I'm trying to set an http header for multiple handlers. My first thought was to make a custom write function that would set the header before writing the response like the code sample at the bottom.
However, when I pass a pointer to the http.ResponseWriter and try to access it from my function it tells me that "type *http.ResponseWriter has no Header method".
What is the best way to set headers for multiple handlers, and also why isn't the pointer working the way I want it to?
func HelloServer(w http.ResponseWriter, req *http.Request) {
type Message struct {
Name string
Body string
Time int64
}
m := Message{"Alice", "Hello", 1294706395881547000}
b, _ := json.Marshal(m)
WriteJSON(&w, b)
}
func WriteJSON(wr *http.ResponseWriter, rawJSON []byte) {
*wr.Header().Set("Content-Type", "application/json")
io.WriteString(*wr, string(rawJSON))
}
func main() {
http.HandleFunc("/json", HelloServer)
err := http.ListenAndServe(":9000", nil)
if err != nil {
log.Fatal("ListenAndServer: ", err)
}
}
I'm not sure about the multiple handlers thing, but I do know why the code you wrote is failing. The key is that the line:
*wr.Header().Set("Content-Type", "application/json")
is being interpreted, because of operator precedence, as:
*(wr.Header().Set("Content-Type", "application/json"))
Since wr has the type *http.ResponseWriter, which is a pointer to and interface, rather than the interface itself, this won't work. I assume that you knew that, which is why you did *wr. I assume what you meant to imply to the compiler is:
(*wr).Header().Set("Content-Type", "application/json")
If I'm not mistaken, that should compile and behave properly.
You don't need to use *wr as it already references a pointer.
wr.Header().Set("Content-Type", "application/json") should be sufficient.
If you want to set "global" headers for every request, you can create a function that satisfies http.HandleFunc (go.auth has a good example) and then wrap your handlers like so:
http.HandleFunc("/hello", Defaults(helloHandler))
Also take a look at the net/http documentation, which has further examples.
I wrap my handlers with an error handler
which calls my AddSafeHeader function.
I based it on http://golang.org/doc/articles/error_handling.html
but it doesn't use ServeHTTP so it works with appstats:
http.Handle("/", appstats.NewHandler(util.ErrorHandler(rootHandler)))
Here:
package httputil
import (
"appengine"
"net/http"
"html/template"
)
func AddSafeHeaders(w http.ResponseWriter) {
w.Header().Set("X-Content-Type-Options", "nosniff")
w.Header().Set("X-XSS-Protection", "1; mode=block")
w.Header().Set("X-Frame-Options", "SAMEORIGIN")
w.Header().Set("Strict-Transport-Security", "max-age=2592000; includeSubDomains")
}
// Redirect to a fixed URL
type redirectHandler struct {
url string
code int
}
func (rh *redirectHandler) ServeHTTP(w http.ResponseWriter, r *http.Request) {
Redirect(w, r, rh.url, rh.code)
}
func Redirect(w http.ResponseWriter, r *http.Request, urlStr string, code int) {
AddSafeHeaders(w)
http.Redirect(w, r, urlStr, code)
}
// RedirectHandler returns a request handler that redirects
// each request it receives to the given url using the given
// status code.
func RedirectHandler(url string, code int) http.Handler {
return &redirectHandler{url, code}
}
func ErrorHandler(fn func(appengine.Context, http.ResponseWriter, *http.Request)) func(appengine.Context, http.ResponseWriter, *http.Request) {
return func(c appengine.Context, w http.ResponseWriter, r *http.Request) {
defer func() {
if err, ok := recover().(error); ok {
c.Errorf("%v", err)
w.WriteHeader(http.StatusInternalServerError)
errorTemplate.Execute(w, err)
}
}()
AddSafeHeaders(w)
fn(c, w, r)
}
}
// Check aborts the current execution if err is non-nil.
func Check(err error) {
if err != nil {
panic(err)
}
}
var errorTemplate = template.Must(template.New("error").Parse(errorTemplateHTML))
const errorTemplateHTML = `
<html>
<head>
<title>XXX</title>
</head>
<body>
<h2>An error occurred:</h2>
<p>{{.}}</p>
</body>
</html>
`
http.ResponseWriter is an interface.
