EasyExcel是Ailibaba团队提供的一个基于Java的、快速、简洁、解决大文件内存溢出的Excel处理工具。
在不用考虑性能、内存等因素下，快速完成Excel的读、写等功能。

官方地址：https://easyexcel.opensource.alibaba.com/

EasyExcel与SpringBoot集成

创建SpringBoot项目后引入EasyExcel依赖即可，其他依赖可按须引入。思考：这玩意为什么不是一个stater呢？

<dependency><groupId>com.alibaba</groupId><artifactId>easyexcel</artifactId><version>4.0.3</version>
</dependency>

读数据

读取数据就是将excel中的数据读到程序中，可以进行逻辑处理之后将数据根据业务需求存储到数据库、写到另一个文件中或者其他的处理均可。数据读取依据监听器实现，可通过匿名内部类的形式，也可以新建类创建监听器。问：两者的最佳应用场景是什么？

读取数据的流程

• 需要有一个excel，判断excel中的数据格式和excel的sheet页
• 创建对应的实体类，对应类型的字段，存储读取的数据
• 选用不同的读取方式即可

定义实体类

@Data
public class StockEntity {// 数据库Idprivate Long id;// 商品序号@ExcelProperty(value = "序号")private Integer productNo;// 商品编码@ExcelProperty(value = "商品编码")private String productCode;// 商品类型@ExcelProperty(value = "类型")private String productType;// 商品品牌@ExcelProperty(value = "品牌")private String productBrand;// 实物库存@ExcelProperty(value = "实物库存")private String productStock;// 成本 浮点类型@ExcelProperty(value = "成本")private BigDecimal costPrice;}

简单读取

public void readSimpleExcel(MultipartFile file) {try {InputStream inputStream = file.getInputStream();EasyExcel.read(inputStream, StockEntity.class,new PageReadListener<StockEntity>(dataList -> {for (StockEntity stockEntity : dataList) {log.info("读取到一条数据:{}", stockEntity);}})).sheet().doRead();} catch (IOException e) {throw new RuntimeException(e);}
}

自定义监听器

新建类实现ReadListener接口，泛型指定要将数据存到到哪个实体中

package com.stt.listener.read;import com.alibaba.excel.context.AnalysisContext;
import com.alibaba.excel.read.listener.ReadListener;
import com.stt.entity.StockEntity;
import lombok.extern.slf4j.Slf4j;/*** @author Jshuai* @description* @date 2024-10-29 22:14* 自定义简单数据读取的监听器*/
@Slf4j
public class SimpleReadListener implements ReadListener<StockEntity> {/*** 每读取一条数据，就会调用invoke方法* @param stockEntity* @param analysisContext*/@Overridepublic void invoke(StockEntity stockEntity, AnalysisContext analysisContext) {log.info("读取到一条数据：{}", stockEntity);}/*** 读取完成之后，会调用doAfterAllAnalysed方法，做一些数据清理的操作* @param analysisContext*/@Overridepublic void doAfterAllAnalysed(AnalysisContext analysisContext) {log.info("读取完成！");}
}

读取指定sheet和所有sheet

public void readMultiSheetExcel(MultipartFile file) {// 读取所有的// EasyExcel.read(file.getInputStream(), StockEntity.class,new SimpleReadListener()).doReadAll();// 读取指定的sheet页try (ExcelReader excelReader = EasyExcel.read(file.getInputStream()).build()) {// 这里为了简单 所以注册了 同样的head 和Listener 自己使用功能必须不同的ListenerReadSheet readSheet1 =EasyExcel.readSheet(0).head(StockEntity.class).registerReadListener(new SimpleReadListener()).build();ReadSheet readSheet2 =EasyExcel.readSheet(1).head(StockEntity.class).registerReadListener(new SimpleReadListener()).build();// 这里注意 一定要把sheet1 sheet2 一起传进去，不然有个问题就是03版的excel 会读取多次，浪费性能excelReader.read(readSheet1, readSheet2);} catch (IOException e) {throw new RuntimeException(e);}
}

