rack-rewrite

A rack middleware for defining and applying rewrite rules. In many cases you can get away with rack-rewrite instead of writing Apache mod_rewrite rules.

Usage Examples

Usage Details

Sample rackup file

# config.ru
gem 'rack-rewrite', '~> 1.5.0'
require 'rack/rewrite'
use Rack::Rewrite do
  rewrite   '/wiki/John_Trupiano',  '/john'
  r301      '/wiki/Yair_Flicker',   '/yair'
  r302      '/wiki/Greg_Jastrab',   '/greg'
  r301      %r{/wiki/(\w+)_\w+},    '/$1'
end

Sample usage in a rails app

Rails 3 or less:

# config/application.rb
config.middleware.insert_before(Rack::Lock, Rack::Rewrite) do
  rewrite   '/wiki/John_Trupiano',  '/john'
  r301      '/wiki/Yair_Flicker',   '/yair'
  r302      '/wiki/Greg_Jastrab',   '/greg'
  r301      %r{/wiki/(\w+)_\w+},    '/$1'
end

Rails 4+ or if you use config.threadsafe, you'll need to insert_before(Rack::Runtime, Rack::Rewrite) as Rack::Lock does not exist when config.allow_concurrency == true:

config.middleware.insert_before(Rack::Runtime, Rack::Rewrite) do
  rewrite   '/wiki/John_Trupiano',  '/john'
  r301      '/wiki/Yair_Flicker',   '/yair'
  r302      '/wiki/Greg_Jastrab',   '/greg'
  r301      %r{/wiki/(\w+)_\w+},    '/$1'
end

Or insert Rack::Rewrite to the top of the stack:

config.middleware.insert 0, 'Rack::Rewrite' {}

Redirection codes

All redirect status codes from the HTTP spec are supported:

301 moved permanently
302 found
303 see other
307 temporary redirect

These translate to the following methods inside the Rack::Rewrite block:

r301                '/wiki/John_Trupiano', '/john'
moved_permanently   '/wiki/John_Trupiano', '/john'
p                   '/wiki/John_Trupiano', '/john'    # shortcut alias

r302                '/wiki/John_Trupiano', '/john'
found               '/wiki/John_Trupiano', '/john'

r303                '/wiki/John_Trupiano', '/john'
see_other           '/wiki/John_Trupiano', '/john'

r307                '/wiki/John_Trupiano', '/john'
temporary_redirect  '/wiki/John_Trupiano', '/john'
t                   '/wiki/John_Trupiano', '/john'    # shortcut alias

The 303 and 307 codes were added to the HTTP spec to make unambiguously clear what clients should do with the request method. 303 means that the new request should always be made via GET. 307 means that the new request should use the same method as the original request. Status code 302 was left as it is, since it was already in use by the time these issues came to light. In practice it behaves the same as 303.

Use Cases

Rebuild of existing site in a new technology

It's very common for sites built in older technologies to be rebuilt with the latest and greatest. Let's consider a site that has already established quite a bit of "google juice." When we launch the new site, we don't want to lose that hard-earned reputation. By writing rewrite rules that issue 301's for old URL's, we can "transfer" that google ranking to the new site. An example rule might look like:

r301 '/contact-us.php', '/contact-us'
r301 '/wiki/John_Trupiano', '/john'

Retiring old routes

As a web application evolves you will undoubtedly reach a point where you need to change the name of something (a model, e.g.). This name change will typically require a similar change to your routing. The danger here is that any URL's previously generated (in a transactional email for instance) will have the URL hard-coded. In order for your rails app to continue to serve this URL, you'll need to add an extra entry to your routes file. Alternatively, you could use rack-rewrite to redirect or pass through requests to these routes and keep your routes.rb clean.

rewrite %r{/features(.*)}, '/facial_features$1'

CNAME alternative

In the event that you do not control your DNS, you can leverage Rack::Rewrite to redirect to a canonical domain. In the following rule we utilize the $& substitution operator to capture the entire request URI.

r301 %r{.*}, 'http://mynewdomain.com$&', :if => Proc.new {|rack_env|
  rack_env['SERVER_NAME'] != 'mynewdomain.com'
}

