java EasyExcel读写逻辑
程序猿杜小头 人气:0正文
EasyExcel是一款由阿里开源的 Excel 处理工具。相较于原生的Apache POI，它可以更优雅、快速地完成 Excel 的读写功能，同时更加地节约内存。
即使 EasyExcel 已经很优雅了，但面向 Excel 文档的读写逻辑几乎千篇一律，笔者索性将这些模板化的逻辑抽离出来，该组件已经发布到 maven 中央仓库，感兴趣的朋友可以体验一下。
1 快速上手
1.1 引入依赖
<dependency> <groupId>io.github.dk900912</groupId> <artifactId>easyexcel-spring-boot-starter</artifactId> <version>0.0.7</version> </dependency>
1.2 导入与导出
@RestController
@RequestMapping(path = "/easyexcel")
public class ExcelController {
@PostMapping(path = "/v1/upload")
public ResponseEntity<String> upload(
@RequestExcel(sheets = {
@Sheet(index = 0, headClazz = User.class, headRowNumber = 1),
@Sheet(index = 1, headClazz = User.class, headRowNumber = 1),
@Sheet(index = 2, headClazz = User.class, headRowNumber = 1)
})
@Valid List<List<User>> users) {
return ResponseEntity.ok("OK");
}
@ResponseExcel(
name="程序猿",
sheets = {
@Sheet(name = "sheet-0", headClazz = User.class),
@Sheet(name = "sheet-1", headClazz = User.class),
@Sheet(name = "sheet-2", headClazz = User.class)
},
suffix = ExcelTypeEnum.XLSX)
@GetMapping(path = "/v1/export")
public List<List<User>> export() {
List<User> data = Lists.newArrayList();
for (int i = 0; i < 10000; i++) {
User user = User.builder().name("暴风赤红" + (i+1))
.birth(LocalDate.now()).address("江苏省苏州市科技城昆仑山路58号")
.build();
data.add(user);
}
return ImmutableList.of(data, data, data);
}
@ResponseExcel(name="templates/程序猿.xlsx", scene = TEMPLATE)
@GetMapping(path = "/v1/template")
public void template() {}
}
2 实现原理
一切 Java 程序都是基于 Thread 的，当一个 HTTP 请求到达后，Servlet Container 会从其线程池中捞出一个线程来处理该 HTTP 请求。具体地，该 HTTP 请求首先到达 Servlet Container 的FilterChain中；然后，FilterChain 将该 HTTP 请求委派给DispatcherServlet处理，而 DispatcherServlet 恰恰就是 Spring MVC 的门户。在 Spring MVC 中，所有 HTTP 请求都由 DispatcherServlet 进行路由分发。大致流程下图所示。
DispatcherServlet 在 HandlerMapping 的帮助下可以快速匹配到最终的 Controller，由于 Controller 大多由@RequestMapping注解标注，那么RequestMappingHandlerMapping最终脱颖而出。
RequestMappingHandlerMapping 会将 HTTP 请求映射到一个HandlerExecutionChain实例中，每一个 HandlerExecutionChain 实例的内部维护了HandlerMethod和List<HandlerInterceptor>。
其中，HandlerMethod 实例持有一个Object类型的 bean 变量和java.lang.reflect.Method类型的 method 变量，bean 和 method 这俩成员变量组合起来最终可以确定究竟由哪一个 Controller 中的某一方法来处理当前 HTTP 请求。
此时已经知道目标方法了，那直接反射执行目标方法？是不可以的，因为通过反射来执行目标方法需要有参数才行，此外还需要对目标方法的执行结果进行加工处理。既然 HandlerMapping 没有解析请求体和处理目标执行结果的能力，只能再引入一层适配器了，它就是 HandlerAdapter。
在 Spring MVC 所提供的若干种 HandlerAdapter 中，能够适配 HandlerMethod 的只有RequestMappingHandlerAdapter；
RequestMappingHandlerAdapter 实现了InitializingBean接口，用于初始化HandlerMethodArgumentResolverComposite类型的 argumentResolvers 成员变量和HandlerMethodReturnValueHandlerComposite类型的 returnValueHandlers 成员变量，Composite 后缀表明这俩成员变量均是一种复合类，argumentResolvers 持有数十种HandlerMethodArgumentResolver类型的方法参数解析器，而 returnValueHandlers 则持有数十种HandlerMethodReturnValueHandler类型的方法返回值解析器。
重点来了！首先，我们需要一个实现 HandlerMethodArgumentResolver 接口的方法参数解析器，该解析器主要用于解析@RequestExcel注解，以读取 Excel 文档；
此外，我们还需要一个实现 HandlerMethodReturnValueHandler 接口的方法返回值解析器，该解析器主要用于解析@ResponseExcel注解，以将目标方法所返回的数据写入到 Excel 文档中；
最后，将这两个自定义的解析器分别添加到 RequestMappingHandlerAdapter 中的 argumentResolvers 与 returnValueHandlers 这俩成员变量中。
在 Spring MVC 中，由RequestResponseBodyMethodProcessor负责处理@RequestBody与@ResponseBody 注解。基于这一事实，笔者也没有单独设计两个解析器来分别应对 @RequestExcel 与 @ResponseExcel 注解，而是合二为一。
2.1 @RequestExcel 与 @ResponseExcel 解析器
public class RequestResponseExcelMethodProcessor implements HandlerMethodArgumentResolver,
HandlerMethodReturnValueHandler {
private final ResourceLoader resourceLoader;
public RequestResponseExcelMethodProcessor(ResourceLoader resourceLoader) {
this.resourceLoader = resourceLoader;
}
@Override
public boolean supportsParameter(MethodParameter parameter) {
return parameter.hasParameterAnnotation(RequestExcel.class);
}
@Override
public boolean supportsReturnType(MethodParameter returnType) {
return returnType.hasMethodAnnotation(ResponseExcel.class);
}
@Override
public Object resolveArgument(MethodParameter parameter, ModelAndViewContainer mavContainer,
NativeWebRequest webRequest, WebDataBinderFactory binderFactory) throws Exception {
parameter = parameter.nestedIfOptional();
Object data = readWithMessageConverters(webRequest, parameter);