You should probably not be using a pointer to an interface. In net/http/server.go, the unexported response struct is the actual type that implements ResponseWriter when the server calls your handler, and importantly, when the server actually calls the handler's ServeHTTP, it passes a *response. It's already a pointer, but you don't see that because ResonseWriter is an interface. (the response pointer is created here, by (c *conn).readRequest. (The links will likely be for the wrong lines the future, but you should be able to locate them).
That's why the ServeHTTP function required to implement Handler is:
ServeHTTP(w ResponseWriter, r *Request)
i.e. not a pointer to ResponseWriter, as this declaration already permits a pointer to a struct that implements the ResponseWriter interface.
As I am new to Go, I created a minimal contrived example, based on elithrar's answer, which shows how to easily add headers to all your routes / responses. We do so, by creating a function that satisfies the http.HandlerFunc interface, then wraps the route handler functions:
package main
import (
"encoding/json"
"log"
"net/http"
"github.com/gorilla/mux"
)
// Hello world.
func Hello(w http.ResponseWriter, r *http.Request) {
json.NewEncoder(w).Encode("Hello World")
}
// HelloTwo world
func HelloTwo(w http.ResponseWriter, r *http.Request) {
json.NewEncoder(w).Encode("Hello Two World")
}
// JSONHeaders conforms to the http.HandlerFunc interface, and
// adds the Content-Type: application/json header to each response.
func JSONHeaders(handler http.HandlerFunc) http.HandlerFunc {
return func(w http.ResponseWriter, r *http.Request) {
w.Header().Set("Content-Type", "application/json")
handler(w, r)
}
}
func main() {
router := mux.NewRouter()
// Now, instead of calling your handler function directly, pass it into the wrapper function.
router.HandleFunc("/", JSONHeaders(Hello)).Methods("GET")
router.HandleFunc("/hellotwo", JSONHeaders(HelloTwo)).Methods("GET")
log.Fatal(http.ListenAndServe(":3000", router))
}
Results:
$ go run test.go &
$ curl -i localhost:3000/
HTTP/1.1 200 OK
Content-Type: application/json
Date: Thu, 28 Feb 2019 22:27:04 GMT
Content-Length: 14
"Hello World"
What i end up doing:
// Accepts a user supplied http.HandlerFunc and then modifies it in various ways. In this case, it adds two new headers.
func CommonlHandler(h http.HandlerFunc) http.HandlerFunc {
return func (rs http.ResponseWriter, rq *http.Request) {
rs.Header().Add("Server", "Some server")
rs.Header().Add("Cache-Control", "no-store")
h(rs, rq)
}
// somewhere down the line, where you're setting up http request handlers
serveMux := http.NewServeMux()
serveMux.HandleFunc("/", CommonHandler(func(rs http.ResponseWriter, rq *http.Request) {
// Handle request as usual. Since it's wrapped in the CommonHandler and we've set some headers there, responses to requests to "/" will contain those headers.
// Be mindful what modifications you're doing here. If, for ex., you decide you want to apply different caching strategy than the Common one, since this will be passed to the CommonHandler, your changes will be overwritten and lost. So it may be a good idea to introduce checks in CommonHandler to determine whether the header exists, before you decide to create it.
}))
serveMux.HandleFunc("/contact", CommonHandler(func(rs http.ResponseWriter, rq *http.Request) {
// handle another request as usual
}))
Related
My original question here was flagged as a duplicate of this question. I had no luck implementing it and suspect my problem is misunderstood, so with my question closed, I'm starting fresh with a more specific question.
I'm trying to set a cookie based on a response header from within middleware in request that is reverse proxied.
Here's the workflow:
User requests http://example.com/foo/bar
Go app uses ReverseProxy to proxy that request to http://baz.com
baz.com sets a response header X-FOO
Go app modifies response by setting a MYAPPFOO cookie with the value of the X-FOO response header
The cookie is written to the user's browser
It was suggested that a custom http.ResponseWriter will work, but after trying and searching for more information, it is not clear how to approach this.
