英文:
how to close/abort a Golang http.Client POST prematurely
问题
我正在使用http.Client
来实现客户端的长轮询:
resp, err := client.Post(url, "application/json", bytes.NewBuffer(jsonPostBytes))
if err != nil {
panic(err)
}
defer resp.Body.Close()
var results []*ResponseMessage
err = json.NewDecoder(resp.Body).Decode(&results) // 在长轮询时,代码在这里阻塞
有没有一种标准的方法可以在客户端取消请求?
我想调用resp.Body.Close()
可以实现取消请求,但是我必须从另一个goroutine中调用它,因为客户端通常已经在读取长轮询的响应时被阻塞。
我知道可以通过http.Transport
设置超时,但是我的应用逻辑需要根据用户操作来取消请求,而不仅仅是超时。
英文:
I'm using http.Client
for the client-side implementation of a long-poll:
resp, err := client.Post(url, "application/json", bytes.NewBuffer(jsonPostBytes))
if err != nil {
panic(err)
}
defer resp.Body.Close()
var results []*ResponseMessage
err = json.NewDecoder(resp.Body).Decode(&results) // code blocks here on long-poll
Is there a standard way to pre-empt/cancel the request from the client-side?
I imagine that calling resp.Body.Close()
would do it, but I'd have to call that from another goroutine, as the client is normally already blocked in reading the response of the long-poll.
I know that there is a way to set a timeout via http.Transport
, but my app logic need to do the cancellation based on a user action, not just a timeout.
答案1
得分: 27
使用CancelRequest现在已经被弃用。
当前的策略是使用http.Request.WithContext传递一个带有截止时间或将被取消的上下文。
之后就像正常的请求一样使用它。
req, err := http.NewRequest("GET", "http://example.com", nil)
// ...
req.Header.Add("If-None-Match", `W/"wyzzy"`)
req = req.WithContext(ctx)
resp, err := client.Do(req)
// ...
英文:
Using CancelRequest is now deprecated.
The current strategy is to use http.Request.WithContext passing a context with a deadline or that will be canceled otherwise.
Just use it like a normal request afterwards.
req, err := http.NewRequest("GET", "http://example.com", nil)
// ...
req.Header.Add("If-None-Match", `W/"wyzzy"`)
req = req.WithContext(ctx)
resp, err := client.Do(req)
// ...
答案2
得分: 19
标准的方法是使用类型为context.Context
的上下文,并将其传递给所有需要知道请求何时取消的函数。
func httpDo(ctx context.Context, req *http.Request, f func(*http.Response, error) error) error {
// 在一个 goroutine 中运行 HTTP 请求,并将响应传递给 f 函数。
tr := &http.Transport{}
client := &http.Client{Transport: tr}
c := make(chan error, 1)
go func() { c <- f(client.Do(req)) }()
select {
case <-ctx.Done():
tr.CancelRequest(req)
<-c // 等待 f 函数返回。
return ctx.Err()
case err := <-c:
return err
}
}
golang.org/x/net/context
// Context 携带截止时间、取消信号和请求范围的值,跨 API 边界传递。
// 它的方法可以被多个 goroutine 同时使用。
type Context interface {
// Done 返回一个通道,当此 Context 被取消或超时时关闭。
Done() <-chan struct{}
// Err 在 Done 通道关闭后,指示此上下文被取消的原因。
Err() error
// Deadline 返回此 Context 将被取消的时间(如果有)。
Deadline() (deadline time.Time, ok bool)
// Value 返回与键关联的值,如果没有则返回 nil。
Value(key interface{}) interface{}
}
更多信息请参考 https://blog.golang.org/context
更新
正如 Paulo 提到的,Request.Cancel 现在已被弃用,作者应该将上下文传递给请求本身(使用 *Request.WithContext),并使用上下文的取消通道(取消请求)。
package main
import (
"context"
"net/http"
"time"
)
func main() {
cx, cancel := context.WithCancel(context.Background())
req, _ := http.NewRequest("GET", "http://google.com", nil)
req = req.WithContext(cx)
ch := make(chan error)
go func() {
_, err := http.DefaultClient.Do(req)
select {
case <-cx.Done():
// 已超时
default:
ch <- err
}
}()
// 模拟用户取消请求
go func() {
time.Sleep(100 * time.Millisecond)
cancel()
}()
select {
case err := <-ch:
if err != nil {
// HTTP 错误
panic(err)
}
print("没有错误")
case <-cx.Done():
panic(cx.Err())
}
}
英文:
The standard way is to use a context of type context.Context
and pass it around to all the functions that need to know when the request is cancelled.
