SanitizeAttributes

This is a simple plugin for ActiveRecord models to define sanitizable attributes. When an object is saved, those attributes will be run through whatever filter you’ve defined. You can define a default filter for all sanitizations.

Sanitization only happens for non-nil attributes. (Because a nil attribute may be valid for your model, and the sanitzers should only have to worry about working with strings.)

This plugin was created to implement anti-XSS validation. My current gem of choice is Sanitize: github.com/rgrove/sanitize/tree/master

Example

 # probably in environment.rb
 SanitizeAttributes.default_sanitization_method = lambda do |text|
   text.gsub(/[^\w\s]/, "") # very simple, very limited
 end

 # app/models/bookmark.rb
 class Bookmark
   sanitize_attribute :sitename
 end

 # app/models/article.rb
 class Article
   default_sanitization_method_for_class = lambda do |text|
     text.gsub(/[^\w\s'".,?!]/, """) # more reasonable, for titles and such
   end

   sanitize_attributes :title, :author

   sanitize_attributes :body do |body_text|
     # This needs to be safe, renderable HTML, so let's use a real sanization tool
     # I recommend: http://github.com/rgrove/sanitize/tree/master
     Sanitize.clean(body_text)
   end
 end

 # in action...
 b = Bookmark.create(:sitename => "boston.rb!!!", :url => "http://http://bostonrb.org/")
 b.sitename # => "bostonrb"
 a = Article.create(:title => "<b>Hello</b>!", :body => "Please remove the <script>script tags</script>!")
 a.title # => "Hello!"
 a.body  # => "Please remove the script tags!"

Future Work

Things to work on in the future:

• allowing strings/symbols for sanitization methods, not just blocks

   Nacho.default_sanitization_method_for_class = :microwave # uses Nacho.microwave
   Nacho.default_sanitization_method_for_class = "Sanitize.clean"
  

• add validation helpers, if you want to flag problematic text rather than cleaning it.

  class Foo

    validate_sanitzed :value
  

  end Foo.new(:value => "abc").valid? #=> false if a sanitized copy of #value is different than the original

• better functionality for subclasses. Currently, they will share sanitized attributes and sanitization methods across subclasses and the base class.

Etc

© 2009 Dev Purkayastha, released under the MIT license