Since I'm failing to grasp the concept of a custom ResponseWriter for my use case, I'll post code that demonstrates more precisely what I was trying to do at the point I got stuck:
package main
import (
"github.com/gorilla/mux"
"log"
"net/http"
"net/http/httputil"
"net/url"
)
func setCookie(w http.ResponseWriter, name string, value string) {
...
http.SetCookie(w, &cookie)
}
func handler(next http.Handler) http.Handler {
return http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
// setCookie() works here
// but I cannot access w.Header().Get("X-FOO")
next.ServeHTTP(w, r)
// I can access w.Header().Get("X-FOO") here
// but setCookie() does not cookie the user's browser
// If I could do it all in one place, this is what I would do:
if r.Method == "POST" && r.URL.String() == "/login" {
foo := w.Header().Get("X-FOO")
setCookie(w, "MYAPPFOO", foo)
}
})
}
func main() {
r := mux.NewRouter()
r.Use(handler)
proxy := httputil.NewSingleHostReverseProxy("https://baz.example.com/")
r.PathPrefix("/").Handler(proxy)
log.Fatal(http.ListenAndServe(":9001", r))
}
As a side note, I was able to make this work with ReverseProxy.ModifyResponse as recommended in the comments of my last question but I'd really like to achieve this with middleware to keep the code that dynamically creates proxies from config clean. (not in the example code)
From the documentation on http.ResponseWriter methods:
(emphasis added)
Header() http.Header:
Changing the header map after a call to WriteHeader (or Write) has no
effect unless the modified headers are trailers.
WriteHeader(statusCode int):
WriteHeader sends an HTTP response header with the provided status
code.
Write([]byte) (int, error):
If WriteHeader has not yet been called, Write calls
WriteHeader(http.StatusOK) before writing the data.
This should highlight the reason why, you can't set a cookie after the next.ServeHTTP(w, r) call, which is that one of the handlers in the middleware chain executed by that call is calling either WriteHeader or Write directly or indirectly.
So to be able set the cookie after the next.ServeHTTP(w, r) call you need to make sure that none of the handlers in the middleware chain calls WriteHeader or Write on the original http.ResponseWriter instance. One way to do this is to wrap the original instance in a custom http.ResponseWriter implementation that will postpone the writing of the response until after you're done with setting the cookie.
For example something like this:
type responsewriter struct {
w http.ResponseWriter
buf bytes.Buffer
code int
}
func (rw *responsewriter) Header() http.Header {
return rw.w.Header()
}
func (rw *responsewriter) WriteHeader(statusCode int) {
rw.code = statusCode
}
func (rw *responsewriter) Write(data []byte) (int, error) {
return rw.buf.Write(data)
}
func (rw *responsewriter) Done() (int64, error) {
if rw.code > 0 {
rw.w.WriteHeader(rw.code)
}
return io.Copy(rw.w, &rw.buf)
}
And you would use it like this in your middleware:
func handler(next http.Handler) http.Handler {
return http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
rw := &responsewriter{w: w}
next.ServeHTTP(rw, r)
if r.Method == "POST" && r.URL.String() == "/login" {
foo := rw.Header().Get("X-FOO")
setCookie(rw, "MYAPPFOO", foo)
}
if _, err := rw.Done(); err != nil {
log.Println(err)
}
})
}
I am fairly new to Go and have not been able to find any information on this, maybe it is just not possible at this time.
I am trying to delete or replace a mux route (using http.NewServeMux, or gorilla's mux.Router). My end goal is to be able to enable/disable a route or set of routes without having to restart the program.
I can probably accomplish this on a handler to handler basis and just return 404 if that feature is "disabled", but I would rather find a more general way to do this since I would like to implement it for every route in my application.
Or would I be better off just keeping track of disabled url patterns and using some middleware to prevent handler execution?
If someone can at least point me in the right direction, I will absolutely post code examples of a solution assuming there is one. Thanks!
