- https://github.com/flavorjones/loofah
- Docs: http://rubydoc.info/github/flavorjones/loofah/main/frames
- Mailing list: loofah-talk@googlegroups.com
Loofah is a general library for manipulating and transforming HTML/XML documents and fragments, built on top of Nokogiri.
Loofah also includes some HTML sanitizers based on html5lib
's safelist, which are a specific application of the general transformation functionality.
Active Record extensions for HTML sanitization are available in the loofah-activerecord
gem.
- Easily write custom transformations for HTML and XML
- Common HTML sanitizing transformations 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.
- Other common HTML transformations are built-in:
- Add the nofollow attribute to all hyperlinks.
- Add the target=_blank attribute to all hyperlinks.
- Remove unprintable characters from text nodes.
- Format markup as plain text, with (or without) sensible whitespace handling around block elements.
- Replace Rails's
strip_tags
andsanitize
view helper methods.
Loofah is both:
- a general framework for transforming XML, XHTML, and HTML documents
- a specific toolkit for HTML sanitization
Loofah tries to make it easy to write your own custom scrubbers for whatever document transformation you need. You don't like the built-in scrubbers? Build your own, like a boss.
Another Ruby library that provides HTML sanitization is rgrove/sanitize
, another library built on top of Nokogiri, which provides a bit more flexibility on the tags and attributes being scrubbed.
You may also want to look at rails/rails-html-sanitizer
which is built on top of Loofah and provides some useful extensions and additional flexibility in the HTML sanitization.
Loofah wraps Nokogiri in a loving embrace. Nokogiri is a stable, well-maintained parser for XML, HTML4, and HTML5.
Loofah implements the following classes:
Loofah::HTML5::Document
Loofah::HTML5::DocumentFragment
Loofah::HTML4::Document
(aliased asLoofah::HTML::Document
for now)Loofah::HTML4::DocumentFragment
(aliased asLoofah::HTML::DocumentFragment
for now)Loofah::XML::Document
Loofah::XML::DocumentFragment
These document and fragment classes are subclasses of the similarly-named Nokogiri classes Nokogiri::HTML5::Document
et al.
Loofah also implements Loofah::Scrubber
, which represents the document transformation, 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.html5_fragment
or Loofah.html4_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.
HTML documents should be parsed with Loofah.html5_document
or Loofah.html4_document
. The result will have a DOCTYPE declaration, along with html
, head
and body
tags.
XML fragments should be parsed with Loofah.xml_fragment
. The result won't have a DOCTYPE declaration, and multiple root nodes are allowed.
XML documents should be parsed with Loofah.xml_document
. The result will have a DOCTYPE declaration and a single root node.
⚠ HTML5 functionality is not available on JRuby, or with versions of Nokogiri < 1.14.0
.
Currently, Loofah's methods Loofah.document
and Loofah.fragment
are aliases to .html4_document
and .html4_fragment
, which use Nokogiri's HTML4 parser. (Similarly, Loofah::HTML::Document
and Loofah::HTML::DocumentFragment
are aliased to Loofah::HTML4::Document
and Loofah::HTML4::DocumentFragment
.)
Please note that in a future version of Loofah, these methods and classes may switch to using Nokogiri's HTML5 parser and classes on platforms that support it [1].
We strongly recommend that you explicitly use .html5_document
or .html5_fragment
unless you know of a compelling reason not to. If you are sure that you need to use the HTML4 parser, you should explicitly call .html4_document
or .html4_fragment
to avoid breakage in a future version.
[1]: [feature request] HTML5 parser for JRuby implementation · Issue #2227 · sparklemotion/nokogiri
These classes are subclasses of Nokogiri::HTML5::Document
and Nokogiri::HTML5::DocumentFragment
.
The module methods Loofah.html5_document
and Loofah.html5_fragment
will parse either an HTML document and an HTML fragment, respectively.
Loofah.html5_document(unsafe_html).is_a?(Nokogiri::HTML5::Document) # => true
Loofah.html5_fragment(unsafe_html).is_a?(Nokogiri::HTML5::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.html5_fragment(unsafe_html).scrub!(:prune)
doc.to_s # => "ohai! <div>div is safe</div> "
and text
to return plain text:
doc.text # => "ohai! div is safe "
Also, to_text
is available, which does the right thing with whitespace around block-level and line break elements.
doc = Loofah.html5_fragment("<h1>Title</h1><div>Content<br>Next line</div>")
doc.text # => "TitleContentNext line" # probably not what you want
doc.to_text # => "\nTitle\n\nContent\nNext line\n" # better
These classes are subclasses of Nokogiri::HTML4::Document
and Nokogiri::HTML4::DocumentFragment
.
