ruby - How to make Nokogiri transparently return un/encoded Html entities untouched?

Question

How can I use Nokogiri with having html entities (like German umlauts) untouched?

I.e.:

# this is fine
node = Nokogiri::HTML.fragment('<p>&ouml;</p>')
node.to_s # => '<p>&ouml;</p>'

# this is not
node = Nokogiri::HTML.fragment('<p>?</p>')
node.to_s # => '<p>&ouml;</p>'

# this is what I need
node = Nokogiri::HTML.fragment('<p>?</p>')
node.to_s # => '<p>?</p>'

I've tried to mess with both PARSE_OPTIONS and :save_with options but could not come up with a way to have Nokogiri just transparently behave like above.

Any pointers?

Answer 1

Ok, my question has been answered by Aaron via twitter/gist:

require 'rubygems'
require 'nokogiri'

doc = Nokogiri::HTML::Document.new
doc.encoding = 'UTF-8'

# We added a contextual fragment method for the 1.4.2 release. This *might*
# work in 1.4.1. If you want to mess with 1.4.2, build from my github, or
# grab one of our nightly builds:
#
# $ sudo gem install nokogiri -s http://tenderlovemaking.com/
#
# Also, libxml2 had a bug with encoding when handling UTF-8 fragments, so I
# suggest you also upgrade to libxml2 2.7.7.
#
# Hope that helps!
puts doc.fragment('<p>?</p>')