Site Maintenance

Most capistrano users will be familiar with the following Apache rewrite rules:

RewriteCond %{REQUEST_URI} !\.(css|jpg|png)$
RewriteCond %{DOCUMENT_ROOT}/system/maintenance.html -f
RewriteCond %{SCRIPT_FILENAME} !maintenance.html
RewriteRule ^.*$ /system/maintenance.html [L]

This rewrite rule says to render a maintenance page for all non-asset requests if the maintenance file exists. In capistrano, you can quickly upload a maintenance file using:

cap deploy:web:disable REASON=upgrade UNTIL=12:30PM

We can replace the mod_rewrite rules with the following Rack::Rewrite rule:

maintenance_file = File.join(RAILS_ROOT, 'public', 'system', 'maintenance.html')
send_file /.*/, maintenance_file, :if => Proc.new { |rack_env|
  File.exists?(maintenance_file) && rack_env['PATH_INFO'] !~ /\.(css|jpg|png)/
}

If you're running Ruby 1.9, this rule is simplified:

maintenance_file = File.join(RAILS_ROOT, 'public', 'system', 'maintenance.html')
send_file /(.*)$(?<!css|png|jpg)/, maintenance_file, :if => Proc.new { |rack_env|
  File.exists?(maintenance_file)
}

For those using the oniguruma gem with their ruby 1.8 installation, you can get away with:

maintenance_file = File.join(RAILS_ROOT, 'public', 'system', 'maintenance.html')
send_file Oniguruma::ORegexp.new("(.*)$(?<!css|png|jpg)"), maintenance_file, :if => Proc.new { |rack_env|
  File.exists?(maintenance_file)
}

Rewrite Rules

:rewrite

Calls to #rewrite will simply update the PATH_INFO, QUERY_STRING and REQUEST_URI HTTP header values and pass the request onto the next chain in the Rack stack. The URL that a user's browser will show will not be changed. See these examples:

rewrite '/wiki/John_Trupiano', '/john'   # [1]
rewrite %r{/wiki/(\w+)_\w+}, '/$1'       # [2]

For [1], the user's browser will continue to display /wiki/John_Trupiano, but the actual HTTP header values for PATH_INFO and REQUEST_URI in the request will be changed to /john for subsequent nodes in the Rack stack. Rails reads these headers to determine which routes will match.

Rule [2] showcases the use of regular expressions and substitutions. [2] is a generalized version of [1] that will match any /wiki/FirstName_LastName URL's and rewrite them as the first name only. This is an actual catch-all rule we applied when we rebuilt our website in September 2009 ( http://www.smartlogicsolutions.com ).

:r301, :r302, :r303, :r307

Calls to #r301 and #r302 have the same signature as #rewrite. The difference, however, is that these actually short-circuit the rack stack and send back their respective status codes. See these examples:

r301 '/wiki/John_Trupiano', '/john'                # [1]
r301 %r{/wiki/(.*)}, 'http://www.google.com/?q=$1' # [2]

Recall that rules are interpreted from top to bottom. So you can install "default" rewrite rules if you like. [2] is a sample default rule that will redirect all other requests to the wiki to a google search.

:send_file, :x_send_file, :send_data

Calls to #send_file and #x_send_file and #send_data also have the same signature as #rewrite. If the rule matches, the 'to' parameter is interpreted as a path to a file to be rendered instead of passing the application call up the rack stack.

send_file /*/, 'public/spammers.htm', :if => Proc.new { |rack_env|
  rack_env['HTTP_REFERER'] =~ 'spammers.com'
}
x_send_file /^blog\/.*/, 'public/blog_offline.htm', :if => Proc.new { |rack_env|
  File.exists?('public/blog_offline.htm')
}
send_data /^blog\/.*/, 'public/blog_offline.png', :if => Proc.new { |rack_env|
  File.exists?('public/blog_offline.htm')
}

Options Parameter

Each rewrite rule takes an optional options parameter. The following options are supported.