func httpDo(ctx context.Context, req *http.Request, f func(*http.Response, error) error) error {
// Run the HTTP request in a goroutine and pass the response to f.
tr := &http.Transport{}
client := &http.Client{Transport: tr}
c := make(chan error, 1)
go func() { c <- f(client.Do(req)) }()
select {
case <-ctx.Done():
tr.CancelRequest(req)
<-c // Wait for f to return.
return ctx.Err()
case err := <-c:
return err
}
}
golang.org/x/net/context
// A Context carries a deadline, cancelation signal, and request-scoped values
// across API boundaries. Its methods are safe for simultaneous use by multiple
// goroutines.
type Context interface {
// Done returns a channel that is closed when this Context is canceled
// or times out.
Done() <-chan struct{}
// Err indicates why this context was canceled, after the Done channel
// is closed.
Err() error
// Deadline returns the time when this Context will be canceled, if any.
Deadline() (deadline time.Time, ok bool)
// Value returns the value associated with key or nil if none.
Value(key interface{}) interface{}
}
Source and more on https://blog.golang.org/context
Update
As Paulo mentioned, Request.Cancel is now deprecated and the author should pass the context to the request itself(using *Request.WithContext) and use the cancellation channel of the context(to cancel the request).
package main
import (
"context"
"net/http"
"time"
)
func main() {
cx, cancel := context.WithCancel(context.Background())
req, _ := http.NewRequest("GET", "http://google.com", nil)
req = req.WithContext(cx)
ch := make(chan error)
go func() {
_, err := http.DefaultClient.Do(req)
select {
case <-cx.Done():
// Already timedout
default:
ch <- err
}
}()
// Simulating user cancel request
go func() {
time.Sleep(100 * time.Millisecond)
cancel()
}()
select {
case err := <-ch:
if err != nil {
// HTTP error
panic(err)
}
print("no error")
case <-cx.Done():
panic(cx.Err())
}
}
答案3
得分: 7
不,client.Post是一个方便的包装器,适用于90%的情况,不需要请求取消。
可能只需重新实现您的客户端以访问底层的Transport对象,该对象具有CancelRequest()函数。
这是一个快速的示例:
package main
import (
"log"
"net/http"
"time"
)
func main() {
req, _ := http.NewRequest("GET", "http://google.com", nil)
tr := &http.Transport{} // TODO: copy defaults from http.DefaultTransport
client := &http.Client{Transport: tr}
c := make(chan error, 1)
go func() {
resp, err := client.Do(req)
// 处理响应...
_ = resp
c <- err
}()
// 模拟用户取消请求的通道
user := make(chan struct{}, 0)
go func() {
time.Sleep(100 * time.Millisecond)
user <- struct{}{}
}()
for {
select {
case <-user:
log.Println("取消请求")
tr.CancelRequest(req)
case err := <-c:
log.Println("客户端完成:", err)
return
}
}
}
英文:
Nope, client.Post is a handy wrapper for 90% of use-cases where request cancellation is not needed.
Probably it will be enough simply to reimplement your client to get access to underlying Transport object, which has CancelRequest() function.
Just a quick example:
package main
import (
"log"
"net/http"
"time"
)
func main() {
req, _ := http.NewRequest("GET", "http://google.com", nil)
tr := &http.Transport{} // TODO: copy defaults from http.DefaultTransport
client := &http.Client{Transport: tr}
c := make(chan error, 1)
go func() {
resp, err := client.Do(req)
// handle response ...
