Loofah is a general library for manipulating HTML/XML documents and fragments. It’s built on top of Nokogiri and libxml2, so it’s fast and has a nice API.
Loofah excels at HTML sanitization (XSS prevention). It includes some nice HTML sanitizers, which are based on HTML5lib’s whitelist, so it most likely won’t make your codes less secure. (These statements have not been evaluated by Netexperts.)
-
Easily write custom scrubbers for HTML/XML leveraging the sweetness of Nokogiri (and HTML5lib’s whitelists).
-
Common HTML sanitizing tasks are built-in:
-
Strip unsafe tags, leaving behind only the inner text.
-
Prune unsafe tags and their subtrees, removing all traces that they ever existed.
-
Escape unsafe tags and their subtrees, leaving behind lots of
<
and>
entities. -
Whitewash the markup, removing all attributes and namespaced nodes.
-
-
Common HTML transformation tasks are built-in:
-
Add the nofollow attribute to all hyperlinks.
-
-
Format markup as plain text.
-
Replace Rails’s
strip_tags
andsanitize
helper methods. -
Two ActiveRecord extensions:
-
Loofah::XssFoliate, an XssTerminate drop-in replacement, is an opt-out sanitizer. By default all models and attributes are sanitized.
-
Loofah::ActiveRecordExtension is an opt-in sanitizer. You must explicitly declare attributes to be sanitized.
-
Loofah is one of two known Ruby XSS/sanitization solutions that guarantees well-formed and valid markup (the other is Sanitize, which also uses Nokogiri).
Loofah works fine on XML, XHTML and HTML documents.
Also, it’s pretty fast. Here is a benchmark comparing Loofah to other commonly-used libraries (ActionView, Sanitize, HTML5lib and HTMLfilter):
Lastly, Loofah is extensible. It’s super-easy to write your own custom scrubbers for whatever document manipulation you need. You don’t like the built-in scrubbers? Build your own, like a boss.
Loofah wraps Nokogiri in a loving embrace. Nokogiri is an excellent HTML/XML parser. If you don’t know how Nokogiri works, you might want to pause for a moment and go check it out. I’ll wait.
Loofah presents the following classes:
-
Loofah::HTML::Document and Loofah::HTML::DocumentFragment
-
Loofah::XML::Document and Loofah::XML::DocumentFragment
-
Loofah::Scrubber
The documents and fragments are subclasses of the similar Nokogiri classes.
The Scrubber represents the document manipulation, either by wrapping a block,
span2div = Loofah::Scrubber.new do |node| node.name = "div" if node.name == "span" end
or by implementing a method.
Generally speaking, unless you expect to have a DOCTYPE and a single root node, you don’t have a document, you have a fragment. For HTML, another rule of thumb is that documents have <html> and <body> tags, and fragments usually do not.
HTML fragments should be parsed with Loofah.fragment. The result won’t be wrapped in html
or body
tags, won’t have a DOCTYPE declaration, head
elements will be silently ignored, and multiple root nodes are allowed.
XML fragments should be parsed with Loofah.xml_fragment. The result won’t have a DOCTYPE declaration, and multiple root nodes are allowed.
HTML documents should be parsed with Loofah.document. The result will have a DOCTYPE declaration, along with html
, head
and body
tags.
XML documents should be parsed with Loofah.xml_document. The result will have a DOCTYPE declaration and a single root node.
These classes are subclasses of Nokogiri::HTML::Document and Nokogiri::HTML::DocumentFragment, so you get all the markup fixer-uppery and API goodness of Nokogiri.
The module methods Loofah.document and Loofah.fragment will parse an HTML document and an HTML fragment, respectively.
Loofah.document(unsafe_html).is_a?(Nokogiri::HTML::Document) # => true Loofah.fragment(unsafe_html).is_a?(Nokogiri::HTML::DocumentFragment) # => true
Loofah injects a scrub!
method, which takes either a symbol (for built-in scrubbers) or a Loofah::Scrubber object (for custom scrubbers), and modifies the document in-place.
Loofah overrides to_s
to return HTML:
unsafe_html = "ohai! <div>div is safe</div> <script>but script is not</script>" doc = Loofah.fragment(unsafe_html).scrub!(:strip) doc.to_s # => "ohai! <div>div is safe</div> "
and text
to return plain text:
doc.text # => "ohai! div is safe "
These classes are subclasses of Nokogiri::XML::Document and Nokogiri::XML::DocumentFragment, so you get all the markup fixer-uppery and API goodness of Nokogiri.
The module methods Loofah.xml_document and Loofah.xml_fragment will parse an XML document and an XML fragment, respectively.
Loofah.xml_document(bad_xml).is_a?(Nokogiri::XML::Document) # => true Loofah.xml_fragment(bad_xml).is_a?(Nokogiri::XML::DocumentFragment) # => true
Nokogiri::XML::Node and Nokogiri::XML::NodeSet also get a scrub!
method, which makes it easy to scrub subtrees.