:host

Using the :host option you can match requests to a specific hostname.

r301 "/features", "/facial_features", :host => "facerecognizer.com"

This rule will only match when the hostname is "facerecognizer.com".

The :host option accepts a symbol, string, or regexp.

:headers

Using the :headers option you can set custom response headers e.g. for HTTP caching instructions.

r301 "/features", "/facial_features", :headers => {'Cache-Control' => 'no-cache'}

Please be aware that the :headers value above is evaluated only once at app boot and shared amongst all matching requests.

Use a Proc as the :headers option if you wish to determine the additional headers at request-time. For example:

# We want the Expires value to always be 1 year in the future from now. If
# we didn't use a Proc here, then the Expires value would be set just once
# at app startup. The Proc will be evaluated for each matching request.
send_file /^.+\.(?:ico|jpg|jpeg|png|gif|)$/,
          'public/$&',
          :headers => lambda { { 'Expires' => 1.year.from_now.httpdate } }

:scheme

Using the :scheme option you can restrict the matching of a rule by the protocol of a given request.

# Redirect all http traffic to https
r301 %r{.*}, 'https://www.example.tld$&', :scheme => 'http'

The :scheme option accepts a symbol, string, or regexp.

:method

Using the :method option you can restrict the matching of a rule by the HTTP method of a given request.

# redirect GET's one way
r301 "/players", "/current_players", :method => :get

# and redirect POST's another way
r302 "/players", "/no_longer_available.html?message=No&longer&supported", :method => :post

The :method option accepts a symbol, string, or regexp.

:if

Using the :if option you can define arbitrary rule guards. Guards are any object responding to #call that return true or false indicating whether the rule matches. The following example demonstrates how the presence of a maintenance page on the filesystem can be utilized to take your site(s) offline.

maintenance_file = File.join(RAILS_ROOT, 'public', 'system', 'maintenance.html')
x_send_file /.*/, maintenance_file, :if => Proc.new { |rack_env|
  File.exists?(maintenance_file)
}

:not

Using the :not option you can negatively match against the path. This can be useful when writing a regular expression match is difficult.

rewrite %r{^\/features}, '/facial_features', :not => '/features'

This will not match the relative URL /features but would match /features.xml.

Tips

Keeping your querystring

When rewriting a URL, you may want to keep your querystring intact (for example if you're tracking traffic sources). You will need to include a capture group and substitution pattern in your rewrite rule to achieve this.

rewrite %r{/wiki/John_Trupiano(\?.*)?}, '/john$1'

This rule will store the querystring in a capture group (via (?.*) ) and will substitute the querystring back into the rewritten URL (via $1).

Arbitrary Rewriting

All rules support passing a Proc as the first or second argument allowing you to perform arbitrary rewrites. The following rule will rewrite all requests received between 12AM and 8AM to an unavailable page.

  rewrite %r{(.*)}, lambda { |match, rack_env|
    Time.now.hour < 8 ? "/unavailable.html" : match[1]
  }

This rule will redirect all requests paths starting with a current date string to /today.html

  r301 lambda { "/#{Time.current.strftime(%m%d%Y)}.html" }, '/today.html'

##Alternative loaders

rack-rewrite can also be driven by external loaders. Bundled with this library is a loader for YAML files.

config.middleware.insert_before(Rack::Lock, Rack::Rewrite,
    :klass => Rack::Rewrite::YamlRuleSet,
    :options => {:file_name => @file_name})

Using syntax like

-
    method: r301
    from: !ruby/regexp '/(.*)/print'
    to : '$1/printer_friendly'
    options :
        host : 'example.com'

Any class can be used here as long as:

the class take an options hash
#rules returns an array of Rack::Rewrite::Rule instances

Contribute

rack-rewrite is maintained by @travisjeffery.

Here's the most direct way to get your work merged into the project.

Fork the project
Clone down your fork
Create a feature branch
Hack away and add tests, not necessarily in that order
Make sure everything still passes by running tests
If necessary, rebase your commits into logical chunks without errors
Push the branch up to your fork
Send a pull request for your branch

bosko/rack-rewrite