceDataList.size());
for (S sourceData : sourceDataList) {
exportEntityList.add(convertSourceData2ExportEntity(sourceData, lineNum));
lineNum++;
// 當(dāng)前sheet數(shù)據(jù)已經(jīng)到達(dá)最大值,將當(dāng)前數(shù)據(jù)全寫入當(dāng)前sheet,下一條數(shù)據(jù)就會(huì)寫入新sheet
if (lineNum > sheetSize) {
excelWriter.write(exportEntityList, sheet);
exportEntityList.clear();
lineNum = 1;
currentSheet++;
sheet = EasyExcel.writerSheet(sheetName + currentSheet).build();
}
}
// 寫入excel
excelWriter.write(exportEntityList, sheet);
} else {
// 未獲取到數(shù)據(jù),結(jié)束
break;
}
}
excelWriter.finish();
} catch (Exception e) {
log.error("EasyExcel導(dǎo)出異常", e);
}
log.info("EasyExcel導(dǎo)出數(shù)據(jù)結(jié)束,總數(shù)據(jù)量為{},耗時(shí){}ms", totalLine, (System.currentTimeMillis() - startTime));
}
/**
* 不分批次導(dǎo)出excel。一次性獲取所有數(shù)據(jù)寫入excel,確定數(shù)據(jù)量不大時(shí)可以使用該方法,數(shù)據(jù)量過大時(shí)使用分批次導(dǎo)出,否則會(huì)OOM
*
* @param fileName excel文件前綴名
* @param sheetName 表頁名
*/
public void easyExcelExport(String fileName, String sheetName, HttpServletResponse response) {
fileName = fileName + LocalDateTime.now().format(DateTimeFormatter.ofPattern("yyyyMMddHHmmss")) + ".xlsx";
int totalLine = 0; // 總共寫入的條數(shù)
int lineNum = 1; // 行號,當(dāng)前寫入的是第幾條數(shù)據(jù)
long startTime = System.currentTimeMillis();
try {
response.setCharacterEncoding("utf-8");
// 告訴瀏覽器用什么軟件可以打開此文件
response.setHeader("content-Type", "application/vnd.ms-excel");
// 下載文件的默認(rèn)名稱
response.setHeader("Content-Disposition", "attachment;filename=" + URLEncoder.encode(fileName, "utf-8"));
List<S> sourceDataList = getData(1);
List<T> exportEntityList = new ArrayList<>();
if (CollUtil.isNotEmpty(sourceDataList)) {
totalLine += sourceDataList.size();
log.info("EasyExcel開始寫入數(shù)據(jù),數(shù)據(jù)大小為{}", sourceDataList.size());
for (S sourceData : sourceDataList) {
exportEntityList.add(convertSourceData2ExportEntity(sourceData, lineNum));
lineNum++;
}
}
response.setCharacterEncoding("utf-8");
// 告訴瀏覽器用什么軟件可以打開此文件
response.setHeader("content-Type", "application/vnd.ms-excel");
// 下載文件的默認(rèn)名稱
response.setHeader("Content-Disposition", "attachment;filename=" + URLEncoder.encode(fileName, "utf-8"));
EasyExcel.write(response.getOutputStream(), (Class<T>) ((ParameterizedType) getClass().getGenericSuperclass()).getActualTypeArguments()[0]).sheet(sheetName).doWrite(exportEntityList);
} catch (Exception e) {
log.error("EasyExcel導(dǎo)出異常", e);
}
log.info("EasyExcel導(dǎo)出數(shù)據(jù)結(jié)束,總數(shù)據(jù)量為{},耗時(shí){}ms", totalLine, (System.currentTimeMillis() - startTime));
}
/**
* 將原數(shù)據(jù)對象轉(zhuǎn)換為需要導(dǎo)出的目標(biāo)對象