The following code will apply the employee_scrubber
only to the employee
nodes (and their subtrees) in the document:
Loofah.xml_document(bad_xml).xpath("//employee").scrub!(employee_scrubber)
And this code will only scrub the first employee
node and its subtree:
Loofah.xml_document(bad_xml).at_xpath("//employee").scrub!(employee_scrubber)
A Scrubber wraps up a block (or method) that is run on a document node:
# change all <span> tags to <div> tags span2div = Loofah::Scrubber.new do |node| node.name = "div" if node.name == "span" end
This can then be run on a document:
Loofah.fragment("<span>foo</span><p>bar</p>").scrub!(span2div).to_s # => "<div>foo</div><p>bar</p>"
Scrubbers can be run on a document in either a top-down traversal (the default) or bottom-up. Top-down scrubbers can optionally return Scrubber::STOP to terminate the traversal of a subtree. Read below and in the Loofah::Scrubber class for more detailed usage.
Here’s an XML example:
# remove all <employee> tags that have a "deceased" attribute set to true bring_out_your_dead = Loofah::Scrubber.new do |node| if node.name == "employee" and node["deceased"] == "true" node.remove Loofah::Scrubber::STOP # don't bother with the rest of the subtree end end Loofah.xml_document(File.read('plague.xml')).scrub!(bring_out_your_dead)
Loofah comes with a set of sanitizing scrubbers that use HTML5lib’s whitelist algorithm:
doc.scrub!(:strip) # replaces unknown/unsafe tags with their inner text doc.scrub!(:prune) # removes unknown/unsafe tags and their children doc.scrub!(:escape) # escapes unknown/unsafe tags, like this: <script> doc.scrub!(:whitewash) # removes unknown/unsafe/namespaced tags and their children, # and strips all node attributes
Loofah also comes with some common transformation tasks:
doc.scrub!(:nofollow) # adds rel="nofollow" attribute to links
See Loofah::Scrubbers for more details and example usage.
You can chain scrubbers:
Loofah.fragment("<span>hello</span> <script>alert('OHAI')</script>") \ .scrub!(:prune) \ .scrub!(span2div).to_s # => "<div>hello</div> "
The class methods Loofah.scrub_fragment and Loofah.scrub_document are shorthand.
Loofah.scrub_fragment(unsafe_html, :prune) Loofah.scrub_document(unsafe_html, :prune) Loofah.scrub_xml_fragment(bad_xml, custom_scrubber) Loofah.scrub_xml_document(bad_xml, custom_scrubber)
are the same thing as (and arguably semantically clearer than):
Loofah.fragment(unsafe_html).scrub!(:prune) Loofah.document(unsafe_html).scrub!(:prune) Loofah.xml_fragment(bad_xml).scrub!(custom_scrubber) Loofah.xml_document(bad_xml).scrub!(custom_scrubber)
See Loofah::ActiveRecordExtension for full documentation. The methods mixed into ActiveRecord are:
-
Loofah::ActiveRecordExtension.html_document
-
Loofah::ActiveRecordExtension.html_fragment
which are used to declare how specific string and text attributes should be scrubbed at before_validation
.
# app/model/post.rb class Post < ActiveRecord::Base html_fragment :body, :scrub => :prune # scrubs 'body' at before_validation end
See Loofah::XssFoliate::ClassMethods for more documentation. The methods mixed into ActiveRecord are:
-
Loofah::XssFoliate::ClassMethods.xss_foliate
-
Loofah::XssFoliate::ClassMethods.xss_foliated?
which are used to declare how specific string and text attributes should be scrubbed at before_validation
.
Attributes are stripped by default, unless another scrubber is specified or the attribute is present in an :except
clause.
Loofah has two “view helpers”: Loofah::Helpers.sanitize and Loofah::Helpers.strip_tags, both of which are drop-in replacements for the ActionView helpers of the same name.
-
Nokogiri >= 1.3.3
-
Rails 2.3, 2.2, 2.1, 2.0 or 1.2 (if you’re using the ActiveRecord extensions)
Unsurprisingly:
-
gem install loofah
The bug tracker is available here:
And the mailing list is on librelist:
-
loofah@librelist.com / librelist.com
And the IRC channel is #loofah on freenode.
-
Nokogiri: nokogiri.org
-
libxml2: xmlsoft.org
-
html5lib: code.google.com/p/html5lib
-
XssTerminate: github.com/look/xss_terminate/tree/master
-
Bryan Helmkamp
Featuring code contributed by:
-
Aaron Patterson
-
John Barnette
-
Josh Owens
-
Paul Dix
-
Josh Nichols
-
Luke Melia
And a big shout-out to Corey Innis for the name, and feedback on the API.
The following people have generously donated via the Pledgie badge on the Loofah github page:
-
Bill Harding
This library was formerly known as Dryopteris, which was a very bad name that nobody could spell properly.
The MIT License
Copyright © 2009 Mike Dalessio, Bryan Helmkamp
Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files (the “Software”), to deal in the Software without restriction, including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do so, subject to the following conditions:
The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software.
THE SOFTWARE IS PROVIDED “AS IS”, WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.