The module methods Loofah.html4_document
and Loofah.html4_fragment
will parse either an HTML document and an HTML fragment, respectively.
Loofah.html4_document(unsafe_html).is_a?(Nokogiri::HTML4::Document) # => true
Loofah.html4_fragment(unsafe_html).is_a?(Nokogiri::HTML4::DocumentFragment) # => true
These classes are subclasses of Nokogiri::XML::Document
and Nokogiri::XML::DocumentFragment
.
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's Node
and NodeSet
classes 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.html5_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 safelist algorithm:
doc = Loofah.html5_document(input)
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
doc.scrub!(:noopener) # adds rel="noopener" attribute to links
doc.scrub!(:noreferrer) # adds rel="noreferrer" attribute to links
doc.scrub!(:unprintable) # removes unprintable characters from text nodes
doc.scrub!(:targetblank) # adds target="_blank" attribute to links
See Loofah::Scrubbers
for more details and example usage.
You can chain scrubbers:
Loofah.html5_fragment("<span>hello</span> <script>alert('OHAI')</script>") \
.scrub!(:prune) \
.scrub!(span2div).to_s
# => "<div>hello</div> "
The class methods Loofah.scrub_html5_fragment
and Loofah.scrub_html5_document
(and the corresponding HTML4 methods) are shorthand.
These methods:
Loofah.scrub_html5_fragment(unsafe_html, :prune)
Loofah.scrub_html5_document(unsafe_html, :prune)
Loofah.scrub_html4_fragment(unsafe_html, :prune)
Loofah.scrub_html4_document(unsafe_html, :prune)
Loofah.scrub_xml_fragment(bad_xml, custom_scrubber)
Loofah.scrub_xml_document(bad_xml, custom_scrubber)
do the same thing as (and arguably semantically clearer than):
Loofah.html5_fragment(unsafe_html).scrub!(:prune)
Loofah.html5_document(unsafe_html).scrub!(:prune)
Loofah.html4_fragment(unsafe_html).scrub!(:prune)
Loofah.html4_document(unsafe_html).scrub!(:prune)
Loofah.xml_fragment(bad_xml).scrub!(custom_scrubber)
Loofah.xml_document(bad_xml).scrub!(custom_scrubber)
Loofah has two "view helpers": Loofah::Helpers.sanitize
and Loofah::Helpers.strip_tags
, both of which are drop-in replacements for the Rails Action View helpers of the same name.
These are not required automatically. You must require loofah/helpers
to use them.
- Nokogiri >= 1.5.9
Unsurprisingly:
gem install loofah
Requirements:
- Ruby >= 2.5
The bug tracker is available here:
And the mailing list is on Google Groups:
Consider subscribing to Tidelift which provides license assurances and timely security notifications for your open source dependencies, including Loofah. Tidelift subscriptions also help the Loofah maintainers fund our automated testing which in turn allows us to ship releases, bugfixes, and security updates more often.
See SECURITY.md
for vulnerability reporting details.
- loofah-activerecord: https://github.com/flavorjones/loofah-activerecord
- Nokogiri: http://nokogiri.org
- libxml2: http://xmlsoft.org
- html5lib: https://github.com/html5lib/
- Mike Dalessio (@flavorjones)
- Bryan Helmkamp
Featuring code contributed by:
- @flavorjones
- @brynary
- @olleolleolle
- @JuanitoFatas
- @kaspth
- @tenderlove
- @ktdreyer
- @orien
- @asok
- @junaruga
- @MothOnMars
- @nick-desteffen
- @NikoRoberts
- @trans
- @andreynering
- @aried3r
- @baopham
- @batter
- @brendon
- @cjba7
- @christiankisssner
- @dacort
- @danfstucky
- @david-a-wheeler
- @dharamgollapudi
- @georgeclaghorn
- @gogainda
- @jaredbeck
- @ThatHurleyGuy
- @jstorimer
- @jbarnette
- @queso
- @technicalpickles
- @kyoshidajp
- @kristianfreeman
- @louim
- @mrpasquini
- @olivierlacan
- @pauldix
- @sampokuokkanen
- @stefannibrasil
- @tastycode
- @vipulnsward
- @joncalhoun
- @ahorek
- @rmacklin
- @y-yagi
And a big shout-out to Corey Innis for the name, and feedback on the API.
The following people have generously funded Loofah with financial sponsorship:
- Bill Harding
- Sentry @getsentry
This library was once named "Dryopteris", which was a very bad name that nobody could spell properly.
Distributed under the MIT License. See MIT-LICENSE.txt
for details.