Spring webflux代理多部分数据答案

【问题标题】：Spring webflux proxy multipart dataSpring webflux代理多部分数据
【发布时间】：2025-12-17 04:00:01
【问题描述】：

我有获取ServerRequest 然后将其转换为WebClient 并将其代理到目的地的组件。请求的Content-Type为multipart/form-data

return getBody(request)
    .flatMap { body ->
      WebClient.builder()
          .baseUrl(path)
          .defaultCookies { x ->
            request.cookies()
                .forEach { (k, v) -> x[k] = v.map { it.value } }
          }
          .defaultHeaders { x ->
            request.headers().asHttpHeaders()
                .forEach { (k, v) -> x[k] = v }
          }
          .defaultUriVariables(request.queryParams())
          .build()
          .method(request.method() ?: HttpMethod.GET)
          .body(body)
          .exchange()
    }

  private fun getBody(request: ServerRequest): Mono<BodyInserter<*, in ClientHttpRequest>> {
    return request.multipartData()
        .filter { it.isNotEmpty() }
        .map<BodyInserter<*, in ClientHttpRequest>> { multipart ->
          BodyInserters.fromMultipartData(multipart)
        }
        .switchIfEmpty(Mono.fromCallable {
          val body = request.bodyToMono(String::class.java)
          BodyInserters.fromPublisher(body, String::class.java)
        })
  }

我遇到了多部分请求的问题：尝试使用多部分创建正文以将其发送到目的地时失败。

我有以下配置：

@Configuration
open class SwitcherMultipartConfigurer : WebFluxConfigurer {

  override fun configureHttpMessageCodecs(configurer: ServerCodecConfigurer) {
    configurer.customCodecs().reader(
        MultipartHttpMessageReader(SynchronossPartHttpMessageReader())
    )
    configurer.customCodecs().writer(
        MultipartHttpMessageWriter(

        )
    )
  }
}

它可以是什么？错误是：

 observed an error org.springframework.core.codec.CodecException: No suitable writer found for part: file
      | 2019-06-25T13:33:49.069006743Z  at org.springframework.http.codec.multipart.MultipartHttpMessageWriter.encodePart(MultipartHttpMessageWriter.java:300)
      | 2019-06-25T13:33:49.069013261Z  at org.springframework.http.codec.multipart.MultipartHttpMessageWriter.lambda$encodePartValues$4(MultipartHttpMessageWriter.java:253)
      | 2019-06-25T13:33:49.069018616Z  at java.util.stream.ReferencePipeline$3$1.accept(ReferencePipeline.java:193)
      | 2019-06-25T13:33:49.069023613Z  at java.util.LinkedList$LLSpliterator.forEachRemaining(LinkedList.java:1235)
      | 2019-06-25T13:33:49.069028522Z  at java.util.stream.AbstractPipeline.copyInto(AbstractPipeline.java:481)
      | 2019-06-25T13:33:49.069033432Z  at java.util.stream.AbstractPipeline.wrapAndCopyInto(AbstractPipeline.java:471)
      | 2019-06-25T13:33:49.069038361Z  at java.util.stream.ReduceOps$ReduceOp.evaluateSequential(ReduceOps.java:708)
      | 2019-06-25T13:33:49.069043368Z  at java.util.stream.AbstractPipeline.evaluate(AbstractPipeline.java:234)
      | 2019-06-25T13:33:49.069048212Z  at java.util.stream.ReferencePipeline.collect(ReferencePipeline.java:499)
      | 2019-06-25T13:33:49.069053089Z  at org.springframework.http.codec.multipart.MultipartHttpMessageWriter.encodePartValues(MultipartHttpMessageWriter.java:253)
      | 2019-06-25T13:33:49.069058115Z  at org.springframework.http.codec.multipart.MultipartHttpMessageWriter.lambda$writeMultipart$3(MultipartHttpMessageWriter.java:234)

【问题讨论】：

我不太擅长阅读 Kotlin，但在阅读了一些文档后，我认为 request.multipartData() 不是正确的选择。文档状态：ServerWebExchange exposes the following method for access to multipart data: Mono<MultiValueMap<String, Part>> getMultipartData(); The DefaultServerWebExchange uses the configured HttpMessageReader<MultiValueMap<String, Part>> to parse multipart/form-data content into a MultiValueMap. 我看不到您如何访问请求对象，但也许可以尝试ServerWebExchange#getMultipartData()
@ThomasAndolf 我在这里通过调用request.multipartData() 访问请求，其中请求是ServerRequest 的实例。

标签： java kotlin multipartform-data spring-webflux

【解决方案1】：

找到了方法。我代理的多部分数据不是MultipartHttpReader 创建的Part，而是ByteArray。

  val builder = MultipartBodyBuilder()
  multipart
      .toSingleValueMap()
      .forEach {
        builder.asyncPart(it.key, it.value.content(), DataBuffer::class.java)
            .headers { httpHeaders -> it.value.headers().forEach { httpHeaders[it.key] = it.value } }
      }
  BodyInserters.fromMultipartData(builder.build())

需要标头来保存Content-Disposition标头中显示的内容类型和文件名。

【讨论】：