多行头读取

public void readMultiHeadExcel(MultipartFile file) {try {EasyExcel.read(file.getInputStream(), StockEntity.class, new SimpleReadListener()).sheet(0)// 这里可以设置1，因为头就是一行。如果多行头，可以设置其他值。不传入也可以，因为默认会根据DemoData 来解析，他没有指定头，也就是默认1行.headRowNumber(2).doRead();} catch (IOException e) {throw new RuntimeException(e);}
}

数据格式转换

部分数据读写场景涉及到数据格式问题，需要转换，比如性别，数据库中存储的一般是数字类型【0表示男，1表示女】，时间在不同系统中也有固定的格式，EasyExcel实现类型转换有两种方式：

• 通过@DateTimeFormat和@NumberFormat对日期时间和数字格式转换
• 通过实现Converter接口实现读写数据的格式转换

以员工数据为例：

列表数据

姓名	性别	出生日期
加油鸭	男	2024/10/10
酱香鸭	女	2024/10/10
果木鸭	保密	2024/10/10

实体类

其中birthday通过easyexcel提供的注解设置格式，性别通过自定义的GenderConverter实现

public class MyUser {@ExcelProperty(value = "姓名",index = 0)private String name;@ExcelProperty(value = "性别",index = 1,converter = GenderConverter.class)private Integer gender;@ExcelProperty(value = "出生日期",index = 2)@DateTimeFormat("yyyy-MM-dd")private LocalDate birthday;
}

自定义转换器

@Slf4j
public class GenderConverter implements Converter<Integer> {/*** 这里读的时候会调用*/@Overridepublic Integer convertToJavaData(ReadCellData<?> cellData, ExcelContentProperty contentProperty, GlobalConfiguration globalConfiguration) throws Exception {// 获取读取的字符串String gender = cellData.getStringValue();log.info("gender===》{}",gender);// 转换成对应的数字return GenderEnum.getCodeByLabel(gender);}
}

自定义监听器

不同的数据实体需要不同的监听器，此处定义一个读取员工数据的监听器，泛型修改，并且提供list来存储读取到的数据，外部可以通过get方法获取集合数据

@Slf4j
public class UserReadListener implements ReadListener<MyUser> {@Getterprivate List<MyUser> data;public UserReadListener() {this.data = new ArrayList<>();}@Overridepublic void invoke(MyUser myUser, AnalysisContext analysisContext) {data.add(myUser);}@Overridepublic void doAfterAllAnalysed(AnalysisContext analysisContext) {log.info("所有数据解析完成！");}}

数据读取

public void readByConverterData(MultipartFile file) {try {// 创建监听器UserReadListener readListener = new UserReadListener();EasyExcel.read(file.getInputStream(), MyUser.class, readListener).sheet(2).doRead();// 获取读取到的数据List<MyUser> data = readListener.getData();data.forEach(System.out::println);} catch (IOException e) {throw new RuntimeException(e);}
}

写数据

写数据就是将系统内的数据写到excel中，可以将文件写到磁盘或者提供给前端下载，本案例基于MySQL +Mybatis实现从数据库中读取数据写出到excel中

简单数据写出

存储到磁盘

public void demo1(HttpServletResponse response) {List<TbExcel> tbExcels = excelMapper.selectList(null);List<ExcelVO> excelVOS = BeanUtil.copyToList(tbExcels, ExcelVO.class);EasyExcel.write("C:\\demo1.xlsx", ExcelVO.class).sheet("sheet1").doWrite(excelVOS);
}

返回前端下载

public void demo1(HttpServletResponse response) {// 查询数据库数据，查询所有的List<TbExcel> tbExcels = excelMapper.selectList(null);// 将TbExcel转换为ExcelVOList<ExcelVO> excelVOS = BeanUtil.copyToList(tbExcels, ExcelVO.class);// 提供给前端下载，需要使用到HttpServletResponse对象// 这里注意 有同学反应使用swagger 会导致各种问题，请直接用浏览器或者用postmanresponse.setContentType("application/vnd.openxmlformats-officedocument.spreadsheetml.sheet");// 强制调起下载
//        response.setContentType("application/octet-stream;charset=UTF-8");response.setCharacterEncoding("utf-8");try {String fileName = URLEncoder.encode("测试", "UTF-8").replaceAll("\\+", "%20");response.setHeader("Content-disposition", "attachment;filename*=utf-8''" + fileName + ".xlsx");EasyExcel.write(response.getOutputStream(), ExcelVO.class).sheet("模板").doWrite(excelVOS);} catch (IOException e) {throw new RuntimeException(e);}
}

