httprouter pass in many middleware functions - go

I'm coming from node express, and I was able to pass in as many middleware as possible, for example: routes.use('/*', ensureAuth, logImportant, ... n);
How can I do something similar when using r.GET("/", HomeIndex)?
Am I forced to do something like EnsureAuth(HomeIndex)? Because I can get that to work. Unfortunately, I'm not sure what would be a good way to add as many middlewares as I want without chaining functions together.
Is there a more elegant way so I could somehow use variadic type function to do r.GET("/", applyMiddleware(HomeIndex, m1, m2, m3, m4)? I'm trying that out right now, but I feel like there's a better way to do this.
I've looked at the httprouter issues page, can't find anything :(
Thanks!

Here is an example how I did it:
package main
import (
"context"
"fmt"
"log"
"net/http"
"github.com/julienschmidt/httprouter"
"github.com/justinas/alice"
)
// m1 is middleware 1
func m1(next http.Handler) http.Handler {
return http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
//do something with m1
log.Println("m1 start here")
next.ServeHTTP(w, r)
log.Println("m1 end here")
})
}
// m2 is middleware 2
func m2(next http.Handler) http.Handler {
return http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
//do something with m2
log.Println("m2 start here")
next.ServeHTTP(w, r)
log.Println("m2 end here")
})
}
func index(w http.ResponseWriter, r *http.Request) {
// get httprouter.Params from request context
ps := r.Context().Value("params").(httprouter.Params)
fmt.Fprintf(w, "Hello, %s", ps.ByName("name"))
}
// wrapper wraps http.Handler and returns httprouter.Handle
func wrapper(next http.Handler) httprouter.Handle {
return func(w http.ResponseWriter, r *http.Request, ps httprouter.Params) {
//pass httprouter.Params to request context
ctx := context.WithValue(r.Context(), "params", ps)
//call next middleware with new context
next.ServeHTTP(w, r.WithContext(ctx))
}
}
func main() {
router := httprouter.New()
chain := alice.New(m1, m2)
//need to wrap http.Handler to be compatible with httprouter.Handle
router.GET("/user/:name", wrapper(chain.ThenFunc(index)))
log.Fatal(http.ListenAndServe(":9000", router))
}
Link to code (you can't run it from play.golang.org though): https://play.golang.org/p/BOCt97xcoY

Related

How to remove a named route from Gorilla mux in go? [duplicate]

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.

Extending GoLang's http.ResponseWriter functionality to pre/post process responses

I am trying to write a simple http MiddleWare handler that will process an http response. Unfortunately, it does not work and I cannot figure out what mistake I am making. Any/all help is appreciated!
I am using Go Gorilla mux router
Here are illustrative parts of the code:
import (
"fmt"
"log"
"github.com/gorilla/mux"
)
:
func Start() {
router := mux.NewRouter()
router.HandleFunc("/", myHandler)
:
log.Fatal(http.ListenAndServe(":8088", Middleware(router)))
}
func myHandler(w http.ResponseWriter, r *http.Request) {
fmt.Fprintln(w, "myHandler called")
}
func Middleware(h http.Handler) http.Handler {
return http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
neww := NewProcessor(w)
h.ServeHTTP(neww, r)
})
}
type Processor struct {
http.ResponseWriter
}
func (r *Processor) Write(b []byte) (int, error) {
fmt.Printf("******* Processor writing...")
log.Print(string(b)) // log it out
return r.Write(b) // pass it to the original ResponseWriter
}
func NewProcessor(w http.ResponseWriter) http.ResponseWriter {
fmt.Printf("******* Creating new Processor...")
return &Processor{ResponseWriter: w}
}
The output I get is listed below (extra logging text omitted for clarity):
******* Creating new Processor
myHandler called
However, notice the message "******* Processor writing..." was not displayed, suggesting that the "Write" function did not get called.
What changes need to be made to allow the "Write" function to be called?
return r.Write(b) caused an infinite loop of calls to the Processor's Write() method. Replacing it with return r.ResponseWriter.Write(b) fixed the bug.
Here is the corrected code:
package main
import (
"fmt"
"log"
"net/http"
)
func main() {
mux := http.NewServeMux()
mux.HandleFunc("/", myHandler)
log.Fatal(http.ListenAndServe(":8088", Middleware(mux)))
}
func myHandler(w http.ResponseWriter, r *http.Request) {
fmt.Fprintln(w, "myHandler called")
}
func Middleware(h http.Handler) http.Handler {
return http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
new := NewProcessor(w)
h.ServeHTTP(new, r)
})
}
type Processor struct {
http.ResponseWriter
}
func (r *Processor) Write(b []byte) (int, error) {
log.Print("******* Processor writing...")
log.Print(string(b)) // log it out
return r.ResponseWriter.Write(b) // pass it to the original ResponseWriter
}
func NewProcessor(w http.ResponseWriter) http.ResponseWriter {
log.Print("******* Creating new Processor...")
return &Processor{ResponseWriter: w}
}
Output:
2016/07/21 22:59:08 ******* Creating new Processor...
2016/07/21 22:59:08 ******* Processor writing...
2016/07/21 22:59:08 myHandler called

