ASP.NET Core透传API如何直接返回HttpResponseMessage内容

我完全懂你的需求——就是要做一个"透明代理"，把外部服务返回的状态码、所有响应头、响应体原封不动地传给调用方，之前的尝试要么把响应对象序列化返回成JSON了，要么因为流处理不当导致响应体为空对吧？

下面给你两种简洁可靠的实现方式，先看最直接的HttpContext操作版本，这也是代理场景下最常用的：

直接操作HttpContext透传响应 #

先修正你的代码，同时解决几个常见坑（比如HttpClient的正确使用、流处理、头的分类复制）：

[Route("proxy/{input1}")]
[Authorize(Policy = "Admin")]
public async Task ProxyCall(string input1)
{
    // 重点：别每次new HttpClient！推荐注入IHttpClientFactory来获取实例，避免socket泄漏
    var client = _httpClientFactory.CreateClient(); 
    var request = _httpContextAccessor.HttpContext?.Request;
    
    if (request == null)
    {
        _httpContextAccessor.HttpContext.Response.StatusCode = StatusCodes.Status400BadRequest;
        return;
    }

    // 拼接目标URI
    var uriToCall = QueryHelpers.AddQueryString(
        LookupUri(input1), 
        request.Query.ToDictionary(kvp => kvp.Key, kvp => kvp.Value.ToString())
    );

    // 构建转发请求：复制原请求的所有内容头
    var httpContent = new StreamContent(request.Body);
    foreach (var header in request.ContentHeaders)
    {
        httpContent.Headers.TryAddWithoutValidation(header.Key, header.Value);
    }

    var executionRequest = new HttpRequestMessage(
        new HttpMethod(GetMethodForHost(uriToCall)), 
        uriToCall
    ) { Content = httpContent };

    // 复制原请求的非内容头（跳过Host这类HttpClient自动处理的头）
    foreach (var header in request.Headers)
    {
        if (!executionRequest.Headers.TryAddWithoutValidation(header.Key, header.Value) && httpContent != null)
        {
            httpContent.Headers.TryAddWithoutValidation(header.Key, header.Value);
        }
    }

    // 用ResponseHeadersRead提高性能：拿到响应头就开始处理，不用等整个响应体下载完
    using var response = await client.SendAsync(executionRequest, HttpCompletionOption.ResponseHeadersRead);
    var currentResponse = _httpContextAccessor.HttpContext.Response;

    // 1. 设置状态码
    currentResponse.StatusCode = (int)response.StatusCode;

    // 2. 复制响应头：区分普通响应头和内容头
    foreach (var header in response.Headers)
    {
        currentResponse.Headers.TryAdd(header.Key, new StringValues(header.Value.ToArray()));
    }
    foreach (var header in response.Content.Headers)
    {
        currentResponse.Headers.TryAdd(header.Key, new StringValues(header.Value.ToArray()));
    }

    // 3. 复制响应体：用CopyToAsync确保流正确传输，避免手动赋值Body的问题
    await using var responseStream = await response.Content.ReadAsStreamAsync();
    await responseStream.CopyToAsync(currentResponse.Body);
    await currentResponse.Body.FlushAsync();
}

用IActionResult方式返回（更符合ASP.NET Core规范） #

如果你的Action习惯返回IActionResult，可以用FileStreamResult来包装响应：

[Route("proxy/{input1}")]
[Authorize(Policy = "Admin")]
public async Task<IActionResult> ProxyCall(string input1)
{
    var client = _httpClientFactory.CreateClient();
    var request = _httpContextAccessor.HttpContext?.Request;
    
    if (request == null)
    {
        return BadRequest();
    }

    // 前面的请求构建逻辑和上面一致...
    var uriToCall = QueryHelpers.AddQueryString(LookupUri(input1), request.Query.ToDictionary(kvp => kvp.Key, kvp => kvp.Value.ToString()));
    var httpContent = new StreamContent(request.Body);
    foreach (var header in request.ContentHeaders)
    {
        httpContent.Headers.TryAddWithoutValidation(header.Key, header.Value);
    }
    var executionRequest = new HttpRequestMessage(new HttpMethod(GetMethodForHost(uriToCall)), uriToCall) { Content = httpContent };
    foreach (var header in request.Headers)
    {
        if (!executionRequest.Headers.TryAddWithoutValidation(header.Key, header.Value) && httpContent != null)
        {
            httpContent.Headers.TryAddWithoutValidation(header.Key, header.Value);
        }
    }

    using var response = await client.SendAsync(executionRequest, HttpCompletionOption.ResponseHeadersRead);
    
    // 用FileStreamResult包装响应流
    var responseStream = await response.Content.ReadAsStreamAsync();
    var result = new FileStreamResult(
        responseStream, 
        response.Content.Headers.ContentType?.ToString() ?? "application/octet-stream"
    )
    {
        StatusCode = (int)response.StatusCode
    };

    // 添加所有响应头
    foreach (var header in response.Headers)
    {
        result.ResponseHeaders.TryAdd(header.Key, new StringValues(header.Value.ToArray()));
    }
    foreach (var header in response.Content.Headers)
    {
        if (!result.ResponseHeaders.ContainsKey(header.Key))
        {
            result.ResponseHeaders.TryAdd(header.Key, new StringValues(header.Value.ToArray()));
        }
    }

    return result;
}

关键注意事项 #

1. HttpClient的正确使用：绝对不要每次请求都new HttpClient，会导致TCP socket资源泄漏，一定要注入IHttpClientFactory来获取实例。
2. 流处理的坑：之前直接赋值Response.Body = responseStream会为空，是因为流读取后位置在末尾，而且ASP.NET Core对Body流的处理有特定逻辑，用CopyToAsync是最可靠的方式。
3. 响应头的分类：外部响应的头分为Response.Headers（普通响应头如Cache-Control）和Content.Headers（内容相关头如Content-Type），两者都需要复制。
4. 性能优化：使用HttpCompletionOption.ResponseHeadersRead可以在拿到响应头后就开始传输内容，不用等整个响应体下载完成，适合大文件场景。

内容的提问来源于stack exchange，提问作者ElFik