There's no built in way, but it is easy enough to implement play.
type HasHandleFunc interface { //this is just so it would work for gorilla and http.ServerMux
HandleFunc(pattern string, handler func(w http.ResponseWriter, req *http.Request))
}
type Handler struct {
http.HandlerFunc
Enabled bool
}
type Handlers map[string]*Handler
func (h Handlers) ServeHTTP(w http.ResponseWriter, r *http.Request) {
path := r.URL.Path
if handler, ok := h[path]; ok && handler.Enabled {
handler.ServeHTTP(w, r)
} else {
http.Error(w, "Not Found", http.StatusNotFound)
}
}
func (h Handlers) HandleFunc(mux HasHandleFunc, pattern string, handler http.HandlerFunc) {
h[pattern] = &Handler{handler, true}
mux.HandleFunc(pattern, h.ServeHTTP)
}
func main() {
mux := http.NewServeMux()
handlers := Handlers{}
handlers.HandleFunc(mux, "/", func(w http.ResponseWriter, r *http.Request) {
w.Write([]byte("this will show once"))
handlers["/"].Enabled = false
})
http.Handle("/", mux)
http.ListenAndServe(":9020", nil)
}
Yes you can.
One way to do it is to have a sturct that implement http.Handle interface with the method
ServeHTTP.
Then have the struct contain another muxer like gorilla's
and finally have an atomic Switch to enable/ disable the subrouting
This is a working example of what I mean:
package main
import (
"fmt"
"github.com/gorilla/mux"
"net/http"
"sync/atomic"
)
var recording int32
func isRecording() bool {
return atomic.LoadInt32(&recording) != 0
}
func setRecording(shouldRecord bool) {
if shouldRecord {
atomic.StoreInt32(&recording, 1)
} else {
atomic.StoreInt32(&recording, 0)
}
}
type SwitchHandler struct {
mux http.Handler
}
func (s *SwitchHandler) ServeHTTP(w http.ResponseWriter, r *http.Request) {
if isRecording() {
fmt.Printf("Switch Handler is Recording\n")
s.mux.ServeHTTP(w, r)
return
}
fmt.Printf("Switch Handler is NOT Recording\n")
w.WriteHeader(http.StatusNotFound)
fmt.Fprintf(w, "NOT Recording\n")
}
func main() {
router := mux.NewRouter()
router.HandleFunc("/success/", func(w http.ResponseWriter, r *http.Request) {
fmt.Fprintf(w, "Recording\n")
})
handler := &SwitchHandler{mux: router}
setRecording(false)
http.Handle("/", handler)
http.ListenAndServe(":8080", nil)
}
According to https://github.com/gorilla/mux/issues/82 it is suggested to swap the router instead of deleting routes. Existing connections will stay open.
I am designing my handlers to return a http.Handler. Here's the design of my handlers:
func Handler() http.Handler {
return http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
})
}
My middleware is designed to accept an http.Handler and then call the handler once the middleware has finished performing its operations. Here's the design of my middleware:
func Middleware(next http.Handler) http.Handler {
return http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
// Middleware operations
next.ServeHTTP(w, r)
})
}
Considering the design of my middleware and handlers, what is the proper way of passing information from the middleware to the handler? The information that I am trying to pass from my middleware to the handlers is a JSON web token parsed from the request body. If I do not pass the parsed JWT to the handler, then I will need to parse the JWT again in my handlers. Parsing the request body for a JWT in both the middleware and handler seems wasteful. Just in case this information is relevant, I am using the standard net/http library with gorilla mux.
Since you're already using Gorilla take a look at the context package.
(This is nice if you don't want to change your method signatures.)
import (
"github.com/gorilla/context"
)
...
func Middleware(next http.Handler) http.Handler {
return http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
// Middleware operations
// Parse body/get token.
context.Set(r, "token", token)
next.ServeHTTP(w, r)
})
}
...
func Handler() http.Handler {
return http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
token := context.Get(r, "token")
})
}
Update
The Gorilla context package is now in maintenance mode
per the repo:
Note gorilla/context, having been born well before context.Context existed, does not play well with the shallow copying of the request that http.Request.WithContext (added to net/http Go 1.7 onwards) performs.