validateIfNecessary(data, parameter);
return data;
}
@Override
public void handleReturnValue(Object returnValue, MethodParameter returnType,
ModelAndViewContainer mavContainer, NativeWebRequest webRequest) throws Exception {
// There is no need to render view
mavContainer.setRequestHandled(true);
writeWithMessageConverters(returnValue, returnType, webRequest);
}
// +----------------------------------------------------------------------------+
// | private method for read |
// +----------------------------------------------------------------------------+
protected <T> Object readWithMessageConverters(NativeWebRequest webRequest, MethodParameter parameter)
throws IOException, UnsatisfiedMethodSignatureException {
validateArgParamOrReturnValueType(parameter);
HttpServletRequest servletRequest = webRequest.getNativeRequest(HttpServletRequest.class);
Assert.state(servletRequest != null, "No HttpServletRequest");
return readWithMessageConverters(servletRequest, parameter);
}
protected Object readWithMessageConverters(HttpServletRequest servletRequest, MethodParameter parameter)
throws IOException {
RequestExcelInfo requestExcelInfo =
new RequestExcelInfo(parameter.getParameterAnnotation(RequestExcel.class));
InputStream inputStream;
if (servletRequest instanceof MultipartRequest) {
inputStream = ((MultipartRequest) servletRequest)
.getMultiFileMap()
.values()
.stream()
.flatMap(Collection::stream)
.findFirst()
.map(multipartFile -> {
try {
return multipartFile.getInputStream();
} catch (IOException e) {
return null;
}
})
.get();
} else {
inputStream = servletRequest.getInputStream();
}
CollectorReadListener collectorReadListener = new CollectorReadListener();
try (ExcelReader excelReader = EasyExcel.read(inputStream).build()) {
List<ReadSheet> readSheetList = requestExcelInfo.getSheetInfoList()
.stream()
.map(sheetInfo -> EasyExcel.readSheet(sheetInfo.getIndex())
.head(sheetInfo.getHeadClazz())
.registerReadListener(collectorReadListener)
.build()
)
.collect(Collectors.toList());
excelReader.read(readSheetList);
}
return collectorReadListener.groupByHeadClazz();
}
protected void validateIfNecessary(Object data, MethodParameter parameter) throws ExcelCellContentNotValidException {
if (parameter.hasParameterAnnotation(Validated.class)
|| parameter.hasParameterAnnotation(Valid.class)) {
List<Object> flattenData = ((List<List<Object>>) data).stream()
.flatMap(Collection::stream)
.collect(Collectors.toList());
for (Object target : flattenData) {
Set<ConstraintViolation<Object>> constraintViolationSet = ValidationUtil.validate(target);
if (CollectionUtils.isNotEmpty(constraintViolationSet)) {
String errorMsg = constraintViolationSet.stream()
.map(ConstraintViolation::getMessage)
.distinct()
.findFirst()
.get();
throw new ExcelCellContentNotValidException(errorMsg);
}
}
}
}
// +----------------------------------------------------------------------------+
// | private method for write |
// +----------------------------------------------------------------------------+
protected void writeWithMessageConverters(Object value, MethodParameter returnType, NativeWebRequest webRequest)
throws IOException, HttpMessageNotWritableException, UnsatisfiedMethodSignatureException {
HttpServletResponse response = webRequest.getNativeResponse(HttpServletResponse.class);
Assert.state(response != null, "No HttpServletResponse");
ResponseExcelInfo responseExcelInfo =
new ResponseExcelInfo(returnType.getMethodAnnotation(ResponseExcel.class));
final String fileName = responseExcelInfo.getName();
response.setContentType("application/vnd.openxmlformats-officedocument.spreadsheetml.sheet");
response.setCharacterEncoding(StandardCharsets.UTF_8.name());
if (TEMPLATE.equals(responseExcelInfo.getScene())) {
response.setHeader("Content-disposition",
"attachment;filename=" + URLEncoder.encode(
fileName.substring(fileName.indexOf("/") + 1), StandardCharsets.UTF_8.name()));
BufferedInputStream bufferedInputStream =
new BufferedInputStream(resourceLoader.getResource(CLASSPATH_URL_PREFIX + fileName).getInputStream());
BufferedOutputStream bufferedOutputStream = new BufferedOutputStream(response.getOutputStream());
