"Document already has a root node" while using Builder

[yokolet]

This bug was found while investigating issue #641

require 'rubygems'
require 'nokogiri'

builder = Nokogiri::XML::Builder.new do |xml|
  0.upto(81900) do
    xml.text "test"
  end
end

raised:

RuntimeError: Document already has a root node
   add_child at /Users/yoko/Documents/workspace/Ranunculus/lib/nokogiri/xml/document.rb:212
     parent= at /Users/yoko/Documents/workspace/Ranunculus/lib/nokogiri/xml/node.rb:549
      insert at /Users/yoko/Documents/workspace/Ranunculus/lib/nokogiri/xml/builder.rb:371
        text at /Users/yoko/Documents/workspace/Ranunculus/lib/nokogiri/xml/builder.rb:302
    __file__ at ../Canna/src/snippet/issue641.rb:6
        upto at org/jruby/RubyInteger.java:139
    __file__ at ../Canna/src/snippet/issue641.rb:5
  initialize at /Users/yoko/Documents/workspace/Ranunculus/lib/nokogiri/xml/builder.rb:293
      (root) at ../Canna/src/snippet/issue641.rb:4

[yokolet]

By investigation, JRuby might be wrong with block processing.

I changed the test as in below and added some puts statement in node.rb/builder.rb. This is becasuse there's no line to be a good breakpoint in Java code.

require 'rubygems'
require 'nokogiri'

builder = Nokogiri::XML::Builder.new do |xml|
  0.upto(3) do |i|
    xml.text "test" + i.to_s
  end
end

libxml version create:

<?xml version="1.0"?>
test0test1test2

while JRuby created:

<?xml version="1.0"?>
<text>test0</text>

then raised an error.

[yokolet]

Oh, well, comment helps.

I found this in XmlNode.java

        /*
         * This is a bit of a hack.  C-Nokogiri allows adding a bare
         * text node as the root element.  Java (and XML spec?) does
         * not.  So we wrap the text node in an element.
         */

This case is the one. So, post processing is needed.

[msavy]

I get this issue when using standard MRI/YARV. Any other random document I read in responds normally to #root (and not as if you're trying to assign), just seems to be builder as you've identified.

Code

require 'rubygems'
require 'nokogiri'

document = Nokogiri::XML::Builder.new{ something("other") { blah("fleg") } }

puts document.to_xml
document.root

Error

<?xml version="1.0"?>
<something>other<blah>fleg</blah></something>
/Users/msavy/.rvm/gems/ruby-1.9.3-p125/gems/nokogiri-1.5.0/lib/nokogiri/xml/document.rb:198:in `add_child': Document already has a root node (RuntimeError)
from /Users/msavy/.rvm/gems/ruby-1.9.3-p125/gems/nokogiri-1.5.0/lib/nokogiri/xml/node.rb:539:in `parent='
from /Users/msavy/.rvm/gems/ruby-1.9.3-p125/gems/nokogiri-1.5.0/lib/nokogiri/xml/builder.rb:365:in `insert'
from /Users/msavy/.rvm/gems/ruby-1.9.3-p125/gems/nokogiri-1.5.0/lib/nokogiri/xml/builder.rb:357:in `method_missing'
from example.rb:7:in `<main>'

Nokogiri (1.5.0)

---
warnings: []
nokogiri: 1.5.0
ruby:
  version: 1.9.3
  platform: x86_64-darwin10.8.0
  description: ruby 1.9.3p125 (2012-02-16 revision 34643) [x86_64-darwin10.8.0]
  engine: ruby
libxml:
  binding: extension
  compiled: 2.7.3
  loaded: 2.7.3

(breaks on 1.8.7 too)

[flavorjones]

@msavy - you're seeing something very different than this issue was opened to address; and it's not a bug.

The variable you've named document in your snippet is not, in fact, a Nokogiri::XML::Document -- it's a Nokogiri::XML::Builder.

See http://nokogiri.org/Nokogiri/XML/Builder.html for supporting documentation, but Builder understands #to_xml and that's about it. Nearly any other method you call on it will be routed to #method_missing and cause the creation of an XML node. That's how Builder works its magic.

If you call #root on a bare Builder object, that act is interpreted as wanting to create an XML node <root /> under the root node. If you've already got a root node, then that's going to create an ill-formed document and raise the error you see above.