Using gorilla/context may lead to memory leaks under those conditions, as the pointers to each http.Request become "islanded" and will not be cleaned up when the response is sent.
You should use the http.Request.Context() feature in Go 1.7.
The proper way to pass request scoped data would now be the context package in the standard library.
https://golang.org/pkg/context/
You can access it with request.Context on an http.Request.
A first approach, similar to the question, is in codemodus/chain by Daved.
Package chain aids the composition of Handler wrapper chains that carry request-scoped data.
It uses the notion of Context, coupled with a Context handler:
func ctxHandler(ctx context.Context, w http.ResponseWriter, r *http.Request) {
// ...
if s, ok := getMyString(ctx); ok {
// s = "Send this down the line."
}
// ...
}
Another approach: You can have a look at "Custom Handlers and Avoiding Globals in Go Web Applications", by Matt Silverlock (elithrar). (full example here)
The idea is to define ServeHTTP on a type which include the relevant context.
// We've turned our original appHandler into a struct with two fields:
// - A function type similar to our original handler type (but that now takes an *appContext)
// - An embedded field of type *appContext
type appHandler struct {
*appContext
h func(*appContext, http.ResponseWriter, *http.Request) (int, error)
}
// Our ServeHTTP method is mostly the same, and also has the ability to
// access our *appContext's fields (templates, loggers, etc.) as well.
func (ah appHandler) ServeHTTP(w http.ResponseWriter, r *http.Request) {
// Updated to pass ah.appContext as a parameter to our handler type.
status, err := ah.h(ah.appContext, w, r)
if err != nil {
log.Printf("HTTP %d: %q", status, err)
switch status {
case http.StatusNotFound:
http.NotFound(w, r)
// And if we wanted a friendlier error page, we can
// now leverage our context instance - e.g.
// err := ah.renderTemplate(w, "http_404.tmpl", nil)
case http.StatusInternalServerError:
http.Error(w, http.StatusText(status), status)
default:
http.Error(w, http.StatusText(status), status)
}
}
}
In the appContext struct, you would put any data you want to pass around.
I'm reasonably new to golang and am trying to do work out the best way to do this idiomatically.
I have an array of routes I am statically defining and passing to gorilla/mux. I am wrapping each handler function with something to time the request and handle panics (mainly so I could understand how the wrapping worked).
I want them each to be able to have access to a 'context' - a struct that's going to be one-per-http-server, which might have things like database handles, config etc. What I don't want to do is use a static global variable.
The way I'm currently doing it I can give the wrappers access to the context structure, but I can't see how to get this into the actual handler, as it wants that to be an http.HandlerFunc. I thought what I could do is convert http.HandlerFunc into a type of my own that was a receiver for Context (and do similarly for the wrappers, but (after much playing about) I couldn't then get Handler() to accept this.
I can't help but think I'm missing something obvious here. Code below.
package main
import (
"fmt"
"github.com/gorilla/mux"
"html"
"log"
"net/http"
"time"
)
type Route struct {
Name string
Method string
Pattern string
HandlerFunc http.HandlerFunc
}
type Context struct {
route *Route
// imagine other stuff here, like database handles, config etc.
}
type Routes []Route
var routes = Routes{
Route{
"Index",
"GET",
"/",
index,
},
// imagine lots more routes here
}
func wrapLogger(inner http.Handler, context *Context) http.Handler {
return http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
start := time.Now()
inner.ServeHTTP(w, r)
log.Printf(
"%s\t%s\t%s\t%s",
r.Method,
r.RequestURI,
context.route.Name,
time.Since(start),
)
})
}
func wrapPanic(inner http.Handler, context *Context) http.Handler {
return http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
defer func() {
if err := recover(); err != nil {
log.Printf("panic caught: %+v", err)
http.Error(w, http.StatusText(500), 500)
}
}()
inner.ServeHTTP(w, r)
})
}
func newRouter() *mux.Router {
router := mux.NewRouter().StrictSlash(true)
for _, route := range routes {
// the context object is created here
context := Context {
&route,
// imagine more stuff here
}
router.
Methods(route.Method).
