Reader outer_xml is decoding escaped entities in JRuby

[jcronk]

Example:

require 'nokogiri'

reader = Nokogiri::XML::Reader(File.open 'entities.xml')
reader.each do |node|
  puts node.outer_xml if node.node_type == 1 && node.name == 'Main'
end

using this XML:

<Root>
    <Main>
        <Element>This has &amp; in it</Element>
    </Main>
    <Main>
        <Element>This has &gt; in it</Element>
    </Main>
    <Main>
        <Element>This has &lt; in it</Element>
    </Main>
</Root>

Produces this output:

<Main>
        <Element>This has & in it</Element>
    </Main>
<Main>
        <Element>This has > in it</Element>
    </Main>
<Main>
        <Element>This has < in it</Element>
    </Main>

This makes it impossible to pass outer_xml to Nokogiri::XML() because with strict mode on, it will throw an error, and with strict off, it will drop parts of the XML. I tested this on jruby-1.7.24 and jruby-9.0.5.0.

[flavorjones]

Confirmed, MRI generates this output in the above test case:

<Main>
        <Element>This has &amp; in it</Element>
    </Main>
<Main>
        <Element>This has &gt; in it</Element>
    </Main>
<Main>
        <Element>This has &lt; in it</Element>
    </Main>

while JRuby generates:

<Main>
        <Element>This has & in it</Element>
    </Main>
<Main>
        <Element>This has > in it</Element>
    </Main>
<Main>
        <Element>This has < in it</Element>
    </Main>

[flavorjones]

Worth noting: Reader#inner_xml has the same issue on JRuby.

#! /usr/bin/env ruby

require "bundler/inline"

gemfile do
  source "https://rubygems.org"
  gem "nokogiri", path: "."
end

require 'nokogiri'

xml = <<~XML
<Root>
    <Main>
        <Element>This has &amp; in it</Element>
    </Main>
    <Main>
        <Element>This has &gt; in it</Element>
    </Main>
    <Main>
        <Element>This has &lt; in it</Element>
    </Main>
</Root>
XML

outer_xml = +""
reader = Nokogiri::XML::Reader(xml)
reader.each do |node|
  outer_xml << node.outer_xml if node.node_type == 1 && node.name == 'Main'
end

pp ["outer:", outer_xml]

inner_xml = +""
reader = Nokogiri::XML::Reader(xml)
reader.each do |node|
  inner_xml << node.inner_xml if node.node_type == 1 && node.name == 'Main'
end

pp ["inner:", inner_xml]

unless outer_xml.include?("&amp;") && inner_xml.include?("&amp;")
  puts "Failed"
  exit 1
end

puts "OK"

[flavorjones]

Proposed fix is in #3268