Skip to content

Instantly share code, notes, and snippets.

@jordanorelli
Created April 7, 2012 19:22
Show Gist options
  • Star 5 You must be signed in to star a gist
  • Fork 2 You must be signed in to fork a gist
  • Save jordanorelli/2331520 to your computer and use it in GitHub Desktop.
Save jordanorelli/2331520 to your computer and use it in GitHub Desktop.
a regex router in go.
package main
import (
"fmt"
"io"
"log"
"routes"
"net/http"
)
func BlankHandler(w http.ResponseWriter, req *routes.Request) {
io.WriteString(w, "Hello.")
}
func HelloHandler(w http.ResponseWriter, req *routes.Request) {
fmt.Fprintf(w, "Hello, %s", req.Kwargs["name"])
}
func main() {
router := routes.NewRouter()
router.AddRoute("^/$", BlankHandler)
router.AddRoute("^/(?P<name>\\w+)$", HelloHandler)
if err := http.ListenAndServe(":8000", router); err != nil {
log.Fatal(err)
}
}
package routes
import (
"net/http"
"regexp"
)
// the Request struct wraps the http.Request struct, providing a slice of
// strings representing the positional arguments found in a url pattern, and a
// map[string]string called kwargs representing the named parameters captured
// in url parsing.
type Request struct {
*http.Request
Args []string
Kwargs map[string]string
}
// HandlerFunc is nearly the same as http.HandlerFunc, it simply takes a
// routes.Request object instead of an http.Request object.
type HandlerFunc func(http.ResponseWriter, *Request)
// struct Router implements http.Handler, so that it may be used with the
// default http library. It keeps a registry mapping regexes to functions for
// easier url parsing.
type Router struct {
routes []*entry
}
type entry struct {
route *Route
handler HandlerFunc
}
// this doesn't really do anything yet, but it probably will in the future.
func NewRouter() *Router {
return &Router{}
}
// implements the http.Handler interface, so that we may use our router with
// the default http package.
func (r *Router) ServeHTTP(w http.ResponseWriter, raw *http.Request) {
req, fn := r.match(raw)
if req == nil {
http.NotFound(w, raw)
return
}
fn(w, req)
}
// checks an incoming http request against our list of known routes. If the
// request matches one of the routes, the request is transformed into a
// routes.Request, and its Args and Kwargs fields are filled in based on the
// url. If no match is found, returns (nil, nil)
func (r *Router) match(req *http.Request) (*Request, HandlerFunc) {
for _, e := range r.routes {
if match := e.route.Match(req.URL.Path); match != nil {
return &Request{req, match.Args, match.Kwargs}, e.handler
}
}
return nil, nil
}
// adds a regex-based route in the normal human fashion.
func (router *Router) AddRoute(pattern string, fn HandlerFunc) {
router.routes = append(router.routes, &entry{route: NewRoute(pattern), handler: fn})
}
// right now, just embeds a regex. A "name" field should also be added here.
type Route struct {
*regexp.Regexp
}
type RouteMatch struct {
Args []string
Kwargs map[string]string
}
func NewRoute(pattern string) *Route {
return &Route{regexp.MustCompile(pattern)}
}
func (r *Route) Match(target string) *RouteMatch {
submatches := r.FindStringSubmatch(target)
if submatches == nil {
return nil
}
if len(submatches) == 1 {
return new(RouteMatch)
}
m := new(RouteMatch)
submatches = submatches[1:]
for i, name := range r.SubexpNames()[1:] {
if name == "" {
m.Args = append(m.Args, submatches[i])
} else {
if m.Kwargs == nil {
m.Kwargs = make(map[string]string)
}
m.Kwargs[name] = submatches[i]
}
}
return m
}
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment