Sanitize broken in JRuby

[aselder]

Case #1 - MRI (1.9.2-p290, 1.9.3-p194)

Nokogiri::VERSION   # => "1.5.0"
Sanitize.clean("a<a> b </a>") # => "a b " 

Case #2 (JRuby-1.6.7.2)

Nokogiri::VERSION   # => "1.5.3"
Sanitize.clean("a<a> b </a>") # => "a" 

[aselder]

A little more info. Adding a "puts node" before line 151 of santize.rb

MRI

a
 b 
<a></a>
a b 

JRuby

a
 b 
<a> b </a>
a

[rgrove]

The JRuby version of Nokogiri behaves very differently from the MRI version. There's not much Sanitize can do about this short of switching to a custom HTML parser. For this reason, I don't consider JRuby an officially supported platform. Sorry.

[aselder]

Looking like an inconsistency in Nokogiri between JRuby and MRI

[aselder]

FYI, I tracked down this to an issue in Nokogiri: https://github.com/tenderlove/nokogiri/issues/691

[LeifWarner]

Sanitize.clean("a<a> b </a>") seems to work for me on JRuby.

[aselder]

Yes, the underlying nokogiri bug was fixed in 1.5.5