*
* @param sourceData 原對象
* @param lineNum 行號
*/
public abstract T convertSourceData2ExportEntity(S sourceData, Integer lineNum);
/**
* 獲取原始數(shù)據(jù),通過currentBatch參數(shù)分頁獲取數(shù)據(jù)。
*
* @param currentBatch 獲取第幾批(頁)數(shù)據(jù),通過該參數(shù)分頁查詢,每次調(diào)用自動(dòng)遞增。不分批次導(dǎo)出時(shí)可以忽略該參數(shù)
*/
public abstract List<S> getData(int currentBatch);
}
首先,這是EasyExcelExport是一個(gè)抽象類,指定了泛型 T 和 S,T是target目標(biāo)類,也就是導(dǎo)出時(shí)對應(yīng)的類,S是source原對象所對應(yīng)的類。
EasyExcelExport里還有兩個(gè)抽象方法,getData() 和 convertSourceData2ExportEntity() 。這兩個(gè)方法是需要在平時(shí)使用時(shí)自己去實(shí)現(xiàn)的,getData是數(shù)據(jù)查詢的方法,可以在這里面去實(shí)現(xiàn)分頁查詢的邏輯,currentBatch參數(shù)是用來控制分頁查詢頁碼的,從1開始,會(huì)自動(dòng)遞增。如果確定數(shù)據(jù)量不大不需要分批次導(dǎo)出的話,那么getData()里只需要進(jìn)行普通的查詢即可,忽略currentBatch參數(shù)不用分頁查詢。還有一個(gè)方法是convertSourceData2ExportEntity(),這個(gè)是用來將對象S轉(zhuǎn)為對象T的方法,因?yàn)閺臄?shù)據(jù)庫查詢或者是從其他地方獲取到的對象類型可能是S,而導(dǎo)出時(shí)需要的對象類型是T,所以通過該方法進(jìn)行對象轉(zhuǎn)換。
最核心的是 easyExcelBatchExport() 方法,里面有一個(gè)while循環(huán),while循環(huán)里首先會(huì)去調(diào)用getData()方法獲取數(shù)據(jù),然后將currentBatch加1便于下次獲取數(shù)據(jù),接下來有個(gè)for循環(huán)去進(jìn)行對象的轉(zhuǎn)換并添加到exportEntityList集合中,這個(gè)集合中裝的是最終寫到Excel里的對象。當(dāng)轉(zhuǎn)換完成后就將當(dāng)前批次的數(shù)據(jù)寫入Excel中,然后進(jìn)行下一次循環(huán),當(dāng)getData()方法未獲取到數(shù)據(jù)時(shí),就結(jié)束循環(huán)。
同時(shí)支持指定每個(gè)sheet頁的最大行數(shù)。在對對象進(jìn)行轉(zhuǎn)換時(shí)有一個(gè)判斷,當(dāng)前sheet頁的數(shù)據(jù)是否到達(dá)指定值,到達(dá)后,直接寫入excel,然后新建一個(gè)sheet頁,這樣新的數(shù)據(jù)就會(huì)寫入新的sheet頁。
使用
那么如何使用這個(gè)工具類呢。很簡單,只要new出EasyExcelExport的對象,然后實(shí)現(xiàn)一下 convertSourceData2ExportEntity() 方法和 getData() 方法即可,然后再根據(jù)需要去調(diào)用不同的導(dǎo)出方法即可。導(dǎo)出方法有指定和不指定sheet數(shù)據(jù)頁大小的分批寫入方法 easyExcelBatchExport() 和不分批次直接一次性寫入的 easyExcelExport() 方法。
下面通過一個(gè)小案例展示一下。假設(shè)現(xiàn)在有個(gè)導(dǎo)出用戶列表的需求,數(shù)據(jù)庫User表對應(yīng)的是UserPO類:
@Data
@Builder
@NoArgsConstructor
@AllArgsConstructor
public class UserPO {
private Long id;
/**
* 用戶編號
*/
private String code;
/**
* 姓名
*/
private String name;
/**
* 手機(jī)號碼
*/
private String phone;
/**
* 性別。1-男,2-女
*/
private Integer sex;
}
導(dǎo)出對應(yīng)的類是UserExportEntity:
@Data
public class UserExportEntity {
@ColumnWidth(10)
@ContentStyle(horizontalAlignment = HorizontalAlignmentEnum.CENTER, verticalAlignment = VerticalAlignmentEnum.CENTER)
@ExcelProperty(index = 0, value = "序號")
private Integer line;
@ColumnWidth(35)
@ContentStyle(horizontalAlignment = HorizontalAlignmentEnum.CENTER, verticalAlignment = VerticalAlignmentEnum.CENTER)
@ExcelProperty(index = 1, value = "用戶編號")
private String code;
