Which behaves correctly, Hpricot or Nokogiri?

[Just Another Victim of the Ambient Morality]

I've been considering switching to Nokogiri instead of Hpricot, mostly
'cause Mechanize has switched. However, the two actually behave quite
differently. The Nokogiri objects don't simulate standard container
behavior nearly as well as Hpricot. I also noticed that this:


require 'nokogiri'
require 'hpricot'

xml = '<first look="Big &amp; small...">content</first>'

doc = Nokogiri::XML(xml)
puts doc.search('first')[0].attributes['look']
doc = Hpricot(xml)
puts doc.search('first')[0].attributes['look']


...produces this output:


Big &amp; small...
Big & smal...


I don't know which output is the correct one. Does anyone know what's
going on here?
Thank you...

 

[Rados³aw Bu³at]

xml =3D '<first look=3D"Big & small...">content</first>'

It's not valid xml. It should be "Big &amp; small..."
I guess that for non-valid xml there is now "valid" behavior. Ask
hpricot and nokogiri developers what happen when xml is not valid
(they try to fix it or smth?)

Big & small...
Big & smal...

Strange. I get:
Big small...
Big & small...

The difference is about '&' which is not valid in xml (&amp; should be
used instead).

--=20
Pozdrawiam

Rados=B3aw Bu=B3at
http://radarek.jogger.pl - m=F3j blog

 

[Just Another Victim of the Ambient Morality]

It's not valid xml. It should be "Big &amp; small..."
I guess that for non-valid xml there is now "valid" behavior. Ask
hpricot and nokogiri developers what happen when xml is not valid
(they try to fix it or smth?)

Actually, "Big &amp; small" is what I wrote in the example. The second
output is erroneously missing an "l" but I think that's understood...
I'm wondering if anyone knows what the correct behaviour is supposed to
be...

Oh, I get it. Maybe my use of & amp ; was translated in whatever client
you're using?

 

[Phlip]

doc = Nokogiri::XML(xml)
puts doc.search('first')[0].attributes['look']
doc = Hpricot(xml)
puts doc.search('first')[0].attributes['look']

...produces this output:


Big &amp; small...
Big & smal...

I don't know which output is the correct one. Does anyone know what's
going on here?

The second one is correct, because &amp; is an encoding, and an XML tool should
use & outside its interface and &amp; inside its interface.

Now try these XPaths in Hpricot and NokoGiri - which combinations find the node?

first[ @look = 'Big &amp; small...' ]
first[ @look = 'Big & small...' ]