Path(route.Pattern).
Name(route.Name).
Handler(wrapLogger(wrapPanic(route.HandlerFunc, &context), &context))
}
return router
}
func index(w http.ResponseWriter, r *http.Request) {
// I want this function to be able to have access to 'context'
fmt.Fprintf(w, "Hello, %q", html.EscapeString(r.URL.Path))
}
func main() {
fmt.Print("Starting\n");
router := newRouter()
log.Fatal(http.ListenAndServe("127.0.0.1:8080", router))
}
Here's a way to do it, but it seems pretty horrible. I can't help but think there must be some better way to do it - perhaps to subclass (?) http.Handler.
package main
import (
"fmt"
"github.com/gorilla/mux"
"html"
"log"
"net/http"
"time"
)
type Route struct {
Name string
Method string
Pattern string
HandlerFunc ContextHandlerFunc
}
type Context struct {
route *Route
secret string
}
type ContextHandlerFunc func(c *Context, w http.ResponseWriter, r *http.Request)
type Routes []Route
var routes = Routes{
Route{
"Index",
"GET",
"/",
index,
},
}
func wrapLogger(inner ContextHandlerFunc) ContextHandlerFunc {
return func(c *Context, w http.ResponseWriter, r *http.Request) {
start := time.Now()
inner(c, w, r)
log.Printf(
"%s\t%s\t%s\t%s",
r.Method,
r.RequestURI,
c.route.Name,
time.Since(start),
)
}
}
func wrapPanic(inner ContextHandlerFunc) ContextHandlerFunc {
return func(c *Context, w http.ResponseWriter, r *http.Request) {
defer func() {
if err := recover(); err != nil {
log.Printf("panic caught: %+v", err)
http.Error(w, http.StatusText(500), 500)
}
}()
inner(c, w, r)
}
}
func newRouter() *mux.Router {
router := mux.NewRouter().StrictSlash(true)
for _, route := range routes {
context := Context{
&route,
"test",
}
router.Methods(route.Method).
Path(route.Pattern).
Name(route.Name).
HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
wrapLogger(wrapPanic(route.HandlerFunc))(&context, w, r)
})
}
return router
}
func index(c *Context, w http.ResponseWriter, r *http.Request) {
fmt.Fprintf(w, "Hello, %q secret is %s\n", html.EscapeString(r.URL.Path), c.secret)
}
func main() {
fmt.Print("Starting\n")
router := newRouter()
log.Fatal(http.ListenAndServe("127.0.0.1:8080", router))
}
I am learning Go and currently in the middle of a nearly identical problem, and this is how I've dealt with it:
First, I think you missed an important detail: There are no global variables in Go. The widest scope you can have for a variable is package scope. The only true globals in Go are predeclared identifiers like true and false (and you can't change these or make your own).
So, it's perfectly fine to set a variable scoped to package main to hold context for your program. Coming from a C/C++ background this took me a little time to get used to. Since the variables are package scoped, they do not suffer from the problems of global variables. If something in another package needs such a variable, you will have to pass it explicitly.
Don't be afraid to use package variables when it makes sense. This can help you reduce complexity in your program, and in a lot of cases make your custom handlers much simpler (where calling http.HandlerFunc() and passing a closure will suffice).
Such a simple handler might look like this:
func simpleHandler(c Context, next http.Handler) http.Handler {
return http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
// FIXME Do something with our context
next.ServeHTTP(w, r)
})
}
and be used by:
r = mux.NewRouter()
http.Handle("/", simpleHandler(c, r))
If your needs are more complex, you may need to implement your own http.Handler. Remember that an http.Handler is just an interface which implements ServeHTTP(w http.ResponseWriter, r *http.Request).