@ColumnWidth(35)
@ContentStyle(horizontalAlignment = HorizontalAlignmentEnum.CENTER, verticalAlignment = VerticalAlignmentEnum.CENTER)
@ExcelProperty(index = 2, value = "姓名")
private String name;
@ColumnWidth(35)
@ContentStyle(horizontalAlignment = HorizontalAlignmentEnum.CENTER, verticalAlignment = VerticalAlignmentEnum.CENTER)
@ExcelProperty(index = 3, value = "手機(jī)號碼")
private String phone;
@ColumnWidth(10)
@ContentStyle(horizontalAlignment = HorizontalAlignmentEnum.CENTER, verticalAlignment = VerticalAlignmentEnum.CENTER)
@ExcelProperty(index = 4, value = "性別")
private String sexStr;
@ColumnWidth(10)
@ContentStyle(horizontalAlignment = HorizontalAlignmentEnum.CENTER, verticalAlignment = VerticalAlignmentEnum.CENTER)
@ExcelProperty(index = 5, value = "fieldA")
private String fieldA;
@ColumnWidth(10)
@ContentStyle(horizontalAlignment = HorizontalAlignmentEnum.CENTER, verticalAlignment = VerticalAlignmentEnum.CENTER)
@ExcelProperty(index = 6, value = "fieldB")
private String fieldB;
@ColumnWidth(10)
@ContentStyle(horizontalAlignment = HorizontalAlignmentEnum.CENTER, verticalAlignment = VerticalAlignmentEnum.CENTER)
@ExcelProperty(index = 7, value = "fieldC")
private String fieldC;
@ColumnWidth(10)
@ContentStyle(horizontalAlignment = HorizontalAlignmentEnum.CENTER, verticalAlignment = VerticalAlignmentEnum.CENTER)
@ExcelProperty(index = 8, value = "fieldD")
private String fieldD;
@ColumnWidth(10)
@ContentStyle(horizontalAlignment = HorizontalAlignmentEnum.CENTER, verticalAlignment = VerticalAlignmentEnum.CENTER)
@ExcelProperty(index = 9, value = "fieldE")
private String fieldE;
@ColumnWidth(10)
@ContentStyle(horizontalAlignment = HorizontalAlignmentEnum.CENTER, verticalAlignment = VerticalAlignmentEnum.CENTER)
@ExcelProperty(index = 10, value = "fieldF")
private String fieldF;
@ColumnWidth(10)
@ContentStyle(horizontalAlignment = HorizontalAlignmentEnum.CENTER, verticalAlignment = VerticalAlignmentEnum.CENTER)
@ExcelProperty(index = 11, value = "fieldG")
private String fieldG;
@ColumnWidth(10)
@ContentStyle(horizontalAlignment = HorizontalAlignmentEnum.CENTER, verticalAlignment = VerticalAlignmentEnum.CENTER)
@ExcelProperty(index = 12, value = "fieldH")
private String fieldH;
@ColumnWidth(10)
@ContentStyle(horizontalAlignment = HorizontalAlignmentEnum.CENTER, verticalAlignment = VerticalAlignmentEnum.CENTER)
@ExcelProperty(index = 13, value = "fieldI")
private String fieldI;
}
先測試一下不分批次導(dǎo)出,導(dǎo)出123456條數(shù)據(jù)。
@GetMapping("/testExport")
public void testExport(HttpServletResponse response) {
new EasyExcelExport<UserExportEntity, UserPO>() {
@Override
public UserExportEntity convertSourceData2ExportEntity(UserPO sourceData, Integer lineNum) {
UserExportEntity entity = new UserExportEntity();
entity.setLine(lineNum);
entity.setCode(sourceData.getCode());
entity.setName(sourceData.getName());
entity.setPhone(sourceData.getPhone());
entity.setSexStr(Objects.equals(sourceData.getSex(), 1) ? "男" : Objects.equals(sourceData.getSex(), 2) ? "女" : StrUtil.EMPTY);
return entity;
}
@Override
public List<UserPO> getData(int currentBatch) {
List<UserPO> userPOList = new ArrayList<>();
// 模擬查詢數(shù)據(jù)庫,假設(shè)每次查詢會(huì)查出123456條數(shù)據(jù)
for (int i = 0; i < 123456; i++) {
userPOList.add(UserPO.builder()
.code("USER_" + RandomUtil.randomString("1234567890", 6))
.name(RandomUtil.randomString("qwertyuiopasdfghjklzxcvbnm", 10))
.phone("138" + RandomUtil.randomString("1234567890", 8))
.sex(RandomUtil.randomInt(1, 3))
.build());
}
log.info("userPOList-->{}", JSONUtil.toJsonStr(userPOList));
return userPOList;
}
}.easyExcelExport("測試不分批次導(dǎo)出", "測試不分批次導(dǎo)出", response);
}
為了更清晰地看到效果,我將內(nèi)存大小限制為128M。
調(diào)用一下測試接口,可以看到,導(dǎo)出十幾萬條數(shù)據(jù)時(shí)發(fā)生了OOM。
再來看看分批次導(dǎo)出的效果,模擬一下分頁查詢,假設(shè)有200頁數(shù)據(jù),每頁8888條,一共是170多萬條數(shù)據(jù)。
@GetMapping("/testBatchExport")
public void testBatchExport(HttpServletResponse response) {
new EasyExcelExport<UserExportEntity, UserPO>() {
@Override
public UserExportEntity convertSourceData2ExportEntity(UserPO sourceData, Integer lineNum) {
UserExportEntity entity = new UserExportEntity();
entity.setLine(lineNum);
entity.setCode(sourceData.getCode());
entity.setName(sourceData.getName());
entity.setPhone(sourceData.getPhone());
entity.setSexStr(Objects.equals(sourceData.getSex(), 1) ? "男" : Objects.equals(sourceData.getSex(), 2) ? "女" : StrUtil.EMPTY);
return entity;
}
@Override
public List<UserPO> getData(int currentBatch) {
// 模擬分頁查詢,假設(shè)數(shù)據(jù)庫中有200頁數(shù)據(jù)
if (currentBatch <= 200) {
List<UserPO> userPOList = new ArrayList<>();
// 模擬查詢數(shù)據(jù)庫,假設(shè)每次查詢會(huì)查出8888條數(shù)據(jù)
for (int i = 0; i < 8888; i++) {
userPOList.add(UserPO.builder()
.code("USER_" + RandomUtil.randomString("1234567890", 6))
.name(RandomUtil.randomString("qwertyuiopasdfghjklzxcvbnm", 10))
.phone("138" + RandomUtil.randomString("1234567890", 8))
.sex(RandomUtil.randomInt(1, 3))
.build());
}
return userPOList;
} else {
return new ArrayList<>();
}
}
}.easyExcelBatchExport("測試分批次導(dǎo)出", "測試分批次導(dǎo)出", response);
}
通過分批次寫入Excel的方式,成功導(dǎo)出了170多萬條數(shù)據(jù),相較于不分批次導(dǎo)出,效果顯而易見。而且通過調(diào)用工具類的方式,進(jìn)一步簡化了導(dǎo)出時(shí)代碼的編寫。