douwen3362 2019-03-08 05:50
浏览 31
已采纳

重写中间件中的内容长度

below code rewrites the http body response of some queries.

However, it fails to update the "content length" header field, it always remains the same original value.

How can i update the content length header field of the http response ?

type writeReplacer struct {
    http.ResponseWriter
    search  []byte
    replace func(*http.Request) string
    buf     []byte
    r       *http.Request
    dir     string
}

func (w *writeReplacer) Write(in []byte) (int, error) {
    if w.buf == nil {
        w.buf = []byte{}
    }
    w.buf = append(w.buf, in...)
    n := len(in)
    if index := bytes.LastIndex(w.buf, w.search); index > -1 {
        var r []byte
        if w.dir == "before" {
            g := []byte(w.replace(w.r))
            n += len(g)
            r = append(g, w.buf[index:]...)
            w.buf = append(w.buf[:index], r...)
        } else {
            g := []byte(w.replace(w.r))
            n += len(g)
            r = append(r, w.buf[:index+len(w.search)]...)
            r = append(r, g...)
            r = append(r, w.buf[index:]...)
            w.buf = r
        }
    }
    return n, nil
}

func (w *writeReplacer) Flush() {
    w.ResponseWriter.Header().Set("Content-Length", fmt.Sprint(len(w.buf)))
    w.ResponseWriter.Write(w.buf[:])
    w.buf = w.buf[:0]
}

func InsertAfter(h http.Handler, path string, search []byte, replace func(*http.Request) string) http.Handler {
    return http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
        if r.URL.Path == path {
            w.Header().Del("Content-length")
            w = &writeReplacer{ResponseWriter: w, search: search, replace: replace, r: r, dir: "after"}
            defer w.(http.Flusher).Flush()
        }
        h.ServeHTTP(w, r)
    })
}

func InsertBefore(h http.Handler, path string, search []byte, replace func(*http.Request) string) http.Handler {
    return http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
        if r.URL.Path == path {
            w.Header().Del("Content-length")
            w = &writeReplacer{ResponseWriter: w, search: search, replace: replace, r: r, dir: "before"}
            defer w.(http.Flusher).Flush()
        }
        h.ServeHTTP(w, r)
    })
}

I receive two errors message, the first one from nginx

2019/03/08 05:58:37 [error] 31194#0: *19 
upstream prematurely closed connection while reading upstream, 
client: 82.21.18.16, server: buycoffee.online, request: 
"GET / HTTP/1.1", upstream: "http://127.0.0.1:8081/", host: "buycoffee.online"

the second one from curl

curl: (18) transfer closed with 6237 bytes remaining to read
  • 写回答

1条回答 默认 最新

  • doudun5009 2019-03-08 07:45
    关注

    as mkopriva suggested the trick was to rewrite ResponseWriter.WriteHeader.

    I believe it has to do with the fact that write might call writeheader at first call, and from there i was not putting the instructions at the right place.

    in order to prevent further difficulties i prefer to use chunked transfer.

    the code change is:

    func (w *writeReplacer) WriteHeader(statusCode int) {
        w.Header().Del("Content-length")
        w.Header().Set("Transfer-Encoding", "chunked")
        w.ResponseWriter.WriteHeader(statusCode)
    }
    
    本回答被题主选为最佳回答 , 对您是否有帮助呢?
    评论

报告相同问题?

悬赏问题

  • ¥15 树莓派与pix飞控通信
  • ¥15 自动转发微信群信息到另外一个微信群
  • ¥15 outlook无法配置成功
  • ¥30 这是哪个作者做的宝宝起名网站
  • ¥60 版本过低apk如何修改可以兼容新的安卓系统
  • ¥25 由IPR导致的DRIVER_POWER_STATE_FAILURE蓝屏
  • ¥50 有数据,怎么建立模型求影响全要素生产率的因素
  • ¥50 有数据,怎么用matlab求全要素生产率
  • ¥15 TI的insta-spin例程
  • ¥15 完成下列问题完成下列问题