写出指定列宽，和数值精度丢失问题

设置列宽

通过@ColumnWidth注解设置

public class ExcelVO {@ExcelProperty(value = "ID", index = 0)private Long id;@ExcelProperty(value = "字符串", index = 1)private String strCol;@ExcelProperty(value = "浮点数字", index = 2)@ColumnWidth(12)private BigDecimal decCol;@ExcelProperty(value = "日期时间", index = 3)@ColumnWidth(18)private LocalDateTime datetimeCol;
}

精度丢失问题

通过设置转换器实现，将Long转为String导出

EasyExcel.write(response.getOutputStream(), ExcelVO.class).registerConverter(new LongStringConverter()).sheet("模板").doWrite(excelVOS);

同一sheet写多次

public void demo2(HttpServletResponse response) {List<TbExcel> tbExcels = excelMapper.selectList(null);List<ExcelVO> excelVOS = BeanUtil.copyToList(tbExcels, ExcelVO.class);try (ExcelWriter excelWriter = EasyExcel.write(response.getOutputStream(), ExcelVO.class).build()) {// 这里注意 有伙伴反应使用swagger 会导致各种问题，请直接用浏览器或者用postmanresponse.setContentType("application/vnd.openxmlformats-officedocument.spreadsheetml.sheet");response.setCharacterEncoding("utf-8");// 这里URLEncoder.encode可以防止中文乱码 当然和easyexcel没有关系String fileName = URLEncoder.encode("测试", "UTF-8").replaceAll("\\+", "%20");response.setHeader("Content-disposition", "attachment;filename*=utf-8''" + fileName + ".xlsx");// 这里注意 如果同一个sheet只要创建一次WriteSheet writeSheet = EasyExcel.writerSheet("模板").build();// 去调用写入,这里我调用了五次，实际使用时根据数据库分页的总的页数来for (int i = 0; i < 5; i++) {excelWriter.write(excelVOS, writeSheet);}} catch (IOException e) {throw new RuntimeException(e);}
}

写出多个sheet

根据不同的维度将数据分别存到不同的sheet表中，比如根据日期分类

public void demo3(HttpServletResponse response) {List<TbExcel> tbExcels = excelMapper.selectList(null);List<ExcelVO> excelVOS = BeanUtil.copyToList(tbExcels, ExcelVO.class);try (ExcelWriter excelWriter = EasyExcel.write(response.getOutputStream(), ExcelVO.class).build()) {response.setContentType("application/vnd.openxmlformats-officedocument.spreadsheetml.sheet");response.setCharacterEncoding("utf-8");// 这里URLEncoder.encode可以防止中文乱码 当然和easyexcel没有关系String fileName = URLEncoder.encode("测试", "UTF-8").replaceAll("\\+", "%20");response.setHeader("Content-disposition", "attachment;filename*=utf-8''" + fileName + ".xlsx");// 去调用写入,这里我调用了五次，实际使用时根据数据库分页的总的页数来。这里最终会写到5个sheet里面Map<LocalDateTime, List<ExcelVO>> collect = excelVOS.stream().collect(Collectors.groupingBy(ExcelVO::getDatetimeCol));int index = 0;for (LocalDateTime localDateTime : collect.keySet()) {WriteSheet writeSheet = EasyExcel.writerSheet(index, localDateTime.format(DateTimeFormatter.ofPattern("yyyy-MM-dd"))).build();// 分页去数据库查询数据 这里可以去数据库查询每一页的数据excelWriter.write(collect.get(localDateTime), writeSheet);index++;}} catch (IOException e) {throw new RuntimeException(e);}
}