FileCopyUtils.copy(bufferedInputStream, bufferedOutputStream);
} else {
validateArgParamOrReturnValueType(returnType);
response.setHeader("Content-disposition",
"attachment;filename=" + URLEncoder.encode(
fileName, StandardCharsets.UTF_8.name()) + responseExcelInfo.getSuffix().getValue());
try (ExcelWriter excelWriter = EasyExcel.write(response.getOutputStream()).build()) {
List<WriteSheet> writeSheetList = responseExcelInfo.getSheetInfoList()
.stream()
.map(sheetInfo -> EasyExcel.writerSheet(sheetInfo.getName())
.head(sheetInfo.getHeadClazz())
.build()
)
.collect(Collectors.toList());
List<List<Object>> multiSheetData = (List<List<Object>>) value;
for (int i = 0; i < writeSheetList.size(); i++) {
WriteSheet writeSheet = writeSheetList.get(i);
List<Object> singleSheetData = multiSheetData.get(i);
excelWriter.write(singleSheetData, writeSheet);
}
}
}
}
// +----------------------------------------------------------------------------+
// | common private method |
// +----------------------------------------------------------------------------+
private void validateArgParamOrReturnValueType(MethodParameter target) throws UnsatisfiedMethodSignatureException {
try {
ResolvableType resolvableType = ResolvableType.forMethodParameter(target);
if (!List.class.isAssignableFrom(resolvableType.resolve())) {
throw new UnsatisfiedMethodSignatureException(
"@RequestExcel or @ResponseExcel Must Be Annotated With List<List<>>");
}
if (!List.class.isAssignableFrom(resolvableType.getGeneric(0).resolve())) {
throw new UnsatisfiedMethodSignatureException(
"@RequestExcel or @ResponseExcel Must Be Annotated With List<List<>>");
}
} catch (Exception exception) {
throw new UnsatisfiedMethodSignatureException(
"@RequestExcel or @ResponseExcel Must Be Annotated With List<List<>>");
}
}
}
2.2 RequestMappingHandlerAdapter 后置处理器
public class RequestMappingHandlerAdapterPostProcessor implements BeanPostProcessor,
PriorityOrdered, ResourceLoaderAware {
private ResourceLoader resourceLoader;
@Override
public Object postProcessAfterInitialization(Object bean, String beanName) throws BeansException {
if (!supports(bean)) {
return bean;
}
RequestMappingHandlerAdapter requestMappingHandlerAdapter = (RequestMappingHandlerAdapter) bean;
List<HandlerMethodArgumentResolver> argumentResolvers = requestMappingHandlerAdapter.getArgumentResolvers();
List<HandlerMethodReturnValueHandler> returnValueHandlers = requestMappingHandlerAdapter.getReturnValueHandlers();
Assert.notEmpty(argumentResolvers,
"RequestMappingHandlerAdapter's argument resolver is empty, this is illegal state");
Assert.notEmpty(returnValueHandlers,
"RequestMappingHandlerAdapter's return-value handler is empty, this is illegal state");
List<HandlerMethodArgumentResolver> copyArgumentResolvers = new ArrayList<>(argumentResolvers);
RequestResponseExcelMethodProcessor argumentResolver4RequestExcel = new RequestResponseExcelMethodProcessor(null);
copyArgumentResolvers.add(0, argumentResolver4RequestExcel);
requestMappingHandlerAdapter.setArgumentResolvers(Collections.unmodifiableList(copyArgumentResolvers));
List<HandlerMethodReturnValueHandler> copyReturnValueHandlers = new ArrayList<>(returnValueHandlers);
RequestResponseExcelMethodProcessor returnValueHandler4ResponseExcel = new RequestResponseExcelMethodProcessor(resourceLoader);
copyReturnValueHandlers.add(0, returnValueHandler4ResponseExcel);
requestMappingHandlerAdapter.setReturnValueHandlers(Collections.unmodifiableList(copyReturnValueHandlers));
return requestMappingHandlerAdapter;
}
@Override
public int getOrder() {
return Ordered.LOWEST_PRECEDENCE;
}
@Override
public void setResourceLoader(ResourceLoader resourceLoader) {
this.resourceLoader = resourceLoader;
}
private boolean supports(Object bean) {
return bean instanceof RequestMappingHandlerAdapter;
}
}
3 总结
目前该版本仅支持针对单个 Excel 文档的导入与导出（多Sheet是支持的哈），所以由 @RequestExcel 注解修饰的方法参数必须是一个List<List<>>类型，而由 @RequestExcel 注解修饰的方法返回类型也必须是一个List<List<>>类型，否则将抛出UnsatisfiedMethodSignatureException类型的自定义异常。
坦白来说，该组件的设计初衷只是为了帮助大家从公式化、模板化的Excel 读写逻辑中解放出来，从而专注于核心业务逻辑的开发，并不是为了增强 EasyExcel，后续也不会朝着这一方向演进。
加载全部内容