This is untested but should get you about 95% of the way there:
package main
import (
"net/http"
)
type complicatedHandler struct {
h http.Handler
opts ComplicatedOptions
}
type ComplicatedOptions struct {
// FIXME All of the variables you want to set for this handler
}
func (m complicatedHandler) ServeHTTP(w http.ResponseWriter, r *http.Request) {
// FIXME Do stuff before serving page
// Call the next handler
m.h.ServeHTTP(w, r)
// FIXME Do stuff after serving page
}
func ComplicatedHandler(o ComplicatedOptions) func(http.Handler) http.Handler {
return func(h http.Handler) http.Handler {
return complicatedHandler{h, o}
}
}
To use it:
r := mux.NewRouter()
// FIXME: Add routes to the mux
opts := ComplicatedOptions{/* FIXME */}
myHandler := ComplicatedHandler(opts)
http.Handle("/", myHandler(r))
For a more developed handler example see basicAuth in goji/httpauth, from which this example was shamelessly ripped off.
Some further reading:
A Recap of Request Handling
Making and Using HTTP Middleware
justinas/alice (for chaining lots of handlers)
I am fairly new to Go and have not been able to find any information on this, maybe it is just not possible at this time.
I am trying to delete or replace a mux route (using http.NewServeMux, or gorilla's mux.Router). My end goal is to be able to enable/disable a route or set of routes without having to restart the program.
I can probably accomplish this on a handler to handler basis and just return 404 if that feature is "disabled", but I would rather find a more general way to do this since I would like to implement it for every route in my application.
Or would I be better off just keeping track of disabled url patterns and using some middleware to prevent handler execution?
If someone can at least point me in the right direction, I will absolutely post code examples of a solution assuming there is one. Thanks!
There's no built in way, but it is easy enough to implement play.
type HasHandleFunc interface { //this is just so it would work for gorilla and http.ServerMux
HandleFunc(pattern string, handler func(w http.ResponseWriter, req *http.Request))
}
type Handler struct {
http.HandlerFunc
Enabled bool
}
type Handlers map[string]*Handler
func (h Handlers) ServeHTTP(w http.ResponseWriter, r *http.Request) {
path := r.URL.Path
if handler, ok := h[path]; ok && handler.Enabled {
handler.ServeHTTP(w, r)
} else {
http.Error(w, "Not Found", http.StatusNotFound)
}
}
func (h Handlers) HandleFunc(mux HasHandleFunc, pattern string, handler http.HandlerFunc) {
h[pattern] = &Handler{handler, true}
mux.HandleFunc(pattern, h.ServeHTTP)
}
func main() {
mux := http.NewServeMux()
handlers := Handlers{}
handlers.HandleFunc(mux, "/", func(w http.ResponseWriter, r *http.Request) {
w.Write([]byte("this will show once"))
handlers["/"].Enabled = false
})
http.Handle("/", mux)
http.ListenAndServe(":9020", nil)
}
Yes you can.
One way to do it is to have a sturct that implement http.Handle interface with the method
ServeHTTP.
Then have the struct contain another muxer like gorilla's
and finally have an atomic Switch to enable/ disable the subrouting
This is a working example of what I mean:
package main
import (
"fmt"
"github.com/gorilla/mux"
"net/http"
"sync/atomic"
)
var recording int32
func isRecording() bool {
return atomic.LoadInt32(&recording) != 0
}
func setRecording(shouldRecord bool) {
if shouldRecord {
atomic.StoreInt32(&recording, 1)
} else {
atomic.StoreInt32(&recording, 0)
}
}
type SwitchHandler struct {
mux http.Handler
}
func (s *SwitchHandler) ServeHTTP(w http.ResponseWriter, r *http.Request) {
if isRecording() {
fmt.Printf("Switch Handler is Recording\n")
s.mux.ServeHTTP(w, r)
return
}
fmt.Printf("Switch Handler is NOT Recording\n")
w.WriteHeader(http.StatusNotFound)
fmt.Fprintf(w, "NOT Recording\n")
}
func main() {
router := mux.NewRouter()
router.HandleFunc("/success/", func(w http.ResponseWriter, r *http.Request) {
fmt.Fprintf(w, "Recording\n")
})
handler := &SwitchHandler{mux: router}
setRecording(false)
http.Handle("/", handler)
http.ListenAndServe(":8080", nil)
}
According to https://github.com/gorilla/mux/issues/82 it is suggested to swap the router instead of deleting routes. Existing connections will stay open.