自定义样式

easyexcel导出的表格有默认样式，我们还可以自定义头，表格内容和字体样式

public void demo4(HttpServletResponse response) {// 头的策略WriteCellStyle headWriteCellStyle = new WriteCellStyle();// 背景设置为红色headWriteCellStyle.setFillForegroundColor(IndexedColors.RED.getIndex());WriteFont headWriteFont = new WriteFont();headWriteFont.setFontHeightInPoints((short)20);headWriteCellStyle.setWriteFont(headWriteFont);// 内容的策略WriteCellStyle contentWriteCellStyle = new WriteCellStyle();// 这里需要指定 FillPatternType 为FillPatternType.SOLID_FOREGROUND 不然无法显示背景颜色.头默认了 FillPatternType所以可以不指定contentWriteCellStyle.setFillPatternType(FillPatternType.SOLID_FOREGROUND);// 背景绿色contentWriteCellStyle.setFillForegroundColor(IndexedColors.GREEN.getIndex());WriteFont contentWriteFont = new WriteFont();// 字体大小contentWriteFont.setFontHeightInPoints((short)20);contentWriteCellStyle.setWriteFont(contentWriteFont);// 这个策略是 头是头的样式 内容是内容的样式 其他的策略可以自己实现HorizontalCellStyleStrategy horizontalCellStyleStrategy =new HorizontalCellStyleStrategy(headWriteCellStyle, contentWriteCellStyle);List<TbExcel> tbExcels = excelMapper.selectList(null);List<ExcelVO> excelVOS = BeanUtil.copyToList(tbExcels, ExcelVO.class);response.setContentType("application/vnd.openxmlformats-officedocument.spreadsheetml.sheet");response.setCharacterEncoding("utf-8");String fileName = null;try {fileName = URLEncoder.encode("测试", "UTF-8").replaceAll("\\+", "%20");response.setHeader("Content-disposition", "attachment;filename*=utf-8''" + fileName + ".xlsx");EasyExcel.write(response.getOutputStream(), ExcelVO.class).registerWriteHandler(horizontalCellStyleStrategy).sheet("模板").doWrite(excelVOS);} catch (IOException e) {throw new RuntimeException(e);}
}

合并单元格

通过注解方式合并

@Data
@AllArgsConstructor
@NoArgsConstructor
// 将第6-7行的2-3列合并成一个单元格
// @OnceAbsoluteMerge(firstRowIndex = 5, lastRowIndex = 6, firstColumnIndex = 1, lastColumnIndex = 2)
public class ExcelVO {// 每两行合并@ContentLoopMerge(eachRow = 2)@ExcelProperty(value = "ID", index = 0)private Long id;@ExcelProperty(value = "字符串", index = 1)private String strCol;@ExcelProperty(value = "浮点数字", index = 2)private BigDecimal decCol;@ExcelProperty(value = "日期时间", index = 3)private LocalDateTime datetimeCol;
}

自定义合并

public void demo5(HttpServletResponse response) {List<ExcelVO> excelVOS = new ArrayList<>();excelVOS.add(new ExcelVO(1L,"字符串1",BigDecimal.ONE, LocalDateTime.now()));excelVOS.add(new ExcelVO(1L,"字符串2",BigDecimal.ONE, LocalDateTime.now()));excelVOS.add(new ExcelVO(2L,"字符串3",BigDecimal.ONE, LocalDateTime.now()));excelVOS.add(new ExcelVO(2L,"字符串4",BigDecimal.ONE, LocalDateTime.now()));// 相加BigDecimal sum = excelVOS.stream().map(ExcelVO::getDecCol).reduce(BigDecimal.ZERO, BigDecimal::add);// 将合计追加到集合中excelVOS.add(new ExcelVO(null,"合计", sum , null));response.setContentType("application/vnd.openxmlformats-officedocument.spreadsheetml.sheet");response.setCharacterEncoding("utf-8");String fileName = null;try {fileName = URLEncoder.encode("测试", "UTF-8").replaceAll("\\+", "%20");response.setHeader("Content-disposition", "attachment;filename*=utf-8''" + fileName + ".xlsx");// 设置合并规则LoopMergeStrategy loopMergeStrategy = new LoopMergeStrategy(2, 0);EasyExcel.write(response.getOutputStream(), ExcelVO.class).sheet("模板").registerWriteHandler(loopMergeStrategy).doWrite(excelVOS);} catch (IOException e) {throw new RuntimeException(e);}}