_ = resp
c <- err
}()
// Simulating user cancel request channel
user := make(chan struct{}, 0)
go func() {
time.Sleep(100 * time.Millisecond)
user <- struct{}{}
}()
for {
select {
case <-user:
log.Println("Cancelling request")
tr.CancelRequest(req)
case err := <-c:
log.Println("Client finished:", err)
return
}
}
}
答案4
得分: 2
自Go 1.13版本开始,除了使用NewRequest
和Request.WithContext
之外,还可以使用NewRequestWithContext
函数来为HTTP请求附加context.Context
。这个函数接受一个控制创建的传出请求整个生命周期的Context
,适用于Client.Do
和Transport.RoundTrip
的使用。
以下是示例代码的转换:
req, err := http.NewRequest(...)
if err != nil {...}
req.WithContext(ctx)
转换为:
req, err := http.NewRequestWithContext(ctx, ...)
if err != nil {...}
英文:
To add to the other answers that attach context.Context
to http requests, since 1.13 we have:
> A new function NewRequestWithContext
has been added and it accepts a Context
that controls the entire lifetime of the created outgoing Request
, suitable for use with Client.Do
and Transport.RoundTrip
.
https://golang.org/doc/go1.13#net/http
This function can be used instead of using NewRequest
and then Request.WithContext
.
req, err := http.NewRequest(...)
if err != nil {...}
req.WithContext(ctx)
becomes
req, err := http.NewRequestWithContext(ctx, ...)
if err != nil {...}
</details>
# 答案5
**得分**: 1
@Paulo Casaretto的答案是正确的,应该使用[http.Request.WithContext](https://golang.org/pkg/net/http/#Request.WithContext)。
以下是一个完整的示例(请注意时间数字:5、10、30秒)。
HTTP服务器:
```go
package main
import (
"fmt"
"log"
"net/http"
"time"
)
func handler(w http.ResponseWriter, r *http.Request) {
fmt.Println("before sleep")
time.Sleep(10 * time.Second)
fmt.Println("after sleep")
fmt.Fprintf(w, "Hi")
}
func main() {
http.HandleFunc("/", handler)
log.Fatal(http.ListenAndServe(":9191", nil))
}
HTTP服务器控制台输出:
before sleep
after sleep
HTTP客户端:
package main
import (
"context"
"fmt"
"net/http"
"time"
)
func main() {
ctx, cancel := context.WithCancel(context.Background())
go func() {
fmt.Println("before request")
client := &http.Client{Timeout: 30 * time.Second}
req, err := http.NewRequest("GET", "http://127.0.0.1:9191", nil)
if err != nil {
panic(err)
}
req = req.WithContext(ctx)
_, err = client.Do(req)
if err != nil {
panic(err)
}
fmt.Println("will not reach here")
}()
time.Sleep(5 * time.Second)
cancel()
fmt.Println("finished")
}
HTTP客户端控制台输出:
before request
finished
英文:
@Paulo Casaretto 's answer is right, should using
http.Request.WithContext.
Here is a full demo (be aware of the time numbers: 5, 10, 30 seconds).
HTTP Server:
package main
import (
"fmt"
"log"
"net/http"
"time"
)
func handler(w http.ResponseWriter, r *http.Request) {
fmt.Println("before sleep")
time.Sleep(10 * time.Second)
fmt.Println("after sleep")
fmt.Fprintf(w, "Hi")
}
func main() {
http.HandleFunc("/", handler)
log.Fatal(http.ListenAndServe(":9191", nil))
}
The HTTP Server console print:
before sleep
after sleep
HTTP Client:
package main
import (
"context"
"fmt"
"net/http"
"time"
)
func main() {
ctx, cancel := context.WithCancel(context.Background())
go func() {
fmt.Println("before request")
client := &http.Client{Timeout: 30 * time.Second}
req, err := http.NewRequest("GET", "http://127.0.0.1:9191", nil)
if err != nil {
panic(err)
}
req = req.WithContext(ctx)
_, err = client.Do(req)
if err != nil {
panic(err)
}
fmt.Println("will not reach here")
}()
time.Sleep(5 * time.Second)
cancel()
fmt.Println("finished")
}
The HTTP Client console print:
before request
finished
通过集体智慧和协作来改善编程学习和解决问题的方式。致力于成为全球开发者共同参与的知识库,让每个人都能够通过互相帮助和分享经验来进步。
评论