Skip to content

HTTPS clone URL

Subversion checkout URL

You can clone with
or
.
Download ZIP
A kick-ass HTTP router for use in Rack

Fetching latest commit…

Cannot retrieve the latest commit at this time

Failed to load latest commit information.
benchmarks
examples
lib
spec
.gitignore
CHANGELOG
Gemfile
README.rdoc
Rakefile
http_router.gemspec

README.rdoc

HTTP Router

Introduction

When I wrote Usher, I made a few compromises in design that I wasn't totally happy with. More and more features got added to it, and eventually, it became harder to maintain. I took a few moments to work in Node.js, and wrote a router there called Sherpa, which I was happier with. But I felt that by losing more abstraction, and tackling just the problem of HTTP routing, I could come up with something even better.

Warning

This is very new code. Lots of stuff probably doesn't work right. I will likely never support all the features I had in Usher. Documentation is super-sparse.

Features

  • Supports variables, and globbing, both named and unnamed.

  • Regex support for variables.

  • Request condition support.

  • Partial matches.

  • Supports interstitial variables (e.g. /my-:variable-brings.all.the.boys/yard) and unnamed variable /one/:/two

  • Very fast and small code base (~1,000 loc).

  • Sinatra compatibility.

Usage

Please see the examples directory for a bunch of awesome rackup file examples, with tonnes of commentary. As well, the rdocs should provide a lot of useful specifics and exact usage.

HttpRouter.new

Takes the following options:

  • :default_app - The default #call made on non-matches. Defaults to a 404 generator.

  • :ignore_trailing_slash - Ignores the trailing slash when matching. Defaults to true.

  • :redirect_trailing_slash - Redirect on trailing slash matches to non-trailing slash paths. Defaults to false.

  • :middleware - Perform matching without deferring to matched route. Defaults to false.

#add(name, options)

Maps a route. The format for variables in paths is:

:variable
*glob

Everything else is treated literally. Optional parts are surrounded by brackets. Partially matching paths have a trailing *. Optional trailing slash matching is done with /?.

As well, you can escape the following characters with a backslash: ( ) : *

Once you have a route object, use HttpRouter::Route#to to add a destination and HttpRouter::Route#name to name it.

e.g.

r = HttpRouter.new
r.add('/test/:variable(.:format)').name(:my_test_path).to {|env| [200, {}, "Hey dude #{env['router.params'][:variable]}"]}
r.add('/test').redirect("http://www.google.com/")
r.add('/static').static('/my_file_system')

As well, you can support regex matching and request conditions. To add a regex match, use matching(:id => /\d+/). To match on a request condition you can use condition(:request_method => %w(POST HEAD)) or more succinctly request_method('POST', 'HEAD').

There are convenience methods HttpRouter#get, HttpRouter#post, etc for each request method.

Routes will not be recognized unless #to has been called on it.

#url(name or route, *args)

Generates a route. The args can either be a hash, a list, or a mix of both.

#call(env or Rack::Request)

Recognizes and dispatches the request.

#recognize(env or Rack::Request)

Only performs recognition.

Something went wrong with that request. Please try again.