百万级别数据读写

多线程读取

• 思路：

  • 分析瓶颈在哪里？
  • 数据读取的性能：读取可以使用多线程【重点】
  • 数据库中插入数据，这个性能是比较不错的，可以使用批量插入【使用数据库连接池】

• 多线程读取数据的问题：

  • 需要避免重复读取，每个线程需要设置读取哪个区间的数据

public void importExcel(MultipartFile file) {try {// 多线程读取EasyExcel.read(file.getInputStream(),ExcelVO.class,new MutliReadListener(excelMapper)).sheet().doRead();} catch (IOException e) {throw new RuntimeException(e);}}

多线程读取监听器

@Service
@Slf4j
@NoArgsConstructor
public class MutliReadListener extends AnalysisEventListener<ExcelVO> {private ExcelMapper excelMapper;public MutliReadListener(ExcelMapper excelMapper) {this.excelMapper = excelMapper;}/*** 使用线程安全集合*/private List<ExcelVO> dataList = Collections.synchronizedList(new ArrayList<>());/*** 创建线程池必要参数*/private static final int CORE_POOL_SIZE = 5;//核心线程数private static final int MAX_POOL_SIZE = 10;//最大线程数private static final int QUEUE_CAPACITY = 100;//队列大小private static final Long KEEP_ALIVE_TIME = 1L;//存活时间@Overridepublic void invoke(ExcelVO data, AnalysisContext context) {if (dataList != null) {dataList.add(data);}}@Overridepublic void doAfterAllAnalysed(AnalysisContext context) {//创建线程池ExecutorService executor = new ThreadPoolExecutor(CORE_POOL_SIZE,MAX_POOL_SIZE,KEEP_ALIVE_TIME,TimeUnit.SECONDS,new ArrayBlockingQueue<>(QUEUE_CAPACITY),new ThreadPoolExecutor.CallerRunsPolicy());//指定每个线程需要处理的导入数量，假设每个线程处理1000条int singleThreadDealCount = 100000;//根据假设每个线程需要处理的数量以及总数，计算需要提交到线程池的线程数量int threadsize = (dataList.size() / singleThreadDealCount) + 1;log.info("threadsize===>{}",threadsize);//计算需要导入的数据总数，用于拆分时线程需要处理数据时使用int rowsize = dataList.size() + 1;//测试开始时间long startTime = System.currentTimeMillis();//申明该线程需要处理数据的开始位置int startPosition = 0;//申明该线程需要处理数据的结束位置int endPosition = 0;//为了让每个线程执行完后回到当前线程，使用CountDownLatch,值为线程数，每次线程执行完就会执行countDown方法减1，为0后回到主CountDownLatch count = new CountDownLatch(threadsize);//计算每个线程要处理的数据for (int i = 0; i < threadsize; i++) {//如果是最后一个线程，为保证程序不发生空指针异常，特殊判断结束位置if ((i + 1) == threadsize) {//计算开始位置startPosition = (i * singleThreadDealCount);//当前线程为划分的最后一个线程，则取总数据的最后为此线程的结束位置endPosition = rowsize - 1;} else {//计算开始位置startPosition = (i * singleThreadDealCount);//计算结束位置endPosition = (i + 1) * singleThreadDealCount;}log.info("线程开启====》count:{},startPosition:{},endPosition:{}",count,startPosition,endPosition);DeadMainThread deadMainThread = new DeadMainThread(count, excelMapper, dataList, startPosition, endPosition);executor.execute(deadMainThread);}try {count.await();} catch (InterruptedException e) {e.printStackTrace();}//逻辑处理完，关闭线程池executor.shutdown();long endTime = System.currentTimeMillis();System.out.println("总耗时：" + (endTime - startTime));}
}

线程任务

@Component
@Slf4j
public class DeadMainThread implements Runnable {/*** 当前线程需要处理的总数据中的开始位置*/private int startPosition;/*** 当前线程需要处理的，总数据中的结束位置*/private int endPosition;/*** 需要处理的拆分之前的全部数据*/private List<ExcelVO> list = Collections.synchronizedList(new ArrayList<>());private CountDownLatch count;private ExcelMapper excelMapper;public DeadMainThread() {}public DeadMainThread(CountDownLatch count, ExcelMapper excelMapper, List<ExcelVO> list, int startPosition, int endPosition) {this.startPosition = startPosition;this.endPosition = endPosition;this.excelMapper = excelMapper;this.list = list;this.count = count;}@Overridepublic void run() {try {List<ExcelVO> newList = list.subList(startPosition, endPosition);//批量新增excelMapper.insertBatch(BeanUtil.copyToList(newList, TbExcel.class));} catch (Exception e) {e.printStackTrace();} finally {//当一个线程执行完了计数要减一不然这个线程会被一直挂起count.countDown();log.info("减一===》{}",count.getCount());}}
}