How to implement HandlerFunc without using DefaultServeMux

If I were to use the DefaultServeMux (which I designate by passing nil as the second argument to ListenAndServe), then I have access to http.HandleFunc, which you see used below in this example from the Go wiki:
func handler(w http.ResponseWriter, r *http.Request) {
fmt.Fprintf(w, "Hi there, I love %s!", r.URL.Path[1:])
}
func main() {
http.HandleFunc("/", handler)
http.ListenAndServe(":8080", nil)
}
In my current code, I am not able to use the DefaultServeMux i.e. I'm passing a custom handler to ListenAndServe
h := &mypackage.Handler{
Database: mydb
}
http.ListenAndServe(":8080", h)
so I don't get the http.HandleFunc built in. However, I have to adapt some authorization code to my code base that requires something like http.HandleFunc. For example, if I had been using DefaultServeMux, when I hit the "/protected" route, I would want to go to the Protected handler, but only after passing through the h.AuthorizationHandlerFunc like this
h.AuthorizationHandlerFunc(Protected)
However, since I'm not using DefaultServeMux, it's not working i.e. I'm not able to pass the Protected function (and have it called) to the AuthorizationHandlerFunc. This is the implementation of the AuthorizationHandlerFunc below. You can see below that Protected never gets called.
Question: how do I implement HandlerFunc in this situation (without using DefaultServeMux)?
func (h *Handler) AuthorizationHandlerFunc(next http.HandlerFunc) http.Handler{
return http.HandlerFunc(func(w http.ResponseWriter, r *http.Request){
h.AuthorizationMiddleWare(w, r, next)
})
}
func (h *Handler) AuthorizationMiddleWare(w http.ResponseWriter, r *http.Request, next http.HandlerFunc){
//other stuff happens
log.Println("this is never getting called")
next(w,r)
}
func (h *Handler)Protected(w http.ResponseWriter, r *http.Request){
log.Println("this is never getting called")
}
Update
ServeHTTP is implemented on mypackage.Handler. Why is the Protected function not getting called, or, for that matter, the relevant code in the AuthorizationMiddleWare?
Re-implement your authorization middleware as a http.Handler :
type auth struct {
DB *sql.DB
UnauthorizedHandler http.Handler
}
func NewAuth(db *sql.DB, unauthorized http.Handler) *auth {
return auth{db, unauthorized}
}
func (a *auth) Protected(h http.Handler) http.Handler {
fn := func(w http.ResponseWriter, r *http.Request) {
// Check whether the request is valid
// If it's invalid, call your error func and make sure to *return* early!
if !valid {
a.UnauthorizedHandler.ServeHTTP(w, r)
return
}
// Call the next handler on success
h.ServeHTTP(w, r)
return
}
return http.HandlerFunc(fn)
}
func someHandler(w http.ResponseWriter, r *http.Request) {
io.WriteString(w, "Hello!\n")
}
func main() {
auth := NewAuth(db, errorHandler)
r := http.NewServeMux()
// We have a http.Handler implementation that wraps a http.HandlerFunc
// ... so we call r.Handle on our ServeMux and type-cast the wrapped func
r.Handle("/protected", auth.Protected(http.HandlerFunc(someHandler)))
// Just a simple http.HandlerFunc here
r.HandleFunc("/public", someOtherHandler)
log.Fatal(http.ListenAndServe(":8000", r))
}
Take a look at the httpauth lib I wrote for a different example with a ServeHTTP method. Both the above and explicitly creating a ServeHTTP method on your type are valid approaches.

Passing context to gorilla mux - go idioms

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)

Golang http mux change handler function

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.

Resources