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.
gem 'rack-rewrite', '~> 1.2.1' 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
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
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.
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'
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'
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' }
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) }
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 ( www.smartlogicsolutions.com ).
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 '/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.
Calls to #send_file and #x_send_file 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')
}
Each rewrite rule takes an optional options parameter. The following options are supported.
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”
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'}
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
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) }
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.
When rewriting a URL, you may want to keep your querystring in tact (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).
All rules support passing a Proc as the 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] }
Copyright © 2009-2011 John Trupiano. See LICENSE for details.