多线程写出

造数据

public void init() {LocalDateTime now = LocalDateTime.now();// 通过计数器，信号量CountDownLatch countDownLatch = new CountDownLatch(100);for (int i = 0; i < 100; i++) {// 开启虚拟线程Thread.ofVirtual().start(() -> {List<TbExcel> list = new ArrayList<>();for (int j = 0; j < 10000; j++) {list.add(new TbExcel("字符串" + j, new BigDecimal(j), now));}excelMapper.insertBatch(list);// 计数器减一countDownLatch.countDown();log.info("线程{}执行完毕", Thread.currentThread().getName());});}// 是否结束try {countDownLatch.await();log.info("初始化完成");} catch (InterruptedException e) {throw new RuntimeException(e);}
}

实现分析

• 百万级别的数据考虑两个核心问题

  • 内存不能爆掉，不能产生OOM
  • 速度，性能

• 实现逻辑就是多线程，导出数据瓶颈在：

  • 数据查询，一次性查询多少数据
  • 内存不能溢出

至于数据写到excel中，这个就是easyexcel的表现了。

public void exportData(HttpServletResponse response) {long start1 = System.currentTimeMillis();// 每次查询10万条，考虑性能，内存Integer pageSize = 100000;// 线程池大小跟cpu有关，一般是cpu数量 * 2 + 1Integer poolSize = 10;// 随机文件名String fileName = String.valueOf(UUID.randomUUID());// 查询数据总数Long totalCount = excelMapper.selectCount(null);if (totalCount == 0) {log.info("没有数据需要导出");return; // 如果没有数据，直接返回}int loopCount = (int) Math.ceil((double) totalCount / pageSize);  // 使用 Math.ceil 计算循环次数// 设置final CountDownLatch latch = new CountDownLatch(loopCount);log.info("要查询的次数===>{}", loopCount);ExecutorService executorService = Executors.newFixedThreadPool(poolSize);OutputStream outputStream = null;try {outputStream = response.getOutputStream();// 创建写对象ExcelWriter excelWriter = EasyExcel.write(outputStream).build();for (int i = 0; i < loopCount; i++) {final int pageNum = i + 1; // 改为从1开始，直接使用 i + 1 作为页码executorService.execute(() -> {long start = System.currentTimeMillis();// 查询数据IPage<TbExcel> data = excelMapper.selectPage(new Page<>(pageNum, pageSize), null);List<TbExcel> records = data.getRecords();log.info("第{}页,查询耗时===>{}", pageNum,System.currentTimeMillis() - start);WriteSheet writeSheet = EasyExcel.writerSheet(pageNum ,"第" + pageNum + "页").head(ExcelVO.class).registerWriteHandler(new LongestMatchColumnWidthStyleStrategy()).build();synchronized (excelWriter) {long start2 = System.currentTimeMillis();excelWriter.write(BeanUtil.copyToList(records, ExcelVO.class), writeSheet);log.info("数据写出耗时===》{}",System.currentTimeMillis() - start2);}latch.countDown();});}latch.await();response.setContentType("application/octet-stream");response.setCharacterEncoding("utf-8");response.setHeader("Content-disposition", "attachment;filename*=utf-8''" + fileName + ".xlsx");excelWriter.finish();outputStream.flush();executorService.shutdown(); // 关闭线程池outputStream.close();log.info("总耗时====》{}",System.currentTimeMillis() - start1);} catch (IOException | InterruptedException e) {throw new RuntimeException(e);}}/*** 虚拟线程* @param response*/
@Override
public void exportData2(HttpServletResponse response) {long start1 = System.currentTimeMillis();Integer pageSize = 100000;Integer poolSize = 10;String fileName = String.valueOf(UUID.randomUUID());// 查询数据总数Long totalCount = excelMapper.selectCount(null);if (totalCount == 0) {log.info("没有数据需要导出");return; // 如果没有数据，直接返回}int loopCount = (int) Math.ceil((double) totalCount / pageSize);  // 使用 Math.ceil 计算循环次数// 设置final CountDownLatch latch = new CountDownLatch(loopCount);log.info("要查询的次数===>{}", loopCount);ExecutorService executorService = Executors.newFixedThreadPool(poolSize);OutputStream outputStream = null;try {outputStream = response.getOutputStream();// 创建写对象ExcelWriter excelWriter = EasyExcel.write(outputStream).build();for (int i = 0; i < loopCount; i++) {final int pageNum = i + 1; // 改为从1开始，直接使用 i + 1 作为页码Thread.ofVirtual().start(() -> {long start = System.currentTimeMillis();// 查询数据IPage<TbExcel> data = excelMapper.selectPage(new Page<>(pageNum, pageSize), null);List<TbExcel> records = data.getRecords();log.info("第{}页,查询耗时===>{}", pageNum,System.currentTimeMillis() - start);WriteSheet writeSheet = EasyExcel.writerSheet(pageNum ,"第" + pageNum + "页").head(ExcelVO.class).registerWriteHandler(new LongestMatchColumnWidthStyleStrategy()).build();synchronized (excelWriter) {long start2 = System.currentTimeMillis();excelWriter.write(BeanUtil.copyToList(records, ExcelVO.class), writeSheet);log.info("数据写出耗时===》{}",System.currentTimeMillis() - start2);}latch.countDown();});}latch.await();response.setContentType("application/octet-stream");response.setCharacterEncoding("utf-8");response.setHeader("Content-disposition", "attachment;filename*=utf-8''" + fileName + ".xlsx");excelWriter.finish();outputStream.flush();executorService.shutdown(); // 关闭线程池outputStream.close();log.info("总耗时====》{}",System.currentTimeMillis() - start1);} catch (IOException | InterruptedException e) {throw new RuntimeException(e);}}/*** 单sheet页超过100万条数据就写不进去了* @param response*/
@Override
public void exportData3(HttpServletResponse response) {long start1 = System.currentTimeMillis();Integer pageSize = 100000;Integer poolSize = 10;String fileName = String.valueOf(UUID.randomUUID());// 查询数据总数Long totalCount = excelMapper.selectCount(null);if (totalCount == 0) {log.info("没有数据需要导出");return; // 如果没有数据，直接返回}int loopCount = (int) Math.ceil((double) totalCount / pageSize);  // 使用 Math.ceil 计算循环次数// 设置final CountDownLatch latch = new CountDownLatch(loopCount);log.info("要查询的次数===>{}", loopCount);ExecutorService executorService = Executors.newFixedThreadPool(poolSize);OutputStream outputStream = null;try {outputStream = response.getOutputStream();// 创建写对象ExcelWriter excelWriter = EasyExcel.write(outputStream).build();WriteSheet writeSheet = EasyExcel.writerSheet(1 ,"第" + 1 + "页").head(ExcelVO.class).registerWriteHandler(new LongestMatchColumnWidthStyleStrategy()).registerConverter(new LongStringConverter()).build();for (int i = 0; i < loopCount; i++) {final int pageNum = i + 1; // 改为从1开始，直接使用 i + 1 作为页码Thread.ofVirtual().start(() -> {long start = System.currentTimeMillis();// 查询数据IPage<TbExcel> data = excelMapper.selectPage(new Page<>(pageNum, pageSize), null);List<TbExcel> records = data.getRecords();log.info("第{}页,查询耗时===>{}", pageNum,System.currentTimeMillis() - start);synchronized (excelWriter) {long start2 = System.currentTimeMillis();excelWriter.write(BeanUtil.copyToList(records, ExcelVO.class), writeSheet);log.info("数据写出耗时===》{}",System.currentTimeMillis() - start2);}latch.countDown();});}latch.await();response.setContentType("application/octet-stream");response.setCharacterEncoding("utf-8");response.setHeader("Content-disposition", "attachment;filename*=utf-8''" + fileName + ".xlsx");excelWriter.finish();outputStream.flush();executorService.shutdown(); // 关闭线程池outputStream.close();log.info("总耗时====》{}",System.currentTimeMillis() - start1);} catch (IOException | InterruptedException e) {throw new RuntimeException(e);}}

---