github.com/pinoOgni/urlpath

urlpath

urlpath is a Golang library for matching paths against a template, or constructing paths using a template. It's meant for applications that take in REST-like URL paths, and need to validate and extract data from those paths.

This is easiest explained with an example:

import "github.com/ucarion/urlpath"

var getBookPath = urlpath.New("/shelves/:shelf/books/:book")

func main() {
  inputPath := "/shelves/foo/books/bar"
  match, ok := getBookPath.Match(inputPath)
  if !ok {
    // handle the input not being valid
    return
  }

  // Output:
  //
  // foo
  // bar
  fmt.Println(match.Params["shelf"])
  fmt.Println(match.Params["book"])
}

One slightly fancier feature is support for trailing segments, like if you have a path that ends with a filename. For example, a GitHub-like API might need to deal with paths like:

/ucarion/urlpath/blob/master/src/foo/bar/baz.go

You can do this with a path that ends with "*". This works like:

path := urlpath.New("/:user/:repo/blob/:branch/*")

match, ok := path.Match("/ucarion/urlpath/blob/master/src/foo/bar/baz.go")
fmt.Println(match.Params["user"])   // ucarion
fmt.Println(match.Params["repo"])   // urlpath
fmt.Println(match.Params["branch"]) // master
fmt.Println(match.Trailing)         // src/foo/bar/baz.go

Additionally, you can call Build to construct a path from a template:

path := urlpath.New("/:user/:repo/blob/:branch/*")

res, ok := path.Build(urlpath.Match{
  Params: map[string]string{
    "user": "ucarion",
    "repo": "urlpath",
    "branch": "master",
  },
  Trailing: "src/foo/bar/baz.go",
})

fmt.Println(res) // /ucarion/urlpath/blob/master/src/foo/bar/baz.go

How it works

urlpath operates on the basis of "segments", which is basically the result of splitting a path by slashes. When you call urlpath.New, each of the segments in the input is treated as either:

• A parameterized segment, like :user. All segments starting with : are considered parameterized. Any corresponding segment in the input (even the empty string!) will be satisfactory, and will be sent to Params in the outputted Match. For example, data corresponding to :user would go in Params["user"].
• An exact-match segment, like users. Only segments exactly equal to users will be satisfactory.
• A "trailing" segment, *. This is only treated specially when it's the last segment -- otherwise, it's just a usual exact-match segment. Any leftover data in the input, after all previous segments were satisfied, goes into Trailing in the outputted Match.

Performance

Although performance wasn't the top priority for this library, urlpath does typically perform better than an equivalent regular expression. In other words, this:

path := urlpath.New("/test/:foo/bar/:baz")
matches := path.Match(...)

Will usually perform better than this:

r := regexp.MustCompile("/test/(?P<foo>[^/]+)/bar/(?P<baz>[^/]+)")
matches := r.FindStringSubmatch(...)

The results of go test -benchmem -bench .:

goos: darwin
goarch: amd64
pkg: github.com/ucarion/urlpath
BenchmarkMatch/without_trailing_segments/urlpath-8 	 1436247	       819 ns/op	     784 B/op	      10 allocs/op
BenchmarkMatch/without_trailing_segments/regex-8   	  693924	      1816 ns/op	     338 B/op	      10 allocs/op
BenchmarkMatch/with_trailing_segments/urlpath-8    	 1454750	       818 ns/op	     784 B/op	      10 allocs/op
BenchmarkMatch/with_trailing_segments/regex-8      	  592644	      2365 ns/op	     225 B/op	       8 allocs/op

Do your own benchmarking if performance matters a lot to you. See BenchmarkMatch in urlpath_test.go for the code that gives these results.